Update libvpx and incorporate new build system
Move libvpx down a directory. Separate libwebm and rename it. It would
be more accurate to rename external/libvpx to external/webm.
Use file lists directly from upstream libvpx. This allows adding new
targets and new features (such as the encoder) easily. MIPS and x86
options are in progress. See new file "UPDATING"
The new libvpx checkout is not from a release branch. The decoder is
stable but it should be checked and potentially updated if the encoder
is enabled.
Requires Iec8566bce0a4763a9a17b4a46f15aa39b5bcd141 which updates
libstagefright to account for new paths and library names.
Change-Id: I2c0a87b0fc049e855e21f455b806287b578091de
diff --git a/AUTHORS b/AUTHORS
deleted file mode 100644
index b8fc45e..0000000
--- a/AUTHORS
+++ /dev/null
@@ -1,45 +0,0 @@
-# This file is automatically generated from the git commit history
-# by tools/gen_authors.sh.
-
-Aaron Watry <awatry@gmail.com>
-Adrian Grange <agrange@google.com>
-Alex Converse <alex.converse@gmail.com>
-Andoni Morales Alastruey <ylatuya@gmail.com>
-Andres Mejia <mcitadel@gmail.com>
-Attila Nagy <attilanagy@google.com>
-Fabio Pedretti <fabio.ped@libero.it>
-Frank Galligan <fgalligan@google.com>
-Fredrik Söderquist <fs@opera.com>
-Fritz Koenig <frkoenig@google.com>
-Gaute Strokkenes <gaute.strokkenes@broadcom.com>
-Giuseppe Scrivano <gscrivano@gnu.org>
-Guillermo Ballester Valor <gbvalor@gmail.com>
-Henrik Lundin <hlundin@google.com>
-James Berry <jamesberry@google.com>
-James Zern <jzern@google.com>
-Jan Kratochvil <jan.kratochvil@redhat.com>
-Jeff Muizelaar <jmuizelaar@mozilla.com>
-Jim Bankoski <jimbankoski@google.com>
-Johann Koenig <johannkoenig@google.com>
-John Koleszar <jkoleszar@google.com>
-Justin Clift <justin@salasaga.org>
-Justin Lebar <justin.lebar@gmail.com>
-Luca Barbato <lu_zero@gentoo.org>
-Makoto Kato <makoto.kt@gmail.com>
-Martin Ettl <ettl.martin78@googlemail.com>
-Michael Kohler <michaelkohler@live.com>
-Mikhal Shemer <mikhal@google.com>
-Pascal Massimino <pascal.massimino@gmail.com>
-Patrik Westin <patrik.westin@gmail.com>
-Paul Wilkins <paulwilkins@google.com>
-Pavol Rusnak <stick@gk2.sk>
-Philip Jägenstedt <philipj@opera.com>
-Scott LaVarnway <slavarnway@google.com>
-Tero Rintaluoma <teror@google.com>
-Timothy B. Terriberry <tterribe@xiph.org>
-Tom Finegan <tomfinegan@google.com>
-Yaowu Xu <yaowu@google.com>
-Yunqing Wang <yunqingwang@google.com>
-Google Inc.
-The Mozilla Foundation
-The Xiph.Org Foundation
diff --git a/Android.mk b/Android.mk
index 53eea93..d0fe10a 100644
--- a/Android.mk
+++ b/Android.mk
@@ -1,199 +1,9 @@
LOCAL_PATH := $(call my-dir)
include $(CLEAR_VARS)
-LOCAL_SRC_FILES = \
- mkvparser/mkvparser.cpp \
- vpx/src/vpx_codec.c \
- vpx/src/vpx_decoder.c \
- vpx/src/vpx_image.c \
- vpx_mem/vpx_mem.c \
- vpx_scale/generic/vpxscale.c \
- vpx_scale/generic/yv12config.c \
- vpx_scale/generic/yv12extend.c \
- vpx_scale/generic/gen_scalers.c \
- vpx_scale/generic/scalesystemdependent.c \
- vp8/common/alloccommon.c \
- vp8/common/arm/arm_systemdependent.c \
- vp8/common/arm/reconintra_arm.c \
- vp8/common/blockd.c \
- vp8/common/debugmodes.c \
- vp8/common/entropy.c \
- vp8/common/entropymode.c \
- vp8/common/entropymv.c \
- vp8/common/extend.c \
- vp8/common/filter.c \
- vp8/common/findnearmv.c \
- vp8/common/generic/systemdependent.c \
- vp8/common/idctllm.c \
- vp8/common/invtrans.c \
- vp8/common/loopfilter.c \
- vp8/common/loopfilter_filters.c \
- vp8/common/mbpitch.c \
- vp8/common/modecont.c \
- vp8/common/modecontext.c \
- vp8/common/quant_common.c \
- vp8/common/recon.c \
- vp8/common/reconinter.c \
- vp8/common/reconintra.c \
- vp8/common/reconintra4x4.c \
- vp8/common/setupintrarecon.c \
- vp8/common/swapyv12buffer.c \
- vp8/common/textblit.c \
- vp8/common/treecoder.c \
- vp8/common/postproc.c \
- vp8/vp8_cx_iface.c \
- vp8/vp8_dx_iface.c \
- vp8/decoder/arm/arm_dsystemdependent.c \
- vp8/decoder/dboolhuff.c \
- vp8/decoder/decodemv.c \
- vp8/decoder/decodframe.c \
- vp8/decoder/dequantize.c \
- vp8/decoder/detokenize.c \
- vp8/decoder/generic/dsystemdependent.c \
- vp8/decoder/onyxd_if.c \
- vp8/decoder/reconintra_mt.c \
- vp8/decoder/threading.c \
- vpx_config.c \
- vp8/decoder/arm/neon/idct_blk_neon.c
+# libvpx
+# if ARMv7 + NEON etc blah blah
+include external/libvpx/libvpx.mk
-LOCAL_CFLAGS := \
- -DHAVE_CONFIG_H=vpx_config.h
-
-LOCAL_MODULE := libvpx
-
-ifeq ($(TARGET_ARCH),mips)
- ifneq ($(ARCH_HAS_BIGENDIAN),true)
- ifeq ($(ARCH_MIPS_DSP_REV),2)
- LOCAL_SRC_FILES += \
- vp8/common/mips/idct_mips.c \
- vp8/common/mips/mips_systemdependent.c \
- vp8/common/mips/subpixel_mips.c \
- vp8/common/mips/loopfilter_filters_mips.c \
- vp8/common/mips/loopfilter_mips.c \
- vp8/common/mips/reconinter_mips.c \
- vp8/decoder/mips/dequantize_mips.c \
- vp8/decoder/mips/idct_blk_mips.c
-
- LOCAL_CFLAGS += -DMIPS_DSP_REV=$(ARCH_MIPS_DSP_REV)
-
- else
- ifeq ($(ARCH_MIPS_DSP_REV),1)
- LOCAL_SRC_FILES += \
- vp8/common/mips/idct_mips.c \
- vp8/common/mips/mips_systemdependent.c \
- vp8/common/mips/reconinter_mips.c \
-
- LOCAL_CFLAGS += -DMIPS_DSP_REV=$(ARCH_MIPS_DSP_REV)
-
- else
- LOCAL_CFLAGS += -DMIPS_DSP_REV=0
- endif # mips_dsp_rev1
- endif # mips_dsp_rev2
-
- endif #bigendian
-endif #mips
-
-ifeq ($(TARGET_ARCH),arm)
-
-LOCAL_MODULE_CLASS := STATIC_LIBRARIES
-intermediates := $(call local-intermediates-dir)
-
-LOCAL_SRC_FILES += \
- vp8/common/arm/loopfilter_arm.c \
- vp8/decoder/arm/dequantize_arm.c \
-
-ifeq ($(ARCH_ARM_HAVE_NEON),true)
-
-LOCAL_CFLAGS += -D__ARM_HAVE_NEON
-
-ASM_FILES = \
- vp8/common/arm/neon/bilinearpredict16x16_neon.s \
- vp8/common/arm/neon/bilinearpredict4x4_neon.s \
- vp8/common/arm/neon/bilinearpredict8x4_neon.s \
- vp8/common/arm/neon/bilinearpredict8x8_neon.s \
- vp8/common/arm/neon/buildintrapredictorsmby_neon.s \
- vp8/common/arm/neon/copymem16x16_neon.s \
- vp8/common/arm/neon/copymem8x4_neon.s \
- vp8/common/arm/neon/copymem8x8_neon.s \
- vp8/common/arm/neon/iwalsh_neon.s \
- vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.s \
- vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.s \
- vp8/common/arm/neon/recon16x16mb_neon.s \
- vp8/common/arm/neon/recon2b_neon.s \
- vp8/common/arm/neon/recon4b_neon.s \
- vp8/common/arm/neon/reconb_neon.s \
- vp8/common/arm/neon/save_neon_reg.s \
- vp8/common/arm/neon/shortidct4x4llm_1_neon.s \
- vp8/common/arm/neon/shortidct4x4llm_neon.s \
- vp8/common/arm/neon/sixtappredict16x16_neon.s \
- vp8/common/arm/neon/sixtappredict4x4_neon.s \
- vp8/common/arm/neon/sixtappredict8x4_neon.s \
- vp8/common/arm/neon/sixtappredict8x8_neon.s \
- vp8/common/arm/neon/dc_only_idct_add_neon.s \
- vp8/decoder/arm/neon/dequantizeb_neon.s \
- vp8/decoder/arm/neon/dequant_idct_neon.s \
- vp8/decoder/arm/neon/idct_dequant_0_2x_neon.s \
- vp8/decoder/arm/neon/idct_dequant_dc_0_2x_neon.s \
- vp8/decoder/arm/neon/idct_dequant_dc_full_2x_neon.s \
- vp8/decoder/arm/neon/idct_dequant_full_2x_neon.s \
- vp8/common/arm/neon/loopfilter_neon.s \
- vp8/common/arm/neon/mbloopfilter_neon.s \
-
-else ifeq ($(ARCH_ARM_HAVE_ARMV7A),true)
-
-# Really, we only need ARMV6 support but I could not find an environment
-# variable to query that...
-
-LOCAL_CFLAGS += -D__ARM_HAVE_ARMV6
-
-ASM_FILES = \
- vp8/common/arm/armv6/bilinearfilter_v6.s \
- vp8/common/arm/armv6/copymem8x4_v6.s \
- vp8/common/arm/armv6/copymem8x8_v6.s \
- vp8/common/arm/armv6/copymem16x16_v6.s \
- vp8/common/arm/armv6/dc_only_idct_add_v6.s \
- vp8/common/arm/armv6/filter_v6.s \
- vp8/common/arm/armv6/iwalsh_v6.s \
- vp8/common/arm/armv6/loopfilter_v6.s \
- vp8/common/arm/armv6/recon_v6.s \
- vp8/common/arm/armv6/simpleloopfilter_v6.s \
- vp8/common/arm/armv6/sixtappredict8x4_v6.s \
- vp8/decoder/arm/armv6/dequant_dc_idct_v6.s \
- vp8/decoder/arm/armv6/dequant_idct_v6.s \
- vp8/decoder/arm/armv6/dequantize_v6.s \
-
-LOCAL_SRC_FILES += \
- vp8/common/arm/bilinearfilter_arm.c \
- vp8/common/arm/filter_arm.c \
- vp8/decoder/arm/armv6/idct_blk_v6.c \
-
-else
-
-LOCAL_SRC_FILES += vp8/decoder/idct_blk.c
-
-endif
-
-# All the assembly sources must be converted from ADS to GAS compatible format
-VPX_GEN := $(addprefix $(intermediates)/, $(ASM_FILES))
-$(VPX_GEN) : PRIVATE_PATH := $(LOCAL_PATH)
-$(VPX_GEN) : PRIVATE_CUSTOM_TOOL = cat $< | perl external/libvpx/build/make/ads2gas.pl > $@
-$(VPX_GEN) : $(intermediates)/%.s : $(LOCAL_PATH)/%.asm
- $(transform-generated-source)
-
-LOCAL_GENERATED_SOURCES += $(VPX_GEN)
-
-else # non-ARM
-
-LOCAL_SRC_FILES += vp8/decoder/idct_blk.c
-
-endif
-
-LOCAL_C_INCLUDES := \
- $(LOCAL_PATH)/vpx_ports \
- $(LOCAL_PATH)/vp8/common \
- $(LOCAL_PATH)/vp8/encoder \
- $(LOCAL_PATH)/vp8/decoder \
- $(LOCAL_PATH)/vp8 \
- $(LOCAL_PATH)/vpx_codec
-
-include $(BUILD_STATIC_LIBRARY)
+# libwebm
+include external/libvpx/libwebm.mk
diff --git a/CHANGELOG b/CHANGELOG
deleted file mode 100644
index e8760d1..0000000
--- a/CHANGELOG
+++ /dev/null
@@ -1,253 +0,0 @@
-2011-03-07 v0.9.6 "Bali"
- Our second named release, focused on a faster, higher quality, encoder.
-
- - Upgrading:
- This release is backwards compatible with Aylesbury (v0.9.5). Users
- of older releases should refer to the Upgrading notes in this
- document for that release.
-
- - Enhancements:
- vpxenc --psnr shows a summary when encode completes
- --tune=ssim option to enable activity masking
- improved postproc visualizations for development
- updated support for Apple iOS to SDK 4.2
- query decoder to determine which reference frames were updated
- implemented error tracking in the decoder
- fix pipe support on windows
-
- - Speed:
- Primary focus was on good quality mode, speed 0. Average improvement
- on x86 about 40%, up to 100% on user-generated content at that speed.
- Best quality mode speed improved 35%, and realtime speed 10-20%. This
- release also saw significant improvement in realtime encoding speed
- on ARM platforms.
-
- Improved encoder threading
- Dont pick encoder filter level when loopfilter is disabled.
- Avoid double copying of key frames into alt and golden buffer
- FDCT optimizations.
- x86 sse2 temporal filter
- SSSE3 version of fast quantizer
- vp8_rd_pick_best_mbsegmentation code restructure
- Adjusted breakout RD for SPLITMV
- Changed segmentation check order
- Improved rd_pick_intra4x4block
- Adds armv6 optimized variance calculation
- ARMv6 optimized sad16x16
- ARMv6 optimized half pixel variance calculations
- Full search SAD function optimization in SSE4.1
- Improve MV prediction accuracy to achieve performance gain
- Improve MV prediction in vp8_pick_inter_mode() for speed>3
-
- - Quality:
- Best quality mode improved PSNR 6.3%, and SSIM 6.1%. This release
- also includes support for "activity masking," which greatly improves
- SSIM at the expense of PSNR. For now, this feature is available with
- the --tune=ssim option. Further experimentation in this area
- is ongoing. This release also introduces a new rate control mode
- called "CQ," which changes the allocation of bits within a clip to
- the sections where they will have the most visual impact.
-
- Tuning for the more exact quantizer.
- Relax rate control for last few frames
- CQ Mode
- Limit key frame quantizer for forced key frames.
- KF/GF Pulsing
- Add simple version of activity masking.
- make rdmult adaptive for intra in quantizer RDO
- cap the best quantizer for 2nd order DC
- change the threshold of DC check for encode breakout
-
- - Bug Fixes:
- Fix crash on Sparc Solaris.
- Fix counter of fixed keyframe distance
- ARNR filter pointer update bug fix
- Fixed use of motion percentage in KF/GF group calc
- Changed condition for using RD in Intra Mode
- Fix encoder real-time only configuration.
- Fix ARM encoder crash with multiple token partitions
- Fixed bug first cluster timecode of webm file is wrong.
- Fixed various encoder bugs with odd-sized images
- vp8e_get_preview fixed when spatial resampling enabled
- quantizer: fix assertion in fast quantizer path
- Allocate source buffers to be multiples of 16
- Fix for manual Golden frame frequency
- Fix drastic undershoot in long form content
-
-
-2010-10-28 v0.9.5 "Aylesbury"
- Our first named release, focused on a faster decoder, and a better encoder.
-
- - Upgrading:
- This release incorporates backwards-incompatible changes to the
- ivfenc and ivfdec tools. These tools are now called vpxenc and vpxdec.
-
- vpxdec
- * the -q (quiet) option has been removed, and replaced with
- -v (verbose). the output is quiet by default. Use -v to see
- the version number of the binary.
-
- * The default behavior is now to write output to a single file
- instead of individual frames. The -y option has been removed.
- Y4M output is the default.
-
- * For raw I420/YV12 output instead of Y4M, the --i420 or --yv12
- options must be specified.
-
- $ ivfdec -o OUTPUT INPUT
- $ vpxdec --i420 -o OUTPUT INPUT
-
- * If an output file is not specified, the default is to write
- Y4M to stdout. This makes piping more natural.
-
- $ ivfdec -y -o - INPUT | ...
- $ vpxdec INPUT | ...
-
- * The output file has additional flexibility for formatting the
- filename. It supports escape characters for constructing a
- filename from the width, height, and sequence number. This
- replaces the -p option. To get the equivalent:
-
- $ ivfdec -p frame INPUT
- $ vpxdec --i420 -o frame-%wx%h-%4.i420 INPUT
-
- vpxenc
- * The output file must be specified with -o, rather than as the
- last argument.
-
- $ ivfenc <options> INPUT OUTPUT
- $ vpxenc <options> -o OUTPUT INPUT
-
- * The output defaults to webm. To get IVF output, use the --ivf
- option.
-
- $ ivfenc <options> INPUT OUTPUT.ivf
- $ vpxenc <options> -o OUTPUT.ivf --ivf INPUT
-
-
- - Enhancements:
- ivfenc and ivfdec have been renamed to vpxenc, vpxdec.
- vpxdec supports .webm input
- vpxdec writes .y4m by default
- vpxenc writes .webm output by default
- vpxenc --psnr now shows the average/overall PSNR at the end
- ARM platforms now support runtime cpu detection
- vpxdec visualizations added for motion vectors, block modes, references
- vpxdec now silent by default
- vpxdec --progress shows frame-by-frame timing information
- vpxenc supports the distinction between --fps and --timebase
- NASM is now a supported assembler
- configure: enable PIC for shared libs by default
- configure: add --enable-small
- configure: support for ppc32-linux-gcc
- configure: support for sparc-solaris-gcc
-
- - Bugs:
- Improve handling of invalid frames
- Fix valgrind errors in the NEON loop filters.
- Fix loopfilter delta zero transitions
- Fix valgrind errors in vp8_sixtap_predict8x4_armv6().
- Build fixes for darwin-icc
-
- - Speed:
- 20-40% (average 28%) improvement in libvpx decoder speed,
- including:
- Rewrite vp8_short_walsh4x4_sse2()
- Optimizations on the loopfilters.
- Miscellaneous improvements for Atom
- Add 4-tap version of 2nd-pass ARMv6 MC filter.
- Improved multithread utilization
- Better instruction choices on x86
- reorder data to use wider instructions
- Update NEON wide idcts
- Make block access to frame buffer sequential
- Improved subset block search
- Bilinear subpixel optimizations for ssse3.
- Decrease memory footprint
-
- Encoder speed improvements (percentage gain not measured):
- Skip unnecessary search of identical frames
- Add SSE2 subtract functions
- Improve bounds checking in vp8_diamond_search_sadx4()
- Added vp8_fast_quantize_b_sse2
-
- - Quality:
- Over 7% overall PSNR improvement (6.3% SSIM) in "best" quality
- encoding mode, and up to 60% improvement on very noisy, still
- or slow moving source video
-
- Motion compensated temporal filter for Alt-Ref Noise Reduction
- Improved use of trellis quantization on 2nd order Y blocks
- Tune effect of motion on KF/GF boost in two pass
- Allow coefficient optimization for good quality speed 0.
- Improved control of active min quantizer for two pass.
- Enable ARFs for non-lagged compress
-
-2010-09-02 v0.9.2
- - Enhancements:
- Disable frame dropping by default
- Improved multithreaded performance
- Improved Force Key Frame Behaviour
- Increased rate control buffer level precision
- Fix bug in 1st pass motion compensation
- ivfenc: correct fixed kf interval, --disable-kf
- - Speed:
- Changed above and left context data layout
- Rework idct calling structure.
- Removed unnecessary MB_MODE_INFO copies
- x86: SSSE3 sixtap prediction
- Reworked IDCT to include reconstruction (add) step
- Swap alt/gold/new/last frame buffer ptrs instead of copying.
- Improve SSE2 loopfilter functions
- Change bitreader to use a larger window.
- Avoid loopfilter reinitialization when possible
- - Quality:
- Normalize quantizer's zero bin and rounding factors
- Add trellis quantization.
- Make the quantizer exact.
- Updates to ARNR filtering algorithm
- Fix breakout thresh computation for golden & AltRef frames
- Redo the forward 4x4 dct
- Improve the accuracy of forward walsh-hadamard transform
- Further adjustment of RD behaviour with Q and Zbin.
- - Build System:
- Allow linking of libs built with MinGW to MSVC
- Fix target auto-detection on mingw32
- Allow --cpu= to work for x86.
- configure: pass original arguments through to make dist
- Fix builds without runtime CPU detection
- msvs: fix install of codec sources
- msvs: Change devenv.com command line for better msys support
- msvs: Add vs9 targets.
- Add x86_64-linux-icc target
- - Bugs:
- Potential crashes on older MinGW builds
- Fix two-pass framrate for Y4M input.
- Fixed simple loop filter, other crashes on ARM v6
- arm: fix missing dependency with --enable-shared
- configure: support directories containing .o
- Replace pinsrw (SSE) with MMX instructions
- apple: include proper mach primatives
- Fixed rate control bug with long key frame interval.
- Fix DSO link errors on x86-64 when not using a version script
- Fixed buffer selection for UV in AltRef filtering
-
-
-2010-06-17 v0.9.1
- - Enhancements:
- * ivfenc/ivfdec now support YUV4MPEG2 input and pipe I/O
- * Speed optimizations
- - Bugfixes:
- * Rate control
- * Prevent out-of-bounds accesses on invalid data
- - Build system updates:
- * Detect toolchain to be used automatically for native builds
- * Support building shared libraries
- * Better autotools emulation (--prefix, --libdir, DESTDIR)
- - Updated LICENSE
- * http://webmproject.blogspot.com/2010/06/changes-to-webm-open-source-license.html
-
-
-2010-05-18 v0.9.0
- - Initial open source release. Welcome to WebM and VP8!
-
diff --git a/LICENSE b/LICENSE
deleted file mode 100644
index 7a6f995..0000000
--- a/LICENSE
+++ /dev/null
@@ -1,30 +0,0 @@
-Copyright (c) 2010, Google Inc. All rights reserved.
-
-Redistribution and use in source and binary forms, with or without
-modification, are permitted provided that the following conditions are
-met:
-
- * Redistributions of source code must retain the above copyright
- notice, this list of conditions and the following disclaimer.
-
- * Redistributions in binary form must reproduce the above copyright
- notice, this list of conditions and the following disclaimer in
- the documentation and/or other materials provided with the
- distribution.
-
- * Neither the name of Google nor the names of its contributors may
- be used to endorse or promote products derived from this software
- without specific prior written permission.
-
-THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-
diff --git a/NOTICE b/NOTICE
index 7a6f995..1ce4434 100644
--- a/NOTICE
+++ b/NOTICE
@@ -1,4 +1,4 @@
-Copyright (c) 2010, Google Inc. All rights reserved.
+Copyright (c) 2010, The WebM Project authors. All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are
@@ -12,9 +12,10 @@
the documentation and/or other materials provided with the
distribution.
- * Neither the name of Google nor the names of its contributors may
- be used to endorse or promote products derived from this software
- without specific prior written permission.
+ * Neither the name of Google, nor the WebM Project, nor the names
+ of its contributors may be used to endorse or promote products
+ derived from this software without specific prior written
+ permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
diff --git a/README b/README
deleted file mode 100644
index dddc5ea..0000000
--- a/README
+++ /dev/null
@@ -1,102 +0,0 @@
-vpx Multi-Format Codec SDK
-README - 19 May 2010
-
-Welcome to the WebM VP8 Codec SDK!
-
-COMPILING THE APPLICATIONS/LIBRARIES:
- The build system used is similar to autotools. Building generally consists of
- "configuring" with your desired build options, then using GNU make to build
- the application.
-
- 1. Prerequisites
-
- * All x86 targets require the Yasm[1] assembler be installed.
- * All Windows builds require that Cygwin[2] be installed.
- * Building the documentation requires PHP[3] and Doxygen[4]. If you do not
- have these packages, you must pass --disable-install-docs to the
- configure script.
-
- [1]: http://www.tortall.net/projects/yasm
- [2]: http://www.cygwin.com
- [3]: http://php.net
- [4]: http://www.doxygen.org
-
- 2. Out-of-tree builds
- Out of tree builds are a supported method of building the application. For
- an out of tree build, the source tree is kept separate from the object
- files produced during compilation. For instance:
-
- $ mkdir build
- $ cd build
- $ ../libvpx/configure <options>
- $ make
-
- 3. Configuration options
- The 'configure' script supports a number of options. The --help option can be
- used to get a list of supported options:
- $ ../libvpx/configure --help
-
- 4. Cross development
- For cross development, the most notable option is the --target option. The
- most up-to-date list of supported targets can be found at the bottom of the
- --help output of the configure script. As of this writing, the list of
- available targets is:
-
- armv5te-linux-rvct
- armv5te-linux-gcc
- armv5te-symbian-gcc
- armv6-darwin-gcc
- armv6-linux-rvct
- armv6-linux-gcc
- armv6-symbian-gcc
- iwmmxt-linux-rvct
- iwmmxt-linux-gcc
- iwmmxt2-linux-rvct
- iwmmxt2-linux-gcc
- armv7-linux-rvct
- armv7-linux-gcc
- mips32-linux-gcc
- ppc32-darwin8-gcc
- ppc32-darwin9-gcc
- ppc64-darwin8-gcc
- ppc64-darwin9-gcc
- ppc64-linux-gcc
- x86-darwin8-gcc
- x86-darwin8-icc
- x86-darwin9-gcc
- x86-darwin9-icc
- x86-linux-gcc
- x86-linux-icc
- x86-solaris-gcc
- x86-win32-vs7
- x86-win32-vs8
- x86_64-darwin9-gcc
- x86_64-linux-gcc
- x86_64-solaris-gcc
- x86_64-win64-vs8
- universal-darwin8-gcc
- universal-darwin9-gcc
- generic-gnu
-
- The generic-gnu target, in conjunction with the CROSS environment variable,
- can be used to cross compile architectures that aren't explicitly listed, if
- the toolchain is a cross GNU (gcc/binutils) toolchain. Other POSIX toolchains
- will likely work as well. For instance, to build using the mipsel-linux-uclibc
- toolchain, the following command could be used (note, POSIX SH syntax, adapt
- to your shell as necessary):
-
- $ CROSS=mipsel-linux-uclibc- ../libvpx/configure
-
- In addition, the executables to be invoked can be overridden by specifying the
- environment variables: CC, AR, LD, AS, STRIP, NM. Additional flags can be
- passed to these executables with CFLAGS, LDFLAGS, and ASFLAGS.
-
- 5. Configuration errors
- If the configuration step fails, the first step is to look in the error log.
- This defaults to config.err. This should give a good indication of what went
- wrong. If not, contact us for support.
-
-SUPPORT
- This library is an open source project supported by its community. Please
- please email webm-users@webmproject.org for help.
-
diff --git a/UPDATING b/UPDATING
new file mode 100644
index 0000000..0633ccc
--- /dev/null
+++ b/UPDATING
@@ -0,0 +1,47 @@
+Pull the new libvpx checkout into external/libvpx/libvpx:
+$ cd external/libvpx/
+$ rm -rf libvpx
+$ git clone http://git.chromium.org/webm/libvpx.git
+$ cd libvpx
+$ git checkout <branch>
+$ rm -rf .git*
+
+Enter the subdirectory for the relevant platform. For example, armv7-neon for
+armv7 targets with neon extensions. We disable many features. Some for
+functional reasons and some for aesthetic ones.
+
+Functional:
+--force-target=$TARGET
+ The "Android" support in libvpx is targeted at the NDK and does not yet
+ include support for architectures such as MIPS. However, we can still generate
+ the necessary files for them.
+--disable-runtime-cpu-detect
+--disable-neon
+ The platform knows at build time what extensions are supported.
+--sdk-path=$SDK_PATH
+ For configuration we do some compiler tests. It is much easier to accept them
+ than to work around them. This uses the compilers included in the NDK.
+--disable-vp8-encoder
+ Build only the decoder.
+
+Aesthetic:
+--disable-examples
+--disable-docs
+ Skip unnecessary extra makefiles.
+
+Example:
+$ cd external/libvpx/armv7a
+$ ../libvpx/configure --target=armv7-android-gcc --disable-runtime-cpu-detect \
+ --disable-neon --sdk-path=$ANDROID_NDK_ROOT --disable-vp8-encoder \
+ --disable-examples --disable-docs
+
+Run 'make libvpx_srcs.txt'
+This will generate a file listing all of the required sources. It will also
+generate vpx_rtcd.h and vpx_version.h
+
+Remove the unused files leaving only:
+libvpx_srcs.txt
+vpx_config.c
+vpx_config.h
+vpx_rtcd.h
+vpx_version.h
diff --git a/args.c b/args.c
deleted file mode 100644
index 7b2cc3a..0000000
--- a/args.c
+++ /dev/null
@@ -1,265 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdlib.h>
-#include <string.h>
-#include <limits.h>
-#include "args.h"
-
-#ifdef _MSC_VER
-#define snprintf _snprintf
-#endif
-
-#if defined(__GNUC__) && __GNUC__
-extern void die(const char *fmt, ...) __attribute__((noreturn));
-#else
-extern void die(const char *fmt, ...);
-#endif
-
-
-struct arg arg_init(char **argv)
-{
- struct arg a;
-
- a.argv = argv;
- a.argv_step = 1;
- a.name = NULL;
- a.val = NULL;
- a.def = NULL;
- return a;
-}
-
-int arg_match(struct arg *arg_, const struct arg_def *def, char **argv)
-{
- struct arg arg;
-
- if (!argv[0] || argv[0][0] != '-')
- return 0;
-
- arg = arg_init(argv);
-
- if (def->short_name
- && strlen(arg.argv[0]) == strlen(def->short_name) + 1
- && !strcmp(arg.argv[0] + 1, def->short_name))
- {
-
- arg.name = arg.argv[0] + 1;
- arg.val = def->has_val ? arg.argv[1] : NULL;
- arg.argv_step = def->has_val ? 2 : 1;
- }
- else if (def->long_name)
- {
- int name_len = strlen(def->long_name);
-
- if (strlen(arg.argv[0]) >= name_len + 2
- && arg.argv[0][1] == '-'
- && !strncmp(arg.argv[0] + 2, def->long_name, name_len)
- && (arg.argv[0][name_len+2] == '='
- || arg.argv[0][name_len+2] == '\0'))
- {
-
- arg.name = arg.argv[0] + 2;
- arg.val = arg.name[name_len] == '=' ? arg.name + name_len + 1 : NULL;
- arg.argv_step = 1;
- }
- }
-
- if (arg.name && !arg.val && def->has_val)
- die("Error: option %s requires argument.\n", arg.name);
-
- if (arg.name && arg.val && !def->has_val)
- die("Error: option %s requires no argument.\n", arg.name);
-
- if (arg.name
- && (arg.val || !def->has_val))
- {
- arg.def = def;
- *arg_ = arg;
- return 1;
- }
-
- return 0;
-}
-
-
-const char *arg_next(struct arg *arg)
-{
- if (arg->argv[0])
- arg->argv += arg->argv_step;
-
- return *arg->argv;
-}
-
-
-char **argv_dup(int argc, const char **argv)
-{
- char **new_argv = malloc((argc + 1) * sizeof(*argv));
-
- memcpy(new_argv, argv, argc * sizeof(*argv));
- new_argv[argc] = NULL;
- return new_argv;
-}
-
-
-void arg_show_usage(FILE *fp, const struct arg_def *const *defs)
-{
- char option_text[40] = {0};
-
- for (; *defs; defs++)
- {
- const struct arg_def *def = *defs;
- char *short_val = def->has_val ? " <arg>" : "";
- char *long_val = def->has_val ? "=<arg>" : "";
-
- if (def->short_name && def->long_name)
- {
- char *comma = def->has_val ? "," : ", ";
-
- snprintf(option_text, 37, "-%s%s%s --%s%6s",
- def->short_name, short_val, comma,
- def->long_name, long_val);
- }
- else if (def->short_name)
- snprintf(option_text, 37, "-%s%s",
- def->short_name, short_val);
- else if (def->long_name)
- snprintf(option_text, 37, " --%s%s",
- def->long_name, long_val);
-
- fprintf(fp, " %-37s\t%s\n", option_text, def->desc);
-
- if(def->enums)
- {
- const struct arg_enum_list *listptr;
-
- fprintf(fp, " %-37s\t ", "");
-
- for(listptr = def->enums; listptr->name; listptr++)
- fprintf(fp, "%s%s", listptr->name,
- listptr[1].name ? ", " : "\n");
- }
- }
-}
-
-
-unsigned int arg_parse_uint(const struct arg *arg)
-{
- long int rawval;
- char *endptr;
-
- rawval = strtol(arg->val, &endptr, 10);
-
- if (arg->val[0] != '\0' && endptr[0] == '\0')
- {
- if (rawval >= 0 && rawval <= UINT_MAX)
- return rawval;
-
- die("Option %s: Value %ld out of range for unsigned int\n",
- arg->name, rawval);
- }
-
- die("Option %s: Invalid character '%c'\n", arg->name, *endptr);
- return 0;
-}
-
-
-int arg_parse_int(const struct arg *arg)
-{
- long int rawval;
- char *endptr;
-
- rawval = strtol(arg->val, &endptr, 10);
-
- if (arg->val[0] != '\0' && endptr[0] == '\0')
- {
- if (rawval >= INT_MIN && rawval <= INT_MAX)
- return rawval;
-
- die("Option %s: Value %ld out of range for signed int\n",
- arg->name, rawval);
- }
-
- die("Option %s: Invalid character '%c'\n", arg->name, *endptr);
- return 0;
-}
-
-
-struct vpx_rational
-{
- int num; /**< fraction numerator */
- int den; /**< fraction denominator */
-};
-struct vpx_rational arg_parse_rational(const struct arg *arg)
-{
- long int rawval;
- char *endptr;
- struct vpx_rational rat;
-
- /* parse numerator */
- rawval = strtol(arg->val, &endptr, 10);
-
- if (arg->val[0] != '\0' && endptr[0] == '/')
- {
- if (rawval >= INT_MIN && rawval <= INT_MAX)
- rat.num = rawval;
- else die("Option %s: Value %ld out of range for signed int\n",
- arg->name, rawval);
- }
- else die("Option %s: Expected / at '%c'\n", arg->name, *endptr);
-
- /* parse denominator */
- rawval = strtol(endptr + 1, &endptr, 10);
-
- if (arg->val[0] != '\0' && endptr[0] == '\0')
- {
- if (rawval >= INT_MIN && rawval <= INT_MAX)
- rat.den = rawval;
- else die("Option %s: Value %ld out of range for signed int\n",
- arg->name, rawval);
- }
- else die("Option %s: Invalid character '%c'\n", arg->name, *endptr);
-
- return rat;
-}
-
-
-int arg_parse_enum(const struct arg *arg)
-{
- const struct arg_enum_list *listptr;
- long int rawval;
- char *endptr;
-
- /* First see if the value can be parsed as a raw value */
- rawval = strtol(arg->val, &endptr, 10);
- if (arg->val[0] != '\0' && endptr[0] == '\0')
- {
- /* Got a raw value, make sure it's valid */
- for(listptr = arg->def->enums; listptr->name; listptr++)
- if(listptr->val == rawval)
- return rawval;
- }
-
- /* Next see if it can be parsed as a string */
- for(listptr = arg->def->enums; listptr->name; listptr++)
- if(!strcmp(arg->val, listptr->name))
- return listptr->val;
-
- die("Option %s: Invalid value '%s'\n", arg->name, arg->val);
- return 0;
-}
-
-
-int arg_parse_enum_or_int(const struct arg *arg)
-{
- if(arg->def->enums)
- return arg_parse_enum(arg);
- return arg_parse_int(arg);
-}
diff --git a/armv7a-neon/libvpx_srcs.txt b/armv7a-neon/libvpx_srcs.txt
new file mode 100644
index 0000000..3b2e0b2
--- /dev/null
+++ b/armv7a-neon/libvpx_srcs.txt
@@ -0,0 +1,179 @@
+build/make/rtcd.sh
+build/make/version.sh
+CHANGELOG
+libs.mk
+vp8/common/alloccommon.c
+vp8/common/alloccommon.h
+vp8/common/arm/armv6/bilinearfilter_v6.asm.s
+vp8/common/arm/armv6/copymem16x16_v6.asm.s
+vp8/common/arm/armv6/copymem8x4_v6.asm.s
+vp8/common/arm/armv6/copymem8x8_v6.asm.s
+vp8/common/arm/armv6/dc_only_idct_add_v6.asm.s
+vp8/common/arm/armv6/dequant_idct_v6.asm.s
+vp8/common/arm/armv6/dequantize_v6.asm.s
+vp8/common/arm/armv6/filter_v6.asm.s
+vp8/common/arm/armv6/idct_blk_v6.c
+vp8/common/arm/armv6/idct_v6.asm.s
+vp8/common/arm/armv6/intra4x4_predict_v6.asm.s
+vp8/common/arm/armv6/iwalsh_v6.asm.s
+vp8/common/arm/armv6/loopfilter_v6.asm.s
+vp8/common/arm/armv6/simpleloopfilter_v6.asm.s
+vp8/common/arm/armv6/sixtappredict8x4_v6.asm.s
+vp8/common/arm/armv6/vp8_sad16x16_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance16x16_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance8x8_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6.asm.s
+vp8/common/arm/bilinearfilter_arm.c
+vp8/common/arm/bilinearfilter_arm.h
+vp8/common/arm/dequantize_arm.c
+vp8/common/arm/filter_arm.c
+vp8/common/arm/loopfilter_arm.c
+vp8/common/arm/neon/bilinearpredict16x16_neon.asm.s
+vp8/common/arm/neon/bilinearpredict4x4_neon.asm.s
+vp8/common/arm/neon/bilinearpredict8x4_neon.asm.s
+vp8/common/arm/neon/bilinearpredict8x8_neon.asm.s
+vp8/common/arm/neon/buildintrapredictorsmby_neon.asm.s
+vp8/common/arm/neon/copymem16x16_neon.asm.s
+vp8/common/arm/neon/copymem8x4_neon.asm.s
+vp8/common/arm/neon/copymem8x8_neon.asm.s
+vp8/common/arm/neon/dc_only_idct_add_neon.asm.s
+vp8/common/arm/neon/dequant_idct_neon.asm.s
+vp8/common/arm/neon/dequantizeb_neon.asm.s
+vp8/common/arm/neon/idct_blk_neon.c
+vp8/common/arm/neon/idct_dequant_0_2x_neon.asm.s
+vp8/common/arm/neon/idct_dequant_full_2x_neon.asm.s
+vp8/common/arm/neon/iwalsh_neon.asm.s
+vp8/common/arm/neon/loopfilter_neon.asm.s
+vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm.s
+vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm.s
+vp8/common/arm/neon/mbloopfilter_neon.asm.s
+vp8/common/arm/neon/sad16_neon.asm.s
+vp8/common/arm/neon/sad8_neon.asm.s
+vp8/common/arm/neon/save_reg_neon.asm.s
+vp8/common/arm/neon/shortidct4x4llm_neon.asm.s
+vp8/common/arm/neon/sixtappredict16x16_neon.asm.s
+vp8/common/arm/neon/sixtappredict4x4_neon.asm.s
+vp8/common/arm/neon/sixtappredict8x4_neon.asm.s
+vp8/common/arm/neon/sixtappredict8x8_neon.asm.s
+vp8/common/arm/neon/variance_neon.asm.s
+vp8/common/arm/neon/vp8_subpixelvariance16x16_neon.asm.s
+vp8/common/arm/neon/vp8_subpixelvariance16x16s_neon.asm.s
+vp8/common/arm/neon/vp8_subpixelvariance8x8_neon.asm.s
+vp8/common/arm/reconintra_arm.c
+vp8/common/arm/variance_arm.c
+vp8/common/asm_com_offsets.c
+vp8/common/blockd.c
+vp8/common/blockd.h
+vp8/common/coefupdateprobs.h
+vp8/common/common.h
+vp8/common/debugmodes.c
+vp8/common/default_coef_probs.h
+vp8/common/dequantize.c
+vp8/common/entropy.c
+vp8/common/entropy.h
+vp8/common/entropymode.c
+vp8/common/entropymode.h
+vp8/common/entropymv.c
+vp8/common/entropymv.h
+vp8/common/extend.c
+vp8/common/extend.h
+vp8/common/filter.c
+vp8/common/filter.h
+vp8/common/findnearmv.c
+vp8/common/findnearmv.h
+vp8/common/generic/systemdependent.c
+vp8/common/header.h
+vp8/common/idct_blk.c
+vp8/common/idctllm.c
+vp8/common/invtrans.h
+vp8/common/loopfilter.c
+vp8/common/loopfilter_filters.c
+vp8/common/loopfilter.h
+vp8/common/mbpitch.c
+vp8/common/modecont.c
+vp8/common/modecont.h
+vp8/common/mv.h
+vp8/common/onyxc_int.h
+vp8/common/onyxd.h
+vp8/common/onyx.h
+vp8/common/ppflags.h
+vp8/common/pragmas.h
+vp8/common/quant_common.c
+vp8/common/quant_common.h
+vp8/common/reconinter.c
+vp8/common/reconinter.h
+vp8/common/reconintra4x4.c
+vp8/common/reconintra4x4.h
+vp8/common/reconintra.c
+vp8/common/rtcd.c
+vp8/common/rtcd_defs.sh
+vp8/common/sad_c.c
+vp8/common/setupintrarecon.c
+vp8/common/setupintrarecon.h
+vp8/common/swapyv12buffer.c
+vp8/common/swapyv12buffer.h
+vp8/common/systemdependent.h
+vp8/common/threading.h
+vp8/common/treecoder.c
+vp8/common/treecoder.h
+vp8/common/variance_c.c
+vp8/common/variance.h
+vp8/common/vp8_entropymodedata.h
+vp8/decoder/asm_dec_offsets.c
+vp8/decoder/dboolhuff.c
+vp8/decoder/dboolhuff.h
+vp8/decoder/decodemv.c
+vp8/decoder/decodemv.h
+vp8/decoder/decoderthreading.h
+vp8/decoder/decodframe.c
+vp8/decoder/detokenize.c
+vp8/decoder/detokenize.h
+vp8/decoder/onyxd_if.c
+vp8/decoder/onyxd_int.h
+vp8/decoder/threading.c
+vp8/decoder/treereader.h
+vp8/vp8_common.mk
+vp8/vp8_dx_iface.c
+vp8/vp8dx.mk
+vpx_config.c
+vpx/internal/vpx_codec_internal.h
+vpx_mem/include/vpx_mem_intrnl.h
+vpx_mem/vpx_mem.c
+vpx_mem/vpx_mem.h
+vpx_mem/vpx_mem.mk
+vpx_ports/arm_cpudetect.c
+vpx_ports/arm.h
+vpx_ports/asm_offsets.h
+vpx_ports/mem.h
+vpx_ports/vpx_ports.mk
+vpx_ports/vpx_timer.h
+vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm.s
+vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm.s
+vpx_scale/arm/neon/vp8_vpxyv12_copy_y_neon.asm.s
+vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm.s
+vpx_scale/arm/neon/yv12extend_arm.c
+vpx_scale/generic/gen_scalers.c
+vpx_scale/generic/vpxscale.c
+vpx_scale/generic/yv12config.c
+vpx_scale/generic/yv12extend.c
+vpx_scale/generic/yv12extend_generic.h
+vpx_scale/scale_mode.h
+vpx_scale/vpxscale.h
+vpx_scale/vpx_scale.mk
+vpx_scale/yv12config.h
+vpx/src/vpx_codec.c
+vpx/src/vpx_decoder.c
+vpx/src/vpx_encoder.c
+vpx/src/vpx_image.c
+vpx/vp8dx.h
+vpx/vp8.h
+vpx/vpx_codec.h
+vpx/vpx_codec_impl_bottom.h
+vpx/vpx_codec_impl_top.h
+vpx/vpx_codec.mk
+vpx/vpx_decoder.h
+vpx/vpx_encoder.h
+vpx/vpx_image.h
+vpx/vpx_integer.h
diff --git a/armv7a-neon/vpx_config.c b/armv7a-neon/vpx_config.c
new file mode 100644
index 0000000..7bb8e40
--- /dev/null
+++ b/armv7a-neon/vpx_config.c
@@ -0,0 +1,9 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+static const char* const cfg = "--force-target=armv7-android-gcc --disable-runtime-cpu-detect --sdk-path=/usr/local/google/home/johannkoenig/android-ndk --disable-vp8-encoder --disable-examples --disable-docs";
+const char *vpx_codec_build_config(void) {return cfg;}
diff --git a/armv7a-neon/vpx_config.h b/armv7a-neon/vpx_config.h
new file mode 100644
index 0000000..fdea8f0
--- /dev/null
+++ b/armv7a-neon/vpx_config.h
@@ -0,0 +1,80 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+/* This file automatically generated by configure. Do not edit! */
+#ifndef VPX_CONFIG_H
+#define VPX_CONFIG_H
+#define RESTRICT
+#define ARCH_ARM 1
+#define ARCH_MIPS 0
+#define ARCH_X86 0
+#define ARCH_X86_64 0
+#define ARCH_PPC32 0
+#define ARCH_PPC64 0
+#define HAVE_EDSP 1
+#define HAVE_MEDIA 1
+#define HAVE_NEON 1
+#define HAVE_MIPS32 0
+#define HAVE_DSPR2 0
+#define HAVE_MMX 0
+#define HAVE_SSE 0
+#define HAVE_SSE2 0
+#define HAVE_SSE3 0
+#define HAVE_SSSE3 0
+#define HAVE_SSE4_1 0
+#define HAVE_ALTIVEC 0
+#define HAVE_VPX_PORTS 1
+#define HAVE_STDINT_H 1
+#define HAVE_ALT_TREE_LAYOUT 0
+#define HAVE_PTHREAD_H 1
+#define HAVE_SYS_MMAN_H 1
+#define HAVE_UNISTD_H 1
+#define CONFIG_EXTERNAL_BUILD 0
+#define CONFIG_INSTALL_DOCS 1
+#define CONFIG_INSTALL_BINS 1
+#define CONFIG_INSTALL_LIBS 1
+#define CONFIG_INSTALL_SRCS 0
+#define CONFIG_DEBUG 0
+#define CONFIG_GPROF 0
+#define CONFIG_GCOV 0
+#define CONFIG_RVCT 0
+#define CONFIG_GCC 1
+#define CONFIG_MSVS 0
+#define CONFIG_PIC 1
+#define CONFIG_BIG_ENDIAN 0
+#define CONFIG_CODEC_SRCS 0
+#define CONFIG_DEBUG_LIBS 0
+#define CONFIG_FAST_UNALIGNED 1
+#define CONFIG_MEM_MANAGER 0
+#define CONFIG_MEM_TRACKER 0
+#define CONFIG_MEM_CHECKS 0
+#define CONFIG_MD5 1
+#define CONFIG_DEQUANT_TOKENS 0
+#define CONFIG_DC_RECON 0
+#define CONFIG_RUNTIME_CPU_DETECT 0
+#define CONFIG_POSTPROC 0
+#define CONFIG_MULTITHREAD 1
+#define CONFIG_INTERNAL_STATS 0
+#define CONFIG_VP8_ENCODER 0
+#define CONFIG_VP8_DECODER 1
+#define CONFIG_VP8 1
+#define CONFIG_ENCODERS 0
+#define CONFIG_DECODERS 1
+#define CONFIG_STATIC_MSVCRT 0
+#define CONFIG_SPATIAL_RESAMPLING 1
+#define CONFIG_REALTIME_ONLY 1
+#define CONFIG_ONTHEFLY_BITPACKING 0
+#define CONFIG_ERROR_CONCEALMENT 0
+#define CONFIG_SHARED 0
+#define CONFIG_STATIC 1
+#define CONFIG_SMALL 0
+#define CONFIG_POSTPROC_VISUALIZER 0
+#define CONFIG_OS_SUPPORT 1
+#define CONFIG_UNIT_TESTS 0
+#define CONFIG_MULTI_RES_ENCODING 0
+#define CONFIG_TEMPORAL_DENOISING 1
+#endif /* VPX_CONFIG_H */
diff --git a/armv7a-neon/vpx_rtcd.h b/armv7a-neon/vpx_rtcd.h
new file mode 100644
index 0000000..2aef442
--- /dev/null
+++ b/armv7a-neon/vpx_rtcd.h
@@ -0,0 +1,379 @@
+#ifndef VPX_RTCD_
+#define VPX_RTCD_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+#include "vp8/common/blockd.h"
+
+struct blockd;
+struct macroblockd;
+struct loop_filter_info;
+
+/* Encoder forward decls */
+struct block;
+struct macroblock;
+struct variance_vtable;
+union int_mv;
+struct yv12_buffer_config;
+
+void vp8_dequantize_b_c(struct blockd*, short *dqc);
+void vp8_dequantize_b_v6(struct blockd*, short *dqc);
+void vp8_dequantize_b_neon(struct blockd*, short *dqc);
+#define vp8_dequantize_b vp8_dequantize_b_neon
+
+void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *output, int stride);
+void vp8_dequant_idct_add_v6(short *input, short *dq, unsigned char *output, int stride);
+void vp8_dequant_idct_add_neon(short *input, short *dq, unsigned char *output, int stride);
+#define vp8_dequant_idct_add vp8_dequant_idct_add_neon
+
+void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+void vp8_dequant_idct_add_y_block_v6(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+void vp8_dequant_idct_add_y_block_neon(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_neon
+
+void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+void vp8_dequant_idct_add_uv_block_v6(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+void vp8_dequant_idct_add_uv_block_neon(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_neon
+
+void vp8_loop_filter_mbv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbv_armv6(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbv_neon(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbv vp8_loop_filter_mbv_neon
+
+void vp8_loop_filter_bv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bv_armv6(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bv_neon(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bv vp8_loop_filter_bv_neon
+
+void vp8_loop_filter_mbh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbh_armv6(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbh_neon(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbh vp8_loop_filter_mbh_neon
+
+void vp8_loop_filter_bh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bh_armv6(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bh_neon(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bh vp8_loop_filter_bh_neon
+
+void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_simple_vertical_edge_armv6(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_mbvs_neon(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbv vp8_loop_filter_mbvs_neon
+
+void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_simple_horizontal_edge_armv6(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_mbhs_neon(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbh vp8_loop_filter_mbhs_neon
+
+void vp8_loop_filter_bvs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bvs_armv6(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bvs_neon(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bv vp8_loop_filter_bvs_neon
+
+void vp8_loop_filter_bhs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bhs_armv6(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bhs_neon(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bh vp8_loop_filter_bhs_neon
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+void vp8_short_idct4x4llm_v6_dual(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+void vp8_short_idct4x4llm_neon(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+#define vp8_short_idct4x4llm vp8_short_idct4x4llm_neon
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_c
+
+void vp8_short_inv_walsh4x4_c(short *input, short *output);
+void vp8_short_inv_walsh4x4_v6(short *input, short *output);
+void vp8_short_inv_walsh4x4_neon(short *input, short *output);
+#define vp8_short_inv_walsh4x4 vp8_short_inv_walsh4x4_neon
+
+void vp8_dc_only_idct_add_c(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+void vp8_dc_only_idct_add_v6(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+void vp8_dc_only_idct_add_neon(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+#define vp8_dc_only_idct_add vp8_dc_only_idct_add_neon
+
+void vp8_copy_mem16x16_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem16x16_v6(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem16x16_neon(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem16x16 vp8_copy_mem16x16_neon
+
+void vp8_copy_mem8x8_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x8_v6(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x8_neon(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x8 vp8_copy_mem8x8_neon
+
+void vp8_copy_mem8x4_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x4_v6(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x4_neon(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x4 vp8_copy_mem8x4_neon
+
+void vp8_build_intra_predictors_mby_s_c(struct macroblockd *x, unsigned char * yabove_row, unsigned char * yleft, int left_stride, unsigned char * ypred_ptr, int y_stride);
+#define vp8_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s_c
+
+void vp8_build_intra_predictors_mbuv_s_c(struct macroblockd *x, unsigned char * uabove_row, unsigned char * vabove_row, unsigned char *uleft, unsigned char *vleft, int left_stride, unsigned char * upred_ptr, unsigned char * vpred_ptr, int pred_stride);
+#define vp8_build_intra_predictors_mbuv_s vp8_build_intra_predictors_mbuv_s_c
+
+void vp8_intra4x4_predict_c(unsigned char *Above, unsigned char *yleft, int left_stride, B_PREDICTION_MODE b_mode, unsigned char *dst, int dst_stride, unsigned char top_left);
+void vp8_intra4x4_predict_armv6(unsigned char *Above, unsigned char *yleft, int left_stride, B_PREDICTION_MODE b_mode, unsigned char *dst, int dst_stride, unsigned char top_left);
+#define vp8_intra4x4_predict vp8_intra4x4_predict_armv6
+
+void vp8_sixtap_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict16x16_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict16x16_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict16x16 vp8_sixtap_predict16x16_neon
+
+void vp8_sixtap_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x8_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x8_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x8 vp8_sixtap_predict8x8_neon
+
+void vp8_sixtap_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x4_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x4_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x4 vp8_sixtap_predict8x4_neon
+
+void vp8_sixtap_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict4x4_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict4x4_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict4x4 vp8_sixtap_predict4x4_neon
+
+void vp8_bilinear_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict16x16_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict16x16_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict16x16 vp8_bilinear_predict16x16_neon
+
+void vp8_bilinear_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x8_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x8_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x8 vp8_bilinear_predict8x8_neon
+
+void vp8_bilinear_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x4_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x4_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x4 vp8_bilinear_predict8x4_neon
+
+void vp8_bilinear_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict4x4_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict4x4_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict4x4 vp8_bilinear_predict4x4_neon
+
+unsigned int vp8_variance4x4_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance4x4 vp8_variance4x4_c
+
+unsigned int vp8_variance8x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance8x8_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance8x8_neon(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x8 vp8_variance8x8_neon
+
+unsigned int vp8_variance8x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance8x16_neon(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x16 vp8_variance8x16_neon
+
+unsigned int vp8_variance16x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance16x8_neon(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x8 vp8_variance16x8_neon
+
+unsigned int vp8_variance16x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance16x16_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance16x16_neon(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x16 vp8_variance16x16_neon
+
+unsigned int vp8_sub_pixel_variance4x4_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance4x4 vp8_sub_pixel_variance4x4_c
+
+unsigned int vp8_sub_pixel_variance8x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+unsigned int vp8_sub_pixel_variance8x8_armv6(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+unsigned int vp8_sub_pixel_variance8x8_neon(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x8 vp8_sub_pixel_variance8x8_neon
+
+unsigned int vp8_sub_pixel_variance8x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x16 vp8_sub_pixel_variance8x16_c
+
+unsigned int vp8_sub_pixel_variance16x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x8 vp8_sub_pixel_variance16x8_c
+
+unsigned int vp8_sub_pixel_variance16x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+unsigned int vp8_sub_pixel_variance16x16_armv6(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+unsigned int vp8_sub_pixel_variance16x16_neon(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x16 vp8_sub_pixel_variance16x16_neon
+
+unsigned int vp8_variance_halfpixvar16x16_h_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_h_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_h_neon(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_neon
+
+unsigned int vp8_variance_halfpixvar16x16_v_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_v_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_v_neon(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_neon
+
+unsigned int vp8_variance_halfpixvar16x16_hv_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_hv_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_hv_neon(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_neon
+
+unsigned int vp8_sad4x4_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+unsigned int vp8_sad4x4_neon(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad4x4 vp8_sad4x4_neon
+
+unsigned int vp8_sad8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+unsigned int vp8_sad8x8_neon(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x8 vp8_sad8x8_neon
+
+unsigned int vp8_sad8x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+unsigned int vp8_sad8x16_neon(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x16 vp8_sad8x16_neon
+
+unsigned int vp8_sad16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+unsigned int vp8_sad16x8_neon(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x8 vp8_sad16x8_neon
+
+unsigned int vp8_sad16x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+unsigned int vp8_sad16x16_armv6(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+unsigned int vp8_sad16x16_neon(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x16 vp8_sad16x16_neon
+
+void vp8_sad4x4x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x3 vp8_sad4x4x3_c
+
+void vp8_sad8x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x3 vp8_sad8x8x3_c
+
+void vp8_sad8x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x3 vp8_sad8x16x3_c
+
+void vp8_sad16x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x3 vp8_sad16x8x3_c
+
+void vp8_sad16x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x3 vp8_sad16x16x3_c
+
+void vp8_sad4x4x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad4x4x8 vp8_sad4x4x8_c
+
+void vp8_sad8x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x8x8 vp8_sad8x8x8_c
+
+void vp8_sad8x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x16x8 vp8_sad8x16x8_c
+
+void vp8_sad16x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x8x8 vp8_sad16x8x8_c
+
+void vp8_sad16x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x16x8 vp8_sad16x16x8_c
+
+void vp8_sad4x4x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x4d vp8_sad4x4x4d_c
+
+void vp8_sad8x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x4d vp8_sad8x8x4d_c
+
+void vp8_sad8x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x4d vp8_sad8x16x4d_c
+
+void vp8_sad16x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x4d vp8_sad16x8x4d_c
+
+void vp8_sad16x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x4d vp8_sad16x16x4d_c
+
+void vp8_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_4_5_scale vp8_horizontal_line_4_5_scale_c
+
+void vp8_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_4_5_scale vp8_vertical_band_4_5_scale_c
+
+void vp8_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_4_5_scale vp8_last_vertical_band_4_5_scale_c
+
+void vp8_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_3_scale vp8_horizontal_line_2_3_scale_c
+
+void vp8_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_3_scale vp8_vertical_band_2_3_scale_c
+
+void vp8_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_2_3_scale vp8_last_vertical_band_2_3_scale_c
+
+void vp8_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_5_scale vp8_horizontal_line_3_5_scale_c
+
+void vp8_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_5_scale vp8_vertical_band_3_5_scale_c
+
+void vp8_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_5_scale vp8_last_vertical_band_3_5_scale_c
+
+void vp8_horizontal_line_3_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_4_scale vp8_horizontal_line_3_4_scale_c
+
+void vp8_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_4_scale vp8_vertical_band_3_4_scale_c
+
+void vp8_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_4_scale vp8_last_vertical_band_3_4_scale_c
+
+void vp8_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_1_2_scale vp8_horizontal_line_1_2_scale_c
+
+void vp8_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_1_2_scale vp8_vertical_band_1_2_scale_c
+
+void vp8_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_1_2_scale vp8_last_vertical_band_1_2_scale_c
+
+void vp8_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_4_scale vp8_horizontal_line_5_4_scale_c
+
+void vp8_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_4_scale vp8_vertical_band_5_4_scale_c
+
+void vp8_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_3_scale vp8_horizontal_line_5_3_scale_c
+
+void vp8_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_3_scale vp8_vertical_band_5_3_scale_c
+
+void vp8_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_1_scale vp8_horizontal_line_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale vp8_vertical_band_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale_i vp8_vertical_band_2_1_scale_i_c
+
+void vp8_yv12_extend_frame_borders_c(struct yv12_buffer_config *ybf);
+void vp8_yv12_extend_frame_borders_neon(struct yv12_buffer_config *ybf);
+#define vp8_yv12_extend_frame_borders vp8_yv12_extend_frame_borders_neon
+
+void vp8_yv12_copy_frame_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+void vp8_yv12_copy_frame_neon(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_frame vp8_yv12_copy_frame_neon
+
+void vp8_yv12_copy_y_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+void vp8_yv12_copy_y_neon(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_y vp8_yv12_copy_y_neon
+
+void vpx_rtcd(void);
+#include "vpx_config.h"
+
+#ifdef RTCD_C
+#include "vpx_ports/arm.h"
+static void setup_rtcd_internal(void)
+{
+ int flags = arm_cpu_caps();
+
+ (void)flags;
+
+
+}
+#endif
+#endif
diff --git a/armv7a-neon/vpx_version.h b/armv7a-neon/vpx_version.h
new file mode 100644
index 0000000..663dd49
--- /dev/null
+++ b/armv7a-neon/vpx_version.h
@@ -0,0 +1,7 @@
+#define VERSION_MAJOR 1
+#define VERSION_MINOR 1
+#define VERSION_PATCH 0
+#define VERSION_EXTRA ""
+#define VERSION_PACKED ((VERSION_MAJOR<<16)|(VERSION_MINOR<<8)|(VERSION_PATCH))
+#define VERSION_STRING_NOSP "v1.1.0"
+#define VERSION_STRING " v1.1.0"
diff --git a/armv7a/libvpx_srcs.txt b/armv7a/libvpx_srcs.txt
new file mode 100644
index 0000000..cd05158
--- /dev/null
+++ b/armv7a/libvpx_srcs.txt
@@ -0,0 +1,143 @@
+build/make/rtcd.sh
+build/make/version.sh
+CHANGELOG
+libs.mk
+vp8/common/alloccommon.c
+vp8/common/alloccommon.h
+vp8/common/arm/armv6/bilinearfilter_v6.asm.s
+vp8/common/arm/armv6/copymem16x16_v6.asm.s
+vp8/common/arm/armv6/copymem8x4_v6.asm.s
+vp8/common/arm/armv6/copymem8x8_v6.asm.s
+vp8/common/arm/armv6/dc_only_idct_add_v6.asm.s
+vp8/common/arm/armv6/dequant_idct_v6.asm.s
+vp8/common/arm/armv6/dequantize_v6.asm.s
+vp8/common/arm/armv6/filter_v6.asm.s
+vp8/common/arm/armv6/idct_blk_v6.c
+vp8/common/arm/armv6/idct_v6.asm.s
+vp8/common/arm/armv6/intra4x4_predict_v6.asm.s
+vp8/common/arm/armv6/iwalsh_v6.asm.s
+vp8/common/arm/armv6/loopfilter_v6.asm.s
+vp8/common/arm/armv6/simpleloopfilter_v6.asm.s
+vp8/common/arm/armv6/sixtappredict8x4_v6.asm.s
+vp8/common/arm/armv6/vp8_sad16x16_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance16x16_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance8x8_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6.asm.s
+vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6.asm.s
+vp8/common/arm/bilinearfilter_arm.c
+vp8/common/arm/bilinearfilter_arm.h
+vp8/common/arm/dequantize_arm.c
+vp8/common/arm/filter_arm.c
+vp8/common/arm/loopfilter_arm.c
+vp8/common/arm/reconintra_arm.c
+vp8/common/arm/variance_arm.c
+vp8/common/asm_com_offsets.c
+vp8/common/blockd.c
+vp8/common/blockd.h
+vp8/common/coefupdateprobs.h
+vp8/common/common.h
+vp8/common/debugmodes.c
+vp8/common/default_coef_probs.h
+vp8/common/dequantize.c
+vp8/common/entropy.c
+vp8/common/entropy.h
+vp8/common/entropymode.c
+vp8/common/entropymode.h
+vp8/common/entropymv.c
+vp8/common/entropymv.h
+vp8/common/extend.c
+vp8/common/extend.h
+vp8/common/filter.c
+vp8/common/filter.h
+vp8/common/findnearmv.c
+vp8/common/findnearmv.h
+vp8/common/generic/systemdependent.c
+vp8/common/header.h
+vp8/common/idct_blk.c
+vp8/common/idctllm.c
+vp8/common/invtrans.h
+vp8/common/loopfilter.c
+vp8/common/loopfilter_filters.c
+vp8/common/loopfilter.h
+vp8/common/mbpitch.c
+vp8/common/modecont.c
+vp8/common/modecont.h
+vp8/common/mv.h
+vp8/common/onyxc_int.h
+vp8/common/onyxd.h
+vp8/common/onyx.h
+vp8/common/ppflags.h
+vp8/common/pragmas.h
+vp8/common/quant_common.c
+vp8/common/quant_common.h
+vp8/common/reconinter.c
+vp8/common/reconinter.h
+vp8/common/reconintra4x4.c
+vp8/common/reconintra4x4.h
+vp8/common/reconintra.c
+vp8/common/rtcd.c
+vp8/common/rtcd_defs.sh
+vp8/common/sad_c.c
+vp8/common/setupintrarecon.c
+vp8/common/setupintrarecon.h
+vp8/common/swapyv12buffer.c
+vp8/common/swapyv12buffer.h
+vp8/common/systemdependent.h
+vp8/common/threading.h
+vp8/common/treecoder.c
+vp8/common/treecoder.h
+vp8/common/variance_c.c
+vp8/common/variance.h
+vp8/common/vp8_entropymodedata.h
+vp8/decoder/asm_dec_offsets.c
+vp8/decoder/dboolhuff.c
+vp8/decoder/dboolhuff.h
+vp8/decoder/decodemv.c
+vp8/decoder/decodemv.h
+vp8/decoder/decoderthreading.h
+vp8/decoder/decodframe.c
+vp8/decoder/detokenize.c
+vp8/decoder/detokenize.h
+vp8/decoder/onyxd_if.c
+vp8/decoder/onyxd_int.h
+vp8/decoder/threading.c
+vp8/decoder/treereader.h
+vp8/vp8_common.mk
+vp8/vp8_dx_iface.c
+vp8/vp8dx.mk
+vpx_config.c
+vpx/internal/vpx_codec_internal.h
+vpx_mem/include/vpx_mem_intrnl.h
+vpx_mem/vpx_mem.c
+vpx_mem/vpx_mem.h
+vpx_mem/vpx_mem.mk
+vpx_ports/arm_cpudetect.c
+vpx_ports/arm.h
+vpx_ports/asm_offsets.h
+vpx_ports/mem.h
+vpx_ports/vpx_ports.mk
+vpx_ports/vpx_timer.h
+vpx_scale/generic/gen_scalers.c
+vpx_scale/generic/vpxscale.c
+vpx_scale/generic/yv12config.c
+vpx_scale/generic/yv12extend.c
+vpx_scale/generic/yv12extend_generic.h
+vpx_scale/scale_mode.h
+vpx_scale/vpxscale.h
+vpx_scale/vpx_scale.mk
+vpx_scale/yv12config.h
+vpx/src/vpx_codec.c
+vpx/src/vpx_decoder.c
+vpx/src/vpx_encoder.c
+vpx/src/vpx_image.c
+vpx/vp8dx.h
+vpx/vp8.h
+vpx/vpx_codec.h
+vpx/vpx_codec_impl_bottom.h
+vpx/vpx_codec_impl_top.h
+vpx/vpx_codec.mk
+vpx/vpx_decoder.h
+vpx/vpx_encoder.h
+vpx/vpx_image.h
+vpx/vpx_integer.h
diff --git a/armv7a/vpx_config.c b/armv7a/vpx_config.c
new file mode 100644
index 0000000..9ff4335
--- /dev/null
+++ b/armv7a/vpx_config.c
@@ -0,0 +1,9 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+static const char* const cfg = "--force-target=armv7-android-gcc --disable-runtime-cpu-detect --sdk-path=/usr/local/google/home/johannkoenig/android-ndk --disable-vp8-encoder --disable-examples --disable-docs --disable-neon";
+const char *vpx_codec_build_config(void) {return cfg;}
diff --git a/armv7a/vpx_config.h b/armv7a/vpx_config.h
new file mode 100644
index 0000000..84b75c2
--- /dev/null
+++ b/armv7a/vpx_config.h
@@ -0,0 +1,80 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+/* This file automatically generated by configure. Do not edit! */
+#ifndef VPX_CONFIG_H
+#define VPX_CONFIG_H
+#define RESTRICT
+#define ARCH_ARM 1
+#define ARCH_MIPS 0
+#define ARCH_X86 0
+#define ARCH_X86_64 0
+#define ARCH_PPC32 0
+#define ARCH_PPC64 0
+#define HAVE_EDSP 1
+#define HAVE_MEDIA 1
+#define HAVE_NEON 0
+#define HAVE_MIPS32 0
+#define HAVE_DSPR2 0
+#define HAVE_MMX 0
+#define HAVE_SSE 0
+#define HAVE_SSE2 0
+#define HAVE_SSE3 0
+#define HAVE_SSSE3 0
+#define HAVE_SSE4_1 0
+#define HAVE_ALTIVEC 0
+#define HAVE_VPX_PORTS 1
+#define HAVE_STDINT_H 1
+#define HAVE_ALT_TREE_LAYOUT 0
+#define HAVE_PTHREAD_H 1
+#define HAVE_SYS_MMAN_H 1
+#define HAVE_UNISTD_H 1
+#define CONFIG_EXTERNAL_BUILD 0
+#define CONFIG_INSTALL_DOCS 1
+#define CONFIG_INSTALL_BINS 1
+#define CONFIG_INSTALL_LIBS 1
+#define CONFIG_INSTALL_SRCS 0
+#define CONFIG_DEBUG 0
+#define CONFIG_GPROF 0
+#define CONFIG_GCOV 0
+#define CONFIG_RVCT 0
+#define CONFIG_GCC 1
+#define CONFIG_MSVS 0
+#define CONFIG_PIC 1
+#define CONFIG_BIG_ENDIAN 0
+#define CONFIG_CODEC_SRCS 0
+#define CONFIG_DEBUG_LIBS 0
+#define CONFIG_FAST_UNALIGNED 1
+#define CONFIG_MEM_MANAGER 0
+#define CONFIG_MEM_TRACKER 0
+#define CONFIG_MEM_CHECKS 0
+#define CONFIG_MD5 1
+#define CONFIG_DEQUANT_TOKENS 0
+#define CONFIG_DC_RECON 0
+#define CONFIG_RUNTIME_CPU_DETECT 0
+#define CONFIG_POSTPROC 0
+#define CONFIG_MULTITHREAD 1
+#define CONFIG_INTERNAL_STATS 0
+#define CONFIG_VP8_ENCODER 0
+#define CONFIG_VP8_DECODER 1
+#define CONFIG_VP8 1
+#define CONFIG_ENCODERS 0
+#define CONFIG_DECODERS 1
+#define CONFIG_STATIC_MSVCRT 0
+#define CONFIG_SPATIAL_RESAMPLING 1
+#define CONFIG_REALTIME_ONLY 1
+#define CONFIG_ONTHEFLY_BITPACKING 0
+#define CONFIG_ERROR_CONCEALMENT 0
+#define CONFIG_SHARED 0
+#define CONFIG_STATIC 1
+#define CONFIG_SMALL 0
+#define CONFIG_POSTPROC_VISUALIZER 0
+#define CONFIG_OS_SUPPORT 1
+#define CONFIG_UNIT_TESTS 0
+#define CONFIG_MULTI_RES_ENCODING 0
+#define CONFIG_TEMPORAL_DENOISING 1
+#endif /* VPX_CONFIG_H */
diff --git a/armv7a/vpx_rtcd.h b/armv7a/vpx_rtcd.h
new file mode 100644
index 0000000..f9b97ca
--- /dev/null
+++ b/armv7a/vpx_rtcd.h
@@ -0,0 +1,336 @@
+#ifndef VPX_RTCD_
+#define VPX_RTCD_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+#include "vp8/common/blockd.h"
+
+struct blockd;
+struct macroblockd;
+struct loop_filter_info;
+
+/* Encoder forward decls */
+struct block;
+struct macroblock;
+struct variance_vtable;
+union int_mv;
+struct yv12_buffer_config;
+
+void vp8_dequantize_b_c(struct blockd*, short *dqc);
+void vp8_dequantize_b_v6(struct blockd*, short *dqc);
+#define vp8_dequantize_b vp8_dequantize_b_v6
+
+void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *output, int stride);
+void vp8_dequant_idct_add_v6(short *input, short *dq, unsigned char *output, int stride);
+#define vp8_dequant_idct_add vp8_dequant_idct_add_v6
+
+void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+void vp8_dequant_idct_add_y_block_v6(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_v6
+
+void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+void vp8_dequant_idct_add_uv_block_v6(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_v6
+
+void vp8_loop_filter_mbv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbv_armv6(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbv vp8_loop_filter_mbv_armv6
+
+void vp8_loop_filter_bv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bv_armv6(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bv vp8_loop_filter_bv_armv6
+
+void vp8_loop_filter_mbh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbh_armv6(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbh vp8_loop_filter_mbh_armv6
+
+void vp8_loop_filter_bh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bh_armv6(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bh vp8_loop_filter_bh_armv6
+
+void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_simple_vertical_edge_armv6(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbv vp8_loop_filter_simple_vertical_edge_armv6
+
+void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_simple_horizontal_edge_armv6(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbh vp8_loop_filter_simple_horizontal_edge_armv6
+
+void vp8_loop_filter_bvs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bvs_armv6(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bv vp8_loop_filter_bvs_armv6
+
+void vp8_loop_filter_bhs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bhs_armv6(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bh vp8_loop_filter_bhs_armv6
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+void vp8_short_idct4x4llm_v6_dual(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+#define vp8_short_idct4x4llm vp8_short_idct4x4llm_v6_dual
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_c
+
+void vp8_short_inv_walsh4x4_c(short *input, short *output);
+void vp8_short_inv_walsh4x4_v6(short *input, short *output);
+#define vp8_short_inv_walsh4x4 vp8_short_inv_walsh4x4_v6
+
+void vp8_dc_only_idct_add_c(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+void vp8_dc_only_idct_add_v6(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+#define vp8_dc_only_idct_add vp8_dc_only_idct_add_v6
+
+void vp8_copy_mem16x16_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem16x16_v6(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem16x16 vp8_copy_mem16x16_v6
+
+void vp8_copy_mem8x8_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x8_v6(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x8 vp8_copy_mem8x8_v6
+
+void vp8_copy_mem8x4_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x4_v6(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x4 vp8_copy_mem8x4_v6
+
+void vp8_build_intra_predictors_mby_s_c(struct macroblockd *x, unsigned char * yabove_row, unsigned char * yleft, int left_stride, unsigned char * ypred_ptr, int y_stride);
+#define vp8_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s_c
+
+void vp8_build_intra_predictors_mbuv_s_c(struct macroblockd *x, unsigned char * uabove_row, unsigned char * vabove_row, unsigned char *uleft, unsigned char *vleft, int left_stride, unsigned char * upred_ptr, unsigned char * vpred_ptr, int pred_stride);
+#define vp8_build_intra_predictors_mbuv_s vp8_build_intra_predictors_mbuv_s_c
+
+void vp8_intra4x4_predict_c(unsigned char *Above, unsigned char *yleft, int left_stride, B_PREDICTION_MODE b_mode, unsigned char *dst, int dst_stride, unsigned char top_left);
+void vp8_intra4x4_predict_armv6(unsigned char *Above, unsigned char *yleft, int left_stride, B_PREDICTION_MODE b_mode, unsigned char *dst, int dst_stride, unsigned char top_left);
+#define vp8_intra4x4_predict vp8_intra4x4_predict_armv6
+
+void vp8_sixtap_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict16x16_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict16x16 vp8_sixtap_predict16x16_armv6
+
+void vp8_sixtap_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x8_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x8 vp8_sixtap_predict8x8_armv6
+
+void vp8_sixtap_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x4_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x4 vp8_sixtap_predict8x4_armv6
+
+void vp8_sixtap_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict4x4_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict4x4 vp8_sixtap_predict4x4_armv6
+
+void vp8_bilinear_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict16x16_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict16x16 vp8_bilinear_predict16x16_armv6
+
+void vp8_bilinear_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x8_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x8 vp8_bilinear_predict8x8_armv6
+
+void vp8_bilinear_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x4_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x4 vp8_bilinear_predict8x4_armv6
+
+void vp8_bilinear_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict4x4_armv6(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict4x4 vp8_bilinear_predict4x4_armv6
+
+unsigned int vp8_variance4x4_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance4x4 vp8_variance4x4_c
+
+unsigned int vp8_variance8x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance8x8_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x8 vp8_variance8x8_armv6
+
+unsigned int vp8_variance8x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x16 vp8_variance8x16_c
+
+unsigned int vp8_variance16x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x8 vp8_variance16x8_c
+
+unsigned int vp8_variance16x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance16x16_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x16 vp8_variance16x16_armv6
+
+unsigned int vp8_sub_pixel_variance4x4_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance4x4 vp8_sub_pixel_variance4x4_c
+
+unsigned int vp8_sub_pixel_variance8x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+unsigned int vp8_sub_pixel_variance8x8_armv6(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x8 vp8_sub_pixel_variance8x8_armv6
+
+unsigned int vp8_sub_pixel_variance8x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x16 vp8_sub_pixel_variance8x16_c
+
+unsigned int vp8_sub_pixel_variance16x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x8 vp8_sub_pixel_variance16x8_c
+
+unsigned int vp8_sub_pixel_variance16x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+unsigned int vp8_sub_pixel_variance16x16_armv6(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x16 vp8_sub_pixel_variance16x16_armv6
+
+unsigned int vp8_variance_halfpixvar16x16_h_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_h_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_armv6
+
+unsigned int vp8_variance_halfpixvar16x16_v_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_v_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_armv6
+
+unsigned int vp8_variance_halfpixvar16x16_hv_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+unsigned int vp8_variance_halfpixvar16x16_hv_armv6(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_armv6
+
+unsigned int vp8_sad4x4_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad4x4 vp8_sad4x4_c
+
+unsigned int vp8_sad8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x8 vp8_sad8x8_c
+
+unsigned int vp8_sad8x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x16 vp8_sad8x16_c
+
+unsigned int vp8_sad16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x8 vp8_sad16x8_c
+
+unsigned int vp8_sad16x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+unsigned int vp8_sad16x16_armv6(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x16 vp8_sad16x16_armv6
+
+void vp8_sad4x4x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x3 vp8_sad4x4x3_c
+
+void vp8_sad8x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x3 vp8_sad8x8x3_c
+
+void vp8_sad8x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x3 vp8_sad8x16x3_c
+
+void vp8_sad16x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x3 vp8_sad16x8x3_c
+
+void vp8_sad16x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x3 vp8_sad16x16x3_c
+
+void vp8_sad4x4x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad4x4x8 vp8_sad4x4x8_c
+
+void vp8_sad8x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x8x8 vp8_sad8x8x8_c
+
+void vp8_sad8x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x16x8 vp8_sad8x16x8_c
+
+void vp8_sad16x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x8x8 vp8_sad16x8x8_c
+
+void vp8_sad16x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x16x8 vp8_sad16x16x8_c
+
+void vp8_sad4x4x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x4d vp8_sad4x4x4d_c
+
+void vp8_sad8x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x4d vp8_sad8x8x4d_c
+
+void vp8_sad8x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x4d vp8_sad8x16x4d_c
+
+void vp8_sad16x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x4d vp8_sad16x8x4d_c
+
+void vp8_sad16x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x4d vp8_sad16x16x4d_c
+
+void vp8_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_4_5_scale vp8_horizontal_line_4_5_scale_c
+
+void vp8_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_4_5_scale vp8_vertical_band_4_5_scale_c
+
+void vp8_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_4_5_scale vp8_last_vertical_band_4_5_scale_c
+
+void vp8_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_3_scale vp8_horizontal_line_2_3_scale_c
+
+void vp8_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_3_scale vp8_vertical_band_2_3_scale_c
+
+void vp8_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_2_3_scale vp8_last_vertical_band_2_3_scale_c
+
+void vp8_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_5_scale vp8_horizontal_line_3_5_scale_c
+
+void vp8_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_5_scale vp8_vertical_band_3_5_scale_c
+
+void vp8_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_5_scale vp8_last_vertical_band_3_5_scale_c
+
+void vp8_horizontal_line_3_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_4_scale vp8_horizontal_line_3_4_scale_c
+
+void vp8_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_4_scale vp8_vertical_band_3_4_scale_c
+
+void vp8_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_4_scale vp8_last_vertical_band_3_4_scale_c
+
+void vp8_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_1_2_scale vp8_horizontal_line_1_2_scale_c
+
+void vp8_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_1_2_scale vp8_vertical_band_1_2_scale_c
+
+void vp8_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_1_2_scale vp8_last_vertical_band_1_2_scale_c
+
+void vp8_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_4_scale vp8_horizontal_line_5_4_scale_c
+
+void vp8_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_4_scale vp8_vertical_band_5_4_scale_c
+
+void vp8_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_3_scale vp8_horizontal_line_5_3_scale_c
+
+void vp8_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_3_scale vp8_vertical_band_5_3_scale_c
+
+void vp8_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_1_scale vp8_horizontal_line_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale vp8_vertical_band_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale_i vp8_vertical_band_2_1_scale_i_c
+
+void vp8_yv12_extend_frame_borders_c(struct yv12_buffer_config *ybf);
+#define vp8_yv12_extend_frame_borders vp8_yv12_extend_frame_borders_c
+
+void vp8_yv12_copy_frame_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_frame vp8_yv12_copy_frame_c
+
+void vp8_yv12_copy_y_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_y vp8_yv12_copy_y_c
+
+void vpx_rtcd(void);
+#include "vpx_config.h"
+
+#ifdef RTCD_C
+#include "vpx_ports/arm.h"
+static void setup_rtcd_internal(void)
+{
+ int flags = arm_cpu_caps();
+
+ (void)flags;
+
+
+}
+#endif
+#endif
diff --git a/armv7a/vpx_version.h b/armv7a/vpx_version.h
new file mode 100644
index 0000000..663dd49
--- /dev/null
+++ b/armv7a/vpx_version.h
@@ -0,0 +1,7 @@
+#define VERSION_MAJOR 1
+#define VERSION_MINOR 1
+#define VERSION_PATCH 0
+#define VERSION_EXTRA ""
+#define VERSION_PACKED ((VERSION_MAJOR<<16)|(VERSION_MINOR<<8)|(VERSION_PATCH))
+#define VERSION_STRING_NOSP "v1.1.0"
+#define VERSION_STRING " v1.1.0"
diff --git a/build/make/Makefile b/build/make/Makefile
deleted file mode 100755
index 64d3c93..0000000
--- a/build/make/Makefile
+++ /dev/null
@@ -1,357 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-include config.mk
-quiet?=true
-ifeq ($(target),)
-# If a target wasn't specified, invoke for all enabled targets.
-.DEFAULT:
- @for t in $(ALL_TARGETS); do \
- $(MAKE) --no-print-directory target=$$t $(MAKECMDGOALS) || exit $$?;\
- done
-all: .DEFAULT
-clean:: .DEFAULT
-install:: .DEFAULT
-
-
-# Note: md5sum is not installed on OS X, but openssl is. Openssl may not be
-# installed on cygwin, so we need to autodetect here.
-md5sum := $(firstword $(wildcard \
- $(foreach e,md5sum openssl,\
- $(foreach p,$(subst :, ,$(PATH)),$(p)/$(e)*))\
- ))
-md5sum := $(if $(filter %openssl,$(md5sum)),$(md5sum) dgst -md5,$(md5sum))
-
-TGT_CC:=$(word 3, $(subst -, ,$(TOOLCHAIN)))
-dist:
- @for t in $(ALL_TARGETS); do \
- $(MAKE) --no-print-directory target=$$t $(MAKECMDGOALS) || exit $$?;\
- done
- # Run configure for the user with the current toolchain.
- @if [ -d "$(DIST_DIR)/src" ]; then \
- mkdir -p "$(DIST_DIR)/build"; \
- cd "$(DIST_DIR)/build"; \
- echo "Rerunning configure $(CONFIGURE_ARGS)"; \
- ../src/configure $(CONFIGURE_ARGS); \
- $(if $(filter vs%,$(TGT_CC)),make NO_LAUNCH_DEVENV=1;) \
- fi
- @if [ -d "$(DIST_DIR)" ]; then \
- echo " [MD5SUM] $(DIST_DIR)"; \
- cd $(DIST_DIR) && \
- $(md5sum) `find . -name md5sums.txt -prune -o -type f -print` \
- | sed -e 's/MD5(\(.*\))= \([0-9a-f]\{32\}\)/\2 \1/' \
- > md5sums.txt;\
- fi
-
-
-endif
-
-ifneq ($(target),)
-# Normally, we want to build the filename from the target and the toolchain.
-# This disambiguates from the $(target).mk file that exists in the source tree.
-# However, the toolchain is part of the target in universal builds, so we
-# don't want to include TOOLCHAIN in that case. FAT_ARCHS is used to test
-# if we're in the universal case.
-include $(target)$(if $(FAT_ARCHS),,-$(TOOLCHAIN)).mk
-endif
-BUILD_ROOT?=.
-VPATH=$(SRC_PATH_BARE)
-CFLAGS+=-I$(BUILD_PFX)$(BUILD_ROOT) -I$(SRC_PATH)
-ASFLAGS+=-I$(BUILD_PFX)$(BUILD_ROOT)/ -I$(SRC_PATH)/
-DIST_DIR?=dist
-HOSTCC?=gcc
-TGT_ISA:=$(word 1, $(subst -, ,$(TOOLCHAIN)))
-TGT_OS:=$(word 2, $(subst -, ,$(TOOLCHAIN)))
-TGT_CC:=$(word 3, $(subst -, ,$(TOOLCHAIN)))
-quiet:=$(if $(verbose),,yes)
-qexec=$(if $(quiet),@)
-
-# Cancel built-in implicit rules
-%: %.o
-%.asm:
-%.a:
-
-#
-# Common rules"
-#
-.PHONY: all-$(target)
-all-$(target):
-
-.PHONY: clean
-clean::
- rm -f $(OBJS-yes) $(OBJS-yes:.o=.d) $(OBJS-yes:.asm.s.o=.asm.s)
- rm -f $(CLEAN-OBJS)
-
-.PHONY: dist
-dist:
-.PHONY: install
-install::
-
-$(BUILD_PFX)%.c.d: %.c
- $(if $(quiet),@echo " [DEP] $@")
- $(qexec)mkdir -p $(dir $@)
- $(qexec)$(CC) $(CFLAGS) -M $< | $(fmt_deps) > $@
-
-$(BUILD_PFX)%.c.o: %.c
- $(if $(quiet),@echo " [CC] $@")
- $(qexec)$(CC) $(CFLAGS) -c -o $@ $<
-
-$(BUILD_PFX)%.asm.d: %.asm
- $(if $(quiet),@echo " [DEP] $@")
- $(qexec)mkdir -p $(dir $@)
- $(qexec)$(SRC_PATH_BARE)/build/make/gen_asm_deps.sh \
- --build-pfx=$(BUILD_PFX) --depfile=$@ $(ASFLAGS) $< > $@
-
-$(BUILD_PFX)%.asm.o: %.asm
- $(if $(quiet),@echo " [AS] $@")
- $(qexec)$(AS) $(ASFLAGS) -o $@ $<
-
-$(BUILD_PFX)%.s.d: %.s
- $(if $(quiet),@echo " [DEP] $@")
- $(qexec)mkdir -p $(dir $@)
- $(qexec)$(SRC_PATH_BARE)/build/make/gen_asm_deps.sh \
- --build-pfx=$(BUILD_PFX) --depfile=$@ $(ASFLAGS) $< > $@
-
-$(BUILD_PFX)%.s.o: %.s
- $(if $(quiet),@echo " [AS] $@")
- $(qexec)$(AS) $(ASFLAGS) -o $@ $<
-
-.PRECIOUS: %.asm.s
-$(BUILD_PFX)%.asm.s: %.asm
- $(if $(quiet),@echo " [ASM CONVERSION] $@")
- $(qexec)mkdir -p $(dir $@)
- $(qexec)$(ASM_CONVERSION) <$< >$@
-
-# If we're in debug mode, pretend we don't have GNU strip, to fall back to
-# the copy implementation
-HAVE_GNU_STRIP := $(if $(CONFIG_DEBUG),,$(HAVE_GNU_STRIP))
-ifeq ($(HAVE_GNU_STRIP),yes)
-# Older binutils strip global sybols not needed for relocation processing
-# when given --strip-unneeded. Use nm and awk to identify globals and
-# keep them.
-%.a: %_g.a
- $(if $(quiet),@echo " [STRIP] $@ < $<")
- $(qexec)$(STRIP) --strip-unneeded \
- `$(NM) $< | grep ' [A-TV-Z] ' | awk '{print "-K"$$3'}`\
- -o $@ $<
-else
-%.a: %_g.a
- $(if $(quiet),@echo " [CP] $@ < $<")
- $(qexec)cp $< $@
-endif
-
-#
-# Rule to extract assembly constants from C sources
-#
-obj_int_extract: build/make/obj_int_extract.c
- $(if $(quiet),@echo " [HOSTCC] $@")
- $(qexec)$(HOSTCC) -I. -I$(SRC_PATH_BARE) -o $@ $<
-CLEAN-OBJS += obj_int_extract
-
-#
-# Utility functions
-#
-pairmap=$(if $(strip $(2)),\
- $(call $(1),$(word 1,$(2)),$(word 2,$(2)))\
- $(call pairmap,$(1),$(wordlist 3,$(words $(2)),$(2)))\
-)
-
-enabled=$(filter-out $($(1)-no),$($(1)-yes))
-cond_enabled=$(if $(filter yes,$($(1))), $(call enabled,$(2)))
-
-find_file1=$(word 1,$(wildcard $(subst //,/,$(addsuffix /$(1),$(2)))))
-find_file=$(foreach f,$(1),$(call find_file1,$(strip $(f)),$(strip $(2))) )
-obj_pats=.c=.c.o $(AS_SFX)=$(AS_SFX).o
-objs=$(addprefix $(BUILD_PFX),$(foreach p,$(obj_pats),$(filter %.o,$(1:$(p))) ))
-
-install_map_templates=$(eval $(call install_map_template,$(1),$(2)))
-
-not=$(subst yes,no,$(1))
-
-ifeq ($(CONFIG_MSVS),yes)
-lib_file_name=$(1).lib
-else
-lib_file_name=lib$(1).a
-endif
-#
-# Rule Templates
-#
-define linker_template
-$(1): $(filter-out -%,$(2))
-$(1):
- $(if $(quiet),@echo " [LD] $$@")
- $(qexec)$$(LD) $$(strip $$(LDFLAGS) -o $$@ $(2) $(3) $$(extralibs))
-endef
-# make-3.80 has a bug with expanding large input strings to the eval function,
-# which was triggered in some cases by the following component of
-# linker_template:
-# $(1): $$(call find_file, $(patsubst -l%,lib%.a,$(filter -l%,$(2))),\
-# $$(patsubst -L%,%,$$(filter -L%,$$(LDFLAGS) $(2))))
-# This may be useful to revisit in the future (it tries to locate libraries
-# in a search path and add them as prerequisites
-
-define install_map_template
-$(DIST_DIR)/$(1): $(2)
- $(if $(quiet),@echo " [INSTALL] $$@")
- $(qexec)mkdir -p $$(dir $$@)
- $(qexec)cp -p $$< $$@
-endef
-
-define archive_template
-# Not using a pattern rule here because we don't want to generate empty
-# archives when they are listed as a dependency in files not responsible
-# for creating them.
-$(1):
- $(if $(quiet),@echo " [AR] $$@")
- $(qexec)$$(AR) $$(ARFLAGS) $$@ $$?
-endef
-
-define so_template
-# Not using a pattern rule here because we don't want to generate empty
-# archives when they are listed as a dependency in files not responsible
-# for creating them.
-#
-# This needs further abstraction for dealing with non-GNU linkers.
-$(1):
- $(if $(quiet),@echo " [LD] $$@")
- $(qexec)$$(LD) -shared $$(LDFLAGS) \
- -Wl,--no-undefined -Wl,-soname,$$(SONAME) \
- -Wl,--version-script,$$(SO_VERSION_SCRIPT) -o $$@ \
- $$(filter %.o,$$?) $$(extralibs)
-endef
-
-define lipo_lib_template
-$(1): $(addsuffix /$(1),$(FAT_ARCHS))
- $(if $(quiet),@echo " [LIPO] $$@")
- $(qexec)libtool -static -o $$@ $$?
-endef
-
-define lipo_bin_template
-$(1): $(addsuffix /$(1),$(FAT_ARCHS))
- $(if $(quiet),@echo " [LIPO] $$@")
- $(qexec)lipo -output $$@ -create $$?
-endef
-
-
-#
-# Get current configuration
-#
-ifneq ($(target),)
-include $(SRC_PATH_BARE)/$(target:-$(TOOLCHAIN)=).mk
-endif
-ifeq ($(filter clean,$(MAKECMDGOALS)),)
- # Older versions of make don't like -include directives with no arguments
- ifneq ($(filter %.d,$(OBJS-yes:.o=.d)),)
- -include $(filter %.d,$(OBJS-yes:.o=.d))
- endif
-endif
-
-#
-# Configuration dependent rules
-#
-$(call pairmap,install_map_templates,$(INSTALL_MAPS))
-
-DOCS=$(call cond_enabled,CONFIG_INSTALL_DOCS,DOCS)
-.docs: $(DOCS)
- @touch $@
-
-INSTALL-DOCS=$(call cond_enabled,CONFIG_INSTALL_DOCS,INSTALL-DOCS)
-ifeq ($(MAKECMDGOALS),dist)
-INSTALL-DOCS+=$(call cond_enabled,CONFIG_INSTALL_DOCS,DIST-DOCS)
-endif
-.install-docs: .docs $(addprefix $(DIST_DIR)/,$(INSTALL-DOCS))
- @touch $@
-
-clean::
- rm -f .docs .install-docs $(DOCS)
-
-BINS=$(call enabled,BINS)
-.bins: $(BINS)
- @touch $@
-
-INSTALL-BINS=$(call cond_enabled,CONFIG_INSTALL_BINS,INSTALL-BINS)
-ifeq ($(MAKECMDGOALS),dist)
-INSTALL-BINS+=$(call cond_enabled,CONFIG_INSTALL_BINS,DIST-BINS)
-endif
-.install-bins: .bins $(addprefix $(DIST_DIR)/,$(INSTALL-BINS))
- @touch $@
-
-clean::
- rm -f .bins .install-bins $(BINS)
-
-LIBS=$(call enabled,LIBS)
-.libs: $(LIBS)
- @touch $@
-$(foreach lib,$(filter %_g.a,$(LIBS)),$(eval $(call archive_template,$(lib))))
-$(foreach lib,$(filter %so.$(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH),$(LIBS)),$(eval $(call so_template,$(lib))))
-
-INSTALL-LIBS=$(call cond_enabled,CONFIG_INSTALL_LIBS,INSTALL-LIBS)
-ifeq ($(MAKECMDGOALS),dist)
-INSTALL-LIBS+=$(call cond_enabled,CONFIG_INSTALL_LIBS,DIST-LIBS)
-endif
-.install-libs: .libs $(addprefix $(DIST_DIR)/,$(INSTALL-LIBS))
- @touch $@
-
-clean::
- rm -f .libs .install-libs $(LIBS)
-
-ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
-PROJECTS=$(call enabled,PROJECTS)
-.projects: $(PROJECTS)
- @touch $@
-
-INSTALL-PROJECTS=$(call cond_enabled,CONFIG_INSTALL_PROJECTS,INSTALL-PROJECTS)
-ifeq ($(MAKECMDGOALS),dist)
-INSTALL-PROJECTS+=$(call cond_enabled,CONFIG_INSTALL_PROJECTS,DIST-PROJECTS)
-endif
-.install-projects: .projects $(addprefix $(DIST_DIR)/,$(INSTALL-PROJECTS))
- @touch $@
-
-clean::
- rm -f .projects .install-projects $(PROJECTS)
-endif
-
-# If there are any source files to be distributed, then include the build
-# system too.
-ifneq ($(call enabled,DIST-SRCS),)
- DIST-SRCS-yes += configure
- DIST-SRCS-yes += build/make/configure.sh
- DIST-SRCS-yes += build/make/gen_asm_deps.sh
- DIST-SRCS-yes += build/make/Makefile
- DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_def.sh
- DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_proj.sh
- DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_sln.sh
- DIST-SRCS-$(CONFIG_MSVS) += build/x86-msvs/yasm.rules
- DIST-SRCS-$(CONFIG_RVCT) += build/make/armlink_adapter.sh
- # Include obj_int_extract if we use offsets from asm_*_offsets
- DIST-SRCS-$(ARCH_ARM)$(ARCH_X86)$(ARCH_X86_64) += build/make/obj_int_extract.c
- DIST-SRCS-$(ARCH_ARM) += build/make/ads2gas.pl
- DIST-SRCS-yes += $(target:-$(TOOLCHAIN)=).mk
-endif
-INSTALL-SRCS := $(call cond_enabled,CONFIG_INSTALL_SRCS,INSTALL-SRCS)
-ifeq ($(MAKECMDGOALS),dist)
-INSTALL-SRCS += $(call cond_enabled,CONFIG_INSTALL_SRCS,DIST-SRCS)
-endif
-.install-srcs: $(addprefix $(DIST_DIR)/src/,$(INSTALL-SRCS))
- @touch $@
-
-clean::
- rm -f .install-srcs
-
-ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
- BUILD_TARGETS += .projects
- INSTALL_TARGETS += .install-projects
-endif
-BUILD_TARGETS += .docs .libs .bins
-INSTALL_TARGETS += .install-docs .install-srcs .install-libs .install-bins
-all-$(target): $(BUILD_TARGETS)
-install:: $(INSTALL_TARGETS)
-dist: $(INSTALL_TARGETS)
diff --git a/build/make/ads2gas.pl b/build/make/ads2gas.pl
deleted file mode 100755
index 3dff048..0000000
--- a/build/make/ads2gas.pl
+++ /dev/null
@@ -1,150 +0,0 @@
-#!/usr/bin/perl
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-# ads2gas.pl
-# Author: Eric Fung (efung (at) acm.org)
-#
-# Convert ARM Developer Suite 1.0.1 syntax assembly source to GNU as format
-#
-# Usage: cat inputfile | perl ads2gas.pl > outputfile
-#
-print "@ This file was created from a .asm file\n";
-print "@ using the ads2gas.pl script.\n";
-print "\t.equ DO1STROUNDING, 0\n";
-
-while (<STDIN>)
-{
- # Comment character
- s/;/@/g;
-
- # Hexadecimal constants prefaced by 0x
- s/#&/#0x/g;
-
- # Convert :OR: to |
- s/:OR:/ | /g;
-
- # Convert :AND: to &
- s/:AND:/ & /g;
-
- # Convert :NOT: to ~
- s/:NOT:/ ~ /g;
-
- # Convert :SHL: to <<
- s/:SHL:/ << /g;
-
- # Convert :SHR: to >>
- s/:SHR:/ >> /g;
-
- # Convert ELSE to .else
- s/ELSE/.else/g;
-
- # Convert ENDIF to .endif
- s/ENDIF/.endif/g;
-
- # Convert ELSEIF to .elseif
- s/ELSEIF/.elseif/g;
-
- # Convert LTORG to .ltorg
- s/LTORG/.ltorg/g;
-
- # Convert IF :DEF:to .if
- # gcc doesn't have the ability to do a conditional
- # if defined variable that is set by IF :DEF: on
- # armasm, so convert it to a normal .if and then
- # make sure to define a value elesewhere
- if (s/\bIF :DEF:\b/.if /g)
- {
- s/=/==/g;
- }
-
- # Convert IF to .if
- if (s/\bIF\b/.if/g)
- {
- s/=+/==/g;
- }
-
- # Convert INCLUDE to .INCLUDE "file"
- s/INCLUDE(\s*)(.*)$/.include $1\"$2\"/;
-
- # Code directive (ARM vs Thumb)
- s/CODE([0-9][0-9])/.code $1/;
-
- # No AREA required
- s/^\s*AREA.*$/.text/;
-
- # DCD to .word
- # This one is for incoming symbols
- s/DCD\s+\|(\w*)\|/.long $1/;
-
- # DCW to .short
- s/DCW\s+\|(\w*)\|/.short $1/;
- s/DCW(.*)/.short $1/;
-
- # Constants defined in scope
- s/DCD(.*)/.long $1/;
- s/DCB(.*)/.byte $1/;
-
- # RN to .req
- if (s/RN\s+([Rr]\d+|lr)/.req $1/)
- {
- print;
- next;
- }
-
- # Make function visible to linker, and make additional symbol with
- # prepended underscore
- s/EXPORT\s+\|([\$\w]*)\|/.global $1 \n\t.type $1, function/;
- s/IMPORT\s+\|([\$\w]*)\|/.global $1/;
-
- # No vertical bars required; make additional symbol with prepended
- # underscore
- s/^\|(\$?\w+)\|/_$1\n\t$1:/g;
-
- # Labels need trailing colon
-# s/^(\w+)/$1:/ if !/EQU/;
- # put the colon at the end of the line in the macro
- s/^([a-zA-Z_0-9\$]+)/$1:/ if !/EQU/;
-
- # Strip ALIGN
- s/\sALIGN/@ ALIGN/g;
-
- # Strip ARM
- s/\sARM/@ ARM/g;
-
- # Strip REQUIRE8
- #s/\sREQUIRE8/@ REQUIRE8/g;
- s/\sREQUIRE8/@ /g; #EQU cause problem
-
- # Strip PRESERVE8
- s/\sPRESERVE8/@ PRESERVE8/g;
-
- # Strip PROC and ENDPROC
- s/\sPROC/@/g;
- s/\sENDP/@/g;
-
- # EQU directive
- s/(.*)EQU(.*)/.equ $1, $2/;
-
- # Begin macro definition
- if (/MACRO/) {
- $_ = <STDIN>;
- s/^/.macro/;
- s/\$//g; # remove formal param reference
- s/;/@/g; # change comment characters
- }
-
- # For macros, use \ to reference formal params
- s/\$/\\/g; # End macro definition
- s/MEND/.endm/; # No need to tell it where to stop assembling
- next if /^\s*END\s*$/;
- print;
-}
diff --git a/build/make/ads2gas_apple.pl b/build/make/ads2gas_apple.pl
deleted file mode 100755
index 5014c61..0000000
--- a/build/make/ads2gas_apple.pl
+++ /dev/null
@@ -1,195 +0,0 @@
-#!/usr/bin/env perl
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-# ads2gas.pl
-# Author: Eric Fung (efung (at) acm.org)
-#
-# Convert ARM Developer Suite 1.0.1 syntax assembly source to GNU as format
-#
-# Usage: cat inputfile | perl ads2gas.pl > outputfile
-#
-print "@ This file was created from a .asm file\n";
-print "@ using the ads2gas_apple.pl script.\n\n";
-print "\t.set WIDE_REFERENCE, 0\n";
-print "\t.set ARCHITECTURE, 5\n";
-print "\t.set DO1STROUNDING, 0\n";
-
-my %register_aliases;
-my %macro_aliases;
-
-my @mapping_list = ("\$0", "\$1", "\$2", "\$3", "\$4", "\$5", "\$6", "\$7", "\$8", "\$9");
-
-my @incoming_array;
-
-# Perl trim function to remove whitespace from the start and end of the string
-sub trim($)
-{
- my $string = shift;
- $string =~ s/^\s+//;
- $string =~ s/\s+$//;
- return $string;
-}
-
-while (<STDIN>)
-{
- # Comment character
- s/;/@/g;
-
- # Hexadecimal constants prefaced by 0x
- s/#&/#0x/g;
-
- # Convert :OR: to |
- s/:OR:/ | /g;
-
- # Convert :AND: to &
- s/:AND:/ & /g;
-
- # Convert :NOT: to ~
- s/:NOT:/ ~ /g;
-
- # Convert :SHL: to <<
- s/:SHL:/ << /g;
-
- # Convert :SHR: to >>
- s/:SHR:/ >> /g;
-
- # Convert ELSE to .else
- s/ELSE/.else/g;
-
- # Convert ENDIF to .endif
- s/ENDIF/.endif/g;
-
- # Convert ELSEIF to .elseif
- s/ELSEIF/.elseif/g;
-
- # Convert LTORG to .ltorg
- s/LTORG/.ltorg/g;
-
- # Convert IF :DEF:to .if
- # gcc doesn't have the ability to do a conditional
- # if defined variable that is set by IF :DEF: on
- # armasm, so convert it to a normal .if and then
- # make sure to define a value elesewhere
- if (s/\bIF :DEF:\b/.if /g)
- {
- s/=/==/g;
- }
-
- # Convert IF to .if
- if (s/\bIF\b/.if/g)
- {
- s/=/==/g;
- }
-
- # Convert INCLUDE to .INCLUDE "file"
- s/INCLUDE(\s*)(.*)$/.include $1\"$2\"/;
-
- # Code directive (ARM vs Thumb)
- s/CODE([0-9][0-9])/.code $1/;
-
- # No AREA required
- s/^\s*AREA.*$/.text/;
-
- # DCD to .word
- # This one is for incoming symbols
- s/DCD\s+\|(\w*)\|/.long $1/;
-
- # DCW to .short
- s/DCW\s+\|(\w*)\|/.short $1/;
- s/DCW(.*)/.short $1/;
-
- # Constants defined in scope
- s/DCD(.*)/.long $1/;
- s/DCB(.*)/.byte $1/;
-
- # Build a hash of all the register - alias pairs.
- if (s/(.*)RN(.*)/$1 .req $2/g)
- {
- $register_aliases{trim($1)} = trim($2);
- next;
- }
-
- while (($key, $value) = each(%register_aliases))
- {
- s/\b$key\b/$value/g;
- }
-
- # Make function visible to linker, and make additional symbol with
- # prepended underscore
- s/EXPORT\s+\|([\$\w]*)\|/.globl _$1\n\t.globl $1/;
- s/IMPORT\s+\|([\$\w]*)\|/.globl $1/;
-
- # No vertical bars required; make additional symbol with prepended
- # underscore
- s/^\|(\$?\w+)\|/_$1\n\t$1:/g;
-
- # Labels need trailing colon
-# s/^(\w+)/$1:/ if !/EQU/;
- # put the colon at the end of the line in the macro
- s/^([a-zA-Z_0-9\$]+)/$1:/ if !/EQU/;
-
- # Strip ALIGN
- s/\sALIGN/@ ALIGN/g;
-
- # Strip ARM
- s/\sARM/@ ARM/g;
-
- # Strip REQUIRE8
- #s/\sREQUIRE8/@ REQUIRE8/g;
- s/\sREQUIRE8/@ /g;
-
- # Strip PRESERVE8
- s/\sPRESERVE8/@ PRESERVE8/g;
-
- # Strip PROC and ENDPROC
- s/PROC/@/g;
- s/ENDP/@/g;
-
- # EQU directive
- s/(.*)EQU(.*)/.set $1, $2/;
-
- # Begin macro definition
- if (/MACRO/)
- {
- # Process next line down, which will be the macro definition
- $_ = <STDIN>;
-
- $trimmed = trim($_);
-
- # remove commas that are separating list
- $trimmed =~ s/,//g;
-
- # string to array
- @incoming_array = split(/ /, $trimmed);
-
- print ".macro @incoming_array[0]\n";
-
- # remove the first element, as that is the name of the macro
- shift (@incoming_array);
-
- @macro_aliases{@incoming_array} = @mapping_list;
-
- next;
- }
-
- while (($key, $value) = each(%macro_aliases))
- {
- $key =~ s/\$/\\\$/;
- s/$key\b/$value/g;
- }
-
- # For macros, use \ to reference formal params
-# s/\$/\\/g; # End macro definition
- s/MEND/.endm/; # No need to tell it where to stop assembling
- next if /^\s*END\s*$/;
- print;
-}
diff --git a/build/make/configure.sh b/build/make/configure.sh
deleted file mode 100755
index a48fd9f..0000000
--- a/build/make/configure.sh
+++ /dev/null
@@ -1,1061 +0,0 @@
-#!/bin/bash
-##
-## configure.sh
-##
-## This script is sourced by the main configure script and contains
-## utility functions and other common bits that aren't strictly libvpx
-## related.
-##
-## This build system is based in part on the FFmpeg configure script.
-##
-
-
-#
-# Logging / Output Functions
-#
-die_unknown(){
- echo "Unknown option \"$1\"."
- echo "See $0 --help for available options."
- clean_temp_files
- exit 1
-}
-
-
-die() {
- echo "$@"
- echo
- echo "Configuration failed. This could reflect a misconfiguration of your"
- echo "toolchains, improper options selected, or another problem. If you"
- echo "don't see any useful error messages above, the next step is to look"
- echo "at the configure error log file ($logfile) to determine what"
- echo "configure was trying to do when it died."
- clean_temp_files
- exit 1
-}
-
-
-log(){
- echo "$@" >>$logfile
-}
-
-
-log_file(){
- log BEGIN $1
- pr -n -t $1 >>$logfile
- log END $1
-}
-
-
-log_echo() {
- echo "$@"
- log "$@"
-}
-
-
-fwrite () {
- outfile=$1
- shift
- echo "$@" >> ${outfile}
-}
-
-
-show_help_pre(){
- for opt in ${CMDLINE_SELECT}; do
- opt2=`echo $opt | sed -e 's;_;-;g'`
- if enabled $opt; then
- eval "toggle_${opt}=\"--disable-${opt2}\""
- else
- eval "toggle_${opt}=\"--enable-${opt2} \""
- fi
- done
-
- cat <<EOF
-Usage: configure [options]
-Options:
-
-Build options:
- --help print this message
- --log=yes|no|FILE file configure log is written to [config.err]
- --target=TARGET target platform tuple [generic-gnu]
- --cpu=CPU optimize for a specific cpu rather than a family
- --extra-cflags=ECFLAGS add ECFLAGS to CFLAGS [$CFLAGS]
- ${toggle_extra_warnings} emit harmless warnings (always non-fatal)
- ${toggle_werror} treat warnings as errors, if possible
- (not available with all compilers)
- ${toggle_optimizations} turn on/off compiler optimization flags
- ${toggle_pic} turn on/off Position Independent Code
- ${toggle_ccache} turn on/off compiler cache
- ${toggle_debug} enable/disable debug mode
- ${toggle_gprof} enable/disable gprof profiling instrumentation
- ${toggle_gcov} enable/disable gcov coverage instrumentation
-
-Install options:
- ${toggle_install_docs} control whether docs are installed
- ${toggle_install_bins} control whether binaries are installed
- ${toggle_install_libs} control whether libraries are installed
- ${toggle_install_srcs} control whether sources are installed
-
-
-EOF
-}
-
-
-show_help_post(){
- cat <<EOF
-
-
-NOTES:
- Object files are built at the place where configure is launched.
-
- All boolean options can be negated. The default value is the opposite
- of that shown above. If the option --disable-foo is listed, then
- the default value for foo is enabled.
-
-Supported targets:
-EOF
- show_targets ${all_platforms}
- echo
- exit 1
-}
-
-
-show_targets() {
- while [ -n "$*" ]; do
- if [ "${1%%-*}" = "${2%%-*}" ]; then
- if [ "${2%%-*}" = "${3%%-*}" ]; then
- printf " %-24s %-24s %-24s\n" "$1" "$2" "$3"
- shift; shift; shift
- else
- printf " %-24s %-24s\n" "$1" "$2"
- shift; shift
- fi
- else
- printf " %-24s\n" "$1"
- shift
- fi
- done
-}
-
-
-show_help() {
- show_help_pre
- show_help_post
-}
-
-#
-# List Processing Functions
-#
-set_all(){
- value=$1
- shift
- for var in $*; do
- eval $var=$value
- done
-}
-
-
-is_in(){
- value=$1
- shift
- for var in $*; do
- [ $var = $value ] && return 0
- done
- return 1
-}
-
-
-add_cflags() {
- CFLAGS="${CFLAGS} $@"
-}
-
-
-add_ldflags() {
- LDFLAGS="${LDFLAGS} $@"
-}
-
-
-add_asflags() {
- ASFLAGS="${ASFLAGS} $@"
-}
-
-
-add_extralibs() {
- extralibs="${extralibs} $@"
-}
-
-#
-# Boolean Manipulation Functions
-#
-enable(){
- set_all yes $*
-}
-
-disable(){
- set_all no $*
-}
-
-enabled(){
- eval test "x\$$1" = "xyes"
-}
-
-disabled(){
- eval test "x\$$1" = "xno"
-}
-
-
-soft_enable() {
- for var in $*; do
- if ! disabled $var; then
- log_echo " enabling $var"
- enable $var
- fi
- done
-}
-
-soft_disable() {
- for var in $*; do
- if ! enabled $var; then
- log_echo " disabling $var"
- disable $var
- fi
- done
-}
-
-
-#
-# Text Processing Functions
-#
-toupper(){
- echo "$@" | tr abcdefghijklmnopqrstuvwxyz ABCDEFGHIJKLMNOPQRSTUVWXYZ
-}
-
-
-tolower(){
- echo "$@" | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz
-}
-
-
-#
-# Temporary File Functions
-#
-source_path=${0%/*}
-enable source_path_used
-if test -z "$source_path" -o "$source_path" = "." ; then
- source_path="`pwd`"
- disable source_path_used
-fi
-
-if test ! -z "$TMPDIR" ; then
- TMPDIRx="${TMPDIR}"
-elif test ! -z "$TEMPDIR" ; then
- TMPDIRx="${TEMPDIR}"
-else
- TMPDIRx="/tmp"
-fi
-TMP_H="${TMPDIRx}/vpx-conf-$$-${RANDOM}.h"
-TMP_C="${TMPDIRx}/vpx-conf-$$-${RANDOM}.c"
-TMP_O="${TMPDIRx}/vpx-conf-$$-${RANDOM}.o"
-TMP_X="${TMPDIRx}/vpx-conf-$$-${RANDOM}.x"
-TMP_ASM="${TMPDIRx}/vpx-conf-$$-${RANDOM}.asm"
-
-clean_temp_files() {
- rm -f ${TMP_C} ${TMP_H} ${TMP_O} ${TMP_X} ${TMP_ASM}
-}
-
-#
-# Toolchain Check Functions
-#
-check_cmd() {
- log "$@"
- "$@" >>${logfile} 2>&1
-}
-
-check_cc() {
- log check_cc "$@"
- cat >${TMP_C}
- log_file ${TMP_C}
- check_cmd ${CC} ${CFLAGS} "$@" -c -o ${TMP_O} ${TMP_C}
-}
-
-check_cpp() {
- log check_cpp "$@"
- cat > ${TMP_C}
- log_file ${TMP_C}
- check_cmd ${CC} ${CFLAGS} "$@" -E -o ${TMP_O} ${TMP_C}
-}
-
-check_ld() {
- log check_ld "$@"
- check_cc $@ \
- && check_cmd ${LD} ${LDFLAGS} "$@" -o ${TMP_X} ${TMP_O} ${extralibs}
-}
-
-check_header(){
- log check_header "$@"
- header=$1
- shift
- var=`echo $header | sed 's/[^A-Za-z0-9_]/_/g'`
- disable $var
- check_cpp "$@" <<EOF && enable $var
-#include "$header"
-int x;
-EOF
-}
-
-
-check_cflags() {
- log check_cflags "$@"
- check_cc "$@" <<EOF
-int x;
-EOF
-}
-
-check_add_cflags() {
- check_cflags "$@" && add_cflags "$@"
-}
-
-check_add_asflags() {
- log add_asflags "$@"
- add_asflags "$@"
-}
-
-check_add_ldflags() {
- log add_ldflags "$@"
- add_ldflags "$@"
-}
-
-check_asm_align() {
- log check_asm_align "$@"
- cat >${TMP_ASM} <<EOF
-section .rodata
-align 16
-EOF
- log_file ${TMP_ASM}
- check_cmd ${AS} ${ASFLAGS} -o ${TMP_O} ${TMP_ASM}
- readelf -WS ${TMP_O} >${TMP_X}
- log_file ${TMP_X}
- if ! grep -q '\.rodata .* 16$' ${TMP_X}; then
- die "${AS} ${ASFLAGS} does not support section alignment (nasm <=2.08?)"
- fi
-}
-
-write_common_config_banner() {
- echo '# This file automatically generated by configure. Do not edit!' > config.mk
- echo "TOOLCHAIN := ${toolchain}" >> config.mk
-
- case ${toolchain} in
- *-linux-rvct)
- echo "ALT_LIBC := ${alt_libc}" >> config.mk
- ;;
- esac
-}
-
-write_common_config_targets() {
- for t in ${all_targets}; do
- if enabled ${t}; then
- if enabled universal || enabled child; then
- fwrite config.mk "ALL_TARGETS += ${t}-${toolchain}"
- else
- fwrite config.mk "ALL_TARGETS += ${t}"
- fi
- fi
- true;
- done
-true
-}
-
-write_common_target_config_mk() {
- local CC=${CC}
- enabled ccache && CC="ccache ${CC}"
-
- cat > $1 << EOF
-# This file automatically generated by configure. Do not edit!
-SRC_PATH="$source_path"
-SRC_PATH_BARE=$source_path
-BUILD_PFX=${BUILD_PFX}
-TOOLCHAIN=${toolchain}
-ASM_CONVERSION=${asm_conversion_cmd:-${source_path}/build/make/ads2gas.pl}
-
-CC=${CC}
-AR=${AR}
-LD=${LD}
-AS=${AS}
-STRIP=${STRIP}
-NM=${NM}
-
-CFLAGS = ${CFLAGS}
-ARFLAGS = -rus\$(if \$(quiet),c,v)
-LDFLAGS = ${LDFLAGS}
-ASFLAGS = ${ASFLAGS}
-extralibs = ${extralibs}
-AS_SFX = ${AS_SFX:-.asm}
-EOF
-
- if enabled rvct; then cat >> $1 << EOF
-fmt_deps = sed -e 's;^__image.axf;\$(dir \$@)\$(notdir \$<).o \$@;' #hide
-EOF
- else cat >> $1 << EOF
-fmt_deps = sed -e 's;^\([a-zA-Z0-9_]*\)\.o;\$(dir \$@)\1\$(suffix \$<).o \$@;'
-EOF
- fi
-
- print_config_mk ARCH "${1}" ${ARCH_LIST}
- print_config_mk HAVE "${1}" ${HAVE_LIST}
- print_config_mk CONFIG "${1}" ${CONFIG_LIST}
- print_config_mk HAVE "${1}" gnu_strip
-
- enabled msvs && echo "CONFIG_VS_VERSION=${vs_version}" >> "${1}"
-
-}
-
-
-write_common_target_config_h() {
- cat > ${TMP_H} << EOF
-/* This file automatically generated by configure. Do not edit! */
-#define RESTRICT ${RESTRICT}
-EOF
- print_config_h ARCH "${TMP_H}" ${ARCH_LIST}
- print_config_h HAVE "${TMP_H}" ${HAVE_LIST}
- print_config_h CONFIG "${TMP_H}" ${CONFIG_LIST}
- mkdir -p `dirname "$1"`
- cmp "$1" ${TMP_H} >/dev/null 2>&1 || mv ${TMP_H} "$1"
-}
-
-process_common_cmdline() {
- for opt in "$@"; do
- optval="${opt#*=}"
- case "$opt" in
- --child) enable child
- ;;
- --log*)
- logging="$optval"
- if ! disabled logging ; then
- enabled logging || logfile="$logging"
- else
- logfile=/dev/null
- fi
- ;;
- --target=*) toolchain="${toolchain:-${optval}}"
- ;;
- --force-target=*) toolchain="${toolchain:-${optval}}"; enable force_toolchain
- ;;
- --cpu)
- ;;
- --cpu=*) tune_cpu="$optval"
- ;;
- --extra-cflags=*)
- extra_cflags="${optval}"
- ;;
- --enable-?*|--disable-?*)
- eval `echo "$opt" | sed 's/--/action=/;s/-/ option=/;s/-/_/g'`
- echo "${CMDLINE_SELECT} ${ARCH_EXT_LIST}" | grep "^ *$option\$" >/dev/null || die_unknown $opt
- $action $option
- ;;
- --force-enable-?*|--force-disable-?*)
- eval `echo "$opt" | sed 's/--force-/action=/;s/-/ option=/;s/-/_/g'`
- $action $option
- ;;
- --libc=*)
- [ -d "${optval}" ] || die "Not a directory: ${optval}"
- disable builtin_libc
- alt_libc="${optval}"
- ;;
- --as=*)
- [ "${optval}" = yasm -o "${optval}" = nasm -o "${optval}" = auto ] \
- || die "Must be yasm, nasm or auto: ${optval}"
- alt_as="${optval}"
- ;;
- --prefix=*)
- prefix="${optval}"
- ;;
- --libdir=*)
- libdir="${optval}"
- ;;
- --libc|--as|--prefix|--libdir)
- die "Option ${opt} requires argument"
- ;;
- --help|-h) show_help
- ;;
- *) die_unknown $opt
- ;;
- esac
- done
-}
-
-process_cmdline() {
- for opt do
- optval="${opt#*=}"
- case "$opt" in
- *) process_common_cmdline $opt
- ;;
- esac
- done
-}
-
-
-post_process_common_cmdline() {
- prefix="${prefix:-/usr/local}"
- prefix="${prefix%/}"
- libdir="${libdir:-${prefix}/lib}"
- libdir="${libdir%/}"
- if [ "${libdir#${prefix}}" = "${libdir}" ]; then
- die "Libdir ${libdir} must be a subdirectory of ${prefix}"
- fi
-}
-
-
-post_process_cmdline() {
- true;
-}
-
-setup_gnu_toolchain() {
- CC=${CC:-${CROSS}gcc}
- AR=${AR:-${CROSS}ar}
- LD=${LD:-${CROSS}${link_with_cc:-ld}}
- AS=${AS:-${CROSS}as}
- STRIP=${STRIP:-${CROSS}strip}
- NM=${NM:-${CROSS}nm}
- AS_SFX=.s
-}
-
-process_common_toolchain() {
- if [ -z "$toolchain" ]; then
- gcctarget="$(gcc -dumpmachine 2> /dev/null)"
-
- # detect tgt_isa
- case "$gcctarget" in
- *x86_64*|*amd64*)
- tgt_isa=x86_64
- ;;
- *i[3456]86*)
- tgt_isa=x86
- ;;
- *powerpc64*)
- tgt_isa=ppc64
- ;;
- *powerpc*)
- tgt_isa=ppc32
- ;;
- *sparc*)
- tgt_isa=sparc
- ;;
- esac
-
- # detect tgt_os
- case "$gcctarget" in
- *darwin8*)
- tgt_isa=universal
- tgt_os=darwin8
- ;;
- *darwin9*)
- tgt_isa=universal
- tgt_os=darwin9
- ;;
- *darwin10*)
- tgt_isa=x86_64
- tgt_os=darwin10
- ;;
- *mingw32*|*cygwin*)
- [ -z "$tgt_isa" ] && tgt_isa=x86
- tgt_os=win32
- ;;
- *linux*|*bsd*)
- tgt_os=linux
- ;;
- *solaris2.10)
- tgt_os=solaris
- ;;
- esac
-
- if [ -n "$tgt_isa" ] && [ -n "$tgt_os" ]; then
- toolchain=${tgt_isa}-${tgt_os}-gcc
- fi
- fi
-
- toolchain=${toolchain:-generic-gnu}
-
- is_in ${toolchain} ${all_platforms} || enabled force_toolchain \
- || die "Unrecognized toolchain '${toolchain}'"
-
- enabled child || log_echo "Configuring for target '${toolchain}'"
-
- #
- # Set up toolchain variables
- #
- tgt_isa=$(echo ${toolchain} | awk 'BEGIN{FS="-"}{print $1}')
- tgt_os=$(echo ${toolchain} | awk 'BEGIN{FS="-"}{print $2}')
- tgt_cc=$(echo ${toolchain} | awk 'BEGIN{FS="-"}{print $3}')
-
- # Mark the specific ISA requested as enabled
- soft_enable ${tgt_isa}
- enable ${tgt_os}
- enable ${tgt_cc}
-
- # Enable the architecture family
- case ${tgt_isa} in
- arm*|iwmmxt*) enable arm;;
- mips*) enable mips;;
- esac
-
- # PIC is probably what we want when building shared libs
- enabled shared && soft_enable pic
-
- # Handle darwin variants
- case ${toolchain} in
- *-darwin8-*)
- add_cflags "-isysroot /Developer/SDKs/MacOSX10.4u.sdk"
- add_cflags "-mmacosx-version-min=10.4"
- add_ldflags "-isysroot /Developer/SDKs/MacOSX10.4u.sdk"
- add_ldflags "-mmacosx-version-min=10.4"
- ;;
- *-darwin9-*)
- add_cflags "-isysroot /Developer/SDKs/MacOSX10.5.sdk"
- add_cflags "-mmacosx-version-min=10.5"
- add_ldflags "-isysroot /Developer/SDKs/MacOSX10.5.sdk"
- add_ldflags "-mmacosx-version-min=10.5"
- ;;
- *-darwin10-*)
- add_cflags "-isysroot /Developer/SDKs/MacOSX10.6.sdk"
- add_cflags "-mmacosx-version-min=10.6"
- add_ldflags "-isysroot /Developer/SDKs/MacOSX10.6.sdk"
- add_ldflags "-mmacosx-version-min=10.6"
- ;;
- esac
-
- # Handle Solaris variants. Solaris 10 needs -lposix4
- case ${toolchain} in
- sparc-solaris-*)
- add_extralibs -lposix4
- add_cflags "-DMUST_BE_ALIGNED"
- ;;
- *-solaris-*)
- add_extralibs -lposix4
- ;;
- esac
-
- # Process ARM architecture variants
- case ${toolchain} in
- arm*|iwmmxt*)
- # on arm, isa versions are supersets
- enabled armv7a && soft_enable armv7 ### DEBUG
- enabled armv7 && soft_enable armv6
- enabled armv6 && soft_enable armv5te
- enabled armv6 && soft_enable fast_unaligned
- enabled iwmmxt2 && soft_enable iwmmxt
- enabled iwmmxt && soft_enable armv5te
-
- asm_conversion_cmd="cat"
-
- case ${tgt_cc} in
- gcc)
- if enabled iwmmxt || enabled iwmmxt2
- then
- CROSS=${CROSS:-arm-iwmmxt-linux-gnueabi-}
- elif enabled symbian; then
- CROSS=${CROSS:-arm-none-symbianelf-}
- else
- CROSS=${CROSS:-arm-none-linux-gnueabi-}
- fi
- link_with_cc=gcc
- setup_gnu_toolchain
- arch_int=${tgt_isa##armv}
- arch_int=${arch_int%%te}
- check_add_asflags --defsym ARCHITECTURE=${arch_int}
- tune_cflags="-mtune="
- if enabled iwmmxt || enabled iwmmxt2
- then
- check_add_asflags -mcpu=${tgt_isa}
- elif enabled armv7
- then
- check_add_cflags -march=armv7-a -mcpu=cortex-a8 -mfpu=neon -mfloat-abi=softfp #-ftree-vectorize
- check_add_asflags -mcpu=cortex-a8 -mfpu=neon -mfloat-abi=softfp #-march=armv7-a
- else
- check_add_cflags -march=${tgt_isa}
- check_add_asflags -march=${tgt_isa}
- fi
- enabled debug && add_asflags -g
- asm_conversion_cmd="${source_path}/build/make/ads2gas.pl"
- ;;
- rvct)
- CC=armcc
- AR=armar
- AS=armasm
- LD=${source_path}/build/make/armlink_adapter.sh
- STRIP=arm-none-linux-gnueabi-strip
- NM=arm-none-linux-gnueabi-nm
- tune_cflags="--cpu="
- tune_asflags="--cpu="
- if [ -z "${tune_cpu}" ]; then
- if enabled armv7
- then
- check_add_cflags --cpu=Cortex-A8 --fpu=softvfp+vfpv3
- check_add_asflags --cpu=Cortex-A8 --fpu=none
- else
- check_add_cflags --cpu=${tgt_isa##armv}
- check_add_asflags --cpu=${tgt_isa##armv}
- fi
- fi
- arch_int=${tgt_isa##armv}
- arch_int=${arch_int%%te}
- check_add_asflags --pd "\"ARCHITECTURE SETA ${arch_int}\""
- enabled debug && add_asflags -g
- add_cflags --gnu
- add_cflags --enum_is_int
- add_cflags --wchar32
- ;;
- esac
-
- case ${tgt_os} in
- none*)
- disable multithread
- disable os_support
- ;;
- darwin*)
- SDK_PATH=/Developer/Platforms/iPhoneOS.platform/Developer
- TOOLCHAIN_PATH=${SDK_PATH}/usr/bin
- CC=${TOOLCHAIN_PATH}/gcc
- AR=${TOOLCHAIN_PATH}/ar
- LD=${TOOLCHAIN_PATH}/arm-apple-darwin10-gcc-4.2.1
- AS=${TOOLCHAIN_PATH}/as
- STRIP=${TOOLCHAIN_PATH}/strip
- NM=${TOOLCHAIN_PATH}/nm
- AS_SFX=.s
-
- # ASFLAGS is written here instead of using check_add_asflags
- # because we need to overwrite all of ASFLAGS and purge the
- # options that were put in above
- ASFLAGS="-version -arch ${tgt_isa} -g"
-
- add_cflags -arch ${tgt_isa}
- add_ldflags -arch_only ${tgt_isa}
-
- add_cflags "-isysroot /Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS4.2.sdk"
-
- # This should be overridable
- alt_libc=${SDK_PATH}/SDKs/iPhoneOS4.2.sdk
-
- # Add the paths for the alternate libc
-# for d in usr/include usr/include/gcc/darwin/4.0/; do
- for d in usr/include usr/include/gcc/darwin/4.0/ usr/lib/gcc/arm-apple-darwin10/4.2.1/include/; do
- try_dir="${alt_libc}/${d}"
- [ -d "${try_dir}" ] && add_cflags -I"${try_dir}"
- done
-
- for d in lib usr/lib; do
- try_dir="${alt_libc}/${d}"
- [ -d "${try_dir}" ] && add_ldflags -L"${try_dir}"
- done
-
- asm_conversion_cmd="${source_path}/build/make/ads2gas_apple.pl"
- ;;
-
- linux*)
- enable linux
- if enabled rvct; then
- # Compiling with RVCT requires an alternate libc (glibc) when
- # targetting linux.
- disabled builtin_libc \
- || die "Must supply --libc when targetting *-linux-rvct"
-
- # Set up compiler
- add_cflags --library_interface=aeabi_glibc
- add_cflags --no_hide_all
- add_cflags --dwarf2
-
- # Set up linker
- add_ldflags --sysv --no_startup --no_ref_cpp_init
- add_ldflags --entry=_start
- add_ldflags --keep '"*(.init)"' --keep '"*(.fini)"'
- add_ldflags --keep '"*(.init_array)"' --keep '"*(.fini_array)"'
- add_ldflags --dynamiclinker=/lib/ld-linux.so.3
- add_extralibs libc.so.6 -lc_nonshared crt1.o crti.o crtn.o
-
- # Add the paths for the alternate libc
- for d in usr/include; do
- try_dir="${alt_libc}/${d}"
- [ -d "${try_dir}" ] && add_cflags -J"${try_dir}"
- done
- add_cflags -J"${RVCT31INC}"
- for d in lib usr/lib; do
- try_dir="${alt_libc}/${d}"
- [ -d "${try_dir}" ] && add_ldflags -L"${try_dir}"
- done
-
-
- # glibc has some struct members named __align, which is a
- # storage modifier in RVCT. If we need to use this modifier,
- # we'll have to #undef it in our code. Note that this must
- # happen AFTER all libc inclues.
- add_cflags -D__align=x_align_x
- fi
- ;;
-
- symbian*)
- enable symbian
- # Add the paths for the alternate libc
- for d in include/libc; do
- try_dir="${alt_libc}/${d}"
- [ -d "${try_dir}" ] && add_cflags -I"${try_dir}"
- done
- for d in release/armv5/urel; do
- try_dir="${alt_libc}/${d}"
- [ -d "${try_dir}" ] && add_ldflags -L"${try_dir}"
- done
- add_cflags -DIMPORT_C=
-
- esac
- ;;
- mips*)
- CROSS=${CROSS:-mipsel-linux-uclibc-}
- link_with_cc=gcc
- setup_gnu_toolchain
- tune_cflags="-mtune="
- check_add_cflags -march=${tgt_isa}
- check_add_asflags -march=${tgt_isa}
- check_add_asflags -KPIC
- ;;
- ppc*)
- enable ppc
- bits=${tgt_isa##ppc}
- link_with_cc=gcc
- setup_gnu_toolchain
- add_asflags -force_cpusubtype_ALL -I"\$(dir \$<)darwin"
- soft_enable altivec
- enabled altivec && add_cflags -maltivec
-
- case "$tgt_os" in
- linux*)
- add_asflags -maltivec -mregnames -I"\$(dir \$<)linux"
- ;;
- darwin*)
- darwin_arch="-arch ppc"
- enabled ppc64 && darwin_arch="${darwin_arch}64"
- add_cflags ${darwin_arch} -m${bits} -fasm-blocks
- add_asflags ${darwin_arch} -force_cpusubtype_ALL -I"\$(dir \$<)darwin"
- add_ldflags ${darwin_arch} -m${bits}
- enabled altivec && add_cflags -faltivec
- ;;
- esac
- ;;
- x86*)
- bits=32
- enabled x86_64 && bits=64
- soft_enable runtime_cpu_detect
- soft_enable mmx
- soft_enable sse
- soft_enable sse2
- soft_enable sse3
- soft_enable ssse3
- soft_enable sse4_1
-
- case ${tgt_os} in
- win*)
- enabled gcc && add_cflags -fno-common
- ;;
- solaris*)
- CC=${CC:-${CROSS}gcc}
- LD=${LD:-${CROSS}gcc}
- CROSS=${CROSS:-g}
- ;;
- esac
-
- AS="${alt_as:-${AS:-auto}}"
- case ${tgt_cc} in
- icc*)
- CC=${CC:-icc}
- LD=${LD:-icc}
- setup_gnu_toolchain
- add_cflags -use-msasm -use-asm
- add_ldflags -i-static
- enabled x86_64 && add_cflags -ipo -no-prec-div -static -xSSE2 -axSSE2
- enabled x86_64 && AR=xiar
- case ${tune_cpu} in
- atom*)
- tune_cflags="-x"
- tune_cpu="SSE3_ATOM"
- ;;
- *)
- tune_cflags="-march="
- ;;
- esac
- ;;
- gcc*)
- add_cflags -m${bits}
- add_ldflags -m${bits}
- link_with_cc=gcc
- tune_cflags="-march="
- setup_gnu_toolchain
- ;;
- esac
-
- case "${AS}" in
- auto|"")
- which nasm >/dev/null 2>&1 && AS=nasm
- which yasm >/dev/null 2>&1 && AS=yasm
- [ "${AS}" = auto -o -z "${AS}" ] \
- && die "Neither yasm nor nasm have been found"
- ;;
- esac
- log_echo " using $AS"
- [ "${AS##*/}" = nasm ] && add_asflags -Ox
- AS_SFX=.asm
- case ${tgt_os} in
- win*)
- add_asflags -f win${bits}
- enabled debug && add_asflags -g cv8
- ;;
- linux*|solaris*)
- add_asflags -f elf${bits}
- enabled debug && [ "${AS}" = yasm ] && add_asflags -g dwarf2
- enabled debug && [ "${AS}" = nasm ] && add_asflags -g
- [ "${AS##*/}" = nasm ] && check_asm_align
- ;;
- darwin*)
- add_asflags -f macho${bits}
- enabled x86 && darwin_arch="-arch i386" || darwin_arch="-arch x86_64"
- add_cflags ${darwin_arch}
- add_ldflags ${darwin_arch}
- # -mdynamic-no-pic is still a bit of voodoo -- it was required at
- # one time, but does not seem to be now, and it breaks some of the
- # code that still relies on inline assembly.
- # enabled icc && ! enabled pic && add_cflags -fno-pic -mdynamic-no-pic
- enabled icc && ! enabled pic && add_cflags -fno-pic
- ;;
- *) log "Warning: Unknown os $tgt_os while setting up $AS flags"
- ;;
- esac
- ;;
- universal*|*-gcc|generic-gnu)
- link_with_cc=gcc
- enable gcc
- setup_gnu_toolchain
- ;;
- esac
-
- # Try to enable CPU specific tuning
- if [ -n "${tune_cpu}" ]; then
- if [ -n "${tune_cflags}" ]; then
- check_add_cflags ${tune_cflags}${tune_cpu} || \
- die "Requested CPU '${tune_cpu}' not supported by compiler"
- fi
- if [ -n "${tune_asflags}" ]; then
- check_add_asflags ${tune_asflags}${tune_cpu} || \
- die "Requested CPU '${tune_cpu}' not supported by assembler"
- fi
- if [ -z "${tune_cflags}${tune_asflags}" ]; then
- log_echo "Warning: CPU tuning not supported by this toolchain"
- fi
- fi
-
- enabled debug && check_add_cflags -g && check_add_ldflags -g
- enabled gprof && check_add_cflags -pg && check_add_ldflags -pg
- enabled gcov &&
- check_add_cflags -fprofile-arcs -ftest-coverage &&
- check_add_ldflags -fprofile-arcs -ftest-coverage
- if enabled optimizations; then
- enabled rvct && check_add_cflags -Otime
- enabled small && check_add_cflags -O2 || check_add_cflags -O3
- fi
-
- # Position Independent Code (PIC) support, for building relocatable
- # shared objects
- enabled gcc && enabled pic && check_add_cflags -fPIC
-
- # Check for strip utility variant
- ${STRIP} -V 2>/dev/null | grep GNU >/dev/null && enable gnu_strip
-
- # Try to determine target endianness
- check_cc <<EOF
- unsigned int e = 'O'<<24 | '2'<<16 | 'B'<<8 | 'E';
-EOF
- [ -f "${TMP_O}" ] && od -A n -t x1 "${TMP_O}" | tr -d '\n' |
- grep '4f *32 *42 *45' >/dev/null 2>&1 && enable big_endian
-
- # Almost every platform uses pthreads.
- if enabled multithread; then
- case ${toolchain} in
- *-win*);;
- *) check_header pthread.h && add_extralibs -lpthread
- esac
- fi
-
- # glibc needs these
- if enabled linux; then
- add_cflags -D_LARGEFILE_SOURCE
- add_cflags -D_FILE_OFFSET_BITS=64
- fi
-
- # append any user defined extra cflags
- if [ -n "${extra_cflags}" ] ; then
- check_add_cflags ${extra_cflags} || \
- die "Requested extra CFLAGS '${extra_cflags}' not supported by compiler"
- fi
-}
-
-process_toolchain() {
- process_common_toolchain
-}
-
-print_config_mk() {
- local prefix=$1
- local makefile=$2
- shift 2
- for cfg; do
- upname="`toupper $cfg`"
- if enabled $cfg; then
- echo "${prefix}_${upname}=yes" >> $makefile
- fi
- done
-}
-
-print_config_h() {
- local prefix=$1
- local header=$2
- shift 2
- for cfg; do
- upname="`toupper $cfg`"
- if enabled $cfg; then
- echo "#define ${prefix}_${upname} 1" >> $header
- else
- echo "#define ${prefix}_${upname} 0" >> $header
- fi
- done
-}
-
-process_targets() {
- true;
-}
-
-process_detect() {
- true;
-}
-
-enable logging
-logfile="config.err"
-self=$0
-process() {
- cmdline_args="$@"
- process_cmdline "$@"
- if enabled child; then
- echo "# ${self} $@" >> ${logfile}
- else
- echo "# ${self} $@" > ${logfile}
- fi
- post_process_common_cmdline
- post_process_cmdline
- process_toolchain
- process_detect
- process_targets
-
- OOT_INSTALLS="${OOT_INSTALLS}"
- if enabled source_path_used; then
- # Prepare the PWD for building.
- for f in ${OOT_INSTALLS}; do
- install -D ${source_path}/$f $f
- done
- fi
- cp ${source_path}/build/make/Makefile .
-
- clean_temp_files
- true
-}
diff --git a/build/make/gen_asm_deps.sh b/build/make/gen_asm_deps.sh
deleted file mode 100755
index 7c6c5d5..0000000
--- a/build/make/gen_asm_deps.sh
+++ /dev/null
@@ -1,64 +0,0 @@
-#!/bin/bash
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-self=$0
-show_help() {
- echo "usage: $self [options] <srcfile>"
- echo
- echo "Generate Makefile dependency information from assembly code source"
- echo
- exit 1
-}
-die_unknown(){
- echo "Unknown option \"$1\"."
- echo "See $0 --help for available options."
- exit 1
-}
-for opt do
- optval="${opt#*=}"
- case "$opt" in
- --build-pfx=*) pfx="${optval}"
- ;;
- --depfile=*) out="${optval}"
- ;;
- -I*) raw_inc_paths="${raw_inc_paths} ${opt}"
- inc_path="${inc_path} ${opt#-I}"
- ;;
- -h|--help) show_help
- ;;
- *) [ -f "$opt" ] && srcfile="$opt"
- ;;
- esac
-done
-
-[ -n "$srcfile" ] || show_help
-sfx=${sfx:-asm}
-includes=$(egrep -i "include +\"?+[a-z0-9_/]+\.${sfx}" $srcfile |
- perl -p -e "s;.*?([a-z0-9_/]+.${sfx}).*;\1;")
-#" restore editor state
-for inc in ${includes}; do
- found_inc_path=
- for idir in ${inc_path}; do
- [ -f "${idir}/${inc}" ] && found_inc_path="${idir}" && break
- done
- if [ -f `dirname $srcfile`/$inc ]; then
- # Handle include files in the same directory as the source
- $self --build-pfx=$pfx --depfile=$out ${raw_inc_paths} `dirname $srcfile`/$inc
- elif [ -n "${found_inc_path}" ]; then
- # Handle include files on the include path
- $self --build-pfx=$pfx --depfile=$out ${raw_inc_paths} "${found_inc_path}/$inc"
- else
- # Handle generated includes in the build root (which may not exist yet)
- echo ${out} ${out%d}o: "${pfx}${inc}"
- fi
-done
-echo ${out} ${out%d}o: $srcfile
diff --git a/build/make/gen_msvs_proj.sh b/build/make/gen_msvs_proj.sh
deleted file mode 100755
index c2ef44a..0000000
--- a/build/make/gen_msvs_proj.sh
+++ /dev/null
@@ -1,567 +0,0 @@
-#!/bin/bash
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-self=$0
-self_basename=${self##*/}
-self_dirname=$(dirname "$0")
-EOL=$'\n'
-
-show_help() {
- cat <<EOF
-Usage: ${self_basename} --name=projname [options] file1 [file2 ...]
-
-This script generates a Visual Studio project file from a list of source
-code files.
-
-Options:
- --help Print this message
- --exe Generate a project for building an Application
- --lib Generate a project for creating a static library
- --static-crt Use the static C runtime (/MT)
- --target=isa-os-cc Target specifier (required)
- --out=filename Write output to a file [stdout]
- --name=project_name Name of the project (required)
- --proj-guid=GUID GUID to use for the project
- --module-def=filename File containing export definitions (for DLLs)
- --ver=version Version (7,8,9) of visual studio to generate for
- --src-path-bare=dir Path to root of source tree
- -Ipath/to/include Additional include directories
- -DFLAG[=value] Preprocessor macros to define
- -Lpath/to/lib Additional library search paths
- -llibname Library to link against
-EOF
- exit 1
-}
-
-die() {
- echo "${self_basename}: $@" >&2
- exit 1
-}
-
-die_unknown(){
- echo "Unknown option \"$1\"." >&2
- echo "See ${self_basename} --help for available options." >&2
- exit 1
-}
-
-generate_uuid() {
- local hex="0123456789ABCDEF"
- local i
- local uuid=""
- local j
- #93995380-89BD-4b04-88EB-625FBE52EBFB
- for ((i=0; i<32; i++)); do
- (( j = $RANDOM % 16 ))
- uuid="${uuid}${hex:$j:1}"
- done
- echo "${uuid:0:8}-${uuid:8:4}-${uuid:12:4}-${uuid:16:4}-${uuid:20:12}"
-}
-
-indent1=" "
-indent=""
-indent_push() {
- indent="${indent}${indent1}"
-}
-indent_pop() {
- indent="${indent%${indent1}}"
-}
-
-tag_attributes() {
- for opt in "$@"; do
- optval="${opt#*=}"
- [ -n "${optval}" ] ||
- die "Missing attribute value in '$opt' while generating $tag tag"
- echo "${indent}${opt%%=*}=\"${optval}\""
- done
-}
-
-open_tag() {
- local tag=$1
- shift
- if [ $# -ne 0 ]; then
- echo "${indent}<${tag}"
- indent_push
- tag_attributes "$@"
- echo "${indent}>"
- else
- echo "${indent}<${tag}>"
- indent_push
- fi
-}
-
-close_tag() {
- local tag=$1
- indent_pop
- echo "${indent}</${tag}>"
-}
-
-tag() {
- local tag=$1
- shift
- if [ $# -ne 0 ]; then
- echo "${indent}<${tag}"
- indent_push
- tag_attributes "$@"
- indent_pop
- echo "${indent}/>"
- else
- echo "${indent}<${tag}/>"
- fi
-}
-
-generate_filter() {
- local var=$1
- local name=$2
- local pats=$3
- local file_list_sz
- local i
- local f
- local saveIFS="$IFS"
- local pack
- echo "generating filter '$name' from ${#file_list[@]} files" >&2
- IFS=*
-
- open_tag Filter \
- Name=$name \
- Filter=$pats \
- UniqueIdentifier=`generate_uuid` \
-
- file_list_sz=${#file_list[@]}
- for i in ${!file_list[@]}; do
- f=${file_list[i]}
- for pat in ${pats//;/$IFS}; do
- if [ "${f##*.}" == "$pat" ]; then
- unset file_list[i]
-
- open_tag File RelativePath="./$f"
- if [ "$pat" == "asm" ] && $asm_use_custom_step; then
- for plat in "${platforms[@]}"; do
- for cfg in Debug Release; do
- open_tag FileConfiguration \
- Name="${cfg}|${plat}" \
-
- tag Tool \
- Name="VCCustomBuildTool" \
- Description="Assembling \$(InputFileName)" \
- CommandLine="$(eval echo \$asm_${cfg}_cmdline)" \
- Outputs="\$(InputName).obj" \
-
- close_tag FileConfiguration
- done
- done
- fi
-
- close_tag File
-
- break
- fi
- done
- done
-
- close_tag Filter
- IFS="$saveIFS"
-}
-
-# Process command line
-unset target
-for opt in "$@"; do
- optval="${opt#*=}"
- case "$opt" in
- --help|-h) show_help
- ;;
- --target=*) target="${optval}"
- ;;
- --out=*) outfile="$optval"
- ;;
- --name=*) name="${optval}"
- ;;
- --proj-guid=*) guid="${optval}"
- ;;
- --module-def=*) link_opts="${link_opts} ModuleDefinitionFile=${optval}"
- ;;
- --exe) proj_kind="exe"
- ;;
- --lib) proj_kind="lib"
- ;;
- --src-path-bare=*) src_path_bare="$optval"
- ;;
- --static-crt) use_static_runtime=true
- ;;
- --ver=*)
- vs_ver="$optval"
- case "$optval" in
- [789])
- ;;
- *) die Unrecognized Visual Studio Version in $opt
- ;;
- esac
- ;;
- -I*)
- opt="${opt%/}"
- incs="${incs}${incs:+;}"${opt##-I}""
- yasmincs="${yasmincs} ${opt}"
- ;;
- -D*) defines="${defines}${defines:+;}${opt##-D}"
- ;;
- -L*) # fudge . to $(OutDir)
- if [ "${opt##-L}" == "." ]; then
- libdirs="${libdirs}${libdirs:+;}"\$(OutDir)""
- else
- # Also try directories for this platform/configuration
- libdirs="${libdirs}${libdirs:+;}"${opt##-L}""
- libdirs="${libdirs}${libdirs:+;}"${opt##-L}/\$(PlatformName)/\$(ConfigurationName)""
- libdirs="${libdirs}${libdirs:+;}"${opt##-L}/\$(PlatformName)""
- fi
- ;;
- -l*) libs="${libs}${libs:+ }${opt##-l}.lib"
- ;;
- -*) die_unknown $opt
- ;;
- *)
- file_list[${#file_list[@]}]="$opt"
- case "$opt" in
- *.asm) uses_asm=true
- ;;
- esac
- ;;
- esac
-done
-outfile=${outfile:-/dev/stdout}
-guid=${guid:-`generate_uuid`}
-asm_use_custom_step=false
-uses_asm=${uses_asm:-false}
-case "${vs_ver:-8}" in
- 7) vs_ver_id="7.10"
- asm_use_custom_step=$uses_asm
- ;;
- 8) vs_ver_id="8.00"
- ;;
- 9) vs_ver_id="9.00"
- ;;
-esac
-
-[ -n "$name" ] || die "Project name (--name) must be specified!"
-[ -n "$target" ] || die "Target (--target) must be specified!"
-
-if ${use_static_runtime:-false}; then
- release_runtime=0
- debug_runtime=1
- lib_sfx=mt
-else
- release_runtime=2
- debug_runtime=3
- lib_sfx=md
-fi
-
-# Calculate debug lib names: If a lib ends in ${lib_sfx}.lib, then rename
-# it to ${lib_sfx}d.lib. This precludes linking to release libs from a
-# debug exe, so this may need to be refactored later.
-for lib in ${libs}; do
- if [ "$lib" != "${lib%${lib_sfx}.lib}" ]; then
- lib=${lib%.lib}d.lib
- fi
- debug_libs="${debug_libs}${debug_libs:+ }${lib}"
-done
-
-
-# List Keyword for this target
-case "$target" in
- x86*) keyword="ManagedCProj"
- ;;
- *) die "Unsupported target $target!"
-esac
-
-# List of all platforms supported for this target
-case "$target" in
- x86_64*)
- platforms[0]="x64"
- ;;
- x86*)
- platforms[0]="Win32"
- # these are only used by vs7
- asm_Debug_cmdline="yasm -Xvc -g cv8 -f \$(PlatformName) ${yasmincs} "\$(InputPath)""
- asm_Release_cmdline="yasm -Xvc -f \$(PlatformName) ${yasmincs} "\$(InputPath)""
- ;;
- *) die "Unsupported target $target!"
- ;;
-esac
-
-generate_vcproj() {
- case "$proj_kind" in
- exe) vs_ConfigurationType=1
- ;;
- *) vs_ConfigurationType=4
- ;;
- esac
-
- echo "<?xml version=\"1.0\" encoding=\"Windows-1252\"?>"
- open_tag VisualStudioProject \
- ProjectType="Visual C++" \
- Version="${vs_ver_id}" \
- Name="${name}" \
- ProjectGUID="{${guid}}" \
- RootNamespace="${name}" \
- Keyword="${keyword}" \
-
- open_tag Platforms
- for plat in "${platforms[@]}"; do
- tag Platform Name="$plat"
- done
- close_tag Platforms
-
- open_tag ToolFiles
- case "$target" in
- x86*) $uses_asm && tag ToolFile RelativePath="$self_dirname/../x86-msvs/yasm.rules"
- ;;
- esac
- close_tag ToolFiles
-
- open_tag Configurations
- for plat in "${platforms[@]}"; do
- plat_no_ws=`echo $plat | sed 's/[^A-Za-z0-9_]/_/g'`
- open_tag Configuration \
- Name="Debug|$plat" \
- OutputDirectory="\$(SolutionDir)$plat_no_ws/\$(ConfigurationName)" \
- IntermediateDirectory="$plat_no_ws/\$(ConfigurationName)/${name}" \
- ConfigurationType="$vs_ConfigurationType" \
- CharacterSet="1" \
-
- case "$target" in
- x86*)
- case "$name" in
- obj_int_extract)
- tag Tool \
- Name="VCCLCompilerTool" \
- Optimization="0" \
- AdditionalIncludeDirectories="$incs" \
- PreprocessorDefinitions="WIN32;DEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE" \
- RuntimeLibrary="$debug_runtime" \
- WarningLevel="3" \
- Detect64BitPortabilityProblems="true" \
- DebugInformationFormat="1" \
- ;;
- vpx)
- tag Tool \
- Name="VCPreBuildEventTool" \
- CommandLine="call obj_int_extract.bat $src_path_bare" \
-
- tag Tool \
- Name="VCCLCompilerTool" \
- Optimization="0" \
- AdditionalIncludeDirectories="$incs" \
- PreprocessorDefinitions="WIN32;_DEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
- RuntimeLibrary="$debug_runtime" \
- UsePrecompiledHeader="0" \
- WarningLevel="3" \
- DebugInformationFormat="1" \
- Detect64BitPortabilityProblems="true" \
-
- $uses_asm && tag Tool Name="YASM" IncludePaths="$incs" Debug="1"
- ;;
- *)
- tag Tool \
- Name="VCCLCompilerTool" \
- Optimization="0" \
- AdditionalIncludeDirectories="$incs" \
- PreprocessorDefinitions="WIN32;_DEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
- RuntimeLibrary="$debug_runtime" \
- UsePrecompiledHeader="0" \
- WarningLevel="3" \
- DebugInformationFormat="1" \
- Detect64BitPortabilityProblems="true" \
-
- $uses_asm && tag Tool Name="YASM" IncludePaths="$incs" Debug="1"
- ;;
- esac
- ;;
- esac
-
- case "$proj_kind" in
- exe)
- case "$target" in
- x86*)
- case "$name" in
- obj_int_extract)
- tag Tool \
- Name="VCLinkerTool" \
- OutputFile="${name}.exe" \
- GenerateDebugInformation="true" \
- ;;
- *)
- tag Tool \
- Name="VCLinkerTool" \
- AdditionalDependencies="$debug_libs \$(NoInherit)" \
- AdditionalLibraryDirectories="$libdirs" \
- GenerateDebugInformation="true" \
- ProgramDatabaseFile="\$(OutDir)/${name}.pdb" \
- ;;
- esac
- ;;
- esac
- ;;
- lib)
- case "$target" in
- x86*)
- tag Tool \
- Name="VCLibrarianTool" \
- OutputFile="\$(OutDir)/${name}${lib_sfx}d.lib" \
-
- ;;
- esac
- ;;
- dll)
- tag Tool \
- Name="VCLinkerTool" \
- AdditionalDependencies="\$(NoInherit)" \
- LinkIncremental="2" \
- GenerateDebugInformation="true" \
- AssemblyDebug="1" \
- TargetMachine="1" \
- $link_opts \
-
- ;;
- esac
-
- close_tag Configuration
-
- open_tag Configuration \
- Name="Release|$plat" \
- OutputDirectory="\$(SolutionDir)$plat_no_ws/\$(ConfigurationName)" \
- IntermediateDirectory="$plat_no_ws/\$(ConfigurationName)/${name}" \
- ConfigurationType="$vs_ConfigurationType" \
- CharacterSet="1" \
- WholeProgramOptimization="0" \
-
- case "$target" in
- x86*)
- case "$name" in
- obj_int_extract)
- tag Tool \
- Name="VCCLCompilerTool" \
- AdditionalIncludeDirectories="$incs" \
- PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE" \
- RuntimeLibrary="$release_runtime" \
- UsePrecompiledHeader="0" \
- WarningLevel="3" \
- Detect64BitPortabilityProblems="true" \
- DebugInformationFormat="0" \
- ;;
- vpx)
- tag Tool \
- Name="VCPreBuildEventTool" \
- CommandLine="call obj_int_extract.bat $src_path_bare" \
-
- tag Tool \
- Name="VCCLCompilerTool" \
- AdditionalIncludeDirectories="$incs" \
- PreprocessorDefinitions="WIN32;NDEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
- RuntimeLibrary="$release_runtime" \
- UsePrecompiledHeader="0" \
- WarningLevel="3" \
- DebugInformationFormat="0" \
- Detect64BitPortabilityProblems="true" \
-
- $uses_asm && tag Tool Name="YASM" IncludePaths="$incs"
- ;;
- *)
- tag Tool \
- Name="VCCLCompilerTool" \
- AdditionalIncludeDirectories="$incs" \
- PreprocessorDefinitions="WIN32;NDEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
- RuntimeLibrary="$release_runtime" \
- UsePrecompiledHeader="0" \
- WarningLevel="3" \
- DebugInformationFormat="0" \
- Detect64BitPortabilityProblems="true" \
-
- $uses_asm && tag Tool Name="YASM" IncludePaths="$incs"
- ;;
- esac
- ;;
- esac
-
- case "$proj_kind" in
- exe)
- case "$target" in
- x86*)
- case "$name" in
- obj_int_extract)
- tag Tool \
- Name="VCLinkerTool" \
- OutputFile="${name}.exe" \
- GenerateDebugInformation="true" \
- ;;
- *)
- tag Tool \
- Name="VCLinkerTool" \
- AdditionalDependencies="$libs \$(NoInherit)" \
- AdditionalLibraryDirectories="$libdirs" \
-
- ;;
- esac
- ;;
- esac
- ;;
- lib)
- case "$target" in
- x86*)
- tag Tool \
- Name="VCLibrarianTool" \
- OutputFile="\$(OutDir)/${name}${lib_sfx}.lib" \
-
- ;;
- esac
- ;;
- dll) # note differences to debug version: LinkIncremental, AssemblyDebug
- tag Tool \
- Name="VCLinkerTool" \
- AdditionalDependencies="\$(NoInherit)" \
- LinkIncremental="1" \
- GenerateDebugInformation="true" \
- TargetMachine="1" \
- $link_opts \
-
- ;;
- esac
-
- close_tag Configuration
- done
- close_tag Configurations
-
- open_tag Files
- generate_filter srcs "Source Files" "c;def;odl;idl;hpj;bat;asm;asmx"
- generate_filter hdrs "Header Files" "h;hm;inl;inc;xsd"
- generate_filter resrcs "Resource Files" "rc;ico;cur;bmp;dlg;rc2;rct;bin;rgs;gif;jpg;jpeg;jpe;resx;tiff;tif;png;wav"
- generate_filter resrcs "Build Files" "mk"
- close_tag Files
-
- tag Globals
- close_tag VisualStudioProject
-
- # This must be done from within the {} subshell
- echo "Ignored files list (${#file_list[@]} items) is:" >&2
- for f in "${file_list[@]}"; do
- echo " $f" >&2
- done
-}
-
-generate_vcproj |
- sed -e '/"/s;\([^ "]\)/;\1\\;g' > ${outfile}
-
-exit
-<!--
-TODO: Add any files not captured by filters.
- <File
- RelativePath=".\ReadMe.txt"
- >
- </File>
--->
diff --git a/build/make/obj_int_extract.c b/build/make/obj_int_extract.c
deleted file mode 100644
index c46d9d5..0000000
--- a/build/make/obj_int_extract.c
+++ /dev/null
@@ -1,1048 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdio.h>
-#include <stdlib.h>
-
-#include "vpx_config.h"
-
-#if defined(_MSC_VER) || defined(__MINGW32__)
-#include <io.h>
-#include <share.h>
-#include "vpx/vpx_integer.h"
-#else
-#include <stdint.h>
-#include <unistd.h>
-#endif
-
-#include <string.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <fcntl.h>
-#include <stdarg.h>
-
-typedef enum
-{
- OUTPUT_FMT_PLAIN,
- OUTPUT_FMT_RVDS,
- OUTPUT_FMT_GAS,
-} output_fmt_t;
-
-int log_msg(const char *fmt, ...)
-{
- int res;
- va_list ap;
- va_start(ap, fmt);
- res = vfprintf(stderr, fmt, ap);
- va_end(ap);
- return res;
-}
-
-#if defined(__GNUC__) && __GNUC__
-
-#if defined(__MACH__)
-
-#include <mach-o/loader.h>
-#include <mach-o/nlist.h>
-
-int parse_macho(uint8_t *base_buf, size_t sz)
-{
- int i, j;
- struct mach_header header;
- uint8_t *buf = base_buf;
- int base_data_section = 0;
- int bits = 0;
-
- /* We can read in mach_header for 32 and 64 bit architectures
- * because it's identical to mach_header_64 except for the last
- * element (uint32_t reserved), which we don't use. Then, when
- * we know which architecture we're looking at, increment buf
- * appropriately.
- */
- memcpy(&header, buf, sizeof(struct mach_header));
-
- if (header.magic == MH_MAGIC)
- {
- if (header.cputype == CPU_TYPE_ARM
- || header.cputype == CPU_TYPE_X86)
- {
- bits = 32;
- buf += sizeof(struct mach_header);
- }
- else
- {
- log_msg("Bad cputype for object file. Currently only tested for CPU_TYPE_[ARM|X86].\n");
- goto bail;
- }
- }
- else if (header.magic == MH_MAGIC_64)
- {
- if (header.cputype == CPU_TYPE_X86_64)
- {
- bits = 64;
- buf += sizeof(struct mach_header_64);
- }
- else
- {
- log_msg("Bad cputype for object file. Currently only tested for CPU_TYPE_X86_64.\n");
- goto bail;
- }
- }
- else
- {
- log_msg("Bad magic number for object file. 0x%x or 0x%x expected, 0x%x found.\n",
- MH_MAGIC, MH_MAGIC_64, header.magic);
- goto bail;
- }
-
- if (header.filetype != MH_OBJECT)
- {
- log_msg("Bad filetype for object file. Currently only tested for MH_OBJECT.\n");
- goto bail;
- }
-
- for (i = 0; i < header.ncmds; i++)
- {
- struct load_command lc;
-
- memcpy(&lc, buf, sizeof(struct load_command));
-
- if (lc.cmd == LC_SEGMENT)
- {
- uint8_t *seg_buf = buf;
- struct section s;
- struct segment_command seg_c;
-
- memcpy(&seg_c, seg_buf, sizeof(struct segment_command));
- seg_buf += sizeof(struct segment_command);
-
- /* Although each section is given it's own offset, nlist.n_value
- * references the offset of the first section. This isn't
- * apparent without debug information because the offset of the
- * data section is the same as the first section. However, with
- * debug sections mixed in, the offset of the debug section
- * increases but n_value still references the first section.
- */
- if (seg_c.nsects < 1)
- {
- log_msg("Not enough sections\n");
- goto bail;
- }
-
- memcpy(&s, seg_buf, sizeof(struct section));
- base_data_section = s.offset;
- }
- else if (lc.cmd == LC_SEGMENT_64)
- {
- uint8_t *seg_buf = buf;
- struct section_64 s;
- struct segment_command_64 seg_c;
-
- memcpy(&seg_c, seg_buf, sizeof(struct segment_command_64));
- seg_buf += sizeof(struct segment_command_64);
-
- /* Explanation in LG_SEGMENT */
- if (seg_c.nsects < 1)
- {
- log_msg("Not enough sections\n");
- goto bail;
- }
-
- memcpy(&s, seg_buf, sizeof(struct section_64));
- base_data_section = s.offset;
- }
- else if (lc.cmd == LC_SYMTAB)
- {
- if (base_data_section != 0)
- {
- struct symtab_command sc;
- uint8_t *sym_buf = base_buf;
- uint8_t *str_buf = base_buf;
-
- memcpy(&sc, buf, sizeof(struct symtab_command));
-
- if (sc.cmdsize != sizeof(struct symtab_command))
- {
- log_msg("Can't find symbol table!\n");
- goto bail;
- }
-
- sym_buf += sc.symoff;
- str_buf += sc.stroff;
-
- for (j = 0; j < sc.nsyms; j++)
- {
- /* Location of string is cacluated each time from the
- * start of the string buffer. On darwin the symbols
- * are prefixed by "_", so we bump the pointer by 1.
- * The target value is defined as an int in asm_*_offsets.c,
- * which is 4 bytes on all targets we currently use.
- */
- if (bits == 32)
- {
- struct nlist nl;
- int val;
-
- memcpy(&nl, sym_buf, sizeof(struct nlist));
- sym_buf += sizeof(struct nlist);
-
- memcpy(&val, base_buf + base_data_section + nl.n_value,
- sizeof(val));
- printf("%-40s EQU %5d\n",
- str_buf + nl.n_un.n_strx + 1, val);
- }
- else /* if (bits == 64) */
- {
- struct nlist_64 nl;
- int val;
-
- memcpy(&nl, sym_buf, sizeof(struct nlist_64));
- sym_buf += sizeof(struct nlist_64);
-
- memcpy(&val, base_buf + base_data_section + nl.n_value,
- sizeof(val));
- printf("%-40s EQU %5d\n",
- str_buf + nl.n_un.n_strx + 1, val);
- }
- }
- }
- }
-
- buf += lc.cmdsize;
- }
-
- return 0;
-bail:
- return 1;
-
-}
-
-int main(int argc, char **argv)
-{
- int fd;
- char *f;
- struct stat stat_buf;
- uint8_t *file_buf;
- int res;
-
- if (argc < 2 || argc > 3)
- {
- fprintf(stderr, "Usage: %s [output format] <obj file>\n\n", argv[0]);
- fprintf(stderr, " <obj file>\tMachO format object file to parse\n");
- fprintf(stderr, "Output Formats:\n");
- fprintf(stderr, " gas - compatible with GNU assembler\n");
- fprintf(stderr, " rvds - compatible with armasm\n");
- goto bail;
- }
-
- f = argv[2];
-
- if (!((!strcmp(argv[1], "rvds")) || (!strcmp(argv[1], "gas"))))
- f = argv[1];
-
- fd = open(f, O_RDONLY);
-
- if (fd < 0)
- {
- perror("Unable to open file");
- goto bail;
- }
-
- if (fstat(fd, &stat_buf))
- {
- perror("stat");
- goto bail;
- }
-
- file_buf = malloc(stat_buf.st_size);
-
- if (!file_buf)
- {
- perror("malloc");
- goto bail;
- }
-
- if (read(fd, file_buf, stat_buf.st_size) != stat_buf.st_size)
- {
- perror("read");
- goto bail;
- }
-
- if (close(fd))
- {
- perror("close");
- goto bail;
- }
-
- res = parse_macho(file_buf, stat_buf.st_size);
- free(file_buf);
-
- if (!res)
- return EXIT_SUCCESS;
-
-bail:
- return EXIT_FAILURE;
-}
-
-#elif defined(__ELF__)
-#include "elf.h"
-
-#define COPY_STRUCT(dst, buf, ofst, sz) do {\
- if(ofst + sizeof((*(dst))) > sz) goto bail;\
- memcpy(dst, buf+ofst, sizeof((*(dst))));\
- } while(0)
-
-#define ENDIAN_ASSIGN(val, memb) do {\
- if(!elf->le_data) {log_msg("Big Endian data not supported yet!\n");goto bail;}\
- (val) = (memb);\
- } while(0)
-
-#define ENDIAN_ASSIGN_IN_PLACE(memb) do {\
- ENDIAN_ASSIGN(memb, memb);\
- } while(0)
-
-typedef struct
-{
- uint8_t *buf; /* Buffer containing ELF data */
- size_t sz; /* Buffer size */
- int le_data; /* Data is little-endian */
- unsigned char e_ident[EI_NIDENT]; /* Magic number and other info */
- int bits; /* 32 or 64 */
- Elf32_Ehdr hdr32;
- Elf64_Ehdr hdr64;
-} elf_obj_t;
-
-int parse_elf_header(elf_obj_t *elf)
-{
- int res;
- /* Verify ELF Magic numbers */
- COPY_STRUCT(&elf->e_ident, elf->buf, 0, elf->sz);
- res = elf->e_ident[EI_MAG0] == ELFMAG0;
- res &= elf->e_ident[EI_MAG1] == ELFMAG1;
- res &= elf->e_ident[EI_MAG2] == ELFMAG2;
- res &= elf->e_ident[EI_MAG3] == ELFMAG3;
- res &= elf->e_ident[EI_CLASS] == ELFCLASS32
- || elf->e_ident[EI_CLASS] == ELFCLASS64;
- res &= elf->e_ident[EI_DATA] == ELFDATA2LSB;
-
- if (!res) goto bail;
-
- elf->le_data = elf->e_ident[EI_DATA] == ELFDATA2LSB;
-
- /* Read in relevant values */
- if (elf->e_ident[EI_CLASS] == ELFCLASS32)
- {
- elf->bits = 32;
- COPY_STRUCT(&elf->hdr32, elf->buf, 0, elf->sz);
-
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_type);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_machine);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_version);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_entry);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phoff);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shoff);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_flags);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_ehsize);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phentsize);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phnum);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shentsize);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shnum);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shstrndx);
- }
- else /* if (elf->e_ident[EI_CLASS] == ELFCLASS64) */
- {
- elf->bits = 64;
- COPY_STRUCT(&elf->hdr64, elf->buf, 0, elf->sz);
-
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_type);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_machine);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_version);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_entry);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phoff);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shoff);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_flags);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_ehsize);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phentsize);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phnum);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shentsize);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shnum);
- ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shstrndx);
- }
-
- return 0;
-bail:
- log_msg("Failed to parse ELF file header");
- return 1;
-}
-
-int parse_elf_section(elf_obj_t *elf, int idx, Elf32_Shdr *hdr32, Elf64_Shdr *hdr64)
-{
- if (hdr32)
- {
- if (idx >= elf->hdr32.e_shnum)
- goto bail;
-
- COPY_STRUCT(hdr32, elf->buf, elf->hdr32.e_shoff + idx * elf->hdr32.e_shentsize,
- elf->sz);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_name);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_type);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_flags);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_addr);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_offset);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_size);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_link);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_info);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_addralign);
- ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_entsize);
- }
- else /* if (hdr64) */
- {
- if (idx >= elf->hdr64.e_shnum)
- goto bail;
-
- COPY_STRUCT(hdr64, elf->buf, elf->hdr64.e_shoff + idx * elf->hdr64.e_shentsize,
- elf->sz);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_name);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_type);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_flags);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_addr);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_offset);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_size);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_link);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_info);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_addralign);
- ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_entsize);
- }
-
- return 0;
-bail:
- return 1;
-}
-
-char *parse_elf_string_table(elf_obj_t *elf, int s_idx, int idx)
-{
- if (elf->bits == 32)
- {
- Elf32_Shdr shdr;
-
- if (parse_elf_section(elf, s_idx, &shdr, NULL))
- {
- log_msg("Failed to parse ELF string table: section %d, index %d\n",
- s_idx, idx);
- return "";
- }
-
- return (char *)(elf->buf + shdr.sh_offset + idx);
- }
- else /* if (elf->bits == 64) */
- {
- Elf64_Shdr shdr;
-
- if (parse_elf_section(elf, s_idx, NULL, &shdr))
- {
- log_msg("Failed to parse ELF string table: section %d, index %d\n",
- s_idx, idx);
- return "";
- }
-
- return (char *)(elf->buf + shdr.sh_offset + idx);
- }
-}
-
-int parse_elf_symbol(elf_obj_t *elf, unsigned int ofst, Elf32_Sym *sym32, Elf64_Sym *sym64)
-{
- if (sym32)
- {
- COPY_STRUCT(sym32, elf->buf, ofst, elf->sz);
- ENDIAN_ASSIGN_IN_PLACE(sym32->st_name);
- ENDIAN_ASSIGN_IN_PLACE(sym32->st_value);
- ENDIAN_ASSIGN_IN_PLACE(sym32->st_size);
- ENDIAN_ASSIGN_IN_PLACE(sym32->st_info);
- ENDIAN_ASSIGN_IN_PLACE(sym32->st_other);
- ENDIAN_ASSIGN_IN_PLACE(sym32->st_shndx);
- }
- else /* if (sym64) */
- {
- COPY_STRUCT(sym64, elf->buf, ofst, elf->sz);
- ENDIAN_ASSIGN_IN_PLACE(sym64->st_name);
- ENDIAN_ASSIGN_IN_PLACE(sym64->st_value);
- ENDIAN_ASSIGN_IN_PLACE(sym64->st_size);
- ENDIAN_ASSIGN_IN_PLACE(sym64->st_info);
- ENDIAN_ASSIGN_IN_PLACE(sym64->st_other);
- ENDIAN_ASSIGN_IN_PLACE(sym64->st_shndx);
- }
- return 0;
-bail:
- return 1;
-}
-
-int parse_elf(uint8_t *buf, size_t sz, output_fmt_t mode)
-{
- elf_obj_t elf;
- unsigned int ofst;
- int i;
- Elf32_Off strtab_off32;
- Elf64_Off strtab_off64; /* save String Table offset for later use */
-
- memset(&elf, 0, sizeof(elf));
- elf.buf = buf;
- elf.sz = sz;
-
- /* Parse Header */
- if (parse_elf_header(&elf))
- goto bail;
-
- if (elf.bits == 32)
- {
- Elf32_Shdr shdr;
- for (i = 0; i < elf.hdr32.e_shnum; i++)
- {
- parse_elf_section(&elf, i, &shdr, NULL);
-
- if (shdr.sh_type == SHT_STRTAB)
- {
- char strtsb_name[128];
-
- strcpy(strtsb_name, (char *)(elf.buf + shdr.sh_offset + shdr.sh_name));
-
- if (!(strcmp(strtsb_name, ".shstrtab")))
- {
- /* log_msg("found section: %s\n", strtsb_name); */
- strtab_off32 = shdr.sh_offset;
- break;
- }
- }
- }
- }
- else /* if (elf.bits == 64) */
- {
- Elf64_Shdr shdr;
- for (i = 0; i < elf.hdr64.e_shnum; i++)
- {
- parse_elf_section(&elf, i, NULL, &shdr);
-
- if (shdr.sh_type == SHT_STRTAB)
- {
- char strtsb_name[128];
-
- strcpy(strtsb_name, (char *)(elf.buf + shdr.sh_offset + shdr.sh_name));
-
- if (!(strcmp(strtsb_name, ".shstrtab")))
- {
- /* log_msg("found section: %s\n", strtsb_name); */
- strtab_off64 = shdr.sh_offset;
- break;
- }
- }
- }
- }
-
- /* Parse all Symbol Tables */
- if (elf.bits == 32)
- {
- Elf32_Shdr shdr;
- for (i = 0; i < elf.hdr32.e_shnum; i++)
- {
- parse_elf_section(&elf, i, &shdr, NULL);
-
- if (shdr.sh_type == SHT_SYMTAB)
- {
- for (ofst = shdr.sh_offset;
- ofst < shdr.sh_offset + shdr.sh_size;
- ofst += shdr.sh_entsize)
- {
- Elf32_Sym sym;
-
- parse_elf_symbol(&elf, ofst, &sym, NULL);
-
- /* For all OBJECTS (data objects), extract the value from the
- * proper data segment.
- */
- /* if (ELF32_ST_TYPE(sym.st_info) == STT_OBJECT && sym.st_name)
- log_msg("found data object %s\n",
- parse_elf_string_table(&elf,
- shdr.sh_link,
- sym.st_name));
- */
-
- if (ELF32_ST_TYPE(sym.st_info) == STT_OBJECT
- && sym.st_size == 4)
- {
- Elf32_Shdr dhdr;
- int val = 0;
- char section_name[128];
-
- parse_elf_section(&elf, sym.st_shndx, &dhdr, NULL);
-
- /* For explanition - refer to _MSC_VER version of code */
- strcpy(section_name, (char *)(elf.buf + strtab_off32 + dhdr.sh_name));
- /* log_msg("Section_name: %s, Section_type: %d\n", section_name, dhdr.sh_type); */
-
- if (strcmp(section_name, ".bss"))
- {
- if (sizeof(val) != sym.st_size)
- {
- /* The target value is declared as an int in
- * asm_*_offsets.c, which is 4 bytes on all
- * targets we currently use. Complain loudly if
- * this is not true.
- */
- log_msg("Symbol size is wrong\n");
- goto bail;
- }
-
- memcpy(&val,
- elf.buf + dhdr.sh_offset + sym.st_value,
- sym.st_size);
- }
-
- if (!elf.le_data)
- {
- log_msg("Big Endian data not supported yet!\n");
- goto bail;
- }
-
- switch (mode)
- {
- case OUTPUT_FMT_RVDS:
- printf("%-40s EQU %5d\n",
- parse_elf_string_table(&elf,
- shdr.sh_link,
- sym.st_name),
- val);
- break;
- case OUTPUT_FMT_GAS:
- printf(".equ %-40s, %5d\n",
- parse_elf_string_table(&elf,
- shdr.sh_link,
- sym.st_name),
- val);
- break;
- default:
- printf("%s = %d\n",
- parse_elf_string_table(&elf,
- shdr.sh_link,
- sym.st_name),
- val);
- }
- }
- }
- }
- }
- }
- else /* if (elf.bits == 64) */
- {
- Elf64_Shdr shdr;
- for (i = 0; i < elf.hdr64.e_shnum; i++)
- {
- parse_elf_section(&elf, i, NULL, &shdr);
-
- if (shdr.sh_type == SHT_SYMTAB)
- {
- for (ofst = shdr.sh_offset;
- ofst < shdr.sh_offset + shdr.sh_size;
- ofst += shdr.sh_entsize)
- {
- Elf64_Sym sym;
-
- parse_elf_symbol(&elf, ofst, NULL, &sym);
-
- /* For all OBJECTS (data objects), extract the value from the
- * proper data segment.
- */
- /* if (ELF64_ST_TYPE(sym.st_info) == STT_OBJECT && sym.st_name)
- log_msg("found data object %s\n",
- parse_elf_string_table(&elf,
- shdr.sh_link,
- sym.st_name));
- */
-
- if (ELF64_ST_TYPE(sym.st_info) == STT_OBJECT
- && sym.st_size == 4)
- {
- Elf64_Shdr dhdr;
- int val = 0;
- char section_name[128];
-
- parse_elf_section(&elf, sym.st_shndx, NULL, &dhdr);
-
- /* For explanition - refer to _MSC_VER version of code */
- strcpy(section_name, (char *)(elf.buf + strtab_off64 + dhdr.sh_name));
- /* log_msg("Section_name: %s, Section_type: %d\n", section_name, dhdr.sh_type); */
-
- if ((strcmp(section_name, ".bss")))
- {
- if (sizeof(val) != sym.st_size)
- {
- /* The target value is declared as an int in
- * asm_*_offsets.c, which is 4 bytes on all
- * targets we currently use. Complain loudly if
- * this is not true.
- */
- log_msg("Symbol size is wrong\n");
- goto bail;
- }
-
- memcpy(&val,
- elf.buf + dhdr.sh_offset + sym.st_value,
- sym.st_size);
- }
-
- if (!elf.le_data)
- {
- log_msg("Big Endian data not supported yet!\n");
- goto bail;
- }
-
- switch (mode)
- {
- case OUTPUT_FMT_RVDS:
- printf("%-40s EQU %5d\n",
- parse_elf_string_table(&elf,
- shdr.sh_link,
- sym.st_name),
- val);
- break;
- case OUTPUT_FMT_GAS:
- printf(".equ %-40s, %5d\n",
- parse_elf_string_table(&elf,
- shdr.sh_link,
- sym.st_name),
- val);
- break;
- default:
- printf("%s = %d\n",
- parse_elf_string_table(&elf,
- shdr.sh_link,
- sym.st_name),
- val);
- }
- }
- }
- }
- }
- }
-
- if (mode == OUTPUT_FMT_RVDS)
- printf(" END\n");
-
- return 0;
-bail:
- log_msg("Parse error: File does not appear to be valid ELF32 or ELF64\n");
- return 1;
-}
-
-int main(int argc, char **argv)
-{
- int fd;
- output_fmt_t mode;
- char *f;
- struct stat stat_buf;
- uint8_t *file_buf;
- int res;
-
- if (argc < 2 || argc > 3)
- {
- fprintf(stderr, "Usage: %s [output format] <obj file>\n\n", argv[0]);
- fprintf(stderr, " <obj file>\tELF format object file to parse\n");
- fprintf(stderr, "Output Formats:\n");
- fprintf(stderr, " gas - compatible with GNU assembler\n");
- fprintf(stderr, " rvds - compatible with armasm\n");
- goto bail;
- }
-
- f = argv[2];
-
- if (!strcmp(argv[1], "rvds"))
- mode = OUTPUT_FMT_RVDS;
- else if (!strcmp(argv[1], "gas"))
- mode = OUTPUT_FMT_GAS;
- else
- f = argv[1];
-
-
- fd = open(f, O_RDONLY);
-
- if (fd < 0)
- {
- perror("Unable to open file");
- goto bail;
- }
-
- if (fstat(fd, &stat_buf))
- {
- perror("stat");
- goto bail;
- }
-
- file_buf = malloc(stat_buf.st_size);
-
- if (!file_buf)
- {
- perror("malloc");
- goto bail;
- }
-
- if (read(fd, file_buf, stat_buf.st_size) != stat_buf.st_size)
- {
- perror("read");
- goto bail;
- }
-
- if (close(fd))
- {
- perror("close");
- goto bail;
- }
-
- res = parse_elf(file_buf, stat_buf.st_size, mode);
- free(file_buf);
-
- if (!res)
- return EXIT_SUCCESS;
-
-bail:
- return EXIT_FAILURE;
-}
-#endif
-#endif
-
-
-#if defined(_MSC_VER) || defined(__MINGW32__)
-/* See "Microsoft Portable Executable and Common Object File Format Specification"
- for reference.
-*/
-#define get_le32(x) ((*(x)) | (*(x+1)) << 8 |(*(x+2)) << 16 | (*(x+3)) << 24 )
-#define get_le16(x) ((*(x)) | (*(x+1)) << 8)
-
-int parse_coff(unsigned __int8 *buf, size_t sz)
-{
- unsigned int nsections, symtab_ptr, symtab_sz, strtab_ptr;
- unsigned int sectionrawdata_ptr;
- unsigned int i;
- unsigned __int8 *ptr;
- unsigned __int32 symoffset;
-
- char **sectionlist; //this array holds all section names in their correct order.
- //it is used to check if the symbol is in .bss or .data section.
-
- nsections = get_le16(buf + 2);
- symtab_ptr = get_le32(buf + 8);
- symtab_sz = get_le32(buf + 12);
- strtab_ptr = symtab_ptr + symtab_sz * 18;
-
- if (nsections > 96)
- {
- log_msg("Too many sections\n");
- return 1;
- }
-
- sectionlist = malloc(nsections * sizeof(sectionlist));
-
- if (sectionlist == NULL)
- {
- log_msg("Allocating first level of section list failed\n");
- return 1;
- }
-
- //log_msg("COFF: Found %u symbols in %u sections.\n", symtab_sz, nsections);
-
- /*
- The size of optional header is always zero for an obj file. So, the section header
- follows the file header immediately.
- */
-
- ptr = buf + 20; //section header
-
- for (i = 0; i < nsections; i++)
- {
- char sectionname[9] = {0, 0, 0, 0, 0, 0, 0, 0, 0};
- strncpy(sectionname, ptr, 8);
- //log_msg("COFF: Parsing section %s\n",sectionname);
-
- sectionlist[i] = malloc(strlen(sectionname) + 1);
-
- if (sectionlist[i] == NULL)
- {
- log_msg("Allocating storage for %s failed\n", sectionname);
- goto bail;
- }
- strcpy(sectionlist[i], sectionname);
-
- if (!strcmp(sectionname, ".data")) sectionrawdata_ptr = get_le32(ptr + 20);
-
- ptr += 40;
- }
-
- //log_msg("COFF: Symbol table at offset %u\n", symtab_ptr);
- //log_msg("COFF: raw data pointer ofset for section .data is %u\n", sectionrawdata_ptr);
-
- /* The compiler puts the data with non-zero offset in .data section, but puts the data with
- zero offset in .bss section. So, if the data in in .bss section, set offset=0.
- Note from Wiki: In an object module compiled from C, the bss section contains
- the local variables (but not functions) that were declared with the static keyword,
- except for those with non-zero initial values. (In C, static variables are initialized
- to zero by default.) It also contains the non-local (both extern and static) variables
- that are also initialized to zero (either explicitly or by default).
- */
- //move to symbol table
- /* COFF symbol table:
- offset field
- 0 Name(*)
- 8 Value
- 12 SectionNumber
- 14 Type
- 16 StorageClass
- 17 NumberOfAuxSymbols
- */
- ptr = buf + symtab_ptr;
-
- for (i = 0; i < symtab_sz; i++)
- {
- __int16 section = get_le16(ptr + 12); //section number
-
- if (section > 0 && ptr[16] == 2)
- {
- //if(section > 0 && ptr[16] == 3 && get_le32(ptr+8)) {
-
- if (get_le32(ptr))
- {
- char name[9] = {0, 0, 0, 0, 0, 0, 0, 0, 0};
- strncpy(name, ptr, 8);
- //log_msg("COFF: Parsing symbol %s\n",name);
- /* The 64bit Windows compiler doesn't prefix with an _.
- * Check what's there, and bump if necessary
- */
- if (name[0] == '_')
- printf("%-40s EQU ", name + 1);
- else
- printf("%-40s EQU ", name);
- }
- else
- {
- //log_msg("COFF: Parsing symbol %s\n",
- // buf + strtab_ptr + get_le32(ptr+4));
- if ((buf + strtab_ptr + get_le32(ptr + 4))[0] == '_')
- printf("%-40s EQU ",
- buf + strtab_ptr + get_le32(ptr + 4) + 1);
- else
- printf("%-40s EQU ", buf + strtab_ptr + get_le32(ptr + 4));
- }
-
- if (!(strcmp(sectionlist[section-1], ".bss")))
- {
- symoffset = 0;
- }
- else
- {
- symoffset = get_le32(buf + sectionrawdata_ptr + get_le32(ptr + 8));
- }
-
- //log_msg(" Section: %d\n",section);
- //log_msg(" Class: %d\n",ptr[16]);
- //log_msg(" Address: %u\n",get_le32(ptr+8));
- //log_msg(" Offset: %u\n", symoffset);
-
- printf("%5d\n", symoffset);
- }
-
- ptr += 18;
- }
-
- printf(" END\n");
-
- for (i = 0; i < nsections; i++)
- {
- free(sectionlist[i]);
- }
-
- free(sectionlist);
-
- return 0;
-bail:
-
- for (i = 0; i < nsections; i++)
- {
- free(sectionlist[i]);
- }
-
- free(sectionlist);
-
- return 1;
-}
-
-int main(int argc, char **argv)
-{
- int fd;
- output_fmt_t mode;
- const char *f;
- struct _stat stat_buf;
- unsigned __int8 *file_buf;
- int res;
-
- if (argc < 2 || argc > 3)
- {
- fprintf(stderr, "Usage: %s [output format] <obj file>\n\n", argv[0]);
- fprintf(stderr, " <obj file>\tELF format object file to parse\n");
- fprintf(stderr, "Output Formats:\n");
- fprintf(stderr, " gas - compatible with GNU assembler\n");
- fprintf(stderr, " rvds - compatible with armasm\n");
- goto bail;
- }
-
- f = argv[2];
-
- if (!strcmp(argv[1], "rvds"))
- mode = OUTPUT_FMT_RVDS;
- else if (!strcmp(argv[1], "gas"))
- mode = OUTPUT_FMT_GAS;
- else
- f = argv[1];
-
- fd = _sopen(f, _O_BINARY, _SH_DENYNO, _S_IREAD | _S_IWRITE);
-
- if (_fstat(fd, &stat_buf))
- {
- perror("stat");
- goto bail;
- }
-
- file_buf = malloc(stat_buf.st_size);
-
- if (!file_buf)
- {
- perror("malloc");
- goto bail;
- }
-
- if (_read(fd, file_buf, stat_buf.st_size) != stat_buf.st_size)
- {
- perror("read");
- goto bail;
- }
-
- if (_close(fd))
- {
- perror("close");
- goto bail;
- }
-
- res = parse_coff(file_buf, stat_buf.st_size);
-
- free(file_buf);
-
- if (!res)
- return EXIT_SUCCESS;
-
-bail:
- return EXIT_FAILURE;
-}
-#endif
diff --git a/configure b/configure
deleted file mode 100755
index ab3936d..0000000
--- a/configure
+++ /dev/null
@@ -1,559 +0,0 @@
-#!/bin/bash
-##
-## configure
-##
-## This script is the front-end to the build system. It provides a similar
-## interface to standard configure scripts with some extra bits for dealing
-## with toolchains that differ from the standard POSIX interface and
-## for extracting subsets of the source tree. In theory, reusable parts
-## of this script were intended to live in build/make/configure.sh,
-## but in practice, the line is pretty blurry.
-##
-## This build system is based in part on the FFmpeg configure script.
-##
-
-#source_path="`dirname \"$0\"`"
-source_path=${0%/*}
-. "${source_path}/build/make/configure.sh"
-
-show_help(){
- show_help_pre
- cat << EOF
-Advanced options:
- ${toggle_libs} don't build libraries
- ${toggle_examples} don't build examples
- --libc=PATH path to alternate libc
- --as={yasm|nasm|auto} use specified assembler [auto, yasm preferred]
- ${toggle_fast_unaligned} don't use unaligned accesses, even when
- supported by hardware [auto]
- ${toggle_codec_srcs} in/exclude codec library source code
- ${toggle_debug_libs} in/exclude debug version of libraries
- ${toggle_md5} support for output of checksum data
- ${toggle_static_msvcrt} use static MSVCRT (VS builds only)
- ${toggle_vp8} VP8 codec support
- ${toggle_psnr} output of PSNR data, if supported (encoders)
- ${toggle_mem_tracker} track memory usage
- ${toggle_postproc} postprocessing
- ${toggle_multithread} multithreaded encoding and decoding.
- ${toggle_spatial_resampling} spatial sampling (scaling) support
- ${toggle_realtime_only} enable this option while building for real-time encoding
- ${toggle_runtime_cpu_detect} runtime cpu detection
- ${toggle_shared} shared library support
- ${toggle_small} favor smaller size over speed
- ${toggle_postproc_visualizer} macro block / block level visualizers
-
-Codecs:
- Codecs can be selectively enabled or disabled individually, or by family:
- --disable-<codec>
- is equivalent to:
- --disable-<codec>-encoder
- --disable-<codec>-decoder
-
- Codecs available in this distribution:
-EOF
-#restore editor state '
-
- local family;
- local last_family;
- local c;
- local str;
- for c in ${CODECS}; do
- family=${c%_*}
- if [ "${family}" != "${last_family}" ]; then
- [ -z "${str}" ] || echo "${str}"
- str="$(printf ' %10s:' ${family})"
- fi
- str="${str} $(printf '%10s' ${c#*_})"
- last_family=${family}
- done
- echo "${str}"
- show_help_post
-}
-
-##
-## BEGIN APPLICATION SPECIFIC CONFIGURATION
-##
-
-# all_platforms is a list of all supported target platforms. Maintain
-# alphabetically by architecture, generic-gnu last.
-all_platforms="${all_platforms} armv5te-linux-rvct"
-all_platforms="${all_platforms} armv5te-linux-gcc"
-all_platforms="${all_platforms} armv5te-none-rvct"
-all_platforms="${all_platforms} armv5te-symbian-gcc"
-all_platforms="${all_platforms} armv6-darwin-gcc"
-all_platforms="${all_platforms} armv6-linux-rvct"
-all_platforms="${all_platforms} armv6-linux-gcc"
-all_platforms="${all_platforms} armv6-none-rvct"
-all_platforms="${all_platforms} armv6-symbian-gcc"
-all_platforms="${all_platforms} iwmmxt-linux-rvct"
-all_platforms="${all_platforms} iwmmxt-linux-gcc"
-all_platforms="${all_platforms} iwmmxt2-linux-rvct"
-all_platforms="${all_platforms} iwmmxt2-linux-gcc"
-all_platforms="${all_platforms} armv7-darwin-gcc" #neon Cortex-A8
-all_platforms="${all_platforms} armv7-linux-rvct" #neon Cortex-A8
-all_platforms="${all_platforms} armv7-linux-gcc" #neon Cortex-A8
-all_platforms="${all_platforms} armv7-none-rvct" #neon Cortex-A8
-all_platforms="${all_platforms} mips32-linux-gcc"
-all_platforms="${all_platforms} ppc32-darwin8-gcc"
-all_platforms="${all_platforms} ppc32-darwin9-gcc"
-all_platforms="${all_platforms} ppc32-linux-gcc"
-all_platforms="${all_platforms} ppc64-darwin8-gcc"
-all_platforms="${all_platforms} ppc64-darwin9-gcc"
-all_platforms="${all_platforms} ppc64-linux-gcc"
-all_platforms="${all_platforms} sparc-solaris-gcc"
-all_platforms="${all_platforms} x86-darwin8-gcc"
-all_platforms="${all_platforms} x86-darwin8-icc"
-all_platforms="${all_platforms} x86-darwin9-gcc"
-all_platforms="${all_platforms} x86-darwin9-icc"
-all_platforms="${all_platforms} x86-linux-gcc"
-all_platforms="${all_platforms} x86-linux-icc"
-all_platforms="${all_platforms} x86-solaris-gcc"
-all_platforms="${all_platforms} x86-win32-gcc"
-all_platforms="${all_platforms} x86-win32-vs7"
-all_platforms="${all_platforms} x86-win32-vs8"
-all_platforms="${all_platforms} x86-win32-vs9"
-all_platforms="${all_platforms} x86_64-darwin9-gcc"
-all_platforms="${all_platforms} x86_64-darwin10-gcc"
-all_platforms="${all_platforms} x86_64-linux-gcc"
-all_platforms="${all_platforms} x86_64-linux-icc"
-all_platforms="${all_platforms} x86_64-solaris-gcc"
-all_platforms="${all_platforms} x86_64-win64-vs8"
-all_platforms="${all_platforms} x86_64-win64-vs9"
-all_platforms="${all_platforms} universal-darwin8-gcc"
-all_platforms="${all_platforms} universal-darwin9-gcc"
-all_platforms="${all_platforms} generic-gnu"
-
-# all_targets is a list of all targets that can be configured
-# note that these should be in dependency order for now.
-all_targets="libs examples docs"
-
-# all targets available are enabled, by default.
-for t in ${all_targets}; do
- [ -f ${source_path}/${t}.mk ] && enable ${t}
-done
-
-# check installed doxygen version
-doxy_version=$(doxygen --version 2>/dev/null)
-doxy_major=${doxy_version%%.*}
-if [ ${doxy_major:-0} -ge 1 ]; then
- doxy_version=${doxy_version#*.}
- doxy_minor=${doxy_version%%.*}
- doxy_patch=${doxy_version##*.}
-
- [ $doxy_major -gt 1 ] && enable doxygen
- [ $doxy_minor -gt 5 ] && enable doxygen
- [ $doxy_minor -eq 5 ] && [ $doxy_patch -ge 3 ] && enable doxygen
-fi
-
-# install everything except the sources, by default. sources will have
-# to be enabled when doing dist builds, since that's no longer a common
-# case.
-enabled doxygen && php -v >/dev/null 2>&1 && enable install_docs
-enable install_bins
-enable install_libs
-
-enable optimizations
-enable fast_unaligned #allow unaligned accesses, if supported by hw
-enable md5
-enable spatial_resampling
-enable multithread
-enable os_support
-
-[ -d ${source_path}/../include ] && enable alt_tree_layout
-for d in vp8; do
- [ -d ${source_path}/${d} ] && disable alt_tree_layout;
-done
-
-if ! enabled alt_tree_layout; then
-# development environment
-[ -d ${source_path}/vp8 ] && CODECS="${CODECS} vp8_encoder vp8_decoder"
-else
-# customer environment
-[ -f ${source_path}/../include/vpx/vp8cx.h ] && CODECS="${CODECS} vp8_encoder"
-[ -f ${source_path}/../include/vpx/vp8dx.h ] && CODECS="${CODECS} vp8_decoder"
-
-[ -f ${source_path}/../lib/*/*mt.lib ] && soft_enable static_msvcrt
-fi
-
-CODECS="$(echo ${CODECS} | tr ' ' '\n')"
-CODEC_FAMILIES="$(for c in ${CODECS}; do echo ${c%_*}; done | sort | uniq)"
-
-ARCH_LIST="
- arm
- mips
- x86
- x86_64
- ppc32
- ppc64
-"
-ARCH_EXT_LIST="
- armv5te
- armv6
- armv7
- iwmmxt
- iwmmxt2
-
- mips32
-
- mmx
- sse
- sse2
- sse3
- ssse3
- sse4_1
-
- altivec
-"
-HAVE_LIST="
- ${ARCH_EXT_LIST}
- vpx_ports
- stdint_h
- alt_tree_layout
- pthread_h
- sys_mman_h
-"
-CONFIG_LIST="
- external_build
- install_docs
- install_bins
- install_libs
- install_srcs
- debug
- gprof
- gcov
- rvct
- gcc
- msvs
- pic
- big_endian
-
- codec_srcs
- debug_libs
- fast_unaligned
- mem_manager
- mem_tracker
- mem_checks
- md5
-
- dequant_tokens
- dc_recon
- runtime_cpu_detect
- postproc
- multithread
- psnr
- ${CODECS}
- ${CODEC_FAMILIES}
- encoders
- decoders
- static_msvcrt
- spatial_resampling
- realtime_only
- shared
- small
- postproc_visualizer
- os_support
-"
-CMDLINE_SELECT="
- extra_warnings
- werror
- install_docs
- install_bins
- install_libs
- install_srcs
- debug
- gprof
- gcov
- pic
- optimizations
- ccache
- runtime_cpu_detect
-
- libs
- examples
- libc
- as
- fast_unaligned
- codec_srcs
- debug_libs
- md5
-
- dequant_tokens
- dc_recon
- postproc
- multithread
- psnr
- ${CODECS}
- ${CODEC_FAMILIES}
- static_msvcrt
- mem_tracker
- spatial_resampling
- realtime_only
- shared
- small
- postproc_visualizer
-"
-
-process_cmdline() {
- for opt do
- optval="${opt#*=}"
- case "$opt" in
- --disable-codecs) for c in ${CODECS}; do disable $c; done ;;
- *) process_common_cmdline "$opt"
- ;;
- esac
- done
-}
-
-post_process_cmdline() {
- local c
-
- # If the codec family is disabled, disable all components of that family.
- # If the codec family is enabled, enable all components of that family.
- log_echo "Configuring selected codecs"
- for c in ${CODECS}; do
- disabled ${c%%_*} && disable ${c}
- enabled ${c%%_*} && enable ${c}
- done
-
- # Enable all detected codecs, if they haven't been disabled
- for c in ${CODECS}; do soft_enable $c; done
-
- # Enable the codec family if any component of that family is enabled
- for c in ${CODECS}; do
- enabled $c && enable ${c%_*}
- done
-
- # Set the {en,de}coders variable if any algorithm in that class is enabled
- for c in ${CODECS}; do
- enabled ${c} && enable ${c##*_}s
- done
-}
-
-
-process_targets() {
- enabled child || write_common_config_banner
- enabled universal || write_common_target_config_h ${BUILD_PFX}vpx_config.h
-
- # TODO: add host tools target (obj_int_extract, etc)
-
- # For fat binaries, call configure recursively to configure for each
- # binary architecture to be included.
- if enabled universal; then
- # Call configure (ourselves) for each subarchitecture
- for arch in $fat_bin_archs; do
- BUILD_PFX=${arch}/ toolchain=${arch} $self --child $cmdline_args || exit $?
- done
- fi
-
- # The write_common_config (config.mk) logic is deferred until after the
- # recursive calls to configure complete, becuase we want our universal
- # targets to be executed last.
- write_common_config_targets
- enabled universal && echo "FAT_ARCHS=${fat_bin_archs}" >> config.mk
-
- # Calculate the default distribution name, based on the enabled features
- local cf
- local DIST_DIR=vpx
- for cf in $CODEC_FAMILIES; do
- if enabled ${cf}_encoder && enabled ${cf}_decoder; then
- DIST_DIR="${DIST_DIR}-${cf}"
- elif enabled ${cf}_encoder; then
- DIST_DIR="${DIST_DIR}-${cf}cx"
- elif enabled ${cf}_decoder; then
- DIST_DIR="${DIST_DIR}-${cf}dx"
- fi
- done
- enabled debug_libs && DIST_DIR="${DIST_DIR}-debug"
- enabled codec_srcs && DIST_DIR="${DIST_DIR}-src"
- ! enabled postproc && DIST_DIR="${DIST_DIR}-nopost"
- ! enabled multithread && DIST_DIR="${DIST_DIR}-nomt"
- ! enabled install_docs && DIST_DIR="${DIST_DIR}-nodocs"
- DIST_DIR="${DIST_DIR}-${tgt_isa}-${tgt_os}"
- case "${tgt_os}" in
- win*) enabled static_msvcrt && DIST_DIR="${DIST_DIR}mt" || DIST_DIR="${DIST_DIR}md"
- DIST_DIR="${DIST_DIR}-${tgt_cc}"
- ;;
- esac
- if [ -f "${source_path}/build/make/version.sh" ]; then
- local ver=`"$source_path/build/make/version.sh" --bare $source_path`
- DIST_DIR="${DIST_DIR}-${ver}"
- VERSION_STRING=${ver}
- ver=${ver%%-*}
- VERSION_PATCH=${ver##*.}
- ver=${ver%.*}
- VERSION_MINOR=${ver##*.}
- ver=${ver#v}
- VERSION_MAJOR=${ver%.*}
- fi
- enabled child || cat <<EOF >> config.mk
-
-PREFIX=${prefix}
-ifeq (\$(MAKECMDGOALS),dist)
-DIST_DIR?=${DIST_DIR}
-else
-DIST_DIR?=\$(DESTDIR)${prefix}
-endif
-LIBSUBDIR=${libdir##${prefix}/}
-
-VERSION_STRING=${VERSION_STRING}
-
-VERSION_MAJOR=${VERSION_MAJOR}
-VERSION_MINOR=${VERSION_MINOR}
-VERSION_PATCH=${VERSION_PATCH}
-
-CONFIGURE_ARGS=${CONFIGURE_ARGS}
-EOF
- enabled child || echo "CONFIGURE_ARGS?=${CONFIGURE_ARGS}" >> config.mk
-
- #
- # Write makefiles for all enabled targets
- #
- for tgt in libs examples docs solution; do
- local tgt_fn="$tgt-$toolchain.mk"
-
- if enabled $tgt; then
- echo "Creating makefiles for ${toolchain} ${tgt}"
- write_common_target_config_mk $tgt_fn ${BUILD_PFX}vpx_config.h
- #write_${tgt}_config
- fi
- done
-
-}
-
-process_detect() {
- if enabled shared; then
- # Can only build shared libs on a subset of platforms. Doing this check
- # here rather than at option parse time because the target auto-detect
- # magic happens after the command line has been parsed.
- enabled linux || die "--enable-shared only supported on ELF for now"
- fi
- if [ -z "$CC" ]; then
- echo "Bypassing toolchain for environment detection."
- enable external_build
- check_header() {
- log fake_check_header "$@"
- header=$1
- shift
- var=`echo $header | sed 's/[^A-Za-z0-9_]/_/g'`
- disable $var
- case $header in
- stdio.h)
- true;
- ;;
- *)
- local result=false
- for d in "$@"; do
- [ -f "${d##-I}/$header" ] && result=true && break
- done
- ${result:-true}
- esac && enable $var
- }
- check_ld() {
- true
- }
- fi
- check_header stdio.h || die "Unable to invoke compiler: ${CC} ${CFLAGS}"
- check_ld <<EOF || die "Toolchain is unable to link executables"
-int main(void) {return 0;}
-EOF
- # check system headers
- check_header stdint.h
- check_header pthread.h
- check_header sys/mman.h
-
- check_header vpx/vpx_integer.h -I${source_path} && enable vpx_ports
-}
-
-process_toolchain() {
- process_common_toolchain
-
- # Handle universal binaries for this architecture
- case $toolchain in
- universal-darwin*)
- local darwin_ver=${tgt_os##darwin}
- fat_bin_archs="$fat_bin_archs ppc32-${tgt_os}-gcc"
-
- # Intel
- fat_bin_archs="$fat_bin_archs x86-${tgt_os}-${tgt_cc}"
- if [ $darwin_ver -gt 8 ]; then
- fat_bin_archs="$fat_bin_archs x86_64-${tgt_os}-${tgt_cc}"
- fi
- ;;
- esac
-
-
- # Enable some useful compiler flags
- if enabled gcc; then
- enabled werror && check_add_cflags -Werror
- check_add_cflags -Wall
- check_add_cflags -Wdeclaration-after-statement
- check_add_cflags -Wdisabled-optimization
- check_add_cflags -Wpointer-arith
- check_add_cflags -Wtype-limits
- check_add_cflags -Wcast-qual
- enabled extra_warnings || check_add_cflags -Wno-unused-function
- fi
-
- if enabled icc; then
- enabled werror && check_add_cflags -Werror
- check_add_cflags -Wall
- check_add_cflags -Wpointer-arith
-
- # ICC has a number of floating point optimizations that we disable
- # in favor of deterministic output WRT to other compilers
- add_cflags -fp-model precise
- fi
-
- # Enable extra, harmless warnings. These might provide additional insight
- # to what the compiler is doing and why, but in general, but they shouldn't
- # be treated as fatal, even if we're treating warnings as errors.
- GCC_EXTRA_WARNINGS="
- -Wdisabled-optimization
- -Winline
- "
- enabled gcc && EXTRA_WARNINGS="${GCC_EXTRA_WARNINGS}"
- RVCT_EXTRA_WARNINGS="
- --remarks
- "
- enabled rvct && EXTRA_WARNINGS="${RVCT_EXTRA_WARNINGS}"
- if enabled extra_warnings; then
- for w in ${EXTRA_WARNINGS}; do
- check_add_cflags ${w}
- enabled gcc && enabled werror && check_add_cflags -Wno-error=${w}
- done
- fi
-
- # ccache only really works on gcc toolchains
- enabled gcc || soft_disable ccache
- if enabled mips; then
- enable dequant_tokens
- enable dc_recon
- fi
-
- # Enable the postbuild target if building for visual studio.
- case "$tgt_cc" in
- vs*) enable msvs
- enable solution
- vs_version=${tgt_cc##vs}
- all_targets="${all_targets} solution"
- ;;
- esac
-
- # Other toolchain specific defaults
- case $toolchain in x86*|ppc*|universal*) soft_enable postproc;; esac
-
- if enabled postproc_visualizer; then
- enabled postproc || die "postproc_visualizer requires postproc to be enabled"
- fi
-}
-
-
-##
-## END APPLICATION SPECIFIC CONFIGURATION
-##
-CONFIGURE_ARGS="$@"
-process "$@"
-cat <<EOF > ${BUILD_PFX}vpx_config.c
-static const char* const cfg = "$CONFIGURE_ARGS";
-const char *vpx_codec_build_config(void) {return cfg;}
-EOF
diff --git a/docs.mk b/docs.mk
deleted file mode 100644
index 98332a2..0000000
--- a/docs.mk
+++ /dev/null
@@ -1,56 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-INSTALL_MAPS += docs/% docs/%
-INSTALL_MAPS += src/% %
-INSTALL_MAPS += % %
-
-# Static documentation authored in doxygen
-CODEC_DOX := mainpage.dox \
- keywords.dox \
- usage.dox \
- usage_cx.dox \
- usage_dx.dox \
-
-# Other doxy files sourced in Markdown
-TXT_DOX-$(CONFIG_VP8) += vp8_api1_migration.dox
-vp8_api1_migration.dox.DESC = VP8 API 1.x Migration
-
-TXT_DOX = $(call enabled,TXT_DOX)
-
-%.dox: %.txt
- @echo " [DOXY] $@"
- @$(SRC_PATH_BARE)/examples/gen_example_doxy.php \
- $(@:.dox=) "$($@.DESC)" > $@ < $<
-
-
-EXAMPLE_PATH += $(SRC_PATH_BARE) #for CHANGELOG, README, etc
-
-doxyfile: $(if $(findstring examples, $(ALL_TARGETS)),examples.doxy)
-doxyfile: libs.doxy_template libs.doxy
- @echo " [CREATE] $@"
- @cat $^ > $@
- @echo "STRIP_FROM_PATH += $(SRC_PATH_BARE) $(BUILD_ROOT)" >> $@
- @echo "INPUT += $(addprefix $(SRC_PATH_BARE)/,$(CODEC_DOX))" >> $@;
- @echo "INPUT += $(TXT_DOX)" >> $@;
- @echo "EXAMPLE_PATH += $(EXAMPLE_PATH)" >> $@
-
-CLEAN-OBJS += doxyfile $(wildcard docs/html/*)
-docs/html/index.html: doxyfile $(CODEC_DOX) $(TXT_DOX)
- @echo " [DOXYGEN] $<"
- @doxygen $<
-DOCS-yes += docs/html/index.html
-
-DIST-DOCS-yes = $(wildcard docs/html/*)
-DIST-DOCS-$(CONFIG_CODEC_SRCS) += $(addprefix src/,$(CODEC_DOX))
-DIST-DOCS-$(CONFIG_CODEC_SRCS) += src/libs.doxy_template
-DIST-DOCS-yes += CHANGELOG
-DIST-DOCS-yes += README
diff --git a/examples.mk b/examples.mk
deleted file mode 100644
index 89c93db..0000000
--- a/examples.mk
+++ /dev/null
@@ -1,265 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-# List of examples to build. UTILS are files that are taken from the source
-# tree directly, and GEN_EXAMPLES are files that are created from the
-# examples folder.
-UTILS-$(CONFIG_DECODERS) += vpxdec.c
-vpxdec.SRCS += md5_utils.c md5_utils.h
-vpxdec.SRCS += vpx_ports/vpx_timer.h
-vpxdec.SRCS += vpx/vpx_integer.h
-vpxdec.SRCS += args.c args.h vpx_ports/config.h
-vpxdec.SRCS += tools_common.c tools_common.h
-vpxdec.SRCS += nestegg/halloc/halloc.h
-vpxdec.SRCS += nestegg/halloc/src/align.h
-vpxdec.SRCS += nestegg/halloc/src/halloc.c
-vpxdec.SRCS += nestegg/halloc/src/hlist.h
-vpxdec.SRCS += nestegg/halloc/src/macros.h
-vpxdec.SRCS += nestegg/include/nestegg/nestegg.h
-vpxdec.SRCS += nestegg/src/nestegg.c
-vpxdec.GUID = BA5FE66F-38DD-E034-F542-B1578C5FB950
-vpxdec.DESCRIPTION = Full featured decoder
-UTILS-$(CONFIG_ENCODERS) += vpxenc.c
-vpxenc.SRCS += args.c args.h y4minput.c y4minput.h
-vpxenc.SRCS += tools_common.c tools_common.h
-vpxenc.SRCS += vpx_ports/config.h vpx_ports/mem_ops.h
-vpxenc.SRCS += vpx_ports/mem_ops_aligned.h
-vpxenc.SRCS += libmkv/EbmlIDs.h
-vpxenc.SRCS += libmkv/EbmlWriter.c
-vpxenc.SRCS += libmkv/EbmlWriter.h
-vpxenc.GUID = 548DEC74-7A15-4B2B-AFC3-AA102E7C25C1
-vpxenc.DESCRIPTION = Full featured encoder
-
-# Clean up old ivfenc, ivfdec binaries.
-ifeq ($(CONFIG_MSVS),yes)
-CLEAN-OBJS += $(foreach p,$(VS_PLATFORMS),$(p)/Release/ivfenc.exe)
-CLEAN-OBJS += $(foreach p,$(VS_PLATFORMS),$(p)/Release/ivfdec.exe)
-else
-CLEAN-OBJS += ivfenc{.c.o,.c.d,.dox,.exe,}
-CLEAN-OBJS += ivfdec{.c.o,.c.d,.dox,.exe,}
-endif
-
-# XMA example disabled for now, not used in VP8
-#UTILS-$(CONFIG_DECODERS) += example_xma.c
-#example_xma.GUID = A955FC4A-73F1-44F7-135E-30D84D32F022
-#example_xma.DESCRIPTION = External Memory Allocation mode usage
-
-GEN_EXAMPLES-$(CONFIG_DECODERS) += simple_decoder.c
-simple_decoder.GUID = D3BBF1E9-2427-450D-BBFF-B2843C1D44CC
-simple_decoder.DESCRIPTION = Simplified decoder loop
-GEN_EXAMPLES-$(CONFIG_DECODERS) += postproc.c
-postproc.GUID = 65E33355-F35E-4088-884D-3FD4905881D7
-postproc.DESCRIPTION = Decoder postprocessor control
-GEN_EXAMPLES-$(CONFIG_DECODERS) += decode_to_md5.c
-decode_to_md5.SRCS += md5_utils.h md5_utils.c
-decode_to_md5.GUID = 59120B9B-2735-4BFE-B022-146CA340FE42
-decode_to_md5.DESCRIPTION = Frame by frame MD5 checksum
-
-GEN_EXAMPLES-$(CONFIG_ENCODERS) += simple_encoder.c
-simple_encoder.GUID = 4607D299-8A71-4D2C-9B1D-071899B6FBFD
-simple_encoder.DESCRIPTION = Simplified encoder loop
-GEN_EXAMPLES-$(CONFIG_ENCODERS) += twopass_encoder.c
-twopass_encoder.GUID = 73494FA6-4AF9-4763-8FBB-265C92402FD8
-twopass_encoder.DESCRIPTION = Two-pass encoder loop
-GEN_EXAMPLES-$(CONFIG_ENCODERS) += force_keyframe.c
-force_keyframe.GUID = 3C67CADF-029F-4C86-81F5-D6D4F51177F0
-force_keyframe.DESCRIPTION = Force generation of keyframes
-ifeq ($(CONFIG_DECODERS),yes)
-GEN_EXAMPLES-$(CONFIG_ENCODERS) += decode_with_drops.c
-endif
-decode_with_drops.GUID = CE5C53C4-8DDA-438A-86ED-0DDD3CDB8D26
-decode_with_drops.DESCRIPTION = Drops frames while decoding
-GEN_EXAMPLES-$(CONFIG_ENCODERS) += error_resilient.c
-error_resilient.GUID = DF5837B9-4145-4F92-A031-44E4F832E00C
-error_resilient.DESCRIPTION = Error Resiliency Feature
-
-GEN_EXAMPLES-$(CONFIG_VP8_ENCODER) += vp8_scalable_patterns.c
-vp8_scalable_patterns.GUID = 0D6A210B-F482-4D6F-8570-4A9C01ACC88C
-vp8_scalable_patterns.DESCRIPTION = VP8 Scalable Bitstream Patterns
-GEN_EXAMPLES-$(CONFIG_VP8_ENCODER) += vp8_set_maps.c
-vp8_set_maps.GUID = ECB2D24D-98B8-4015-A465-A4AF3DCC145F
-vp8_set_maps.DESCRIPTION = VP8 set active and ROI maps
-GEN_EXAMPLES-$(CONFIG_VP8_ENCODER) += vp8cx_set_ref.c
-vp8cx_set_ref.GUID = C5E31F7F-96F6-48BD-BD3E-10EBF6E8057A
-vp8cx_set_ref.DESCRIPTION = VP8 set encoder reference frame
-
-
-# Handle extra library flags depending on codec configuration
-
-# We should not link to math library (libm) on RVCT
-# when building for bare-metal targets
-ifeq ($(CONFIG_OS_SUPPORT), yes)
-CODEC_EXTRA_LIBS-$(CONFIG_VP8) += m
-else
- ifeq ($(CONFIG_GCC), yes)
- CODEC_EXTRA_LIBS-$(CONFIG_VP8) += m
- endif
-endif
-#
-# End of specified files. The rest of the build rules should happen
-# automagically from here.
-#
-
-
-# Examples need different flags based on whether we're building
-# from an installed tree or a version controlled tree. Determine
-# the proper paths.
-ifeq ($(HAVE_ALT_TREE_LAYOUT),yes)
- LIB_PATH := $(SRC_PATH_BARE)/../lib
- INC_PATH := $(SRC_PATH_BARE)/../include
-else
- LIB_PATH-yes += $(if $(BUILD_PFX),$(BUILD_PFX),.)
- INC_PATH-$(CONFIG_VP8_DECODER) += $(SRC_PATH_BARE)/vp8
- INC_PATH-$(CONFIG_VP8_ENCODER) += $(SRC_PATH_BARE)/vp8
- LIB_PATH := $(call enabled,LIB_PATH)
- INC_PATH := $(call enabled,INC_PATH)
-endif
-CFLAGS += $(addprefix -I,$(INC_PATH))
-LDFLAGS += $(addprefix -L,$(LIB_PATH))
-
-
-# Expand list of selected examples to build (as specified above)
-UTILS = $(call enabled,UTILS)
-GEN_EXAMPLES = $(call enabled,GEN_EXAMPLES)
-ALL_EXAMPLES = $(UTILS) $(GEN_EXAMPLES)
-UTIL_SRCS = $(foreach ex,$(UTILS),$($(ex:.c=).SRCS))
-ALL_SRCS = $(foreach ex,$(ALL_EXAMPLES),$($(ex:.c=).SRCS))
-CODEC_EXTRA_LIBS=$(sort $(call enabled,CODEC_EXTRA_LIBS))
-
-
-# Expand all example sources into a variable containing all sources
-# for that example (not just them main one specified in UTILS/GEN_EXAMPLES)
-# and add this file to the list (for MSVS workspace generation)
-$(foreach ex,$(ALL_EXAMPLES),$(eval $(ex:.c=).SRCS += $(ex) examples.mk))
-
-
-# If this is a universal (fat) binary, then all the subarchitectures have
-# already been built and our job is to stitch them together. The
-# BUILD_OBJS variable indicates whether we should be building
-# (compiling, linking) the library. The LIPO_OBJS variable indicates
-# that we're stitching.
-$(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_OBJS,BUILD_OBJS):=yes)
-
-
-# Create build/install dependencies for all examples. The common case
-# is handled here. The MSVS case is handled below.
-NOT_MSVS = $(if $(CONFIG_MSVS),,yes)
-DIST-BINS-$(NOT_MSVS) += $(addprefix bin/,$(ALL_EXAMPLES:.c=))
-INSTALL-BINS-$(NOT_MSVS) += $(addprefix bin/,$(UTILS:.c=))
-DIST-SRCS-yes += $(ALL_SRCS)
-INSTALL-SRCS-yes += $(UTIL_SRCS)
-OBJS-$(NOT_MSVS) += $(if $(BUILD_OBJS),$(call objs,$(ALL_SRCS)))
-BINS-$(NOT_MSVS) += $(addprefix $(BUILD_PFX),$(ALL_EXAMPLES:.c=))
-
-
-# Instantiate linker template for all examples.
-CODEC_LIB=$(if $(CONFIG_DEBUG_LIBS),vpx_g,vpx)
-$(foreach bin,$(BINS-yes),\
- $(if $(BUILD_OBJS),$(eval $(bin): $(LIB_PATH)/lib$(CODEC_LIB).a))\
- $(if $(BUILD_OBJS),$(eval $(call linker_template,$(bin),\
- $(call objs,$($(notdir $(bin)).SRCS)) \
- -l$(CODEC_LIB) $(addprefix -l,$(CODEC_EXTRA_LIBS))\
- )))\
- $(if $(LIPO_OBJS),$(eval $(call lipo_bin_template,$(bin))))\
- )
-
-
-# Rules to generate the GEN_EXAMPLES sources
-.PRECIOUS: %.c
-CLEAN-OBJS += $(GEN_EXAMPLES)
-%.c: examples/%.txt
- @echo " [EXAMPLE] $@"
- @$(SRC_PATH_BARE)/examples/gen_example_code.sh $< > $@
-
-
-# The following pairs define a mapping of locations in the distribution
-# tree to locations in the source/build trees.
-INSTALL_MAPS += src/%.c %.c
-INSTALL_MAPS += src/% $(SRC_PATH_BARE)/%
-INSTALL_MAPS += bin/% %
-INSTALL_MAPS += % %
-
-
-# Set up additional MSVS environment
-ifeq ($(CONFIG_MSVS),yes)
-CODEC_LIB=$(if $(CONFIG_STATIC_MSVCRT),vpxmt,vpxmd)
-# This variable uses deferred expansion intentionally, since the results of
-# $(wildcard) may change during the course of the Make.
-VS_PLATFORMS = $(foreach d,$(wildcard */Release/$(CODEC_LIB).lib),$(word 1,$(subst /, ,$(d))))
-INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),bin/$(p)/% $(p)/Release/%)
-endif
-
-# Build Visual Studio Projects. We use a template here to instantiate
-# explicit rules rather than using an implicit rule because we want to
-# leverage make's VPATH searching rather than specifying the paths on
-# each file in ALL_EXAMPLES. This has the unfortunate side effect that
-# touching the source files trigger a rebuild of the project files
-# even though there is no real dependency there (the dependency is on
-# the makefiles). We may want to revisit this.
-define vcproj_template
-$(1): $($(1:.vcproj=).SRCS)
- @echo " [vcproj] $$@"
- $$(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh\
- --exe\
- --target=$$(TOOLCHAIN)\
- --name=$$(@:.vcproj=)\
- --ver=$$(CONFIG_VS_VERSION)\
- --proj-guid=$$($$(@:.vcproj=).GUID)\
- $$(if $$(CONFIG_STATIC_MSVCRT),--static-crt) \
- --out=$$@ $$(CFLAGS) $$(LDFLAGS) -l$$(CODEC_LIB) -lwinmm $$^
-endef
-PROJECTS-$(CONFIG_MSVS) += $(ALL_EXAMPLES:.c=.vcproj)
-INSTALL-BINS-$(CONFIG_MSVS) += $(foreach p,$(VS_PLATFORMS),\
- $(addprefix bin/$(p)/,$(ALL_EXAMPLES:.c=.exe)))
-$(foreach proj,$(call enabled,PROJECTS),\
- $(eval $(call vcproj_template,$(proj))))
-
-
-
-#
-# Documentation Rules
-#
-%.dox: examples/%.txt
- @echo " [DOXY] $@"
- @$(SRC_PATH_BARE)/examples/gen_example_text.sh $< | \
- $(SRC_PATH_BARE)/examples/gen_example_doxy.php \
- example_$(@:.dox=) $(@:.dox=.c) > $@
-
-%.dox: %.c
- @echo " [DOXY] $@"
- @echo "/*!\page example_$(@:.dox=) $(@:.dox=)" > $@
- @echo " \includelineno $(notdir $<)" >> $@
- @echo "*/" >> $@
-
-samples.dox: examples.mk
- @echo " [DOXY] $@"
- @echo "/*!\page samples Sample Code" > $@
- @echo " This SDK includes a number of sample applications."\
- "each sample documents a feature of the SDK in both prose"\
- "and the associated C code. In general, later samples"\
- "build upon prior samples, so it is best to work through the"\
- "list in order. The following samples are included: ">>$@
- @$(foreach ex,$(GEN_EXAMPLES:.c=),\
- echo " - \subpage example_$(ex) $($(ex).DESCRIPTION)" >> $@;)
- @echo >> $@
- @echo " In addition, the SDK contains a number of utilities."\
- "Since these utilities are built upon the concepts described"\
- "in the sample code listed above, they are not documented in"\
- "pieces like the samples are. Thir sourcre is included here"\
- "for reference. The following utilities are included:" >> $@
- @$(foreach ex,$(UTILS:.c=),\
- echo " - \subpage example_$(ex) $($(ex).DESCRIPTION)" >> $@;)
- @echo "*/" >> $@
-
-CLEAN-OBJS += examples.doxy samples.dox $(ALL_EXAMPLES:.c=.dox)
-DOCS-yes += examples.doxy samples.dox $(ALL_EXAMPLES:.c=.dox)
-examples.doxy: samples.dox $(ALL_EXAMPLES:.c=.dox)
- @echo "INPUT += $^" > $@
diff --git a/examples/decode_to_md5.txt b/examples/decode_to_md5.txt
deleted file mode 100644
index b3dd568..0000000
--- a/examples/decode_to_md5.txt
+++ /dev/null
@@ -1,48 +0,0 @@
-@TEMPLATE decoder_tmpl.c
-Frame-by-frame MD5 Checksum
-===========================
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
-This example builds upon the simple decoder loop to show how checksums
-of the decoded output can be generated. These are used for validating
-decoder implementations against the reference implementation, for example.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
-
-MD5 algorithm
--------------
-The Message-Digest 5 (MD5) is a well known hash function. We have provided
-an implementation derived from the RSA Data Security, Inc. MD5 Message-Digest
-Algorithm for your use. Our implmentation only changes the interface of this
-reference code. You must include the `md5_utils.h` header for access to these
-functions.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ EXTRA_INCLUDES
-#include "md5_utils.h"
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ EXTRA_INCLUDES
-
-
-Processing The Decoded Data
----------------------------
-Each row of the image is passed to the MD5 accumulator. First the Y plane
-is processed, then U, then V. It is important to honor the image's `stride`
-values.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX
-unsigned char md5_sum[16];
-MD5Context md5;
-int i;
-
-MD5Init(&md5);
-
-for(plane=0; plane < 3; plane++) {
- unsigned char *buf =img->planes[plane];
-
- for(y=0; y<img->d_h >> (plane?1:0); y++) {
- MD5Update(&md5, buf, img->d_w >> (plane?1:0));
- buf += img->stride[plane];
- }
-}
-
-MD5Final(md5_sum, &md5);
-for(i=0; i<16; i++)
- fprintf(outfile, "%02x",md5_sum[i]);
-fprintf(outfile, " img-%dx%d-%04d.i420\n", img->d_w, img->d_h,
- frame_cnt);
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX
diff --git a/examples/decoder_tmpl.c b/examples/decoder_tmpl.c
deleted file mode 100644
index c70681b..0000000
--- a/examples/decoder_tmpl.c
+++ /dev/null
@@ -1,101 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*
-@*INTRODUCTION
- */
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdarg.h>
-#include <string.h>
-#define VPX_CODEC_DISABLE_COMPAT 1
-#include "vpx/vpx_decoder.h"
-#include "vpx/vp8dx.h"
-#define interface (vpx_codec_vp8_dx())
-@EXTRA_INCLUDES
-
-
-#define IVF_FILE_HDR_SZ (32)
-#define IVF_FRAME_HDR_SZ (12)
-
-static unsigned int mem_get_le32(const unsigned char *mem) {
- return (mem[3] << 24)|(mem[2] << 16)|(mem[1] << 8)|(mem[0]);
-}
-
-static void die(const char *fmt, ...) {
- va_list ap;
-
- va_start(ap, fmt);
- vprintf(fmt, ap);
- if(fmt[strlen(fmt)-1] != '\n')
- printf("\n");
- exit(EXIT_FAILURE);
-}
-
-@DIE_CODEC
-
-int main(int argc, char **argv) {
- FILE *infile, *outfile;
- vpx_codec_ctx_t codec;
- int flags = 0, frame_cnt = 0;
- unsigned char file_hdr[IVF_FILE_HDR_SZ];
- unsigned char frame_hdr[IVF_FRAME_HDR_SZ];
- unsigned char frame[256*1024];
- vpx_codec_err_t res;
-@@@@EXTRA_VARS
-
- (void)res;
- /* Open files */
-@@@@USAGE
- if(!(infile = fopen(argv[1], "rb")))
- die("Failed to open %s for reading", argv[1]);
- if(!(outfile = fopen(argv[2], "wb")))
- die("Failed to open %s for writing", argv[2]);
-
- /* Read file header */
- if(!(fread(file_hdr, 1, IVF_FILE_HDR_SZ, infile) == IVF_FILE_HDR_SZ
- && file_hdr[0]=='D' && file_hdr[1]=='K' && file_hdr[2]=='I'
- && file_hdr[3]=='F'))
- die("%s is not an IVF file.", argv[1]);
-
- printf("Using %s\n",vpx_codec_iface_name(interface));
-@@@@DEC_INIT
-
- /* Read each frame */
- while(fread(frame_hdr, 1, IVF_FRAME_HDR_SZ, infile) == IVF_FRAME_HDR_SZ) {
- int frame_sz = mem_get_le32(frame_hdr);
- vpx_codec_iter_t iter = NULL;
- vpx_image_t *img;
-
-
- frame_cnt++;
- if(frame_sz > sizeof(frame))
- die("Frame %d data too big for example code buffer", frame_sz);
- if(fread(frame, 1, frame_sz, infile) != frame_sz)
- die("Frame %d failed to read complete frame", frame_cnt);
-
-@@@@@@@@PRE_DECODE
-@@@@@@@@DECODE
-
- /* Write decoded data to disk */
-@@@@@@@@GET_FRAME
- unsigned int plane, y;
-
-@@@@@@@@@@@@PROCESS_DX
- }
- }
- printf("Processed %d frames.\n",frame_cnt);
-@@@@DESTROY
-
- fclose(outfile);
- fclose(infile);
- return EXIT_SUCCESS;
-}
diff --git a/examples/decoder_tmpl.txt b/examples/decoder_tmpl.txt
deleted file mode 100644
index 7dd05d1..0000000
--- a/examples/decoder_tmpl.txt
+++ /dev/null
@@ -1,61 +0,0 @@
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INCLUDES
-#define VPX_CODEC_DISABLE_COMPAT 1
-#include "vpx/vpx_decoder.h"
-#include "vpx/vp8dx.h"
-#define interface (vpx_codec_vp8_dx())
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INCLUDES
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DIE_CODEC
-static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
- const char *detail = vpx_codec_error_detail(ctx);
-
- printf("%s: %s\n", s, vpx_codec_error(ctx));
- if(detail)
- printf(" %s\n",detail);
- exit(EXIT_FAILURE);
-}
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DIE_CODEC
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
-if(argc!=3)
- die("Usage: %s <infile> <outfile>\n", argv[0]);
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
-/* Initialize codec */
-if(vpx_codec_dec_init(&codec, interface, NULL, flags))
- die_codec(&codec, "Failed to initialize decoder");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DECODE
-/* Decode the frame */
-if(vpx_codec_decode(&codec, frame, frame_sz, NULL, 0))
- die_codec(&codec, "Failed to decode frame");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DECODE
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ GET_FRAME
-while((img = vpx_codec_get_frame(&codec, &iter))) {
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ GET_FRAME
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX
-for(plane=0; plane < 3; plane++) {
- unsigned char *buf =img->planes[plane];
-
- for(y=0; y<img->d_h >> (plane?1:0); y++) {
- if(fwrite(buf, 1, img->d_w >> (plane?1:0), outfile));
- buf += img->stride[plane];
- }
-}
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
-if(vpx_codec_destroy(&codec))
- die_codec(&codec, "Failed to destroy codec");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
diff --git a/examples/encoder_tmpl.c b/examples/encoder_tmpl.c
deleted file mode 100644
index f109e62..0000000
--- a/examples/encoder_tmpl.c
+++ /dev/null
@@ -1,188 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*
-@*INTRODUCTION
- */
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdarg.h>
-#include <string.h>
-#define VPX_CODEC_DISABLE_COMPAT 1
-#include "vpx/vpx_encoder.h"
-#include "vpx/vp8cx.h"
-#define interface (vpx_codec_vp8_cx())
-#define fourcc 0x30385056
-@EXTRA_INCLUDES
-
-#define IVF_FILE_HDR_SZ (32)
-#define IVF_FRAME_HDR_SZ (12)
-
-static void mem_put_le16(char *mem, unsigned int val) {
- mem[0] = val;
- mem[1] = val>>8;
-}
-
-static void mem_put_le32(char *mem, unsigned int val) {
- mem[0] = val;
- mem[1] = val>>8;
- mem[2] = val>>16;
- mem[3] = val>>24;
-}
-
-static void die(const char *fmt, ...) {
- va_list ap;
-
- va_start(ap, fmt);
- vprintf(fmt, ap);
- if(fmt[strlen(fmt)-1] != '\n')
- printf("\n");
- exit(EXIT_FAILURE);
-}
-
-@DIE_CODEC
-
-static int read_frame(FILE *f, vpx_image_t *img) {
- size_t nbytes, to_read;
- int res = 1;
-
- to_read = img->w*img->h*3/2;
- nbytes = fread(img->planes[0], 1, to_read, f);
- if(nbytes != to_read) {
- res = 0;
- if(nbytes > 0)
- printf("Warning: Read partial frame. Check your width & height!\n");
- }
- return res;
-}
-
-static void write_ivf_file_header(FILE *outfile,
- const vpx_codec_enc_cfg_t *cfg,
- int frame_cnt) {
- char header[32];
-
- if(cfg->g_pass != VPX_RC_ONE_PASS && cfg->g_pass != VPX_RC_LAST_PASS)
- return;
- header[0] = 'D';
- header[1] = 'K';
- header[2] = 'I';
- header[3] = 'F';
- mem_put_le16(header+4, 0); /* version */
- mem_put_le16(header+6, 32); /* headersize */
- mem_put_le32(header+8, fourcc); /* headersize */
- mem_put_le16(header+12, cfg->g_w); /* width */
- mem_put_le16(header+14, cfg->g_h); /* height */
- mem_put_le32(header+16, cfg->g_timebase.den); /* rate */
- mem_put_le32(header+20, cfg->g_timebase.num); /* scale */
- mem_put_le32(header+24, frame_cnt); /* length */
- mem_put_le32(header+28, 0); /* unused */
-
- if(fwrite(header, 1, 32, outfile));
-}
-
-
-static void write_ivf_frame_header(FILE *outfile,
- const vpx_codec_cx_pkt_t *pkt)
-{
- char header[12];
- vpx_codec_pts_t pts;
-
- if(pkt->kind != VPX_CODEC_CX_FRAME_PKT)
- return;
-
- pts = pkt->data.frame.pts;
- mem_put_le32(header, pkt->data.frame.sz);
- mem_put_le32(header+4, pts&0xFFFFFFFF);
- mem_put_le32(header+8, pts >> 32);
-
- if(fwrite(header, 1, 12, outfile));
-}
-
-int main(int argc, char **argv) {
- FILE *infile, *outfile;
- vpx_codec_ctx_t codec;
- vpx_codec_enc_cfg_t cfg;
- int frame_cnt = 0;
- unsigned char file_hdr[IVF_FILE_HDR_SZ];
- unsigned char frame_hdr[IVF_FRAME_HDR_SZ];
- vpx_image_t raw;
- vpx_codec_err_t res;
- long width;
- long height;
- int frame_avail;
- int got_data;
- int flags = 0;
-@@@@TWOPASS_VARS
-
- /* Open files */
-@@@@USAGE
- width = strtol(argv[1], NULL, 0);
- height = strtol(argv[2], NULL, 0);
- if(width < 16 || width%2 || height <16 || height%2)
- die("Invalid resolution: %ldx%ld", width, height);
- if(!vpx_img_alloc(&raw, VPX_IMG_FMT_I420, width, height, 1))
- die("Faile to allocate image", width, height);
- if(!(outfile = fopen(argv[4], "wb")))
- die("Failed to open %s for writing", argv[4]);
-
- printf("Using %s\n",vpx_codec_iface_name(interface));
-
-@@@@ENC_DEF_CFG
-
-@@@@ENC_SET_CFG
-@@@@ENC_SET_CFG2
-
- write_ivf_file_header(outfile, &cfg, 0);
-
-@@@@TWOPASS_LOOP_BEGIN
-
- /* Open input file for this encoding pass */
- if(!(infile = fopen(argv[3], "rb")))
- die("Failed to open %s for reading", argv[3]);
-
-@@@@@@@@ENC_INIT
-
- frame_avail = 1;
- got_data = 0;
- while(frame_avail || got_data) {
- vpx_codec_iter_t iter = NULL;
- const vpx_codec_cx_pkt_t *pkt;
-
-@@@@@@@@@@@@PER_FRAME_CFG
-@@@@@@@@@@@@ENCODE_FRAME
- got_data = 0;
- while( (pkt = vpx_codec_get_cx_data(&codec, &iter)) ) {
- got_data = 1;
- switch(pkt->kind) {
-@@@@@@@@@@@@@@@@PROCESS_FRAME
-@@@@@@@@@@@@@@@@PROCESS_STATS
- default:
- break;
- }
- printf(pkt->kind == VPX_CODEC_CX_FRAME_PKT
- && (pkt->data.frame.flags & VPX_FRAME_IS_KEY)? "K":".");
- fflush(stdout);
- }
- frame_cnt++;
- }
- printf("\n");
- fclose(infile);
-@@@@TWOPASS_LOOP_END
-
- printf("Processed %d frames.\n",frame_cnt-1);
-@@@@DESTROY
-
- /* Try to rewrite the file header with the actual frame count */
- if(!fseek(outfile, 0, SEEK_SET))
- write_ivf_file_header(outfile, &cfg, frame_cnt-1);
- fclose(outfile);
- return EXIT_SUCCESS;
-}
diff --git a/examples/encoder_tmpl.txt b/examples/encoder_tmpl.txt
deleted file mode 100644
index 0042071..0000000
--- a/examples/encoder_tmpl.txt
+++ /dev/null
@@ -1,73 +0,0 @@
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INCLUDES
-#define VPX_CODEC_DISABLE_COMPAT 1
-#include "vpx/vpx_encoder.h"
-#include "vpx/vp8cx.h"
-#define interface (vpx_codec_vp8_cx())
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INCLUDES
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DIE_CODEC
-static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
- const char *detail = vpx_codec_error_detail(ctx);
-
- printf("%s: %s\n", s, vpx_codec_error(ctx));
- if(detail)
- printf(" %s\n",detail);
- exit(EXIT_FAILURE);
-}
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DIE_CODEC
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
-if(argc!=5)
- die("Usage: %s <width> <height> <infile> <outfile>\n", argv[0]);
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_DEF_CFG
-/* Populate encoder configuration */
-res = vpx_codec_enc_config_default(interface, &cfg, 0);
-if(res) {
- printf("Failed to get config: %s\n", vpx_codec_err_to_string(res));
- return EXIT_FAILURE;
-}
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_DEF_CFG
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_SET_CFG
-/* Update the default configuration with our settings */
-cfg.rc_target_bitrate = width * height * cfg.rc_target_bitrate
- / cfg.g_w / cfg.g_h;
-cfg.g_w = width;
-cfg.g_h = height;
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_SET_CFG
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INIT
-/* Initialize codec */
-if(vpx_codec_enc_init(&codec, interface, &cfg, 0))
- die_codec(&codec, "Failed to initialize encoder");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INIT
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENCODE_FRAME
-frame_avail = read_frame(infile, &raw);
-if(vpx_codec_encode(&codec, frame_avail? &raw : NULL, frame_cnt,
- 1, flags, VPX_DL_REALTIME))
- die_codec(&codec, "Failed to encode frame");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENCODE_FRAME
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_FRAME
-case VPX_CODEC_CX_FRAME_PKT:
- write_ivf_frame_header(outfile, pkt);
- if(fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
- outfile));
- break;
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_FRAME
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
-if(vpx_codec_destroy(&codec))
- die_codec(&codec, "Failed to destroy codec");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
diff --git a/examples/postproc.txt b/examples/postproc.txt
deleted file mode 100644
index c65ddda..0000000
--- a/examples/postproc.txt
+++ /dev/null
@@ -1,67 +0,0 @@
-@TEMPLATE decoder_tmpl.c
-Postprocessing Decoder
-======================
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
-This example adds postprocessing to the simple decoder loop.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
-
-
-Initializing Postprocessing
----------------------------
-You must inform the codec that you might request postprocessing at
-initialization time. This is done by passing the VPX_CODEC_USE_POSTPROC
-flag to `vpx_codec_dec_init`. If the codec does not support
-postprocessing, this call will return VPX_CODEC_INCAPABLE. For
-demonstration purposes, we also fall back to default initialization if
-the codec does not provide support.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
-/* Initialize codec */
-res = vpx_codec_dec_init(&codec, interface, NULL,
- VPX_CODEC_USE_POSTPROC);
-if(res == VPX_CODEC_INCAPABLE) {
- printf("NOTICE: Postproc not supported by %s\n",
- vpx_codec_iface_name(interface));
- res = vpx_codec_dec_init(&codec, interface, NULL, 0);
-}
-if(res)
- die_codec(&codec, "Failed to initialize decoder");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
-
-
-Using Adaptive Postprocessing
------------------------------
-VP6 provides "adaptive postprocessing." It will automatically select the
-best postprocessing filter on a frame by frame basis based on the amount
-of time remaining before the user's specified deadline expires. The
-special value 0 indicates that the codec should take as long as
-necessary to provide the best quality frame. This example gives the
-codec 15ms (15000us) to return a frame. Remember that this is a soft
-deadline, and the codec may exceed it doing its regular processing. In
-these cases, no additional postprocessing will be done.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DECODE
-/* Decode the frame with 15ms deadline */
-if(vpx_codec_decode(&codec, frame, frame_sz, NULL, 15000))
- die_codec(&codec, "Failed to decode frame");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DECODE
-
-
-Codec Specific Postprocessing Controls
---------------------------------------
-Some codecs provide fine grained controls over their built-in
-postprocessors. VP8 is one example. The following sample code toggles
-postprocessing on and off every 15 frames.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PRE_DECODE
-#if CONFIG_VP8_DECODER
-if(frame_cnt%30 == 1) {
- vp8_postproc_cfg_t pp = {0, 0, 0};
-
- if(vpx_codec_control(&codec, VP8_SET_POSTPROC, &pp))
- die_codec(&codec, "Failed to turn off postproc");
-} else if(frame_cnt%30 == 16) {
- vp8_postproc_cfg_t pp = {VP8_DEBLOCK | VP8_DEMACROBLOCK, 4, 0};
-
- if(vpx_codec_control(&codec, VP8_SET_POSTPROC, &pp))
- die_codec(&codec, "Failed to turn on postproc");
-};
-#endif
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PRE_DECODE
diff --git a/examples/twopass_encoder.txt b/examples/twopass_encoder.txt
deleted file mode 100644
index 4683bc7..0000000
--- a/examples/twopass_encoder.txt
+++ /dev/null
@@ -1,75 +0,0 @@
-@TEMPLATE encoder_tmpl.c
-Two Pass Encoder
-================
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
-This is an example of a two pass encoder loop. It takes an input file in
-YV12 format, passes it through the encoder twice, and writes the compressed
-frames to disk in IVF format. It builds upon the simple_encoder example.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
-
-
-Twopass Variables
------------------
-Twopass mode needs to track the current pass number and the buffer of
-statistics packets.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_VARS
-int pass;
-vpx_fixed_buf_t stats = {0};
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_VARS
-
-
-Updating The Configuration
----------------------------------
-In two pass mode, the configuration has to be updated on each pass. The
-statistics buffer is passed on the last pass.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_LOOP_BEGIN
-for(pass=0; pass<2; pass++) {
- frame_cnt = 0;
-
- if(pass == 0)
- cfg.g_pass = VPX_RC_FIRST_PASS;
- else {
- cfg.g_pass = VPX_RC_LAST_PASS;
- cfg.rc_twopass_stats_in = stats;
- }
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_LOOP_BEGIN
-
-
-Encoding A Frame
-----------------
-Encoding a frame in two pass mode is identical to the simple encoder
-example, except the deadline is set to VPX_DL_BEST_QUALITY to get the
-best quality possible. VPX_DL_GOOD_QUALITY could also be used.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENCODE_FRAME
-frame_avail = read_frame(infile, &raw);
-if(vpx_codec_encode(&codec, frame_avail? &raw : NULL, frame_cnt,
- 1, flags, VPX_DL_BEST_QUALITY))
- die_codec(&codec, "Failed to encode frame");
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENCODE_FRAME
-
-
-Processing Statistics Packets
------------------------------
-Each packet of type `VPX_CODEC_CX_FRAME_PKT` contains the encoded data
-for this frame. We write a IVF frame header, followed by the raw data.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_STATS
-case VPX_CODEC_STATS_PKT:
- stats.buf = realloc(stats.buf, stats.sz
- + pkt->data.twopass_stats.sz);
- if(!stats.buf)
- die("Memory reallocation failed.\n");
- memcpy((char*)stats.buf + stats.sz,
- pkt->data.twopass_stats.buf,
- pkt->data.twopass_stats.sz);
- stats.sz += pkt->data.twopass_stats.sz;
- break;
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_STATS
-
-
-Pass Progress Reporting
------------------------------
-It's sometimes helpful to see when each pass completes.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_LOOP_END
- printf("Pass %d complete.\n", pass+1);
-}
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_LOOP_END
diff --git a/examples/vp8_scalable_patterns.txt b/examples/vp8_scalable_patterns.txt
deleted file mode 100644
index e1d5dbd..0000000
--- a/examples/vp8_scalable_patterns.txt
+++ /dev/null
@@ -1,143 +0,0 @@
-@TEMPLATE encoder_tmpl.c
-VP8 Scalable Frame Patterns
-===========================
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
-This is an example demonstrating how to control the VP8 encoder's
-reference frame selection and update mechanism for video applications
-that benefit from a scalable bitstream.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
-
-
-Configuration
--------------
-Scalable frame patterns are most useful in an error resilient context,
-so error resiliency mode is enabled, as in the `error_resilient.c`
-example. In addition, we want to disable automatic keyframe selection,
-so we force an interval of 1000 frames.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_SET_CFG2
-
-/* Enable error resilient mode */
-cfg.g_error_resilient = 1;
-cfg.g_lag_in_frames = 0;
-cfg.kf_mode = VPX_KF_FIXED;
-
-/* Disable automatic keyframe placement */
-cfg.kf_min_dist = cfg.kf_max_dist = 1000;
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_SET_CFG2
-
-This example uses the following frame pattern (L->last_frame,
-G->golden_frame, A->alt_ref_frame):
-
-* Frame 0 Intra, use none, update L&G&A
-* Frame 1 Inter, use LGA, update none
-* Frame 2 Inter, use LGA, update L
-* Frame 3 Inter, use LGA, update none
-* Frame 4 Inter, use GA, update L&G
-* Frame 5 Inter, use LGA, update none
-* Frame 6 Inter, use LGA, update L
-* Frame 7 Inter, use LGA, update none
-* Frame 8 Inter, use A, update L&G&A
-* Frame 9 Inter, use LGA, update none
-* Frame 10 Inter, use LGA, update L
-* Frame 11 Inter, use LGA, update none
-* Frame 12 Inter, use GA, update L&G
-* Frame 13 Inter, use LGA, update none
-* Frame 14 Inter, use LGA, update L
-* Frame 15 Inter, use LGA, update none
-* ...Repeats the pattern from frame 0
-
-Change this variable to test the 3 decodable streams case.
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_VARS
-int num_streams = 5;
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_VARS
-
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PER_FRAME_CFG
-flags = 0;
-if(num_streams == 5)
-{
- switch(frame_cnt % 16) {
- case 0:
- flags |= VPX_EFLAG_FORCE_KF;
- flags |= VP8_EFLAG_FORCE_GF;
- flags |= VP8_EFLAG_FORCE_ARF;
- break;
- case 1:
- case 3:
- case 5:
- case 7:
- case 9:
- case 11:
- case 13:
- case 15:
- flags |= VP8_EFLAG_NO_UPD_LAST;
- flags |= VP8_EFLAG_NO_UPD_GF;
- flags |= VP8_EFLAG_NO_UPD_ARF;
- break;
- case 2:
- case 6:
- case 10:
- case 14:
- break;
- case 4:
- flags |= VP8_EFLAG_NO_REF_LAST;
- flags |= VP8_EFLAG_FORCE_GF;
- break;
- case 8:
- flags |= VP8_EFLAG_NO_REF_LAST;
- flags |= VP8_EFLAG_NO_REF_GF;
- flags |= VP8_EFLAG_FORCE_GF;
- flags |= VP8_EFLAG_FORCE_ARF;
- break;
- case 12:
- flags |= VP8_EFLAG_NO_REF_LAST;
- flags |= VP8_EFLAG_FORCE_GF;
- break;
- }
-}
-else
-{
- switch(frame_cnt % 9) {
- case 0:
- if(frame_cnt==0)
- {
- flags |= VPX_EFLAG_FORCE_KF;
- }
- else
- {
- cfg.rc_max_quantizer = 26;
- cfg.rc_min_quantizer = 0;
- cfg.rc_target_bitrate = 300;
- flags |= VP8_EFLAG_NO_REF_LAST;
- flags |= VP8_EFLAG_NO_REF_ARF;
- }
- flags |= VP8_EFLAG_FORCE_GF;
- flags |= VP8_EFLAG_FORCE_ARF;
- break;
- case 1:
- case 2:
- case 4:
- case 5:
- case 7:
- case 8:
- cfg.rc_max_quantizer = 45;
- cfg.rc_min_quantizer = 0;
- cfg.rc_target_bitrate = 230;
- break;
- case 3:
- case 6:
- cfg.rc_max_quantizer = 45;
- cfg.rc_min_quantizer = 0;
- cfg.rc_target_bitrate = 215;
- flags |= VP8_EFLAG_NO_REF_LAST;
- flags |= VP8_EFLAG_FORCE_ARF;
- break;
- }
-}
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PER_FRAME_CFG
-
-Observing The Effects
----------------------
-Use the `decode_with_drops` example to decode with various dropped frame
-patterns. Good patterns to start with are 1/2, 3/4, 7/8, and 15/16
-drops.
diff --git a/generic/libvpx_srcs.txt b/generic/libvpx_srcs.txt
new file mode 100644
index 0000000..4ee59ef
--- /dev/null
+++ b/generic/libvpx_srcs.txt
@@ -0,0 +1,113 @@
+build/make/rtcd.sh
+build/make/version.sh
+CHANGELOG
+libs.mk
+vp8/common/alloccommon.c
+vp8/common/alloccommon.h
+vp8/common/asm_com_offsets.c
+vp8/common/blockd.c
+vp8/common/blockd.h
+vp8/common/coefupdateprobs.h
+vp8/common/common.h
+vp8/common/debugmodes.c
+vp8/common/default_coef_probs.h
+vp8/common/dequantize.c
+vp8/common/entropy.c
+vp8/common/entropy.h
+vp8/common/entropymode.c
+vp8/common/entropymode.h
+vp8/common/entropymv.c
+vp8/common/entropymv.h
+vp8/common/extend.c
+vp8/common/extend.h
+vp8/common/filter.c
+vp8/common/filter.h
+vp8/common/findnearmv.c
+vp8/common/findnearmv.h
+vp8/common/generic/systemdependent.c
+vp8/common/header.h
+vp8/common/idct_blk.c
+vp8/common/idctllm.c
+vp8/common/invtrans.h
+vp8/common/loopfilter.c
+vp8/common/loopfilter_filters.c
+vp8/common/loopfilter.h
+vp8/common/mbpitch.c
+vp8/common/modecont.c
+vp8/common/modecont.h
+vp8/common/mv.h
+vp8/common/onyxc_int.h
+vp8/common/onyxd.h
+vp8/common/onyx.h
+vp8/common/ppflags.h
+vp8/common/pragmas.h
+vp8/common/quant_common.c
+vp8/common/quant_common.h
+vp8/common/reconinter.c
+vp8/common/reconinter.h
+vp8/common/reconintra4x4.c
+vp8/common/reconintra4x4.h
+vp8/common/reconintra.c
+vp8/common/rtcd.c
+vp8/common/rtcd_defs.sh
+vp8/common/sad_c.c
+vp8/common/setupintrarecon.c
+vp8/common/setupintrarecon.h
+vp8/common/swapyv12buffer.c
+vp8/common/swapyv12buffer.h
+vp8/common/systemdependent.h
+vp8/common/threading.h
+vp8/common/treecoder.c
+vp8/common/treecoder.h
+vp8/common/variance_c.c
+vp8/common/variance.h
+vp8/common/vp8_entropymodedata.h
+vp8/decoder/asm_dec_offsets.c
+vp8/decoder/dboolhuff.c
+vp8/decoder/dboolhuff.h
+vp8/decoder/decodemv.c
+vp8/decoder/decodemv.h
+vp8/decoder/decoderthreading.h
+vp8/decoder/decodframe.c
+vp8/decoder/detokenize.c
+vp8/decoder/detokenize.h
+vp8/decoder/onyxd_if.c
+vp8/decoder/onyxd_int.h
+vp8/decoder/threading.c
+vp8/decoder/treereader.h
+vp8/vp8_common.mk
+vp8/vp8_dx_iface.c
+vp8/vp8dx.mk
+vpx_config.c
+vpx/internal/vpx_codec_internal.h
+vpx_mem/include/vpx_mem_intrnl.h
+vpx_mem/vpx_mem.c
+vpx_mem/vpx_mem.h
+vpx_mem/vpx_mem.mk
+vpx_ports/asm_offsets.h
+vpx_ports/mem.h
+vpx_ports/vpx_ports.mk
+vpx_ports/vpx_timer.h
+vpx_scale/generic/gen_scalers.c
+vpx_scale/generic/vpxscale.c
+vpx_scale/generic/yv12config.c
+vpx_scale/generic/yv12extend.c
+vpx_scale/generic/yv12extend_generic.h
+vpx_scale/scale_mode.h
+vpx_scale/vpxscale.h
+vpx_scale/vpx_scale.mk
+vpx_scale/yv12config.h
+vpx/src/vpx_codec.c
+vpx/src/vpx_decoder.c
+vpx/src/vpx_encoder.c
+vpx/src/vpx_image.c
+vpx/vp8dx.h
+vpx/vp8.h
+vpx/vpx_codec.h
+vpx/vpx_codec_impl_bottom.h
+vpx/vpx_codec_impl_top.h
+vpx/vpx_codec.mk
+vpx/vpx_decoder.h
+vpx/vpx_encoder.h
+vpx/vpx_image.h
+vpx/vpx_integer.h
diff --git a/generic/vpx_config.c b/generic/vpx_config.c
new file mode 100644
index 0000000..15c48f4
--- /dev/null
+++ b/generic/vpx_config.c
@@ -0,0 +1,9 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+static const char* const cfg = "--force-target=generic-gnu --disable-vp8-encoder --disable-examples --disable-docs";
+const char *vpx_codec_build_config(void) {return cfg;}
diff --git a/generic/vpx_config.h b/generic/vpx_config.h
new file mode 100644
index 0000000..84eb822
--- /dev/null
+++ b/generic/vpx_config.h
@@ -0,0 +1,80 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+/* This file automatically generated by configure. Do not edit! */
+#ifndef VPX_CONFIG_H
+#define VPX_CONFIG_H
+#define RESTRICT
+#define ARCH_ARM 0
+#define ARCH_MIPS 0
+#define ARCH_X86 0
+#define ARCH_X86_64 0
+#define ARCH_PPC32 0
+#define ARCH_PPC64 0
+#define HAVE_EDSP 0
+#define HAVE_MEDIA 0
+#define HAVE_NEON 0
+#define HAVE_MIPS32 0
+#define HAVE_DSPR2 0
+#define HAVE_MMX 0
+#define HAVE_SSE 0
+#define HAVE_SSE2 0
+#define HAVE_SSE3 0
+#define HAVE_SSSE3 0
+#define HAVE_SSE4_1 0
+#define HAVE_ALTIVEC 0
+#define HAVE_VPX_PORTS 1
+#define HAVE_STDINT_H 1
+#define HAVE_ALT_TREE_LAYOUT 0
+#define HAVE_PTHREAD_H 1
+#define HAVE_SYS_MMAN_H 1
+#define HAVE_UNISTD_H 1
+#define CONFIG_EXTERNAL_BUILD 0
+#define CONFIG_INSTALL_DOCS 1
+#define CONFIG_INSTALL_BINS 1
+#define CONFIG_INSTALL_LIBS 1
+#define CONFIG_INSTALL_SRCS 0
+#define CONFIG_DEBUG 0
+#define CONFIG_GPROF 0
+#define CONFIG_GCOV 0
+#define CONFIG_RVCT 0
+#define CONFIG_GCC 1
+#define CONFIG_MSVS 0
+#define CONFIG_PIC 0
+#define CONFIG_BIG_ENDIAN 0
+#define CONFIG_CODEC_SRCS 0
+#define CONFIG_DEBUG_LIBS 0
+#define CONFIG_FAST_UNALIGNED 1
+#define CONFIG_MEM_MANAGER 0
+#define CONFIG_MEM_TRACKER 0
+#define CONFIG_MEM_CHECKS 0
+#define CONFIG_MD5 1
+#define CONFIG_DEQUANT_TOKENS 0
+#define CONFIG_DC_RECON 0
+#define CONFIG_RUNTIME_CPU_DETECT 0
+#define CONFIG_POSTPROC 0
+#define CONFIG_MULTITHREAD 1
+#define CONFIG_INTERNAL_STATS 0
+#define CONFIG_VP8_ENCODER 0
+#define CONFIG_VP8_DECODER 1
+#define CONFIG_VP8 1
+#define CONFIG_ENCODERS 0
+#define CONFIG_DECODERS 1
+#define CONFIG_STATIC_MSVCRT 0
+#define CONFIG_SPATIAL_RESAMPLING 1
+#define CONFIG_REALTIME_ONLY 0
+#define CONFIG_ONTHEFLY_BITPACKING 0
+#define CONFIG_ERROR_CONCEALMENT 0
+#define CONFIG_SHARED 0
+#define CONFIG_STATIC 1
+#define CONFIG_SMALL 0
+#define CONFIG_POSTPROC_VISUALIZER 0
+#define CONFIG_OS_SUPPORT 1
+#define CONFIG_UNIT_TESTS 1
+#define CONFIG_MULTI_RES_ENCODING 0
+#define CONFIG_TEMPORAL_DENOISING 1
+#endif /* VPX_CONFIG_H */
diff --git a/generic/vpx_rtcd.h b/generic/vpx_rtcd.h
new file mode 100644
index 0000000..7d57578
--- /dev/null
+++ b/generic/vpx_rtcd.h
@@ -0,0 +1,296 @@
+#ifndef VPX_RTCD_
+#define VPX_RTCD_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+#include "vp8/common/blockd.h"
+
+struct blockd;
+struct macroblockd;
+struct loop_filter_info;
+
+/* Encoder forward decls */
+struct block;
+struct macroblock;
+struct variance_vtable;
+union int_mv;
+struct yv12_buffer_config;
+
+void vp8_dequantize_b_c(struct blockd*, short *dqc);
+#define vp8_dequantize_b vp8_dequantize_b_c
+
+void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *output, int stride);
+#define vp8_dequant_idct_add vp8_dequant_idct_add_c
+
+void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_c
+
+void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_c
+
+void vp8_loop_filter_mbv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbv vp8_loop_filter_mbv_c
+
+void vp8_loop_filter_bv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bv vp8_loop_filter_bv_c
+
+void vp8_loop_filter_mbh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbh vp8_loop_filter_mbh_c
+
+void vp8_loop_filter_bh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bh vp8_loop_filter_bh_c
+
+void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbv vp8_loop_filter_simple_vertical_edge_c
+
+void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbh vp8_loop_filter_simple_horizontal_edge_c
+
+void vp8_loop_filter_bvs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bv vp8_loop_filter_bvs_c
+
+void vp8_loop_filter_bhs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bh vp8_loop_filter_bhs_c
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+#define vp8_short_idct4x4llm vp8_short_idct4x4llm_c
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_c
+
+void vp8_short_inv_walsh4x4_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4 vp8_short_inv_walsh4x4_c
+
+void vp8_dc_only_idct_add_c(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+#define vp8_dc_only_idct_add vp8_dc_only_idct_add_c
+
+void vp8_copy_mem16x16_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem16x16 vp8_copy_mem16x16_c
+
+void vp8_copy_mem8x8_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x8 vp8_copy_mem8x8_c
+
+void vp8_copy_mem8x4_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x4 vp8_copy_mem8x4_c
+
+void vp8_build_intra_predictors_mby_s_c(struct macroblockd *x, unsigned char * yabove_row, unsigned char * yleft, int left_stride, unsigned char * ypred_ptr, int y_stride);
+#define vp8_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s_c
+
+void vp8_build_intra_predictors_mbuv_s_c(struct macroblockd *x, unsigned char * uabove_row, unsigned char * vabove_row, unsigned char *uleft, unsigned char *vleft, int left_stride, unsigned char * upred_ptr, unsigned char * vpred_ptr, int pred_stride);
+#define vp8_build_intra_predictors_mbuv_s vp8_build_intra_predictors_mbuv_s_c
+
+void vp8_intra4x4_predict_c(unsigned char *Above, unsigned char *yleft, int left_stride, B_PREDICTION_MODE b_mode, unsigned char *dst, int dst_stride, unsigned char top_left);
+#define vp8_intra4x4_predict vp8_intra4x4_predict_c
+
+void vp8_sixtap_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict16x16 vp8_sixtap_predict16x16_c
+
+void vp8_sixtap_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x8 vp8_sixtap_predict8x8_c
+
+void vp8_sixtap_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x4 vp8_sixtap_predict8x4_c
+
+void vp8_sixtap_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict4x4 vp8_sixtap_predict4x4_c
+
+void vp8_bilinear_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict16x16 vp8_bilinear_predict16x16_c
+
+void vp8_bilinear_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x8 vp8_bilinear_predict8x8_c
+
+void vp8_bilinear_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x4 vp8_bilinear_predict8x4_c
+
+void vp8_bilinear_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict4x4 vp8_bilinear_predict4x4_c
+
+unsigned int vp8_variance4x4_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance4x4 vp8_variance4x4_c
+
+unsigned int vp8_variance8x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x8 vp8_variance8x8_c
+
+unsigned int vp8_variance8x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x16 vp8_variance8x16_c
+
+unsigned int vp8_variance16x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x8 vp8_variance16x8_c
+
+unsigned int vp8_variance16x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x16 vp8_variance16x16_c
+
+unsigned int vp8_sub_pixel_variance4x4_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance4x4 vp8_sub_pixel_variance4x4_c
+
+unsigned int vp8_sub_pixel_variance8x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x8 vp8_sub_pixel_variance8x8_c
+
+unsigned int vp8_sub_pixel_variance8x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x16 vp8_sub_pixel_variance8x16_c
+
+unsigned int vp8_sub_pixel_variance16x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x8 vp8_sub_pixel_variance16x8_c
+
+unsigned int vp8_sub_pixel_variance16x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x16 vp8_sub_pixel_variance16x16_c
+
+unsigned int vp8_variance_halfpixvar16x16_h_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_c
+
+unsigned int vp8_variance_halfpixvar16x16_v_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_c
+
+unsigned int vp8_variance_halfpixvar16x16_hv_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_c
+
+unsigned int vp8_sad4x4_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad4x4 vp8_sad4x4_c
+
+unsigned int vp8_sad8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x8 vp8_sad8x8_c
+
+unsigned int vp8_sad8x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x16 vp8_sad8x16_c
+
+unsigned int vp8_sad16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x8 vp8_sad16x8_c
+
+unsigned int vp8_sad16x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x16 vp8_sad16x16_c
+
+void vp8_sad4x4x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x3 vp8_sad4x4x3_c
+
+void vp8_sad8x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x3 vp8_sad8x8x3_c
+
+void vp8_sad8x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x3 vp8_sad8x16x3_c
+
+void vp8_sad16x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x3 vp8_sad16x8x3_c
+
+void vp8_sad16x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x3 vp8_sad16x16x3_c
+
+void vp8_sad4x4x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad4x4x8 vp8_sad4x4x8_c
+
+void vp8_sad8x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x8x8 vp8_sad8x8x8_c
+
+void vp8_sad8x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x16x8 vp8_sad8x16x8_c
+
+void vp8_sad16x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x8x8 vp8_sad16x8x8_c
+
+void vp8_sad16x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x16x8 vp8_sad16x16x8_c
+
+void vp8_sad4x4x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x4d vp8_sad4x4x4d_c
+
+void vp8_sad8x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x4d vp8_sad8x8x4d_c
+
+void vp8_sad8x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x4d vp8_sad8x16x4d_c
+
+void vp8_sad16x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x4d vp8_sad16x8x4d_c
+
+void vp8_sad16x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x4d vp8_sad16x16x4d_c
+
+void vp8_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_4_5_scale vp8_horizontal_line_4_5_scale_c
+
+void vp8_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_4_5_scale vp8_vertical_band_4_5_scale_c
+
+void vp8_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_4_5_scale vp8_last_vertical_band_4_5_scale_c
+
+void vp8_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_3_scale vp8_horizontal_line_2_3_scale_c
+
+void vp8_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_3_scale vp8_vertical_band_2_3_scale_c
+
+void vp8_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_2_3_scale vp8_last_vertical_band_2_3_scale_c
+
+void vp8_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_5_scale vp8_horizontal_line_3_5_scale_c
+
+void vp8_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_5_scale vp8_vertical_band_3_5_scale_c
+
+void vp8_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_5_scale vp8_last_vertical_band_3_5_scale_c
+
+void vp8_horizontal_line_3_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_4_scale vp8_horizontal_line_3_4_scale_c
+
+void vp8_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_4_scale vp8_vertical_band_3_4_scale_c
+
+void vp8_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_4_scale vp8_last_vertical_band_3_4_scale_c
+
+void vp8_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_1_2_scale vp8_horizontal_line_1_2_scale_c
+
+void vp8_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_1_2_scale vp8_vertical_band_1_2_scale_c
+
+void vp8_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_1_2_scale vp8_last_vertical_band_1_2_scale_c
+
+void vp8_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_4_scale vp8_horizontal_line_5_4_scale_c
+
+void vp8_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_4_scale vp8_vertical_band_5_4_scale_c
+
+void vp8_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_3_scale vp8_horizontal_line_5_3_scale_c
+
+void vp8_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_3_scale vp8_vertical_band_5_3_scale_c
+
+void vp8_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_1_scale vp8_horizontal_line_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale vp8_vertical_band_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale_i vp8_vertical_band_2_1_scale_i_c
+
+void vp8_yv12_extend_frame_borders_c(struct yv12_buffer_config *ybf);
+#define vp8_yv12_extend_frame_borders vp8_yv12_extend_frame_borders_c
+
+void vp8_yv12_copy_frame_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_frame vp8_yv12_copy_frame_c
+
+void vp8_yv12_copy_y_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_y vp8_yv12_copy_y_c
+
+void vpx_rtcd(void);
+#include "vpx_config.h"
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+
+}
+#endif
+#endif
diff --git a/generic/vpx_version.h b/generic/vpx_version.h
new file mode 100644
index 0000000..663dd49
--- /dev/null
+++ b/generic/vpx_version.h
@@ -0,0 +1,7 @@
+#define VERSION_MAJOR 1
+#define VERSION_MINOR 1
+#define VERSION_PATCH 0
+#define VERSION_EXTRA ""
+#define VERSION_PACKED ((VERSION_MAJOR<<16)|(VERSION_MINOR<<8)|(VERSION_PATCH))
+#define VERSION_STRING_NOSP "v1.1.0"
+#define VERSION_STRING " v1.1.0"
diff --git a/libmkv/EbmlIDs.h b/libmkv/EbmlIDs.h
deleted file mode 100644
index 4297470..0000000
--- a/libmkv/EbmlIDs.h
+++ /dev/null
@@ -1,231 +0,0 @@
-// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the LICENSE file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-
-
-#ifndef MKV_DEFS_HPP
-#define MKV_DEFS_HPP 1
-
-//Commenting out values not available in webm, but available in matroska
-
-enum mkv
-{
- EBML = 0x1A45DFA3,
- EBMLVersion = 0x4286,
- EBMLReadVersion = 0x42F7,
- EBMLMaxIDLength = 0x42F2,
- EBMLMaxSizeLength = 0x42F3,
- DocType = 0x4282,
- DocTypeVersion = 0x4287,
- DocTypeReadVersion = 0x4285,
-// CRC_32 = 0xBF,
- Void = 0xEC,
- SignatureSlot = 0x1B538667,
- SignatureAlgo = 0x7E8A,
- SignatureHash = 0x7E9A,
- SignaturePublicKey = 0x7EA5,
- Signature = 0x7EB5,
- SignatureElements = 0x7E5B,
- SignatureElementList = 0x7E7B,
- SignedElement = 0x6532,
- //segment
- Segment = 0x18538067,
- //Meta Seek Information
- SeekHead = 0x114D9B74,
- Seek = 0x4DBB,
- SeekID = 0x53AB,
- SeekPosition = 0x53AC,
- //Segment Information
- Info = 0x1549A966,
-// SegmentUID = 0x73A4,
-// SegmentFilename = 0x7384,
-// PrevUID = 0x3CB923,
-// PrevFilename = 0x3C83AB,
-// NextUID = 0x3EB923,
-// NextFilename = 0x3E83BB,
-// SegmentFamily = 0x4444,
-// ChapterTranslate = 0x6924,
-// ChapterTranslateEditionUID = 0x69FC,
-// ChapterTranslateCodec = 0x69BF,
-// ChapterTranslateID = 0x69A5,
- TimecodeScale = 0x2AD7B1,
- Segment_Duration = 0x4489,
- DateUTC = 0x4461,
-// Title = 0x7BA9,
- MuxingApp = 0x4D80,
- WritingApp = 0x5741,
- //Cluster
- Cluster = 0x1F43B675,
- Timecode = 0xE7,
-// SilentTracks = 0x5854,
-// SilentTrackNumber = 0x58D7,
-// Position = 0xA7,
- PrevSize = 0xAB,
- BlockGroup = 0xA0,
- Block = 0xA1,
-// BlockVirtual = 0xA2,
-// BlockAdditions = 0x75A1,
-// BlockMore = 0xA6,
-// BlockAddID = 0xEE,
-// BlockAdditional = 0xA5,
- BlockDuration = 0x9B,
-// ReferencePriority = 0xFA,
- ReferenceBlock = 0xFB,
-// ReferenceVirtual = 0xFD,
-// CodecState = 0xA4,
-// Slices = 0x8E,
-// TimeSlice = 0xE8,
- LaceNumber = 0xCC,
-// FrameNumber = 0xCD,
-// BlockAdditionID = 0xCB,
-// MkvDelay = 0xCE,
-// Cluster_Duration = 0xCF,
- SimpleBlock = 0xA3,
-// EncryptedBlock = 0xAF,
- //Track
- Tracks = 0x1654AE6B,
- TrackEntry = 0xAE,
- TrackNumber = 0xD7,
- TrackUID = 0x73C5,
- TrackType = 0x83,
- FlagEnabled = 0xB9,
- FlagDefault = 0x88,
- FlagForced = 0x55AA,
- FlagLacing = 0x9C,
-// MinCache = 0x6DE7,
-// MaxCache = 0x6DF8,
- DefaultDuration = 0x23E383,
-// TrackTimecodeScale = 0x23314F,
-// TrackOffset = 0x537F,
-// MaxBlockAdditionID = 0x55EE,
- Name = 0x536E,
- Language = 0x22B59C,
- CodecID = 0x86,
- CodecPrivate = 0x63A2,
- CodecName = 0x258688,
-// AttachmentLink = 0x7446,
-// CodecSettings = 0x3A9697,
-// CodecInfoURL = 0x3B4040,
-// CodecDownloadURL = 0x26B240,
-// CodecDecodeAll = 0xAA,
-// TrackOverlay = 0x6FAB,
-// TrackTranslate = 0x6624,
-// TrackTranslateEditionUID = 0x66FC,
-// TrackTranslateCodec = 0x66BF,
-// TrackTranslateTrackID = 0x66A5,
- //video
- Video = 0xE0,
- FlagInterlaced = 0x9A,
-// StereoMode = 0x53B8,
- PixelWidth = 0xB0,
- PixelHeight = 0xBA,
- PixelCropBottom = 0x54AA,
- PixelCropTop = 0x54BB,
- PixelCropLeft = 0x54CC,
- PixelCropRight = 0x54DD,
- DisplayWidth = 0x54B0,
- DisplayHeight = 0x54BA,
- DisplayUnit = 0x54B2,
- AspectRatioType = 0x54B3,
-// ColourSpace = 0x2EB524,
-// GammaValue = 0x2FB523,
- FrameRate = 0x2383E3,
- //end video
- //audio
- Audio = 0xE1,
- SamplingFrequency = 0xB5,
- OutputSamplingFrequency = 0x78B5,
- Channels = 0x9F,
-// ChannelPositions = 0x7D7B,
- BitDepth = 0x6264,
- //end audio
- //content encoding
-// ContentEncodings = 0x6d80,
-// ContentEncoding = 0x6240,
-// ContentEncodingOrder = 0x5031,
-// ContentEncodingScope = 0x5032,
-// ContentEncodingType = 0x5033,
-// ContentCompression = 0x5034,
-// ContentCompAlgo = 0x4254,
-// ContentCompSettings = 0x4255,
-// ContentEncryption = 0x5035,
-// ContentEncAlgo = 0x47e1,
-// ContentEncKeyID = 0x47e2,
-// ContentSignature = 0x47e3,
-// ContentSigKeyID = 0x47e4,
-// ContentSigAlgo = 0x47e5,
-// ContentSigHashAlgo = 0x47e6,
- //end content encoding
- //Cueing Data
- Cues = 0x1C53BB6B,
- CuePoint = 0xBB,
- CueTime = 0xB3,
- CueTrackPositions = 0xB7,
- CueTrack = 0xF7,
- CueClusterPosition = 0xF1,
- CueBlockNumber = 0x5378,
-// CueCodecState = 0xEA,
-// CueReference = 0xDB,
-// CueRefTime = 0x96,
-// CueRefCluster = 0x97,
-// CueRefNumber = 0x535F,
-// CueRefCodecState = 0xEB,
- //Attachment
-// Attachments = 0x1941A469,
-// AttachedFile = 0x61A7,
-// FileDescription = 0x467E,
-// FileName = 0x466E,
-// FileMimeType = 0x4660,
-// FileData = 0x465C,
-// FileUID = 0x46AE,
-// FileReferral = 0x4675,
- //Chapters
-// Chapters = 0x1043A770,
-// EditionEntry = 0x45B9,
-// EditionUID = 0x45BC,
-// EditionFlagHidden = 0x45BD,
-// EditionFlagDefault = 0x45DB,
-// EditionFlagOrdered = 0x45DD,
-// ChapterAtom = 0xB6,
-// ChapterUID = 0x73C4,
-// ChapterTimeStart = 0x91,
-// ChapterTimeEnd = 0x92,
-// ChapterFlagHidden = 0x98,
-// ChapterFlagEnabled = 0x4598,
-// ChapterSegmentUID = 0x6E67,
-// ChapterSegmentEditionUID = 0x6EBC,
-// ChapterPhysicalEquiv = 0x63C3,
-// ChapterTrack = 0x8F,
-// ChapterTrackNumber = 0x89,
-// ChapterDisplay = 0x80,
-// ChapString = 0x85,
-// ChapLanguage = 0x437C,
-// ChapCountry = 0x437E,
-// ChapProcess = 0x6944,
-// ChapProcessCodecID = 0x6955,
-// ChapProcessPrivate = 0x450D,
-// ChapProcessCommand = 0x6911,
-// ChapProcessTime = 0x6922,
-// ChapProcessData = 0x6933,
- //Tagging
-// Tags = 0x1254C367,
-// Tag = 0x7373,
-// Targets = 0x63C0,
-// TargetTypeValue = 0x68CA,
-// TargetType = 0x63CA,
-// Tagging_TrackUID = 0x63C5,
-// Tagging_EditionUID = 0x63C9,
-// Tagging_ChapterUID = 0x63C4,
-// AttachmentUID = 0x63C6,
-// SimpleTag = 0x67C8,
-// TagName = 0x45A3,
-// TagLanguage = 0x447A,
-// TagDefault = 0x4484,
-// TagString = 0x4487,
-// TagBinary = 0x4485,
-};
-#endif
diff --git a/libmkv/EbmlWriter.c b/libmkv/EbmlWriter.c
deleted file mode 100644
index 9d564c1..0000000
--- a/libmkv/EbmlWriter.c
+++ /dev/null
@@ -1,166 +0,0 @@
-// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the LICENSE file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-
-
-#include "EbmlWriter.h"
-#include <stdlib.h>
-#include <wchar.h>
-#include <string.h>
-#if defined(_MSC_VER)
-#define LITERALU64(n) n
-#else
-#define LITERALU64(n) n##LLU
-#endif
-
-void Ebml_WriteLen(EbmlGlobal *glob, long long val)
-{
- //TODO check and make sure we are not > than 0x0100000000000000LLU
- unsigned char size = 8; //size in bytes to output
- unsigned long long minVal = LITERALU64(0x00000000000000ff); //mask to compare for byte size
-
- for (size = 1; size < 8; size ++)
- {
- if (val < minVal)
- break;
-
- minVal = (minVal << 7);
- }
-
- val |= (LITERALU64(0x000000000000080) << ((size - 1) * 7));
-
- Ebml_Serialize(glob, (void *) &val, size);
-}
-
-void Ebml_WriteString(EbmlGlobal *glob, const char *str)
-{
- const size_t size_ = strlen(str);
- const unsigned long long size = size_;
- Ebml_WriteLen(glob, size);
- //TODO: it's not clear from the spec whether the nul terminator
- //should be serialized too. For now we omit the null terminator.
- Ebml_Write(glob, str, size);
-}
-
-void Ebml_WriteUTF8(EbmlGlobal *glob, const wchar_t *wstr)
-{
- const size_t strlen = wcslen(wstr);
-
- //TODO: it's not clear from the spec whether the nul terminator
- //should be serialized too. For now we include it.
- const unsigned long long size = strlen;
-
- Ebml_WriteLen(glob, size);
- Ebml_Write(glob, wstr, size);
-}
-
-void Ebml_WriteID(EbmlGlobal *glob, unsigned long class_id)
-{
- if (class_id >= 0x01000000)
- Ebml_Serialize(glob, (void *)&class_id, 4);
- else if (class_id >= 0x00010000)
- Ebml_Serialize(glob, (void *)&class_id, 3);
- else if (class_id >= 0x00000100)
- Ebml_Serialize(glob, (void *)&class_id, 2);
- else
- Ebml_Serialize(glob, (void *)&class_id, 1);
-}
-void Ebml_SerializeUnsigned64(EbmlGlobal *glob, unsigned long class_id, uint64_t ui)
-{
- unsigned char sizeSerialized = 8 | 0x80;
- Ebml_WriteID(glob, class_id);
- Ebml_Serialize(glob, &sizeSerialized, 1);
- Ebml_Serialize(glob, &ui, 8);
-}
-
-void Ebml_SerializeUnsigned(EbmlGlobal *glob, unsigned long class_id, unsigned long ui)
-{
- unsigned char size = 8; //size in bytes to output
- unsigned char sizeSerialized = 0;
- unsigned long minVal;
-
- Ebml_WriteID(glob, class_id);
- minVal = 0x7fLU; //mask to compare for byte size
-
- for (size = 1; size < 4; size ++)
- {
- if (ui < minVal)
- {
- break;
- }
-
- minVal <<= 7;
- }
-
- sizeSerialized = 0x80 | size;
- Ebml_Serialize(glob, &sizeSerialized, 1);
- Ebml_Serialize(glob, &ui, size);
-}
-//TODO: perhaps this is a poor name for this id serializer helper function
-void Ebml_SerializeBinary(EbmlGlobal *glob, unsigned long class_id, unsigned long bin)
-{
- int size;
- for (size=4; size > 1; size--)
- {
- if (bin & 0x000000ff << ((size-1) * 8))
- break;
- }
- Ebml_WriteID(glob, class_id);
- Ebml_WriteLen(glob, size);
- Ebml_WriteID(glob, bin);
-}
-
-void Ebml_SerializeFloat(EbmlGlobal *glob, unsigned long class_id, double d)
-{
- unsigned char len = 0x88;
-
- Ebml_WriteID(glob, class_id);
- Ebml_Serialize(glob, &len, 1);
- Ebml_Serialize(glob, &d, 8);
-}
-
-void Ebml_WriteSigned16(EbmlGlobal *glob, short val)
-{
- signed long out = ((val & 0x003FFFFF) | 0x00200000) << 8;
- Ebml_Serialize(glob, &out, 3);
-}
-
-void Ebml_SerializeString(EbmlGlobal *glob, unsigned long class_id, const char *s)
-{
- Ebml_WriteID(glob, class_id);
- Ebml_WriteString(glob, s);
-}
-
-void Ebml_SerializeUTF8(EbmlGlobal *glob, unsigned long class_id, wchar_t *s)
-{
- Ebml_WriteID(glob, class_id);
- Ebml_WriteUTF8(glob, s);
-}
-
-void Ebml_SerializeData(EbmlGlobal *glob, unsigned long class_id, unsigned char *data, unsigned long data_length)
-{
- unsigned char size = 4;
- Ebml_WriteID(glob, class_id);
- Ebml_WriteLen(glob, data_length);
- Ebml_Write(glob, data, data_length);
-}
-
-void Ebml_WriteVoid(EbmlGlobal *glob, unsigned long vSize)
-{
- unsigned char tmp = 0;
- unsigned long i = 0;
-
- Ebml_WriteID(glob, 0xEC);
- Ebml_WriteLen(glob, vSize);
-
- for (i = 0; i < vSize; i++)
- {
- Ebml_Write(glob, &tmp, 1);
- }
-}
-
-//TODO Serialize Date
diff --git a/libmkv/EbmlWriter.h b/libmkv/EbmlWriter.h
deleted file mode 100644
index 8c7fe7c..0000000
--- a/libmkv/EbmlWriter.h
+++ /dev/null
@@ -1,38 +0,0 @@
-#ifndef EBMLWRITER_HPP
-#define EBMLWRITER_HPP
-
-// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the LICENSE file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-
-//note: you must define write and serialize functions as well as your own EBML_GLOBAL
-//These functions MUST be implemented
-#include <stddef.h>
-#include "vpx/vpx_integer.h"
-
-typedef struct EbmlGlobal EbmlGlobal;
-void Ebml_Serialize(EbmlGlobal *glob, const void *, unsigned long);
-void Ebml_Write(EbmlGlobal *glob, const void *, unsigned long);
-/////
-
-
-void Ebml_WriteLen(EbmlGlobal *glob, long long val);
-void Ebml_WriteString(EbmlGlobal *glob, const char *str);
-void Ebml_WriteUTF8(EbmlGlobal *glob, const wchar_t *wstr);
-void Ebml_WriteID(EbmlGlobal *glob, unsigned long class_id);
-void Ebml_SerializeUnsigned64(EbmlGlobal *glob, unsigned long class_id, uint64_t ui);
-void Ebml_SerializeUnsigned(EbmlGlobal *glob, unsigned long class_id, unsigned long ui);
-void Ebml_SerializeBinary(EbmlGlobal *glob, unsigned long class_id, unsigned long ui);
-void Ebml_SerializeFloat(EbmlGlobal *glob, unsigned long class_id, double d);
-//TODO make this more generic to signed
-void Ebml_WriteSigned16(EbmlGlobal *glob, short val);
-void Ebml_SerializeString(EbmlGlobal *glob, unsigned long class_id, const char *s);
-void Ebml_SerializeUTF8(EbmlGlobal *glob, unsigned long class_id, wchar_t *s);
-void Ebml_SerializeData(EbmlGlobal *glob, unsigned long class_id, unsigned char *data, unsigned long data_length);
-void Ebml_WriteVoid(EbmlGlobal *glob, unsigned long vSize);
-//TODO need date function
-#endif
diff --git a/libmkv/WebMElement.c b/libmkv/WebMElement.c
deleted file mode 100644
index 25a9024..0000000
--- a/libmkv/WebMElement.c
+++ /dev/null
@@ -1,220 +0,0 @@
-// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the LICENSE file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-
-
-#include "EbmlBufferWriter.h"
-#include "EbmlIDs.h"
-#include "WebMElement.h"
-#include <stdio.h>
-
-#define kVorbisPrivateMaxSize 4000
-
-void writeHeader(EbmlGlobal *glob)
-{
- EbmlLoc start;
- Ebml_StartSubElement(glob, &start, EBML);
- Ebml_SerializeUnsigned(glob, EBMLVersion, 1);
- Ebml_SerializeUnsigned(glob, EBMLReadVersion, 1); //EBML Read Version
- Ebml_SerializeUnsigned(glob, EBMLMaxIDLength, 4); //EBML Max ID Length
- Ebml_SerializeUnsigned(glob, EBMLMaxSizeLength, 8); //EBML Max Size Length
- Ebml_SerializeString(glob, DocType, "webm"); //Doc Type
- Ebml_SerializeUnsigned(glob, DocTypeVersion, 2); //Doc Type Version
- Ebml_SerializeUnsigned(glob, DocTypeReadVersion, 2); //Doc Type Read Version
- Ebml_EndSubElement(glob, &start);
-}
-
-void writeSimpleBlock(EbmlGlobal *glob, unsigned char trackNumber, short timeCode,
- int isKeyframe, unsigned char lacingFlag, int discardable,
- unsigned char *data, unsigned long dataLength)
-{
- Ebml_WriteID(glob, SimpleBlock);
- unsigned long blockLength = 4 + dataLength;
- blockLength |= 0x10000000; //TODO check length < 0x0FFFFFFFF
- Ebml_Serialize(glob, &blockLength, 4);
- trackNumber |= 0x80; //TODO check track nubmer < 128
- Ebml_Write(glob, &trackNumber, 1);
- //Ebml_WriteSigned16(glob, timeCode,2); //this is 3 bytes
- Ebml_Serialize(glob, &timeCode, 2);
- unsigned char flags = 0x00 | (isKeyframe ? 0x80 : 0x00) | (lacingFlag << 1) | discardable;
- Ebml_Write(glob, &flags, 1);
- Ebml_Write(glob, data, dataLength);
-}
-
-static UInt64 generateTrackID(unsigned int trackNumber)
-{
- UInt64 t = time(NULL) * trackNumber;
- UInt64 r = rand();
- r = r << 32;
- r += rand();
- UInt64 rval = t ^ r;
- return rval;
-}
-
-void writeVideoTrack(EbmlGlobal *glob, unsigned int trackNumber, int flagLacing,
- char *codecId, unsigned int pixelWidth, unsigned int pixelHeight,
- double frameRate)
-{
- EbmlLoc start;
- Ebml_StartSubElement(glob, &start, TrackEntry);
- Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber);
- UInt64 trackID = generateTrackID(trackNumber);
- Ebml_SerializeUnsigned(glob, TrackUID, trackID);
- Ebml_SerializeString(glob, CodecName, "VP8"); //TODO shouldn't be fixed
-
- Ebml_SerializeUnsigned(glob, TrackType, 1); //video is always 1
- Ebml_SerializeString(glob, CodecID, codecId);
- {
- EbmlLoc videoStart;
- Ebml_StartSubElement(glob, &videoStart, Video);
- Ebml_SerializeUnsigned(glob, PixelWidth, pixelWidth);
- Ebml_SerializeUnsigned(glob, PixelHeight, pixelHeight);
- Ebml_SerializeFloat(glob, FrameRate, frameRate);
- Ebml_EndSubElement(glob, &videoStart); //Video
- }
- Ebml_EndSubElement(glob, &start); //Track Entry
-}
-void writeAudioTrack(EbmlGlobal *glob, unsigned int trackNumber, int flagLacing,
- char *codecId, double samplingFrequency, unsigned int channels,
- unsigned char *private, unsigned long privateSize)
-{
- EbmlLoc start;
- Ebml_StartSubElement(glob, &start, TrackEntry);
- Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber);
- UInt64 trackID = generateTrackID(trackNumber);
- Ebml_SerializeUnsigned(glob, TrackUID, trackID);
- Ebml_SerializeUnsigned(glob, TrackType, 2); //audio is always 2
- //I am using defaults for thesed required fields
- /* Ebml_SerializeUnsigned(glob, FlagEnabled, 1);
- Ebml_SerializeUnsigned(glob, FlagDefault, 1);
- Ebml_SerializeUnsigned(glob, FlagForced, 1);
- Ebml_SerializeUnsigned(glob, FlagLacing, flagLacing);*/
- Ebml_SerializeString(glob, CodecID, codecId);
- Ebml_SerializeData(glob, CodecPrivate, private, privateSize);
-
- Ebml_SerializeString(glob, CodecName, "VORBIS"); //fixed for now
- {
- EbmlLoc AudioStart;
- Ebml_StartSubElement(glob, &AudioStart, Audio);
- Ebml_SerializeFloat(glob, SamplingFrequency, samplingFrequency);
- Ebml_SerializeUnsigned(glob, Channels, channels);
- Ebml_EndSubElement(glob, &AudioStart);
- }
- Ebml_EndSubElement(glob, &start);
-}
-void writeSegmentInformation(EbmlGlobal *ebml, EbmlLoc* startInfo, unsigned long timeCodeScale, double duration)
-{
- Ebml_StartSubElement(ebml, startInfo, Info);
- Ebml_SerializeUnsigned(ebml, TimecodeScale, timeCodeScale);
- Ebml_SerializeFloat(ebml, Segment_Duration, duration * 1000.0); //Currently fixed to using milliseconds
- Ebml_SerializeString(ebml, 0x4D80, "QTmuxingAppLibWebM-0.0.1");
- Ebml_SerializeString(ebml, 0x5741, "QTwritingAppLibWebM-0.0.1");
- Ebml_EndSubElement(ebml, startInfo);
-}
-
-/*
-void Mkv_InitializeSegment(Ebml& ebml_out, EbmlLoc& ebmlLoc)
-{
- Ebml_StartSubElement(ebml_out, ebmlLoc, 0x18538067);
-}
-
-void Mkv_InitializeSeek(Ebml& ebml_out, EbmlLoc& ebmlLoc)
-{
- Ebml_StartSubElement(ebml_out, ebmlLoc, 0x114d9b74);
-}
-void Mkv_WriteSeekInformation(Ebml& ebml_out, SeekStruct& seekInformation)
-{
- EbmlLoc ebmlLoc;
- Ebml_StartSubElement(ebml_out, ebmlLoc, 0x4dbb);
- Ebml_SerializeString(ebml_out, 0x53ab, seekInformation.SeekID);
- Ebml_SerializeUnsigned(ebml_out, 0x53ac, seekInformation.SeekPosition);
- Ebml_EndSubElement(ebml_out, ebmlLoc);
-}
-
-void Mkv_WriteSegmentInformation(Ebml& ebml_out, SegmentInformationStruct& segmentInformation)
-{
- Ebml_SerializeUnsigned(ebml_out, 0x73a4, segmentInformation.segmentUID);
- if (segmentInformation.filename != 0)
- Ebml_SerializeString(ebml_out, 0x7384, segmentInformation.filename);
- Ebml_SerializeUnsigned(ebml_out, 0x2AD7B1, segmentInformation.TimecodeScale);
- Ebml_SerializeUnsigned(ebml_out, 0x4489, segmentInformation.Duration);
- //TODO date
- Ebml_SerializeWString(ebml_out, 0x4D80, L"MKVMUX");
- Ebml_SerializeWString(ebml_out, 0x5741, segmentInformation.WritingApp);
-}
-
-void Mkv_InitializeTrack(Ebml& ebml_out, EbmlLoc& ebmlLoc)
-{
- Ebml_StartSubElement(ebml_out, ebmlLoc, 0x1654AE6B);
-}
-
-static void Mkv_WriteGenericTrackData(Ebml& ebml_out, TrackStruct& track)
-{
- Ebml_SerializeUnsigned(ebml_out, 0xD7, track.TrackNumber);
- Ebml_SerializeUnsigned(ebml_out, 0x73C5, track.TrackUID);
- Ebml_SerializeUnsigned(ebml_out, 0x83, track.TrackType);
- Ebml_SerializeUnsigned(ebml_out, 0xB9, track.FlagEnabled ? 1 :0);
- Ebml_SerializeUnsigned(ebml_out, 0x88, track.FlagDefault ? 1 :0);
- Ebml_SerializeUnsigned(ebml_out, 0x55AA, track.FlagForced ? 1 :0);
- if (track.Language != 0)
- Ebml_SerializeString(ebml_out, 0x22B59C, track.Language);
- if (track.CodecID != 0)
- Ebml_SerializeString(ebml_out, 0x86, track.CodecID);
- if (track.CodecPrivate != 0)
- Ebml_SerializeData(ebml_out, 0x63A2, track.CodecPrivate, track.CodecPrivateLength);
- if (track.CodecName != 0)
- Ebml_SerializeWString(ebml_out, 0x258688, track.CodecName);
-}
-
-void Mkv_WriteVideoTrack(Ebml& ebml_out, TrackStruct & track, VideoTrackStruct& video)
-{
- EbmlLoc trackHeadLoc, videoHeadLoc;
- Ebml_StartSubElement(ebml_out, trackHeadLoc, 0xAE); //start Track
- Mkv_WriteGenericTrackData(ebml_out, track);
- Ebml_StartSubElement(ebml_out, videoHeadLoc, 0xE0); //start Video
- Ebml_SerializeUnsigned(ebml_out, 0x9A, video.FlagInterlaced ? 1 :0);
- Ebml_SerializeUnsigned(ebml_out, 0xB0, video.PixelWidth);
- Ebml_SerializeUnsigned(ebml_out, 0xBA, video.PixelHeight);
- Ebml_SerializeUnsigned(ebml_out, 0x54B0, video.PixelDisplayWidth);
- Ebml_SerializeUnsigned(ebml_out, 0x54BA, video.PixelDisplayHeight);
- Ebml_SerializeUnsigned(ebml_out, 0x54B2, video.displayUnit);
- Ebml_SerializeFloat(ebml_out, 0x2383E3, video.FrameRate);
- Ebml_EndSubElement(ebml_out, videoHeadLoc);
- Ebml_EndSubElement(ebml_out, trackHeadLoc);
-
-}
-
-void Mkv_WriteAudioTrack(Ebml& ebml_out, TrackStruct & track, AudioTrackStruct& video)
-{
- EbmlLoc trackHeadLoc, audioHeadLoc;
- Ebml_StartSubElement(ebml_out, trackHeadLoc, 0xAE);
- Mkv_WriteGenericTrackData(ebml_out, track);
- Ebml_StartSubElement(ebml_out, audioHeadLoc, 0xE0); //start Audio
- Ebml_SerializeFloat(ebml_out, 0xB5, video.SamplingFrequency);
- Ebml_SerializeUnsigned(ebml_out, 0x9F, video.Channels);
- Ebml_SerializeUnsigned(ebml_out, 0x6264, video.BitDepth);
- Ebml_EndSubElement(ebml_out, audioHeadLoc); // end audio
- Ebml_EndSubElement(ebml_out, trackHeadLoc);
-}
-
-void Mkv_WriteEbmlClusterHead(Ebml& ebml_out, EbmlLoc& ebmlLoc, ClusterHeadStruct & clusterHead)
-{
- Ebml_StartSubElement(ebml_out, ebmlLoc, 0x1F43B675);
- Ebml_SerializeUnsigned(ebml_out, 0x6264, clusterHead.TimeCode);
-}
-
-void Mkv_WriteSimpleBlockHead(Ebml& ebml_out, EbmlLoc& ebmlLoc, SimpleBlockStruct& block)
-{
- Ebml_StartSubElement(ebml_out, ebmlLoc, 0xA3);
- Ebml_Write1UInt(ebml_out, block.TrackNumber);
- Ebml_WriteSigned16(ebml_out,block.TimeCode);
- unsigned char flags = 0x00 | (block.iskey ? 0x80:0x00) | (block.lacing << 1) | block.discardable;
- Ebml_Write1UInt(ebml_out, flags); //TODO this may be the wrong function
- Ebml_Serialize(ebml_out, block.data, block.dataLength);
- Ebml_EndSubElement(ebml_out,ebmlLoc);
-}
-*/
diff --git a/libs.mk b/libs.mk
deleted file mode 100644
index 2cb7f49..0000000
--- a/libs.mk
+++ /dev/null
@@ -1,309 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-# ARM assembly files are written in RVCT-style. We use some make magic to
-# filter those files to allow GCC compilation
-ifeq ($(ARCH_ARM),yes)
- ASM:=$(if $(filter yes,$(CONFIG_GCC)),.asm.s,.asm)
-else
- ASM:=.asm
-endif
-
-CODEC_SRCS-yes += libs.mk
-
-include $(SRC_PATH_BARE)/vpx/vpx_codec.mk
-CODEC_SRCS-yes += $(addprefix vpx/,$(call enabled,API_SRCS))
-
-include $(SRC_PATH_BARE)/vpx_mem/vpx_mem.mk
-CODEC_SRCS-yes += $(addprefix vpx_mem/,$(call enabled,MEM_SRCS))
-
-include $(SRC_PATH_BARE)/vpx_scale/vpx_scale.mk
-CODEC_SRCS-yes += $(addprefix vpx_scale/,$(call enabled,SCALE_SRCS))
-
-
-ifeq ($(CONFIG_VP8_ENCODER),yes)
- VP8_PREFIX=vp8/
- include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8cx.mk
- CODEC_SRCS-yes += $(addprefix $(VP8_PREFIX),$(call enabled,VP8_CX_SRCS))
- CODEC_EXPORTS-yes += $(addprefix $(VP8_PREFIX),$(VP8_CX_EXPORTS))
- CODEC_SRCS-yes += $(VP8_PREFIX)vp8cx.mk vpx/vp8.h vpx/vp8cx.h vpx/vp8e.h
- INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8e.h include/vpx/vp8cx.h
- INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/%
- CODEC_DOC_SRCS += vpx/vp8.h vpx/vp8cx.h
- CODEC_DOC_SECTIONS += vp8 vp8_encoder
-endif
-
-ifeq ($(CONFIG_VP8_DECODER),yes)
- VP8_PREFIX=vp8/
- include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8dx.mk
- CODEC_SRCS-yes += $(addprefix $(VP8_PREFIX),$(call enabled,VP8_DX_SRCS))
- CODEC_EXPORTS-yes += $(addprefix $(VP8_PREFIX),$(VP8_DX_EXPORTS))
- CODEC_SRCS-yes += $(VP8_PREFIX)vp8dx.mk vpx/vp8.h vpx/vp8dx.h
- INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8dx.h
- INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/%
- CODEC_DOC_SRCS += vpx/vp8.h vpx/vp8dx.h
- CODEC_DOC_SECTIONS += vp8 vp8_decoder
-endif
-
-
-ifeq ($(CONFIG_ENCODERS),yes)
- CODEC_DOC_SECTIONS += encoder
-endif
-ifeq ($(CONFIG_DECODERS),yes)
- CODEC_DOC_SECTIONS += decoder
-endif
-
-
-ifeq ($(CONFIG_MSVS),yes)
-CODEC_LIB=$(if $(CONFIG_STATIC_MSVCRT),vpxmt,vpxmd)
-# This variable uses deferred expansion intentionally, since the results of
-# $(wildcard) may change during the course of the Make.
-VS_PLATFORMS = $(foreach d,$(wildcard */Release/$(CODEC_LIB).lib),$(word 1,$(subst /, ,$(d))))
-endif
-
-# The following pairs define a mapping of locations in the distribution
-# tree to locations in the source/build trees.
-INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/vpx/%
-INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/vpx_ports/%
-INSTALL_MAPS += $(LIBSUBDIR)/% %
-INSTALL_MAPS += src/% $(SRC_PATH_BARE)/%
-ifeq ($(CONFIG_MSVS),yes)
-INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/% $(p)/Release/%)
-INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/% $(p)/Debug/%)
-endif
-
-# If this is a universal (fat) binary, then all the subarchitectures have
-# already been built and our job is to stitch them together. The
-# BUILD_LIBVPX variable indicates whether we should be building
-# (compiling, linking) the library. The LIPO_LIBVPX variable indicates
-# that we're stitching.
-$(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_LIBVPX,BUILD_LIBVPX):=yes)
-
-CODEC_SRCS-$(BUILD_LIBVPX) += build/make/version.sh
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx/vpx_integer.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/vpx_timer.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/mem.h
-CODEC_SRCS-$(BUILD_LIBVPX) += $(BUILD_PFX)vpx_config.c
-INSTALL-SRCS-no += $(BUILD_PFX)vpx_config.c
-ifeq ($(ARCH_X86)$(ARCH_X86_64),yes)
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/emms.asm
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86_abi_support.asm
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86_cpuid.c
-endif
-CODEC_SRCS-$(ARCH_ARM) += vpx_ports/arm_cpudetect.c
-CODEC_SRCS-$(ARCH_ARM) += $(BUILD_PFX)vpx_config.asm
-CODEC_EXPORTS-$(BUILD_LIBVPX) += vpx/exports_com
-CODEC_EXPORTS-$(CONFIG_ENCODERS) += vpx/exports_enc
-CODEC_EXPORTS-$(CONFIG_DECODERS) += vpx/exports_dec
-
-INSTALL-LIBS-yes += include/vpx/vpx_codec.h
-INSTALL-LIBS-yes += include/vpx/vpx_image.h
-INSTALL-LIBS-yes += include/vpx/vpx_integer.h
-INSTALL-LIBS-yes += include/vpx/vpx_codec_impl_top.h
-INSTALL-LIBS-yes += include/vpx/vpx_codec_impl_bottom.h
-INSTALL-LIBS-$(CONFIG_DECODERS) += include/vpx/vpx_decoder.h
-INSTALL-LIBS-$(CONFIG_DECODERS) += include/vpx/vpx_decoder_compat.h
-INSTALL-LIBS-$(CONFIG_ENCODERS) += include/vpx/vpx_encoder.h
-ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
-ifeq ($(CONFIG_MSVS),yes)
-INSTALL-LIBS-yes += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/$(CODEC_LIB).lib)
-INSTALL-LIBS-$(CONFIG_DEBUG_LIBS) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/$(CODEC_LIB)d.lib)
-INSTALL-LIBS-$(CONFIG_SHARED) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/vpx.dll)
-INSTALL-LIBS-$(CONFIG_SHARED) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/vpx.exp)
-endif
-else
-INSTALL-LIBS-yes += $(LIBSUBDIR)/libvpx.a
-INSTALL-LIBS-$(CONFIG_DEBUG_LIBS) += $(LIBSUBDIR)/libvpx_g.a
-endif
-
-CODEC_SRCS=$(call enabled,CODEC_SRCS)
-INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(CODEC_SRCS)
-INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(call enabled,CODEC_EXPORTS)
-
-ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
-ifeq ($(CONFIG_MSVS),yes)
-
-obj_int_extract.vcproj: $(SRC_PATH_BARE)/build/make/obj_int_extract.c
- @cp $(SRC_PATH_BARE)/build/x86-msvs/obj_int_extract.bat .
- @echo " [CREATE] $@"
- $(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh \
- --exe \
- --target=$(TOOLCHAIN) \
- --name=obj_int_extract \
- --ver=$(CONFIG_VS_VERSION) \
- --proj-guid=E1360C65-D375-4335-8057-7ED99CC3F9B2 \
- $(if $(CONFIG_STATIC_MSVCRT),--static-crt) \
- --out=$@ $^ \
- -I. \
- -I"$(SRC_PATH_BARE)" \
-
-PROJECTS-$(BUILD_LIBVPX) += obj_int_extract.vcproj
-PROJECTS-$(BUILD_LIBVPX) += obj_int_extract.bat
-
-vpx.def: $(call enabled,CODEC_EXPORTS)
- @echo " [CREATE] $@"
- $(SRC_PATH_BARE)/build/make/gen_msvs_def.sh\
- --name=vpx\
- --out=$@ $^
-CLEAN-OBJS += vpx.def
-
-vpx.vcproj: $(CODEC_SRCS) vpx.def
- @echo " [CREATE] $@"
- $(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh \
- --lib \
- --target=$(TOOLCHAIN) \
- $(if $(CONFIG_STATIC_MSVCRT),--static-crt) \
- --name=vpx \
- --proj-guid=DCE19DAF-69AC-46DB-B14A-39F0FAA5DB74 \
- --module-def=vpx.def \
- --ver=$(CONFIG_VS_VERSION) \
- --out=$@ $(CFLAGS) $^ \
- --src-path-bare="$(SRC_PATH_BARE)" \
-
-PROJECTS-$(BUILD_LIBVPX) += vpx.vcproj
-
-vpx.vcproj: vpx_config.asm
-
-endif
-else
-LIBVPX_OBJS=$(call objs,$(CODEC_SRCS))
-OBJS-$(BUILD_LIBVPX) += $(LIBVPX_OBJS)
-LIBS-$(BUILD_LIBVPX) += $(BUILD_PFX)libvpx.a $(BUILD_PFX)libvpx_g.a
-$(BUILD_PFX)libvpx_g.a: $(LIBVPX_OBJS)
-
-BUILD_LIBVPX_SO := $(if $(BUILD_LIBVPX),$(CONFIG_SHARED))
-LIBVPX_SO := libvpx.so.$(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH)
-LIBS-$(BUILD_LIBVPX_SO) += $(BUILD_PFX)$(LIBVPX_SO)
-$(BUILD_PFX)$(LIBVPX_SO): $(LIBVPX_OBJS) libvpx.ver
-$(BUILD_PFX)$(LIBVPX_SO): extralibs += -lm -pthread
-$(BUILD_PFX)$(LIBVPX_SO): SONAME = libvpx.so.$(VERSION_MAJOR)
-$(BUILD_PFX)$(LIBVPX_SO): SO_VERSION_SCRIPT = libvpx.ver
-LIBVPX_SO_SYMLINKS := $(addprefix $(LIBSUBDIR)/, \
- libvpx.so libvpx.so.$(VERSION_MAJOR) \
- libvpx.so.$(VERSION_MAJOR).$(VERSION_MINOR))
-
-libvpx.ver: $(call enabled,CODEC_EXPORTS)
- @echo " [CREATE] $@"
- $(qexec)echo "{ global:" > $@
- $(qexec)for f in $?; do awk '{print $$2";"}' < $$f >>$@; done
- $(qexec)echo "local: *; };" >> $@
-CLEAN-OBJS += libvpx.ver
-
-$(addprefix $(DIST_DIR)/,$(LIBVPX_SO_SYMLINKS)):
- @echo " [LN] $@"
- $(qexec)ln -sf $(LIBVPX_SO) $@
-
-INSTALL-LIBS-$(CONFIG_SHARED) += $(LIBVPX_SO_SYMLINKS)
-INSTALL-LIBS-$(CONFIG_SHARED) += $(LIBSUBDIR)/$(LIBVPX_SO)
-
-LIBS-$(BUILD_LIBVPX) += vpx.pc
-vpx.pc: config.mk libs.mk
- @echo " [CREATE] $@"
- $(qexec)echo '# pkg-config file from libvpx $(VERSION_STRING)' > $@
- $(qexec)echo 'prefix=$(PREFIX)' >> $@
- $(qexec)echo 'exec_prefix=$${prefix}' >> $@
- $(qexec)echo 'libdir=$${prefix}/lib' >> $@
- $(qexec)echo 'includedir=$${prefix}/include' >> $@
- $(qexec)echo '' >> $@
- $(qexec)echo 'Name: vpx' >> $@
- $(qexec)echo 'Description: WebM Project VPx codec implementation' >> $@
- $(qexec)echo 'Version: $(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH)' >> $@
- $(qexec)echo 'Requires:' >> $@
- $(qexec)echo 'Conflicts:' >> $@
- $(qexec)echo 'Libs: -L$${libdir} -lvpx' >> $@
- $(qexec)echo 'Cflags: -I$${includedir}' >> $@
-INSTALL-LIBS-yes += $(LIBSUBDIR)/pkgconfig/vpx.pc
-INSTALL_MAPS += $(LIBSUBDIR)/pkgconfig/%.pc %.pc
-CLEAN-OBJS += vpx.pc
-endif
-
-LIBS-$(LIPO_LIBVPX) += libvpx.a
-$(eval $(if $(LIPO_LIBVPX),$(call lipo_lib_template,libvpx.a)))
-
-#
-# Rule to make assembler configuration file from C configuration file
-#
-ifeq ($(ARCH_X86)$(ARCH_X86_64),yes)
-# YASM
-$(BUILD_PFX)vpx_config.asm: $(BUILD_PFX)vpx_config.h
- @echo " [CREATE] $@"
- @egrep "#define [A-Z0-9_]+ [01]" $< \
- | awk '{print $$2 " equ " $$3}' > $@
-else
-ADS2GAS=$(if $(filter yes,$(CONFIG_GCC)),| $(ASM_CONVERSION))
-$(BUILD_PFX)vpx_config.asm: $(BUILD_PFX)vpx_config.h
- @echo " [CREATE] $@"
- @egrep "#define [A-Z0-9_]+ [01]" $< \
- | awk '{print $$2 " EQU " $$3}' $(ADS2GAS) > $@
- @echo " END" $(ADS2GAS) >> $@
-CLEAN-OBJS += $(BUILD_PFX)vpx_config.asm
-endif
-
-#
-# Add assembler dependencies for configuration and offsets
-#
-$(filter %.s.o,$(OBJS-yes)): $(BUILD_PFX)vpx_config.asm
-$(filter %$(ASM).o,$(OBJS-yes)): $(BUILD_PFX)vpx_config.asm
-
-#
-# Calculate platform- and compiler-specific offsets for hand coded assembly
-#
-ifeq ($(CONFIG_EXTERNAL_BUILD),) # Visual Studio uses obj_int_extract.bat
- ifeq ($(ARCH_ARM), yes)
- asm_com_offsets.asm: obj_int_extract
- asm_com_offsets.asm: $(VP8_PREFIX)common/asm_com_offsets.c.o
- ./obj_int_extract rvds $< $(ADS2GAS) > $@
- OBJS-yes += $(VP8_PREFIX)common/asm_com_offsets.c.o
- CLEAN-OBJS += asm_com_offsets.asm
- $(filter %$(ASM).o,$(OBJS-yes)): $(BUILD_PFX)asm_com_offsets.asm
- endif
-
- ifeq ($(ARCH_ARM)$(ARCH_X86)$(ARCH_X86_64), yes)
- ifeq ($(CONFIG_VP8_ENCODER), yes)
- asm_enc_offsets.asm: obj_int_extract
- asm_enc_offsets.asm: $(VP8_PREFIX)encoder/asm_enc_offsets.c.o
- ./obj_int_extract rvds $< $(ADS2GAS) > $@
- OBJS-yes += $(VP8_PREFIX)encoder/asm_enc_offsets.c.o
- CLEAN-OBJS += asm_enc_offsets.asm
- $(filter %$(ASM).o,$(OBJS-yes)): $(BUILD_PFX)asm_enc_offsets.asm
- endif
- endif
-
- ifeq ($(ARCH_ARM), yes)
- ifeq ($(CONFIG_VP8_DECODER), yes)
- asm_dec_offsets.asm: obj_int_extract
- asm_dec_offsets.asm: $(VP8_PREFIX)decoder/asm_dec_offsets.c.o
- ./obj_int_extract rvds $< $(ADS2GAS) > $@
- OBJS-yes += $(VP8_PREFIX)decoder/asm_dec_offsets.c.o
- CLEAN-OBJS += asm_dec_offsets.asm
- $(filter %$(ASM).o,$(OBJS-yes)): $(BUILD_PFX)asm_dec_offsets.asm
- endif
- endif
-endif
-
-$(shell $(SRC_PATH_BARE)/build/make/version.sh "$(SRC_PATH_BARE)" $(BUILD_PFX)vpx_version.h)
-CLEAN-OBJS += $(BUILD_PFX)vpx_version.h
-
-CODEC_DOC_SRCS += vpx/vpx_codec.h \
- vpx/vpx_decoder.h \
- vpx/vpx_encoder.h \
- vpx/vpx_image.h
-
-CLEAN-OBJS += libs.doxy
-DOCS-yes += libs.doxy
-libs.doxy: $(CODEC_DOC_SRCS)
- @echo " [CREATE] $@"
- @rm -f $@
- @echo "INPUT += $^" >> $@
- @echo "PREDEFINED = VPX_CODEC_DISABLE_COMPAT" >> $@
- @echo "INCLUDE_PATH += ." >> $@;
- @echo "ENABLED_SECTIONS += $(sort $(CODEC_DOC_SECTIONS))" >> $@
diff --git a/libvpx.mk b/libvpx.mk
new file mode 100644
index 0000000..7b3cd9b
--- /dev/null
+++ b/libvpx.mk
@@ -0,0 +1,127 @@
+LOCAL_PATH := $(call my-dir)
+include $(CLEAR_VARS)
+
+ifeq ($(ARCH_ARM_HAVE_NEON),true)
+ libvpx_target := armv7a-neon
+ libvpx_asm := .asm.s
+else ifeq ($(ARCH_ARM_HAVE_ARMV7A),true)
+ libvpx_target := armv7a
+ libvpx_asm := .asm.s
+endif
+
+ifeq ($(TARGET_ARCH),mips)
+ ifneq ($(ARCH_HAS_BIGENDIAN),true)
+ ifeq ($(ARCH_MIPS_DSP_REV),2)
+ libvpx_target := mips-dspr2
+ LOCAL_CFLAGS += -DMIPS_DSP_REV=$(ARCH_MIPS_DSP_REV)
+ else
+ libvpx_target := mips
+ ifeq ($(ARCH_MIPS_DSP_REV),1)
+ LOCAL_CFLAGS += -DMIPS_DSP_REV=$(ARCH_MIPS_DSP_REV)
+ else
+ LOCAL_CFLAGS += -DMIPS_DSP_REV=0
+ endif #mips_dsp_rev1
+ endif #mips_dsp_rev2
+ endif #bigendian
+endif #mips
+
+# Un-optimized targets
+libvpx_target ?= generic
+libvpx_asm ?= .asm
+
+libvpx_config_dir := $(LOCAL_PATH)/$(libvpx_target)
+libvpx_source_dir := $(LOCAL_PATH)/libvpx
+
+libvpx_codec_srcs := $(shell cat $(libvpx_config_dir)/libvpx_srcs.txt)
+
+LOCAL_CFLAGS := -DHAVE_CONFIG_H=vpx_config.h
+
+LOCAL_MODULE := libvpx
+
+LOCAL_MODULE_CLASS := STATIC_LIBRARIES
+libvpx_intermediates := $(call local-intermediates-dir)
+
+# Extract the C files from the list and add them to LOCAL_SRC_FILES.
+libvpx_codec_srcs_unique := $(sort $(libvpx_codec_srcs))
+libvpx_codec_srcs_c := $(filter %.c, $(libvpx_codec_srcs_unique))
+# vpx_config.c is an auto-generated file in $(config_dir)
+libvpx_codec_srcs_c_static := $(filter-out vpx_config.c, $(libvpx_codec_srcs_c))
+LOCAL_SRC_FILES += $(addprefix libvpx/, $(libvpx_codec_srcs_c_static))
+LOCAL_SRC_FILES += $(libvpx_target)/vpx_config.c
+
+# ARM and x86 use an 'offsets' file in the assembly. It is generated by
+# tricking the compiler and generating non-functional output which is then
+# processed with grep. For ARM, this must be additionally converted from
+# RVCT (ARM's in-house compiler) format to GNU Assembler Format for gcc.
+
+# Offset files are currently used in vpx_scale for NEON and some encoder
+# functions used in both ARM and x86. These files can not be compiled and need
+# to be named accordingly to avoid auto-build rules. The encoder files are not
+# used yet but are included in the comments for future reference.
+
+libvpx_asm_offsets_intermediates := \
+ vp8/common/asm_com_offsets.intermediate \
+ vp8/decoder/asm_dec_offsets.intermediate \
+
+# vp8/encoder/asm_enc_offsets.intermediate \
+
+libvpx_asm_offsets_files := \
+ vp8/common/asm_com_offsets.asm \
+ vp8/decoder/asm_dec_offsets.asm \
+
+# vp8/encoder/asm_enc_offsets.asm \
+
+# Build the S files with inline assembly.
+COMPILE_TO_S := $(addprefix $(libvpx_intermediates)/, $(libvpx_asm_offsets_intermediates))
+$(COMPILE_TO_S) : PRIVATE_INTERMEDIATES := $(libvpx_intermediates)
+$(COMPILE_TO_S) : PRIVATE_SOURCE_DIR := $(libvpx_source_dir)
+$(COMPILE_TO_S) : PRIVATE_CONFIG_DIR := $(libvpx_config_dir)
+$(COMPILE_TO_S) : PRIVATE_CUSTOM_TOOL = $(TARGET_CC) -S $(addprefix -I, $(TARGET_C_INCLUDES)) -I $(PRIVATE_INTERMEDIATES) -I $(PRIVATE_SOURCE_DIR) -I $(PRIVATE_CONFIG_DIR) -DINLINE_ASM -o $@ $<
+$(COMPILE_TO_S) : $(libvpx_intermediates)/%.intermediate : $(libvpx_source_dir)/%.c
+ $(transform-generated-source)
+
+# Extract the offsets from the inline assembly.
+OFFSETS_GEN := $(addprefix $(libvpx_intermediates)/, $(libvpx_asm_offsets_files))
+$(OFFSETS_GEN) : PRIVATE_OFFSET_PATTERN := '^[a-zA-Z0-9_]* EQU'
+$(OFFSETS_GEN) : PRIVATE_SOURCE_DIR := $(libvpx_source_dir)
+$(OFFSETS_GEN) : PRIVATE_CUSTOM_TOOL = grep $(PRIVATE_OFFSET_PATTERN) $< | tr -d '$$\#' | perl $(PRIVATE_SOURCE_DIR)/build/make/ads2gas.pl > $@
+$(OFFSETS_GEN) : %.asm : %.intermediate
+ $(transform-generated-source)
+
+LOCAL_GENERATED_SOURCES += $(OFFSETS_GEN)
+
+# This step is only required for ARM. MIPS uses intrinsics and x86 requires an
+# assembler to pre-process its assembly files.
+libvpx_asm_srcs := $(filter %.asm.s, $(libvpx_codec_srcs_unique))
+
+# The ARM assembly sources must be converted from ADS to GAS compatible format.
+VPX_GEN := $(addprefix $(libvpx_intermediates)/, $(libvpx_asm_srcs))
+$(VPX_GEN) : PRIVATE_SOURCE_DIR := $(libvpx_source_dir)
+$(VPX_GEN) : PRIVATE_CUSTOM_TOOL = cat $< | perl $(PRIVATE_SOURCE_DIR)/build/make/ads2gas.pl > $@
+$(VPX_GEN) : $(libvpx_intermediates)/%.s : $(libvpx_source_dir)/%
+ $(transform-generated-source)
+
+LOCAL_GENERATED_SOURCES += $(VPX_GEN)
+
+LOCAL_C_INCLUDES := \
+ $(libvpx_source_dir) \
+ $(libvpx_config_dir) \
+ $(libvpx_intermediates)/vp8/common \
+ $(libvpx_intermediates)/vp8/decoder \
+
+# $(libvpx_intermediates)/vp8/encoder \
+
+libvpx_target :=
+libvpx_asm :=
+libvpx_config_dir :=
+libvpx_source_dir :=
+libvpx_codec_srcs :=
+libvpx_intermediates :=
+libvpx_codec_srcs_unique :=
+libvpx_codec_srcs_c :=
+libvpx_codec_srcs_c_static :=
+libvpx_asm_offsets_intermediates :=
+libvpx_asm_offsets_files :=
+libvpx_asm_srcs :=
+
+include $(BUILD_STATIC_LIBRARY)
diff --git a/libvpx/.mailmap b/libvpx/.mailmap
new file mode 100644
index 0000000..ba1279b
--- /dev/null
+++ b/libvpx/.mailmap
@@ -0,0 +1,8 @@
+Adrian Grange <agrange@google.com>
+Johann Koenig <johannkoenig@google.com>
+Tero Rintaluoma <teror@google.com> <tero.rintaluoma@on2.com>
+Tom Finegan <tomfinegan@google.com>
+Ralph Giles <giles@xiph.org> <giles@entropywave.com>
+Ralph Giles <giles@xiph.org> <giles@mozilla.com>
+Alpha Lam <hclam@google.com> <hclam@chromium.org>
+Deb Mukherjee <debargha@google.com>
diff --git a/libvpx/AUTHORS b/libvpx/AUTHORS
new file mode 100644
index 0000000..0937d5d
--- /dev/null
+++ b/libvpx/AUTHORS
@@ -0,0 +1,65 @@
+# This file is automatically generated from the git commit history
+# by tools/gen_authors.sh.
+
+Aaron Watry <awatry@gmail.com>
+Adrian Grange <agrange@google.com>
+Alex Converse <alex.converse@gmail.com>
+Alexis Ballier <aballier@gentoo.org>
+Alok Ahuja <waveletcoeff@gmail.com>
+Alpha Lam <hclam@google.com>
+Andoni Morales Alastruey <ylatuya@gmail.com>
+Andres Mejia <mcitadel@gmail.com>
+Aron Rosenberg <arosenberg@logitech.com>
+Attila Nagy <attilanagy@google.com>
+Deb Mukherjee <debargha@google.com>
+Fabio Pedretti <fabio.ped@libero.it>
+Frank Galligan <fgalligan@google.com>
+Fredrik Söderquist <fs@opera.com>
+Fritz Koenig <frkoenig@google.com>
+Gaute Strokkenes <gaute.strokkenes@broadcom.com>
+Giuseppe Scrivano <gscrivano@gnu.org>
+Guillermo Ballester Valor <gbvalor@gmail.com>
+Henrik Lundin <hlundin@google.com>
+James Berry <jamesberry@google.com>
+James Zern <jzern@google.com>
+Jan Kratochvil <jan.kratochvil@redhat.com>
+Jeff Faust <jfaust@google.com>
+Jeff Muizelaar <jmuizelaar@mozilla.com>
+Jim Bankoski <jimbankoski@google.com>
+Johann Koenig <johannkoenig@google.com>
+John Koleszar <jkoleszar@google.com>
+Joshua Bleecher Snyder <josh@treelinelabs.com>
+Justin Clift <justin@salasaga.org>
+Justin Lebar <justin.lebar@gmail.com>
+KO Myung-Hun <komh@chollian.net>
+Lou Quillio <louquillio@google.com>
+Luca Barbato <lu_zero@gentoo.org>
+Makoto Kato <makoto.kt@gmail.com>
+Marco Paniconi <marpan@google.com>
+Martin Ettl <ettl.martin78@googlemail.com>
+Michael Kohler <michaelkohler@live.com>
+Mike Hommey <mhommey@mozilla.com>
+Mikhal Shemer <mikhal@google.com>
+Pascal Massimino <pascal.massimino@gmail.com>
+Patrik Westin <patrik.westin@gmail.com>
+Paul Wilkins <paulwilkins@google.com>
+Pavol Rusnak <stick@gk2.sk>
+Philip Jägenstedt <philipj@opera.com>
+Priit Laes <plaes@plaes.org>
+Rafael Ávila de Espíndola <rafael.espindola@gmail.com>
+Rafaël Carré <funman@videolan.org>
+Ralph Giles <giles@xiph.org>
+Ronald S. Bultje <rbultje@google.com>
+Scott LaVarnway <slavarnway@google.com>
+Stefan Holmer <holmer@google.com>
+Taekhyun Kim <takim@nvidia.com>
+Takanori MATSUURA <t.matsuu@gmail.com>
+Tero Rintaluoma <teror@google.com>
+Thijs Vermeir <thijsvermeir@gmail.com>
+Timothy B. Terriberry <tterribe@xiph.org>
+Tom Finegan <tomfinegan@google.com>
+Yaowu Xu <yaowu@google.com>
+Yunqing Wang <yunqingwang@google.com>
+Google Inc.
+The Mozilla Foundation
+The Xiph.Org Foundation
diff --git a/libvpx/CHANGELOG b/libvpx/CHANGELOG
new file mode 100644
index 0000000..dcb9f73
--- /dev/null
+++ b/libvpx/CHANGELOG
@@ -0,0 +1,505 @@
+2012-05-09 v1.1.0 "Eider"
+ This introduces a number of enhancements, mostly focused on real-time
+ encoding. In addition, it fixes a decoder bug (first introduced in
+ Duclair) so all users of that release are encouraged to upgrade.
+
+ - Upgrading:
+ This release is ABI and API compatible with Duclair (v1.0.0). Users
+ of older releases should refer to the Upgrading notes in this
+ document for that release.
+
+ This release introduces a new temporal denoiser, controlled by the
+ VP8E_SET_NOISE_SENSITIVITY control. The temporal denoiser does not
+ currently take a strength parameter, so the control is effectively
+ a boolean - zero (off) or non-zero (on). For compatibility with
+ existing applications, the values accepted are the same as those
+ for the spatial denoiser (0-6). The temporal denoiser is enabled
+ by default, and the older spatial denoiser may be restored by
+ configuring with --disable-temporal-denoising. The temporal denoiser
+ is more computationally intensive than the spatial one.
+
+ This release removes support for a legacy, decode only API that was
+ supported, but deprecated, at the initial release of libvpx
+ (v0.9.0). This is not expected to have any impact. If you are
+ impacted, you can apply a reversion to commit 2bf8fb58 locally.
+ Please update to the latest libvpx API if you are affected.
+
+ - Enhancements:
+ Adds a motion compensated temporal denoiser to the encoder, which
+ gives higher quality than the older spatial denoiser. (See above
+ for notes on upgrading).
+
+ In addition, support for new compilers and platforms were added,
+ including:
+ improved support for XCode
+ Android x86 NDK build
+ OS/2 support
+ SunCC support
+
+ Changing resolution with vpx_codec_enc_config_set() is now
+ supported. Previously, reinitializing the codec was required to
+ change the input resolution.
+
+ The vpxenc application has initial support for producing multiple
+ encodes from the same input in one call. Resizing is not yet
+ supported, but varying other codec parameters is. Use -- to
+ delineate output streams. Options persist from one stream to the
+ next.
+
+ Also, the vpxenc application will now use a keyframe interval of
+ 5 seconds by default. Use the --kf-max-dist option to override.
+
+ - Speed:
+ Decoder performance improved 2.5% versus Duclair. Encoder speed is
+ consistent with Duclair for most material. Two pass encoding of
+ slideshow-like material will see significant improvements.
+
+ Large realtime encoding speed gains at a small quality expense are
+ possible by configuring the on-the-fly bitpacking experiment with
+ --enable-onthefly-bitpacking. Realtime encoder can be up to 13%
+ faster (ARM) depending on the number of threads and bitrate
+ settings. This technique sees constant gain over the 5-16 speed
+ range. For VC style input the loss seen is up to 0.2dB. See commit
+ 52cf4dca for further details.
+
+ - Quality:
+ On the whole, quality is consistent with the Duclair release. Some
+ tweaks:
+
+ Reduced blockiness in easy sections by applying a penalty to
+ intra modes.
+
+ Improved quality of static sections (like slideshows) with
+ two pass encoding.
+
+ Improved keyframe sizing with multiple temporal layers
+
+ - Bug Fixes:
+ Corrected alt-ref contribution to frame rate for visible updates
+ to the alt-ref buffer. This affected applications making manual
+ usage of the frame reference flags, or temporal layers.
+
+ Additional constraints were added to disable multi-frame quality
+ enhancement (MFQE) in sections of the frame where there is motion.
+ (#392)
+
+ Fixed corruption issues when vpx_codec_enc_config_set() was called
+ with spatial resampling enabled.
+
+ Fixed a decoder error introduced in Duclair where the segmentation
+ map was not being reinitialized on keyframes (#378)
+
+
+2012-01-27 v1.0.0 "Duclair"
+ Our fourth named release, focused on performance and features related to
+ real-time encoding. It also fixes a decoder crash bug introduced in
+ v0.9.7, so all users of that release are encouraged to upgrade.
+
+ - Upgrading:
+ This release is ABI incompatible with prior releases of libvpx, so the
+ "major" version number has been bumped to 1. You must recompile your
+ applications against the latest version of the libvpx headers. The
+ API remains compatible, and this should not require code changes in most
+ applications.
+
+ - Enhancements:
+ This release introduces several substantial new features to the encoder,
+ of particular interest to real time streaming applications.
+
+ Temporal scalability allows the encoder to produce a stream that can
+ be decimated to different frame rates, with independent rate targetting
+ for each substream.
+
+ Multiframe quality enhancement postprocessing can make visual quality
+ more consistent in the presence of frames that are substantially
+ different quality than the surrounding frames, as in the temporal
+ scalability case and in some forced keyframe scenarios.
+
+ Multiple-resolution encoding support allows the encoding of the
+ same content at different resolutions faster than encoding them
+ separately.
+
+ - Speed:
+ Optimization targets for this release included the decoder and the real-
+ time modes of the encoder. Decoder speed on x86 has improved 10.5% with
+ this release. Encoder improvements followed a curve where speeds 1-3
+ improved 4.0%-1.5%, speeds 4-8 improved <1%, and speeds 9-16 improved
+ 1.5% to 10.5%, respectively. "Best" mode speed is consistent with the
+ Cayuga release.
+
+ - Quality:
+ Encoder quality in the single stream case is consistent with the Cayuga
+ release.
+
+ - Bug Fixes:
+ This release fixes an OOB read decoder crash bug present in v0.9.7
+ related to the clamping of motion vectors in SPLITMV blocks. This
+ behavior could be triggered by corrupt input or by starting
+ decoding from a P-frame.
+
+
+2011-08-15 v0.9.7-p1 "Cayuga" patch 1
+ This is an incremental bugfix release against Cayuga. All users of that
+ release are strongly encouraged to upgrade.
+
+ - Fix potential OOB reads (cdae03a)
+
+ An unbounded out of bounds read was discovered when the
+ decoder was requested to perform error concealment (new in
+ Cayuga) given a frame with corrupt partition sizes.
+
+ A bounded out of bounds read was discovered affecting all
+ versions of libvpx. Given an multipartition input frame that
+ is truncated between the mode/mv partition and the first
+ residiual paritition (in the block of partition offsets), up
+ to 3 extra bytes could have been read from the source buffer.
+ The code will not take any action regardless of the contents
+ of these undefined bytes, as the truncated buffer is detected
+ immediately following the read based on the calculated
+ starting position of the coefficient partition.
+
+ - Fix potential error concealment crash when the very first frame
+ is missing or corrupt (a609be5)
+
+ - Fix significant artifacts in error concealment (a4c2211, 99d870a)
+
+ - Revert 1-pass CBR rate control changes (e961317)
+ Further testing showed this change produced undesirable visual
+ artifacts, rolling back for now.
+
+
+2011-08-02 v0.9.7 "Cayuga"
+ Our third named release, focused on a faster, higher quality, encoder.
+
+ - Upgrading:
+ This release is backwards compatible with Aylesbury (v0.9.5) and
+ Bali (v0.9.6). Users of older releases should refer to the Upgrading
+ notes in this document for that release.
+
+ - Enhancements:
+ Stereo 3D format support for vpxenc
+ Runtime detection of available processor cores.
+ Allow specifying --end-usage by enum name
+ vpxdec: test for frame corruption
+ vpxenc: add quantizer histogram display
+ vpxenc: add rate histogram display
+ Set VPX_FRAME_IS_DROPPABLE
+ update configure for ios sdk 4.3
+ Avoid text relocations in ARM vp8 decoder
+ Generate a vpx.pc file for pkg-config.
+ New ways of passing encoded data between encoder and decoder.
+
+ - Speed:
+ This release includes across-the-board speed improvements to the
+ encoder. On x86, these measure at approximately 11.5% in Best mode,
+ 21.5% in Good mode (speed 0), and 22.5% in Realtime mode (speed 6).
+ On ARM Cortex A9 with Neon extensions, real-time encoding of video
+ telephony content is 35% faster than Bali on single core and 48%
+ faster on multi-core. On the NVidia Tegra2 platform, real time
+ encoding is 40% faster than Bali.
+
+ Decoder speed was not a priority for this release, but improved
+ approximately 8.4% on x86.
+
+ Reduce motion vector search on alt-ref frame.
+ Encoder loopfilter running in its own thread
+ Reworked loopfilter to precalculate more parameters
+ SSE2/SSSE3 optimizations for build_predictors_mbuv{,_s}().
+ Make hor UV predict ~2x faster (73 vs 132 cycles) using SSSE3.
+ Removed redundant checks
+ Reduced structure sizes
+ utilize preload in ARMv6 MC/LPF/Copy routines
+ ARM optimized quantization, dfct, variance, subtract
+ Increase chrow row alignment to 16 bytes.
+ disable trellis optimization for first pass
+ Write SSSE3 sub-pixel filter function
+ Improve SSE2 half-pixel filter funtions
+ Add vp8_sub_pixel_variance16x8_ssse3 function
+ Reduce unnecessary distortion computation
+ Use diamond search to replace full search
+ Preload reference area in sub-pixel motion search (real-time mode)
+
+ - Quality:
+ This release focused primarily on one-pass use cases, including
+ video conferencing. Low latency data rate control was significantly
+ improved, improving streamability over bandwidth constrained links.
+ Added support for error concealment, allowing frames to maintain
+ visual quality in the presence of substantial packet loss.
+
+ Add rc_max_intra_bitrate_pct control
+ Limit size of initial keyframe in one-pass.
+ Improve framerate adaptation
+ Improved 1-pass CBR rate control
+ Improved KF insertion after fades to still.
+ Improved key frame detection.
+ Improved activity masking (lower PSNR impact for same SSIM boost)
+ Improved interaction between GF and ARFs
+ Adding error-concealment to the decoder.
+ Adding support for independent partitions
+ Adjusted rate-distortion constants
+
+
+ - Bug Fixes:
+ Removed firstpass motion map
+ Fix parallel make install
+ Fix multithreaded encoding for 1 MB wide frame
+ Fixed iwalsh_neon build problems with RVDS4.1
+ Fix semaphore emulation, spin-wait intrinsics on Windows
+ Fix build with xcode4 and simplify GLOBAL.
+ Mark ARM asm objects as allowing a non-executable stack.
+ Fix vpxenc encoding incorrect webm file header on big endian
+
+
+2011-03-07 v0.9.6 "Bali"
+ Our second named release, focused on a faster, higher quality, encoder.
+
+ - Upgrading:
+ This release is backwards compatible with Aylesbury (v0.9.5). Users
+ of older releases should refer to the Upgrading notes in this
+ document for that release.
+
+ - Enhancements:
+ vpxenc --psnr shows a summary when encode completes
+ --tune=ssim option to enable activity masking
+ improved postproc visualizations for development
+ updated support for Apple iOS to SDK 4.2
+ query decoder to determine which reference frames were updated
+ implemented error tracking in the decoder
+ fix pipe support on windows
+
+ - Speed:
+ Primary focus was on good quality mode, speed 0. Average improvement
+ on x86 about 40%, up to 100% on user-generated content at that speed.
+ Best quality mode speed improved 35%, and realtime speed 10-20%. This
+ release also saw significant improvement in realtime encoding speed
+ on ARM platforms.
+
+ Improved encoder threading
+ Dont pick encoder filter level when loopfilter is disabled.
+ Avoid double copying of key frames into alt and golden buffer
+ FDCT optimizations.
+ x86 sse2 temporal filter
+ SSSE3 version of fast quantizer
+ vp8_rd_pick_best_mbsegmentation code restructure
+ Adjusted breakout RD for SPLITMV
+ Changed segmentation check order
+ Improved rd_pick_intra4x4block
+ Adds armv6 optimized variance calculation
+ ARMv6 optimized sad16x16
+ ARMv6 optimized half pixel variance calculations
+ Full search SAD function optimization in SSE4.1
+ Improve MV prediction accuracy to achieve performance gain
+ Improve MV prediction in vp8_pick_inter_mode() for speed>3
+
+ - Quality:
+ Best quality mode improved PSNR 6.3%, and SSIM 6.1%. This release
+ also includes support for "activity masking," which greatly improves
+ SSIM at the expense of PSNR. For now, this feature is available with
+ the --tune=ssim option. Further experimentation in this area
+ is ongoing. This release also introduces a new rate control mode
+ called "CQ," which changes the allocation of bits within a clip to
+ the sections where they will have the most visual impact.
+
+ Tuning for the more exact quantizer.
+ Relax rate control for last few frames
+ CQ Mode
+ Limit key frame quantizer for forced key frames.
+ KF/GF Pulsing
+ Add simple version of activity masking.
+ make rdmult adaptive for intra in quantizer RDO
+ cap the best quantizer for 2nd order DC
+ change the threshold of DC check for encode breakout
+
+ - Bug Fixes:
+ Fix crash on Sparc Solaris.
+ Fix counter of fixed keyframe distance
+ ARNR filter pointer update bug fix
+ Fixed use of motion percentage in KF/GF group calc
+ Changed condition for using RD in Intra Mode
+ Fix encoder real-time only configuration.
+ Fix ARM encoder crash with multiple token partitions
+ Fixed bug first cluster timecode of webm file is wrong.
+ Fixed various encoder bugs with odd-sized images
+ vp8e_get_preview fixed when spatial resampling enabled
+ quantizer: fix assertion in fast quantizer path
+ Allocate source buffers to be multiples of 16
+ Fix for manual Golden frame frequency
+ Fix drastic undershoot in long form content
+
+
+2010-10-28 v0.9.5 "Aylesbury"
+ Our first named release, focused on a faster decoder, and a better encoder.
+
+ - Upgrading:
+ This release incorporates backwards-incompatible changes to the
+ ivfenc and ivfdec tools. These tools are now called vpxenc and vpxdec.
+
+ vpxdec
+ * the -q (quiet) option has been removed, and replaced with
+ -v (verbose). the output is quiet by default. Use -v to see
+ the version number of the binary.
+
+ * The default behavior is now to write output to a single file
+ instead of individual frames. The -y option has been removed.
+ Y4M output is the default.
+
+ * For raw I420/YV12 output instead of Y4M, the --i420 or --yv12
+ options must be specified.
+
+ $ ivfdec -o OUTPUT INPUT
+ $ vpxdec --i420 -o OUTPUT INPUT
+
+ * If an output file is not specified, the default is to write
+ Y4M to stdout. This makes piping more natural.
+
+ $ ivfdec -y -o - INPUT | ...
+ $ vpxdec INPUT | ...
+
+ * The output file has additional flexibility for formatting the
+ filename. It supports escape characters for constructing a
+ filename from the width, height, and sequence number. This
+ replaces the -p option. To get the equivalent:
+
+ $ ivfdec -p frame INPUT
+ $ vpxdec --i420 -o frame-%wx%h-%4.i420 INPUT
+
+ vpxenc
+ * The output file must be specified with -o, rather than as the
+ last argument.
+
+ $ ivfenc <options> INPUT OUTPUT
+ $ vpxenc <options> -o OUTPUT INPUT
+
+ * The output defaults to webm. To get IVF output, use the --ivf
+ option.
+
+ $ ivfenc <options> INPUT OUTPUT.ivf
+ $ vpxenc <options> -o OUTPUT.ivf --ivf INPUT
+
+
+ - Enhancements:
+ ivfenc and ivfdec have been renamed to vpxenc, vpxdec.
+ vpxdec supports .webm input
+ vpxdec writes .y4m by default
+ vpxenc writes .webm output by default
+ vpxenc --psnr now shows the average/overall PSNR at the end
+ ARM platforms now support runtime cpu detection
+ vpxdec visualizations added for motion vectors, block modes, references
+ vpxdec now silent by default
+ vpxdec --progress shows frame-by-frame timing information
+ vpxenc supports the distinction between --fps and --timebase
+ NASM is now a supported assembler
+ configure: enable PIC for shared libs by default
+ configure: add --enable-small
+ configure: support for ppc32-linux-gcc
+ configure: support for sparc-solaris-gcc
+
+ - Bugs:
+ Improve handling of invalid frames
+ Fix valgrind errors in the NEON loop filters.
+ Fix loopfilter delta zero transitions
+ Fix valgrind errors in vp8_sixtap_predict8x4_armv6().
+ Build fixes for darwin-icc
+
+ - Speed:
+ 20-40% (average 28%) improvement in libvpx decoder speed,
+ including:
+ Rewrite vp8_short_walsh4x4_sse2()
+ Optimizations on the loopfilters.
+ Miscellaneous improvements for Atom
+ Add 4-tap version of 2nd-pass ARMv6 MC filter.
+ Improved multithread utilization
+ Better instruction choices on x86
+ reorder data to use wider instructions
+ Update NEON wide idcts
+ Make block access to frame buffer sequential
+ Improved subset block search
+ Bilinear subpixel optimizations for ssse3.
+ Decrease memory footprint
+
+ Encoder speed improvements (percentage gain not measured):
+ Skip unnecessary search of identical frames
+ Add SSE2 subtract functions
+ Improve bounds checking in vp8_diamond_search_sadx4()
+ Added vp8_fast_quantize_b_sse2
+
+ - Quality:
+ Over 7% overall PSNR improvement (6.3% SSIM) in "best" quality
+ encoding mode, and up to 60% improvement on very noisy, still
+ or slow moving source video
+
+ Motion compensated temporal filter for Alt-Ref Noise Reduction
+ Improved use of trellis quantization on 2nd order Y blocks
+ Tune effect of motion on KF/GF boost in two pass
+ Allow coefficient optimization for good quality speed 0.
+ Improved control of active min quantizer for two pass.
+ Enable ARFs for non-lagged compress
+
+2010-09-02 v0.9.2
+ - Enhancements:
+ Disable frame dropping by default
+ Improved multithreaded performance
+ Improved Force Key Frame Behaviour
+ Increased rate control buffer level precision
+ Fix bug in 1st pass motion compensation
+ ivfenc: correct fixed kf interval, --disable-kf
+ - Speed:
+ Changed above and left context data layout
+ Rework idct calling structure.
+ Removed unnecessary MB_MODE_INFO copies
+ x86: SSSE3 sixtap prediction
+ Reworked IDCT to include reconstruction (add) step
+ Swap alt/gold/new/last frame buffer ptrs instead of copying.
+ Improve SSE2 loopfilter functions
+ Change bitreader to use a larger window.
+ Avoid loopfilter reinitialization when possible
+ - Quality:
+ Normalize quantizer's zero bin and rounding factors
+ Add trellis quantization.
+ Make the quantizer exact.
+ Updates to ARNR filtering algorithm
+ Fix breakout thresh computation for golden & AltRef frames
+ Redo the forward 4x4 dct
+ Improve the accuracy of forward walsh-hadamard transform
+ Further adjustment of RD behaviour with Q and Zbin.
+ - Build System:
+ Allow linking of libs built with MinGW to MSVC
+ Fix target auto-detection on mingw32
+ Allow --cpu= to work for x86.
+ configure: pass original arguments through to make dist
+ Fix builds without runtime CPU detection
+ msvs: fix install of codec sources
+ msvs: Change devenv.com command line for better msys support
+ msvs: Add vs9 targets.
+ Add x86_64-linux-icc target
+ - Bugs:
+ Potential crashes on older MinGW builds
+ Fix two-pass framrate for Y4M input.
+ Fixed simple loop filter, other crashes on ARM v6
+ arm: fix missing dependency with --enable-shared
+ configure: support directories containing .o
+ Replace pinsrw (SSE) with MMX instructions
+ apple: include proper mach primatives
+ Fixed rate control bug with long key frame interval.
+ Fix DSO link errors on x86-64 when not using a version script
+ Fixed buffer selection for UV in AltRef filtering
+
+
+2010-06-17 v0.9.1
+ - Enhancements:
+ * ivfenc/ivfdec now support YUV4MPEG2 input and pipe I/O
+ * Speed optimizations
+ - Bugfixes:
+ * Rate control
+ * Prevent out-of-bounds accesses on invalid data
+ - Build system updates:
+ * Detect toolchain to be used automatically for native builds
+ * Support building shared libraries
+ * Better autotools emulation (--prefix, --libdir, DESTDIR)
+ - Updated LICENSE
+ * http://webmproject.blogspot.com/2010/06/changes-to-webm-open-source-license.html
+
+
+2010-05-18 v0.9.0
+ - Initial open source release. Welcome to WebM and VP8!
+
diff --git a/libvpx/LICENSE b/libvpx/LICENSE
new file mode 100644
index 0000000..1ce4434
--- /dev/null
+++ b/libvpx/LICENSE
@@ -0,0 +1,31 @@
+Copyright (c) 2010, The WebM Project authors. All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are
+met:
+
+ * Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ * Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in
+ the documentation and/or other materials provided with the
+ distribution.
+
+ * Neither the name of Google, nor the WebM Project, nor the names
+ of its contributors may be used to endorse or promote products
+ derived from this software without specific prior written
+ permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
diff --git a/PATENTS b/libvpx/PATENTS
similarity index 100%
rename from PATENTS
rename to libvpx/PATENTS
diff --git a/libvpx/README b/libvpx/README
new file mode 100644
index 0000000..0475dad
--- /dev/null
+++ b/libvpx/README
@@ -0,0 +1,106 @@
+vpx Multi-Format Codec SDK
+README - 21 June 2012
+
+Welcome to the WebM VP8 Codec SDK!
+
+COMPILING THE APPLICATIONS/LIBRARIES:
+ The build system used is similar to autotools. Building generally consists of
+ "configuring" with your desired build options, then using GNU make to build
+ the application.
+
+ 1. Prerequisites
+
+ * All x86 targets require the Yasm[1] assembler be installed.
+ * All Windows builds require that Cygwin[2] be installed.
+ * Building the documentation requires PHP[3] and Doxygen[4]. If you do not
+ have these packages, you must pass --disable-install-docs to the
+ configure script.
+ * Downloading the data for the unit tests requires curl[5] and sha1sum.
+ sha1sum is provided via the GNU coreutils, installed by default on
+ many *nix platforms, as well as MinGW and Cygwin. If coreutils is not
+ available, a compatible version of sha1sum can be built from
+ source[6]. These requirements are optional if not running the unit
+ tests.
+
+ [1]: http://www.tortall.net/projects/yasm
+ [2]: http://www.cygwin.com
+ [3]: http://php.net
+ [4]: http://www.doxygen.org
+ [5]: http://curl.haxx.se
+ [6]: http://www.microbrew.org/tools/md5sha1sum/
+
+ 2. Out-of-tree builds
+ Out of tree builds are a supported method of building the application. For
+ an out of tree build, the source tree is kept separate from the object
+ files produced during compilation. For instance:
+
+ $ mkdir build
+ $ cd build
+ $ ../libvpx/configure <options>
+ $ make
+
+ 3. Configuration options
+ The 'configure' script supports a number of options. The --help option can be
+ used to get a list of supported options:
+ $ ../libvpx/configure --help
+
+ 4. Cross development
+ For cross development, the most notable option is the --target option. The
+ most up-to-date list of supported targets can be found at the bottom of the
+ --help output of the configure script. As of this writing, the list of
+ available targets is:
+
+ armv5te-android-gcc
+ armv5te-linux-rvct
+ armv5te-linux-gcc
+ armv6-darwin-gcc
+ armv6-linux-rvct
+ armv6-linux-gcc
+ armv7-android-gcc
+ armv7-linux-rvct
+ armv7-linux-gcc
+ mips32-linux-gcc
+ ppc32-darwin8-gcc
+ ppc32-darwin9-gcc
+ ppc64-darwin8-gcc
+ ppc64-darwin9-gcc
+ ppc64-linux-gcc
+ x86-darwin8-gcc
+ x86-darwin8-icc
+ x86-darwin9-gcc
+ x86-darwin9-icc
+ x86-linux-gcc
+ x86-linux-icc
+ x86-solaris-gcc
+ x86-win32-vs7
+ x86-win32-vs8
+ x86_64-darwin9-gcc
+ x86_64-linux-gcc
+ x86_64-solaris-gcc
+ x86_64-win64-vs8
+ universal-darwin8-gcc
+ universal-darwin9-gcc
+ generic-gnu
+
+ The generic-gnu target, in conjunction with the CROSS environment variable,
+ can be used to cross compile architectures that aren't explicitly listed, if
+ the toolchain is a cross GNU (gcc/binutils) toolchain. Other POSIX toolchains
+ will likely work as well. For instance, to build using the mipsel-linux-uclibc
+ toolchain, the following command could be used (note, POSIX SH syntax, adapt
+ to your shell as necessary):
+
+ $ CROSS=mipsel-linux-uclibc- ../libvpx/configure
+
+ In addition, the executables to be invoked can be overridden by specifying the
+ environment variables: CC, AR, LD, AS, STRIP, NM. Additional flags can be
+ passed to these executables with CFLAGS, LDFLAGS, and ASFLAGS.
+
+ 5. Configuration errors
+ If the configuration step fails, the first step is to look in the error log.
+ This defaults to config.err. This should give a good indication of what went
+ wrong. If not, contact us for support.
+
+SUPPORT
+ This library is an open source project supported by its community. Please
+ please email webm-discuss@webmproject.org for help.
+
diff --git a/libvpx/args.c b/libvpx/args.c
new file mode 100644
index 0000000..37ba778
--- /dev/null
+++ b/libvpx/args.c
@@ -0,0 +1,265 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdlib.h>
+#include <string.h>
+#include <limits.h>
+#include "args.h"
+
+#ifdef _MSC_VER
+#define snprintf _snprintf
+#endif
+
+#if defined(__GNUC__) && __GNUC__
+extern void die(const char *fmt, ...) __attribute__((noreturn));
+#else
+extern void die(const char *fmt, ...);
+#endif
+
+
+struct arg arg_init(char **argv)
+{
+ struct arg a;
+
+ a.argv = argv;
+ a.argv_step = 1;
+ a.name = NULL;
+ a.val = NULL;
+ a.def = NULL;
+ return a;
+}
+
+int arg_match(struct arg *arg_, const struct arg_def *def, char **argv)
+{
+ struct arg arg;
+
+ if (!argv[0] || argv[0][0] != '-')
+ return 0;
+
+ arg = arg_init(argv);
+
+ if (def->short_name
+ && strlen(arg.argv[0]) == strlen(def->short_name) + 1
+ && !strcmp(arg.argv[0] + 1, def->short_name))
+ {
+
+ arg.name = arg.argv[0] + 1;
+ arg.val = def->has_val ? arg.argv[1] : NULL;
+ arg.argv_step = def->has_val ? 2 : 1;
+ }
+ else if (def->long_name)
+ {
+ const size_t name_len = strlen(def->long_name);
+
+ if (strlen(arg.argv[0]) >= name_len + 2
+ && arg.argv[0][1] == '-'
+ && !strncmp(arg.argv[0] + 2, def->long_name, name_len)
+ && (arg.argv[0][name_len+2] == '='
+ || arg.argv[0][name_len+2] == '\0'))
+ {
+
+ arg.name = arg.argv[0] + 2;
+ arg.val = arg.name[name_len] == '=' ? arg.name + name_len + 1 : NULL;
+ arg.argv_step = 1;
+ }
+ }
+
+ if (arg.name && !arg.val && def->has_val)
+ die("Error: option %s requires argument.\n", arg.name);
+
+ if (arg.name && arg.val && !def->has_val)
+ die("Error: option %s requires no argument.\n", arg.name);
+
+ if (arg.name
+ && (arg.val || !def->has_val))
+ {
+ arg.def = def;
+ *arg_ = arg;
+ return 1;
+ }
+
+ return 0;
+}
+
+
+const char *arg_next(struct arg *arg)
+{
+ if (arg->argv[0])
+ arg->argv += arg->argv_step;
+
+ return *arg->argv;
+}
+
+
+char **argv_dup(int argc, const char **argv)
+{
+ char **new_argv = malloc((argc + 1) * sizeof(*argv));
+
+ memcpy(new_argv, argv, argc * sizeof(*argv));
+ new_argv[argc] = NULL;
+ return new_argv;
+}
+
+
+void arg_show_usage(FILE *fp, const struct arg_def *const *defs)
+{
+ char option_text[40] = {0};
+
+ for (; *defs; defs++)
+ {
+ const struct arg_def *def = *defs;
+ char *short_val = def->has_val ? " <arg>" : "";
+ char *long_val = def->has_val ? "=<arg>" : "";
+
+ if (def->short_name && def->long_name)
+ {
+ char *comma = def->has_val ? "," : ", ";
+
+ snprintf(option_text, 37, "-%s%s%s --%s%6s",
+ def->short_name, short_val, comma,
+ def->long_name, long_val);
+ }
+ else if (def->short_name)
+ snprintf(option_text, 37, "-%s%s",
+ def->short_name, short_val);
+ else if (def->long_name)
+ snprintf(option_text, 37, " --%s%s",
+ def->long_name, long_val);
+
+ fprintf(fp, " %-37s\t%s\n", option_text, def->desc);
+
+ if(def->enums)
+ {
+ const struct arg_enum_list *listptr;
+
+ fprintf(fp, " %-37s\t ", "");
+
+ for(listptr = def->enums; listptr->name; listptr++)
+ fprintf(fp, "%s%s", listptr->name,
+ listptr[1].name ? ", " : "\n");
+ }
+ }
+}
+
+
+unsigned int arg_parse_uint(const struct arg *arg)
+{
+ long int rawval;
+ char *endptr;
+
+ rawval = strtol(arg->val, &endptr, 10);
+
+ if (arg->val[0] != '\0' && endptr[0] == '\0')
+ {
+ if (rawval >= 0 && rawval <= UINT_MAX)
+ return rawval;
+
+ die("Option %s: Value %ld out of range for unsigned int\n",
+ arg->name, rawval);
+ }
+
+ die("Option %s: Invalid character '%c'\n", arg->name, *endptr);
+ return 0;
+}
+
+
+int arg_parse_int(const struct arg *arg)
+{
+ long int rawval;
+ char *endptr;
+
+ rawval = strtol(arg->val, &endptr, 10);
+
+ if (arg->val[0] != '\0' && endptr[0] == '\0')
+ {
+ if (rawval >= INT_MIN && rawval <= INT_MAX)
+ return rawval;
+
+ die("Option %s: Value %ld out of range for signed int\n",
+ arg->name, rawval);
+ }
+
+ die("Option %s: Invalid character '%c'\n", arg->name, *endptr);
+ return 0;
+}
+
+
+struct vpx_rational
+{
+ int num; /**< fraction numerator */
+ int den; /**< fraction denominator */
+};
+struct vpx_rational arg_parse_rational(const struct arg *arg)
+{
+ long int rawval;
+ char *endptr;
+ struct vpx_rational rat;
+
+ /* parse numerator */
+ rawval = strtol(arg->val, &endptr, 10);
+
+ if (arg->val[0] != '\0' && endptr[0] == '/')
+ {
+ if (rawval >= INT_MIN && rawval <= INT_MAX)
+ rat.num = rawval;
+ else die("Option %s: Value %ld out of range for signed int\n",
+ arg->name, rawval);
+ }
+ else die("Option %s: Expected / at '%c'\n", arg->name, *endptr);
+
+ /* parse denominator */
+ rawval = strtol(endptr + 1, &endptr, 10);
+
+ if (arg->val[0] != '\0' && endptr[0] == '\0')
+ {
+ if (rawval >= INT_MIN && rawval <= INT_MAX)
+ rat.den = rawval;
+ else die("Option %s: Value %ld out of range for signed int\n",
+ arg->name, rawval);
+ }
+ else die("Option %s: Invalid character '%c'\n", arg->name, *endptr);
+
+ return rat;
+}
+
+
+int arg_parse_enum(const struct arg *arg)
+{
+ const struct arg_enum_list *listptr;
+ long int rawval;
+ char *endptr;
+
+ /* First see if the value can be parsed as a raw value */
+ rawval = strtol(arg->val, &endptr, 10);
+ if (arg->val[0] != '\0' && endptr[0] == '\0')
+ {
+ /* Got a raw value, make sure it's valid */
+ for(listptr = arg->def->enums; listptr->name; listptr++)
+ if(listptr->val == rawval)
+ return rawval;
+ }
+
+ /* Next see if it can be parsed as a string */
+ for(listptr = arg->def->enums; listptr->name; listptr++)
+ if(!strcmp(arg->val, listptr->name))
+ return listptr->val;
+
+ die("Option %s: Invalid value '%s'\n", arg->name, arg->val);
+ return 0;
+}
+
+
+int arg_parse_enum_or_int(const struct arg *arg)
+{
+ if(arg->def->enums)
+ return arg_parse_enum(arg);
+ return arg_parse_int(arg);
+}
diff --git a/args.h b/libvpx/args.h
similarity index 100%
rename from args.h
rename to libvpx/args.h
diff --git a/libvpx/build/.gitattributes b/libvpx/build/.gitattributes
new file mode 100644
index 0000000..03db79b
--- /dev/null
+++ b/libvpx/build/.gitattributes
@@ -0,0 +1,2 @@
+*-vs8/*.rules -crlf
+*-msvs/*.rules -crlf
diff --git a/libvpx/build/.gitignore b/libvpx/build/.gitignore
new file mode 100644
index 0000000..1350fcb
--- /dev/null
+++ b/libvpx/build/.gitignore
@@ -0,0 +1 @@
+x86*-win32-vs*
diff --git a/libvpx/build/make/Android.mk b/libvpx/build/make/Android.mk
new file mode 100644
index 0000000..d54639a
--- /dev/null
+++ b/libvpx/build/make/Android.mk
@@ -0,0 +1,211 @@
+##
+## Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+#
+# This file is to be used for compiling libvpx for Android using the NDK.
+# In an Android project place a libvpx checkout in the jni directory.
+# Run the configure script from the jni directory. Base libvpx
+# encoder/decoder configuration will look similar to:
+# ./libvpx/configure --target=armv7-android-gcc --disable-examples \
+# --sdk-path=/opt/android-ndk-r6b/
+#
+# When targeting Android, realtime-only is enabled by default. This can
+# be overridden by adding the command line flag:
+# --disable-realtime-only
+#
+# This will create .mk files that contain variables that contain the
+# source files to compile.
+#
+# Place an Android.mk file in the jni directory that references the
+# Android.mk file in the libvpx directory:
+# LOCAL_PATH := $(call my-dir)
+# include $(CLEAR_VARS)
+# include libvpx/build/make/Android.mk
+#
+# There are currently two TARGET_ARCH_ABI targets for ARM.
+# armeabi and armeabi-v7a. armeabi-v7a is selected by creating an
+# Application.mk in the jni directory that contains:
+# APP_ABI := armeabi-v7a
+#
+# By default libvpx will detect at runtime the existance of NEON extension.
+# For this we import the 'cpufeatures' module from the NDK sources.
+# libvpx can also be configured without this runtime detection method.
+# Configuring with --disable-runtime-cpu-detect will assume presence of NEON.
+# Configuring with --disable-runtime-cpu-detect --disable-neon will remove any
+# NEON dependency.
+
+# To change to building armeabi, run ./libvpx/configure again, but with
+# --target=arm5te-android-gcc and modify the Application.mk file to
+# set APP_ABI := armeabi
+#
+# Running ndk-build will build libvpx and include it in your project.
+#
+
+CONFIG_DIR := $(LOCAL_PATH)
+LIBVPX_PATH := $(LOCAL_PATH)/libvpx
+ASM_CNV_PATH_LOCAL := $(TARGET_ARCH_ABI)/ads2gas
+ASM_CNV_PATH := $(LOCAL_PATH)/$(ASM_CNV_PATH_LOCAL)
+
+# Makefiles created by the libvpx configure process
+# This will need to be fixed to handle x86.
+ifeq ($(TARGET_ARCH_ABI),armeabi-v7a)
+ include $(CONFIG_DIR)/libs-armv7-android-gcc.mk
+else
+ include $(CONFIG_DIR)/libs-armv5te-android-gcc.mk
+endif
+
+# Rule that is normally in Makefile created by libvpx
+# configure. Used to filter out source files based on configuration.
+enabled=$(filter-out $($(1)-no),$($(1)-yes))
+
+# Override the relative path that is defined by the libvpx
+# configure process
+SRC_PATH_BARE := $(LIBVPX_PATH)
+
+# Include the list of files to be built
+include $(LIBVPX_PATH)/libs.mk
+
+# Want arm, not thumb, optimized
+LOCAL_ARM_MODE := arm
+LOCAL_CFLAGS := -O3
+
+# -----------------------------------------------------------------------------
+# Template : asm_offsets_template
+# Arguments : 1: assembly offsets file to be created
+# 2: c file to base assembly offsets on
+# Returns : None
+# Usage : $(eval $(call asm_offsets_template,<asmfile>, <srcfile>
+# Rationale : Create offsets at compile time using for structures that are
+# defined in c, but used in assembly functions.
+# -----------------------------------------------------------------------------
+define asm_offsets_template
+
+_SRC:=$(2)
+_OBJ:=$(ASM_CNV_PATH)/$$(notdir $(2)).S
+
+_FLAGS = $$($$(my)CFLAGS) \
+ $$(call get-src-file-target-cflags,$(2)) \
+ $$(call host-c-includes,$$(LOCAL_C_INCLUDES) $$(CONFIG_DIR)) \
+ $$(LOCAL_CFLAGS) \
+ $$(NDK_APP_CFLAGS) \
+ $$(call host-c-includes,$$($(my)C_INCLUDES)) \
+ -DINLINE_ASM \
+ -S \
+
+_TEXT = "Compile $$(call get-src-file-text,$(2))"
+_CC = $$(TARGET_CC)
+
+$$(eval $$(call ev-build-file))
+
+$(1) : $$(_OBJ) $(2)
+ @mkdir -p $$(dir $$@)
+ @grep $(OFFSET_PATTERN) $$< | tr -d '\#' | $(CONFIG_DIR)/$(ASM_CONVERSION) > $$@
+endef
+
+# Use ads2gas script to convert from RVCT format to GAS format. This passes
+# puts the processed file under $(ASM_CNV_PATH). Local clean rule
+# to handle removing these
+ASM_CNV_OFFSETS_DEPEND = $(ASM_CNV_PATH)/asm_com_offsets.asm
+ifeq ($(CONFIG_VP8_DECODER), yes)
+ ASM_CNV_OFFSETS_DEPEND += $(ASM_CNV_PATH)/asm_dec_offsets.asm
+endif
+ifeq ($(CONFIG_VP8_ENCODER), yes)
+ ASM_CNV_OFFSETS_DEPEND += $(ASM_CNV_PATH)/asm_enc_offsets.asm
+endif
+
+.PRECIOUS: %.asm.s
+$(ASM_CNV_PATH)/libvpx/%.asm.s: $(LIBVPX_PATH)/%.asm $(ASM_CNV_OFFSETS_DEPEND)
+ @mkdir -p $(dir $@)
+ @$(CONFIG_DIR)/$(ASM_CONVERSION) <$< > $@
+
+# For building vpx_rtcd.h, which has a rule in libs.mk
+TGT_ISA:=$(word 1, $(subst -, ,$(TOOLCHAIN)))
+target := libs
+
+LOCAL_SRC_FILES += vpx_config.c
+
+# Remove duplicate entries
+CODEC_SRCS_UNIQUE = $(sort $(CODEC_SRCS))
+
+# Pull out C files. vpx_config.c is in the immediate directory and
+# so it does not need libvpx/ prefixed like the rest of the source files.
+CODEC_SRCS_C = $(filter %.c, $(CODEC_SRCS_UNIQUE))
+LOCAL_CODEC_SRCS_C = $(filter-out vpx_config.c, $(CODEC_SRCS_C))
+
+LOCAL_SRC_FILES += $(foreach file, $(LOCAL_CODEC_SRCS_C), libvpx/$(file))
+
+# Pull out assembly files, splitting NEON from the rest. This is
+# done to specify that the NEON assembly files use NEON assembler flags.
+CODEC_SRCS_ASM_ALL = $(filter %.asm.s, $(CODEC_SRCS_UNIQUE))
+CODEC_SRCS_ASM = $(foreach v, \
+ $(CODEC_SRCS_ASM_ALL), \
+ $(if $(findstring neon,$(v)),,$(v)))
+CODEC_SRCS_ASM_ADS2GAS = $(patsubst %.s, \
+ $(ASM_CNV_PATH_LOCAL)/libvpx/%.s, \
+ $(CODEC_SRCS_ASM))
+LOCAL_SRC_FILES += $(CODEC_SRCS_ASM_ADS2GAS)
+
+ifeq ($(TARGET_ARCH_ABI),armeabi-v7a)
+ CODEC_SRCS_ASM_NEON = $(foreach v, \
+ $(CODEC_SRCS_ASM_ALL),\
+ $(if $(findstring neon,$(v)),$(v),))
+ CODEC_SRCS_ASM_NEON_ADS2GAS = $(patsubst %.s, \
+ $(ASM_CNV_PATH_LOCAL)/libvpx/%.s, \
+ $(CODEC_SRCS_ASM_NEON))
+ LOCAL_SRC_FILES += $(patsubst %.s, \
+ %.s.neon, \
+ $(CODEC_SRCS_ASM_NEON_ADS2GAS))
+endif
+
+LOCAL_CFLAGS += \
+ -DHAVE_CONFIG_H=vpx_config.h \
+ -I$(LIBVPX_PATH) \
+ -I$(ASM_CNV_PATH)
+
+LOCAL_MODULE := libvpx
+
+LOCAL_LDLIBS := -llog
+
+ifeq ($(CONFIG_RUNTIME_CPU_DETECT),yes)
+ LOCAL_STATIC_LIBRARIES := cpufeatures
+endif
+
+$(foreach file, $(LOCAL_SRC_FILES), $(LOCAL_PATH)/$(file)): vpx_rtcd.h
+
+.PHONY: clean
+clean:
+ @echo "Clean: ads2gas files [$(TARGET_ARCH_ABI)]"
+ @$(RM) $(CODEC_SRCS_ASM_ADS2GAS) $(CODEC_SRCS_ASM_NEON_ADS2GAS)
+ @$(RM) $(patsubst %.asm, %.*, $(ASM_CNV_OFFSETS_DEPEND))
+ @$(RM) -r $(ASM_CNV_PATH)
+ @$(RM) $(CLEAN-OBJS)
+
+include $(BUILD_SHARED_LIBRARY)
+
+$(eval $(call asm_offsets_template,\
+ $(ASM_CNV_PATH)/asm_com_offsets.asm, \
+ $(LIBVPX_PATH)/vp8/common/asm_com_offsets.c))
+
+ifeq ($(CONFIG_VP8_DECODER), yes)
+ $(eval $(call asm_offsets_template,\
+ $(ASM_CNV_PATH)/asm_dec_offsets.asm, \
+ $(LIBVPX_PATH)/vp8/decoder/asm_dec_offsets.c))
+endif
+
+ifeq ($(CONFIG_VP8_ENCODER), yes)
+ $(eval $(call asm_offsets_template,\
+ $(ASM_CNV_PATH)/asm_enc_offsets.asm, \
+ $(LIBVPX_PATH)/vp8/encoder/asm_enc_offsets.c))
+endif
+
+ifeq ($(CONFIG_RUNTIME_CPU_DETECT),yes)
+$(call import-module,cpufeatures)
+endif
+
diff --git a/libvpx/build/make/Makefile b/libvpx/build/make/Makefile
new file mode 100644
index 0000000..1088c84
--- /dev/null
+++ b/libvpx/build/make/Makefile
@@ -0,0 +1,388 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+include config.mk
+quiet?=true
+ifeq ($(target),)
+# If a target wasn't specified, invoke for all enabled targets.
+.DEFAULT:
+ @for t in $(ALL_TARGETS); do \
+ $(MAKE) --no-print-directory target=$$t $(MAKECMDGOALS) || exit $$?;\
+ done
+all: .DEFAULT
+clean:: .DEFAULT
+install:: .DEFAULT
+test:: .DEFAULT
+testdata:: .DEFAULT
+
+
+# Note: md5sum is not installed on OS X, but openssl is. Openssl may not be
+# installed on cygwin, so we need to autodetect here.
+md5sum := $(firstword $(wildcard \
+ $(foreach e,md5sum openssl,\
+ $(foreach p,$(subst :, ,$(PATH)),$(p)/$(e)*))\
+ ))
+md5sum := $(if $(filter %openssl,$(md5sum)),$(md5sum) dgst -md5,$(md5sum))
+
+TGT_CC:=$(word 3, $(subst -, ,$(TOOLCHAIN)))
+dist:
+ @for t in $(ALL_TARGETS); do \
+ $(MAKE) --no-print-directory target=$$t $(MAKECMDGOALS) || exit $$?;\
+ done
+ # Run configure for the user with the current toolchain.
+ @if [ -d "$(DIST_DIR)/src" ]; then \
+ mkdir -p "$(DIST_DIR)/build"; \
+ cd "$(DIST_DIR)/build"; \
+ echo "Rerunning configure $(CONFIGURE_ARGS)"; \
+ ../src/configure $(CONFIGURE_ARGS); \
+ $(if $(filter vs%,$(TGT_CC)),make NO_LAUNCH_DEVENV=1;) \
+ fi
+ @if [ -d "$(DIST_DIR)" ]; then \
+ echo " [MD5SUM] $(DIST_DIR)"; \
+ cd $(DIST_DIR) && \
+ $(md5sum) `find . -name md5sums.txt -prune -o -type f -print` \
+ | sed -e 's/MD5(\(.*\))= \([0-9a-f]\{32\}\)/\2 \1/' \
+ > md5sums.txt;\
+ fi
+
+
+endif
+
+ifneq ($(target),)
+# Normally, we want to build the filename from the target and the toolchain.
+# This disambiguates from the $(target).mk file that exists in the source tree.
+# However, the toolchain is part of the target in universal builds, so we
+# don't want to include TOOLCHAIN in that case. FAT_ARCHS is used to test
+# if we're in the universal case.
+include $(target)$(if $(FAT_ARCHS),,-$(TOOLCHAIN)).mk
+endif
+BUILD_ROOT?=.
+VPATH=$(SRC_PATH_BARE)
+CFLAGS+=-I$(BUILD_PFX)$(BUILD_ROOT) -I$(SRC_PATH)
+CXXFLAGS+=-I$(BUILD_PFX)$(BUILD_ROOT) -I$(SRC_PATH)
+ASFLAGS+=-I$(BUILD_PFX)$(BUILD_ROOT)/ -I$(SRC_PATH)/
+DIST_DIR?=dist
+HOSTCC?=gcc
+TGT_ISA:=$(word 1, $(subst -, ,$(TOOLCHAIN)))
+TGT_OS:=$(word 2, $(subst -, ,$(TOOLCHAIN)))
+TGT_CC:=$(word 3, $(subst -, ,$(TOOLCHAIN)))
+quiet:=$(if $(verbose),,yes)
+qexec=$(if $(quiet),@)
+
+# Cancel built-in implicit rules
+%: %.o
+%.asm:
+%.a:
+%: %.cc
+
+#
+# Common rules"
+#
+.PHONY: all
+all:
+
+.PHONY: clean
+clean::
+ rm -f $(OBJS-yes) $(OBJS-yes:.o=.d) $(OBJS-yes:.asm.s.o=.asm.s)
+ rm -f $(CLEAN-OBJS)
+
+.PHONY: dist
+dist:
+.PHONY: install
+install::
+.PHONY: test
+test::
+.PHONY: testdata
+testdata::
+
+$(BUILD_PFX)%.c.d: %.c
+ $(if $(quiet),@echo " [DEP] $@")
+ $(qexec)mkdir -p $(dir $@)
+ $(qexec)$(CC) $(INTERNAL_CFLAGS) $(CFLAGS) -M $< | $(fmt_deps) > $@
+
+$(BUILD_PFX)%.c.o: %.c
+ $(if $(quiet),@echo " [CC] $@")
+ $(qexec)$(CC) $(INTERNAL_CFLAGS) $(CFLAGS) -c -o $@ $<
+
+$(BUILD_PFX)%.cc.d: %.cc
+ $(if $(quiet),@echo " [DEP] $@")
+ $(qexec)mkdir -p $(dir $@)
+ $(qexec)$(CXX) $(INTERNAL_CFLAGS) $(CXXFLAGS) -M $< | $(fmt_deps) > $@
+
+$(BUILD_PFX)%.cc.o: %.cc
+ $(if $(quiet),@echo " [CXX] $@")
+ $(qexec)$(CXX) $(INTERNAL_CFLAGS) $(CXXFLAGS) -c -o $@ $<
+
+$(BUILD_PFX)%.asm.d: %.asm
+ $(if $(quiet),@echo " [DEP] $@")
+ $(qexec)mkdir -p $(dir $@)
+ $(qexec)$(SRC_PATH_BARE)/build/make/gen_asm_deps.sh \
+ --build-pfx=$(BUILD_PFX) --depfile=$@ $(ASFLAGS) $< > $@
+
+$(BUILD_PFX)%.asm.o: %.asm
+ $(if $(quiet),@echo " [AS] $@")
+ $(qexec)$(AS) $(ASFLAGS) -o $@ $<
+
+$(BUILD_PFX)%.s.d: %.s
+ $(if $(quiet),@echo " [DEP] $@")
+ $(qexec)mkdir -p $(dir $@)
+ $(qexec)$(SRC_PATH_BARE)/build/make/gen_asm_deps.sh \
+ --build-pfx=$(BUILD_PFX) --depfile=$@ $(ASFLAGS) $< > $@
+
+$(BUILD_PFX)%.s.o: %.s
+ $(if $(quiet),@echo " [AS] $@")
+ $(qexec)$(AS) $(ASFLAGS) -o $@ $<
+
+.PRECIOUS: %.c.S
+%.c.S: CFLAGS += -DINLINE_ASM
+$(BUILD_PFX)%.c.S: %.c
+ $(if $(quiet),@echo " [GEN] $@")
+ $(qexec)$(CC) -S $(CFLAGS) -o $@ $<
+
+.PRECIOUS: %.asm.s
+$(BUILD_PFX)%.asm.s: %.asm
+ $(if $(quiet),@echo " [ASM CONVERSION] $@")
+ $(qexec)mkdir -p $(dir $@)
+ $(qexec)$(ASM_CONVERSION) <$< >$@
+
+# If we're in debug mode, pretend we don't have GNU strip, to fall back to
+# the copy implementation
+HAVE_GNU_STRIP := $(if $(CONFIG_DEBUG),,$(HAVE_GNU_STRIP))
+ifeq ($(HAVE_GNU_STRIP),yes)
+# Older binutils strip global sybols not needed for relocation processing
+# when given --strip-unneeded. Use nm and awk to identify globals and
+# keep them.
+%.a: %_g.a
+ $(if $(quiet),@echo " [STRIP] $@ < $<")
+ $(qexec)$(STRIP) --strip-unneeded \
+ `$(NM) $< | grep ' [A-TV-Z] ' | awk '{print "-K"$$3'}`\
+ -o $@ $<
+else
+%.a: %_g.a
+ $(if $(quiet),@echo " [CP] $@ < $<")
+ $(qexec)cp $< $@
+endif
+
+#
+# Rule to extract assembly constants from C sources
+#
+obj_int_extract: build/make/obj_int_extract.c
+ $(if $(quiet),@echo " [HOSTCC] $@")
+ $(qexec)$(HOSTCC) -I. -I$(SRC_PATH_BARE) -o $@ $<
+CLEAN-OBJS += obj_int_extract
+
+#
+# Utility functions
+#
+pairmap=$(if $(strip $(2)),\
+ $(call $(1),$(word 1,$(2)),$(word 2,$(2)))\
+ $(call pairmap,$(1),$(wordlist 3,$(words $(2)),$(2)))\
+)
+
+enabled=$(filter-out $($(1)-no),$($(1)-yes))
+cond_enabled=$(if $(filter yes,$($(1))), $(call enabled,$(2)))
+
+find_file1=$(word 1,$(wildcard $(subst //,/,$(addsuffix /$(1),$(2)))))
+find_file=$(foreach f,$(1),$(call find_file1,$(strip $(f)),$(strip $(2))) )
+obj_pats=.c=.c.o $(AS_SFX)=$(AS_SFX).o .cc=.cc.o
+objs=$(addprefix $(BUILD_PFX),$(foreach p,$(obj_pats),$(filter %.o,$(1:$(p))) ))
+
+install_map_templates=$(eval $(call install_map_template,$(1),$(2)))
+
+not=$(subst yes,no,$(1))
+
+ifeq ($(CONFIG_MSVS),yes)
+lib_file_name=$(1).lib
+else
+lib_file_name=lib$(1).a
+endif
+#
+# Rule Templates
+#
+define linker_template
+$(1): $(filter-out -%,$(2))
+$(1):
+ $(if $(quiet),@echo " [LD] $$@")
+ $(qexec)$$(LD) $$(strip $$(INTERNAL_LDFLAGS) $$(LDFLAGS) -o $$@ $(2) $(3) $$(extralibs))
+endef
+define linkerxx_template
+$(1): $(filter-out -%,$(2))
+$(1):
+ $(if $(quiet),@echo " [LD] $$@")
+ $(qexec)$$(CXX) $$(strip $$(INTERNAL_LDFLAGS) $$(LDFLAGS) -o $$@ $(2) $(3) $$(extralibs))
+endef
+# make-3.80 has a bug with expanding large input strings to the eval function,
+# which was triggered in some cases by the following component of
+# linker_template:
+# $(1): $$(call find_file, $(patsubst -l%,lib%.a,$(filter -l%,$(2))),\
+# $$(patsubst -L%,%,$$(filter -L%,$$(LDFLAGS) $(2))))
+# This may be useful to revisit in the future (it tries to locate libraries
+# in a search path and add them as prerequisites
+
+define install_map_template
+$(DIST_DIR)/$(1): $(2)
+ $(if $(quiet),@echo " [INSTALL] $$@")
+ $(qexec)mkdir -p $$(dir $$@)
+ $(qexec)cp -p $$< $$@
+endef
+
+define archive_template
+# Not using a pattern rule here because we don't want to generate empty
+# archives when they are listed as a dependency in files not responsible
+# for creating them.
+$(1):
+ $(if $(quiet),@echo " [AR] $$@")
+ $(qexec)$$(AR) $$(ARFLAGS) $$@ $$?
+endef
+
+define so_template
+# Not using a pattern rule here because we don't want to generate empty
+# archives when they are listed as a dependency in files not responsible
+# for creating them.
+#
+# This needs further abstraction for dealing with non-GNU linkers.
+$(1):
+ $(if $(quiet),@echo " [LD] $$@")
+ $(qexec)$$(LD) -shared $$(LDFLAGS) \
+ -Wl,--no-undefined -Wl,-soname,$$(SONAME) \
+ -Wl,--version-script,$$(SO_VERSION_SCRIPT) -o $$@ \
+ $$(filter %.o,$$?) $$(extralibs)
+endef
+
+define lipo_lib_template
+$(1): $(addsuffix /$(1),$(FAT_ARCHS))
+ $(if $(quiet),@echo " [LIPO] $$@")
+ $(qexec)libtool -static -o $$@ $$?
+endef
+
+define lipo_bin_template
+$(1): $(addsuffix /$(1),$(FAT_ARCHS))
+ $(if $(quiet),@echo " [LIPO] $$@")
+ $(qexec)lipo -output $$@ -create $$?
+endef
+
+
+#
+# Get current configuration
+#
+ifneq ($(target),)
+include $(SRC_PATH_BARE)/$(target:-$(TOOLCHAIN)=).mk
+endif
+ifeq ($(filter clean,$(MAKECMDGOALS)),)
+ # Older versions of make don't like -include directives with no arguments
+ ifneq ($(filter %.d,$(OBJS-yes:.o=.d)),)
+ -include $(filter %.d,$(OBJS-yes:.o=.d))
+ endif
+endif
+
+#
+# Configuration dependent rules
+#
+$(call pairmap,install_map_templates,$(INSTALL_MAPS))
+
+DOCS=$(call cond_enabled,CONFIG_INSTALL_DOCS,DOCS)
+.docs: $(DOCS)
+ @touch $@
+
+INSTALL-DOCS=$(call cond_enabled,CONFIG_INSTALL_DOCS,INSTALL-DOCS)
+ifeq ($(MAKECMDGOALS),dist)
+INSTALL-DOCS+=$(call cond_enabled,CONFIG_INSTALL_DOCS,DIST-DOCS)
+endif
+.install-docs: .docs $(addprefix $(DIST_DIR)/,$(INSTALL-DOCS))
+ @touch $@
+
+clean::
+ rm -f .docs .install-docs $(DOCS)
+
+BINS=$(call enabled,BINS)
+.bins: $(BINS)
+ @touch $@
+
+INSTALL-BINS=$(call cond_enabled,CONFIG_INSTALL_BINS,INSTALL-BINS)
+ifeq ($(MAKECMDGOALS),dist)
+INSTALL-BINS+=$(call cond_enabled,CONFIG_INSTALL_BINS,DIST-BINS)
+endif
+.install-bins: .bins $(addprefix $(DIST_DIR)/,$(INSTALL-BINS))
+ @touch $@
+
+clean::
+ rm -f .bins .install-bins $(BINS)
+
+LIBS=$(call enabled,LIBS)
+.libs: $(LIBS)
+ @touch $@
+$(foreach lib,$(filter %_g.a,$(LIBS)),$(eval $(call archive_template,$(lib))))
+$(foreach lib,$(filter %so.$(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH),$(LIBS)),$(eval $(call so_template,$(lib))))
+
+INSTALL-LIBS=$(call cond_enabled,CONFIG_INSTALL_LIBS,INSTALL-LIBS)
+ifeq ($(MAKECMDGOALS),dist)
+INSTALL-LIBS+=$(call cond_enabled,CONFIG_INSTALL_LIBS,DIST-LIBS)
+endif
+.install-libs: .libs $(addprefix $(DIST_DIR)/,$(INSTALL-LIBS))
+ @touch $@
+
+clean::
+ rm -f .libs .install-libs $(LIBS)
+
+ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
+PROJECTS=$(call enabled,PROJECTS)
+.projects: $(PROJECTS)
+ @touch $@
+
+INSTALL-PROJECTS=$(call cond_enabled,CONFIG_INSTALL_PROJECTS,INSTALL-PROJECTS)
+ifeq ($(MAKECMDGOALS),dist)
+INSTALL-PROJECTS+=$(call cond_enabled,CONFIG_INSTALL_PROJECTS,DIST-PROJECTS)
+endif
+.install-projects: .projects $(addprefix $(DIST_DIR)/,$(INSTALL-PROJECTS))
+ @touch $@
+
+clean::
+ rm -f .projects .install-projects $(PROJECTS)
+endif
+
+# If there are any source files to be distributed, then include the build
+# system too.
+ifneq ($(call enabled,DIST-SRCS),)
+ DIST-SRCS-yes += configure
+ DIST-SRCS-yes += build/make/configure.sh
+ DIST-SRCS-yes += build/make/gen_asm_deps.sh
+ DIST-SRCS-yes += build/make/Makefile
+ DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_def.sh
+ DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_proj.sh
+ DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_sln.sh
+ DIST-SRCS-$(CONFIG_MSVS) += build/x86-msvs/yasm.rules
+ DIST-SRCS-$(CONFIG_MSVS) += build/x86-msvs/obj_int_extract.bat
+ DIST-SRCS-$(CONFIG_RVCT) += build/make/armlink_adapter.sh
+ # Include obj_int_extract if we use offsets from asm_*_offsets
+ DIST-SRCS-$(ARCH_ARM)$(ARCH_X86)$(ARCH_X86_64) += build/make/obj_int_extract.c
+ DIST-SRCS-$(ARCH_ARM) += build/make/ads2gas.pl
+ DIST-SRCS-yes += $(target:-$(TOOLCHAIN)=).mk
+endif
+INSTALL-SRCS := $(call cond_enabled,CONFIG_INSTALL_SRCS,INSTALL-SRCS)
+ifeq ($(MAKECMDGOALS),dist)
+INSTALL-SRCS += $(call cond_enabled,CONFIG_INSTALL_SRCS,DIST-SRCS)
+endif
+.install-srcs: $(addprefix $(DIST_DIR)/src/,$(INSTALL-SRCS))
+ @touch $@
+
+clean::
+ rm -f .install-srcs
+
+ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
+ BUILD_TARGETS += .projects
+ INSTALL_TARGETS += .install-projects
+endif
+BUILD_TARGETS += .docs .libs .bins
+INSTALL_TARGETS += .install-docs .install-srcs .install-libs .install-bins
+all: $(BUILD_TARGETS)
+install:: $(INSTALL_TARGETS)
+dist: $(INSTALL_TARGETS)
+test::
diff --git a/libvpx/build/make/ads2gas.pl b/libvpx/build/make/ads2gas.pl
new file mode 100755
index 0000000..95be467
--- /dev/null
+++ b/libvpx/build/make/ads2gas.pl
@@ -0,0 +1,212 @@
+#!/usr/bin/perl
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+# ads2gas.pl
+# Author: Eric Fung (efung (at) acm.org)
+#
+# Convert ARM Developer Suite 1.0.1 syntax assembly source to GNU as format
+#
+# Usage: cat inputfile | perl ads2gas.pl > outputfile
+#
+print "@ This file was created from a .asm file\n";
+print "@ using the ads2gas.pl script.\n";
+print "\t.equ DO1STROUNDING, 0\n";
+
+# Stack of procedure names.
+@proc_stack = ();
+
+while (<STDIN>)
+{
+ undef $comment;
+ undef $line;
+ $comment_char = ";";
+ $comment_sub = "@";
+
+ # Handle comments.
+ if (/$comment_char/)
+ {
+ $comment = "";
+ ($line, $comment) = /(.*?)$comment_char(.*)/;
+ $_ = $line;
+ }
+
+ # Load and store alignment
+ s/@/,:/g;
+
+ # Hexadecimal constants prefaced by 0x
+ s/#&/#0x/g;
+
+ # Convert :OR: to |
+ s/:OR:/ | /g;
+
+ # Convert :AND: to &
+ s/:AND:/ & /g;
+
+ # Convert :NOT: to ~
+ s/:NOT:/ ~ /g;
+
+ # Convert :SHL: to <<
+ s/:SHL:/ << /g;
+
+ # Convert :SHR: to >>
+ s/:SHR:/ >> /g;
+
+ # Convert ELSE to .else
+ s/\bELSE\b/.else/g;
+
+ # Convert ENDIF to .endif
+ s/\bENDIF\b/.endif/g;
+
+ # Convert ELSEIF to .elseif
+ s/\bELSEIF\b/.elseif/g;
+
+ # Convert LTORG to .ltorg
+ s/\bLTORG\b/.ltorg/g;
+
+ # Convert endfunc to nothing.
+ s/\bendfunc\b//ig;
+
+ # Convert FUNCTION to nothing.
+ s/\bFUNCTION\b//g;
+ s/\bfunction\b//g;
+
+ s/\bENTRY\b//g;
+ s/\bMSARMASM\b/0/g;
+ s/^\s+end\s+$//g;
+
+ # Convert IF :DEF:to .if
+ # gcc doesn't have the ability to do a conditional
+ # if defined variable that is set by IF :DEF: on
+ # armasm, so convert it to a normal .if and then
+ # make sure to define a value elesewhere
+ if (s/\bIF :DEF:\b/.if /g)
+ {
+ s/=/==/g;
+ }
+
+ # Convert IF to .if
+ if (s/\bIF\b/.if/g)
+ {
+ s/=+/==/g;
+ }
+
+ # Convert INCLUDE to .INCLUDE "file"
+ s/INCLUDE(\s*)(.*)$/.include $1\"$2\"/;
+
+ # Code directive (ARM vs Thumb)
+ s/CODE([0-9][0-9])/.code $1/;
+
+ # No AREA required
+ # But ALIGNs in AREA must be obeyed
+ s/^\s*AREA.*ALIGN=([0-9])$/.text\n.p2align $1/;
+ # If no ALIGN, strip the AREA and align to 4 bytes
+ s/^\s*AREA.*$/.text\n.p2align 2/;
+
+ # DCD to .word
+ # This one is for incoming symbols
+ s/DCD\s+\|(\w*)\|/.long $1/;
+
+ # DCW to .short
+ s/DCW\s+\|(\w*)\|/.short $1/;
+ s/DCW(.*)/.short $1/;
+
+ # Constants defined in scope
+ s/DCD(.*)/.long $1/;
+ s/DCB(.*)/.byte $1/;
+
+ # RN to .req
+ if (s/RN\s+([Rr]\d+|lr)/.req $1/)
+ {
+ print;
+ print "$comment_sub$comment\n" if defined $comment;
+ next;
+ }
+
+ # Make function visible to linker, and make additional symbol with
+ # prepended underscore
+ s/EXPORT\s+\|([\$\w]*)\|/.global $1 \n\t.type $1, function/;
+ s/IMPORT\s+\|([\$\w]*)\|/.global $1/;
+
+ s/EXPORT\s+([\$\w]*)/.global $1/;
+ s/export\s+([\$\w]*)/.global $1/;
+
+ # No vertical bars required; make additional symbol with prepended
+ # underscore
+ s/^\|(\$?\w+)\|/_$1\n\t$1:/g;
+
+ # Labels need trailing colon
+# s/^(\w+)/$1:/ if !/EQU/;
+ # put the colon at the end of the line in the macro
+ s/^([a-zA-Z_0-9\$]+)/$1:/ if !/EQU/;
+
+ # ALIGN directive
+ s/\bALIGN\b/.balign/g;
+
+ # ARM code
+ s/\sARM/.arm/g;
+
+ # push/pop
+ s/(push\s+)(r\d+)/stmdb sp\!, \{$2\}/g;
+ s/(pop\s+)(r\d+)/ldmia sp\!, \{$2\}/g;
+
+ # NEON code
+ s/(vld1.\d+\s+)(q\d+)/$1\{$2\}/g;
+ s/(vtbl.\d+\s+[^,]+),([^,]+)/$1,\{$2\}/g;
+
+ # eabi_attributes numerical equivalents can be found in the
+ # "ARM IHI 0045C" document.
+
+ # REQUIRE8 Stack is required to be 8-byte aligned
+ s/\sREQUIRE8/.eabi_attribute 24, 1 \@Tag_ABI_align_needed/g;
+
+ # PRESERVE8 Stack 8-byte align is preserved
+ s/\sPRESERVE8/.eabi_attribute 25, 1 \@Tag_ABI_align_preserved/g;
+
+ # Use PROC and ENDP to give the symbols a .size directive.
+ # This makes them show up properly in debugging tools like gdb and valgrind.
+ if (/\bPROC\b/)
+ {
+ my $proc;
+ /^_([\.0-9A-Z_a-z]\w+)\b/;
+ $proc = $1;
+ push(@proc_stack, $proc) if ($proc);
+ s/\bPROC\b/@ $&/;
+ }
+ if (/\bENDP\b/)
+ {
+ my $proc;
+ s/\bENDP\b/@ $&/;
+ $proc = pop(@proc_stack);
+ $_ = "\t.size $proc, .-$proc".$_ if ($proc);
+ }
+
+ # EQU directive
+ s/(\S+\s+)EQU(\s+\S+)/.equ $1, $2/;
+
+ # Begin macro definition
+ if (/\bMACRO\b/) {
+ $_ = <STDIN>;
+ s/^/.macro/;
+ s/\$//g; # remove formal param reference
+ s/;/@/g; # change comment characters
+ }
+
+ # For macros, use \ to reference formal params
+ s/\$/\\/g; # End macro definition
+ s/\bMEND\b/.endm/; # No need to tell it where to stop assembling
+ next if /^\s*END\s*$/;
+ print;
+ print "$comment_sub$comment\n" if defined $comment;
+}
+
+# Mark that this object doesn't need an executable stack.
+printf ("\t.section\t.note.GNU-stack,\"\",\%\%progbits\n");
diff --git a/libvpx/build/make/ads2gas_apple.pl b/libvpx/build/make/ads2gas_apple.pl
new file mode 100755
index 0000000..81280bf
--- /dev/null
+++ b/libvpx/build/make/ads2gas_apple.pl
@@ -0,0 +1,214 @@
+#!/usr/bin/env perl
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+# ads2gas.pl
+# Author: Eric Fung (efung (at) acm.org)
+#
+# Convert ARM Developer Suite 1.0.1 syntax assembly source to GNU as format
+#
+# Usage: cat inputfile | perl ads2gas.pl > outputfile
+#
+print "@ This file was created from a .asm file\n";
+print "@ using the ads2gas_apple.pl script.\n\n";
+print "\t.set WIDE_REFERENCE, 0\n";
+print "\t.set ARCHITECTURE, 5\n";
+print "\t.set DO1STROUNDING, 0\n";
+
+my %register_aliases;
+my %macro_aliases;
+
+my @mapping_list = ("\$0", "\$1", "\$2", "\$3", "\$4", "\$5", "\$6", "\$7", "\$8", "\$9");
+
+my @incoming_array;
+
+my @imported_functions;
+
+# Perl trim function to remove whitespace from the start and end of the string
+sub trim($)
+{
+ my $string = shift;
+ $string =~ s/^\s+//;
+ $string =~ s/\s+$//;
+ return $string;
+}
+
+while (<STDIN>)
+{
+ # Load and store alignment
+ s/@/,:/g;
+
+ # Comment character
+ s/;/@/g;
+
+ # Hexadecimal constants prefaced by 0x
+ s/#&/#0x/g;
+
+ # Convert :OR: to |
+ s/:OR:/ | /g;
+
+ # Convert :AND: to &
+ s/:AND:/ & /g;
+
+ # Convert :NOT: to ~
+ s/:NOT:/ ~ /g;
+
+ # Convert :SHL: to <<
+ s/:SHL:/ << /g;
+
+ # Convert :SHR: to >>
+ s/:SHR:/ >> /g;
+
+ # Convert ELSE to .else
+ s/ELSE/.else/g;
+
+ # Convert ENDIF to .endif
+ s/ENDIF/.endif/g;
+
+ # Convert ELSEIF to .elseif
+ s/ELSEIF/.elseif/g;
+
+ # Convert LTORG to .ltorg
+ s/LTORG/.ltorg/g;
+
+ # Convert IF :DEF:to .if
+ # gcc doesn't have the ability to do a conditional
+ # if defined variable that is set by IF :DEF: on
+ # armasm, so convert it to a normal .if and then
+ # make sure to define a value elesewhere
+ if (s/\bIF :DEF:\b/.if /g)
+ {
+ s/=/==/g;
+ }
+
+ # Convert IF to .if
+ if (s/\bIF\b/.if/g)
+ {
+ s/=/==/g;
+ }
+
+ # Convert INCLUDE to .INCLUDE "file"
+ s/INCLUDE(\s*)(.*)$/.include $1\"$2\"/;
+
+ # Code directive (ARM vs Thumb)
+ s/CODE([0-9][0-9])/.code $1/;
+
+ # No AREA required
+ # But ALIGNs in AREA must be obeyed
+ s/^\s*AREA.*ALIGN=([0-9])$/.text\n.p2align $1/;
+ # If no ALIGN, strip the AREA and align to 4 bytes
+ s/^\s*AREA.*$/.text\n.p2align 2/;
+
+ # DCD to .word
+ # This one is for incoming symbols
+ s/DCD\s+\|(\w*)\|/.long $1/;
+
+ # DCW to .short
+ s/DCW\s+\|(\w*)\|/.short $1/;
+ s/DCW(.*)/.short $1/;
+
+ # Constants defined in scope
+ s/DCD(.*)/.long $1/;
+ s/DCB(.*)/.byte $1/;
+
+ # Build a hash of all the register - alias pairs.
+ if (s/(.*)RN(.*)/$1 .req $2/g)
+ {
+ $register_aliases{trim($1)} = trim($2);
+ next;
+ }
+
+ while (($key, $value) = each(%register_aliases))
+ {
+ s/\b$key\b/$value/g;
+ }
+
+ # Make function visible to linker, and make additional symbol with
+ # prepended underscore
+ s/EXPORT\s+\|([\$\w]*)\|/.globl _$1\n\t.globl $1/;
+
+ # Prepend imported functions with _
+ if (s/IMPORT\s+\|([\$\w]*)\|/.globl $1/)
+ {
+ $function = trim($1);
+ push(@imported_functions, $function);
+ }
+
+ foreach $function (@imported_functions)
+ {
+ s/$function/_$function/;
+ }
+
+ # No vertical bars required; make additional symbol with prepended
+ # underscore
+ s/^\|(\$?\w+)\|/_$1\n\t$1:/g;
+
+ # Labels need trailing colon
+# s/^(\w+)/$1:/ if !/EQU/;
+ # put the colon at the end of the line in the macro
+ s/^([a-zA-Z_0-9\$]+)/$1:/ if !/EQU/;
+
+ # ALIGN directive
+ s/ALIGN/.balign/g;
+
+ # Strip ARM
+ s/\sARM/@ ARM/g;
+
+ # Strip REQUIRE8
+ #s/\sREQUIRE8/@ REQUIRE8/g;
+ s/\sREQUIRE8/@ /g;
+
+ # Strip PRESERVE8
+ s/\sPRESERVE8/@ PRESERVE8/g;
+
+ # Strip PROC and ENDPROC
+ s/\bPROC\b/@/g;
+ s/\bENDP\b/@/g;
+
+ # EQU directive
+ s/(.*)EQU(.*)/.set $1, $2/;
+
+ # Begin macro definition
+ if (/MACRO/)
+ {
+ # Process next line down, which will be the macro definition
+ $_ = <STDIN>;
+
+ $trimmed = trim($_);
+
+ # remove commas that are separating list
+ $trimmed =~ s/,//g;
+
+ # string to array
+ @incoming_array = split(/ /, $trimmed);
+
+ print ".macro @incoming_array[0]\n";
+
+ # remove the first element, as that is the name of the macro
+ shift (@incoming_array);
+
+ @macro_aliases{@incoming_array} = @mapping_list;
+
+ next;
+ }
+
+ while (($key, $value) = each(%macro_aliases))
+ {
+ $key =~ s/\$/\\\$/;
+ s/$key\b/$value/g;
+ }
+
+ # For macros, use \ to reference formal params
+# s/\$/\\/g; # End macro definition
+ s/MEND/.endm/; # No need to tell it where to stop assembling
+ next if /^\s*END\s*$/;
+ print;
+}
diff --git a/build/make/armlink_adapter.sh b/libvpx/build/make/armlink_adapter.sh
similarity index 100%
rename from build/make/armlink_adapter.sh
rename to libvpx/build/make/armlink_adapter.sh
diff --git a/libvpx/build/make/configure.sh b/libvpx/build/make/configure.sh
new file mode 100755
index 0000000..05bbabe
--- /dev/null
+++ b/libvpx/build/make/configure.sh
@@ -0,0 +1,1276 @@
+#!/bin/bash
+##
+## configure.sh
+##
+## This script is sourced by the main configure script and contains
+## utility functions and other common bits that aren't strictly libvpx
+## related.
+##
+## This build system is based in part on the FFmpeg configure script.
+##
+
+
+#
+# Logging / Output Functions
+#
+die_unknown(){
+ echo "Unknown option \"$1\"."
+ echo "See $0 --help for available options."
+ clean_temp_files
+ exit 1
+}
+
+
+die() {
+ echo "$@"
+ echo
+ echo "Configuration failed. This could reflect a misconfiguration of your"
+ echo "toolchains, improper options selected, or another problem. If you"
+ echo "don't see any useful error messages above, the next step is to look"
+ echo "at the configure error log file ($logfile) to determine what"
+ echo "configure was trying to do when it died."
+ clean_temp_files
+ exit 1
+}
+
+
+log(){
+ echo "$@" >>$logfile
+}
+
+
+log_file(){
+ log BEGIN $1
+ pr -n -t $1 >>$logfile
+ log END $1
+}
+
+
+log_echo() {
+ echo "$@"
+ log "$@"
+}
+
+
+fwrite () {
+ outfile=$1
+ shift
+ echo "$@" >> ${outfile}
+}
+
+
+show_help_pre(){
+ for opt in ${CMDLINE_SELECT}; do
+ opt2=`echo $opt | sed -e 's;_;-;g'`
+ if enabled $opt; then
+ eval "toggle_${opt}=\"--disable-${opt2}\""
+ else
+ eval "toggle_${opt}=\"--enable-${opt2} \""
+ fi
+ done
+
+ cat <<EOF
+Usage: configure [options]
+Options:
+
+Build options:
+ --help print this message
+ --log=yes|no|FILE file configure log is written to [config.err]
+ --target=TARGET target platform tuple [generic-gnu]
+ --cpu=CPU optimize for a specific cpu rather than a family
+ --extra-cflags=ECFLAGS add ECFLAGS to CFLAGS [$CFLAGS]
+ ${toggle_extra_warnings} emit harmless warnings (always non-fatal)
+ ${toggle_werror} treat warnings as errors, if possible
+ (not available with all compilers)
+ ${toggle_optimizations} turn on/off compiler optimization flags
+ ${toggle_pic} turn on/off Position Independent Code
+ ${toggle_ccache} turn on/off compiler cache
+ ${toggle_debug} enable/disable debug mode
+ ${toggle_gprof} enable/disable gprof profiling instrumentation
+ ${toggle_gcov} enable/disable gcov coverage instrumentation
+
+Install options:
+ ${toggle_install_docs} control whether docs are installed
+ ${toggle_install_bins} control whether binaries are installed
+ ${toggle_install_libs} control whether libraries are installed
+ ${toggle_install_srcs} control whether sources are installed
+
+
+EOF
+}
+
+
+show_help_post(){
+ cat <<EOF
+
+
+NOTES:
+ Object files are built at the place where configure is launched.
+
+ All boolean options can be negated. The default value is the opposite
+ of that shown above. If the option --disable-foo is listed, then
+ the default value for foo is enabled.
+
+Supported targets:
+EOF
+ show_targets ${all_platforms}
+ echo
+ exit 1
+}
+
+
+show_targets() {
+ while [ -n "$*" ]; do
+ if [ "${1%%-*}" = "${2%%-*}" ]; then
+ if [ "${2%%-*}" = "${3%%-*}" ]; then
+ printf " %-24s %-24s %-24s\n" "$1" "$2" "$3"
+ shift; shift; shift
+ else
+ printf " %-24s %-24s\n" "$1" "$2"
+ shift; shift
+ fi
+ else
+ printf " %-24s\n" "$1"
+ shift
+ fi
+ done
+}
+
+
+show_help() {
+ show_help_pre
+ show_help_post
+}
+
+#
+# List Processing Functions
+#
+set_all(){
+ value=$1
+ shift
+ for var in $*; do
+ eval $var=$value
+ done
+}
+
+
+is_in(){
+ value=$1
+ shift
+ for var in $*; do
+ [ $var = $value ] && return 0
+ done
+ return 1
+}
+
+
+add_cflags() {
+ CFLAGS="${CFLAGS} $@"
+ CXXFLAGS="${CXXFLAGS} $@"
+}
+
+
+add_cflags_only() {
+ CFLAGS="${CFLAGS} $@"
+}
+
+
+add_cxxflags_only() {
+ CXXFLAGS="${CXXFLAGS} $@"
+}
+
+
+add_ldflags() {
+ LDFLAGS="${LDFLAGS} $@"
+}
+
+
+add_asflags() {
+ ASFLAGS="${ASFLAGS} $@"
+}
+
+
+add_extralibs() {
+ extralibs="${extralibs} $@"
+}
+
+#
+# Boolean Manipulation Functions
+#
+enable(){
+ set_all yes $*
+}
+
+disable(){
+ set_all no $*
+}
+
+enabled(){
+ eval test "x\$$1" = "xyes"
+}
+
+disabled(){
+ eval test "x\$$1" = "xno"
+}
+
+
+soft_enable() {
+ for var in $*; do
+ if ! disabled $var; then
+ log_echo " enabling $var"
+ enable $var
+ fi
+ done
+}
+
+soft_disable() {
+ for var in $*; do
+ if ! enabled $var; then
+ log_echo " disabling $var"
+ disable $var
+ fi
+ done
+}
+
+
+#
+# Text Processing Functions
+#
+toupper(){
+ echo "$@" | tr abcdefghijklmnopqrstuvwxyz ABCDEFGHIJKLMNOPQRSTUVWXYZ
+}
+
+
+tolower(){
+ echo "$@" | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz
+}
+
+
+#
+# Temporary File Functions
+#
+source_path=${0%/*}
+enable source_path_used
+if test -z "$source_path" -o "$source_path" = "." ; then
+ source_path="`pwd`"
+ disable source_path_used
+fi
+
+if test ! -z "$TMPDIR" ; then
+ TMPDIRx="${TMPDIR}"
+elif test ! -z "$TEMPDIR" ; then
+ TMPDIRx="${TEMPDIR}"
+else
+ TMPDIRx="/tmp"
+fi
+TMP_H="${TMPDIRx}/vpx-conf-$$-${RANDOM}.h"
+TMP_C="${TMPDIRx}/vpx-conf-$$-${RANDOM}.c"
+TMP_O="${TMPDIRx}/vpx-conf-$$-${RANDOM}.o"
+TMP_X="${TMPDIRx}/vpx-conf-$$-${RANDOM}.x"
+TMP_ASM="${TMPDIRx}/vpx-conf-$$-${RANDOM}.asm"
+
+clean_temp_files() {
+ rm -f ${TMP_C} ${TMP_H} ${TMP_O} ${TMP_X} ${TMP_ASM}
+}
+
+#
+# Toolchain Check Functions
+#
+check_cmd() {
+ log "$@"
+ "$@" >>${logfile} 2>&1
+}
+
+check_cc() {
+ log check_cc "$@"
+ cat >${TMP_C}
+ log_file ${TMP_C}
+ check_cmd ${CC} ${CFLAGS} "$@" -c -o ${TMP_O} ${TMP_C}
+}
+
+check_cxx() {
+ log check_cxx "$@"
+ cat >${TMP_C}
+ log_file ${TMP_C}
+ check_cmd ${CXX} ${CXXFLAGS} "$@" -c -o ${TMP_O} ${TMP_C}
+}
+
+check_cpp() {
+ log check_cpp "$@"
+ cat > ${TMP_C}
+ log_file ${TMP_C}
+ check_cmd ${CC} ${CFLAGS} "$@" -E -o ${TMP_O} ${TMP_C}
+}
+
+check_ld() {
+ log check_ld "$@"
+ check_cc $@ \
+ && check_cmd ${LD} ${LDFLAGS} "$@" -o ${TMP_X} ${TMP_O} ${extralibs}
+}
+
+check_header(){
+ log check_header "$@"
+ header=$1
+ shift
+ var=`echo $header | sed 's/[^A-Za-z0-9_]/_/g'`
+ disable $var
+ check_cpp "$@" <<EOF && enable $var
+#include "$header"
+int x;
+EOF
+}
+
+
+check_cflags() {
+ log check_cflags "$@"
+ check_cc "$@" <<EOF
+int x;
+EOF
+}
+
+check_cxxflags() {
+ log check_cxxflags "$@"
+
+ # Catch CFLAGS that trigger CXX warnings
+ case "$CXX" in
+ *g++*) check_cxx -Werror "$@" <<EOF
+int x;
+EOF
+ ;;
+ *) check_cxx "$@" <<EOF
+int x;
+EOF
+ ;;
+ esac
+}
+
+check_add_cflags() {
+ check_cxxflags "$@" && add_cxxflags_only "$@"
+ check_cflags "$@" && add_cflags_only "$@"
+}
+
+check_add_asflags() {
+ log add_asflags "$@"
+ add_asflags "$@"
+}
+
+check_add_ldflags() {
+ log add_ldflags "$@"
+ add_ldflags "$@"
+}
+
+check_asm_align() {
+ log check_asm_align "$@"
+ cat >${TMP_ASM} <<EOF
+section .rodata
+align 16
+EOF
+ log_file ${TMP_ASM}
+ check_cmd ${AS} ${ASFLAGS} -o ${TMP_O} ${TMP_ASM}
+ readelf -WS ${TMP_O} >${TMP_X}
+ log_file ${TMP_X}
+ if ! grep -q '\.rodata .* 16$' ${TMP_X}; then
+ die "${AS} ${ASFLAGS} does not support section alignment (nasm <=2.08?)"
+ fi
+}
+
+write_common_config_banner() {
+ print_webm_license config.mk "##" ""
+ echo '# This file automatically generated by configure. Do not edit!' >> config.mk
+ echo "TOOLCHAIN := ${toolchain}" >> config.mk
+
+ case ${toolchain} in
+ *-linux-rvct)
+ echo "ALT_LIBC := ${alt_libc}" >> config.mk
+ ;;
+ esac
+}
+
+write_common_config_targets() {
+ for t in ${all_targets}; do
+ if enabled ${t}; then
+ if enabled universal || enabled child; then
+ fwrite config.mk "ALL_TARGETS += ${t}-${toolchain}"
+ else
+ fwrite config.mk "ALL_TARGETS += ${t}"
+ fi
+ fi
+ true;
+ done
+true
+}
+
+write_common_target_config_mk() {
+ local CC=${CC}
+ local CXX=${CXX}
+ enabled ccache && CC="ccache ${CC}"
+ enabled ccache && CXX="ccache ${CXX}"
+ print_webm_license $1 "##" ""
+
+ cat >> $1 << EOF
+# This file automatically generated by configure. Do not edit!
+SRC_PATH="$source_path"
+SRC_PATH_BARE=$source_path
+BUILD_PFX=${BUILD_PFX}
+TOOLCHAIN=${toolchain}
+ASM_CONVERSION=${asm_conversion_cmd:-${source_path}/build/make/ads2gas.pl}
+
+CC=${CC}
+CXX=${CXX}
+AR=${AR}
+LD=${LD}
+AS=${AS}
+STRIP=${STRIP}
+NM=${NM}
+
+CFLAGS = ${CFLAGS}
+CXXFLAGS = ${CXXFLAGS}
+ARFLAGS = -rus\$(if \$(quiet),c,v)
+LDFLAGS = ${LDFLAGS}
+ASFLAGS = ${ASFLAGS}
+extralibs = ${extralibs}
+AS_SFX = ${AS_SFX:-.asm}
+EXE_SFX = ${EXE_SFX}
+RTCD_OPTIONS = ${RTCD_OPTIONS}
+EOF
+
+ if enabled rvct; then cat >> $1 << EOF
+fmt_deps = sed -e 's;^__image.axf;\$(dir \$@)\$(notdir \$<).o \$@;' #hide
+EOF
+ else cat >> $1 << EOF
+fmt_deps = sed -e 's;^\([a-zA-Z0-9_]*\)\.o;\$(dir \$@)\1\$(suffix \$<).o \$@;'
+EOF
+ fi
+
+ print_config_mk ARCH "${1}" ${ARCH_LIST}
+ print_config_mk HAVE "${1}" ${HAVE_LIST}
+ print_config_mk CONFIG "${1}" ${CONFIG_LIST}
+ print_config_mk HAVE "${1}" gnu_strip
+
+ enabled msvs && echo "CONFIG_VS_VERSION=${vs_version}" >> "${1}"
+
+}
+
+
+write_common_target_config_h() {
+ print_webm_license ${TMP_H} "/*" " */"
+ cat >> ${TMP_H} << EOF
+/* This file automatically generated by configure. Do not edit! */
+#ifndef VPX_CONFIG_H
+#define VPX_CONFIG_H
+#define RESTRICT ${RESTRICT}
+EOF
+ print_config_h ARCH "${TMP_H}" ${ARCH_LIST}
+ print_config_h HAVE "${TMP_H}" ${HAVE_LIST}
+ print_config_h CONFIG "${TMP_H}" ${CONFIG_LIST}
+ echo "#endif /* VPX_CONFIG_H */" >> ${TMP_H}
+ mkdir -p `dirname "$1"`
+ cmp "$1" ${TMP_H} >/dev/null 2>&1 || mv ${TMP_H} "$1"
+}
+
+process_common_cmdline() {
+ for opt in "$@"; do
+ optval="${opt#*=}"
+ case "$opt" in
+ --child) enable child
+ ;;
+ --log*)
+ logging="$optval"
+ if ! disabled logging ; then
+ enabled logging || logfile="$logging"
+ else
+ logfile=/dev/null
+ fi
+ ;;
+ --target=*) toolchain="${toolchain:-${optval}}"
+ ;;
+ --force-target=*) toolchain="${toolchain:-${optval}}"; enable force_toolchain
+ ;;
+ --cpu)
+ ;;
+ --cpu=*) tune_cpu="$optval"
+ ;;
+ --extra-cflags=*)
+ extra_cflags="${optval}"
+ ;;
+ --enable-?*|--disable-?*)
+ eval `echo "$opt" | sed 's/--/action=/;s/-/ option=/;s/-/_/g'`
+ if echo "${ARCH_EXT_LIST}" | grep "^ *$option\$" >/dev/null; then
+ [ $action = "disable" ] && RTCD_OPTIONS="${RTCD_OPTIONS}${opt} "
+ elif [ $action = "disable" ] && ! disabled $option ; then
+ echo "${CMDLINE_SELECT}" | grep "^ *$option\$" >/dev/null ||
+ die_unknown $opt
+ elif [ $action = "enable" ] && ! enabled $option ; then
+ echo "${CMDLINE_SELECT}" | grep "^ *$option\$" >/dev/null ||
+ die_unknown $opt
+ fi
+ $action $option
+ ;;
+ --require-?*)
+ eval `echo "$opt" | sed 's/--/action=/;s/-/ option=/;s/-/_/g'`
+ if echo "${ARCH_EXT_LIST}" none | grep "^ *$option\$" >/dev/null; then
+ RTCD_OPTIONS="${RTCD_OPTIONS}${opt} "
+ else
+ die_unknown $opt
+ fi
+ ;;
+ --force-enable-?*|--force-disable-?*)
+ eval `echo "$opt" | sed 's/--force-/action=/;s/-/ option=/;s/-/_/g'`
+ $action $option
+ ;;
+ --libc=*)
+ [ -d "${optval}" ] || die "Not a directory: ${optval}"
+ disable builtin_libc
+ alt_libc="${optval}"
+ ;;
+ --as=*)
+ [ "${optval}" = yasm -o "${optval}" = nasm -o "${optval}" = auto ] \
+ || die "Must be yasm, nasm or auto: ${optval}"
+ alt_as="${optval}"
+ ;;
+ --prefix=*)
+ prefix="${optval}"
+ ;;
+ --libdir=*)
+ libdir="${optval}"
+ ;;
+ --sdk-path=*)
+ [ -d "${optval}" ] || die "Not a directory: ${optval}"
+ sdk_path="${optval}"
+ ;;
+ --libc|--as|--prefix|--libdir|--sdk-path)
+ die "Option ${opt} requires argument"
+ ;;
+ --help|-h) show_help
+ ;;
+ *) die_unknown $opt
+ ;;
+ esac
+ done
+}
+
+process_cmdline() {
+ for opt do
+ optval="${opt#*=}"
+ case "$opt" in
+ *) process_common_cmdline $opt
+ ;;
+ esac
+ done
+}
+
+
+post_process_common_cmdline() {
+ prefix="${prefix:-/usr/local}"
+ prefix="${prefix%/}"
+ libdir="${libdir:-${prefix}/lib}"
+ libdir="${libdir%/}"
+ if [ "${libdir#${prefix}}" = "${libdir}" ]; then
+ die "Libdir ${libdir} must be a subdirectory of ${prefix}"
+ fi
+}
+
+
+post_process_cmdline() {
+ true;
+}
+
+setup_gnu_toolchain() {
+ CC=${CC:-${CROSS}gcc}
+ CXX=${CXX:-${CROSS}g++}
+ AR=${AR:-${CROSS}ar}
+ LD=${LD:-${CROSS}${link_with_cc:-ld}}
+ AS=${AS:-${CROSS}as}
+ STRIP=${STRIP:-${CROSS}strip}
+ NM=${NM:-${CROSS}nm}
+ AS_SFX=.s
+ EXE_SFX=
+}
+
+process_common_toolchain() {
+ if [ -z "$toolchain" ]; then
+ gcctarget="${CHOST:-$(gcc -dumpmachine 2> /dev/null)}"
+
+ # detect tgt_isa
+ case "$gcctarget" in
+ armv6*)
+ tgt_isa=armv6
+ ;;
+ armv7*)
+ tgt_isa=armv7
+ ;;
+ armv5te*)
+ tgt_isa=armv5te
+ ;;
+ *x86_64*|*amd64*)
+ tgt_isa=x86_64
+ ;;
+ *i[3456]86*)
+ tgt_isa=x86
+ ;;
+ *powerpc64*)
+ tgt_isa=ppc64
+ ;;
+ *powerpc*)
+ tgt_isa=ppc32
+ ;;
+ *sparc*)
+ tgt_isa=sparc
+ ;;
+ esac
+
+ # detect tgt_os
+ case "$gcctarget" in
+ *darwin8*)
+ tgt_isa=universal
+ tgt_os=darwin8
+ ;;
+ *darwin9*)
+ tgt_isa=universal
+ tgt_os=darwin9
+ ;;
+ *darwin10*)
+ tgt_isa=x86_64
+ tgt_os=darwin10
+ ;;
+ *darwin11*)
+ tgt_isa=x86_64
+ tgt_os=darwin11
+ ;;
+ *darwin12*)
+ tgt_isa=x86_64
+ tgt_os=darwin12
+ ;;
+ *mingw32*|*cygwin*)
+ [ -z "$tgt_isa" ] && tgt_isa=x86
+ tgt_os=win32
+ ;;
+ *linux*|*bsd*)
+ tgt_os=linux
+ ;;
+ *solaris2.10)
+ tgt_os=solaris
+ ;;
+ *os2*)
+ tgt_os=os2
+ ;;
+ esac
+
+ if [ -n "$tgt_isa" ] && [ -n "$tgt_os" ]; then
+ toolchain=${tgt_isa}-${tgt_os}-gcc
+ fi
+ fi
+
+ toolchain=${toolchain:-generic-gnu}
+
+ is_in ${toolchain} ${all_platforms} || enabled force_toolchain \
+ || die "Unrecognized toolchain '${toolchain}'"
+
+ enabled child || log_echo "Configuring for target '${toolchain}'"
+
+ #
+ # Set up toolchain variables
+ #
+ tgt_isa=$(echo ${toolchain} | awk 'BEGIN{FS="-"}{print $1}')
+ tgt_os=$(echo ${toolchain} | awk 'BEGIN{FS="-"}{print $2}')
+ tgt_cc=$(echo ${toolchain} | awk 'BEGIN{FS="-"}{print $3}')
+
+ # Mark the specific ISA requested as enabled
+ soft_enable ${tgt_isa}
+ enable ${tgt_os}
+ enable ${tgt_cc}
+
+ # Enable the architecture family
+ case ${tgt_isa} in
+ arm*) enable arm;;
+ mips*) enable mips;;
+ esac
+
+ # PIC is probably what we want when building shared libs
+ enabled shared && soft_enable pic
+
+ # Handle darwin variants. Newer SDKs allow targeting older
+ # platforms, so find the newest SDK available.
+ case ${toolchain} in
+ *-darwin*)
+ if [ -z "${DEVELOPER_DIR}" ]; then
+ DEVELOPER_DIR=`xcode-select -print-path 2> /dev/null`
+ [ $? -ne 0 ] && OSX_SKIP_DIR_CHECK=1
+ fi
+ if [ -z "${OSX_SKIP_DIR_CHECK}" ]; then
+ OSX_SDK_ROOTS="${DEVELOPER_DIR}/SDKs"
+ OSX_SDK_VERSIONS="MacOSX10.4u.sdk MacOSX10.5.sdk MacOSX10.6.sdk"
+ OSX_SDK_VERSIONS="${OSX_SDK_VERSIONS} MacOSX10.7.sdk"
+ for v in ${OSX_SDK_VERSIONS}; do
+ if [ -d "${OSX_SDK_ROOTS}/${v}" ]; then
+ osx_sdk_dir="${OSX_SDK_ROOTS}/${v}"
+ fi
+ done
+ fi
+ ;;
+ esac
+
+ if [ -d "${osx_sdk_dir}" ]; then
+ add_cflags "-isysroot ${osx_sdk_dir}"
+ add_ldflags "-isysroot ${osx_sdk_dir}"
+ fi
+
+ case ${toolchain} in
+ *-darwin8-*)
+ add_cflags "-mmacosx-version-min=10.4"
+ add_ldflags "-mmacosx-version-min=10.4"
+ ;;
+ *-darwin9-*)
+ add_cflags "-mmacosx-version-min=10.5"
+ add_ldflags "-mmacosx-version-min=10.5"
+ ;;
+ *-darwin10-*)
+ add_cflags "-mmacosx-version-min=10.6"
+ add_ldflags "-mmacosx-version-min=10.6"
+ ;;
+ *-darwin11-*)
+ add_cflags "-mmacosx-version-min=10.7"
+ add_ldflags "-mmacosx-version-min=10.7"
+ ;;
+ *-darwin12-*)
+ add_cflags "-mmacosx-version-min=10.8"
+ add_ldflags "-mmacosx-version-min=10.8"
+ ;;
+ esac
+
+ # Handle Solaris variants. Solaris 10 needs -lposix4
+ case ${toolchain} in
+ sparc-solaris-*)
+ add_extralibs -lposix4
+ disable fast_unaligned
+ ;;
+ *-solaris-*)
+ add_extralibs -lposix4
+ ;;
+ esac
+
+ # Process ARM architecture variants
+ case ${toolchain} in
+ arm*)
+ # on arm, isa versions are supersets
+ case ${tgt_isa} in
+ armv7)
+ soft_enable neon
+ soft_enable media
+ soft_enable edsp
+ soft_enable fast_unaligned
+ ;;
+ armv6)
+ soft_enable media
+ soft_enable edsp
+ soft_enable fast_unaligned
+ ;;
+ armv5te)
+ soft_enable edsp
+ ;;
+ esac
+
+ asm_conversion_cmd="cat"
+
+ case ${tgt_cc} in
+ gcc)
+ CROSS=${CROSS:-arm-none-linux-gnueabi-}
+ link_with_cc=gcc
+ setup_gnu_toolchain
+ arch_int=${tgt_isa##armv}
+ arch_int=${arch_int%%te}
+ check_add_asflags --defsym ARCHITECTURE=${arch_int}
+ tune_cflags="-mtune="
+ if [ ${tgt_isa} == "armv7" ]; then
+ check_add_cflags -march=armv7-a -mfloat-abi=softfp
+ check_add_asflags -march=armv7-a -mfloat-abi=softfp
+
+ if enabled neon
+ then
+ check_add_cflags -mfpu=neon #-ftree-vectorize
+ check_add_asflags -mfpu=neon
+ fi
+
+ if [ -z "${tune_cpu}" ]; then
+ tune_cpu=cortex-a8
+ fi
+ else
+ check_add_cflags -march=${tgt_isa}
+ check_add_asflags -march=${tgt_isa}
+ fi
+
+ enabled debug && add_asflags -g
+ asm_conversion_cmd="${source_path}/build/make/ads2gas.pl"
+ ;;
+ rvct)
+ CC=armcc
+ AR=armar
+ AS=armasm
+ LD=${source_path}/build/make/armlink_adapter.sh
+ STRIP=arm-none-linux-gnueabi-strip
+ NM=arm-none-linux-gnueabi-nm
+ tune_cflags="--cpu="
+ tune_asflags="--cpu="
+ if [ -z "${tune_cpu}" ]; then
+ if [ ${tgt_isa} == "armv7" ]; then
+ if enabled neon
+ then
+ check_add_cflags --fpu=softvfp+vfpv3
+ check_add_asflags --fpu=softvfp+vfpv3
+ fi
+ check_add_cflags --cpu=Cortex-A8
+ check_add_asflags --cpu=Cortex-A8
+ else
+ check_add_cflags --cpu=${tgt_isa##armv}
+ check_add_asflags --cpu=${tgt_isa##armv}
+ fi
+ fi
+ arch_int=${tgt_isa##armv}
+ arch_int=${arch_int%%te}
+ check_add_asflags --pd "\"ARCHITECTURE SETA ${arch_int}\""
+ enabled debug && add_asflags -g
+ add_cflags --gnu
+ add_cflags --enum_is_int
+ add_cflags --wchar32
+ ;;
+ esac
+
+ case ${tgt_os} in
+ none*)
+ disable multithread
+ disable os_support
+ ;;
+
+ android*)
+ SDK_PATH=${sdk_path}
+ COMPILER_LOCATION=`find "${SDK_PATH}" \
+ -name "arm-linux-androideabi-gcc*" -print -quit`
+ TOOLCHAIN_PATH=${COMPILER_LOCATION%/*}/arm-linux-androideabi-
+ CC=${TOOLCHAIN_PATH}gcc
+ CXX=${TOOLCHAIN_PATH}g++
+ AR=${TOOLCHAIN_PATH}ar
+ LD=${TOOLCHAIN_PATH}gcc
+ AS=${TOOLCHAIN_PATH}as
+ STRIP=${TOOLCHAIN_PATH}strip
+ NM=${TOOLCHAIN_PATH}nm
+
+ if [ -z "${alt_libc}" ]; then
+ alt_libc=`find "${SDK_PATH}" -name arch-arm -print | \
+ awk '{n = split($0,a,"/"); \
+ split(a[n-1],b,"-"); \
+ print $0 " " b[2]}' | \
+ sort -g -k 2 | \
+ awk '{ print $1 }' | tail -1`
+ fi
+
+ add_cflags "--sysroot=${alt_libc}"
+ add_ldflags "--sysroot=${alt_libc}"
+
+ # linker flag that routes around a CPU bug in some
+ # Cortex-A8 implementations (NDK Dev Guide)
+ add_ldflags "-Wl,--fix-cortex-a8"
+
+ enable pic
+ soft_enable realtime_only
+ if [ ${tgt_isa} == "armv7" ]; then
+ soft_enable runtime_cpu_detect
+ fi
+ if enabled runtime_cpu_detect; then
+ add_cflags "-I${SDK_PATH}/sources/android/cpufeatures"
+ fi
+ ;;
+
+ darwin*)
+ if [ -z "${sdk_path}" ]; then
+ SDK_PATH=`xcode-select -print-path 2> /dev/null`
+ SDK_PATH=${SDK_PATH}/Platforms/iPhoneOS.platform/Developer
+ else
+ SDK_PATH=${sdk_path}
+ fi
+ TOOLCHAIN_PATH=${SDK_PATH}/usr/bin
+ CXX=${TOOLCHAIN_PATH}/g++
+ CC=${TOOLCHAIN_PATH}/gcc
+ AR=${TOOLCHAIN_PATH}/ar
+ LD=${TOOLCHAIN_PATH}/arm-apple-darwin10-llvm-gcc-4.2
+ AS=${TOOLCHAIN_PATH}/as
+ STRIP=${TOOLCHAIN_PATH}/strip
+ NM=${TOOLCHAIN_PATH}/nm
+ AS_SFX=.s
+
+ # ASFLAGS is written here instead of using check_add_asflags
+ # because we need to overwrite all of ASFLAGS and purge the
+ # options that were put in above
+ ASFLAGS="-version -arch ${tgt_isa} -g"
+
+ add_cflags -arch ${tgt_isa}
+ add_ldflags -arch_only ${tgt_isa}
+
+ if [ -z "${alt_libc}" ]; then
+ alt_libc=${SDK_PATH}/SDKs/iPhoneOS5.1.sdk
+ fi
+
+ add_cflags "-isysroot ${alt_libc}"
+
+ # Add the paths for the alternate libc
+ for d in usr/include; do
+ try_dir="${alt_libc}/${d}"
+ [ -d "${try_dir}" ] && add_cflags -I"${try_dir}"
+ done
+
+ for d in lib usr/lib usr/lib/system; do
+ try_dir="${alt_libc}/${d}"
+ [ -d "${try_dir}" ] && add_ldflags -L"${try_dir}"
+ done
+
+ asm_conversion_cmd="${source_path}/build/make/ads2gas_apple.pl"
+ ;;
+
+ linux*)
+ enable linux
+ if enabled rvct; then
+ # Check if we have CodeSourcery GCC in PATH. Needed for
+ # libraries
+ hash arm-none-linux-gnueabi-gcc 2>&- || \
+ die "Couldn't find CodeSourcery GCC from PATH"
+
+ # Use armcc as a linker to enable translation of
+ # some gcc specific options such as -lm and -lpthread.
+ LD="armcc --translate_gcc"
+
+ # create configuration file (uses path to CodeSourcery GCC)
+ armcc --arm_linux_configure --arm_linux_config_file=arm_linux.cfg
+
+ add_cflags --arm_linux_paths --arm_linux_config_file=arm_linux.cfg
+ add_asflags --no_hide_all --apcs=/interwork
+ add_ldflags --arm_linux_paths --arm_linux_config_file=arm_linux.cfg
+ enabled pic && add_cflags --apcs=/fpic
+ enabled pic && add_asflags --apcs=/fpic
+ enabled shared && add_cflags --shared
+ fi
+ ;;
+
+ esac
+ ;;
+ mips*)
+ link_with_cc=gcc
+ setup_gnu_toolchain
+ tune_cflags="-mtune="
+ if enabled dspr2; then
+ check_add_cflags -mips32r2 -mdspr2
+ disable fast_unaligned
+ fi
+ check_add_cflags -march=${tgt_isa}
+ check_add_asflags -march=${tgt_isa}
+ check_add_asflags -KPIC
+ ;;
+ ppc*)
+ enable ppc
+ bits=${tgt_isa##ppc}
+ link_with_cc=gcc
+ setup_gnu_toolchain
+ add_asflags -force_cpusubtype_ALL -I"\$(dir \$<)darwin"
+ soft_enable altivec
+ enabled altivec && add_cflags -maltivec
+
+ case "$tgt_os" in
+ linux*)
+ add_asflags -maltivec -mregnames -I"\$(dir \$<)linux"
+ ;;
+ darwin*)
+ darwin_arch="-arch ppc"
+ enabled ppc64 && darwin_arch="${darwin_arch}64"
+ add_cflags ${darwin_arch} -m${bits} -fasm-blocks
+ add_asflags ${darwin_arch} -force_cpusubtype_ALL -I"\$(dir \$<)darwin"
+ add_ldflags ${darwin_arch} -m${bits}
+ enabled altivec && add_cflags -faltivec
+ ;;
+ esac
+ ;;
+ x86*)
+ bits=32
+ enabled x86_64 && bits=64
+ check_cpp <<EOF && bits=x32
+#ifndef __ILP32__
+#error "not x32"
+#endif
+EOF
+ soft_enable runtime_cpu_detect
+ soft_enable mmx
+ soft_enable sse
+ soft_enable sse2
+ soft_enable sse3
+ soft_enable ssse3
+ soft_enable sse4_1
+
+ case ${tgt_os} in
+ win*)
+ enabled gcc && add_cflags -fno-common
+ ;;
+ solaris*)
+ CC=${CC:-${CROSS}gcc}
+ CXX=${CXX:-${CROSS}g++}
+ LD=${LD:-${CROSS}gcc}
+ CROSS=${CROSS:-g}
+ ;;
+ os2)
+ AS=${AS:-nasm}
+ ;;
+ esac
+
+ AS="${alt_as:-${AS:-auto}}"
+ case ${tgt_cc} in
+ icc*)
+ CC=${CC:-icc}
+ LD=${LD:-icc}
+ setup_gnu_toolchain
+ add_cflags -use-msasm -use-asm
+ add_ldflags -i-static
+ enabled x86_64 && add_cflags -ipo -no-prec-div -static -xSSE2 -axSSE2
+ enabled x86_64 && AR=xiar
+ case ${tune_cpu} in
+ atom*)
+ tune_cflags="-x"
+ tune_cpu="SSE3_ATOM"
+ ;;
+ *)
+ tune_cflags="-march="
+ ;;
+ esac
+ ;;
+ gcc*)
+ add_cflags -m${bits}
+ add_ldflags -m${bits}
+ link_with_cc=gcc
+ tune_cflags="-march="
+ setup_gnu_toolchain
+ #for 32 bit x86 builds, -O3 did not turn on this flag
+ enabled optimizations && check_add_cflags -fomit-frame-pointer
+ ;;
+ vs*)
+ # When building with Microsoft Visual Studio the assembler is
+ # invoked directly. Checking at configure time is unnecessary.
+ # Skip the check by setting AS arbitrarily
+ AS=msvs
+ ;;
+ esac
+
+ case "${AS}" in
+ auto|"")
+ which nasm >/dev/null 2>&1 && AS=nasm
+ which yasm >/dev/null 2>&1 && AS=yasm
+ [ "${AS}" = auto -o -z "${AS}" ] \
+ && die "Neither yasm nor nasm have been found"
+ ;;
+ esac
+ log_echo " using $AS"
+ [ "${AS##*/}" = nasm ] && add_asflags -Ox
+ AS_SFX=.asm
+ case ${tgt_os} in
+ win32)
+ add_asflags -f win32
+ enabled debug && add_asflags -g cv8
+ ;;
+ win64)
+ add_asflags -f x64
+ enabled debug && add_asflags -g cv8
+ ;;
+ linux*|solaris*)
+ add_asflags -f elf${bits}
+ enabled debug && [ "${AS}" = yasm ] && add_asflags -g dwarf2
+ enabled debug && [ "${AS}" = nasm ] && add_asflags -g
+ [ "${AS##*/}" = nasm ] && check_asm_align
+ ;;
+ darwin*)
+ add_asflags -f macho${bits}
+ enabled x86 && darwin_arch="-arch i386" || darwin_arch="-arch x86_64"
+ add_cflags ${darwin_arch}
+ add_ldflags ${darwin_arch}
+ # -mdynamic-no-pic is still a bit of voodoo -- it was required at
+ # one time, but does not seem to be now, and it breaks some of the
+ # code that still relies on inline assembly.
+ # enabled icc && ! enabled pic && add_cflags -fno-pic -mdynamic-no-pic
+ enabled icc && ! enabled pic && add_cflags -fno-pic
+ ;;
+ os2)
+ add_asflags -f aout
+ enabled debug && add_asflags -g
+ EXE_SFX=.exe
+ ;;
+ *) log "Warning: Unknown os $tgt_os while setting up $AS flags"
+ ;;
+ esac
+ ;;
+ universal*|*-gcc|generic-gnu)
+ link_with_cc=gcc
+ enable gcc
+ setup_gnu_toolchain
+ ;;
+ esac
+
+ # Try to enable CPU specific tuning
+ if [ -n "${tune_cpu}" ]; then
+ if [ -n "${tune_cflags}" ]; then
+ check_add_cflags ${tune_cflags}${tune_cpu} || \
+ die "Requested CPU '${tune_cpu}' not supported by compiler"
+ fi
+ if [ -n "${tune_asflags}" ]; then
+ check_add_asflags ${tune_asflags}${tune_cpu} || \
+ die "Requested CPU '${tune_cpu}' not supported by assembler"
+ fi
+ if [ -z "${tune_cflags}${tune_asflags}" ]; then
+ log_echo "Warning: CPU tuning not supported by this toolchain"
+ fi
+ fi
+
+ enabled debug && check_add_cflags -g && check_add_ldflags -g
+ enabled gprof && check_add_cflags -pg && check_add_ldflags -pg
+ enabled gcov &&
+ check_add_cflags -fprofile-arcs -ftest-coverage &&
+ check_add_ldflags -fprofile-arcs -ftest-coverage
+
+ if enabled optimizations; then
+ if enabled rvct; then
+ enabled small && check_add_cflags -Ospace || check_add_cflags -Otime
+ else
+ enabled small && check_add_cflags -O2 || check_add_cflags -O3
+ fi
+ fi
+
+ # Position Independent Code (PIC) support, for building relocatable
+ # shared objects
+ enabled gcc && enabled pic && check_add_cflags -fPIC
+
+ # Work around longjmp interception on glibc >= 2.11, to improve binary
+ # compatibility. See http://code.google.com/p/webm/issues/detail?id=166
+ enabled linux && check_add_cflags -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=0
+
+ # Check for strip utility variant
+ ${STRIP} -V 2>/dev/null | grep GNU >/dev/null && enable gnu_strip
+
+ # Try to determine target endianness
+ check_cc <<EOF
+ unsigned int e = 'O'<<24 | '2'<<16 | 'B'<<8 | 'E';
+EOF
+ [ -f "${TMP_O}" ] && od -A n -t x1 "${TMP_O}" | tr -d '\n' |
+ grep '4f *32 *42 *45' >/dev/null 2>&1 && enable big_endian
+
+ # Almost every platform uses pthreads.
+ if enabled multithread; then
+ case ${toolchain} in
+ *-win*-vs*);;
+ *-android-gcc);;
+ *) check_header pthread.h && add_extralibs -lpthread
+ esac
+ fi
+
+ # only for MIPS platforms
+ case ${toolchain} in
+ mips*)
+ if enabled dspr2; then
+ if enabled big_endian; then
+ echo "dspr2 optimizations are available only for little endian platforms"
+ disable dspr2
+ fi
+ fi
+ ;;
+ esac
+
+ # for sysconf(3) and friends.
+ check_header unistd.h
+
+ # glibc needs these
+ if enabled linux; then
+ add_cflags -D_LARGEFILE_SOURCE
+ add_cflags -D_FILE_OFFSET_BITS=64
+ fi
+
+ # append any user defined extra cflags
+ if [ -n "${extra_cflags}" ] ; then
+ check_add_cflags ${extra_cflags} || \
+ die "Requested extra CFLAGS '${extra_cflags}' not supported by compiler"
+ fi
+}
+
+process_toolchain() {
+ process_common_toolchain
+}
+
+print_config_mk() {
+ local prefix=$1
+ local makefile=$2
+ shift 2
+ for cfg; do
+ upname="`toupper $cfg`"
+ if enabled $cfg; then
+ echo "${prefix}_${upname}=yes" >> $makefile
+ fi
+ done
+}
+
+print_config_h() {
+ local prefix=$1
+ local header=$2
+ shift 2
+ for cfg; do
+ upname="`toupper $cfg`"
+ if enabled $cfg; then
+ echo "#define ${prefix}_${upname} 1" >> $header
+ else
+ echo "#define ${prefix}_${upname} 0" >> $header
+ fi
+ done
+}
+
+print_webm_license() {
+ local destination=$1
+ local prefix=$2
+ local suffix=$3
+ shift 3
+ cat <<EOF > ${destination}
+${prefix} Copyright (c) 2011 The WebM project authors. All Rights Reserved.${suffix}
+${prefix} ${suffix}
+${prefix} Use of this source code is governed by a BSD-style license${suffix}
+${prefix} that can be found in the LICENSE file in the root of the source${suffix}
+${prefix} tree. An additional intellectual property rights grant can be found${suffix}
+${prefix} in the file PATENTS. All contributing project authors may${suffix}
+${prefix} be found in the AUTHORS file in the root of the source tree.${suffix}
+EOF
+}
+
+process_targets() {
+ true;
+}
+
+process_detect() {
+ true;
+}
+
+enable logging
+logfile="config.err"
+self=$0
+process() {
+ cmdline_args="$@"
+ process_cmdline "$@"
+ if enabled child; then
+ echo "# ${self} $@" >> ${logfile}
+ else
+ echo "# ${self} $@" > ${logfile}
+ fi
+ post_process_common_cmdline
+ post_process_cmdline
+ process_toolchain
+ process_detect
+ process_targets
+
+ OOT_INSTALLS="${OOT_INSTALLS}"
+ if enabled source_path_used; then
+ # Prepare the PWD for building.
+ for f in ${OOT_INSTALLS}; do
+ install -D ${source_path}/$f $f
+ done
+ fi
+ cp ${source_path}/build/make/Makefile .
+
+ clean_temp_files
+ true
+}
diff --git a/libvpx/build/make/gen_asm_deps.sh b/libvpx/build/make/gen_asm_deps.sh
new file mode 100755
index 0000000..0b4e3aa
--- /dev/null
+++ b/libvpx/build/make/gen_asm_deps.sh
@@ -0,0 +1,64 @@
+#!/bin/bash
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+self=$0
+show_help() {
+ echo "usage: $self [options] <srcfile>"
+ echo
+ echo "Generate Makefile dependency information from assembly code source"
+ echo
+ exit 1
+}
+die_unknown(){
+ echo "Unknown option \"$1\"."
+ echo "See $0 --help for available options."
+ exit 1
+}
+for opt do
+ optval="${opt#*=}"
+ case "$opt" in
+ --build-pfx=*) pfx="${optval}"
+ ;;
+ --depfile=*) out="${optval}"
+ ;;
+ -I*) raw_inc_paths="${raw_inc_paths} ${opt}"
+ inc_path="${inc_path} ${opt#-I}"
+ ;;
+ -h|--help) show_help
+ ;;
+ *) [ -f "$opt" ] && srcfile="$opt"
+ ;;
+ esac
+done
+
+[ -n "$srcfile" ] || show_help
+sfx=${sfx:-asm}
+includes=$(LC_ALL=C egrep -i "include +\"?[a-z0-9_/]+\.${sfx}" $srcfile |
+ perl -p -e "s;.*?([a-z0-9_/]+.${sfx}).*;\1;")
+#" restore editor state
+for inc in ${includes}; do
+ found_inc_path=
+ for idir in ${inc_path}; do
+ [ -f "${idir}/${inc}" ] && found_inc_path="${idir}" && break
+ done
+ if [ -f `dirname $srcfile`/$inc ]; then
+ # Handle include files in the same directory as the source
+ $self --build-pfx=$pfx --depfile=$out ${raw_inc_paths} `dirname $srcfile`/$inc
+ elif [ -n "${found_inc_path}" ]; then
+ # Handle include files on the include path
+ $self --build-pfx=$pfx --depfile=$out ${raw_inc_paths} "${found_inc_path}/$inc"
+ else
+ # Handle generated includes in the build root (which may not exist yet)
+ echo ${out} ${out%d}o: "${pfx}${inc}"
+ fi
+done
+echo ${out} ${out%d}o: $srcfile
diff --git a/build/make/gen_msvs_def.sh b/libvpx/build/make/gen_msvs_def.sh
similarity index 100%
rename from build/make/gen_msvs_def.sh
rename to libvpx/build/make/gen_msvs_def.sh
diff --git a/libvpx/build/make/gen_msvs_proj.sh b/libvpx/build/make/gen_msvs_proj.sh
new file mode 100755
index 0000000..6d42941
--- /dev/null
+++ b/libvpx/build/make/gen_msvs_proj.sh
@@ -0,0 +1,573 @@
+#!/bin/bash
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+self=$0
+self_basename=${self##*/}
+self_dirname=$(dirname "$0")
+EOL=$'\n'
+
+show_help() {
+ cat <<EOF
+Usage: ${self_basename} --name=projname [options] file1 [file2 ...]
+
+This script generates a Visual Studio project file from a list of source
+code files.
+
+Options:
+ --help Print this message
+ --exe Generate a project for building an Application
+ --lib Generate a project for creating a static library
+ --static-crt Use the static C runtime (/MT)
+ --target=isa-os-cc Target specifier (required)
+ --out=filename Write output to a file [stdout]
+ --name=project_name Name of the project (required)
+ --proj-guid=GUID GUID to use for the project
+ --module-def=filename File containing export definitions (for DLLs)
+ --ver=version Version (7,8,9) of visual studio to generate for
+ --src-path-bare=dir Path to root of source tree
+ -Ipath/to/include Additional include directories
+ -DFLAG[=value] Preprocessor macros to define
+ -Lpath/to/lib Additional library search paths
+ -llibname Library to link against
+EOF
+ exit 1
+}
+
+die() {
+ echo "${self_basename}: $@" >&2
+ exit 1
+}
+
+die_unknown(){
+ echo "Unknown option \"$1\"." >&2
+ echo "See ${self_basename} --help for available options." >&2
+ exit 1
+}
+
+generate_uuid() {
+ local hex="0123456789ABCDEF"
+ local i
+ local uuid=""
+ local j
+ #93995380-89BD-4b04-88EB-625FBE52EBFB
+ for ((i=0; i<32; i++)); do
+ (( j = $RANDOM % 16 ))
+ uuid="${uuid}${hex:$j:1}"
+ done
+ echo "${uuid:0:8}-${uuid:8:4}-${uuid:12:4}-${uuid:16:4}-${uuid:20:12}"
+}
+
+indent1=" "
+indent=""
+indent_push() {
+ indent="${indent}${indent1}"
+}
+indent_pop() {
+ indent="${indent%${indent1}}"
+}
+
+tag_attributes() {
+ for opt in "$@"; do
+ optval="${opt#*=}"
+ [ -n "${optval}" ] ||
+ die "Missing attribute value in '$opt' while generating $tag tag"
+ echo "${indent}${opt%%=*}=\"${optval}\""
+ done
+}
+
+open_tag() {
+ local tag=$1
+ shift
+ if [ $# -ne 0 ]; then
+ echo "${indent}<${tag}"
+ indent_push
+ tag_attributes "$@"
+ echo "${indent}>"
+ else
+ echo "${indent}<${tag}>"
+ indent_push
+ fi
+}
+
+close_tag() {
+ local tag=$1
+ indent_pop
+ echo "${indent}</${tag}>"
+}
+
+tag() {
+ local tag=$1
+ shift
+ if [ $# -ne 0 ]; then
+ echo "${indent}<${tag}"
+ indent_push
+ tag_attributes "$@"
+ indent_pop
+ echo "${indent}/>"
+ else
+ echo "${indent}<${tag}/>"
+ fi
+}
+
+generate_filter() {
+ local var=$1
+ local name=$2
+ local pats=$3
+ local file_list_sz
+ local i
+ local f
+ local saveIFS="$IFS"
+ local pack
+ echo "generating filter '$name' from ${#file_list[@]} files" >&2
+ IFS=*
+
+ open_tag Filter \
+ Name=$name \
+ Filter=$pats \
+ UniqueIdentifier=`generate_uuid` \
+
+ file_list_sz=${#file_list[@]}
+ for i in ${!file_list[@]}; do
+ f=${file_list[i]}
+ for pat in ${pats//;/$IFS}; do
+ if [ "${f##*.}" == "$pat" ]; then
+ unset file_list[i]
+
+ open_tag File RelativePath="./$f"
+ if [ "$pat" == "asm" ] && $asm_use_custom_step; then
+ for plat in "${platforms[@]}"; do
+ for cfg in Debug Release; do
+ open_tag FileConfiguration \
+ Name="${cfg}|${plat}" \
+
+ tag Tool \
+ Name="VCCustomBuildTool" \
+ Description="Assembling \$(InputFileName)" \
+ CommandLine="$(eval echo \$asm_${cfg}_cmdline)" \
+ Outputs="\$(InputName).obj" \
+
+ close_tag FileConfiguration
+ done
+ done
+ fi
+
+ close_tag File
+
+ break
+ fi
+ done
+ done
+
+ close_tag Filter
+ IFS="$saveIFS"
+}
+
+# Process command line
+unset target
+for opt in "$@"; do
+ optval="${opt#*=}"
+ case "$opt" in
+ --help|-h) show_help
+ ;;
+ --target=*) target="${optval}"
+ ;;
+ --out=*) outfile="$optval"
+ ;;
+ --name=*) name="${optval}"
+ ;;
+ --proj-guid=*) guid="${optval}"
+ ;;
+ --module-def=*) link_opts="${link_opts} ModuleDefinitionFile=${optval}"
+ ;;
+ --exe) proj_kind="exe"
+ ;;
+ --lib) proj_kind="lib"
+ ;;
+ --src-path-bare=*) src_path_bare="$optval"
+ ;;
+ --static-crt) use_static_runtime=true
+ ;;
+ --ver=*)
+ vs_ver="$optval"
+ case "$optval" in
+ [789])
+ ;;
+ *) die Unrecognized Visual Studio Version in $opt
+ ;;
+ esac
+ ;;
+ -I*)
+ opt="${opt%/}"
+ incs="${incs}${incs:+;}"${opt##-I}""
+ yasmincs="${yasmincs} ${opt}"
+ ;;
+ -D*) defines="${defines}${defines:+;}${opt##-D}"
+ ;;
+ -L*) # fudge . to $(OutDir)
+ if [ "${opt##-L}" == "." ]; then
+ libdirs="${libdirs}${libdirs:+;}"\$(OutDir)""
+ else
+ # Also try directories for this platform/configuration
+ libdirs="${libdirs}${libdirs:+;}"${opt##-L}""
+ libdirs="${libdirs}${libdirs:+;}"${opt##-L}/\$(PlatformName)/\$(ConfigurationName)""
+ libdirs="${libdirs}${libdirs:+;}"${opt##-L}/\$(PlatformName)""
+ fi
+ ;;
+ -l*) libs="${libs}${libs:+ }${opt##-l}.lib"
+ ;;
+ -*) die_unknown $opt
+ ;;
+ *)
+ file_list[${#file_list[@]}]="$opt"
+ case "$opt" in
+ *.asm) uses_asm=true
+ ;;
+ esac
+ ;;
+ esac
+done
+outfile=${outfile:-/dev/stdout}
+guid=${guid:-`generate_uuid`}
+asm_use_custom_step=false
+uses_asm=${uses_asm:-false}
+case "${vs_ver:-8}" in
+ 7) vs_ver_id="7.10"
+ asm_use_custom_step=$uses_asm
+ ;;
+ 8) vs_ver_id="8.00"
+ ;;
+ 9) vs_ver_id="9.00"
+ ;;
+esac
+
+[ -n "$name" ] || die "Project name (--name) must be specified!"
+[ -n "$target" ] || die "Target (--target) must be specified!"
+
+if ${use_static_runtime:-false}; then
+ release_runtime=0
+ debug_runtime=1
+ lib_sfx=mt
+else
+ release_runtime=2
+ debug_runtime=3
+ lib_sfx=md
+fi
+
+# Calculate debug lib names: If a lib ends in ${lib_sfx}.lib, then rename
+# it to ${lib_sfx}d.lib. This precludes linking to release libs from a
+# debug exe, so this may need to be refactored later.
+for lib in ${libs}; do
+ if [ "$lib" != "${lib%${lib_sfx}.lib}" ]; then
+ lib=${lib%.lib}d.lib
+ fi
+ debug_libs="${debug_libs}${debug_libs:+ }${lib}"
+done
+
+
+# List Keyword for this target
+case "$target" in
+ x86*) keyword="ManagedCProj"
+ ;;
+ *) die "Unsupported target $target!"
+esac
+
+# List of all platforms supported for this target
+case "$target" in
+ x86_64*)
+ platforms[0]="x64"
+ ;;
+ x86*)
+ platforms[0]="Win32"
+ # these are only used by vs7
+ asm_Debug_cmdline="yasm -Xvc -g cv8 -f \$(PlatformName) ${yasmincs} "\$(InputPath)""
+ asm_Release_cmdline="yasm -Xvc -f \$(PlatformName) ${yasmincs} "\$(InputPath)""
+ ;;
+ *) die "Unsupported target $target!"
+ ;;
+esac
+
+generate_vcproj() {
+ case "$proj_kind" in
+ exe) vs_ConfigurationType=1
+ ;;
+ *) vs_ConfigurationType=4
+ ;;
+ esac
+
+ echo "<?xml version=\"1.0\" encoding=\"Windows-1252\"?>"
+ open_tag VisualStudioProject \
+ ProjectType="Visual C++" \
+ Version="${vs_ver_id}" \
+ Name="${name}" \
+ ProjectGUID="{${guid}}" \
+ RootNamespace="${name}" \
+ Keyword="${keyword}" \
+
+ open_tag Platforms
+ for plat in "${platforms[@]}"; do
+ tag Platform Name="$plat"
+ done
+ close_tag Platforms
+
+ open_tag ToolFiles
+ case "$target" in
+ x86*) $uses_asm && tag ToolFile RelativePath="$self_dirname/../x86-msvs/yasm.rules"
+ ;;
+ esac
+ close_tag ToolFiles
+
+ open_tag Configurations
+ for plat in "${platforms[@]}"; do
+ plat_no_ws=`echo $plat | sed 's/[^A-Za-z0-9_]/_/g'`
+ open_tag Configuration \
+ Name="Debug|$plat" \
+ OutputDirectory="\$(SolutionDir)$plat_no_ws/\$(ConfigurationName)" \
+ IntermediateDirectory="$plat_no_ws/\$(ConfigurationName)/${name}" \
+ ConfigurationType="$vs_ConfigurationType" \
+ CharacterSet="1" \
+
+ case "$target" in
+ x86*)
+ case "$name" in
+ obj_int_extract)
+ tag Tool \
+ Name="VCCLCompilerTool" \
+ Optimization="0" \
+ AdditionalIncludeDirectories="$incs" \
+ PreprocessorDefinitions="WIN32;DEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE" \
+ RuntimeLibrary="$debug_runtime" \
+ WarningLevel="3" \
+ Detect64BitPortabilityProblems="true" \
+ DebugInformationFormat="1" \
+ ;;
+ vpx)
+ tag Tool \
+ Name="VCPreBuildEventTool" \
+ CommandLine="call obj_int_extract.bat $src_path_bare" \
+
+ tag Tool \
+ Name="VCCLCompilerTool" \
+ Optimization="0" \
+ AdditionalIncludeDirectories="$incs" \
+ PreprocessorDefinitions="WIN32;_DEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
+ RuntimeLibrary="$debug_runtime" \
+ UsePrecompiledHeader="0" \
+ WarningLevel="3" \
+ DebugInformationFormat="1" \
+ Detect64BitPortabilityProblems="true" \
+
+ $uses_asm && tag Tool Name="YASM" IncludePaths="$incs" Debug="true"
+ ;;
+ *)
+ tag Tool \
+ Name="VCCLCompilerTool" \
+ Optimization="0" \
+ AdditionalIncludeDirectories="$incs" \
+ PreprocessorDefinitions="WIN32;_DEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
+ RuntimeLibrary="$debug_runtime" \
+ UsePrecompiledHeader="0" \
+ WarningLevel="3" \
+ DebugInformationFormat="1" \
+ Detect64BitPortabilityProblems="true" \
+
+ $uses_asm && tag Tool Name="YASM" IncludePaths="$incs" Debug="true"
+ ;;
+ esac
+ ;;
+ esac
+
+ case "$proj_kind" in
+ exe)
+ case "$target" in
+ x86*)
+ case "$name" in
+ obj_int_extract)
+ tag Tool \
+ Name="VCLinkerTool" \
+ OutputFile="${name}.exe" \
+ GenerateDebugInformation="true" \
+ ;;
+ *)
+ tag Tool \
+ Name="VCLinkerTool" \
+ AdditionalDependencies="$debug_libs \$(NoInherit)" \
+ AdditionalLibraryDirectories="$libdirs" \
+ GenerateDebugInformation="true" \
+ ProgramDatabaseFile="\$(OutDir)/${name}.pdb" \
+ ;;
+ esac
+ ;;
+ esac
+ ;;
+ lib)
+ case "$target" in
+ x86*)
+ tag Tool \
+ Name="VCLibrarianTool" \
+ OutputFile="\$(OutDir)/${name}${lib_sfx}d.lib" \
+
+ ;;
+ esac
+ ;;
+ dll)
+ tag Tool \
+ Name="VCLinkerTool" \
+ AdditionalDependencies="\$(NoInherit)" \
+ LinkIncremental="2" \
+ GenerateDebugInformation="true" \
+ AssemblyDebug="1" \
+ TargetMachine="1" \
+ $link_opts \
+
+ ;;
+ esac
+
+ close_tag Configuration
+
+ open_tag Configuration \
+ Name="Release|$plat" \
+ OutputDirectory="\$(SolutionDir)$plat_no_ws/\$(ConfigurationName)" \
+ IntermediateDirectory="$plat_no_ws/\$(ConfigurationName)/${name}" \
+ ConfigurationType="$vs_ConfigurationType" \
+ CharacterSet="1" \
+ WholeProgramOptimization="0" \
+
+ case "$target" in
+ x86*)
+ case "$name" in
+ obj_int_extract)
+ tag Tool \
+ Name="VCCLCompilerTool" \
+ Optimization="2" \
+ FavorSizeorSpeed="1" \
+ AdditionalIncludeDirectories="$incs" \
+ PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE" \
+ RuntimeLibrary="$release_runtime" \
+ UsePrecompiledHeader="0" \
+ WarningLevel="3" \
+ Detect64BitPortabilityProblems="true" \
+ DebugInformationFormat="0" \
+ ;;
+ vpx)
+ tag Tool \
+ Name="VCPreBuildEventTool" \
+ CommandLine="call obj_int_extract.bat $src_path_bare" \
+
+ tag Tool \
+ Name="VCCLCompilerTool" \
+ Optimization="2" \
+ FavorSizeorSpeed="1" \
+ AdditionalIncludeDirectories="$incs" \
+ PreprocessorDefinitions="WIN32;NDEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
+ RuntimeLibrary="$release_runtime" \
+ UsePrecompiledHeader="0" \
+ WarningLevel="3" \
+ DebugInformationFormat="0" \
+ Detect64BitPortabilityProblems="true" \
+
+ $uses_asm && tag Tool Name="YASM" IncludePaths="$incs"
+ ;;
+ *)
+ tag Tool \
+ Name="VCCLCompilerTool" \
+ AdditionalIncludeDirectories="$incs" \
+ Optimization="2" \
+ FavorSizeorSpeed="1" \
+ PreprocessorDefinitions="WIN32;NDEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
+ RuntimeLibrary="$release_runtime" \
+ UsePrecompiledHeader="0" \
+ WarningLevel="3" \
+ DebugInformationFormat="0" \
+ Detect64BitPortabilityProblems="true" \
+
+ $uses_asm && tag Tool Name="YASM" IncludePaths="$incs"
+ ;;
+ esac
+ ;;
+ esac
+
+ case "$proj_kind" in
+ exe)
+ case "$target" in
+ x86*)
+ case "$name" in
+ obj_int_extract)
+ tag Tool \
+ Name="VCLinkerTool" \
+ OutputFile="${name}.exe" \
+ GenerateDebugInformation="true" \
+ ;;
+ *)
+ tag Tool \
+ Name="VCLinkerTool" \
+ AdditionalDependencies="$libs \$(NoInherit)" \
+ AdditionalLibraryDirectories="$libdirs" \
+
+ ;;
+ esac
+ ;;
+ esac
+ ;;
+ lib)
+ case "$target" in
+ x86*)
+ tag Tool \
+ Name="VCLibrarianTool" \
+ OutputFile="\$(OutDir)/${name}${lib_sfx}.lib" \
+
+ ;;
+ esac
+ ;;
+ dll) # note differences to debug version: LinkIncremental, AssemblyDebug
+ tag Tool \
+ Name="VCLinkerTool" \
+ AdditionalDependencies="\$(NoInherit)" \
+ LinkIncremental="1" \
+ GenerateDebugInformation="true" \
+ TargetMachine="1" \
+ $link_opts \
+
+ ;;
+ esac
+
+ close_tag Configuration
+ done
+ close_tag Configurations
+
+ open_tag Files
+ generate_filter srcs "Source Files" "c;cc;def;odl;idl;hpj;bat;asm;asmx"
+ generate_filter hdrs "Header Files" "h;hm;inl;inc;xsd"
+ generate_filter resrcs "Resource Files" "rc;ico;cur;bmp;dlg;rc2;rct;bin;rgs;gif;jpg;jpeg;jpe;resx;tiff;tif;png;wav"
+ generate_filter resrcs "Build Files" "mk"
+ close_tag Files
+
+ tag Globals
+ close_tag VisualStudioProject
+
+ # This must be done from within the {} subshell
+ echo "Ignored files list (${#file_list[@]} items) is:" >&2
+ for f in "${file_list[@]}"; do
+ echo " $f" >&2
+ done
+}
+
+generate_vcproj |
+ sed -e '/"/s;\([^ "]\)/;\1\\;g' > ${outfile}
+
+exit
+<!--
+TODO: Add any files not captured by filters.
+ <File
+ RelativePath=".\ReadMe.txt"
+ >
+ </File>
+-->
diff --git a/build/make/gen_msvs_sln.sh b/libvpx/build/make/gen_msvs_sln.sh
similarity index 100%
rename from build/make/gen_msvs_sln.sh
rename to libvpx/build/make/gen_msvs_sln.sh
diff --git a/libvpx/build/make/obj_int_extract.c b/libvpx/build/make/obj_int_extract.c
new file mode 100644
index 0000000..bf317bd
--- /dev/null
+++ b/libvpx/build/make/obj_int_extract.c
@@ -0,0 +1,914 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "vpx_config.h"
+#include "vpx/vpx_integer.h"
+
+typedef enum
+{
+ OUTPUT_FMT_PLAIN,
+ OUTPUT_FMT_RVDS,
+ OUTPUT_FMT_GAS,
+} output_fmt_t;
+
+int log_msg(const char *fmt, ...)
+{
+ int res;
+ va_list ap;
+ va_start(ap, fmt);
+ res = vfprintf(stderr, fmt, ap);
+ va_end(ap);
+ return res;
+}
+
+#if defined(__GNUC__) && __GNUC__
+#if defined(__MACH__)
+
+#include <mach-o/loader.h>
+#include <mach-o/nlist.h>
+
+int parse_macho(uint8_t *base_buf, size_t sz)
+{
+ int i, j;
+ struct mach_header header;
+ uint8_t *buf = base_buf;
+ int base_data_section = 0;
+ int bits = 0;
+
+ /* We can read in mach_header for 32 and 64 bit architectures
+ * because it's identical to mach_header_64 except for the last
+ * element (uint32_t reserved), which we don't use. Then, when
+ * we know which architecture we're looking at, increment buf
+ * appropriately.
+ */
+ memcpy(&header, buf, sizeof(struct mach_header));
+
+ if (header.magic == MH_MAGIC)
+ {
+ if (header.cputype == CPU_TYPE_ARM
+ || header.cputype == CPU_TYPE_X86)
+ {
+ bits = 32;
+ buf += sizeof(struct mach_header);
+ }
+ else
+ {
+ log_msg("Bad cputype for object file. Currently only tested for CPU_TYPE_[ARM|X86].\n");
+ goto bail;
+ }
+ }
+ else if (header.magic == MH_MAGIC_64)
+ {
+ if (header.cputype == CPU_TYPE_X86_64)
+ {
+ bits = 64;
+ buf += sizeof(struct mach_header_64);
+ }
+ else
+ {
+ log_msg("Bad cputype for object file. Currently only tested for CPU_TYPE_X86_64.\n");
+ goto bail;
+ }
+ }
+ else
+ {
+ log_msg("Bad magic number for object file. 0x%x or 0x%x expected, 0x%x found.\n",
+ MH_MAGIC, MH_MAGIC_64, header.magic);
+ goto bail;
+ }
+
+ if (header.filetype != MH_OBJECT)
+ {
+ log_msg("Bad filetype for object file. Currently only tested for MH_OBJECT.\n");
+ goto bail;
+ }
+
+ for (i = 0; i < header.ncmds; i++)
+ {
+ struct load_command lc;
+
+ memcpy(&lc, buf, sizeof(struct load_command));
+
+ if (lc.cmd == LC_SEGMENT)
+ {
+ uint8_t *seg_buf = buf;
+ struct section s;
+ struct segment_command seg_c;
+
+ memcpy(&seg_c, seg_buf, sizeof(struct segment_command));
+ seg_buf += sizeof(struct segment_command);
+
+ /* Although each section is given it's own offset, nlist.n_value
+ * references the offset of the first section. This isn't
+ * apparent without debug information because the offset of the
+ * data section is the same as the first section. However, with
+ * debug sections mixed in, the offset of the debug section
+ * increases but n_value still references the first section.
+ */
+ if (seg_c.nsects < 1)
+ {
+ log_msg("Not enough sections\n");
+ goto bail;
+ }
+
+ memcpy(&s, seg_buf, sizeof(struct section));
+ base_data_section = s.offset;
+ }
+ else if (lc.cmd == LC_SEGMENT_64)
+ {
+ uint8_t *seg_buf = buf;
+ struct section_64 s;
+ struct segment_command_64 seg_c;
+
+ memcpy(&seg_c, seg_buf, sizeof(struct segment_command_64));
+ seg_buf += sizeof(struct segment_command_64);
+
+ /* Explanation in LG_SEGMENT */
+ if (seg_c.nsects < 1)
+ {
+ log_msg("Not enough sections\n");
+ goto bail;
+ }
+
+ memcpy(&s, seg_buf, sizeof(struct section_64));
+ base_data_section = s.offset;
+ }
+ else if (lc.cmd == LC_SYMTAB)
+ {
+ if (base_data_section != 0)
+ {
+ struct symtab_command sc;
+ uint8_t *sym_buf = base_buf;
+ uint8_t *str_buf = base_buf;
+
+ memcpy(&sc, buf, sizeof(struct symtab_command));
+
+ if (sc.cmdsize != sizeof(struct symtab_command))
+ {
+ log_msg("Can't find symbol table!\n");
+ goto bail;
+ }
+
+ sym_buf += sc.symoff;
+ str_buf += sc.stroff;
+
+ for (j = 0; j < sc.nsyms; j++)
+ {
+ /* Location of string is cacluated each time from the
+ * start of the string buffer. On darwin the symbols
+ * are prefixed by "_", so we bump the pointer by 1.
+ * The target value is defined as an int in asm_*_offsets.c,
+ * which is 4 bytes on all targets we currently use.
+ */
+ if (bits == 32)
+ {
+ struct nlist nl;
+ int val;
+
+ memcpy(&nl, sym_buf, sizeof(struct nlist));
+ sym_buf += sizeof(struct nlist);
+
+ memcpy(&val, base_buf + base_data_section + nl.n_value,
+ sizeof(val));
+ printf("%-40s EQU %5d\n",
+ str_buf + nl.n_un.n_strx + 1, val);
+ }
+ else /* if (bits == 64) */
+ {
+ struct nlist_64 nl;
+ int val;
+
+ memcpy(&nl, sym_buf, sizeof(struct nlist_64));
+ sym_buf += sizeof(struct nlist_64);
+
+ memcpy(&val, base_buf + base_data_section + nl.n_value,
+ sizeof(val));
+ printf("%-40s EQU %5d\n",
+ str_buf + nl.n_un.n_strx + 1, val);
+ }
+ }
+ }
+ }
+
+ buf += lc.cmdsize;
+ }
+
+ return 0;
+bail:
+ return 1;
+
+}
+
+#elif defined(__ELF__)
+#include "elf.h"
+
+#define COPY_STRUCT(dst, buf, ofst, sz) do {\
+ if(ofst + sizeof((*(dst))) > sz) goto bail;\
+ memcpy(dst, buf+ofst, sizeof((*(dst))));\
+ } while(0)
+
+#define ENDIAN_ASSIGN(val, memb) do {\
+ if(!elf->le_data) {log_msg("Big Endian data not supported yet!\n");goto bail;}\
+ (val) = (memb);\
+ } while(0)
+
+#define ENDIAN_ASSIGN_IN_PLACE(memb) do {\
+ ENDIAN_ASSIGN(memb, memb);\
+ } while(0)
+
+typedef struct
+{
+ uint8_t *buf; /* Buffer containing ELF data */
+ size_t sz; /* Buffer size */
+ int le_data; /* Data is little-endian */
+ unsigned char e_ident[EI_NIDENT]; /* Magic number and other info */
+ int bits; /* 32 or 64 */
+ Elf32_Ehdr hdr32;
+ Elf64_Ehdr hdr64;
+} elf_obj_t;
+
+int parse_elf_header(elf_obj_t *elf)
+{
+ int res;
+ /* Verify ELF Magic numbers */
+ COPY_STRUCT(&elf->e_ident, elf->buf, 0, elf->sz);
+ res = elf->e_ident[EI_MAG0] == ELFMAG0;
+ res &= elf->e_ident[EI_MAG1] == ELFMAG1;
+ res &= elf->e_ident[EI_MAG2] == ELFMAG2;
+ res &= elf->e_ident[EI_MAG3] == ELFMAG3;
+ res &= elf->e_ident[EI_CLASS] == ELFCLASS32
+ || elf->e_ident[EI_CLASS] == ELFCLASS64;
+ res &= elf->e_ident[EI_DATA] == ELFDATA2LSB;
+
+ if (!res) goto bail;
+
+ elf->le_data = elf->e_ident[EI_DATA] == ELFDATA2LSB;
+
+ /* Read in relevant values */
+ if (elf->e_ident[EI_CLASS] == ELFCLASS32)
+ {
+ elf->bits = 32;
+ COPY_STRUCT(&elf->hdr32, elf->buf, 0, elf->sz);
+
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_type);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_machine);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_version);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_entry);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phoff);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shoff);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_flags);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_ehsize);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phentsize);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phnum);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shentsize);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shnum);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shstrndx);
+ }
+ else /* if (elf->e_ident[EI_CLASS] == ELFCLASS64) */
+ {
+ elf->bits = 64;
+ COPY_STRUCT(&elf->hdr64, elf->buf, 0, elf->sz);
+
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_type);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_machine);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_version);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_entry);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phoff);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shoff);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_flags);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_ehsize);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phentsize);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phnum);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shentsize);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shnum);
+ ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shstrndx);
+ }
+
+ return 0;
+bail:
+ log_msg("Failed to parse ELF file header");
+ return 1;
+}
+
+int parse_elf_section(elf_obj_t *elf, int idx, Elf32_Shdr *hdr32, Elf64_Shdr *hdr64)
+{
+ if (hdr32)
+ {
+ if (idx >= elf->hdr32.e_shnum)
+ goto bail;
+
+ COPY_STRUCT(hdr32, elf->buf, elf->hdr32.e_shoff + idx * elf->hdr32.e_shentsize,
+ elf->sz);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_name);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_type);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_flags);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_addr);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_offset);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_size);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_link);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_info);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_addralign);
+ ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_entsize);
+ }
+ else /* if (hdr64) */
+ {
+ if (idx >= elf->hdr64.e_shnum)
+ goto bail;
+
+ COPY_STRUCT(hdr64, elf->buf, elf->hdr64.e_shoff + idx * elf->hdr64.e_shentsize,
+ elf->sz);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_name);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_type);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_flags);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_addr);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_offset);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_size);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_link);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_info);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_addralign);
+ ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_entsize);
+ }
+
+ return 0;
+bail:
+ return 1;
+}
+
+char *parse_elf_string_table(elf_obj_t *elf, int s_idx, int idx)
+{
+ if (elf->bits == 32)
+ {
+ Elf32_Shdr shdr;
+
+ if (parse_elf_section(elf, s_idx, &shdr, NULL))
+ {
+ log_msg("Failed to parse ELF string table: section %d, index %d\n",
+ s_idx, idx);
+ return "";
+ }
+
+ return (char *)(elf->buf + shdr.sh_offset + idx);
+ }
+ else /* if (elf->bits == 64) */
+ {
+ Elf64_Shdr shdr;
+
+ if (parse_elf_section(elf, s_idx, NULL, &shdr))
+ {
+ log_msg("Failed to parse ELF string table: section %d, index %d\n",
+ s_idx, idx);
+ return "";
+ }
+
+ return (char *)(elf->buf + shdr.sh_offset + idx);
+ }
+}
+
+int parse_elf_symbol(elf_obj_t *elf, unsigned int ofst, Elf32_Sym *sym32, Elf64_Sym *sym64)
+{
+ if (sym32)
+ {
+ COPY_STRUCT(sym32, elf->buf, ofst, elf->sz);
+ ENDIAN_ASSIGN_IN_PLACE(sym32->st_name);
+ ENDIAN_ASSIGN_IN_PLACE(sym32->st_value);
+ ENDIAN_ASSIGN_IN_PLACE(sym32->st_size);
+ ENDIAN_ASSIGN_IN_PLACE(sym32->st_info);
+ ENDIAN_ASSIGN_IN_PLACE(sym32->st_other);
+ ENDIAN_ASSIGN_IN_PLACE(sym32->st_shndx);
+ }
+ else /* if (sym64) */
+ {
+ COPY_STRUCT(sym64, elf->buf, ofst, elf->sz);
+ ENDIAN_ASSIGN_IN_PLACE(sym64->st_name);
+ ENDIAN_ASSIGN_IN_PLACE(sym64->st_value);
+ ENDIAN_ASSIGN_IN_PLACE(sym64->st_size);
+ ENDIAN_ASSIGN_IN_PLACE(sym64->st_info);
+ ENDIAN_ASSIGN_IN_PLACE(sym64->st_other);
+ ENDIAN_ASSIGN_IN_PLACE(sym64->st_shndx);
+ }
+ return 0;
+bail:
+ return 1;
+}
+
+int parse_elf(uint8_t *buf, size_t sz, output_fmt_t mode)
+{
+ elf_obj_t elf;
+ unsigned int ofst;
+ int i;
+ Elf32_Off strtab_off32;
+ Elf64_Off strtab_off64; /* save String Table offset for later use */
+
+ memset(&elf, 0, sizeof(elf));
+ elf.buf = buf;
+ elf.sz = sz;
+
+ /* Parse Header */
+ if (parse_elf_header(&elf))
+ goto bail;
+
+ if (elf.bits == 32)
+ {
+ Elf32_Shdr shdr;
+ for (i = 0; i < elf.hdr32.e_shnum; i++)
+ {
+ parse_elf_section(&elf, i, &shdr, NULL);
+
+ if (shdr.sh_type == SHT_STRTAB)
+ {
+ char strtsb_name[128];
+
+ strcpy(strtsb_name, (char *)(elf.buf + shdr.sh_offset + shdr.sh_name));
+
+ if (!(strcmp(strtsb_name, ".shstrtab")))
+ {
+ /* log_msg("found section: %s\n", strtsb_name); */
+ strtab_off32 = shdr.sh_offset;
+ break;
+ }
+ }
+ }
+ }
+ else /* if (elf.bits == 64) */
+ {
+ Elf64_Shdr shdr;
+ for (i = 0; i < elf.hdr64.e_shnum; i++)
+ {
+ parse_elf_section(&elf, i, NULL, &shdr);
+
+ if (shdr.sh_type == SHT_STRTAB)
+ {
+ char strtsb_name[128];
+
+ strcpy(strtsb_name, (char *)(elf.buf + shdr.sh_offset + shdr.sh_name));
+
+ if (!(strcmp(strtsb_name, ".shstrtab")))
+ {
+ /* log_msg("found section: %s\n", strtsb_name); */
+ strtab_off64 = shdr.sh_offset;
+ break;
+ }
+ }
+ }
+ }
+
+ /* Parse all Symbol Tables */
+ if (elf.bits == 32)
+ {
+ Elf32_Shdr shdr;
+ for (i = 0; i < elf.hdr32.e_shnum; i++)
+ {
+ parse_elf_section(&elf, i, &shdr, NULL);
+
+ if (shdr.sh_type == SHT_SYMTAB)
+ {
+ for (ofst = shdr.sh_offset;
+ ofst < shdr.sh_offset + shdr.sh_size;
+ ofst += shdr.sh_entsize)
+ {
+ Elf32_Sym sym;
+
+ parse_elf_symbol(&elf, ofst, &sym, NULL);
+
+ /* For all OBJECTS (data objects), extract the value from the
+ * proper data segment.
+ */
+ /* if (ELF32_ST_TYPE(sym.st_info) == STT_OBJECT && sym.st_name)
+ log_msg("found data object %s\n",
+ parse_elf_string_table(&elf,
+ shdr.sh_link,
+ sym.st_name));
+ */
+
+ if (ELF32_ST_TYPE(sym.st_info) == STT_OBJECT
+ && sym.st_size == 4)
+ {
+ Elf32_Shdr dhdr;
+ int val = 0;
+ char section_name[128];
+
+ parse_elf_section(&elf, sym.st_shndx, &dhdr, NULL);
+
+ /* For explanition - refer to _MSC_VER version of code */
+ strcpy(section_name, (char *)(elf.buf + strtab_off32 + dhdr.sh_name));
+ /* log_msg("Section_name: %s, Section_type: %d\n", section_name, dhdr.sh_type); */
+
+ if (strcmp(section_name, ".bss"))
+ {
+ if (sizeof(val) != sym.st_size)
+ {
+ /* The target value is declared as an int in
+ * asm_*_offsets.c, which is 4 bytes on all
+ * targets we currently use. Complain loudly if
+ * this is not true.
+ */
+ log_msg("Symbol size is wrong\n");
+ goto bail;
+ }
+
+ memcpy(&val,
+ elf.buf + dhdr.sh_offset + sym.st_value,
+ sym.st_size);
+ }
+
+ if (!elf.le_data)
+ {
+ log_msg("Big Endian data not supported yet!\n");
+ goto bail;
+ }
+
+ switch (mode)
+ {
+ case OUTPUT_FMT_RVDS:
+ printf("%-40s EQU %5d\n",
+ parse_elf_string_table(&elf,
+ shdr.sh_link,
+ sym.st_name),
+ val);
+ break;
+ case OUTPUT_FMT_GAS:
+ printf(".equ %-40s, %5d\n",
+ parse_elf_string_table(&elf,
+ shdr.sh_link,
+ sym.st_name),
+ val);
+ break;
+ default:
+ printf("%s = %d\n",
+ parse_elf_string_table(&elf,
+ shdr.sh_link,
+ sym.st_name),
+ val);
+ }
+ }
+ }
+ }
+ }
+ }
+ else /* if (elf.bits == 64) */
+ {
+ Elf64_Shdr shdr;
+ for (i = 0; i < elf.hdr64.e_shnum; i++)
+ {
+ parse_elf_section(&elf, i, NULL, &shdr);
+
+ if (shdr.sh_type == SHT_SYMTAB)
+ {
+ for (ofst = shdr.sh_offset;
+ ofst < shdr.sh_offset + shdr.sh_size;
+ ofst += shdr.sh_entsize)
+ {
+ Elf64_Sym sym;
+
+ parse_elf_symbol(&elf, ofst, NULL, &sym);
+
+ /* For all OBJECTS (data objects), extract the value from the
+ * proper data segment.
+ */
+ /* if (ELF64_ST_TYPE(sym.st_info) == STT_OBJECT && sym.st_name)
+ log_msg("found data object %s\n",
+ parse_elf_string_table(&elf,
+ shdr.sh_link,
+ sym.st_name));
+ */
+
+ if (ELF64_ST_TYPE(sym.st_info) == STT_OBJECT
+ && sym.st_size == 4)
+ {
+ Elf64_Shdr dhdr;
+ int val = 0;
+ char section_name[128];
+
+ parse_elf_section(&elf, sym.st_shndx, NULL, &dhdr);
+
+ /* For explanition - refer to _MSC_VER version of code */
+ strcpy(section_name, (char *)(elf.buf + strtab_off64 + dhdr.sh_name));
+ /* log_msg("Section_name: %s, Section_type: %d\n", section_name, dhdr.sh_type); */
+
+ if ((strcmp(section_name, ".bss")))
+ {
+ if (sizeof(val) != sym.st_size)
+ {
+ /* The target value is declared as an int in
+ * asm_*_offsets.c, which is 4 bytes on all
+ * targets we currently use. Complain loudly if
+ * this is not true.
+ */
+ log_msg("Symbol size is wrong\n");
+ goto bail;
+ }
+
+ memcpy(&val,
+ elf.buf + dhdr.sh_offset + sym.st_value,
+ sym.st_size);
+ }
+
+ if (!elf.le_data)
+ {
+ log_msg("Big Endian data not supported yet!\n");
+ goto bail;
+ }
+
+ switch (mode)
+ {
+ case OUTPUT_FMT_RVDS:
+ printf("%-40s EQU %5d\n",
+ parse_elf_string_table(&elf,
+ shdr.sh_link,
+ sym.st_name),
+ val);
+ break;
+ case OUTPUT_FMT_GAS:
+ printf(".equ %-40s, %5d\n",
+ parse_elf_string_table(&elf,
+ shdr.sh_link,
+ sym.st_name),
+ val);
+ break;
+ default:
+ printf("%s = %d\n",
+ parse_elf_string_table(&elf,
+ shdr.sh_link,
+ sym.st_name),
+ val);
+ }
+ }
+ }
+ }
+ }
+ }
+
+ if (mode == OUTPUT_FMT_RVDS)
+ printf(" END\n");
+
+ return 0;
+bail:
+ log_msg("Parse error: File does not appear to be valid ELF32 or ELF64\n");
+ return 1;
+}
+
+#endif
+#endif /* defined(__GNUC__) && __GNUC__ */
+
+
+#if defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__)
+/* See "Microsoft Portable Executable and Common Object File Format Specification"
+ for reference.
+*/
+#define get_le32(x) ((*(x)) | (*(x+1)) << 8 |(*(x+2)) << 16 | (*(x+3)) << 24 )
+#define get_le16(x) ((*(x)) | (*(x+1)) << 8)
+
+int parse_coff(uint8_t *buf, size_t sz)
+{
+ unsigned int nsections, symtab_ptr, symtab_sz, strtab_ptr;
+ unsigned int sectionrawdata_ptr;
+ unsigned int i;
+ uint8_t *ptr;
+ uint32_t symoffset;
+
+ char **sectionlist; //this array holds all section names in their correct order.
+ //it is used to check if the symbol is in .bss or .rdata section.
+
+ nsections = get_le16(buf + 2);
+ symtab_ptr = get_le32(buf + 8);
+ symtab_sz = get_le32(buf + 12);
+ strtab_ptr = symtab_ptr + symtab_sz * 18;
+
+ if (nsections > 96)
+ {
+ log_msg("Too many sections\n");
+ return 1;
+ }
+
+ sectionlist = malloc(nsections * sizeof(sectionlist));
+
+ if (sectionlist == NULL)
+ {
+ log_msg("Allocating first level of section list failed\n");
+ return 1;
+ }
+
+ //log_msg("COFF: Found %u symbols in %u sections.\n", symtab_sz, nsections);
+
+ /*
+ The size of optional header is always zero for an obj file. So, the section header
+ follows the file header immediately.
+ */
+
+ ptr = buf + 20; //section header
+
+ for (i = 0; i < nsections; i++)
+ {
+ char sectionname[9] = {0, 0, 0, 0, 0, 0, 0, 0, 0};
+ strncpy(sectionname, ptr, 8);
+ //log_msg("COFF: Parsing section %s\n",sectionname);
+
+ sectionlist[i] = malloc(strlen(sectionname) + 1);
+
+ if (sectionlist[i] == NULL)
+ {
+ log_msg("Allocating storage for %s failed\n", sectionname);
+ goto bail;
+ }
+ strcpy(sectionlist[i], sectionname);
+
+ if (!strcmp(sectionname, ".rdata")) sectionrawdata_ptr = get_le32(ptr + 20);
+
+ ptr += 40;
+ }
+
+ //log_msg("COFF: Symbol table at offset %u\n", symtab_ptr);
+ //log_msg("COFF: raw data pointer ofset for section .rdata is %u\n", sectionrawdata_ptr);
+
+ /* The compiler puts the data with non-zero offset in .rdata section, but puts the data with
+ zero offset in .bss section. So, if the data in in .bss section, set offset=0.
+ Note from Wiki: In an object module compiled from C, the bss section contains
+ the local variables (but not functions) that were declared with the static keyword,
+ except for those with non-zero initial values. (In C, static variables are initialized
+ to zero by default.) It also contains the non-local (both extern and static) variables
+ that are also initialized to zero (either explicitly or by default).
+ */
+ //move to symbol table
+ /* COFF symbol table:
+ offset field
+ 0 Name(*)
+ 8 Value
+ 12 SectionNumber
+ 14 Type
+ 16 StorageClass
+ 17 NumberOfAuxSymbols
+ */
+ ptr = buf + symtab_ptr;
+
+ for (i = 0; i < symtab_sz; i++)
+ {
+ int16_t section = get_le16(ptr + 12); //section number
+
+ if (section > 0 && ptr[16] == 2)
+ {
+ //if(section > 0 && ptr[16] == 3 && get_le32(ptr+8)) {
+
+ if (get_le32(ptr))
+ {
+ char name[9] = {0, 0, 0, 0, 0, 0, 0, 0, 0};
+ strncpy(name, ptr, 8);
+ //log_msg("COFF: Parsing symbol %s\n",name);
+ /* The 64bit Windows compiler doesn't prefix with an _.
+ * Check what's there, and bump if necessary
+ */
+ if (name[0] == '_')
+ printf("%-40s EQU ", name + 1);
+ else
+ printf("%-40s EQU ", name);
+ }
+ else
+ {
+ //log_msg("COFF: Parsing symbol %s\n",
+ // buf + strtab_ptr + get_le32(ptr+4));
+ if ((buf + strtab_ptr + get_le32(ptr + 4))[0] == '_')
+ printf("%-40s EQU ",
+ buf + strtab_ptr + get_le32(ptr + 4) + 1);
+ else
+ printf("%-40s EQU ", buf + strtab_ptr + get_le32(ptr + 4));
+ }
+
+ if (!(strcmp(sectionlist[section-1], ".bss")))
+ {
+ symoffset = 0;
+ }
+ else
+ {
+ symoffset = get_le32(buf + sectionrawdata_ptr + get_le32(ptr + 8));
+ }
+
+ //log_msg(" Section: %d\n",section);
+ //log_msg(" Class: %d\n",ptr[16]);
+ //log_msg(" Address: %u\n",get_le32(ptr+8));
+ //log_msg(" Offset: %u\n", symoffset);
+
+ printf("%5d\n", symoffset);
+ }
+
+ ptr += 18;
+ }
+
+ printf(" END\n");
+
+ for (i = 0; i < nsections; i++)
+ {
+ free(sectionlist[i]);
+ }
+
+ free(sectionlist);
+
+ return 0;
+bail:
+
+ for (i = 0; i < nsections; i++)
+ {
+ free(sectionlist[i]);
+ }
+
+ free(sectionlist);
+
+ return 1;
+}
+#endif /* defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__) */
+
+int main(int argc, char **argv)
+{
+ output_fmt_t mode = OUTPUT_FMT_PLAIN;
+ const char *f;
+ uint8_t *file_buf;
+ int res;
+ FILE *fp;
+ long int file_size;
+
+ if (argc < 2 || argc > 3)
+ {
+ fprintf(stderr, "Usage: %s [output format] <obj file>\n\n", argv[0]);
+ fprintf(stderr, " <obj file>\tobject file to parse\n");
+ fprintf(stderr, "Output Formats:\n");
+ fprintf(stderr, " gas - compatible with GNU assembler\n");
+ fprintf(stderr, " rvds - compatible with armasm\n");
+ goto bail;
+ }
+
+ f = argv[2];
+
+ if (!strcmp(argv[1], "rvds"))
+ mode = OUTPUT_FMT_RVDS;
+ else if (!strcmp(argv[1], "gas"))
+ mode = OUTPUT_FMT_GAS;
+ else
+ f = argv[1];
+
+ fp = fopen(f, "rb");
+
+ if (!fp)
+ {
+ perror("Unable to open file");
+ goto bail;
+ }
+
+ if (fseek(fp, 0, SEEK_END))
+ {
+ perror("stat");
+ goto bail;
+ }
+
+ file_size = ftell(fp);
+ file_buf = malloc(file_size);
+
+ if (!file_buf)
+ {
+ perror("malloc");
+ goto bail;
+ }
+
+ rewind(fp);
+
+ if (fread(file_buf, sizeof(char), file_size, fp) != file_size)
+ {
+ perror("read");
+ goto bail;
+ }
+
+ if (fclose(fp))
+ {
+ perror("close");
+ goto bail;
+ }
+
+#if defined(__GNUC__) && __GNUC__
+#if defined(__MACH__)
+ res = parse_macho(file_buf, file_size);
+#elif defined(__ELF__)
+ res = parse_elf(file_buf, file_size, mode);
+#endif
+#endif
+#if defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__)
+ res = parse_coff(file_buf, file_size);
+#endif
+
+ free(file_buf);
+
+ if (!res)
+ return EXIT_SUCCESS;
+
+bail:
+ return EXIT_FAILURE;
+}
diff --git a/libvpx/build/make/rtcd.sh b/libvpx/build/make/rtcd.sh
new file mode 100755
index 0000000..ddf9e09
--- /dev/null
+++ b/libvpx/build/make/rtcd.sh
@@ -0,0 +1,356 @@
+#!/bin/sh
+self=$0
+
+usage() {
+ cat <<EOF >&2
+Usage: $self [options] FILE
+
+Reads the Run Time CPU Detections definitions from FILE and generates a
+C header file on stdout.
+
+Options:
+ --arch=ARCH Architecture to generate defs for (required)
+ --disable-EXT Disable support for EXT extensions
+ --require-EXT Require support for EXT extensions
+ --sym=SYMBOL Unique symbol to use for RTCD initialization function
+ --config=FILE File with CONFIG_FOO=yes lines to parse
+EOF
+ exit 1
+}
+
+die() {
+ echo "$@" >&2
+ exit 1
+}
+
+die_argument_required() {
+ die "Option $opt requires argument"
+}
+
+for opt; do
+ optval="${opt#*=}"
+ case "$opt" in
+ --arch) die_argument_required;;
+ --arch=*) arch=${optval};;
+ --disable-*) eval "disable_${opt#--disable-}=true";;
+ --require-*) REQUIRES="${REQUIRES}${opt#--require-} ";;
+ --sym) die_argument_required;;
+ --sym=*) symbol=${optval};;
+ --config=*) config_file=${optval};;
+ -h|--help)
+ usage
+ ;;
+ -*)
+ die "Unrecognized option: ${opt%%=*}"
+ ;;
+ *)
+ defs_file="$defs_file $opt"
+ ;;
+ esac
+ shift
+done
+for f in $defs_file; do [ -f "$f" ] || usage; done
+[ -n "$arch" ] || usage
+
+# Import the configuration
+[ -f "$config_file" ] && eval $(grep CONFIG_ "$config_file")
+
+#
+# Routines for the RTCD DSL to call
+#
+prototype() {
+ local rtyp
+ case "$1" in
+ unsigned) rtyp="$1 "; shift;;
+ esac
+ rtyp="${rtyp}$1"
+ local fn="$2"
+ local args="$3"
+
+ eval "${2}_rtyp='$rtyp'"
+ eval "${2}_args='$3'"
+ ALL_FUNCS="$ALL_FUNCS $fn"
+ specialize $fn c
+}
+
+specialize() {
+ local fn="$1"
+ shift
+ for opt in "$@"; do
+ eval "${fn}_${opt}=${fn}_${opt}"
+ done
+}
+
+require() {
+ for fn in $ALL_FUNCS; do
+ for opt in "$@"; do
+ local ofn=$(eval "echo \$${fn}_${opt}")
+ [ -z "$ofn" ] && continue
+
+ # if we already have a default, then we can disable it, as we know
+ # we can do better.
+ local best=$(eval "echo \$${fn}_default")
+ local best_ofn=$(eval "echo \$${best}")
+ [ -n "$best" ] && [ "$best_ofn" != "$ofn" ] && eval "${best}_link=false"
+ eval "${fn}_default=${fn}_${opt}"
+ eval "${fn}_${opt}_link=true"
+ done
+ done
+}
+
+forward_decls() {
+ ALL_FORWARD_DECLS="$ALL_FORWARD_DECLS $1"
+}
+
+#
+# Include the user's directives
+#
+for f in $defs_file; do
+ . $f
+done
+
+#
+# Process the directives according to the command line
+#
+process_forward_decls() {
+ for fn in $ALL_FORWARD_DECLS; do
+ eval $fn
+ done
+}
+
+determine_indirection() {
+ [ "$CONFIG_RUNTIME_CPU_DETECT" = "yes" ] || require $ALL_ARCHS
+ for fn in $ALL_FUNCS; do
+ local n=""
+ local rtyp="$(eval "echo \$${fn}_rtyp")"
+ local args="$(eval "echo \"\$${fn}_args\"")"
+ local dfn="$(eval "echo \$${fn}_default")"
+ dfn=$(eval "echo \$${dfn}")
+ for opt in "$@"; do
+ local ofn=$(eval "echo \$${fn}_${opt}")
+ [ -z "$ofn" ] && continue
+ local link=$(eval "echo \$${fn}_${opt}_link")
+ [ "$link" = "false" ] && continue
+ n="${n}x"
+ done
+ if [ "$n" = "x" ]; then
+ eval "${fn}_indirect=false"
+ else
+ eval "${fn}_indirect=true"
+ fi
+ done
+}
+
+declare_function_pointers() {
+ for fn in $ALL_FUNCS; do
+ local rtyp="$(eval "echo \$${fn}_rtyp")"
+ local args="$(eval "echo \"\$${fn}_args\"")"
+ local dfn="$(eval "echo \$${fn}_default")"
+ dfn=$(eval "echo \$${dfn}")
+ for opt in "$@"; do
+ local ofn=$(eval "echo \$${fn}_${opt}")
+ [ -z "$ofn" ] && continue
+ echo "$rtyp ${ofn}($args);"
+ done
+ if [ "$(eval "echo \$${fn}_indirect")" = "false" ]; then
+ echo "#define ${fn} ${dfn}"
+ else
+ echo "RTCD_EXTERN $rtyp (*${fn})($args);"
+ fi
+ echo
+ done
+}
+
+set_function_pointers() {
+ for fn in $ALL_FUNCS; do
+ local n=""
+ local rtyp="$(eval "echo \$${fn}_rtyp")"
+ local args="$(eval "echo \"\$${fn}_args\"")"
+ local dfn="$(eval "echo \$${fn}_default")"
+ dfn=$(eval "echo \$${dfn}")
+ if $(eval "echo \$${fn}_indirect"); then
+ echo " $fn = $dfn;"
+ for opt in "$@"; do
+ local ofn=$(eval "echo \$${fn}_${opt}")
+ [ -z "$ofn" ] && continue
+ [ "$ofn" = "$dfn" ] && continue;
+ local link=$(eval "echo \$${fn}_${opt}_link")
+ [ "$link" = "false" ] && continue
+ local cond="$(eval "echo \$have_${opt}")"
+ echo " if (${cond}) $fn = $ofn;"
+ done
+ fi
+ echo
+ done
+}
+
+filter() {
+ local filtered
+ for opt in "$@"; do
+ [ -z $(eval "echo \$disable_${opt}") ] && filtered="$filtered $opt"
+ done
+ echo $filtered
+}
+
+#
+# Helper functions for generating the arch specific RTCD files
+#
+common_top() {
+ local outfile_basename=$(basename ${symbol:-rtcd.h})
+ local include_guard=$(echo $outfile_basename | tr '[a-z]' '[A-Z]' | tr -c '[A-Z]' _)
+ cat <<EOF
+#ifndef ${include_guard}
+#define ${include_guard}
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+$(process_forward_decls)
+
+$(declare_function_pointers c $ALL_ARCHS)
+
+void ${symbol:-rtcd}(void);
+EOF
+}
+
+common_bottom() {
+ cat <<EOF
+#endif
+EOF
+}
+
+x86() {
+ determine_indirection c $ALL_ARCHS
+
+ # Assign the helper variable for each enabled extension
+ for opt in $ALL_ARCHS; do
+ local uc=$(echo $opt | tr '[a-z]' '[A-Z]')
+ eval "have_${opt}=\"flags & HAS_${uc}\""
+ done
+
+ cat <<EOF
+$(common_top)
+
+#ifdef RTCD_C
+#include "vpx_ports/x86.h"
+static void setup_rtcd_internal(void)
+{
+ int flags = x86_simd_caps();
+
+ (void)flags;
+
+$(set_function_pointers c $ALL_ARCHS)
+}
+#endif
+$(common_bottom)
+EOF
+}
+
+arm() {
+ determine_indirection c $ALL_ARCHS
+
+ # Assign the helper variable for each enabled extension
+ for opt in $ALL_ARCHS; do
+ local uc=$(echo $opt | tr '[a-z]' '[A-Z]')
+ eval "have_${opt}=\"flags & HAS_${uc}\""
+ done
+
+ cat <<EOF
+$(common_top)
+#include "vpx_config.h"
+
+#ifdef RTCD_C
+#include "vpx_ports/arm.h"
+static void setup_rtcd_internal(void)
+{
+ int flags = arm_cpu_caps();
+
+ (void)flags;
+
+$(set_function_pointers c $ALL_ARCHS)
+}
+#endif
+$(common_bottom)
+EOF
+}
+
+
+mips() {
+ determine_indirection c $ALL_ARCHS
+ cat <<EOF
+$(common_top)
+#include "vpx_config.h"
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+$(set_function_pointers c $ALL_ARCHS)
+#if HAVE_DSPR2
+void dsputil_static_init();
+dsputil_static_init();
+#endif
+}
+#endif
+$(common_bottom)
+EOF
+}
+
+unoptimized() {
+ determine_indirection c
+ cat <<EOF
+$(common_top)
+#include "vpx_config.h"
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+$(set_function_pointers c)
+}
+#endif
+$(common_bottom)
+EOF
+
+}
+#
+# Main Driver
+#
+require c
+case $arch in
+ x86)
+ ALL_ARCHS=$(filter mmx sse sse2 sse3 ssse3 sse4_1)
+ x86
+ ;;
+ x86_64)
+ ALL_ARCHS=$(filter mmx sse sse2 sse3 ssse3 sse4_1)
+ REQUIRES=${REQUIRES:-mmx sse sse2}
+ require $(filter $REQUIRES)
+ x86
+ ;;
+ mips32)
+ ALL_ARCHS=$(filter mips32)
+ dspr2=$([ -f "$config_file" ] && eval echo $(grep HAVE_DSPR2 "$config_file"))
+ HAVE_DSPR2="${dspr2#*=}"
+ if [ "$HAVE_DSPR2" = "yes" ]; then
+ ALL_ARCHS=$(filter mips32 dspr2)
+ fi
+ mips
+ ;;
+ armv5te)
+ ALL_ARCHS=$(filter edsp)
+ arm
+ ;;
+ armv6)
+ ALL_ARCHS=$(filter edsp media)
+ arm
+ ;;
+ armv7)
+ ALL_ARCHS=$(filter edsp media neon)
+ arm
+ ;;
+ *)
+ unoptimized
+ ;;
+esac
diff --git a/build/make/version.sh b/libvpx/build/make/version.sh
similarity index 100%
rename from build/make/version.sh
rename to libvpx/build/make/version.sh
diff --git a/build/x86-msvs/obj_int_extract.bat b/libvpx/build/x86-msvs/obj_int_extract.bat
similarity index 100%
rename from build/x86-msvs/obj_int_extract.bat
rename to libvpx/build/x86-msvs/obj_int_extract.bat
diff --git a/build/x86-msvs/yasm.rules b/libvpx/build/x86-msvs/yasm.rules
similarity index 100%
rename from build/x86-msvs/yasm.rules
rename to libvpx/build/x86-msvs/yasm.rules
diff --git a/libvpx/configure b/libvpx/configure
new file mode 100755
index 0000000..b3c5fe9
--- /dev/null
+++ b/libvpx/configure
@@ -0,0 +1,625 @@
+#!/bin/bash
+##
+## configure
+##
+## This script is the front-end to the build system. It provides a similar
+## interface to standard configure scripts with some extra bits for dealing
+## with toolchains that differ from the standard POSIX interface and
+## for extracting subsets of the source tree. In theory, reusable parts
+## of this script were intended to live in build/make/configure.sh,
+## but in practice, the line is pretty blurry.
+##
+## This build system is based in part on the FFmpeg configure script.
+##
+
+#source_path="`dirname \"$0\"`"
+source_path=${0%/*}
+. "${source_path}/build/make/configure.sh"
+
+show_help(){
+ show_help_pre
+ cat << EOF
+Advanced options:
+ ${toggle_libs} libraries
+ ${toggle_examples} examples
+ ${toggle_docs} documentation
+ ${toggle_unit_tests} unit tests
+ --libc=PATH path to alternate libc
+ --as={yasm|nasm|auto} use specified assembler [auto, yasm preferred]
+ --sdk-path=PATH path to root of sdk (iOS, android builds only)
+ ${toggle_fast_unaligned} don't use unaligned accesses, even when
+ supported by hardware [auto]
+ ${toggle_codec_srcs} in/exclude codec library source code
+ ${toggle_debug_libs} in/exclude debug version of libraries
+ ${toggle_md5} support for output of checksum data
+ ${toggle_static_msvcrt} use static MSVCRT (VS builds only)
+ ${toggle_vp8} VP8 codec support
+ ${toggle_internal_stats} output of encoder internal stats for debug, if supported (encoders)
+ ${toggle_mem_tracker} track memory usage
+ ${toggle_postproc} postprocessing
+ ${toggle_multithread} multithreaded encoding and decoding
+ ${toggle_spatial_resampling} spatial sampling (scaling) support
+ ${toggle_realtime_only} enable this option while building for real-time encoding
+ ${toggle_onthefly_bitpacking} enable on-the-fly bitpacking in real-time encoding
+ ${toggle_error_concealment} enable this option to get a decoder which is able to conceal losses
+ ${toggle_runtime_cpu_detect} runtime cpu detection
+ ${toggle_shared} shared library support
+ ${toggle_static} static library support
+ ${toggle_small} favor smaller size over speed
+ ${toggle_postproc_visualizer} macro block / block level visualizers
+ ${toggle_multi_res_encoding} enable multiple-resolution encoding
+ ${toggle_temporal_denoising} enable temporal denoising and disable the spatial denoiser
+
+Codecs:
+ Codecs can be selectively enabled or disabled individually, or by family:
+ --disable-<codec>
+ is equivalent to:
+ --disable-<codec>-encoder
+ --disable-<codec>-decoder
+
+ Codecs available in this distribution:
+EOF
+#restore editor state '
+
+ local family;
+ local last_family;
+ local c;
+ local str;
+ for c in ${CODECS}; do
+ family=${c%_*}
+ if [ "${family}" != "${last_family}" ]; then
+ [ -z "${str}" ] || echo "${str}"
+ str="$(printf ' %10s:' ${family})"
+ fi
+ str="${str} $(printf '%10s' ${c#*_})"
+ last_family=${family}
+ done
+ echo "${str}"
+ show_help_post
+}
+
+##
+## BEGIN APPLICATION SPECIFIC CONFIGURATION
+##
+
+# all_platforms is a list of all supported target platforms. Maintain
+# alphabetically by architecture, generic-gnu last.
+all_platforms="${all_platforms} armv5te-android-gcc"
+all_platforms="${all_platforms} armv5te-linux-rvct"
+all_platforms="${all_platforms} armv5te-linux-gcc"
+all_platforms="${all_platforms} armv5te-none-rvct"
+all_platforms="${all_platforms} armv6-darwin-gcc"
+all_platforms="${all_platforms} armv6-linux-rvct"
+all_platforms="${all_platforms} armv6-linux-gcc"
+all_platforms="${all_platforms} armv6-none-rvct"
+all_platforms="${all_platforms} armv7-android-gcc" #neon Cortex-A8
+all_platforms="${all_platforms} armv7-darwin-gcc" #neon Cortex-A8
+all_platforms="${all_platforms} armv7-linux-rvct" #neon Cortex-A8
+all_platforms="${all_platforms} armv7-linux-gcc" #neon Cortex-A8
+all_platforms="${all_platforms} armv7-none-rvct" #neon Cortex-A8
+all_platforms="${all_platforms} mips32-linux-gcc"
+all_platforms="${all_platforms} ppc32-darwin8-gcc"
+all_platforms="${all_platforms} ppc32-darwin9-gcc"
+all_platforms="${all_platforms} ppc32-linux-gcc"
+all_platforms="${all_platforms} ppc64-darwin8-gcc"
+all_platforms="${all_platforms} ppc64-darwin9-gcc"
+all_platforms="${all_platforms} ppc64-linux-gcc"
+all_platforms="${all_platforms} sparc-solaris-gcc"
+all_platforms="${all_platforms} x86-darwin8-gcc"
+all_platforms="${all_platforms} x86-darwin8-icc"
+all_platforms="${all_platforms} x86-darwin9-gcc"
+all_platforms="${all_platforms} x86-darwin9-icc"
+all_platforms="${all_platforms} x86-darwin10-gcc"
+all_platforms="${all_platforms} x86-darwin11-gcc"
+all_platforms="${all_platforms} x86-darwin12-gcc"
+all_platforms="${all_platforms} x86-linux-gcc"
+all_platforms="${all_platforms} x86-linux-icc"
+all_platforms="${all_platforms} x86-os2-gcc"
+all_platforms="${all_platforms} x86-solaris-gcc"
+all_platforms="${all_platforms} x86-win32-gcc"
+all_platforms="${all_platforms} x86-win32-vs7"
+all_platforms="${all_platforms} x86-win32-vs8"
+all_platforms="${all_platforms} x86-win32-vs9"
+all_platforms="${all_platforms} x86_64-darwin9-gcc"
+all_platforms="${all_platforms} x86_64-darwin10-gcc"
+all_platforms="${all_platforms} x86_64-darwin11-gcc"
+all_platforms="${all_platforms} x86_64-darwin12-gcc"
+all_platforms="${all_platforms} x86_64-linux-gcc"
+all_platforms="${all_platforms} x86_64-linux-icc"
+all_platforms="${all_platforms} x86_64-solaris-gcc"
+all_platforms="${all_platforms} x86_64-win64-gcc"
+all_platforms="${all_platforms} x86_64-win64-vs8"
+all_platforms="${all_platforms} x86_64-win64-vs9"
+all_platforms="${all_platforms} universal-darwin8-gcc"
+all_platforms="${all_platforms} universal-darwin9-gcc"
+all_platforms="${all_platforms} universal-darwin10-gcc"
+all_platforms="${all_platforms} universal-darwin11-gcc"
+all_platforms="${all_platforms} universal-darwin12-gcc"
+all_platforms="${all_platforms} generic-gnu"
+
+# all_targets is a list of all targets that can be configured
+# note that these should be in dependency order for now.
+all_targets="libs examples docs"
+
+# all targets available are enabled, by default.
+for t in ${all_targets}; do
+ [ -f ${source_path}/${t}.mk ] && enable ${t}
+done
+
+# check installed doxygen version
+doxy_version=$(doxygen --version 2>/dev/null)
+doxy_major=${doxy_version%%.*}
+if [ ${doxy_major:-0} -ge 1 ]; then
+ doxy_version=${doxy_version#*.}
+ doxy_minor=${doxy_version%%.*}
+ doxy_patch=${doxy_version##*.}
+
+ [ $doxy_major -gt 1 ] && enable doxygen
+ [ $doxy_minor -gt 5 ] && enable doxygen
+ [ $doxy_minor -eq 5 ] && [ $doxy_patch -ge 3 ] && enable doxygen
+fi
+
+# install everything except the sources, by default. sources will have
+# to be enabled when doing dist builds, since that's no longer a common
+# case.
+enabled doxygen && php -v >/dev/null 2>&1 && enable install_docs
+enable install_bins
+enable install_libs
+
+enable static
+enable optimizations
+enable fast_unaligned #allow unaligned accesses, if supported by hw
+enable md5
+enable spatial_resampling
+enable multithread
+enable os_support
+enable temporal_denoising
+
+[ -d ${source_path}/../include ] && enable alt_tree_layout
+for d in vp8; do
+ [ -d ${source_path}/${d} ] && disable alt_tree_layout;
+done
+
+if ! enabled alt_tree_layout; then
+# development environment
+[ -d ${source_path}/vp8 ] && CODECS="${CODECS} vp8_encoder vp8_decoder"
+else
+# customer environment
+[ -f ${source_path}/../include/vpx/vp8cx.h ] && CODECS="${CODECS} vp8_encoder"
+[ -f ${source_path}/../include/vpx/vp8dx.h ] && CODECS="${CODECS} vp8_decoder"
+[ -f ${source_path}/../include/vpx/vp8cx.h ] || disable vp8_encoder
+[ -f ${source_path}/../include/vpx/vp8dx.h ] || disable vp8_decoder
+
+[ -f ${source_path}/../lib/*/*mt.lib ] && soft_enable static_msvcrt
+fi
+
+CODECS="$(echo ${CODECS} | tr ' ' '\n')"
+CODEC_FAMILIES="$(for c in ${CODECS}; do echo ${c%_*}; done | sort | uniq)"
+
+ARCH_LIST="
+ arm
+ mips
+ x86
+ x86_64
+ ppc32
+ ppc64
+"
+ARCH_EXT_LIST="
+ edsp
+ media
+ neon
+
+ mips32
+ dspr2
+
+ mmx
+ sse
+ sse2
+ sse3
+ ssse3
+ sse4_1
+
+ altivec
+"
+HAVE_LIST="
+ ${ARCH_EXT_LIST}
+ vpx_ports
+ stdint_h
+ alt_tree_layout
+ pthread_h
+ sys_mman_h
+ unistd_h
+"
+CONFIG_LIST="
+ external_build
+ install_docs
+ install_bins
+ install_libs
+ install_srcs
+ debug
+ gprof
+ gcov
+ rvct
+ gcc
+ msvs
+ pic
+ big_endian
+
+ codec_srcs
+ debug_libs
+ fast_unaligned
+ mem_manager
+ mem_tracker
+ mem_checks
+ md5
+
+ dequant_tokens
+ dc_recon
+ runtime_cpu_detect
+ postproc
+ multithread
+ internal_stats
+ ${CODECS}
+ ${CODEC_FAMILIES}
+ encoders
+ decoders
+ static_msvcrt
+ spatial_resampling
+ realtime_only
+ onthefly_bitpacking
+ error_concealment
+ shared
+ static
+ small
+ postproc_visualizer
+ os_support
+ unit_tests
+ multi_res_encoding
+ temporal_denoising
+"
+CMDLINE_SELECT="
+ extra_warnings
+ werror
+ install_docs
+ install_bins
+ install_libs
+ install_srcs
+ debug
+ gprof
+ gcov
+ pic
+ optimizations
+ ccache
+ runtime_cpu_detect
+
+ libs
+ examples
+ docs
+ libc
+ as
+ fast_unaligned
+ codec_srcs
+ debug_libs
+ md5
+
+ dequant_tokens
+ dc_recon
+ postproc
+ multithread
+ internal_stats
+ ${CODECS}
+ ${CODEC_FAMILIES}
+ static_msvcrt
+ mem_tracker
+ spatial_resampling
+ realtime_only
+ onthefly_bitpacking
+ error_concealment
+ shared
+ static
+ small
+ postproc_visualizer
+ unit_tests
+ multi_res_encoding
+ temporal_denoising
+"
+
+process_cmdline() {
+ for opt do
+ optval="${opt#*=}"
+ case "$opt" in
+ --disable-codecs) for c in ${CODECS}; do disable $c; done ;;
+ *) process_common_cmdline "$opt"
+ ;;
+ esac
+ done
+}
+
+post_process_cmdline() {
+ local c
+
+ # If the codec family is disabled, disable all components of that family.
+ # If the codec family is enabled, enable all components of that family.
+ log_echo "Configuring selected codecs"
+ for c in ${CODECS}; do
+ disabled ${c%%_*} && disable ${c}
+ enabled ${c%%_*} && enable ${c}
+ done
+
+ # Enable all detected codecs, if they haven't been disabled
+ for c in ${CODECS}; do soft_enable $c; done
+
+ # Enable the codec family if any component of that family is enabled
+ for c in ${CODECS}; do
+ enabled $c && enable ${c%_*}
+ done
+
+ # Set the {en,de}coders variable if any algorithm in that class is enabled
+ for c in ${CODECS}; do
+ enabled ${c} && enable ${c##*_}s
+ done
+}
+
+
+process_targets() {
+ enabled child || write_common_config_banner
+ enabled universal || write_common_target_config_h ${BUILD_PFX}vpx_config.h
+
+ # TODO: add host tools target (obj_int_extract, etc)
+
+ # For fat binaries, call configure recursively to configure for each
+ # binary architecture to be included.
+ if enabled universal; then
+ # Call configure (ourselves) for each subarchitecture
+ for arch in $fat_bin_archs; do
+ BUILD_PFX=${arch}/ toolchain=${arch} $self --child $cmdline_args || exit $?
+ done
+ fi
+
+ # The write_common_config (config.mk) logic is deferred until after the
+ # recursive calls to configure complete, becuase we want our universal
+ # targets to be executed last.
+ write_common_config_targets
+ enabled universal && echo "FAT_ARCHS=${fat_bin_archs}" >> config.mk
+
+ # Calculate the default distribution name, based on the enabled features
+ local cf
+ local DIST_DIR=vpx
+ for cf in $CODEC_FAMILIES; do
+ if enabled ${cf}_encoder && enabled ${cf}_decoder; then
+ DIST_DIR="${DIST_DIR}-${cf}"
+ elif enabled ${cf}_encoder; then
+ DIST_DIR="${DIST_DIR}-${cf}cx"
+ elif enabled ${cf}_decoder; then
+ DIST_DIR="${DIST_DIR}-${cf}dx"
+ fi
+ done
+ enabled debug_libs && DIST_DIR="${DIST_DIR}-debug"
+ enabled codec_srcs && DIST_DIR="${DIST_DIR}-src"
+ ! enabled postproc && DIST_DIR="${DIST_DIR}-nopost"
+ ! enabled multithread && DIST_DIR="${DIST_DIR}-nomt"
+ ! enabled install_docs && DIST_DIR="${DIST_DIR}-nodocs"
+ DIST_DIR="${DIST_DIR}-${tgt_isa}-${tgt_os}"
+ case "${tgt_os}" in
+ win*) enabled static_msvcrt && DIST_DIR="${DIST_DIR}mt" || DIST_DIR="${DIST_DIR}md"
+ DIST_DIR="${DIST_DIR}-${tgt_cc}"
+ ;;
+ esac
+ if [ -f "${source_path}/build/make/version.sh" ]; then
+ local ver=`"$source_path/build/make/version.sh" --bare $source_path`
+ DIST_DIR="${DIST_DIR}-${ver}"
+ VERSION_STRING=${ver}
+ ver=${ver%%-*}
+ VERSION_PATCH=${ver##*.}
+ ver=${ver%.*}
+ VERSION_MINOR=${ver##*.}
+ ver=${ver#v}
+ VERSION_MAJOR=${ver%.*}
+ fi
+ enabled child || cat <<EOF >> config.mk
+
+PREFIX=${prefix}
+ifeq (\$(MAKECMDGOALS),dist)
+DIST_DIR?=${DIST_DIR}
+else
+DIST_DIR?=\$(DESTDIR)${prefix}
+endif
+LIBSUBDIR=${libdir##${prefix}/}
+
+VERSION_STRING=${VERSION_STRING}
+
+VERSION_MAJOR=${VERSION_MAJOR}
+VERSION_MINOR=${VERSION_MINOR}
+VERSION_PATCH=${VERSION_PATCH}
+
+CONFIGURE_ARGS=${CONFIGURE_ARGS}
+EOF
+ enabled child || echo "CONFIGURE_ARGS?=${CONFIGURE_ARGS}" >> config.mk
+
+ #
+ # Write makefiles for all enabled targets
+ #
+ for tgt in libs examples docs solution; do
+ local tgt_fn="$tgt-$toolchain.mk"
+
+ if enabled $tgt; then
+ echo "Creating makefiles for ${toolchain} ${tgt}"
+ write_common_target_config_mk $tgt_fn ${BUILD_PFX}vpx_config.h
+ #write_${tgt}_config
+ fi
+ done
+
+}
+
+process_detect() {
+ if enabled shared; then
+ # Can only build shared libs on a subset of platforms. Doing this check
+ # here rather than at option parse time because the target auto-detect
+ # magic happens after the command line has been parsed.
+ if ! enabled linux; then
+ if enabled gnu; then
+ echo "--enable-shared is only supported on ELF; assuming this is OK"
+ else
+ die "--enable-shared only supported on ELF for now"
+ fi
+ fi
+ fi
+ if [ -z "$CC" ]; then
+ echo "Bypassing toolchain for environment detection."
+ enable external_build
+ check_header() {
+ log fake_check_header "$@"
+ header=$1
+ shift
+ var=`echo $header | sed 's/[^A-Za-z0-9_]/_/g'`
+ disable $var
+ case $header in
+ stdio.h)
+ true;
+ ;;
+ *)
+ local result=false
+ for d in "$@"; do
+ [ -f "${d##-I}/$header" ] && result=true && break
+ done
+ ${result:-true}
+ esac && enable $var
+ }
+ check_ld() {
+ true
+ }
+ fi
+ check_header stdio.h || die "Unable to invoke compiler: ${CC} ${CFLAGS}"
+ check_ld <<EOF || die "Toolchain is unable to link executables"
+int main(void) {return 0;}
+EOF
+ # check system headers
+ check_header stdint.h
+ check_header pthread.h
+ check_header sys/mman.h
+
+ check_header vpx/vpx_integer.h -I${source_path} && enable vpx_ports
+}
+
+process_toolchain() {
+ process_common_toolchain
+
+ # Handle universal binaries for this architecture
+ case $toolchain in
+ universal-darwin*)
+ local darwin_ver=${tgt_os##darwin}
+
+ # Snow Leopard (10.6/darwin10) dropped support for PPC
+ # Include PPC support for all prior versions
+ if [ $darwin_ver -lt 10 ]; then
+ fat_bin_archs="$fat_bin_archs ppc32-${tgt_os}-gcc"
+ fi
+
+ # Tiger (10.4/darwin8) brought support for x86
+ if [ $darwin_ver -ge 8 ]; then
+ fat_bin_archs="$fat_bin_archs x86-${tgt_os}-${tgt_cc}"
+ fi
+
+ # Leopard (10.5/darwin9) brought 64 bit support
+ if [ $darwin_ver -ge 9 ]; then
+ fat_bin_archs="$fat_bin_archs x86_64-${tgt_os}-${tgt_cc}"
+ fi
+ ;;
+ esac
+
+
+ # Enable some useful compiler flags
+ if enabled gcc; then
+ enabled werror && check_add_cflags -Werror
+ check_add_cflags -Wall
+ check_add_cflags -Wdeclaration-after-statement
+ check_add_cflags -Wdisabled-optimization
+ check_add_cflags -Wpointer-arith
+ check_add_cflags -Wtype-limits
+ check_add_cflags -Wcast-qual
+ check_add_cflags -Wimplicit-function-declaration
+ check_add_cflags -Wuninitialized
+ check_add_cflags -Wunused-variable
+ check_add_cflags -Wunused-but-set-variable
+ enabled extra_warnings || check_add_cflags -Wno-unused-function
+ fi
+
+ if enabled icc; then
+ enabled werror && check_add_cflags -Werror
+ check_add_cflags -Wall
+ check_add_cflags -Wpointer-arith
+
+ # ICC has a number of floating point optimizations that we disable
+ # in favor of deterministic output WRT to other compilers
+ add_cflags -fp-model precise
+ fi
+
+ # Enable extra, harmless warnings. These might provide additional insight
+ # to what the compiler is doing and why, but in general, but they shouldn't
+ # be treated as fatal, even if we're treating warnings as errors.
+ GCC_EXTRA_WARNINGS="
+ -Wdisabled-optimization
+ -Winline
+ "
+ enabled gcc && EXTRA_WARNINGS="${GCC_EXTRA_WARNINGS}"
+ RVCT_EXTRA_WARNINGS="
+ --remarks
+ "
+ enabled rvct && EXTRA_WARNINGS="${RVCT_EXTRA_WARNINGS}"
+ if enabled extra_warnings; then
+ for w in ${EXTRA_WARNINGS}; do
+ check_add_cflags ${w}
+ enabled gcc && enabled werror && check_add_cflags -Wno-error=${w}
+ done
+ fi
+
+ # ccache only really works on gcc toolchains
+ enabled gcc || soft_disable ccache
+ if enabled mips; then
+ enable dequant_tokens
+ enable dc_recon
+ fi
+
+ # Enable the postbuild target if building for visual studio.
+ case "$tgt_cc" in
+ vs*) enable msvs
+ enable solution
+ vs_version=${tgt_cc##vs}
+ all_targets="${all_targets} solution"
+ ;;
+ esac
+
+ # Other toolchain specific defaults
+ case $toolchain in x86*|ppc*|universal*) soft_enable postproc;; esac
+
+ if enabled postproc_visualizer; then
+ enabled postproc || die "postproc_visualizer requires postproc to be enabled"
+ fi
+
+ # Enable unit tests if we have a working C++ compiler
+ case "$toolchain" in
+ *-vs*)
+ soft_enable unit_tests
+ ;;
+ *-android-*)
+ # GTestLog must be modified to use Android logging utilities.
+ ;;
+ *)
+ check_cxx "$@" <<EOF && soft_enable unit_tests
+int z;
+EOF
+ ;;
+ esac
+}
+
+
+##
+## END APPLICATION SPECIFIC CONFIGURATION
+##
+CONFIGURE_ARGS="$@"
+process "$@"
+print_webm_license ${BUILD_PFX}vpx_config.c "/*" " */"
+cat <<EOF >> ${BUILD_PFX}vpx_config.c
+static const char* const cfg = "$CONFIGURE_ARGS";
+const char *vpx_codec_build_config(void) {return cfg;}
+EOF
diff --git a/libvpx/docs.mk b/libvpx/docs.mk
new file mode 100644
index 0000000..cfe57ed
--- /dev/null
+++ b/libvpx/docs.mk
@@ -0,0 +1,53 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+INSTALL_MAPS += docs/% docs/%
+INSTALL_MAPS += src/% %
+INSTALL_MAPS += % %
+
+# Static documentation authored in doxygen
+CODEC_DOX := mainpage.dox \
+ keywords.dox \
+ usage.dox \
+ usage_cx.dox \
+ usage_dx.dox \
+
+# Other doxy files sourced in Markdown
+TXT_DOX = $(call enabled,TXT_DOX)
+
+%.dox: %.txt
+ @echo " [DOXY] $@"
+ @$(SRC_PATH_BARE)/examples/gen_example_doxy.php \
+ $(@:.dox=) "$($@.DESC)" > $@ < $<
+
+
+EXAMPLE_PATH += $(SRC_PATH_BARE) #for CHANGELOG, README, etc
+
+doxyfile: $(if $(findstring examples, $(ALL_TARGETS)),examples.doxy)
+doxyfile: libs.doxy_template libs.doxy
+ @echo " [CREATE] $@"
+ @cat $^ > $@
+ @echo "STRIP_FROM_PATH += $(SRC_PATH_BARE) $(BUILD_ROOT)" >> $@
+ @echo "INPUT += $(addprefix $(SRC_PATH_BARE)/,$(CODEC_DOX))" >> $@;
+ @echo "INPUT += $(TXT_DOX)" >> $@;
+ @echo "EXAMPLE_PATH += $(EXAMPLE_PATH)" >> $@
+
+CLEAN-OBJS += doxyfile $(wildcard docs/html/*)
+docs/html/index.html: doxyfile $(CODEC_DOX) $(TXT_DOX)
+ @echo " [DOXYGEN] $<"
+ @doxygen $<
+DOCS-yes += docs/html/index.html
+
+DIST-DOCS-yes = $(wildcard docs/html/*)
+DIST-DOCS-$(CONFIG_CODEC_SRCS) += $(addprefix src/,$(CODEC_DOX))
+DIST-DOCS-$(CONFIG_CODEC_SRCS) += src/libs.doxy_template
+DIST-DOCS-yes += CHANGELOG
+DIST-DOCS-yes += README
diff --git a/example_xma.c b/libvpx/example_xma.c
similarity index 100%
rename from example_xma.c
rename to libvpx/example_xma.c
diff --git a/libvpx/examples.mk b/libvpx/examples.mk
new file mode 100644
index 0000000..90913e6
--- /dev/null
+++ b/libvpx/examples.mk
@@ -0,0 +1,285 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+# List of examples to build. UTILS are files that are taken from the source
+# tree directly, and GEN_EXAMPLES are files that are created from the
+# examples folder.
+UTILS-$(CONFIG_DECODERS) += vpxdec.c
+vpxdec.SRCS += md5_utils.c md5_utils.h
+vpxdec.SRCS += vpx_ports/vpx_timer.h
+vpxdec.SRCS += vpx/vpx_integer.h
+vpxdec.SRCS += args.c args.h
+vpxdec.SRCS += tools_common.c tools_common.h
+vpxdec.SRCS += nestegg/halloc/halloc.h
+vpxdec.SRCS += nestegg/halloc/src/align.h
+vpxdec.SRCS += nestegg/halloc/src/halloc.c
+vpxdec.SRCS += nestegg/halloc/src/hlist.h
+vpxdec.SRCS += nestegg/halloc/src/macros.h
+vpxdec.SRCS += nestegg/include/nestegg/nestegg.h
+vpxdec.SRCS += nestegg/src/nestegg.c
+vpxdec.GUID = BA5FE66F-38DD-E034-F542-B1578C5FB950
+vpxdec.DESCRIPTION = Full featured decoder
+UTILS-$(CONFIG_ENCODERS) += vpxenc.c
+vpxenc.SRCS += args.c args.h y4minput.c y4minput.h
+vpxenc.SRCS += tools_common.c tools_common.h
+vpxenc.SRCS += vpx_ports/mem_ops.h
+vpxenc.SRCS += vpx_ports/mem_ops_aligned.h
+vpxenc.SRCS += vpx_ports/vpx_timer.h
+vpxenc.SRCS += libmkv/EbmlIDs.h
+vpxenc.SRCS += libmkv/EbmlWriter.c
+vpxenc.SRCS += libmkv/EbmlWriter.h
+vpxenc.GUID = 548DEC74-7A15-4B2B-AFC3-AA102E7C25C1
+vpxenc.DESCRIPTION = Full featured encoder
+UTILS-$(CONFIG_ENCODERS) += vp8_scalable_patterns.c
+vp8_scalable_patterns.GUID = 0D6A210B-F482-4D6F-8570-4A9C01ACC88C
+vp8_scalable_patterns.DESCRIPTION = Temporal Scalability Encoder
+
+# Clean up old ivfenc, ivfdec binaries.
+ifeq ($(CONFIG_MSVS),yes)
+CLEAN-OBJS += $(foreach p,$(VS_PLATFORMS),$(p)/Release/ivfenc.exe)
+CLEAN-OBJS += $(foreach p,$(VS_PLATFORMS),$(p)/Release/ivfdec.exe)
+else
+CLEAN-OBJS += ivfenc{.c.o,.c.d,.dox,.exe,}
+CLEAN-OBJS += ivfdec{.c.o,.c.d,.dox,.exe,}
+endif
+
+# XMA example disabled for now, not used in VP8
+#UTILS-$(CONFIG_DECODERS) += example_xma.c
+#example_xma.GUID = A955FC4A-73F1-44F7-135E-30D84D32F022
+#example_xma.DESCRIPTION = External Memory Allocation mode usage
+
+GEN_EXAMPLES-$(CONFIG_DECODERS) += simple_decoder.c
+simple_decoder.GUID = D3BBF1E9-2427-450D-BBFF-B2843C1D44CC
+simple_decoder.DESCRIPTION = Simplified decoder loop
+GEN_EXAMPLES-$(CONFIG_DECODERS) += postproc.c
+postproc.GUID = 65E33355-F35E-4088-884D-3FD4905881D7
+postproc.DESCRIPTION = Decoder postprocessor control
+GEN_EXAMPLES-$(CONFIG_DECODERS) += decode_to_md5.c
+decode_to_md5.SRCS += md5_utils.h md5_utils.c
+decode_to_md5.GUID = 59120B9B-2735-4BFE-B022-146CA340FE42
+decode_to_md5.DESCRIPTION = Frame by frame MD5 checksum
+
+GEN_EXAMPLES-$(CONFIG_ENCODERS) += simple_encoder.c
+simple_encoder.GUID = 4607D299-8A71-4D2C-9B1D-071899B6FBFD
+simple_encoder.DESCRIPTION = Simplified encoder loop
+GEN_EXAMPLES-$(CONFIG_ENCODERS) += twopass_encoder.c
+twopass_encoder.GUID = 73494FA6-4AF9-4763-8FBB-265C92402FD8
+twopass_encoder.DESCRIPTION = Two-pass encoder loop
+GEN_EXAMPLES-$(CONFIG_ENCODERS) += force_keyframe.c
+force_keyframe.GUID = 3C67CADF-029F-4C86-81F5-D6D4F51177F0
+force_keyframe.DESCRIPTION = Force generation of keyframes
+ifeq ($(CONFIG_DECODERS),yes)
+GEN_EXAMPLES-$(CONFIG_ENCODERS) += decode_with_drops.c
+endif
+decode_with_drops.GUID = CE5C53C4-8DDA-438A-86ED-0DDD3CDB8D26
+decode_with_drops.DESCRIPTION = Drops frames while decoding
+ifeq ($(CONFIG_DECODERS),yes)
+GEN_EXAMPLES-$(CONFIG_ERROR_CONCEALMENT) += decode_with_partial_drops.c
+endif
+decode_with_partial_drops.GUID = 61C2D026-5754-46AC-916F-1343ECC5537E
+decode_with_partial_drops.DESCRIPTION = Drops parts of frames while decoding
+GEN_EXAMPLES-$(CONFIG_ENCODERS) += error_resilient.c
+error_resilient.GUID = DF5837B9-4145-4F92-A031-44E4F832E00C
+error_resilient.DESCRIPTION = Error Resiliency Feature
+
+GEN_EXAMPLES-$(CONFIG_VP8_ENCODER) += vp8_set_maps.c
+vp8_set_maps.GUID = ECB2D24D-98B8-4015-A465-A4AF3DCC145F
+vp8_set_maps.DESCRIPTION = VP8 set active and ROI maps
+GEN_EXAMPLES-$(CONFIG_VP8_ENCODER) += vp8cx_set_ref.c
+vp8cx_set_ref.GUID = C5E31F7F-96F6-48BD-BD3E-10EBF6E8057A
+vp8cx_set_ref.DESCRIPTION = VP8 set encoder reference frame
+
+# C file is provided, not generated automatically.
+UTILS-$(CONFIG_MULTI_RES_ENCODING) += vp8_multi_resolution_encoder.c
+vp8_multi_resolution_encoder.SRCS \
+ += third_party/libyuv/include/libyuv/basic_types.h \
+ third_party/libyuv/include/libyuv/cpu_id.h \
+ third_party/libyuv/include/libyuv/scale.h \
+ third_party/libyuv/source/row.h \
+ third_party/libyuv/source/scale.c \
+ third_party/libyuv/source/cpu_id.c
+vp8_multi_resolution_encoder.GUID = 04f8738e-63c8-423b-90fa-7c2703a374de
+vp8_multi_resolution_encoder.DESCRIPTION = VP8 Multiple-resolution Encoding
+
+# Handle extra library flags depending on codec configuration
+
+# We should not link to math library (libm) on RVCT
+# when building for bare-metal targets
+ifeq ($(CONFIG_OS_SUPPORT), yes)
+CODEC_EXTRA_LIBS-$(CONFIG_VP8) += m
+else
+ ifeq ($(CONFIG_GCC), yes)
+ CODEC_EXTRA_LIBS-$(CONFIG_VP8) += m
+ endif
+endif
+#
+# End of specified files. The rest of the build rules should happen
+# automagically from here.
+#
+
+
+# Examples need different flags based on whether we're building
+# from an installed tree or a version controlled tree. Determine
+# the proper paths.
+ifeq ($(HAVE_ALT_TREE_LAYOUT),yes)
+ LIB_PATH := $(SRC_PATH_BARE)/../lib
+ INC_PATH := $(SRC_PATH_BARE)/../include
+else
+ LIB_PATH-yes += $(if $(BUILD_PFX),$(BUILD_PFX),.)
+ INC_PATH-$(CONFIG_VP8_DECODER) += $(SRC_PATH_BARE)/vp8
+ INC_PATH-$(CONFIG_VP8_ENCODER) += $(SRC_PATH_BARE)/vp8
+ LIB_PATH := $(call enabled,LIB_PATH)
+ INC_PATH := $(call enabled,INC_PATH)
+endif
+INTERNAL_CFLAGS = $(addprefix -I,$(INC_PATH))
+INTERNAL_LDFLAGS += $(addprefix -L,$(LIB_PATH))
+
+
+# Expand list of selected examples to build (as specified above)
+UTILS = $(call enabled,UTILS)
+GEN_EXAMPLES = $(call enabled,GEN_EXAMPLES)
+ALL_EXAMPLES = $(UTILS) $(GEN_EXAMPLES)
+UTIL_SRCS = $(foreach ex,$(UTILS),$($(ex:.c=).SRCS))
+ALL_SRCS = $(foreach ex,$(ALL_EXAMPLES),$($(ex:.c=).SRCS))
+CODEC_EXTRA_LIBS=$(sort $(call enabled,CODEC_EXTRA_LIBS))
+
+
+# Expand all example sources into a variable containing all sources
+# for that example (not just them main one specified in UTILS/GEN_EXAMPLES)
+# and add this file to the list (for MSVS workspace generation)
+$(foreach ex,$(ALL_EXAMPLES),$(eval $(ex:.c=).SRCS += $(ex) examples.mk))
+
+
+# If this is a universal (fat) binary, then all the subarchitectures have
+# already been built and our job is to stitch them together. The
+# BUILD_OBJS variable indicates whether we should be building
+# (compiling, linking) the library. The LIPO_OBJS variable indicates
+# that we're stitching.
+$(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_OBJS,BUILD_OBJS):=yes)
+
+
+# Create build/install dependencies for all examples. The common case
+# is handled here. The MSVS case is handled below.
+NOT_MSVS = $(if $(CONFIG_MSVS),,yes)
+DIST-BINS-$(NOT_MSVS) += $(addprefix bin/,$(ALL_EXAMPLES:.c=$(EXE_SFX)))
+INSTALL-BINS-$(NOT_MSVS) += $(addprefix bin/,$(UTILS:.c=$(EXE_SFX)))
+DIST-SRCS-yes += $(ALL_SRCS)
+INSTALL-SRCS-yes += $(UTIL_SRCS)
+OBJS-$(NOT_MSVS) += $(if $(BUILD_OBJS),$(call objs,$(ALL_SRCS)))
+BINS-$(NOT_MSVS) += $(addprefix $(BUILD_PFX),$(ALL_EXAMPLES:.c=$(EXE_SFX)))
+
+
+# Instantiate linker template for all examples.
+CODEC_LIB=$(if $(CONFIG_DEBUG_LIBS),vpx_g,vpx)
+CODEC_LIB_SUF=$(if $(CONFIG_SHARED),.so,.a)
+$(foreach bin,$(BINS-yes),\
+ $(if $(BUILD_OBJS),$(eval $(bin):\
+ $(LIB_PATH)/lib$(CODEC_LIB)$(CODEC_LIB_SUF)))\
+ $(if $(BUILD_OBJS),$(eval $(call linker_template,$(bin),\
+ $(call objs,$($(notdir $(bin:$(EXE_SFX)=)).SRCS)) \
+ -l$(CODEC_LIB) $(addprefix -l,$(CODEC_EXTRA_LIBS))\
+ )))\
+ $(if $(LIPO_OBJS),$(eval $(call lipo_bin_template,$(bin))))\
+ )
+
+
+# Rules to generate the GEN_EXAMPLES sources
+.PRECIOUS: %.c
+CLEAN-OBJS += $(GEN_EXAMPLES)
+%.c: examples/%.txt
+ @echo " [EXAMPLE] $@"
+ @$(SRC_PATH_BARE)/examples/gen_example_code.sh $< > $@
+
+
+# The following pairs define a mapping of locations in the distribution
+# tree to locations in the source/build trees.
+INSTALL_MAPS += src/%.c %.c
+INSTALL_MAPS += src/% $(SRC_PATH_BARE)/%
+INSTALL_MAPS += bin/% %
+INSTALL_MAPS += % %
+
+
+# Set up additional MSVS environment
+ifeq ($(CONFIG_MSVS),yes)
+CODEC_LIB=$(if $(CONFIG_STATIC_MSVCRT),vpxmt,vpxmd)
+# This variable uses deferred expansion intentionally, since the results of
+# $(wildcard) may change during the course of the Make.
+VS_PLATFORMS = $(foreach d,$(wildcard */Release/$(CODEC_LIB).lib),$(word 1,$(subst /, ,$(d))))
+INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),bin/$(p)/% $(p)/Release/%)
+endif
+
+# Build Visual Studio Projects. We use a template here to instantiate
+# explicit rules rather than using an implicit rule because we want to
+# leverage make's VPATH searching rather than specifying the paths on
+# each file in ALL_EXAMPLES. This has the unfortunate side effect that
+# touching the source files trigger a rebuild of the project files
+# even though there is no real dependency there (the dependency is on
+# the makefiles). We may want to revisit this.
+define vcproj_template
+$(1): $($(1:.vcproj=).SRCS)
+ @echo " [vcproj] $$@"
+ $$(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh\
+ --exe\
+ --target=$$(TOOLCHAIN)\
+ --name=$$(@:.vcproj=)\
+ --ver=$$(CONFIG_VS_VERSION)\
+ --proj-guid=$$($$(@:.vcproj=).GUID)\
+ $$(if $$(CONFIG_STATIC_MSVCRT),--static-crt) \
+ --out=$$@ $$(INTERNAL_CFLAGS) $$(CFLAGS) \
+ $$(INTERNAL_LDFLAGS) $$(LDFLAGS) -l$$(CODEC_LIB) -lwinmm $$^
+endef
+PROJECTS-$(CONFIG_MSVS) += $(ALL_EXAMPLES:.c=.vcproj)
+INSTALL-BINS-$(CONFIG_MSVS) += $(foreach p,$(VS_PLATFORMS),\
+ $(addprefix bin/$(p)/,$(ALL_EXAMPLES:.c=.exe)))
+$(foreach proj,$(call enabled,PROJECTS),\
+ $(eval $(call vcproj_template,$(proj))))
+
+
+
+#
+# Documentation Rules
+#
+%.dox: examples/%.txt
+ @echo " [DOXY] $@"
+ @$(SRC_PATH_BARE)/examples/gen_example_text.sh $< | \
+ $(SRC_PATH_BARE)/examples/gen_example_doxy.php \
+ example_$(@:.dox=) $(@:.dox=.c) > $@
+
+%.dox: %.c
+ @echo " [DOXY] $@"
+ @echo "/*!\page example_$(@:.dox=) $(@:.dox=)" > $@
+ @echo " \includelineno $(notdir $<)" >> $@
+ @echo "*/" >> $@
+
+samples.dox: examples.mk
+ @echo " [DOXY] $@"
+ @echo "/*!\page samples Sample Code" > $@
+ @echo " This SDK includes a number of sample applications."\
+ "each sample documents a feature of the SDK in both prose"\
+ "and the associated C code. In general, later samples"\
+ "build upon prior samples, so it is best to work through the"\
+ "list in order. The following samples are included: ">>$@
+ @$(foreach ex,$(GEN_EXAMPLES:.c=),\
+ echo " - \subpage example_$(ex) $($(ex).DESCRIPTION)" >> $@;)
+ @echo >> $@
+ @echo " In addition, the SDK contains a number of utilities."\
+ "Since these utilities are built upon the concepts described"\
+ "in the sample code listed above, they are not documented in"\
+ "pieces like the samples are. Thir sourcre is included here"\
+ "for reference. The following utilities are included:" >> $@
+ @$(foreach ex,$(UTILS:.c=),\
+ echo " - \subpage example_$(ex) $($(ex).DESCRIPTION)" >> $@;)
+ @echo "*/" >> $@
+
+CLEAN-OBJS += examples.doxy samples.dox $(ALL_EXAMPLES:.c=.dox)
+DOCS-yes += examples.doxy samples.dox $(ALL_EXAMPLES:.c=.dox)
+examples.doxy: samples.dox $(ALL_EXAMPLES:.c=.dox)
+ @echo "INPUT += $^" > $@
diff --git a/libvpx/examples/decode_to_md5.txt b/libvpx/examples/decode_to_md5.txt
new file mode 100644
index 0000000..ea0eb69
--- /dev/null
+++ b/libvpx/examples/decode_to_md5.txt
@@ -0,0 +1,48 @@
+@TEMPLATE decoder_tmpl.c
+Frame-by-frame MD5 Checksum
+===========================
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
+This example builds upon the simple decoder loop to show how checksums
+of the decoded output can be generated. These are used for validating
+decoder implementations against the reference implementation, for example.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
+
+MD5 algorithm
+-------------
+The Message-Digest 5 (MD5) is a well known hash function. We have provided
+an implementation derived from the RSA Data Security, Inc. MD5 Message-Digest
+Algorithm for your use. Our implmentation only changes the interface of this
+reference code. You must include the `md5_utils.h` header for access to these
+functions.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ EXTRA_INCLUDES
+#include "md5_utils.h"
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ EXTRA_INCLUDES
+
+
+Processing The Decoded Data
+---------------------------
+Each row of the image is passed to the MD5 accumulator. First the Y plane
+is processed, then U, then V. It is important to honor the image's `stride`
+values.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX
+unsigned char md5_sum[16];
+MD5Context md5;
+int i;
+
+MD5Init(&md5);
+
+for(plane=0; plane < 3; plane++) {
+ unsigned char *buf =img->planes[plane];
+
+ for(y=0; y < (plane ? (img->d_h + 1) >> 1 : img->d_h); y++) {
+ MD5Update(&md5, buf, (plane ? (img->d_w + 1) >> 1 : img->d_w));
+ buf += img->stride[plane];
+ }
+}
+
+MD5Final(md5_sum, &md5);
+for(i=0; i<16; i++)
+ fprintf(outfile, "%02x",md5_sum[i]);
+fprintf(outfile, " img-%dx%d-%04d.i420\n", img->d_w, img->d_h,
+ frame_cnt);
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX
diff --git a/examples/decode_with_drops.txt b/libvpx/examples/decode_with_drops.txt
similarity index 100%
rename from examples/decode_with_drops.txt
rename to libvpx/examples/decode_with_drops.txt
diff --git a/libvpx/examples/decode_with_partial_drops.txt b/libvpx/examples/decode_with_partial_drops.txt
new file mode 100644
index 0000000..7b0d3d2
--- /dev/null
+++ b/libvpx/examples/decode_with_partial_drops.txt
@@ -0,0 +1,238 @@
+@TEMPLATE decoder_tmpl.c
+Decode With Partial Drops Example
+=========================
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
+This is an example utility which drops a series of frames (or parts of frames),
+as specified on the command line. This is useful for observing the error
+recovery features of the codec.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ EXTRA_INCLUDES
+#include <time.h>
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ EXTRA_INCLUDES
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ HELPERS
+struct parsed_header
+{
+ char key_frame;
+ int version;
+ char show_frame;
+ int first_part_size;
+};
+
+int next_packet(struct parsed_header* hdr, int pos, int length, int mtu)
+{
+ int size = 0;
+ int remaining = length - pos;
+ /* Uncompressed part is 3 bytes for P frames and 10 bytes for I frames */
+ int uncomp_part_size = (hdr->key_frame ? 10 : 3);
+ /* number of bytes yet to send from header and the first partition */
+ int remainFirst = uncomp_part_size + hdr->first_part_size - pos;
+ if (remainFirst > 0)
+ {
+ if (remainFirst <= mtu)
+ {
+ size = remainFirst;
+ }
+ else
+ {
+ size = mtu;
+ }
+
+ return size;
+ }
+
+ /* second partition; just slot it up according to MTU */
+ if (remaining <= mtu)
+ {
+ size = remaining;
+ return size;
+ }
+ return mtu;
+}
+
+void throw_packets(unsigned char* frame, int* size, int loss_rate,
+ int* thrown, int* kept)
+{
+ unsigned char loss_frame[256*1024];
+ int pkg_size = 1;
+ int pos = 0;
+ int loss_pos = 0;
+ struct parsed_header hdr;
+ unsigned int tmp;
+ int mtu = 1500;
+
+ if (*size < 3)
+ {
+ return;
+ }
+ putc('|', stdout);
+ /* parse uncompressed 3 bytes */
+ tmp = (frame[2] << 16) | (frame[1] << 8) | frame[0];
+ hdr.key_frame = !(tmp & 0x1); /* inverse logic */
+ hdr.version = (tmp >> 1) & 0x7;
+ hdr.show_frame = (tmp >> 4) & 0x1;
+ hdr.first_part_size = (tmp >> 5) & 0x7FFFF;
+
+ /* don't drop key frames */
+ if (hdr.key_frame)
+ {
+ int i;
+ *kept = *size/mtu + ((*size % mtu > 0) ? 1 : 0); /* approximate */
+ for (i=0; i < *kept; i++)
+ putc('.', stdout);
+ return;
+ }
+
+ while ((pkg_size = next_packet(&hdr, pos, *size, mtu)) > 0)
+ {
+ int loss_event = ((rand() + 1.0)/(RAND_MAX + 1.0) < loss_rate/100.0);
+ if (*thrown == 0 && !loss_event)
+ {
+ memcpy(loss_frame + loss_pos, frame + pos, pkg_size);
+ loss_pos += pkg_size;
+ (*kept)++;
+ putc('.', stdout);
+ }
+ else
+ {
+ (*thrown)++;
+ putc('X', stdout);
+ }
+ pos += pkg_size;
+ }
+ memcpy(frame, loss_frame, loss_pos);
+ memset(frame + loss_pos, 0, *size - loss_pos);
+ *size = loss_pos;
+}
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ HELPERS
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
+/* Initialize codec */
+flags = VPX_CODEC_USE_ERROR_CONCEALMENT;
+res = vpx_codec_dec_init(&codec, interface, &dec_cfg, flags);
+if(res)
+ die_codec(&codec, "Failed to initialize decoder");
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
+
+Usage
+-----
+This example adds a single argument to the `simple_decoder` example,
+which specifies the range or pattern of frames to drop. The parameter is
+parsed as follows:
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
+if(argc < 4 || argc > 6)
+ die("Usage: %s <infile> <outfile> [-t <num threads>] <N-M|N/M|L,S>\n",
+ argv[0]);
+{
+ char *nptr;
+ int arg_num = 3;
+ if (argc == 6 && strncmp(argv[arg_num++], "-t", 2) == 0)
+ dec_cfg.threads = strtol(argv[arg_num++], NULL, 0);
+ n = strtol(argv[arg_num], &nptr, 0);
+ mode = (*nptr == '\0' || *nptr == ',') ? 2 : (*nptr == '-') ? 1 : 0;
+
+ m = strtol(nptr+1, NULL, 0);
+ if((!n && !m) || (*nptr != '-' && *nptr != '/' &&
+ *nptr != '\0' && *nptr != ','))
+ die("Couldn't parse pattern %s\n", argv[3]);
+}
+seed = (m > 0) ? m : (unsigned int)time(NULL);
+srand(seed);thrown_frame = 0;
+printf("Seed: %u\n", seed);
+printf("Threads: %d\n", dec_cfg.threads);
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
+
+
+Dropping A Range Of Frames
+--------------------------
+To drop a range of frames, specify the starting frame and the ending
+frame to drop, separated by a dash. The following command will drop
+frames 5 through 10 (base 1).
+
+ $ ./decode_with_partial_drops in.ivf out.i420 5-10
+
+
+Dropping A Pattern Of Frames
+----------------------------
+To drop a pattern of frames, specify the number of frames to drop and
+the number of frames after which to repeat the pattern, separated by
+a forward-slash. The following command will drop 3 of 7 frames.
+Specifically, it will decode 4 frames, then drop 3 frames, and then
+repeat.
+
+ $ ./decode_with_partial_drops in.ivf out.i420 3/7
+
+Dropping Random Parts Of Frames
+-------------------------------
+A third argument tuple is available to split the frame into 1500 bytes pieces
+and randomly drop pieces rather than frames. The frame will be split at
+partition boundaries where possible. The following example will seed the RNG
+with the seed 123 and drop approximately 5% of the pieces. Pieces which
+are depending on an already dropped piece will also be dropped.
+
+ $ ./decode_with_partial_drops in.ivf out.i420 5,123
+
+
+Extra Variables
+---------------
+This example maintains the pattern passed on the command line in the
+`n`, `m`, and `is_range` variables:
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ EXTRA_VARS
+int n, m, mode;
+unsigned int seed;
+int thrown=0, kept=0;
+int thrown_frame=0, kept_frame=0;
+vpx_codec_dec_cfg_t dec_cfg = {0};
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ EXTRA_VARS
+
+
+Making The Drop Decision
+------------------------
+The example decides whether to drop the frame based on the current
+frame number, immediately before decoding the frame.
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PRE_DECODE
+/* Decide whether to throw parts of the frame or the whole frame
+ depending on the drop mode */
+thrown_frame = 0;
+kept_frame = 0;
+switch (mode)
+{
+case 0:
+ if (m - (frame_cnt-1)%m <= n)
+ {
+ frame_sz = 0;
+ }
+ break;
+case 1:
+ if (frame_cnt >= n && frame_cnt <= m)
+ {
+ frame_sz = 0;
+ }
+ break;
+case 2:
+ throw_packets(frame, &frame_sz, n, &thrown_frame, &kept_frame);
+ break;
+default: break;
+}
+if (mode < 2)
+{
+ if (frame_sz == 0)
+ {
+ putc('X', stdout);
+ thrown_frame++;
+ }
+ else
+ {
+ putc('.', stdout);
+ kept_frame++;
+ }
+}
+thrown += thrown_frame;
+kept += kept_frame;
+fflush(stdout);
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PRE_DECODE
diff --git a/libvpx/examples/decoder_tmpl.c b/libvpx/examples/decoder_tmpl.c
new file mode 100644
index 0000000..8194f0a
--- /dev/null
+++ b/libvpx/examples/decoder_tmpl.c
@@ -0,0 +1,103 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*
+@*INTRODUCTION
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#define VPX_CODEC_DISABLE_COMPAT 1
+#include "vpx/vpx_decoder.h"
+#include "vpx/vp8dx.h"
+#define interface (vpx_codec_vp8_dx())
+@EXTRA_INCLUDES
+
+
+#define IVF_FILE_HDR_SZ (32)
+#define IVF_FRAME_HDR_SZ (12)
+
+static unsigned int mem_get_le32(const unsigned char *mem) {
+ return (mem[3] << 24)|(mem[2] << 16)|(mem[1] << 8)|(mem[0]);
+}
+
+static void die(const char *fmt, ...) {
+ va_list ap;
+
+ va_start(ap, fmt);
+ vprintf(fmt, ap);
+ if(fmt[strlen(fmt)-1] != '\n')
+ printf("\n");
+ exit(EXIT_FAILURE);
+}
+
+@DIE_CODEC
+
+@HELPERS
+
+int main(int argc, char **argv) {
+ FILE *infile, *outfile;
+ vpx_codec_ctx_t codec;
+ int flags = 0, frame_cnt = 0;
+ unsigned char file_hdr[IVF_FILE_HDR_SZ];
+ unsigned char frame_hdr[IVF_FRAME_HDR_SZ];
+ unsigned char frame[256*1024];
+ vpx_codec_err_t res;
+@@@@EXTRA_VARS
+
+ (void)res;
+ /* Open files */
+@@@@USAGE
+ if(!(infile = fopen(argv[1], "rb")))
+ die("Failed to open %s for reading", argv[1]);
+ if(!(outfile = fopen(argv[2], "wb")))
+ die("Failed to open %s for writing", argv[2]);
+
+ /* Read file header */
+ if(!(fread(file_hdr, 1, IVF_FILE_HDR_SZ, infile) == IVF_FILE_HDR_SZ
+ && file_hdr[0]=='D' && file_hdr[1]=='K' && file_hdr[2]=='I'
+ && file_hdr[3]=='F'))
+ die("%s is not an IVF file.", argv[1]);
+
+ printf("Using %s\n",vpx_codec_iface_name(interface));
+@@@@DEC_INIT
+
+ /* Read each frame */
+ while(fread(frame_hdr, 1, IVF_FRAME_HDR_SZ, infile) == IVF_FRAME_HDR_SZ) {
+ int frame_sz = mem_get_le32(frame_hdr);
+ vpx_codec_iter_t iter = NULL;
+ vpx_image_t *img;
+
+
+ frame_cnt++;
+ if(frame_sz > sizeof(frame))
+ die("Frame %d data too big for example code buffer", frame_sz);
+ if(fread(frame, 1, frame_sz, infile) != frame_sz)
+ die("Frame %d failed to read complete frame", frame_cnt);
+
+@@@@@@@@PRE_DECODE
+@@@@@@@@DECODE
+
+ /* Write decoded data to disk */
+@@@@@@@@GET_FRAME
+ unsigned int plane, y;
+
+@@@@@@@@@@@@PROCESS_DX
+ }
+ }
+ printf("Processed %d frames.\n",frame_cnt);
+@@@@DESTROY
+
+ fclose(outfile);
+ fclose(infile);
+ return EXIT_SUCCESS;
+}
diff --git a/libvpx/examples/decoder_tmpl.txt b/libvpx/examples/decoder_tmpl.txt
new file mode 100644
index 0000000..e652a63
--- /dev/null
+++ b/libvpx/examples/decoder_tmpl.txt
@@ -0,0 +1,62 @@
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INCLUDES
+#define VPX_CODEC_DISABLE_COMPAT 1
+#include "vpx/vpx_decoder.h"
+#include "vpx/vp8dx.h"
+#define interface (vpx_codec_vp8_dx())
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INCLUDES
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DIE_CODEC
+static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
+ const char *detail = vpx_codec_error_detail(ctx);
+
+ printf("%s: %s\n", s, vpx_codec_error(ctx));
+ if(detail)
+ printf(" %s\n",detail);
+ exit(EXIT_FAILURE);
+}
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DIE_CODEC
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
+if(argc!=3)
+ die("Usage: %s <infile> <outfile>\n", argv[0]);
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
+/* Initialize codec */
+if(vpx_codec_dec_init(&codec, interface, NULL, flags))
+ die_codec(&codec, "Failed to initialize decoder");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DECODE
+/* Decode the frame */
+if(vpx_codec_decode(&codec, frame, frame_sz, NULL, 0))
+ die_codec(&codec, "Failed to decode frame");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DECODE
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ GET_FRAME
+while((img = vpx_codec_get_frame(&codec, &iter))) {
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ GET_FRAME
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX
+for(plane=0; plane < 3; plane++) {
+ unsigned char *buf =img->planes[plane];
+
+ for(y=0; y < (plane ? (img->d_h + 1) >> 1 : img->d_h); y++) {
+ (void) fwrite(buf, 1, (plane ? (img->d_w + 1) >> 1 : img->d_w),
+ outfile);
+ buf += img->stride[plane];
+ }
+}
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
+if(vpx_codec_destroy(&codec))
+ die_codec(&codec, "Failed to destroy codec");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
diff --git a/libvpx/examples/encoder_tmpl.c b/libvpx/examples/encoder_tmpl.c
new file mode 100644
index 0000000..e2b65ec
--- /dev/null
+++ b/libvpx/examples/encoder_tmpl.c
@@ -0,0 +1,186 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*
+@*INTRODUCTION
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#define VPX_CODEC_DISABLE_COMPAT 1
+#include "vpx/vpx_encoder.h"
+#include "vpx/vp8cx.h"
+#define interface (vpx_codec_vp8_cx())
+#define fourcc 0x30385056
+@EXTRA_INCLUDES
+
+#define IVF_FILE_HDR_SZ (32)
+#define IVF_FRAME_HDR_SZ (12)
+
+static void mem_put_le16(char *mem, unsigned int val) {
+ mem[0] = val;
+ mem[1] = val>>8;
+}
+
+static void mem_put_le32(char *mem, unsigned int val) {
+ mem[0] = val;
+ mem[1] = val>>8;
+ mem[2] = val>>16;
+ mem[3] = val>>24;
+}
+
+static void die(const char *fmt, ...) {
+ va_list ap;
+
+ va_start(ap, fmt);
+ vprintf(fmt, ap);
+ if(fmt[strlen(fmt)-1] != '\n')
+ printf("\n");
+ exit(EXIT_FAILURE);
+}
+
+@DIE_CODEC
+
+static int read_frame(FILE *f, vpx_image_t *img) {
+ size_t nbytes, to_read;
+ int res = 1;
+
+ to_read = img->w*img->h*3/2;
+ nbytes = fread(img->planes[0], 1, to_read, f);
+ if(nbytes != to_read) {
+ res = 0;
+ if(nbytes > 0)
+ printf("Warning: Read partial frame. Check your width & height!\n");
+ }
+ return res;
+}
+
+static void write_ivf_file_header(FILE *outfile,
+ const vpx_codec_enc_cfg_t *cfg,
+ int frame_cnt) {
+ char header[32];
+
+ if(cfg->g_pass != VPX_RC_ONE_PASS && cfg->g_pass != VPX_RC_LAST_PASS)
+ return;
+ header[0] = 'D';
+ header[1] = 'K';
+ header[2] = 'I';
+ header[3] = 'F';
+ mem_put_le16(header+4, 0); /* version */
+ mem_put_le16(header+6, 32); /* headersize */
+ mem_put_le32(header+8, fourcc); /* headersize */
+ mem_put_le16(header+12, cfg->g_w); /* width */
+ mem_put_le16(header+14, cfg->g_h); /* height */
+ mem_put_le32(header+16, cfg->g_timebase.den); /* rate */
+ mem_put_le32(header+20, cfg->g_timebase.num); /* scale */
+ mem_put_le32(header+24, frame_cnt); /* length */
+ mem_put_le32(header+28, 0); /* unused */
+
+ (void) fwrite(header, 1, 32, outfile);
+}
+
+
+static void write_ivf_frame_header(FILE *outfile,
+ const vpx_codec_cx_pkt_t *pkt)
+{
+ char header[12];
+ vpx_codec_pts_t pts;
+
+ if(pkt->kind != VPX_CODEC_CX_FRAME_PKT)
+ return;
+
+ pts = pkt->data.frame.pts;
+ mem_put_le32(header, pkt->data.frame.sz);
+ mem_put_le32(header+4, pts&0xFFFFFFFF);
+ mem_put_le32(header+8, pts >> 32);
+
+ (void) fwrite(header, 1, 12, outfile);
+}
+
+int main(int argc, char **argv) {
+ FILE *infile, *outfile;
+ vpx_codec_ctx_t codec;
+ vpx_codec_enc_cfg_t cfg;
+ int frame_cnt = 0;
+ vpx_image_t raw;
+ vpx_codec_err_t res;
+ long width;
+ long height;
+ int frame_avail;
+ int got_data;
+ int flags = 0;
+@@@@TWOPASS_VARS
+
+ /* Open files */
+@@@@USAGE
+ width = strtol(argv[1], NULL, 0);
+ height = strtol(argv[2], NULL, 0);
+ if(width < 16 || width%2 || height <16 || height%2)
+ die("Invalid resolution: %ldx%ld", width, height);
+ if(!vpx_img_alloc(&raw, VPX_IMG_FMT_I420, width, height, 1))
+ die("Faile to allocate image", width, height);
+ if(!(outfile = fopen(argv[4], "wb")))
+ die("Failed to open %s for writing", argv[4]);
+
+ printf("Using %s\n",vpx_codec_iface_name(interface));
+
+@@@@ENC_DEF_CFG
+
+@@@@ENC_SET_CFG
+@@@@ENC_SET_CFG2
+
+ write_ivf_file_header(outfile, &cfg, 0);
+
+@@@@TWOPASS_LOOP_BEGIN
+
+ /* Open input file for this encoding pass */
+ if(!(infile = fopen(argv[3], "rb")))
+ die("Failed to open %s for reading", argv[3]);
+
+@@@@@@@@ENC_INIT
+
+ frame_avail = 1;
+ got_data = 0;
+ while(frame_avail || got_data) {
+ vpx_codec_iter_t iter = NULL;
+ const vpx_codec_cx_pkt_t *pkt;
+
+@@@@@@@@@@@@PER_FRAME_CFG
+@@@@@@@@@@@@ENCODE_FRAME
+ got_data = 0;
+ while( (pkt = vpx_codec_get_cx_data(&codec, &iter)) ) {
+ got_data = 1;
+ switch(pkt->kind) {
+@@@@@@@@@@@@@@@@PROCESS_FRAME
+@@@@@@@@@@@@@@@@PROCESS_STATS
+ default:
+ break;
+ }
+ printf(pkt->kind == VPX_CODEC_CX_FRAME_PKT
+ && (pkt->data.frame.flags & VPX_FRAME_IS_KEY)? "K":".");
+ fflush(stdout);
+ }
+ frame_cnt++;
+ }
+ printf("\n");
+ fclose(infile);
+@@@@TWOPASS_LOOP_END
+
+ printf("Processed %d frames.\n",frame_cnt-1);
+@@@@DESTROY
+
+ /* Try to rewrite the file header with the actual frame count */
+ if(!fseek(outfile, 0, SEEK_SET))
+ write_ivf_file_header(outfile, &cfg, frame_cnt-1);
+ fclose(outfile);
+ return EXIT_SUCCESS;
+}
diff --git a/libvpx/examples/encoder_tmpl.txt b/libvpx/examples/encoder_tmpl.txt
new file mode 100644
index 0000000..1afbd8b
--- /dev/null
+++ b/libvpx/examples/encoder_tmpl.txt
@@ -0,0 +1,74 @@
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INCLUDES
+#define VPX_CODEC_DISABLE_COMPAT 1
+#include "vpx/vpx_encoder.h"
+#include "vpx/vp8cx.h"
+#define interface (vpx_codec_vp8_cx())
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INCLUDES
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DIE_CODEC
+static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
+ const char *detail = vpx_codec_error_detail(ctx);
+
+ printf("%s: %s\n", s, vpx_codec_error(ctx));
+ if(detail)
+ printf(" %s\n",detail);
+ exit(EXIT_FAILURE);
+}
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DIE_CODEC
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
+if(argc!=5)
+ die("Usage: %s <width> <height> <infile> <outfile>\n", argv[0]);
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ USAGE
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_DEF_CFG
+/* Populate encoder configuration */
+res = vpx_codec_enc_config_default(interface, &cfg, 0);
+if(res) {
+ printf("Failed to get config: %s\n", vpx_codec_err_to_string(res));
+ return EXIT_FAILURE;
+}
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_DEF_CFG
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_SET_CFG
+/* Update the default configuration with our settings */
+cfg.rc_target_bitrate = width * height * cfg.rc_target_bitrate
+ / cfg.g_w / cfg.g_h;
+cfg.g_w = width;
+cfg.g_h = height;
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_SET_CFG
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INIT
+/* Initialize codec */
+if(vpx_codec_enc_init(&codec, interface, &cfg, 0))
+ die_codec(&codec, "Failed to initialize encoder");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INIT
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENCODE_FRAME
+frame_avail = read_frame(infile, &raw);
+if(vpx_codec_encode(&codec, frame_avail? &raw : NULL, frame_cnt,
+ 1, flags, VPX_DL_REALTIME))
+ die_codec(&codec, "Failed to encode frame");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENCODE_FRAME
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_FRAME
+case VPX_CODEC_CX_FRAME_PKT:
+ write_ivf_frame_header(outfile, pkt);
+ (void) fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
+ outfile);
+ break;
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_FRAME
+
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
+vpx_img_free(&raw);
+if(vpx_codec_destroy(&codec))
+ die_codec(&codec, "Failed to destroy codec");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
diff --git a/examples/error_resilient.txt b/libvpx/examples/error_resilient.txt
similarity index 100%
rename from examples/error_resilient.txt
rename to libvpx/examples/error_resilient.txt
diff --git a/examples/force_keyframe.txt b/libvpx/examples/force_keyframe.txt
similarity index 100%
rename from examples/force_keyframe.txt
rename to libvpx/examples/force_keyframe.txt
diff --git a/examples/gen_example_code.sh b/libvpx/examples/gen_example_code.sh
similarity index 100%
rename from examples/gen_example_code.sh
rename to libvpx/examples/gen_example_code.sh
diff --git a/examples/gen_example_doxy.php b/libvpx/examples/gen_example_doxy.php
similarity index 100%
rename from examples/gen_example_doxy.php
rename to libvpx/examples/gen_example_doxy.php
diff --git a/examples/gen_example_text.sh b/libvpx/examples/gen_example_text.sh
similarity index 100%
rename from examples/gen_example_text.sh
rename to libvpx/examples/gen_example_text.sh
diff --git a/examples/includes/ASCIIMathPHP-2.0/ASCIIMathPHP-2.0.cfg.php b/libvpx/examples/includes/ASCIIMathPHP-2.0/ASCIIMathPHP-2.0.cfg.php
similarity index 100%
rename from examples/includes/ASCIIMathPHP-2.0/ASCIIMathPHP-2.0.cfg.php
rename to libvpx/examples/includes/ASCIIMathPHP-2.0/ASCIIMathPHP-2.0.cfg.php
diff --git a/examples/includes/ASCIIMathPHP-2.0/ASCIIMathPHP-2.0.class.php b/libvpx/examples/includes/ASCIIMathPHP-2.0/ASCIIMathPHP-2.0.class.php
similarity index 100%
rename from examples/includes/ASCIIMathPHP-2.0/ASCIIMathPHP-2.0.class.php
rename to libvpx/examples/includes/ASCIIMathPHP-2.0/ASCIIMathPHP-2.0.class.php
diff --git a/examples/includes/ASCIIMathPHP-2.0/htmlMathML.js b/libvpx/examples/includes/ASCIIMathPHP-2.0/htmlMathML.js
similarity index 100%
rename from examples/includes/ASCIIMathPHP-2.0/htmlMathML.js
rename to libvpx/examples/includes/ASCIIMathPHP-2.0/htmlMathML.js
diff --git a/libvpx/examples/includes/HTML-Toc-0.91/.gitattributes b/libvpx/examples/includes/HTML-Toc-0.91/.gitattributes
new file mode 100644
index 0000000..aecf250
--- /dev/null
+++ b/libvpx/examples/includes/HTML-Toc-0.91/.gitattributes
@@ -0,0 +1 @@
+* -crlf
diff --git a/examples/includes/HTML-Toc-0.91/Changes b/libvpx/examples/includes/HTML-Toc-0.91/Changes
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/Changes
rename to libvpx/examples/includes/HTML-Toc-0.91/Changes
diff --git a/examples/includes/HTML-Toc-0.91/MANIFEST b/libvpx/examples/includes/HTML-Toc-0.91/MANIFEST
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/MANIFEST
rename to libvpx/examples/includes/HTML-Toc-0.91/MANIFEST
diff --git a/examples/includes/HTML-Toc-0.91/Makefile.PL b/libvpx/examples/includes/HTML-Toc-0.91/Makefile.PL
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/Makefile.PL
rename to libvpx/examples/includes/HTML-Toc-0.91/Makefile.PL
diff --git a/examples/includes/HTML-Toc-0.91/Toc.pm b/libvpx/examples/includes/HTML-Toc-0.91/Toc.pm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/Toc.pm
rename to libvpx/examples/includes/HTML-Toc-0.91/Toc.pm
diff --git a/examples/includes/HTML-Toc-0.91/Toc.pod b/libvpx/examples/includes/HTML-Toc-0.91/Toc.pod
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/Toc.pod
rename to libvpx/examples/includes/HTML-Toc-0.91/Toc.pod
diff --git a/examples/includes/HTML-Toc-0.91/TocGenerator.pm b/libvpx/examples/includes/HTML-Toc-0.91/TocGenerator.pm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/TocGenerator.pm
rename to libvpx/examples/includes/HTML-Toc-0.91/TocGenerator.pm
diff --git a/examples/includes/HTML-Toc-0.91/TocInsertor.pm b/libvpx/examples/includes/HTML-Toc-0.91/TocInsertor.pm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/TocInsertor.pm
rename to libvpx/examples/includes/HTML-Toc-0.91/TocInsertor.pm
diff --git a/examples/includes/HTML-Toc-0.91/TocUpdator.pm b/libvpx/examples/includes/HTML-Toc-0.91/TocUpdator.pm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/TocUpdator.pm
rename to libvpx/examples/includes/HTML-Toc-0.91/TocUpdator.pm
diff --git a/examples/includes/HTML-Toc-0.91/t/ManualTest/manualTest1.htm b/libvpx/examples/includes/HTML-Toc-0.91/t/ManualTest/manualTest1.htm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/ManualTest/manualTest1.htm
rename to libvpx/examples/includes/HTML-Toc-0.91/t/ManualTest/manualTest1.htm
diff --git a/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir1/SubSubDir1/index.htm b/libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir1/SubSubDir1/index.htm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir1/SubSubDir1/index.htm
rename to libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir1/SubSubDir1/index.htm
diff --git a/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir1/index.htm b/libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir1/index.htm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir1/index.htm
rename to libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir1/index.htm
diff --git a/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/SubSubDir1/index.htm b/libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/SubSubDir1/index.htm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/SubSubDir1/index.htm
rename to libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/SubSubDir1/index.htm
diff --git a/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/SubSubDir2/index.htm b/libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/SubSubDir2/index.htm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/SubSubDir2/index.htm
rename to libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/SubSubDir2/index.htm
diff --git a/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/index.htm b/libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/index.htm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/index.htm
rename to libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir2/index.htm
diff --git a/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir3/index.htm b/libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir3/index.htm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir3/index.htm
rename to libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/SubDir3/index.htm
diff --git a/examples/includes/HTML-Toc-0.91/t/SiteMap/index.htm b/libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/index.htm
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/SiteMap/index.htm
rename to libvpx/examples/includes/HTML-Toc-0.91/t/SiteMap/index.htm
diff --git a/examples/includes/HTML-Toc-0.91/t/extend.t b/libvpx/examples/includes/HTML-Toc-0.91/t/extend.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/extend.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/extend.t
diff --git a/examples/includes/HTML-Toc-0.91/t/format.t b/libvpx/examples/includes/HTML-Toc-0.91/t/format.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/format.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/format.t
diff --git a/examples/includes/HTML-Toc-0.91/t/generate.t b/libvpx/examples/includes/HTML-Toc-0.91/t/generate.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/generate.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/generate.t
diff --git a/examples/includes/HTML-Toc-0.91/t/insert.t b/libvpx/examples/includes/HTML-Toc-0.91/t/insert.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/insert.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/insert.t
diff --git a/examples/includes/HTML-Toc-0.91/t/manualTest.t b/libvpx/examples/includes/HTML-Toc-0.91/t/manualTest.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/manualTest.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/manualTest.t
diff --git a/examples/includes/HTML-Toc-0.91/t/options.t b/libvpx/examples/includes/HTML-Toc-0.91/t/options.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/options.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/options.t
diff --git a/examples/includes/HTML-Toc-0.91/t/podExamples.t b/libvpx/examples/includes/HTML-Toc-0.91/t/podExamples.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/podExamples.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/podExamples.t
diff --git a/examples/includes/HTML-Toc-0.91/t/propagate.t b/libvpx/examples/includes/HTML-Toc-0.91/t/propagate.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/propagate.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/propagate.t
diff --git a/examples/includes/HTML-Toc-0.91/t/siteMap.t b/libvpx/examples/includes/HTML-Toc-0.91/t/siteMap.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/siteMap.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/siteMap.t
diff --git a/examples/includes/HTML-Toc-0.91/t/update.t b/libvpx/examples/includes/HTML-Toc-0.91/t/update.t
similarity index 100%
rename from examples/includes/HTML-Toc-0.91/t/update.t
rename to libvpx/examples/includes/HTML-Toc-0.91/t/update.t
diff --git a/examples/includes/PHP-Markdown-Extra-1.2.3/License.text b/libvpx/examples/includes/PHP-Markdown-Extra-1.2.3/License.text
similarity index 100%
rename from examples/includes/PHP-Markdown-Extra-1.2.3/License.text
rename to libvpx/examples/includes/PHP-Markdown-Extra-1.2.3/License.text
diff --git a/examples/includes/PHP-Markdown-Extra-1.2.3/PHP Markdown Extra Readme.text b/libvpx/examples/includes/PHP-Markdown-Extra-1.2.3/PHP Markdown Extra Readme.text
similarity index 100%
rename from examples/includes/PHP-Markdown-Extra-1.2.3/PHP Markdown Extra Readme.text
rename to libvpx/examples/includes/PHP-Markdown-Extra-1.2.3/PHP Markdown Extra Readme.text
diff --git a/examples/includes/PHP-Markdown-Extra-1.2.3/markdown.php b/libvpx/examples/includes/PHP-Markdown-Extra-1.2.3/markdown.php
similarity index 100%
rename from examples/includes/PHP-Markdown-Extra-1.2.3/markdown.php
rename to libvpx/examples/includes/PHP-Markdown-Extra-1.2.3/markdown.php
diff --git a/examples/includes/PHP-SmartyPants-1.5.1e/PHP SmartyPants Readme.txt b/libvpx/examples/includes/PHP-SmartyPants-1.5.1e/PHP SmartyPants Readme.txt
similarity index 100%
rename from examples/includes/PHP-SmartyPants-1.5.1e/PHP SmartyPants Readme.txt
rename to libvpx/examples/includes/PHP-SmartyPants-1.5.1e/PHP SmartyPants Readme.txt
diff --git a/examples/includes/PHP-SmartyPants-1.5.1e/smartypants.php b/libvpx/examples/includes/PHP-SmartyPants-1.5.1e/smartypants.php
similarity index 100%
rename from examples/includes/PHP-SmartyPants-1.5.1e/smartypants.php
rename to libvpx/examples/includes/PHP-SmartyPants-1.5.1e/smartypants.php
diff --git a/examples/includes/geshi/contrib/aliased.php b/libvpx/examples/includes/geshi/contrib/aliased.php
similarity index 100%
rename from examples/includes/geshi/contrib/aliased.php
rename to libvpx/examples/includes/geshi/contrib/aliased.php
diff --git a/examples/includes/geshi/contrib/cssgen.php b/libvpx/examples/includes/geshi/contrib/cssgen.php
similarity index 100%
rename from examples/includes/geshi/contrib/cssgen.php
rename to libvpx/examples/includes/geshi/contrib/cssgen.php
diff --git a/examples/includes/geshi/contrib/cssgen2.php b/libvpx/examples/includes/geshi/contrib/cssgen2.php
similarity index 100%
rename from examples/includes/geshi/contrib/cssgen2.php
rename to libvpx/examples/includes/geshi/contrib/cssgen2.php
diff --git a/examples/includes/geshi/contrib/example.php b/libvpx/examples/includes/geshi/contrib/example.php
similarity index 100%
rename from examples/includes/geshi/contrib/example.php
rename to libvpx/examples/includes/geshi/contrib/example.php
diff --git a/examples/includes/geshi/contrib/langcheck.php b/libvpx/examples/includes/geshi/contrib/langcheck.php
similarity index 100%
rename from examples/includes/geshi/contrib/langcheck.php
rename to libvpx/examples/includes/geshi/contrib/langcheck.php
diff --git a/examples/includes/geshi/docs/BUGS b/libvpx/examples/includes/geshi/docs/BUGS
similarity index 100%
rename from examples/includes/geshi/docs/BUGS
rename to libvpx/examples/includes/geshi/docs/BUGS
diff --git a/examples/includes/geshi/docs/CHANGES b/libvpx/examples/includes/geshi/docs/CHANGES
similarity index 100%
rename from examples/includes/geshi/docs/CHANGES
rename to libvpx/examples/includes/geshi/docs/CHANGES
diff --git a/examples/includes/geshi/docs/COPYING b/libvpx/examples/includes/geshi/docs/COPYING
similarity index 100%
rename from examples/includes/geshi/docs/COPYING
rename to libvpx/examples/includes/geshi/docs/COPYING
diff --git a/examples/includes/geshi/docs/README b/libvpx/examples/includes/geshi/docs/README
similarity index 100%
rename from examples/includes/geshi/docs/README
rename to libvpx/examples/includes/geshi/docs/README
diff --git a/examples/includes/geshi/docs/THANKS b/libvpx/examples/includes/geshi/docs/THANKS
similarity index 100%
rename from examples/includes/geshi/docs/THANKS
rename to libvpx/examples/includes/geshi/docs/THANKS
diff --git a/examples/includes/geshi/docs/TODO b/libvpx/examples/includes/geshi/docs/TODO
similarity index 100%
rename from examples/includes/geshi/docs/TODO
rename to libvpx/examples/includes/geshi/docs/TODO
diff --git a/examples/includes/geshi/docs/api/__filesource/fsource_geshi_core_geshi.php.html b/libvpx/examples/includes/geshi/docs/api/__filesource/fsource_geshi_core_geshi.php.html
similarity index 100%
rename from examples/includes/geshi/docs/api/__filesource/fsource_geshi_core_geshi.php.html
rename to libvpx/examples/includes/geshi/docs/api/__filesource/fsource_geshi_core_geshi.php.html
diff --git a/examples/includes/geshi/docs/api/blank.html b/libvpx/examples/includes/geshi/docs/api/blank.html
similarity index 100%
rename from examples/includes/geshi/docs/api/blank.html
rename to libvpx/examples/includes/geshi/docs/api/blank.html
diff --git a/examples/includes/geshi/docs/api/classtrees_geshi.html b/libvpx/examples/includes/geshi/docs/api/classtrees_geshi.html
similarity index 100%
rename from examples/includes/geshi/docs/api/classtrees_geshi.html
rename to libvpx/examples/includes/geshi/docs/api/classtrees_geshi.html
diff --git a/examples/includes/geshi/docs/api/elementindex.html b/libvpx/examples/includes/geshi/docs/api/elementindex.html
similarity index 100%
rename from examples/includes/geshi/docs/api/elementindex.html
rename to libvpx/examples/includes/geshi/docs/api/elementindex.html
diff --git a/examples/includes/geshi/docs/api/elementindex_geshi.html b/libvpx/examples/includes/geshi/docs/api/elementindex_geshi.html
similarity index 100%
rename from examples/includes/geshi/docs/api/elementindex_geshi.html
rename to libvpx/examples/includes/geshi/docs/api/elementindex_geshi.html
diff --git a/examples/includes/geshi/docs/api/errors.html b/libvpx/examples/includes/geshi/docs/api/errors.html
similarity index 100%
rename from examples/includes/geshi/docs/api/errors.html
rename to libvpx/examples/includes/geshi/docs/api/errors.html
diff --git a/examples/includes/geshi/docs/api/geshi/core/GeSHi.html b/libvpx/examples/includes/geshi/docs/api/geshi/core/GeSHi.html
similarity index 100%
rename from examples/includes/geshi/docs/api/geshi/core/GeSHi.html
rename to libvpx/examples/includes/geshi/docs/api/geshi/core/GeSHi.html
diff --git a/examples/includes/geshi/docs/api/geshi/core/_geshi.php.html b/libvpx/examples/includes/geshi/docs/api/geshi/core/_geshi.php.html
similarity index 100%
rename from examples/includes/geshi/docs/api/geshi/core/_geshi.php.html
rename to libvpx/examples/includes/geshi/docs/api/geshi/core/_geshi.php.html
diff --git a/examples/includes/geshi/docs/api/index.html b/libvpx/examples/includes/geshi/docs/api/index.html
similarity index 100%
rename from examples/includes/geshi/docs/api/index.html
rename to libvpx/examples/includes/geshi/docs/api/index.html
diff --git a/examples/includes/geshi/docs/api/li_geshi.html b/libvpx/examples/includes/geshi/docs/api/li_geshi.html
similarity index 100%
rename from examples/includes/geshi/docs/api/li_geshi.html
rename to libvpx/examples/includes/geshi/docs/api/li_geshi.html
diff --git a/examples/includes/geshi/docs/api/media/banner.css b/libvpx/examples/includes/geshi/docs/api/media/banner.css
similarity index 100%
rename from examples/includes/geshi/docs/api/media/banner.css
rename to libvpx/examples/includes/geshi/docs/api/media/banner.css
diff --git a/examples/includes/geshi/docs/api/media/images/AbstractClass.png b/libvpx/examples/includes/geshi/docs/api/media/images/AbstractClass.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/AbstractClass.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/AbstractClass.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/AbstractClass_logo.png b/libvpx/examples/includes/geshi/docs/api/media/images/AbstractClass_logo.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/AbstractClass_logo.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/AbstractClass_logo.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/AbstractMethod.png b/libvpx/examples/includes/geshi/docs/api/media/images/AbstractMethod.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/AbstractMethod.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/AbstractMethod.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/AbstractPrivateClass.png b/libvpx/examples/includes/geshi/docs/api/media/images/AbstractPrivateClass.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/AbstractPrivateClass.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/AbstractPrivateClass.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/AbstractPrivateClass_logo.png b/libvpx/examples/includes/geshi/docs/api/media/images/AbstractPrivateClass_logo.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/AbstractPrivateClass_logo.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/AbstractPrivateClass_logo.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/AbstractPrivateMethod.png b/libvpx/examples/includes/geshi/docs/api/media/images/AbstractPrivateMethod.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/AbstractPrivateMethod.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/AbstractPrivateMethod.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Class.png b/libvpx/examples/includes/geshi/docs/api/media/images/Class.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Class.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Class.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Class_logo.png b/libvpx/examples/includes/geshi/docs/api/media/images/Class_logo.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Class_logo.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Class_logo.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Constant.png b/libvpx/examples/includes/geshi/docs/api/media/images/Constant.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Constant.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Constant.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Constructor.png b/libvpx/examples/includes/geshi/docs/api/media/images/Constructor.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Constructor.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Constructor.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Destructor.png b/libvpx/examples/includes/geshi/docs/api/media/images/Destructor.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Destructor.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Destructor.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Function.png b/libvpx/examples/includes/geshi/docs/api/media/images/Function.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Function.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Function.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Global.png b/libvpx/examples/includes/geshi/docs/api/media/images/Global.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Global.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Global.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/I.png b/libvpx/examples/includes/geshi/docs/api/media/images/I.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/I.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/I.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Index.png b/libvpx/examples/includes/geshi/docs/api/media/images/Index.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Index.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Index.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Interface.png b/libvpx/examples/includes/geshi/docs/api/media/images/Interface.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Interface.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Interface.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Interface_logo.png b/libvpx/examples/includes/geshi/docs/api/media/images/Interface_logo.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Interface_logo.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Interface_logo.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/L.png b/libvpx/examples/includes/geshi/docs/api/media/images/L.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/L.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/L.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Lminus.png b/libvpx/examples/includes/geshi/docs/api/media/images/Lminus.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Lminus.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Lminus.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Lplus.png b/libvpx/examples/includes/geshi/docs/api/media/images/Lplus.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Lplus.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Lplus.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Method.png b/libvpx/examples/includes/geshi/docs/api/media/images/Method.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Method.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Method.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Page.png b/libvpx/examples/includes/geshi/docs/api/media/images/Page.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Page.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Page.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Page_logo.png b/libvpx/examples/includes/geshi/docs/api/media/images/Page_logo.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Page_logo.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Page_logo.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/PrivateClass.png b/libvpx/examples/includes/geshi/docs/api/media/images/PrivateClass.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/PrivateClass.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/PrivateClass.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/PrivateClass_logo.png b/libvpx/examples/includes/geshi/docs/api/media/images/PrivateClass_logo.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/PrivateClass_logo.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/PrivateClass_logo.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/PrivateMethod.png b/libvpx/examples/includes/geshi/docs/api/media/images/PrivateMethod.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/PrivateMethod.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/PrivateMethod.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/PrivateVariable.png b/libvpx/examples/includes/geshi/docs/api/media/images/PrivateVariable.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/PrivateVariable.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/PrivateVariable.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/StaticMethod.png b/libvpx/examples/includes/geshi/docs/api/media/images/StaticMethod.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/StaticMethod.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/StaticMethod.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/StaticVariable.png b/libvpx/examples/includes/geshi/docs/api/media/images/StaticVariable.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/StaticVariable.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/StaticVariable.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/T.png b/libvpx/examples/includes/geshi/docs/api/media/images/T.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/T.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/T.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Tminus.png b/libvpx/examples/includes/geshi/docs/api/media/images/Tminus.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Tminus.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Tminus.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Tplus.png b/libvpx/examples/includes/geshi/docs/api/media/images/Tplus.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Tplus.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Tplus.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/Variable.png b/libvpx/examples/includes/geshi/docs/api/media/images/Variable.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/Variable.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/Variable.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/blank.png b/libvpx/examples/includes/geshi/docs/api/media/images/blank.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/blank.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/blank.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/class_folder.png b/libvpx/examples/includes/geshi/docs/api/media/images/class_folder.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/class_folder.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/class_folder.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/empty.png b/libvpx/examples/includes/geshi/docs/api/media/images/empty.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/empty.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/empty.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/file.png b/libvpx/examples/includes/geshi/docs/api/media/images/file.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/file.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/file.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/folder.png b/libvpx/examples/includes/geshi/docs/api/media/images/folder.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/folder.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/folder.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/function_folder.png b/libvpx/examples/includes/geshi/docs/api/media/images/function_folder.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/function_folder.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/function_folder.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/next_button.png b/libvpx/examples/includes/geshi/docs/api/media/images/next_button.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/next_button.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/next_button.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/next_button_disabled.png b/libvpx/examples/includes/geshi/docs/api/media/images/next_button_disabled.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/next_button_disabled.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/next_button_disabled.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/package.png b/libvpx/examples/includes/geshi/docs/api/media/images/package.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/package.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/package.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/package_folder.png b/libvpx/examples/includes/geshi/docs/api/media/images/package_folder.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/package_folder.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/package_folder.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/previous_button.png b/libvpx/examples/includes/geshi/docs/api/media/images/previous_button.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/previous_button.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/previous_button.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/previous_button_disabled.png b/libvpx/examples/includes/geshi/docs/api/media/images/previous_button_disabled.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/previous_button_disabled.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/previous_button_disabled.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/private_class_logo.png b/libvpx/examples/includes/geshi/docs/api/media/images/private_class_logo.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/private_class_logo.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/private_class_logo.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/tutorial.png b/libvpx/examples/includes/geshi/docs/api/media/images/tutorial.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/tutorial.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/tutorial.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/tutorial_folder.png b/libvpx/examples/includes/geshi/docs/api/media/images/tutorial_folder.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/tutorial_folder.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/tutorial_folder.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/images/up_button.png b/libvpx/examples/includes/geshi/docs/api/media/images/up_button.png
similarity index 100%
rename from examples/includes/geshi/docs/api/media/images/up_button.png
rename to libvpx/examples/includes/geshi/docs/api/media/images/up_button.png
Binary files differ
diff --git a/examples/includes/geshi/docs/api/media/stylesheet.css b/libvpx/examples/includes/geshi/docs/api/media/stylesheet.css
similarity index 100%
rename from examples/includes/geshi/docs/api/media/stylesheet.css
rename to libvpx/examples/includes/geshi/docs/api/media/stylesheet.css
diff --git a/examples/includes/geshi/docs/api/packages.html b/libvpx/examples/includes/geshi/docs/api/packages.html
similarity index 100%
rename from examples/includes/geshi/docs/api/packages.html
rename to libvpx/examples/includes/geshi/docs/api/packages.html
diff --git a/examples/includes/geshi/docs/api/todolist.html b/libvpx/examples/includes/geshi/docs/api/todolist.html
similarity index 100%
rename from examples/includes/geshi/docs/api/todolist.html
rename to libvpx/examples/includes/geshi/docs/api/todolist.html
diff --git a/examples/includes/geshi/docs/geshi-doc.html b/libvpx/examples/includes/geshi/docs/geshi-doc.html
similarity index 100%
rename from examples/includes/geshi/docs/geshi-doc.html
rename to libvpx/examples/includes/geshi/docs/geshi-doc.html
diff --git a/examples/includes/geshi/docs/geshi-doc.txt b/libvpx/examples/includes/geshi/docs/geshi-doc.txt
similarity index 100%
rename from examples/includes/geshi/docs/geshi-doc.txt
rename to libvpx/examples/includes/geshi/docs/geshi-doc.txt
diff --git a/examples/includes/geshi/docs/phpdoc.ini b/libvpx/examples/includes/geshi/docs/phpdoc.ini
similarity index 100%
rename from examples/includes/geshi/docs/phpdoc.ini
rename to libvpx/examples/includes/geshi/docs/phpdoc.ini
diff --git a/examples/includes/geshi/geshi.php b/libvpx/examples/includes/geshi/geshi.php
similarity index 100%
rename from examples/includes/geshi/geshi.php
rename to libvpx/examples/includes/geshi/geshi.php
diff --git a/examples/includes/geshi/geshi/abap.php b/libvpx/examples/includes/geshi/geshi/abap.php
similarity index 100%
rename from examples/includes/geshi/geshi/abap.php
rename to libvpx/examples/includes/geshi/geshi/abap.php
diff --git a/examples/includes/geshi/geshi/actionscript.php b/libvpx/examples/includes/geshi/geshi/actionscript.php
similarity index 100%
rename from examples/includes/geshi/geshi/actionscript.php
rename to libvpx/examples/includes/geshi/geshi/actionscript.php
diff --git a/examples/includes/geshi/geshi/actionscript3.php b/libvpx/examples/includes/geshi/geshi/actionscript3.php
similarity index 100%
rename from examples/includes/geshi/geshi/actionscript3.php
rename to libvpx/examples/includes/geshi/geshi/actionscript3.php
diff --git a/examples/includes/geshi/geshi/ada.php b/libvpx/examples/includes/geshi/geshi/ada.php
similarity index 100%
rename from examples/includes/geshi/geshi/ada.php
rename to libvpx/examples/includes/geshi/geshi/ada.php
diff --git a/examples/includes/geshi/geshi/apache.php b/libvpx/examples/includes/geshi/geshi/apache.php
similarity index 100%
rename from examples/includes/geshi/geshi/apache.php
rename to libvpx/examples/includes/geshi/geshi/apache.php
diff --git a/examples/includes/geshi/geshi/applescript.php b/libvpx/examples/includes/geshi/geshi/applescript.php
similarity index 100%
rename from examples/includes/geshi/geshi/applescript.php
rename to libvpx/examples/includes/geshi/geshi/applescript.php
diff --git a/examples/includes/geshi/geshi/apt_sources.php b/libvpx/examples/includes/geshi/geshi/apt_sources.php
similarity index 100%
rename from examples/includes/geshi/geshi/apt_sources.php
rename to libvpx/examples/includes/geshi/geshi/apt_sources.php
diff --git a/examples/includes/geshi/geshi/asm.php b/libvpx/examples/includes/geshi/geshi/asm.php
similarity index 100%
rename from examples/includes/geshi/geshi/asm.php
rename to libvpx/examples/includes/geshi/geshi/asm.php
diff --git a/examples/includes/geshi/geshi/asp.php b/libvpx/examples/includes/geshi/geshi/asp.php
similarity index 100%
rename from examples/includes/geshi/geshi/asp.php
rename to libvpx/examples/includes/geshi/geshi/asp.php
diff --git a/examples/includes/geshi/geshi/autoit.php b/libvpx/examples/includes/geshi/geshi/autoit.php
similarity index 100%
rename from examples/includes/geshi/geshi/autoit.php
rename to libvpx/examples/includes/geshi/geshi/autoit.php
diff --git a/examples/includes/geshi/geshi/avisynth.php b/libvpx/examples/includes/geshi/geshi/avisynth.php
similarity index 100%
rename from examples/includes/geshi/geshi/avisynth.php
rename to libvpx/examples/includes/geshi/geshi/avisynth.php
diff --git a/examples/includes/geshi/geshi/bash.php b/libvpx/examples/includes/geshi/geshi/bash.php
similarity index 100%
rename from examples/includes/geshi/geshi/bash.php
rename to libvpx/examples/includes/geshi/geshi/bash.php
diff --git a/examples/includes/geshi/geshi/basic4gl.php b/libvpx/examples/includes/geshi/geshi/basic4gl.php
similarity index 100%
rename from examples/includes/geshi/geshi/basic4gl.php
rename to libvpx/examples/includes/geshi/geshi/basic4gl.php
diff --git a/examples/includes/geshi/geshi/bf.php b/libvpx/examples/includes/geshi/geshi/bf.php
similarity index 100%
rename from examples/includes/geshi/geshi/bf.php
rename to libvpx/examples/includes/geshi/geshi/bf.php
diff --git a/examples/includes/geshi/geshi/blitzbasic.php b/libvpx/examples/includes/geshi/geshi/blitzbasic.php
similarity index 100%
rename from examples/includes/geshi/geshi/blitzbasic.php
rename to libvpx/examples/includes/geshi/geshi/blitzbasic.php
diff --git a/examples/includes/geshi/geshi/bnf.php b/libvpx/examples/includes/geshi/geshi/bnf.php
similarity index 100%
rename from examples/includes/geshi/geshi/bnf.php
rename to libvpx/examples/includes/geshi/geshi/bnf.php
diff --git a/examples/includes/geshi/geshi/boo.php b/libvpx/examples/includes/geshi/geshi/boo.php
similarity index 100%
rename from examples/includes/geshi/geshi/boo.php
rename to libvpx/examples/includes/geshi/geshi/boo.php
diff --git a/examples/includes/geshi/geshi/c.php b/libvpx/examples/includes/geshi/geshi/c.php
similarity index 100%
rename from examples/includes/geshi/geshi/c.php
rename to libvpx/examples/includes/geshi/geshi/c.php
diff --git a/examples/includes/geshi/geshi/c_mac.php b/libvpx/examples/includes/geshi/geshi/c_mac.php
similarity index 100%
rename from examples/includes/geshi/geshi/c_mac.php
rename to libvpx/examples/includes/geshi/geshi/c_mac.php
diff --git a/examples/includes/geshi/geshi/caddcl.php b/libvpx/examples/includes/geshi/geshi/caddcl.php
similarity index 100%
rename from examples/includes/geshi/geshi/caddcl.php
rename to libvpx/examples/includes/geshi/geshi/caddcl.php
diff --git a/examples/includes/geshi/geshi/cadlisp.php b/libvpx/examples/includes/geshi/geshi/cadlisp.php
similarity index 100%
rename from examples/includes/geshi/geshi/cadlisp.php
rename to libvpx/examples/includes/geshi/geshi/cadlisp.php
diff --git a/examples/includes/geshi/geshi/cfdg.php b/libvpx/examples/includes/geshi/geshi/cfdg.php
similarity index 100%
rename from examples/includes/geshi/geshi/cfdg.php
rename to libvpx/examples/includes/geshi/geshi/cfdg.php
diff --git a/examples/includes/geshi/geshi/cfm.php b/libvpx/examples/includes/geshi/geshi/cfm.php
similarity index 100%
rename from examples/includes/geshi/geshi/cfm.php
rename to libvpx/examples/includes/geshi/geshi/cfm.php
diff --git a/examples/includes/geshi/geshi/cil.php b/libvpx/examples/includes/geshi/geshi/cil.php
similarity index 100%
rename from examples/includes/geshi/geshi/cil.php
rename to libvpx/examples/includes/geshi/geshi/cil.php
diff --git a/examples/includes/geshi/geshi/cobol.php b/libvpx/examples/includes/geshi/geshi/cobol.php
similarity index 100%
rename from examples/includes/geshi/geshi/cobol.php
rename to libvpx/examples/includes/geshi/geshi/cobol.php
diff --git a/examples/includes/geshi/geshi/cpp-qt.php b/libvpx/examples/includes/geshi/geshi/cpp-qt.php
similarity index 100%
rename from examples/includes/geshi/geshi/cpp-qt.php
rename to libvpx/examples/includes/geshi/geshi/cpp-qt.php
diff --git a/examples/includes/geshi/geshi/cpp.php b/libvpx/examples/includes/geshi/geshi/cpp.php
similarity index 100%
rename from examples/includes/geshi/geshi/cpp.php
rename to libvpx/examples/includes/geshi/geshi/cpp.php
diff --git a/examples/includes/geshi/geshi/csharp.php b/libvpx/examples/includes/geshi/geshi/csharp.php
similarity index 100%
rename from examples/includes/geshi/geshi/csharp.php
rename to libvpx/examples/includes/geshi/geshi/csharp.php
diff --git a/examples/includes/geshi/geshi/css.php b/libvpx/examples/includes/geshi/geshi/css.php
similarity index 100%
rename from examples/includes/geshi/geshi/css.php
rename to libvpx/examples/includes/geshi/geshi/css.php
diff --git a/examples/includes/geshi/geshi/d.php b/libvpx/examples/includes/geshi/geshi/d.php
similarity index 100%
rename from examples/includes/geshi/geshi/d.php
rename to libvpx/examples/includes/geshi/geshi/d.php
diff --git a/examples/includes/geshi/geshi/dcs.php b/libvpx/examples/includes/geshi/geshi/dcs.php
similarity index 100%
rename from examples/includes/geshi/geshi/dcs.php
rename to libvpx/examples/includes/geshi/geshi/dcs.php
diff --git a/examples/includes/geshi/geshi/delphi.php b/libvpx/examples/includes/geshi/geshi/delphi.php
similarity index 100%
rename from examples/includes/geshi/geshi/delphi.php
rename to libvpx/examples/includes/geshi/geshi/delphi.php
diff --git a/examples/includes/geshi/geshi/diff.php b/libvpx/examples/includes/geshi/geshi/diff.php
similarity index 100%
rename from examples/includes/geshi/geshi/diff.php
rename to libvpx/examples/includes/geshi/geshi/diff.php
diff --git a/examples/includes/geshi/geshi/div.php b/libvpx/examples/includes/geshi/geshi/div.php
similarity index 100%
rename from examples/includes/geshi/geshi/div.php
rename to libvpx/examples/includes/geshi/geshi/div.php
diff --git a/examples/includes/geshi/geshi/dos.php b/libvpx/examples/includes/geshi/geshi/dos.php
similarity index 100%
rename from examples/includes/geshi/geshi/dos.php
rename to libvpx/examples/includes/geshi/geshi/dos.php
diff --git a/examples/includes/geshi/geshi/dot.php b/libvpx/examples/includes/geshi/geshi/dot.php
similarity index 100%
rename from examples/includes/geshi/geshi/dot.php
rename to libvpx/examples/includes/geshi/geshi/dot.php
diff --git a/examples/includes/geshi/geshi/eiffel.php b/libvpx/examples/includes/geshi/geshi/eiffel.php
similarity index 100%
rename from examples/includes/geshi/geshi/eiffel.php
rename to libvpx/examples/includes/geshi/geshi/eiffel.php
diff --git a/examples/includes/geshi/geshi/email.php b/libvpx/examples/includes/geshi/geshi/email.php
similarity index 100%
rename from examples/includes/geshi/geshi/email.php
rename to libvpx/examples/includes/geshi/geshi/email.php
diff --git a/examples/includes/geshi/geshi/fortran.php b/libvpx/examples/includes/geshi/geshi/fortran.php
similarity index 100%
rename from examples/includes/geshi/geshi/fortran.php
rename to libvpx/examples/includes/geshi/geshi/fortran.php
diff --git a/examples/includes/geshi/geshi/freebasic.php b/libvpx/examples/includes/geshi/geshi/freebasic.php
similarity index 100%
rename from examples/includes/geshi/geshi/freebasic.php
rename to libvpx/examples/includes/geshi/geshi/freebasic.php
diff --git a/examples/includes/geshi/geshi/genero.php b/libvpx/examples/includes/geshi/geshi/genero.php
similarity index 100%
rename from examples/includes/geshi/geshi/genero.php
rename to libvpx/examples/includes/geshi/geshi/genero.php
diff --git a/examples/includes/geshi/geshi/gettext.php b/libvpx/examples/includes/geshi/geshi/gettext.php
similarity index 100%
rename from examples/includes/geshi/geshi/gettext.php
rename to libvpx/examples/includes/geshi/geshi/gettext.php
diff --git a/examples/includes/geshi/geshi/glsl.php b/libvpx/examples/includes/geshi/geshi/glsl.php
similarity index 100%
rename from examples/includes/geshi/geshi/glsl.php
rename to libvpx/examples/includes/geshi/geshi/glsl.php
diff --git a/examples/includes/geshi/geshi/gml.php b/libvpx/examples/includes/geshi/geshi/gml.php
similarity index 100%
rename from examples/includes/geshi/geshi/gml.php
rename to libvpx/examples/includes/geshi/geshi/gml.php
diff --git a/examples/includes/geshi/geshi/gnuplot.php b/libvpx/examples/includes/geshi/geshi/gnuplot.php
similarity index 100%
rename from examples/includes/geshi/geshi/gnuplot.php
rename to libvpx/examples/includes/geshi/geshi/gnuplot.php
diff --git a/examples/includes/geshi/geshi/groovy.php b/libvpx/examples/includes/geshi/geshi/groovy.php
similarity index 100%
rename from examples/includes/geshi/geshi/groovy.php
rename to libvpx/examples/includes/geshi/geshi/groovy.php
diff --git a/examples/includes/geshi/geshi/haskell.php b/libvpx/examples/includes/geshi/geshi/haskell.php
similarity index 100%
rename from examples/includes/geshi/geshi/haskell.php
rename to libvpx/examples/includes/geshi/geshi/haskell.php
diff --git a/examples/includes/geshi/geshi/hq9plus.php b/libvpx/examples/includes/geshi/geshi/hq9plus.php
similarity index 100%
rename from examples/includes/geshi/geshi/hq9plus.php
rename to libvpx/examples/includes/geshi/geshi/hq9plus.php
diff --git a/examples/includes/geshi/geshi/html4strict.php b/libvpx/examples/includes/geshi/geshi/html4strict.php
similarity index 100%
rename from examples/includes/geshi/geshi/html4strict.php
rename to libvpx/examples/includes/geshi/geshi/html4strict.php
diff --git a/examples/includes/geshi/geshi/idl.php b/libvpx/examples/includes/geshi/geshi/idl.php
similarity index 100%
rename from examples/includes/geshi/geshi/idl.php
rename to libvpx/examples/includes/geshi/geshi/idl.php
diff --git a/examples/includes/geshi/geshi/ini.php b/libvpx/examples/includes/geshi/geshi/ini.php
similarity index 100%
rename from examples/includes/geshi/geshi/ini.php
rename to libvpx/examples/includes/geshi/geshi/ini.php
diff --git a/examples/includes/geshi/geshi/inno.php b/libvpx/examples/includes/geshi/geshi/inno.php
similarity index 100%
rename from examples/includes/geshi/geshi/inno.php
rename to libvpx/examples/includes/geshi/geshi/inno.php
diff --git a/examples/includes/geshi/geshi/intercal.php b/libvpx/examples/includes/geshi/geshi/intercal.php
similarity index 100%
rename from examples/includes/geshi/geshi/intercal.php
rename to libvpx/examples/includes/geshi/geshi/intercal.php
diff --git a/examples/includes/geshi/geshi/io.php b/libvpx/examples/includes/geshi/geshi/io.php
similarity index 100%
rename from examples/includes/geshi/geshi/io.php
rename to libvpx/examples/includes/geshi/geshi/io.php
diff --git a/examples/includes/geshi/geshi/java.php b/libvpx/examples/includes/geshi/geshi/java.php
similarity index 100%
rename from examples/includes/geshi/geshi/java.php
rename to libvpx/examples/includes/geshi/geshi/java.php
diff --git a/examples/includes/geshi/geshi/java5.php b/libvpx/examples/includes/geshi/geshi/java5.php
similarity index 100%
rename from examples/includes/geshi/geshi/java5.php
rename to libvpx/examples/includes/geshi/geshi/java5.php
diff --git a/examples/includes/geshi/geshi/javascript.php b/libvpx/examples/includes/geshi/geshi/javascript.php
similarity index 100%
rename from examples/includes/geshi/geshi/javascript.php
rename to libvpx/examples/includes/geshi/geshi/javascript.php
diff --git a/examples/includes/geshi/geshi/kixtart.php b/libvpx/examples/includes/geshi/geshi/kixtart.php
similarity index 100%
rename from examples/includes/geshi/geshi/kixtart.php
rename to libvpx/examples/includes/geshi/geshi/kixtart.php
diff --git a/examples/includes/geshi/geshi/klonec.php b/libvpx/examples/includes/geshi/geshi/klonec.php
similarity index 100%
rename from examples/includes/geshi/geshi/klonec.php
rename to libvpx/examples/includes/geshi/geshi/klonec.php
diff --git a/examples/includes/geshi/geshi/klonecpp.php b/libvpx/examples/includes/geshi/geshi/klonecpp.php
similarity index 100%
rename from examples/includes/geshi/geshi/klonecpp.php
rename to libvpx/examples/includes/geshi/geshi/klonecpp.php
diff --git a/examples/includes/geshi/geshi/latex.php b/libvpx/examples/includes/geshi/geshi/latex.php
similarity index 100%
rename from examples/includes/geshi/geshi/latex.php
rename to libvpx/examples/includes/geshi/geshi/latex.php
diff --git a/examples/includes/geshi/geshi/lisp.php b/libvpx/examples/includes/geshi/geshi/lisp.php
similarity index 100%
rename from examples/includes/geshi/geshi/lisp.php
rename to libvpx/examples/includes/geshi/geshi/lisp.php
diff --git a/examples/includes/geshi/geshi/locobasic.php b/libvpx/examples/includes/geshi/geshi/locobasic.php
similarity index 100%
rename from examples/includes/geshi/geshi/locobasic.php
rename to libvpx/examples/includes/geshi/geshi/locobasic.php
diff --git a/examples/includes/geshi/geshi/lolcode.php b/libvpx/examples/includes/geshi/geshi/lolcode.php
similarity index 100%
rename from examples/includes/geshi/geshi/lolcode.php
rename to libvpx/examples/includes/geshi/geshi/lolcode.php
diff --git a/examples/includes/geshi/geshi/lotusformulas.php b/libvpx/examples/includes/geshi/geshi/lotusformulas.php
similarity index 100%
rename from examples/includes/geshi/geshi/lotusformulas.php
rename to libvpx/examples/includes/geshi/geshi/lotusformulas.php
diff --git a/examples/includes/geshi/geshi/lotusscript.php b/libvpx/examples/includes/geshi/geshi/lotusscript.php
similarity index 100%
rename from examples/includes/geshi/geshi/lotusscript.php
rename to libvpx/examples/includes/geshi/geshi/lotusscript.php
diff --git a/examples/includes/geshi/geshi/lscript.php b/libvpx/examples/includes/geshi/geshi/lscript.php
similarity index 100%
rename from examples/includes/geshi/geshi/lscript.php
rename to libvpx/examples/includes/geshi/geshi/lscript.php
diff --git a/examples/includes/geshi/geshi/lsl2.php b/libvpx/examples/includes/geshi/geshi/lsl2.php
similarity index 100%
rename from examples/includes/geshi/geshi/lsl2.php
rename to libvpx/examples/includes/geshi/geshi/lsl2.php
diff --git a/examples/includes/geshi/geshi/lua.php b/libvpx/examples/includes/geshi/geshi/lua.php
similarity index 100%
rename from examples/includes/geshi/geshi/lua.php
rename to libvpx/examples/includes/geshi/geshi/lua.php
diff --git a/examples/includes/geshi/geshi/m68k.php b/libvpx/examples/includes/geshi/geshi/m68k.php
similarity index 100%
rename from examples/includes/geshi/geshi/m68k.php
rename to libvpx/examples/includes/geshi/geshi/m68k.php
diff --git a/examples/includes/geshi/geshi/make.php b/libvpx/examples/includes/geshi/geshi/make.php
similarity index 100%
rename from examples/includes/geshi/geshi/make.php
rename to libvpx/examples/includes/geshi/geshi/make.php
diff --git a/examples/includes/geshi/geshi/matlab.php b/libvpx/examples/includes/geshi/geshi/matlab.php
similarity index 100%
rename from examples/includes/geshi/geshi/matlab.php
rename to libvpx/examples/includes/geshi/geshi/matlab.php
diff --git a/examples/includes/geshi/geshi/mirc.php b/libvpx/examples/includes/geshi/geshi/mirc.php
similarity index 100%
rename from examples/includes/geshi/geshi/mirc.php
rename to libvpx/examples/includes/geshi/geshi/mirc.php
diff --git a/examples/includes/geshi/geshi/modula3.php b/libvpx/examples/includes/geshi/geshi/modula3.php
similarity index 100%
rename from examples/includes/geshi/geshi/modula3.php
rename to libvpx/examples/includes/geshi/geshi/modula3.php
diff --git a/examples/includes/geshi/geshi/mpasm.php b/libvpx/examples/includes/geshi/geshi/mpasm.php
similarity index 100%
rename from examples/includes/geshi/geshi/mpasm.php
rename to libvpx/examples/includes/geshi/geshi/mpasm.php
diff --git a/examples/includes/geshi/geshi/mxml.php b/libvpx/examples/includes/geshi/geshi/mxml.php
similarity index 100%
rename from examples/includes/geshi/geshi/mxml.php
rename to libvpx/examples/includes/geshi/geshi/mxml.php
diff --git a/examples/includes/geshi/geshi/mysql.php b/libvpx/examples/includes/geshi/geshi/mysql.php
similarity index 100%
rename from examples/includes/geshi/geshi/mysql.php
rename to libvpx/examples/includes/geshi/geshi/mysql.php
diff --git a/examples/includes/geshi/geshi/nsis.php b/libvpx/examples/includes/geshi/geshi/nsis.php
similarity index 100%
rename from examples/includes/geshi/geshi/nsis.php
rename to libvpx/examples/includes/geshi/geshi/nsis.php
diff --git a/examples/includes/geshi/geshi/objc.php b/libvpx/examples/includes/geshi/geshi/objc.php
similarity index 100%
rename from examples/includes/geshi/geshi/objc.php
rename to libvpx/examples/includes/geshi/geshi/objc.php
diff --git a/examples/includes/geshi/geshi/ocaml-brief.php b/libvpx/examples/includes/geshi/geshi/ocaml-brief.php
similarity index 100%
rename from examples/includes/geshi/geshi/ocaml-brief.php
rename to libvpx/examples/includes/geshi/geshi/ocaml-brief.php
diff --git a/examples/includes/geshi/geshi/ocaml.php b/libvpx/examples/includes/geshi/geshi/ocaml.php
similarity index 100%
rename from examples/includes/geshi/geshi/ocaml.php
rename to libvpx/examples/includes/geshi/geshi/ocaml.php
diff --git a/examples/includes/geshi/geshi/oobas.php b/libvpx/examples/includes/geshi/geshi/oobas.php
similarity index 100%
rename from examples/includes/geshi/geshi/oobas.php
rename to libvpx/examples/includes/geshi/geshi/oobas.php
diff --git a/examples/includes/geshi/geshi/oracle11.php b/libvpx/examples/includes/geshi/geshi/oracle11.php
similarity index 100%
rename from examples/includes/geshi/geshi/oracle11.php
rename to libvpx/examples/includes/geshi/geshi/oracle11.php
diff --git a/examples/includes/geshi/geshi/oracle8.php b/libvpx/examples/includes/geshi/geshi/oracle8.php
similarity index 100%
rename from examples/includes/geshi/geshi/oracle8.php
rename to libvpx/examples/includes/geshi/geshi/oracle8.php
diff --git a/examples/includes/geshi/geshi/pascal.php b/libvpx/examples/includes/geshi/geshi/pascal.php
similarity index 100%
rename from examples/includes/geshi/geshi/pascal.php
rename to libvpx/examples/includes/geshi/geshi/pascal.php
diff --git a/examples/includes/geshi/geshi/per.php b/libvpx/examples/includes/geshi/geshi/per.php
similarity index 100%
rename from examples/includes/geshi/geshi/per.php
rename to libvpx/examples/includes/geshi/geshi/per.php
diff --git a/examples/includes/geshi/geshi/perl.php b/libvpx/examples/includes/geshi/geshi/perl.php
similarity index 100%
rename from examples/includes/geshi/geshi/perl.php
rename to libvpx/examples/includes/geshi/geshi/perl.php
diff --git a/examples/includes/geshi/geshi/php-brief.php b/libvpx/examples/includes/geshi/geshi/php-brief.php
similarity index 100%
rename from examples/includes/geshi/geshi/php-brief.php
rename to libvpx/examples/includes/geshi/geshi/php-brief.php
diff --git a/examples/includes/geshi/geshi/php.php b/libvpx/examples/includes/geshi/geshi/php.php
similarity index 100%
rename from examples/includes/geshi/geshi/php.php
rename to libvpx/examples/includes/geshi/geshi/php.php
diff --git a/examples/includes/geshi/geshi/pic16.php b/libvpx/examples/includes/geshi/geshi/pic16.php
similarity index 100%
rename from examples/includes/geshi/geshi/pic16.php
rename to libvpx/examples/includes/geshi/geshi/pic16.php
diff --git a/examples/includes/geshi/geshi/pixelbender.php b/libvpx/examples/includes/geshi/geshi/pixelbender.php
similarity index 100%
rename from examples/includes/geshi/geshi/pixelbender.php
rename to libvpx/examples/includes/geshi/geshi/pixelbender.php
diff --git a/examples/includes/geshi/geshi/plsql.php b/libvpx/examples/includes/geshi/geshi/plsql.php
similarity index 100%
rename from examples/includes/geshi/geshi/plsql.php
rename to libvpx/examples/includes/geshi/geshi/plsql.php
diff --git a/examples/includes/geshi/geshi/povray.php b/libvpx/examples/includes/geshi/geshi/povray.php
similarity index 100%
rename from examples/includes/geshi/geshi/povray.php
rename to libvpx/examples/includes/geshi/geshi/povray.php
diff --git a/examples/includes/geshi/geshi/powershell.php b/libvpx/examples/includes/geshi/geshi/powershell.php
similarity index 100%
rename from examples/includes/geshi/geshi/powershell.php
rename to libvpx/examples/includes/geshi/geshi/powershell.php
diff --git a/examples/includes/geshi/geshi/progress.php b/libvpx/examples/includes/geshi/geshi/progress.php
similarity index 100%
rename from examples/includes/geshi/geshi/progress.php
rename to libvpx/examples/includes/geshi/geshi/progress.php
diff --git a/examples/includes/geshi/geshi/prolog.php b/libvpx/examples/includes/geshi/geshi/prolog.php
similarity index 100%
rename from examples/includes/geshi/geshi/prolog.php
rename to libvpx/examples/includes/geshi/geshi/prolog.php
diff --git a/examples/includes/geshi/geshi/providex.php b/libvpx/examples/includes/geshi/geshi/providex.php
similarity index 100%
rename from examples/includes/geshi/geshi/providex.php
rename to libvpx/examples/includes/geshi/geshi/providex.php
diff --git a/examples/includes/geshi/geshi/python.php b/libvpx/examples/includes/geshi/geshi/python.php
similarity index 100%
rename from examples/includes/geshi/geshi/python.php
rename to libvpx/examples/includes/geshi/geshi/python.php
diff --git a/examples/includes/geshi/geshi/qbasic.php b/libvpx/examples/includes/geshi/geshi/qbasic.php
similarity index 100%
rename from examples/includes/geshi/geshi/qbasic.php
rename to libvpx/examples/includes/geshi/geshi/qbasic.php
diff --git a/examples/includes/geshi/geshi/rails.php b/libvpx/examples/includes/geshi/geshi/rails.php
similarity index 100%
rename from examples/includes/geshi/geshi/rails.php
rename to libvpx/examples/includes/geshi/geshi/rails.php
diff --git a/examples/includes/geshi/geshi/rebol.php b/libvpx/examples/includes/geshi/geshi/rebol.php
similarity index 100%
rename from examples/includes/geshi/geshi/rebol.php
rename to libvpx/examples/includes/geshi/geshi/rebol.php
diff --git a/examples/includes/geshi/geshi/reg.php b/libvpx/examples/includes/geshi/geshi/reg.php
similarity index 100%
rename from examples/includes/geshi/geshi/reg.php
rename to libvpx/examples/includes/geshi/geshi/reg.php
diff --git a/examples/includes/geshi/geshi/robots.php b/libvpx/examples/includes/geshi/geshi/robots.php
similarity index 100%
rename from examples/includes/geshi/geshi/robots.php
rename to libvpx/examples/includes/geshi/geshi/robots.php
diff --git a/examples/includes/geshi/geshi/ruby.php b/libvpx/examples/includes/geshi/geshi/ruby.php
similarity index 100%
rename from examples/includes/geshi/geshi/ruby.php
rename to libvpx/examples/includes/geshi/geshi/ruby.php
diff --git a/examples/includes/geshi/geshi/sas.php b/libvpx/examples/includes/geshi/geshi/sas.php
similarity index 100%
rename from examples/includes/geshi/geshi/sas.php
rename to libvpx/examples/includes/geshi/geshi/sas.php
diff --git a/examples/includes/geshi/geshi/scala.php b/libvpx/examples/includes/geshi/geshi/scala.php
similarity index 100%
rename from examples/includes/geshi/geshi/scala.php
rename to libvpx/examples/includes/geshi/geshi/scala.php
diff --git a/examples/includes/geshi/geshi/scheme.php b/libvpx/examples/includes/geshi/geshi/scheme.php
similarity index 100%
rename from examples/includes/geshi/geshi/scheme.php
rename to libvpx/examples/includes/geshi/geshi/scheme.php
diff --git a/examples/includes/geshi/geshi/scilab.php b/libvpx/examples/includes/geshi/geshi/scilab.php
similarity index 100%
rename from examples/includes/geshi/geshi/scilab.php
rename to libvpx/examples/includes/geshi/geshi/scilab.php
diff --git a/examples/includes/geshi/geshi/sdlbasic.php b/libvpx/examples/includes/geshi/geshi/sdlbasic.php
similarity index 100%
rename from examples/includes/geshi/geshi/sdlbasic.php
rename to libvpx/examples/includes/geshi/geshi/sdlbasic.php
diff --git a/examples/includes/geshi/geshi/smalltalk.php b/libvpx/examples/includes/geshi/geshi/smalltalk.php
similarity index 100%
rename from examples/includes/geshi/geshi/smalltalk.php
rename to libvpx/examples/includes/geshi/geshi/smalltalk.php
diff --git a/examples/includes/geshi/geshi/smarty.php b/libvpx/examples/includes/geshi/geshi/smarty.php
similarity index 100%
rename from examples/includes/geshi/geshi/smarty.php
rename to libvpx/examples/includes/geshi/geshi/smarty.php
diff --git a/examples/includes/geshi/geshi/sql.php b/libvpx/examples/includes/geshi/geshi/sql.php
similarity index 100%
rename from examples/includes/geshi/geshi/sql.php
rename to libvpx/examples/includes/geshi/geshi/sql.php
diff --git a/examples/includes/geshi/geshi/tcl.php b/libvpx/examples/includes/geshi/geshi/tcl.php
similarity index 100%
rename from examples/includes/geshi/geshi/tcl.php
rename to libvpx/examples/includes/geshi/geshi/tcl.php
diff --git a/examples/includes/geshi/geshi/teraterm.php b/libvpx/examples/includes/geshi/geshi/teraterm.php
similarity index 100%
rename from examples/includes/geshi/geshi/teraterm.php
rename to libvpx/examples/includes/geshi/geshi/teraterm.php
diff --git a/examples/includes/geshi/geshi/text.php b/libvpx/examples/includes/geshi/geshi/text.php
similarity index 100%
rename from examples/includes/geshi/geshi/text.php
rename to libvpx/examples/includes/geshi/geshi/text.php
diff --git a/examples/includes/geshi/geshi/thinbasic.php b/libvpx/examples/includes/geshi/geshi/thinbasic.php
similarity index 100%
rename from examples/includes/geshi/geshi/thinbasic.php
rename to libvpx/examples/includes/geshi/geshi/thinbasic.php
diff --git a/examples/includes/geshi/geshi/tsql.php b/libvpx/examples/includes/geshi/geshi/tsql.php
similarity index 100%
rename from examples/includes/geshi/geshi/tsql.php
rename to libvpx/examples/includes/geshi/geshi/tsql.php
diff --git a/examples/includes/geshi/geshi/typoscript.php b/libvpx/examples/includes/geshi/geshi/typoscript.php
similarity index 100%
rename from examples/includes/geshi/geshi/typoscript.php
rename to libvpx/examples/includes/geshi/geshi/typoscript.php
diff --git a/examples/includes/geshi/geshi/vb.php b/libvpx/examples/includes/geshi/geshi/vb.php
similarity index 100%
rename from examples/includes/geshi/geshi/vb.php
rename to libvpx/examples/includes/geshi/geshi/vb.php
diff --git a/examples/includes/geshi/geshi/vbnet.php b/libvpx/examples/includes/geshi/geshi/vbnet.php
similarity index 100%
rename from examples/includes/geshi/geshi/vbnet.php
rename to libvpx/examples/includes/geshi/geshi/vbnet.php
diff --git a/examples/includes/geshi/geshi/verilog.php b/libvpx/examples/includes/geshi/geshi/verilog.php
similarity index 100%
rename from examples/includes/geshi/geshi/verilog.php
rename to libvpx/examples/includes/geshi/geshi/verilog.php
diff --git a/examples/includes/geshi/geshi/vhdl.php b/libvpx/examples/includes/geshi/geshi/vhdl.php
similarity index 100%
rename from examples/includes/geshi/geshi/vhdl.php
rename to libvpx/examples/includes/geshi/geshi/vhdl.php
diff --git a/examples/includes/geshi/geshi/vim.php b/libvpx/examples/includes/geshi/geshi/vim.php
similarity index 100%
rename from examples/includes/geshi/geshi/vim.php
rename to libvpx/examples/includes/geshi/geshi/vim.php
diff --git a/examples/includes/geshi/geshi/visualfoxpro.php b/libvpx/examples/includes/geshi/geshi/visualfoxpro.php
similarity index 100%
rename from examples/includes/geshi/geshi/visualfoxpro.php
rename to libvpx/examples/includes/geshi/geshi/visualfoxpro.php
diff --git a/examples/includes/geshi/geshi/visualprolog.php b/libvpx/examples/includes/geshi/geshi/visualprolog.php
similarity index 100%
rename from examples/includes/geshi/geshi/visualprolog.php
rename to libvpx/examples/includes/geshi/geshi/visualprolog.php
diff --git a/examples/includes/geshi/geshi/whitespace.php b/libvpx/examples/includes/geshi/geshi/whitespace.php
similarity index 100%
rename from examples/includes/geshi/geshi/whitespace.php
rename to libvpx/examples/includes/geshi/geshi/whitespace.php
diff --git a/examples/includes/geshi/geshi/winbatch.php b/libvpx/examples/includes/geshi/geshi/winbatch.php
similarity index 100%
rename from examples/includes/geshi/geshi/winbatch.php
rename to libvpx/examples/includes/geshi/geshi/winbatch.php
diff --git a/examples/includes/geshi/geshi/xml.php b/libvpx/examples/includes/geshi/geshi/xml.php
similarity index 100%
rename from examples/includes/geshi/geshi/xml.php
rename to libvpx/examples/includes/geshi/geshi/xml.php
diff --git a/examples/includes/geshi/geshi/xorg_conf.php b/libvpx/examples/includes/geshi/geshi/xorg_conf.php
similarity index 100%
rename from examples/includes/geshi/geshi/xorg_conf.php
rename to libvpx/examples/includes/geshi/geshi/xorg_conf.php
diff --git a/examples/includes/geshi/geshi/xpp.php b/libvpx/examples/includes/geshi/geshi/xpp.php
similarity index 100%
rename from examples/includes/geshi/geshi/xpp.php
rename to libvpx/examples/includes/geshi/geshi/xpp.php
diff --git a/examples/includes/geshi/geshi/z80.php b/libvpx/examples/includes/geshi/geshi/z80.php
similarity index 100%
rename from examples/includes/geshi/geshi/z80.php
rename to libvpx/examples/includes/geshi/geshi/z80.php
diff --git a/examples/includes/vp8_doc_tools.php b/libvpx/examples/includes/vp8_doc_tools.php
similarity index 100%
rename from examples/includes/vp8_doc_tools.php
rename to libvpx/examples/includes/vp8_doc_tools.php
diff --git a/libvpx/examples/postproc.txt b/libvpx/examples/postproc.txt
new file mode 100644
index 0000000..51b251a
--- /dev/null
+++ b/libvpx/examples/postproc.txt
@@ -0,0 +1,67 @@
+@TEMPLATE decoder_tmpl.c
+Postprocessing Decoder
+======================
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
+This example adds postprocessing to the simple decoder loop.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
+
+
+Initializing Postprocessing
+---------------------------
+You must inform the codec that you might request postprocessing at
+initialization time. This is done by passing the VPX_CODEC_USE_POSTPROC
+flag to `vpx_codec_dec_init`. If the codec does not support
+postprocessing, this call will return VPX_CODEC_INCAPABLE. For
+demonstration purposes, we also fall back to default initialization if
+the codec does not provide support.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
+/* Initialize codec */
+res = vpx_codec_dec_init(&codec, interface, NULL,
+ VPX_CODEC_USE_POSTPROC);
+if(res == VPX_CODEC_INCAPABLE) {
+ printf("NOTICE: Postproc not supported by %s\n",
+ vpx_codec_iface_name(interface));
+ res = vpx_codec_dec_init(&codec, interface, NULL, flags);
+}
+if(res)
+ die_codec(&codec, "Failed to initialize decoder");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INIT
+
+
+Using Adaptive Postprocessing
+-----------------------------
+VP6 provides "adaptive postprocessing." It will automatically select the
+best postprocessing filter on a frame by frame basis based on the amount
+of time remaining before the user's specified deadline expires. The
+special value 0 indicates that the codec should take as long as
+necessary to provide the best quality frame. This example gives the
+codec 15ms (15000us) to return a frame. Remember that this is a soft
+deadline, and the codec may exceed it doing its regular processing. In
+these cases, no additional postprocessing will be done.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DECODE
+/* Decode the frame with 15ms deadline */
+if(vpx_codec_decode(&codec, frame, frame_sz, NULL, 15000))
+ die_codec(&codec, "Failed to decode frame");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DECODE
+
+
+Codec Specific Postprocessing Controls
+--------------------------------------
+Some codecs provide fine grained controls over their built-in
+postprocessors. VP8 is one example. The following sample code toggles
+postprocessing on and off every 15 frames.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PRE_DECODE
+#if CONFIG_VP8_DECODER
+if(frame_cnt%30 == 1) {
+ vp8_postproc_cfg_t pp = {0, 0, 0};
+
+ if(vpx_codec_control(&codec, VP8_SET_POSTPROC, &pp))
+ die_codec(&codec, "Failed to turn off postproc");
+} else if(frame_cnt%30 == 16) {
+ vp8_postproc_cfg_t pp = {VP8_DEBLOCK | VP8_DEMACROBLOCK | VP8_MFQE, 4, 0};
+
+ if(vpx_codec_control(&codec, VP8_SET_POSTPROC, &pp))
+ die_codec(&codec, "Failed to turn on postproc");
+};
+#endif
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PRE_DECODE
diff --git a/examples/simple_decoder.txt b/libvpx/examples/simple_decoder.txt
similarity index 100%
rename from examples/simple_decoder.txt
rename to libvpx/examples/simple_decoder.txt
diff --git a/examples/simple_encoder.txt b/libvpx/examples/simple_encoder.txt
similarity index 100%
rename from examples/simple_encoder.txt
rename to libvpx/examples/simple_encoder.txt
diff --git a/libvpx/examples/twopass_encoder.txt b/libvpx/examples/twopass_encoder.txt
new file mode 100644
index 0000000..2f81a90
--- /dev/null
+++ b/libvpx/examples/twopass_encoder.txt
@@ -0,0 +1,87 @@
+@TEMPLATE encoder_tmpl.c
+Two Pass Encoder
+================
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
+This is an example of a two pass encoder loop. It takes an input file in
+YV12 format, passes it through the encoder twice, and writes the compressed
+frames to disk in IVF format. It builds upon the simple_encoder example.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INTRODUCTION
+
+
+Twopass Variables
+-----------------
+Twopass mode needs to track the current pass number and the buffer of
+statistics packets.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_VARS
+int pass;
+vpx_fixed_buf_t stats = {0};
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_VARS
+
+
+Updating The Configuration
+---------------------------------
+In two pass mode, the configuration has to be updated on each pass. The
+statistics buffer is passed on the last pass.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_LOOP_BEGIN
+for(pass=0; pass<2; pass++) {
+ frame_cnt = 0;
+
+ if(pass == 0)
+ cfg.g_pass = VPX_RC_FIRST_PASS;
+ else {
+ cfg.g_pass = VPX_RC_LAST_PASS;
+ cfg.rc_twopass_stats_in = stats;
+ }
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_LOOP_BEGIN
+
+
+Encoding A Frame
+----------------
+Encoding a frame in two pass mode is identical to the simple encoder
+example, except the deadline is set to VPX_DL_BEST_QUALITY to get the
+best quality possible. VPX_DL_GOOD_QUALITY could also be used.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENCODE_FRAME
+frame_avail = read_frame(infile, &raw);
+if(vpx_codec_encode(&codec, frame_avail? &raw : NULL, frame_cnt,
+ 1, flags, VPX_DL_BEST_QUALITY))
+ die_codec(&codec, "Failed to encode frame");
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENCODE_FRAME
+
+
+Processing Statistics Packets
+-----------------------------
+Each packet of type `VPX_CODEC_CX_FRAME_PKT` contains the encoded data
+for this frame. We write a IVF frame header, followed by the raw data.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_STATS
+case VPX_CODEC_STATS_PKT:
+ stats.buf = realloc(stats.buf, stats.sz
+ + pkt->data.twopass_stats.sz);
+ if(!stats.buf)
+ die("Memory reallocation failed.\n");
+ memcpy((char*)stats.buf + stats.sz,
+ pkt->data.twopass_stats.buf,
+ pkt->data.twopass_stats.sz);
+ stats.sz += pkt->data.twopass_stats.sz;
+ break;
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_STATS
+
+
+Pass Progress Reporting
+-----------------------------
+It's sometimes helpful to see when each pass completes.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_LOOP_END
+ printf("Pass %d complete.\n", pass+1);
+ if(vpx_codec_destroy(&codec))
+ die_codec(&codec, "Failed to destroy codec");
+}
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ TWOPASS_LOOP_END
+
+
+Clean-up
+-----------------------------
+Destruction of the encoder instance must be done on each pass. The
+raw image should be destroyed at the end as usual.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
+vpx_img_free(&raw);
+free(stats.buf);
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DESTROY
diff --git a/examples/vp8_set_maps.txt b/libvpx/examples/vp8_set_maps.txt
similarity index 100%
rename from examples/vp8_set_maps.txt
rename to libvpx/examples/vp8_set_maps.txt
diff --git a/examples/vp8cx_set_ref.txt b/libvpx/examples/vp8cx_set_ref.txt
similarity index 100%
rename from examples/vp8cx_set_ref.txt
rename to libvpx/examples/vp8cx_set_ref.txt
diff --git a/keywords.dox b/libvpx/keywords.dox
similarity index 100%
rename from keywords.dox
rename to libvpx/keywords.dox
diff --git a/libmkv/EbmlBufferWriter.c b/libvpx/libmkv/EbmlBufferWriter.c
similarity index 100%
rename from libmkv/EbmlBufferWriter.c
rename to libvpx/libmkv/EbmlBufferWriter.c
diff --git a/libmkv/EbmlBufferWriter.h b/libvpx/libmkv/EbmlBufferWriter.h
similarity index 100%
rename from libmkv/EbmlBufferWriter.h
rename to libvpx/libmkv/EbmlBufferWriter.h
diff --git a/libvpx/libmkv/EbmlIDs.h b/libvpx/libmkv/EbmlIDs.h
new file mode 100644
index 0000000..e3ce585
--- /dev/null
+++ b/libvpx/libmkv/EbmlIDs.h
@@ -0,0 +1,231 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#ifndef MKV_DEFS_HPP
+#define MKV_DEFS_HPP 1
+
+/* Commenting out values not available in webm, but available in matroska */
+
+enum mkv
+{
+ EBML = 0x1A45DFA3,
+ EBMLVersion = 0x4286,
+ EBMLReadVersion = 0x42F7,
+ EBMLMaxIDLength = 0x42F2,
+ EBMLMaxSizeLength = 0x42F3,
+ DocType = 0x4282,
+ DocTypeVersion = 0x4287,
+ DocTypeReadVersion = 0x4285,
+/* CRC_32 = 0xBF, */
+ Void = 0xEC,
+ SignatureSlot = 0x1B538667,
+ SignatureAlgo = 0x7E8A,
+ SignatureHash = 0x7E9A,
+ SignaturePublicKey = 0x7EA5,
+ Signature = 0x7EB5,
+ SignatureElements = 0x7E5B,
+ SignatureElementList = 0x7E7B,
+ SignedElement = 0x6532,
+ /* segment */
+ Segment = 0x18538067,
+ /* Meta Seek Information */
+ SeekHead = 0x114D9B74,
+ Seek = 0x4DBB,
+ SeekID = 0x53AB,
+ SeekPosition = 0x53AC,
+ /* Segment Information */
+ Info = 0x1549A966,
+/* SegmentUID = 0x73A4, */
+/* SegmentFilename = 0x7384, */
+/* PrevUID = 0x3CB923, */
+/* PrevFilename = 0x3C83AB, */
+/* NextUID = 0x3EB923, */
+/* NextFilename = 0x3E83BB, */
+/* SegmentFamily = 0x4444, */
+/* ChapterTranslate = 0x6924, */
+/* ChapterTranslateEditionUID = 0x69FC, */
+/* ChapterTranslateCodec = 0x69BF, */
+/* ChapterTranslateID = 0x69A5, */
+ TimecodeScale = 0x2AD7B1,
+ Segment_Duration = 0x4489,
+ DateUTC = 0x4461,
+/* Title = 0x7BA9, */
+ MuxingApp = 0x4D80,
+ WritingApp = 0x5741,
+ /* Cluster */
+ Cluster = 0x1F43B675,
+ Timecode = 0xE7,
+/* SilentTracks = 0x5854, */
+/* SilentTrackNumber = 0x58D7, */
+/* Position = 0xA7, */
+ PrevSize = 0xAB,
+ BlockGroup = 0xA0,
+ Block = 0xA1,
+/* BlockVirtual = 0xA2, */
+/* BlockAdditions = 0x75A1, */
+/* BlockMore = 0xA6, */
+/* BlockAddID = 0xEE, */
+/* BlockAdditional = 0xA5, */
+ BlockDuration = 0x9B,
+/* ReferencePriority = 0xFA, */
+ ReferenceBlock = 0xFB,
+/* ReferenceVirtual = 0xFD, */
+/* CodecState = 0xA4, */
+/* Slices = 0x8E, */
+/* TimeSlice = 0xE8, */
+ LaceNumber = 0xCC,
+/* FrameNumber = 0xCD, */
+/* BlockAdditionID = 0xCB, */
+/* MkvDelay = 0xCE, */
+/* Cluster_Duration = 0xCF, */
+ SimpleBlock = 0xA3,
+/* EncryptedBlock = 0xAF, */
+ /* Track */
+ Tracks = 0x1654AE6B,
+ TrackEntry = 0xAE,
+ TrackNumber = 0xD7,
+ TrackUID = 0x73C5,
+ TrackType = 0x83,
+ FlagEnabled = 0xB9,
+ FlagDefault = 0x88,
+ FlagForced = 0x55AA,
+ FlagLacing = 0x9C,
+/* MinCache = 0x6DE7, */
+/* MaxCache = 0x6DF8, */
+ DefaultDuration = 0x23E383,
+/* TrackTimecodeScale = 0x23314F, */
+/* TrackOffset = 0x537F, */
+/* MaxBlockAdditionID = 0x55EE, */
+ Name = 0x536E,
+ Language = 0x22B59C,
+ CodecID = 0x86,
+ CodecPrivate = 0x63A2,
+ CodecName = 0x258688,
+/* AttachmentLink = 0x7446, */
+/* CodecSettings = 0x3A9697, */
+/* CodecInfoURL = 0x3B4040, */
+/* CodecDownloadURL = 0x26B240, */
+/* CodecDecodeAll = 0xAA, */
+/* TrackOverlay = 0x6FAB, */
+/* TrackTranslate = 0x6624, */
+/* TrackTranslateEditionUID = 0x66FC, */
+/* TrackTranslateCodec = 0x66BF, */
+/* TrackTranslateTrackID = 0x66A5, */
+ /* video */
+ Video = 0xE0,
+ FlagInterlaced = 0x9A,
+ StereoMode = 0x53B8,
+ PixelWidth = 0xB0,
+ PixelHeight = 0xBA,
+ PixelCropBottom = 0x54AA,
+ PixelCropTop = 0x54BB,
+ PixelCropLeft = 0x54CC,
+ PixelCropRight = 0x54DD,
+ DisplayWidth = 0x54B0,
+ DisplayHeight = 0x54BA,
+ DisplayUnit = 0x54B2,
+ AspectRatioType = 0x54B3,
+/* ColourSpace = 0x2EB524, */
+/* GammaValue = 0x2FB523, */
+ FrameRate = 0x2383E3,
+ /* end video */
+ /* audio */
+ Audio = 0xE1,
+ SamplingFrequency = 0xB5,
+ OutputSamplingFrequency = 0x78B5,
+ Channels = 0x9F,
+/* ChannelPositions = 0x7D7B, */
+ BitDepth = 0x6264,
+ /* end audio */
+ /* content encoding */
+/* ContentEncodings = 0x6d80, */
+/* ContentEncoding = 0x6240, */
+/* ContentEncodingOrder = 0x5031, */
+/* ContentEncodingScope = 0x5032, */
+/* ContentEncodingType = 0x5033, */
+/* ContentCompression = 0x5034, */
+/* ContentCompAlgo = 0x4254, */
+/* ContentCompSettings = 0x4255, */
+/* ContentEncryption = 0x5035, */
+/* ContentEncAlgo = 0x47e1, */
+/* ContentEncKeyID = 0x47e2, */
+/* ContentSignature = 0x47e3, */
+/* ContentSigKeyID = 0x47e4, */
+/* ContentSigAlgo = 0x47e5, */
+/* ContentSigHashAlgo = 0x47e6, */
+ /* end content encoding */
+ /* Cueing Data */
+ Cues = 0x1C53BB6B,
+ CuePoint = 0xBB,
+ CueTime = 0xB3,
+ CueTrackPositions = 0xB7,
+ CueTrack = 0xF7,
+ CueClusterPosition = 0xF1,
+ CueBlockNumber = 0x5378
+/* CueCodecState = 0xEA, */
+/* CueReference = 0xDB, */
+/* CueRefTime = 0x96, */
+/* CueRefCluster = 0x97, */
+/* CueRefNumber = 0x535F, */
+/* CueRefCodecState = 0xEB, */
+ /* Attachment */
+/* Attachments = 0x1941A469, */
+/* AttachedFile = 0x61A7, */
+/* FileDescription = 0x467E, */
+/* FileName = 0x466E, */
+/* FileMimeType = 0x4660, */
+/* FileData = 0x465C, */
+/* FileUID = 0x46AE, */
+/* FileReferral = 0x4675, */
+ /* Chapters */
+/* Chapters = 0x1043A770, */
+/* EditionEntry = 0x45B9, */
+/* EditionUID = 0x45BC, */
+/* EditionFlagHidden = 0x45BD, */
+/* EditionFlagDefault = 0x45DB, */
+/* EditionFlagOrdered = 0x45DD, */
+/* ChapterAtom = 0xB6, */
+/* ChapterUID = 0x73C4, */
+/* ChapterTimeStart = 0x91, */
+/* ChapterTimeEnd = 0x92, */
+/* ChapterFlagHidden = 0x98, */
+/* ChapterFlagEnabled = 0x4598, */
+/* ChapterSegmentUID = 0x6E67, */
+/* ChapterSegmentEditionUID = 0x6EBC, */
+/* ChapterPhysicalEquiv = 0x63C3, */
+/* ChapterTrack = 0x8F, */
+/* ChapterTrackNumber = 0x89, */
+/* ChapterDisplay = 0x80, */
+/* ChapString = 0x85, */
+/* ChapLanguage = 0x437C, */
+/* ChapCountry = 0x437E, */
+/* ChapProcess = 0x6944, */
+/* ChapProcessCodecID = 0x6955, */
+/* ChapProcessPrivate = 0x450D, */
+/* ChapProcessCommand = 0x6911, */
+/* ChapProcessTime = 0x6922, */
+/* ChapProcessData = 0x6933, */
+ /* Tagging */
+/* Tags = 0x1254C367, */
+/* Tag = 0x7373, */
+/* Targets = 0x63C0, */
+/* TargetTypeValue = 0x68CA, */
+/* TargetType = 0x63CA, */
+/* Tagging_TrackUID = 0x63C5, */
+/* Tagging_EditionUID = 0x63C9, */
+/* Tagging_ChapterUID = 0x63C4, */
+/* AttachmentUID = 0x63C6, */
+/* SimpleTag = 0x67C8, */
+/* TagName = 0x45A3, */
+/* TagLanguage = 0x447A, */
+/* TagDefault = 0x4484, */
+/* TagString = 0x4487, */
+/* TagBinary = 0x4485, */
+};
+#endif
diff --git a/libvpx/libmkv/EbmlWriter.c b/libvpx/libmkv/EbmlWriter.c
new file mode 100644
index 0000000..d70f06e
--- /dev/null
+++ b/libvpx/libmkv/EbmlWriter.c
@@ -0,0 +1,175 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include "EbmlWriter.h"
+#include <stdlib.h>
+#include <wchar.h>
+#include <string.h>
+#include <limits.h>
+#if defined(_MSC_VER)
+#define LITERALU64(n) n
+#else
+#define LITERALU64(n) n##LLU
+#endif
+
+void Ebml_WriteLen(EbmlGlobal *glob, int64_t val)
+{
+ /* TODO check and make sure we are not > than 0x0100000000000000LLU */
+ unsigned char size = 8; /* size in bytes to output */
+
+ /* mask to compare for byte size */
+ int64_t minVal = 0xff;
+
+ for (size = 1; size < 8; size ++)
+ {
+ if (val < minVal)
+ break;
+
+ minVal = (minVal << 7);
+ }
+
+ val |= (((uint64_t)0x80) << ((size - 1) * 7));
+
+ Ebml_Serialize(glob, (void *) &val, sizeof(val), size);
+}
+
+void Ebml_WriteString(EbmlGlobal *glob, const char *str)
+{
+ const size_t size_ = strlen(str);
+ const uint64_t size = size_;
+ Ebml_WriteLen(glob, size);
+ /* TODO: it's not clear from the spec whether the nul terminator
+ * should be serialized too. For now we omit the null terminator.
+ */
+ Ebml_Write(glob, str, (unsigned long)size);
+}
+
+void Ebml_WriteUTF8(EbmlGlobal *glob, const wchar_t *wstr)
+{
+ const size_t strlen = wcslen(wstr);
+
+ /* TODO: it's not clear from the spec whether the nul terminator
+ * should be serialized too. For now we include it.
+ */
+ const uint64_t size = strlen;
+
+ Ebml_WriteLen(glob, size);
+ Ebml_Write(glob, wstr, (unsigned long)size);
+}
+
+void Ebml_WriteID(EbmlGlobal *glob, unsigned long class_id)
+{
+ int len;
+
+ if (class_id >= 0x01000000)
+ len = 4;
+ else if (class_id >= 0x00010000)
+ len = 3;
+ else if (class_id >= 0x00000100)
+ len = 2;
+ else
+ len = 1;
+
+ Ebml_Serialize(glob, (void *)&class_id, sizeof(class_id), len);
+}
+
+void Ebml_SerializeUnsigned64(EbmlGlobal *glob, unsigned long class_id, uint64_t ui)
+{
+ unsigned char sizeSerialized = 8 | 0x80;
+ Ebml_WriteID(glob, class_id);
+ Ebml_Serialize(glob, &sizeSerialized, sizeof(sizeSerialized), 1);
+ Ebml_Serialize(glob, &ui, sizeof(ui), 8);
+}
+
+void Ebml_SerializeUnsigned(EbmlGlobal *glob, unsigned long class_id, unsigned long ui)
+{
+ unsigned char size = 8; /* size in bytes to output */
+ unsigned char sizeSerialized = 0;
+ unsigned long minVal;
+
+ Ebml_WriteID(glob, class_id);
+ minVal = 0x7fLU; /* mask to compare for byte size */
+
+ for (size = 1; size < 4; size ++)
+ {
+ if (ui < minVal)
+ {
+ break;
+ }
+
+ minVal <<= 7;
+ }
+
+ sizeSerialized = 0x80 | size;
+ Ebml_Serialize(glob, &sizeSerialized, sizeof(sizeSerialized), 1);
+ Ebml_Serialize(glob, &ui, sizeof(ui), size);
+}
+/* TODO: perhaps this is a poor name for this id serializer helper function */
+void Ebml_SerializeBinary(EbmlGlobal *glob, unsigned long class_id, unsigned long bin)
+{
+ int size;
+ for (size=4; size > 1; size--)
+ {
+ if (bin & 0x000000ff << ((size-1) * 8))
+ break;
+ }
+ Ebml_WriteID(glob, class_id);
+ Ebml_WriteLen(glob, size);
+ Ebml_WriteID(glob, bin);
+}
+
+void Ebml_SerializeFloat(EbmlGlobal *glob, unsigned long class_id, double d)
+{
+ unsigned char len = 0x88;
+
+ Ebml_WriteID(glob, class_id);
+ Ebml_Serialize(glob, &len, sizeof(len), 1);
+ Ebml_Serialize(glob, &d, sizeof(d), 8);
+}
+
+void Ebml_WriteSigned16(EbmlGlobal *glob, short val)
+{
+ signed long out = ((val & 0x003FFFFF) | 0x00200000) << 8;
+ Ebml_Serialize(glob, &out, sizeof(out), 3);
+}
+
+void Ebml_SerializeString(EbmlGlobal *glob, unsigned long class_id, const char *s)
+{
+ Ebml_WriteID(glob, class_id);
+ Ebml_WriteString(glob, s);
+}
+
+void Ebml_SerializeUTF8(EbmlGlobal *glob, unsigned long class_id, wchar_t *s)
+{
+ Ebml_WriteID(glob, class_id);
+ Ebml_WriteUTF8(glob, s);
+}
+
+void Ebml_SerializeData(EbmlGlobal *glob, unsigned long class_id, unsigned char *data, unsigned long data_length)
+{
+ Ebml_WriteID(glob, class_id);
+ Ebml_WriteLen(glob, data_length);
+ Ebml_Write(glob, data, data_length);
+}
+
+void Ebml_WriteVoid(EbmlGlobal *glob, unsigned long vSize)
+{
+ unsigned char tmp = 0;
+ unsigned long i = 0;
+
+ Ebml_WriteID(glob, 0xEC);
+ Ebml_WriteLen(glob, vSize);
+
+ for (i = 0; i < vSize; i++)
+ {
+ Ebml_Write(glob, &tmp, 1);
+ }
+}
+
+/* TODO Serialize Date */
diff --git a/libvpx/libmkv/EbmlWriter.h b/libvpx/libmkv/EbmlWriter.h
new file mode 100644
index 0000000..b94f757
--- /dev/null
+++ b/libvpx/libmkv/EbmlWriter.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#ifndef EBMLWRITER_HPP
+#define EBMLWRITER_HPP
+#include <stddef.h>
+#include "vpx/vpx_integer.h"
+
+/* note: you must define write and serialize functions as well as your own
+ * EBML_GLOBAL
+ *
+ * These functions MUST be implemented
+ */
+
+typedef struct EbmlGlobal EbmlGlobal;
+void Ebml_Serialize(EbmlGlobal *glob, const void *, int, unsigned long);
+void Ebml_Write(EbmlGlobal *glob, const void *, unsigned long);
+
+/*****/
+
+void Ebml_WriteLen(EbmlGlobal *glob, int64_t val);
+void Ebml_WriteString(EbmlGlobal *glob, const char *str);
+void Ebml_WriteUTF8(EbmlGlobal *glob, const wchar_t *wstr);
+void Ebml_WriteID(EbmlGlobal *glob, unsigned long class_id);
+void Ebml_SerializeUnsigned64(EbmlGlobal *glob, unsigned long class_id, uint64_t ui);
+void Ebml_SerializeUnsigned(EbmlGlobal *glob, unsigned long class_id, unsigned long ui);
+void Ebml_SerializeBinary(EbmlGlobal *glob, unsigned long class_id, unsigned long ui);
+void Ebml_SerializeFloat(EbmlGlobal *glob, unsigned long class_id, double d);
+/* TODO make this more generic to signed */
+void Ebml_WriteSigned16(EbmlGlobal *glob, short val);
+void Ebml_SerializeString(EbmlGlobal *glob, unsigned long class_id, const char *s);
+void Ebml_SerializeUTF8(EbmlGlobal *glob, unsigned long class_id, wchar_t *s);
+void Ebml_SerializeData(EbmlGlobal *glob, unsigned long class_id, unsigned char *data, unsigned long data_length);
+void Ebml_WriteVoid(EbmlGlobal *glob, unsigned long vSize);
+/* TODO need date function */
+#endif
diff --git a/libmkv/Makefile b/libvpx/libmkv/Makefile
similarity index 100%
rename from libmkv/Makefile
rename to libvpx/libmkv/Makefile
diff --git a/libvpx/libmkv/WebMElement.c b/libvpx/libmkv/WebMElement.c
new file mode 100644
index 0000000..0ef5100
--- /dev/null
+++ b/libvpx/libmkv/WebMElement.c
@@ -0,0 +1,220 @@
+// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+//
+// Use of this source code is governed by a BSD-style license
+// that can be found in the LICENSE file in the root of the source
+// tree. An additional intellectual property rights grant can be found
+// in the file PATENTS. All contributing project authors may
+// be found in the AUTHORS file in the root of the source tree.
+
+
+#include "EbmlBufferWriter.h"
+#include "EbmlIDs.h"
+#include "WebMElement.h"
+#include <stdio.h>
+
+#define kVorbisPrivateMaxSize 4000
+
+void writeHeader(EbmlGlobal *glob)
+{
+ EbmlLoc start;
+ Ebml_StartSubElement(glob, &start, EBML);
+ Ebml_SerializeUnsigned(glob, EBMLVersion, 1);
+ Ebml_SerializeUnsigned(glob, EBMLReadVersion, 1); //EBML Read Version
+ Ebml_SerializeUnsigned(glob, EBMLMaxIDLength, 4); //EBML Max ID Length
+ Ebml_SerializeUnsigned(glob, EBMLMaxSizeLength, 8); //EBML Max Size Length
+ Ebml_SerializeString(glob, DocType, "webm"); //Doc Type
+ Ebml_SerializeUnsigned(glob, DocTypeVersion, 2); //Doc Type Version
+ Ebml_SerializeUnsigned(glob, DocTypeReadVersion, 2); //Doc Type Read Version
+ Ebml_EndSubElement(glob, &start);
+}
+
+void writeSimpleBlock(EbmlGlobal *glob, unsigned char trackNumber, short timeCode,
+ int isKeyframe, unsigned char lacingFlag, int discardable,
+ unsigned char *data, unsigned long dataLength)
+{
+ Ebml_WriteID(glob, SimpleBlock);
+ unsigned long blockLength = 4 + dataLength;
+ blockLength |= 0x10000000; //TODO check length < 0x0FFFFFFFF
+ Ebml_Serialize(glob, &blockLength, sizeof(blockLength), 4);
+ trackNumber |= 0x80; //TODO check track nubmer < 128
+ Ebml_Write(glob, &trackNumber, 1);
+ //Ebml_WriteSigned16(glob, timeCode,2); //this is 3 bytes
+ Ebml_Serialize(glob, &timeCode, sizeof(timeCode), 2);
+ unsigned char flags = 0x00 | (isKeyframe ? 0x80 : 0x00) | (lacingFlag << 1) | discardable;
+ Ebml_Write(glob, &flags, 1);
+ Ebml_Write(glob, data, dataLength);
+}
+
+static UInt64 generateTrackID(unsigned int trackNumber)
+{
+ UInt64 t = time(NULL) * trackNumber;
+ UInt64 r = rand();
+ r = r << 32;
+ r += rand();
+ UInt64 rval = t ^ r;
+ return rval;
+}
+
+void writeVideoTrack(EbmlGlobal *glob, unsigned int trackNumber, int flagLacing,
+ char *codecId, unsigned int pixelWidth, unsigned int pixelHeight,
+ double frameRate)
+{
+ EbmlLoc start;
+ Ebml_StartSubElement(glob, &start, TrackEntry);
+ Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber);
+ UInt64 trackID = generateTrackID(trackNumber);
+ Ebml_SerializeUnsigned(glob, TrackUID, trackID);
+ Ebml_SerializeString(glob, CodecName, "VP8"); //TODO shouldn't be fixed
+
+ Ebml_SerializeUnsigned(glob, TrackType, 1); //video is always 1
+ Ebml_SerializeString(glob, CodecID, codecId);
+ {
+ EbmlLoc videoStart;
+ Ebml_StartSubElement(glob, &videoStart, Video);
+ Ebml_SerializeUnsigned(glob, PixelWidth, pixelWidth);
+ Ebml_SerializeUnsigned(glob, PixelHeight, pixelHeight);
+ Ebml_SerializeFloat(glob, FrameRate, frameRate);
+ Ebml_EndSubElement(glob, &videoStart); //Video
+ }
+ Ebml_EndSubElement(glob, &start); //Track Entry
+}
+void writeAudioTrack(EbmlGlobal *glob, unsigned int trackNumber, int flagLacing,
+ char *codecId, double samplingFrequency, unsigned int channels,
+ unsigned char *private, unsigned long privateSize)
+{
+ EbmlLoc start;
+ Ebml_StartSubElement(glob, &start, TrackEntry);
+ Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber);
+ UInt64 trackID = generateTrackID(trackNumber);
+ Ebml_SerializeUnsigned(glob, TrackUID, trackID);
+ Ebml_SerializeUnsigned(glob, TrackType, 2); //audio is always 2
+ //I am using defaults for thesed required fields
+ /* Ebml_SerializeUnsigned(glob, FlagEnabled, 1);
+ Ebml_SerializeUnsigned(glob, FlagDefault, 1);
+ Ebml_SerializeUnsigned(glob, FlagForced, 1);
+ Ebml_SerializeUnsigned(glob, FlagLacing, flagLacing);*/
+ Ebml_SerializeString(glob, CodecID, codecId);
+ Ebml_SerializeData(glob, CodecPrivate, private, privateSize);
+
+ Ebml_SerializeString(glob, CodecName, "VORBIS"); //fixed for now
+ {
+ EbmlLoc AudioStart;
+ Ebml_StartSubElement(glob, &AudioStart, Audio);
+ Ebml_SerializeFloat(glob, SamplingFrequency, samplingFrequency);
+ Ebml_SerializeUnsigned(glob, Channels, channels);
+ Ebml_EndSubElement(glob, &AudioStart);
+ }
+ Ebml_EndSubElement(glob, &start);
+}
+void writeSegmentInformation(EbmlGlobal *ebml, EbmlLoc* startInfo, unsigned long timeCodeScale, double duration)
+{
+ Ebml_StartSubElement(ebml, startInfo, Info);
+ Ebml_SerializeUnsigned(ebml, TimecodeScale, timeCodeScale);
+ Ebml_SerializeFloat(ebml, Segment_Duration, duration * 1000.0); //Currently fixed to using milliseconds
+ Ebml_SerializeString(ebml, 0x4D80, "QTmuxingAppLibWebM-0.0.1");
+ Ebml_SerializeString(ebml, 0x5741, "QTwritingAppLibWebM-0.0.1");
+ Ebml_EndSubElement(ebml, startInfo);
+}
+
+/*
+void Mkv_InitializeSegment(Ebml& ebml_out, EbmlLoc& ebmlLoc)
+{
+ Ebml_StartSubElement(ebml_out, ebmlLoc, 0x18538067);
+}
+
+void Mkv_InitializeSeek(Ebml& ebml_out, EbmlLoc& ebmlLoc)
+{
+ Ebml_StartSubElement(ebml_out, ebmlLoc, 0x114d9b74);
+}
+void Mkv_WriteSeekInformation(Ebml& ebml_out, SeekStruct& seekInformation)
+{
+ EbmlLoc ebmlLoc;
+ Ebml_StartSubElement(ebml_out, ebmlLoc, 0x4dbb);
+ Ebml_SerializeString(ebml_out, 0x53ab, seekInformation.SeekID);
+ Ebml_SerializeUnsigned(ebml_out, 0x53ac, seekInformation.SeekPosition);
+ Ebml_EndSubElement(ebml_out, ebmlLoc);
+}
+
+void Mkv_WriteSegmentInformation(Ebml& ebml_out, SegmentInformationStruct& segmentInformation)
+{
+ Ebml_SerializeUnsigned(ebml_out, 0x73a4, segmentInformation.segmentUID);
+ if (segmentInformation.filename != 0)
+ Ebml_SerializeString(ebml_out, 0x7384, segmentInformation.filename);
+ Ebml_SerializeUnsigned(ebml_out, 0x2AD7B1, segmentInformation.TimecodeScale);
+ Ebml_SerializeUnsigned(ebml_out, 0x4489, segmentInformation.Duration);
+ //TODO date
+ Ebml_SerializeWString(ebml_out, 0x4D80, L"MKVMUX");
+ Ebml_SerializeWString(ebml_out, 0x5741, segmentInformation.WritingApp);
+}
+
+void Mkv_InitializeTrack(Ebml& ebml_out, EbmlLoc& ebmlLoc)
+{
+ Ebml_StartSubElement(ebml_out, ebmlLoc, 0x1654AE6B);
+}
+
+static void Mkv_WriteGenericTrackData(Ebml& ebml_out, TrackStruct& track)
+{
+ Ebml_SerializeUnsigned(ebml_out, 0xD7, track.TrackNumber);
+ Ebml_SerializeUnsigned(ebml_out, 0x73C5, track.TrackUID);
+ Ebml_SerializeUnsigned(ebml_out, 0x83, track.TrackType);
+ Ebml_SerializeUnsigned(ebml_out, 0xB9, track.FlagEnabled ? 1 :0);
+ Ebml_SerializeUnsigned(ebml_out, 0x88, track.FlagDefault ? 1 :0);
+ Ebml_SerializeUnsigned(ebml_out, 0x55AA, track.FlagForced ? 1 :0);
+ if (track.Language != 0)
+ Ebml_SerializeString(ebml_out, 0x22B59C, track.Language);
+ if (track.CodecID != 0)
+ Ebml_SerializeString(ebml_out, 0x86, track.CodecID);
+ if (track.CodecPrivate != 0)
+ Ebml_SerializeData(ebml_out, 0x63A2, track.CodecPrivate, track.CodecPrivateLength);
+ if (track.CodecName != 0)
+ Ebml_SerializeWString(ebml_out, 0x258688, track.CodecName);
+}
+
+void Mkv_WriteVideoTrack(Ebml& ebml_out, TrackStruct & track, VideoTrackStruct& video)
+{
+ EbmlLoc trackHeadLoc, videoHeadLoc;
+ Ebml_StartSubElement(ebml_out, trackHeadLoc, 0xAE); //start Track
+ Mkv_WriteGenericTrackData(ebml_out, track);
+ Ebml_StartSubElement(ebml_out, videoHeadLoc, 0xE0); //start Video
+ Ebml_SerializeUnsigned(ebml_out, 0x9A, video.FlagInterlaced ? 1 :0);
+ Ebml_SerializeUnsigned(ebml_out, 0xB0, video.PixelWidth);
+ Ebml_SerializeUnsigned(ebml_out, 0xBA, video.PixelHeight);
+ Ebml_SerializeUnsigned(ebml_out, 0x54B0, video.PixelDisplayWidth);
+ Ebml_SerializeUnsigned(ebml_out, 0x54BA, video.PixelDisplayHeight);
+ Ebml_SerializeUnsigned(ebml_out, 0x54B2, video.displayUnit);
+ Ebml_SerializeFloat(ebml_out, 0x2383E3, video.FrameRate);
+ Ebml_EndSubElement(ebml_out, videoHeadLoc);
+ Ebml_EndSubElement(ebml_out, trackHeadLoc);
+
+}
+
+void Mkv_WriteAudioTrack(Ebml& ebml_out, TrackStruct & track, AudioTrackStruct& video)
+{
+ EbmlLoc trackHeadLoc, audioHeadLoc;
+ Ebml_StartSubElement(ebml_out, trackHeadLoc, 0xAE);
+ Mkv_WriteGenericTrackData(ebml_out, track);
+ Ebml_StartSubElement(ebml_out, audioHeadLoc, 0xE0); //start Audio
+ Ebml_SerializeFloat(ebml_out, 0xB5, video.SamplingFrequency);
+ Ebml_SerializeUnsigned(ebml_out, 0x9F, video.Channels);
+ Ebml_SerializeUnsigned(ebml_out, 0x6264, video.BitDepth);
+ Ebml_EndSubElement(ebml_out, audioHeadLoc); // end audio
+ Ebml_EndSubElement(ebml_out, trackHeadLoc);
+}
+
+void Mkv_WriteEbmlClusterHead(Ebml& ebml_out, EbmlLoc& ebmlLoc, ClusterHeadStruct & clusterHead)
+{
+ Ebml_StartSubElement(ebml_out, ebmlLoc, 0x1F43B675);
+ Ebml_SerializeUnsigned(ebml_out, 0x6264, clusterHead.TimeCode);
+}
+
+void Mkv_WriteSimpleBlockHead(Ebml& ebml_out, EbmlLoc& ebmlLoc, SimpleBlockStruct& block)
+{
+ Ebml_StartSubElement(ebml_out, ebmlLoc, 0xA3);
+ Ebml_Write1UInt(ebml_out, block.TrackNumber);
+ Ebml_WriteSigned16(ebml_out,block.TimeCode);
+ unsigned char flags = 0x00 | (block.iskey ? 0x80:0x00) | (block.lacing << 1) | block.discardable;
+ Ebml_Write1UInt(ebml_out, flags); //TODO this may be the wrong function
+ Ebml_Serialize(ebml_out, block.data, block.dataLength);
+ Ebml_EndSubElement(ebml_out,ebmlLoc);
+}
+*/
diff --git a/libmkv/WebMElement.h b/libvpx/libmkv/WebMElement.h
similarity index 100%
rename from libmkv/WebMElement.h
rename to libvpx/libmkv/WebMElement.h
diff --git a/libmkv/testlibmkv.c b/libvpx/libmkv/testlibmkv.c
similarity index 100%
rename from libmkv/testlibmkv.c
rename to libvpx/libmkv/testlibmkv.c
diff --git a/libs.doxy_template b/libvpx/libs.doxy_template
similarity index 100%
rename from libs.doxy_template
rename to libvpx/libs.doxy_template
diff --git a/libvpx/libs.mk b/libvpx/libs.mk
new file mode 100644
index 0000000..4115dd8
--- /dev/null
+++ b/libvpx/libs.mk
@@ -0,0 +1,454 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+# ARM assembly files are written in RVCT-style. We use some make magic to
+# filter those files to allow GCC compilation
+ifeq ($(ARCH_ARM),yes)
+ ASM:=$(if $(filter yes,$(CONFIG_GCC)),.asm.s,.asm)
+else
+ ASM:=.asm
+endif
+
+CODEC_SRCS-yes += CHANGELOG
+CODEC_SRCS-yes += libs.mk
+
+# If this is a universal (fat) binary, then all the subarchitectures have
+# already been built and our job is to stitch them together. The
+# BUILD_LIBVPX variable indicates whether we should be building
+# (compiling, linking) the library. The LIPO_LIBVPX variable indicates
+# that we're stitching.
+$(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_LIBVPX,BUILD_LIBVPX):=yes)
+
+include $(SRC_PATH_BARE)/vpx/vpx_codec.mk
+CODEC_SRCS-yes += $(addprefix vpx/,$(call enabled,API_SRCS))
+CODEC_DOC_SRCS += $(addprefix vpx/,$(call enabled,API_DOC_SRCS))
+
+include $(SRC_PATH_BARE)/vpx_mem/vpx_mem.mk
+CODEC_SRCS-yes += $(addprefix vpx_mem/,$(call enabled,MEM_SRCS))
+
+include $(SRC_PATH_BARE)/vpx_scale/vpx_scale.mk
+CODEC_SRCS-yes += $(addprefix vpx_scale/,$(call enabled,SCALE_SRCS))
+
+include $(SRC_PATH_BARE)/vpx_ports/vpx_ports.mk
+CODEC_SRCS-yes += $(addprefix vpx_ports/,$(call enabled,PORTS_SRCS))
+
+
+ifeq ($(CONFIG_VP8_ENCODER),yes)
+ VP8_PREFIX=vp8/
+ include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8cx.mk
+ CODEC_SRCS-yes += $(addprefix $(VP8_PREFIX),$(call enabled,VP8_CX_SRCS))
+ CODEC_EXPORTS-yes += $(addprefix $(VP8_PREFIX),$(VP8_CX_EXPORTS))
+ INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8cx.h
+ INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/%
+ CODEC_DOC_SECTIONS += vp8 vp8_encoder
+endif
+
+ifeq ($(CONFIG_VP8_DECODER),yes)
+ VP8_PREFIX=vp8/
+ include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8dx.mk
+ CODEC_SRCS-yes += $(addprefix $(VP8_PREFIX),$(call enabled,VP8_DX_SRCS))
+ CODEC_EXPORTS-yes += $(addprefix $(VP8_PREFIX),$(VP8_DX_EXPORTS))
+ INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8dx.h
+ INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/%
+ CODEC_DOC_SECTIONS += vp8 vp8_decoder
+endif
+
+
+ifeq ($(CONFIG_ENCODERS),yes)
+ CODEC_DOC_SECTIONS += encoder
+endif
+ifeq ($(CONFIG_DECODERS),yes)
+ CODEC_DOC_SECTIONS += decoder
+endif
+
+
+ifeq ($(CONFIG_MSVS),yes)
+CODEC_LIB=$(if $(CONFIG_STATIC_MSVCRT),vpxmt,vpxmd)
+GTEST_LIB=$(if $(CONFIG_STATIC_MSVCRT),gtestmt,gtestmd)
+# This variable uses deferred expansion intentionally, since the results of
+# $(wildcard) may change during the course of the Make.
+VS_PLATFORMS = $(foreach d,$(wildcard */Release/$(CODEC_LIB).lib),$(word 1,$(subst /, ,$(d))))
+endif
+
+# The following pairs define a mapping of locations in the distribution
+# tree to locations in the source/build trees.
+INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/vpx/%
+INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/vpx_ports/%
+INSTALL_MAPS += $(LIBSUBDIR)/% %
+INSTALL_MAPS += src/% $(SRC_PATH_BARE)/%
+ifeq ($(CONFIG_MSVS),yes)
+INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/% $(p)/Release/%)
+INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/% $(p)/Debug/%)
+endif
+
+CODEC_SRCS-$(BUILD_LIBVPX) += build/make/version.sh
+CODEC_SRCS-$(BUILD_LIBVPX) += build/make/rtcd.sh
+CODEC_SRCS-$(BUILD_LIBVPX) += $(BUILD_PFX)vpx_config.c
+INSTALL-SRCS-no += $(BUILD_PFX)vpx_config.c
+CODEC_EXPORTS-$(BUILD_LIBVPX) += vpx/exports_com
+CODEC_EXPORTS-$(CONFIG_ENCODERS) += vpx/exports_enc
+CODEC_EXPORTS-$(CONFIG_DECODERS) += vpx/exports_dec
+
+INSTALL-LIBS-yes += include/vpx/vpx_codec.h
+INSTALL-LIBS-yes += include/vpx/vpx_image.h
+INSTALL-LIBS-yes += include/vpx/vpx_integer.h
+INSTALL-LIBS-yes += include/vpx/vpx_codec_impl_top.h
+INSTALL-LIBS-yes += include/vpx/vpx_codec_impl_bottom.h
+INSTALL-LIBS-$(CONFIG_DECODERS) += include/vpx/vpx_decoder.h
+INSTALL-LIBS-$(CONFIG_ENCODERS) += include/vpx/vpx_encoder.h
+ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
+ifeq ($(CONFIG_MSVS),yes)
+INSTALL-LIBS-yes += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/$(CODEC_LIB).lib)
+INSTALL-LIBS-$(CONFIG_DEBUG_LIBS) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/$(CODEC_LIB)d.lib)
+INSTALL-LIBS-$(CONFIG_SHARED) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/vpx.dll)
+INSTALL-LIBS-$(CONFIG_SHARED) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/vpx.exp)
+endif
+else
+INSTALL-LIBS-$(CONFIG_STATIC) += $(LIBSUBDIR)/libvpx.a
+INSTALL-LIBS-$(CONFIG_DEBUG_LIBS) += $(LIBSUBDIR)/libvpx_g.a
+endif
+
+CODEC_SRCS=$(filter-out %_test.cc,$(call enabled,CODEC_SRCS))
+INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(CODEC_SRCS)
+INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(call enabled,CODEC_EXPORTS)
+
+
+# Generate a list of all enabled sources, in particular for exporting to gyp
+# based build systems.
+libvpx_srcs.txt:
+ @echo " [CREATE] $@"
+ @echo $(CODEC_SRCS) | xargs -n1 echo | sort -u > $@
+
+
+ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
+ifeq ($(CONFIG_MSVS),yes)
+
+obj_int_extract.vcproj: $(SRC_PATH_BARE)/build/make/obj_int_extract.c
+ @cp $(SRC_PATH_BARE)/build/x86-msvs/obj_int_extract.bat .
+ @echo " [CREATE] $@"
+ $(qexec)$(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh \
+ --exe \
+ --target=$(TOOLCHAIN) \
+ --name=obj_int_extract \
+ --ver=$(CONFIG_VS_VERSION) \
+ --proj-guid=E1360C65-D375-4335-8057-7ED99CC3F9B2 \
+ $(if $(CONFIG_STATIC_MSVCRT),--static-crt) \
+ --out=$@ $^ \
+ -I. \
+ -I"$(SRC_PATH_BARE)" \
+
+PROJECTS-$(BUILD_LIBVPX) += obj_int_extract.vcproj
+PROJECTS-$(BUILD_LIBVPX) += obj_int_extract.bat
+
+vpx.def: $(call enabled,CODEC_EXPORTS)
+ @echo " [CREATE] $@"
+ $(qexec)$(SRC_PATH_BARE)/build/make/gen_msvs_def.sh\
+ --name=vpx\
+ --out=$@ $^
+CLEAN-OBJS += vpx.def
+
+vpx.vcproj: $(CODEC_SRCS) vpx.def
+ @echo " [CREATE] $@"
+ $(qexec)$(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh \
+ --lib \
+ --target=$(TOOLCHAIN) \
+ $(if $(CONFIG_STATIC_MSVCRT),--static-crt) \
+ --name=vpx \
+ --proj-guid=DCE19DAF-69AC-46DB-B14A-39F0FAA5DB74 \
+ --module-def=vpx.def \
+ --ver=$(CONFIG_VS_VERSION) \
+ --out=$@ $(CFLAGS) $^ \
+ --src-path-bare="$(SRC_PATH_BARE)" \
+
+PROJECTS-$(BUILD_LIBVPX) += vpx.vcproj
+
+vpx.vcproj: vpx_config.asm
+vpx.vcproj: vpx_rtcd.h
+
+endif
+else
+LIBVPX_OBJS=$(call objs,$(CODEC_SRCS))
+OBJS-$(BUILD_LIBVPX) += $(LIBVPX_OBJS)
+LIBS-$(if $(BUILD_LIBVPX),$(CONFIG_STATIC)) += $(BUILD_PFX)libvpx.a $(BUILD_PFX)libvpx_g.a
+$(BUILD_PFX)libvpx_g.a: $(LIBVPX_OBJS)
+
+BUILD_LIBVPX_SO := $(if $(BUILD_LIBVPX),$(CONFIG_SHARED))
+LIBVPX_SO := libvpx.so.$(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH)
+LIBS-$(BUILD_LIBVPX_SO) += $(BUILD_PFX)$(LIBVPX_SO)\
+ $(notdir $(LIBVPX_SO_SYMLINKS))
+$(BUILD_PFX)$(LIBVPX_SO): $(LIBVPX_OBJS) libvpx.ver
+$(BUILD_PFX)$(LIBVPX_SO): extralibs += -lm
+$(BUILD_PFX)$(LIBVPX_SO): SONAME = libvpx.so.$(VERSION_MAJOR)
+$(BUILD_PFX)$(LIBVPX_SO): SO_VERSION_SCRIPT = libvpx.ver
+LIBVPX_SO_SYMLINKS := $(addprefix $(LIBSUBDIR)/, \
+ libvpx.so libvpx.so.$(VERSION_MAJOR) \
+ libvpx.so.$(VERSION_MAJOR).$(VERSION_MINOR))
+
+libvpx.ver: $(call enabled,CODEC_EXPORTS)
+ @echo " [CREATE] $@"
+ $(qexec)echo "{ global:" > $@
+ $(qexec)for f in $?; do awk '{print $$2";"}' < $$f >>$@; done
+ $(qexec)echo "local: *; };" >> $@
+CLEAN-OBJS += libvpx.ver
+
+define libvpx_symlink_template
+$(1): $(2)
+ @echo " [LN] $$@"
+ $(qexec)ln -sf $(LIBVPX_SO) $$@
+endef
+
+$(eval $(call libvpx_symlink_template,\
+ $(addprefix $(BUILD_PFX),$(notdir $(LIBVPX_SO_SYMLINKS))),\
+ $(BUILD_PFX)$(LIBVPX_SO)))
+$(eval $(call libvpx_symlink_template,\
+ $(addprefix $(DIST_DIR)/,$(LIBVPX_SO_SYMLINKS)),\
+ $(DIST_DIR)/$(LIBSUBDIR)/$(LIBVPX_SO)))
+
+INSTALL-LIBS-$(CONFIG_SHARED) += $(LIBVPX_SO_SYMLINKS)
+INSTALL-LIBS-$(CONFIG_SHARED) += $(LIBSUBDIR)/$(LIBVPX_SO)
+
+LIBS-$(BUILD_LIBVPX) += vpx.pc
+vpx.pc: config.mk libs.mk
+ @echo " [CREATE] $@"
+ $(qexec)echo '# pkg-config file from libvpx $(VERSION_STRING)' > $@
+ $(qexec)echo 'prefix=$(PREFIX)' >> $@
+ $(qexec)echo 'exec_prefix=$${prefix}' >> $@
+ $(qexec)echo 'libdir=$${prefix}/$(LIBSUBDIR)' >> $@
+ $(qexec)echo 'includedir=$${prefix}/include' >> $@
+ $(qexec)echo '' >> $@
+ $(qexec)echo 'Name: vpx' >> $@
+ $(qexec)echo 'Description: WebM Project VPx codec implementation' >> $@
+ $(qexec)echo 'Version: $(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH)' >> $@
+ $(qexec)echo 'Requires:' >> $@
+ $(qexec)echo 'Conflicts:' >> $@
+ $(qexec)echo 'Libs: -L$${libdir} -lvpx' >> $@
+ $(qexec)echo 'Libs.private: -lm -lpthread' >> $@
+ $(qexec)echo 'Cflags: -I$${includedir}' >> $@
+INSTALL-LIBS-yes += $(LIBSUBDIR)/pkgconfig/vpx.pc
+INSTALL_MAPS += $(LIBSUBDIR)/pkgconfig/%.pc %.pc
+CLEAN-OBJS += vpx.pc
+endif
+
+LIBS-$(LIPO_LIBVPX) += libvpx.a
+$(eval $(if $(LIPO_LIBVPX),$(call lipo_lib_template,libvpx.a)))
+
+#
+# Rule to make assembler configuration file from C configuration file
+#
+ifeq ($(ARCH_X86)$(ARCH_X86_64),yes)
+# YASM
+$(BUILD_PFX)vpx_config.asm: $(BUILD_PFX)vpx_config.h
+ @echo " [CREATE] $@"
+ @egrep "#define [A-Z0-9_]+ [01]" $< \
+ | awk '{print $$2 " equ " $$3}' > $@
+else
+ADS2GAS=$(if $(filter yes,$(CONFIG_GCC)),| $(ASM_CONVERSION))
+$(BUILD_PFX)vpx_config.asm: $(BUILD_PFX)vpx_config.h
+ @echo " [CREATE] $@"
+ @egrep "#define [A-Z0-9_]+ [01]" $< \
+ | awk '{print $$2 " EQU " $$3}' $(ADS2GAS) > $@
+ @echo " END" $(ADS2GAS) >> $@
+CLEAN-OBJS += $(BUILD_PFX)vpx_config.asm
+endif
+
+#
+# Add assembler dependencies for configuration and offsets
+#
+$(filter %.s.o,$(OBJS-yes)): $(BUILD_PFX)vpx_config.asm
+$(filter %$(ASM).o,$(OBJS-yes)): $(BUILD_PFX)vpx_config.asm
+
+#
+# Calculate platform- and compiler-specific offsets for hand coded assembly
+#
+
+OFFSET_PATTERN:='^[a-zA-Z0-9_]* EQU'
+
+ifeq ($(filter icc gcc,$(TGT_CC)), $(TGT_CC))
+ $(BUILD_PFX)asm_com_offsets.asm: $(BUILD_PFX)$(VP8_PREFIX)common/asm_com_offsets.c.S
+ @echo " [CREATE] $@"
+ $(qexec)LC_ALL=C grep $(OFFSET_PATTERN) $< | tr -d '$$\#' $(ADS2GAS) > $@
+ $(BUILD_PFX)$(VP8_PREFIX)common/asm_com_offsets.c.S: $(VP8_PREFIX)common/asm_com_offsets.c
+ CLEAN-OBJS += $(BUILD_PFX)asm_com_offsets.asm $(BUILD_PFX)$(VP8_PREFIX)common/asm_com_offsets.c.S
+
+ $(BUILD_PFX)asm_enc_offsets.asm: $(BUILD_PFX)$(VP8_PREFIX)encoder/asm_enc_offsets.c.S
+ @echo " [CREATE] $@"
+ $(qexec)LC_ALL=C grep $(OFFSET_PATTERN) $< | tr -d '$$\#' $(ADS2GAS) > $@
+ $(BUILD_PFX)$(VP8_PREFIX)encoder/asm_enc_offsets.c.S: $(VP8_PREFIX)encoder/asm_enc_offsets.c
+ CLEAN-OBJS += $(BUILD_PFX)asm_enc_offsets.asm $(BUILD_PFX)$(VP8_PREFIX)encoder/asm_enc_offsets.c.S
+
+ $(BUILD_PFX)asm_dec_offsets.asm: $(BUILD_PFX)$(VP8_PREFIX)decoder/asm_dec_offsets.c.S
+ @echo " [CREATE] $@"
+ $(qexec)LC_ALL=C grep $(OFFSET_PATTERN) $< | tr -d '$$\#' $(ADS2GAS) > $@
+ $(BUILD_PFX)$(VP8_PREFIX)decoder/asm_dec_offsets.c.S: $(VP8_PREFIX)decoder/asm_dec_offsets.c
+ CLEAN-OBJS += $(BUILD_PFX)asm_dec_offsets.asm $(BUILD_PFX)$(VP8_PREFIX)decoder/asm_dec_offsets.c.S
+else
+ ifeq ($(filter rvct,$(TGT_CC)), $(TGT_CC))
+ asm_com_offsets.asm: obj_int_extract
+ asm_com_offsets.asm: $(VP8_PREFIX)common/asm_com_offsets.c.o
+ @echo " [CREATE] $@"
+ $(qexec)./obj_int_extract rvds $< $(ADS2GAS) > $@
+ OBJS-yes += $(VP8_PREFIX)common/asm_com_offsets.c.o
+ CLEAN-OBJS += asm_com_offsets.asm
+ $(filter %$(ASM).o,$(OBJS-yes)): $(BUILD_PFX)asm_com_offsets.asm
+
+ asm_enc_offsets.asm: obj_int_extract
+ asm_enc_offsets.asm: $(VP8_PREFIX)encoder/asm_enc_offsets.c.o
+ @echo " [CREATE] $@"
+ $(qexec)./obj_int_extract rvds $< $(ADS2GAS) > $@
+ OBJS-yes += $(VP8_PREFIX)encoder/asm_enc_offsets.c.o
+ CLEAN-OBJS += asm_enc_offsets.asm
+ $(filter %$(ASM).o,$(OBJS-yes)): $(BUILD_PFX)asm_enc_offsets.asm
+
+ asm_dec_offsets.asm: obj_int_extract
+ asm_dec_offsets.asm: $(VP8_PREFIX)decoder/asm_dec_offsets.c.o
+ @echo " [CREATE] $@"
+ $(qexec)./obj_int_extract rvds $< $(ADS2GAS) > $@
+ OBJS-yes += $(VP8_PREFIX)decoder/asm_dec_offsets.c.o
+ CLEAN-OBJS += asm_dec_offsets.asm
+ $(filter %$(ASM).o,$(OBJS-yes)): $(BUILD_PFX)asm_dec_offsets.asm
+ endif
+endif
+
+$(shell $(SRC_PATH_BARE)/build/make/version.sh "$(SRC_PATH_BARE)" $(BUILD_PFX)vpx_version.h)
+CLEAN-OBJS += $(BUILD_PFX)vpx_version.h
+
+#
+# Rule to generate runtime cpu detection files
+#
+$(BUILD_PFX)vpx_rtcd.h: $(SRC_PATH_BARE)/$(sort $(filter %rtcd_defs.sh,$(CODEC_SRCS)))
+ @echo " [CREATE] $@"
+ $(qexec)$(SRC_PATH_BARE)/build/make/rtcd.sh --arch=$(TGT_ISA) \
+ --sym=vpx_rtcd \
+ --config=$(target)$(if $(FAT_ARCHS),,-$(TOOLCHAIN)).mk \
+ $(RTCD_OPTIONS) $^ > $@
+CLEAN-OBJS += $(BUILD_PFX)vpx_rtcd.h
+
+##
+## libvpx test directives
+##
+ifeq ($(CONFIG_UNIT_TESTS),yes)
+LIBVPX_TEST_DATA_PATH ?= .
+
+include $(SRC_PATH_BARE)/test/test.mk
+LIBVPX_TEST_SRCS=$(addprefix test/,$(call enabled,LIBVPX_TEST_SRCS))
+LIBVPX_TEST_BINS=./test_libvpx
+LIBVPX_TEST_DATA=$(addprefix $(LIBVPX_TEST_DATA_PATH)/,\
+ $(call enabled,LIBVPX_TEST_DATA))
+libvpx_test_data_url=http://downloads.webmproject.org/test_data/libvpx/$(1)
+
+$(LIBVPX_TEST_DATA):
+ @echo " [DOWNLOAD] $@"
+ $(qexec)trap 'rm -f $@' INT TERM &&\
+ curl -L -o $@ $(call libvpx_test_data_url,$(@F))
+
+testdata:: $(LIBVPX_TEST_DATA)
+ $(qexec)if [ -x "$$(which sha1sum)" ]; then\
+ echo "Checking test data:";\
+ if [ -n "$(LIBVPX_TEST_DATA)" ]; then\
+ for f in $(call enabled,LIBVPX_TEST_DATA); do\
+ grep $$f $(SRC_PATH_BARE)/test/test-data.sha1 |\
+ (cd $(LIBVPX_TEST_DATA_PATH); sha1sum -c);\
+ done; \
+ fi; \
+ else\
+ echo "Skipping test data integrity check, sha1sum not found.";\
+ fi
+
+ifeq ($(CONFIG_EXTERNAL_BUILD),yes)
+ifeq ($(CONFIG_MSVS),yes)
+
+gtest.vcproj: $(SRC_PATH_BARE)/third_party/googletest/src/src/gtest-all.cc
+ @echo " [CREATE] $@"
+ $(qexec)$(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh \
+ --lib \
+ --target=$(TOOLCHAIN) \
+ $(if $(CONFIG_STATIC_MSVCRT),--static-crt) \
+ --name=gtest \
+ --proj-guid=EC00E1EC-AF68-4D92-A255-181690D1C9B1 \
+ --ver=$(CONFIG_VS_VERSION) \
+ --src-path-bare="$(SRC_PATH_BARE)" \
+ --out=gtest.vcproj $(SRC_PATH_BARE)/third_party/googletest/src/src/gtest-all.cc \
+ -I. -I"$(SRC_PATH_BARE)/third_party/googletest/src/include" -I"$(SRC_PATH_BARE)/third_party/googletest/src"
+
+PROJECTS-$(CONFIG_MSVS) += gtest.vcproj
+
+test_libvpx.vcproj: $(LIBVPX_TEST_SRCS)
+ @echo " [CREATE] $@"
+ $(qexec)$(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh \
+ --exe \
+ --target=$(TOOLCHAIN) \
+ --name=test_libvpx \
+ --proj-guid=CD837F5F-52D8-4314-A370-895D614166A7 \
+ --ver=$(CONFIG_VS_VERSION) \
+ $(if $(CONFIG_STATIC_MSVCRT),--static-crt) \
+ --out=$@ $(INTERNAL_CFLAGS) $(CFLAGS) \
+ -I. -I"$(SRC_PATH_BARE)/third_party/googletest/src/include" \
+ -L. -l$(CODEC_LIB) -lwinmm -l$(GTEST_LIB) $^
+
+PROJECTS-$(CONFIG_MSVS) += test_libvpx.vcproj
+
+test:: testdata
+ @set -e; for t in $(addprefix Win32/Release/,$(notdir $(LIBVPX_TEST_BINS:.cc=.exe))); do $$t; done
+endif
+else
+
+include $(SRC_PATH_BARE)/third_party/googletest/gtest.mk
+GTEST_SRCS := $(addprefix third_party/googletest/src/,$(call enabled,GTEST_SRCS))
+GTEST_OBJS=$(call objs,$(GTEST_SRCS))
+$(GTEST_OBJS) $(GTEST_OBJS:.o=.d): CXXFLAGS += -I$(SRC_PATH_BARE)/third_party/googletest/src
+$(GTEST_OBJS) $(GTEST_OBJS:.o=.d): CXXFLAGS += -I$(SRC_PATH_BARE)/third_party/googletest/src/include
+OBJS-$(BUILD_LIBVPX) += $(GTEST_OBJS)
+LIBS-$(BUILD_LIBVPX) += $(BUILD_PFX)libgtest.a $(BUILD_PFX)libgtest_g.a
+$(BUILD_PFX)libgtest_g.a: $(GTEST_OBJS)
+
+LIBVPX_TEST_OBJS=$(sort $(call objs,$(LIBVPX_TEST_SRCS)))
+$(LIBVPX_TEST_OBJS) $(LIBVPX_TEST_OBJS:.o=.d): CXXFLAGS += -I$(SRC_PATH_BARE)/third_party/googletest/src
+$(LIBVPX_TEST_OBJS) $(LIBVPX_TEST_OBJS:.o=.d): CXXFLAGS += -I$(SRC_PATH_BARE)/third_party/googletest/src/include
+OBJS-$(BUILD_LIBVPX) += $(LIBVPX_TEST_OBJS)
+BINS-$(BUILD_LIBVPX) += $(LIBVPX_TEST_BINS)
+
+# Install test sources only if codec source is included
+INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(patsubst $(SRC_PATH_BARE)/%,%,\
+ $(shell find $(SRC_PATH_BARE)/third_party/googletest -type f))
+INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(LIBVPX_TEST_SRCS)
+
+CODEC_LIB=$(if $(CONFIG_DEBUG_LIBS),vpx_g,vpx)
+CODEC_LIB_SUF=$(if $(CONFIG_SHARED),.so,.a)
+$(foreach bin,$(LIBVPX_TEST_BINS),\
+ $(if $(BUILD_LIBVPX),$(eval $(bin): \
+ lib$(CODEC_LIB)$(CODEC_LIB_SUF) libgtest.a ))\
+ $(if $(BUILD_LIBVPX),$(eval $(call linkerxx_template,$(bin),\
+ $(LIBVPX_TEST_OBJS) \
+ -L. -lvpx -lgtest -lpthread -lm)\
+ )))\
+ $(if $(LIPO_LIBS),$(eval $(call lipo_bin_template,$(bin))))\
+
+test:: $(LIBVPX_TEST_BINS) testdata
+ @set -e; for t in $(LIBVPX_TEST_BINS); do $$t; done
+
+endif
+endif
+
+##
+## documentation directives
+##
+CLEAN-OBJS += libs.doxy
+DOCS-yes += libs.doxy
+libs.doxy: $(CODEC_DOC_SRCS)
+ @echo " [CREATE] $@"
+ @rm -f $@
+ @echo "INPUT += $^" >> $@
+ @echo "PREDEFINED = VPX_CODEC_DISABLE_COMPAT" >> $@
+ @echo "INCLUDE_PATH += ." >> $@;
+ @echo "ENABLED_SECTIONS += $(sort $(CODEC_DOC_SECTIONS))" >> $@
+
+## Generate vpx_rtcd.h for all objects
+$(OBJS-yes:.o=.d): $(BUILD_PFX)vpx_rtcd.h
diff --git a/libvpx/mainpage.dox b/libvpx/mainpage.dox
new file mode 100644
index 0000000..e2ec280
--- /dev/null
+++ b/libvpx/mainpage.dox
@@ -0,0 +1,53 @@
+/*!\mainpage WebM VP8 Codec SDK
+
+ \section main_contents Page Contents
+ - \ref main_intro
+ - \ref main_startpoints
+ - \ref main_support
+
+ \section main_intro Introduction
+ Welcome to the WebM VP8 Codec SDK. This SDK allows you to integrate your
+ applications with the VP8 video codec, a high quality, royalty free, open
+ source codec deployed on millions of computers and devices worldwide.
+
+ This distribution of the WebM VP8 Codec SDK includes the following support:
+
+ \if vp8_encoder
+ - \ref vp8_encoder
+ \endif
+ \if vp8_decoder
+ - \ref vp8_decoder
+ \endif
+
+
+ \section main_startpoints Starting Points
+ - Consult the \ref changelog for a complete list of improvements in this
+ release.
+ - The \ref readme contains instructions on recompiling the sample applications.
+ - Read the \ref usage "usage" for a narrative on codec usage.
+ - Read the \ref samples "sample code" for examples of how to interact with the
+ codec.
+ - \ref codec reference
+ \if encoder
+ - \ref encoder reference
+ \endif
+ \if decoder
+ - \ref decoder reference
+ \endif
+
+ \section main_support Support Options & FAQ
+ The WebM project is an open source project supported by its community. For
+ questions about this SDK, please mail the apps-devel@webmproject.org list.
+ To contribute, see http://www.webmproject.org/code/contribute and mail
+ codec-devel@webmproject.org.
+*/
+
+/*!\page changelog CHANGELOG
+ \verbinclude CHANGELOG
+*/
+
+/*!\page readme README
+ \verbinclude README
+*/
+
+/*!\defgroup codecs Supported Codecs */
diff --git a/md5_utils.c b/libvpx/md5_utils.c
similarity index 100%
rename from md5_utils.c
rename to libvpx/md5_utils.c
diff --git a/md5_utils.h b/libvpx/md5_utils.h
similarity index 100%
rename from md5_utils.h
rename to libvpx/md5_utils.h
diff --git a/libvpx/nestegg/.gitignore b/libvpx/nestegg/.gitignore
new file mode 100644
index 0000000..b2ba99c
--- /dev/null
+++ b/libvpx/nestegg/.gitignore
@@ -0,0 +1,40 @@
+*.lo
+*.o
+*.swp
+*~
+.deps
+.dirstamp
+.libs
+Makefile
+Makefile.in
+_stdint.h
+aclocal.m4
+autom4te.cache
+compile
+config.guess
+config.h
+config.h.in
+config.log
+config.status
+config.sub
+configure
+depcomp
+docs/Doxyfile
+docs/doxygen-build.stamp
+docs/html
+install-sh
+libtool
+ltmain.sh
+m4/libtool.m4
+m4/ltoptions.m4
+m4/ltsugar.m4
+m4/ltversion.m4
+m4/lt~obsolete.m4
+missing
+nestegg-uninstalled.pc
+nestegg.pc
+src/.dirstamp
+src/libnestegg.la
+stamp-h1
+test/test
+include/nestegg/nestegg-stdint.h
diff --git a/nestegg/AUTHORS b/libvpx/nestegg/AUTHORS
similarity index 100%
rename from nestegg/AUTHORS
rename to libvpx/nestegg/AUTHORS
diff --git a/nestegg/INSTALL b/libvpx/nestegg/INSTALL
similarity index 100%
rename from nestegg/INSTALL
rename to libvpx/nestegg/INSTALL
diff --git a/nestegg/LICENSE b/libvpx/nestegg/LICENSE
similarity index 100%
rename from nestegg/LICENSE
rename to libvpx/nestegg/LICENSE
diff --git a/nestegg/Makefile.am b/libvpx/nestegg/Makefile.am
similarity index 100%
rename from nestegg/Makefile.am
rename to libvpx/nestegg/Makefile.am
diff --git a/nestegg/README b/libvpx/nestegg/README
similarity index 100%
rename from nestegg/README
rename to libvpx/nestegg/README
diff --git a/nestegg/TODO b/libvpx/nestegg/TODO
similarity index 100%
rename from nestegg/TODO
rename to libvpx/nestegg/TODO
diff --git a/nestegg/configure.ac b/libvpx/nestegg/configure.ac
similarity index 100%
rename from nestegg/configure.ac
rename to libvpx/nestegg/configure.ac
diff --git a/nestegg/docs/Doxyfile.in b/libvpx/nestegg/docs/Doxyfile.in
similarity index 100%
rename from nestegg/docs/Doxyfile.in
rename to libvpx/nestegg/docs/Doxyfile.in
diff --git a/nestegg/docs/Makefile.am b/libvpx/nestegg/docs/Makefile.am
similarity index 100%
rename from nestegg/docs/Makefile.am
rename to libvpx/nestegg/docs/Makefile.am
diff --git a/nestegg/halloc/README b/libvpx/nestegg/halloc/README
similarity index 100%
rename from nestegg/halloc/README
rename to libvpx/nestegg/halloc/README
diff --git a/nestegg/halloc/halloc.h b/libvpx/nestegg/halloc/halloc.h
similarity index 100%
rename from nestegg/halloc/halloc.h
rename to libvpx/nestegg/halloc/halloc.h
diff --git a/nestegg/halloc/src/align.h b/libvpx/nestegg/halloc/src/align.h
similarity index 100%
rename from nestegg/halloc/src/align.h
rename to libvpx/nestegg/halloc/src/align.h
diff --git a/nestegg/halloc/src/halloc.c b/libvpx/nestegg/halloc/src/halloc.c
similarity index 100%
rename from nestegg/halloc/src/halloc.c
rename to libvpx/nestegg/halloc/src/halloc.c
diff --git a/nestegg/halloc/src/hlist.h b/libvpx/nestegg/halloc/src/hlist.h
similarity index 100%
rename from nestegg/halloc/src/hlist.h
rename to libvpx/nestegg/halloc/src/hlist.h
diff --git a/nestegg/halloc/src/macros.h b/libvpx/nestegg/halloc/src/macros.h
similarity index 100%
rename from nestegg/halloc/src/macros.h
rename to libvpx/nestegg/halloc/src/macros.h
diff --git a/nestegg/include/nestegg/nestegg.h b/libvpx/nestegg/include/nestegg/nestegg.h
similarity index 100%
rename from nestegg/include/nestegg/nestegg.h
rename to libvpx/nestegg/include/nestegg/nestegg.h
diff --git a/nestegg/m4/as-ac-expand.m4 b/libvpx/nestegg/m4/as-ac-expand.m4
similarity index 100%
rename from nestegg/m4/as-ac-expand.m4
rename to libvpx/nestegg/m4/as-ac-expand.m4
diff --git a/nestegg/m4/ax_create_stdint_h.m4 b/libvpx/nestegg/m4/ax_create_stdint_h.m4
similarity index 100%
rename from nestegg/m4/ax_create_stdint_h.m4
rename to libvpx/nestegg/m4/ax_create_stdint_h.m4
diff --git a/nestegg/m4/pkg.m4 b/libvpx/nestegg/m4/pkg.m4
similarity index 100%
rename from nestegg/m4/pkg.m4
rename to libvpx/nestegg/m4/pkg.m4
diff --git a/nestegg/nestegg-uninstalled.pc.in b/libvpx/nestegg/nestegg-uninstalled.pc.in
similarity index 100%
rename from nestegg/nestegg-uninstalled.pc.in
rename to libvpx/nestegg/nestegg-uninstalled.pc.in
diff --git a/nestegg/nestegg.pc.in b/libvpx/nestegg/nestegg.pc.in
similarity index 100%
rename from nestegg/nestegg.pc.in
rename to libvpx/nestegg/nestegg.pc.in
diff --git a/libvpx/nestegg/src/nestegg.c b/libvpx/nestegg/src/nestegg.c
new file mode 100644
index 0000000..cc87788
--- /dev/null
+++ b/libvpx/nestegg/src/nestegg.c
@@ -0,0 +1,1938 @@
+/*
+ * Copyright © 2010 Mozilla Foundation
+ *
+ * This program is made available under an ISC-style license. See the
+ * accompanying file LICENSE for details.
+ */
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "nestegg/halloc/halloc.h"
+#include "nestegg/include/nestegg/nestegg.h"
+
+/* EBML Elements */
+#define ID_EBML 0x1a45dfa3
+#define ID_EBML_VERSION 0x4286
+#define ID_EBML_READ_VERSION 0x42f7
+#define ID_EBML_MAX_ID_LENGTH 0x42f2
+#define ID_EBML_MAX_SIZE_LENGTH 0x42f3
+#define ID_DOCTYPE 0x4282
+#define ID_DOCTYPE_VERSION 0x4287
+#define ID_DOCTYPE_READ_VERSION 0x4285
+
+/* Global Elements */
+#define ID_VOID 0xec
+#define ID_CRC32 0xbf
+
+/* WebMedia Elements */
+#define ID_SEGMENT 0x18538067
+
+/* Seek Head Elements */
+#define ID_SEEK_HEAD 0x114d9b74
+#define ID_SEEK 0x4dbb
+#define ID_SEEK_ID 0x53ab
+#define ID_SEEK_POSITION 0x53ac
+
+/* Info Elements */
+#define ID_INFO 0x1549a966
+#define ID_TIMECODE_SCALE 0x2ad7b1
+#define ID_DURATION 0x4489
+
+/* Cluster Elements */
+#define ID_CLUSTER 0x1f43b675
+#define ID_TIMECODE 0xe7
+#define ID_BLOCK_GROUP 0xa0
+#define ID_SIMPLE_BLOCK 0xa3
+
+/* BlockGroup Elements */
+#define ID_BLOCK 0xa1
+#define ID_BLOCK_DURATION 0x9b
+#define ID_REFERENCE_BLOCK 0xfb
+
+/* Tracks Elements */
+#define ID_TRACKS 0x1654ae6b
+#define ID_TRACK_ENTRY 0xae
+#define ID_TRACK_NUMBER 0xd7
+#define ID_TRACK_UID 0x73c5
+#define ID_TRACK_TYPE 0x83
+#define ID_FLAG_ENABLED 0xb9
+#define ID_FLAG_DEFAULT 0x88
+#define ID_FLAG_LACING 0x9c
+#define ID_TRACK_TIMECODE_SCALE 0x23314f
+#define ID_LANGUAGE 0x22b59c
+#define ID_CODEC_ID 0x86
+#define ID_CODEC_PRIVATE 0x63a2
+
+/* Video Elements */
+#define ID_VIDEO 0xe0
+#define ID_PIXEL_WIDTH 0xb0
+#define ID_PIXEL_HEIGHT 0xba
+#define ID_PIXEL_CROP_BOTTOM 0x54aa
+#define ID_PIXEL_CROP_TOP 0x54bb
+#define ID_PIXEL_CROP_LEFT 0x54cc
+#define ID_PIXEL_CROP_RIGHT 0x54dd
+#define ID_DISPLAY_WIDTH 0x54b0
+#define ID_DISPLAY_HEIGHT 0x54ba
+
+/* Audio Elements */
+#define ID_AUDIO 0xe1
+#define ID_SAMPLING_FREQUENCY 0xb5
+#define ID_CHANNELS 0x9f
+#define ID_BIT_DEPTH 0x6264
+
+/* Cues Elements */
+#define ID_CUES 0x1c53bb6b
+#define ID_CUE_POINT 0xbb
+#define ID_CUE_TIME 0xb3
+#define ID_CUE_TRACK_POSITIONS 0xb7
+#define ID_CUE_TRACK 0xf7
+#define ID_CUE_CLUSTER_POSITION 0xf1
+#define ID_CUE_BLOCK_NUMBER 0x5378
+
+/* EBML Types */
+enum ebml_type_enum {
+ TYPE_UNKNOWN,
+ TYPE_MASTER,
+ TYPE_UINT,
+ TYPE_FLOAT,
+ TYPE_INT,
+ TYPE_STRING,
+ TYPE_BINARY
+};
+
+#define LIMIT_STRING (1 << 20)
+#define LIMIT_BINARY (1 << 24)
+#define LIMIT_BLOCK (1 << 30)
+#define LIMIT_FRAME (1 << 28)
+
+/* Field Flags */
+#define DESC_FLAG_NONE 0
+#define DESC_FLAG_MULTI (1 << 0)
+#define DESC_FLAG_SUSPEND (1 << 1)
+#define DESC_FLAG_OFFSET (1 << 2)
+
+/* Block Header Flags */
+#define BLOCK_FLAGS_LACING 6
+
+/* Lacing Constants */
+#define LACING_NONE 0
+#define LACING_XIPH 1
+#define LACING_FIXED 2
+#define LACING_EBML 3
+
+/* Track Types */
+#define TRACK_TYPE_VIDEO 1
+#define TRACK_TYPE_AUDIO 2
+
+/* Track IDs */
+#define TRACK_ID_VP8 "V_VP8"
+#define TRACK_ID_VORBIS "A_VORBIS"
+
+enum vint_mask {
+ MASK_NONE,
+ MASK_FIRST_BIT
+};
+
+struct ebml_binary {
+ unsigned char * data;
+ size_t length;
+};
+
+struct ebml_list_node {
+ struct ebml_list_node * next;
+ uint64_t id;
+ void * data;
+};
+
+struct ebml_list {
+ struct ebml_list_node * head;
+ struct ebml_list_node * tail;
+};
+
+struct ebml_type {
+ union ebml_value {
+ uint64_t u;
+ double f;
+ int64_t i;
+ char * s;
+ struct ebml_binary b;
+ } v;
+ enum ebml_type_enum type;
+ int read;
+};
+
+/* EBML Definitions */
+struct ebml {
+ struct ebml_type ebml_version;
+ struct ebml_type ebml_read_version;
+ struct ebml_type ebml_max_id_length;
+ struct ebml_type ebml_max_size_length;
+ struct ebml_type doctype;
+ struct ebml_type doctype_version;
+ struct ebml_type doctype_read_version;
+};
+
+/* Matroksa Definitions */
+struct seek {
+ struct ebml_type id;
+ struct ebml_type position;
+};
+
+struct seek_head {
+ struct ebml_list seek;
+};
+
+struct info {
+ struct ebml_type timecode_scale;
+ struct ebml_type duration;
+};
+
+struct block_group {
+ struct ebml_type duration;
+ struct ebml_type reference_block;
+};
+
+struct cluster {
+ struct ebml_type timecode;
+ struct ebml_list block_group;
+};
+
+struct video {
+ struct ebml_type pixel_width;
+ struct ebml_type pixel_height;
+ struct ebml_type pixel_crop_bottom;
+ struct ebml_type pixel_crop_top;
+ struct ebml_type pixel_crop_left;
+ struct ebml_type pixel_crop_right;
+ struct ebml_type display_width;
+ struct ebml_type display_height;
+};
+
+struct audio {
+ struct ebml_type sampling_frequency;
+ struct ebml_type channels;
+ struct ebml_type bit_depth;
+};
+
+struct track_entry {
+ struct ebml_type number;
+ struct ebml_type uid;
+ struct ebml_type type;
+ struct ebml_type flag_enabled;
+ struct ebml_type flag_default;
+ struct ebml_type flag_lacing;
+ struct ebml_type track_timecode_scale;
+ struct ebml_type language;
+ struct ebml_type codec_id;
+ struct ebml_type codec_private;
+ struct video video;
+ struct audio audio;
+};
+
+struct tracks {
+ struct ebml_list track_entry;
+};
+
+struct cue_track_positions {
+ struct ebml_type track;
+ struct ebml_type cluster_position;
+ struct ebml_type block_number;
+};
+
+struct cue_point {
+ struct ebml_type time;
+ struct ebml_list cue_track_positions;
+};
+
+struct cues {
+ struct ebml_list cue_point;
+};
+
+struct segment {
+ struct ebml_list seek_head;
+ struct info info;
+ struct ebml_list cluster;
+ struct tracks tracks;
+ struct cues cues;
+};
+
+/* Misc. */
+struct pool_ctx {
+ char dummy;
+};
+
+struct list_node {
+ struct list_node * previous;
+ struct ebml_element_desc * node;
+ unsigned char * data;
+};
+
+struct saved_state {
+ int64_t stream_offset;
+ struct list_node * ancestor;
+ uint64_t last_id;
+ uint64_t last_size;
+};
+
+struct frame {
+ unsigned char * data;
+ size_t length;
+ struct frame * next;
+};
+
+/* Public (opaque) Structures */
+struct nestegg {
+ nestegg_io * io;
+ nestegg_log log;
+ struct pool_ctx * alloc_pool;
+ uint64_t last_id;
+ uint64_t last_size;
+ struct list_node * ancestor;
+ struct ebml ebml;
+ struct segment segment;
+ int64_t segment_offset;
+ unsigned int track_count;
+};
+
+struct nestegg_packet {
+ uint64_t track;
+ uint64_t timecode;
+ struct frame * frame;
+};
+
+/* Element Descriptor */
+struct ebml_element_desc {
+ char const * name;
+ uint64_t id;
+ enum ebml_type_enum type;
+ size_t offset;
+ unsigned int flags;
+ struct ebml_element_desc * children;
+ size_t size;
+ size_t data_offset;
+};
+
+#define E_FIELD(ID, TYPE, STRUCT, FIELD) \
+ { #ID, ID, TYPE, offsetof(STRUCT, FIELD), DESC_FLAG_NONE, NULL, 0, 0 }
+#define E_MASTER(ID, TYPE, STRUCT, FIELD) \
+ { #ID, ID, TYPE, offsetof(STRUCT, FIELD), DESC_FLAG_MULTI, ne_ ## FIELD ## _elements, \
+ sizeof(struct FIELD), 0 }
+#define E_SINGLE_MASTER_O(ID, TYPE, STRUCT, FIELD) \
+ { #ID, ID, TYPE, offsetof(STRUCT, FIELD), DESC_FLAG_OFFSET, ne_ ## FIELD ## _elements, 0, \
+ offsetof(STRUCT, FIELD ## _offset) }
+#define E_SINGLE_MASTER(ID, TYPE, STRUCT, FIELD) \
+ { #ID, ID, TYPE, offsetof(STRUCT, FIELD), DESC_FLAG_NONE, ne_ ## FIELD ## _elements, 0, 0 }
+#define E_SUSPEND(ID, TYPE) \
+ { #ID, ID, TYPE, 0, DESC_FLAG_SUSPEND, NULL, 0, 0 }
+#define E_LAST \
+ { NULL, 0, 0, 0, DESC_FLAG_NONE, NULL, 0, 0 }
+
+/* EBML Element Lists */
+static struct ebml_element_desc ne_ebml_elements[] = {
+ E_FIELD(ID_EBML_VERSION, TYPE_UINT, struct ebml, ebml_version),
+ E_FIELD(ID_EBML_READ_VERSION, TYPE_UINT, struct ebml, ebml_read_version),
+ E_FIELD(ID_EBML_MAX_ID_LENGTH, TYPE_UINT, struct ebml, ebml_max_id_length),
+ E_FIELD(ID_EBML_MAX_SIZE_LENGTH, TYPE_UINT, struct ebml, ebml_max_size_length),
+ E_FIELD(ID_DOCTYPE, TYPE_STRING, struct ebml, doctype),
+ E_FIELD(ID_DOCTYPE_VERSION, TYPE_UINT, struct ebml, doctype_version),
+ E_FIELD(ID_DOCTYPE_READ_VERSION, TYPE_UINT, struct ebml, doctype_read_version),
+ E_LAST
+};
+
+/* WebMedia Element Lists */
+static struct ebml_element_desc ne_seek_elements[] = {
+ E_FIELD(ID_SEEK_ID, TYPE_BINARY, struct seek, id),
+ E_FIELD(ID_SEEK_POSITION, TYPE_UINT, struct seek, position),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_seek_head_elements[] = {
+ E_MASTER(ID_SEEK, TYPE_MASTER, struct seek_head, seek),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_info_elements[] = {
+ E_FIELD(ID_TIMECODE_SCALE, TYPE_UINT, struct info, timecode_scale),
+ E_FIELD(ID_DURATION, TYPE_FLOAT, struct info, duration),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_block_group_elements[] = {
+ E_SUSPEND(ID_BLOCK, TYPE_BINARY),
+ E_FIELD(ID_BLOCK_DURATION, TYPE_UINT, struct block_group, duration),
+ E_FIELD(ID_REFERENCE_BLOCK, TYPE_INT, struct block_group, reference_block),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_cluster_elements[] = {
+ E_FIELD(ID_TIMECODE, TYPE_UINT, struct cluster, timecode),
+ E_MASTER(ID_BLOCK_GROUP, TYPE_MASTER, struct cluster, block_group),
+ E_SUSPEND(ID_SIMPLE_BLOCK, TYPE_BINARY),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_video_elements[] = {
+ E_FIELD(ID_PIXEL_WIDTH, TYPE_UINT, struct video, pixel_width),
+ E_FIELD(ID_PIXEL_HEIGHT, TYPE_UINT, struct video, pixel_height),
+ E_FIELD(ID_PIXEL_CROP_BOTTOM, TYPE_UINT, struct video, pixel_crop_bottom),
+ E_FIELD(ID_PIXEL_CROP_TOP, TYPE_UINT, struct video, pixel_crop_top),
+ E_FIELD(ID_PIXEL_CROP_LEFT, TYPE_UINT, struct video, pixel_crop_left),
+ E_FIELD(ID_PIXEL_CROP_RIGHT, TYPE_UINT, struct video, pixel_crop_right),
+ E_FIELD(ID_DISPLAY_WIDTH, TYPE_UINT, struct video, display_width),
+ E_FIELD(ID_DISPLAY_HEIGHT, TYPE_UINT, struct video, display_height),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_audio_elements[] = {
+ E_FIELD(ID_SAMPLING_FREQUENCY, TYPE_FLOAT, struct audio, sampling_frequency),
+ E_FIELD(ID_CHANNELS, TYPE_UINT, struct audio, channels),
+ E_FIELD(ID_BIT_DEPTH, TYPE_UINT, struct audio, bit_depth),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_track_entry_elements[] = {
+ E_FIELD(ID_TRACK_NUMBER, TYPE_UINT, struct track_entry, number),
+ E_FIELD(ID_TRACK_UID, TYPE_UINT, struct track_entry, uid),
+ E_FIELD(ID_TRACK_TYPE, TYPE_UINT, struct track_entry, type),
+ E_FIELD(ID_FLAG_ENABLED, TYPE_UINT, struct track_entry, flag_enabled),
+ E_FIELD(ID_FLAG_DEFAULT, TYPE_UINT, struct track_entry, flag_default),
+ E_FIELD(ID_FLAG_LACING, TYPE_UINT, struct track_entry, flag_lacing),
+ E_FIELD(ID_TRACK_TIMECODE_SCALE, TYPE_FLOAT, struct track_entry, track_timecode_scale),
+ E_FIELD(ID_LANGUAGE, TYPE_STRING, struct track_entry, language),
+ E_FIELD(ID_CODEC_ID, TYPE_STRING, struct track_entry, codec_id),
+ E_FIELD(ID_CODEC_PRIVATE, TYPE_BINARY, struct track_entry, codec_private),
+ E_SINGLE_MASTER(ID_VIDEO, TYPE_MASTER, struct track_entry, video),
+ E_SINGLE_MASTER(ID_AUDIO, TYPE_MASTER, struct track_entry, audio),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_tracks_elements[] = {
+ E_MASTER(ID_TRACK_ENTRY, TYPE_MASTER, struct tracks, track_entry),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_cue_track_positions_elements[] = {
+ E_FIELD(ID_CUE_TRACK, TYPE_UINT, struct cue_track_positions, track),
+ E_FIELD(ID_CUE_CLUSTER_POSITION, TYPE_UINT, struct cue_track_positions, cluster_position),
+ E_FIELD(ID_CUE_BLOCK_NUMBER, TYPE_UINT, struct cue_track_positions, block_number),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_cue_point_elements[] = {
+ E_FIELD(ID_CUE_TIME, TYPE_UINT, struct cue_point, time),
+ E_MASTER(ID_CUE_TRACK_POSITIONS, TYPE_MASTER, struct cue_point, cue_track_positions),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_cues_elements[] = {
+ E_MASTER(ID_CUE_POINT, TYPE_MASTER, struct cues, cue_point),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_segment_elements[] = {
+ E_MASTER(ID_SEEK_HEAD, TYPE_MASTER, struct segment, seek_head),
+ E_SINGLE_MASTER(ID_INFO, TYPE_MASTER, struct segment, info),
+ E_MASTER(ID_CLUSTER, TYPE_MASTER, struct segment, cluster),
+ E_SINGLE_MASTER(ID_TRACKS, TYPE_MASTER, struct segment, tracks),
+ E_SINGLE_MASTER(ID_CUES, TYPE_MASTER, struct segment, cues),
+ E_LAST
+};
+
+static struct ebml_element_desc ne_top_level_elements[] = {
+ E_SINGLE_MASTER(ID_EBML, TYPE_MASTER, nestegg, ebml),
+ E_SINGLE_MASTER_O(ID_SEGMENT, TYPE_MASTER, nestegg, segment),
+ E_LAST
+};
+
+#undef E_FIELD
+#undef E_MASTER
+#undef E_SINGLE_MASTER_O
+#undef E_SINGLE_MASTER
+#undef E_SUSPEND
+#undef E_LAST
+
+static struct pool_ctx *
+ne_pool_init(void)
+{
+ struct pool_ctx * pool;
+
+ pool = h_malloc(sizeof(*pool));
+ if (!pool)
+ abort();
+ return pool;
+}
+
+static void
+ne_pool_destroy(struct pool_ctx * pool)
+{
+ h_free(pool);
+}
+
+static void *
+ne_pool_alloc(size_t size, struct pool_ctx * pool)
+{
+ void * p;
+
+ p = h_malloc(size);
+ if (!p)
+ abort();
+ hattach(p, pool);
+ memset(p, 0, size);
+ return p;
+}
+
+static void *
+ne_alloc(size_t size)
+{
+ void * p;
+
+ p = calloc(1, size);
+ if (!p)
+ abort();
+ return p;
+}
+
+static int
+ne_io_read(nestegg_io * io, void * buffer, size_t length)
+{
+ return io->read(buffer, length, io->userdata);
+}
+
+static int
+ne_io_seek(nestegg_io * io, int64_t offset, int whence)
+{
+ return io->seek(offset, whence, io->userdata);
+}
+
+static int
+ne_io_read_skip(nestegg_io * io, size_t length)
+{
+ size_t get;
+ unsigned char buf[8192];
+ int r = 1;
+
+ while (length > 0) {
+ get = length < sizeof(buf) ? length : sizeof(buf);
+ r = ne_io_read(io, buf, get);
+ if (r != 1)
+ break;
+ length -= get;
+ }
+
+ return r;
+}
+
+static int64_t
+ne_io_tell(nestegg_io * io)
+{
+ return io->tell(io->userdata);
+}
+
+static int
+ne_bare_read_vint(nestegg_io * io, uint64_t * value, uint64_t * length, enum vint_mask maskflag)
+{
+ int r;
+ unsigned char b;
+ size_t maxlen = 8;
+ unsigned int count = 1, mask = 1 << 7;
+
+ r = ne_io_read(io, &b, 1);
+ if (r != 1)
+ return r;
+
+ while (count < maxlen) {
+ if ((b & mask) != 0)
+ break;
+ mask >>= 1;
+ count += 1;
+ }
+
+ if (length)
+ *length = count;
+ *value = b;
+
+ if (maskflag == MASK_FIRST_BIT)
+ *value = b & ~mask;
+
+ while (--count) {
+ r = ne_io_read(io, &b, 1);
+ if (r != 1)
+ return r;
+ *value <<= 8;
+ *value |= b;
+ }
+
+ return 1;
+}
+
+static int
+ne_read_id(nestegg_io * io, uint64_t * value, uint64_t * length)
+{
+ return ne_bare_read_vint(io, value, length, MASK_NONE);
+}
+
+static int
+ne_read_vint(nestegg_io * io, uint64_t * value, uint64_t * length)
+{
+ return ne_bare_read_vint(io, value, length, MASK_FIRST_BIT);
+}
+
+static int
+ne_read_svint(nestegg_io * io, int64_t * value, uint64_t * length)
+{
+ int r;
+ uint64_t uvalue;
+ uint64_t ulength;
+ int64_t svint_subtr[] = {
+ 0x3f, 0x1fff,
+ 0xfffff, 0x7ffffff,
+ 0x3ffffffffLL, 0x1ffffffffffLL,
+ 0xffffffffffffLL, 0x7fffffffffffffLL
+ };
+
+ r = ne_bare_read_vint(io, &uvalue, &ulength, MASK_FIRST_BIT);
+ if (r != 1)
+ return r;
+ *value = uvalue - svint_subtr[ulength - 1];
+ if (length)
+ *length = ulength;
+ return r;
+}
+
+static int
+ne_read_uint(nestegg_io * io, uint64_t * val, uint64_t length)
+{
+ unsigned char b;
+ int r;
+
+ if (length == 0 || length > 8)
+ return -1;
+ r = ne_io_read(io, &b, 1);
+ if (r != 1)
+ return r;
+ *val = b;
+ while (--length) {
+ r = ne_io_read(io, &b, 1);
+ if (r != 1)
+ return r;
+ *val <<= 8;
+ *val |= b;
+ }
+ return 1;
+}
+
+static int
+ne_read_int(nestegg_io * io, int64_t * val, uint64_t length)
+{
+ int r;
+ uint64_t uval, base;
+
+ r = ne_read_uint(io, &uval, length);
+ if (r != 1)
+ return r;
+
+ if (length < sizeof(int64_t)) {
+ base = 1;
+ base <<= length * 8 - 1;
+ if (uval >= base) {
+ base = 1;
+ base <<= length * 8;
+ } else {
+ base = 0;
+ }
+ *val = uval - base;
+ } else {
+ *val = (int64_t) uval;
+ }
+
+ return 1;
+}
+
+static int
+ne_read_float(nestegg_io * io, double * val, uint64_t length)
+{
+ union {
+ uint64_t u;
+ float f;
+ double d;
+ } value;
+ int r;
+
+ /* length == 10 not implemented */
+ if (length != 4 && length != 8)
+ return -1;
+ r = ne_read_uint(io, &value.u, length);
+ if (r != 1)
+ return r;
+ if (length == 4)
+ *val = value.f;
+ else
+ *val = value.d;
+ return 1;
+}
+
+static int
+ne_read_string(nestegg * ctx, char ** val, uint64_t length)
+{
+ char * str;
+ int r;
+
+ if (length == 0 || length > LIMIT_STRING)
+ return -1;
+ str = ne_pool_alloc(length + 1, ctx->alloc_pool);
+ r = ne_io_read(ctx->io, (unsigned char *) str, length);
+ if (r != 1)
+ return r;
+ str[length] = '\0';
+ *val = str;
+ return 1;
+}
+
+static int
+ne_read_binary(nestegg * ctx, struct ebml_binary * val, uint64_t length)
+{
+ if (length == 0 || length > LIMIT_BINARY)
+ return -1;
+ val->data = ne_pool_alloc(length, ctx->alloc_pool);
+ val->length = length;
+ return ne_io_read(ctx->io, val->data, length);
+}
+
+static int
+ne_get_uint(struct ebml_type type, uint64_t * value)
+{
+ if (!type.read)
+ return -1;
+
+ assert(type.type == TYPE_UINT);
+
+ *value = type.v.u;
+
+ return 0;
+}
+
+static int
+ne_get_float(struct ebml_type type, double * value)
+{
+ if (!type.read)
+ return -1;
+
+ assert(type.type == TYPE_FLOAT);
+
+ *value = type.v.f;
+
+ return 0;
+}
+
+static int
+ne_get_string(struct ebml_type type, char ** value)
+{
+ if (!type.read)
+ return -1;
+
+ assert(type.type == TYPE_STRING);
+
+ *value = type.v.s;
+
+ return 0;
+}
+
+static int
+ne_get_binary(struct ebml_type type, struct ebml_binary * value)
+{
+ if (!type.read)
+ return -1;
+
+ assert(type.type == TYPE_BINARY);
+
+ *value = type.v.b;
+
+ return 0;
+}
+
+static int
+ne_is_ancestor_element(uint64_t id, struct list_node * ancestor)
+{
+ struct ebml_element_desc * element;
+
+ for (; ancestor; ancestor = ancestor->previous)
+ for (element = ancestor->node; element->id; ++element)
+ if (element->id == id)
+ return 1;
+
+ return 0;
+}
+
+static struct ebml_element_desc *
+ne_find_element(uint64_t id, struct ebml_element_desc * elements)
+{
+ struct ebml_element_desc * element;
+
+ for (element = elements; element->id; ++element)
+ if (element->id == id)
+ return element;
+
+ return NULL;
+}
+
+static void
+ne_ctx_push(nestegg * ctx, struct ebml_element_desc * ancestor, void * data)
+{
+ struct list_node * item;
+
+ item = ne_alloc(sizeof(*item));
+ item->previous = ctx->ancestor;
+ item->node = ancestor;
+ item->data = data;
+ ctx->ancestor = item;
+}
+
+static void
+ne_ctx_pop(nestegg * ctx)
+{
+ struct list_node * item;
+
+ item = ctx->ancestor;
+ ctx->ancestor = item->previous;
+ free(item);
+}
+
+static int
+ne_ctx_save(nestegg * ctx, struct saved_state * s)
+{
+ s->stream_offset = ne_io_tell(ctx->io);
+ if (s->stream_offset < 0)
+ return -1;
+ s->ancestor = ctx->ancestor;
+ s->last_id = ctx->last_id;
+ s->last_size = ctx->last_size;
+ return 0;
+}
+
+static int
+ne_ctx_restore(nestegg * ctx, struct saved_state * s)
+{
+ int r;
+
+ r = ne_io_seek(ctx->io, s->stream_offset, NESTEGG_SEEK_SET);
+ if (r != 0)
+ return -1;
+ ctx->ancestor = s->ancestor;
+ ctx->last_id = s->last_id;
+ ctx->last_size = s->last_size;
+ return 0;
+}
+
+static int
+ne_peek_element(nestegg * ctx, uint64_t * id, uint64_t * size)
+{
+ int r;
+
+ if (ctx->last_id && ctx->last_size) {
+ if (id)
+ *id = ctx->last_id;
+ if (size)
+ *size = ctx->last_size;
+ return 1;
+ }
+
+ r = ne_read_id(ctx->io, &ctx->last_id, NULL);
+ if (r != 1)
+ return r;
+
+ r = ne_read_vint(ctx->io, &ctx->last_size, NULL);
+ if (r != 1)
+ return r;
+
+ if (id)
+ *id = ctx->last_id;
+ if (size)
+ *size = ctx->last_size;
+
+ return 1;
+}
+
+static int
+ne_read_element(nestegg * ctx, uint64_t * id, uint64_t * size)
+{
+ int r;
+
+ r = ne_peek_element(ctx, id, size);
+ if (r != 1)
+ return r;
+
+ ctx->last_id = 0;
+ ctx->last_size = 0;
+
+ return 1;
+}
+
+static void
+ne_read_master(nestegg * ctx, struct ebml_element_desc * desc)
+{
+ struct ebml_list * list;
+ struct ebml_list_node * node, * oldtail;
+
+ assert(desc->type == TYPE_MASTER && desc->flags & DESC_FLAG_MULTI);
+
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "multi master element %llx (%s)",
+ desc->id, desc->name);
+
+ list = (struct ebml_list *) (ctx->ancestor->data + desc->offset);
+
+ node = ne_pool_alloc(sizeof(*node), ctx->alloc_pool);
+ node->id = desc->id;
+ node->data = ne_pool_alloc(desc->size, ctx->alloc_pool);
+
+ oldtail = list->tail;
+ if (oldtail)
+ oldtail->next = node;
+ list->tail = node;
+ if (!list->head)
+ list->head = node;
+
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, " -> using data %p", node->data);
+
+ ne_ctx_push(ctx, desc->children, node->data);
+}
+
+static void
+ne_read_single_master(nestegg * ctx, struct ebml_element_desc * desc)
+{
+ assert(desc->type == TYPE_MASTER && !(desc->flags & DESC_FLAG_MULTI));
+
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "single master element %llx (%s)",
+ desc->id, desc->name);
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, " -> using data %p (%u)",
+ ctx->ancestor->data + desc->offset, desc->offset);
+
+ ne_ctx_push(ctx, desc->children, ctx->ancestor->data + desc->offset);
+}
+
+static int
+ne_read_simple(nestegg * ctx, struct ebml_element_desc * desc, size_t length)
+{
+ struct ebml_type * storage;
+ int r;
+
+ storage = (struct ebml_type *) (ctx->ancestor->data + desc->offset);
+
+ if (storage->read) {
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "element %llx (%s) already read, skipping",
+ desc->id, desc->name);
+ return 0;
+ }
+
+ storage->type = desc->type;
+
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "element %llx (%s) -> %p (%u)",
+ desc->id, desc->name, storage, desc->offset);
+
+ r = -1;
+
+ switch (desc->type) {
+ case TYPE_UINT:
+ r = ne_read_uint(ctx->io, &storage->v.u, length);
+ break;
+ case TYPE_FLOAT:
+ r = ne_read_float(ctx->io, &storage->v.f, length);
+ break;
+ case TYPE_INT:
+ r = ne_read_int(ctx->io, &storage->v.i, length);
+ break;
+ case TYPE_STRING:
+ r = ne_read_string(ctx, &storage->v.s, length);
+ break;
+ case TYPE_BINARY:
+ r = ne_read_binary(ctx, &storage->v.b, length);
+ break;
+ case TYPE_MASTER:
+ case TYPE_UNKNOWN:
+ assert(0);
+ break;
+ }
+
+ if (r == 1)
+ storage->read = 1;
+
+ return r;
+}
+
+static int
+ne_parse(nestegg * ctx, struct ebml_element_desc * top_level)
+{
+ int r;
+ int64_t * data_offset;
+ uint64_t id, size;
+ struct ebml_element_desc * element;
+
+ /* loop until we need to return:
+ - hit suspend point
+ - parse complete
+ - error occurred */
+
+ /* loop over elements at current level reading them if sublevel found,
+ push ctx onto stack and continue if sublevel ended, pop ctx off stack
+ and continue */
+
+ if (!ctx->ancestor)
+ return -1;
+
+ for (;;) {
+ r = ne_peek_element(ctx, &id, &size);
+ if (r != 1)
+ break;
+
+ element = ne_find_element(id, ctx->ancestor->node);
+ if (element) {
+ if (element->flags & DESC_FLAG_SUSPEND) {
+ assert(element->type == TYPE_BINARY);
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "suspend parse at %llx", id);
+ r = 1;
+ break;
+ }
+
+ r = ne_read_element(ctx, &id, &size);
+ if (r != 1)
+ break;
+
+ if (element->flags & DESC_FLAG_OFFSET) {
+ data_offset = (int64_t *) (ctx->ancestor->data + element->data_offset);
+ *data_offset = ne_io_tell(ctx->io);
+ if (*data_offset < 0) {
+ r = -1;
+ break;
+ }
+ }
+
+ if (element->type == TYPE_MASTER) {
+ if (element->flags & DESC_FLAG_MULTI)
+ ne_read_master(ctx, element);
+ else
+ ne_read_single_master(ctx, element);
+ continue;
+ } else {
+ r = ne_read_simple(ctx, element, size);
+ if (r < 0)
+ break;
+ }
+ } else if (ne_is_ancestor_element(id, ctx->ancestor->previous)) {
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "parent element %llx", id);
+ if (top_level && ctx->ancestor->node == top_level) {
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "*** parse about to back up past top_level");
+ r = 1;
+ break;
+ }
+ ne_ctx_pop(ctx);
+ } else {
+ r = ne_read_element(ctx, &id, &size);
+ if (r != 1)
+ break;
+
+ if (id != ID_VOID && id != ID_CRC32)
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "unknown element %llx", id);
+ r = ne_io_read_skip(ctx->io, size);
+ if (r != 1)
+ break;
+ }
+ }
+
+ if (r != 1)
+ while (ctx->ancestor)
+ ne_ctx_pop(ctx);
+
+ return r;
+}
+
+static uint64_t
+ne_xiph_lace_value(unsigned char ** np)
+{
+ uint64_t lace;
+ uint64_t value;
+ unsigned char * p = *np;
+
+ lace = *p++;
+ value = lace;
+ while (lace == 255) {
+ lace = *p++;
+ value += lace;
+ }
+
+ *np = p;
+
+ return value;
+}
+
+static int
+ne_read_xiph_lace_value(nestegg_io * io, uint64_t * value, size_t * consumed)
+{
+ int r;
+ uint64_t lace;
+
+ r = ne_read_uint(io, &lace, 1);
+ if (r != 1)
+ return r;
+ *consumed += 1;
+
+ *value = lace;
+ while (lace == 255) {
+ r = ne_read_uint(io, &lace, 1);
+ if (r != 1)
+ return r;
+ *consumed += 1;
+ *value += lace;
+ }
+
+ return 1;
+}
+
+static int
+ne_read_xiph_lacing(nestegg_io * io, size_t block, size_t * read, uint64_t n, uint64_t * sizes)
+{
+ int r;
+ size_t i = 0;
+ uint64_t sum = 0;
+
+ while (--n) {
+ r = ne_read_xiph_lace_value(io, &sizes[i], read);
+ if (r != 1)
+ return r;
+ sum += sizes[i];
+ i += 1;
+ }
+
+ if (*read + sum > block)
+ return -1;
+
+ /* last frame is the remainder of the block */
+ sizes[i] = block - *read - sum;
+ return 1;
+}
+
+static int
+ne_read_ebml_lacing(nestegg_io * io, size_t block, size_t * read, uint64_t n, uint64_t * sizes)
+{
+ int r;
+ uint64_t lace, sum, length;
+ int64_t slace;
+ size_t i = 0;
+
+ r = ne_read_vint(io, &lace, &length);
+ if (r != 1)
+ return r;
+ *read += length;
+
+ sizes[i] = lace;
+ sum = sizes[i];
+
+ i += 1;
+ n -= 1;
+
+ while (--n) {
+ r = ne_read_svint(io, &slace, &length);
+ if (r != 1)
+ return r;
+ *read += length;
+ sizes[i] = sizes[i - 1] + slace;
+ sum += sizes[i];
+ i += 1;
+ }
+
+ if (*read + sum > block)
+ return -1;
+
+ /* last frame is the remainder of the block */
+ sizes[i] = block - *read - sum;
+ return 1;
+}
+
+static uint64_t
+ne_get_timecode_scale(nestegg * ctx)
+{
+ uint64_t scale;
+
+ if (ne_get_uint(ctx->segment.info.timecode_scale, &scale) != 0)
+ scale = 1000000;
+
+ return scale;
+}
+
+static struct track_entry *
+ne_find_track_entry(nestegg * ctx, unsigned int track)
+{
+ struct ebml_list_node * node;
+ unsigned int tracks = 0;
+
+ node = ctx->segment.tracks.track_entry.head;
+ while (node) {
+ assert(node->id == ID_TRACK_ENTRY);
+ if (track == tracks)
+ return node->data;
+ tracks += 1;
+ node = node->next;
+ }
+
+ return NULL;
+}
+
+static int
+ne_read_block(nestegg * ctx, uint64_t block_id, uint64_t block_size, nestegg_packet ** data)
+{
+ int r;
+ int64_t timecode, abs_timecode;
+ nestegg_packet * pkt;
+ struct cluster * cluster;
+ struct frame * f, * last;
+ struct track_entry * entry;
+ double track_scale;
+ uint64_t track, length, frame_sizes[256], cluster_tc, flags, frames, tc_scale, total;
+ unsigned int i, lacing;
+ size_t consumed = 0;
+
+ *data = NULL;
+
+ if (block_size > LIMIT_BLOCK)
+ return -1;
+
+ r = ne_read_vint(ctx->io, &track, &length);
+ if (r != 1)
+ return r;
+
+ if (track == 0 || track > ctx->track_count)
+ return -1;
+
+ consumed += length;
+
+ r = ne_read_int(ctx->io, &timecode, 2);
+ if (r != 1)
+ return r;
+
+ consumed += 2;
+
+ r = ne_read_uint(ctx->io, &flags, 1);
+ if (r != 1)
+ return r;
+
+ consumed += 1;
+
+ frames = 0;
+
+ /* flags are different between block and simpleblock, but lacing is
+ encoded the same way */
+ lacing = (flags & BLOCK_FLAGS_LACING) >> 1;
+
+ switch (lacing) {
+ case LACING_NONE:
+ frames = 1;
+ break;
+ case LACING_XIPH:
+ case LACING_FIXED:
+ case LACING_EBML:
+ r = ne_read_uint(ctx->io, &frames, 1);
+ if (r != 1)
+ return r;
+ consumed += 1;
+ frames += 1;
+ }
+
+ if (frames > 256)
+ return -1;
+
+ switch (lacing) {
+ case LACING_NONE:
+ frame_sizes[0] = block_size - consumed;
+ break;
+ case LACING_XIPH:
+ if (frames == 1)
+ return -1;
+ r = ne_read_xiph_lacing(ctx->io, block_size, &consumed, frames, frame_sizes);
+ if (r != 1)
+ return r;
+ break;
+ case LACING_FIXED:
+ if ((block_size - consumed) % frames)
+ return -1;
+ for (i = 0; i < frames; ++i)
+ frame_sizes[i] = (block_size - consumed) / frames;
+ break;
+ case LACING_EBML:
+ if (frames == 1)
+ return -1;
+ r = ne_read_ebml_lacing(ctx->io, block_size, &consumed, frames, frame_sizes);
+ if (r != 1)
+ return r;
+ break;
+ }
+
+ /* sanity check unlaced frame sizes against total block size. */
+ total = consumed;
+ for (i = 0; i < frames; ++i)
+ total += frame_sizes[i];
+ if (total > block_size)
+ return -1;
+
+ entry = ne_find_track_entry(ctx, (unsigned int)(track - 1));
+ if (!entry)
+ return -1;
+
+ track_scale = 1.0;
+
+ tc_scale = ne_get_timecode_scale(ctx);
+
+ assert(ctx->segment.cluster.tail->id == ID_CLUSTER);
+ cluster = ctx->segment.cluster.tail->data;
+ if (ne_get_uint(cluster->timecode, &cluster_tc) != 0)
+ return -1;
+
+ abs_timecode = timecode + cluster_tc;
+ if (abs_timecode < 0)
+ return -1;
+
+ pkt = ne_alloc(sizeof(*pkt));
+ pkt->track = track - 1;
+ pkt->timecode = (uint64_t)(abs_timecode * tc_scale * track_scale);
+
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "%sblock t %lld pts %f f %llx frames: %llu",
+ block_id == ID_BLOCK ? "" : "simple", pkt->track, pkt->timecode / 1e9, flags, frames);
+
+ last = NULL;
+ for (i = 0; i < frames; ++i) {
+ if (frame_sizes[i] > LIMIT_FRAME) {
+ nestegg_free_packet(pkt);
+ return -1;
+ }
+ f = ne_alloc(sizeof(*f));
+ f->data = ne_alloc(frame_sizes[i]);
+ f->length = frame_sizes[i];
+ r = ne_io_read(ctx->io, f->data, frame_sizes[i]);
+ if (r != 1) {
+ free(f->data);
+ free(f);
+ nestegg_free_packet(pkt);
+ return -1;
+ }
+
+ if (!last)
+ pkt->frame = f;
+ else
+ last->next = f;
+ last = f;
+ }
+
+ *data = pkt;
+
+ return 1;
+}
+
+static uint64_t
+ne_buf_read_id(unsigned char const * p, size_t length)
+{
+ uint64_t id = 0;
+
+ while (length--) {
+ id <<= 8;
+ id |= *p++;
+ }
+
+ return id;
+}
+
+static struct seek *
+ne_find_seek_for_id(struct ebml_list_node * seek_head, uint64_t id)
+{
+ struct ebml_list * head;
+ struct ebml_list_node * seek;
+ struct ebml_binary binary_id;
+ struct seek * s;
+
+ while (seek_head) {
+ assert(seek_head->id == ID_SEEK_HEAD);
+ head = seek_head->data;
+ seek = head->head;
+
+ while (seek) {
+ assert(seek->id == ID_SEEK);
+ s = seek->data;
+
+ if (ne_get_binary(s->id, &binary_id) == 0 &&
+ ne_buf_read_id(binary_id.data, binary_id.length) == id)
+ return s;
+
+ seek = seek->next;
+ }
+
+ seek_head = seek_head->next;
+ }
+
+ return NULL;
+}
+
+static struct cue_point *
+ne_find_cue_point_for_tstamp(struct ebml_list_node * cue_point, uint64_t scale, uint64_t tstamp)
+{
+ uint64_t time;
+ struct cue_point * c, * prev = NULL;
+
+ while (cue_point) {
+ assert(cue_point->id == ID_CUE_POINT);
+ c = cue_point->data;
+
+ if (!prev)
+ prev = c;
+
+ if (ne_get_uint(c->time, &time) == 0 && time * scale > tstamp)
+ break;
+
+ prev = cue_point->data;
+ cue_point = cue_point->next;
+ }
+
+ return prev;
+}
+
+static int
+ne_is_suspend_element(uint64_t id)
+{
+ /* this could search the tree of elements for DESC_FLAG_SUSPEND */
+ if (id == ID_SIMPLE_BLOCK || id == ID_BLOCK)
+ return 1;
+ return 0;
+}
+
+static void
+ne_null_log_callback(nestegg * ctx, unsigned int severity, char const * fmt, ...)
+{
+ if (ctx && severity && fmt)
+ return;
+}
+
+int
+nestegg_init(nestegg ** context, nestegg_io io, nestegg_log callback)
+{
+ int r;
+ uint64_t id, version, docversion;
+ struct ebml_list_node * track;
+ char * doctype;
+ nestegg * ctx = NULL;
+
+ if (!(io.read && io.seek && io.tell))
+ return -1;
+
+ ctx = ne_alloc(sizeof(*ctx));
+
+ ctx->io = ne_alloc(sizeof(*ctx->io));
+ *ctx->io = io;
+ ctx->log = callback;
+ ctx->alloc_pool = ne_pool_init();
+
+ if (!ctx->log)
+ ctx->log = ne_null_log_callback;
+
+ r = ne_peek_element(ctx, &id, NULL);
+ if (r != 1) {
+ nestegg_destroy(ctx);
+ return -1;
+ }
+
+ if (id != ID_EBML) {
+ nestegg_destroy(ctx);
+ return -1;
+ }
+
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "ctx %p", ctx);
+
+ ne_ctx_push(ctx, ne_top_level_elements, ctx);
+
+ r = ne_parse(ctx, NULL);
+
+ if (r != 1) {
+ nestegg_destroy(ctx);
+ return -1;
+ }
+
+ if (ne_get_uint(ctx->ebml.ebml_read_version, &version) != 0)
+ version = 1;
+ if (version != 1) {
+ nestegg_destroy(ctx);
+ return -1;
+ }
+
+ if (ne_get_string(ctx->ebml.doctype, &doctype) != 0)
+ doctype = "matroska";
+ if (strcmp(doctype, "webm") != 0) {
+ nestegg_destroy(ctx);
+ return -1;
+ }
+
+ if (ne_get_uint(ctx->ebml.doctype_read_version, &docversion) != 0)
+ docversion = 1;
+ if (docversion < 1 || docversion > 2) {
+ nestegg_destroy(ctx);
+ return -1;
+ }
+
+ if (!ctx->segment.tracks.track_entry.head) {
+ nestegg_destroy(ctx);
+ return -1;
+ }
+
+ track = ctx->segment.tracks.track_entry.head;
+ ctx->track_count = 0;
+
+ while (track) {
+ ctx->track_count += 1;
+ track = track->next;
+ }
+
+ *context = ctx;
+
+ return 0;
+}
+
+void
+nestegg_destroy(nestegg * ctx)
+{
+ while (ctx->ancestor)
+ ne_ctx_pop(ctx);
+ ne_pool_destroy(ctx->alloc_pool);
+ free(ctx->io);
+ free(ctx);
+}
+
+int
+nestegg_duration(nestegg * ctx, uint64_t * duration)
+{
+ uint64_t tc_scale;
+ double unscaled_duration;
+
+ if (ne_get_float(ctx->segment.info.duration, &unscaled_duration) != 0)
+ return -1;
+
+ tc_scale = ne_get_timecode_scale(ctx);
+
+ *duration = (uint64_t) (unscaled_duration * tc_scale);
+ return 0;
+}
+
+int
+nestegg_tstamp_scale(nestegg * ctx, uint64_t * scale)
+{
+ *scale = ne_get_timecode_scale(ctx);
+ return 0;
+}
+
+int
+nestegg_track_count(nestegg * ctx, unsigned int * tracks)
+{
+ *tracks = ctx->track_count;
+ return 0;
+}
+
+int
+nestegg_track_seek(nestegg * ctx, unsigned int track, uint64_t tstamp)
+{
+ int r;
+ struct cue_point * cue_point;
+ struct cue_track_positions * pos;
+ struct saved_state state;
+ struct seek * found;
+ uint64_t seek_pos, tc_scale, t, id;
+ struct ebml_list_node * node = ctx->segment.cues.cue_point.head;
+
+ /* If there are no cues loaded, check for cues element in the seek head
+ and load it. */
+ if (!node) {
+ found = ne_find_seek_for_id(ctx->segment.seek_head.head, ID_CUES);
+ if (!found)
+ return -1;
+
+ if (ne_get_uint(found->position, &seek_pos) != 0)
+ return -1;
+
+ /* Save old parser state. */
+ r = ne_ctx_save(ctx, &state);
+ if (r != 0)
+ return -1;
+
+ /* Seek and set up parser state for segment-level element (Cues). */
+ r = ne_io_seek(ctx->io, ctx->segment_offset + seek_pos, NESTEGG_SEEK_SET);
+ if (r != 0)
+ return -1;
+ ctx->last_id = 0;
+ ctx->last_size = 0;
+
+ r = ne_read_element(ctx, &id, NULL);
+ if (r != 1)
+ return -1;
+
+ if (id != ID_CUES)
+ return -1;
+
+ ctx->ancestor = NULL;
+ ne_ctx_push(ctx, ne_top_level_elements, ctx);
+ ne_ctx_push(ctx, ne_segment_elements, &ctx->segment);
+ ne_ctx_push(ctx, ne_cues_elements, &ctx->segment.cues);
+ /* parser will run until end of cues element. */
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "seek: parsing cue elements");
+ r = ne_parse(ctx, ne_cues_elements);
+ while (ctx->ancestor)
+ ne_ctx_pop(ctx);
+
+ /* Reset parser state to original state and seek back to old position. */
+ if (ne_ctx_restore(ctx, &state) != 0)
+ return -1;
+
+ if (r < 0)
+ return -1;
+ }
+
+ tc_scale = ne_get_timecode_scale(ctx);
+
+ cue_point = ne_find_cue_point_for_tstamp(ctx->segment.cues.cue_point.head, tc_scale, tstamp);
+ if (!cue_point)
+ return -1;
+
+ node = cue_point->cue_track_positions.head;
+
+ seek_pos = 0;
+
+ while (node) {
+ assert(node->id == ID_CUE_TRACK_POSITIONS);
+ pos = node->data;
+ if (ne_get_uint(pos->track, &t) == 0 && t - 1 == track) {
+ if (ne_get_uint(pos->cluster_position, &seek_pos) != 0)
+ return -1;
+ break;
+ }
+ node = node->next;
+ }
+
+ /* Seek and set up parser state for segment-level element (Cluster). */
+ r = ne_io_seek(ctx->io, ctx->segment_offset + seek_pos, NESTEGG_SEEK_SET);
+ if (r != 0)
+ return -1;
+ ctx->last_id = 0;
+ ctx->last_size = 0;
+
+ while (ctx->ancestor)
+ ne_ctx_pop(ctx);
+
+ ne_ctx_push(ctx, ne_top_level_elements, ctx);
+ ne_ctx_push(ctx, ne_segment_elements, &ctx->segment);
+ ctx->log(ctx, NESTEGG_LOG_DEBUG, "seek: parsing cluster elements");
+ r = ne_parse(ctx, NULL);
+ if (r != 1)
+ return -1;
+
+ if (!ne_is_suspend_element(ctx->last_id))
+ return -1;
+
+ return 0;
+}
+
+int
+nestegg_track_type(nestegg * ctx, unsigned int track)
+{
+ struct track_entry * entry;
+ uint64_t type;
+
+ entry = ne_find_track_entry(ctx, track);
+ if (!entry)
+ return -1;
+
+ if (ne_get_uint(entry->type, &type) != 0)
+ return -1;
+
+ if (type & TRACK_TYPE_VIDEO)
+ return NESTEGG_TRACK_VIDEO;
+
+ if (type & TRACK_TYPE_AUDIO)
+ return NESTEGG_TRACK_AUDIO;
+
+ return -1;
+}
+
+int
+nestegg_track_codec_id(nestegg * ctx, unsigned int track)
+{
+ char * codec_id;
+ struct track_entry * entry;
+
+ entry = ne_find_track_entry(ctx, track);
+ if (!entry)
+ return -1;
+
+ if (ne_get_string(entry->codec_id, &codec_id) != 0)
+ return -1;
+
+ if (strcmp(codec_id, TRACK_ID_VP8) == 0)
+ return NESTEGG_CODEC_VP8;
+
+ if (strcmp(codec_id, TRACK_ID_VORBIS) == 0)
+ return NESTEGG_CODEC_VORBIS;
+
+ return -1;
+}
+
+int
+nestegg_track_codec_data_count(nestegg * ctx, unsigned int track,
+ unsigned int * count)
+{
+ struct track_entry * entry;
+ struct ebml_binary codec_private;
+ unsigned char * p;
+
+ *count = 0;
+
+ entry = ne_find_track_entry(ctx, track);
+ if (!entry)
+ return -1;
+
+ if (nestegg_track_codec_id(ctx, track) != NESTEGG_CODEC_VORBIS)
+ return -1;
+
+ if (ne_get_binary(entry->codec_private, &codec_private) != 0)
+ return -1;
+
+ if (codec_private.length < 1)
+ return -1;
+
+ p = codec_private.data;
+ *count = *p + 1;
+
+ if (*count > 3)
+ return -1;
+
+ return 0;
+}
+
+int
+nestegg_track_codec_data(nestegg * ctx, unsigned int track, unsigned int item,
+ unsigned char ** data, size_t * length)
+{
+ struct track_entry * entry;
+ struct ebml_binary codec_private;
+ uint64_t sizes[3], total;
+ unsigned char * p;
+ unsigned int count, i;
+
+ *data = NULL;
+ *length = 0;
+
+ entry = ne_find_track_entry(ctx, track);
+ if (!entry)
+ return -1;
+
+ if (nestegg_track_codec_id(ctx, track) != NESTEGG_CODEC_VORBIS)
+ return -1;
+
+ if (ne_get_binary(entry->codec_private, &codec_private) != 0)
+ return -1;
+
+ p = codec_private.data;
+ count = *p++ + 1;
+
+ if (count > 3)
+ return -1;
+
+ i = 0;
+ total = 0;
+ while (--count) {
+ sizes[i] = ne_xiph_lace_value(&p);
+ total += sizes[i];
+ i += 1;
+ }
+ sizes[i] = codec_private.length - total - (p - codec_private.data);
+
+ for (i = 0; i < item; ++i) {
+ if (sizes[i] > LIMIT_FRAME)
+ return -1;
+ p += sizes[i];
+ }
+ *data = p;
+ *length = sizes[item];
+
+ return 0;
+}
+
+int
+nestegg_track_video_params(nestegg * ctx, unsigned int track,
+ nestegg_video_params * params)
+{
+ struct track_entry * entry;
+ uint64_t value;
+
+ memset(params, 0, sizeof(*params));
+
+ entry = ne_find_track_entry(ctx, track);
+ if (!entry)
+ return -1;
+
+ if (nestegg_track_type(ctx, track) != NESTEGG_TRACK_VIDEO)
+ return -1;
+
+ if (ne_get_uint(entry->video.pixel_width, &value) != 0)
+ return -1;
+ params->width = (unsigned int)value;
+
+ if (ne_get_uint(entry->video.pixel_height, &value) != 0)
+ return -1;
+ params->height = (unsigned int)value;
+
+ value = 0;
+ ne_get_uint(entry->video.pixel_crop_bottom, &value);
+ params->crop_bottom = (unsigned int)value;
+
+ value = 0;
+ ne_get_uint(entry->video.pixel_crop_top, &value);
+ params->crop_top = (unsigned int)value;
+
+ value = 0;
+ ne_get_uint(entry->video.pixel_crop_left, &value);
+ params->crop_left = (unsigned int)value;
+
+ value = 0;
+ ne_get_uint(entry->video.pixel_crop_right, &value);
+ params->crop_right = (unsigned int)value;
+
+ value = params->width;
+ ne_get_uint(entry->video.display_width, &value);
+ params->display_width = (unsigned int)value;
+
+ value = params->height;
+ ne_get_uint(entry->video.display_height, &value);
+ params->display_height = (unsigned int)value;
+
+ return 0;
+}
+
+int
+nestegg_track_audio_params(nestegg * ctx, unsigned int track,
+ nestegg_audio_params * params)
+{
+ struct track_entry * entry;
+ uint64_t value;
+
+ memset(params, 0, sizeof(*params));
+
+ entry = ne_find_track_entry(ctx, track);
+ if (!entry)
+ return -1;
+
+ if (nestegg_track_type(ctx, track) != NESTEGG_TRACK_AUDIO)
+ return -1;
+
+ params->rate = 8000;
+ ne_get_float(entry->audio.sampling_frequency, ¶ms->rate);
+
+ value = 1;
+ ne_get_uint(entry->audio.channels, &value);
+ params->channels = (unsigned int)value;
+
+ value = 16;
+ ne_get_uint(entry->audio.bit_depth, &value);
+ params->depth = (unsigned int)value;
+
+ return 0;
+}
+
+int
+nestegg_read_packet(nestegg * ctx, nestegg_packet ** pkt)
+{
+ int r;
+ uint64_t id, size;
+
+ *pkt = NULL;
+
+ for (;;) {
+ r = ne_peek_element(ctx, &id, &size);
+ if (r != 1)
+ return r;
+
+ /* any suspend fields must be handled here */
+ if (ne_is_suspend_element(id)) {
+ r = ne_read_element(ctx, &id, &size);
+ if (r != 1)
+ return r;
+
+ /* the only suspend fields are blocks and simple blocks, which we
+ handle directly. */
+ r = ne_read_block(ctx, id, size, pkt);
+ return r;
+ }
+
+ r = ne_parse(ctx, NULL);
+ if (r != 1)
+ return r;
+ }
+
+ return 1;
+}
+
+void
+nestegg_free_packet(nestegg_packet * pkt)
+{
+ struct frame * frame;
+
+ while (pkt->frame) {
+ frame = pkt->frame;
+ pkt->frame = frame->next;
+ free(frame->data);
+ free(frame);
+ }
+
+ free(pkt);
+}
+
+int
+nestegg_packet_track(nestegg_packet * pkt, unsigned int * track)
+{
+ *track = (unsigned int)pkt->track;
+ return 0;
+}
+
+int
+nestegg_packet_tstamp(nestegg_packet * pkt, uint64_t * tstamp)
+{
+ *tstamp = pkt->timecode;
+ return 0;
+}
+
+int
+nestegg_packet_count(nestegg_packet * pkt, unsigned int * count)
+{
+ struct frame * f = pkt->frame;
+
+ *count = 0;
+
+ while (f) {
+ *count += 1;
+ f = f->next;
+ }
+
+ return 0;
+}
+
+int
+nestegg_packet_data(nestegg_packet * pkt, unsigned int item,
+ unsigned char ** data, size_t * length)
+{
+ struct frame * f = pkt->frame;
+ unsigned int count = 0;
+
+ *data = NULL;
+ *length = 0;
+
+ while (f) {
+ if (count == item) {
+ *data = f->data;
+ *length = f->length;
+ return 0;
+ }
+ count += 1;
+ f = f->next;
+ }
+
+ return -1;
+}
diff --git a/nestegg/test/test.c b/libvpx/nestegg/test/test.c
similarity index 100%
rename from nestegg/test/test.c
rename to libvpx/nestegg/test/test.c
diff --git a/libvpx/solution.mk b/libvpx/solution.mk
new file mode 100644
index 0000000..948305f
--- /dev/null
+++ b/libvpx/solution.mk
@@ -0,0 +1,32 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+# libvpx reverse dependencies (targets that depend on libvpx)
+VPX_NONDEPS=$(addsuffix .vcproj,vpx gtest obj_int_extract)
+VPX_RDEPS=$(foreach vcp,\
+ $(filter-out $(VPX_NONDEPS),$^), --dep=$(vcp:.vcproj=):vpx)
+
+vpx.sln: $(wildcard *.vcproj)
+ @echo " [CREATE] $@"
+ $(SRC_PATH_BARE)/build/make/gen_msvs_sln.sh \
+ $(if $(filter vpx.vcproj,$^),$(VPX_RDEPS)) \
+ --dep=vpx:obj_int_extract \
+ --dep=test_libvpx:gtest \
+ --ver=$(CONFIG_VS_VERSION)\
+ --out=$@ $^
+vpx.sln.mk: vpx.sln
+ @true
+
+PROJECTS-yes += vpx.sln vpx.sln.mk
+-include vpx.sln.mk
+
+# Always install this file, as it is an unconditional post-build rule.
+INSTALL_MAPS += src/% $(SRC_PATH_BARE)/%
+INSTALL-SRCS-yes += $(target).mk
diff --git a/libvpx/test/acm_random.h b/libvpx/test/acm_random.h
new file mode 100644
index 0000000..514894e
--- /dev/null
+++ b/libvpx/test/acm_random.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef LIBVPX_TEST_ACM_RANDOM_H_
+#define LIBVPX_TEST_ACM_RANDOM_H_
+
+#include <stdlib.h>
+
+#include "vpx/vpx_integer.h"
+
+namespace libvpx_test {
+
+class ACMRandom {
+ public:
+ ACMRandom() {
+ Reset(DeterministicSeed());
+ }
+
+ explicit ACMRandom(int seed) {
+ Reset(seed);
+ }
+
+ void Reset(int seed) {
+ srand(seed);
+ }
+
+ uint8_t Rand8(void) {
+ return (rand() >> 8) & 0xff;
+ }
+
+ int PseudoUniform(int range) {
+ return (rand() >> 8) % range;
+ }
+
+ int operator()(int n) {
+ return PseudoUniform(n);
+ }
+
+ static int DeterministicSeed(void) {
+ return 0xbaba;
+ }
+};
+
+} // namespace libvpx_test
+
+#endif // LIBVPX_TEST_ACM_RANDOM_H_
diff --git a/libvpx/test/altref_test.cc b/libvpx/test/altref_test.cc
new file mode 100644
index 0000000..ca05577
--- /dev/null
+++ b/libvpx/test/altref_test.cc
@@ -0,0 +1,71 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/encode_test_driver.h"
+#include "test/i420_video_source.h"
+
+namespace {
+
+// lookahead range: [kLookAheadMin, kLookAheadMax).
+const int kLookAheadMin = 5;
+const int kLookAheadMax = 26;
+
+class AltRefTest : public libvpx_test::EncoderTest,
+ public ::testing::TestWithParam<int> {
+ protected:
+ AltRefTest() : altref_count_(0) {}
+ virtual ~AltRefTest() {}
+
+ virtual void SetUp() {
+ InitializeConfig();
+ SetMode(libvpx_test::kTwoPassGood);
+ }
+
+ virtual void BeginPassHook(unsigned int pass) {
+ altref_count_ = 0;
+ }
+
+ virtual bool Continue() const {
+ return !HasFatalFailure() && !abort_;
+ }
+
+ virtual void PreEncodeFrameHook(libvpx_test::VideoSource *video,
+ libvpx_test::Encoder *encoder) {
+ if (video->frame() == 1) {
+ encoder->Control(VP8E_SET_ENABLEAUTOALTREF, 1);
+ encoder->Control(VP8E_SET_CPUUSED, 3);
+ }
+ }
+
+ virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {
+ if (pkt->data.frame.flags & VPX_FRAME_IS_INVISIBLE) ++altref_count_;
+ }
+
+ int altref_count() const { return altref_count_; }
+
+ private:
+ int altref_count_;
+};
+
+TEST_P(AltRefTest, MonotonicTimestamps) {
+ const vpx_rational timebase = { 33333333, 1000000000 };
+ cfg_.g_timebase = timebase;
+ cfg_.rc_target_bitrate = 1000;
+ cfg_.g_lag_in_frames = GetParam();
+
+ libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
+ timebase.den, timebase.num, 0, 30);
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+ EXPECT_GE(altref_count(), 1);
+}
+
+INSTANTIATE_TEST_CASE_P(NonZeroLag, AltRefTest,
+ ::testing::Range(kLookAheadMin, kLookAheadMax));
+} // namespace
diff --git a/libvpx/test/boolcoder_test.cc b/libvpx/test/boolcoder_test.cc
new file mode 100644
index 0000000..4e21be8
--- /dev/null
+++ b/libvpx/test/boolcoder_test.cc
@@ -0,0 +1,90 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+extern "C" {
+#include "vp8/encoder/boolhuff.h"
+#include "vp8/decoder/dboolhuff.h"
+}
+
+#include <math.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+
+#include "test/acm_random.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "vpx/vpx_integer.h"
+
+namespace {
+const int num_tests = 10;
+} // namespace
+
+using libvpx_test::ACMRandom;
+
+TEST(VP8, TestBitIO) {
+ ACMRandom rnd(ACMRandom::DeterministicSeed());
+ for (int n = 0; n < num_tests; ++n) {
+ for (int method = 0; method <= 7; ++method) { // we generate various proba
+ const int bits_to_test = 1000;
+ uint8_t probas[bits_to_test];
+
+ for (int i = 0; i < bits_to_test; ++i) {
+ const int parity = i & 1;
+ probas[i] =
+ (method == 0) ? 0 : (method == 1) ? 255 :
+ (method == 2) ? 128 :
+ (method == 3) ? rnd.Rand8() :
+ (method == 4) ? (parity ? 0 : 255) :
+ // alternate between low and high proba:
+ (method == 5) ? (parity ? rnd(128) : 255 - rnd(128)) :
+ (method == 6) ?
+ (parity ? rnd(64) : 255 - rnd(64)) :
+ (parity ? rnd(32) : 255 - rnd(32));
+ }
+ for (int bit_method = 0; bit_method <= 3; ++bit_method) {
+ const int random_seed = 6432;
+ const int buffer_size = 10000;
+ ACMRandom bit_rnd(random_seed);
+ BOOL_CODER bw;
+ uint8_t bw_buffer[buffer_size];
+ vp8_start_encode(&bw, bw_buffer, bw_buffer + buffer_size);
+
+ int bit = (bit_method == 0) ? 0 : (bit_method == 1) ? 1 : 0;
+ for (int i = 0; i < bits_to_test; ++i) {
+ if (bit_method == 2) {
+ bit = (i & 1);
+ } else if (bit_method == 3) {
+ bit = bit_rnd(2);
+ }
+ vp8_encode_bool(&bw, bit, static_cast<int>(probas[i]));
+ }
+
+ vp8_stop_encode(&bw);
+
+ BOOL_DECODER br;
+ vp8dx_start_decode(&br, bw_buffer, buffer_size);
+ bit_rnd.Reset(random_seed);
+ for (int i = 0; i < bits_to_test; ++i) {
+ if (bit_method == 2) {
+ bit = (i & 1);
+ } else if (bit_method == 3) {
+ bit = bit_rnd(2);
+ }
+ GTEST_ASSERT_EQ(vp8dx_decode_bool(&br, probas[i]), bit)
+ << "pos: "<< i << " / " << bits_to_test
+ << " bit_method: " << bit_method
+ << " method: " << method;
+ }
+ }
+ }
+ }
+}
diff --git a/libvpx/test/config_test.cc b/libvpx/test/config_test.cc
new file mode 100644
index 0000000..c4da46e
--- /dev/null
+++ b/libvpx/test/config_test.cc
@@ -0,0 +1,61 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/encode_test_driver.h"
+#include "test/video_source.h"
+
+namespace {
+
+class ConfigTest : public ::libvpx_test::EncoderTest,
+ public ::testing::TestWithParam<enum libvpx_test::TestMode> {
+ public:
+ ConfigTest() : frame_count_in_(0), frame_count_out_(0), frame_count_max_(0) {}
+
+ protected:
+ virtual void SetUp() {
+ InitializeConfig();
+ SetMode(GetParam());
+ }
+
+ virtual void BeginPassHook(unsigned int /*pass*/) {
+ frame_count_in_ = 0;
+ frame_count_out_ = 0;
+ }
+
+ virtual void PreEncodeFrameHook(libvpx_test::VideoSource* /*video*/) {
+ ++frame_count_in_;
+ abort_ |= (frame_count_in_ >= frame_count_max_);
+ }
+
+ virtual void FramePktHook(const vpx_codec_cx_pkt_t* /*pkt*/) {
+ ++frame_count_out_;
+ }
+
+ virtual bool Continue() const {
+ return !HasFatalFailure() && !abort_;
+ }
+
+ unsigned int frame_count_in_;
+ unsigned int frame_count_out_;
+ unsigned int frame_count_max_;
+};
+
+TEST_P(ConfigTest, LagIsDisabled) {
+ frame_count_max_ = 2;
+ cfg_.g_lag_in_frames = 15;
+
+ libvpx_test::DummyVideoSource video;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+
+ EXPECT_EQ(frame_count_in_, frame_count_out_);
+}
+
+INSTANTIATE_TEST_CASE_P(OnePassModes, ConfigTest, ONE_PASS_TEST_MODES);
+} // namespace
diff --git a/libvpx/test/cq_test.cc b/libvpx/test/cq_test.cc
new file mode 100644
index 0000000..42ee2a2
--- /dev/null
+++ b/libvpx/test/cq_test.cc
@@ -0,0 +1,106 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include <cmath>
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/encode_test_driver.h"
+#include "test/i420_video_source.h"
+
+// CQ level range: [kCQLevelMin, kCQLevelMax).
+const int kCQLevelMin = 4;
+const int kCQLevelMax = 63;
+const int kCQLevelStep = 8;
+const int kCQTargetBitrate = 2000;
+
+namespace {
+
+class CQTest : public libvpx_test::EncoderTest,
+ public ::testing::TestWithParam<int> {
+ protected:
+ CQTest() : cq_level_(GetParam()) { init_flags_ = VPX_CODEC_USE_PSNR; }
+ virtual ~CQTest() {}
+
+ virtual void SetUp() {
+ InitializeConfig();
+ SetMode(libvpx_test::kTwoPassGood);
+ }
+
+ virtual void BeginPassHook(unsigned int /*pass*/) {
+ file_size_ = 0;
+ psnr_ = 0.0;
+ n_frames_ = 0;
+ }
+
+ virtual bool Continue() const {
+ return !HasFatalFailure() && !abort_;
+ }
+
+ virtual void PreEncodeFrameHook(libvpx_test::VideoSource *video,
+ libvpx_test::Encoder *encoder) {
+ if (video->frame() == 1) {
+ if (cfg_.rc_end_usage == VPX_CQ) {
+ encoder->Control(VP8E_SET_CQ_LEVEL, cq_level_);
+ }
+ encoder->Control(VP8E_SET_CPUUSED, 3);
+ }
+ }
+
+ virtual void PSNRPktHook(const vpx_codec_cx_pkt_t *pkt) {
+ psnr_ += pow(10.0, pkt->data.psnr.psnr[0] / 10.0);
+ n_frames_++;
+ }
+
+ virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {
+ file_size_ += pkt->data.frame.sz;
+ }
+
+ double GetLinearPSNROverBitrate() const {
+ double avg_psnr = log10(psnr_ / n_frames_) * 10.0;
+ return pow(10.0, avg_psnr / 10.0) / file_size_;
+ }
+
+ int file_size() const { return file_size_; }
+ int n_frames() const { return n_frames_; }
+
+ private:
+ int cq_level_;
+ int file_size_;
+ double psnr_;
+ int n_frames_;
+};
+
+int prev_actual_bitrate = kCQTargetBitrate;
+TEST_P(CQTest, LinearPSNRIsHigherForCQLevel) {
+ const vpx_rational timebase = { 33333333, 1000000000 };
+ cfg_.g_timebase = timebase;
+ cfg_.rc_target_bitrate = kCQTargetBitrate;
+ cfg_.g_lag_in_frames = 25;
+
+ cfg_.rc_end_usage = VPX_CQ;
+ libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
+ timebase.den, timebase.num, 0, 30);
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+ const double cq_psnr_lin = GetLinearPSNROverBitrate();
+ const int cq_actual_bitrate = file_size() * 8 * 30 / (n_frames() * 1000);
+ EXPECT_LE(cq_actual_bitrate, kCQTargetBitrate);
+ EXPECT_LE(cq_actual_bitrate, prev_actual_bitrate);
+ prev_actual_bitrate = cq_actual_bitrate;
+
+ // try targeting the approximate same bitrate with VBR mode
+ cfg_.rc_end_usage = VPX_VBR;
+ cfg_.rc_target_bitrate = cq_actual_bitrate;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+ const double vbr_psnr_lin = GetLinearPSNROverBitrate();
+ EXPECT_GE(cq_psnr_lin, vbr_psnr_lin);
+}
+
+INSTANTIATE_TEST_CASE_P(CQLevelRange, CQTest,
+ ::testing::Range(kCQLevelMin, kCQLevelMax,
+ kCQLevelStep));
+} // namespace
diff --git a/libvpx/test/datarate_test.cc b/libvpx/test/datarate_test.cc
new file mode 100644
index 0000000..6fbcb64
--- /dev/null
+++ b/libvpx/test/datarate_test.cc
@@ -0,0 +1,178 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include "test/encode_test_driver.h"
+#include "test/i420_video_source.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+namespace {
+
+class DatarateTest : public ::libvpx_test::EncoderTest,
+ public ::testing::TestWithParam<enum libvpx_test::TestMode> {
+ protected:
+ virtual void SetUp() {
+ InitializeConfig();
+ SetMode(GetParam());
+ ResetModel();
+ }
+
+ virtual void ResetModel() {
+ last_pts_ = 0;
+ bits_in_buffer_model_ = cfg_.rc_target_bitrate * cfg_.rc_buf_initial_sz;
+ frame_number_ = 0;
+ first_drop_ = 0;
+ bits_total_ = 0;
+ duration_ = 0.0;
+ }
+
+ virtual bool Continue() const {
+ return !HasFatalFailure() && !abort_;
+ }
+
+ virtual void PreEncodeFrameHook(::libvpx_test::VideoSource *video,
+ ::libvpx_test::Encoder *encoder) {
+ const vpx_rational_t tb = video->timebase();
+ timebase_ = static_cast<double>(tb.num) / tb.den;
+ duration_ = 0;
+ }
+
+ virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {
+ // Time since last timestamp = duration.
+ vpx_codec_pts_t duration = pkt->data.frame.pts - last_pts_;
+
+ // TODO(jimbankoski): Remove these lines when the issue:
+ // http://code.google.com/p/webm/issues/detail?id=496 is fixed.
+ // For now the codec assumes buffer starts at starting buffer rate
+ // plus one frame's time.
+ if (last_pts_ == 0)
+ duration = 1;
+
+ // Add to the buffer the bits we'd expect from a constant bitrate server.
+ bits_in_buffer_model_ += duration * timebase_ * cfg_.rc_target_bitrate
+ * 1000;
+
+ /* Test the buffer model here before subtracting the frame. Do so because
+ * the way the leaky bucket model works in libvpx is to allow the buffer to
+ * empty - and then stop showing frames until we've got enough bits to
+ * show one. As noted in comment below (issue 495), this does not currently
+ * apply to key frames. For now exclude key frames in condition below. */
+ bool key_frame = (pkt->data.frame.flags & VPX_FRAME_IS_KEY) ? true: false;
+ if (!key_frame) {
+ ASSERT_GE(bits_in_buffer_model_, 0) << "Buffer Underrun at frame "
+ << pkt->data.frame.pts;
+ }
+
+ const int frame_size_in_bits = pkt->data.frame.sz * 8;
+
+ // Subtract from the buffer the bits associated with a played back frame.
+ bits_in_buffer_model_ -= frame_size_in_bits;
+
+ // Update the running total of bits for end of test datarate checks.
+ bits_total_ += frame_size_in_bits ;
+
+ // If first drop not set and we have a drop set it to this time.
+ if (!first_drop_ && duration > 1)
+ first_drop_ = last_pts_ + 1;
+
+ // Update the most recent pts.
+ last_pts_ = pkt->data.frame.pts;
+
+ // We update this so that we can calculate the datarate minus the last
+ // frame encoded in the file.
+ bits_in_last_frame_ = frame_size_in_bits;
+
+ ++frame_number_;
+ }
+
+ virtual void EndPassHook(void) {
+ if (bits_total_) {
+ const double file_size_in_kb = bits_total_ / 1000; /* bits per kilobit */
+
+ duration_ = (last_pts_ + 1) * timebase_;
+
+ // Effective file datarate includes the time spent prebuffering.
+ effective_datarate_ = (bits_total_ - bits_in_last_frame_) / 1000.0
+ / (cfg_.rc_buf_initial_sz / 1000.0 + duration_);
+
+ file_datarate_ = file_size_in_kb / duration_;
+ }
+ }
+
+ vpx_codec_pts_t last_pts_;
+ int bits_in_buffer_model_;
+ double timebase_;
+ int frame_number_;
+ vpx_codec_pts_t first_drop_;
+ int64_t bits_total_;
+ double duration_;
+ double file_datarate_;
+ double effective_datarate_;
+ int bits_in_last_frame_;
+};
+
+TEST_P(DatarateTest, BasicBufferModel) {
+ cfg_.rc_buf_initial_sz = 500;
+ cfg_.rc_dropframe_thresh = 1;
+ cfg_.rc_max_quantizer = 56;
+ cfg_.rc_end_usage = VPX_CBR;
+ // 2 pass cbr datarate control has a bug hidden by the small # of
+ // frames selected in this encode. The problem is that even if the buffer is
+ // negative we produce a keyframe on a cutscene. Ignoring datarate
+ // constraints
+ // TODO(jimbankoski): ( Fix when issue
+ // http://code.google.com/p/webm/issues/detail?id=495 is addressed. )
+ ::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
+ 30, 1, 0, 140);
+
+ // There is an issue for low bitrates in real-time mode, where the
+ // effective_datarate slightly overshoots the target bitrate.
+ // This is same the issue as noted about (#495).
+ // TODO(jimbankoski/marpan): Update test to run for lower bitrates (< 100),
+ // when the issue is resolved.
+ for (int i = 100; i < 800; i += 200) {
+ cfg_.rc_target_bitrate = i;
+ ResetModel();
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+ ASSERT_GE(cfg_.rc_target_bitrate, effective_datarate_)
+ << " The datarate for the file exceeds the target!";
+
+ ASSERT_LE(cfg_.rc_target_bitrate, file_datarate_ * 1.3)
+ << " The datarate for the file missed the target!";
+ }
+}
+
+TEST_P(DatarateTest, ChangingDropFrameThresh) {
+ cfg_.rc_buf_initial_sz = 500;
+ cfg_.rc_max_quantizer = 36;
+ cfg_.rc_end_usage = VPX_CBR;
+ cfg_.rc_target_bitrate = 200;
+ cfg_.kf_mode = VPX_KF_DISABLED;
+
+ const int frame_count = 40;
+ ::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
+ 30, 1, 0, frame_count);
+
+ // Here we check that the first dropped frame gets earlier and earlier
+ // as the drop frame threshold is increased.
+
+ const int kDropFrameThreshTestStep = 30;
+ vpx_codec_pts_t last_drop = frame_count;
+ for (int i = 1; i < 91; i += kDropFrameThreshTestStep) {
+ cfg_.rc_dropframe_thresh = i;
+ ResetModel();
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+ ASSERT_LE(first_drop_, last_drop)
+ << " The first dropped frame for drop_thresh " << i
+ << " > first dropped frame for drop_thresh "
+ << i - kDropFrameThreshTestStep;
+ last_drop = first_drop_;
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(AllModes, DatarateTest, ALL_TEST_MODES);
+} // namespace
diff --git a/libvpx/test/decode_test_driver.cc b/libvpx/test/decode_test_driver.cc
new file mode 100644
index 0000000..3610f02
--- /dev/null
+++ b/libvpx/test/decode_test_driver.cc
@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include "test/decode_test_driver.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/video_source.h"
+
+namespace libvpx_test {
+#if CONFIG_VP8_DECODER
+void Decoder::DecodeFrame(const uint8_t *cxdata, int size) {
+ if (!decoder_.priv) {
+ const vpx_codec_err_t res_init = vpx_codec_dec_init(&decoder_,
+ &vpx_codec_vp8_dx_algo,
+ &cfg_, 0);
+ ASSERT_EQ(VPX_CODEC_OK, res_init) << DecodeError();
+ }
+
+ const vpx_codec_err_t res_dec = vpx_codec_decode(&decoder_,
+ cxdata, size, NULL, 0);
+ ASSERT_EQ(VPX_CODEC_OK, res_dec) << DecodeError();
+}
+
+void DecoderTest::RunLoop(CompressedVideoSource *video) {
+ vpx_codec_dec_cfg_t dec_cfg = {0};
+ Decoder decoder(dec_cfg, 0);
+
+ // Decode frames.
+ for (video->Begin(); video->cxdata(); video->Next()) {
+ decoder.DecodeFrame(video->cxdata(), video->frame_size());
+
+ DxDataIterator dec_iter = decoder.GetDxData();
+ const vpx_image_t *img = NULL;
+
+ // Get decompressed data
+ while ((img = dec_iter.Next()))
+ DecompressedFrameHook(*img, video->frame_number());
+ }
+}
+#endif
+} // namespace libvpx_test
diff --git a/libvpx/test/decode_test_driver.h b/libvpx/test/decode_test_driver.h
new file mode 100644
index 0000000..6408bee
--- /dev/null
+++ b/libvpx/test/decode_test_driver.h
@@ -0,0 +1,97 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef TEST_DECODE_TEST_DRIVER_H_
+#define TEST_DECODE_TEST_DRIVER_H_
+#include <cstring>
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "vpx_config.h"
+#include "vpx/vpx_decoder.h"
+#include "vpx/vp8dx.h"
+
+namespace libvpx_test {
+
+class CompressedVideoSource;
+
+// Provides an object to handle decoding output
+class DxDataIterator {
+ public:
+ explicit DxDataIterator(vpx_codec_ctx_t *decoder)
+ : decoder_(decoder), iter_(NULL) {}
+
+ const vpx_image_t *Next() {
+ return vpx_codec_get_frame(decoder_, &iter_);
+ }
+
+ private:
+ vpx_codec_ctx_t *decoder_;
+ vpx_codec_iter_t iter_;
+};
+
+// Provides a simplified interface to manage one video decoding.
+//
+// TODO: similar to Encoder class, the exact services should be
+// added as more tests are added.
+class Decoder {
+ public:
+ Decoder(vpx_codec_dec_cfg_t cfg, unsigned long deadline)
+ : cfg_(cfg), deadline_(deadline) {
+ memset(&decoder_, 0, sizeof(decoder_));
+ }
+
+ ~Decoder() {
+ vpx_codec_destroy(&decoder_);
+ }
+
+ void DecodeFrame(const uint8_t *cxdata, int size);
+
+ DxDataIterator GetDxData() {
+ return DxDataIterator(&decoder_);
+ }
+
+ void set_deadline(unsigned long deadline) {
+ deadline_ = deadline;
+ }
+
+ void Control(int ctrl_id, int arg) {
+ const vpx_codec_err_t res = vpx_codec_control_(&decoder_, ctrl_id, arg);
+ ASSERT_EQ(VPX_CODEC_OK, res) << DecodeError();
+ }
+
+ protected:
+ const char *DecodeError() {
+ const char *detail = vpx_codec_error_detail(&decoder_);
+ return detail ? detail : vpx_codec_error(&decoder_);
+ }
+
+ vpx_codec_ctx_t decoder_;
+ vpx_codec_dec_cfg_t cfg_;
+ unsigned int deadline_;
+};
+
+// Common test functionality for all Decoder tests.
+class DecoderTest {
+ public:
+ // Main loop.
+ virtual void RunLoop(CompressedVideoSource *video);
+
+ // Hook to be called on every decompressed frame.
+ virtual void DecompressedFrameHook(const vpx_image_t& img,
+ const unsigned int frame_number) {}
+
+ protected:
+ DecoderTest() {}
+
+ virtual ~DecoderTest() {}
+};
+
+} // namespace libvpx_test
+
+#endif // TEST_DECODE_TEST_DRIVER_H_
diff --git a/libvpx/test/encode_test_driver.cc b/libvpx/test/encode_test_driver.cc
new file mode 100644
index 0000000..ebb3959
--- /dev/null
+++ b/libvpx/test/encode_test_driver.cc
@@ -0,0 +1,204 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include "vpx_config.h"
+#include "test/encode_test_driver.h"
+#if CONFIG_VP8_DECODER
+#include "test/decode_test_driver.h"
+#endif
+#include "test/video_source.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+namespace libvpx_test {
+void Encoder::EncodeFrame(VideoSource *video, const unsigned long frame_flags) {
+ if (video->img())
+ EncodeFrameInternal(*video, frame_flags);
+ else
+ Flush();
+
+ // Handle twopass stats
+ CxDataIterator iter = GetCxData();
+
+ while (const vpx_codec_cx_pkt_t *pkt = iter.Next()) {
+ if (pkt->kind != VPX_CODEC_STATS_PKT)
+ continue;
+
+ stats_->Append(*pkt);
+ }
+}
+
+void Encoder::EncodeFrameInternal(const VideoSource &video,
+ const unsigned long frame_flags) {
+ vpx_codec_err_t res;
+ const vpx_image_t *img = video.img();
+
+ // Handle first frame initialization
+ if (!encoder_.priv) {
+ cfg_.g_w = img->d_w;
+ cfg_.g_h = img->d_h;
+ cfg_.g_timebase = video.timebase();
+ cfg_.rc_twopass_stats_in = stats_->buf();
+ res = vpx_codec_enc_init(&encoder_, &vpx_codec_vp8_cx_algo, &cfg_,
+ init_flags_);
+ ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
+ }
+
+ // Handle frame resizing
+ if (cfg_.g_w != img->d_w || cfg_.g_h != img->d_h) {
+ cfg_.g_w = img->d_w;
+ cfg_.g_h = img->d_h;
+ res = vpx_codec_enc_config_set(&encoder_, &cfg_);
+ ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
+ }
+
+ // Encode the frame
+ res = vpx_codec_encode(&encoder_,
+ video.img(), video.pts(), video.duration(),
+ frame_flags, deadline_);
+ ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
+}
+
+void Encoder::Flush() {
+ const vpx_codec_err_t res = vpx_codec_encode(&encoder_, NULL, 0, 0, 0,
+ deadline_);
+ ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
+}
+
+void EncoderTest::SetMode(TestMode mode) {
+ switch (mode) {
+ case kRealTime:
+ deadline_ = VPX_DL_REALTIME;
+ break;
+
+ case kOnePassGood:
+ case kTwoPassGood:
+ deadline_ = VPX_DL_GOOD_QUALITY;
+ break;
+
+ case kOnePassBest:
+ case kTwoPassBest:
+ deadline_ = VPX_DL_BEST_QUALITY;
+ break;
+
+ default:
+ ASSERT_TRUE(false) << "Unexpected mode " << mode;
+ }
+
+ if (mode == kTwoPassGood || mode == kTwoPassBest)
+ passes_ = 2;
+ else
+ passes_ = 1;
+}
+// The function should return "true" most of the time, therefore no early
+// break-out is implemented within the match checking process.
+static bool compare_img(const vpx_image_t *img1,
+ const vpx_image_t *img2) {
+ bool match = (img1->fmt == img2->fmt) &&
+ (img1->d_w == img2->d_w) &&
+ (img1->d_h == img2->d_h);
+
+ const unsigned int width_y = img1->d_w;
+ const unsigned int height_y = img1->d_h;
+ unsigned int i;
+ for (i = 0; i < height_y; ++i)
+ match = ( memcmp(img1->planes[VPX_PLANE_Y] + i * img1->stride[VPX_PLANE_Y],
+ img2->planes[VPX_PLANE_Y] + i * img2->stride[VPX_PLANE_Y],
+ width_y) == 0) && match;
+ const unsigned int width_uv = (img1->d_w + 1) >> 1;
+ const unsigned int height_uv = (img1->d_h + 1) >> 1;
+ for (i = 0; i < height_uv; ++i)
+ match = ( memcmp(img1->planes[VPX_PLANE_U] + i * img1->stride[VPX_PLANE_U],
+ img2->planes[VPX_PLANE_U] + i * img2->stride[VPX_PLANE_U],
+ width_uv) == 0) && match;
+ for (i = 0; i < height_uv; ++i)
+ match = ( memcmp(img1->planes[VPX_PLANE_V] + i * img1->stride[VPX_PLANE_V],
+ img2->planes[VPX_PLANE_V] + i * img2->stride[VPX_PLANE_V],
+ width_uv) == 0) && match;
+ return match;
+}
+
+void EncoderTest::RunLoop(VideoSource *video) {
+#if CONFIG_VP8_DECODER
+ vpx_codec_dec_cfg_t dec_cfg = {0};
+#endif
+
+ stats_.Reset();
+
+ for (unsigned int pass = 0; pass < passes_; pass++) {
+ last_pts_ = 0;
+
+ if (passes_ == 1)
+ cfg_.g_pass = VPX_RC_ONE_PASS;
+ else if (pass == 0)
+ cfg_.g_pass = VPX_RC_FIRST_PASS;
+ else
+ cfg_.g_pass = VPX_RC_LAST_PASS;
+
+ BeginPassHook(pass);
+ Encoder encoder(cfg_, deadline_, init_flags_, &stats_);
+#if CONFIG_VP8_DECODER
+ Decoder decoder(dec_cfg, 0);
+ bool has_cxdata = false;
+#endif
+ bool again;
+ for (again = true, video->Begin(); again; video->Next()) {
+ again = video->img() != NULL;
+
+ PreEncodeFrameHook(video);
+ PreEncodeFrameHook(video, &encoder);
+ encoder.EncodeFrame(video, frame_flags_);
+
+ CxDataIterator iter = encoder.GetCxData();
+
+ while (const vpx_codec_cx_pkt_t *pkt = iter.Next()) {
+ again = true;
+
+ switch (pkt->kind) {
+ case VPX_CODEC_CX_FRAME_PKT:
+#if CONFIG_VP8_DECODER
+ has_cxdata = true;
+ decoder.DecodeFrame((const uint8_t*)pkt->data.frame.buf,
+ pkt->data.frame.sz);
+#endif
+ ASSERT_GE(pkt->data.frame.pts, last_pts_);
+ last_pts_ = pkt->data.frame.pts;
+ FramePktHook(pkt);
+ break;
+
+ case VPX_CODEC_PSNR_PKT:
+ PSNRPktHook(pkt);
+ break;
+
+ default:
+ break;
+ }
+ }
+
+#if CONFIG_VP8_DECODER
+ if (has_cxdata) {
+ const vpx_image_t *img_enc = encoder.GetPreviewFrame();
+ DxDataIterator dec_iter = decoder.GetDxData();
+ const vpx_image_t *img_dec = dec_iter.Next();
+ if(img_enc && img_dec) {
+ const bool res = compare_img(img_enc, img_dec);
+ ASSERT_TRUE(res)<< "Encoder/Decoder mismatch found.";
+ }
+ }
+#endif
+ if (!Continue())
+ break;
+ }
+
+ EndPassHook();
+
+ if (!Continue())
+ break;
+ }
+}
+} // namespace libvpx_test
diff --git a/libvpx/test/encode_test_driver.h b/libvpx/test/encode_test_driver.h
new file mode 100644
index 0000000..0141fa9
--- /dev/null
+++ b/libvpx/test/encode_test_driver.h
@@ -0,0 +1,197 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#ifndef TEST_ENCODE_TEST_DRIVER_H_
+#define TEST_ENCODE_TEST_DRIVER_H_
+#include <string>
+#include <vector>
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "vpx/vpx_encoder.h"
+#include "vpx/vp8cx.h"
+
+namespace libvpx_test {
+
+class VideoSource;
+
+enum TestMode {
+ kRealTime,
+ kOnePassGood,
+ kOnePassBest,
+ kTwoPassGood,
+ kTwoPassBest
+};
+#define ALL_TEST_MODES ::testing::Values(::libvpx_test::kRealTime, \
+ ::libvpx_test::kOnePassGood, \
+ ::libvpx_test::kOnePassBest, \
+ ::libvpx_test::kTwoPassGood, \
+ ::libvpx_test::kTwoPassBest)
+
+#define ONE_PASS_TEST_MODES ::testing::Values(::libvpx_test::kRealTime, \
+ ::libvpx_test::kOnePassGood, \
+ ::libvpx_test::kOnePassBest)
+
+
+// Provides an object to handle the libvpx get_cx_data() iteration pattern
+class CxDataIterator {
+ public:
+ explicit CxDataIterator(vpx_codec_ctx_t *encoder)
+ : encoder_(encoder), iter_(NULL) {}
+
+ const vpx_codec_cx_pkt_t *Next() {
+ return vpx_codec_get_cx_data(encoder_, &iter_);
+ }
+
+ private:
+ vpx_codec_ctx_t *encoder_;
+ vpx_codec_iter_t iter_;
+};
+
+// Implements an in-memory store for libvpx twopass statistics
+class TwopassStatsStore {
+ public:
+ void Append(const vpx_codec_cx_pkt_t &pkt) {
+ buffer_.append(reinterpret_cast<char *>(pkt.data.twopass_stats.buf),
+ pkt.data.twopass_stats.sz);
+ }
+
+ vpx_fixed_buf_t buf() {
+ const vpx_fixed_buf_t buf = { &buffer_[0], buffer_.size() };
+ return buf;
+ }
+
+ void Reset() {
+ buffer_.clear();
+ }
+
+ protected:
+ std::string buffer_;
+};
+
+
+// Provides a simplified interface to manage one video encoding pass, given
+// a configuration and video source.
+//
+// TODO(jkoleszar): The exact services it provides and the appropriate
+// level of abstraction will be fleshed out as more tests are written.
+class Encoder {
+ public:
+ Encoder(vpx_codec_enc_cfg_t cfg, unsigned long deadline,
+ const unsigned long init_flags, TwopassStatsStore *stats)
+ : cfg_(cfg), deadline_(deadline), init_flags_(init_flags), stats_(stats) {
+ memset(&encoder_, 0, sizeof(encoder_));
+ }
+
+ ~Encoder() {
+ vpx_codec_destroy(&encoder_);
+ }
+
+ CxDataIterator GetCxData() {
+ return CxDataIterator(&encoder_);
+ }
+
+ const vpx_image_t *GetPreviewFrame() {
+ return vpx_codec_get_preview_frame(&encoder_);
+ }
+ // This is a thin wrapper around vpx_codec_encode(), so refer to
+ // vpx_encoder.h for its semantics.
+ void EncodeFrame(VideoSource *video, const unsigned long frame_flags);
+
+ // Convenience wrapper for EncodeFrame()
+ void EncodeFrame(VideoSource *video) {
+ EncodeFrame(video, 0);
+ }
+
+ void Control(int ctrl_id, int arg) {
+ const vpx_codec_err_t res = vpx_codec_control_(&encoder_, ctrl_id, arg);
+ ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
+ }
+
+ void set_deadline(unsigned long deadline) {
+ deadline_ = deadline;
+ }
+
+ protected:
+ const char *EncoderError() {
+ const char *detail = vpx_codec_error_detail(&encoder_);
+ return detail ? detail : vpx_codec_error(&encoder_);
+ }
+
+ // Encode an image
+ void EncodeFrameInternal(const VideoSource &video,
+ const unsigned long frame_flags);
+
+ // Flush the encoder on EOS
+ void Flush();
+
+ vpx_codec_ctx_t encoder_;
+ vpx_codec_enc_cfg_t cfg_;
+ unsigned long deadline_;
+ unsigned long init_flags_;
+ TwopassStatsStore *stats_;
+};
+
+// Common test functionality for all Encoder tests.
+//
+// This class is a mixin which provides the main loop common to all
+// encoder tests. It provides hooks which can be overridden by subclasses
+// to implement each test's specific behavior, while centralizing the bulk
+// of the boilerplate. Note that it doesn't inherit the gtest testing
+// classes directly, so that tests can be parameterized differently.
+class EncoderTest {
+ protected:
+ EncoderTest() : abort_(false), init_flags_(0), frame_flags_(0),
+ last_pts_(0) {}
+
+ virtual ~EncoderTest() {}
+
+ // Initialize the cfg_ member with the default configuration.
+ void InitializeConfig() {
+ const vpx_codec_err_t res = vpx_codec_enc_config_default(
+ &vpx_codec_vp8_cx_algo, &cfg_, 0);
+ ASSERT_EQ(VPX_CODEC_OK, res);
+ }
+
+ // Map the TestMode enum to the deadline_ and passes_ variables.
+ void SetMode(TestMode mode);
+
+ // Main loop.
+ virtual void RunLoop(VideoSource *video);
+
+ // Hook to be called at the beginning of a pass.
+ virtual void BeginPassHook(unsigned int pass) {}
+
+ // Hook to be called at the end of a pass.
+ virtual void EndPassHook() {}
+
+ // Hook to be called before encoding a frame.
+ virtual void PreEncodeFrameHook(VideoSource *video) {}
+ virtual void PreEncodeFrameHook(VideoSource *video, Encoder *encoder) {}
+
+ // Hook to be called on every compressed data packet.
+ virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {}
+
+ // Hook to be called on every PSNR packet.
+ virtual void PSNRPktHook(const vpx_codec_cx_pkt_t *pkt) {}
+
+ // Hook to determine whether the encode loop should continue.
+ virtual bool Continue() const { return !abort_; }
+
+ bool abort_;
+ vpx_codec_enc_cfg_t cfg_;
+ unsigned int passes_;
+ unsigned long deadline_;
+ TwopassStatsStore stats_;
+ unsigned long init_flags_;
+ unsigned long frame_flags_;
+ vpx_codec_pts_t last_pts_;
+};
+
+} // namespace libvpx_test
+
+#endif // TEST_ENCODE_TEST_DRIVER_H_
diff --git a/libvpx/test/error_resilience_test.cc b/libvpx/test/error_resilience_test.cc
new file mode 100644
index 0000000..25c6731
--- /dev/null
+++ b/libvpx/test/error_resilience_test.cc
@@ -0,0 +1,90 @@
+/*
+ Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+
+ Use of this source code is governed by a BSD-style license
+ that can be found in the LICENSE file in the root of the source
+ tree. An additional intellectual property rights grant can be found
+ in the file PATENTS. All contributing project authors may
+ be found in the AUTHORS file in the root of the source tree.
+*/
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/encode_test_driver.h"
+#include "test/i420_video_source.h"
+
+namespace {
+
+class ErrorResilienceTest : public libvpx_test::EncoderTest,
+ public ::testing::TestWithParam<int> {
+ protected:
+ ErrorResilienceTest() {
+ psnr_ = 0.0;
+ nframes_ = 0;
+ encoding_mode_ = static_cast<libvpx_test::TestMode>(GetParam());
+ }
+ virtual ~ErrorResilienceTest() {}
+
+ virtual void SetUp() {
+ InitializeConfig();
+ SetMode(encoding_mode_);
+ }
+
+ virtual void BeginPassHook(unsigned int /*pass*/) {
+ psnr_ = 0.0;
+ nframes_ = 0;
+ }
+
+ virtual bool Continue() const {
+ return !HasFatalFailure() && !abort_;
+ }
+
+ virtual void PSNRPktHook(const vpx_codec_cx_pkt_t *pkt) {
+ psnr_ += pkt->data.psnr.psnr[0];
+ nframes_++;
+ }
+
+ double GetAveragePsnr() const {
+ if (nframes_)
+ return psnr_ / nframes_;
+ return 0.0;
+ }
+
+ private:
+ double psnr_;
+ unsigned int nframes_;
+ libvpx_test::TestMode encoding_mode_;
+};
+
+TEST_P(ErrorResilienceTest, OnVersusOff) {
+ const vpx_rational timebase = { 33333333, 1000000000 };
+ cfg_.g_timebase = timebase;
+ cfg_.rc_target_bitrate = 2000;
+ cfg_.g_lag_in_frames = 25;
+
+ init_flags_ = VPX_CODEC_USE_PSNR;
+
+ libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
+ timebase.den, timebase.num, 0, 30);
+
+ // Error resilient mode OFF.
+ cfg_.g_error_resilient = 0;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+ const double psnr_resilience_off = GetAveragePsnr();
+ EXPECT_GT(psnr_resilience_off, 25.0);
+
+ // Error resilient mode ON.
+ cfg_.g_error_resilient = 1;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+ const double psnr_resilience_on = GetAveragePsnr();
+ EXPECT_GT(psnr_resilience_on, 25.0);
+
+ // Test that turning on error resilient mode hurts by 10% at most.
+ if (psnr_resilience_off > 0.0) {
+ const double psnr_ratio = psnr_resilience_on / psnr_resilience_off;
+ EXPECT_GE(psnr_ratio, 0.9);
+ EXPECT_LE(psnr_ratio, 1.1);
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(OnOffTest, ErrorResilienceTest,
+ ONE_PASS_TEST_MODES);
+} // namespace
diff --git a/libvpx/test/fdct4x4_test.cc b/libvpx/test/fdct4x4_test.cc
new file mode 100644
index 0000000..619b23d
--- /dev/null
+++ b/libvpx/test/fdct4x4_test.cc
@@ -0,0 +1,169 @@
+/*
+* Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+*
+* Use of this source code is governed by a BSD-style license
+* that can be found in the LICENSE file in the root of the source
+* tree. An additional intellectual property rights grant can be found
+* in the file PATENTS. All contributing project authors may
+* be found in the AUTHORS file in the root of the source tree.
+*/
+
+
+#include <math.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+
+
+extern "C" {
+#include "vpx_rtcd.h"
+}
+
+#include "test/acm_random.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "vpx/vpx_integer.h"
+
+
+namespace {
+
+const int cospi8sqrt2minus1 = 20091;
+const int sinpi8sqrt2 = 35468;
+
+void reference_idct4x4(const int16_t *input, int16_t *output) {
+ const int16_t *ip = input;
+ int16_t *op = output;
+
+ for (int i = 0; i < 4; ++i) {
+ const int a1 = ip[0] + ip[8];
+ const int b1 = ip[0] - ip[8];
+ const int temp1 = (ip[4] * sinpi8sqrt2) >> 16;
+ const int temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
+ const int c1 = temp1 - temp2;
+ const int temp3 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
+ const int temp4 = (ip[12] * sinpi8sqrt2) >> 16;
+ const int d1 = temp3 + temp4;
+ op[0] = a1 + d1;
+ op[12] = a1 - d1;
+ op[4] = b1 + c1;
+ op[8] = b1 - c1;
+ ++ip;
+ ++op;
+ }
+ ip = output;
+ op = output;
+ for (int i = 0; i < 4; ++i) {
+ const int a1 = ip[0] + ip[2];
+ const int b1 = ip[0] - ip[2];
+ const int temp1 = (ip[1] * sinpi8sqrt2) >> 16;
+ const int temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
+ const int c1 = temp1 - temp2;
+ const int temp3 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
+ const int temp4 = (ip[3] * sinpi8sqrt2) >> 16;
+ const int d1 = temp3 + temp4;
+ op[0] = (a1 + d1 + 4) >> 3;
+ op[3] = (a1 - d1 + 4) >> 3;
+ op[1] = (b1 + c1 + 4) >> 3;
+ op[2] = (b1 - c1 + 4) >> 3;
+ ip += 4;
+ op += 4;
+ }
+}
+
+using libvpx_test::ACMRandom;
+
+TEST(Vp8FdctTest, SignBiasCheck) {
+ ACMRandom rnd(ACMRandom::DeterministicSeed());
+ int16_t test_input_block[16];
+ int16_t test_output_block[16];
+ const int pitch = 8;
+ int count_sign_block[16][2];
+ const int count_test_block = 1000000;
+
+ memset(count_sign_block, 0, sizeof(count_sign_block));
+
+ for (int i = 0; i < count_test_block; ++i) {
+ // Initialize a test block with input range [-255, 255].
+ for (int j = 0; j < 16; ++j)
+ test_input_block[j] = rnd.Rand8() - rnd.Rand8();
+
+ vp8_short_fdct4x4_c(test_input_block, test_output_block, pitch);
+
+ for (int j = 0; j < 16; ++j) {
+ if (test_output_block[j] < 0)
+ ++count_sign_block[j][0];
+ else if (test_output_block[j] > 0)
+ ++count_sign_block[j][1];
+ }
+ }
+
+ bool bias_acceptable = true;
+ for (int j = 0; j < 16; ++j)
+ bias_acceptable = bias_acceptable &&
+ (abs(count_sign_block[j][0] - count_sign_block[j][1]) < 10000);
+
+ EXPECT_EQ(true, bias_acceptable)
+ << "Error: 4x4 FDCT has a sign bias > 1% for input range [-255, 255]";
+
+ memset(count_sign_block, 0, sizeof(count_sign_block));
+
+ for (int i = 0; i < count_test_block; ++i) {
+ // Initialize a test block with input range [-15, 15].
+ for (int j = 0; j < 16; ++j)
+ test_input_block[j] = (rnd.Rand8() >> 4) - (rnd.Rand8() >> 4);
+
+ vp8_short_fdct4x4_c(test_input_block, test_output_block, pitch);
+
+ for (int j = 0; j < 16; ++j) {
+ if (test_output_block[j] < 0)
+ ++count_sign_block[j][0];
+ else if (test_output_block[j] > 0)
+ ++count_sign_block[j][1];
+ }
+ }
+
+ bias_acceptable = true;
+ for (int j = 0; j < 16; ++j)
+ bias_acceptable = bias_acceptable &&
+ (abs(count_sign_block[j][0] - count_sign_block[j][1]) < 100000);
+
+ EXPECT_EQ(true, bias_acceptable)
+ << "Error: 4x4 FDCT has a sign bias > 10% for input range [-15, 15]";
+};
+
+TEST(Vp8FdctTest, RoundTripErrorCheck) {
+ ACMRandom rnd(ACMRandom::DeterministicSeed());
+ int max_error = 0;
+ double total_error = 0;
+ const int count_test_block = 1000000;
+ for (int i = 0; i < count_test_block; ++i) {
+ int16_t test_input_block[16];
+ int16_t test_temp_block[16];
+ int16_t test_output_block[16];
+
+ // Initialize a test block with input range [-255, 255].
+ for (int j = 0; j < 16; ++j)
+ test_input_block[j] = rnd.Rand8() - rnd.Rand8();
+
+ const int pitch = 8;
+ vp8_short_fdct4x4_c(test_input_block, test_temp_block, pitch);
+ reference_idct4x4(test_temp_block, test_output_block);
+
+ for (int j = 0; j < 16; ++j) {
+ const int diff = test_input_block[j] - test_output_block[j];
+ const int error = diff * diff;
+ if (max_error < error)
+ max_error = error;
+ total_error += error;
+ }
+ }
+
+ EXPECT_GE(1, max_error )
+ << "Error: FDCT/IDCT has an individual roundtrip error > 1";
+
+ EXPECT_GE(count_test_block, total_error)
+ << "Error: FDCT/IDCT has average roundtrip error > 1 per block";
+};
+
+} // namespace
diff --git a/libvpx/test/i420_video_source.h b/libvpx/test/i420_video_source.h
new file mode 100644
index 0000000..219bd33
--- /dev/null
+++ b/libvpx/test/i420_video_source.h
@@ -0,0 +1,117 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#ifndef TEST_I420_VIDEO_SOURCE_H_
+#define TEST_I420_VIDEO_SOURCE_H_
+#include <cstdio>
+#include <cstdlib>
+
+#include "test/video_source.h"
+
+namespace libvpx_test {
+
+// This class extends VideoSource to allow parsing of raw yv12
+// so that we can do actual file encodes.
+class I420VideoSource : public VideoSource {
+ public:
+ I420VideoSource(const std::string &file_name,
+ unsigned int width, unsigned int height,
+ int rate_numerator, int rate_denominator,
+ unsigned int start, int limit)
+ : file_name_(file_name),
+ input_file_(NULL),
+ img_(NULL),
+ start_(start),
+ limit_(limit),
+ frame_(0),
+ width_(0),
+ height_(0),
+ framerate_numerator_(rate_numerator),
+ framerate_denominator_(rate_denominator) {
+
+ // This initializes raw_sz_, width_, height_ and allocates an img.
+ SetSize(width, height);
+ }
+
+ virtual ~I420VideoSource() {
+ vpx_img_free(img_);
+ if (input_file_)
+ fclose(input_file_);
+ }
+
+ virtual void Begin() {
+ if (input_file_)
+ fclose(input_file_);
+ input_file_ = OpenTestDataFile(file_name_);
+ ASSERT_TRUE(input_file_) << "Input file open failed. Filename: "
+ << file_name_;
+ if (start_) {
+ fseek(input_file_, raw_sz_ * start_, SEEK_SET);
+ }
+
+ frame_ = start_;
+ FillFrame();
+ }
+
+ virtual void Next() {
+ ++frame_;
+ FillFrame();
+ }
+
+ virtual vpx_image_t *img() const { return (frame_ < limit_) ? img_ : NULL; }
+
+ // Models a stream where Timebase = 1/FPS, so pts == frame.
+ virtual vpx_codec_pts_t pts() const { return frame_; }
+
+ virtual unsigned long duration() const { return 1; }
+
+ virtual vpx_rational_t timebase() const {
+ const vpx_rational_t t = { framerate_denominator_, framerate_numerator_ };
+ return t;
+ }
+
+ virtual unsigned int frame() const { return frame_; }
+
+ virtual unsigned int limit() const { return limit_; }
+
+ void SetSize(unsigned int width, unsigned int height) {
+ if (width != width_ || height != height_) {
+ vpx_img_free(img_);
+ img_ = vpx_img_alloc(NULL, VPX_IMG_FMT_VPXI420, width, height, 1);
+ ASSERT_TRUE(img_ != NULL);
+ width_ = width;
+ height_ = height;
+ raw_sz_ = width * height * 3 / 2;
+ }
+ }
+
+ virtual void FillFrame() {
+ // Read a frame from input_file.
+ if (fread(img_->img_data, raw_sz_, 1, input_file_) == 0) {
+ limit_ = frame_;
+ }
+ }
+
+ protected:
+ std::string file_name_;
+ FILE *input_file_;
+ vpx_image_t *img_;
+ size_t raw_sz_;
+ unsigned int start_;
+ unsigned int limit_;
+ unsigned int frame_;
+ unsigned int width_;
+ unsigned int height_;
+ unsigned int framerate_numerator_;
+ unsigned int framerate_denominator_;
+};
+
+} // namespace libvpx_test
+
+#endif // TEST_I420_VIDEO_SOURCE_H_
diff --git a/libvpx/test/idctllm_test.cc b/libvpx/test/idctllm_test.cc
new file mode 100644
index 0000000..dd42e22
--- /dev/null
+++ b/libvpx/test/idctllm_test.cc
@@ -0,0 +1,125 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+extern "C" {
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+}
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+typedef void (*idct_fn_t)(short *input, unsigned char *pred_ptr,
+ int pred_stride, unsigned char *dst_ptr,
+ int dst_stride);
+namespace {
+class IDCTTest : public ::testing::TestWithParam<idct_fn_t>
+{
+ protected:
+ virtual void SetUp()
+ {
+ int i;
+
+ UUT = GetParam();
+ memset(input, 0, sizeof(input));
+ /* Set up guard blocks */
+ for(i=0; i<256; i++)
+ output[i] = ((i&0xF)<4&&(i<64))?0:-1;
+ }
+
+ idct_fn_t UUT;
+ short input[16];
+ unsigned char output[256];
+ unsigned char predict[256];
+};
+
+TEST_P(IDCTTest, TestGuardBlocks)
+{
+ int i;
+
+ for(i=0; i<256; i++)
+ if((i&0xF) < 4 && i<64)
+ EXPECT_EQ(0, output[i]) << i;
+ else
+ EXPECT_EQ(255, output[i]);
+}
+
+TEST_P(IDCTTest, TestAllZeros)
+{
+ int i;
+
+ UUT(input, output, 16, output, 16);
+
+ for(i=0; i<256; i++)
+ if((i&0xF) < 4 && i<64)
+ EXPECT_EQ(0, output[i]) << "i==" << i;
+ else
+ EXPECT_EQ(255, output[i]) << "i==" << i;
+}
+
+TEST_P(IDCTTest, TestAllOnes)
+{
+ int i;
+
+ input[0] = 4;
+ UUT(input, output, 16, output, 16);
+
+ for(i=0; i<256; i++)
+ if((i&0xF) < 4 && i<64)
+ EXPECT_EQ(1, output[i]) << "i==" << i;
+ else
+ EXPECT_EQ(255, output[i]) << "i==" << i;
+}
+
+TEST_P(IDCTTest, TestAddOne)
+{
+ int i;
+
+ for(i=0; i<256; i++)
+ predict[i] = i;
+
+ input[0] = 4;
+ UUT(input, predict, 16, output, 16);
+
+ for(i=0; i<256; i++)
+ if((i&0xF) < 4 && i<64)
+ EXPECT_EQ(i+1, output[i]) << "i==" << i;
+ else
+ EXPECT_EQ(255, output[i]) << "i==" << i;
+}
+
+TEST_P(IDCTTest, TestWithData)
+{
+ int i;
+
+ for(i=0; i<16; i++)
+ input[i] = i;
+
+ UUT(input, output, 16, output, 16);
+
+ for(i=0; i<256; i++)
+ if((i&0xF) > 3 || i>63)
+ EXPECT_EQ(255, output[i]) << "i==" << i;
+ else if(i == 0)
+ EXPECT_EQ(11, output[i]) << "i==" << i;
+ else if(i == 34)
+ EXPECT_EQ(1, output[i]) << "i==" << i;
+ else if(i == 2 || i == 17 || i == 32)
+ EXPECT_EQ(3, output[i]) << "i==" << i;
+ else
+ EXPECT_EQ(0, output[i]) << "i==" << i;
+}
+
+INSTANTIATE_TEST_CASE_P(C, IDCTTest,
+ ::testing::Values(vp8_short_idct4x4llm_c));
+#if HAVE_MMX
+INSTANTIATE_TEST_CASE_P(MMX, IDCTTest,
+ ::testing::Values(vp8_short_idct4x4llm_mmx));
+#endif
+}
diff --git a/libvpx/test/intrapred_test.cc b/libvpx/test/intrapred_test.cc
new file mode 100644
index 0000000..d2e0d61
--- /dev/null
+++ b/libvpx/test/intrapred_test.cc
@@ -0,0 +1,354 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <string.h>
+#include "test/acm_random.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+extern "C" {
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/common/blockd.h"
+#include "vpx_mem/vpx_mem.h"
+}
+
+namespace {
+
+using libvpx_test::ACMRandom;
+
+class IntraPredBase {
+ protected:
+ void SetupMacroblock(uint8_t *data, int block_size, int stride,
+ int num_planes) {
+ memset(&mb_, 0, sizeof(mb_));
+ memset(&mi_, 0, sizeof(mi_));
+ mb_.up_available = 1;
+ mb_.left_available = 1;
+ mb_.mode_info_context = &mi_;
+ stride_ = stride;
+ block_size_ = block_size;
+ num_planes_ = num_planes;
+ for (int p = 0; p < num_planes; p++)
+ data_ptr_[p] = data + stride * (block_size + 1) * p +
+ stride + block_size;
+ }
+
+ void FillRandom() {
+ // Fill edges with random data
+ ACMRandom rnd(ACMRandom::DeterministicSeed());
+ for (int p = 0; p < num_planes_; p++) {
+ for (int x = -1 ; x <= block_size_; x++)
+ data_ptr_[p][x - stride_] = rnd.Rand8();
+ for (int y = 0; y < block_size_; y++)
+ data_ptr_[p][y * stride_ - 1] = rnd.Rand8();
+ }
+ }
+
+ virtual void Predict(MB_PREDICTION_MODE mode) = 0;
+
+ void SetLeftUnavailable() {
+ mb_.left_available = 0;
+ for (int p = 0; p < num_planes_; p++)
+ for (int i = -1; i < block_size_; ++i)
+ data_ptr_[p][stride_ * i - 1] = 129;
+ }
+
+ void SetTopUnavailable() {
+ mb_.up_available = 0;
+ for (int p = 0; p < num_planes_; p++)
+ memset(&data_ptr_[p][-1 - stride_], 127, block_size_ + 2);
+ }
+
+ void SetTopLeftUnavailable() {
+ SetLeftUnavailable();
+ SetTopUnavailable();
+ }
+
+ int BlockSizeLog2Min1() const {
+ switch (block_size_) {
+ case 16:
+ return 3;
+ case 8:
+ return 2;
+ default:
+ return 0;
+ }
+ }
+
+ // check DC prediction output against a reference
+ void CheckDCPrediction() const {
+ for (int p = 0; p < num_planes_; p++) {
+ // calculate expected DC
+ int expected;
+ if (mb_.up_available || mb_.left_available) {
+ int sum = 0, shift = BlockSizeLog2Min1() + mb_.up_available +
+ mb_.left_available;
+ if (mb_.up_available)
+ for (int x = 0; x < block_size_; x++)
+ sum += data_ptr_[p][x - stride_];
+ if (mb_.left_available)
+ for (int y = 0; y < block_size_; y++)
+ sum += data_ptr_[p][y * stride_ - 1];
+ expected = (sum + (1 << (shift - 1))) >> shift;
+ } else
+ expected = 0x80;
+
+ // check that all subsequent lines are equal to the first
+ for (int y = 1; y < block_size_; ++y)
+ ASSERT_EQ(0, memcmp(data_ptr_[p], &data_ptr_[p][y * stride_],
+ block_size_));
+ // within the first line, ensure that each pixel has the same value
+ for (int x = 1; x < block_size_; ++x)
+ ASSERT_EQ(data_ptr_[p][0], data_ptr_[p][x]);
+ // now ensure that that pixel has the expected (DC) value
+ ASSERT_EQ(expected, data_ptr_[p][0]);
+ }
+ }
+
+ // check V prediction output against a reference
+ void CheckVPrediction() const {
+ // check that all lines equal the top border
+ for (int p = 0; p < num_planes_; p++)
+ for (int y = 0; y < block_size_; y++)
+ ASSERT_EQ(0, memcmp(&data_ptr_[p][-stride_],
+ &data_ptr_[p][y * stride_], block_size_));
+ }
+
+ // check H prediction output against a reference
+ void CheckHPrediction() const {
+ // for each line, ensure that each pixel is equal to the left border
+ for (int p = 0; p < num_planes_; p++)
+ for (int y = 0; y < block_size_; y++)
+ for (int x = 0; x < block_size_; x++)
+ ASSERT_EQ(data_ptr_[p][-1 + y * stride_],
+ data_ptr_[p][x + y * stride_]);
+ }
+
+ static int ClipByte(int value) {
+ if (value > 255)
+ return 255;
+ else if (value < 0)
+ return 0;
+ return value;
+ }
+
+ // check TM prediction output against a reference
+ void CheckTMPrediction() const {
+ for (int p = 0; p < num_planes_; p++)
+ for (int y = 0; y < block_size_; y++)
+ for (int x = 0; x < block_size_; x++) {
+ const int expected = ClipByte(data_ptr_[p][x - stride_]
+ + data_ptr_[p][stride_ * y - 1]
+ - data_ptr_[p][-1 - stride_]);
+ ASSERT_EQ(expected, data_ptr_[p][y * stride_ + x]);
+ }
+ }
+
+ // Actual test
+ void RunTest() {
+ {
+ SCOPED_TRACE("DC_PRED");
+ FillRandom();
+ Predict(DC_PRED);
+ CheckDCPrediction();
+ }
+ {
+ SCOPED_TRACE("DC_PRED LEFT");
+ FillRandom();
+ SetLeftUnavailable();
+ Predict(DC_PRED);
+ CheckDCPrediction();
+ }
+ {
+ SCOPED_TRACE("DC_PRED TOP");
+ FillRandom();
+ SetTopUnavailable();
+ Predict(DC_PRED);
+ CheckDCPrediction();
+ }
+ {
+ SCOPED_TRACE("DC_PRED TOP_LEFT");
+ FillRandom();
+ SetTopLeftUnavailable();
+ Predict(DC_PRED);
+ CheckDCPrediction();
+ }
+ {
+ SCOPED_TRACE("H_PRED");
+ FillRandom();
+ Predict(H_PRED);
+ CheckHPrediction();
+ }
+ {
+ SCOPED_TRACE("V_PRED");
+ FillRandom();
+ Predict(V_PRED);
+ CheckVPrediction();
+ }
+ {
+ SCOPED_TRACE("TM_PRED");
+ FillRandom();
+ Predict(TM_PRED);
+ CheckTMPrediction();
+ }
+ }
+
+ MACROBLOCKD mb_;
+ MODE_INFO mi_;
+ uint8_t *data_ptr_[2]; // in the case of Y, only [0] is used
+ int stride_;
+ int block_size_;
+ int num_planes_;
+};
+
+typedef void (*intra_pred_y_fn_t)(MACROBLOCKD *x,
+ uint8_t *yabove_row,
+ uint8_t *yleft,
+ int left_stride,
+ uint8_t *ypred_ptr,
+ int y_stride);
+
+class IntraPredYTest : public ::testing::TestWithParam<intra_pred_y_fn_t>,
+ protected IntraPredBase {
+ public:
+ static void SetUpTestCase() {
+ data_array_ = reinterpret_cast<uint8_t*>(
+ vpx_memalign(kDataAlignment, kDataBufferSize));
+ }
+
+ static void TearDownTestCase() {
+ vpx_free(data_array_);
+ data_array_ = NULL;
+ }
+
+ protected:
+ static const int kBlockSize = 16;
+ static const int kDataAlignment = 16;
+ static const int kStride = kBlockSize * 3;
+ // We use 48 so that the data pointer of the first pixel in each row of
+ // each macroblock is 16-byte aligned, and this gives us access to the
+ // top-left and top-right corner pixels belonging to the top-left/right
+ // macroblocks.
+ // We use 17 lines so we have one line above us for top-prediction.
+ static const int kDataBufferSize = kStride * (kBlockSize + 1);
+
+ virtual void SetUp() {
+ pred_fn_ = GetParam();
+ SetupMacroblock(data_array_, kBlockSize, kStride, 1);
+ }
+
+ virtual void Predict(MB_PREDICTION_MODE mode) {
+ mb_.mode_info_context->mbmi.mode = mode;
+ pred_fn_(&mb_, data_ptr_[0] - kStride, data_ptr_[0] - 1, kStride,
+ data_ptr_[0], kStride);
+ }
+
+ intra_pred_y_fn_t pred_fn_;
+ static uint8_t* data_array_;
+};
+
+uint8_t* IntraPredYTest::data_array_ = NULL;
+
+TEST_P(IntraPredYTest, IntraPredTests) {
+ RunTest();
+}
+
+INSTANTIATE_TEST_CASE_P(C, IntraPredYTest,
+ ::testing::Values(
+ vp8_build_intra_predictors_mby_s_c));
+#if HAVE_SSE2
+INSTANTIATE_TEST_CASE_P(SSE2, IntraPredYTest,
+ ::testing::Values(
+ vp8_build_intra_predictors_mby_s_sse2));
+#endif
+#if HAVE_SSSE3
+INSTANTIATE_TEST_CASE_P(SSSE3, IntraPredYTest,
+ ::testing::Values(
+ vp8_build_intra_predictors_mby_s_ssse3));
+#endif
+
+typedef void (*intra_pred_uv_fn_t)(MACROBLOCKD *x,
+ uint8_t *uabove_row,
+ uint8_t *vabove_row,
+ uint8_t *uleft,
+ uint8_t *vleft,
+ int left_stride,
+ uint8_t *upred_ptr,
+ uint8_t *vpred_ptr,
+ int pred_stride);
+
+class IntraPredUVTest : public ::testing::TestWithParam<intra_pred_uv_fn_t>,
+ protected IntraPredBase {
+ public:
+ static void SetUpTestCase() {
+ data_array_ = reinterpret_cast<uint8_t*>(
+ vpx_memalign(kDataAlignment, kDataBufferSize));
+ }
+
+ static void TearDownTestCase() {
+ vpx_free(data_array_);
+ data_array_ = NULL;
+ }
+
+ protected:
+ static const int kBlockSize = 8;
+ static const int kDataAlignment = 8;
+ static const int kStride = kBlockSize * 3;
+ // We use 24 so that the data pointer of the first pixel in each row of
+ // each macroblock is 8-byte aligned, and this gives us access to the
+ // top-left and top-right corner pixels belonging to the top-left/right
+ // macroblocks.
+ // We use 9 lines so we have one line above us for top-prediction.
+ // [0] = U, [1] = V
+ static const int kDataBufferSize = 2 * kStride * (kBlockSize + 1);
+
+ virtual void SetUp() {
+ pred_fn_ = GetParam();
+ SetupMacroblock(data_array_, kBlockSize, kStride, 2);
+ }
+
+ virtual void Predict(MB_PREDICTION_MODE mode) {
+ mb_.mode_info_context->mbmi.uv_mode = mode;
+ pred_fn_(&mb_, data_ptr_[0] - kStride, data_ptr_[1] - kStride,
+ data_ptr_[0] - 1, data_ptr_[1] - 1, kStride,
+ data_ptr_[0], data_ptr_[1], kStride);
+ }
+
+ intra_pred_uv_fn_t pred_fn_;
+ // We use 24 so that the data pointer of the first pixel in each row of
+ // each macroblock is 8-byte aligned, and this gives us access to the
+ // top-left and top-right corner pixels belonging to the top-left/right
+ // macroblocks.
+ // We use 9 lines so we have one line above us for top-prediction.
+ // [0] = U, [1] = V
+ static uint8_t* data_array_;
+};
+
+uint8_t* IntraPredUVTest::data_array_ = NULL;
+
+TEST_P(IntraPredUVTest, IntraPredTests) {
+ RunTest();
+}
+
+INSTANTIATE_TEST_CASE_P(C, IntraPredUVTest,
+ ::testing::Values(
+ vp8_build_intra_predictors_mbuv_s_c));
+#if HAVE_SSE2
+INSTANTIATE_TEST_CASE_P(SSE2, IntraPredUVTest,
+ ::testing::Values(
+ vp8_build_intra_predictors_mbuv_s_sse2));
+#endif
+#if HAVE_SSSE3
+INSTANTIATE_TEST_CASE_P(SSSE3, IntraPredUVTest,
+ ::testing::Values(
+ vp8_build_intra_predictors_mbuv_s_ssse3));
+#endif
+
+} // namespace
diff --git a/libvpx/test/ivf_video_source.h b/libvpx/test/ivf_video_source.h
new file mode 100644
index 0000000..48c3a7d
--- /dev/null
+++ b/libvpx/test/ivf_video_source.h
@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#ifndef TEST_IVF_VIDEO_SOURCE_H_
+#define TEST_IVF_VIDEO_SOURCE_H_
+#include <cstdio>
+#include <cstdlib>
+#include <new>
+#include <string>
+#include "test/video_source.h"
+
+namespace libvpx_test {
+const unsigned int kCodeBufferSize = 256 * 1024;
+const unsigned int kIvfFileHdrSize = 32;
+const unsigned int kIvfFrameHdrSize = 12;
+
+static unsigned int MemGetLe32(const uint8_t *mem) {
+ return (mem[3] << 24) | (mem[2] << 16) | (mem[1] << 8) | (mem[0]);
+}
+
+// This class extends VideoSource to allow parsing of ivf files,
+// so that we can do actual file decodes.
+class IVFVideoSource : public CompressedVideoSource {
+ public:
+ IVFVideoSource(const std::string &file_name)
+ : file_name_(file_name),
+ input_file_(NULL),
+ compressed_frame_buf_(NULL),
+ frame_sz_(0),
+ frame_(0),
+ end_of_file_(false) {
+ }
+
+ virtual ~IVFVideoSource() {
+ delete[] compressed_frame_buf_;
+
+ if (input_file_)
+ fclose(input_file_);
+ }
+
+ virtual void Init() {
+ // Allocate a buffer for read in the compressed video frame.
+ compressed_frame_buf_ = new uint8_t[libvpx_test::kCodeBufferSize];
+ ASSERT_TRUE(compressed_frame_buf_) << "Allocate frame buffer failed";
+ }
+
+ virtual void Begin() {
+ input_file_ = OpenTestDataFile(file_name_);
+ ASSERT_TRUE(input_file_) << "Input file open failed. Filename: "
+ << file_name_;
+
+ // Read file header
+ uint8_t file_hdr[kIvfFileHdrSize];
+ ASSERT_EQ(kIvfFileHdrSize, fread(file_hdr, 1, kIvfFileHdrSize, input_file_))
+ << "File header read failed.";
+ // Check file header
+ ASSERT_TRUE(file_hdr[0] == 'D' && file_hdr[1] == 'K' && file_hdr[2] == 'I'
+ && file_hdr[3] == 'F') << "Input is not an IVF file.";
+
+ FillFrame();
+ }
+
+ virtual void Next() {
+ ++frame_;
+ FillFrame();
+ }
+
+ void FillFrame() {
+ uint8_t frame_hdr[kIvfFrameHdrSize];
+ // Check frame header and read a frame from input_file.
+ if (fread(frame_hdr, 1, kIvfFrameHdrSize, input_file_)
+ != kIvfFrameHdrSize) {
+ end_of_file_ = true;
+ } else {
+ end_of_file_ = false;
+
+ frame_sz_ = MemGetLe32(frame_hdr);
+ ASSERT_LE(frame_sz_, kCodeBufferSize)
+ << "Frame is too big for allocated code buffer";
+ ASSERT_EQ(frame_sz_,
+ fread(compressed_frame_buf_, 1, frame_sz_, input_file_))
+ << "Failed to read complete frame";
+ }
+ }
+
+ virtual const uint8_t *cxdata() const {
+ return end_of_file_ ? NULL : compressed_frame_buf_;
+ }
+ virtual const unsigned int frame_size() const { return frame_sz_; }
+ virtual const unsigned int frame_number() const { return frame_; }
+
+ protected:
+ std::string file_name_;
+ FILE *input_file_;
+ uint8_t *compressed_frame_buf_;
+ unsigned int frame_sz_;
+ unsigned int frame_;
+ bool end_of_file_;
+};
+
+} // namespace libvpx_test
+
+#endif // TEST_IVF_VIDEO_SOURCE_H_
diff --git a/libvpx/test/keyframe_test.cc b/libvpx/test/keyframe_test.cc
new file mode 100644
index 0000000..d0c81df
--- /dev/null
+++ b/libvpx/test/keyframe_test.cc
@@ -0,0 +1,145 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include <climits>
+#include <vector>
+#include "test/encode_test_driver.h"
+#include "test/i420_video_source.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+namespace {
+
+class KeyframeTest : public ::libvpx_test::EncoderTest,
+ public ::testing::TestWithParam<enum libvpx_test::TestMode> {
+ protected:
+ virtual void SetUp() {
+ InitializeConfig();
+ SetMode(GetParam());
+ kf_count_ = 0;
+ kf_count_max_ = INT_MAX;
+ kf_do_force_kf_ = false;
+ set_cpu_used_ = 0;
+ }
+
+ virtual bool Continue() const {
+ return !HasFatalFailure() && !abort_;
+ }
+
+ virtual void PreEncodeFrameHook(::libvpx_test::VideoSource *video,
+ ::libvpx_test::Encoder *encoder) {
+ if (kf_do_force_kf_)
+ frame_flags_ = (video->frame() % 3) ? 0 : VPX_EFLAG_FORCE_KF;
+ if (set_cpu_used_ && video->frame() == 1)
+ encoder->Control(VP8E_SET_CPUUSED, set_cpu_used_);
+ }
+
+ virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {
+ if (pkt->data.frame.flags & VPX_FRAME_IS_KEY) {
+ kf_pts_list_.push_back(pkt->data.frame.pts);
+ kf_count_++;
+ abort_ |= kf_count_ > kf_count_max_;
+ }
+ }
+
+ bool kf_do_force_kf_;
+ int kf_count_;
+ int kf_count_max_;
+ std::vector<vpx_codec_pts_t> kf_pts_list_;
+ int set_cpu_used_;
+};
+
+TEST_P(KeyframeTest, TestRandomVideoSource) {
+ // Validate that encoding the RandomVideoSource produces multiple keyframes.
+ // This validates the results of the TestDisableKeyframes test.
+ kf_count_max_ = 2; // early exit successful tests.
+
+ ::libvpx_test::RandomVideoSource video;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+
+ // In realtime mode - auto placed keyframes are exceedingly rare, don't
+ // bother with this check if(GetParam() > 0)
+ if(GetParam() > 0)
+ EXPECT_GT(kf_count_, 1);
+}
+
+TEST_P(KeyframeTest, TestDisableKeyframes) {
+ cfg_.kf_mode = VPX_KF_DISABLED;
+ kf_count_max_ = 1; // early exit failed tests.
+
+ ::libvpx_test::RandomVideoSource video;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+
+ EXPECT_EQ(1, kf_count_);
+}
+
+TEST_P(KeyframeTest, TestForceKeyframe) {
+ cfg_.kf_mode = VPX_KF_DISABLED;
+ kf_do_force_kf_ = true;
+
+ ::libvpx_test::DummyVideoSource video;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+
+ // verify that every third frame is a keyframe.
+ for (std::vector<vpx_codec_pts_t>::const_iterator iter = kf_pts_list_.begin();
+ iter != kf_pts_list_.end(); ++iter) {
+ ASSERT_EQ(0, *iter % 3) << "Unexpected keyframe at frame " << *iter;
+ }
+}
+
+TEST_P(KeyframeTest, TestKeyframeMaxDistance) {
+ cfg_.kf_max_dist = 25;
+
+ ::libvpx_test::DummyVideoSource video;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+
+ // verify that keyframe interval matches kf_max_dist
+ for (std::vector<vpx_codec_pts_t>::const_iterator iter = kf_pts_list_.begin();
+ iter != kf_pts_list_.end(); ++iter) {
+ ASSERT_EQ(0, *iter % 25) << "Unexpected keyframe at frame " << *iter;
+ }
+}
+
+TEST_P(KeyframeTest, TestAutoKeyframe) {
+ cfg_.kf_mode = VPX_KF_AUTO;
+ kf_do_force_kf_ = false;
+
+ // Force a deterministic speed step in Real Time mode, as the faster modes
+ // may not produce a keyframe like we expect. This is necessary when running
+ // on very slow environments (like Valgrind). The step -11 was determined
+ // experimentally as the fastest mode that still throws the keyframe.
+ if (deadline_ == VPX_DL_REALTIME)
+ set_cpu_used_ = -11;
+
+ // This clip has a cut scene every 30 frames -> Frame 0, 30, 60, 90, 120.
+ // I check only the first 40 frames to make sure there's a keyframe at frame
+ // 0 and 30.
+ ::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
+ 30, 1, 0, 40);
+
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+
+ // In realtime mode - auto placed keyframes are exceedingly rare, don't
+ // bother with this check
+ if(GetParam() > 0)
+ EXPECT_EQ(2u, kf_pts_list_.size()) << " Not the right number of keyframes ";
+
+ // Verify that keyframes match the file keyframes in the file.
+ for (std::vector<vpx_codec_pts_t>::const_iterator iter = kf_pts_list_.begin();
+ iter != kf_pts_list_.end(); ++iter) {
+
+ if (deadline_ == VPX_DL_REALTIME && *iter > 0)
+ EXPECT_EQ(0, (*iter - 1) % 30) << "Unexpected keyframe at frame "
+ << *iter;
+ else
+ EXPECT_EQ(0, *iter % 30) << "Unexpected keyframe at frame " << *iter;
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(AllModes, KeyframeTest, ALL_TEST_MODES);
+} // namespace
diff --git a/libvpx/test/pp_filter_test.cc b/libvpx/test/pp_filter_test.cc
new file mode 100644
index 0000000..af2f3bd
--- /dev/null
+++ b/libvpx/test/pp_filter_test.cc
@@ -0,0 +1,106 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include "third_party/googletest/src/include/gtest/gtest.h"
+extern "C" {
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx/vpx_integer.h"
+#include "vpx_mem/vpx_mem.h"
+}
+
+typedef void (*post_proc_func_t)(unsigned char *src_ptr,
+ unsigned char *dst_ptr,
+ int src_pixels_per_line,
+ int dst_pixels_per_line,
+ int cols,
+ unsigned char *flimit,
+ int size);
+
+namespace {
+
+class Vp8PostProcessingFilterTest
+ : public ::testing::TestWithParam<post_proc_func_t> {};
+
+// Test routine for the VP8 post-processing function
+// vp8_post_proc_down_and_across_mb_row_c.
+
+TEST_P(Vp8PostProcessingFilterTest, FilterOutputCheck) {
+ // Size of the underlying data block that will be filtered.
+ const int block_width = 16;
+ const int block_height = 16;
+
+ // 5-tap filter needs 2 padding rows above and below the block in the input.
+ const int input_width = block_width;
+ const int input_height = block_height + 4;
+ const int input_stride = input_width;
+ const int input_size = input_width * input_height;
+
+ // Filter extends output block by 8 samples at left and right edges.
+ const int output_width = block_width + 16;
+ const int output_height = block_height;
+ const int output_stride = output_width;
+ const int output_size = output_width * output_height;
+
+ uint8_t *const src_image =
+ reinterpret_cast<uint8_t*>(vpx_calloc(input_size, 1));
+ uint8_t *const dst_image =
+ reinterpret_cast<uint8_t*>(vpx_calloc(output_size, 1));
+
+ // Pointers to top-left pixel of block in the input and output images.
+ uint8_t *const src_image_ptr = src_image + (input_stride << 1);
+ uint8_t *const dst_image_ptr = dst_image + 8;
+ uint8_t *const flimits = reinterpret_cast<uint8_t *>(vpx_memalign(16, block_width));
+ (void)vpx_memset(flimits, 255, block_width);
+
+ // Initialize pixels in the input:
+ // block pixels to value 1,
+ // border pixels to value 10.
+ (void)vpx_memset(src_image, 10, input_size);
+ uint8_t *pixel_ptr = src_image_ptr;
+ for (int i = 0; i < block_height; ++i) {
+ for (int j = 0; j < block_width; ++j) {
+ pixel_ptr[j] = 1;
+ }
+ pixel_ptr += input_stride;
+ }
+
+ // Initialize pixels in the output to 99.
+ (void)vpx_memset(dst_image, 99, output_size);
+
+ GetParam()(src_image_ptr, dst_image_ptr, input_stride,
+ output_stride, block_width, flimits, 16);
+
+ static const uint8_t expected_data[block_height] = {
+ 4, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 4
+ };
+
+ pixel_ptr = dst_image_ptr;
+ for (int i = 0; i < block_height; ++i) {
+ for (int j = 0; j < block_width; ++j) {
+ EXPECT_EQ(expected_data[i], pixel_ptr[j])
+ << "Vp8PostProcessingFilterTest failed with invalid filter output";
+ }
+ pixel_ptr += output_stride;
+ }
+
+ vpx_free(src_image);
+ vpx_free(dst_image);
+ vpx_free(flimits);
+};
+
+INSTANTIATE_TEST_CASE_P(C, Vp8PostProcessingFilterTest,
+ ::testing::Values(vp8_post_proc_down_and_across_mb_row_c));
+
+#if HAVE_SSE2
+INSTANTIATE_TEST_CASE_P(SSE2, Vp8PostProcessingFilterTest,
+ ::testing::Values(vp8_post_proc_down_and_across_mb_row_sse2));
+#endif
+
+} // namespace
diff --git a/libvpx/test/resize_test.cc b/libvpx/test/resize_test.cc
new file mode 100644
index 0000000..c846157
--- /dev/null
+++ b/libvpx/test/resize_test.cc
@@ -0,0 +1,104 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include <climits>
+#include <vector>
+#include "test/encode_test_driver.h"
+#include "test/video_source.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+namespace {
+
+const unsigned int kInitialWidth = 320;
+const unsigned int kInitialHeight = 240;
+
+unsigned int ScaleForFrameNumber(unsigned int frame, unsigned int val) {
+ if (frame < 10)
+ return val;
+ if (frame < 20)
+ return val / 2;
+ if (frame < 30)
+ return val * 2 / 3;
+ if (frame < 40)
+ return val / 4;
+ if (frame < 50)
+ return val * 7 / 8;
+ return val;
+}
+
+class ResizingVideoSource : public ::libvpx_test::DummyVideoSource {
+ public:
+ ResizingVideoSource() {
+ SetSize(kInitialWidth, kInitialHeight);
+ limit_ = 60;
+ }
+
+ protected:
+ virtual void Next() {
+ ++frame_;
+ SetSize(ScaleForFrameNumber(frame_, kInitialWidth),
+ ScaleForFrameNumber(frame_, kInitialHeight));
+ FillFrame();
+ }
+};
+
+class ResizeTest : public ::libvpx_test::EncoderTest,
+ public ::testing::TestWithParam<enum libvpx_test::TestMode> {
+ protected:
+ struct FrameInfo {
+ FrameInfo(vpx_codec_pts_t _pts, unsigned int _w, unsigned int _h)
+ : pts(_pts), w(_w), h(_h) {}
+
+ vpx_codec_pts_t pts;
+ unsigned int w;
+ unsigned int h;
+ };
+
+ virtual void SetUp() {
+ InitializeConfig();
+ SetMode(GetParam());
+ }
+
+ virtual bool Continue() const {
+ return !HasFatalFailure() && !abort_;
+ }
+
+ virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {
+ if (pkt->data.frame.flags & VPX_FRAME_IS_KEY) {
+ const unsigned char *buf =
+ reinterpret_cast<const unsigned char *>(pkt->data.frame.buf);
+ const unsigned int w = (buf[6] | (buf[7] << 8)) & 0x3fff;
+ const unsigned int h = (buf[8] | (buf[9] << 8)) & 0x3fff;
+
+ frame_info_list_.push_back(FrameInfo(pkt->data.frame.pts, w, h));
+ }
+ }
+
+ std::vector< FrameInfo > frame_info_list_;
+};
+
+TEST_P(ResizeTest, TestExternalResizeWorks) {
+ ResizingVideoSource video;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+
+ for (std::vector<FrameInfo>::iterator info = frame_info_list_.begin();
+ info != frame_info_list_.end(); ++info) {
+ const vpx_codec_pts_t pts = info->pts;
+ const unsigned int expected_w = ScaleForFrameNumber(pts, kInitialWidth);
+ const unsigned int expected_h = ScaleForFrameNumber(pts, kInitialHeight);
+
+ EXPECT_EQ(expected_w, info->w)
+ << "Frame " << pts << "had unexpected width";
+ EXPECT_EQ(expected_h, info->h)
+ << "Frame " << pts << "had unexpected height";
+ }
+}
+
+INSTANTIATE_TEST_CASE_P(OnePass, ResizeTest, ONE_PASS_TEST_MODES);
+} // namespace
diff --git a/libvpx/test/sad_test.cc b/libvpx/test/sad_test.cc
new file mode 100644
index 0000000..2b562e6
--- /dev/null
+++ b/libvpx/test/sad_test.cc
@@ -0,0 +1,250 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <string.h>
+#include <limits.h>
+#include <stdio.h>
+
+extern "C" {
+#include "./vpx_config.h"
+#include "./vpx_rtcd.h"
+#include "vp8/common/blockd.h"
+#include "vpx_mem/vpx_mem.h"
+}
+
+#include "test/acm_random.h"
+#include "test/util.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+
+typedef unsigned int (*sad_m_by_n_fn_t)(const unsigned char *source_ptr,
+ int source_stride,
+ const unsigned char *reference_ptr,
+ int reference_stride,
+ unsigned int max_sad);
+
+using libvpx_test::ACMRandom;
+
+namespace {
+class SADTest : public PARAMS(int, int, sad_m_by_n_fn_t) {
+ public:
+ static void SetUpTestCase() {
+ source_data_ = reinterpret_cast<uint8_t*>(
+ vpx_memalign(kDataAlignment, kDataBufferSize));
+ reference_data_ = reinterpret_cast<uint8_t*>(
+ vpx_memalign(kDataAlignment, kDataBufferSize));
+ }
+
+ static void TearDownTestCase() {
+ vpx_free(source_data_);
+ source_data_ = NULL;
+ vpx_free(reference_data_);
+ reference_data_ = NULL;
+ }
+
+ protected:
+ static const int kDataAlignment = 16;
+ static const int kDataBufferSize = 16 * 32;
+
+ virtual void SetUp() {
+ sad_fn_ = GET_PARAM(2);
+ height_ = GET_PARAM(1);
+ width_ = GET_PARAM(0);
+ source_stride_ = width_ * 2;
+ reference_stride_ = width_ * 2;
+ rnd_.Reset(ACMRandom::DeterministicSeed());
+ }
+
+ sad_m_by_n_fn_t sad_fn_;
+ virtual unsigned int SAD(unsigned int max_sad) {
+ return sad_fn_(source_data_, source_stride_,
+ reference_data_, reference_stride_,
+ max_sad);
+ }
+
+ // Sum of Absolute Differences. Given two blocks, calculate the absolute
+ // difference between two pixels in the same relative location; accumulate.
+ unsigned int ReferenceSAD(unsigned int max_sad) {
+ unsigned int sad = 0;
+
+ for (int h = 0; h < height_; ++h) {
+ for (int w = 0; w < width_; ++w) {
+ sad += abs(source_data_[h * source_stride_ + w]
+ - reference_data_[h * reference_stride_ + w]);
+ }
+ if (sad > max_sad) {
+ break;
+ }
+ }
+ return sad;
+ }
+
+ void FillConstant(uint8_t *data, int stride, uint8_t fill_constant) {
+ for (int h = 0; h < height_; ++h) {
+ for (int w = 0; w < width_; ++w) {
+ data[h * stride + w] = fill_constant;
+ }
+ }
+ }
+
+ void FillRandom(uint8_t *data, int stride) {
+ for (int h = 0; h < height_; ++h) {
+ for (int w = 0; w < width_; ++w) {
+ data[h * stride + w] = rnd_.Rand8();
+ }
+ }
+ }
+
+ void CheckSad(unsigned int max_sad) {
+ unsigned int reference_sad, exp_sad;
+
+ reference_sad = ReferenceSAD(max_sad);
+ exp_sad = SAD(max_sad);
+
+ if (reference_sad <= max_sad) {
+ ASSERT_EQ(exp_sad, reference_sad);
+ } else {
+ // Alternative implementations are not required to check max_sad
+ ASSERT_GE(exp_sad, reference_sad);
+ }
+ }
+
+ // Handle blocks up to 16x16 with stride up to 32
+ int height_, width_;
+ static uint8_t* source_data_;
+ int source_stride_;
+ static uint8_t* reference_data_;
+ int reference_stride_;
+
+ ACMRandom rnd_;
+};
+
+uint8_t* SADTest::source_data_ = NULL;
+uint8_t* SADTest::reference_data_ = NULL;
+
+TEST_P(SADTest, MaxRef) {
+ FillConstant(source_data_, source_stride_, 0);
+ FillConstant(reference_data_, reference_stride_, 255);
+ CheckSad(UINT_MAX);
+}
+
+TEST_P(SADTest, MaxSrc) {
+ FillConstant(source_data_, source_stride_, 255);
+ FillConstant(reference_data_, reference_stride_, 0);
+ CheckSad(UINT_MAX);
+}
+
+TEST_P(SADTest, ShortRef) {
+ int tmp_stride = reference_stride_;
+ reference_stride_ >>= 1;
+ FillRandom(source_data_, source_stride_);
+ FillRandom(reference_data_, reference_stride_);
+ CheckSad(UINT_MAX);
+ reference_stride_ = tmp_stride;
+}
+
+TEST_P(SADTest, UnalignedRef) {
+ // The reference frame, but not the source frame, may be unaligned for
+ // certain types of searches.
+ int tmp_stride = reference_stride_;
+ reference_stride_ -= 1;
+ FillRandom(source_data_, source_stride_);
+ FillRandom(reference_data_, reference_stride_);
+ CheckSad(UINT_MAX);
+ reference_stride_ = tmp_stride;
+}
+
+TEST_P(SADTest, ShortSrc) {
+ int tmp_stride = source_stride_;
+ source_stride_ >>= 1;
+ FillRandom(source_data_, source_stride_);
+ FillRandom(reference_data_, reference_stride_);
+ CheckSad(UINT_MAX);
+ source_stride_ = tmp_stride;
+}
+
+TEST_P(SADTest, MaxSAD) {
+ // Verify that, when max_sad is set, the implementation does not return a
+ // value lower than the reference.
+ FillConstant(source_data_, source_stride_, 255);
+ FillConstant(reference_data_, reference_stride_, 0);
+ CheckSad(128);
+}
+
+using std::tr1::make_tuple;
+
+const sad_m_by_n_fn_t sad_16x16_c = vp8_sad16x16_c;
+const sad_m_by_n_fn_t sad_8x16_c = vp8_sad8x16_c;
+const sad_m_by_n_fn_t sad_16x8_c = vp8_sad16x8_c;
+const sad_m_by_n_fn_t sad_8x8_c = vp8_sad8x8_c;
+const sad_m_by_n_fn_t sad_4x4_c = vp8_sad4x4_c;
+INSTANTIATE_TEST_CASE_P(C, SADTest, ::testing::Values(
+ make_tuple(16, 16, sad_16x16_c),
+ make_tuple(8, 16, sad_8x16_c),
+ make_tuple(16, 8, sad_16x8_c),
+ make_tuple(8, 8, sad_8x8_c),
+ make_tuple(4, 4, sad_4x4_c)));
+
+// ARM tests
+#if HAVE_MEDIA
+const sad_m_by_n_fn_t sad_16x16_armv6 = vp8_sad16x16_armv6;
+INSTANTIATE_TEST_CASE_P(MEDIA, SADTest, ::testing::Values(
+ make_tuple(16, 16, sad_16x16_armv6)));
+
+#endif
+#if HAVE_NEON
+const sad_m_by_n_fn_t sad_16x16_neon = vp8_sad16x16_neon;
+const sad_m_by_n_fn_t sad_8x16_neon = vp8_sad8x16_neon;
+const sad_m_by_n_fn_t sad_16x8_neon = vp8_sad16x8_neon;
+const sad_m_by_n_fn_t sad_8x8_neon = vp8_sad8x8_neon;
+const sad_m_by_n_fn_t sad_4x4_neon = vp8_sad4x4_neon;
+INSTANTIATE_TEST_CASE_P(NEON, SADTest, ::testing::Values(
+ make_tuple(16, 16, sad_16x16_neon),
+ make_tuple(8, 16, sad_8x16_neon),
+ make_tuple(16, 8, sad_16x8_neon),
+ make_tuple(8, 8, sad_8x8_neon),
+ make_tuple(4, 4, sad_4x4_neon)));
+#endif
+
+// X86 tests
+#if HAVE_MMX
+const sad_m_by_n_fn_t sad_16x16_mmx = vp8_sad16x16_mmx;
+const sad_m_by_n_fn_t sad_8x16_mmx = vp8_sad8x16_mmx;
+const sad_m_by_n_fn_t sad_16x8_mmx = vp8_sad16x8_mmx;
+const sad_m_by_n_fn_t sad_8x8_mmx = vp8_sad8x8_mmx;
+const sad_m_by_n_fn_t sad_4x4_mmx = vp8_sad4x4_mmx;
+INSTANTIATE_TEST_CASE_P(MMX, SADTest, ::testing::Values(
+ make_tuple(16, 16, sad_16x16_mmx),
+ make_tuple(8, 16, sad_8x16_mmx),
+ make_tuple(16, 8, sad_16x8_mmx),
+ make_tuple(8, 8, sad_8x8_mmx),
+ make_tuple(4, 4, sad_4x4_mmx)));
+#endif
+#if HAVE_SSE2
+const sad_m_by_n_fn_t sad_16x16_wmt = vp8_sad16x16_wmt;
+const sad_m_by_n_fn_t sad_8x16_wmt = vp8_sad8x16_wmt;
+const sad_m_by_n_fn_t sad_16x8_wmt = vp8_sad16x8_wmt;
+const sad_m_by_n_fn_t sad_8x8_wmt = vp8_sad8x8_wmt;
+const sad_m_by_n_fn_t sad_4x4_wmt = vp8_sad4x4_wmt;
+INSTANTIATE_TEST_CASE_P(SSE2, SADTest, ::testing::Values(
+ make_tuple(16, 16, sad_16x16_wmt),
+ make_tuple(8, 16, sad_8x16_wmt),
+ make_tuple(16, 8, sad_16x8_wmt),
+ make_tuple(8, 8, sad_8x8_wmt),
+ make_tuple(4, 4, sad_4x4_wmt)));
+#endif
+#if HAVE_SSSE3
+const sad_m_by_n_fn_t sad_16x16_sse3 = vp8_sad16x16_sse3;
+INSTANTIATE_TEST_CASE_P(SSE3, SADTest, ::testing::Values(
+ make_tuple(16, 16, sad_16x16_sse3)));
+#endif
+
+} // namespace
diff --git a/libvpx/test/set_roi.cc b/libvpx/test/set_roi.cc
new file mode 100644
index 0000000..3b6112e
--- /dev/null
+++ b/libvpx/test/set_roi.cc
@@ -0,0 +1,182 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <math.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "vpx/vpx_integer.h"
+#include "vpx_mem/vpx_mem.h"
+extern "C" {
+#include "vp8/encoder/onyx_int.h"
+}
+
+namespace {
+
+TEST(Vp8RoiMapTest, ParameterCheck) {
+ int delta_q[MAX_MB_SEGMENTS] = { -2, -25, 0, 31 };
+ int delta_lf[MAX_MB_SEGMENTS] = { -2, -25, 0, 31 };
+ unsigned int threshold[MAX_MB_SEGMENTS] = { 0, 100, 200, 300 };
+
+ const int internalq_trans[] = {
+ 0, 1, 2, 3, 4, 5, 7, 8,
+ 9, 10, 12, 13, 15, 17, 18, 19,
+ 20, 21, 23, 24, 25, 26, 27, 28,
+ 29, 30, 31, 33, 35, 37, 39, 41,
+ 43, 45, 47, 49, 51, 53, 55, 57,
+ 59, 61, 64, 67, 70, 73, 76, 79,
+ 82, 85, 88, 91, 94, 97, 100, 103,
+ 106, 109, 112, 115, 118, 121, 124, 127,
+ };
+
+ // Initialize elements of cpi with valid defaults.
+ VP8_COMP cpi;
+ cpi.mb.e_mbd.mb_segement_abs_delta = SEGMENT_DELTADATA;
+ cpi.cyclic_refresh_mode_enabled = 0;
+ cpi.mb.e_mbd.segmentation_enabled = 0;
+ cpi.mb.e_mbd.update_mb_segmentation_map = 0;
+ cpi.mb.e_mbd.update_mb_segmentation_data = 0;
+ cpi.common.mb_rows = 240 >> 4;
+ cpi.common.mb_cols = 320 >> 4;
+ const int mbs = (cpi.common.mb_rows * cpi.common.mb_cols);
+ vpx_memset(cpi.segment_feature_data, 0, sizeof(cpi.segment_feature_data));
+
+ // Segment map
+ cpi.segmentation_map = reinterpret_cast<unsigned char *>(vpx_calloc(mbs, 1));
+
+ // Allocate memory for the source memory map.
+ unsigned char *roi_map =
+ reinterpret_cast<unsigned char *>(vpx_calloc(mbs, 1));
+ vpx_memset(&roi_map[mbs >> 2], 1, (mbs >> 2));
+ vpx_memset(&roi_map[mbs >> 1], 2, (mbs >> 2));
+ vpx_memset(&roi_map[mbs -(mbs >> 2)], 3, (mbs >> 2));
+
+ // Do a test call with valid parameters.
+ int roi_retval = vp8_set_roimap(&cpi, roi_map, cpi.common.mb_rows,
+ cpi.common.mb_cols, delta_q, delta_lf,
+ threshold);
+ EXPECT_EQ(0, roi_retval)
+ << "vp8_set_roimap roi failed with default test parameters";
+
+ // Check that the values in the cpi structure get set as expected.
+ if (roi_retval == 0) {
+ // Check that the segment map got set.
+ const int mapcompare = memcmp(roi_map, cpi.segmentation_map, mbs);
+ EXPECT_EQ(0, mapcompare) << "segment map error";
+
+ // Check the q deltas (note the need to translate into
+ // the interanl range of 0-127.
+ for (int i = 0; i < MAX_MB_SEGMENTS; ++i) {
+ const int transq = internalq_trans[abs(delta_q[i])];
+ if (abs(cpi.segment_feature_data[MB_LVL_ALT_Q][i]) != transq) {
+ EXPECT_EQ(transq, cpi.segment_feature_data[MB_LVL_ALT_Q][i])
+ << "segment delta_q error";
+ break;
+ }
+ }
+
+ // Check the loop filter deltas
+ for (int i = 0; i < MAX_MB_SEGMENTS; ++i) {
+ if (cpi.segment_feature_data[MB_LVL_ALT_LF][i] != delta_lf[i]) {
+ EXPECT_EQ(delta_lf[i], cpi.segment_feature_data[MB_LVL_ALT_LF][i])
+ << "segment delta_lf error";
+ break;
+ }
+ }
+
+ // Check the breakout thresholds
+ for (int i = 0; i < MAX_MB_SEGMENTS; ++i) {
+ unsigned int breakout =
+ static_cast<unsigned int>(cpi.segment_encode_breakout[i]);
+
+ if (threshold[i] != breakout) {
+ EXPECT_EQ(threshold[i], breakout)
+ << "breakout threshold error";
+ break;
+ }
+ }
+
+ // Segmentation, and segmentation update flages should be set.
+ EXPECT_EQ(1, cpi.mb.e_mbd.segmentation_enabled)
+ << "segmentation_enabled error";
+ EXPECT_EQ(1, cpi.mb.e_mbd.update_mb_segmentation_map)
+ << "update_mb_segmentation_map error";
+ EXPECT_EQ(1, cpi.mb.e_mbd.update_mb_segmentation_data)
+ << "update_mb_segmentation_data error";
+
+
+ // Try a range of delta q and lf parameters (some legal, some not)
+ for (int i = 0; i < 1000; ++i) {
+ int rand_deltas[4];
+ int deltas_valid;
+ rand_deltas[0] = (rand() % 160) - 80;
+ rand_deltas[1] = (rand() % 160) - 80;
+ rand_deltas[2] = (rand() % 160) - 80;
+ rand_deltas[3] = (rand() % 160) - 80;
+
+ deltas_valid = ((abs(rand_deltas[0]) <= 63) &&
+ (abs(rand_deltas[1]) <= 63) &&
+ (abs(rand_deltas[2]) <= 63) &&
+ (abs(rand_deltas[3]) <= 63)) ? 0 : -1;
+
+ // Test with random delta q values.
+ roi_retval = vp8_set_roimap(&cpi, roi_map, cpi.common.mb_rows,
+ cpi.common.mb_cols, rand_deltas,
+ delta_lf, threshold);
+ EXPECT_EQ(deltas_valid, roi_retval) << "dq range check error";
+
+ // One delta_q error shown at a time
+ if (deltas_valid != roi_retval)
+ break;
+
+ // Test with random loop filter values.
+ roi_retval = vp8_set_roimap(&cpi, roi_map, cpi.common.mb_rows,
+ cpi.common.mb_cols, delta_q,
+ rand_deltas, threshold);
+ EXPECT_EQ(deltas_valid, roi_retval) << "dlf range check error";
+
+ // One delta loop filter error shown at a time
+ if (deltas_valid != roi_retval)
+ break;
+ }
+
+ // Test that we report and error if cyclic refresh is enabled.
+ cpi.cyclic_refresh_mode_enabled = 1;
+ roi_retval = vp8_set_roimap(&cpi, roi_map, cpi.common.mb_rows,
+ cpi.common.mb_cols, delta_q,
+ delta_lf, threshold);
+ EXPECT_EQ(-1, roi_retval) << "cyclic refresh check error";
+ cpi.cyclic_refresh_mode_enabled = 0;
+
+ // Test invalid number of rows or colums.
+ roi_retval = vp8_set_roimap(&cpi, roi_map, cpi.common.mb_rows + 1,
+ cpi.common.mb_cols, delta_q,
+ delta_lf, threshold);
+ EXPECT_EQ(-1, roi_retval) << "MB rows bounds check error";
+
+ roi_retval = vp8_set_roimap(&cpi, roi_map, cpi.common.mb_rows,
+ cpi.common.mb_cols - 1, delta_q,
+ delta_lf, threshold);
+ EXPECT_EQ(-1, roi_retval) << "MB cols bounds check error";
+ }
+
+ // Free allocated memory
+ if (cpi.segmentation_map)
+ vpx_free(cpi.segmentation_map);
+ if (roi_map)
+ vpx_free(roi_map);
+};
+
+} // namespace
diff --git a/libvpx/test/sixtap_predict_test.cc b/libvpx/test/sixtap_predict_test.cc
new file mode 100644
index 0000000..06f14a1
--- /dev/null
+++ b/libvpx/test/sixtap_predict_test.cc
@@ -0,0 +1,222 @@
+/*
+* Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+*
+* Use of this source code is governed by a BSD-style license
+* that can be found in the LICENSE file in the root of the source
+* tree. An additional intellectual property rights grant can be found
+* in the file PATENTS. All contributing project authors may
+* be found in the AUTHORS file in the root of the source tree.
+*/
+
+#include <math.h>
+#include <stdlib.h>
+#include <string.h>
+#include "test/acm_random.h"
+#include "test/util.h"
+#include "third_party/googletest/src/include/gtest/gtest.h"
+extern "C" {
+#include "./vpx_config.h"
+#include "./vpx_rtcd.h"
+#include "vpx/vpx_integer.h"
+#include "vpx_mem/vpx_mem.h"
+}
+
+namespace {
+
+typedef void (*sixtap_predict_fn_t)(uint8_t *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ uint8_t *dst_ptr,
+ int dst_pitch);
+
+class SixtapPredictTest : public PARAMS(int, int, sixtap_predict_fn_t) {
+ public:
+ static void SetUpTestCase() {
+ src_ = reinterpret_cast<uint8_t*>(vpx_memalign(kDataAlignment, kSrcSize));
+ dst_ = reinterpret_cast<uint8_t*>(vpx_memalign(kDataAlignment, kDstSize));
+ dst_c_ = reinterpret_cast<uint8_t*>(vpx_memalign(kDataAlignment, kDstSize));
+ }
+
+ static void TearDownTestCase() {
+ vpx_free(src_);
+ src_ = NULL;
+ vpx_free(dst_);
+ dst_ = NULL;
+ vpx_free(dst_c_);
+ dst_c_ = NULL;
+ }
+
+ protected:
+ // Make test arrays big enough for 16x16 functions. Six-tap filters
+ // need 5 extra pixels outside of the macroblock.
+ static const int kSrcStride = 21;
+ static const int kDstStride = 16;
+ static const int kDataAlignment = 16;
+ static const int kSrcSize = kSrcStride * kSrcStride + 1;
+ static const int kDstSize = kDstStride * kDstStride;
+
+ virtual void SetUp() {
+ width_ = GET_PARAM(0);
+ height_ = GET_PARAM(1);
+ sixtap_predict_ = GET_PARAM(2);
+ memset(src_, 0, sizeof(src_));
+ memset(dst_, 0, sizeof(dst_));
+ memset(dst_c_, 0, sizeof(dst_c_));
+ }
+
+ int width_;
+ int height_;
+ sixtap_predict_fn_t sixtap_predict_;
+ // The src stores the macroblock we will filter on, and makes it 1 byte larger
+ // in order to test unaligned access. The result is stored in dst and dst_c(c
+ // reference code result).
+ static uint8_t* src_;
+ static uint8_t* dst_;
+ static uint8_t* dst_c_;
+};
+
+uint8_t* SixtapPredictTest::src_ = NULL;
+uint8_t* SixtapPredictTest::dst_ = NULL;
+uint8_t* SixtapPredictTest::dst_c_ = NULL;
+
+TEST_P(SixtapPredictTest, TestWithPresetData) {
+ // Test input
+ static const uint8_t test_data[kSrcSize] = {
+ 216, 184, 4, 191, 82, 92, 41, 0, 1, 226, 236, 172, 20, 182, 42, 226, 177,
+ 79, 94, 77, 179, 203, 206, 198, 22, 192, 19, 75, 17, 192, 44, 233, 120,
+ 48, 168, 203, 141, 210, 203, 143, 180, 184, 59, 201, 110, 102, 171, 32,
+ 182, 10, 109, 105, 213, 60, 47, 236, 253, 67, 55, 14, 3, 99, 247, 124,
+ 148, 159, 71, 34, 114, 19, 177, 38, 203, 237, 239, 58, 83, 155, 91, 10,
+ 166, 201, 115, 124, 5, 163, 104, 2, 231, 160, 16, 234, 4, 8, 103, 153,
+ 167, 174, 187, 26, 193, 109, 64, 141, 90, 48, 200, 174, 204, 36, 184,
+ 114, 237, 43, 238, 242, 207, 86, 245, 182, 247, 6, 161, 251, 14, 8, 148,
+ 182, 182, 79, 208, 120, 188, 17, 6, 23, 65, 206, 197, 13, 242, 126, 128,
+ 224, 170, 110, 211, 121, 197, 200, 47, 188, 207, 208, 184, 221, 216, 76,
+ 148, 143, 156, 100, 8, 89, 117, 14, 112, 183, 221, 54, 197, 208, 180, 69,
+ 176, 94, 180, 131, 215, 121, 76, 7, 54, 28, 216, 238, 249, 176, 58, 142,
+ 64, 215, 242, 72, 49, 104, 87, 161, 32, 52, 216, 230, 4, 141, 44, 181,
+ 235, 224, 57, 195, 89, 134, 203, 144, 162, 163, 126, 156, 84, 185, 42,
+ 148, 145, 29, 221, 194, 134, 52, 100, 166, 105, 60, 140, 110, 201, 184,
+ 35, 181, 153, 93, 121, 243, 227, 68, 131, 134, 232, 2, 35, 60, 187, 77,
+ 209, 76, 106, 174, 15, 241, 227, 115, 151, 77, 175, 36, 187, 121, 221,
+ 223, 47, 118, 61, 168, 105, 32, 237, 236, 167, 213, 238, 202, 17, 170,
+ 24, 226, 247, 131, 145, 6, 116, 117, 121, 11, 194, 41, 48, 126, 162, 13,
+ 93, 209, 131, 154, 122, 237, 187, 103, 217, 99, 60, 200, 45, 78, 115, 69,
+ 49, 106, 200, 194, 112, 60, 56, 234, 72, 251, 19, 120, 121, 182, 134, 215,
+ 135, 10, 114, 2, 247, 46, 105, 209, 145, 165, 153, 191, 243, 12, 5, 36,
+ 119, 206, 231, 231, 11, 32, 209, 83, 27, 229, 204, 149, 155, 83, 109, 35,
+ 93, 223, 37, 84, 14, 142, 37, 160, 52, 191, 96, 40, 204, 101, 77, 67, 52,
+ 53, 43, 63, 85, 253, 147, 113, 226, 96, 6, 125, 179, 115, 161, 17, 83,
+ 198, 101, 98, 85, 139, 3, 137, 75, 99, 178, 23, 201, 255, 91, 253, 52,
+ 134, 60, 138, 131, 208, 251, 101, 48, 2, 227, 228, 118, 132, 245, 202,
+ 75, 91, 44, 160, 231, 47, 41, 50, 147, 220, 74, 92, 219, 165, 89, 16
+ };
+
+ // Expected result
+ static const uint8_t expected_dst[kDstSize] = {
+ 117, 102, 74, 135, 42, 98, 175, 206, 70, 73, 222, 197, 50, 24, 39, 49, 38,
+ 105, 90, 47, 169, 40, 171, 215, 200, 73, 109, 141, 53, 85, 177, 164, 79,
+ 208, 124, 89, 212, 18, 81, 145, 151, 164, 217, 153, 91, 154, 102, 102,
+ 159, 75, 164, 152, 136, 51, 213, 219, 186, 116, 193, 224, 186, 36, 231,
+ 208, 84, 211, 155, 167, 35, 59, 42, 76, 216, 149, 73, 201, 78, 149, 184,
+ 100, 96, 196, 189, 198, 188, 235, 195, 117, 129, 120, 129, 49, 25, 133,
+ 113, 69, 221, 114, 70, 143, 99, 157, 108, 189, 140, 78, 6, 55, 65, 240,
+ 255, 245, 184, 72, 90, 100, 116, 131, 39, 60, 234, 167, 33, 160, 88, 185,
+ 200, 157, 159, 176, 127, 151, 138, 102, 168, 106, 170, 86, 82, 219, 189,
+ 76, 33, 115, 197, 106, 96, 198, 136, 97, 141, 237, 151, 98, 137, 191,
+ 185, 2, 57, 95, 142, 91, 255, 185, 97, 137, 76, 162, 94, 173, 131, 193,
+ 161, 81, 106, 72, 135, 222, 234, 137, 66, 137, 106, 243, 210, 147, 95,
+ 15, 137, 110, 85, 66, 16, 96, 167, 147, 150, 173, 203, 140, 118, 196,
+ 84, 147, 160, 19, 95, 101, 123, 74, 132, 202, 82, 166, 12, 131, 166,
+ 189, 170, 159, 85, 79, 66, 57, 152, 132, 203, 194, 0, 1, 56, 146, 180,
+ 224, 156, 28, 83, 181, 79, 76, 80, 46, 160, 175, 59, 106, 43, 87, 75,
+ 136, 85, 189, 46, 71, 200, 90
+ };
+
+ uint8_t *src = const_cast<uint8_t*>(test_data);
+
+ sixtap_predict_(&src[kSrcStride * 2 + 2 + 1], kSrcStride,
+ 2, 2, dst_, kDstStride);
+
+ for (int i = 0; i < height_; ++i)
+ for (int j = 0; j < width_; ++j)
+ ASSERT_EQ(expected_dst[i * kDstStride + j], dst_[i * kDstStride + j])
+ << "i==" << (i * width_ + j);
+}
+
+using libvpx_test::ACMRandom;
+
+TEST_P(SixtapPredictTest, TestWithRandomData) {
+ ACMRandom rnd(ACMRandom::DeterministicSeed());
+ for (int i = 0; i < kSrcSize; ++i)
+ src_[i] = rnd.Rand8();
+
+ // Run tests for all possible offsets.
+ for (int xoffset = 0; xoffset < 8; ++xoffset) {
+ for (int yoffset = 0; yoffset < 8; ++yoffset) {
+ // Call c reference function.
+ // Move start point to next pixel to test if the function reads
+ // unaligned data correctly.
+ vp8_sixtap_predict16x16_c(&src_[kSrcStride * 2 + 2 + 1], kSrcStride,
+ xoffset, yoffset, dst_c_, kDstStride);
+
+ // Run test.
+ sixtap_predict_(&src_[kSrcStride * 2 + 2 + 1], kSrcStride,
+ xoffset, yoffset, dst_, kDstStride);
+
+ for (int i = 0; i < height_; ++i)
+ for (int j = 0; j < width_; ++j)
+ ASSERT_EQ(dst_c_[i * kDstStride + j], dst_[i * kDstStride + j])
+ << "i==" << (i * width_ + j);
+ }
+ }
+}
+
+using std::tr1::make_tuple;
+
+const sixtap_predict_fn_t sixtap_16x16_c = vp8_sixtap_predict16x16_c;
+const sixtap_predict_fn_t sixtap_8x8_c = vp8_sixtap_predict8x8_c;
+const sixtap_predict_fn_t sixtap_8x4_c = vp8_sixtap_predict8x4_c;
+const sixtap_predict_fn_t sixtap_4x4_c = vp8_sixtap_predict4x4_c;
+INSTANTIATE_TEST_CASE_P(
+ C, SixtapPredictTest, ::testing::Values(
+ make_tuple(16, 16, sixtap_16x16_c),
+ make_tuple(8, 8, sixtap_8x8_c),
+ make_tuple(8, 4, sixtap_8x4_c),
+ make_tuple(4, 4, sixtap_4x4_c)));
+#if HAVE_MMX
+const sixtap_predict_fn_t sixtap_16x16_mmx = vp8_sixtap_predict16x16_mmx;
+const sixtap_predict_fn_t sixtap_8x8_mmx = vp8_sixtap_predict8x8_mmx;
+const sixtap_predict_fn_t sixtap_8x4_mmx = vp8_sixtap_predict8x4_mmx;
+const sixtap_predict_fn_t sixtap_4x4_mmx = vp8_sixtap_predict4x4_mmx;
+INSTANTIATE_TEST_CASE_P(
+ MMX, SixtapPredictTest, ::testing::Values(
+ make_tuple(16, 16, sixtap_16x16_mmx),
+ make_tuple(8, 8, sixtap_8x8_mmx),
+ make_tuple(8, 4, sixtap_8x4_mmx),
+ make_tuple(4, 4, sixtap_4x4_mmx)));
+#endif
+#if HAVE_SSE2
+const sixtap_predict_fn_t sixtap_16x16_sse2 = vp8_sixtap_predict16x16_sse2;
+const sixtap_predict_fn_t sixtap_8x8_sse2 = vp8_sixtap_predict8x8_sse2;
+const sixtap_predict_fn_t sixtap_8x4_sse2 = vp8_sixtap_predict8x4_sse2;
+INSTANTIATE_TEST_CASE_P(
+ SSE2, SixtapPredictTest, ::testing::Values(
+ make_tuple(16, 16, sixtap_16x16_sse2),
+ make_tuple(8, 8, sixtap_8x8_sse2),
+ make_tuple(8, 4, sixtap_8x4_sse2)));
+#endif
+#if HAVE_SSSE3
+const sixtap_predict_fn_t sixtap_16x16_ssse3 = vp8_sixtap_predict16x16_ssse3;
+const sixtap_predict_fn_t sixtap_8x8_ssse3 = vp8_sixtap_predict8x8_ssse3;
+const sixtap_predict_fn_t sixtap_8x4_ssse3 = vp8_sixtap_predict8x4_ssse3;
+const sixtap_predict_fn_t sixtap_4x4_ssse3 = vp8_sixtap_predict4x4_ssse3;
+INSTANTIATE_TEST_CASE_P(
+ SSSE3, SixtapPredictTest, ::testing::Values(
+ make_tuple(16, 16, sixtap_16x16_ssse3),
+ make_tuple(8, 8, sixtap_8x8_ssse3),
+ make_tuple(8, 4, sixtap_8x4_ssse3),
+ make_tuple(4, 4, sixtap_4x4_ssse3)));
+#endif
+} // namespace
diff --git a/libvpx/test/subtract_test.cc b/libvpx/test/subtract_test.cc
new file mode 100644
index 0000000..99363de
--- /dev/null
+++ b/libvpx/test/subtract_test.cc
@@ -0,0 +1,113 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/acm_random.h"
+extern "C" {
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/common/blockd.h"
+#include "vp8/encoder/block.h"
+#include "vpx_mem/vpx_mem.h"
+}
+
+typedef void (*subtract_b_fn_t)(BLOCK *be, BLOCKD *bd, int pitch);
+
+namespace {
+
+class SubtractBlockTest : public ::testing::TestWithParam<subtract_b_fn_t> {};
+
+using libvpx_test::ACMRandom;
+
+TEST_P(SubtractBlockTest, SimpleSubtract) {
+ ACMRandom rnd(ACMRandom::DeterministicSeed());
+ BLOCK be;
+ BLOCKD bd;
+ // in libvpx, this stride is always 16
+ const int kDiffPredStride = 16;
+ const int kSrcStride[] = {32, 16, 8, 4, 0};
+ const int kBlockWidth = 4;
+ const int kBlockHeight = 4;
+
+ // Allocate... align to 16 for mmx/sse tests
+ uint8_t *source = reinterpret_cast<uint8_t*>(
+ vpx_memalign(16, kBlockHeight * kSrcStride[0] * sizeof(*source)));
+ be.src_diff = reinterpret_cast<int16_t*>(
+ vpx_memalign(16, kBlockHeight * kDiffPredStride * sizeof(*be.src_diff)));
+ bd.predictor = reinterpret_cast<unsigned char*>(
+ vpx_memalign(16, kBlockHeight * kDiffPredStride * sizeof(*bd.predictor)));
+
+ for(int i = 0; kSrcStride[i] > 0; ++i) {
+ // start at block0
+ be.src = 0;
+ be.base_src = &source;
+ be.src_stride = kSrcStride[i];
+
+ // set difference
+ int16_t *src_diff = be.src_diff;
+ for (int r = 0; r < kBlockHeight; ++r) {
+ for (int c = 0; c < kBlockWidth; ++c) {
+ src_diff[c] = 0xa5a5;
+ }
+ src_diff += kDiffPredStride;
+ }
+
+ // set destination
+ uint8_t *base_src = *be.base_src;
+ for (int r = 0; r < kBlockHeight; ++r) {
+ for (int c = 0; c < kBlockWidth; ++c) {
+ base_src[c] = rnd.Rand8();
+ }
+ base_src += be.src_stride;
+ }
+
+ // set predictor
+ uint8_t *predictor = bd.predictor;
+ for (int r = 0; r < kBlockHeight; ++r) {
+ for (int c = 0; c < kBlockWidth; ++c) {
+ predictor[c] = rnd.Rand8();
+ }
+ predictor += kDiffPredStride;
+ }
+
+ GetParam()(&be, &bd, kDiffPredStride);
+
+ base_src = *be.base_src;
+ src_diff = be.src_diff;
+ predictor = bd.predictor;
+ for (int r = 0; r < kBlockHeight; ++r) {
+ for (int c = 0; c < kBlockWidth; ++c) {
+ EXPECT_EQ(base_src[c], (src_diff[c] + predictor[c])) << "r = " << r
+ << ", c = " << c;
+ }
+ src_diff += kDiffPredStride;
+ predictor += kDiffPredStride;
+ base_src += be.src_stride;
+ }
+ }
+ vpx_free(be.src_diff);
+ vpx_free(source);
+ vpx_free(bd.predictor);
+}
+
+INSTANTIATE_TEST_CASE_P(C, SubtractBlockTest,
+ ::testing::Values(vp8_subtract_b_c));
+
+#if HAVE_MMX
+INSTANTIATE_TEST_CASE_P(MMX, SubtractBlockTest,
+ ::testing::Values(vp8_subtract_b_mmx));
+#endif
+
+#if HAVE_SSE2
+INSTANTIATE_TEST_CASE_P(SSE2, SubtractBlockTest,
+ ::testing::Values(vp8_subtract_b_sse2));
+#endif
+
+} // namespace
diff --git a/libvpx/test/test-data.sha1 b/libvpx/test/test-data.sha1
new file mode 100644
index 0000000..c1b6a83
--- /dev/null
+++ b/libvpx/test/test-data.sha1
@@ -0,0 +1,123 @@
+d5dfb0151c9051f8c85999255645d7a23916d3c0 hantro_collage_w352h288.yuv
+5184c46ddca8b1fadd16742e8500115bc8f749da vp80-00-comprehensive-001.ivf
+65bf1bbbced81b97bd030f376d1b7f61a224793f vp80-00-comprehensive-002.ivf
+906b4c1e99eb734504c504b3f1ad8052137ce672 vp80-00-comprehensive-003.ivf
+ec144b1af53af895db78355785650b96dd3f0ade vp80-00-comprehensive-004.ivf
+afc7091785c62f1c121c4554a2830c30704587d9 vp80-00-comprehensive-005.ivf
+42ea9d55c818145d06a9b633b8e85c6a6164fd3e vp80-00-comprehensive-006.ivf
+e5b3a73ab79fe024c14309d653d6bed92902ee3b vp80-00-comprehensive-007.ivf
+f3c50a58875930adfb84525c0ef59d7e4c08540c vp80-00-comprehensive-008.ivf
+4b2841fdb83db51ae322096ae468bbb9dc2c8362 vp80-00-comprehensive-009.ivf
+efbff736e3a91ab6a98c5bc2dce65d645944c7b1 vp80-00-comprehensive-010.ivf
+6b315102cae008d22a3d2c231be92cb704a222f8 vp80-00-comprehensive-011.ivf
+f3214a4fea14c2d5ec689936c1613f274c859ee8 vp80-00-comprehensive-012.ivf
+e4094e96d308c8a35b74c480a43d853c5294cd34 vp80-00-comprehensive-013.ivf
+5b0adfaf60a69e0aaf3ec021a39d0a68fc0e1b5a vp80-00-comprehensive-014.ivf
+e8467688ddf26b5000664f904faf0d70506aa653 vp80-00-comprehensive-015.ivf
+aab55582337dfd2a39ff54fb2576a91910d49337 vp80-00-comprehensive-016.ivf
+1ba24724f80203c9bae4f1d0f99d534721980016 vp80-00-comprehensive-017.ivf
+143a15512b46f436280ddb4d0e6411eb4af434f2 vp80-00-comprehensive-018.ivf
+c5baeaf5714fdfb3a8bc960a8e33ac438e83b16b vp80-01-intra-1400.ivf
+f383955229afe3408453e316d11553d923ca60d5 vp80-01-intra-1411.ivf
+84e1f4343f174c9f3c83f834bac3196fb325bf2c vp80-01-intra-1416.ivf
+fb6e712a47dd57a28a3727d2ae2c97a8b7c7ca51 vp80-01-intra-1417.ivf
+71ea772d3e9d315b8cbecf41207b8a237c34853b vp80-02-inter-1402.ivf
+d85dbc4271525dcd128c503f936fe69091d1f8d0 vp80-02-inter-1412.ivf
+d4e5d3ad56511867d025f93724d090f92ba6ec3d vp80-02-inter-1418.ivf
+91791cbcc37c60f35dbd8090bacb54e5ec6dd4fa vp80-02-inter-1424.ivf
+17fbfe2fea70f6e2f3fa6ca4efaae6c0b03b5f02 vp80-03-segmentation-01.ivf
+3c3600dbbcde08e20d54c66fe3b7eadd4f09bdbb vp80-03-segmentation-02.ivf
+c156778d5340967d4b369c490848076e92f1f875 vp80-03-segmentation-03.ivf
+d25dcff6c60e87a1af70945b8911b6b4998533b0 vp80-03-segmentation-04.ivf
+362baba2ce454c9db21218f35e81c27a5ed0b730 vp80-03-segmentation-1401.ivf
+d223ae7ee748ce07e74c4679bfd219e84aa9f4b0 vp80-03-segmentation-1403.ivf
+033adf7f3a13836a3f1cffcb87c1972900f2b5c6 vp80-03-segmentation-1407.ivf
+4d51dfbf9f3e2c590ec99d1d6f59dd731d04375f vp80-03-segmentation-1408.ivf
+f37a62b197c2600d75e0ccfbb31b60efdedac251 vp80-03-segmentation-1409.ivf
+eb25bd7bfba5b2f6935018a930f42d123b1e7fcd vp80-03-segmentation-1410.ivf
+b9d5c436663a30c27cfff84b53a002e501258843 vp80-03-segmentation-1413.ivf
+6da92b9d1a180cc3a8afe348ab12258f5a37be1a vp80-03-segmentation-1414.ivf
+a4f5842602886bd669f115f93d8a35c035cb0948 vp80-03-segmentation-1415.ivf
+f295dceb8ef278b77251b3f9df8aee22e161d547 vp80-03-segmentation-1425.ivf
+198dbf9f36f733200e432664cc8c5752d59779de vp80-03-segmentation-1426.ivf
+7704804e32f5de976803929934a7fafe101ac7b0 vp80-03-segmentation-1427.ivf
+831ccd862ea95ca025d2f3bd8b88678752f5416d vp80-03-segmentation-1432.ivf
+b3c11978529289f9109f2766fcaba3ebc40e11ef vp80-03-segmentation-1435.ivf
+a835a731f5520ebfc1002c40121264d0020559ac vp80-03-segmentation-1436.ivf
+1d1732942f773bb2a5775fcb9689b1579ce28eab vp80-03-segmentation-1437.ivf
+db04799adfe089dfdf74dbd43cc05ede7161f99e vp80-03-segmentation-1441.ivf
+7caf39b3f20cfd52b998210878062e52a5edf1e6 vp80-03-segmentation-1442.ivf
+3607f6bb4ee106c38fa1ea370dc4ff8b8cde2261 vp80-04-partitions-1404.ivf
+93cc323b6b6867f1b12dd48773424549c6960a6b vp80-04-partitions-1405.ivf
+047eedb14b865bdac8a3538e63801054e0295e9c vp80-04-partitions-1406.ivf
+0f1233bd2bc33f56ce5e495dbd455d122339f384 vp80-05-sharpness-1428.ivf
+51767fc136488a9535c2a4c38067c542ee2048df vp80-05-sharpness-1429.ivf
+9805aa107672de25d6fb8c35e20d06deca5efe18 vp80-05-sharpness-1430.ivf
+61db6b965f9c27aebe71b85bf2d5877e58e4bbdf vp80-05-sharpness-1431.ivf
+10420d266290d2923555f84af38eeb96edbd3ae8 vp80-05-sharpness-1433.ivf
+3ed24f9a80cddfdf75824ba95cdb4ff9286cb443 vp80-05-sharpness-1434.ivf
+c87599cbecd72d4cd4f7ace3313b7a6bc6eb8163 vp80-05-sharpness-1438.ivf
+aff51d865c2621b60510459244ea83e958e4baed vp80-05-sharpness-1439.ivf
+da386e72b19b5485a6af199c5eb60ef25e510dd1 vp80-05-sharpness-1440.ivf
+6759a095203d96ccd267ce09b1b050b8cc4c2f1f vp80-05-sharpness-1443.ivf
+db55ec7fd02c864ba996ff060b25b1e08611330b vp80-00-comprehensive-001.ivf.md5
+29db0ad011cba1e45f856d5623cd38dac3e3bf19 vp80-00-comprehensive-002.ivf.md5
+e84f258f69e173e7d68f8f8c037a0a3766902182 vp80-00-comprehensive-003.ivf.md5
+eb7912eaf69559a16fd82bc3f5fb1524cf4a4466 vp80-00-comprehensive-004.ivf.md5
+4206f71c94894bd5b5b376f6c09b3817dbc65206 vp80-00-comprehensive-005.ivf.md5
+4f89b356f6f2fecb928f330a10f804f00f5325f5 vp80-00-comprehensive-006.ivf.md5
+2813236a32964dd8007e17648bcf035a20fcda6c vp80-00-comprehensive-007.ivf.md5
+10746c72098f872803c900e17c5680e451f5f498 vp80-00-comprehensive-008.ivf.md5
+39a23d0692ce64421a7bb7cdf6ccec5928d37fff vp80-00-comprehensive-009.ivf.md5
+f6e3de8931a0cc659bda8fbc14050346955e72d4 vp80-00-comprehensive-010.ivf.md5
+101683ec195b6e944f7cd1e468fc8921439363e6 vp80-00-comprehensive-011.ivf.md5
+1f592751ce46d8688998fa0fa4fbdcda0fd4058c vp80-00-comprehensive-012.ivf.md5
+6066176f90ca790251e795fca1a5797d59999841 vp80-00-comprehensive-013.ivf.md5
+2656da94ba93691f23edc4d60b3a09e2be46c217 vp80-00-comprehensive-014.ivf.md5
+c6e0d5f5d61460c8ac8edfa4e701f10312c03133 vp80-00-comprehensive-015.ivf.md5
+ee60fee501d8493e34e8d6a1fe315b51ed09b24a vp80-00-comprehensive-016.ivf.md5
+9f1914ceffcad4546c0a29de3ef591d8bea304dc vp80-00-comprehensive-017.ivf.md5
+e0305178fe288a9fd8082b39e2d03181edb19054 vp80-00-comprehensive-018.ivf.md5
+612494da2fa799cc9d76dcdd835ae6c7cb2e5c05 vp80-01-intra-1400.ivf.md5
+48ea06097ac8269c5e8c2131d3d0639f431fcf0e vp80-01-intra-1411.ivf.md5
+6e2ab4e7677ad0ba868083ca6bc387ee922b400c vp80-01-intra-1416.ivf.md5
+eca0a90348959ce3854142f8d8641b13050e8349 vp80-01-intra-1417.ivf.md5
+920feea203145d5c2258a91c4e6991934a79a99e vp80-02-inter-1402.ivf.md5
+f71d97909fe2b3dd65be7e1f56c72237f0cef200 vp80-02-inter-1412.ivf.md5
+e911254569a30bbb2a237ff8b79f69ed9da0672d vp80-02-inter-1418.ivf.md5
+58c789c50c9bb9cc90580bed291164a0939d28ba vp80-02-inter-1424.ivf.md5
+ff3e2f441327b9c20a0b37c524e0f5a48a36de7b vp80-03-segmentation-01.ivf.md5
+0791f417f076a542ae66fbc3426ab4d94cbd6c75 vp80-03-segmentation-02.ivf.md5
+722e50f1a6a91c34302d68681faffc1c26d1cc57 vp80-03-segmentation-03.ivf.md5
+c701f1885bcfb27fb8e70cc65606b289172ef889 vp80-03-segmentation-04.ivf.md5
+f79bc9ec189a2b4807632a3d0c5bf04a178b5300 vp80-03-segmentation-1401.ivf.md5
+b9aa4c74c0219b639811c44760d0b24cd8bb436a vp80-03-segmentation-1403.ivf.md5
+70d5a2207ca1891bcaebd5cf6dd88ce8d57b4334 vp80-03-segmentation-1407.ivf.md5
+265f962ee781531f9a93b9309461316fd32b2a1d vp80-03-segmentation-1408.ivf.md5
+0c4ecbbd6dc042d30e626d951b65f460dd6cd563 vp80-03-segmentation-1409.ivf.md5
+cf779af36a937f06570a0fca9db64ba133451dee vp80-03-segmentation-1410.ivf.md5
+0e6c5036d51ab078842f133934926c598a9cff02 vp80-03-segmentation-1413.ivf.md5
+eb3930aaf229116c80d507516c34759c3f6cdf69 vp80-03-segmentation-1414.ivf.md5
+123d6c0f72ee87911c4ae7538e87b7d163b22d6c vp80-03-segmentation-1415.ivf.md5
+e70551d1a38920e097a5d8782390b79ecaeb7505 vp80-03-segmentation-1425.ivf.md5
+44e8f4117e46dbb302b2cfd81171cc1a1846e431 vp80-03-segmentation-1426.ivf.md5
+52636e54aee5f95bbace37021bd67de5db767e9a vp80-03-segmentation-1427.ivf.md5
+b1ad3eff20215c28e295b15ef3636ed926d59cba vp80-03-segmentation-1432.ivf.md5
+24c22a552fa28a90e5978f67f57181cc2d7546d7 vp80-03-segmentation-1435.ivf.md5
+96c49c390abfced18a7a8c9b9ea10af778e10edb vp80-03-segmentation-1436.ivf.md5
+f95eb6214571434f1f73ab7833b9ccdf47588020 vp80-03-segmentation-1437.ivf.md5
+1c0700ca27c9b0090a7747a4b0b4dc21d1843181 vp80-03-segmentation-1441.ivf.md5
+81d4f23ca32667ee958bae579c8f5e97ba72eb97 vp80-03-segmentation-1442.ivf.md5
+272efcef07a3a30fbca51bfd566063d8258ec0be vp80-04-partitions-1404.ivf.md5
+66ed219ab812ac801b256d35cf495d193d4cf478 vp80-04-partitions-1405.ivf.md5
+36083f37f56f502bd60ec5e07502ee9e6b8699b0 vp80-04-partitions-1406.ivf.md5
+6ca909bf168a64c09415626294665dc1be3d1973 vp80-05-sharpness-1428.ivf.md5
+1667d2ee2334e5fdea8a8a866f4ccf3cf76f033a vp80-05-sharpness-1429.ivf.md5
+71bcbe5357d36a19df5b07fbe3e27bffa8893f0a vp80-05-sharpness-1430.ivf.md5
+89a09b1dffce2d55770a89e58d9925c70ef79bf8 vp80-05-sharpness-1431.ivf.md5
+08444a18b4e6ba3450c0796dd728d48c399a2dc9 vp80-05-sharpness-1433.ivf.md5
+6d6223719a90c13e848aa2a8a6642098cdb5977a vp80-05-sharpness-1434.ivf.md5
+41d70bb5fa45bc88da1604a0af466930b8dd77b5 vp80-05-sharpness-1438.ivf.md5
+086c56378df81b6cee264d7540a7b8f2b405c7a4 vp80-05-sharpness-1439.ivf.md5
+d32dc2c4165eb266ea4c23c14a45459b363def32 vp80-05-sharpness-1440.ivf.md5
+8c69dc3d8e563f56ffab5ad1e400d9e689dd23df vp80-05-sharpness-1443.ivf.md5
\ No newline at end of file
diff --git a/libvpx/test/test.mk b/libvpx/test/test.mk
new file mode 100644
index 0000000..7a11a27
--- /dev/null
+++ b/libvpx/test/test.mk
@@ -0,0 +1,178 @@
+LIBVPX_TEST_SRCS-yes += acm_random.h
+LIBVPX_TEST_SRCS-yes += test.mk
+LIBVPX_TEST_SRCS-yes += test_libvpx.cc
+LIBVPX_TEST_SRCS-yes += util.h
+LIBVPX_TEST_SRCS-yes += video_source.h
+
+##
+## BLACK BOX TESTS
+##
+## Black box tests only use the public API.
+##
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += altref_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += config_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += cq_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += datarate_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += encode_test_driver.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += encode_test_driver.h
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += error_resilience_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += i420_video_source.h
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += keyframe_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += resize_test.cc
+
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_DECODER) += ../md5_utils.h ../md5_utils.c
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_DECODER) += decode_test_driver.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_DECODER) += decode_test_driver.h
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_DECODER) += ivf_video_source.h
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_DECODER) += test_vector_test.cc
+##
+## WHITE BOX TESTS
+##
+## Whitebox tests invoke functions not exposed via the public API. Certain
+## shared library builds don't make these functions accessible.
+##
+ifeq ($(CONFIG_SHARED),)
+
+# These tests require both the encoder and decoder to be built.
+ifeq ($(CONFIG_VP8_ENCODER)$(CONFIG_VP8_DECODER),yesyes)
+LIBVPX_TEST_SRCS-yes += boolcoder_test.cc
+endif
+
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += fdct4x4_test.cc
+LIBVPX_TEST_SRCS-yes += idctllm_test.cc
+LIBVPX_TEST_SRCS-yes += intrapred_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_POSTPROC) += pp_filter_test.cc
+LIBVPX_TEST_SRCS-yes += sad_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += set_roi.cc
+LIBVPX_TEST_SRCS-yes += sixtap_predict_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += subtract_test.cc
+
+endif
+
+
+##
+## TEST DATA
+##
+LIBVPX_TEST_DATA-$(CONFIG_VP8_ENCODER) += hantro_collage_w352h288.yuv
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-001.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-002.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-003.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-004.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-005.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-006.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-007.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-008.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-009.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-010.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-011.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-012.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-013.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-014.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-015.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-016.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-017.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-018.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-01-intra-1400.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-01-intra-1411.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-01-intra-1416.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-01-intra-1417.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-02-inter-1402.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-02-inter-1412.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-02-inter-1418.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-02-inter-1424.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-01.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-02.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-03.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-04.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1401.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1403.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1407.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1408.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1409.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1410.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1413.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1414.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1415.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1425.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1426.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1427.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1432.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1435.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1436.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1437.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1441.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1442.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-04-partitions-1404.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-04-partitions-1405.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-04-partitions-1406.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1428.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1429.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1430.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1431.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1433.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1434.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1438.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1439.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1440.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1443.ivf
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-001.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-002.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-003.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-004.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-005.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-006.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-007.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-008.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-009.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-010.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-011.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-012.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-013.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-014.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-015.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-016.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-017.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-00-comprehensive-018.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-01-intra-1400.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-01-intra-1411.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-01-intra-1416.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-01-intra-1417.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-02-inter-1402.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-02-inter-1412.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-02-inter-1418.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-02-inter-1424.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1401.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1403.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1407.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1408.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1409.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1410.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1413.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1414.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1415.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1425.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1426.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1427.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1432.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1435.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1436.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1437.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1441.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-1442.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-01.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-02.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-03.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-03-segmentation-04.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-04-partitions-1404.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-04-partitions-1405.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-04-partitions-1406.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1428.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1429.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1430.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1431.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1433.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1434.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1438.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1439.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1440.ivf.md5
+LIBVPX_TEST_DATA-$(CONFIG_VP8_DECODER) += vp80-05-sharpness-1443.ivf.md5
diff --git a/libvpx/test/test_libvpx.cc b/libvpx/test/test_libvpx.cc
new file mode 100644
index 0000000..cfd5d28
--- /dev/null
+++ b/libvpx/test/test_libvpx.cc
@@ -0,0 +1,45 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include <string>
+#include "vpx_config.h"
+#if ARCH_X86 || ARCH_X86_64
+extern "C" {
+#include "vpx_ports/x86.h"
+}
+#endif
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+static void append_gtest_filter(const char *str) {
+ std::string filter = ::testing::FLAGS_gtest_filter;
+ filter += str;
+ ::testing::FLAGS_gtest_filter = filter;
+}
+
+int main(int argc, char **argv) {
+ ::testing::InitGoogleTest(&argc, argv);
+
+#if ARCH_X86 || ARCH_X86_64
+ const int simd_caps = x86_simd_caps();
+ if(!(simd_caps & HAS_MMX))
+ append_gtest_filter(":-MMX/*");
+ if(!(simd_caps & HAS_SSE))
+ append_gtest_filter(":-SSE/*");
+ if(!(simd_caps & HAS_SSE2))
+ append_gtest_filter(":-SSE2/*");
+ if(!(simd_caps & HAS_SSE3))
+ append_gtest_filter(":-SSE3/*");
+ if(!(simd_caps & HAS_SSSE3))
+ append_gtest_filter(":-SSSE3/*");
+ if(!(simd_caps & HAS_SSE4_1))
+ append_gtest_filter(":-SSE4_1/*");
+#endif
+
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/test/test_vector_test.cc b/libvpx/test/test_vector_test.cc
new file mode 100644
index 0000000..938457b
--- /dev/null
+++ b/libvpx/test/test_vector_test.cc
@@ -0,0 +1,144 @@
+/*
+ Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+
+ Use of this source code is governed by a BSD-style license
+ that can be found in the LICENSE file in the root of the source
+ tree. An additional intellectual property rights grant can be found
+ in the file PATENTS. All contributing project authors may
+ be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include <cstdio>
+#include <cstdlib>
+#include <string>
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/decode_test_driver.h"
+#include "test/ivf_video_source.h"
+extern "C" {
+#include "./md5_utils.h"
+#include "vpx_mem/vpx_mem.h"
+}
+
+#if defined(_MSC_VER)
+#define snprintf sprintf_s
+#endif
+
+namespace {
+// There are 61 test vectors in total.
+const char *kTestVectors[] = {
+ "vp80-00-comprehensive-001.ivf",
+ "vp80-00-comprehensive-002.ivf", "vp80-00-comprehensive-003.ivf",
+ "vp80-00-comprehensive-004.ivf", "vp80-00-comprehensive-005.ivf",
+ "vp80-00-comprehensive-006.ivf", "vp80-00-comprehensive-007.ivf",
+ "vp80-00-comprehensive-008.ivf", "vp80-00-comprehensive-009.ivf",
+ "vp80-00-comprehensive-010.ivf", "vp80-00-comprehensive-011.ivf",
+ "vp80-00-comprehensive-012.ivf", "vp80-00-comprehensive-013.ivf",
+ "vp80-00-comprehensive-014.ivf", "vp80-00-comprehensive-015.ivf",
+ "vp80-00-comprehensive-016.ivf", "vp80-00-comprehensive-017.ivf",
+ "vp80-00-comprehensive-018.ivf", "vp80-01-intra-1400.ivf",
+ "vp80-01-intra-1411.ivf", "vp80-01-intra-1416.ivf",
+ "vp80-01-intra-1417.ivf", "vp80-02-inter-1402.ivf",
+ "vp80-02-inter-1412.ivf", "vp80-02-inter-1418.ivf",
+ "vp80-02-inter-1424.ivf", "vp80-03-segmentation-01.ivf",
+ "vp80-03-segmentation-02.ivf", "vp80-03-segmentation-03.ivf",
+ "vp80-03-segmentation-04.ivf", "vp80-03-segmentation-1401.ivf",
+ "vp80-03-segmentation-1403.ivf", "vp80-03-segmentation-1407.ivf",
+ "vp80-03-segmentation-1408.ivf", "vp80-03-segmentation-1409.ivf",
+ "vp80-03-segmentation-1410.ivf", "vp80-03-segmentation-1413.ivf",
+ "vp80-03-segmentation-1414.ivf", "vp80-03-segmentation-1415.ivf",
+ "vp80-03-segmentation-1425.ivf", "vp80-03-segmentation-1426.ivf",
+ "vp80-03-segmentation-1427.ivf", "vp80-03-segmentation-1432.ivf",
+ "vp80-03-segmentation-1435.ivf", "vp80-03-segmentation-1436.ivf",
+ "vp80-03-segmentation-1437.ivf", "vp80-03-segmentation-1441.ivf",
+ "vp80-03-segmentation-1442.ivf", "vp80-04-partitions-1404.ivf",
+ "vp80-04-partitions-1405.ivf", "vp80-04-partitions-1406.ivf",
+ "vp80-05-sharpness-1428.ivf", "vp80-05-sharpness-1429.ivf",
+ "vp80-05-sharpness-1430.ivf", "vp80-05-sharpness-1431.ivf",
+ "vp80-05-sharpness-1433.ivf", "vp80-05-sharpness-1434.ivf",
+ "vp80-05-sharpness-1438.ivf", "vp80-05-sharpness-1439.ivf",
+ "vp80-05-sharpness-1440.ivf", "vp80-05-sharpness-1443.ivf"
+};
+
+class TestVectorTest : public libvpx_test::DecoderTest,
+ public ::testing::TestWithParam<const char*> {
+ protected:
+ TestVectorTest() : md5_file_(NULL) {}
+
+ virtual ~TestVectorTest() {
+ if (md5_file_)
+ fclose(md5_file_);
+ }
+
+ void OpenMD5File(const std::string& md5_file_name_) {
+ md5_file_ = libvpx_test::OpenTestDataFile(md5_file_name_);
+ ASSERT_TRUE(md5_file_) << "Md5 file open failed. Filename: "
+ << md5_file_name_;
+ }
+
+ virtual void DecompressedFrameHook(const vpx_image_t& img,
+ const unsigned int frame_number) {
+ char expected_md5[33];
+ char junk[128];
+
+ // Read correct md5 checksums.
+ const int res = fscanf(md5_file_, "%s %s", expected_md5, junk);
+ ASSERT_NE(res, EOF) << "Read md5 data failed";
+ expected_md5[32] = '\0';
+
+ MD5Context md5;
+ MD5Init(&md5);
+
+ // Compute and update md5 for each raw in decompressed data.
+ for (int plane = 0; plane < 3; ++plane) {
+ uint8_t *buf = img.planes[plane];
+
+ for (unsigned int y = 0; y < (plane ? (img.d_h + 1) >> 1 : img.d_h);
+ ++y) {
+ MD5Update(&md5, buf, (plane ? (img.d_w + 1) >> 1 : img.d_w));
+ buf += img.stride[plane];
+ }
+ }
+
+ uint8_t md5_sum[16];
+ MD5Final(md5_sum, &md5);
+
+ char actual_md5[33];
+ // Convert to get the actual md5.
+ for (int i = 0; i < 16; i++) {
+ snprintf(&actual_md5[i * 2], sizeof(actual_md5) - i * 2, "%02x",
+ md5_sum[i]);
+ }
+ actual_md5[32] = '\0';
+
+ // Check md5 match.
+ ASSERT_STREQ(expected_md5, actual_md5)
+ << "Md5 checksums don't match: frame number = " << frame_number;
+ }
+
+ private:
+ FILE *md5_file_;
+};
+
+// This test runs through the whole set of test vectors, and decodes them.
+// The md5 checksums are computed for each frame in the video file. If md5
+// checksums match the correct md5 data, then the test is passed. Otherwise,
+// the test failed.
+TEST_P(TestVectorTest, MD5Match) {
+ const std::string filename = GetParam();
+ // Open compressed video file.
+ libvpx_test::IVFVideoSource video(filename);
+
+ video.Init();
+
+ // Construct md5 file name.
+ const std::string md5_filename = filename + ".md5";
+ OpenMD5File(md5_filename);
+
+ // Decode frame, and check the md5 matching.
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+}
+
+INSTANTIATE_TEST_CASE_P(TestVectorSequence, TestVectorTest,
+ ::testing::ValuesIn(kTestVectors));
+
+} // namespace
diff --git a/libvpx/test/util.h b/libvpx/test/util.h
new file mode 100644
index 0000000..06a70cc
--- /dev/null
+++ b/libvpx/test/util.h
@@ -0,0 +1,18 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef TEST_UTIL_H_
+#define TEST_UTIL_H_
+
+// Macros
+#define PARAMS(...) ::testing::TestWithParam< std::tr1::tuple< __VA_ARGS__ > >
+#define GET_PARAM(k) std::tr1::get< k >(GetParam())
+
+#endif // TEST_UTIL_H_
diff --git a/libvpx/test/video_source.h b/libvpx/test/video_source.h
new file mode 100644
index 0000000..9772657
--- /dev/null
+++ b/libvpx/test/video_source.h
@@ -0,0 +1,175 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#ifndef TEST_VIDEO_SOURCE_H_
+#define TEST_VIDEO_SOURCE_H_
+
+#include <cstdio>
+#include <cstdlib>
+#include <string>
+#include "test/acm_random.h"
+#include "vpx/vpx_encoder.h"
+
+namespace libvpx_test {
+
+static FILE *OpenTestDataFile(const std::string& file_name) {
+ std::string path_to_source = file_name;
+ const char *kDataPath = getenv("LIBVPX_TEST_DATA_PATH");
+
+ if (kDataPath) {
+ path_to_source = kDataPath;
+ path_to_source += "/";
+ path_to_source += file_name;
+ }
+
+ return fopen(path_to_source.c_str(), "rb");
+}
+
+// Abstract base class for test video sources, which provide a stream of
+// vpx_image_t images with associated timestamps and duration.
+class VideoSource {
+ public:
+ virtual ~VideoSource() {}
+
+ // Prepare the stream for reading, rewind/open as necessary.
+ virtual void Begin() = 0;
+
+ // Advance the cursor to the next frame
+ virtual void Next() = 0;
+
+ // Get the current video frame, or NULL on End-Of-Stream.
+ virtual vpx_image_t *img() const = 0;
+
+ // Get the presentation timestamp of the current frame.
+ virtual vpx_codec_pts_t pts() const = 0;
+
+ // Get the current frame's duration
+ virtual unsigned long duration() const = 0;
+
+ // Get the timebase for the stream
+ virtual vpx_rational_t timebase() const = 0;
+
+ // Get the current frame counter, starting at 0.
+ virtual unsigned int frame() const = 0;
+
+ // Get the current file limit.
+ virtual unsigned int limit() const = 0;
+};
+
+
+class DummyVideoSource : public VideoSource {
+ public:
+ DummyVideoSource() : img_(NULL), limit_(100), width_(0), height_(0) {
+ SetSize(80, 64);
+ }
+
+ virtual ~DummyVideoSource() { vpx_img_free(img_); }
+
+ virtual void Begin() {
+ frame_ = 0;
+ FillFrame();
+ }
+
+ virtual void Next() {
+ ++frame_;
+ FillFrame();
+ }
+
+ virtual vpx_image_t *img() const {
+ return (frame_ < limit_) ? img_ : NULL;
+ }
+
+ // Models a stream where Timebase = 1/FPS, so pts == frame.
+ virtual vpx_codec_pts_t pts() const { return frame_; }
+
+ virtual unsigned long duration() const { return 1; }
+
+ virtual vpx_rational_t timebase() const {
+ const vpx_rational_t t = {1, 30};
+ return t;
+ }
+
+ virtual unsigned int frame() const { return frame_; }
+
+ virtual unsigned int limit() const { return limit_; }
+
+ void SetSize(unsigned int width, unsigned int height) {
+ if (width != width_ || height != height_) {
+ vpx_img_free(img_);
+ raw_sz_ = ((width + 31)&~31) * height * 3 / 2;
+ img_ = vpx_img_alloc(NULL, VPX_IMG_FMT_VPXI420, width, height, 32);
+ width_ = width;
+ height_ = height;
+ }
+ }
+
+ protected:
+ virtual void FillFrame() { memset(img_->img_data, 0, raw_sz_); }
+
+ vpx_image_t *img_;
+ size_t raw_sz_;
+ unsigned int limit_;
+ unsigned int frame_;
+ unsigned int width_;
+ unsigned int height_;
+};
+
+
+class RandomVideoSource : public DummyVideoSource {
+ public:
+ RandomVideoSource(int seed = ACMRandom::DeterministicSeed())
+ : rnd_(seed),
+ seed_(seed) { }
+
+ protected:
+ // Reset the RNG to get a matching stream for the second pass
+ virtual void Begin() {
+ frame_ = 0;
+ rnd_.Reset(seed_);
+ FillFrame();
+ }
+
+ // 15 frames of noise, followed by 15 static frames. Reset to 0 rather
+ // than holding previous frames to encourage keyframes to be thrown.
+ virtual void FillFrame() {
+ if (frame_ % 30 < 15)
+ for (size_t i = 0; i < raw_sz_; ++i)
+ img_->img_data[i] = rnd_.Rand8();
+ else
+ memset(img_->img_data, 0, raw_sz_);
+ }
+
+ ACMRandom rnd_;
+ int seed_;
+};
+
+// Abstract base class for test video sources, which provide a stream of
+// decompressed images to the decoder.
+class CompressedVideoSource {
+ public:
+ virtual ~CompressedVideoSource() {}
+
+ virtual void Init() = 0;
+
+ // Prepare the stream for reading, rewind/open as necessary.
+ virtual void Begin() = 0;
+
+ // Advance the cursor to the next frame
+ virtual void Next() = 0;
+
+ virtual const uint8_t *cxdata() const = 0;
+
+ virtual const unsigned int frame_size() const = 0;
+
+ virtual const unsigned int frame_number() const = 0;
+};
+
+} // namespace libvpx_test
+
+#endif // TEST_VIDEO_SOURCE_H_
diff --git a/libvpx/third_party/googletest/README.webm b/libvpx/third_party/googletest/README.webm
new file mode 100644
index 0000000..571d890
--- /dev/null
+++ b/libvpx/third_party/googletest/README.webm
@@ -0,0 +1,15 @@
+URL: http://code.google.com/p/googletest/
+Version: 1.6.0
+License: BSD
+License File: COPYING
+
+Description:
+Google's framework for writing C++ tests on a variety of platforms
+(Linux, Mac OS X, Windows, Windows CE, Symbian, etc). Based on the
+xUnit architecture. Supports automatic test discovery, a rich set of
+assertions, user-defined assertions, death tests, fatal and non-fatal
+failures, various options for running the tests, and XML test report
+generation.
+
+Local Modifications:
+None.
\ No newline at end of file
diff --git a/libvpx/third_party/googletest/gtest.mk b/libvpx/third_party/googletest/gtest.mk
new file mode 100644
index 0000000..0de3113
--- /dev/null
+++ b/libvpx/third_party/googletest/gtest.mk
@@ -0,0 +1 @@
+GTEST_SRCS-yes += src/gtest-all.cc
diff --git a/libvpx/third_party/googletest/src/CHANGES b/libvpx/third_party/googletest/src/CHANGES
new file mode 100644
index 0000000..5919245
--- /dev/null
+++ b/libvpx/third_party/googletest/src/CHANGES
@@ -0,0 +1,130 @@
+Changes for 1.6.0:
+
+* New feature: ADD_FAILURE_AT() for reporting a test failure at the
+ given source location -- useful for writing testing utilities.
+* New feature: the universal value printer is moved from Google Mock
+ to Google Test.
+* New feature: type parameters and value parameters are reported in
+ the XML report now.
+* A gtest_disable_pthreads CMake option.
+* Colored output works in GNU Screen sessions now.
+* Parameters of value-parameterized tests are now printed in the
+ textual output.
+* Failures from ad hoc test assertions run before RUN_ALL_TESTS() are
+ now correctly reported.
+* Arguments of ASSERT_XY and EXPECT_XY no longer need to support << to
+ ostream.
+* More complete handling of exceptions.
+* GTEST_ASSERT_XY can be used instead of ASSERT_XY in case the latter
+ name is already used by another library.
+* --gtest_catch_exceptions is now true by default, allowing a test
+ program to continue after an exception is thrown.
+* Value-parameterized test fixtures can now derive from Test and
+ WithParamInterface<T> separately, easing conversion of legacy tests.
+* Death test messages are clearly marked to make them more
+ distinguishable from other messages.
+* Compatibility fixes for Android, Google Native Client, MinGW, HP UX,
+ PowerPC, Lucid autotools, libCStd, Sun C++, Borland C++ Builder (Code Gear),
+ IBM XL C++ (Visual Age C++), and C++0x.
+* Bug fixes and implementation clean-ups.
+* Potentially incompatible changes: disables the harmful 'make install'
+ command in autotools.
+
+Changes for 1.5.0:
+
+ * New feature: assertions can be safely called in multiple threads
+ where the pthreads library is available.
+ * New feature: predicates used inside EXPECT_TRUE() and friends
+ can now generate custom failure messages.
+ * New feature: Google Test can now be compiled as a DLL.
+ * New feature: fused source files are included.
+ * New feature: prints help when encountering unrecognized Google Test flags.
+ * Experimental feature: CMake build script (requires CMake 2.6.4+).
+ * Experimental feature: the Pump script for meta programming.
+ * double values streamed to an assertion are printed with enough precision
+ to differentiate any two different values.
+ * Google Test now works on Solaris and AIX.
+ * Build and test script improvements.
+ * Bug fixes and implementation clean-ups.
+
+ Potentially breaking changes:
+
+ * Stopped supporting VC++ 7.1 with exceptions disabled.
+ * Dropped support for 'make install'.
+
+Changes for 1.4.0:
+
+ * New feature: the event listener API
+ * New feature: test shuffling
+ * New feature: the XML report format is closer to junitreport and can
+ be parsed by Hudson now.
+ * New feature: when a test runs under Visual Studio, its failures are
+ integrated in the IDE.
+ * New feature: /MD(d) versions of VC++ projects.
+ * New feature: elapsed time for the tests is printed by default.
+ * New feature: comes with a TR1 tuple implementation such that Boost
+ is no longer needed for Combine().
+ * New feature: EXPECT_DEATH_IF_SUPPORTED macro and friends.
+ * New feature: the Xcode project can now produce static gtest
+ libraries in addition to a framework.
+ * Compatibility fixes for Solaris, Cygwin, minGW, Windows Mobile,
+ Symbian, gcc, and C++Builder.
+ * Bug fixes and implementation clean-ups.
+
+Changes for 1.3.0:
+
+ * New feature: death tests on Windows, Cygwin, and Mac.
+ * New feature: ability to use Google Test assertions in other testing
+ frameworks.
+ * New feature: ability to run disabled test via
+ --gtest_also_run_disabled_tests.
+ * New feature: the --help flag for printing the usage.
+ * New feature: access to Google Test flag values in user code.
+ * New feature: a script that packs Google Test into one .h and one
+ .cc file for easy deployment.
+ * New feature: support for distributing test functions to multiple
+ machines (requires support from the test runner).
+ * Bug fixes and implementation clean-ups.
+
+Changes for 1.2.1:
+
+ * Compatibility fixes for Linux IA-64 and IBM z/OS.
+ * Added support for using Boost and other TR1 implementations.
+ * Changes to the build scripts to support upcoming release of Google C++
+ Mocking Framework.
+ * Added Makefile to the distribution package.
+ * Improved build instructions in README.
+
+Changes for 1.2.0:
+
+ * New feature: value-parameterized tests.
+ * New feature: the ASSERT/EXPECT_(NON)FATAL_FAILURE(_ON_ALL_THREADS)
+ macros.
+ * Changed the XML report format to match JUnit/Ant's.
+ * Added tests to the Xcode project.
+ * Added scons/SConscript for building with SCons.
+ * Added src/gtest-all.cc for building Google Test from a single file.
+ * Fixed compatibility with Solaris and z/OS.
+ * Enabled running Python tests on systems with python 2.3 installed,
+ e.g. Mac OS X 10.4.
+ * Bug fixes.
+
+Changes for 1.1.0:
+
+ * New feature: type-parameterized tests.
+ * New feature: exception assertions.
+ * New feature: printing elapsed time of tests.
+ * Improved the robustness of death tests.
+ * Added an Xcode project and samples.
+ * Adjusted the output format on Windows to be understandable by Visual Studio.
+ * Minor bug fixes.
+
+Changes for 1.0.1:
+
+ * Added project files for Visual Studio 7.1.
+ * Fixed issues with compiling on Mac OS X.
+ * Fixed issues with compiling on Cygwin.
+
+Changes for 1.0.0:
+
+ * Initial Open Source release of Google Test
diff --git a/libvpx/third_party/googletest/src/CMakeLists.txt b/libvpx/third_party/googletest/src/CMakeLists.txt
new file mode 100644
index 0000000..0fe2654
--- /dev/null
+++ b/libvpx/third_party/googletest/src/CMakeLists.txt
@@ -0,0 +1,240 @@
+########################################################################
+# CMake build script for Google Test.
+#
+# To run the tests for Google Test itself on Linux, use 'make test' or
+# ctest. You can select which tests to run using 'ctest -R regex'.
+# For more options, run 'ctest --help'.
+
+# BUILD_SHARED_LIBS is a standard CMake variable, but we declare it here to
+# make it prominent in the GUI.
+option(BUILD_SHARED_LIBS "Build shared libraries (DLLs)." OFF)
+
+# When other libraries are using a shared version of runtime libraries,
+# Google Test also has to use one.
+option(
+ gtest_force_shared_crt
+ "Use shared (DLL) run-time lib even when Google Test is built as static lib."
+ OFF)
+
+option(gtest_build_tests "Build all of gtest's own tests." OFF)
+
+option(gtest_build_samples "Build gtest's sample programs." OFF)
+
+option(gtest_disable_pthreads "Disable uses of pthreads in gtest." OFF)
+
+# Defines pre_project_set_up_hermetic_build() and set_up_hermetic_build().
+include(cmake/hermetic_build.cmake OPTIONAL)
+
+if (COMMAND pre_project_set_up_hermetic_build)
+ pre_project_set_up_hermetic_build()
+endif()
+
+########################################################################
+#
+# Project-wide settings
+
+# Name of the project.
+#
+# CMake files in this project can refer to the root source directory
+# as ${gtest_SOURCE_DIR} and to the root binary directory as
+# ${gtest_BINARY_DIR}.
+# Language "C" is required for find_package(Threads).
+project(gtest CXX C)
+cmake_minimum_required(VERSION 2.6.2)
+
+if (COMMAND set_up_hermetic_build)
+ set_up_hermetic_build()
+endif()
+
+# Define helper functions and macros used by Google Test.
+include(cmake/internal_utils.cmake)
+
+config_compiler_and_linker() # Defined in internal_utils.cmake.
+
+# Where Google Test's .h files can be found.
+include_directories(
+ ${gtest_SOURCE_DIR}/include
+ ${gtest_SOURCE_DIR})
+
+# Where Google Test's libraries can be found.
+link_directories(${gtest_BINARY_DIR}/src)
+
+########################################################################
+#
+# Defines the gtest & gtest_main libraries. User tests should link
+# with one of them.
+
+# Google Test libraries. We build them using more strict warnings than what
+# are used for other targets, to ensure that gtest can be compiled by a user
+# aggressive about warnings.
+cxx_library(gtest "${cxx_strict}" src/gtest-all.cc)
+cxx_library(gtest_main "${cxx_strict}" src/gtest_main.cc)
+target_link_libraries(gtest_main gtest)
+
+########################################################################
+#
+# Samples on how to link user tests with gtest or gtest_main.
+#
+# They are not built by default. To build them, set the
+# gtest_build_samples option to ON. You can do it by running ccmake
+# or specifying the -Dbuild_gtest_samples=ON flag when running cmake.
+
+if (gtest_build_samples)
+ cxx_executable(sample1_unittest samples gtest_main samples/sample1.cc)
+ cxx_executable(sample2_unittest samples gtest_main samples/sample2.cc)
+ cxx_executable(sample3_unittest samples gtest_main)
+ cxx_executable(sample4_unittest samples gtest_main samples/sample4.cc)
+ cxx_executable(sample5_unittest samples gtest_main samples/sample1.cc)
+ cxx_executable(sample6_unittest samples gtest_main)
+ cxx_executable(sample7_unittest samples gtest_main)
+ cxx_executable(sample8_unittest samples gtest_main)
+ cxx_executable(sample9_unittest samples gtest)
+ cxx_executable(sample10_unittest samples gtest)
+endif()
+
+########################################################################
+#
+# Google Test's own tests.
+#
+# You can skip this section if you aren't interested in testing
+# Google Test itself.
+#
+# The tests are not built by default. To build them, set the
+# gtest_build_tests option to ON. You can do it by running ccmake
+# or specifying the -Dgtest_build_tests=ON flag when running cmake.
+
+if (gtest_build_tests)
+ # This must be set in the root directory for the tests to be run by
+ # 'make test' or ctest.
+ enable_testing()
+
+ ############################################################
+ # C++ tests built with standard compiler flags.
+
+ cxx_test(gtest-death-test_test gtest_main)
+ cxx_test(gtest_environment_test gtest)
+ cxx_test(gtest-filepath_test gtest_main)
+ cxx_test(gtest-linked_ptr_test gtest_main)
+ cxx_test(gtest-listener_test gtest_main)
+ cxx_test(gtest_main_unittest gtest_main)
+ cxx_test(gtest-message_test gtest_main)
+ cxx_test(gtest_no_test_unittest gtest)
+ cxx_test(gtest-options_test gtest_main)
+ cxx_test(gtest-param-test_test gtest
+ test/gtest-param-test2_test.cc)
+ cxx_test(gtest-port_test gtest_main)
+ cxx_test(gtest_pred_impl_unittest gtest_main)
+ cxx_test(gtest-printers_test gtest_main)
+ cxx_test(gtest_prod_test gtest_main
+ test/production.cc)
+ cxx_test(gtest_repeat_test gtest)
+ cxx_test(gtest_sole_header_test gtest_main)
+ cxx_test(gtest_stress_test gtest)
+ cxx_test(gtest-test-part_test gtest_main)
+ cxx_test(gtest_throw_on_failure_ex_test gtest)
+ cxx_test(gtest-typed-test_test gtest_main
+ test/gtest-typed-test2_test.cc)
+ cxx_test(gtest_unittest gtest_main)
+ cxx_test(gtest-unittest-api_test gtest)
+
+ ############################################################
+ # C++ tests built with non-standard compiler flags.
+
+ cxx_library(gtest_no_exception "${cxx_no_exception}"
+ src/gtest-all.cc)
+ cxx_library(gtest_main_no_exception "${cxx_no_exception}"
+ src/gtest-all.cc src/gtest_main.cc)
+ cxx_library(gtest_main_no_rtti "${cxx_no_rtti}"
+ src/gtest-all.cc src/gtest_main.cc)
+
+ cxx_test_with_flags(gtest-death-test_ex_nocatch_test
+ "${cxx_exception} -DGTEST_ENABLE_CATCH_EXCEPTIONS_=0"
+ gtest test/gtest-death-test_ex_test.cc)
+ cxx_test_with_flags(gtest-death-test_ex_catch_test
+ "${cxx_exception} -DGTEST_ENABLE_CATCH_EXCEPTIONS_=1"
+ gtest test/gtest-death-test_ex_test.cc)
+
+ cxx_test_with_flags(gtest_no_rtti_unittest "${cxx_no_rtti}"
+ gtest_main_no_rtti test/gtest_unittest.cc)
+
+ cxx_shared_library(gtest_dll "${cxx_default}"
+ src/gtest-all.cc src/gtest_main.cc)
+
+ cxx_executable_with_flags(gtest_dll_test_ "${cxx_default}"
+ gtest_dll test/gtest_all_test.cc)
+ set_target_properties(gtest_dll_test_
+ PROPERTIES
+ COMPILE_DEFINITIONS "GTEST_LINKED_AS_SHARED_LIBRARY=1")
+
+ if (NOT MSVC OR NOT MSVC_VERSION EQUAL 1600)
+ # The C++ Standard specifies tuple_element<int, class>.
+ # Yet MSVC 10's <utility> declares tuple_element<size_t, class>.
+ # That declaration conflicts with our own standard-conforming
+ # tuple implementation. Therefore using our own tuple with
+ # MSVC 10 doesn't compile.
+ cxx_library(gtest_main_use_own_tuple "${cxx_use_own_tuple}"
+ src/gtest-all.cc src/gtest_main.cc)
+
+ cxx_test_with_flags(gtest-tuple_test "${cxx_use_own_tuple}"
+ gtest_main_use_own_tuple test/gtest-tuple_test.cc)
+
+ cxx_test_with_flags(gtest_use_own_tuple_test "${cxx_use_own_tuple}"
+ gtest_main_use_own_tuple
+ test/gtest-param-test_test.cc test/gtest-param-test2_test.cc)
+ endif()
+
+ ############################################################
+ # Python tests.
+
+ cxx_executable(gtest_break_on_failure_unittest_ test gtest)
+ py_test(gtest_break_on_failure_unittest)
+
+ cxx_executable_with_flags(
+ gtest_catch_exceptions_no_ex_test_
+ "${cxx_no_exception}"
+ gtest_main_no_exception
+ test/gtest_catch_exceptions_test_.cc)
+ cxx_executable_with_flags(
+ gtest_catch_exceptions_ex_test_
+ "${cxx_exception}"
+ gtest_main
+ test/gtest_catch_exceptions_test_.cc)
+ py_test(gtest_catch_exceptions_test)
+
+ cxx_executable(gtest_color_test_ test gtest)
+ py_test(gtest_color_test)
+
+ cxx_executable(gtest_env_var_test_ test gtest)
+ py_test(gtest_env_var_test)
+
+ cxx_executable(gtest_filter_unittest_ test gtest)
+ py_test(gtest_filter_unittest)
+
+ cxx_executable(gtest_help_test_ test gtest_main)
+ py_test(gtest_help_test)
+
+ cxx_executable(gtest_list_tests_unittest_ test gtest)
+ py_test(gtest_list_tests_unittest)
+
+ cxx_executable(gtest_output_test_ test gtest)
+ py_test(gtest_output_test)
+
+ cxx_executable(gtest_shuffle_test_ test gtest)
+ py_test(gtest_shuffle_test)
+
+ cxx_executable(gtest_throw_on_failure_test_ test gtest_no_exception)
+ set_target_properties(gtest_throw_on_failure_test_
+ PROPERTIES
+ COMPILE_FLAGS "${cxx_no_exception}")
+ py_test(gtest_throw_on_failure_test)
+
+ cxx_executable(gtest_uninitialized_test_ test gtest)
+ py_test(gtest_uninitialized_test)
+
+ cxx_executable(gtest_xml_outfile1_test_ test gtest_main)
+ cxx_executable(gtest_xml_outfile2_test_ test gtest_main)
+ py_test(gtest_xml_outfiles_test)
+
+ cxx_executable(gtest_xml_output_unittest_ test gtest)
+ py_test(gtest_xml_output_unittest)
+endif()
diff --git a/libvpx/third_party/googletest/src/CONTRIBUTORS b/libvpx/third_party/googletest/src/CONTRIBUTORS
new file mode 100644
index 0000000..feae2fc
--- /dev/null
+++ b/libvpx/third_party/googletest/src/CONTRIBUTORS
@@ -0,0 +1,37 @@
+# This file contains a list of people who've made non-trivial
+# contribution to the Google C++ Testing Framework project. People
+# who commit code to the project are encouraged to add their names
+# here. Please keep the list sorted by first names.
+
+Ajay Joshi <jaj@google.com>
+Balázs Dán <balazs.dan@gmail.com>
+Bharat Mediratta <bharat@menalto.com>
+Chandler Carruth <chandlerc@google.com>
+Chris Prince <cprince@google.com>
+Chris Taylor <taylorc@google.com>
+Dan Egnor <egnor@google.com>
+Eric Roman <eroman@chromium.org>
+Hady Zalek <hady.zalek@gmail.com>
+Jeffrey Yasskin <jyasskin@google.com>
+Jói Sigurðsson <joi@google.com>
+Keir Mierle <mierle@gmail.com>
+Keith Ray <keith.ray@gmail.com>
+Kenton Varda <kenton@google.com>
+Manuel Klimek <klimek@google.com>
+Markus Heule <markus.heule@gmail.com>
+Mika Raento <mikie@iki.fi>
+Miklós Fazekas <mfazekas@szemafor.com>
+Pasi Valminen <pasi.valminen@gmail.com>
+Patrick Hanna <phanna@google.com>
+Patrick Riley <pfr@google.com>
+Peter Kaminski <piotrk@google.com>
+Preston Jackson <preston.a.jackson@gmail.com>
+Rainer Klaffenboeck <rainer.klaffenboeck@dynatrace.com>
+Russ Cox <rsc@google.com>
+Russ Rufer <russ@pentad.com>
+Sean Mcafee <eefacm@gmail.com>
+Sigurður Ásgeirsson <siggi@google.com>
+Tracy Bialik <tracy@pentad.com>
+Vadim Berman <vadimb@google.com>
+Vlad Losev <vladl@google.com>
+Zhanyong Wan <wan@google.com>
diff --git a/libvpx/third_party/googletest/src/COPYING b/libvpx/third_party/googletest/src/COPYING
new file mode 100644
index 0000000..1941a11
--- /dev/null
+++ b/libvpx/third_party/googletest/src/COPYING
@@ -0,0 +1,28 @@
+Copyright 2008, Google Inc.
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are
+met:
+
+ * Redistributions of source code must retain the above copyright
+notice, this list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above
+copyright notice, this list of conditions and the following disclaimer
+in the documentation and/or other materials provided with the
+distribution.
+ * Neither the name of Google Inc. nor the names of its
+contributors may be used to endorse or promote products derived from
+this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
diff --git a/libvpx/third_party/googletest/src/Makefile.am b/libvpx/third_party/googletest/src/Makefile.am
new file mode 100644
index 0000000..cb350b7
--- /dev/null
+++ b/libvpx/third_party/googletest/src/Makefile.am
@@ -0,0 +1,302 @@
+# Automake file
+
+ACLOCAL_AMFLAGS = -I m4
+
+# Nonstandard package files for distribution
+EXTRA_DIST = \
+ CHANGES \
+ CONTRIBUTORS \
+ include/gtest/gtest-param-test.h.pump \
+ include/gtest/internal/gtest-param-util-generated.h.pump \
+ include/gtest/internal/gtest-tuple.h.pump \
+ include/gtest/internal/gtest-type-util.h.pump \
+ make/Makefile \
+ scripts/fuse_gtest_files.py \
+ scripts/gen_gtest_pred_impl.py \
+ scripts/pump.py \
+ scripts/test/Makefile
+
+# gtest source files that we don't compile directly. They are
+# #included by gtest-all.cc.
+GTEST_SRC = \
+ src/gtest-death-test.cc \
+ src/gtest-filepath.cc \
+ src/gtest-internal-inl.h \
+ src/gtest-port.cc \
+ src/gtest-printers.cc \
+ src/gtest-test-part.cc \
+ src/gtest-typed-test.cc \
+ src/gtest.cc
+
+EXTRA_DIST += $(GTEST_SRC)
+
+# Sample files that we don't compile.
+EXTRA_DIST += \
+ samples/prime_tables.h \
+ samples/sample2_unittest.cc \
+ samples/sample3_unittest.cc \
+ samples/sample4_unittest.cc \
+ samples/sample5_unittest.cc \
+ samples/sample6_unittest.cc \
+ samples/sample7_unittest.cc \
+ samples/sample8_unittest.cc \
+ samples/sample9_unittest.cc
+
+# C++ test files that we don't compile directly.
+EXTRA_DIST += \
+ test/gtest-death-test_ex_test.cc \
+ test/gtest-death-test_test.cc \
+ test/gtest-filepath_test.cc \
+ test/gtest-linked_ptr_test.cc \
+ test/gtest-listener_test.cc \
+ test/gtest-message_test.cc \
+ test/gtest-options_test.cc \
+ test/gtest-param-test2_test.cc \
+ test/gtest-param-test2_test.cc \
+ test/gtest-param-test_test.cc \
+ test/gtest-param-test_test.cc \
+ test/gtest-param-test_test.h \
+ test/gtest-port_test.cc \
+ test/gtest-printers_test.cc \
+ test/gtest-test-part_test.cc \
+ test/gtest-tuple_test.cc \
+ test/gtest-typed-test2_test.cc \
+ test/gtest-typed-test_test.cc \
+ test/gtest-typed-test_test.h \
+ test/gtest-unittest-api_test.cc \
+ test/gtest_break_on_failure_unittest_.cc \
+ test/gtest_catch_exceptions_test_.cc \
+ test/gtest_color_test_.cc \
+ test/gtest_env_var_test_.cc \
+ test/gtest_environment_test.cc \
+ test/gtest_filter_unittest_.cc \
+ test/gtest_help_test_.cc \
+ test/gtest_list_tests_unittest_.cc \
+ test/gtest_main_unittest.cc \
+ test/gtest_no_test_unittest.cc \
+ test/gtest_output_test_.cc \
+ test/gtest_pred_impl_unittest.cc \
+ test/gtest_prod_test.cc \
+ test/gtest_repeat_test.cc \
+ test/gtest_shuffle_test_.cc \
+ test/gtest_sole_header_test.cc \
+ test/gtest_stress_test.cc \
+ test/gtest_throw_on_failure_ex_test.cc \
+ test/gtest_throw_on_failure_test_.cc \
+ test/gtest_uninitialized_test_.cc \
+ test/gtest_unittest.cc \
+ test/gtest_unittest.cc \
+ test/gtest_xml_outfile1_test_.cc \
+ test/gtest_xml_outfile2_test_.cc \
+ test/gtest_xml_output_unittest_.cc \
+ test/production.cc \
+ test/production.h
+
+# Python tests that we don't run.
+EXTRA_DIST += \
+ test/gtest_break_on_failure_unittest.py \
+ test/gtest_catch_exceptions_test.py \
+ test/gtest_color_test.py \
+ test/gtest_env_var_test.py \
+ test/gtest_filter_unittest.py \
+ test/gtest_help_test.py \
+ test/gtest_list_tests_unittest.py \
+ test/gtest_output_test.py \
+ test/gtest_output_test_golden_lin.txt \
+ test/gtest_shuffle_test.py \
+ test/gtest_test_utils.py \
+ test/gtest_throw_on_failure_test.py \
+ test/gtest_uninitialized_test.py \
+ test/gtest_xml_outfiles_test.py \
+ test/gtest_xml_output_unittest.py \
+ test/gtest_xml_test_utils.py
+
+# CMake script
+EXTRA_DIST += \
+ CMakeLists.txt \
+ cmake/internal_utils.cmake
+
+# MSVC project files
+EXTRA_DIST += \
+ msvc/gtest-md.sln \
+ msvc/gtest-md.vcproj \
+ msvc/gtest.sln \
+ msvc/gtest.vcproj \
+ msvc/gtest_main-md.vcproj \
+ msvc/gtest_main.vcproj \
+ msvc/gtest_prod_test-md.vcproj \
+ msvc/gtest_prod_test.vcproj \
+ msvc/gtest_unittest-md.vcproj \
+ msvc/gtest_unittest.vcproj
+
+# xcode project files
+EXTRA_DIST += \
+ xcode/Config/DebugProject.xcconfig \
+ xcode/Config/FrameworkTarget.xcconfig \
+ xcode/Config/General.xcconfig \
+ xcode/Config/ReleaseProject.xcconfig \
+ xcode/Config/StaticLibraryTarget.xcconfig \
+ xcode/Config/TestTarget.xcconfig \
+ xcode/Resources/Info.plist \
+ xcode/Scripts/runtests.sh \
+ xcode/Scripts/versiongenerate.py \
+ xcode/gtest.xcodeproj/project.pbxproj
+
+# xcode sample files
+EXTRA_DIST += \
+ xcode/Samples/FrameworkSample/Info.plist \
+ xcode/Samples/FrameworkSample/WidgetFramework.xcodeproj/project.pbxproj \
+ xcode/Samples/FrameworkSample/runtests.sh \
+ xcode/Samples/FrameworkSample/widget.cc \
+ xcode/Samples/FrameworkSample/widget.h \
+ xcode/Samples/FrameworkSample/widget_test.cc
+
+# C++Builder project files
+EXTRA_DIST += \
+ codegear/gtest.cbproj \
+ codegear/gtest.groupproj \
+ codegear/gtest_all.cc \
+ codegear/gtest_link.cc \
+ codegear/gtest_main.cbproj \
+ codegear/gtest_unittest.cbproj
+
+# Distribute and install M4 macro
+m4datadir = $(datadir)/aclocal
+m4data_DATA = m4/gtest.m4
+EXTRA_DIST += $(m4data_DATA)
+
+# We define the global AM_CPPFLAGS as everything we compile includes from these
+# directories.
+AM_CPPFLAGS = -I$(srcdir) -I$(srcdir)/include
+
+# Modifies compiler and linker flags for pthreads compatibility.
+if HAVE_PTHREADS
+ AM_CXXFLAGS = @PTHREAD_CFLAGS@ -DGTEST_HAS_PTHREAD=1
+ AM_LIBS = @PTHREAD_LIBS@
+else
+ AM_CXXFLAGS = -DGTEST_HAS_PTHREAD=0
+endif
+
+# Build rules for libraries.
+lib_LTLIBRARIES = lib/libgtest.la lib/libgtest_main.la
+
+lib_libgtest_la_SOURCES = src/gtest-all.cc
+
+pkginclude_HEADERS = \
+ include/gtest/gtest-death-test.h \
+ include/gtest/gtest-message.h \
+ include/gtest/gtest-param-test.h \
+ include/gtest/gtest-printers.h \
+ include/gtest/gtest-spi.h \
+ include/gtest/gtest-test-part.h \
+ include/gtest/gtest-typed-test.h \
+ include/gtest/gtest.h \
+ include/gtest/gtest_pred_impl.h \
+ include/gtest/gtest_prod.h
+
+pkginclude_internaldir = $(pkgincludedir)/internal
+pkginclude_internal_HEADERS = \
+ include/gtest/internal/gtest-death-test-internal.h \
+ include/gtest/internal/gtest-filepath.h \
+ include/gtest/internal/gtest-internal.h \
+ include/gtest/internal/gtest-linked_ptr.h \
+ include/gtest/internal/gtest-param-util-generated.h \
+ include/gtest/internal/gtest-param-util.h \
+ include/gtest/internal/gtest-port.h \
+ include/gtest/internal/gtest-string.h \
+ include/gtest/internal/gtest-tuple.h \
+ include/gtest/internal/gtest-type-util.h
+
+lib_libgtest_main_la_SOURCES = src/gtest_main.cc
+lib_libgtest_main_la_LIBADD = lib/libgtest.la
+
+# Bulid rules for samples and tests. Automake's naming for some of
+# these variables isn't terribly obvious, so this is a brief
+# reference:
+#
+# TESTS -- Programs run automatically by "make check"
+# check_PROGRAMS -- Programs built by "make check" but not necessarily run
+
+noinst_LTLIBRARIES = samples/libsamples.la
+
+samples_libsamples_la_SOURCES = \
+ samples/sample1.cc \
+ samples/sample1.h \
+ samples/sample2.cc \
+ samples/sample2.h \
+ samples/sample3-inl.h \
+ samples/sample4.cc \
+ samples/sample4.h
+
+TESTS=
+TESTS_ENVIRONMENT = GTEST_SOURCE_DIR="$(srcdir)/test" \
+ GTEST_BUILD_DIR="$(top_builddir)/test"
+check_PROGRAMS=
+
+# A simple sample on using gtest.
+TESTS += samples/sample1_unittest
+check_PROGRAMS += samples/sample1_unittest
+samples_sample1_unittest_SOURCES = samples/sample1_unittest.cc
+samples_sample1_unittest_LDADD = lib/libgtest_main.la \
+ lib/libgtest.la \
+ samples/libsamples.la
+
+# Another sample. It also verifies that libgtest works.
+TESTS += samples/sample10_unittest
+check_PROGRAMS += samples/sample10_unittest
+samples_sample10_unittest_SOURCES = samples/sample10_unittest.cc
+samples_sample10_unittest_LDADD = lib/libgtest.la
+
+# This tests most constructs of gtest and verifies that libgtest_main
+# and libgtest work.
+TESTS += test/gtest_all_test
+check_PROGRAMS += test/gtest_all_test
+test_gtest_all_test_SOURCES = test/gtest_all_test.cc
+test_gtest_all_test_LDADD = lib/libgtest_main.la \
+ lib/libgtest.la
+
+# Tests that fused gtest files compile and work.
+FUSED_GTEST_SRC = \
+ fused-src/gtest/gtest-all.cc \
+ fused-src/gtest/gtest.h \
+ fused-src/gtest/gtest_main.cc
+
+TESTS += test/fused_gtest_test
+check_PROGRAMS += test/fused_gtest_test
+test_fused_gtest_test_SOURCES = $(FUSED_GTEST_SRC) \
+ samples/sample1.cc samples/sample1_unittest.cc
+test_fused_gtest_test_CPPFLAGS = -I"$(srcdir)/fused-src"
+
+# Build rules for putting fused Google Test files into the distribution
+# package. The user can also create those files by manually running
+# scripts/fuse_gtest_files.py.
+$(test_fused_gtest_test_SOURCES): fused-gtest
+
+fused-gtest: $(pkginclude_HEADERS) $(pkginclude_internal_HEADERS) \
+ $(GTEST_SRC) src/gtest-all.cc src/gtest_main.cc \
+ scripts/fuse_gtest_files.py
+ mkdir -p "$(srcdir)/fused-src"
+ chmod -R u+w "$(srcdir)/fused-src"
+ rm -f "$(srcdir)/fused-src/gtest/gtest-all.cc"
+ rm -f "$(srcdir)/fused-src/gtest/gtest.h"
+ "$(srcdir)/scripts/fuse_gtest_files.py" "$(srcdir)/fused-src"
+ cp -f "$(srcdir)/src/gtest_main.cc" "$(srcdir)/fused-src/gtest/"
+
+maintainer-clean-local:
+ rm -rf "$(srcdir)/fused-src"
+
+# Death tests may produce core dumps in the build directory. In case
+# this happens, clean them to keep distcleancheck happy.
+CLEANFILES = core
+
+# Disables 'make install' as installing a compiled version of Google
+# Test can lead to undefined behavior due to violation of the
+# One-Definition Rule.
+
+install-exec-local:
+ echo "'make install' is dangerous and not supported. Instead, see README for how to integrate Google Test into your build system."
+ false
+
+install-data-local:
+ echo "'make install' is dangerous and not supported. Instead, see README for how to integrate Google Test into your build system."
+ false
diff --git a/libvpx/third_party/googletest/src/README b/libvpx/third_party/googletest/src/README
new file mode 100644
index 0000000..51a9376
--- /dev/null
+++ b/libvpx/third_party/googletest/src/README
@@ -0,0 +1,424 @@
+Google C++ Testing Framework
+============================
+
+http://code.google.com/p/googletest/
+
+Overview
+--------
+
+Google's framework for writing C++ tests on a variety of platforms
+(Linux, Mac OS X, Windows, Windows CE, Symbian, etc). Based on the
+xUnit architecture. Supports automatic test discovery, a rich set of
+assertions, user-defined assertions, death tests, fatal and non-fatal
+failures, various options for running the tests, and XML test report
+generation.
+
+Please see the project page above for more information as well as the
+mailing list for questions, discussions, and development. There is
+also an IRC channel on OFTC (irc.oftc.net) #gtest available. Please
+join us!
+
+Requirements for End Users
+--------------------------
+
+Google Test is designed to have fairly minimal requirements to build
+and use with your projects, but there are some. Currently, we support
+Linux, Windows, Mac OS X, and Cygwin. We will also make our best
+effort to support other platforms (e.g. Solaris, AIX, and z/OS).
+However, since core members of the Google Test project have no access
+to these platforms, Google Test may have outstanding issues there. If
+you notice any problems on your platform, please notify
+googletestframework@googlegroups.com. Patches for fixing them are
+even more welcome!
+
+### Linux Requirements ###
+
+These are the base requirements to build and use Google Test from a source
+package (as described below):
+ * GNU-compatible Make or gmake
+ * POSIX-standard shell
+ * POSIX(-2) Regular Expressions (regex.h)
+ * A C++98-standard-compliant compiler
+
+### Windows Requirements ###
+
+ * Microsoft Visual C++ 7.1 or newer
+
+### Cygwin Requirements ###
+
+ * Cygwin 1.5.25-14 or newer
+
+### Mac OS X Requirements ###
+
+ * Mac OS X 10.4 Tiger or newer
+ * Developer Tools Installed
+
+Also, you'll need CMake 2.6.4 or higher if you want to build the
+samples using the provided CMake script, regardless of the platform.
+
+Requirements for Contributors
+-----------------------------
+
+We welcome patches. If you plan to contribute a patch, you need to
+build Google Test and its own tests from an SVN checkout (described
+below), which has further requirements:
+
+ * Python version 2.3 or newer (for running some of the tests and
+ re-generating certain source files from templates)
+ * CMake 2.6.4 or newer
+
+Getting the Source
+------------------
+
+There are two primary ways of getting Google Test's source code: you
+can download a stable source release in your preferred archive format,
+or directly check out the source from our Subversion (SVN) repositary.
+The SVN checkout requires a few extra steps and some extra software
+packages on your system, but lets you track the latest development and
+make patches much more easily, so we highly encourage it.
+
+### Source Package ###
+
+Google Test is released in versioned source packages which can be
+downloaded from the download page [1]. Several different archive
+formats are provided, but the only difference is the tools used to
+manipulate them, and the size of the resulting file. Download
+whichever you are most comfortable with.
+
+ [1] http://code.google.com/p/googletest/downloads/list
+
+Once the package is downloaded, expand it using whichever tools you
+prefer for that type. This will result in a new directory with the
+name "gtest-X.Y.Z" which contains all of the source code. Here are
+some examples on Linux:
+
+ tar -xvzf gtest-X.Y.Z.tar.gz
+ tar -xvjf gtest-X.Y.Z.tar.bz2
+ unzip gtest-X.Y.Z.zip
+
+### SVN Checkout ###
+
+To check out the main branch (also known as the "trunk") of Google
+Test, run the following Subversion command:
+
+ svn checkout http://googletest.googlecode.com/svn/trunk/ gtest-svn
+
+Setting up the Build
+--------------------
+
+To build Google Test and your tests that use it, you need to tell your
+build system where to find its headers and source files. The exact
+way to do it depends on which build system you use, and is usually
+straightforward.
+
+### Generic Build Instructions ###
+
+Suppose you put Google Test in directory ${GTEST_DIR}. To build it,
+create a library build target (or a project as called by Visual Studio
+and Xcode) to compile
+
+ ${GTEST_DIR}/src/gtest-all.cc
+
+with
+
+ ${GTEST_DIR}/include and ${GTEST_DIR}
+
+in the header search path. Assuming a Linux-like system and gcc,
+something like the following will do:
+
+ g++ -I${GTEST_DIR}/include -I${GTEST_DIR} -c ${GTEST_DIR}/src/gtest-all.cc
+ ar -rv libgtest.a gtest-all.o
+
+Next, you should compile your test source file with
+${GTEST_DIR}/include in the header search path, and link it with gtest
+and any other necessary libraries:
+
+ g++ -I${GTEST_DIR}/include path/to/your_test.cc libgtest.a -o your_test
+
+As an example, the make/ directory contains a Makefile that you can
+use to build Google Test on systems where GNU make is available
+(e.g. Linux, Mac OS X, and Cygwin). It doesn't try to build Google
+Test's own tests. Instead, it just builds the Google Test library and
+a sample test. You can use it as a starting point for your own build
+script.
+
+If the default settings are correct for your environment, the
+following commands should succeed:
+
+ cd ${GTEST_DIR}/make
+ make
+ ./sample1_unittest
+
+If you see errors, try to tweak the contents of make/Makefile to make
+them go away. There are instructions in make/Makefile on how to do
+it.
+
+### Using CMake ###
+
+Google Test comes with a CMake build script (CMakeLists.txt) that can
+be used on a wide range of platforms ("C" stands for cross-platofrm.).
+If you don't have CMake installed already, you can download it for
+free from http://www.cmake.org/.
+
+CMake works by generating native makefiles or build projects that can
+be used in the compiler environment of your choice. The typical
+workflow starts with:
+
+ mkdir mybuild # Create a directory to hold the build output.
+ cd mybuild
+ cmake ${GTEST_DIR} # Generate native build scripts.
+
+If you want to build Google Test's samples, you should replace the
+last command with
+
+ cmake -Dgtest_build_samples=ON ${GTEST_DIR}
+
+If you are on a *nix system, you should now see a Makefile in the
+current directory. Just type 'make' to build gtest.
+
+If you use Windows and have Vistual Studio installed, a gtest.sln file
+and several .vcproj files will be created. You can then build them
+using Visual Studio.
+
+On Mac OS X with Xcode installed, a .xcodeproj file will be generated.
+
+### Legacy Build Scripts ###
+
+Before settling on CMake, we have been providing hand-maintained build
+projects/scripts for Visual Studio, Xcode, and Autotools. While we
+continue to provide them for convenience, they are not actively
+maintained any more. We highly recommend that you follow the
+instructions in the previous two sections to integrate Google Test
+with your existing build system.
+
+If you still need to use the legacy build scripts, here's how:
+
+The msvc\ folder contains two solutions with Visual C++ projects.
+Open the gtest.sln or gtest-md.sln file using Visual Studio, and you
+are ready to build Google Test the same way you build any Visual
+Studio project. Files that have names ending with -md use DLL
+versions of Microsoft runtime libraries (the /MD or the /MDd compiler
+option). Files without that suffix use static versions of the runtime
+libraries (the /MT or the /MTd option). Please note that one must use
+the same option to compile both gtest and the test code. If you use
+Visual Studio 2005 or above, we recommend the -md version as /MD is
+the default for new projects in these versions of Visual Studio.
+
+On Mac OS X, open the gtest.xcodeproj in the xcode/ folder using
+Xcode. Build the "gtest" target. The universal binary framework will
+end up in your selected build directory (selected in the Xcode
+"Preferences..." -> "Building" pane and defaults to xcode/build).
+Alternatively, at the command line, enter:
+
+ xcodebuild
+
+This will build the "Release" configuration of gtest.framework in your
+default build location. See the "xcodebuild" man page for more
+information about building different configurations and building in
+different locations.
+
+Tweaking Google Test
+--------------------
+
+Google Test can be used in diverse environments. The default
+configuration may not work (or may not work well) out of the box in
+some environments. However, you can easily tweak Google Test by
+defining control macros on the compiler command line. Generally,
+these macros are named like GTEST_XYZ and you define them to either 1
+or 0 to enable or disable a certain feature.
+
+We list the most frequently used macros below. For a complete list,
+see file include/gtest/internal/gtest-port.h.
+
+### Choosing a TR1 Tuple Library ###
+
+Some Google Test features require the C++ Technical Report 1 (TR1)
+tuple library, which is not yet available with all compilers. The
+good news is that Google Test implements a subset of TR1 tuple that's
+enough for its own need, and will automatically use this when the
+compiler doesn't provide TR1 tuple.
+
+Usually you don't need to care about which tuple library Google Test
+uses. However, if your project already uses TR1 tuple, you need to
+tell Google Test to use the same TR1 tuple library the rest of your
+project uses, or the two tuple implementations will clash. To do
+that, add
+
+ -DGTEST_USE_OWN_TR1_TUPLE=0
+
+to the compiler flags while compiling Google Test and your tests. If
+you want to force Google Test to use its own tuple library, just add
+
+ -DGTEST_USE_OWN_TR1_TUPLE=1
+
+to the compiler flags instead.
+
+If you don't want Google Test to use tuple at all, add
+
+ -DGTEST_HAS_TR1_TUPLE=0
+
+and all features using tuple will be disabled.
+
+### Multi-threaded Tests ###
+
+Google Test is thread-safe where the pthread library is available.
+After #include "gtest/gtest.h", you can check the GTEST_IS_THREADSAFE
+macro to see whether this is the case (yes if the macro is #defined to
+1, no if it's undefined.).
+
+If Google Test doesn't correctly detect whether pthread is available
+in your environment, you can force it with
+
+ -DGTEST_HAS_PTHREAD=1
+
+or
+
+ -DGTEST_HAS_PTHREAD=0
+
+When Google Test uses pthread, you may need to add flags to your
+compiler and/or linker to select the pthread library, or you'll get
+link errors. If you use the CMake script or the deprecated Autotools
+script, this is taken care of for you. If you use your own build
+script, you'll need to read your compiler and linker's manual to
+figure out what flags to add.
+
+### As a Shared Library (DLL) ###
+
+Google Test is compact, so most users can build and link it as a
+static library for the simplicity. You can choose to use Google Test
+as a shared library (known as a DLL on Windows) if you prefer.
+
+To compile *gtest* as a shared library, add
+
+ -DGTEST_CREATE_SHARED_LIBRARY=1
+
+to the compiler flags. You'll also need to tell the linker to produce
+a shared library instead - consult your linker's manual for how to do
+it.
+
+To compile your *tests* that use the gtest shared library, add
+
+ -DGTEST_LINKED_AS_SHARED_LIBRARY=1
+
+to the compiler flags.
+
+Note: while the above steps aren't technically necessary today when
+using some compilers (e.g. GCC), they may become necessary in the
+future, if we decide to improve the speed of loading the library (see
+http://gcc.gnu.org/wiki/Visibility for details). Therefore you are
+recommended to always add the above flags when using Google Test as a
+shared library. Otherwise a future release of Google Test may break
+your build script.
+
+### Avoiding Macro Name Clashes ###
+
+In C++, macros don't obey namespaces. Therefore two libraries that
+both define a macro of the same name will clash if you #include both
+definitions. In case a Google Test macro clashes with another
+library, you can force Google Test to rename its macro to avoid the
+conflict.
+
+Specifically, if both Google Test and some other code define macro
+FOO, you can add
+
+ -DGTEST_DONT_DEFINE_FOO=1
+
+to the compiler flags to tell Google Test to change the macro's name
+from FOO to GTEST_FOO. Currently FOO can be FAIL, SUCCEED, or TEST.
+For example, with -DGTEST_DONT_DEFINE_TEST=1, you'll need to write
+
+ GTEST_TEST(SomeTest, DoesThis) { ... }
+
+instead of
+
+ TEST(SomeTest, DoesThis) { ... }
+
+in order to define a test.
+
+Upgrating from an Earlier Version
+---------------------------------
+
+We strive to keep Google Test releases backward compatible.
+Sometimes, though, we have to make some breaking changes for the
+users' long-term benefits. This section describes what you'll need to
+do if you are upgrading from an earlier version of Google Test.
+
+### Upgrading from 1.3.0 or Earlier ###
+
+You may need to explicitly enable or disable Google Test's own TR1
+tuple library. See the instructions in section "Choosing a TR1 Tuple
+Library".
+
+### Upgrading from 1.4.0 or Earlier ###
+
+The Autotools build script (configure + make) is no longer officially
+supportted. You are encouraged to migrate to your own build system or
+use CMake. If you still need to use Autotools, you can find
+instructions in the README file from Google Test 1.4.0.
+
+On platforms where the pthread library is available, Google Test uses
+it in order to be thread-safe. See the "Multi-threaded Tests" section
+for what this means to your build script.
+
+If you use Microsoft Visual C++ 7.1 with exceptions disabled, Google
+Test will no longer compile. This should affect very few people, as a
+large portion of STL (including <string>) doesn't compile in this mode
+anyway. We decided to stop supporting it in order to greatly simplify
+Google Test's implementation.
+
+Developing Google Test
+----------------------
+
+This section discusses how to make your own changes to Google Test.
+
+### Testing Google Test Itself ###
+
+To make sure your changes work as intended and don't break existing
+functionality, you'll want to compile and run Google Test's own tests.
+For that you can use CMake:
+
+ mkdir mybuild
+ cd mybuild
+ cmake -Dgtest_build_tests=ON ${GTEST_DIR}
+
+Make sure you have Python installed, as some of Google Test's tests
+are written in Python. If the cmake command complains about not being
+able to find Python ("Could NOT find PythonInterp (missing:
+PYTHON_EXECUTABLE)"), try telling it explicitly where your Python
+executable can be found:
+
+ cmake -DPYTHON_EXECUTABLE=path/to/python -Dgtest_build_tests=ON ${GTEST_DIR}
+
+Next, you can build Google Test and all of its own tests. On *nix,
+this is usually done by 'make'. To run the tests, do
+
+ make test
+
+All tests should pass.
+
+### Regenerating Source Files ###
+
+Some of Google Test's source files are generated from templates (not
+in the C++ sense) using a script. A template file is named FOO.pump,
+where FOO is the name of the file it will generate. For example, the
+file include/gtest/internal/gtest-type-util.h.pump is used to generate
+gtest-type-util.h in the same directory.
+
+Normally you don't need to worry about regenerating the source files,
+unless you need to modify them. In that case, you should modify the
+corresponding .pump files instead and run the pump.py Python script to
+regenerate them. You can find pump.py in the scripts/ directory.
+Read the Pump manual [2] for how to use it.
+
+ [2] http://code.google.com/p/googletest/wiki/PumpManual
+
+### Contributing a Patch ###
+
+We welcome patches. Please read the Google Test developer's guide [3]
+for how you can contribute. In particular, make sure you have signed
+the Contributor License Agreement, or we won't be able to accept the
+patch.
+
+ [3] http://code.google.com/p/googletest/wiki/GoogleTestDevGuide
+
+Happy testing!
diff --git a/libvpx/third_party/googletest/src/build-aux/.keep b/libvpx/third_party/googletest/src/build-aux/.keep
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/libvpx/third_party/googletest/src/build-aux/.keep
diff --git a/libvpx/third_party/googletest/src/cmake/internal_utils.cmake b/libvpx/third_party/googletest/src/cmake/internal_utils.cmake
new file mode 100644
index 0000000..7efc2ac
--- /dev/null
+++ b/libvpx/third_party/googletest/src/cmake/internal_utils.cmake
@@ -0,0 +1,216 @@
+# Defines functions and macros useful for building Google Test and
+# Google Mock.
+#
+# Note:
+#
+# - This file will be run twice when building Google Mock (once via
+# Google Test's CMakeLists.txt, and once via Google Mock's).
+# Therefore it shouldn't have any side effects other than defining
+# the functions and macros.
+#
+# - The functions/macros defined in this file may depend on Google
+# Test and Google Mock's option() definitions, and thus must be
+# called *after* the options have been defined.
+
+# Tweaks CMake's default compiler/linker settings to suit Google Test's needs.
+#
+# This must be a macro(), as inside a function string() can only
+# update variables in the function scope.
+macro(fix_default_compiler_settings_)
+ if (MSVC)
+ # For MSVC, CMake sets certain flags to defaults we want to override.
+ # This replacement code is taken from sample in the CMake Wiki at
+ # http://www.cmake.org/Wiki/CMake_FAQ#Dynamic_Replace.
+ foreach (flag_var
+ CMAKE_CXX_FLAGS CMAKE_CXX_FLAGS_DEBUG CMAKE_CXX_FLAGS_RELEASE
+ CMAKE_CXX_FLAGS_MINSIZEREL CMAKE_CXX_FLAGS_RELWITHDEBINFO)
+ if (NOT BUILD_SHARED_LIBS AND NOT gtest_force_shared_crt)
+ # When Google Test is built as a shared library, it should also use
+ # shared runtime libraries. Otherwise, it may end up with multiple
+ # copies of runtime library data in different modules, resulting in
+ # hard-to-find crashes. When it is built as a static library, it is
+ # preferable to use CRT as static libraries, as we don't have to rely
+ # on CRT DLLs being available. CMake always defaults to using shared
+ # CRT libraries, so we override that default here.
+ string(REPLACE "/MD" "-MT" ${flag_var} "${${flag_var}}")
+ endif()
+
+ # We prefer more strict warning checking for building Google Test.
+ # Replaces /W3 with /W4 in defaults.
+ string(REPLACE "/W3" "-W4" ${flag_var} "${${flag_var}}")
+ endforeach()
+ endif()
+endmacro()
+
+# Defines the compiler/linker flags used to build Google Test and
+# Google Mock. You can tweak these definitions to suit your need. A
+# variable's value is empty before it's explicitly assigned to.
+macro(config_compiler_and_linker)
+ if (NOT gtest_disable_pthreads)
+ # Defines CMAKE_USE_PTHREADS_INIT and CMAKE_THREAD_LIBS_INIT.
+ find_package(Threads)
+ endif()
+
+ fix_default_compiler_settings_()
+ if (MSVC)
+ # Newlines inside flags variables break CMake's NMake generator.
+ # TODO(vladl@google.com): Add -RTCs and -RTCu to debug builds.
+ set(cxx_base_flags "-GS -W4 -WX -wd4127 -wd4251 -wd4275 -nologo -J -Zi")
+ set(cxx_base_flags "${cxx_base_flags} -D_UNICODE -DUNICODE -DWIN32 -D_WIN32")
+ set(cxx_base_flags "${cxx_base_flags} -DSTRICT -DWIN32_LEAN_AND_MEAN")
+ set(cxx_exception_flags "-EHsc -D_HAS_EXCEPTIONS=1")
+ set(cxx_no_exception_flags "-D_HAS_EXCEPTIONS=0")
+ set(cxx_no_rtti_flags "-GR-")
+ elseif (CMAKE_COMPILER_IS_GNUCXX)
+ set(cxx_base_flags "-Wall -Wshadow")
+ set(cxx_exception_flags "-fexceptions")
+ set(cxx_no_exception_flags "-fno-exceptions")
+ # Until version 4.3.2, GCC doesn't define a macro to indicate
+ # whether RTTI is enabled. Therefore we define GTEST_HAS_RTTI
+ # explicitly.
+ set(cxx_no_rtti_flags "-fno-rtti -DGTEST_HAS_RTTI=0")
+ set(cxx_strict_flags "-Wextra")
+ elseif (CMAKE_CXX_COMPILER_ID STREQUAL "SunPro")
+ set(cxx_exception_flags "-features=except")
+ # Sun Pro doesn't provide macros to indicate whether exceptions and
+ # RTTI are enabled, so we define GTEST_HAS_* explicitly.
+ set(cxx_no_exception_flags "-features=no%except -DGTEST_HAS_EXCEPTIONS=0")
+ set(cxx_no_rtti_flags "-features=no%rtti -DGTEST_HAS_RTTI=0")
+ elseif (CMAKE_CXX_COMPILER_ID STREQUAL "VisualAge" OR
+ CMAKE_CXX_COMPILER_ID STREQUAL "XL")
+ # CMake 2.8 changes Visual Age's compiler ID to "XL".
+ set(cxx_exception_flags "-qeh")
+ set(cxx_no_exception_flags "-qnoeh")
+ # Until version 9.0, Visual Age doesn't define a macro to indicate
+ # whether RTTI is enabled. Therefore we define GTEST_HAS_RTTI
+ # explicitly.
+ set(cxx_no_rtti_flags "-qnortti -DGTEST_HAS_RTTI=0")
+ elseif (CMAKE_CXX_COMPILER_ID STREQUAL "HP")
+ set(cxx_base_flags "-AA -mt")
+ set(cxx_exception_flags "-DGTEST_HAS_EXCEPTIONS=1")
+ set(cxx_no_exception_flags "+noeh -DGTEST_HAS_EXCEPTIONS=0")
+ # RTTI can not be disabled in HP aCC compiler.
+ set(cxx_no_rtti_flags "")
+ endif()
+
+ if (CMAKE_USE_PTHREADS_INIT) # The pthreads library is available and allowed.
+ set(cxx_base_flags "${cxx_base_flags} -DGTEST_HAS_PTHREAD=1")
+ else()
+ set(cxx_base_flags "${cxx_base_flags} -DGTEST_HAS_PTHREAD=0")
+ endif()
+
+ # For building gtest's own tests and samples.
+ set(cxx_exception "${CMAKE_CXX_FLAGS} ${cxx_base_flags} ${cxx_exception_flags}")
+ set(cxx_no_exception
+ "${CMAKE_CXX_FLAGS} ${cxx_base_flags} ${cxx_no_exception_flags}")
+ set(cxx_default "${cxx_exception}")
+ set(cxx_no_rtti "${cxx_default} ${cxx_no_rtti_flags}")
+ set(cxx_use_own_tuple "${cxx_default} -DGTEST_USE_OWN_TR1_TUPLE=1")
+
+ # For building the gtest libraries.
+ set(cxx_strict "${cxx_default} ${cxx_strict_flags}")
+endmacro()
+
+# Defines the gtest & gtest_main libraries. User tests should link
+# with one of them.
+function(cxx_library_with_type name type cxx_flags)
+ # type can be either STATIC or SHARED to denote a static or shared library.
+ # ARGN refers to additional arguments after 'cxx_flags'.
+ add_library(${name} ${type} ${ARGN})
+ set_target_properties(${name}
+ PROPERTIES
+ COMPILE_FLAGS "${cxx_flags}")
+ if (BUILD_SHARED_LIBS OR type STREQUAL "SHARED")
+ set_target_properties(${name}
+ PROPERTIES
+ COMPILE_DEFINITIONS "GTEST_CREATE_SHARED_LIBRARY=1")
+ endif()
+ if (CMAKE_USE_PTHREADS_INIT)
+ target_link_libraries(${name} ${CMAKE_THREAD_LIBS_INIT})
+ endif()
+endfunction()
+
+########################################################################
+#
+# Helper functions for creating build targets.
+
+function(cxx_shared_library name cxx_flags)
+ cxx_library_with_type(${name} SHARED "${cxx_flags}" ${ARGN})
+endfunction()
+
+function(cxx_library name cxx_flags)
+ cxx_library_with_type(${name} "" "${cxx_flags}" ${ARGN})
+endfunction()
+
+# cxx_executable_with_flags(name cxx_flags libs srcs...)
+#
+# creates a named C++ executable that depends on the given libraries and
+# is built from the given source files with the given compiler flags.
+function(cxx_executable_with_flags name cxx_flags libs)
+ add_executable(${name} ${ARGN})
+ if (cxx_flags)
+ set_target_properties(${name}
+ PROPERTIES
+ COMPILE_FLAGS "${cxx_flags}")
+ endif()
+ if (BUILD_SHARED_LIBS)
+ set_target_properties(${name}
+ PROPERTIES
+ COMPILE_DEFINITIONS "GTEST_LINKED_AS_SHARED_LIBRARY=1")
+ endif()
+ # To support mixing linking in static and dynamic libraries, link each
+ # library in with an extra call to target_link_libraries.
+ foreach (lib "${libs}")
+ target_link_libraries(${name} ${lib})
+ endforeach()
+endfunction()
+
+# cxx_executable(name dir lib srcs...)
+#
+# creates a named target that depends on the given libs and is built
+# from the given source files. dir/name.cc is implicitly included in
+# the source file list.
+function(cxx_executable name dir libs)
+ cxx_executable_with_flags(
+ ${name} "${cxx_default}" "${libs}" "${dir}/${name}.cc" ${ARGN})
+endfunction()
+
+# Sets PYTHONINTERP_FOUND and PYTHON_EXECUTABLE.
+find_package(PythonInterp)
+
+# cxx_test_with_flags(name cxx_flags libs srcs...)
+#
+# creates a named C++ test that depends on the given libs and is built
+# from the given source files with the given compiler flags.
+function(cxx_test_with_flags name cxx_flags libs)
+ cxx_executable_with_flags(${name} "${cxx_flags}" "${libs}" ${ARGN})
+ add_test(${name} ${name})
+endfunction()
+
+# cxx_test(name libs srcs...)
+#
+# creates a named test target that depends on the given libs and is
+# built from the given source files. Unlike cxx_test_with_flags,
+# test/name.cc is already implicitly included in the source file list.
+function(cxx_test name libs)
+ cxx_test_with_flags("${name}" "${cxx_default}" "${libs}"
+ "test/${name}.cc" ${ARGN})
+endfunction()
+
+# py_test(name)
+#
+# creates a Python test with the given name whose main module is in
+# test/name.py. It does nothing if Python is not installed.
+function(py_test name)
+ # We are not supporting Python tests on Linux yet as they consider
+ # all Linux environments to be google3 and try to use google3 features.
+ if (PYTHONINTERP_FOUND)
+ # ${CMAKE_BINARY_DIR} is known at configuration time, so we can
+ # directly bind it from cmake. ${CTEST_CONFIGURATION_TYPE} is known
+ # only at ctest runtime (by calling ctest -c <Configuration>), so
+ # we have to escape $ to delay variable substitution here.
+ add_test(${name}
+ ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/test/${name}.py
+ --build_dir=${CMAKE_CURRENT_BINARY_DIR}/\${CTEST_CONFIGURATION_TYPE})
+ endif()
+endfunction()
diff --git a/libvpx/third_party/googletest/src/codegear/gtest.cbproj b/libvpx/third_party/googletest/src/codegear/gtest.cbproj
new file mode 100644
index 0000000..95c3054
--- /dev/null
+++ b/libvpx/third_party/googletest/src/codegear/gtest.cbproj
@@ -0,0 +1,138 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+ <PropertyGroup>
+ <ProjectGuid>{bca37a72-5b07-46cf-b44e-89f8e06451a2}</ProjectGuid>
+ <Config Condition="'$(Config)'==''">Release</Config>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Base' or '$(Base)'!=''">
+ <Base>true</Base>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Debug' or '$(Cfg_1)'!=''">
+ <Base>true</Base>
+ <Cfg_1>true</Cfg_1>
+ <CfgParent>Base</CfgParent>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Release' or '$(Cfg_2)'!=''">
+ <Base>true</Base>
+ <Cfg_2>true</Cfg_2>
+ <CfgParent>Base</CfgParent>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Base)'!=''">
+ <BCC_OptimizeForSpeed>true</BCC_OptimizeForSpeed>
+ <OutputExt>lib</OutputExt>
+ <DCC_CBuilderOutput>JPHNE</DCC_CBuilderOutput>
+ <Defines>NO_STRICT</Defines>
+ <DynamicRTL>true</DynamicRTL>
+ <UsePackages>true</UsePackages>
+ <ProjectType>CppStaticLibrary</ProjectType>
+ <BCC_CPPCompileAlways>true</BCC_CPPCompileAlways>
+ <PackageImports>rtl.bpi;vcl.bpi;bcbie.bpi;vclx.bpi;vclactnband.bpi;xmlrtl.bpi;bcbsmp.bpi;dbrtl.bpi;vcldb.bpi;bdertl.bpi;vcldbx.bpi;dsnap.bpi;dsnapcon.bpi;vclib.bpi;ibxpress.bpi;adortl.bpi;dbxcds.bpi;dbexpress.bpi;DbxCommonDriver.bpi;websnap.bpi;vclie.bpi;webdsnap.bpi;inet.bpi;inetdbbde.bpi;inetdbxpress.bpi;soaprtl.bpi;Rave75VCL.bpi;teeUI.bpi;tee.bpi;teedb.bpi;IndyCore.bpi;IndySystem.bpi;IndyProtocols.bpi;IntrawebDB_90_100.bpi;Intraweb_90_100.bpi;dclZipForged11.bpi;vclZipForged11.bpi;GR32_BDS2006.bpi;GR32_DSGN_BDS2006.bpi;Jcl.bpi;JclVcl.bpi;JvCoreD11R.bpi;JvSystemD11R.bpi;JvStdCtrlsD11R.bpi;JvAppFrmD11R.bpi;JvBandsD11R.bpi;JvDBD11R.bpi;JvDlgsD11R.bpi;JvBDED11R.bpi;JvCmpD11R.bpi;JvCryptD11R.bpi;JvCtrlsD11R.bpi;JvCustomD11R.bpi;JvDockingD11R.bpi;JvDotNetCtrlsD11R.bpi;JvEDID11R.bpi;JvGlobusD11R.bpi;JvHMID11R.bpi;JvInterpreterD11R.bpi;JvJansD11R.bpi;JvManagedThreadsD11R.bpi;JvMMD11R.bpi;JvNetD11R.bpi;JvPageCompsD11R.bpi;JvPluginD11R.bpi;JvPrintPreviewD11R.bpi;JvRuntimeDesignD11R.bpi;JvTimeFrameworkD11R.bpi;JvValidatorsD11R.bpi;JvWizardD11R.bpi;JvXPCtrlsD11R.bpi;VclSmp.bpi;CExceptionExpert11.bpi</PackageImports>
+ <BCC_wpar>false</BCC_wpar>
+ <IncludePath>$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\src;..\include;..</IncludePath>
+ <AllPackageLibs>rtl.lib;vcl.lib</AllPackageLibs>
+ <TLIB_PageSize>32</TLIB_PageSize>
+ <ILINK_LibraryPath>$(BDS)\lib;$(BDS)\lib\obj;$(BDS)\lib\psdk</ILINK_LibraryPath>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Cfg_1)'!=''">
+ <BCC_OptimizeForSpeed>false</BCC_OptimizeForSpeed>
+ <DCC_Optimize>false</DCC_Optimize>
+ <DCC_DebugInfoInExe>true</DCC_DebugInfoInExe>
+ <Defines>_DEBUG;$(Defines)</Defines>
+ <ILINK_FullDebugInfo>true</ILINK_FullDebugInfo>
+ <BCC_InlineFunctionExpansion>false</BCC_InlineFunctionExpansion>
+ <ILINK_DisableIncrementalLinking>true</ILINK_DisableIncrementalLinking>
+ <BCC_UseRegisterVariables>None</BCC_UseRegisterVariables>
+ <DCC_Define>DEBUG</DCC_Define>
+ <BCC_DebugLineNumbers>true</BCC_DebugLineNumbers>
+ <IntermediateOutputDir>Debug</IntermediateOutputDir>
+ <TASM_DisplaySourceLines>true</TASM_DisplaySourceLines>
+ <BCC_StackFrames>true</BCC_StackFrames>
+ <BCC_DisableOptimizations>true</BCC_DisableOptimizations>
+ <ILINK_LibraryPath>$(BDS)\lib\debug;$(ILINK_LibraryPath)</ILINK_LibraryPath>
+ <TASM_Debugging>Full</TASM_Debugging>
+ <BCC_SourceDebuggingOn>true</BCC_SourceDebuggingOn>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Cfg_2)'!=''">
+ <Defines>NDEBUG;$(Defines)</Defines>
+ <IntermediateOutputDir>Release</IntermediateOutputDir>
+ <ILINK_LibraryPath>$(BDS)\lib\release;$(ILINK_LibraryPath)</ILINK_LibraryPath>
+ <TASM_Debugging>None</TASM_Debugging>
+ </PropertyGroup>
+ <ProjectExtensions>
+ <Borland.Personality>CPlusPlusBuilder.Personality</Borland.Personality>
+ <Borland.ProjectType>CppStaticLibrary</Borland.ProjectType>
+ <BorlandProject>
+<BorlandProject><CPlusPlusBuilder.Personality><VersionInfo><VersionInfo Name="IncludeVerInfo">False</VersionInfo><VersionInfo Name="AutoIncBuild">False</VersionInfo><VersionInfo Name="MajorVer">1</VersionInfo><VersionInfo Name="MinorVer">0</VersionInfo><VersionInfo Name="Release">0</VersionInfo><VersionInfo Name="Build">0</VersionInfo><VersionInfo Name="Debug">False</VersionInfo><VersionInfo Name="PreRelease">False</VersionInfo><VersionInfo Name="Special">False</VersionInfo><VersionInfo Name="Private">False</VersionInfo><VersionInfo Name="DLL">False</VersionInfo><VersionInfo Name="Locale">1033</VersionInfo><VersionInfo Name="CodePage">1252</VersionInfo></VersionInfo><VersionInfoKeys><VersionInfoKeys Name="CompanyName"></VersionInfoKeys><VersionInfoKeys Name="FileDescription"></VersionInfoKeys><VersionInfoKeys Name="FileVersion">1.0.0.0</VersionInfoKeys><VersionInfoKeys Name="InternalName"></VersionInfoKeys><VersionInfoKeys Name="LegalCopyright"></VersionInfoKeys><VersionInfoKeys Name="LegalTrademarks"></VersionInfoKeys><VersionInfoKeys Name="OriginalFilename"></VersionInfoKeys><VersionInfoKeys Name="ProductName"></VersionInfoKeys><VersionInfoKeys Name="ProductVersion">1.0.0.0</VersionInfoKeys><VersionInfoKeys Name="Comments"></VersionInfoKeys></VersionInfoKeys><Debugging><Debugging Name="DebugSourceDirs"></Debugging></Debugging><Parameters><Parameters Name="RunParams"></Parameters><Parameters Name="Launcher"></Parameters><Parameters Name="UseLauncher">False</Parameters><Parameters Name="DebugCWD"></Parameters><Parameters Name="HostApplication"></Parameters><Parameters Name="RemoteHost"></Parameters><Parameters Name="RemotePath"></Parameters><Parameters Name="RemoteParams"></Parameters><Parameters Name="RemoteLauncher"></Parameters><Parameters Name="UseRemoteLauncher">False</Parameters><Parameters Name="RemoteCWD"></Parameters><Parameters Name="RemoteDebug">False</Parameters><Parameters Name="Debug Symbols Search Path"></Parameters><Parameters Name="LoadAllSymbols">True</Parameters><Parameters Name="LoadUnspecifiedSymbols">False</Parameters></Parameters><Excluded_Packages>
+
+
+ <Excluded_Packages Name="$(BDS)\bin\bcboffice2k100.bpl">CodeGear C++Builder Office 2000 Servers Package</Excluded_Packages>
+ <Excluded_Packages Name="$(BDS)\bin\bcbofficexp100.bpl">CodeGear C++Builder Office XP Servers Package</Excluded_Packages>
+ </Excluded_Packages><Linker><Linker Name="LibPrefix"></Linker><Linker Name="LibSuffix"></Linker><Linker Name="LibVersion"></Linker></Linker><ProjectProperties><ProjectProperties Name="AutoShowDeps">False</ProjectProperties><ProjectProperties Name="ManagePaths">True</ProjectProperties><ProjectProperties Name="VerifyPackages">True</ProjectProperties></ProjectProperties><HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Count">3</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item0">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\src;..\include;..</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item1">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\src;..\include;..</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item2">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\src;..\src;..\include</HistoryLists_hlIncludePath></HistoryLists_hlIncludePath><HistoryLists_hlILINK_LibraryPath><HistoryLists_hlILINK_LibraryPath Name="Count">1</HistoryLists_hlILINK_LibraryPath><HistoryLists_hlILINK_LibraryPath Name="Item0">$(BDS)\lib;$(BDS)\lib\obj;$(BDS)\lib\psdk</HistoryLists_hlILINK_LibraryPath></HistoryLists_hlILINK_LibraryPath><HistoryLists_hlDefines><HistoryLists_hlDefines Name="Count">1</HistoryLists_hlDefines><HistoryLists_hlDefines Name="Item0">NO_STRICT</HistoryLists_hlDefines></HistoryLists_hlDefines><HistoryLists_hlTLIB_PageSize><HistoryLists_hlTLIB_PageSize Name="Count">1</HistoryLists_hlTLIB_PageSize><HistoryLists_hlTLIB_PageSize Name="Item0">32</HistoryLists_hlTLIB_PageSize><HistoryLists_hlTLIB_PageSize Name="Item1">16</HistoryLists_hlTLIB_PageSize></HistoryLists_hlTLIB_PageSize></CPlusPlusBuilder.Personality></BorlandProject></BorlandProject>
+ </ProjectExtensions>
+ <Import Project="$(MSBuildBinPath)\Borland.Cpp.Targets" />
+ <ItemGroup>
+ <None Include="..\include\gtest\gtest-death-test.h">
+ <BuildOrder>3</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\gtest-message.h">
+ <BuildOrder>4</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\gtest-param-test.h">
+ <BuildOrder>5</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\gtest-spi.h">
+ <BuildOrder>6</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\gtest-test-part.h">
+ <BuildOrder>7</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\gtest-typed-test.h">
+ <BuildOrder>8</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\gtest.h">
+ <BuildOrder>0</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\gtest_pred_impl.h">
+ <BuildOrder>1</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\gtest_prod.h">
+ <BuildOrder>2</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-death-test-internal.h">
+ <BuildOrder>9</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-filepath.h">
+ <BuildOrder>10</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-internal.h">
+ <BuildOrder>11</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-linked_ptr.h">
+ <BuildOrder>12</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-param-util-generated.h">
+ <BuildOrder>14</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-param-util.h">
+ <BuildOrder>13</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-port.h">
+ <BuildOrder>15</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-string.h">
+ <BuildOrder>16</BuildOrder>
+ </None>
+ <None Include="..\include\gtest\internal\gtest-type-util.h">
+ <BuildOrder>17</BuildOrder>
+ </None>
+ <CppCompile Include="gtest_all.cc">
+ <BuildOrder>18</BuildOrder>
+ </CppCompile>
+ <BuildConfiguration Include="Debug">
+ <Key>Cfg_1</Key>
+ </BuildConfiguration>
+ <BuildConfiguration Include="Release">
+ <Key>Cfg_2</Key>
+ </BuildConfiguration>
+ </ItemGroup>
+</Project>
\ No newline at end of file
diff --git a/libvpx/third_party/googletest/src/codegear/gtest.groupproj b/libvpx/third_party/googletest/src/codegear/gtest.groupproj
new file mode 100644
index 0000000..faf31ca
--- /dev/null
+++ b/libvpx/third_party/googletest/src/codegear/gtest.groupproj
@@ -0,0 +1,54 @@
+<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+ <PropertyGroup>
+ <ProjectGuid>{c1d923e0-6cba-4332-9b6f-3420acbf5091}</ProjectGuid>
+ </PropertyGroup>
+ <ItemGroup />
+ <ItemGroup>
+ <Projects Include="gtest.cbproj" />
+ <Projects Include="gtest_main.cbproj" />
+ <Projects Include="gtest_unittest.cbproj" />
+ </ItemGroup>
+ <ProjectExtensions>
+ <Borland.Personality>Default.Personality</Borland.Personality>
+ <Borland.ProjectType />
+ <BorlandProject>
+<BorlandProject xmlns=""><Default.Personality></Default.Personality></BorlandProject></BorlandProject>
+ </ProjectExtensions>
+ <Target Name="gtest">
+ <MSBuild Projects="gtest.cbproj" Targets="" />
+ </Target>
+ <Target Name="gtest:Clean">
+ <MSBuild Projects="gtest.cbproj" Targets="Clean" />
+ </Target>
+ <Target Name="gtest:Make">
+ <MSBuild Projects="gtest.cbproj" Targets="Make" />
+ </Target>
+ <Target Name="gtest_main">
+ <MSBuild Projects="gtest_main.cbproj" Targets="" />
+ </Target>
+ <Target Name="gtest_main:Clean">
+ <MSBuild Projects="gtest_main.cbproj" Targets="Clean" />
+ </Target>
+ <Target Name="gtest_main:Make">
+ <MSBuild Projects="gtest_main.cbproj" Targets="Make" />
+ </Target>
+ <Target Name="gtest_unittest">
+ <MSBuild Projects="gtest_unittest.cbproj" Targets="" />
+ </Target>
+ <Target Name="gtest_unittest:Clean">
+ <MSBuild Projects="gtest_unittest.cbproj" Targets="Clean" />
+ </Target>
+ <Target Name="gtest_unittest:Make">
+ <MSBuild Projects="gtest_unittest.cbproj" Targets="Make" />
+ </Target>
+ <Target Name="Build">
+ <CallTarget Targets="gtest;gtest_main;gtest_unittest" />
+ </Target>
+ <Target Name="Clean">
+ <CallTarget Targets="gtest:Clean;gtest_main:Clean;gtest_unittest:Clean" />
+ </Target>
+ <Target Name="Make">
+ <CallTarget Targets="gtest:Make;gtest_main:Make;gtest_unittest:Make" />
+ </Target>
+ <Import Condition="Exists('$(MSBuildBinPath)\Borland.Group.Targets')" Project="$(MSBuildBinPath)\Borland.Group.Targets" />
+</Project>
\ No newline at end of file
diff --git a/libvpx/third_party/googletest/src/codegear/gtest_all.cc b/libvpx/third_party/googletest/src/codegear/gtest_all.cc
new file mode 100644
index 0000000..121b2d8
--- /dev/null
+++ b/libvpx/third_party/googletest/src/codegear/gtest_all.cc
@@ -0,0 +1,38 @@
+// Copyright 2009, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: Josh Kelley (joshkel@gmail.com)
+//
+// Google C++ Testing Framework (Google Test)
+//
+// C++Builder's IDE cannot build a static library from files with hyphens
+// in their name. See http://qc.codegear.com/wc/qcmain.aspx?d=70977 .
+// This file serves as a workaround.
+
+#include "src/gtest-all.cc"
diff --git a/libvpx/third_party/googletest/src/codegear/gtest_link.cc b/libvpx/third_party/googletest/src/codegear/gtest_link.cc
new file mode 100644
index 0000000..918eccd
--- /dev/null
+++ b/libvpx/third_party/googletest/src/codegear/gtest_link.cc
@@ -0,0 +1,40 @@
+// Copyright 2009, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: Josh Kelley (joshkel@gmail.com)
+//
+// Google C++ Testing Framework (Google Test)
+//
+// Links gtest.lib and gtest_main.lib into the current project in C++Builder.
+// This means that these libraries can't be renamed, but it's the only way to
+// ensure that Debug versus Release test builds are linked against the
+// appropriate Debug or Release build of the libraries.
+
+#pragma link "gtest.lib"
+#pragma link "gtest_main.lib"
diff --git a/libvpx/third_party/googletest/src/codegear/gtest_main.cbproj b/libvpx/third_party/googletest/src/codegear/gtest_main.cbproj
new file mode 100644
index 0000000..d76ce13
--- /dev/null
+++ b/libvpx/third_party/googletest/src/codegear/gtest_main.cbproj
@@ -0,0 +1,82 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+ <PropertyGroup>
+ <ProjectGuid>{bca37a72-5b07-46cf-b44e-89f8e06451a2}</ProjectGuid>
+ <Config Condition="'$(Config)'==''">Release</Config>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Base' or '$(Base)'!=''">
+ <Base>true</Base>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Debug' or '$(Cfg_1)'!=''">
+ <Base>true</Base>
+ <Cfg_1>true</Cfg_1>
+ <CfgParent>Base</CfgParent>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Release' or '$(Cfg_2)'!=''">
+ <Base>true</Base>
+ <Cfg_2>true</Cfg_2>
+ <CfgParent>Base</CfgParent>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Base)'!=''">
+ <BCC_OptimizeForSpeed>true</BCC_OptimizeForSpeed>
+ <OutputExt>lib</OutputExt>
+ <DCC_CBuilderOutput>JPHNE</DCC_CBuilderOutput>
+ <Defines>NO_STRICT</Defines>
+ <DynamicRTL>true</DynamicRTL>
+ <UsePackages>true</UsePackages>
+ <ProjectType>CppStaticLibrary</ProjectType>
+ <BCC_CPPCompileAlways>true</BCC_CPPCompileAlways>
+ <PackageImports>rtl.bpi;vcl.bpi;bcbie.bpi;vclx.bpi;vclactnband.bpi;xmlrtl.bpi;bcbsmp.bpi;dbrtl.bpi;vcldb.bpi;bdertl.bpi;vcldbx.bpi;dsnap.bpi;dsnapcon.bpi;vclib.bpi;ibxpress.bpi;adortl.bpi;dbxcds.bpi;dbexpress.bpi;DbxCommonDriver.bpi;websnap.bpi;vclie.bpi;webdsnap.bpi;inet.bpi;inetdbbde.bpi;inetdbxpress.bpi;soaprtl.bpi;Rave75VCL.bpi;teeUI.bpi;tee.bpi;teedb.bpi;IndyCore.bpi;IndySystem.bpi;IndyProtocols.bpi;IntrawebDB_90_100.bpi;Intraweb_90_100.bpi;dclZipForged11.bpi;vclZipForged11.bpi;GR32_BDS2006.bpi;GR32_DSGN_BDS2006.bpi;Jcl.bpi;JclVcl.bpi;JvCoreD11R.bpi;JvSystemD11R.bpi;JvStdCtrlsD11R.bpi;JvAppFrmD11R.bpi;JvBandsD11R.bpi;JvDBD11R.bpi;JvDlgsD11R.bpi;JvBDED11R.bpi;JvCmpD11R.bpi;JvCryptD11R.bpi;JvCtrlsD11R.bpi;JvCustomD11R.bpi;JvDockingD11R.bpi;JvDotNetCtrlsD11R.bpi;JvEDID11R.bpi;JvGlobusD11R.bpi;JvHMID11R.bpi;JvInterpreterD11R.bpi;JvJansD11R.bpi;JvManagedThreadsD11R.bpi;JvMMD11R.bpi;JvNetD11R.bpi;JvPageCompsD11R.bpi;JvPluginD11R.bpi;JvPrintPreviewD11R.bpi;JvRuntimeDesignD11R.bpi;JvTimeFrameworkD11R.bpi;JvValidatorsD11R.bpi;JvWizardD11R.bpi;JvXPCtrlsD11R.bpi;VclSmp.bpi;CExceptionExpert11.bpi</PackageImports>
+ <BCC_wpar>false</BCC_wpar>
+ <IncludePath>$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\src;..\include;..</IncludePath>
+ <AllPackageLibs>rtl.lib;vcl.lib</AllPackageLibs>
+ <TLIB_PageSize>32</TLIB_PageSize>
+ <ILINK_LibraryPath>$(BDS)\lib;$(BDS)\lib\obj;$(BDS)\lib\psdk</ILINK_LibraryPath>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Cfg_1)'!=''">
+ <BCC_OptimizeForSpeed>false</BCC_OptimizeForSpeed>
+ <DCC_Optimize>false</DCC_Optimize>
+ <DCC_DebugInfoInExe>true</DCC_DebugInfoInExe>
+ <Defines>_DEBUG;$(Defines)</Defines>
+ <ILINK_FullDebugInfo>true</ILINK_FullDebugInfo>
+ <BCC_InlineFunctionExpansion>false</BCC_InlineFunctionExpansion>
+ <ILINK_DisableIncrementalLinking>true</ILINK_DisableIncrementalLinking>
+ <BCC_UseRegisterVariables>None</BCC_UseRegisterVariables>
+ <DCC_Define>DEBUG</DCC_Define>
+ <BCC_DebugLineNumbers>true</BCC_DebugLineNumbers>
+ <IntermediateOutputDir>Debug</IntermediateOutputDir>
+ <TASM_DisplaySourceLines>true</TASM_DisplaySourceLines>
+ <BCC_StackFrames>true</BCC_StackFrames>
+ <BCC_DisableOptimizations>true</BCC_DisableOptimizations>
+ <ILINK_LibraryPath>$(BDS)\lib\debug;$(ILINK_LibraryPath)</ILINK_LibraryPath>
+ <TASM_Debugging>Full</TASM_Debugging>
+ <BCC_SourceDebuggingOn>true</BCC_SourceDebuggingOn>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Cfg_2)'!=''">
+ <Defines>NDEBUG;$(Defines)</Defines>
+ <IntermediateOutputDir>Release</IntermediateOutputDir>
+ <ILINK_LibraryPath>$(BDS)\lib\release;$(ILINK_LibraryPath)</ILINK_LibraryPath>
+ <TASM_Debugging>None</TASM_Debugging>
+ </PropertyGroup>
+ <ProjectExtensions>
+ <Borland.Personality>CPlusPlusBuilder.Personality</Borland.Personality>
+ <Borland.ProjectType>CppStaticLibrary</Borland.ProjectType>
+ <BorlandProject>
+<BorlandProject><CPlusPlusBuilder.Personality><VersionInfo><VersionInfo Name="IncludeVerInfo">False</VersionInfo><VersionInfo Name="AutoIncBuild">False</VersionInfo><VersionInfo Name="MajorVer">1</VersionInfo><VersionInfo Name="MinorVer">0</VersionInfo><VersionInfo Name="Release">0</VersionInfo><VersionInfo Name="Build">0</VersionInfo><VersionInfo Name="Debug">False</VersionInfo><VersionInfo Name="PreRelease">False</VersionInfo><VersionInfo Name="Special">False</VersionInfo><VersionInfo Name="Private">False</VersionInfo><VersionInfo Name="DLL">False</VersionInfo><VersionInfo Name="Locale">1033</VersionInfo><VersionInfo Name="CodePage">1252</VersionInfo></VersionInfo><VersionInfoKeys><VersionInfoKeys Name="CompanyName"></VersionInfoKeys><VersionInfoKeys Name="FileDescription"></VersionInfoKeys><VersionInfoKeys Name="FileVersion">1.0.0.0</VersionInfoKeys><VersionInfoKeys Name="InternalName"></VersionInfoKeys><VersionInfoKeys Name="LegalCopyright"></VersionInfoKeys><VersionInfoKeys Name="LegalTrademarks"></VersionInfoKeys><VersionInfoKeys Name="OriginalFilename"></VersionInfoKeys><VersionInfoKeys Name="ProductName"></VersionInfoKeys><VersionInfoKeys Name="ProductVersion">1.0.0.0</VersionInfoKeys><VersionInfoKeys Name="Comments"></VersionInfoKeys></VersionInfoKeys><Debugging><Debugging Name="DebugSourceDirs"></Debugging></Debugging><Parameters><Parameters Name="RunParams"></Parameters><Parameters Name="Launcher"></Parameters><Parameters Name="UseLauncher">False</Parameters><Parameters Name="DebugCWD"></Parameters><Parameters Name="HostApplication"></Parameters><Parameters Name="RemoteHost"></Parameters><Parameters Name="RemotePath"></Parameters><Parameters Name="RemoteParams"></Parameters><Parameters Name="RemoteLauncher"></Parameters><Parameters Name="UseRemoteLauncher">False</Parameters><Parameters Name="RemoteCWD"></Parameters><Parameters Name="RemoteDebug">False</Parameters><Parameters Name="Debug Symbols Search Path"></Parameters><Parameters Name="LoadAllSymbols">True</Parameters><Parameters Name="LoadUnspecifiedSymbols">False</Parameters></Parameters><Excluded_Packages>
+ <Excluded_Packages Name="$(BDS)\bin\bcboffice2k100.bpl">CodeGear C++Builder Office 2000 Servers Package</Excluded_Packages>
+ <Excluded_Packages Name="$(BDS)\bin\bcbofficexp100.bpl">CodeGear C++Builder Office XP Servers Package</Excluded_Packages>
+ </Excluded_Packages><Linker><Linker Name="LibPrefix"></Linker><Linker Name="LibSuffix"></Linker><Linker Name="LibVersion"></Linker></Linker><ProjectProperties><ProjectProperties Name="AutoShowDeps">False</ProjectProperties><ProjectProperties Name="ManagePaths">True</ProjectProperties><ProjectProperties Name="VerifyPackages">True</ProjectProperties></ProjectProperties><HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Count">3</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item0">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\src;..\include;..</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item1">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\src;..\include;..</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item2">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\src;..\src;..\include</HistoryLists_hlIncludePath></HistoryLists_hlIncludePath><HistoryLists_hlILINK_LibraryPath><HistoryLists_hlILINK_LibraryPath Name="Count">1</HistoryLists_hlILINK_LibraryPath><HistoryLists_hlILINK_LibraryPath Name="Item0">$(BDS)\lib;$(BDS)\lib\obj;$(BDS)\lib\psdk</HistoryLists_hlILINK_LibraryPath></HistoryLists_hlILINK_LibraryPath><HistoryLists_hlDefines><HistoryLists_hlDefines Name="Count">1</HistoryLists_hlDefines><HistoryLists_hlDefines Name="Item0">NO_STRICT</HistoryLists_hlDefines></HistoryLists_hlDefines><HistoryLists_hlTLIB_PageSize><HistoryLists_hlTLIB_PageSize Name="Count">1</HistoryLists_hlTLIB_PageSize><HistoryLists_hlTLIB_PageSize Name="Item0">32</HistoryLists_hlTLIB_PageSize><HistoryLists_hlTLIB_PageSize Name="Item1">16</HistoryLists_hlTLIB_PageSize></HistoryLists_hlTLIB_PageSize></CPlusPlusBuilder.Personality></BorlandProject></BorlandProject>
+ </ProjectExtensions>
+ <Import Project="$(MSBuildBinPath)\Borland.Cpp.Targets" />
+ <ItemGroup>
+ <CppCompile Include="..\src\gtest_main.cc">
+ <BuildOrder>0</BuildOrder>
+ </CppCompile>
+ <BuildConfiguration Include="Debug">
+ <Key>Cfg_1</Key>
+ </BuildConfiguration>
+ <BuildConfiguration Include="Release">
+ <Key>Cfg_2</Key>
+ </BuildConfiguration>
+ </ItemGroup>
+</Project>
diff --git a/libvpx/third_party/googletest/src/codegear/gtest_unittest.cbproj b/libvpx/third_party/googletest/src/codegear/gtest_unittest.cbproj
new file mode 100644
index 0000000..dc5db8e
--- /dev/null
+++ b/libvpx/third_party/googletest/src/codegear/gtest_unittest.cbproj
@@ -0,0 +1,88 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+ <PropertyGroup>
+ <ProjectGuid>{eea63393-5ac5-4b9c-8909-d75fef2daa41}</ProjectGuid>
+ <Config Condition="'$(Config)'==''">Release</Config>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Base' or '$(Base)'!=''">
+ <Base>true</Base>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Debug' or '$(Cfg_1)'!=''">
+ <Base>true</Base>
+ <Cfg_1>true</Cfg_1>
+ <CfgParent>Base</CfgParent>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Config)'=='Release' or '$(Cfg_2)'!=''">
+ <Base>true</Base>
+ <Cfg_2>true</Cfg_2>
+ <CfgParent>Base</CfgParent>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Base)'!=''">
+ <OutputExt>exe</OutputExt>
+ <BCC_OptimizeForSpeed>true</BCC_OptimizeForSpeed>
+ <Defines>NO_STRICT</Defines>
+ <DCC_CBuilderOutput>JPHNE</DCC_CBuilderOutput>
+ <DynamicRTL>true</DynamicRTL>
+ <ILINK_ObjectSearchPath>..\test</ILINK_ObjectSearchPath>
+ <UsePackages>true</UsePackages>
+ <ProjectType>CppConsoleApplication</ProjectType>
+ <NoVCL>true</NoVCL>
+ <BCC_CPPCompileAlways>true</BCC_CPPCompileAlways>
+ <PackageImports>rtl.bpi;vcl.bpi;bcbie.bpi;vclx.bpi;vclactnband.bpi;xmlrtl.bpi;bcbsmp.bpi;dbrtl.bpi;vcldb.bpi;bdertl.bpi;vcldbx.bpi;dsnap.bpi;dsnapcon.bpi;vclib.bpi;ibxpress.bpi;adortl.bpi;dbxcds.bpi;dbexpress.bpi;DbxCommonDriver.bpi;websnap.bpi;vclie.bpi;webdsnap.bpi;inet.bpi;inetdbbde.bpi;inetdbxpress.bpi;soaprtl.bpi;Rave75VCL.bpi;teeUI.bpi;tee.bpi;teedb.bpi;IndyCore.bpi;IndySystem.bpi;IndyProtocols.bpi;IntrawebDB_90_100.bpi;Intraweb_90_100.bpi;Jcl.bpi;JclVcl.bpi;JvCoreD11R.bpi;JvSystemD11R.bpi;JvStdCtrlsD11R.bpi;JvAppFrmD11R.bpi;JvBandsD11R.bpi;JvDBD11R.bpi;JvDlgsD11R.bpi;JvBDED11R.bpi;JvCmpD11R.bpi;JvCryptD11R.bpi;JvCtrlsD11R.bpi;JvCustomD11R.bpi;JvDockingD11R.bpi;JvDotNetCtrlsD11R.bpi;JvEDID11R.bpi;JvGlobusD11R.bpi;JvHMID11R.bpi;JvInterpreterD11R.bpi;JvJansD11R.bpi;JvManagedThreadsD11R.bpi;JvMMD11R.bpi;JvNetD11R.bpi;JvPageCompsD11R.bpi;JvPluginD11R.bpi;JvPrintPreviewD11R.bpi;JvRuntimeDesignD11R.bpi;JvTimeFrameworkD11R.bpi;JvValidatorsD11R.bpi;JvWizardD11R.bpi;JvXPCtrlsD11R.bpi;VclSmp.bpi</PackageImports>
+ <BCC_wpar>false</BCC_wpar>
+ <IncludePath>$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\include;..\test;..</IncludePath>
+ <ILINK_LibraryPath>$(BDS)\lib;$(BDS)\lib\obj;$(BDS)\lib\psdk;..\test</ILINK_LibraryPath>
+ <Multithreaded>true</Multithreaded>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Cfg_1)'!=''">
+ <BCC_OptimizeForSpeed>false</BCC_OptimizeForSpeed>
+ <DCC_Optimize>false</DCC_Optimize>
+ <DCC_DebugInfoInExe>true</DCC_DebugInfoInExe>
+ <Defines>_DEBUG;$(Defines)</Defines>
+ <ILINK_FullDebugInfo>true</ILINK_FullDebugInfo>
+ <BCC_InlineFunctionExpansion>false</BCC_InlineFunctionExpansion>
+ <ILINK_DisableIncrementalLinking>true</ILINK_DisableIncrementalLinking>
+ <BCC_UseRegisterVariables>None</BCC_UseRegisterVariables>
+ <DCC_Define>DEBUG</DCC_Define>
+ <BCC_DebugLineNumbers>true</BCC_DebugLineNumbers>
+ <IntermediateOutputDir>Debug</IntermediateOutputDir>
+ <TASM_DisplaySourceLines>true</TASM_DisplaySourceLines>
+ <BCC_StackFrames>true</BCC_StackFrames>
+ <BCC_DisableOptimizations>true</BCC_DisableOptimizations>
+ <ILINK_LibraryPath>$(BDS)\lib\debug;$(ILINK_LibraryPath)</ILINK_LibraryPath>
+ <TASM_Debugging>Full</TASM_Debugging>
+ <BCC_SourceDebuggingOn>true</BCC_SourceDebuggingOn>
+ </PropertyGroup>
+ <PropertyGroup Condition="'$(Cfg_2)'!=''">
+ <Defines>NDEBUG;$(Defines)</Defines>
+ <IntermediateOutputDir>Release</IntermediateOutputDir>
+ <ILINK_LibraryPath>$(BDS)\lib\release;$(ILINK_LibraryPath)</ILINK_LibraryPath>
+ <TASM_Debugging>None</TASM_Debugging>
+ </PropertyGroup>
+ <ProjectExtensions>
+ <Borland.Personality>CPlusPlusBuilder.Personality</Borland.Personality>
+ <Borland.ProjectType>CppConsoleApplication</Borland.ProjectType>
+ <BorlandProject>
+<BorlandProject><CPlusPlusBuilder.Personality><VersionInfo><VersionInfo Name="IncludeVerInfo">False</VersionInfo><VersionInfo Name="AutoIncBuild">False</VersionInfo><VersionInfo Name="MajorVer">1</VersionInfo><VersionInfo Name="MinorVer">0</VersionInfo><VersionInfo Name="Release">0</VersionInfo><VersionInfo Name="Build">0</VersionInfo><VersionInfo Name="Debug">False</VersionInfo><VersionInfo Name="PreRelease">False</VersionInfo><VersionInfo Name="Special">False</VersionInfo><VersionInfo Name="Private">False</VersionInfo><VersionInfo Name="DLL">False</VersionInfo><VersionInfo Name="Locale">1033</VersionInfo><VersionInfo Name="CodePage">1252</VersionInfo></VersionInfo><VersionInfoKeys><VersionInfoKeys Name="CompanyName"></VersionInfoKeys><VersionInfoKeys Name="FileDescription"></VersionInfoKeys><VersionInfoKeys Name="FileVersion">1.0.0.0</VersionInfoKeys><VersionInfoKeys Name="InternalName"></VersionInfoKeys><VersionInfoKeys Name="LegalCopyright"></VersionInfoKeys><VersionInfoKeys Name="LegalTrademarks"></VersionInfoKeys><VersionInfoKeys Name="OriginalFilename"></VersionInfoKeys><VersionInfoKeys Name="ProductName"></VersionInfoKeys><VersionInfoKeys Name="ProductVersion">1.0.0.0</VersionInfoKeys><VersionInfoKeys Name="Comments"></VersionInfoKeys></VersionInfoKeys><Debugging><Debugging Name="DebugSourceDirs"></Debugging></Debugging><Parameters><Parameters Name="RunParams"></Parameters><Parameters Name="Launcher"></Parameters><Parameters Name="UseLauncher">False</Parameters><Parameters Name="DebugCWD"></Parameters><Parameters Name="HostApplication"></Parameters><Parameters Name="RemoteHost"></Parameters><Parameters Name="RemotePath"></Parameters><Parameters Name="RemoteParams"></Parameters><Parameters Name="RemoteLauncher"></Parameters><Parameters Name="UseRemoteLauncher">False</Parameters><Parameters Name="RemoteCWD"></Parameters><Parameters Name="RemoteDebug">False</Parameters><Parameters Name="Debug Symbols Search Path"></Parameters><Parameters Name="LoadAllSymbols">True</Parameters><Parameters Name="LoadUnspecifiedSymbols">False</Parameters></Parameters><Excluded_Packages>
+
+
+ <Excluded_Packages Name="$(BDS)\bin\bcboffice2k100.bpl">CodeGear C++Builder Office 2000 Servers Package</Excluded_Packages>
+ <Excluded_Packages Name="$(BDS)\bin\bcbofficexp100.bpl">CodeGear C++Builder Office XP Servers Package</Excluded_Packages>
+ </Excluded_Packages><Linker><Linker Name="LibPrefix"></Linker><Linker Name="LibSuffix"></Linker><Linker Name="LibVersion"></Linker></Linker><ProjectProperties><ProjectProperties Name="AutoShowDeps">False</ProjectProperties><ProjectProperties Name="ManagePaths">True</ProjectProperties><ProjectProperties Name="VerifyPackages">True</ProjectProperties></ProjectProperties><HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Count">3</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item0">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\include;..\test;..</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item1">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\include;..\test</HistoryLists_hlIncludePath><HistoryLists_hlIncludePath Name="Item2">$(BDS)\include;$(BDS)\include\dinkumware;$(BDS)\include\vcl;..\include</HistoryLists_hlIncludePath></HistoryLists_hlIncludePath><HistoryLists_hlILINK_LibraryPath><HistoryLists_hlILINK_LibraryPath Name="Count">1</HistoryLists_hlILINK_LibraryPath><HistoryLists_hlILINK_LibraryPath Name="Item0">$(BDS)\lib;$(BDS)\lib\obj;$(BDS)\lib\psdk;..\test</HistoryLists_hlILINK_LibraryPath><HistoryLists_hlILINK_LibraryPath Name="Item1">$(BDS)\lib;$(BDS)\lib\obj;$(BDS)\lib\psdk;..\test</HistoryLists_hlILINK_LibraryPath><HistoryLists_hlILINK_LibraryPath Name="Item2">$(BDS)\lib;$(BDS)\lib\obj;$(BDS)\lib\psdk;$(OUTPUTDIR);..\test</HistoryLists_hlILINK_LibraryPath></HistoryLists_hlILINK_LibraryPath><HistoryLists_hlDefines><HistoryLists_hlDefines Name="Count">2</HistoryLists_hlDefines><HistoryLists_hlDefines Name="Item0">NO_STRICT</HistoryLists_hlDefines><HistoryLists_hlDefines Name="Item1">STRICT</HistoryLists_hlDefines></HistoryLists_hlDefines></CPlusPlusBuilder.Personality></BorlandProject></BorlandProject>
+ </ProjectExtensions>
+ <Import Project="$(MSBuildBinPath)\Borland.Cpp.Targets" />
+ <ItemGroup>
+ <CppCompile Include="..\test\gtest_unittest.cc">
+ <BuildOrder>0</BuildOrder>
+ </CppCompile>
+ <CppCompile Include="gtest_link.cc">
+ <BuildOrder>1</BuildOrder>
+ </CppCompile>
+ <BuildConfiguration Include="Debug">
+ <Key>Cfg_1</Key>
+ </BuildConfiguration>
+ <BuildConfiguration Include="Release">
+ <Key>Cfg_2</Key>
+ </BuildConfiguration>
+ </ItemGroup>
+</Project>
\ No newline at end of file
diff --git a/libvpx/third_party/googletest/src/configure.ac b/libvpx/third_party/googletest/src/configure.ac
new file mode 100644
index 0000000..fa66029
--- /dev/null
+++ b/libvpx/third_party/googletest/src/configure.ac
@@ -0,0 +1,68 @@
+m4_include(m4/acx_pthread.m4)
+
+# At this point, the Xcode project assumes the version string will be three
+# integers separated by periods and surrounded by square brackets (e.g.
+# "[1.0.1]"). It also asumes that there won't be any closing parenthesis
+# between "AC_INIT(" and the closing ")" including comments and strings.
+AC_INIT([Google C++ Testing Framework],
+ [1.6.0],
+ [googletestframework@googlegroups.com],
+ [gtest])
+
+# Provide various options to initialize the Autoconf and configure processes.
+AC_PREREQ([2.59])
+AC_CONFIG_SRCDIR([./COPYING])
+AC_CONFIG_MACRO_DIR([m4])
+AC_CONFIG_AUX_DIR([build-aux])
+AC_CONFIG_HEADERS([build-aux/config.h])
+AC_CONFIG_FILES([Makefile])
+AC_CONFIG_FILES([scripts/gtest-config], [chmod +x scripts/gtest-config])
+
+# Initialize Automake with various options. We require at least v1.9, prevent
+# pedantic complaints about package files, and enable various distribution
+# targets.
+AM_INIT_AUTOMAKE([1.9 dist-bzip2 dist-zip foreign subdir-objects])
+
+# Check for programs used in building Google Test.
+AC_PROG_CC
+AC_PROG_CXX
+AC_LANG([C++])
+AC_PROG_LIBTOOL
+
+# TODO(chandlerc@google.com): Currently we aren't running the Python tests
+# against the interpreter detected by AM_PATH_PYTHON, and so we condition
+# HAVE_PYTHON by requiring "python" to be in the PATH, and that interpreter's
+# version to be >= 2.3. This will allow the scripts to use a "/usr/bin/env"
+# hashbang.
+PYTHON= # We *do not* allow the user to specify a python interpreter
+AC_PATH_PROG([PYTHON],[python],[:])
+AS_IF([test "$PYTHON" != ":"],
+ [AM_PYTHON_CHECK_VERSION([$PYTHON],[2.3],[:],[PYTHON=":"])])
+AM_CONDITIONAL([HAVE_PYTHON],[test "$PYTHON" != ":"])
+
+# Configure pthreads.
+AC_ARG_WITH([pthreads],
+ [AS_HELP_STRING([--with-pthreads],
+ [use pthreads (default is yes)])],
+ [with_pthreads=$withval],
+ [with_pthreads=check])
+
+have_pthreads=no
+AS_IF([test "x$with_pthreads" != "xno"],
+ [ACX_PTHREAD(
+ [],
+ [AS_IF([test "x$with_pthreads" != "xcheck"],
+ [AC_MSG_FAILURE(
+ [--with-pthreads was specified, but unable to be used])])])
+ have_pthreads="$acx_pthread_ok"])
+AM_CONDITIONAL([HAVE_PTHREADS],[test "x$have_pthreads" == "xyes"])
+AC_SUBST(PTHREAD_CFLAGS)
+AC_SUBST(PTHREAD_LIBS)
+
+# TODO(chandlerc@google.com) Check for the necessary system headers.
+
+# TODO(chandlerc@google.com) Check the types, structures, and other compiler
+# and architecture characteristics.
+
+# Output the generated files. No further autoconf macros may be used.
+AC_OUTPUT
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest-death-test.h b/libvpx/third_party/googletest/src/include/gtest/gtest-death-test.h
new file mode 100644
index 0000000..a27883f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest-death-test.h
@@ -0,0 +1,283 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the public API for death tests. It is
+// #included by gtest.h so a user doesn't need to include this
+// directly.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
+
+#include "gtest/internal/gtest-death-test-internal.h"
+
+namespace testing {
+
+// This flag controls the style of death tests. Valid values are "threadsafe",
+// meaning that the death test child process will re-execute the test binary
+// from the start, running only a single death test, or "fast",
+// meaning that the child process will execute the test logic immediately
+// after forking.
+GTEST_DECLARE_string_(death_test_style);
+
+#if GTEST_HAS_DEATH_TEST
+
+// The following macros are useful for writing death tests.
+
+// Here's what happens when an ASSERT_DEATH* or EXPECT_DEATH* is
+// executed:
+//
+// 1. It generates a warning if there is more than one active
+// thread. This is because it's safe to fork() or clone() only
+// when there is a single thread.
+//
+// 2. The parent process clone()s a sub-process and runs the death
+// test in it; the sub-process exits with code 0 at the end of the
+// death test, if it hasn't exited already.
+//
+// 3. The parent process waits for the sub-process to terminate.
+//
+// 4. The parent process checks the exit code and error message of
+// the sub-process.
+//
+// Examples:
+//
+// ASSERT_DEATH(server.SendMessage(56, "Hello"), "Invalid port number");
+// for (int i = 0; i < 5; i++) {
+// EXPECT_DEATH(server.ProcessRequest(i),
+// "Invalid request .* in ProcessRequest()")
+// << "Failed to die on request " << i);
+// }
+//
+// ASSERT_EXIT(server.ExitNow(), ::testing::ExitedWithCode(0), "Exiting");
+//
+// bool KilledBySIGHUP(int exit_code) {
+// return WIFSIGNALED(exit_code) && WTERMSIG(exit_code) == SIGHUP;
+// }
+//
+// ASSERT_EXIT(client.HangUpServer(), KilledBySIGHUP, "Hanging up!");
+//
+// On the regular expressions used in death tests:
+//
+// On POSIX-compliant systems (*nix), we use the <regex.h> library,
+// which uses the POSIX extended regex syntax.
+//
+// On other platforms (e.g. Windows), we only support a simple regex
+// syntax implemented as part of Google Test. This limited
+// implementation should be enough most of the time when writing
+// death tests; though it lacks many features you can find in PCRE
+// or POSIX extended regex syntax. For example, we don't support
+// union ("x|y"), grouping ("(xy)"), brackets ("[xy]"), and
+// repetition count ("x{5,7}"), among others.
+//
+// Below is the syntax that we do support. We chose it to be a
+// subset of both PCRE and POSIX extended regex, so it's easy to
+// learn wherever you come from. In the following: 'A' denotes a
+// literal character, period (.), or a single \\ escape sequence;
+// 'x' and 'y' denote regular expressions; 'm' and 'n' are for
+// natural numbers.
+//
+// c matches any literal character c
+// \\d matches any decimal digit
+// \\D matches any character that's not a decimal digit
+// \\f matches \f
+// \\n matches \n
+// \\r matches \r
+// \\s matches any ASCII whitespace, including \n
+// \\S matches any character that's not a whitespace
+// \\t matches \t
+// \\v matches \v
+// \\w matches any letter, _, or decimal digit
+// \\W matches any character that \\w doesn't match
+// \\c matches any literal character c, which must be a punctuation
+// . matches any single character except \n
+// A? matches 0 or 1 occurrences of A
+// A* matches 0 or many occurrences of A
+// A+ matches 1 or many occurrences of A
+// ^ matches the beginning of a string (not that of each line)
+// $ matches the end of a string (not that of each line)
+// xy matches x followed by y
+//
+// If you accidentally use PCRE or POSIX extended regex features
+// not implemented by us, you will get a run-time failure. In that
+// case, please try to rewrite your regular expression within the
+// above syntax.
+//
+// This implementation is *not* meant to be as highly tuned or robust
+// as a compiled regex library, but should perform well enough for a
+// death test, which already incurs significant overhead by launching
+// a child process.
+//
+// Known caveats:
+//
+// A "threadsafe" style death test obtains the path to the test
+// program from argv[0] and re-executes it in the sub-process. For
+// simplicity, the current implementation doesn't search the PATH
+// when launching the sub-process. This means that the user must
+// invoke the test program via a path that contains at least one
+// path separator (e.g. path/to/foo_test and
+// /absolute/path/to/bar_test are fine, but foo_test is not). This
+// is rarely a problem as people usually don't put the test binary
+// directory in PATH.
+//
+// TODO(wan@google.com): make thread-safe death tests search the PATH.
+
+// Asserts that a given statement causes the program to exit, with an
+// integer exit status that satisfies predicate, and emitting error output
+// that matches regex.
+# define ASSERT_EXIT(statement, predicate, regex) \
+ GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_FATAL_FAILURE_)
+
+// Like ASSERT_EXIT, but continues on to successive tests in the
+// test case, if any:
+# define EXPECT_EXIT(statement, predicate, regex) \
+ GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_NONFATAL_FAILURE_)
+
+// Asserts that a given statement causes the program to exit, either by
+// explicitly exiting with a nonzero exit code or being killed by a
+// signal, and emitting error output that matches regex.
+# define ASSERT_DEATH(statement, regex) \
+ ASSERT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex)
+
+// Like ASSERT_DEATH, but continues on to successive tests in the
+// test case, if any:
+# define EXPECT_DEATH(statement, regex) \
+ EXPECT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex)
+
+// Two predicate classes that can be used in {ASSERT,EXPECT}_EXIT*:
+
+// Tests that an exit code describes a normal exit with a given exit code.
+class GTEST_API_ ExitedWithCode {
+ public:
+ explicit ExitedWithCode(int exit_code);
+ bool operator()(int exit_status) const;
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ExitedWithCode& other);
+
+ const int exit_code_;
+};
+
+# if !GTEST_OS_WINDOWS
+// Tests that an exit code describes an exit due to termination by a
+// given signal.
+class GTEST_API_ KilledBySignal {
+ public:
+ explicit KilledBySignal(int signum);
+ bool operator()(int exit_status) const;
+ private:
+ const int signum_;
+};
+# endif // !GTEST_OS_WINDOWS
+
+// EXPECT_DEBUG_DEATH asserts that the given statements die in debug mode.
+// The death testing framework causes this to have interesting semantics,
+// since the sideeffects of the call are only visible in opt mode, and not
+// in debug mode.
+//
+// In practice, this can be used to test functions that utilize the
+// LOG(DFATAL) macro using the following style:
+//
+// int DieInDebugOr12(int* sideeffect) {
+// if (sideeffect) {
+// *sideeffect = 12;
+// }
+// LOG(DFATAL) << "death";
+// return 12;
+// }
+//
+// TEST(TestCase, TestDieOr12WorksInDgbAndOpt) {
+// int sideeffect = 0;
+// // Only asserts in dbg.
+// EXPECT_DEBUG_DEATH(DieInDebugOr12(&sideeffect), "death");
+//
+// #ifdef NDEBUG
+// // opt-mode has sideeffect visible.
+// EXPECT_EQ(12, sideeffect);
+// #else
+// // dbg-mode no visible sideeffect.
+// EXPECT_EQ(0, sideeffect);
+// #endif
+// }
+//
+// This will assert that DieInDebugReturn12InOpt() crashes in debug
+// mode, usually due to a DCHECK or LOG(DFATAL), but returns the
+// appropriate fallback value (12 in this case) in opt mode. If you
+// need to test that a function has appropriate side-effects in opt
+// mode, include assertions against the side-effects. A general
+// pattern for this is:
+//
+// EXPECT_DEBUG_DEATH({
+// // Side-effects here will have an effect after this statement in
+// // opt mode, but none in debug mode.
+// EXPECT_EQ(12, DieInDebugOr12(&sideeffect));
+// }, "death");
+//
+# ifdef NDEBUG
+
+# define EXPECT_DEBUG_DEATH(statement, regex) \
+ do { statement; } while (::testing::internal::AlwaysFalse())
+
+# define ASSERT_DEBUG_DEATH(statement, regex) \
+ do { statement; } while (::testing::internal::AlwaysFalse())
+
+# else
+
+# define EXPECT_DEBUG_DEATH(statement, regex) \
+ EXPECT_DEATH(statement, regex)
+
+# define ASSERT_DEBUG_DEATH(statement, regex) \
+ ASSERT_DEATH(statement, regex)
+
+# endif // NDEBUG for EXPECT_DEBUG_DEATH
+#endif // GTEST_HAS_DEATH_TEST
+
+// EXPECT_DEATH_IF_SUPPORTED(statement, regex) and
+// ASSERT_DEATH_IF_SUPPORTED(statement, regex) expand to real death tests if
+// death tests are supported; otherwise they just issue a warning. This is
+// useful when you are combining death test assertions with normal test
+// assertions in one test.
+#if GTEST_HAS_DEATH_TEST
+# define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \
+ EXPECT_DEATH(statement, regex)
+# define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \
+ ASSERT_DEATH(statement, regex)
+#else
+# define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \
+ GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, )
+# define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \
+ GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, return)
+#endif
+
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest-message.h b/libvpx/third_party/googletest/src/include/gtest/gtest-message.h
new file mode 100644
index 0000000..9b7142f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest-message.h
@@ -0,0 +1,230 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the Message class.
+//
+// IMPORTANT NOTE: Due to limitation of the C++ language, we have to
+// leave some internal implementation details in this header file.
+// They are clearly marked by comments like this:
+//
+// // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+//
+// Such code is NOT meant to be used by a user directly, and is subject
+// to CHANGE WITHOUT NOTICE. Therefore DO NOT DEPEND ON IT in a user
+// program!
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
+#define GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
+
+#include <limits>
+
+#include "gtest/internal/gtest-string.h"
+#include "gtest/internal/gtest-internal.h"
+
+namespace testing {
+
+// The Message class works like an ostream repeater.
+//
+// Typical usage:
+//
+// 1. You stream a bunch of values to a Message object.
+// It will remember the text in a stringstream.
+// 2. Then you stream the Message object to an ostream.
+// This causes the text in the Message to be streamed
+// to the ostream.
+//
+// For example;
+//
+// testing::Message foo;
+// foo << 1 << " != " << 2;
+// std::cout << foo;
+//
+// will print "1 != 2".
+//
+// Message is not intended to be inherited from. In particular, its
+// destructor is not virtual.
+//
+// Note that stringstream behaves differently in gcc and in MSVC. You
+// can stream a NULL char pointer to it in the former, but not in the
+// latter (it causes an access violation if you do). The Message
+// class hides this difference by treating a NULL char pointer as
+// "(null)".
+class GTEST_API_ Message {
+ private:
+ // The type of basic IO manipulators (endl, ends, and flush) for
+ // narrow streams.
+ typedef std::ostream& (*BasicNarrowIoManip)(std::ostream&);
+
+ public:
+ // Constructs an empty Message.
+ // We allocate the stringstream separately because otherwise each use of
+ // ASSERT/EXPECT in a procedure adds over 200 bytes to the procedure's
+ // stack frame leading to huge stack frames in some cases; gcc does not reuse
+ // the stack space.
+ Message() : ss_(new ::std::stringstream) {
+ // By default, we want there to be enough precision when printing
+ // a double to a Message.
+ *ss_ << std::setprecision(std::numeric_limits<double>::digits10 + 2);
+ }
+
+ // Copy constructor.
+ Message(const Message& msg) : ss_(new ::std::stringstream) { // NOLINT
+ *ss_ << msg.GetString();
+ }
+
+ // Constructs a Message from a C-string.
+ explicit Message(const char* str) : ss_(new ::std::stringstream) {
+ *ss_ << str;
+ }
+
+#if GTEST_OS_SYMBIAN
+ // Streams a value (either a pointer or not) to this object.
+ template <typename T>
+ inline Message& operator <<(const T& value) {
+ StreamHelper(typename internal::is_pointer<T>::type(), value);
+ return *this;
+ }
+#else
+ // Streams a non-pointer value to this object.
+ template <typename T>
+ inline Message& operator <<(const T& val) {
+ ::GTestStreamToHelper(ss_.get(), val);
+ return *this;
+ }
+
+ // Streams a pointer value to this object.
+ //
+ // This function is an overload of the previous one. When you
+ // stream a pointer to a Message, this definition will be used as it
+ // is more specialized. (The C++ Standard, section
+ // [temp.func.order].) If you stream a non-pointer, then the
+ // previous definition will be used.
+ //
+ // The reason for this overload is that streaming a NULL pointer to
+ // ostream is undefined behavior. Depending on the compiler, you
+ // may get "0", "(nil)", "(null)", or an access violation. To
+ // ensure consistent result across compilers, we always treat NULL
+ // as "(null)".
+ template <typename T>
+ inline Message& operator <<(T* const& pointer) { // NOLINT
+ if (pointer == NULL) {
+ *ss_ << "(null)";
+ } else {
+ ::GTestStreamToHelper(ss_.get(), pointer);
+ }
+ return *this;
+ }
+#endif // GTEST_OS_SYMBIAN
+
+ // Since the basic IO manipulators are overloaded for both narrow
+ // and wide streams, we have to provide this specialized definition
+ // of operator <<, even though its body is the same as the
+ // templatized version above. Without this definition, streaming
+ // endl or other basic IO manipulators to Message will confuse the
+ // compiler.
+ Message& operator <<(BasicNarrowIoManip val) {
+ *ss_ << val;
+ return *this;
+ }
+
+ // Instead of 1/0, we want to see true/false for bool values.
+ Message& operator <<(bool b) {
+ return *this << (b ? "true" : "false");
+ }
+
+ // These two overloads allow streaming a wide C string to a Message
+ // using the UTF-8 encoding.
+ Message& operator <<(const wchar_t* wide_c_str) {
+ return *this << internal::String::ShowWideCString(wide_c_str);
+ }
+ Message& operator <<(wchar_t* wide_c_str) {
+ return *this << internal::String::ShowWideCString(wide_c_str);
+ }
+
+#if GTEST_HAS_STD_WSTRING
+ // Converts the given wide string to a narrow string using the UTF-8
+ // encoding, and streams the result to this Message object.
+ Message& operator <<(const ::std::wstring& wstr);
+#endif // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+ // Converts the given wide string to a narrow string using the UTF-8
+ // encoding, and streams the result to this Message object.
+ Message& operator <<(const ::wstring& wstr);
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+ // Gets the text streamed to this object so far as a String.
+ // Each '\0' character in the buffer is replaced with "\\0".
+ //
+ // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+ internal::String GetString() const {
+ return internal::StringStreamToString(ss_.get());
+ }
+
+ private:
+
+#if GTEST_OS_SYMBIAN
+ // These are needed as the Nokia Symbian Compiler cannot decide between
+ // const T& and const T* in a function template. The Nokia compiler _can_
+ // decide between class template specializations for T and T*, so a
+ // tr1::type_traits-like is_pointer works, and we can overload on that.
+ template <typename T>
+ inline void StreamHelper(internal::true_type /*dummy*/, T* pointer) {
+ if (pointer == NULL) {
+ *ss_ << "(null)";
+ } else {
+ ::GTestStreamToHelper(ss_.get(), pointer);
+ }
+ }
+ template <typename T>
+ inline void StreamHelper(internal::false_type /*dummy*/, const T& value) {
+ ::GTestStreamToHelper(ss_.get(), value);
+ }
+#endif // GTEST_OS_SYMBIAN
+
+ // We'll hold the text streamed to this object here.
+ const internal::scoped_ptr< ::std::stringstream> ss_;
+
+ // We declare (but don't implement) this to prevent the compiler
+ // from implementing the assignment operator.
+ void operator=(const Message&);
+};
+
+// Streams a Message to an ostream.
+inline std::ostream& operator <<(std::ostream& os, const Message& sb) {
+ return os << sb.GetString();
+}
+
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest-param-test.h b/libvpx/third_party/googletest/src/include/gtest/gtest-param-test.h
new file mode 100644
index 0000000..6407cfd
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest-param-test.h
@@ -0,0 +1,1421 @@
+// This file was GENERATED by command:
+// pump.py gtest-param-test.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: vladl@google.com (Vlad Losev)
+//
+// Macros and functions for implementing parameterized tests
+// in Google C++ Testing Framework (Google Test)
+//
+// This file is generated by a SCRIPT. DO NOT EDIT BY HAND!
+//
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+
+
+// Value-parameterized tests allow you to test your code with different
+// parameters without writing multiple copies of the same test.
+//
+// Here is how you use value-parameterized tests:
+
+#if 0
+
+// To write value-parameterized tests, first you should define a fixture
+// class. It is usually derived from testing::TestWithParam<T> (see below for
+// another inheritance scheme that's sometimes useful in more complicated
+// class hierarchies), where the type of your parameter values.
+// TestWithParam<T> is itself derived from testing::Test. T can be any
+// copyable type. If it's a raw pointer, you are responsible for managing the
+// lifespan of the pointed values.
+
+class FooTest : public ::testing::TestWithParam<const char*> {
+ // You can implement all the usual class fixture members here.
+};
+
+// Then, use the TEST_P macro to define as many parameterized tests
+// for this fixture as you want. The _P suffix is for "parameterized"
+// or "pattern", whichever you prefer to think.
+
+TEST_P(FooTest, DoesBlah) {
+ // Inside a test, access the test parameter with the GetParam() method
+ // of the TestWithParam<T> class:
+ EXPECT_TRUE(foo.Blah(GetParam()));
+ ...
+}
+
+TEST_P(FooTest, HasBlahBlah) {
+ ...
+}
+
+// Finally, you can use INSTANTIATE_TEST_CASE_P to instantiate the test
+// case with any set of parameters you want. Google Test defines a number
+// of functions for generating test parameters. They return what we call
+// (surprise!) parameter generators. Here is a summary of them, which
+// are all in the testing namespace:
+//
+//
+// Range(begin, end [, step]) - Yields values {begin, begin+step,
+// begin+step+step, ...}. The values do not
+// include end. step defaults to 1.
+// Values(v1, v2, ..., vN) - Yields values {v1, v2, ..., vN}.
+// ValuesIn(container) - Yields values from a C-style array, an STL
+// ValuesIn(begin,end) container, or an iterator range [begin, end).
+// Bool() - Yields sequence {false, true}.
+// Combine(g1, g2, ..., gN) - Yields all combinations (the Cartesian product
+// for the math savvy) of the values generated
+// by the N generators.
+//
+// For more details, see comments at the definitions of these functions below
+// in this file.
+//
+// The following statement will instantiate tests from the FooTest test case
+// each with parameter values "meeny", "miny", and "moe".
+
+INSTANTIATE_TEST_CASE_P(InstantiationName,
+ FooTest,
+ Values("meeny", "miny", "moe"));
+
+// To distinguish different instances of the pattern, (yes, you
+// can instantiate it more then once) the first argument to the
+// INSTANTIATE_TEST_CASE_P macro is a prefix that will be added to the
+// actual test case name. Remember to pick unique prefixes for different
+// instantiations. The tests from the instantiation above will have
+// these names:
+//
+// * InstantiationName/FooTest.DoesBlah/0 for "meeny"
+// * InstantiationName/FooTest.DoesBlah/1 for "miny"
+// * InstantiationName/FooTest.DoesBlah/2 for "moe"
+// * InstantiationName/FooTest.HasBlahBlah/0 for "meeny"
+// * InstantiationName/FooTest.HasBlahBlah/1 for "miny"
+// * InstantiationName/FooTest.HasBlahBlah/2 for "moe"
+//
+// You can use these names in --gtest_filter.
+//
+// This statement will instantiate all tests from FooTest again, each
+// with parameter values "cat" and "dog":
+
+const char* pets[] = {"cat", "dog"};
+INSTANTIATE_TEST_CASE_P(AnotherInstantiationName, FooTest, ValuesIn(pets));
+
+// The tests from the instantiation above will have these names:
+//
+// * AnotherInstantiationName/FooTest.DoesBlah/0 for "cat"
+// * AnotherInstantiationName/FooTest.DoesBlah/1 for "dog"
+// * AnotherInstantiationName/FooTest.HasBlahBlah/0 for "cat"
+// * AnotherInstantiationName/FooTest.HasBlahBlah/1 for "dog"
+//
+// Please note that INSTANTIATE_TEST_CASE_P will instantiate all tests
+// in the given test case, whether their definitions come before or
+// AFTER the INSTANTIATE_TEST_CASE_P statement.
+//
+// Please also note that generator expressions (including parameters to the
+// generators) are evaluated in InitGoogleTest(), after main() has started.
+// This allows the user on one hand, to adjust generator parameters in order
+// to dynamically determine a set of tests to run and on the other hand,
+// give the user a chance to inspect the generated tests with Google Test
+// reflection API before RUN_ALL_TESTS() is executed.
+//
+// You can see samples/sample7_unittest.cc and samples/sample8_unittest.cc
+// for more examples.
+//
+// In the future, we plan to publish the API for defining new parameter
+// generators. But for now this interface remains part of the internal
+// implementation and is subject to change.
+//
+//
+// A parameterized test fixture must be derived from testing::Test and from
+// testing::WithParamInterface<T>, where T is the type of the parameter
+// values. Inheriting from TestWithParam<T> satisfies that requirement because
+// TestWithParam<T> inherits from both Test and WithParamInterface. In more
+// complicated hierarchies, however, it is occasionally useful to inherit
+// separately from Test and WithParamInterface. For example:
+
+class BaseTest : public ::testing::Test {
+ // You can inherit all the usual members for a non-parameterized test
+ // fixture here.
+};
+
+class DerivedTest : public BaseTest, public ::testing::WithParamInterface<int> {
+ // The usual test fixture members go here too.
+};
+
+TEST_F(BaseTest, HasFoo) {
+ // This is an ordinary non-parameterized test.
+}
+
+TEST_P(DerivedTest, DoesBlah) {
+ // GetParam works just the same here as if you inherit from TestWithParam.
+ EXPECT_TRUE(foo.Blah(GetParam()));
+}
+
+#endif // 0
+
+#include "gtest/internal/gtest-port.h"
+
+#if !GTEST_OS_SYMBIAN
+# include <utility>
+#endif
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*. Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+#include "gtest/internal/gtest-internal.h"
+#include "gtest/internal/gtest-param-util.h"
+#include "gtest/internal/gtest-param-util-generated.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing {
+
+// Functions producing parameter generators.
+//
+// Google Test uses these generators to produce parameters for value-
+// parameterized tests. When a parameterized test case is instantiated
+// with a particular generator, Google Test creates and runs tests
+// for each element in the sequence produced by the generator.
+//
+// In the following sample, tests from test case FooTest are instantiated
+// each three times with parameter values 3, 5, and 8:
+//
+// class FooTest : public TestWithParam<int> { ... };
+//
+// TEST_P(FooTest, TestThis) {
+// }
+// TEST_P(FooTest, TestThat) {
+// }
+// INSTANTIATE_TEST_CASE_P(TestSequence, FooTest, Values(3, 5, 8));
+//
+
+// Range() returns generators providing sequences of values in a range.
+//
+// Synopsis:
+// Range(start, end)
+// - returns a generator producing a sequence of values {start, start+1,
+// start+2, ..., }.
+// Range(start, end, step)
+// - returns a generator producing a sequence of values {start, start+step,
+// start+step+step, ..., }.
+// Notes:
+// * The generated sequences never include end. For example, Range(1, 5)
+// returns a generator producing a sequence {1, 2, 3, 4}. Range(1, 9, 2)
+// returns a generator producing {1, 3, 5, 7}.
+// * start and end must have the same type. That type may be any integral or
+// floating-point type or a user defined type satisfying these conditions:
+// * It must be assignable (have operator=() defined).
+// * It must have operator+() (operator+(int-compatible type) for
+// two-operand version).
+// * It must have operator<() defined.
+// Elements in the resulting sequences will also have that type.
+// * Condition start < end must be satisfied in order for resulting sequences
+// to contain any elements.
+//
+template <typename T, typename IncrementT>
+internal::ParamGenerator<T> Range(T start, T end, IncrementT step) {
+ return internal::ParamGenerator<T>(
+ new internal::RangeGenerator<T, IncrementT>(start, end, step));
+}
+
+template <typename T>
+internal::ParamGenerator<T> Range(T start, T end) {
+ return Range(start, end, 1);
+}
+
+// ValuesIn() function allows generation of tests with parameters coming from
+// a container.
+//
+// Synopsis:
+// ValuesIn(const T (&array)[N])
+// - returns a generator producing sequences with elements from
+// a C-style array.
+// ValuesIn(const Container& container)
+// - returns a generator producing sequences with elements from
+// an STL-style container.
+// ValuesIn(Iterator begin, Iterator end)
+// - returns a generator producing sequences with elements from
+// a range [begin, end) defined by a pair of STL-style iterators. These
+// iterators can also be plain C pointers.
+//
+// Please note that ValuesIn copies the values from the containers
+// passed in and keeps them to generate tests in RUN_ALL_TESTS().
+//
+// Examples:
+//
+// This instantiates tests from test case StringTest
+// each with C-string values of "foo", "bar", and "baz":
+//
+// const char* strings[] = {"foo", "bar", "baz"};
+// INSTANTIATE_TEST_CASE_P(StringSequence, SrtingTest, ValuesIn(strings));
+//
+// This instantiates tests from test case StlStringTest
+// each with STL strings with values "a" and "b":
+//
+// ::std::vector< ::std::string> GetParameterStrings() {
+// ::std::vector< ::std::string> v;
+// v.push_back("a");
+// v.push_back("b");
+// return v;
+// }
+//
+// INSTANTIATE_TEST_CASE_P(CharSequence,
+// StlStringTest,
+// ValuesIn(GetParameterStrings()));
+//
+//
+// This will also instantiate tests from CharTest
+// each with parameter values 'a' and 'b':
+//
+// ::std::list<char> GetParameterChars() {
+// ::std::list<char> list;
+// list.push_back('a');
+// list.push_back('b');
+// return list;
+// }
+// ::std::list<char> l = GetParameterChars();
+// INSTANTIATE_TEST_CASE_P(CharSequence2,
+// CharTest,
+// ValuesIn(l.begin(), l.end()));
+//
+template <typename ForwardIterator>
+internal::ParamGenerator<
+ typename ::testing::internal::IteratorTraits<ForwardIterator>::value_type>
+ValuesIn(ForwardIterator begin, ForwardIterator end) {
+ typedef typename ::testing::internal::IteratorTraits<ForwardIterator>
+ ::value_type ParamType;
+ return internal::ParamGenerator<ParamType>(
+ new internal::ValuesInIteratorRangeGenerator<ParamType>(begin, end));
+}
+
+template <typename T, size_t N>
+internal::ParamGenerator<T> ValuesIn(const T (&array)[N]) {
+ return ValuesIn(array, array + N);
+}
+
+template <class Container>
+internal::ParamGenerator<typename Container::value_type> ValuesIn(
+ const Container& container) {
+ return ValuesIn(container.begin(), container.end());
+}
+
+// Values() allows generating tests from explicitly specified list of
+// parameters.
+//
+// Synopsis:
+// Values(T v1, T v2, ..., T vN)
+// - returns a generator producing sequences with elements v1, v2, ..., vN.
+//
+// For example, this instantiates tests from test case BarTest each
+// with values "one", "two", and "three":
+//
+// INSTANTIATE_TEST_CASE_P(NumSequence, BarTest, Values("one", "two", "three"));
+//
+// This instantiates tests from test case BazTest each with values 1, 2, 3.5.
+// The exact type of values will depend on the type of parameter in BazTest.
+//
+// INSTANTIATE_TEST_CASE_P(FloatingNumbers, BazTest, Values(1, 2, 3.5));
+//
+// Currently, Values() supports from 1 to 50 parameters.
+//
+template <typename T1>
+internal::ValueArray1<T1> Values(T1 v1) {
+ return internal::ValueArray1<T1>(v1);
+}
+
+template <typename T1, typename T2>
+internal::ValueArray2<T1, T2> Values(T1 v1, T2 v2) {
+ return internal::ValueArray2<T1, T2>(v1, v2);
+}
+
+template <typename T1, typename T2, typename T3>
+internal::ValueArray3<T1, T2, T3> Values(T1 v1, T2 v2, T3 v3) {
+ return internal::ValueArray3<T1, T2, T3>(v1, v2, v3);
+}
+
+template <typename T1, typename T2, typename T3, typename T4>
+internal::ValueArray4<T1, T2, T3, T4> Values(T1 v1, T2 v2, T3 v3, T4 v4) {
+ return internal::ValueArray4<T1, T2, T3, T4>(v1, v2, v3, v4);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5>
+internal::ValueArray5<T1, T2, T3, T4, T5> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+ T5 v5) {
+ return internal::ValueArray5<T1, T2, T3, T4, T5>(v1, v2, v3, v4, v5);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6>
+internal::ValueArray6<T1, T2, T3, T4, T5, T6> Values(T1 v1, T2 v2, T3 v3,
+ T4 v4, T5 v5, T6 v6) {
+ return internal::ValueArray6<T1, T2, T3, T4, T5, T6>(v1, v2, v3, v4, v5, v6);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7>
+internal::ValueArray7<T1, T2, T3, T4, T5, T6, T7> Values(T1 v1, T2 v2, T3 v3,
+ T4 v4, T5 v5, T6 v6, T7 v7) {
+ return internal::ValueArray7<T1, T2, T3, T4, T5, T6, T7>(v1, v2, v3, v4, v5,
+ v6, v7);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8>
+internal::ValueArray8<T1, T2, T3, T4, T5, T6, T7, T8> Values(T1 v1, T2 v2,
+ T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8) {
+ return internal::ValueArray8<T1, T2, T3, T4, T5, T6, T7, T8>(v1, v2, v3, v4,
+ v5, v6, v7, v8);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9>
+internal::ValueArray9<T1, T2, T3, T4, T5, T6, T7, T8, T9> Values(T1 v1, T2 v2,
+ T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9) {
+ return internal::ValueArray9<T1, T2, T3, T4, T5, T6, T7, T8, T9>(v1, v2, v3,
+ v4, v5, v6, v7, v8, v9);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10>
+internal::ValueArray10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> Values(T1 v1,
+ T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10) {
+ return internal::ValueArray10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10>(v1,
+ v2, v3, v4, v5, v6, v7, v8, v9, v10);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11>
+internal::ValueArray11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10,
+ T11> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11) {
+ return internal::ValueArray11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10,
+ T11>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12>
+internal::ValueArray12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12) {
+ return internal::ValueArray12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13>
+internal::ValueArray13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13) {
+ return internal::ValueArray13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14>
+internal::ValueArray14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14) {
+ return internal::ValueArray14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+ v14);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15>
+internal::ValueArray15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+ T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15) {
+ return internal::ValueArray15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+ v13, v14, v15);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16>
+internal::ValueArray16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+ T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16) {
+ return internal::ValueArray16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+ v12, v13, v14, v15, v16);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17>
+internal::ValueArray17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+ T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16, T17 v17) {
+ return internal::ValueArray17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+ v11, v12, v13, v14, v15, v16, v17);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18>
+internal::ValueArray18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+ T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16, T17 v17, T18 v18) {
+ return internal::ValueArray18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+ v10, v11, v12, v13, v14, v15, v16, v17, v18);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19>
+internal::ValueArray19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+ T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+ T15 v15, T16 v16, T17 v17, T18 v18, T19 v19) {
+ return internal::ValueArray19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19>(v1, v2, v3, v4, v5, v6, v7, v8,
+ v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20>
+internal::ValueArray20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+ T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+ T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20) {
+ return internal::ValueArray20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20>(v1, v2, v3, v4, v5, v6, v7,
+ v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21>
+internal::ValueArray21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+ T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+ T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21) {
+ return internal::ValueArray21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21>(v1, v2, v3, v4, v5, v6,
+ v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22>
+internal::ValueArray22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22> Values(T1 v1, T2 v2, T3 v3,
+ T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+ T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+ T21 v21, T22 v22) {
+ return internal::ValueArray22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22>(v1, v2, v3, v4,
+ v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+ v20, v21, v22);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23>
+internal::ValueArray23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> Values(T1 v1, T2 v2,
+ T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+ T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+ T21 v21, T22 v22, T23 v23) {
+ return internal::ValueArray23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23>(v1, v2, v3,
+ v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+ v20, v21, v22, v23);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24>
+internal::ValueArray24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> Values(T1 v1, T2 v2,
+ T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+ T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+ T21 v21, T22 v22, T23 v23, T24 v24) {
+ return internal::ValueArray24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24>(v1, v2,
+ v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18,
+ v19, v20, v21, v22, v23, v24);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25>
+internal::ValueArray25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Values(T1 v1,
+ T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11,
+ T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19,
+ T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25) {
+ return internal::ValueArray25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25>(v1,
+ v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17,
+ v18, v19, v20, v21, v22, v23, v24, v25);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26>
+internal::ValueArray26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26) {
+ return internal::ValueArray26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15,
+ v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27>
+internal::ValueArray27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27) {
+ return internal::ValueArray27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14,
+ v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28>
+internal::ValueArray28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28) {
+ return internal::ValueArray28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+ v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27,
+ v28);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29>
+internal::ValueArray29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29) {
+ return internal::ValueArray29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+ v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26,
+ v27, v28, v29);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30>
+internal::ValueArray30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+ T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+ T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+ T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30) {
+ return internal::ValueArray30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+ v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25,
+ v26, v27, v28, v29, v30);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31>
+internal::ValueArray31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+ T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+ T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31) {
+ return internal::ValueArray31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+ v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24,
+ v25, v26, v27, v28, v29, v30, v31);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32>
+internal::ValueArray32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+ T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+ T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+ T32 v32) {
+ return internal::ValueArray32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+ v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+ v24, v25, v26, v27, v28, v29, v30, v31, v32);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33>
+internal::ValueArray33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+ T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+ T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+ T32 v32, T33 v33) {
+ return internal::ValueArray33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33>(v1, v2, v3, v4, v5, v6, v7, v8,
+ v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+ v24, v25, v26, v27, v28, v29, v30, v31, v32, v33);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34>
+internal::ValueArray34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+ T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+ T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22,
+ T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30,
+ T31 v31, T32 v32, T33 v33, T34 v34) {
+ return internal::ValueArray34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34>(v1, v2, v3, v4, v5, v6, v7,
+ v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22,
+ v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35>
+internal::ValueArray35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+ T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+ T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+ T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+ T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35) {
+ return internal::ValueArray35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35>(v1, v2, v3, v4, v5, v6,
+ v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21,
+ v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36>
+internal::ValueArray36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+ T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+ T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+ T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+ T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36) {
+ return internal::ValueArray36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36>(v1, v2, v3, v4,
+ v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+ v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+ v34, v35, v36);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37>
+internal::ValueArray37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37> Values(T1 v1, T2 v2, T3 v3,
+ T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+ T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+ T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+ T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+ T37 v37) {
+ return internal::ValueArray37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37>(v1, v2, v3,
+ v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+ v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+ v34, v35, v36, v37);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38>
+internal::ValueArray38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> Values(T1 v1, T2 v2,
+ T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+ T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+ T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+ T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+ T37 v37, T38 v38) {
+ return internal::ValueArray38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38>(v1, v2,
+ v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18,
+ v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32,
+ v33, v34, v35, v36, v37, v38);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39>
+internal::ValueArray39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Values(T1 v1, T2 v2,
+ T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+ T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+ T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+ T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+ T37 v37, T38 v38, T39 v39) {
+ return internal::ValueArray39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39>(v1,
+ v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17,
+ v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31,
+ v32, v33, v34, v35, v36, v37, v38, v39);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40>
+internal::ValueArray40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Values(T1 v1,
+ T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11,
+ T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19,
+ T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27,
+ T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35,
+ T36 v36, T37 v37, T38 v38, T39 v39, T40 v40) {
+ return internal::ValueArray40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15,
+ v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29,
+ v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41>
+internal::ValueArray41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41) {
+ return internal::ValueArray41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14,
+ v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28,
+ v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42>
+internal::ValueArray42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42) {
+ return internal::ValueArray42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+ v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27,
+ v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41,
+ v42);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43>
+internal::ValueArray43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43) {
+ return internal::ValueArray43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42, T43>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+ v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26,
+ v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40,
+ v41, v42, v43);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44>
+internal::ValueArray44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43, T44 v44) {
+ return internal::ValueArray44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42, T43, T44>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+ v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25,
+ v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39,
+ v40, v41, v42, v43, v44);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45>
+internal::ValueArray45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+ T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+ T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+ T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32,
+ T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40,
+ T41 v41, T42 v42, T43 v43, T44 v44, T45 v45) {
+ return internal::ValueArray45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42, T43, T44, T45>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+ v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24,
+ v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38,
+ v39, v40, v41, v42, v43, v44, v45);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46>
+internal::ValueArray46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+ T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+ T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+ T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+ T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46) {
+ return internal::ValueArray46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42, T43, T44, T45, T46>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+ v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+ v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37,
+ v38, v39, v40, v41, v42, v43, v44, v45, v46);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47>
+internal::ValueArray47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+ T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+ T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+ T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+ T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47) {
+ return internal::ValueArray47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42, T43, T44, T45, T46, T47>(v1, v2, v3, v4, v5, v6, v7, v8,
+ v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+ v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37,
+ v38, v39, v40, v41, v42, v43, v44, v45, v46, v47);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48>
+internal::ValueArray48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47, T48> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+ T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+ T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+ T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+ T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+ T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47,
+ T48 v48) {
+ return internal::ValueArray48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42, T43, T44, T45, T46, T47, T48>(v1, v2, v3, v4, v5, v6, v7,
+ v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22,
+ v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36,
+ v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48, typename T49>
+internal::ValueArray49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47, T48, T49> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+ T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+ T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22,
+ T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30,
+ T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38,
+ T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46,
+ T47 v47, T48 v48, T49 v49) {
+ return internal::ValueArray49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42, T43, T44, T45, T46, T47, T48, T49>(v1, v2, v3, v4, v5, v6,
+ v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21,
+ v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35,
+ v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48, v49);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48, typename T49, typename T50>
+internal::ValueArray50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47, T48, T49, T50> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+ T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+ T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+ T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+ T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37,
+ T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45,
+ T46 v46, T47 v47, T48 v48, T49 v49, T50 v50) {
+ return internal::ValueArray50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40, T41, T42, T43, T44, T45, T46, T47, T48, T49, T50>(v1, v2, v3, v4,
+ v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+ v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+ v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47,
+ v48, v49, v50);
+}
+
+// Bool() allows generating tests with parameters in a set of (false, true).
+//
+// Synopsis:
+// Bool()
+// - returns a generator producing sequences with elements {false, true}.
+//
+// It is useful when testing code that depends on Boolean flags. Combinations
+// of multiple flags can be tested when several Bool()'s are combined using
+// Combine() function.
+//
+// In the following example all tests in the test case FlagDependentTest
+// will be instantiated twice with parameters false and true.
+//
+// class FlagDependentTest : public testing::TestWithParam<bool> {
+// virtual void SetUp() {
+// external_flag = GetParam();
+// }
+// }
+// INSTANTIATE_TEST_CASE_P(BoolSequence, FlagDependentTest, Bool());
+//
+inline internal::ParamGenerator<bool> Bool() {
+ return Values(false, true);
+}
+
+# if GTEST_HAS_COMBINE
+// Combine() allows the user to combine two or more sequences to produce
+// values of a Cartesian product of those sequences' elements.
+//
+// Synopsis:
+// Combine(gen1, gen2, ..., genN)
+// - returns a generator producing sequences with elements coming from
+// the Cartesian product of elements from the sequences generated by
+// gen1, gen2, ..., genN. The sequence elements will have a type of
+// tuple<T1, T2, ..., TN> where T1, T2, ..., TN are the types
+// of elements from sequences produces by gen1, gen2, ..., genN.
+//
+// Combine can have up to 10 arguments. This number is currently limited
+// by the maximum number of elements in the tuple implementation used by Google
+// Test.
+//
+// Example:
+//
+// This will instantiate tests in test case AnimalTest each one with
+// the parameter values tuple("cat", BLACK), tuple("cat", WHITE),
+// tuple("dog", BLACK), and tuple("dog", WHITE):
+//
+// enum Color { BLACK, GRAY, WHITE };
+// class AnimalTest
+// : public testing::TestWithParam<tuple<const char*, Color> > {...};
+//
+// TEST_P(AnimalTest, AnimalLooksNice) {...}
+//
+// INSTANTIATE_TEST_CASE_P(AnimalVariations, AnimalTest,
+// Combine(Values("cat", "dog"),
+// Values(BLACK, WHITE)));
+//
+// This will instantiate tests in FlagDependentTest with all variations of two
+// Boolean flags:
+//
+// class FlagDependentTest
+// : public testing::TestWithParam<tuple(bool, bool)> > {
+// virtual void SetUp() {
+// // Assigns external_flag_1 and external_flag_2 values from the tuple.
+// tie(external_flag_1, external_flag_2) = GetParam();
+// }
+// };
+//
+// TEST_P(FlagDependentTest, TestFeature1) {
+// // Test your code using external_flag_1 and external_flag_2 here.
+// }
+// INSTANTIATE_TEST_CASE_P(TwoBoolSequence, FlagDependentTest,
+// Combine(Bool(), Bool()));
+//
+template <typename Generator1, typename Generator2>
+internal::CartesianProductHolder2<Generator1, Generator2> Combine(
+ const Generator1& g1, const Generator2& g2) {
+ return internal::CartesianProductHolder2<Generator1, Generator2>(
+ g1, g2);
+}
+
+template <typename Generator1, typename Generator2, typename Generator3>
+internal::CartesianProductHolder3<Generator1, Generator2, Generator3> Combine(
+ const Generator1& g1, const Generator2& g2, const Generator3& g3) {
+ return internal::CartesianProductHolder3<Generator1, Generator2, Generator3>(
+ g1, g2, g3);
+}
+
+template <typename Generator1, typename Generator2, typename Generator3,
+ typename Generator4>
+internal::CartesianProductHolder4<Generator1, Generator2, Generator3,
+ Generator4> Combine(
+ const Generator1& g1, const Generator2& g2, const Generator3& g3,
+ const Generator4& g4) {
+ return internal::CartesianProductHolder4<Generator1, Generator2, Generator3,
+ Generator4>(
+ g1, g2, g3, g4);
+}
+
+template <typename Generator1, typename Generator2, typename Generator3,
+ typename Generator4, typename Generator5>
+internal::CartesianProductHolder5<Generator1, Generator2, Generator3,
+ Generator4, Generator5> Combine(
+ const Generator1& g1, const Generator2& g2, const Generator3& g3,
+ const Generator4& g4, const Generator5& g5) {
+ return internal::CartesianProductHolder5<Generator1, Generator2, Generator3,
+ Generator4, Generator5>(
+ g1, g2, g3, g4, g5);
+}
+
+template <typename Generator1, typename Generator2, typename Generator3,
+ typename Generator4, typename Generator5, typename Generator6>
+internal::CartesianProductHolder6<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6> Combine(
+ const Generator1& g1, const Generator2& g2, const Generator3& g3,
+ const Generator4& g4, const Generator5& g5, const Generator6& g6) {
+ return internal::CartesianProductHolder6<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6>(
+ g1, g2, g3, g4, g5, g6);
+}
+
+template <typename Generator1, typename Generator2, typename Generator3,
+ typename Generator4, typename Generator5, typename Generator6,
+ typename Generator7>
+internal::CartesianProductHolder7<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6, Generator7> Combine(
+ const Generator1& g1, const Generator2& g2, const Generator3& g3,
+ const Generator4& g4, const Generator5& g5, const Generator6& g6,
+ const Generator7& g7) {
+ return internal::CartesianProductHolder7<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6, Generator7>(
+ g1, g2, g3, g4, g5, g6, g7);
+}
+
+template <typename Generator1, typename Generator2, typename Generator3,
+ typename Generator4, typename Generator5, typename Generator6,
+ typename Generator7, typename Generator8>
+internal::CartesianProductHolder8<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6, Generator7, Generator8> Combine(
+ const Generator1& g1, const Generator2& g2, const Generator3& g3,
+ const Generator4& g4, const Generator5& g5, const Generator6& g6,
+ const Generator7& g7, const Generator8& g8) {
+ return internal::CartesianProductHolder8<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6, Generator7, Generator8>(
+ g1, g2, g3, g4, g5, g6, g7, g8);
+}
+
+template <typename Generator1, typename Generator2, typename Generator3,
+ typename Generator4, typename Generator5, typename Generator6,
+ typename Generator7, typename Generator8, typename Generator9>
+internal::CartesianProductHolder9<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6, Generator7, Generator8,
+ Generator9> Combine(
+ const Generator1& g1, const Generator2& g2, const Generator3& g3,
+ const Generator4& g4, const Generator5& g5, const Generator6& g6,
+ const Generator7& g7, const Generator8& g8, const Generator9& g9) {
+ return internal::CartesianProductHolder9<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6, Generator7, Generator8, Generator9>(
+ g1, g2, g3, g4, g5, g6, g7, g8, g9);
+}
+
+template <typename Generator1, typename Generator2, typename Generator3,
+ typename Generator4, typename Generator5, typename Generator6,
+ typename Generator7, typename Generator8, typename Generator9,
+ typename Generator10>
+internal::CartesianProductHolder10<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6, Generator7, Generator8, Generator9,
+ Generator10> Combine(
+ const Generator1& g1, const Generator2& g2, const Generator3& g3,
+ const Generator4& g4, const Generator5& g5, const Generator6& g6,
+ const Generator7& g7, const Generator8& g8, const Generator9& g9,
+ const Generator10& g10) {
+ return internal::CartesianProductHolder10<Generator1, Generator2, Generator3,
+ Generator4, Generator5, Generator6, Generator7, Generator8, Generator9,
+ Generator10>(
+ g1, g2, g3, g4, g5, g6, g7, g8, g9, g10);
+}
+# endif // GTEST_HAS_COMBINE
+
+
+
+# define TEST_P(test_case_name, test_name) \
+ class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \
+ : public test_case_name { \
+ public: \
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {} \
+ virtual void TestBody(); \
+ private: \
+ static int AddToRegistry() { \
+ ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \
+ GetTestCasePatternHolder<test_case_name>(\
+ #test_case_name, __FILE__, __LINE__)->AddTestPattern(\
+ #test_case_name, \
+ #test_name, \
+ new ::testing::internal::TestMetaFactory< \
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)>()); \
+ return 0; \
+ } \
+ static int gtest_registering_dummy_; \
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(\
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)); \
+ }; \
+ int GTEST_TEST_CLASS_NAME_(test_case_name, \
+ test_name)::gtest_registering_dummy_ = \
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::AddToRegistry(); \
+ void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody()
+
+# define INSTANTIATE_TEST_CASE_P(prefix, test_case_name, generator) \
+ ::testing::internal::ParamGenerator<test_case_name::ParamType> \
+ gtest_##prefix##test_case_name##_EvalGenerator_() { return generator; } \
+ int gtest_##prefix##test_case_name##_dummy_ = \
+ ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \
+ GetTestCasePatternHolder<test_case_name>(\
+ #test_case_name, __FILE__, __LINE__)->AddTestCaseInstantiation(\
+ #prefix, \
+ >est_##prefix##test_case_name##_EvalGenerator_, \
+ __FILE__, __LINE__)
+
+} // namespace testing
+
+#endif // GTEST_HAS_PARAM_TEST
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest-param-test.h.pump b/libvpx/third_party/googletest/src/include/gtest/gtest-param-test.h.pump
new file mode 100644
index 0000000..401cb51
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest-param-test.h.pump
@@ -0,0 +1,487 @@
+$$ -*- mode: c++; -*-
+$var n = 50 $$ Maximum length of Values arguments we want to support.
+$var maxtuple = 10 $$ Maximum number of Combine arguments we want to support.
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: vladl@google.com (Vlad Losev)
+//
+// Macros and functions for implementing parameterized tests
+// in Google C++ Testing Framework (Google Test)
+//
+// This file is generated by a SCRIPT. DO NOT EDIT BY HAND!
+//
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+
+
+// Value-parameterized tests allow you to test your code with different
+// parameters without writing multiple copies of the same test.
+//
+// Here is how you use value-parameterized tests:
+
+#if 0
+
+// To write value-parameterized tests, first you should define a fixture
+// class. It is usually derived from testing::TestWithParam<T> (see below for
+// another inheritance scheme that's sometimes useful in more complicated
+// class hierarchies), where the type of your parameter values.
+// TestWithParam<T> is itself derived from testing::Test. T can be any
+// copyable type. If it's a raw pointer, you are responsible for managing the
+// lifespan of the pointed values.
+
+class FooTest : public ::testing::TestWithParam<const char*> {
+ // You can implement all the usual class fixture members here.
+};
+
+// Then, use the TEST_P macro to define as many parameterized tests
+// for this fixture as you want. The _P suffix is for "parameterized"
+// or "pattern", whichever you prefer to think.
+
+TEST_P(FooTest, DoesBlah) {
+ // Inside a test, access the test parameter with the GetParam() method
+ // of the TestWithParam<T> class:
+ EXPECT_TRUE(foo.Blah(GetParam()));
+ ...
+}
+
+TEST_P(FooTest, HasBlahBlah) {
+ ...
+}
+
+// Finally, you can use INSTANTIATE_TEST_CASE_P to instantiate the test
+// case with any set of parameters you want. Google Test defines a number
+// of functions for generating test parameters. They return what we call
+// (surprise!) parameter generators. Here is a summary of them, which
+// are all in the testing namespace:
+//
+//
+// Range(begin, end [, step]) - Yields values {begin, begin+step,
+// begin+step+step, ...}. The values do not
+// include end. step defaults to 1.
+// Values(v1, v2, ..., vN) - Yields values {v1, v2, ..., vN}.
+// ValuesIn(container) - Yields values from a C-style array, an STL
+// ValuesIn(begin,end) container, or an iterator range [begin, end).
+// Bool() - Yields sequence {false, true}.
+// Combine(g1, g2, ..., gN) - Yields all combinations (the Cartesian product
+// for the math savvy) of the values generated
+// by the N generators.
+//
+// For more details, see comments at the definitions of these functions below
+// in this file.
+//
+// The following statement will instantiate tests from the FooTest test case
+// each with parameter values "meeny", "miny", and "moe".
+
+INSTANTIATE_TEST_CASE_P(InstantiationName,
+ FooTest,
+ Values("meeny", "miny", "moe"));
+
+// To distinguish different instances of the pattern, (yes, you
+// can instantiate it more then once) the first argument to the
+// INSTANTIATE_TEST_CASE_P macro is a prefix that will be added to the
+// actual test case name. Remember to pick unique prefixes for different
+// instantiations. The tests from the instantiation above will have
+// these names:
+//
+// * InstantiationName/FooTest.DoesBlah/0 for "meeny"
+// * InstantiationName/FooTest.DoesBlah/1 for "miny"
+// * InstantiationName/FooTest.DoesBlah/2 for "moe"
+// * InstantiationName/FooTest.HasBlahBlah/0 for "meeny"
+// * InstantiationName/FooTest.HasBlahBlah/1 for "miny"
+// * InstantiationName/FooTest.HasBlahBlah/2 for "moe"
+//
+// You can use these names in --gtest_filter.
+//
+// This statement will instantiate all tests from FooTest again, each
+// with parameter values "cat" and "dog":
+
+const char* pets[] = {"cat", "dog"};
+INSTANTIATE_TEST_CASE_P(AnotherInstantiationName, FooTest, ValuesIn(pets));
+
+// The tests from the instantiation above will have these names:
+//
+// * AnotherInstantiationName/FooTest.DoesBlah/0 for "cat"
+// * AnotherInstantiationName/FooTest.DoesBlah/1 for "dog"
+// * AnotherInstantiationName/FooTest.HasBlahBlah/0 for "cat"
+// * AnotherInstantiationName/FooTest.HasBlahBlah/1 for "dog"
+//
+// Please note that INSTANTIATE_TEST_CASE_P will instantiate all tests
+// in the given test case, whether their definitions come before or
+// AFTER the INSTANTIATE_TEST_CASE_P statement.
+//
+// Please also note that generator expressions (including parameters to the
+// generators) are evaluated in InitGoogleTest(), after main() has started.
+// This allows the user on one hand, to adjust generator parameters in order
+// to dynamically determine a set of tests to run and on the other hand,
+// give the user a chance to inspect the generated tests with Google Test
+// reflection API before RUN_ALL_TESTS() is executed.
+//
+// You can see samples/sample7_unittest.cc and samples/sample8_unittest.cc
+// for more examples.
+//
+// In the future, we plan to publish the API for defining new parameter
+// generators. But for now this interface remains part of the internal
+// implementation and is subject to change.
+//
+//
+// A parameterized test fixture must be derived from testing::Test and from
+// testing::WithParamInterface<T>, where T is the type of the parameter
+// values. Inheriting from TestWithParam<T> satisfies that requirement because
+// TestWithParam<T> inherits from both Test and WithParamInterface. In more
+// complicated hierarchies, however, it is occasionally useful to inherit
+// separately from Test and WithParamInterface. For example:
+
+class BaseTest : public ::testing::Test {
+ // You can inherit all the usual members for a non-parameterized test
+ // fixture here.
+};
+
+class DerivedTest : public BaseTest, public ::testing::WithParamInterface<int> {
+ // The usual test fixture members go here too.
+};
+
+TEST_F(BaseTest, HasFoo) {
+ // This is an ordinary non-parameterized test.
+}
+
+TEST_P(DerivedTest, DoesBlah) {
+ // GetParam works just the same here as if you inherit from TestWithParam.
+ EXPECT_TRUE(foo.Blah(GetParam()));
+}
+
+#endif // 0
+
+#include "gtest/internal/gtest-port.h"
+
+#if !GTEST_OS_SYMBIAN
+# include <utility>
+#endif
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*. Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+#include "gtest/internal/gtest-internal.h"
+#include "gtest/internal/gtest-param-util.h"
+#include "gtest/internal/gtest-param-util-generated.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing {
+
+// Functions producing parameter generators.
+//
+// Google Test uses these generators to produce parameters for value-
+// parameterized tests. When a parameterized test case is instantiated
+// with a particular generator, Google Test creates and runs tests
+// for each element in the sequence produced by the generator.
+//
+// In the following sample, tests from test case FooTest are instantiated
+// each three times with parameter values 3, 5, and 8:
+//
+// class FooTest : public TestWithParam<int> { ... };
+//
+// TEST_P(FooTest, TestThis) {
+// }
+// TEST_P(FooTest, TestThat) {
+// }
+// INSTANTIATE_TEST_CASE_P(TestSequence, FooTest, Values(3, 5, 8));
+//
+
+// Range() returns generators providing sequences of values in a range.
+//
+// Synopsis:
+// Range(start, end)
+// - returns a generator producing a sequence of values {start, start+1,
+// start+2, ..., }.
+// Range(start, end, step)
+// - returns a generator producing a sequence of values {start, start+step,
+// start+step+step, ..., }.
+// Notes:
+// * The generated sequences never include end. For example, Range(1, 5)
+// returns a generator producing a sequence {1, 2, 3, 4}. Range(1, 9, 2)
+// returns a generator producing {1, 3, 5, 7}.
+// * start and end must have the same type. That type may be any integral or
+// floating-point type or a user defined type satisfying these conditions:
+// * It must be assignable (have operator=() defined).
+// * It must have operator+() (operator+(int-compatible type) for
+// two-operand version).
+// * It must have operator<() defined.
+// Elements in the resulting sequences will also have that type.
+// * Condition start < end must be satisfied in order for resulting sequences
+// to contain any elements.
+//
+template <typename T, typename IncrementT>
+internal::ParamGenerator<T> Range(T start, T end, IncrementT step) {
+ return internal::ParamGenerator<T>(
+ new internal::RangeGenerator<T, IncrementT>(start, end, step));
+}
+
+template <typename T>
+internal::ParamGenerator<T> Range(T start, T end) {
+ return Range(start, end, 1);
+}
+
+// ValuesIn() function allows generation of tests with parameters coming from
+// a container.
+//
+// Synopsis:
+// ValuesIn(const T (&array)[N])
+// - returns a generator producing sequences with elements from
+// a C-style array.
+// ValuesIn(const Container& container)
+// - returns a generator producing sequences with elements from
+// an STL-style container.
+// ValuesIn(Iterator begin, Iterator end)
+// - returns a generator producing sequences with elements from
+// a range [begin, end) defined by a pair of STL-style iterators. These
+// iterators can also be plain C pointers.
+//
+// Please note that ValuesIn copies the values from the containers
+// passed in and keeps them to generate tests in RUN_ALL_TESTS().
+//
+// Examples:
+//
+// This instantiates tests from test case StringTest
+// each with C-string values of "foo", "bar", and "baz":
+//
+// const char* strings[] = {"foo", "bar", "baz"};
+// INSTANTIATE_TEST_CASE_P(StringSequence, SrtingTest, ValuesIn(strings));
+//
+// This instantiates tests from test case StlStringTest
+// each with STL strings with values "a" and "b":
+//
+// ::std::vector< ::std::string> GetParameterStrings() {
+// ::std::vector< ::std::string> v;
+// v.push_back("a");
+// v.push_back("b");
+// return v;
+// }
+//
+// INSTANTIATE_TEST_CASE_P(CharSequence,
+// StlStringTest,
+// ValuesIn(GetParameterStrings()));
+//
+//
+// This will also instantiate tests from CharTest
+// each with parameter values 'a' and 'b':
+//
+// ::std::list<char> GetParameterChars() {
+// ::std::list<char> list;
+// list.push_back('a');
+// list.push_back('b');
+// return list;
+// }
+// ::std::list<char> l = GetParameterChars();
+// INSTANTIATE_TEST_CASE_P(CharSequence2,
+// CharTest,
+// ValuesIn(l.begin(), l.end()));
+//
+template <typename ForwardIterator>
+internal::ParamGenerator<
+ typename ::testing::internal::IteratorTraits<ForwardIterator>::value_type>
+ValuesIn(ForwardIterator begin, ForwardIterator end) {
+ typedef typename ::testing::internal::IteratorTraits<ForwardIterator>
+ ::value_type ParamType;
+ return internal::ParamGenerator<ParamType>(
+ new internal::ValuesInIteratorRangeGenerator<ParamType>(begin, end));
+}
+
+template <typename T, size_t N>
+internal::ParamGenerator<T> ValuesIn(const T (&array)[N]) {
+ return ValuesIn(array, array + N);
+}
+
+template <class Container>
+internal::ParamGenerator<typename Container::value_type> ValuesIn(
+ const Container& container) {
+ return ValuesIn(container.begin(), container.end());
+}
+
+// Values() allows generating tests from explicitly specified list of
+// parameters.
+//
+// Synopsis:
+// Values(T v1, T v2, ..., T vN)
+// - returns a generator producing sequences with elements v1, v2, ..., vN.
+//
+// For example, this instantiates tests from test case BarTest each
+// with values "one", "two", and "three":
+//
+// INSTANTIATE_TEST_CASE_P(NumSequence, BarTest, Values("one", "two", "three"));
+//
+// This instantiates tests from test case BazTest each with values 1, 2, 3.5.
+// The exact type of values will depend on the type of parameter in BazTest.
+//
+// INSTANTIATE_TEST_CASE_P(FloatingNumbers, BazTest, Values(1, 2, 3.5));
+//
+// Currently, Values() supports from 1 to $n parameters.
+//
+$range i 1..n
+$for i [[
+$range j 1..i
+
+template <$for j, [[typename T$j]]>
+internal::ValueArray$i<$for j, [[T$j]]> Values($for j, [[T$j v$j]]) {
+ return internal::ValueArray$i<$for j, [[T$j]]>($for j, [[v$j]]);
+}
+
+]]
+
+// Bool() allows generating tests with parameters in a set of (false, true).
+//
+// Synopsis:
+// Bool()
+// - returns a generator producing sequences with elements {false, true}.
+//
+// It is useful when testing code that depends on Boolean flags. Combinations
+// of multiple flags can be tested when several Bool()'s are combined using
+// Combine() function.
+//
+// In the following example all tests in the test case FlagDependentTest
+// will be instantiated twice with parameters false and true.
+//
+// class FlagDependentTest : public testing::TestWithParam<bool> {
+// virtual void SetUp() {
+// external_flag = GetParam();
+// }
+// }
+// INSTANTIATE_TEST_CASE_P(BoolSequence, FlagDependentTest, Bool());
+//
+inline internal::ParamGenerator<bool> Bool() {
+ return Values(false, true);
+}
+
+# if GTEST_HAS_COMBINE
+// Combine() allows the user to combine two or more sequences to produce
+// values of a Cartesian product of those sequences' elements.
+//
+// Synopsis:
+// Combine(gen1, gen2, ..., genN)
+// - returns a generator producing sequences with elements coming from
+// the Cartesian product of elements from the sequences generated by
+// gen1, gen2, ..., genN. The sequence elements will have a type of
+// tuple<T1, T2, ..., TN> where T1, T2, ..., TN are the types
+// of elements from sequences produces by gen1, gen2, ..., genN.
+//
+// Combine can have up to $maxtuple arguments. This number is currently limited
+// by the maximum number of elements in the tuple implementation used by Google
+// Test.
+//
+// Example:
+//
+// This will instantiate tests in test case AnimalTest each one with
+// the parameter values tuple("cat", BLACK), tuple("cat", WHITE),
+// tuple("dog", BLACK), and tuple("dog", WHITE):
+//
+// enum Color { BLACK, GRAY, WHITE };
+// class AnimalTest
+// : public testing::TestWithParam<tuple<const char*, Color> > {...};
+//
+// TEST_P(AnimalTest, AnimalLooksNice) {...}
+//
+// INSTANTIATE_TEST_CASE_P(AnimalVariations, AnimalTest,
+// Combine(Values("cat", "dog"),
+// Values(BLACK, WHITE)));
+//
+// This will instantiate tests in FlagDependentTest with all variations of two
+// Boolean flags:
+//
+// class FlagDependentTest
+// : public testing::TestWithParam<tuple(bool, bool)> > {
+// virtual void SetUp() {
+// // Assigns external_flag_1 and external_flag_2 values from the tuple.
+// tie(external_flag_1, external_flag_2) = GetParam();
+// }
+// };
+//
+// TEST_P(FlagDependentTest, TestFeature1) {
+// // Test your code using external_flag_1 and external_flag_2 here.
+// }
+// INSTANTIATE_TEST_CASE_P(TwoBoolSequence, FlagDependentTest,
+// Combine(Bool(), Bool()));
+//
+$range i 2..maxtuple
+$for i [[
+$range j 1..i
+
+template <$for j, [[typename Generator$j]]>
+internal::CartesianProductHolder$i<$for j, [[Generator$j]]> Combine(
+ $for j, [[const Generator$j& g$j]]) {
+ return internal::CartesianProductHolder$i<$for j, [[Generator$j]]>(
+ $for j, [[g$j]]);
+}
+
+]]
+# endif // GTEST_HAS_COMBINE
+
+
+
+# define TEST_P(test_case_name, test_name) \
+ class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \
+ : public test_case_name { \
+ public: \
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {} \
+ virtual void TestBody(); \
+ private: \
+ static int AddToRegistry() { \
+ ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \
+ GetTestCasePatternHolder<test_case_name>(\
+ #test_case_name, __FILE__, __LINE__)->AddTestPattern(\
+ #test_case_name, \
+ #test_name, \
+ new ::testing::internal::TestMetaFactory< \
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)>()); \
+ return 0; \
+ } \
+ static int gtest_registering_dummy_; \
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(\
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)); \
+ }; \
+ int GTEST_TEST_CLASS_NAME_(test_case_name, \
+ test_name)::gtest_registering_dummy_ = \
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::AddToRegistry(); \
+ void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody()
+
+# define INSTANTIATE_TEST_CASE_P(prefix, test_case_name, generator) \
+ ::testing::internal::ParamGenerator<test_case_name::ParamType> \
+ gtest_##prefix##test_case_name##_EvalGenerator_() { return generator; } \
+ int gtest_##prefix##test_case_name##_dummy_ = \
+ ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \
+ GetTestCasePatternHolder<test_case_name>(\
+ #test_case_name, __FILE__, __LINE__)->AddTestCaseInstantiation(\
+ #prefix, \
+ >est_##prefix##test_case_name##_EvalGenerator_, \
+ __FILE__, __LINE__)
+
+} // namespace testing
+
+#endif // GTEST_HAS_PARAM_TEST
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest-printers.h b/libvpx/third_party/googletest/src/include/gtest/gtest-printers.h
new file mode 100644
index 0000000..9cbab3f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest-printers.h
@@ -0,0 +1,796 @@
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Google Test - The Google C++ Testing Framework
+//
+// This file implements a universal value printer that can print a
+// value of any type T:
+//
+// void ::testing::internal::UniversalPrinter<T>::Print(value, ostream_ptr);
+//
+// A user can teach this function how to print a class type T by
+// defining either operator<<() or PrintTo() in the namespace that
+// defines T. More specifically, the FIRST defined function in the
+// following list will be used (assuming T is defined in namespace
+// foo):
+//
+// 1. foo::PrintTo(const T&, ostream*)
+// 2. operator<<(ostream&, const T&) defined in either foo or the
+// global namespace.
+//
+// If none of the above is defined, it will print the debug string of
+// the value if it is a protocol buffer, or print the raw bytes in the
+// value otherwise.
+//
+// To aid debugging: when T is a reference type, the address of the
+// value is also printed; when T is a (const) char pointer, both the
+// pointer value and the NUL-terminated string it points to are
+// printed.
+//
+// We also provide some convenient wrappers:
+//
+// // Prints a value to a string. For a (const or not) char
+// // pointer, the NUL-terminated string (but not the pointer) is
+// // printed.
+// std::string ::testing::PrintToString(const T& value);
+//
+// // Prints a value tersely: for a reference type, the referenced
+// // value (but not the address) is printed; for a (const or not) char
+// // pointer, the NUL-terminated string (but not the pointer) is
+// // printed.
+// void ::testing::internal::UniversalTersePrint(const T& value, ostream*);
+//
+// // Prints value using the type inferred by the compiler. The difference
+// // from UniversalTersePrint() is that this function prints both the
+// // pointer and the NUL-terminated string for a (const or not) char pointer.
+// void ::testing::internal::UniversalPrint(const T& value, ostream*);
+//
+// // Prints the fields of a tuple tersely to a string vector, one
+// // element for each field. Tuple support must be enabled in
+// // gtest-port.h.
+// std::vector<string> UniversalTersePrintTupleFieldsToStrings(
+// const Tuple& value);
+//
+// Known limitation:
+//
+// The print primitives print the elements of an STL-style container
+// using the compiler-inferred type of *iter where iter is a
+// const_iterator of the container. When const_iterator is an input
+// iterator but not a forward iterator, this inferred type may not
+// match value_type, and the print output may be incorrect. In
+// practice, this is rarely a problem as for most containers
+// const_iterator is a forward iterator. We'll fix this if there's an
+// actual need for it. Note that this fix cannot rely on value_type
+// being defined as many user-defined container types don't have
+// value_type.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
+
+#include <ostream> // NOLINT
+#include <sstream>
+#include <string>
+#include <utility>
+#include <vector>
+#include "gtest/internal/gtest-port.h"
+#include "gtest/internal/gtest-internal.h"
+
+namespace testing {
+
+// Definitions in the 'internal' and 'internal2' name spaces are
+// subject to change without notice. DO NOT USE THEM IN USER CODE!
+namespace internal2 {
+
+// Prints the given number of bytes in the given object to the given
+// ostream.
+GTEST_API_ void PrintBytesInObjectTo(const unsigned char* obj_bytes,
+ size_t count,
+ ::std::ostream* os);
+
+// For selecting which printer to use when a given type has neither <<
+// nor PrintTo().
+enum TypeKind {
+ kProtobuf, // a protobuf type
+ kConvertibleToInteger, // a type implicitly convertible to BiggestInt
+ // (e.g. a named or unnamed enum type)
+ kOtherType // anything else
+};
+
+// TypeWithoutFormatter<T, kTypeKind>::PrintValue(value, os) is called
+// by the universal printer to print a value of type T when neither
+// operator<< nor PrintTo() is defined for T, where kTypeKind is the
+// "kind" of T as defined by enum TypeKind.
+template <typename T, TypeKind kTypeKind>
+class TypeWithoutFormatter {
+ public:
+ // This default version is called when kTypeKind is kOtherType.
+ static void PrintValue(const T& value, ::std::ostream* os) {
+ PrintBytesInObjectTo(reinterpret_cast<const unsigned char*>(&value),
+ sizeof(value), os);
+ }
+};
+
+// We print a protobuf using its ShortDebugString() when the string
+// doesn't exceed this many characters; otherwise we print it using
+// DebugString() for better readability.
+const size_t kProtobufOneLinerMaxLength = 50;
+
+template <typename T>
+class TypeWithoutFormatter<T, kProtobuf> {
+ public:
+ static void PrintValue(const T& value, ::std::ostream* os) {
+ const ::testing::internal::string short_str = value.ShortDebugString();
+ const ::testing::internal::string pretty_str =
+ short_str.length() <= kProtobufOneLinerMaxLength ?
+ short_str : ("\n" + value.DebugString());
+ *os << ("<" + pretty_str + ">");
+ }
+};
+
+template <typename T>
+class TypeWithoutFormatter<T, kConvertibleToInteger> {
+ public:
+ // Since T has no << operator or PrintTo() but can be implicitly
+ // converted to BiggestInt, we print it as a BiggestInt.
+ //
+ // Most likely T is an enum type (either named or unnamed), in which
+ // case printing it as an integer is the desired behavior. In case
+ // T is not an enum, printing it as an integer is the best we can do
+ // given that it has no user-defined printer.
+ static void PrintValue(const T& value, ::std::ostream* os) {
+ const internal::BiggestInt kBigInt = value;
+ *os << kBigInt;
+ }
+};
+
+// Prints the given value to the given ostream. If the value is a
+// protocol message, its debug string is printed; if it's an enum or
+// of a type implicitly convertible to BiggestInt, it's printed as an
+// integer; otherwise the bytes in the value are printed. This is
+// what UniversalPrinter<T>::Print() does when it knows nothing about
+// type T and T has neither << operator nor PrintTo().
+//
+// A user can override this behavior for a class type Foo by defining
+// a << operator in the namespace where Foo is defined.
+//
+// We put this operator in namespace 'internal2' instead of 'internal'
+// to simplify the implementation, as much code in 'internal' needs to
+// use << in STL, which would conflict with our own << were it defined
+// in 'internal'.
+//
+// Note that this operator<< takes a generic std::basic_ostream<Char,
+// CharTraits> type instead of the more restricted std::ostream. If
+// we define it to take an std::ostream instead, we'll get an
+// "ambiguous overloads" compiler error when trying to print a type
+// Foo that supports streaming to std::basic_ostream<Char,
+// CharTraits>, as the compiler cannot tell whether
+// operator<<(std::ostream&, const T&) or
+// operator<<(std::basic_stream<Char, CharTraits>, const Foo&) is more
+// specific.
+template <typename Char, typename CharTraits, typename T>
+::std::basic_ostream<Char, CharTraits>& operator<<(
+ ::std::basic_ostream<Char, CharTraits>& os, const T& x) {
+ TypeWithoutFormatter<T,
+ (internal::IsAProtocolMessage<T>::value ? kProtobuf :
+ internal::ImplicitlyConvertible<const T&, internal::BiggestInt>::value ?
+ kConvertibleToInteger : kOtherType)>::PrintValue(x, &os);
+ return os;
+}
+
+} // namespace internal2
+} // namespace testing
+
+// This namespace MUST NOT BE NESTED IN ::testing, or the name look-up
+// magic needed for implementing UniversalPrinter won't work.
+namespace testing_internal {
+
+// Used to print a value that is not an STL-style container when the
+// user doesn't define PrintTo() for it.
+template <typename T>
+void DefaultPrintNonContainerTo(const T& value, ::std::ostream* os) {
+ // With the following statement, during unqualified name lookup,
+ // testing::internal2::operator<< appears as if it was declared in
+ // the nearest enclosing namespace that contains both
+ // ::testing_internal and ::testing::internal2, i.e. the global
+ // namespace. For more details, refer to the C++ Standard section
+ // 7.3.4-1 [namespace.udir]. This allows us to fall back onto
+ // testing::internal2::operator<< in case T doesn't come with a <<
+ // operator.
+ //
+ // We cannot write 'using ::testing::internal2::operator<<;', which
+ // gcc 3.3 fails to compile due to a compiler bug.
+ using namespace ::testing::internal2; // NOLINT
+
+ // Assuming T is defined in namespace foo, in the next statement,
+ // the compiler will consider all of:
+ //
+ // 1. foo::operator<< (thanks to Koenig look-up),
+ // 2. ::operator<< (as the current namespace is enclosed in ::),
+ // 3. testing::internal2::operator<< (thanks to the using statement above).
+ //
+ // The operator<< whose type matches T best will be picked.
+ //
+ // We deliberately allow #2 to be a candidate, as sometimes it's
+ // impossible to define #1 (e.g. when foo is ::std, defining
+ // anything in it is undefined behavior unless you are a compiler
+ // vendor.).
+ *os << value;
+}
+
+} // namespace testing_internal
+
+namespace testing {
+namespace internal {
+
+// UniversalPrinter<T>::Print(value, ostream_ptr) prints the given
+// value to the given ostream. The caller must ensure that
+// 'ostream_ptr' is not NULL, or the behavior is undefined.
+//
+// We define UniversalPrinter as a class template (as opposed to a
+// function template), as we need to partially specialize it for
+// reference types, which cannot be done with function templates.
+template <typename T>
+class UniversalPrinter;
+
+template <typename T>
+void UniversalPrint(const T& value, ::std::ostream* os);
+
+// Used to print an STL-style container when the user doesn't define
+// a PrintTo() for it.
+template <typename C>
+void DefaultPrintTo(IsContainer /* dummy */,
+ false_type /* is not a pointer */,
+ const C& container, ::std::ostream* os) {
+ const size_t kMaxCount = 32; // The maximum number of elements to print.
+ *os << '{';
+ size_t count = 0;
+ for (typename C::const_iterator it = container.begin();
+ it != container.end(); ++it, ++count) {
+ if (count > 0) {
+ *os << ',';
+ if (count == kMaxCount) { // Enough has been printed.
+ *os << " ...";
+ break;
+ }
+ }
+ *os << ' ';
+ // We cannot call PrintTo(*it, os) here as PrintTo() doesn't
+ // handle *it being a native array.
+ internal::UniversalPrint(*it, os);
+ }
+
+ if (count > 0) {
+ *os << ' ';
+ }
+ *os << '}';
+}
+
+// Used to print a pointer that is neither a char pointer nor a member
+// pointer, when the user doesn't define PrintTo() for it. (A member
+// variable pointer or member function pointer doesn't really point to
+// a location in the address space. Their representation is
+// implementation-defined. Therefore they will be printed as raw
+// bytes.)
+template <typename T>
+void DefaultPrintTo(IsNotContainer /* dummy */,
+ true_type /* is a pointer */,
+ T* p, ::std::ostream* os) {
+ if (p == NULL) {
+ *os << "NULL";
+ } else {
+ // C++ doesn't allow casting from a function pointer to any object
+ // pointer.
+ //
+ // IsTrue() silences warnings: "Condition is always true",
+ // "unreachable code".
+ if (IsTrue(ImplicitlyConvertible<T*, const void*>::value)) {
+ // T is not a function type. We just call << to print p,
+ // relying on ADL to pick up user-defined << for their pointer
+ // types, if any.
+ *os << p;
+ } else {
+ // T is a function type, so '*os << p' doesn't do what we want
+ // (it just prints p as bool). We want to print p as a const
+ // void*. However, we cannot cast it to const void* directly,
+ // even using reinterpret_cast, as earlier versions of gcc
+ // (e.g. 3.4.5) cannot compile the cast when p is a function
+ // pointer. Casting to UInt64 first solves the problem.
+ *os << reinterpret_cast<const void*>(
+ reinterpret_cast<internal::UInt64>(p));
+ }
+ }
+}
+
+// Used to print a non-container, non-pointer value when the user
+// doesn't define PrintTo() for it.
+template <typename T>
+void DefaultPrintTo(IsNotContainer /* dummy */,
+ false_type /* is not a pointer */,
+ const T& value, ::std::ostream* os) {
+ ::testing_internal::DefaultPrintNonContainerTo(value, os);
+}
+
+// Prints the given value using the << operator if it has one;
+// otherwise prints the bytes in it. This is what
+// UniversalPrinter<T>::Print() does when PrintTo() is not specialized
+// or overloaded for type T.
+//
+// A user can override this behavior for a class type Foo by defining
+// an overload of PrintTo() in the namespace where Foo is defined. We
+// give the user this option as sometimes defining a << operator for
+// Foo is not desirable (e.g. the coding style may prevent doing it,
+// or there is already a << operator but it doesn't do what the user
+// wants).
+template <typename T>
+void PrintTo(const T& value, ::std::ostream* os) {
+ // DefaultPrintTo() is overloaded. The type of its first two
+ // arguments determine which version will be picked. If T is an
+ // STL-style container, the version for container will be called; if
+ // T is a pointer, the pointer version will be called; otherwise the
+ // generic version will be called.
+ //
+ // Note that we check for container types here, prior to we check
+ // for protocol message types in our operator<<. The rationale is:
+ //
+ // For protocol messages, we want to give people a chance to
+ // override Google Mock's format by defining a PrintTo() or
+ // operator<<. For STL containers, other formats can be
+ // incompatible with Google Mock's format for the container
+ // elements; therefore we check for container types here to ensure
+ // that our format is used.
+ //
+ // The second argument of DefaultPrintTo() is needed to bypass a bug
+ // in Symbian's C++ compiler that prevents it from picking the right
+ // overload between:
+ //
+ // PrintTo(const T& x, ...);
+ // PrintTo(T* x, ...);
+ DefaultPrintTo(IsContainerTest<T>(0), is_pointer<T>(), value, os);
+}
+
+// The following list of PrintTo() overloads tells
+// UniversalPrinter<T>::Print() how to print standard types (built-in
+// types, strings, plain arrays, and pointers).
+
+// Overloads for various char types.
+GTEST_API_ void PrintTo(unsigned char c, ::std::ostream* os);
+GTEST_API_ void PrintTo(signed char c, ::std::ostream* os);
+inline void PrintTo(char c, ::std::ostream* os) {
+ // When printing a plain char, we always treat it as unsigned. This
+ // way, the output won't be affected by whether the compiler thinks
+ // char is signed or not.
+ PrintTo(static_cast<unsigned char>(c), os);
+}
+
+// Overloads for other simple built-in types.
+inline void PrintTo(bool x, ::std::ostream* os) {
+ *os << (x ? "true" : "false");
+}
+
+// Overload for wchar_t type.
+// Prints a wchar_t as a symbol if it is printable or as its internal
+// code otherwise and also as its decimal code (except for L'\0').
+// The L'\0' char is printed as "L'\\0'". The decimal code is printed
+// as signed integer when wchar_t is implemented by the compiler
+// as a signed type and is printed as an unsigned integer when wchar_t
+// is implemented as an unsigned type.
+GTEST_API_ void PrintTo(wchar_t wc, ::std::ostream* os);
+
+// Overloads for C strings.
+GTEST_API_ void PrintTo(const char* s, ::std::ostream* os);
+inline void PrintTo(char* s, ::std::ostream* os) {
+ PrintTo(ImplicitCast_<const char*>(s), os);
+}
+
+// signed/unsigned char is often used for representing binary data, so
+// we print pointers to it as void* to be safe.
+inline void PrintTo(const signed char* s, ::std::ostream* os) {
+ PrintTo(ImplicitCast_<const void*>(s), os);
+}
+inline void PrintTo(signed char* s, ::std::ostream* os) {
+ PrintTo(ImplicitCast_<const void*>(s), os);
+}
+inline void PrintTo(const unsigned char* s, ::std::ostream* os) {
+ PrintTo(ImplicitCast_<const void*>(s), os);
+}
+inline void PrintTo(unsigned char* s, ::std::ostream* os) {
+ PrintTo(ImplicitCast_<const void*>(s), os);
+}
+
+// MSVC can be configured to define wchar_t as a typedef of unsigned
+// short. It defines _NATIVE_WCHAR_T_DEFINED when wchar_t is a native
+// type. When wchar_t is a typedef, defining an overload for const
+// wchar_t* would cause unsigned short* be printed as a wide string,
+// possibly causing invalid memory accesses.
+#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED)
+// Overloads for wide C strings
+GTEST_API_ void PrintTo(const wchar_t* s, ::std::ostream* os);
+inline void PrintTo(wchar_t* s, ::std::ostream* os) {
+ PrintTo(ImplicitCast_<const wchar_t*>(s), os);
+}
+#endif
+
+// Overload for C arrays. Multi-dimensional arrays are printed
+// properly.
+
+// Prints the given number of elements in an array, without printing
+// the curly braces.
+template <typename T>
+void PrintRawArrayTo(const T a[], size_t count, ::std::ostream* os) {
+ UniversalPrint(a[0], os);
+ for (size_t i = 1; i != count; i++) {
+ *os << ", ";
+ UniversalPrint(a[i], os);
+ }
+}
+
+// Overloads for ::string and ::std::string.
+#if GTEST_HAS_GLOBAL_STRING
+GTEST_API_ void PrintStringTo(const ::string&s, ::std::ostream* os);
+inline void PrintTo(const ::string& s, ::std::ostream* os) {
+ PrintStringTo(s, os);
+}
+#endif // GTEST_HAS_GLOBAL_STRING
+
+GTEST_API_ void PrintStringTo(const ::std::string&s, ::std::ostream* os);
+inline void PrintTo(const ::std::string& s, ::std::ostream* os) {
+ PrintStringTo(s, os);
+}
+
+// Overloads for ::wstring and ::std::wstring.
+#if GTEST_HAS_GLOBAL_WSTRING
+GTEST_API_ void PrintWideStringTo(const ::wstring&s, ::std::ostream* os);
+inline void PrintTo(const ::wstring& s, ::std::ostream* os) {
+ PrintWideStringTo(s, os);
+}
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+#if GTEST_HAS_STD_WSTRING
+GTEST_API_ void PrintWideStringTo(const ::std::wstring&s, ::std::ostream* os);
+inline void PrintTo(const ::std::wstring& s, ::std::ostream* os) {
+ PrintWideStringTo(s, os);
+}
+#endif // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_TR1_TUPLE
+// Overload for ::std::tr1::tuple. Needed for printing function arguments,
+// which are packed as tuples.
+
+// Helper function for printing a tuple. T must be instantiated with
+// a tuple type.
+template <typename T>
+void PrintTupleTo(const T& t, ::std::ostream* os);
+
+// Overloaded PrintTo() for tuples of various arities. We support
+// tuples of up-to 10 fields. The following implementation works
+// regardless of whether tr1::tuple is implemented using the
+// non-standard variadic template feature or not.
+
+inline void PrintTo(const ::std::tr1::tuple<>& t, ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1>
+void PrintTo(const ::std::tr1::tuple<T1>& t, ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2>
+void PrintTo(const ::std::tr1::tuple<T1, T2>& t, ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2, typename T3>
+void PrintTo(const ::std::tr1::tuple<T1, T2, T3>& t, ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2, typename T3, typename T4>
+void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4>& t, ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5>
+void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5>& t,
+ ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6>
+void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6>& t,
+ ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7>
+void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7>& t,
+ ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8>
+void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8>& t,
+ ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9>
+void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9>& t,
+ ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10>
+void PrintTo(
+ const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10>& t,
+ ::std::ostream* os) {
+ PrintTupleTo(t, os);
+}
+#endif // GTEST_HAS_TR1_TUPLE
+
+// Overload for std::pair.
+template <typename T1, typename T2>
+void PrintTo(const ::std::pair<T1, T2>& value, ::std::ostream* os) {
+ *os << '(';
+ // We cannot use UniversalPrint(value.first, os) here, as T1 may be
+ // a reference type. The same for printing value.second.
+ UniversalPrinter<T1>::Print(value.first, os);
+ *os << ", ";
+ UniversalPrinter<T2>::Print(value.second, os);
+ *os << ')';
+}
+
+// Implements printing a non-reference type T by letting the compiler
+// pick the right overload of PrintTo() for T.
+template <typename T>
+class UniversalPrinter {
+ public:
+ // MSVC warns about adding const to a function type, so we want to
+ // disable the warning.
+#ifdef _MSC_VER
+# pragma warning(push) // Saves the current warning state.
+# pragma warning(disable:4180) // Temporarily disables warning 4180.
+#endif // _MSC_VER
+
+ // Note: we deliberately don't call this PrintTo(), as that name
+ // conflicts with ::testing::internal::PrintTo in the body of the
+ // function.
+ static void Print(const T& value, ::std::ostream* os) {
+ // By default, ::testing::internal::PrintTo() is used for printing
+ // the value.
+ //
+ // Thanks to Koenig look-up, if T is a class and has its own
+ // PrintTo() function defined in its namespace, that function will
+ // be visible here. Since it is more specific than the generic ones
+ // in ::testing::internal, it will be picked by the compiler in the
+ // following statement - exactly what we want.
+ PrintTo(value, os);
+ }
+
+#ifdef _MSC_VER
+# pragma warning(pop) // Restores the warning state.
+#endif // _MSC_VER
+};
+
+// UniversalPrintArray(begin, len, os) prints an array of 'len'
+// elements, starting at address 'begin'.
+template <typename T>
+void UniversalPrintArray(const T* begin, size_t len, ::std::ostream* os) {
+ if (len == 0) {
+ *os << "{}";
+ } else {
+ *os << "{ ";
+ const size_t kThreshold = 18;
+ const size_t kChunkSize = 8;
+ // If the array has more than kThreshold elements, we'll have to
+ // omit some details by printing only the first and the last
+ // kChunkSize elements.
+ // TODO(wan@google.com): let the user control the threshold using a flag.
+ if (len <= kThreshold) {
+ PrintRawArrayTo(begin, len, os);
+ } else {
+ PrintRawArrayTo(begin, kChunkSize, os);
+ *os << ", ..., ";
+ PrintRawArrayTo(begin + len - kChunkSize, kChunkSize, os);
+ }
+ *os << " }";
+ }
+}
+// This overload prints a (const) char array compactly.
+GTEST_API_ void UniversalPrintArray(const char* begin,
+ size_t len,
+ ::std::ostream* os);
+
+// Implements printing an array type T[N].
+template <typename T, size_t N>
+class UniversalPrinter<T[N]> {
+ public:
+ // Prints the given array, omitting some elements when there are too
+ // many.
+ static void Print(const T (&a)[N], ::std::ostream* os) {
+ UniversalPrintArray(a, N, os);
+ }
+};
+
+// Implements printing a reference type T&.
+template <typename T>
+class UniversalPrinter<T&> {
+ public:
+ // MSVC warns about adding const to a function type, so we want to
+ // disable the warning.
+#ifdef _MSC_VER
+# pragma warning(push) // Saves the current warning state.
+# pragma warning(disable:4180) // Temporarily disables warning 4180.
+#endif // _MSC_VER
+
+ static void Print(const T& value, ::std::ostream* os) {
+ // Prints the address of the value. We use reinterpret_cast here
+ // as static_cast doesn't compile when T is a function type.
+ *os << "@" << reinterpret_cast<const void*>(&value) << " ";
+
+ // Then prints the value itself.
+ UniversalPrint(value, os);
+ }
+
+#ifdef _MSC_VER
+# pragma warning(pop) // Restores the warning state.
+#endif // _MSC_VER
+};
+
+// Prints a value tersely: for a reference type, the referenced value
+// (but not the address) is printed; for a (const) char pointer, the
+// NUL-terminated string (but not the pointer) is printed.
+template <typename T>
+void UniversalTersePrint(const T& value, ::std::ostream* os) {
+ UniversalPrint(value, os);
+}
+inline void UniversalTersePrint(const char* str, ::std::ostream* os) {
+ if (str == NULL) {
+ *os << "NULL";
+ } else {
+ UniversalPrint(string(str), os);
+ }
+}
+inline void UniversalTersePrint(char* str, ::std::ostream* os) {
+ UniversalTersePrint(static_cast<const char*>(str), os);
+}
+
+// Prints a value using the type inferred by the compiler. The
+// difference between this and UniversalTersePrint() is that for a
+// (const) char pointer, this prints both the pointer and the
+// NUL-terminated string.
+template <typename T>
+void UniversalPrint(const T& value, ::std::ostream* os) {
+ UniversalPrinter<T>::Print(value, os);
+}
+
+#if GTEST_HAS_TR1_TUPLE
+typedef ::std::vector<string> Strings;
+
+// This helper template allows PrintTo() for tuples and
+// UniversalTersePrintTupleFieldsToStrings() to be defined by
+// induction on the number of tuple fields. The idea is that
+// TuplePrefixPrinter<N>::PrintPrefixTo(t, os) prints the first N
+// fields in tuple t, and can be defined in terms of
+// TuplePrefixPrinter<N - 1>.
+
+// The inductive case.
+template <size_t N>
+struct TuplePrefixPrinter {
+ // Prints the first N fields of a tuple.
+ template <typename Tuple>
+ static void PrintPrefixTo(const Tuple& t, ::std::ostream* os) {
+ TuplePrefixPrinter<N - 1>::PrintPrefixTo(t, os);
+ *os << ", ";
+ UniversalPrinter<typename ::std::tr1::tuple_element<N - 1, Tuple>::type>
+ ::Print(::std::tr1::get<N - 1>(t), os);
+ }
+
+ // Tersely prints the first N fields of a tuple to a string vector,
+ // one element for each field.
+ template <typename Tuple>
+ static void TersePrintPrefixToStrings(const Tuple& t, Strings* strings) {
+ TuplePrefixPrinter<N - 1>::TersePrintPrefixToStrings(t, strings);
+ ::std::stringstream ss;
+ UniversalTersePrint(::std::tr1::get<N - 1>(t), &ss);
+ strings->push_back(ss.str());
+ }
+};
+
+// Base cases.
+template <>
+struct TuplePrefixPrinter<0> {
+ template <typename Tuple>
+ static void PrintPrefixTo(const Tuple&, ::std::ostream*) {}
+
+ template <typename Tuple>
+ static void TersePrintPrefixToStrings(const Tuple&, Strings*) {}
+};
+// We have to specialize the entire TuplePrefixPrinter<> class
+// template here, even though the definition of
+// TersePrintPrefixToStrings() is the same as the generic version, as
+// Embarcadero (formerly CodeGear, formerly Borland) C++ doesn't
+// support specializing a method template of a class template.
+template <>
+struct TuplePrefixPrinter<1> {
+ template <typename Tuple>
+ static void PrintPrefixTo(const Tuple& t, ::std::ostream* os) {
+ UniversalPrinter<typename ::std::tr1::tuple_element<0, Tuple>::type>::
+ Print(::std::tr1::get<0>(t), os);
+ }
+
+ template <typename Tuple>
+ static void TersePrintPrefixToStrings(const Tuple& t, Strings* strings) {
+ ::std::stringstream ss;
+ UniversalTersePrint(::std::tr1::get<0>(t), &ss);
+ strings->push_back(ss.str());
+ }
+};
+
+// Helper function for printing a tuple. T must be instantiated with
+// a tuple type.
+template <typename T>
+void PrintTupleTo(const T& t, ::std::ostream* os) {
+ *os << "(";
+ TuplePrefixPrinter< ::std::tr1::tuple_size<T>::value>::
+ PrintPrefixTo(t, os);
+ *os << ")";
+}
+
+// Prints the fields of a tuple tersely to a string vector, one
+// element for each field. See the comment before
+// UniversalTersePrint() for how we define "tersely".
+template <typename Tuple>
+Strings UniversalTersePrintTupleFieldsToStrings(const Tuple& value) {
+ Strings result;
+ TuplePrefixPrinter< ::std::tr1::tuple_size<Tuple>::value>::
+ TersePrintPrefixToStrings(value, &result);
+ return result;
+}
+#endif // GTEST_HAS_TR1_TUPLE
+
+} // namespace internal
+
+template <typename T>
+::std::string PrintToString(const T& value) {
+ ::std::stringstream ss;
+ internal::UniversalTersePrint(value, &ss);
+ return ss.str();
+}
+
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest-spi.h b/libvpx/third_party/googletest/src/include/gtest/gtest-spi.h
new file mode 100644
index 0000000..b226e55
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest-spi.h
@@ -0,0 +1,232 @@
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Utilities for testing Google Test itself and code that uses Google Test
+// (e.g. frameworks built on top of Google Test).
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_SPI_H_
+#define GTEST_INCLUDE_GTEST_GTEST_SPI_H_
+
+#include "gtest/gtest.h"
+
+namespace testing {
+
+// This helper class can be used to mock out Google Test failure reporting
+// so that we can test Google Test or code that builds on Google Test.
+//
+// An object of this class appends a TestPartResult object to the
+// TestPartResultArray object given in the constructor whenever a Google Test
+// failure is reported. It can either intercept only failures that are
+// generated in the same thread that created this object or it can intercept
+// all generated failures. The scope of this mock object can be controlled with
+// the second argument to the two arguments constructor.
+class GTEST_API_ ScopedFakeTestPartResultReporter
+ : public TestPartResultReporterInterface {
+ public:
+ // The two possible mocking modes of this object.
+ enum InterceptMode {
+ INTERCEPT_ONLY_CURRENT_THREAD, // Intercepts only thread local failures.
+ INTERCEPT_ALL_THREADS // Intercepts all failures.
+ };
+
+ // The c'tor sets this object as the test part result reporter used
+ // by Google Test. The 'result' parameter specifies where to report the
+ // results. This reporter will only catch failures generated in the current
+ // thread. DEPRECATED
+ explicit ScopedFakeTestPartResultReporter(TestPartResultArray* result);
+
+ // Same as above, but you can choose the interception scope of this object.
+ ScopedFakeTestPartResultReporter(InterceptMode intercept_mode,
+ TestPartResultArray* result);
+
+ // The d'tor restores the previous test part result reporter.
+ virtual ~ScopedFakeTestPartResultReporter();
+
+ // Appends the TestPartResult object to the TestPartResultArray
+ // received in the constructor.
+ //
+ // This method is from the TestPartResultReporterInterface
+ // interface.
+ virtual void ReportTestPartResult(const TestPartResult& result);
+ private:
+ void Init();
+
+ const InterceptMode intercept_mode_;
+ TestPartResultReporterInterface* old_reporter_;
+ TestPartResultArray* const result_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedFakeTestPartResultReporter);
+};
+
+namespace internal {
+
+// A helper class for implementing EXPECT_FATAL_FAILURE() and
+// EXPECT_NONFATAL_FAILURE(). Its destructor verifies that the given
+// TestPartResultArray contains exactly one failure that has the given
+// type and contains the given substring. If that's not the case, a
+// non-fatal failure will be generated.
+class GTEST_API_ SingleFailureChecker {
+ public:
+ // The constructor remembers the arguments.
+ SingleFailureChecker(const TestPartResultArray* results,
+ TestPartResult::Type type,
+ const string& substr);
+ ~SingleFailureChecker();
+ private:
+ const TestPartResultArray* const results_;
+ const TestPartResult::Type type_;
+ const string substr_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(SingleFailureChecker);
+};
+
+} // namespace internal
+
+} // namespace testing
+
+// A set of macros for testing Google Test assertions or code that's expected
+// to generate Google Test fatal failures. It verifies that the given
+// statement will cause exactly one fatal Google Test failure with 'substr'
+// being part of the failure message.
+//
+// There are two different versions of this macro. EXPECT_FATAL_FAILURE only
+// affects and considers failures generated in the current thread and
+// EXPECT_FATAL_FAILURE_ON_ALL_THREADS does the same but for all threads.
+//
+// The verification of the assertion is done correctly even when the statement
+// throws an exception or aborts the current function.
+//
+// Known restrictions:
+// - 'statement' cannot reference local non-static variables or
+// non-static members of the current object.
+// - 'statement' cannot return a value.
+// - You cannot stream a failure message to this macro.
+//
+// Note that even though the implementations of the following two
+// macros are much alike, we cannot refactor them to use a common
+// helper macro, due to some peculiarity in how the preprocessor
+// works. The AcceptsMacroThatExpandsToUnprotectedComma test in
+// gtest_unittest.cc will fail to compile if we do that.
+#define EXPECT_FATAL_FAILURE(statement, substr) \
+ do { \
+ class GTestExpectFatalFailureHelper {\
+ public:\
+ static void Execute() { statement; }\
+ };\
+ ::testing::TestPartResultArray gtest_failures;\
+ ::testing::internal::SingleFailureChecker gtest_checker(\
+ >est_failures, ::testing::TestPartResult::kFatalFailure, (substr));\
+ {\
+ ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+ ::testing::ScopedFakeTestPartResultReporter:: \
+ INTERCEPT_ONLY_CURRENT_THREAD, >est_failures);\
+ GTestExpectFatalFailureHelper::Execute();\
+ }\
+ } while (::testing::internal::AlwaysFalse())
+
+#define EXPECT_FATAL_FAILURE_ON_ALL_THREADS(statement, substr) \
+ do { \
+ class GTestExpectFatalFailureHelper {\
+ public:\
+ static void Execute() { statement; }\
+ };\
+ ::testing::TestPartResultArray gtest_failures;\
+ ::testing::internal::SingleFailureChecker gtest_checker(\
+ >est_failures, ::testing::TestPartResult::kFatalFailure, (substr));\
+ {\
+ ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+ ::testing::ScopedFakeTestPartResultReporter:: \
+ INTERCEPT_ALL_THREADS, >est_failures);\
+ GTestExpectFatalFailureHelper::Execute();\
+ }\
+ } while (::testing::internal::AlwaysFalse())
+
+// A macro for testing Google Test assertions or code that's expected to
+// generate Google Test non-fatal failures. It asserts that the given
+// statement will cause exactly one non-fatal Google Test failure with 'substr'
+// being part of the failure message.
+//
+// There are two different versions of this macro. EXPECT_NONFATAL_FAILURE only
+// affects and considers failures generated in the current thread and
+// EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS does the same but for all threads.
+//
+// 'statement' is allowed to reference local variables and members of
+// the current object.
+//
+// The verification of the assertion is done correctly even when the statement
+// throws an exception or aborts the current function.
+//
+// Known restrictions:
+// - You cannot stream a failure message to this macro.
+//
+// Note that even though the implementations of the following two
+// macros are much alike, we cannot refactor them to use a common
+// helper macro, due to some peculiarity in how the preprocessor
+// works. If we do that, the code won't compile when the user gives
+// EXPECT_NONFATAL_FAILURE() a statement that contains a macro that
+// expands to code containing an unprotected comma. The
+// AcceptsMacroThatExpandsToUnprotectedComma test in gtest_unittest.cc
+// catches that.
+//
+// For the same reason, we have to write
+// if (::testing::internal::AlwaysTrue()) { statement; }
+// instead of
+// GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement)
+// to avoid an MSVC warning on unreachable code.
+#define EXPECT_NONFATAL_FAILURE(statement, substr) \
+ do {\
+ ::testing::TestPartResultArray gtest_failures;\
+ ::testing::internal::SingleFailureChecker gtest_checker(\
+ >est_failures, ::testing::TestPartResult::kNonFatalFailure, \
+ (substr));\
+ {\
+ ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+ ::testing::ScopedFakeTestPartResultReporter:: \
+ INTERCEPT_ONLY_CURRENT_THREAD, >est_failures);\
+ if (::testing::internal::AlwaysTrue()) { statement; }\
+ }\
+ } while (::testing::internal::AlwaysFalse())
+
+#define EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(statement, substr) \
+ do {\
+ ::testing::TestPartResultArray gtest_failures;\
+ ::testing::internal::SingleFailureChecker gtest_checker(\
+ >est_failures, ::testing::TestPartResult::kNonFatalFailure, \
+ (substr));\
+ {\
+ ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+ ::testing::ScopedFakeTestPartResultReporter::INTERCEPT_ALL_THREADS,\
+ >est_failures);\
+ if (::testing::internal::AlwaysTrue()) { statement; }\
+ }\
+ } while (::testing::internal::AlwaysFalse())
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_SPI_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest-test-part.h b/libvpx/third_party/googletest/src/include/gtest/gtest-test-part.h
new file mode 100644
index 0000000..8aeea14
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest-test-part.h
@@ -0,0 +1,176 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
+#define GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
+
+#include <iosfwd>
+#include <vector>
+#include "gtest/internal/gtest-internal.h"
+#include "gtest/internal/gtest-string.h"
+
+namespace testing {
+
+// A copyable object representing the result of a test part (i.e. an
+// assertion or an explicit FAIL(), ADD_FAILURE(), or SUCCESS()).
+//
+// Don't inherit from TestPartResult as its destructor is not virtual.
+class GTEST_API_ TestPartResult {
+ public:
+ // The possible outcomes of a test part (i.e. an assertion or an
+ // explicit SUCCEED(), FAIL(), or ADD_FAILURE()).
+ enum Type {
+ kSuccess, // Succeeded.
+ kNonFatalFailure, // Failed but the test can continue.
+ kFatalFailure // Failed and the test should be terminated.
+ };
+
+ // C'tor. TestPartResult does NOT have a default constructor.
+ // Always use this constructor (with parameters) to create a
+ // TestPartResult object.
+ TestPartResult(Type a_type,
+ const char* a_file_name,
+ int a_line_number,
+ const char* a_message)
+ : type_(a_type),
+ file_name_(a_file_name),
+ line_number_(a_line_number),
+ summary_(ExtractSummary(a_message)),
+ message_(a_message) {
+ }
+
+ // Gets the outcome of the test part.
+ Type type() const { return type_; }
+
+ // Gets the name of the source file where the test part took place, or
+ // NULL if it's unknown.
+ const char* file_name() const { return file_name_.c_str(); }
+
+ // Gets the line in the source file where the test part took place,
+ // or -1 if it's unknown.
+ int line_number() const { return line_number_; }
+
+ // Gets the summary of the failure message.
+ const char* summary() const { return summary_.c_str(); }
+
+ // Gets the message associated with the test part.
+ const char* message() const { return message_.c_str(); }
+
+ // Returns true iff the test part passed.
+ bool passed() const { return type_ == kSuccess; }
+
+ // Returns true iff the test part failed.
+ bool failed() const { return type_ != kSuccess; }
+
+ // Returns true iff the test part non-fatally failed.
+ bool nonfatally_failed() const { return type_ == kNonFatalFailure; }
+
+ // Returns true iff the test part fatally failed.
+ bool fatally_failed() const { return type_ == kFatalFailure; }
+ private:
+ Type type_;
+
+ // Gets the summary of the failure message by omitting the stack
+ // trace in it.
+ static internal::String ExtractSummary(const char* message);
+
+ // The name of the source file where the test part took place, or
+ // NULL if the source file is unknown.
+ internal::String file_name_;
+ // The line in the source file where the test part took place, or -1
+ // if the line number is unknown.
+ int line_number_;
+ internal::String summary_; // The test failure summary.
+ internal::String message_; // The test failure message.
+};
+
+// Prints a TestPartResult object.
+std::ostream& operator<<(std::ostream& os, const TestPartResult& result);
+
+// An array of TestPartResult objects.
+//
+// Don't inherit from TestPartResultArray as its destructor is not
+// virtual.
+class GTEST_API_ TestPartResultArray {
+ public:
+ TestPartResultArray() {}
+
+ // Appends the given TestPartResult to the array.
+ void Append(const TestPartResult& result);
+
+ // Returns the TestPartResult at the given index (0-based).
+ const TestPartResult& GetTestPartResult(int index) const;
+
+ // Returns the number of TestPartResult objects in the array.
+ int size() const;
+
+ private:
+ std::vector<TestPartResult> array_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestPartResultArray);
+};
+
+// This interface knows how to report a test part result.
+class TestPartResultReporterInterface {
+ public:
+ virtual ~TestPartResultReporterInterface() {}
+
+ virtual void ReportTestPartResult(const TestPartResult& result) = 0;
+};
+
+namespace internal {
+
+// This helper class is used by {ASSERT|EXPECT}_NO_FATAL_FAILURE to check if a
+// statement generates new fatal failures. To do so it registers itself as the
+// current test part result reporter. Besides checking if fatal failures were
+// reported, it only delegates the reporting to the former result reporter.
+// The original result reporter is restored in the destructor.
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+class GTEST_API_ HasNewFatalFailureHelper
+ : public TestPartResultReporterInterface {
+ public:
+ HasNewFatalFailureHelper();
+ virtual ~HasNewFatalFailureHelper();
+ virtual void ReportTestPartResult(const TestPartResult& result);
+ bool has_new_fatal_failure() const { return has_new_fatal_failure_; }
+ private:
+ bool has_new_fatal_failure_;
+ TestPartResultReporterInterface* original_reporter_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(HasNewFatalFailureHelper);
+};
+
+} // namespace internal
+
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest-typed-test.h b/libvpx/third_party/googletest/src/include/gtest/gtest-typed-test.h
new file mode 100644
index 0000000..fe1e83b
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest-typed-test.h
@@ -0,0 +1,259 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
+
+// This header implements typed tests and type-parameterized tests.
+
+// Typed (aka type-driven) tests repeat the same test for types in a
+// list. You must know which types you want to test with when writing
+// typed tests. Here's how you do it:
+
+#if 0
+
+// First, define a fixture class template. It should be parameterized
+// by a type. Remember to derive it from testing::Test.
+template <typename T>
+class FooTest : public testing::Test {
+ public:
+ ...
+ typedef std::list<T> List;
+ static T shared_;
+ T value_;
+};
+
+// Next, associate a list of types with the test case, which will be
+// repeated for each type in the list. The typedef is necessary for
+// the macro to parse correctly.
+typedef testing::Types<char, int, unsigned int> MyTypes;
+TYPED_TEST_CASE(FooTest, MyTypes);
+
+// If the type list contains only one type, you can write that type
+// directly without Types<...>:
+// TYPED_TEST_CASE(FooTest, int);
+
+// Then, use TYPED_TEST() instead of TEST_F() to define as many typed
+// tests for this test case as you want.
+TYPED_TEST(FooTest, DoesBlah) {
+ // Inside a test, refer to TypeParam to get the type parameter.
+ // Since we are inside a derived class template, C++ requires use to
+ // visit the members of FooTest via 'this'.
+ TypeParam n = this->value_;
+
+ // To visit static members of the fixture, add the TestFixture::
+ // prefix.
+ n += TestFixture::shared_;
+
+ // To refer to typedefs in the fixture, add the "typename
+ // TestFixture::" prefix.
+ typename TestFixture::List values;
+ values.push_back(n);
+ ...
+}
+
+TYPED_TEST(FooTest, HasPropertyA) { ... }
+
+#endif // 0
+
+// Type-parameterized tests are abstract test patterns parameterized
+// by a type. Compared with typed tests, type-parameterized tests
+// allow you to define the test pattern without knowing what the type
+// parameters are. The defined pattern can be instantiated with
+// different types any number of times, in any number of translation
+// units.
+//
+// If you are designing an interface or concept, you can define a
+// suite of type-parameterized tests to verify properties that any
+// valid implementation of the interface/concept should have. Then,
+// each implementation can easily instantiate the test suite to verify
+// that it conforms to the requirements, without having to write
+// similar tests repeatedly. Here's an example:
+
+#if 0
+
+// First, define a fixture class template. It should be parameterized
+// by a type. Remember to derive it from testing::Test.
+template <typename T>
+class FooTest : public testing::Test {
+ ...
+};
+
+// Next, declare that you will define a type-parameterized test case
+// (the _P suffix is for "parameterized" or "pattern", whichever you
+// prefer):
+TYPED_TEST_CASE_P(FooTest);
+
+// Then, use TYPED_TEST_P() to define as many type-parameterized tests
+// for this type-parameterized test case as you want.
+TYPED_TEST_P(FooTest, DoesBlah) {
+ // Inside a test, refer to TypeParam to get the type parameter.
+ TypeParam n = 0;
+ ...
+}
+
+TYPED_TEST_P(FooTest, HasPropertyA) { ... }
+
+// Now the tricky part: you need to register all test patterns before
+// you can instantiate them. The first argument of the macro is the
+// test case name; the rest are the names of the tests in this test
+// case.
+REGISTER_TYPED_TEST_CASE_P(FooTest,
+ DoesBlah, HasPropertyA);
+
+// Finally, you are free to instantiate the pattern with the types you
+// want. If you put the above code in a header file, you can #include
+// it in multiple C++ source files and instantiate it multiple times.
+//
+// To distinguish different instances of the pattern, the first
+// argument to the INSTANTIATE_* macro is a prefix that will be added
+// to the actual test case name. Remember to pick unique prefixes for
+// different instances.
+typedef testing::Types<char, int, unsigned int> MyTypes;
+INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, MyTypes);
+
+// If the type list contains only one type, you can write that type
+// directly without Types<...>:
+// INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, int);
+
+#endif // 0
+
+#include "gtest/internal/gtest-port.h"
+#include "gtest/internal/gtest-type-util.h"
+
+// Implements typed tests.
+
+#if GTEST_HAS_TYPED_TEST
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the name of the typedef for the type parameters of the
+// given test case.
+# define GTEST_TYPE_PARAMS_(TestCaseName) gtest_type_params_##TestCaseName##_
+
+// The 'Types' template argument below must have spaces around it
+// since some compilers may choke on '>>' when passing a template
+// instance (e.g. Types<int>)
+# define TYPED_TEST_CASE(CaseName, Types) \
+ typedef ::testing::internal::TypeList< Types >::type \
+ GTEST_TYPE_PARAMS_(CaseName)
+
+# define TYPED_TEST(CaseName, TestName) \
+ template <typename gtest_TypeParam_> \
+ class GTEST_TEST_CLASS_NAME_(CaseName, TestName) \
+ : public CaseName<gtest_TypeParam_> { \
+ private: \
+ typedef CaseName<gtest_TypeParam_> TestFixture; \
+ typedef gtest_TypeParam_ TypeParam; \
+ virtual void TestBody(); \
+ }; \
+ bool gtest_##CaseName##_##TestName##_registered_ GTEST_ATTRIBUTE_UNUSED_ = \
+ ::testing::internal::TypeParameterizedTest< \
+ CaseName, \
+ ::testing::internal::TemplateSel< \
+ GTEST_TEST_CLASS_NAME_(CaseName, TestName)>, \
+ GTEST_TYPE_PARAMS_(CaseName)>::Register(\
+ "", #CaseName, #TestName, 0); \
+ template <typename gtest_TypeParam_> \
+ void GTEST_TEST_CLASS_NAME_(CaseName, TestName)<gtest_TypeParam_>::TestBody()
+
+#endif // GTEST_HAS_TYPED_TEST
+
+// Implements type-parameterized tests.
+
+#if GTEST_HAS_TYPED_TEST_P
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the namespace name that the type-parameterized tests for
+// the given type-parameterized test case are defined in. The exact
+// name of the namespace is subject to change without notice.
+# define GTEST_CASE_NAMESPACE_(TestCaseName) \
+ gtest_case_##TestCaseName##_
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the name of the variable used to remember the names of
+// the defined tests in the given test case.
+# define GTEST_TYPED_TEST_CASE_P_STATE_(TestCaseName) \
+ gtest_typed_test_case_p_state_##TestCaseName##_
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE DIRECTLY.
+//
+// Expands to the name of the variable used to remember the names of
+// the registered tests in the given test case.
+# define GTEST_REGISTERED_TEST_NAMES_(TestCaseName) \
+ gtest_registered_test_names_##TestCaseName##_
+
+// The variables defined in the type-parameterized test macros are
+// static as typically these macros are used in a .h file that can be
+// #included in multiple translation units linked together.
+# define TYPED_TEST_CASE_P(CaseName) \
+ static ::testing::internal::TypedTestCasePState \
+ GTEST_TYPED_TEST_CASE_P_STATE_(CaseName)
+
+# define TYPED_TEST_P(CaseName, TestName) \
+ namespace GTEST_CASE_NAMESPACE_(CaseName) { \
+ template <typename gtest_TypeParam_> \
+ class TestName : public CaseName<gtest_TypeParam_> { \
+ private: \
+ typedef CaseName<gtest_TypeParam_> TestFixture; \
+ typedef gtest_TypeParam_ TypeParam; \
+ virtual void TestBody(); \
+ }; \
+ static bool gtest_##TestName##_defined_ GTEST_ATTRIBUTE_UNUSED_ = \
+ GTEST_TYPED_TEST_CASE_P_STATE_(CaseName).AddTestName(\
+ __FILE__, __LINE__, #CaseName, #TestName); \
+ } \
+ template <typename gtest_TypeParam_> \
+ void GTEST_CASE_NAMESPACE_(CaseName)::TestName<gtest_TypeParam_>::TestBody()
+
+# define REGISTER_TYPED_TEST_CASE_P(CaseName, ...) \
+ namespace GTEST_CASE_NAMESPACE_(CaseName) { \
+ typedef ::testing::internal::Templates<__VA_ARGS__>::type gtest_AllTests_; \
+ } \
+ static const char* const GTEST_REGISTERED_TEST_NAMES_(CaseName) = \
+ GTEST_TYPED_TEST_CASE_P_STATE_(CaseName).VerifyRegisteredTestNames(\
+ __FILE__, __LINE__, #__VA_ARGS__)
+
+// The 'Types' template argument below must have spaces around it
+// since some compilers may choke on '>>' when passing a template
+// instance (e.g. Types<int>)
+# define INSTANTIATE_TYPED_TEST_CASE_P(Prefix, CaseName, Types) \
+ bool gtest_##Prefix##_##CaseName GTEST_ATTRIBUTE_UNUSED_ = \
+ ::testing::internal::TypeParameterizedTestCase<CaseName, \
+ GTEST_CASE_NAMESPACE_(CaseName)::gtest_AllTests_, \
+ ::testing::internal::TypeList< Types >::type>::Register(\
+ #Prefix, #CaseName, GTEST_REGISTERED_TEST_NAMES_(CaseName))
+
+#endif // GTEST_HAS_TYPED_TEST_P
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest.h b/libvpx/third_party/googletest/src/include/gtest/gtest.h
new file mode 100644
index 0000000..cd01c7b
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest.h
@@ -0,0 +1,2155 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the public API for Google Test. It should be
+// included by any test program that uses Google Test.
+//
+// IMPORTANT NOTE: Due to limitation of the C++ language, we have to
+// leave some internal implementation details in this header file.
+// They are clearly marked by comments like this:
+//
+// // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+//
+// Such code is NOT meant to be used by a user directly, and is subject
+// to CHANGE WITHOUT NOTICE. Therefore DO NOT DEPEND ON IT in a user
+// program!
+//
+// Acknowledgment: Google Test borrowed the idea of automatic test
+// registration from Barthelemy Dagenais' (barthelemy@prologique.com)
+// easyUnit framework.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_H_
+
+#include <limits>
+#include <vector>
+
+#include "gtest/internal/gtest-internal.h"
+#include "gtest/internal/gtest-string.h"
+#include "gtest/gtest-death-test.h"
+#include "gtest/gtest-message.h"
+#include "gtest/gtest-param-test.h"
+#include "gtest/gtest-printers.h"
+#include "gtest/gtest_prod.h"
+#include "gtest/gtest-test-part.h"
+#include "gtest/gtest-typed-test.h"
+
+// Depending on the platform, different string classes are available.
+// On Linux, in addition to ::std::string, Google also makes use of
+// class ::string, which has the same interface as ::std::string, but
+// has a different implementation.
+//
+// The user can define GTEST_HAS_GLOBAL_STRING to 1 to indicate that
+// ::string is available AND is a distinct type to ::std::string, or
+// define it to 0 to indicate otherwise.
+//
+// If the user's ::std::string and ::string are the same class due to
+// aliasing, he should define GTEST_HAS_GLOBAL_STRING to 0.
+//
+// If the user doesn't define GTEST_HAS_GLOBAL_STRING, it is defined
+// heuristically.
+
+namespace testing {
+
+// Declares the flags.
+
+// This flag temporary enables the disabled tests.
+GTEST_DECLARE_bool_(also_run_disabled_tests);
+
+// This flag brings the debugger on an assertion failure.
+GTEST_DECLARE_bool_(break_on_failure);
+
+// This flag controls whether Google Test catches all test-thrown exceptions
+// and logs them as failures.
+GTEST_DECLARE_bool_(catch_exceptions);
+
+// This flag enables using colors in terminal output. Available values are
+// "yes" to enable colors, "no" (disable colors), or "auto" (the default)
+// to let Google Test decide.
+GTEST_DECLARE_string_(color);
+
+// This flag sets up the filter to select by name using a glob pattern
+// the tests to run. If the filter is not given all tests are executed.
+GTEST_DECLARE_string_(filter);
+
+// This flag causes the Google Test to list tests. None of the tests listed
+// are actually run if the flag is provided.
+GTEST_DECLARE_bool_(list_tests);
+
+// This flag controls whether Google Test emits a detailed XML report to a file
+// in addition to its normal textual output.
+GTEST_DECLARE_string_(output);
+
+// This flags control whether Google Test prints the elapsed time for each
+// test.
+GTEST_DECLARE_bool_(print_time);
+
+// This flag specifies the random number seed.
+GTEST_DECLARE_int32_(random_seed);
+
+// This flag sets how many times the tests are repeated. The default value
+// is 1. If the value is -1 the tests are repeating forever.
+GTEST_DECLARE_int32_(repeat);
+
+// This flag controls whether Google Test includes Google Test internal
+// stack frames in failure stack traces.
+GTEST_DECLARE_bool_(show_internal_stack_frames);
+
+// When this flag is specified, tests' order is randomized on every iteration.
+GTEST_DECLARE_bool_(shuffle);
+
+// This flag specifies the maximum number of stack frames to be
+// printed in a failure message.
+GTEST_DECLARE_int32_(stack_trace_depth);
+
+// When this flag is specified, a failed assertion will throw an
+// exception if exceptions are enabled, or exit the program with a
+// non-zero code otherwise.
+GTEST_DECLARE_bool_(throw_on_failure);
+
+// When this flag is set with a "host:port" string, on supported
+// platforms test results are streamed to the specified port on
+// the specified host machine.
+GTEST_DECLARE_string_(stream_result_to);
+
+// The upper limit for valid stack trace depths.
+const int kMaxStackTraceDepth = 100;
+
+namespace internal {
+
+class AssertHelper;
+class DefaultGlobalTestPartResultReporter;
+class ExecDeathTest;
+class NoExecDeathTest;
+class FinalSuccessChecker;
+class GTestFlagSaver;
+class TestResultAccessor;
+class TestEventListenersAccessor;
+class TestEventRepeater;
+class WindowsDeathTest;
+class UnitTestImpl* GetUnitTestImpl();
+void ReportFailureInUnknownLocation(TestPartResult::Type result_type,
+ const String& message);
+
+// Converts a streamable value to a String. A NULL pointer is
+// converted to "(null)". When the input value is a ::string,
+// ::std::string, ::wstring, or ::std::wstring object, each NUL
+// character in it is replaced with "\\0".
+// Declared in gtest-internal.h but defined here, so that it has access
+// to the definition of the Message class, required by the ARM
+// compiler.
+template <typename T>
+String StreamableToString(const T& streamable) {
+ return (Message() << streamable).GetString();
+}
+
+} // namespace internal
+
+// The friend relationship of some of these classes is cyclic.
+// If we don't forward declare them the compiler might confuse the classes
+// in friendship clauses with same named classes on the scope.
+class Test;
+class TestCase;
+class TestInfo;
+class UnitTest;
+
+// A class for indicating whether an assertion was successful. When
+// the assertion wasn't successful, the AssertionResult object
+// remembers a non-empty message that describes how it failed.
+//
+// To create an instance of this class, use one of the factory functions
+// (AssertionSuccess() and AssertionFailure()).
+//
+// This class is useful for two purposes:
+// 1. Defining predicate functions to be used with Boolean test assertions
+// EXPECT_TRUE/EXPECT_FALSE and their ASSERT_ counterparts
+// 2. Defining predicate-format functions to be
+// used with predicate assertions (ASSERT_PRED_FORMAT*, etc).
+//
+// For example, if you define IsEven predicate:
+//
+// testing::AssertionResult IsEven(int n) {
+// if ((n % 2) == 0)
+// return testing::AssertionSuccess();
+// else
+// return testing::AssertionFailure() << n << " is odd";
+// }
+//
+// Then the failed expectation EXPECT_TRUE(IsEven(Fib(5)))
+// will print the message
+//
+// Value of: IsEven(Fib(5))
+// Actual: false (5 is odd)
+// Expected: true
+//
+// instead of a more opaque
+//
+// Value of: IsEven(Fib(5))
+// Actual: false
+// Expected: true
+//
+// in case IsEven is a simple Boolean predicate.
+//
+// If you expect your predicate to be reused and want to support informative
+// messages in EXPECT_FALSE and ASSERT_FALSE (negative assertions show up
+// about half as often as positive ones in our tests), supply messages for
+// both success and failure cases:
+//
+// testing::AssertionResult IsEven(int n) {
+// if ((n % 2) == 0)
+// return testing::AssertionSuccess() << n << " is even";
+// else
+// return testing::AssertionFailure() << n << " is odd";
+// }
+//
+// Then a statement EXPECT_FALSE(IsEven(Fib(6))) will print
+//
+// Value of: IsEven(Fib(6))
+// Actual: true (8 is even)
+// Expected: false
+//
+// NB: Predicates that support negative Boolean assertions have reduced
+// performance in positive ones so be careful not to use them in tests
+// that have lots (tens of thousands) of positive Boolean assertions.
+//
+// To use this class with EXPECT_PRED_FORMAT assertions such as:
+//
+// // Verifies that Foo() returns an even number.
+// EXPECT_PRED_FORMAT1(IsEven, Foo());
+//
+// you need to define:
+//
+// testing::AssertionResult IsEven(const char* expr, int n) {
+// if ((n % 2) == 0)
+// return testing::AssertionSuccess();
+// else
+// return testing::AssertionFailure()
+// << "Expected: " << expr << " is even\n Actual: it's " << n;
+// }
+//
+// If Foo() returns 5, you will see the following message:
+//
+// Expected: Foo() is even
+// Actual: it's 5
+//
+class GTEST_API_ AssertionResult {
+ public:
+ // Copy constructor.
+ // Used in EXPECT_TRUE/FALSE(assertion_result).
+ AssertionResult(const AssertionResult& other);
+ // Used in the EXPECT_TRUE/FALSE(bool_expression).
+ explicit AssertionResult(bool success) : success_(success) {}
+
+ // Returns true iff the assertion succeeded.
+ operator bool() const { return success_; } // NOLINT
+
+ // Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE.
+ AssertionResult operator!() const;
+
+ // Returns the text streamed into this AssertionResult. Test assertions
+ // use it when they fail (i.e., the predicate's outcome doesn't match the
+ // assertion's expectation). When nothing has been streamed into the
+ // object, returns an empty string.
+ const char* message() const {
+ return message_.get() != NULL ? message_->c_str() : "";
+ }
+ // TODO(vladl@google.com): Remove this after making sure no clients use it.
+ // Deprecated; please use message() instead.
+ const char* failure_message() const { return message(); }
+
+ // Streams a custom failure message into this object.
+ template <typename T> AssertionResult& operator<<(const T& value) {
+ AppendMessage(Message() << value);
+ return *this;
+ }
+
+ // Allows streaming basic output manipulators such as endl or flush into
+ // this object.
+ AssertionResult& operator<<(
+ ::std::ostream& (*basic_manipulator)(::std::ostream& stream)) {
+ AppendMessage(Message() << basic_manipulator);
+ return *this;
+ }
+
+ private:
+ // Appends the contents of message to message_.
+ void AppendMessage(const Message& a_message) {
+ if (message_.get() == NULL)
+ message_.reset(new ::std::string);
+ message_->append(a_message.GetString().c_str());
+ }
+
+ // Stores result of the assertion predicate.
+ bool success_;
+ // Stores the message describing the condition in case the expectation
+ // construct is not satisfied with the predicate's outcome.
+ // Referenced via a pointer to avoid taking too much stack frame space
+ // with test assertions.
+ internal::scoped_ptr< ::std::string> message_;
+
+ GTEST_DISALLOW_ASSIGN_(AssertionResult);
+};
+
+// Makes a successful assertion result.
+GTEST_API_ AssertionResult AssertionSuccess();
+
+// Makes a failed assertion result.
+GTEST_API_ AssertionResult AssertionFailure();
+
+// Makes a failed assertion result with the given failure message.
+// Deprecated; use AssertionFailure() << msg.
+GTEST_API_ AssertionResult AssertionFailure(const Message& msg);
+
+// The abstract class that all tests inherit from.
+//
+// In Google Test, a unit test program contains one or many TestCases, and
+// each TestCase contains one or many Tests.
+//
+// When you define a test using the TEST macro, you don't need to
+// explicitly derive from Test - the TEST macro automatically does
+// this for you.
+//
+// The only time you derive from Test is when defining a test fixture
+// to be used a TEST_F. For example:
+//
+// class FooTest : public testing::Test {
+// protected:
+// virtual void SetUp() { ... }
+// virtual void TearDown() { ... }
+// ...
+// };
+//
+// TEST_F(FooTest, Bar) { ... }
+// TEST_F(FooTest, Baz) { ... }
+//
+// Test is not copyable.
+class GTEST_API_ Test {
+ public:
+ friend class TestInfo;
+
+ // Defines types for pointers to functions that set up and tear down
+ // a test case.
+ typedef internal::SetUpTestCaseFunc SetUpTestCaseFunc;
+ typedef internal::TearDownTestCaseFunc TearDownTestCaseFunc;
+
+ // The d'tor is virtual as we intend to inherit from Test.
+ virtual ~Test();
+
+ // Sets up the stuff shared by all tests in this test case.
+ //
+ // Google Test will call Foo::SetUpTestCase() before running the first
+ // test in test case Foo. Hence a sub-class can define its own
+ // SetUpTestCase() method to shadow the one defined in the super
+ // class.
+ static void SetUpTestCase() {}
+
+ // Tears down the stuff shared by all tests in this test case.
+ //
+ // Google Test will call Foo::TearDownTestCase() after running the last
+ // test in test case Foo. Hence a sub-class can define its own
+ // TearDownTestCase() method to shadow the one defined in the super
+ // class.
+ static void TearDownTestCase() {}
+
+ // Returns true iff the current test has a fatal failure.
+ static bool HasFatalFailure();
+
+ // Returns true iff the current test has a non-fatal failure.
+ static bool HasNonfatalFailure();
+
+ // Returns true iff the current test has a (either fatal or
+ // non-fatal) failure.
+ static bool HasFailure() { return HasFatalFailure() || HasNonfatalFailure(); }
+
+ // Logs a property for the current test. Only the last value for a given
+ // key is remembered.
+ // These are public static so they can be called from utility functions
+ // that are not members of the test fixture.
+ // The arguments are const char* instead strings, as Google Test is used
+ // on platforms where string doesn't compile.
+ //
+ // Note that a driving consideration for these RecordProperty methods
+ // was to produce xml output suited to the Greenspan charting utility,
+ // which at present will only chart values that fit in a 32-bit int. It
+ // is the user's responsibility to restrict their values to 32-bit ints
+ // if they intend them to be used with Greenspan.
+ static void RecordProperty(const char* key, const char* value);
+ static void RecordProperty(const char* key, int value);
+
+ protected:
+ // Creates a Test object.
+ Test();
+
+ // Sets up the test fixture.
+ virtual void SetUp();
+
+ // Tears down the test fixture.
+ virtual void TearDown();
+
+ private:
+ // Returns true iff the current test has the same fixture class as
+ // the first test in the current test case.
+ static bool HasSameFixtureClass();
+
+ // Runs the test after the test fixture has been set up.
+ //
+ // A sub-class must implement this to define the test logic.
+ //
+ // DO NOT OVERRIDE THIS FUNCTION DIRECTLY IN A USER PROGRAM.
+ // Instead, use the TEST or TEST_F macro.
+ virtual void TestBody() = 0;
+
+ // Sets up, executes, and tears down the test.
+ void Run();
+
+ // Deletes self. We deliberately pick an unusual name for this
+ // internal method to avoid clashing with names used in user TESTs.
+ void DeleteSelf_() { delete this; }
+
+ // Uses a GTestFlagSaver to save and restore all Google Test flags.
+ const internal::GTestFlagSaver* const gtest_flag_saver_;
+
+ // Often a user mis-spells SetUp() as Setup() and spends a long time
+ // wondering why it is never called by Google Test. The declaration of
+ // the following method is solely for catching such an error at
+ // compile time:
+ //
+ // - The return type is deliberately chosen to be not void, so it
+ // will be a conflict if a user declares void Setup() in his test
+ // fixture.
+ //
+ // - This method is private, so it will be another compiler error
+ // if a user calls it from his test fixture.
+ //
+ // DO NOT OVERRIDE THIS FUNCTION.
+ //
+ // If you see an error about overriding the following function or
+ // about it being private, you have mis-spelled SetUp() as Setup().
+ struct Setup_should_be_spelled_SetUp {};
+ virtual Setup_should_be_spelled_SetUp* Setup() { return NULL; }
+
+ // We disallow copying Tests.
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(Test);
+};
+
+typedef internal::TimeInMillis TimeInMillis;
+
+// A copyable object representing a user specified test property which can be
+// output as a key/value string pair.
+//
+// Don't inherit from TestProperty as its destructor is not virtual.
+class TestProperty {
+ public:
+ // C'tor. TestProperty does NOT have a default constructor.
+ // Always use this constructor (with parameters) to create a
+ // TestProperty object.
+ TestProperty(const char* a_key, const char* a_value) :
+ key_(a_key), value_(a_value) {
+ }
+
+ // Gets the user supplied key.
+ const char* key() const {
+ return key_.c_str();
+ }
+
+ // Gets the user supplied value.
+ const char* value() const {
+ return value_.c_str();
+ }
+
+ // Sets a new value, overriding the one supplied in the constructor.
+ void SetValue(const char* new_value) {
+ value_ = new_value;
+ }
+
+ private:
+ // The key supplied by the user.
+ internal::String key_;
+ // The value supplied by the user.
+ internal::String value_;
+};
+
+// The result of a single Test. This includes a list of
+// TestPartResults, a list of TestProperties, a count of how many
+// death tests there are in the Test, and how much time it took to run
+// the Test.
+//
+// TestResult is not copyable.
+class GTEST_API_ TestResult {
+ public:
+ // Creates an empty TestResult.
+ TestResult();
+
+ // D'tor. Do not inherit from TestResult.
+ ~TestResult();
+
+ // Gets the number of all test parts. This is the sum of the number
+ // of successful test parts and the number of failed test parts.
+ int total_part_count() const;
+
+ // Returns the number of the test properties.
+ int test_property_count() const;
+
+ // Returns true iff the test passed (i.e. no test part failed).
+ bool Passed() const { return !Failed(); }
+
+ // Returns true iff the test failed.
+ bool Failed() const;
+
+ // Returns true iff the test fatally failed.
+ bool HasFatalFailure() const;
+
+ // Returns true iff the test has a non-fatal failure.
+ bool HasNonfatalFailure() const;
+
+ // Returns the elapsed time, in milliseconds.
+ TimeInMillis elapsed_time() const { return elapsed_time_; }
+
+ // Returns the i-th test part result among all the results. i can range
+ // from 0 to test_property_count() - 1. If i is not in that range, aborts
+ // the program.
+ const TestPartResult& GetTestPartResult(int i) const;
+
+ // Returns the i-th test property. i can range from 0 to
+ // test_property_count() - 1. If i is not in that range, aborts the
+ // program.
+ const TestProperty& GetTestProperty(int i) const;
+
+ private:
+ friend class TestInfo;
+ friend class UnitTest;
+ friend class internal::DefaultGlobalTestPartResultReporter;
+ friend class internal::ExecDeathTest;
+ friend class internal::TestResultAccessor;
+ friend class internal::UnitTestImpl;
+ friend class internal::WindowsDeathTest;
+
+ // Gets the vector of TestPartResults.
+ const std::vector<TestPartResult>& test_part_results() const {
+ return test_part_results_;
+ }
+
+ // Gets the vector of TestProperties.
+ const std::vector<TestProperty>& test_properties() const {
+ return test_properties_;
+ }
+
+ // Sets the elapsed time.
+ void set_elapsed_time(TimeInMillis elapsed) { elapsed_time_ = elapsed; }
+
+ // Adds a test property to the list. The property is validated and may add
+ // a non-fatal failure if invalid (e.g., if it conflicts with reserved
+ // key names). If a property is already recorded for the same key, the
+ // value will be updated, rather than storing multiple values for the same
+ // key.
+ void RecordProperty(const TestProperty& test_property);
+
+ // Adds a failure if the key is a reserved attribute of Google Test
+ // testcase tags. Returns true if the property is valid.
+ // TODO(russr): Validate attribute names are legal and human readable.
+ static bool ValidateTestProperty(const TestProperty& test_property);
+
+ // Adds a test part result to the list.
+ void AddTestPartResult(const TestPartResult& test_part_result);
+
+ // Returns the death test count.
+ int death_test_count() const { return death_test_count_; }
+
+ // Increments the death test count, returning the new count.
+ int increment_death_test_count() { return ++death_test_count_; }
+
+ // Clears the test part results.
+ void ClearTestPartResults();
+
+ // Clears the object.
+ void Clear();
+
+ // Protects mutable state of the property vector and of owned
+ // properties, whose values may be updated.
+ internal::Mutex test_properites_mutex_;
+
+ // The vector of TestPartResults
+ std::vector<TestPartResult> test_part_results_;
+ // The vector of TestProperties
+ std::vector<TestProperty> test_properties_;
+ // Running count of death tests.
+ int death_test_count_;
+ // The elapsed time, in milliseconds.
+ TimeInMillis elapsed_time_;
+
+ // We disallow copying TestResult.
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestResult);
+}; // class TestResult
+
+// A TestInfo object stores the following information about a test:
+//
+// Test case name
+// Test name
+// Whether the test should be run
+// A function pointer that creates the test object when invoked
+// Test result
+//
+// The constructor of TestInfo registers itself with the UnitTest
+// singleton such that the RUN_ALL_TESTS() macro knows which tests to
+// run.
+class GTEST_API_ TestInfo {
+ public:
+ // Destructs a TestInfo object. This function is not virtual, so
+ // don't inherit from TestInfo.
+ ~TestInfo();
+
+ // Returns the test case name.
+ const char* test_case_name() const { return test_case_name_.c_str(); }
+
+ // Returns the test name.
+ const char* name() const { return name_.c_str(); }
+
+ // Returns the name of the parameter type, or NULL if this is not a typed
+ // or a type-parameterized test.
+ const char* type_param() const {
+ if (type_param_.get() != NULL)
+ return type_param_->c_str();
+ return NULL;
+ }
+
+ // Returns the text representation of the value parameter, or NULL if this
+ // is not a value-parameterized test.
+ const char* value_param() const {
+ if (value_param_.get() != NULL)
+ return value_param_->c_str();
+ return NULL;
+ }
+
+ // Returns true if this test should run, that is if the test is not disabled
+ // (or it is disabled but the also_run_disabled_tests flag has been specified)
+ // and its full name matches the user-specified filter.
+ //
+ // Google Test allows the user to filter the tests by their full names.
+ // The full name of a test Bar in test case Foo is defined as
+ // "Foo.Bar". Only the tests that match the filter will run.
+ //
+ // A filter is a colon-separated list of glob (not regex) patterns,
+ // optionally followed by a '-' and a colon-separated list of
+ // negative patterns (tests to exclude). A test is run if it
+ // matches one of the positive patterns and does not match any of
+ // the negative patterns.
+ //
+ // For example, *A*:Foo.* is a filter that matches any string that
+ // contains the character 'A' or starts with "Foo.".
+ bool should_run() const { return should_run_; }
+
+ // Returns the result of the test.
+ const TestResult* result() const { return &result_; }
+
+ private:
+
+#if GTEST_HAS_DEATH_TEST
+ friend class internal::DefaultDeathTestFactory;
+#endif // GTEST_HAS_DEATH_TEST
+ friend class Test;
+ friend class TestCase;
+ friend class internal::UnitTestImpl;
+ friend TestInfo* internal::MakeAndRegisterTestInfo(
+ const char* test_case_name, const char* name,
+ const char* type_param,
+ const char* value_param,
+ internal::TypeId fixture_class_id,
+ Test::SetUpTestCaseFunc set_up_tc,
+ Test::TearDownTestCaseFunc tear_down_tc,
+ internal::TestFactoryBase* factory);
+
+ // Constructs a TestInfo object. The newly constructed instance assumes
+ // ownership of the factory object.
+ TestInfo(const char* test_case_name, const char* name,
+ const char* a_type_param,
+ const char* a_value_param,
+ internal::TypeId fixture_class_id,
+ internal::TestFactoryBase* factory);
+
+ // Increments the number of death tests encountered in this test so
+ // far.
+ int increment_death_test_count() {
+ return result_.increment_death_test_count();
+ }
+
+ // Creates the test object, runs it, records its result, and then
+ // deletes it.
+ void Run();
+
+ static void ClearTestResult(TestInfo* test_info) {
+ test_info->result_.Clear();
+ }
+
+ // These fields are immutable properties of the test.
+ const std::string test_case_name_; // Test case name
+ const std::string name_; // Test name
+ // Name of the parameter type, or NULL if this is not a typed or a
+ // type-parameterized test.
+ const internal::scoped_ptr<const ::std::string> type_param_;
+ // Text representation of the value parameter, or NULL if this is not a
+ // value-parameterized test.
+ const internal::scoped_ptr<const ::std::string> value_param_;
+ const internal::TypeId fixture_class_id_; // ID of the test fixture class
+ bool should_run_; // True iff this test should run
+ bool is_disabled_; // True iff this test is disabled
+ bool matches_filter_; // True if this test matches the
+ // user-specified filter.
+ internal::TestFactoryBase* const factory_; // The factory that creates
+ // the test object
+
+ // This field is mutable and needs to be reset before running the
+ // test for the second time.
+ TestResult result_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestInfo);
+};
+
+// A test case, which consists of a vector of TestInfos.
+//
+// TestCase is not copyable.
+class GTEST_API_ TestCase {
+ public:
+ // Creates a TestCase with the given name.
+ //
+ // TestCase does NOT have a default constructor. Always use this
+ // constructor to create a TestCase object.
+ //
+ // Arguments:
+ //
+ // name: name of the test case
+ // a_type_param: the name of the test's type parameter, or NULL if
+ // this is not a type-parameterized test.
+ // set_up_tc: pointer to the function that sets up the test case
+ // tear_down_tc: pointer to the function that tears down the test case
+ TestCase(const char* name, const char* a_type_param,
+ Test::SetUpTestCaseFunc set_up_tc,
+ Test::TearDownTestCaseFunc tear_down_tc);
+
+ // Destructor of TestCase.
+ virtual ~TestCase();
+
+ // Gets the name of the TestCase.
+ const char* name() const { return name_.c_str(); }
+
+ // Returns the name of the parameter type, or NULL if this is not a
+ // type-parameterized test case.
+ const char* type_param() const {
+ if (type_param_.get() != NULL)
+ return type_param_->c_str();
+ return NULL;
+ }
+
+ // Returns true if any test in this test case should run.
+ bool should_run() const { return should_run_; }
+
+ // Gets the number of successful tests in this test case.
+ int successful_test_count() const;
+
+ // Gets the number of failed tests in this test case.
+ int failed_test_count() const;
+
+ // Gets the number of disabled tests in this test case.
+ int disabled_test_count() const;
+
+ // Get the number of tests in this test case that should run.
+ int test_to_run_count() const;
+
+ // Gets the number of all tests in this test case.
+ int total_test_count() const;
+
+ // Returns true iff the test case passed.
+ bool Passed() const { return !Failed(); }
+
+ // Returns true iff the test case failed.
+ bool Failed() const { return failed_test_count() > 0; }
+
+ // Returns the elapsed time, in milliseconds.
+ TimeInMillis elapsed_time() const { return elapsed_time_; }
+
+ // Returns the i-th test among all the tests. i can range from 0 to
+ // total_test_count() - 1. If i is not in that range, returns NULL.
+ const TestInfo* GetTestInfo(int i) const;
+
+ private:
+ friend class Test;
+ friend class internal::UnitTestImpl;
+
+ // Gets the (mutable) vector of TestInfos in this TestCase.
+ std::vector<TestInfo*>& test_info_list() { return test_info_list_; }
+
+ // Gets the (immutable) vector of TestInfos in this TestCase.
+ const std::vector<TestInfo*>& test_info_list() const {
+ return test_info_list_;
+ }
+
+ // Returns the i-th test among all the tests. i can range from 0 to
+ // total_test_count() - 1. If i is not in that range, returns NULL.
+ TestInfo* GetMutableTestInfo(int i);
+
+ // Sets the should_run member.
+ void set_should_run(bool should) { should_run_ = should; }
+
+ // Adds a TestInfo to this test case. Will delete the TestInfo upon
+ // destruction of the TestCase object.
+ void AddTestInfo(TestInfo * test_info);
+
+ // Clears the results of all tests in this test case.
+ void ClearResult();
+
+ // Clears the results of all tests in the given test case.
+ static void ClearTestCaseResult(TestCase* test_case) {
+ test_case->ClearResult();
+ }
+
+ // Runs every test in this TestCase.
+ void Run();
+
+ // Runs SetUpTestCase() for this TestCase. This wrapper is needed
+ // for catching exceptions thrown from SetUpTestCase().
+ void RunSetUpTestCase() { (*set_up_tc_)(); }
+
+ // Runs TearDownTestCase() for this TestCase. This wrapper is
+ // needed for catching exceptions thrown from TearDownTestCase().
+ void RunTearDownTestCase() { (*tear_down_tc_)(); }
+
+ // Returns true iff test passed.
+ static bool TestPassed(const TestInfo* test_info) {
+ return test_info->should_run() && test_info->result()->Passed();
+ }
+
+ // Returns true iff test failed.
+ static bool TestFailed(const TestInfo* test_info) {
+ return test_info->should_run() && test_info->result()->Failed();
+ }
+
+ // Returns true iff test is disabled.
+ static bool TestDisabled(const TestInfo* test_info) {
+ return test_info->is_disabled_;
+ }
+
+ // Returns true if the given test should run.
+ static bool ShouldRunTest(const TestInfo* test_info) {
+ return test_info->should_run();
+ }
+
+ // Shuffles the tests in this test case.
+ void ShuffleTests(internal::Random* random);
+
+ // Restores the test order to before the first shuffle.
+ void UnshuffleTests();
+
+ // Name of the test case.
+ internal::String name_;
+ // Name of the parameter type, or NULL if this is not a typed or a
+ // type-parameterized test.
+ const internal::scoped_ptr<const ::std::string> type_param_;
+ // The vector of TestInfos in their original order. It owns the
+ // elements in the vector.
+ std::vector<TestInfo*> test_info_list_;
+ // Provides a level of indirection for the test list to allow easy
+ // shuffling and restoring the test order. The i-th element in this
+ // vector is the index of the i-th test in the shuffled test list.
+ std::vector<int> test_indices_;
+ // Pointer to the function that sets up the test case.
+ Test::SetUpTestCaseFunc set_up_tc_;
+ // Pointer to the function that tears down the test case.
+ Test::TearDownTestCaseFunc tear_down_tc_;
+ // True iff any test in this test case should run.
+ bool should_run_;
+ // Elapsed time, in milliseconds.
+ TimeInMillis elapsed_time_;
+
+ // We disallow copying TestCases.
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestCase);
+};
+
+// An Environment object is capable of setting up and tearing down an
+// environment. The user should subclass this to define his own
+// environment(s).
+//
+// An Environment object does the set-up and tear-down in virtual
+// methods SetUp() and TearDown() instead of the constructor and the
+// destructor, as:
+//
+// 1. You cannot safely throw from a destructor. This is a problem
+// as in some cases Google Test is used where exceptions are enabled, and
+// we may want to implement ASSERT_* using exceptions where they are
+// available.
+// 2. You cannot use ASSERT_* directly in a constructor or
+// destructor.
+class Environment {
+ public:
+ // The d'tor is virtual as we need to subclass Environment.
+ virtual ~Environment() {}
+
+ // Override this to define how to set up the environment.
+ virtual void SetUp() {}
+
+ // Override this to define how to tear down the environment.
+ virtual void TearDown() {}
+ private:
+ // If you see an error about overriding the following function or
+ // about it being private, you have mis-spelled SetUp() as Setup().
+ struct Setup_should_be_spelled_SetUp {};
+ virtual Setup_should_be_spelled_SetUp* Setup() { return NULL; }
+};
+
+// The interface for tracing execution of tests. The methods are organized in
+// the order the corresponding events are fired.
+class TestEventListener {
+ public:
+ virtual ~TestEventListener() {}
+
+ // Fired before any test activity starts.
+ virtual void OnTestProgramStart(const UnitTest& unit_test) = 0;
+
+ // Fired before each iteration of tests starts. There may be more than
+ // one iteration if GTEST_FLAG(repeat) is set. iteration is the iteration
+ // index, starting from 0.
+ virtual void OnTestIterationStart(const UnitTest& unit_test,
+ int iteration) = 0;
+
+ // Fired before environment set-up for each iteration of tests starts.
+ virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test) = 0;
+
+ // Fired after environment set-up for each iteration of tests ends.
+ virtual void OnEnvironmentsSetUpEnd(const UnitTest& unit_test) = 0;
+
+ // Fired before the test case starts.
+ virtual void OnTestCaseStart(const TestCase& test_case) = 0;
+
+ // Fired before the test starts.
+ virtual void OnTestStart(const TestInfo& test_info) = 0;
+
+ // Fired after a failed assertion or a SUCCEED() invocation.
+ virtual void OnTestPartResult(const TestPartResult& test_part_result) = 0;
+
+ // Fired after the test ends.
+ virtual void OnTestEnd(const TestInfo& test_info) = 0;
+
+ // Fired after the test case ends.
+ virtual void OnTestCaseEnd(const TestCase& test_case) = 0;
+
+ // Fired before environment tear-down for each iteration of tests starts.
+ virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test) = 0;
+
+ // Fired after environment tear-down for each iteration of tests ends.
+ virtual void OnEnvironmentsTearDownEnd(const UnitTest& unit_test) = 0;
+
+ // Fired after each iteration of tests finishes.
+ virtual void OnTestIterationEnd(const UnitTest& unit_test,
+ int iteration) = 0;
+
+ // Fired after all test activities have ended.
+ virtual void OnTestProgramEnd(const UnitTest& unit_test) = 0;
+};
+
+// The convenience class for users who need to override just one or two
+// methods and are not concerned that a possible change to a signature of
+// the methods they override will not be caught during the build. For
+// comments about each method please see the definition of TestEventListener
+// above.
+class EmptyTestEventListener : public TestEventListener {
+ public:
+ virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {}
+ virtual void OnTestIterationStart(const UnitTest& /*unit_test*/,
+ int /*iteration*/) {}
+ virtual void OnEnvironmentsSetUpStart(const UnitTest& /*unit_test*/) {}
+ virtual void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) {}
+ virtual void OnTestCaseStart(const TestCase& /*test_case*/) {}
+ virtual void OnTestStart(const TestInfo& /*test_info*/) {}
+ virtual void OnTestPartResult(const TestPartResult& /*test_part_result*/) {}
+ virtual void OnTestEnd(const TestInfo& /*test_info*/) {}
+ virtual void OnTestCaseEnd(const TestCase& /*test_case*/) {}
+ virtual void OnEnvironmentsTearDownStart(const UnitTest& /*unit_test*/) {}
+ virtual void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) {}
+ virtual void OnTestIterationEnd(const UnitTest& /*unit_test*/,
+ int /*iteration*/) {}
+ virtual void OnTestProgramEnd(const UnitTest& /*unit_test*/) {}
+};
+
+// TestEventListeners lets users add listeners to track events in Google Test.
+class GTEST_API_ TestEventListeners {
+ public:
+ TestEventListeners();
+ ~TestEventListeners();
+
+ // Appends an event listener to the end of the list. Google Test assumes
+ // the ownership of the listener (i.e. it will delete the listener when
+ // the test program finishes).
+ void Append(TestEventListener* listener);
+
+ // Removes the given event listener from the list and returns it. It then
+ // becomes the caller's responsibility to delete the listener. Returns
+ // NULL if the listener is not found in the list.
+ TestEventListener* Release(TestEventListener* listener);
+
+ // Returns the standard listener responsible for the default console
+ // output. Can be removed from the listeners list to shut down default
+ // console output. Note that removing this object from the listener list
+ // with Release transfers its ownership to the caller and makes this
+ // function return NULL the next time.
+ TestEventListener* default_result_printer() const {
+ return default_result_printer_;
+ }
+
+ // Returns the standard listener responsible for the default XML output
+ // controlled by the --gtest_output=xml flag. Can be removed from the
+ // listeners list by users who want to shut down the default XML output
+ // controlled by this flag and substitute it with custom one. Note that
+ // removing this object from the listener list with Release transfers its
+ // ownership to the caller and makes this function return NULL the next
+ // time.
+ TestEventListener* default_xml_generator() const {
+ return default_xml_generator_;
+ }
+
+ private:
+ friend class TestCase;
+ friend class TestInfo;
+ friend class internal::DefaultGlobalTestPartResultReporter;
+ friend class internal::NoExecDeathTest;
+ friend class internal::TestEventListenersAccessor;
+ friend class internal::UnitTestImpl;
+
+ // Returns repeater that broadcasts the TestEventListener events to all
+ // subscribers.
+ TestEventListener* repeater();
+
+ // Sets the default_result_printer attribute to the provided listener.
+ // The listener is also added to the listener list and previous
+ // default_result_printer is removed from it and deleted. The listener can
+ // also be NULL in which case it will not be added to the list. Does
+ // nothing if the previous and the current listener objects are the same.
+ void SetDefaultResultPrinter(TestEventListener* listener);
+
+ // Sets the default_xml_generator attribute to the provided listener. The
+ // listener is also added to the listener list and previous
+ // default_xml_generator is removed from it and deleted. The listener can
+ // also be NULL in which case it will not be added to the list. Does
+ // nothing if the previous and the current listener objects are the same.
+ void SetDefaultXmlGenerator(TestEventListener* listener);
+
+ // Controls whether events will be forwarded by the repeater to the
+ // listeners in the list.
+ bool EventForwardingEnabled() const;
+ void SuppressEventForwarding();
+
+ // The actual list of listeners.
+ internal::TestEventRepeater* repeater_;
+ // Listener responsible for the standard result output.
+ TestEventListener* default_result_printer_;
+ // Listener responsible for the creation of the XML output file.
+ TestEventListener* default_xml_generator_;
+
+ // We disallow copying TestEventListeners.
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventListeners);
+};
+
+// A UnitTest consists of a vector of TestCases.
+//
+// This is a singleton class. The only instance of UnitTest is
+// created when UnitTest::GetInstance() is first called. This
+// instance is never deleted.
+//
+// UnitTest is not copyable.
+//
+// This class is thread-safe as long as the methods are called
+// according to their specification.
+class GTEST_API_ UnitTest {
+ public:
+ // Gets the singleton UnitTest object. The first time this method
+ // is called, a UnitTest object is constructed and returned.
+ // Consecutive calls will return the same object.
+ static UnitTest* GetInstance();
+
+ // Runs all tests in this UnitTest object and prints the result.
+ // Returns 0 if successful, or 1 otherwise.
+ //
+ // This method can only be called from the main thread.
+ //
+ // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+ int Run() GTEST_MUST_USE_RESULT_;
+
+ // Returns the working directory when the first TEST() or TEST_F()
+ // was executed. The UnitTest object owns the string.
+ const char* original_working_dir() const;
+
+ // Returns the TestCase object for the test that's currently running,
+ // or NULL if no test is running.
+ const TestCase* current_test_case() const;
+
+ // Returns the TestInfo object for the test that's currently running,
+ // or NULL if no test is running.
+ const TestInfo* current_test_info() const;
+
+ // Returns the random seed used at the start of the current test run.
+ int random_seed() const;
+
+#if GTEST_HAS_PARAM_TEST
+ // Returns the ParameterizedTestCaseRegistry object used to keep track of
+ // value-parameterized tests and instantiate and register them.
+ //
+ // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+ internal::ParameterizedTestCaseRegistry& parameterized_test_registry();
+#endif // GTEST_HAS_PARAM_TEST
+
+ // Gets the number of successful test cases.
+ int successful_test_case_count() const;
+
+ // Gets the number of failed test cases.
+ int failed_test_case_count() const;
+
+ // Gets the number of all test cases.
+ int total_test_case_count() const;
+
+ // Gets the number of all test cases that contain at least one test
+ // that should run.
+ int test_case_to_run_count() const;
+
+ // Gets the number of successful tests.
+ int successful_test_count() const;
+
+ // Gets the number of failed tests.
+ int failed_test_count() const;
+
+ // Gets the number of disabled tests.
+ int disabled_test_count() const;
+
+ // Gets the number of all tests.
+ int total_test_count() const;
+
+ // Gets the number of tests that should run.
+ int test_to_run_count() const;
+
+ // Gets the elapsed time, in milliseconds.
+ TimeInMillis elapsed_time() const;
+
+ // Returns true iff the unit test passed (i.e. all test cases passed).
+ bool Passed() const;
+
+ // Returns true iff the unit test failed (i.e. some test case failed
+ // or something outside of all tests failed).
+ bool Failed() const;
+
+ // Gets the i-th test case among all the test cases. i can range from 0 to
+ // total_test_case_count() - 1. If i is not in that range, returns NULL.
+ const TestCase* GetTestCase(int i) const;
+
+ // Returns the list of event listeners that can be used to track events
+ // inside Google Test.
+ TestEventListeners& listeners();
+
+ private:
+ // Registers and returns a global test environment. When a test
+ // program is run, all global test environments will be set-up in
+ // the order they were registered. After all tests in the program
+ // have finished, all global test environments will be torn-down in
+ // the *reverse* order they were registered.
+ //
+ // The UnitTest object takes ownership of the given environment.
+ //
+ // This method can only be called from the main thread.
+ Environment* AddEnvironment(Environment* env);
+
+ // Adds a TestPartResult to the current TestResult object. All
+ // Google Test assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc)
+ // eventually call this to report their results. The user code
+ // should use the assertion macros instead of calling this directly.
+ void AddTestPartResult(TestPartResult::Type result_type,
+ const char* file_name,
+ int line_number,
+ const internal::String& message,
+ const internal::String& os_stack_trace);
+
+ // Adds a TestProperty to the current TestResult object. If the result already
+ // contains a property with the same key, the value will be updated.
+ void RecordPropertyForCurrentTest(const char* key, const char* value);
+
+ // Gets the i-th test case among all the test cases. i can range from 0 to
+ // total_test_case_count() - 1. If i is not in that range, returns NULL.
+ TestCase* GetMutableTestCase(int i);
+
+ // Accessors for the implementation object.
+ internal::UnitTestImpl* impl() { return impl_; }
+ const internal::UnitTestImpl* impl() const { return impl_; }
+
+ // These classes and funcions are friends as they need to access private
+ // members of UnitTest.
+ friend class Test;
+ friend class internal::AssertHelper;
+ friend class internal::ScopedTrace;
+ friend Environment* AddGlobalTestEnvironment(Environment* env);
+ friend internal::UnitTestImpl* internal::GetUnitTestImpl();
+ friend void internal::ReportFailureInUnknownLocation(
+ TestPartResult::Type result_type,
+ const internal::String& message);
+
+ // Creates an empty UnitTest.
+ UnitTest();
+
+ // D'tor
+ virtual ~UnitTest();
+
+ // Pushes a trace defined by SCOPED_TRACE() on to the per-thread
+ // Google Test trace stack.
+ void PushGTestTrace(const internal::TraceInfo& trace);
+
+ // Pops a trace from the per-thread Google Test trace stack.
+ void PopGTestTrace();
+
+ // Protects mutable state in *impl_. This is mutable as some const
+ // methods need to lock it too.
+ mutable internal::Mutex mutex_;
+
+ // Opaque implementation object. This field is never changed once
+ // the object is constructed. We don't mark it as const here, as
+ // doing so will cause a warning in the constructor of UnitTest.
+ // Mutable state in *impl_ is protected by mutex_.
+ internal::UnitTestImpl* impl_;
+
+ // We disallow copying UnitTest.
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTest);
+};
+
+// A convenient wrapper for adding an environment for the test
+// program.
+//
+// You should call this before RUN_ALL_TESTS() is called, probably in
+// main(). If you use gtest_main, you need to call this before main()
+// starts for it to take effect. For example, you can define a global
+// variable like this:
+//
+// testing::Environment* const foo_env =
+// testing::AddGlobalTestEnvironment(new FooEnvironment);
+//
+// However, we strongly recommend you to write your own main() and
+// call AddGlobalTestEnvironment() there, as relying on initialization
+// of global variables makes the code harder to read and may cause
+// problems when you register multiple environments from different
+// translation units and the environments have dependencies among them
+// (remember that the compiler doesn't guarantee the order in which
+// global variables from different translation units are initialized).
+inline Environment* AddGlobalTestEnvironment(Environment* env) {
+ return UnitTest::GetInstance()->AddEnvironment(env);
+}
+
+// Initializes Google Test. This must be called before calling
+// RUN_ALL_TESTS(). In particular, it parses a command line for the
+// flags that Google Test recognizes. Whenever a Google Test flag is
+// seen, it is removed from argv, and *argc is decremented.
+//
+// No value is returned. Instead, the Google Test flag variables are
+// updated.
+//
+// Calling the function for the second time has no user-visible effect.
+GTEST_API_ void InitGoogleTest(int* argc, char** argv);
+
+// This overloaded version can be used in Windows programs compiled in
+// UNICODE mode.
+GTEST_API_ void InitGoogleTest(int* argc, wchar_t** argv);
+
+namespace internal {
+
+// Formats a comparison assertion (e.g. ASSERT_EQ, EXPECT_LT, and etc)
+// operand to be used in a failure message. The type (but not value)
+// of the other operand may affect the format. This allows us to
+// print a char* as a raw pointer when it is compared against another
+// char*, and print it as a C string when it is compared against an
+// std::string object, for example.
+//
+// The default implementation ignores the type of the other operand.
+// Some specialized versions are used to handle formatting wide or
+// narrow C strings.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+template <typename T1, typename T2>
+String FormatForComparisonFailureMessage(const T1& value,
+ const T2& /* other_operand */) {
+ // C++Builder compiles this incorrectly if the namespace isn't explicitly
+ // given.
+ return ::testing::PrintToString(value);
+}
+
+// The helper function for {ASSERT|EXPECT}_EQ.
+template <typename T1, typename T2>
+AssertionResult CmpHelperEQ(const char* expected_expression,
+ const char* actual_expression,
+ const T1& expected,
+ const T2& actual) {
+#ifdef _MSC_VER
+# pragma warning(push) // Saves the current warning state.
+# pragma warning(disable:4389) // Temporarily disables warning on
+ // signed/unsigned mismatch.
+#endif
+
+ if (expected == actual) {
+ return AssertionSuccess();
+ }
+
+#ifdef _MSC_VER
+# pragma warning(pop) // Restores the warning state.
+#endif
+
+ return EqFailure(expected_expression,
+ actual_expression,
+ FormatForComparisonFailureMessage(expected, actual),
+ FormatForComparisonFailureMessage(actual, expected),
+ false);
+}
+
+// With this overloaded version, we allow anonymous enums to be used
+// in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous enums
+// can be implicitly cast to BiggestInt.
+GTEST_API_ AssertionResult CmpHelperEQ(const char* expected_expression,
+ const char* actual_expression,
+ BiggestInt expected,
+ BiggestInt actual);
+
+// The helper class for {ASSERT|EXPECT}_EQ. The template argument
+// lhs_is_null_literal is true iff the first argument to ASSERT_EQ()
+// is a null pointer literal. The following default implementation is
+// for lhs_is_null_literal being false.
+template <bool lhs_is_null_literal>
+class EqHelper {
+ public:
+ // This templatized version is for the general case.
+ template <typename T1, typename T2>
+ static AssertionResult Compare(const char* expected_expression,
+ const char* actual_expression,
+ const T1& expected,
+ const T2& actual) {
+ return CmpHelperEQ(expected_expression, actual_expression, expected,
+ actual);
+ }
+
+ // With this overloaded version, we allow anonymous enums to be used
+ // in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous
+ // enums can be implicitly cast to BiggestInt.
+ //
+ // Even though its body looks the same as the above version, we
+ // cannot merge the two, as it will make anonymous enums unhappy.
+ static AssertionResult Compare(const char* expected_expression,
+ const char* actual_expression,
+ BiggestInt expected,
+ BiggestInt actual) {
+ return CmpHelperEQ(expected_expression, actual_expression, expected,
+ actual);
+ }
+};
+
+// This specialization is used when the first argument to ASSERT_EQ()
+// is a null pointer literal, like NULL, false, or 0.
+template <>
+class EqHelper<true> {
+ public:
+ // We define two overloaded versions of Compare(). The first
+ // version will be picked when the second argument to ASSERT_EQ() is
+ // NOT a pointer, e.g. ASSERT_EQ(0, AnIntFunction()) or
+ // EXPECT_EQ(false, a_bool).
+ template <typename T1, typename T2>
+ static AssertionResult Compare(
+ const char* expected_expression,
+ const char* actual_expression,
+ const T1& expected,
+ const T2& actual,
+ // The following line prevents this overload from being considered if T2
+ // is not a pointer type. We need this because ASSERT_EQ(NULL, my_ptr)
+ // expands to Compare("", "", NULL, my_ptr), which requires a conversion
+ // to match the Secret* in the other overload, which would otherwise make
+ // this template match better.
+ typename EnableIf<!is_pointer<T2>::value>::type* = 0) {
+ return CmpHelperEQ(expected_expression, actual_expression, expected,
+ actual);
+ }
+
+ // This version will be picked when the second argument to ASSERT_EQ() is a
+ // pointer, e.g. ASSERT_EQ(NULL, a_pointer).
+ template <typename T>
+ static AssertionResult Compare(
+ const char* expected_expression,
+ const char* actual_expression,
+ // We used to have a second template parameter instead of Secret*. That
+ // template parameter would deduce to 'long', making this a better match
+ // than the first overload even without the first overload's EnableIf.
+ // Unfortunately, gcc with -Wconversion-null warns when "passing NULL to
+ // non-pointer argument" (even a deduced integral argument), so the old
+ // implementation caused warnings in user code.
+ Secret* /* expected (NULL) */,
+ T* actual) {
+ // We already know that 'expected' is a null pointer.
+ return CmpHelperEQ(expected_expression, actual_expression,
+ static_cast<T*>(NULL), actual);
+ }
+};
+
+// A macro for implementing the helper functions needed to implement
+// ASSERT_?? and EXPECT_??. It is here just to avoid copy-and-paste
+// of similar code.
+//
+// For each templatized helper function, we also define an overloaded
+// version for BiggestInt in order to reduce code bloat and allow
+// anonymous enums to be used with {ASSERT|EXPECT}_?? when compiled
+// with gcc 4.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+#define GTEST_IMPL_CMP_HELPER_(op_name, op)\
+template <typename T1, typename T2>\
+AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \
+ const T1& val1, const T2& val2) {\
+ if (val1 op val2) {\
+ return AssertionSuccess();\
+ } else {\
+ return AssertionFailure() \
+ << "Expected: (" << expr1 << ") " #op " (" << expr2\
+ << "), actual: " << FormatForComparisonFailureMessage(val1, val2)\
+ << " vs " << FormatForComparisonFailureMessage(val2, val1);\
+ }\
+}\
+GTEST_API_ AssertionResult CmpHelper##op_name(\
+ const char* expr1, const char* expr2, BiggestInt val1, BiggestInt val2)
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+
+// Implements the helper function for {ASSERT|EXPECT}_NE
+GTEST_IMPL_CMP_HELPER_(NE, !=);
+// Implements the helper function for {ASSERT|EXPECT}_LE
+GTEST_IMPL_CMP_HELPER_(LE, <=);
+// Implements the helper function for {ASSERT|EXPECT}_LT
+GTEST_IMPL_CMP_HELPER_(LT, < );
+// Implements the helper function for {ASSERT|EXPECT}_GE
+GTEST_IMPL_CMP_HELPER_(GE, >=);
+// Implements the helper function for {ASSERT|EXPECT}_GT
+GTEST_IMPL_CMP_HELPER_(GT, > );
+
+#undef GTEST_IMPL_CMP_HELPER_
+
+// The helper function for {ASSERT|EXPECT}_STREQ.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+GTEST_API_ AssertionResult CmpHelperSTREQ(const char* expected_expression,
+ const char* actual_expression,
+ const char* expected,
+ const char* actual);
+
+// The helper function for {ASSERT|EXPECT}_STRCASEEQ.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+GTEST_API_ AssertionResult CmpHelperSTRCASEEQ(const char* expected_expression,
+ const char* actual_expression,
+ const char* expected,
+ const char* actual);
+
+// The helper function for {ASSERT|EXPECT}_STRNE.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+GTEST_API_ AssertionResult CmpHelperSTRNE(const char* s1_expression,
+ const char* s2_expression,
+ const char* s1,
+ const char* s2);
+
+// The helper function for {ASSERT|EXPECT}_STRCASENE.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+GTEST_API_ AssertionResult CmpHelperSTRCASENE(const char* s1_expression,
+ const char* s2_expression,
+ const char* s1,
+ const char* s2);
+
+
+// Helper function for *_STREQ on wide strings.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+GTEST_API_ AssertionResult CmpHelperSTREQ(const char* expected_expression,
+ const char* actual_expression,
+ const wchar_t* expected,
+ const wchar_t* actual);
+
+// Helper function for *_STRNE on wide strings.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+GTEST_API_ AssertionResult CmpHelperSTRNE(const char* s1_expression,
+ const char* s2_expression,
+ const wchar_t* s1,
+ const wchar_t* s2);
+
+} // namespace internal
+
+// IsSubstring() and IsNotSubstring() are intended to be used as the
+// first argument to {EXPECT,ASSERT}_PRED_FORMAT2(), not by
+// themselves. They check whether needle is a substring of haystack
+// (NULL is considered a substring of itself only), and return an
+// appropriate error message when they fail.
+//
+// The {needle,haystack}_expr arguments are the stringified
+// expressions that generated the two real arguments.
+GTEST_API_ AssertionResult IsSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const char* needle, const char* haystack);
+GTEST_API_ AssertionResult IsSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const wchar_t* needle, const wchar_t* haystack);
+GTEST_API_ AssertionResult IsNotSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const char* needle, const char* haystack);
+GTEST_API_ AssertionResult IsNotSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const wchar_t* needle, const wchar_t* haystack);
+GTEST_API_ AssertionResult IsSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const ::std::string& needle, const ::std::string& haystack);
+GTEST_API_ AssertionResult IsNotSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const ::std::string& needle, const ::std::string& haystack);
+
+#if GTEST_HAS_STD_WSTRING
+GTEST_API_ AssertionResult IsSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const ::std::wstring& needle, const ::std::wstring& haystack);
+GTEST_API_ AssertionResult IsNotSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const ::std::wstring& needle, const ::std::wstring& haystack);
+#endif // GTEST_HAS_STD_WSTRING
+
+namespace internal {
+
+// Helper template function for comparing floating-points.
+//
+// Template parameter:
+//
+// RawType: the raw floating-point type (either float or double)
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+template <typename RawType>
+AssertionResult CmpHelperFloatingPointEQ(const char* expected_expression,
+ const char* actual_expression,
+ RawType expected,
+ RawType actual) {
+ const FloatingPoint<RawType> lhs(expected), rhs(actual);
+
+ if (lhs.AlmostEquals(rhs)) {
+ return AssertionSuccess();
+ }
+
+ ::std::stringstream expected_ss;
+ expected_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+ << expected;
+
+ ::std::stringstream actual_ss;
+ actual_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+ << actual;
+
+ return EqFailure(expected_expression,
+ actual_expression,
+ StringStreamToString(&expected_ss),
+ StringStreamToString(&actual_ss),
+ false);
+}
+
+// Helper function for implementing ASSERT_NEAR.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+GTEST_API_ AssertionResult DoubleNearPredFormat(const char* expr1,
+ const char* expr2,
+ const char* abs_error_expr,
+ double val1,
+ double val2,
+ double abs_error);
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+// A class that enables one to stream messages to assertion macros
+class GTEST_API_ AssertHelper {
+ public:
+ // Constructor.
+ AssertHelper(TestPartResult::Type type,
+ const char* file,
+ int line,
+ const char* message);
+ ~AssertHelper();
+
+ // Message assignment is a semantic trick to enable assertion
+ // streaming; see the GTEST_MESSAGE_ macro below.
+ void operator=(const Message& message) const;
+
+ private:
+ // We put our data in a struct so that the size of the AssertHelper class can
+ // be as small as possible. This is important because gcc is incapable of
+ // re-using stack space even for temporary variables, so every EXPECT_EQ
+ // reserves stack space for another AssertHelper.
+ struct AssertHelperData {
+ AssertHelperData(TestPartResult::Type t,
+ const char* srcfile,
+ int line_num,
+ const char* msg)
+ : type(t), file(srcfile), line(line_num), message(msg) { }
+
+ TestPartResult::Type const type;
+ const char* const file;
+ int const line;
+ String const message;
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelperData);
+ };
+
+ AssertHelperData* const data_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelper);
+};
+
+} // namespace internal
+
+#if GTEST_HAS_PARAM_TEST
+// The pure interface class that all value-parameterized tests inherit from.
+// A value-parameterized class must inherit from both ::testing::Test and
+// ::testing::WithParamInterface. In most cases that just means inheriting
+// from ::testing::TestWithParam, but more complicated test hierarchies
+// may need to inherit from Test and WithParamInterface at different levels.
+//
+// This interface has support for accessing the test parameter value via
+// the GetParam() method.
+//
+// Use it with one of the parameter generator defining functions, like Range(),
+// Values(), ValuesIn(), Bool(), and Combine().
+//
+// class FooTest : public ::testing::TestWithParam<int> {
+// protected:
+// FooTest() {
+// // Can use GetParam() here.
+// }
+// virtual ~FooTest() {
+// // Can use GetParam() here.
+// }
+// virtual void SetUp() {
+// // Can use GetParam() here.
+// }
+// virtual void TearDown {
+// // Can use GetParam() here.
+// }
+// };
+// TEST_P(FooTest, DoesBar) {
+// // Can use GetParam() method here.
+// Foo foo;
+// ASSERT_TRUE(foo.DoesBar(GetParam()));
+// }
+// INSTANTIATE_TEST_CASE_P(OneToTenRange, FooTest, ::testing::Range(1, 10));
+
+template <typename T>
+class WithParamInterface {
+ public:
+ typedef T ParamType;
+ virtual ~WithParamInterface() {}
+
+ // The current parameter value. Is also available in the test fixture's
+ // constructor. This member function is non-static, even though it only
+ // references static data, to reduce the opportunity for incorrect uses
+ // like writing 'WithParamInterface<bool>::GetParam()' for a test that
+ // uses a fixture whose parameter type is int.
+ const ParamType& GetParam() const { return *parameter_; }
+
+ private:
+ // Sets parameter value. The caller is responsible for making sure the value
+ // remains alive and unchanged throughout the current test.
+ static void SetParam(const ParamType* parameter) {
+ parameter_ = parameter;
+ }
+
+ // Static value used for accessing parameter during a test lifetime.
+ static const ParamType* parameter_;
+
+ // TestClass must be a subclass of WithParamInterface<T> and Test.
+ template <class TestClass> friend class internal::ParameterizedTestFactory;
+};
+
+template <typename T>
+const T* WithParamInterface<T>::parameter_ = NULL;
+
+// Most value-parameterized classes can ignore the existence of
+// WithParamInterface, and can just inherit from ::testing::TestWithParam.
+
+template <typename T>
+class TestWithParam : public Test, public WithParamInterface<T> {
+};
+
+#endif // GTEST_HAS_PARAM_TEST
+
+// Macros for indicating success/failure in test code.
+
+// ADD_FAILURE unconditionally adds a failure to the current test.
+// SUCCEED generates a success - it doesn't automatically make the
+// current test successful, as a test is only successful when it has
+// no failure.
+//
+// EXPECT_* verifies that a certain condition is satisfied. If not,
+// it behaves like ADD_FAILURE. In particular:
+//
+// EXPECT_TRUE verifies that a Boolean condition is true.
+// EXPECT_FALSE verifies that a Boolean condition is false.
+//
+// FAIL and ASSERT_* are similar to ADD_FAILURE and EXPECT_*, except
+// that they will also abort the current function on failure. People
+// usually want the fail-fast behavior of FAIL and ASSERT_*, but those
+// writing data-driven tests often find themselves using ADD_FAILURE
+// and EXPECT_* more.
+//
+// Examples:
+//
+// EXPECT_TRUE(server.StatusIsOK());
+// ASSERT_FALSE(server.HasPendingRequest(port))
+// << "There are still pending requests " << "on port " << port;
+
+// Generates a nonfatal failure with a generic message.
+#define ADD_FAILURE() GTEST_NONFATAL_FAILURE_("Failed")
+
+// Generates a nonfatal failure at the given source file location with
+// a generic message.
+#define ADD_FAILURE_AT(file, line) \
+ GTEST_MESSAGE_AT_(file, line, "Failed", \
+ ::testing::TestPartResult::kNonFatalFailure)
+
+// Generates a fatal failure with a generic message.
+#define GTEST_FAIL() GTEST_FATAL_FAILURE_("Failed")
+
+// Define this macro to 1 to omit the definition of FAIL(), which is a
+// generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_FAIL
+# define FAIL() GTEST_FAIL()
+#endif
+
+// Generates a success with a generic message.
+#define GTEST_SUCCEED() GTEST_SUCCESS_("Succeeded")
+
+// Define this macro to 1 to omit the definition of SUCCEED(), which
+// is a generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_SUCCEED
+# define SUCCEED() GTEST_SUCCEED()
+#endif
+
+// Macros for testing exceptions.
+//
+// * {ASSERT|EXPECT}_THROW(statement, expected_exception):
+// Tests that the statement throws the expected exception.
+// * {ASSERT|EXPECT}_NO_THROW(statement):
+// Tests that the statement doesn't throw any exception.
+// * {ASSERT|EXPECT}_ANY_THROW(statement):
+// Tests that the statement throws an exception.
+
+#define EXPECT_THROW(statement, expected_exception) \
+ GTEST_TEST_THROW_(statement, expected_exception, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_NO_THROW(statement) \
+ GTEST_TEST_NO_THROW_(statement, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_ANY_THROW(statement) \
+ GTEST_TEST_ANY_THROW_(statement, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_THROW(statement, expected_exception) \
+ GTEST_TEST_THROW_(statement, expected_exception, GTEST_FATAL_FAILURE_)
+#define ASSERT_NO_THROW(statement) \
+ GTEST_TEST_NO_THROW_(statement, GTEST_FATAL_FAILURE_)
+#define ASSERT_ANY_THROW(statement) \
+ GTEST_TEST_ANY_THROW_(statement, GTEST_FATAL_FAILURE_)
+
+// Boolean assertions. Condition can be either a Boolean expression or an
+// AssertionResult. For more information on how to use AssertionResult with
+// these macros see comments on that class.
+#define EXPECT_TRUE(condition) \
+ GTEST_TEST_BOOLEAN_(condition, #condition, false, true, \
+ GTEST_NONFATAL_FAILURE_)
+#define EXPECT_FALSE(condition) \
+ GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \
+ GTEST_NONFATAL_FAILURE_)
+#define ASSERT_TRUE(condition) \
+ GTEST_TEST_BOOLEAN_(condition, #condition, false, true, \
+ GTEST_FATAL_FAILURE_)
+#define ASSERT_FALSE(condition) \
+ GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \
+ GTEST_FATAL_FAILURE_)
+
+// Includes the auto-generated header that implements a family of
+// generic predicate assertion macros.
+#include "gtest/gtest_pred_impl.h"
+
+// Macros for testing equalities and inequalities.
+//
+// * {ASSERT|EXPECT}_EQ(expected, actual): Tests that expected == actual
+// * {ASSERT|EXPECT}_NE(v1, v2): Tests that v1 != v2
+// * {ASSERT|EXPECT}_LT(v1, v2): Tests that v1 < v2
+// * {ASSERT|EXPECT}_LE(v1, v2): Tests that v1 <= v2
+// * {ASSERT|EXPECT}_GT(v1, v2): Tests that v1 > v2
+// * {ASSERT|EXPECT}_GE(v1, v2): Tests that v1 >= v2
+//
+// When they are not, Google Test prints both the tested expressions and
+// their actual values. The values must be compatible built-in types,
+// or you will get a compiler error. By "compatible" we mean that the
+// values can be compared by the respective operator.
+//
+// Note:
+//
+// 1. It is possible to make a user-defined type work with
+// {ASSERT|EXPECT}_??(), but that requires overloading the
+// comparison operators and is thus discouraged by the Google C++
+// Usage Guide. Therefore, you are advised to use the
+// {ASSERT|EXPECT}_TRUE() macro to assert that two objects are
+// equal.
+//
+// 2. The {ASSERT|EXPECT}_??() macros do pointer comparisons on
+// pointers (in particular, C strings). Therefore, if you use it
+// with two C strings, you are testing how their locations in memory
+// are related, not how their content is related. To compare two C
+// strings by content, use {ASSERT|EXPECT}_STR*().
+//
+// 3. {ASSERT|EXPECT}_EQ(expected, actual) is preferred to
+// {ASSERT|EXPECT}_TRUE(expected == actual), as the former tells you
+// what the actual value is when it fails, and similarly for the
+// other comparisons.
+//
+// 4. Do not depend on the order in which {ASSERT|EXPECT}_??()
+// evaluate their arguments, which is undefined.
+//
+// 5. These macros evaluate their arguments exactly once.
+//
+// Examples:
+//
+// EXPECT_NE(5, Foo());
+// EXPECT_EQ(NULL, a_pointer);
+// ASSERT_LT(i, array_size);
+// ASSERT_GT(records.size(), 0) << "There is no record left.";
+
+#define EXPECT_EQ(expected, actual) \
+ EXPECT_PRED_FORMAT2(::testing::internal:: \
+ EqHelper<GTEST_IS_NULL_LITERAL_(expected)>::Compare, \
+ expected, actual)
+#define EXPECT_NE(expected, actual) \
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperNE, expected, actual)
+#define EXPECT_LE(val1, val2) \
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2)
+#define EXPECT_LT(val1, val2) \
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2)
+#define EXPECT_GE(val1, val2) \
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2)
+#define EXPECT_GT(val1, val2) \
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2)
+
+#define GTEST_ASSERT_EQ(expected, actual) \
+ ASSERT_PRED_FORMAT2(::testing::internal:: \
+ EqHelper<GTEST_IS_NULL_LITERAL_(expected)>::Compare, \
+ expected, actual)
+#define GTEST_ASSERT_NE(val1, val2) \
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperNE, val1, val2)
+#define GTEST_ASSERT_LE(val1, val2) \
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2)
+#define GTEST_ASSERT_LT(val1, val2) \
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2)
+#define GTEST_ASSERT_GE(val1, val2) \
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2)
+#define GTEST_ASSERT_GT(val1, val2) \
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2)
+
+// Define macro GTEST_DONT_DEFINE_ASSERT_XY to 1 to omit the definition of
+// ASSERT_XY(), which clashes with some users' own code.
+
+#if !GTEST_DONT_DEFINE_ASSERT_EQ
+# define ASSERT_EQ(val1, val2) GTEST_ASSERT_EQ(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_NE
+# define ASSERT_NE(val1, val2) GTEST_ASSERT_NE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_LE
+# define ASSERT_LE(val1, val2) GTEST_ASSERT_LE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_LT
+# define ASSERT_LT(val1, val2) GTEST_ASSERT_LT(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_GE
+# define ASSERT_GE(val1, val2) GTEST_ASSERT_GE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_GT
+# define ASSERT_GT(val1, val2) GTEST_ASSERT_GT(val1, val2)
+#endif
+
+// C String Comparisons. All tests treat NULL and any non-NULL string
+// as different. Two NULLs are equal.
+//
+// * {ASSERT|EXPECT}_STREQ(s1, s2): Tests that s1 == s2
+// * {ASSERT|EXPECT}_STRNE(s1, s2): Tests that s1 != s2
+// * {ASSERT|EXPECT}_STRCASEEQ(s1, s2): Tests that s1 == s2, ignoring case
+// * {ASSERT|EXPECT}_STRCASENE(s1, s2): Tests that s1 != s2, ignoring case
+//
+// For wide or narrow string objects, you can use the
+// {ASSERT|EXPECT}_??() macros.
+//
+// Don't depend on the order in which the arguments are evaluated,
+// which is undefined.
+//
+// These macros evaluate their arguments exactly once.
+
+#define EXPECT_STREQ(expected, actual) \
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, expected, actual)
+#define EXPECT_STRNE(s1, s2) \
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2)
+#define EXPECT_STRCASEEQ(expected, actual) \
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, expected, actual)
+#define EXPECT_STRCASENE(s1, s2)\
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2)
+
+#define ASSERT_STREQ(expected, actual) \
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, expected, actual)
+#define ASSERT_STRNE(s1, s2) \
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2)
+#define ASSERT_STRCASEEQ(expected, actual) \
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, expected, actual)
+#define ASSERT_STRCASENE(s1, s2)\
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2)
+
+// Macros for comparing floating-point numbers.
+//
+// * {ASSERT|EXPECT}_FLOAT_EQ(expected, actual):
+// Tests that two float values are almost equal.
+// * {ASSERT|EXPECT}_DOUBLE_EQ(expected, actual):
+// Tests that two double values are almost equal.
+// * {ASSERT|EXPECT}_NEAR(v1, v2, abs_error):
+// Tests that v1 and v2 are within the given distance to each other.
+//
+// Google Test uses ULP-based comparison to automatically pick a default
+// error bound that is appropriate for the operands. See the
+// FloatingPoint template class in gtest-internal.h if you are
+// interested in the implementation details.
+
+#define EXPECT_FLOAT_EQ(expected, actual)\
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<float>, \
+ expected, actual)
+
+#define EXPECT_DOUBLE_EQ(expected, actual)\
+ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<double>, \
+ expected, actual)
+
+#define ASSERT_FLOAT_EQ(expected, actual)\
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<float>, \
+ expected, actual)
+
+#define ASSERT_DOUBLE_EQ(expected, actual)\
+ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<double>, \
+ expected, actual)
+
+#define EXPECT_NEAR(val1, val2, abs_error)\
+ EXPECT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \
+ val1, val2, abs_error)
+
+#define ASSERT_NEAR(val1, val2, abs_error)\
+ ASSERT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \
+ val1, val2, abs_error)
+
+// These predicate format functions work on floating-point values, and
+// can be used in {ASSERT|EXPECT}_PRED_FORMAT2*(), e.g.
+//
+// EXPECT_PRED_FORMAT2(testing::DoubleLE, Foo(), 5.0);
+
+// Asserts that val1 is less than, or almost equal to, val2. Fails
+// otherwise. In particular, it fails if either val1 or val2 is NaN.
+GTEST_API_ AssertionResult FloatLE(const char* expr1, const char* expr2,
+ float val1, float val2);
+GTEST_API_ AssertionResult DoubleLE(const char* expr1, const char* expr2,
+ double val1, double val2);
+
+
+#if GTEST_OS_WINDOWS
+
+// Macros that test for HRESULT failure and success, these are only useful
+// on Windows, and rely on Windows SDK macros and APIs to compile.
+//
+// * {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED}(expr)
+//
+// When expr unexpectedly fails or succeeds, Google Test prints the
+// expected result and the actual result with both a human-readable
+// string representation of the error, if available, as well as the
+// hex result code.
+# define EXPECT_HRESULT_SUCCEEDED(expr) \
+ EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr))
+
+# define ASSERT_HRESULT_SUCCEEDED(expr) \
+ ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr))
+
+# define EXPECT_HRESULT_FAILED(expr) \
+ EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr))
+
+# define ASSERT_HRESULT_FAILED(expr) \
+ ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr))
+
+#endif // GTEST_OS_WINDOWS
+
+// Macros that execute statement and check that it doesn't generate new fatal
+// failures in the current thread.
+//
+// * {ASSERT|EXPECT}_NO_FATAL_FAILURE(statement);
+//
+// Examples:
+//
+// EXPECT_NO_FATAL_FAILURE(Process());
+// ASSERT_NO_FATAL_FAILURE(Process()) << "Process() failed";
+//
+#define ASSERT_NO_FATAL_FAILURE(statement) \
+ GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_FATAL_FAILURE_)
+#define EXPECT_NO_FATAL_FAILURE(statement) \
+ GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_NONFATAL_FAILURE_)
+
+// Causes a trace (including the source file path, the current line
+// number, and the given message) to be included in every test failure
+// message generated by code in the current scope. The effect is
+// undone when the control leaves the current scope.
+//
+// The message argument can be anything streamable to std::ostream.
+//
+// In the implementation, we include the current line number as part
+// of the dummy variable name, thus allowing multiple SCOPED_TRACE()s
+// to appear in the same block - as long as they are on different
+// lines.
+#define SCOPED_TRACE(message) \
+ ::testing::internal::ScopedTrace GTEST_CONCAT_TOKEN_(gtest_trace_, __LINE__)(\
+ __FILE__, __LINE__, ::testing::Message() << (message))
+
+// Compile-time assertion for type equality.
+// StaticAssertTypeEq<type1, type2>() compiles iff type1 and type2 are
+// the same type. The value it returns is not interesting.
+//
+// Instead of making StaticAssertTypeEq a class template, we make it a
+// function template that invokes a helper class template. This
+// prevents a user from misusing StaticAssertTypeEq<T1, T2> by
+// defining objects of that type.
+//
+// CAVEAT:
+//
+// When used inside a method of a class template,
+// StaticAssertTypeEq<T1, T2>() is effective ONLY IF the method is
+// instantiated. For example, given:
+//
+// template <typename T> class Foo {
+// public:
+// void Bar() { testing::StaticAssertTypeEq<int, T>(); }
+// };
+//
+// the code:
+//
+// void Test1() { Foo<bool> foo; }
+//
+// will NOT generate a compiler error, as Foo<bool>::Bar() is never
+// actually instantiated. Instead, you need:
+//
+// void Test2() { Foo<bool> foo; foo.Bar(); }
+//
+// to cause a compiler error.
+template <typename T1, typename T2>
+bool StaticAssertTypeEq() {
+ (void)internal::StaticAssertTypeEqHelper<T1, T2>();
+ return true;
+}
+
+// Defines a test.
+//
+// The first parameter is the name of the test case, and the second
+// parameter is the name of the test within the test case.
+//
+// The convention is to end the test case name with "Test". For
+// example, a test case for the Foo class can be named FooTest.
+//
+// The user should put his test code between braces after using this
+// macro. Example:
+//
+// TEST(FooTest, InitializesCorrectly) {
+// Foo foo;
+// EXPECT_TRUE(foo.StatusIsOK());
+// }
+
+// Note that we call GetTestTypeId() instead of GetTypeId<
+// ::testing::Test>() here to get the type ID of testing::Test. This
+// is to work around a suspected linker bug when using Google Test as
+// a framework on Mac OS X. The bug causes GetTypeId<
+// ::testing::Test>() to return different values depending on whether
+// the call is from the Google Test framework itself or from user test
+// code. GetTestTypeId() is guaranteed to always return the same
+// value, as it always calls GetTypeId<>() from the Google Test
+// framework.
+#define GTEST_TEST(test_case_name, test_name)\
+ GTEST_TEST_(test_case_name, test_name, \
+ ::testing::Test, ::testing::internal::GetTestTypeId())
+
+// Define this macro to 1 to omit the definition of TEST(), which
+// is a generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_TEST
+# define TEST(test_case_name, test_name) GTEST_TEST(test_case_name, test_name)
+#endif
+
+// Defines a test that uses a test fixture.
+//
+// The first parameter is the name of the test fixture class, which
+// also doubles as the test case name. The second parameter is the
+// name of the test within the test case.
+//
+// A test fixture class must be declared earlier. The user should put
+// his test code between braces after using this macro. Example:
+//
+// class FooTest : public testing::Test {
+// protected:
+// virtual void SetUp() { b_.AddElement(3); }
+//
+// Foo a_;
+// Foo b_;
+// };
+//
+// TEST_F(FooTest, InitializesCorrectly) {
+// EXPECT_TRUE(a_.StatusIsOK());
+// }
+//
+// TEST_F(FooTest, ReturnsElementCountCorrectly) {
+// EXPECT_EQ(0, a_.size());
+// EXPECT_EQ(1, b_.size());
+// }
+
+#define TEST_F(test_fixture, test_name)\
+ GTEST_TEST_(test_fixture, test_name, test_fixture, \
+ ::testing::internal::GetTypeId<test_fixture>())
+
+// Use this macro in main() to run all tests. It returns 0 if all
+// tests are successful, or 1 otherwise.
+//
+// RUN_ALL_TESTS() should be invoked after the command line has been
+// parsed by InitGoogleTest().
+
+#define RUN_ALL_TESTS()\
+ (::testing::UnitTest::GetInstance()->Run())
+
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest_pred_impl.h b/libvpx/third_party/googletest/src/include/gtest/gtest_pred_impl.h
new file mode 100644
index 0000000..3805f85
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest_pred_impl.h
@@ -0,0 +1,358 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// This file is AUTOMATICALLY GENERATED on 09/24/2010 by command
+// 'gen_gtest_pred_impl.py 5'. DO NOT EDIT BY HAND!
+//
+// Implements a family of generic predicate assertion macros.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+
+// Makes sure this header is not included before gtest.h.
+#ifndef GTEST_INCLUDE_GTEST_GTEST_H_
+# error Do not include gtest_pred_impl.h directly. Include gtest.h instead.
+#endif // GTEST_INCLUDE_GTEST_GTEST_H_
+
+// This header implements a family of generic predicate assertion
+// macros:
+//
+// ASSERT_PRED_FORMAT1(pred_format, v1)
+// ASSERT_PRED_FORMAT2(pred_format, v1, v2)
+// ...
+//
+// where pred_format is a function or functor that takes n (in the
+// case of ASSERT_PRED_FORMATn) values and their source expression
+// text, and returns a testing::AssertionResult. See the definition
+// of ASSERT_EQ in gtest.h for an example.
+//
+// If you don't care about formatting, you can use the more
+// restrictive version:
+//
+// ASSERT_PRED1(pred, v1)
+// ASSERT_PRED2(pred, v1, v2)
+// ...
+//
+// where pred is an n-ary function or functor that returns bool,
+// and the values v1, v2, ..., must support the << operator for
+// streaming to std::ostream.
+//
+// We also define the EXPECT_* variations.
+//
+// For now we only support predicates whose arity is at most 5.
+// Please email googletestframework@googlegroups.com if you need
+// support for higher arities.
+
+// GTEST_ASSERT_ is the basic statement to which all of the assertions
+// in this file reduce. Don't use this in your code.
+
+#define GTEST_ASSERT_(expression, on_failure) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (const ::testing::AssertionResult gtest_ar = (expression)) \
+ ; \
+ else \
+ on_failure(gtest_ar.failure_message())
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED1. Don't use
+// this in your code.
+template <typename Pred,
+ typename T1>
+AssertionResult AssertPred1Helper(const char* pred_text,
+ const char* e1,
+ Pred pred,
+ const T1& v1) {
+ if (pred(v1)) return AssertionSuccess();
+
+ return AssertionFailure() << pred_text << "("
+ << e1 << ") evaluates to false, where"
+ << "\n" << e1 << " evaluates to " << v1;
+}
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT1.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT1_(pred_format, v1, on_failure)\
+ GTEST_ASSERT_(pred_format(#v1, v1),\
+ on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED1. Don't use
+// this in your code.
+#define GTEST_PRED1_(pred, v1, on_failure)\
+ GTEST_ASSERT_(::testing::AssertPred1Helper(#pred, \
+ #v1, \
+ pred, \
+ v1), on_failure)
+
+// Unary predicate assertion macros.
+#define EXPECT_PRED_FORMAT1(pred_format, v1) \
+ GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED1(pred, v1) \
+ GTEST_PRED1_(pred, v1, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT1(pred_format, v1) \
+ GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED1(pred, v1) \
+ GTEST_PRED1_(pred, v1, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED2. Don't use
+// this in your code.
+template <typename Pred,
+ typename T1,
+ typename T2>
+AssertionResult AssertPred2Helper(const char* pred_text,
+ const char* e1,
+ const char* e2,
+ Pred pred,
+ const T1& v1,
+ const T2& v2) {
+ if (pred(v1, v2)) return AssertionSuccess();
+
+ return AssertionFailure() << pred_text << "("
+ << e1 << ", "
+ << e2 << ") evaluates to false, where"
+ << "\n" << e1 << " evaluates to " << v1
+ << "\n" << e2 << " evaluates to " << v2;
+}
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT2.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT2_(pred_format, v1, v2, on_failure)\
+ GTEST_ASSERT_(pred_format(#v1, #v2, v1, v2),\
+ on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED2. Don't use
+// this in your code.
+#define GTEST_PRED2_(pred, v1, v2, on_failure)\
+ GTEST_ASSERT_(::testing::AssertPred2Helper(#pred, \
+ #v1, \
+ #v2, \
+ pred, \
+ v1, \
+ v2), on_failure)
+
+// Binary predicate assertion macros.
+#define EXPECT_PRED_FORMAT2(pred_format, v1, v2) \
+ GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED2(pred, v1, v2) \
+ GTEST_PRED2_(pred, v1, v2, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT2(pred_format, v1, v2) \
+ GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED2(pred, v1, v2) \
+ GTEST_PRED2_(pred, v1, v2, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED3. Don't use
+// this in your code.
+template <typename Pred,
+ typename T1,
+ typename T2,
+ typename T3>
+AssertionResult AssertPred3Helper(const char* pred_text,
+ const char* e1,
+ const char* e2,
+ const char* e3,
+ Pred pred,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3) {
+ if (pred(v1, v2, v3)) return AssertionSuccess();
+
+ return AssertionFailure() << pred_text << "("
+ << e1 << ", "
+ << e2 << ", "
+ << e3 << ") evaluates to false, where"
+ << "\n" << e1 << " evaluates to " << v1
+ << "\n" << e2 << " evaluates to " << v2
+ << "\n" << e3 << " evaluates to " << v3;
+}
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT3.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, on_failure)\
+ GTEST_ASSERT_(pred_format(#v1, #v2, #v3, v1, v2, v3),\
+ on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED3. Don't use
+// this in your code.
+#define GTEST_PRED3_(pred, v1, v2, v3, on_failure)\
+ GTEST_ASSERT_(::testing::AssertPred3Helper(#pred, \
+ #v1, \
+ #v2, \
+ #v3, \
+ pred, \
+ v1, \
+ v2, \
+ v3), on_failure)
+
+// Ternary predicate assertion macros.
+#define EXPECT_PRED_FORMAT3(pred_format, v1, v2, v3) \
+ GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED3(pred, v1, v2, v3) \
+ GTEST_PRED3_(pred, v1, v2, v3, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT3(pred_format, v1, v2, v3) \
+ GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED3(pred, v1, v2, v3) \
+ GTEST_PRED3_(pred, v1, v2, v3, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED4. Don't use
+// this in your code.
+template <typename Pred,
+ typename T1,
+ typename T2,
+ typename T3,
+ typename T4>
+AssertionResult AssertPred4Helper(const char* pred_text,
+ const char* e1,
+ const char* e2,
+ const char* e3,
+ const char* e4,
+ Pred pred,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3,
+ const T4& v4) {
+ if (pred(v1, v2, v3, v4)) return AssertionSuccess();
+
+ return AssertionFailure() << pred_text << "("
+ << e1 << ", "
+ << e2 << ", "
+ << e3 << ", "
+ << e4 << ") evaluates to false, where"
+ << "\n" << e1 << " evaluates to " << v1
+ << "\n" << e2 << " evaluates to " << v2
+ << "\n" << e3 << " evaluates to " << v3
+ << "\n" << e4 << " evaluates to " << v4;
+}
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT4.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, on_failure)\
+ GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, v1, v2, v3, v4),\
+ on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED4. Don't use
+// this in your code.
+#define GTEST_PRED4_(pred, v1, v2, v3, v4, on_failure)\
+ GTEST_ASSERT_(::testing::AssertPred4Helper(#pred, \
+ #v1, \
+ #v2, \
+ #v3, \
+ #v4, \
+ pred, \
+ v1, \
+ v2, \
+ v3, \
+ v4), on_failure)
+
+// 4-ary predicate assertion macros.
+#define EXPECT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \
+ GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED4(pred, v1, v2, v3, v4) \
+ GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \
+ GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED4(pred, v1, v2, v3, v4) \
+ GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED5. Don't use
+// this in your code.
+template <typename Pred,
+ typename T1,
+ typename T2,
+ typename T3,
+ typename T4,
+ typename T5>
+AssertionResult AssertPred5Helper(const char* pred_text,
+ const char* e1,
+ const char* e2,
+ const char* e3,
+ const char* e4,
+ const char* e5,
+ Pred pred,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3,
+ const T4& v4,
+ const T5& v5) {
+ if (pred(v1, v2, v3, v4, v5)) return AssertionSuccess();
+
+ return AssertionFailure() << pred_text << "("
+ << e1 << ", "
+ << e2 << ", "
+ << e3 << ", "
+ << e4 << ", "
+ << e5 << ") evaluates to false, where"
+ << "\n" << e1 << " evaluates to " << v1
+ << "\n" << e2 << " evaluates to " << v2
+ << "\n" << e3 << " evaluates to " << v3
+ << "\n" << e4 << " evaluates to " << v4
+ << "\n" << e5 << " evaluates to " << v5;
+}
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT5.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, on_failure)\
+ GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, #v5, v1, v2, v3, v4, v5),\
+ on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED5. Don't use
+// this in your code.
+#define GTEST_PRED5_(pred, v1, v2, v3, v4, v5, on_failure)\
+ GTEST_ASSERT_(::testing::AssertPred5Helper(#pred, \
+ #v1, \
+ #v2, \
+ #v3, \
+ #v4, \
+ #v5, \
+ pred, \
+ v1, \
+ v2, \
+ v3, \
+ v4, \
+ v5), on_failure)
+
+// 5-ary predicate assertion macros.
+#define EXPECT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \
+ GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED5(pred, v1, v2, v3, v4, v5) \
+ GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \
+ GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED5(pred, v1, v2, v3, v4, v5) \
+ GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_)
+
+
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/gtest_prod.h b/libvpx/third_party/googletest/src/include/gtest/gtest_prod.h
new file mode 100644
index 0000000..da80ddc
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/gtest_prod.h
@@ -0,0 +1,58 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Google C++ Testing Framework definitions useful in production code.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PROD_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PROD_H_
+
+// When you need to test the private or protected members of a class,
+// use the FRIEND_TEST macro to declare your tests as friends of the
+// class. For example:
+//
+// class MyClass {
+// private:
+// void MyMethod();
+// FRIEND_TEST(MyClassTest, MyMethod);
+// };
+//
+// class MyClassTest : public testing::Test {
+// // ...
+// };
+//
+// TEST_F(MyClassTest, MyMethod) {
+// // Can call MyClass::MyMethod() here.
+// }
+
+#define FRIEND_TEST(test_case_name, test_name)\
+friend class test_case_name##_##test_name##_Test
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_PROD_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-death-test-internal.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-death-test-internal.h
new file mode 100644
index 0000000..1d9f83b
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-death-test-internal.h
@@ -0,0 +1,308 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines internal utilities needed for implementing
+// death tests. They are subject to change without notice.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
+
+#include "gtest/internal/gtest-internal.h"
+
+#include <stdio.h>
+
+namespace testing {
+namespace internal {
+
+GTEST_DECLARE_string_(internal_run_death_test);
+
+// Names of the flags (needed for parsing Google Test flags).
+const char kDeathTestStyleFlag[] = "death_test_style";
+const char kDeathTestUseFork[] = "death_test_use_fork";
+const char kInternalRunDeathTestFlag[] = "internal_run_death_test";
+
+#if GTEST_HAS_DEATH_TEST
+
+// DeathTest is a class that hides much of the complexity of the
+// GTEST_DEATH_TEST_ macro. It is abstract; its static Create method
+// returns a concrete class that depends on the prevailing death test
+// style, as defined by the --gtest_death_test_style and/or
+// --gtest_internal_run_death_test flags.
+
+// In describing the results of death tests, these terms are used with
+// the corresponding definitions:
+//
+// exit status: The integer exit information in the format specified
+// by wait(2)
+// exit code: The integer code passed to exit(3), _exit(2), or
+// returned from main()
+class GTEST_API_ DeathTest {
+ public:
+ // Create returns false if there was an error determining the
+ // appropriate action to take for the current death test; for example,
+ // if the gtest_death_test_style flag is set to an invalid value.
+ // The LastMessage method will return a more detailed message in that
+ // case. Otherwise, the DeathTest pointer pointed to by the "test"
+ // argument is set. If the death test should be skipped, the pointer
+ // is set to NULL; otherwise, it is set to the address of a new concrete
+ // DeathTest object that controls the execution of the current test.
+ static bool Create(const char* statement, const RE* regex,
+ const char* file, int line, DeathTest** test);
+ DeathTest();
+ virtual ~DeathTest() { }
+
+ // A helper class that aborts a death test when it's deleted.
+ class ReturnSentinel {
+ public:
+ explicit ReturnSentinel(DeathTest* test) : test_(test) { }
+ ~ReturnSentinel() { test_->Abort(TEST_ENCOUNTERED_RETURN_STATEMENT); }
+ private:
+ DeathTest* const test_;
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ReturnSentinel);
+ } GTEST_ATTRIBUTE_UNUSED_;
+
+ // An enumeration of possible roles that may be taken when a death
+ // test is encountered. EXECUTE means that the death test logic should
+ // be executed immediately. OVERSEE means that the program should prepare
+ // the appropriate environment for a child process to execute the death
+ // test, then wait for it to complete.
+ enum TestRole { OVERSEE_TEST, EXECUTE_TEST };
+
+ // An enumeration of the three reasons that a test might be aborted.
+ enum AbortReason {
+ TEST_ENCOUNTERED_RETURN_STATEMENT,
+ TEST_THREW_EXCEPTION,
+ TEST_DID_NOT_DIE
+ };
+
+ // Assumes one of the above roles.
+ virtual TestRole AssumeRole() = 0;
+
+ // Waits for the death test to finish and returns its status.
+ virtual int Wait() = 0;
+
+ // Returns true if the death test passed; that is, the test process
+ // exited during the test, its exit status matches a user-supplied
+ // predicate, and its stderr output matches a user-supplied regular
+ // expression.
+ // The user-supplied predicate may be a macro expression rather
+ // than a function pointer or functor, or else Wait and Passed could
+ // be combined.
+ virtual bool Passed(bool exit_status_ok) = 0;
+
+ // Signals that the death test did not die as expected.
+ virtual void Abort(AbortReason reason) = 0;
+
+ // Returns a human-readable outcome message regarding the outcome of
+ // the last death test.
+ static const char* LastMessage();
+
+ static void set_last_death_test_message(const String& message);
+
+ private:
+ // A string containing a description of the outcome of the last death test.
+ static String last_death_test_message_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(DeathTest);
+};
+
+// Factory interface for death tests. May be mocked out for testing.
+class DeathTestFactory {
+ public:
+ virtual ~DeathTestFactory() { }
+ virtual bool Create(const char* statement, const RE* regex,
+ const char* file, int line, DeathTest** test) = 0;
+};
+
+// A concrete DeathTestFactory implementation for normal use.
+class DefaultDeathTestFactory : public DeathTestFactory {
+ public:
+ virtual bool Create(const char* statement, const RE* regex,
+ const char* file, int line, DeathTest** test);
+};
+
+// Returns true if exit_status describes a process that was terminated
+// by a signal, or exited normally with a nonzero exit code.
+GTEST_API_ bool ExitedUnsuccessfully(int exit_status);
+
+// Traps C++ exceptions escaping statement and reports them as test
+// failures. Note that trapping SEH exceptions is not implemented here.
+# if GTEST_HAS_EXCEPTIONS
+# define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \
+ try { \
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+ } catch (const ::std::exception& gtest_exception) { \
+ fprintf(\
+ stderr, \
+ "\n%s: Caught std::exception-derived exception escaping the " \
+ "death test statement. Exception message: %s\n", \
+ ::testing::internal::FormatFileLocation(__FILE__, __LINE__).c_str(), \
+ gtest_exception.what()); \
+ fflush(stderr); \
+ death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \
+ } catch (...) { \
+ death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \
+ }
+
+# else
+# define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement)
+
+# endif
+
+// This macro is for implementing ASSERT_DEATH*, EXPECT_DEATH*,
+// ASSERT_EXIT*, and EXPECT_EXIT*.
+# define GTEST_DEATH_TEST_(statement, predicate, regex, fail) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (::testing::internal::AlwaysTrue()) { \
+ const ::testing::internal::RE& gtest_regex = (regex); \
+ ::testing::internal::DeathTest* gtest_dt; \
+ if (!::testing::internal::DeathTest::Create(#statement, >est_regex, \
+ __FILE__, __LINE__, >est_dt)) { \
+ goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \
+ } \
+ if (gtest_dt != NULL) { \
+ ::testing::internal::scoped_ptr< ::testing::internal::DeathTest> \
+ gtest_dt_ptr(gtest_dt); \
+ switch (gtest_dt->AssumeRole()) { \
+ case ::testing::internal::DeathTest::OVERSEE_TEST: \
+ if (!gtest_dt->Passed(predicate(gtest_dt->Wait()))) { \
+ goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \
+ } \
+ break; \
+ case ::testing::internal::DeathTest::EXECUTE_TEST: { \
+ ::testing::internal::DeathTest::ReturnSentinel \
+ gtest_sentinel(gtest_dt); \
+ GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, gtest_dt); \
+ gtest_dt->Abort(::testing::internal::DeathTest::TEST_DID_NOT_DIE); \
+ break; \
+ } \
+ default: \
+ break; \
+ } \
+ } \
+ } else \
+ GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__): \
+ fail(::testing::internal::DeathTest::LastMessage())
+// The symbol "fail" here expands to something into which a message
+// can be streamed.
+
+// A class representing the parsed contents of the
+// --gtest_internal_run_death_test flag, as it existed when
+// RUN_ALL_TESTS was called.
+class InternalRunDeathTestFlag {
+ public:
+ InternalRunDeathTestFlag(const String& a_file,
+ int a_line,
+ int an_index,
+ int a_write_fd)
+ : file_(a_file), line_(a_line), index_(an_index),
+ write_fd_(a_write_fd) {}
+
+ ~InternalRunDeathTestFlag() {
+ if (write_fd_ >= 0)
+ posix::Close(write_fd_);
+ }
+
+ String file() const { return file_; }
+ int line() const { return line_; }
+ int index() const { return index_; }
+ int write_fd() const { return write_fd_; }
+
+ private:
+ String file_;
+ int line_;
+ int index_;
+ int write_fd_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(InternalRunDeathTestFlag);
+};
+
+// Returns a newly created InternalRunDeathTestFlag object with fields
+// initialized from the GTEST_FLAG(internal_run_death_test) flag if
+// the flag is specified; otherwise returns NULL.
+InternalRunDeathTestFlag* ParseInternalRunDeathTestFlag();
+
+#else // GTEST_HAS_DEATH_TEST
+
+// This macro is used for implementing macros such as
+// EXPECT_DEATH_IF_SUPPORTED and ASSERT_DEATH_IF_SUPPORTED on systems where
+// death tests are not supported. Those macros must compile on such systems
+// iff EXPECT_DEATH and ASSERT_DEATH compile with the same parameters on
+// systems that support death tests. This allows one to write such a macro
+// on a system that does not support death tests and be sure that it will
+// compile on a death-test supporting system.
+//
+// Parameters:
+// statement - A statement that a macro such as EXPECT_DEATH would test
+// for program termination. This macro has to make sure this
+// statement is compiled but not executed, to ensure that
+// EXPECT_DEATH_IF_SUPPORTED compiles with a certain
+// parameter iff EXPECT_DEATH compiles with it.
+// regex - A regex that a macro such as EXPECT_DEATH would use to test
+// the output of statement. This parameter has to be
+// compiled but not evaluated by this macro, to ensure that
+// this macro only accepts expressions that a macro such as
+// EXPECT_DEATH would accept.
+// terminator - Must be an empty statement for EXPECT_DEATH_IF_SUPPORTED
+// and a return statement for ASSERT_DEATH_IF_SUPPORTED.
+// This ensures that ASSERT_DEATH_IF_SUPPORTED will not
+// compile inside functions where ASSERT_DEATH doesn't
+// compile.
+//
+// The branch that has an always false condition is used to ensure that
+// statement and regex are compiled (and thus syntactically correct) but
+// never executed. The unreachable code macro protects the terminator
+// statement from generating an 'unreachable code' warning in case
+// statement unconditionally returns or throws. The Message constructor at
+// the end allows the syntax of streaming additional messages into the
+// macro, for compilational compatibility with EXPECT_DEATH/ASSERT_DEATH.
+# define GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, terminator) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (::testing::internal::AlwaysTrue()) { \
+ GTEST_LOG_(WARNING) \
+ << "Death tests are not supported on this platform.\n" \
+ << "Statement '" #statement "' cannot be verified."; \
+ } else if (::testing::internal::AlwaysFalse()) { \
+ ::testing::internal::RE::PartialMatch(".*", (regex)); \
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+ terminator; \
+ } else \
+ ::testing::Message()
+
+#endif // GTEST_HAS_DEATH_TEST
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-filepath.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-filepath.h
new file mode 100644
index 0000000..b36b3cf
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-filepath.h
@@ -0,0 +1,210 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: keith.ray@gmail.com (Keith Ray)
+//
+// Google Test filepath utilities
+//
+// This header file declares classes and functions used internally by
+// Google Test. They are subject to change without notice.
+//
+// This file is #included in <gtest/internal/gtest-internal.h>.
+// Do not include this header file separately!
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
+
+#include "gtest/internal/gtest-string.h"
+
+namespace testing {
+namespace internal {
+
+// FilePath - a class for file and directory pathname manipulation which
+// handles platform-specific conventions (like the pathname separator).
+// Used for helper functions for naming files in a directory for xml output.
+// Except for Set methods, all methods are const or static, which provides an
+// "immutable value object" -- useful for peace of mind.
+// A FilePath with a value ending in a path separator ("like/this/") represents
+// a directory, otherwise it is assumed to represent a file. In either case,
+// it may or may not represent an actual file or directory in the file system.
+// Names are NOT checked for syntax correctness -- no checking for illegal
+// characters, malformed paths, etc.
+
+class GTEST_API_ FilePath {
+ public:
+ FilePath() : pathname_("") { }
+ FilePath(const FilePath& rhs) : pathname_(rhs.pathname_) { }
+
+ explicit FilePath(const char* pathname) : pathname_(pathname) {
+ Normalize();
+ }
+
+ explicit FilePath(const String& pathname) : pathname_(pathname) {
+ Normalize();
+ }
+
+ FilePath& operator=(const FilePath& rhs) {
+ Set(rhs);
+ return *this;
+ }
+
+ void Set(const FilePath& rhs) {
+ pathname_ = rhs.pathname_;
+ }
+
+ String ToString() const { return pathname_; }
+ const char* c_str() const { return pathname_.c_str(); }
+
+ // Returns the current working directory, or "" if unsuccessful.
+ static FilePath GetCurrentDir();
+
+ // Given directory = "dir", base_name = "test", number = 0,
+ // extension = "xml", returns "dir/test.xml". If number is greater
+ // than zero (e.g., 12), returns "dir/test_12.xml".
+ // On Windows platform, uses \ as the separator rather than /.
+ static FilePath MakeFileName(const FilePath& directory,
+ const FilePath& base_name,
+ int number,
+ const char* extension);
+
+ // Given directory = "dir", relative_path = "test.xml",
+ // returns "dir/test.xml".
+ // On Windows, uses \ as the separator rather than /.
+ static FilePath ConcatPaths(const FilePath& directory,
+ const FilePath& relative_path);
+
+ // Returns a pathname for a file that does not currently exist. The pathname
+ // will be directory/base_name.extension or
+ // directory/base_name_<number>.extension if directory/base_name.extension
+ // already exists. The number will be incremented until a pathname is found
+ // that does not already exist.
+ // Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'.
+ // There could be a race condition if two or more processes are calling this
+ // function at the same time -- they could both pick the same filename.
+ static FilePath GenerateUniqueFileName(const FilePath& directory,
+ const FilePath& base_name,
+ const char* extension);
+
+ // Returns true iff the path is NULL or "".
+ bool IsEmpty() const { return c_str() == NULL || *c_str() == '\0'; }
+
+ // If input name has a trailing separator character, removes it and returns
+ // the name, otherwise return the name string unmodified.
+ // On Windows platform, uses \ as the separator, other platforms use /.
+ FilePath RemoveTrailingPathSeparator() const;
+
+ // Returns a copy of the FilePath with the directory part removed.
+ // Example: FilePath("path/to/file").RemoveDirectoryName() returns
+ // FilePath("file"). If there is no directory part ("just_a_file"), it returns
+ // the FilePath unmodified. If there is no file part ("just_a_dir/") it
+ // returns an empty FilePath ("").
+ // On Windows platform, '\' is the path separator, otherwise it is '/'.
+ FilePath RemoveDirectoryName() const;
+
+ // RemoveFileName returns the directory path with the filename removed.
+ // Example: FilePath("path/to/file").RemoveFileName() returns "path/to/".
+ // If the FilePath is "a_file" or "/a_file", RemoveFileName returns
+ // FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does
+ // not have a file, like "just/a/dir/", it returns the FilePath unmodified.
+ // On Windows platform, '\' is the path separator, otherwise it is '/'.
+ FilePath RemoveFileName() const;
+
+ // Returns a copy of the FilePath with the case-insensitive extension removed.
+ // Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns
+ // FilePath("dir/file"). If a case-insensitive extension is not
+ // found, returns a copy of the original FilePath.
+ FilePath RemoveExtension(const char* extension) const;
+
+ // Creates directories so that path exists. Returns true if successful or if
+ // the directories already exist; returns false if unable to create
+ // directories for any reason. Will also return false if the FilePath does
+ // not represent a directory (that is, it doesn't end with a path separator).
+ bool CreateDirectoriesRecursively() const;
+
+ // Create the directory so that path exists. Returns true if successful or
+ // if the directory already exists; returns false if unable to create the
+ // directory for any reason, including if the parent directory does not
+ // exist. Not named "CreateDirectory" because that's a macro on Windows.
+ bool CreateFolder() const;
+
+ // Returns true if FilePath describes something in the file-system,
+ // either a file, directory, or whatever, and that something exists.
+ bool FileOrDirectoryExists() const;
+
+ // Returns true if pathname describes a directory in the file-system
+ // that exists.
+ bool DirectoryExists() const;
+
+ // Returns true if FilePath ends with a path separator, which indicates that
+ // it is intended to represent a directory. Returns false otherwise.
+ // This does NOT check that a directory (or file) actually exists.
+ bool IsDirectory() const;
+
+ // Returns true if pathname describes a root directory. (Windows has one
+ // root directory per disk drive.)
+ bool IsRootDirectory() const;
+
+ // Returns true if pathname describes an absolute path.
+ bool IsAbsolutePath() const;
+
+ private:
+ // Replaces multiple consecutive separators with a single separator.
+ // For example, "bar///foo" becomes "bar/foo". Does not eliminate other
+ // redundancies that might be in a pathname involving "." or "..".
+ //
+ // A pathname with multiple consecutive separators may occur either through
+ // user error or as a result of some scripts or APIs that generate a pathname
+ // with a trailing separator. On other platforms the same API or script
+ // may NOT generate a pathname with a trailing "/". Then elsewhere that
+ // pathname may have another "/" and pathname components added to it,
+ // without checking for the separator already being there.
+ // The script language and operating system may allow paths like "foo//bar"
+ // but some of the functions in FilePath will not handle that correctly. In
+ // particular, RemoveTrailingPathSeparator() only removes one separator, and
+ // it is called in CreateDirectoriesRecursively() assuming that it will change
+ // a pathname from directory syntax (trailing separator) to filename syntax.
+ //
+ // On Windows this method also replaces the alternate path separator '/' with
+ // the primary path separator '\\', so that for example "bar\\/\\foo" becomes
+ // "bar\\foo".
+
+ void Normalize();
+
+ // Returns a pointer to the last occurence of a valid path separator in
+ // the FilePath. On Windows, for example, both '/' and '\' are valid path
+ // separators. Returns NULL if no path separator was found.
+ const char* FindLastPathSeparator() const;
+
+ String pathname_;
+}; // class FilePath
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-internal.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-internal.h
new file mode 100644
index 0000000..7aa1197
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-internal.h
@@ -0,0 +1,1226 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file declares functions and macros used internally by
+// Google Test. They are subject to change without notice.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
+
+#include "gtest/internal/gtest-port.h"
+
+#if GTEST_OS_LINUX
+# include <stdlib.h>
+# include <sys/types.h>
+# include <sys/wait.h>
+# include <unistd.h>
+#endif // GTEST_OS_LINUX
+
+#include <ctype.h>
+#include <string.h>
+#include <iomanip>
+#include <limits>
+#include <set>
+
+#include "gtest/internal/gtest-string.h"
+#include "gtest/internal/gtest-filepath.h"
+#include "gtest/internal/gtest-type-util.h"
+
+// Due to C++ preprocessor weirdness, we need double indirection to
+// concatenate two tokens when one of them is __LINE__. Writing
+//
+// foo ## __LINE__
+//
+// will result in the token foo__LINE__, instead of foo followed by
+// the current line number. For more details, see
+// http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.6
+#define GTEST_CONCAT_TOKEN_(foo, bar) GTEST_CONCAT_TOKEN_IMPL_(foo, bar)
+#define GTEST_CONCAT_TOKEN_IMPL_(foo, bar) foo ## bar
+
+// Google Test defines the testing::Message class to allow construction of
+// test messages via the << operator. The idea is that anything
+// streamable to std::ostream can be streamed to a testing::Message.
+// This allows a user to use his own types in Google Test assertions by
+// overloading the << operator.
+//
+// util/gtl/stl_logging-inl.h overloads << for STL containers. These
+// overloads cannot be defined in the std namespace, as that will be
+// undefined behavior. Therefore, they are defined in the global
+// namespace instead.
+//
+// C++'s symbol lookup rule (i.e. Koenig lookup) says that these
+// overloads are visible in either the std namespace or the global
+// namespace, but not other namespaces, including the testing
+// namespace which Google Test's Message class is in.
+//
+// To allow STL containers (and other types that has a << operator
+// defined in the global namespace) to be used in Google Test assertions,
+// testing::Message must access the custom << operator from the global
+// namespace. Hence this helper function.
+//
+// Note: Jeffrey Yasskin suggested an alternative fix by "using
+// ::operator<<;" in the definition of Message's operator<<. That fix
+// doesn't require a helper function, but unfortunately doesn't
+// compile with MSVC.
+template <typename T>
+inline void GTestStreamToHelper(std::ostream* os, const T& val) {
+ *os << val;
+}
+
+class ProtocolMessage;
+namespace proto2 { class Message; }
+
+namespace testing {
+
+// Forward declarations.
+
+class AssertionResult; // Result of an assertion.
+class Message; // Represents a failure message.
+class Test; // Represents a test.
+class TestInfo; // Information about a test.
+class TestPartResult; // Result of a test part.
+class UnitTest; // A collection of test cases.
+
+template <typename T>
+::std::string PrintToString(const T& value);
+
+namespace internal {
+
+struct TraceInfo; // Information about a trace point.
+class ScopedTrace; // Implements scoped trace.
+class TestInfoImpl; // Opaque implementation of TestInfo
+class UnitTestImpl; // Opaque implementation of UnitTest
+
+// How many times InitGoogleTest() has been called.
+extern int g_init_gtest_count;
+
+// The text used in failure messages to indicate the start of the
+// stack trace.
+GTEST_API_ extern const char kStackTraceMarker[];
+
+// A secret type that Google Test users don't know about. It has no
+// definition on purpose. Therefore it's impossible to create a
+// Secret object, which is what we want.
+class Secret;
+
+// Two overloaded helpers for checking at compile time whether an
+// expression is a null pointer literal (i.e. NULL or any 0-valued
+// compile-time integral constant). Their return values have
+// different sizes, so we can use sizeof() to test which version is
+// picked by the compiler. These helpers have no implementations, as
+// we only need their signatures.
+//
+// Given IsNullLiteralHelper(x), the compiler will pick the first
+// version if x can be implicitly converted to Secret*, and pick the
+// second version otherwise. Since Secret is a secret and incomplete
+// type, the only expression a user can write that has type Secret* is
+// a null pointer literal. Therefore, we know that x is a null
+// pointer literal if and only if the first version is picked by the
+// compiler.
+char IsNullLiteralHelper(Secret* p);
+char (&IsNullLiteralHelper(...))[2]; // NOLINT
+
+// A compile-time bool constant that is true if and only if x is a
+// null pointer literal (i.e. NULL or any 0-valued compile-time
+// integral constant).
+#ifdef GTEST_ELLIPSIS_NEEDS_POD_
+// We lose support for NULL detection where the compiler doesn't like
+// passing non-POD classes through ellipsis (...).
+# define GTEST_IS_NULL_LITERAL_(x) false
+#else
+# define GTEST_IS_NULL_LITERAL_(x) \
+ (sizeof(::testing::internal::IsNullLiteralHelper(x)) == 1)
+#endif // GTEST_ELLIPSIS_NEEDS_POD_
+
+// Appends the user-supplied message to the Google-Test-generated message.
+GTEST_API_ String AppendUserMessage(const String& gtest_msg,
+ const Message& user_msg);
+
+// A helper class for creating scoped traces in user programs.
+class GTEST_API_ ScopedTrace {
+ public:
+ // The c'tor pushes the given source file location and message onto
+ // a trace stack maintained by Google Test.
+ ScopedTrace(const char* file, int line, const Message& message);
+
+ // The d'tor pops the info pushed by the c'tor.
+ //
+ // Note that the d'tor is not virtual in order to be efficient.
+ // Don't inherit from ScopedTrace!
+ ~ScopedTrace();
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedTrace);
+} GTEST_ATTRIBUTE_UNUSED_; // A ScopedTrace object does its job in its
+ // c'tor and d'tor. Therefore it doesn't
+ // need to be used otherwise.
+
+// Converts a streamable value to a String. A NULL pointer is
+// converted to "(null)". When the input value is a ::string,
+// ::std::string, ::wstring, or ::std::wstring object, each NUL
+// character in it is replaced with "\\0".
+// Declared here but defined in gtest.h, so that it has access
+// to the definition of the Message class, required by the ARM
+// compiler.
+template <typename T>
+String StreamableToString(const T& streamable);
+
+// The Symbian compiler has a bug that prevents it from selecting the
+// correct overload of FormatForComparisonFailureMessage (see below)
+// unless we pass the first argument by reference. If we do that,
+// however, Visual Age C++ 10.1 generates a compiler error. Therefore
+// we only apply the work-around for Symbian.
+#if defined(__SYMBIAN32__)
+# define GTEST_CREF_WORKAROUND_ const&
+#else
+# define GTEST_CREF_WORKAROUND_
+#endif
+
+// When this operand is a const char* or char*, if the other operand
+// is a ::std::string or ::string, we print this operand as a C string
+// rather than a pointer (we do the same for wide strings); otherwise
+// we print it as a pointer to be safe.
+
+// This internal macro is used to avoid duplicated code.
+#define GTEST_FORMAT_IMPL_(operand2_type, operand1_printer)\
+inline String FormatForComparisonFailureMessage(\
+ operand2_type::value_type* GTEST_CREF_WORKAROUND_ str, \
+ const operand2_type& /*operand2*/) {\
+ return operand1_printer(str);\
+}\
+inline String FormatForComparisonFailureMessage(\
+ const operand2_type::value_type* GTEST_CREF_WORKAROUND_ str, \
+ const operand2_type& /*operand2*/) {\
+ return operand1_printer(str);\
+}
+
+GTEST_FORMAT_IMPL_(::std::string, String::ShowCStringQuoted)
+#if GTEST_HAS_STD_WSTRING
+GTEST_FORMAT_IMPL_(::std::wstring, String::ShowWideCStringQuoted)
+#endif // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_GLOBAL_STRING
+GTEST_FORMAT_IMPL_(::string, String::ShowCStringQuoted)
+#endif // GTEST_HAS_GLOBAL_STRING
+#if GTEST_HAS_GLOBAL_WSTRING
+GTEST_FORMAT_IMPL_(::wstring, String::ShowWideCStringQuoted)
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+#undef GTEST_FORMAT_IMPL_
+
+// The next four overloads handle the case where the operand being
+// printed is a char/wchar_t pointer and the other operand is not a
+// string/wstring object. In such cases, we just print the operand as
+// a pointer to be safe.
+#define GTEST_FORMAT_CHAR_PTR_IMPL_(CharType) \
+ template <typename T> \
+ String FormatForComparisonFailureMessage(CharType* GTEST_CREF_WORKAROUND_ p, \
+ const T&) { \
+ return PrintToString(static_cast<const void*>(p)); \
+ }
+
+GTEST_FORMAT_CHAR_PTR_IMPL_(char)
+GTEST_FORMAT_CHAR_PTR_IMPL_(const char)
+GTEST_FORMAT_CHAR_PTR_IMPL_(wchar_t)
+GTEST_FORMAT_CHAR_PTR_IMPL_(const wchar_t)
+
+#undef GTEST_FORMAT_CHAR_PTR_IMPL_
+
+// Constructs and returns the message for an equality assertion
+// (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure.
+//
+// The first four parameters are the expressions used in the assertion
+// and their values, as strings. For example, for ASSERT_EQ(foo, bar)
+// where foo is 5 and bar is 6, we have:
+//
+// expected_expression: "foo"
+// actual_expression: "bar"
+// expected_value: "5"
+// actual_value: "6"
+//
+// The ignoring_case parameter is true iff the assertion is a
+// *_STRCASEEQ*. When it's true, the string " (ignoring case)" will
+// be inserted into the message.
+GTEST_API_ AssertionResult EqFailure(const char* expected_expression,
+ const char* actual_expression,
+ const String& expected_value,
+ const String& actual_value,
+ bool ignoring_case);
+
+// Constructs a failure message for Boolean assertions such as EXPECT_TRUE.
+GTEST_API_ String GetBoolAssertionFailureMessage(
+ const AssertionResult& assertion_result,
+ const char* expression_text,
+ const char* actual_predicate_value,
+ const char* expected_predicate_value);
+
+// This template class represents an IEEE floating-point number
+// (either single-precision or double-precision, depending on the
+// template parameters).
+//
+// The purpose of this class is to do more sophisticated number
+// comparison. (Due to round-off error, etc, it's very unlikely that
+// two floating-points will be equal exactly. Hence a naive
+// comparison by the == operation often doesn't work.)
+//
+// Format of IEEE floating-point:
+//
+// The most-significant bit being the leftmost, an IEEE
+// floating-point looks like
+//
+// sign_bit exponent_bits fraction_bits
+//
+// Here, sign_bit is a single bit that designates the sign of the
+// number.
+//
+// For float, there are 8 exponent bits and 23 fraction bits.
+//
+// For double, there are 11 exponent bits and 52 fraction bits.
+//
+// More details can be found at
+// http://en.wikipedia.org/wiki/IEEE_floating-point_standard.
+//
+// Template parameter:
+//
+// RawType: the raw floating-point type (either float or double)
+template <typename RawType>
+class FloatingPoint {
+ public:
+ // Defines the unsigned integer type that has the same size as the
+ // floating point number.
+ typedef typename TypeWithSize<sizeof(RawType)>::UInt Bits;
+
+ // Constants.
+
+ // # of bits in a number.
+ static const size_t kBitCount = 8*sizeof(RawType);
+
+ // # of fraction bits in a number.
+ static const size_t kFractionBitCount =
+ std::numeric_limits<RawType>::digits - 1;
+
+ // # of exponent bits in a number.
+ static const size_t kExponentBitCount = kBitCount - 1 - kFractionBitCount;
+
+ // The mask for the sign bit.
+ static const Bits kSignBitMask = static_cast<Bits>(1) << (kBitCount - 1);
+
+ // The mask for the fraction bits.
+ static const Bits kFractionBitMask =
+ ~static_cast<Bits>(0) >> (kExponentBitCount + 1);
+
+ // The mask for the exponent bits.
+ static const Bits kExponentBitMask = ~(kSignBitMask | kFractionBitMask);
+
+ // How many ULP's (Units in the Last Place) we want to tolerate when
+ // comparing two numbers. The larger the value, the more error we
+ // allow. A 0 value means that two numbers must be exactly the same
+ // to be considered equal.
+ //
+ // The maximum error of a single floating-point operation is 0.5
+ // units in the last place. On Intel CPU's, all floating-point
+ // calculations are done with 80-bit precision, while double has 64
+ // bits. Therefore, 4 should be enough for ordinary use.
+ //
+ // See the following article for more details on ULP:
+ // http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm.
+ static const size_t kMaxUlps = 4;
+
+ // Constructs a FloatingPoint from a raw floating-point number.
+ //
+ // On an Intel CPU, passing a non-normalized NAN (Not a Number)
+ // around may change its bits, although the new value is guaranteed
+ // to be also a NAN. Therefore, don't expect this constructor to
+ // preserve the bits in x when x is a NAN.
+ explicit FloatingPoint(const RawType& x) { u_.value_ = x; }
+
+ // Static methods
+
+ // Reinterprets a bit pattern as a floating-point number.
+ //
+ // This function is needed to test the AlmostEquals() method.
+ static RawType ReinterpretBits(const Bits bits) {
+ FloatingPoint fp(0);
+ fp.u_.bits_ = bits;
+ return fp.u_.value_;
+ }
+
+ // Returns the floating-point number that represent positive infinity.
+ static RawType Infinity() {
+ return ReinterpretBits(kExponentBitMask);
+ }
+
+ // Non-static methods
+
+ // Returns the bits that represents this number.
+ const Bits &bits() const { return u_.bits_; }
+
+ // Returns the exponent bits of this number.
+ Bits exponent_bits() const { return kExponentBitMask & u_.bits_; }
+
+ // Returns the fraction bits of this number.
+ Bits fraction_bits() const { return kFractionBitMask & u_.bits_; }
+
+ // Returns the sign bit of this number.
+ Bits sign_bit() const { return kSignBitMask & u_.bits_; }
+
+ // Returns true iff this is NAN (not a number).
+ bool is_nan() const {
+ // It's a NAN if the exponent bits are all ones and the fraction
+ // bits are not entirely zeros.
+ return (exponent_bits() == kExponentBitMask) && (fraction_bits() != 0);
+ }
+
+ // Returns true iff this number is at most kMaxUlps ULP's away from
+ // rhs. In particular, this function:
+ //
+ // - returns false if either number is (or both are) NAN.
+ // - treats really large numbers as almost equal to infinity.
+ // - thinks +0.0 and -0.0 are 0 DLP's apart.
+ bool AlmostEquals(const FloatingPoint& rhs) const {
+ // The IEEE standard says that any comparison operation involving
+ // a NAN must return false.
+ if (is_nan() || rhs.is_nan()) return false;
+
+ return DistanceBetweenSignAndMagnitudeNumbers(u_.bits_, rhs.u_.bits_)
+ <= kMaxUlps;
+ }
+
+ private:
+ // The data type used to store the actual floating-point number.
+ union FloatingPointUnion {
+ RawType value_; // The raw floating-point number.
+ Bits bits_; // The bits that represent the number.
+ };
+
+ // Converts an integer from the sign-and-magnitude representation to
+ // the biased representation. More precisely, let N be 2 to the
+ // power of (kBitCount - 1), an integer x is represented by the
+ // unsigned number x + N.
+ //
+ // For instance,
+ //
+ // -N + 1 (the most negative number representable using
+ // sign-and-magnitude) is represented by 1;
+ // 0 is represented by N; and
+ // N - 1 (the biggest number representable using
+ // sign-and-magnitude) is represented by 2N - 1.
+ //
+ // Read http://en.wikipedia.org/wiki/Signed_number_representations
+ // for more details on signed number representations.
+ static Bits SignAndMagnitudeToBiased(const Bits &sam) {
+ if (kSignBitMask & sam) {
+ // sam represents a negative number.
+ return ~sam + 1;
+ } else {
+ // sam represents a positive number.
+ return kSignBitMask | sam;
+ }
+ }
+
+ // Given two numbers in the sign-and-magnitude representation,
+ // returns the distance between them as an unsigned number.
+ static Bits DistanceBetweenSignAndMagnitudeNumbers(const Bits &sam1,
+ const Bits &sam2) {
+ const Bits biased1 = SignAndMagnitudeToBiased(sam1);
+ const Bits biased2 = SignAndMagnitudeToBiased(sam2);
+ return (biased1 >= biased2) ? (biased1 - biased2) : (biased2 - biased1);
+ }
+
+ FloatingPointUnion u_;
+};
+
+// Typedefs the instances of the FloatingPoint template class that we
+// care to use.
+typedef FloatingPoint<float> Float;
+typedef FloatingPoint<double> Double;
+
+// In order to catch the mistake of putting tests that use different
+// test fixture classes in the same test case, we need to assign
+// unique IDs to fixture classes and compare them. The TypeId type is
+// used to hold such IDs. The user should treat TypeId as an opaque
+// type: the only operation allowed on TypeId values is to compare
+// them for equality using the == operator.
+typedef const void* TypeId;
+
+template <typename T>
+class TypeIdHelper {
+ public:
+ // dummy_ must not have a const type. Otherwise an overly eager
+ // compiler (e.g. MSVC 7.1 & 8.0) may try to merge
+ // TypeIdHelper<T>::dummy_ for different Ts as an "optimization".
+ static bool dummy_;
+};
+
+template <typename T>
+bool TypeIdHelper<T>::dummy_ = false;
+
+// GetTypeId<T>() returns the ID of type T. Different values will be
+// returned for different types. Calling the function twice with the
+// same type argument is guaranteed to return the same ID.
+template <typename T>
+TypeId GetTypeId() {
+ // The compiler is required to allocate a different
+ // TypeIdHelper<T>::dummy_ variable for each T used to instantiate
+ // the template. Therefore, the address of dummy_ is guaranteed to
+ // be unique.
+ return &(TypeIdHelper<T>::dummy_);
+}
+
+// Returns the type ID of ::testing::Test. Always call this instead
+// of GetTypeId< ::testing::Test>() to get the type ID of
+// ::testing::Test, as the latter may give the wrong result due to a
+// suspected linker bug when compiling Google Test as a Mac OS X
+// framework.
+GTEST_API_ TypeId GetTestTypeId();
+
+// Defines the abstract factory interface that creates instances
+// of a Test object.
+class TestFactoryBase {
+ public:
+ virtual ~TestFactoryBase() {}
+
+ // Creates a test instance to run. The instance is both created and destroyed
+ // within TestInfoImpl::Run()
+ virtual Test* CreateTest() = 0;
+
+ protected:
+ TestFactoryBase() {}
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestFactoryBase);
+};
+
+// This class provides implementation of TeastFactoryBase interface.
+// It is used in TEST and TEST_F macros.
+template <class TestClass>
+class TestFactoryImpl : public TestFactoryBase {
+ public:
+ virtual Test* CreateTest() { return new TestClass; }
+};
+
+#if GTEST_OS_WINDOWS
+
+// Predicate-formatters for implementing the HRESULT checking macros
+// {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED}
+// We pass a long instead of HRESULT to avoid causing an
+// include dependency for the HRESULT type.
+GTEST_API_ AssertionResult IsHRESULTSuccess(const char* expr,
+ long hr); // NOLINT
+GTEST_API_ AssertionResult IsHRESULTFailure(const char* expr,
+ long hr); // NOLINT
+
+#endif // GTEST_OS_WINDOWS
+
+// Types of SetUpTestCase() and TearDownTestCase() functions.
+typedef void (*SetUpTestCaseFunc)();
+typedef void (*TearDownTestCaseFunc)();
+
+// Creates a new TestInfo object and registers it with Google Test;
+// returns the created object.
+//
+// Arguments:
+//
+// test_case_name: name of the test case
+// name: name of the test
+// type_param the name of the test's type parameter, or NULL if
+// this is not a typed or a type-parameterized test.
+// value_param text representation of the test's value parameter,
+// or NULL if this is not a type-parameterized test.
+// fixture_class_id: ID of the test fixture class
+// set_up_tc: pointer to the function that sets up the test case
+// tear_down_tc: pointer to the function that tears down the test case
+// factory: pointer to the factory that creates a test object.
+// The newly created TestInfo instance will assume
+// ownership of the factory object.
+GTEST_API_ TestInfo* MakeAndRegisterTestInfo(
+ const char* test_case_name, const char* name,
+ const char* type_param,
+ const char* value_param,
+ TypeId fixture_class_id,
+ SetUpTestCaseFunc set_up_tc,
+ TearDownTestCaseFunc tear_down_tc,
+ TestFactoryBase* factory);
+
+// If *pstr starts with the given prefix, modifies *pstr to be right
+// past the prefix and returns true; otherwise leaves *pstr unchanged
+// and returns false. None of pstr, *pstr, and prefix can be NULL.
+GTEST_API_ bool SkipPrefix(const char* prefix, const char** pstr);
+
+#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// State of the definition of a type-parameterized test case.
+class GTEST_API_ TypedTestCasePState {
+ public:
+ TypedTestCasePState() : registered_(false) {}
+
+ // Adds the given test name to defined_test_names_ and return true
+ // if the test case hasn't been registered; otherwise aborts the
+ // program.
+ bool AddTestName(const char* file, int line, const char* case_name,
+ const char* test_name) {
+ if (registered_) {
+ fprintf(stderr, "%s Test %s must be defined before "
+ "REGISTER_TYPED_TEST_CASE_P(%s, ...).\n",
+ FormatFileLocation(file, line).c_str(), test_name, case_name);
+ fflush(stderr);
+ posix::Abort();
+ }
+ defined_test_names_.insert(test_name);
+ return true;
+ }
+
+ // Verifies that registered_tests match the test names in
+ // defined_test_names_; returns registered_tests if successful, or
+ // aborts the program otherwise.
+ const char* VerifyRegisteredTestNames(
+ const char* file, int line, const char* registered_tests);
+
+ private:
+ bool registered_;
+ ::std::set<const char*> defined_test_names_;
+};
+
+// Skips to the first non-space char after the first comma in 'str';
+// returns NULL if no comma is found in 'str'.
+inline const char* SkipComma(const char* str) {
+ const char* comma = strchr(str, ',');
+ if (comma == NULL) {
+ return NULL;
+ }
+ while (IsSpace(*(++comma))) {}
+ return comma;
+}
+
+// Returns the prefix of 'str' before the first comma in it; returns
+// the entire string if it contains no comma.
+inline String GetPrefixUntilComma(const char* str) {
+ const char* comma = strchr(str, ',');
+ return comma == NULL ? String(str) : String(str, comma - str);
+}
+
+// TypeParameterizedTest<Fixture, TestSel, Types>::Register()
+// registers a list of type-parameterized tests with Google Test. The
+// return value is insignificant - we just need to return something
+// such that we can call this function in a namespace scope.
+//
+// Implementation note: The GTEST_TEMPLATE_ macro declares a template
+// template parameter. It's defined in gtest-type-util.h.
+template <GTEST_TEMPLATE_ Fixture, class TestSel, typename Types>
+class TypeParameterizedTest {
+ public:
+ // 'index' is the index of the test in the type list 'Types'
+ // specified in INSTANTIATE_TYPED_TEST_CASE_P(Prefix, TestCase,
+ // Types). Valid values for 'index' are [0, N - 1] where N is the
+ // length of Types.
+ static bool Register(const char* prefix, const char* case_name,
+ const char* test_names, int index) {
+ typedef typename Types::Head Type;
+ typedef Fixture<Type> FixtureClass;
+ typedef typename GTEST_BIND_(TestSel, Type) TestClass;
+
+ // First, registers the first type-parameterized test in the type
+ // list.
+ MakeAndRegisterTestInfo(
+ String::Format("%s%s%s/%d", prefix, prefix[0] == '\0' ? "" : "/",
+ case_name, index).c_str(),
+ GetPrefixUntilComma(test_names).c_str(),
+ GetTypeName<Type>().c_str(),
+ NULL, // No value parameter.
+ GetTypeId<FixtureClass>(),
+ TestClass::SetUpTestCase,
+ TestClass::TearDownTestCase,
+ new TestFactoryImpl<TestClass>);
+
+ // Next, recurses (at compile time) with the tail of the type list.
+ return TypeParameterizedTest<Fixture, TestSel, typename Types::Tail>
+ ::Register(prefix, case_name, test_names, index + 1);
+ }
+};
+
+// The base case for the compile time recursion.
+template <GTEST_TEMPLATE_ Fixture, class TestSel>
+class TypeParameterizedTest<Fixture, TestSel, Types0> {
+ public:
+ static bool Register(const char* /*prefix*/, const char* /*case_name*/,
+ const char* /*test_names*/, int /*index*/) {
+ return true;
+ }
+};
+
+// TypeParameterizedTestCase<Fixture, Tests, Types>::Register()
+// registers *all combinations* of 'Tests' and 'Types' with Google
+// Test. The return value is insignificant - we just need to return
+// something such that we can call this function in a namespace scope.
+template <GTEST_TEMPLATE_ Fixture, typename Tests, typename Types>
+class TypeParameterizedTestCase {
+ public:
+ static bool Register(const char* prefix, const char* case_name,
+ const char* test_names) {
+ typedef typename Tests::Head Head;
+
+ // First, register the first test in 'Test' for each type in 'Types'.
+ TypeParameterizedTest<Fixture, Head, Types>::Register(
+ prefix, case_name, test_names, 0);
+
+ // Next, recurses (at compile time) with the tail of the test list.
+ return TypeParameterizedTestCase<Fixture, typename Tests::Tail, Types>
+ ::Register(prefix, case_name, SkipComma(test_names));
+ }
+};
+
+// The base case for the compile time recursion.
+template <GTEST_TEMPLATE_ Fixture, typename Types>
+class TypeParameterizedTestCase<Fixture, Templates0, Types> {
+ public:
+ static bool Register(const char* /*prefix*/, const char* /*case_name*/,
+ const char* /*test_names*/) {
+ return true;
+ }
+};
+
+#endif // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// Returns the current OS stack trace as a String.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag. The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in
+// the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't.
+GTEST_API_ String GetCurrentOsStackTraceExceptTop(UnitTest* unit_test,
+ int skip_count);
+
+// Helpers for suppressing warnings on unreachable code or constant
+// condition.
+
+// Always returns true.
+GTEST_API_ bool AlwaysTrue();
+
+// Always returns false.
+inline bool AlwaysFalse() { return !AlwaysTrue(); }
+
+// Helper for suppressing false warning from Clang on a const char*
+// variable declared in a conditional expression always being NULL in
+// the else branch.
+struct GTEST_API_ ConstCharPtr {
+ ConstCharPtr(const char* str) : value(str) {}
+ operator bool() const { return true; }
+ const char* value;
+};
+
+// A simple Linear Congruential Generator for generating random
+// numbers with a uniform distribution. Unlike rand() and srand(), it
+// doesn't use global state (and therefore can't interfere with user
+// code). Unlike rand_r(), it's portable. An LCG isn't very random,
+// but it's good enough for our purposes.
+class GTEST_API_ Random {
+ public:
+ static const UInt32 kMaxRange = 1u << 31;
+
+ explicit Random(UInt32 seed) : state_(seed) {}
+
+ void Reseed(UInt32 seed) { state_ = seed; }
+
+ // Generates a random number from [0, range). Crashes if 'range' is
+ // 0 or greater than kMaxRange.
+ UInt32 Generate(UInt32 range);
+
+ private:
+ UInt32 state_;
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(Random);
+};
+
+// Defining a variable of type CompileAssertTypesEqual<T1, T2> will cause a
+// compiler error iff T1 and T2 are different types.
+template <typename T1, typename T2>
+struct CompileAssertTypesEqual;
+
+template <typename T>
+struct CompileAssertTypesEqual<T, T> {
+};
+
+// Removes the reference from a type if it is a reference type,
+// otherwise leaves it unchanged. This is the same as
+// tr1::remove_reference, which is not widely available yet.
+template <typename T>
+struct RemoveReference { typedef T type; }; // NOLINT
+template <typename T>
+struct RemoveReference<T&> { typedef T type; }; // NOLINT
+
+// A handy wrapper around RemoveReference that works when the argument
+// T depends on template parameters.
+#define GTEST_REMOVE_REFERENCE_(T) \
+ typename ::testing::internal::RemoveReference<T>::type
+
+// Removes const from a type if it is a const type, otherwise leaves
+// it unchanged. This is the same as tr1::remove_const, which is not
+// widely available yet.
+template <typename T>
+struct RemoveConst { typedef T type; }; // NOLINT
+template <typename T>
+struct RemoveConst<const T> { typedef T type; }; // NOLINT
+
+// MSVC 8.0, Sun C++, and IBM XL C++ have a bug which causes the above
+// definition to fail to remove the const in 'const int[3]' and 'const
+// char[3][4]'. The following specialization works around the bug.
+// However, it causes trouble with GCC and thus needs to be
+// conditionally compiled.
+#if defined(_MSC_VER) || defined(__SUNPRO_CC) || defined(__IBMCPP__)
+template <typename T, size_t N>
+struct RemoveConst<const T[N]> {
+ typedef typename RemoveConst<T>::type type[N];
+};
+#endif
+
+// A handy wrapper around RemoveConst that works when the argument
+// T depends on template parameters.
+#define GTEST_REMOVE_CONST_(T) \
+ typename ::testing::internal::RemoveConst<T>::type
+
+// Turns const U&, U&, const U, and U all into U.
+#define GTEST_REMOVE_REFERENCE_AND_CONST_(T) \
+ GTEST_REMOVE_CONST_(GTEST_REMOVE_REFERENCE_(T))
+
+// Adds reference to a type if it is not a reference type,
+// otherwise leaves it unchanged. This is the same as
+// tr1::add_reference, which is not widely available yet.
+template <typename T>
+struct AddReference { typedef T& type; }; // NOLINT
+template <typename T>
+struct AddReference<T&> { typedef T& type; }; // NOLINT
+
+// A handy wrapper around AddReference that works when the argument T
+// depends on template parameters.
+#define GTEST_ADD_REFERENCE_(T) \
+ typename ::testing::internal::AddReference<T>::type
+
+// Adds a reference to const on top of T as necessary. For example,
+// it transforms
+//
+// char ==> const char&
+// const char ==> const char&
+// char& ==> const char&
+// const char& ==> const char&
+//
+// The argument T must depend on some template parameters.
+#define GTEST_REFERENCE_TO_CONST_(T) \
+ GTEST_ADD_REFERENCE_(const GTEST_REMOVE_REFERENCE_(T))
+
+// ImplicitlyConvertible<From, To>::value is a compile-time bool
+// constant that's true iff type From can be implicitly converted to
+// type To.
+template <typename From, typename To>
+class ImplicitlyConvertible {
+ private:
+ // We need the following helper functions only for their types.
+ // They have no implementations.
+
+ // MakeFrom() is an expression whose type is From. We cannot simply
+ // use From(), as the type From may not have a public default
+ // constructor.
+ static From MakeFrom();
+
+ // These two functions are overloaded. Given an expression
+ // Helper(x), the compiler will pick the first version if x can be
+ // implicitly converted to type To; otherwise it will pick the
+ // second version.
+ //
+ // The first version returns a value of size 1, and the second
+ // version returns a value of size 2. Therefore, by checking the
+ // size of Helper(x), which can be done at compile time, we can tell
+ // which version of Helper() is used, and hence whether x can be
+ // implicitly converted to type To.
+ static char Helper(To);
+ static char (&Helper(...))[2]; // NOLINT
+
+ // We have to put the 'public' section after the 'private' section,
+ // or MSVC refuses to compile the code.
+ public:
+ // MSVC warns about implicitly converting from double to int for
+ // possible loss of data, so we need to temporarily disable the
+ // warning.
+#ifdef _MSC_VER
+# pragma warning(push) // Saves the current warning state.
+# pragma warning(disable:4244) // Temporarily disables warning 4244.
+
+ static const bool value =
+ sizeof(Helper(ImplicitlyConvertible::MakeFrom())) == 1;
+# pragma warning(pop) // Restores the warning state.
+#elif defined(__BORLANDC__)
+ // C++Builder cannot use member overload resolution during template
+ // instantiation. The simplest workaround is to use its C++0x type traits
+ // functions (C++Builder 2009 and above only).
+ static const bool value = __is_convertible(From, To);
+#else
+ static const bool value =
+ sizeof(Helper(ImplicitlyConvertible::MakeFrom())) == 1;
+#endif // _MSV_VER
+};
+template <typename From, typename To>
+const bool ImplicitlyConvertible<From, To>::value;
+
+// IsAProtocolMessage<T>::value is a compile-time bool constant that's
+// true iff T is type ProtocolMessage, proto2::Message, or a subclass
+// of those.
+template <typename T>
+struct IsAProtocolMessage
+ : public bool_constant<
+ ImplicitlyConvertible<const T*, const ::ProtocolMessage*>::value ||
+ ImplicitlyConvertible<const T*, const ::proto2::Message*>::value> {
+};
+
+// When the compiler sees expression IsContainerTest<C>(0), if C is an
+// STL-style container class, the first overload of IsContainerTest
+// will be viable (since both C::iterator* and C::const_iterator* are
+// valid types and NULL can be implicitly converted to them). It will
+// be picked over the second overload as 'int' is a perfect match for
+// the type of argument 0. If C::iterator or C::const_iterator is not
+// a valid type, the first overload is not viable, and the second
+// overload will be picked. Therefore, we can determine whether C is
+// a container class by checking the type of IsContainerTest<C>(0).
+// The value of the expression is insignificant.
+//
+// Note that we look for both C::iterator and C::const_iterator. The
+// reason is that C++ injects the name of a class as a member of the
+// class itself (e.g. you can refer to class iterator as either
+// 'iterator' or 'iterator::iterator'). If we look for C::iterator
+// only, for example, we would mistakenly think that a class named
+// iterator is an STL container.
+//
+// Also note that the simpler approach of overloading
+// IsContainerTest(typename C::const_iterator*) and
+// IsContainerTest(...) doesn't work with Visual Age C++ and Sun C++.
+typedef int IsContainer;
+template <class C>
+IsContainer IsContainerTest(int /* dummy */,
+ typename C::iterator* /* it */ = NULL,
+ typename C::const_iterator* /* const_it */ = NULL) {
+ return 0;
+}
+
+typedef char IsNotContainer;
+template <class C>
+IsNotContainer IsContainerTest(long /* dummy */) { return '\0'; }
+
+// EnableIf<condition>::type is void when 'Cond' is true, and
+// undefined when 'Cond' is false. To use SFINAE to make a function
+// overload only apply when a particular expression is true, add
+// "typename EnableIf<expression>::type* = 0" as the last parameter.
+template<bool> struct EnableIf;
+template<> struct EnableIf<true> { typedef void type; }; // NOLINT
+
+// Utilities for native arrays.
+
+// ArrayEq() compares two k-dimensional native arrays using the
+// elements' operator==, where k can be any integer >= 0. When k is
+// 0, ArrayEq() degenerates into comparing a single pair of values.
+
+template <typename T, typename U>
+bool ArrayEq(const T* lhs, size_t size, const U* rhs);
+
+// This generic version is used when k is 0.
+template <typename T, typename U>
+inline bool ArrayEq(const T& lhs, const U& rhs) { return lhs == rhs; }
+
+// This overload is used when k >= 1.
+template <typename T, typename U, size_t N>
+inline bool ArrayEq(const T(&lhs)[N], const U(&rhs)[N]) {
+ return internal::ArrayEq(lhs, N, rhs);
+}
+
+// This helper reduces code bloat. If we instead put its logic inside
+// the previous ArrayEq() function, arrays with different sizes would
+// lead to different copies of the template code.
+template <typename T, typename U>
+bool ArrayEq(const T* lhs, size_t size, const U* rhs) {
+ for (size_t i = 0; i != size; i++) {
+ if (!internal::ArrayEq(lhs[i], rhs[i]))
+ return false;
+ }
+ return true;
+}
+
+// Finds the first element in the iterator range [begin, end) that
+// equals elem. Element may be a native array type itself.
+template <typename Iter, typename Element>
+Iter ArrayAwareFind(Iter begin, Iter end, const Element& elem) {
+ for (Iter it = begin; it != end; ++it) {
+ if (internal::ArrayEq(*it, elem))
+ return it;
+ }
+ return end;
+}
+
+// CopyArray() copies a k-dimensional native array using the elements'
+// operator=, where k can be any integer >= 0. When k is 0,
+// CopyArray() degenerates into copying a single value.
+
+template <typename T, typename U>
+void CopyArray(const T* from, size_t size, U* to);
+
+// This generic version is used when k is 0.
+template <typename T, typename U>
+inline void CopyArray(const T& from, U* to) { *to = from; }
+
+// This overload is used when k >= 1.
+template <typename T, typename U, size_t N>
+inline void CopyArray(const T(&from)[N], U(*to)[N]) {
+ internal::CopyArray(from, N, *to);
+}
+
+// This helper reduces code bloat. If we instead put its logic inside
+// the previous CopyArray() function, arrays with different sizes
+// would lead to different copies of the template code.
+template <typename T, typename U>
+void CopyArray(const T* from, size_t size, U* to) {
+ for (size_t i = 0; i != size; i++) {
+ internal::CopyArray(from[i], to + i);
+ }
+}
+
+// The relation between an NativeArray object (see below) and the
+// native array it represents.
+enum RelationToSource {
+ kReference, // The NativeArray references the native array.
+ kCopy // The NativeArray makes a copy of the native array and
+ // owns the copy.
+};
+
+// Adapts a native array to a read-only STL-style container. Instead
+// of the complete STL container concept, this adaptor only implements
+// members useful for Google Mock's container matchers. New members
+// should be added as needed. To simplify the implementation, we only
+// support Element being a raw type (i.e. having no top-level const or
+// reference modifier). It's the client's responsibility to satisfy
+// this requirement. Element can be an array type itself (hence
+// multi-dimensional arrays are supported).
+template <typename Element>
+class NativeArray {
+ public:
+ // STL-style container typedefs.
+ typedef Element value_type;
+ typedef Element* iterator;
+ typedef const Element* const_iterator;
+
+ // Constructs from a native array.
+ NativeArray(const Element* array, size_t count, RelationToSource relation) {
+ Init(array, count, relation);
+ }
+
+ // Copy constructor.
+ NativeArray(const NativeArray& rhs) {
+ Init(rhs.array_, rhs.size_, rhs.relation_to_source_);
+ }
+
+ ~NativeArray() {
+ // Ensures that the user doesn't instantiate NativeArray with a
+ // const or reference type.
+ static_cast<void>(StaticAssertTypeEqHelper<Element,
+ GTEST_REMOVE_REFERENCE_AND_CONST_(Element)>());
+ if (relation_to_source_ == kCopy)
+ delete[] array_;
+ }
+
+ // STL-style container methods.
+ size_t size() const { return size_; }
+ const_iterator begin() const { return array_; }
+ const_iterator end() const { return array_ + size_; }
+ bool operator==(const NativeArray& rhs) const {
+ return size() == rhs.size() &&
+ ArrayEq(begin(), size(), rhs.begin());
+ }
+
+ private:
+ // Initializes this object; makes a copy of the input array if
+ // 'relation' is kCopy.
+ void Init(const Element* array, size_t a_size, RelationToSource relation) {
+ if (relation == kReference) {
+ array_ = array;
+ } else {
+ Element* const copy = new Element[a_size];
+ CopyArray(array, a_size, copy);
+ array_ = copy;
+ }
+ size_ = a_size;
+ relation_to_source_ = relation;
+ }
+
+ const Element* array_;
+ size_t size_;
+ RelationToSource relation_to_source_;
+
+ GTEST_DISALLOW_ASSIGN_(NativeArray);
+};
+
+} // namespace internal
+} // namespace testing
+
+#define GTEST_MESSAGE_AT_(file, line, message, result_type) \
+ ::testing::internal::AssertHelper(result_type, file, line, message) \
+ = ::testing::Message()
+
+#define GTEST_MESSAGE_(message, result_type) \
+ GTEST_MESSAGE_AT_(__FILE__, __LINE__, message, result_type)
+
+#define GTEST_FATAL_FAILURE_(message) \
+ return GTEST_MESSAGE_(message, ::testing::TestPartResult::kFatalFailure)
+
+#define GTEST_NONFATAL_FAILURE_(message) \
+ GTEST_MESSAGE_(message, ::testing::TestPartResult::kNonFatalFailure)
+
+#define GTEST_SUCCESS_(message) \
+ GTEST_MESSAGE_(message, ::testing::TestPartResult::kSuccess)
+
+// Suppresses MSVC warnings 4072 (unreachable code) for the code following
+// statement if it returns or throws (or doesn't return or throw in some
+// situations).
+#define GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) \
+ if (::testing::internal::AlwaysTrue()) { statement; }
+
+#define GTEST_TEST_THROW_(statement, expected_exception, fail) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (::testing::internal::ConstCharPtr gtest_msg = "") { \
+ bool gtest_caught_expected = false; \
+ try { \
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+ } \
+ catch (expected_exception const&) { \
+ gtest_caught_expected = true; \
+ } \
+ catch (...) { \
+ gtest_msg.value = \
+ "Expected: " #statement " throws an exception of type " \
+ #expected_exception ".\n Actual: it throws a different type."; \
+ goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \
+ } \
+ if (!gtest_caught_expected) { \
+ gtest_msg.value = \
+ "Expected: " #statement " throws an exception of type " \
+ #expected_exception ".\n Actual: it throws nothing."; \
+ goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \
+ } \
+ } else \
+ GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__): \
+ fail(gtest_msg.value)
+
+#define GTEST_TEST_NO_THROW_(statement, fail) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (::testing::internal::AlwaysTrue()) { \
+ try { \
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+ } \
+ catch (...) { \
+ goto GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__); \
+ } \
+ } else \
+ GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__): \
+ fail("Expected: " #statement " doesn't throw an exception.\n" \
+ " Actual: it throws.")
+
+#define GTEST_TEST_ANY_THROW_(statement, fail) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (::testing::internal::AlwaysTrue()) { \
+ bool gtest_caught_any = false; \
+ try { \
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+ } \
+ catch (...) { \
+ gtest_caught_any = true; \
+ } \
+ if (!gtest_caught_any) { \
+ goto GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__); \
+ } \
+ } else \
+ GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__): \
+ fail("Expected: " #statement " throws an exception.\n" \
+ " Actual: it doesn't.")
+
+
+// Implements Boolean test assertions such as EXPECT_TRUE. expression can be
+// either a boolean expression or an AssertionResult. text is a textual
+// represenation of expression as it was passed into the EXPECT_TRUE.
+#define GTEST_TEST_BOOLEAN_(expression, text, actual, expected, fail) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (const ::testing::AssertionResult gtest_ar_ = \
+ ::testing::AssertionResult(expression)) \
+ ; \
+ else \
+ fail(::testing::internal::GetBoolAssertionFailureMessage(\
+ gtest_ar_, text, #actual, #expected).c_str())
+
+#define GTEST_TEST_NO_FATAL_FAILURE_(statement, fail) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (::testing::internal::AlwaysTrue()) { \
+ ::testing::internal::HasNewFatalFailureHelper gtest_fatal_failure_checker; \
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+ if (gtest_fatal_failure_checker.has_new_fatal_failure()) { \
+ goto GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__); \
+ } \
+ } else \
+ GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__): \
+ fail("Expected: " #statement " doesn't generate new fatal " \
+ "failures in the current thread.\n" \
+ " Actual: it does.")
+
+// Expands to the name of the class that implements the given test.
+#define GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \
+ test_case_name##_##test_name##_Test
+
+// Helper macro for defining tests.
+#define GTEST_TEST_(test_case_name, test_name, parent_class, parent_id)\
+class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) : public parent_class {\
+ public:\
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {}\
+ private:\
+ virtual void TestBody();\
+ static ::testing::TestInfo* const test_info_ GTEST_ATTRIBUTE_UNUSED_;\
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(\
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name));\
+};\
+\
+::testing::TestInfo* const GTEST_TEST_CLASS_NAME_(test_case_name, test_name)\
+ ::test_info_ =\
+ ::testing::internal::MakeAndRegisterTestInfo(\
+ #test_case_name, #test_name, NULL, NULL, \
+ (parent_id), \
+ parent_class::SetUpTestCase, \
+ parent_class::TearDownTestCase, \
+ new ::testing::internal::TestFactoryImpl<\
+ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)>);\
+void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody()
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-linked_ptr.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-linked_ptr.h
new file mode 100644
index 0000000..57147b4
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-linked_ptr.h
@@ -0,0 +1,233 @@
+// Copyright 2003 Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: Dan Egnor (egnor@google.com)
+//
+// A "smart" pointer type with reference tracking. Every pointer to a
+// particular object is kept on a circular linked list. When the last pointer
+// to an object is destroyed or reassigned, the object is deleted.
+//
+// Used properly, this deletes the object when the last reference goes away.
+// There are several caveats:
+// - Like all reference counting schemes, cycles lead to leaks.
+// - Each smart pointer is actually two pointers (8 bytes instead of 4).
+// - Every time a pointer is assigned, the entire list of pointers to that
+// object is traversed. This class is therefore NOT SUITABLE when there
+// will often be more than two or three pointers to a particular object.
+// - References are only tracked as long as linked_ptr<> objects are copied.
+// If a linked_ptr<> is converted to a raw pointer and back, BAD THINGS
+// will happen (double deletion).
+//
+// A good use of this class is storing object references in STL containers.
+// You can safely put linked_ptr<> in a vector<>.
+// Other uses may not be as good.
+//
+// Note: If you use an incomplete type with linked_ptr<>, the class
+// *containing* linked_ptr<> must have a constructor and destructor (even
+// if they do nothing!).
+//
+// Bill Gibbons suggested we use something like this.
+//
+// Thread Safety:
+// Unlike other linked_ptr implementations, in this implementation
+// a linked_ptr object is thread-safe in the sense that:
+// - it's safe to copy linked_ptr objects concurrently,
+// - it's safe to copy *from* a linked_ptr and read its underlying
+// raw pointer (e.g. via get()) concurrently, and
+// - it's safe to write to two linked_ptrs that point to the same
+// shared object concurrently.
+// TODO(wan@google.com): rename this to safe_linked_ptr to avoid
+// confusion with normal linked_ptr.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
+
+#include <stdlib.h>
+#include <assert.h>
+
+#include "gtest/internal/gtest-port.h"
+
+namespace testing {
+namespace internal {
+
+// Protects copying of all linked_ptr objects.
+GTEST_API_ GTEST_DECLARE_STATIC_MUTEX_(g_linked_ptr_mutex);
+
+// This is used internally by all instances of linked_ptr<>. It needs to be
+// a non-template class because different types of linked_ptr<> can refer to
+// the same object (linked_ptr<Superclass>(obj) vs linked_ptr<Subclass>(obj)).
+// So, it needs to be possible for different types of linked_ptr to participate
+// in the same circular linked list, so we need a single class type here.
+//
+// DO NOT USE THIS CLASS DIRECTLY YOURSELF. Use linked_ptr<T>.
+class linked_ptr_internal {
+ public:
+ // Create a new circle that includes only this instance.
+ void join_new() {
+ next_ = this;
+ }
+
+ // Many linked_ptr operations may change p.link_ for some linked_ptr
+ // variable p in the same circle as this object. Therefore we need
+ // to prevent two such operations from occurring concurrently.
+ //
+ // Note that different types of linked_ptr objects can coexist in a
+ // circle (e.g. linked_ptr<Base>, linked_ptr<Derived1>, and
+ // linked_ptr<Derived2>). Therefore we must use a single mutex to
+ // protect all linked_ptr objects. This can create serious
+ // contention in production code, but is acceptable in a testing
+ // framework.
+
+ // Join an existing circle.
+ // L < g_linked_ptr_mutex
+ void join(linked_ptr_internal const* ptr) {
+ MutexLock lock(&g_linked_ptr_mutex);
+
+ linked_ptr_internal const* p = ptr;
+ while (p->next_ != ptr) p = p->next_;
+ p->next_ = this;
+ next_ = ptr;
+ }
+
+ // Leave whatever circle we're part of. Returns true if we were the
+ // last member of the circle. Once this is done, you can join() another.
+ // L < g_linked_ptr_mutex
+ bool depart() {
+ MutexLock lock(&g_linked_ptr_mutex);
+
+ if (next_ == this) return true;
+ linked_ptr_internal const* p = next_;
+ while (p->next_ != this) p = p->next_;
+ p->next_ = next_;
+ return false;
+ }
+
+ private:
+ mutable linked_ptr_internal const* next_;
+};
+
+template <typename T>
+class linked_ptr {
+ public:
+ typedef T element_type;
+
+ // Take over ownership of a raw pointer. This should happen as soon as
+ // possible after the object is created.
+ explicit linked_ptr(T* ptr = NULL) { capture(ptr); }
+ ~linked_ptr() { depart(); }
+
+ // Copy an existing linked_ptr<>, adding ourselves to the list of references.
+ template <typename U> linked_ptr(linked_ptr<U> const& ptr) { copy(&ptr); }
+ linked_ptr(linked_ptr const& ptr) { // NOLINT
+ assert(&ptr != this);
+ copy(&ptr);
+ }
+
+ // Assignment releases the old value and acquires the new.
+ template <typename U> linked_ptr& operator=(linked_ptr<U> const& ptr) {
+ depart();
+ copy(&ptr);
+ return *this;
+ }
+
+ linked_ptr& operator=(linked_ptr const& ptr) {
+ if (&ptr != this) {
+ depart();
+ copy(&ptr);
+ }
+ return *this;
+ }
+
+ // Smart pointer members.
+ void reset(T* ptr = NULL) {
+ depart();
+ capture(ptr);
+ }
+ T* get() const { return value_; }
+ T* operator->() const { return value_; }
+ T& operator*() const { return *value_; }
+
+ bool operator==(T* p) const { return value_ == p; }
+ bool operator!=(T* p) const { return value_ != p; }
+ template <typename U>
+ bool operator==(linked_ptr<U> const& ptr) const {
+ return value_ == ptr.get();
+ }
+ template <typename U>
+ bool operator!=(linked_ptr<U> const& ptr) const {
+ return value_ != ptr.get();
+ }
+
+ private:
+ template <typename U>
+ friend class linked_ptr;
+
+ T* value_;
+ linked_ptr_internal link_;
+
+ void depart() {
+ if (link_.depart()) delete value_;
+ }
+
+ void capture(T* ptr) {
+ value_ = ptr;
+ link_.join_new();
+ }
+
+ template <typename U> void copy(linked_ptr<U> const* ptr) {
+ value_ = ptr->get();
+ if (value_)
+ link_.join(&ptr->link_);
+ else
+ link_.join_new();
+ }
+};
+
+template<typename T> inline
+bool operator==(T* ptr, const linked_ptr<T>& x) {
+ return ptr == x.get();
+}
+
+template<typename T> inline
+bool operator!=(T* ptr, const linked_ptr<T>& x) {
+ return ptr != x.get();
+}
+
+// A function to convert T* into linked_ptr<T>
+// Doing e.g. make_linked_ptr(new FooBarBaz<type>(arg)) is a shorter notation
+// for linked_ptr<FooBarBaz<type> >(new FooBarBaz<type>(arg))
+template <typename T>
+linked_ptr<T> make_linked_ptr(T* ptr) {
+ return linked_ptr<T>(ptr);
+}
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util-generated.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util-generated.h
new file mode 100644
index 0000000..2582675
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util-generated.h
@@ -0,0 +1,4822 @@
+// This file was GENERATED by command:
+// pump.py gtest-param-util-generated.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// Type and function utilities for implementing parameterized tests.
+// This file is generated by a SCRIPT. DO NOT EDIT BY HAND!
+//
+// Currently Google Test supports at most 50 arguments in Values,
+// and at most 10 arguments in Combine. Please contact
+// googletestframework@googlegroups.com if you need more.
+// Please note that the number of arguments to Combine is limited
+// by the maximum arity of the implementation of tr1::tuple which is
+// currently set at 10.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*. Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+#include "gtest/internal/gtest-param-util.h"
+#include "gtest/internal/gtest-port.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing {
+
+// Forward declarations of ValuesIn(), which is implemented in
+// include/gtest/gtest-param-test.h.
+template <typename ForwardIterator>
+internal::ParamGenerator<
+ typename ::testing::internal::IteratorTraits<ForwardIterator>::value_type>
+ValuesIn(ForwardIterator begin, ForwardIterator end);
+
+template <typename T, size_t N>
+internal::ParamGenerator<T> ValuesIn(const T (&array)[N]);
+
+template <class Container>
+internal::ParamGenerator<typename Container::value_type> ValuesIn(
+ const Container& container);
+
+namespace internal {
+
+// Used in the Values() function to provide polymorphic capabilities.
+template <typename T1>
+class ValueArray1 {
+ public:
+ explicit ValueArray1(T1 v1) : v1_(v1) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const { return ValuesIn(&v1_, &v1_ + 1); }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray1& other);
+
+ const T1 v1_;
+};
+
+template <typename T1, typename T2>
+class ValueArray2 {
+ public:
+ ValueArray2(T1 v1, T2 v2) : v1_(v1), v2_(v2) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray2& other);
+
+ const T1 v1_;
+ const T2 v2_;
+};
+
+template <typename T1, typename T2, typename T3>
+class ValueArray3 {
+ public:
+ ValueArray3(T1 v1, T2 v2, T3 v3) : v1_(v1), v2_(v2), v3_(v3) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray3& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4>
+class ValueArray4 {
+ public:
+ ValueArray4(T1 v1, T2 v2, T3 v3, T4 v4) : v1_(v1), v2_(v2), v3_(v3),
+ v4_(v4) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray4& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5>
+class ValueArray5 {
+ public:
+ ValueArray5(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5) : v1_(v1), v2_(v2), v3_(v3),
+ v4_(v4), v5_(v5) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray5& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6>
+class ValueArray6 {
+ public:
+ ValueArray6(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6) : v1_(v1), v2_(v2),
+ v3_(v3), v4_(v4), v5_(v5), v6_(v6) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray6& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7>
+class ValueArray7 {
+ public:
+ ValueArray7(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7) : v1_(v1),
+ v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray7& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8>
+class ValueArray8 {
+ public:
+ ValueArray8(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+ T8 v8) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray8& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9>
+class ValueArray9 {
+ public:
+ ValueArray9(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+ T9 v9) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray9& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10>
+class ValueArray10 {
+ public:
+ ValueArray10(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray10& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11>
+class ValueArray11 {
+ public:
+ ValueArray11(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+ v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray11& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12>
+class ValueArray12 {
+ public:
+ ValueArray12(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+ v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray12& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13>
+class ValueArray13 {
+ public:
+ ValueArray13(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+ v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+ v12_(v12), v13_(v13) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray13& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14>
+class ValueArray14 {
+ public:
+ ValueArray14(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14) : v1_(v1), v2_(v2), v3_(v3),
+ v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray14& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15>
+class ValueArray15 {
+ public:
+ ValueArray15(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15) : v1_(v1), v2_(v2),
+ v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray15& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16>
+class ValueArray16 {
+ public:
+ ValueArray16(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16) : v1_(v1),
+ v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+ v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+ v16_(v16) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray16& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17>
+class ValueArray17 {
+ public:
+ ValueArray17(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+ T17 v17) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray17& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18>
+class ValueArray18 {
+ public:
+ ValueArray18(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray18& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19>
+class ValueArray19 {
+ public:
+ ValueArray19(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+ v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+ v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray19& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20>
+class ValueArray20 {
+ public:
+ ValueArray20(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+ v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+ v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+ v19_(v19), v20_(v20) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray20& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21>
+class ValueArray21 {
+ public:
+ ValueArray21(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+ v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+ v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+ v18_(v18), v19_(v19), v20_(v20), v21_(v21) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray21& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22>
+class ValueArray22 {
+ public:
+ ValueArray22(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22) : v1_(v1), v2_(v2), v3_(v3),
+ v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+ v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray22& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23>
+class ValueArray23 {
+ public:
+ ValueArray23(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23) : v1_(v1), v2_(v2),
+ v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+ v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+ v23_(v23) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_,
+ v23_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray23& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24>
+class ValueArray24 {
+ public:
+ ValueArray24(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24) : v1_(v1),
+ v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+ v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+ v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+ v22_(v22), v23_(v23), v24_(v24) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray24& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25>
+class ValueArray25 {
+ public:
+ ValueArray25(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+ T25 v25) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+ v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray25& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26>
+class ValueArray26 {
+ public:
+ ValueArray26(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+ v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray26& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27>
+class ValueArray27 {
+ public:
+ ValueArray27(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+ v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+ v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+ v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+ v26_(v26), v27_(v27) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray27& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28>
+class ValueArray28 {
+ public:
+ ValueArray28(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+ v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+ v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+ v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+ v25_(v25), v26_(v26), v27_(v27), v28_(v28) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray28& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29>
+class ValueArray29 {
+ public:
+ ValueArray29(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+ v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+ v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+ v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+ v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray29& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30>
+class ValueArray30 {
+ public:
+ ValueArray30(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30) : v1_(v1), v2_(v2), v3_(v3),
+ v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+ v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+ v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+ v29_(v29), v30_(v30) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray30& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31>
+class ValueArray31 {
+ public:
+ ValueArray31(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31) : v1_(v1), v2_(v2),
+ v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+ v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+ v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+ v29_(v29), v30_(v30), v31_(v31) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray31& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32>
+class ValueArray32 {
+ public:
+ ValueArray32(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32) : v1_(v1),
+ v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+ v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+ v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+ v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+ v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray32& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33>
+class ValueArray33 {
+ public:
+ ValueArray33(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32,
+ T33 v33) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+ v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+ v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+ v33_(v33) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray33& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34>
+class ValueArray34 {
+ public:
+ ValueArray34(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+ v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+ v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+ v33_(v33), v34_(v34) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray34& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35>
+class ValueArray35 {
+ public:
+ ValueArray35(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+ v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+ v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+ v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+ v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31),
+ v32_(v32), v33_(v33), v34_(v34), v35_(v35) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_,
+ v35_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray35& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36>
+class ValueArray36 {
+ public:
+ ValueArray36(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+ v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+ v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+ v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+ v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30),
+ v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray36& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37>
+class ValueArray37 {
+ public:
+ ValueArray37(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+ v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+ v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+ v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+ v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29),
+ v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35),
+ v36_(v36), v37_(v37) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray37& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38>
+class ValueArray38 {
+ public:
+ ValueArray38(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38) : v1_(v1), v2_(v2), v3_(v3),
+ v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+ v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+ v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+ v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+ v35_(v35), v36_(v36), v37_(v37), v38_(v38) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray38& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39>
+class ValueArray39 {
+ public:
+ ValueArray39(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39) : v1_(v1), v2_(v2),
+ v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+ v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+ v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+ v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+ v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray39& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40>
+class ValueArray40 {
+ public:
+ ValueArray40(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40) : v1_(v1),
+ v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+ v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+ v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+ v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+ v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33),
+ v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39),
+ v40_(v40) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray40& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41>
+class ValueArray41 {
+ public:
+ ValueArray41(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40,
+ T41 v41) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+ v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+ v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+ v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+ v39_(v39), v40_(v40), v41_(v41) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray41& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42>
+class ValueArray42 {
+ public:
+ ValueArray42(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+ v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+ v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+ v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+ v39_(v39), v40_(v40), v41_(v41), v42_(v42) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray42& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43>
+class ValueArray43 {
+ public:
+ ValueArray43(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+ v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+ v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+ v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+ v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31),
+ v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37),
+ v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_, v43_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray43& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+ const T43 v43_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44>
+class ValueArray44 {
+ public:
+ ValueArray44(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43, T44 v44) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+ v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+ v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+ v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+ v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30),
+ v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36),
+ v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42),
+ v43_(v43), v44_(v44) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_, v43_, v44_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray44& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+ const T43 v43_;
+ const T44 v44_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45>
+class ValueArray45 {
+ public:
+ ValueArray45(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43, T44 v44, T45 v45) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+ v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+ v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+ v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+ v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29),
+ v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35),
+ v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41),
+ v42_(v42), v43_(v43), v44_(v44), v45_(v45) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_, v43_, v44_, v45_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray45& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+ const T43 v43_;
+ const T44 v44_;
+ const T45 v45_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46>
+class ValueArray46 {
+ public:
+ ValueArray46(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43, T44 v44, T45 v45, T46 v46) : v1_(v1), v2_(v2), v3_(v3),
+ v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+ v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+ v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+ v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+ v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40),
+ v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_, v43_, v44_, v45_, v46_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray46& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+ const T43 v43_;
+ const T44 v44_;
+ const T45 v45_;
+ const T46 v46_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47>
+class ValueArray47 {
+ public:
+ ValueArray47(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47) : v1_(v1), v2_(v2),
+ v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+ v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+ v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+ v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+ v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+ v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40),
+ v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46),
+ v47_(v47) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_, v43_, v44_, v45_, v46_,
+ v47_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray47& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+ const T43 v43_;
+ const T44 v44_;
+ const T45 v45_;
+ const T46 v46_;
+ const T47 v47_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48>
+class ValueArray48 {
+ public:
+ ValueArray48(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48) : v1_(v1),
+ v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+ v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+ v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+ v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+ v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33),
+ v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39),
+ v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45),
+ v46_(v46), v47_(v47), v48_(v48) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_, v43_, v44_, v45_, v46_, v47_,
+ v48_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray48& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+ const T43 v43_;
+ const T44 v44_;
+ const T45 v45_;
+ const T46 v46_;
+ const T47 v47_;
+ const T48 v48_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48, typename T49>
+class ValueArray49 {
+ public:
+ ValueArray49(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48,
+ T49 v49) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+ v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+ v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+ v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+ v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44),
+ v45_(v45), v46_(v46), v47_(v47), v48_(v48), v49_(v49) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_, v43_, v44_, v45_, v46_, v47_,
+ v48_, v49_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray49& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+ const T43 v43_;
+ const T44 v44_;
+ const T45 v45_;
+ const T46 v46_;
+ const T47 v47_;
+ const T48 v48_;
+ const T49 v49_;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48, typename T49, typename T50>
+class ValueArray50 {
+ public:
+ ValueArray50(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+ T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+ T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+ T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+ T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+ T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48, T49 v49,
+ T50 v50) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+ v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+ v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+ v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+ v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+ v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+ v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44),
+ v45_(v45), v46_(v46), v47_(v47), v48_(v48), v49_(v49), v50_(v50) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {v1_, v2_, v3_, v4_, v5_, v6_, v7_, v8_, v9_, v10_, v11_,
+ v12_, v13_, v14_, v15_, v16_, v17_, v18_, v19_, v20_, v21_, v22_, v23_,
+ v24_, v25_, v26_, v27_, v28_, v29_, v30_, v31_, v32_, v33_, v34_, v35_,
+ v36_, v37_, v38_, v39_, v40_, v41_, v42_, v43_, v44_, v45_, v46_, v47_,
+ v48_, v49_, v50_};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray50& other);
+
+ const T1 v1_;
+ const T2 v2_;
+ const T3 v3_;
+ const T4 v4_;
+ const T5 v5_;
+ const T6 v6_;
+ const T7 v7_;
+ const T8 v8_;
+ const T9 v9_;
+ const T10 v10_;
+ const T11 v11_;
+ const T12 v12_;
+ const T13 v13_;
+ const T14 v14_;
+ const T15 v15_;
+ const T16 v16_;
+ const T17 v17_;
+ const T18 v18_;
+ const T19 v19_;
+ const T20 v20_;
+ const T21 v21_;
+ const T22 v22_;
+ const T23 v23_;
+ const T24 v24_;
+ const T25 v25_;
+ const T26 v26_;
+ const T27 v27_;
+ const T28 v28_;
+ const T29 v29_;
+ const T30 v30_;
+ const T31 v31_;
+ const T32 v32_;
+ const T33 v33_;
+ const T34 v34_;
+ const T35 v35_;
+ const T36 v36_;
+ const T37 v37_;
+ const T38 v38_;
+ const T39 v39_;
+ const T40 v40_;
+ const T41 v41_;
+ const T42 v42_;
+ const T43 v43_;
+ const T44 v44_;
+ const T45 v45_;
+ const T46 v46_;
+ const T47 v47_;
+ const T48 v48_;
+ const T49 v49_;
+ const T50 v50_;
+};
+
+# if GTEST_HAS_COMBINE
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Generates values from the Cartesian product of values produced
+// by the argument generators.
+//
+template <typename T1, typename T2>
+class CartesianProductGenerator2
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2> ParamType;
+
+ CartesianProductGenerator2(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2)
+ : g1_(g1), g2_(g2) {}
+ virtual ~CartesianProductGenerator2() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current2_;
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator2::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator2& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+}; // class CartesianProductGenerator2
+
+
+template <typename T1, typename T2, typename T3>
+class CartesianProductGenerator3
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2, T3> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2, T3> ParamType;
+
+ CartesianProductGenerator3(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2, const ParamGenerator<T3>& g3)
+ : g1_(g1), g2_(g2), g3_(g3) {}
+ virtual ~CartesianProductGenerator3() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+ g3_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2,
+ const ParamGenerator<T3>& g3,
+ const typename ParamGenerator<T3>::iterator& current3)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+ begin3_(g3.begin()), end3_(g3.end()), current3_(current3) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current3_;
+ if (current3_ == end3_) {
+ current3_ = begin3_;
+ ++current2_;
+ }
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_ &&
+ current3_ == typed_other->current3_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_),
+ begin3_(other.begin3_),
+ end3_(other.end3_),
+ current3_(other.current3_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_, *current3_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_ ||
+ current3_ == end3_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ const typename ParamGenerator<T3>::iterator begin3_;
+ const typename ParamGenerator<T3>::iterator end3_;
+ typename ParamGenerator<T3>::iterator current3_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator3::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator3& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+ const ParamGenerator<T3> g3_;
+}; // class CartesianProductGenerator3
+
+
+template <typename T1, typename T2, typename T3, typename T4>
+class CartesianProductGenerator4
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2, T3, T4> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2, T3, T4> ParamType;
+
+ CartesianProductGenerator4(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2, const ParamGenerator<T3>& g3,
+ const ParamGenerator<T4>& g4)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4) {}
+ virtual ~CartesianProductGenerator4() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+ g3_.begin(), g4_, g4_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+ g4_, g4_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2,
+ const ParamGenerator<T3>& g3,
+ const typename ParamGenerator<T3>::iterator& current3,
+ const ParamGenerator<T4>& g4,
+ const typename ParamGenerator<T4>::iterator& current4)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+ begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+ begin4_(g4.begin()), end4_(g4.end()), current4_(current4) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current4_;
+ if (current4_ == end4_) {
+ current4_ = begin4_;
+ ++current3_;
+ }
+ if (current3_ == end3_) {
+ current3_ = begin3_;
+ ++current2_;
+ }
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_ &&
+ current3_ == typed_other->current3_ &&
+ current4_ == typed_other->current4_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_),
+ begin3_(other.begin3_),
+ end3_(other.end3_),
+ current3_(other.current3_),
+ begin4_(other.begin4_),
+ end4_(other.end4_),
+ current4_(other.current4_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_, *current3_,
+ *current4_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_ ||
+ current3_ == end3_ ||
+ current4_ == end4_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ const typename ParamGenerator<T3>::iterator begin3_;
+ const typename ParamGenerator<T3>::iterator end3_;
+ typename ParamGenerator<T3>::iterator current3_;
+ const typename ParamGenerator<T4>::iterator begin4_;
+ const typename ParamGenerator<T4>::iterator end4_;
+ typename ParamGenerator<T4>::iterator current4_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator4::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator4& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+ const ParamGenerator<T3> g3_;
+ const ParamGenerator<T4> g4_;
+}; // class CartesianProductGenerator4
+
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5>
+class CartesianProductGenerator5
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2, T3, T4, T5> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2, T3, T4, T5> ParamType;
+
+ CartesianProductGenerator5(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2, const ParamGenerator<T3>& g3,
+ const ParamGenerator<T4>& g4, const ParamGenerator<T5>& g5)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5) {}
+ virtual ~CartesianProductGenerator5() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+ g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+ g4_, g4_.end(), g5_, g5_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2,
+ const ParamGenerator<T3>& g3,
+ const typename ParamGenerator<T3>::iterator& current3,
+ const ParamGenerator<T4>& g4,
+ const typename ParamGenerator<T4>::iterator& current4,
+ const ParamGenerator<T5>& g5,
+ const typename ParamGenerator<T5>::iterator& current5)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+ begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+ begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+ begin5_(g5.begin()), end5_(g5.end()), current5_(current5) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current5_;
+ if (current5_ == end5_) {
+ current5_ = begin5_;
+ ++current4_;
+ }
+ if (current4_ == end4_) {
+ current4_ = begin4_;
+ ++current3_;
+ }
+ if (current3_ == end3_) {
+ current3_ = begin3_;
+ ++current2_;
+ }
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_ &&
+ current3_ == typed_other->current3_ &&
+ current4_ == typed_other->current4_ &&
+ current5_ == typed_other->current5_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_),
+ begin3_(other.begin3_),
+ end3_(other.end3_),
+ current3_(other.current3_),
+ begin4_(other.begin4_),
+ end4_(other.end4_),
+ current4_(other.current4_),
+ begin5_(other.begin5_),
+ end5_(other.end5_),
+ current5_(other.current5_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_, *current3_,
+ *current4_, *current5_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_ ||
+ current3_ == end3_ ||
+ current4_ == end4_ ||
+ current5_ == end5_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ const typename ParamGenerator<T3>::iterator begin3_;
+ const typename ParamGenerator<T3>::iterator end3_;
+ typename ParamGenerator<T3>::iterator current3_;
+ const typename ParamGenerator<T4>::iterator begin4_;
+ const typename ParamGenerator<T4>::iterator end4_;
+ typename ParamGenerator<T4>::iterator current4_;
+ const typename ParamGenerator<T5>::iterator begin5_;
+ const typename ParamGenerator<T5>::iterator end5_;
+ typename ParamGenerator<T5>::iterator current5_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator5::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator5& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+ const ParamGenerator<T3> g3_;
+ const ParamGenerator<T4> g4_;
+ const ParamGenerator<T5> g5_;
+}; // class CartesianProductGenerator5
+
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6>
+class CartesianProductGenerator6
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2, T3, T4, T5,
+ T6> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2, T3, T4, T5, T6> ParamType;
+
+ CartesianProductGenerator6(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2, const ParamGenerator<T3>& g3,
+ const ParamGenerator<T4>& g4, const ParamGenerator<T5>& g5,
+ const ParamGenerator<T6>& g6)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6) {}
+ virtual ~CartesianProductGenerator6() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+ g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+ g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2,
+ const ParamGenerator<T3>& g3,
+ const typename ParamGenerator<T3>::iterator& current3,
+ const ParamGenerator<T4>& g4,
+ const typename ParamGenerator<T4>::iterator& current4,
+ const ParamGenerator<T5>& g5,
+ const typename ParamGenerator<T5>::iterator& current5,
+ const ParamGenerator<T6>& g6,
+ const typename ParamGenerator<T6>::iterator& current6)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+ begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+ begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+ begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+ begin6_(g6.begin()), end6_(g6.end()), current6_(current6) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current6_;
+ if (current6_ == end6_) {
+ current6_ = begin6_;
+ ++current5_;
+ }
+ if (current5_ == end5_) {
+ current5_ = begin5_;
+ ++current4_;
+ }
+ if (current4_ == end4_) {
+ current4_ = begin4_;
+ ++current3_;
+ }
+ if (current3_ == end3_) {
+ current3_ = begin3_;
+ ++current2_;
+ }
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_ &&
+ current3_ == typed_other->current3_ &&
+ current4_ == typed_other->current4_ &&
+ current5_ == typed_other->current5_ &&
+ current6_ == typed_other->current6_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_),
+ begin3_(other.begin3_),
+ end3_(other.end3_),
+ current3_(other.current3_),
+ begin4_(other.begin4_),
+ end4_(other.end4_),
+ current4_(other.current4_),
+ begin5_(other.begin5_),
+ end5_(other.end5_),
+ current5_(other.current5_),
+ begin6_(other.begin6_),
+ end6_(other.end6_),
+ current6_(other.current6_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_, *current3_,
+ *current4_, *current5_, *current6_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_ ||
+ current3_ == end3_ ||
+ current4_ == end4_ ||
+ current5_ == end5_ ||
+ current6_ == end6_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ const typename ParamGenerator<T3>::iterator begin3_;
+ const typename ParamGenerator<T3>::iterator end3_;
+ typename ParamGenerator<T3>::iterator current3_;
+ const typename ParamGenerator<T4>::iterator begin4_;
+ const typename ParamGenerator<T4>::iterator end4_;
+ typename ParamGenerator<T4>::iterator current4_;
+ const typename ParamGenerator<T5>::iterator begin5_;
+ const typename ParamGenerator<T5>::iterator end5_;
+ typename ParamGenerator<T5>::iterator current5_;
+ const typename ParamGenerator<T6>::iterator begin6_;
+ const typename ParamGenerator<T6>::iterator end6_;
+ typename ParamGenerator<T6>::iterator current6_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator6::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator6& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+ const ParamGenerator<T3> g3_;
+ const ParamGenerator<T4> g4_;
+ const ParamGenerator<T5> g5_;
+ const ParamGenerator<T6> g6_;
+}; // class CartesianProductGenerator6
+
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7>
+class CartesianProductGenerator7
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6,
+ T7> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7> ParamType;
+
+ CartesianProductGenerator7(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2, const ParamGenerator<T3>& g3,
+ const ParamGenerator<T4>& g4, const ParamGenerator<T5>& g5,
+ const ParamGenerator<T6>& g6, const ParamGenerator<T7>& g7)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7) {}
+ virtual ~CartesianProductGenerator7() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+ g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+ g7_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+ g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2,
+ const ParamGenerator<T3>& g3,
+ const typename ParamGenerator<T3>::iterator& current3,
+ const ParamGenerator<T4>& g4,
+ const typename ParamGenerator<T4>::iterator& current4,
+ const ParamGenerator<T5>& g5,
+ const typename ParamGenerator<T5>::iterator& current5,
+ const ParamGenerator<T6>& g6,
+ const typename ParamGenerator<T6>::iterator& current6,
+ const ParamGenerator<T7>& g7,
+ const typename ParamGenerator<T7>::iterator& current7)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+ begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+ begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+ begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+ begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+ begin7_(g7.begin()), end7_(g7.end()), current7_(current7) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current7_;
+ if (current7_ == end7_) {
+ current7_ = begin7_;
+ ++current6_;
+ }
+ if (current6_ == end6_) {
+ current6_ = begin6_;
+ ++current5_;
+ }
+ if (current5_ == end5_) {
+ current5_ = begin5_;
+ ++current4_;
+ }
+ if (current4_ == end4_) {
+ current4_ = begin4_;
+ ++current3_;
+ }
+ if (current3_ == end3_) {
+ current3_ = begin3_;
+ ++current2_;
+ }
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_ &&
+ current3_ == typed_other->current3_ &&
+ current4_ == typed_other->current4_ &&
+ current5_ == typed_other->current5_ &&
+ current6_ == typed_other->current6_ &&
+ current7_ == typed_other->current7_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_),
+ begin3_(other.begin3_),
+ end3_(other.end3_),
+ current3_(other.current3_),
+ begin4_(other.begin4_),
+ end4_(other.end4_),
+ current4_(other.current4_),
+ begin5_(other.begin5_),
+ end5_(other.end5_),
+ current5_(other.current5_),
+ begin6_(other.begin6_),
+ end6_(other.end6_),
+ current6_(other.current6_),
+ begin7_(other.begin7_),
+ end7_(other.end7_),
+ current7_(other.current7_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_, *current3_,
+ *current4_, *current5_, *current6_, *current7_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_ ||
+ current3_ == end3_ ||
+ current4_ == end4_ ||
+ current5_ == end5_ ||
+ current6_ == end6_ ||
+ current7_ == end7_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ const typename ParamGenerator<T3>::iterator begin3_;
+ const typename ParamGenerator<T3>::iterator end3_;
+ typename ParamGenerator<T3>::iterator current3_;
+ const typename ParamGenerator<T4>::iterator begin4_;
+ const typename ParamGenerator<T4>::iterator end4_;
+ typename ParamGenerator<T4>::iterator current4_;
+ const typename ParamGenerator<T5>::iterator begin5_;
+ const typename ParamGenerator<T5>::iterator end5_;
+ typename ParamGenerator<T5>::iterator current5_;
+ const typename ParamGenerator<T6>::iterator begin6_;
+ const typename ParamGenerator<T6>::iterator end6_;
+ typename ParamGenerator<T6>::iterator current6_;
+ const typename ParamGenerator<T7>::iterator begin7_;
+ const typename ParamGenerator<T7>::iterator end7_;
+ typename ParamGenerator<T7>::iterator current7_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator7::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator7& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+ const ParamGenerator<T3> g3_;
+ const ParamGenerator<T4> g4_;
+ const ParamGenerator<T5> g5_;
+ const ParamGenerator<T6> g6_;
+ const ParamGenerator<T7> g7_;
+}; // class CartesianProductGenerator7
+
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8>
+class CartesianProductGenerator8
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6,
+ T7, T8> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8> ParamType;
+
+ CartesianProductGenerator8(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2, const ParamGenerator<T3>& g3,
+ const ParamGenerator<T4>& g4, const ParamGenerator<T5>& g5,
+ const ParamGenerator<T6>& g6, const ParamGenerator<T7>& g7,
+ const ParamGenerator<T8>& g8)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7),
+ g8_(g8) {}
+ virtual ~CartesianProductGenerator8() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+ g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+ g7_.begin(), g8_, g8_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+ g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+ g8_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2,
+ const ParamGenerator<T3>& g3,
+ const typename ParamGenerator<T3>::iterator& current3,
+ const ParamGenerator<T4>& g4,
+ const typename ParamGenerator<T4>::iterator& current4,
+ const ParamGenerator<T5>& g5,
+ const typename ParamGenerator<T5>::iterator& current5,
+ const ParamGenerator<T6>& g6,
+ const typename ParamGenerator<T6>::iterator& current6,
+ const ParamGenerator<T7>& g7,
+ const typename ParamGenerator<T7>::iterator& current7,
+ const ParamGenerator<T8>& g8,
+ const typename ParamGenerator<T8>::iterator& current8)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+ begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+ begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+ begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+ begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+ begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+ begin8_(g8.begin()), end8_(g8.end()), current8_(current8) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current8_;
+ if (current8_ == end8_) {
+ current8_ = begin8_;
+ ++current7_;
+ }
+ if (current7_ == end7_) {
+ current7_ = begin7_;
+ ++current6_;
+ }
+ if (current6_ == end6_) {
+ current6_ = begin6_;
+ ++current5_;
+ }
+ if (current5_ == end5_) {
+ current5_ = begin5_;
+ ++current4_;
+ }
+ if (current4_ == end4_) {
+ current4_ = begin4_;
+ ++current3_;
+ }
+ if (current3_ == end3_) {
+ current3_ = begin3_;
+ ++current2_;
+ }
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_ &&
+ current3_ == typed_other->current3_ &&
+ current4_ == typed_other->current4_ &&
+ current5_ == typed_other->current5_ &&
+ current6_ == typed_other->current6_ &&
+ current7_ == typed_other->current7_ &&
+ current8_ == typed_other->current8_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_),
+ begin3_(other.begin3_),
+ end3_(other.end3_),
+ current3_(other.current3_),
+ begin4_(other.begin4_),
+ end4_(other.end4_),
+ current4_(other.current4_),
+ begin5_(other.begin5_),
+ end5_(other.end5_),
+ current5_(other.current5_),
+ begin6_(other.begin6_),
+ end6_(other.end6_),
+ current6_(other.current6_),
+ begin7_(other.begin7_),
+ end7_(other.end7_),
+ current7_(other.current7_),
+ begin8_(other.begin8_),
+ end8_(other.end8_),
+ current8_(other.current8_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_, *current3_,
+ *current4_, *current5_, *current6_, *current7_, *current8_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_ ||
+ current3_ == end3_ ||
+ current4_ == end4_ ||
+ current5_ == end5_ ||
+ current6_ == end6_ ||
+ current7_ == end7_ ||
+ current8_ == end8_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ const typename ParamGenerator<T3>::iterator begin3_;
+ const typename ParamGenerator<T3>::iterator end3_;
+ typename ParamGenerator<T3>::iterator current3_;
+ const typename ParamGenerator<T4>::iterator begin4_;
+ const typename ParamGenerator<T4>::iterator end4_;
+ typename ParamGenerator<T4>::iterator current4_;
+ const typename ParamGenerator<T5>::iterator begin5_;
+ const typename ParamGenerator<T5>::iterator end5_;
+ typename ParamGenerator<T5>::iterator current5_;
+ const typename ParamGenerator<T6>::iterator begin6_;
+ const typename ParamGenerator<T6>::iterator end6_;
+ typename ParamGenerator<T6>::iterator current6_;
+ const typename ParamGenerator<T7>::iterator begin7_;
+ const typename ParamGenerator<T7>::iterator end7_;
+ typename ParamGenerator<T7>::iterator current7_;
+ const typename ParamGenerator<T8>::iterator begin8_;
+ const typename ParamGenerator<T8>::iterator end8_;
+ typename ParamGenerator<T8>::iterator current8_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator8::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator8& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+ const ParamGenerator<T3> g3_;
+ const ParamGenerator<T4> g4_;
+ const ParamGenerator<T5> g5_;
+ const ParamGenerator<T6> g6_;
+ const ParamGenerator<T7> g7_;
+ const ParamGenerator<T8> g8_;
+}; // class CartesianProductGenerator8
+
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9>
+class CartesianProductGenerator9
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6,
+ T7, T8, T9> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9> ParamType;
+
+ CartesianProductGenerator9(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2, const ParamGenerator<T3>& g3,
+ const ParamGenerator<T4>& g4, const ParamGenerator<T5>& g5,
+ const ParamGenerator<T6>& g6, const ParamGenerator<T7>& g7,
+ const ParamGenerator<T8>& g8, const ParamGenerator<T9>& g9)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+ g9_(g9) {}
+ virtual ~CartesianProductGenerator9() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+ g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+ g7_.begin(), g8_, g8_.begin(), g9_, g9_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+ g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+ g8_.end(), g9_, g9_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2,
+ const ParamGenerator<T3>& g3,
+ const typename ParamGenerator<T3>::iterator& current3,
+ const ParamGenerator<T4>& g4,
+ const typename ParamGenerator<T4>::iterator& current4,
+ const ParamGenerator<T5>& g5,
+ const typename ParamGenerator<T5>::iterator& current5,
+ const ParamGenerator<T6>& g6,
+ const typename ParamGenerator<T6>::iterator& current6,
+ const ParamGenerator<T7>& g7,
+ const typename ParamGenerator<T7>::iterator& current7,
+ const ParamGenerator<T8>& g8,
+ const typename ParamGenerator<T8>::iterator& current8,
+ const ParamGenerator<T9>& g9,
+ const typename ParamGenerator<T9>::iterator& current9)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+ begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+ begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+ begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+ begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+ begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+ begin8_(g8.begin()), end8_(g8.end()), current8_(current8),
+ begin9_(g9.begin()), end9_(g9.end()), current9_(current9) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current9_;
+ if (current9_ == end9_) {
+ current9_ = begin9_;
+ ++current8_;
+ }
+ if (current8_ == end8_) {
+ current8_ = begin8_;
+ ++current7_;
+ }
+ if (current7_ == end7_) {
+ current7_ = begin7_;
+ ++current6_;
+ }
+ if (current6_ == end6_) {
+ current6_ = begin6_;
+ ++current5_;
+ }
+ if (current5_ == end5_) {
+ current5_ = begin5_;
+ ++current4_;
+ }
+ if (current4_ == end4_) {
+ current4_ = begin4_;
+ ++current3_;
+ }
+ if (current3_ == end3_) {
+ current3_ = begin3_;
+ ++current2_;
+ }
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_ &&
+ current3_ == typed_other->current3_ &&
+ current4_ == typed_other->current4_ &&
+ current5_ == typed_other->current5_ &&
+ current6_ == typed_other->current6_ &&
+ current7_ == typed_other->current7_ &&
+ current8_ == typed_other->current8_ &&
+ current9_ == typed_other->current9_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_),
+ begin3_(other.begin3_),
+ end3_(other.end3_),
+ current3_(other.current3_),
+ begin4_(other.begin4_),
+ end4_(other.end4_),
+ current4_(other.current4_),
+ begin5_(other.begin5_),
+ end5_(other.end5_),
+ current5_(other.current5_),
+ begin6_(other.begin6_),
+ end6_(other.end6_),
+ current6_(other.current6_),
+ begin7_(other.begin7_),
+ end7_(other.end7_),
+ current7_(other.current7_),
+ begin8_(other.begin8_),
+ end8_(other.end8_),
+ current8_(other.current8_),
+ begin9_(other.begin9_),
+ end9_(other.end9_),
+ current9_(other.current9_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_, *current3_,
+ *current4_, *current5_, *current6_, *current7_, *current8_,
+ *current9_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_ ||
+ current3_ == end3_ ||
+ current4_ == end4_ ||
+ current5_ == end5_ ||
+ current6_ == end6_ ||
+ current7_ == end7_ ||
+ current8_ == end8_ ||
+ current9_ == end9_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ const typename ParamGenerator<T3>::iterator begin3_;
+ const typename ParamGenerator<T3>::iterator end3_;
+ typename ParamGenerator<T3>::iterator current3_;
+ const typename ParamGenerator<T4>::iterator begin4_;
+ const typename ParamGenerator<T4>::iterator end4_;
+ typename ParamGenerator<T4>::iterator current4_;
+ const typename ParamGenerator<T5>::iterator begin5_;
+ const typename ParamGenerator<T5>::iterator end5_;
+ typename ParamGenerator<T5>::iterator current5_;
+ const typename ParamGenerator<T6>::iterator begin6_;
+ const typename ParamGenerator<T6>::iterator end6_;
+ typename ParamGenerator<T6>::iterator current6_;
+ const typename ParamGenerator<T7>::iterator begin7_;
+ const typename ParamGenerator<T7>::iterator end7_;
+ typename ParamGenerator<T7>::iterator current7_;
+ const typename ParamGenerator<T8>::iterator begin8_;
+ const typename ParamGenerator<T8>::iterator end8_;
+ typename ParamGenerator<T8>::iterator current8_;
+ const typename ParamGenerator<T9>::iterator begin9_;
+ const typename ParamGenerator<T9>::iterator end9_;
+ typename ParamGenerator<T9>::iterator current9_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator9::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator9& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+ const ParamGenerator<T3> g3_;
+ const ParamGenerator<T4> g4_;
+ const ParamGenerator<T5> g5_;
+ const ParamGenerator<T6> g6_;
+ const ParamGenerator<T7> g7_;
+ const ParamGenerator<T8> g8_;
+ const ParamGenerator<T9> g9_;
+}; // class CartesianProductGenerator9
+
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10>
+class CartesianProductGenerator10
+ : public ParamGeneratorInterface< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6,
+ T7, T8, T9, T10> > {
+ public:
+ typedef ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> ParamType;
+
+ CartesianProductGenerator10(const ParamGenerator<T1>& g1,
+ const ParamGenerator<T2>& g2, const ParamGenerator<T3>& g3,
+ const ParamGenerator<T4>& g4, const ParamGenerator<T5>& g5,
+ const ParamGenerator<T6>& g6, const ParamGenerator<T7>& g7,
+ const ParamGenerator<T8>& g8, const ParamGenerator<T9>& g9,
+ const ParamGenerator<T10>& g10)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+ g9_(g9), g10_(g10) {}
+ virtual ~CartesianProductGenerator10() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+ g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+ g7_.begin(), g8_, g8_.begin(), g9_, g9_.begin(), g10_, g10_.begin());
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+ g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+ g8_.end(), g9_, g9_.end(), g10_, g10_.end());
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base,
+ const ParamGenerator<T1>& g1,
+ const typename ParamGenerator<T1>::iterator& current1,
+ const ParamGenerator<T2>& g2,
+ const typename ParamGenerator<T2>::iterator& current2,
+ const ParamGenerator<T3>& g3,
+ const typename ParamGenerator<T3>::iterator& current3,
+ const ParamGenerator<T4>& g4,
+ const typename ParamGenerator<T4>::iterator& current4,
+ const ParamGenerator<T5>& g5,
+ const typename ParamGenerator<T5>::iterator& current5,
+ const ParamGenerator<T6>& g6,
+ const typename ParamGenerator<T6>::iterator& current6,
+ const ParamGenerator<T7>& g7,
+ const typename ParamGenerator<T7>::iterator& current7,
+ const ParamGenerator<T8>& g8,
+ const typename ParamGenerator<T8>::iterator& current8,
+ const ParamGenerator<T9>& g9,
+ const typename ParamGenerator<T9>::iterator& current9,
+ const ParamGenerator<T10>& g10,
+ const typename ParamGenerator<T10>::iterator& current10)
+ : base_(base),
+ begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+ begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+ begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+ begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+ begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+ begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+ begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+ begin8_(g8.begin()), end8_(g8.end()), current8_(current8),
+ begin9_(g9.begin()), end9_(g9.end()), current9_(current9),
+ begin10_(g10.begin()), end10_(g10.end()), current10_(current10) {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current10_;
+ if (current10_ == end10_) {
+ current10_ = begin10_;
+ ++current9_;
+ }
+ if (current9_ == end9_) {
+ current9_ = begin9_;
+ ++current8_;
+ }
+ if (current8_ == end8_) {
+ current8_ = begin8_;
+ ++current7_;
+ }
+ if (current7_ == end7_) {
+ current7_ = begin7_;
+ ++current6_;
+ }
+ if (current6_ == end6_) {
+ current6_ = begin6_;
+ ++current5_;
+ }
+ if (current5_ == end5_) {
+ current5_ = begin5_;
+ ++current4_;
+ }
+ if (current4_ == end4_) {
+ current4_ = begin4_;
+ ++current3_;
+ }
+ if (current3_ == end3_) {
+ current3_ = begin3_;
+ ++current2_;
+ }
+ if (current2_ == end2_) {
+ current2_ = begin2_;
+ ++current1_;
+ }
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ (
+ current1_ == typed_other->current1_ &&
+ current2_ == typed_other->current2_ &&
+ current3_ == typed_other->current3_ &&
+ current4_ == typed_other->current4_ &&
+ current5_ == typed_other->current5_ &&
+ current6_ == typed_other->current6_ &&
+ current7_ == typed_other->current7_ &&
+ current8_ == typed_other->current8_ &&
+ current9_ == typed_other->current9_ &&
+ current10_ == typed_other->current10_);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_),
+ begin1_(other.begin1_),
+ end1_(other.end1_),
+ current1_(other.current1_),
+ begin2_(other.begin2_),
+ end2_(other.end2_),
+ current2_(other.current2_),
+ begin3_(other.begin3_),
+ end3_(other.end3_),
+ current3_(other.current3_),
+ begin4_(other.begin4_),
+ end4_(other.end4_),
+ current4_(other.current4_),
+ begin5_(other.begin5_),
+ end5_(other.end5_),
+ current5_(other.current5_),
+ begin6_(other.begin6_),
+ end6_(other.end6_),
+ current6_(other.current6_),
+ begin7_(other.begin7_),
+ end7_(other.end7_),
+ current7_(other.current7_),
+ begin8_(other.begin8_),
+ end8_(other.end8_),
+ current8_(other.current8_),
+ begin9_(other.begin9_),
+ end9_(other.end9_),
+ current9_(other.current9_),
+ begin10_(other.begin10_),
+ end10_(other.end10_),
+ current10_(other.current10_) {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType(*current1_, *current2_, *current3_,
+ *current4_, *current5_, *current6_, *current7_, *current8_,
+ *current9_, *current10_);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+ current1_ == end1_ ||
+ current2_ == end2_ ||
+ current3_ == end3_ ||
+ current4_ == end4_ ||
+ current5_ == end5_ ||
+ current6_ == end6_ ||
+ current7_ == end7_ ||
+ current8_ == end8_ ||
+ current9_ == end9_ ||
+ current10_ == end10_;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+ const typename ParamGenerator<T1>::iterator begin1_;
+ const typename ParamGenerator<T1>::iterator end1_;
+ typename ParamGenerator<T1>::iterator current1_;
+ const typename ParamGenerator<T2>::iterator begin2_;
+ const typename ParamGenerator<T2>::iterator end2_;
+ typename ParamGenerator<T2>::iterator current2_;
+ const typename ParamGenerator<T3>::iterator begin3_;
+ const typename ParamGenerator<T3>::iterator end3_;
+ typename ParamGenerator<T3>::iterator current3_;
+ const typename ParamGenerator<T4>::iterator begin4_;
+ const typename ParamGenerator<T4>::iterator end4_;
+ typename ParamGenerator<T4>::iterator current4_;
+ const typename ParamGenerator<T5>::iterator begin5_;
+ const typename ParamGenerator<T5>::iterator end5_;
+ typename ParamGenerator<T5>::iterator current5_;
+ const typename ParamGenerator<T6>::iterator begin6_;
+ const typename ParamGenerator<T6>::iterator end6_;
+ typename ParamGenerator<T6>::iterator current6_;
+ const typename ParamGenerator<T7>::iterator begin7_;
+ const typename ParamGenerator<T7>::iterator end7_;
+ typename ParamGenerator<T7>::iterator current7_;
+ const typename ParamGenerator<T8>::iterator begin8_;
+ const typename ParamGenerator<T8>::iterator end8_;
+ typename ParamGenerator<T8>::iterator current8_;
+ const typename ParamGenerator<T9>::iterator begin9_;
+ const typename ParamGenerator<T9>::iterator end9_;
+ typename ParamGenerator<T9>::iterator current9_;
+ const typename ParamGenerator<T10>::iterator begin10_;
+ const typename ParamGenerator<T10>::iterator end10_;
+ typename ParamGenerator<T10>::iterator current10_;
+ ParamType current_value_;
+ }; // class CartesianProductGenerator10::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator10& other);
+
+ const ParamGenerator<T1> g1_;
+ const ParamGenerator<T2> g2_;
+ const ParamGenerator<T3> g3_;
+ const ParamGenerator<T4> g4_;
+ const ParamGenerator<T5> g5_;
+ const ParamGenerator<T6> g6_;
+ const ParamGenerator<T7> g7_;
+ const ParamGenerator<T8> g8_;
+ const ParamGenerator<T9> g9_;
+ const ParamGenerator<T10> g10_;
+}; // class CartesianProductGenerator10
+
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Helper classes providing Combine() with polymorphic features. They allow
+// casting CartesianProductGeneratorN<T> to ParamGenerator<U> if T is
+// convertible to U.
+//
+template <class Generator1, class Generator2>
+class CartesianProductHolder2 {
+ public:
+CartesianProductHolder2(const Generator1& g1, const Generator2& g2)
+ : g1_(g1), g2_(g2) {}
+ template <typename T1, typename T2>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2> >(
+ new CartesianProductGenerator2<T1, T2>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder2& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+}; // class CartesianProductHolder2
+
+template <class Generator1, class Generator2, class Generator3>
+class CartesianProductHolder3 {
+ public:
+CartesianProductHolder3(const Generator1& g1, const Generator2& g2,
+ const Generator3& g3)
+ : g1_(g1), g2_(g2), g3_(g3) {}
+ template <typename T1, typename T2, typename T3>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2, T3> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2, T3> >(
+ new CartesianProductGenerator3<T1, T2, T3>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_),
+ static_cast<ParamGenerator<T3> >(g3_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder3& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+ const Generator3 g3_;
+}; // class CartesianProductHolder3
+
+template <class Generator1, class Generator2, class Generator3,
+ class Generator4>
+class CartesianProductHolder4 {
+ public:
+CartesianProductHolder4(const Generator1& g1, const Generator2& g2,
+ const Generator3& g3, const Generator4& g4)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4) {}
+ template <typename T1, typename T2, typename T3, typename T4>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4> >(
+ new CartesianProductGenerator4<T1, T2, T3, T4>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_),
+ static_cast<ParamGenerator<T3> >(g3_),
+ static_cast<ParamGenerator<T4> >(g4_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder4& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+ const Generator3 g3_;
+ const Generator4 g4_;
+}; // class CartesianProductHolder4
+
+template <class Generator1, class Generator2, class Generator3,
+ class Generator4, class Generator5>
+class CartesianProductHolder5 {
+ public:
+CartesianProductHolder5(const Generator1& g1, const Generator2& g2,
+ const Generator3& g3, const Generator4& g4, const Generator5& g5)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5) {}
+ template <typename T1, typename T2, typename T3, typename T4, typename T5>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5> >(
+ new CartesianProductGenerator5<T1, T2, T3, T4, T5>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_),
+ static_cast<ParamGenerator<T3> >(g3_),
+ static_cast<ParamGenerator<T4> >(g4_),
+ static_cast<ParamGenerator<T5> >(g5_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder5& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+ const Generator3 g3_;
+ const Generator4 g4_;
+ const Generator5 g5_;
+}; // class CartesianProductHolder5
+
+template <class Generator1, class Generator2, class Generator3,
+ class Generator4, class Generator5, class Generator6>
+class CartesianProductHolder6 {
+ public:
+CartesianProductHolder6(const Generator1& g1, const Generator2& g2,
+ const Generator3& g3, const Generator4& g4, const Generator5& g5,
+ const Generator6& g6)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6) {}
+ template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6> >(
+ new CartesianProductGenerator6<T1, T2, T3, T4, T5, T6>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_),
+ static_cast<ParamGenerator<T3> >(g3_),
+ static_cast<ParamGenerator<T4> >(g4_),
+ static_cast<ParamGenerator<T5> >(g5_),
+ static_cast<ParamGenerator<T6> >(g6_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder6& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+ const Generator3 g3_;
+ const Generator4 g4_;
+ const Generator5 g5_;
+ const Generator6 g6_;
+}; // class CartesianProductHolder6
+
+template <class Generator1, class Generator2, class Generator3,
+ class Generator4, class Generator5, class Generator6, class Generator7>
+class CartesianProductHolder7 {
+ public:
+CartesianProductHolder7(const Generator1& g1, const Generator2& g2,
+ const Generator3& g3, const Generator4& g4, const Generator5& g5,
+ const Generator6& g6, const Generator7& g7)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7) {}
+ template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6,
+ T7> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7> >(
+ new CartesianProductGenerator7<T1, T2, T3, T4, T5, T6, T7>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_),
+ static_cast<ParamGenerator<T3> >(g3_),
+ static_cast<ParamGenerator<T4> >(g4_),
+ static_cast<ParamGenerator<T5> >(g5_),
+ static_cast<ParamGenerator<T6> >(g6_),
+ static_cast<ParamGenerator<T7> >(g7_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder7& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+ const Generator3 g3_;
+ const Generator4 g4_;
+ const Generator5 g5_;
+ const Generator6 g6_;
+ const Generator7 g7_;
+}; // class CartesianProductHolder7
+
+template <class Generator1, class Generator2, class Generator3,
+ class Generator4, class Generator5, class Generator6, class Generator7,
+ class Generator8>
+class CartesianProductHolder8 {
+ public:
+CartesianProductHolder8(const Generator1& g1, const Generator2& g2,
+ const Generator3& g3, const Generator4& g4, const Generator5& g5,
+ const Generator6& g6, const Generator7& g7, const Generator8& g8)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7),
+ g8_(g8) {}
+ template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7,
+ T8> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8> >(
+ new CartesianProductGenerator8<T1, T2, T3, T4, T5, T6, T7, T8>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_),
+ static_cast<ParamGenerator<T3> >(g3_),
+ static_cast<ParamGenerator<T4> >(g4_),
+ static_cast<ParamGenerator<T5> >(g5_),
+ static_cast<ParamGenerator<T6> >(g6_),
+ static_cast<ParamGenerator<T7> >(g7_),
+ static_cast<ParamGenerator<T8> >(g8_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder8& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+ const Generator3 g3_;
+ const Generator4 g4_;
+ const Generator5 g5_;
+ const Generator6 g6_;
+ const Generator7 g7_;
+ const Generator8 g8_;
+}; // class CartesianProductHolder8
+
+template <class Generator1, class Generator2, class Generator3,
+ class Generator4, class Generator5, class Generator6, class Generator7,
+ class Generator8, class Generator9>
+class CartesianProductHolder9 {
+ public:
+CartesianProductHolder9(const Generator1& g1, const Generator2& g2,
+ const Generator3& g3, const Generator4& g4, const Generator5& g5,
+ const Generator6& g6, const Generator7& g7, const Generator8& g8,
+ const Generator9& g9)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+ g9_(g9) {}
+ template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8,
+ T9> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8,
+ T9> >(
+ new CartesianProductGenerator9<T1, T2, T3, T4, T5, T6, T7, T8, T9>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_),
+ static_cast<ParamGenerator<T3> >(g3_),
+ static_cast<ParamGenerator<T4> >(g4_),
+ static_cast<ParamGenerator<T5> >(g5_),
+ static_cast<ParamGenerator<T6> >(g6_),
+ static_cast<ParamGenerator<T7> >(g7_),
+ static_cast<ParamGenerator<T8> >(g8_),
+ static_cast<ParamGenerator<T9> >(g9_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder9& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+ const Generator3 g3_;
+ const Generator4 g4_;
+ const Generator5 g5_;
+ const Generator6 g6_;
+ const Generator7 g7_;
+ const Generator8 g8_;
+ const Generator9 g9_;
+}; // class CartesianProductHolder9
+
+template <class Generator1, class Generator2, class Generator3,
+ class Generator4, class Generator5, class Generator6, class Generator7,
+ class Generator8, class Generator9, class Generator10>
+class CartesianProductHolder10 {
+ public:
+CartesianProductHolder10(const Generator1& g1, const Generator2& g2,
+ const Generator3& g3, const Generator4& g4, const Generator5& g5,
+ const Generator6& g6, const Generator7& g7, const Generator8& g8,
+ const Generator9& g9, const Generator10& g10)
+ : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+ g9_(g9), g10_(g10) {}
+ template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10>
+ operator ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8,
+ T9, T10> >() const {
+ return ParamGenerator< ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8,
+ T9, T10> >(
+ new CartesianProductGenerator10<T1, T2, T3, T4, T5, T6, T7, T8, T9,
+ T10>(
+ static_cast<ParamGenerator<T1> >(g1_),
+ static_cast<ParamGenerator<T2> >(g2_),
+ static_cast<ParamGenerator<T3> >(g3_),
+ static_cast<ParamGenerator<T4> >(g4_),
+ static_cast<ParamGenerator<T5> >(g5_),
+ static_cast<ParamGenerator<T6> >(g6_),
+ static_cast<ParamGenerator<T7> >(g7_),
+ static_cast<ParamGenerator<T8> >(g8_),
+ static_cast<ParamGenerator<T9> >(g9_),
+ static_cast<ParamGenerator<T10> >(g10_)));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder10& other);
+
+ const Generator1 g1_;
+ const Generator2 g2_;
+ const Generator3 g3_;
+ const Generator4 g4_;
+ const Generator5 g5_;
+ const Generator6 g6_;
+ const Generator7 g7_;
+ const Generator8 g8_;
+ const Generator9 g9_;
+ const Generator10 g10_;
+}; // class CartesianProductHolder10
+
+# endif // GTEST_HAS_COMBINE
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_HAS_PARAM_TEST
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util-generated.h.pump b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util-generated.h.pump
new file mode 100644
index 0000000..dbe9386
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util-generated.h.pump
@@ -0,0 +1,301 @@
+$$ -*- mode: c++; -*-
+$var n = 50 $$ Maximum length of Values arguments we want to support.
+$var maxtuple = 10 $$ Maximum number of Combine arguments we want to support.
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// Type and function utilities for implementing parameterized tests.
+// This file is generated by a SCRIPT. DO NOT EDIT BY HAND!
+//
+// Currently Google Test supports at most $n arguments in Values,
+// and at most $maxtuple arguments in Combine. Please contact
+// googletestframework@googlegroups.com if you need more.
+// Please note that the number of arguments to Combine is limited
+// by the maximum arity of the implementation of tr1::tuple which is
+// currently set at $maxtuple.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*. Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+#include "gtest/internal/gtest-param-util.h"
+#include "gtest/internal/gtest-port.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing {
+
+// Forward declarations of ValuesIn(), which is implemented in
+// include/gtest/gtest-param-test.h.
+template <typename ForwardIterator>
+internal::ParamGenerator<
+ typename ::testing::internal::IteratorTraits<ForwardIterator>::value_type>
+ValuesIn(ForwardIterator begin, ForwardIterator end);
+
+template <typename T, size_t N>
+internal::ParamGenerator<T> ValuesIn(const T (&array)[N]);
+
+template <class Container>
+internal::ParamGenerator<typename Container::value_type> ValuesIn(
+ const Container& container);
+
+namespace internal {
+
+// Used in the Values() function to provide polymorphic capabilities.
+template <typename T1>
+class ValueArray1 {
+ public:
+ explicit ValueArray1(T1 v1) : v1_(v1) {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const { return ValuesIn(&v1_, &v1_ + 1); }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray1& other);
+
+ const T1 v1_;
+};
+
+$range i 2..n
+$for i [[
+$range j 1..i
+
+template <$for j, [[typename T$j]]>
+class ValueArray$i {
+ public:
+ ValueArray$i($for j, [[T$j v$j]]) : $for j, [[v$(j)_(v$j)]] {}
+
+ template <typename T>
+ operator ParamGenerator<T>() const {
+ const T array[] = {$for j, [[v$(j)_]]};
+ return ValuesIn(array);
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const ValueArray$i& other);
+
+$for j [[
+
+ const T$j v$(j)_;
+]]
+
+};
+
+]]
+
+# if GTEST_HAS_COMBINE
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Generates values from the Cartesian product of values produced
+// by the argument generators.
+//
+$range i 2..maxtuple
+$for i [[
+$range j 1..i
+$range k 2..i
+
+template <$for j, [[typename T$j]]>
+class CartesianProductGenerator$i
+ : public ParamGeneratorInterface< ::std::tr1::tuple<$for j, [[T$j]]> > {
+ public:
+ typedef ::std::tr1::tuple<$for j, [[T$j]]> ParamType;
+
+ CartesianProductGenerator$i($for j, [[const ParamGenerator<T$j>& g$j]])
+ : $for j, [[g$(j)_(g$j)]] {}
+ virtual ~CartesianProductGenerator$i() {}
+
+ virtual ParamIteratorInterface<ParamType>* Begin() const {
+ return new Iterator(this, $for j, [[g$(j)_, g$(j)_.begin()]]);
+ }
+ virtual ParamIteratorInterface<ParamType>* End() const {
+ return new Iterator(this, $for j, [[g$(j)_, g$(j)_.end()]]);
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<ParamType> {
+ public:
+ Iterator(const ParamGeneratorInterface<ParamType>* base, $for j, [[
+
+ const ParamGenerator<T$j>& g$j,
+ const typename ParamGenerator<T$j>::iterator& current$(j)]])
+ : base_(base),
+$for j, [[
+
+ begin$(j)_(g$j.begin()), end$(j)_(g$j.end()), current$(j)_(current$j)
+]] {
+ ComputeCurrentValue();
+ }
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<ParamType>* BaseGenerator() const {
+ return base_;
+ }
+ // Advance should not be called on beyond-of-range iterators
+ // so no component iterators must be beyond end of range, either.
+ virtual void Advance() {
+ assert(!AtEnd());
+ ++current$(i)_;
+
+$for k [[
+ if (current$(i+2-k)_ == end$(i+2-k)_) {
+ current$(i+2-k)_ = begin$(i+2-k)_;
+ ++current$(i+2-k-1)_;
+ }
+
+]]
+ ComputeCurrentValue();
+ }
+ virtual ParamIteratorInterface<ParamType>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const ParamType* Current() const { return ¤t_value_; }
+ virtual bool Equals(const ParamIteratorInterface<ParamType>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const Iterator* typed_other =
+ CheckedDowncastToActualType<const Iterator>(&other);
+ // We must report iterators equal if they both point beyond their
+ // respective ranges. That can happen in a variety of fashions,
+ // so we have to consult AtEnd().
+ return (AtEnd() && typed_other->AtEnd()) ||
+ ($for j && [[
+
+ current$(j)_ == typed_other->current$(j)_
+]]);
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : base_(other.base_), $for j, [[
+
+ begin$(j)_(other.begin$(j)_),
+ end$(j)_(other.end$(j)_),
+ current$(j)_(other.current$(j)_)
+]] {
+ ComputeCurrentValue();
+ }
+
+ void ComputeCurrentValue() {
+ if (!AtEnd())
+ current_value_ = ParamType($for j, [[*current$(j)_]]);
+ }
+ bool AtEnd() const {
+ // We must report iterator past the end of the range when either of the
+ // component iterators has reached the end of its range.
+ return
+$for j || [[
+
+ current$(j)_ == end$(j)_
+]];
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<ParamType>* const base_;
+ // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+ // current[i]_ is the actual traversing iterator.
+$for j [[
+
+ const typename ParamGenerator<T$j>::iterator begin$(j)_;
+ const typename ParamGenerator<T$j>::iterator end$(j)_;
+ typename ParamGenerator<T$j>::iterator current$(j)_;
+]]
+
+ ParamType current_value_;
+ }; // class CartesianProductGenerator$i::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductGenerator$i& other);
+
+
+$for j [[
+ const ParamGenerator<T$j> g$(j)_;
+
+]]
+}; // class CartesianProductGenerator$i
+
+
+]]
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Helper classes providing Combine() with polymorphic features. They allow
+// casting CartesianProductGeneratorN<T> to ParamGenerator<U> if T is
+// convertible to U.
+//
+$range i 2..maxtuple
+$for i [[
+$range j 1..i
+
+template <$for j, [[class Generator$j]]>
+class CartesianProductHolder$i {
+ public:
+CartesianProductHolder$i($for j, [[const Generator$j& g$j]])
+ : $for j, [[g$(j)_(g$j)]] {}
+ template <$for j, [[typename T$j]]>
+ operator ParamGenerator< ::std::tr1::tuple<$for j, [[T$j]]> >() const {
+ return ParamGenerator< ::std::tr1::tuple<$for j, [[T$j]]> >(
+ new CartesianProductGenerator$i<$for j, [[T$j]]>(
+$for j,[[
+
+ static_cast<ParamGenerator<T$j> >(g$(j)_)
+]]));
+ }
+
+ private:
+ // No implementation - assignment is unsupported.
+ void operator=(const CartesianProductHolder$i& other);
+
+
+$for j [[
+ const Generator$j g$(j)_;
+
+]]
+}; // class CartesianProductHolder$i
+
+]]
+
+# endif // GTEST_HAS_COMBINE
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_HAS_PARAM_TEST
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util.h
new file mode 100644
index 0000000..0ef9718
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-param-util.h
@@ -0,0 +1,619 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// Type and function utilities for implementing parameterized tests.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
+
+#include <iterator>
+#include <utility>
+#include <vector>
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*. Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+#include "gtest/internal/gtest-internal.h"
+#include "gtest/internal/gtest-linked_ptr.h"
+#include "gtest/internal/gtest-port.h"
+#include "gtest/gtest-printers.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing {
+namespace internal {
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Outputs a message explaining invalid registration of different
+// fixture class for the same test case. This may happen when
+// TEST_P macro is used to define two tests with the same name
+// but in different namespaces.
+GTEST_API_ void ReportInvalidTestCaseType(const char* test_case_name,
+ const char* file, int line);
+
+template <typename> class ParamGeneratorInterface;
+template <typename> class ParamGenerator;
+
+// Interface for iterating over elements provided by an implementation
+// of ParamGeneratorInterface<T>.
+template <typename T>
+class ParamIteratorInterface {
+ public:
+ virtual ~ParamIteratorInterface() {}
+ // A pointer to the base generator instance.
+ // Used only for the purposes of iterator comparison
+ // to make sure that two iterators belong to the same generator.
+ virtual const ParamGeneratorInterface<T>* BaseGenerator() const = 0;
+ // Advances iterator to point to the next element
+ // provided by the generator. The caller is responsible
+ // for not calling Advance() on an iterator equal to
+ // BaseGenerator()->End().
+ virtual void Advance() = 0;
+ // Clones the iterator object. Used for implementing copy semantics
+ // of ParamIterator<T>.
+ virtual ParamIteratorInterface* Clone() const = 0;
+ // Dereferences the current iterator and provides (read-only) access
+ // to the pointed value. It is the caller's responsibility not to call
+ // Current() on an iterator equal to BaseGenerator()->End().
+ // Used for implementing ParamGenerator<T>::operator*().
+ virtual const T* Current() const = 0;
+ // Determines whether the given iterator and other point to the same
+ // element in the sequence generated by the generator.
+ // Used for implementing ParamGenerator<T>::operator==().
+ virtual bool Equals(const ParamIteratorInterface& other) const = 0;
+};
+
+// Class iterating over elements provided by an implementation of
+// ParamGeneratorInterface<T>. It wraps ParamIteratorInterface<T>
+// and implements the const forward iterator concept.
+template <typename T>
+class ParamIterator {
+ public:
+ typedef T value_type;
+ typedef const T& reference;
+ typedef ptrdiff_t difference_type;
+
+ // ParamIterator assumes ownership of the impl_ pointer.
+ ParamIterator(const ParamIterator& other) : impl_(other.impl_->Clone()) {}
+ ParamIterator& operator=(const ParamIterator& other) {
+ if (this != &other)
+ impl_.reset(other.impl_->Clone());
+ return *this;
+ }
+
+ const T& operator*() const { return *impl_->Current(); }
+ const T* operator->() const { return impl_->Current(); }
+ // Prefix version of operator++.
+ ParamIterator& operator++() {
+ impl_->Advance();
+ return *this;
+ }
+ // Postfix version of operator++.
+ ParamIterator operator++(int /*unused*/) {
+ ParamIteratorInterface<T>* clone = impl_->Clone();
+ impl_->Advance();
+ return ParamIterator(clone);
+ }
+ bool operator==(const ParamIterator& other) const {
+ return impl_.get() == other.impl_.get() || impl_->Equals(*other.impl_);
+ }
+ bool operator!=(const ParamIterator& other) const {
+ return !(*this == other);
+ }
+
+ private:
+ friend class ParamGenerator<T>;
+ explicit ParamIterator(ParamIteratorInterface<T>* impl) : impl_(impl) {}
+ scoped_ptr<ParamIteratorInterface<T> > impl_;
+};
+
+// ParamGeneratorInterface<T> is the binary interface to access generators
+// defined in other translation units.
+template <typename T>
+class ParamGeneratorInterface {
+ public:
+ typedef T ParamType;
+
+ virtual ~ParamGeneratorInterface() {}
+
+ // Generator interface definition
+ virtual ParamIteratorInterface<T>* Begin() const = 0;
+ virtual ParamIteratorInterface<T>* End() const = 0;
+};
+
+// Wraps ParamGeneratorInterface<T> and provides general generator syntax
+// compatible with the STL Container concept.
+// This class implements copy initialization semantics and the contained
+// ParamGeneratorInterface<T> instance is shared among all copies
+// of the original object. This is possible because that instance is immutable.
+template<typename T>
+class ParamGenerator {
+ public:
+ typedef ParamIterator<T> iterator;
+
+ explicit ParamGenerator(ParamGeneratorInterface<T>* impl) : impl_(impl) {}
+ ParamGenerator(const ParamGenerator& other) : impl_(other.impl_) {}
+
+ ParamGenerator& operator=(const ParamGenerator& other) {
+ impl_ = other.impl_;
+ return *this;
+ }
+
+ iterator begin() const { return iterator(impl_->Begin()); }
+ iterator end() const { return iterator(impl_->End()); }
+
+ private:
+ linked_ptr<const ParamGeneratorInterface<T> > impl_;
+};
+
+// Generates values from a range of two comparable values. Can be used to
+// generate sequences of user-defined types that implement operator+() and
+// operator<().
+// This class is used in the Range() function.
+template <typename T, typename IncrementT>
+class RangeGenerator : public ParamGeneratorInterface<T> {
+ public:
+ RangeGenerator(T begin, T end, IncrementT step)
+ : begin_(begin), end_(end),
+ step_(step), end_index_(CalculateEndIndex(begin, end, step)) {}
+ virtual ~RangeGenerator() {}
+
+ virtual ParamIteratorInterface<T>* Begin() const {
+ return new Iterator(this, begin_, 0, step_);
+ }
+ virtual ParamIteratorInterface<T>* End() const {
+ return new Iterator(this, end_, end_index_, step_);
+ }
+
+ private:
+ class Iterator : public ParamIteratorInterface<T> {
+ public:
+ Iterator(const ParamGeneratorInterface<T>* base, T value, int index,
+ IncrementT step)
+ : base_(base), value_(value), index_(index), step_(step) {}
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<T>* BaseGenerator() const {
+ return base_;
+ }
+ virtual void Advance() {
+ value_ = value_ + step_;
+ index_++;
+ }
+ virtual ParamIteratorInterface<T>* Clone() const {
+ return new Iterator(*this);
+ }
+ virtual const T* Current() const { return &value_; }
+ virtual bool Equals(const ParamIteratorInterface<T>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ const int other_index =
+ CheckedDowncastToActualType<const Iterator>(&other)->index_;
+ return index_ == other_index;
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ : ParamIteratorInterface<T>(),
+ base_(other.base_), value_(other.value_), index_(other.index_),
+ step_(other.step_) {}
+
+ // No implementation - assignment is unsupported.
+ void operator=(const Iterator& other);
+
+ const ParamGeneratorInterface<T>* const base_;
+ T value_;
+ int index_;
+ const IncrementT step_;
+ }; // class RangeGenerator::Iterator
+
+ static int CalculateEndIndex(const T& begin,
+ const T& end,
+ const IncrementT& step) {
+ int end_index = 0;
+ for (T i = begin; i < end; i = i + step)
+ end_index++;
+ return end_index;
+ }
+
+ // No implementation - assignment is unsupported.
+ void operator=(const RangeGenerator& other);
+
+ const T begin_;
+ const T end_;
+ const IncrementT step_;
+ // The index for the end() iterator. All the elements in the generated
+ // sequence are indexed (0-based) to aid iterator comparison.
+ const int end_index_;
+}; // class RangeGenerator
+
+
+// Generates values from a pair of STL-style iterators. Used in the
+// ValuesIn() function. The elements are copied from the source range
+// since the source can be located on the stack, and the generator
+// is likely to persist beyond that stack frame.
+template <typename T>
+class ValuesInIteratorRangeGenerator : public ParamGeneratorInterface<T> {
+ public:
+ template <typename ForwardIterator>
+ ValuesInIteratorRangeGenerator(ForwardIterator begin, ForwardIterator end)
+ : container_(begin, end) {}
+ virtual ~ValuesInIteratorRangeGenerator() {}
+
+ virtual ParamIteratorInterface<T>* Begin() const {
+ return new Iterator(this, container_.begin());
+ }
+ virtual ParamIteratorInterface<T>* End() const {
+ return new Iterator(this, container_.end());
+ }
+
+ private:
+ typedef typename ::std::vector<T> ContainerType;
+
+ class Iterator : public ParamIteratorInterface<T> {
+ public:
+ Iterator(const ParamGeneratorInterface<T>* base,
+ typename ContainerType::const_iterator iterator)
+ : base_(base), iterator_(iterator) {}
+ virtual ~Iterator() {}
+
+ virtual const ParamGeneratorInterface<T>* BaseGenerator() const {
+ return base_;
+ }
+ virtual void Advance() {
+ ++iterator_;
+ value_.reset();
+ }
+ virtual ParamIteratorInterface<T>* Clone() const {
+ return new Iterator(*this);
+ }
+ // We need to use cached value referenced by iterator_ because *iterator_
+ // can return a temporary object (and of type other then T), so just
+ // having "return &*iterator_;" doesn't work.
+ // value_ is updated here and not in Advance() because Advance()
+ // can advance iterator_ beyond the end of the range, and we cannot
+ // detect that fact. The client code, on the other hand, is
+ // responsible for not calling Current() on an out-of-range iterator.
+ virtual const T* Current() const {
+ if (value_.get() == NULL)
+ value_.reset(new T(*iterator_));
+ return value_.get();
+ }
+ virtual bool Equals(const ParamIteratorInterface<T>& other) const {
+ // Having the same base generator guarantees that the other
+ // iterator is of the same type and we can downcast.
+ GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+ << "The program attempted to compare iterators "
+ << "from different generators." << std::endl;
+ return iterator_ ==
+ CheckedDowncastToActualType<const Iterator>(&other)->iterator_;
+ }
+
+ private:
+ Iterator(const Iterator& other)
+ // The explicit constructor call suppresses a false warning
+ // emitted by gcc when supplied with the -Wextra option.
+ : ParamIteratorInterface<T>(),
+ base_(other.base_),
+ iterator_(other.iterator_) {}
+
+ const ParamGeneratorInterface<T>* const base_;
+ typename ContainerType::const_iterator iterator_;
+ // A cached value of *iterator_. We keep it here to allow access by
+ // pointer in the wrapping iterator's operator->().
+ // value_ needs to be mutable to be accessed in Current().
+ // Use of scoped_ptr helps manage cached value's lifetime,
+ // which is bound by the lifespan of the iterator itself.
+ mutable scoped_ptr<const T> value_;
+ }; // class ValuesInIteratorRangeGenerator::Iterator
+
+ // No implementation - assignment is unsupported.
+ void operator=(const ValuesInIteratorRangeGenerator& other);
+
+ const ContainerType container_;
+}; // class ValuesInIteratorRangeGenerator
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Stores a parameter value and later creates tests parameterized with that
+// value.
+template <class TestClass>
+class ParameterizedTestFactory : public TestFactoryBase {
+ public:
+ typedef typename TestClass::ParamType ParamType;
+ explicit ParameterizedTestFactory(ParamType parameter) :
+ parameter_(parameter) {}
+ virtual Test* CreateTest() {
+ TestClass::SetParam(¶meter_);
+ return new TestClass();
+ }
+
+ private:
+ const ParamType parameter_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestFactory);
+};
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// TestMetaFactoryBase is a base class for meta-factories that create
+// test factories for passing into MakeAndRegisterTestInfo function.
+template <class ParamType>
+class TestMetaFactoryBase {
+ public:
+ virtual ~TestMetaFactoryBase() {}
+
+ virtual TestFactoryBase* CreateTestFactory(ParamType parameter) = 0;
+};
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// TestMetaFactory creates test factories for passing into
+// MakeAndRegisterTestInfo function. Since MakeAndRegisterTestInfo receives
+// ownership of test factory pointer, same factory object cannot be passed
+// into that method twice. But ParameterizedTestCaseInfo is going to call
+// it for each Test/Parameter value combination. Thus it needs meta factory
+// creator class.
+template <class TestCase>
+class TestMetaFactory
+ : public TestMetaFactoryBase<typename TestCase::ParamType> {
+ public:
+ typedef typename TestCase::ParamType ParamType;
+
+ TestMetaFactory() {}
+
+ virtual TestFactoryBase* CreateTestFactory(ParamType parameter) {
+ return new ParameterizedTestFactory<TestCase>(parameter);
+ }
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestMetaFactory);
+};
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseInfoBase is a generic interface
+// to ParameterizedTestCaseInfo classes. ParameterizedTestCaseInfoBase
+// accumulates test information provided by TEST_P macro invocations
+// and generators provided by INSTANTIATE_TEST_CASE_P macro invocations
+// and uses that information to register all resulting test instances
+// in RegisterTests method. The ParameterizeTestCaseRegistry class holds
+// a collection of pointers to the ParameterizedTestCaseInfo objects
+// and calls RegisterTests() on each of them when asked.
+class ParameterizedTestCaseInfoBase {
+ public:
+ virtual ~ParameterizedTestCaseInfoBase() {}
+
+ // Base part of test case name for display purposes.
+ virtual const string& GetTestCaseName() const = 0;
+ // Test case id to verify identity.
+ virtual TypeId GetTestCaseTypeId() const = 0;
+ // UnitTest class invokes this method to register tests in this
+ // test case right before running them in RUN_ALL_TESTS macro.
+ // This method should not be called more then once on any single
+ // instance of a ParameterizedTestCaseInfoBase derived class.
+ virtual void RegisterTests() = 0;
+
+ protected:
+ ParameterizedTestCaseInfoBase() {}
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseInfoBase);
+};
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseInfo accumulates tests obtained from TEST_P
+// macro invocations for a particular test case and generators
+// obtained from INSTANTIATE_TEST_CASE_P macro invocations for that
+// test case. It registers tests with all values generated by all
+// generators when asked.
+template <class TestCase>
+class ParameterizedTestCaseInfo : public ParameterizedTestCaseInfoBase {
+ public:
+ // ParamType and GeneratorCreationFunc are private types but are required
+ // for declarations of public methods AddTestPattern() and
+ // AddTestCaseInstantiation().
+ typedef typename TestCase::ParamType ParamType;
+ // A function that returns an instance of appropriate generator type.
+ typedef ParamGenerator<ParamType>(GeneratorCreationFunc)();
+
+ explicit ParameterizedTestCaseInfo(const char* name)
+ : test_case_name_(name) {}
+
+ // Test case base name for display purposes.
+ virtual const string& GetTestCaseName() const { return test_case_name_; }
+ // Test case id to verify identity.
+ virtual TypeId GetTestCaseTypeId() const { return GetTypeId<TestCase>(); }
+ // TEST_P macro uses AddTestPattern() to record information
+ // about a single test in a LocalTestInfo structure.
+ // test_case_name is the base name of the test case (without invocation
+ // prefix). test_base_name is the name of an individual test without
+ // parameter index. For the test SequenceA/FooTest.DoBar/1 FooTest is
+ // test case base name and DoBar is test base name.
+ void AddTestPattern(const char* test_case_name,
+ const char* test_base_name,
+ TestMetaFactoryBase<ParamType>* meta_factory) {
+ tests_.push_back(linked_ptr<TestInfo>(new TestInfo(test_case_name,
+ test_base_name,
+ meta_factory)));
+ }
+ // INSTANTIATE_TEST_CASE_P macro uses AddGenerator() to record information
+ // about a generator.
+ int AddTestCaseInstantiation(const string& instantiation_name,
+ GeneratorCreationFunc* func,
+ const char* /* file */,
+ int /* line */) {
+ instantiations_.push_back(::std::make_pair(instantiation_name, func));
+ return 0; // Return value used only to run this method in namespace scope.
+ }
+ // UnitTest class invokes this method to register tests in this test case
+ // test cases right before running tests in RUN_ALL_TESTS macro.
+ // This method should not be called more then once on any single
+ // instance of a ParameterizedTestCaseInfoBase derived class.
+ // UnitTest has a guard to prevent from calling this method more then once.
+ virtual void RegisterTests() {
+ for (typename TestInfoContainer::iterator test_it = tests_.begin();
+ test_it != tests_.end(); ++test_it) {
+ linked_ptr<TestInfo> test_info = *test_it;
+ for (typename InstantiationContainer::iterator gen_it =
+ instantiations_.begin(); gen_it != instantiations_.end();
+ ++gen_it) {
+ const string& instantiation_name = gen_it->first;
+ ParamGenerator<ParamType> generator((*gen_it->second)());
+
+ Message test_case_name_stream;
+ if ( !instantiation_name.empty() )
+ test_case_name_stream << instantiation_name << "/";
+ test_case_name_stream << test_info->test_case_base_name;
+
+ int i = 0;
+ for (typename ParamGenerator<ParamType>::iterator param_it =
+ generator.begin();
+ param_it != generator.end(); ++param_it, ++i) {
+ Message test_name_stream;
+ test_name_stream << test_info->test_base_name << "/" << i;
+ MakeAndRegisterTestInfo(
+ test_case_name_stream.GetString().c_str(),
+ test_name_stream.GetString().c_str(),
+ NULL, // No type parameter.
+ PrintToString(*param_it).c_str(),
+ GetTestCaseTypeId(),
+ TestCase::SetUpTestCase,
+ TestCase::TearDownTestCase,
+ test_info->test_meta_factory->CreateTestFactory(*param_it));
+ } // for param_it
+ } // for gen_it
+ } // for test_it
+ } // RegisterTests
+
+ private:
+ // LocalTestInfo structure keeps information about a single test registered
+ // with TEST_P macro.
+ struct TestInfo {
+ TestInfo(const char* a_test_case_base_name,
+ const char* a_test_base_name,
+ TestMetaFactoryBase<ParamType>* a_test_meta_factory) :
+ test_case_base_name(a_test_case_base_name),
+ test_base_name(a_test_base_name),
+ test_meta_factory(a_test_meta_factory) {}
+
+ const string test_case_base_name;
+ const string test_base_name;
+ const scoped_ptr<TestMetaFactoryBase<ParamType> > test_meta_factory;
+ };
+ typedef ::std::vector<linked_ptr<TestInfo> > TestInfoContainer;
+ // Keeps pairs of <Instantiation name, Sequence generator creation function>
+ // received from INSTANTIATE_TEST_CASE_P macros.
+ typedef ::std::vector<std::pair<string, GeneratorCreationFunc*> >
+ InstantiationContainer;
+
+ const string test_case_name_;
+ TestInfoContainer tests_;
+ InstantiationContainer instantiations_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseInfo);
+}; // class ParameterizedTestCaseInfo
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseRegistry contains a map of ParameterizedTestCaseInfoBase
+// classes accessed by test case names. TEST_P and INSTANTIATE_TEST_CASE_P
+// macros use it to locate their corresponding ParameterizedTestCaseInfo
+// descriptors.
+class ParameterizedTestCaseRegistry {
+ public:
+ ParameterizedTestCaseRegistry() {}
+ ~ParameterizedTestCaseRegistry() {
+ for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+ it != test_case_infos_.end(); ++it) {
+ delete *it;
+ }
+ }
+
+ // Looks up or creates and returns a structure containing information about
+ // tests and instantiations of a particular test case.
+ template <class TestCase>
+ ParameterizedTestCaseInfo<TestCase>* GetTestCasePatternHolder(
+ const char* test_case_name,
+ const char* file,
+ int line) {
+ ParameterizedTestCaseInfo<TestCase>* typed_test_info = NULL;
+ for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+ it != test_case_infos_.end(); ++it) {
+ if ((*it)->GetTestCaseName() == test_case_name) {
+ if ((*it)->GetTestCaseTypeId() != GetTypeId<TestCase>()) {
+ // Complain about incorrect usage of Google Test facilities
+ // and terminate the program since we cannot guaranty correct
+ // test case setup and tear-down in this case.
+ ReportInvalidTestCaseType(test_case_name, file, line);
+ posix::Abort();
+ } else {
+ // At this point we are sure that the object we found is of the same
+ // type we are looking for, so we downcast it to that type
+ // without further checks.
+ typed_test_info = CheckedDowncastToActualType<
+ ParameterizedTestCaseInfo<TestCase> >(*it);
+ }
+ break;
+ }
+ }
+ if (typed_test_info == NULL) {
+ typed_test_info = new ParameterizedTestCaseInfo<TestCase>(test_case_name);
+ test_case_infos_.push_back(typed_test_info);
+ }
+ return typed_test_info;
+ }
+ void RegisterTests() {
+ for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+ it != test_case_infos_.end(); ++it) {
+ (*it)->RegisterTests();
+ }
+ }
+
+ private:
+ typedef ::std::vector<ParameterizedTestCaseInfoBase*> TestCaseInfoContainer;
+
+ TestCaseInfoContainer test_case_infos_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseRegistry);
+};
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_HAS_PARAM_TEST
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-port.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-port.h
new file mode 100644
index 0000000..157b47f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-port.h
@@ -0,0 +1,1775 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan)
+//
+// Low-level types and utilities for porting Google Test to various
+// platforms. They are subject to change without notice. DO NOT USE
+// THEM IN USER CODE.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
+
+// The user can define the following macros in the build script to
+// control Google Test's behavior. If the user doesn't define a macro
+// in this list, Google Test will define it.
+//
+// GTEST_HAS_CLONE - Define it to 1/0 to indicate that clone(2)
+// is/isn't available.
+// GTEST_HAS_EXCEPTIONS - Define it to 1/0 to indicate that exceptions
+// are enabled.
+// GTEST_HAS_GLOBAL_STRING - Define it to 1/0 to indicate that ::string
+// is/isn't available (some systems define
+// ::string, which is different to std::string).
+// GTEST_HAS_GLOBAL_WSTRING - Define it to 1/0 to indicate that ::string
+// is/isn't available (some systems define
+// ::wstring, which is different to std::wstring).
+// GTEST_HAS_POSIX_RE - Define it to 1/0 to indicate that POSIX regular
+// expressions are/aren't available.
+// GTEST_HAS_PTHREAD - Define it to 1/0 to indicate that <pthread.h>
+// is/isn't available.
+// GTEST_HAS_RTTI - Define it to 1/0 to indicate that RTTI is/isn't
+// enabled.
+// GTEST_HAS_STD_WSTRING - Define it to 1/0 to indicate that
+// std::wstring does/doesn't work (Google Test can
+// be used where std::wstring is unavailable).
+// GTEST_HAS_TR1_TUPLE - Define it to 1/0 to indicate tr1::tuple
+// is/isn't available.
+// GTEST_HAS_SEH - Define it to 1/0 to indicate whether the
+// compiler supports Microsoft's "Structured
+// Exception Handling".
+// GTEST_HAS_STREAM_REDIRECTION
+// - Define it to 1/0 to indicate whether the
+// platform supports I/O stream redirection using
+// dup() and dup2().
+// GTEST_USE_OWN_TR1_TUPLE - Define it to 1/0 to indicate whether Google
+// Test's own tr1 tuple implementation should be
+// used. Unused when the user sets
+// GTEST_HAS_TR1_TUPLE to 0.
+// GTEST_LINKED_AS_SHARED_LIBRARY
+// - Define to 1 when compiling tests that use
+// Google Test as a shared library (known as
+// DLL on Windows).
+// GTEST_CREATE_SHARED_LIBRARY
+// - Define to 1 when compiling Google Test itself
+// as a shared library.
+
+// This header defines the following utilities:
+//
+// Macros indicating the current platform (defined to 1 if compiled on
+// the given platform; otherwise undefined):
+// GTEST_OS_AIX - IBM AIX
+// GTEST_OS_CYGWIN - Cygwin
+// GTEST_OS_HPUX - HP-UX
+// GTEST_OS_LINUX - Linux
+// GTEST_OS_LINUX_ANDROID - Google Android
+// GTEST_OS_MAC - Mac OS X
+// GTEST_OS_NACL - Google Native Client (NaCl)
+// GTEST_OS_SOLARIS - Sun Solaris
+// GTEST_OS_SYMBIAN - Symbian
+// GTEST_OS_WINDOWS - Windows (Desktop, MinGW, or Mobile)
+// GTEST_OS_WINDOWS_DESKTOP - Windows Desktop
+// GTEST_OS_WINDOWS_MINGW - MinGW
+// GTEST_OS_WINDOWS_MOBILE - Windows Mobile
+// GTEST_OS_ZOS - z/OS
+//
+// Among the platforms, Cygwin, Linux, Max OS X, and Windows have the
+// most stable support. Since core members of the Google Test project
+// don't have access to other platforms, support for them may be less
+// stable. If you notice any problems on your platform, please notify
+// googletestframework@googlegroups.com (patches for fixing them are
+// even more welcome!).
+//
+// Note that it is possible that none of the GTEST_OS_* macros are defined.
+//
+// Macros indicating available Google Test features (defined to 1 if
+// the corresponding feature is supported; otherwise undefined):
+// GTEST_HAS_COMBINE - the Combine() function (for value-parameterized
+// tests)
+// GTEST_HAS_DEATH_TEST - death tests
+// GTEST_HAS_PARAM_TEST - value-parameterized tests
+// GTEST_HAS_TYPED_TEST - typed tests
+// GTEST_HAS_TYPED_TEST_P - type-parameterized tests
+// GTEST_USES_POSIX_RE - enhanced POSIX regex is used. Do not confuse with
+// GTEST_HAS_POSIX_RE (see above) which users can
+// define themselves.
+// GTEST_USES_SIMPLE_RE - our own simple regex is used;
+// the above two are mutually exclusive.
+// GTEST_CAN_COMPARE_NULL - accepts untyped NULL in EXPECT_EQ().
+//
+// Macros for basic C++ coding:
+// GTEST_AMBIGUOUS_ELSE_BLOCKER_ - for disabling a gcc warning.
+// GTEST_ATTRIBUTE_UNUSED_ - declares that a class' instances or a
+// variable don't have to be used.
+// GTEST_DISALLOW_ASSIGN_ - disables operator=.
+// GTEST_DISALLOW_COPY_AND_ASSIGN_ - disables copy ctor and operator=.
+// GTEST_MUST_USE_RESULT_ - declares that a function's result must be used.
+//
+// Synchronization:
+// Mutex, MutexLock, ThreadLocal, GetThreadCount()
+// - synchronization primitives.
+// GTEST_IS_THREADSAFE - defined to 1 to indicate that the above
+// synchronization primitives have real implementations
+// and Google Test is thread-safe; or 0 otherwise.
+//
+// Template meta programming:
+// is_pointer - as in TR1; needed on Symbian and IBM XL C/C++ only.
+// IteratorTraits - partial implementation of std::iterator_traits, which
+// is not available in libCstd when compiled with Sun C++.
+//
+// Smart pointers:
+// scoped_ptr - as in TR2.
+//
+// Regular expressions:
+// RE - a simple regular expression class using the POSIX
+// Extended Regular Expression syntax on UNIX-like
+// platforms, or a reduced regular exception syntax on
+// other platforms, including Windows.
+//
+// Logging:
+// GTEST_LOG_() - logs messages at the specified severity level.
+// LogToStderr() - directs all log messages to stderr.
+// FlushInfoLog() - flushes informational log messages.
+//
+// Stdout and stderr capturing:
+// CaptureStdout() - starts capturing stdout.
+// GetCapturedStdout() - stops capturing stdout and returns the captured
+// string.
+// CaptureStderr() - starts capturing stderr.
+// GetCapturedStderr() - stops capturing stderr and returns the captured
+// string.
+//
+// Integer types:
+// TypeWithSize - maps an integer to a int type.
+// Int32, UInt32, Int64, UInt64, TimeInMillis
+// - integers of known sizes.
+// BiggestInt - the biggest signed integer type.
+//
+// Command-line utilities:
+// GTEST_FLAG() - references a flag.
+// GTEST_DECLARE_*() - declares a flag.
+// GTEST_DEFINE_*() - defines a flag.
+// GetArgvs() - returns the command line as a vector of strings.
+//
+// Environment variable utilities:
+// GetEnv() - gets the value of an environment variable.
+// BoolFromGTestEnv() - parses a bool environment variable.
+// Int32FromGTestEnv() - parses an Int32 environment variable.
+// StringFromGTestEnv() - parses a string environment variable.
+
+#include <ctype.h> // for isspace, etc
+#include <stddef.h> // for ptrdiff_t
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#ifndef _WIN32_WCE
+# include <sys/types.h>
+# include <sys/stat.h>
+#endif // !_WIN32_WCE
+
+#include <iostream> // NOLINT
+#include <sstream> // NOLINT
+#include <string> // NOLINT
+
+#define GTEST_DEV_EMAIL_ "googletestframework@@googlegroups.com"
+#define GTEST_FLAG_PREFIX_ "gtest_"
+#define GTEST_FLAG_PREFIX_DASH_ "gtest-"
+#define GTEST_FLAG_PREFIX_UPPER_ "GTEST_"
+#define GTEST_NAME_ "Google Test"
+#define GTEST_PROJECT_URL_ "http://code.google.com/p/googletest/"
+
+// Determines the version of gcc that is used to compile this.
+#ifdef __GNUC__
+// 40302 means version 4.3.2.
+# define GTEST_GCC_VER_ \
+ (__GNUC__*10000 + __GNUC_MINOR__*100 + __GNUC_PATCHLEVEL__)
+#endif // __GNUC__
+
+// Determines the platform on which Google Test is compiled.
+#ifdef __CYGWIN__
+# define GTEST_OS_CYGWIN 1
+#elif defined __SYMBIAN32__
+# define GTEST_OS_SYMBIAN 1
+#elif defined _WIN32
+# define GTEST_OS_WINDOWS 1
+# ifdef _WIN32_WCE
+# define GTEST_OS_WINDOWS_MOBILE 1
+# elif defined(__MINGW__) || defined(__MINGW32__)
+# define GTEST_OS_WINDOWS_MINGW 1
+# else
+# define GTEST_OS_WINDOWS_DESKTOP 1
+# endif // _WIN32_WCE
+#elif defined __APPLE__
+# define GTEST_OS_MAC 1
+#elif defined __linux__
+# define GTEST_OS_LINUX 1
+# ifdef ANDROID
+# define GTEST_OS_LINUX_ANDROID 1
+# endif // ANDROID
+#elif defined __MVS__
+# define GTEST_OS_ZOS 1
+#elif defined(__sun) && defined(__SVR4)
+# define GTEST_OS_SOLARIS 1
+#elif defined(_AIX)
+# define GTEST_OS_AIX 1
+#elif defined(__hpux)
+# define GTEST_OS_HPUX 1
+#elif defined __native_client__
+# define GTEST_OS_NACL 1
+#endif // __CYGWIN__
+
+// Brings in definitions for functions used in the testing::internal::posix
+// namespace (read, write, close, chdir, isatty, stat). We do not currently
+// use them on Windows Mobile.
+#if !GTEST_OS_WINDOWS
+// This assumes that non-Windows OSes provide unistd.h. For OSes where this
+// is not the case, we need to include headers that provide the functions
+// mentioned above.
+# include <unistd.h>
+# if !GTEST_OS_NACL
+// TODO(vladl@google.com): Remove this condition when Native Client SDK adds
+// strings.h (tracked in
+// http://code.google.com/p/nativeclient/issues/detail?id=1175).
+# include <strings.h> // Native Client doesn't provide strings.h.
+# endif
+#elif !GTEST_OS_WINDOWS_MOBILE
+# include <direct.h>
+# include <io.h>
+#endif
+
+// Defines this to true iff Google Test can use POSIX regular expressions.
+#ifndef GTEST_HAS_POSIX_RE
+# define GTEST_HAS_POSIX_RE (!GTEST_OS_WINDOWS)
+#endif
+
+#if GTEST_HAS_POSIX_RE
+
+// On some platforms, <regex.h> needs someone to define size_t, and
+// won't compile otherwise. We can #include it here as we already
+// included <stdlib.h>, which is guaranteed to define size_t through
+// <stddef.h>.
+# include <regex.h> // NOLINT
+
+# define GTEST_USES_POSIX_RE 1
+
+#elif GTEST_OS_WINDOWS
+
+// <regex.h> is not available on Windows. Use our own simple regex
+// implementation instead.
+# define GTEST_USES_SIMPLE_RE 1
+
+#else
+
+// <regex.h> may not be available on this platform. Use our own
+// simple regex implementation instead.
+# define GTEST_USES_SIMPLE_RE 1
+
+#endif // GTEST_HAS_POSIX_RE
+
+#ifndef GTEST_HAS_EXCEPTIONS
+// The user didn't tell us whether exceptions are enabled, so we need
+// to figure it out.
+# if defined(_MSC_VER) || defined(__BORLANDC__)
+// MSVC's and C++Builder's implementations of the STL use the _HAS_EXCEPTIONS
+// macro to enable exceptions, so we'll do the same.
+// Assumes that exceptions are enabled by default.
+# ifndef _HAS_EXCEPTIONS
+# define _HAS_EXCEPTIONS 1
+# endif // _HAS_EXCEPTIONS
+# define GTEST_HAS_EXCEPTIONS _HAS_EXCEPTIONS
+# elif defined(__GNUC__) && __EXCEPTIONS
+// gcc defines __EXCEPTIONS to 1 iff exceptions are enabled.
+# define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__SUNPRO_CC)
+// Sun Pro CC supports exceptions. However, there is no compile-time way of
+// detecting whether they are enabled or not. Therefore, we assume that
+// they are enabled unless the user tells us otherwise.
+# define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__IBMCPP__) && __EXCEPTIONS
+// xlC defines __EXCEPTIONS to 1 iff exceptions are enabled.
+# define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__HP_aCC)
+// Exception handling is in effect by default in HP aCC compiler. It has to
+// be turned of by +noeh compiler option if desired.
+# define GTEST_HAS_EXCEPTIONS 1
+# else
+// For other compilers, we assume exceptions are disabled to be
+// conservative.
+# define GTEST_HAS_EXCEPTIONS 0
+# endif // defined(_MSC_VER) || defined(__BORLANDC__)
+#endif // GTEST_HAS_EXCEPTIONS
+
+#if !defined(GTEST_HAS_STD_STRING)
+// Even though we don't use this macro any longer, we keep it in case
+// some clients still depend on it.
+# define GTEST_HAS_STD_STRING 1
+#elif !GTEST_HAS_STD_STRING
+// The user told us that ::std::string isn't available.
+# error "Google Test cannot be used where ::std::string isn't available."
+#endif // !defined(GTEST_HAS_STD_STRING)
+
+#ifndef GTEST_HAS_GLOBAL_STRING
+// The user didn't tell us whether ::string is available, so we need
+// to figure it out.
+
+# define GTEST_HAS_GLOBAL_STRING 0
+
+#endif // GTEST_HAS_GLOBAL_STRING
+
+#ifndef GTEST_HAS_STD_WSTRING
+// The user didn't tell us whether ::std::wstring is available, so we need
+// to figure it out.
+// TODO(wan@google.com): uses autoconf to detect whether ::std::wstring
+// is available.
+
+// Cygwin 1.7 and below doesn't support ::std::wstring.
+// Solaris' libc++ doesn't support it either. Android has
+// no support for it at least as recent as Froyo (2.2).
+# define GTEST_HAS_STD_WSTRING \
+ (!(GTEST_OS_LINUX_ANDROID || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS))
+
+#endif // GTEST_HAS_STD_WSTRING
+
+#ifndef GTEST_HAS_GLOBAL_WSTRING
+// The user didn't tell us whether ::wstring is available, so we need
+// to figure it out.
+# define GTEST_HAS_GLOBAL_WSTRING \
+ (GTEST_HAS_STD_WSTRING && GTEST_HAS_GLOBAL_STRING)
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+// Determines whether RTTI is available.
+#ifndef GTEST_HAS_RTTI
+// The user didn't tell us whether RTTI is enabled, so we need to
+// figure it out.
+
+# ifdef _MSC_VER
+
+# ifdef _CPPRTTI // MSVC defines this macro iff RTTI is enabled.
+# define GTEST_HAS_RTTI 1
+# else
+# define GTEST_HAS_RTTI 0
+# endif
+
+// Starting with version 4.3.2, gcc defines __GXX_RTTI iff RTTI is enabled.
+# elif defined(__GNUC__) && (GTEST_GCC_VER_ >= 40302)
+
+# ifdef __GXX_RTTI
+# define GTEST_HAS_RTTI 1
+# else
+# define GTEST_HAS_RTTI 0
+# endif // __GXX_RTTI
+
+// Starting with version 9.0 IBM Visual Age defines __RTTI_ALL__ to 1 if
+// both the typeid and dynamic_cast features are present.
+# elif defined(__IBMCPP__) && (__IBMCPP__ >= 900)
+
+# ifdef __RTTI_ALL__
+# define GTEST_HAS_RTTI 1
+# else
+# define GTEST_HAS_RTTI 0
+# endif
+
+# else
+
+// For all other compilers, we assume RTTI is enabled.
+# define GTEST_HAS_RTTI 1
+
+# endif // _MSC_VER
+
+#endif // GTEST_HAS_RTTI
+
+// It's this header's responsibility to #include <typeinfo> when RTTI
+// is enabled.
+#if GTEST_HAS_RTTI
+# include <typeinfo>
+#endif
+
+// Determines whether Google Test can use the pthreads library.
+#ifndef GTEST_HAS_PTHREAD
+// The user didn't tell us explicitly, so we assume pthreads support is
+// available on Linux and Mac.
+//
+// To disable threading support in Google Test, add -DGTEST_HAS_PTHREAD=0
+// to your compiler flags.
+# define GTEST_HAS_PTHREAD (GTEST_OS_LINUX || GTEST_OS_MAC || GTEST_OS_HPUX)
+#endif // GTEST_HAS_PTHREAD
+
+#if GTEST_HAS_PTHREAD
+// gtest-port.h guarantees to #include <pthread.h> when GTEST_HAS_PTHREAD is
+// true.
+# include <pthread.h> // NOLINT
+
+// For timespec and nanosleep, used below.
+# include <time.h> // NOLINT
+#endif
+
+// Determines whether Google Test can use tr1/tuple. You can define
+// this macro to 0 to prevent Google Test from using tuple (any
+// feature depending on tuple with be disabled in this mode).
+#ifndef GTEST_HAS_TR1_TUPLE
+// The user didn't tell us not to do it, so we assume it's OK.
+# define GTEST_HAS_TR1_TUPLE 1
+#endif // GTEST_HAS_TR1_TUPLE
+
+// Determines whether Google Test's own tr1 tuple implementation
+// should be used.
+#ifndef GTEST_USE_OWN_TR1_TUPLE
+// The user didn't tell us, so we need to figure it out.
+
+// We use our own TR1 tuple if we aren't sure the user has an
+// implementation of it already. At this time, GCC 4.0.0+ and MSVC
+// 2010 are the only mainstream compilers that come with a TR1 tuple
+// implementation. NVIDIA's CUDA NVCC compiler pretends to be GCC by
+// defining __GNUC__ and friends, but cannot compile GCC's tuple
+// implementation. MSVC 2008 (9.0) provides TR1 tuple in a 323 MB
+// Feature Pack download, which we cannot assume the user has.
+# if (defined(__GNUC__) && !defined(__CUDACC__) && (GTEST_GCC_VER_ >= 40000)) \
+ || _MSC_VER >= 1600
+# define GTEST_USE_OWN_TR1_TUPLE 0
+# else
+# define GTEST_USE_OWN_TR1_TUPLE 1
+# endif
+
+#endif // GTEST_USE_OWN_TR1_TUPLE
+
+// To avoid conditional compilation everywhere, we make it
+// gtest-port.h's responsibility to #include the header implementing
+// tr1/tuple.
+#if GTEST_HAS_TR1_TUPLE
+
+# if GTEST_USE_OWN_TR1_TUPLE
+# include "gtest/internal/gtest-tuple.h"
+# elif GTEST_OS_SYMBIAN
+
+// On Symbian, BOOST_HAS_TR1_TUPLE causes Boost's TR1 tuple library to
+// use STLport's tuple implementation, which unfortunately doesn't
+// work as the copy of STLport distributed with Symbian is incomplete.
+// By making sure BOOST_HAS_TR1_TUPLE is undefined, we force Boost to
+// use its own tuple implementation.
+# ifdef BOOST_HAS_TR1_TUPLE
+# undef BOOST_HAS_TR1_TUPLE
+# endif // BOOST_HAS_TR1_TUPLE
+
+// This prevents <boost/tr1/detail/config.hpp>, which defines
+// BOOST_HAS_TR1_TUPLE, from being #included by Boost's <tuple>.
+# define BOOST_TR1_DETAIL_CONFIG_HPP_INCLUDED
+# include <tuple>
+
+# elif defined(__GNUC__) && (GTEST_GCC_VER_ >= 40000)
+// GCC 4.0+ implements tr1/tuple in the <tr1/tuple> header. This does
+// not conform to the TR1 spec, which requires the header to be <tuple>.
+
+# if !GTEST_HAS_RTTI && GTEST_GCC_VER_ < 40302
+// Until version 4.3.2, gcc has a bug that causes <tr1/functional>,
+// which is #included by <tr1/tuple>, to not compile when RTTI is
+// disabled. _TR1_FUNCTIONAL is the header guard for
+// <tr1/functional>. Hence the following #define is a hack to prevent
+// <tr1/functional> from being included.
+# define _TR1_FUNCTIONAL 1
+# include <tr1/tuple>
+# undef _TR1_FUNCTIONAL // Allows the user to #include
+ // <tr1/functional> if he chooses to.
+# else
+# include <tr1/tuple> // NOLINT
+# endif // !GTEST_HAS_RTTI && GTEST_GCC_VER_ < 40302
+
+# else
+// If the compiler is not GCC 4.0+, we assume the user is using a
+// spec-conforming TR1 implementation.
+# include <tuple> // NOLINT
+# endif // GTEST_USE_OWN_TR1_TUPLE
+
+#endif // GTEST_HAS_TR1_TUPLE
+
+// Determines whether clone(2) is supported.
+// Usually it will only be available on Linux, excluding
+// Linux on the Itanium architecture.
+// Also see http://linux.die.net/man/2/clone.
+#ifndef GTEST_HAS_CLONE
+// The user didn't tell us, so we need to figure it out.
+
+# if GTEST_OS_LINUX && !defined(__ia64__)
+# define GTEST_HAS_CLONE 1
+# else
+# define GTEST_HAS_CLONE 0
+# endif // GTEST_OS_LINUX && !defined(__ia64__)
+
+#endif // GTEST_HAS_CLONE
+
+// Determines whether to support stream redirection. This is used to test
+// output correctness and to implement death tests.
+#ifndef GTEST_HAS_STREAM_REDIRECTION
+// By default, we assume that stream redirection is supported on all
+// platforms except known mobile ones.
+# if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN
+# define GTEST_HAS_STREAM_REDIRECTION 0
+# else
+# define GTEST_HAS_STREAM_REDIRECTION 1
+# endif // !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_SYMBIAN
+#endif // GTEST_HAS_STREAM_REDIRECTION
+
+// Determines whether to support death tests.
+// Google Test does not support death tests for VC 7.1 and earlier as
+// abort() in a VC 7.1 application compiled as GUI in debug config
+// pops up a dialog window that cannot be suppressed programmatically.
+#if (GTEST_OS_LINUX || GTEST_OS_MAC || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS || \
+ (GTEST_OS_WINDOWS_DESKTOP && _MSC_VER >= 1400) || \
+ GTEST_OS_WINDOWS_MINGW || GTEST_OS_AIX || GTEST_OS_HPUX)
+# define GTEST_HAS_DEATH_TEST 1
+# include <vector> // NOLINT
+#endif
+
+// We don't support MSVC 7.1 with exceptions disabled now. Therefore
+// all the compilers we care about are adequate for supporting
+// value-parameterized tests.
+#define GTEST_HAS_PARAM_TEST 1
+
+// Determines whether to support type-driven tests.
+
+// Typed tests need <typeinfo> and variadic macros, which GCC, VC++ 8.0,
+// Sun Pro CC, IBM Visual Age, and HP aCC support.
+#if defined(__GNUC__) || (_MSC_VER >= 1400) || defined(__SUNPRO_CC) || \
+ defined(__IBMCPP__) || defined(__HP_aCC)
+# define GTEST_HAS_TYPED_TEST 1
+# define GTEST_HAS_TYPED_TEST_P 1
+#endif
+
+// Determines whether to support Combine(). This only makes sense when
+// value-parameterized tests are enabled. The implementation doesn't
+// work on Sun Studio since it doesn't understand templated conversion
+// operators.
+#if GTEST_HAS_PARAM_TEST && GTEST_HAS_TR1_TUPLE && !defined(__SUNPRO_CC)
+# define GTEST_HAS_COMBINE 1
+#endif
+
+// Determines whether the system compiler uses UTF-16 for encoding wide strings.
+#define GTEST_WIDE_STRING_USES_UTF16_ \
+ (GTEST_OS_WINDOWS || GTEST_OS_CYGWIN || GTEST_OS_SYMBIAN || GTEST_OS_AIX)
+
+// Determines whether test results can be streamed to a socket.
+#if GTEST_OS_LINUX
+# define GTEST_CAN_STREAM_RESULTS_ 1
+#endif
+
+// Defines some utility macros.
+
+// The GNU compiler emits a warning if nested "if" statements are followed by
+// an "else" statement and braces are not used to explicitly disambiguate the
+// "else" binding. This leads to problems with code like:
+//
+// if (gate)
+// ASSERT_*(condition) << "Some message";
+//
+// The "switch (0) case 0:" idiom is used to suppress this.
+#ifdef __INTEL_COMPILER
+# define GTEST_AMBIGUOUS_ELSE_BLOCKER_
+#else
+# define GTEST_AMBIGUOUS_ELSE_BLOCKER_ switch (0) case 0: default: // NOLINT
+#endif
+
+// Use this annotation at the end of a struct/class definition to
+// prevent the compiler from optimizing away instances that are never
+// used. This is useful when all interesting logic happens inside the
+// c'tor and / or d'tor. Example:
+//
+// struct Foo {
+// Foo() { ... }
+// } GTEST_ATTRIBUTE_UNUSED_;
+//
+// Also use it after a variable or parameter declaration to tell the
+// compiler the variable/parameter does not have to be used.
+#if defined(__GNUC__) && !defined(COMPILER_ICC)
+# define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused))
+#else
+# define GTEST_ATTRIBUTE_UNUSED_
+#endif
+
+// A macro to disallow operator=
+// This should be used in the private: declarations for a class.
+#define GTEST_DISALLOW_ASSIGN_(type)\
+ void operator=(type const &)
+
+// A macro to disallow copy constructor and operator=
+// This should be used in the private: declarations for a class.
+#define GTEST_DISALLOW_COPY_AND_ASSIGN_(type)\
+ type(type const &);\
+ GTEST_DISALLOW_ASSIGN_(type)
+
+// Tell the compiler to warn about unused return values for functions declared
+// with this macro. The macro should be used on function declarations
+// following the argument list:
+//
+// Sprocket* AllocateSprocket() GTEST_MUST_USE_RESULT_;
+#if defined(__GNUC__) && (GTEST_GCC_VER_ >= 30400) && !defined(COMPILER_ICC)
+# define GTEST_MUST_USE_RESULT_ __attribute__ ((warn_unused_result))
+#else
+# define GTEST_MUST_USE_RESULT_
+#endif // __GNUC__ && (GTEST_GCC_VER_ >= 30400) && !COMPILER_ICC
+
+// Determine whether the compiler supports Microsoft's Structured Exception
+// Handling. This is supported by several Windows compilers but generally
+// does not exist on any other system.
+#ifndef GTEST_HAS_SEH
+// The user didn't tell us, so we need to figure it out.
+
+# if defined(_MSC_VER) || defined(__BORLANDC__)
+// These two compilers are known to support SEH.
+# define GTEST_HAS_SEH 1
+# else
+// Assume no SEH.
+# define GTEST_HAS_SEH 0
+# endif
+
+#endif // GTEST_HAS_SEH
+
+#ifdef _MSC_VER
+
+# if GTEST_LINKED_AS_SHARED_LIBRARY
+# define GTEST_API_ __declspec(dllimport)
+# elif GTEST_CREATE_SHARED_LIBRARY
+# define GTEST_API_ __declspec(dllexport)
+# endif
+
+#endif // _MSC_VER
+
+#ifndef GTEST_API_
+# define GTEST_API_
+#endif
+
+#ifdef __GNUC__
+// Ask the compiler to never inline a given function.
+# define GTEST_NO_INLINE_ __attribute__((noinline))
+#else
+# define GTEST_NO_INLINE_
+#endif
+
+namespace testing {
+
+class Message;
+
+namespace internal {
+
+class String;
+
+// The GTEST_COMPILE_ASSERT_ macro can be used to verify that a compile time
+// expression is true. For example, you could use it to verify the
+// size of a static array:
+//
+// GTEST_COMPILE_ASSERT_(ARRAYSIZE(content_type_names) == CONTENT_NUM_TYPES,
+// content_type_names_incorrect_size);
+//
+// or to make sure a struct is smaller than a certain size:
+//
+// GTEST_COMPILE_ASSERT_(sizeof(foo) < 128, foo_too_large);
+//
+// The second argument to the macro is the name of the variable. If
+// the expression is false, most compilers will issue a warning/error
+// containing the name of the variable.
+
+template <bool>
+struct CompileAssert {
+};
+
+#define GTEST_COMPILE_ASSERT_(expr, msg) \
+ typedef ::testing::internal::CompileAssert<(bool(expr))> \
+ msg[bool(expr) ? 1 : -1]
+
+// Implementation details of GTEST_COMPILE_ASSERT_:
+//
+// - GTEST_COMPILE_ASSERT_ works by defining an array type that has -1
+// elements (and thus is invalid) when the expression is false.
+//
+// - The simpler definition
+//
+// #define GTEST_COMPILE_ASSERT_(expr, msg) typedef char msg[(expr) ? 1 : -1]
+//
+// does not work, as gcc supports variable-length arrays whose sizes
+// are determined at run-time (this is gcc's extension and not part
+// of the C++ standard). As a result, gcc fails to reject the
+// following code with the simple definition:
+//
+// int foo;
+// GTEST_COMPILE_ASSERT_(foo, msg); // not supposed to compile as foo is
+// // not a compile-time constant.
+//
+// - By using the type CompileAssert<(bool(expr))>, we ensures that
+// expr is a compile-time constant. (Template arguments must be
+// determined at compile-time.)
+//
+// - The outter parentheses in CompileAssert<(bool(expr))> are necessary
+// to work around a bug in gcc 3.4.4 and 4.0.1. If we had written
+//
+// CompileAssert<bool(expr)>
+//
+// instead, these compilers will refuse to compile
+//
+// GTEST_COMPILE_ASSERT_(5 > 0, some_message);
+//
+// (They seem to think the ">" in "5 > 0" marks the end of the
+// template argument list.)
+//
+// - The array size is (bool(expr) ? 1 : -1), instead of simply
+//
+// ((expr) ? 1 : -1).
+//
+// This is to avoid running into a bug in MS VC 7.1, which
+// causes ((0.0) ? 1 : -1) to incorrectly evaluate to 1.
+
+// StaticAssertTypeEqHelper is used by StaticAssertTypeEq defined in gtest.h.
+//
+// This template is declared, but intentionally undefined.
+template <typename T1, typename T2>
+struct StaticAssertTypeEqHelper;
+
+template <typename T>
+struct StaticAssertTypeEqHelper<T, T> {};
+
+#if GTEST_HAS_GLOBAL_STRING
+typedef ::string string;
+#else
+typedef ::std::string string;
+#endif // GTEST_HAS_GLOBAL_STRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+typedef ::wstring wstring;
+#elif GTEST_HAS_STD_WSTRING
+typedef ::std::wstring wstring;
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+// A helper for suppressing warnings on constant condition. It just
+// returns 'condition'.
+GTEST_API_ bool IsTrue(bool condition);
+
+// Defines scoped_ptr.
+
+// This implementation of scoped_ptr is PARTIAL - it only contains
+// enough stuff to satisfy Google Test's need.
+template <typename T>
+class scoped_ptr {
+ public:
+ typedef T element_type;
+
+ explicit scoped_ptr(T* p = NULL) : ptr_(p) {}
+ ~scoped_ptr() { reset(); }
+
+ T& operator*() const { return *ptr_; }
+ T* operator->() const { return ptr_; }
+ T* get() const { return ptr_; }
+
+ T* release() {
+ T* const ptr = ptr_;
+ ptr_ = NULL;
+ return ptr;
+ }
+
+ void reset(T* p = NULL) {
+ if (p != ptr_) {
+ if (IsTrue(sizeof(T) > 0)) { // Makes sure T is a complete type.
+ delete ptr_;
+ }
+ ptr_ = p;
+ }
+ }
+ private:
+ T* ptr_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(scoped_ptr);
+};
+
+// Defines RE.
+
+// A simple C++ wrapper for <regex.h>. It uses the POSIX Extended
+// Regular Expression syntax.
+class GTEST_API_ RE {
+ public:
+ // A copy constructor is required by the Standard to initialize object
+ // references from r-values.
+ RE(const RE& other) { Init(other.pattern()); }
+
+ // Constructs an RE from a string.
+ RE(const ::std::string& regex) { Init(regex.c_str()); } // NOLINT
+
+#if GTEST_HAS_GLOBAL_STRING
+
+ RE(const ::string& regex) { Init(regex.c_str()); } // NOLINT
+
+#endif // GTEST_HAS_GLOBAL_STRING
+
+ RE(const char* regex) { Init(regex); } // NOLINT
+ ~RE();
+
+ // Returns the string representation of the regex.
+ const char* pattern() const { return pattern_; }
+
+ // FullMatch(str, re) returns true iff regular expression re matches
+ // the entire str.
+ // PartialMatch(str, re) returns true iff regular expression re
+ // matches a substring of str (including str itself).
+ //
+ // TODO(wan@google.com): make FullMatch() and PartialMatch() work
+ // when str contains NUL characters.
+ static bool FullMatch(const ::std::string& str, const RE& re) {
+ return FullMatch(str.c_str(), re);
+ }
+ static bool PartialMatch(const ::std::string& str, const RE& re) {
+ return PartialMatch(str.c_str(), re);
+ }
+
+#if GTEST_HAS_GLOBAL_STRING
+
+ static bool FullMatch(const ::string& str, const RE& re) {
+ return FullMatch(str.c_str(), re);
+ }
+ static bool PartialMatch(const ::string& str, const RE& re) {
+ return PartialMatch(str.c_str(), re);
+ }
+
+#endif // GTEST_HAS_GLOBAL_STRING
+
+ static bool FullMatch(const char* str, const RE& re);
+ static bool PartialMatch(const char* str, const RE& re);
+
+ private:
+ void Init(const char* regex);
+
+ // We use a const char* instead of a string, as Google Test may be used
+ // where string is not available. We also do not use Google Test's own
+ // String type here, in order to simplify dependencies between the
+ // files.
+ const char* pattern_;
+ bool is_valid_;
+
+#if GTEST_USES_POSIX_RE
+
+ regex_t full_regex_; // For FullMatch().
+ regex_t partial_regex_; // For PartialMatch().
+
+#else // GTEST_USES_SIMPLE_RE
+
+ const char* full_pattern_; // For FullMatch();
+
+#endif
+
+ GTEST_DISALLOW_ASSIGN_(RE);
+};
+
+// Formats a source file path and a line number as they would appear
+// in an error message from the compiler used to compile this code.
+GTEST_API_ ::std::string FormatFileLocation(const char* file, int line);
+
+// Formats a file location for compiler-independent XML output.
+// Although this function is not platform dependent, we put it next to
+// FormatFileLocation in order to contrast the two functions.
+GTEST_API_ ::std::string FormatCompilerIndependentFileLocation(const char* file,
+ int line);
+
+// Defines logging utilities:
+// GTEST_LOG_(severity) - logs messages at the specified severity level. The
+// message itself is streamed into the macro.
+// LogToStderr() - directs all log messages to stderr.
+// FlushInfoLog() - flushes informational log messages.
+
+enum GTestLogSeverity {
+ GTEST_INFO,
+ GTEST_WARNING,
+ GTEST_ERROR,
+ GTEST_FATAL
+};
+
+// Formats log entry severity, provides a stream object for streaming the
+// log message, and terminates the message with a newline when going out of
+// scope.
+class GTEST_API_ GTestLog {
+ public:
+ GTestLog(GTestLogSeverity severity, const char* file, int line);
+
+ // Flushes the buffers and, if severity is GTEST_FATAL, aborts the program.
+ ~GTestLog();
+
+ ::std::ostream& GetStream() { return ::std::cerr; }
+
+ private:
+ const GTestLogSeverity severity_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestLog);
+};
+
+#define GTEST_LOG_(severity) \
+ ::testing::internal::GTestLog(::testing::internal::GTEST_##severity, \
+ __FILE__, __LINE__).GetStream()
+
+inline void LogToStderr() {}
+inline void FlushInfoLog() { fflush(NULL); }
+
+// INTERNAL IMPLEMENTATION - DO NOT USE.
+//
+// GTEST_CHECK_ is an all-mode assert. It aborts the program if the condition
+// is not satisfied.
+// Synopsys:
+// GTEST_CHECK_(boolean_condition);
+// or
+// GTEST_CHECK_(boolean_condition) << "Additional message";
+//
+// This checks the condition and if the condition is not satisfied
+// it prints message about the condition violation, including the
+// condition itself, plus additional message streamed into it, if any,
+// and then it aborts the program. It aborts the program irrespective of
+// whether it is built in the debug mode or not.
+#define GTEST_CHECK_(condition) \
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+ if (::testing::internal::IsTrue(condition)) \
+ ; \
+ else \
+ GTEST_LOG_(FATAL) << "Condition " #condition " failed. "
+
+// An all-mode assert to verify that the given POSIX-style function
+// call returns 0 (indicating success). Known limitation: this
+// doesn't expand to a balanced 'if' statement, so enclose the macro
+// in {} if you need to use it as the only statement in an 'if'
+// branch.
+#define GTEST_CHECK_POSIX_SUCCESS_(posix_call) \
+ if (const int gtest_error = (posix_call)) \
+ GTEST_LOG_(FATAL) << #posix_call << "failed with error " \
+ << gtest_error
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Use ImplicitCast_ as a safe version of static_cast for upcasting in
+// the type hierarchy (e.g. casting a Foo* to a SuperclassOfFoo* or a
+// const Foo*). When you use ImplicitCast_, the compiler checks that
+// the cast is safe. Such explicit ImplicitCast_s are necessary in
+// surprisingly many situations where C++ demands an exact type match
+// instead of an argument type convertable to a target type.
+//
+// The syntax for using ImplicitCast_ is the same as for static_cast:
+//
+// ImplicitCast_<ToType>(expr)
+//
+// ImplicitCast_ would have been part of the C++ standard library,
+// but the proposal was submitted too late. It will probably make
+// its way into the language in the future.
+//
+// This relatively ugly name is intentional. It prevents clashes with
+// similar functions users may have (e.g., implicit_cast). The internal
+// namespace alone is not enough because the function can be found by ADL.
+template<typename To>
+inline To ImplicitCast_(To x) { return x; }
+
+// When you upcast (that is, cast a pointer from type Foo to type
+// SuperclassOfFoo), it's fine to use ImplicitCast_<>, since upcasts
+// always succeed. When you downcast (that is, cast a pointer from
+// type Foo to type SubclassOfFoo), static_cast<> isn't safe, because
+// how do you know the pointer is really of type SubclassOfFoo? It
+// could be a bare Foo, or of type DifferentSubclassOfFoo. Thus,
+// when you downcast, you should use this macro. In debug mode, we
+// use dynamic_cast<> to double-check the downcast is legal (we die
+// if it's not). In normal mode, we do the efficient static_cast<>
+// instead. Thus, it's important to test in debug mode to make sure
+// the cast is legal!
+// This is the only place in the code we should use dynamic_cast<>.
+// In particular, you SHOULDN'T be using dynamic_cast<> in order to
+// do RTTI (eg code like this:
+// if (dynamic_cast<Subclass1>(foo)) HandleASubclass1Object(foo);
+// if (dynamic_cast<Subclass2>(foo)) HandleASubclass2Object(foo);
+// You should design the code some other way not to need this.
+//
+// This relatively ugly name is intentional. It prevents clashes with
+// similar functions users may have (e.g., down_cast). The internal
+// namespace alone is not enough because the function can be found by ADL.
+template<typename To, typename From> // use like this: DownCast_<T*>(foo);
+inline To DownCast_(From* f) { // so we only accept pointers
+ // Ensures that To is a sub-type of From *. This test is here only
+ // for compile-time type checking, and has no overhead in an
+ // optimized build at run-time, as it will be optimized away
+ // completely.
+ if (false) {
+ const To to = NULL;
+ ::testing::internal::ImplicitCast_<From*>(to);
+ }
+
+#if GTEST_HAS_RTTI
+ // RTTI: debug mode only!
+ GTEST_CHECK_(f == NULL || dynamic_cast<To>(f) != NULL);
+#endif
+ return static_cast<To>(f);
+}
+
+// Downcasts the pointer of type Base to Derived.
+// Derived must be a subclass of Base. The parameter MUST
+// point to a class of type Derived, not any subclass of it.
+// When RTTI is available, the function performs a runtime
+// check to enforce this.
+template <class Derived, class Base>
+Derived* CheckedDowncastToActualType(Base* base) {
+#if GTEST_HAS_RTTI
+ GTEST_CHECK_(typeid(*base) == typeid(Derived));
+ return dynamic_cast<Derived*>(base); // NOLINT
+#else
+ return static_cast<Derived*>(base); // Poor man's downcast.
+#endif
+}
+
+#if GTEST_HAS_STREAM_REDIRECTION
+
+// Defines the stderr capturer:
+// CaptureStdout - starts capturing stdout.
+// GetCapturedStdout - stops capturing stdout and returns the captured string.
+// CaptureStderr - starts capturing stderr.
+// GetCapturedStderr - stops capturing stderr and returns the captured string.
+//
+GTEST_API_ void CaptureStdout();
+GTEST_API_ String GetCapturedStdout();
+GTEST_API_ void CaptureStderr();
+GTEST_API_ String GetCapturedStderr();
+
+#endif // GTEST_HAS_STREAM_REDIRECTION
+
+
+#if GTEST_HAS_DEATH_TEST
+
+// A copy of all command line arguments. Set by InitGoogleTest().
+extern ::std::vector<String> g_argvs;
+
+// GTEST_HAS_DEATH_TEST implies we have ::std::string.
+const ::std::vector<String>& GetArgvs();
+
+#endif // GTEST_HAS_DEATH_TEST
+
+// Defines synchronization primitives.
+
+#if GTEST_HAS_PTHREAD
+
+// Sleeps for (roughly) n milli-seconds. This function is only for
+// testing Google Test's own constructs. Don't use it in user tests,
+// either directly or indirectly.
+inline void SleepMilliseconds(int n) {
+ const timespec time = {
+ 0, // 0 seconds.
+ n * 1000L * 1000L, // And n ms.
+ };
+ nanosleep(&time, NULL);
+}
+
+// Allows a controller thread to pause execution of newly created
+// threads until notified. Instances of this class must be created
+// and destroyed in the controller thread.
+//
+// This class is only for testing Google Test's own constructs. Do not
+// use it in user tests, either directly or indirectly.
+class Notification {
+ public:
+ Notification() : notified_(false) {}
+
+ // Notifies all threads created with this notification to start. Must
+ // be called from the controller thread.
+ void Notify() { notified_ = true; }
+
+ // Blocks until the controller thread notifies. Must be called from a test
+ // thread.
+ void WaitForNotification() {
+ while(!notified_) {
+ SleepMilliseconds(10);
+ }
+ }
+
+ private:
+ volatile bool notified_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification);
+};
+
+// As a C-function, ThreadFuncWithCLinkage cannot be templated itself.
+// Consequently, it cannot select a correct instantiation of ThreadWithParam
+// in order to call its Run(). Introducing ThreadWithParamBase as a
+// non-templated base class for ThreadWithParam allows us to bypass this
+// problem.
+class ThreadWithParamBase {
+ public:
+ virtual ~ThreadWithParamBase() {}
+ virtual void Run() = 0;
+};
+
+// pthread_create() accepts a pointer to a function type with the C linkage.
+// According to the Standard (7.5/1), function types with different linkages
+// are different even if they are otherwise identical. Some compilers (for
+// example, SunStudio) treat them as different types. Since class methods
+// cannot be defined with C-linkage we need to define a free C-function to
+// pass into pthread_create().
+extern "C" inline void* ThreadFuncWithCLinkage(void* thread) {
+ static_cast<ThreadWithParamBase*>(thread)->Run();
+ return NULL;
+}
+
+// Helper class for testing Google Test's multi-threading constructs.
+// To use it, write:
+//
+// void ThreadFunc(int param) { /* Do things with param */ }
+// Notification thread_can_start;
+// ...
+// // The thread_can_start parameter is optional; you can supply NULL.
+// ThreadWithParam<int> thread(&ThreadFunc, 5, &thread_can_start);
+// thread_can_start.Notify();
+//
+// These classes are only for testing Google Test's own constructs. Do
+// not use them in user tests, either directly or indirectly.
+template <typename T>
+class ThreadWithParam : public ThreadWithParamBase {
+ public:
+ typedef void (*UserThreadFunc)(T);
+
+ ThreadWithParam(
+ UserThreadFunc func, T param, Notification* thread_can_start)
+ : func_(func),
+ param_(param),
+ thread_can_start_(thread_can_start),
+ finished_(false) {
+ ThreadWithParamBase* const base = this;
+ // The thread can be created only after all fields except thread_
+ // have been initialized.
+ GTEST_CHECK_POSIX_SUCCESS_(
+ pthread_create(&thread_, 0, &ThreadFuncWithCLinkage, base));
+ }
+ ~ThreadWithParam() { Join(); }
+
+ void Join() {
+ if (!finished_) {
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_join(thread_, 0));
+ finished_ = true;
+ }
+ }
+
+ virtual void Run() {
+ if (thread_can_start_ != NULL)
+ thread_can_start_->WaitForNotification();
+ func_(param_);
+ }
+
+ private:
+ const UserThreadFunc func_; // User-supplied thread function.
+ const T param_; // User-supplied parameter to the thread function.
+ // When non-NULL, used to block execution until the controller thread
+ // notifies.
+ Notification* const thread_can_start_;
+ bool finished_; // true iff we know that the thread function has finished.
+ pthread_t thread_; // The native thread object.
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam);
+};
+
+// MutexBase and Mutex implement mutex on pthreads-based platforms. They
+// are used in conjunction with class MutexLock:
+//
+// Mutex mutex;
+// ...
+// MutexLock lock(&mutex); // Acquires the mutex and releases it at the end
+// // of the current scope.
+//
+// MutexBase implements behavior for both statically and dynamically
+// allocated mutexes. Do not use MutexBase directly. Instead, write
+// the following to define a static mutex:
+//
+// GTEST_DEFINE_STATIC_MUTEX_(g_some_mutex);
+//
+// You can forward declare a static mutex like this:
+//
+// GTEST_DECLARE_STATIC_MUTEX_(g_some_mutex);
+//
+// To create a dynamic mutex, just define an object of type Mutex.
+class MutexBase {
+ public:
+ // Acquires this mutex.
+ void Lock() {
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_lock(&mutex_));
+ owner_ = pthread_self();
+ }
+
+ // Releases this mutex.
+ void Unlock() {
+ // We don't protect writing to owner_ here, as it's the caller's
+ // responsibility to ensure that the current thread holds the
+ // mutex when this is called.
+ owner_ = 0;
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_unlock(&mutex_));
+ }
+
+ // Does nothing if the current thread holds the mutex. Otherwise, crashes
+ // with high probability.
+ void AssertHeld() const {
+ GTEST_CHECK_(owner_ == pthread_self())
+ << "The current thread is not holding the mutex @" << this;
+ }
+
+ // A static mutex may be used before main() is entered. It may even
+ // be used before the dynamic initialization stage. Therefore we
+ // must be able to initialize a static mutex object at link time.
+ // This means MutexBase has to be a POD and its member variables
+ // have to be public.
+ public:
+ pthread_mutex_t mutex_; // The underlying pthread mutex.
+ pthread_t owner_; // The thread holding the mutex; 0 means no one holds it.
+};
+
+// Forward-declares a static mutex.
+# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \
+ extern ::testing::internal::MutexBase mutex
+
+// Defines and statically (i.e. at link time) initializes a static mutex.
+# define GTEST_DEFINE_STATIC_MUTEX_(mutex) \
+ ::testing::internal::MutexBase mutex = { PTHREAD_MUTEX_INITIALIZER, 0 }
+
+// The Mutex class can only be used for mutexes created at runtime. It
+// shares its API with MutexBase otherwise.
+class Mutex : public MutexBase {
+ public:
+ Mutex() {
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, NULL));
+ owner_ = 0;
+ }
+ ~Mutex() {
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_destroy(&mutex_));
+ }
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex);
+};
+
+// We cannot name this class MutexLock as the ctor declaration would
+// conflict with a macro named MutexLock, which is defined on some
+// platforms. Hence the typedef trick below.
+class GTestMutexLock {
+ public:
+ explicit GTestMutexLock(MutexBase* mutex)
+ : mutex_(mutex) { mutex_->Lock(); }
+
+ ~GTestMutexLock() { mutex_->Unlock(); }
+
+ private:
+ MutexBase* const mutex_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock);
+};
+
+typedef GTestMutexLock MutexLock;
+
+// Helpers for ThreadLocal.
+
+// pthread_key_create() requires DeleteThreadLocalValue() to have
+// C-linkage. Therefore it cannot be templatized to access
+// ThreadLocal<T>. Hence the need for class
+// ThreadLocalValueHolderBase.
+class ThreadLocalValueHolderBase {
+ public:
+ virtual ~ThreadLocalValueHolderBase() {}
+};
+
+// Called by pthread to delete thread-local data stored by
+// pthread_setspecific().
+extern "C" inline void DeleteThreadLocalValue(void* value_holder) {
+ delete static_cast<ThreadLocalValueHolderBase*>(value_holder);
+}
+
+// Implements thread-local storage on pthreads-based systems.
+//
+// // Thread 1
+// ThreadLocal<int> tl(100); // 100 is the default value for each thread.
+//
+// // Thread 2
+// tl.set(150); // Changes the value for thread 2 only.
+// EXPECT_EQ(150, tl.get());
+//
+// // Thread 1
+// EXPECT_EQ(100, tl.get()); // In thread 1, tl has the original value.
+// tl.set(200);
+// EXPECT_EQ(200, tl.get());
+//
+// The template type argument T must have a public copy constructor.
+// In addition, the default ThreadLocal constructor requires T to have
+// a public default constructor.
+//
+// An object managed for a thread by a ThreadLocal instance is deleted
+// when the thread exits. Or, if the ThreadLocal instance dies in
+// that thread, when the ThreadLocal dies. It's the user's
+// responsibility to ensure that all other threads using a ThreadLocal
+// have exited when it dies, or the per-thread objects for those
+// threads will not be deleted.
+//
+// Google Test only uses global ThreadLocal objects. That means they
+// will die after main() has returned. Therefore, no per-thread
+// object managed by Google Test will be leaked as long as all threads
+// using Google Test have exited when main() returns.
+template <typename T>
+class ThreadLocal {
+ public:
+ ThreadLocal() : key_(CreateKey()),
+ default_() {}
+ explicit ThreadLocal(const T& value) : key_(CreateKey()),
+ default_(value) {}
+
+ ~ThreadLocal() {
+ // Destroys the managed object for the current thread, if any.
+ DeleteThreadLocalValue(pthread_getspecific(key_));
+
+ // Releases resources associated with the key. This will *not*
+ // delete managed objects for other threads.
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_key_delete(key_));
+ }
+
+ T* pointer() { return GetOrCreateValue(); }
+ const T* pointer() const { return GetOrCreateValue(); }
+ const T& get() const { return *pointer(); }
+ void set(const T& value) { *pointer() = value; }
+
+ private:
+ // Holds a value of type T.
+ class ValueHolder : public ThreadLocalValueHolderBase {
+ public:
+ explicit ValueHolder(const T& value) : value_(value) {}
+
+ T* pointer() { return &value_; }
+
+ private:
+ T value_;
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder);
+ };
+
+ static pthread_key_t CreateKey() {
+ pthread_key_t key;
+ // When a thread exits, DeleteThreadLocalValue() will be called on
+ // the object managed for that thread.
+ GTEST_CHECK_POSIX_SUCCESS_(
+ pthread_key_create(&key, &DeleteThreadLocalValue));
+ return key;
+ }
+
+ T* GetOrCreateValue() const {
+ ThreadLocalValueHolderBase* const holder =
+ static_cast<ThreadLocalValueHolderBase*>(pthread_getspecific(key_));
+ if (holder != NULL) {
+ return CheckedDowncastToActualType<ValueHolder>(holder)->pointer();
+ }
+
+ ValueHolder* const new_holder = new ValueHolder(default_);
+ ThreadLocalValueHolderBase* const holder_base = new_holder;
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_setspecific(key_, holder_base));
+ return new_holder->pointer();
+ }
+
+ // A key pthreads uses for looking up per-thread values.
+ const pthread_key_t key_;
+ const T default_; // The default value for each thread.
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal);
+};
+
+# define GTEST_IS_THREADSAFE 1
+
+#else // GTEST_HAS_PTHREAD
+
+// A dummy implementation of synchronization primitives (mutex, lock,
+// and thread-local variable). Necessary for compiling Google Test where
+// mutex is not supported - using Google Test in multiple threads is not
+// supported on such platforms.
+
+class Mutex {
+ public:
+ Mutex() {}
+ void AssertHeld() const {}
+};
+
+# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \
+ extern ::testing::internal::Mutex mutex
+
+# define GTEST_DEFINE_STATIC_MUTEX_(mutex) ::testing::internal::Mutex mutex
+
+class GTestMutexLock {
+ public:
+ explicit GTestMutexLock(Mutex*) {} // NOLINT
+};
+
+typedef GTestMutexLock MutexLock;
+
+template <typename T>
+class ThreadLocal {
+ public:
+ ThreadLocal() : value_() {}
+ explicit ThreadLocal(const T& value) : value_(value) {}
+ T* pointer() { return &value_; }
+ const T* pointer() const { return &value_; }
+ const T& get() const { return value_; }
+ void set(const T& value) { value_ = value; }
+ private:
+ T value_;
+};
+
+// The above synchronization primitives have dummy implementations.
+// Therefore Google Test is not thread-safe.
+# define GTEST_IS_THREADSAFE 0
+
+#endif // GTEST_HAS_PTHREAD
+
+// Returns the number of threads running in the process, or 0 to indicate that
+// we cannot detect it.
+GTEST_API_ size_t GetThreadCount();
+
+// Passing non-POD classes through ellipsis (...) crashes the ARM
+// compiler and generates a warning in Sun Studio. The Nokia Symbian
+// and the IBM XL C/C++ compiler try to instantiate a copy constructor
+// for objects passed through ellipsis (...), failing for uncopyable
+// objects. We define this to ensure that only POD is passed through
+// ellipsis on these systems.
+#if defined(__SYMBIAN32__) || defined(__IBMCPP__) || defined(__SUNPRO_CC)
+// We lose support for NULL detection where the compiler doesn't like
+// passing non-POD classes through ellipsis (...).
+# define GTEST_ELLIPSIS_NEEDS_POD_ 1
+#else
+# define GTEST_CAN_COMPARE_NULL 1
+#endif
+
+// The Nokia Symbian and IBM XL C/C++ compilers cannot decide between
+// const T& and const T* in a function template. These compilers
+// _can_ decide between class template specializations for T and T*,
+// so a tr1::type_traits-like is_pointer works.
+#if defined(__SYMBIAN32__) || defined(__IBMCPP__)
+# define GTEST_NEEDS_IS_POINTER_ 1
+#endif
+
+template <bool bool_value>
+struct bool_constant {
+ typedef bool_constant<bool_value> type;
+ static const bool value = bool_value;
+};
+template <bool bool_value> const bool bool_constant<bool_value>::value;
+
+typedef bool_constant<false> false_type;
+typedef bool_constant<true> true_type;
+
+template <typename T>
+struct is_pointer : public false_type {};
+
+template <typename T>
+struct is_pointer<T*> : public true_type {};
+
+template <typename Iterator>
+struct IteratorTraits {
+ typedef typename Iterator::value_type value_type;
+};
+
+template <typename T>
+struct IteratorTraits<T*> {
+ typedef T value_type;
+};
+
+template <typename T>
+struct IteratorTraits<const T*> {
+ typedef T value_type;
+};
+
+#if GTEST_OS_WINDOWS
+# define GTEST_PATH_SEP_ "\\"
+# define GTEST_HAS_ALT_PATH_SEP_ 1
+// The biggest signed integer type the compiler supports.
+typedef __int64 BiggestInt;
+#else
+# define GTEST_PATH_SEP_ "/"
+# define GTEST_HAS_ALT_PATH_SEP_ 0
+typedef long long BiggestInt; // NOLINT
+#endif // GTEST_OS_WINDOWS
+
+// Utilities for char.
+
+// isspace(int ch) and friends accept an unsigned char or EOF. char
+// may be signed, depending on the compiler (or compiler flags).
+// Therefore we need to cast a char to unsigned char before calling
+// isspace(), etc.
+
+inline bool IsAlpha(char ch) {
+ return isalpha(static_cast<unsigned char>(ch)) != 0;
+}
+inline bool IsAlNum(char ch) {
+ return isalnum(static_cast<unsigned char>(ch)) != 0;
+}
+inline bool IsDigit(char ch) {
+ return isdigit(static_cast<unsigned char>(ch)) != 0;
+}
+inline bool IsLower(char ch) {
+ return islower(static_cast<unsigned char>(ch)) != 0;
+}
+inline bool IsSpace(char ch) {
+ return isspace(static_cast<unsigned char>(ch)) != 0;
+}
+inline bool IsUpper(char ch) {
+ return isupper(static_cast<unsigned char>(ch)) != 0;
+}
+inline bool IsXDigit(char ch) {
+ return isxdigit(static_cast<unsigned char>(ch)) != 0;
+}
+
+inline char ToLower(char ch) {
+ return static_cast<char>(tolower(static_cast<unsigned char>(ch)));
+}
+inline char ToUpper(char ch) {
+ return static_cast<char>(toupper(static_cast<unsigned char>(ch)));
+}
+
+// The testing::internal::posix namespace holds wrappers for common
+// POSIX functions. These wrappers hide the differences between
+// Windows/MSVC and POSIX systems. Since some compilers define these
+// standard functions as macros, the wrapper cannot have the same name
+// as the wrapped function.
+
+namespace posix {
+
+// Functions with a different name on Windows.
+
+#if GTEST_OS_WINDOWS
+
+typedef struct _stat StatStruct;
+
+# ifdef __BORLANDC__
+inline int IsATTY(int fd) { return isatty(fd); }
+inline int StrCaseCmp(const char* s1, const char* s2) {
+ return stricmp(s1, s2);
+}
+inline char* StrDup(const char* src) { return strdup(src); }
+# else // !__BORLANDC__
+# if GTEST_OS_WINDOWS_MOBILE
+inline int IsATTY(int /* fd */) { return 0; }
+# else
+inline int IsATTY(int fd) { return _isatty(fd); }
+# endif // GTEST_OS_WINDOWS_MOBILE
+inline int StrCaseCmp(const char* s1, const char* s2) {
+ return _stricmp(s1, s2);
+}
+inline char* StrDup(const char* src) { return _strdup(src); }
+# endif // __BORLANDC__
+
+# if GTEST_OS_WINDOWS_MOBILE
+inline int FileNo(FILE* file) { return reinterpret_cast<int>(_fileno(file)); }
+// Stat(), RmDir(), and IsDir() are not needed on Windows CE at this
+// time and thus not defined there.
+# else
+inline int FileNo(FILE* file) { return _fileno(file); }
+inline int Stat(const char* path, StatStruct* buf) { return _stat(path, buf); }
+inline int RmDir(const char* dir) { return _rmdir(dir); }
+inline bool IsDir(const StatStruct& st) {
+ return (_S_IFDIR & st.st_mode) != 0;
+}
+# endif // GTEST_OS_WINDOWS_MOBILE
+
+#else
+
+typedef struct stat StatStruct;
+
+inline int FileNo(FILE* file) { return fileno(file); }
+inline int IsATTY(int fd) { return isatty(fd); }
+inline int Stat(const char* path, StatStruct* buf) { return stat(path, buf); }
+inline int StrCaseCmp(const char* s1, const char* s2) {
+ return strcasecmp(s1, s2);
+}
+inline char* StrDup(const char* src) { return strdup(src); }
+inline int RmDir(const char* dir) { return rmdir(dir); }
+inline bool IsDir(const StatStruct& st) { return S_ISDIR(st.st_mode); }
+
+#endif // GTEST_OS_WINDOWS
+
+// Functions deprecated by MSVC 8.0.
+
+#ifdef _MSC_VER
+// Temporarily disable warning 4996 (deprecated function).
+# pragma warning(push)
+# pragma warning(disable:4996)
+#endif
+
+inline const char* StrNCpy(char* dest, const char* src, size_t n) {
+ return strncpy(dest, src, n);
+}
+
+// ChDir(), FReopen(), FDOpen(), Read(), Write(), Close(), and
+// StrError() aren't needed on Windows CE at this time and thus not
+// defined there.
+
+#if !GTEST_OS_WINDOWS_MOBILE
+inline int ChDir(const char* dir) { return chdir(dir); }
+#endif
+inline FILE* FOpen(const char* path, const char* mode) {
+ return fopen(path, mode);
+}
+#if !GTEST_OS_WINDOWS_MOBILE
+inline FILE *FReopen(const char* path, const char* mode, FILE* stream) {
+ return freopen(path, mode, stream);
+}
+inline FILE* FDOpen(int fd, const char* mode) { return fdopen(fd, mode); }
+#endif
+inline int FClose(FILE* fp) { return fclose(fp); }
+#if !GTEST_OS_WINDOWS_MOBILE
+inline int Read(int fd, void* buf, unsigned int count) {
+ return static_cast<int>(read(fd, buf, count));
+}
+inline int Write(int fd, const void* buf, unsigned int count) {
+ return static_cast<int>(write(fd, buf, count));
+}
+inline int Close(int fd) { return close(fd); }
+inline const char* StrError(int errnum) { return strerror(errnum); }
+#endif
+inline const char* GetEnv(const char* name) {
+#if GTEST_OS_WINDOWS_MOBILE
+ // We are on Windows CE, which has no environment variables.
+ return NULL;
+#elif defined(__BORLANDC__) || defined(__SunOS_5_8) || defined(__SunOS_5_9)
+ // Environment variables which we programmatically clear will be set to the
+ // empty string rather than unset (NULL). Handle that case.
+ const char* const env = getenv(name);
+ return (env != NULL && env[0] != '\0') ? env : NULL;
+#else
+ return getenv(name);
+#endif
+}
+
+#ifdef _MSC_VER
+# pragma warning(pop) // Restores the warning state.
+#endif
+
+#if GTEST_OS_WINDOWS_MOBILE
+// Windows CE has no C library. The abort() function is used in
+// several places in Google Test. This implementation provides a reasonable
+// imitation of standard behaviour.
+void Abort();
+#else
+inline void Abort() { abort(); }
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+} // namespace posix
+
+// The maximum number a BiggestInt can represent. This definition
+// works no matter BiggestInt is represented in one's complement or
+// two's complement.
+//
+// We cannot rely on numeric_limits in STL, as __int64 and long long
+// are not part of standard C++ and numeric_limits doesn't need to be
+// defined for them.
+const BiggestInt kMaxBiggestInt =
+ ~(static_cast<BiggestInt>(1) << (8*sizeof(BiggestInt) - 1));
+
+// This template class serves as a compile-time function from size to
+// type. It maps a size in bytes to a primitive type with that
+// size. e.g.
+//
+// TypeWithSize<4>::UInt
+//
+// is typedef-ed to be unsigned int (unsigned integer made up of 4
+// bytes).
+//
+// Such functionality should belong to STL, but I cannot find it
+// there.
+//
+// Google Test uses this class in the implementation of floating-point
+// comparison.
+//
+// For now it only handles UInt (unsigned int) as that's all Google Test
+// needs. Other types can be easily added in the future if need
+// arises.
+template <size_t size>
+class TypeWithSize {
+ public:
+ // This prevents the user from using TypeWithSize<N> with incorrect
+ // values of N.
+ typedef void UInt;
+};
+
+// The specialization for size 4.
+template <>
+class TypeWithSize<4> {
+ public:
+ // unsigned int has size 4 in both gcc and MSVC.
+ //
+ // As base/basictypes.h doesn't compile on Windows, we cannot use
+ // uint32, uint64, and etc here.
+ typedef int Int;
+ typedef unsigned int UInt;
+};
+
+// The specialization for size 8.
+template <>
+class TypeWithSize<8> {
+ public:
+
+#if GTEST_OS_WINDOWS
+ typedef __int64 Int;
+ typedef unsigned __int64 UInt;
+#else
+ typedef long long Int; // NOLINT
+ typedef unsigned long long UInt; // NOLINT
+#endif // GTEST_OS_WINDOWS
+};
+
+// Integer types of known sizes.
+typedef TypeWithSize<4>::Int Int32;
+typedef TypeWithSize<4>::UInt UInt32;
+typedef TypeWithSize<8>::Int Int64;
+typedef TypeWithSize<8>::UInt UInt64;
+typedef TypeWithSize<8>::Int TimeInMillis; // Represents time in milliseconds.
+
+// Utilities for command line flags and environment variables.
+
+// Macro for referencing flags.
+#define GTEST_FLAG(name) FLAGS_gtest_##name
+
+// Macros for declaring flags.
+#define GTEST_DECLARE_bool_(name) GTEST_API_ extern bool GTEST_FLAG(name)
+#define GTEST_DECLARE_int32_(name) \
+ GTEST_API_ extern ::testing::internal::Int32 GTEST_FLAG(name)
+#define GTEST_DECLARE_string_(name) \
+ GTEST_API_ extern ::testing::internal::String GTEST_FLAG(name)
+
+// Macros for defining flags.
+#define GTEST_DEFINE_bool_(name, default_val, doc) \
+ GTEST_API_ bool GTEST_FLAG(name) = (default_val)
+#define GTEST_DEFINE_int32_(name, default_val, doc) \
+ GTEST_API_ ::testing::internal::Int32 GTEST_FLAG(name) = (default_val)
+#define GTEST_DEFINE_string_(name, default_val, doc) \
+ GTEST_API_ ::testing::internal::String GTEST_FLAG(name) = (default_val)
+
+// Parses 'str' for a 32-bit signed integer. If successful, writes the result
+// to *value and returns true; otherwise leaves *value unchanged and returns
+// false.
+// TODO(chandlerc): Find a better way to refactor flag and environment parsing
+// out of both gtest-port.cc and gtest.cc to avoid exporting this utility
+// function.
+bool ParseInt32(const Message& src_text, const char* str, Int32* value);
+
+// Parses a bool/Int32/string from the environment variable
+// corresponding to the given Google Test flag.
+bool BoolFromGTestEnv(const char* flag, bool default_val);
+GTEST_API_ Int32 Int32FromGTestEnv(const char* flag, Int32 default_val);
+const char* StringFromGTestEnv(const char* flag, const char* default_val);
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-string.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-string.h
new file mode 100644
index 0000000..dc3a07b
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-string.h
@@ -0,0 +1,350 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file declares the String class and functions used internally by
+// Google Test. They are subject to change without notice. They should not used
+// by code external to Google Test.
+//
+// This header file is #included by <gtest/internal/gtest-internal.h>.
+// It should not be #included by other files.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
+
+#ifdef __BORLANDC__
+// string.h is not guaranteed to provide strcpy on C++ Builder.
+# include <mem.h>
+#endif
+
+#include <string.h>
+#include "gtest/internal/gtest-port.h"
+
+#include <string>
+
+namespace testing {
+namespace internal {
+
+// String - a UTF-8 string class.
+//
+// For historic reasons, we don't use std::string.
+//
+// TODO(wan@google.com): replace this class with std::string or
+// implement it in terms of the latter.
+//
+// Note that String can represent both NULL and the empty string,
+// while std::string cannot represent NULL.
+//
+// NULL and the empty string are considered different. NULL is less
+// than anything (including the empty string) except itself.
+//
+// This class only provides minimum functionality necessary for
+// implementing Google Test. We do not intend to implement a full-fledged
+// string class here.
+//
+// Since the purpose of this class is to provide a substitute for
+// std::string on platforms where it cannot be used, we define a copy
+// constructor and assignment operators such that we don't need
+// conditional compilation in a lot of places.
+//
+// In order to make the representation efficient, the d'tor of String
+// is not virtual. Therefore DO NOT INHERIT FROM String.
+class GTEST_API_ String {
+ public:
+ // Static utility methods
+
+ // Returns the input enclosed in double quotes if it's not NULL;
+ // otherwise returns "(null)". For example, "\"Hello\"" is returned
+ // for input "Hello".
+ //
+ // This is useful for printing a C string in the syntax of a literal.
+ //
+ // Known issue: escape sequences are not handled yet.
+ static String ShowCStringQuoted(const char* c_str);
+
+ // Clones a 0-terminated C string, allocating memory using new. The
+ // caller is responsible for deleting the return value using
+ // delete[]. Returns the cloned string, or NULL if the input is
+ // NULL.
+ //
+ // This is different from strdup() in string.h, which allocates
+ // memory using malloc().
+ static const char* CloneCString(const char* c_str);
+
+#if GTEST_OS_WINDOWS_MOBILE
+ // Windows CE does not have the 'ANSI' versions of Win32 APIs. To be
+ // able to pass strings to Win32 APIs on CE we need to convert them
+ // to 'Unicode', UTF-16.
+
+ // Creates a UTF-16 wide string from the given ANSI string, allocating
+ // memory using new. The caller is responsible for deleting the return
+ // value using delete[]. Returns the wide string, or NULL if the
+ // input is NULL.
+ //
+ // The wide string is created using the ANSI codepage (CP_ACP) to
+ // match the behaviour of the ANSI versions of Win32 calls and the
+ // C runtime.
+ static LPCWSTR AnsiToUtf16(const char* c_str);
+
+ // Creates an ANSI string from the given wide string, allocating
+ // memory using new. The caller is responsible for deleting the return
+ // value using delete[]. Returns the ANSI string, or NULL if the
+ // input is NULL.
+ //
+ // The returned string is created using the ANSI codepage (CP_ACP) to
+ // match the behaviour of the ANSI versions of Win32 calls and the
+ // C runtime.
+ static const char* Utf16ToAnsi(LPCWSTR utf16_str);
+#endif
+
+ // Compares two C strings. Returns true iff they have the same content.
+ //
+ // Unlike strcmp(), this function can handle NULL argument(s). A
+ // NULL C string is considered different to any non-NULL C string,
+ // including the empty string.
+ static bool CStringEquals(const char* lhs, const char* rhs);
+
+ // Converts a wide C string to a String using the UTF-8 encoding.
+ // NULL will be converted to "(null)". If an error occurred during
+ // the conversion, "(failed to convert from wide string)" is
+ // returned.
+ static String ShowWideCString(const wchar_t* wide_c_str);
+
+ // Similar to ShowWideCString(), except that this function encloses
+ // the converted string in double quotes.
+ static String ShowWideCStringQuoted(const wchar_t* wide_c_str);
+
+ // Compares two wide C strings. Returns true iff they have the same
+ // content.
+ //
+ // Unlike wcscmp(), this function can handle NULL argument(s). A
+ // NULL C string is considered different to any non-NULL C string,
+ // including the empty string.
+ static bool WideCStringEquals(const wchar_t* lhs, const wchar_t* rhs);
+
+ // Compares two C strings, ignoring case. Returns true iff they
+ // have the same content.
+ //
+ // Unlike strcasecmp(), this function can handle NULL argument(s).
+ // A NULL C string is considered different to any non-NULL C string,
+ // including the empty string.
+ static bool CaseInsensitiveCStringEquals(const char* lhs,
+ const char* rhs);
+
+ // Compares two wide C strings, ignoring case. Returns true iff they
+ // have the same content.
+ //
+ // Unlike wcscasecmp(), this function can handle NULL argument(s).
+ // A NULL C string is considered different to any non-NULL wide C string,
+ // including the empty string.
+ // NB: The implementations on different platforms slightly differ.
+ // On windows, this method uses _wcsicmp which compares according to LC_CTYPE
+ // environment variable. On GNU platform this method uses wcscasecmp
+ // which compares according to LC_CTYPE category of the current locale.
+ // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the
+ // current locale.
+ static bool CaseInsensitiveWideCStringEquals(const wchar_t* lhs,
+ const wchar_t* rhs);
+
+ // Formats a list of arguments to a String, using the same format
+ // spec string as for printf.
+ //
+ // We do not use the StringPrintf class as it is not universally
+ // available.
+ //
+ // The result is limited to 4096 characters (including the tailing
+ // 0). If 4096 characters are not enough to format the input,
+ // "<buffer exceeded>" is returned.
+ static String Format(const char* format, ...);
+
+ // C'tors
+
+ // The default c'tor constructs a NULL string.
+ String() : c_str_(NULL), length_(0) {}
+
+ // Constructs a String by cloning a 0-terminated C string.
+ String(const char* a_c_str) { // NOLINT
+ if (a_c_str == NULL) {
+ c_str_ = NULL;
+ length_ = 0;
+ } else {
+ ConstructNonNull(a_c_str, strlen(a_c_str));
+ }
+ }
+
+ // Constructs a String by copying a given number of chars from a
+ // buffer. E.g. String("hello", 3) creates the string "hel",
+ // String("a\0bcd", 4) creates "a\0bc", String(NULL, 0) creates "",
+ // and String(NULL, 1) results in access violation.
+ String(const char* buffer, size_t a_length) {
+ ConstructNonNull(buffer, a_length);
+ }
+
+ // The copy c'tor creates a new copy of the string. The two
+ // String objects do not share content.
+ String(const String& str) : c_str_(NULL), length_(0) { *this = str; }
+
+ // D'tor. String is intended to be a final class, so the d'tor
+ // doesn't need to be virtual.
+ ~String() { delete[] c_str_; }
+
+ // Allows a String to be implicitly converted to an ::std::string or
+ // ::string, and vice versa. Converting a String containing a NULL
+ // pointer to ::std::string or ::string is undefined behavior.
+ // Converting a ::std::string or ::string containing an embedded NUL
+ // character to a String will result in the prefix up to the first
+ // NUL character.
+ String(const ::std::string& str) {
+ ConstructNonNull(str.c_str(), str.length());
+ }
+
+ operator ::std::string() const { return ::std::string(c_str(), length()); }
+
+#if GTEST_HAS_GLOBAL_STRING
+ String(const ::string& str) {
+ ConstructNonNull(str.c_str(), str.length());
+ }
+
+ operator ::string() const { return ::string(c_str(), length()); }
+#endif // GTEST_HAS_GLOBAL_STRING
+
+ // Returns true iff this is an empty string (i.e. "").
+ bool empty() const { return (c_str() != NULL) && (length() == 0); }
+
+ // Compares this with another String.
+ // Returns < 0 if this is less than rhs, 0 if this is equal to rhs, or > 0
+ // if this is greater than rhs.
+ int Compare(const String& rhs) const;
+
+ // Returns true iff this String equals the given C string. A NULL
+ // string and a non-NULL string are considered not equal.
+ bool operator==(const char* a_c_str) const { return Compare(a_c_str) == 0; }
+
+ // Returns true iff this String is less than the given String. A
+ // NULL string is considered less than "".
+ bool operator<(const String& rhs) const { return Compare(rhs) < 0; }
+
+ // Returns true iff this String doesn't equal the given C string. A NULL
+ // string and a non-NULL string are considered not equal.
+ bool operator!=(const char* a_c_str) const { return !(*this == a_c_str); }
+
+ // Returns true iff this String ends with the given suffix. *Any*
+ // String is considered to end with a NULL or empty suffix.
+ bool EndsWith(const char* suffix) const;
+
+ // Returns true iff this String ends with the given suffix, not considering
+ // case. Any String is considered to end with a NULL or empty suffix.
+ bool EndsWithCaseInsensitive(const char* suffix) const;
+
+ // Returns the length of the encapsulated string, or 0 if the
+ // string is NULL.
+ size_t length() const { return length_; }
+
+ // Gets the 0-terminated C string this String object represents.
+ // The String object still owns the string. Therefore the caller
+ // should NOT delete the return value.
+ const char* c_str() const { return c_str_; }
+
+ // Assigns a C string to this object. Self-assignment works.
+ const String& operator=(const char* a_c_str) {
+ return *this = String(a_c_str);
+ }
+
+ // Assigns a String object to this object. Self-assignment works.
+ const String& operator=(const String& rhs) {
+ if (this != &rhs) {
+ delete[] c_str_;
+ if (rhs.c_str() == NULL) {
+ c_str_ = NULL;
+ length_ = 0;
+ } else {
+ ConstructNonNull(rhs.c_str(), rhs.length());
+ }
+ }
+
+ return *this;
+ }
+
+ private:
+ // Constructs a non-NULL String from the given content. This
+ // function can only be called when c_str_ has not been allocated.
+ // ConstructNonNull(NULL, 0) results in an empty string ("").
+ // ConstructNonNull(NULL, non_zero) is undefined behavior.
+ void ConstructNonNull(const char* buffer, size_t a_length) {
+ char* const str = new char[a_length + 1];
+ memcpy(str, buffer, a_length);
+ str[a_length] = '\0';
+ c_str_ = str;
+ length_ = a_length;
+ }
+
+ const char* c_str_;
+ size_t length_;
+}; // class String
+
+// Streams a String to an ostream. Each '\0' character in the String
+// is replaced with "\\0".
+inline ::std::ostream& operator<<(::std::ostream& os, const String& str) {
+ if (str.c_str() == NULL) {
+ os << "(null)";
+ } else {
+ const char* const c_str = str.c_str();
+ for (size_t i = 0; i != str.length(); i++) {
+ if (c_str[i] == '\0') {
+ os << "\\0";
+ } else {
+ os << c_str[i];
+ }
+ }
+ }
+ return os;
+}
+
+// Gets the content of the stringstream's buffer as a String. Each '\0'
+// character in the buffer is replaced with "\\0".
+GTEST_API_ String StringStreamToString(::std::stringstream* stream);
+
+// Converts a streamable value to a String. A NULL pointer is
+// converted to "(null)". When the input value is a ::string,
+// ::std::string, ::wstring, or ::std::wstring object, each NUL
+// character in it is replaced with "\\0".
+
+// Declared here but defined in gtest.h, so that it has access
+// to the definition of the Message class, required by the ARM
+// compiler.
+template <typename T>
+String StreamableToString(const T& streamable);
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-tuple.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-tuple.h
new file mode 100644
index 0000000..d1af50e
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-tuple.h
@@ -0,0 +1,968 @@
+// This file was GENERATED by a script. DO NOT EDIT BY HAND!!!
+
+// Copyright 2009 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Implements a subset of TR1 tuple needed by Google Test and Google Mock.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+
+#include <utility> // For ::std::pair.
+
+// The compiler used in Symbian has a bug that prevents us from declaring the
+// tuple template as a friend (it complains that tuple is redefined). This
+// hack bypasses the bug by declaring the members that should otherwise be
+// private as public.
+// Sun Studio versions < 12 also have the above bug.
+#if defined(__SYMBIAN32__) || (defined(__SUNPRO_CC) && __SUNPRO_CC < 0x590)
+# define GTEST_DECLARE_TUPLE_AS_FRIEND_ public:
+#else
+# define GTEST_DECLARE_TUPLE_AS_FRIEND_ \
+ template <GTEST_10_TYPENAMES_(U)> friend class tuple; \
+ private:
+#endif
+
+// GTEST_n_TUPLE_(T) is the type of an n-tuple.
+#define GTEST_0_TUPLE_(T) tuple<>
+#define GTEST_1_TUPLE_(T) tuple<T##0, void, void, void, void, void, void, \
+ void, void, void>
+#define GTEST_2_TUPLE_(T) tuple<T##0, T##1, void, void, void, void, void, \
+ void, void, void>
+#define GTEST_3_TUPLE_(T) tuple<T##0, T##1, T##2, void, void, void, void, \
+ void, void, void>
+#define GTEST_4_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, void, void, void, \
+ void, void, void>
+#define GTEST_5_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, void, void, \
+ void, void, void>
+#define GTEST_6_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, void, \
+ void, void, void>
+#define GTEST_7_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+ void, void, void>
+#define GTEST_8_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+ T##7, void, void>
+#define GTEST_9_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+ T##7, T##8, void>
+#define GTEST_10_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+ T##7, T##8, T##9>
+
+// GTEST_n_TYPENAMES_(T) declares a list of n typenames.
+#define GTEST_0_TYPENAMES_(T)
+#define GTEST_1_TYPENAMES_(T) typename T##0
+#define GTEST_2_TYPENAMES_(T) typename T##0, typename T##1
+#define GTEST_3_TYPENAMES_(T) typename T##0, typename T##1, typename T##2
+#define GTEST_4_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+ typename T##3
+#define GTEST_5_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+ typename T##3, typename T##4
+#define GTEST_6_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+ typename T##3, typename T##4, typename T##5
+#define GTEST_7_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+ typename T##3, typename T##4, typename T##5, typename T##6
+#define GTEST_8_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+ typename T##3, typename T##4, typename T##5, typename T##6, typename T##7
+#define GTEST_9_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+ typename T##3, typename T##4, typename T##5, typename T##6, \
+ typename T##7, typename T##8
+#define GTEST_10_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+ typename T##3, typename T##4, typename T##5, typename T##6, \
+ typename T##7, typename T##8, typename T##9
+
+// In theory, defining stuff in the ::std namespace is undefined
+// behavior. We can do this as we are playing the role of a standard
+// library vendor.
+namespace std {
+namespace tr1 {
+
+template <typename T0 = void, typename T1 = void, typename T2 = void,
+ typename T3 = void, typename T4 = void, typename T5 = void,
+ typename T6 = void, typename T7 = void, typename T8 = void,
+ typename T9 = void>
+class tuple;
+
+// Anything in namespace gtest_internal is Google Test's INTERNAL
+// IMPLEMENTATION DETAIL and MUST NOT BE USED DIRECTLY in user code.
+namespace gtest_internal {
+
+// ByRef<T>::type is T if T is a reference; otherwise it's const T&.
+template <typename T>
+struct ByRef { typedef const T& type; }; // NOLINT
+template <typename T>
+struct ByRef<T&> { typedef T& type; }; // NOLINT
+
+// A handy wrapper for ByRef.
+#define GTEST_BY_REF_(T) typename ::std::tr1::gtest_internal::ByRef<T>::type
+
+// AddRef<T>::type is T if T is a reference; otherwise it's T&. This
+// is the same as tr1::add_reference<T>::type.
+template <typename T>
+struct AddRef { typedef T& type; }; // NOLINT
+template <typename T>
+struct AddRef<T&> { typedef T& type; }; // NOLINT
+
+// A handy wrapper for AddRef.
+#define GTEST_ADD_REF_(T) typename ::std::tr1::gtest_internal::AddRef<T>::type
+
+// A helper for implementing get<k>().
+template <int k> class Get;
+
+// A helper for implementing tuple_element<k, T>. kIndexValid is true
+// iff k < the number of fields in tuple type T.
+template <bool kIndexValid, int kIndex, class Tuple>
+struct TupleElement;
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 0, GTEST_10_TUPLE_(T)> { typedef T0 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 1, GTEST_10_TUPLE_(T)> { typedef T1 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 2, GTEST_10_TUPLE_(T)> { typedef T2 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 3, GTEST_10_TUPLE_(T)> { typedef T3 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 4, GTEST_10_TUPLE_(T)> { typedef T4 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 5, GTEST_10_TUPLE_(T)> { typedef T5 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 6, GTEST_10_TUPLE_(T)> { typedef T6 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 7, GTEST_10_TUPLE_(T)> { typedef T7 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 8, GTEST_10_TUPLE_(T)> { typedef T8 type; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct TupleElement<true, 9, GTEST_10_TUPLE_(T)> { typedef T9 type; };
+
+} // namespace gtest_internal
+
+template <>
+class tuple<> {
+ public:
+ tuple() {}
+ tuple(const tuple& /* t */) {}
+ tuple& operator=(const tuple& /* t */) { return *this; }
+};
+
+template <GTEST_1_TYPENAMES_(T)>
+class GTEST_1_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0) : f0_(f0) {}
+
+ tuple(const tuple& t) : f0_(t.f0_) {}
+
+ template <GTEST_1_TYPENAMES_(U)>
+ tuple(const GTEST_1_TUPLE_(U)& t) : f0_(t.f0_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_1_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_1_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_1_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_1_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ return *this;
+ }
+
+ T0 f0_;
+};
+
+template <GTEST_2_TYPENAMES_(T)>
+class GTEST_2_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1) : f0_(f0),
+ f1_(f1) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_) {}
+
+ template <GTEST_2_TYPENAMES_(U)>
+ tuple(const GTEST_2_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_) {}
+ template <typename U0, typename U1>
+ tuple(const ::std::pair<U0, U1>& p) : f0_(p.first), f1_(p.second) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_2_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_2_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+ template <typename U0, typename U1>
+ tuple& operator=(const ::std::pair<U0, U1>& p) {
+ f0_ = p.first;
+ f1_ = p.second;
+ return *this;
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_2_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_2_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+};
+
+template <GTEST_3_TYPENAMES_(T)>
+class GTEST_3_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_(), f2_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+ GTEST_BY_REF_(T2) f2) : f0_(f0), f1_(f1), f2_(f2) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_) {}
+
+ template <GTEST_3_TYPENAMES_(U)>
+ tuple(const GTEST_3_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_3_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_3_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_3_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_3_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ f2_ = t.f2_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+ T2 f2_;
+};
+
+template <GTEST_4_TYPENAMES_(T)>
+class GTEST_4_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_(), f2_(), f3_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+ GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3) : f0_(f0), f1_(f1), f2_(f2),
+ f3_(f3) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_) {}
+
+ template <GTEST_4_TYPENAMES_(U)>
+ tuple(const GTEST_4_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+ f3_(t.f3_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_4_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_4_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_4_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_4_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ f2_ = t.f2_;
+ f3_ = t.f3_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+ T2 f2_;
+ T3 f3_;
+};
+
+template <GTEST_5_TYPENAMES_(T)>
+class GTEST_5_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_(), f2_(), f3_(), f4_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+ GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3,
+ GTEST_BY_REF_(T4) f4) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+ f4_(t.f4_) {}
+
+ template <GTEST_5_TYPENAMES_(U)>
+ tuple(const GTEST_5_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+ f3_(t.f3_), f4_(t.f4_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_5_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_5_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_5_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_5_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ f2_ = t.f2_;
+ f3_ = t.f3_;
+ f4_ = t.f4_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+ T2 f2_;
+ T3 f3_;
+ T4 f4_;
+};
+
+template <GTEST_6_TYPENAMES_(T)>
+class GTEST_6_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+ GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+ GTEST_BY_REF_(T5) f5) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+ f5_(f5) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+ f4_(t.f4_), f5_(t.f5_) {}
+
+ template <GTEST_6_TYPENAMES_(U)>
+ tuple(const GTEST_6_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+ f3_(t.f3_), f4_(t.f4_), f5_(t.f5_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_6_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_6_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_6_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_6_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ f2_ = t.f2_;
+ f3_ = t.f3_;
+ f4_ = t.f4_;
+ f5_ = t.f5_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+ T2 f2_;
+ T3 f3_;
+ T4 f4_;
+ T5 f5_;
+};
+
+template <GTEST_7_TYPENAMES_(T)>
+class GTEST_7_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+ GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+ GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6) : f0_(f0), f1_(f1), f2_(f2),
+ f3_(f3), f4_(f4), f5_(f5), f6_(f6) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+ f4_(t.f4_), f5_(t.f5_), f6_(t.f6_) {}
+
+ template <GTEST_7_TYPENAMES_(U)>
+ tuple(const GTEST_7_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+ f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_7_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_7_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_7_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_7_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ f2_ = t.f2_;
+ f3_ = t.f3_;
+ f4_ = t.f4_;
+ f5_ = t.f5_;
+ f6_ = t.f6_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+ T2 f2_;
+ T3 f3_;
+ T4 f4_;
+ T5 f5_;
+ T6 f6_;
+};
+
+template <GTEST_8_TYPENAMES_(T)>
+class GTEST_8_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+ GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+ GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6,
+ GTEST_BY_REF_(T7) f7) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+ f5_(f5), f6_(f6), f7_(f7) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+ f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_) {}
+
+ template <GTEST_8_TYPENAMES_(U)>
+ tuple(const GTEST_8_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+ f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_8_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_8_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_8_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_8_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ f2_ = t.f2_;
+ f3_ = t.f3_;
+ f4_ = t.f4_;
+ f5_ = t.f5_;
+ f6_ = t.f6_;
+ f7_ = t.f7_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+ T2 f2_;
+ T3 f3_;
+ T4 f4_;
+ T5 f5_;
+ T6 f6_;
+ T7 f7_;
+};
+
+template <GTEST_9_TYPENAMES_(T)>
+class GTEST_9_TUPLE_(T) {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_(), f8_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+ GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+ GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7,
+ GTEST_BY_REF_(T8) f8) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+ f5_(f5), f6_(f6), f7_(f7), f8_(f8) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+ f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_) {}
+
+ template <GTEST_9_TYPENAMES_(U)>
+ tuple(const GTEST_9_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+ f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_9_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_9_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_9_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_9_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ f2_ = t.f2_;
+ f3_ = t.f3_;
+ f4_ = t.f4_;
+ f5_ = t.f5_;
+ f6_ = t.f6_;
+ f7_ = t.f7_;
+ f8_ = t.f8_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+ T2 f2_;
+ T3 f3_;
+ T4 f4_;
+ T5 f5_;
+ T6 f6_;
+ T7 f7_;
+ T8 f8_;
+};
+
+template <GTEST_10_TYPENAMES_(T)>
+class tuple {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_(), f8_(),
+ f9_() {}
+
+ explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+ GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+ GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7,
+ GTEST_BY_REF_(T8) f8, GTEST_BY_REF_(T9) f9) : f0_(f0), f1_(f1), f2_(f2),
+ f3_(f3), f4_(f4), f5_(f5), f6_(f6), f7_(f7), f8_(f8), f9_(f9) {}
+
+ tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+ f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_), f9_(t.f9_) {}
+
+ template <GTEST_10_TYPENAMES_(U)>
+ tuple(const GTEST_10_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+ f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_),
+ f9_(t.f9_) {}
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_10_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_10_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_10_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_10_TUPLE_(U)& t) {
+ f0_ = t.f0_;
+ f1_ = t.f1_;
+ f2_ = t.f2_;
+ f3_ = t.f3_;
+ f4_ = t.f4_;
+ f5_ = t.f5_;
+ f6_ = t.f6_;
+ f7_ = t.f7_;
+ f8_ = t.f8_;
+ f9_ = t.f9_;
+ return *this;
+ }
+
+ T0 f0_;
+ T1 f1_;
+ T2 f2_;
+ T3 f3_;
+ T4 f4_;
+ T5 f5_;
+ T6 f6_;
+ T7 f7_;
+ T8 f8_;
+ T9 f9_;
+};
+
+// 6.1.3.2 Tuple creation functions.
+
+// Known limitations: we don't support passing an
+// std::tr1::reference_wrapper<T> to make_tuple(). And we don't
+// implement tie().
+
+inline tuple<> make_tuple() { return tuple<>(); }
+
+template <GTEST_1_TYPENAMES_(T)>
+inline GTEST_1_TUPLE_(T) make_tuple(const T0& f0) {
+ return GTEST_1_TUPLE_(T)(f0);
+}
+
+template <GTEST_2_TYPENAMES_(T)>
+inline GTEST_2_TUPLE_(T) make_tuple(const T0& f0, const T1& f1) {
+ return GTEST_2_TUPLE_(T)(f0, f1);
+}
+
+template <GTEST_3_TYPENAMES_(T)>
+inline GTEST_3_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2) {
+ return GTEST_3_TUPLE_(T)(f0, f1, f2);
+}
+
+template <GTEST_4_TYPENAMES_(T)>
+inline GTEST_4_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2,
+ const T3& f3) {
+ return GTEST_4_TUPLE_(T)(f0, f1, f2, f3);
+}
+
+template <GTEST_5_TYPENAMES_(T)>
+inline GTEST_5_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2,
+ const T3& f3, const T4& f4) {
+ return GTEST_5_TUPLE_(T)(f0, f1, f2, f3, f4);
+}
+
+template <GTEST_6_TYPENAMES_(T)>
+inline GTEST_6_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2,
+ const T3& f3, const T4& f4, const T5& f5) {
+ return GTEST_6_TUPLE_(T)(f0, f1, f2, f3, f4, f5);
+}
+
+template <GTEST_7_TYPENAMES_(T)>
+inline GTEST_7_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2,
+ const T3& f3, const T4& f4, const T5& f5, const T6& f6) {
+ return GTEST_7_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6);
+}
+
+template <GTEST_8_TYPENAMES_(T)>
+inline GTEST_8_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2,
+ const T3& f3, const T4& f4, const T5& f5, const T6& f6, const T7& f7) {
+ return GTEST_8_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7);
+}
+
+template <GTEST_9_TYPENAMES_(T)>
+inline GTEST_9_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2,
+ const T3& f3, const T4& f4, const T5& f5, const T6& f6, const T7& f7,
+ const T8& f8) {
+ return GTEST_9_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7, f8);
+}
+
+template <GTEST_10_TYPENAMES_(T)>
+inline GTEST_10_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2,
+ const T3& f3, const T4& f4, const T5& f5, const T6& f6, const T7& f7,
+ const T8& f8, const T9& f9) {
+ return GTEST_10_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7, f8, f9);
+}
+
+// 6.1.3.3 Tuple helper classes.
+
+template <typename Tuple> struct tuple_size;
+
+template <GTEST_0_TYPENAMES_(T)>
+struct tuple_size<GTEST_0_TUPLE_(T)> { static const int value = 0; };
+
+template <GTEST_1_TYPENAMES_(T)>
+struct tuple_size<GTEST_1_TUPLE_(T)> { static const int value = 1; };
+
+template <GTEST_2_TYPENAMES_(T)>
+struct tuple_size<GTEST_2_TUPLE_(T)> { static const int value = 2; };
+
+template <GTEST_3_TYPENAMES_(T)>
+struct tuple_size<GTEST_3_TUPLE_(T)> { static const int value = 3; };
+
+template <GTEST_4_TYPENAMES_(T)>
+struct tuple_size<GTEST_4_TUPLE_(T)> { static const int value = 4; };
+
+template <GTEST_5_TYPENAMES_(T)>
+struct tuple_size<GTEST_5_TUPLE_(T)> { static const int value = 5; };
+
+template <GTEST_6_TYPENAMES_(T)>
+struct tuple_size<GTEST_6_TUPLE_(T)> { static const int value = 6; };
+
+template <GTEST_7_TYPENAMES_(T)>
+struct tuple_size<GTEST_7_TUPLE_(T)> { static const int value = 7; };
+
+template <GTEST_8_TYPENAMES_(T)>
+struct tuple_size<GTEST_8_TUPLE_(T)> { static const int value = 8; };
+
+template <GTEST_9_TYPENAMES_(T)>
+struct tuple_size<GTEST_9_TUPLE_(T)> { static const int value = 9; };
+
+template <GTEST_10_TYPENAMES_(T)>
+struct tuple_size<GTEST_10_TUPLE_(T)> { static const int value = 10; };
+
+template <int k, class Tuple>
+struct tuple_element {
+ typedef typename gtest_internal::TupleElement<
+ k < (tuple_size<Tuple>::value), k, Tuple>::type type;
+};
+
+#define GTEST_TUPLE_ELEMENT_(k, Tuple) typename tuple_element<k, Tuple >::type
+
+// 6.1.3.4 Element access.
+
+namespace gtest_internal {
+
+template <>
+class Get<0> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(0, Tuple))
+ Field(Tuple& t) { return t.f0_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(0, Tuple))
+ ConstField(const Tuple& t) { return t.f0_; }
+};
+
+template <>
+class Get<1> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(1, Tuple))
+ Field(Tuple& t) { return t.f1_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(1, Tuple))
+ ConstField(const Tuple& t) { return t.f1_; }
+};
+
+template <>
+class Get<2> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(2, Tuple))
+ Field(Tuple& t) { return t.f2_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(2, Tuple))
+ ConstField(const Tuple& t) { return t.f2_; }
+};
+
+template <>
+class Get<3> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(3, Tuple))
+ Field(Tuple& t) { return t.f3_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(3, Tuple))
+ ConstField(const Tuple& t) { return t.f3_; }
+};
+
+template <>
+class Get<4> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(4, Tuple))
+ Field(Tuple& t) { return t.f4_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(4, Tuple))
+ ConstField(const Tuple& t) { return t.f4_; }
+};
+
+template <>
+class Get<5> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(5, Tuple))
+ Field(Tuple& t) { return t.f5_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(5, Tuple))
+ ConstField(const Tuple& t) { return t.f5_; }
+};
+
+template <>
+class Get<6> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(6, Tuple))
+ Field(Tuple& t) { return t.f6_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(6, Tuple))
+ ConstField(const Tuple& t) { return t.f6_; }
+};
+
+template <>
+class Get<7> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(7, Tuple))
+ Field(Tuple& t) { return t.f7_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(7, Tuple))
+ ConstField(const Tuple& t) { return t.f7_; }
+};
+
+template <>
+class Get<8> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(8, Tuple))
+ Field(Tuple& t) { return t.f8_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(8, Tuple))
+ ConstField(const Tuple& t) { return t.f8_; }
+};
+
+template <>
+class Get<9> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(9, Tuple))
+ Field(Tuple& t) { return t.f9_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(9, Tuple))
+ ConstField(const Tuple& t) { return t.f9_; }
+};
+
+} // namespace gtest_internal
+
+template <int k, GTEST_10_TYPENAMES_(T)>
+GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(k, GTEST_10_TUPLE_(T)))
+get(GTEST_10_TUPLE_(T)& t) {
+ return gtest_internal::Get<k>::Field(t);
+}
+
+template <int k, GTEST_10_TYPENAMES_(T)>
+GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(k, GTEST_10_TUPLE_(T)))
+get(const GTEST_10_TUPLE_(T)& t) {
+ return gtest_internal::Get<k>::ConstField(t);
+}
+
+// 6.1.3.5 Relational operators
+
+// We only implement == and !=, as we don't have a need for the rest yet.
+
+namespace gtest_internal {
+
+// SameSizeTuplePrefixComparator<k, k>::Eq(t1, t2) returns true if the
+// first k fields of t1 equals the first k fields of t2.
+// SameSizeTuplePrefixComparator(k1, k2) would be a compiler error if
+// k1 != k2.
+template <int kSize1, int kSize2>
+struct SameSizeTuplePrefixComparator;
+
+template <>
+struct SameSizeTuplePrefixComparator<0, 0> {
+ template <class Tuple1, class Tuple2>
+ static bool Eq(const Tuple1& /* t1 */, const Tuple2& /* t2 */) {
+ return true;
+ }
+};
+
+template <int k>
+struct SameSizeTuplePrefixComparator<k, k> {
+ template <class Tuple1, class Tuple2>
+ static bool Eq(const Tuple1& t1, const Tuple2& t2) {
+ return SameSizeTuplePrefixComparator<k - 1, k - 1>::Eq(t1, t2) &&
+ ::std::tr1::get<k - 1>(t1) == ::std::tr1::get<k - 1>(t2);
+ }
+};
+
+} // namespace gtest_internal
+
+template <GTEST_10_TYPENAMES_(T), GTEST_10_TYPENAMES_(U)>
+inline bool operator==(const GTEST_10_TUPLE_(T)& t,
+ const GTEST_10_TUPLE_(U)& u) {
+ return gtest_internal::SameSizeTuplePrefixComparator<
+ tuple_size<GTEST_10_TUPLE_(T)>::value,
+ tuple_size<GTEST_10_TUPLE_(U)>::value>::Eq(t, u);
+}
+
+template <GTEST_10_TYPENAMES_(T), GTEST_10_TYPENAMES_(U)>
+inline bool operator!=(const GTEST_10_TUPLE_(T)& t,
+ const GTEST_10_TUPLE_(U)& u) { return !(t == u); }
+
+// 6.1.4 Pairs.
+// Unimplemented.
+
+} // namespace tr1
+} // namespace std
+
+#undef GTEST_0_TUPLE_
+#undef GTEST_1_TUPLE_
+#undef GTEST_2_TUPLE_
+#undef GTEST_3_TUPLE_
+#undef GTEST_4_TUPLE_
+#undef GTEST_5_TUPLE_
+#undef GTEST_6_TUPLE_
+#undef GTEST_7_TUPLE_
+#undef GTEST_8_TUPLE_
+#undef GTEST_9_TUPLE_
+#undef GTEST_10_TUPLE_
+
+#undef GTEST_0_TYPENAMES_
+#undef GTEST_1_TYPENAMES_
+#undef GTEST_2_TYPENAMES_
+#undef GTEST_3_TYPENAMES_
+#undef GTEST_4_TYPENAMES_
+#undef GTEST_5_TYPENAMES_
+#undef GTEST_6_TYPENAMES_
+#undef GTEST_7_TYPENAMES_
+#undef GTEST_8_TYPENAMES_
+#undef GTEST_9_TYPENAMES_
+#undef GTEST_10_TYPENAMES_
+
+#undef GTEST_DECLARE_TUPLE_AS_FRIEND_
+#undef GTEST_BY_REF_
+#undef GTEST_ADD_REF_
+#undef GTEST_TUPLE_ELEMENT_
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-tuple.h.pump b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-tuple.h.pump
new file mode 100644
index 0000000..ef51909
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-tuple.h.pump
@@ -0,0 +1,336 @@
+$$ -*- mode: c++; -*-
+$var n = 10 $$ Maximum number of tuple fields we want to support.
+$$ This meta comment fixes auto-indentation in Emacs. }}
+// Copyright 2009 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Implements a subset of TR1 tuple needed by Google Test and Google Mock.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+
+#include <utility> // For ::std::pair.
+
+// The compiler used in Symbian has a bug that prevents us from declaring the
+// tuple template as a friend (it complains that tuple is redefined). This
+// hack bypasses the bug by declaring the members that should otherwise be
+// private as public.
+// Sun Studio versions < 12 also have the above bug.
+#if defined(__SYMBIAN32__) || (defined(__SUNPRO_CC) && __SUNPRO_CC < 0x590)
+# define GTEST_DECLARE_TUPLE_AS_FRIEND_ public:
+#else
+# define GTEST_DECLARE_TUPLE_AS_FRIEND_ \
+ template <GTEST_$(n)_TYPENAMES_(U)> friend class tuple; \
+ private:
+#endif
+
+
+$range i 0..n-1
+$range j 0..n
+$range k 1..n
+// GTEST_n_TUPLE_(T) is the type of an n-tuple.
+#define GTEST_0_TUPLE_(T) tuple<>
+
+$for k [[
+$range m 0..k-1
+$range m2 k..n-1
+#define GTEST_$(k)_TUPLE_(T) tuple<$for m, [[T##$m]]$for m2 [[, void]]>
+
+]]
+
+// GTEST_n_TYPENAMES_(T) declares a list of n typenames.
+
+$for j [[
+$range m 0..j-1
+#define GTEST_$(j)_TYPENAMES_(T) $for m, [[typename T##$m]]
+
+
+]]
+
+// In theory, defining stuff in the ::std namespace is undefined
+// behavior. We can do this as we are playing the role of a standard
+// library vendor.
+namespace std {
+namespace tr1 {
+
+template <$for i, [[typename T$i = void]]>
+class tuple;
+
+// Anything in namespace gtest_internal is Google Test's INTERNAL
+// IMPLEMENTATION DETAIL and MUST NOT BE USED DIRECTLY in user code.
+namespace gtest_internal {
+
+// ByRef<T>::type is T if T is a reference; otherwise it's const T&.
+template <typename T>
+struct ByRef { typedef const T& type; }; // NOLINT
+template <typename T>
+struct ByRef<T&> { typedef T& type; }; // NOLINT
+
+// A handy wrapper for ByRef.
+#define GTEST_BY_REF_(T) typename ::std::tr1::gtest_internal::ByRef<T>::type
+
+// AddRef<T>::type is T if T is a reference; otherwise it's T&. This
+// is the same as tr1::add_reference<T>::type.
+template <typename T>
+struct AddRef { typedef T& type; }; // NOLINT
+template <typename T>
+struct AddRef<T&> { typedef T& type; }; // NOLINT
+
+// A handy wrapper for AddRef.
+#define GTEST_ADD_REF_(T) typename ::std::tr1::gtest_internal::AddRef<T>::type
+
+// A helper for implementing get<k>().
+template <int k> class Get;
+
+// A helper for implementing tuple_element<k, T>. kIndexValid is true
+// iff k < the number of fields in tuple type T.
+template <bool kIndexValid, int kIndex, class Tuple>
+struct TupleElement;
+
+
+$for i [[
+template <GTEST_$(n)_TYPENAMES_(T)>
+struct TupleElement<true, $i, GTEST_$(n)_TUPLE_(T)> [[]]
+{ typedef T$i type; };
+
+
+]]
+} // namespace gtest_internal
+
+template <>
+class tuple<> {
+ public:
+ tuple() {}
+ tuple(const tuple& /* t */) {}
+ tuple& operator=(const tuple& /* t */) { return *this; }
+};
+
+
+$for k [[
+$range m 0..k-1
+template <GTEST_$(k)_TYPENAMES_(T)>
+class $if k < n [[GTEST_$(k)_TUPLE_(T)]] $else [[tuple]] {
+ public:
+ template <int k> friend class gtest_internal::Get;
+
+ tuple() : $for m, [[f$(m)_()]] {}
+
+ explicit tuple($for m, [[GTEST_BY_REF_(T$m) f$m]]) : [[]]
+$for m, [[f$(m)_(f$m)]] {}
+
+ tuple(const tuple& t) : $for m, [[f$(m)_(t.f$(m)_)]] {}
+
+ template <GTEST_$(k)_TYPENAMES_(U)>
+ tuple(const GTEST_$(k)_TUPLE_(U)& t) : $for m, [[f$(m)_(t.f$(m)_)]] {}
+
+$if k == 2 [[
+ template <typename U0, typename U1>
+ tuple(const ::std::pair<U0, U1>& p) : f0_(p.first), f1_(p.second) {}
+
+]]
+
+ tuple& operator=(const tuple& t) { return CopyFrom(t); }
+
+ template <GTEST_$(k)_TYPENAMES_(U)>
+ tuple& operator=(const GTEST_$(k)_TUPLE_(U)& t) {
+ return CopyFrom(t);
+ }
+
+$if k == 2 [[
+ template <typename U0, typename U1>
+ tuple& operator=(const ::std::pair<U0, U1>& p) {
+ f0_ = p.first;
+ f1_ = p.second;
+ return *this;
+ }
+
+]]
+
+ GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+ template <GTEST_$(k)_TYPENAMES_(U)>
+ tuple& CopyFrom(const GTEST_$(k)_TUPLE_(U)& t) {
+
+$for m [[
+ f$(m)_ = t.f$(m)_;
+
+]]
+ return *this;
+ }
+
+
+$for m [[
+ T$m f$(m)_;
+
+]]
+};
+
+
+]]
+// 6.1.3.2 Tuple creation functions.
+
+// Known limitations: we don't support passing an
+// std::tr1::reference_wrapper<T> to make_tuple(). And we don't
+// implement tie().
+
+inline tuple<> make_tuple() { return tuple<>(); }
+
+$for k [[
+$range m 0..k-1
+
+template <GTEST_$(k)_TYPENAMES_(T)>
+inline GTEST_$(k)_TUPLE_(T) make_tuple($for m, [[const T$m& f$m]]) {
+ return GTEST_$(k)_TUPLE_(T)($for m, [[f$m]]);
+}
+
+]]
+
+// 6.1.3.3 Tuple helper classes.
+
+template <typename Tuple> struct tuple_size;
+
+
+$for j [[
+template <GTEST_$(j)_TYPENAMES_(T)>
+struct tuple_size<GTEST_$(j)_TUPLE_(T)> { static const int value = $j; };
+
+
+]]
+template <int k, class Tuple>
+struct tuple_element {
+ typedef typename gtest_internal::TupleElement<
+ k < (tuple_size<Tuple>::value), k, Tuple>::type type;
+};
+
+#define GTEST_TUPLE_ELEMENT_(k, Tuple) typename tuple_element<k, Tuple >::type
+
+// 6.1.3.4 Element access.
+
+namespace gtest_internal {
+
+
+$for i [[
+template <>
+class Get<$i> {
+ public:
+ template <class Tuple>
+ static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_($i, Tuple))
+ Field(Tuple& t) { return t.f$(i)_; } // NOLINT
+
+ template <class Tuple>
+ static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_($i, Tuple))
+ ConstField(const Tuple& t) { return t.f$(i)_; }
+};
+
+
+]]
+} // namespace gtest_internal
+
+template <int k, GTEST_$(n)_TYPENAMES_(T)>
+GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(k, GTEST_$(n)_TUPLE_(T)))
+get(GTEST_$(n)_TUPLE_(T)& t) {
+ return gtest_internal::Get<k>::Field(t);
+}
+
+template <int k, GTEST_$(n)_TYPENAMES_(T)>
+GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(k, GTEST_$(n)_TUPLE_(T)))
+get(const GTEST_$(n)_TUPLE_(T)& t) {
+ return gtest_internal::Get<k>::ConstField(t);
+}
+
+// 6.1.3.5 Relational operators
+
+// We only implement == and !=, as we don't have a need for the rest yet.
+
+namespace gtest_internal {
+
+// SameSizeTuplePrefixComparator<k, k>::Eq(t1, t2) returns true if the
+// first k fields of t1 equals the first k fields of t2.
+// SameSizeTuplePrefixComparator(k1, k2) would be a compiler error if
+// k1 != k2.
+template <int kSize1, int kSize2>
+struct SameSizeTuplePrefixComparator;
+
+template <>
+struct SameSizeTuplePrefixComparator<0, 0> {
+ template <class Tuple1, class Tuple2>
+ static bool Eq(const Tuple1& /* t1 */, const Tuple2& /* t2 */) {
+ return true;
+ }
+};
+
+template <int k>
+struct SameSizeTuplePrefixComparator<k, k> {
+ template <class Tuple1, class Tuple2>
+ static bool Eq(const Tuple1& t1, const Tuple2& t2) {
+ return SameSizeTuplePrefixComparator<k - 1, k - 1>::Eq(t1, t2) &&
+ ::std::tr1::get<k - 1>(t1) == ::std::tr1::get<k - 1>(t2);
+ }
+};
+
+} // namespace gtest_internal
+
+template <GTEST_$(n)_TYPENAMES_(T), GTEST_$(n)_TYPENAMES_(U)>
+inline bool operator==(const GTEST_$(n)_TUPLE_(T)& t,
+ const GTEST_$(n)_TUPLE_(U)& u) {
+ return gtest_internal::SameSizeTuplePrefixComparator<
+ tuple_size<GTEST_$(n)_TUPLE_(T)>::value,
+ tuple_size<GTEST_$(n)_TUPLE_(U)>::value>::Eq(t, u);
+}
+
+template <GTEST_$(n)_TYPENAMES_(T), GTEST_$(n)_TYPENAMES_(U)>
+inline bool operator!=(const GTEST_$(n)_TUPLE_(T)& t,
+ const GTEST_$(n)_TUPLE_(U)& u) { return !(t == u); }
+
+// 6.1.4 Pairs.
+// Unimplemented.
+
+} // namespace tr1
+} // namespace std
+
+
+$for j [[
+#undef GTEST_$(j)_TUPLE_
+
+]]
+
+
+$for j [[
+#undef GTEST_$(j)_TYPENAMES_
+
+]]
+
+#undef GTEST_DECLARE_TUPLE_AS_FRIEND_
+#undef GTEST_BY_REF_
+#undef GTEST_ADD_REF_
+#undef GTEST_TUPLE_ELEMENT_
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-type-util.h b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-type-util.h
new file mode 100644
index 0000000..b7b01b0
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-type-util.h
@@ -0,0 +1,3330 @@
+// This file was GENERATED by command:
+// pump.py gtest-type-util.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Type utilities needed for implementing typed and type-parameterized
+// tests. This file is generated by a SCRIPT. DO NOT EDIT BY HAND!
+//
+// Currently we support at most 50 types in a list, and at most 50
+// type-parameterized tests in one type-parameterized test case.
+// Please contact googletestframework@googlegroups.com if you need
+// more.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+
+#include "gtest/internal/gtest-port.h"
+#include "gtest/internal/gtest-string.h"
+
+// #ifdef __GNUC__ is too general here. It is possible to use gcc without using
+// libstdc++ (which is where cxxabi.h comes from).
+# ifdef __GLIBCXX__
+# include <cxxabi.h>
+# elif defined(__HP_aCC)
+# include <acxx_demangle.h>
+# endif // __GLIBCXX__
+
+namespace testing {
+namespace internal {
+
+// GetTypeName<T>() returns a human-readable name of type T.
+// NB: This function is also used in Google Mock, so don't move it inside of
+// the typed-test-only section below.
+template <typename T>
+String GetTypeName() {
+# if GTEST_HAS_RTTI
+
+ const char* const name = typeid(T).name();
+# if defined(__GLIBCXX__) || defined(__HP_aCC)
+ int status = 0;
+ // gcc's implementation of typeid(T).name() mangles the type name,
+ // so we have to demangle it.
+# ifdef __GLIBCXX__
+ using abi::__cxa_demangle;
+# endif // __GLIBCXX__
+ char* const readable_name = __cxa_demangle(name, 0, 0, &status);
+ const String name_str(status == 0 ? readable_name : name);
+ free(readable_name);
+ return name_str;
+# else
+ return name;
+# endif // __GLIBCXX__ || __HP_aCC
+
+# else
+
+ return "<type>";
+
+# endif // GTEST_HAS_RTTI
+}
+
+#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// AssertyTypeEq<T1, T2>::type is defined iff T1 and T2 are the same
+// type. This can be used as a compile-time assertion to ensure that
+// two types are equal.
+
+template <typename T1, typename T2>
+struct AssertTypeEq;
+
+template <typename T>
+struct AssertTypeEq<T, T> {
+ typedef bool type;
+};
+
+// A unique type used as the default value for the arguments of class
+// template Types. This allows us to simulate variadic templates
+// (e.g. Types<int>, Type<int, double>, and etc), which C++ doesn't
+// support directly.
+struct None {};
+
+// The following family of struct and struct templates are used to
+// represent type lists. In particular, TypesN<T1, T2, ..., TN>
+// represents a type list with N types (T1, T2, ..., and TN) in it.
+// Except for Types0, every struct in the family has two member types:
+// Head for the first type in the list, and Tail for the rest of the
+// list.
+
+// The empty type list.
+struct Types0 {};
+
+// Type lists of length 1, 2, 3, and so on.
+
+template <typename T1>
+struct Types1 {
+ typedef T1 Head;
+ typedef Types0 Tail;
+};
+template <typename T1, typename T2>
+struct Types2 {
+ typedef T1 Head;
+ typedef Types1<T2> Tail;
+};
+
+template <typename T1, typename T2, typename T3>
+struct Types3 {
+ typedef T1 Head;
+ typedef Types2<T2, T3> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4>
+struct Types4 {
+ typedef T1 Head;
+ typedef Types3<T2, T3, T4> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5>
+struct Types5 {
+ typedef T1 Head;
+ typedef Types4<T2, T3, T4, T5> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6>
+struct Types6 {
+ typedef T1 Head;
+ typedef Types5<T2, T3, T4, T5, T6> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7>
+struct Types7 {
+ typedef T1 Head;
+ typedef Types6<T2, T3, T4, T5, T6, T7> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8>
+struct Types8 {
+ typedef T1 Head;
+ typedef Types7<T2, T3, T4, T5, T6, T7, T8> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9>
+struct Types9 {
+ typedef T1 Head;
+ typedef Types8<T2, T3, T4, T5, T6, T7, T8, T9> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10>
+struct Types10 {
+ typedef T1 Head;
+ typedef Types9<T2, T3, T4, T5, T6, T7, T8, T9, T10> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11>
+struct Types11 {
+ typedef T1 Head;
+ typedef Types10<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12>
+struct Types12 {
+ typedef T1 Head;
+ typedef Types11<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13>
+struct Types13 {
+ typedef T1 Head;
+ typedef Types12<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14>
+struct Types14 {
+ typedef T1 Head;
+ typedef Types13<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15>
+struct Types15 {
+ typedef T1 Head;
+ typedef Types14<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16>
+struct Types16 {
+ typedef T1 Head;
+ typedef Types15<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17>
+struct Types17 {
+ typedef T1 Head;
+ typedef Types16<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18>
+struct Types18 {
+ typedef T1 Head;
+ typedef Types17<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19>
+struct Types19 {
+ typedef T1 Head;
+ typedef Types18<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20>
+struct Types20 {
+ typedef T1 Head;
+ typedef Types19<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21>
+struct Types21 {
+ typedef T1 Head;
+ typedef Types20<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22>
+struct Types22 {
+ typedef T1 Head;
+ typedef Types21<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23>
+struct Types23 {
+ typedef T1 Head;
+ typedef Types22<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24>
+struct Types24 {
+ typedef T1 Head;
+ typedef Types23<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25>
+struct Types25 {
+ typedef T1 Head;
+ typedef Types24<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26>
+struct Types26 {
+ typedef T1 Head;
+ typedef Types25<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27>
+struct Types27 {
+ typedef T1 Head;
+ typedef Types26<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28>
+struct Types28 {
+ typedef T1 Head;
+ typedef Types27<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29>
+struct Types29 {
+ typedef T1 Head;
+ typedef Types28<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30>
+struct Types30 {
+ typedef T1 Head;
+ typedef Types29<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31>
+struct Types31 {
+ typedef T1 Head;
+ typedef Types30<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32>
+struct Types32 {
+ typedef T1 Head;
+ typedef Types31<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33>
+struct Types33 {
+ typedef T1 Head;
+ typedef Types32<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34>
+struct Types34 {
+ typedef T1 Head;
+ typedef Types33<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35>
+struct Types35 {
+ typedef T1 Head;
+ typedef Types34<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36>
+struct Types36 {
+ typedef T1 Head;
+ typedef Types35<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37>
+struct Types37 {
+ typedef T1 Head;
+ typedef Types36<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38>
+struct Types38 {
+ typedef T1 Head;
+ typedef Types37<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39>
+struct Types39 {
+ typedef T1 Head;
+ typedef Types38<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40>
+struct Types40 {
+ typedef T1 Head;
+ typedef Types39<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41>
+struct Types41 {
+ typedef T1 Head;
+ typedef Types40<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42>
+struct Types42 {
+ typedef T1 Head;
+ typedef Types41<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43>
+struct Types43 {
+ typedef T1 Head;
+ typedef Types42<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44>
+struct Types44 {
+ typedef T1 Head;
+ typedef Types43<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45>
+struct Types45 {
+ typedef T1 Head;
+ typedef Types44<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46>
+struct Types46 {
+ typedef T1 Head;
+ typedef Types45<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47>
+struct Types47 {
+ typedef T1 Head;
+ typedef Types46<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48>
+struct Types48 {
+ typedef T1 Head;
+ typedef Types47<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47, T48> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48, typename T49>
+struct Types49 {
+ typedef T1 Head;
+ typedef Types48<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47, T48, T49> Tail;
+};
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48, typename T49, typename T50>
+struct Types50 {
+ typedef T1 Head;
+ typedef Types49<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47, T48, T49, T50> Tail;
+};
+
+
+} // namespace internal
+
+// We don't want to require the users to write TypesN<...> directly,
+// as that would require them to count the length. Types<...> is much
+// easier to write, but generates horrible messages when there is a
+// compiler error, as gcc insists on printing out each template
+// argument, even if it has the default value (this means Types<int>
+// will appear as Types<int, None, None, ..., None> in the compiler
+// errors).
+//
+// Our solution is to combine the best part of the two approaches: a
+// user would write Types<T1, ..., TN>, and Google Test will translate
+// that to TypesN<T1, ..., TN> internally to make error messages
+// readable. The translation is done by the 'type' member of the
+// Types template.
+template <typename T1 = internal::None, typename T2 = internal::None,
+ typename T3 = internal::None, typename T4 = internal::None,
+ typename T5 = internal::None, typename T6 = internal::None,
+ typename T7 = internal::None, typename T8 = internal::None,
+ typename T9 = internal::None, typename T10 = internal::None,
+ typename T11 = internal::None, typename T12 = internal::None,
+ typename T13 = internal::None, typename T14 = internal::None,
+ typename T15 = internal::None, typename T16 = internal::None,
+ typename T17 = internal::None, typename T18 = internal::None,
+ typename T19 = internal::None, typename T20 = internal::None,
+ typename T21 = internal::None, typename T22 = internal::None,
+ typename T23 = internal::None, typename T24 = internal::None,
+ typename T25 = internal::None, typename T26 = internal::None,
+ typename T27 = internal::None, typename T28 = internal::None,
+ typename T29 = internal::None, typename T30 = internal::None,
+ typename T31 = internal::None, typename T32 = internal::None,
+ typename T33 = internal::None, typename T34 = internal::None,
+ typename T35 = internal::None, typename T36 = internal::None,
+ typename T37 = internal::None, typename T38 = internal::None,
+ typename T39 = internal::None, typename T40 = internal::None,
+ typename T41 = internal::None, typename T42 = internal::None,
+ typename T43 = internal::None, typename T44 = internal::None,
+ typename T45 = internal::None, typename T46 = internal::None,
+ typename T47 = internal::None, typename T48 = internal::None,
+ typename T49 = internal::None, typename T50 = internal::None>
+struct Types {
+ typedef internal::Types50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43, T44, T45, T46, T47, T48, T49, T50> type;
+};
+
+template <>
+struct Types<internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types0 type;
+};
+template <typename T1>
+struct Types<T1, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types1<T1> type;
+};
+template <typename T1, typename T2>
+struct Types<T1, T2, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types2<T1, T2> type;
+};
+template <typename T1, typename T2, typename T3>
+struct Types<T1, T2, T3, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types3<T1, T2, T3> type;
+};
+template <typename T1, typename T2, typename T3, typename T4>
+struct Types<T1, T2, T3, T4, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types4<T1, T2, T3, T4> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5>
+struct Types<T1, T2, T3, T4, T5, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types5<T1, T2, T3, T4, T5> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6>
+struct Types<T1, T2, T3, T4, T5, T6, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types6<T1, T2, T3, T4, T5, T6> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7>
+struct Types<T1, T2, T3, T4, T5, T6, T7, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types7<T1, T2, T3, T4, T5, T6, T7> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types8<T1, T2, T3, T4, T5, T6, T7, T8> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types9<T1, T2, T3, T4, T5, T6, T7, T8, T9> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+ T12> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+ T26> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+ T40> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, internal::None,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None, internal::None> {
+ typedef internal::Types43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None, internal::None> {
+ typedef internal::Types44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43, T44> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+ internal::None, internal::None, internal::None, internal::None,
+ internal::None> {
+ typedef internal::Types45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43, T44, T45> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+ T46, internal::None, internal::None, internal::None, internal::None> {
+ typedef internal::Types46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43, T44, T45, T46> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+ T46, T47, internal::None, internal::None, internal::None> {
+ typedef internal::Types47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43, T44, T45, T46, T47> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+ T46, T47, T48, internal::None, internal::None> {
+ typedef internal::Types48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43, T44, T45, T46, T47, T48> type;
+};
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48, typename T49>
+struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+ T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+ T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+ T46, T47, T48, T49, internal::None> {
+ typedef internal::Types49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43, T44, T45, T46, T47, T48, T49> type;
+};
+
+namespace internal {
+
+# define GTEST_TEMPLATE_ template <typename T> class
+
+// The template "selector" struct TemplateSel<Tmpl> is used to
+// represent Tmpl, which must be a class template with one type
+// parameter, as a type. TemplateSel<Tmpl>::Bind<T>::type is defined
+// as the type Tmpl<T>. This allows us to actually instantiate the
+// template "selected" by TemplateSel<Tmpl>.
+//
+// This trick is necessary for simulating typedef for class templates,
+// which C++ doesn't support directly.
+template <GTEST_TEMPLATE_ Tmpl>
+struct TemplateSel {
+ template <typename T>
+ struct Bind {
+ typedef Tmpl<T> type;
+ };
+};
+
+# define GTEST_BIND_(TmplSel, T) \
+ TmplSel::template Bind<T>::type
+
+// A unique struct template used as the default value for the
+// arguments of class template Templates. This allows us to simulate
+// variadic templates (e.g. Templates<int>, Templates<int, double>,
+// and etc), which C++ doesn't support directly.
+template <typename T>
+struct NoneT {};
+
+// The following family of struct and struct templates are used to
+// represent template lists. In particular, TemplatesN<T1, T2, ...,
+// TN> represents a list of N templates (T1, T2, ..., and TN). Except
+// for Templates0, every struct in the family has two member types:
+// Head for the selector of the first template in the list, and Tail
+// for the rest of the list.
+
+// The empty template list.
+struct Templates0 {};
+
+// Template lists of length 1, 2, 3, and so on.
+
+template <GTEST_TEMPLATE_ T1>
+struct Templates1 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates0 Tail;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2>
+struct Templates2 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates1<T2> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3>
+struct Templates3 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates2<T2, T3> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4>
+struct Templates4 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates3<T2, T3, T4> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5>
+struct Templates5 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates4<T2, T3, T4, T5> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6>
+struct Templates6 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates5<T2, T3, T4, T5, T6> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7>
+struct Templates7 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates6<T2, T3, T4, T5, T6, T7> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8>
+struct Templates8 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates7<T2, T3, T4, T5, T6, T7, T8> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9>
+struct Templates9 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates8<T2, T3, T4, T5, T6, T7, T8, T9> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10>
+struct Templates10 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates9<T2, T3, T4, T5, T6, T7, T8, T9, T10> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11>
+struct Templates11 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates10<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12>
+struct Templates12 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates11<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13>
+struct Templates13 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates12<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14>
+struct Templates14 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates13<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15>
+struct Templates15 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates14<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16>
+struct Templates16 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates15<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17>
+struct Templates17 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates16<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18>
+struct Templates18 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates17<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19>
+struct Templates19 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates18<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20>
+struct Templates20 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates19<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21>
+struct Templates21 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates20<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22>
+struct Templates22 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates21<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23>
+struct Templates23 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates22<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24>
+struct Templates24 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates23<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25>
+struct Templates25 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates24<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26>
+struct Templates26 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates25<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27>
+struct Templates27 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates26<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28>
+struct Templates28 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates27<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29>
+struct Templates29 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates28<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30>
+struct Templates30 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates29<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31>
+struct Templates31 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates30<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32>
+struct Templates32 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates31<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33>
+struct Templates33 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates32<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34>
+struct Templates34 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates33<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35>
+struct Templates35 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates34<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36>
+struct Templates36 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates35<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37>
+struct Templates37 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates36<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38>
+struct Templates38 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates37<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39>
+struct Templates39 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates38<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40>
+struct Templates40 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates39<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41>
+struct Templates41 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates40<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42>
+struct Templates42 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates41<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43>
+struct Templates43 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates42<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44>
+struct Templates44 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates43<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43, T44> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45>
+struct Templates45 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates44<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43, T44, T45> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46>
+struct Templates46 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates45<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43, T44, T45, T46> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47>
+struct Templates47 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates46<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43, T44, T45, T46, T47> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48>
+struct Templates48 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates47<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43, T44, T45, T46, T47, T48> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+ GTEST_TEMPLATE_ T49>
+struct Templates49 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates48<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43, T44, T45, T46, T47, T48, T49> Tail;
+};
+
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+ GTEST_TEMPLATE_ T49, GTEST_TEMPLATE_ T50>
+struct Templates50 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates49<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+ T43, T44, T45, T46, T47, T48, T49, T50> Tail;
+};
+
+
+// We don't want to require the users to write TemplatesN<...> directly,
+// as that would require them to count the length. Templates<...> is much
+// easier to write, but generates horrible messages when there is a
+// compiler error, as gcc insists on printing out each template
+// argument, even if it has the default value (this means Templates<list>
+// will appear as Templates<list, NoneT, NoneT, ..., NoneT> in the compiler
+// errors).
+//
+// Our solution is to combine the best part of the two approaches: a
+// user would write Templates<T1, ..., TN>, and Google Test will translate
+// that to TemplatesN<T1, ..., TN> internally to make error messages
+// readable. The translation is done by the 'type' member of the
+// Templates template.
+template <GTEST_TEMPLATE_ T1 = NoneT, GTEST_TEMPLATE_ T2 = NoneT,
+ GTEST_TEMPLATE_ T3 = NoneT, GTEST_TEMPLATE_ T4 = NoneT,
+ GTEST_TEMPLATE_ T5 = NoneT, GTEST_TEMPLATE_ T6 = NoneT,
+ GTEST_TEMPLATE_ T7 = NoneT, GTEST_TEMPLATE_ T8 = NoneT,
+ GTEST_TEMPLATE_ T9 = NoneT, GTEST_TEMPLATE_ T10 = NoneT,
+ GTEST_TEMPLATE_ T11 = NoneT, GTEST_TEMPLATE_ T12 = NoneT,
+ GTEST_TEMPLATE_ T13 = NoneT, GTEST_TEMPLATE_ T14 = NoneT,
+ GTEST_TEMPLATE_ T15 = NoneT, GTEST_TEMPLATE_ T16 = NoneT,
+ GTEST_TEMPLATE_ T17 = NoneT, GTEST_TEMPLATE_ T18 = NoneT,
+ GTEST_TEMPLATE_ T19 = NoneT, GTEST_TEMPLATE_ T20 = NoneT,
+ GTEST_TEMPLATE_ T21 = NoneT, GTEST_TEMPLATE_ T22 = NoneT,
+ GTEST_TEMPLATE_ T23 = NoneT, GTEST_TEMPLATE_ T24 = NoneT,
+ GTEST_TEMPLATE_ T25 = NoneT, GTEST_TEMPLATE_ T26 = NoneT,
+ GTEST_TEMPLATE_ T27 = NoneT, GTEST_TEMPLATE_ T28 = NoneT,
+ GTEST_TEMPLATE_ T29 = NoneT, GTEST_TEMPLATE_ T30 = NoneT,
+ GTEST_TEMPLATE_ T31 = NoneT, GTEST_TEMPLATE_ T32 = NoneT,
+ GTEST_TEMPLATE_ T33 = NoneT, GTEST_TEMPLATE_ T34 = NoneT,
+ GTEST_TEMPLATE_ T35 = NoneT, GTEST_TEMPLATE_ T36 = NoneT,
+ GTEST_TEMPLATE_ T37 = NoneT, GTEST_TEMPLATE_ T38 = NoneT,
+ GTEST_TEMPLATE_ T39 = NoneT, GTEST_TEMPLATE_ T40 = NoneT,
+ GTEST_TEMPLATE_ T41 = NoneT, GTEST_TEMPLATE_ T42 = NoneT,
+ GTEST_TEMPLATE_ T43 = NoneT, GTEST_TEMPLATE_ T44 = NoneT,
+ GTEST_TEMPLATE_ T45 = NoneT, GTEST_TEMPLATE_ T46 = NoneT,
+ GTEST_TEMPLATE_ T47 = NoneT, GTEST_TEMPLATE_ T48 = NoneT,
+ GTEST_TEMPLATE_ T49 = NoneT, GTEST_TEMPLATE_ T50 = NoneT>
+struct Templates {
+ typedef Templates50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42, T43, T44, T45, T46, T47, T48, T49, T50> type;
+};
+
+template <>
+struct Templates<NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT> {
+ typedef Templates0 type;
+};
+template <GTEST_TEMPLATE_ T1>
+struct Templates<T1, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT> {
+ typedef Templates1<T1> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2>
+struct Templates<T1, T2, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT> {
+ typedef Templates2<T1, T2> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3>
+struct Templates<T1, T2, T3, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates3<T1, T2, T3> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4>
+struct Templates<T1, T2, T3, T4, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates4<T1, T2, T3, T4> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5>
+struct Templates<T1, T2, T3, T4, T5, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates5<T1, T2, T3, T4, T5> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6>
+struct Templates<T1, T2, T3, T4, T5, T6, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates6<T1, T2, T3, T4, T5, T6> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates7<T1, T2, T3, T4, T5, T6, T7> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates8<T1, T2, T3, T4, T5, T6, T7, T8> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates9<T1, T2, T3, T4, T5, T6, T7, T8, T9> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT> {
+ typedef Templates22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT> {
+ typedef Templates23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT> {
+ typedef Templates24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT> {
+ typedef Templates25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT> {
+ typedef Templates26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT> {
+ typedef Templates27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT> {
+ typedef Templates28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT> {
+ typedef Templates29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, NoneT, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, NoneT, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, NoneT, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, NoneT, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, NoneT,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42, T43> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+ NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42, T43, T44> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+ T45, NoneT, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42, T43, T44, T45> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+ T45, T46, NoneT, NoneT, NoneT, NoneT> {
+ typedef Templates46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42, T43, T44, T45, T46> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+ T45, T46, T47, NoneT, NoneT, NoneT> {
+ typedef Templates47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42, T43, T44, T45, T46, T47> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+ T45, T46, T47, T48, NoneT, NoneT> {
+ typedef Templates48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42, T43, T44, T45, T46, T47, T48> type;
+};
+template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+ GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+ GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+ GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+ GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+ GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+ GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+ GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+ GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+ GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+ GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+ GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+ GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+ GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+ GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+ GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+ GTEST_TEMPLATE_ T49>
+struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+ T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+ T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+ T45, T46, T47, T48, T49, NoneT> {
+ typedef Templates49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+ T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+ T42, T43, T44, T45, T46, T47, T48, T49> type;
+};
+
+// The TypeList template makes it possible to use either a single type
+// or a Types<...> list in TYPED_TEST_CASE() and
+// INSTANTIATE_TYPED_TEST_CASE_P().
+
+template <typename T>
+struct TypeList { typedef Types1<T> type; };
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10,
+ typename T11, typename T12, typename T13, typename T14, typename T15,
+ typename T16, typename T17, typename T18, typename T19, typename T20,
+ typename T21, typename T22, typename T23, typename T24, typename T25,
+ typename T26, typename T27, typename T28, typename T29, typename T30,
+ typename T31, typename T32, typename T33, typename T34, typename T35,
+ typename T36, typename T37, typename T38, typename T39, typename T40,
+ typename T41, typename T42, typename T43, typename T44, typename T45,
+ typename T46, typename T47, typename T48, typename T49, typename T50>
+struct TypeList<Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+ T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+ T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+ T44, T45, T46, T47, T48, T49, T50> > {
+ typedef typename Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+ T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+ T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+ T41, T42, T43, T44, T45, T46, T47, T48, T49, T50>::type type;
+};
+
+#endif // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
diff --git a/libvpx/third_party/googletest/src/include/gtest/internal/gtest-type-util.h.pump b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-type-util.h.pump
new file mode 100644
index 0000000..27f331d
--- /dev/null
+++ b/libvpx/third_party/googletest/src/include/gtest/internal/gtest-type-util.h.pump
@@ -0,0 +1,296 @@
+$$ -*- mode: c++; -*-
+$var n = 50 $$ Maximum length of type lists we want to support.
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Type utilities needed for implementing typed and type-parameterized
+// tests. This file is generated by a SCRIPT. DO NOT EDIT BY HAND!
+//
+// Currently we support at most $n types in a list, and at most $n
+// type-parameterized tests in one type-parameterized test case.
+// Please contact googletestframework@googlegroups.com if you need
+// more.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+
+#include "gtest/internal/gtest-port.h"
+#include "gtest/internal/gtest-string.h"
+
+// #ifdef __GNUC__ is too general here. It is possible to use gcc without using
+// libstdc++ (which is where cxxabi.h comes from).
+# ifdef __GLIBCXX__
+# include <cxxabi.h>
+# elif defined(__HP_aCC)
+# include <acxx_demangle.h>
+# endif // __GLIBCXX__
+
+namespace testing {
+namespace internal {
+
+// GetTypeName<T>() returns a human-readable name of type T.
+// NB: This function is also used in Google Mock, so don't move it inside of
+// the typed-test-only section below.
+template <typename T>
+String GetTypeName() {
+# if GTEST_HAS_RTTI
+
+ const char* const name = typeid(T).name();
+# if defined(__GLIBCXX__) || defined(__HP_aCC)
+ int status = 0;
+ // gcc's implementation of typeid(T).name() mangles the type name,
+ // so we have to demangle it.
+# ifdef __GLIBCXX__
+ using abi::__cxa_demangle;
+# endif // __GLIBCXX__
+ char* const readable_name = __cxa_demangle(name, 0, 0, &status);
+ const String name_str(status == 0 ? readable_name : name);
+ free(readable_name);
+ return name_str;
+# else
+ return name;
+# endif // __GLIBCXX__ || __HP_aCC
+
+# else
+
+ return "<type>";
+
+# endif // GTEST_HAS_RTTI
+}
+
+#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// AssertyTypeEq<T1, T2>::type is defined iff T1 and T2 are the same
+// type. This can be used as a compile-time assertion to ensure that
+// two types are equal.
+
+template <typename T1, typename T2>
+struct AssertTypeEq;
+
+template <typename T>
+struct AssertTypeEq<T, T> {
+ typedef bool type;
+};
+
+// A unique type used as the default value for the arguments of class
+// template Types. This allows us to simulate variadic templates
+// (e.g. Types<int>, Type<int, double>, and etc), which C++ doesn't
+// support directly.
+struct None {};
+
+// The following family of struct and struct templates are used to
+// represent type lists. In particular, TypesN<T1, T2, ..., TN>
+// represents a type list with N types (T1, T2, ..., and TN) in it.
+// Except for Types0, every struct in the family has two member types:
+// Head for the first type in the list, and Tail for the rest of the
+// list.
+
+// The empty type list.
+struct Types0 {};
+
+// Type lists of length 1, 2, 3, and so on.
+
+template <typename T1>
+struct Types1 {
+ typedef T1 Head;
+ typedef Types0 Tail;
+};
+
+$range i 2..n
+
+$for i [[
+$range j 1..i
+$range k 2..i
+template <$for j, [[typename T$j]]>
+struct Types$i {
+ typedef T1 Head;
+ typedef Types$(i-1)<$for k, [[T$k]]> Tail;
+};
+
+
+]]
+
+} // namespace internal
+
+// We don't want to require the users to write TypesN<...> directly,
+// as that would require them to count the length. Types<...> is much
+// easier to write, but generates horrible messages when there is a
+// compiler error, as gcc insists on printing out each template
+// argument, even if it has the default value (this means Types<int>
+// will appear as Types<int, None, None, ..., None> in the compiler
+// errors).
+//
+// Our solution is to combine the best part of the two approaches: a
+// user would write Types<T1, ..., TN>, and Google Test will translate
+// that to TypesN<T1, ..., TN> internally to make error messages
+// readable. The translation is done by the 'type' member of the
+// Types template.
+
+$range i 1..n
+template <$for i, [[typename T$i = internal::None]]>
+struct Types {
+ typedef internal::Types$n<$for i, [[T$i]]> type;
+};
+
+template <>
+struct Types<$for i, [[internal::None]]> {
+ typedef internal::Types0 type;
+};
+
+$range i 1..n-1
+$for i [[
+$range j 1..i
+$range k i+1..n
+template <$for j, [[typename T$j]]>
+struct Types<$for j, [[T$j]]$for k[[, internal::None]]> {
+ typedef internal::Types$i<$for j, [[T$j]]> type;
+};
+
+]]
+
+namespace internal {
+
+# define GTEST_TEMPLATE_ template <typename T> class
+
+// The template "selector" struct TemplateSel<Tmpl> is used to
+// represent Tmpl, which must be a class template with one type
+// parameter, as a type. TemplateSel<Tmpl>::Bind<T>::type is defined
+// as the type Tmpl<T>. This allows us to actually instantiate the
+// template "selected" by TemplateSel<Tmpl>.
+//
+// This trick is necessary for simulating typedef for class templates,
+// which C++ doesn't support directly.
+template <GTEST_TEMPLATE_ Tmpl>
+struct TemplateSel {
+ template <typename T>
+ struct Bind {
+ typedef Tmpl<T> type;
+ };
+};
+
+# define GTEST_BIND_(TmplSel, T) \
+ TmplSel::template Bind<T>::type
+
+// A unique struct template used as the default value for the
+// arguments of class template Templates. This allows us to simulate
+// variadic templates (e.g. Templates<int>, Templates<int, double>,
+// and etc), which C++ doesn't support directly.
+template <typename T>
+struct NoneT {};
+
+// The following family of struct and struct templates are used to
+// represent template lists. In particular, TemplatesN<T1, T2, ...,
+// TN> represents a list of N templates (T1, T2, ..., and TN). Except
+// for Templates0, every struct in the family has two member types:
+// Head for the selector of the first template in the list, and Tail
+// for the rest of the list.
+
+// The empty template list.
+struct Templates0 {};
+
+// Template lists of length 1, 2, 3, and so on.
+
+template <GTEST_TEMPLATE_ T1>
+struct Templates1 {
+ typedef TemplateSel<T1> Head;
+ typedef Templates0 Tail;
+};
+
+$range i 2..n
+
+$for i [[
+$range j 1..i
+$range k 2..i
+template <$for j, [[GTEST_TEMPLATE_ T$j]]>
+struct Templates$i {
+ typedef TemplateSel<T1> Head;
+ typedef Templates$(i-1)<$for k, [[T$k]]> Tail;
+};
+
+
+]]
+
+// We don't want to require the users to write TemplatesN<...> directly,
+// as that would require them to count the length. Templates<...> is much
+// easier to write, but generates horrible messages when there is a
+// compiler error, as gcc insists on printing out each template
+// argument, even if it has the default value (this means Templates<list>
+// will appear as Templates<list, NoneT, NoneT, ..., NoneT> in the compiler
+// errors).
+//
+// Our solution is to combine the best part of the two approaches: a
+// user would write Templates<T1, ..., TN>, and Google Test will translate
+// that to TemplatesN<T1, ..., TN> internally to make error messages
+// readable. The translation is done by the 'type' member of the
+// Templates template.
+
+$range i 1..n
+template <$for i, [[GTEST_TEMPLATE_ T$i = NoneT]]>
+struct Templates {
+ typedef Templates$n<$for i, [[T$i]]> type;
+};
+
+template <>
+struct Templates<$for i, [[NoneT]]> {
+ typedef Templates0 type;
+};
+
+$range i 1..n-1
+$for i [[
+$range j 1..i
+$range k i+1..n
+template <$for j, [[GTEST_TEMPLATE_ T$j]]>
+struct Templates<$for j, [[T$j]]$for k[[, NoneT]]> {
+ typedef Templates$i<$for j, [[T$j]]> type;
+};
+
+]]
+
+// The TypeList template makes it possible to use either a single type
+// or a Types<...> list in TYPED_TEST_CASE() and
+// INSTANTIATE_TYPED_TEST_CASE_P().
+
+template <typename T>
+struct TypeList { typedef Types1<T> type; };
+
+
+$range i 1..n
+template <$for i, [[typename T$i]]>
+struct TypeList<Types<$for i, [[T$i]]> > {
+ typedef typename Types<$for i, [[T$i]]>::type type;
+};
+
+#endif // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
diff --git a/libvpx/third_party/googletest/src/m4/acx_pthread.m4 b/libvpx/third_party/googletest/src/m4/acx_pthread.m4
new file mode 100644
index 0000000..2cf20de
--- /dev/null
+++ b/libvpx/third_party/googletest/src/m4/acx_pthread.m4
@@ -0,0 +1,363 @@
+# This was retrieved from
+# http://svn.0pointer.de/viewvc/trunk/common/acx_pthread.m4?revision=1277&root=avahi
+# See also (perhaps for new versions?)
+# http://svn.0pointer.de/viewvc/trunk/common/acx_pthread.m4?root=avahi
+#
+# We've rewritten the inconsistency check code (from avahi), to work
+# more broadly. In particular, it no longer assumes ld accepts -zdefs.
+# This caused a restructing of the code, but the functionality has only
+# changed a little.
+
+dnl @synopsis ACX_PTHREAD([ACTION-IF-FOUND[, ACTION-IF-NOT-FOUND]])
+dnl
+dnl @summary figure out how to build C programs using POSIX threads
+dnl
+dnl This macro figures out how to build C programs using POSIX threads.
+dnl It sets the PTHREAD_LIBS output variable to the threads library and
+dnl linker flags, and the PTHREAD_CFLAGS output variable to any special
+dnl C compiler flags that are needed. (The user can also force certain
+dnl compiler flags/libs to be tested by setting these environment
+dnl variables.)
+dnl
+dnl Also sets PTHREAD_CC to any special C compiler that is needed for
+dnl multi-threaded programs (defaults to the value of CC otherwise).
+dnl (This is necessary on AIX to use the special cc_r compiler alias.)
+dnl
+dnl NOTE: You are assumed to not only compile your program with these
+dnl flags, but also link it with them as well. e.g. you should link
+dnl with $PTHREAD_CC $CFLAGS $PTHREAD_CFLAGS $LDFLAGS ... $PTHREAD_LIBS
+dnl $LIBS
+dnl
+dnl If you are only building threads programs, you may wish to use
+dnl these variables in your default LIBS, CFLAGS, and CC:
+dnl
+dnl LIBS="$PTHREAD_LIBS $LIBS"
+dnl CFLAGS="$CFLAGS $PTHREAD_CFLAGS"
+dnl CC="$PTHREAD_CC"
+dnl
+dnl In addition, if the PTHREAD_CREATE_JOINABLE thread-attribute
+dnl constant has a nonstandard name, defines PTHREAD_CREATE_JOINABLE to
+dnl that name (e.g. PTHREAD_CREATE_UNDETACHED on AIX).
+dnl
+dnl ACTION-IF-FOUND is a list of shell commands to run if a threads
+dnl library is found, and ACTION-IF-NOT-FOUND is a list of commands to
+dnl run it if it is not found. If ACTION-IF-FOUND is not specified, the
+dnl default action will define HAVE_PTHREAD.
+dnl
+dnl Please let the authors know if this macro fails on any platform, or
+dnl if you have any other suggestions or comments. This macro was based
+dnl on work by SGJ on autoconf scripts for FFTW (www.fftw.org) (with
+dnl help from M. Frigo), as well as ac_pthread and hb_pthread macros
+dnl posted by Alejandro Forero Cuervo to the autoconf macro repository.
+dnl We are also grateful for the helpful feedback of numerous users.
+dnl
+dnl @category InstalledPackages
+dnl @author Steven G. Johnson <stevenj@alum.mit.edu>
+dnl @version 2006-05-29
+dnl @license GPLWithACException
+dnl
+dnl Checks for GCC shared/pthread inconsistency based on work by
+dnl Marcin Owsiany <marcin@owsiany.pl>
+
+
+AC_DEFUN([ACX_PTHREAD], [
+AC_REQUIRE([AC_CANONICAL_HOST])
+AC_LANG_SAVE
+AC_LANG_C
+acx_pthread_ok=no
+
+# We used to check for pthread.h first, but this fails if pthread.h
+# requires special compiler flags (e.g. on True64 or Sequent).
+# It gets checked for in the link test anyway.
+
+# First of all, check if the user has set any of the PTHREAD_LIBS,
+# etcetera environment variables, and if threads linking works using
+# them:
+if test x"$PTHREAD_LIBS$PTHREAD_CFLAGS" != x; then
+ save_CFLAGS="$CFLAGS"
+ CFLAGS="$CFLAGS $PTHREAD_CFLAGS"
+ save_LIBS="$LIBS"
+ LIBS="$PTHREAD_LIBS $LIBS"
+ AC_MSG_CHECKING([for pthread_join in LIBS=$PTHREAD_LIBS with CFLAGS=$PTHREAD_CFLAGS])
+ AC_TRY_LINK_FUNC(pthread_join, acx_pthread_ok=yes)
+ AC_MSG_RESULT($acx_pthread_ok)
+ if test x"$acx_pthread_ok" = xno; then
+ PTHREAD_LIBS=""
+ PTHREAD_CFLAGS=""
+ fi
+ LIBS="$save_LIBS"
+ CFLAGS="$save_CFLAGS"
+fi
+
+# We must check for the threads library under a number of different
+# names; the ordering is very important because some systems
+# (e.g. DEC) have both -lpthread and -lpthreads, where one of the
+# libraries is broken (non-POSIX).
+
+# Create a list of thread flags to try. Items starting with a "-" are
+# C compiler flags, and other items are library names, except for "none"
+# which indicates that we try without any flags at all, and "pthread-config"
+# which is a program returning the flags for the Pth emulation library.
+
+acx_pthread_flags="pthreads none -Kthread -kthread lthread -pthread -pthreads -mthreads pthread --thread-safe -mt pthread-config"
+
+# The ordering *is* (sometimes) important. Some notes on the
+# individual items follow:
+
+# pthreads: AIX (must check this before -lpthread)
+# none: in case threads are in libc; should be tried before -Kthread and
+# other compiler flags to prevent continual compiler warnings
+# -Kthread: Sequent (threads in libc, but -Kthread needed for pthread.h)
+# -kthread: FreeBSD kernel threads (preferred to -pthread since SMP-able)
+# lthread: LinuxThreads port on FreeBSD (also preferred to -pthread)
+# -pthread: Linux/gcc (kernel threads), BSD/gcc (userland threads)
+# -pthreads: Solaris/gcc
+# -mthreads: Mingw32/gcc, Lynx/gcc
+# -mt: Sun Workshop C (may only link SunOS threads [-lthread], but it
+# doesn't hurt to check since this sometimes defines pthreads too;
+# also defines -D_REENTRANT)
+# ... -mt is also the pthreads flag for HP/aCC
+# pthread: Linux, etcetera
+# --thread-safe: KAI C++
+# pthread-config: use pthread-config program (for GNU Pth library)
+
+case "${host_cpu}-${host_os}" in
+ *solaris*)
+
+ # On Solaris (at least, for some versions), libc contains stubbed
+ # (non-functional) versions of the pthreads routines, so link-based
+ # tests will erroneously succeed. (We need to link with -pthreads/-mt/
+ # -lpthread.) (The stubs are missing pthread_cleanup_push, or rather
+ # a function called by this macro, so we could check for that, but
+ # who knows whether they'll stub that too in a future libc.) So,
+ # we'll just look for -pthreads and -lpthread first:
+
+ acx_pthread_flags="-pthreads pthread -mt -pthread $acx_pthread_flags"
+ ;;
+esac
+
+if test x"$acx_pthread_ok" = xno; then
+for flag in $acx_pthread_flags; do
+
+ case $flag in
+ none)
+ AC_MSG_CHECKING([whether pthreads work without any flags])
+ ;;
+
+ -*)
+ AC_MSG_CHECKING([whether pthreads work with $flag])
+ PTHREAD_CFLAGS="$flag"
+ ;;
+
+ pthread-config)
+ AC_CHECK_PROG(acx_pthread_config, pthread-config, yes, no)
+ if test x"$acx_pthread_config" = xno; then continue; fi
+ PTHREAD_CFLAGS="`pthread-config --cflags`"
+ PTHREAD_LIBS="`pthread-config --ldflags` `pthread-config --libs`"
+ ;;
+
+ *)
+ AC_MSG_CHECKING([for the pthreads library -l$flag])
+ PTHREAD_LIBS="-l$flag"
+ ;;
+ esac
+
+ save_LIBS="$LIBS"
+ save_CFLAGS="$CFLAGS"
+ LIBS="$PTHREAD_LIBS $LIBS"
+ CFLAGS="$CFLAGS $PTHREAD_CFLAGS"
+
+ # Check for various functions. We must include pthread.h,
+ # since some functions may be macros. (On the Sequent, we
+ # need a special flag -Kthread to make this header compile.)
+ # We check for pthread_join because it is in -lpthread on IRIX
+ # while pthread_create is in libc. We check for pthread_attr_init
+ # due to DEC craziness with -lpthreads. We check for
+ # pthread_cleanup_push because it is one of the few pthread
+ # functions on Solaris that doesn't have a non-functional libc stub.
+ # We try pthread_create on general principles.
+ AC_TRY_LINK([#include <pthread.h>],
+ [pthread_t th; pthread_join(th, 0);
+ pthread_attr_init(0); pthread_cleanup_push(0, 0);
+ pthread_create(0,0,0,0); pthread_cleanup_pop(0); ],
+ [acx_pthread_ok=yes])
+
+ LIBS="$save_LIBS"
+ CFLAGS="$save_CFLAGS"
+
+ AC_MSG_RESULT($acx_pthread_ok)
+ if test "x$acx_pthread_ok" = xyes; then
+ break;
+ fi
+
+ PTHREAD_LIBS=""
+ PTHREAD_CFLAGS=""
+done
+fi
+
+# Various other checks:
+if test "x$acx_pthread_ok" = xyes; then
+ save_LIBS="$LIBS"
+ LIBS="$PTHREAD_LIBS $LIBS"
+ save_CFLAGS="$CFLAGS"
+ CFLAGS="$CFLAGS $PTHREAD_CFLAGS"
+
+ # Detect AIX lossage: JOINABLE attribute is called UNDETACHED.
+ AC_MSG_CHECKING([for joinable pthread attribute])
+ attr_name=unknown
+ for attr in PTHREAD_CREATE_JOINABLE PTHREAD_CREATE_UNDETACHED; do
+ AC_TRY_LINK([#include <pthread.h>], [int attr=$attr; return attr;],
+ [attr_name=$attr; break])
+ done
+ AC_MSG_RESULT($attr_name)
+ if test "$attr_name" != PTHREAD_CREATE_JOINABLE; then
+ AC_DEFINE_UNQUOTED(PTHREAD_CREATE_JOINABLE, $attr_name,
+ [Define to necessary symbol if this constant
+ uses a non-standard name on your system.])
+ fi
+
+ AC_MSG_CHECKING([if more special flags are required for pthreads])
+ flag=no
+ case "${host_cpu}-${host_os}" in
+ *-aix* | *-freebsd* | *-darwin*) flag="-D_THREAD_SAFE";;
+ *solaris* | *-osf* | *-hpux*) flag="-D_REENTRANT";;
+ esac
+ AC_MSG_RESULT(${flag})
+ if test "x$flag" != xno; then
+ PTHREAD_CFLAGS="$flag $PTHREAD_CFLAGS"
+ fi
+
+ LIBS="$save_LIBS"
+ CFLAGS="$save_CFLAGS"
+ # More AIX lossage: must compile with xlc_r or cc_r
+ if test x"$GCC" != xyes; then
+ AC_CHECK_PROGS(PTHREAD_CC, xlc_r cc_r, ${CC})
+ else
+ PTHREAD_CC=$CC
+ fi
+
+ # The next part tries to detect GCC inconsistency with -shared on some
+ # architectures and systems. The problem is that in certain
+ # configurations, when -shared is specified, GCC "forgets" to
+ # internally use various flags which are still necessary.
+
+ #
+ # Prepare the flags
+ #
+ save_CFLAGS="$CFLAGS"
+ save_LIBS="$LIBS"
+ save_CC="$CC"
+
+ # Try with the flags determined by the earlier checks.
+ #
+ # -Wl,-z,defs forces link-time symbol resolution, so that the
+ # linking checks with -shared actually have any value
+ #
+ # FIXME: -fPIC is required for -shared on many architectures,
+ # so we specify it here, but the right way would probably be to
+ # properly detect whether it is actually required.
+ CFLAGS="-shared -fPIC -Wl,-z,defs $CFLAGS $PTHREAD_CFLAGS"
+ LIBS="$PTHREAD_LIBS $LIBS"
+ CC="$PTHREAD_CC"
+
+ # In order not to create several levels of indentation, we test
+ # the value of "$done" until we find the cure or run out of ideas.
+ done="no"
+
+ # First, make sure the CFLAGS we added are actually accepted by our
+ # compiler. If not (and OS X's ld, for instance, does not accept -z),
+ # then we can't do this test.
+ if test x"$done" = xno; then
+ AC_MSG_CHECKING([whether to check for GCC pthread/shared inconsistencies])
+ AC_TRY_LINK(,, , [done=yes])
+
+ if test "x$done" = xyes ; then
+ AC_MSG_RESULT([no])
+ else
+ AC_MSG_RESULT([yes])
+ fi
+ fi
+
+ if test x"$done" = xno; then
+ AC_MSG_CHECKING([whether -pthread is sufficient with -shared])
+ AC_TRY_LINK([#include <pthread.h>],
+ [pthread_t th; pthread_join(th, 0);
+ pthread_attr_init(0); pthread_cleanup_push(0, 0);
+ pthread_create(0,0,0,0); pthread_cleanup_pop(0); ],
+ [done=yes])
+
+ if test "x$done" = xyes; then
+ AC_MSG_RESULT([yes])
+ else
+ AC_MSG_RESULT([no])
+ fi
+ fi
+
+ #
+ # Linux gcc on some architectures such as mips/mipsel forgets
+ # about -lpthread
+ #
+ if test x"$done" = xno; then
+ AC_MSG_CHECKING([whether -lpthread fixes that])
+ LIBS="-lpthread $PTHREAD_LIBS $save_LIBS"
+ AC_TRY_LINK([#include <pthread.h>],
+ [pthread_t th; pthread_join(th, 0);
+ pthread_attr_init(0); pthread_cleanup_push(0, 0);
+ pthread_create(0,0,0,0); pthread_cleanup_pop(0); ],
+ [done=yes])
+
+ if test "x$done" = xyes; then
+ AC_MSG_RESULT([yes])
+ PTHREAD_LIBS="-lpthread $PTHREAD_LIBS"
+ else
+ AC_MSG_RESULT([no])
+ fi
+ fi
+ #
+ # FreeBSD 4.10 gcc forgets to use -lc_r instead of -lc
+ #
+ if test x"$done" = xno; then
+ AC_MSG_CHECKING([whether -lc_r fixes that])
+ LIBS="-lc_r $PTHREAD_LIBS $save_LIBS"
+ AC_TRY_LINK([#include <pthread.h>],
+ [pthread_t th; pthread_join(th, 0);
+ pthread_attr_init(0); pthread_cleanup_push(0, 0);
+ pthread_create(0,0,0,0); pthread_cleanup_pop(0); ],
+ [done=yes])
+
+ if test "x$done" = xyes; then
+ AC_MSG_RESULT([yes])
+ PTHREAD_LIBS="-lc_r $PTHREAD_LIBS"
+ else
+ AC_MSG_RESULT([no])
+ fi
+ fi
+ if test x"$done" = xno; then
+ # OK, we have run out of ideas
+ AC_MSG_WARN([Impossible to determine how to use pthreads with shared libraries])
+
+ # so it's not safe to assume that we may use pthreads
+ acx_pthread_ok=no
+ fi
+
+ CFLAGS="$save_CFLAGS"
+ LIBS="$save_LIBS"
+ CC="$save_CC"
+else
+ PTHREAD_CC="$CC"
+fi
+
+AC_SUBST(PTHREAD_LIBS)
+AC_SUBST(PTHREAD_CFLAGS)
+AC_SUBST(PTHREAD_CC)
+
+# Finally, execute ACTION-IF-FOUND/ACTION-IF-NOT-FOUND:
+if test x"$acx_pthread_ok" = xyes; then
+ ifelse([$1],,AC_DEFINE(HAVE_PTHREAD,1,[Define if you have POSIX threads libraries and header files.]),[$1])
+ :
+else
+ acx_pthread_ok=no
+ $2
+fi
+AC_LANG_RESTORE
+])dnl ACX_PTHREAD
diff --git a/libvpx/third_party/googletest/src/m4/gtest.m4 b/libvpx/third_party/googletest/src/m4/gtest.m4
new file mode 100644
index 0000000..6598ba7
--- /dev/null
+++ b/libvpx/third_party/googletest/src/m4/gtest.m4
@@ -0,0 +1,74 @@
+dnl GTEST_LIB_CHECK([minimum version [,
+dnl action if found [,action if not found]]])
+dnl
+dnl Check for the presence of the Google Test library, optionally at a minimum
+dnl version, and indicate a viable version with the HAVE_GTEST flag. It defines
+dnl standard variables for substitution including GTEST_CPPFLAGS,
+dnl GTEST_CXXFLAGS, GTEST_LDFLAGS, and GTEST_LIBS. It also defines
+dnl GTEST_VERSION as the version of Google Test found. Finally, it provides
+dnl optional custom action slots in the event GTEST is found or not.
+AC_DEFUN([GTEST_LIB_CHECK],
+[
+dnl Provide a flag to enable or disable Google Test usage.
+AC_ARG_ENABLE([gtest],
+ [AS_HELP_STRING([--enable-gtest],
+ [Enable tests using the Google C++ Testing Framework.
+ (Default is enabled.)])],
+ [],
+ [enable_gtest=])
+AC_ARG_VAR([GTEST_CONFIG],
+ [The exact path of Google Test's 'gtest-config' script.])
+AC_ARG_VAR([GTEST_CPPFLAGS],
+ [C-like preprocessor flags for Google Test.])
+AC_ARG_VAR([GTEST_CXXFLAGS],
+ [C++ compile flags for Google Test.])
+AC_ARG_VAR([GTEST_LDFLAGS],
+ [Linker path and option flags for Google Test.])
+AC_ARG_VAR([GTEST_LIBS],
+ [Library linking flags for Google Test.])
+AC_ARG_VAR([GTEST_VERSION],
+ [The version of Google Test available.])
+HAVE_GTEST="no"
+AS_IF([test "x${enable_gtest}" != "xno"],
+ [AC_MSG_CHECKING([for 'gtest-config'])
+ AS_IF([test "x${enable_gtest}" != "xyes"],
+ [AS_IF([test -x "${enable_gtest}/scripts/gtest-config"],
+ [GTEST_CONFIG="${enable_gtest}/scripts/gtest-config"],
+ [GTEST_CONFIG="${enable_gtest}/bin/gtest-config"])
+ AS_IF([test -x "${GTEST_CONFIG}"], [],
+ [AC_MSG_RESULT([no])
+ AC_MSG_ERROR([dnl
+Unable to locate either a built or installed Google Test.
+The specific location '${enable_gtest}' was provided for a built or installed
+Google Test, but no 'gtest-config' script could be found at this location.])
+ ])],
+ [AC_PATH_PROG([GTEST_CONFIG], [gtest-config])])
+ AS_IF([test -x "${GTEST_CONFIG}"],
+ [AC_MSG_RESULT([${GTEST_CONFIG}])
+ m4_ifval([$1],
+ [_gtest_min_version="--min-version=$1"
+ AC_MSG_CHECKING([for Google Test at least version >= $1])],
+ [_gtest_min_version="--min-version=0"
+ AC_MSG_CHECKING([for Google Test])])
+ AS_IF([${GTEST_CONFIG} ${_gtest_min_version}],
+ [AC_MSG_RESULT([yes])
+ HAVE_GTEST='yes'],
+ [AC_MSG_RESULT([no])])],
+ [AC_MSG_RESULT([no])])
+ AS_IF([test "x${HAVE_GTEST}" = "xyes"],
+ [GTEST_CPPFLAGS=`${GTEST_CONFIG} --cppflags`
+ GTEST_CXXFLAGS=`${GTEST_CONFIG} --cxxflags`
+ GTEST_LDFLAGS=`${GTEST_CONFIG} --ldflags`
+ GTEST_LIBS=`${GTEST_CONFIG} --libs`
+ GTEST_VERSION=`${GTEST_CONFIG} --version`
+ AC_DEFINE([HAVE_GTEST],[1],[Defined when Google Test is available.])],
+ [AS_IF([test "x${enable_gtest}" = "xyes"],
+ [AC_MSG_ERROR([dnl
+Google Test was enabled, but no viable version could be found.])
+ ])])])
+AC_SUBST([HAVE_GTEST])
+AM_CONDITIONAL([HAVE_GTEST],[test "x$HAVE_GTEST" = "xyes"])
+AS_IF([test "x$HAVE_GTEST" = "xyes"],
+ [m4_ifval([$2], [$2])],
+ [m4_ifval([$3], [$3])])
+])
diff --git a/libvpx/third_party/googletest/src/make/Makefile b/libvpx/third_party/googletest/src/make/Makefile
new file mode 100644
index 0000000..5b27b6a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/make/Makefile
@@ -0,0 +1,80 @@
+# A sample Makefile for building Google Test and using it in user
+# tests. Please tweak it to suit your environment and project. You
+# may want to move it to your project's root directory.
+#
+# SYNOPSIS:
+#
+# make [all] - makes everything.
+# make TARGET - makes the given target.
+# make clean - removes all files generated by make.
+
+# Please tweak the following variable definitions as needed by your
+# project, except GTEST_HEADERS, which you can use in your own targets
+# but shouldn't modify.
+
+# Points to the root of Google Test, relative to where this file is.
+# Remember to tweak this if you move this file.
+GTEST_DIR = ..
+
+# Where to find user code.
+USER_DIR = ../samples
+
+# Flags passed to the preprocessor.
+CPPFLAGS += -I$(GTEST_DIR)/include
+
+# Flags passed to the C++ compiler.
+CXXFLAGS += -g -Wall -Wextra
+
+# All tests produced by this Makefile. Remember to add new tests you
+# created to the list.
+TESTS = sample1_unittest
+
+# All Google Test headers. Usually you shouldn't change this
+# definition.
+GTEST_HEADERS = $(GTEST_DIR)/include/gtest/*.h \
+ $(GTEST_DIR)/include/gtest/internal/*.h
+
+# House-keeping build targets.
+
+all : $(TESTS)
+
+clean :
+ rm -f $(TESTS) gtest.a gtest_main.a *.o
+
+# Builds gtest.a and gtest_main.a.
+
+# Usually you shouldn't tweak such internal variables, indicated by a
+# trailing _.
+GTEST_SRCS_ = $(GTEST_DIR)/src/*.cc $(GTEST_DIR)/src/*.h $(GTEST_HEADERS)
+
+# For simplicity and to avoid depending on Google Test's
+# implementation details, the dependencies specified below are
+# conservative and not optimized. This is fine as Google Test
+# compiles fast and for ordinary users its source rarely changes.
+gtest-all.o : $(GTEST_SRCS_)
+ $(CXX) $(CPPFLAGS) -I$(GTEST_DIR) $(CXXFLAGS) -c \
+ $(GTEST_DIR)/src/gtest-all.cc
+
+gtest_main.o : $(GTEST_SRCS_)
+ $(CXX) $(CPPFLAGS) -I$(GTEST_DIR) $(CXXFLAGS) -c \
+ $(GTEST_DIR)/src/gtest_main.cc
+
+gtest.a : gtest-all.o
+ $(AR) $(ARFLAGS) $@ $^
+
+gtest_main.a : gtest-all.o gtest_main.o
+ $(AR) $(ARFLAGS) $@ $^
+
+# Builds a sample test. A test should link with either gtest.a or
+# gtest_main.a, depending on whether it defines its own main()
+# function.
+
+sample1.o : $(USER_DIR)/sample1.cc $(USER_DIR)/sample1.h $(GTEST_HEADERS)
+ $(CXX) $(CPPFLAGS) $(CXXFLAGS) -c $(USER_DIR)/sample1.cc
+
+sample1_unittest.o : $(USER_DIR)/sample1_unittest.cc \
+ $(USER_DIR)/sample1.h $(GTEST_HEADERS)
+ $(CXX) $(CPPFLAGS) $(CXXFLAGS) -c $(USER_DIR)/sample1_unittest.cc
+
+sample1_unittest : sample1.o sample1_unittest.o gtest_main.a
+ $(CXX) $(CPPFLAGS) $(CXXFLAGS) -lpthread $^ -o $@
diff --git a/libvpx/third_party/googletest/src/msvc/gtest-md.sln b/libvpx/third_party/googletest/src/msvc/gtest-md.sln
new file mode 100644
index 0000000..f7908da
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest-md.sln
@@ -0,0 +1,45 @@
+Microsoft Visual Studio Solution File, Format Version 8.00
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "gtest-md", "gtest-md.vcproj", "{C8F6C172-56F2-4E76-B5FA-C3B423B31BE8}"
+ ProjectSection(ProjectDependencies) = postProject
+ EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "gtest_main-md", "gtest_main-md.vcproj", "{3AF54C8A-10BF-4332-9147-F68ED9862033}"
+ ProjectSection(ProjectDependencies) = postProject
+ EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "gtest_prod_test-md", "gtest_prod_test-md.vcproj", "{24848551-EF4F-47E8-9A9D-EA4D49BC3ECB}"
+ ProjectSection(ProjectDependencies) = postProject
+ EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "gtest_unittest-md", "gtest_unittest-md.vcproj", "{4D9FDFB5-986A-4139-823C-F4EE0ED481A2}"
+ ProjectSection(ProjectDependencies) = postProject
+ EndProjectSection
+EndProject
+Global
+ GlobalSection(SolutionConfiguration) = preSolution
+ Debug = Debug
+ Release = Release
+ EndGlobalSection
+ GlobalSection(ProjectConfiguration) = postSolution
+ {C8F6C172-56F2-4E76-B5FA-C3B423B31BE8}.Debug.ActiveCfg = Debug|Win32
+ {C8F6C172-56F2-4E76-B5FA-C3B423B31BE8}.Debug.Build.0 = Debug|Win32
+ {C8F6C172-56F2-4E76-B5FA-C3B423B31BE8}.Release.ActiveCfg = Release|Win32
+ {C8F6C172-56F2-4E76-B5FA-C3B423B31BE8}.Release.Build.0 = Release|Win32
+ {3AF54C8A-10BF-4332-9147-F68ED9862033}.Debug.ActiveCfg = Debug|Win32
+ {3AF54C8A-10BF-4332-9147-F68ED9862033}.Debug.Build.0 = Debug|Win32
+ {3AF54C8A-10BF-4332-9147-F68ED9862033}.Release.ActiveCfg = Release|Win32
+ {3AF54C8A-10BF-4332-9147-F68ED9862033}.Release.Build.0 = Release|Win32
+ {24848551-EF4F-47E8-9A9D-EA4D49BC3ECB}.Debug.ActiveCfg = Debug|Win32
+ {24848551-EF4F-47E8-9A9D-EA4D49BC3ECB}.Debug.Build.0 = Debug|Win32
+ {24848551-EF4F-47E8-9A9D-EA4D49BC3ECB}.Release.ActiveCfg = Release|Win32
+ {24848551-EF4F-47E8-9A9D-EA4D49BC3ECB}.Release.Build.0 = Release|Win32
+ {4D9FDFB5-986A-4139-823C-F4EE0ED481A2}.Debug.ActiveCfg = Debug|Win32
+ {4D9FDFB5-986A-4139-823C-F4EE0ED481A2}.Debug.Build.0 = Debug|Win32
+ {4D9FDFB5-986A-4139-823C-F4EE0ED481A2}.Release.ActiveCfg = Release|Win32
+ {4D9FDFB5-986A-4139-823C-F4EE0ED481A2}.Release.Build.0 = Release|Win32
+ EndGlobalSection
+ GlobalSection(ExtensibilityGlobals) = postSolution
+ EndGlobalSection
+ GlobalSection(ExtensibilityAddIns) = postSolution
+ EndGlobalSection
+EndGlobal
diff --git a/libvpx/third_party/googletest/src/msvc/gtest-md.vcproj b/libvpx/third_party/googletest/src/msvc/gtest-md.vcproj
new file mode 100644
index 0000000..1c35c3a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest-md.vcproj
@@ -0,0 +1,126 @@
+<?xml version="1.0" encoding="Windows-1252"?>
+<VisualStudioProject
+ ProjectType="Visual C++"
+ Version="7.10"
+ Name="gtest-md"
+ ProjectGUID="{C8F6C172-56F2-4E76-B5FA-C3B423B31BE8}"
+ Keyword="Win32Proj">
+ <Platforms>
+ <Platform
+ Name="Win32"/>
+ </Platforms>
+ <Configurations>
+ <Configuration
+ Name="Debug|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="4"
+ CharacterSet="2"
+ ReferencesPath="">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="0"
+ PreprocessorDefinitions="WIN32;_DEBUG;_LIB"
+ MinimalRebuild="TRUE"
+ BasicRuntimeChecks="3"
+ RuntimeLibrary="3"
+ UsePrecompiledHeader="0"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="4"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLibrarianTool"
+ OutputFile="$(OutDir)/gtestd.lib"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ <Configuration
+ Name="Release|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="4"
+ CharacterSet="2"
+ ReferencesPath=""..\include";".."">
+ <Tool
+ Name="VCCLCompilerTool"
+ PreprocessorDefinitions="WIN32;NDEBUG;_LIB"
+ RuntimeLibrary="2"
+ UsePrecompiledHeader="0"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="3"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLibrarianTool"
+ OutputFile="$(OutDir)/gtest.lib"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ </Configurations>
+ <References>
+ </References>
+ <Files>
+ <Filter
+ Name="Source Files"
+ Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm;asmx"
+ UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}">
+ <File
+ RelativePath="..\src\gtest-all.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""/>
+ </FileConfiguration>
+ </File>
+ </Filter>
+ <Filter
+ Name="Header Files"
+ Filter="h;hpp;hxx;hm;inl;inc;xsd"
+ UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}">
+ </Filter>
+ </Files>
+ <Globals>
+ </Globals>
+</VisualStudioProject>
diff --git a/libvpx/third_party/googletest/src/msvc/gtest.sln b/libvpx/third_party/googletest/src/msvc/gtest.sln
new file mode 100644
index 0000000..ef4b057
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest.sln
@@ -0,0 +1,45 @@
+Microsoft Visual Studio Solution File, Format Version 8.00
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "gtest", "gtest.vcproj", "{C8F6C172-56F2-4E76-B5FA-C3B423B31BE7}"
+ ProjectSection(ProjectDependencies) = postProject
+ EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "gtest_main", "gtest_main.vcproj", "{3AF54C8A-10BF-4332-9147-F68ED9862032}"
+ ProjectSection(ProjectDependencies) = postProject
+ EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "gtest_unittest", "gtest_unittest.vcproj", "{4D9FDFB5-986A-4139-823C-F4EE0ED481A1}"
+ ProjectSection(ProjectDependencies) = postProject
+ EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "gtest_prod_test", "gtest_prod_test.vcproj", "{24848551-EF4F-47E8-9A9D-EA4D49BC3ECA}"
+ ProjectSection(ProjectDependencies) = postProject
+ EndProjectSection
+EndProject
+Global
+ GlobalSection(SolutionConfiguration) = preSolution
+ Debug = Debug
+ Release = Release
+ EndGlobalSection
+ GlobalSection(ProjectConfiguration) = postSolution
+ {C8F6C172-56F2-4E76-B5FA-C3B423B31BE7}.Debug.ActiveCfg = Debug|Win32
+ {C8F6C172-56F2-4E76-B5FA-C3B423B31BE7}.Debug.Build.0 = Debug|Win32
+ {C8F6C172-56F2-4E76-B5FA-C3B423B31BE7}.Release.ActiveCfg = Release|Win32
+ {C8F6C172-56F2-4E76-B5FA-C3B423B31BE7}.Release.Build.0 = Release|Win32
+ {3AF54C8A-10BF-4332-9147-F68ED9862032}.Debug.ActiveCfg = Debug|Win32
+ {3AF54C8A-10BF-4332-9147-F68ED9862032}.Debug.Build.0 = Debug|Win32
+ {3AF54C8A-10BF-4332-9147-F68ED9862032}.Release.ActiveCfg = Release|Win32
+ {3AF54C8A-10BF-4332-9147-F68ED9862032}.Release.Build.0 = Release|Win32
+ {4D9FDFB5-986A-4139-823C-F4EE0ED481A1}.Debug.ActiveCfg = Debug|Win32
+ {4D9FDFB5-986A-4139-823C-F4EE0ED481A1}.Debug.Build.0 = Debug|Win32
+ {4D9FDFB5-986A-4139-823C-F4EE0ED481A1}.Release.ActiveCfg = Release|Win32
+ {4D9FDFB5-986A-4139-823C-F4EE0ED481A1}.Release.Build.0 = Release|Win32
+ {24848551-EF4F-47E8-9A9D-EA4D49BC3ECA}.Debug.ActiveCfg = Debug|Win32
+ {24848551-EF4F-47E8-9A9D-EA4D49BC3ECA}.Debug.Build.0 = Debug|Win32
+ {24848551-EF4F-47E8-9A9D-EA4D49BC3ECA}.Release.ActiveCfg = Release|Win32
+ {24848551-EF4F-47E8-9A9D-EA4D49BC3ECA}.Release.Build.0 = Release|Win32
+ EndGlobalSection
+ GlobalSection(ExtensibilityGlobals) = postSolution
+ EndGlobalSection
+ GlobalSection(ExtensibilityAddIns) = postSolution
+ EndGlobalSection
+EndGlobal
diff --git a/libvpx/third_party/googletest/src/msvc/gtest.vcproj b/libvpx/third_party/googletest/src/msvc/gtest.vcproj
new file mode 100644
index 0000000..a8373ce
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest.vcproj
@@ -0,0 +1,126 @@
+<?xml version="1.0" encoding="Windows-1252"?>
+<VisualStudioProject
+ ProjectType="Visual C++"
+ Version="7.10"
+ Name="gtest"
+ ProjectGUID="{C8F6C172-56F2-4E76-B5FA-C3B423B31BE7}"
+ Keyword="Win32Proj">
+ <Platforms>
+ <Platform
+ Name="Win32"/>
+ </Platforms>
+ <Configurations>
+ <Configuration
+ Name="Debug|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="4"
+ CharacterSet="2"
+ ReferencesPath="">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="0"
+ PreprocessorDefinitions="WIN32;_DEBUG;_LIB"
+ MinimalRebuild="TRUE"
+ BasicRuntimeChecks="3"
+ RuntimeLibrary="5"
+ UsePrecompiledHeader="0"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="4"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLibrarianTool"
+ OutputFile="$(OutDir)/gtestd.lib"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ <Configuration
+ Name="Release|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="4"
+ CharacterSet="2"
+ ReferencesPath=""..\include";".."">
+ <Tool
+ Name="VCCLCompilerTool"
+ PreprocessorDefinitions="WIN32;NDEBUG;_LIB"
+ RuntimeLibrary="4"
+ UsePrecompiledHeader="0"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="3"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLibrarianTool"
+ OutputFile="$(OutDir)/gtest.lib"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ </Configurations>
+ <References>
+ </References>
+ <Files>
+ <Filter
+ Name="Source Files"
+ Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm;asmx"
+ UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}">
+ <File
+ RelativePath="..\src\gtest-all.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""/>
+ </FileConfiguration>
+ </File>
+ </Filter>
+ <Filter
+ Name="Header Files"
+ Filter="h;hpp;hxx;hm;inl;inc;xsd"
+ UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}">
+ </Filter>
+ </Files>
+ <Globals>
+ </Globals>
+</VisualStudioProject>
diff --git a/libvpx/third_party/googletest/src/msvc/gtest_main-md.vcproj b/libvpx/third_party/googletest/src/msvc/gtest_main-md.vcproj
new file mode 100644
index 0000000..b5379fe
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest_main-md.vcproj
@@ -0,0 +1,129 @@
+<?xml version="1.0" encoding="Windows-1252"?>
+<VisualStudioProject
+ ProjectType="Visual C++"
+ Version="7.10"
+ Name="gtest_main-md"
+ ProjectGUID="{3AF54C8A-10BF-4332-9147-F68ED9862033}"
+ Keyword="Win32Proj">
+ <Platforms>
+ <Platform
+ Name="Win32"/>
+ </Platforms>
+ <Configurations>
+ <Configuration
+ Name="Debug|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="4"
+ CharacterSet="2"
+ ReferencesPath="">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="0"
+ PreprocessorDefinitions="WIN32;_DEBUG;_LIB"
+ MinimalRebuild="TRUE"
+ BasicRuntimeChecks="3"
+ RuntimeLibrary="3"
+ UsePrecompiledHeader="0"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="4"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLibrarianTool"
+ OutputFile="$(OutDir)/$(ProjectName)d.lib"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ <Configuration
+ Name="Release|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="4"
+ CharacterSet="2"
+ ReferencesPath=""..\include";".."">
+ <Tool
+ Name="VCCLCompilerTool"
+ PreprocessorDefinitions="WIN32;NDEBUG;_LIB"
+ RuntimeLibrary="2"
+ UsePrecompiledHeader="0"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="3"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLibrarianTool"
+ OutputFile="$(OutDir)/$(ProjectName).lib"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ </Configurations>
+ <References>
+ <ProjectReference
+ ReferencedProjectIdentifier="{C8F6C172-56F2-4E76-B5FA-C3B423B31BE8}"
+ Name="gtest-md"/>
+ </References>
+ <Files>
+ <Filter
+ Name="Source Files"
+ Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm;asmx"
+ UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}">
+ <File
+ RelativePath="..\src\gtest_main.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""/>
+ </FileConfiguration>
+ </File>
+ </Filter>
+ <Filter
+ Name="Header Files"
+ Filter="h;hpp;hxx;hm;inl;inc;xsd"
+ UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}">
+ </Filter>
+ </Files>
+ <Globals>
+ </Globals>
+</VisualStudioProject>
diff --git a/libvpx/third_party/googletest/src/msvc/gtest_main.vcproj b/libvpx/third_party/googletest/src/msvc/gtest_main.vcproj
new file mode 100644
index 0000000..e8b763c
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest_main.vcproj
@@ -0,0 +1,129 @@
+<?xml version="1.0" encoding="Windows-1252"?>
+<VisualStudioProject
+ ProjectType="Visual C++"
+ Version="7.10"
+ Name="gtest_main"
+ ProjectGUID="{3AF54C8A-10BF-4332-9147-F68ED9862032}"
+ Keyword="Win32Proj">
+ <Platforms>
+ <Platform
+ Name="Win32"/>
+ </Platforms>
+ <Configurations>
+ <Configuration
+ Name="Debug|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="4"
+ CharacterSet="2"
+ ReferencesPath="">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="0"
+ PreprocessorDefinitions="WIN32;_DEBUG;_LIB"
+ MinimalRebuild="TRUE"
+ BasicRuntimeChecks="3"
+ RuntimeLibrary="5"
+ UsePrecompiledHeader="0"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="4"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLibrarianTool"
+ OutputFile="$(OutDir)/$(ProjectName)d.lib"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ <Configuration
+ Name="Release|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="4"
+ CharacterSet="2"
+ ReferencesPath=""..\include";".."">
+ <Tool
+ Name="VCCLCompilerTool"
+ PreprocessorDefinitions="WIN32;NDEBUG;_LIB"
+ RuntimeLibrary="4"
+ UsePrecompiledHeader="0"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="3"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLibrarianTool"
+ OutputFile="$(OutDir)/$(ProjectName).lib"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ </Configurations>
+ <References>
+ <ProjectReference
+ ReferencedProjectIdentifier="{C8F6C172-56F2-4E76-B5FA-C3B423B31BE7}"
+ Name="gtest"/>
+ </References>
+ <Files>
+ <Filter
+ Name="Source Files"
+ Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm;asmx"
+ UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}">
+ <File
+ RelativePath="..\src\gtest_main.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""/>
+ </FileConfiguration>
+ </File>
+ </Filter>
+ <Filter
+ Name="Header Files"
+ Filter="h;hpp;hxx;hm;inl;inc;xsd"
+ UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}">
+ </Filter>
+ </Files>
+ <Globals>
+ </Globals>
+</VisualStudioProject>
diff --git a/libvpx/third_party/googletest/src/msvc/gtest_prod_test-md.vcproj b/libvpx/third_party/googletest/src/msvc/gtest_prod_test-md.vcproj
new file mode 100644
index 0000000..05b05d9
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest_prod_test-md.vcproj
@@ -0,0 +1,164 @@
+<?xml version="1.0" encoding="Windows-1252"?>
+<VisualStudioProject
+ ProjectType="Visual C++"
+ Version="7.10"
+ Name="gtest_prod_test-md"
+ ProjectGUID="{24848551-EF4F-47E8-9A9D-EA4D49BC3ECB}"
+ Keyword="Win32Proj">
+ <Platforms>
+ <Platform
+ Name="Win32"/>
+ </Platforms>
+ <Configurations>
+ <Configuration
+ Name="Debug|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="1"
+ CharacterSet="2">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="0"
+ PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE"
+ MinimalRebuild="TRUE"
+ BasicRuntimeChecks="3"
+ RuntimeLibrary="3"
+ UsePrecompiledHeader="3"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="4"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLinkerTool"
+ OutputFile="$(OutDir)/gtest_prod_test.exe"
+ LinkIncremental="2"
+ GenerateDebugInformation="TRUE"
+ ProgramDatabaseFile="$(OutDir)/gtest_prod_test.pdb"
+ SubSystem="1"
+ TargetMachine="1"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCWebDeploymentTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ <Configuration
+ Name="Release|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="1"
+ CharacterSet="2">
+ <Tool
+ Name="VCCLCompilerTool"
+ PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE"
+ RuntimeLibrary="2"
+ UsePrecompiledHeader="3"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="3"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLinkerTool"
+ OutputFile="$(OutDir)/gtest_prod_test.exe"
+ LinkIncremental="1"
+ GenerateDebugInformation="TRUE"
+ SubSystem="1"
+ OptimizeReferences="2"
+ EnableCOMDATFolding="2"
+ TargetMachine="1"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCWebDeploymentTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ </Configurations>
+ <References>
+ <ProjectReference
+ ReferencedProjectIdentifier="{3AF54C8A-10BF-4332-9147-F68ED9862033}"
+ Name="gtest_main-md"/>
+ </References>
+ <Files>
+ <Filter
+ Name="Source Files"
+ Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm;asmx"
+ UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}">
+ <File
+ RelativePath="..\test\gtest_prod_test.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ </File>
+ <File
+ RelativePath="..\test\production.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ </File>
+ </Filter>
+ <Filter
+ Name="Header Files"
+ Filter="h;hpp;hxx;hm;inl;inc;xsd"
+ UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}">
+ <File
+ RelativePath="..\test\production.h">
+ </File>
+ </Filter>
+ </Files>
+ <Globals>
+ </Globals>
+</VisualStudioProject>
diff --git a/libvpx/third_party/googletest/src/msvc/gtest_prod_test.vcproj b/libvpx/third_party/googletest/src/msvc/gtest_prod_test.vcproj
new file mode 100644
index 0000000..6d7a2f0
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest_prod_test.vcproj
@@ -0,0 +1,164 @@
+<?xml version="1.0" encoding="Windows-1252"?>
+<VisualStudioProject
+ ProjectType="Visual C++"
+ Version="7.10"
+ Name="gtest_prod_test"
+ ProjectGUID="{24848551-EF4F-47E8-9A9D-EA4D49BC3ECA}"
+ Keyword="Win32Proj">
+ <Platforms>
+ <Platform
+ Name="Win32"/>
+ </Platforms>
+ <Configurations>
+ <Configuration
+ Name="Debug|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="1"
+ CharacterSet="2">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="0"
+ PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE"
+ MinimalRebuild="TRUE"
+ BasicRuntimeChecks="3"
+ RuntimeLibrary="5"
+ UsePrecompiledHeader="3"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="4"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLinkerTool"
+ OutputFile="$(OutDir)/gtest_prod_test.exe"
+ LinkIncremental="2"
+ GenerateDebugInformation="TRUE"
+ ProgramDatabaseFile="$(OutDir)/gtest_prod_test.pdb"
+ SubSystem="1"
+ TargetMachine="1"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCWebDeploymentTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ <Configuration
+ Name="Release|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="1"
+ CharacterSet="2">
+ <Tool
+ Name="VCCLCompilerTool"
+ PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE"
+ RuntimeLibrary="4"
+ UsePrecompiledHeader="3"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="3"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLinkerTool"
+ OutputFile="$(OutDir)/gtest_prod_test.exe"
+ LinkIncremental="1"
+ GenerateDebugInformation="TRUE"
+ SubSystem="1"
+ OptimizeReferences="2"
+ EnableCOMDATFolding="2"
+ TargetMachine="1"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCWebDeploymentTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ </Configurations>
+ <References>
+ <ProjectReference
+ ReferencedProjectIdentifier="{3AF54C8A-10BF-4332-9147-F68ED9862032}"
+ Name="gtest_main"/>
+ </References>
+ <Files>
+ <Filter
+ Name="Source Files"
+ Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm;asmx"
+ UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}">
+ <File
+ RelativePath="..\test\gtest_prod_test.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ </File>
+ <File
+ RelativePath="..\test\production.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ </File>
+ </Filter>
+ <Filter
+ Name="Header Files"
+ Filter="h;hpp;hxx;hm;inl;inc;xsd"
+ UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}">
+ <File
+ RelativePath="..\test\production.h">
+ </File>
+ </Filter>
+ </Files>
+ <Globals>
+ </Globals>
+</VisualStudioProject>
diff --git a/libvpx/third_party/googletest/src/msvc/gtest_unittest-md.vcproj b/libvpx/third_party/googletest/src/msvc/gtest_unittest-md.vcproj
new file mode 100644
index 0000000..38a5e56
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest_unittest-md.vcproj
@@ -0,0 +1,147 @@
+<?xml version="1.0" encoding="Windows-1252"?>
+<VisualStudioProject
+ ProjectType="Visual C++"
+ Version="7.10"
+ Name="gtest_unittest-md"
+ ProjectGUID="{4D9FDFB5-986A-4139-823C-F4EE0ED481A2}"
+ Keyword="Win32Proj">
+ <Platforms>
+ <Platform
+ Name="Win32"/>
+ </Platforms>
+ <Configurations>
+ <Configuration
+ Name="Debug|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="1"
+ CharacterSet="2">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="0"
+ PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE"
+ MinimalRebuild="TRUE"
+ BasicRuntimeChecks="3"
+ RuntimeLibrary="3"
+ UsePrecompiledHeader="3"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="4"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLinkerTool"
+ OutputFile="$(OutDir)/gtest_unittest.exe"
+ LinkIncremental="2"
+ GenerateDebugInformation="TRUE"
+ ProgramDatabaseFile="$(OutDir)/gtest_unittest.pdb"
+ SubSystem="1"
+ TargetMachine="1"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCWebDeploymentTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ <Configuration
+ Name="Release|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="1"
+ CharacterSet="2">
+ <Tool
+ Name="VCCLCompilerTool"
+ PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE"
+ RuntimeLibrary="2"
+ UsePrecompiledHeader="3"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="3"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLinkerTool"
+ OutputFile="$(OutDir)/gtest_unittest.exe"
+ LinkIncremental="1"
+ GenerateDebugInformation="TRUE"
+ SubSystem="1"
+ OptimizeReferences="2"
+ EnableCOMDATFolding="2"
+ TargetMachine="1"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCWebDeploymentTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ </Configurations>
+ <References>
+ <ProjectReference
+ ReferencedProjectIdentifier="{3AF54C8A-10BF-4332-9147-F68ED9862033}"
+ Name="gtest_main-md"/>
+ </References>
+ <Files>
+ <Filter
+ Name="Source Files"
+ Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm;asmx"
+ UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}">
+ <File
+ RelativePath="..\test\gtest_unittest.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="1"
+ AdditionalIncludeDirectories=""..";"..\include""
+ BasicRuntimeChecks="0"
+ UsePrecompiledHeader="0"
+ DebugInformationFormat="3"/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ </File>
+ </Filter>
+ <Filter
+ Name="Header Files"
+ Filter="h;hpp;hxx;hm;inl;inc;xsd"
+ UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}">
+ </Filter>
+ </Files>
+ <Globals>
+ </Globals>
+</VisualStudioProject>
diff --git a/libvpx/third_party/googletest/src/msvc/gtest_unittest.vcproj b/libvpx/third_party/googletest/src/msvc/gtest_unittest.vcproj
new file mode 100644
index 0000000..cb1f52b
--- /dev/null
+++ b/libvpx/third_party/googletest/src/msvc/gtest_unittest.vcproj
@@ -0,0 +1,147 @@
+<?xml version="1.0" encoding="Windows-1252"?>
+<VisualStudioProject
+ ProjectType="Visual C++"
+ Version="7.10"
+ Name="gtest_unittest"
+ ProjectGUID="{4D9FDFB5-986A-4139-823C-F4EE0ED481A1}"
+ Keyword="Win32Proj">
+ <Platforms>
+ <Platform
+ Name="Win32"/>
+ </Platforms>
+ <Configurations>
+ <Configuration
+ Name="Debug|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="1"
+ CharacterSet="2">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="0"
+ PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE"
+ MinimalRebuild="TRUE"
+ BasicRuntimeChecks="3"
+ RuntimeLibrary="5"
+ UsePrecompiledHeader="3"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="4"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLinkerTool"
+ OutputFile="$(OutDir)/gtest_unittest.exe"
+ LinkIncremental="2"
+ GenerateDebugInformation="TRUE"
+ ProgramDatabaseFile="$(OutDir)/gtest_unittest.pdb"
+ SubSystem="1"
+ TargetMachine="1"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCWebDeploymentTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ <Configuration
+ Name="Release|Win32"
+ OutputDirectory="$(SolutionName)/$(ConfigurationName)"
+ IntermediateDirectory="$(OutDir)/$(ProjectName)"
+ ConfigurationType="1"
+ CharacterSet="2">
+ <Tool
+ Name="VCCLCompilerTool"
+ PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE"
+ RuntimeLibrary="4"
+ UsePrecompiledHeader="3"
+ WarningLevel="3"
+ Detect64BitPortabilityProblems="FALSE"
+ DebugInformationFormat="3"/>
+ <Tool
+ Name="VCCustomBuildTool"/>
+ <Tool
+ Name="VCLinkerTool"
+ OutputFile="$(OutDir)/gtest_unittest.exe"
+ LinkIncremental="1"
+ GenerateDebugInformation="TRUE"
+ SubSystem="1"
+ OptimizeReferences="2"
+ EnableCOMDATFolding="2"
+ TargetMachine="1"/>
+ <Tool
+ Name="VCMIDLTool"/>
+ <Tool
+ Name="VCPostBuildEventTool"/>
+ <Tool
+ Name="VCPreBuildEventTool"/>
+ <Tool
+ Name="VCPreLinkEventTool"/>
+ <Tool
+ Name="VCResourceCompilerTool"/>
+ <Tool
+ Name="VCWebServiceProxyGeneratorTool"/>
+ <Tool
+ Name="VCXMLDataGeneratorTool"/>
+ <Tool
+ Name="VCWebDeploymentTool"/>
+ <Tool
+ Name="VCManagedWrapperGeneratorTool"/>
+ <Tool
+ Name="VCAuxiliaryManagedWrapperGeneratorTool"/>
+ </Configuration>
+ </Configurations>
+ <References>
+ <ProjectReference
+ ReferencedProjectIdentifier="{3AF54C8A-10BF-4332-9147-F68ED9862032}"
+ Name="gtest_main"/>
+ </References>
+ <Files>
+ <Filter
+ Name="Source Files"
+ Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm;asmx"
+ UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}">
+ <File
+ RelativePath="..\test\gtest_unittest.cc">
+ <FileConfiguration
+ Name="Debug|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ Optimization="1"
+ AdditionalIncludeDirectories=""..";"..\include""
+ BasicRuntimeChecks="0"
+ UsePrecompiledHeader="0"
+ DebugInformationFormat="3"/>
+ </FileConfiguration>
+ <FileConfiguration
+ Name="Release|Win32">
+ <Tool
+ Name="VCCLCompilerTool"
+ AdditionalIncludeDirectories=""..";"..\include""
+ UsePrecompiledHeader="0"/>
+ </FileConfiguration>
+ </File>
+ </Filter>
+ <Filter
+ Name="Header Files"
+ Filter="h;hpp;hxx;hm;inl;inc;xsd"
+ UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}">
+ </Filter>
+ </Files>
+ <Globals>
+ </Globals>
+</VisualStudioProject>
diff --git a/libvpx/third_party/googletest/src/samples/prime_tables.h b/libvpx/third_party/googletest/src/samples/prime_tables.h
new file mode 100644
index 0000000..92ce16a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/prime_tables.h
@@ -0,0 +1,123 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+// Author: vladl@google.com (Vlad Losev)
+
+// This provides interface PrimeTable that determines whether a number is a
+// prime and determines a next prime number. This interface is used
+// in Google Test samples demonstrating use of parameterized tests.
+
+#ifndef GTEST_SAMPLES_PRIME_TABLES_H_
+#define GTEST_SAMPLES_PRIME_TABLES_H_
+
+#include <algorithm>
+
+// The prime table interface.
+class PrimeTable {
+ public:
+ virtual ~PrimeTable() {}
+
+ // Returns true iff n is a prime number.
+ virtual bool IsPrime(int n) const = 0;
+
+ // Returns the smallest prime number greater than p; or returns -1
+ // if the next prime is beyond the capacity of the table.
+ virtual int GetNextPrime(int p) const = 0;
+};
+
+// Implementation #1 calculates the primes on-the-fly.
+class OnTheFlyPrimeTable : public PrimeTable {
+ public:
+ virtual bool IsPrime(int n) const {
+ if (n <= 1) return false;
+
+ for (int i = 2; i*i <= n; i++) {
+ // n is divisible by an integer other than 1 and itself.
+ if ((n % i) == 0) return false;
+ }
+
+ return true;
+ }
+
+ virtual int GetNextPrime(int p) const {
+ for (int n = p + 1; n > 0; n++) {
+ if (IsPrime(n)) return n;
+ }
+
+ return -1;
+ }
+};
+
+// Implementation #2 pre-calculates the primes and stores the result
+// in an array.
+class PreCalculatedPrimeTable : public PrimeTable {
+ public:
+ // 'max' specifies the maximum number the prime table holds.
+ explicit PreCalculatedPrimeTable(int max)
+ : is_prime_size_(max + 1), is_prime_(new bool[max + 1]) {
+ CalculatePrimesUpTo(max);
+ }
+ virtual ~PreCalculatedPrimeTable() { delete[] is_prime_; }
+
+ virtual bool IsPrime(int n) const {
+ return 0 <= n && n < is_prime_size_ && is_prime_[n];
+ }
+
+ virtual int GetNextPrime(int p) const {
+ for (int n = p + 1; n < is_prime_size_; n++) {
+ if (is_prime_[n]) return n;
+ }
+
+ return -1;
+ }
+
+ private:
+ void CalculatePrimesUpTo(int max) {
+ ::std::fill(is_prime_, is_prime_ + is_prime_size_, true);
+ is_prime_[0] = is_prime_[1] = false;
+
+ for (int i = 2; i <= max; i++) {
+ if (!is_prime_[i]) continue;
+
+ // Marks all multiples of i (except i itself) as non-prime.
+ for (int j = 2*i; j <= max; j += i) {
+ is_prime_[j] = false;
+ }
+ }
+ }
+
+ const int is_prime_size_;
+ bool* const is_prime_;
+
+ // Disables compiler warning "assignment operator could not be generated."
+ void operator=(const PreCalculatedPrimeTable& rhs);
+};
+
+#endif // GTEST_SAMPLES_PRIME_TABLES_H_
diff --git a/libvpx/third_party/googletest/src/samples/sample1.cc b/libvpx/third_party/googletest/src/samples/sample1.cc
new file mode 100644
index 0000000..f171e26
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample1.cc
@@ -0,0 +1,68 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "sample1.h"
+
+// Returns n! (the factorial of n). For negative n, n! is defined to be 1.
+int Factorial(int n) {
+ int result = 1;
+ for (int i = 1; i <= n; i++) {
+ result *= i;
+ }
+
+ return result;
+}
+
+// Returns true iff n is a prime number.
+bool IsPrime(int n) {
+ // Trivial case 1: small numbers
+ if (n <= 1) return false;
+
+ // Trivial case 2: even numbers
+ if (n % 2 == 0) return n == 2;
+
+ // Now, we have that n is odd and n >= 3.
+
+ // Try to divide n by every odd number i, starting from 3
+ for (int i = 3; ; i += 2) {
+ // We only have to try i up to the squre root of n
+ if (i > n/i) break;
+
+ // Now, we have i <= n/i < n.
+ // If n is divisible by i, n is not prime.
+ if (n % i == 0) return false;
+ }
+
+ // n has no integer factor in the range (1, n), and thus is prime.
+ return true;
+}
diff --git a/libvpx/third_party/googletest/src/samples/sample1.h b/libvpx/third_party/googletest/src/samples/sample1.h
new file mode 100644
index 0000000..3dfeb98
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample1.h
@@ -0,0 +1,43 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#ifndef GTEST_SAMPLES_SAMPLE1_H_
+#define GTEST_SAMPLES_SAMPLE1_H_
+
+// Returns n! (the factorial of n). For negative n, n! is defined to be 1.
+int Factorial(int n);
+
+// Returns true iff n is a prime number.
+bool IsPrime(int n);
+
+#endif // GTEST_SAMPLES_SAMPLE1_H_
diff --git a/libvpx/third_party/googletest/src/samples/sample10_unittest.cc b/libvpx/third_party/googletest/src/samples/sample10_unittest.cc
new file mode 100644
index 0000000..2813d04
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample10_unittest.cc
@@ -0,0 +1,145 @@
+// Copyright 2009 Google Inc. All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// This sample shows how to use Google Test listener API to implement
+// a primitive leak checker.
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "gtest/gtest.h"
+
+using ::testing::EmptyTestEventListener;
+using ::testing::InitGoogleTest;
+using ::testing::Test;
+using ::testing::TestCase;
+using ::testing::TestEventListeners;
+using ::testing::TestInfo;
+using ::testing::TestPartResult;
+using ::testing::UnitTest;
+
+namespace {
+
+// We will track memory used by this class.
+class Water {
+ public:
+ // Normal Water declarations go here.
+
+ // operator new and operator delete help us control water allocation.
+ void* operator new(size_t allocation_size) {
+ allocated_++;
+ return malloc(allocation_size);
+ }
+
+ void operator delete(void* block, size_t /* allocation_size */) {
+ allocated_--;
+ free(block);
+ }
+
+ static int allocated() { return allocated_; }
+
+ private:
+ static int allocated_;
+};
+
+int Water::allocated_ = 0;
+
+// This event listener monitors how many Water objects are created and
+// destroyed by each test, and reports a failure if a test leaks some Water
+// objects. It does this by comparing the number of live Water objects at
+// the beginning of a test and at the end of a test.
+class LeakChecker : public EmptyTestEventListener {
+ private:
+ // Called before a test starts.
+ virtual void OnTestStart(const TestInfo& /* test_info */) {
+ initially_allocated_ = Water::allocated();
+ }
+
+ // Called after a test ends.
+ virtual void OnTestEnd(const TestInfo& /* test_info */) {
+ int difference = Water::allocated() - initially_allocated_;
+
+ // You can generate a failure in any event handler except
+ // OnTestPartResult. Just use an appropriate Google Test assertion to do
+ // it.
+ EXPECT_TRUE(difference <= 0)
+ << "Leaked " << difference << " unit(s) of Water!";
+ }
+
+ int initially_allocated_;
+};
+
+TEST(ListenersTest, DoesNotLeak) {
+ Water* water = new Water;
+ delete water;
+}
+
+// This should fail when the --check_for_leaks command line flag is
+// specified.
+TEST(ListenersTest, LeaksWater) {
+ Water* water = new Water;
+ EXPECT_TRUE(water != NULL);
+}
+
+} // namespace
+
+int main(int argc, char **argv) {
+ InitGoogleTest(&argc, argv);
+
+ bool check_for_leaks = false;
+ if (argc > 1 && strcmp(argv[1], "--check_for_leaks") == 0 )
+ check_for_leaks = true;
+ else
+ printf("%s\n", "Run this program with --check_for_leaks to enable "
+ "custom leak checking in the tests.");
+
+ // If we are given the --check_for_leaks command line flag, installs the
+ // leak checker.
+ if (check_for_leaks) {
+ TestEventListeners& listeners = UnitTest::GetInstance()->listeners();
+
+ // Adds the leak checker to the end of the test event listener list,
+ // after the default text output printer and the default XML report
+ // generator.
+ //
+ // The order is important - it ensures that failures generated in the
+ // leak checker's OnTestEnd() method are processed by the text and XML
+ // printers *before* their OnTestEnd() methods are called, such that
+ // they are attributed to the right test. Remember that a listener
+ // receives an OnXyzStart event *after* listeners preceding it in the
+ // list received that event, and receives an OnXyzEnd event *before*
+ // listeners preceding it.
+ //
+ // We don't need to worry about deleting the new listener later, as
+ // Google Test will do it.
+ listeners.Append(new LeakChecker);
+ }
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/samples/sample1_unittest.cc b/libvpx/third_party/googletest/src/samples/sample1_unittest.cc
new file mode 100644
index 0000000..a8a7c79
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample1_unittest.cc
@@ -0,0 +1,153 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+
+// This sample shows how to write a simple unit test for a function,
+// using Google C++ testing framework.
+//
+// Writing a unit test using Google C++ testing framework is easy as 1-2-3:
+
+
+// Step 1. Include necessary header files such that the stuff your
+// test logic needs is declared.
+//
+// Don't forget gtest.h, which declares the testing framework.
+
+#include <limits.h>
+#include "sample1.h"
+#include "gtest/gtest.h"
+
+
+// Step 2. Use the TEST macro to define your tests.
+//
+// TEST has two parameters: the test case name and the test name.
+// After using the macro, you should define your test logic between a
+// pair of braces. You can use a bunch of macros to indicate the
+// success or failure of a test. EXPECT_TRUE and EXPECT_EQ are
+// examples of such macros. For a complete list, see gtest.h.
+//
+// <TechnicalDetails>
+//
+// In Google Test, tests are grouped into test cases. This is how we
+// keep test code organized. You should put logically related tests
+// into the same test case.
+//
+// The test case name and the test name should both be valid C++
+// identifiers. And you should not use underscore (_) in the names.
+//
+// Google Test guarantees that each test you define is run exactly
+// once, but it makes no guarantee on the order the tests are
+// executed. Therefore, you should write your tests in such a way
+// that their results don't depend on their order.
+//
+// </TechnicalDetails>
+
+
+// Tests Factorial().
+
+// Tests factorial of negative numbers.
+TEST(FactorialTest, Negative) {
+ // This test is named "Negative", and belongs to the "FactorialTest"
+ // test case.
+ EXPECT_EQ(1, Factorial(-5));
+ EXPECT_EQ(1, Factorial(-1));
+ EXPECT_TRUE(Factorial(-10) > 0);
+
+ // <TechnicalDetails>
+ //
+ // EXPECT_EQ(expected, actual) is the same as
+ //
+ // EXPECT_TRUE((expected) == (actual))
+ //
+ // except that it will print both the expected value and the actual
+ // value when the assertion fails. This is very helpful for
+ // debugging. Therefore in this case EXPECT_EQ is preferred.
+ //
+ // On the other hand, EXPECT_TRUE accepts any Boolean expression,
+ // and is thus more general.
+ //
+ // </TechnicalDetails>
+}
+
+// Tests factorial of 0.
+TEST(FactorialTest, Zero) {
+ EXPECT_EQ(1, Factorial(0));
+}
+
+// Tests factorial of positive numbers.
+TEST(FactorialTest, Positive) {
+ EXPECT_EQ(1, Factorial(1));
+ EXPECT_EQ(2, Factorial(2));
+ EXPECT_EQ(6, Factorial(3));
+ EXPECT_EQ(40320, Factorial(8));
+}
+
+
+// Tests IsPrime()
+
+// Tests negative input.
+TEST(IsPrimeTest, Negative) {
+ // This test belongs to the IsPrimeTest test case.
+
+ EXPECT_FALSE(IsPrime(-1));
+ EXPECT_FALSE(IsPrime(-2));
+ EXPECT_FALSE(IsPrime(INT_MIN));
+}
+
+// Tests some trivial cases.
+TEST(IsPrimeTest, Trivial) {
+ EXPECT_FALSE(IsPrime(0));
+ EXPECT_FALSE(IsPrime(1));
+ EXPECT_TRUE(IsPrime(2));
+ EXPECT_TRUE(IsPrime(3));
+}
+
+// Tests positive input.
+TEST(IsPrimeTest, Positive) {
+ EXPECT_FALSE(IsPrime(4));
+ EXPECT_TRUE(IsPrime(5));
+ EXPECT_FALSE(IsPrime(6));
+ EXPECT_TRUE(IsPrime(23));
+}
+
+// Step 3. Call RUN_ALL_TESTS() in main().
+//
+// We do this by linking in src/gtest_main.cc file, which consists of
+// a main() function which calls RUN_ALL_TESTS() for us.
+//
+// This runs all the tests you've defined, prints the result, and
+// returns 0 if successful, or 1 otherwise.
+//
+// Did you notice that we didn't register the tests? The
+// RUN_ALL_TESTS() macro magically knows about all the tests we
+// defined. Isn't this convenient?
diff --git a/libvpx/third_party/googletest/src/samples/sample2.cc b/libvpx/third_party/googletest/src/samples/sample2.cc
new file mode 100644
index 0000000..5f763b9
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample2.cc
@@ -0,0 +1,56 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "sample2.h"
+
+#include <string.h>
+
+// Clones a 0-terminated C string, allocating memory using new.
+const char* MyString::CloneCString(const char* a_c_string) {
+ if (a_c_string == NULL) return NULL;
+
+ const size_t len = strlen(a_c_string);
+ char* const clone = new char[ len + 1 ];
+ memcpy(clone, a_c_string, len + 1);
+
+ return clone;
+}
+
+// Sets the 0-terminated C string this MyString object
+// represents.
+void MyString::Set(const char* a_c_string) {
+ // Makes sure this works when c_string == c_string_
+ const char* const temp = MyString::CloneCString(a_c_string);
+ delete[] c_string_;
+ c_string_ = temp;
+}
diff --git a/libvpx/third_party/googletest/src/samples/sample2.h b/libvpx/third_party/googletest/src/samples/sample2.h
new file mode 100644
index 0000000..5b57e60
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample2.h
@@ -0,0 +1,86 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#ifndef GTEST_SAMPLES_SAMPLE2_H_
+#define GTEST_SAMPLES_SAMPLE2_H_
+
+#include <string.h>
+
+
+// A simple string class.
+class MyString {
+ private:
+ const char* c_string_;
+ const MyString& operator=(const MyString& rhs);
+
+ public:
+
+ // Clones a 0-terminated C string, allocating memory using new.
+ static const char* CloneCString(const char* a_c_string);
+
+ ////////////////////////////////////////////////////////////
+ //
+ // C'tors
+
+ // The default c'tor constructs a NULL string.
+ MyString() : c_string_(NULL) {}
+
+ // Constructs a MyString by cloning a 0-terminated C string.
+ explicit MyString(const char* a_c_string) : c_string_(NULL) {
+ Set(a_c_string);
+ }
+
+ // Copy c'tor
+ MyString(const MyString& string) : c_string_(NULL) {
+ Set(string.c_string_);
+ }
+
+ ////////////////////////////////////////////////////////////
+ //
+ // D'tor. MyString is intended to be a final class, so the d'tor
+ // doesn't need to be virtual.
+ ~MyString() { delete[] c_string_; }
+
+ // Gets the 0-terminated C string this MyString object represents.
+ const char* c_string() const { return c_string_; }
+
+ size_t Length() const {
+ return c_string_ == NULL ? 0 : strlen(c_string_);
+ }
+
+ // Sets the 0-terminated C string this MyString object represents.
+ void Set(const char* c_string);
+};
+
+
+#endif // GTEST_SAMPLES_SAMPLE2_H_
diff --git a/libvpx/third_party/googletest/src/samples/sample2_unittest.cc b/libvpx/third_party/googletest/src/samples/sample2_unittest.cc
new file mode 100644
index 0000000..3792fa5
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample2_unittest.cc
@@ -0,0 +1,109 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+
+// This sample shows how to write a more complex unit test for a class
+// that has multiple member functions.
+//
+// Usually, it's a good idea to have one test for each method in your
+// class. You don't have to do that exactly, but it helps to keep
+// your tests organized. You may also throw in additional tests as
+// needed.
+
+#include "sample2.h"
+#include "gtest/gtest.h"
+
+// In this example, we test the MyString class (a simple string).
+
+// Tests the default c'tor.
+TEST(MyString, DefaultConstructor) {
+ const MyString s;
+
+ // Asserts that s.c_string() returns NULL.
+ //
+ // <TechnicalDetails>
+ //
+ // If we write NULL instead of
+ //
+ // static_cast<const char *>(NULL)
+ //
+ // in this assertion, it will generate a warning on gcc 3.4. The
+ // reason is that EXPECT_EQ needs to know the types of its
+ // arguments in order to print them when it fails. Since NULL is
+ // #defined as 0, the compiler will use the formatter function for
+ // int to print it. However, gcc thinks that NULL should be used as
+ // a pointer, not an int, and therefore complains.
+ //
+ // The root of the problem is C++'s lack of distinction between the
+ // integer number 0 and the null pointer constant. Unfortunately,
+ // we have to live with this fact.
+ //
+ // </TechnicalDetails>
+ EXPECT_STREQ(NULL, s.c_string());
+
+ EXPECT_EQ(0u, s.Length());
+}
+
+const char kHelloString[] = "Hello, world!";
+
+// Tests the c'tor that accepts a C string.
+TEST(MyString, ConstructorFromCString) {
+ const MyString s(kHelloString);
+ EXPECT_TRUE(strcmp(s.c_string(), kHelloString) == 0);
+ EXPECT_EQ(sizeof(kHelloString)/sizeof(kHelloString[0]) - 1,
+ s.Length());
+}
+
+// Tests the copy c'tor.
+TEST(MyString, CopyConstructor) {
+ const MyString s1(kHelloString);
+ const MyString s2 = s1;
+ EXPECT_TRUE(strcmp(s2.c_string(), kHelloString) == 0);
+}
+
+// Tests the Set method.
+TEST(MyString, Set) {
+ MyString s;
+
+ s.Set(kHelloString);
+ EXPECT_TRUE(strcmp(s.c_string(), kHelloString) == 0);
+
+ // Set should work when the input pointer is the same as the one
+ // already in the MyString object.
+ s.Set(s.c_string());
+ EXPECT_TRUE(strcmp(s.c_string(), kHelloString) == 0);
+
+ // Can we set the MyString to NULL?
+ s.Set(NULL);
+ EXPECT_STREQ(NULL, s.c_string());
+}
diff --git a/libvpx/third_party/googletest/src/samples/sample3-inl.h b/libvpx/third_party/googletest/src/samples/sample3-inl.h
new file mode 100644
index 0000000..46369a0
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample3-inl.h
@@ -0,0 +1,173 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#ifndef GTEST_SAMPLES_SAMPLE3_INL_H_
+#define GTEST_SAMPLES_SAMPLE3_INL_H_
+
+#include <stddef.h>
+
+
+// Queue is a simple queue implemented as a singled-linked list.
+//
+// The element type must support copy constructor.
+template <typename E> // E is the element type
+class Queue;
+
+// QueueNode is a node in a Queue, which consists of an element of
+// type E and a pointer to the next node.
+template <typename E> // E is the element type
+class QueueNode {
+ friend class Queue<E>;
+
+ public:
+ // Gets the element in this node.
+ const E& element() const { return element_; }
+
+ // Gets the next node in the queue.
+ QueueNode* next() { return next_; }
+ const QueueNode* next() const { return next_; }
+
+ private:
+ // Creates a node with a given element value. The next pointer is
+ // set to NULL.
+ QueueNode(const E& an_element) : element_(an_element), next_(NULL) {}
+
+ // We disable the default assignment operator and copy c'tor.
+ const QueueNode& operator = (const QueueNode&);
+ QueueNode(const QueueNode&);
+
+ E element_;
+ QueueNode* next_;
+};
+
+template <typename E> // E is the element type.
+class Queue {
+public:
+
+ // Creates an empty queue.
+ Queue() : head_(NULL), last_(NULL), size_(0) {}
+
+ // D'tor. Clears the queue.
+ ~Queue() { Clear(); }
+
+ // Clears the queue.
+ void Clear() {
+ if (size_ > 0) {
+ // 1. Deletes every node.
+ QueueNode<E>* node = head_;
+ QueueNode<E>* next = node->next();
+ for (; ;) {
+ delete node;
+ node = next;
+ if (node == NULL) break;
+ next = node->next();
+ }
+
+ // 2. Resets the member variables.
+ head_ = last_ = NULL;
+ size_ = 0;
+ }
+ }
+
+ // Gets the number of elements.
+ size_t Size() const { return size_; }
+
+ // Gets the first element of the queue, or NULL if the queue is empty.
+ QueueNode<E>* Head() { return head_; }
+ const QueueNode<E>* Head() const { return head_; }
+
+ // Gets the last element of the queue, or NULL if the queue is empty.
+ QueueNode<E>* Last() { return last_; }
+ const QueueNode<E>* Last() const { return last_; }
+
+ // Adds an element to the end of the queue. A copy of the element is
+ // created using the copy constructor, and then stored in the queue.
+ // Changes made to the element in the queue doesn't affect the source
+ // object, and vice versa.
+ void Enqueue(const E& element) {
+ QueueNode<E>* new_node = new QueueNode<E>(element);
+
+ if (size_ == 0) {
+ head_ = last_ = new_node;
+ size_ = 1;
+ } else {
+ last_->next_ = new_node;
+ last_ = new_node;
+ size_++;
+ }
+ }
+
+ // Removes the head of the queue and returns it. Returns NULL if
+ // the queue is empty.
+ E* Dequeue() {
+ if (size_ == 0) {
+ return NULL;
+ }
+
+ const QueueNode<E>* const old_head = head_;
+ head_ = head_->next_;
+ size_--;
+ if (size_ == 0) {
+ last_ = NULL;
+ }
+
+ E* element = new E(old_head->element());
+ delete old_head;
+
+ return element;
+ }
+
+ // Applies a function/functor on each element of the queue, and
+ // returns the result in a new queue. The original queue is not
+ // affected.
+ template <typename F>
+ Queue* Map(F function) const {
+ Queue* new_queue = new Queue();
+ for (const QueueNode<E>* node = head_; node != NULL; node = node->next_) {
+ new_queue->Enqueue(function(node->element()));
+ }
+
+ return new_queue;
+ }
+
+ private:
+ QueueNode<E>* head_; // The first node of the queue.
+ QueueNode<E>* last_; // The last node of the queue.
+ size_t size_; // The number of elements in the queue.
+
+ // We disallow copying a queue.
+ Queue(const Queue&);
+ const Queue& operator = (const Queue&);
+ };
+
+#endif // GTEST_SAMPLES_SAMPLE3_INL_H_
diff --git a/libvpx/third_party/googletest/src/samples/sample3_unittest.cc b/libvpx/third_party/googletest/src/samples/sample3_unittest.cc
new file mode 100644
index 0000000..bf3877d
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample3_unittest.cc
@@ -0,0 +1,151 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+
+// In this example, we use a more advanced feature of Google Test called
+// test fixture.
+//
+// A test fixture is a place to hold objects and functions shared by
+// all tests in a test case. Using a test fixture avoids duplicating
+// the test code necessary to initialize and cleanup those common
+// objects for each test. It is also useful for defining sub-routines
+// that your tests need to invoke a lot.
+//
+// <TechnicalDetails>
+//
+// The tests share the test fixture in the sense of code sharing, not
+// data sharing. Each test is given its own fresh copy of the
+// fixture. You cannot expect the data modified by one test to be
+// passed on to another test, which is a bad idea.
+//
+// The reason for this design is that tests should be independent and
+// repeatable. In particular, a test should not fail as the result of
+// another test's failure. If one test depends on info produced by
+// another test, then the two tests should really be one big test.
+//
+// The macros for indicating the success/failure of a test
+// (EXPECT_TRUE, FAIL, etc) need to know what the current test is
+// (when Google Test prints the test result, it tells you which test
+// each failure belongs to). Technically, these macros invoke a
+// member function of the Test class. Therefore, you cannot use them
+// in a global function. That's why you should put test sub-routines
+// in a test fixture.
+//
+// </TechnicalDetails>
+
+#include "sample3-inl.h"
+#include "gtest/gtest.h"
+
+// To use a test fixture, derive a class from testing::Test.
+class QueueTest : public testing::Test {
+ protected: // You should make the members protected s.t. they can be
+ // accessed from sub-classes.
+
+ // virtual void SetUp() will be called before each test is run. You
+ // should define it if you need to initialize the varaibles.
+ // Otherwise, this can be skipped.
+ virtual void SetUp() {
+ q1_.Enqueue(1);
+ q2_.Enqueue(2);
+ q2_.Enqueue(3);
+ }
+
+ // virtual void TearDown() will be called after each test is run.
+ // You should define it if there is cleanup work to do. Otherwise,
+ // you don't have to provide it.
+ //
+ // virtual void TearDown() {
+ // }
+
+ // A helper function that some test uses.
+ static int Double(int n) {
+ return 2*n;
+ }
+
+ // A helper function for testing Queue::Map().
+ void MapTester(const Queue<int> * q) {
+ // Creates a new queue, where each element is twice as big as the
+ // corresponding one in q.
+ const Queue<int> * const new_q = q->Map(Double);
+
+ // Verifies that the new queue has the same size as q.
+ ASSERT_EQ(q->Size(), new_q->Size());
+
+ // Verifies the relationship between the elements of the two queues.
+ for ( const QueueNode<int> * n1 = q->Head(), * n2 = new_q->Head();
+ n1 != NULL; n1 = n1->next(), n2 = n2->next() ) {
+ EXPECT_EQ(2 * n1->element(), n2->element());
+ }
+
+ delete new_q;
+ }
+
+ // Declares the variables your tests want to use.
+ Queue<int> q0_;
+ Queue<int> q1_;
+ Queue<int> q2_;
+};
+
+// When you have a test fixture, you define a test using TEST_F
+// instead of TEST.
+
+// Tests the default c'tor.
+TEST_F(QueueTest, DefaultConstructor) {
+ // You can access data in the test fixture here.
+ EXPECT_EQ(0u, q0_.Size());
+}
+
+// Tests Dequeue().
+TEST_F(QueueTest, Dequeue) {
+ int * n = q0_.Dequeue();
+ EXPECT_TRUE(n == NULL);
+
+ n = q1_.Dequeue();
+ ASSERT_TRUE(n != NULL);
+ EXPECT_EQ(1, *n);
+ EXPECT_EQ(0u, q1_.Size());
+ delete n;
+
+ n = q2_.Dequeue();
+ ASSERT_TRUE(n != NULL);
+ EXPECT_EQ(2, *n);
+ EXPECT_EQ(1u, q2_.Size());
+ delete n;
+}
+
+// Tests the Queue::Map() function.
+TEST_F(QueueTest, Map) {
+ MapTester(&q0_);
+ MapTester(&q1_);
+ MapTester(&q2_);
+}
diff --git a/libvpx/third_party/googletest/src/samples/sample4.cc b/libvpx/third_party/googletest/src/samples/sample4.cc
new file mode 100644
index 0000000..ae44bda
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample4.cc
@@ -0,0 +1,46 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include <stdio.h>
+
+#include "sample4.h"
+
+// Returns the current counter value, and increments it.
+int Counter::Increment() {
+ return counter_++;
+}
+
+// Prints the current counter value to STDOUT.
+void Counter::Print() const {
+ printf("%d", counter_);
+}
diff --git a/libvpx/third_party/googletest/src/samples/sample4.h b/libvpx/third_party/googletest/src/samples/sample4.h
new file mode 100644
index 0000000..cd60f0d
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample4.h
@@ -0,0 +1,53 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// A sample program demonstrating using Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#ifndef GTEST_SAMPLES_SAMPLE4_H_
+#define GTEST_SAMPLES_SAMPLE4_H_
+
+// A simple monotonic counter.
+class Counter {
+ private:
+ int counter_;
+
+ public:
+ // Creates a counter that starts at 0.
+ Counter() : counter_(0) {}
+
+ // Returns the current counter value, and increments it.
+ int Increment();
+
+ // Prints the current counter value to STDOUT.
+ void Print() const;
+};
+
+#endif // GTEST_SAMPLES_SAMPLE4_H_
diff --git a/libvpx/third_party/googletest/src/samples/sample4_unittest.cc b/libvpx/third_party/googletest/src/samples/sample4_unittest.cc
new file mode 100644
index 0000000..fa5afc7
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample4_unittest.cc
@@ -0,0 +1,45 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "gtest/gtest.h"
+#include "sample4.h"
+
+// Tests the Increment() method.
+TEST(Counter, Increment) {
+ Counter c;
+
+ // EXPECT_EQ() evaluates its arguments exactly once, so they
+ // can have side effects.
+
+ EXPECT_EQ(0, c.Increment());
+ EXPECT_EQ(1, c.Increment());
+ EXPECT_EQ(2, c.Increment());
+}
diff --git a/libvpx/third_party/googletest/src/samples/sample5_unittest.cc b/libvpx/third_party/googletest/src/samples/sample5_unittest.cc
new file mode 100644
index 0000000..e7cab01
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample5_unittest.cc
@@ -0,0 +1,199 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// This sample teaches how to reuse a test fixture in multiple test
+// cases by deriving sub-fixtures from it.
+//
+// When you define a test fixture, you specify the name of the test
+// case that will use this fixture. Therefore, a test fixture can
+// be used by only one test case.
+//
+// Sometimes, more than one test cases may want to use the same or
+// slightly different test fixtures. For example, you may want to
+// make sure that all tests for a GUI library don't leak important
+// system resources like fonts and brushes. In Google Test, you do
+// this by putting the shared logic in a super (as in "super class")
+// test fixture, and then have each test case use a fixture derived
+// from this super fixture.
+
+#include <limits.h>
+#include <time.h>
+#include "sample3-inl.h"
+#include "gtest/gtest.h"
+#include "sample1.h"
+
+// In this sample, we want to ensure that every test finishes within
+// ~5 seconds. If a test takes longer to run, we consider it a
+// failure.
+//
+// We put the code for timing a test in a test fixture called
+// "QuickTest". QuickTest is intended to be the super fixture that
+// other fixtures derive from, therefore there is no test case with
+// the name "QuickTest". This is OK.
+//
+// Later, we will derive multiple test fixtures from QuickTest.
+class QuickTest : public testing::Test {
+ protected:
+ // Remember that SetUp() is run immediately before a test starts.
+ // This is a good place to record the start time.
+ virtual void SetUp() {
+ start_time_ = time(NULL);
+ }
+
+ // TearDown() is invoked immediately after a test finishes. Here we
+ // check if the test was too slow.
+ virtual void TearDown() {
+ // Gets the time when the test finishes
+ const time_t end_time = time(NULL);
+
+ // Asserts that the test took no more than ~5 seconds. Did you
+ // know that you can use assertions in SetUp() and TearDown() as
+ // well?
+ EXPECT_TRUE(end_time - start_time_ <= 5) << "The test took too long.";
+ }
+
+ // The UTC time (in seconds) when the test starts
+ time_t start_time_;
+};
+
+
+// We derive a fixture named IntegerFunctionTest from the QuickTest
+// fixture. All tests using this fixture will be automatically
+// required to be quick.
+class IntegerFunctionTest : public QuickTest {
+ // We don't need any more logic than already in the QuickTest fixture.
+ // Therefore the body is empty.
+};
+
+
+// Now we can write tests in the IntegerFunctionTest test case.
+
+// Tests Factorial()
+TEST_F(IntegerFunctionTest, Factorial) {
+ // Tests factorial of negative numbers.
+ EXPECT_EQ(1, Factorial(-5));
+ EXPECT_EQ(1, Factorial(-1));
+ EXPECT_TRUE(Factorial(-10) > 0);
+
+ // Tests factorial of 0.
+ EXPECT_EQ(1, Factorial(0));
+
+ // Tests factorial of positive numbers.
+ EXPECT_EQ(1, Factorial(1));
+ EXPECT_EQ(2, Factorial(2));
+ EXPECT_EQ(6, Factorial(3));
+ EXPECT_EQ(40320, Factorial(8));
+}
+
+
+// Tests IsPrime()
+TEST_F(IntegerFunctionTest, IsPrime) {
+ // Tests negative input.
+ EXPECT_TRUE(!IsPrime(-1));
+ EXPECT_TRUE(!IsPrime(-2));
+ EXPECT_TRUE(!IsPrime(INT_MIN));
+
+ // Tests some trivial cases.
+ EXPECT_TRUE(!IsPrime(0));
+ EXPECT_TRUE(!IsPrime(1));
+ EXPECT_TRUE(IsPrime(2));
+ EXPECT_TRUE(IsPrime(3));
+
+ // Tests positive input.
+ EXPECT_TRUE(!IsPrime(4));
+ EXPECT_TRUE(IsPrime(5));
+ EXPECT_TRUE(!IsPrime(6));
+ EXPECT_TRUE(IsPrime(23));
+}
+
+
+// The next test case (named "QueueTest") also needs to be quick, so
+// we derive another fixture from QuickTest.
+//
+// The QueueTest test fixture has some logic and shared objects in
+// addition to what's in QuickTest already. We define the additional
+// stuff inside the body of the test fixture, as usual.
+class QueueTest : public QuickTest {
+ protected:
+ virtual void SetUp() {
+ // First, we need to set up the super fixture (QuickTest).
+ QuickTest::SetUp();
+
+ // Second, some additional setup for this fixture.
+ q1_.Enqueue(1);
+ q2_.Enqueue(2);
+ q2_.Enqueue(3);
+ }
+
+ // By default, TearDown() inherits the behavior of
+ // QuickTest::TearDown(). As we have no additional cleaning work
+ // for QueueTest, we omit it here.
+ //
+ // virtual void TearDown() {
+ // QuickTest::TearDown();
+ // }
+
+ Queue<int> q0_;
+ Queue<int> q1_;
+ Queue<int> q2_;
+};
+
+
+// Now, let's write tests using the QueueTest fixture.
+
+// Tests the default constructor.
+TEST_F(QueueTest, DefaultConstructor) {
+ EXPECT_EQ(0u, q0_.Size());
+}
+
+// Tests Dequeue().
+TEST_F(QueueTest, Dequeue) {
+ int* n = q0_.Dequeue();
+ EXPECT_TRUE(n == NULL);
+
+ n = q1_.Dequeue();
+ EXPECT_TRUE(n != NULL);
+ EXPECT_EQ(1, *n);
+ EXPECT_EQ(0u, q1_.Size());
+ delete n;
+
+ n = q2_.Dequeue();
+ EXPECT_TRUE(n != NULL);
+ EXPECT_EQ(2, *n);
+ EXPECT_EQ(1u, q2_.Size());
+ delete n;
+}
+
+// If necessary, you can derive further test fixtures from a derived
+// fixture itself. For example, you can derive another fixture from
+// QueueTest. Google Test imposes no limit on how deep the hierarchy
+// can be. In practice, however, you probably don't want it to be too
+// deep as to be confusing.
diff --git a/libvpx/third_party/googletest/src/samples/sample6_unittest.cc b/libvpx/third_party/googletest/src/samples/sample6_unittest.cc
new file mode 100644
index 0000000..8f2036a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample6_unittest.cc
@@ -0,0 +1,224 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// This sample shows how to test common properties of multiple
+// implementations of the same interface (aka interface tests).
+
+// The interface and its implementations are in this header.
+#include "prime_tables.h"
+
+#include "gtest/gtest.h"
+
+// First, we define some factory functions for creating instances of
+// the implementations. You may be able to skip this step if all your
+// implementations can be constructed the same way.
+
+template <class T>
+PrimeTable* CreatePrimeTable();
+
+template <>
+PrimeTable* CreatePrimeTable<OnTheFlyPrimeTable>() {
+ return new OnTheFlyPrimeTable;
+}
+
+template <>
+PrimeTable* CreatePrimeTable<PreCalculatedPrimeTable>() {
+ return new PreCalculatedPrimeTable(10000);
+}
+
+// Then we define a test fixture class template.
+template <class T>
+class PrimeTableTest : public testing::Test {
+ protected:
+ // The ctor calls the factory function to create a prime table
+ // implemented by T.
+ PrimeTableTest() : table_(CreatePrimeTable<T>()) {}
+
+ virtual ~PrimeTableTest() { delete table_; }
+
+ // Note that we test an implementation via the base interface
+ // instead of the actual implementation class. This is important
+ // for keeping the tests close to the real world scenario, where the
+ // implementation is invoked via the base interface. It avoids
+ // got-yas where the implementation class has a method that shadows
+ // a method with the same name (but slightly different argument
+ // types) in the base interface, for example.
+ PrimeTable* const table_;
+};
+
+#if GTEST_HAS_TYPED_TEST
+
+using testing::Types;
+
+// Google Test offers two ways for reusing tests for different types.
+// The first is called "typed tests". You should use it if you
+// already know *all* the types you are gonna exercise when you write
+// the tests.
+
+// To write a typed test case, first use
+//
+// TYPED_TEST_CASE(TestCaseName, TypeList);
+//
+// to declare it and specify the type parameters. As with TEST_F,
+// TestCaseName must match the test fixture name.
+
+// The list of types we want to test.
+typedef Types<OnTheFlyPrimeTable, PreCalculatedPrimeTable> Implementations;
+
+TYPED_TEST_CASE(PrimeTableTest, Implementations);
+
+// Then use TYPED_TEST(TestCaseName, TestName) to define a typed test,
+// similar to TEST_F.
+TYPED_TEST(PrimeTableTest, ReturnsFalseForNonPrimes) {
+ // Inside the test body, you can refer to the type parameter by
+ // TypeParam, and refer to the fixture class by TestFixture. We
+ // don't need them in this example.
+
+ // Since we are in the template world, C++ requires explicitly
+ // writing 'this->' when referring to members of the fixture class.
+ // This is something you have to learn to live with.
+ EXPECT_FALSE(this->table_->IsPrime(-5));
+ EXPECT_FALSE(this->table_->IsPrime(0));
+ EXPECT_FALSE(this->table_->IsPrime(1));
+ EXPECT_FALSE(this->table_->IsPrime(4));
+ EXPECT_FALSE(this->table_->IsPrime(6));
+ EXPECT_FALSE(this->table_->IsPrime(100));
+}
+
+TYPED_TEST(PrimeTableTest, ReturnsTrueForPrimes) {
+ EXPECT_TRUE(this->table_->IsPrime(2));
+ EXPECT_TRUE(this->table_->IsPrime(3));
+ EXPECT_TRUE(this->table_->IsPrime(5));
+ EXPECT_TRUE(this->table_->IsPrime(7));
+ EXPECT_TRUE(this->table_->IsPrime(11));
+ EXPECT_TRUE(this->table_->IsPrime(131));
+}
+
+TYPED_TEST(PrimeTableTest, CanGetNextPrime) {
+ EXPECT_EQ(2, this->table_->GetNextPrime(0));
+ EXPECT_EQ(3, this->table_->GetNextPrime(2));
+ EXPECT_EQ(5, this->table_->GetNextPrime(3));
+ EXPECT_EQ(7, this->table_->GetNextPrime(5));
+ EXPECT_EQ(11, this->table_->GetNextPrime(7));
+ EXPECT_EQ(131, this->table_->GetNextPrime(128));
+}
+
+// That's it! Google Test will repeat each TYPED_TEST for each type
+// in the type list specified in TYPED_TEST_CASE. Sit back and be
+// happy that you don't have to define them multiple times.
+
+#endif // GTEST_HAS_TYPED_TEST
+
+#if GTEST_HAS_TYPED_TEST_P
+
+using testing::Types;
+
+// Sometimes, however, you don't yet know all the types that you want
+// to test when you write the tests. For example, if you are the
+// author of an interface and expect other people to implement it, you
+// might want to write a set of tests to make sure each implementation
+// conforms to some basic requirements, but you don't know what
+// implementations will be written in the future.
+//
+// How can you write the tests without committing to the type
+// parameters? That's what "type-parameterized tests" can do for you.
+// It is a bit more involved than typed tests, but in return you get a
+// test pattern that can be reused in many contexts, which is a big
+// win. Here's how you do it:
+
+// First, define a test fixture class template. Here we just reuse
+// the PrimeTableTest fixture defined earlier:
+
+template <class T>
+class PrimeTableTest2 : public PrimeTableTest<T> {
+};
+
+// Then, declare the test case. The argument is the name of the test
+// fixture, and also the name of the test case (as usual). The _P
+// suffix is for "parameterized" or "pattern".
+TYPED_TEST_CASE_P(PrimeTableTest2);
+
+// Next, use TYPED_TEST_P(TestCaseName, TestName) to define a test,
+// similar to what you do with TEST_F.
+TYPED_TEST_P(PrimeTableTest2, ReturnsFalseForNonPrimes) {
+ EXPECT_FALSE(this->table_->IsPrime(-5));
+ EXPECT_FALSE(this->table_->IsPrime(0));
+ EXPECT_FALSE(this->table_->IsPrime(1));
+ EXPECT_FALSE(this->table_->IsPrime(4));
+ EXPECT_FALSE(this->table_->IsPrime(6));
+ EXPECT_FALSE(this->table_->IsPrime(100));
+}
+
+TYPED_TEST_P(PrimeTableTest2, ReturnsTrueForPrimes) {
+ EXPECT_TRUE(this->table_->IsPrime(2));
+ EXPECT_TRUE(this->table_->IsPrime(3));
+ EXPECT_TRUE(this->table_->IsPrime(5));
+ EXPECT_TRUE(this->table_->IsPrime(7));
+ EXPECT_TRUE(this->table_->IsPrime(11));
+ EXPECT_TRUE(this->table_->IsPrime(131));
+}
+
+TYPED_TEST_P(PrimeTableTest2, CanGetNextPrime) {
+ EXPECT_EQ(2, this->table_->GetNextPrime(0));
+ EXPECT_EQ(3, this->table_->GetNextPrime(2));
+ EXPECT_EQ(5, this->table_->GetNextPrime(3));
+ EXPECT_EQ(7, this->table_->GetNextPrime(5));
+ EXPECT_EQ(11, this->table_->GetNextPrime(7));
+ EXPECT_EQ(131, this->table_->GetNextPrime(128));
+}
+
+// Type-parameterized tests involve one extra step: you have to
+// enumerate the tests you defined:
+REGISTER_TYPED_TEST_CASE_P(
+ PrimeTableTest2, // The first argument is the test case name.
+ // The rest of the arguments are the test names.
+ ReturnsFalseForNonPrimes, ReturnsTrueForPrimes, CanGetNextPrime);
+
+// At this point the test pattern is done. However, you don't have
+// any real test yet as you haven't said which types you want to run
+// the tests with.
+
+// To turn the abstract test pattern into real tests, you instantiate
+// it with a list of types. Usually the test pattern will be defined
+// in a .h file, and anyone can #include and instantiate it. You can
+// even instantiate it more than once in the same program. To tell
+// different instances apart, you give each of them a name, which will
+// become part of the test case name and can be used in test filters.
+
+// The list of types we want to test. Note that it doesn't have to be
+// defined at the time we write the TYPED_TEST_P()s.
+typedef Types<OnTheFlyPrimeTable, PreCalculatedPrimeTable>
+ PrimeTableImplementations;
+INSTANTIATE_TYPED_TEST_CASE_P(OnTheFlyAndPreCalculated, // Instance name
+ PrimeTableTest2, // Test case name
+ PrimeTableImplementations); // Type list
+
+#endif // GTEST_HAS_TYPED_TEST_P
diff --git a/libvpx/third_party/googletest/src/samples/sample7_unittest.cc b/libvpx/third_party/googletest/src/samples/sample7_unittest.cc
new file mode 100644
index 0000000..1b651a2
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample7_unittest.cc
@@ -0,0 +1,130 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// This sample shows how to test common properties of multiple
+// implementations of an interface (aka interface tests) using
+// value-parameterized tests. Each test in the test case has
+// a parameter that is an interface pointer to an implementation
+// tested.
+
+// The interface and its implementations are in this header.
+#include "prime_tables.h"
+
+#include "gtest/gtest.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+using ::testing::TestWithParam;
+using ::testing::Values;
+
+// As a general rule, to prevent a test from affecting the tests that come
+// after it, you should create and destroy the tested objects for each test
+// instead of reusing them. In this sample we will define a simple factory
+// function for PrimeTable objects. We will instantiate objects in test's
+// SetUp() method and delete them in TearDown() method.
+typedef PrimeTable* CreatePrimeTableFunc();
+
+PrimeTable* CreateOnTheFlyPrimeTable() {
+ return new OnTheFlyPrimeTable();
+}
+
+template <size_t max_precalculated>
+PrimeTable* CreatePreCalculatedPrimeTable() {
+ return new PreCalculatedPrimeTable(max_precalculated);
+}
+
+// Inside the test body, fixture constructor, SetUp(), and TearDown() you
+// can refer to the test parameter by GetParam(). In this case, the test
+// parameter is a factory function which we call in fixture's SetUp() to
+// create and store an instance of PrimeTable.
+class PrimeTableTest : public TestWithParam<CreatePrimeTableFunc*> {
+ public:
+ virtual ~PrimeTableTest() { delete table_; }
+ virtual void SetUp() { table_ = (*GetParam())(); }
+ virtual void TearDown() {
+ delete table_;
+ table_ = NULL;
+ }
+
+ protected:
+ PrimeTable* table_;
+};
+
+TEST_P(PrimeTableTest, ReturnsFalseForNonPrimes) {
+ EXPECT_FALSE(table_->IsPrime(-5));
+ EXPECT_FALSE(table_->IsPrime(0));
+ EXPECT_FALSE(table_->IsPrime(1));
+ EXPECT_FALSE(table_->IsPrime(4));
+ EXPECT_FALSE(table_->IsPrime(6));
+ EXPECT_FALSE(table_->IsPrime(100));
+}
+
+TEST_P(PrimeTableTest, ReturnsTrueForPrimes) {
+ EXPECT_TRUE(table_->IsPrime(2));
+ EXPECT_TRUE(table_->IsPrime(3));
+ EXPECT_TRUE(table_->IsPrime(5));
+ EXPECT_TRUE(table_->IsPrime(7));
+ EXPECT_TRUE(table_->IsPrime(11));
+ EXPECT_TRUE(table_->IsPrime(131));
+}
+
+TEST_P(PrimeTableTest, CanGetNextPrime) {
+ EXPECT_EQ(2, table_->GetNextPrime(0));
+ EXPECT_EQ(3, table_->GetNextPrime(2));
+ EXPECT_EQ(5, table_->GetNextPrime(3));
+ EXPECT_EQ(7, table_->GetNextPrime(5));
+ EXPECT_EQ(11, table_->GetNextPrime(7));
+ EXPECT_EQ(131, table_->GetNextPrime(128));
+}
+
+// In order to run value-parameterized tests, you need to instantiate them,
+// or bind them to a list of values which will be used as test parameters.
+// You can instantiate them in a different translation module, or even
+// instantiate them several times.
+//
+// Here, we instantiate our tests with a list of two PrimeTable object
+// factory functions:
+INSTANTIATE_TEST_CASE_P(
+ OnTheFlyAndPreCalculated,
+ PrimeTableTest,
+ Values(&CreateOnTheFlyPrimeTable, &CreatePreCalculatedPrimeTable<1000>));
+
+#else
+
+// Google Test may not support value-parameterized tests with some
+// compilers. If we use conditional compilation to compile out all
+// code referring to the gtest_main library, MSVC linker will not link
+// that library at all and consequently complain about missing entry
+// point defined in that library (fatal error LNK1561: entry point
+// must be defined). This dummy test keeps gtest_main linked in.
+TEST(DummyTest, ValueParameterizedTestsAreNotSupportedOnThisPlatform) {}
+
+#endif // GTEST_HAS_PARAM_TEST
diff --git a/libvpx/third_party/googletest/src/samples/sample8_unittest.cc b/libvpx/third_party/googletest/src/samples/sample8_unittest.cc
new file mode 100644
index 0000000..5ad2e2c
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample8_unittest.cc
@@ -0,0 +1,173 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// This sample shows how to test code relying on some global flag variables.
+// Combine() helps with generating all possible combinations of such flags,
+// and each test is given one combination as a parameter.
+
+// Use class definitions to test from this header.
+#include "prime_tables.h"
+
+#include "gtest/gtest.h"
+
+#if GTEST_HAS_COMBINE
+
+// Suppose we want to introduce a new, improved implementation of PrimeTable
+// which combines speed of PrecalcPrimeTable and versatility of
+// OnTheFlyPrimeTable (see prime_tables.h). Inside it instantiates both
+// PrecalcPrimeTable and OnTheFlyPrimeTable and uses the one that is more
+// appropriate under the circumstances. But in low memory conditions, it can be
+// told to instantiate without PrecalcPrimeTable instance at all and use only
+// OnTheFlyPrimeTable.
+class HybridPrimeTable : public PrimeTable {
+ public:
+ HybridPrimeTable(bool force_on_the_fly, int max_precalculated)
+ : on_the_fly_impl_(new OnTheFlyPrimeTable),
+ precalc_impl_(force_on_the_fly ? NULL :
+ new PreCalculatedPrimeTable(max_precalculated)),
+ max_precalculated_(max_precalculated) {}
+ virtual ~HybridPrimeTable() {
+ delete on_the_fly_impl_;
+ delete precalc_impl_;
+ }
+
+ virtual bool IsPrime(int n) const {
+ if (precalc_impl_ != NULL && n < max_precalculated_)
+ return precalc_impl_->IsPrime(n);
+ else
+ return on_the_fly_impl_->IsPrime(n);
+ }
+
+ virtual int GetNextPrime(int p) const {
+ int next_prime = -1;
+ if (precalc_impl_ != NULL && p < max_precalculated_)
+ next_prime = precalc_impl_->GetNextPrime(p);
+
+ return next_prime != -1 ? next_prime : on_the_fly_impl_->GetNextPrime(p);
+ }
+
+ private:
+ OnTheFlyPrimeTable* on_the_fly_impl_;
+ PreCalculatedPrimeTable* precalc_impl_;
+ int max_precalculated_;
+};
+
+using ::testing::TestWithParam;
+using ::testing::Bool;
+using ::testing::Values;
+using ::testing::Combine;
+
+// To test all code paths for HybridPrimeTable we must test it with numbers
+// both within and outside PreCalculatedPrimeTable's capacity and also with
+// PreCalculatedPrimeTable disabled. We do this by defining fixture which will
+// accept different combinations of parameters for instantiating a
+// HybridPrimeTable instance.
+class PrimeTableTest : public TestWithParam< ::std::tr1::tuple<bool, int> > {
+ protected:
+ virtual void SetUp() {
+ // This can be written as
+ //
+ // bool force_on_the_fly;
+ // int max_precalculated;
+ // tie(force_on_the_fly, max_precalculated) = GetParam();
+ //
+ // once the Google C++ Style Guide allows use of ::std::tr1::tie.
+ //
+ bool force_on_the_fly = ::std::tr1::get<0>(GetParam());
+ int max_precalculated = ::std::tr1::get<1>(GetParam());
+ table_ = new HybridPrimeTable(force_on_the_fly, max_precalculated);
+ }
+ virtual void TearDown() {
+ delete table_;
+ table_ = NULL;
+ }
+ HybridPrimeTable* table_;
+};
+
+TEST_P(PrimeTableTest, ReturnsFalseForNonPrimes) {
+ // Inside the test body, you can refer to the test parameter by GetParam().
+ // In this case, the test parameter is a PrimeTable interface pointer which
+ // we can use directly.
+ // Please note that you can also save it in the fixture's SetUp() method
+ // or constructor and use saved copy in the tests.
+
+ EXPECT_FALSE(table_->IsPrime(-5));
+ EXPECT_FALSE(table_->IsPrime(0));
+ EXPECT_FALSE(table_->IsPrime(1));
+ EXPECT_FALSE(table_->IsPrime(4));
+ EXPECT_FALSE(table_->IsPrime(6));
+ EXPECT_FALSE(table_->IsPrime(100));
+}
+
+TEST_P(PrimeTableTest, ReturnsTrueForPrimes) {
+ EXPECT_TRUE(table_->IsPrime(2));
+ EXPECT_TRUE(table_->IsPrime(3));
+ EXPECT_TRUE(table_->IsPrime(5));
+ EXPECT_TRUE(table_->IsPrime(7));
+ EXPECT_TRUE(table_->IsPrime(11));
+ EXPECT_TRUE(table_->IsPrime(131));
+}
+
+TEST_P(PrimeTableTest, CanGetNextPrime) {
+ EXPECT_EQ(2, table_->GetNextPrime(0));
+ EXPECT_EQ(3, table_->GetNextPrime(2));
+ EXPECT_EQ(5, table_->GetNextPrime(3));
+ EXPECT_EQ(7, table_->GetNextPrime(5));
+ EXPECT_EQ(11, table_->GetNextPrime(7));
+ EXPECT_EQ(131, table_->GetNextPrime(128));
+}
+
+// In order to run value-parameterized tests, you need to instantiate them,
+// or bind them to a list of values which will be used as test parameters.
+// You can instantiate them in a different translation module, or even
+// instantiate them several times.
+//
+// Here, we instantiate our tests with a list of parameters. We must combine
+// all variations of the boolean flag suppressing PrecalcPrimeTable and some
+// meaningful values for tests. We choose a small value (1), and a value that
+// will put some of the tested numbers beyond the capability of the
+// PrecalcPrimeTable instance and some inside it (10). Combine will produce all
+// possible combinations.
+INSTANTIATE_TEST_CASE_P(MeaningfulTestParameters,
+ PrimeTableTest,
+ Combine(Bool(), Values(1, 10)));
+
+#else
+
+// Google Test may not support Combine() with some compilers. If we
+// use conditional compilation to compile out all code referring to
+// the gtest_main library, MSVC linker will not link that library at
+// all and consequently complain about missing entry point defined in
+// that library (fatal error LNK1561: entry point must be
+// defined). This dummy test keeps gtest_main linked in.
+TEST(DummyTest, CombineIsNotSupportedOnThisPlatform) {}
+
+#endif // GTEST_HAS_COMBINE
diff --git a/libvpx/third_party/googletest/src/samples/sample9_unittest.cc b/libvpx/third_party/googletest/src/samples/sample9_unittest.cc
new file mode 100644
index 0000000..b2e2079
--- /dev/null
+++ b/libvpx/third_party/googletest/src/samples/sample9_unittest.cc
@@ -0,0 +1,160 @@
+// Copyright 2009 Google Inc. All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// This sample shows how to use Google Test listener API to implement
+// an alternative console output and how to use the UnitTest reflection API
+// to enumerate test cases and tests and to inspect their results.
+
+#include <stdio.h>
+
+#include "gtest/gtest.h"
+
+using ::testing::EmptyTestEventListener;
+using ::testing::InitGoogleTest;
+using ::testing::Test;
+using ::testing::TestCase;
+using ::testing::TestEventListeners;
+using ::testing::TestInfo;
+using ::testing::TestPartResult;
+using ::testing::UnitTest;
+
+namespace {
+
+// Provides alternative output mode which produces minimal amount of
+// information about tests.
+class TersePrinter : public EmptyTestEventListener {
+ private:
+ // Called before any test activity starts.
+ virtual void OnTestProgramStart(const UnitTest& /* unit_test */) {}
+
+ // Called after all test activities have ended.
+ virtual void OnTestProgramEnd(const UnitTest& unit_test) {
+ fprintf(stdout, "TEST %s\n", unit_test.Passed() ? "PASSED" : "FAILED");
+ fflush(stdout);
+ }
+
+ // Called before a test starts.
+ virtual void OnTestStart(const TestInfo& test_info) {
+ fprintf(stdout,
+ "*** Test %s.%s starting.\n",
+ test_info.test_case_name(),
+ test_info.name());
+ fflush(stdout);
+ }
+
+ // Called after a failed assertion or a SUCCEED() invocation.
+ virtual void OnTestPartResult(const TestPartResult& test_part_result) {
+ fprintf(stdout,
+ "%s in %s:%d\n%s\n",
+ test_part_result.failed() ? "*** Failure" : "Success",
+ test_part_result.file_name(),
+ test_part_result.line_number(),
+ test_part_result.summary());
+ fflush(stdout);
+ }
+
+ // Called after a test ends.
+ virtual void OnTestEnd(const TestInfo& test_info) {
+ fprintf(stdout,
+ "*** Test %s.%s ending.\n",
+ test_info.test_case_name(),
+ test_info.name());
+ fflush(stdout);
+ }
+}; // class TersePrinter
+
+TEST(CustomOutputTest, PrintsMessage) {
+ printf("Printing something from the test body...\n");
+}
+
+TEST(CustomOutputTest, Succeeds) {
+ SUCCEED() << "SUCCEED() has been invoked from here";
+}
+
+TEST(CustomOutputTest, Fails) {
+ EXPECT_EQ(1, 2)
+ << "This test fails in order to demonstrate alternative failure messages";
+}
+
+} // namespace
+
+int main(int argc, char **argv) {
+ InitGoogleTest(&argc, argv);
+
+ bool terse_output = false;
+ if (argc > 1 && strcmp(argv[1], "--terse_output") == 0 )
+ terse_output = true;
+ else
+ printf("%s\n", "Run this program with --terse_output to change the way "
+ "it prints its output.");
+
+ UnitTest& unit_test = *UnitTest::GetInstance();
+
+ // If we are given the --terse_output command line flag, suppresses the
+ // standard output and attaches own result printer.
+ if (terse_output) {
+ TestEventListeners& listeners = unit_test.listeners();
+
+ // Removes the default console output listener from the list so it will
+ // not receive events from Google Test and won't print any output. Since
+ // this operation transfers ownership of the listener to the caller we
+ // have to delete it as well.
+ delete listeners.Release(listeners.default_result_printer());
+
+ // Adds the custom output listener to the list. It will now receive
+ // events from Google Test and print the alternative output. We don't
+ // have to worry about deleting it since Google Test assumes ownership
+ // over it after adding it to the list.
+ listeners.Append(new TersePrinter);
+ }
+ int ret_val = RUN_ALL_TESTS();
+
+ // This is an example of using the UnitTest reflection API to inspect test
+ // results. Here we discount failures from the tests we expected to fail.
+ int unexpectedly_failed_tests = 0;
+ for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+ const TestCase& test_case = *unit_test.GetTestCase(i);
+ for (int j = 0; j < test_case.total_test_count(); ++j) {
+ const TestInfo& test_info = *test_case.GetTestInfo(j);
+ // Counts failed tests that were not meant to fail (those without
+ // 'Fails' in the name).
+ if (test_info.result()->Failed() &&
+ strcmp(test_info.name(), "Fails") != 0) {
+ unexpectedly_failed_tests++;
+ }
+ }
+ }
+
+ // Test that were meant to fail should not affect the test program outcome.
+ if (unexpectedly_failed_tests == 0)
+ ret_val = 0;
+
+ return ret_val;
+}
diff --git a/libvpx/third_party/googletest/src/scripts/fuse_gtest_files.py b/libvpx/third_party/googletest/src/scripts/fuse_gtest_files.py
new file mode 100755
index 0000000..57ef72f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/scripts/fuse_gtest_files.py
@@ -0,0 +1,250 @@
+#!/usr/bin/env python
+#
+# Copyright 2009, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""fuse_gtest_files.py v0.2.0
+Fuses Google Test source code into a .h file and a .cc file.
+
+SYNOPSIS
+ fuse_gtest_files.py [GTEST_ROOT_DIR] OUTPUT_DIR
+
+ Scans GTEST_ROOT_DIR for Google Test source code, and generates
+ two files: OUTPUT_DIR/gtest/gtest.h and OUTPUT_DIR/gtest/gtest-all.cc.
+ Then you can build your tests by adding OUTPUT_DIR to the include
+ search path and linking with OUTPUT_DIR/gtest/gtest-all.cc. These
+ two files contain everything you need to use Google Test. Hence
+ you can "install" Google Test by copying them to wherever you want.
+
+ GTEST_ROOT_DIR can be omitted and defaults to the parent
+ directory of the directory holding this script.
+
+EXAMPLES
+ ./fuse_gtest_files.py fused_gtest
+ ./fuse_gtest_files.py path/to/unpacked/gtest fused_gtest
+
+This tool is experimental. In particular, it assumes that there is no
+conditional inclusion of Google Test headers. Please report any
+problems to googletestframework@googlegroups.com. You can read
+http://code.google.com/p/googletest/wiki/GoogleTestAdvancedGuide for
+more information.
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import re
+import sets
+import sys
+
+# We assume that this file is in the scripts/ directory in the Google
+# Test root directory.
+DEFAULT_GTEST_ROOT_DIR = os.path.join(os.path.dirname(__file__), '..')
+
+# Regex for matching '#include "gtest/..."'.
+INCLUDE_GTEST_FILE_REGEX = re.compile(r'^\s*#\s*include\s*"(gtest/.+)"')
+
+# Regex for matching '#include "src/..."'.
+INCLUDE_SRC_FILE_REGEX = re.compile(r'^\s*#\s*include\s*"(src/.+)"')
+
+# Where to find the source seed files.
+GTEST_H_SEED = 'include/gtest/gtest.h'
+GTEST_SPI_H_SEED = 'include/gtest/gtest-spi.h'
+GTEST_ALL_CC_SEED = 'src/gtest-all.cc'
+
+# Where to put the generated files.
+GTEST_H_OUTPUT = 'gtest/gtest.h'
+GTEST_ALL_CC_OUTPUT = 'gtest/gtest-all.cc'
+
+
+def VerifyFileExists(directory, relative_path):
+ """Verifies that the given file exists; aborts on failure.
+
+ relative_path is the file path relative to the given directory.
+ """
+
+ if not os.path.isfile(os.path.join(directory, relative_path)):
+ print 'ERROR: Cannot find %s in directory %s.' % (relative_path,
+ directory)
+ print ('Please either specify a valid project root directory '
+ 'or omit it on the command line.')
+ sys.exit(1)
+
+
+def ValidateGTestRootDir(gtest_root):
+ """Makes sure gtest_root points to a valid gtest root directory.
+
+ The function aborts the program on failure.
+ """
+
+ VerifyFileExists(gtest_root, GTEST_H_SEED)
+ VerifyFileExists(gtest_root, GTEST_ALL_CC_SEED)
+
+
+def VerifyOutputFile(output_dir, relative_path):
+ """Verifies that the given output file path is valid.
+
+ relative_path is relative to the output_dir directory.
+ """
+
+ # Makes sure the output file either doesn't exist or can be overwritten.
+ output_file = os.path.join(output_dir, relative_path)
+ if os.path.exists(output_file):
+ # TODO(wan@google.com): The following user-interaction doesn't
+ # work with automated processes. We should provide a way for the
+ # Makefile to force overwriting the files.
+ print ('%s already exists in directory %s - overwrite it? (y/N) ' %
+ (relative_path, output_dir))
+ answer = sys.stdin.readline().strip()
+ if answer not in ['y', 'Y']:
+ print 'ABORTED.'
+ sys.exit(1)
+
+ # Makes sure the directory holding the output file exists; creates
+ # it and all its ancestors if necessary.
+ parent_directory = os.path.dirname(output_file)
+ if not os.path.isdir(parent_directory):
+ os.makedirs(parent_directory)
+
+
+def ValidateOutputDir(output_dir):
+ """Makes sure output_dir points to a valid output directory.
+
+ The function aborts the program on failure.
+ """
+
+ VerifyOutputFile(output_dir, GTEST_H_OUTPUT)
+ VerifyOutputFile(output_dir, GTEST_ALL_CC_OUTPUT)
+
+
+def FuseGTestH(gtest_root, output_dir):
+ """Scans folder gtest_root to generate gtest/gtest.h in output_dir."""
+
+ output_file = file(os.path.join(output_dir, GTEST_H_OUTPUT), 'w')
+ processed_files = sets.Set() # Holds all gtest headers we've processed.
+
+ def ProcessFile(gtest_header_path):
+ """Processes the given gtest header file."""
+
+ # We don't process the same header twice.
+ if gtest_header_path in processed_files:
+ return
+
+ processed_files.add(gtest_header_path)
+
+ # Reads each line in the given gtest header.
+ for line in file(os.path.join(gtest_root, gtest_header_path), 'r'):
+ m = INCLUDE_GTEST_FILE_REGEX.match(line)
+ if m:
+ # It's '#include "gtest/..."' - let's process it recursively.
+ ProcessFile('include/' + m.group(1))
+ else:
+ # Otherwise we copy the line unchanged to the output file.
+ output_file.write(line)
+
+ ProcessFile(GTEST_H_SEED)
+ output_file.close()
+
+
+def FuseGTestAllCcToFile(gtest_root, output_file):
+ """Scans folder gtest_root to generate gtest/gtest-all.cc in output_file."""
+
+ processed_files = sets.Set()
+
+ def ProcessFile(gtest_source_file):
+ """Processes the given gtest source file."""
+
+ # We don't process the same #included file twice.
+ if gtest_source_file in processed_files:
+ return
+
+ processed_files.add(gtest_source_file)
+
+ # Reads each line in the given gtest source file.
+ for line in file(os.path.join(gtest_root, gtest_source_file), 'r'):
+ m = INCLUDE_GTEST_FILE_REGEX.match(line)
+ if m:
+ if 'include/' + m.group(1) == GTEST_SPI_H_SEED:
+ # It's '#include "gtest/gtest-spi.h"'. This file is not
+ # #included by "gtest/gtest.h", so we need to process it.
+ ProcessFile(GTEST_SPI_H_SEED)
+ else:
+ # It's '#include "gtest/foo.h"' where foo is not gtest-spi.
+ # We treat it as '#include "gtest/gtest.h"', as all other
+ # gtest headers are being fused into gtest.h and cannot be
+ # #included directly.
+
+ # There is no need to #include "gtest/gtest.h" more than once.
+ if not GTEST_H_SEED in processed_files:
+ processed_files.add(GTEST_H_SEED)
+ output_file.write('#include "%s"\n' % (GTEST_H_OUTPUT,))
+ else:
+ m = INCLUDE_SRC_FILE_REGEX.match(line)
+ if m:
+ # It's '#include "src/foo"' - let's process it recursively.
+ ProcessFile(m.group(1))
+ else:
+ output_file.write(line)
+
+ ProcessFile(GTEST_ALL_CC_SEED)
+
+
+def FuseGTestAllCc(gtest_root, output_dir):
+ """Scans folder gtest_root to generate gtest/gtest-all.cc in output_dir."""
+
+ output_file = file(os.path.join(output_dir, GTEST_ALL_CC_OUTPUT), 'w')
+ FuseGTestAllCcToFile(gtest_root, output_file)
+ output_file.close()
+
+
+def FuseGTest(gtest_root, output_dir):
+ """Fuses gtest.h and gtest-all.cc."""
+
+ ValidateGTestRootDir(gtest_root)
+ ValidateOutputDir(output_dir)
+
+ FuseGTestH(gtest_root, output_dir)
+ FuseGTestAllCc(gtest_root, output_dir)
+
+
+def main():
+ argc = len(sys.argv)
+ if argc == 2:
+ # fuse_gtest_files.py OUTPUT_DIR
+ FuseGTest(DEFAULT_GTEST_ROOT_DIR, sys.argv[1])
+ elif argc == 3:
+ # fuse_gtest_files.py GTEST_ROOT_DIR OUTPUT_DIR
+ FuseGTest(sys.argv[1], sys.argv[2])
+ else:
+ print __doc__
+ sys.exit(1)
+
+
+if __name__ == '__main__':
+ main()
diff --git a/libvpx/third_party/googletest/src/scripts/gen_gtest_pred_impl.py b/libvpx/third_party/googletest/src/scripts/gen_gtest_pred_impl.py
new file mode 100755
index 0000000..d35b4f0
--- /dev/null
+++ b/libvpx/third_party/googletest/src/scripts/gen_gtest_pred_impl.py
@@ -0,0 +1,730 @@
+#!/usr/bin/env python
+#
+# Copyright 2006, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""gen_gtest_pred_impl.py v0.1
+
+Generates the implementation of Google Test predicate assertions and
+accompanying tests.
+
+Usage:
+
+ gen_gtest_pred_impl.py MAX_ARITY
+
+where MAX_ARITY is a positive integer.
+
+The command generates the implementation of up-to MAX_ARITY-ary
+predicate assertions, and writes it to file gtest_pred_impl.h in the
+directory where the script is. It also generates the accompanying
+unit test in file gtest_pred_impl_unittest.cc.
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import sys
+import time
+
+# Where this script is.
+SCRIPT_DIR = os.path.dirname(sys.argv[0])
+
+# Where to store the generated header.
+HEADER = os.path.join(SCRIPT_DIR, '../include/gtest/gtest_pred_impl.h')
+
+# Where to store the generated unit test.
+UNIT_TEST = os.path.join(SCRIPT_DIR, '../test/gtest_pred_impl_unittest.cc')
+
+
+def HeaderPreamble(n):
+ """Returns the preamble for the header file.
+
+ Args:
+ n: the maximum arity of the predicate macros to be generated.
+ """
+
+ # A map that defines the values used in the preamble template.
+ DEFS = {
+ 'today' : time.strftime('%m/%d/%Y'),
+ 'year' : time.strftime('%Y'),
+ 'command' : '%s %s' % (os.path.basename(sys.argv[0]), n),
+ 'n' : n
+ }
+
+ return (
+"""// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// This file is AUTOMATICALLY GENERATED on %(today)s by command
+// '%(command)s'. DO NOT EDIT BY HAND!
+//
+// Implements a family of generic predicate assertion macros.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+
+// Makes sure this header is not included before gtest.h.
+#ifndef GTEST_INCLUDE_GTEST_GTEST_H_
+#error Do not include gtest_pred_impl.h directly. Include gtest.h instead.
+#endif // GTEST_INCLUDE_GTEST_GTEST_H_
+
+// This header implements a family of generic predicate assertion
+// macros:
+//
+// ASSERT_PRED_FORMAT1(pred_format, v1)
+// ASSERT_PRED_FORMAT2(pred_format, v1, v2)
+// ...
+//
+// where pred_format is a function or functor that takes n (in the
+// case of ASSERT_PRED_FORMATn) values and their source expression
+// text, and returns a testing::AssertionResult. See the definition
+// of ASSERT_EQ in gtest.h for an example.
+//
+// If you don't care about formatting, you can use the more
+// restrictive version:
+//
+// ASSERT_PRED1(pred, v1)
+// ASSERT_PRED2(pred, v1, v2)
+// ...
+//
+// where pred is an n-ary function or functor that returns bool,
+// and the values v1, v2, ..., must support the << operator for
+// streaming to std::ostream.
+//
+// We also define the EXPECT_* variations.
+//
+// For now we only support predicates whose arity is at most %(n)s.
+// Please email googletestframework@googlegroups.com if you need
+// support for higher arities.
+
+// GTEST_ASSERT_ is the basic statement to which all of the assertions
+// in this file reduce. Don't use this in your code.
+
+#define GTEST_ASSERT_(expression, on_failure) \\
+ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \\
+ if (const ::testing::AssertionResult gtest_ar = (expression)) \\
+ ; \\
+ else \\
+ on_failure(gtest_ar.failure_message())
+""" % DEFS)
+
+
+def Arity(n):
+ """Returns the English name of the given arity."""
+
+ if n < 0:
+ return None
+ elif n <= 3:
+ return ['nullary', 'unary', 'binary', 'ternary'][n]
+ else:
+ return '%s-ary' % n
+
+
+def Title(word):
+ """Returns the given word in title case. The difference between
+ this and string's title() method is that Title('4-ary') is '4-ary'
+ while '4-ary'.title() is '4-Ary'."""
+
+ return word[0].upper() + word[1:]
+
+
+def OneTo(n):
+ """Returns the list [1, 2, 3, ..., n]."""
+
+ return range(1, n + 1)
+
+
+def Iter(n, format, sep=''):
+ """Given a positive integer n, a format string that contains 0 or
+ more '%s' format specs, and optionally a separator string, returns
+ the join of n strings, each formatted with the format string on an
+ iterator ranged from 1 to n.
+
+ Example:
+
+ Iter(3, 'v%s', sep=', ') returns 'v1, v2, v3'.
+ """
+
+ # How many '%s' specs are in format?
+ spec_count = len(format.split('%s')) - 1
+ return sep.join([format % (spec_count * (i,)) for i in OneTo(n)])
+
+
+def ImplementationForArity(n):
+ """Returns the implementation of n-ary predicate assertions."""
+
+ # A map the defines the values used in the implementation template.
+ DEFS = {
+ 'n' : str(n),
+ 'vs' : Iter(n, 'v%s', sep=', '),
+ 'vts' : Iter(n, '#v%s', sep=', '),
+ 'arity' : Arity(n),
+ 'Arity' : Title(Arity(n))
+ }
+
+ impl = """
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED%(n)s. Don't use
+// this in your code.
+template <typename Pred""" % DEFS
+
+ impl += Iter(n, """,
+ typename T%s""")
+
+ impl += """>
+AssertionResult AssertPred%(n)sHelper(const char* pred_text""" % DEFS
+
+ impl += Iter(n, """,
+ const char* e%s""")
+
+ impl += """,
+ Pred pred"""
+
+ impl += Iter(n, """,
+ const T%s& v%s""")
+
+ impl += """) {
+ if (pred(%(vs)s)) return AssertionSuccess();
+
+""" % DEFS
+
+ impl += ' return AssertionFailure() << pred_text << "("'
+
+ impl += Iter(n, """
+ << e%s""", sep=' << ", "')
+
+ impl += ' << ") evaluates to false, where"'
+
+ impl += Iter(n, """
+ << "\\n" << e%s << " evaluates to " << v%s""")
+
+ impl += """;
+}
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT%(n)s.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT%(n)s_(pred_format, %(vs)s, on_failure)\\
+ GTEST_ASSERT_(pred_format(%(vts)s, %(vs)s),\\
+ on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED%(n)s. Don't use
+// this in your code.
+#define GTEST_PRED%(n)s_(pred, %(vs)s, on_failure)\\
+ GTEST_ASSERT_(::testing::AssertPred%(n)sHelper(#pred""" % DEFS
+
+ impl += Iter(n, """, \\
+ #v%s""")
+
+ impl += """, \\
+ pred"""
+
+ impl += Iter(n, """, \\
+ v%s""")
+
+ impl += """), on_failure)
+
+// %(Arity)s predicate assertion macros.
+#define EXPECT_PRED_FORMAT%(n)s(pred_format, %(vs)s) \\
+ GTEST_PRED_FORMAT%(n)s_(pred_format, %(vs)s, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED%(n)s(pred, %(vs)s) \\
+ GTEST_PRED%(n)s_(pred, %(vs)s, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT%(n)s(pred_format, %(vs)s) \\
+ GTEST_PRED_FORMAT%(n)s_(pred_format, %(vs)s, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED%(n)s(pred, %(vs)s) \\
+ GTEST_PRED%(n)s_(pred, %(vs)s, GTEST_FATAL_FAILURE_)
+
+""" % DEFS
+
+ return impl
+
+
+def HeaderPostamble():
+ """Returns the postamble for the header file."""
+
+ return """
+
+#endif // GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+"""
+
+
+def GenerateFile(path, content):
+ """Given a file path and a content string, overwrites it with the
+ given content."""
+
+ print 'Updating file %s . . .' % path
+
+ f = file(path, 'w+')
+ print >>f, content,
+ f.close()
+
+ print 'File %s has been updated.' % path
+
+
+def GenerateHeader(n):
+ """Given the maximum arity n, updates the header file that implements
+ the predicate assertions."""
+
+ GenerateFile(HEADER,
+ HeaderPreamble(n)
+ + ''.join([ImplementationForArity(i) for i in OneTo(n)])
+ + HeaderPostamble())
+
+
+def UnitTestPreamble():
+ """Returns the preamble for the unit test file."""
+
+ # A map that defines the values used in the preamble template.
+ DEFS = {
+ 'today' : time.strftime('%m/%d/%Y'),
+ 'year' : time.strftime('%Y'),
+ 'command' : '%s %s' % (os.path.basename(sys.argv[0]), sys.argv[1]),
+ }
+
+ return (
+"""// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// This file is AUTOMATICALLY GENERATED on %(today)s by command
+// '%(command)s'. DO NOT EDIT BY HAND!
+
+// Regression test for gtest_pred_impl.h
+//
+// This file is generated by a script and quite long. If you intend to
+// learn how Google Test works by reading its unit tests, read
+// gtest_unittest.cc instead.
+//
+// This is intended as a regression test for the Google Test predicate
+// assertions. We compile it as part of the gtest_unittest target
+// only to keep the implementation tidy and compact, as it is quite
+// involved to set up the stage for testing Google Test using Google
+// Test itself.
+//
+// Currently, gtest_unittest takes ~11 seconds to run in the testing
+// daemon. In the future, if it grows too large and needs much more
+// time to finish, we should consider separating this file into a
+// stand-alone regression test.
+
+#include <iostream>
+
+#include "gtest/gtest.h"
+#include "gtest/gtest-spi.h"
+
+// A user-defined data type.
+struct Bool {
+ explicit Bool(int val) : value(val != 0) {}
+
+ bool operator>(int n) const { return value > Bool(n).value; }
+
+ Bool operator+(const Bool& rhs) const { return Bool(value + rhs.value); }
+
+ bool operator==(const Bool& rhs) const { return value == rhs.value; }
+
+ bool value;
+};
+
+// Enables Bool to be used in assertions.
+std::ostream& operator<<(std::ostream& os, const Bool& x) {
+ return os << (x.value ? "true" : "false");
+}
+
+""" % DEFS)
+
+
+def TestsForArity(n):
+ """Returns the tests for n-ary predicate assertions."""
+
+ # A map that defines the values used in the template for the tests.
+ DEFS = {
+ 'n' : n,
+ 'es' : Iter(n, 'e%s', sep=', '),
+ 'vs' : Iter(n, 'v%s', sep=', '),
+ 'vts' : Iter(n, '#v%s', sep=', '),
+ 'tvs' : Iter(n, 'T%s v%s', sep=', '),
+ 'int_vs' : Iter(n, 'int v%s', sep=', '),
+ 'Bool_vs' : Iter(n, 'Bool v%s', sep=', '),
+ 'types' : Iter(n, 'typename T%s', sep=', '),
+ 'v_sum' : Iter(n, 'v%s', sep=' + '),
+ 'arity' : Arity(n),
+ 'Arity' : Title(Arity(n)),
+ }
+
+ tests = (
+"""// Sample functions/functors for testing %(arity)s predicate assertions.
+
+// A %(arity)s predicate function.
+template <%(types)s>
+bool PredFunction%(n)s(%(tvs)s) {
+ return %(v_sum)s > 0;
+}
+
+// The following two functions are needed to circumvent a bug in
+// gcc 2.95.3, which sometimes has problem with the above template
+// function.
+bool PredFunction%(n)sInt(%(int_vs)s) {
+ return %(v_sum)s > 0;
+}
+bool PredFunction%(n)sBool(%(Bool_vs)s) {
+ return %(v_sum)s > 0;
+}
+""" % DEFS)
+
+ tests += """
+// A %(arity)s predicate functor.
+struct PredFunctor%(n)s {
+ template <%(types)s>
+ bool operator()(""" % DEFS
+
+ tests += Iter(n, 'const T%s& v%s', sep=""",
+ """)
+
+ tests += """) {
+ return %(v_sum)s > 0;
+ }
+};
+""" % DEFS
+
+ tests += """
+// A %(arity)s predicate-formatter function.
+template <%(types)s>
+testing::AssertionResult PredFormatFunction%(n)s(""" % DEFS
+
+ tests += Iter(n, 'const char* e%s', sep=""",
+ """)
+
+ tests += Iter(n, """,
+ const T%s& v%s""")
+
+ tests += """) {
+ if (PredFunction%(n)s(%(vs)s))
+ return testing::AssertionSuccess();
+
+ return testing::AssertionFailure()
+ << """ % DEFS
+
+ tests += Iter(n, 'e%s', sep=' << " + " << ')
+
+ tests += """
+ << " is expected to be positive, but evaluates to "
+ << %(v_sum)s << ".";
+}
+""" % DEFS
+
+ tests += """
+// A %(arity)s predicate-formatter functor.
+struct PredFormatFunctor%(n)s {
+ template <%(types)s>
+ testing::AssertionResult operator()(""" % DEFS
+
+ tests += Iter(n, 'const char* e%s', sep=""",
+ """)
+
+ tests += Iter(n, """,
+ const T%s& v%s""")
+
+ tests += """) const {
+ return PredFormatFunction%(n)s(%(es)s, %(vs)s);
+ }
+};
+""" % DEFS
+
+ tests += """
+// Tests for {EXPECT|ASSERT}_PRED_FORMAT%(n)s.
+
+class Predicate%(n)sTest : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ expected_to_finish_ = true;
+ finished_ = false;""" % DEFS
+
+ tests += """
+ """ + Iter(n, 'n%s_ = ') + """0;
+ }
+"""
+
+ tests += """
+ virtual void TearDown() {
+ // Verifies that each of the predicate's arguments was evaluated
+ // exactly once."""
+
+ tests += ''.join(["""
+ EXPECT_EQ(1, n%s_) <<
+ "The predicate assertion didn't evaluate argument %s "
+ "exactly once.";""" % (i, i + 1) for i in OneTo(n)])
+
+ tests += """
+
+ // Verifies that the control flow in the test function is expected.
+ if (expected_to_finish_ && !finished_) {
+ FAIL() << "The predicate assertion unexpactedly aborted the test.";
+ } else if (!expected_to_finish_ && finished_) {
+ FAIL() << "The failed predicate assertion didn't abort the test "
+ "as expected.";
+ }
+ }
+
+ // true iff the test function is expected to run to finish.
+ static bool expected_to_finish_;
+
+ // true iff the test function did run to finish.
+ static bool finished_;
+""" % DEFS
+
+ tests += Iter(n, """
+ static int n%s_;""")
+
+ tests += """
+};
+
+bool Predicate%(n)sTest::expected_to_finish_;
+bool Predicate%(n)sTest::finished_;
+""" % DEFS
+
+ tests += Iter(n, """int Predicate%%(n)sTest::n%s_;
+""") % DEFS
+
+ tests += """
+typedef Predicate%(n)sTest EXPECT_PRED_FORMAT%(n)sTest;
+typedef Predicate%(n)sTest ASSERT_PRED_FORMAT%(n)sTest;
+typedef Predicate%(n)sTest EXPECT_PRED%(n)sTest;
+typedef Predicate%(n)sTest ASSERT_PRED%(n)sTest;
+""" % DEFS
+
+ def GenTest(use_format, use_assert, expect_failure,
+ use_functor, use_user_type):
+ """Returns the test for a predicate assertion macro.
+
+ Args:
+ use_format: true iff the assertion is a *_PRED_FORMAT*.
+ use_assert: true iff the assertion is a ASSERT_*.
+ expect_failure: true iff the assertion is expected to fail.
+ use_functor: true iff the first argument of the assertion is
+ a functor (as opposed to a function)
+ use_user_type: true iff the predicate functor/function takes
+ argument(s) of a user-defined type.
+
+ Example:
+
+ GenTest(1, 0, 0, 1, 0) returns a test that tests the behavior
+ of a successful EXPECT_PRED_FORMATn() that takes a functor
+ whose arguments have built-in types."""
+
+ if use_assert:
+ assrt = 'ASSERT' # 'assert' is reserved, so we cannot use
+ # that identifier here.
+ else:
+ assrt = 'EXPECT'
+
+ assertion = assrt + '_PRED'
+
+ if use_format:
+ pred_format = 'PredFormat'
+ assertion += '_FORMAT'
+ else:
+ pred_format = 'Pred'
+
+ assertion += '%(n)s' % DEFS
+
+ if use_functor:
+ pred_format_type = 'functor'
+ pred_format += 'Functor%(n)s()'
+ else:
+ pred_format_type = 'function'
+ pred_format += 'Function%(n)s'
+ if not use_format:
+ if use_user_type:
+ pred_format += 'Bool'
+ else:
+ pred_format += 'Int'
+
+ test_name = pred_format_type.title()
+
+ if use_user_type:
+ arg_type = 'user-defined type (Bool)'
+ test_name += 'OnUserType'
+ if expect_failure:
+ arg = 'Bool(n%s_++)'
+ else:
+ arg = 'Bool(++n%s_)'
+ else:
+ arg_type = 'built-in type (int)'
+ test_name += 'OnBuiltInType'
+ if expect_failure:
+ arg = 'n%s_++'
+ else:
+ arg = '++n%s_'
+
+ if expect_failure:
+ successful_or_failed = 'failed'
+ expected_or_not = 'expected.'
+ test_name += 'Failure'
+ else:
+ successful_or_failed = 'successful'
+ expected_or_not = 'UNEXPECTED!'
+ test_name += 'Success'
+
+ # A map that defines the values used in the test template.
+ defs = DEFS.copy()
+ defs.update({
+ 'assert' : assrt,
+ 'assertion' : assertion,
+ 'test_name' : test_name,
+ 'pf_type' : pred_format_type,
+ 'pf' : pred_format,
+ 'arg_type' : arg_type,
+ 'arg' : arg,
+ 'successful' : successful_or_failed,
+ 'expected' : expected_or_not,
+ })
+
+ test = """
+// Tests a %(successful)s %(assertion)s where the
+// predicate-formatter is a %(pf_type)s on a %(arg_type)s.
+TEST_F(%(assertion)sTest, %(test_name)s) {""" % defs
+
+ indent = (len(assertion) + 3)*' '
+ extra_indent = ''
+
+ if expect_failure:
+ extra_indent = ' '
+ if use_assert:
+ test += """
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT"""
+ else:
+ test += """
+ EXPECT_NONFATAL_FAILURE({ // NOLINT"""
+
+ test += '\n' + extra_indent + """ %(assertion)s(%(pf)s""" % defs
+
+ test = test % defs
+ test += Iter(n, ',\n' + indent + extra_indent + '%(arg)s' % defs)
+ test += ');\n' + extra_indent + ' finished_ = true;\n'
+
+ if expect_failure:
+ test += ' }, "");\n'
+
+ test += '}\n'
+ return test
+
+ # Generates tests for all 2**6 = 64 combinations.
+ tests += ''.join([GenTest(use_format, use_assert, expect_failure,
+ use_functor, use_user_type)
+ for use_format in [0, 1]
+ for use_assert in [0, 1]
+ for expect_failure in [0, 1]
+ for use_functor in [0, 1]
+ for use_user_type in [0, 1]
+ ])
+
+ return tests
+
+
+def UnitTestPostamble():
+ """Returns the postamble for the tests."""
+
+ return ''
+
+
+def GenerateUnitTest(n):
+ """Returns the tests for up-to n-ary predicate assertions."""
+
+ GenerateFile(UNIT_TEST,
+ UnitTestPreamble()
+ + ''.join([TestsForArity(i) for i in OneTo(n)])
+ + UnitTestPostamble())
+
+
+def _Main():
+ """The entry point of the script. Generates the header file and its
+ unit test."""
+
+ if len(sys.argv) != 2:
+ print __doc__
+ print 'Author: ' + __author__
+ sys.exit(1)
+
+ n = int(sys.argv[1])
+ GenerateHeader(n)
+ GenerateUnitTest(n)
+
+
+if __name__ == '__main__':
+ _Main()
diff --git a/libvpx/third_party/googletest/src/scripts/gtest-config.in b/libvpx/third_party/googletest/src/scripts/gtest-config.in
new file mode 100644
index 0000000..9c72638
--- /dev/null
+++ b/libvpx/third_party/googletest/src/scripts/gtest-config.in
@@ -0,0 +1,274 @@
+#!/bin/sh
+
+# These variables are automatically filled in by the configure script.
+name="@PACKAGE_TARNAME@"
+version="@PACKAGE_VERSION@"
+
+show_usage()
+{
+ echo "Usage: gtest-config [OPTIONS...]"
+}
+
+show_help()
+{
+ show_usage
+ cat <<\EOF
+
+The `gtest-config' script provides access to the necessary compile and linking
+flags to connect with Google C++ Testing Framework, both in a build prior to
+installation, and on the system proper after installation. The installation
+overrides may be issued in combination with any other queries, but will only
+affect installation queries if called on a built but not installed gtest. The
+installation queries may not be issued with any other types of queries, and
+only one installation query may be made at a time. The version queries and
+compiler flag queries may be combined as desired but not mixed. Different
+version queries are always combined with logical "and" semantics, and only the
+last of any particular query is used while all previous ones ignored. All
+versions must be specified as a sequence of numbers separated by periods.
+Compiler flag queries output the union of the sets of flags when combined.
+
+ Examples:
+ gtest-config --min-version=1.0 || echo "Insufficient Google Test version."
+
+ g++ $(gtest-config --cppflags --cxxflags) -o foo.o -c foo.cpp
+ g++ $(gtest-config --ldflags --libs) -o foo foo.o
+
+ # When using a built but not installed Google Test:
+ g++ $(../../my_gtest_build/scripts/gtest-config ...) ...
+
+ # When using an installed Google Test, but with installation overrides:
+ export GTEST_PREFIX="/opt"
+ g++ $(gtest-config --libdir="/opt/lib64" ...) ...
+
+ Help:
+ --usage brief usage information
+ --help display this help message
+
+ Installation Overrides:
+ --prefix=<dir> overrides the installation prefix
+ --exec-prefix=<dir> overrides the executable installation prefix
+ --libdir=<dir> overrides the library installation prefix
+ --includedir=<dir> overrides the header file installation prefix
+
+ Installation Queries:
+ --prefix installation prefix
+ --exec-prefix executable installation prefix
+ --libdir library installation directory
+ --includedir header file installation directory
+ --version the version of the Google Test installation
+
+ Version Queries:
+ --min-version=VERSION return 0 if the version is at least VERSION
+ --exact-version=VERSION return 0 if the version is exactly VERSION
+ --max-version=VERSION return 0 if the version is at most VERSION
+
+ Compilation Flag Queries:
+ --cppflags compile flags specific to the C-like preprocessors
+ --cxxflags compile flags appropriate for C++ programs
+ --ldflags linker flags
+ --libs libraries for linking
+
+EOF
+}
+
+# This function bounds our version with a min and a max. It uses some clever
+# POSIX-compliant variable expansion to portably do all the work in the shell
+# and avoid any dependency on a particular "sed" or "awk" implementation.
+# Notable is that it will only ever compare the first 3 components of versions.
+# Further components will be cleanly stripped off. All versions must be
+# unadorned, so "v1.0" will *not* work. The minimum version must be in $1, and
+# the max in $2. TODO(chandlerc@google.com): If this ever breaks, we should
+# investigate expanding this via autom4te from AS_VERSION_COMPARE rather than
+# continuing to maintain our own shell version.
+check_versions()
+{
+ major_version=${version%%.*}
+ minor_version="0"
+ point_version="0"
+ if test "${version#*.}" != "${version}"; then
+ minor_version=${version#*.}
+ minor_version=${minor_version%%.*}
+ fi
+ if test "${version#*.*.}" != "${version}"; then
+ point_version=${version#*.*.}
+ point_version=${point_version%%.*}
+ fi
+
+ min_version="$1"
+ min_major_version=${min_version%%.*}
+ min_minor_version="0"
+ min_point_version="0"
+ if test "${min_version#*.}" != "${min_version}"; then
+ min_minor_version=${min_version#*.}
+ min_minor_version=${min_minor_version%%.*}
+ fi
+ if test "${min_version#*.*.}" != "${min_version}"; then
+ min_point_version=${min_version#*.*.}
+ min_point_version=${min_point_version%%.*}
+ fi
+
+ max_version="$2"
+ max_major_version=${max_version%%.*}
+ max_minor_version="0"
+ max_point_version="0"
+ if test "${max_version#*.}" != "${max_version}"; then
+ max_minor_version=${max_version#*.}
+ max_minor_version=${max_minor_version%%.*}
+ fi
+ if test "${max_version#*.*.}" != "${max_version}"; then
+ max_point_version=${max_version#*.*.}
+ max_point_version=${max_point_version%%.*}
+ fi
+
+ test $(($major_version)) -lt $(($min_major_version)) && exit 1
+ if test $(($major_version)) -eq $(($min_major_version)); then
+ test $(($minor_version)) -lt $(($min_minor_version)) && exit 1
+ if test $(($minor_version)) -eq $(($min_minor_version)); then
+ test $(($point_version)) -lt $(($min_point_version)) && exit 1
+ fi
+ fi
+
+ test $(($major_version)) -gt $(($max_major_version)) && exit 1
+ if test $(($major_version)) -eq $(($max_major_version)); then
+ test $(($minor_version)) -gt $(($max_minor_version)) && exit 1
+ if test $(($minor_version)) -eq $(($max_minor_version)); then
+ test $(($point_version)) -gt $(($max_point_version)) && exit 1
+ fi
+ fi
+
+ exit 0
+}
+
+# Show the usage line when no arguments are specified.
+if test $# -eq 0; then
+ show_usage
+ exit 1
+fi
+
+while test $# -gt 0; do
+ case $1 in
+ --usage) show_usage; exit 0;;
+ --help) show_help; exit 0;;
+
+ # Installation overrides
+ --prefix=*) GTEST_PREFIX=${1#--prefix=};;
+ --exec-prefix=*) GTEST_EXEC_PREFIX=${1#--exec-prefix=};;
+ --libdir=*) GTEST_LIBDIR=${1#--libdir=};;
+ --includedir=*) GTEST_INCLUDEDIR=${1#--includedir=};;
+
+ # Installation queries
+ --prefix|--exec-prefix|--libdir|--includedir|--version)
+ if test -n "${do_query}"; then
+ show_usage
+ exit 1
+ fi
+ do_query=${1#--}
+ ;;
+
+ # Version checking
+ --min-version=*)
+ do_check_versions=yes
+ min_version=${1#--min-version=}
+ ;;
+ --max-version=*)
+ do_check_versions=yes
+ max_version=${1#--max-version=}
+ ;;
+ --exact-version=*)
+ do_check_versions=yes
+ exact_version=${1#--exact-version=}
+ ;;
+
+ # Compiler flag output
+ --cppflags) echo_cppflags=yes;;
+ --cxxflags) echo_cxxflags=yes;;
+ --ldflags) echo_ldflags=yes;;
+ --libs) echo_libs=yes;;
+
+ # Everything else is an error
+ *) show_usage; exit 1;;
+ esac
+ shift
+done
+
+# These have defaults filled in by the configure script but can also be
+# overridden by environment variables or command line parameters.
+prefix="${GTEST_PREFIX:-@prefix@}"
+exec_prefix="${GTEST_EXEC_PREFIX:-@exec_prefix@}"
+libdir="${GTEST_LIBDIR:-@libdir@}"
+includedir="${GTEST_INCLUDEDIR:-@includedir@}"
+
+# We try and detect if our binary is not located at its installed location. If
+# it's not, we provide variables pointing to the source and build tree rather
+# than to the install tree. This allows building against a just-built gtest
+# rather than an installed gtest.
+bindir="@bindir@"
+this_relative_bindir=`dirname $0`
+this_bindir=`cd ${this_relative_bindir}; pwd -P`
+if test "${this_bindir}" = "${this_bindir%${bindir}}"; then
+ # The path to the script doesn't end in the bindir sequence from Autoconf,
+ # assume that we are in a build tree.
+ build_dir=`dirname ${this_bindir}`
+ src_dir=`cd ${this_bindir}/@top_srcdir@; pwd -P`
+
+ # TODO(chandlerc@google.com): This is a dangerous dependency on libtool, we
+ # should work to remove it, and/or remove libtool altogether, replacing it
+ # with direct references to the library and a link path.
+ gtest_libs="${build_dir}/lib/libgtest.la @PTHREAD_CFLAGS@ @PTHREAD_LIBS@"
+ gtest_ldflags=""
+
+ # We provide hooks to include from either the source or build dir, where the
+ # build dir is always preferred. This will potentially allow us to write
+ # build rules for generated headers and have them automatically be preferred
+ # over provided versions.
+ gtest_cppflags="-I${build_dir}/include -I${src_dir}/include"
+ gtest_cxxflags="@PTHREAD_CFLAGS@"
+else
+ # We're using an installed gtest, although it may be staged under some
+ # prefix. Assume (as our own libraries do) that we can resolve the prefix,
+ # and are present in the dynamic link paths.
+ gtest_ldflags="-L${libdir}"
+ gtest_libs="-l${name} @PTHREAD_CFLAGS@ @PTHREAD_LIBS@"
+ gtest_cppflags="-I${includedir}"
+ gtest_cxxflags="@PTHREAD_CFLAGS@"
+fi
+
+# Do an installation query if requested.
+if test -n "$do_query"; then
+ case $do_query in
+ prefix) echo $prefix; exit 0;;
+ exec-prefix) echo $exec_prefix; exit 0;;
+ libdir) echo $libdir; exit 0;;
+ includedir) echo $includedir; exit 0;;
+ version) echo $version; exit 0;;
+ *) show_usage; exit 1;;
+ esac
+fi
+
+# Do a version check if requested.
+if test "$do_check_versions" = "yes"; then
+ # Make sure we didn't receive a bad combination of parameters.
+ test "$echo_cppflags" = "yes" && show_usage && exit 1
+ test "$echo_cxxflags" = "yes" && show_usage && exit 1
+ test "$echo_ldflags" = "yes" && show_usage && exit 1
+ test "$echo_libs" = "yes" && show_usage && exit 1
+
+ if test "$exact_version" != ""; then
+ check_versions $exact_version $exact_version
+ # unreachable
+ else
+ check_versions ${min_version:-0.0.0} ${max_version:-9999.9999.9999}
+ # unreachable
+ fi
+fi
+
+# Do the output in the correct order so that these can be used in-line of
+# a compiler invocation.
+output=""
+test "$echo_cppflags" = "yes" && output="$output $gtest_cppflags"
+test "$echo_cxxflags" = "yes" && output="$output $gtest_cxxflags"
+test "$echo_ldflags" = "yes" && output="$output $gtest_ldflags"
+test "$echo_libs" = "yes" && output="$output $gtest_libs"
+echo $output
+
+exit 0
diff --git a/libvpx/third_party/googletest/src/scripts/pump.py b/libvpx/third_party/googletest/src/scripts/pump.py
new file mode 100755
index 0000000..8afe808
--- /dev/null
+++ b/libvpx/third_party/googletest/src/scripts/pump.py
@@ -0,0 +1,847 @@
+#!/usr/bin/env python
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""pump v0.2.0 - Pretty Useful for Meta Programming.
+
+A tool for preprocessor meta programming. Useful for generating
+repetitive boilerplate code. Especially useful for writing C++
+classes, functions, macros, and templates that need to work with
+various number of arguments.
+
+USAGE:
+ pump.py SOURCE_FILE
+
+EXAMPLES:
+ pump.py foo.cc.pump
+ Converts foo.cc.pump to foo.cc.
+
+GRAMMAR:
+ CODE ::= ATOMIC_CODE*
+ ATOMIC_CODE ::= $var ID = EXPRESSION
+ | $var ID = [[ CODE ]]
+ | $range ID EXPRESSION..EXPRESSION
+ | $for ID SEPARATOR [[ CODE ]]
+ | $($)
+ | $ID
+ | $(EXPRESSION)
+ | $if EXPRESSION [[ CODE ]] ELSE_BRANCH
+ | [[ CODE ]]
+ | RAW_CODE
+ SEPARATOR ::= RAW_CODE | EMPTY
+ ELSE_BRANCH ::= $else [[ CODE ]]
+ | $elif EXPRESSION [[ CODE ]] ELSE_BRANCH
+ | EMPTY
+ EXPRESSION has Python syntax.
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import re
+import sys
+
+
+TOKEN_TABLE = [
+ (re.compile(r'\$var\s+'), '$var'),
+ (re.compile(r'\$elif\s+'), '$elif'),
+ (re.compile(r'\$else\s+'), '$else'),
+ (re.compile(r'\$for\s+'), '$for'),
+ (re.compile(r'\$if\s+'), '$if'),
+ (re.compile(r'\$range\s+'), '$range'),
+ (re.compile(r'\$[_A-Za-z]\w*'), '$id'),
+ (re.compile(r'\$\(\$\)'), '$($)'),
+ (re.compile(r'\$'), '$'),
+ (re.compile(r'\[\[\n?'), '[['),
+ (re.compile(r'\]\]\n?'), ']]'),
+ ]
+
+
+class Cursor:
+ """Represents a position (line and column) in a text file."""
+
+ def __init__(self, line=-1, column=-1):
+ self.line = line
+ self.column = column
+
+ def __eq__(self, rhs):
+ return self.line == rhs.line and self.column == rhs.column
+
+ def __ne__(self, rhs):
+ return not self == rhs
+
+ def __lt__(self, rhs):
+ return self.line < rhs.line or (
+ self.line == rhs.line and self.column < rhs.column)
+
+ def __le__(self, rhs):
+ return self < rhs or self == rhs
+
+ def __gt__(self, rhs):
+ return rhs < self
+
+ def __ge__(self, rhs):
+ return rhs <= self
+
+ def __str__(self):
+ if self == Eof():
+ return 'EOF'
+ else:
+ return '%s(%s)' % (self.line + 1, self.column)
+
+ def __add__(self, offset):
+ return Cursor(self.line, self.column + offset)
+
+ def __sub__(self, offset):
+ return Cursor(self.line, self.column - offset)
+
+ def Clone(self):
+ """Returns a copy of self."""
+
+ return Cursor(self.line, self.column)
+
+
+# Special cursor to indicate the end-of-file.
+def Eof():
+ """Returns the special cursor to denote the end-of-file."""
+ return Cursor(-1, -1)
+
+
+class Token:
+ """Represents a token in a Pump source file."""
+
+ def __init__(self, start=None, end=None, value=None, token_type=None):
+ if start is None:
+ self.start = Eof()
+ else:
+ self.start = start
+ if end is None:
+ self.end = Eof()
+ else:
+ self.end = end
+ self.value = value
+ self.token_type = token_type
+
+ def __str__(self):
+ return 'Token @%s: \'%s\' type=%s' % (
+ self.start, self.value, self.token_type)
+
+ def Clone(self):
+ """Returns a copy of self."""
+
+ return Token(self.start.Clone(), self.end.Clone(), self.value,
+ self.token_type)
+
+
+def StartsWith(lines, pos, string):
+ """Returns True iff the given position in lines starts with 'string'."""
+
+ return lines[pos.line][pos.column:].startswith(string)
+
+
+def FindFirstInLine(line, token_table):
+ best_match_start = -1
+ for (regex, token_type) in token_table:
+ m = regex.search(line)
+ if m:
+ # We found regex in lines
+ if best_match_start < 0 or m.start() < best_match_start:
+ best_match_start = m.start()
+ best_match_length = m.end() - m.start()
+ best_match_token_type = token_type
+
+ if best_match_start < 0:
+ return None
+
+ return (best_match_start, best_match_length, best_match_token_type)
+
+
+def FindFirst(lines, token_table, cursor):
+ """Finds the first occurrence of any string in strings in lines."""
+
+ start = cursor.Clone()
+ cur_line_number = cursor.line
+ for line in lines[start.line:]:
+ if cur_line_number == start.line:
+ line = line[start.column:]
+ m = FindFirstInLine(line, token_table)
+ if m:
+ # We found a regex in line.
+ (start_column, length, token_type) = m
+ if cur_line_number == start.line:
+ start_column += start.column
+ found_start = Cursor(cur_line_number, start_column)
+ found_end = found_start + length
+ return MakeToken(lines, found_start, found_end, token_type)
+ cur_line_number += 1
+ # We failed to find str in lines
+ return None
+
+
+def SubString(lines, start, end):
+ """Returns a substring in lines."""
+
+ if end == Eof():
+ end = Cursor(len(lines) - 1, len(lines[-1]))
+
+ if start >= end:
+ return ''
+
+ if start.line == end.line:
+ return lines[start.line][start.column:end.column]
+
+ result_lines = ([lines[start.line][start.column:]] +
+ lines[start.line + 1:end.line] +
+ [lines[end.line][:end.column]])
+ return ''.join(result_lines)
+
+
+def StripMetaComments(str):
+ """Strip meta comments from each line in the given string."""
+
+ # First, completely remove lines containing nothing but a meta
+ # comment, including the trailing \n.
+ str = re.sub(r'^\s*\$\$.*\n', '', str)
+
+ # Then, remove meta comments from contentful lines.
+ return re.sub(r'\s*\$\$.*', '', str)
+
+
+def MakeToken(lines, start, end, token_type):
+ """Creates a new instance of Token."""
+
+ return Token(start, end, SubString(lines, start, end), token_type)
+
+
+def ParseToken(lines, pos, regex, token_type):
+ line = lines[pos.line][pos.column:]
+ m = regex.search(line)
+ if m and not m.start():
+ return MakeToken(lines, pos, pos + m.end(), token_type)
+ else:
+ print 'ERROR: %s expected at %s.' % (token_type, pos)
+ sys.exit(1)
+
+
+ID_REGEX = re.compile(r'[_A-Za-z]\w*')
+EQ_REGEX = re.compile(r'=')
+REST_OF_LINE_REGEX = re.compile(r'.*?(?=$|\$\$)')
+OPTIONAL_WHITE_SPACES_REGEX = re.compile(r'\s*')
+WHITE_SPACE_REGEX = re.compile(r'\s')
+DOT_DOT_REGEX = re.compile(r'\.\.')
+
+
+def Skip(lines, pos, regex):
+ line = lines[pos.line][pos.column:]
+ m = re.search(regex, line)
+ if m and not m.start():
+ return pos + m.end()
+ else:
+ return pos
+
+
+def SkipUntil(lines, pos, regex, token_type):
+ line = lines[pos.line][pos.column:]
+ m = re.search(regex, line)
+ if m:
+ return pos + m.start()
+ else:
+ print ('ERROR: %s expected on line %s after column %s.' %
+ (token_type, pos.line + 1, pos.column))
+ sys.exit(1)
+
+
+def ParseExpTokenInParens(lines, pos):
+ def ParseInParens(pos):
+ pos = Skip(lines, pos, OPTIONAL_WHITE_SPACES_REGEX)
+ pos = Skip(lines, pos, r'\(')
+ pos = Parse(pos)
+ pos = Skip(lines, pos, r'\)')
+ return pos
+
+ def Parse(pos):
+ pos = SkipUntil(lines, pos, r'\(|\)', ')')
+ if SubString(lines, pos, pos + 1) == '(':
+ pos = Parse(pos + 1)
+ pos = Skip(lines, pos, r'\)')
+ return Parse(pos)
+ else:
+ return pos
+
+ start = pos.Clone()
+ pos = ParseInParens(pos)
+ return MakeToken(lines, start, pos, 'exp')
+
+
+def RStripNewLineFromToken(token):
+ if token.value.endswith('\n'):
+ return Token(token.start, token.end, token.value[:-1], token.token_type)
+ else:
+ return token
+
+
+def TokenizeLines(lines, pos):
+ while True:
+ found = FindFirst(lines, TOKEN_TABLE, pos)
+ if not found:
+ yield MakeToken(lines, pos, Eof(), 'code')
+ return
+
+ if found.start == pos:
+ prev_token = None
+ prev_token_rstripped = None
+ else:
+ prev_token = MakeToken(lines, pos, found.start, 'code')
+ prev_token_rstripped = RStripNewLineFromToken(prev_token)
+
+ if found.token_type == '$var':
+ if prev_token_rstripped:
+ yield prev_token_rstripped
+ yield found
+ id_token = ParseToken(lines, found.end, ID_REGEX, 'id')
+ yield id_token
+ pos = Skip(lines, id_token.end, OPTIONAL_WHITE_SPACES_REGEX)
+
+ eq_token = ParseToken(lines, pos, EQ_REGEX, '=')
+ yield eq_token
+ pos = Skip(lines, eq_token.end, r'\s*')
+
+ if SubString(lines, pos, pos + 2) != '[[':
+ exp_token = ParseToken(lines, pos, REST_OF_LINE_REGEX, 'exp')
+ yield exp_token
+ pos = Cursor(exp_token.end.line + 1, 0)
+ elif found.token_type == '$for':
+ if prev_token_rstripped:
+ yield prev_token_rstripped
+ yield found
+ id_token = ParseToken(lines, found.end, ID_REGEX, 'id')
+ yield id_token
+ pos = Skip(lines, id_token.end, WHITE_SPACE_REGEX)
+ elif found.token_type == '$range':
+ if prev_token_rstripped:
+ yield prev_token_rstripped
+ yield found
+ id_token = ParseToken(lines, found.end, ID_REGEX, 'id')
+ yield id_token
+ pos = Skip(lines, id_token.end, OPTIONAL_WHITE_SPACES_REGEX)
+
+ dots_pos = SkipUntil(lines, pos, DOT_DOT_REGEX, '..')
+ yield MakeToken(lines, pos, dots_pos, 'exp')
+ yield MakeToken(lines, dots_pos, dots_pos + 2, '..')
+ pos = dots_pos + 2
+ new_pos = Cursor(pos.line + 1, 0)
+ yield MakeToken(lines, pos, new_pos, 'exp')
+ pos = new_pos
+ elif found.token_type == '$':
+ if prev_token:
+ yield prev_token
+ yield found
+ exp_token = ParseExpTokenInParens(lines, found.end)
+ yield exp_token
+ pos = exp_token.end
+ elif (found.token_type == ']]' or found.token_type == '$if' or
+ found.token_type == '$elif' or found.token_type == '$else'):
+ if prev_token_rstripped:
+ yield prev_token_rstripped
+ yield found
+ pos = found.end
+ else:
+ if prev_token:
+ yield prev_token
+ yield found
+ pos = found.end
+
+
+def Tokenize(s):
+ """A generator that yields the tokens in the given string."""
+ if s != '':
+ lines = s.splitlines(True)
+ for token in TokenizeLines(lines, Cursor(0, 0)):
+ yield token
+
+
+class CodeNode:
+ def __init__(self, atomic_code_list=None):
+ self.atomic_code = atomic_code_list
+
+
+class VarNode:
+ def __init__(self, identifier=None, atomic_code=None):
+ self.identifier = identifier
+ self.atomic_code = atomic_code
+
+
+class RangeNode:
+ def __init__(self, identifier=None, exp1=None, exp2=None):
+ self.identifier = identifier
+ self.exp1 = exp1
+ self.exp2 = exp2
+
+
+class ForNode:
+ def __init__(self, identifier=None, sep=None, code=None):
+ self.identifier = identifier
+ self.sep = sep
+ self.code = code
+
+
+class ElseNode:
+ def __init__(self, else_branch=None):
+ self.else_branch = else_branch
+
+
+class IfNode:
+ def __init__(self, exp=None, then_branch=None, else_branch=None):
+ self.exp = exp
+ self.then_branch = then_branch
+ self.else_branch = else_branch
+
+
+class RawCodeNode:
+ def __init__(self, token=None):
+ self.raw_code = token
+
+
+class LiteralDollarNode:
+ def __init__(self, token):
+ self.token = token
+
+
+class ExpNode:
+ def __init__(self, token, python_exp):
+ self.token = token
+ self.python_exp = python_exp
+
+
+def PopFront(a_list):
+ head = a_list[0]
+ a_list[:1] = []
+ return head
+
+
+def PushFront(a_list, elem):
+ a_list[:0] = [elem]
+
+
+def PopToken(a_list, token_type=None):
+ token = PopFront(a_list)
+ if token_type is not None and token.token_type != token_type:
+ print 'ERROR: %s expected at %s' % (token_type, token.start)
+ print 'ERROR: %s found instead' % (token,)
+ sys.exit(1)
+
+ return token
+
+
+def PeekToken(a_list):
+ if not a_list:
+ return None
+
+ return a_list[0]
+
+
+def ParseExpNode(token):
+ python_exp = re.sub(r'([_A-Za-z]\w*)', r'self.GetValue("\1")', token.value)
+ return ExpNode(token, python_exp)
+
+
+def ParseElseNode(tokens):
+ def Pop(token_type=None):
+ return PopToken(tokens, token_type)
+
+ next = PeekToken(tokens)
+ if not next:
+ return None
+ if next.token_type == '$else':
+ Pop('$else')
+ Pop('[[')
+ code_node = ParseCodeNode(tokens)
+ Pop(']]')
+ return code_node
+ elif next.token_type == '$elif':
+ Pop('$elif')
+ exp = Pop('code')
+ Pop('[[')
+ code_node = ParseCodeNode(tokens)
+ Pop(']]')
+ inner_else_node = ParseElseNode(tokens)
+ return CodeNode([IfNode(ParseExpNode(exp), code_node, inner_else_node)])
+ elif not next.value.strip():
+ Pop('code')
+ return ParseElseNode(tokens)
+ else:
+ return None
+
+
+def ParseAtomicCodeNode(tokens):
+ def Pop(token_type=None):
+ return PopToken(tokens, token_type)
+
+ head = PopFront(tokens)
+ t = head.token_type
+ if t == 'code':
+ return RawCodeNode(head)
+ elif t == '$var':
+ id_token = Pop('id')
+ Pop('=')
+ next = PeekToken(tokens)
+ if next.token_type == 'exp':
+ exp_token = Pop()
+ return VarNode(id_token, ParseExpNode(exp_token))
+ Pop('[[')
+ code_node = ParseCodeNode(tokens)
+ Pop(']]')
+ return VarNode(id_token, code_node)
+ elif t == '$for':
+ id_token = Pop('id')
+ next_token = PeekToken(tokens)
+ if next_token.token_type == 'code':
+ sep_token = next_token
+ Pop('code')
+ else:
+ sep_token = None
+ Pop('[[')
+ code_node = ParseCodeNode(tokens)
+ Pop(']]')
+ return ForNode(id_token, sep_token, code_node)
+ elif t == '$if':
+ exp_token = Pop('code')
+ Pop('[[')
+ code_node = ParseCodeNode(tokens)
+ Pop(']]')
+ else_node = ParseElseNode(tokens)
+ return IfNode(ParseExpNode(exp_token), code_node, else_node)
+ elif t == '$range':
+ id_token = Pop('id')
+ exp1_token = Pop('exp')
+ Pop('..')
+ exp2_token = Pop('exp')
+ return RangeNode(id_token, ParseExpNode(exp1_token),
+ ParseExpNode(exp2_token))
+ elif t == '$id':
+ return ParseExpNode(Token(head.start + 1, head.end, head.value[1:], 'id'))
+ elif t == '$($)':
+ return LiteralDollarNode(head)
+ elif t == '$':
+ exp_token = Pop('exp')
+ return ParseExpNode(exp_token)
+ elif t == '[[':
+ code_node = ParseCodeNode(tokens)
+ Pop(']]')
+ return code_node
+ else:
+ PushFront(tokens, head)
+ return None
+
+
+def ParseCodeNode(tokens):
+ atomic_code_list = []
+ while True:
+ if not tokens:
+ break
+ atomic_code_node = ParseAtomicCodeNode(tokens)
+ if atomic_code_node:
+ atomic_code_list.append(atomic_code_node)
+ else:
+ break
+ return CodeNode(atomic_code_list)
+
+
+def ParseToAST(pump_src_text):
+ """Convert the given Pump source text into an AST."""
+ tokens = list(Tokenize(pump_src_text))
+ code_node = ParseCodeNode(tokens)
+ return code_node
+
+
+class Env:
+ def __init__(self):
+ self.variables = []
+ self.ranges = []
+
+ def Clone(self):
+ clone = Env()
+ clone.variables = self.variables[:]
+ clone.ranges = self.ranges[:]
+ return clone
+
+ def PushVariable(self, var, value):
+ # If value looks like an int, store it as an int.
+ try:
+ int_value = int(value)
+ if ('%s' % int_value) == value:
+ value = int_value
+ except Exception:
+ pass
+ self.variables[:0] = [(var, value)]
+
+ def PopVariable(self):
+ self.variables[:1] = []
+
+ def PushRange(self, var, lower, upper):
+ self.ranges[:0] = [(var, lower, upper)]
+
+ def PopRange(self):
+ self.ranges[:1] = []
+
+ def GetValue(self, identifier):
+ for (var, value) in self.variables:
+ if identifier == var:
+ return value
+
+ print 'ERROR: meta variable %s is undefined.' % (identifier,)
+ sys.exit(1)
+
+ def EvalExp(self, exp):
+ try:
+ result = eval(exp.python_exp)
+ except Exception, e:
+ print 'ERROR: caught exception %s: %s' % (e.__class__.__name__, e)
+ print ('ERROR: failed to evaluate meta expression %s at %s' %
+ (exp.python_exp, exp.token.start))
+ sys.exit(1)
+ return result
+
+ def GetRange(self, identifier):
+ for (var, lower, upper) in self.ranges:
+ if identifier == var:
+ return (lower, upper)
+
+ print 'ERROR: range %s is undefined.' % (identifier,)
+ sys.exit(1)
+
+
+class Output:
+ def __init__(self):
+ self.string = ''
+
+ def GetLastLine(self):
+ index = self.string.rfind('\n')
+ if index < 0:
+ return ''
+
+ return self.string[index + 1:]
+
+ def Append(self, s):
+ self.string += s
+
+
+def RunAtomicCode(env, node, output):
+ if isinstance(node, VarNode):
+ identifier = node.identifier.value.strip()
+ result = Output()
+ RunAtomicCode(env.Clone(), node.atomic_code, result)
+ value = result.string
+ env.PushVariable(identifier, value)
+ elif isinstance(node, RangeNode):
+ identifier = node.identifier.value.strip()
+ lower = int(env.EvalExp(node.exp1))
+ upper = int(env.EvalExp(node.exp2))
+ env.PushRange(identifier, lower, upper)
+ elif isinstance(node, ForNode):
+ identifier = node.identifier.value.strip()
+ if node.sep is None:
+ sep = ''
+ else:
+ sep = node.sep.value
+ (lower, upper) = env.GetRange(identifier)
+ for i in range(lower, upper + 1):
+ new_env = env.Clone()
+ new_env.PushVariable(identifier, i)
+ RunCode(new_env, node.code, output)
+ if i != upper:
+ output.Append(sep)
+ elif isinstance(node, RawCodeNode):
+ output.Append(node.raw_code.value)
+ elif isinstance(node, IfNode):
+ cond = env.EvalExp(node.exp)
+ if cond:
+ RunCode(env.Clone(), node.then_branch, output)
+ elif node.else_branch is not None:
+ RunCode(env.Clone(), node.else_branch, output)
+ elif isinstance(node, ExpNode):
+ value = env.EvalExp(node)
+ output.Append('%s' % (value,))
+ elif isinstance(node, LiteralDollarNode):
+ output.Append('$')
+ elif isinstance(node, CodeNode):
+ RunCode(env.Clone(), node, output)
+ else:
+ print 'BAD'
+ print node
+ sys.exit(1)
+
+
+def RunCode(env, code_node, output):
+ for atomic_code in code_node.atomic_code:
+ RunAtomicCode(env, atomic_code, output)
+
+
+def IsComment(cur_line):
+ return '//' in cur_line
+
+
+def IsInPreprocessorDirevative(prev_lines, cur_line):
+ if cur_line.lstrip().startswith('#'):
+ return True
+ return prev_lines != [] and prev_lines[-1].endswith('\\')
+
+
+def WrapComment(line, output):
+ loc = line.find('//')
+ before_comment = line[:loc].rstrip()
+ if before_comment == '':
+ indent = loc
+ else:
+ output.append(before_comment)
+ indent = len(before_comment) - len(before_comment.lstrip())
+ prefix = indent*' ' + '// '
+ max_len = 80 - len(prefix)
+ comment = line[loc + 2:].strip()
+ segs = [seg for seg in re.split(r'(\w+\W*)', comment) if seg != '']
+ cur_line = ''
+ for seg in segs:
+ if len((cur_line + seg).rstrip()) < max_len:
+ cur_line += seg
+ else:
+ if cur_line.strip() != '':
+ output.append(prefix + cur_line.rstrip())
+ cur_line = seg.lstrip()
+ if cur_line.strip() != '':
+ output.append(prefix + cur_line.strip())
+
+
+def WrapCode(line, line_concat, output):
+ indent = len(line) - len(line.lstrip())
+ prefix = indent*' ' # Prefix of the current line
+ max_len = 80 - indent - len(line_concat) # Maximum length of the current line
+ new_prefix = prefix + 4*' ' # Prefix of a continuation line
+ new_max_len = max_len - 4 # Maximum length of a continuation line
+ # Prefers to wrap a line after a ',' or ';'.
+ segs = [seg for seg in re.split(r'([^,;]+[,;]?)', line.strip()) if seg != '']
+ cur_line = '' # The current line without leading spaces.
+ for seg in segs:
+ # If the line is still too long, wrap at a space.
+ while cur_line == '' and len(seg.strip()) > max_len:
+ seg = seg.lstrip()
+ split_at = seg.rfind(' ', 0, max_len)
+ output.append(prefix + seg[:split_at].strip() + line_concat)
+ seg = seg[split_at + 1:]
+ prefix = new_prefix
+ max_len = new_max_len
+
+ if len((cur_line + seg).rstrip()) < max_len:
+ cur_line = (cur_line + seg).lstrip()
+ else:
+ output.append(prefix + cur_line.rstrip() + line_concat)
+ prefix = new_prefix
+ max_len = new_max_len
+ cur_line = seg.lstrip()
+ if cur_line.strip() != '':
+ output.append(prefix + cur_line.strip())
+
+
+def WrapPreprocessorDirevative(line, output):
+ WrapCode(line, ' \\', output)
+
+
+def WrapPlainCode(line, output):
+ WrapCode(line, '', output)
+
+
+def IsHeaderGuardOrInclude(line):
+ return (re.match(r'^#(ifndef|define|endif\s*//)\s*[\w_]+\s*$', line) or
+ re.match(r'^#include\s', line))
+
+
+def WrapLongLine(line, output):
+ line = line.rstrip()
+ if len(line) <= 80:
+ output.append(line)
+ elif IsComment(line):
+ if IsHeaderGuardOrInclude(line):
+ # The style guide made an exception to allow long header guard lines
+ # and includes.
+ output.append(line)
+ else:
+ WrapComment(line, output)
+ elif IsInPreprocessorDirevative(output, line):
+ if IsHeaderGuardOrInclude(line):
+ # The style guide made an exception to allow long header guard lines
+ # and includes.
+ output.append(line)
+ else:
+ WrapPreprocessorDirevative(line, output)
+ else:
+ WrapPlainCode(line, output)
+
+
+def BeautifyCode(string):
+ lines = string.splitlines()
+ output = []
+ for line in lines:
+ WrapLongLine(line, output)
+ output2 = [line.rstrip() for line in output]
+ return '\n'.join(output2) + '\n'
+
+
+def ConvertFromPumpSource(src_text):
+ """Return the text generated from the given Pump source text."""
+ ast = ParseToAST(StripMetaComments(src_text))
+ output = Output()
+ RunCode(Env(), ast, output)
+ return BeautifyCode(output.string)
+
+
+def main(argv):
+ if len(argv) == 1:
+ print __doc__
+ sys.exit(1)
+
+ file_path = argv[-1]
+ output_str = ConvertFromPumpSource(file(file_path, 'r').read())
+ if file_path.endswith('.pump'):
+ output_file_path = file_path[:-5]
+ else:
+ output_file_path = '-'
+ if output_file_path == '-':
+ print output_str,
+ else:
+ output_file = file(output_file_path, 'w')
+ output_file.write('// This file was GENERATED by command:\n')
+ output_file.write('// %s %s\n' %
+ (os.path.basename(__file__), os.path.basename(file_path)))
+ output_file.write('// DO NOT EDIT BY HAND!!!\n\n')
+ output_file.write(output_str)
+ output_file.close()
+
+
+if __name__ == '__main__':
+ main(sys.argv)
diff --git a/libvpx/third_party/googletest/src/scripts/test/Makefile b/libvpx/third_party/googletest/src/scripts/test/Makefile
new file mode 100644
index 0000000..cdff584
--- /dev/null
+++ b/libvpx/third_party/googletest/src/scripts/test/Makefile
@@ -0,0 +1,59 @@
+# A Makefile for fusing Google Test and building a sample test against it.
+#
+# SYNOPSIS:
+#
+# make [all] - makes everything.
+# make TARGET - makes the given target.
+# make check - makes everything and runs the built sample test.
+# make clean - removes all files generated by make.
+
+# Points to the root of fused Google Test, relative to where this file is.
+FUSED_GTEST_DIR = output
+
+# Paths to the fused gtest files.
+FUSED_GTEST_H = $(FUSED_GTEST_DIR)/gtest/gtest.h
+FUSED_GTEST_ALL_CC = $(FUSED_GTEST_DIR)/gtest/gtest-all.cc
+
+# Where to find the sample test.
+SAMPLE_DIR = ../../samples
+
+# Where to find gtest_main.cc.
+GTEST_MAIN_CC = ../../src/gtest_main.cc
+
+# Flags passed to the preprocessor.
+# We have no idea here whether pthreads is available in the system, so
+# disable its use.
+CPPFLAGS += -I$(FUSED_GTEST_DIR) -DGTEST_HAS_PTHREAD=0
+
+# Flags passed to the C++ compiler.
+CXXFLAGS += -g
+
+all : sample1_unittest
+
+check : all
+ ./sample1_unittest
+
+clean :
+ rm -rf $(FUSED_GTEST_DIR) sample1_unittest *.o
+
+$(FUSED_GTEST_H) :
+ ../fuse_gtest_files.py $(FUSED_GTEST_DIR)
+
+$(FUSED_GTEST_ALL_CC) :
+ ../fuse_gtest_files.py $(FUSED_GTEST_DIR)
+
+gtest-all.o : $(FUSED_GTEST_H) $(FUSED_GTEST_ALL_CC)
+ $(CXX) $(CPPFLAGS) $(CXXFLAGS) -c $(FUSED_GTEST_DIR)/gtest/gtest-all.cc
+
+gtest_main.o : $(FUSED_GTEST_H) $(GTEST_MAIN_CC)
+ $(CXX) $(CPPFLAGS) $(CXXFLAGS) -c $(GTEST_MAIN_CC)
+
+sample1.o : $(SAMPLE_DIR)/sample1.cc $(SAMPLE_DIR)/sample1.h
+ $(CXX) $(CPPFLAGS) $(CXXFLAGS) -c $(SAMPLE_DIR)/sample1.cc
+
+sample1_unittest.o : $(SAMPLE_DIR)/sample1_unittest.cc \
+ $(SAMPLE_DIR)/sample1.h $(FUSED_GTEST_H)
+ $(CXX) $(CPPFLAGS) $(CXXFLAGS) -c $(SAMPLE_DIR)/sample1_unittest.cc
+
+sample1_unittest : sample1.o sample1_unittest.o gtest-all.o gtest_main.o
+ $(CXX) $(CPPFLAGS) $(CXXFLAGS) $^ -o $@
diff --git a/libvpx/third_party/googletest/src/scripts/upload.py b/libvpx/third_party/googletest/src/scripts/upload.py
new file mode 100755
index 0000000..6e6f9a1
--- /dev/null
+++ b/libvpx/third_party/googletest/src/scripts/upload.py
@@ -0,0 +1,1387 @@
+#!/usr/bin/env python
+#
+# Copyright 2007 Google Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Tool for uploading diffs from a version control system to the codereview app.
+
+Usage summary: upload.py [options] [-- diff_options]
+
+Diff options are passed to the diff command of the underlying system.
+
+Supported version control systems:
+ Git
+ Mercurial
+ Subversion
+
+It is important for Git/Mercurial users to specify a tree/node/branch to diff
+against by using the '--rev' option.
+"""
+# This code is derived from appcfg.py in the App Engine SDK (open source),
+# and from ASPN recipe #146306.
+
+import cookielib
+import getpass
+import logging
+import md5
+import mimetypes
+import optparse
+import os
+import re
+import socket
+import subprocess
+import sys
+import urllib
+import urllib2
+import urlparse
+
+try:
+ import readline
+except ImportError:
+ pass
+
+# The logging verbosity:
+# 0: Errors only.
+# 1: Status messages.
+# 2: Info logs.
+# 3: Debug logs.
+verbosity = 1
+
+# Max size of patch or base file.
+MAX_UPLOAD_SIZE = 900 * 1024
+
+
+def GetEmail(prompt):
+ """Prompts the user for their email address and returns it.
+
+ The last used email address is saved to a file and offered up as a suggestion
+ to the user. If the user presses enter without typing in anything the last
+ used email address is used. If the user enters a new address, it is saved
+ for next time we prompt.
+
+ """
+ last_email_file_name = os.path.expanduser("~/.last_codereview_email_address")
+ last_email = ""
+ if os.path.exists(last_email_file_name):
+ try:
+ last_email_file = open(last_email_file_name, "r")
+ last_email = last_email_file.readline().strip("\n")
+ last_email_file.close()
+ prompt += " [%s]" % last_email
+ except IOError, e:
+ pass
+ email = raw_input(prompt + ": ").strip()
+ if email:
+ try:
+ last_email_file = open(last_email_file_name, "w")
+ last_email_file.write(email)
+ last_email_file.close()
+ except IOError, e:
+ pass
+ else:
+ email = last_email
+ return email
+
+
+def StatusUpdate(msg):
+ """Print a status message to stdout.
+
+ If 'verbosity' is greater than 0, print the message.
+
+ Args:
+ msg: The string to print.
+ """
+ if verbosity > 0:
+ print msg
+
+
+def ErrorExit(msg):
+ """Print an error message to stderr and exit."""
+ print >>sys.stderr, msg
+ sys.exit(1)
+
+
+class ClientLoginError(urllib2.HTTPError):
+ """Raised to indicate there was an error authenticating with ClientLogin."""
+
+ def __init__(self, url, code, msg, headers, args):
+ urllib2.HTTPError.__init__(self, url, code, msg, headers, None)
+ self.args = args
+ self.reason = args["Error"]
+
+
+class AbstractRpcServer(object):
+ """Provides a common interface for a simple RPC server."""
+
+ def __init__(self, host, auth_function, host_override=None, extra_headers={},
+ save_cookies=False):
+ """Creates a new HttpRpcServer.
+
+ Args:
+ host: The host to send requests to.
+ auth_function: A function that takes no arguments and returns an
+ (email, password) tuple when called. Will be called if authentication
+ is required.
+ host_override: The host header to send to the server (defaults to host).
+ extra_headers: A dict of extra headers to append to every request.
+ save_cookies: If True, save the authentication cookies to local disk.
+ If False, use an in-memory cookiejar instead. Subclasses must
+ implement this functionality. Defaults to False.
+ """
+ self.host = host
+ self.host_override = host_override
+ self.auth_function = auth_function
+ self.authenticated = False
+ self.extra_headers = extra_headers
+ self.save_cookies = save_cookies
+ self.opener = self._GetOpener()
+ if self.host_override:
+ logging.info("Server: %s; Host: %s", self.host, self.host_override)
+ else:
+ logging.info("Server: %s", self.host)
+
+ def _GetOpener(self):
+ """Returns an OpenerDirector for making HTTP requests.
+
+ Returns:
+ A urllib2.OpenerDirector object.
+ """
+ raise NotImplementedError()
+
+ def _CreateRequest(self, url, data=None):
+ """Creates a new urllib request."""
+ logging.debug("Creating request for: '%s' with payload:\n%s", url, data)
+ req = urllib2.Request(url, data=data)
+ if self.host_override:
+ req.add_header("Host", self.host_override)
+ for key, value in self.extra_headers.iteritems():
+ req.add_header(key, value)
+ return req
+
+ def _GetAuthToken(self, email, password):
+ """Uses ClientLogin to authenticate the user, returning an auth token.
+
+ Args:
+ email: The user's email address
+ password: The user's password
+
+ Raises:
+ ClientLoginError: If there was an error authenticating with ClientLogin.
+ HTTPError: If there was some other form of HTTP error.
+
+ Returns:
+ The authentication token returned by ClientLogin.
+ """
+ account_type = "GOOGLE"
+ if self.host.endswith(".google.com"):
+ # Needed for use inside Google.
+ account_type = "HOSTED"
+ req = self._CreateRequest(
+ url="https://www.google.com/accounts/ClientLogin",
+ data=urllib.urlencode({
+ "Email": email,
+ "Passwd": password,
+ "service": "ah",
+ "source": "rietveld-codereview-upload",
+ "accountType": account_type,
+ }),
+ )
+ try:
+ response = self.opener.open(req)
+ response_body = response.read()
+ response_dict = dict(x.split("=")
+ for x in response_body.split("\n") if x)
+ return response_dict["Auth"]
+ except urllib2.HTTPError, e:
+ if e.code == 403:
+ body = e.read()
+ response_dict = dict(x.split("=", 1) for x in body.split("\n") if x)
+ raise ClientLoginError(req.get_full_url(), e.code, e.msg,
+ e.headers, response_dict)
+ else:
+ raise
+
+ def _GetAuthCookie(self, auth_token):
+ """Fetches authentication cookies for an authentication token.
+
+ Args:
+ auth_token: The authentication token returned by ClientLogin.
+
+ Raises:
+ HTTPError: If there was an error fetching the authentication cookies.
+ """
+ # This is a dummy value to allow us to identify when we're successful.
+ continue_location = "http://localhost/"
+ args = {"continue": continue_location, "auth": auth_token}
+ req = self._CreateRequest("http://%s/_ah/login?%s" %
+ (self.host, urllib.urlencode(args)))
+ try:
+ response = self.opener.open(req)
+ except urllib2.HTTPError, e:
+ response = e
+ if (response.code != 302 or
+ response.info()["location"] != continue_location):
+ raise urllib2.HTTPError(req.get_full_url(), response.code, response.msg,
+ response.headers, response.fp)
+ self.authenticated = True
+
+ def _Authenticate(self):
+ """Authenticates the user.
+
+ The authentication process works as follows:
+ 1) We get a username and password from the user
+ 2) We use ClientLogin to obtain an AUTH token for the user
+ (see http://code.google.com/apis/accounts/AuthForInstalledApps.html).
+ 3) We pass the auth token to /_ah/login on the server to obtain an
+ authentication cookie. If login was successful, it tries to redirect
+ us to the URL we provided.
+
+ If we attempt to access the upload API without first obtaining an
+ authentication cookie, it returns a 401 response and directs us to
+ authenticate ourselves with ClientLogin.
+ """
+ for i in range(3):
+ credentials = self.auth_function()
+ try:
+ auth_token = self._GetAuthToken(credentials[0], credentials[1])
+ except ClientLoginError, e:
+ if e.reason == "BadAuthentication":
+ print >>sys.stderr, "Invalid username or password."
+ continue
+ if e.reason == "CaptchaRequired":
+ print >>sys.stderr, (
+ "Please go to\n"
+ "https://www.google.com/accounts/DisplayUnlockCaptcha\n"
+ "and verify you are a human. Then try again.")
+ break
+ if e.reason == "NotVerified":
+ print >>sys.stderr, "Account not verified."
+ break
+ if e.reason == "TermsNotAgreed":
+ print >>sys.stderr, "User has not agreed to TOS."
+ break
+ if e.reason == "AccountDeleted":
+ print >>sys.stderr, "The user account has been deleted."
+ break
+ if e.reason == "AccountDisabled":
+ print >>sys.stderr, "The user account has been disabled."
+ break
+ if e.reason == "ServiceDisabled":
+ print >>sys.stderr, ("The user's access to the service has been "
+ "disabled.")
+ break
+ if e.reason == "ServiceUnavailable":
+ print >>sys.stderr, "The service is not available; try again later."
+ break
+ raise
+ self._GetAuthCookie(auth_token)
+ return
+
+ def Send(self, request_path, payload=None,
+ content_type="application/octet-stream",
+ timeout=None,
+ **kwargs):
+ """Sends an RPC and returns the response.
+
+ Args:
+ request_path: The path to send the request to, eg /api/appversion/create.
+ payload: The body of the request, or None to send an empty request.
+ content_type: The Content-Type header to use.
+ timeout: timeout in seconds; default None i.e. no timeout.
+ (Note: for large requests on OS X, the timeout doesn't work right.)
+ kwargs: Any keyword arguments are converted into query string parameters.
+
+ Returns:
+ The response body, as a string.
+ """
+ # TODO: Don't require authentication. Let the server say
+ # whether it is necessary.
+ if not self.authenticated:
+ self._Authenticate()
+
+ old_timeout = socket.getdefaulttimeout()
+ socket.setdefaulttimeout(timeout)
+ try:
+ tries = 0
+ while True:
+ tries += 1
+ args = dict(kwargs)
+ url = "http://%s%s" % (self.host, request_path)
+ if args:
+ url += "?" + urllib.urlencode(args)
+ req = self._CreateRequest(url=url, data=payload)
+ req.add_header("Content-Type", content_type)
+ try:
+ f = self.opener.open(req)
+ response = f.read()
+ f.close()
+ return response
+ except urllib2.HTTPError, e:
+ if tries > 3:
+ raise
+ elif e.code == 401:
+ self._Authenticate()
+## elif e.code >= 500 and e.code < 600:
+## # Server Error - try again.
+## continue
+ else:
+ raise
+ finally:
+ socket.setdefaulttimeout(old_timeout)
+
+
+class HttpRpcServer(AbstractRpcServer):
+ """Provides a simplified RPC-style interface for HTTP requests."""
+
+ def _Authenticate(self):
+ """Save the cookie jar after authentication."""
+ super(HttpRpcServer, self)._Authenticate()
+ if self.save_cookies:
+ StatusUpdate("Saving authentication cookies to %s" % self.cookie_file)
+ self.cookie_jar.save()
+
+ def _GetOpener(self):
+ """Returns an OpenerDirector that supports cookies and ignores redirects.
+
+ Returns:
+ A urllib2.OpenerDirector object.
+ """
+ opener = urllib2.OpenerDirector()
+ opener.add_handler(urllib2.ProxyHandler())
+ opener.add_handler(urllib2.UnknownHandler())
+ opener.add_handler(urllib2.HTTPHandler())
+ opener.add_handler(urllib2.HTTPDefaultErrorHandler())
+ opener.add_handler(urllib2.HTTPSHandler())
+ opener.add_handler(urllib2.HTTPErrorProcessor())
+ if self.save_cookies:
+ self.cookie_file = os.path.expanduser("~/.codereview_upload_cookies")
+ self.cookie_jar = cookielib.MozillaCookieJar(self.cookie_file)
+ if os.path.exists(self.cookie_file):
+ try:
+ self.cookie_jar.load()
+ self.authenticated = True
+ StatusUpdate("Loaded authentication cookies from %s" %
+ self.cookie_file)
+ except (cookielib.LoadError, IOError):
+ # Failed to load cookies - just ignore them.
+ pass
+ else:
+ # Create an empty cookie file with mode 600
+ fd = os.open(self.cookie_file, os.O_CREAT, 0600)
+ os.close(fd)
+ # Always chmod the cookie file
+ os.chmod(self.cookie_file, 0600)
+ else:
+ # Don't save cookies across runs of update.py.
+ self.cookie_jar = cookielib.CookieJar()
+ opener.add_handler(urllib2.HTTPCookieProcessor(self.cookie_jar))
+ return opener
+
+
+parser = optparse.OptionParser(usage="%prog [options] [-- diff_options]")
+parser.add_option("-y", "--assume_yes", action="store_true",
+ dest="assume_yes", default=False,
+ help="Assume that the answer to yes/no questions is 'yes'.")
+# Logging
+group = parser.add_option_group("Logging options")
+group.add_option("-q", "--quiet", action="store_const", const=0,
+ dest="verbose", help="Print errors only.")
+group.add_option("-v", "--verbose", action="store_const", const=2,
+ dest="verbose", default=1,
+ help="Print info level logs (default).")
+group.add_option("--noisy", action="store_const", const=3,
+ dest="verbose", help="Print all logs.")
+# Review server
+group = parser.add_option_group("Review server options")
+group.add_option("-s", "--server", action="store", dest="server",
+ default="codereview.appspot.com",
+ metavar="SERVER",
+ help=("The server to upload to. The format is host[:port]. "
+ "Defaults to 'codereview.appspot.com'."))
+group.add_option("-e", "--email", action="store", dest="email",
+ metavar="EMAIL", default=None,
+ help="The username to use. Will prompt if omitted.")
+group.add_option("-H", "--host", action="store", dest="host",
+ metavar="HOST", default=None,
+ help="Overrides the Host header sent with all RPCs.")
+group.add_option("--no_cookies", action="store_false",
+ dest="save_cookies", default=True,
+ help="Do not save authentication cookies to local disk.")
+# Issue
+group = parser.add_option_group("Issue options")
+group.add_option("-d", "--description", action="store", dest="description",
+ metavar="DESCRIPTION", default=None,
+ help="Optional description when creating an issue.")
+group.add_option("-f", "--description_file", action="store",
+ dest="description_file", metavar="DESCRIPTION_FILE",
+ default=None,
+ help="Optional path of a file that contains "
+ "the description when creating an issue.")
+group.add_option("-r", "--reviewers", action="store", dest="reviewers",
+ metavar="REVIEWERS", default=None,
+ help="Add reviewers (comma separated email addresses).")
+group.add_option("--cc", action="store", dest="cc",
+ metavar="CC", default=None,
+ help="Add CC (comma separated email addresses).")
+# Upload options
+group = parser.add_option_group("Patch options")
+group.add_option("-m", "--message", action="store", dest="message",
+ metavar="MESSAGE", default=None,
+ help="A message to identify the patch. "
+ "Will prompt if omitted.")
+group.add_option("-i", "--issue", type="int", action="store",
+ metavar="ISSUE", default=None,
+ help="Issue number to which to add. Defaults to new issue.")
+group.add_option("--download_base", action="store_true",
+ dest="download_base", default=False,
+ help="Base files will be downloaded by the server "
+ "(side-by-side diffs may not work on files with CRs).")
+group.add_option("--rev", action="store", dest="revision",
+ metavar="REV", default=None,
+ help="Branch/tree/revision to diff against (used by DVCS).")
+group.add_option("--send_mail", action="store_true",
+ dest="send_mail", default=False,
+ help="Send notification email to reviewers.")
+
+
+def GetRpcServer(options):
+ """Returns an instance of an AbstractRpcServer.
+
+ Returns:
+ A new AbstractRpcServer, on which RPC calls can be made.
+ """
+
+ rpc_server_class = HttpRpcServer
+
+ def GetUserCredentials():
+ """Prompts the user for a username and password."""
+ email = options.email
+ if email is None:
+ email = GetEmail("Email (login for uploading to %s)" % options.server)
+ password = getpass.getpass("Password for %s: " % email)
+ return (email, password)
+
+ # If this is the dev_appserver, use fake authentication.
+ host = (options.host or options.server).lower()
+ if host == "localhost" or host.startswith("localhost:"):
+ email = options.email
+ if email is None:
+ email = "test@example.com"
+ logging.info("Using debug user %s. Override with --email" % email)
+ server = rpc_server_class(
+ options.server,
+ lambda: (email, "password"),
+ host_override=options.host,
+ extra_headers={"Cookie":
+ 'dev_appserver_login="%s:False"' % email},
+ save_cookies=options.save_cookies)
+ # Don't try to talk to ClientLogin.
+ server.authenticated = True
+ return server
+
+ return rpc_server_class(options.server, GetUserCredentials,
+ host_override=options.host,
+ save_cookies=options.save_cookies)
+
+
+def EncodeMultipartFormData(fields, files):
+ """Encode form fields for multipart/form-data.
+
+ Args:
+ fields: A sequence of (name, value) elements for regular form fields.
+ files: A sequence of (name, filename, value) elements for data to be
+ uploaded as files.
+ Returns:
+ (content_type, body) ready for httplib.HTTP instance.
+
+ Source:
+ http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/146306
+ """
+ BOUNDARY = '-M-A-G-I-C---B-O-U-N-D-A-R-Y-'
+ CRLF = '\r\n'
+ lines = []
+ for (key, value) in fields:
+ lines.append('--' + BOUNDARY)
+ lines.append('Content-Disposition: form-data; name="%s"' % key)
+ lines.append('')
+ lines.append(value)
+ for (key, filename, value) in files:
+ lines.append('--' + BOUNDARY)
+ lines.append('Content-Disposition: form-data; name="%s"; filename="%s"' %
+ (key, filename))
+ lines.append('Content-Type: %s' % GetContentType(filename))
+ lines.append('')
+ lines.append(value)
+ lines.append('--' + BOUNDARY + '--')
+ lines.append('')
+ body = CRLF.join(lines)
+ content_type = 'multipart/form-data; boundary=%s' % BOUNDARY
+ return content_type, body
+
+
+def GetContentType(filename):
+ """Helper to guess the content-type from the filename."""
+ return mimetypes.guess_type(filename)[0] or 'application/octet-stream'
+
+
+# Use a shell for subcommands on Windows to get a PATH search.
+use_shell = sys.platform.startswith("win")
+
+def RunShellWithReturnCode(command, print_output=False,
+ universal_newlines=True):
+ """Executes a command and returns the output from stdout and the return code.
+
+ Args:
+ command: Command to execute.
+ print_output: If True, the output is printed to stdout.
+ If False, both stdout and stderr are ignored.
+ universal_newlines: Use universal_newlines flag (default: True).
+
+ Returns:
+ Tuple (output, return code)
+ """
+ logging.info("Running %s", command)
+ p = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE,
+ shell=use_shell, universal_newlines=universal_newlines)
+ if print_output:
+ output_array = []
+ while True:
+ line = p.stdout.readline()
+ if not line:
+ break
+ print line.strip("\n")
+ output_array.append(line)
+ output = "".join(output_array)
+ else:
+ output = p.stdout.read()
+ p.wait()
+ errout = p.stderr.read()
+ if print_output and errout:
+ print >>sys.stderr, errout
+ p.stdout.close()
+ p.stderr.close()
+ return output, p.returncode
+
+
+def RunShell(command, silent_ok=False, universal_newlines=True,
+ print_output=False):
+ data, retcode = RunShellWithReturnCode(command, print_output,
+ universal_newlines)
+ if retcode:
+ ErrorExit("Got error status from %s:\n%s" % (command, data))
+ if not silent_ok and not data:
+ ErrorExit("No output from %s" % command)
+ return data
+
+
+class VersionControlSystem(object):
+ """Abstract base class providing an interface to the VCS."""
+
+ def __init__(self, options):
+ """Constructor.
+
+ Args:
+ options: Command line options.
+ """
+ self.options = options
+
+ def GenerateDiff(self, args):
+ """Return the current diff as a string.
+
+ Args:
+ args: Extra arguments to pass to the diff command.
+ """
+ raise NotImplementedError(
+ "abstract method -- subclass %s must override" % self.__class__)
+
+ def GetUnknownFiles(self):
+ """Return a list of files unknown to the VCS."""
+ raise NotImplementedError(
+ "abstract method -- subclass %s must override" % self.__class__)
+
+ def CheckForUnknownFiles(self):
+ """Show an "are you sure?" prompt if there are unknown files."""
+ unknown_files = self.GetUnknownFiles()
+ if unknown_files:
+ print "The following files are not added to version control:"
+ for line in unknown_files:
+ print line
+ prompt = "Are you sure to continue?(y/N) "
+ answer = raw_input(prompt).strip()
+ if answer != "y":
+ ErrorExit("User aborted")
+
+ def GetBaseFile(self, filename):
+ """Get the content of the upstream version of a file.
+
+ Returns:
+ A tuple (base_content, new_content, is_binary, status)
+ base_content: The contents of the base file.
+ new_content: For text files, this is empty. For binary files, this is
+ the contents of the new file, since the diff output won't contain
+ information to reconstruct the current file.
+ is_binary: True iff the file is binary.
+ status: The status of the file.
+ """
+
+ raise NotImplementedError(
+ "abstract method -- subclass %s must override" % self.__class__)
+
+
+ def GetBaseFiles(self, diff):
+ """Helper that calls GetBase file for each file in the patch.
+
+ Returns:
+ A dictionary that maps from filename to GetBaseFile's tuple. Filenames
+ are retrieved based on lines that start with "Index:" or
+ "Property changes on:".
+ """
+ files = {}
+ for line in diff.splitlines(True):
+ if line.startswith('Index:') or line.startswith('Property changes on:'):
+ unused, filename = line.split(':', 1)
+ # On Windows if a file has property changes its filename uses '\'
+ # instead of '/'.
+ filename = filename.strip().replace('\\', '/')
+ files[filename] = self.GetBaseFile(filename)
+ return files
+
+
+ def UploadBaseFiles(self, issue, rpc_server, patch_list, patchset, options,
+ files):
+ """Uploads the base files (and if necessary, the current ones as well)."""
+
+ def UploadFile(filename, file_id, content, is_binary, status, is_base):
+ """Uploads a file to the server."""
+ file_too_large = False
+ if is_base:
+ type = "base"
+ else:
+ type = "current"
+ if len(content) > MAX_UPLOAD_SIZE:
+ print ("Not uploading the %s file for %s because it's too large." %
+ (type, filename))
+ file_too_large = True
+ content = ""
+ checksum = md5.new(content).hexdigest()
+ if options.verbose > 0 and not file_too_large:
+ print "Uploading %s file for %s" % (type, filename)
+ url = "/%d/upload_content/%d/%d" % (int(issue), int(patchset), file_id)
+ form_fields = [("filename", filename),
+ ("status", status),
+ ("checksum", checksum),
+ ("is_binary", str(is_binary)),
+ ("is_current", str(not is_base)),
+ ]
+ if file_too_large:
+ form_fields.append(("file_too_large", "1"))
+ if options.email:
+ form_fields.append(("user", options.email))
+ ctype, body = EncodeMultipartFormData(form_fields,
+ [("data", filename, content)])
+ response_body = rpc_server.Send(url, body,
+ content_type=ctype)
+ if not response_body.startswith("OK"):
+ StatusUpdate(" --> %s" % response_body)
+ sys.exit(1)
+
+ patches = dict()
+ [patches.setdefault(v, k) for k, v in patch_list]
+ for filename in patches.keys():
+ base_content, new_content, is_binary, status = files[filename]
+ file_id_str = patches.get(filename)
+ if file_id_str.find("nobase") != -1:
+ base_content = None
+ file_id_str = file_id_str[file_id_str.rfind("_") + 1:]
+ file_id = int(file_id_str)
+ if base_content != None:
+ UploadFile(filename, file_id, base_content, is_binary, status, True)
+ if new_content != None:
+ UploadFile(filename, file_id, new_content, is_binary, status, False)
+
+ def IsImage(self, filename):
+ """Returns true if the filename has an image extension."""
+ mimetype = mimetypes.guess_type(filename)[0]
+ if not mimetype:
+ return False
+ return mimetype.startswith("image/")
+
+
+class SubversionVCS(VersionControlSystem):
+ """Implementation of the VersionControlSystem interface for Subversion."""
+
+ def __init__(self, options):
+ super(SubversionVCS, self).__init__(options)
+ if self.options.revision:
+ match = re.match(r"(\d+)(:(\d+))?", self.options.revision)
+ if not match:
+ ErrorExit("Invalid Subversion revision %s." % self.options.revision)
+ self.rev_start = match.group(1)
+ self.rev_end = match.group(3)
+ else:
+ self.rev_start = self.rev_end = None
+ # Cache output from "svn list -r REVNO dirname".
+ # Keys: dirname, Values: 2-tuple (ouput for start rev and end rev).
+ self.svnls_cache = {}
+ # SVN base URL is required to fetch files deleted in an older revision.
+ # Result is cached to not guess it over and over again in GetBaseFile().
+ required = self.options.download_base or self.options.revision is not None
+ self.svn_base = self._GuessBase(required)
+
+ def GuessBase(self, required):
+ """Wrapper for _GuessBase."""
+ return self.svn_base
+
+ def _GuessBase(self, required):
+ """Returns the SVN base URL.
+
+ Args:
+ required: If true, exits if the url can't be guessed, otherwise None is
+ returned.
+ """
+ info = RunShell(["svn", "info"])
+ for line in info.splitlines():
+ words = line.split()
+ if len(words) == 2 and words[0] == "URL:":
+ url = words[1]
+ scheme, netloc, path, params, query, fragment = urlparse.urlparse(url)
+ username, netloc = urllib.splituser(netloc)
+ if username:
+ logging.info("Removed username from base URL")
+ if netloc.endswith("svn.python.org"):
+ if netloc == "svn.python.org":
+ if path.startswith("/projects/"):
+ path = path[9:]
+ elif netloc != "pythondev@svn.python.org":
+ ErrorExit("Unrecognized Python URL: %s" % url)
+ base = "http://svn.python.org/view/*checkout*%s/" % path
+ logging.info("Guessed Python base = %s", base)
+ elif netloc.endswith("svn.collab.net"):
+ if path.startswith("/repos/"):
+ path = path[6:]
+ base = "http://svn.collab.net/viewvc/*checkout*%s/" % path
+ logging.info("Guessed CollabNet base = %s", base)
+ elif netloc.endswith(".googlecode.com"):
+ path = path + "/"
+ base = urlparse.urlunparse(("http", netloc, path, params,
+ query, fragment))
+ logging.info("Guessed Google Code base = %s", base)
+ else:
+ path = path + "/"
+ base = urlparse.urlunparse((scheme, netloc, path, params,
+ query, fragment))
+ logging.info("Guessed base = %s", base)
+ return base
+ if required:
+ ErrorExit("Can't find URL in output from svn info")
+ return None
+
+ def GenerateDiff(self, args):
+ cmd = ["svn", "diff"]
+ if self.options.revision:
+ cmd += ["-r", self.options.revision]
+ cmd.extend(args)
+ data = RunShell(cmd)
+ count = 0
+ for line in data.splitlines():
+ if line.startswith("Index:") or line.startswith("Property changes on:"):
+ count += 1
+ logging.info(line)
+ if not count:
+ ErrorExit("No valid patches found in output from svn diff")
+ return data
+
+ def _CollapseKeywords(self, content, keyword_str):
+ """Collapses SVN keywords."""
+ # svn cat translates keywords but svn diff doesn't. As a result of this
+ # behavior patching.PatchChunks() fails with a chunk mismatch error.
+ # This part was originally written by the Review Board development team
+ # who had the same problem (http://reviews.review-board.org/r/276/).
+ # Mapping of keywords to known aliases
+ svn_keywords = {
+ # Standard keywords
+ 'Date': ['Date', 'LastChangedDate'],
+ 'Revision': ['Revision', 'LastChangedRevision', 'Rev'],
+ 'Author': ['Author', 'LastChangedBy'],
+ 'HeadURL': ['HeadURL', 'URL'],
+ 'Id': ['Id'],
+
+ # Aliases
+ 'LastChangedDate': ['LastChangedDate', 'Date'],
+ 'LastChangedRevision': ['LastChangedRevision', 'Rev', 'Revision'],
+ 'LastChangedBy': ['LastChangedBy', 'Author'],
+ 'URL': ['URL', 'HeadURL'],
+ }
+
+ def repl(m):
+ if m.group(2):
+ return "$%s::%s$" % (m.group(1), " " * len(m.group(3)))
+ return "$%s$" % m.group(1)
+ keywords = [keyword
+ for name in keyword_str.split(" ")
+ for keyword in svn_keywords.get(name, [])]
+ return re.sub(r"\$(%s):(:?)([^\$]+)\$" % '|'.join(keywords), repl, content)
+
+ def GetUnknownFiles(self):
+ status = RunShell(["svn", "status", "--ignore-externals"], silent_ok=True)
+ unknown_files = []
+ for line in status.split("\n"):
+ if line and line[0] == "?":
+ unknown_files.append(line)
+ return unknown_files
+
+ def ReadFile(self, filename):
+ """Returns the contents of a file."""
+ file = open(filename, 'rb')
+ result = ""
+ try:
+ result = file.read()
+ finally:
+ file.close()
+ return result
+
+ def GetStatus(self, filename):
+ """Returns the status of a file."""
+ if not self.options.revision:
+ status = RunShell(["svn", "status", "--ignore-externals", filename])
+ if not status:
+ ErrorExit("svn status returned no output for %s" % filename)
+ status_lines = status.splitlines()
+ # If file is in a cl, the output will begin with
+ # "\n--- Changelist 'cl_name':\n". See
+ # http://svn.collab.net/repos/svn/trunk/notes/changelist-design.txt
+ if (len(status_lines) == 3 and
+ not status_lines[0] and
+ status_lines[1].startswith("--- Changelist")):
+ status = status_lines[2]
+ else:
+ status = status_lines[0]
+ # If we have a revision to diff against we need to run "svn list"
+ # for the old and the new revision and compare the results to get
+ # the correct status for a file.
+ else:
+ dirname, relfilename = os.path.split(filename)
+ if dirname not in self.svnls_cache:
+ cmd = ["svn", "list", "-r", self.rev_start, dirname or "."]
+ out, returncode = RunShellWithReturnCode(cmd)
+ if returncode:
+ ErrorExit("Failed to get status for %s." % filename)
+ old_files = out.splitlines()
+ args = ["svn", "list"]
+ if self.rev_end:
+ args += ["-r", self.rev_end]
+ cmd = args + [dirname or "."]
+ out, returncode = RunShellWithReturnCode(cmd)
+ if returncode:
+ ErrorExit("Failed to run command %s" % cmd)
+ self.svnls_cache[dirname] = (old_files, out.splitlines())
+ old_files, new_files = self.svnls_cache[dirname]
+ if relfilename in old_files and relfilename not in new_files:
+ status = "D "
+ elif relfilename in old_files and relfilename in new_files:
+ status = "M "
+ else:
+ status = "A "
+ return status
+
+ def GetBaseFile(self, filename):
+ status = self.GetStatus(filename)
+ base_content = None
+ new_content = None
+
+ # If a file is copied its status will be "A +", which signifies
+ # "addition-with-history". See "svn st" for more information. We need to
+ # upload the original file or else diff parsing will fail if the file was
+ # edited.
+ if status[0] == "A" and status[3] != "+":
+ # We'll need to upload the new content if we're adding a binary file
+ # since diff's output won't contain it.
+ mimetype = RunShell(["svn", "propget", "svn:mime-type", filename],
+ silent_ok=True)
+ base_content = ""
+ is_binary = mimetype and not mimetype.startswith("text/")
+ if is_binary and self.IsImage(filename):
+ new_content = self.ReadFile(filename)
+ elif (status[0] in ("M", "D", "R") or
+ (status[0] == "A" and status[3] == "+") or # Copied file.
+ (status[0] == " " and status[1] == "M")): # Property change.
+ args = []
+ if self.options.revision:
+ url = "%s/%s@%s" % (self.svn_base, filename, self.rev_start)
+ else:
+ # Don't change filename, it's needed later.
+ url = filename
+ args += ["-r", "BASE"]
+ cmd = ["svn"] + args + ["propget", "svn:mime-type", url]
+ mimetype, returncode = RunShellWithReturnCode(cmd)
+ if returncode:
+ # File does not exist in the requested revision.
+ # Reset mimetype, it contains an error message.
+ mimetype = ""
+ get_base = False
+ is_binary = mimetype and not mimetype.startswith("text/")
+ if status[0] == " ":
+ # Empty base content just to force an upload.
+ base_content = ""
+ elif is_binary:
+ if self.IsImage(filename):
+ get_base = True
+ if status[0] == "M":
+ if not self.rev_end:
+ new_content = self.ReadFile(filename)
+ else:
+ url = "%s/%s@%s" % (self.svn_base, filename, self.rev_end)
+ new_content = RunShell(["svn", "cat", url],
+ universal_newlines=True, silent_ok=True)
+ else:
+ base_content = ""
+ else:
+ get_base = True
+
+ if get_base:
+ if is_binary:
+ universal_newlines = False
+ else:
+ universal_newlines = True
+ if self.rev_start:
+ # "svn cat -r REV delete_file.txt" doesn't work. cat requires
+ # the full URL with "@REV" appended instead of using "-r" option.
+ url = "%s/%s@%s" % (self.svn_base, filename, self.rev_start)
+ base_content = RunShell(["svn", "cat", url],
+ universal_newlines=universal_newlines,
+ silent_ok=True)
+ else:
+ base_content = RunShell(["svn", "cat", filename],
+ universal_newlines=universal_newlines,
+ silent_ok=True)
+ if not is_binary:
+ args = []
+ if self.rev_start:
+ url = "%s/%s@%s" % (self.svn_base, filename, self.rev_start)
+ else:
+ url = filename
+ args += ["-r", "BASE"]
+ cmd = ["svn"] + args + ["propget", "svn:keywords", url]
+ keywords, returncode = RunShellWithReturnCode(cmd)
+ if keywords and not returncode:
+ base_content = self._CollapseKeywords(base_content, keywords)
+ else:
+ StatusUpdate("svn status returned unexpected output: %s" % status)
+ sys.exit(1)
+ return base_content, new_content, is_binary, status[0:5]
+
+
+class GitVCS(VersionControlSystem):
+ """Implementation of the VersionControlSystem interface for Git."""
+
+ def __init__(self, options):
+ super(GitVCS, self).__init__(options)
+ # Map of filename -> hash of base file.
+ self.base_hashes = {}
+
+ def GenerateDiff(self, extra_args):
+ # This is more complicated than svn's GenerateDiff because we must convert
+ # the diff output to include an svn-style "Index:" line as well as record
+ # the hashes of the base files, so we can upload them along with our diff.
+ if self.options.revision:
+ extra_args = [self.options.revision] + extra_args
+ gitdiff = RunShell(["git", "diff", "--full-index"] + extra_args)
+ svndiff = []
+ filecount = 0
+ filename = None
+ for line in gitdiff.splitlines():
+ match = re.match(r"diff --git a/(.*) b/.*$", line)
+ if match:
+ filecount += 1
+ filename = match.group(1)
+ svndiff.append("Index: %s\n" % filename)
+ else:
+ # The "index" line in a git diff looks like this (long hashes elided):
+ # index 82c0d44..b2cee3f 100755
+ # We want to save the left hash, as that identifies the base file.
+ match = re.match(r"index (\w+)\.\.", line)
+ if match:
+ self.base_hashes[filename] = match.group(1)
+ svndiff.append(line + "\n")
+ if not filecount:
+ ErrorExit("No valid patches found in output from git diff")
+ return "".join(svndiff)
+
+ def GetUnknownFiles(self):
+ status = RunShell(["git", "ls-files", "--exclude-standard", "--others"],
+ silent_ok=True)
+ return status.splitlines()
+
+ def GetBaseFile(self, filename):
+ hash = self.base_hashes[filename]
+ base_content = None
+ new_content = None
+ is_binary = False
+ if hash == "0" * 40: # All-zero hash indicates no base file.
+ status = "A"
+ base_content = ""
+ else:
+ status = "M"
+ base_content, returncode = RunShellWithReturnCode(["git", "show", hash])
+ if returncode:
+ ErrorExit("Got error status from 'git show %s'" % hash)
+ return (base_content, new_content, is_binary, status)
+
+
+class MercurialVCS(VersionControlSystem):
+ """Implementation of the VersionControlSystem interface for Mercurial."""
+
+ def __init__(self, options, repo_dir):
+ super(MercurialVCS, self).__init__(options)
+ # Absolute path to repository (we can be in a subdir)
+ self.repo_dir = os.path.normpath(repo_dir)
+ # Compute the subdir
+ cwd = os.path.normpath(os.getcwd())
+ assert cwd.startswith(self.repo_dir)
+ self.subdir = cwd[len(self.repo_dir):].lstrip(r"\/")
+ if self.options.revision:
+ self.base_rev = self.options.revision
+ else:
+ self.base_rev = RunShell(["hg", "parent", "-q"]).split(':')[1].strip()
+
+ def _GetRelPath(self, filename):
+ """Get relative path of a file according to the current directory,
+ given its logical path in the repo."""
+ assert filename.startswith(self.subdir), filename
+ return filename[len(self.subdir):].lstrip(r"\/")
+
+ def GenerateDiff(self, extra_args):
+ # If no file specified, restrict to the current subdir
+ extra_args = extra_args or ["."]
+ cmd = ["hg", "diff", "--git", "-r", self.base_rev] + extra_args
+ data = RunShell(cmd, silent_ok=True)
+ svndiff = []
+ filecount = 0
+ for line in data.splitlines():
+ m = re.match("diff --git a/(\S+) b/(\S+)", line)
+ if m:
+ # Modify line to make it look like as it comes from svn diff.
+ # With this modification no changes on the server side are required
+ # to make upload.py work with Mercurial repos.
+ # NOTE: for proper handling of moved/copied files, we have to use
+ # the second filename.
+ filename = m.group(2)
+ svndiff.append("Index: %s" % filename)
+ svndiff.append("=" * 67)
+ filecount += 1
+ logging.info(line)
+ else:
+ svndiff.append(line)
+ if not filecount:
+ ErrorExit("No valid patches found in output from hg diff")
+ return "\n".join(svndiff) + "\n"
+
+ def GetUnknownFiles(self):
+ """Return a list of files unknown to the VCS."""
+ args = []
+ status = RunShell(["hg", "status", "--rev", self.base_rev, "-u", "."],
+ silent_ok=True)
+ unknown_files = []
+ for line in status.splitlines():
+ st, fn = line.split(" ", 1)
+ if st == "?":
+ unknown_files.append(fn)
+ return unknown_files
+
+ def GetBaseFile(self, filename):
+ # "hg status" and "hg cat" both take a path relative to the current subdir
+ # rather than to the repo root, but "hg diff" has given us the full path
+ # to the repo root.
+ base_content = ""
+ new_content = None
+ is_binary = False
+ oldrelpath = relpath = self._GetRelPath(filename)
+ # "hg status -C" returns two lines for moved/copied files, one otherwise
+ out = RunShell(["hg", "status", "-C", "--rev", self.base_rev, relpath])
+ out = out.splitlines()
+ # HACK: strip error message about missing file/directory if it isn't in
+ # the working copy
+ if out[0].startswith('%s: ' % relpath):
+ out = out[1:]
+ if len(out) > 1:
+ # Moved/copied => considered as modified, use old filename to
+ # retrieve base contents
+ oldrelpath = out[1].strip()
+ status = "M"
+ else:
+ status, _ = out[0].split(' ', 1)
+ if status != "A":
+ base_content = RunShell(["hg", "cat", "-r", self.base_rev, oldrelpath],
+ silent_ok=True)
+ is_binary = "\0" in base_content # Mercurial's heuristic
+ if status != "R":
+ new_content = open(relpath, "rb").read()
+ is_binary = is_binary or "\0" in new_content
+ if is_binary and base_content:
+ # Fetch again without converting newlines
+ base_content = RunShell(["hg", "cat", "-r", self.base_rev, oldrelpath],
+ silent_ok=True, universal_newlines=False)
+ if not is_binary or not self.IsImage(relpath):
+ new_content = None
+ return base_content, new_content, is_binary, status
+
+
+# NOTE: The SplitPatch function is duplicated in engine.py, keep them in sync.
+def SplitPatch(data):
+ """Splits a patch into separate pieces for each file.
+
+ Args:
+ data: A string containing the output of svn diff.
+
+ Returns:
+ A list of 2-tuple (filename, text) where text is the svn diff output
+ pertaining to filename.
+ """
+ patches = []
+ filename = None
+ diff = []
+ for line in data.splitlines(True):
+ new_filename = None
+ if line.startswith('Index:'):
+ unused, new_filename = line.split(':', 1)
+ new_filename = new_filename.strip()
+ elif line.startswith('Property changes on:'):
+ unused, temp_filename = line.split(':', 1)
+ # When a file is modified, paths use '/' between directories, however
+ # when a property is modified '\' is used on Windows. Make them the same
+ # otherwise the file shows up twice.
+ temp_filename = temp_filename.strip().replace('\\', '/')
+ if temp_filename != filename:
+ # File has property changes but no modifications, create a new diff.
+ new_filename = temp_filename
+ if new_filename:
+ if filename and diff:
+ patches.append((filename, ''.join(diff)))
+ filename = new_filename
+ diff = [line]
+ continue
+ if diff is not None:
+ diff.append(line)
+ if filename and diff:
+ patches.append((filename, ''.join(diff)))
+ return patches
+
+
+def UploadSeparatePatches(issue, rpc_server, patchset, data, options):
+ """Uploads a separate patch for each file in the diff output.
+
+ Returns a list of [patch_key, filename] for each file.
+ """
+ patches = SplitPatch(data)
+ rv = []
+ for patch in patches:
+ if len(patch[1]) > MAX_UPLOAD_SIZE:
+ print ("Not uploading the patch for " + patch[0] +
+ " because the file is too large.")
+ continue
+ form_fields = [("filename", patch[0])]
+ if not options.download_base:
+ form_fields.append(("content_upload", "1"))
+ files = [("data", "data.diff", patch[1])]
+ ctype, body = EncodeMultipartFormData(form_fields, files)
+ url = "/%d/upload_patch/%d" % (int(issue), int(patchset))
+ print "Uploading patch for " + patch[0]
+ response_body = rpc_server.Send(url, body, content_type=ctype)
+ lines = response_body.splitlines()
+ if not lines or lines[0] != "OK":
+ StatusUpdate(" --> %s" % response_body)
+ sys.exit(1)
+ rv.append([lines[1], patch[0]])
+ return rv
+
+
+def GuessVCS(options):
+ """Helper to guess the version control system.
+
+ This examines the current directory, guesses which VersionControlSystem
+ we're using, and returns an instance of the appropriate class. Exit with an
+ error if we can't figure it out.
+
+ Returns:
+ A VersionControlSystem instance. Exits if the VCS can't be guessed.
+ """
+ # Mercurial has a command to get the base directory of a repository
+ # Try running it, but don't die if we don't have hg installed.
+ # NOTE: we try Mercurial first as it can sit on top of an SVN working copy.
+ try:
+ out, returncode = RunShellWithReturnCode(["hg", "root"])
+ if returncode == 0:
+ return MercurialVCS(options, out.strip())
+ except OSError, (errno, message):
+ if errno != 2: # ENOENT -- they don't have hg installed.
+ raise
+
+ # Subversion has a .svn in all working directories.
+ if os.path.isdir('.svn'):
+ logging.info("Guessed VCS = Subversion")
+ return SubversionVCS(options)
+
+ # Git has a command to test if you're in a git tree.
+ # Try running it, but don't die if we don't have git installed.
+ try:
+ out, returncode = RunShellWithReturnCode(["git", "rev-parse",
+ "--is-inside-work-tree"])
+ if returncode == 0:
+ return GitVCS(options)
+ except OSError, (errno, message):
+ if errno != 2: # ENOENT -- they don't have git installed.
+ raise
+
+ ErrorExit(("Could not guess version control system. "
+ "Are you in a working copy directory?"))
+
+
+def RealMain(argv, data=None):
+ """The real main function.
+
+ Args:
+ argv: Command line arguments.
+ data: Diff contents. If None (default) the diff is generated by
+ the VersionControlSystem implementation returned by GuessVCS().
+
+ Returns:
+ A 2-tuple (issue id, patchset id).
+ The patchset id is None if the base files are not uploaded by this
+ script (applies only to SVN checkouts).
+ """
+ logging.basicConfig(format=("%(asctime).19s %(levelname)s %(filename)s:"
+ "%(lineno)s %(message)s "))
+ os.environ['LC_ALL'] = 'C'
+ options, args = parser.parse_args(argv[1:])
+ global verbosity
+ verbosity = options.verbose
+ if verbosity >= 3:
+ logging.getLogger().setLevel(logging.DEBUG)
+ elif verbosity >= 2:
+ logging.getLogger().setLevel(logging.INFO)
+ vcs = GuessVCS(options)
+ if isinstance(vcs, SubversionVCS):
+ # base field is only allowed for Subversion.
+ # Note: Fetching base files may become deprecated in future releases.
+ base = vcs.GuessBase(options.download_base)
+ else:
+ base = None
+ if not base and options.download_base:
+ options.download_base = True
+ logging.info("Enabled upload of base file")
+ if not options.assume_yes:
+ vcs.CheckForUnknownFiles()
+ if data is None:
+ data = vcs.GenerateDiff(args)
+ files = vcs.GetBaseFiles(data)
+ if verbosity >= 1:
+ print "Upload server:", options.server, "(change with -s/--server)"
+ if options.issue:
+ prompt = "Message describing this patch set: "
+ else:
+ prompt = "New issue subject: "
+ message = options.message or raw_input(prompt).strip()
+ if not message:
+ ErrorExit("A non-empty message is required")
+ rpc_server = GetRpcServer(options)
+ form_fields = [("subject", message)]
+ if base:
+ form_fields.append(("base", base))
+ if options.issue:
+ form_fields.append(("issue", str(options.issue)))
+ if options.email:
+ form_fields.append(("user", options.email))
+ if options.reviewers:
+ for reviewer in options.reviewers.split(','):
+ if "@" in reviewer and not reviewer.split("@")[1].count(".") == 1:
+ ErrorExit("Invalid email address: %s" % reviewer)
+ form_fields.append(("reviewers", options.reviewers))
+ if options.cc:
+ for cc in options.cc.split(','):
+ if "@" in cc and not cc.split("@")[1].count(".") == 1:
+ ErrorExit("Invalid email address: %s" % cc)
+ form_fields.append(("cc", options.cc))
+ description = options.description
+ if options.description_file:
+ if options.description:
+ ErrorExit("Can't specify description and description_file")
+ file = open(options.description_file, 'r')
+ description = file.read()
+ file.close()
+ if description:
+ form_fields.append(("description", description))
+ # Send a hash of all the base file so the server can determine if a copy
+ # already exists in an earlier patchset.
+ base_hashes = ""
+ for file, info in files.iteritems():
+ if not info[0] is None:
+ checksum = md5.new(info[0]).hexdigest()
+ if base_hashes:
+ base_hashes += "|"
+ base_hashes += checksum + ":" + file
+ form_fields.append(("base_hashes", base_hashes))
+ # If we're uploading base files, don't send the email before the uploads, so
+ # that it contains the file status.
+ if options.send_mail and options.download_base:
+ form_fields.append(("send_mail", "1"))
+ if not options.download_base:
+ form_fields.append(("content_upload", "1"))
+ if len(data) > MAX_UPLOAD_SIZE:
+ print "Patch is large, so uploading file patches separately."
+ uploaded_diff_file = []
+ form_fields.append(("separate_patches", "1"))
+ else:
+ uploaded_diff_file = [("data", "data.diff", data)]
+ ctype, body = EncodeMultipartFormData(form_fields, uploaded_diff_file)
+ response_body = rpc_server.Send("/upload", body, content_type=ctype)
+ patchset = None
+ if not options.download_base or not uploaded_diff_file:
+ lines = response_body.splitlines()
+ if len(lines) >= 2:
+ msg = lines[0]
+ patchset = lines[1].strip()
+ patches = [x.split(" ", 1) for x in lines[2:]]
+ else:
+ msg = response_body
+ else:
+ msg = response_body
+ StatusUpdate(msg)
+ if not response_body.startswith("Issue created.") and \
+ not response_body.startswith("Issue updated."):
+ sys.exit(0)
+ issue = msg[msg.rfind("/")+1:]
+
+ if not uploaded_diff_file:
+ result = UploadSeparatePatches(issue, rpc_server, patchset, data, options)
+ if not options.download_base:
+ patches = result
+
+ if not options.download_base:
+ vcs.UploadBaseFiles(issue, rpc_server, patches, patchset, options, files)
+ if options.send_mail:
+ rpc_server.Send("/" + issue + "/mail", payload="")
+ return issue, patchset
+
+
+def main():
+ try:
+ RealMain(sys.argv)
+ except KeyboardInterrupt:
+ print
+ StatusUpdate("Interrupted.")
+ sys.exit(1)
+
+
+if __name__ == "__main__":
+ main()
diff --git a/libvpx/third_party/googletest/src/scripts/upload_gtest.py b/libvpx/third_party/googletest/src/scripts/upload_gtest.py
new file mode 100755
index 0000000..be19ae8
--- /dev/null
+++ b/libvpx/third_party/googletest/src/scripts/upload_gtest.py
@@ -0,0 +1,78 @@
+#!/usr/bin/env python
+#
+# Copyright 2009, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""upload_gtest.py v0.1.0 -- uploads a Google Test patch for review.
+
+This simple wrapper passes all command line flags and
+--cc=googletestframework@googlegroups.com to upload.py.
+
+USAGE: upload_gtest.py [options for upload.py]
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import sys
+
+CC_FLAG = '--cc='
+GTEST_GROUP = 'googletestframework@googlegroups.com'
+
+
+def main():
+ # Finds the path to upload.py, assuming it is in the same directory
+ # as this file.
+ my_dir = os.path.dirname(os.path.abspath(__file__))
+ upload_py_path = os.path.join(my_dir, 'upload.py')
+
+ # Adds Google Test discussion group to the cc line if it's not there
+ # already.
+ upload_py_argv = [upload_py_path]
+ found_cc_flag = False
+ for arg in sys.argv[1:]:
+ if arg.startswith(CC_FLAG):
+ found_cc_flag = True
+ cc_line = arg[len(CC_FLAG):]
+ cc_list = [addr for addr in cc_line.split(',') if addr]
+ if GTEST_GROUP not in cc_list:
+ cc_list.append(GTEST_GROUP)
+ upload_py_argv.append(CC_FLAG + ','.join(cc_list))
+ else:
+ upload_py_argv.append(arg)
+
+ if not found_cc_flag:
+ upload_py_argv.append(CC_FLAG + GTEST_GROUP)
+
+ # Invokes upload.py with the modified command line flags.
+ os.execv(upload_py_path, upload_py_argv)
+
+
+if __name__ == '__main__':
+ main()
diff --git a/libvpx/third_party/googletest/src/src/gtest-all.cc b/libvpx/third_party/googletest/src/src/gtest-all.cc
new file mode 100644
index 0000000..0a9cee5
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest-all.cc
@@ -0,0 +1,48 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+// Google C++ Testing Framework (Google Test)
+//
+// Sometimes it's desirable to build Google Test by compiling a single file.
+// This file serves this purpose.
+
+// This line ensures that gtest.h can be compiled on its own, even
+// when it's fused.
+#include "gtest/gtest.h"
+
+// The following lines pull in the real gtest *.cc files.
+#include "src/gtest.cc"
+#include "src/gtest-death-test.cc"
+#include "src/gtest-filepath.cc"
+#include "src/gtest-port.cc"
+#include "src/gtest-printers.cc"
+#include "src/gtest-test-part.cc"
+#include "src/gtest-typed-test.cc"
diff --git a/libvpx/third_party/googletest/src/src/gtest-death-test.cc b/libvpx/third_party/googletest/src/src/gtest-death-test.cc
new file mode 100644
index 0000000..8b2e413
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest-death-test.cc
@@ -0,0 +1,1234 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan), vladl@google.com (Vlad Losev)
+//
+// This file implements death tests.
+
+#include "gtest/gtest-death-test.h"
+#include "gtest/internal/gtest-port.h"
+
+#if GTEST_HAS_DEATH_TEST
+
+# if GTEST_OS_MAC
+# include <crt_externs.h>
+# endif // GTEST_OS_MAC
+
+# include <errno.h>
+# include <fcntl.h>
+# include <limits.h>
+# include <stdarg.h>
+
+# if GTEST_OS_WINDOWS
+# include <windows.h>
+# else
+# include <sys/mman.h>
+# include <sys/wait.h>
+# endif // GTEST_OS_WINDOWS
+
+#endif // GTEST_HAS_DEATH_TEST
+
+#include "gtest/gtest-message.h"
+#include "gtest/internal/gtest-string.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+
+// Constants.
+
+// The default death test style.
+static const char kDefaultDeathTestStyle[] = "fast";
+
+GTEST_DEFINE_string_(
+ death_test_style,
+ internal::StringFromGTestEnv("death_test_style", kDefaultDeathTestStyle),
+ "Indicates how to run a death test in a forked child process: "
+ "\"threadsafe\" (child process re-executes the test binary "
+ "from the beginning, running only the specific death test) or "
+ "\"fast\" (child process runs the death test immediately "
+ "after forking).");
+
+GTEST_DEFINE_bool_(
+ death_test_use_fork,
+ internal::BoolFromGTestEnv("death_test_use_fork", false),
+ "Instructs to use fork()/_exit() instead of clone() in death tests. "
+ "Ignored and always uses fork() on POSIX systems where clone() is not "
+ "implemented. Useful when running under valgrind or similar tools if "
+ "those do not support clone(). Valgrind 3.3.1 will just fail if "
+ "it sees an unsupported combination of clone() flags. "
+ "It is not recommended to use this flag w/o valgrind though it will "
+ "work in 99% of the cases. Once valgrind is fixed, this flag will "
+ "most likely be removed.");
+
+namespace internal {
+GTEST_DEFINE_string_(
+ internal_run_death_test, "",
+ "Indicates the file, line number, temporal index of "
+ "the single death test to run, and a file descriptor to "
+ "which a success code may be sent, all separated by "
+ "colons. This flag is specified if and only if the current "
+ "process is a sub-process launched for running a thread-safe "
+ "death test. FOR INTERNAL USE ONLY.");
+} // namespace internal
+
+#if GTEST_HAS_DEATH_TEST
+
+// ExitedWithCode constructor.
+ExitedWithCode::ExitedWithCode(int exit_code) : exit_code_(exit_code) {
+}
+
+// ExitedWithCode function-call operator.
+bool ExitedWithCode::operator()(int exit_status) const {
+# if GTEST_OS_WINDOWS
+
+ return exit_status == exit_code_;
+
+# else
+
+ return WIFEXITED(exit_status) && WEXITSTATUS(exit_status) == exit_code_;
+
+# endif // GTEST_OS_WINDOWS
+}
+
+# if !GTEST_OS_WINDOWS
+// KilledBySignal constructor.
+KilledBySignal::KilledBySignal(int signum) : signum_(signum) {
+}
+
+// KilledBySignal function-call operator.
+bool KilledBySignal::operator()(int exit_status) const {
+ return WIFSIGNALED(exit_status) && WTERMSIG(exit_status) == signum_;
+}
+# endif // !GTEST_OS_WINDOWS
+
+namespace internal {
+
+// Utilities needed for death tests.
+
+// Generates a textual description of a given exit code, in the format
+// specified by wait(2).
+static String ExitSummary(int exit_code) {
+ Message m;
+
+# if GTEST_OS_WINDOWS
+
+ m << "Exited with exit status " << exit_code;
+
+# else
+
+ if (WIFEXITED(exit_code)) {
+ m << "Exited with exit status " << WEXITSTATUS(exit_code);
+ } else if (WIFSIGNALED(exit_code)) {
+ m << "Terminated by signal " << WTERMSIG(exit_code);
+ }
+# ifdef WCOREDUMP
+ if (WCOREDUMP(exit_code)) {
+ m << " (core dumped)";
+ }
+# endif
+# endif // GTEST_OS_WINDOWS
+
+ return m.GetString();
+}
+
+// Returns true if exit_status describes a process that was terminated
+// by a signal, or exited normally with a nonzero exit code.
+bool ExitedUnsuccessfully(int exit_status) {
+ return !ExitedWithCode(0)(exit_status);
+}
+
+# if !GTEST_OS_WINDOWS
+// Generates a textual failure message when a death test finds more than
+// one thread running, or cannot determine the number of threads, prior
+// to executing the given statement. It is the responsibility of the
+// caller not to pass a thread_count of 1.
+static String DeathTestThreadWarning(size_t thread_count) {
+ Message msg;
+ msg << "Death tests use fork(), which is unsafe particularly"
+ << " in a threaded context. For this test, " << GTEST_NAME_ << " ";
+ if (thread_count == 0)
+ msg << "couldn't detect the number of threads.";
+ else
+ msg << "detected " << thread_count << " threads.";
+ return msg.GetString();
+}
+# endif // !GTEST_OS_WINDOWS
+
+// Flag characters for reporting a death test that did not die.
+static const char kDeathTestLived = 'L';
+static const char kDeathTestReturned = 'R';
+static const char kDeathTestThrew = 'T';
+static const char kDeathTestInternalError = 'I';
+
+// An enumeration describing all of the possible ways that a death test can
+// conclude. DIED means that the process died while executing the test
+// code; LIVED means that process lived beyond the end of the test code;
+// RETURNED means that the test statement attempted to execute a return
+// statement, which is not allowed; THREW means that the test statement
+// returned control by throwing an exception. IN_PROGRESS means the test
+// has not yet concluded.
+// TODO(vladl@google.com): Unify names and possibly values for
+// AbortReason, DeathTestOutcome, and flag characters above.
+enum DeathTestOutcome { IN_PROGRESS, DIED, LIVED, RETURNED, THREW };
+
+// Routine for aborting the program which is safe to call from an
+// exec-style death test child process, in which case the error
+// message is propagated back to the parent process. Otherwise, the
+// message is simply printed to stderr. In either case, the program
+// then exits with status 1.
+void DeathTestAbort(const String& message) {
+ // On a POSIX system, this function may be called from a threadsafe-style
+ // death test child process, which operates on a very small stack. Use
+ // the heap for any additional non-minuscule memory requirements.
+ const InternalRunDeathTestFlag* const flag =
+ GetUnitTestImpl()->internal_run_death_test_flag();
+ if (flag != NULL) {
+ FILE* parent = posix::FDOpen(flag->write_fd(), "w");
+ fputc(kDeathTestInternalError, parent);
+ fprintf(parent, "%s", message.c_str());
+ fflush(parent);
+ _exit(1);
+ } else {
+ fprintf(stderr, "%s", message.c_str());
+ fflush(stderr);
+ posix::Abort();
+ }
+}
+
+// A replacement for CHECK that calls DeathTestAbort if the assertion
+// fails.
+# define GTEST_DEATH_TEST_CHECK_(expression) \
+ do { \
+ if (!::testing::internal::IsTrue(expression)) { \
+ DeathTestAbort(::testing::internal::String::Format( \
+ "CHECK failed: File %s, line %d: %s", \
+ __FILE__, __LINE__, #expression)); \
+ } \
+ } while (::testing::internal::AlwaysFalse())
+
+// This macro is similar to GTEST_DEATH_TEST_CHECK_, but it is meant for
+// evaluating any system call that fulfills two conditions: it must return
+// -1 on failure, and set errno to EINTR when it is interrupted and
+// should be tried again. The macro expands to a loop that repeatedly
+// evaluates the expression as long as it evaluates to -1 and sets
+// errno to EINTR. If the expression evaluates to -1 but errno is
+// something other than EINTR, DeathTestAbort is called.
+# define GTEST_DEATH_TEST_CHECK_SYSCALL_(expression) \
+ do { \
+ int gtest_retval; \
+ do { \
+ gtest_retval = (expression); \
+ } while (gtest_retval == -1 && errno == EINTR); \
+ if (gtest_retval == -1) { \
+ DeathTestAbort(::testing::internal::String::Format( \
+ "CHECK failed: File %s, line %d: %s != -1", \
+ __FILE__, __LINE__, #expression)); \
+ } \
+ } while (::testing::internal::AlwaysFalse())
+
+// Returns the message describing the last system error in errno.
+String GetLastErrnoDescription() {
+ return String(errno == 0 ? "" : posix::StrError(errno));
+}
+
+// This is called from a death test parent process to read a failure
+// message from the death test child process and log it with the FATAL
+// severity. On Windows, the message is read from a pipe handle. On other
+// platforms, it is read from a file descriptor.
+static void FailFromInternalError(int fd) {
+ Message error;
+ char buffer[256];
+ int num_read;
+
+ do {
+ while ((num_read = posix::Read(fd, buffer, 255)) > 0) {
+ buffer[num_read] = '\0';
+ error << buffer;
+ }
+ } while (num_read == -1 && errno == EINTR);
+
+ if (num_read == 0) {
+ GTEST_LOG_(FATAL) << error.GetString();
+ } else {
+ const int last_error = errno;
+ GTEST_LOG_(FATAL) << "Error while reading death test internal: "
+ << GetLastErrnoDescription() << " [" << last_error << "]";
+ }
+}
+
+// Death test constructor. Increments the running death test count
+// for the current test.
+DeathTest::DeathTest() {
+ TestInfo* const info = GetUnitTestImpl()->current_test_info();
+ if (info == NULL) {
+ DeathTestAbort("Cannot run a death test outside of a TEST or "
+ "TEST_F construct");
+ }
+}
+
+// Creates and returns a death test by dispatching to the current
+// death test factory.
+bool DeathTest::Create(const char* statement, const RE* regex,
+ const char* file, int line, DeathTest** test) {
+ return GetUnitTestImpl()->death_test_factory()->Create(
+ statement, regex, file, line, test);
+}
+
+const char* DeathTest::LastMessage() {
+ return last_death_test_message_.c_str();
+}
+
+void DeathTest::set_last_death_test_message(const String& message) {
+ last_death_test_message_ = message;
+}
+
+String DeathTest::last_death_test_message_;
+
+// Provides cross platform implementation for some death functionality.
+class DeathTestImpl : public DeathTest {
+ protected:
+ DeathTestImpl(const char* a_statement, const RE* a_regex)
+ : statement_(a_statement),
+ regex_(a_regex),
+ spawned_(false),
+ status_(-1),
+ outcome_(IN_PROGRESS),
+ read_fd_(-1),
+ write_fd_(-1) {}
+
+ // read_fd_ is expected to be closed and cleared by a derived class.
+ ~DeathTestImpl() { GTEST_DEATH_TEST_CHECK_(read_fd_ == -1); }
+
+ void Abort(AbortReason reason);
+ virtual bool Passed(bool status_ok);
+
+ const char* statement() const { return statement_; }
+ const RE* regex() const { return regex_; }
+ bool spawned() const { return spawned_; }
+ void set_spawned(bool is_spawned) { spawned_ = is_spawned; }
+ int status() const { return status_; }
+ void set_status(int a_status) { status_ = a_status; }
+ DeathTestOutcome outcome() const { return outcome_; }
+ void set_outcome(DeathTestOutcome an_outcome) { outcome_ = an_outcome; }
+ int read_fd() const { return read_fd_; }
+ void set_read_fd(int fd) { read_fd_ = fd; }
+ int write_fd() const { return write_fd_; }
+ void set_write_fd(int fd) { write_fd_ = fd; }
+
+ // Called in the parent process only. Reads the result code of the death
+ // test child process via a pipe, interprets it to set the outcome_
+ // member, and closes read_fd_. Outputs diagnostics and terminates in
+ // case of unexpected codes.
+ void ReadAndInterpretStatusByte();
+
+ private:
+ // The textual content of the code this object is testing. This class
+ // doesn't own this string and should not attempt to delete it.
+ const char* const statement_;
+ // The regular expression which test output must match. DeathTestImpl
+ // doesn't own this object and should not attempt to delete it.
+ const RE* const regex_;
+ // True if the death test child process has been successfully spawned.
+ bool spawned_;
+ // The exit status of the child process.
+ int status_;
+ // How the death test concluded.
+ DeathTestOutcome outcome_;
+ // Descriptor to the read end of the pipe to the child process. It is
+ // always -1 in the child process. The child keeps its write end of the
+ // pipe in write_fd_.
+ int read_fd_;
+ // Descriptor to the child's write end of the pipe to the parent process.
+ // It is always -1 in the parent process. The parent keeps its end of the
+ // pipe in read_fd_.
+ int write_fd_;
+};
+
+// Called in the parent process only. Reads the result code of the death
+// test child process via a pipe, interprets it to set the outcome_
+// member, and closes read_fd_. Outputs diagnostics and terminates in
+// case of unexpected codes.
+void DeathTestImpl::ReadAndInterpretStatusByte() {
+ char flag;
+ int bytes_read;
+
+ // The read() here blocks until data is available (signifying the
+ // failure of the death test) or until the pipe is closed (signifying
+ // its success), so it's okay to call this in the parent before
+ // the child process has exited.
+ do {
+ bytes_read = posix::Read(read_fd(), &flag, 1);
+ } while (bytes_read == -1 && errno == EINTR);
+
+ if (bytes_read == 0) {
+ set_outcome(DIED);
+ } else if (bytes_read == 1) {
+ switch (flag) {
+ case kDeathTestReturned:
+ set_outcome(RETURNED);
+ break;
+ case kDeathTestThrew:
+ set_outcome(THREW);
+ break;
+ case kDeathTestLived:
+ set_outcome(LIVED);
+ break;
+ case kDeathTestInternalError:
+ FailFromInternalError(read_fd()); // Does not return.
+ break;
+ default:
+ GTEST_LOG_(FATAL) << "Death test child process reported "
+ << "unexpected status byte ("
+ << static_cast<unsigned int>(flag) << ")";
+ }
+ } else {
+ GTEST_LOG_(FATAL) << "Read from death test child process failed: "
+ << GetLastErrnoDescription();
+ }
+ GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Close(read_fd()));
+ set_read_fd(-1);
+}
+
+// Signals that the death test code which should have exited, didn't.
+// Should be called only in a death test child process.
+// Writes a status byte to the child's status file descriptor, then
+// calls _exit(1).
+void DeathTestImpl::Abort(AbortReason reason) {
+ // The parent process considers the death test to be a failure if
+ // it finds any data in our pipe. So, here we write a single flag byte
+ // to the pipe, then exit.
+ const char status_ch =
+ reason == TEST_DID_NOT_DIE ? kDeathTestLived :
+ reason == TEST_THREW_EXCEPTION ? kDeathTestThrew : kDeathTestReturned;
+
+ GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Write(write_fd(), &status_ch, 1));
+ // We are leaking the descriptor here because on some platforms (i.e.,
+ // when built as Windows DLL), destructors of global objects will still
+ // run after calling _exit(). On such systems, write_fd_ will be
+ // indirectly closed from the destructor of UnitTestImpl, causing double
+ // close if it is also closed here. On debug configurations, double close
+ // may assert. As there are no in-process buffers to flush here, we are
+ // relying on the OS to close the descriptor after the process terminates
+ // when the destructors are not run.
+ _exit(1); // Exits w/o any normal exit hooks (we were supposed to crash)
+}
+
+// Returns an indented copy of stderr output for a death test.
+// This makes distinguishing death test output lines from regular log lines
+// much easier.
+static ::std::string FormatDeathTestOutput(const ::std::string& output) {
+ ::std::string ret;
+ for (size_t at = 0; ; ) {
+ const size_t line_end = output.find('\n', at);
+ ret += "[ DEATH ] ";
+ if (line_end == ::std::string::npos) {
+ ret += output.substr(at);
+ break;
+ }
+ ret += output.substr(at, line_end + 1 - at);
+ at = line_end + 1;
+ }
+ return ret;
+}
+
+// Assesses the success or failure of a death test, using both private
+// members which have previously been set, and one argument:
+//
+// Private data members:
+// outcome: An enumeration describing how the death test
+// concluded: DIED, LIVED, THREW, or RETURNED. The death test
+// fails in the latter three cases.
+// status: The exit status of the child process. On *nix, it is in the
+// in the format specified by wait(2). On Windows, this is the
+// value supplied to the ExitProcess() API or a numeric code
+// of the exception that terminated the program.
+// regex: A regular expression object to be applied to
+// the test's captured standard error output; the death test
+// fails if it does not match.
+//
+// Argument:
+// status_ok: true if exit_status is acceptable in the context of
+// this particular death test, which fails if it is false
+//
+// Returns true iff all of the above conditions are met. Otherwise, the
+// first failing condition, in the order given above, is the one that is
+// reported. Also sets the last death test message string.
+bool DeathTestImpl::Passed(bool status_ok) {
+ if (!spawned())
+ return false;
+
+ const String error_message = GetCapturedStderr();
+
+ bool success = false;
+ Message buffer;
+
+ buffer << "Death test: " << statement() << "\n";
+ switch (outcome()) {
+ case LIVED:
+ buffer << " Result: failed to die.\n"
+ << " Error msg:\n" << FormatDeathTestOutput(error_message);
+ break;
+ case THREW:
+ buffer << " Result: threw an exception.\n"
+ << " Error msg:\n" << FormatDeathTestOutput(error_message);
+ break;
+ case RETURNED:
+ buffer << " Result: illegal return in test statement.\n"
+ << " Error msg:\n" << FormatDeathTestOutput(error_message);
+ break;
+ case DIED:
+ if (status_ok) {
+ const bool matched = RE::PartialMatch(error_message.c_str(), *regex());
+ if (matched) {
+ success = true;
+ } else {
+ buffer << " Result: died but not with expected error.\n"
+ << " Expected: " << regex()->pattern() << "\n"
+ << "Actual msg:\n" << FormatDeathTestOutput(error_message);
+ }
+ } else {
+ buffer << " Result: died but not with expected exit code:\n"
+ << " " << ExitSummary(status()) << "\n"
+ << "Actual msg:\n" << FormatDeathTestOutput(error_message);
+ }
+ break;
+ case IN_PROGRESS:
+ default:
+ GTEST_LOG_(FATAL)
+ << "DeathTest::Passed somehow called before conclusion of test";
+ }
+
+ DeathTest::set_last_death_test_message(buffer.GetString());
+ return success;
+}
+
+# if GTEST_OS_WINDOWS
+// WindowsDeathTest implements death tests on Windows. Due to the
+// specifics of starting new processes on Windows, death tests there are
+// always threadsafe, and Google Test considers the
+// --gtest_death_test_style=fast setting to be equivalent to
+// --gtest_death_test_style=threadsafe there.
+//
+// A few implementation notes: Like the Linux version, the Windows
+// implementation uses pipes for child-to-parent communication. But due to
+// the specifics of pipes on Windows, some extra steps are required:
+//
+// 1. The parent creates a communication pipe and stores handles to both
+// ends of it.
+// 2. The parent starts the child and provides it with the information
+// necessary to acquire the handle to the write end of the pipe.
+// 3. The child acquires the write end of the pipe and signals the parent
+// using a Windows event.
+// 4. Now the parent can release the write end of the pipe on its side. If
+// this is done before step 3, the object's reference count goes down to
+// 0 and it is destroyed, preventing the child from acquiring it. The
+// parent now has to release it, or read operations on the read end of
+// the pipe will not return when the child terminates.
+// 5. The parent reads child's output through the pipe (outcome code and
+// any possible error messages) from the pipe, and its stderr and then
+// determines whether to fail the test.
+//
+// Note: to distinguish Win32 API calls from the local method and function
+// calls, the former are explicitly resolved in the global namespace.
+//
+class WindowsDeathTest : public DeathTestImpl {
+ public:
+ WindowsDeathTest(const char* a_statement,
+ const RE* a_regex,
+ const char* file,
+ int line)
+ : DeathTestImpl(a_statement, a_regex), file_(file), line_(line) {}
+
+ // All of these virtual functions are inherited from DeathTest.
+ virtual int Wait();
+ virtual TestRole AssumeRole();
+
+ private:
+ // The name of the file in which the death test is located.
+ const char* const file_;
+ // The line number on which the death test is located.
+ const int line_;
+ // Handle to the write end of the pipe to the child process.
+ AutoHandle write_handle_;
+ // Child process handle.
+ AutoHandle child_handle_;
+ // Event the child process uses to signal the parent that it has
+ // acquired the handle to the write end of the pipe. After seeing this
+ // event the parent can release its own handles to make sure its
+ // ReadFile() calls return when the child terminates.
+ AutoHandle event_handle_;
+};
+
+// Waits for the child in a death test to exit, returning its exit
+// status, or 0 if no child process exists. As a side effect, sets the
+// outcome data member.
+int WindowsDeathTest::Wait() {
+ if (!spawned())
+ return 0;
+
+ // Wait until the child either signals that it has acquired the write end
+ // of the pipe or it dies.
+ const HANDLE wait_handles[2] = { child_handle_.Get(), event_handle_.Get() };
+ switch (::WaitForMultipleObjects(2,
+ wait_handles,
+ FALSE, // Waits for any of the handles.
+ INFINITE)) {
+ case WAIT_OBJECT_0:
+ case WAIT_OBJECT_0 + 1:
+ break;
+ default:
+ GTEST_DEATH_TEST_CHECK_(false); // Should not get here.
+ }
+
+ // The child has acquired the write end of the pipe or exited.
+ // We release the handle on our side and continue.
+ write_handle_.Reset();
+ event_handle_.Reset();
+
+ ReadAndInterpretStatusByte();
+
+ // Waits for the child process to exit if it haven't already. This
+ // returns immediately if the child has already exited, regardless of
+ // whether previous calls to WaitForMultipleObjects synchronized on this
+ // handle or not.
+ GTEST_DEATH_TEST_CHECK_(
+ WAIT_OBJECT_0 == ::WaitForSingleObject(child_handle_.Get(),
+ INFINITE));
+ DWORD status_code;
+ GTEST_DEATH_TEST_CHECK_(
+ ::GetExitCodeProcess(child_handle_.Get(), &status_code) != FALSE);
+ child_handle_.Reset();
+ set_status(static_cast<int>(status_code));
+ return status();
+}
+
+// The AssumeRole process for a Windows death test. It creates a child
+// process with the same executable as the current process to run the
+// death test. The child process is given the --gtest_filter and
+// --gtest_internal_run_death_test flags such that it knows to run the
+// current death test only.
+DeathTest::TestRole WindowsDeathTest::AssumeRole() {
+ const UnitTestImpl* const impl = GetUnitTestImpl();
+ const InternalRunDeathTestFlag* const flag =
+ impl->internal_run_death_test_flag();
+ const TestInfo* const info = impl->current_test_info();
+ const int death_test_index = info->result()->death_test_count();
+
+ if (flag != NULL) {
+ // ParseInternalRunDeathTestFlag() has performed all the necessary
+ // processing.
+ set_write_fd(flag->write_fd());
+ return EXECUTE_TEST;
+ }
+
+ // WindowsDeathTest uses an anonymous pipe to communicate results of
+ // a death test.
+ SECURITY_ATTRIBUTES handles_are_inheritable = {
+ sizeof(SECURITY_ATTRIBUTES), NULL, TRUE };
+ HANDLE read_handle, write_handle;
+ GTEST_DEATH_TEST_CHECK_(
+ ::CreatePipe(&read_handle, &write_handle, &handles_are_inheritable,
+ 0) // Default buffer size.
+ != FALSE);
+ set_read_fd(::_open_osfhandle(reinterpret_cast<intptr_t>(read_handle),
+ O_RDONLY));
+ write_handle_.Reset(write_handle);
+ event_handle_.Reset(::CreateEvent(
+ &handles_are_inheritable,
+ TRUE, // The event will automatically reset to non-signaled state.
+ FALSE, // The initial state is non-signalled.
+ NULL)); // The even is unnamed.
+ GTEST_DEATH_TEST_CHECK_(event_handle_.Get() != NULL);
+ const String filter_flag = String::Format("--%s%s=%s.%s",
+ GTEST_FLAG_PREFIX_, kFilterFlag,
+ info->test_case_name(),
+ info->name());
+ const String internal_flag = String::Format(
+ "--%s%s=%s|%d|%d|%u|%Iu|%Iu",
+ GTEST_FLAG_PREFIX_,
+ kInternalRunDeathTestFlag,
+ file_, line_,
+ death_test_index,
+ static_cast<unsigned int>(::GetCurrentProcessId()),
+ // size_t has the same with as pointers on both 32-bit and 64-bit
+ // Windows platforms.
+ // See http://msdn.microsoft.com/en-us/library/tcxf1dw6.aspx.
+ reinterpret_cast<size_t>(write_handle),
+ reinterpret_cast<size_t>(event_handle_.Get()));
+
+ char executable_path[_MAX_PATH + 1]; // NOLINT
+ GTEST_DEATH_TEST_CHECK_(
+ _MAX_PATH + 1 != ::GetModuleFileNameA(NULL,
+ executable_path,
+ _MAX_PATH));
+
+ String command_line = String::Format("%s %s \"%s\"",
+ ::GetCommandLineA(),
+ filter_flag.c_str(),
+ internal_flag.c_str());
+
+ DeathTest::set_last_death_test_message("");
+
+ CaptureStderr();
+ // Flush the log buffers since the log streams are shared with the child.
+ FlushInfoLog();
+
+ // The child process will share the standard handles with the parent.
+ STARTUPINFOA startup_info;
+ memset(&startup_info, 0, sizeof(STARTUPINFO));
+ startup_info.dwFlags = STARTF_USESTDHANDLES;
+ startup_info.hStdInput = ::GetStdHandle(STD_INPUT_HANDLE);
+ startup_info.hStdOutput = ::GetStdHandle(STD_OUTPUT_HANDLE);
+ startup_info.hStdError = ::GetStdHandle(STD_ERROR_HANDLE);
+
+ PROCESS_INFORMATION process_info;
+ GTEST_DEATH_TEST_CHECK_(::CreateProcessA(
+ executable_path,
+ const_cast<char*>(command_line.c_str()),
+ NULL, // Retuned process handle is not inheritable.
+ NULL, // Retuned thread handle is not inheritable.
+ TRUE, // Child inherits all inheritable handles (for write_handle_).
+ 0x0, // Default creation flags.
+ NULL, // Inherit the parent's environment.
+ UnitTest::GetInstance()->original_working_dir(),
+ &startup_info,
+ &process_info) != FALSE);
+ child_handle_.Reset(process_info.hProcess);
+ ::CloseHandle(process_info.hThread);
+ set_spawned(true);
+ return OVERSEE_TEST;
+}
+# else // We are not on Windows.
+
+// ForkingDeathTest provides implementations for most of the abstract
+// methods of the DeathTest interface. Only the AssumeRole method is
+// left undefined.
+class ForkingDeathTest : public DeathTestImpl {
+ public:
+ ForkingDeathTest(const char* statement, const RE* regex);
+
+ // All of these virtual functions are inherited from DeathTest.
+ virtual int Wait();
+
+ protected:
+ void set_child_pid(pid_t child_pid) { child_pid_ = child_pid; }
+
+ private:
+ // PID of child process during death test; 0 in the child process itself.
+ pid_t child_pid_;
+};
+
+// Constructs a ForkingDeathTest.
+ForkingDeathTest::ForkingDeathTest(const char* a_statement, const RE* a_regex)
+ : DeathTestImpl(a_statement, a_regex),
+ child_pid_(-1) {}
+
+// Waits for the child in a death test to exit, returning its exit
+// status, or 0 if no child process exists. As a side effect, sets the
+// outcome data member.
+int ForkingDeathTest::Wait() {
+ if (!spawned())
+ return 0;
+
+ ReadAndInterpretStatusByte();
+
+ int status_value;
+ GTEST_DEATH_TEST_CHECK_SYSCALL_(waitpid(child_pid_, &status_value, 0));
+ set_status(status_value);
+ return status_value;
+}
+
+// A concrete death test class that forks, then immediately runs the test
+// in the child process.
+class NoExecDeathTest : public ForkingDeathTest {
+ public:
+ NoExecDeathTest(const char* a_statement, const RE* a_regex) :
+ ForkingDeathTest(a_statement, a_regex) { }
+ virtual TestRole AssumeRole();
+};
+
+// The AssumeRole process for a fork-and-run death test. It implements a
+// straightforward fork, with a simple pipe to transmit the status byte.
+DeathTest::TestRole NoExecDeathTest::AssumeRole() {
+ const size_t thread_count = GetThreadCount();
+ if (thread_count != 1) {
+ GTEST_LOG_(WARNING) << DeathTestThreadWarning(thread_count);
+ }
+
+ int pipe_fd[2];
+ GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1);
+
+ DeathTest::set_last_death_test_message("");
+ CaptureStderr();
+ // When we fork the process below, the log file buffers are copied, but the
+ // file descriptors are shared. We flush all log files here so that closing
+ // the file descriptors in the child process doesn't throw off the
+ // synchronization between descriptors and buffers in the parent process.
+ // This is as close to the fork as possible to avoid a race condition in case
+ // there are multiple threads running before the death test, and another
+ // thread writes to the log file.
+ FlushInfoLog();
+
+ const pid_t child_pid = fork();
+ GTEST_DEATH_TEST_CHECK_(child_pid != -1);
+ set_child_pid(child_pid);
+ if (child_pid == 0) {
+ GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[0]));
+ set_write_fd(pipe_fd[1]);
+ // Redirects all logging to stderr in the child process to prevent
+ // concurrent writes to the log files. We capture stderr in the parent
+ // process and append the child process' output to a log.
+ LogToStderr();
+ // Event forwarding to the listeners of event listener API mush be shut
+ // down in death test subprocesses.
+ GetUnitTestImpl()->listeners()->SuppressEventForwarding();
+ return EXECUTE_TEST;
+ } else {
+ GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1]));
+ set_read_fd(pipe_fd[0]);
+ set_spawned(true);
+ return OVERSEE_TEST;
+ }
+}
+
+// A concrete death test class that forks and re-executes the main
+// program from the beginning, with command-line flags set that cause
+// only this specific death test to be run.
+class ExecDeathTest : public ForkingDeathTest {
+ public:
+ ExecDeathTest(const char* a_statement, const RE* a_regex,
+ const char* file, int line) :
+ ForkingDeathTest(a_statement, a_regex), file_(file), line_(line) { }
+ virtual TestRole AssumeRole();
+ private:
+ // The name of the file in which the death test is located.
+ const char* const file_;
+ // The line number on which the death test is located.
+ const int line_;
+};
+
+// Utility class for accumulating command-line arguments.
+class Arguments {
+ public:
+ Arguments() {
+ args_.push_back(NULL);
+ }
+
+ ~Arguments() {
+ for (std::vector<char*>::iterator i = args_.begin(); i != args_.end();
+ ++i) {
+ free(*i);
+ }
+ }
+ void AddArgument(const char* argument) {
+ args_.insert(args_.end() - 1, posix::StrDup(argument));
+ }
+
+ template <typename Str>
+ void AddArguments(const ::std::vector<Str>& arguments) {
+ for (typename ::std::vector<Str>::const_iterator i = arguments.begin();
+ i != arguments.end();
+ ++i) {
+ args_.insert(args_.end() - 1, posix::StrDup(i->c_str()));
+ }
+ }
+ char* const* Argv() {
+ return &args_[0];
+ }
+ private:
+ std::vector<char*> args_;
+};
+
+// A struct that encompasses the arguments to the child process of a
+// threadsafe-style death test process.
+struct ExecDeathTestArgs {
+ char* const* argv; // Command-line arguments for the child's call to exec
+ int close_fd; // File descriptor to close; the read end of a pipe
+};
+
+# if GTEST_OS_MAC
+inline char** GetEnviron() {
+ // When Google Test is built as a framework on MacOS X, the environ variable
+ // is unavailable. Apple's documentation (man environ) recommends using
+ // _NSGetEnviron() instead.
+ return *_NSGetEnviron();
+}
+# else
+// Some POSIX platforms expect you to declare environ. extern "C" makes
+// it reside in the global namespace.
+extern "C" char** environ;
+inline char** GetEnviron() { return environ; }
+# endif // GTEST_OS_MAC
+
+// The main function for a threadsafe-style death test child process.
+// This function is called in a clone()-ed process and thus must avoid
+// any potentially unsafe operations like malloc or libc functions.
+static int ExecDeathTestChildMain(void* child_arg) {
+ ExecDeathTestArgs* const args = static_cast<ExecDeathTestArgs*>(child_arg);
+ GTEST_DEATH_TEST_CHECK_SYSCALL_(close(args->close_fd));
+
+ // We need to execute the test program in the same environment where
+ // it was originally invoked. Therefore we change to the original
+ // working directory first.
+ const char* const original_dir =
+ UnitTest::GetInstance()->original_working_dir();
+ // We can safely call chdir() as it's a direct system call.
+ if (chdir(original_dir) != 0) {
+ DeathTestAbort(String::Format("chdir(\"%s\") failed: %s",
+ original_dir,
+ GetLastErrnoDescription().c_str()));
+ return EXIT_FAILURE;
+ }
+
+ // We can safely call execve() as it's a direct system call. We
+ // cannot use execvp() as it's a libc function and thus potentially
+ // unsafe. Since execve() doesn't search the PATH, the user must
+ // invoke the test program via a valid path that contains at least
+ // one path separator.
+ execve(args->argv[0], args->argv, GetEnviron());
+ DeathTestAbort(String::Format("execve(%s, ...) in %s failed: %s",
+ args->argv[0],
+ original_dir,
+ GetLastErrnoDescription().c_str()));
+ return EXIT_FAILURE;
+}
+
+// Two utility routines that together determine the direction the stack
+// grows.
+// This could be accomplished more elegantly by a single recursive
+// function, but we want to guard against the unlikely possibility of
+// a smart compiler optimizing the recursion away.
+//
+// GTEST_NO_INLINE_ is required to prevent GCC 4.6 from inlining
+// StackLowerThanAddress into StackGrowsDown, which then doesn't give
+// correct answer.
+bool StackLowerThanAddress(const void* ptr) GTEST_NO_INLINE_;
+bool StackLowerThanAddress(const void* ptr) {
+ int dummy;
+ return &dummy < ptr;
+}
+
+bool StackGrowsDown() {
+ int dummy;
+ return StackLowerThanAddress(&dummy);
+}
+
+// A threadsafe implementation of fork(2) for threadsafe-style death tests
+// that uses clone(2). It dies with an error message if anything goes
+// wrong.
+static pid_t ExecDeathTestFork(char* const* argv, int close_fd) {
+ ExecDeathTestArgs args = { argv, close_fd };
+ pid_t child_pid = -1;
+
+# if GTEST_HAS_CLONE
+ const bool use_fork = GTEST_FLAG(death_test_use_fork);
+
+ if (!use_fork) {
+ static const bool stack_grows_down = StackGrowsDown();
+ const size_t stack_size = getpagesize();
+ // MMAP_ANONYMOUS is not defined on Mac, so we use MAP_ANON instead.
+ void* const stack = mmap(NULL, stack_size, PROT_READ | PROT_WRITE,
+ MAP_ANON | MAP_PRIVATE, -1, 0);
+ GTEST_DEATH_TEST_CHECK_(stack != MAP_FAILED);
+ void* const stack_top =
+ static_cast<char*>(stack) + (stack_grows_down ? stack_size : 0);
+
+ child_pid = clone(&ExecDeathTestChildMain, stack_top, SIGCHLD, &args);
+
+ GTEST_DEATH_TEST_CHECK_(munmap(stack, stack_size) != -1);
+ }
+# else
+ const bool use_fork = true;
+# endif // GTEST_HAS_CLONE
+
+ if (use_fork && (child_pid = fork()) == 0) {
+ ExecDeathTestChildMain(&args);
+ _exit(0);
+ }
+
+ GTEST_DEATH_TEST_CHECK_(child_pid != -1);
+ return child_pid;
+}
+
+// The AssumeRole process for a fork-and-exec death test. It re-executes the
+// main program from the beginning, setting the --gtest_filter
+// and --gtest_internal_run_death_test flags to cause only the current
+// death test to be re-run.
+DeathTest::TestRole ExecDeathTest::AssumeRole() {
+ const UnitTestImpl* const impl = GetUnitTestImpl();
+ const InternalRunDeathTestFlag* const flag =
+ impl->internal_run_death_test_flag();
+ const TestInfo* const info = impl->current_test_info();
+ const int death_test_index = info->result()->death_test_count();
+
+ if (flag != NULL) {
+ set_write_fd(flag->write_fd());
+ return EXECUTE_TEST;
+ }
+
+ int pipe_fd[2];
+ GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1);
+ // Clear the close-on-exec flag on the write end of the pipe, lest
+ // it be closed when the child process does an exec:
+ GTEST_DEATH_TEST_CHECK_(fcntl(pipe_fd[1], F_SETFD, 0) != -1);
+
+ const String filter_flag =
+ String::Format("--%s%s=%s.%s",
+ GTEST_FLAG_PREFIX_, kFilterFlag,
+ info->test_case_name(), info->name());
+ const String internal_flag =
+ String::Format("--%s%s=%s|%d|%d|%d",
+ GTEST_FLAG_PREFIX_, kInternalRunDeathTestFlag,
+ file_, line_, death_test_index, pipe_fd[1]);
+ Arguments args;
+ args.AddArguments(GetArgvs());
+ args.AddArgument(filter_flag.c_str());
+ args.AddArgument(internal_flag.c_str());
+
+ DeathTest::set_last_death_test_message("");
+
+ CaptureStderr();
+ // See the comment in NoExecDeathTest::AssumeRole for why the next line
+ // is necessary.
+ FlushInfoLog();
+
+ const pid_t child_pid = ExecDeathTestFork(args.Argv(), pipe_fd[0]);
+ GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1]));
+ set_child_pid(child_pid);
+ set_read_fd(pipe_fd[0]);
+ set_spawned(true);
+ return OVERSEE_TEST;
+}
+
+# endif // !GTEST_OS_WINDOWS
+
+// Creates a concrete DeathTest-derived class that depends on the
+// --gtest_death_test_style flag, and sets the pointer pointed to
+// by the "test" argument to its address. If the test should be
+// skipped, sets that pointer to NULL. Returns true, unless the
+// flag is set to an invalid value.
+bool DefaultDeathTestFactory::Create(const char* statement, const RE* regex,
+ const char* file, int line,
+ DeathTest** test) {
+ UnitTestImpl* const impl = GetUnitTestImpl();
+ const InternalRunDeathTestFlag* const flag =
+ impl->internal_run_death_test_flag();
+ const int death_test_index = impl->current_test_info()
+ ->increment_death_test_count();
+
+ if (flag != NULL) {
+ if (death_test_index > flag->index()) {
+ DeathTest::set_last_death_test_message(String::Format(
+ "Death test count (%d) somehow exceeded expected maximum (%d)",
+ death_test_index, flag->index()));
+ return false;
+ }
+
+ if (!(flag->file() == file && flag->line() == line &&
+ flag->index() == death_test_index)) {
+ *test = NULL;
+ return true;
+ }
+ }
+
+# if GTEST_OS_WINDOWS
+
+ if (GTEST_FLAG(death_test_style) == "threadsafe" ||
+ GTEST_FLAG(death_test_style) == "fast") {
+ *test = new WindowsDeathTest(statement, regex, file, line);
+ }
+
+# else
+
+ if (GTEST_FLAG(death_test_style) == "threadsafe") {
+ *test = new ExecDeathTest(statement, regex, file, line);
+ } else if (GTEST_FLAG(death_test_style) == "fast") {
+ *test = new NoExecDeathTest(statement, regex);
+ }
+
+# endif // GTEST_OS_WINDOWS
+
+ else { // NOLINT - this is more readable than unbalanced brackets inside #if.
+ DeathTest::set_last_death_test_message(String::Format(
+ "Unknown death test style \"%s\" encountered",
+ GTEST_FLAG(death_test_style).c_str()));
+ return false;
+ }
+
+ return true;
+}
+
+// Splits a given string on a given delimiter, populating a given
+// vector with the fields. GTEST_HAS_DEATH_TEST implies that we have
+// ::std::string, so we can use it here.
+static void SplitString(const ::std::string& str, char delimiter,
+ ::std::vector< ::std::string>* dest) {
+ ::std::vector< ::std::string> parsed;
+ ::std::string::size_type pos = 0;
+ while (::testing::internal::AlwaysTrue()) {
+ const ::std::string::size_type colon = str.find(delimiter, pos);
+ if (colon == ::std::string::npos) {
+ parsed.push_back(str.substr(pos));
+ break;
+ } else {
+ parsed.push_back(str.substr(pos, colon - pos));
+ pos = colon + 1;
+ }
+ }
+ dest->swap(parsed);
+}
+
+# if GTEST_OS_WINDOWS
+// Recreates the pipe and event handles from the provided parameters,
+// signals the event, and returns a file descriptor wrapped around the pipe
+// handle. This function is called in the child process only.
+int GetStatusFileDescriptor(unsigned int parent_process_id,
+ size_t write_handle_as_size_t,
+ size_t event_handle_as_size_t) {
+ AutoHandle parent_process_handle(::OpenProcess(PROCESS_DUP_HANDLE,
+ FALSE, // Non-inheritable.
+ parent_process_id));
+ if (parent_process_handle.Get() == INVALID_HANDLE_VALUE) {
+ DeathTestAbort(String::Format("Unable to open parent process %u",
+ parent_process_id));
+ }
+
+ // TODO(vladl@google.com): Replace the following check with a
+ // compile-time assertion when available.
+ GTEST_CHECK_(sizeof(HANDLE) <= sizeof(size_t));
+
+ const HANDLE write_handle =
+ reinterpret_cast<HANDLE>(write_handle_as_size_t);
+ HANDLE dup_write_handle;
+
+ // The newly initialized handle is accessible only in in the parent
+ // process. To obtain one accessible within the child, we need to use
+ // DuplicateHandle.
+ if (!::DuplicateHandle(parent_process_handle.Get(), write_handle,
+ ::GetCurrentProcess(), &dup_write_handle,
+ 0x0, // Requested privileges ignored since
+ // DUPLICATE_SAME_ACCESS is used.
+ FALSE, // Request non-inheritable handler.
+ DUPLICATE_SAME_ACCESS)) {
+ DeathTestAbort(String::Format(
+ "Unable to duplicate the pipe handle %Iu from the parent process %u",
+ write_handle_as_size_t, parent_process_id));
+ }
+
+ const HANDLE event_handle = reinterpret_cast<HANDLE>(event_handle_as_size_t);
+ HANDLE dup_event_handle;
+
+ if (!::DuplicateHandle(parent_process_handle.Get(), event_handle,
+ ::GetCurrentProcess(), &dup_event_handle,
+ 0x0,
+ FALSE,
+ DUPLICATE_SAME_ACCESS)) {
+ DeathTestAbort(String::Format(
+ "Unable to duplicate the event handle %Iu from the parent process %u",
+ event_handle_as_size_t, parent_process_id));
+ }
+
+ const int write_fd =
+ ::_open_osfhandle(reinterpret_cast<intptr_t>(dup_write_handle), O_APPEND);
+ if (write_fd == -1) {
+ DeathTestAbort(String::Format(
+ "Unable to convert pipe handle %Iu to a file descriptor",
+ write_handle_as_size_t));
+ }
+
+ // Signals the parent that the write end of the pipe has been acquired
+ // so the parent can release its own write end.
+ ::SetEvent(dup_event_handle);
+
+ return write_fd;
+}
+# endif // GTEST_OS_WINDOWS
+
+// Returns a newly created InternalRunDeathTestFlag object with fields
+// initialized from the GTEST_FLAG(internal_run_death_test) flag if
+// the flag is specified; otherwise returns NULL.
+InternalRunDeathTestFlag* ParseInternalRunDeathTestFlag() {
+ if (GTEST_FLAG(internal_run_death_test) == "") return NULL;
+
+ // GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we
+ // can use it here.
+ int line = -1;
+ int index = -1;
+ ::std::vector< ::std::string> fields;
+ SplitString(GTEST_FLAG(internal_run_death_test).c_str(), '|', &fields);
+ int write_fd = -1;
+
+# if GTEST_OS_WINDOWS
+
+ unsigned int parent_process_id = 0;
+ size_t write_handle_as_size_t = 0;
+ size_t event_handle_as_size_t = 0;
+
+ if (fields.size() != 6
+ || !ParseNaturalNumber(fields[1], &line)
+ || !ParseNaturalNumber(fields[2], &index)
+ || !ParseNaturalNumber(fields[3], &parent_process_id)
+ || !ParseNaturalNumber(fields[4], &write_handle_as_size_t)
+ || !ParseNaturalNumber(fields[5], &event_handle_as_size_t)) {
+ DeathTestAbort(String::Format(
+ "Bad --gtest_internal_run_death_test flag: %s",
+ GTEST_FLAG(internal_run_death_test).c_str()));
+ }
+ write_fd = GetStatusFileDescriptor(parent_process_id,
+ write_handle_as_size_t,
+ event_handle_as_size_t);
+# else
+
+ if (fields.size() != 4
+ || !ParseNaturalNumber(fields[1], &line)
+ || !ParseNaturalNumber(fields[2], &index)
+ || !ParseNaturalNumber(fields[3], &write_fd)) {
+ DeathTestAbort(String::Format(
+ "Bad --gtest_internal_run_death_test flag: %s",
+ GTEST_FLAG(internal_run_death_test).c_str()));
+ }
+
+# endif // GTEST_OS_WINDOWS
+
+ return new InternalRunDeathTestFlag(fields[0], line, index, write_fd);
+}
+
+} // namespace internal
+
+#endif // GTEST_HAS_DEATH_TEST
+
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/src/gtest-filepath.cc b/libvpx/third_party/googletest/src/src/gtest-filepath.cc
new file mode 100644
index 0000000..91b2571
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest-filepath.cc
@@ -0,0 +1,380 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: keith.ray@gmail.com (Keith Ray)
+
+#include "gtest/internal/gtest-filepath.h"
+#include "gtest/internal/gtest-port.h"
+
+#include <stdlib.h>
+
+#if GTEST_OS_WINDOWS_MOBILE
+# include <windows.h>
+#elif GTEST_OS_WINDOWS
+# include <direct.h>
+# include <io.h>
+#elif GTEST_OS_SYMBIAN || GTEST_OS_NACL
+// Symbian OpenC and NaCl have PATH_MAX in sys/syslimits.h
+# include <sys/syslimits.h>
+#else
+# include <limits.h>
+# include <climits> // Some Linux distributions define PATH_MAX here.
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+#if GTEST_OS_WINDOWS
+# define GTEST_PATH_MAX_ _MAX_PATH
+#elif defined(PATH_MAX)
+# define GTEST_PATH_MAX_ PATH_MAX
+#elif defined(_XOPEN_PATH_MAX)
+# define GTEST_PATH_MAX_ _XOPEN_PATH_MAX
+#else
+# define GTEST_PATH_MAX_ _POSIX_PATH_MAX
+#endif // GTEST_OS_WINDOWS
+
+#include "gtest/internal/gtest-string.h"
+
+namespace testing {
+namespace internal {
+
+#if GTEST_OS_WINDOWS
+// On Windows, '\\' is the standard path separator, but many tools and the
+// Windows API also accept '/' as an alternate path separator. Unless otherwise
+// noted, a file path can contain either kind of path separators, or a mixture
+// of them.
+const char kPathSeparator = '\\';
+const char kAlternatePathSeparator = '/';
+const char kPathSeparatorString[] = "\\";
+const char kAlternatePathSeparatorString[] = "/";
+# if GTEST_OS_WINDOWS_MOBILE
+// Windows CE doesn't have a current directory. You should not use
+// the current directory in tests on Windows CE, but this at least
+// provides a reasonable fallback.
+const char kCurrentDirectoryString[] = "\\";
+// Windows CE doesn't define INVALID_FILE_ATTRIBUTES
+const DWORD kInvalidFileAttributes = 0xffffffff;
+# else
+const char kCurrentDirectoryString[] = ".\\";
+# endif // GTEST_OS_WINDOWS_MOBILE
+#else
+const char kPathSeparator = '/';
+const char kPathSeparatorString[] = "/";
+const char kCurrentDirectoryString[] = "./";
+#endif // GTEST_OS_WINDOWS
+
+// Returns whether the given character is a valid path separator.
+static bool IsPathSeparator(char c) {
+#if GTEST_HAS_ALT_PATH_SEP_
+ return (c == kPathSeparator) || (c == kAlternatePathSeparator);
+#else
+ return c == kPathSeparator;
+#endif
+}
+
+// Returns the current working directory, or "" if unsuccessful.
+FilePath FilePath::GetCurrentDir() {
+#if GTEST_OS_WINDOWS_MOBILE
+ // Windows CE doesn't have a current directory, so we just return
+ // something reasonable.
+ return FilePath(kCurrentDirectoryString);
+#elif GTEST_OS_WINDOWS
+ char cwd[GTEST_PATH_MAX_ + 1] = { '\0' };
+ return FilePath(_getcwd(cwd, sizeof(cwd)) == NULL ? "" : cwd);
+#else
+ char cwd[GTEST_PATH_MAX_ + 1] = { '\0' };
+ return FilePath(getcwd(cwd, sizeof(cwd)) == NULL ? "" : cwd);
+#endif // GTEST_OS_WINDOWS_MOBILE
+}
+
+// Returns a copy of the FilePath with the case-insensitive extension removed.
+// Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns
+// FilePath("dir/file"). If a case-insensitive extension is not
+// found, returns a copy of the original FilePath.
+FilePath FilePath::RemoveExtension(const char* extension) const {
+ String dot_extension(String::Format(".%s", extension));
+ if (pathname_.EndsWithCaseInsensitive(dot_extension.c_str())) {
+ return FilePath(String(pathname_.c_str(), pathname_.length() - 4));
+ }
+ return *this;
+}
+
+// Returns a pointer to the last occurence of a valid path separator in
+// the FilePath. On Windows, for example, both '/' and '\' are valid path
+// separators. Returns NULL if no path separator was found.
+const char* FilePath::FindLastPathSeparator() const {
+ const char* const last_sep = strrchr(c_str(), kPathSeparator);
+#if GTEST_HAS_ALT_PATH_SEP_
+ const char* const last_alt_sep = strrchr(c_str(), kAlternatePathSeparator);
+ // Comparing two pointers of which only one is NULL is undefined.
+ if (last_alt_sep != NULL &&
+ (last_sep == NULL || last_alt_sep > last_sep)) {
+ return last_alt_sep;
+ }
+#endif
+ return last_sep;
+}
+
+// Returns a copy of the FilePath with the directory part removed.
+// Example: FilePath("path/to/file").RemoveDirectoryName() returns
+// FilePath("file"). If there is no directory part ("just_a_file"), it returns
+// the FilePath unmodified. If there is no file part ("just_a_dir/") it
+// returns an empty FilePath ("").
+// On Windows platform, '\' is the path separator, otherwise it is '/'.
+FilePath FilePath::RemoveDirectoryName() const {
+ const char* const last_sep = FindLastPathSeparator();
+ return last_sep ? FilePath(String(last_sep + 1)) : *this;
+}
+
+// RemoveFileName returns the directory path with the filename removed.
+// Example: FilePath("path/to/file").RemoveFileName() returns "path/to/".
+// If the FilePath is "a_file" or "/a_file", RemoveFileName returns
+// FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does
+// not have a file, like "just/a/dir/", it returns the FilePath unmodified.
+// On Windows platform, '\' is the path separator, otherwise it is '/'.
+FilePath FilePath::RemoveFileName() const {
+ const char* const last_sep = FindLastPathSeparator();
+ String dir;
+ if (last_sep) {
+ dir = String(c_str(), last_sep + 1 - c_str());
+ } else {
+ dir = kCurrentDirectoryString;
+ }
+ return FilePath(dir);
+}
+
+// Helper functions for naming files in a directory for xml output.
+
+// Given directory = "dir", base_name = "test", number = 0,
+// extension = "xml", returns "dir/test.xml". If number is greater
+// than zero (e.g., 12), returns "dir/test_12.xml".
+// On Windows platform, uses \ as the separator rather than /.
+FilePath FilePath::MakeFileName(const FilePath& directory,
+ const FilePath& base_name,
+ int number,
+ const char* extension) {
+ String file;
+ if (number == 0) {
+ file = String::Format("%s.%s", base_name.c_str(), extension);
+ } else {
+ file = String::Format("%s_%d.%s", base_name.c_str(), number, extension);
+ }
+ return ConcatPaths(directory, FilePath(file));
+}
+
+// Given directory = "dir", relative_path = "test.xml", returns "dir/test.xml".
+// On Windows, uses \ as the separator rather than /.
+FilePath FilePath::ConcatPaths(const FilePath& directory,
+ const FilePath& relative_path) {
+ if (directory.IsEmpty())
+ return relative_path;
+ const FilePath dir(directory.RemoveTrailingPathSeparator());
+ return FilePath(String::Format("%s%c%s", dir.c_str(), kPathSeparator,
+ relative_path.c_str()));
+}
+
+// Returns true if pathname describes something findable in the file-system,
+// either a file, directory, or whatever.
+bool FilePath::FileOrDirectoryExists() const {
+#if GTEST_OS_WINDOWS_MOBILE
+ LPCWSTR unicode = String::AnsiToUtf16(pathname_.c_str());
+ const DWORD attributes = GetFileAttributes(unicode);
+ delete [] unicode;
+ return attributes != kInvalidFileAttributes;
+#else
+ posix::StatStruct file_stat;
+ return posix::Stat(pathname_.c_str(), &file_stat) == 0;
+#endif // GTEST_OS_WINDOWS_MOBILE
+}
+
+// Returns true if pathname describes a directory in the file-system
+// that exists.
+bool FilePath::DirectoryExists() const {
+ bool result = false;
+#if GTEST_OS_WINDOWS
+ // Don't strip off trailing separator if path is a root directory on
+ // Windows (like "C:\\").
+ const FilePath& path(IsRootDirectory() ? *this :
+ RemoveTrailingPathSeparator());
+#else
+ const FilePath& path(*this);
+#endif
+
+#if GTEST_OS_WINDOWS_MOBILE
+ LPCWSTR unicode = String::AnsiToUtf16(path.c_str());
+ const DWORD attributes = GetFileAttributes(unicode);
+ delete [] unicode;
+ if ((attributes != kInvalidFileAttributes) &&
+ (attributes & FILE_ATTRIBUTE_DIRECTORY)) {
+ result = true;
+ }
+#else
+ posix::StatStruct file_stat;
+ result = posix::Stat(path.c_str(), &file_stat) == 0 &&
+ posix::IsDir(file_stat);
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+ return result;
+}
+
+// Returns true if pathname describes a root directory. (Windows has one
+// root directory per disk drive.)
+bool FilePath::IsRootDirectory() const {
+#if GTEST_OS_WINDOWS
+ // TODO(wan@google.com): on Windows a network share like
+ // \\server\share can be a root directory, although it cannot be the
+ // current directory. Handle this properly.
+ return pathname_.length() == 3 && IsAbsolutePath();
+#else
+ return pathname_.length() == 1 && IsPathSeparator(pathname_.c_str()[0]);
+#endif
+}
+
+// Returns true if pathname describes an absolute path.
+bool FilePath::IsAbsolutePath() const {
+ const char* const name = pathname_.c_str();
+#if GTEST_OS_WINDOWS
+ return pathname_.length() >= 3 &&
+ ((name[0] >= 'a' && name[0] <= 'z') ||
+ (name[0] >= 'A' && name[0] <= 'Z')) &&
+ name[1] == ':' &&
+ IsPathSeparator(name[2]);
+#else
+ return IsPathSeparator(name[0]);
+#endif
+}
+
+// Returns a pathname for a file that does not currently exist. The pathname
+// will be directory/base_name.extension or
+// directory/base_name_<number>.extension if directory/base_name.extension
+// already exists. The number will be incremented until a pathname is found
+// that does not already exist.
+// Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'.
+// There could be a race condition if two or more processes are calling this
+// function at the same time -- they could both pick the same filename.
+FilePath FilePath::GenerateUniqueFileName(const FilePath& directory,
+ const FilePath& base_name,
+ const char* extension) {
+ FilePath full_pathname;
+ int number = 0;
+ do {
+ full_pathname.Set(MakeFileName(directory, base_name, number++, extension));
+ } while (full_pathname.FileOrDirectoryExists());
+ return full_pathname;
+}
+
+// Returns true if FilePath ends with a path separator, which indicates that
+// it is intended to represent a directory. Returns false otherwise.
+// This does NOT check that a directory (or file) actually exists.
+bool FilePath::IsDirectory() const {
+ return !pathname_.empty() &&
+ IsPathSeparator(pathname_.c_str()[pathname_.length() - 1]);
+}
+
+// Create directories so that path exists. Returns true if successful or if
+// the directories already exist; returns false if unable to create directories
+// for any reason.
+bool FilePath::CreateDirectoriesRecursively() const {
+ if (!this->IsDirectory()) {
+ return false;
+ }
+
+ if (pathname_.length() == 0 || this->DirectoryExists()) {
+ return true;
+ }
+
+ const FilePath parent(this->RemoveTrailingPathSeparator().RemoveFileName());
+ return parent.CreateDirectoriesRecursively() && this->CreateFolder();
+}
+
+// Create the directory so that path exists. Returns true if successful or
+// if the directory already exists; returns false if unable to create the
+// directory for any reason, including if the parent directory does not
+// exist. Not named "CreateDirectory" because that's a macro on Windows.
+bool FilePath::CreateFolder() const {
+#if GTEST_OS_WINDOWS_MOBILE
+ FilePath removed_sep(this->RemoveTrailingPathSeparator());
+ LPCWSTR unicode = String::AnsiToUtf16(removed_sep.c_str());
+ int result = CreateDirectory(unicode, NULL) ? 0 : -1;
+ delete [] unicode;
+#elif GTEST_OS_WINDOWS
+ int result = _mkdir(pathname_.c_str());
+#else
+ int result = mkdir(pathname_.c_str(), 0777);
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+ if (result == -1) {
+ return this->DirectoryExists(); // An error is OK if the directory exists.
+ }
+ return true; // No error.
+}
+
+// If input name has a trailing separator character, remove it and return the
+// name, otherwise return the name string unmodified.
+// On Windows platform, uses \ as the separator, other platforms use /.
+FilePath FilePath::RemoveTrailingPathSeparator() const {
+ return IsDirectory()
+ ? FilePath(String(pathname_.c_str(), pathname_.length() - 1))
+ : *this;
+}
+
+// Removes any redundant separators that might be in the pathname.
+// For example, "bar///foo" becomes "bar/foo". Does not eliminate other
+// redundancies that might be in a pathname involving "." or "..".
+// TODO(wan@google.com): handle Windows network shares (e.g. \\server\share).
+void FilePath::Normalize() {
+ if (pathname_.c_str() == NULL) {
+ pathname_ = "";
+ return;
+ }
+ const char* src = pathname_.c_str();
+ char* const dest = new char[pathname_.length() + 1];
+ char* dest_ptr = dest;
+ memset(dest_ptr, 0, pathname_.length() + 1);
+
+ while (*src != '\0') {
+ *dest_ptr = *src;
+ if (!IsPathSeparator(*src)) {
+ src++;
+ } else {
+#if GTEST_HAS_ALT_PATH_SEP_
+ if (*dest_ptr == kAlternatePathSeparator) {
+ *dest_ptr = kPathSeparator;
+ }
+#endif
+ while (IsPathSeparator(*src))
+ src++;
+ }
+ dest_ptr++;
+ }
+ *dest_ptr = '\0';
+ pathname_ = dest;
+ delete[] dest;
+}
+
+} // namespace internal
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/src/gtest-internal-inl.h b/libvpx/third_party/googletest/src/src/gtest-internal-inl.h
new file mode 100644
index 0000000..65a2101
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest-internal-inl.h
@@ -0,0 +1,1038 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// Utility functions and classes used by the Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// This file contains purely Google Test's internal implementation. Please
+// DO NOT #INCLUDE IT IN A USER PROGRAM.
+
+#ifndef GTEST_SRC_GTEST_INTERNAL_INL_H_
+#define GTEST_SRC_GTEST_INTERNAL_INL_H_
+
+// GTEST_IMPLEMENTATION_ is defined to 1 iff the current translation unit is
+// part of Google Test's implementation; otherwise it's undefined.
+#if !GTEST_IMPLEMENTATION_
+// A user is trying to include this from his code - just say no.
+# error "gtest-internal-inl.h is part of Google Test's internal implementation."
+# error "It must not be included except by Google Test itself."
+#endif // GTEST_IMPLEMENTATION_
+
+#ifndef _WIN32_WCE
+# include <errno.h>
+#endif // !_WIN32_WCE
+#include <stddef.h>
+#include <stdlib.h> // For strtoll/_strtoul64/malloc/free.
+#include <string.h> // For memmove.
+
+#include <algorithm>
+#include <string>
+#include <vector>
+
+#include "gtest/internal/gtest-port.h"
+
+#if GTEST_OS_WINDOWS
+# include <windows.h> // NOLINT
+#endif // GTEST_OS_WINDOWS
+
+#include "gtest/gtest.h" // NOLINT
+#include "gtest/gtest-spi.h"
+
+namespace testing {
+
+// Declares the flags.
+//
+// We don't want the users to modify this flag in the code, but want
+// Google Test's own unit tests to be able to access it. Therefore we
+// declare it here as opposed to in gtest.h.
+GTEST_DECLARE_bool_(death_test_use_fork);
+
+namespace internal {
+
+// The value of GetTestTypeId() as seen from within the Google Test
+// library. This is solely for testing GetTestTypeId().
+GTEST_API_ extern const TypeId kTestTypeIdInGoogleTest;
+
+// Names of the flags (needed for parsing Google Test flags).
+const char kAlsoRunDisabledTestsFlag[] = "also_run_disabled_tests";
+const char kBreakOnFailureFlag[] = "break_on_failure";
+const char kCatchExceptionsFlag[] = "catch_exceptions";
+const char kColorFlag[] = "color";
+const char kFilterFlag[] = "filter";
+const char kListTestsFlag[] = "list_tests";
+const char kOutputFlag[] = "output";
+const char kPrintTimeFlag[] = "print_time";
+const char kRandomSeedFlag[] = "random_seed";
+const char kRepeatFlag[] = "repeat";
+const char kShuffleFlag[] = "shuffle";
+const char kStackTraceDepthFlag[] = "stack_trace_depth";
+const char kStreamResultToFlag[] = "stream_result_to";
+const char kThrowOnFailureFlag[] = "throw_on_failure";
+
+// A valid random seed must be in [1, kMaxRandomSeed].
+const int kMaxRandomSeed = 99999;
+
+// g_help_flag is true iff the --help flag or an equivalent form is
+// specified on the command line.
+GTEST_API_ extern bool g_help_flag;
+
+// Returns the current time in milliseconds.
+GTEST_API_ TimeInMillis GetTimeInMillis();
+
+// Returns true iff Google Test should use colors in the output.
+GTEST_API_ bool ShouldUseColor(bool stdout_is_tty);
+
+// Formats the given time in milliseconds as seconds.
+GTEST_API_ std::string FormatTimeInMillisAsSeconds(TimeInMillis ms);
+
+// Parses a string for an Int32 flag, in the form of "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true. On failure, returns false without changing *value.
+GTEST_API_ bool ParseInt32Flag(
+ const char* str, const char* flag, Int32* value);
+
+// Returns a random seed in range [1, kMaxRandomSeed] based on the
+// given --gtest_random_seed flag value.
+inline int GetRandomSeedFromFlag(Int32 random_seed_flag) {
+ const unsigned int raw_seed = (random_seed_flag == 0) ?
+ static_cast<unsigned int>(GetTimeInMillis()) :
+ static_cast<unsigned int>(random_seed_flag);
+
+ // Normalizes the actual seed to range [1, kMaxRandomSeed] such that
+ // it's easy to type.
+ const int normalized_seed =
+ static_cast<int>((raw_seed - 1U) %
+ static_cast<unsigned int>(kMaxRandomSeed)) + 1;
+ return normalized_seed;
+}
+
+// Returns the first valid random seed after 'seed'. The behavior is
+// undefined if 'seed' is invalid. The seed after kMaxRandomSeed is
+// considered to be 1.
+inline int GetNextRandomSeed(int seed) {
+ GTEST_CHECK_(1 <= seed && seed <= kMaxRandomSeed)
+ << "Invalid random seed " << seed << " - must be in [1, "
+ << kMaxRandomSeed << "].";
+ const int next_seed = seed + 1;
+ return (next_seed > kMaxRandomSeed) ? 1 : next_seed;
+}
+
+// This class saves the values of all Google Test flags in its c'tor, and
+// restores them in its d'tor.
+class GTestFlagSaver {
+ public:
+ // The c'tor.
+ GTestFlagSaver() {
+ also_run_disabled_tests_ = GTEST_FLAG(also_run_disabled_tests);
+ break_on_failure_ = GTEST_FLAG(break_on_failure);
+ catch_exceptions_ = GTEST_FLAG(catch_exceptions);
+ color_ = GTEST_FLAG(color);
+ death_test_style_ = GTEST_FLAG(death_test_style);
+ death_test_use_fork_ = GTEST_FLAG(death_test_use_fork);
+ filter_ = GTEST_FLAG(filter);
+ internal_run_death_test_ = GTEST_FLAG(internal_run_death_test);
+ list_tests_ = GTEST_FLAG(list_tests);
+ output_ = GTEST_FLAG(output);
+ print_time_ = GTEST_FLAG(print_time);
+ random_seed_ = GTEST_FLAG(random_seed);
+ repeat_ = GTEST_FLAG(repeat);
+ shuffle_ = GTEST_FLAG(shuffle);
+ stack_trace_depth_ = GTEST_FLAG(stack_trace_depth);
+ stream_result_to_ = GTEST_FLAG(stream_result_to);
+ throw_on_failure_ = GTEST_FLAG(throw_on_failure);
+ }
+
+ // The d'tor is not virtual. DO NOT INHERIT FROM THIS CLASS.
+ ~GTestFlagSaver() {
+ GTEST_FLAG(also_run_disabled_tests) = also_run_disabled_tests_;
+ GTEST_FLAG(break_on_failure) = break_on_failure_;
+ GTEST_FLAG(catch_exceptions) = catch_exceptions_;
+ GTEST_FLAG(color) = color_;
+ GTEST_FLAG(death_test_style) = death_test_style_;
+ GTEST_FLAG(death_test_use_fork) = death_test_use_fork_;
+ GTEST_FLAG(filter) = filter_;
+ GTEST_FLAG(internal_run_death_test) = internal_run_death_test_;
+ GTEST_FLAG(list_tests) = list_tests_;
+ GTEST_FLAG(output) = output_;
+ GTEST_FLAG(print_time) = print_time_;
+ GTEST_FLAG(random_seed) = random_seed_;
+ GTEST_FLAG(repeat) = repeat_;
+ GTEST_FLAG(shuffle) = shuffle_;
+ GTEST_FLAG(stack_trace_depth) = stack_trace_depth_;
+ GTEST_FLAG(stream_result_to) = stream_result_to_;
+ GTEST_FLAG(throw_on_failure) = throw_on_failure_;
+ }
+ private:
+ // Fields for saving the original values of flags.
+ bool also_run_disabled_tests_;
+ bool break_on_failure_;
+ bool catch_exceptions_;
+ String color_;
+ String death_test_style_;
+ bool death_test_use_fork_;
+ String filter_;
+ String internal_run_death_test_;
+ bool list_tests_;
+ String output_;
+ bool print_time_;
+ bool pretty_;
+ internal::Int32 random_seed_;
+ internal::Int32 repeat_;
+ bool shuffle_;
+ internal::Int32 stack_trace_depth_;
+ String stream_result_to_;
+ bool throw_on_failure_;
+} GTEST_ATTRIBUTE_UNUSED_;
+
+// Converts a Unicode code point to a narrow string in UTF-8 encoding.
+// code_point parameter is of type UInt32 because wchar_t may not be
+// wide enough to contain a code point.
+// The output buffer str must containt at least 32 characters.
+// The function returns the address of the output buffer.
+// If the code_point is not a valid Unicode code point
+// (i.e. outside of Unicode range U+0 to U+10FFFF) it will be output
+// as '(Invalid Unicode 0xXXXXXXXX)'.
+GTEST_API_ char* CodePointToUtf8(UInt32 code_point, char* str);
+
+// Converts a wide string to a narrow string in UTF-8 encoding.
+// The wide string is assumed to have the following encoding:
+// UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin, Symbian OS)
+// UTF-32 if sizeof(wchar_t) == 4 (on Linux)
+// Parameter str points to a null-terminated wide string.
+// Parameter num_chars may additionally limit the number
+// of wchar_t characters processed. -1 is used when the entire string
+// should be processed.
+// If the string contains code points that are not valid Unicode code points
+// (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output
+// as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding
+// and contains invalid UTF-16 surrogate pairs, values in those pairs
+// will be encoded as individual Unicode characters from Basic Normal Plane.
+GTEST_API_ String WideStringToUtf8(const wchar_t* str, int num_chars);
+
+// Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file
+// if the variable is present. If a file already exists at this location, this
+// function will write over it. If the variable is present, but the file cannot
+// be created, prints an error and exits.
+void WriteToShardStatusFileIfNeeded();
+
+// Checks whether sharding is enabled by examining the relevant
+// environment variable values. If the variables are present,
+// but inconsistent (e.g., shard_index >= total_shards), prints
+// an error and exits. If in_subprocess_for_death_test, sharding is
+// disabled because it must only be applied to the original test
+// process. Otherwise, we could filter out death tests we intended to execute.
+GTEST_API_ bool ShouldShard(const char* total_shards_str,
+ const char* shard_index_str,
+ bool in_subprocess_for_death_test);
+
+// Parses the environment variable var as an Int32. If it is unset,
+// returns default_val. If it is not an Int32, prints an error and
+// and aborts.
+GTEST_API_ Int32 Int32FromEnvOrDie(const char* env_var, Int32 default_val);
+
+// Given the total number of shards, the shard index, and the test id,
+// returns true iff the test should be run on this shard. The test id is
+// some arbitrary but unique non-negative integer assigned to each test
+// method. Assumes that 0 <= shard_index < total_shards.
+GTEST_API_ bool ShouldRunTestOnShard(
+ int total_shards, int shard_index, int test_id);
+
+// STL container utilities.
+
+// Returns the number of elements in the given container that satisfy
+// the given predicate.
+template <class Container, typename Predicate>
+inline int CountIf(const Container& c, Predicate predicate) {
+ // Implemented as an explicit loop since std::count_if() in libCstd on
+ // Solaris has a non-standard signature.
+ int count = 0;
+ for (typename Container::const_iterator it = c.begin(); it != c.end(); ++it) {
+ if (predicate(*it))
+ ++count;
+ }
+ return count;
+}
+
+// Applies a function/functor to each element in the container.
+template <class Container, typename Functor>
+void ForEach(const Container& c, Functor functor) {
+ std::for_each(c.begin(), c.end(), functor);
+}
+
+// Returns the i-th element of the vector, or default_value if i is not
+// in range [0, v.size()).
+template <typename E>
+inline E GetElementOr(const std::vector<E>& v, int i, E default_value) {
+ return (i < 0 || i >= static_cast<int>(v.size())) ? default_value : v[i];
+}
+
+// Performs an in-place shuffle of a range of the vector's elements.
+// 'begin' and 'end' are element indices as an STL-style range;
+// i.e. [begin, end) are shuffled, where 'end' == size() means to
+// shuffle to the end of the vector.
+template <typename E>
+void ShuffleRange(internal::Random* random, int begin, int end,
+ std::vector<E>* v) {
+ const int size = static_cast<int>(v->size());
+ GTEST_CHECK_(0 <= begin && begin <= size)
+ << "Invalid shuffle range start " << begin << ": must be in range [0, "
+ << size << "].";
+ GTEST_CHECK_(begin <= end && end <= size)
+ << "Invalid shuffle range finish " << end << ": must be in range ["
+ << begin << ", " << size << "].";
+
+ // Fisher-Yates shuffle, from
+ // http://en.wikipedia.org/wiki/Fisher-Yates_shuffle
+ for (int range_width = end - begin; range_width >= 2; range_width--) {
+ const int last_in_range = begin + range_width - 1;
+ const int selected = begin + random->Generate(range_width);
+ std::swap((*v)[selected], (*v)[last_in_range]);
+ }
+}
+
+// Performs an in-place shuffle of the vector's elements.
+template <typename E>
+inline void Shuffle(internal::Random* random, std::vector<E>* v) {
+ ShuffleRange(random, 0, static_cast<int>(v->size()), v);
+}
+
+// A function for deleting an object. Handy for being used as a
+// functor.
+template <typename T>
+static void Delete(T* x) {
+ delete x;
+}
+
+// A predicate that checks the key of a TestProperty against a known key.
+//
+// TestPropertyKeyIs is copyable.
+class TestPropertyKeyIs {
+ public:
+ // Constructor.
+ //
+ // TestPropertyKeyIs has NO default constructor.
+ explicit TestPropertyKeyIs(const char* key)
+ : key_(key) {}
+
+ // Returns true iff the test name of test property matches on key_.
+ bool operator()(const TestProperty& test_property) const {
+ return String(test_property.key()).Compare(key_) == 0;
+ }
+
+ private:
+ String key_;
+};
+
+// Class UnitTestOptions.
+//
+// This class contains functions for processing options the user
+// specifies when running the tests. It has only static members.
+//
+// In most cases, the user can specify an option using either an
+// environment variable or a command line flag. E.g. you can set the
+// test filter using either GTEST_FILTER or --gtest_filter. If both
+// the variable and the flag are present, the latter overrides the
+// former.
+class GTEST_API_ UnitTestOptions {
+ public:
+ // Functions for processing the gtest_output flag.
+
+ // Returns the output format, or "" for normal printed output.
+ static String GetOutputFormat();
+
+ // Returns the absolute path of the requested output file, or the
+ // default (test_detail.xml in the original working directory) if
+ // none was explicitly specified.
+ static String GetAbsolutePathToOutputFile();
+
+ // Functions for processing the gtest_filter flag.
+
+ // Returns true iff the wildcard pattern matches the string. The
+ // first ':' or '\0' character in pattern marks the end of it.
+ //
+ // This recursive algorithm isn't very efficient, but is clear and
+ // works well enough for matching test names, which are short.
+ static bool PatternMatchesString(const char *pattern, const char *str);
+
+ // Returns true iff the user-specified filter matches the test case
+ // name and the test name.
+ static bool FilterMatchesTest(const String &test_case_name,
+ const String &test_name);
+
+#if GTEST_OS_WINDOWS
+ // Function for supporting the gtest_catch_exception flag.
+
+ // Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the
+ // given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise.
+ // This function is useful as an __except condition.
+ static int GTestShouldProcessSEH(DWORD exception_code);
+#endif // GTEST_OS_WINDOWS
+
+ // Returns true if "name" matches the ':' separated list of glob-style
+ // filters in "filter".
+ static bool MatchesFilter(const String& name, const char* filter);
+};
+
+// Returns the current application's name, removing directory path if that
+// is present. Used by UnitTestOptions::GetOutputFile.
+GTEST_API_ FilePath GetCurrentExecutableName();
+
+// The role interface for getting the OS stack trace as a string.
+class OsStackTraceGetterInterface {
+ public:
+ OsStackTraceGetterInterface() {}
+ virtual ~OsStackTraceGetterInterface() {}
+
+ // Returns the current OS stack trace as a String. Parameters:
+ //
+ // max_depth - the maximum number of stack frames to be included
+ // in the trace.
+ // skip_count - the number of top frames to be skipped; doesn't count
+ // against max_depth.
+ virtual String CurrentStackTrace(int max_depth, int skip_count) = 0;
+
+ // UponLeavingGTest() should be called immediately before Google Test calls
+ // user code. It saves some information about the current stack that
+ // CurrentStackTrace() will use to find and hide Google Test stack frames.
+ virtual void UponLeavingGTest() = 0;
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetterInterface);
+};
+
+// A working implementation of the OsStackTraceGetterInterface interface.
+class OsStackTraceGetter : public OsStackTraceGetterInterface {
+ public:
+ OsStackTraceGetter() : caller_frame_(NULL) {}
+ virtual String CurrentStackTrace(int max_depth, int skip_count);
+ virtual void UponLeavingGTest();
+
+ // This string is inserted in place of stack frames that are part of
+ // Google Test's implementation.
+ static const char* const kElidedFramesMarker;
+
+ private:
+ Mutex mutex_; // protects all internal state
+
+ // We save the stack frame below the frame that calls user code.
+ // We do this because the address of the frame immediately below
+ // the user code changes between the call to UponLeavingGTest()
+ // and any calls to CurrentStackTrace() from within the user code.
+ void* caller_frame_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetter);
+};
+
+// Information about a Google Test trace point.
+struct TraceInfo {
+ const char* file;
+ int line;
+ String message;
+};
+
+// This is the default global test part result reporter used in UnitTestImpl.
+// This class should only be used by UnitTestImpl.
+class DefaultGlobalTestPartResultReporter
+ : public TestPartResultReporterInterface {
+ public:
+ explicit DefaultGlobalTestPartResultReporter(UnitTestImpl* unit_test);
+ // Implements the TestPartResultReporterInterface. Reports the test part
+ // result in the current test.
+ virtual void ReportTestPartResult(const TestPartResult& result);
+
+ private:
+ UnitTestImpl* const unit_test_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultGlobalTestPartResultReporter);
+};
+
+// This is the default per thread test part result reporter used in
+// UnitTestImpl. This class should only be used by UnitTestImpl.
+class DefaultPerThreadTestPartResultReporter
+ : public TestPartResultReporterInterface {
+ public:
+ explicit DefaultPerThreadTestPartResultReporter(UnitTestImpl* unit_test);
+ // Implements the TestPartResultReporterInterface. The implementation just
+ // delegates to the current global test part result reporter of *unit_test_.
+ virtual void ReportTestPartResult(const TestPartResult& result);
+
+ private:
+ UnitTestImpl* const unit_test_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultPerThreadTestPartResultReporter);
+};
+
+// The private implementation of the UnitTest class. We don't protect
+// the methods under a mutex, as this class is not accessible by a
+// user and the UnitTest class that delegates work to this class does
+// proper locking.
+class GTEST_API_ UnitTestImpl {
+ public:
+ explicit UnitTestImpl(UnitTest* parent);
+ virtual ~UnitTestImpl();
+
+ // There are two different ways to register your own TestPartResultReporter.
+ // You can register your own repoter to listen either only for test results
+ // from the current thread or for results from all threads.
+ // By default, each per-thread test result repoter just passes a new
+ // TestPartResult to the global test result reporter, which registers the
+ // test part result for the currently running test.
+
+ // Returns the global test part result reporter.
+ TestPartResultReporterInterface* GetGlobalTestPartResultReporter();
+
+ // Sets the global test part result reporter.
+ void SetGlobalTestPartResultReporter(
+ TestPartResultReporterInterface* reporter);
+
+ // Returns the test part result reporter for the current thread.
+ TestPartResultReporterInterface* GetTestPartResultReporterForCurrentThread();
+
+ // Sets the test part result reporter for the current thread.
+ void SetTestPartResultReporterForCurrentThread(
+ TestPartResultReporterInterface* reporter);
+
+ // Gets the number of successful test cases.
+ int successful_test_case_count() const;
+
+ // Gets the number of failed test cases.
+ int failed_test_case_count() const;
+
+ // Gets the number of all test cases.
+ int total_test_case_count() const;
+
+ // Gets the number of all test cases that contain at least one test
+ // that should run.
+ int test_case_to_run_count() const;
+
+ // Gets the number of successful tests.
+ int successful_test_count() const;
+
+ // Gets the number of failed tests.
+ int failed_test_count() const;
+
+ // Gets the number of disabled tests.
+ int disabled_test_count() const;
+
+ // Gets the number of all tests.
+ int total_test_count() const;
+
+ // Gets the number of tests that should run.
+ int test_to_run_count() const;
+
+ // Gets the elapsed time, in milliseconds.
+ TimeInMillis elapsed_time() const { return elapsed_time_; }
+
+ // Returns true iff the unit test passed (i.e. all test cases passed).
+ bool Passed() const { return !Failed(); }
+
+ // Returns true iff the unit test failed (i.e. some test case failed
+ // or something outside of all tests failed).
+ bool Failed() const {
+ return failed_test_case_count() > 0 || ad_hoc_test_result()->Failed();
+ }
+
+ // Gets the i-th test case among all the test cases. i can range from 0 to
+ // total_test_case_count() - 1. If i is not in that range, returns NULL.
+ const TestCase* GetTestCase(int i) const {
+ const int index = GetElementOr(test_case_indices_, i, -1);
+ return index < 0 ? NULL : test_cases_[i];
+ }
+
+ // Gets the i-th test case among all the test cases. i can range from 0 to
+ // total_test_case_count() - 1. If i is not in that range, returns NULL.
+ TestCase* GetMutableTestCase(int i) {
+ const int index = GetElementOr(test_case_indices_, i, -1);
+ return index < 0 ? NULL : test_cases_[index];
+ }
+
+ // Provides access to the event listener list.
+ TestEventListeners* listeners() { return &listeners_; }
+
+ // Returns the TestResult for the test that's currently running, or
+ // the TestResult for the ad hoc test if no test is running.
+ TestResult* current_test_result();
+
+ // Returns the TestResult for the ad hoc test.
+ const TestResult* ad_hoc_test_result() const { return &ad_hoc_test_result_; }
+
+ // Sets the OS stack trace getter.
+ //
+ // Does nothing if the input and the current OS stack trace getter
+ // are the same; otherwise, deletes the old getter and makes the
+ // input the current getter.
+ void set_os_stack_trace_getter(OsStackTraceGetterInterface* getter);
+
+ // Returns the current OS stack trace getter if it is not NULL;
+ // otherwise, creates an OsStackTraceGetter, makes it the current
+ // getter, and returns it.
+ OsStackTraceGetterInterface* os_stack_trace_getter();
+
+ // Returns the current OS stack trace as a String.
+ //
+ // The maximum number of stack frames to be included is specified by
+ // the gtest_stack_trace_depth flag. The skip_count parameter
+ // specifies the number of top frames to be skipped, which doesn't
+ // count against the number of frames to be included.
+ //
+ // For example, if Foo() calls Bar(), which in turn calls
+ // CurrentOsStackTraceExceptTop(1), Foo() will be included in the
+ // trace but Bar() and CurrentOsStackTraceExceptTop() won't.
+ String CurrentOsStackTraceExceptTop(int skip_count);
+
+ // Finds and returns a TestCase with the given name. If one doesn't
+ // exist, creates one and returns it.
+ //
+ // Arguments:
+ //
+ // test_case_name: name of the test case
+ // type_param: the name of the test's type parameter, or NULL if
+ // this is not a typed or a type-parameterized test.
+ // set_up_tc: pointer to the function that sets up the test case
+ // tear_down_tc: pointer to the function that tears down the test case
+ TestCase* GetTestCase(const char* test_case_name,
+ const char* type_param,
+ Test::SetUpTestCaseFunc set_up_tc,
+ Test::TearDownTestCaseFunc tear_down_tc);
+
+ // Adds a TestInfo to the unit test.
+ //
+ // Arguments:
+ //
+ // set_up_tc: pointer to the function that sets up the test case
+ // tear_down_tc: pointer to the function that tears down the test case
+ // test_info: the TestInfo object
+ void AddTestInfo(Test::SetUpTestCaseFunc set_up_tc,
+ Test::TearDownTestCaseFunc tear_down_tc,
+ TestInfo* test_info) {
+ // In order to support thread-safe death tests, we need to
+ // remember the original working directory when the test program
+ // was first invoked. We cannot do this in RUN_ALL_TESTS(), as
+ // the user may have changed the current directory before calling
+ // RUN_ALL_TESTS(). Therefore we capture the current directory in
+ // AddTestInfo(), which is called to register a TEST or TEST_F
+ // before main() is reached.
+ if (original_working_dir_.IsEmpty()) {
+ original_working_dir_.Set(FilePath::GetCurrentDir());
+ GTEST_CHECK_(!original_working_dir_.IsEmpty())
+ << "Failed to get the current working directory.";
+ }
+
+ GetTestCase(test_info->test_case_name(),
+ test_info->type_param(),
+ set_up_tc,
+ tear_down_tc)->AddTestInfo(test_info);
+ }
+
+#if GTEST_HAS_PARAM_TEST
+ // Returns ParameterizedTestCaseRegistry object used to keep track of
+ // value-parameterized tests and instantiate and register them.
+ internal::ParameterizedTestCaseRegistry& parameterized_test_registry() {
+ return parameterized_test_registry_;
+ }
+#endif // GTEST_HAS_PARAM_TEST
+
+ // Sets the TestCase object for the test that's currently running.
+ void set_current_test_case(TestCase* a_current_test_case) {
+ current_test_case_ = a_current_test_case;
+ }
+
+ // Sets the TestInfo object for the test that's currently running. If
+ // current_test_info is NULL, the assertion results will be stored in
+ // ad_hoc_test_result_.
+ void set_current_test_info(TestInfo* a_current_test_info) {
+ current_test_info_ = a_current_test_info;
+ }
+
+ // Registers all parameterized tests defined using TEST_P and
+ // INSTANTIATE_TEST_CASE_P, creating regular tests for each test/parameter
+ // combination. This method can be called more then once; it has guards
+ // protecting from registering the tests more then once. If
+ // value-parameterized tests are disabled, RegisterParameterizedTests is
+ // present but does nothing.
+ void RegisterParameterizedTests();
+
+ // Runs all tests in this UnitTest object, prints the result, and
+ // returns true if all tests are successful. If any exception is
+ // thrown during a test, this test is considered to be failed, but
+ // the rest of the tests will still be run.
+ bool RunAllTests();
+
+ // Clears the results of all tests, except the ad hoc tests.
+ void ClearNonAdHocTestResult() {
+ ForEach(test_cases_, TestCase::ClearTestCaseResult);
+ }
+
+ // Clears the results of ad-hoc test assertions.
+ void ClearAdHocTestResult() {
+ ad_hoc_test_result_.Clear();
+ }
+
+ enum ReactionToSharding {
+ HONOR_SHARDING_PROTOCOL,
+ IGNORE_SHARDING_PROTOCOL
+ };
+
+ // Matches the full name of each test against the user-specified
+ // filter to decide whether the test should run, then records the
+ // result in each TestCase and TestInfo object.
+ // If shard_tests == HONOR_SHARDING_PROTOCOL, further filters tests
+ // based on sharding variables in the environment.
+ // Returns the number of tests that should run.
+ int FilterTests(ReactionToSharding shard_tests);
+
+ // Prints the names of the tests matching the user-specified filter flag.
+ void ListTestsMatchingFilter();
+
+ const TestCase* current_test_case() const { return current_test_case_; }
+ TestInfo* current_test_info() { return current_test_info_; }
+ const TestInfo* current_test_info() const { return current_test_info_; }
+
+ // Returns the vector of environments that need to be set-up/torn-down
+ // before/after the tests are run.
+ std::vector<Environment*>& environments() { return environments_; }
+
+ // Getters for the per-thread Google Test trace stack.
+ std::vector<TraceInfo>& gtest_trace_stack() {
+ return *(gtest_trace_stack_.pointer());
+ }
+ const std::vector<TraceInfo>& gtest_trace_stack() const {
+ return gtest_trace_stack_.get();
+ }
+
+#if GTEST_HAS_DEATH_TEST
+ void InitDeathTestSubprocessControlInfo() {
+ internal_run_death_test_flag_.reset(ParseInternalRunDeathTestFlag());
+ }
+ // Returns a pointer to the parsed --gtest_internal_run_death_test
+ // flag, or NULL if that flag was not specified.
+ // This information is useful only in a death test child process.
+ // Must not be called before a call to InitGoogleTest.
+ const InternalRunDeathTestFlag* internal_run_death_test_flag() const {
+ return internal_run_death_test_flag_.get();
+ }
+
+ // Returns a pointer to the current death test factory.
+ internal::DeathTestFactory* death_test_factory() {
+ return death_test_factory_.get();
+ }
+
+ void SuppressTestEventsIfInSubprocess();
+
+ friend class ReplaceDeathTestFactory;
+#endif // GTEST_HAS_DEATH_TEST
+
+ // Initializes the event listener performing XML output as specified by
+ // UnitTestOptions. Must not be called before InitGoogleTest.
+ void ConfigureXmlOutput();
+
+#if GTEST_CAN_STREAM_RESULTS_
+ // Initializes the event listener for streaming test results to a socket.
+ // Must not be called before InitGoogleTest.
+ void ConfigureStreamingOutput();
+#endif
+
+ // Performs initialization dependent upon flag values obtained in
+ // ParseGoogleTestFlagsOnly. Is called from InitGoogleTest after the call to
+ // ParseGoogleTestFlagsOnly. In case a user neglects to call InitGoogleTest
+ // this function is also called from RunAllTests. Since this function can be
+ // called more than once, it has to be idempotent.
+ void PostFlagParsingInit();
+
+ // Gets the random seed used at the start of the current test iteration.
+ int random_seed() const { return random_seed_; }
+
+ // Gets the random number generator.
+ internal::Random* random() { return &random_; }
+
+ // Shuffles all test cases, and the tests within each test case,
+ // making sure that death tests are still run first.
+ void ShuffleTests();
+
+ // Restores the test cases and tests to their order before the first shuffle.
+ void UnshuffleTests();
+
+ // Returns the value of GTEST_FLAG(catch_exceptions) at the moment
+ // UnitTest::Run() starts.
+ bool catch_exceptions() const { return catch_exceptions_; }
+
+ private:
+ friend class ::testing::UnitTest;
+
+ // Used by UnitTest::Run() to capture the state of
+ // GTEST_FLAG(catch_exceptions) at the moment it starts.
+ void set_catch_exceptions(bool value) { catch_exceptions_ = value; }
+
+ // The UnitTest object that owns this implementation object.
+ UnitTest* const parent_;
+
+ // The working directory when the first TEST() or TEST_F() was
+ // executed.
+ internal::FilePath original_working_dir_;
+
+ // The default test part result reporters.
+ DefaultGlobalTestPartResultReporter default_global_test_part_result_reporter_;
+ DefaultPerThreadTestPartResultReporter
+ default_per_thread_test_part_result_reporter_;
+
+ // Points to (but doesn't own) the global test part result reporter.
+ TestPartResultReporterInterface* global_test_part_result_repoter_;
+
+ // Protects read and write access to global_test_part_result_reporter_.
+ internal::Mutex global_test_part_result_reporter_mutex_;
+
+ // Points to (but doesn't own) the per-thread test part result reporter.
+ internal::ThreadLocal<TestPartResultReporterInterface*>
+ per_thread_test_part_result_reporter_;
+
+ // The vector of environments that need to be set-up/torn-down
+ // before/after the tests are run.
+ std::vector<Environment*> environments_;
+
+ // The vector of TestCases in their original order. It owns the
+ // elements in the vector.
+ std::vector<TestCase*> test_cases_;
+
+ // Provides a level of indirection for the test case list to allow
+ // easy shuffling and restoring the test case order. The i-th
+ // element of this vector is the index of the i-th test case in the
+ // shuffled order.
+ std::vector<int> test_case_indices_;
+
+#if GTEST_HAS_PARAM_TEST
+ // ParameterizedTestRegistry object used to register value-parameterized
+ // tests.
+ internal::ParameterizedTestCaseRegistry parameterized_test_registry_;
+
+ // Indicates whether RegisterParameterizedTests() has been called already.
+ bool parameterized_tests_registered_;
+#endif // GTEST_HAS_PARAM_TEST
+
+ // Index of the last death test case registered. Initially -1.
+ int last_death_test_case_;
+
+ // This points to the TestCase for the currently running test. It
+ // changes as Google Test goes through one test case after another.
+ // When no test is running, this is set to NULL and Google Test
+ // stores assertion results in ad_hoc_test_result_. Initially NULL.
+ TestCase* current_test_case_;
+
+ // This points to the TestInfo for the currently running test. It
+ // changes as Google Test goes through one test after another. When
+ // no test is running, this is set to NULL and Google Test stores
+ // assertion results in ad_hoc_test_result_. Initially NULL.
+ TestInfo* current_test_info_;
+
+ // Normally, a user only writes assertions inside a TEST or TEST_F,
+ // or inside a function called by a TEST or TEST_F. Since Google
+ // Test keeps track of which test is current running, it can
+ // associate such an assertion with the test it belongs to.
+ //
+ // If an assertion is encountered when no TEST or TEST_F is running,
+ // Google Test attributes the assertion result to an imaginary "ad hoc"
+ // test, and records the result in ad_hoc_test_result_.
+ TestResult ad_hoc_test_result_;
+
+ // The list of event listeners that can be used to track events inside
+ // Google Test.
+ TestEventListeners listeners_;
+
+ // The OS stack trace getter. Will be deleted when the UnitTest
+ // object is destructed. By default, an OsStackTraceGetter is used,
+ // but the user can set this field to use a custom getter if that is
+ // desired.
+ OsStackTraceGetterInterface* os_stack_trace_getter_;
+
+ // True iff PostFlagParsingInit() has been called.
+ bool post_flag_parse_init_performed_;
+
+ // The random number seed used at the beginning of the test run.
+ int random_seed_;
+
+ // Our random number generator.
+ internal::Random random_;
+
+ // How long the test took to run, in milliseconds.
+ TimeInMillis elapsed_time_;
+
+#if GTEST_HAS_DEATH_TEST
+ // The decomposed components of the gtest_internal_run_death_test flag,
+ // parsed when RUN_ALL_TESTS is called.
+ internal::scoped_ptr<InternalRunDeathTestFlag> internal_run_death_test_flag_;
+ internal::scoped_ptr<internal::DeathTestFactory> death_test_factory_;
+#endif // GTEST_HAS_DEATH_TEST
+
+ // A per-thread stack of traces created by the SCOPED_TRACE() macro.
+ internal::ThreadLocal<std::vector<TraceInfo> > gtest_trace_stack_;
+
+ // The value of GTEST_FLAG(catch_exceptions) at the moment RunAllTests()
+ // starts.
+ bool catch_exceptions_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTestImpl);
+}; // class UnitTestImpl
+
+// Convenience function for accessing the global UnitTest
+// implementation object.
+inline UnitTestImpl* GetUnitTestImpl() {
+ return UnitTest::GetInstance()->impl();
+}
+
+#if GTEST_USES_SIMPLE_RE
+
+// Internal helper functions for implementing the simple regular
+// expression matcher.
+GTEST_API_ bool IsInSet(char ch, const char* str);
+GTEST_API_ bool IsAsciiDigit(char ch);
+GTEST_API_ bool IsAsciiPunct(char ch);
+GTEST_API_ bool IsRepeat(char ch);
+GTEST_API_ bool IsAsciiWhiteSpace(char ch);
+GTEST_API_ bool IsAsciiWordChar(char ch);
+GTEST_API_ bool IsValidEscape(char ch);
+GTEST_API_ bool AtomMatchesChar(bool escaped, char pattern, char ch);
+GTEST_API_ bool ValidateRegex(const char* regex);
+GTEST_API_ bool MatchRegexAtHead(const char* regex, const char* str);
+GTEST_API_ bool MatchRepetitionAndRegexAtHead(
+ bool escaped, char ch, char repeat, const char* regex, const char* str);
+GTEST_API_ bool MatchRegexAnywhere(const char* regex, const char* str);
+
+#endif // GTEST_USES_SIMPLE_RE
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test.
+GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, char** argv);
+GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv);
+
+#if GTEST_HAS_DEATH_TEST
+
+// Returns the message describing the last system error, regardless of the
+// platform.
+GTEST_API_ String GetLastErrnoDescription();
+
+# if GTEST_OS_WINDOWS
+// Provides leak-safe Windows kernel handle ownership.
+class AutoHandle {
+ public:
+ AutoHandle() : handle_(INVALID_HANDLE_VALUE) {}
+ explicit AutoHandle(HANDLE handle) : handle_(handle) {}
+
+ ~AutoHandle() { Reset(); }
+
+ HANDLE Get() const { return handle_; }
+ void Reset() { Reset(INVALID_HANDLE_VALUE); }
+ void Reset(HANDLE handle) {
+ if (handle != handle_) {
+ if (handle_ != INVALID_HANDLE_VALUE)
+ ::CloseHandle(handle_);
+ handle_ = handle;
+ }
+ }
+
+ private:
+ HANDLE handle_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(AutoHandle);
+};
+# endif // GTEST_OS_WINDOWS
+
+// Attempts to parse a string into a positive integer pointed to by the
+// number parameter. Returns true if that is possible.
+// GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we can use
+// it here.
+template <typename Integer>
+bool ParseNaturalNumber(const ::std::string& str, Integer* number) {
+ // Fail fast if the given string does not begin with a digit;
+ // this bypasses strtoXXX's "optional leading whitespace and plus
+ // or minus sign" semantics, which are undesirable here.
+ if (str.empty() || !IsDigit(str[0])) {
+ return false;
+ }
+ errno = 0;
+
+ char* end;
+ // BiggestConvertible is the largest integer type that system-provided
+ // string-to-number conversion routines can return.
+
+# if GTEST_OS_WINDOWS && !defined(__GNUC__)
+
+ // MSVC and C++ Builder define __int64 instead of the standard long long.
+ typedef unsigned __int64 BiggestConvertible;
+ const BiggestConvertible parsed = _strtoui64(str.c_str(), &end, 10);
+
+# else
+
+ typedef unsigned long long BiggestConvertible; // NOLINT
+ const BiggestConvertible parsed = strtoull(str.c_str(), &end, 10);
+
+# endif // GTEST_OS_WINDOWS && !defined(__GNUC__)
+
+ const bool parse_success = *end == '\0' && errno == 0;
+
+ // TODO(vladl@google.com): Convert this to compile time assertion when it is
+ // available.
+ GTEST_CHECK_(sizeof(Integer) <= sizeof(parsed));
+
+ const Integer result = static_cast<Integer>(parsed);
+ if (parse_success && static_cast<BiggestConvertible>(result) == parsed) {
+ *number = result;
+ return true;
+ }
+ return false;
+}
+#endif // GTEST_HAS_DEATH_TEST
+
+// TestResult contains some private methods that should be hidden from
+// Google Test user but are required for testing. This class allow our tests
+// to access them.
+//
+// This class is supplied only for the purpose of testing Google Test's own
+// constructs. Do not use it in user tests, either directly or indirectly.
+class TestResultAccessor {
+ public:
+ static void RecordProperty(TestResult* test_result,
+ const TestProperty& property) {
+ test_result->RecordProperty(property);
+ }
+
+ static void ClearTestPartResults(TestResult* test_result) {
+ test_result->ClearTestPartResults();
+ }
+
+ static const std::vector<testing::TestPartResult>& test_part_results(
+ const TestResult& test_result) {
+ return test_result.test_part_results();
+ }
+};
+
+} // namespace internal
+} // namespace testing
+
+#endif // GTEST_SRC_GTEST_INTERNAL_INL_H_
diff --git a/libvpx/third_party/googletest/src/src/gtest-port.cc b/libvpx/third_party/googletest/src/src/gtest-port.cc
new file mode 100644
index 0000000..b860d48
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest-port.cc
@@ -0,0 +1,746 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "gtest/internal/gtest-port.h"
+
+#include <limits.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+
+#if GTEST_OS_WINDOWS_MOBILE
+# include <windows.h> // For TerminateProcess()
+#elif GTEST_OS_WINDOWS
+# include <io.h>
+# include <sys/stat.h>
+#else
+# include <unistd.h>
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+#if GTEST_OS_MAC
+# include <mach/mach_init.h>
+# include <mach/task.h>
+# include <mach/vm_map.h>
+#endif // GTEST_OS_MAC
+
+#include "gtest/gtest-spi.h"
+#include "gtest/gtest-message.h"
+#include "gtest/internal/gtest-internal.h"
+#include "gtest/internal/gtest-string.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+namespace internal {
+
+#if defined(_MSC_VER) || defined(__BORLANDC__)
+// MSVC and C++Builder do not provide a definition of STDERR_FILENO.
+const int kStdOutFileno = 1;
+const int kStdErrFileno = 2;
+#else
+const int kStdOutFileno = STDOUT_FILENO;
+const int kStdErrFileno = STDERR_FILENO;
+#endif // _MSC_VER
+
+#if GTEST_OS_MAC
+
+// Returns the number of threads running in the process, or 0 to indicate that
+// we cannot detect it.
+size_t GetThreadCount() {
+ const task_t task = mach_task_self();
+ mach_msg_type_number_t thread_count;
+ thread_act_array_t thread_list;
+ const kern_return_t status = task_threads(task, &thread_list, &thread_count);
+ if (status == KERN_SUCCESS) {
+ // task_threads allocates resources in thread_list and we need to free them
+ // to avoid leaks.
+ vm_deallocate(task,
+ reinterpret_cast<vm_address_t>(thread_list),
+ sizeof(thread_t) * thread_count);
+ return static_cast<size_t>(thread_count);
+ } else {
+ return 0;
+ }
+}
+
+#else
+
+size_t GetThreadCount() {
+ // There's no portable way to detect the number of threads, so we just
+ // return 0 to indicate that we cannot detect it.
+ return 0;
+}
+
+#endif // GTEST_OS_MAC
+
+#if GTEST_USES_POSIX_RE
+
+// Implements RE. Currently only needed for death tests.
+
+RE::~RE() {
+ if (is_valid_) {
+ // regfree'ing an invalid regex might crash because the content
+ // of the regex is undefined. Since the regex's are essentially
+ // the same, one cannot be valid (or invalid) without the other
+ // being so too.
+ regfree(&partial_regex_);
+ regfree(&full_regex_);
+ }
+ free(const_cast<char*>(pattern_));
+}
+
+// Returns true iff regular expression re matches the entire str.
+bool RE::FullMatch(const char* str, const RE& re) {
+ if (!re.is_valid_) return false;
+
+ regmatch_t match;
+ return regexec(&re.full_regex_, str, 1, &match, 0) == 0;
+}
+
+// Returns true iff regular expression re matches a substring of str
+// (including str itself).
+bool RE::PartialMatch(const char* str, const RE& re) {
+ if (!re.is_valid_) return false;
+
+ regmatch_t match;
+ return regexec(&re.partial_regex_, str, 1, &match, 0) == 0;
+}
+
+// Initializes an RE from its string representation.
+void RE::Init(const char* regex) {
+ pattern_ = posix::StrDup(regex);
+
+ // Reserves enough bytes to hold the regular expression used for a
+ // full match.
+ const size_t full_regex_len = strlen(regex) + 10;
+ char* const full_pattern = new char[full_regex_len];
+
+ snprintf(full_pattern, full_regex_len, "^(%s)$", regex);
+ is_valid_ = regcomp(&full_regex_, full_pattern, REG_EXTENDED) == 0;
+ // We want to call regcomp(&partial_regex_, ...) even if the
+ // previous expression returns false. Otherwise partial_regex_ may
+ // not be properly initialized can may cause trouble when it's
+ // freed.
+ //
+ // Some implementation of POSIX regex (e.g. on at least some
+ // versions of Cygwin) doesn't accept the empty string as a valid
+ // regex. We change it to an equivalent form "()" to be safe.
+ if (is_valid_) {
+ const char* const partial_regex = (*regex == '\0') ? "()" : regex;
+ is_valid_ = regcomp(&partial_regex_, partial_regex, REG_EXTENDED) == 0;
+ }
+ EXPECT_TRUE(is_valid_)
+ << "Regular expression \"" << regex
+ << "\" is not a valid POSIX Extended regular expression.";
+
+ delete[] full_pattern;
+}
+
+#elif GTEST_USES_SIMPLE_RE
+
+// Returns true iff ch appears anywhere in str (excluding the
+// terminating '\0' character).
+bool IsInSet(char ch, const char* str) {
+ return ch != '\0' && strchr(str, ch) != NULL;
+}
+
+// Returns true iff ch belongs to the given classification. Unlike
+// similar functions in <ctype.h>, these aren't affected by the
+// current locale.
+bool IsAsciiDigit(char ch) { return '0' <= ch && ch <= '9'; }
+bool IsAsciiPunct(char ch) {
+ return IsInSet(ch, "^-!\"#$%&'()*+,./:;<=>?@[\\]_`{|}~");
+}
+bool IsRepeat(char ch) { return IsInSet(ch, "?*+"); }
+bool IsAsciiWhiteSpace(char ch) { return IsInSet(ch, " \f\n\r\t\v"); }
+bool IsAsciiWordChar(char ch) {
+ return ('a' <= ch && ch <= 'z') || ('A' <= ch && ch <= 'Z') ||
+ ('0' <= ch && ch <= '9') || ch == '_';
+}
+
+// Returns true iff "\\c" is a supported escape sequence.
+bool IsValidEscape(char c) {
+ return (IsAsciiPunct(c) || IsInSet(c, "dDfnrsStvwW"));
+}
+
+// Returns true iff the given atom (specified by escaped and pattern)
+// matches ch. The result is undefined if the atom is invalid.
+bool AtomMatchesChar(bool escaped, char pattern_char, char ch) {
+ if (escaped) { // "\\p" where p is pattern_char.
+ switch (pattern_char) {
+ case 'd': return IsAsciiDigit(ch);
+ case 'D': return !IsAsciiDigit(ch);
+ case 'f': return ch == '\f';
+ case 'n': return ch == '\n';
+ case 'r': return ch == '\r';
+ case 's': return IsAsciiWhiteSpace(ch);
+ case 'S': return !IsAsciiWhiteSpace(ch);
+ case 't': return ch == '\t';
+ case 'v': return ch == '\v';
+ case 'w': return IsAsciiWordChar(ch);
+ case 'W': return !IsAsciiWordChar(ch);
+ }
+ return IsAsciiPunct(pattern_char) && pattern_char == ch;
+ }
+
+ return (pattern_char == '.' && ch != '\n') || pattern_char == ch;
+}
+
+// Helper function used by ValidateRegex() to format error messages.
+String FormatRegexSyntaxError(const char* regex, int index) {
+ return (Message() << "Syntax error at index " << index
+ << " in simple regular expression \"" << regex << "\": ").GetString();
+}
+
+// Generates non-fatal failures and returns false if regex is invalid;
+// otherwise returns true.
+bool ValidateRegex(const char* regex) {
+ if (regex == NULL) {
+ // TODO(wan@google.com): fix the source file location in the
+ // assertion failures to match where the regex is used in user
+ // code.
+ ADD_FAILURE() << "NULL is not a valid simple regular expression.";
+ return false;
+ }
+
+ bool is_valid = true;
+
+ // True iff ?, *, or + can follow the previous atom.
+ bool prev_repeatable = false;
+ for (int i = 0; regex[i]; i++) {
+ if (regex[i] == '\\') { // An escape sequence
+ i++;
+ if (regex[i] == '\0') {
+ ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1)
+ << "'\\' cannot appear at the end.";
+ return false;
+ }
+
+ if (!IsValidEscape(regex[i])) {
+ ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1)
+ << "invalid escape sequence \"\\" << regex[i] << "\".";
+ is_valid = false;
+ }
+ prev_repeatable = true;
+ } else { // Not an escape sequence.
+ const char ch = regex[i];
+
+ if (ch == '^' && i > 0) {
+ ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+ << "'^' can only appear at the beginning.";
+ is_valid = false;
+ } else if (ch == '$' && regex[i + 1] != '\0') {
+ ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+ << "'$' can only appear at the end.";
+ is_valid = false;
+ } else if (IsInSet(ch, "()[]{}|")) {
+ ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+ << "'" << ch << "' is unsupported.";
+ is_valid = false;
+ } else if (IsRepeat(ch) && !prev_repeatable) {
+ ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+ << "'" << ch << "' can only follow a repeatable token.";
+ is_valid = false;
+ }
+
+ prev_repeatable = !IsInSet(ch, "^$?*+");
+ }
+ }
+
+ return is_valid;
+}
+
+// Matches a repeated regex atom followed by a valid simple regular
+// expression. The regex atom is defined as c if escaped is false,
+// or \c otherwise. repeat is the repetition meta character (?, *,
+// or +). The behavior is undefined if str contains too many
+// characters to be indexable by size_t, in which case the test will
+// probably time out anyway. We are fine with this limitation as
+// std::string has it too.
+bool MatchRepetitionAndRegexAtHead(
+ bool escaped, char c, char repeat, const char* regex,
+ const char* str) {
+ const size_t min_count = (repeat == '+') ? 1 : 0;
+ const size_t max_count = (repeat == '?') ? 1 :
+ static_cast<size_t>(-1) - 1;
+ // We cannot call numeric_limits::max() as it conflicts with the
+ // max() macro on Windows.
+
+ for (size_t i = 0; i <= max_count; ++i) {
+ // We know that the atom matches each of the first i characters in str.
+ if (i >= min_count && MatchRegexAtHead(regex, str + i)) {
+ // We have enough matches at the head, and the tail matches too.
+ // Since we only care about *whether* the pattern matches str
+ // (as opposed to *how* it matches), there is no need to find a
+ // greedy match.
+ return true;
+ }
+ if (str[i] == '\0' || !AtomMatchesChar(escaped, c, str[i]))
+ return false;
+ }
+ return false;
+}
+
+// Returns true iff regex matches a prefix of str. regex must be a
+// valid simple regular expression and not start with "^", or the
+// result is undefined.
+bool MatchRegexAtHead(const char* regex, const char* str) {
+ if (*regex == '\0') // An empty regex matches a prefix of anything.
+ return true;
+
+ // "$" only matches the end of a string. Note that regex being
+ // valid guarantees that there's nothing after "$" in it.
+ if (*regex == '$')
+ return *str == '\0';
+
+ // Is the first thing in regex an escape sequence?
+ const bool escaped = *regex == '\\';
+ if (escaped)
+ ++regex;
+ if (IsRepeat(regex[1])) {
+ // MatchRepetitionAndRegexAtHead() calls MatchRegexAtHead(), so
+ // here's an indirect recursion. It terminates as the regex gets
+ // shorter in each recursion.
+ return MatchRepetitionAndRegexAtHead(
+ escaped, regex[0], regex[1], regex + 2, str);
+ } else {
+ // regex isn't empty, isn't "$", and doesn't start with a
+ // repetition. We match the first atom of regex with the first
+ // character of str and recurse.
+ return (*str != '\0') && AtomMatchesChar(escaped, *regex, *str) &&
+ MatchRegexAtHead(regex + 1, str + 1);
+ }
+}
+
+// Returns true iff regex matches any substring of str. regex must be
+// a valid simple regular expression, or the result is undefined.
+//
+// The algorithm is recursive, but the recursion depth doesn't exceed
+// the regex length, so we won't need to worry about running out of
+// stack space normally. In rare cases the time complexity can be
+// exponential with respect to the regex length + the string length,
+// but usually it's must faster (often close to linear).
+bool MatchRegexAnywhere(const char* regex, const char* str) {
+ if (regex == NULL || str == NULL)
+ return false;
+
+ if (*regex == '^')
+ return MatchRegexAtHead(regex + 1, str);
+
+ // A successful match can be anywhere in str.
+ do {
+ if (MatchRegexAtHead(regex, str))
+ return true;
+ } while (*str++ != '\0');
+ return false;
+}
+
+// Implements the RE class.
+
+RE::~RE() {
+ free(const_cast<char*>(pattern_));
+ free(const_cast<char*>(full_pattern_));
+}
+
+// Returns true iff regular expression re matches the entire str.
+bool RE::FullMatch(const char* str, const RE& re) {
+ return re.is_valid_ && MatchRegexAnywhere(re.full_pattern_, str);
+}
+
+// Returns true iff regular expression re matches a substring of str
+// (including str itself).
+bool RE::PartialMatch(const char* str, const RE& re) {
+ return re.is_valid_ && MatchRegexAnywhere(re.pattern_, str);
+}
+
+// Initializes an RE from its string representation.
+void RE::Init(const char* regex) {
+ pattern_ = full_pattern_ = NULL;
+ if (regex != NULL) {
+ pattern_ = posix::StrDup(regex);
+ }
+
+ is_valid_ = ValidateRegex(regex);
+ if (!is_valid_) {
+ // No need to calculate the full pattern when the regex is invalid.
+ return;
+ }
+
+ const size_t len = strlen(regex);
+ // Reserves enough bytes to hold the regular expression used for a
+ // full match: we need space to prepend a '^', append a '$', and
+ // terminate the string with '\0'.
+ char* buffer = static_cast<char*>(malloc(len + 3));
+ full_pattern_ = buffer;
+
+ if (*regex != '^')
+ *buffer++ = '^'; // Makes sure full_pattern_ starts with '^'.
+
+ // We don't use snprintf or strncpy, as they trigger a warning when
+ // compiled with VC++ 8.0.
+ memcpy(buffer, regex, len);
+ buffer += len;
+
+ if (len == 0 || regex[len - 1] != '$')
+ *buffer++ = '$'; // Makes sure full_pattern_ ends with '$'.
+
+ *buffer = '\0';
+}
+
+#endif // GTEST_USES_POSIX_RE
+
+const char kUnknownFile[] = "unknown file";
+
+// Formats a source file path and a line number as they would appear
+// in an error message from the compiler used to compile this code.
+GTEST_API_ ::std::string FormatFileLocation(const char* file, int line) {
+ const char* const file_name = file == NULL ? kUnknownFile : file;
+
+ if (line < 0) {
+ return String::Format("%s:", file_name).c_str();
+ }
+#ifdef _MSC_VER
+ return String::Format("%s(%d):", file_name, line).c_str();
+#else
+ return String::Format("%s:%d:", file_name, line).c_str();
+#endif // _MSC_VER
+}
+
+// Formats a file location for compiler-independent XML output.
+// Although this function is not platform dependent, we put it next to
+// FormatFileLocation in order to contrast the two functions.
+// Note that FormatCompilerIndependentFileLocation() does NOT append colon
+// to the file location it produces, unlike FormatFileLocation().
+GTEST_API_ ::std::string FormatCompilerIndependentFileLocation(
+ const char* file, int line) {
+ const char* const file_name = file == NULL ? kUnknownFile : file;
+
+ if (line < 0)
+ return file_name;
+ else
+ return String::Format("%s:%d", file_name, line).c_str();
+}
+
+
+GTestLog::GTestLog(GTestLogSeverity severity, const char* file, int line)
+ : severity_(severity) {
+ const char* const marker =
+ severity == GTEST_INFO ? "[ INFO ]" :
+ severity == GTEST_WARNING ? "[WARNING]" :
+ severity == GTEST_ERROR ? "[ ERROR ]" : "[ FATAL ]";
+ GetStream() << ::std::endl << marker << " "
+ << FormatFileLocation(file, line).c_str() << ": ";
+}
+
+// Flushes the buffers and, if severity is GTEST_FATAL, aborts the program.
+GTestLog::~GTestLog() {
+ GetStream() << ::std::endl;
+ if (severity_ == GTEST_FATAL) {
+ fflush(stderr);
+ posix::Abort();
+ }
+}
+// Disable Microsoft deprecation warnings for POSIX functions called from
+// this class (creat, dup, dup2, and close)
+#ifdef _MSC_VER
+# pragma warning(push)
+# pragma warning(disable: 4996)
+#endif // _MSC_VER
+
+#if GTEST_HAS_STREAM_REDIRECTION
+
+// Object that captures an output stream (stdout/stderr).
+class CapturedStream {
+ public:
+ // The ctor redirects the stream to a temporary file.
+ CapturedStream(int fd) : fd_(fd), uncaptured_fd_(dup(fd)) {
+
+# if GTEST_OS_WINDOWS
+ char temp_dir_path[MAX_PATH + 1] = { '\0' }; // NOLINT
+ char temp_file_path[MAX_PATH + 1] = { '\0' }; // NOLINT
+
+ ::GetTempPathA(sizeof(temp_dir_path), temp_dir_path);
+ const UINT success = ::GetTempFileNameA(temp_dir_path,
+ "gtest_redir",
+ 0, // Generate unique file name.
+ temp_file_path);
+ GTEST_CHECK_(success != 0)
+ << "Unable to create a temporary file in " << temp_dir_path;
+ const int captured_fd = creat(temp_file_path, _S_IREAD | _S_IWRITE);
+ GTEST_CHECK_(captured_fd != -1) << "Unable to open temporary file "
+ << temp_file_path;
+ filename_ = temp_file_path;
+# else
+ // There's no guarantee that a test has write access to the
+ // current directory, so we create the temporary file in the /tmp
+ // directory instead.
+ char name_template[] = "/tmp/captured_stream.XXXXXX";
+ const int captured_fd = mkstemp(name_template);
+ filename_ = name_template;
+# endif // GTEST_OS_WINDOWS
+ fflush(NULL);
+ dup2(captured_fd, fd_);
+ close(captured_fd);
+ }
+
+ ~CapturedStream() {
+ remove(filename_.c_str());
+ }
+
+ String GetCapturedString() {
+ if (uncaptured_fd_ != -1) {
+ // Restores the original stream.
+ fflush(NULL);
+ dup2(uncaptured_fd_, fd_);
+ close(uncaptured_fd_);
+ uncaptured_fd_ = -1;
+ }
+
+ FILE* const file = posix::FOpen(filename_.c_str(), "r");
+ const String content = ReadEntireFile(file);
+ posix::FClose(file);
+ return content;
+ }
+
+ private:
+ // Reads the entire content of a file as a String.
+ static String ReadEntireFile(FILE* file);
+
+ // Returns the size (in bytes) of a file.
+ static size_t GetFileSize(FILE* file);
+
+ const int fd_; // A stream to capture.
+ int uncaptured_fd_;
+ // Name of the temporary file holding the stderr output.
+ ::std::string filename_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(CapturedStream);
+};
+
+// Returns the size (in bytes) of a file.
+size_t CapturedStream::GetFileSize(FILE* file) {
+ fseek(file, 0, SEEK_END);
+ return static_cast<size_t>(ftell(file));
+}
+
+// Reads the entire content of a file as a string.
+String CapturedStream::ReadEntireFile(FILE* file) {
+ const size_t file_size = GetFileSize(file);
+ char* const buffer = new char[file_size];
+
+ size_t bytes_last_read = 0; // # of bytes read in the last fread()
+ size_t bytes_read = 0; // # of bytes read so far
+
+ fseek(file, 0, SEEK_SET);
+
+ // Keeps reading the file until we cannot read further or the
+ // pre-determined file size is reached.
+ do {
+ bytes_last_read = fread(buffer+bytes_read, 1, file_size-bytes_read, file);
+ bytes_read += bytes_last_read;
+ } while (bytes_last_read > 0 && bytes_read < file_size);
+
+ const String content(buffer, bytes_read);
+ delete[] buffer;
+
+ return content;
+}
+
+# ifdef _MSC_VER
+# pragma warning(pop)
+# endif // _MSC_VER
+
+static CapturedStream* g_captured_stderr = NULL;
+static CapturedStream* g_captured_stdout = NULL;
+
+// Starts capturing an output stream (stdout/stderr).
+void CaptureStream(int fd, const char* stream_name, CapturedStream** stream) {
+ if (*stream != NULL) {
+ GTEST_LOG_(FATAL) << "Only one " << stream_name
+ << " capturer can exist at a time.";
+ }
+ *stream = new CapturedStream(fd);
+}
+
+// Stops capturing the output stream and returns the captured string.
+String GetCapturedStream(CapturedStream** captured_stream) {
+ const String content = (*captured_stream)->GetCapturedString();
+
+ delete *captured_stream;
+ *captured_stream = NULL;
+
+ return content;
+}
+
+// Starts capturing stdout.
+void CaptureStdout() {
+ CaptureStream(kStdOutFileno, "stdout", &g_captured_stdout);
+}
+
+// Starts capturing stderr.
+void CaptureStderr() {
+ CaptureStream(kStdErrFileno, "stderr", &g_captured_stderr);
+}
+
+// Stops capturing stdout and returns the captured string.
+String GetCapturedStdout() { return GetCapturedStream(&g_captured_stdout); }
+
+// Stops capturing stderr and returns the captured string.
+String GetCapturedStderr() { return GetCapturedStream(&g_captured_stderr); }
+
+#endif // GTEST_HAS_STREAM_REDIRECTION
+
+#if GTEST_HAS_DEATH_TEST
+
+// A copy of all command line arguments. Set by InitGoogleTest().
+::std::vector<String> g_argvs;
+
+// Returns the command line as a vector of strings.
+const ::std::vector<String>& GetArgvs() { return g_argvs; }
+
+#endif // GTEST_HAS_DEATH_TEST
+
+#if GTEST_OS_WINDOWS_MOBILE
+namespace posix {
+void Abort() {
+ DebugBreak();
+ TerminateProcess(GetCurrentProcess(), 1);
+}
+} // namespace posix
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+// Returns the name of the environment variable corresponding to the
+// given flag. For example, FlagToEnvVar("foo") will return
+// "GTEST_FOO" in the open-source version.
+static String FlagToEnvVar(const char* flag) {
+ const String full_flag =
+ (Message() << GTEST_FLAG_PREFIX_ << flag).GetString();
+
+ Message env_var;
+ for (size_t i = 0; i != full_flag.length(); i++) {
+ env_var << ToUpper(full_flag.c_str()[i]);
+ }
+
+ return env_var.GetString();
+}
+
+// Parses 'str' for a 32-bit signed integer. If successful, writes
+// the result to *value and returns true; otherwise leaves *value
+// unchanged and returns false.
+bool ParseInt32(const Message& src_text, const char* str, Int32* value) {
+ // Parses the environment variable as a decimal integer.
+ char* end = NULL;
+ const long long_value = strtol(str, &end, 10); // NOLINT
+
+ // Has strtol() consumed all characters in the string?
+ if (*end != '\0') {
+ // No - an invalid character was encountered.
+ Message msg;
+ msg << "WARNING: " << src_text
+ << " is expected to be a 32-bit integer, but actually"
+ << " has value \"" << str << "\".\n";
+ printf("%s", msg.GetString().c_str());
+ fflush(stdout);
+ return false;
+ }
+
+ // Is the parsed value in the range of an Int32?
+ const Int32 result = static_cast<Int32>(long_value);
+ if (long_value == LONG_MAX || long_value == LONG_MIN ||
+ // The parsed value overflows as a long. (strtol() returns
+ // LONG_MAX or LONG_MIN when the input overflows.)
+ result != long_value
+ // The parsed value overflows as an Int32.
+ ) {
+ Message msg;
+ msg << "WARNING: " << src_text
+ << " is expected to be a 32-bit integer, but actually"
+ << " has value " << str << ", which overflows.\n";
+ printf("%s", msg.GetString().c_str());
+ fflush(stdout);
+ return false;
+ }
+
+ *value = result;
+ return true;
+}
+
+// Reads and returns the Boolean environment variable corresponding to
+// the given flag; if it's not set, returns default_value.
+//
+// The value is considered true iff it's not "0".
+bool BoolFromGTestEnv(const char* flag, bool default_value) {
+ const String env_var = FlagToEnvVar(flag);
+ const char* const string_value = posix::GetEnv(env_var.c_str());
+ return string_value == NULL ?
+ default_value : strcmp(string_value, "0") != 0;
+}
+
+// Reads and returns a 32-bit integer stored in the environment
+// variable corresponding to the given flag; if it isn't set or
+// doesn't represent a valid 32-bit integer, returns default_value.
+Int32 Int32FromGTestEnv(const char* flag, Int32 default_value) {
+ const String env_var = FlagToEnvVar(flag);
+ const char* const string_value = posix::GetEnv(env_var.c_str());
+ if (string_value == NULL) {
+ // The environment variable is not set.
+ return default_value;
+ }
+
+ Int32 result = default_value;
+ if (!ParseInt32(Message() << "Environment variable " << env_var,
+ string_value, &result)) {
+ printf("The default value %s is used.\n",
+ (Message() << default_value).GetString().c_str());
+ fflush(stdout);
+ return default_value;
+ }
+
+ return result;
+}
+
+// Reads and returns the string environment variable corresponding to
+// the given flag; if it's not set, returns default_value.
+const char* StringFromGTestEnv(const char* flag, const char* default_value) {
+ const String env_var = FlagToEnvVar(flag);
+ const char* const value = posix::GetEnv(env_var.c_str());
+ return value == NULL ? default_value : value;
+}
+
+} // namespace internal
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/src/gtest-printers.cc b/libvpx/third_party/googletest/src/src/gtest-printers.cc
new file mode 100644
index 0000000..ed63c7b
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest-printers.cc
@@ -0,0 +1,356 @@
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Google Test - The Google C++ Testing Framework
+//
+// This file implements a universal value printer that can print a
+// value of any type T:
+//
+// void ::testing::internal::UniversalPrinter<T>::Print(value, ostream_ptr);
+//
+// It uses the << operator when possible, and prints the bytes in the
+// object otherwise. A user can override its behavior for a class
+// type Foo by defining either operator<<(::std::ostream&, const Foo&)
+// or void PrintTo(const Foo&, ::std::ostream*) in the namespace that
+// defines Foo.
+
+#include "gtest/gtest-printers.h"
+#include <ctype.h>
+#include <stdio.h>
+#include <ostream> // NOLINT
+#include <string>
+#include "gtest/internal/gtest-port.h"
+
+namespace testing {
+
+namespace {
+
+using ::std::ostream;
+
+#if GTEST_OS_WINDOWS_MOBILE // Windows CE does not define _snprintf_s.
+# define snprintf _snprintf
+#elif _MSC_VER >= 1400 // VC 8.0 and later deprecate snprintf and _snprintf.
+# define snprintf _snprintf_s
+#elif _MSC_VER
+# define snprintf _snprintf
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+// Prints a segment of bytes in the given object.
+void PrintByteSegmentInObjectTo(const unsigned char* obj_bytes, size_t start,
+ size_t count, ostream* os) {
+ char text[5] = "";
+ for (size_t i = 0; i != count; i++) {
+ const size_t j = start + i;
+ if (i != 0) {
+ // Organizes the bytes into groups of 2 for easy parsing by
+ // human.
+ if ((j % 2) == 0)
+ *os << ' ';
+ else
+ *os << '-';
+ }
+ snprintf(text, sizeof(text), "%02X", obj_bytes[j]);
+ *os << text;
+ }
+}
+
+// Prints the bytes in the given value to the given ostream.
+void PrintBytesInObjectToImpl(const unsigned char* obj_bytes, size_t count,
+ ostream* os) {
+ // Tells the user how big the object is.
+ *os << count << "-byte object <";
+
+ const size_t kThreshold = 132;
+ const size_t kChunkSize = 64;
+ // If the object size is bigger than kThreshold, we'll have to omit
+ // some details by printing only the first and the last kChunkSize
+ // bytes.
+ // TODO(wan): let the user control the threshold using a flag.
+ if (count < kThreshold) {
+ PrintByteSegmentInObjectTo(obj_bytes, 0, count, os);
+ } else {
+ PrintByteSegmentInObjectTo(obj_bytes, 0, kChunkSize, os);
+ *os << " ... ";
+ // Rounds up to 2-byte boundary.
+ const size_t resume_pos = (count - kChunkSize + 1)/2*2;
+ PrintByteSegmentInObjectTo(obj_bytes, resume_pos, count - resume_pos, os);
+ }
+ *os << ">";
+}
+
+} // namespace
+
+namespace internal2 {
+
+// Delegates to PrintBytesInObjectToImpl() to print the bytes in the
+// given object. The delegation simplifies the implementation, which
+// uses the << operator and thus is easier done outside of the
+// ::testing::internal namespace, which contains a << operator that
+// sometimes conflicts with the one in STL.
+void PrintBytesInObjectTo(const unsigned char* obj_bytes, size_t count,
+ ostream* os) {
+ PrintBytesInObjectToImpl(obj_bytes, count, os);
+}
+
+} // namespace internal2
+
+namespace internal {
+
+// Depending on the value of a char (or wchar_t), we print it in one
+// of three formats:
+// - as is if it's a printable ASCII (e.g. 'a', '2', ' '),
+// - as a hexidecimal escape sequence (e.g. '\x7F'), or
+// - as a special escape sequence (e.g. '\r', '\n').
+enum CharFormat {
+ kAsIs,
+ kHexEscape,
+ kSpecialEscape
+};
+
+// Returns true if c is a printable ASCII character. We test the
+// value of c directly instead of calling isprint(), which is buggy on
+// Windows Mobile.
+inline bool IsPrintableAscii(wchar_t c) {
+ return 0x20 <= c && c <= 0x7E;
+}
+
+// Prints a wide or narrow char c as a character literal without the
+// quotes, escaping it when necessary; returns how c was formatted.
+// The template argument UnsignedChar is the unsigned version of Char,
+// which is the type of c.
+template <typename UnsignedChar, typename Char>
+static CharFormat PrintAsCharLiteralTo(Char c, ostream* os) {
+ switch (static_cast<wchar_t>(c)) {
+ case L'\0':
+ *os << "\\0";
+ break;
+ case L'\'':
+ *os << "\\'";
+ break;
+ case L'\\':
+ *os << "\\\\";
+ break;
+ case L'\a':
+ *os << "\\a";
+ break;
+ case L'\b':
+ *os << "\\b";
+ break;
+ case L'\f':
+ *os << "\\f";
+ break;
+ case L'\n':
+ *os << "\\n";
+ break;
+ case L'\r':
+ *os << "\\r";
+ break;
+ case L'\t':
+ *os << "\\t";
+ break;
+ case L'\v':
+ *os << "\\v";
+ break;
+ default:
+ if (IsPrintableAscii(c)) {
+ *os << static_cast<char>(c);
+ return kAsIs;
+ } else {
+ *os << String::Format("\\x%X", static_cast<UnsignedChar>(c));
+ return kHexEscape;
+ }
+ }
+ return kSpecialEscape;
+}
+
+// Prints a char c as if it's part of a string literal, escaping it when
+// necessary; returns how c was formatted.
+static CharFormat PrintAsWideStringLiteralTo(wchar_t c, ostream* os) {
+ switch (c) {
+ case L'\'':
+ *os << "'";
+ return kAsIs;
+ case L'"':
+ *os << "\\\"";
+ return kSpecialEscape;
+ default:
+ return PrintAsCharLiteralTo<wchar_t>(c, os);
+ }
+}
+
+// Prints a char c as if it's part of a string literal, escaping it when
+// necessary; returns how c was formatted.
+static CharFormat PrintAsNarrowStringLiteralTo(char c, ostream* os) {
+ return PrintAsWideStringLiteralTo(static_cast<unsigned char>(c), os);
+}
+
+// Prints a wide or narrow character c and its code. '\0' is printed
+// as "'\\0'", other unprintable characters are also properly escaped
+// using the standard C++ escape sequence. The template argument
+// UnsignedChar is the unsigned version of Char, which is the type of c.
+template <typename UnsignedChar, typename Char>
+void PrintCharAndCodeTo(Char c, ostream* os) {
+ // First, print c as a literal in the most readable form we can find.
+ *os << ((sizeof(c) > 1) ? "L'" : "'");
+ const CharFormat format = PrintAsCharLiteralTo<UnsignedChar>(c, os);
+ *os << "'";
+
+ // To aid user debugging, we also print c's code in decimal, unless
+ // it's 0 (in which case c was printed as '\\0', making the code
+ // obvious).
+ if (c == 0)
+ return;
+ *os << " (" << String::Format("%d", c).c_str();
+
+ // For more convenience, we print c's code again in hexidecimal,
+ // unless c was already printed in the form '\x##' or the code is in
+ // [1, 9].
+ if (format == kHexEscape || (1 <= c && c <= 9)) {
+ // Do nothing.
+ } else {
+ *os << String::Format(", 0x%X",
+ static_cast<UnsignedChar>(c)).c_str();
+ }
+ *os << ")";
+}
+
+void PrintTo(unsigned char c, ::std::ostream* os) {
+ PrintCharAndCodeTo<unsigned char>(c, os);
+}
+void PrintTo(signed char c, ::std::ostream* os) {
+ PrintCharAndCodeTo<unsigned char>(c, os);
+}
+
+// Prints a wchar_t as a symbol if it is printable or as its internal
+// code otherwise and also as its code. L'\0' is printed as "L'\\0'".
+void PrintTo(wchar_t wc, ostream* os) {
+ PrintCharAndCodeTo<wchar_t>(wc, os);
+}
+
+// Prints the given array of characters to the ostream.
+// The array starts at *begin, the length is len, it may include '\0' characters
+// and may not be null-terminated.
+static void PrintCharsAsStringTo(const char* begin, size_t len, ostream* os) {
+ *os << "\"";
+ bool is_previous_hex = false;
+ for (size_t index = 0; index < len; ++index) {
+ const char cur = begin[index];
+ if (is_previous_hex && IsXDigit(cur)) {
+ // Previous character is of '\x..' form and this character can be
+ // interpreted as another hexadecimal digit in its number. Break string to
+ // disambiguate.
+ *os << "\" \"";
+ }
+ is_previous_hex = PrintAsNarrowStringLiteralTo(cur, os) == kHexEscape;
+ }
+ *os << "\"";
+}
+
+// Prints a (const) char array of 'len' elements, starting at address 'begin'.
+void UniversalPrintArray(const char* begin, size_t len, ostream* os) {
+ PrintCharsAsStringTo(begin, len, os);
+}
+
+// Prints the given array of wide characters to the ostream.
+// The array starts at *begin, the length is len, it may include L'\0'
+// characters and may not be null-terminated.
+static void PrintWideCharsAsStringTo(const wchar_t* begin, size_t len,
+ ostream* os) {
+ *os << "L\"";
+ bool is_previous_hex = false;
+ for (size_t index = 0; index < len; ++index) {
+ const wchar_t cur = begin[index];
+ if (is_previous_hex && isascii(cur) && IsXDigit(static_cast<char>(cur))) {
+ // Previous character is of '\x..' form and this character can be
+ // interpreted as another hexadecimal digit in its number. Break string to
+ // disambiguate.
+ *os << "\" L\"";
+ }
+ is_previous_hex = PrintAsWideStringLiteralTo(cur, os) == kHexEscape;
+ }
+ *os << "\"";
+}
+
+// Prints the given C string to the ostream.
+void PrintTo(const char* s, ostream* os) {
+ if (s == NULL) {
+ *os << "NULL";
+ } else {
+ *os << ImplicitCast_<const void*>(s) << " pointing to ";
+ PrintCharsAsStringTo(s, strlen(s), os);
+ }
+}
+
+// MSVC compiler can be configured to define whar_t as a typedef
+// of unsigned short. Defining an overload for const wchar_t* in that case
+// would cause pointers to unsigned shorts be printed as wide strings,
+// possibly accessing more memory than intended and causing invalid
+// memory accesses. MSVC defines _NATIVE_WCHAR_T_DEFINED symbol when
+// wchar_t is implemented as a native type.
+#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED)
+// Prints the given wide C string to the ostream.
+void PrintTo(const wchar_t* s, ostream* os) {
+ if (s == NULL) {
+ *os << "NULL";
+ } else {
+ *os << ImplicitCast_<const void*>(s) << " pointing to ";
+ PrintWideCharsAsStringTo(s, wcslen(s), os);
+ }
+}
+#endif // wchar_t is native
+
+// Prints a ::string object.
+#if GTEST_HAS_GLOBAL_STRING
+void PrintStringTo(const ::string& s, ostream* os) {
+ PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif // GTEST_HAS_GLOBAL_STRING
+
+void PrintStringTo(const ::std::string& s, ostream* os) {
+ PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+
+// Prints a ::wstring object.
+#if GTEST_HAS_GLOBAL_WSTRING
+void PrintWideStringTo(const ::wstring& s, ostream* os) {
+ PrintWideCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+#if GTEST_HAS_STD_WSTRING
+void PrintWideStringTo(const ::std::wstring& s, ostream* os) {
+ PrintWideCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif // GTEST_HAS_STD_WSTRING
+
+} // namespace internal
+
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/src/gtest-test-part.cc b/libvpx/third_party/googletest/src/src/gtest-test-part.cc
new file mode 100644
index 0000000..5ddc67c
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest-test-part.cc
@@ -0,0 +1,110 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+// The Google C++ Testing Framework (Google Test)
+
+#include "gtest/gtest-test-part.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+
+using internal::GetUnitTestImpl;
+
+// Gets the summary of the failure message by omitting the stack trace
+// in it.
+internal::String TestPartResult::ExtractSummary(const char* message) {
+ const char* const stack_trace = strstr(message, internal::kStackTraceMarker);
+ return stack_trace == NULL ? internal::String(message) :
+ internal::String(message, stack_trace - message);
+}
+
+// Prints a TestPartResult object.
+std::ostream& operator<<(std::ostream& os, const TestPartResult& result) {
+ return os
+ << result.file_name() << ":" << result.line_number() << ": "
+ << (result.type() == TestPartResult::kSuccess ? "Success" :
+ result.type() == TestPartResult::kFatalFailure ? "Fatal failure" :
+ "Non-fatal failure") << ":\n"
+ << result.message() << std::endl;
+}
+
+// Appends a TestPartResult to the array.
+void TestPartResultArray::Append(const TestPartResult& result) {
+ array_.push_back(result);
+}
+
+// Returns the TestPartResult at the given index (0-based).
+const TestPartResult& TestPartResultArray::GetTestPartResult(int index) const {
+ if (index < 0 || index >= size()) {
+ printf("\nInvalid index (%d) into TestPartResultArray.\n", index);
+ internal::posix::Abort();
+ }
+
+ return array_[index];
+}
+
+// Returns the number of TestPartResult objects in the array.
+int TestPartResultArray::size() const {
+ return static_cast<int>(array_.size());
+}
+
+namespace internal {
+
+HasNewFatalFailureHelper::HasNewFatalFailureHelper()
+ : has_new_fatal_failure_(false),
+ original_reporter_(GetUnitTestImpl()->
+ GetTestPartResultReporterForCurrentThread()) {
+ GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread(this);
+}
+
+HasNewFatalFailureHelper::~HasNewFatalFailureHelper() {
+ GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread(
+ original_reporter_);
+}
+
+void HasNewFatalFailureHelper::ReportTestPartResult(
+ const TestPartResult& result) {
+ if (result.fatally_failed())
+ has_new_fatal_failure_ = true;
+ original_reporter_->ReportTestPartResult(result);
+}
+
+} // namespace internal
+
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/src/gtest-typed-test.cc b/libvpx/third_party/googletest/src/src/gtest-typed-test.cc
new file mode 100644
index 0000000..a5cc88f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest-typed-test.cc
@@ -0,0 +1,110 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "gtest/gtest-typed-test.h"
+#include "gtest/gtest.h"
+
+namespace testing {
+namespace internal {
+
+#if GTEST_HAS_TYPED_TEST_P
+
+// Skips to the first non-space char in str. Returns an empty string if str
+// contains only whitespace characters.
+static const char* SkipSpaces(const char* str) {
+ while (IsSpace(*str))
+ str++;
+ return str;
+}
+
+// Verifies that registered_tests match the test names in
+// defined_test_names_; returns registered_tests if successful, or
+// aborts the program otherwise.
+const char* TypedTestCasePState::VerifyRegisteredTestNames(
+ const char* file, int line, const char* registered_tests) {
+ typedef ::std::set<const char*>::const_iterator DefinedTestIter;
+ registered_ = true;
+
+ // Skip initial whitespace in registered_tests since some
+ // preprocessors prefix stringizied literals with whitespace.
+ registered_tests = SkipSpaces(registered_tests);
+
+ Message errors;
+ ::std::set<String> tests;
+ for (const char* names = registered_tests; names != NULL;
+ names = SkipComma(names)) {
+ const String name = GetPrefixUntilComma(names);
+ if (tests.count(name) != 0) {
+ errors << "Test " << name << " is listed more than once.\n";
+ continue;
+ }
+
+ bool found = false;
+ for (DefinedTestIter it = defined_test_names_.begin();
+ it != defined_test_names_.end();
+ ++it) {
+ if (name == *it) {
+ found = true;
+ break;
+ }
+ }
+
+ if (found) {
+ tests.insert(name);
+ } else {
+ errors << "No test named " << name
+ << " can be found in this test case.\n";
+ }
+ }
+
+ for (DefinedTestIter it = defined_test_names_.begin();
+ it != defined_test_names_.end();
+ ++it) {
+ if (tests.count(*it) == 0) {
+ errors << "You forgot to list test " << *it << ".\n";
+ }
+ }
+
+ const String& errors_str = errors.GetString();
+ if (errors_str != "") {
+ fprintf(stderr, "%s %s", FormatFileLocation(file, line).c_str(),
+ errors_str.c_str());
+ fflush(stderr);
+ posix::Abort();
+ }
+
+ return registered_tests;
+}
+
+#endif // GTEST_HAS_TYPED_TEST_P
+
+} // namespace internal
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/src/gtest.cc b/libvpx/third_party/googletest/src/src/gtest.cc
new file mode 100644
index 0000000..904d9d7
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest.cc
@@ -0,0 +1,4898 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+
+#include "gtest/gtest.h"
+#include "gtest/gtest-spi.h"
+
+#include <ctype.h>
+#include <math.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <wchar.h>
+#include <wctype.h>
+
+#include <algorithm>
+#include <ostream> // NOLINT
+#include <sstream>
+#include <vector>
+
+#if GTEST_OS_LINUX
+
+// TODO(kenton@google.com): Use autoconf to detect availability of
+// gettimeofday().
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+
+# include <fcntl.h> // NOLINT
+# include <limits.h> // NOLINT
+# include <sched.h> // NOLINT
+// Declares vsnprintf(). This header is not available on Windows.
+# include <strings.h> // NOLINT
+# include <sys/mman.h> // NOLINT
+# include <sys/time.h> // NOLINT
+# include <unistd.h> // NOLINT
+# include <string>
+
+#elif GTEST_OS_SYMBIAN
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+# include <sys/time.h> // NOLINT
+
+#elif GTEST_OS_ZOS
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+# include <sys/time.h> // NOLINT
+
+// On z/OS we additionally need strings.h for strcasecmp.
+# include <strings.h> // NOLINT
+
+#elif GTEST_OS_WINDOWS_MOBILE // We are on Windows CE.
+
+# include <windows.h> // NOLINT
+
+#elif GTEST_OS_WINDOWS // We are on Windows proper.
+
+# include <io.h> // NOLINT
+# include <sys/timeb.h> // NOLINT
+# include <sys/types.h> // NOLINT
+# include <sys/stat.h> // NOLINT
+
+# if GTEST_OS_WINDOWS_MINGW
+// MinGW has gettimeofday() but not _ftime64().
+// TODO(kenton@google.com): Use autoconf to detect availability of
+// gettimeofday().
+// TODO(kenton@google.com): There are other ways to get the time on
+// Windows, like GetTickCount() or GetSystemTimeAsFileTime(). MinGW
+// supports these. consider using them instead.
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+# include <sys/time.h> // NOLINT
+# endif // GTEST_OS_WINDOWS_MINGW
+
+// cpplint thinks that the header is already included, so we want to
+// silence it.
+# include <windows.h> // NOLINT
+
+#else
+
+// Assume other platforms have gettimeofday().
+// TODO(kenton@google.com): Use autoconf to detect availability of
+// gettimeofday().
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+
+// cpplint thinks that the header is already included, so we want to
+// silence it.
+# include <sys/time.h> // NOLINT
+# include <unistd.h> // NOLINT
+
+#endif // GTEST_OS_LINUX
+
+#if GTEST_HAS_EXCEPTIONS
+# include <stdexcept>
+#endif
+
+#if GTEST_CAN_STREAM_RESULTS_
+# include <arpa/inet.h> // NOLINT
+# include <netdb.h> // NOLINT
+#endif
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+#if GTEST_OS_WINDOWS
+# define vsnprintf _vsnprintf
+#endif // GTEST_OS_WINDOWS
+
+namespace testing {
+
+using internal::CountIf;
+using internal::ForEach;
+using internal::GetElementOr;
+using internal::Shuffle;
+
+// Constants.
+
+// A test whose test case name or test name matches this filter is
+// disabled and not run.
+static const char kDisableTestFilter[] = "DISABLED_*:*/DISABLED_*";
+
+// A test case whose name matches this filter is considered a death
+// test case and will be run before test cases whose name doesn't
+// match this filter.
+static const char kDeathTestCaseFilter[] = "*DeathTest:*DeathTest/*";
+
+// A test filter that matches everything.
+static const char kUniversalFilter[] = "*";
+
+// The default output file for XML output.
+static const char kDefaultOutputFile[] = "test_detail.xml";
+
+// The environment variable name for the test shard index.
+static const char kTestShardIndex[] = "GTEST_SHARD_INDEX";
+// The environment variable name for the total number of test shards.
+static const char kTestTotalShards[] = "GTEST_TOTAL_SHARDS";
+// The environment variable name for the test shard status file.
+static const char kTestShardStatusFile[] = "GTEST_SHARD_STATUS_FILE";
+
+namespace internal {
+
+// The text used in failure messages to indicate the start of the
+// stack trace.
+const char kStackTraceMarker[] = "\nStack trace:\n";
+
+// g_help_flag is true iff the --help flag or an equivalent form is
+// specified on the command line.
+bool g_help_flag = false;
+
+} // namespace internal
+
+GTEST_DEFINE_bool_(
+ also_run_disabled_tests,
+ internal::BoolFromGTestEnv("also_run_disabled_tests", false),
+ "Run disabled tests too, in addition to the tests normally being run.");
+
+GTEST_DEFINE_bool_(
+ break_on_failure,
+ internal::BoolFromGTestEnv("break_on_failure", false),
+ "True iff a failed assertion should be a debugger break-point.");
+
+GTEST_DEFINE_bool_(
+ catch_exceptions,
+ internal::BoolFromGTestEnv("catch_exceptions", true),
+ "True iff " GTEST_NAME_
+ " should catch exceptions and treat them as test failures.");
+
+GTEST_DEFINE_string_(
+ color,
+ internal::StringFromGTestEnv("color", "auto"),
+ "Whether to use colors in the output. Valid values: yes, no, "
+ "and auto. 'auto' means to use colors if the output is "
+ "being sent to a terminal and the TERM environment variable "
+ "is set to xterm, xterm-color, xterm-256color, linux or cygwin.");
+
+GTEST_DEFINE_string_(
+ filter,
+ internal::StringFromGTestEnv("filter", kUniversalFilter),
+ "A colon-separated list of glob (not regex) patterns "
+ "for filtering the tests to run, optionally followed by a "
+ "'-' and a : separated list of negative patterns (tests to "
+ "exclude). A test is run if it matches one of the positive "
+ "patterns and does not match any of the negative patterns.");
+
+GTEST_DEFINE_bool_(list_tests, false,
+ "List all tests without running them.");
+
+GTEST_DEFINE_string_(
+ output,
+ internal::StringFromGTestEnv("output", ""),
+ "A format (currently must be \"xml\"), optionally followed "
+ "by a colon and an output file name or directory. A directory "
+ "is indicated by a trailing pathname separator. "
+ "Examples: \"xml:filename.xml\", \"xml::directoryname/\". "
+ "If a directory is specified, output files will be created "
+ "within that directory, with file-names based on the test "
+ "executable's name and, if necessary, made unique by adding "
+ "digits.");
+
+GTEST_DEFINE_bool_(
+ print_time,
+ internal::BoolFromGTestEnv("print_time", true),
+ "True iff " GTEST_NAME_
+ " should display elapsed time in text output.");
+
+GTEST_DEFINE_int32_(
+ random_seed,
+ internal::Int32FromGTestEnv("random_seed", 0),
+ "Random number seed to use when shuffling test orders. Must be in range "
+ "[1, 99999], or 0 to use a seed based on the current time.");
+
+GTEST_DEFINE_int32_(
+ repeat,
+ internal::Int32FromGTestEnv("repeat", 1),
+ "How many times to repeat each test. Specify a negative number "
+ "for repeating forever. Useful for shaking out flaky tests.");
+
+GTEST_DEFINE_bool_(
+ show_internal_stack_frames, false,
+ "True iff " GTEST_NAME_ " should include internal stack frames when "
+ "printing test failure stack traces.");
+
+GTEST_DEFINE_bool_(
+ shuffle,
+ internal::BoolFromGTestEnv("shuffle", false),
+ "True iff " GTEST_NAME_
+ " should randomize tests' order on every run.");
+
+GTEST_DEFINE_int32_(
+ stack_trace_depth,
+ internal::Int32FromGTestEnv("stack_trace_depth", kMaxStackTraceDepth),
+ "The maximum number of stack frames to print when an "
+ "assertion fails. The valid range is 0 through 100, inclusive.");
+
+GTEST_DEFINE_string_(
+ stream_result_to,
+ internal::StringFromGTestEnv("stream_result_to", ""),
+ "This flag specifies the host name and the port number on which to stream "
+ "test results. Example: \"localhost:555\". The flag is effective only on "
+ "Linux.");
+
+GTEST_DEFINE_bool_(
+ throw_on_failure,
+ internal::BoolFromGTestEnv("throw_on_failure", false),
+ "When this flag is specified, a failed assertion will throw an exception "
+ "if exceptions are enabled or exit the program with a non-zero code "
+ "otherwise.");
+
+namespace internal {
+
+// Generates a random number from [0, range), using a Linear
+// Congruential Generator (LCG). Crashes if 'range' is 0 or greater
+// than kMaxRange.
+UInt32 Random::Generate(UInt32 range) {
+ // These constants are the same as are used in glibc's rand(3).
+ state_ = (1103515245U*state_ + 12345U) % kMaxRange;
+
+ GTEST_CHECK_(range > 0)
+ << "Cannot generate a number in the range [0, 0).";
+ GTEST_CHECK_(range <= kMaxRange)
+ << "Generation of a number in [0, " << range << ") was requested, "
+ << "but this can only generate numbers in [0, " << kMaxRange << ").";
+
+ // Converting via modulus introduces a bit of downward bias, but
+ // it's simple, and a linear congruential generator isn't too good
+ // to begin with.
+ return state_ % range;
+}
+
+// GTestIsInitialized() returns true iff the user has initialized
+// Google Test. Useful for catching the user mistake of not initializing
+// Google Test before calling RUN_ALL_TESTS().
+//
+// A user must call testing::InitGoogleTest() to initialize Google
+// Test. g_init_gtest_count is set to the number of times
+// InitGoogleTest() has been called. We don't protect this variable
+// under a mutex as it is only accessed in the main thread.
+int g_init_gtest_count = 0;
+static bool GTestIsInitialized() { return g_init_gtest_count != 0; }
+
+// Iterates over a vector of TestCases, keeping a running sum of the
+// results of calling a given int-returning method on each.
+// Returns the sum.
+static int SumOverTestCaseList(const std::vector<TestCase*>& case_list,
+ int (TestCase::*method)() const) {
+ int sum = 0;
+ for (size_t i = 0; i < case_list.size(); i++) {
+ sum += (case_list[i]->*method)();
+ }
+ return sum;
+}
+
+// Returns true iff the test case passed.
+static bool TestCasePassed(const TestCase* test_case) {
+ return test_case->should_run() && test_case->Passed();
+}
+
+// Returns true iff the test case failed.
+static bool TestCaseFailed(const TestCase* test_case) {
+ return test_case->should_run() && test_case->Failed();
+}
+
+// Returns true iff test_case contains at least one test that should
+// run.
+static bool ShouldRunTestCase(const TestCase* test_case) {
+ return test_case->should_run();
+}
+
+// AssertHelper constructor.
+AssertHelper::AssertHelper(TestPartResult::Type type,
+ const char* file,
+ int line,
+ const char* message)
+ : data_(new AssertHelperData(type, file, line, message)) {
+}
+
+AssertHelper::~AssertHelper() {
+ delete data_;
+}
+
+// Message assignment, for assertion streaming support.
+void AssertHelper::operator=(const Message& message) const {
+ UnitTest::GetInstance()->
+ AddTestPartResult(data_->type, data_->file, data_->line,
+ AppendUserMessage(data_->message, message),
+ UnitTest::GetInstance()->impl()
+ ->CurrentOsStackTraceExceptTop(1)
+ // Skips the stack frame for this function itself.
+ ); // NOLINT
+}
+
+// Mutex for linked pointers.
+GTEST_DEFINE_STATIC_MUTEX_(g_linked_ptr_mutex);
+
+// Application pathname gotten in InitGoogleTest.
+String g_executable_path;
+
+// Returns the current application's name, removing directory path if that
+// is present.
+FilePath GetCurrentExecutableName() {
+ FilePath result;
+
+#if GTEST_OS_WINDOWS
+ result.Set(FilePath(g_executable_path).RemoveExtension("exe"));
+#else
+ result.Set(FilePath(g_executable_path));
+#endif // GTEST_OS_WINDOWS
+
+ return result.RemoveDirectoryName();
+}
+
+// Functions for processing the gtest_output flag.
+
+// Returns the output format, or "" for normal printed output.
+String UnitTestOptions::GetOutputFormat() {
+ const char* const gtest_output_flag = GTEST_FLAG(output).c_str();
+ if (gtest_output_flag == NULL) return String("");
+
+ const char* const colon = strchr(gtest_output_flag, ':');
+ return (colon == NULL) ?
+ String(gtest_output_flag) :
+ String(gtest_output_flag, colon - gtest_output_flag);
+}
+
+// Returns the name of the requested output file, or the default if none
+// was explicitly specified.
+String UnitTestOptions::GetAbsolutePathToOutputFile() {
+ const char* const gtest_output_flag = GTEST_FLAG(output).c_str();
+ if (gtest_output_flag == NULL)
+ return String("");
+
+ const char* const colon = strchr(gtest_output_flag, ':');
+ if (colon == NULL)
+ return String(internal::FilePath::ConcatPaths(
+ internal::FilePath(
+ UnitTest::GetInstance()->original_working_dir()),
+ internal::FilePath(kDefaultOutputFile)).ToString() );
+
+ internal::FilePath output_name(colon + 1);
+ if (!output_name.IsAbsolutePath())
+ // TODO(wan@google.com): on Windows \some\path is not an absolute
+ // path (as its meaning depends on the current drive), yet the
+ // following logic for turning it into an absolute path is wrong.
+ // Fix it.
+ output_name = internal::FilePath::ConcatPaths(
+ internal::FilePath(UnitTest::GetInstance()->original_working_dir()),
+ internal::FilePath(colon + 1));
+
+ if (!output_name.IsDirectory())
+ return output_name.ToString();
+
+ internal::FilePath result(internal::FilePath::GenerateUniqueFileName(
+ output_name, internal::GetCurrentExecutableName(),
+ GetOutputFormat().c_str()));
+ return result.ToString();
+}
+
+// Returns true iff the wildcard pattern matches the string. The
+// first ':' or '\0' character in pattern marks the end of it.
+//
+// This recursive algorithm isn't very efficient, but is clear and
+// works well enough for matching test names, which are short.
+bool UnitTestOptions::PatternMatchesString(const char *pattern,
+ const char *str) {
+ switch (*pattern) {
+ case '\0':
+ case ':': // Either ':' or '\0' marks the end of the pattern.
+ return *str == '\0';
+ case '?': // Matches any single character.
+ return *str != '\0' && PatternMatchesString(pattern + 1, str + 1);
+ case '*': // Matches any string (possibly empty) of characters.
+ return (*str != '\0' && PatternMatchesString(pattern, str + 1)) ||
+ PatternMatchesString(pattern + 1, str);
+ default: // Non-special character. Matches itself.
+ return *pattern == *str &&
+ PatternMatchesString(pattern + 1, str + 1);
+ }
+}
+
+bool UnitTestOptions::MatchesFilter(const String& name, const char* filter) {
+ const char *cur_pattern = filter;
+ for (;;) {
+ if (PatternMatchesString(cur_pattern, name.c_str())) {
+ return true;
+ }
+
+ // Finds the next pattern in the filter.
+ cur_pattern = strchr(cur_pattern, ':');
+
+ // Returns if no more pattern can be found.
+ if (cur_pattern == NULL) {
+ return false;
+ }
+
+ // Skips the pattern separater (the ':' character).
+ cur_pattern++;
+ }
+}
+
+// TODO(keithray): move String function implementations to gtest-string.cc.
+
+// Returns true iff the user-specified filter matches the test case
+// name and the test name.
+bool UnitTestOptions::FilterMatchesTest(const String &test_case_name,
+ const String &test_name) {
+ const String& full_name = String::Format("%s.%s",
+ test_case_name.c_str(),
+ test_name.c_str());
+
+ // Split --gtest_filter at '-', if there is one, to separate into
+ // positive filter and negative filter portions
+ const char* const p = GTEST_FLAG(filter).c_str();
+ const char* const dash = strchr(p, '-');
+ String positive;
+ String negative;
+ if (dash == NULL) {
+ positive = GTEST_FLAG(filter).c_str(); // Whole string is a positive filter
+ negative = String("");
+ } else {
+ positive = String(p, dash - p); // Everything up to the dash
+ negative = String(dash+1); // Everything after the dash
+ if (positive.empty()) {
+ // Treat '-test1' as the same as '*-test1'
+ positive = kUniversalFilter;
+ }
+ }
+
+ // A filter is a colon-separated list of patterns. It matches a
+ // test if any pattern in it matches the test.
+ return (MatchesFilter(full_name, positive.c_str()) &&
+ !MatchesFilter(full_name, negative.c_str()));
+}
+
+#if GTEST_HAS_SEH
+// Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the
+// given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise.
+// This function is useful as an __except condition.
+int UnitTestOptions::GTestShouldProcessSEH(DWORD exception_code) {
+ // Google Test should handle a SEH exception if:
+ // 1. the user wants it to, AND
+ // 2. this is not a breakpoint exception, AND
+ // 3. this is not a C++ exception (VC++ implements them via SEH,
+ // apparently).
+ //
+ // SEH exception code for C++ exceptions.
+ // (see http://support.microsoft.com/kb/185294 for more information).
+ const DWORD kCxxExceptionCode = 0xe06d7363;
+
+ bool should_handle = true;
+
+ if (!GTEST_FLAG(catch_exceptions))
+ should_handle = false;
+ else if (exception_code == EXCEPTION_BREAKPOINT)
+ should_handle = false;
+ else if (exception_code == kCxxExceptionCode)
+ should_handle = false;
+
+ return should_handle ? EXCEPTION_EXECUTE_HANDLER : EXCEPTION_CONTINUE_SEARCH;
+}
+#endif // GTEST_HAS_SEH
+
+} // namespace internal
+
+// The c'tor sets this object as the test part result reporter used by
+// Google Test. The 'result' parameter specifies where to report the
+// results. Intercepts only failures from the current thread.
+ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter(
+ TestPartResultArray* result)
+ : intercept_mode_(INTERCEPT_ONLY_CURRENT_THREAD),
+ result_(result) {
+ Init();
+}
+
+// The c'tor sets this object as the test part result reporter used by
+// Google Test. The 'result' parameter specifies where to report the
+// results.
+ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter(
+ InterceptMode intercept_mode, TestPartResultArray* result)
+ : intercept_mode_(intercept_mode),
+ result_(result) {
+ Init();
+}
+
+void ScopedFakeTestPartResultReporter::Init() {
+ internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+ if (intercept_mode_ == INTERCEPT_ALL_THREADS) {
+ old_reporter_ = impl->GetGlobalTestPartResultReporter();
+ impl->SetGlobalTestPartResultReporter(this);
+ } else {
+ old_reporter_ = impl->GetTestPartResultReporterForCurrentThread();
+ impl->SetTestPartResultReporterForCurrentThread(this);
+ }
+}
+
+// The d'tor restores the test part result reporter used by Google Test
+// before.
+ScopedFakeTestPartResultReporter::~ScopedFakeTestPartResultReporter() {
+ internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+ if (intercept_mode_ == INTERCEPT_ALL_THREADS) {
+ impl->SetGlobalTestPartResultReporter(old_reporter_);
+ } else {
+ impl->SetTestPartResultReporterForCurrentThread(old_reporter_);
+ }
+}
+
+// Increments the test part result count and remembers the result.
+// This method is from the TestPartResultReporterInterface interface.
+void ScopedFakeTestPartResultReporter::ReportTestPartResult(
+ const TestPartResult& result) {
+ result_->Append(result);
+}
+
+namespace internal {
+
+// Returns the type ID of ::testing::Test. We should always call this
+// instead of GetTypeId< ::testing::Test>() to get the type ID of
+// testing::Test. This is to work around a suspected linker bug when
+// using Google Test as a framework on Mac OS X. The bug causes
+// GetTypeId< ::testing::Test>() to return different values depending
+// on whether the call is from the Google Test framework itself or
+// from user test code. GetTestTypeId() is guaranteed to always
+// return the same value, as it always calls GetTypeId<>() from the
+// gtest.cc, which is within the Google Test framework.
+TypeId GetTestTypeId() {
+ return GetTypeId<Test>();
+}
+
+// The value of GetTestTypeId() as seen from within the Google Test
+// library. This is solely for testing GetTestTypeId().
+extern const TypeId kTestTypeIdInGoogleTest = GetTestTypeId();
+
+// This predicate-formatter checks that 'results' contains a test part
+// failure of the given type and that the failure message contains the
+// given substring.
+AssertionResult HasOneFailure(const char* /* results_expr */,
+ const char* /* type_expr */,
+ const char* /* substr_expr */,
+ const TestPartResultArray& results,
+ TestPartResult::Type type,
+ const string& substr) {
+ const String expected(type == TestPartResult::kFatalFailure ?
+ "1 fatal failure" :
+ "1 non-fatal failure");
+ Message msg;
+ if (results.size() != 1) {
+ msg << "Expected: " << expected << "\n"
+ << " Actual: " << results.size() << " failures";
+ for (int i = 0; i < results.size(); i++) {
+ msg << "\n" << results.GetTestPartResult(i);
+ }
+ return AssertionFailure() << msg;
+ }
+
+ const TestPartResult& r = results.GetTestPartResult(0);
+ if (r.type() != type) {
+ return AssertionFailure() << "Expected: " << expected << "\n"
+ << " Actual:\n"
+ << r;
+ }
+
+ if (strstr(r.message(), substr.c_str()) == NULL) {
+ return AssertionFailure() << "Expected: " << expected << " containing \""
+ << substr << "\"\n"
+ << " Actual:\n"
+ << r;
+ }
+
+ return AssertionSuccess();
+}
+
+// The constructor of SingleFailureChecker remembers where to look up
+// test part results, what type of failure we expect, and what
+// substring the failure message should contain.
+SingleFailureChecker:: SingleFailureChecker(
+ const TestPartResultArray* results,
+ TestPartResult::Type type,
+ const string& substr)
+ : results_(results),
+ type_(type),
+ substr_(substr) {}
+
+// The destructor of SingleFailureChecker verifies that the given
+// TestPartResultArray contains exactly one failure that has the given
+// type and contains the given substring. If that's not the case, a
+// non-fatal failure will be generated.
+SingleFailureChecker::~SingleFailureChecker() {
+ EXPECT_PRED_FORMAT3(HasOneFailure, *results_, type_, substr_);
+}
+
+DefaultGlobalTestPartResultReporter::DefaultGlobalTestPartResultReporter(
+ UnitTestImpl* unit_test) : unit_test_(unit_test) {}
+
+void DefaultGlobalTestPartResultReporter::ReportTestPartResult(
+ const TestPartResult& result) {
+ unit_test_->current_test_result()->AddTestPartResult(result);
+ unit_test_->listeners()->repeater()->OnTestPartResult(result);
+}
+
+DefaultPerThreadTestPartResultReporter::DefaultPerThreadTestPartResultReporter(
+ UnitTestImpl* unit_test) : unit_test_(unit_test) {}
+
+void DefaultPerThreadTestPartResultReporter::ReportTestPartResult(
+ const TestPartResult& result) {
+ unit_test_->GetGlobalTestPartResultReporter()->ReportTestPartResult(result);
+}
+
+// Returns the global test part result reporter.
+TestPartResultReporterInterface*
+UnitTestImpl::GetGlobalTestPartResultReporter() {
+ internal::MutexLock lock(&global_test_part_result_reporter_mutex_);
+ return global_test_part_result_repoter_;
+}
+
+// Sets the global test part result reporter.
+void UnitTestImpl::SetGlobalTestPartResultReporter(
+ TestPartResultReporterInterface* reporter) {
+ internal::MutexLock lock(&global_test_part_result_reporter_mutex_);
+ global_test_part_result_repoter_ = reporter;
+}
+
+// Returns the test part result reporter for the current thread.
+TestPartResultReporterInterface*
+UnitTestImpl::GetTestPartResultReporterForCurrentThread() {
+ return per_thread_test_part_result_reporter_.get();
+}
+
+// Sets the test part result reporter for the current thread.
+void UnitTestImpl::SetTestPartResultReporterForCurrentThread(
+ TestPartResultReporterInterface* reporter) {
+ per_thread_test_part_result_reporter_.set(reporter);
+}
+
+// Gets the number of successful test cases.
+int UnitTestImpl::successful_test_case_count() const {
+ return CountIf(test_cases_, TestCasePassed);
+}
+
+// Gets the number of failed test cases.
+int UnitTestImpl::failed_test_case_count() const {
+ return CountIf(test_cases_, TestCaseFailed);
+}
+
+// Gets the number of all test cases.
+int UnitTestImpl::total_test_case_count() const {
+ return static_cast<int>(test_cases_.size());
+}
+
+// Gets the number of all test cases that contain at least one test
+// that should run.
+int UnitTestImpl::test_case_to_run_count() const {
+ return CountIf(test_cases_, ShouldRunTestCase);
+}
+
+// Gets the number of successful tests.
+int UnitTestImpl::successful_test_count() const {
+ return SumOverTestCaseList(test_cases_, &TestCase::successful_test_count);
+}
+
+// Gets the number of failed tests.
+int UnitTestImpl::failed_test_count() const {
+ return SumOverTestCaseList(test_cases_, &TestCase::failed_test_count);
+}
+
+// Gets the number of disabled tests.
+int UnitTestImpl::disabled_test_count() const {
+ return SumOverTestCaseList(test_cases_, &TestCase::disabled_test_count);
+}
+
+// Gets the number of all tests.
+int UnitTestImpl::total_test_count() const {
+ return SumOverTestCaseList(test_cases_, &TestCase::total_test_count);
+}
+
+// Gets the number of tests that should run.
+int UnitTestImpl::test_to_run_count() const {
+ return SumOverTestCaseList(test_cases_, &TestCase::test_to_run_count);
+}
+
+// Returns the current OS stack trace as a String.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag. The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// CurrentOsStackTraceExceptTop(1), Foo() will be included in the
+// trace but Bar() and CurrentOsStackTraceExceptTop() won't.
+String UnitTestImpl::CurrentOsStackTraceExceptTop(int skip_count) {
+ (void)skip_count;
+ return String("");
+}
+
+// Returns the current time in milliseconds.
+TimeInMillis GetTimeInMillis() {
+#if GTEST_OS_WINDOWS_MOBILE || defined(__BORLANDC__)
+ // Difference between 1970-01-01 and 1601-01-01 in milliseconds.
+ // http://analogous.blogspot.com/2005/04/epoch.html
+ const TimeInMillis kJavaEpochToWinFileTimeDelta =
+ static_cast<TimeInMillis>(116444736UL) * 100000UL;
+ const DWORD kTenthMicrosInMilliSecond = 10000;
+
+ SYSTEMTIME now_systime;
+ FILETIME now_filetime;
+ ULARGE_INTEGER now_int64;
+ // TODO(kenton@google.com): Shouldn't this just use
+ // GetSystemTimeAsFileTime()?
+ GetSystemTime(&now_systime);
+ if (SystemTimeToFileTime(&now_systime, &now_filetime)) {
+ now_int64.LowPart = now_filetime.dwLowDateTime;
+ now_int64.HighPart = now_filetime.dwHighDateTime;
+ now_int64.QuadPart = (now_int64.QuadPart / kTenthMicrosInMilliSecond) -
+ kJavaEpochToWinFileTimeDelta;
+ return now_int64.QuadPart;
+ }
+ return 0;
+#elif GTEST_OS_WINDOWS && !GTEST_HAS_GETTIMEOFDAY_
+ __timeb64 now;
+
+# ifdef _MSC_VER
+
+ // MSVC 8 deprecates _ftime64(), so we want to suppress warning 4996
+ // (deprecated function) there.
+ // TODO(kenton@google.com): Use GetTickCount()? Or use
+ // SystemTimeToFileTime()
+# pragma warning(push) // Saves the current warning state.
+# pragma warning(disable:4996) // Temporarily disables warning 4996.
+ _ftime64(&now);
+# pragma warning(pop) // Restores the warning state.
+# else
+
+ _ftime64(&now);
+
+# endif // _MSC_VER
+
+ return static_cast<TimeInMillis>(now.time) * 1000 + now.millitm;
+#elif GTEST_HAS_GETTIMEOFDAY_
+ struct timeval now;
+ gettimeofday(&now, NULL);
+ return static_cast<TimeInMillis>(now.tv_sec) * 1000 + now.tv_usec / 1000;
+#else
+# error "Don't know how to get the current time on your system."
+#endif
+}
+
+// Utilities
+
+// class String
+
+// Returns the input enclosed in double quotes if it's not NULL;
+// otherwise returns "(null)". For example, "\"Hello\"" is returned
+// for input "Hello".
+//
+// This is useful for printing a C string in the syntax of a literal.
+//
+// Known issue: escape sequences are not handled yet.
+String String::ShowCStringQuoted(const char* c_str) {
+ return c_str ? String::Format("\"%s\"", c_str) : String("(null)");
+}
+
+// Copies at most length characters from str into a newly-allocated
+// piece of memory of size length+1. The memory is allocated with new[].
+// A terminating null byte is written to the memory, and a pointer to it
+// is returned. If str is NULL, NULL is returned.
+static char* CloneString(const char* str, size_t length) {
+ if (str == NULL) {
+ return NULL;
+ } else {
+ char* const clone = new char[length + 1];
+ posix::StrNCpy(clone, str, length);
+ clone[length] = '\0';
+ return clone;
+ }
+}
+
+// Clones a 0-terminated C string, allocating memory using new. The
+// caller is responsible for deleting[] the return value. Returns the
+// cloned string, or NULL if the input is NULL.
+const char * String::CloneCString(const char* c_str) {
+ return (c_str == NULL) ?
+ NULL : CloneString(c_str, strlen(c_str));
+}
+
+#if GTEST_OS_WINDOWS_MOBILE
+// Creates a UTF-16 wide string from the given ANSI string, allocating
+// memory using new. The caller is responsible for deleting the return
+// value using delete[]. Returns the wide string, or NULL if the
+// input is NULL.
+LPCWSTR String::AnsiToUtf16(const char* ansi) {
+ if (!ansi) return NULL;
+ const int length = strlen(ansi);
+ const int unicode_length =
+ MultiByteToWideChar(CP_ACP, 0, ansi, length,
+ NULL, 0);
+ WCHAR* unicode = new WCHAR[unicode_length + 1];
+ MultiByteToWideChar(CP_ACP, 0, ansi, length,
+ unicode, unicode_length);
+ unicode[unicode_length] = 0;
+ return unicode;
+}
+
+// Creates an ANSI string from the given wide string, allocating
+// memory using new. The caller is responsible for deleting the return
+// value using delete[]. Returns the ANSI string, or NULL if the
+// input is NULL.
+const char* String::Utf16ToAnsi(LPCWSTR utf16_str) {
+ if (!utf16_str) return NULL;
+ const int ansi_length =
+ WideCharToMultiByte(CP_ACP, 0, utf16_str, -1,
+ NULL, 0, NULL, NULL);
+ char* ansi = new char[ansi_length + 1];
+ WideCharToMultiByte(CP_ACP, 0, utf16_str, -1,
+ ansi, ansi_length, NULL, NULL);
+ ansi[ansi_length] = 0;
+ return ansi;
+}
+
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+// Compares two C strings. Returns true iff they have the same content.
+//
+// Unlike strcmp(), this function can handle NULL argument(s). A NULL
+// C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::CStringEquals(const char * lhs, const char * rhs) {
+ if ( lhs == NULL ) return rhs == NULL;
+
+ if ( rhs == NULL ) return false;
+
+ return strcmp(lhs, rhs) == 0;
+}
+
+#if GTEST_HAS_STD_WSTRING || GTEST_HAS_GLOBAL_WSTRING
+
+// Converts an array of wide chars to a narrow string using the UTF-8
+// encoding, and streams the result to the given Message object.
+static void StreamWideCharsToMessage(const wchar_t* wstr, size_t length,
+ Message* msg) {
+ // TODO(wan): consider allowing a testing::String object to
+ // contain '\0'. This will make it behave more like std::string,
+ // and will allow ToUtf8String() to return the correct encoding
+ // for '\0' s.t. we can get rid of the conditional here (and in
+ // several other places).
+ for (size_t i = 0; i != length; ) { // NOLINT
+ if (wstr[i] != L'\0') {
+ *msg << WideStringToUtf8(wstr + i, static_cast<int>(length - i));
+ while (i != length && wstr[i] != L'\0')
+ i++;
+ } else {
+ *msg << '\0';
+ i++;
+ }
+ }
+}
+
+#endif // GTEST_HAS_STD_WSTRING || GTEST_HAS_GLOBAL_WSTRING
+
+} // namespace internal
+
+#if GTEST_HAS_STD_WSTRING
+// Converts the given wide string to a narrow string using the UTF-8
+// encoding, and streams the result to this Message object.
+Message& Message::operator <<(const ::std::wstring& wstr) {
+ internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this);
+ return *this;
+}
+#endif // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+// Converts the given wide string to a narrow string using the UTF-8
+// encoding, and streams the result to this Message object.
+Message& Message::operator <<(const ::wstring& wstr) {
+ internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this);
+ return *this;
+}
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+// AssertionResult constructors.
+// Used in EXPECT_TRUE/FALSE(assertion_result).
+AssertionResult::AssertionResult(const AssertionResult& other)
+ : success_(other.success_),
+ message_(other.message_.get() != NULL ?
+ new ::std::string(*other.message_) :
+ static_cast< ::std::string*>(NULL)) {
+}
+
+// Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE.
+AssertionResult AssertionResult::operator!() const {
+ AssertionResult negation(!success_);
+ if (message_.get() != NULL)
+ negation << *message_;
+ return negation;
+}
+
+// Makes a successful assertion result.
+AssertionResult AssertionSuccess() {
+ return AssertionResult(true);
+}
+
+// Makes a failed assertion result.
+AssertionResult AssertionFailure() {
+ return AssertionResult(false);
+}
+
+// Makes a failed assertion result with the given failure message.
+// Deprecated; use AssertionFailure() << message.
+AssertionResult AssertionFailure(const Message& message) {
+ return AssertionFailure() << message;
+}
+
+namespace internal {
+
+// Constructs and returns the message for an equality assertion
+// (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure.
+//
+// The first four parameters are the expressions used in the assertion
+// and their values, as strings. For example, for ASSERT_EQ(foo, bar)
+// where foo is 5 and bar is 6, we have:
+//
+// expected_expression: "foo"
+// actual_expression: "bar"
+// expected_value: "5"
+// actual_value: "6"
+//
+// The ignoring_case parameter is true iff the assertion is a
+// *_STRCASEEQ*. When it's true, the string " (ignoring case)" will
+// be inserted into the message.
+AssertionResult EqFailure(const char* expected_expression,
+ const char* actual_expression,
+ const String& expected_value,
+ const String& actual_value,
+ bool ignoring_case) {
+ Message msg;
+ msg << "Value of: " << actual_expression;
+ if (actual_value != actual_expression) {
+ msg << "\n Actual: " << actual_value;
+ }
+
+ msg << "\nExpected: " << expected_expression;
+ if (ignoring_case) {
+ msg << " (ignoring case)";
+ }
+ if (expected_value != expected_expression) {
+ msg << "\nWhich is: " << expected_value;
+ }
+
+ return AssertionFailure() << msg;
+}
+
+// Constructs a failure message for Boolean assertions such as EXPECT_TRUE.
+String GetBoolAssertionFailureMessage(const AssertionResult& assertion_result,
+ const char* expression_text,
+ const char* actual_predicate_value,
+ const char* expected_predicate_value) {
+ const char* actual_message = assertion_result.message();
+ Message msg;
+ msg << "Value of: " << expression_text
+ << "\n Actual: " << actual_predicate_value;
+ if (actual_message[0] != '\0')
+ msg << " (" << actual_message << ")";
+ msg << "\nExpected: " << expected_predicate_value;
+ return msg.GetString();
+}
+
+// Helper function for implementing ASSERT_NEAR.
+AssertionResult DoubleNearPredFormat(const char* expr1,
+ const char* expr2,
+ const char* abs_error_expr,
+ double val1,
+ double val2,
+ double abs_error) {
+ const double diff = fabs(val1 - val2);
+ if (diff <= abs_error) return AssertionSuccess();
+
+ // TODO(wan): do not print the value of an expression if it's
+ // already a literal.
+ return AssertionFailure()
+ << "The difference between " << expr1 << " and " << expr2
+ << " is " << diff << ", which exceeds " << abs_error_expr << ", where\n"
+ << expr1 << " evaluates to " << val1 << ",\n"
+ << expr2 << " evaluates to " << val2 << ", and\n"
+ << abs_error_expr << " evaluates to " << abs_error << ".";
+}
+
+
+// Helper template for implementing FloatLE() and DoubleLE().
+template <typename RawType>
+AssertionResult FloatingPointLE(const char* expr1,
+ const char* expr2,
+ RawType val1,
+ RawType val2) {
+ // Returns success if val1 is less than val2,
+ if (val1 < val2) {
+ return AssertionSuccess();
+ }
+
+ // or if val1 is almost equal to val2.
+ const FloatingPoint<RawType> lhs(val1), rhs(val2);
+ if (lhs.AlmostEquals(rhs)) {
+ return AssertionSuccess();
+ }
+
+ // Note that the above two checks will both fail if either val1 or
+ // val2 is NaN, as the IEEE floating-point standard requires that
+ // any predicate involving a NaN must return false.
+
+ ::std::stringstream val1_ss;
+ val1_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+ << val1;
+
+ ::std::stringstream val2_ss;
+ val2_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+ << val2;
+
+ return AssertionFailure()
+ << "Expected: (" << expr1 << ") <= (" << expr2 << ")\n"
+ << " Actual: " << StringStreamToString(&val1_ss) << " vs "
+ << StringStreamToString(&val2_ss);
+}
+
+} // namespace internal
+
+// Asserts that val1 is less than, or almost equal to, val2. Fails
+// otherwise. In particular, it fails if either val1 or val2 is NaN.
+AssertionResult FloatLE(const char* expr1, const char* expr2,
+ float val1, float val2) {
+ return internal::FloatingPointLE<float>(expr1, expr2, val1, val2);
+}
+
+// Asserts that val1 is less than, or almost equal to, val2. Fails
+// otherwise. In particular, it fails if either val1 or val2 is NaN.
+AssertionResult DoubleLE(const char* expr1, const char* expr2,
+ double val1, double val2) {
+ return internal::FloatingPointLE<double>(expr1, expr2, val1, val2);
+}
+
+namespace internal {
+
+// The helper function for {ASSERT|EXPECT}_EQ with int or enum
+// arguments.
+AssertionResult CmpHelperEQ(const char* expected_expression,
+ const char* actual_expression,
+ BiggestInt expected,
+ BiggestInt actual) {
+ if (expected == actual) {
+ return AssertionSuccess();
+ }
+
+ return EqFailure(expected_expression,
+ actual_expression,
+ FormatForComparisonFailureMessage(expected, actual),
+ FormatForComparisonFailureMessage(actual, expected),
+ false);
+}
+
+// A macro for implementing the helper functions needed to implement
+// ASSERT_?? and EXPECT_?? with integer or enum arguments. It is here
+// just to avoid copy-and-paste of similar code.
+#define GTEST_IMPL_CMP_HELPER_(op_name, op)\
+AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \
+ BiggestInt val1, BiggestInt val2) {\
+ if (val1 op val2) {\
+ return AssertionSuccess();\
+ } else {\
+ return AssertionFailure() \
+ << "Expected: (" << expr1 << ") " #op " (" << expr2\
+ << "), actual: " << FormatForComparisonFailureMessage(val1, val2)\
+ << " vs " << FormatForComparisonFailureMessage(val2, val1);\
+ }\
+}
+
+// Implements the helper function for {ASSERT|EXPECT}_NE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(NE, !=)
+// Implements the helper function for {ASSERT|EXPECT}_LE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(LE, <=)
+// Implements the helper function for {ASSERT|EXPECT}_LT with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(LT, < )
+// Implements the helper function for {ASSERT|EXPECT}_GE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(GE, >=)
+// Implements the helper function for {ASSERT|EXPECT}_GT with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(GT, > )
+
+#undef GTEST_IMPL_CMP_HELPER_
+
+// The helper function for {ASSERT|EXPECT}_STREQ.
+AssertionResult CmpHelperSTREQ(const char* expected_expression,
+ const char* actual_expression,
+ const char* expected,
+ const char* actual) {
+ if (String::CStringEquals(expected, actual)) {
+ return AssertionSuccess();
+ }
+
+ return EqFailure(expected_expression,
+ actual_expression,
+ String::ShowCStringQuoted(expected),
+ String::ShowCStringQuoted(actual),
+ false);
+}
+
+// The helper function for {ASSERT|EXPECT}_STRCASEEQ.
+AssertionResult CmpHelperSTRCASEEQ(const char* expected_expression,
+ const char* actual_expression,
+ const char* expected,
+ const char* actual) {
+ if (String::CaseInsensitiveCStringEquals(expected, actual)) {
+ return AssertionSuccess();
+ }
+
+ return EqFailure(expected_expression,
+ actual_expression,
+ String::ShowCStringQuoted(expected),
+ String::ShowCStringQuoted(actual),
+ true);
+}
+
+// The helper function for {ASSERT|EXPECT}_STRNE.
+AssertionResult CmpHelperSTRNE(const char* s1_expression,
+ const char* s2_expression,
+ const char* s1,
+ const char* s2) {
+ if (!String::CStringEquals(s1, s2)) {
+ return AssertionSuccess();
+ } else {
+ return AssertionFailure() << "Expected: (" << s1_expression << ") != ("
+ << s2_expression << "), actual: \""
+ << s1 << "\" vs \"" << s2 << "\"";
+ }
+}
+
+// The helper function for {ASSERT|EXPECT}_STRCASENE.
+AssertionResult CmpHelperSTRCASENE(const char* s1_expression,
+ const char* s2_expression,
+ const char* s1,
+ const char* s2) {
+ if (!String::CaseInsensitiveCStringEquals(s1, s2)) {
+ return AssertionSuccess();
+ } else {
+ return AssertionFailure()
+ << "Expected: (" << s1_expression << ") != ("
+ << s2_expression << ") (ignoring case), actual: \""
+ << s1 << "\" vs \"" << s2 << "\"";
+ }
+}
+
+} // namespace internal
+
+namespace {
+
+// Helper functions for implementing IsSubString() and IsNotSubstring().
+
+// This group of overloaded functions return true iff needle is a
+// substring of haystack. NULL is considered a substring of itself
+// only.
+
+bool IsSubstringPred(const char* needle, const char* haystack) {
+ if (needle == NULL || haystack == NULL)
+ return needle == haystack;
+
+ return strstr(haystack, needle) != NULL;
+}
+
+bool IsSubstringPred(const wchar_t* needle, const wchar_t* haystack) {
+ if (needle == NULL || haystack == NULL)
+ return needle == haystack;
+
+ return wcsstr(haystack, needle) != NULL;
+}
+
+// StringType here can be either ::std::string or ::std::wstring.
+template <typename StringType>
+bool IsSubstringPred(const StringType& needle,
+ const StringType& haystack) {
+ return haystack.find(needle) != StringType::npos;
+}
+
+// This function implements either IsSubstring() or IsNotSubstring(),
+// depending on the value of the expected_to_be_substring parameter.
+// StringType here can be const char*, const wchar_t*, ::std::string,
+// or ::std::wstring.
+template <typename StringType>
+AssertionResult IsSubstringImpl(
+ bool expected_to_be_substring,
+ const char* needle_expr, const char* haystack_expr,
+ const StringType& needle, const StringType& haystack) {
+ if (IsSubstringPred(needle, haystack) == expected_to_be_substring)
+ return AssertionSuccess();
+
+ const bool is_wide_string = sizeof(needle[0]) > 1;
+ const char* const begin_string_quote = is_wide_string ? "L\"" : "\"";
+ return AssertionFailure()
+ << "Value of: " << needle_expr << "\n"
+ << " Actual: " << begin_string_quote << needle << "\"\n"
+ << "Expected: " << (expected_to_be_substring ? "" : "not ")
+ << "a substring of " << haystack_expr << "\n"
+ << "Which is: " << begin_string_quote << haystack << "\"";
+}
+
+} // namespace
+
+// IsSubstring() and IsNotSubstring() check whether needle is a
+// substring of haystack (NULL is considered a substring of itself
+// only), and return an appropriate error message when they fail.
+
+AssertionResult IsSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const char* needle, const char* haystack) {
+ return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const wchar_t* needle, const wchar_t* haystack) {
+ return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const char* needle, const char* haystack) {
+ return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const wchar_t* needle, const wchar_t* haystack) {
+ return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const ::std::string& needle, const ::std::string& haystack) {
+ return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const ::std::string& needle, const ::std::string& haystack) {
+ return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+#if GTEST_HAS_STD_WSTRING
+AssertionResult IsSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const ::std::wstring& needle, const ::std::wstring& haystack) {
+ return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+ const char* needle_expr, const char* haystack_expr,
+ const ::std::wstring& needle, const ::std::wstring& haystack) {
+ return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+#endif // GTEST_HAS_STD_WSTRING
+
+namespace internal {
+
+#if GTEST_OS_WINDOWS
+
+namespace {
+
+// Helper function for IsHRESULT{SuccessFailure} predicates
+AssertionResult HRESULTFailureHelper(const char* expr,
+ const char* expected,
+ long hr) { // NOLINT
+# if GTEST_OS_WINDOWS_MOBILE
+
+ // Windows CE doesn't support FormatMessage.
+ const char error_text[] = "";
+
+# else
+
+ // Looks up the human-readable system message for the HRESULT code
+ // and since we're not passing any params to FormatMessage, we don't
+ // want inserts expanded.
+ const DWORD kFlags = FORMAT_MESSAGE_FROM_SYSTEM |
+ FORMAT_MESSAGE_IGNORE_INSERTS;
+ const DWORD kBufSize = 4096; // String::Format can't exceed this length.
+ // Gets the system's human readable message string for this HRESULT.
+ char error_text[kBufSize] = { '\0' };
+ DWORD message_length = ::FormatMessageA(kFlags,
+ 0, // no source, we're asking system
+ hr, // the error
+ 0, // no line width restrictions
+ error_text, // output buffer
+ kBufSize, // buf size
+ NULL); // no arguments for inserts
+ // Trims tailing white space (FormatMessage leaves a trailing cr-lf)
+ for (; message_length && IsSpace(error_text[message_length - 1]);
+ --message_length) {
+ error_text[message_length - 1] = '\0';
+ }
+
+# endif // GTEST_OS_WINDOWS_MOBILE
+
+ const String error_hex(String::Format("0x%08X ", hr));
+ return ::testing::AssertionFailure()
+ << "Expected: " << expr << " " << expected << ".\n"
+ << " Actual: " << error_hex << error_text << "\n";
+}
+
+} // namespace
+
+AssertionResult IsHRESULTSuccess(const char* expr, long hr) { // NOLINT
+ if (SUCCEEDED(hr)) {
+ return AssertionSuccess();
+ }
+ return HRESULTFailureHelper(expr, "succeeds", hr);
+}
+
+AssertionResult IsHRESULTFailure(const char* expr, long hr) { // NOLINT
+ if (FAILED(hr)) {
+ return AssertionSuccess();
+ }
+ return HRESULTFailureHelper(expr, "fails", hr);
+}
+
+#endif // GTEST_OS_WINDOWS
+
+// Utility functions for encoding Unicode text (wide strings) in
+// UTF-8.
+
+// A Unicode code-point can have upto 21 bits, and is encoded in UTF-8
+// like this:
+//
+// Code-point length Encoding
+// 0 - 7 bits 0xxxxxxx
+// 8 - 11 bits 110xxxxx 10xxxxxx
+// 12 - 16 bits 1110xxxx 10xxxxxx 10xxxxxx
+// 17 - 21 bits 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
+
+// The maximum code-point a one-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint1 = (static_cast<UInt32>(1) << 7) - 1;
+
+// The maximum code-point a two-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint2 = (static_cast<UInt32>(1) << (5 + 6)) - 1;
+
+// The maximum code-point a three-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint3 = (static_cast<UInt32>(1) << (4 + 2*6)) - 1;
+
+// The maximum code-point a four-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint4 = (static_cast<UInt32>(1) << (3 + 3*6)) - 1;
+
+// Chops off the n lowest bits from a bit pattern. Returns the n
+// lowest bits. As a side effect, the original bit pattern will be
+// shifted to the right by n bits.
+inline UInt32 ChopLowBits(UInt32* bits, int n) {
+ const UInt32 low_bits = *bits & ((static_cast<UInt32>(1) << n) - 1);
+ *bits >>= n;
+ return low_bits;
+}
+
+// Converts a Unicode code point to a narrow string in UTF-8 encoding.
+// code_point parameter is of type UInt32 because wchar_t may not be
+// wide enough to contain a code point.
+// The output buffer str must containt at least 32 characters.
+// The function returns the address of the output buffer.
+// If the code_point is not a valid Unicode code point
+// (i.e. outside of Unicode range U+0 to U+10FFFF) it will be output
+// as '(Invalid Unicode 0xXXXXXXXX)'.
+char* CodePointToUtf8(UInt32 code_point, char* str) {
+ if (code_point <= kMaxCodePoint1) {
+ str[1] = '\0';
+ str[0] = static_cast<char>(code_point); // 0xxxxxxx
+ } else if (code_point <= kMaxCodePoint2) {
+ str[2] = '\0';
+ str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx
+ str[0] = static_cast<char>(0xC0 | code_point); // 110xxxxx
+ } else if (code_point <= kMaxCodePoint3) {
+ str[3] = '\0';
+ str[2] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx
+ str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx
+ str[0] = static_cast<char>(0xE0 | code_point); // 1110xxxx
+ } else if (code_point <= kMaxCodePoint4) {
+ str[4] = '\0';
+ str[3] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx
+ str[2] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx
+ str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx
+ str[0] = static_cast<char>(0xF0 | code_point); // 11110xxx
+ } else {
+ // The longest string String::Format can produce when invoked
+ // with these parameters is 28 character long (not including
+ // the terminating nul character). We are asking for 32 character
+ // buffer just in case. This is also enough for strncpy to
+ // null-terminate the destination string.
+ posix::StrNCpy(
+ str, String::Format("(Invalid Unicode 0x%X)", code_point).c_str(), 32);
+ str[31] = '\0'; // Makes sure no change in the format to strncpy leaves
+ // the result unterminated.
+ }
+ return str;
+}
+
+// The following two functions only make sense if the the system
+// uses UTF-16 for wide string encoding. All supported systems
+// with 16 bit wchar_t (Windows, Cygwin, Symbian OS) do use UTF-16.
+
+// Determines if the arguments constitute UTF-16 surrogate pair
+// and thus should be combined into a single Unicode code point
+// using CreateCodePointFromUtf16SurrogatePair.
+inline bool IsUtf16SurrogatePair(wchar_t first, wchar_t second) {
+ return sizeof(wchar_t) == 2 &&
+ (first & 0xFC00) == 0xD800 && (second & 0xFC00) == 0xDC00;
+}
+
+// Creates a Unicode code point from UTF16 surrogate pair.
+inline UInt32 CreateCodePointFromUtf16SurrogatePair(wchar_t first,
+ wchar_t second) {
+ const UInt32 mask = (1 << 10) - 1;
+ return (sizeof(wchar_t) == 2) ?
+ (((first & mask) << 10) | (second & mask)) + 0x10000 :
+ // This function should not be called when the condition is
+ // false, but we provide a sensible default in case it is.
+ static_cast<UInt32>(first);
+}
+
+// Converts a wide string to a narrow string in UTF-8 encoding.
+// The wide string is assumed to have the following encoding:
+// UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin, Symbian OS)
+// UTF-32 if sizeof(wchar_t) == 4 (on Linux)
+// Parameter str points to a null-terminated wide string.
+// Parameter num_chars may additionally limit the number
+// of wchar_t characters processed. -1 is used when the entire string
+// should be processed.
+// If the string contains code points that are not valid Unicode code points
+// (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output
+// as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding
+// and contains invalid UTF-16 surrogate pairs, values in those pairs
+// will be encoded as individual Unicode characters from Basic Normal Plane.
+String WideStringToUtf8(const wchar_t* str, int num_chars) {
+ if (num_chars == -1)
+ num_chars = static_cast<int>(wcslen(str));
+
+ ::std::stringstream stream;
+ for (int i = 0; i < num_chars; ++i) {
+ UInt32 unicode_code_point;
+
+ if (str[i] == L'\0') {
+ break;
+ } else if (i + 1 < num_chars && IsUtf16SurrogatePair(str[i], str[i + 1])) {
+ unicode_code_point = CreateCodePointFromUtf16SurrogatePair(str[i],
+ str[i + 1]);
+ i++;
+ } else {
+ unicode_code_point = static_cast<UInt32>(str[i]);
+ }
+
+ char buffer[32]; // CodePointToUtf8 requires a buffer this big.
+ stream << CodePointToUtf8(unicode_code_point, buffer);
+ }
+ return StringStreamToString(&stream);
+}
+
+// Converts a wide C string to a String using the UTF-8 encoding.
+// NULL will be converted to "(null)".
+String String::ShowWideCString(const wchar_t * wide_c_str) {
+ if (wide_c_str == NULL) return String("(null)");
+
+ return String(internal::WideStringToUtf8(wide_c_str, -1).c_str());
+}
+
+// Similar to ShowWideCString(), except that this function encloses
+// the converted string in double quotes.
+String String::ShowWideCStringQuoted(const wchar_t* wide_c_str) {
+ if (wide_c_str == NULL) return String("(null)");
+
+ return String::Format("L\"%s\"",
+ String::ShowWideCString(wide_c_str).c_str());
+}
+
+// Compares two wide C strings. Returns true iff they have the same
+// content.
+//
+// Unlike wcscmp(), this function can handle NULL argument(s). A NULL
+// C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::WideCStringEquals(const wchar_t * lhs, const wchar_t * rhs) {
+ if (lhs == NULL) return rhs == NULL;
+
+ if (rhs == NULL) return false;
+
+ return wcscmp(lhs, rhs) == 0;
+}
+
+// Helper function for *_STREQ on wide strings.
+AssertionResult CmpHelperSTREQ(const char* expected_expression,
+ const char* actual_expression,
+ const wchar_t* expected,
+ const wchar_t* actual) {
+ if (String::WideCStringEquals(expected, actual)) {
+ return AssertionSuccess();
+ }
+
+ return EqFailure(expected_expression,
+ actual_expression,
+ String::ShowWideCStringQuoted(expected),
+ String::ShowWideCStringQuoted(actual),
+ false);
+}
+
+// Helper function for *_STRNE on wide strings.
+AssertionResult CmpHelperSTRNE(const char* s1_expression,
+ const char* s2_expression,
+ const wchar_t* s1,
+ const wchar_t* s2) {
+ if (!String::WideCStringEquals(s1, s2)) {
+ return AssertionSuccess();
+ }
+
+ return AssertionFailure() << "Expected: (" << s1_expression << ") != ("
+ << s2_expression << "), actual: "
+ << String::ShowWideCStringQuoted(s1)
+ << " vs " << String::ShowWideCStringQuoted(s2);
+}
+
+// Compares two C strings, ignoring case. Returns true iff they have
+// the same content.
+//
+// Unlike strcasecmp(), this function can handle NULL argument(s). A
+// NULL C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::CaseInsensitiveCStringEquals(const char * lhs, const char * rhs) {
+ if (lhs == NULL)
+ return rhs == NULL;
+ if (rhs == NULL)
+ return false;
+ return posix::StrCaseCmp(lhs, rhs) == 0;
+}
+
+ // Compares two wide C strings, ignoring case. Returns true iff they
+ // have the same content.
+ //
+ // Unlike wcscasecmp(), this function can handle NULL argument(s).
+ // A NULL C string is considered different to any non-NULL wide C string,
+ // including the empty string.
+ // NB: The implementations on different platforms slightly differ.
+ // On windows, this method uses _wcsicmp which compares according to LC_CTYPE
+ // environment variable. On GNU platform this method uses wcscasecmp
+ // which compares according to LC_CTYPE category of the current locale.
+ // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the
+ // current locale.
+bool String::CaseInsensitiveWideCStringEquals(const wchar_t* lhs,
+ const wchar_t* rhs) {
+ if (lhs == NULL) return rhs == NULL;
+
+ if (rhs == NULL) return false;
+
+#if GTEST_OS_WINDOWS
+ return _wcsicmp(lhs, rhs) == 0;
+#elif GTEST_OS_LINUX && !GTEST_OS_LINUX_ANDROID
+ return wcscasecmp(lhs, rhs) == 0;
+#else
+ // Android, Mac OS X and Cygwin don't define wcscasecmp.
+ // Other unknown OSes may not define it either.
+ wint_t left, right;
+ do {
+ left = towlower(*lhs++);
+ right = towlower(*rhs++);
+ } while (left && left == right);
+ return left == right;
+#endif // OS selector
+}
+
+// Compares this with another String.
+// Returns < 0 if this is less than rhs, 0 if this is equal to rhs, or > 0
+// if this is greater than rhs.
+int String::Compare(const String & rhs) const {
+ const char* const lhs_c_str = c_str();
+ const char* const rhs_c_str = rhs.c_str();
+
+ if (lhs_c_str == NULL) {
+ return rhs_c_str == NULL ? 0 : -1; // NULL < anything except NULL
+ } else if (rhs_c_str == NULL) {
+ return 1;
+ }
+
+ const size_t shorter_str_len =
+ length() <= rhs.length() ? length() : rhs.length();
+ for (size_t i = 0; i != shorter_str_len; i++) {
+ if (lhs_c_str[i] < rhs_c_str[i]) {
+ return -1;
+ } else if (lhs_c_str[i] > rhs_c_str[i]) {
+ return 1;
+ }
+ }
+ return (length() < rhs.length()) ? -1 :
+ (length() > rhs.length()) ? 1 : 0;
+}
+
+// Returns true iff this String ends with the given suffix. *Any*
+// String is considered to end with a NULL or empty suffix.
+bool String::EndsWith(const char* suffix) const {
+ if (suffix == NULL || CStringEquals(suffix, "")) return true;
+
+ if (c_str() == NULL) return false;
+
+ const size_t this_len = strlen(c_str());
+ const size_t suffix_len = strlen(suffix);
+ return (this_len >= suffix_len) &&
+ CStringEquals(c_str() + this_len - suffix_len, suffix);
+}
+
+// Returns true iff this String ends with the given suffix, ignoring case.
+// Any String is considered to end with a NULL or empty suffix.
+bool String::EndsWithCaseInsensitive(const char* suffix) const {
+ if (suffix == NULL || CStringEquals(suffix, "")) return true;
+
+ if (c_str() == NULL) return false;
+
+ const size_t this_len = strlen(c_str());
+ const size_t suffix_len = strlen(suffix);
+ return (this_len >= suffix_len) &&
+ CaseInsensitiveCStringEquals(c_str() + this_len - suffix_len, suffix);
+}
+
+// Formats a list of arguments to a String, using the same format
+// spec string as for printf.
+//
+// We do not use the StringPrintf class as it is not universally
+// available.
+//
+// The result is limited to 4096 characters (including the tailing 0).
+// If 4096 characters are not enough to format the input, or if
+// there's an error, "<formatting error or buffer exceeded>" is
+// returned.
+String String::Format(const char * format, ...) {
+ va_list args;
+ va_start(args, format);
+
+ char buffer[4096];
+ const int kBufferSize = sizeof(buffer)/sizeof(buffer[0]);
+
+ // MSVC 8 deprecates vsnprintf(), so we want to suppress warning
+ // 4996 (deprecated function) there.
+#ifdef _MSC_VER // We are using MSVC.
+# pragma warning(push) // Saves the current warning state.
+# pragma warning(disable:4996) // Temporarily disables warning 4996.
+
+ const int size = vsnprintf(buffer, kBufferSize, format, args);
+
+# pragma warning(pop) // Restores the warning state.
+#else // We are not using MSVC.
+ const int size = vsnprintf(buffer, kBufferSize, format, args);
+#endif // _MSC_VER
+ va_end(args);
+
+ // vsnprintf()'s behavior is not portable. When the buffer is not
+ // big enough, it returns a negative value in MSVC, and returns the
+ // needed buffer size on Linux. When there is an output error, it
+ // always returns a negative value. For simplicity, we lump the two
+ // error cases together.
+ if (size < 0 || size >= kBufferSize) {
+ return String("<formatting error or buffer exceeded>");
+ } else {
+ return String(buffer, size);
+ }
+}
+
+// Converts the buffer in a stringstream to a String, converting NUL
+// bytes to "\\0" along the way.
+String StringStreamToString(::std::stringstream* ss) {
+ const ::std::string& str = ss->str();
+ const char* const start = str.c_str();
+ const char* const end = start + str.length();
+
+ // We need to use a helper stringstream to do this transformation
+ // because String doesn't support push_back().
+ ::std::stringstream helper;
+ for (const char* ch = start; ch != end; ++ch) {
+ if (*ch == '\0') {
+ helper << "\\0"; // Replaces NUL with "\\0";
+ } else {
+ helper.put(*ch);
+ }
+ }
+
+ return String(helper.str().c_str());
+}
+
+// Appends the user-supplied message to the Google-Test-generated message.
+String AppendUserMessage(const String& gtest_msg,
+ const Message& user_msg) {
+ // Appends the user message if it's non-empty.
+ const String user_msg_string = user_msg.GetString();
+ if (user_msg_string.empty()) {
+ return gtest_msg;
+ }
+
+ Message msg;
+ msg << gtest_msg << "\n" << user_msg_string;
+
+ return msg.GetString();
+}
+
+} // namespace internal
+
+// class TestResult
+
+// Creates an empty TestResult.
+TestResult::TestResult()
+ : death_test_count_(0),
+ elapsed_time_(0) {
+}
+
+// D'tor.
+TestResult::~TestResult() {
+}
+
+// Returns the i-th test part result among all the results. i can
+// range from 0 to total_part_count() - 1. If i is not in that range,
+// aborts the program.
+const TestPartResult& TestResult::GetTestPartResult(int i) const {
+ if (i < 0 || i >= total_part_count())
+ internal::posix::Abort();
+ return test_part_results_.at(i);
+}
+
+// Returns the i-th test property. i can range from 0 to
+// test_property_count() - 1. If i is not in that range, aborts the
+// program.
+const TestProperty& TestResult::GetTestProperty(int i) const {
+ if (i < 0 || i >= test_property_count())
+ internal::posix::Abort();
+ return test_properties_.at(i);
+}
+
+// Clears the test part results.
+void TestResult::ClearTestPartResults() {
+ test_part_results_.clear();
+}
+
+// Adds a test part result to the list.
+void TestResult::AddTestPartResult(const TestPartResult& test_part_result) {
+ test_part_results_.push_back(test_part_result);
+}
+
+// Adds a test property to the list. If a property with the same key as the
+// supplied property is already represented, the value of this test_property
+// replaces the old value for that key.
+void TestResult::RecordProperty(const TestProperty& test_property) {
+ if (!ValidateTestProperty(test_property)) {
+ return;
+ }
+ internal::MutexLock lock(&test_properites_mutex_);
+ const std::vector<TestProperty>::iterator property_with_matching_key =
+ std::find_if(test_properties_.begin(), test_properties_.end(),
+ internal::TestPropertyKeyIs(test_property.key()));
+ if (property_with_matching_key == test_properties_.end()) {
+ test_properties_.push_back(test_property);
+ return;
+ }
+ property_with_matching_key->SetValue(test_property.value());
+}
+
+// Adds a failure if the key is a reserved attribute of Google Test
+// testcase tags. Returns true if the property is valid.
+bool TestResult::ValidateTestProperty(const TestProperty& test_property) {
+ internal::String key(test_property.key());
+ if (key == "name" || key == "status" || key == "time" || key == "classname") {
+ ADD_FAILURE()
+ << "Reserved key used in RecordProperty(): "
+ << key
+ << " ('name', 'status', 'time', and 'classname' are reserved by "
+ << GTEST_NAME_ << ")";
+ return false;
+ }
+ return true;
+}
+
+// Clears the object.
+void TestResult::Clear() {
+ test_part_results_.clear();
+ test_properties_.clear();
+ death_test_count_ = 0;
+ elapsed_time_ = 0;
+}
+
+// Returns true iff the test failed.
+bool TestResult::Failed() const {
+ for (int i = 0; i < total_part_count(); ++i) {
+ if (GetTestPartResult(i).failed())
+ return true;
+ }
+ return false;
+}
+
+// Returns true iff the test part fatally failed.
+static bool TestPartFatallyFailed(const TestPartResult& result) {
+ return result.fatally_failed();
+}
+
+// Returns true iff the test fatally failed.
+bool TestResult::HasFatalFailure() const {
+ return CountIf(test_part_results_, TestPartFatallyFailed) > 0;
+}
+
+// Returns true iff the test part non-fatally failed.
+static bool TestPartNonfatallyFailed(const TestPartResult& result) {
+ return result.nonfatally_failed();
+}
+
+// Returns true iff the test has a non-fatal failure.
+bool TestResult::HasNonfatalFailure() const {
+ return CountIf(test_part_results_, TestPartNonfatallyFailed) > 0;
+}
+
+// Gets the number of all test parts. This is the sum of the number
+// of successful test parts and the number of failed test parts.
+int TestResult::total_part_count() const {
+ return static_cast<int>(test_part_results_.size());
+}
+
+// Returns the number of the test properties.
+int TestResult::test_property_count() const {
+ return static_cast<int>(test_properties_.size());
+}
+
+// class Test
+
+// Creates a Test object.
+
+// The c'tor saves the values of all Google Test flags.
+Test::Test()
+ : gtest_flag_saver_(new internal::GTestFlagSaver) {
+}
+
+// The d'tor restores the values of all Google Test flags.
+Test::~Test() {
+ delete gtest_flag_saver_;
+}
+
+// Sets up the test fixture.
+//
+// A sub-class may override this.
+void Test::SetUp() {
+}
+
+// Tears down the test fixture.
+//
+// A sub-class may override this.
+void Test::TearDown() {
+}
+
+// Allows user supplied key value pairs to be recorded for later output.
+void Test::RecordProperty(const char* key, const char* value) {
+ UnitTest::GetInstance()->RecordPropertyForCurrentTest(key, value);
+}
+
+// Allows user supplied key value pairs to be recorded for later output.
+void Test::RecordProperty(const char* key, int value) {
+ Message value_message;
+ value_message << value;
+ RecordProperty(key, value_message.GetString().c_str());
+}
+
+namespace internal {
+
+void ReportFailureInUnknownLocation(TestPartResult::Type result_type,
+ const String& message) {
+ // This function is a friend of UnitTest and as such has access to
+ // AddTestPartResult.
+ UnitTest::GetInstance()->AddTestPartResult(
+ result_type,
+ NULL, // No info about the source file where the exception occurred.
+ -1, // We have no info on which line caused the exception.
+ message,
+ String()); // No stack trace, either.
+}
+
+} // namespace internal
+
+// Google Test requires all tests in the same test case to use the same test
+// fixture class. This function checks if the current test has the
+// same fixture class as the first test in the current test case. If
+// yes, it returns true; otherwise it generates a Google Test failure and
+// returns false.
+bool Test::HasSameFixtureClass() {
+ internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+ const TestCase* const test_case = impl->current_test_case();
+
+ // Info about the first test in the current test case.
+ const TestInfo* const first_test_info = test_case->test_info_list()[0];
+ const internal::TypeId first_fixture_id = first_test_info->fixture_class_id_;
+ const char* const first_test_name = first_test_info->name();
+
+ // Info about the current test.
+ const TestInfo* const this_test_info = impl->current_test_info();
+ const internal::TypeId this_fixture_id = this_test_info->fixture_class_id_;
+ const char* const this_test_name = this_test_info->name();
+
+ if (this_fixture_id != first_fixture_id) {
+ // Is the first test defined using TEST?
+ const bool first_is_TEST = first_fixture_id == internal::GetTestTypeId();
+ // Is this test defined using TEST?
+ const bool this_is_TEST = this_fixture_id == internal::GetTestTypeId();
+
+ if (first_is_TEST || this_is_TEST) {
+ // The user mixed TEST and TEST_F in this test case - we'll tell
+ // him/her how to fix it.
+
+ // Gets the name of the TEST and the name of the TEST_F. Note
+ // that first_is_TEST and this_is_TEST cannot both be true, as
+ // the fixture IDs are different for the two tests.
+ const char* const TEST_name =
+ first_is_TEST ? first_test_name : this_test_name;
+ const char* const TEST_F_name =
+ first_is_TEST ? this_test_name : first_test_name;
+
+ ADD_FAILURE()
+ << "All tests in the same test case must use the same test fixture\n"
+ << "class, so mixing TEST_F and TEST in the same test case is\n"
+ << "illegal. In test case " << this_test_info->test_case_name()
+ << ",\n"
+ << "test " << TEST_F_name << " is defined using TEST_F but\n"
+ << "test " << TEST_name << " is defined using TEST. You probably\n"
+ << "want to change the TEST to TEST_F or move it to another test\n"
+ << "case.";
+ } else {
+ // The user defined two fixture classes with the same name in
+ // two namespaces - we'll tell him/her how to fix it.
+ ADD_FAILURE()
+ << "All tests in the same test case must use the same test fixture\n"
+ << "class. However, in test case "
+ << this_test_info->test_case_name() << ",\n"
+ << "you defined test " << first_test_name
+ << " and test " << this_test_name << "\n"
+ << "using two different test fixture classes. This can happen if\n"
+ << "the two classes are from different namespaces or translation\n"
+ << "units and have the same name. You should probably rename one\n"
+ << "of the classes to put the tests into different test cases.";
+ }
+ return false;
+ }
+
+ return true;
+}
+
+#if GTEST_HAS_SEH
+
+// Adds an "exception thrown" fatal failure to the current test. This
+// function returns its result via an output parameter pointer because VC++
+// prohibits creation of objects with destructors on stack in functions
+// using __try (see error C2712).
+static internal::String* FormatSehExceptionMessage(DWORD exception_code,
+ const char* location) {
+ Message message;
+ message << "SEH exception with code 0x" << std::setbase(16) <<
+ exception_code << std::setbase(10) << " thrown in " << location << ".";
+
+ return new internal::String(message.GetString());
+}
+
+#endif // GTEST_HAS_SEH
+
+#if GTEST_HAS_EXCEPTIONS
+
+// Adds an "exception thrown" fatal failure to the current test.
+static internal::String FormatCxxExceptionMessage(const char* description,
+ const char* location) {
+ Message message;
+ if (description != NULL) {
+ message << "C++ exception with description \"" << description << "\"";
+ } else {
+ message << "Unknown C++ exception";
+ }
+ message << " thrown in " << location << ".";
+
+ return message.GetString();
+}
+
+static internal::String PrintTestPartResultToString(
+ const TestPartResult& test_part_result);
+
+// A failed Google Test assertion will throw an exception of this type when
+// GTEST_FLAG(throw_on_failure) is true (if exceptions are enabled). We
+// derive it from std::runtime_error, which is for errors presumably
+// detectable only at run time. Since std::runtime_error inherits from
+// std::exception, many testing frameworks know how to extract and print the
+// message inside it.
+class GoogleTestFailureException : public ::std::runtime_error {
+ public:
+ explicit GoogleTestFailureException(const TestPartResult& failure)
+ : ::std::runtime_error(PrintTestPartResultToString(failure).c_str()) {}
+};
+#endif // GTEST_HAS_EXCEPTIONS
+
+namespace internal {
+// We put these helper functions in the internal namespace as IBM's xlC
+// compiler rejects the code if they were declared static.
+
+// Runs the given method and handles SEH exceptions it throws, when
+// SEH is supported; returns the 0-value for type Result in case of an
+// SEH exception. (Microsoft compilers cannot handle SEH and C++
+// exceptions in the same function. Therefore, we provide a separate
+// wrapper function for handling SEH exceptions.)
+template <class T, typename Result>
+Result HandleSehExceptionsInMethodIfSupported(
+ T* object, Result (T::*method)(), const char* location) {
+#if GTEST_HAS_SEH
+ __try {
+ return (object->*method)();
+ } __except (internal::UnitTestOptions::GTestShouldProcessSEH( // NOLINT
+ GetExceptionCode())) {
+ // We create the exception message on the heap because VC++ prohibits
+ // creation of objects with destructors on stack in functions using __try
+ // (see error C2712).
+ internal::String* exception_message = FormatSehExceptionMessage(
+ GetExceptionCode(), location);
+ internal::ReportFailureInUnknownLocation(TestPartResult::kFatalFailure,
+ *exception_message);
+ delete exception_message;
+ return static_cast<Result>(0);
+ }
+#else
+ (void)location;
+ return (object->*method)();
+#endif // GTEST_HAS_SEH
+}
+
+// Runs the given method and catches and reports C++ and/or SEH-style
+// exceptions, if they are supported; returns the 0-value for type
+// Result in case of an SEH exception.
+template <class T, typename Result>
+Result HandleExceptionsInMethodIfSupported(
+ T* object, Result (T::*method)(), const char* location) {
+ // NOTE: The user code can affect the way in which Google Test handles
+ // exceptions by setting GTEST_FLAG(catch_exceptions), but only before
+ // RUN_ALL_TESTS() starts. It is technically possible to check the flag
+ // after the exception is caught and either report or re-throw the
+ // exception based on the flag's value:
+ //
+ // try {
+ // // Perform the test method.
+ // } catch (...) {
+ // if (GTEST_FLAG(catch_exceptions))
+ // // Report the exception as failure.
+ // else
+ // throw; // Re-throws the original exception.
+ // }
+ //
+ // However, the purpose of this flag is to allow the program to drop into
+ // the debugger when the exception is thrown. On most platforms, once the
+ // control enters the catch block, the exception origin information is
+ // lost and the debugger will stop the program at the point of the
+ // re-throw in this function -- instead of at the point of the original
+ // throw statement in the code under test. For this reason, we perform
+ // the check early, sacrificing the ability to affect Google Test's
+ // exception handling in the method where the exception is thrown.
+ if (internal::GetUnitTestImpl()->catch_exceptions()) {
+#if GTEST_HAS_EXCEPTIONS
+ try {
+ return HandleSehExceptionsInMethodIfSupported(object, method, location);
+ } catch (const GoogleTestFailureException&) { // NOLINT
+ // This exception doesn't originate in code under test. It makes no
+ // sense to report it as a test failure.
+ throw;
+ } catch (const std::exception& e) { // NOLINT
+ internal::ReportFailureInUnknownLocation(
+ TestPartResult::kFatalFailure,
+ FormatCxxExceptionMessage(e.what(), location));
+ } catch (...) { // NOLINT
+ internal::ReportFailureInUnknownLocation(
+ TestPartResult::kFatalFailure,
+ FormatCxxExceptionMessage(NULL, location));
+ }
+ return static_cast<Result>(0);
+#else
+ return HandleSehExceptionsInMethodIfSupported(object, method, location);
+#endif // GTEST_HAS_EXCEPTIONS
+ } else {
+ return (object->*method)();
+ }
+}
+
+} // namespace internal
+
+// Runs the test and updates the test result.
+void Test::Run() {
+ if (!HasSameFixtureClass()) return;
+
+ internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+ impl->os_stack_trace_getter()->UponLeavingGTest();
+ internal::HandleExceptionsInMethodIfSupported(this, &Test::SetUp, "SetUp()");
+ // We will run the test only if SetUp() was successful.
+ if (!HasFatalFailure()) {
+ impl->os_stack_trace_getter()->UponLeavingGTest();
+ internal::HandleExceptionsInMethodIfSupported(
+ this, &Test::TestBody, "the test body");
+ }
+
+ // However, we want to clean up as much as possible. Hence we will
+ // always call TearDown(), even if SetUp() or the test body has
+ // failed.
+ impl->os_stack_trace_getter()->UponLeavingGTest();
+ internal::HandleExceptionsInMethodIfSupported(
+ this, &Test::TearDown, "TearDown()");
+}
+
+// Returns true iff the current test has a fatal failure.
+bool Test::HasFatalFailure() {
+ return internal::GetUnitTestImpl()->current_test_result()->HasFatalFailure();
+}
+
+// Returns true iff the current test has a non-fatal failure.
+bool Test::HasNonfatalFailure() {
+ return internal::GetUnitTestImpl()->current_test_result()->
+ HasNonfatalFailure();
+}
+
+// class TestInfo
+
+// Constructs a TestInfo object. It assumes ownership of the test factory
+// object.
+// TODO(vladl@google.com): Make a_test_case_name and a_name const string&'s
+// to signify they cannot be NULLs.
+TestInfo::TestInfo(const char* a_test_case_name,
+ const char* a_name,
+ const char* a_type_param,
+ const char* a_value_param,
+ internal::TypeId fixture_class_id,
+ internal::TestFactoryBase* factory)
+ : test_case_name_(a_test_case_name),
+ name_(a_name),
+ type_param_(a_type_param ? new std::string(a_type_param) : NULL),
+ value_param_(a_value_param ? new std::string(a_value_param) : NULL),
+ fixture_class_id_(fixture_class_id),
+ should_run_(false),
+ is_disabled_(false),
+ matches_filter_(false),
+ factory_(factory),
+ result_() {}
+
+// Destructs a TestInfo object.
+TestInfo::~TestInfo() { delete factory_; }
+
+namespace internal {
+
+// Creates a new TestInfo object and registers it with Google Test;
+// returns the created object.
+//
+// Arguments:
+//
+// test_case_name: name of the test case
+// name: name of the test
+// type_param: the name of the test's type parameter, or NULL if
+// this is not a typed or a type-parameterized test.
+// value_param: text representation of the test's value parameter,
+// or NULL if this is not a value-parameterized test.
+// fixture_class_id: ID of the test fixture class
+// set_up_tc: pointer to the function that sets up the test case
+// tear_down_tc: pointer to the function that tears down the test case
+// factory: pointer to the factory that creates a test object.
+// The newly created TestInfo instance will assume
+// ownership of the factory object.
+TestInfo* MakeAndRegisterTestInfo(
+ const char* test_case_name, const char* name,
+ const char* type_param,
+ const char* value_param,
+ TypeId fixture_class_id,
+ SetUpTestCaseFunc set_up_tc,
+ TearDownTestCaseFunc tear_down_tc,
+ TestFactoryBase* factory) {
+ TestInfo* const test_info =
+ new TestInfo(test_case_name, name, type_param, value_param,
+ fixture_class_id, factory);
+ GetUnitTestImpl()->AddTestInfo(set_up_tc, tear_down_tc, test_info);
+ return test_info;
+}
+
+#if GTEST_HAS_PARAM_TEST
+void ReportInvalidTestCaseType(const char* test_case_name,
+ const char* file, int line) {
+ Message errors;
+ errors
+ << "Attempted redefinition of test case " << test_case_name << ".\n"
+ << "All tests in the same test case must use the same test fixture\n"
+ << "class. However, in test case " << test_case_name << ", you tried\n"
+ << "to define a test using a fixture class different from the one\n"
+ << "used earlier. This can happen if the two fixture classes are\n"
+ << "from different namespaces and have the same name. You should\n"
+ << "probably rename one of the classes to put the tests into different\n"
+ << "test cases.";
+
+ fprintf(stderr, "%s %s", FormatFileLocation(file, line).c_str(),
+ errors.GetString().c_str());
+}
+#endif // GTEST_HAS_PARAM_TEST
+
+} // namespace internal
+
+namespace {
+
+// A predicate that checks the test name of a TestInfo against a known
+// value.
+//
+// This is used for implementation of the TestCase class only. We put
+// it in the anonymous namespace to prevent polluting the outer
+// namespace.
+//
+// TestNameIs is copyable.
+class TestNameIs {
+ public:
+ // Constructor.
+ //
+ // TestNameIs has NO default constructor.
+ explicit TestNameIs(const char* name)
+ : name_(name) {}
+
+ // Returns true iff the test name of test_info matches name_.
+ bool operator()(const TestInfo * test_info) const {
+ return test_info && internal::String(test_info->name()).Compare(name_) == 0;
+ }
+
+ private:
+ internal::String name_;
+};
+
+} // namespace
+
+namespace internal {
+
+// This method expands all parameterized tests registered with macros TEST_P
+// and INSTANTIATE_TEST_CASE_P into regular tests and registers those.
+// This will be done just once during the program runtime.
+void UnitTestImpl::RegisterParameterizedTests() {
+#if GTEST_HAS_PARAM_TEST
+ if (!parameterized_tests_registered_) {
+ parameterized_test_registry_.RegisterTests();
+ parameterized_tests_registered_ = true;
+ }
+#endif
+}
+
+} // namespace internal
+
+// Creates the test object, runs it, records its result, and then
+// deletes it.
+void TestInfo::Run() {
+ if (!should_run_) return;
+
+ // Tells UnitTest where to store test result.
+ internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+ impl->set_current_test_info(this);
+
+ TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater();
+
+ // Notifies the unit test event listeners that a test is about to start.
+ repeater->OnTestStart(*this);
+
+ const TimeInMillis start = internal::GetTimeInMillis();
+
+ impl->os_stack_trace_getter()->UponLeavingGTest();
+
+ // Creates the test object.
+ Test* const test = internal::HandleExceptionsInMethodIfSupported(
+ factory_, &internal::TestFactoryBase::CreateTest,
+ "the test fixture's constructor");
+
+ // Runs the test only if the test object was created and its
+ // constructor didn't generate a fatal failure.
+ if ((test != NULL) && !Test::HasFatalFailure()) {
+ // This doesn't throw as all user code that can throw are wrapped into
+ // exception handling code.
+ test->Run();
+ }
+
+ // Deletes the test object.
+ impl->os_stack_trace_getter()->UponLeavingGTest();
+ internal::HandleExceptionsInMethodIfSupported(
+ test, &Test::DeleteSelf_, "the test fixture's destructor");
+
+ result_.set_elapsed_time(internal::GetTimeInMillis() - start);
+
+ // Notifies the unit test event listener that a test has just finished.
+ repeater->OnTestEnd(*this);
+
+ // Tells UnitTest to stop associating assertion results to this
+ // test.
+ impl->set_current_test_info(NULL);
+}
+
+// class TestCase
+
+// Gets the number of successful tests in this test case.
+int TestCase::successful_test_count() const {
+ return CountIf(test_info_list_, TestPassed);
+}
+
+// Gets the number of failed tests in this test case.
+int TestCase::failed_test_count() const {
+ return CountIf(test_info_list_, TestFailed);
+}
+
+int TestCase::disabled_test_count() const {
+ return CountIf(test_info_list_, TestDisabled);
+}
+
+// Get the number of tests in this test case that should run.
+int TestCase::test_to_run_count() const {
+ return CountIf(test_info_list_, ShouldRunTest);
+}
+
+// Gets the number of all tests.
+int TestCase::total_test_count() const {
+ return static_cast<int>(test_info_list_.size());
+}
+
+// Creates a TestCase with the given name.
+//
+// Arguments:
+//
+// name: name of the test case
+// a_type_param: the name of the test case's type parameter, or NULL if
+// this is not a typed or a type-parameterized test case.
+// set_up_tc: pointer to the function that sets up the test case
+// tear_down_tc: pointer to the function that tears down the test case
+TestCase::TestCase(const char* a_name, const char* a_type_param,
+ Test::SetUpTestCaseFunc set_up_tc,
+ Test::TearDownTestCaseFunc tear_down_tc)
+ : name_(a_name),
+ type_param_(a_type_param ? new std::string(a_type_param) : NULL),
+ set_up_tc_(set_up_tc),
+ tear_down_tc_(tear_down_tc),
+ should_run_(false),
+ elapsed_time_(0) {
+}
+
+// Destructor of TestCase.
+TestCase::~TestCase() {
+ // Deletes every Test in the collection.
+ ForEach(test_info_list_, internal::Delete<TestInfo>);
+}
+
+// Returns the i-th test among all the tests. i can range from 0 to
+// total_test_count() - 1. If i is not in that range, returns NULL.
+const TestInfo* TestCase::GetTestInfo(int i) const {
+ const int index = GetElementOr(test_indices_, i, -1);
+ return index < 0 ? NULL : test_info_list_[index];
+}
+
+// Returns the i-th test among all the tests. i can range from 0 to
+// total_test_count() - 1. If i is not in that range, returns NULL.
+TestInfo* TestCase::GetMutableTestInfo(int i) {
+ const int index = GetElementOr(test_indices_, i, -1);
+ return index < 0 ? NULL : test_info_list_[index];
+}
+
+// Adds a test to this test case. Will delete the test upon
+// destruction of the TestCase object.
+void TestCase::AddTestInfo(TestInfo * test_info) {
+ test_info_list_.push_back(test_info);
+ test_indices_.push_back(static_cast<int>(test_indices_.size()));
+}
+
+// Runs every test in this TestCase.
+void TestCase::Run() {
+ if (!should_run_) return;
+
+ internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+ impl->set_current_test_case(this);
+
+ TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater();
+
+ repeater->OnTestCaseStart(*this);
+ impl->os_stack_trace_getter()->UponLeavingGTest();
+ internal::HandleExceptionsInMethodIfSupported(
+ this, &TestCase::RunSetUpTestCase, "SetUpTestCase()");
+
+ const internal::TimeInMillis start = internal::GetTimeInMillis();
+ for (int i = 0; i < total_test_count(); i++) {
+ GetMutableTestInfo(i)->Run();
+ }
+ elapsed_time_ = internal::GetTimeInMillis() - start;
+
+ impl->os_stack_trace_getter()->UponLeavingGTest();
+ internal::HandleExceptionsInMethodIfSupported(
+ this, &TestCase::RunTearDownTestCase, "TearDownTestCase()");
+
+ repeater->OnTestCaseEnd(*this);
+ impl->set_current_test_case(NULL);
+}
+
+// Clears the results of all tests in this test case.
+void TestCase::ClearResult() {
+ ForEach(test_info_list_, TestInfo::ClearTestResult);
+}
+
+// Shuffles the tests in this test case.
+void TestCase::ShuffleTests(internal::Random* random) {
+ Shuffle(random, &test_indices_);
+}
+
+// Restores the test order to before the first shuffle.
+void TestCase::UnshuffleTests() {
+ for (size_t i = 0; i < test_indices_.size(); i++) {
+ test_indices_[i] = static_cast<int>(i);
+ }
+}
+
+// Formats a countable noun. Depending on its quantity, either the
+// singular form or the plural form is used. e.g.
+//
+// FormatCountableNoun(1, "formula", "formuli") returns "1 formula".
+// FormatCountableNoun(5, "book", "books") returns "5 books".
+static internal::String FormatCountableNoun(int count,
+ const char * singular_form,
+ const char * plural_form) {
+ return internal::String::Format("%d %s", count,
+ count == 1 ? singular_form : plural_form);
+}
+
+// Formats the count of tests.
+static internal::String FormatTestCount(int test_count) {
+ return FormatCountableNoun(test_count, "test", "tests");
+}
+
+// Formats the count of test cases.
+static internal::String FormatTestCaseCount(int test_case_count) {
+ return FormatCountableNoun(test_case_count, "test case", "test cases");
+}
+
+// Converts a TestPartResult::Type enum to human-friendly string
+// representation. Both kNonFatalFailure and kFatalFailure are translated
+// to "Failure", as the user usually doesn't care about the difference
+// between the two when viewing the test result.
+static const char * TestPartResultTypeToString(TestPartResult::Type type) {
+ switch (type) {
+ case TestPartResult::kSuccess:
+ return "Success";
+
+ case TestPartResult::kNonFatalFailure:
+ case TestPartResult::kFatalFailure:
+#ifdef _MSC_VER
+ return "error: ";
+#else
+ return "Failure\n";
+#endif
+ default:
+ return "Unknown result type";
+ }
+}
+
+// Prints a TestPartResult to a String.
+static internal::String PrintTestPartResultToString(
+ const TestPartResult& test_part_result) {
+ return (Message()
+ << internal::FormatFileLocation(test_part_result.file_name(),
+ test_part_result.line_number())
+ << " " << TestPartResultTypeToString(test_part_result.type())
+ << test_part_result.message()).GetString();
+}
+
+// Prints a TestPartResult.
+static void PrintTestPartResult(const TestPartResult& test_part_result) {
+ const internal::String& result =
+ PrintTestPartResultToString(test_part_result);
+ printf("%s\n", result.c_str());
+ fflush(stdout);
+ // If the test program runs in Visual Studio or a debugger, the
+ // following statements add the test part result message to the Output
+ // window such that the user can double-click on it to jump to the
+ // corresponding source code location; otherwise they do nothing.
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+ // We don't call OutputDebugString*() on Windows Mobile, as printing
+ // to stdout is done by OutputDebugString() there already - we don't
+ // want the same message printed twice.
+ ::OutputDebugStringA(result.c_str());
+ ::OutputDebugStringA("\n");
+#endif
+}
+
+// class PrettyUnitTestResultPrinter
+
+namespace internal {
+
+enum GTestColor {
+ COLOR_DEFAULT,
+ COLOR_RED,
+ COLOR_GREEN,
+ COLOR_YELLOW
+};
+
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+
+// Returns the character attribute for the given color.
+WORD GetColorAttribute(GTestColor color) {
+ switch (color) {
+ case COLOR_RED: return FOREGROUND_RED;
+ case COLOR_GREEN: return FOREGROUND_GREEN;
+ case COLOR_YELLOW: return FOREGROUND_RED | FOREGROUND_GREEN;
+ default: return 0;
+ }
+}
+
+#else
+
+// Returns the ANSI color code for the given color. COLOR_DEFAULT is
+// an invalid input.
+const char* GetAnsiColorCode(GTestColor color) {
+ switch (color) {
+ case COLOR_RED: return "1";
+ case COLOR_GREEN: return "2";
+ case COLOR_YELLOW: return "3";
+ default: return NULL;
+ };
+}
+
+#endif // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+
+// Returns true iff Google Test should use colors in the output.
+bool ShouldUseColor(bool stdout_is_tty) {
+ const char* const gtest_color = GTEST_FLAG(color).c_str();
+
+ if (String::CaseInsensitiveCStringEquals(gtest_color, "auto")) {
+#if GTEST_OS_WINDOWS
+ // On Windows the TERM variable is usually not set, but the
+ // console there does support colors.
+ return stdout_is_tty;
+#else
+ // On non-Windows platforms, we rely on the TERM variable.
+ const char* const term = posix::GetEnv("TERM");
+ const bool term_supports_color =
+ String::CStringEquals(term, "xterm") ||
+ String::CStringEquals(term, "xterm-color") ||
+ String::CStringEquals(term, "xterm-256color") ||
+ String::CStringEquals(term, "screen") ||
+ String::CStringEquals(term, "linux") ||
+ String::CStringEquals(term, "cygwin");
+ return stdout_is_tty && term_supports_color;
+#endif // GTEST_OS_WINDOWS
+ }
+
+ return String::CaseInsensitiveCStringEquals(gtest_color, "yes") ||
+ String::CaseInsensitiveCStringEquals(gtest_color, "true") ||
+ String::CaseInsensitiveCStringEquals(gtest_color, "t") ||
+ String::CStringEquals(gtest_color, "1");
+ // We take "yes", "true", "t", and "1" as meaning "yes". If the
+ // value is neither one of these nor "auto", we treat it as "no" to
+ // be conservative.
+}
+
+// Helpers for printing colored strings to stdout. Note that on Windows, we
+// cannot simply emit special characters and have the terminal change colors.
+// This routine must actually emit the characters rather than return a string
+// that would be colored when printed, as can be done on Linux.
+void ColoredPrintf(GTestColor color, const char* fmt, ...) {
+ va_list args;
+ va_start(args, fmt);
+
+#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || GTEST_OS_ZOS
+ const bool use_color = false;
+#else
+ static const bool in_color_mode =
+ ShouldUseColor(posix::IsATTY(posix::FileNo(stdout)) != 0);
+ const bool use_color = in_color_mode && (color != COLOR_DEFAULT);
+#endif // GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || GTEST_OS_ZOS
+ // The '!= 0' comparison is necessary to satisfy MSVC 7.1.
+
+ if (!use_color) {
+ vprintf(fmt, args);
+ va_end(args);
+ return;
+ }
+
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+ const HANDLE stdout_handle = GetStdHandle(STD_OUTPUT_HANDLE);
+
+ // Gets the current text color.
+ CONSOLE_SCREEN_BUFFER_INFO buffer_info;
+ GetConsoleScreenBufferInfo(stdout_handle, &buffer_info);
+ const WORD old_color_attrs = buffer_info.wAttributes;
+
+ // We need to flush the stream buffers into the console before each
+ // SetConsoleTextAttribute call lest it affect the text that is already
+ // printed but has not yet reached the console.
+ fflush(stdout);
+ SetConsoleTextAttribute(stdout_handle,
+ GetColorAttribute(color) | FOREGROUND_INTENSITY);
+ vprintf(fmt, args);
+
+ fflush(stdout);
+ // Restores the text color.
+ SetConsoleTextAttribute(stdout_handle, old_color_attrs);
+#else
+ printf("\033[0;3%sm", GetAnsiColorCode(color));
+ vprintf(fmt, args);
+ printf("\033[m"); // Resets the terminal to default.
+#endif // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+ va_end(args);
+}
+
+void PrintFullTestCommentIfPresent(const TestInfo& test_info) {
+ const char* const type_param = test_info.type_param();
+ const char* const value_param = test_info.value_param();
+
+ if (type_param != NULL || value_param != NULL) {
+ printf(", where ");
+ if (type_param != NULL) {
+ printf("TypeParam = %s", type_param);
+ if (value_param != NULL)
+ printf(" and ");
+ }
+ if (value_param != NULL) {
+ printf("GetParam() = %s", value_param);
+ }
+ }
+}
+
+// This class implements the TestEventListener interface.
+//
+// Class PrettyUnitTestResultPrinter is copyable.
+class PrettyUnitTestResultPrinter : public TestEventListener {
+ public:
+ PrettyUnitTestResultPrinter() {}
+ static void PrintTestName(const char * test_case, const char * test) {
+ printf("%s.%s", test_case, test);
+ }
+
+ // The following methods override what's in the TestEventListener class.
+ virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {}
+ virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration);
+ virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test);
+ virtual void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) {}
+ virtual void OnTestCaseStart(const TestCase& test_case);
+ virtual void OnTestStart(const TestInfo& test_info);
+ virtual void OnTestPartResult(const TestPartResult& result);
+ virtual void OnTestEnd(const TestInfo& test_info);
+ virtual void OnTestCaseEnd(const TestCase& test_case);
+ virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test);
+ virtual void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) {}
+ virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+ virtual void OnTestProgramEnd(const UnitTest& /*unit_test*/) {}
+
+ private:
+ static void PrintFailedTests(const UnitTest& unit_test);
+
+ internal::String test_case_name_;
+};
+
+ // Fired before each iteration of tests starts.
+void PrettyUnitTestResultPrinter::OnTestIterationStart(
+ const UnitTest& unit_test, int iteration) {
+ if (GTEST_FLAG(repeat) != 1)
+ printf("\nRepeating all tests (iteration %d) . . .\n\n", iteration + 1);
+
+ const char* const filter = GTEST_FLAG(filter).c_str();
+
+ // Prints the filter if it's not *. This reminds the user that some
+ // tests may be skipped.
+ if (!internal::String::CStringEquals(filter, kUniversalFilter)) {
+ ColoredPrintf(COLOR_YELLOW,
+ "Note: %s filter = %s\n", GTEST_NAME_, filter);
+ }
+
+ if (internal::ShouldShard(kTestTotalShards, kTestShardIndex, false)) {
+ const Int32 shard_index = Int32FromEnvOrDie(kTestShardIndex, -1);
+ ColoredPrintf(COLOR_YELLOW,
+ "Note: This is test shard %d of %s.\n",
+ static_cast<int>(shard_index) + 1,
+ internal::posix::GetEnv(kTestTotalShards));
+ }
+
+ if (GTEST_FLAG(shuffle)) {
+ ColoredPrintf(COLOR_YELLOW,
+ "Note: Randomizing tests' orders with a seed of %d .\n",
+ unit_test.random_seed());
+ }
+
+ ColoredPrintf(COLOR_GREEN, "[==========] ");
+ printf("Running %s from %s.\n",
+ FormatTestCount(unit_test.test_to_run_count()).c_str(),
+ FormatTestCaseCount(unit_test.test_case_to_run_count()).c_str());
+ fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnEnvironmentsSetUpStart(
+ const UnitTest& /*unit_test*/) {
+ ColoredPrintf(COLOR_GREEN, "[----------] ");
+ printf("Global test environment set-up.\n");
+ fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestCaseStart(const TestCase& test_case) {
+ test_case_name_ = test_case.name();
+ const internal::String counts =
+ FormatCountableNoun(test_case.test_to_run_count(), "test", "tests");
+ ColoredPrintf(COLOR_GREEN, "[----------] ");
+ printf("%s from %s", counts.c_str(), test_case_name_.c_str());
+ if (test_case.type_param() == NULL) {
+ printf("\n");
+ } else {
+ printf(", where TypeParam = %s\n", test_case.type_param());
+ }
+ fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestStart(const TestInfo& test_info) {
+ ColoredPrintf(COLOR_GREEN, "[ RUN ] ");
+ PrintTestName(test_case_name_.c_str(), test_info.name());
+ printf("\n");
+ fflush(stdout);
+}
+
+// Called after an assertion failure.
+void PrettyUnitTestResultPrinter::OnTestPartResult(
+ const TestPartResult& result) {
+ // If the test part succeeded, we don't need to do anything.
+ if (result.type() == TestPartResult::kSuccess)
+ return;
+
+ // Print failure message from the assertion (e.g. expected this and got that).
+ PrintTestPartResult(result);
+ fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestEnd(const TestInfo& test_info) {
+ if (test_info.result()->Passed()) {
+ ColoredPrintf(COLOR_GREEN, "[ OK ] ");
+ } else {
+ ColoredPrintf(COLOR_RED, "[ FAILED ] ");
+ }
+ PrintTestName(test_case_name_.c_str(), test_info.name());
+ if (test_info.result()->Failed())
+ PrintFullTestCommentIfPresent(test_info);
+
+ if (GTEST_FLAG(print_time)) {
+ printf(" (%s ms)\n", internal::StreamableToString(
+ test_info.result()->elapsed_time()).c_str());
+ } else {
+ printf("\n");
+ }
+ fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestCaseEnd(const TestCase& test_case) {
+ if (!GTEST_FLAG(print_time)) return;
+
+ test_case_name_ = test_case.name();
+ const internal::String counts =
+ FormatCountableNoun(test_case.test_to_run_count(), "test", "tests");
+ ColoredPrintf(COLOR_GREEN, "[----------] ");
+ printf("%s from %s (%s ms total)\n\n",
+ counts.c_str(), test_case_name_.c_str(),
+ internal::StreamableToString(test_case.elapsed_time()).c_str());
+ fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnEnvironmentsTearDownStart(
+ const UnitTest& /*unit_test*/) {
+ ColoredPrintf(COLOR_GREEN, "[----------] ");
+ printf("Global test environment tear-down\n");
+ fflush(stdout);
+}
+
+// Internal helper for printing the list of failed tests.
+void PrettyUnitTestResultPrinter::PrintFailedTests(const UnitTest& unit_test) {
+ const int failed_test_count = unit_test.failed_test_count();
+ if (failed_test_count == 0) {
+ return;
+ }
+
+ for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+ const TestCase& test_case = *unit_test.GetTestCase(i);
+ if (!test_case.should_run() || (test_case.failed_test_count() == 0)) {
+ continue;
+ }
+ for (int j = 0; j < test_case.total_test_count(); ++j) {
+ const TestInfo& test_info = *test_case.GetTestInfo(j);
+ if (!test_info.should_run() || test_info.result()->Passed()) {
+ continue;
+ }
+ ColoredPrintf(COLOR_RED, "[ FAILED ] ");
+ printf("%s.%s", test_case.name(), test_info.name());
+ PrintFullTestCommentIfPresent(test_info);
+ printf("\n");
+ }
+ }
+}
+
+void PrettyUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test,
+ int /*iteration*/) {
+ ColoredPrintf(COLOR_GREEN, "[==========] ");
+ printf("%s from %s ran.",
+ FormatTestCount(unit_test.test_to_run_count()).c_str(),
+ FormatTestCaseCount(unit_test.test_case_to_run_count()).c_str());
+ if (GTEST_FLAG(print_time)) {
+ printf(" (%s ms total)",
+ internal::StreamableToString(unit_test.elapsed_time()).c_str());
+ }
+ printf("\n");
+ ColoredPrintf(COLOR_GREEN, "[ PASSED ] ");
+ printf("%s.\n", FormatTestCount(unit_test.successful_test_count()).c_str());
+
+ int num_failures = unit_test.failed_test_count();
+ if (!unit_test.Passed()) {
+ const int failed_test_count = unit_test.failed_test_count();
+ ColoredPrintf(COLOR_RED, "[ FAILED ] ");
+ printf("%s, listed below:\n", FormatTestCount(failed_test_count).c_str());
+ PrintFailedTests(unit_test);
+ printf("\n%2d FAILED %s\n", num_failures,
+ num_failures == 1 ? "TEST" : "TESTS");
+ }
+
+ int num_disabled = unit_test.disabled_test_count();
+ if (num_disabled && !GTEST_FLAG(also_run_disabled_tests)) {
+ if (!num_failures) {
+ printf("\n"); // Add a spacer if no FAILURE banner is displayed.
+ }
+ ColoredPrintf(COLOR_YELLOW,
+ " YOU HAVE %d DISABLED %s\n\n",
+ num_disabled,
+ num_disabled == 1 ? "TEST" : "TESTS");
+ }
+ // Ensure that Google Test output is printed before, e.g., heapchecker output.
+ fflush(stdout);
+}
+
+// End PrettyUnitTestResultPrinter
+
+// class TestEventRepeater
+//
+// This class forwards events to other event listeners.
+class TestEventRepeater : public TestEventListener {
+ public:
+ TestEventRepeater() : forwarding_enabled_(true) {}
+ virtual ~TestEventRepeater();
+ void Append(TestEventListener *listener);
+ TestEventListener* Release(TestEventListener* listener);
+
+ // Controls whether events will be forwarded to listeners_. Set to false
+ // in death test child processes.
+ bool forwarding_enabled() const { return forwarding_enabled_; }
+ void set_forwarding_enabled(bool enable) { forwarding_enabled_ = enable; }
+
+ virtual void OnTestProgramStart(const UnitTest& unit_test);
+ virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration);
+ virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test);
+ virtual void OnEnvironmentsSetUpEnd(const UnitTest& unit_test);
+ virtual void OnTestCaseStart(const TestCase& test_case);
+ virtual void OnTestStart(const TestInfo& test_info);
+ virtual void OnTestPartResult(const TestPartResult& result);
+ virtual void OnTestEnd(const TestInfo& test_info);
+ virtual void OnTestCaseEnd(const TestCase& test_case);
+ virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test);
+ virtual void OnEnvironmentsTearDownEnd(const UnitTest& unit_test);
+ virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+ virtual void OnTestProgramEnd(const UnitTest& unit_test);
+
+ private:
+ // Controls whether events will be forwarded to listeners_. Set to false
+ // in death test child processes.
+ bool forwarding_enabled_;
+ // The list of listeners that receive events.
+ std::vector<TestEventListener*> listeners_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventRepeater);
+};
+
+TestEventRepeater::~TestEventRepeater() {
+ ForEach(listeners_, Delete<TestEventListener>);
+}
+
+void TestEventRepeater::Append(TestEventListener *listener) {
+ listeners_.push_back(listener);
+}
+
+// TODO(vladl@google.com): Factor the search functionality into Vector::Find.
+TestEventListener* TestEventRepeater::Release(TestEventListener *listener) {
+ for (size_t i = 0; i < listeners_.size(); ++i) {
+ if (listeners_[i] == listener) {
+ listeners_.erase(listeners_.begin() + i);
+ return listener;
+ }
+ }
+
+ return NULL;
+}
+
+// Since most methods are very similar, use macros to reduce boilerplate.
+// This defines a member that forwards the call to all listeners.
+#define GTEST_REPEATER_METHOD_(Name, Type) \
+void TestEventRepeater::Name(const Type& parameter) { \
+ if (forwarding_enabled_) { \
+ for (size_t i = 0; i < listeners_.size(); i++) { \
+ listeners_[i]->Name(parameter); \
+ } \
+ } \
+}
+// This defines a member that forwards the call to all listeners in reverse
+// order.
+#define GTEST_REVERSE_REPEATER_METHOD_(Name, Type) \
+void TestEventRepeater::Name(const Type& parameter) { \
+ if (forwarding_enabled_) { \
+ for (int i = static_cast<int>(listeners_.size()) - 1; i >= 0; i--) { \
+ listeners_[i]->Name(parameter); \
+ } \
+ } \
+}
+
+GTEST_REPEATER_METHOD_(OnTestProgramStart, UnitTest)
+GTEST_REPEATER_METHOD_(OnEnvironmentsSetUpStart, UnitTest)
+GTEST_REPEATER_METHOD_(OnTestCaseStart, TestCase)
+GTEST_REPEATER_METHOD_(OnTestStart, TestInfo)
+GTEST_REPEATER_METHOD_(OnTestPartResult, TestPartResult)
+GTEST_REPEATER_METHOD_(OnEnvironmentsTearDownStart, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsSetUpEnd, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsTearDownEnd, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestEnd, TestInfo)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestCaseEnd, TestCase)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestProgramEnd, UnitTest)
+
+#undef GTEST_REPEATER_METHOD_
+#undef GTEST_REVERSE_REPEATER_METHOD_
+
+void TestEventRepeater::OnTestIterationStart(const UnitTest& unit_test,
+ int iteration) {
+ if (forwarding_enabled_) {
+ for (size_t i = 0; i < listeners_.size(); i++) {
+ listeners_[i]->OnTestIterationStart(unit_test, iteration);
+ }
+ }
+}
+
+void TestEventRepeater::OnTestIterationEnd(const UnitTest& unit_test,
+ int iteration) {
+ if (forwarding_enabled_) {
+ for (int i = static_cast<int>(listeners_.size()) - 1; i >= 0; i--) {
+ listeners_[i]->OnTestIterationEnd(unit_test, iteration);
+ }
+ }
+}
+
+// End TestEventRepeater
+
+// This class generates an XML output file.
+class XmlUnitTestResultPrinter : public EmptyTestEventListener {
+ public:
+ explicit XmlUnitTestResultPrinter(const char* output_file);
+
+ virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+
+ private:
+ // Is c a whitespace character that is normalized to a space character
+ // when it appears in an XML attribute value?
+ static bool IsNormalizableWhitespace(char c) {
+ return c == 0x9 || c == 0xA || c == 0xD;
+ }
+
+ // May c appear in a well-formed XML document?
+ static bool IsValidXmlCharacter(char c) {
+ return IsNormalizableWhitespace(c) || c >= 0x20;
+ }
+
+ // Returns an XML-escaped copy of the input string str. If
+ // is_attribute is true, the text is meant to appear as an attribute
+ // value, and normalizable whitespace is preserved by replacing it
+ // with character references.
+ static String EscapeXml(const char* str, bool is_attribute);
+
+ // Returns the given string with all characters invalid in XML removed.
+ static string RemoveInvalidXmlCharacters(const string& str);
+
+ // Convenience wrapper around EscapeXml when str is an attribute value.
+ static String EscapeXmlAttribute(const char* str) {
+ return EscapeXml(str, true);
+ }
+
+ // Convenience wrapper around EscapeXml when str is not an attribute value.
+ static String EscapeXmlText(const char* str) { return EscapeXml(str, false); }
+
+ // Streams an XML CDATA section, escaping invalid CDATA sequences as needed.
+ static void OutputXmlCDataSection(::std::ostream* stream, const char* data);
+
+ // Streams an XML representation of a TestInfo object.
+ static void OutputXmlTestInfo(::std::ostream* stream,
+ const char* test_case_name,
+ const TestInfo& test_info);
+
+ // Prints an XML representation of a TestCase object
+ static void PrintXmlTestCase(FILE* out, const TestCase& test_case);
+
+ // Prints an XML summary of unit_test to output stream out.
+ static void PrintXmlUnitTest(FILE* out, const UnitTest& unit_test);
+
+ // Produces a string representing the test properties in a result as space
+ // delimited XML attributes based on the property key="value" pairs.
+ // When the String is not empty, it includes a space at the beginning,
+ // to delimit this attribute from prior attributes.
+ static String TestPropertiesAsXmlAttributes(const TestResult& result);
+
+ // The output file.
+ const String output_file_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(XmlUnitTestResultPrinter);
+};
+
+// Creates a new XmlUnitTestResultPrinter.
+XmlUnitTestResultPrinter::XmlUnitTestResultPrinter(const char* output_file)
+ : output_file_(output_file) {
+ if (output_file_.c_str() == NULL || output_file_.empty()) {
+ fprintf(stderr, "XML output file may not be null\n");
+ fflush(stderr);
+ exit(EXIT_FAILURE);
+ }
+}
+
+// Called after the unit test ends.
+void XmlUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test,
+ int /*iteration*/) {
+ FILE* xmlout = NULL;
+ FilePath output_file(output_file_);
+ FilePath output_dir(output_file.RemoveFileName());
+
+ if (output_dir.CreateDirectoriesRecursively()) {
+ xmlout = posix::FOpen(output_file_.c_str(), "w");
+ }
+ if (xmlout == NULL) {
+ // TODO(wan): report the reason of the failure.
+ //
+ // We don't do it for now as:
+ //
+ // 1. There is no urgent need for it.
+ // 2. It's a bit involved to make the errno variable thread-safe on
+ // all three operating systems (Linux, Windows, and Mac OS).
+ // 3. To interpret the meaning of errno in a thread-safe way,
+ // we need the strerror_r() function, which is not available on
+ // Windows.
+ fprintf(stderr,
+ "Unable to open file \"%s\"\n",
+ output_file_.c_str());
+ fflush(stderr);
+ exit(EXIT_FAILURE);
+ }
+ PrintXmlUnitTest(xmlout, unit_test);
+ fclose(xmlout);
+}
+
+// Returns an XML-escaped copy of the input string str. If is_attribute
+// is true, the text is meant to appear as an attribute value, and
+// normalizable whitespace is preserved by replacing it with character
+// references.
+//
+// Invalid XML characters in str, if any, are stripped from the output.
+// It is expected that most, if not all, of the text processed by this
+// module will consist of ordinary English text.
+// If this module is ever modified to produce version 1.1 XML output,
+// most invalid characters can be retained using character references.
+// TODO(wan): It might be nice to have a minimally invasive, human-readable
+// escaping scheme for invalid characters, rather than dropping them.
+String XmlUnitTestResultPrinter::EscapeXml(const char* str, bool is_attribute) {
+ Message m;
+
+ if (str != NULL) {
+ for (const char* src = str; *src; ++src) {
+ switch (*src) {
+ case '<':
+ m << "<";
+ break;
+ case '>':
+ m << ">";
+ break;
+ case '&':
+ m << "&";
+ break;
+ case '\'':
+ if (is_attribute)
+ m << "'";
+ else
+ m << '\'';
+ break;
+ case '"':
+ if (is_attribute)
+ m << """;
+ else
+ m << '"';
+ break;
+ default:
+ if (IsValidXmlCharacter(*src)) {
+ if (is_attribute && IsNormalizableWhitespace(*src))
+ m << String::Format("&#x%02X;", unsigned(*src));
+ else
+ m << *src;
+ }
+ break;
+ }
+ }
+ }
+
+ return m.GetString();
+}
+
+// Returns the given string with all characters invalid in XML removed.
+// Currently invalid characters are dropped from the string. An
+// alternative is to replace them with certain characters such as . or ?.
+string XmlUnitTestResultPrinter::RemoveInvalidXmlCharacters(const string& str) {
+ string output;
+ output.reserve(str.size());
+ for (string::const_iterator it = str.begin(); it != str.end(); ++it)
+ if (IsValidXmlCharacter(*it))
+ output.push_back(*it);
+
+ return output;
+}
+
+// The following routines generate an XML representation of a UnitTest
+// object.
+//
+// This is how Google Test concepts map to the DTD:
+//
+// <testsuites name="AllTests"> <-- corresponds to a UnitTest object
+// <testsuite name="testcase-name"> <-- corresponds to a TestCase object
+// <testcase name="test-name"> <-- corresponds to a TestInfo object
+// <failure message="...">...</failure>
+// <failure message="...">...</failure>
+// <failure message="...">...</failure>
+// <-- individual assertion failures
+// </testcase>
+// </testsuite>
+// </testsuites>
+
+// Formats the given time in milliseconds as seconds.
+std::string FormatTimeInMillisAsSeconds(TimeInMillis ms) {
+ ::std::stringstream ss;
+ ss << ms/1000.0;
+ return ss.str();
+}
+
+// Streams an XML CDATA section, escaping invalid CDATA sequences as needed.
+void XmlUnitTestResultPrinter::OutputXmlCDataSection(::std::ostream* stream,
+ const char* data) {
+ const char* segment = data;
+ *stream << "<![CDATA[";
+ for (;;) {
+ const char* const next_segment = strstr(segment, "]]>");
+ if (next_segment != NULL) {
+ stream->write(
+ segment, static_cast<std::streamsize>(next_segment - segment));
+ *stream << "]]>]]><![CDATA[";
+ segment = next_segment + strlen("]]>");
+ } else {
+ *stream << segment;
+ break;
+ }
+ }
+ *stream << "]]>";
+}
+
+// Prints an XML representation of a TestInfo object.
+// TODO(wan): There is also value in printing properties with the plain printer.
+void XmlUnitTestResultPrinter::OutputXmlTestInfo(::std::ostream* stream,
+ const char* test_case_name,
+ const TestInfo& test_info) {
+ const TestResult& result = *test_info.result();
+ *stream << " <testcase name=\""
+ << EscapeXmlAttribute(test_info.name()).c_str() << "\"";
+
+ if (test_info.value_param() != NULL) {
+ *stream << " value_param=\"" << EscapeXmlAttribute(test_info.value_param())
+ << "\"";
+ }
+ if (test_info.type_param() != NULL) {
+ *stream << " type_param=\"" << EscapeXmlAttribute(test_info.type_param())
+ << "\"";
+ }
+
+ *stream << " status=\""
+ << (test_info.should_run() ? "run" : "notrun")
+ << "\" time=\""
+ << FormatTimeInMillisAsSeconds(result.elapsed_time())
+ << "\" classname=\"" << EscapeXmlAttribute(test_case_name).c_str()
+ << "\"" << TestPropertiesAsXmlAttributes(result).c_str();
+
+ int failures = 0;
+ for (int i = 0; i < result.total_part_count(); ++i) {
+ const TestPartResult& part = result.GetTestPartResult(i);
+ if (part.failed()) {
+ if (++failures == 1)
+ *stream << ">\n";
+ *stream << " <failure message=\""
+ << EscapeXmlAttribute(part.summary()).c_str()
+ << "\" type=\"\">";
+ const string location = internal::FormatCompilerIndependentFileLocation(
+ part.file_name(), part.line_number());
+ const string message = location + "\n" + part.message();
+ OutputXmlCDataSection(stream,
+ RemoveInvalidXmlCharacters(message).c_str());
+ *stream << "</failure>\n";
+ }
+ }
+
+ if (failures == 0)
+ *stream << " />\n";
+ else
+ *stream << " </testcase>\n";
+}
+
+// Prints an XML representation of a TestCase object
+void XmlUnitTestResultPrinter::PrintXmlTestCase(FILE* out,
+ const TestCase& test_case) {
+ fprintf(out,
+ " <testsuite name=\"%s\" tests=\"%d\" failures=\"%d\" "
+ "disabled=\"%d\" ",
+ EscapeXmlAttribute(test_case.name()).c_str(),
+ test_case.total_test_count(),
+ test_case.failed_test_count(),
+ test_case.disabled_test_count());
+ fprintf(out,
+ "errors=\"0\" time=\"%s\">\n",
+ FormatTimeInMillisAsSeconds(test_case.elapsed_time()).c_str());
+ for (int i = 0; i < test_case.total_test_count(); ++i) {
+ ::std::stringstream stream;
+ OutputXmlTestInfo(&stream, test_case.name(), *test_case.GetTestInfo(i));
+ fprintf(out, "%s", StringStreamToString(&stream).c_str());
+ }
+ fprintf(out, " </testsuite>\n");
+}
+
+// Prints an XML summary of unit_test to output stream out.
+void XmlUnitTestResultPrinter::PrintXmlUnitTest(FILE* out,
+ const UnitTest& unit_test) {
+ fprintf(out, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
+ fprintf(out,
+ "<testsuites tests=\"%d\" failures=\"%d\" disabled=\"%d\" "
+ "errors=\"0\" time=\"%s\" ",
+ unit_test.total_test_count(),
+ unit_test.failed_test_count(),
+ unit_test.disabled_test_count(),
+ FormatTimeInMillisAsSeconds(unit_test.elapsed_time()).c_str());
+ if (GTEST_FLAG(shuffle)) {
+ fprintf(out, "random_seed=\"%d\" ", unit_test.random_seed());
+ }
+ fprintf(out, "name=\"AllTests\">\n");
+ for (int i = 0; i < unit_test.total_test_case_count(); ++i)
+ PrintXmlTestCase(out, *unit_test.GetTestCase(i));
+ fprintf(out, "</testsuites>\n");
+}
+
+// Produces a string representing the test properties in a result as space
+// delimited XML attributes based on the property key="value" pairs.
+String XmlUnitTestResultPrinter::TestPropertiesAsXmlAttributes(
+ const TestResult& result) {
+ Message attributes;
+ for (int i = 0; i < result.test_property_count(); ++i) {
+ const TestProperty& property = result.GetTestProperty(i);
+ attributes << " " << property.key() << "="
+ << "\"" << EscapeXmlAttribute(property.value()) << "\"";
+ }
+ return attributes.GetString();
+}
+
+// End XmlUnitTestResultPrinter
+
+#if GTEST_CAN_STREAM_RESULTS_
+
+// Streams test results to the given port on the given host machine.
+class StreamingListener : public EmptyTestEventListener {
+ public:
+ // Escapes '=', '&', '%', and '\n' characters in str as "%xx".
+ static string UrlEncode(const char* str);
+
+ StreamingListener(const string& host, const string& port)
+ : sockfd_(-1), host_name_(host), port_num_(port) {
+ MakeConnection();
+ Send("gtest_streaming_protocol_version=1.0\n");
+ }
+
+ virtual ~StreamingListener() {
+ if (sockfd_ != -1)
+ CloseConnection();
+ }
+
+ void OnTestProgramStart(const UnitTest& /* unit_test */) {
+ Send("event=TestProgramStart\n");
+ }
+
+ void OnTestProgramEnd(const UnitTest& unit_test) {
+ // Note that Google Test current only report elapsed time for each
+ // test iteration, not for the entire test program.
+ Send(String::Format("event=TestProgramEnd&passed=%d\n",
+ unit_test.Passed()));
+
+ // Notify the streaming server to stop.
+ CloseConnection();
+ }
+
+ void OnTestIterationStart(const UnitTest& /* unit_test */, int iteration) {
+ Send(String::Format("event=TestIterationStart&iteration=%d\n",
+ iteration));
+ }
+
+ void OnTestIterationEnd(const UnitTest& unit_test, int /* iteration */) {
+ Send(String::Format("event=TestIterationEnd&passed=%d&elapsed_time=%sms\n",
+ unit_test.Passed(),
+ StreamableToString(unit_test.elapsed_time()).c_str()));
+ }
+
+ void OnTestCaseStart(const TestCase& test_case) {
+ Send(String::Format("event=TestCaseStart&name=%s\n", test_case.name()));
+ }
+
+ void OnTestCaseEnd(const TestCase& test_case) {
+ Send(String::Format("event=TestCaseEnd&passed=%d&elapsed_time=%sms\n",
+ test_case.Passed(),
+ StreamableToString(test_case.elapsed_time()).c_str()));
+ }
+
+ void OnTestStart(const TestInfo& test_info) {
+ Send(String::Format("event=TestStart&name=%s\n", test_info.name()));
+ }
+
+ void OnTestEnd(const TestInfo& test_info) {
+ Send(String::Format(
+ "event=TestEnd&passed=%d&elapsed_time=%sms\n",
+ (test_info.result())->Passed(),
+ StreamableToString((test_info.result())->elapsed_time()).c_str()));
+ }
+
+ void OnTestPartResult(const TestPartResult& test_part_result) {
+ const char* file_name = test_part_result.file_name();
+ if (file_name == NULL)
+ file_name = "";
+ Send(String::Format("event=TestPartResult&file=%s&line=%d&message=",
+ UrlEncode(file_name).c_str(),
+ test_part_result.line_number()));
+ Send(UrlEncode(test_part_result.message()) + "\n");
+ }
+
+ private:
+ // Creates a client socket and connects to the server.
+ void MakeConnection();
+
+ // Closes the socket.
+ void CloseConnection() {
+ GTEST_CHECK_(sockfd_ != -1)
+ << "CloseConnection() can be called only when there is a connection.";
+
+ close(sockfd_);
+ sockfd_ = -1;
+ }
+
+ // Sends a string to the socket.
+ void Send(const string& message) {
+ GTEST_CHECK_(sockfd_ != -1)
+ << "Send() can be called only when there is a connection.";
+
+ const int len = static_cast<int>(message.length());
+ if (write(sockfd_, message.c_str(), len) != len) {
+ GTEST_LOG_(WARNING)
+ << "stream_result_to: failed to stream to "
+ << host_name_ << ":" << port_num_;
+ }
+ }
+
+ int sockfd_; // socket file descriptor
+ const string host_name_;
+ const string port_num_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(StreamingListener);
+}; // class StreamingListener
+
+// Checks if str contains '=', '&', '%' or '\n' characters. If yes,
+// replaces them by "%xx" where xx is their hexadecimal value. For
+// example, replaces "=" with "%3D". This algorithm is O(strlen(str))
+// in both time and space -- important as the input str may contain an
+// arbitrarily long test failure message and stack trace.
+string StreamingListener::UrlEncode(const char* str) {
+ string result;
+ result.reserve(strlen(str) + 1);
+ for (char ch = *str; ch != '\0'; ch = *++str) {
+ switch (ch) {
+ case '%':
+ case '=':
+ case '&':
+ case '\n':
+ result.append(String::Format("%%%02x", static_cast<unsigned char>(ch)));
+ break;
+ default:
+ result.push_back(ch);
+ break;
+ }
+ }
+ return result;
+}
+
+void StreamingListener::MakeConnection() {
+ GTEST_CHECK_(sockfd_ == -1)
+ << "MakeConnection() can't be called when there is already a connection.";
+
+ addrinfo hints;
+ memset(&hints, 0, sizeof(hints));
+ hints.ai_family = AF_UNSPEC; // To allow both IPv4 and IPv6 addresses.
+ hints.ai_socktype = SOCK_STREAM;
+ addrinfo* servinfo = NULL;
+
+ // Use the getaddrinfo() to get a linked list of IP addresses for
+ // the given host name.
+ const int error_num = getaddrinfo(
+ host_name_.c_str(), port_num_.c_str(), &hints, &servinfo);
+ if (error_num != 0) {
+ GTEST_LOG_(WARNING) << "stream_result_to: getaddrinfo() failed: "
+ << gai_strerror(error_num);
+ }
+
+ // Loop through all the results and connect to the first we can.
+ for (addrinfo* cur_addr = servinfo; sockfd_ == -1 && cur_addr != NULL;
+ cur_addr = cur_addr->ai_next) {
+ sockfd_ = socket(
+ cur_addr->ai_family, cur_addr->ai_socktype, cur_addr->ai_protocol);
+ if (sockfd_ != -1) {
+ // Connect the client socket to the server socket.
+ if (connect(sockfd_, cur_addr->ai_addr, cur_addr->ai_addrlen) == -1) {
+ close(sockfd_);
+ sockfd_ = -1;
+ }
+ }
+ }
+
+ freeaddrinfo(servinfo); // all done with this structure
+
+ if (sockfd_ == -1) {
+ GTEST_LOG_(WARNING) << "stream_result_to: failed to connect to "
+ << host_name_ << ":" << port_num_;
+ }
+}
+
+// End of class Streaming Listener
+#endif // GTEST_CAN_STREAM_RESULTS__
+
+// Class ScopedTrace
+
+// Pushes the given source file location and message onto a per-thread
+// trace stack maintained by Google Test.
+// L < UnitTest::mutex_
+ScopedTrace::ScopedTrace(const char* file, int line, const Message& message) {
+ TraceInfo trace;
+ trace.file = file;
+ trace.line = line;
+ trace.message = message.GetString();
+
+ UnitTest::GetInstance()->PushGTestTrace(trace);
+}
+
+// Pops the info pushed by the c'tor.
+// L < UnitTest::mutex_
+ScopedTrace::~ScopedTrace() {
+ UnitTest::GetInstance()->PopGTestTrace();
+}
+
+
+// class OsStackTraceGetter
+
+// Returns the current OS stack trace as a String. Parameters:
+//
+// max_depth - the maximum number of stack frames to be included
+// in the trace.
+// skip_count - the number of top frames to be skipped; doesn't count
+// against max_depth.
+//
+// L < mutex_
+// We use "L < mutex_" to denote that the function may acquire mutex_.
+String OsStackTraceGetter::CurrentStackTrace(int, int) {
+ return String("");
+}
+
+// L < mutex_
+void OsStackTraceGetter::UponLeavingGTest() {
+}
+
+const char* const
+OsStackTraceGetter::kElidedFramesMarker =
+ "... " GTEST_NAME_ " internal frames ...";
+
+} // namespace internal
+
+// class TestEventListeners
+
+TestEventListeners::TestEventListeners()
+ : repeater_(new internal::TestEventRepeater()),
+ default_result_printer_(NULL),
+ default_xml_generator_(NULL) {
+}
+
+TestEventListeners::~TestEventListeners() { delete repeater_; }
+
+// Returns the standard listener responsible for the default console
+// output. Can be removed from the listeners list to shut down default
+// console output. Note that removing this object from the listener list
+// with Release transfers its ownership to the user.
+void TestEventListeners::Append(TestEventListener* listener) {
+ repeater_->Append(listener);
+}
+
+// Removes the given event listener from the list and returns it. It then
+// becomes the caller's responsibility to delete the listener. Returns
+// NULL if the listener is not found in the list.
+TestEventListener* TestEventListeners::Release(TestEventListener* listener) {
+ if (listener == default_result_printer_)
+ default_result_printer_ = NULL;
+ else if (listener == default_xml_generator_)
+ default_xml_generator_ = NULL;
+ return repeater_->Release(listener);
+}
+
+// Returns repeater that broadcasts the TestEventListener events to all
+// subscribers.
+TestEventListener* TestEventListeners::repeater() { return repeater_; }
+
+// Sets the default_result_printer attribute to the provided listener.
+// The listener is also added to the listener list and previous
+// default_result_printer is removed from it and deleted. The listener can
+// also be NULL in which case it will not be added to the list. Does
+// nothing if the previous and the current listener objects are the same.
+void TestEventListeners::SetDefaultResultPrinter(TestEventListener* listener) {
+ if (default_result_printer_ != listener) {
+ // It is an error to pass this method a listener that is already in the
+ // list.
+ delete Release(default_result_printer_);
+ default_result_printer_ = listener;
+ if (listener != NULL)
+ Append(listener);
+ }
+}
+
+// Sets the default_xml_generator attribute to the provided listener. The
+// listener is also added to the listener list and previous
+// default_xml_generator is removed from it and deleted. The listener can
+// also be NULL in which case it will not be added to the list. Does
+// nothing if the previous and the current listener objects are the same.
+void TestEventListeners::SetDefaultXmlGenerator(TestEventListener* listener) {
+ if (default_xml_generator_ != listener) {
+ // It is an error to pass this method a listener that is already in the
+ // list.
+ delete Release(default_xml_generator_);
+ default_xml_generator_ = listener;
+ if (listener != NULL)
+ Append(listener);
+ }
+}
+
+// Controls whether events will be forwarded by the repeater to the
+// listeners in the list.
+bool TestEventListeners::EventForwardingEnabled() const {
+ return repeater_->forwarding_enabled();
+}
+
+void TestEventListeners::SuppressEventForwarding() {
+ repeater_->set_forwarding_enabled(false);
+}
+
+// class UnitTest
+
+// Gets the singleton UnitTest object. The first time this method is
+// called, a UnitTest object is constructed and returned. Consecutive
+// calls will return the same object.
+//
+// We don't protect this under mutex_ as a user is not supposed to
+// call this before main() starts, from which point on the return
+// value will never change.
+UnitTest * UnitTest::GetInstance() {
+ // When compiled with MSVC 7.1 in optimized mode, destroying the
+ // UnitTest object upon exiting the program messes up the exit code,
+ // causing successful tests to appear failed. We have to use a
+ // different implementation in this case to bypass the compiler bug.
+ // This implementation makes the compiler happy, at the cost of
+ // leaking the UnitTest object.
+
+ // CodeGear C++Builder insists on a public destructor for the
+ // default implementation. Use this implementation to keep good OO
+ // design with private destructor.
+
+#if (_MSC_VER == 1310 && !defined(_DEBUG)) || defined(__BORLANDC__)
+ static UnitTest* const instance = new UnitTest;
+ return instance;
+#else
+ static UnitTest instance;
+ return &instance;
+#endif // (_MSC_VER == 1310 && !defined(_DEBUG)) || defined(__BORLANDC__)
+}
+
+// Gets the number of successful test cases.
+int UnitTest::successful_test_case_count() const {
+ return impl()->successful_test_case_count();
+}
+
+// Gets the number of failed test cases.
+int UnitTest::failed_test_case_count() const {
+ return impl()->failed_test_case_count();
+}
+
+// Gets the number of all test cases.
+int UnitTest::total_test_case_count() const {
+ return impl()->total_test_case_count();
+}
+
+// Gets the number of all test cases that contain at least one test
+// that should run.
+int UnitTest::test_case_to_run_count() const {
+ return impl()->test_case_to_run_count();
+}
+
+// Gets the number of successful tests.
+int UnitTest::successful_test_count() const {
+ return impl()->successful_test_count();
+}
+
+// Gets the number of failed tests.
+int UnitTest::failed_test_count() const { return impl()->failed_test_count(); }
+
+// Gets the number of disabled tests.
+int UnitTest::disabled_test_count() const {
+ return impl()->disabled_test_count();
+}
+
+// Gets the number of all tests.
+int UnitTest::total_test_count() const { return impl()->total_test_count(); }
+
+// Gets the number of tests that should run.
+int UnitTest::test_to_run_count() const { return impl()->test_to_run_count(); }
+
+// Gets the elapsed time, in milliseconds.
+internal::TimeInMillis UnitTest::elapsed_time() const {
+ return impl()->elapsed_time();
+}
+
+// Returns true iff the unit test passed (i.e. all test cases passed).
+bool UnitTest::Passed() const { return impl()->Passed(); }
+
+// Returns true iff the unit test failed (i.e. some test case failed
+// or something outside of all tests failed).
+bool UnitTest::Failed() const { return impl()->Failed(); }
+
+// Gets the i-th test case among all the test cases. i can range from 0 to
+// total_test_case_count() - 1. If i is not in that range, returns NULL.
+const TestCase* UnitTest::GetTestCase(int i) const {
+ return impl()->GetTestCase(i);
+}
+
+// Gets the i-th test case among all the test cases. i can range from 0 to
+// total_test_case_count() - 1. If i is not in that range, returns NULL.
+TestCase* UnitTest::GetMutableTestCase(int i) {
+ return impl()->GetMutableTestCase(i);
+}
+
+// Returns the list of event listeners that can be used to track events
+// inside Google Test.
+TestEventListeners& UnitTest::listeners() {
+ return *impl()->listeners();
+}
+
+// Registers and returns a global test environment. When a test
+// program is run, all global test environments will be set-up in the
+// order they were registered. After all tests in the program have
+// finished, all global test environments will be torn-down in the
+// *reverse* order they were registered.
+//
+// The UnitTest object takes ownership of the given environment.
+//
+// We don't protect this under mutex_, as we only support calling it
+// from the main thread.
+Environment* UnitTest::AddEnvironment(Environment* env) {
+ if (env == NULL) {
+ return NULL;
+ }
+
+ impl_->environments().push_back(env);
+ return env;
+}
+
+// Adds a TestPartResult to the current TestResult object. All Google Test
+// assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc) eventually call
+// this to report their results. The user code should use the
+// assertion macros instead of calling this directly.
+// L < mutex_
+void UnitTest::AddTestPartResult(TestPartResult::Type result_type,
+ const char* file_name,
+ int line_number,
+ const internal::String& message,
+ const internal::String& os_stack_trace) {
+ Message msg;
+ msg << message;
+
+ internal::MutexLock lock(&mutex_);
+ if (impl_->gtest_trace_stack().size() > 0) {
+ msg << "\n" << GTEST_NAME_ << " trace:";
+
+ for (int i = static_cast<int>(impl_->gtest_trace_stack().size());
+ i > 0; --i) {
+ const internal::TraceInfo& trace = impl_->gtest_trace_stack()[i - 1];
+ msg << "\n" << internal::FormatFileLocation(trace.file, trace.line)
+ << " " << trace.message;
+ }
+ }
+
+ if (os_stack_trace.c_str() != NULL && !os_stack_trace.empty()) {
+ msg << internal::kStackTraceMarker << os_stack_trace;
+ }
+
+ const TestPartResult result =
+ TestPartResult(result_type, file_name, line_number,
+ msg.GetString().c_str());
+ impl_->GetTestPartResultReporterForCurrentThread()->
+ ReportTestPartResult(result);
+
+ if (result_type != TestPartResult::kSuccess) {
+ // gtest_break_on_failure takes precedence over
+ // gtest_throw_on_failure. This allows a user to set the latter
+ // in the code (perhaps in order to use Google Test assertions
+ // with another testing framework) and specify the former on the
+ // command line for debugging.
+ if (GTEST_FLAG(break_on_failure)) {
+#if GTEST_OS_WINDOWS
+ // Using DebugBreak on Windows allows gtest to still break into a debugger
+ // when a failure happens and both the --gtest_break_on_failure and
+ // the --gtest_catch_exceptions flags are specified.
+ DebugBreak();
+#else
+ // Dereference NULL through a volatile pointer to prevent the compiler
+ // from removing. We use this rather than abort() or __builtin_trap() for
+ // portability: Symbian doesn't implement abort() well, and some debuggers
+ // don't correctly trap abort().
+ *static_cast<volatile int*>(NULL) = 1;
+#endif // GTEST_OS_WINDOWS
+ } else if (GTEST_FLAG(throw_on_failure)) {
+#if GTEST_HAS_EXCEPTIONS
+ throw GoogleTestFailureException(result);
+#else
+ // We cannot call abort() as it generates a pop-up in debug mode
+ // that cannot be suppressed in VC 7.1 or below.
+ exit(1);
+#endif
+ }
+ }
+}
+
+// Creates and adds a property to the current TestResult. If a property matching
+// the supplied value already exists, updates its value instead.
+void UnitTest::RecordPropertyForCurrentTest(const char* key,
+ const char* value) {
+ const TestProperty test_property(key, value);
+ impl_->current_test_result()->RecordProperty(test_property);
+}
+
+// Runs all tests in this UnitTest object and prints the result.
+// Returns 0 if successful, or 1 otherwise.
+//
+// We don't protect this under mutex_, as we only support calling it
+// from the main thread.
+int UnitTest::Run() {
+ // Captures the value of GTEST_FLAG(catch_exceptions). This value will be
+ // used for the duration of the program.
+ impl()->set_catch_exceptions(GTEST_FLAG(catch_exceptions));
+
+#if GTEST_HAS_SEH
+ const bool in_death_test_child_process =
+ internal::GTEST_FLAG(internal_run_death_test).length() > 0;
+
+ // Either the user wants Google Test to catch exceptions thrown by the
+ // tests or this is executing in the context of death test child
+ // process. In either case the user does not want to see pop-up dialogs
+ // about crashes - they are expected.
+ if (impl()->catch_exceptions() || in_death_test_child_process) {
+
+# if !GTEST_OS_WINDOWS_MOBILE
+ // SetErrorMode doesn't exist on CE.
+ SetErrorMode(SEM_FAILCRITICALERRORS | SEM_NOALIGNMENTFAULTEXCEPT |
+ SEM_NOGPFAULTERRORBOX | SEM_NOOPENFILEERRORBOX);
+# endif // !GTEST_OS_WINDOWS_MOBILE
+
+# if (defined(_MSC_VER) || GTEST_OS_WINDOWS_MINGW) && !GTEST_OS_WINDOWS_MOBILE
+ // Death test children can be terminated with _abort(). On Windows,
+ // _abort() can show a dialog with a warning message. This forces the
+ // abort message to go to stderr instead.
+ _set_error_mode(_OUT_TO_STDERR);
+# endif
+
+# if _MSC_VER >= 1400 && !GTEST_OS_WINDOWS_MOBILE
+ // In the debug version, Visual Studio pops up a separate dialog
+ // offering a choice to debug the aborted program. We need to suppress
+ // this dialog or it will pop up for every EXPECT/ASSERT_DEATH statement
+ // executed. Google Test will notify the user of any unexpected
+ // failure via stderr.
+ //
+ // VC++ doesn't define _set_abort_behavior() prior to the version 8.0.
+ // Users of prior VC versions shall suffer the agony and pain of
+ // clicking through the countless debug dialogs.
+ // TODO(vladl@google.com): find a way to suppress the abort dialog() in the
+ // debug mode when compiled with VC 7.1 or lower.
+ if (!GTEST_FLAG(break_on_failure))
+ _set_abort_behavior(
+ 0x0, // Clear the following flags:
+ _WRITE_ABORT_MSG | _CALL_REPORTFAULT); // pop-up window, core dump.
+# endif
+
+ }
+#endif // GTEST_HAS_SEH
+
+ return internal::HandleExceptionsInMethodIfSupported(
+ impl(),
+ &internal::UnitTestImpl::RunAllTests,
+ "auxiliary test code (environments or event listeners)") ? 0 : 1;
+}
+
+// Returns the working directory when the first TEST() or TEST_F() was
+// executed.
+const char* UnitTest::original_working_dir() const {
+ return impl_->original_working_dir_.c_str();
+}
+
+// Returns the TestCase object for the test that's currently running,
+// or NULL if no test is running.
+// L < mutex_
+const TestCase* UnitTest::current_test_case() const {
+ internal::MutexLock lock(&mutex_);
+ return impl_->current_test_case();
+}
+
+// Returns the TestInfo object for the test that's currently running,
+// or NULL if no test is running.
+// L < mutex_
+const TestInfo* UnitTest::current_test_info() const {
+ internal::MutexLock lock(&mutex_);
+ return impl_->current_test_info();
+}
+
+// Returns the random seed used at the start of the current test run.
+int UnitTest::random_seed() const { return impl_->random_seed(); }
+
+#if GTEST_HAS_PARAM_TEST
+// Returns ParameterizedTestCaseRegistry object used to keep track of
+// value-parameterized tests and instantiate and register them.
+// L < mutex_
+internal::ParameterizedTestCaseRegistry&
+ UnitTest::parameterized_test_registry() {
+ return impl_->parameterized_test_registry();
+}
+#endif // GTEST_HAS_PARAM_TEST
+
+// Creates an empty UnitTest.
+UnitTest::UnitTest() {
+ impl_ = new internal::UnitTestImpl(this);
+}
+
+// Destructor of UnitTest.
+UnitTest::~UnitTest() {
+ delete impl_;
+}
+
+// Pushes a trace defined by SCOPED_TRACE() on to the per-thread
+// Google Test trace stack.
+// L < mutex_
+void UnitTest::PushGTestTrace(const internal::TraceInfo& trace) {
+ internal::MutexLock lock(&mutex_);
+ impl_->gtest_trace_stack().push_back(trace);
+}
+
+// Pops a trace from the per-thread Google Test trace stack.
+// L < mutex_
+void UnitTest::PopGTestTrace() {
+ internal::MutexLock lock(&mutex_);
+ impl_->gtest_trace_stack().pop_back();
+}
+
+namespace internal {
+
+UnitTestImpl::UnitTestImpl(UnitTest* parent)
+ : parent_(parent),
+#ifdef _MSC_VER
+# pragma warning(push) // Saves the current warning state.
+# pragma warning(disable:4355) // Temporarily disables warning 4355
+ // (using this in initializer).
+ default_global_test_part_result_reporter_(this),
+ default_per_thread_test_part_result_reporter_(this),
+# pragma warning(pop) // Restores the warning state again.
+#else
+ default_global_test_part_result_reporter_(this),
+ default_per_thread_test_part_result_reporter_(this),
+#endif // _MSC_VER
+ global_test_part_result_repoter_(
+ &default_global_test_part_result_reporter_),
+ per_thread_test_part_result_reporter_(
+ &default_per_thread_test_part_result_reporter_),
+#if GTEST_HAS_PARAM_TEST
+ parameterized_test_registry_(),
+ parameterized_tests_registered_(false),
+#endif // GTEST_HAS_PARAM_TEST
+ last_death_test_case_(-1),
+ current_test_case_(NULL),
+ current_test_info_(NULL),
+ ad_hoc_test_result_(),
+ os_stack_trace_getter_(NULL),
+ post_flag_parse_init_performed_(false),
+ random_seed_(0), // Will be overridden by the flag before first use.
+ random_(0), // Will be reseeded before first use.
+ elapsed_time_(0),
+#if GTEST_HAS_DEATH_TEST
+ internal_run_death_test_flag_(NULL),
+ death_test_factory_(new DefaultDeathTestFactory),
+#endif
+ // Will be overridden by the flag before first use.
+ catch_exceptions_(false) {
+ listeners()->SetDefaultResultPrinter(new PrettyUnitTestResultPrinter);
+}
+
+UnitTestImpl::~UnitTestImpl() {
+ // Deletes every TestCase.
+ ForEach(test_cases_, internal::Delete<TestCase>);
+
+ // Deletes every Environment.
+ ForEach(environments_, internal::Delete<Environment>);
+
+ delete os_stack_trace_getter_;
+}
+
+#if GTEST_HAS_DEATH_TEST
+// Disables event forwarding if the control is currently in a death test
+// subprocess. Must not be called before InitGoogleTest.
+void UnitTestImpl::SuppressTestEventsIfInSubprocess() {
+ if (internal_run_death_test_flag_.get() != NULL)
+ listeners()->SuppressEventForwarding();
+}
+#endif // GTEST_HAS_DEATH_TEST
+
+// Initializes event listeners performing XML output as specified by
+// UnitTestOptions. Must not be called before InitGoogleTest.
+void UnitTestImpl::ConfigureXmlOutput() {
+ const String& output_format = UnitTestOptions::GetOutputFormat();
+ if (output_format == "xml") {
+ listeners()->SetDefaultXmlGenerator(new XmlUnitTestResultPrinter(
+ UnitTestOptions::GetAbsolutePathToOutputFile().c_str()));
+ } else if (output_format != "") {
+ printf("WARNING: unrecognized output format \"%s\" ignored.\n",
+ output_format.c_str());
+ fflush(stdout);
+ }
+}
+
+#if GTEST_CAN_STREAM_RESULTS_
+// Initializes event listeners for streaming test results in String form.
+// Must not be called before InitGoogleTest.
+void UnitTestImpl::ConfigureStreamingOutput() {
+ const string& target = GTEST_FLAG(stream_result_to);
+ if (!target.empty()) {
+ const size_t pos = target.find(':');
+ if (pos != string::npos) {
+ listeners()->Append(new StreamingListener(target.substr(0, pos),
+ target.substr(pos+1)));
+ } else {
+ printf("WARNING: unrecognized streaming target \"%s\" ignored.\n",
+ target.c_str());
+ fflush(stdout);
+ }
+ }
+}
+#endif // GTEST_CAN_STREAM_RESULTS_
+
+// Performs initialization dependent upon flag values obtained in
+// ParseGoogleTestFlagsOnly. Is called from InitGoogleTest after the call to
+// ParseGoogleTestFlagsOnly. In case a user neglects to call InitGoogleTest
+// this function is also called from RunAllTests. Since this function can be
+// called more than once, it has to be idempotent.
+void UnitTestImpl::PostFlagParsingInit() {
+ // Ensures that this function does not execute more than once.
+ if (!post_flag_parse_init_performed_) {
+ post_flag_parse_init_performed_ = true;
+
+#if GTEST_HAS_DEATH_TEST
+ InitDeathTestSubprocessControlInfo();
+ SuppressTestEventsIfInSubprocess();
+#endif // GTEST_HAS_DEATH_TEST
+
+ // Registers parameterized tests. This makes parameterized tests
+ // available to the UnitTest reflection API without running
+ // RUN_ALL_TESTS.
+ RegisterParameterizedTests();
+
+ // Configures listeners for XML output. This makes it possible for users
+ // to shut down the default XML output before invoking RUN_ALL_TESTS.
+ ConfigureXmlOutput();
+
+#if GTEST_CAN_STREAM_RESULTS_
+ // Configures listeners for streaming test results to the specified server.
+ ConfigureStreamingOutput();
+#endif // GTEST_CAN_STREAM_RESULTS_
+ }
+}
+
+// A predicate that checks the name of a TestCase against a known
+// value.
+//
+// This is used for implementation of the UnitTest class only. We put
+// it in the anonymous namespace to prevent polluting the outer
+// namespace.
+//
+// TestCaseNameIs is copyable.
+class TestCaseNameIs {
+ public:
+ // Constructor.
+ explicit TestCaseNameIs(const String& name)
+ : name_(name) {}
+
+ // Returns true iff the name of test_case matches name_.
+ bool operator()(const TestCase* test_case) const {
+ return test_case != NULL && strcmp(test_case->name(), name_.c_str()) == 0;
+ }
+
+ private:
+ String name_;
+};
+
+// Finds and returns a TestCase with the given name. If one doesn't
+// exist, creates one and returns it. It's the CALLER'S
+// RESPONSIBILITY to ensure that this function is only called WHEN THE
+// TESTS ARE NOT SHUFFLED.
+//
+// Arguments:
+//
+// test_case_name: name of the test case
+// type_param: the name of the test case's type parameter, or NULL if
+// this is not a typed or a type-parameterized test case.
+// set_up_tc: pointer to the function that sets up the test case
+// tear_down_tc: pointer to the function that tears down the test case
+TestCase* UnitTestImpl::GetTestCase(const char* test_case_name,
+ const char* type_param,
+ Test::SetUpTestCaseFunc set_up_tc,
+ Test::TearDownTestCaseFunc tear_down_tc) {
+ // Can we find a TestCase with the given name?
+ const std::vector<TestCase*>::const_iterator test_case =
+ std::find_if(test_cases_.begin(), test_cases_.end(),
+ TestCaseNameIs(test_case_name));
+
+ if (test_case != test_cases_.end())
+ return *test_case;
+
+ // No. Let's create one.
+ TestCase* const new_test_case =
+ new TestCase(test_case_name, type_param, set_up_tc, tear_down_tc);
+
+ // Is this a death test case?
+ if (internal::UnitTestOptions::MatchesFilter(String(test_case_name),
+ kDeathTestCaseFilter)) {
+ // Yes. Inserts the test case after the last death test case
+ // defined so far. This only works when the test cases haven't
+ // been shuffled. Otherwise we may end up running a death test
+ // after a non-death test.
+ ++last_death_test_case_;
+ test_cases_.insert(test_cases_.begin() + last_death_test_case_,
+ new_test_case);
+ } else {
+ // No. Appends to the end of the list.
+ test_cases_.push_back(new_test_case);
+ }
+
+ test_case_indices_.push_back(static_cast<int>(test_case_indices_.size()));
+ return new_test_case;
+}
+
+// Helpers for setting up / tearing down the given environment. They
+// are for use in the ForEach() function.
+static void SetUpEnvironment(Environment* env) { env->SetUp(); }
+static void TearDownEnvironment(Environment* env) { env->TearDown(); }
+
+// Runs all tests in this UnitTest object, prints the result, and
+// returns true if all tests are successful. If any exception is
+// thrown during a test, the test is considered to be failed, but the
+// rest of the tests will still be run.
+//
+// When parameterized tests are enabled, it expands and registers
+// parameterized tests first in RegisterParameterizedTests().
+// All other functions called from RunAllTests() may safely assume that
+// parameterized tests are ready to be counted and run.
+bool UnitTestImpl::RunAllTests() {
+ // Makes sure InitGoogleTest() was called.
+ if (!GTestIsInitialized()) {
+ printf("%s",
+ "\nThis test program did NOT call ::testing::InitGoogleTest "
+ "before calling RUN_ALL_TESTS(). Please fix it.\n");
+ return false;
+ }
+
+ // Do not run any test if the --help flag was specified.
+ if (g_help_flag)
+ return true;
+
+ // Repeats the call to the post-flag parsing initialization in case the
+ // user didn't call InitGoogleTest.
+ PostFlagParsingInit();
+
+ // Even if sharding is not on, test runners may want to use the
+ // GTEST_SHARD_STATUS_FILE to query whether the test supports the sharding
+ // protocol.
+ internal::WriteToShardStatusFileIfNeeded();
+
+ // True iff we are in a subprocess for running a thread-safe-style
+ // death test.
+ bool in_subprocess_for_death_test = false;
+
+#if GTEST_HAS_DEATH_TEST
+ in_subprocess_for_death_test = (internal_run_death_test_flag_.get() != NULL);
+#endif // GTEST_HAS_DEATH_TEST
+
+ const bool should_shard = ShouldShard(kTestTotalShards, kTestShardIndex,
+ in_subprocess_for_death_test);
+
+ // Compares the full test names with the filter to decide which
+ // tests to run.
+ const bool has_tests_to_run = FilterTests(should_shard
+ ? HONOR_SHARDING_PROTOCOL
+ : IGNORE_SHARDING_PROTOCOL) > 0;
+
+ // Lists the tests and exits if the --gtest_list_tests flag was specified.
+ if (GTEST_FLAG(list_tests)) {
+ // This must be called *after* FilterTests() has been called.
+ ListTestsMatchingFilter();
+ return true;
+ }
+
+ random_seed_ = GTEST_FLAG(shuffle) ?
+ GetRandomSeedFromFlag(GTEST_FLAG(random_seed)) : 0;
+
+ // True iff at least one test has failed.
+ bool failed = false;
+
+ TestEventListener* repeater = listeners()->repeater();
+
+ repeater->OnTestProgramStart(*parent_);
+
+ // How many times to repeat the tests? We don't want to repeat them
+ // when we are inside the subprocess of a death test.
+ const int repeat = in_subprocess_for_death_test ? 1 : GTEST_FLAG(repeat);
+ // Repeats forever if the repeat count is negative.
+ const bool forever = repeat < 0;
+ for (int i = 0; forever || i != repeat; i++) {
+ // We want to preserve failures generated by ad-hoc test
+ // assertions executed before RUN_ALL_TESTS().
+ ClearNonAdHocTestResult();
+
+ const TimeInMillis start = GetTimeInMillis();
+
+ // Shuffles test cases and tests if requested.
+ if (has_tests_to_run && GTEST_FLAG(shuffle)) {
+ random()->Reseed(random_seed_);
+ // This should be done before calling OnTestIterationStart(),
+ // such that a test event listener can see the actual test order
+ // in the event.
+ ShuffleTests();
+ }
+
+ // Tells the unit test event listeners that the tests are about to start.
+ repeater->OnTestIterationStart(*parent_, i);
+
+ // Runs each test case if there is at least one test to run.
+ if (has_tests_to_run) {
+ // Sets up all environments beforehand.
+ repeater->OnEnvironmentsSetUpStart(*parent_);
+ ForEach(environments_, SetUpEnvironment);
+ repeater->OnEnvironmentsSetUpEnd(*parent_);
+
+ // Runs the tests only if there was no fatal failure during global
+ // set-up.
+ if (!Test::HasFatalFailure()) {
+ for (int test_index = 0; test_index < total_test_case_count();
+ test_index++) {
+ GetMutableTestCase(test_index)->Run();
+ }
+ }
+
+ // Tears down all environments in reverse order afterwards.
+ repeater->OnEnvironmentsTearDownStart(*parent_);
+ std::for_each(environments_.rbegin(), environments_.rend(),
+ TearDownEnvironment);
+ repeater->OnEnvironmentsTearDownEnd(*parent_);
+ }
+
+ elapsed_time_ = GetTimeInMillis() - start;
+
+ // Tells the unit test event listener that the tests have just finished.
+ repeater->OnTestIterationEnd(*parent_, i);
+
+ // Gets the result and clears it.
+ if (!Passed()) {
+ failed = true;
+ }
+
+ // Restores the original test order after the iteration. This
+ // allows the user to quickly repro a failure that happens in the
+ // N-th iteration without repeating the first (N - 1) iterations.
+ // This is not enclosed in "if (GTEST_FLAG(shuffle)) { ... }", in
+ // case the user somehow changes the value of the flag somewhere
+ // (it's always safe to unshuffle the tests).
+ UnshuffleTests();
+
+ if (GTEST_FLAG(shuffle)) {
+ // Picks a new random seed for each iteration.
+ random_seed_ = GetNextRandomSeed(random_seed_);
+ }
+ }
+
+ repeater->OnTestProgramEnd(*parent_);
+
+ return !failed;
+}
+
+// Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file
+// if the variable is present. If a file already exists at this location, this
+// function will write over it. If the variable is present, but the file cannot
+// be created, prints an error and exits.
+void WriteToShardStatusFileIfNeeded() {
+ const char* const test_shard_file = posix::GetEnv(kTestShardStatusFile);
+ if (test_shard_file != NULL) {
+ FILE* const file = posix::FOpen(test_shard_file, "w");
+ if (file == NULL) {
+ ColoredPrintf(COLOR_RED,
+ "Could not write to the test shard status file \"%s\" "
+ "specified by the %s environment variable.\n",
+ test_shard_file, kTestShardStatusFile);
+ fflush(stdout);
+ exit(EXIT_FAILURE);
+ }
+ fclose(file);
+ }
+}
+
+// Checks whether sharding is enabled by examining the relevant
+// environment variable values. If the variables are present,
+// but inconsistent (i.e., shard_index >= total_shards), prints
+// an error and exits. If in_subprocess_for_death_test, sharding is
+// disabled because it must only be applied to the original test
+// process. Otherwise, we could filter out death tests we intended to execute.
+bool ShouldShard(const char* total_shards_env,
+ const char* shard_index_env,
+ bool in_subprocess_for_death_test) {
+ if (in_subprocess_for_death_test) {
+ return false;
+ }
+
+ const Int32 total_shards = Int32FromEnvOrDie(total_shards_env, -1);
+ const Int32 shard_index = Int32FromEnvOrDie(shard_index_env, -1);
+
+ if (total_shards == -1 && shard_index == -1) {
+ return false;
+ } else if (total_shards == -1 && shard_index != -1) {
+ const Message msg = Message()
+ << "Invalid environment variables: you have "
+ << kTestShardIndex << " = " << shard_index
+ << ", but have left " << kTestTotalShards << " unset.\n";
+ ColoredPrintf(COLOR_RED, msg.GetString().c_str());
+ fflush(stdout);
+ exit(EXIT_FAILURE);
+ } else if (total_shards != -1 && shard_index == -1) {
+ const Message msg = Message()
+ << "Invalid environment variables: you have "
+ << kTestTotalShards << " = " << total_shards
+ << ", but have left " << kTestShardIndex << " unset.\n";
+ ColoredPrintf(COLOR_RED, msg.GetString().c_str());
+ fflush(stdout);
+ exit(EXIT_FAILURE);
+ } else if (shard_index < 0 || shard_index >= total_shards) {
+ const Message msg = Message()
+ << "Invalid environment variables: we require 0 <= "
+ << kTestShardIndex << " < " << kTestTotalShards
+ << ", but you have " << kTestShardIndex << "=" << shard_index
+ << ", " << kTestTotalShards << "=" << total_shards << ".\n";
+ ColoredPrintf(COLOR_RED, msg.GetString().c_str());
+ fflush(stdout);
+ exit(EXIT_FAILURE);
+ }
+
+ return total_shards > 1;
+}
+
+// Parses the environment variable var as an Int32. If it is unset,
+// returns default_val. If it is not an Int32, prints an error
+// and aborts.
+Int32 Int32FromEnvOrDie(const char* var, Int32 default_val) {
+ const char* str_val = posix::GetEnv(var);
+ if (str_val == NULL) {
+ return default_val;
+ }
+
+ Int32 result;
+ if (!ParseInt32(Message() << "The value of environment variable " << var,
+ str_val, &result)) {
+ exit(EXIT_FAILURE);
+ }
+ return result;
+}
+
+// Given the total number of shards, the shard index, and the test id,
+// returns true iff the test should be run on this shard. The test id is
+// some arbitrary but unique non-negative integer assigned to each test
+// method. Assumes that 0 <= shard_index < total_shards.
+bool ShouldRunTestOnShard(int total_shards, int shard_index, int test_id) {
+ return (test_id % total_shards) == shard_index;
+}
+
+// Compares the name of each test with the user-specified filter to
+// decide whether the test should be run, then records the result in
+// each TestCase and TestInfo object.
+// If shard_tests == true, further filters tests based on sharding
+// variables in the environment - see
+// http://code.google.com/p/googletest/wiki/GoogleTestAdvancedGuide.
+// Returns the number of tests that should run.
+int UnitTestImpl::FilterTests(ReactionToSharding shard_tests) {
+ const Int32 total_shards = shard_tests == HONOR_SHARDING_PROTOCOL ?
+ Int32FromEnvOrDie(kTestTotalShards, -1) : -1;
+ const Int32 shard_index = shard_tests == HONOR_SHARDING_PROTOCOL ?
+ Int32FromEnvOrDie(kTestShardIndex, -1) : -1;
+
+ // num_runnable_tests are the number of tests that will
+ // run across all shards (i.e., match filter and are not disabled).
+ // num_selected_tests are the number of tests to be run on
+ // this shard.
+ int num_runnable_tests = 0;
+ int num_selected_tests = 0;
+ for (size_t i = 0; i < test_cases_.size(); i++) {
+ TestCase* const test_case = test_cases_[i];
+ const String &test_case_name = test_case->name();
+ test_case->set_should_run(false);
+
+ for (size_t j = 0; j < test_case->test_info_list().size(); j++) {
+ TestInfo* const test_info = test_case->test_info_list()[j];
+ const String test_name(test_info->name());
+ // A test is disabled if test case name or test name matches
+ // kDisableTestFilter.
+ const bool is_disabled =
+ internal::UnitTestOptions::MatchesFilter(test_case_name,
+ kDisableTestFilter) ||
+ internal::UnitTestOptions::MatchesFilter(test_name,
+ kDisableTestFilter);
+ test_info->is_disabled_ = is_disabled;
+
+ const bool matches_filter =
+ internal::UnitTestOptions::FilterMatchesTest(test_case_name,
+ test_name);
+ test_info->matches_filter_ = matches_filter;
+
+ const bool is_runnable =
+ (GTEST_FLAG(also_run_disabled_tests) || !is_disabled) &&
+ matches_filter;
+
+ const bool is_selected = is_runnable &&
+ (shard_tests == IGNORE_SHARDING_PROTOCOL ||
+ ShouldRunTestOnShard(total_shards, shard_index,
+ num_runnable_tests));
+
+ num_runnable_tests += is_runnable;
+ num_selected_tests += is_selected;
+
+ test_info->should_run_ = is_selected;
+ test_case->set_should_run(test_case->should_run() || is_selected);
+ }
+ }
+ return num_selected_tests;
+}
+
+// Prints the names of the tests matching the user-specified filter flag.
+void UnitTestImpl::ListTestsMatchingFilter() {
+ for (size_t i = 0; i < test_cases_.size(); i++) {
+ const TestCase* const test_case = test_cases_[i];
+ bool printed_test_case_name = false;
+
+ for (size_t j = 0; j < test_case->test_info_list().size(); j++) {
+ const TestInfo* const test_info =
+ test_case->test_info_list()[j];
+ if (test_info->matches_filter_) {
+ if (!printed_test_case_name) {
+ printed_test_case_name = true;
+ printf("%s.\n", test_case->name());
+ }
+ printf(" %s\n", test_info->name());
+ }
+ }
+ }
+ fflush(stdout);
+}
+
+// Sets the OS stack trace getter.
+//
+// Does nothing if the input and the current OS stack trace getter are
+// the same; otherwise, deletes the old getter and makes the input the
+// current getter.
+void UnitTestImpl::set_os_stack_trace_getter(
+ OsStackTraceGetterInterface* getter) {
+ if (os_stack_trace_getter_ != getter) {
+ delete os_stack_trace_getter_;
+ os_stack_trace_getter_ = getter;
+ }
+}
+
+// Returns the current OS stack trace getter if it is not NULL;
+// otherwise, creates an OsStackTraceGetter, makes it the current
+// getter, and returns it.
+OsStackTraceGetterInterface* UnitTestImpl::os_stack_trace_getter() {
+ if (os_stack_trace_getter_ == NULL) {
+ os_stack_trace_getter_ = new OsStackTraceGetter;
+ }
+
+ return os_stack_trace_getter_;
+}
+
+// Returns the TestResult for the test that's currently running, or
+// the TestResult for the ad hoc test if no test is running.
+TestResult* UnitTestImpl::current_test_result() {
+ return current_test_info_ ?
+ &(current_test_info_->result_) : &ad_hoc_test_result_;
+}
+
+// Shuffles all test cases, and the tests within each test case,
+// making sure that death tests are still run first.
+void UnitTestImpl::ShuffleTests() {
+ // Shuffles the death test cases.
+ ShuffleRange(random(), 0, last_death_test_case_ + 1, &test_case_indices_);
+
+ // Shuffles the non-death test cases.
+ ShuffleRange(random(), last_death_test_case_ + 1,
+ static_cast<int>(test_cases_.size()), &test_case_indices_);
+
+ // Shuffles the tests inside each test case.
+ for (size_t i = 0; i < test_cases_.size(); i++) {
+ test_cases_[i]->ShuffleTests(random());
+ }
+}
+
+// Restores the test cases and tests to their order before the first shuffle.
+void UnitTestImpl::UnshuffleTests() {
+ for (size_t i = 0; i < test_cases_.size(); i++) {
+ // Unshuffles the tests in each test case.
+ test_cases_[i]->UnshuffleTests();
+ // Resets the index of each test case.
+ test_case_indices_[i] = static_cast<int>(i);
+ }
+}
+
+// Returns the current OS stack trace as a String.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag. The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in
+// the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't.
+String GetCurrentOsStackTraceExceptTop(UnitTest* /*unit_test*/,
+ int skip_count) {
+ // We pass skip_count + 1 to skip this wrapper function in addition
+ // to what the user really wants to skip.
+ return GetUnitTestImpl()->CurrentOsStackTraceExceptTop(skip_count + 1);
+}
+
+// Used by the GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_ macro to
+// suppress unreachable code warnings.
+namespace {
+class ClassUniqueToAlwaysTrue {};
+}
+
+bool IsTrue(bool condition) { return condition; }
+
+bool AlwaysTrue() {
+#if GTEST_HAS_EXCEPTIONS
+ // This condition is always false so AlwaysTrue() never actually throws,
+ // but it makes the compiler think that it may throw.
+ if (IsTrue(false))
+ throw ClassUniqueToAlwaysTrue();
+#endif // GTEST_HAS_EXCEPTIONS
+ return true;
+}
+
+// If *pstr starts with the given prefix, modifies *pstr to be right
+// past the prefix and returns true; otherwise leaves *pstr unchanged
+// and returns false. None of pstr, *pstr, and prefix can be NULL.
+bool SkipPrefix(const char* prefix, const char** pstr) {
+ const size_t prefix_len = strlen(prefix);
+ if (strncmp(*pstr, prefix, prefix_len) == 0) {
+ *pstr += prefix_len;
+ return true;
+ }
+ return false;
+}
+
+// Parses a string as a command line flag. The string should have
+// the format "--flag=value". When def_optional is true, the "=value"
+// part can be omitted.
+//
+// Returns the value of the flag, or NULL if the parsing failed.
+const char* ParseFlagValue(const char* str,
+ const char* flag,
+ bool def_optional) {
+ // str and flag must not be NULL.
+ if (str == NULL || flag == NULL) return NULL;
+
+ // The flag must start with "--" followed by GTEST_FLAG_PREFIX_.
+ const String flag_str = String::Format("--%s%s", GTEST_FLAG_PREFIX_, flag);
+ const size_t flag_len = flag_str.length();
+ if (strncmp(str, flag_str.c_str(), flag_len) != 0) return NULL;
+
+ // Skips the flag name.
+ const char* flag_end = str + flag_len;
+
+ // When def_optional is true, it's OK to not have a "=value" part.
+ if (def_optional && (flag_end[0] == '\0')) {
+ return flag_end;
+ }
+
+ // If def_optional is true and there are more characters after the
+ // flag name, or if def_optional is false, there must be a '=' after
+ // the flag name.
+ if (flag_end[0] != '=') return NULL;
+
+ // Returns the string after "=".
+ return flag_end + 1;
+}
+
+// Parses a string for a bool flag, in the form of either
+// "--flag=value" or "--flag".
+//
+// In the former case, the value is taken as true as long as it does
+// not start with '0', 'f', or 'F'.
+//
+// In the latter case, the value is taken as true.
+//
+// On success, stores the value of the flag in *value, and returns
+// true. On failure, returns false without changing *value.
+bool ParseBoolFlag(const char* str, const char* flag, bool* value) {
+ // Gets the value of the flag as a string.
+ const char* const value_str = ParseFlagValue(str, flag, true);
+
+ // Aborts if the parsing failed.
+ if (value_str == NULL) return false;
+
+ // Converts the string value to a bool.
+ *value = !(*value_str == '0' || *value_str == 'f' || *value_str == 'F');
+ return true;
+}
+
+// Parses a string for an Int32 flag, in the form of
+// "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true. On failure, returns false without changing *value.
+bool ParseInt32Flag(const char* str, const char* flag, Int32* value) {
+ // Gets the value of the flag as a string.
+ const char* const value_str = ParseFlagValue(str, flag, false);
+
+ // Aborts if the parsing failed.
+ if (value_str == NULL) return false;
+
+ // Sets *value to the value of the flag.
+ return ParseInt32(Message() << "The value of flag --" << flag,
+ value_str, value);
+}
+
+// Parses a string for a string flag, in the form of
+// "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true. On failure, returns false without changing *value.
+bool ParseStringFlag(const char* str, const char* flag, String* value) {
+ // Gets the value of the flag as a string.
+ const char* const value_str = ParseFlagValue(str, flag, false);
+
+ // Aborts if the parsing failed.
+ if (value_str == NULL) return false;
+
+ // Sets *value to the value of the flag.
+ *value = value_str;
+ return true;
+}
+
+// Determines whether a string has a prefix that Google Test uses for its
+// flags, i.e., starts with GTEST_FLAG_PREFIX_ or GTEST_FLAG_PREFIX_DASH_.
+// If Google Test detects that a command line flag has its prefix but is not
+// recognized, it will print its help message. Flags starting with
+// GTEST_INTERNAL_PREFIX_ followed by "internal_" are considered Google Test
+// internal flags and do not trigger the help message.
+static bool HasGoogleTestFlagPrefix(const char* str) {
+ return (SkipPrefix("--", &str) ||
+ SkipPrefix("-", &str) ||
+ SkipPrefix("/", &str)) &&
+ !SkipPrefix(GTEST_FLAG_PREFIX_ "internal_", &str) &&
+ (SkipPrefix(GTEST_FLAG_PREFIX_, &str) ||
+ SkipPrefix(GTEST_FLAG_PREFIX_DASH_, &str));
+}
+
+// Prints a string containing code-encoded text. The following escape
+// sequences can be used in the string to control the text color:
+//
+// @@ prints a single '@' character.
+// @R changes the color to red.
+// @G changes the color to green.
+// @Y changes the color to yellow.
+// @D changes to the default terminal text color.
+//
+// TODO(wan@google.com): Write tests for this once we add stdout
+// capturing to Google Test.
+static void PrintColorEncoded(const char* str) {
+ GTestColor color = COLOR_DEFAULT; // The current color.
+
+ // Conceptually, we split the string into segments divided by escape
+ // sequences. Then we print one segment at a time. At the end of
+ // each iteration, the str pointer advances to the beginning of the
+ // next segment.
+ for (;;) {
+ const char* p = strchr(str, '@');
+ if (p == NULL) {
+ ColoredPrintf(color, "%s", str);
+ return;
+ }
+
+ ColoredPrintf(color, "%s", String(str, p - str).c_str());
+
+ const char ch = p[1];
+ str = p + 2;
+ if (ch == '@') {
+ ColoredPrintf(color, "@");
+ } else if (ch == 'D') {
+ color = COLOR_DEFAULT;
+ } else if (ch == 'R') {
+ color = COLOR_RED;
+ } else if (ch == 'G') {
+ color = COLOR_GREEN;
+ } else if (ch == 'Y') {
+ color = COLOR_YELLOW;
+ } else {
+ --str;
+ }
+ }
+}
+
+static const char kColorEncodedHelpMessage[] =
+"This program contains tests written using " GTEST_NAME_ ". You can use the\n"
+"following command line flags to control its behavior:\n"
+"\n"
+"Test Selection:\n"
+" @G--" GTEST_FLAG_PREFIX_ "list_tests@D\n"
+" List the names of all tests instead of running them. The name of\n"
+" TEST(Foo, Bar) is \"Foo.Bar\".\n"
+" @G--" GTEST_FLAG_PREFIX_ "filter=@YPOSTIVE_PATTERNS"
+ "[@G-@YNEGATIVE_PATTERNS]@D\n"
+" Run only the tests whose name matches one of the positive patterns but\n"
+" none of the negative patterns. '?' matches any single character; '*'\n"
+" matches any substring; ':' separates two patterns.\n"
+" @G--" GTEST_FLAG_PREFIX_ "also_run_disabled_tests@D\n"
+" Run all disabled tests too.\n"
+"\n"
+"Test Execution:\n"
+" @G--" GTEST_FLAG_PREFIX_ "repeat=@Y[COUNT]@D\n"
+" Run the tests repeatedly; use a negative count to repeat forever.\n"
+" @G--" GTEST_FLAG_PREFIX_ "shuffle@D\n"
+" Randomize tests' orders on every iteration.\n"
+" @G--" GTEST_FLAG_PREFIX_ "random_seed=@Y[NUMBER]@D\n"
+" Random number seed to use for shuffling test orders (between 1 and\n"
+" 99999, or 0 to use a seed based on the current time).\n"
+"\n"
+"Test Output:\n"
+" @G--" GTEST_FLAG_PREFIX_ "color=@Y(@Gyes@Y|@Gno@Y|@Gauto@Y)@D\n"
+" Enable/disable colored output. The default is @Gauto@D.\n"
+" -@G-" GTEST_FLAG_PREFIX_ "print_time=0@D\n"
+" Don't print the elapsed time of each test.\n"
+" @G--" GTEST_FLAG_PREFIX_ "output=xml@Y[@G:@YDIRECTORY_PATH@G"
+ GTEST_PATH_SEP_ "@Y|@G:@YFILE_PATH]@D\n"
+" Generate an XML report in the given directory or with the given file\n"
+" name. @YFILE_PATH@D defaults to @Gtest_details.xml@D.\n"
+#if GTEST_CAN_STREAM_RESULTS_
+" @G--" GTEST_FLAG_PREFIX_ "stream_result_to=@YHOST@G:@YPORT@D\n"
+" Stream test results to the given server.\n"
+#endif // GTEST_CAN_STREAM_RESULTS_
+"\n"
+"Assertion Behavior:\n"
+#if GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS
+" @G--" GTEST_FLAG_PREFIX_ "death_test_style=@Y(@Gfast@Y|@Gthreadsafe@Y)@D\n"
+" Set the default death test style.\n"
+#endif // GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS
+" @G--" GTEST_FLAG_PREFIX_ "break_on_failure@D\n"
+" Turn assertion failures into debugger break-points.\n"
+" @G--" GTEST_FLAG_PREFIX_ "throw_on_failure@D\n"
+" Turn assertion failures into C++ exceptions.\n"
+" @G--" GTEST_FLAG_PREFIX_ "catch_exceptions=0@D\n"
+" Do not report exceptions as test failures. Instead, allow them\n"
+" to crash the program or throw a pop-up (on Windows).\n"
+"\n"
+"Except for @G--" GTEST_FLAG_PREFIX_ "list_tests@D, you can alternatively set "
+ "the corresponding\n"
+"environment variable of a flag (all letters in upper-case). For example, to\n"
+"disable colored text output, you can either specify @G--" GTEST_FLAG_PREFIX_
+ "color=no@D or set\n"
+"the @G" GTEST_FLAG_PREFIX_UPPER_ "COLOR@D environment variable to @Gno@D.\n"
+"\n"
+"For more information, please read the " GTEST_NAME_ " documentation at\n"
+"@G" GTEST_PROJECT_URL_ "@D. If you find a bug in " GTEST_NAME_ "\n"
+"(not one in your own code or tests), please report it to\n"
+"@G<" GTEST_DEV_EMAIL_ ">@D.\n";
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test. The type parameter CharType can be
+// instantiated to either char or wchar_t.
+template <typename CharType>
+void ParseGoogleTestFlagsOnlyImpl(int* argc, CharType** argv) {
+ for (int i = 1; i < *argc; i++) {
+ const String arg_string = StreamableToString(argv[i]);
+ const char* const arg = arg_string.c_str();
+
+ using internal::ParseBoolFlag;
+ using internal::ParseInt32Flag;
+ using internal::ParseStringFlag;
+
+ // Do we see a Google Test flag?
+ if (ParseBoolFlag(arg, kAlsoRunDisabledTestsFlag,
+ >EST_FLAG(also_run_disabled_tests)) ||
+ ParseBoolFlag(arg, kBreakOnFailureFlag,
+ >EST_FLAG(break_on_failure)) ||
+ ParseBoolFlag(arg, kCatchExceptionsFlag,
+ >EST_FLAG(catch_exceptions)) ||
+ ParseStringFlag(arg, kColorFlag, >EST_FLAG(color)) ||
+ ParseStringFlag(arg, kDeathTestStyleFlag,
+ >EST_FLAG(death_test_style)) ||
+ ParseBoolFlag(arg, kDeathTestUseFork,
+ >EST_FLAG(death_test_use_fork)) ||
+ ParseStringFlag(arg, kFilterFlag, >EST_FLAG(filter)) ||
+ ParseStringFlag(arg, kInternalRunDeathTestFlag,
+ >EST_FLAG(internal_run_death_test)) ||
+ ParseBoolFlag(arg, kListTestsFlag, >EST_FLAG(list_tests)) ||
+ ParseStringFlag(arg, kOutputFlag, >EST_FLAG(output)) ||
+ ParseBoolFlag(arg, kPrintTimeFlag, >EST_FLAG(print_time)) ||
+ ParseInt32Flag(arg, kRandomSeedFlag, >EST_FLAG(random_seed)) ||
+ ParseInt32Flag(arg, kRepeatFlag, >EST_FLAG(repeat)) ||
+ ParseBoolFlag(arg, kShuffleFlag, >EST_FLAG(shuffle)) ||
+ ParseInt32Flag(arg, kStackTraceDepthFlag,
+ >EST_FLAG(stack_trace_depth)) ||
+ ParseStringFlag(arg, kStreamResultToFlag,
+ >EST_FLAG(stream_result_to)) ||
+ ParseBoolFlag(arg, kThrowOnFailureFlag,
+ >EST_FLAG(throw_on_failure))
+ ) {
+ // Yes. Shift the remainder of the argv list left by one. Note
+ // that argv has (*argc + 1) elements, the last one always being
+ // NULL. The following loop moves the trailing NULL element as
+ // well.
+ for (int j = i; j != *argc; j++) {
+ argv[j] = argv[j + 1];
+ }
+
+ // Decrements the argument count.
+ (*argc)--;
+
+ // We also need to decrement the iterator as we just removed
+ // an element.
+ i--;
+ } else if (arg_string == "--help" || arg_string == "-h" ||
+ arg_string == "-?" || arg_string == "/?" ||
+ HasGoogleTestFlagPrefix(arg)) {
+ // Both help flag and unrecognized Google Test flags (excluding
+ // internal ones) trigger help display.
+ g_help_flag = true;
+ }
+ }
+
+ if (g_help_flag) {
+ // We print the help here instead of in RUN_ALL_TESTS(), as the
+ // latter may not be called at all if the user is using Google
+ // Test with another testing framework.
+ PrintColorEncoded(kColorEncodedHelpMessage);
+ }
+}
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test.
+void ParseGoogleTestFlagsOnly(int* argc, char** argv) {
+ ParseGoogleTestFlagsOnlyImpl(argc, argv);
+}
+void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv) {
+ ParseGoogleTestFlagsOnlyImpl(argc, argv);
+}
+
+// The internal implementation of InitGoogleTest().
+//
+// The type parameter CharType can be instantiated to either char or
+// wchar_t.
+template <typename CharType>
+void InitGoogleTestImpl(int* argc, CharType** argv) {
+ g_init_gtest_count++;
+
+ // We don't want to run the initialization code twice.
+ if (g_init_gtest_count != 1) return;
+
+ if (*argc <= 0) return;
+
+ internal::g_executable_path = internal::StreamableToString(argv[0]);
+
+#if GTEST_HAS_DEATH_TEST
+
+ g_argvs.clear();
+ for (int i = 0; i != *argc; i++) {
+ g_argvs.push_back(StreamableToString(argv[i]));
+ }
+
+#endif // GTEST_HAS_DEATH_TEST
+
+ ParseGoogleTestFlagsOnly(argc, argv);
+ GetUnitTestImpl()->PostFlagParsingInit();
+}
+
+} // namespace internal
+
+// Initializes Google Test. This must be called before calling
+// RUN_ALL_TESTS(). In particular, it parses a command line for the
+// flags that Google Test recognizes. Whenever a Google Test flag is
+// seen, it is removed from argv, and *argc is decremented.
+//
+// No value is returned. Instead, the Google Test flag variables are
+// updated.
+//
+// Calling the function for the second time has no user-visible effect.
+void InitGoogleTest(int* argc, char** argv) {
+ internal::InitGoogleTestImpl(argc, argv);
+}
+
+// This overloaded version can be used in Windows programs compiled in
+// UNICODE mode.
+void InitGoogleTest(int* argc, wchar_t** argv) {
+ internal::InitGoogleTestImpl(argc, argv);
+}
+
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/src/gtest_main.cc b/libvpx/third_party/googletest/src/src/gtest_main.cc
new file mode 100644
index 0000000..a09bbe0
--- /dev/null
+++ b/libvpx/third_party/googletest/src/src/gtest_main.cc
@@ -0,0 +1,39 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#include <iostream>
+
+#include "gtest/gtest.h"
+
+GTEST_API_ int main(int argc, char **argv) {
+ std::cout << "Running main() from gtest_main.cc\n";
+
+ testing::InitGoogleTest(&argc, argv);
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest-death-test_ex_test.cc b/libvpx/third_party/googletest/src/test/gtest-death-test_ex_test.cc
new file mode 100644
index 0000000..b50a13d
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-death-test_ex_test.cc
@@ -0,0 +1,93 @@
+// Copyright 2010, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+//
+// Tests that verify interaction of exceptions and death tests.
+
+#include "gtest/gtest-death-test.h"
+#include "gtest/gtest.h"
+
+#if GTEST_HAS_DEATH_TEST
+
+# if GTEST_HAS_SEH
+# include <windows.h> // For RaiseException().
+# endif
+
+# include "gtest/gtest-spi.h"
+
+# if GTEST_HAS_EXCEPTIONS
+
+# include <exception> // For std::exception.
+
+// Tests that death tests report thrown exceptions as failures and that the
+// exceptions do not escape death test macros.
+TEST(CxxExceptionDeathTest, ExceptionIsFailure) {
+ try {
+ EXPECT_NONFATAL_FAILURE(EXPECT_DEATH(throw 1, ""), "threw an exception");
+ } catch (...) { // NOLINT
+ FAIL() << "An exception escaped a death test macro invocation "
+ << "with catch_exceptions "
+ << (testing::GTEST_FLAG(catch_exceptions) ? "enabled" : "disabled");
+ }
+}
+
+class TestException : public std::exception {
+ public:
+ virtual const char* what() const throw() { return "exceptional message"; }
+};
+
+TEST(CxxExceptionDeathTest, PrintsMessageForStdExceptions) {
+ // Verifies that the exception message is quoted in the failure text.
+ EXPECT_NONFATAL_FAILURE(EXPECT_DEATH(throw TestException(), ""),
+ "exceptional message");
+ // Verifies that the location is mentioned in the failure text.
+ EXPECT_NONFATAL_FAILURE(EXPECT_DEATH(throw TestException(), ""),
+ "gtest-death-test_ex_test.cc");
+}
+# endif // GTEST_HAS_EXCEPTIONS
+
+# if GTEST_HAS_SEH
+// Tests that enabling interception of SEH exceptions with the
+// catch_exceptions flag does not interfere with SEH exceptions being
+// treated as death by death tests.
+TEST(SehExceptionDeasTest, CatchExceptionsDoesNotInterfere) {
+ EXPECT_DEATH(RaiseException(42, 0x0, 0, NULL), "")
+ << "with catch_exceptions "
+ << (testing::GTEST_FLAG(catch_exceptions) ? "enabled" : "disabled");
+}
+# endif
+
+#endif // GTEST_HAS_DEATH_TEST
+
+int main(int argc, char** argv) {
+ testing::InitGoogleTest(&argc, argv);
+ testing::GTEST_FLAG(catch_exceptions) = GTEST_ENABLE_CATCH_EXCEPTIONS_ != 0;
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest-death-test_test.cc b/libvpx/third_party/googletest/src/test/gtest-death-test_test.cc
new file mode 100644
index 0000000..bcf8e2a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-death-test_test.cc
@@ -0,0 +1,1296 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Tests for death tests.
+
+#include "gtest/gtest-death-test.h"
+#include "gtest/gtest.h"
+#include "gtest/internal/gtest-filepath.h"
+
+using testing::internal::AlwaysFalse;
+using testing::internal::AlwaysTrue;
+
+#if GTEST_HAS_DEATH_TEST
+
+# if GTEST_OS_WINDOWS
+# include <direct.h> // For chdir().
+# else
+# include <unistd.h>
+# include <sys/wait.h> // For waitpid.
+# include <limits> // For std::numeric_limits.
+# endif // GTEST_OS_WINDOWS
+
+# include <limits.h>
+# include <signal.h>
+# include <stdio.h>
+
+# include "gtest/gtest-spi.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+# define GTEST_IMPLEMENTATION_ 1
+# include "src/gtest-internal-inl.h"
+# undef GTEST_IMPLEMENTATION_
+
+namespace posix = ::testing::internal::posix;
+
+using testing::Message;
+using testing::internal::DeathTest;
+using testing::internal::DeathTestFactory;
+using testing::internal::FilePath;
+using testing::internal::GetLastErrnoDescription;
+using testing::internal::GetUnitTestImpl;
+using testing::internal::ParseNaturalNumber;
+using testing::internal::String;
+
+namespace testing {
+namespace internal {
+
+// A helper class whose objects replace the death test factory for a
+// single UnitTest object during their lifetimes.
+class ReplaceDeathTestFactory {
+ public:
+ explicit ReplaceDeathTestFactory(DeathTestFactory* new_factory)
+ : unit_test_impl_(GetUnitTestImpl()) {
+ old_factory_ = unit_test_impl_->death_test_factory_.release();
+ unit_test_impl_->death_test_factory_.reset(new_factory);
+ }
+
+ ~ReplaceDeathTestFactory() {
+ unit_test_impl_->death_test_factory_.release();
+ unit_test_impl_->death_test_factory_.reset(old_factory_);
+ }
+ private:
+ // Prevents copying ReplaceDeathTestFactory objects.
+ ReplaceDeathTestFactory(const ReplaceDeathTestFactory&);
+ void operator=(const ReplaceDeathTestFactory&);
+
+ UnitTestImpl* unit_test_impl_;
+ DeathTestFactory* old_factory_;
+};
+
+} // namespace internal
+} // namespace testing
+
+void DieWithMessage(const ::std::string& message) {
+ fprintf(stderr, "%s", message.c_str());
+ fflush(stderr); // Make sure the text is printed before the process exits.
+
+ // We call _exit() instead of exit(), as the former is a direct
+ // system call and thus safer in the presence of threads. exit()
+ // will invoke user-defined exit-hooks, which may do dangerous
+ // things that conflict with death tests.
+ //
+ // Some compilers can recognize that _exit() never returns and issue the
+ // 'unreachable code' warning for code following this function, unless
+ // fooled by a fake condition.
+ if (AlwaysTrue())
+ _exit(1);
+}
+
+void DieInside(const ::std::string& function) {
+ DieWithMessage("death inside " + function + "().");
+}
+
+// Tests that death tests work.
+
+class TestForDeathTest : public testing::Test {
+ protected:
+ TestForDeathTest() : original_dir_(FilePath::GetCurrentDir()) {}
+
+ virtual ~TestForDeathTest() {
+ posix::ChDir(original_dir_.c_str());
+ }
+
+ // A static member function that's expected to die.
+ static void StaticMemberFunction() { DieInside("StaticMemberFunction"); }
+
+ // A method of the test fixture that may die.
+ void MemberFunction() {
+ if (should_die_)
+ DieInside("MemberFunction");
+ }
+
+ // True iff MemberFunction() should die.
+ bool should_die_;
+ const FilePath original_dir_;
+};
+
+// A class with a member function that may die.
+class MayDie {
+ public:
+ explicit MayDie(bool should_die) : should_die_(should_die) {}
+
+ // A member function that may die.
+ void MemberFunction() const {
+ if (should_die_)
+ DieInside("MayDie::MemberFunction");
+ }
+
+ private:
+ // True iff MemberFunction() should die.
+ bool should_die_;
+};
+
+// A global function that's expected to die.
+void GlobalFunction() { DieInside("GlobalFunction"); }
+
+// A non-void function that's expected to die.
+int NonVoidFunction() {
+ DieInside("NonVoidFunction");
+ return 1;
+}
+
+// A unary function that may die.
+void DieIf(bool should_die) {
+ if (should_die)
+ DieInside("DieIf");
+}
+
+// A binary function that may die.
+bool DieIfLessThan(int x, int y) {
+ if (x < y) {
+ DieInside("DieIfLessThan");
+ }
+ return true;
+}
+
+// Tests that ASSERT_DEATH can be used outside a TEST, TEST_F, or test fixture.
+void DeathTestSubroutine() {
+ EXPECT_DEATH(GlobalFunction(), "death.*GlobalFunction");
+ ASSERT_DEATH(GlobalFunction(), "death.*GlobalFunction");
+}
+
+// Death in dbg, not opt.
+int DieInDebugElse12(int* sideeffect) {
+ if (sideeffect) *sideeffect = 12;
+
+# ifndef NDEBUG
+
+ DieInside("DieInDebugElse12");
+
+# endif // NDEBUG
+
+ return 12;
+}
+
+# if GTEST_OS_WINDOWS
+
+// Tests the ExitedWithCode predicate.
+TEST(ExitStatusPredicateTest, ExitedWithCode) {
+ // On Windows, the process's exit code is the same as its exit status,
+ // so the predicate just compares the its input with its parameter.
+ EXPECT_TRUE(testing::ExitedWithCode(0)(0));
+ EXPECT_TRUE(testing::ExitedWithCode(1)(1));
+ EXPECT_TRUE(testing::ExitedWithCode(42)(42));
+ EXPECT_FALSE(testing::ExitedWithCode(0)(1));
+ EXPECT_FALSE(testing::ExitedWithCode(1)(0));
+}
+
+# else
+
+// Returns the exit status of a process that calls _exit(2) with a
+// given exit code. This is a helper function for the
+// ExitStatusPredicateTest test suite.
+static int NormalExitStatus(int exit_code) {
+ pid_t child_pid = fork();
+ if (child_pid == 0) {
+ _exit(exit_code);
+ }
+ int status;
+ waitpid(child_pid, &status, 0);
+ return status;
+}
+
+// Returns the exit status of a process that raises a given signal.
+// If the signal does not cause the process to die, then it returns
+// instead the exit status of a process that exits normally with exit
+// code 1. This is a helper function for the ExitStatusPredicateTest
+// test suite.
+static int KilledExitStatus(int signum) {
+ pid_t child_pid = fork();
+ if (child_pid == 0) {
+ raise(signum);
+ _exit(1);
+ }
+ int status;
+ waitpid(child_pid, &status, 0);
+ return status;
+}
+
+// Tests the ExitedWithCode predicate.
+TEST(ExitStatusPredicateTest, ExitedWithCode) {
+ const int status0 = NormalExitStatus(0);
+ const int status1 = NormalExitStatus(1);
+ const int status42 = NormalExitStatus(42);
+ const testing::ExitedWithCode pred0(0);
+ const testing::ExitedWithCode pred1(1);
+ const testing::ExitedWithCode pred42(42);
+ EXPECT_PRED1(pred0, status0);
+ EXPECT_PRED1(pred1, status1);
+ EXPECT_PRED1(pred42, status42);
+ EXPECT_FALSE(pred0(status1));
+ EXPECT_FALSE(pred42(status0));
+ EXPECT_FALSE(pred1(status42));
+}
+
+// Tests the KilledBySignal predicate.
+TEST(ExitStatusPredicateTest, KilledBySignal) {
+ const int status_segv = KilledExitStatus(SIGSEGV);
+ const int status_kill = KilledExitStatus(SIGKILL);
+ const testing::KilledBySignal pred_segv(SIGSEGV);
+ const testing::KilledBySignal pred_kill(SIGKILL);
+ EXPECT_PRED1(pred_segv, status_segv);
+ EXPECT_PRED1(pred_kill, status_kill);
+ EXPECT_FALSE(pred_segv(status_kill));
+ EXPECT_FALSE(pred_kill(status_segv));
+}
+
+# endif // GTEST_OS_WINDOWS
+
+// Tests that the death test macros expand to code which may or may not
+// be followed by operator<<, and that in either case the complete text
+// comprises only a single C++ statement.
+TEST_F(TestForDeathTest, SingleStatement) {
+ if (AlwaysFalse())
+ // This would fail if executed; this is a compilation test only
+ ASSERT_DEATH(return, "");
+
+ if (AlwaysTrue())
+ EXPECT_DEATH(_exit(1), "");
+ else
+ // This empty "else" branch is meant to ensure that EXPECT_DEATH
+ // doesn't expand into an "if" statement without an "else"
+ ;
+
+ if (AlwaysFalse())
+ ASSERT_DEATH(return, "") << "did not die";
+
+ if (AlwaysFalse())
+ ;
+ else
+ EXPECT_DEATH(_exit(1), "") << 1 << 2 << 3;
+}
+
+void DieWithEmbeddedNul() {
+ fprintf(stderr, "Hello%cmy null world.\n", '\0');
+ fflush(stderr);
+ _exit(1);
+}
+
+# if GTEST_USES_PCRE
+// Tests that EXPECT_DEATH and ASSERT_DEATH work when the error
+// message has a NUL character in it.
+TEST_F(TestForDeathTest, EmbeddedNulInMessage) {
+ // TODO(wan@google.com): <regex.h> doesn't support matching strings
+ // with embedded NUL characters - find a way to workaround it.
+ EXPECT_DEATH(DieWithEmbeddedNul(), "my null world");
+ ASSERT_DEATH(DieWithEmbeddedNul(), "my null world");
+}
+# endif // GTEST_USES_PCRE
+
+// Tests that death test macros expand to code which interacts well with switch
+// statements.
+TEST_F(TestForDeathTest, SwitchStatement) {
+// Microsoft compiler usually complains about switch statements without
+// case labels. We suppress that warning for this test.
+# ifdef _MSC_VER
+# pragma warning(push)
+# pragma warning(disable: 4065)
+# endif // _MSC_VER
+
+ switch (0)
+ default:
+ ASSERT_DEATH(_exit(1), "") << "exit in default switch handler";
+
+ switch (0)
+ case 0:
+ EXPECT_DEATH(_exit(1), "") << "exit in switch case";
+
+# ifdef _MSC_VER
+# pragma warning(pop)
+# endif // _MSC_VER
+}
+
+// Tests that a static member function can be used in a "fast" style
+// death test.
+TEST_F(TestForDeathTest, StaticMemberFunctionFastStyle) {
+ testing::GTEST_FLAG(death_test_style) = "fast";
+ ASSERT_DEATH(StaticMemberFunction(), "death.*StaticMember");
+}
+
+// Tests that a method of the test fixture can be used in a "fast"
+// style death test.
+TEST_F(TestForDeathTest, MemberFunctionFastStyle) {
+ testing::GTEST_FLAG(death_test_style) = "fast";
+ should_die_ = true;
+ EXPECT_DEATH(MemberFunction(), "inside.*MemberFunction");
+}
+
+void ChangeToRootDir() { posix::ChDir(GTEST_PATH_SEP_); }
+
+// Tests that death tests work even if the current directory has been
+// changed.
+TEST_F(TestForDeathTest, FastDeathTestInChangedDir) {
+ testing::GTEST_FLAG(death_test_style) = "fast";
+
+ ChangeToRootDir();
+ EXPECT_EXIT(_exit(1), testing::ExitedWithCode(1), "");
+
+ ChangeToRootDir();
+ ASSERT_DEATH(_exit(1), "");
+}
+
+// Repeats a representative sample of death tests in the "threadsafe" style:
+
+TEST_F(TestForDeathTest, StaticMemberFunctionThreadsafeStyle) {
+ testing::GTEST_FLAG(death_test_style) = "threadsafe";
+ ASSERT_DEATH(StaticMemberFunction(), "death.*StaticMember");
+}
+
+TEST_F(TestForDeathTest, MemberFunctionThreadsafeStyle) {
+ testing::GTEST_FLAG(death_test_style) = "threadsafe";
+ should_die_ = true;
+ EXPECT_DEATH(MemberFunction(), "inside.*MemberFunction");
+}
+
+TEST_F(TestForDeathTest, ThreadsafeDeathTestInLoop) {
+ testing::GTEST_FLAG(death_test_style) = "threadsafe";
+
+ for (int i = 0; i < 3; ++i)
+ EXPECT_EXIT(_exit(i), testing::ExitedWithCode(i), "") << ": i = " << i;
+}
+
+TEST_F(TestForDeathTest, ThreadsafeDeathTestInChangedDir) {
+ testing::GTEST_FLAG(death_test_style) = "threadsafe";
+
+ ChangeToRootDir();
+ EXPECT_EXIT(_exit(1), testing::ExitedWithCode(1), "");
+
+ ChangeToRootDir();
+ ASSERT_DEATH(_exit(1), "");
+}
+
+TEST_F(TestForDeathTest, MixedStyles) {
+ testing::GTEST_FLAG(death_test_style) = "threadsafe";
+ EXPECT_DEATH(_exit(1), "");
+ testing::GTEST_FLAG(death_test_style) = "fast";
+ EXPECT_DEATH(_exit(1), "");
+}
+
+namespace {
+
+bool pthread_flag;
+
+void SetPthreadFlag() {
+ pthread_flag = true;
+}
+
+} // namespace
+
+# if GTEST_HAS_CLONE && GTEST_HAS_PTHREAD
+
+TEST_F(TestForDeathTest, DoesNotExecuteAtforkHooks) {
+ if (!testing::GTEST_FLAG(death_test_use_fork)) {
+ testing::GTEST_FLAG(death_test_style) = "threadsafe";
+ pthread_flag = false;
+ ASSERT_EQ(0, pthread_atfork(&SetPthreadFlag, NULL, NULL));
+ ASSERT_DEATH(_exit(1), "");
+ ASSERT_FALSE(pthread_flag);
+ }
+}
+
+# endif // GTEST_HAS_CLONE && GTEST_HAS_PTHREAD
+
+// Tests that a method of another class can be used in a death test.
+TEST_F(TestForDeathTest, MethodOfAnotherClass) {
+ const MayDie x(true);
+ ASSERT_DEATH(x.MemberFunction(), "MayDie\\:\\:MemberFunction");
+}
+
+// Tests that a global function can be used in a death test.
+TEST_F(TestForDeathTest, GlobalFunction) {
+ EXPECT_DEATH(GlobalFunction(), "GlobalFunction");
+}
+
+// Tests that any value convertible to an RE works as a second
+// argument to EXPECT_DEATH.
+TEST_F(TestForDeathTest, AcceptsAnythingConvertibleToRE) {
+ static const char regex_c_str[] = "GlobalFunction";
+ EXPECT_DEATH(GlobalFunction(), regex_c_str);
+
+ const testing::internal::RE regex(regex_c_str);
+ EXPECT_DEATH(GlobalFunction(), regex);
+
+# if GTEST_HAS_GLOBAL_STRING
+
+ const string regex_str(regex_c_str);
+ EXPECT_DEATH(GlobalFunction(), regex_str);
+
+# endif // GTEST_HAS_GLOBAL_STRING
+
+ const ::std::string regex_std_str(regex_c_str);
+ EXPECT_DEATH(GlobalFunction(), regex_std_str);
+}
+
+// Tests that a non-void function can be used in a death test.
+TEST_F(TestForDeathTest, NonVoidFunction) {
+ ASSERT_DEATH(NonVoidFunction(), "NonVoidFunction");
+}
+
+// Tests that functions that take parameter(s) can be used in a death test.
+TEST_F(TestForDeathTest, FunctionWithParameter) {
+ EXPECT_DEATH(DieIf(true), "DieIf\\(\\)");
+ EXPECT_DEATH(DieIfLessThan(2, 3), "DieIfLessThan");
+}
+
+// Tests that ASSERT_DEATH can be used outside a TEST, TEST_F, or test fixture.
+TEST_F(TestForDeathTest, OutsideFixture) {
+ DeathTestSubroutine();
+}
+
+// Tests that death tests can be done inside a loop.
+TEST_F(TestForDeathTest, InsideLoop) {
+ for (int i = 0; i < 5; i++) {
+ EXPECT_DEATH(DieIfLessThan(-1, i), "DieIfLessThan") << "where i == " << i;
+ }
+}
+
+// Tests that a compound statement can be used in a death test.
+TEST_F(TestForDeathTest, CompoundStatement) {
+ EXPECT_DEATH({ // NOLINT
+ const int x = 2;
+ const int y = x + 1;
+ DieIfLessThan(x, y);
+ },
+ "DieIfLessThan");
+}
+
+// Tests that code that doesn't die causes a death test to fail.
+TEST_F(TestForDeathTest, DoesNotDie) {
+ EXPECT_NONFATAL_FAILURE(EXPECT_DEATH(DieIf(false), "DieIf"),
+ "failed to die");
+}
+
+// Tests that a death test fails when the error message isn't expected.
+TEST_F(TestForDeathTest, ErrorMessageMismatch) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_DEATH(DieIf(true), "DieIfLessThan") << "End of death test message.";
+ }, "died but not with expected error");
+}
+
+// On exit, *aborted will be true iff the EXPECT_DEATH() statement
+// aborted the function.
+void ExpectDeathTestHelper(bool* aborted) {
+ *aborted = true;
+ EXPECT_DEATH(DieIf(false), "DieIf"); // This assertion should fail.
+ *aborted = false;
+}
+
+// Tests that EXPECT_DEATH doesn't abort the test on failure.
+TEST_F(TestForDeathTest, EXPECT_DEATH) {
+ bool aborted = true;
+ EXPECT_NONFATAL_FAILURE(ExpectDeathTestHelper(&aborted),
+ "failed to die");
+ EXPECT_FALSE(aborted);
+}
+
+// Tests that ASSERT_DEATH does abort the test on failure.
+TEST_F(TestForDeathTest, ASSERT_DEATH) {
+ static bool aborted;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ aborted = true;
+ ASSERT_DEATH(DieIf(false), "DieIf"); // This assertion should fail.
+ aborted = false;
+ }, "failed to die");
+ EXPECT_TRUE(aborted);
+}
+
+// Tests that EXPECT_DEATH evaluates the arguments exactly once.
+TEST_F(TestForDeathTest, SingleEvaluation) {
+ int x = 3;
+ EXPECT_DEATH(DieIf((++x) == 4), "DieIf");
+
+ const char* regex = "DieIf";
+ const char* regex_save = regex;
+ EXPECT_DEATH(DieIfLessThan(3, 4), regex++);
+ EXPECT_EQ(regex_save + 1, regex);
+}
+
+// Tests that run-away death tests are reported as failures.
+TEST_F(TestForDeathTest, RunawayIsFailure) {
+ EXPECT_NONFATAL_FAILURE(EXPECT_DEATH(static_cast<void>(0), "Foo"),
+ "failed to die.");
+}
+
+// Tests that death tests report executing 'return' in the statement as
+// failure.
+TEST_F(TestForDeathTest, ReturnIsFailure) {
+ EXPECT_FATAL_FAILURE(ASSERT_DEATH(return, "Bar"),
+ "illegal return in test statement.");
+}
+
+// Tests that EXPECT_DEBUG_DEATH works as expected,
+// that is, in debug mode, it:
+// 1. Asserts on death.
+// 2. Has no side effect.
+//
+// And in opt mode, it:
+// 1. Has side effects but does not assert.
+TEST_F(TestForDeathTest, TestExpectDebugDeath) {
+ int sideeffect = 0;
+
+ EXPECT_DEBUG_DEATH(DieInDebugElse12(&sideeffect),
+ "death.*DieInDebugElse12");
+
+# ifdef NDEBUG
+
+ // Checks that the assignment occurs in opt mode (sideeffect).
+ EXPECT_EQ(12, sideeffect);
+
+# else
+
+ // Checks that the assignment does not occur in dbg mode (no sideeffect).
+ EXPECT_EQ(0, sideeffect);
+
+# endif
+}
+
+// Tests that ASSERT_DEBUG_DEATH works as expected
+// In debug mode:
+// 1. Asserts on debug death.
+// 2. Has no side effect.
+//
+// In opt mode:
+// 1. Has side effects and returns the expected value (12).
+TEST_F(TestForDeathTest, TestAssertDebugDeath) {
+ int sideeffect = 0;
+
+ ASSERT_DEBUG_DEATH({ // NOLINT
+ // Tests that the return value is 12 in opt mode.
+ EXPECT_EQ(12, DieInDebugElse12(&sideeffect));
+ // Tests that the side effect occurred in opt mode.
+ EXPECT_EQ(12, sideeffect);
+ }, "death.*DieInDebugElse12");
+
+# ifdef NDEBUG
+
+ // Checks that the assignment occurs in opt mode (sideeffect).
+ EXPECT_EQ(12, sideeffect);
+
+# else
+
+ // Checks that the assignment does not occur in dbg mode (no sideeffect).
+ EXPECT_EQ(0, sideeffect);
+
+# endif
+}
+
+# ifndef NDEBUG
+
+void ExpectDebugDeathHelper(bool* aborted) {
+ *aborted = true;
+ EXPECT_DEBUG_DEATH(return, "") << "This is expected to fail.";
+ *aborted = false;
+}
+
+# if GTEST_OS_WINDOWS
+TEST(PopUpDeathTest, DoesNotShowPopUpOnAbort) {
+ printf("This test should be considered failing if it shows "
+ "any pop-up dialogs.\n");
+ fflush(stdout);
+
+ EXPECT_DEATH({
+ testing::GTEST_FLAG(catch_exceptions) = false;
+ abort();
+ }, "");
+}
+# endif // GTEST_OS_WINDOWS
+
+// Tests that EXPECT_DEBUG_DEATH in debug mode does not abort
+// the function.
+TEST_F(TestForDeathTest, ExpectDebugDeathDoesNotAbort) {
+ bool aborted = true;
+ EXPECT_NONFATAL_FAILURE(ExpectDebugDeathHelper(&aborted), "");
+ EXPECT_FALSE(aborted);
+}
+
+void AssertDebugDeathHelper(bool* aborted) {
+ *aborted = true;
+ ASSERT_DEBUG_DEATH(return, "") << "This is expected to fail.";
+ *aborted = false;
+}
+
+// Tests that ASSERT_DEBUG_DEATH in debug mode aborts the function on
+// failure.
+TEST_F(TestForDeathTest, AssertDebugDeathAborts) {
+ static bool aborted;
+ aborted = false;
+ EXPECT_FATAL_FAILURE(AssertDebugDeathHelper(&aborted), "");
+ EXPECT_TRUE(aborted);
+}
+
+# endif // _NDEBUG
+
+// Tests the *_EXIT family of macros, using a variety of predicates.
+static void TestExitMacros() {
+ EXPECT_EXIT(_exit(1), testing::ExitedWithCode(1), "");
+ ASSERT_EXIT(_exit(42), testing::ExitedWithCode(42), "");
+
+# if GTEST_OS_WINDOWS
+
+ // Of all signals effects on the process exit code, only those of SIGABRT
+ // are documented on Windows.
+ // See http://msdn.microsoft.com/en-us/library/dwwzkt4c(VS.71).aspx.
+ EXPECT_EXIT(raise(SIGABRT), testing::ExitedWithCode(3), "");
+
+# else
+
+ EXPECT_EXIT(raise(SIGKILL), testing::KilledBySignal(SIGKILL), "") << "foo";
+ ASSERT_EXIT(raise(SIGUSR2), testing::KilledBySignal(SIGUSR2), "") << "bar";
+
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_EXIT(_exit(0), testing::KilledBySignal(SIGSEGV), "")
+ << "This failure is expected, too.";
+ }, "This failure is expected, too.");
+
+# endif // GTEST_OS_WINDOWS
+
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_EXIT(raise(SIGSEGV), testing::ExitedWithCode(0), "")
+ << "This failure is expected.";
+ }, "This failure is expected.");
+}
+
+TEST_F(TestForDeathTest, ExitMacros) {
+ TestExitMacros();
+}
+
+TEST_F(TestForDeathTest, ExitMacrosUsingFork) {
+ testing::GTEST_FLAG(death_test_use_fork) = true;
+ TestExitMacros();
+}
+
+TEST_F(TestForDeathTest, InvalidStyle) {
+ testing::GTEST_FLAG(death_test_style) = "rococo";
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_DEATH(_exit(0), "") << "This failure is expected.";
+ }, "This failure is expected.");
+}
+
+TEST_F(TestForDeathTest, DeathTestFailedOutput) {
+ testing::GTEST_FLAG(death_test_style) = "fast";
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_DEATH(DieWithMessage("death\n"),
+ "expected message"),
+ "Actual msg:\n"
+ "[ DEATH ] death\n");
+}
+
+TEST_F(TestForDeathTest, DeathTestUnexpectedReturnOutput) {
+ testing::GTEST_FLAG(death_test_style) = "fast";
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_DEATH({
+ fprintf(stderr, "returning\n");
+ fflush(stderr);
+ return;
+ }, ""),
+ " Result: illegal return in test statement.\n"
+ " Error msg:\n"
+ "[ DEATH ] returning\n");
+}
+
+TEST_F(TestForDeathTest, DeathTestBadExitCodeOutput) {
+ testing::GTEST_FLAG(death_test_style) = "fast";
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_EXIT(DieWithMessage("exiting with rc 1\n"),
+ testing::ExitedWithCode(3),
+ "expected message"),
+ " Result: died but not with expected exit code:\n"
+ " Exited with exit status 1\n"
+ "Actual msg:\n"
+ "[ DEATH ] exiting with rc 1\n");
+}
+
+TEST_F(TestForDeathTest, DeathTestMultiLineMatchFail) {
+ testing::GTEST_FLAG(death_test_style) = "fast";
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_DEATH(DieWithMessage("line 1\nline 2\nline 3\n"),
+ "line 1\nxyz\nline 3\n"),
+ "Actual msg:\n"
+ "[ DEATH ] line 1\n"
+ "[ DEATH ] line 2\n"
+ "[ DEATH ] line 3\n");
+}
+
+TEST_F(TestForDeathTest, DeathTestMultiLineMatchPass) {
+ testing::GTEST_FLAG(death_test_style) = "fast";
+ EXPECT_DEATH(DieWithMessage("line 1\nline 2\nline 3\n"),
+ "line 1\nline 2\nline 3\n");
+}
+
+// A DeathTestFactory that returns MockDeathTests.
+class MockDeathTestFactory : public DeathTestFactory {
+ public:
+ MockDeathTestFactory();
+ virtual bool Create(const char* statement,
+ const ::testing::internal::RE* regex,
+ const char* file, int line, DeathTest** test);
+
+ // Sets the parameters for subsequent calls to Create.
+ void SetParameters(bool create, DeathTest::TestRole role,
+ int status, bool passed);
+
+ // Accessors.
+ int AssumeRoleCalls() const { return assume_role_calls_; }
+ int WaitCalls() const { return wait_calls_; }
+ int PassedCalls() const { return passed_args_.size(); }
+ bool PassedArgument(int n) const { return passed_args_[n]; }
+ int AbortCalls() const { return abort_args_.size(); }
+ DeathTest::AbortReason AbortArgument(int n) const {
+ return abort_args_[n];
+ }
+ bool TestDeleted() const { return test_deleted_; }
+
+ private:
+ friend class MockDeathTest;
+ // If true, Create will return a MockDeathTest; otherwise it returns
+ // NULL.
+ bool create_;
+ // The value a MockDeathTest will return from its AssumeRole method.
+ DeathTest::TestRole role_;
+ // The value a MockDeathTest will return from its Wait method.
+ int status_;
+ // The value a MockDeathTest will return from its Passed method.
+ bool passed_;
+
+ // Number of times AssumeRole was called.
+ int assume_role_calls_;
+ // Number of times Wait was called.
+ int wait_calls_;
+ // The arguments to the calls to Passed since the last call to
+ // SetParameters.
+ std::vector<bool> passed_args_;
+ // The arguments to the calls to Abort since the last call to
+ // SetParameters.
+ std::vector<DeathTest::AbortReason> abort_args_;
+ // True if the last MockDeathTest returned by Create has been
+ // deleted.
+ bool test_deleted_;
+};
+
+
+// A DeathTest implementation useful in testing. It returns values set
+// at its creation from its various inherited DeathTest methods, and
+// reports calls to those methods to its parent MockDeathTestFactory
+// object.
+class MockDeathTest : public DeathTest {
+ public:
+ MockDeathTest(MockDeathTestFactory *parent,
+ TestRole role, int status, bool passed) :
+ parent_(parent), role_(role), status_(status), passed_(passed) {
+ }
+ virtual ~MockDeathTest() {
+ parent_->test_deleted_ = true;
+ }
+ virtual TestRole AssumeRole() {
+ ++parent_->assume_role_calls_;
+ return role_;
+ }
+ virtual int Wait() {
+ ++parent_->wait_calls_;
+ return status_;
+ }
+ virtual bool Passed(bool exit_status_ok) {
+ parent_->passed_args_.push_back(exit_status_ok);
+ return passed_;
+ }
+ virtual void Abort(AbortReason reason) {
+ parent_->abort_args_.push_back(reason);
+ }
+ private:
+ MockDeathTestFactory* const parent_;
+ const TestRole role_;
+ const int status_;
+ const bool passed_;
+};
+
+
+// MockDeathTestFactory constructor.
+MockDeathTestFactory::MockDeathTestFactory()
+ : create_(true),
+ role_(DeathTest::OVERSEE_TEST),
+ status_(0),
+ passed_(true),
+ assume_role_calls_(0),
+ wait_calls_(0),
+ passed_args_(),
+ abort_args_() {
+}
+
+
+// Sets the parameters for subsequent calls to Create.
+void MockDeathTestFactory::SetParameters(bool create,
+ DeathTest::TestRole role,
+ int status, bool passed) {
+ create_ = create;
+ role_ = role;
+ status_ = status;
+ passed_ = passed;
+
+ assume_role_calls_ = 0;
+ wait_calls_ = 0;
+ passed_args_.clear();
+ abort_args_.clear();
+}
+
+
+// Sets test to NULL (if create_ is false) or to the address of a new
+// MockDeathTest object with parameters taken from the last call
+// to SetParameters (if create_ is true). Always returns true.
+bool MockDeathTestFactory::Create(const char* /*statement*/,
+ const ::testing::internal::RE* /*regex*/,
+ const char* /*file*/,
+ int /*line*/,
+ DeathTest** test) {
+ test_deleted_ = false;
+ if (create_) {
+ *test = new MockDeathTest(this, role_, status_, passed_);
+ } else {
+ *test = NULL;
+ }
+ return true;
+}
+
+// A test fixture for testing the logic of the GTEST_DEATH_TEST_ macro.
+// It installs a MockDeathTestFactory that is used for the duration
+// of the test case.
+class MacroLogicDeathTest : public testing::Test {
+ protected:
+ static testing::internal::ReplaceDeathTestFactory* replacer_;
+ static MockDeathTestFactory* factory_;
+
+ static void SetUpTestCase() {
+ factory_ = new MockDeathTestFactory;
+ replacer_ = new testing::internal::ReplaceDeathTestFactory(factory_);
+ }
+
+ static void TearDownTestCase() {
+ delete replacer_;
+ replacer_ = NULL;
+ delete factory_;
+ factory_ = NULL;
+ }
+
+ // Runs a death test that breaks the rules by returning. Such a death
+ // test cannot be run directly from a test routine that uses a
+ // MockDeathTest, or the remainder of the routine will not be executed.
+ static void RunReturningDeathTest(bool* flag) {
+ ASSERT_DEATH({ // NOLINT
+ *flag = true;
+ return;
+ }, "");
+ }
+};
+
+testing::internal::ReplaceDeathTestFactory* MacroLogicDeathTest::replacer_
+ = NULL;
+MockDeathTestFactory* MacroLogicDeathTest::factory_ = NULL;
+
+
+// Test that nothing happens when the factory doesn't return a DeathTest:
+TEST_F(MacroLogicDeathTest, NothingHappens) {
+ bool flag = false;
+ factory_->SetParameters(false, DeathTest::OVERSEE_TEST, 0, true);
+ EXPECT_DEATH(flag = true, "");
+ EXPECT_FALSE(flag);
+ EXPECT_EQ(0, factory_->AssumeRoleCalls());
+ EXPECT_EQ(0, factory_->WaitCalls());
+ EXPECT_EQ(0, factory_->PassedCalls());
+ EXPECT_EQ(0, factory_->AbortCalls());
+ EXPECT_FALSE(factory_->TestDeleted());
+}
+
+// Test that the parent process doesn't run the death test code,
+// and that the Passed method returns false when the (simulated)
+// child process exits with status 0:
+TEST_F(MacroLogicDeathTest, ChildExitsSuccessfully) {
+ bool flag = false;
+ factory_->SetParameters(true, DeathTest::OVERSEE_TEST, 0, true);
+ EXPECT_DEATH(flag = true, "");
+ EXPECT_FALSE(flag);
+ EXPECT_EQ(1, factory_->AssumeRoleCalls());
+ EXPECT_EQ(1, factory_->WaitCalls());
+ ASSERT_EQ(1, factory_->PassedCalls());
+ EXPECT_FALSE(factory_->PassedArgument(0));
+ EXPECT_EQ(0, factory_->AbortCalls());
+ EXPECT_TRUE(factory_->TestDeleted());
+}
+
+// Tests that the Passed method was given the argument "true" when
+// the (simulated) child process exits with status 1:
+TEST_F(MacroLogicDeathTest, ChildExitsUnsuccessfully) {
+ bool flag = false;
+ factory_->SetParameters(true, DeathTest::OVERSEE_TEST, 1, true);
+ EXPECT_DEATH(flag = true, "");
+ EXPECT_FALSE(flag);
+ EXPECT_EQ(1, factory_->AssumeRoleCalls());
+ EXPECT_EQ(1, factory_->WaitCalls());
+ ASSERT_EQ(1, factory_->PassedCalls());
+ EXPECT_TRUE(factory_->PassedArgument(0));
+ EXPECT_EQ(0, factory_->AbortCalls());
+ EXPECT_TRUE(factory_->TestDeleted());
+}
+
+// Tests that the (simulated) child process executes the death test
+// code, and is aborted with the correct AbortReason if it
+// executes a return statement.
+TEST_F(MacroLogicDeathTest, ChildPerformsReturn) {
+ bool flag = false;
+ factory_->SetParameters(true, DeathTest::EXECUTE_TEST, 0, true);
+ RunReturningDeathTest(&flag);
+ EXPECT_TRUE(flag);
+ EXPECT_EQ(1, factory_->AssumeRoleCalls());
+ EXPECT_EQ(0, factory_->WaitCalls());
+ EXPECT_EQ(0, factory_->PassedCalls());
+ EXPECT_EQ(1, factory_->AbortCalls());
+ EXPECT_EQ(DeathTest::TEST_ENCOUNTERED_RETURN_STATEMENT,
+ factory_->AbortArgument(0));
+ EXPECT_TRUE(factory_->TestDeleted());
+}
+
+// Tests that the (simulated) child process is aborted with the
+// correct AbortReason if it does not die.
+TEST_F(MacroLogicDeathTest, ChildDoesNotDie) {
+ bool flag = false;
+ factory_->SetParameters(true, DeathTest::EXECUTE_TEST, 0, true);
+ EXPECT_DEATH(flag = true, "");
+ EXPECT_TRUE(flag);
+ EXPECT_EQ(1, factory_->AssumeRoleCalls());
+ EXPECT_EQ(0, factory_->WaitCalls());
+ EXPECT_EQ(0, factory_->PassedCalls());
+ // This time there are two calls to Abort: one since the test didn't
+ // die, and another from the ReturnSentinel when it's destroyed. The
+ // sentinel normally isn't destroyed if a test doesn't die, since
+ // _exit(2) is called in that case by ForkingDeathTest, but not by
+ // our MockDeathTest.
+ ASSERT_EQ(2, factory_->AbortCalls());
+ EXPECT_EQ(DeathTest::TEST_DID_NOT_DIE,
+ factory_->AbortArgument(0));
+ EXPECT_EQ(DeathTest::TEST_ENCOUNTERED_RETURN_STATEMENT,
+ factory_->AbortArgument(1));
+ EXPECT_TRUE(factory_->TestDeleted());
+}
+
+// Tests that a successful death test does not register a successful
+// test part.
+TEST(SuccessRegistrationDeathTest, NoSuccessPart) {
+ EXPECT_DEATH(_exit(1), "");
+ EXPECT_EQ(0, GetUnitTestImpl()->current_test_result()->total_part_count());
+}
+
+TEST(StreamingAssertionsDeathTest, DeathTest) {
+ EXPECT_DEATH(_exit(1), "") << "unexpected failure";
+ ASSERT_DEATH(_exit(1), "") << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_DEATH(_exit(0), "") << "expected failure";
+ }, "expected failure");
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_DEATH(_exit(0), "") << "expected failure";
+ }, "expected failure");
+}
+
+// Tests that GetLastErrnoDescription returns an empty string when the
+// last error is 0 and non-empty string when it is non-zero.
+TEST(GetLastErrnoDescription, GetLastErrnoDescriptionWorks) {
+ errno = ENOENT;
+ EXPECT_STRNE("", GetLastErrnoDescription().c_str());
+ errno = 0;
+ EXPECT_STREQ("", GetLastErrnoDescription().c_str());
+}
+
+# if GTEST_OS_WINDOWS
+TEST(AutoHandleTest, AutoHandleWorks) {
+ HANDLE handle = ::CreateEvent(NULL, FALSE, FALSE, NULL);
+ ASSERT_NE(INVALID_HANDLE_VALUE, handle);
+
+ // Tests that the AutoHandle is correctly initialized with a handle.
+ testing::internal::AutoHandle auto_handle(handle);
+ EXPECT_EQ(handle, auto_handle.Get());
+
+ // Tests that Reset assigns INVALID_HANDLE_VALUE.
+ // Note that this cannot verify whether the original handle is closed.
+ auto_handle.Reset();
+ EXPECT_EQ(INVALID_HANDLE_VALUE, auto_handle.Get());
+
+ // Tests that Reset assigns the new handle.
+ // Note that this cannot verify whether the original handle is closed.
+ handle = ::CreateEvent(NULL, FALSE, FALSE, NULL);
+ ASSERT_NE(INVALID_HANDLE_VALUE, handle);
+ auto_handle.Reset(handle);
+ EXPECT_EQ(handle, auto_handle.Get());
+
+ // Tests that AutoHandle contains INVALID_HANDLE_VALUE by default.
+ testing::internal::AutoHandle auto_handle2;
+ EXPECT_EQ(INVALID_HANDLE_VALUE, auto_handle2.Get());
+}
+# endif // GTEST_OS_WINDOWS
+
+# if GTEST_OS_WINDOWS
+typedef unsigned __int64 BiggestParsable;
+typedef signed __int64 BiggestSignedParsable;
+const BiggestParsable kBiggestParsableMax = ULLONG_MAX;
+const BiggestSignedParsable kBiggestSignedParsableMax = LLONG_MAX;
+# else
+typedef unsigned long long BiggestParsable;
+typedef signed long long BiggestSignedParsable;
+const BiggestParsable kBiggestParsableMax =
+ ::std::numeric_limits<BiggestParsable>::max();
+const BiggestSignedParsable kBiggestSignedParsableMax =
+ ::std::numeric_limits<BiggestSignedParsable>::max();
+# endif // GTEST_OS_WINDOWS
+
+TEST(ParseNaturalNumberTest, RejectsInvalidFormat) {
+ BiggestParsable result = 0;
+
+ // Rejects non-numbers.
+ EXPECT_FALSE(ParseNaturalNumber(String("non-number string"), &result));
+
+ // Rejects numbers with whitespace prefix.
+ EXPECT_FALSE(ParseNaturalNumber(String(" 123"), &result));
+
+ // Rejects negative numbers.
+ EXPECT_FALSE(ParseNaturalNumber(String("-123"), &result));
+
+ // Rejects numbers starting with a plus sign.
+ EXPECT_FALSE(ParseNaturalNumber(String("+123"), &result));
+ errno = 0;
+}
+
+TEST(ParseNaturalNumberTest, RejectsOverflownNumbers) {
+ BiggestParsable result = 0;
+
+ EXPECT_FALSE(ParseNaturalNumber(String("99999999999999999999999"), &result));
+
+ signed char char_result = 0;
+ EXPECT_FALSE(ParseNaturalNumber(String("200"), &char_result));
+ errno = 0;
+}
+
+TEST(ParseNaturalNumberTest, AcceptsValidNumbers) {
+ BiggestParsable result = 0;
+
+ result = 0;
+ ASSERT_TRUE(ParseNaturalNumber(String("123"), &result));
+ EXPECT_EQ(123U, result);
+
+ // Check 0 as an edge case.
+ result = 1;
+ ASSERT_TRUE(ParseNaturalNumber(String("0"), &result));
+ EXPECT_EQ(0U, result);
+
+ result = 1;
+ ASSERT_TRUE(ParseNaturalNumber(String("00000"), &result));
+ EXPECT_EQ(0U, result);
+}
+
+TEST(ParseNaturalNumberTest, AcceptsTypeLimits) {
+ Message msg;
+ msg << kBiggestParsableMax;
+
+ BiggestParsable result = 0;
+ EXPECT_TRUE(ParseNaturalNumber(msg.GetString(), &result));
+ EXPECT_EQ(kBiggestParsableMax, result);
+
+ Message msg2;
+ msg2 << kBiggestSignedParsableMax;
+
+ BiggestSignedParsable signed_result = 0;
+ EXPECT_TRUE(ParseNaturalNumber(msg2.GetString(), &signed_result));
+ EXPECT_EQ(kBiggestSignedParsableMax, signed_result);
+
+ Message msg3;
+ msg3 << INT_MAX;
+
+ int int_result = 0;
+ EXPECT_TRUE(ParseNaturalNumber(msg3.GetString(), &int_result));
+ EXPECT_EQ(INT_MAX, int_result);
+
+ Message msg4;
+ msg4 << UINT_MAX;
+
+ unsigned int uint_result = 0;
+ EXPECT_TRUE(ParseNaturalNumber(msg4.GetString(), &uint_result));
+ EXPECT_EQ(UINT_MAX, uint_result);
+}
+
+TEST(ParseNaturalNumberTest, WorksForShorterIntegers) {
+ short short_result = 0;
+ ASSERT_TRUE(ParseNaturalNumber(String("123"), &short_result));
+ EXPECT_EQ(123, short_result);
+
+ signed char char_result = 0;
+ ASSERT_TRUE(ParseNaturalNumber(String("123"), &char_result));
+ EXPECT_EQ(123, char_result);
+}
+
+# if GTEST_OS_WINDOWS
+TEST(EnvironmentTest, HandleFitsIntoSizeT) {
+ // TODO(vladl@google.com): Remove this test after this condition is verified
+ // in a static assertion in gtest-death-test.cc in the function
+ // GetStatusFileDescriptor.
+ ASSERT_TRUE(sizeof(HANDLE) <= sizeof(size_t));
+}
+# endif // GTEST_OS_WINDOWS
+
+// Tests that EXPECT_DEATH_IF_SUPPORTED/ASSERT_DEATH_IF_SUPPORTED trigger
+// failures when death tests are available on the system.
+TEST(ConditionalDeathMacrosDeathTest, ExpectsDeathWhenDeathTestsAvailable) {
+ EXPECT_DEATH_IF_SUPPORTED(DieInside("CondDeathTestExpectMacro"),
+ "death inside CondDeathTestExpectMacro");
+ ASSERT_DEATH_IF_SUPPORTED(DieInside("CondDeathTestAssertMacro"),
+ "death inside CondDeathTestAssertMacro");
+
+ // Empty statement will not crash, which must trigger a failure.
+ EXPECT_NONFATAL_FAILURE(EXPECT_DEATH_IF_SUPPORTED(;, ""), "");
+ EXPECT_FATAL_FAILURE(ASSERT_DEATH_IF_SUPPORTED(;, ""), "");
+}
+
+#else
+
+using testing::internal::CaptureStderr;
+using testing::internal::GetCapturedStderr;
+using testing::internal::String;
+
+// Tests that EXPECT_DEATH_IF_SUPPORTED/ASSERT_DEATH_IF_SUPPORTED are still
+// defined but do not trigger failures when death tests are not available on
+// the system.
+TEST(ConditionalDeathMacrosTest, WarnsWhenDeathTestsNotAvailable) {
+ // Empty statement will not crash, but that should not trigger a failure
+ // when death tests are not supported.
+ CaptureStderr();
+ EXPECT_DEATH_IF_SUPPORTED(;, "");
+ String output = GetCapturedStderr();
+ ASSERT_TRUE(NULL != strstr(output.c_str(),
+ "Death tests are not supported on this platform"));
+ ASSERT_TRUE(NULL != strstr(output.c_str(), ";"));
+
+ // The streamed message should not be printed as there is no test failure.
+ CaptureStderr();
+ EXPECT_DEATH_IF_SUPPORTED(;, "") << "streamed message";
+ output = GetCapturedStderr();
+ ASSERT_TRUE(NULL == strstr(output.c_str(), "streamed message"));
+
+ CaptureStderr();
+ ASSERT_DEATH_IF_SUPPORTED(;, ""); // NOLINT
+ output = GetCapturedStderr();
+ ASSERT_TRUE(NULL != strstr(output.c_str(),
+ "Death tests are not supported on this platform"));
+ ASSERT_TRUE(NULL != strstr(output.c_str(), ";"));
+
+ CaptureStderr();
+ ASSERT_DEATH_IF_SUPPORTED(;, "") << "streamed message"; // NOLINT
+ output = GetCapturedStderr();
+ ASSERT_TRUE(NULL == strstr(output.c_str(), "streamed message"));
+}
+
+void FuncWithAssert(int* n) {
+ ASSERT_DEATH_IF_SUPPORTED(return;, "");
+ (*n)++;
+}
+
+// Tests that ASSERT_DEATH_IF_SUPPORTED does not return from the current
+// function (as ASSERT_DEATH does) if death tests are not supported.
+TEST(ConditionalDeathMacrosTest, AssertDeatDoesNotReturnhIfUnsupported) {
+ int n = 0;
+ FuncWithAssert(&n);
+ EXPECT_EQ(1, n);
+}
+#endif // GTEST_HAS_DEATH_TEST
+
+// Tests that the death test macros expand to code which may or may not
+// be followed by operator<<, and that in either case the complete text
+// comprises only a single C++ statement.
+//
+// The syntax should work whether death tests are available or not.
+TEST(ConditionalDeathMacrosSyntaxDeathTest, SingleStatement) {
+ if (AlwaysFalse())
+ // This would fail if executed; this is a compilation test only
+ ASSERT_DEATH_IF_SUPPORTED(return, "");
+
+ if (AlwaysTrue())
+ EXPECT_DEATH_IF_SUPPORTED(_exit(1), "");
+ else
+ // This empty "else" branch is meant to ensure that EXPECT_DEATH
+ // doesn't expand into an "if" statement without an "else"
+ ; // NOLINT
+
+ if (AlwaysFalse())
+ ASSERT_DEATH_IF_SUPPORTED(return, "") << "did not die";
+
+ if (AlwaysFalse())
+ ; // NOLINT
+ else
+ EXPECT_DEATH_IF_SUPPORTED(_exit(1), "") << 1 << 2 << 3;
+}
+
+// Tests that conditional death test macros expand to code which interacts
+// well with switch statements.
+TEST(ConditionalDeathMacrosSyntaxDeathTest, SwitchStatement) {
+// Microsoft compiler usually complains about switch statements without
+// case labels. We suppress that warning for this test.
+#ifdef _MSC_VER
+# pragma warning(push)
+# pragma warning(disable: 4065)
+#endif // _MSC_VER
+
+ switch (0)
+ default:
+ ASSERT_DEATH_IF_SUPPORTED(_exit(1), "")
+ << "exit in default switch handler";
+
+ switch (0)
+ case 0:
+ EXPECT_DEATH_IF_SUPPORTED(_exit(1), "") << "exit in switch case";
+
+#ifdef _MSC_VER
+# pragma warning(pop)
+#endif // _MSC_VER
+}
+
+// Tests that a test case whose name ends with "DeathTest" works fine
+// on Windows.
+TEST(NotADeathTest, Test) {
+ SUCCEED();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest-filepath_test.cc b/libvpx/third_party/googletest/src/test/gtest-filepath_test.cc
new file mode 100644
index 0000000..66d4118
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-filepath_test.cc
@@ -0,0 +1,696 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: keith.ray@gmail.com (Keith Ray)
+//
+// Google Test filepath utilities
+//
+// This file tests classes and functions used internally by
+// Google Test. They are subject to change without notice.
+//
+// This file is #included from gtest_unittest.cc, to avoid changing
+// build or make-files for some existing Google Test clients. Do not
+// #include this file anywhere else!
+
+#include "gtest/internal/gtest-filepath.h"
+#include "gtest/gtest.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+#if GTEST_OS_WINDOWS_MOBILE
+# include <windows.h> // NOLINT
+#elif GTEST_OS_WINDOWS
+# include <direct.h> // NOLINT
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+namespace testing {
+namespace internal {
+namespace {
+
+#if GTEST_OS_WINDOWS_MOBILE
+// TODO(wan@google.com): Move these to the POSIX adapter section in
+// gtest-port.h.
+
+// Windows CE doesn't have the remove C function.
+int remove(const char* path) {
+ LPCWSTR wpath = String::AnsiToUtf16(path);
+ int ret = DeleteFile(wpath) ? 0 : -1;
+ delete [] wpath;
+ return ret;
+}
+// Windows CE doesn't have the _rmdir C function.
+int _rmdir(const char* path) {
+ FilePath filepath(path);
+ LPCWSTR wpath = String::AnsiToUtf16(
+ filepath.RemoveTrailingPathSeparator().c_str());
+ int ret = RemoveDirectory(wpath) ? 0 : -1;
+ delete [] wpath;
+ return ret;
+}
+
+#else
+
+TEST(GetCurrentDirTest, ReturnsCurrentDir) {
+ const FilePath original_dir = FilePath::GetCurrentDir();
+ EXPECT_FALSE(original_dir.IsEmpty());
+
+ posix::ChDir(GTEST_PATH_SEP_);
+ const FilePath cwd = FilePath::GetCurrentDir();
+ posix::ChDir(original_dir.c_str());
+
+# if GTEST_OS_WINDOWS
+
+ // Skips the ":".
+ const char* const cwd_without_drive = strchr(cwd.c_str(), ':');
+ ASSERT_TRUE(cwd_without_drive != NULL);
+ EXPECT_STREQ(GTEST_PATH_SEP_, cwd_without_drive + 1);
+
+# else
+
+ EXPECT_STREQ(GTEST_PATH_SEP_, cwd.c_str());
+
+# endif
+}
+
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+TEST(IsEmptyTest, ReturnsTrueForEmptyPath) {
+ EXPECT_TRUE(FilePath("").IsEmpty());
+ EXPECT_TRUE(FilePath(NULL).IsEmpty());
+}
+
+TEST(IsEmptyTest, ReturnsFalseForNonEmptyPath) {
+ EXPECT_FALSE(FilePath("a").IsEmpty());
+ EXPECT_FALSE(FilePath(".").IsEmpty());
+ EXPECT_FALSE(FilePath("a/b").IsEmpty());
+ EXPECT_FALSE(FilePath("a\\b\\").IsEmpty());
+}
+
+// RemoveDirectoryName "" -> ""
+TEST(RemoveDirectoryNameTest, WhenEmptyName) {
+ EXPECT_STREQ("", FilePath("").RemoveDirectoryName().c_str());
+}
+
+// RemoveDirectoryName "afile" -> "afile"
+TEST(RemoveDirectoryNameTest, ButNoDirectory) {
+ EXPECT_STREQ("afile",
+ FilePath("afile").RemoveDirectoryName().c_str());
+}
+
+// RemoveDirectoryName "/afile" -> "afile"
+TEST(RemoveDirectoryNameTest, RootFileShouldGiveFileName) {
+ EXPECT_STREQ("afile",
+ FilePath(GTEST_PATH_SEP_ "afile").RemoveDirectoryName().c_str());
+}
+
+// RemoveDirectoryName "adir/" -> ""
+TEST(RemoveDirectoryNameTest, WhereThereIsNoFileName) {
+ EXPECT_STREQ("",
+ FilePath("adir" GTEST_PATH_SEP_).RemoveDirectoryName().c_str());
+}
+
+// RemoveDirectoryName "adir/afile" -> "afile"
+TEST(RemoveDirectoryNameTest, ShouldGiveFileName) {
+ EXPECT_STREQ("afile",
+ FilePath("adir" GTEST_PATH_SEP_ "afile").RemoveDirectoryName().c_str());
+}
+
+// RemoveDirectoryName "adir/subdir/afile" -> "afile"
+TEST(RemoveDirectoryNameTest, ShouldAlsoGiveFileName) {
+ EXPECT_STREQ("afile",
+ FilePath("adir" GTEST_PATH_SEP_ "subdir" GTEST_PATH_SEP_ "afile")
+ .RemoveDirectoryName().c_str());
+}
+
+#if GTEST_HAS_ALT_PATH_SEP_
+
+// Tests that RemoveDirectoryName() works with the alternate separator
+// on Windows.
+
+// RemoveDirectoryName("/afile") -> "afile"
+TEST(RemoveDirectoryNameTest, RootFileShouldGiveFileNameForAlternateSeparator) {
+ EXPECT_STREQ("afile",
+ FilePath("/afile").RemoveDirectoryName().c_str());
+}
+
+// RemoveDirectoryName("adir/") -> ""
+TEST(RemoveDirectoryNameTest, WhereThereIsNoFileNameForAlternateSeparator) {
+ EXPECT_STREQ("",
+ FilePath("adir/").RemoveDirectoryName().c_str());
+}
+
+// RemoveDirectoryName("adir/afile") -> "afile"
+TEST(RemoveDirectoryNameTest, ShouldGiveFileNameForAlternateSeparator) {
+ EXPECT_STREQ("afile",
+ FilePath("adir/afile").RemoveDirectoryName().c_str());
+}
+
+// RemoveDirectoryName("adir/subdir/afile") -> "afile"
+TEST(RemoveDirectoryNameTest, ShouldAlsoGiveFileNameForAlternateSeparator) {
+ EXPECT_STREQ("afile",
+ FilePath("adir/subdir/afile").RemoveDirectoryName().c_str());
+}
+
+#endif
+
+// RemoveFileName "" -> "./"
+TEST(RemoveFileNameTest, EmptyName) {
+#if GTEST_OS_WINDOWS_MOBILE
+ // On Windows CE, we use the root as the current directory.
+ EXPECT_STREQ(GTEST_PATH_SEP_,
+ FilePath("").RemoveFileName().c_str());
+#else
+ EXPECT_STREQ("." GTEST_PATH_SEP_,
+ FilePath("").RemoveFileName().c_str());
+#endif
+}
+
+// RemoveFileName "adir/" -> "adir/"
+TEST(RemoveFileNameTest, ButNoFile) {
+ EXPECT_STREQ("adir" GTEST_PATH_SEP_,
+ FilePath("adir" GTEST_PATH_SEP_).RemoveFileName().c_str());
+}
+
+// RemoveFileName "adir/afile" -> "adir/"
+TEST(RemoveFileNameTest, GivesDirName) {
+ EXPECT_STREQ("adir" GTEST_PATH_SEP_,
+ FilePath("adir" GTEST_PATH_SEP_ "afile")
+ .RemoveFileName().c_str());
+}
+
+// RemoveFileName "adir/subdir/afile" -> "adir/subdir/"
+TEST(RemoveFileNameTest, GivesDirAndSubDirName) {
+ EXPECT_STREQ("adir" GTEST_PATH_SEP_ "subdir" GTEST_PATH_SEP_,
+ FilePath("adir" GTEST_PATH_SEP_ "subdir" GTEST_PATH_SEP_ "afile")
+ .RemoveFileName().c_str());
+}
+
+// RemoveFileName "/afile" -> "/"
+TEST(RemoveFileNameTest, GivesRootDir) {
+ EXPECT_STREQ(GTEST_PATH_SEP_,
+ FilePath(GTEST_PATH_SEP_ "afile").RemoveFileName().c_str());
+}
+
+#if GTEST_HAS_ALT_PATH_SEP_
+
+// Tests that RemoveFileName() works with the alternate separator on
+// Windows.
+
+// RemoveFileName("adir/") -> "adir/"
+TEST(RemoveFileNameTest, ButNoFileForAlternateSeparator) {
+ EXPECT_STREQ("adir" GTEST_PATH_SEP_,
+ FilePath("adir/").RemoveFileName().c_str());
+}
+
+// RemoveFileName("adir/afile") -> "adir/"
+TEST(RemoveFileNameTest, GivesDirNameForAlternateSeparator) {
+ EXPECT_STREQ("adir" GTEST_PATH_SEP_,
+ FilePath("adir/afile").RemoveFileName().c_str());
+}
+
+// RemoveFileName("adir/subdir/afile") -> "adir/subdir/"
+TEST(RemoveFileNameTest, GivesDirAndSubDirNameForAlternateSeparator) {
+ EXPECT_STREQ("adir" GTEST_PATH_SEP_ "subdir" GTEST_PATH_SEP_,
+ FilePath("adir/subdir/afile").RemoveFileName().c_str());
+}
+
+// RemoveFileName("/afile") -> "\"
+TEST(RemoveFileNameTest, GivesRootDirForAlternateSeparator) {
+ EXPECT_STREQ(GTEST_PATH_SEP_,
+ FilePath("/afile").RemoveFileName().c_str());
+}
+
+#endif
+
+TEST(MakeFileNameTest, GenerateWhenNumberIsZero) {
+ FilePath actual = FilePath::MakeFileName(FilePath("foo"), FilePath("bar"),
+ 0, "xml");
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar.xml", actual.c_str());
+}
+
+TEST(MakeFileNameTest, GenerateFileNameNumberGtZero) {
+ FilePath actual = FilePath::MakeFileName(FilePath("foo"), FilePath("bar"),
+ 12, "xml");
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar_12.xml", actual.c_str());
+}
+
+TEST(MakeFileNameTest, GenerateFileNameWithSlashNumberIsZero) {
+ FilePath actual = FilePath::MakeFileName(FilePath("foo" GTEST_PATH_SEP_),
+ FilePath("bar"), 0, "xml");
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar.xml", actual.c_str());
+}
+
+TEST(MakeFileNameTest, GenerateFileNameWithSlashNumberGtZero) {
+ FilePath actual = FilePath::MakeFileName(FilePath("foo" GTEST_PATH_SEP_),
+ FilePath("bar"), 12, "xml");
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar_12.xml", actual.c_str());
+}
+
+TEST(MakeFileNameTest, GenerateWhenNumberIsZeroAndDirIsEmpty) {
+ FilePath actual = FilePath::MakeFileName(FilePath(""), FilePath("bar"),
+ 0, "xml");
+ EXPECT_STREQ("bar.xml", actual.c_str());
+}
+
+TEST(MakeFileNameTest, GenerateWhenNumberIsNotZeroAndDirIsEmpty) {
+ FilePath actual = FilePath::MakeFileName(FilePath(""), FilePath("bar"),
+ 14, "xml");
+ EXPECT_STREQ("bar_14.xml", actual.c_str());
+}
+
+TEST(ConcatPathsTest, WorksWhenDirDoesNotEndWithPathSep) {
+ FilePath actual = FilePath::ConcatPaths(FilePath("foo"),
+ FilePath("bar.xml"));
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar.xml", actual.c_str());
+}
+
+TEST(ConcatPathsTest, WorksWhenPath1EndsWithPathSep) {
+ FilePath actual = FilePath::ConcatPaths(FilePath("foo" GTEST_PATH_SEP_),
+ FilePath("bar.xml"));
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar.xml", actual.c_str());
+}
+
+TEST(ConcatPathsTest, Path1BeingEmpty) {
+ FilePath actual = FilePath::ConcatPaths(FilePath(""),
+ FilePath("bar.xml"));
+ EXPECT_STREQ("bar.xml", actual.c_str());
+}
+
+TEST(ConcatPathsTest, Path2BeingEmpty) {
+ FilePath actual = FilePath::ConcatPaths(FilePath("foo"),
+ FilePath(""));
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_, actual.c_str());
+}
+
+TEST(ConcatPathsTest, BothPathBeingEmpty) {
+ FilePath actual = FilePath::ConcatPaths(FilePath(""),
+ FilePath(""));
+ EXPECT_STREQ("", actual.c_str());
+}
+
+TEST(ConcatPathsTest, Path1ContainsPathSep) {
+ FilePath actual = FilePath::ConcatPaths(FilePath("foo" GTEST_PATH_SEP_ "bar"),
+ FilePath("foobar.xml"));
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar" GTEST_PATH_SEP_ "foobar.xml",
+ actual.c_str());
+}
+
+TEST(ConcatPathsTest, Path2ContainsPathSep) {
+ FilePath actual = FilePath::ConcatPaths(
+ FilePath("foo" GTEST_PATH_SEP_),
+ FilePath("bar" GTEST_PATH_SEP_ "bar.xml"));
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar" GTEST_PATH_SEP_ "bar.xml",
+ actual.c_str());
+}
+
+TEST(ConcatPathsTest, Path2EndsWithPathSep) {
+ FilePath actual = FilePath::ConcatPaths(FilePath("foo"),
+ FilePath("bar" GTEST_PATH_SEP_));
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar" GTEST_PATH_SEP_, actual.c_str());
+}
+
+// RemoveTrailingPathSeparator "" -> ""
+TEST(RemoveTrailingPathSeparatorTest, EmptyString) {
+ EXPECT_STREQ("",
+ FilePath("").RemoveTrailingPathSeparator().c_str());
+}
+
+// RemoveTrailingPathSeparator "foo" -> "foo"
+TEST(RemoveTrailingPathSeparatorTest, FileNoSlashString) {
+ EXPECT_STREQ("foo",
+ FilePath("foo").RemoveTrailingPathSeparator().c_str());
+}
+
+// RemoveTrailingPathSeparator "foo/" -> "foo"
+TEST(RemoveTrailingPathSeparatorTest, ShouldRemoveTrailingSeparator) {
+ EXPECT_STREQ(
+ "foo",
+ FilePath("foo" GTEST_PATH_SEP_).RemoveTrailingPathSeparator().c_str());
+#if GTEST_HAS_ALT_PATH_SEP_
+ EXPECT_STREQ("foo",
+ FilePath("foo/").RemoveTrailingPathSeparator().c_str());
+#endif
+}
+
+// RemoveTrailingPathSeparator "foo/bar/" -> "foo/bar/"
+TEST(RemoveTrailingPathSeparatorTest, ShouldRemoveLastSeparator) {
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar",
+ FilePath("foo" GTEST_PATH_SEP_ "bar" GTEST_PATH_SEP_)
+ .RemoveTrailingPathSeparator().c_str());
+}
+
+// RemoveTrailingPathSeparator "foo/bar" -> "foo/bar"
+TEST(RemoveTrailingPathSeparatorTest, ShouldReturnUnmodified) {
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar",
+ FilePath("foo" GTEST_PATH_SEP_ "bar")
+ .RemoveTrailingPathSeparator().c_str());
+}
+
+TEST(DirectoryTest, RootDirectoryExists) {
+#if GTEST_OS_WINDOWS // We are on Windows.
+ char current_drive[_MAX_PATH]; // NOLINT
+ current_drive[0] = static_cast<char>(_getdrive() + 'A' - 1);
+ current_drive[1] = ':';
+ current_drive[2] = '\\';
+ current_drive[3] = '\0';
+ EXPECT_TRUE(FilePath(current_drive).DirectoryExists());
+#else
+ EXPECT_TRUE(FilePath("/").DirectoryExists());
+#endif // GTEST_OS_WINDOWS
+}
+
+#if GTEST_OS_WINDOWS
+TEST(DirectoryTest, RootOfWrongDriveDoesNotExists) {
+ const int saved_drive_ = _getdrive();
+ // Find a drive that doesn't exist. Start with 'Z' to avoid common ones.
+ for (char drive = 'Z'; drive >= 'A'; drive--)
+ if (_chdrive(drive - 'A' + 1) == -1) {
+ char non_drive[_MAX_PATH]; // NOLINT
+ non_drive[0] = drive;
+ non_drive[1] = ':';
+ non_drive[2] = '\\';
+ non_drive[3] = '\0';
+ EXPECT_FALSE(FilePath(non_drive).DirectoryExists());
+ break;
+ }
+ _chdrive(saved_drive_);
+}
+#endif // GTEST_OS_WINDOWS
+
+#if !GTEST_OS_WINDOWS_MOBILE
+// Windows CE _does_ consider an empty directory to exist.
+TEST(DirectoryTest, EmptyPathDirectoryDoesNotExist) {
+ EXPECT_FALSE(FilePath("").DirectoryExists());
+}
+#endif // !GTEST_OS_WINDOWS_MOBILE
+
+TEST(DirectoryTest, CurrentDirectoryExists) {
+#if GTEST_OS_WINDOWS // We are on Windows.
+# ifndef _WIN32_CE // Windows CE doesn't have a current directory.
+
+ EXPECT_TRUE(FilePath(".").DirectoryExists());
+ EXPECT_TRUE(FilePath(".\\").DirectoryExists());
+
+# endif // _WIN32_CE
+#else
+ EXPECT_TRUE(FilePath(".").DirectoryExists());
+ EXPECT_TRUE(FilePath("./").DirectoryExists());
+#endif // GTEST_OS_WINDOWS
+}
+
+TEST(NormalizeTest, NullStringsEqualEmptyDirectory) {
+ EXPECT_STREQ("", FilePath(NULL).c_str());
+ EXPECT_STREQ("", FilePath(String(NULL)).c_str());
+}
+
+// "foo/bar" == foo//bar" == "foo///bar"
+TEST(NormalizeTest, MultipleConsecutiveSepaparatorsInMidstring) {
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar",
+ FilePath("foo" GTEST_PATH_SEP_ "bar").c_str());
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar",
+ FilePath("foo" GTEST_PATH_SEP_ GTEST_PATH_SEP_ "bar").c_str());
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_ "bar",
+ FilePath("foo" GTEST_PATH_SEP_ GTEST_PATH_SEP_
+ GTEST_PATH_SEP_ "bar").c_str());
+}
+
+// "/bar" == //bar" == "///bar"
+TEST(NormalizeTest, MultipleConsecutiveSepaparatorsAtStringStart) {
+ EXPECT_STREQ(GTEST_PATH_SEP_ "bar",
+ FilePath(GTEST_PATH_SEP_ "bar").c_str());
+ EXPECT_STREQ(GTEST_PATH_SEP_ "bar",
+ FilePath(GTEST_PATH_SEP_ GTEST_PATH_SEP_ "bar").c_str());
+ EXPECT_STREQ(GTEST_PATH_SEP_ "bar",
+ FilePath(GTEST_PATH_SEP_ GTEST_PATH_SEP_ GTEST_PATH_SEP_ "bar").c_str());
+}
+
+// "foo/" == foo//" == "foo///"
+TEST(NormalizeTest, MultipleConsecutiveSepaparatorsAtStringEnd) {
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_,
+ FilePath("foo" GTEST_PATH_SEP_).c_str());
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_,
+ FilePath("foo" GTEST_PATH_SEP_ GTEST_PATH_SEP_).c_str());
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_,
+ FilePath("foo" GTEST_PATH_SEP_ GTEST_PATH_SEP_ GTEST_PATH_SEP_).c_str());
+}
+
+#if GTEST_HAS_ALT_PATH_SEP_
+
+// Tests that separators at the end of the string are normalized
+// regardless of their combination (e.g. "foo\" =="foo/\" ==
+// "foo\\/").
+TEST(NormalizeTest, MixAlternateSeparatorAtStringEnd) {
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_,
+ FilePath("foo/").c_str());
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_,
+ FilePath("foo" GTEST_PATH_SEP_ "/").c_str());
+ EXPECT_STREQ("foo" GTEST_PATH_SEP_,
+ FilePath("foo//" GTEST_PATH_SEP_).c_str());
+}
+
+#endif
+
+TEST(AssignmentOperatorTest, DefaultAssignedToNonDefault) {
+ FilePath default_path;
+ FilePath non_default_path("path");
+ non_default_path = default_path;
+ EXPECT_STREQ("", non_default_path.c_str());
+ EXPECT_STREQ("", default_path.c_str()); // RHS var is unchanged.
+}
+
+TEST(AssignmentOperatorTest, NonDefaultAssignedToDefault) {
+ FilePath non_default_path("path");
+ FilePath default_path;
+ default_path = non_default_path;
+ EXPECT_STREQ("path", default_path.c_str());
+ EXPECT_STREQ("path", non_default_path.c_str()); // RHS var is unchanged.
+}
+
+TEST(AssignmentOperatorTest, ConstAssignedToNonConst) {
+ const FilePath const_default_path("const_path");
+ FilePath non_default_path("path");
+ non_default_path = const_default_path;
+ EXPECT_STREQ("const_path", non_default_path.c_str());
+}
+
+class DirectoryCreationTest : public Test {
+ protected:
+ virtual void SetUp() {
+ testdata_path_.Set(FilePath(String::Format("%s%s%s",
+ TempDir().c_str(), GetCurrentExecutableName().c_str(),
+ "_directory_creation" GTEST_PATH_SEP_ "test" GTEST_PATH_SEP_)));
+ testdata_file_.Set(testdata_path_.RemoveTrailingPathSeparator());
+
+ unique_file0_.Set(FilePath::MakeFileName(testdata_path_, FilePath("unique"),
+ 0, "txt"));
+ unique_file1_.Set(FilePath::MakeFileName(testdata_path_, FilePath("unique"),
+ 1, "txt"));
+
+ remove(testdata_file_.c_str());
+ remove(unique_file0_.c_str());
+ remove(unique_file1_.c_str());
+ posix::RmDir(testdata_path_.c_str());
+ }
+
+ virtual void TearDown() {
+ remove(testdata_file_.c_str());
+ remove(unique_file0_.c_str());
+ remove(unique_file1_.c_str());
+ posix::RmDir(testdata_path_.c_str());
+ }
+
+ String TempDir() const {
+#if GTEST_OS_WINDOWS_MOBILE
+ return String("\\temp\\");
+#elif GTEST_OS_WINDOWS
+ const char* temp_dir = posix::GetEnv("TEMP");
+ if (temp_dir == NULL || temp_dir[0] == '\0')
+ return String("\\temp\\");
+ else if (String(temp_dir).EndsWith("\\"))
+ return String(temp_dir);
+ else
+ return String::Format("%s\\", temp_dir);
+#else
+ return String("/tmp/");
+#endif // GTEST_OS_WINDOWS_MOBILE
+ }
+
+ void CreateTextFile(const char* filename) {
+ FILE* f = posix::FOpen(filename, "w");
+ fprintf(f, "text\n");
+ fclose(f);
+ }
+
+ // Strings representing a directory and a file, with identical paths
+ // except for the trailing separator character that distinquishes
+ // a directory named 'test' from a file named 'test'. Example names:
+ FilePath testdata_path_; // "/tmp/directory_creation/test/"
+ FilePath testdata_file_; // "/tmp/directory_creation/test"
+ FilePath unique_file0_; // "/tmp/directory_creation/test/unique.txt"
+ FilePath unique_file1_; // "/tmp/directory_creation/test/unique_1.txt"
+};
+
+TEST_F(DirectoryCreationTest, CreateDirectoriesRecursively) {
+ EXPECT_FALSE(testdata_path_.DirectoryExists()) << testdata_path_.c_str();
+ EXPECT_TRUE(testdata_path_.CreateDirectoriesRecursively());
+ EXPECT_TRUE(testdata_path_.DirectoryExists());
+}
+
+TEST_F(DirectoryCreationTest, CreateDirectoriesForAlreadyExistingPath) {
+ EXPECT_FALSE(testdata_path_.DirectoryExists()) << testdata_path_.c_str();
+ EXPECT_TRUE(testdata_path_.CreateDirectoriesRecursively());
+ // Call 'create' again... should still succeed.
+ EXPECT_TRUE(testdata_path_.CreateDirectoriesRecursively());
+}
+
+TEST_F(DirectoryCreationTest, CreateDirectoriesAndUniqueFilename) {
+ FilePath file_path(FilePath::GenerateUniqueFileName(testdata_path_,
+ FilePath("unique"), "txt"));
+ EXPECT_STREQ(unique_file0_.c_str(), file_path.c_str());
+ EXPECT_FALSE(file_path.FileOrDirectoryExists()); // file not there
+
+ testdata_path_.CreateDirectoriesRecursively();
+ EXPECT_FALSE(file_path.FileOrDirectoryExists()); // file still not there
+ CreateTextFile(file_path.c_str());
+ EXPECT_TRUE(file_path.FileOrDirectoryExists());
+
+ FilePath file_path2(FilePath::GenerateUniqueFileName(testdata_path_,
+ FilePath("unique"), "txt"));
+ EXPECT_STREQ(unique_file1_.c_str(), file_path2.c_str());
+ EXPECT_FALSE(file_path2.FileOrDirectoryExists()); // file not there
+ CreateTextFile(file_path2.c_str());
+ EXPECT_TRUE(file_path2.FileOrDirectoryExists());
+}
+
+TEST_F(DirectoryCreationTest, CreateDirectoriesFail) {
+ // force a failure by putting a file where we will try to create a directory.
+ CreateTextFile(testdata_file_.c_str());
+ EXPECT_TRUE(testdata_file_.FileOrDirectoryExists());
+ EXPECT_FALSE(testdata_file_.DirectoryExists());
+ EXPECT_FALSE(testdata_file_.CreateDirectoriesRecursively());
+}
+
+TEST(NoDirectoryCreationTest, CreateNoDirectoriesForDefaultXmlFile) {
+ const FilePath test_detail_xml("test_detail.xml");
+ EXPECT_FALSE(test_detail_xml.CreateDirectoriesRecursively());
+}
+
+TEST(FilePathTest, DefaultConstructor) {
+ FilePath fp;
+ EXPECT_STREQ("", fp.c_str());
+}
+
+TEST(FilePathTest, CharAndCopyConstructors) {
+ const FilePath fp("spicy");
+ EXPECT_STREQ("spicy", fp.c_str());
+
+ const FilePath fp_copy(fp);
+ EXPECT_STREQ("spicy", fp_copy.c_str());
+}
+
+TEST(FilePathTest, StringConstructor) {
+ const FilePath fp(String("cider"));
+ EXPECT_STREQ("cider", fp.c_str());
+}
+
+TEST(FilePathTest, Set) {
+ const FilePath apple("apple");
+ FilePath mac("mac");
+ mac.Set(apple); // Implement Set() since overloading operator= is forbidden.
+ EXPECT_STREQ("apple", mac.c_str());
+ EXPECT_STREQ("apple", apple.c_str());
+}
+
+TEST(FilePathTest, ToString) {
+ const FilePath file("drink");
+ String str(file.ToString());
+ EXPECT_STREQ("drink", str.c_str());
+}
+
+TEST(FilePathTest, RemoveExtension) {
+ EXPECT_STREQ("app", FilePath("app.exe").RemoveExtension("exe").c_str());
+ EXPECT_STREQ("APP", FilePath("APP.EXE").RemoveExtension("exe").c_str());
+}
+
+TEST(FilePathTest, RemoveExtensionWhenThereIsNoExtension) {
+ EXPECT_STREQ("app", FilePath("app").RemoveExtension("exe").c_str());
+}
+
+TEST(FilePathTest, IsDirectory) {
+ EXPECT_FALSE(FilePath("cola").IsDirectory());
+ EXPECT_TRUE(FilePath("koala" GTEST_PATH_SEP_).IsDirectory());
+#if GTEST_HAS_ALT_PATH_SEP_
+ EXPECT_TRUE(FilePath("koala/").IsDirectory());
+#endif
+}
+
+TEST(FilePathTest, IsAbsolutePath) {
+ EXPECT_FALSE(FilePath("is" GTEST_PATH_SEP_ "relative").IsAbsolutePath());
+ EXPECT_FALSE(FilePath("").IsAbsolutePath());
+#if GTEST_OS_WINDOWS
+ EXPECT_TRUE(FilePath("c:\\" GTEST_PATH_SEP_ "is_not"
+ GTEST_PATH_SEP_ "relative").IsAbsolutePath());
+ EXPECT_FALSE(FilePath("c:foo" GTEST_PATH_SEP_ "bar").IsAbsolutePath());
+ EXPECT_TRUE(FilePath("c:/" GTEST_PATH_SEP_ "is_not"
+ GTEST_PATH_SEP_ "relative").IsAbsolutePath());
+#else
+ EXPECT_TRUE(FilePath(GTEST_PATH_SEP_ "is_not" GTEST_PATH_SEP_ "relative")
+ .IsAbsolutePath());
+#endif // GTEST_OS_WINDOWS
+}
+
+TEST(FilePathTest, IsRootDirectory) {
+#if GTEST_OS_WINDOWS
+ EXPECT_TRUE(FilePath("a:\\").IsRootDirectory());
+ EXPECT_TRUE(FilePath("Z:/").IsRootDirectory());
+ EXPECT_TRUE(FilePath("e://").IsRootDirectory());
+ EXPECT_FALSE(FilePath("").IsRootDirectory());
+ EXPECT_FALSE(FilePath("b:").IsRootDirectory());
+ EXPECT_FALSE(FilePath("b:a").IsRootDirectory());
+ EXPECT_FALSE(FilePath("8:/").IsRootDirectory());
+ EXPECT_FALSE(FilePath("c|/").IsRootDirectory());
+#else
+ EXPECT_TRUE(FilePath("/").IsRootDirectory());
+ EXPECT_TRUE(FilePath("//").IsRootDirectory());
+ EXPECT_FALSE(FilePath("").IsRootDirectory());
+ EXPECT_FALSE(FilePath("\\").IsRootDirectory());
+ EXPECT_FALSE(FilePath("/x").IsRootDirectory());
+#endif
+}
+
+} // namespace
+} // namespace internal
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/test/gtest-linked_ptr_test.cc b/libvpx/third_party/googletest/src/test/gtest-linked_ptr_test.cc
new file mode 100644
index 0000000..0d5508a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-linked_ptr_test.cc
@@ -0,0 +1,155 @@
+// Copyright 2003, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: Dan Egnor (egnor@google.com)
+// Ported to Windows: Vadim Berman (vadimb@google.com)
+
+#include "gtest/internal/gtest-linked_ptr.h"
+
+#include <stdlib.h>
+#include "gtest/gtest.h"
+
+namespace {
+
+using testing::Message;
+using testing::internal::linked_ptr;
+
+int num;
+Message* history = NULL;
+
+// Class which tracks allocation/deallocation
+class A {
+ public:
+ A(): mynum(num++) { *history << "A" << mynum << " ctor\n"; }
+ virtual ~A() { *history << "A" << mynum << " dtor\n"; }
+ virtual void Use() { *history << "A" << mynum << " use\n"; }
+ protected:
+ int mynum;
+};
+
+// Subclass
+class B : public A {
+ public:
+ B() { *history << "B" << mynum << " ctor\n"; }
+ ~B() { *history << "B" << mynum << " dtor\n"; }
+ virtual void Use() { *history << "B" << mynum << " use\n"; }
+};
+
+class LinkedPtrTest : public testing::Test {
+ public:
+ LinkedPtrTest() {
+ num = 0;
+ history = new Message;
+ }
+
+ virtual ~LinkedPtrTest() {
+ delete history;
+ history = NULL;
+ }
+};
+
+TEST_F(LinkedPtrTest, GeneralTest) {
+ {
+ linked_ptr<A> a0, a1, a2;
+ // Use explicit function call notation here to suppress self-assign warning.
+ a0.operator=(a0);
+ a1 = a2;
+ ASSERT_EQ(a0.get(), static_cast<A*>(NULL));
+ ASSERT_EQ(a1.get(), static_cast<A*>(NULL));
+ ASSERT_EQ(a2.get(), static_cast<A*>(NULL));
+ ASSERT_TRUE(a0 == NULL);
+ ASSERT_TRUE(a1 == NULL);
+ ASSERT_TRUE(a2 == NULL);
+
+ {
+ linked_ptr<A> a3(new A);
+ a0 = a3;
+ ASSERT_TRUE(a0 == a3);
+ ASSERT_TRUE(a0 != NULL);
+ ASSERT_TRUE(a0.get() == a3);
+ ASSERT_TRUE(a0 == a3.get());
+ linked_ptr<A> a4(a0);
+ a1 = a4;
+ linked_ptr<A> a5(new A);
+ ASSERT_TRUE(a5.get() != a3);
+ ASSERT_TRUE(a5 != a3.get());
+ a2 = a5;
+ linked_ptr<B> b0(new B);
+ linked_ptr<A> a6(b0);
+ ASSERT_TRUE(b0 == a6);
+ ASSERT_TRUE(a6 == b0);
+ ASSERT_TRUE(b0 != NULL);
+ a5 = b0;
+ a5 = b0;
+ a3->Use();
+ a4->Use();
+ a5->Use();
+ a6->Use();
+ b0->Use();
+ (*b0).Use();
+ b0.get()->Use();
+ }
+
+ a0->Use();
+ a1->Use();
+ a2->Use();
+
+ a1 = a2;
+ a2.reset(new A);
+ a0.reset();
+
+ linked_ptr<A> a7;
+ }
+
+ ASSERT_STREQ(
+ "A0 ctor\n"
+ "A1 ctor\n"
+ "A2 ctor\n"
+ "B2 ctor\n"
+ "A0 use\n"
+ "A0 use\n"
+ "B2 use\n"
+ "B2 use\n"
+ "B2 use\n"
+ "B2 use\n"
+ "B2 use\n"
+ "B2 dtor\n"
+ "A2 dtor\n"
+ "A0 use\n"
+ "A0 use\n"
+ "A1 use\n"
+ "A3 ctor\n"
+ "A0 dtor\n"
+ "A3 dtor\n"
+ "A1 dtor\n",
+ history->GetString().c_str()
+ );
+}
+
+} // Unnamed namespace
diff --git a/libvpx/third_party/googletest/src/test/gtest-listener_test.cc b/libvpx/third_party/googletest/src/test/gtest-listener_test.cc
new file mode 100644
index 0000000..2aa08ef
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-listener_test.cc
@@ -0,0 +1,313 @@
+// Copyright 2009 Google Inc. All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This file verifies Google Test event listeners receive events at the
+// right times.
+
+#include "gtest/gtest.h"
+#include <vector>
+
+using ::testing::AddGlobalTestEnvironment;
+using ::testing::Environment;
+using ::testing::InitGoogleTest;
+using ::testing::Test;
+using ::testing::TestCase;
+using ::testing::TestEventListener;
+using ::testing::TestInfo;
+using ::testing::TestPartResult;
+using ::testing::UnitTest;
+using ::testing::internal::String;
+
+// Used by tests to register their events.
+std::vector<String>* g_events = NULL;
+
+namespace testing {
+namespace internal {
+
+class EventRecordingListener : public TestEventListener {
+ public:
+ EventRecordingListener(const char* name) : name_(name) {}
+
+ protected:
+ virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {
+ g_events->push_back(GetFullMethodName("OnTestProgramStart"));
+ }
+
+ virtual void OnTestIterationStart(const UnitTest& /*unit_test*/,
+ int iteration) {
+ Message message;
+ message << GetFullMethodName("OnTestIterationStart")
+ << "(" << iteration << ")";
+ g_events->push_back(message.GetString());
+ }
+
+ virtual void OnEnvironmentsSetUpStart(const UnitTest& /*unit_test*/) {
+ g_events->push_back(GetFullMethodName("OnEnvironmentsSetUpStart"));
+ }
+
+ virtual void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) {
+ g_events->push_back(GetFullMethodName("OnEnvironmentsSetUpEnd"));
+ }
+
+ virtual void OnTestCaseStart(const TestCase& /*test_case*/) {
+ g_events->push_back(GetFullMethodName("OnTestCaseStart"));
+ }
+
+ virtual void OnTestStart(const TestInfo& /*test_info*/) {
+ g_events->push_back(GetFullMethodName("OnTestStart"));
+ }
+
+ virtual void OnTestPartResult(const TestPartResult& /*test_part_result*/) {
+ g_events->push_back(GetFullMethodName("OnTestPartResult"));
+ }
+
+ virtual void OnTestEnd(const TestInfo& /*test_info*/) {
+ g_events->push_back(GetFullMethodName("OnTestEnd"));
+ }
+
+ virtual void OnTestCaseEnd(const TestCase& /*test_case*/) {
+ g_events->push_back(GetFullMethodName("OnTestCaseEnd"));
+ }
+
+ virtual void OnEnvironmentsTearDownStart(const UnitTest& /*unit_test*/) {
+ g_events->push_back(GetFullMethodName("OnEnvironmentsTearDownStart"));
+ }
+
+ virtual void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) {
+ g_events->push_back(GetFullMethodName("OnEnvironmentsTearDownEnd"));
+ }
+
+ virtual void OnTestIterationEnd(const UnitTest& /*unit_test*/,
+ int iteration) {
+ Message message;
+ message << GetFullMethodName("OnTestIterationEnd")
+ << "(" << iteration << ")";
+ g_events->push_back(message.GetString());
+ }
+
+ virtual void OnTestProgramEnd(const UnitTest& /*unit_test*/) {
+ g_events->push_back(GetFullMethodName("OnTestProgramEnd"));
+ }
+
+ private:
+ String GetFullMethodName(const char* name) {
+ Message message;
+ message << name_ << "." << name;
+ return message.GetString();
+ }
+
+ String name_;
+};
+
+class EnvironmentInvocationCatcher : public Environment {
+ protected:
+ virtual void SetUp() {
+ g_events->push_back(String("Environment::SetUp"));
+ }
+
+ virtual void TearDown() {
+ g_events->push_back(String("Environment::TearDown"));
+ }
+};
+
+class ListenerTest : public Test {
+ protected:
+ static void SetUpTestCase() {
+ g_events->push_back(String("ListenerTest::SetUpTestCase"));
+ }
+
+ static void TearDownTestCase() {
+ g_events->push_back(String("ListenerTest::TearDownTestCase"));
+ }
+
+ virtual void SetUp() {
+ g_events->push_back(String("ListenerTest::SetUp"));
+ }
+
+ virtual void TearDown() {
+ g_events->push_back(String("ListenerTest::TearDown"));
+ }
+};
+
+TEST_F(ListenerTest, DoesFoo) {
+ // Test execution order within a test case is not guaranteed so we are not
+ // recording the test name.
+ g_events->push_back(String("ListenerTest::* Test Body"));
+ SUCCEED(); // Triggers OnTestPartResult.
+}
+
+TEST_F(ListenerTest, DoesBar) {
+ g_events->push_back(String("ListenerTest::* Test Body"));
+ SUCCEED(); // Triggers OnTestPartResult.
+}
+
+} // namespace internal
+
+} // namespace testing
+
+using ::testing::internal::EnvironmentInvocationCatcher;
+using ::testing::internal::EventRecordingListener;
+
+void VerifyResults(const std::vector<String>& data,
+ const char* const* expected_data,
+ int expected_data_size) {
+ const int actual_size = data.size();
+ // If the following assertion fails, a new entry will be appended to
+ // data. Hence we save data.size() first.
+ EXPECT_EQ(expected_data_size, actual_size);
+
+ // Compares the common prefix.
+ const int shorter_size = expected_data_size <= actual_size ?
+ expected_data_size : actual_size;
+ int i = 0;
+ for (; i < shorter_size; ++i) {
+ ASSERT_STREQ(expected_data[i], data[i].c_str())
+ << "at position " << i;
+ }
+
+ // Prints extra elements in the actual data.
+ for (; i < actual_size; ++i) {
+ printf(" Actual event #%d: %s\n", i, data[i].c_str());
+ }
+}
+
+int main(int argc, char **argv) {
+ std::vector<String> events;
+ g_events = &events;
+ InitGoogleTest(&argc, argv);
+
+ UnitTest::GetInstance()->listeners().Append(
+ new EventRecordingListener("1st"));
+ UnitTest::GetInstance()->listeners().Append(
+ new EventRecordingListener("2nd"));
+
+ AddGlobalTestEnvironment(new EnvironmentInvocationCatcher);
+
+ GTEST_CHECK_(events.size() == 0)
+ << "AddGlobalTestEnvironment should not generate any events itself.";
+
+ ::testing::GTEST_FLAG(repeat) = 2;
+ int ret_val = RUN_ALL_TESTS();
+
+ const char* const expected_events[] = {
+ "1st.OnTestProgramStart",
+ "2nd.OnTestProgramStart",
+ "1st.OnTestIterationStart(0)",
+ "2nd.OnTestIterationStart(0)",
+ "1st.OnEnvironmentsSetUpStart",
+ "2nd.OnEnvironmentsSetUpStart",
+ "Environment::SetUp",
+ "2nd.OnEnvironmentsSetUpEnd",
+ "1st.OnEnvironmentsSetUpEnd",
+ "1st.OnTestCaseStart",
+ "2nd.OnTestCaseStart",
+ "ListenerTest::SetUpTestCase",
+ "1st.OnTestStart",
+ "2nd.OnTestStart",
+ "ListenerTest::SetUp",
+ "ListenerTest::* Test Body",
+ "1st.OnTestPartResult",
+ "2nd.OnTestPartResult",
+ "ListenerTest::TearDown",
+ "2nd.OnTestEnd",
+ "1st.OnTestEnd",
+ "1st.OnTestStart",
+ "2nd.OnTestStart",
+ "ListenerTest::SetUp",
+ "ListenerTest::* Test Body",
+ "1st.OnTestPartResult",
+ "2nd.OnTestPartResult",
+ "ListenerTest::TearDown",
+ "2nd.OnTestEnd",
+ "1st.OnTestEnd",
+ "ListenerTest::TearDownTestCase",
+ "2nd.OnTestCaseEnd",
+ "1st.OnTestCaseEnd",
+ "1st.OnEnvironmentsTearDownStart",
+ "2nd.OnEnvironmentsTearDownStart",
+ "Environment::TearDown",
+ "2nd.OnEnvironmentsTearDownEnd",
+ "1st.OnEnvironmentsTearDownEnd",
+ "2nd.OnTestIterationEnd(0)",
+ "1st.OnTestIterationEnd(0)",
+ "1st.OnTestIterationStart(1)",
+ "2nd.OnTestIterationStart(1)",
+ "1st.OnEnvironmentsSetUpStart",
+ "2nd.OnEnvironmentsSetUpStart",
+ "Environment::SetUp",
+ "2nd.OnEnvironmentsSetUpEnd",
+ "1st.OnEnvironmentsSetUpEnd",
+ "1st.OnTestCaseStart",
+ "2nd.OnTestCaseStart",
+ "ListenerTest::SetUpTestCase",
+ "1st.OnTestStart",
+ "2nd.OnTestStart",
+ "ListenerTest::SetUp",
+ "ListenerTest::* Test Body",
+ "1st.OnTestPartResult",
+ "2nd.OnTestPartResult",
+ "ListenerTest::TearDown",
+ "2nd.OnTestEnd",
+ "1st.OnTestEnd",
+ "1st.OnTestStart",
+ "2nd.OnTestStart",
+ "ListenerTest::SetUp",
+ "ListenerTest::* Test Body",
+ "1st.OnTestPartResult",
+ "2nd.OnTestPartResult",
+ "ListenerTest::TearDown",
+ "2nd.OnTestEnd",
+ "1st.OnTestEnd",
+ "ListenerTest::TearDownTestCase",
+ "2nd.OnTestCaseEnd",
+ "1st.OnTestCaseEnd",
+ "1st.OnEnvironmentsTearDownStart",
+ "2nd.OnEnvironmentsTearDownStart",
+ "Environment::TearDown",
+ "2nd.OnEnvironmentsTearDownEnd",
+ "1st.OnEnvironmentsTearDownEnd",
+ "2nd.OnTestIterationEnd(1)",
+ "1st.OnTestIterationEnd(1)",
+ "2nd.OnTestProgramEnd",
+ "1st.OnTestProgramEnd"
+ };
+ VerifyResults(events,
+ expected_events,
+ sizeof(expected_events)/sizeof(expected_events[0]));
+
+ // We need to check manually for ad hoc test failures that happen after
+ // RUN_ALL_TESTS finishes.
+ if (UnitTest::GetInstance()->Failed())
+ ret_val = 1;
+
+ return ret_val;
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest-message_test.cc b/libvpx/third_party/googletest/src/test/gtest-message_test.cc
new file mode 100644
index 0000000..c09c6a8
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-message_test.cc
@@ -0,0 +1,166 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Tests for the Message class.
+
+#include "gtest/gtest-message.h"
+
+#include "gtest/gtest.h"
+
+namespace {
+
+using ::testing::Message;
+
+// A helper function that turns a Message into a C string.
+const char* ToCString(const Message& msg) {
+ static testing::internal::String result;
+ result = msg.GetString();
+ return result.c_str();
+}
+
+// Tests the testing::Message class
+
+// Tests the default constructor.
+TEST(MessageTest, DefaultConstructor) {
+ const Message msg;
+ EXPECT_STREQ("", ToCString(msg));
+}
+
+// Tests the copy constructor.
+TEST(MessageTest, CopyConstructor) {
+ const Message msg1("Hello");
+ const Message msg2(msg1);
+ EXPECT_STREQ("Hello", ToCString(msg2));
+}
+
+// Tests constructing a Message from a C-string.
+TEST(MessageTest, ConstructsFromCString) {
+ Message msg("Hello");
+ EXPECT_STREQ("Hello", ToCString(msg));
+}
+
+// Tests streaming a float.
+TEST(MessageTest, StreamsFloat) {
+ const char* const s = ToCString(Message() << 1.23456F << " " << 2.34567F);
+ // Both numbers should be printed with enough precision.
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "1.234560", s);
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, " 2.345669", s);
+}
+
+// Tests streaming a double.
+TEST(MessageTest, StreamsDouble) {
+ const char* const s = ToCString(Message() << 1260570880.4555497 << " "
+ << 1260572265.1954534);
+ // Both numbers should be printed with enough precision.
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "1260570880.45", s);
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, " 1260572265.19", s);
+}
+
+// Tests streaming a non-char pointer.
+TEST(MessageTest, StreamsPointer) {
+ int n = 0;
+ int* p = &n;
+ EXPECT_STRNE("(null)", ToCString(Message() << p));
+}
+
+// Tests streaming a NULL non-char pointer.
+TEST(MessageTest, StreamsNullPointer) {
+ int* p = NULL;
+ EXPECT_STREQ("(null)", ToCString(Message() << p));
+}
+
+// Tests streaming a C string.
+TEST(MessageTest, StreamsCString) {
+ EXPECT_STREQ("Foo", ToCString(Message() << "Foo"));
+}
+
+// Tests streaming a NULL C string.
+TEST(MessageTest, StreamsNullCString) {
+ char* p = NULL;
+ EXPECT_STREQ("(null)", ToCString(Message() << p));
+}
+
+// Tests streaming std::string.
+TEST(MessageTest, StreamsString) {
+ const ::std::string str("Hello");
+ EXPECT_STREQ("Hello", ToCString(Message() << str));
+}
+
+// Tests that we can output strings containing embedded NULs.
+TEST(MessageTest, StreamsStringWithEmbeddedNUL) {
+ const char char_array_with_nul[] =
+ "Here's a NUL\0 and some more string";
+ const ::std::string string_with_nul(char_array_with_nul,
+ sizeof(char_array_with_nul) - 1);
+ EXPECT_STREQ("Here's a NUL\\0 and some more string",
+ ToCString(Message() << string_with_nul));
+}
+
+// Tests streaming a NUL char.
+TEST(MessageTest, StreamsNULChar) {
+ EXPECT_STREQ("\\0", ToCString(Message() << '\0'));
+}
+
+// Tests streaming int.
+TEST(MessageTest, StreamsInt) {
+ EXPECT_STREQ("123", ToCString(Message() << 123));
+}
+
+// Tests that basic IO manipulators (endl, ends, and flush) can be
+// streamed to Message.
+TEST(MessageTest, StreamsBasicIoManip) {
+ EXPECT_STREQ("Line 1.\nA NUL char \\0 in line 2.",
+ ToCString(Message() << "Line 1." << std::endl
+ << "A NUL char " << std::ends << std::flush
+ << " in line 2."));
+}
+
+// Tests Message::GetString()
+TEST(MessageTest, GetString) {
+ Message msg;
+ msg << 1 << " lamb";
+ EXPECT_STREQ("1 lamb", msg.GetString().c_str());
+}
+
+// Tests streaming a Message object to an ostream.
+TEST(MessageTest, StreamsToOStream) {
+ Message msg("Hello");
+ ::std::stringstream ss;
+ ss << msg;
+ EXPECT_STREQ("Hello", testing::internal::StringStreamToString(&ss).c_str());
+}
+
+// Tests that a Message object doesn't take up too much stack space.
+TEST(MessageTest, DoesNotTakeUpMuchStackSpace) {
+ EXPECT_LE(sizeof(Message), 16U);
+}
+
+} // namespace
diff --git a/libvpx/third_party/googletest/src/test/gtest-options_test.cc b/libvpx/third_party/googletest/src/test/gtest-options_test.cc
new file mode 100644
index 0000000..9e98f3f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-options_test.cc
@@ -0,0 +1,212 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: keith.ray@gmail.com (Keith Ray)
+//
+// Google Test UnitTestOptions tests
+//
+// This file tests classes and functions used internally by
+// Google Test. They are subject to change without notice.
+//
+// This file is #included from gtest.cc, to avoid changing build or
+// make-files on Windows and other platforms. Do not #include this file
+// anywhere else!
+
+#include "gtest/gtest.h"
+
+#if GTEST_OS_WINDOWS_MOBILE
+# include <windows.h>
+#elif GTEST_OS_WINDOWS
+# include <direct.h>
+#endif // GTEST_OS_WINDOWS_MOBILE
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+namespace internal {
+namespace {
+
+// Turns the given relative path into an absolute path.
+FilePath GetAbsolutePathOf(const FilePath& relative_path) {
+ return FilePath::ConcatPaths(FilePath::GetCurrentDir(), relative_path);
+}
+
+// Testing UnitTestOptions::GetOutputFormat/GetOutputFile.
+
+TEST(XmlOutputTest, GetOutputFormatDefault) {
+ GTEST_FLAG(output) = "";
+ EXPECT_STREQ("", UnitTestOptions::GetOutputFormat().c_str());
+}
+
+TEST(XmlOutputTest, GetOutputFormat) {
+ GTEST_FLAG(output) = "xml:filename";
+ EXPECT_STREQ("xml", UnitTestOptions::GetOutputFormat().c_str());
+}
+
+TEST(XmlOutputTest, GetOutputFileDefault) {
+ GTEST_FLAG(output) = "";
+ EXPECT_STREQ(GetAbsolutePathOf(FilePath("test_detail.xml")).c_str(),
+ UnitTestOptions::GetAbsolutePathToOutputFile().c_str());
+}
+
+TEST(XmlOutputTest, GetOutputFileSingleFile) {
+ GTEST_FLAG(output) = "xml:filename.abc";
+ EXPECT_STREQ(GetAbsolutePathOf(FilePath("filename.abc")).c_str(),
+ UnitTestOptions::GetAbsolutePathToOutputFile().c_str());
+}
+
+TEST(XmlOutputTest, GetOutputFileFromDirectoryPath) {
+ GTEST_FLAG(output) = "xml:path" GTEST_PATH_SEP_;
+ const std::string expected_output_file =
+ GetAbsolutePathOf(
+ FilePath(std::string("path") + GTEST_PATH_SEP_ +
+ GetCurrentExecutableName().c_str() + ".xml")).c_str();
+ const String& output_file = UnitTestOptions::GetAbsolutePathToOutputFile();
+#if GTEST_OS_WINDOWS
+ EXPECT_STRCASEEQ(expected_output_file.c_str(), output_file.c_str());
+#else
+ EXPECT_EQ(expected_output_file, output_file.c_str());
+#endif
+}
+
+TEST(OutputFileHelpersTest, GetCurrentExecutableName) {
+ const std::string exe_str = GetCurrentExecutableName().c_str();
+#if GTEST_OS_WINDOWS
+ const bool success =
+ _strcmpi("gtest-options_test", exe_str.c_str()) == 0 ||
+ _strcmpi("gtest-options-ex_test", exe_str.c_str()) == 0 ||
+ _strcmpi("gtest_all_test", exe_str.c_str()) == 0 ||
+ _strcmpi("gtest_dll_test", exe_str.c_str()) == 0;
+#else
+ // TODO(wan@google.com): remove the hard-coded "lt-" prefix when
+ // Chandler Carruth's libtool replacement is ready.
+ const bool success =
+ exe_str == "gtest-options_test" ||
+ exe_str == "gtest_all_test" ||
+ exe_str == "lt-gtest_all_test" ||
+ exe_str == "gtest_dll_test";
+#endif // GTEST_OS_WINDOWS
+ if (!success)
+ FAIL() << "GetCurrentExecutableName() returns " << exe_str;
+}
+
+class XmlOutputChangeDirTest : public Test {
+ protected:
+ virtual void SetUp() {
+ original_working_dir_ = FilePath::GetCurrentDir();
+ posix::ChDir("..");
+ // This will make the test fail if run from the root directory.
+ EXPECT_STRNE(original_working_dir_.c_str(),
+ FilePath::GetCurrentDir().c_str());
+ }
+
+ virtual void TearDown() {
+ posix::ChDir(original_working_dir_.c_str());
+ }
+
+ FilePath original_working_dir_;
+};
+
+TEST_F(XmlOutputChangeDirTest, PreserveOriginalWorkingDirWithDefault) {
+ GTEST_FLAG(output) = "";
+ EXPECT_STREQ(FilePath::ConcatPaths(original_working_dir_,
+ FilePath("test_detail.xml")).c_str(),
+ UnitTestOptions::GetAbsolutePathToOutputFile().c_str());
+}
+
+TEST_F(XmlOutputChangeDirTest, PreserveOriginalWorkingDirWithDefaultXML) {
+ GTEST_FLAG(output) = "xml";
+ EXPECT_STREQ(FilePath::ConcatPaths(original_working_dir_,
+ FilePath("test_detail.xml")).c_str(),
+ UnitTestOptions::GetAbsolutePathToOutputFile().c_str());
+}
+
+TEST_F(XmlOutputChangeDirTest, PreserveOriginalWorkingDirWithRelativeFile) {
+ GTEST_FLAG(output) = "xml:filename.abc";
+ EXPECT_STREQ(FilePath::ConcatPaths(original_working_dir_,
+ FilePath("filename.abc")).c_str(),
+ UnitTestOptions::GetAbsolutePathToOutputFile().c_str());
+}
+
+TEST_F(XmlOutputChangeDirTest, PreserveOriginalWorkingDirWithRelativePath) {
+ GTEST_FLAG(output) = "xml:path" GTEST_PATH_SEP_;
+ const std::string expected_output_file =
+ FilePath::ConcatPaths(
+ original_working_dir_,
+ FilePath(std::string("path") + GTEST_PATH_SEP_ +
+ GetCurrentExecutableName().c_str() + ".xml")).c_str();
+ const String& output_file = UnitTestOptions::GetAbsolutePathToOutputFile();
+#if GTEST_OS_WINDOWS
+ EXPECT_STRCASEEQ(expected_output_file.c_str(), output_file.c_str());
+#else
+ EXPECT_EQ(expected_output_file, output_file.c_str());
+#endif
+}
+
+TEST_F(XmlOutputChangeDirTest, PreserveOriginalWorkingDirWithAbsoluteFile) {
+#if GTEST_OS_WINDOWS
+ GTEST_FLAG(output) = "xml:c:\\tmp\\filename.abc";
+ EXPECT_STREQ(FilePath("c:\\tmp\\filename.abc").c_str(),
+ UnitTestOptions::GetAbsolutePathToOutputFile().c_str());
+#else
+ GTEST_FLAG(output) ="xml:/tmp/filename.abc";
+ EXPECT_STREQ(FilePath("/tmp/filename.abc").c_str(),
+ UnitTestOptions::GetAbsolutePathToOutputFile().c_str());
+#endif
+}
+
+TEST_F(XmlOutputChangeDirTest, PreserveOriginalWorkingDirWithAbsolutePath) {
+#if GTEST_OS_WINDOWS
+ const std::string path = "c:\\tmp\\";
+#else
+ const std::string path = "/tmp/";
+#endif
+
+ GTEST_FLAG(output) = "xml:" + path;
+ const std::string expected_output_file =
+ path + GetCurrentExecutableName().c_str() + ".xml";
+ const String& output_file = UnitTestOptions::GetAbsolutePathToOutputFile();
+
+#if GTEST_OS_WINDOWS
+ EXPECT_STRCASEEQ(expected_output_file.c_str(), output_file.c_str());
+#else
+ EXPECT_EQ(expected_output_file, output_file.c_str());
+#endif
+}
+
+} // namespace
+} // namespace internal
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/test/gtest-param-test2_test.cc b/libvpx/third_party/googletest/src/test/gtest-param-test2_test.cc
new file mode 100644
index 0000000..4a782fe
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-param-test2_test.cc
@@ -0,0 +1,65 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+//
+// Tests for Google Test itself. This verifies that the basic constructs of
+// Google Test work.
+
+#include "gtest/gtest.h"
+
+#include "test/gtest-param-test_test.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+using ::testing::Values;
+using ::testing::internal::ParamGenerator;
+
+// Tests that generators defined in a different translation unit
+// are functional. The test using extern_gen is defined
+// in gtest-param-test_test.cc.
+ParamGenerator<int> extern_gen = Values(33);
+
+// Tests that a parameterized test case can be defined in one translation unit
+// and instantiated in another. The test is defined in gtest-param-test_test.cc
+// and ExternalInstantiationTest fixture class is defined in
+// gtest-param-test_test.h.
+INSTANTIATE_TEST_CASE_P(MultiplesOf33,
+ ExternalInstantiationTest,
+ Values(33, 66));
+
+// Tests that a parameterized test case can be instantiated
+// in multiple translation units. Another instantiation is defined
+// in gtest-param-test_test.cc and InstantiationInMultipleTranslaionUnitsTest
+// fixture is defined in gtest-param-test_test.h
+INSTANTIATE_TEST_CASE_P(Sequence2,
+ InstantiationInMultipleTranslaionUnitsTest,
+ Values(42*3, 42*4, 42*5));
+
+#endif // GTEST_HAS_PARAM_TEST
diff --git a/libvpx/third_party/googletest/src/test/gtest-param-test_test.cc b/libvpx/third_party/googletest/src/test/gtest-param-test_test.cc
new file mode 100644
index 0000000..94a53d9
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-param-test_test.cc
@@ -0,0 +1,895 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+//
+// Tests for Google Test itself. This file verifies that the parameter
+// generators objects produce correct parameter sequences and that
+// Google Test runtime instantiates correct tests from those sequences.
+
+#include "gtest/gtest.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+# include <algorithm>
+# include <iostream>
+# include <list>
+# include <sstream>
+# include <string>
+# include <vector>
+
+// To include gtest-internal-inl.h.
+# define GTEST_IMPLEMENTATION_ 1
+# include "src/gtest-internal-inl.h" // for UnitTestOptions
+# undef GTEST_IMPLEMENTATION_
+
+# include "test/gtest-param-test_test.h"
+
+using ::std::vector;
+using ::std::sort;
+
+using ::testing::AddGlobalTestEnvironment;
+using ::testing::Bool;
+using ::testing::Message;
+using ::testing::Range;
+using ::testing::TestWithParam;
+using ::testing::Values;
+using ::testing::ValuesIn;
+
+# if GTEST_HAS_COMBINE
+using ::testing::Combine;
+using ::std::tr1::get;
+using ::std::tr1::make_tuple;
+using ::std::tr1::tuple;
+# endif // GTEST_HAS_COMBINE
+
+using ::testing::internal::ParamGenerator;
+using ::testing::internal::UnitTestOptions;
+
+// Prints a value to a string.
+//
+// TODO(wan@google.com): remove PrintValue() when we move matchers and
+// EXPECT_THAT() from Google Mock to Google Test. At that time, we
+// can write EXPECT_THAT(x, Eq(y)) to compare two tuples x and y, as
+// EXPECT_THAT() and the matchers know how to print tuples.
+template <typename T>
+::std::string PrintValue(const T& value) {
+ ::std::stringstream stream;
+ stream << value;
+ return stream.str();
+}
+
+# if GTEST_HAS_COMBINE
+
+// These overloads allow printing tuples in our tests. We cannot
+// define an operator<< for tuples, as that definition needs to be in
+// the std namespace in order to be picked up by Google Test via
+// Argument-Dependent Lookup, yet defining anything in the std
+// namespace in non-STL code is undefined behavior.
+
+template <typename T1, typename T2>
+::std::string PrintValue(const tuple<T1, T2>& value) {
+ ::std::stringstream stream;
+ stream << "(" << get<0>(value) << ", " << get<1>(value) << ")";
+ return stream.str();
+}
+
+template <typename T1, typename T2, typename T3>
+::std::string PrintValue(const tuple<T1, T2, T3>& value) {
+ ::std::stringstream stream;
+ stream << "(" << get<0>(value) << ", " << get<1>(value)
+ << ", "<< get<2>(value) << ")";
+ return stream.str();
+}
+
+template <typename T1, typename T2, typename T3, typename T4, typename T5,
+ typename T6, typename T7, typename T8, typename T9, typename T10>
+::std::string PrintValue(
+ const tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10>& value) {
+ ::std::stringstream stream;
+ stream << "(" << get<0>(value) << ", " << get<1>(value)
+ << ", "<< get<2>(value) << ", " << get<3>(value)
+ << ", "<< get<4>(value) << ", " << get<5>(value)
+ << ", "<< get<6>(value) << ", " << get<7>(value)
+ << ", "<< get<8>(value) << ", " << get<9>(value) << ")";
+ return stream.str();
+}
+
+# endif // GTEST_HAS_COMBINE
+
+// Verifies that a sequence generated by the generator and accessed
+// via the iterator object matches the expected one using Google Test
+// assertions.
+template <typename T, size_t N>
+void VerifyGenerator(const ParamGenerator<T>& generator,
+ const T (&expected_values)[N]) {
+ typename ParamGenerator<T>::iterator it = generator.begin();
+ for (size_t i = 0; i < N; ++i) {
+ ASSERT_FALSE(it == generator.end())
+ << "At element " << i << " when accessing via an iterator "
+ << "created with the copy constructor.\n";
+ // We cannot use EXPECT_EQ() here as the values may be tuples,
+ // which don't support <<.
+ EXPECT_TRUE(expected_values[i] == *it)
+ << "where i is " << i
+ << ", expected_values[i] is " << PrintValue(expected_values[i])
+ << ", *it is " << PrintValue(*it)
+ << ", and 'it' is an iterator created with the copy constructor.\n";
+ it++;
+ }
+ EXPECT_TRUE(it == generator.end())
+ << "At the presumed end of sequence when accessing via an iterator "
+ << "created with the copy constructor.\n";
+
+ // Test the iterator assignment. The following lines verify that
+ // the sequence accessed via an iterator initialized via the
+ // assignment operator (as opposed to a copy constructor) matches
+ // just the same.
+ it = generator.begin();
+ for (size_t i = 0; i < N; ++i) {
+ ASSERT_FALSE(it == generator.end())
+ << "At element " << i << " when accessing via an iterator "
+ << "created with the assignment operator.\n";
+ EXPECT_TRUE(expected_values[i] == *it)
+ << "where i is " << i
+ << ", expected_values[i] is " << PrintValue(expected_values[i])
+ << ", *it is " << PrintValue(*it)
+ << ", and 'it' is an iterator created with the copy constructor.\n";
+ it++;
+ }
+ EXPECT_TRUE(it == generator.end())
+ << "At the presumed end of sequence when accessing via an iterator "
+ << "created with the assignment operator.\n";
+}
+
+template <typename T>
+void VerifyGeneratorIsEmpty(const ParamGenerator<T>& generator) {
+ typename ParamGenerator<T>::iterator it = generator.begin();
+ EXPECT_TRUE(it == generator.end());
+
+ it = generator.begin();
+ EXPECT_TRUE(it == generator.end());
+}
+
+// Generator tests. They test that each of the provided generator functions
+// generates an expected sequence of values. The general test pattern
+// instantiates a generator using one of the generator functions,
+// checks the sequence produced by the generator using its iterator API,
+// and then resets the iterator back to the beginning of the sequence
+// and checks the sequence again.
+
+// Tests that iterators produced by generator functions conform to the
+// ForwardIterator concept.
+TEST(IteratorTest, ParamIteratorConformsToForwardIteratorConcept) {
+ const ParamGenerator<int> gen = Range(0, 10);
+ ParamGenerator<int>::iterator it = gen.begin();
+
+ // Verifies that iterator initialization works as expected.
+ ParamGenerator<int>::iterator it2 = it;
+ EXPECT_TRUE(*it == *it2) << "Initialized iterators must point to the "
+ << "element same as its source points to";
+
+ // Verifies that iterator assignment works as expected.
+ it++;
+ EXPECT_FALSE(*it == *it2);
+ it2 = it;
+ EXPECT_TRUE(*it == *it2) << "Assigned iterators must point to the "
+ << "element same as its source points to";
+
+ // Verifies that prefix operator++() returns *this.
+ EXPECT_EQ(&it, &(++it)) << "Result of the prefix operator++ must be "
+ << "refer to the original object";
+
+ // Verifies that the result of the postfix operator++ points to the value
+ // pointed to by the original iterator.
+ int original_value = *it; // Have to compute it outside of macro call to be
+ // unaffected by the parameter evaluation order.
+ EXPECT_EQ(original_value, *(it++));
+
+ // Verifies that prefix and postfix operator++() advance an iterator
+ // all the same.
+ it2 = it;
+ it++;
+ ++it2;
+ EXPECT_TRUE(*it == *it2);
+}
+
+// Tests that Range() generates the expected sequence.
+TEST(RangeTest, IntRangeWithDefaultStep) {
+ const ParamGenerator<int> gen = Range(0, 3);
+ const int expected_values[] = {0, 1, 2};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Edge case. Tests that Range() generates the single element sequence
+// as expected when provided with range limits that are equal.
+TEST(RangeTest, IntRangeSingleValue) {
+ const ParamGenerator<int> gen = Range(0, 1);
+ const int expected_values[] = {0};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Edge case. Tests that Range() with generates empty sequence when
+// supplied with an empty range.
+TEST(RangeTest, IntRangeEmpty) {
+ const ParamGenerator<int> gen = Range(0, 0);
+ VerifyGeneratorIsEmpty(gen);
+}
+
+// Tests that Range() with custom step (greater then one) generates
+// the expected sequence.
+TEST(RangeTest, IntRangeWithCustomStep) {
+ const ParamGenerator<int> gen = Range(0, 9, 3);
+ const int expected_values[] = {0, 3, 6};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Tests that Range() with custom step (greater then one) generates
+// the expected sequence when the last element does not fall on the
+// upper range limit. Sequences generated by Range() must not have
+// elements beyond the range limits.
+TEST(RangeTest, IntRangeWithCustomStepOverUpperBound) {
+ const ParamGenerator<int> gen = Range(0, 4, 3);
+ const int expected_values[] = {0, 3};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Verifies that Range works with user-defined types that define
+// copy constructor, operator=(), operator+(), and operator<().
+class DogAdder {
+ public:
+ explicit DogAdder(const char* a_value) : value_(a_value) {}
+ DogAdder(const DogAdder& other) : value_(other.value_.c_str()) {}
+
+ DogAdder operator=(const DogAdder& other) {
+ if (this != &other)
+ value_ = other.value_;
+ return *this;
+ }
+ DogAdder operator+(const DogAdder& other) const {
+ Message msg;
+ msg << value_.c_str() << other.value_.c_str();
+ return DogAdder(msg.GetString().c_str());
+ }
+ bool operator<(const DogAdder& other) const {
+ return value_ < other.value_;
+ }
+ const ::testing::internal::String& value() const { return value_; }
+
+ private:
+ ::testing::internal::String value_;
+};
+
+TEST(RangeTest, WorksWithACustomType) {
+ const ParamGenerator<DogAdder> gen =
+ Range(DogAdder("cat"), DogAdder("catdogdog"), DogAdder("dog"));
+ ParamGenerator<DogAdder>::iterator it = gen.begin();
+
+ ASSERT_FALSE(it == gen.end());
+ EXPECT_STREQ("cat", it->value().c_str());
+
+ ASSERT_FALSE(++it == gen.end());
+ EXPECT_STREQ("catdog", it->value().c_str());
+
+ EXPECT_TRUE(++it == gen.end());
+}
+
+class IntWrapper {
+ public:
+ explicit IntWrapper(int a_value) : value_(a_value) {}
+ IntWrapper(const IntWrapper& other) : value_(other.value_) {}
+
+ IntWrapper operator=(const IntWrapper& other) {
+ value_ = other.value_;
+ return *this;
+ }
+ // operator+() adds a different type.
+ IntWrapper operator+(int other) const { return IntWrapper(value_ + other); }
+ bool operator<(const IntWrapper& other) const {
+ return value_ < other.value_;
+ }
+ int value() const { return value_; }
+
+ private:
+ int value_;
+};
+
+TEST(RangeTest, WorksWithACustomTypeWithDifferentIncrementType) {
+ const ParamGenerator<IntWrapper> gen = Range(IntWrapper(0), IntWrapper(2));
+ ParamGenerator<IntWrapper>::iterator it = gen.begin();
+
+ ASSERT_FALSE(it == gen.end());
+ EXPECT_EQ(0, it->value());
+
+ ASSERT_FALSE(++it == gen.end());
+ EXPECT_EQ(1, it->value());
+
+ EXPECT_TRUE(++it == gen.end());
+}
+
+// Tests that ValuesIn() with an array parameter generates
+// the expected sequence.
+TEST(ValuesInTest, ValuesInArray) {
+ int array[] = {3, 5, 8};
+ const ParamGenerator<int> gen = ValuesIn(array);
+ VerifyGenerator(gen, array);
+}
+
+// Tests that ValuesIn() with a const array parameter generates
+// the expected sequence.
+TEST(ValuesInTest, ValuesInConstArray) {
+ const int array[] = {3, 5, 8};
+ const ParamGenerator<int> gen = ValuesIn(array);
+ VerifyGenerator(gen, array);
+}
+
+// Edge case. Tests that ValuesIn() with an array parameter containing a
+// single element generates the single element sequence.
+TEST(ValuesInTest, ValuesInSingleElementArray) {
+ int array[] = {42};
+ const ParamGenerator<int> gen = ValuesIn(array);
+ VerifyGenerator(gen, array);
+}
+
+// Tests that ValuesIn() generates the expected sequence for an STL
+// container (vector).
+TEST(ValuesInTest, ValuesInVector) {
+ typedef ::std::vector<int> ContainerType;
+ ContainerType values;
+ values.push_back(3);
+ values.push_back(5);
+ values.push_back(8);
+ const ParamGenerator<int> gen = ValuesIn(values);
+
+ const int expected_values[] = {3, 5, 8};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Tests that ValuesIn() generates the expected sequence.
+TEST(ValuesInTest, ValuesInIteratorRange) {
+ typedef ::std::vector<int> ContainerType;
+ ContainerType values;
+ values.push_back(3);
+ values.push_back(5);
+ values.push_back(8);
+ const ParamGenerator<int> gen = ValuesIn(values.begin(), values.end());
+
+ const int expected_values[] = {3, 5, 8};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Edge case. Tests that ValuesIn() provided with an iterator range specifying a
+// single value generates a single-element sequence.
+TEST(ValuesInTest, ValuesInSingleElementIteratorRange) {
+ typedef ::std::vector<int> ContainerType;
+ ContainerType values;
+ values.push_back(42);
+ const ParamGenerator<int> gen = ValuesIn(values.begin(), values.end());
+
+ const int expected_values[] = {42};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Edge case. Tests that ValuesIn() provided with an empty iterator range
+// generates an empty sequence.
+TEST(ValuesInTest, ValuesInEmptyIteratorRange) {
+ typedef ::std::vector<int> ContainerType;
+ ContainerType values;
+ const ParamGenerator<int> gen = ValuesIn(values.begin(), values.end());
+
+ VerifyGeneratorIsEmpty(gen);
+}
+
+// Tests that the Values() generates the expected sequence.
+TEST(ValuesTest, ValuesWorks) {
+ const ParamGenerator<int> gen = Values(3, 5, 8);
+
+ const int expected_values[] = {3, 5, 8};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Tests that Values() generates the expected sequences from elements of
+// different types convertible to ParamGenerator's parameter type.
+TEST(ValuesTest, ValuesWorksForValuesOfCompatibleTypes) {
+ const ParamGenerator<double> gen = Values(3, 5.0f, 8.0);
+
+ const double expected_values[] = {3.0, 5.0, 8.0};
+ VerifyGenerator(gen, expected_values);
+}
+
+TEST(ValuesTest, ValuesWorksForMaxLengthList) {
+ const ParamGenerator<int> gen = Values(
+ 10, 20, 30, 40, 50, 60, 70, 80, 90, 100,
+ 110, 120, 130, 140, 150, 160, 170, 180, 190, 200,
+ 210, 220, 230, 240, 250, 260, 270, 280, 290, 300,
+ 310, 320, 330, 340, 350, 360, 370, 380, 390, 400,
+ 410, 420, 430, 440, 450, 460, 470, 480, 490, 500);
+
+ const int expected_values[] = {
+ 10, 20, 30, 40, 50, 60, 70, 80, 90, 100,
+ 110, 120, 130, 140, 150, 160, 170, 180, 190, 200,
+ 210, 220, 230, 240, 250, 260, 270, 280, 290, 300,
+ 310, 320, 330, 340, 350, 360, 370, 380, 390, 400,
+ 410, 420, 430, 440, 450, 460, 470, 480, 490, 500};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Edge case test. Tests that single-parameter Values() generates the sequence
+// with the single value.
+TEST(ValuesTest, ValuesWithSingleParameter) {
+ const ParamGenerator<int> gen = Values(42);
+
+ const int expected_values[] = {42};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Tests that Bool() generates sequence (false, true).
+TEST(BoolTest, BoolWorks) {
+ const ParamGenerator<bool> gen = Bool();
+
+ const bool expected_values[] = {false, true};
+ VerifyGenerator(gen, expected_values);
+}
+
+# if GTEST_HAS_COMBINE
+
+// Tests that Combine() with two parameters generates the expected sequence.
+TEST(CombineTest, CombineWithTwoParameters) {
+ const char* foo = "foo";
+ const char* bar = "bar";
+ const ParamGenerator<tuple<const char*, int> > gen =
+ Combine(Values(foo, bar), Values(3, 4));
+
+ tuple<const char*, int> expected_values[] = {
+ make_tuple(foo, 3), make_tuple(foo, 4),
+ make_tuple(bar, 3), make_tuple(bar, 4)};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Tests that Combine() with three parameters generates the expected sequence.
+TEST(CombineTest, CombineWithThreeParameters) {
+ const ParamGenerator<tuple<int, int, int> > gen = Combine(Values(0, 1),
+ Values(3, 4),
+ Values(5, 6));
+ tuple<int, int, int> expected_values[] = {
+ make_tuple(0, 3, 5), make_tuple(0, 3, 6),
+ make_tuple(0, 4, 5), make_tuple(0, 4, 6),
+ make_tuple(1, 3, 5), make_tuple(1, 3, 6),
+ make_tuple(1, 4, 5), make_tuple(1, 4, 6)};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Tests that the Combine() with the first parameter generating a single value
+// sequence generates a sequence with the number of elements equal to the
+// number of elements in the sequence generated by the second parameter.
+TEST(CombineTest, CombineWithFirstParameterSingleValue) {
+ const ParamGenerator<tuple<int, int> > gen = Combine(Values(42),
+ Values(0, 1));
+
+ tuple<int, int> expected_values[] = {make_tuple(42, 0), make_tuple(42, 1)};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Tests that the Combine() with the second parameter generating a single value
+// sequence generates a sequence with the number of elements equal to the
+// number of elements in the sequence generated by the first parameter.
+TEST(CombineTest, CombineWithSecondParameterSingleValue) {
+ const ParamGenerator<tuple<int, int> > gen = Combine(Values(0, 1),
+ Values(42));
+
+ tuple<int, int> expected_values[] = {make_tuple(0, 42), make_tuple(1, 42)};
+ VerifyGenerator(gen, expected_values);
+}
+
+// Tests that when the first parameter produces an empty sequence,
+// Combine() produces an empty sequence, too.
+TEST(CombineTest, CombineWithFirstParameterEmptyRange) {
+ const ParamGenerator<tuple<int, int> > gen = Combine(Range(0, 0),
+ Values(0, 1));
+ VerifyGeneratorIsEmpty(gen);
+}
+
+// Tests that when the second parameter produces an empty sequence,
+// Combine() produces an empty sequence, too.
+TEST(CombineTest, CombineWithSecondParameterEmptyRange) {
+ const ParamGenerator<tuple<int, int> > gen = Combine(Values(0, 1),
+ Range(1, 1));
+ VerifyGeneratorIsEmpty(gen);
+}
+
+// Edge case. Tests that combine works with the maximum number
+// of parameters supported by Google Test (currently 10).
+TEST(CombineTest, CombineWithMaxNumberOfParameters) {
+ const char* foo = "foo";
+ const char* bar = "bar";
+ const ParamGenerator<tuple<const char*, int, int, int, int, int, int, int,
+ int, int> > gen = Combine(Values(foo, bar),
+ Values(1), Values(2),
+ Values(3), Values(4),
+ Values(5), Values(6),
+ Values(7), Values(8),
+ Values(9));
+
+ tuple<const char*, int, int, int, int, int, int, int, int, int>
+ expected_values[] = {make_tuple(foo, 1, 2, 3, 4, 5, 6, 7, 8, 9),
+ make_tuple(bar, 1, 2, 3, 4, 5, 6, 7, 8, 9)};
+ VerifyGenerator(gen, expected_values);
+}
+
+# endif // GTEST_HAS_COMBINE
+
+// Tests that an generator produces correct sequence after being
+// assigned from another generator.
+TEST(ParamGeneratorTest, AssignmentWorks) {
+ ParamGenerator<int> gen = Values(1, 2);
+ const ParamGenerator<int> gen2 = Values(3, 4);
+ gen = gen2;
+
+ const int expected_values[] = {3, 4};
+ VerifyGenerator(gen, expected_values);
+}
+
+// This test verifies that the tests are expanded and run as specified:
+// one test per element from the sequence produced by the generator
+// specified in INSTANTIATE_TEST_CASE_P. It also verifies that the test's
+// fixture constructor, SetUp(), and TearDown() have run and have been
+// supplied with the correct parameters.
+
+// The use of environment object allows detection of the case where no test
+// case functionality is run at all. In this case TestCaseTearDown will not
+// be able to detect missing tests, naturally.
+template <int kExpectedCalls>
+class TestGenerationEnvironment : public ::testing::Environment {
+ public:
+ static TestGenerationEnvironment* Instance() {
+ static TestGenerationEnvironment* instance = new TestGenerationEnvironment;
+ return instance;
+ }
+
+ void FixtureConstructorExecuted() { fixture_constructor_count_++; }
+ void SetUpExecuted() { set_up_count_++; }
+ void TearDownExecuted() { tear_down_count_++; }
+ void TestBodyExecuted() { test_body_count_++; }
+
+ virtual void TearDown() {
+ // If all MultipleTestGenerationTest tests have been de-selected
+ // by the filter flag, the following checks make no sense.
+ bool perform_check = false;
+
+ for (int i = 0; i < kExpectedCalls; ++i) {
+ Message msg;
+ msg << "TestsExpandedAndRun/" << i;
+ if (UnitTestOptions::FilterMatchesTest(
+ "TestExpansionModule/MultipleTestGenerationTest",
+ msg.GetString().c_str())) {
+ perform_check = true;
+ }
+ }
+ if (perform_check) {
+ EXPECT_EQ(kExpectedCalls, fixture_constructor_count_)
+ << "Fixture constructor of ParamTestGenerationTest test case "
+ << "has not been run as expected.";
+ EXPECT_EQ(kExpectedCalls, set_up_count_)
+ << "Fixture SetUp method of ParamTestGenerationTest test case "
+ << "has not been run as expected.";
+ EXPECT_EQ(kExpectedCalls, tear_down_count_)
+ << "Fixture TearDown method of ParamTestGenerationTest test case "
+ << "has not been run as expected.";
+ EXPECT_EQ(kExpectedCalls, test_body_count_)
+ << "Test in ParamTestGenerationTest test case "
+ << "has not been run as expected.";
+ }
+ }
+ private:
+ TestGenerationEnvironment() : fixture_constructor_count_(0), set_up_count_(0),
+ tear_down_count_(0), test_body_count_(0) {}
+
+ int fixture_constructor_count_;
+ int set_up_count_;
+ int tear_down_count_;
+ int test_body_count_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestGenerationEnvironment);
+};
+
+const int test_generation_params[] = {36, 42, 72};
+
+class TestGenerationTest : public TestWithParam<int> {
+ public:
+ enum {
+ PARAMETER_COUNT =
+ sizeof(test_generation_params)/sizeof(test_generation_params[0])
+ };
+
+ typedef TestGenerationEnvironment<PARAMETER_COUNT> Environment;
+
+ TestGenerationTest() {
+ Environment::Instance()->FixtureConstructorExecuted();
+ current_parameter_ = GetParam();
+ }
+ virtual void SetUp() {
+ Environment::Instance()->SetUpExecuted();
+ EXPECT_EQ(current_parameter_, GetParam());
+ }
+ virtual void TearDown() {
+ Environment::Instance()->TearDownExecuted();
+ EXPECT_EQ(current_parameter_, GetParam());
+ }
+
+ static void SetUpTestCase() {
+ bool all_tests_in_test_case_selected = true;
+
+ for (int i = 0; i < PARAMETER_COUNT; ++i) {
+ Message test_name;
+ test_name << "TestsExpandedAndRun/" << i;
+ if ( !UnitTestOptions::FilterMatchesTest(
+ "TestExpansionModule/MultipleTestGenerationTest",
+ test_name.GetString())) {
+ all_tests_in_test_case_selected = false;
+ }
+ }
+ EXPECT_TRUE(all_tests_in_test_case_selected)
+ << "When running the TestGenerationTest test case all of its tests\n"
+ << "must be selected by the filter flag for the test case to pass.\n"
+ << "If not all of them are enabled, we can't reliably conclude\n"
+ << "that the correct number of tests have been generated.";
+
+ collected_parameters_.clear();
+ }
+
+ static void TearDownTestCase() {
+ vector<int> expected_values(test_generation_params,
+ test_generation_params + PARAMETER_COUNT);
+ // Test execution order is not guaranteed by Google Test,
+ // so the order of values in collected_parameters_ can be
+ // different and we have to sort to compare.
+ sort(expected_values.begin(), expected_values.end());
+ sort(collected_parameters_.begin(), collected_parameters_.end());
+
+ EXPECT_TRUE(collected_parameters_ == expected_values);
+ }
+ protected:
+ int current_parameter_;
+ static vector<int> collected_parameters_;
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(TestGenerationTest);
+};
+vector<int> TestGenerationTest::collected_parameters_;
+
+TEST_P(TestGenerationTest, TestsExpandedAndRun) {
+ Environment::Instance()->TestBodyExecuted();
+ EXPECT_EQ(current_parameter_, GetParam());
+ collected_parameters_.push_back(GetParam());
+}
+INSTANTIATE_TEST_CASE_P(TestExpansionModule, TestGenerationTest,
+ ValuesIn(test_generation_params));
+
+// This test verifies that the element sequence (third parameter of
+// INSTANTIATE_TEST_CASE_P) is evaluated in InitGoogleTest() and neither at
+// the call site of INSTANTIATE_TEST_CASE_P nor in RUN_ALL_TESTS(). For
+// that, we declare param_value_ to be a static member of
+// GeneratorEvaluationTest and initialize it to 0. We set it to 1 in
+// main(), just before invocation of InitGoogleTest(). After calling
+// InitGoogleTest(), we set the value to 2. If the sequence is evaluated
+// before or after InitGoogleTest, INSTANTIATE_TEST_CASE_P will create a
+// test with parameter other than 1, and the test body will fail the
+// assertion.
+class GeneratorEvaluationTest : public TestWithParam<int> {
+ public:
+ static int param_value() { return param_value_; }
+ static void set_param_value(int param_value) { param_value_ = param_value; }
+
+ private:
+ static int param_value_;
+};
+int GeneratorEvaluationTest::param_value_ = 0;
+
+TEST_P(GeneratorEvaluationTest, GeneratorsEvaluatedInMain) {
+ EXPECT_EQ(1, GetParam());
+}
+INSTANTIATE_TEST_CASE_P(GenEvalModule,
+ GeneratorEvaluationTest,
+ Values(GeneratorEvaluationTest::param_value()));
+
+// Tests that generators defined in a different translation unit are
+// functional. Generator extern_gen is defined in gtest-param-test_test2.cc.
+extern ParamGenerator<int> extern_gen;
+class ExternalGeneratorTest : public TestWithParam<int> {};
+TEST_P(ExternalGeneratorTest, ExternalGenerator) {
+ // Sequence produced by extern_gen contains only a single value
+ // which we verify here.
+ EXPECT_EQ(GetParam(), 33);
+}
+INSTANTIATE_TEST_CASE_P(ExternalGeneratorModule,
+ ExternalGeneratorTest,
+ extern_gen);
+
+// Tests that a parameterized test case can be defined in one translation
+// unit and instantiated in another. This test will be instantiated in
+// gtest-param-test_test2.cc. ExternalInstantiationTest fixture class is
+// defined in gtest-param-test_test.h.
+TEST_P(ExternalInstantiationTest, IsMultipleOf33) {
+ EXPECT_EQ(0, GetParam() % 33);
+}
+
+// Tests that a parameterized test case can be instantiated with multiple
+// generators.
+class MultipleInstantiationTest : public TestWithParam<int> {};
+TEST_P(MultipleInstantiationTest, AllowsMultipleInstances) {
+}
+INSTANTIATE_TEST_CASE_P(Sequence1, MultipleInstantiationTest, Values(1, 2));
+INSTANTIATE_TEST_CASE_P(Sequence2, MultipleInstantiationTest, Range(3, 5));
+
+// Tests that a parameterized test case can be instantiated
+// in multiple translation units. This test will be instantiated
+// here and in gtest-param-test_test2.cc.
+// InstantiationInMultipleTranslationUnitsTest fixture class
+// is defined in gtest-param-test_test.h.
+TEST_P(InstantiationInMultipleTranslaionUnitsTest, IsMultipleOf42) {
+ EXPECT_EQ(0, GetParam() % 42);
+}
+INSTANTIATE_TEST_CASE_P(Sequence1,
+ InstantiationInMultipleTranslaionUnitsTest,
+ Values(42, 42*2));
+
+// Tests that each iteration of parameterized test runs in a separate test
+// object.
+class SeparateInstanceTest : public TestWithParam<int> {
+ public:
+ SeparateInstanceTest() : count_(0) {}
+
+ static void TearDownTestCase() {
+ EXPECT_GE(global_count_, 2)
+ << "If some (but not all) SeparateInstanceTest tests have been "
+ << "filtered out this test will fail. Make sure that all "
+ << "GeneratorEvaluationTest are selected or de-selected together "
+ << "by the test filter.";
+ }
+
+ protected:
+ int count_;
+ static int global_count_;
+};
+int SeparateInstanceTest::global_count_ = 0;
+
+TEST_P(SeparateInstanceTest, TestsRunInSeparateInstances) {
+ EXPECT_EQ(0, count_++);
+ global_count_++;
+}
+INSTANTIATE_TEST_CASE_P(FourElemSequence, SeparateInstanceTest, Range(1, 4));
+
+// Tests that all instantiations of a test have named appropriately. Test
+// defined with TEST_P(TestCaseName, TestName) and instantiated with
+// INSTANTIATE_TEST_CASE_P(SequenceName, TestCaseName, generator) must be named
+// SequenceName/TestCaseName.TestName/i, where i is the 0-based index of the
+// sequence element used to instantiate the test.
+class NamingTest : public TestWithParam<int> {};
+
+TEST_P(NamingTest, TestsReportCorrectNamesAndParameters) {
+ const ::testing::TestInfo* const test_info =
+ ::testing::UnitTest::GetInstance()->current_test_info();
+
+ EXPECT_STREQ("ZeroToFiveSequence/NamingTest", test_info->test_case_name());
+
+ Message index_stream;
+ index_stream << "TestsReportCorrectNamesAndParameters/" << GetParam();
+ EXPECT_STREQ(index_stream.GetString().c_str(), test_info->name());
+
+ EXPECT_EQ(::testing::PrintToString(GetParam()), test_info->value_param());
+}
+
+INSTANTIATE_TEST_CASE_P(ZeroToFiveSequence, NamingTest, Range(0, 5));
+
+// Class that cannot be streamed into an ostream. It needs to be copyable
+// (and, in case of MSVC, also assignable) in order to be a test parameter
+// type. Its default copy constructor and assignment operator do exactly
+// what we need.
+class Unstreamable {
+ public:
+ explicit Unstreamable(int value) : value_(value) {}
+
+ private:
+ int value_;
+};
+
+class CommentTest : public TestWithParam<Unstreamable> {};
+
+TEST_P(CommentTest, TestsCorrectlyReportUnstreamableParams) {
+ const ::testing::TestInfo* const test_info =
+ ::testing::UnitTest::GetInstance()->current_test_info();
+
+ EXPECT_EQ(::testing::PrintToString(GetParam()), test_info->value_param());
+}
+
+INSTANTIATE_TEST_CASE_P(InstantiationWithComments,
+ CommentTest,
+ Values(Unstreamable(1)));
+
+// Verify that we can create a hierarchy of test fixtures, where the base
+// class fixture is not parameterized and the derived class is. In this case
+// ParameterizedDerivedTest inherits from NonParameterizedBaseTest. We
+// perform simple tests on both.
+class NonParameterizedBaseTest : public ::testing::Test {
+ public:
+ NonParameterizedBaseTest() : n_(17) { }
+ protected:
+ int n_;
+};
+
+class ParameterizedDerivedTest : public NonParameterizedBaseTest,
+ public ::testing::WithParamInterface<int> {
+ protected:
+ ParameterizedDerivedTest() : count_(0) { }
+ int count_;
+ static int global_count_;
+};
+
+int ParameterizedDerivedTest::global_count_ = 0;
+
+TEST_F(NonParameterizedBaseTest, FixtureIsInitialized) {
+ EXPECT_EQ(17, n_);
+}
+
+TEST_P(ParameterizedDerivedTest, SeesSequence) {
+ EXPECT_EQ(17, n_);
+ EXPECT_EQ(0, count_++);
+ EXPECT_EQ(GetParam(), global_count_++);
+}
+
+INSTANTIATE_TEST_CASE_P(RangeZeroToFive, ParameterizedDerivedTest, Range(0, 5));
+
+#endif // GTEST_HAS_PARAM_TEST
+
+TEST(CompileTest, CombineIsDefinedOnlyWhenGtestHasParamTestIsDefined) {
+#if GTEST_HAS_COMBINE && !GTEST_HAS_PARAM_TEST
+ FAIL() << "GTEST_HAS_COMBINE is defined while GTEST_HAS_PARAM_TEST is not\n"
+#endif
+}
+
+int main(int argc, char **argv) {
+#if GTEST_HAS_PARAM_TEST
+ // Used in TestGenerationTest test case.
+ AddGlobalTestEnvironment(TestGenerationTest::Environment::Instance());
+ // Used in GeneratorEvaluationTest test case. Tests that the updated value
+ // will be picked up for instantiating tests in GeneratorEvaluationTest.
+ GeneratorEvaluationTest::set_param_value(1);
+#endif // GTEST_HAS_PARAM_TEST
+
+ ::testing::InitGoogleTest(&argc, argv);
+
+#if GTEST_HAS_PARAM_TEST
+ // Used in GeneratorEvaluationTest test case. Tests that value updated
+ // here will NOT be used for instantiating tests in
+ // GeneratorEvaluationTest.
+ GeneratorEvaluationTest::set_param_value(2);
+#endif // GTEST_HAS_PARAM_TEST
+
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest-param-test_test.h b/libvpx/third_party/googletest/src/test/gtest-param-test_test.h
new file mode 100644
index 0000000..d0f6556
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-param-test_test.h
@@ -0,0 +1,55 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: vladl@google.com (Vlad Losev)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file provides classes and functions used internally
+// for testing Google Test itself.
+
+#ifndef GTEST_TEST_GTEST_PARAM_TEST_TEST_H_
+#define GTEST_TEST_GTEST_PARAM_TEST_TEST_H_
+
+#include "gtest/gtest.h"
+
+#if GTEST_HAS_PARAM_TEST
+
+// Test fixture for testing definition and instantiation of a test
+// in separate translation units.
+class ExternalInstantiationTest : public ::testing::TestWithParam<int> {};
+
+// Test fixture for testing instantiation of a test in multiple
+// translation units.
+class InstantiationInMultipleTranslaionUnitsTest
+ : public ::testing::TestWithParam<int> {};
+
+#endif // GTEST_HAS_PARAM_TEST
+
+#endif // GTEST_TEST_GTEST_PARAM_TEST_TEST_H_
diff --git a/libvpx/third_party/googletest/src/test/gtest-port_test.cc b/libvpx/third_party/googletest/src/test/gtest-port_test.cc
new file mode 100644
index 0000000..1c6e2b0
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-port_test.cc
@@ -0,0 +1,1206 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: vladl@google.com (Vlad Losev), wan@google.com (Zhanyong Wan)
+//
+// This file tests the internal cross-platform support utilities.
+
+#include "gtest/internal/gtest-port.h"
+
+#include <stdio.h>
+
+#if GTEST_OS_MAC
+# include <time.h>
+#endif // GTEST_OS_MAC
+
+#include <list>
+#include <utility> // For std::pair and std::make_pair.
+#include <vector>
+
+#include "gtest/gtest.h"
+#include "gtest/gtest-spi.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+using std::make_pair;
+using std::pair;
+
+namespace testing {
+namespace internal {
+
+class Base {
+ public:
+ // Copy constructor and assignment operator do exactly what we need, so we
+ // use them.
+ Base() : member_(0) {}
+ explicit Base(int n) : member_(n) {}
+ virtual ~Base() {}
+ int member() { return member_; }
+
+ private:
+ int member_;
+};
+
+class Derived : public Base {
+ public:
+ explicit Derived(int n) : Base(n) {}
+};
+
+TEST(ImplicitCastTest, ConvertsPointers) {
+ Derived derived(0);
+ EXPECT_TRUE(&derived == ::testing::internal::ImplicitCast_<Base*>(&derived));
+}
+
+TEST(ImplicitCastTest, CanUseInheritance) {
+ Derived derived(1);
+ Base base = ::testing::internal::ImplicitCast_<Base>(derived);
+ EXPECT_EQ(derived.member(), base.member());
+}
+
+class Castable {
+ public:
+ Castable(bool* converted) : converted_(converted) {}
+ operator Base() {
+ *converted_ = true;
+ return Base();
+ }
+
+ private:
+ bool* converted_;
+};
+
+TEST(ImplicitCastTest, CanUseNonConstCastOperator) {
+ bool converted = false;
+ Castable castable(&converted);
+ Base base = ::testing::internal::ImplicitCast_<Base>(castable);
+ EXPECT_TRUE(converted);
+}
+
+class ConstCastable {
+ public:
+ ConstCastable(bool* converted) : converted_(converted) {}
+ operator Base() const {
+ *converted_ = true;
+ return Base();
+ }
+
+ private:
+ bool* converted_;
+};
+
+TEST(ImplicitCastTest, CanUseConstCastOperatorOnConstValues) {
+ bool converted = false;
+ const ConstCastable const_castable(&converted);
+ Base base = ::testing::internal::ImplicitCast_<Base>(const_castable);
+ EXPECT_TRUE(converted);
+}
+
+class ConstAndNonConstCastable {
+ public:
+ ConstAndNonConstCastable(bool* converted, bool* const_converted)
+ : converted_(converted), const_converted_(const_converted) {}
+ operator Base() {
+ *converted_ = true;
+ return Base();
+ }
+ operator Base() const {
+ *const_converted_ = true;
+ return Base();
+ }
+
+ private:
+ bool* converted_;
+ bool* const_converted_;
+};
+
+TEST(ImplicitCastTest, CanSelectBetweenConstAndNonConstCasrAppropriately) {
+ bool converted = false;
+ bool const_converted = false;
+ ConstAndNonConstCastable castable(&converted, &const_converted);
+ Base base = ::testing::internal::ImplicitCast_<Base>(castable);
+ EXPECT_TRUE(converted);
+ EXPECT_FALSE(const_converted);
+
+ converted = false;
+ const_converted = false;
+ const ConstAndNonConstCastable const_castable(&converted, &const_converted);
+ base = ::testing::internal::ImplicitCast_<Base>(const_castable);
+ EXPECT_FALSE(converted);
+ EXPECT_TRUE(const_converted);
+}
+
+class To {
+ public:
+ To(bool* converted) { *converted = true; } // NOLINT
+};
+
+TEST(ImplicitCastTest, CanUseImplicitConstructor) {
+ bool converted = false;
+ To to = ::testing::internal::ImplicitCast_<To>(&converted);
+ (void)to;
+ EXPECT_TRUE(converted);
+}
+
+TEST(IteratorTraitsTest, WorksForSTLContainerIterators) {
+ StaticAssertTypeEq<int,
+ IteratorTraits< ::std::vector<int>::const_iterator>::value_type>();
+ StaticAssertTypeEq<bool,
+ IteratorTraits< ::std::list<bool>::iterator>::value_type>();
+}
+
+TEST(IteratorTraitsTest, WorksForPointerToNonConst) {
+ StaticAssertTypeEq<char, IteratorTraits<char*>::value_type>();
+ StaticAssertTypeEq<const void*, IteratorTraits<const void**>::value_type>();
+}
+
+TEST(IteratorTraitsTest, WorksForPointerToConst) {
+ StaticAssertTypeEq<char, IteratorTraits<const char*>::value_type>();
+ StaticAssertTypeEq<const void*,
+ IteratorTraits<const void* const*>::value_type>();
+}
+
+// Tests that the element_type typedef is available in scoped_ptr and refers
+// to the parameter type.
+TEST(ScopedPtrTest, DefinesElementType) {
+ StaticAssertTypeEq<int, ::testing::internal::scoped_ptr<int>::element_type>();
+}
+
+// TODO(vladl@google.com): Implement THE REST of scoped_ptr tests.
+
+TEST(GtestCheckSyntaxTest, BehavesLikeASingleStatement) {
+ if (AlwaysFalse())
+ GTEST_CHECK_(false) << "This should never be executed; "
+ "It's a compilation test only.";
+
+ if (AlwaysTrue())
+ GTEST_CHECK_(true);
+ else
+ ; // NOLINT
+
+ if (AlwaysFalse())
+ ; // NOLINT
+ else
+ GTEST_CHECK_(true) << "";
+}
+
+TEST(GtestCheckSyntaxTest, WorksWithSwitch) {
+ switch (0) {
+ case 1:
+ break;
+ default:
+ GTEST_CHECK_(true);
+ }
+
+ switch(0)
+ case 0:
+ GTEST_CHECK_(true) << "Check failed in switch case";
+}
+
+// Verifies behavior of FormatFileLocation.
+TEST(FormatFileLocationTest, FormatsFileLocation) {
+ EXPECT_PRED_FORMAT2(IsSubstring, "foo.cc", FormatFileLocation("foo.cc", 42));
+ EXPECT_PRED_FORMAT2(IsSubstring, "42", FormatFileLocation("foo.cc", 42));
+}
+
+TEST(FormatFileLocationTest, FormatsUnknownFile) {
+ EXPECT_PRED_FORMAT2(
+ IsSubstring, "unknown file", FormatFileLocation(NULL, 42));
+ EXPECT_PRED_FORMAT2(IsSubstring, "42", FormatFileLocation(NULL, 42));
+}
+
+TEST(FormatFileLocationTest, FormatsUknownLine) {
+ EXPECT_EQ("foo.cc:", FormatFileLocation("foo.cc", -1));
+}
+
+TEST(FormatFileLocationTest, FormatsUknownFileAndLine) {
+ EXPECT_EQ("unknown file:", FormatFileLocation(NULL, -1));
+}
+
+// Verifies behavior of FormatCompilerIndependentFileLocation.
+TEST(FormatCompilerIndependentFileLocationTest, FormatsFileLocation) {
+ EXPECT_EQ("foo.cc:42", FormatCompilerIndependentFileLocation("foo.cc", 42));
+}
+
+TEST(FormatCompilerIndependentFileLocationTest, FormatsUknownFile) {
+ EXPECT_EQ("unknown file:42",
+ FormatCompilerIndependentFileLocation(NULL, 42));
+}
+
+TEST(FormatCompilerIndependentFileLocationTest, FormatsUknownLine) {
+ EXPECT_EQ("foo.cc", FormatCompilerIndependentFileLocation("foo.cc", -1));
+}
+
+TEST(FormatCompilerIndependentFileLocationTest, FormatsUknownFileAndLine) {
+ EXPECT_EQ("unknown file", FormatCompilerIndependentFileLocation(NULL, -1));
+}
+
+#if GTEST_OS_MAC
+void* ThreadFunc(void* data) {
+ pthread_mutex_t* mutex = static_cast<pthread_mutex_t*>(data);
+ pthread_mutex_lock(mutex);
+ pthread_mutex_unlock(mutex);
+ return NULL;
+}
+
+TEST(GetThreadCountTest, ReturnsCorrectValue) {
+ EXPECT_EQ(1U, GetThreadCount());
+ pthread_mutex_t mutex;
+ pthread_attr_t attr;
+ pthread_t thread_id;
+
+ // TODO(vladl@google.com): turn mutex into internal::Mutex for automatic
+ // destruction.
+ pthread_mutex_init(&mutex, NULL);
+ pthread_mutex_lock(&mutex);
+ ASSERT_EQ(0, pthread_attr_init(&attr));
+ ASSERT_EQ(0, pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_JOINABLE));
+
+ const int status = pthread_create(&thread_id, &attr, &ThreadFunc, &mutex);
+ ASSERT_EQ(0, pthread_attr_destroy(&attr));
+ ASSERT_EQ(0, status);
+ EXPECT_EQ(2U, GetThreadCount());
+ pthread_mutex_unlock(&mutex);
+
+ void* dummy;
+ ASSERT_EQ(0, pthread_join(thread_id, &dummy));
+
+ // MacOS X may not immediately report the updated thread count after
+ // joining a thread, causing flakiness in this test. To counter that, we
+ // wait for up to .5 seconds for the OS to report the correct value.
+ for (int i = 0; i < 5; ++i) {
+ if (GetThreadCount() == 1)
+ break;
+
+ SleepMilliseconds(100);
+ }
+ EXPECT_EQ(1U, GetThreadCount());
+ pthread_mutex_destroy(&mutex);
+}
+#else
+TEST(GetThreadCountTest, ReturnsZeroWhenUnableToCountThreads) {
+ EXPECT_EQ(0U, GetThreadCount());
+}
+#endif // GTEST_OS_MAC
+
+TEST(GtestCheckDeathTest, DiesWithCorrectOutputOnFailure) {
+ const bool a_false_condition = false;
+ const char regex[] =
+#ifdef _MSC_VER
+ "gtest-port_test\\.cc\\(\\d+\\):"
+#elif GTEST_USES_POSIX_RE
+ "gtest-port_test\\.cc:[0-9]+"
+#else
+ "gtest-port_test\\.cc:\\d+"
+#endif // _MSC_VER
+ ".*a_false_condition.*Extra info.*";
+
+ EXPECT_DEATH_IF_SUPPORTED(GTEST_CHECK_(a_false_condition) << "Extra info",
+ regex);
+}
+
+#if GTEST_HAS_DEATH_TEST
+
+TEST(GtestCheckDeathTest, LivesSilentlyOnSuccess) {
+ EXPECT_EXIT({
+ GTEST_CHECK_(true) << "Extra info";
+ ::std::cerr << "Success\n";
+ exit(0); },
+ ::testing::ExitedWithCode(0), "Success");
+}
+
+#endif // GTEST_HAS_DEATH_TEST
+
+// Verifies that Google Test choose regular expression engine appropriate to
+// the platform. The test will produce compiler errors in case of failure.
+// For simplicity, we only cover the most important platforms here.
+TEST(RegexEngineSelectionTest, SelectsCorrectRegexEngine) {
+#if GTEST_HAS_POSIX_RE
+
+ EXPECT_TRUE(GTEST_USES_POSIX_RE);
+
+#else
+
+ EXPECT_TRUE(GTEST_USES_SIMPLE_RE);
+
+#endif
+}
+
+#if GTEST_USES_POSIX_RE
+
+# if GTEST_HAS_TYPED_TEST
+
+template <typename Str>
+class RETest : public ::testing::Test {};
+
+// Defines StringTypes as the list of all string types that class RE
+// supports.
+typedef testing::Types<
+ ::std::string,
+# if GTEST_HAS_GLOBAL_STRING
+ ::string,
+# endif // GTEST_HAS_GLOBAL_STRING
+ const char*> StringTypes;
+
+TYPED_TEST_CASE(RETest, StringTypes);
+
+// Tests RE's implicit constructors.
+TYPED_TEST(RETest, ImplicitConstructorWorks) {
+ const RE empty(TypeParam(""));
+ EXPECT_STREQ("", empty.pattern());
+
+ const RE simple(TypeParam("hello"));
+ EXPECT_STREQ("hello", simple.pattern());
+
+ const RE normal(TypeParam(".*(\\w+)"));
+ EXPECT_STREQ(".*(\\w+)", normal.pattern());
+}
+
+// Tests that RE's constructors reject invalid regular expressions.
+TYPED_TEST(RETest, RejectsInvalidRegex) {
+ EXPECT_NONFATAL_FAILURE({
+ const RE invalid(TypeParam("?"));
+ }, "\"?\" is not a valid POSIX Extended regular expression.");
+}
+
+// Tests RE::FullMatch().
+TYPED_TEST(RETest, FullMatchWorks) {
+ const RE empty(TypeParam(""));
+ EXPECT_TRUE(RE::FullMatch(TypeParam(""), empty));
+ EXPECT_FALSE(RE::FullMatch(TypeParam("a"), empty));
+
+ const RE re(TypeParam("a.*z"));
+ EXPECT_TRUE(RE::FullMatch(TypeParam("az"), re));
+ EXPECT_TRUE(RE::FullMatch(TypeParam("axyz"), re));
+ EXPECT_FALSE(RE::FullMatch(TypeParam("baz"), re));
+ EXPECT_FALSE(RE::FullMatch(TypeParam("azy"), re));
+}
+
+// Tests RE::PartialMatch().
+TYPED_TEST(RETest, PartialMatchWorks) {
+ const RE empty(TypeParam(""));
+ EXPECT_TRUE(RE::PartialMatch(TypeParam(""), empty));
+ EXPECT_TRUE(RE::PartialMatch(TypeParam("a"), empty));
+
+ const RE re(TypeParam("a.*z"));
+ EXPECT_TRUE(RE::PartialMatch(TypeParam("az"), re));
+ EXPECT_TRUE(RE::PartialMatch(TypeParam("axyz"), re));
+ EXPECT_TRUE(RE::PartialMatch(TypeParam("baz"), re));
+ EXPECT_TRUE(RE::PartialMatch(TypeParam("azy"), re));
+ EXPECT_FALSE(RE::PartialMatch(TypeParam("zza"), re));
+}
+
+# endif // GTEST_HAS_TYPED_TEST
+
+#elif GTEST_USES_SIMPLE_RE
+
+TEST(IsInSetTest, NulCharIsNotInAnySet) {
+ EXPECT_FALSE(IsInSet('\0', ""));
+ EXPECT_FALSE(IsInSet('\0', "\0"));
+ EXPECT_FALSE(IsInSet('\0', "a"));
+}
+
+TEST(IsInSetTest, WorksForNonNulChars) {
+ EXPECT_FALSE(IsInSet('a', "Ab"));
+ EXPECT_FALSE(IsInSet('c', ""));
+
+ EXPECT_TRUE(IsInSet('b', "bcd"));
+ EXPECT_TRUE(IsInSet('b', "ab"));
+}
+
+TEST(IsAsciiDigitTest, IsFalseForNonDigit) {
+ EXPECT_FALSE(IsAsciiDigit('\0'));
+ EXPECT_FALSE(IsAsciiDigit(' '));
+ EXPECT_FALSE(IsAsciiDigit('+'));
+ EXPECT_FALSE(IsAsciiDigit('-'));
+ EXPECT_FALSE(IsAsciiDigit('.'));
+ EXPECT_FALSE(IsAsciiDigit('a'));
+}
+
+TEST(IsAsciiDigitTest, IsTrueForDigit) {
+ EXPECT_TRUE(IsAsciiDigit('0'));
+ EXPECT_TRUE(IsAsciiDigit('1'));
+ EXPECT_TRUE(IsAsciiDigit('5'));
+ EXPECT_TRUE(IsAsciiDigit('9'));
+}
+
+TEST(IsAsciiPunctTest, IsFalseForNonPunct) {
+ EXPECT_FALSE(IsAsciiPunct('\0'));
+ EXPECT_FALSE(IsAsciiPunct(' '));
+ EXPECT_FALSE(IsAsciiPunct('\n'));
+ EXPECT_FALSE(IsAsciiPunct('a'));
+ EXPECT_FALSE(IsAsciiPunct('0'));
+}
+
+TEST(IsAsciiPunctTest, IsTrueForPunct) {
+ for (const char* p = "^-!\"#$%&'()*+,./:;<=>?@[\\]_`{|}~"; *p; p++) {
+ EXPECT_PRED1(IsAsciiPunct, *p);
+ }
+}
+
+TEST(IsRepeatTest, IsFalseForNonRepeatChar) {
+ EXPECT_FALSE(IsRepeat('\0'));
+ EXPECT_FALSE(IsRepeat(' '));
+ EXPECT_FALSE(IsRepeat('a'));
+ EXPECT_FALSE(IsRepeat('1'));
+ EXPECT_FALSE(IsRepeat('-'));
+}
+
+TEST(IsRepeatTest, IsTrueForRepeatChar) {
+ EXPECT_TRUE(IsRepeat('?'));
+ EXPECT_TRUE(IsRepeat('*'));
+ EXPECT_TRUE(IsRepeat('+'));
+}
+
+TEST(IsAsciiWhiteSpaceTest, IsFalseForNonWhiteSpace) {
+ EXPECT_FALSE(IsAsciiWhiteSpace('\0'));
+ EXPECT_FALSE(IsAsciiWhiteSpace('a'));
+ EXPECT_FALSE(IsAsciiWhiteSpace('1'));
+ EXPECT_FALSE(IsAsciiWhiteSpace('+'));
+ EXPECT_FALSE(IsAsciiWhiteSpace('_'));
+}
+
+TEST(IsAsciiWhiteSpaceTest, IsTrueForWhiteSpace) {
+ EXPECT_TRUE(IsAsciiWhiteSpace(' '));
+ EXPECT_TRUE(IsAsciiWhiteSpace('\n'));
+ EXPECT_TRUE(IsAsciiWhiteSpace('\r'));
+ EXPECT_TRUE(IsAsciiWhiteSpace('\t'));
+ EXPECT_TRUE(IsAsciiWhiteSpace('\v'));
+ EXPECT_TRUE(IsAsciiWhiteSpace('\f'));
+}
+
+TEST(IsAsciiWordCharTest, IsFalseForNonWordChar) {
+ EXPECT_FALSE(IsAsciiWordChar('\0'));
+ EXPECT_FALSE(IsAsciiWordChar('+'));
+ EXPECT_FALSE(IsAsciiWordChar('.'));
+ EXPECT_FALSE(IsAsciiWordChar(' '));
+ EXPECT_FALSE(IsAsciiWordChar('\n'));
+}
+
+TEST(IsAsciiWordCharTest, IsTrueForLetter) {
+ EXPECT_TRUE(IsAsciiWordChar('a'));
+ EXPECT_TRUE(IsAsciiWordChar('b'));
+ EXPECT_TRUE(IsAsciiWordChar('A'));
+ EXPECT_TRUE(IsAsciiWordChar('Z'));
+}
+
+TEST(IsAsciiWordCharTest, IsTrueForDigit) {
+ EXPECT_TRUE(IsAsciiWordChar('0'));
+ EXPECT_TRUE(IsAsciiWordChar('1'));
+ EXPECT_TRUE(IsAsciiWordChar('7'));
+ EXPECT_TRUE(IsAsciiWordChar('9'));
+}
+
+TEST(IsAsciiWordCharTest, IsTrueForUnderscore) {
+ EXPECT_TRUE(IsAsciiWordChar('_'));
+}
+
+TEST(IsValidEscapeTest, IsFalseForNonPrintable) {
+ EXPECT_FALSE(IsValidEscape('\0'));
+ EXPECT_FALSE(IsValidEscape('\007'));
+}
+
+TEST(IsValidEscapeTest, IsFalseForDigit) {
+ EXPECT_FALSE(IsValidEscape('0'));
+ EXPECT_FALSE(IsValidEscape('9'));
+}
+
+TEST(IsValidEscapeTest, IsFalseForWhiteSpace) {
+ EXPECT_FALSE(IsValidEscape(' '));
+ EXPECT_FALSE(IsValidEscape('\n'));
+}
+
+TEST(IsValidEscapeTest, IsFalseForSomeLetter) {
+ EXPECT_FALSE(IsValidEscape('a'));
+ EXPECT_FALSE(IsValidEscape('Z'));
+}
+
+TEST(IsValidEscapeTest, IsTrueForPunct) {
+ EXPECT_TRUE(IsValidEscape('.'));
+ EXPECT_TRUE(IsValidEscape('-'));
+ EXPECT_TRUE(IsValidEscape('^'));
+ EXPECT_TRUE(IsValidEscape('$'));
+ EXPECT_TRUE(IsValidEscape('('));
+ EXPECT_TRUE(IsValidEscape(']'));
+ EXPECT_TRUE(IsValidEscape('{'));
+ EXPECT_TRUE(IsValidEscape('|'));
+}
+
+TEST(IsValidEscapeTest, IsTrueForSomeLetter) {
+ EXPECT_TRUE(IsValidEscape('d'));
+ EXPECT_TRUE(IsValidEscape('D'));
+ EXPECT_TRUE(IsValidEscape('s'));
+ EXPECT_TRUE(IsValidEscape('S'));
+ EXPECT_TRUE(IsValidEscape('w'));
+ EXPECT_TRUE(IsValidEscape('W'));
+}
+
+TEST(AtomMatchesCharTest, EscapedPunct) {
+ EXPECT_FALSE(AtomMatchesChar(true, '\\', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, '\\', ' '));
+ EXPECT_FALSE(AtomMatchesChar(true, '_', '.'));
+ EXPECT_FALSE(AtomMatchesChar(true, '.', 'a'));
+
+ EXPECT_TRUE(AtomMatchesChar(true, '\\', '\\'));
+ EXPECT_TRUE(AtomMatchesChar(true, '_', '_'));
+ EXPECT_TRUE(AtomMatchesChar(true, '+', '+'));
+ EXPECT_TRUE(AtomMatchesChar(true, '.', '.'));
+}
+
+TEST(AtomMatchesCharTest, Escaped_d) {
+ EXPECT_FALSE(AtomMatchesChar(true, 'd', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'd', 'a'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'd', '.'));
+
+ EXPECT_TRUE(AtomMatchesChar(true, 'd', '0'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'd', '9'));
+}
+
+TEST(AtomMatchesCharTest, Escaped_D) {
+ EXPECT_FALSE(AtomMatchesChar(true, 'D', '0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'D', '9'));
+
+ EXPECT_TRUE(AtomMatchesChar(true, 'D', '\0'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'D', 'a'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'D', '-'));
+}
+
+TEST(AtomMatchesCharTest, Escaped_s) {
+ EXPECT_FALSE(AtomMatchesChar(true, 's', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 's', 'a'));
+ EXPECT_FALSE(AtomMatchesChar(true, 's', '.'));
+ EXPECT_FALSE(AtomMatchesChar(true, 's', '9'));
+
+ EXPECT_TRUE(AtomMatchesChar(true, 's', ' '));
+ EXPECT_TRUE(AtomMatchesChar(true, 's', '\n'));
+ EXPECT_TRUE(AtomMatchesChar(true, 's', '\t'));
+}
+
+TEST(AtomMatchesCharTest, Escaped_S) {
+ EXPECT_FALSE(AtomMatchesChar(true, 'S', ' '));
+ EXPECT_FALSE(AtomMatchesChar(true, 'S', '\r'));
+
+ EXPECT_TRUE(AtomMatchesChar(true, 'S', '\0'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'S', 'a'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'S', '9'));
+}
+
+TEST(AtomMatchesCharTest, Escaped_w) {
+ EXPECT_FALSE(AtomMatchesChar(true, 'w', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'w', '+'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'w', ' '));
+ EXPECT_FALSE(AtomMatchesChar(true, 'w', '\n'));
+
+ EXPECT_TRUE(AtomMatchesChar(true, 'w', '0'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'w', 'b'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'w', 'C'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'w', '_'));
+}
+
+TEST(AtomMatchesCharTest, Escaped_W) {
+ EXPECT_FALSE(AtomMatchesChar(true, 'W', 'A'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'W', 'b'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'W', '9'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'W', '_'));
+
+ EXPECT_TRUE(AtomMatchesChar(true, 'W', '\0'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'W', '*'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'W', '\n'));
+}
+
+TEST(AtomMatchesCharTest, EscapedWhiteSpace) {
+ EXPECT_FALSE(AtomMatchesChar(true, 'f', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'f', '\n'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'n', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'n', '\r'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'r', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'r', 'a'));
+ EXPECT_FALSE(AtomMatchesChar(true, 't', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 't', 't'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'v', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(true, 'v', '\f'));
+
+ EXPECT_TRUE(AtomMatchesChar(true, 'f', '\f'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'n', '\n'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'r', '\r'));
+ EXPECT_TRUE(AtomMatchesChar(true, 't', '\t'));
+ EXPECT_TRUE(AtomMatchesChar(true, 'v', '\v'));
+}
+
+TEST(AtomMatchesCharTest, UnescapedDot) {
+ EXPECT_FALSE(AtomMatchesChar(false, '.', '\n'));
+
+ EXPECT_TRUE(AtomMatchesChar(false, '.', '\0'));
+ EXPECT_TRUE(AtomMatchesChar(false, '.', '.'));
+ EXPECT_TRUE(AtomMatchesChar(false, '.', 'a'));
+ EXPECT_TRUE(AtomMatchesChar(false, '.', ' '));
+}
+
+TEST(AtomMatchesCharTest, UnescapedChar) {
+ EXPECT_FALSE(AtomMatchesChar(false, 'a', '\0'));
+ EXPECT_FALSE(AtomMatchesChar(false, 'a', 'b'));
+ EXPECT_FALSE(AtomMatchesChar(false, '$', 'a'));
+
+ EXPECT_TRUE(AtomMatchesChar(false, '$', '$'));
+ EXPECT_TRUE(AtomMatchesChar(false, '5', '5'));
+ EXPECT_TRUE(AtomMatchesChar(false, 'Z', 'Z'));
+}
+
+TEST(ValidateRegexTest, GeneratesFailureAndReturnsFalseForInvalid) {
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex(NULL)),
+ "NULL is not a valid simple regular expression");
+ EXPECT_NONFATAL_FAILURE(
+ ASSERT_FALSE(ValidateRegex("a\\")),
+ "Syntax error at index 1 in simple regular expression \"a\\\": ");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("a\\")),
+ "'\\' cannot appear at the end");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("\\n\\")),
+ "'\\' cannot appear at the end");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("\\s\\hb")),
+ "invalid escape sequence \"\\h\"");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("^^")),
+ "'^' can only appear at the beginning");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex(".*^b")),
+ "'^' can only appear at the beginning");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("$$")),
+ "'$' can only appear at the end");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("^$a")),
+ "'$' can only appear at the end");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("a(b")),
+ "'(' is unsupported");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("ab)")),
+ "')' is unsupported");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("[ab")),
+ "'[' is unsupported");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("a{2")),
+ "'{' is unsupported");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("?")),
+ "'?' can only follow a repeatable token");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("^*")),
+ "'*' can only follow a repeatable token");
+ EXPECT_NONFATAL_FAILURE(ASSERT_FALSE(ValidateRegex("5*+")),
+ "'+' can only follow a repeatable token");
+}
+
+TEST(ValidateRegexTest, ReturnsTrueForValid) {
+ EXPECT_TRUE(ValidateRegex(""));
+ EXPECT_TRUE(ValidateRegex("a"));
+ EXPECT_TRUE(ValidateRegex(".*"));
+ EXPECT_TRUE(ValidateRegex("^a_+"));
+ EXPECT_TRUE(ValidateRegex("^a\\t\\&?"));
+ EXPECT_TRUE(ValidateRegex("09*$"));
+ EXPECT_TRUE(ValidateRegex("^Z$"));
+ EXPECT_TRUE(ValidateRegex("a\\^Z\\$\\(\\)\\|\\[\\]\\{\\}"));
+}
+
+TEST(MatchRepetitionAndRegexAtHeadTest, WorksForZeroOrOne) {
+ EXPECT_FALSE(MatchRepetitionAndRegexAtHead(false, 'a', '?', "a", "ba"));
+ // Repeating more than once.
+ EXPECT_FALSE(MatchRepetitionAndRegexAtHead(false, 'a', '?', "b", "aab"));
+
+ // Repeating zero times.
+ EXPECT_TRUE(MatchRepetitionAndRegexAtHead(false, 'a', '?', "b", "ba"));
+ // Repeating once.
+ EXPECT_TRUE(MatchRepetitionAndRegexAtHead(false, 'a', '?', "b", "ab"));
+ EXPECT_TRUE(MatchRepetitionAndRegexAtHead(false, '#', '?', ".", "##"));
+}
+
+TEST(MatchRepetitionAndRegexAtHeadTest, WorksForZeroOrMany) {
+ EXPECT_FALSE(MatchRepetitionAndRegexAtHead(false, '.', '*', "a$", "baab"));
+
+ // Repeating zero times.
+ EXPECT_TRUE(MatchRepetitionAndRegexAtHead(false, '.', '*', "b", "bc"));
+ // Repeating once.
+ EXPECT_TRUE(MatchRepetitionAndRegexAtHead(false, '.', '*', "b", "abc"));
+ // Repeating more than once.
+ EXPECT_TRUE(MatchRepetitionAndRegexAtHead(true, 'w', '*', "-", "ab_1-g"));
+}
+
+TEST(MatchRepetitionAndRegexAtHeadTest, WorksForOneOrMany) {
+ EXPECT_FALSE(MatchRepetitionAndRegexAtHead(false, '.', '+', "a$", "baab"));
+ // Repeating zero times.
+ EXPECT_FALSE(MatchRepetitionAndRegexAtHead(false, '.', '+', "b", "bc"));
+
+ // Repeating once.
+ EXPECT_TRUE(MatchRepetitionAndRegexAtHead(false, '.', '+', "b", "abc"));
+ // Repeating more than once.
+ EXPECT_TRUE(MatchRepetitionAndRegexAtHead(true, 'w', '+', "-", "ab_1-g"));
+}
+
+TEST(MatchRegexAtHeadTest, ReturnsTrueForEmptyRegex) {
+ EXPECT_TRUE(MatchRegexAtHead("", ""));
+ EXPECT_TRUE(MatchRegexAtHead("", "ab"));
+}
+
+TEST(MatchRegexAtHeadTest, WorksWhenDollarIsInRegex) {
+ EXPECT_FALSE(MatchRegexAtHead("$", "a"));
+
+ EXPECT_TRUE(MatchRegexAtHead("$", ""));
+ EXPECT_TRUE(MatchRegexAtHead("a$", "a"));
+}
+
+TEST(MatchRegexAtHeadTest, WorksWhenRegexStartsWithEscapeSequence) {
+ EXPECT_FALSE(MatchRegexAtHead("\\w", "+"));
+ EXPECT_FALSE(MatchRegexAtHead("\\W", "ab"));
+
+ EXPECT_TRUE(MatchRegexAtHead("\\sa", "\nab"));
+ EXPECT_TRUE(MatchRegexAtHead("\\d", "1a"));
+}
+
+TEST(MatchRegexAtHeadTest, WorksWhenRegexStartsWithRepetition) {
+ EXPECT_FALSE(MatchRegexAtHead(".+a", "abc"));
+ EXPECT_FALSE(MatchRegexAtHead("a?b", "aab"));
+
+ EXPECT_TRUE(MatchRegexAtHead(".*a", "bc12-ab"));
+ EXPECT_TRUE(MatchRegexAtHead("a?b", "b"));
+ EXPECT_TRUE(MatchRegexAtHead("a?b", "ab"));
+}
+
+TEST(MatchRegexAtHeadTest,
+ WorksWhenRegexStartsWithRepetionOfEscapeSequence) {
+ EXPECT_FALSE(MatchRegexAtHead("\\.+a", "abc"));
+ EXPECT_FALSE(MatchRegexAtHead("\\s?b", " b"));
+
+ EXPECT_TRUE(MatchRegexAtHead("\\(*a", "((((ab"));
+ EXPECT_TRUE(MatchRegexAtHead("\\^?b", "^b"));
+ EXPECT_TRUE(MatchRegexAtHead("\\\\?b", "b"));
+ EXPECT_TRUE(MatchRegexAtHead("\\\\?b", "\\b"));
+}
+
+TEST(MatchRegexAtHeadTest, MatchesSequentially) {
+ EXPECT_FALSE(MatchRegexAtHead("ab.*c", "acabc"));
+
+ EXPECT_TRUE(MatchRegexAtHead("ab.*c", "ab-fsc"));
+}
+
+TEST(MatchRegexAnywhereTest, ReturnsFalseWhenStringIsNull) {
+ EXPECT_FALSE(MatchRegexAnywhere("", NULL));
+}
+
+TEST(MatchRegexAnywhereTest, WorksWhenRegexStartsWithCaret) {
+ EXPECT_FALSE(MatchRegexAnywhere("^a", "ba"));
+ EXPECT_FALSE(MatchRegexAnywhere("^$", "a"));
+
+ EXPECT_TRUE(MatchRegexAnywhere("^a", "ab"));
+ EXPECT_TRUE(MatchRegexAnywhere("^", "ab"));
+ EXPECT_TRUE(MatchRegexAnywhere("^$", ""));
+}
+
+TEST(MatchRegexAnywhereTest, ReturnsFalseWhenNoMatch) {
+ EXPECT_FALSE(MatchRegexAnywhere("a", "bcde123"));
+ EXPECT_FALSE(MatchRegexAnywhere("a.+a", "--aa88888888"));
+}
+
+TEST(MatchRegexAnywhereTest, ReturnsTrueWhenMatchingPrefix) {
+ EXPECT_TRUE(MatchRegexAnywhere("\\w+", "ab1_ - 5"));
+ EXPECT_TRUE(MatchRegexAnywhere(".*=", "="));
+ EXPECT_TRUE(MatchRegexAnywhere("x.*ab?.*bc", "xaaabc"));
+}
+
+TEST(MatchRegexAnywhereTest, ReturnsTrueWhenMatchingNonPrefix) {
+ EXPECT_TRUE(MatchRegexAnywhere("\\w+", "$$$ ab1_ - 5"));
+ EXPECT_TRUE(MatchRegexAnywhere("\\.+=", "= ...="));
+}
+
+// Tests RE's implicit constructors.
+TEST(RETest, ImplicitConstructorWorks) {
+ const RE empty("");
+ EXPECT_STREQ("", empty.pattern());
+
+ const RE simple("hello");
+ EXPECT_STREQ("hello", simple.pattern());
+}
+
+// Tests that RE's constructors reject invalid regular expressions.
+TEST(RETest, RejectsInvalidRegex) {
+ EXPECT_NONFATAL_FAILURE({
+ const RE normal(NULL);
+ }, "NULL is not a valid simple regular expression");
+
+ EXPECT_NONFATAL_FAILURE({
+ const RE normal(".*(\\w+");
+ }, "'(' is unsupported");
+
+ EXPECT_NONFATAL_FAILURE({
+ const RE invalid("^?");
+ }, "'?' can only follow a repeatable token");
+}
+
+// Tests RE::FullMatch().
+TEST(RETest, FullMatchWorks) {
+ const RE empty("");
+ EXPECT_TRUE(RE::FullMatch("", empty));
+ EXPECT_FALSE(RE::FullMatch("a", empty));
+
+ const RE re1("a");
+ EXPECT_TRUE(RE::FullMatch("a", re1));
+
+ const RE re("a.*z");
+ EXPECT_TRUE(RE::FullMatch("az", re));
+ EXPECT_TRUE(RE::FullMatch("axyz", re));
+ EXPECT_FALSE(RE::FullMatch("baz", re));
+ EXPECT_FALSE(RE::FullMatch("azy", re));
+}
+
+// Tests RE::PartialMatch().
+TEST(RETest, PartialMatchWorks) {
+ const RE empty("");
+ EXPECT_TRUE(RE::PartialMatch("", empty));
+ EXPECT_TRUE(RE::PartialMatch("a", empty));
+
+ const RE re("a.*z");
+ EXPECT_TRUE(RE::PartialMatch("az", re));
+ EXPECT_TRUE(RE::PartialMatch("axyz", re));
+ EXPECT_TRUE(RE::PartialMatch("baz", re));
+ EXPECT_TRUE(RE::PartialMatch("azy", re));
+ EXPECT_FALSE(RE::PartialMatch("zza", re));
+}
+
+#endif // GTEST_USES_POSIX_RE
+
+#if !GTEST_OS_WINDOWS_MOBILE
+
+TEST(CaptureTest, CapturesStdout) {
+ CaptureStdout();
+ fprintf(stdout, "abc");
+ EXPECT_STREQ("abc", GetCapturedStdout().c_str());
+
+ CaptureStdout();
+ fprintf(stdout, "def%cghi", '\0');
+ EXPECT_EQ(::std::string("def\0ghi", 7), ::std::string(GetCapturedStdout()));
+}
+
+TEST(CaptureTest, CapturesStderr) {
+ CaptureStderr();
+ fprintf(stderr, "jkl");
+ EXPECT_STREQ("jkl", GetCapturedStderr().c_str());
+
+ CaptureStderr();
+ fprintf(stderr, "jkl%cmno", '\0');
+ EXPECT_EQ(::std::string("jkl\0mno", 7), ::std::string(GetCapturedStderr()));
+}
+
+// Tests that stdout and stderr capture don't interfere with each other.
+TEST(CaptureTest, CapturesStdoutAndStderr) {
+ CaptureStdout();
+ CaptureStderr();
+ fprintf(stdout, "pqr");
+ fprintf(stderr, "stu");
+ EXPECT_STREQ("pqr", GetCapturedStdout().c_str());
+ EXPECT_STREQ("stu", GetCapturedStderr().c_str());
+}
+
+TEST(CaptureDeathTest, CannotReenterStdoutCapture) {
+ CaptureStdout();
+ EXPECT_DEATH_IF_SUPPORTED(CaptureStdout();,
+ "Only one stdout capturer can exist at a time");
+ GetCapturedStdout();
+
+ // We cannot test stderr capturing using death tests as they use it
+ // themselves.
+}
+
+#endif // !GTEST_OS_WINDOWS_MOBILE
+
+TEST(ThreadLocalTest, DefaultConstructorInitializesToDefaultValues) {
+ ThreadLocal<int> t1;
+ EXPECT_EQ(0, t1.get());
+
+ ThreadLocal<void*> t2;
+ EXPECT_TRUE(t2.get() == NULL);
+}
+
+TEST(ThreadLocalTest, SingleParamConstructorInitializesToParam) {
+ ThreadLocal<int> t1(123);
+ EXPECT_EQ(123, t1.get());
+
+ int i = 0;
+ ThreadLocal<int*> t2(&i);
+ EXPECT_EQ(&i, t2.get());
+}
+
+class NoDefaultContructor {
+ public:
+ explicit NoDefaultContructor(const char*) {}
+ NoDefaultContructor(const NoDefaultContructor&) {}
+};
+
+TEST(ThreadLocalTest, ValueDefaultContructorIsNotRequiredForParamVersion) {
+ ThreadLocal<NoDefaultContructor> bar(NoDefaultContructor("foo"));
+ bar.pointer();
+}
+
+TEST(ThreadLocalTest, GetAndPointerReturnSameValue) {
+ ThreadLocal<String> thread_local;
+
+ EXPECT_EQ(thread_local.pointer(), &(thread_local.get()));
+
+ // Verifies the condition still holds after calling set.
+ thread_local.set("foo");
+ EXPECT_EQ(thread_local.pointer(), &(thread_local.get()));
+}
+
+TEST(ThreadLocalTest, PointerAndConstPointerReturnSameValue) {
+ ThreadLocal<String> thread_local;
+ const ThreadLocal<String>& const_thread_local = thread_local;
+
+ EXPECT_EQ(thread_local.pointer(), const_thread_local.pointer());
+
+ thread_local.set("foo");
+ EXPECT_EQ(thread_local.pointer(), const_thread_local.pointer());
+}
+
+#if GTEST_IS_THREADSAFE
+
+void AddTwo(int* param) { *param += 2; }
+
+TEST(ThreadWithParamTest, ConstructorExecutesThreadFunc) {
+ int i = 40;
+ ThreadWithParam<int*> thread(&AddTwo, &i, NULL);
+ thread.Join();
+ EXPECT_EQ(42, i);
+}
+
+TEST(MutexDeathTest, AssertHeldShouldAssertWhenNotLocked) {
+ // AssertHeld() is flaky only in the presence of multiple threads accessing
+ // the lock. In this case, the test is robust.
+ EXPECT_DEATH_IF_SUPPORTED({
+ Mutex m;
+ { MutexLock lock(&m); }
+ m.AssertHeld();
+ },
+ "thread .*hold");
+}
+
+TEST(MutexTest, AssertHeldShouldNotAssertWhenLocked) {
+ Mutex m;
+ MutexLock lock(&m);
+ m.AssertHeld();
+}
+
+class AtomicCounterWithMutex {
+ public:
+ explicit AtomicCounterWithMutex(Mutex* mutex) :
+ value_(0), mutex_(mutex), random_(42) {}
+
+ void Increment() {
+ MutexLock lock(mutex_);
+ int temp = value_;
+ {
+ // Locking a mutex puts up a memory barrier, preventing reads and
+ // writes to value_ rearranged when observed from other threads.
+ //
+ // We cannot use Mutex and MutexLock here or rely on their memory
+ // barrier functionality as we are testing them here.
+ pthread_mutex_t memory_barrier_mutex;
+ GTEST_CHECK_POSIX_SUCCESS_(
+ pthread_mutex_init(&memory_barrier_mutex, NULL));
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_lock(&memory_barrier_mutex));
+
+ SleepMilliseconds(random_.Generate(30));
+
+ GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_unlock(&memory_barrier_mutex));
+ }
+ value_ = temp + 1;
+ }
+ int value() const { return value_; }
+
+ private:
+ volatile int value_;
+ Mutex* const mutex_; // Protects value_.
+ Random random_;
+};
+
+void CountingThreadFunc(pair<AtomicCounterWithMutex*, int> param) {
+ for (int i = 0; i < param.second; ++i)
+ param.first->Increment();
+}
+
+// Tests that the mutex only lets one thread at a time to lock it.
+TEST(MutexTest, OnlyOneThreadCanLockAtATime) {
+ Mutex mutex;
+ AtomicCounterWithMutex locked_counter(&mutex);
+
+ typedef ThreadWithParam<pair<AtomicCounterWithMutex*, int> > ThreadType;
+ const int kCycleCount = 20;
+ const int kThreadCount = 7;
+ scoped_ptr<ThreadType> counting_threads[kThreadCount];
+ Notification threads_can_start;
+ // Creates and runs kThreadCount threads that increment locked_counter
+ // kCycleCount times each.
+ for (int i = 0; i < kThreadCount; ++i) {
+ counting_threads[i].reset(new ThreadType(&CountingThreadFunc,
+ make_pair(&locked_counter,
+ kCycleCount),
+ &threads_can_start));
+ }
+ threads_can_start.Notify();
+ for (int i = 0; i < kThreadCount; ++i)
+ counting_threads[i]->Join();
+
+ // If the mutex lets more than one thread to increment the counter at a
+ // time, they are likely to encounter a race condition and have some
+ // increments overwritten, resulting in the lower then expected counter
+ // value.
+ EXPECT_EQ(kCycleCount * kThreadCount, locked_counter.value());
+}
+
+template <typename T>
+void RunFromThread(void (func)(T), T param) {
+ ThreadWithParam<T> thread(func, param, NULL);
+ thread.Join();
+}
+
+void RetrieveThreadLocalValue(pair<ThreadLocal<String>*, String*> param) {
+ *param.second = param.first->get();
+}
+
+TEST(ThreadLocalTest, ParameterizedConstructorSetsDefault) {
+ ThreadLocal<String> thread_local("foo");
+ EXPECT_STREQ("foo", thread_local.get().c_str());
+
+ thread_local.set("bar");
+ EXPECT_STREQ("bar", thread_local.get().c_str());
+
+ String result;
+ RunFromThread(&RetrieveThreadLocalValue, make_pair(&thread_local, &result));
+ EXPECT_STREQ("foo", result.c_str());
+}
+
+// DestructorTracker keeps track of whether its instances have been
+// destroyed.
+static std::vector<bool> g_destroyed;
+
+class DestructorTracker {
+ public:
+ DestructorTracker() : index_(GetNewIndex()) {}
+ DestructorTracker(const DestructorTracker& /* rhs */)
+ : index_(GetNewIndex()) {}
+ ~DestructorTracker() {
+ // We never access g_destroyed concurrently, so we don't need to
+ // protect the write operation under a mutex.
+ g_destroyed[index_] = true;
+ }
+
+ private:
+ static int GetNewIndex() {
+ g_destroyed.push_back(false);
+ return g_destroyed.size() - 1;
+ }
+ const int index_;
+};
+
+typedef ThreadLocal<DestructorTracker>* ThreadParam;
+
+void CallThreadLocalGet(ThreadParam thread_local) {
+ thread_local->get();
+}
+
+// Tests that when a ThreadLocal object dies in a thread, it destroys
+// the managed object for that thread.
+TEST(ThreadLocalTest, DestroysManagedObjectForOwnThreadWhenDying) {
+ g_destroyed.clear();
+
+ {
+ // The next line default constructs a DestructorTracker object as
+ // the default value of objects managed by thread_local.
+ ThreadLocal<DestructorTracker> thread_local;
+ ASSERT_EQ(1U, g_destroyed.size());
+ ASSERT_FALSE(g_destroyed[0]);
+
+ // This creates another DestructorTracker object for the main thread.
+ thread_local.get();
+ ASSERT_EQ(2U, g_destroyed.size());
+ ASSERT_FALSE(g_destroyed[0]);
+ ASSERT_FALSE(g_destroyed[1]);
+ }
+
+ // Now thread_local has died. It should have destroyed both the
+ // default value shared by all threads and the value for the main
+ // thread.
+ ASSERT_EQ(2U, g_destroyed.size());
+ EXPECT_TRUE(g_destroyed[0]);
+ EXPECT_TRUE(g_destroyed[1]);
+
+ g_destroyed.clear();
+}
+
+// Tests that when a thread exits, the thread-local object for that
+// thread is destroyed.
+TEST(ThreadLocalTest, DestroysManagedObjectAtThreadExit) {
+ g_destroyed.clear();
+
+ {
+ // The next line default constructs a DestructorTracker object as
+ // the default value of objects managed by thread_local.
+ ThreadLocal<DestructorTracker> thread_local;
+ ASSERT_EQ(1U, g_destroyed.size());
+ ASSERT_FALSE(g_destroyed[0]);
+
+ // This creates another DestructorTracker object in the new thread.
+ ThreadWithParam<ThreadParam> thread(
+ &CallThreadLocalGet, &thread_local, NULL);
+ thread.Join();
+
+ // Now the new thread has exited. The per-thread object for it
+ // should have been destroyed.
+ ASSERT_EQ(2U, g_destroyed.size());
+ ASSERT_FALSE(g_destroyed[0]);
+ ASSERT_TRUE(g_destroyed[1]);
+ }
+
+ // Now thread_local has died. The default value should have been
+ // destroyed too.
+ ASSERT_EQ(2U, g_destroyed.size());
+ EXPECT_TRUE(g_destroyed[0]);
+ EXPECT_TRUE(g_destroyed[1]);
+
+ g_destroyed.clear();
+}
+
+TEST(ThreadLocalTest, ThreadLocalMutationsAffectOnlyCurrentThread) {
+ ThreadLocal<String> thread_local;
+ thread_local.set("Foo");
+ EXPECT_STREQ("Foo", thread_local.get().c_str());
+
+ String result;
+ RunFromThread(&RetrieveThreadLocalValue, make_pair(&thread_local, &result));
+ EXPECT_TRUE(result.c_str() == NULL);
+}
+
+#endif // GTEST_IS_THREADSAFE
+
+} // namespace internal
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/test/gtest-printers_test.cc b/libvpx/third_party/googletest/src/test/gtest-printers_test.cc
new file mode 100644
index 0000000..6292c7f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-printers_test.cc
@@ -0,0 +1,1307 @@
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Google Test - The Google C++ Testing Framework
+//
+// This file tests the universal value printer.
+
+#include "gtest/gtest-printers.h"
+
+#include <ctype.h>
+#include <limits.h>
+#include <string.h>
+#include <algorithm>
+#include <deque>
+#include <list>
+#include <map>
+#include <set>
+#include <sstream>
+#include <string>
+#include <utility>
+#include <vector>
+
+#include "gtest/gtest.h"
+
+// hash_map and hash_set are available under Visual C++.
+#if _MSC_VER
+# define GTEST_HAS_HASH_MAP_ 1 // Indicates that hash_map is available.
+# include <hash_map> // NOLINT
+# define GTEST_HAS_HASH_SET_ 1 // Indicates that hash_set is available.
+# include <hash_set> // NOLINT
+#endif // GTEST_OS_WINDOWS
+
+// Some user-defined types for testing the universal value printer.
+
+// An anonymous enum type.
+enum AnonymousEnum {
+ kAE1 = -1,
+ kAE2 = 1
+};
+
+// An enum without a user-defined printer.
+enum EnumWithoutPrinter {
+ kEWP1 = -2,
+ kEWP2 = 42
+};
+
+// An enum with a << operator.
+enum EnumWithStreaming {
+ kEWS1 = 10
+};
+
+std::ostream& operator<<(std::ostream& os, EnumWithStreaming e) {
+ return os << (e == kEWS1 ? "kEWS1" : "invalid");
+}
+
+// An enum with a PrintTo() function.
+enum EnumWithPrintTo {
+ kEWPT1 = 1
+};
+
+void PrintTo(EnumWithPrintTo e, std::ostream* os) {
+ *os << (e == kEWPT1 ? "kEWPT1" : "invalid");
+}
+
+// A class implicitly convertible to BiggestInt.
+class BiggestIntConvertible {
+ public:
+ operator ::testing::internal::BiggestInt() const { return 42; }
+};
+
+// A user-defined unprintable class template in the global namespace.
+template <typename T>
+class UnprintableTemplateInGlobal {
+ public:
+ UnprintableTemplateInGlobal() : value_() {}
+ private:
+ T value_;
+};
+
+// A user-defined streamable type in the global namespace.
+class StreamableInGlobal {
+ public:
+ virtual ~StreamableInGlobal() {}
+};
+
+inline void operator<<(::std::ostream& os, const StreamableInGlobal& /* x */) {
+ os << "StreamableInGlobal";
+}
+
+void operator<<(::std::ostream& os, const StreamableInGlobal* /* x */) {
+ os << "StreamableInGlobal*";
+}
+
+namespace foo {
+
+// A user-defined unprintable type in a user namespace.
+class UnprintableInFoo {
+ public:
+ UnprintableInFoo() : z_(0) { memcpy(xy_, "\xEF\x12\x0\x0\x34\xAB\x0\x0", 8); }
+ private:
+ char xy_[8];
+ double z_;
+};
+
+// A user-defined printable type in a user-chosen namespace.
+struct PrintableViaPrintTo {
+ PrintableViaPrintTo() : value() {}
+ int value;
+};
+
+void PrintTo(const PrintableViaPrintTo& x, ::std::ostream* os) {
+ *os << "PrintableViaPrintTo: " << x.value;
+}
+
+// A type with a user-defined << for printing its pointer.
+struct PointerPrintable {
+};
+
+::std::ostream& operator<<(::std::ostream& os,
+ const PointerPrintable* /* x */) {
+ return os << "PointerPrintable*";
+}
+
+// A user-defined printable class template in a user-chosen namespace.
+template <typename T>
+class PrintableViaPrintToTemplate {
+ public:
+ explicit PrintableViaPrintToTemplate(const T& a_value) : value_(a_value) {}
+
+ const T& value() const { return value_; }
+ private:
+ T value_;
+};
+
+template <typename T>
+void PrintTo(const PrintableViaPrintToTemplate<T>& x, ::std::ostream* os) {
+ *os << "PrintableViaPrintToTemplate: " << x.value();
+}
+
+// A user-defined streamable class template in a user namespace.
+template <typename T>
+class StreamableTemplateInFoo {
+ public:
+ StreamableTemplateInFoo() : value_() {}
+
+ const T& value() const { return value_; }
+ private:
+ T value_;
+};
+
+template <typename T>
+inline ::std::ostream& operator<<(::std::ostream& os,
+ const StreamableTemplateInFoo<T>& x) {
+ return os << "StreamableTemplateInFoo: " << x.value();
+}
+
+} // namespace foo
+
+namespace testing {
+namespace gtest_printers_test {
+
+using ::std::deque;
+using ::std::list;
+using ::std::make_pair;
+using ::std::map;
+using ::std::multimap;
+using ::std::multiset;
+using ::std::pair;
+using ::std::set;
+using ::std::vector;
+using ::testing::PrintToString;
+using ::testing::internal::NativeArray;
+using ::testing::internal::RE;
+using ::testing::internal::Strings;
+using ::testing::internal::UniversalTersePrint;
+using ::testing::internal::UniversalPrint;
+using ::testing::internal::UniversalTersePrintTupleFieldsToStrings;
+using ::testing::internal::UniversalPrinter;
+using ::testing::internal::kReference;
+using ::testing::internal::string;
+
+#if GTEST_HAS_TR1_TUPLE
+using ::std::tr1::make_tuple;
+using ::std::tr1::tuple;
+#endif
+
+#if _MSC_VER
+// MSVC defines the following classes in the ::stdext namespace while
+// gcc defines them in the :: namespace. Note that they are not part
+// of the C++ standard.
+using ::stdext::hash_map;
+using ::stdext::hash_set;
+using ::stdext::hash_multimap;
+using ::stdext::hash_multiset;
+#endif
+
+// Prints a value to a string using the universal value printer. This
+// is a helper for testing UniversalPrinter<T>::Print() for various types.
+template <typename T>
+string Print(const T& value) {
+ ::std::stringstream ss;
+ UniversalPrinter<T>::Print(value, &ss);
+ return ss.str();
+}
+
+// Prints a value passed by reference to a string, using the universal
+// value printer. This is a helper for testing
+// UniversalPrinter<T&>::Print() for various types.
+template <typename T>
+string PrintByRef(const T& value) {
+ ::std::stringstream ss;
+ UniversalPrinter<T&>::Print(value, &ss);
+ return ss.str();
+}
+
+// Tests printing various enum types.
+
+TEST(PrintEnumTest, AnonymousEnum) {
+ EXPECT_EQ("-1", Print(kAE1));
+ EXPECT_EQ("1", Print(kAE2));
+}
+
+TEST(PrintEnumTest, EnumWithoutPrinter) {
+ EXPECT_EQ("-2", Print(kEWP1));
+ EXPECT_EQ("42", Print(kEWP2));
+}
+
+TEST(PrintEnumTest, EnumWithStreaming) {
+ EXPECT_EQ("kEWS1", Print(kEWS1));
+ EXPECT_EQ("invalid", Print(static_cast<EnumWithStreaming>(0)));
+}
+
+TEST(PrintEnumTest, EnumWithPrintTo) {
+ EXPECT_EQ("kEWPT1", Print(kEWPT1));
+ EXPECT_EQ("invalid", Print(static_cast<EnumWithPrintTo>(0)));
+}
+
+// Tests printing a class implicitly convertible to BiggestInt.
+
+TEST(PrintClassTest, BiggestIntConvertible) {
+ EXPECT_EQ("42", Print(BiggestIntConvertible()));
+}
+
+// Tests printing various char types.
+
+// char.
+TEST(PrintCharTest, PlainChar) {
+ EXPECT_EQ("'\\0'", Print('\0'));
+ EXPECT_EQ("'\\'' (39, 0x27)", Print('\''));
+ EXPECT_EQ("'\"' (34, 0x22)", Print('"'));
+ EXPECT_EQ("'?' (63, 0x3F)", Print('?'));
+ EXPECT_EQ("'\\\\' (92, 0x5C)", Print('\\'));
+ EXPECT_EQ("'\\a' (7)", Print('\a'));
+ EXPECT_EQ("'\\b' (8)", Print('\b'));
+ EXPECT_EQ("'\\f' (12, 0xC)", Print('\f'));
+ EXPECT_EQ("'\\n' (10, 0xA)", Print('\n'));
+ EXPECT_EQ("'\\r' (13, 0xD)", Print('\r'));
+ EXPECT_EQ("'\\t' (9)", Print('\t'));
+ EXPECT_EQ("'\\v' (11, 0xB)", Print('\v'));
+ EXPECT_EQ("'\\x7F' (127)", Print('\x7F'));
+ EXPECT_EQ("'\\xFF' (255)", Print('\xFF'));
+ EXPECT_EQ("' ' (32, 0x20)", Print(' '));
+ EXPECT_EQ("'a' (97, 0x61)", Print('a'));
+}
+
+// signed char.
+TEST(PrintCharTest, SignedChar) {
+ EXPECT_EQ("'\\0'", Print(static_cast<signed char>('\0')));
+ EXPECT_EQ("'\\xCE' (-50)",
+ Print(static_cast<signed char>(-50)));
+}
+
+// unsigned char.
+TEST(PrintCharTest, UnsignedChar) {
+ EXPECT_EQ("'\\0'", Print(static_cast<unsigned char>('\0')));
+ EXPECT_EQ("'b' (98, 0x62)",
+ Print(static_cast<unsigned char>('b')));
+}
+
+// Tests printing other simple, built-in types.
+
+// bool.
+TEST(PrintBuiltInTypeTest, Bool) {
+ EXPECT_EQ("false", Print(false));
+ EXPECT_EQ("true", Print(true));
+}
+
+// wchar_t.
+TEST(PrintBuiltInTypeTest, Wchar_t) {
+ EXPECT_EQ("L'\\0'", Print(L'\0'));
+ EXPECT_EQ("L'\\'' (39, 0x27)", Print(L'\''));
+ EXPECT_EQ("L'\"' (34, 0x22)", Print(L'"'));
+ EXPECT_EQ("L'?' (63, 0x3F)", Print(L'?'));
+ EXPECT_EQ("L'\\\\' (92, 0x5C)", Print(L'\\'));
+ EXPECT_EQ("L'\\a' (7)", Print(L'\a'));
+ EXPECT_EQ("L'\\b' (8)", Print(L'\b'));
+ EXPECT_EQ("L'\\f' (12, 0xC)", Print(L'\f'));
+ EXPECT_EQ("L'\\n' (10, 0xA)", Print(L'\n'));
+ EXPECT_EQ("L'\\r' (13, 0xD)", Print(L'\r'));
+ EXPECT_EQ("L'\\t' (9)", Print(L'\t'));
+ EXPECT_EQ("L'\\v' (11, 0xB)", Print(L'\v'));
+ EXPECT_EQ("L'\\x7F' (127)", Print(L'\x7F'));
+ EXPECT_EQ("L'\\xFF' (255)", Print(L'\xFF'));
+ EXPECT_EQ("L' ' (32, 0x20)", Print(L' '));
+ EXPECT_EQ("L'a' (97, 0x61)", Print(L'a'));
+ EXPECT_EQ("L'\\x576' (1398)", Print(static_cast<wchar_t>(0x576)));
+ EXPECT_EQ("L'\\xC74D' (51021)", Print(static_cast<wchar_t>(0xC74D)));
+}
+
+// Test that Int64 provides more storage than wchar_t.
+TEST(PrintTypeSizeTest, Wchar_t) {
+ EXPECT_LT(sizeof(wchar_t), sizeof(testing::internal::Int64));
+}
+
+// Various integer types.
+TEST(PrintBuiltInTypeTest, Integer) {
+ EXPECT_EQ("'\\xFF' (255)", Print(static_cast<unsigned char>(255))); // uint8
+ EXPECT_EQ("'\\x80' (-128)", Print(static_cast<signed char>(-128))); // int8
+ EXPECT_EQ("65535", Print(USHRT_MAX)); // uint16
+ EXPECT_EQ("-32768", Print(SHRT_MIN)); // int16
+ EXPECT_EQ("4294967295", Print(UINT_MAX)); // uint32
+ EXPECT_EQ("-2147483648", Print(INT_MIN)); // int32
+ EXPECT_EQ("18446744073709551615",
+ Print(static_cast<testing::internal::UInt64>(-1))); // uint64
+ EXPECT_EQ("-9223372036854775808",
+ Print(static_cast<testing::internal::Int64>(1) << 63)); // int64
+}
+
+// Size types.
+TEST(PrintBuiltInTypeTest, Size_t) {
+ EXPECT_EQ("1", Print(sizeof('a'))); // size_t.
+#if !GTEST_OS_WINDOWS
+ // Windows has no ssize_t type.
+ EXPECT_EQ("-2", Print(static_cast<ssize_t>(-2))); // ssize_t.
+#endif // !GTEST_OS_WINDOWS
+}
+
+// Floating-points.
+TEST(PrintBuiltInTypeTest, FloatingPoints) {
+ EXPECT_EQ("1.5", Print(1.5f)); // float
+ EXPECT_EQ("-2.5", Print(-2.5)); // double
+}
+
+// Since ::std::stringstream::operator<<(const void *) formats the pointer
+// output differently with different compilers, we have to create the expected
+// output first and use it as our expectation.
+static string PrintPointer(const void *p) {
+ ::std::stringstream expected_result_stream;
+ expected_result_stream << p;
+ return expected_result_stream.str();
+}
+
+// Tests printing C strings.
+
+// const char*.
+TEST(PrintCStringTest, Const) {
+ const char* p = "World";
+ EXPECT_EQ(PrintPointer(p) + " pointing to \"World\"", Print(p));
+}
+
+// char*.
+TEST(PrintCStringTest, NonConst) {
+ char p[] = "Hi";
+ EXPECT_EQ(PrintPointer(p) + " pointing to \"Hi\"",
+ Print(static_cast<char*>(p)));
+}
+
+// NULL C string.
+TEST(PrintCStringTest, Null) {
+ const char* p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// Tests that C strings are escaped properly.
+TEST(PrintCStringTest, EscapesProperly) {
+ const char* p = "'\"?\\\a\b\f\n\r\t\v\x7F\xFF a";
+ EXPECT_EQ(PrintPointer(p) + " pointing to \"'\\\"?\\\\\\a\\b\\f"
+ "\\n\\r\\t\\v\\x7F\\xFF a\"",
+ Print(p));
+}
+
+
+
+// MSVC compiler can be configured to define whar_t as a typedef
+// of unsigned short. Defining an overload for const wchar_t* in that case
+// would cause pointers to unsigned shorts be printed as wide strings,
+// possibly accessing more memory than intended and causing invalid
+// memory accesses. MSVC defines _NATIVE_WCHAR_T_DEFINED symbol when
+// wchar_t is implemented as a native type.
+#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED)
+
+// const wchar_t*.
+TEST(PrintWideCStringTest, Const) {
+ const wchar_t* p = L"World";
+ EXPECT_EQ(PrintPointer(p) + " pointing to L\"World\"", Print(p));
+}
+
+// wchar_t*.
+TEST(PrintWideCStringTest, NonConst) {
+ wchar_t p[] = L"Hi";
+ EXPECT_EQ(PrintPointer(p) + " pointing to L\"Hi\"",
+ Print(static_cast<wchar_t*>(p)));
+}
+
+// NULL wide C string.
+TEST(PrintWideCStringTest, Null) {
+ const wchar_t* p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// Tests that wide C strings are escaped properly.
+TEST(PrintWideCStringTest, EscapesProperly) {
+ const wchar_t s[] = {'\'', '"', '?', '\\', '\a', '\b', '\f', '\n', '\r',
+ '\t', '\v', 0xD3, 0x576, 0x8D3, 0xC74D, ' ', 'a', '\0'};
+ EXPECT_EQ(PrintPointer(s) + " pointing to L\"'\\\"?\\\\\\a\\b\\f"
+ "\\n\\r\\t\\v\\xD3\\x576\\x8D3\\xC74D a\"",
+ Print(static_cast<const wchar_t*>(s)));
+}
+#endif // native wchar_t
+
+// Tests printing pointers to other char types.
+
+// signed char*.
+TEST(PrintCharPointerTest, SignedChar) {
+ signed char* p = reinterpret_cast<signed char*>(0x1234);
+ EXPECT_EQ(PrintPointer(p), Print(p));
+ p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// const signed char*.
+TEST(PrintCharPointerTest, ConstSignedChar) {
+ signed char* p = reinterpret_cast<signed char*>(0x1234);
+ EXPECT_EQ(PrintPointer(p), Print(p));
+ p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// unsigned char*.
+TEST(PrintCharPointerTest, UnsignedChar) {
+ unsigned char* p = reinterpret_cast<unsigned char*>(0x1234);
+ EXPECT_EQ(PrintPointer(p), Print(p));
+ p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// const unsigned char*.
+TEST(PrintCharPointerTest, ConstUnsignedChar) {
+ const unsigned char* p = reinterpret_cast<const unsigned char*>(0x1234);
+ EXPECT_EQ(PrintPointer(p), Print(p));
+ p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// Tests printing pointers to simple, built-in types.
+
+// bool*.
+TEST(PrintPointerToBuiltInTypeTest, Bool) {
+ bool* p = reinterpret_cast<bool*>(0xABCD);
+ EXPECT_EQ(PrintPointer(p), Print(p));
+ p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// void*.
+TEST(PrintPointerToBuiltInTypeTest, Void) {
+ void* p = reinterpret_cast<void*>(0xABCD);
+ EXPECT_EQ(PrintPointer(p), Print(p));
+ p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// const void*.
+TEST(PrintPointerToBuiltInTypeTest, ConstVoid) {
+ const void* p = reinterpret_cast<const void*>(0xABCD);
+ EXPECT_EQ(PrintPointer(p), Print(p));
+ p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// Tests printing pointers to pointers.
+TEST(PrintPointerToPointerTest, IntPointerPointer) {
+ int** p = reinterpret_cast<int**>(0xABCD);
+ EXPECT_EQ(PrintPointer(p), Print(p));
+ p = NULL;
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// Tests printing (non-member) function pointers.
+
+void MyFunction(int /* n */) {}
+
+TEST(PrintPointerTest, NonMemberFunctionPointer) {
+ // We cannot directly cast &MyFunction to const void* because the
+ // standard disallows casting between pointers to functions and
+ // pointers to objects, and some compilers (e.g. GCC 3.4) enforce
+ // this limitation.
+ EXPECT_EQ(
+ PrintPointer(reinterpret_cast<const void*>(
+ reinterpret_cast<internal::BiggestInt>(&MyFunction))),
+ Print(&MyFunction));
+ int (*p)(bool) = NULL; // NOLINT
+ EXPECT_EQ("NULL", Print(p));
+}
+
+// An assertion predicate determining whether a one string is a prefix for
+// another.
+template <typename StringType>
+AssertionResult HasPrefix(const StringType& str, const StringType& prefix) {
+ if (str.find(prefix, 0) == 0)
+ return AssertionSuccess();
+
+ const bool is_wide_string = sizeof(prefix[0]) > 1;
+ const char* const begin_string_quote = is_wide_string ? "L\"" : "\"";
+ return AssertionFailure()
+ << begin_string_quote << prefix << "\" is not a prefix of "
+ << begin_string_quote << str << "\"\n";
+}
+
+// Tests printing member variable pointers. Although they are called
+// pointers, they don't point to a location in the address space.
+// Their representation is implementation-defined. Thus they will be
+// printed as raw bytes.
+
+struct Foo {
+ public:
+ virtual ~Foo() {}
+ int MyMethod(char x) { return x + 1; }
+ virtual char MyVirtualMethod(int /* n */) { return 'a'; }
+
+ int value;
+};
+
+TEST(PrintPointerTest, MemberVariablePointer) {
+ EXPECT_TRUE(HasPrefix(Print(&Foo::value),
+ Print(sizeof(&Foo::value)) + "-byte object "));
+ int (Foo::*p) = NULL; // NOLINT
+ EXPECT_TRUE(HasPrefix(Print(p),
+ Print(sizeof(p)) + "-byte object "));
+}
+
+// Tests printing member function pointers. Although they are called
+// pointers, they don't point to a location in the address space.
+// Their representation is implementation-defined. Thus they will be
+// printed as raw bytes.
+TEST(PrintPointerTest, MemberFunctionPointer) {
+ EXPECT_TRUE(HasPrefix(Print(&Foo::MyMethod),
+ Print(sizeof(&Foo::MyMethod)) + "-byte object "));
+ EXPECT_TRUE(
+ HasPrefix(Print(&Foo::MyVirtualMethod),
+ Print(sizeof((&Foo::MyVirtualMethod))) + "-byte object "));
+ int (Foo::*p)(char) = NULL; // NOLINT
+ EXPECT_TRUE(HasPrefix(Print(p),
+ Print(sizeof(p)) + "-byte object "));
+}
+
+// Tests printing C arrays.
+
+// The difference between this and Print() is that it ensures that the
+// argument is a reference to an array.
+template <typename T, size_t N>
+string PrintArrayHelper(T (&a)[N]) {
+ return Print(a);
+}
+
+// One-dimensional array.
+TEST(PrintArrayTest, OneDimensionalArray) {
+ int a[5] = { 1, 2, 3, 4, 5 };
+ EXPECT_EQ("{ 1, 2, 3, 4, 5 }", PrintArrayHelper(a));
+}
+
+// Two-dimensional array.
+TEST(PrintArrayTest, TwoDimensionalArray) {
+ int a[2][5] = {
+ { 1, 2, 3, 4, 5 },
+ { 6, 7, 8, 9, 0 }
+ };
+ EXPECT_EQ("{ { 1, 2, 3, 4, 5 }, { 6, 7, 8, 9, 0 } }", PrintArrayHelper(a));
+}
+
+// Array of const elements.
+TEST(PrintArrayTest, ConstArray) {
+ const bool a[1] = { false };
+ EXPECT_EQ("{ false }", PrintArrayHelper(a));
+}
+
+// Char array.
+TEST(PrintArrayTest, CharArray) {
+ // Array a contains '\0' in the middle and doesn't end with '\0'.
+ char a[3] = { 'H', '\0', 'i' };
+ EXPECT_EQ("\"H\\0i\"", PrintArrayHelper(a));
+}
+
+// Const char array.
+TEST(PrintArrayTest, ConstCharArray) {
+ const char a[4] = "\0Hi";
+ EXPECT_EQ("\"\\0Hi\\0\"", PrintArrayHelper(a));
+}
+
+// Array of objects.
+TEST(PrintArrayTest, ObjectArray) {
+ string a[3] = { "Hi", "Hello", "Ni hao" };
+ EXPECT_EQ("{ \"Hi\", \"Hello\", \"Ni hao\" }", PrintArrayHelper(a));
+}
+
+// Array with many elements.
+TEST(PrintArrayTest, BigArray) {
+ int a[100] = { 1, 2, 3 };
+ EXPECT_EQ("{ 1, 2, 3, 0, 0, 0, 0, 0, ..., 0, 0, 0, 0, 0, 0, 0, 0 }",
+ PrintArrayHelper(a));
+}
+
+// Tests printing ::string and ::std::string.
+
+#if GTEST_HAS_GLOBAL_STRING
+// ::string.
+TEST(PrintStringTest, StringInGlobalNamespace) {
+ const char s[] = "'\"?\\\a\b\f\n\0\r\t\v\x7F\xFF a";
+ const ::string str(s, sizeof(s));
+ EXPECT_EQ("\"'\\\"?\\\\\\a\\b\\f\\n\\0\\r\\t\\v\\x7F\\xFF a\\0\"",
+ Print(str));
+}
+#endif // GTEST_HAS_GLOBAL_STRING
+
+// ::std::string.
+TEST(PrintStringTest, StringInStdNamespace) {
+ const char s[] = "'\"?\\\a\b\f\n\0\r\t\v\x7F\xFF a";
+ const ::std::string str(s, sizeof(s));
+ EXPECT_EQ("\"'\\\"?\\\\\\a\\b\\f\\n\\0\\r\\t\\v\\x7F\\xFF a\\0\"",
+ Print(str));
+}
+
+TEST(PrintStringTest, StringAmbiguousHex) {
+ // "\x6BANANA" is ambiguous, it can be interpreted as starting with either of:
+ // '\x6', '\x6B', or '\x6BA'.
+
+ // a hex escaping sequence following by a decimal digit
+ EXPECT_EQ("\"0\\x12\" \"3\"", Print(::std::string("0\x12" "3")));
+ // a hex escaping sequence following by a hex digit (lower-case)
+ EXPECT_EQ("\"mm\\x6\" \"bananas\"", Print(::std::string("mm\x6" "bananas")));
+ // a hex escaping sequence following by a hex digit (upper-case)
+ EXPECT_EQ("\"NOM\\x6\" \"BANANA\"", Print(::std::string("NOM\x6" "BANANA")));
+ // a hex escaping sequence following by a non-xdigit
+ EXPECT_EQ("\"!\\x5-!\"", Print(::std::string("!\x5-!")));
+}
+
+// Tests printing ::wstring and ::std::wstring.
+
+#if GTEST_HAS_GLOBAL_WSTRING
+// ::wstring.
+TEST(PrintWideStringTest, StringInGlobalNamespace) {
+ const wchar_t s[] = L"'\"?\\\a\b\f\n\0\r\t\v\xD3\x576\x8D3\xC74D a";
+ const ::wstring str(s, sizeof(s)/sizeof(wchar_t));
+ EXPECT_EQ("L\"'\\\"?\\\\\\a\\b\\f\\n\\0\\r\\t\\v"
+ "\\xD3\\x576\\x8D3\\xC74D a\\0\"",
+ Print(str));
+}
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+#if GTEST_HAS_STD_WSTRING
+// ::std::wstring.
+TEST(PrintWideStringTest, StringInStdNamespace) {
+ const wchar_t s[] = L"'\"?\\\a\b\f\n\0\r\t\v\xD3\x576\x8D3\xC74D a";
+ const ::std::wstring str(s, sizeof(s)/sizeof(wchar_t));
+ EXPECT_EQ("L\"'\\\"?\\\\\\a\\b\\f\\n\\0\\r\\t\\v"
+ "\\xD3\\x576\\x8D3\\xC74D a\\0\"",
+ Print(str));
+}
+
+TEST(PrintWideStringTest, StringAmbiguousHex) {
+ // same for wide strings.
+ EXPECT_EQ("L\"0\\x12\" L\"3\"", Print(::std::wstring(L"0\x12" L"3")));
+ EXPECT_EQ("L\"mm\\x6\" L\"bananas\"",
+ Print(::std::wstring(L"mm\x6" L"bananas")));
+ EXPECT_EQ("L\"NOM\\x6\" L\"BANANA\"",
+ Print(::std::wstring(L"NOM\x6" L"BANANA")));
+ EXPECT_EQ("L\"!\\x5-!\"", Print(::std::wstring(L"!\x5-!")));
+}
+#endif // GTEST_HAS_STD_WSTRING
+
+// Tests printing types that support generic streaming (i.e. streaming
+// to std::basic_ostream<Char, CharTraits> for any valid Char and
+// CharTraits types).
+
+// Tests printing a non-template type that supports generic streaming.
+
+class AllowsGenericStreaming {};
+
+template <typename Char, typename CharTraits>
+std::basic_ostream<Char, CharTraits>& operator<<(
+ std::basic_ostream<Char, CharTraits>& os,
+ const AllowsGenericStreaming& /* a */) {
+ return os << "AllowsGenericStreaming";
+}
+
+TEST(PrintTypeWithGenericStreamingTest, NonTemplateType) {
+ AllowsGenericStreaming a;
+ EXPECT_EQ("AllowsGenericStreaming", Print(a));
+}
+
+// Tests printing a template type that supports generic streaming.
+
+template <typename T>
+class AllowsGenericStreamingTemplate {};
+
+template <typename Char, typename CharTraits, typename T>
+std::basic_ostream<Char, CharTraits>& operator<<(
+ std::basic_ostream<Char, CharTraits>& os,
+ const AllowsGenericStreamingTemplate<T>& /* a */) {
+ return os << "AllowsGenericStreamingTemplate";
+}
+
+TEST(PrintTypeWithGenericStreamingTest, TemplateType) {
+ AllowsGenericStreamingTemplate<int> a;
+ EXPECT_EQ("AllowsGenericStreamingTemplate", Print(a));
+}
+
+// Tests printing a type that supports generic streaming and can be
+// implicitly converted to another printable type.
+
+template <typename T>
+class AllowsGenericStreamingAndImplicitConversionTemplate {
+ public:
+ operator bool() const { return false; }
+};
+
+template <typename Char, typename CharTraits, typename T>
+std::basic_ostream<Char, CharTraits>& operator<<(
+ std::basic_ostream<Char, CharTraits>& os,
+ const AllowsGenericStreamingAndImplicitConversionTemplate<T>& /* a */) {
+ return os << "AllowsGenericStreamingAndImplicitConversionTemplate";
+}
+
+TEST(PrintTypeWithGenericStreamingTest, TypeImplicitlyConvertible) {
+ AllowsGenericStreamingAndImplicitConversionTemplate<int> a;
+ EXPECT_EQ("AllowsGenericStreamingAndImplicitConversionTemplate", Print(a));
+}
+
+#if GTEST_HAS_STRING_PIECE_
+
+// Tests printing StringPiece.
+
+TEST(PrintStringPieceTest, SimpleStringPiece) {
+ const StringPiece sp = "Hello";
+ EXPECT_EQ("\"Hello\"", Print(sp));
+}
+
+TEST(PrintStringPieceTest, UnprintableCharacters) {
+ const char str[] = "NUL (\0) and \r\t";
+ const StringPiece sp(str, sizeof(str) - 1);
+ EXPECT_EQ("\"NUL (\\0) and \\r\\t\"", Print(sp));
+}
+
+#endif // GTEST_HAS_STRING_PIECE_
+
+// Tests printing STL containers.
+
+TEST(PrintStlContainerTest, EmptyDeque) {
+ deque<char> empty;
+ EXPECT_EQ("{}", Print(empty));
+}
+
+TEST(PrintStlContainerTest, NonEmptyDeque) {
+ deque<int> non_empty;
+ non_empty.push_back(1);
+ non_empty.push_back(3);
+ EXPECT_EQ("{ 1, 3 }", Print(non_empty));
+}
+
+#if GTEST_HAS_HASH_MAP_
+
+TEST(PrintStlContainerTest, OneElementHashMap) {
+ hash_map<int, char> map1;
+ map1[1] = 'a';
+ EXPECT_EQ("{ (1, 'a' (97, 0x61)) }", Print(map1));
+}
+
+TEST(PrintStlContainerTest, HashMultiMap) {
+ hash_multimap<int, bool> map1;
+ map1.insert(make_pair(5, true));
+ map1.insert(make_pair(5, false));
+
+ // Elements of hash_multimap can be printed in any order.
+ const string result = Print(map1);
+ EXPECT_TRUE(result == "{ (5, true), (5, false) }" ||
+ result == "{ (5, false), (5, true) }")
+ << " where Print(map1) returns \"" << result << "\".";
+}
+
+#endif // GTEST_HAS_HASH_MAP_
+
+#if GTEST_HAS_HASH_SET_
+
+TEST(PrintStlContainerTest, HashSet) {
+ hash_set<string> set1;
+ set1.insert("hello");
+ EXPECT_EQ("{ \"hello\" }", Print(set1));
+}
+
+TEST(PrintStlContainerTest, HashMultiSet) {
+ const int kSize = 5;
+ int a[kSize] = { 1, 1, 2, 5, 1 };
+ hash_multiset<int> set1(a, a + kSize);
+
+ // Elements of hash_multiset can be printed in any order.
+ const string result = Print(set1);
+ const string expected_pattern = "{ d, d, d, d, d }"; // d means a digit.
+
+ // Verifies the result matches the expected pattern; also extracts
+ // the numbers in the result.
+ ASSERT_EQ(expected_pattern.length(), result.length());
+ std::vector<int> numbers;
+ for (size_t i = 0; i != result.length(); i++) {
+ if (expected_pattern[i] == 'd') {
+ ASSERT_TRUE(isdigit(static_cast<unsigned char>(result[i])) != 0);
+ numbers.push_back(result[i] - '0');
+ } else {
+ EXPECT_EQ(expected_pattern[i], result[i]) << " where result is "
+ << result;
+ }
+ }
+
+ // Makes sure the result contains the right numbers.
+ std::sort(numbers.begin(), numbers.end());
+ std::sort(a, a + kSize);
+ EXPECT_TRUE(std::equal(a, a + kSize, numbers.begin()));
+}
+
+#endif // GTEST_HAS_HASH_SET_
+
+TEST(PrintStlContainerTest, List) {
+ const string a[] = {
+ "hello",
+ "world"
+ };
+ const list<string> strings(a, a + 2);
+ EXPECT_EQ("{ \"hello\", \"world\" }", Print(strings));
+}
+
+TEST(PrintStlContainerTest, Map) {
+ map<int, bool> map1;
+ map1[1] = true;
+ map1[5] = false;
+ map1[3] = true;
+ EXPECT_EQ("{ (1, true), (3, true), (5, false) }", Print(map1));
+}
+
+TEST(PrintStlContainerTest, MultiMap) {
+ multimap<bool, int> map1;
+ // The make_pair template function would deduce the type as
+ // pair<bool, int> here, and since the key part in a multimap has to
+ // be constant, without a templated ctor in the pair class (as in
+ // libCstd on Solaris), make_pair call would fail to compile as no
+ // implicit conversion is found. Thus explicit typename is used
+ // here instead.
+ map1.insert(pair<const bool, int>(true, 0));
+ map1.insert(pair<const bool, int>(true, 1));
+ map1.insert(pair<const bool, int>(false, 2));
+ EXPECT_EQ("{ (false, 2), (true, 0), (true, 1) }", Print(map1));
+}
+
+TEST(PrintStlContainerTest, Set) {
+ const unsigned int a[] = { 3, 0, 5 };
+ set<unsigned int> set1(a, a + 3);
+ EXPECT_EQ("{ 0, 3, 5 }", Print(set1));
+}
+
+TEST(PrintStlContainerTest, MultiSet) {
+ const int a[] = { 1, 1, 2, 5, 1 };
+ multiset<int> set1(a, a + 5);
+ EXPECT_EQ("{ 1, 1, 1, 2, 5 }", Print(set1));
+}
+
+TEST(PrintStlContainerTest, Pair) {
+ pair<const bool, int> p(true, 5);
+ EXPECT_EQ("(true, 5)", Print(p));
+}
+
+TEST(PrintStlContainerTest, Vector) {
+ vector<int> v;
+ v.push_back(1);
+ v.push_back(2);
+ EXPECT_EQ("{ 1, 2 }", Print(v));
+}
+
+TEST(PrintStlContainerTest, LongSequence) {
+ const int a[100] = { 1, 2, 3 };
+ const vector<int> v(a, a + 100);
+ EXPECT_EQ("{ 1, 2, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ... }", Print(v));
+}
+
+TEST(PrintStlContainerTest, NestedContainer) {
+ const int a1[] = { 1, 2 };
+ const int a2[] = { 3, 4, 5 };
+ const list<int> l1(a1, a1 + 2);
+ const list<int> l2(a2, a2 + 3);
+
+ vector<list<int> > v;
+ v.push_back(l1);
+ v.push_back(l2);
+ EXPECT_EQ("{ { 1, 2 }, { 3, 4, 5 } }", Print(v));
+}
+
+TEST(PrintStlContainerTest, OneDimensionalNativeArray) {
+ const int a[3] = { 1, 2, 3 };
+ NativeArray<int> b(a, 3, kReference);
+ EXPECT_EQ("{ 1, 2, 3 }", Print(b));
+}
+
+TEST(PrintStlContainerTest, TwoDimensionalNativeArray) {
+ const int a[2][3] = { { 1, 2, 3 }, { 4, 5, 6 } };
+ NativeArray<int[3]> b(a, 2, kReference);
+ EXPECT_EQ("{ { 1, 2, 3 }, { 4, 5, 6 } }", Print(b));
+}
+
+// Tests that a class named iterator isn't treated as a container.
+
+struct iterator {
+ char x;
+};
+
+TEST(PrintStlContainerTest, Iterator) {
+ iterator it = {};
+ EXPECT_EQ("1-byte object <00>", Print(it));
+}
+
+// Tests that a class named const_iterator isn't treated as a container.
+
+struct const_iterator {
+ char x;
+};
+
+TEST(PrintStlContainerTest, ConstIterator) {
+ const_iterator it = {};
+ EXPECT_EQ("1-byte object <00>", Print(it));
+}
+
+#if GTEST_HAS_TR1_TUPLE
+// Tests printing tuples.
+
+// Tuples of various arities.
+TEST(PrintTupleTest, VariousSizes) {
+ tuple<> t0;
+ EXPECT_EQ("()", Print(t0));
+
+ tuple<int> t1(5);
+ EXPECT_EQ("(5)", Print(t1));
+
+ tuple<char, bool> t2('a', true);
+ EXPECT_EQ("('a' (97, 0x61), true)", Print(t2));
+
+ tuple<bool, int, int> t3(false, 2, 3);
+ EXPECT_EQ("(false, 2, 3)", Print(t3));
+
+ tuple<bool, int, int, int> t4(false, 2, 3, 4);
+ EXPECT_EQ("(false, 2, 3, 4)", Print(t4));
+
+ tuple<bool, int, int, int, bool> t5(false, 2, 3, 4, true);
+ EXPECT_EQ("(false, 2, 3, 4, true)", Print(t5));
+
+ tuple<bool, int, int, int, bool, int> t6(false, 2, 3, 4, true, 6);
+ EXPECT_EQ("(false, 2, 3, 4, true, 6)", Print(t6));
+
+ tuple<bool, int, int, int, bool, int, int> t7(false, 2, 3, 4, true, 6, 7);
+ EXPECT_EQ("(false, 2, 3, 4, true, 6, 7)", Print(t7));
+
+ tuple<bool, int, int, int, bool, int, int, bool> t8(
+ false, 2, 3, 4, true, 6, 7, true);
+ EXPECT_EQ("(false, 2, 3, 4, true, 6, 7, true)", Print(t8));
+
+ tuple<bool, int, int, int, bool, int, int, bool, int> t9(
+ false, 2, 3, 4, true, 6, 7, true, 9);
+ EXPECT_EQ("(false, 2, 3, 4, true, 6, 7, true, 9)", Print(t9));
+
+ const char* const str = "8";
+ tuple<bool, char, short, testing::internal::Int32, // NOLINT
+ testing::internal::Int64, float, double, const char*, void*, string>
+ t10(false, 'a', 3, 4, 5, 1.5F, -2.5, str, NULL, "10");
+ EXPECT_EQ("(false, 'a' (97, 0x61), 3, 4, 5, 1.5, -2.5, " + PrintPointer(str) +
+ " pointing to \"8\", NULL, \"10\")",
+ Print(t10));
+}
+
+// Nested tuples.
+TEST(PrintTupleTest, NestedTuple) {
+ tuple<tuple<int, bool>, char> nested(make_tuple(5, true), 'a');
+ EXPECT_EQ("((5, true), 'a' (97, 0x61))", Print(nested));
+}
+
+#endif // GTEST_HAS_TR1_TUPLE
+
+// Tests printing user-defined unprintable types.
+
+// Unprintable types in the global namespace.
+TEST(PrintUnprintableTypeTest, InGlobalNamespace) {
+ EXPECT_EQ("1-byte object <00>",
+ Print(UnprintableTemplateInGlobal<char>()));
+}
+
+// Unprintable types in a user namespace.
+TEST(PrintUnprintableTypeTest, InUserNamespace) {
+ EXPECT_EQ("16-byte object <EF-12 00-00 34-AB 00-00 00-00 00-00 00-00 00-00>",
+ Print(::foo::UnprintableInFoo()));
+}
+
+// Unprintable types are that too big to be printed completely.
+
+struct Big {
+ Big() { memset(array, 0, sizeof(array)); }
+ char array[257];
+};
+
+TEST(PrintUnpritableTypeTest, BigObject) {
+ EXPECT_EQ("257-byte object <00-00 00-00 00-00 00-00 00-00 00-00 "
+ "00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 "
+ "00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 "
+ "00-00 00-00 00-00 00-00 00-00 00-00 ... 00-00 00-00 00-00 "
+ "00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 "
+ "00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 "
+ "00-00 00-00 00-00 00-00 00-00 00-00 00-00 00-00 00>",
+ Print(Big()));
+}
+
+// Tests printing user-defined streamable types.
+
+// Streamable types in the global namespace.
+TEST(PrintStreamableTypeTest, InGlobalNamespace) {
+ StreamableInGlobal x;
+ EXPECT_EQ("StreamableInGlobal", Print(x));
+ EXPECT_EQ("StreamableInGlobal*", Print(&x));
+}
+
+// Printable template types in a user namespace.
+TEST(PrintStreamableTypeTest, TemplateTypeInUserNamespace) {
+ EXPECT_EQ("StreamableTemplateInFoo: 0",
+ Print(::foo::StreamableTemplateInFoo<int>()));
+}
+
+// Tests printing user-defined types that have a PrintTo() function.
+TEST(PrintPrintableTypeTest, InUserNamespace) {
+ EXPECT_EQ("PrintableViaPrintTo: 0",
+ Print(::foo::PrintableViaPrintTo()));
+}
+
+// Tests printing a pointer to a user-defined type that has a <<
+// operator for its pointer.
+TEST(PrintPrintableTypeTest, PointerInUserNamespace) {
+ ::foo::PointerPrintable x;
+ EXPECT_EQ("PointerPrintable*", Print(&x));
+}
+
+// Tests printing user-defined class template that have a PrintTo() function.
+TEST(PrintPrintableTypeTest, TemplateInUserNamespace) {
+ EXPECT_EQ("PrintableViaPrintToTemplate: 5",
+ Print(::foo::PrintableViaPrintToTemplate<int>(5)));
+}
+
+#if GTEST_HAS_PROTOBUF_
+
+// Tests printing a protocol message.
+TEST(PrintProtocolMessageTest, PrintsShortDebugString) {
+ testing::internal::TestMessage msg;
+ msg.set_member("yes");
+ EXPECT_EQ("<member:\"yes\">", Print(msg));
+}
+
+// Tests printing a short proto2 message.
+TEST(PrintProto2MessageTest, PrintsShortDebugStringWhenItIsShort) {
+ testing::internal::FooMessage msg;
+ msg.set_int_field(2);
+ msg.set_string_field("hello");
+ EXPECT_PRED2(RE::FullMatch, Print(msg),
+ "<int_field:\\s*2\\s+string_field:\\s*\"hello\">");
+}
+
+// Tests printing a long proto2 message.
+TEST(PrintProto2MessageTest, PrintsDebugStringWhenItIsLong) {
+ testing::internal::FooMessage msg;
+ msg.set_int_field(2);
+ msg.set_string_field("hello");
+ msg.add_names("peter");
+ msg.add_names("paul");
+ msg.add_names("mary");
+ EXPECT_PRED2(RE::FullMatch, Print(msg),
+ "<\n"
+ "int_field:\\s*2\n"
+ "string_field:\\s*\"hello\"\n"
+ "names:\\s*\"peter\"\n"
+ "names:\\s*\"paul\"\n"
+ "names:\\s*\"mary\"\n"
+ ">");
+}
+
+#endif // GTEST_HAS_PROTOBUF_
+
+// Tests that the universal printer prints both the address and the
+// value of a reference.
+TEST(PrintReferenceTest, PrintsAddressAndValue) {
+ int n = 5;
+ EXPECT_EQ("@" + PrintPointer(&n) + " 5", PrintByRef(n));
+
+ int a[2][3] = {
+ { 0, 1, 2 },
+ { 3, 4, 5 }
+ };
+ EXPECT_EQ("@" + PrintPointer(a) + " { { 0, 1, 2 }, { 3, 4, 5 } }",
+ PrintByRef(a));
+
+ const ::foo::UnprintableInFoo x;
+ EXPECT_EQ("@" + PrintPointer(&x) + " 16-byte object "
+ "<EF-12 00-00 34-AB 00-00 00-00 00-00 00-00 00-00>",
+ PrintByRef(x));
+}
+
+// Tests that the universal printer prints a function pointer passed by
+// reference.
+TEST(PrintReferenceTest, HandlesFunctionPointer) {
+ void (*fp)(int n) = &MyFunction;
+ const string fp_pointer_string =
+ PrintPointer(reinterpret_cast<const void*>(&fp));
+ // We cannot directly cast &MyFunction to const void* because the
+ // standard disallows casting between pointers to functions and
+ // pointers to objects, and some compilers (e.g. GCC 3.4) enforce
+ // this limitation.
+ const string fp_string = PrintPointer(reinterpret_cast<const void*>(
+ reinterpret_cast<internal::BiggestInt>(fp)));
+ EXPECT_EQ("@" + fp_pointer_string + " " + fp_string,
+ PrintByRef(fp));
+}
+
+// Tests that the universal printer prints a member function pointer
+// passed by reference.
+TEST(PrintReferenceTest, HandlesMemberFunctionPointer) {
+ int (Foo::*p)(char ch) = &Foo::MyMethod;
+ EXPECT_TRUE(HasPrefix(
+ PrintByRef(p),
+ "@" + PrintPointer(reinterpret_cast<const void*>(&p)) + " " +
+ Print(sizeof(p)) + "-byte object "));
+
+ char (Foo::*p2)(int n) = &Foo::MyVirtualMethod;
+ EXPECT_TRUE(HasPrefix(
+ PrintByRef(p2),
+ "@" + PrintPointer(reinterpret_cast<const void*>(&p2)) + " " +
+ Print(sizeof(p2)) + "-byte object "));
+}
+
+// Tests that the universal printer prints a member variable pointer
+// passed by reference.
+TEST(PrintReferenceTest, HandlesMemberVariablePointer) {
+ int (Foo::*p) = &Foo::value; // NOLINT
+ EXPECT_TRUE(HasPrefix(
+ PrintByRef(p),
+ "@" + PrintPointer(&p) + " " + Print(sizeof(p)) + "-byte object "));
+}
+
+// Useful for testing PrintToString(). We cannot use EXPECT_EQ()
+// there as its implementation uses PrintToString(). The caller must
+// ensure that 'value' has no side effect.
+#define EXPECT_PRINT_TO_STRING_(value, expected_string) \
+ EXPECT_TRUE(PrintToString(value) == (expected_string)) \
+ << " where " #value " prints as " << (PrintToString(value))
+
+TEST(PrintToStringTest, WorksForScalar) {
+ EXPECT_PRINT_TO_STRING_(123, "123");
+}
+
+TEST(PrintToStringTest, WorksForPointerToConstChar) {
+ const char* p = "hello";
+ EXPECT_PRINT_TO_STRING_(p, "\"hello\"");
+}
+
+TEST(PrintToStringTest, WorksForPointerToNonConstChar) {
+ char s[] = "hello";
+ char* p = s;
+ EXPECT_PRINT_TO_STRING_(p, "\"hello\"");
+}
+
+TEST(PrintToStringTest, WorksForArray) {
+ int n[3] = { 1, 2, 3 };
+ EXPECT_PRINT_TO_STRING_(n, "{ 1, 2, 3 }");
+}
+
+#undef EXPECT_PRINT_TO_STRING_
+
+TEST(UniversalTersePrintTest, WorksForNonReference) {
+ ::std::stringstream ss;
+ UniversalTersePrint(123, &ss);
+ EXPECT_EQ("123", ss.str());
+}
+
+TEST(UniversalTersePrintTest, WorksForReference) {
+ const int& n = 123;
+ ::std::stringstream ss;
+ UniversalTersePrint(n, &ss);
+ EXPECT_EQ("123", ss.str());
+}
+
+TEST(UniversalTersePrintTest, WorksForCString) {
+ const char* s1 = "abc";
+ ::std::stringstream ss1;
+ UniversalTersePrint(s1, &ss1);
+ EXPECT_EQ("\"abc\"", ss1.str());
+
+ char* s2 = const_cast<char*>(s1);
+ ::std::stringstream ss2;
+ UniversalTersePrint(s2, &ss2);
+ EXPECT_EQ("\"abc\"", ss2.str());
+
+ const char* s3 = NULL;
+ ::std::stringstream ss3;
+ UniversalTersePrint(s3, &ss3);
+ EXPECT_EQ("NULL", ss3.str());
+}
+
+TEST(UniversalPrintTest, WorksForNonReference) {
+ ::std::stringstream ss;
+ UniversalPrint(123, &ss);
+ EXPECT_EQ("123", ss.str());
+}
+
+TEST(UniversalPrintTest, WorksForReference) {
+ const int& n = 123;
+ ::std::stringstream ss;
+ UniversalPrint(n, &ss);
+ EXPECT_EQ("123", ss.str());
+}
+
+TEST(UniversalPrintTest, WorksForCString) {
+ const char* s1 = "abc";
+ ::std::stringstream ss1;
+ UniversalPrint(s1, &ss1);
+ EXPECT_EQ(PrintPointer(s1) + " pointing to \"abc\"", string(ss1.str()));
+
+ char* s2 = const_cast<char*>(s1);
+ ::std::stringstream ss2;
+ UniversalPrint(s2, &ss2);
+ EXPECT_EQ(PrintPointer(s2) + " pointing to \"abc\"", string(ss2.str()));
+
+ const char* s3 = NULL;
+ ::std::stringstream ss3;
+ UniversalPrint(s3, &ss3);
+ EXPECT_EQ("NULL", ss3.str());
+}
+
+
+#if GTEST_HAS_TR1_TUPLE
+
+TEST(UniversalTersePrintTupleFieldsToStringsTest, PrintsEmptyTuple) {
+ Strings result = UniversalTersePrintTupleFieldsToStrings(make_tuple());
+ EXPECT_EQ(0u, result.size());
+}
+
+TEST(UniversalTersePrintTupleFieldsToStringsTest, PrintsOneTuple) {
+ Strings result = UniversalTersePrintTupleFieldsToStrings(make_tuple(1));
+ ASSERT_EQ(1u, result.size());
+ EXPECT_EQ("1", result[0]);
+}
+
+TEST(UniversalTersePrintTupleFieldsToStringsTest, PrintsTwoTuple) {
+ Strings result = UniversalTersePrintTupleFieldsToStrings(make_tuple(1, 'a'));
+ ASSERT_EQ(2u, result.size());
+ EXPECT_EQ("1", result[0]);
+ EXPECT_EQ("'a' (97, 0x61)", result[1]);
+}
+
+TEST(UniversalTersePrintTupleFieldsToStringsTest, PrintsTersely) {
+ const int n = 1;
+ Strings result = UniversalTersePrintTupleFieldsToStrings(
+ tuple<const int&, const char*>(n, "a"));
+ ASSERT_EQ(2u, result.size());
+ EXPECT_EQ("1", result[0]);
+ EXPECT_EQ("\"a\"", result[1]);
+}
+
+#endif // GTEST_HAS_TR1_TUPLE
+
+} // namespace gtest_printers_test
+} // namespace testing
diff --git a/libvpx/third_party/googletest/src/test/gtest-test-part_test.cc b/libvpx/third_party/googletest/src/test/gtest-test-part_test.cc
new file mode 100644
index 0000000..ca8ba93
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-test-part_test.cc
@@ -0,0 +1,208 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+
+#include "gtest/gtest-test-part.h"
+
+#include "gtest/gtest.h"
+
+using testing::Message;
+using testing::Test;
+using testing::TestPartResult;
+using testing::TestPartResultArray;
+
+namespace {
+
+// Tests the TestPartResult class.
+
+// The test fixture for testing TestPartResult.
+class TestPartResultTest : public Test {
+ protected:
+ TestPartResultTest()
+ : r1_(TestPartResult::kSuccess, "foo/bar.cc", 10, "Success!"),
+ r2_(TestPartResult::kNonFatalFailure, "foo/bar.cc", -1, "Failure!"),
+ r3_(TestPartResult::kFatalFailure, NULL, -1, "Failure!") {}
+
+ TestPartResult r1_, r2_, r3_;
+};
+
+
+TEST_F(TestPartResultTest, ConstructorWorks) {
+ Message message;
+ message << "something is terribly wrong";
+ message << static_cast<const char*>(testing::internal::kStackTraceMarker);
+ message << "some unimportant stack trace";
+
+ const TestPartResult result(TestPartResult::kNonFatalFailure,
+ "some_file.cc",
+ 42,
+ message.GetString().c_str());
+
+ EXPECT_EQ(TestPartResult::kNonFatalFailure, result.type());
+ EXPECT_STREQ("some_file.cc", result.file_name());
+ EXPECT_EQ(42, result.line_number());
+ EXPECT_STREQ(message.GetString().c_str(), result.message());
+ EXPECT_STREQ("something is terribly wrong", result.summary());
+}
+
+TEST_F(TestPartResultTest, ResultAccessorsWork) {
+ const TestPartResult success(TestPartResult::kSuccess,
+ "file.cc",
+ 42,
+ "message");
+ EXPECT_TRUE(success.passed());
+ EXPECT_FALSE(success.failed());
+ EXPECT_FALSE(success.nonfatally_failed());
+ EXPECT_FALSE(success.fatally_failed());
+
+ const TestPartResult nonfatal_failure(TestPartResult::kNonFatalFailure,
+ "file.cc",
+ 42,
+ "message");
+ EXPECT_FALSE(nonfatal_failure.passed());
+ EXPECT_TRUE(nonfatal_failure.failed());
+ EXPECT_TRUE(nonfatal_failure.nonfatally_failed());
+ EXPECT_FALSE(nonfatal_failure.fatally_failed());
+
+ const TestPartResult fatal_failure(TestPartResult::kFatalFailure,
+ "file.cc",
+ 42,
+ "message");
+ EXPECT_FALSE(fatal_failure.passed());
+ EXPECT_TRUE(fatal_failure.failed());
+ EXPECT_FALSE(fatal_failure.nonfatally_failed());
+ EXPECT_TRUE(fatal_failure.fatally_failed());
+}
+
+// Tests TestPartResult::type().
+TEST_F(TestPartResultTest, type) {
+ EXPECT_EQ(TestPartResult::kSuccess, r1_.type());
+ EXPECT_EQ(TestPartResult::kNonFatalFailure, r2_.type());
+ EXPECT_EQ(TestPartResult::kFatalFailure, r3_.type());
+}
+
+// Tests TestPartResult::file_name().
+TEST_F(TestPartResultTest, file_name) {
+ EXPECT_STREQ("foo/bar.cc", r1_.file_name());
+ EXPECT_STREQ(NULL, r3_.file_name());
+}
+
+// Tests TestPartResult::line_number().
+TEST_F(TestPartResultTest, line_number) {
+ EXPECT_EQ(10, r1_.line_number());
+ EXPECT_EQ(-1, r2_.line_number());
+}
+
+// Tests TestPartResult::message().
+TEST_F(TestPartResultTest, message) {
+ EXPECT_STREQ("Success!", r1_.message());
+}
+
+// Tests TestPartResult::passed().
+TEST_F(TestPartResultTest, Passed) {
+ EXPECT_TRUE(r1_.passed());
+ EXPECT_FALSE(r2_.passed());
+ EXPECT_FALSE(r3_.passed());
+}
+
+// Tests TestPartResult::failed().
+TEST_F(TestPartResultTest, Failed) {
+ EXPECT_FALSE(r1_.failed());
+ EXPECT_TRUE(r2_.failed());
+ EXPECT_TRUE(r3_.failed());
+}
+
+// Tests TestPartResult::fatally_failed().
+TEST_F(TestPartResultTest, FatallyFailed) {
+ EXPECT_FALSE(r1_.fatally_failed());
+ EXPECT_FALSE(r2_.fatally_failed());
+ EXPECT_TRUE(r3_.fatally_failed());
+}
+
+// Tests TestPartResult::nonfatally_failed().
+TEST_F(TestPartResultTest, NonfatallyFailed) {
+ EXPECT_FALSE(r1_.nonfatally_failed());
+ EXPECT_TRUE(r2_.nonfatally_failed());
+ EXPECT_FALSE(r3_.nonfatally_failed());
+}
+
+// Tests the TestPartResultArray class.
+
+class TestPartResultArrayTest : public Test {
+ protected:
+ TestPartResultArrayTest()
+ : r1_(TestPartResult::kNonFatalFailure, "foo/bar.cc", -1, "Failure 1"),
+ r2_(TestPartResult::kFatalFailure, "foo/bar.cc", -1, "Failure 2") {}
+
+ const TestPartResult r1_, r2_;
+};
+
+// Tests that TestPartResultArray initially has size 0.
+TEST_F(TestPartResultArrayTest, InitialSizeIsZero) {
+ TestPartResultArray results;
+ EXPECT_EQ(0, results.size());
+}
+
+// Tests that TestPartResultArray contains the given TestPartResult
+// after one Append() operation.
+TEST_F(TestPartResultArrayTest, ContainsGivenResultAfterAppend) {
+ TestPartResultArray results;
+ results.Append(r1_);
+ EXPECT_EQ(1, results.size());
+ EXPECT_STREQ("Failure 1", results.GetTestPartResult(0).message());
+}
+
+// Tests that TestPartResultArray contains the given TestPartResults
+// after two Append() operations.
+TEST_F(TestPartResultArrayTest, ContainsGivenResultsAfterTwoAppends) {
+ TestPartResultArray results;
+ results.Append(r1_);
+ results.Append(r2_);
+ EXPECT_EQ(2, results.size());
+ EXPECT_STREQ("Failure 1", results.GetTestPartResult(0).message());
+ EXPECT_STREQ("Failure 2", results.GetTestPartResult(1).message());
+}
+
+typedef TestPartResultArrayTest TestPartResultArrayDeathTest;
+
+// Tests that the program dies when GetTestPartResult() is called with
+// an invalid index.
+TEST_F(TestPartResultArrayDeathTest, DiesWhenIndexIsOutOfBound) {
+ TestPartResultArray results;
+ results.Append(r1_);
+
+ EXPECT_DEATH_IF_SUPPORTED(results.GetTestPartResult(-1), "");
+ EXPECT_DEATH_IF_SUPPORTED(results.GetTestPartResult(1), "");
+}
+
+// TODO(mheule@google.com): Add a test for the class HasNewFatalFailureHelper.
+
+} // namespace
diff --git a/libvpx/third_party/googletest/src/test/gtest-tuple_test.cc b/libvpx/third_party/googletest/src/test/gtest-tuple_test.cc
new file mode 100644
index 0000000..bfaa3e0
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-tuple_test.cc
@@ -0,0 +1,320 @@
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "gtest/internal/gtest-tuple.h"
+#include <utility>
+#include "gtest/gtest.h"
+
+namespace {
+
+using ::std::tr1::get;
+using ::std::tr1::make_tuple;
+using ::std::tr1::tuple;
+using ::std::tr1::tuple_element;
+using ::std::tr1::tuple_size;
+using ::testing::StaticAssertTypeEq;
+
+// Tests that tuple_element<K, tuple<T0, T1, ..., TN> >::type returns TK.
+TEST(tuple_element_Test, ReturnsElementType) {
+ StaticAssertTypeEq<int, tuple_element<0, tuple<int, char> >::type>();
+ StaticAssertTypeEq<int&, tuple_element<1, tuple<double, int&> >::type>();
+ StaticAssertTypeEq<bool, tuple_element<2, tuple<double, int, bool> >::type>();
+}
+
+// Tests that tuple_size<T>::value gives the number of fields in tuple
+// type T.
+TEST(tuple_size_Test, ReturnsNumberOfFields) {
+ EXPECT_EQ(0, +tuple_size<tuple<> >::value);
+ EXPECT_EQ(1, +tuple_size<tuple<void*> >::value);
+ EXPECT_EQ(1, +tuple_size<tuple<char> >::value);
+ EXPECT_EQ(1, +(tuple_size<tuple<tuple<int, double> > >::value));
+ EXPECT_EQ(2, +(tuple_size<tuple<int&, const char> >::value));
+ EXPECT_EQ(3, +(tuple_size<tuple<char*, void, const bool&> >::value));
+}
+
+// Tests comparing a tuple with itself.
+TEST(ComparisonTest, ComparesWithSelf) {
+ const tuple<int, char, bool> a(5, 'a', false);
+
+ EXPECT_TRUE(a == a);
+ EXPECT_FALSE(a != a);
+}
+
+// Tests comparing two tuples with the same value.
+TEST(ComparisonTest, ComparesEqualTuples) {
+ const tuple<int, bool> a(5, true), b(5, true);
+
+ EXPECT_TRUE(a == b);
+ EXPECT_FALSE(a != b);
+}
+
+// Tests comparing two different tuples that have no reference fields.
+TEST(ComparisonTest, ComparesUnequalTuplesWithoutReferenceFields) {
+ typedef tuple<const int, char> FooTuple;
+
+ const FooTuple a(0, 'x');
+ const FooTuple b(1, 'a');
+
+ EXPECT_TRUE(a != b);
+ EXPECT_FALSE(a == b);
+
+ const FooTuple c(1, 'b');
+
+ EXPECT_TRUE(b != c);
+ EXPECT_FALSE(b == c);
+}
+
+// Tests comparing two different tuples that have reference fields.
+TEST(ComparisonTest, ComparesUnequalTuplesWithReferenceFields) {
+ typedef tuple<int&, const char&> FooTuple;
+
+ int i = 5;
+ const char ch = 'a';
+ const FooTuple a(i, ch);
+
+ int j = 6;
+ const FooTuple b(j, ch);
+
+ EXPECT_TRUE(a != b);
+ EXPECT_FALSE(a == b);
+
+ j = 5;
+ const char ch2 = 'b';
+ const FooTuple c(j, ch2);
+
+ EXPECT_TRUE(b != c);
+ EXPECT_FALSE(b == c);
+}
+
+// Tests that a tuple field with a reference type is an alias of the
+// variable it's supposed to reference.
+TEST(ReferenceFieldTest, IsAliasOfReferencedVariable) {
+ int n = 0;
+ tuple<bool, int&> t(true, n);
+
+ n = 1;
+ EXPECT_EQ(n, get<1>(t))
+ << "Changing a underlying variable should update the reference field.";
+
+ // Makes sure that the implementation doesn't do anything funny with
+ // the & operator for the return type of get<>().
+ EXPECT_EQ(&n, &(get<1>(t)))
+ << "The address of a reference field should equal the address of "
+ << "the underlying variable.";
+
+ get<1>(t) = 2;
+ EXPECT_EQ(2, n)
+ << "Changing a reference field should update the underlying variable.";
+}
+
+// Tests that tuple's default constructor default initializes each field.
+// This test needs to compile without generating warnings.
+TEST(TupleConstructorTest, DefaultConstructorDefaultInitializesEachField) {
+ // The TR1 report requires that tuple's default constructor default
+ // initializes each field, even if it's a primitive type. If the
+ // implementation forgets to do this, this test will catch it by
+ // generating warnings about using uninitialized variables (assuming
+ // a decent compiler).
+
+ tuple<> empty;
+
+ tuple<int> a1, b1;
+ b1 = a1;
+ EXPECT_EQ(0, get<0>(b1));
+
+ tuple<int, double> a2, b2;
+ b2 = a2;
+ EXPECT_EQ(0, get<0>(b2));
+ EXPECT_EQ(0.0, get<1>(b2));
+
+ tuple<double, char, bool*> a3, b3;
+ b3 = a3;
+ EXPECT_EQ(0.0, get<0>(b3));
+ EXPECT_EQ('\0', get<1>(b3));
+ EXPECT_TRUE(get<2>(b3) == NULL);
+
+ tuple<int, int, int, int, int, int, int, int, int, int> a10, b10;
+ b10 = a10;
+ EXPECT_EQ(0, get<0>(b10));
+ EXPECT_EQ(0, get<1>(b10));
+ EXPECT_EQ(0, get<2>(b10));
+ EXPECT_EQ(0, get<3>(b10));
+ EXPECT_EQ(0, get<4>(b10));
+ EXPECT_EQ(0, get<5>(b10));
+ EXPECT_EQ(0, get<6>(b10));
+ EXPECT_EQ(0, get<7>(b10));
+ EXPECT_EQ(0, get<8>(b10));
+ EXPECT_EQ(0, get<9>(b10));
+}
+
+// Tests constructing a tuple from its fields.
+TEST(TupleConstructorTest, ConstructsFromFields) {
+ int n = 1;
+ // Reference field.
+ tuple<int&> a(n);
+ EXPECT_EQ(&n, &(get<0>(a)));
+
+ // Non-reference fields.
+ tuple<int, char> b(5, 'a');
+ EXPECT_EQ(5, get<0>(b));
+ EXPECT_EQ('a', get<1>(b));
+
+ // Const reference field.
+ const int m = 2;
+ tuple<bool, const int&> c(true, m);
+ EXPECT_TRUE(get<0>(c));
+ EXPECT_EQ(&m, &(get<1>(c)));
+}
+
+// Tests tuple's copy constructor.
+TEST(TupleConstructorTest, CopyConstructor) {
+ tuple<double, bool> a(0.0, true);
+ tuple<double, bool> b(a);
+
+ EXPECT_DOUBLE_EQ(0.0, get<0>(b));
+ EXPECT_TRUE(get<1>(b));
+}
+
+// Tests constructing a tuple from another tuple that has a compatible
+// but different type.
+TEST(TupleConstructorTest, ConstructsFromDifferentTupleType) {
+ tuple<int, int, char> a(0, 1, 'a');
+ tuple<double, long, int> b(a);
+
+ EXPECT_DOUBLE_EQ(0.0, get<0>(b));
+ EXPECT_EQ(1, get<1>(b));
+ EXPECT_EQ('a', get<2>(b));
+}
+
+// Tests constructing a 2-tuple from an std::pair.
+TEST(TupleConstructorTest, ConstructsFromPair) {
+ ::std::pair<int, char> a(1, 'a');
+ tuple<int, char> b(a);
+ tuple<int, const char&> c(a);
+}
+
+// Tests assigning a tuple to another tuple with the same type.
+TEST(TupleAssignmentTest, AssignsToSameTupleType) {
+ const tuple<int, long> a(5, 7L);
+ tuple<int, long> b;
+ b = a;
+ EXPECT_EQ(5, get<0>(b));
+ EXPECT_EQ(7L, get<1>(b));
+}
+
+// Tests assigning a tuple to another tuple with a different but
+// compatible type.
+TEST(TupleAssignmentTest, AssignsToDifferentTupleType) {
+ const tuple<int, long, bool> a(1, 7L, true);
+ tuple<long, int, bool> b;
+ b = a;
+ EXPECT_EQ(1L, get<0>(b));
+ EXPECT_EQ(7, get<1>(b));
+ EXPECT_TRUE(get<2>(b));
+}
+
+// Tests assigning an std::pair to a 2-tuple.
+TEST(TupleAssignmentTest, AssignsFromPair) {
+ const ::std::pair<int, bool> a(5, true);
+ tuple<int, bool> b;
+ b = a;
+ EXPECT_EQ(5, get<0>(b));
+ EXPECT_TRUE(get<1>(b));
+
+ tuple<long, bool> c;
+ c = a;
+ EXPECT_EQ(5L, get<0>(c));
+ EXPECT_TRUE(get<1>(c));
+}
+
+// A fixture for testing big tuples.
+class BigTupleTest : public testing::Test {
+ protected:
+ typedef tuple<int, int, int, int, int, int, int, int, int, int> BigTuple;
+
+ BigTupleTest() :
+ a_(1, 0, 0, 0, 0, 0, 0, 0, 0, 2),
+ b_(1, 0, 0, 0, 0, 0, 0, 0, 0, 3) {}
+
+ BigTuple a_, b_;
+};
+
+// Tests constructing big tuples.
+TEST_F(BigTupleTest, Construction) {
+ BigTuple a;
+ BigTuple b(b_);
+}
+
+// Tests that get<N>(t) returns the N-th (0-based) field of tuple t.
+TEST_F(BigTupleTest, get) {
+ EXPECT_EQ(1, get<0>(a_));
+ EXPECT_EQ(2, get<9>(a_));
+
+ // Tests that get() works on a const tuple too.
+ const BigTuple a(a_);
+ EXPECT_EQ(1, get<0>(a));
+ EXPECT_EQ(2, get<9>(a));
+}
+
+// Tests comparing big tuples.
+TEST_F(BigTupleTest, Comparisons) {
+ EXPECT_TRUE(a_ == a_);
+ EXPECT_FALSE(a_ != a_);
+
+ EXPECT_TRUE(a_ != b_);
+ EXPECT_FALSE(a_ == b_);
+}
+
+TEST(MakeTupleTest, WorksForScalarTypes) {
+ tuple<bool, int> a;
+ a = make_tuple(true, 5);
+ EXPECT_TRUE(get<0>(a));
+ EXPECT_EQ(5, get<1>(a));
+
+ tuple<char, int, long> b;
+ b = make_tuple('a', 'b', 5);
+ EXPECT_EQ('a', get<0>(b));
+ EXPECT_EQ('b', get<1>(b));
+ EXPECT_EQ(5, get<2>(b));
+}
+
+TEST(MakeTupleTest, WorksForPointers) {
+ int a[] = { 1, 2, 3, 4 };
+ const char* const str = "hi";
+ int* const p = a;
+
+ tuple<const char*, int*> t;
+ t = make_tuple(str, p);
+ EXPECT_EQ(str, get<0>(t));
+ EXPECT_EQ(p, get<1>(t));
+}
+
+} // namespace
diff --git a/libvpx/third_party/googletest/src/test/gtest-typed-test2_test.cc b/libvpx/third_party/googletest/src/test/gtest-typed-test2_test.cc
new file mode 100644
index 0000000..c284700
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-typed-test2_test.cc
@@ -0,0 +1,45 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include <vector>
+
+#include "test/gtest-typed-test_test.h"
+#include "gtest/gtest.h"
+
+#if GTEST_HAS_TYPED_TEST_P
+
+// Tests that the same type-parameterized test case can be
+// instantiated in different translation units linked together.
+// (ContainerTest is also instantiated in gtest-typed-test_test.cc.)
+INSTANTIATE_TYPED_TEST_CASE_P(Vector, ContainerTest,
+ testing::Types<std::vector<int> >);
+
+#endif // GTEST_HAS_TYPED_TEST_P
diff --git a/libvpx/third_party/googletest/src/test/gtest-typed-test_test.cc b/libvpx/third_party/googletest/src/test/gtest-typed-test_test.cc
new file mode 100644
index 0000000..dd4ba43
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-typed-test_test.cc
@@ -0,0 +1,360 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include <set>
+#include <vector>
+
+#include "test/gtest-typed-test_test.h"
+#include "gtest/gtest.h"
+
+using testing::Test;
+
+// Used for testing that SetUpTestCase()/TearDownTestCase(), fixture
+// ctor/dtor, and SetUp()/TearDown() work correctly in typed tests and
+// type-parameterized test.
+template <typename T>
+class CommonTest : public Test {
+ // For some technical reason, SetUpTestCase() and TearDownTestCase()
+ // must be public.
+ public:
+ static void SetUpTestCase() {
+ shared_ = new T(5);
+ }
+
+ static void TearDownTestCase() {
+ delete shared_;
+ shared_ = NULL;
+ }
+
+ // This 'protected:' is optional. There's no harm in making all
+ // members of this fixture class template public.
+ protected:
+ // We used to use std::list here, but switched to std::vector since
+ // MSVC's <list> doesn't compile cleanly with /W4.
+ typedef std::vector<T> Vector;
+ typedef std::set<int> IntSet;
+
+ CommonTest() : value_(1) {}
+
+ virtual ~CommonTest() { EXPECT_EQ(3, value_); }
+
+ virtual void SetUp() {
+ EXPECT_EQ(1, value_);
+ value_++;
+ }
+
+ virtual void TearDown() {
+ EXPECT_EQ(2, value_);
+ value_++;
+ }
+
+ T value_;
+ static T* shared_;
+};
+
+template <typename T>
+T* CommonTest<T>::shared_ = NULL;
+
+// This #ifdef block tests typed tests.
+#if GTEST_HAS_TYPED_TEST
+
+using testing::Types;
+
+// Tests that SetUpTestCase()/TearDownTestCase(), fixture ctor/dtor,
+// and SetUp()/TearDown() work correctly in typed tests
+
+typedef Types<char, int> TwoTypes;
+TYPED_TEST_CASE(CommonTest, TwoTypes);
+
+TYPED_TEST(CommonTest, ValuesAreCorrect) {
+ // Static members of the fixture class template can be visited via
+ // the TestFixture:: prefix.
+ EXPECT_EQ(5, *TestFixture::shared_);
+
+ // Typedefs in the fixture class template can be visited via the
+ // "typename TestFixture::" prefix.
+ typename TestFixture::Vector empty;
+ EXPECT_EQ(0U, empty.size());
+
+ typename TestFixture::IntSet empty2;
+ EXPECT_EQ(0U, empty2.size());
+
+ // Non-static members of the fixture class must be visited via
+ // 'this', as required by C++ for class templates.
+ EXPECT_EQ(2, this->value_);
+}
+
+// The second test makes sure shared_ is not deleted after the first
+// test.
+TYPED_TEST(CommonTest, ValuesAreStillCorrect) {
+ // Static members of the fixture class template can also be visited
+ // via 'this'.
+ ASSERT_TRUE(this->shared_ != NULL);
+ EXPECT_EQ(5, *this->shared_);
+
+ // TypeParam can be used to refer to the type parameter.
+ EXPECT_EQ(static_cast<TypeParam>(2), this->value_);
+}
+
+// Tests that multiple TYPED_TEST_CASE's can be defined in the same
+// translation unit.
+
+template <typename T>
+class TypedTest1 : public Test {
+};
+
+// Verifies that the second argument of TYPED_TEST_CASE can be a
+// single type.
+TYPED_TEST_CASE(TypedTest1, int);
+TYPED_TEST(TypedTest1, A) {}
+
+template <typename T>
+class TypedTest2 : public Test {
+};
+
+// Verifies that the second argument of TYPED_TEST_CASE can be a
+// Types<...> type list.
+TYPED_TEST_CASE(TypedTest2, Types<int>);
+
+// This also verifies that tests from different typed test cases can
+// share the same name.
+TYPED_TEST(TypedTest2, A) {}
+
+// Tests that a typed test case can be defined in a namespace.
+
+namespace library1 {
+
+template <typename T>
+class NumericTest : public Test {
+};
+
+typedef Types<int, long> NumericTypes;
+TYPED_TEST_CASE(NumericTest, NumericTypes);
+
+TYPED_TEST(NumericTest, DefaultIsZero) {
+ EXPECT_EQ(0, TypeParam());
+}
+
+} // namespace library1
+
+#endif // GTEST_HAS_TYPED_TEST
+
+// This #ifdef block tests type-parameterized tests.
+#if GTEST_HAS_TYPED_TEST_P
+
+using testing::Types;
+using testing::internal::TypedTestCasePState;
+
+// Tests TypedTestCasePState.
+
+class TypedTestCasePStateTest : public Test {
+ protected:
+ virtual void SetUp() {
+ state_.AddTestName("foo.cc", 0, "FooTest", "A");
+ state_.AddTestName("foo.cc", 0, "FooTest", "B");
+ state_.AddTestName("foo.cc", 0, "FooTest", "C");
+ }
+
+ TypedTestCasePState state_;
+};
+
+TEST_F(TypedTestCasePStateTest, SucceedsForMatchingList) {
+ const char* tests = "A, B, C";
+ EXPECT_EQ(tests,
+ state_.VerifyRegisteredTestNames("foo.cc", 1, tests));
+}
+
+// Makes sure that the order of the tests and spaces around the names
+// don't matter.
+TEST_F(TypedTestCasePStateTest, IgnoresOrderAndSpaces) {
+ const char* tests = "A,C, B";
+ EXPECT_EQ(tests,
+ state_.VerifyRegisteredTestNames("foo.cc", 1, tests));
+}
+
+typedef TypedTestCasePStateTest TypedTestCasePStateDeathTest;
+
+TEST_F(TypedTestCasePStateDeathTest, DetectsDuplicates) {
+ EXPECT_DEATH_IF_SUPPORTED(
+ state_.VerifyRegisteredTestNames("foo.cc", 1, "A, B, A, C"),
+ "foo\\.cc.1.?: Test A is listed more than once\\.");
+}
+
+TEST_F(TypedTestCasePStateDeathTest, DetectsExtraTest) {
+ EXPECT_DEATH_IF_SUPPORTED(
+ state_.VerifyRegisteredTestNames("foo.cc", 1, "A, B, C, D"),
+ "foo\\.cc.1.?: No test named D can be found in this test case\\.");
+}
+
+TEST_F(TypedTestCasePStateDeathTest, DetectsMissedTest) {
+ EXPECT_DEATH_IF_SUPPORTED(
+ state_.VerifyRegisteredTestNames("foo.cc", 1, "A, C"),
+ "foo\\.cc.1.?: You forgot to list test B\\.");
+}
+
+// Tests that defining a test for a parameterized test case generates
+// a run-time error if the test case has been registered.
+TEST_F(TypedTestCasePStateDeathTest, DetectsTestAfterRegistration) {
+ state_.VerifyRegisteredTestNames("foo.cc", 1, "A, B, C");
+ EXPECT_DEATH_IF_SUPPORTED(
+ state_.AddTestName("foo.cc", 2, "FooTest", "D"),
+ "foo\\.cc.2.?: Test D must be defined before REGISTER_TYPED_TEST_CASE_P"
+ "\\(FooTest, \\.\\.\\.\\)\\.");
+}
+
+// Tests that SetUpTestCase()/TearDownTestCase(), fixture ctor/dtor,
+// and SetUp()/TearDown() work correctly in type-parameterized tests.
+
+template <typename T>
+class DerivedTest : public CommonTest<T> {
+};
+
+TYPED_TEST_CASE_P(DerivedTest);
+
+TYPED_TEST_P(DerivedTest, ValuesAreCorrect) {
+ // Static members of the fixture class template can be visited via
+ // the TestFixture:: prefix.
+ EXPECT_EQ(5, *TestFixture::shared_);
+
+ // Non-static members of the fixture class must be visited via
+ // 'this', as required by C++ for class templates.
+ EXPECT_EQ(2, this->value_);
+}
+
+// The second test makes sure shared_ is not deleted after the first
+// test.
+TYPED_TEST_P(DerivedTest, ValuesAreStillCorrect) {
+ // Static members of the fixture class template can also be visited
+ // via 'this'.
+ ASSERT_TRUE(this->shared_ != NULL);
+ EXPECT_EQ(5, *this->shared_);
+ EXPECT_EQ(2, this->value_);
+}
+
+REGISTER_TYPED_TEST_CASE_P(DerivedTest,
+ ValuesAreCorrect, ValuesAreStillCorrect);
+
+typedef Types<short, long> MyTwoTypes;
+INSTANTIATE_TYPED_TEST_CASE_P(My, DerivedTest, MyTwoTypes);
+
+// Tests that multiple TYPED_TEST_CASE_P's can be defined in the same
+// translation unit.
+
+template <typename T>
+class TypedTestP1 : public Test {
+};
+
+TYPED_TEST_CASE_P(TypedTestP1);
+
+// For testing that the code between TYPED_TEST_CASE_P() and
+// TYPED_TEST_P() is not enclosed in a namespace.
+typedef int IntAfterTypedTestCaseP;
+
+TYPED_TEST_P(TypedTestP1, A) {}
+TYPED_TEST_P(TypedTestP1, B) {}
+
+// For testing that the code between TYPED_TEST_P() and
+// REGISTER_TYPED_TEST_CASE_P() is not enclosed in a namespace.
+typedef int IntBeforeRegisterTypedTestCaseP;
+
+REGISTER_TYPED_TEST_CASE_P(TypedTestP1, A, B);
+
+template <typename T>
+class TypedTestP2 : public Test {
+};
+
+TYPED_TEST_CASE_P(TypedTestP2);
+
+// This also verifies that tests from different type-parameterized
+// test cases can share the same name.
+TYPED_TEST_P(TypedTestP2, A) {}
+
+REGISTER_TYPED_TEST_CASE_P(TypedTestP2, A);
+
+// Verifies that the code between TYPED_TEST_CASE_P() and
+// REGISTER_TYPED_TEST_CASE_P() is not enclosed in a namespace.
+IntAfterTypedTestCaseP after = 0;
+IntBeforeRegisterTypedTestCaseP before = 0;
+
+// Verifies that the last argument of INSTANTIATE_TYPED_TEST_CASE_P()
+// can be either a single type or a Types<...> type list.
+INSTANTIATE_TYPED_TEST_CASE_P(Int, TypedTestP1, int);
+INSTANTIATE_TYPED_TEST_CASE_P(Int, TypedTestP2, Types<int>);
+
+// Tests that the same type-parameterized test case can be
+// instantiated more than once in the same translation unit.
+INSTANTIATE_TYPED_TEST_CASE_P(Double, TypedTestP2, Types<double>);
+
+// Tests that the same type-parameterized test case can be
+// instantiated in different translation units linked together.
+// (ContainerTest is also instantiated in gtest-typed-test_test.cc.)
+typedef Types<std::vector<double>, std::set<char> > MyContainers;
+INSTANTIATE_TYPED_TEST_CASE_P(My, ContainerTest, MyContainers);
+
+// Tests that a type-parameterized test case can be defined and
+// instantiated in a namespace.
+
+namespace library2 {
+
+template <typename T>
+class NumericTest : public Test {
+};
+
+TYPED_TEST_CASE_P(NumericTest);
+
+TYPED_TEST_P(NumericTest, DefaultIsZero) {
+ EXPECT_EQ(0, TypeParam());
+}
+
+TYPED_TEST_P(NumericTest, ZeroIsLessThanOne) {
+ EXPECT_LT(TypeParam(0), TypeParam(1));
+}
+
+REGISTER_TYPED_TEST_CASE_P(NumericTest,
+ DefaultIsZero, ZeroIsLessThanOne);
+typedef Types<int, double> NumericTypes;
+INSTANTIATE_TYPED_TEST_CASE_P(My, NumericTest, NumericTypes);
+
+} // namespace library2
+
+#endif // GTEST_HAS_TYPED_TEST_P
+
+#if !defined(GTEST_HAS_TYPED_TEST) && !defined(GTEST_HAS_TYPED_TEST_P)
+
+// Google Test may not support type-parameterized tests with some
+// compilers. If we use conditional compilation to compile out all
+// code referring to the gtest_main library, MSVC linker will not link
+// that library at all and consequently complain about missing entry
+// point defined in that library (fatal error LNK1561: entry point
+// must be defined). This dummy test keeps gtest_main linked in.
+TEST(DummyTest, TypedTestsAreNotSupportedOnThisPlatform) {}
+
+#endif // #if !defined(GTEST_HAS_TYPED_TEST) && !defined(GTEST_HAS_TYPED_TEST_P)
diff --git a/libvpx/third_party/googletest/src/test/gtest-typed-test_test.h b/libvpx/third_party/googletest/src/test/gtest-typed-test_test.h
new file mode 100644
index 0000000..41d7570
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-typed-test_test.h
@@ -0,0 +1,66 @@
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#ifndef GTEST_TEST_GTEST_TYPED_TEST_TEST_H_
+#define GTEST_TEST_GTEST_TYPED_TEST_TEST_H_
+
+#include "gtest/gtest.h"
+
+#if GTEST_HAS_TYPED_TEST_P
+
+using testing::Test;
+
+// For testing that the same type-parameterized test case can be
+// instantiated in different translation units linked together.
+// ContainerTest will be instantiated in both gtest-typed-test_test.cc
+// and gtest-typed-test2_test.cc.
+
+template <typename T>
+class ContainerTest : public Test {
+};
+
+TYPED_TEST_CASE_P(ContainerTest);
+
+TYPED_TEST_P(ContainerTest, CanBeDefaultConstructed) {
+ TypeParam container;
+}
+
+TYPED_TEST_P(ContainerTest, InitialSizeIsZero) {
+ TypeParam container;
+ EXPECT_EQ(0U, container.size());
+}
+
+REGISTER_TYPED_TEST_CASE_P(ContainerTest,
+ CanBeDefaultConstructed, InitialSizeIsZero);
+
+#endif // GTEST_HAS_TYPED_TEST_P
+
+#endif // GTEST_TEST_GTEST_TYPED_TEST_TEST_H_
diff --git a/libvpx/third_party/googletest/src/test/gtest-unittest-api_test.cc b/libvpx/third_party/googletest/src/test/gtest-unittest-api_test.cc
new file mode 100644
index 0000000..07083e5
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest-unittest-api_test.cc
@@ -0,0 +1,341 @@
+// Copyright 2009 Google Inc. All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This file contains tests verifying correctness of data provided via
+// UnitTest's public methods.
+
+#include "gtest/gtest.h"
+
+#include <string.h> // For strcmp.
+#include <algorithm>
+
+using ::testing::InitGoogleTest;
+
+namespace testing {
+namespace internal {
+
+template <typename T>
+struct LessByName {
+ bool operator()(const T* a, const T* b) {
+ return strcmp(a->name(), b->name()) < 0;
+ }
+};
+
+class UnitTestHelper {
+ public:
+ // Returns the array of pointers to all test cases sorted by the test case
+ // name. The caller is responsible for deleting the array.
+ static TestCase const** const GetSortedTestCases() {
+ UnitTest& unit_test = *UnitTest::GetInstance();
+ TestCase const** const test_cases =
+ new const TestCase*[unit_test.total_test_case_count()];
+
+ for (int i = 0; i < unit_test.total_test_case_count(); ++i)
+ test_cases[i] = unit_test.GetTestCase(i);
+
+ std::sort(test_cases,
+ test_cases + unit_test.total_test_case_count(),
+ LessByName<TestCase>());
+ return test_cases;
+ }
+
+ // Returns the test case by its name. The caller doesn't own the returned
+ // pointer.
+ static const TestCase* FindTestCase(const char* name) {
+ UnitTest& unit_test = *UnitTest::GetInstance();
+ for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+ const TestCase* test_case = unit_test.GetTestCase(i);
+ if (0 == strcmp(test_case->name(), name))
+ return test_case;
+ }
+ return NULL;
+ }
+
+ // Returns the array of pointers to all tests in a particular test case
+ // sorted by the test name. The caller is responsible for deleting the
+ // array.
+ static TestInfo const** const GetSortedTests(const TestCase* test_case) {
+ TestInfo const** const tests =
+ new const TestInfo*[test_case->total_test_count()];
+
+ for (int i = 0; i < test_case->total_test_count(); ++i)
+ tests[i] = test_case->GetTestInfo(i);
+
+ std::sort(tests, tests + test_case->total_test_count(),
+ LessByName<TestInfo>());
+ return tests;
+ }
+};
+
+#if GTEST_HAS_TYPED_TEST
+template <typename T> class TestCaseWithCommentTest : public Test {};
+TYPED_TEST_CASE(TestCaseWithCommentTest, Types<int>);
+TYPED_TEST(TestCaseWithCommentTest, Dummy) {}
+
+const int kTypedTestCases = 1;
+const int kTypedTests = 1;
+#else
+const int kTypedTestCases = 0;
+const int kTypedTests = 0;
+#endif // GTEST_HAS_TYPED_TEST
+
+// We can only test the accessors that do not change value while tests run.
+// Since tests can be run in any order, the values the accessors that track
+// test execution (such as failed_test_count) can not be predicted.
+TEST(ApiTest, UnitTestImmutableAccessorsWork) {
+ UnitTest* unit_test = UnitTest::GetInstance();
+
+ ASSERT_EQ(2 + kTypedTestCases, unit_test->total_test_case_count());
+ EXPECT_EQ(1 + kTypedTestCases, unit_test->test_case_to_run_count());
+ EXPECT_EQ(2, unit_test->disabled_test_count());
+ EXPECT_EQ(5 + kTypedTests, unit_test->total_test_count());
+ EXPECT_EQ(3 + kTypedTests, unit_test->test_to_run_count());
+
+ const TestCase** const test_cases = UnitTestHelper::GetSortedTestCases();
+
+ EXPECT_STREQ("ApiTest", test_cases[0]->name());
+ EXPECT_STREQ("DISABLED_Test", test_cases[1]->name());
+#if GTEST_HAS_TYPED_TEST
+ EXPECT_STREQ("TestCaseWithCommentTest/0", test_cases[2]->name());
+#endif // GTEST_HAS_TYPED_TEST
+
+ delete[] test_cases;
+
+ // The following lines initiate actions to verify certain methods in
+ // FinalSuccessChecker::TearDown.
+
+ // Records a test property to verify TestResult::GetTestProperty().
+ RecordProperty("key", "value");
+}
+
+AssertionResult IsNull(const char* str) {
+ if (str != NULL) {
+ return testing::AssertionFailure() << "argument is " << str;
+ }
+ return AssertionSuccess();
+}
+
+TEST(ApiTest, TestCaseImmutableAccessorsWork) {
+ const TestCase* test_case = UnitTestHelper::FindTestCase("ApiTest");
+ ASSERT_TRUE(test_case != NULL);
+
+ EXPECT_STREQ("ApiTest", test_case->name());
+ EXPECT_TRUE(IsNull(test_case->type_param()));
+ EXPECT_TRUE(test_case->should_run());
+ EXPECT_EQ(1, test_case->disabled_test_count());
+ EXPECT_EQ(3, test_case->test_to_run_count());
+ ASSERT_EQ(4, test_case->total_test_count());
+
+ const TestInfo** tests = UnitTestHelper::GetSortedTests(test_case);
+
+ EXPECT_STREQ("DISABLED_Dummy1", tests[0]->name());
+ EXPECT_STREQ("ApiTest", tests[0]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[0]->value_param()));
+ EXPECT_TRUE(IsNull(tests[0]->type_param()));
+ EXPECT_FALSE(tests[0]->should_run());
+
+ EXPECT_STREQ("TestCaseDisabledAccessorsWork", tests[1]->name());
+ EXPECT_STREQ("ApiTest", tests[1]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[1]->value_param()));
+ EXPECT_TRUE(IsNull(tests[1]->type_param()));
+ EXPECT_TRUE(tests[1]->should_run());
+
+ EXPECT_STREQ("TestCaseImmutableAccessorsWork", tests[2]->name());
+ EXPECT_STREQ("ApiTest", tests[2]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[2]->value_param()));
+ EXPECT_TRUE(IsNull(tests[2]->type_param()));
+ EXPECT_TRUE(tests[2]->should_run());
+
+ EXPECT_STREQ("UnitTestImmutableAccessorsWork", tests[3]->name());
+ EXPECT_STREQ("ApiTest", tests[3]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[3]->value_param()));
+ EXPECT_TRUE(IsNull(tests[3]->type_param()));
+ EXPECT_TRUE(tests[3]->should_run());
+
+ delete[] tests;
+ tests = NULL;
+
+#if GTEST_HAS_TYPED_TEST
+ test_case = UnitTestHelper::FindTestCase("TestCaseWithCommentTest/0");
+ ASSERT_TRUE(test_case != NULL);
+
+ EXPECT_STREQ("TestCaseWithCommentTest/0", test_case->name());
+ EXPECT_STREQ(GetTypeName<int>().c_str(), test_case->type_param());
+ EXPECT_TRUE(test_case->should_run());
+ EXPECT_EQ(0, test_case->disabled_test_count());
+ EXPECT_EQ(1, test_case->test_to_run_count());
+ ASSERT_EQ(1, test_case->total_test_count());
+
+ tests = UnitTestHelper::GetSortedTests(test_case);
+
+ EXPECT_STREQ("Dummy", tests[0]->name());
+ EXPECT_STREQ("TestCaseWithCommentTest/0", tests[0]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[0]->value_param()));
+ EXPECT_STREQ(GetTypeName<int>().c_str(), tests[0]->type_param());
+ EXPECT_TRUE(tests[0]->should_run());
+
+ delete[] tests;
+#endif // GTEST_HAS_TYPED_TEST
+}
+
+TEST(ApiTest, TestCaseDisabledAccessorsWork) {
+ const TestCase* test_case = UnitTestHelper::FindTestCase("DISABLED_Test");
+ ASSERT_TRUE(test_case != NULL);
+
+ EXPECT_STREQ("DISABLED_Test", test_case->name());
+ EXPECT_TRUE(IsNull(test_case->type_param()));
+ EXPECT_FALSE(test_case->should_run());
+ EXPECT_EQ(1, test_case->disabled_test_count());
+ EXPECT_EQ(0, test_case->test_to_run_count());
+ ASSERT_EQ(1, test_case->total_test_count());
+
+ const TestInfo* const test_info = test_case->GetTestInfo(0);
+ EXPECT_STREQ("Dummy2", test_info->name());
+ EXPECT_STREQ("DISABLED_Test", test_info->test_case_name());
+ EXPECT_TRUE(IsNull(test_info->value_param()));
+ EXPECT_TRUE(IsNull(test_info->type_param()));
+ EXPECT_FALSE(test_info->should_run());
+}
+
+// These two tests are here to provide support for testing
+// test_case_to_run_count, disabled_test_count, and test_to_run_count.
+TEST(ApiTest, DISABLED_Dummy1) {}
+TEST(DISABLED_Test, Dummy2) {}
+
+class FinalSuccessChecker : public Environment {
+ protected:
+ virtual void TearDown() {
+ UnitTest* unit_test = UnitTest::GetInstance();
+
+ EXPECT_EQ(1 + kTypedTestCases, unit_test->successful_test_case_count());
+ EXPECT_EQ(3 + kTypedTests, unit_test->successful_test_count());
+ EXPECT_EQ(0, unit_test->failed_test_case_count());
+ EXPECT_EQ(0, unit_test->failed_test_count());
+ EXPECT_TRUE(unit_test->Passed());
+ EXPECT_FALSE(unit_test->Failed());
+ ASSERT_EQ(2 + kTypedTestCases, unit_test->total_test_case_count());
+
+ const TestCase** const test_cases = UnitTestHelper::GetSortedTestCases();
+
+ EXPECT_STREQ("ApiTest", test_cases[0]->name());
+ EXPECT_TRUE(IsNull(test_cases[0]->type_param()));
+ EXPECT_TRUE(test_cases[0]->should_run());
+ EXPECT_EQ(1, test_cases[0]->disabled_test_count());
+ ASSERT_EQ(4, test_cases[0]->total_test_count());
+ EXPECT_EQ(3, test_cases[0]->successful_test_count());
+ EXPECT_EQ(0, test_cases[0]->failed_test_count());
+ EXPECT_TRUE(test_cases[0]->Passed());
+ EXPECT_FALSE(test_cases[0]->Failed());
+
+ EXPECT_STREQ("DISABLED_Test", test_cases[1]->name());
+ EXPECT_TRUE(IsNull(test_cases[1]->type_param()));
+ EXPECT_FALSE(test_cases[1]->should_run());
+ EXPECT_EQ(1, test_cases[1]->disabled_test_count());
+ ASSERT_EQ(1, test_cases[1]->total_test_count());
+ EXPECT_EQ(0, test_cases[1]->successful_test_count());
+ EXPECT_EQ(0, test_cases[1]->failed_test_count());
+
+#if GTEST_HAS_TYPED_TEST
+ EXPECT_STREQ("TestCaseWithCommentTest/0", test_cases[2]->name());
+ EXPECT_STREQ(GetTypeName<int>().c_str(), test_cases[2]->type_param());
+ EXPECT_TRUE(test_cases[2]->should_run());
+ EXPECT_EQ(0, test_cases[2]->disabled_test_count());
+ ASSERT_EQ(1, test_cases[2]->total_test_count());
+ EXPECT_EQ(1, test_cases[2]->successful_test_count());
+ EXPECT_EQ(0, test_cases[2]->failed_test_count());
+ EXPECT_TRUE(test_cases[2]->Passed());
+ EXPECT_FALSE(test_cases[2]->Failed());
+#endif // GTEST_HAS_TYPED_TEST
+
+ const TestCase* test_case = UnitTestHelper::FindTestCase("ApiTest");
+ const TestInfo** tests = UnitTestHelper::GetSortedTests(test_case);
+ EXPECT_STREQ("DISABLED_Dummy1", tests[0]->name());
+ EXPECT_STREQ("ApiTest", tests[0]->test_case_name());
+ EXPECT_FALSE(tests[0]->should_run());
+
+ EXPECT_STREQ("TestCaseDisabledAccessorsWork", tests[1]->name());
+ EXPECT_STREQ("ApiTest", tests[1]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[1]->value_param()));
+ EXPECT_TRUE(IsNull(tests[1]->type_param()));
+ EXPECT_TRUE(tests[1]->should_run());
+ EXPECT_TRUE(tests[1]->result()->Passed());
+ EXPECT_EQ(0, tests[1]->result()->test_property_count());
+
+ EXPECT_STREQ("TestCaseImmutableAccessorsWork", tests[2]->name());
+ EXPECT_STREQ("ApiTest", tests[2]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[2]->value_param()));
+ EXPECT_TRUE(IsNull(tests[2]->type_param()));
+ EXPECT_TRUE(tests[2]->should_run());
+ EXPECT_TRUE(tests[2]->result()->Passed());
+ EXPECT_EQ(0, tests[2]->result()->test_property_count());
+
+ EXPECT_STREQ("UnitTestImmutableAccessorsWork", tests[3]->name());
+ EXPECT_STREQ("ApiTest", tests[3]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[3]->value_param()));
+ EXPECT_TRUE(IsNull(tests[3]->type_param()));
+ EXPECT_TRUE(tests[3]->should_run());
+ EXPECT_TRUE(tests[3]->result()->Passed());
+ EXPECT_EQ(1, tests[3]->result()->test_property_count());
+ const TestProperty& property = tests[3]->result()->GetTestProperty(0);
+ EXPECT_STREQ("key", property.key());
+ EXPECT_STREQ("value", property.value());
+
+ delete[] tests;
+
+#if GTEST_HAS_TYPED_TEST
+ test_case = UnitTestHelper::FindTestCase("TestCaseWithCommentTest/0");
+ tests = UnitTestHelper::GetSortedTests(test_case);
+
+ EXPECT_STREQ("Dummy", tests[0]->name());
+ EXPECT_STREQ("TestCaseWithCommentTest/0", tests[0]->test_case_name());
+ EXPECT_TRUE(IsNull(tests[0]->value_param()));
+ EXPECT_STREQ(GetTypeName<int>().c_str(), tests[0]->type_param());
+ EXPECT_TRUE(tests[0]->should_run());
+ EXPECT_TRUE(tests[0]->result()->Passed());
+ EXPECT_EQ(0, tests[0]->result()->test_property_count());
+
+ delete[] tests;
+#endif // GTEST_HAS_TYPED_TEST
+ delete[] test_cases;
+ }
+};
+
+} // namespace internal
+} // namespace testing
+
+int main(int argc, char **argv) {
+ InitGoogleTest(&argc, argv);
+
+ AddGlobalTestEnvironment(new testing::internal::FinalSuccessChecker());
+
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_all_test.cc b/libvpx/third_party/googletest/src/test/gtest_all_test.cc
new file mode 100644
index 0000000..955aa62
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_all_test.cc
@@ -0,0 +1,47 @@
+// Copyright 2009, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Tests for Google C++ Testing Framework (Google Test)
+//
+// Sometimes it's desirable to build most of Google Test's own tests
+// by compiling a single file. This file serves this purpose.
+#include "test/gtest-filepath_test.cc"
+#include "test/gtest-linked_ptr_test.cc"
+#include "test/gtest-message_test.cc"
+#include "test/gtest-options_test.cc"
+#include "test/gtest-port_test.cc"
+#include "test/gtest_pred_impl_unittest.cc"
+#include "test/gtest_prod_test.cc"
+#include "test/gtest-test-part_test.cc"
+#include "test/gtest-typed-test_test.cc"
+#include "test/gtest-typed-test2_test.cc"
+#include "test/gtest_unittest.cc"
+#include "test/production.cc"
diff --git a/libvpx/third_party/googletest/src/test/gtest_break_on_failure_unittest.py b/libvpx/third_party/googletest/src/test/gtest_break_on_failure_unittest.py
new file mode 100755
index 0000000..c819183
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_break_on_failure_unittest.py
@@ -0,0 +1,218 @@
+#!/usr/bin/env python
+#
+# Copyright 2006, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Unit test for Google Test's break-on-failure mode.
+
+A user can ask Google Test to seg-fault when an assertion fails, using
+either the GTEST_BREAK_ON_FAILURE environment variable or the
+--gtest_break_on_failure flag. This script tests such functionality
+by invoking gtest_break_on_failure_unittest_ (a program written with
+Google Test) with different environments and command line flags.
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import gtest_test_utils
+import os
+import sys
+
+
+# Constants.
+
+IS_WINDOWS = os.name == 'nt'
+
+# The environment variable for enabling/disabling the break-on-failure mode.
+BREAK_ON_FAILURE_ENV_VAR = 'GTEST_BREAK_ON_FAILURE'
+
+# The command line flag for enabling/disabling the break-on-failure mode.
+BREAK_ON_FAILURE_FLAG = 'gtest_break_on_failure'
+
+# The environment variable for enabling/disabling the throw-on-failure mode.
+THROW_ON_FAILURE_ENV_VAR = 'GTEST_THROW_ON_FAILURE'
+
+# The environment variable for enabling/disabling the catch-exceptions mode.
+CATCH_EXCEPTIONS_ENV_VAR = 'GTEST_CATCH_EXCEPTIONS'
+
+# Path to the gtest_break_on_failure_unittest_ program.
+EXE_PATH = gtest_test_utils.GetTestExecutablePath(
+ 'gtest_break_on_failure_unittest_')
+
+
+# Utilities.
+
+
+environ = os.environ.copy()
+
+
+def SetEnvVar(env_var, value):
+ """Sets an environment variable to a given value; unsets it when the
+ given value is None.
+ """
+
+ if value is not None:
+ environ[env_var] = value
+ elif env_var in environ:
+ del environ[env_var]
+
+
+def Run(command):
+ """Runs a command; returns 1 if it was killed by a signal, or 0 otherwise."""
+
+ p = gtest_test_utils.Subprocess(command, env=environ)
+ if p.terminated_by_signal:
+ return 1
+ else:
+ return 0
+
+
+# The tests.
+
+
+class GTestBreakOnFailureUnitTest(gtest_test_utils.TestCase):
+ """Tests using the GTEST_BREAK_ON_FAILURE environment variable or
+ the --gtest_break_on_failure flag to turn assertion failures into
+ segmentation faults.
+ """
+
+ def RunAndVerify(self, env_var_value, flag_value, expect_seg_fault):
+ """Runs gtest_break_on_failure_unittest_ and verifies that it does
+ (or does not) have a seg-fault.
+
+ Args:
+ env_var_value: value of the GTEST_BREAK_ON_FAILURE environment
+ variable; None if the variable should be unset.
+ flag_value: value of the --gtest_break_on_failure flag;
+ None if the flag should not be present.
+ expect_seg_fault: 1 if the program is expected to generate a seg-fault;
+ 0 otherwise.
+ """
+
+ SetEnvVar(BREAK_ON_FAILURE_ENV_VAR, env_var_value)
+
+ if env_var_value is None:
+ env_var_value_msg = ' is not set'
+ else:
+ env_var_value_msg = '=' + env_var_value
+
+ if flag_value is None:
+ flag = ''
+ elif flag_value == '0':
+ flag = '--%s=0' % BREAK_ON_FAILURE_FLAG
+ else:
+ flag = '--%s' % BREAK_ON_FAILURE_FLAG
+
+ command = [EXE_PATH]
+ if flag:
+ command.append(flag)
+
+ if expect_seg_fault:
+ should_or_not = 'should'
+ else:
+ should_or_not = 'should not'
+
+ has_seg_fault = Run(command)
+
+ SetEnvVar(BREAK_ON_FAILURE_ENV_VAR, None)
+
+ msg = ('when %s%s, an assertion failure in "%s" %s cause a seg-fault.' %
+ (BREAK_ON_FAILURE_ENV_VAR, env_var_value_msg, ' '.join(command),
+ should_or_not))
+ self.assert_(has_seg_fault == expect_seg_fault, msg)
+
+ def testDefaultBehavior(self):
+ """Tests the behavior of the default mode."""
+
+ self.RunAndVerify(env_var_value=None,
+ flag_value=None,
+ expect_seg_fault=0)
+
+ def testEnvVar(self):
+ """Tests using the GTEST_BREAK_ON_FAILURE environment variable."""
+
+ self.RunAndVerify(env_var_value='0',
+ flag_value=None,
+ expect_seg_fault=0)
+ self.RunAndVerify(env_var_value='1',
+ flag_value=None,
+ expect_seg_fault=1)
+
+ def testFlag(self):
+ """Tests using the --gtest_break_on_failure flag."""
+
+ self.RunAndVerify(env_var_value=None,
+ flag_value='0',
+ expect_seg_fault=0)
+ self.RunAndVerify(env_var_value=None,
+ flag_value='1',
+ expect_seg_fault=1)
+
+ def testFlagOverridesEnvVar(self):
+ """Tests that the flag overrides the environment variable."""
+
+ self.RunAndVerify(env_var_value='0',
+ flag_value='0',
+ expect_seg_fault=0)
+ self.RunAndVerify(env_var_value='0',
+ flag_value='1',
+ expect_seg_fault=1)
+ self.RunAndVerify(env_var_value='1',
+ flag_value='0',
+ expect_seg_fault=0)
+ self.RunAndVerify(env_var_value='1',
+ flag_value='1',
+ expect_seg_fault=1)
+
+ def testBreakOnFailureOverridesThrowOnFailure(self):
+ """Tests that gtest_break_on_failure overrides gtest_throw_on_failure."""
+
+ SetEnvVar(THROW_ON_FAILURE_ENV_VAR, '1')
+ try:
+ self.RunAndVerify(env_var_value=None,
+ flag_value='1',
+ expect_seg_fault=1)
+ finally:
+ SetEnvVar(THROW_ON_FAILURE_ENV_VAR, None)
+
+ if IS_WINDOWS:
+ def testCatchExceptionsDoesNotInterfere(self):
+ """Tests that gtest_catch_exceptions doesn't interfere."""
+
+ SetEnvVar(CATCH_EXCEPTIONS_ENV_VAR, '1')
+ try:
+ self.RunAndVerify(env_var_value='1',
+ flag_value='1',
+ expect_seg_fault=1)
+ finally:
+ SetEnvVar(CATCH_EXCEPTIONS_ENV_VAR, None)
+
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_break_on_failure_unittest_.cc b/libvpx/third_party/googletest/src/test/gtest_break_on_failure_unittest_.cc
new file mode 100644
index 0000000..dd07478
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_break_on_failure_unittest_.cc
@@ -0,0 +1,88 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Unit test for Google Test's break-on-failure mode.
+//
+// A user can ask Google Test to seg-fault when an assertion fails, using
+// either the GTEST_BREAK_ON_FAILURE environment variable or the
+// --gtest_break_on_failure flag. This file is used for testing such
+// functionality.
+//
+// This program will be invoked from a Python unit test. It is
+// expected to fail. Don't run it directly.
+
+#include "gtest/gtest.h"
+
+#if GTEST_OS_WINDOWS
+# include <windows.h>
+# include <stdlib.h>
+#endif
+
+namespace {
+
+// A test that's expected to fail.
+TEST(Foo, Bar) {
+ EXPECT_EQ(2, 3);
+}
+
+#if GTEST_HAS_SEH && !GTEST_OS_WINDOWS_MOBILE
+// On Windows Mobile global exception handlers are not supported.
+LONG WINAPI ExitWithExceptionCode(
+ struct _EXCEPTION_POINTERS* exception_pointers) {
+ exit(exception_pointers->ExceptionRecord->ExceptionCode);
+}
+#endif
+
+} // namespace
+
+int main(int argc, char **argv) {
+#if GTEST_OS_WINDOWS
+ // Suppresses display of the Windows error dialog upon encountering
+ // a general protection fault (segment violation).
+ SetErrorMode(SEM_NOGPFAULTERRORBOX | SEM_FAILCRITICALERRORS);
+
+# if GTEST_HAS_SEH && !GTEST_OS_WINDOWS_MOBILE
+
+ // The default unhandled exception filter does not always exit
+ // with the exception code as exit code - for example it exits with
+ // 0 for EXCEPTION_ACCESS_VIOLATION and 1 for EXCEPTION_BREAKPOINT
+ // if the application is compiled in debug mode. Thus we use our own
+ // filter which always exits with the exception code for unhandled
+ // exceptions.
+ SetUnhandledExceptionFilter(ExitWithExceptionCode);
+
+# endif
+#endif
+
+ testing::InitGoogleTest(&argc, argv);
+
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_catch_exceptions_test.py b/libvpx/third_party/googletest/src/test/gtest_catch_exceptions_test.py
new file mode 100755
index 0000000..7fd7dba
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_catch_exceptions_test.py
@@ -0,0 +1,220 @@
+#!/usr/bin/env python
+#
+# Copyright 2010 Google Inc. All Rights Reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Tests Google Test's exception catching behavior.
+
+This script invokes gtest_catch_exceptions_test_ and
+gtest_catch_exceptions_ex_test_ (programs written with
+Google Test) and verifies their output.
+"""
+
+__author__ = 'vladl@google.com (Vlad Losev)'
+
+import os
+
+import gtest_test_utils
+
+# Constants.
+FLAG_PREFIX = '--gtest_'
+LIST_TESTS_FLAG = FLAG_PREFIX + 'list_tests'
+NO_CATCH_EXCEPTIONS_FLAG = FLAG_PREFIX + 'catch_exceptions=0'
+FILTER_FLAG = FLAG_PREFIX + 'filter'
+
+# Path to the gtest_catch_exceptions_ex_test_ binary, compiled with
+# exceptions enabled.
+EX_EXE_PATH = gtest_test_utils.GetTestExecutablePath(
+ 'gtest_catch_exceptions_ex_test_')
+
+# Path to the gtest_catch_exceptions_test_ binary, compiled with
+# exceptions disabled.
+EXE_PATH = gtest_test_utils.GetTestExecutablePath(
+ 'gtest_catch_exceptions_no_ex_test_')
+
+TEST_LIST = gtest_test_utils.Subprocess([EXE_PATH, LIST_TESTS_FLAG]).output
+
+SUPPORTS_SEH_EXCEPTIONS = 'ThrowsSehException' in TEST_LIST
+
+if SUPPORTS_SEH_EXCEPTIONS:
+ BINARY_OUTPUT = gtest_test_utils.Subprocess([EXE_PATH]).output
+
+EX_BINARY_OUTPUT = gtest_test_utils.Subprocess([EX_EXE_PATH]).output
+
+# The tests.
+if SUPPORTS_SEH_EXCEPTIONS:
+ # pylint:disable-msg=C6302
+ class CatchSehExceptionsTest(gtest_test_utils.TestCase):
+ """Tests exception-catching behavior."""
+
+
+ def TestSehExceptions(self, test_output):
+ self.assert_('SEH exception with code 0x2a thrown '
+ 'in the test fixture\'s constructor'
+ in test_output)
+ self.assert_('SEH exception with code 0x2a thrown '
+ 'in the test fixture\'s destructor'
+ in test_output)
+ self.assert_('SEH exception with code 0x2a thrown in SetUpTestCase()'
+ in test_output)
+ self.assert_('SEH exception with code 0x2a thrown in TearDownTestCase()'
+ in test_output)
+ self.assert_('SEH exception with code 0x2a thrown in SetUp()'
+ in test_output)
+ self.assert_('SEH exception with code 0x2a thrown in TearDown()'
+ in test_output)
+ self.assert_('SEH exception with code 0x2a thrown in the test body'
+ in test_output)
+
+ def testCatchesSehExceptionsWithCxxExceptionsEnabled(self):
+ self.TestSehExceptions(EX_BINARY_OUTPUT)
+
+ def testCatchesSehExceptionsWithCxxExceptionsDisabled(self):
+ self.TestSehExceptions(BINARY_OUTPUT)
+
+
+class CatchCxxExceptionsTest(gtest_test_utils.TestCase):
+ """Tests C++ exception-catching behavior.
+
+ Tests in this test case verify that:
+ * C++ exceptions are caught and logged as C++ (not SEH) exceptions
+ * Exception thrown affect the remainder of the test work flow in the
+ expected manner.
+ """
+
+ def testCatchesCxxExceptionsInFixtureConstructor(self):
+ self.assert_('C++ exception with description '
+ '"Standard C++ exception" thrown '
+ 'in the test fixture\'s constructor'
+ in EX_BINARY_OUTPUT)
+ self.assert_('unexpected' not in EX_BINARY_OUTPUT,
+ 'This failure belongs in this test only if '
+ '"CxxExceptionInConstructorTest" (no quotes) '
+ 'appears on the same line as words "called unexpectedly"')
+
+ def testCatchesCxxExceptionsInFixtureDestructor(self):
+ self.assert_('C++ exception with description '
+ '"Standard C++ exception" thrown '
+ 'in the test fixture\'s destructor'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInDestructorTest::TearDownTestCase() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+
+ def testCatchesCxxExceptionsInSetUpTestCase(self):
+ self.assert_('C++ exception with description "Standard C++ exception"'
+ ' thrown in SetUpTestCase()'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInConstructorTest::TearDownTestCase() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInSetUpTestCaseTest constructor '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInSetUpTestCaseTest destructor '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInSetUpTestCaseTest::SetUp() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInSetUpTestCaseTest::TearDown() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInSetUpTestCaseTest test body '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+
+ def testCatchesCxxExceptionsInTearDownTestCase(self):
+ self.assert_('C++ exception with description "Standard C++ exception"'
+ ' thrown in TearDownTestCase()'
+ in EX_BINARY_OUTPUT)
+
+ def testCatchesCxxExceptionsInSetUp(self):
+ self.assert_('C++ exception with description "Standard C++ exception"'
+ ' thrown in SetUp()'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInSetUpTest::TearDownTestCase() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInSetUpTest destructor '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInSetUpTest::TearDown() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('unexpected' not in EX_BINARY_OUTPUT,
+ 'This failure belongs in this test only if '
+ '"CxxExceptionInSetUpTest" (no quotes) '
+ 'appears on the same line as words "called unexpectedly"')
+
+ def testCatchesCxxExceptionsInTearDown(self):
+ self.assert_('C++ exception with description "Standard C++ exception"'
+ ' thrown in TearDown()'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInTearDownTest::TearDownTestCase() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInTearDownTest destructor '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+
+ def testCatchesCxxExceptionsInTestBody(self):
+ self.assert_('C++ exception with description "Standard C++ exception"'
+ ' thrown in the test body'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInTestBodyTest::TearDownTestCase() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInTestBodyTest destructor '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+ self.assert_('CxxExceptionInTestBodyTest::TearDown() '
+ 'called as expected.'
+ in EX_BINARY_OUTPUT)
+
+ def testCatchesNonStdCxxExceptions(self):
+ self.assert_('Unknown C++ exception thrown in the test body'
+ in EX_BINARY_OUTPUT)
+
+ def testUnhandledCxxExceptionsAbortTheProgram(self):
+ # Filters out SEH exception tests on Windows. Unhandled SEH exceptions
+ # cause tests to show pop-up windows there.
+ FITLER_OUT_SEH_TESTS_FLAG = FILTER_FLAG + '=-*Seh*'
+ # By default, Google Test doesn't catch the exceptions.
+ uncaught_exceptions_ex_binary_output = gtest_test_utils.Subprocess(
+ [EX_EXE_PATH,
+ NO_CATCH_EXCEPTIONS_FLAG,
+ FITLER_OUT_SEH_TESTS_FLAG]).output
+
+ self.assert_('Unhandled C++ exception terminating the program'
+ in uncaught_exceptions_ex_binary_output)
+ self.assert_('unexpected' not in uncaught_exceptions_ex_binary_output)
+
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_catch_exceptions_test_.cc b/libvpx/third_party/googletest/src/test/gtest_catch_exceptions_test_.cc
new file mode 100644
index 0000000..a35103f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_catch_exceptions_test_.cc
@@ -0,0 +1,308 @@
+// Copyright 2010, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+//
+// Tests for Google Test itself. Tests in this file throw C++ or SEH
+// exceptions, and the output is verified by gtest_catch_exceptions_test.py.
+
+#include "gtest/gtest.h"
+
+#include <stdio.h> // NOLINT
+#include <stdlib.h> // For exit().
+
+#if GTEST_HAS_SEH
+# include <windows.h>
+#endif
+
+#if GTEST_HAS_EXCEPTIONS
+# include <exception> // For set_terminate().
+# include <stdexcept>
+#endif
+
+using testing::Test;
+
+#if GTEST_HAS_SEH
+
+class SehExceptionInConstructorTest : public Test {
+ public:
+ SehExceptionInConstructorTest() { RaiseException(42, 0, 0, NULL); }
+};
+
+TEST_F(SehExceptionInConstructorTest, ThrowsExceptionInConstructor) {}
+
+class SehExceptionInDestructorTest : public Test {
+ public:
+ ~SehExceptionInDestructorTest() { RaiseException(42, 0, 0, NULL); }
+};
+
+TEST_F(SehExceptionInDestructorTest, ThrowsExceptionInDestructor) {}
+
+class SehExceptionInSetUpTestCaseTest : public Test {
+ public:
+ static void SetUpTestCase() { RaiseException(42, 0, 0, NULL); }
+};
+
+TEST_F(SehExceptionInSetUpTestCaseTest, ThrowsExceptionInSetUpTestCase) {}
+
+class SehExceptionInTearDownTestCaseTest : public Test {
+ public:
+ static void TearDownTestCase() { RaiseException(42, 0, 0, NULL); }
+};
+
+TEST_F(SehExceptionInTearDownTestCaseTest, ThrowsExceptionInTearDownTestCase) {}
+
+class SehExceptionInSetUpTest : public Test {
+ protected:
+ virtual void SetUp() { RaiseException(42, 0, 0, NULL); }
+};
+
+TEST_F(SehExceptionInSetUpTest, ThrowsExceptionInSetUp) {}
+
+class SehExceptionInTearDownTest : public Test {
+ protected:
+ virtual void TearDown() { RaiseException(42, 0, 0, NULL); }
+};
+
+TEST_F(SehExceptionInTearDownTest, ThrowsExceptionInTearDown) {}
+
+TEST(SehExceptionTest, ThrowsSehException) {
+ RaiseException(42, 0, 0, NULL);
+}
+
+#endif // GTEST_HAS_SEH
+
+#if GTEST_HAS_EXCEPTIONS
+
+class CxxExceptionInConstructorTest : public Test {
+ public:
+ CxxExceptionInConstructorTest() {
+ // Without this macro VC++ complains about unreachable code at the end of
+ // the constructor.
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(
+ throw std::runtime_error("Standard C++ exception"));
+ }
+
+ static void TearDownTestCase() {
+ printf("%s",
+ "CxxExceptionInConstructorTest::TearDownTestCase() "
+ "called as expected.\n");
+ }
+
+ protected:
+ ~CxxExceptionInConstructorTest() {
+ ADD_FAILURE() << "CxxExceptionInConstructorTest destructor "
+ << "called unexpectedly.";
+ }
+
+ virtual void SetUp() {
+ ADD_FAILURE() << "CxxExceptionInConstructorTest::SetUp() "
+ << "called unexpectedly.";
+ }
+
+ virtual void TearDown() {
+ ADD_FAILURE() << "CxxExceptionInConstructorTest::TearDown() "
+ << "called unexpectedly.";
+ }
+};
+
+TEST_F(CxxExceptionInConstructorTest, ThrowsExceptionInConstructor) {
+ ADD_FAILURE() << "CxxExceptionInConstructorTest test body "
+ << "called unexpectedly.";
+}
+
+class CxxExceptionInDestructorTest : public Test {
+ public:
+ static void TearDownTestCase() {
+ printf("%s",
+ "CxxExceptionInDestructorTest::TearDownTestCase() "
+ "called as expected.\n");
+ }
+
+ protected:
+ ~CxxExceptionInDestructorTest() {
+ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(
+ throw std::runtime_error("Standard C++ exception"));
+ }
+};
+
+TEST_F(CxxExceptionInDestructorTest, ThrowsExceptionInDestructor) {}
+
+class CxxExceptionInSetUpTestCaseTest : public Test {
+ public:
+ CxxExceptionInSetUpTestCaseTest() {
+ printf("%s",
+ "CxxExceptionInSetUpTestCaseTest constructor "
+ "called as expected.\n");
+ }
+
+ static void SetUpTestCase() {
+ throw std::runtime_error("Standard C++ exception");
+ }
+
+ static void TearDownTestCase() {
+ printf("%s",
+ "CxxExceptionInSetUpTestCaseTest::TearDownTestCase() "
+ "called as expected.\n");
+ }
+
+ protected:
+ ~CxxExceptionInSetUpTestCaseTest() {
+ printf("%s",
+ "CxxExceptionInSetUpTestCaseTest destructor "
+ "called as expected.\n");
+ }
+
+ virtual void SetUp() {
+ printf("%s",
+ "CxxExceptionInSetUpTestCaseTest::SetUp() "
+ "called as expected.\n");
+ }
+
+ virtual void TearDown() {
+ printf("%s",
+ "CxxExceptionInSetUpTestCaseTest::TearDown() "
+ "called as expected.\n");
+ }
+};
+
+TEST_F(CxxExceptionInSetUpTestCaseTest, ThrowsExceptionInSetUpTestCase) {
+ printf("%s",
+ "CxxExceptionInSetUpTestCaseTest test body "
+ "called as expected.\n");
+}
+
+class CxxExceptionInTearDownTestCaseTest : public Test {
+ public:
+ static void TearDownTestCase() {
+ throw std::runtime_error("Standard C++ exception");
+ }
+};
+
+TEST_F(CxxExceptionInTearDownTestCaseTest, ThrowsExceptionInTearDownTestCase) {}
+
+class CxxExceptionInSetUpTest : public Test {
+ public:
+ static void TearDownTestCase() {
+ printf("%s",
+ "CxxExceptionInSetUpTest::TearDownTestCase() "
+ "called as expected.\n");
+ }
+
+ protected:
+ ~CxxExceptionInSetUpTest() {
+ printf("%s",
+ "CxxExceptionInSetUpTest destructor "
+ "called as expected.\n");
+ }
+
+ virtual void SetUp() { throw std::runtime_error("Standard C++ exception"); }
+
+ virtual void TearDown() {
+ printf("%s",
+ "CxxExceptionInSetUpTest::TearDown() "
+ "called as expected.\n");
+ }
+};
+
+TEST_F(CxxExceptionInSetUpTest, ThrowsExceptionInSetUp) {
+ ADD_FAILURE() << "CxxExceptionInSetUpTest test body "
+ << "called unexpectedly.";
+}
+
+class CxxExceptionInTearDownTest : public Test {
+ public:
+ static void TearDownTestCase() {
+ printf("%s",
+ "CxxExceptionInTearDownTest::TearDownTestCase() "
+ "called as expected.\n");
+ }
+
+ protected:
+ ~CxxExceptionInTearDownTest() {
+ printf("%s",
+ "CxxExceptionInTearDownTest destructor "
+ "called as expected.\n");
+ }
+
+ virtual void TearDown() {
+ throw std::runtime_error("Standard C++ exception");
+ }
+};
+
+TEST_F(CxxExceptionInTearDownTest, ThrowsExceptionInTearDown) {}
+
+class CxxExceptionInTestBodyTest : public Test {
+ public:
+ static void TearDownTestCase() {
+ printf("%s",
+ "CxxExceptionInTestBodyTest::TearDownTestCase() "
+ "called as expected.\n");
+ }
+
+ protected:
+ ~CxxExceptionInTestBodyTest() {
+ printf("%s",
+ "CxxExceptionInTestBodyTest destructor "
+ "called as expected.\n");
+ }
+
+ virtual void TearDown() {
+ printf("%s",
+ "CxxExceptionInTestBodyTest::TearDown() "
+ "called as expected.\n");
+ }
+};
+
+TEST_F(CxxExceptionInTestBodyTest, ThrowsStdCxxException) {
+ throw std::runtime_error("Standard C++ exception");
+}
+
+TEST(CxxExceptionTest, ThrowsNonStdCxxException) {
+ throw "C-string";
+}
+
+// This terminate handler aborts the program using exit() rather than abort().
+// This avoids showing pop-ups on Windows systems and core dumps on Unix-like
+// ones.
+void TerminateHandler() {
+ fprintf(stderr, "%s\n", "Unhandled C++ exception terminating the program.");
+ fflush(NULL);
+ exit(3);
+}
+
+#endif // GTEST_HAS_EXCEPTIONS
+
+int main(int argc, char** argv) {
+#if GTEST_HAS_EXCEPTIONS
+ std::set_terminate(&TerminateHandler);
+#endif
+ testing::InitGoogleTest(&argc, argv);
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_color_test.py b/libvpx/third_party/googletest/src/test/gtest_color_test.py
new file mode 100755
index 0000000..d02a53e
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_color_test.py
@@ -0,0 +1,130 @@
+#!/usr/bin/env python
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Verifies that Google Test correctly determines whether to use colors."""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import gtest_test_utils
+
+
+IS_WINDOWS = os.name = 'nt'
+
+COLOR_ENV_VAR = 'GTEST_COLOR'
+COLOR_FLAG = 'gtest_color'
+COMMAND = gtest_test_utils.GetTestExecutablePath('gtest_color_test_')
+
+
+def SetEnvVar(env_var, value):
+ """Sets the env variable to 'value'; unsets it when 'value' is None."""
+
+ if value is not None:
+ os.environ[env_var] = value
+ elif env_var in os.environ:
+ del os.environ[env_var]
+
+
+def UsesColor(term, color_env_var, color_flag):
+ """Runs gtest_color_test_ and returns its exit code."""
+
+ SetEnvVar('TERM', term)
+ SetEnvVar(COLOR_ENV_VAR, color_env_var)
+
+ if color_flag is None:
+ args = []
+ else:
+ args = ['--%s=%s' % (COLOR_FLAG, color_flag)]
+ p = gtest_test_utils.Subprocess([COMMAND] + args)
+ return not p.exited or p.exit_code
+
+
+class GTestColorTest(gtest_test_utils.TestCase):
+ def testNoEnvVarNoFlag(self):
+ """Tests the case when there's neither GTEST_COLOR nor --gtest_color."""
+
+ if not IS_WINDOWS:
+ self.assert_(not UsesColor('dumb', None, None))
+ self.assert_(not UsesColor('emacs', None, None))
+ self.assert_(not UsesColor('xterm-mono', None, None))
+ self.assert_(not UsesColor('unknown', None, None))
+ self.assert_(not UsesColor(None, None, None))
+ self.assert_(UsesColor('linux', None, None))
+ self.assert_(UsesColor('cygwin', None, None))
+ self.assert_(UsesColor('xterm', None, None))
+ self.assert_(UsesColor('xterm-color', None, None))
+ self.assert_(UsesColor('xterm-256color', None, None))
+
+ def testFlagOnly(self):
+ """Tests the case when there's --gtest_color but not GTEST_COLOR."""
+
+ self.assert_(not UsesColor('dumb', None, 'no'))
+ self.assert_(not UsesColor('xterm-color', None, 'no'))
+ if not IS_WINDOWS:
+ self.assert_(not UsesColor('emacs', None, 'auto'))
+ self.assert_(UsesColor('xterm', None, 'auto'))
+ self.assert_(UsesColor('dumb', None, 'yes'))
+ self.assert_(UsesColor('xterm', None, 'yes'))
+
+ def testEnvVarOnly(self):
+ """Tests the case when there's GTEST_COLOR but not --gtest_color."""
+
+ self.assert_(not UsesColor('dumb', 'no', None))
+ self.assert_(not UsesColor('xterm-color', 'no', None))
+ if not IS_WINDOWS:
+ self.assert_(not UsesColor('dumb', 'auto', None))
+ self.assert_(UsesColor('xterm-color', 'auto', None))
+ self.assert_(UsesColor('dumb', 'yes', None))
+ self.assert_(UsesColor('xterm-color', 'yes', None))
+
+ def testEnvVarAndFlag(self):
+ """Tests the case when there are both GTEST_COLOR and --gtest_color."""
+
+ self.assert_(not UsesColor('xterm-color', 'no', 'no'))
+ self.assert_(UsesColor('dumb', 'no', 'yes'))
+ self.assert_(UsesColor('xterm-color', 'no', 'auto'))
+
+ def testAliasesOfYesAndNo(self):
+ """Tests using aliases in specifying --gtest_color."""
+
+ self.assert_(UsesColor('dumb', None, 'true'))
+ self.assert_(UsesColor('dumb', None, 'YES'))
+ self.assert_(UsesColor('dumb', None, 'T'))
+ self.assert_(UsesColor('dumb', None, '1'))
+
+ self.assert_(not UsesColor('xterm', None, 'f'))
+ self.assert_(not UsesColor('xterm', None, 'false'))
+ self.assert_(not UsesColor('xterm', None, '0'))
+ self.assert_(not UsesColor('xterm', None, 'unknown'))
+
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_color_test_.cc b/libvpx/third_party/googletest/src/test/gtest_color_test_.cc
new file mode 100644
index 0000000..f61ebb8
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_color_test_.cc
@@ -0,0 +1,71 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// A helper program for testing how Google Test determines whether to use
+// colors in the output. It prints "YES" and returns 1 if Google Test
+// decides to use colors, and prints "NO" and returns 0 otherwise.
+
+#include <stdio.h>
+
+#include "gtest/gtest.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+using testing::internal::ShouldUseColor;
+
+// The purpose of this is to ensure that the UnitTest singleton is
+// created before main() is entered, and thus that ShouldUseColor()
+// works the same way as in a real Google-Test-based test. We don't actual
+// run the TEST itself.
+TEST(GTestColorTest, Dummy) {
+}
+
+int main(int argc, char** argv) {
+ testing::InitGoogleTest(&argc, argv);
+
+ if (ShouldUseColor(true)) {
+ // Google Test decides to use colors in the output (assuming it
+ // goes to a TTY).
+ printf("YES\n");
+ return 1;
+ } else {
+ // Google Test decides not to use colors in the output.
+ printf("NO\n");
+ return 0;
+ }
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_env_var_test.py b/libvpx/third_party/googletest/src/test/gtest_env_var_test.py
new file mode 100755
index 0000000..ac24337
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_env_var_test.py
@@ -0,0 +1,103 @@
+#!/usr/bin/env python
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Verifies that Google Test correctly parses environment variables."""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import gtest_test_utils
+
+
+IS_WINDOWS = os.name == 'nt'
+IS_LINUX = os.name == 'posix' and os.uname()[0] == 'Linux'
+
+COMMAND = gtest_test_utils.GetTestExecutablePath('gtest_env_var_test_')
+
+environ = os.environ.copy()
+
+
+def AssertEq(expected, actual):
+ if expected != actual:
+ print 'Expected: %s' % (expected,)
+ print ' Actual: %s' % (actual,)
+ raise AssertionError
+
+
+def SetEnvVar(env_var, value):
+ """Sets the env variable to 'value'; unsets it when 'value' is None."""
+
+ if value is not None:
+ environ[env_var] = value
+ elif env_var in environ:
+ del environ[env_var]
+
+
+def GetFlag(flag):
+ """Runs gtest_env_var_test_ and returns its output."""
+
+ args = [COMMAND]
+ if flag is not None:
+ args += [flag]
+ return gtest_test_utils.Subprocess(args, env=environ).output
+
+
+def TestFlag(flag, test_val, default_val):
+ """Verifies that the given flag is affected by the corresponding env var."""
+
+ env_var = 'GTEST_' + flag.upper()
+ SetEnvVar(env_var, test_val)
+ AssertEq(test_val, GetFlag(flag))
+ SetEnvVar(env_var, None)
+ AssertEq(default_val, GetFlag(flag))
+
+
+class GTestEnvVarTest(gtest_test_utils.TestCase):
+ def testEnvVarAffectsFlag(self):
+ """Tests that environment variable should affect the corresponding flag."""
+
+ TestFlag('break_on_failure', '1', '0')
+ TestFlag('color', 'yes', 'auto')
+ TestFlag('filter', 'FooTest.Bar', '*')
+ TestFlag('output', 'xml:tmp/foo.xml', '')
+ TestFlag('print_time', '0', '1')
+ TestFlag('repeat', '999', '1')
+ TestFlag('throw_on_failure', '1', '0')
+ TestFlag('death_test_style', 'threadsafe', 'fast')
+ TestFlag('catch_exceptions', '0', '1')
+
+ if IS_LINUX:
+ TestFlag('death_test_use_fork', '1', '0')
+ TestFlag('stack_trace_depth', '0', '100')
+
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_env_var_test_.cc b/libvpx/third_party/googletest/src/test/gtest_env_var_test_.cc
new file mode 100644
index 0000000..539afc9
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_env_var_test_.cc
@@ -0,0 +1,126 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// A helper program for testing that Google Test parses the environment
+// variables correctly.
+
+#include "gtest/gtest.h"
+
+#include <iostream>
+
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+using ::std::cout;
+
+namespace testing {
+
+// The purpose of this is to make the test more realistic by ensuring
+// that the UnitTest singleton is created before main() is entered.
+// We don't actual run the TEST itself.
+TEST(GTestEnvVarTest, Dummy) {
+}
+
+void PrintFlag(const char* flag) {
+ if (strcmp(flag, "break_on_failure") == 0) {
+ cout << GTEST_FLAG(break_on_failure);
+ return;
+ }
+
+ if (strcmp(flag, "catch_exceptions") == 0) {
+ cout << GTEST_FLAG(catch_exceptions);
+ return;
+ }
+
+ if (strcmp(flag, "color") == 0) {
+ cout << GTEST_FLAG(color);
+ return;
+ }
+
+ if (strcmp(flag, "death_test_style") == 0) {
+ cout << GTEST_FLAG(death_test_style);
+ return;
+ }
+
+ if (strcmp(flag, "death_test_use_fork") == 0) {
+ cout << GTEST_FLAG(death_test_use_fork);
+ return;
+ }
+
+ if (strcmp(flag, "filter") == 0) {
+ cout << GTEST_FLAG(filter);
+ return;
+ }
+
+ if (strcmp(flag, "output") == 0) {
+ cout << GTEST_FLAG(output);
+ return;
+ }
+
+ if (strcmp(flag, "print_time") == 0) {
+ cout << GTEST_FLAG(print_time);
+ return;
+ }
+
+ if (strcmp(flag, "repeat") == 0) {
+ cout << GTEST_FLAG(repeat);
+ return;
+ }
+
+ if (strcmp(flag, "stack_trace_depth") == 0) {
+ cout << GTEST_FLAG(stack_trace_depth);
+ return;
+ }
+
+ if (strcmp(flag, "throw_on_failure") == 0) {
+ cout << GTEST_FLAG(throw_on_failure);
+ return;
+ }
+
+ cout << "Invalid flag name " << flag
+ << ". Valid names are break_on_failure, color, filter, etc.\n";
+ exit(1);
+}
+
+} // namespace testing
+
+int main(int argc, char** argv) {
+ testing::InitGoogleTest(&argc, argv);
+
+ if (argc != 2) {
+ cout << "Usage: gtest_env_var_test_ NAME_OF_FLAG\n";
+ return 1;
+ }
+
+ testing::PrintFlag(argv[1]);
+ return 0;
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_environment_test.cc b/libvpx/third_party/googletest/src/test/gtest_environment_test.cc
new file mode 100644
index 0000000..ec9aa2c
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_environment_test.cc
@@ -0,0 +1,191 @@
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Tests using global test environments.
+
+#include <stdlib.h>
+#include <stdio.h>
+#include "gtest/gtest.h"
+
+#define GTEST_IMPLEMENTATION_ 1 // Required for the next #include.
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+GTEST_DECLARE_string_(filter);
+}
+
+namespace {
+
+enum FailureType {
+ NO_FAILURE, NON_FATAL_FAILURE, FATAL_FAILURE
+};
+
+// For testing using global test environments.
+class MyEnvironment : public testing::Environment {
+ public:
+ MyEnvironment() { Reset(); }
+
+ // Depending on the value of failure_in_set_up_, SetUp() will
+ // generate a non-fatal failure, generate a fatal failure, or
+ // succeed.
+ virtual void SetUp() {
+ set_up_was_run_ = true;
+
+ switch (failure_in_set_up_) {
+ case NON_FATAL_FAILURE:
+ ADD_FAILURE() << "Expected non-fatal failure in global set-up.";
+ break;
+ case FATAL_FAILURE:
+ FAIL() << "Expected fatal failure in global set-up.";
+ break;
+ default:
+ break;
+ }
+ }
+
+ // Generates a non-fatal failure.
+ virtual void TearDown() {
+ tear_down_was_run_ = true;
+ ADD_FAILURE() << "Expected non-fatal failure in global tear-down.";
+ }
+
+ // Resets the state of the environment s.t. it can be reused.
+ void Reset() {
+ failure_in_set_up_ = NO_FAILURE;
+ set_up_was_run_ = false;
+ tear_down_was_run_ = false;
+ }
+
+ // We call this function to set the type of failure SetUp() should
+ // generate.
+ void set_failure_in_set_up(FailureType type) {
+ failure_in_set_up_ = type;
+ }
+
+ // Was SetUp() run?
+ bool set_up_was_run() const { return set_up_was_run_; }
+
+ // Was TearDown() run?
+ bool tear_down_was_run() const { return tear_down_was_run_; }
+ private:
+ FailureType failure_in_set_up_;
+ bool set_up_was_run_;
+ bool tear_down_was_run_;
+};
+
+// Was the TEST run?
+bool test_was_run;
+
+// The sole purpose of this TEST is to enable us to check whether it
+// was run.
+TEST(FooTest, Bar) {
+ test_was_run = true;
+}
+
+// Prints the message and aborts the program if condition is false.
+void Check(bool condition, const char* msg) {
+ if (!condition) {
+ printf("FAILED: %s\n", msg);
+ testing::internal::posix::Abort();
+ }
+}
+
+// Runs the tests. Return true iff successful.
+//
+// The 'failure' parameter specifies the type of failure that should
+// be generated by the global set-up.
+int RunAllTests(MyEnvironment* env, FailureType failure) {
+ env->Reset();
+ env->set_failure_in_set_up(failure);
+ test_was_run = false;
+ testing::internal::GetUnitTestImpl()->ClearAdHocTestResult();
+ return RUN_ALL_TESTS();
+}
+
+} // namespace
+
+int main(int argc, char **argv) {
+ testing::InitGoogleTest(&argc, argv);
+
+ // Registers a global test environment, and verifies that the
+ // registration function returns its argument.
+ MyEnvironment* const env = new MyEnvironment;
+ Check(testing::AddGlobalTestEnvironment(env) == env,
+ "AddGlobalTestEnvironment() should return its argument.");
+
+ // Verifies that RUN_ALL_TESTS() runs the tests when the global
+ // set-up is successful.
+ Check(RunAllTests(env, NO_FAILURE) != 0,
+ "RUN_ALL_TESTS() should return non-zero, as the global tear-down "
+ "should generate a failure.");
+ Check(test_was_run,
+ "The tests should run, as the global set-up should generate no "
+ "failure");
+ Check(env->tear_down_was_run(),
+ "The global tear-down should run, as the global set-up was run.");
+
+ // Verifies that RUN_ALL_TESTS() runs the tests when the global
+ // set-up generates no fatal failure.
+ Check(RunAllTests(env, NON_FATAL_FAILURE) != 0,
+ "RUN_ALL_TESTS() should return non-zero, as both the global set-up "
+ "and the global tear-down should generate a non-fatal failure.");
+ Check(test_was_run,
+ "The tests should run, as the global set-up should generate no "
+ "fatal failure.");
+ Check(env->tear_down_was_run(),
+ "The global tear-down should run, as the global set-up was run.");
+
+ // Verifies that RUN_ALL_TESTS() runs no test when the global set-up
+ // generates a fatal failure.
+ Check(RunAllTests(env, FATAL_FAILURE) != 0,
+ "RUN_ALL_TESTS() should return non-zero, as the global set-up "
+ "should generate a fatal failure.");
+ Check(!test_was_run,
+ "The tests should not run, as the global set-up should generate "
+ "a fatal failure.");
+ Check(env->tear_down_was_run(),
+ "The global tear-down should run, as the global set-up was run.");
+
+ // Verifies that RUN_ALL_TESTS() doesn't do global set-up or
+ // tear-down when there is no test to run.
+ testing::GTEST_FLAG(filter) = "-*";
+ Check(RunAllTests(env, NO_FAILURE) == 0,
+ "RUN_ALL_TESTS() should return zero, as there is no test to run.");
+ Check(!env->set_up_was_run(),
+ "The global set-up should not run, as there is no test to run.");
+ Check(!env->tear_down_was_run(),
+ "The global tear-down should not run, "
+ "as the global set-up was not run.");
+
+ printf("PASS\n");
+ return 0;
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_filter_unittest.py b/libvpx/third_party/googletest/src/test/gtest_filter_unittest.py
new file mode 100755
index 0000000..0d1a770
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_filter_unittest.py
@@ -0,0 +1,633 @@
+#!/usr/bin/env python
+#
+# Copyright 2005 Google Inc. All Rights Reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Unit test for Google Test test filters.
+
+A user can specify which test(s) in a Google Test program to run via either
+the GTEST_FILTER environment variable or the --gtest_filter flag.
+This script tests such functionality by invoking
+gtest_filter_unittest_ (a program written with Google Test) with different
+environments and command line flags.
+
+Note that test sharding may also influence which tests are filtered. Therefore,
+we test that here also.
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import re
+import sets
+import sys
+
+import gtest_test_utils
+
+# Constants.
+
+# Checks if this platform can pass empty environment variables to child
+# processes. We set an env variable to an empty string and invoke a python
+# script in a subprocess to print whether the variable is STILL in
+# os.environ. We then use 'eval' to parse the child's output so that an
+# exception is thrown if the input is anything other than 'True' nor 'False'.
+os.environ['EMPTY_VAR'] = ''
+child = gtest_test_utils.Subprocess(
+ [sys.executable, '-c', 'import os; print \'EMPTY_VAR\' in os.environ'])
+CAN_PASS_EMPTY_ENV = eval(child.output)
+
+
+# Check if this platform can unset environment variables in child processes.
+# We set an env variable to a non-empty string, unset it, and invoke
+# a python script in a subprocess to print whether the variable
+# is NO LONGER in os.environ.
+# We use 'eval' to parse the child's output so that an exception
+# is thrown if the input is neither 'True' nor 'False'.
+os.environ['UNSET_VAR'] = 'X'
+del os.environ['UNSET_VAR']
+child = gtest_test_utils.Subprocess(
+ [sys.executable, '-c', 'import os; print \'UNSET_VAR\' not in os.environ'])
+CAN_UNSET_ENV = eval(child.output)
+
+
+# Checks if we should test with an empty filter. This doesn't
+# make sense on platforms that cannot pass empty env variables (Win32)
+# and on platforms that cannot unset variables (since we cannot tell
+# the difference between "" and NULL -- Borland and Solaris < 5.10)
+CAN_TEST_EMPTY_FILTER = (CAN_PASS_EMPTY_ENV and CAN_UNSET_ENV)
+
+
+# The environment variable for specifying the test filters.
+FILTER_ENV_VAR = 'GTEST_FILTER'
+
+# The environment variables for test sharding.
+TOTAL_SHARDS_ENV_VAR = 'GTEST_TOTAL_SHARDS'
+SHARD_INDEX_ENV_VAR = 'GTEST_SHARD_INDEX'
+SHARD_STATUS_FILE_ENV_VAR = 'GTEST_SHARD_STATUS_FILE'
+
+# The command line flag for specifying the test filters.
+FILTER_FLAG = 'gtest_filter'
+
+# The command line flag for including disabled tests.
+ALSO_RUN_DISABED_TESTS_FLAG = 'gtest_also_run_disabled_tests'
+
+# Command to run the gtest_filter_unittest_ program.
+COMMAND = gtest_test_utils.GetTestExecutablePath('gtest_filter_unittest_')
+
+# Regex for determining whether parameterized tests are enabled in the binary.
+PARAM_TEST_REGEX = re.compile(r'/ParamTest')
+
+# Regex for parsing test case names from Google Test's output.
+TEST_CASE_REGEX = re.compile(r'^\[\-+\] \d+ tests? from (\w+(/\w+)?)')
+
+# Regex for parsing test names from Google Test's output.
+TEST_REGEX = re.compile(r'^\[\s*RUN\s*\].*\.(\w+(/\w+)?)')
+
+# The command line flag to tell Google Test to output the list of tests it
+# will run.
+LIST_TESTS_FLAG = '--gtest_list_tests'
+
+# Indicates whether Google Test supports death tests.
+SUPPORTS_DEATH_TESTS = 'HasDeathTest' in gtest_test_utils.Subprocess(
+ [COMMAND, LIST_TESTS_FLAG]).output
+
+# Full names of all tests in gtest_filter_unittests_.
+PARAM_TESTS = [
+ 'SeqP/ParamTest.TestX/0',
+ 'SeqP/ParamTest.TestX/1',
+ 'SeqP/ParamTest.TestY/0',
+ 'SeqP/ParamTest.TestY/1',
+ 'SeqQ/ParamTest.TestX/0',
+ 'SeqQ/ParamTest.TestX/1',
+ 'SeqQ/ParamTest.TestY/0',
+ 'SeqQ/ParamTest.TestY/1',
+ ]
+
+DISABLED_TESTS = [
+ 'BarTest.DISABLED_TestFour',
+ 'BarTest.DISABLED_TestFive',
+ 'BazTest.DISABLED_TestC',
+ 'DISABLED_FoobarTest.Test1',
+ 'DISABLED_FoobarTest.DISABLED_Test2',
+ 'DISABLED_FoobarbazTest.TestA',
+ ]
+
+if SUPPORTS_DEATH_TESTS:
+ DEATH_TESTS = [
+ 'HasDeathTest.Test1',
+ 'HasDeathTest.Test2',
+ ]
+else:
+ DEATH_TESTS = []
+
+# All the non-disabled tests.
+ACTIVE_TESTS = [
+ 'FooTest.Abc',
+ 'FooTest.Xyz',
+
+ 'BarTest.TestOne',
+ 'BarTest.TestTwo',
+ 'BarTest.TestThree',
+
+ 'BazTest.TestOne',
+ 'BazTest.TestA',
+ 'BazTest.TestB',
+ ] + DEATH_TESTS + PARAM_TESTS
+
+param_tests_present = None
+
+# Utilities.
+
+environ = os.environ.copy()
+
+
+def SetEnvVar(env_var, value):
+ """Sets the env variable to 'value'; unsets it when 'value' is None."""
+
+ if value is not None:
+ environ[env_var] = value
+ elif env_var in environ:
+ del environ[env_var]
+
+
+def RunAndReturnOutput(args = None):
+ """Runs the test program and returns its output."""
+
+ return gtest_test_utils.Subprocess([COMMAND] + (args or []),
+ env=environ).output
+
+
+def RunAndExtractTestList(args = None):
+ """Runs the test program and returns its exit code and a list of tests run."""
+
+ p = gtest_test_utils.Subprocess([COMMAND] + (args or []), env=environ)
+ tests_run = []
+ test_case = ''
+ test = ''
+ for line in p.output.split('\n'):
+ match = TEST_CASE_REGEX.match(line)
+ if match is not None:
+ test_case = match.group(1)
+ else:
+ match = TEST_REGEX.match(line)
+ if match is not None:
+ test = match.group(1)
+ tests_run.append(test_case + '.' + test)
+ return (tests_run, p.exit_code)
+
+
+def InvokeWithModifiedEnv(extra_env, function, *args, **kwargs):
+ """Runs the given function and arguments in a modified environment."""
+ try:
+ original_env = environ.copy()
+ environ.update(extra_env)
+ return function(*args, **kwargs)
+ finally:
+ environ.clear()
+ environ.update(original_env)
+
+
+def RunWithSharding(total_shards, shard_index, command):
+ """Runs a test program shard and returns exit code and a list of tests run."""
+
+ extra_env = {SHARD_INDEX_ENV_VAR: str(shard_index),
+ TOTAL_SHARDS_ENV_VAR: str(total_shards)}
+ return InvokeWithModifiedEnv(extra_env, RunAndExtractTestList, command)
+
+# The unit test.
+
+
+class GTestFilterUnitTest(gtest_test_utils.TestCase):
+ """Tests the env variable or the command line flag to filter tests."""
+
+ # Utilities.
+
+ def AssertSetEqual(self, lhs, rhs):
+ """Asserts that two sets are equal."""
+
+ for elem in lhs:
+ self.assert_(elem in rhs, '%s in %s' % (elem, rhs))
+
+ for elem in rhs:
+ self.assert_(elem in lhs, '%s in %s' % (elem, lhs))
+
+ def AssertPartitionIsValid(self, set_var, list_of_sets):
+ """Asserts that list_of_sets is a valid partition of set_var."""
+
+ full_partition = []
+ for slice_var in list_of_sets:
+ full_partition.extend(slice_var)
+ self.assertEqual(len(set_var), len(full_partition))
+ self.assertEqual(sets.Set(set_var), sets.Set(full_partition))
+
+ def AdjustForParameterizedTests(self, tests_to_run):
+ """Adjust tests_to_run in case value parameterized tests are disabled."""
+
+ global param_tests_present
+ if not param_tests_present:
+ return list(sets.Set(tests_to_run) - sets.Set(PARAM_TESTS))
+ else:
+ return tests_to_run
+
+ def RunAndVerify(self, gtest_filter, tests_to_run):
+ """Checks that the binary runs correct set of tests for a given filter."""
+
+ tests_to_run = self.AdjustForParameterizedTests(tests_to_run)
+
+ # First, tests using the environment variable.
+
+ # Windows removes empty variables from the environment when passing it
+ # to a new process. This means it is impossible to pass an empty filter
+ # into a process using the environment variable. However, we can still
+ # test the case when the variable is not supplied (i.e., gtest_filter is
+ # None).
+ # pylint: disable-msg=C6403
+ if CAN_TEST_EMPTY_FILTER or gtest_filter != '':
+ SetEnvVar(FILTER_ENV_VAR, gtest_filter)
+ tests_run = RunAndExtractTestList()[0]
+ SetEnvVar(FILTER_ENV_VAR, None)
+ self.AssertSetEqual(tests_run, tests_to_run)
+ # pylint: enable-msg=C6403
+
+ # Next, tests using the command line flag.
+
+ if gtest_filter is None:
+ args = []
+ else:
+ args = ['--%s=%s' % (FILTER_FLAG, gtest_filter)]
+
+ tests_run = RunAndExtractTestList(args)[0]
+ self.AssertSetEqual(tests_run, tests_to_run)
+
+ def RunAndVerifyWithSharding(self, gtest_filter, total_shards, tests_to_run,
+ args=None, check_exit_0=False):
+ """Checks that binary runs correct tests for the given filter and shard.
+
+ Runs all shards of gtest_filter_unittest_ with the given filter, and
+ verifies that the right set of tests were run. The union of tests run
+ on each shard should be identical to tests_to_run, without duplicates.
+
+ Args:
+ gtest_filter: A filter to apply to the tests.
+ total_shards: A total number of shards to split test run into.
+ tests_to_run: A set of tests expected to run.
+ args : Arguments to pass to the to the test binary.
+ check_exit_0: When set to a true value, make sure that all shards
+ return 0.
+ """
+
+ tests_to_run = self.AdjustForParameterizedTests(tests_to_run)
+
+ # Windows removes empty variables from the environment when passing it
+ # to a new process. This means it is impossible to pass an empty filter
+ # into a process using the environment variable. However, we can still
+ # test the case when the variable is not supplied (i.e., gtest_filter is
+ # None).
+ # pylint: disable-msg=C6403
+ if CAN_TEST_EMPTY_FILTER or gtest_filter != '':
+ SetEnvVar(FILTER_ENV_VAR, gtest_filter)
+ partition = []
+ for i in range(0, total_shards):
+ (tests_run, exit_code) = RunWithSharding(total_shards, i, args)
+ if check_exit_0:
+ self.assertEqual(0, exit_code)
+ partition.append(tests_run)
+
+ self.AssertPartitionIsValid(tests_to_run, partition)
+ SetEnvVar(FILTER_ENV_VAR, None)
+ # pylint: enable-msg=C6403
+
+ def RunAndVerifyAllowingDisabled(self, gtest_filter, tests_to_run):
+ """Checks that the binary runs correct set of tests for the given filter.
+
+ Runs gtest_filter_unittest_ with the given filter, and enables
+ disabled tests. Verifies that the right set of tests were run.
+
+ Args:
+ gtest_filter: A filter to apply to the tests.
+ tests_to_run: A set of tests expected to run.
+ """
+
+ tests_to_run = self.AdjustForParameterizedTests(tests_to_run)
+
+ # Construct the command line.
+ args = ['--%s' % ALSO_RUN_DISABED_TESTS_FLAG]
+ if gtest_filter is not None:
+ args.append('--%s=%s' % (FILTER_FLAG, gtest_filter))
+
+ tests_run = RunAndExtractTestList(args)[0]
+ self.AssertSetEqual(tests_run, tests_to_run)
+
+ def setUp(self):
+ """Sets up test case.
+
+ Determines whether value-parameterized tests are enabled in the binary and
+ sets the flags accordingly.
+ """
+
+ global param_tests_present
+ if param_tests_present is None:
+ param_tests_present = PARAM_TEST_REGEX.search(
+ RunAndReturnOutput()) is not None
+
+ def testDefaultBehavior(self):
+ """Tests the behavior of not specifying the filter."""
+
+ self.RunAndVerify(None, ACTIVE_TESTS)
+
+ def testDefaultBehaviorWithShards(self):
+ """Tests the behavior without the filter, with sharding enabled."""
+
+ self.RunAndVerifyWithSharding(None, 1, ACTIVE_TESTS)
+ self.RunAndVerifyWithSharding(None, 2, ACTIVE_TESTS)
+ self.RunAndVerifyWithSharding(None, len(ACTIVE_TESTS) - 1, ACTIVE_TESTS)
+ self.RunAndVerifyWithSharding(None, len(ACTIVE_TESTS), ACTIVE_TESTS)
+ self.RunAndVerifyWithSharding(None, len(ACTIVE_TESTS) + 1, ACTIVE_TESTS)
+
+ def testEmptyFilter(self):
+ """Tests an empty filter."""
+
+ self.RunAndVerify('', [])
+ self.RunAndVerifyWithSharding('', 1, [])
+ self.RunAndVerifyWithSharding('', 2, [])
+
+ def testBadFilter(self):
+ """Tests a filter that matches nothing."""
+
+ self.RunAndVerify('BadFilter', [])
+ self.RunAndVerifyAllowingDisabled('BadFilter', [])
+
+ def testFullName(self):
+ """Tests filtering by full name."""
+
+ self.RunAndVerify('FooTest.Xyz', ['FooTest.Xyz'])
+ self.RunAndVerifyAllowingDisabled('FooTest.Xyz', ['FooTest.Xyz'])
+ self.RunAndVerifyWithSharding('FooTest.Xyz', 5, ['FooTest.Xyz'])
+
+ def testUniversalFilters(self):
+ """Tests filters that match everything."""
+
+ self.RunAndVerify('*', ACTIVE_TESTS)
+ self.RunAndVerify('*.*', ACTIVE_TESTS)
+ self.RunAndVerifyWithSharding('*.*', len(ACTIVE_TESTS) - 3, ACTIVE_TESTS)
+ self.RunAndVerifyAllowingDisabled('*', ACTIVE_TESTS + DISABLED_TESTS)
+ self.RunAndVerifyAllowingDisabled('*.*', ACTIVE_TESTS + DISABLED_TESTS)
+
+ def testFilterByTestCase(self):
+ """Tests filtering by test case name."""
+
+ self.RunAndVerify('FooTest.*', ['FooTest.Abc', 'FooTest.Xyz'])
+
+ BAZ_TESTS = ['BazTest.TestOne', 'BazTest.TestA', 'BazTest.TestB']
+ self.RunAndVerify('BazTest.*', BAZ_TESTS)
+ self.RunAndVerifyAllowingDisabled('BazTest.*',
+ BAZ_TESTS + ['BazTest.DISABLED_TestC'])
+
+ def testFilterByTest(self):
+ """Tests filtering by test name."""
+
+ self.RunAndVerify('*.TestOne', ['BarTest.TestOne', 'BazTest.TestOne'])
+
+ def testFilterDisabledTests(self):
+ """Select only the disabled tests to run."""
+
+ self.RunAndVerify('DISABLED_FoobarTest.Test1', [])
+ self.RunAndVerifyAllowingDisabled('DISABLED_FoobarTest.Test1',
+ ['DISABLED_FoobarTest.Test1'])
+
+ self.RunAndVerify('*DISABLED_*', [])
+ self.RunAndVerifyAllowingDisabled('*DISABLED_*', DISABLED_TESTS)
+
+ self.RunAndVerify('*.DISABLED_*', [])
+ self.RunAndVerifyAllowingDisabled('*.DISABLED_*', [
+ 'BarTest.DISABLED_TestFour',
+ 'BarTest.DISABLED_TestFive',
+ 'BazTest.DISABLED_TestC',
+ 'DISABLED_FoobarTest.DISABLED_Test2',
+ ])
+
+ self.RunAndVerify('DISABLED_*', [])
+ self.RunAndVerifyAllowingDisabled('DISABLED_*', [
+ 'DISABLED_FoobarTest.Test1',
+ 'DISABLED_FoobarTest.DISABLED_Test2',
+ 'DISABLED_FoobarbazTest.TestA',
+ ])
+
+ def testWildcardInTestCaseName(self):
+ """Tests using wildcard in the test case name."""
+
+ self.RunAndVerify('*a*.*', [
+ 'BarTest.TestOne',
+ 'BarTest.TestTwo',
+ 'BarTest.TestThree',
+
+ 'BazTest.TestOne',
+ 'BazTest.TestA',
+ 'BazTest.TestB', ] + DEATH_TESTS + PARAM_TESTS)
+
+ def testWildcardInTestName(self):
+ """Tests using wildcard in the test name."""
+
+ self.RunAndVerify('*.*A*', ['FooTest.Abc', 'BazTest.TestA'])
+
+ def testFilterWithoutDot(self):
+ """Tests a filter that has no '.' in it."""
+
+ self.RunAndVerify('*z*', [
+ 'FooTest.Xyz',
+
+ 'BazTest.TestOne',
+ 'BazTest.TestA',
+ 'BazTest.TestB',
+ ])
+
+ def testTwoPatterns(self):
+ """Tests filters that consist of two patterns."""
+
+ self.RunAndVerify('Foo*.*:*A*', [
+ 'FooTest.Abc',
+ 'FooTest.Xyz',
+
+ 'BazTest.TestA',
+ ])
+
+ # An empty pattern + a non-empty one
+ self.RunAndVerify(':*A*', ['FooTest.Abc', 'BazTest.TestA'])
+
+ def testThreePatterns(self):
+ """Tests filters that consist of three patterns."""
+
+ self.RunAndVerify('*oo*:*A*:*One', [
+ 'FooTest.Abc',
+ 'FooTest.Xyz',
+
+ 'BarTest.TestOne',
+
+ 'BazTest.TestOne',
+ 'BazTest.TestA',
+ ])
+
+ # The 2nd pattern is empty.
+ self.RunAndVerify('*oo*::*One', [
+ 'FooTest.Abc',
+ 'FooTest.Xyz',
+
+ 'BarTest.TestOne',
+
+ 'BazTest.TestOne',
+ ])
+
+ # The last 2 patterns are empty.
+ self.RunAndVerify('*oo*::', [
+ 'FooTest.Abc',
+ 'FooTest.Xyz',
+ ])
+
+ def testNegativeFilters(self):
+ self.RunAndVerify('*-BazTest.TestOne', [
+ 'FooTest.Abc',
+ 'FooTest.Xyz',
+
+ 'BarTest.TestOne',
+ 'BarTest.TestTwo',
+ 'BarTest.TestThree',
+
+ 'BazTest.TestA',
+ 'BazTest.TestB',
+ ] + DEATH_TESTS + PARAM_TESTS)
+
+ self.RunAndVerify('*-FooTest.Abc:BazTest.*', [
+ 'FooTest.Xyz',
+
+ 'BarTest.TestOne',
+ 'BarTest.TestTwo',
+ 'BarTest.TestThree',
+ ] + DEATH_TESTS + PARAM_TESTS)
+
+ self.RunAndVerify('BarTest.*-BarTest.TestOne', [
+ 'BarTest.TestTwo',
+ 'BarTest.TestThree',
+ ])
+
+ # Tests without leading '*'.
+ self.RunAndVerify('-FooTest.Abc:FooTest.Xyz:BazTest.*', [
+ 'BarTest.TestOne',
+ 'BarTest.TestTwo',
+ 'BarTest.TestThree',
+ ] + DEATH_TESTS + PARAM_TESTS)
+
+ # Value parameterized tests.
+ self.RunAndVerify('*/*', PARAM_TESTS)
+
+ # Value parameterized tests filtering by the sequence name.
+ self.RunAndVerify('SeqP/*', [
+ 'SeqP/ParamTest.TestX/0',
+ 'SeqP/ParamTest.TestX/1',
+ 'SeqP/ParamTest.TestY/0',
+ 'SeqP/ParamTest.TestY/1',
+ ])
+
+ # Value parameterized tests filtering by the test name.
+ self.RunAndVerify('*/0', [
+ 'SeqP/ParamTest.TestX/0',
+ 'SeqP/ParamTest.TestY/0',
+ 'SeqQ/ParamTest.TestX/0',
+ 'SeqQ/ParamTest.TestY/0',
+ ])
+
+ def testFlagOverridesEnvVar(self):
+ """Tests that the filter flag overrides the filtering env. variable."""
+
+ SetEnvVar(FILTER_ENV_VAR, 'Foo*')
+ args = ['--%s=%s' % (FILTER_FLAG, '*One')]
+ tests_run = RunAndExtractTestList(args)[0]
+ SetEnvVar(FILTER_ENV_VAR, None)
+
+ self.AssertSetEqual(tests_run, ['BarTest.TestOne', 'BazTest.TestOne'])
+
+ def testShardStatusFileIsCreated(self):
+ """Tests that the shard file is created if specified in the environment."""
+
+ shard_status_file = os.path.join(gtest_test_utils.GetTempDir(),
+ 'shard_status_file')
+ self.assert_(not os.path.exists(shard_status_file))
+
+ extra_env = {SHARD_STATUS_FILE_ENV_VAR: shard_status_file}
+ try:
+ InvokeWithModifiedEnv(extra_env, RunAndReturnOutput)
+ finally:
+ self.assert_(os.path.exists(shard_status_file))
+ os.remove(shard_status_file)
+
+ def testShardStatusFileIsCreatedWithListTests(self):
+ """Tests that the shard file is created with the "list_tests" flag."""
+
+ shard_status_file = os.path.join(gtest_test_utils.GetTempDir(),
+ 'shard_status_file2')
+ self.assert_(not os.path.exists(shard_status_file))
+
+ extra_env = {SHARD_STATUS_FILE_ENV_VAR: shard_status_file}
+ try:
+ output = InvokeWithModifiedEnv(extra_env,
+ RunAndReturnOutput,
+ [LIST_TESTS_FLAG])
+ finally:
+ # This assertion ensures that Google Test enumerated the tests as
+ # opposed to running them.
+ self.assert_('[==========]' not in output,
+ 'Unexpected output during test enumeration.\n'
+ 'Please ensure that LIST_TESTS_FLAG is assigned the\n'
+ 'correct flag value for listing Google Test tests.')
+
+ self.assert_(os.path.exists(shard_status_file))
+ os.remove(shard_status_file)
+
+ if SUPPORTS_DEATH_TESTS:
+ def testShardingWorksWithDeathTests(self):
+ """Tests integration with death tests and sharding."""
+
+ gtest_filter = 'HasDeathTest.*:SeqP/*'
+ expected_tests = [
+ 'HasDeathTest.Test1',
+ 'HasDeathTest.Test2',
+
+ 'SeqP/ParamTest.TestX/0',
+ 'SeqP/ParamTest.TestX/1',
+ 'SeqP/ParamTest.TestY/0',
+ 'SeqP/ParamTest.TestY/1',
+ ]
+
+ for flag in ['--gtest_death_test_style=threadsafe',
+ '--gtest_death_test_style=fast']:
+ self.RunAndVerifyWithSharding(gtest_filter, 3, expected_tests,
+ check_exit_0=True, args=[flag])
+ self.RunAndVerifyWithSharding(gtest_filter, 5, expected_tests,
+ check_exit_0=True, args=[flag])
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_filter_unittest_.cc b/libvpx/third_party/googletest/src/test/gtest_filter_unittest_.cc
new file mode 100644
index 0000000..77deffc
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_filter_unittest_.cc
@@ -0,0 +1,140 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Unit test for Google Test test filters.
+//
+// A user can specify which test(s) in a Google Test program to run via
+// either the GTEST_FILTER environment variable or the --gtest_filter
+// flag. This is used for testing such functionality.
+//
+// The program will be invoked from a Python unit test. Don't run it
+// directly.
+
+#include "gtest/gtest.h"
+
+namespace {
+
+// Test case FooTest.
+
+class FooTest : public testing::Test {
+};
+
+TEST_F(FooTest, Abc) {
+}
+
+TEST_F(FooTest, Xyz) {
+ FAIL() << "Expected failure.";
+}
+
+// Test case BarTest.
+
+TEST(BarTest, TestOne) {
+}
+
+TEST(BarTest, TestTwo) {
+}
+
+TEST(BarTest, TestThree) {
+}
+
+TEST(BarTest, DISABLED_TestFour) {
+ FAIL() << "Expected failure.";
+}
+
+TEST(BarTest, DISABLED_TestFive) {
+ FAIL() << "Expected failure.";
+}
+
+// Test case BazTest.
+
+TEST(BazTest, TestOne) {
+ FAIL() << "Expected failure.";
+}
+
+TEST(BazTest, TestA) {
+}
+
+TEST(BazTest, TestB) {
+}
+
+TEST(BazTest, DISABLED_TestC) {
+ FAIL() << "Expected failure.";
+}
+
+// Test case HasDeathTest
+
+TEST(HasDeathTest, Test1) {
+ EXPECT_DEATH_IF_SUPPORTED(exit(1), ".*");
+}
+
+// We need at least two death tests to make sure that the all death tests
+// aren't on the first shard.
+TEST(HasDeathTest, Test2) {
+ EXPECT_DEATH_IF_SUPPORTED(exit(1), ".*");
+}
+
+// Test case FoobarTest
+
+TEST(DISABLED_FoobarTest, Test1) {
+ FAIL() << "Expected failure.";
+}
+
+TEST(DISABLED_FoobarTest, DISABLED_Test2) {
+ FAIL() << "Expected failure.";
+}
+
+// Test case FoobarbazTest
+
+TEST(DISABLED_FoobarbazTest, TestA) {
+ FAIL() << "Expected failure.";
+}
+
+#if GTEST_HAS_PARAM_TEST
+class ParamTest : public testing::TestWithParam<int> {
+};
+
+TEST_P(ParamTest, TestX) {
+}
+
+TEST_P(ParamTest, TestY) {
+}
+
+INSTANTIATE_TEST_CASE_P(SeqP, ParamTest, testing::Values(1, 2));
+INSTANTIATE_TEST_CASE_P(SeqQ, ParamTest, testing::Values(5, 6));
+#endif // GTEST_HAS_PARAM_TEST
+
+} // namespace
+
+int main(int argc, char **argv) {
+ ::testing::InitGoogleTest(&argc, argv);
+
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_help_test.py b/libvpx/third_party/googletest/src/test/gtest_help_test.py
new file mode 100755
index 0000000..093c838
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_help_test.py
@@ -0,0 +1,172 @@
+#!/usr/bin/env python
+#
+# Copyright 2009, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Tests the --help flag of Google C++ Testing Framework.
+
+SYNOPSIS
+ gtest_help_test.py --build_dir=BUILD/DIR
+ # where BUILD/DIR contains the built gtest_help_test_ file.
+ gtest_help_test.py
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import re
+import gtest_test_utils
+
+
+IS_LINUX = os.name == 'posix' and os.uname()[0] == 'Linux'
+IS_WINDOWS = os.name == 'nt'
+
+PROGRAM_PATH = gtest_test_utils.GetTestExecutablePath('gtest_help_test_')
+FLAG_PREFIX = '--gtest_'
+DEATH_TEST_STYLE_FLAG = FLAG_PREFIX + 'death_test_style'
+STREAM_RESULT_TO_FLAG = FLAG_PREFIX + 'stream_result_to'
+UNKNOWN_FLAG = FLAG_PREFIX + 'unknown_flag_for_testing'
+LIST_TESTS_FLAG = FLAG_PREFIX + 'list_tests'
+INCORRECT_FLAG_VARIANTS = [re.sub('^--', '-', LIST_TESTS_FLAG),
+ re.sub('^--', '/', LIST_TESTS_FLAG),
+ re.sub('_', '-', LIST_TESTS_FLAG)]
+INTERNAL_FLAG_FOR_TESTING = FLAG_PREFIX + 'internal_flag_for_testing'
+
+SUPPORTS_DEATH_TESTS = "DeathTest" in gtest_test_utils.Subprocess(
+ [PROGRAM_PATH, LIST_TESTS_FLAG]).output
+
+# The help message must match this regex.
+HELP_REGEX = re.compile(
+ FLAG_PREFIX + r'list_tests.*' +
+ FLAG_PREFIX + r'filter=.*' +
+ FLAG_PREFIX + r'also_run_disabled_tests.*' +
+ FLAG_PREFIX + r'repeat=.*' +
+ FLAG_PREFIX + r'shuffle.*' +
+ FLAG_PREFIX + r'random_seed=.*' +
+ FLAG_PREFIX + r'color=.*' +
+ FLAG_PREFIX + r'print_time.*' +
+ FLAG_PREFIX + r'output=.*' +
+ FLAG_PREFIX + r'break_on_failure.*' +
+ FLAG_PREFIX + r'throw_on_failure.*' +
+ FLAG_PREFIX + r'catch_exceptions=0.*',
+ re.DOTALL)
+
+
+def RunWithFlag(flag):
+ """Runs gtest_help_test_ with the given flag.
+
+ Returns:
+ the exit code and the text output as a tuple.
+ Args:
+ flag: the command-line flag to pass to gtest_help_test_, or None.
+ """
+
+ if flag is None:
+ command = [PROGRAM_PATH]
+ else:
+ command = [PROGRAM_PATH, flag]
+ child = gtest_test_utils.Subprocess(command)
+ return child.exit_code, child.output
+
+
+class GTestHelpTest(gtest_test_utils.TestCase):
+ """Tests the --help flag and its equivalent forms."""
+
+ def TestHelpFlag(self, flag):
+ """Verifies correct behavior when help flag is specified.
+
+ The right message must be printed and the tests must
+ skipped when the given flag is specified.
+
+ Args:
+ flag: A flag to pass to the binary or None.
+ """
+
+ exit_code, output = RunWithFlag(flag)
+ self.assertEquals(0, exit_code)
+ self.assert_(HELP_REGEX.search(output), output)
+
+ if IS_LINUX:
+ self.assert_(STREAM_RESULT_TO_FLAG in output, output)
+ else:
+ self.assert_(STREAM_RESULT_TO_FLAG not in output, output)
+
+ if SUPPORTS_DEATH_TESTS and not IS_WINDOWS:
+ self.assert_(DEATH_TEST_STYLE_FLAG in output, output)
+ else:
+ self.assert_(DEATH_TEST_STYLE_FLAG not in output, output)
+
+ def TestNonHelpFlag(self, flag):
+ """Verifies correct behavior when no help flag is specified.
+
+ Verifies that when no help flag is specified, the tests are run
+ and the help message is not printed.
+
+ Args:
+ flag: A flag to pass to the binary or None.
+ """
+
+ exit_code, output = RunWithFlag(flag)
+ self.assert_(exit_code != 0)
+ self.assert_(not HELP_REGEX.search(output), output)
+
+ def testPrintsHelpWithFullFlag(self):
+ self.TestHelpFlag('--help')
+
+ def testPrintsHelpWithShortFlag(self):
+ self.TestHelpFlag('-h')
+
+ def testPrintsHelpWithQuestionFlag(self):
+ self.TestHelpFlag('-?')
+
+ def testPrintsHelpWithWindowsStyleQuestionFlag(self):
+ self.TestHelpFlag('/?')
+
+ def testPrintsHelpWithUnrecognizedGoogleTestFlag(self):
+ self.TestHelpFlag(UNKNOWN_FLAG)
+
+ def testPrintsHelpWithIncorrectFlagStyle(self):
+ for incorrect_flag in INCORRECT_FLAG_VARIANTS:
+ self.TestHelpFlag(incorrect_flag)
+
+ def testRunsTestsWithoutHelpFlag(self):
+ """Verifies that when no help flag is specified, the tests are run
+ and the help message is not printed."""
+
+ self.TestNonHelpFlag(None)
+
+ def testRunsTestsWithGtestInternalFlag(self):
+ """Verifies that the tests are run and no help message is printed when
+ a flag starting with Google Test prefix and 'internal_' is supplied."""
+
+ self.TestNonHelpFlag(INTERNAL_FLAG_FOR_TESTING)
+
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_help_test_.cc b/libvpx/third_party/googletest/src/test/gtest_help_test_.cc
new file mode 100644
index 0000000..31f78c2
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_help_test_.cc
@@ -0,0 +1,46 @@
+// Copyright 2009, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// This program is meant to be run by gtest_help_test.py. Do not run
+// it directly.
+
+#include "gtest/gtest.h"
+
+// When a help flag is specified, this program should skip the tests
+// and exit with 0; otherwise the following test will be executed,
+// causing this program to exit with a non-zero code.
+TEST(HelpFlagTest, ShouldNotBeRun) {
+ ASSERT_TRUE(false) << "Tests shouldn't be run when --help is specified.";
+}
+
+#if GTEST_HAS_DEATH_TEST
+TEST(DeathTest, UsedByPythonScriptToDetectSupportForDeathTestsInThisBinary) {}
+#endif
diff --git a/libvpx/third_party/googletest/src/test/gtest_list_tests_unittest.py b/libvpx/third_party/googletest/src/test/gtest_list_tests_unittest.py
new file mode 100755
index 0000000..ce8c3ef
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_list_tests_unittest.py
@@ -0,0 +1,177 @@
+#!/usr/bin/env python
+#
+# Copyright 2006, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Unit test for Google Test's --gtest_list_tests flag.
+
+A user can ask Google Test to list all tests by specifying the
+--gtest_list_tests flag. This script tests such functionality
+by invoking gtest_list_tests_unittest_ (a program written with
+Google Test) the command line flags.
+"""
+
+__author__ = 'phanna@google.com (Patrick Hanna)'
+
+import gtest_test_utils
+
+
+# Constants.
+
+# The command line flag for enabling/disabling listing all tests.
+LIST_TESTS_FLAG = 'gtest_list_tests'
+
+# Path to the gtest_list_tests_unittest_ program.
+EXE_PATH = gtest_test_utils.GetTestExecutablePath('gtest_list_tests_unittest_')
+
+# The expected output when running gtest_list_tests_unittest_ with
+# --gtest_list_tests
+EXPECTED_OUTPUT_NO_FILTER = """FooDeathTest.
+ Test1
+Foo.
+ Bar1
+ Bar2
+ DISABLED_Bar3
+Abc.
+ Xyz
+ Def
+FooBar.
+ Baz
+FooTest.
+ Test1
+ DISABLED_Test2
+ Test3
+"""
+
+# The expected output when running gtest_list_tests_unittest_ with
+# --gtest_list_tests and --gtest_filter=Foo*.
+EXPECTED_OUTPUT_FILTER_FOO = """FooDeathTest.
+ Test1
+Foo.
+ Bar1
+ Bar2
+ DISABLED_Bar3
+FooBar.
+ Baz
+FooTest.
+ Test1
+ DISABLED_Test2
+ Test3
+"""
+
+# Utilities.
+
+
+def Run(args):
+ """Runs gtest_list_tests_unittest_ and returns the list of tests printed."""
+
+ return gtest_test_utils.Subprocess([EXE_PATH] + args,
+ capture_stderr=False).output
+
+
+# The unit test.
+
+class GTestListTestsUnitTest(gtest_test_utils.TestCase):
+ """Tests using the --gtest_list_tests flag to list all tests."""
+
+ def RunAndVerify(self, flag_value, expected_output, other_flag):
+ """Runs gtest_list_tests_unittest_ and verifies that it prints
+ the correct tests.
+
+ Args:
+ flag_value: value of the --gtest_list_tests flag;
+ None if the flag should not be present.
+
+ expected_output: the expected output after running command;
+
+ other_flag: a different flag to be passed to command
+ along with gtest_list_tests;
+ None if the flag should not be present.
+ """
+
+ if flag_value is None:
+ flag = ''
+ flag_expression = 'not set'
+ elif flag_value == '0':
+ flag = '--%s=0' % LIST_TESTS_FLAG
+ flag_expression = '0'
+ else:
+ flag = '--%s' % LIST_TESTS_FLAG
+ flag_expression = '1'
+
+ args = [flag]
+
+ if other_flag is not None:
+ args += [other_flag]
+
+ output = Run(args)
+
+ msg = ('when %s is %s, the output of "%s" is "%s".' %
+ (LIST_TESTS_FLAG, flag_expression, ' '.join(args), output))
+
+ if expected_output is not None:
+ self.assert_(output == expected_output, msg)
+ else:
+ self.assert_(output != EXPECTED_OUTPUT_NO_FILTER, msg)
+
+ def testDefaultBehavior(self):
+ """Tests the behavior of the default mode."""
+
+ self.RunAndVerify(flag_value=None,
+ expected_output=None,
+ other_flag=None)
+
+ def testFlag(self):
+ """Tests using the --gtest_list_tests flag."""
+
+ self.RunAndVerify(flag_value='0',
+ expected_output=None,
+ other_flag=None)
+ self.RunAndVerify(flag_value='1',
+ expected_output=EXPECTED_OUTPUT_NO_FILTER,
+ other_flag=None)
+
+ def testOverrideNonFilterFlags(self):
+ """Tests that --gtest_list_tests overrides the non-filter flags."""
+
+ self.RunAndVerify(flag_value='1',
+ expected_output=EXPECTED_OUTPUT_NO_FILTER,
+ other_flag='--gtest_break_on_failure')
+
+ def testWithFilterFlags(self):
+ """Tests that --gtest_list_tests takes into account the
+ --gtest_filter flag."""
+
+ self.RunAndVerify(flag_value='1',
+ expected_output=EXPECTED_OUTPUT_FILTER_FOO,
+ other_flag='--gtest_filter=Foo*')
+
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_list_tests_unittest_.cc b/libvpx/third_party/googletest/src/test/gtest_list_tests_unittest_.cc
new file mode 100644
index 0000000..2b1d078
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_list_tests_unittest_.cc
@@ -0,0 +1,85 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: phanna@google.com (Patrick Hanna)
+
+// Unit test for Google Test's --gtest_list_tests flag.
+//
+// A user can ask Google Test to list all tests that will run
+// so that when using a filter, a user will know what
+// tests to look for. The tests will not be run after listing.
+//
+// This program will be invoked from a Python unit test.
+// Don't run it directly.
+
+#include "gtest/gtest.h"
+
+namespace {
+
+// Several different test cases and tests that will be listed.
+TEST(Foo, Bar1) {
+}
+
+TEST(Foo, Bar2) {
+}
+
+TEST(Foo, DISABLED_Bar3) {
+}
+
+TEST(Abc, Xyz) {
+}
+
+TEST(Abc, Def) {
+}
+
+TEST(FooBar, Baz) {
+}
+
+class FooTest : public testing::Test {
+};
+
+TEST_F(FooTest, Test1) {
+}
+
+TEST_F(FooTest, DISABLED_Test2) {
+}
+
+TEST_F(FooTest, Test3) {
+}
+
+TEST(FooDeathTest, Test1) {
+}
+
+} // namespace
+
+int main(int argc, char **argv) {
+ ::testing::InitGoogleTest(&argc, argv);
+
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_main_unittest.cc b/libvpx/third_party/googletest/src/test/gtest_main_unittest.cc
new file mode 100644
index 0000000..ecd9bb8
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_main_unittest.cc
@@ -0,0 +1,45 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "gtest/gtest.h"
+
+// Tests that we don't have to define main() when we link to
+// gtest_main instead of gtest.
+
+namespace {
+
+TEST(GTestMainTest, ShouldSucceed) {
+}
+
+} // namespace
+
+// We are using the main() function defined in src/gtest_main.cc, so
+// we don't define it here.
diff --git a/libvpx/third_party/googletest/src/test/gtest_no_test_unittest.cc b/libvpx/third_party/googletest/src/test/gtest_no_test_unittest.cc
new file mode 100644
index 0000000..e3a85f1
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_no_test_unittest.cc
@@ -0,0 +1,57 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// Tests that a Google Test program that has no test defined can run
+// successfully.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "gtest/gtest.h"
+
+
+int main(int argc, char **argv) {
+ testing::InitGoogleTest(&argc, argv);
+
+ // An ad-hoc assertion outside of all tests.
+ //
+ // This serves three purposes:
+ //
+ // 1. It verifies that an ad-hoc assertion can be executed even if
+ // no test is defined.
+ // 2. It verifies that a failed ad-hoc assertion causes the test
+ // program to fail.
+ // 3. We had a bug where the XML output won't be generated if an
+ // assertion is executed before RUN_ALL_TESTS() is called, even
+ // though --gtest_output=xml is specified. This makes sure the
+ // bug is fixed and doesn't regress.
+ EXPECT_EQ(1, 2);
+
+ // The above EXPECT_EQ() should cause RUN_ALL_TESTS() to return non-zero.
+ return RUN_ALL_TESTS() ? 0 : 1;
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_output_test.py b/libvpx/third_party/googletest/src/test/gtest_output_test.py
new file mode 100755
index 0000000..f409e2a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_output_test.py
@@ -0,0 +1,335 @@
+#!/usr/bin/env python
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Tests the text output of Google C++ Testing Framework.
+
+SYNOPSIS
+ gtest_output_test.py --build_dir=BUILD/DIR --gengolden
+ # where BUILD/DIR contains the built gtest_output_test_ file.
+ gtest_output_test.py --gengolden
+ gtest_output_test.py
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import re
+import sys
+import gtest_test_utils
+
+
+# The flag for generating the golden file
+GENGOLDEN_FLAG = '--gengolden'
+CATCH_EXCEPTIONS_ENV_VAR_NAME = 'GTEST_CATCH_EXCEPTIONS'
+
+IS_WINDOWS = os.name == 'nt'
+
+# TODO(vladl@google.com): remove the _lin suffix.
+GOLDEN_NAME = 'gtest_output_test_golden_lin.txt'
+
+PROGRAM_PATH = gtest_test_utils.GetTestExecutablePath('gtest_output_test_')
+
+# At least one command we exercise must not have the
+# --gtest_internal_skip_environment_and_ad_hoc_tests flag.
+COMMAND_LIST_TESTS = ({}, [PROGRAM_PATH, '--gtest_list_tests'])
+COMMAND_WITH_COLOR = ({}, [PROGRAM_PATH, '--gtest_color=yes'])
+COMMAND_WITH_TIME = ({}, [PROGRAM_PATH,
+ '--gtest_print_time',
+ '--gtest_internal_skip_environment_and_ad_hoc_tests',
+ '--gtest_filter=FatalFailureTest.*:LoggingTest.*'])
+COMMAND_WITH_DISABLED = (
+ {}, [PROGRAM_PATH,
+ '--gtest_also_run_disabled_tests',
+ '--gtest_internal_skip_environment_and_ad_hoc_tests',
+ '--gtest_filter=*DISABLED_*'])
+COMMAND_WITH_SHARDING = (
+ {'GTEST_SHARD_INDEX': '1', 'GTEST_TOTAL_SHARDS': '2'},
+ [PROGRAM_PATH,
+ '--gtest_internal_skip_environment_and_ad_hoc_tests',
+ '--gtest_filter=PassingTest.*'])
+
+GOLDEN_PATH = os.path.join(gtest_test_utils.GetSourceDir(), GOLDEN_NAME)
+
+
+def ToUnixLineEnding(s):
+ """Changes all Windows/Mac line endings in s to UNIX line endings."""
+
+ return s.replace('\r\n', '\n').replace('\r', '\n')
+
+
+def RemoveLocations(test_output):
+ """Removes all file location info from a Google Test program's output.
+
+ Args:
+ test_output: the output of a Google Test program.
+
+ Returns:
+ output with all file location info (in the form of
+ 'DIRECTORY/FILE_NAME:LINE_NUMBER: 'or
+ 'DIRECTORY\\FILE_NAME(LINE_NUMBER): ') replaced by
+ 'FILE_NAME:#: '.
+ """
+
+ return re.sub(r'.*[/\\](.+)(\:\d+|\(\d+\))\: ', r'\1:#: ', test_output)
+
+
+def RemoveStackTraceDetails(output):
+ """Removes all stack traces from a Google Test program's output."""
+
+ # *? means "find the shortest string that matches".
+ return re.sub(r'Stack trace:(.|\n)*?\n\n',
+ 'Stack trace: (omitted)\n\n', output)
+
+
+def RemoveStackTraces(output):
+ """Removes all traces of stack traces from a Google Test program's output."""
+
+ # *? means "find the shortest string that matches".
+ return re.sub(r'Stack trace:(.|\n)*?\n\n', '', output)
+
+
+def RemoveTime(output):
+ """Removes all time information from a Google Test program's output."""
+
+ return re.sub(r'\(\d+ ms', '(? ms', output)
+
+
+def RemoveTypeInfoDetails(test_output):
+ """Removes compiler-specific type info from Google Test program's output.
+
+ Args:
+ test_output: the output of a Google Test program.
+
+ Returns:
+ output with type information normalized to canonical form.
+ """
+
+ # some compilers output the name of type 'unsigned int' as 'unsigned'
+ return re.sub(r'unsigned int', 'unsigned', test_output)
+
+
+def NormalizeToCurrentPlatform(test_output):
+ """Normalizes platform specific output details for easier comparison."""
+
+ if IS_WINDOWS:
+ # Removes the color information that is not present on Windows.
+ test_output = re.sub('\x1b\\[(0;3\d)?m', '', test_output)
+ # Changes failure message headers into the Windows format.
+ test_output = re.sub(r': Failure\n', r': error: ', test_output)
+ # Changes file(line_number) to file:line_number.
+ test_output = re.sub(r'((\w|\.)+)\((\d+)\):', r'\1:\3:', test_output)
+
+ return test_output
+
+
+def RemoveTestCounts(output):
+ """Removes test counts from a Google Test program's output."""
+
+ output = re.sub(r'\d+ tests?, listed below',
+ '? tests, listed below', output)
+ output = re.sub(r'\d+ FAILED TESTS',
+ '? FAILED TESTS', output)
+ output = re.sub(r'\d+ tests? from \d+ test cases?',
+ '? tests from ? test cases', output)
+ output = re.sub(r'\d+ tests? from ([a-zA-Z_])',
+ r'? tests from \1', output)
+ return re.sub(r'\d+ tests?\.', '? tests.', output)
+
+
+def RemoveMatchingTests(test_output, pattern):
+ """Removes output of specified tests from a Google Test program's output.
+
+ This function strips not only the beginning and the end of a test but also
+ all output in between.
+
+ Args:
+ test_output: A string containing the test output.
+ pattern: A regex string that matches names of test cases or
+ tests to remove.
+
+ Returns:
+ Contents of test_output with tests whose names match pattern removed.
+ """
+
+ test_output = re.sub(
+ r'.*\[ RUN \] .*%s(.|\n)*?\[( FAILED | OK )\] .*%s.*\n' % (
+ pattern, pattern),
+ '',
+ test_output)
+ return re.sub(r'.*%s.*\n' % pattern, '', test_output)
+
+
+def NormalizeOutput(output):
+ """Normalizes output (the output of gtest_output_test_.exe)."""
+
+ output = ToUnixLineEnding(output)
+ output = RemoveLocations(output)
+ output = RemoveStackTraceDetails(output)
+ output = RemoveTime(output)
+ return output
+
+
+def GetShellCommandOutput(env_cmd):
+ """Runs a command in a sub-process, and returns its output in a string.
+
+ Args:
+ env_cmd: The shell command. A 2-tuple where element 0 is a dict of extra
+ environment variables to set, and element 1 is a string with
+ the command and any flags.
+
+ Returns:
+ A string with the command's combined standard and diagnostic output.
+ """
+
+ # Spawns cmd in a sub-process, and gets its standard I/O file objects.
+ # Set and save the environment properly.
+ environ = os.environ.copy()
+ environ.update(env_cmd[0])
+ p = gtest_test_utils.Subprocess(env_cmd[1], env=environ)
+
+ return p.output
+
+
+def GetCommandOutput(env_cmd):
+ """Runs a command and returns its output with all file location
+ info stripped off.
+
+ Args:
+ env_cmd: The shell command. A 2-tuple where element 0 is a dict of extra
+ environment variables to set, and element 1 is a string with
+ the command and any flags.
+ """
+
+ # Disables exception pop-ups on Windows.
+ environ, cmdline = env_cmd
+ environ = dict(environ) # Ensures we are modifying a copy.
+ environ[CATCH_EXCEPTIONS_ENV_VAR_NAME] = '1'
+ return NormalizeOutput(GetShellCommandOutput((environ, cmdline)))
+
+
+def GetOutputOfAllCommands():
+ """Returns concatenated output from several representative commands."""
+
+ return (GetCommandOutput(COMMAND_WITH_COLOR) +
+ GetCommandOutput(COMMAND_WITH_TIME) +
+ GetCommandOutput(COMMAND_WITH_DISABLED) +
+ GetCommandOutput(COMMAND_WITH_SHARDING))
+
+
+test_list = GetShellCommandOutput(COMMAND_LIST_TESTS)
+SUPPORTS_DEATH_TESTS = 'DeathTest' in test_list
+SUPPORTS_TYPED_TESTS = 'TypedTest' in test_list
+SUPPORTS_THREADS = 'ExpectFailureWithThreadsTest' in test_list
+SUPPORTS_STACK_TRACES = False
+
+CAN_GENERATE_GOLDEN_FILE = (SUPPORTS_DEATH_TESTS and
+ SUPPORTS_TYPED_TESTS and
+ SUPPORTS_THREADS)
+
+
+class GTestOutputTest(gtest_test_utils.TestCase):
+ def RemoveUnsupportedTests(self, test_output):
+ if not SUPPORTS_DEATH_TESTS:
+ test_output = RemoveMatchingTests(test_output, 'DeathTest')
+ if not SUPPORTS_TYPED_TESTS:
+ test_output = RemoveMatchingTests(test_output, 'TypedTest')
+ test_output = RemoveMatchingTests(test_output, 'TypedDeathTest')
+ test_output = RemoveMatchingTests(test_output, 'TypeParamDeathTest')
+ if not SUPPORTS_THREADS:
+ test_output = RemoveMatchingTests(test_output,
+ 'ExpectFailureWithThreadsTest')
+ test_output = RemoveMatchingTests(test_output,
+ 'ScopedFakeTestPartResultReporterTest')
+ test_output = RemoveMatchingTests(test_output,
+ 'WorksConcurrently')
+ if not SUPPORTS_STACK_TRACES:
+ test_output = RemoveStackTraces(test_output)
+
+ return test_output
+
+ def testOutput(self):
+ output = GetOutputOfAllCommands()
+
+ golden_file = open(GOLDEN_PATH, 'rb')
+ # A mis-configured source control system can cause \r appear in EOL
+ # sequences when we read the golden file irrespective of an operating
+ # system used. Therefore, we need to strip those \r's from newlines
+ # unconditionally.
+ golden = ToUnixLineEnding(golden_file.read())
+ golden_file.close()
+
+ # We want the test to pass regardless of certain features being
+ # supported or not.
+
+ # We still have to remove type name specifics in all cases.
+ normalized_actual = RemoveTypeInfoDetails(output)
+ normalized_golden = RemoveTypeInfoDetails(golden)
+
+ if CAN_GENERATE_GOLDEN_FILE:
+ self.assertEqual(normalized_golden, normalized_actual)
+ else:
+ normalized_actual = NormalizeToCurrentPlatform(
+ RemoveTestCounts(normalized_actual))
+ normalized_golden = NormalizeToCurrentPlatform(
+ RemoveTestCounts(self.RemoveUnsupportedTests(normalized_golden)))
+
+ # This code is very handy when debugging golden file differences:
+ if os.getenv('DEBUG_GTEST_OUTPUT_TEST'):
+ open(os.path.join(
+ gtest_test_utils.GetSourceDir(),
+ '_gtest_output_test_normalized_actual.txt'), 'wb').write(
+ normalized_actual)
+ open(os.path.join(
+ gtest_test_utils.GetSourceDir(),
+ '_gtest_output_test_normalized_golden.txt'), 'wb').write(
+ normalized_golden)
+
+ self.assertEqual(normalized_golden, normalized_actual)
+
+
+if __name__ == '__main__':
+ if sys.argv[1:] == [GENGOLDEN_FLAG]:
+ if CAN_GENERATE_GOLDEN_FILE:
+ output = GetOutputOfAllCommands()
+ golden_file = open(GOLDEN_PATH, 'wb')
+ golden_file.write(output)
+ golden_file.close()
+ else:
+ message = (
+ """Unable to write a golden file when compiled in an environment
+that does not support all the required features (death tests, typed tests,
+and multiple threads). Please generate the golden file using a binary built
+with those features enabled.""")
+
+ sys.stderr.write(message)
+ sys.exit(1)
+ else:
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_output_test_.cc b/libvpx/third_party/googletest/src/test/gtest_output_test_.cc
new file mode 100644
index 0000000..13dbec4
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_output_test_.cc
@@ -0,0 +1,1020 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// A unit test for Google Test itself. This verifies that the basic
+// constructs of Google Test work.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "gtest/gtest-spi.h"
+#include "gtest/gtest.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+#include <stdlib.h>
+
+#if GTEST_IS_THREADSAFE
+using testing::ScopedFakeTestPartResultReporter;
+using testing::TestPartResultArray;
+
+using testing::internal::Notification;
+using testing::internal::ThreadWithParam;
+#endif
+
+namespace posix = ::testing::internal::posix;
+using testing::internal::String;
+using testing::internal::scoped_ptr;
+
+// Tests catching fatal failures.
+
+// A subroutine used by the following test.
+void TestEq1(int x) {
+ ASSERT_EQ(1, x);
+}
+
+// This function calls a test subroutine, catches the fatal failure it
+// generates, and then returns early.
+void TryTestSubroutine() {
+ // Calls a subrountine that yields a fatal failure.
+ TestEq1(2);
+
+ // Catches the fatal failure and aborts the test.
+ //
+ // The testing::Test:: prefix is necessary when calling
+ // HasFatalFailure() outside of a TEST, TEST_F, or test fixture.
+ if (testing::Test::HasFatalFailure()) return;
+
+ // If we get here, something is wrong.
+ FAIL() << "This should never be reached.";
+}
+
+TEST(PassingTest, PassingTest1) {
+}
+
+TEST(PassingTest, PassingTest2) {
+}
+
+// Tests that parameters of failing parameterized tests are printed in the
+// failing test summary.
+class FailingParamTest : public testing::TestWithParam<int> {};
+
+TEST_P(FailingParamTest, Fails) {
+ EXPECT_EQ(1, GetParam());
+}
+
+// This generates a test which will fail. Google Test is expected to print
+// its parameter when it outputs the list of all failed tests.
+INSTANTIATE_TEST_CASE_P(PrintingFailingParams,
+ FailingParamTest,
+ testing::Values(2));
+
+// Tests catching a fatal failure in a subroutine.
+TEST(FatalFailureTest, FatalFailureInSubroutine) {
+ printf("(expecting a failure that x should be 1)\n");
+
+ TryTestSubroutine();
+}
+
+// Tests catching a fatal failure in a nested subroutine.
+TEST(FatalFailureTest, FatalFailureInNestedSubroutine) {
+ printf("(expecting a failure that x should be 1)\n");
+
+ // Calls a subrountine that yields a fatal failure.
+ TryTestSubroutine();
+
+ // Catches the fatal failure and aborts the test.
+ //
+ // When calling HasFatalFailure() inside a TEST, TEST_F, or test
+ // fixture, the testing::Test:: prefix is not needed.
+ if (HasFatalFailure()) return;
+
+ // If we get here, something is wrong.
+ FAIL() << "This should never be reached.";
+}
+
+// Tests HasFatalFailure() after a failed EXPECT check.
+TEST(FatalFailureTest, NonfatalFailureInSubroutine) {
+ printf("(expecting a failure on false)\n");
+ EXPECT_TRUE(false); // Generates a nonfatal failure
+ ASSERT_FALSE(HasFatalFailure()); // This should succeed.
+}
+
+// Tests interleaving user logging and Google Test assertions.
+TEST(LoggingTest, InterleavingLoggingAndAssertions) {
+ static const int a[4] = {
+ 3, 9, 2, 6
+ };
+
+ printf("(expecting 2 failures on (3) >= (a[i]))\n");
+ for (int i = 0; i < static_cast<int>(sizeof(a)/sizeof(*a)); i++) {
+ printf("i == %d\n", i);
+ EXPECT_GE(3, a[i]);
+ }
+}
+
+// Tests the SCOPED_TRACE macro.
+
+// A helper function for testing SCOPED_TRACE.
+void SubWithoutTrace(int n) {
+ EXPECT_EQ(1, n);
+ ASSERT_EQ(2, n);
+}
+
+// Another helper function for testing SCOPED_TRACE.
+void SubWithTrace(int n) {
+ SCOPED_TRACE(testing::Message() << "n = " << n);
+
+ SubWithoutTrace(n);
+}
+
+// Tests that SCOPED_TRACE() obeys lexical scopes.
+TEST(SCOPED_TRACETest, ObeysScopes) {
+ printf("(expected to fail)\n");
+
+ // There should be no trace before SCOPED_TRACE() is invoked.
+ ADD_FAILURE() << "This failure is expected, and shouldn't have a trace.";
+
+ {
+ SCOPED_TRACE("Expected trace");
+ // After SCOPED_TRACE(), a failure in the current scope should contain
+ // the trace.
+ ADD_FAILURE() << "This failure is expected, and should have a trace.";
+ }
+
+ // Once the control leaves the scope of the SCOPED_TRACE(), there
+ // should be no trace again.
+ ADD_FAILURE() << "This failure is expected, and shouldn't have a trace.";
+}
+
+// Tests that SCOPED_TRACE works inside a loop.
+TEST(SCOPED_TRACETest, WorksInLoop) {
+ printf("(expected to fail)\n");
+
+ for (int i = 1; i <= 2; i++) {
+ SCOPED_TRACE(testing::Message() << "i = " << i);
+
+ SubWithoutTrace(i);
+ }
+}
+
+// Tests that SCOPED_TRACE works in a subroutine.
+TEST(SCOPED_TRACETest, WorksInSubroutine) {
+ printf("(expected to fail)\n");
+
+ SubWithTrace(1);
+ SubWithTrace(2);
+}
+
+// Tests that SCOPED_TRACE can be nested.
+TEST(SCOPED_TRACETest, CanBeNested) {
+ printf("(expected to fail)\n");
+
+ SCOPED_TRACE(""); // A trace without a message.
+
+ SubWithTrace(2);
+}
+
+// Tests that multiple SCOPED_TRACEs can be used in the same scope.
+TEST(SCOPED_TRACETest, CanBeRepeated) {
+ printf("(expected to fail)\n");
+
+ SCOPED_TRACE("A");
+ ADD_FAILURE()
+ << "This failure is expected, and should contain trace point A.";
+
+ SCOPED_TRACE("B");
+ ADD_FAILURE()
+ << "This failure is expected, and should contain trace point A and B.";
+
+ {
+ SCOPED_TRACE("C");
+ ADD_FAILURE() << "This failure is expected, and should contain "
+ << "trace point A, B, and C.";
+ }
+
+ SCOPED_TRACE("D");
+ ADD_FAILURE() << "This failure is expected, and should contain "
+ << "trace point A, B, and D.";
+}
+
+#if GTEST_IS_THREADSAFE
+// Tests that SCOPED_TRACE()s can be used concurrently from multiple
+// threads. Namely, an assertion should be affected by
+// SCOPED_TRACE()s in its own thread only.
+
+// Here's the sequence of actions that happen in the test:
+//
+// Thread A (main) | Thread B (spawned)
+// ===============================|================================
+// spawns thread B |
+// -------------------------------+--------------------------------
+// waits for n1 | SCOPED_TRACE("Trace B");
+// | generates failure #1
+// | notifies n1
+// -------------------------------+--------------------------------
+// SCOPED_TRACE("Trace A"); | waits for n2
+// generates failure #2 |
+// notifies n2 |
+// -------------------------------|--------------------------------
+// waits for n3 | generates failure #3
+// | trace B dies
+// | generates failure #4
+// | notifies n3
+// -------------------------------|--------------------------------
+// generates failure #5 | finishes
+// trace A dies |
+// generates failure #6 |
+// -------------------------------|--------------------------------
+// waits for thread B to finish |
+
+struct CheckPoints {
+ Notification n1;
+ Notification n2;
+ Notification n3;
+};
+
+static void ThreadWithScopedTrace(CheckPoints* check_points) {
+ {
+ SCOPED_TRACE("Trace B");
+ ADD_FAILURE()
+ << "Expected failure #1 (in thread B, only trace B alive).";
+ check_points->n1.Notify();
+ check_points->n2.WaitForNotification();
+
+ ADD_FAILURE()
+ << "Expected failure #3 (in thread B, trace A & B both alive).";
+ } // Trace B dies here.
+ ADD_FAILURE()
+ << "Expected failure #4 (in thread B, only trace A alive).";
+ check_points->n3.Notify();
+}
+
+TEST(SCOPED_TRACETest, WorksConcurrently) {
+ printf("(expecting 6 failures)\n");
+
+ CheckPoints check_points;
+ ThreadWithParam<CheckPoints*> thread(&ThreadWithScopedTrace,
+ &check_points,
+ NULL);
+ check_points.n1.WaitForNotification();
+
+ {
+ SCOPED_TRACE("Trace A");
+ ADD_FAILURE()
+ << "Expected failure #2 (in thread A, trace A & B both alive).";
+ check_points.n2.Notify();
+ check_points.n3.WaitForNotification();
+
+ ADD_FAILURE()
+ << "Expected failure #5 (in thread A, only trace A alive).";
+ } // Trace A dies here.
+ ADD_FAILURE()
+ << "Expected failure #6 (in thread A, no trace alive).";
+ thread.Join();
+}
+#endif // GTEST_IS_THREADSAFE
+
+TEST(DisabledTestsWarningTest,
+ DISABLED_AlsoRunDisabledTestsFlagSuppressesWarning) {
+ // This test body is intentionally empty. Its sole purpose is for
+ // verifying that the --gtest_also_run_disabled_tests flag
+ // suppresses the "YOU HAVE 12 DISABLED TESTS" warning at the end of
+ // the test output.
+}
+
+// Tests using assertions outside of TEST and TEST_F.
+//
+// This function creates two failures intentionally.
+void AdHocTest() {
+ printf("The non-test part of the code is expected to have 2 failures.\n\n");
+ EXPECT_TRUE(false);
+ EXPECT_EQ(2, 3);
+}
+
+// Runs all TESTs, all TEST_Fs, and the ad hoc test.
+int RunAllTests() {
+ AdHocTest();
+ return RUN_ALL_TESTS();
+}
+
+// Tests non-fatal failures in the fixture constructor.
+class NonFatalFailureInFixtureConstructorTest : public testing::Test {
+ protected:
+ NonFatalFailureInFixtureConstructorTest() {
+ printf("(expecting 5 failures)\n");
+ ADD_FAILURE() << "Expected failure #1, in the test fixture c'tor.";
+ }
+
+ ~NonFatalFailureInFixtureConstructorTest() {
+ ADD_FAILURE() << "Expected failure #5, in the test fixture d'tor.";
+ }
+
+ virtual void SetUp() {
+ ADD_FAILURE() << "Expected failure #2, in SetUp().";
+ }
+
+ virtual void TearDown() {
+ ADD_FAILURE() << "Expected failure #4, in TearDown.";
+ }
+};
+
+TEST_F(NonFatalFailureInFixtureConstructorTest, FailureInConstructor) {
+ ADD_FAILURE() << "Expected failure #3, in the test body.";
+}
+
+// Tests fatal failures in the fixture constructor.
+class FatalFailureInFixtureConstructorTest : public testing::Test {
+ protected:
+ FatalFailureInFixtureConstructorTest() {
+ printf("(expecting 2 failures)\n");
+ Init();
+ }
+
+ ~FatalFailureInFixtureConstructorTest() {
+ ADD_FAILURE() << "Expected failure #2, in the test fixture d'tor.";
+ }
+
+ virtual void SetUp() {
+ ADD_FAILURE() << "UNEXPECTED failure in SetUp(). "
+ << "We should never get here, as the test fixture c'tor "
+ << "had a fatal failure.";
+ }
+
+ virtual void TearDown() {
+ ADD_FAILURE() << "UNEXPECTED failure in TearDown(). "
+ << "We should never get here, as the test fixture c'tor "
+ << "had a fatal failure.";
+ }
+ private:
+ void Init() {
+ FAIL() << "Expected failure #1, in the test fixture c'tor.";
+ }
+};
+
+TEST_F(FatalFailureInFixtureConstructorTest, FailureInConstructor) {
+ ADD_FAILURE() << "UNEXPECTED failure in the test body. "
+ << "We should never get here, as the test fixture c'tor "
+ << "had a fatal failure.";
+}
+
+// Tests non-fatal failures in SetUp().
+class NonFatalFailureInSetUpTest : public testing::Test {
+ protected:
+ virtual ~NonFatalFailureInSetUpTest() {
+ Deinit();
+ }
+
+ virtual void SetUp() {
+ printf("(expecting 4 failures)\n");
+ ADD_FAILURE() << "Expected failure #1, in SetUp().";
+ }
+
+ virtual void TearDown() {
+ FAIL() << "Expected failure #3, in TearDown().";
+ }
+ private:
+ void Deinit() {
+ FAIL() << "Expected failure #4, in the test fixture d'tor.";
+ }
+};
+
+TEST_F(NonFatalFailureInSetUpTest, FailureInSetUp) {
+ FAIL() << "Expected failure #2, in the test function.";
+}
+
+// Tests fatal failures in SetUp().
+class FatalFailureInSetUpTest : public testing::Test {
+ protected:
+ virtual ~FatalFailureInSetUpTest() {
+ Deinit();
+ }
+
+ virtual void SetUp() {
+ printf("(expecting 3 failures)\n");
+ FAIL() << "Expected failure #1, in SetUp().";
+ }
+
+ virtual void TearDown() {
+ FAIL() << "Expected failure #2, in TearDown().";
+ }
+ private:
+ void Deinit() {
+ FAIL() << "Expected failure #3, in the test fixture d'tor.";
+ }
+};
+
+TEST_F(FatalFailureInSetUpTest, FailureInSetUp) {
+ FAIL() << "UNEXPECTED failure in the test function. "
+ << "We should never get here, as SetUp() failed.";
+}
+
+TEST(AddFailureAtTest, MessageContainsSpecifiedFileAndLineNumber) {
+ ADD_FAILURE_AT("foo.cc", 42) << "Expected failure in foo.cc";
+}
+
+#if GTEST_IS_THREADSAFE
+
+// A unary function that may die.
+void DieIf(bool should_die) {
+ GTEST_CHECK_(!should_die) << " - death inside DieIf().";
+}
+
+// Tests running death tests in a multi-threaded context.
+
+// Used for coordination between the main and the spawn thread.
+struct SpawnThreadNotifications {
+ SpawnThreadNotifications() {}
+
+ Notification spawn_thread_started;
+ Notification spawn_thread_ok_to_terminate;
+
+ private:
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(SpawnThreadNotifications);
+};
+
+// The function to be executed in the thread spawn by the
+// MultipleThreads test (below).
+static void ThreadRoutine(SpawnThreadNotifications* notifications) {
+ // Signals the main thread that this thread has started.
+ notifications->spawn_thread_started.Notify();
+
+ // Waits for permission to finish from the main thread.
+ notifications->spawn_thread_ok_to_terminate.WaitForNotification();
+}
+
+// This is a death-test test, but it's not named with a DeathTest
+// suffix. It starts threads which might interfere with later
+// death tests, so it must run after all other death tests.
+class DeathTestAndMultiThreadsTest : public testing::Test {
+ protected:
+ // Starts a thread and waits for it to begin.
+ virtual void SetUp() {
+ thread_.reset(new ThreadWithParam<SpawnThreadNotifications*>(
+ &ThreadRoutine, ¬ifications_, NULL));
+ notifications_.spawn_thread_started.WaitForNotification();
+ }
+ // Tells the thread to finish, and reaps it.
+ // Depending on the version of the thread library in use,
+ // a manager thread might still be left running that will interfere
+ // with later death tests. This is unfortunate, but this class
+ // cleans up after itself as best it can.
+ virtual void TearDown() {
+ notifications_.spawn_thread_ok_to_terminate.Notify();
+ }
+
+ private:
+ SpawnThreadNotifications notifications_;
+ scoped_ptr<ThreadWithParam<SpawnThreadNotifications*> > thread_;
+};
+
+#endif // GTEST_IS_THREADSAFE
+
+// The MixedUpTestCaseTest test case verifies that Google Test will fail a
+// test if it uses a different fixture class than what other tests in
+// the same test case use. It deliberately contains two fixture
+// classes with the same name but defined in different namespaces.
+
+// The MixedUpTestCaseWithSameTestNameTest test case verifies that
+// when the user defines two tests with the same test case name AND
+// same test name (but in different namespaces), the second test will
+// fail.
+
+namespace foo {
+
+class MixedUpTestCaseTest : public testing::Test {
+};
+
+TEST_F(MixedUpTestCaseTest, FirstTestFromNamespaceFoo) {}
+TEST_F(MixedUpTestCaseTest, SecondTestFromNamespaceFoo) {}
+
+class MixedUpTestCaseWithSameTestNameTest : public testing::Test {
+};
+
+TEST_F(MixedUpTestCaseWithSameTestNameTest,
+ TheSecondTestWithThisNameShouldFail) {}
+
+} // namespace foo
+
+namespace bar {
+
+class MixedUpTestCaseTest : public testing::Test {
+};
+
+// The following two tests are expected to fail. We rely on the
+// golden file to check that Google Test generates the right error message.
+TEST_F(MixedUpTestCaseTest, ThisShouldFail) {}
+TEST_F(MixedUpTestCaseTest, ThisShouldFailToo) {}
+
+class MixedUpTestCaseWithSameTestNameTest : public testing::Test {
+};
+
+// Expected to fail. We rely on the golden file to check that Google Test
+// generates the right error message.
+TEST_F(MixedUpTestCaseWithSameTestNameTest,
+ TheSecondTestWithThisNameShouldFail) {}
+
+} // namespace bar
+
+// The following two test cases verify that Google Test catches the user
+// error of mixing TEST and TEST_F in the same test case. The first
+// test case checks the scenario where TEST_F appears before TEST, and
+// the second one checks where TEST appears before TEST_F.
+
+class TEST_F_before_TEST_in_same_test_case : public testing::Test {
+};
+
+TEST_F(TEST_F_before_TEST_in_same_test_case, DefinedUsingTEST_F) {}
+
+// Expected to fail. We rely on the golden file to check that Google Test
+// generates the right error message.
+TEST(TEST_F_before_TEST_in_same_test_case, DefinedUsingTESTAndShouldFail) {}
+
+class TEST_before_TEST_F_in_same_test_case : public testing::Test {
+};
+
+TEST(TEST_before_TEST_F_in_same_test_case, DefinedUsingTEST) {}
+
+// Expected to fail. We rely on the golden file to check that Google Test
+// generates the right error message.
+TEST_F(TEST_before_TEST_F_in_same_test_case, DefinedUsingTEST_FAndShouldFail) {
+}
+
+// Used for testing EXPECT_NONFATAL_FAILURE() and EXPECT_FATAL_FAILURE().
+int global_integer = 0;
+
+// Tests that EXPECT_NONFATAL_FAILURE() can reference global variables.
+TEST(ExpectNonfatalFailureTest, CanReferenceGlobalVariables) {
+ global_integer = 0;
+ EXPECT_NONFATAL_FAILURE({
+ EXPECT_EQ(1, global_integer) << "Expected non-fatal failure.";
+ }, "Expected non-fatal failure.");
+}
+
+// Tests that EXPECT_NONFATAL_FAILURE() can reference local variables
+// (static or not).
+TEST(ExpectNonfatalFailureTest, CanReferenceLocalVariables) {
+ int m = 0;
+ static int n;
+ n = 1;
+ EXPECT_NONFATAL_FAILURE({
+ EXPECT_EQ(m, n) << "Expected non-fatal failure.";
+ }, "Expected non-fatal failure.");
+}
+
+// Tests that EXPECT_NONFATAL_FAILURE() succeeds when there is exactly
+// one non-fatal failure and no fatal failure.
+TEST(ExpectNonfatalFailureTest, SucceedsWhenThereIsOneNonfatalFailure) {
+ EXPECT_NONFATAL_FAILURE({
+ ADD_FAILURE() << "Expected non-fatal failure.";
+ }, "Expected non-fatal failure.");
+}
+
+// Tests that EXPECT_NONFATAL_FAILURE() fails when there is no
+// non-fatal failure.
+TEST(ExpectNonfatalFailureTest, FailsWhenThereIsNoNonfatalFailure) {
+ printf("(expecting a failure)\n");
+ EXPECT_NONFATAL_FAILURE({
+ }, "");
+}
+
+// Tests that EXPECT_NONFATAL_FAILURE() fails when there are two
+// non-fatal failures.
+TEST(ExpectNonfatalFailureTest, FailsWhenThereAreTwoNonfatalFailures) {
+ printf("(expecting a failure)\n");
+ EXPECT_NONFATAL_FAILURE({
+ ADD_FAILURE() << "Expected non-fatal failure 1.";
+ ADD_FAILURE() << "Expected non-fatal failure 2.";
+ }, "");
+}
+
+// Tests that EXPECT_NONFATAL_FAILURE() fails when there is one fatal
+// failure.
+TEST(ExpectNonfatalFailureTest, FailsWhenThereIsOneFatalFailure) {
+ printf("(expecting a failure)\n");
+ EXPECT_NONFATAL_FAILURE({
+ FAIL() << "Expected fatal failure.";
+ }, "");
+}
+
+// Tests that EXPECT_NONFATAL_FAILURE() fails when the statement being
+// tested returns.
+TEST(ExpectNonfatalFailureTest, FailsWhenStatementReturns) {
+ printf("(expecting a failure)\n");
+ EXPECT_NONFATAL_FAILURE({
+ return;
+ }, "");
+}
+
+#if GTEST_HAS_EXCEPTIONS
+
+// Tests that EXPECT_NONFATAL_FAILURE() fails when the statement being
+// tested throws.
+TEST(ExpectNonfatalFailureTest, FailsWhenStatementThrows) {
+ printf("(expecting a failure)\n");
+ try {
+ EXPECT_NONFATAL_FAILURE({
+ throw 0;
+ }, "");
+ } catch(int) { // NOLINT
+ }
+}
+
+#endif // GTEST_HAS_EXCEPTIONS
+
+// Tests that EXPECT_FATAL_FAILURE() can reference global variables.
+TEST(ExpectFatalFailureTest, CanReferenceGlobalVariables) {
+ global_integer = 0;
+ EXPECT_FATAL_FAILURE({
+ ASSERT_EQ(1, global_integer) << "Expected fatal failure.";
+ }, "Expected fatal failure.");
+}
+
+// Tests that EXPECT_FATAL_FAILURE() can reference local static
+// variables.
+TEST(ExpectFatalFailureTest, CanReferenceLocalStaticVariables) {
+ static int n;
+ n = 1;
+ EXPECT_FATAL_FAILURE({
+ ASSERT_EQ(0, n) << "Expected fatal failure.";
+ }, "Expected fatal failure.");
+}
+
+// Tests that EXPECT_FATAL_FAILURE() succeeds when there is exactly
+// one fatal failure and no non-fatal failure.
+TEST(ExpectFatalFailureTest, SucceedsWhenThereIsOneFatalFailure) {
+ EXPECT_FATAL_FAILURE({
+ FAIL() << "Expected fatal failure.";
+ }, "Expected fatal failure.");
+}
+
+// Tests that EXPECT_FATAL_FAILURE() fails when there is no fatal
+// failure.
+TEST(ExpectFatalFailureTest, FailsWhenThereIsNoFatalFailure) {
+ printf("(expecting a failure)\n");
+ EXPECT_FATAL_FAILURE({
+ }, "");
+}
+
+// A helper for generating a fatal failure.
+void FatalFailure() {
+ FAIL() << "Expected fatal failure.";
+}
+
+// Tests that EXPECT_FATAL_FAILURE() fails when there are two
+// fatal failures.
+TEST(ExpectFatalFailureTest, FailsWhenThereAreTwoFatalFailures) {
+ printf("(expecting a failure)\n");
+ EXPECT_FATAL_FAILURE({
+ FatalFailure();
+ FatalFailure();
+ }, "");
+}
+
+// Tests that EXPECT_FATAL_FAILURE() fails when there is one non-fatal
+// failure.
+TEST(ExpectFatalFailureTest, FailsWhenThereIsOneNonfatalFailure) {
+ printf("(expecting a failure)\n");
+ EXPECT_FATAL_FAILURE({
+ ADD_FAILURE() << "Expected non-fatal failure.";
+ }, "");
+}
+
+// Tests that EXPECT_FATAL_FAILURE() fails when the statement being
+// tested returns.
+TEST(ExpectFatalFailureTest, FailsWhenStatementReturns) {
+ printf("(expecting a failure)\n");
+ EXPECT_FATAL_FAILURE({
+ return;
+ }, "");
+}
+
+#if GTEST_HAS_EXCEPTIONS
+
+// Tests that EXPECT_FATAL_FAILURE() fails when the statement being
+// tested throws.
+TEST(ExpectFatalFailureTest, FailsWhenStatementThrows) {
+ printf("(expecting a failure)\n");
+ try {
+ EXPECT_FATAL_FAILURE({
+ throw 0;
+ }, "");
+ } catch(int) { // NOLINT
+ }
+}
+
+#endif // GTEST_HAS_EXCEPTIONS
+
+// This #ifdef block tests the output of typed tests.
+#if GTEST_HAS_TYPED_TEST
+
+template <typename T>
+class TypedTest : public testing::Test {
+};
+
+TYPED_TEST_CASE(TypedTest, testing::Types<int>);
+
+TYPED_TEST(TypedTest, Success) {
+ EXPECT_EQ(0, TypeParam());
+}
+
+TYPED_TEST(TypedTest, Failure) {
+ EXPECT_EQ(1, TypeParam()) << "Expected failure";
+}
+
+#endif // GTEST_HAS_TYPED_TEST
+
+// This #ifdef block tests the output of type-parameterized tests.
+#if GTEST_HAS_TYPED_TEST_P
+
+template <typename T>
+class TypedTestP : public testing::Test {
+};
+
+TYPED_TEST_CASE_P(TypedTestP);
+
+TYPED_TEST_P(TypedTestP, Success) {
+ EXPECT_EQ(0U, TypeParam());
+}
+
+TYPED_TEST_P(TypedTestP, Failure) {
+ EXPECT_EQ(1U, TypeParam()) << "Expected failure";
+}
+
+REGISTER_TYPED_TEST_CASE_P(TypedTestP, Success, Failure);
+
+typedef testing::Types<unsigned char, unsigned int> UnsignedTypes;
+INSTANTIATE_TYPED_TEST_CASE_P(Unsigned, TypedTestP, UnsignedTypes);
+
+#endif // GTEST_HAS_TYPED_TEST_P
+
+#if GTEST_HAS_DEATH_TEST
+
+// We rely on the golden file to verify that tests whose test case
+// name ends with DeathTest are run first.
+
+TEST(ADeathTest, ShouldRunFirst) {
+}
+
+# if GTEST_HAS_TYPED_TEST
+
+// We rely on the golden file to verify that typed tests whose test
+// case name ends with DeathTest are run first.
+
+template <typename T>
+class ATypedDeathTest : public testing::Test {
+};
+
+typedef testing::Types<int, double> NumericTypes;
+TYPED_TEST_CASE(ATypedDeathTest, NumericTypes);
+
+TYPED_TEST(ATypedDeathTest, ShouldRunFirst) {
+}
+
+# endif // GTEST_HAS_TYPED_TEST
+
+# if GTEST_HAS_TYPED_TEST_P
+
+
+// We rely on the golden file to verify that type-parameterized tests
+// whose test case name ends with DeathTest are run first.
+
+template <typename T>
+class ATypeParamDeathTest : public testing::Test {
+};
+
+TYPED_TEST_CASE_P(ATypeParamDeathTest);
+
+TYPED_TEST_P(ATypeParamDeathTest, ShouldRunFirst) {
+}
+
+REGISTER_TYPED_TEST_CASE_P(ATypeParamDeathTest, ShouldRunFirst);
+
+INSTANTIATE_TYPED_TEST_CASE_P(My, ATypeParamDeathTest, NumericTypes);
+
+# endif // GTEST_HAS_TYPED_TEST_P
+
+#endif // GTEST_HAS_DEATH_TEST
+
+// Tests various failure conditions of
+// EXPECT_{,NON}FATAL_FAILURE{,_ON_ALL_THREADS}.
+class ExpectFailureTest : public testing::Test {
+ public: // Must be public and not protected due to a bug in g++ 3.4.2.
+ enum FailureMode {
+ FATAL_FAILURE,
+ NONFATAL_FAILURE
+ };
+ static void AddFailure(FailureMode failure) {
+ if (failure == FATAL_FAILURE) {
+ FAIL() << "Expected fatal failure.";
+ } else {
+ ADD_FAILURE() << "Expected non-fatal failure.";
+ }
+ }
+};
+
+TEST_F(ExpectFailureTest, ExpectFatalFailure) {
+ // Expected fatal failure, but succeeds.
+ printf("(expecting 1 failure)\n");
+ EXPECT_FATAL_FAILURE(SUCCEED(), "Expected fatal failure.");
+ // Expected fatal failure, but got a non-fatal failure.
+ printf("(expecting 1 failure)\n");
+ EXPECT_FATAL_FAILURE(AddFailure(NONFATAL_FAILURE), "Expected non-fatal "
+ "failure.");
+ // Wrong message.
+ printf("(expecting 1 failure)\n");
+ EXPECT_FATAL_FAILURE(AddFailure(FATAL_FAILURE), "Some other fatal failure "
+ "expected.");
+}
+
+TEST_F(ExpectFailureTest, ExpectNonFatalFailure) {
+ // Expected non-fatal failure, but succeeds.
+ printf("(expecting 1 failure)\n");
+ EXPECT_NONFATAL_FAILURE(SUCCEED(), "Expected non-fatal failure.");
+ // Expected non-fatal failure, but got a fatal failure.
+ printf("(expecting 1 failure)\n");
+ EXPECT_NONFATAL_FAILURE(AddFailure(FATAL_FAILURE), "Expected fatal failure.");
+ // Wrong message.
+ printf("(expecting 1 failure)\n");
+ EXPECT_NONFATAL_FAILURE(AddFailure(NONFATAL_FAILURE), "Some other non-fatal "
+ "failure.");
+}
+
+#if GTEST_IS_THREADSAFE
+
+class ExpectFailureWithThreadsTest : public ExpectFailureTest {
+ protected:
+ static void AddFailureInOtherThread(FailureMode failure) {
+ ThreadWithParam<FailureMode> thread(&AddFailure, failure, NULL);
+ thread.Join();
+ }
+};
+
+TEST_F(ExpectFailureWithThreadsTest, ExpectFatalFailure) {
+ // We only intercept the current thread.
+ printf("(expecting 2 failures)\n");
+ EXPECT_FATAL_FAILURE(AddFailureInOtherThread(FATAL_FAILURE),
+ "Expected fatal failure.");
+}
+
+TEST_F(ExpectFailureWithThreadsTest, ExpectNonFatalFailure) {
+ // We only intercept the current thread.
+ printf("(expecting 2 failures)\n");
+ EXPECT_NONFATAL_FAILURE(AddFailureInOtherThread(NONFATAL_FAILURE),
+ "Expected non-fatal failure.");
+}
+
+typedef ExpectFailureWithThreadsTest ScopedFakeTestPartResultReporterTest;
+
+// Tests that the ScopedFakeTestPartResultReporter only catches failures from
+// the current thread if it is instantiated with INTERCEPT_ONLY_CURRENT_THREAD.
+TEST_F(ScopedFakeTestPartResultReporterTest, InterceptOnlyCurrentThread) {
+ printf("(expecting 2 failures)\n");
+ TestPartResultArray results;
+ {
+ ScopedFakeTestPartResultReporter reporter(
+ ScopedFakeTestPartResultReporter::INTERCEPT_ONLY_CURRENT_THREAD,
+ &results);
+ AddFailureInOtherThread(FATAL_FAILURE);
+ AddFailureInOtherThread(NONFATAL_FAILURE);
+ }
+ // The two failures should not have been intercepted.
+ EXPECT_EQ(0, results.size()) << "This shouldn't fail.";
+}
+
+#endif // GTEST_IS_THREADSAFE
+
+TEST_F(ExpectFailureTest, ExpectFatalFailureOnAllThreads) {
+ // Expected fatal failure, but succeeds.
+ printf("(expecting 1 failure)\n");
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS(SUCCEED(), "Expected fatal failure.");
+ // Expected fatal failure, but got a non-fatal failure.
+ printf("(expecting 1 failure)\n");
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS(AddFailure(NONFATAL_FAILURE),
+ "Expected non-fatal failure.");
+ // Wrong message.
+ printf("(expecting 1 failure)\n");
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS(AddFailure(FATAL_FAILURE),
+ "Some other fatal failure expected.");
+}
+
+TEST_F(ExpectFailureTest, ExpectNonFatalFailureOnAllThreads) {
+ // Expected non-fatal failure, but succeeds.
+ printf("(expecting 1 failure)\n");
+ EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(SUCCEED(), "Expected non-fatal "
+ "failure.");
+ // Expected non-fatal failure, but got a fatal failure.
+ printf("(expecting 1 failure)\n");
+ EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(AddFailure(FATAL_FAILURE),
+ "Expected fatal failure.");
+ // Wrong message.
+ printf("(expecting 1 failure)\n");
+ EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(AddFailure(NONFATAL_FAILURE),
+ "Some other non-fatal failure.");
+}
+
+
+// Two test environments for testing testing::AddGlobalTestEnvironment().
+
+class FooEnvironment : public testing::Environment {
+ public:
+ virtual void SetUp() {
+ printf("%s", "FooEnvironment::SetUp() called.\n");
+ }
+
+ virtual void TearDown() {
+ printf("%s", "FooEnvironment::TearDown() called.\n");
+ FAIL() << "Expected fatal failure.";
+ }
+};
+
+class BarEnvironment : public testing::Environment {
+ public:
+ virtual void SetUp() {
+ printf("%s", "BarEnvironment::SetUp() called.\n");
+ }
+
+ virtual void TearDown() {
+ printf("%s", "BarEnvironment::TearDown() called.\n");
+ ADD_FAILURE() << "Expected non-fatal failure.";
+ }
+};
+
+bool GTEST_FLAG(internal_skip_environment_and_ad_hoc_tests) = false;
+
+// The main function.
+//
+// The idea is to use Google Test to run all the tests we have defined (some
+// of them are intended to fail), and then compare the test results
+// with the "golden" file.
+int main(int argc, char **argv) {
+ testing::GTEST_FLAG(print_time) = false;
+
+ // We just run the tests, knowing some of them are intended to fail.
+ // We will use a separate Python script to compare the output of
+ // this program with the golden file.
+
+ // It's hard to test InitGoogleTest() directly, as it has many
+ // global side effects. The following line serves as a sanity test
+ // for it.
+ testing::InitGoogleTest(&argc, argv);
+ if (argc >= 2 &&
+ String(argv[1]) == "--gtest_internal_skip_environment_and_ad_hoc_tests")
+ GTEST_FLAG(internal_skip_environment_and_ad_hoc_tests) = true;
+
+#if GTEST_HAS_DEATH_TEST
+ if (testing::internal::GTEST_FLAG(internal_run_death_test) != "") {
+ // Skip the usual output capturing if we're running as the child
+ // process of an threadsafe-style death test.
+# if GTEST_OS_WINDOWS
+ posix::FReopen("nul:", "w", stdout);
+# else
+ posix::FReopen("/dev/null", "w", stdout);
+# endif // GTEST_OS_WINDOWS
+ return RUN_ALL_TESTS();
+ }
+#endif // GTEST_HAS_DEATH_TEST
+
+ if (GTEST_FLAG(internal_skip_environment_and_ad_hoc_tests))
+ return RUN_ALL_TESTS();
+
+ // Registers two global test environments.
+ // The golden file verifies that they are set up in the order they
+ // are registered, and torn down in the reverse order.
+ testing::AddGlobalTestEnvironment(new FooEnvironment);
+ testing::AddGlobalTestEnvironment(new BarEnvironment);
+
+ return RunAllTests();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_output_test_golden_lin.txt b/libvpx/third_party/googletest/src/test/gtest_output_test_golden_lin.txt
new file mode 100644
index 0000000..a1d342d
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_output_test_golden_lin.txt
@@ -0,0 +1,711 @@
+The non-test part of the code is expected to have 2 failures.
+
+gtest_output_test_.cc:#: Failure
+Value of: false
+ Actual: false
+Expected: true
+gtest_output_test_.cc:#: Failure
+Value of: 3
+Expected: 2
+[0;32m[==========] [mRunning 62 tests from 27 test cases.
+[0;32m[----------] [mGlobal test environment set-up.
+FooEnvironment::SetUp() called.
+BarEnvironment::SetUp() called.
+[0;32m[----------] [m1 test from ADeathTest
+[0;32m[ RUN ] [mADeathTest.ShouldRunFirst
+[0;32m[ OK ] [mADeathTest.ShouldRunFirst
+[0;32m[----------] [m1 test from ATypedDeathTest/0, where TypeParam = int
+[0;32m[ RUN ] [mATypedDeathTest/0.ShouldRunFirst
+[0;32m[ OK ] [mATypedDeathTest/0.ShouldRunFirst
+[0;32m[----------] [m1 test from ATypedDeathTest/1, where TypeParam = double
+[0;32m[ RUN ] [mATypedDeathTest/1.ShouldRunFirst
+[0;32m[ OK ] [mATypedDeathTest/1.ShouldRunFirst
+[0;32m[----------] [m1 test from My/ATypeParamDeathTest/0, where TypeParam = int
+[0;32m[ RUN ] [mMy/ATypeParamDeathTest/0.ShouldRunFirst
+[0;32m[ OK ] [mMy/ATypeParamDeathTest/0.ShouldRunFirst
+[0;32m[----------] [m1 test from My/ATypeParamDeathTest/1, where TypeParam = double
+[0;32m[ RUN ] [mMy/ATypeParamDeathTest/1.ShouldRunFirst
+[0;32m[ OK ] [mMy/ATypeParamDeathTest/1.ShouldRunFirst
+[0;32m[----------] [m2 tests from PassingTest
+[0;32m[ RUN ] [mPassingTest.PassingTest1
+[0;32m[ OK ] [mPassingTest.PassingTest1
+[0;32m[ RUN ] [mPassingTest.PassingTest2
+[0;32m[ OK ] [mPassingTest.PassingTest2
+[0;32m[----------] [m3 tests from FatalFailureTest
+[0;32m[ RUN ] [mFatalFailureTest.FatalFailureInSubroutine
+(expecting a failure that x should be 1)
+gtest_output_test_.cc:#: Failure
+Value of: x
+ Actual: 2
+Expected: 1
+[0;31m[ FAILED ] [mFatalFailureTest.FatalFailureInSubroutine
+[0;32m[ RUN ] [mFatalFailureTest.FatalFailureInNestedSubroutine
+(expecting a failure that x should be 1)
+gtest_output_test_.cc:#: Failure
+Value of: x
+ Actual: 2
+Expected: 1
+[0;31m[ FAILED ] [mFatalFailureTest.FatalFailureInNestedSubroutine
+[0;32m[ RUN ] [mFatalFailureTest.NonfatalFailureInSubroutine
+(expecting a failure on false)
+gtest_output_test_.cc:#: Failure
+Value of: false
+ Actual: false
+Expected: true
+[0;31m[ FAILED ] [mFatalFailureTest.NonfatalFailureInSubroutine
+[0;32m[----------] [m1 test from LoggingTest
+[0;32m[ RUN ] [mLoggingTest.InterleavingLoggingAndAssertions
+(expecting 2 failures on (3) >= (a[i]))
+i == 0
+i == 1
+gtest_output_test_.cc:#: Failure
+Expected: (3) >= (a[i]), actual: 3 vs 9
+i == 2
+i == 3
+gtest_output_test_.cc:#: Failure
+Expected: (3) >= (a[i]), actual: 3 vs 6
+[0;31m[ FAILED ] [mLoggingTest.InterleavingLoggingAndAssertions
+[0;32m[----------] [m6 tests from SCOPED_TRACETest
+[0;32m[ RUN ] [mSCOPED_TRACETest.ObeysScopes
+(expected to fail)
+gtest_output_test_.cc:#: Failure
+Failed
+This failure is expected, and shouldn't have a trace.
+gtest_output_test_.cc:#: Failure
+Failed
+This failure is expected, and should have a trace.
+Google Test trace:
+gtest_output_test_.cc:#: Expected trace
+gtest_output_test_.cc:#: Failure
+Failed
+This failure is expected, and shouldn't have a trace.
+[0;31m[ FAILED ] [mSCOPED_TRACETest.ObeysScopes
+[0;32m[ RUN ] [mSCOPED_TRACETest.WorksInLoop
+(expected to fail)
+gtest_output_test_.cc:#: Failure
+Value of: n
+ Actual: 1
+Expected: 2
+Google Test trace:
+gtest_output_test_.cc:#: i = 1
+gtest_output_test_.cc:#: Failure
+Value of: n
+ Actual: 2
+Expected: 1
+Google Test trace:
+gtest_output_test_.cc:#: i = 2
+[0;31m[ FAILED ] [mSCOPED_TRACETest.WorksInLoop
+[0;32m[ RUN ] [mSCOPED_TRACETest.WorksInSubroutine
+(expected to fail)
+gtest_output_test_.cc:#: Failure
+Value of: n
+ Actual: 1
+Expected: 2
+Google Test trace:
+gtest_output_test_.cc:#: n = 1
+gtest_output_test_.cc:#: Failure
+Value of: n
+ Actual: 2
+Expected: 1
+Google Test trace:
+gtest_output_test_.cc:#: n = 2
+[0;31m[ FAILED ] [mSCOPED_TRACETest.WorksInSubroutine
+[0;32m[ RUN ] [mSCOPED_TRACETest.CanBeNested
+(expected to fail)
+gtest_output_test_.cc:#: Failure
+Value of: n
+ Actual: 2
+Expected: 1
+Google Test trace:
+gtest_output_test_.cc:#: n = 2
+gtest_output_test_.cc:#:
+[0;31m[ FAILED ] [mSCOPED_TRACETest.CanBeNested
+[0;32m[ RUN ] [mSCOPED_TRACETest.CanBeRepeated
+(expected to fail)
+gtest_output_test_.cc:#: Failure
+Failed
+This failure is expected, and should contain trace point A.
+Google Test trace:
+gtest_output_test_.cc:#: A
+gtest_output_test_.cc:#: Failure
+Failed
+This failure is expected, and should contain trace point A and B.
+Google Test trace:
+gtest_output_test_.cc:#: B
+gtest_output_test_.cc:#: A
+gtest_output_test_.cc:#: Failure
+Failed
+This failure is expected, and should contain trace point A, B, and C.
+Google Test trace:
+gtest_output_test_.cc:#: C
+gtest_output_test_.cc:#: B
+gtest_output_test_.cc:#: A
+gtest_output_test_.cc:#: Failure
+Failed
+This failure is expected, and should contain trace point A, B, and D.
+Google Test trace:
+gtest_output_test_.cc:#: D
+gtest_output_test_.cc:#: B
+gtest_output_test_.cc:#: A
+[0;31m[ FAILED ] [mSCOPED_TRACETest.CanBeRepeated
+[0;32m[ RUN ] [mSCOPED_TRACETest.WorksConcurrently
+(expecting 6 failures)
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #1 (in thread B, only trace B alive).
+Google Test trace:
+gtest_output_test_.cc:#: Trace B
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #2 (in thread A, trace A & B both alive).
+Google Test trace:
+gtest_output_test_.cc:#: Trace A
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #3 (in thread B, trace A & B both alive).
+Google Test trace:
+gtest_output_test_.cc:#: Trace B
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #4 (in thread B, only trace A alive).
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #5 (in thread A, only trace A alive).
+Google Test trace:
+gtest_output_test_.cc:#: Trace A
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #6 (in thread A, no trace alive).
+[0;31m[ FAILED ] [mSCOPED_TRACETest.WorksConcurrently
+[0;32m[----------] [m1 test from NonFatalFailureInFixtureConstructorTest
+[0;32m[ RUN ] [mNonFatalFailureInFixtureConstructorTest.FailureInConstructor
+(expecting 5 failures)
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #1, in the test fixture c'tor.
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #2, in SetUp().
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #3, in the test body.
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #4, in TearDown.
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #5, in the test fixture d'tor.
+[0;31m[ FAILED ] [mNonFatalFailureInFixtureConstructorTest.FailureInConstructor
+[0;32m[----------] [m1 test from FatalFailureInFixtureConstructorTest
+[0;32m[ RUN ] [mFatalFailureInFixtureConstructorTest.FailureInConstructor
+(expecting 2 failures)
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #1, in the test fixture c'tor.
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #2, in the test fixture d'tor.
+[0;31m[ FAILED ] [mFatalFailureInFixtureConstructorTest.FailureInConstructor
+[0;32m[----------] [m1 test from NonFatalFailureInSetUpTest
+[0;32m[ RUN ] [mNonFatalFailureInSetUpTest.FailureInSetUp
+(expecting 4 failures)
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #1, in SetUp().
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #2, in the test function.
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #3, in TearDown().
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #4, in the test fixture d'tor.
+[0;31m[ FAILED ] [mNonFatalFailureInSetUpTest.FailureInSetUp
+[0;32m[----------] [m1 test from FatalFailureInSetUpTest
+[0;32m[ RUN ] [mFatalFailureInSetUpTest.FailureInSetUp
+(expecting 3 failures)
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #1, in SetUp().
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #2, in TearDown().
+gtest_output_test_.cc:#: Failure
+Failed
+Expected failure #3, in the test fixture d'tor.
+[0;31m[ FAILED ] [mFatalFailureInSetUpTest.FailureInSetUp
+[0;32m[----------] [m1 test from AddFailureAtTest
+[0;32m[ RUN ] [mAddFailureAtTest.MessageContainsSpecifiedFileAndLineNumber
+foo.cc:42: Failure
+Failed
+Expected failure in foo.cc
+[0;31m[ FAILED ] [mAddFailureAtTest.MessageContainsSpecifiedFileAndLineNumber
+[0;32m[----------] [m4 tests from MixedUpTestCaseTest
+[0;32m[ RUN ] [mMixedUpTestCaseTest.FirstTestFromNamespaceFoo
+[0;32m[ OK ] [mMixedUpTestCaseTest.FirstTestFromNamespaceFoo
+[0;32m[ RUN ] [mMixedUpTestCaseTest.SecondTestFromNamespaceFoo
+[0;32m[ OK ] [mMixedUpTestCaseTest.SecondTestFromNamespaceFoo
+[0;32m[ RUN ] [mMixedUpTestCaseTest.ThisShouldFail
+gtest.cc:#: Failure
+Failed
+All tests in the same test case must use the same test fixture
+class. However, in test case MixedUpTestCaseTest,
+you defined test FirstTestFromNamespaceFoo and test ThisShouldFail
+using two different test fixture classes. This can happen if
+the two classes are from different namespaces or translation
+units and have the same name. You should probably rename one
+of the classes to put the tests into different test cases.
+[0;31m[ FAILED ] [mMixedUpTestCaseTest.ThisShouldFail
+[0;32m[ RUN ] [mMixedUpTestCaseTest.ThisShouldFailToo
+gtest.cc:#: Failure
+Failed
+All tests in the same test case must use the same test fixture
+class. However, in test case MixedUpTestCaseTest,
+you defined test FirstTestFromNamespaceFoo and test ThisShouldFailToo
+using two different test fixture classes. This can happen if
+the two classes are from different namespaces or translation
+units and have the same name. You should probably rename one
+of the classes to put the tests into different test cases.
+[0;31m[ FAILED ] [mMixedUpTestCaseTest.ThisShouldFailToo
+[0;32m[----------] [m2 tests from MixedUpTestCaseWithSameTestNameTest
+[0;32m[ RUN ] [mMixedUpTestCaseWithSameTestNameTest.TheSecondTestWithThisNameShouldFail
+[0;32m[ OK ] [mMixedUpTestCaseWithSameTestNameTest.TheSecondTestWithThisNameShouldFail
+[0;32m[ RUN ] [mMixedUpTestCaseWithSameTestNameTest.TheSecondTestWithThisNameShouldFail
+gtest.cc:#: Failure
+Failed
+All tests in the same test case must use the same test fixture
+class. However, in test case MixedUpTestCaseWithSameTestNameTest,
+you defined test TheSecondTestWithThisNameShouldFail and test TheSecondTestWithThisNameShouldFail
+using two different test fixture classes. This can happen if
+the two classes are from different namespaces or translation
+units and have the same name. You should probably rename one
+of the classes to put the tests into different test cases.
+[0;31m[ FAILED ] [mMixedUpTestCaseWithSameTestNameTest.TheSecondTestWithThisNameShouldFail
+[0;32m[----------] [m2 tests from TEST_F_before_TEST_in_same_test_case
+[0;32m[ RUN ] [mTEST_F_before_TEST_in_same_test_case.DefinedUsingTEST_F
+[0;32m[ OK ] [mTEST_F_before_TEST_in_same_test_case.DefinedUsingTEST_F
+[0;32m[ RUN ] [mTEST_F_before_TEST_in_same_test_case.DefinedUsingTESTAndShouldFail
+gtest.cc:#: Failure
+Failed
+All tests in the same test case must use the same test fixture
+class, so mixing TEST_F and TEST in the same test case is
+illegal. In test case TEST_F_before_TEST_in_same_test_case,
+test DefinedUsingTEST_F is defined using TEST_F but
+test DefinedUsingTESTAndShouldFail is defined using TEST. You probably
+want to change the TEST to TEST_F or move it to another test
+case.
+[0;31m[ FAILED ] [mTEST_F_before_TEST_in_same_test_case.DefinedUsingTESTAndShouldFail
+[0;32m[----------] [m2 tests from TEST_before_TEST_F_in_same_test_case
+[0;32m[ RUN ] [mTEST_before_TEST_F_in_same_test_case.DefinedUsingTEST
+[0;32m[ OK ] [mTEST_before_TEST_F_in_same_test_case.DefinedUsingTEST
+[0;32m[ RUN ] [mTEST_before_TEST_F_in_same_test_case.DefinedUsingTEST_FAndShouldFail
+gtest.cc:#: Failure
+Failed
+All tests in the same test case must use the same test fixture
+class, so mixing TEST_F and TEST in the same test case is
+illegal. In test case TEST_before_TEST_F_in_same_test_case,
+test DefinedUsingTEST_FAndShouldFail is defined using TEST_F but
+test DefinedUsingTEST is defined using TEST. You probably
+want to change the TEST to TEST_F or move it to another test
+case.
+[0;31m[ FAILED ] [mTEST_before_TEST_F_in_same_test_case.DefinedUsingTEST_FAndShouldFail
+[0;32m[----------] [m8 tests from ExpectNonfatalFailureTest
+[0;32m[ RUN ] [mExpectNonfatalFailureTest.CanReferenceGlobalVariables
+[0;32m[ OK ] [mExpectNonfatalFailureTest.CanReferenceGlobalVariables
+[0;32m[ RUN ] [mExpectNonfatalFailureTest.CanReferenceLocalVariables
+[0;32m[ OK ] [mExpectNonfatalFailureTest.CanReferenceLocalVariables
+[0;32m[ RUN ] [mExpectNonfatalFailureTest.SucceedsWhenThereIsOneNonfatalFailure
+[0;32m[ OK ] [mExpectNonfatalFailureTest.SucceedsWhenThereIsOneNonfatalFailure
+[0;32m[ RUN ] [mExpectNonfatalFailureTest.FailsWhenThereIsNoNonfatalFailure
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual: 0 failures
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenThereIsNoNonfatalFailure
+[0;32m[ RUN ] [mExpectNonfatalFailureTest.FailsWhenThereAreTwoNonfatalFailures
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual: 2 failures
+gtest_output_test_.cc:#: Non-fatal failure:
+Failed
+Expected non-fatal failure 1.
+
+gtest_output_test_.cc:#: Non-fatal failure:
+Failed
+Expected non-fatal failure 2.
+
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenThereAreTwoNonfatalFailures
+[0;32m[ RUN ] [mExpectNonfatalFailureTest.FailsWhenThereIsOneFatalFailure
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual:
+gtest_output_test_.cc:#: Fatal failure:
+Failed
+Expected fatal failure.
+
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenThereIsOneFatalFailure
+[0;32m[ RUN ] [mExpectNonfatalFailureTest.FailsWhenStatementReturns
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual: 0 failures
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenStatementReturns
+[0;32m[ RUN ] [mExpectNonfatalFailureTest.FailsWhenStatementThrows
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual: 0 failures
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenStatementThrows
+[0;32m[----------] [m8 tests from ExpectFatalFailureTest
+[0;32m[ RUN ] [mExpectFatalFailureTest.CanReferenceGlobalVariables
+[0;32m[ OK ] [mExpectFatalFailureTest.CanReferenceGlobalVariables
+[0;32m[ RUN ] [mExpectFatalFailureTest.CanReferenceLocalStaticVariables
+[0;32m[ OK ] [mExpectFatalFailureTest.CanReferenceLocalStaticVariables
+[0;32m[ RUN ] [mExpectFatalFailureTest.SucceedsWhenThereIsOneFatalFailure
+[0;32m[ OK ] [mExpectFatalFailureTest.SucceedsWhenThereIsOneFatalFailure
+[0;32m[ RUN ] [mExpectFatalFailureTest.FailsWhenThereIsNoFatalFailure
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual: 0 failures
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenThereIsNoFatalFailure
+[0;32m[ RUN ] [mExpectFatalFailureTest.FailsWhenThereAreTwoFatalFailures
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual: 2 failures
+gtest_output_test_.cc:#: Fatal failure:
+Failed
+Expected fatal failure.
+
+gtest_output_test_.cc:#: Fatal failure:
+Failed
+Expected fatal failure.
+
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenThereAreTwoFatalFailures
+[0;32m[ RUN ] [mExpectFatalFailureTest.FailsWhenThereIsOneNonfatalFailure
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual:
+gtest_output_test_.cc:#: Non-fatal failure:
+Failed
+Expected non-fatal failure.
+
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenThereIsOneNonfatalFailure
+[0;32m[ RUN ] [mExpectFatalFailureTest.FailsWhenStatementReturns
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual: 0 failures
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenStatementReturns
+[0;32m[ RUN ] [mExpectFatalFailureTest.FailsWhenStatementThrows
+(expecting a failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual: 0 failures
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenStatementThrows
+[0;32m[----------] [m2 tests from TypedTest/0, where TypeParam = int
+[0;32m[ RUN ] [mTypedTest/0.Success
+[0;32m[ OK ] [mTypedTest/0.Success
+[0;32m[ RUN ] [mTypedTest/0.Failure
+gtest_output_test_.cc:#: Failure
+Value of: TypeParam()
+ Actual: 0
+Expected: 1
+Expected failure
+[0;31m[ FAILED ] [mTypedTest/0.Failure, where TypeParam = int
+[0;32m[----------] [m2 tests from Unsigned/TypedTestP/0, where TypeParam = unsigned char
+[0;32m[ RUN ] [mUnsigned/TypedTestP/0.Success
+[0;32m[ OK ] [mUnsigned/TypedTestP/0.Success
+[0;32m[ RUN ] [mUnsigned/TypedTestP/0.Failure
+gtest_output_test_.cc:#: Failure
+Value of: TypeParam()
+ Actual: '\0'
+Expected: 1U
+Which is: 1
+Expected failure
+[0;31m[ FAILED ] [mUnsigned/TypedTestP/0.Failure, where TypeParam = unsigned char
+[0;32m[----------] [m2 tests from Unsigned/TypedTestP/1, where TypeParam = unsigned int
+[0;32m[ RUN ] [mUnsigned/TypedTestP/1.Success
+[0;32m[ OK ] [mUnsigned/TypedTestP/1.Success
+[0;32m[ RUN ] [mUnsigned/TypedTestP/1.Failure
+gtest_output_test_.cc:#: Failure
+Value of: TypeParam()
+ Actual: 0
+Expected: 1U
+Which is: 1
+Expected failure
+[0;31m[ FAILED ] [mUnsigned/TypedTestP/1.Failure, where TypeParam = unsigned int
+[0;32m[----------] [m4 tests from ExpectFailureTest
+[0;32m[ RUN ] [mExpectFailureTest.ExpectFatalFailure
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual:
+gtest_output_test_.cc:#: Success:
+Succeeded
+
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual:
+gtest_output_test_.cc:#: Non-fatal failure:
+Failed
+Expected non-fatal failure.
+
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure containing "Some other fatal failure expected."
+ Actual:
+gtest_output_test_.cc:#: Fatal failure:
+Failed
+Expected fatal failure.
+
+[0;31m[ FAILED ] [mExpectFailureTest.ExpectFatalFailure
+[0;32m[ RUN ] [mExpectFailureTest.ExpectNonFatalFailure
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual:
+gtest_output_test_.cc:#: Success:
+Succeeded
+
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual:
+gtest_output_test_.cc:#: Fatal failure:
+Failed
+Expected fatal failure.
+
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure containing "Some other non-fatal failure."
+ Actual:
+gtest_output_test_.cc:#: Non-fatal failure:
+Failed
+Expected non-fatal failure.
+
+[0;31m[ FAILED ] [mExpectFailureTest.ExpectNonFatalFailure
+[0;32m[ RUN ] [mExpectFailureTest.ExpectFatalFailureOnAllThreads
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual:
+gtest_output_test_.cc:#: Success:
+Succeeded
+
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual:
+gtest_output_test_.cc:#: Non-fatal failure:
+Failed
+Expected non-fatal failure.
+
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 fatal failure containing "Some other fatal failure expected."
+ Actual:
+gtest_output_test_.cc:#: Fatal failure:
+Failed
+Expected fatal failure.
+
+[0;31m[ FAILED ] [mExpectFailureTest.ExpectFatalFailureOnAllThreads
+[0;32m[ RUN ] [mExpectFailureTest.ExpectNonFatalFailureOnAllThreads
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual:
+gtest_output_test_.cc:#: Success:
+Succeeded
+
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual:
+gtest_output_test_.cc:#: Fatal failure:
+Failed
+Expected fatal failure.
+
+(expecting 1 failure)
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure containing "Some other non-fatal failure."
+ Actual:
+gtest_output_test_.cc:#: Non-fatal failure:
+Failed
+Expected non-fatal failure.
+
+[0;31m[ FAILED ] [mExpectFailureTest.ExpectNonFatalFailureOnAllThreads
+[0;32m[----------] [m2 tests from ExpectFailureWithThreadsTest
+[0;32m[ RUN ] [mExpectFailureWithThreadsTest.ExpectFatalFailure
+(expecting 2 failures)
+gtest_output_test_.cc:#: Failure
+Failed
+Expected fatal failure.
+gtest.cc:#: Failure
+Expected: 1 fatal failure
+ Actual: 0 failures
+[0;31m[ FAILED ] [mExpectFailureWithThreadsTest.ExpectFatalFailure
+[0;32m[ RUN ] [mExpectFailureWithThreadsTest.ExpectNonFatalFailure
+(expecting 2 failures)
+gtest_output_test_.cc:#: Failure
+Failed
+Expected non-fatal failure.
+gtest.cc:#: Failure
+Expected: 1 non-fatal failure
+ Actual: 0 failures
+[0;31m[ FAILED ] [mExpectFailureWithThreadsTest.ExpectNonFatalFailure
+[0;32m[----------] [m1 test from ScopedFakeTestPartResultReporterTest
+[0;32m[ RUN ] [mScopedFakeTestPartResultReporterTest.InterceptOnlyCurrentThread
+(expecting 2 failures)
+gtest_output_test_.cc:#: Failure
+Failed
+Expected fatal failure.
+gtest_output_test_.cc:#: Failure
+Failed
+Expected non-fatal failure.
+[0;31m[ FAILED ] [mScopedFakeTestPartResultReporterTest.InterceptOnlyCurrentThread
+[0;32m[----------] [m1 test from PrintingFailingParams/FailingParamTest
+[0;32m[ RUN ] [mPrintingFailingParams/FailingParamTest.Fails/0
+gtest_output_test_.cc:#: Failure
+Value of: GetParam()
+ Actual: 2
+Expected: 1
+[0;31m[ FAILED ] [mPrintingFailingParams/FailingParamTest.Fails/0, where GetParam() = 2
+[0;32m[----------] [mGlobal test environment tear-down
+BarEnvironment::TearDown() called.
+gtest_output_test_.cc:#: Failure
+Failed
+Expected non-fatal failure.
+FooEnvironment::TearDown() called.
+gtest_output_test_.cc:#: Failure
+Failed
+Expected fatal failure.
+[0;32m[==========] [m62 tests from 27 test cases ran.
+[0;32m[ PASSED ] [m21 tests.
+[0;31m[ FAILED ] [m41 tests, listed below:
+[0;31m[ FAILED ] [mFatalFailureTest.FatalFailureInSubroutine
+[0;31m[ FAILED ] [mFatalFailureTest.FatalFailureInNestedSubroutine
+[0;31m[ FAILED ] [mFatalFailureTest.NonfatalFailureInSubroutine
+[0;31m[ FAILED ] [mLoggingTest.InterleavingLoggingAndAssertions
+[0;31m[ FAILED ] [mSCOPED_TRACETest.ObeysScopes
+[0;31m[ FAILED ] [mSCOPED_TRACETest.WorksInLoop
+[0;31m[ FAILED ] [mSCOPED_TRACETest.WorksInSubroutine
+[0;31m[ FAILED ] [mSCOPED_TRACETest.CanBeNested
+[0;31m[ FAILED ] [mSCOPED_TRACETest.CanBeRepeated
+[0;31m[ FAILED ] [mSCOPED_TRACETest.WorksConcurrently
+[0;31m[ FAILED ] [mNonFatalFailureInFixtureConstructorTest.FailureInConstructor
+[0;31m[ FAILED ] [mFatalFailureInFixtureConstructorTest.FailureInConstructor
+[0;31m[ FAILED ] [mNonFatalFailureInSetUpTest.FailureInSetUp
+[0;31m[ FAILED ] [mFatalFailureInSetUpTest.FailureInSetUp
+[0;31m[ FAILED ] [mAddFailureAtTest.MessageContainsSpecifiedFileAndLineNumber
+[0;31m[ FAILED ] [mMixedUpTestCaseTest.ThisShouldFail
+[0;31m[ FAILED ] [mMixedUpTestCaseTest.ThisShouldFailToo
+[0;31m[ FAILED ] [mMixedUpTestCaseWithSameTestNameTest.TheSecondTestWithThisNameShouldFail
+[0;31m[ FAILED ] [mTEST_F_before_TEST_in_same_test_case.DefinedUsingTESTAndShouldFail
+[0;31m[ FAILED ] [mTEST_before_TEST_F_in_same_test_case.DefinedUsingTEST_FAndShouldFail
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenThereIsNoNonfatalFailure
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenThereAreTwoNonfatalFailures
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenThereIsOneFatalFailure
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenStatementReturns
+[0;31m[ FAILED ] [mExpectNonfatalFailureTest.FailsWhenStatementThrows
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenThereIsNoFatalFailure
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenThereAreTwoFatalFailures
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenThereIsOneNonfatalFailure
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenStatementReturns
+[0;31m[ FAILED ] [mExpectFatalFailureTest.FailsWhenStatementThrows
+[0;31m[ FAILED ] [mTypedTest/0.Failure, where TypeParam = int
+[0;31m[ FAILED ] [mUnsigned/TypedTestP/0.Failure, where TypeParam = unsigned char
+[0;31m[ FAILED ] [mUnsigned/TypedTestP/1.Failure, where TypeParam = unsigned int
+[0;31m[ FAILED ] [mExpectFailureTest.ExpectFatalFailure
+[0;31m[ FAILED ] [mExpectFailureTest.ExpectNonFatalFailure
+[0;31m[ FAILED ] [mExpectFailureTest.ExpectFatalFailureOnAllThreads
+[0;31m[ FAILED ] [mExpectFailureTest.ExpectNonFatalFailureOnAllThreads
+[0;31m[ FAILED ] [mExpectFailureWithThreadsTest.ExpectFatalFailure
+[0;31m[ FAILED ] [mExpectFailureWithThreadsTest.ExpectNonFatalFailure
+[0;31m[ FAILED ] [mScopedFakeTestPartResultReporterTest.InterceptOnlyCurrentThread
+[0;31m[ FAILED ] [mPrintingFailingParams/FailingParamTest.Fails/0, where GetParam() = 2
+
+41 FAILED TESTS
+[0;33m YOU HAVE 1 DISABLED TEST
+
+[mNote: Google Test filter = FatalFailureTest.*:LoggingTest.*
+[==========] Running 4 tests from 2 test cases.
+[----------] Global test environment set-up.
+[----------] 3 tests from FatalFailureTest
+[ RUN ] FatalFailureTest.FatalFailureInSubroutine
+(expecting a failure that x should be 1)
+gtest_output_test_.cc:#: Failure
+Value of: x
+ Actual: 2
+Expected: 1
+[ FAILED ] FatalFailureTest.FatalFailureInSubroutine (? ms)
+[ RUN ] FatalFailureTest.FatalFailureInNestedSubroutine
+(expecting a failure that x should be 1)
+gtest_output_test_.cc:#: Failure
+Value of: x
+ Actual: 2
+Expected: 1
+[ FAILED ] FatalFailureTest.FatalFailureInNestedSubroutine (? ms)
+[ RUN ] FatalFailureTest.NonfatalFailureInSubroutine
+(expecting a failure on false)
+gtest_output_test_.cc:#: Failure
+Value of: false
+ Actual: false
+Expected: true
+[ FAILED ] FatalFailureTest.NonfatalFailureInSubroutine (? ms)
+[----------] 3 tests from FatalFailureTest (? ms total)
+
+[----------] 1 test from LoggingTest
+[ RUN ] LoggingTest.InterleavingLoggingAndAssertions
+(expecting 2 failures on (3) >= (a[i]))
+i == 0
+i == 1
+gtest_output_test_.cc:#: Failure
+Expected: (3) >= (a[i]), actual: 3 vs 9
+i == 2
+i == 3
+gtest_output_test_.cc:#: Failure
+Expected: (3) >= (a[i]), actual: 3 vs 6
+[ FAILED ] LoggingTest.InterleavingLoggingAndAssertions (? ms)
+[----------] 1 test from LoggingTest (? ms total)
+
+[----------] Global test environment tear-down
+[==========] 4 tests from 2 test cases ran. (? ms total)
+[ PASSED ] 0 tests.
+[ FAILED ] 4 tests, listed below:
+[ FAILED ] FatalFailureTest.FatalFailureInSubroutine
+[ FAILED ] FatalFailureTest.FatalFailureInNestedSubroutine
+[ FAILED ] FatalFailureTest.NonfatalFailureInSubroutine
+[ FAILED ] LoggingTest.InterleavingLoggingAndAssertions
+
+ 4 FAILED TESTS
+ YOU HAVE 1 DISABLED TEST
+
+Note: Google Test filter = *DISABLED_*
+[==========] Running 1 test from 1 test case.
+[----------] Global test environment set-up.
+[----------] 1 test from DisabledTestsWarningTest
+[ RUN ] DisabledTestsWarningTest.DISABLED_AlsoRunDisabledTestsFlagSuppressesWarning
+[ OK ] DisabledTestsWarningTest.DISABLED_AlsoRunDisabledTestsFlagSuppressesWarning
+[----------] Global test environment tear-down
+[==========] 1 test from 1 test case ran.
+[ PASSED ] 1 test.
+Note: Google Test filter = PassingTest.*
+Note: This is test shard 2 of 2.
+[==========] Running 1 test from 1 test case.
+[----------] Global test environment set-up.
+[----------] 1 test from PassingTest
+[ RUN ] PassingTest.PassingTest2
+[ OK ] PassingTest.PassingTest2
+[----------] Global test environment tear-down
+[==========] 1 test from 1 test case ran.
+[ PASSED ] 1 test.
+
+ YOU HAVE 1 DISABLED TEST
+
diff --git a/libvpx/third_party/googletest/src/test/gtest_pred_impl_unittest.cc b/libvpx/third_party/googletest/src/test/gtest_pred_impl_unittest.cc
new file mode 100644
index 0000000..35dc9bc
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_pred_impl_unittest.cc
@@ -0,0 +1,2427 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// This file is AUTOMATICALLY GENERATED on 09/24/2010 by command
+// 'gen_gtest_pred_impl.py 5'. DO NOT EDIT BY HAND!
+
+// Regression test for gtest_pred_impl.h
+//
+// This file is generated by a script and quite long. If you intend to
+// learn how Google Test works by reading its unit tests, read
+// gtest_unittest.cc instead.
+//
+// This is intended as a regression test for the Google Test predicate
+// assertions. We compile it as part of the gtest_unittest target
+// only to keep the implementation tidy and compact, as it is quite
+// involved to set up the stage for testing Google Test using Google
+// Test itself.
+//
+// Currently, gtest_unittest takes ~11 seconds to run in the testing
+// daemon. In the future, if it grows too large and needs much more
+// time to finish, we should consider separating this file into a
+// stand-alone regression test.
+
+#include <iostream>
+
+#include "gtest/gtest.h"
+#include "gtest/gtest-spi.h"
+
+// A user-defined data type.
+struct Bool {
+ explicit Bool(int val) : value(val != 0) {}
+
+ bool operator>(int n) const { return value > Bool(n).value; }
+
+ Bool operator+(const Bool& rhs) const { return Bool(value + rhs.value); }
+
+ bool operator==(const Bool& rhs) const { return value == rhs.value; }
+
+ bool value;
+};
+
+// Enables Bool to be used in assertions.
+std::ostream& operator<<(std::ostream& os, const Bool& x) {
+ return os << (x.value ? "true" : "false");
+}
+
+// Sample functions/functors for testing unary predicate assertions.
+
+// A unary predicate function.
+template <typename T1>
+bool PredFunction1(T1 v1) {
+ return v1 > 0;
+}
+
+// The following two functions are needed to circumvent a bug in
+// gcc 2.95.3, which sometimes has problem with the above template
+// function.
+bool PredFunction1Int(int v1) {
+ return v1 > 0;
+}
+bool PredFunction1Bool(Bool v1) {
+ return v1 > 0;
+}
+
+// A unary predicate functor.
+struct PredFunctor1 {
+ template <typename T1>
+ bool operator()(const T1& v1) {
+ return v1 > 0;
+ }
+};
+
+// A unary predicate-formatter function.
+template <typename T1>
+testing::AssertionResult PredFormatFunction1(const char* e1,
+ const T1& v1) {
+ if (PredFunction1(v1))
+ return testing::AssertionSuccess();
+
+ return testing::AssertionFailure()
+ << e1
+ << " is expected to be positive, but evaluates to "
+ << v1 << ".";
+}
+
+// A unary predicate-formatter functor.
+struct PredFormatFunctor1 {
+ template <typename T1>
+ testing::AssertionResult operator()(const char* e1,
+ const T1& v1) const {
+ return PredFormatFunction1(e1, v1);
+ }
+};
+
+// Tests for {EXPECT|ASSERT}_PRED_FORMAT1.
+
+class Predicate1Test : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ expected_to_finish_ = true;
+ finished_ = false;
+ n1_ = 0;
+ }
+
+ virtual void TearDown() {
+ // Verifies that each of the predicate's arguments was evaluated
+ // exactly once.
+ EXPECT_EQ(1, n1_) <<
+ "The predicate assertion didn't evaluate argument 2 "
+ "exactly once.";
+
+ // Verifies that the control flow in the test function is expected.
+ if (expected_to_finish_ && !finished_) {
+ FAIL() << "The predicate assertion unexpactedly aborted the test.";
+ } else if (!expected_to_finish_ && finished_) {
+ FAIL() << "The failed predicate assertion didn't abort the test "
+ "as expected.";
+ }
+ }
+
+ // true iff the test function is expected to run to finish.
+ static bool expected_to_finish_;
+
+ // true iff the test function did run to finish.
+ static bool finished_;
+
+ static int n1_;
+};
+
+bool Predicate1Test::expected_to_finish_;
+bool Predicate1Test::finished_;
+int Predicate1Test::n1_;
+
+typedef Predicate1Test EXPECT_PRED_FORMAT1Test;
+typedef Predicate1Test ASSERT_PRED_FORMAT1Test;
+typedef Predicate1Test EXPECT_PRED1Test;
+typedef Predicate1Test ASSERT_PRED1Test;
+
+// Tests a successful EXPECT_PRED1 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED1Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED1(PredFunction1Int,
+ ++n1_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED1 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED1Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED1(PredFunction1Bool,
+ Bool(++n1_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED1 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED1Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED1(PredFunctor1(),
+ ++n1_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED1 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED1Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED1(PredFunctor1(),
+ Bool(++n1_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED1 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED1Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED1(PredFunction1Int,
+ n1_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED1 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED1Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED1(PredFunction1Bool,
+ Bool(n1_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED1 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED1Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED1(PredFunctor1(),
+ n1_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED1 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED1Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED1(PredFunctor1(),
+ Bool(n1_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED1 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED1Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED1(PredFunction1Int,
+ ++n1_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED1 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED1Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED1(PredFunction1Bool,
+ Bool(++n1_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED1 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED1Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED1(PredFunctor1(),
+ ++n1_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED1 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED1Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED1(PredFunctor1(),
+ Bool(++n1_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED1 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED1Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED1(PredFunction1Int,
+ n1_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED1 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED1Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED1(PredFunction1Bool,
+ Bool(n1_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED1 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED1Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED1(PredFunctor1(),
+ n1_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED1 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED1Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED1(PredFunctor1(),
+ Bool(n1_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful EXPECT_PRED_FORMAT1 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT1Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT1(PredFormatFunction1,
+ ++n1_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT1 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT1Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT1(PredFormatFunction1,
+ Bool(++n1_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT1 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT1Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT1(PredFormatFunctor1(),
+ ++n1_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT1 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT1Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT1(PredFormatFunctor1(),
+ Bool(++n1_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED_FORMAT1 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT1Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT1(PredFormatFunction1,
+ n1_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT1 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT1Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT1(PredFormatFunction1,
+ Bool(n1_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT1 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT1Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT1(PredFormatFunctor1(),
+ n1_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT1 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT1Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT1(PredFormatFunctor1(),
+ Bool(n1_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED_FORMAT1 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT1Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT1(PredFormatFunction1,
+ ++n1_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT1 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT1Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT1(PredFormatFunction1,
+ Bool(++n1_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT1 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT1Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT1(PredFormatFunctor1(),
+ ++n1_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT1 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT1Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT1(PredFormatFunctor1(),
+ Bool(++n1_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED_FORMAT1 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT1Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT1(PredFormatFunction1,
+ n1_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT1 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT1Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT1(PredFormatFunction1,
+ Bool(n1_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT1 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT1Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT1(PredFormatFunctor1(),
+ n1_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT1 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT1Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT1(PredFormatFunctor1(),
+ Bool(n1_++));
+ finished_ = true;
+ }, "");
+}
+// Sample functions/functors for testing binary predicate assertions.
+
+// A binary predicate function.
+template <typename T1, typename T2>
+bool PredFunction2(T1 v1, T2 v2) {
+ return v1 + v2 > 0;
+}
+
+// The following two functions are needed to circumvent a bug in
+// gcc 2.95.3, which sometimes has problem with the above template
+// function.
+bool PredFunction2Int(int v1, int v2) {
+ return v1 + v2 > 0;
+}
+bool PredFunction2Bool(Bool v1, Bool v2) {
+ return v1 + v2 > 0;
+}
+
+// A binary predicate functor.
+struct PredFunctor2 {
+ template <typename T1, typename T2>
+ bool operator()(const T1& v1,
+ const T2& v2) {
+ return v1 + v2 > 0;
+ }
+};
+
+// A binary predicate-formatter function.
+template <typename T1, typename T2>
+testing::AssertionResult PredFormatFunction2(const char* e1,
+ const char* e2,
+ const T1& v1,
+ const T2& v2) {
+ if (PredFunction2(v1, v2))
+ return testing::AssertionSuccess();
+
+ return testing::AssertionFailure()
+ << e1 << " + " << e2
+ << " is expected to be positive, but evaluates to "
+ << v1 + v2 << ".";
+}
+
+// A binary predicate-formatter functor.
+struct PredFormatFunctor2 {
+ template <typename T1, typename T2>
+ testing::AssertionResult operator()(const char* e1,
+ const char* e2,
+ const T1& v1,
+ const T2& v2) const {
+ return PredFormatFunction2(e1, e2, v1, v2);
+ }
+};
+
+// Tests for {EXPECT|ASSERT}_PRED_FORMAT2.
+
+class Predicate2Test : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ expected_to_finish_ = true;
+ finished_ = false;
+ n1_ = n2_ = 0;
+ }
+
+ virtual void TearDown() {
+ // Verifies that each of the predicate's arguments was evaluated
+ // exactly once.
+ EXPECT_EQ(1, n1_) <<
+ "The predicate assertion didn't evaluate argument 2 "
+ "exactly once.";
+ EXPECT_EQ(1, n2_) <<
+ "The predicate assertion didn't evaluate argument 3 "
+ "exactly once.";
+
+ // Verifies that the control flow in the test function is expected.
+ if (expected_to_finish_ && !finished_) {
+ FAIL() << "The predicate assertion unexpactedly aborted the test.";
+ } else if (!expected_to_finish_ && finished_) {
+ FAIL() << "The failed predicate assertion didn't abort the test "
+ "as expected.";
+ }
+ }
+
+ // true iff the test function is expected to run to finish.
+ static bool expected_to_finish_;
+
+ // true iff the test function did run to finish.
+ static bool finished_;
+
+ static int n1_;
+ static int n2_;
+};
+
+bool Predicate2Test::expected_to_finish_;
+bool Predicate2Test::finished_;
+int Predicate2Test::n1_;
+int Predicate2Test::n2_;
+
+typedef Predicate2Test EXPECT_PRED_FORMAT2Test;
+typedef Predicate2Test ASSERT_PRED_FORMAT2Test;
+typedef Predicate2Test EXPECT_PRED2Test;
+typedef Predicate2Test ASSERT_PRED2Test;
+
+// Tests a successful EXPECT_PRED2 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED2Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED2(PredFunction2Int,
+ ++n1_,
+ ++n2_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED2 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED2Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED2(PredFunction2Bool,
+ Bool(++n1_),
+ Bool(++n2_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED2 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED2Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED2(PredFunctor2(),
+ ++n1_,
+ ++n2_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED2 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED2Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED2(PredFunctor2(),
+ Bool(++n1_),
+ Bool(++n2_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED2 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED2Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED2(PredFunction2Int,
+ n1_++,
+ n2_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED2 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED2Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED2(PredFunction2Bool,
+ Bool(n1_++),
+ Bool(n2_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED2 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED2Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED2(PredFunctor2(),
+ n1_++,
+ n2_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED2 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED2Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED2(PredFunctor2(),
+ Bool(n1_++),
+ Bool(n2_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED2 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED2Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED2(PredFunction2Int,
+ ++n1_,
+ ++n2_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED2 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED2Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED2(PredFunction2Bool,
+ Bool(++n1_),
+ Bool(++n2_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED2 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED2Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED2(PredFunctor2(),
+ ++n1_,
+ ++n2_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED2 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED2Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED2(PredFunctor2(),
+ Bool(++n1_),
+ Bool(++n2_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED2 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED2Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED2(PredFunction2Int,
+ n1_++,
+ n2_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED2 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED2Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED2(PredFunction2Bool,
+ Bool(n1_++),
+ Bool(n2_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED2 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED2Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED2(PredFunctor2(),
+ n1_++,
+ n2_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED2 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED2Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED2(PredFunctor2(),
+ Bool(n1_++),
+ Bool(n2_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful EXPECT_PRED_FORMAT2 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT2Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT2(PredFormatFunction2,
+ ++n1_,
+ ++n2_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT2 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT2Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT2(PredFormatFunction2,
+ Bool(++n1_),
+ Bool(++n2_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT2 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT2Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT2(PredFormatFunctor2(),
+ ++n1_,
+ ++n2_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT2 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT2Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT2(PredFormatFunctor2(),
+ Bool(++n1_),
+ Bool(++n2_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED_FORMAT2 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT2Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(PredFormatFunction2,
+ n1_++,
+ n2_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT2 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT2Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(PredFormatFunction2,
+ Bool(n1_++),
+ Bool(n2_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT2 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT2Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(PredFormatFunctor2(),
+ n1_++,
+ n2_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT2 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT2Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(PredFormatFunctor2(),
+ Bool(n1_++),
+ Bool(n2_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED_FORMAT2 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT2Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT2(PredFormatFunction2,
+ ++n1_,
+ ++n2_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT2 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT2Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT2(PredFormatFunction2,
+ Bool(++n1_),
+ Bool(++n2_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT2 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT2Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT2(PredFormatFunctor2(),
+ ++n1_,
+ ++n2_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT2 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT2Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT2(PredFormatFunctor2(),
+ Bool(++n1_),
+ Bool(++n2_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED_FORMAT2 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT2Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT2(PredFormatFunction2,
+ n1_++,
+ n2_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT2 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT2Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT2(PredFormatFunction2,
+ Bool(n1_++),
+ Bool(n2_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT2 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT2Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT2(PredFormatFunctor2(),
+ n1_++,
+ n2_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT2 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT2Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT2(PredFormatFunctor2(),
+ Bool(n1_++),
+ Bool(n2_++));
+ finished_ = true;
+ }, "");
+}
+// Sample functions/functors for testing ternary predicate assertions.
+
+// A ternary predicate function.
+template <typename T1, typename T2, typename T3>
+bool PredFunction3(T1 v1, T2 v2, T3 v3) {
+ return v1 + v2 + v3 > 0;
+}
+
+// The following two functions are needed to circumvent a bug in
+// gcc 2.95.3, which sometimes has problem with the above template
+// function.
+bool PredFunction3Int(int v1, int v2, int v3) {
+ return v1 + v2 + v3 > 0;
+}
+bool PredFunction3Bool(Bool v1, Bool v2, Bool v3) {
+ return v1 + v2 + v3 > 0;
+}
+
+// A ternary predicate functor.
+struct PredFunctor3 {
+ template <typename T1, typename T2, typename T3>
+ bool operator()(const T1& v1,
+ const T2& v2,
+ const T3& v3) {
+ return v1 + v2 + v3 > 0;
+ }
+};
+
+// A ternary predicate-formatter function.
+template <typename T1, typename T2, typename T3>
+testing::AssertionResult PredFormatFunction3(const char* e1,
+ const char* e2,
+ const char* e3,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3) {
+ if (PredFunction3(v1, v2, v3))
+ return testing::AssertionSuccess();
+
+ return testing::AssertionFailure()
+ << e1 << " + " << e2 << " + " << e3
+ << " is expected to be positive, but evaluates to "
+ << v1 + v2 + v3 << ".";
+}
+
+// A ternary predicate-formatter functor.
+struct PredFormatFunctor3 {
+ template <typename T1, typename T2, typename T3>
+ testing::AssertionResult operator()(const char* e1,
+ const char* e2,
+ const char* e3,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3) const {
+ return PredFormatFunction3(e1, e2, e3, v1, v2, v3);
+ }
+};
+
+// Tests for {EXPECT|ASSERT}_PRED_FORMAT3.
+
+class Predicate3Test : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ expected_to_finish_ = true;
+ finished_ = false;
+ n1_ = n2_ = n3_ = 0;
+ }
+
+ virtual void TearDown() {
+ // Verifies that each of the predicate's arguments was evaluated
+ // exactly once.
+ EXPECT_EQ(1, n1_) <<
+ "The predicate assertion didn't evaluate argument 2 "
+ "exactly once.";
+ EXPECT_EQ(1, n2_) <<
+ "The predicate assertion didn't evaluate argument 3 "
+ "exactly once.";
+ EXPECT_EQ(1, n3_) <<
+ "The predicate assertion didn't evaluate argument 4 "
+ "exactly once.";
+
+ // Verifies that the control flow in the test function is expected.
+ if (expected_to_finish_ && !finished_) {
+ FAIL() << "The predicate assertion unexpactedly aborted the test.";
+ } else if (!expected_to_finish_ && finished_) {
+ FAIL() << "The failed predicate assertion didn't abort the test "
+ "as expected.";
+ }
+ }
+
+ // true iff the test function is expected to run to finish.
+ static bool expected_to_finish_;
+
+ // true iff the test function did run to finish.
+ static bool finished_;
+
+ static int n1_;
+ static int n2_;
+ static int n3_;
+};
+
+bool Predicate3Test::expected_to_finish_;
+bool Predicate3Test::finished_;
+int Predicate3Test::n1_;
+int Predicate3Test::n2_;
+int Predicate3Test::n3_;
+
+typedef Predicate3Test EXPECT_PRED_FORMAT3Test;
+typedef Predicate3Test ASSERT_PRED_FORMAT3Test;
+typedef Predicate3Test EXPECT_PRED3Test;
+typedef Predicate3Test ASSERT_PRED3Test;
+
+// Tests a successful EXPECT_PRED3 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED3Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED3(PredFunction3Int,
+ ++n1_,
+ ++n2_,
+ ++n3_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED3 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED3Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED3(PredFunction3Bool,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED3 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED3Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED3(PredFunctor3(),
+ ++n1_,
+ ++n2_,
+ ++n3_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED3 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED3Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED3(PredFunctor3(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED3 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED3Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED3(PredFunction3Int,
+ n1_++,
+ n2_++,
+ n3_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED3 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED3Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED3(PredFunction3Bool,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED3 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED3Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED3(PredFunctor3(),
+ n1_++,
+ n2_++,
+ n3_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED3 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED3Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED3(PredFunctor3(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED3 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED3Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED3(PredFunction3Int,
+ ++n1_,
+ ++n2_,
+ ++n3_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED3 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED3Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED3(PredFunction3Bool,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED3 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED3Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED3(PredFunctor3(),
+ ++n1_,
+ ++n2_,
+ ++n3_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED3 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED3Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED3(PredFunctor3(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED3 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED3Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED3(PredFunction3Int,
+ n1_++,
+ n2_++,
+ n3_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED3 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED3Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED3(PredFunction3Bool,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED3 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED3Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED3(PredFunctor3(),
+ n1_++,
+ n2_++,
+ n3_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED3 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED3Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED3(PredFunctor3(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful EXPECT_PRED_FORMAT3 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT3Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT3(PredFormatFunction3,
+ ++n1_,
+ ++n2_,
+ ++n3_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT3 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT3Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT3(PredFormatFunction3,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT3 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT3Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT3(PredFormatFunctor3(),
+ ++n1_,
+ ++n2_,
+ ++n3_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT3 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT3Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT3(PredFormatFunctor3(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED_FORMAT3 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT3Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT3(PredFormatFunction3,
+ n1_++,
+ n2_++,
+ n3_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT3 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT3Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT3(PredFormatFunction3,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT3 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT3Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT3(PredFormatFunctor3(),
+ n1_++,
+ n2_++,
+ n3_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT3 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT3Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT3(PredFormatFunctor3(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED_FORMAT3 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT3Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT3(PredFormatFunction3,
+ ++n1_,
+ ++n2_,
+ ++n3_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT3 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT3Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT3(PredFormatFunction3,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT3 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT3Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT3(PredFormatFunctor3(),
+ ++n1_,
+ ++n2_,
+ ++n3_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT3 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT3Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT3(PredFormatFunctor3(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED_FORMAT3 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT3Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT3(PredFormatFunction3,
+ n1_++,
+ n2_++,
+ n3_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT3 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT3Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT3(PredFormatFunction3,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT3 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT3Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT3(PredFormatFunctor3(),
+ n1_++,
+ n2_++,
+ n3_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT3 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT3Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT3(PredFormatFunctor3(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++));
+ finished_ = true;
+ }, "");
+}
+// Sample functions/functors for testing 4-ary predicate assertions.
+
+// A 4-ary predicate function.
+template <typename T1, typename T2, typename T3, typename T4>
+bool PredFunction4(T1 v1, T2 v2, T3 v3, T4 v4) {
+ return v1 + v2 + v3 + v4 > 0;
+}
+
+// The following two functions are needed to circumvent a bug in
+// gcc 2.95.3, which sometimes has problem with the above template
+// function.
+bool PredFunction4Int(int v1, int v2, int v3, int v4) {
+ return v1 + v2 + v3 + v4 > 0;
+}
+bool PredFunction4Bool(Bool v1, Bool v2, Bool v3, Bool v4) {
+ return v1 + v2 + v3 + v4 > 0;
+}
+
+// A 4-ary predicate functor.
+struct PredFunctor4 {
+ template <typename T1, typename T2, typename T3, typename T4>
+ bool operator()(const T1& v1,
+ const T2& v2,
+ const T3& v3,
+ const T4& v4) {
+ return v1 + v2 + v3 + v4 > 0;
+ }
+};
+
+// A 4-ary predicate-formatter function.
+template <typename T1, typename T2, typename T3, typename T4>
+testing::AssertionResult PredFormatFunction4(const char* e1,
+ const char* e2,
+ const char* e3,
+ const char* e4,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3,
+ const T4& v4) {
+ if (PredFunction4(v1, v2, v3, v4))
+ return testing::AssertionSuccess();
+
+ return testing::AssertionFailure()
+ << e1 << " + " << e2 << " + " << e3 << " + " << e4
+ << " is expected to be positive, but evaluates to "
+ << v1 + v2 + v3 + v4 << ".";
+}
+
+// A 4-ary predicate-formatter functor.
+struct PredFormatFunctor4 {
+ template <typename T1, typename T2, typename T3, typename T4>
+ testing::AssertionResult operator()(const char* e1,
+ const char* e2,
+ const char* e3,
+ const char* e4,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3,
+ const T4& v4) const {
+ return PredFormatFunction4(e1, e2, e3, e4, v1, v2, v3, v4);
+ }
+};
+
+// Tests for {EXPECT|ASSERT}_PRED_FORMAT4.
+
+class Predicate4Test : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ expected_to_finish_ = true;
+ finished_ = false;
+ n1_ = n2_ = n3_ = n4_ = 0;
+ }
+
+ virtual void TearDown() {
+ // Verifies that each of the predicate's arguments was evaluated
+ // exactly once.
+ EXPECT_EQ(1, n1_) <<
+ "The predicate assertion didn't evaluate argument 2 "
+ "exactly once.";
+ EXPECT_EQ(1, n2_) <<
+ "The predicate assertion didn't evaluate argument 3 "
+ "exactly once.";
+ EXPECT_EQ(1, n3_) <<
+ "The predicate assertion didn't evaluate argument 4 "
+ "exactly once.";
+ EXPECT_EQ(1, n4_) <<
+ "The predicate assertion didn't evaluate argument 5 "
+ "exactly once.";
+
+ // Verifies that the control flow in the test function is expected.
+ if (expected_to_finish_ && !finished_) {
+ FAIL() << "The predicate assertion unexpactedly aborted the test.";
+ } else if (!expected_to_finish_ && finished_) {
+ FAIL() << "The failed predicate assertion didn't abort the test "
+ "as expected.";
+ }
+ }
+
+ // true iff the test function is expected to run to finish.
+ static bool expected_to_finish_;
+
+ // true iff the test function did run to finish.
+ static bool finished_;
+
+ static int n1_;
+ static int n2_;
+ static int n3_;
+ static int n4_;
+};
+
+bool Predicate4Test::expected_to_finish_;
+bool Predicate4Test::finished_;
+int Predicate4Test::n1_;
+int Predicate4Test::n2_;
+int Predicate4Test::n3_;
+int Predicate4Test::n4_;
+
+typedef Predicate4Test EXPECT_PRED_FORMAT4Test;
+typedef Predicate4Test ASSERT_PRED_FORMAT4Test;
+typedef Predicate4Test EXPECT_PRED4Test;
+typedef Predicate4Test ASSERT_PRED4Test;
+
+// Tests a successful EXPECT_PRED4 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED4Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED4(PredFunction4Int,
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED4 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED4Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED4(PredFunction4Bool,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED4 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED4Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED4(PredFunctor4(),
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED4 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED4Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED4(PredFunctor4(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED4 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED4Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED4(PredFunction4Int,
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED4 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED4Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED4(PredFunction4Bool,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED4 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED4Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED4(PredFunctor4(),
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED4 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED4Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED4(PredFunctor4(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED4 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED4Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED4(PredFunction4Int,
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED4 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED4Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED4(PredFunction4Bool,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED4 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED4Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED4(PredFunctor4(),
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED4 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED4Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED4(PredFunctor4(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED4 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED4Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED4(PredFunction4Int,
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED4 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED4Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED4(PredFunction4Bool,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED4 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED4Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED4(PredFunctor4(),
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED4 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED4Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED4(PredFunctor4(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful EXPECT_PRED_FORMAT4 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT4Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT4(PredFormatFunction4,
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT4 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT4Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT4(PredFormatFunction4,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT4 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT4Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT4(PredFormatFunctor4(),
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT4 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT4Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT4(PredFormatFunctor4(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED_FORMAT4 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT4Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT4(PredFormatFunction4,
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT4 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT4Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT4(PredFormatFunction4,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT4 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT4Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT4(PredFormatFunctor4(),
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT4 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT4Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT4(PredFormatFunctor4(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED_FORMAT4 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT4Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT4(PredFormatFunction4,
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT4 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT4Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT4(PredFormatFunction4,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT4 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT4Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT4(PredFormatFunctor4(),
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT4 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT4Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT4(PredFormatFunctor4(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED_FORMAT4 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT4Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT4(PredFormatFunction4,
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT4 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT4Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT4(PredFormatFunction4,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT4 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT4Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT4(PredFormatFunctor4(),
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT4 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT4Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT4(PredFormatFunctor4(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++));
+ finished_ = true;
+ }, "");
+}
+// Sample functions/functors for testing 5-ary predicate assertions.
+
+// A 5-ary predicate function.
+template <typename T1, typename T2, typename T3, typename T4, typename T5>
+bool PredFunction5(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5) {
+ return v1 + v2 + v3 + v4 + v5 > 0;
+}
+
+// The following two functions are needed to circumvent a bug in
+// gcc 2.95.3, which sometimes has problem with the above template
+// function.
+bool PredFunction5Int(int v1, int v2, int v3, int v4, int v5) {
+ return v1 + v2 + v3 + v4 + v5 > 0;
+}
+bool PredFunction5Bool(Bool v1, Bool v2, Bool v3, Bool v4, Bool v5) {
+ return v1 + v2 + v3 + v4 + v5 > 0;
+}
+
+// A 5-ary predicate functor.
+struct PredFunctor5 {
+ template <typename T1, typename T2, typename T3, typename T4, typename T5>
+ bool operator()(const T1& v1,
+ const T2& v2,
+ const T3& v3,
+ const T4& v4,
+ const T5& v5) {
+ return v1 + v2 + v3 + v4 + v5 > 0;
+ }
+};
+
+// A 5-ary predicate-formatter function.
+template <typename T1, typename T2, typename T3, typename T4, typename T5>
+testing::AssertionResult PredFormatFunction5(const char* e1,
+ const char* e2,
+ const char* e3,
+ const char* e4,
+ const char* e5,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3,
+ const T4& v4,
+ const T5& v5) {
+ if (PredFunction5(v1, v2, v3, v4, v5))
+ return testing::AssertionSuccess();
+
+ return testing::AssertionFailure()
+ << e1 << " + " << e2 << " + " << e3 << " + " << e4 << " + " << e5
+ << " is expected to be positive, but evaluates to "
+ << v1 + v2 + v3 + v4 + v5 << ".";
+}
+
+// A 5-ary predicate-formatter functor.
+struct PredFormatFunctor5 {
+ template <typename T1, typename T2, typename T3, typename T4, typename T5>
+ testing::AssertionResult operator()(const char* e1,
+ const char* e2,
+ const char* e3,
+ const char* e4,
+ const char* e5,
+ const T1& v1,
+ const T2& v2,
+ const T3& v3,
+ const T4& v4,
+ const T5& v5) const {
+ return PredFormatFunction5(e1, e2, e3, e4, e5, v1, v2, v3, v4, v5);
+ }
+};
+
+// Tests for {EXPECT|ASSERT}_PRED_FORMAT5.
+
+class Predicate5Test : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ expected_to_finish_ = true;
+ finished_ = false;
+ n1_ = n2_ = n3_ = n4_ = n5_ = 0;
+ }
+
+ virtual void TearDown() {
+ // Verifies that each of the predicate's arguments was evaluated
+ // exactly once.
+ EXPECT_EQ(1, n1_) <<
+ "The predicate assertion didn't evaluate argument 2 "
+ "exactly once.";
+ EXPECT_EQ(1, n2_) <<
+ "The predicate assertion didn't evaluate argument 3 "
+ "exactly once.";
+ EXPECT_EQ(1, n3_) <<
+ "The predicate assertion didn't evaluate argument 4 "
+ "exactly once.";
+ EXPECT_EQ(1, n4_) <<
+ "The predicate assertion didn't evaluate argument 5 "
+ "exactly once.";
+ EXPECT_EQ(1, n5_) <<
+ "The predicate assertion didn't evaluate argument 6 "
+ "exactly once.";
+
+ // Verifies that the control flow in the test function is expected.
+ if (expected_to_finish_ && !finished_) {
+ FAIL() << "The predicate assertion unexpactedly aborted the test.";
+ } else if (!expected_to_finish_ && finished_) {
+ FAIL() << "The failed predicate assertion didn't abort the test "
+ "as expected.";
+ }
+ }
+
+ // true iff the test function is expected to run to finish.
+ static bool expected_to_finish_;
+
+ // true iff the test function did run to finish.
+ static bool finished_;
+
+ static int n1_;
+ static int n2_;
+ static int n3_;
+ static int n4_;
+ static int n5_;
+};
+
+bool Predicate5Test::expected_to_finish_;
+bool Predicate5Test::finished_;
+int Predicate5Test::n1_;
+int Predicate5Test::n2_;
+int Predicate5Test::n3_;
+int Predicate5Test::n4_;
+int Predicate5Test::n5_;
+
+typedef Predicate5Test EXPECT_PRED_FORMAT5Test;
+typedef Predicate5Test ASSERT_PRED_FORMAT5Test;
+typedef Predicate5Test EXPECT_PRED5Test;
+typedef Predicate5Test ASSERT_PRED5Test;
+
+// Tests a successful EXPECT_PRED5 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED5Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED5(PredFunction5Int,
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_,
+ ++n5_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED5 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED5Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED5(PredFunction5Bool,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_),
+ Bool(++n5_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED5 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED5Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED5(PredFunctor5(),
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_,
+ ++n5_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED5 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED5Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED5(PredFunctor5(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_),
+ Bool(++n5_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED5 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED5Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED5(PredFunction5Int,
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++,
+ n5_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED5 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED5Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED5(PredFunction5Bool,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++),
+ Bool(n5_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED5 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED5Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED5(PredFunctor5(),
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++,
+ n5_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED5 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED5Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED5(PredFunctor5(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++),
+ Bool(n5_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED5 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED5Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED5(PredFunction5Int,
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_,
+ ++n5_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED5 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED5Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED5(PredFunction5Bool,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_),
+ Bool(++n5_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED5 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED5Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED5(PredFunctor5(),
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_,
+ ++n5_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED5 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED5Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED5(PredFunctor5(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_),
+ Bool(++n5_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED5 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED5Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED5(PredFunction5Int,
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++,
+ n5_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED5 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED5Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED5(PredFunction5Bool,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++),
+ Bool(n5_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED5 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED5Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED5(PredFunctor5(),
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++,
+ n5_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED5 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED5Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED5(PredFunctor5(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++),
+ Bool(n5_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful EXPECT_PRED_FORMAT5 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT5Test, FunctionOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT5(PredFormatFunction5,
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_,
+ ++n5_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT5 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT5Test, FunctionOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT5(PredFormatFunction5,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_),
+ Bool(++n5_));
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT5 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT5Test, FunctorOnBuiltInTypeSuccess) {
+ EXPECT_PRED_FORMAT5(PredFormatFunctor5(),
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_,
+ ++n5_);
+ finished_ = true;
+}
+
+// Tests a successful EXPECT_PRED_FORMAT5 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT5Test, FunctorOnUserTypeSuccess) {
+ EXPECT_PRED_FORMAT5(PredFormatFunctor5(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_),
+ Bool(++n5_));
+ finished_ = true;
+}
+
+// Tests a failed EXPECT_PRED_FORMAT5 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT5Test, FunctionOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT5(PredFormatFunction5,
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++,
+ n5_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT5 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT5Test, FunctionOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT5(PredFormatFunction5,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++),
+ Bool(n5_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT5 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(EXPECT_PRED_FORMAT5Test, FunctorOnBuiltInTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT5(PredFormatFunctor5(),
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++,
+ n5_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed EXPECT_PRED_FORMAT5 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(EXPECT_PRED_FORMAT5Test, FunctorOnUserTypeFailure) {
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT5(PredFormatFunctor5(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++),
+ Bool(n5_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a successful ASSERT_PRED_FORMAT5 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT5Test, FunctionOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT5(PredFormatFunction5,
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_,
+ ++n5_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT5 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT5Test, FunctionOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT5(PredFormatFunction5,
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_),
+ Bool(++n5_));
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT5 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT5Test, FunctorOnBuiltInTypeSuccess) {
+ ASSERT_PRED_FORMAT5(PredFormatFunctor5(),
+ ++n1_,
+ ++n2_,
+ ++n3_,
+ ++n4_,
+ ++n5_);
+ finished_ = true;
+}
+
+// Tests a successful ASSERT_PRED_FORMAT5 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT5Test, FunctorOnUserTypeSuccess) {
+ ASSERT_PRED_FORMAT5(PredFormatFunctor5(),
+ Bool(++n1_),
+ Bool(++n2_),
+ Bool(++n3_),
+ Bool(++n4_),
+ Bool(++n5_));
+ finished_ = true;
+}
+
+// Tests a failed ASSERT_PRED_FORMAT5 where the
+// predicate-formatter is a function on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT5Test, FunctionOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT5(PredFormatFunction5,
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++,
+ n5_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT5 where the
+// predicate-formatter is a function on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT5Test, FunctionOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT5(PredFormatFunction5,
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++),
+ Bool(n5_++));
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT5 where the
+// predicate-formatter is a functor on a built-in type (int).
+TEST_F(ASSERT_PRED_FORMAT5Test, FunctorOnBuiltInTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT5(PredFormatFunctor5(),
+ n1_++,
+ n2_++,
+ n3_++,
+ n4_++,
+ n5_++);
+ finished_ = true;
+ }, "");
+}
+
+// Tests a failed ASSERT_PRED_FORMAT5 where the
+// predicate-formatter is a functor on a user-defined type (Bool).
+TEST_F(ASSERT_PRED_FORMAT5Test, FunctorOnUserTypeFailure) {
+ expected_to_finish_ = false;
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT5(PredFormatFunctor5(),
+ Bool(n1_++),
+ Bool(n2_++),
+ Bool(n3_++),
+ Bool(n4_++),
+ Bool(n5_++));
+ finished_ = true;
+ }, "");
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_prod_test.cc b/libvpx/third_party/googletest/src/test/gtest_prod_test.cc
new file mode 100644
index 0000000..060abce
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_prod_test.cc
@@ -0,0 +1,57 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Unit test for include/gtest/gtest_prod.h.
+
+#include "gtest/gtest.h"
+#include "test/production.h"
+
+// Tests that private members can be accessed from a TEST declared as
+// a friend of the class.
+TEST(PrivateCodeTest, CanAccessPrivateMembers) {
+ PrivateCode a;
+ EXPECT_EQ(0, a.x_);
+
+ a.set_x(1);
+ EXPECT_EQ(1, a.x_);
+}
+
+typedef testing::Test PrivateCodeFixtureTest;
+
+// Tests that private members can be accessed from a TEST_F declared
+// as a friend of the class.
+TEST_F(PrivateCodeFixtureTest, CanAccessPrivateMembers) {
+ PrivateCode a;
+ EXPECT_EQ(0, a.x_);
+
+ a.set_x(2);
+ EXPECT_EQ(2, a.x_);
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_repeat_test.cc b/libvpx/third_party/googletest/src/test/gtest_repeat_test.cc
new file mode 100644
index 0000000..5223dc0
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_repeat_test.cc
@@ -0,0 +1,253 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Tests the --gtest_repeat=number flag.
+
+#include <stdlib.h>
+#include <iostream>
+#include "gtest/gtest.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+
+GTEST_DECLARE_string_(death_test_style);
+GTEST_DECLARE_string_(filter);
+GTEST_DECLARE_int32_(repeat);
+
+} // namespace testing
+
+using testing::GTEST_FLAG(death_test_style);
+using testing::GTEST_FLAG(filter);
+using testing::GTEST_FLAG(repeat);
+
+namespace {
+
+// We need this when we are testing Google Test itself and therefore
+// cannot use Google Test assertions.
+#define GTEST_CHECK_INT_EQ_(expected, actual) \
+ do {\
+ const int expected_val = (expected);\
+ const int actual_val = (actual);\
+ if (::testing::internal::IsTrue(expected_val != actual_val)) {\
+ ::std::cout << "Value of: " #actual "\n"\
+ << " Actual: " << actual_val << "\n"\
+ << "Expected: " #expected "\n"\
+ << "Which is: " << expected_val << "\n";\
+ ::testing::internal::posix::Abort();\
+ }\
+ } while(::testing::internal::AlwaysFalse())
+
+
+// Used for verifying that global environment set-up and tear-down are
+// inside the gtest_repeat loop.
+
+int g_environment_set_up_count = 0;
+int g_environment_tear_down_count = 0;
+
+class MyEnvironment : public testing::Environment {
+ public:
+ MyEnvironment() {}
+ virtual void SetUp() { g_environment_set_up_count++; }
+ virtual void TearDown() { g_environment_tear_down_count++; }
+};
+
+// A test that should fail.
+
+int g_should_fail_count = 0;
+
+TEST(FooTest, ShouldFail) {
+ g_should_fail_count++;
+ EXPECT_EQ(0, 1) << "Expected failure.";
+}
+
+// A test that should pass.
+
+int g_should_pass_count = 0;
+
+TEST(FooTest, ShouldPass) {
+ g_should_pass_count++;
+}
+
+// A test that contains a thread-safe death test and a fast death
+// test. It should pass.
+
+int g_death_test_count = 0;
+
+TEST(BarDeathTest, ThreadSafeAndFast) {
+ g_death_test_count++;
+
+ GTEST_FLAG(death_test_style) = "threadsafe";
+ EXPECT_DEATH_IF_SUPPORTED(::testing::internal::posix::Abort(), "");
+
+ GTEST_FLAG(death_test_style) = "fast";
+ EXPECT_DEATH_IF_SUPPORTED(::testing::internal::posix::Abort(), "");
+}
+
+#if GTEST_HAS_PARAM_TEST
+int g_param_test_count = 0;
+
+const int kNumberOfParamTests = 10;
+
+class MyParamTest : public testing::TestWithParam<int> {};
+
+TEST_P(MyParamTest, ShouldPass) {
+ // TODO(vladl@google.com): Make parameter value checking robust
+ // WRT order of tests.
+ GTEST_CHECK_INT_EQ_(g_param_test_count % kNumberOfParamTests, GetParam());
+ g_param_test_count++;
+}
+INSTANTIATE_TEST_CASE_P(MyParamSequence,
+ MyParamTest,
+ testing::Range(0, kNumberOfParamTests));
+#endif // GTEST_HAS_PARAM_TEST
+
+// Resets the count for each test.
+void ResetCounts() {
+ g_environment_set_up_count = 0;
+ g_environment_tear_down_count = 0;
+ g_should_fail_count = 0;
+ g_should_pass_count = 0;
+ g_death_test_count = 0;
+#if GTEST_HAS_PARAM_TEST
+ g_param_test_count = 0;
+#endif // GTEST_HAS_PARAM_TEST
+}
+
+// Checks that the count for each test is expected.
+void CheckCounts(int expected) {
+ GTEST_CHECK_INT_EQ_(expected, g_environment_set_up_count);
+ GTEST_CHECK_INT_EQ_(expected, g_environment_tear_down_count);
+ GTEST_CHECK_INT_EQ_(expected, g_should_fail_count);
+ GTEST_CHECK_INT_EQ_(expected, g_should_pass_count);
+ GTEST_CHECK_INT_EQ_(expected, g_death_test_count);
+#if GTEST_HAS_PARAM_TEST
+ GTEST_CHECK_INT_EQ_(expected * kNumberOfParamTests, g_param_test_count);
+#endif // GTEST_HAS_PARAM_TEST
+}
+
+// Tests the behavior of Google Test when --gtest_repeat is not specified.
+void TestRepeatUnspecified() {
+ ResetCounts();
+ GTEST_CHECK_INT_EQ_(1, RUN_ALL_TESTS());
+ CheckCounts(1);
+}
+
+// Tests the behavior of Google Test when --gtest_repeat has the given value.
+void TestRepeat(int repeat) {
+ GTEST_FLAG(repeat) = repeat;
+
+ ResetCounts();
+ GTEST_CHECK_INT_EQ_(repeat > 0 ? 1 : 0, RUN_ALL_TESTS());
+ CheckCounts(repeat);
+}
+
+// Tests using --gtest_repeat when --gtest_filter specifies an empty
+// set of tests.
+void TestRepeatWithEmptyFilter(int repeat) {
+ GTEST_FLAG(repeat) = repeat;
+ GTEST_FLAG(filter) = "None";
+
+ ResetCounts();
+ GTEST_CHECK_INT_EQ_(0, RUN_ALL_TESTS());
+ CheckCounts(0);
+}
+
+// Tests using --gtest_repeat when --gtest_filter specifies a set of
+// successful tests.
+void TestRepeatWithFilterForSuccessfulTests(int repeat) {
+ GTEST_FLAG(repeat) = repeat;
+ GTEST_FLAG(filter) = "*-*ShouldFail";
+
+ ResetCounts();
+ GTEST_CHECK_INT_EQ_(0, RUN_ALL_TESTS());
+ GTEST_CHECK_INT_EQ_(repeat, g_environment_set_up_count);
+ GTEST_CHECK_INT_EQ_(repeat, g_environment_tear_down_count);
+ GTEST_CHECK_INT_EQ_(0, g_should_fail_count);
+ GTEST_CHECK_INT_EQ_(repeat, g_should_pass_count);
+ GTEST_CHECK_INT_EQ_(repeat, g_death_test_count);
+#if GTEST_HAS_PARAM_TEST
+ GTEST_CHECK_INT_EQ_(repeat * kNumberOfParamTests, g_param_test_count);
+#endif // GTEST_HAS_PARAM_TEST
+}
+
+// Tests using --gtest_repeat when --gtest_filter specifies a set of
+// failed tests.
+void TestRepeatWithFilterForFailedTests(int repeat) {
+ GTEST_FLAG(repeat) = repeat;
+ GTEST_FLAG(filter) = "*ShouldFail";
+
+ ResetCounts();
+ GTEST_CHECK_INT_EQ_(1, RUN_ALL_TESTS());
+ GTEST_CHECK_INT_EQ_(repeat, g_environment_set_up_count);
+ GTEST_CHECK_INT_EQ_(repeat, g_environment_tear_down_count);
+ GTEST_CHECK_INT_EQ_(repeat, g_should_fail_count);
+ GTEST_CHECK_INT_EQ_(0, g_should_pass_count);
+ GTEST_CHECK_INT_EQ_(0, g_death_test_count);
+#if GTEST_HAS_PARAM_TEST
+ GTEST_CHECK_INT_EQ_(0, g_param_test_count);
+#endif // GTEST_HAS_PARAM_TEST
+}
+
+} // namespace
+
+int main(int argc, char **argv) {
+ testing::InitGoogleTest(&argc, argv);
+ testing::AddGlobalTestEnvironment(new MyEnvironment);
+
+ TestRepeatUnspecified();
+ TestRepeat(0);
+ TestRepeat(1);
+ TestRepeat(5);
+
+ TestRepeatWithEmptyFilter(2);
+ TestRepeatWithEmptyFilter(3);
+
+ TestRepeatWithFilterForSuccessfulTests(3);
+
+ TestRepeatWithFilterForFailedTests(4);
+
+ // It would be nice to verify that the tests indeed loop forever
+ // when GTEST_FLAG(repeat) is negative, but this test will be quite
+ // complicated to write. Since this flag is for interactive
+ // debugging only and doesn't affect the normal test result, such a
+ // test would be an overkill.
+
+ printf("PASS\n");
+ return 0;
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_shuffle_test.py b/libvpx/third_party/googletest/src/test/gtest_shuffle_test.py
new file mode 100755
index 0000000..30d0303
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_shuffle_test.py
@@ -0,0 +1,325 @@
+#!/usr/bin/env python
+#
+# Copyright 2009 Google Inc. All Rights Reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Verifies that test shuffling works."""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import gtest_test_utils
+
+# Command to run the gtest_shuffle_test_ program.
+COMMAND = gtest_test_utils.GetTestExecutablePath('gtest_shuffle_test_')
+
+# The environment variables for test sharding.
+TOTAL_SHARDS_ENV_VAR = 'GTEST_TOTAL_SHARDS'
+SHARD_INDEX_ENV_VAR = 'GTEST_SHARD_INDEX'
+
+TEST_FILTER = 'A*.A:A*.B:C*'
+
+ALL_TESTS = []
+ACTIVE_TESTS = []
+FILTERED_TESTS = []
+SHARDED_TESTS = []
+
+SHUFFLED_ALL_TESTS = []
+SHUFFLED_ACTIVE_TESTS = []
+SHUFFLED_FILTERED_TESTS = []
+SHUFFLED_SHARDED_TESTS = []
+
+
+def AlsoRunDisabledTestsFlag():
+ return '--gtest_also_run_disabled_tests'
+
+
+def FilterFlag(test_filter):
+ return '--gtest_filter=%s' % (test_filter,)
+
+
+def RepeatFlag(n):
+ return '--gtest_repeat=%s' % (n,)
+
+
+def ShuffleFlag():
+ return '--gtest_shuffle'
+
+
+def RandomSeedFlag(n):
+ return '--gtest_random_seed=%s' % (n,)
+
+
+def RunAndReturnOutput(extra_env, args):
+ """Runs the test program and returns its output."""
+
+ environ_copy = os.environ.copy()
+ environ_copy.update(extra_env)
+
+ return gtest_test_utils.Subprocess([COMMAND] + args, env=environ_copy).output
+
+
+def GetTestsForAllIterations(extra_env, args):
+ """Runs the test program and returns a list of test lists.
+
+ Args:
+ extra_env: a map from environment variables to their values
+ args: command line flags to pass to gtest_shuffle_test_
+
+ Returns:
+ A list where the i-th element is the list of tests run in the i-th
+ test iteration.
+ """
+
+ test_iterations = []
+ for line in RunAndReturnOutput(extra_env, args).split('\n'):
+ if line.startswith('----'):
+ tests = []
+ test_iterations.append(tests)
+ elif line.strip():
+ tests.append(line.strip()) # 'TestCaseName.TestName'
+
+ return test_iterations
+
+
+def GetTestCases(tests):
+ """Returns a list of test cases in the given full test names.
+
+ Args:
+ tests: a list of full test names
+
+ Returns:
+ A list of test cases from 'tests', in their original order.
+ Consecutive duplicates are removed.
+ """
+
+ test_cases = []
+ for test in tests:
+ test_case = test.split('.')[0]
+ if not test_case in test_cases:
+ test_cases.append(test_case)
+
+ return test_cases
+
+
+def CalculateTestLists():
+ """Calculates the list of tests run under different flags."""
+
+ if not ALL_TESTS:
+ ALL_TESTS.extend(
+ GetTestsForAllIterations({}, [AlsoRunDisabledTestsFlag()])[0])
+
+ if not ACTIVE_TESTS:
+ ACTIVE_TESTS.extend(GetTestsForAllIterations({}, [])[0])
+
+ if not FILTERED_TESTS:
+ FILTERED_TESTS.extend(
+ GetTestsForAllIterations({}, [FilterFlag(TEST_FILTER)])[0])
+
+ if not SHARDED_TESTS:
+ SHARDED_TESTS.extend(
+ GetTestsForAllIterations({TOTAL_SHARDS_ENV_VAR: '3',
+ SHARD_INDEX_ENV_VAR: '1'},
+ [])[0])
+
+ if not SHUFFLED_ALL_TESTS:
+ SHUFFLED_ALL_TESTS.extend(GetTestsForAllIterations(
+ {}, [AlsoRunDisabledTestsFlag(), ShuffleFlag(), RandomSeedFlag(1)])[0])
+
+ if not SHUFFLED_ACTIVE_TESTS:
+ SHUFFLED_ACTIVE_TESTS.extend(GetTestsForAllIterations(
+ {}, [ShuffleFlag(), RandomSeedFlag(1)])[0])
+
+ if not SHUFFLED_FILTERED_TESTS:
+ SHUFFLED_FILTERED_TESTS.extend(GetTestsForAllIterations(
+ {}, [ShuffleFlag(), RandomSeedFlag(1), FilterFlag(TEST_FILTER)])[0])
+
+ if not SHUFFLED_SHARDED_TESTS:
+ SHUFFLED_SHARDED_TESTS.extend(
+ GetTestsForAllIterations({TOTAL_SHARDS_ENV_VAR: '3',
+ SHARD_INDEX_ENV_VAR: '1'},
+ [ShuffleFlag(), RandomSeedFlag(1)])[0])
+
+
+class GTestShuffleUnitTest(gtest_test_utils.TestCase):
+ """Tests test shuffling."""
+
+ def setUp(self):
+ CalculateTestLists()
+
+ def testShufflePreservesNumberOfTests(self):
+ self.assertEqual(len(ALL_TESTS), len(SHUFFLED_ALL_TESTS))
+ self.assertEqual(len(ACTIVE_TESTS), len(SHUFFLED_ACTIVE_TESTS))
+ self.assertEqual(len(FILTERED_TESTS), len(SHUFFLED_FILTERED_TESTS))
+ self.assertEqual(len(SHARDED_TESTS), len(SHUFFLED_SHARDED_TESTS))
+
+ def testShuffleChangesTestOrder(self):
+ self.assert_(SHUFFLED_ALL_TESTS != ALL_TESTS, SHUFFLED_ALL_TESTS)
+ self.assert_(SHUFFLED_ACTIVE_TESTS != ACTIVE_TESTS, SHUFFLED_ACTIVE_TESTS)
+ self.assert_(SHUFFLED_FILTERED_TESTS != FILTERED_TESTS,
+ SHUFFLED_FILTERED_TESTS)
+ self.assert_(SHUFFLED_SHARDED_TESTS != SHARDED_TESTS,
+ SHUFFLED_SHARDED_TESTS)
+
+ def testShuffleChangesTestCaseOrder(self):
+ self.assert_(GetTestCases(SHUFFLED_ALL_TESTS) != GetTestCases(ALL_TESTS),
+ GetTestCases(SHUFFLED_ALL_TESTS))
+ self.assert_(
+ GetTestCases(SHUFFLED_ACTIVE_TESTS) != GetTestCases(ACTIVE_TESTS),
+ GetTestCases(SHUFFLED_ACTIVE_TESTS))
+ self.assert_(
+ GetTestCases(SHUFFLED_FILTERED_TESTS) != GetTestCases(FILTERED_TESTS),
+ GetTestCases(SHUFFLED_FILTERED_TESTS))
+ self.assert_(
+ GetTestCases(SHUFFLED_SHARDED_TESTS) != GetTestCases(SHARDED_TESTS),
+ GetTestCases(SHUFFLED_SHARDED_TESTS))
+
+ def testShuffleDoesNotRepeatTest(self):
+ for test in SHUFFLED_ALL_TESTS:
+ self.assertEqual(1, SHUFFLED_ALL_TESTS.count(test),
+ '%s appears more than once' % (test,))
+ for test in SHUFFLED_ACTIVE_TESTS:
+ self.assertEqual(1, SHUFFLED_ACTIVE_TESTS.count(test),
+ '%s appears more than once' % (test,))
+ for test in SHUFFLED_FILTERED_TESTS:
+ self.assertEqual(1, SHUFFLED_FILTERED_TESTS.count(test),
+ '%s appears more than once' % (test,))
+ for test in SHUFFLED_SHARDED_TESTS:
+ self.assertEqual(1, SHUFFLED_SHARDED_TESTS.count(test),
+ '%s appears more than once' % (test,))
+
+ def testShuffleDoesNotCreateNewTest(self):
+ for test in SHUFFLED_ALL_TESTS:
+ self.assert_(test in ALL_TESTS, '%s is an invalid test' % (test,))
+ for test in SHUFFLED_ACTIVE_TESTS:
+ self.assert_(test in ACTIVE_TESTS, '%s is an invalid test' % (test,))
+ for test in SHUFFLED_FILTERED_TESTS:
+ self.assert_(test in FILTERED_TESTS, '%s is an invalid test' % (test,))
+ for test in SHUFFLED_SHARDED_TESTS:
+ self.assert_(test in SHARDED_TESTS, '%s is an invalid test' % (test,))
+
+ def testShuffleIncludesAllTests(self):
+ for test in ALL_TESTS:
+ self.assert_(test in SHUFFLED_ALL_TESTS, '%s is missing' % (test,))
+ for test in ACTIVE_TESTS:
+ self.assert_(test in SHUFFLED_ACTIVE_TESTS, '%s is missing' % (test,))
+ for test in FILTERED_TESTS:
+ self.assert_(test in SHUFFLED_FILTERED_TESTS, '%s is missing' % (test,))
+ for test in SHARDED_TESTS:
+ self.assert_(test in SHUFFLED_SHARDED_TESTS, '%s is missing' % (test,))
+
+ def testShuffleLeavesDeathTestsAtFront(self):
+ non_death_test_found = False
+ for test in SHUFFLED_ACTIVE_TESTS:
+ if 'DeathTest.' in test:
+ self.assert_(not non_death_test_found,
+ '%s appears after a non-death test' % (test,))
+ else:
+ non_death_test_found = True
+
+ def _VerifyTestCasesDoNotInterleave(self, tests):
+ test_cases = []
+ for test in tests:
+ [test_case, _] = test.split('.')
+ if test_cases and test_cases[-1] != test_case:
+ test_cases.append(test_case)
+ self.assertEqual(1, test_cases.count(test_case),
+ 'Test case %s is not grouped together in %s' %
+ (test_case, tests))
+
+ def testShuffleDoesNotInterleaveTestCases(self):
+ self._VerifyTestCasesDoNotInterleave(SHUFFLED_ALL_TESTS)
+ self._VerifyTestCasesDoNotInterleave(SHUFFLED_ACTIVE_TESTS)
+ self._VerifyTestCasesDoNotInterleave(SHUFFLED_FILTERED_TESTS)
+ self._VerifyTestCasesDoNotInterleave(SHUFFLED_SHARDED_TESTS)
+
+ def testShuffleRestoresOrderAfterEachIteration(self):
+ # Get the test lists in all 3 iterations, using random seed 1, 2,
+ # and 3 respectively. Google Test picks a different seed in each
+ # iteration, and this test depends on the current implementation
+ # picking successive numbers. This dependency is not ideal, but
+ # makes the test much easier to write.
+ [tests_in_iteration1, tests_in_iteration2, tests_in_iteration3] = (
+ GetTestsForAllIterations(
+ {}, [ShuffleFlag(), RandomSeedFlag(1), RepeatFlag(3)]))
+
+ # Make sure running the tests with random seed 1 gets the same
+ # order as in iteration 1 above.
+ [tests_with_seed1] = GetTestsForAllIterations(
+ {}, [ShuffleFlag(), RandomSeedFlag(1)])
+ self.assertEqual(tests_in_iteration1, tests_with_seed1)
+
+ # Make sure running the tests with random seed 2 gets the same
+ # order as in iteration 2 above. Success means that Google Test
+ # correctly restores the test order before re-shuffling at the
+ # beginning of iteration 2.
+ [tests_with_seed2] = GetTestsForAllIterations(
+ {}, [ShuffleFlag(), RandomSeedFlag(2)])
+ self.assertEqual(tests_in_iteration2, tests_with_seed2)
+
+ # Make sure running the tests with random seed 3 gets the same
+ # order as in iteration 3 above. Success means that Google Test
+ # correctly restores the test order before re-shuffling at the
+ # beginning of iteration 3.
+ [tests_with_seed3] = GetTestsForAllIterations(
+ {}, [ShuffleFlag(), RandomSeedFlag(3)])
+ self.assertEqual(tests_in_iteration3, tests_with_seed3)
+
+ def testShuffleGeneratesNewOrderInEachIteration(self):
+ [tests_in_iteration1, tests_in_iteration2, tests_in_iteration3] = (
+ GetTestsForAllIterations(
+ {}, [ShuffleFlag(), RandomSeedFlag(1), RepeatFlag(3)]))
+
+ self.assert_(tests_in_iteration1 != tests_in_iteration2,
+ tests_in_iteration1)
+ self.assert_(tests_in_iteration1 != tests_in_iteration3,
+ tests_in_iteration1)
+ self.assert_(tests_in_iteration2 != tests_in_iteration3,
+ tests_in_iteration2)
+
+ def testShuffleShardedTestsPreservesPartition(self):
+ # If we run M tests on N shards, the same M tests should be run in
+ # total, regardless of the random seeds used by the shards.
+ [tests1] = GetTestsForAllIterations({TOTAL_SHARDS_ENV_VAR: '3',
+ SHARD_INDEX_ENV_VAR: '0'},
+ [ShuffleFlag(), RandomSeedFlag(1)])
+ [tests2] = GetTestsForAllIterations({TOTAL_SHARDS_ENV_VAR: '3',
+ SHARD_INDEX_ENV_VAR: '1'},
+ [ShuffleFlag(), RandomSeedFlag(20)])
+ [tests3] = GetTestsForAllIterations({TOTAL_SHARDS_ENV_VAR: '3',
+ SHARD_INDEX_ENV_VAR: '2'},
+ [ShuffleFlag(), RandomSeedFlag(25)])
+ sorted_sharded_tests = tests1 + tests2 + tests3
+ sorted_sharded_tests.sort()
+ sorted_active_tests = []
+ sorted_active_tests.extend(ACTIVE_TESTS)
+ sorted_active_tests.sort()
+ self.assertEqual(sorted_active_tests, sorted_sharded_tests)
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_shuffle_test_.cc b/libvpx/third_party/googletest/src/test/gtest_shuffle_test_.cc
new file mode 100644
index 0000000..0752789
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_shuffle_test_.cc
@@ -0,0 +1,104 @@
+// Copyright 2009, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Verifies that test shuffling works.
+
+#include "gtest/gtest.h"
+
+namespace {
+
+using ::testing::EmptyTestEventListener;
+using ::testing::InitGoogleTest;
+using ::testing::Message;
+using ::testing::Test;
+using ::testing::TestEventListeners;
+using ::testing::TestInfo;
+using ::testing::UnitTest;
+using ::testing::internal::String;
+using ::testing::internal::scoped_ptr;
+
+// The test methods are empty, as the sole purpose of this program is
+// to print the test names before/after shuffling.
+
+class A : public Test {};
+TEST_F(A, A) {}
+TEST_F(A, B) {}
+
+TEST(ADeathTest, A) {}
+TEST(ADeathTest, B) {}
+TEST(ADeathTest, C) {}
+
+TEST(B, A) {}
+TEST(B, B) {}
+TEST(B, C) {}
+TEST(B, DISABLED_D) {}
+TEST(B, DISABLED_E) {}
+
+TEST(BDeathTest, A) {}
+TEST(BDeathTest, B) {}
+
+TEST(C, A) {}
+TEST(C, B) {}
+TEST(C, C) {}
+TEST(C, DISABLED_D) {}
+
+TEST(CDeathTest, A) {}
+
+TEST(DISABLED_D, A) {}
+TEST(DISABLED_D, DISABLED_B) {}
+
+// This printer prints the full test names only, starting each test
+// iteration with a "----" marker.
+class TestNamePrinter : public EmptyTestEventListener {
+ public:
+ virtual void OnTestIterationStart(const UnitTest& /* unit_test */,
+ int /* iteration */) {
+ printf("----\n");
+ }
+
+ virtual void OnTestStart(const TestInfo& test_info) {
+ printf("%s.%s\n", test_info.test_case_name(), test_info.name());
+ }
+};
+
+} // namespace
+
+int main(int argc, char **argv) {
+ InitGoogleTest(&argc, argv);
+
+ // Replaces the default printer with TestNamePrinter, which prints
+ // the test name only.
+ TestEventListeners& listeners = UnitTest::GetInstance()->listeners();
+ delete listeners.Release(listeners.default_result_printer());
+ listeners.Append(new TestNamePrinter);
+
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_sole_header_test.cc b/libvpx/third_party/googletest/src/test/gtest_sole_header_test.cc
new file mode 100644
index 0000000..ccd091a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_sole_header_test.cc
@@ -0,0 +1,57 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+// This test verifies that it's possible to use Google Test by including
+// the gtest.h header file alone.
+
+#include "gtest/gtest.h"
+
+namespace {
+
+void Subroutine() {
+ EXPECT_EQ(42, 42);
+}
+
+TEST(NoFatalFailureTest, ExpectNoFatalFailure) {
+ EXPECT_NO_FATAL_FAILURE(;);
+ EXPECT_NO_FATAL_FAILURE(SUCCEED());
+ EXPECT_NO_FATAL_FAILURE(Subroutine());
+ EXPECT_NO_FATAL_FAILURE({ SUCCEED(); });
+}
+
+TEST(NoFatalFailureTest, AssertNoFatalFailure) {
+ ASSERT_NO_FATAL_FAILURE(;);
+ ASSERT_NO_FATAL_FAILURE(SUCCEED());
+ ASSERT_NO_FATAL_FAILURE(Subroutine());
+ ASSERT_NO_FATAL_FAILURE({ SUCCEED(); });
+}
+
+} // namespace
diff --git a/libvpx/third_party/googletest/src/test/gtest_stress_test.cc b/libvpx/third_party/googletest/src/test/gtest_stress_test.cc
new file mode 100644
index 0000000..4e7d9bf
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_stress_test.cc
@@ -0,0 +1,257 @@
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Tests that SCOPED_TRACE() and various Google Test assertions can be
+// used in a large number of threads concurrently.
+
+#include "gtest/gtest.h"
+
+#include <iostream>
+#include <vector>
+
+// We must define this macro in order to #include
+// gtest-internal-inl.h. This is how Google Test prevents a user from
+// accidentally depending on its internal implementation.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+#if GTEST_IS_THREADSAFE
+
+namespace testing {
+namespace {
+
+using internal::Notification;
+using internal::String;
+using internal::TestPropertyKeyIs;
+using internal::ThreadWithParam;
+using internal::scoped_ptr;
+
+// In order to run tests in this file, for platforms where Google Test is
+// thread safe, implement ThreadWithParam. See the description of its API
+// in gtest-port.h, where it is defined for already supported platforms.
+
+// How many threads to create?
+const int kThreadCount = 50;
+
+String IdToKey(int id, const char* suffix) {
+ Message key;
+ key << "key_" << id << "_" << suffix;
+ return key.GetString();
+}
+
+String IdToString(int id) {
+ Message id_message;
+ id_message << id;
+ return id_message.GetString();
+}
+
+void ExpectKeyAndValueWereRecordedForId(
+ const std::vector<TestProperty>& properties,
+ int id, const char* suffix) {
+ TestPropertyKeyIs matches_key(IdToKey(id, suffix).c_str());
+ const std::vector<TestProperty>::const_iterator property =
+ std::find_if(properties.begin(), properties.end(), matches_key);
+ ASSERT_TRUE(property != properties.end())
+ << "expecting " << suffix << " value for id " << id;
+ EXPECT_STREQ(IdToString(id).c_str(), property->value());
+}
+
+// Calls a large number of Google Test assertions, where exactly one of them
+// will fail.
+void ManyAsserts(int id) {
+ GTEST_LOG_(INFO) << "Thread #" << id << " running...";
+
+ SCOPED_TRACE(Message() << "Thread #" << id);
+
+ for (int i = 0; i < kThreadCount; i++) {
+ SCOPED_TRACE(Message() << "Iteration #" << i);
+
+ // A bunch of assertions that should succeed.
+ EXPECT_TRUE(true);
+ ASSERT_FALSE(false) << "This shouldn't fail.";
+ EXPECT_STREQ("a", "a");
+ ASSERT_LE(5, 6);
+ EXPECT_EQ(i, i) << "This shouldn't fail.";
+
+ // RecordProperty() should interact safely with other threads as well.
+ // The shared_key forces property updates.
+ Test::RecordProperty(IdToKey(id, "string").c_str(), IdToString(id).c_str());
+ Test::RecordProperty(IdToKey(id, "int").c_str(), id);
+ Test::RecordProperty("shared_key", IdToString(id).c_str());
+
+ // This assertion should fail kThreadCount times per thread. It
+ // is for testing whether Google Test can handle failed assertions in a
+ // multi-threaded context.
+ EXPECT_LT(i, 0) << "This should always fail.";
+ }
+}
+
+void CheckTestFailureCount(int expected_failures) {
+ const TestInfo* const info = UnitTest::GetInstance()->current_test_info();
+ const TestResult* const result = info->result();
+ GTEST_CHECK_(expected_failures == result->total_part_count())
+ << "Logged " << result->total_part_count() << " failures "
+ << " vs. " << expected_failures << " expected";
+}
+
+// Tests using SCOPED_TRACE() and Google Test assertions in many threads
+// concurrently.
+TEST(StressTest, CanUseScopedTraceAndAssertionsInManyThreads) {
+ {
+ scoped_ptr<ThreadWithParam<int> > threads[kThreadCount];
+ Notification threads_can_start;
+ for (int i = 0; i != kThreadCount; i++)
+ threads[i].reset(new ThreadWithParam<int>(&ManyAsserts,
+ i,
+ &threads_can_start));
+
+ threads_can_start.Notify();
+
+ // Blocks until all the threads are done.
+ for (int i = 0; i != kThreadCount; i++)
+ threads[i]->Join();
+ }
+
+ // Ensures that kThreadCount*kThreadCount failures have been reported.
+ const TestInfo* const info = UnitTest::GetInstance()->current_test_info();
+ const TestResult* const result = info->result();
+
+ std::vector<TestProperty> properties;
+ // We have no access to the TestResult's list of properties but we can
+ // copy them one by one.
+ for (int i = 0; i < result->test_property_count(); ++i)
+ properties.push_back(result->GetTestProperty(i));
+
+ EXPECT_EQ(kThreadCount * 2 + 1, result->test_property_count())
+ << "String and int values recorded on each thread, "
+ << "as well as one shared_key";
+ for (int i = 0; i < kThreadCount; ++i) {
+ ExpectKeyAndValueWereRecordedForId(properties, i, "string");
+ ExpectKeyAndValueWereRecordedForId(properties, i, "int");
+ }
+ CheckTestFailureCount(kThreadCount*kThreadCount);
+}
+
+void FailingThread(bool is_fatal) {
+ if (is_fatal)
+ FAIL() << "Fatal failure in some other thread. "
+ << "(This failure is expected.)";
+ else
+ ADD_FAILURE() << "Non-fatal failure in some other thread. "
+ << "(This failure is expected.)";
+}
+
+void GenerateFatalFailureInAnotherThread(bool is_fatal) {
+ ThreadWithParam<bool> thread(&FailingThread, is_fatal, NULL);
+ thread.Join();
+}
+
+TEST(NoFatalFailureTest, ExpectNoFatalFailureIgnoresFailuresInOtherThreads) {
+ EXPECT_NO_FATAL_FAILURE(GenerateFatalFailureInAnotherThread(true));
+ // We should only have one failure (the one from
+ // GenerateFatalFailureInAnotherThread()), since the EXPECT_NO_FATAL_FAILURE
+ // should succeed.
+ CheckTestFailureCount(1);
+}
+
+void AssertNoFatalFailureIgnoresFailuresInOtherThreads() {
+ ASSERT_NO_FATAL_FAILURE(GenerateFatalFailureInAnotherThread(true));
+}
+TEST(NoFatalFailureTest, AssertNoFatalFailureIgnoresFailuresInOtherThreads) {
+ // Using a subroutine, to make sure, that the test continues.
+ AssertNoFatalFailureIgnoresFailuresInOtherThreads();
+ // We should only have one failure (the one from
+ // GenerateFatalFailureInAnotherThread()), since the EXPECT_NO_FATAL_FAILURE
+ // should succeed.
+ CheckTestFailureCount(1);
+}
+
+TEST(FatalFailureTest, ExpectFatalFailureIgnoresFailuresInOtherThreads) {
+ // This statement should fail, since the current thread doesn't generate a
+ // fatal failure, only another one does.
+ EXPECT_FATAL_FAILURE(GenerateFatalFailureInAnotherThread(true), "expected");
+ CheckTestFailureCount(2);
+}
+
+TEST(FatalFailureOnAllThreadsTest, ExpectFatalFailureOnAllThreads) {
+ // This statement should succeed, because failures in all threads are
+ // considered.
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS(
+ GenerateFatalFailureInAnotherThread(true), "expected");
+ CheckTestFailureCount(0);
+ // We need to add a failure, because main() checks that there are failures.
+ // But when only this test is run, we shouldn't have any failures.
+ ADD_FAILURE() << "This is an expected non-fatal failure.";
+}
+
+TEST(NonFatalFailureTest, ExpectNonFatalFailureIgnoresFailuresInOtherThreads) {
+ // This statement should fail, since the current thread doesn't generate a
+ // fatal failure, only another one does.
+ EXPECT_NONFATAL_FAILURE(GenerateFatalFailureInAnotherThread(false),
+ "expected");
+ CheckTestFailureCount(2);
+}
+
+TEST(NonFatalFailureOnAllThreadsTest, ExpectNonFatalFailureOnAllThreads) {
+ // This statement should succeed, because failures in all threads are
+ // considered.
+ EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(
+ GenerateFatalFailureInAnotherThread(false), "expected");
+ CheckTestFailureCount(0);
+ // We need to add a failure, because main() checks that there are failures,
+ // But when only this test is run, we shouldn't have any failures.
+ ADD_FAILURE() << "This is an expected non-fatal failure.";
+}
+
+} // namespace
+} // namespace testing
+
+int main(int argc, char **argv) {
+ testing::InitGoogleTest(&argc, argv);
+
+ const int result = RUN_ALL_TESTS(); // Expected to fail.
+ GTEST_CHECK_(result == 1) << "RUN_ALL_TESTS() did not fail as expected";
+
+ printf("\nPASS\n");
+ return 0;
+}
+
+#else
+TEST(StressTest,
+ DISABLED_ThreadSafetyTestsAreSkippedWhenGoogleTestIsNotThreadSafe) {
+}
+
+int main(int argc, char **argv) {
+ testing::InitGoogleTest(&argc, argv);
+ return RUN_ALL_TESTS();
+}
+#endif // GTEST_IS_THREADSAFE
diff --git a/libvpx/third_party/googletest/src/test/gtest_test_utils.py b/libvpx/third_party/googletest/src/test/gtest_test_utils.py
new file mode 100755
index 0000000..4e897bd
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_test_utils.py
@@ -0,0 +1,305 @@
+#!/usr/bin/env python
+#
+# Copyright 2006, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Unit test utilities for Google C++ Testing Framework."""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import atexit
+import os
+import shutil
+import sys
+import tempfile
+import unittest
+_test_module = unittest
+
+# Suppresses the 'Import not at the top of the file' lint complaint.
+# pylint: disable-msg=C6204
+try:
+ import subprocess
+ _SUBPROCESS_MODULE_AVAILABLE = True
+except:
+ import popen2
+ _SUBPROCESS_MODULE_AVAILABLE = False
+# pylint: enable-msg=C6204
+
+GTEST_OUTPUT_VAR_NAME = 'GTEST_OUTPUT'
+
+IS_WINDOWS = os.name == 'nt'
+IS_CYGWIN = os.name == 'posix' and 'CYGWIN' in os.uname()[0]
+
+# Here we expose a class from a particular module, depending on the
+# environment. The comment suppresses the 'Invalid variable name' lint
+# complaint.
+TestCase = _test_module.TestCase # pylint: disable-msg=C6409
+
+# Initially maps a flag to its default value. After
+# _ParseAndStripGTestFlags() is called, maps a flag to its actual value.
+_flag_map = {'source_dir': os.path.dirname(sys.argv[0]),
+ 'build_dir': os.path.dirname(sys.argv[0])}
+_gtest_flags_are_parsed = False
+
+
+def _ParseAndStripGTestFlags(argv):
+ """Parses and strips Google Test flags from argv. This is idempotent."""
+
+ # Suppresses the lint complaint about a global variable since we need it
+ # here to maintain module-wide state.
+ global _gtest_flags_are_parsed # pylint: disable-msg=W0603
+ if _gtest_flags_are_parsed:
+ return
+
+ _gtest_flags_are_parsed = True
+ for flag in _flag_map:
+ # The environment variable overrides the default value.
+ if flag.upper() in os.environ:
+ _flag_map[flag] = os.environ[flag.upper()]
+
+ # The command line flag overrides the environment variable.
+ i = 1 # Skips the program name.
+ while i < len(argv):
+ prefix = '--' + flag + '='
+ if argv[i].startswith(prefix):
+ _flag_map[flag] = argv[i][len(prefix):]
+ del argv[i]
+ break
+ else:
+ # We don't increment i in case we just found a --gtest_* flag
+ # and removed it from argv.
+ i += 1
+
+
+def GetFlag(flag):
+ """Returns the value of the given flag."""
+
+ # In case GetFlag() is called before Main(), we always call
+ # _ParseAndStripGTestFlags() here to make sure the --gtest_* flags
+ # are parsed.
+ _ParseAndStripGTestFlags(sys.argv)
+
+ return _flag_map[flag]
+
+
+def GetSourceDir():
+ """Returns the absolute path of the directory where the .py files are."""
+
+ return os.path.abspath(GetFlag('source_dir'))
+
+
+def GetBuildDir():
+ """Returns the absolute path of the directory where the test binaries are."""
+
+ return os.path.abspath(GetFlag('build_dir'))
+
+
+_temp_dir = None
+
+def _RemoveTempDir():
+ if _temp_dir:
+ shutil.rmtree(_temp_dir, ignore_errors=True)
+
+atexit.register(_RemoveTempDir)
+
+
+def GetTempDir():
+ """Returns a directory for temporary files."""
+
+ global _temp_dir
+ if not _temp_dir:
+ _temp_dir = tempfile.mkdtemp()
+ return _temp_dir
+
+
+def GetTestExecutablePath(executable_name, build_dir=None):
+ """Returns the absolute path of the test binary given its name.
+
+ The function will print a message and abort the program if the resulting file
+ doesn't exist.
+
+ Args:
+ executable_name: name of the test binary that the test script runs.
+ build_dir: directory where to look for executables, by default
+ the result of GetBuildDir().
+
+ Returns:
+ The absolute path of the test binary.
+ """
+
+ path = os.path.abspath(os.path.join(build_dir or GetBuildDir(),
+ executable_name))
+ if (IS_WINDOWS or IS_CYGWIN) and not path.endswith('.exe'):
+ path += '.exe'
+
+ if not os.path.exists(path):
+ message = (
+ 'Unable to find the test binary. Please make sure to provide path\n'
+ 'to the binary via the --build_dir flag or the BUILD_DIR\n'
+ 'environment variable.')
+ print >> sys.stderr, message
+ sys.exit(1)
+
+ return path
+
+
+def GetExitStatus(exit_code):
+ """Returns the argument to exit(), or -1 if exit() wasn't called.
+
+ Args:
+ exit_code: the result value of os.system(command).
+ """
+
+ if os.name == 'nt':
+ # On Windows, os.WEXITSTATUS() doesn't work and os.system() returns
+ # the argument to exit() directly.
+ return exit_code
+ else:
+ # On Unix, os.WEXITSTATUS() must be used to extract the exit status
+ # from the result of os.system().
+ if os.WIFEXITED(exit_code):
+ return os.WEXITSTATUS(exit_code)
+ else:
+ return -1
+
+
+class Subprocess:
+ def __init__(self, command, working_dir=None, capture_stderr=True, env=None):
+ """Changes into a specified directory, if provided, and executes a command.
+
+ Restores the old directory afterwards.
+
+ Args:
+ command: The command to run, in the form of sys.argv.
+ working_dir: The directory to change into.
+ capture_stderr: Determines whether to capture stderr in the output member
+ or to discard it.
+ env: Dictionary with environment to pass to the subprocess.
+
+ Returns:
+ An object that represents outcome of the executed process. It has the
+ following attributes:
+ terminated_by_signal True iff the child process has been terminated
+ by a signal.
+ signal Sygnal that terminated the child process.
+ exited True iff the child process exited normally.
+ exit_code The code with which the child process exited.
+ output Child process's stdout and stderr output
+ combined in a string.
+ """
+
+ # The subprocess module is the preferrable way of running programs
+ # since it is available and behaves consistently on all platforms,
+ # including Windows. But it is only available starting in python 2.4.
+ # In earlier python versions, we revert to the popen2 module, which is
+ # available in python 2.0 and later but doesn't provide required
+ # functionality (Popen4) under Windows. This allows us to support Mac
+ # OS X 10.4 Tiger, which has python 2.3 installed.
+ if _SUBPROCESS_MODULE_AVAILABLE:
+ if capture_stderr:
+ stderr = subprocess.STDOUT
+ else:
+ stderr = subprocess.PIPE
+
+ p = subprocess.Popen(command,
+ stdout=subprocess.PIPE, stderr=stderr,
+ cwd=working_dir, universal_newlines=True, env=env)
+ # communicate returns a tuple with the file obect for the child's
+ # output.
+ self.output = p.communicate()[0]
+ self._return_code = p.returncode
+ else:
+ old_dir = os.getcwd()
+
+ def _ReplaceEnvDict(dest, src):
+ # Changes made by os.environ.clear are not inheritable by child
+ # processes until Python 2.6. To produce inheritable changes we have
+ # to delete environment items with the del statement.
+ for key in dest:
+ del dest[key]
+ dest.update(src)
+
+ # When 'env' is not None, backup the environment variables and replace
+ # them with the passed 'env'. When 'env' is None, we simply use the
+ # current 'os.environ' for compatibility with the subprocess.Popen
+ # semantics used above.
+ if env is not None:
+ old_environ = os.environ.copy()
+ _ReplaceEnvDict(os.environ, env)
+
+ try:
+ if working_dir is not None:
+ os.chdir(working_dir)
+ if capture_stderr:
+ p = popen2.Popen4(command)
+ else:
+ p = popen2.Popen3(command)
+ p.tochild.close()
+ self.output = p.fromchild.read()
+ ret_code = p.wait()
+ finally:
+ os.chdir(old_dir)
+
+ # Restore the old environment variables
+ # if they were replaced.
+ if env is not None:
+ _ReplaceEnvDict(os.environ, old_environ)
+
+ # Converts ret_code to match the semantics of
+ # subprocess.Popen.returncode.
+ if os.WIFSIGNALED(ret_code):
+ self._return_code = -os.WTERMSIG(ret_code)
+ else: # os.WIFEXITED(ret_code) should return True here.
+ self._return_code = os.WEXITSTATUS(ret_code)
+
+ if self._return_code < 0:
+ self.terminated_by_signal = True
+ self.exited = False
+ self.signal = -self._return_code
+ else:
+ self.terminated_by_signal = False
+ self.exited = True
+ self.exit_code = self._return_code
+
+
+def Main():
+ """Runs the unit test."""
+
+ # We must call _ParseAndStripGTestFlags() before calling
+ # unittest.main(). Otherwise the latter will be confused by the
+ # --gtest_* flags.
+ _ParseAndStripGTestFlags(sys.argv)
+ # The tested binaries should not be writing XML output files unless the
+ # script explicitly instructs them to.
+ # TODO(vladl@google.com): Move this into Subprocess when we implement
+ # passing environment into it as a parameter.
+ if GTEST_OUTPUT_VAR_NAME in os.environ:
+ del os.environ[GTEST_OUTPUT_VAR_NAME]
+
+ _test_module.main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_ex_test.cc b/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_ex_test.cc
new file mode 100644
index 0000000..8d46c76
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_ex_test.cc
@@ -0,0 +1,92 @@
+// Copyright 2009, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Tests Google Test's throw-on-failure mode with exceptions enabled.
+
+#include "gtest/gtest.h"
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdexcept>
+
+// Prints the given failure message and exits the program with
+// non-zero. We use this instead of a Google Test assertion to
+// indicate a failure, as the latter is been tested and cannot be
+// relied on.
+void Fail(const char* msg) {
+ printf("FAILURE: %s\n", msg);
+ fflush(stdout);
+ exit(1);
+}
+
+// Tests that an assertion failure throws a subclass of
+// std::runtime_error.
+void TestFailureThrowsRuntimeError() {
+ testing::GTEST_FLAG(throw_on_failure) = true;
+
+ // A successful assertion shouldn't throw.
+ try {
+ EXPECT_EQ(3, 3);
+ } catch(...) {
+ Fail("A successful assertion wrongfully threw.");
+ }
+
+ // A failed assertion should throw a subclass of std::runtime_error.
+ try {
+ EXPECT_EQ(2, 3) << "Expected failure";
+ } catch(const std::runtime_error& e) {
+ if (strstr(e.what(), "Expected failure") != NULL)
+ return;
+
+ printf("%s",
+ "A failed assertion did throw an exception of the right type, "
+ "but the message is incorrect. Instead of containing \"Expected "
+ "failure\", it is:\n");
+ Fail(e.what());
+ } catch(...) {
+ Fail("A failed assertion threw the wrong type of exception.");
+ }
+ Fail("A failed assertion should've thrown but didn't.");
+}
+
+int main(int argc, char** argv) {
+ testing::InitGoogleTest(&argc, argv);
+
+ // We want to ensure that people can use Google Test assertions in
+ // other testing frameworks, as long as they initialize Google Test
+ // properly and set the thrown-on-failure mode. Therefore, we don't
+ // use Google Test's constructs for defining and running tests
+ // (e.g. TEST and RUN_ALL_TESTS) here.
+
+ TestFailureThrowsRuntimeError();
+ return 0;
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_test.py b/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_test.py
new file mode 100755
index 0000000..5678ffe
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_test.py
@@ -0,0 +1,171 @@
+#!/usr/bin/env python
+#
+# Copyright 2009, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Tests Google Test's throw-on-failure mode with exceptions disabled.
+
+This script invokes gtest_throw_on_failure_test_ (a program written with
+Google Test) with different environments and command line flags.
+"""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import os
+import gtest_test_utils
+
+
+# Constants.
+
+# The command line flag for enabling/disabling the throw-on-failure mode.
+THROW_ON_FAILURE = 'gtest_throw_on_failure'
+
+# Path to the gtest_throw_on_failure_test_ program, compiled with
+# exceptions disabled.
+EXE_PATH = gtest_test_utils.GetTestExecutablePath(
+ 'gtest_throw_on_failure_test_')
+
+
+# Utilities.
+
+
+def SetEnvVar(env_var, value):
+ """Sets an environment variable to a given value; unsets it when the
+ given value is None.
+ """
+
+ env_var = env_var.upper()
+ if value is not None:
+ os.environ[env_var] = value
+ elif env_var in os.environ:
+ del os.environ[env_var]
+
+
+def Run(command):
+ """Runs a command; returns True/False if its exit code is/isn't 0."""
+
+ print 'Running "%s". . .' % ' '.join(command)
+ p = gtest_test_utils.Subprocess(command)
+ return p.exited and p.exit_code == 0
+
+
+# The tests. TODO(wan@google.com): refactor the class to share common
+# logic with code in gtest_break_on_failure_unittest.py.
+class ThrowOnFailureTest(gtest_test_utils.TestCase):
+ """Tests the throw-on-failure mode."""
+
+ def RunAndVerify(self, env_var_value, flag_value, should_fail):
+ """Runs gtest_throw_on_failure_test_ and verifies that it does
+ (or does not) exit with a non-zero code.
+
+ Args:
+ env_var_value: value of the GTEST_BREAK_ON_FAILURE environment
+ variable; None if the variable should be unset.
+ flag_value: value of the --gtest_break_on_failure flag;
+ None if the flag should not be present.
+ should_fail: True iff the program is expected to fail.
+ """
+
+ SetEnvVar(THROW_ON_FAILURE, env_var_value)
+
+ if env_var_value is None:
+ env_var_value_msg = ' is not set'
+ else:
+ env_var_value_msg = '=' + env_var_value
+
+ if flag_value is None:
+ flag = ''
+ elif flag_value == '0':
+ flag = '--%s=0' % THROW_ON_FAILURE
+ else:
+ flag = '--%s' % THROW_ON_FAILURE
+
+ command = [EXE_PATH]
+ if flag:
+ command.append(flag)
+
+ if should_fail:
+ should_or_not = 'should'
+ else:
+ should_or_not = 'should not'
+
+ failed = not Run(command)
+
+ SetEnvVar(THROW_ON_FAILURE, None)
+
+ msg = ('when %s%s, an assertion failure in "%s" %s cause a non-zero '
+ 'exit code.' %
+ (THROW_ON_FAILURE, env_var_value_msg, ' '.join(command),
+ should_or_not))
+ self.assert_(failed == should_fail, msg)
+
+ def testDefaultBehavior(self):
+ """Tests the behavior of the default mode."""
+
+ self.RunAndVerify(env_var_value=None, flag_value=None, should_fail=False)
+
+ def testThrowOnFailureEnvVar(self):
+ """Tests using the GTEST_THROW_ON_FAILURE environment variable."""
+
+ self.RunAndVerify(env_var_value='0',
+ flag_value=None,
+ should_fail=False)
+ self.RunAndVerify(env_var_value='1',
+ flag_value=None,
+ should_fail=True)
+
+ def testThrowOnFailureFlag(self):
+ """Tests using the --gtest_throw_on_failure flag."""
+
+ self.RunAndVerify(env_var_value=None,
+ flag_value='0',
+ should_fail=False)
+ self.RunAndVerify(env_var_value=None,
+ flag_value='1',
+ should_fail=True)
+
+ def testThrowOnFailureFlagOverridesEnvVar(self):
+ """Tests that --gtest_throw_on_failure overrides GTEST_THROW_ON_FAILURE."""
+
+ self.RunAndVerify(env_var_value='0',
+ flag_value='0',
+ should_fail=False)
+ self.RunAndVerify(env_var_value='0',
+ flag_value='1',
+ should_fail=True)
+ self.RunAndVerify(env_var_value='1',
+ flag_value='0',
+ should_fail=False)
+ self.RunAndVerify(env_var_value='1',
+ flag_value='1',
+ should_fail=True)
+
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_test_.cc b/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_test_.cc
new file mode 100644
index 0000000..03776ec
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_throw_on_failure_test_.cc
@@ -0,0 +1,56 @@
+// Copyright 2009, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Tests Google Test's throw-on-failure mode with exceptions disabled.
+//
+// This program must be compiled with exceptions disabled. It will be
+// invoked by gtest_throw_on_failure_test.py, and is expected to exit
+// with non-zero in the throw-on-failure mode or 0 otherwise.
+
+#include "gtest/gtest.h"
+
+int main(int argc, char** argv) {
+ testing::InitGoogleTest(&argc, argv);
+
+ // We want to ensure that people can use Google Test assertions in
+ // other testing frameworks, as long as they initialize Google Test
+ // properly and set the thrown-on-failure mode. Therefore, we don't
+ // use Google Test's constructs for defining and running tests
+ // (e.g. TEST and RUN_ALL_TESTS) here.
+
+ // In the throw-on-failure mode with exceptions disabled, this
+ // assertion will cause the program to exit with a non-zero code.
+ EXPECT_EQ(2, 3);
+
+ // When not in the throw-on-failure mode, the control will reach
+ // here.
+ return 0;
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_uninitialized_test.py b/libvpx/third_party/googletest/src/test/gtest_uninitialized_test.py
new file mode 100755
index 0000000..6ae57ee
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_uninitialized_test.py
@@ -0,0 +1,70 @@
+#!/usr/bin/env python
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Verifies that Google Test warns the user when not initialized properly."""
+
+__author__ = 'wan@google.com (Zhanyong Wan)'
+
+import gtest_test_utils
+
+
+COMMAND = gtest_test_utils.GetTestExecutablePath('gtest_uninitialized_test_')
+
+
+def Assert(condition):
+ if not condition:
+ raise AssertionError
+
+
+def AssertEq(expected, actual):
+ if expected != actual:
+ print 'Expected: %s' % (expected,)
+ print ' Actual: %s' % (actual,)
+ raise AssertionError
+
+
+def TestExitCodeAndOutput(command):
+ """Runs the given command and verifies its exit code and output."""
+
+ # Verifies that 'command' exits with code 1.
+ p = gtest_test_utils.Subprocess(command)
+ Assert(p.exited)
+ AssertEq(1, p.exit_code)
+ Assert('InitGoogleTest' in p.output)
+
+
+class GTestUninitializedTest(gtest_test_utils.TestCase):
+ def testExitCodeAndOutput(self):
+ TestExitCodeAndOutput(COMMAND)
+
+
+if __name__ == '__main__':
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_uninitialized_test_.cc b/libvpx/third_party/googletest/src/test/gtest_uninitialized_test_.cc
new file mode 100644
index 0000000..4431698
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_uninitialized_test_.cc
@@ -0,0 +1,43 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#include "gtest/gtest.h"
+
+TEST(DummyTest, Dummy) {
+ // This test doesn't verify anything. We just need it to create a
+ // realistic stage for testing the behavior of Google Test when
+ // RUN_ALL_TESTS() is called without testing::InitGoogleTest() being
+ // called first.
+}
+
+int main() {
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_unittest.cc b/libvpx/third_party/googletest/src/test/gtest_unittest.cc
new file mode 100644
index 0000000..23d6860
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_unittest.cc
@@ -0,0 +1,7337 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Tests for Google Test itself. This verifies that the basic constructs of
+// Google Test work.
+
+#include "gtest/gtest.h"
+#include <vector>
+#include <ostream>
+
+// Verifies that the command line flag variables can be accessed
+// in code once <gtest/gtest.h> has been #included.
+// Do not move it after other #includes.
+TEST(CommandLineFlagsTest, CanBeAccessedInCodeOnceGTestHIsIncluded) {
+ bool dummy = testing::GTEST_FLAG(also_run_disabled_tests)
+ || testing::GTEST_FLAG(break_on_failure)
+ || testing::GTEST_FLAG(catch_exceptions)
+ || testing::GTEST_FLAG(color) != "unknown"
+ || testing::GTEST_FLAG(filter) != "unknown"
+ || testing::GTEST_FLAG(list_tests)
+ || testing::GTEST_FLAG(output) != "unknown"
+ || testing::GTEST_FLAG(print_time)
+ || testing::GTEST_FLAG(random_seed)
+ || testing::GTEST_FLAG(repeat) > 0
+ || testing::GTEST_FLAG(show_internal_stack_frames)
+ || testing::GTEST_FLAG(shuffle)
+ || testing::GTEST_FLAG(stack_trace_depth) > 0
+ || testing::GTEST_FLAG(stream_result_to) != "unknown"
+ || testing::GTEST_FLAG(throw_on_failure);
+ EXPECT_TRUE(dummy || !dummy); // Suppresses warning that dummy is unused.
+}
+
+#include "gtest/gtest-spi.h"
+
+// Indicates that this translation unit is part of Google Test's
+// implementation. It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error. This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+#include "src/gtest-internal-inl.h"
+#undef GTEST_IMPLEMENTATION_
+
+#include <limits.h> // For INT_MAX.
+#include <stdlib.h>
+#include <time.h>
+
+#include <map>
+
+namespace testing {
+namespace internal {
+
+// Provides access to otherwise private parts of the TestEventListeners class
+// that are needed to test it.
+class TestEventListenersAccessor {
+ public:
+ static TestEventListener* GetRepeater(TestEventListeners* listeners) {
+ return listeners->repeater();
+ }
+
+ static void SetDefaultResultPrinter(TestEventListeners* listeners,
+ TestEventListener* listener) {
+ listeners->SetDefaultResultPrinter(listener);
+ }
+ static void SetDefaultXmlGenerator(TestEventListeners* listeners,
+ TestEventListener* listener) {
+ listeners->SetDefaultXmlGenerator(listener);
+ }
+
+ static bool EventForwardingEnabled(const TestEventListeners& listeners) {
+ return listeners.EventForwardingEnabled();
+ }
+
+ static void SuppressEventForwarding(TestEventListeners* listeners) {
+ listeners->SuppressEventForwarding();
+ }
+};
+
+} // namespace internal
+} // namespace testing
+
+using testing::AssertionFailure;
+using testing::AssertionResult;
+using testing::AssertionSuccess;
+using testing::DoubleLE;
+using testing::EmptyTestEventListener;
+using testing::FloatLE;
+using testing::GTEST_FLAG(also_run_disabled_tests);
+using testing::GTEST_FLAG(break_on_failure);
+using testing::GTEST_FLAG(catch_exceptions);
+using testing::GTEST_FLAG(color);
+using testing::GTEST_FLAG(death_test_use_fork);
+using testing::GTEST_FLAG(filter);
+using testing::GTEST_FLAG(list_tests);
+using testing::GTEST_FLAG(output);
+using testing::GTEST_FLAG(print_time);
+using testing::GTEST_FLAG(random_seed);
+using testing::GTEST_FLAG(repeat);
+using testing::GTEST_FLAG(show_internal_stack_frames);
+using testing::GTEST_FLAG(shuffle);
+using testing::GTEST_FLAG(stack_trace_depth);
+using testing::GTEST_FLAG(stream_result_to);
+using testing::GTEST_FLAG(throw_on_failure);
+using testing::IsNotSubstring;
+using testing::IsSubstring;
+using testing::Message;
+using testing::ScopedFakeTestPartResultReporter;
+using testing::StaticAssertTypeEq;
+using testing::Test;
+using testing::TestCase;
+using testing::TestEventListeners;
+using testing::TestPartResult;
+using testing::TestPartResultArray;
+using testing::TestProperty;
+using testing::TestResult;
+using testing::UnitTest;
+using testing::kMaxStackTraceDepth;
+using testing::internal::AddReference;
+using testing::internal::AlwaysFalse;
+using testing::internal::AlwaysTrue;
+using testing::internal::AppendUserMessage;
+using testing::internal::ArrayAwareFind;
+using testing::internal::ArrayEq;
+using testing::internal::CodePointToUtf8;
+using testing::internal::CompileAssertTypesEqual;
+using testing::internal::CopyArray;
+using testing::internal::CountIf;
+using testing::internal::EqFailure;
+using testing::internal::FloatingPoint;
+using testing::internal::ForEach;
+using testing::internal::FormatTimeInMillisAsSeconds;
+using testing::internal::GTestFlagSaver;
+using testing::internal::GetCurrentOsStackTraceExceptTop;
+using testing::internal::GetElementOr;
+using testing::internal::GetNextRandomSeed;
+using testing::internal::GetRandomSeedFromFlag;
+using testing::internal::GetTestTypeId;
+using testing::internal::GetTypeId;
+using testing::internal::GetUnitTestImpl;
+using testing::internal::ImplicitlyConvertible;
+using testing::internal::Int32;
+using testing::internal::Int32FromEnvOrDie;
+using testing::internal::IsAProtocolMessage;
+using testing::internal::IsContainer;
+using testing::internal::IsContainerTest;
+using testing::internal::IsNotContainer;
+using testing::internal::NativeArray;
+using testing::internal::ParseInt32Flag;
+using testing::internal::RemoveConst;
+using testing::internal::RemoveReference;
+using testing::internal::ShouldRunTestOnShard;
+using testing::internal::ShouldShard;
+using testing::internal::ShouldUseColor;
+using testing::internal::Shuffle;
+using testing::internal::ShuffleRange;
+using testing::internal::SkipPrefix;
+using testing::internal::StreamableToString;
+using testing::internal::String;
+using testing::internal::TestEventListenersAccessor;
+using testing::internal::TestResultAccessor;
+using testing::internal::UInt32;
+using testing::internal::WideStringToUtf8;
+using testing::internal::kCopy;
+using testing::internal::kMaxRandomSeed;
+using testing::internal::kReference;
+using testing::internal::kTestTypeIdInGoogleTest;
+using testing::internal::scoped_ptr;
+
+#if GTEST_HAS_STREAM_REDIRECTION
+using testing::internal::CaptureStdout;
+using testing::internal::GetCapturedStdout;
+#endif
+
+#if GTEST_IS_THREADSAFE
+using testing::internal::ThreadWithParam;
+#endif
+
+class TestingVector : public std::vector<int> {
+};
+
+::std::ostream& operator<<(::std::ostream& os,
+ const TestingVector& vector) {
+ os << "{ ";
+ for (size_t i = 0; i < vector.size(); i++) {
+ os << vector[i] << " ";
+ }
+ os << "}";
+ return os;
+}
+
+// This line tests that we can define tests in an unnamed namespace.
+namespace {
+
+TEST(GetRandomSeedFromFlagTest, HandlesZero) {
+ const int seed = GetRandomSeedFromFlag(0);
+ EXPECT_LE(1, seed);
+ EXPECT_LE(seed, static_cast<int>(kMaxRandomSeed));
+}
+
+TEST(GetRandomSeedFromFlagTest, PreservesValidSeed) {
+ EXPECT_EQ(1, GetRandomSeedFromFlag(1));
+ EXPECT_EQ(2, GetRandomSeedFromFlag(2));
+ EXPECT_EQ(kMaxRandomSeed - 1, GetRandomSeedFromFlag(kMaxRandomSeed - 1));
+ EXPECT_EQ(static_cast<int>(kMaxRandomSeed),
+ GetRandomSeedFromFlag(kMaxRandomSeed));
+}
+
+TEST(GetRandomSeedFromFlagTest, NormalizesInvalidSeed) {
+ const int seed1 = GetRandomSeedFromFlag(-1);
+ EXPECT_LE(1, seed1);
+ EXPECT_LE(seed1, static_cast<int>(kMaxRandomSeed));
+
+ const int seed2 = GetRandomSeedFromFlag(kMaxRandomSeed + 1);
+ EXPECT_LE(1, seed2);
+ EXPECT_LE(seed2, static_cast<int>(kMaxRandomSeed));
+}
+
+TEST(GetNextRandomSeedTest, WorksForValidInput) {
+ EXPECT_EQ(2, GetNextRandomSeed(1));
+ EXPECT_EQ(3, GetNextRandomSeed(2));
+ EXPECT_EQ(static_cast<int>(kMaxRandomSeed),
+ GetNextRandomSeed(kMaxRandomSeed - 1));
+ EXPECT_EQ(1, GetNextRandomSeed(kMaxRandomSeed));
+
+ // We deliberately don't test GetNextRandomSeed() with invalid
+ // inputs, as that requires death tests, which are expensive. This
+ // is fine as GetNextRandomSeed() is internal and has a
+ // straightforward definition.
+}
+
+static void ClearCurrentTestPartResults() {
+ TestResultAccessor::ClearTestPartResults(
+ GetUnitTestImpl()->current_test_result());
+}
+
+// Tests GetTypeId.
+
+TEST(GetTypeIdTest, ReturnsSameValueForSameType) {
+ EXPECT_EQ(GetTypeId<int>(), GetTypeId<int>());
+ EXPECT_EQ(GetTypeId<Test>(), GetTypeId<Test>());
+}
+
+class SubClassOfTest : public Test {};
+class AnotherSubClassOfTest : public Test {};
+
+TEST(GetTypeIdTest, ReturnsDifferentValuesForDifferentTypes) {
+ EXPECT_NE(GetTypeId<int>(), GetTypeId<const int>());
+ EXPECT_NE(GetTypeId<int>(), GetTypeId<char>());
+ EXPECT_NE(GetTypeId<int>(), GetTestTypeId());
+ EXPECT_NE(GetTypeId<SubClassOfTest>(), GetTestTypeId());
+ EXPECT_NE(GetTypeId<AnotherSubClassOfTest>(), GetTestTypeId());
+ EXPECT_NE(GetTypeId<AnotherSubClassOfTest>(), GetTypeId<SubClassOfTest>());
+}
+
+// Verifies that GetTestTypeId() returns the same value, no matter it
+// is called from inside Google Test or outside of it.
+TEST(GetTestTypeIdTest, ReturnsTheSameValueInsideOrOutsideOfGoogleTest) {
+ EXPECT_EQ(kTestTypeIdInGoogleTest, GetTestTypeId());
+}
+
+// Tests FormatTimeInMillisAsSeconds().
+
+TEST(FormatTimeInMillisAsSecondsTest, FormatsZero) {
+ EXPECT_EQ("0", FormatTimeInMillisAsSeconds(0));
+}
+
+TEST(FormatTimeInMillisAsSecondsTest, FormatsPositiveNumber) {
+ EXPECT_EQ("0.003", FormatTimeInMillisAsSeconds(3));
+ EXPECT_EQ("0.01", FormatTimeInMillisAsSeconds(10));
+ EXPECT_EQ("0.2", FormatTimeInMillisAsSeconds(200));
+ EXPECT_EQ("1.2", FormatTimeInMillisAsSeconds(1200));
+ EXPECT_EQ("3", FormatTimeInMillisAsSeconds(3000));
+}
+
+TEST(FormatTimeInMillisAsSecondsTest, FormatsNegativeNumber) {
+ EXPECT_EQ("-0.003", FormatTimeInMillisAsSeconds(-3));
+ EXPECT_EQ("-0.01", FormatTimeInMillisAsSeconds(-10));
+ EXPECT_EQ("-0.2", FormatTimeInMillisAsSeconds(-200));
+ EXPECT_EQ("-1.2", FormatTimeInMillisAsSeconds(-1200));
+ EXPECT_EQ("-3", FormatTimeInMillisAsSeconds(-3000));
+}
+
+#if GTEST_CAN_COMPARE_NULL
+
+# ifdef __BORLANDC__
+// Silences warnings: "Condition is always true", "Unreachable code"
+# pragma option push -w-ccc -w-rch
+# endif
+
+// Tests that GTEST_IS_NULL_LITERAL_(x) is true when x is a null
+// pointer literal.
+TEST(NullLiteralTest, IsTrueForNullLiterals) {
+ EXPECT_TRUE(GTEST_IS_NULL_LITERAL_(NULL));
+ EXPECT_TRUE(GTEST_IS_NULL_LITERAL_(0));
+ EXPECT_TRUE(GTEST_IS_NULL_LITERAL_(0U));
+ EXPECT_TRUE(GTEST_IS_NULL_LITERAL_(0L));
+
+# ifndef __BORLANDC__
+
+ // Some compilers may fail to detect some null pointer literals;
+ // as long as users of the framework don't use such literals, this
+ // is harmless.
+ EXPECT_TRUE(GTEST_IS_NULL_LITERAL_(1 - 1));
+
+# endif
+}
+
+// Tests that GTEST_IS_NULL_LITERAL_(x) is false when x is not a null
+// pointer literal.
+TEST(NullLiteralTest, IsFalseForNonNullLiterals) {
+ EXPECT_FALSE(GTEST_IS_NULL_LITERAL_(1));
+ EXPECT_FALSE(GTEST_IS_NULL_LITERAL_(0.0));
+ EXPECT_FALSE(GTEST_IS_NULL_LITERAL_('a'));
+ EXPECT_FALSE(GTEST_IS_NULL_LITERAL_(static_cast<void*>(NULL)));
+}
+
+# ifdef __BORLANDC__
+// Restores warnings after previous "#pragma option push" suppressed them.
+# pragma option pop
+# endif
+
+#endif // GTEST_CAN_COMPARE_NULL
+//
+// Tests CodePointToUtf8().
+
+// Tests that the NUL character L'\0' is encoded correctly.
+TEST(CodePointToUtf8Test, CanEncodeNul) {
+ char buffer[32];
+ EXPECT_STREQ("", CodePointToUtf8(L'\0', buffer));
+}
+
+// Tests that ASCII characters are encoded correctly.
+TEST(CodePointToUtf8Test, CanEncodeAscii) {
+ char buffer[32];
+ EXPECT_STREQ("a", CodePointToUtf8(L'a', buffer));
+ EXPECT_STREQ("Z", CodePointToUtf8(L'Z', buffer));
+ EXPECT_STREQ("&", CodePointToUtf8(L'&', buffer));
+ EXPECT_STREQ("\x7F", CodePointToUtf8(L'\x7F', buffer));
+}
+
+// Tests that Unicode code-points that have 8 to 11 bits are encoded
+// as 110xxxxx 10xxxxxx.
+TEST(CodePointToUtf8Test, CanEncode8To11Bits) {
+ char buffer[32];
+ // 000 1101 0011 => 110-00011 10-010011
+ EXPECT_STREQ("\xC3\x93", CodePointToUtf8(L'\xD3', buffer));
+
+ // 101 0111 0110 => 110-10101 10-110110
+ // Some compilers (e.g., GCC on MinGW) cannot handle non-ASCII codepoints
+ // in wide strings and wide chars. In order to accomodate them, we have to
+ // introduce such character constants as integers.
+ EXPECT_STREQ("\xD5\xB6",
+ CodePointToUtf8(static_cast<wchar_t>(0x576), buffer));
+}
+
+// Tests that Unicode code-points that have 12 to 16 bits are encoded
+// as 1110xxxx 10xxxxxx 10xxxxxx.
+TEST(CodePointToUtf8Test, CanEncode12To16Bits) {
+ char buffer[32];
+ // 0000 1000 1101 0011 => 1110-0000 10-100011 10-010011
+ EXPECT_STREQ("\xE0\xA3\x93",
+ CodePointToUtf8(static_cast<wchar_t>(0x8D3), buffer));
+
+ // 1100 0111 0100 1101 => 1110-1100 10-011101 10-001101
+ EXPECT_STREQ("\xEC\x9D\x8D",
+ CodePointToUtf8(static_cast<wchar_t>(0xC74D), buffer));
+}
+
+#if !GTEST_WIDE_STRING_USES_UTF16_
+// Tests in this group require a wchar_t to hold > 16 bits, and thus
+// are skipped on Windows, Cygwin, and Symbian, where a wchar_t is
+// 16-bit wide. This code may not compile on those systems.
+
+// Tests that Unicode code-points that have 17 to 21 bits are encoded
+// as 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx.
+TEST(CodePointToUtf8Test, CanEncode17To21Bits) {
+ char buffer[32];
+ // 0 0001 0000 1000 1101 0011 => 11110-000 10-010000 10-100011 10-010011
+ EXPECT_STREQ("\xF0\x90\xA3\x93", CodePointToUtf8(L'\x108D3', buffer));
+
+ // 0 0001 0000 0100 0000 0000 => 11110-000 10-010000 10-010000 10-000000
+ EXPECT_STREQ("\xF0\x90\x90\x80", CodePointToUtf8(L'\x10400', buffer));
+
+ // 1 0000 1000 0110 0011 0100 => 11110-100 10-001000 10-011000 10-110100
+ EXPECT_STREQ("\xF4\x88\x98\xB4", CodePointToUtf8(L'\x108634', buffer));
+}
+
+// Tests that encoding an invalid code-point generates the expected result.
+TEST(CodePointToUtf8Test, CanEncodeInvalidCodePoint) {
+ char buffer[32];
+ EXPECT_STREQ("(Invalid Unicode 0x1234ABCD)",
+ CodePointToUtf8(L'\x1234ABCD', buffer));
+}
+
+#endif // !GTEST_WIDE_STRING_USES_UTF16_
+
+// Tests WideStringToUtf8().
+
+// Tests that the NUL character L'\0' is encoded correctly.
+TEST(WideStringToUtf8Test, CanEncodeNul) {
+ EXPECT_STREQ("", WideStringToUtf8(L"", 0).c_str());
+ EXPECT_STREQ("", WideStringToUtf8(L"", -1).c_str());
+}
+
+// Tests that ASCII strings are encoded correctly.
+TEST(WideStringToUtf8Test, CanEncodeAscii) {
+ EXPECT_STREQ("a", WideStringToUtf8(L"a", 1).c_str());
+ EXPECT_STREQ("ab", WideStringToUtf8(L"ab", 2).c_str());
+ EXPECT_STREQ("a", WideStringToUtf8(L"a", -1).c_str());
+ EXPECT_STREQ("ab", WideStringToUtf8(L"ab", -1).c_str());
+}
+
+// Tests that Unicode code-points that have 8 to 11 bits are encoded
+// as 110xxxxx 10xxxxxx.
+TEST(WideStringToUtf8Test, CanEncode8To11Bits) {
+ // 000 1101 0011 => 110-00011 10-010011
+ EXPECT_STREQ("\xC3\x93", WideStringToUtf8(L"\xD3", 1).c_str());
+ EXPECT_STREQ("\xC3\x93", WideStringToUtf8(L"\xD3", -1).c_str());
+
+ // 101 0111 0110 => 110-10101 10-110110
+ const wchar_t s[] = { 0x576, '\0' };
+ EXPECT_STREQ("\xD5\xB6", WideStringToUtf8(s, 1).c_str());
+ EXPECT_STREQ("\xD5\xB6", WideStringToUtf8(s, -1).c_str());
+}
+
+// Tests that Unicode code-points that have 12 to 16 bits are encoded
+// as 1110xxxx 10xxxxxx 10xxxxxx.
+TEST(WideStringToUtf8Test, CanEncode12To16Bits) {
+ // 0000 1000 1101 0011 => 1110-0000 10-100011 10-010011
+ const wchar_t s1[] = { 0x8D3, '\0' };
+ EXPECT_STREQ("\xE0\xA3\x93", WideStringToUtf8(s1, 1).c_str());
+ EXPECT_STREQ("\xE0\xA3\x93", WideStringToUtf8(s1, -1).c_str());
+
+ // 1100 0111 0100 1101 => 1110-1100 10-011101 10-001101
+ const wchar_t s2[] = { 0xC74D, '\0' };
+ EXPECT_STREQ("\xEC\x9D\x8D", WideStringToUtf8(s2, 1).c_str());
+ EXPECT_STREQ("\xEC\x9D\x8D", WideStringToUtf8(s2, -1).c_str());
+}
+
+// Tests that the conversion stops when the function encounters \0 character.
+TEST(WideStringToUtf8Test, StopsOnNulCharacter) {
+ EXPECT_STREQ("ABC", WideStringToUtf8(L"ABC\0XYZ", 100).c_str());
+}
+
+// Tests that the conversion stops when the function reaches the limit
+// specified by the 'length' parameter.
+TEST(WideStringToUtf8Test, StopsWhenLengthLimitReached) {
+ EXPECT_STREQ("ABC", WideStringToUtf8(L"ABCDEF", 3).c_str());
+}
+
+#if !GTEST_WIDE_STRING_USES_UTF16_
+// Tests that Unicode code-points that have 17 to 21 bits are encoded
+// as 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx. This code may not compile
+// on the systems using UTF-16 encoding.
+TEST(WideStringToUtf8Test, CanEncode17To21Bits) {
+ // 0 0001 0000 1000 1101 0011 => 11110-000 10-010000 10-100011 10-010011
+ EXPECT_STREQ("\xF0\x90\xA3\x93", WideStringToUtf8(L"\x108D3", 1).c_str());
+ EXPECT_STREQ("\xF0\x90\xA3\x93", WideStringToUtf8(L"\x108D3", -1).c_str());
+
+ // 1 0000 1000 0110 0011 0100 => 11110-100 10-001000 10-011000 10-110100
+ EXPECT_STREQ("\xF4\x88\x98\xB4", WideStringToUtf8(L"\x108634", 1).c_str());
+ EXPECT_STREQ("\xF4\x88\x98\xB4", WideStringToUtf8(L"\x108634", -1).c_str());
+}
+
+// Tests that encoding an invalid code-point generates the expected result.
+TEST(WideStringToUtf8Test, CanEncodeInvalidCodePoint) {
+ EXPECT_STREQ("(Invalid Unicode 0xABCDFF)",
+ WideStringToUtf8(L"\xABCDFF", -1).c_str());
+}
+#else // !GTEST_WIDE_STRING_USES_UTF16_
+// Tests that surrogate pairs are encoded correctly on the systems using
+// UTF-16 encoding in the wide strings.
+TEST(WideStringToUtf8Test, CanEncodeValidUtf16SUrrogatePairs) {
+ const wchar_t s[] = { 0xD801, 0xDC00, '\0' };
+ EXPECT_STREQ("\xF0\x90\x90\x80", WideStringToUtf8(s, -1).c_str());
+}
+
+// Tests that encoding an invalid UTF-16 surrogate pair
+// generates the expected result.
+TEST(WideStringToUtf8Test, CanEncodeInvalidUtf16SurrogatePair) {
+ // Leading surrogate is at the end of the string.
+ const wchar_t s1[] = { 0xD800, '\0' };
+ EXPECT_STREQ("\xED\xA0\x80", WideStringToUtf8(s1, -1).c_str());
+ // Leading surrogate is not followed by the trailing surrogate.
+ const wchar_t s2[] = { 0xD800, 'M', '\0' };
+ EXPECT_STREQ("\xED\xA0\x80M", WideStringToUtf8(s2, -1).c_str());
+ // Trailing surrogate appearas without a leading surrogate.
+ const wchar_t s3[] = { 0xDC00, 'P', 'Q', 'R', '\0' };
+ EXPECT_STREQ("\xED\xB0\x80PQR", WideStringToUtf8(s3, -1).c_str());
+}
+#endif // !GTEST_WIDE_STRING_USES_UTF16_
+
+// Tests that codepoint concatenation works correctly.
+#if !GTEST_WIDE_STRING_USES_UTF16_
+TEST(WideStringToUtf8Test, ConcatenatesCodepointsCorrectly) {
+ const wchar_t s[] = { 0x108634, 0xC74D, '\n', 0x576, 0x8D3, 0x108634, '\0'};
+ EXPECT_STREQ(
+ "\xF4\x88\x98\xB4"
+ "\xEC\x9D\x8D"
+ "\n"
+ "\xD5\xB6"
+ "\xE0\xA3\x93"
+ "\xF4\x88\x98\xB4",
+ WideStringToUtf8(s, -1).c_str());
+}
+#else
+TEST(WideStringToUtf8Test, ConcatenatesCodepointsCorrectly) {
+ const wchar_t s[] = { 0xC74D, '\n', 0x576, 0x8D3, '\0'};
+ EXPECT_STREQ(
+ "\xEC\x9D\x8D" "\n" "\xD5\xB6" "\xE0\xA3\x93",
+ WideStringToUtf8(s, -1).c_str());
+}
+#endif // !GTEST_WIDE_STRING_USES_UTF16_
+
+// Tests the Random class.
+
+TEST(RandomDeathTest, GeneratesCrashesOnInvalidRange) {
+ testing::internal::Random random(42);
+ EXPECT_DEATH_IF_SUPPORTED(
+ random.Generate(0),
+ "Cannot generate a number in the range \\[0, 0\\)");
+ EXPECT_DEATH_IF_SUPPORTED(
+ random.Generate(testing::internal::Random::kMaxRange + 1),
+ "Generation of a number in \\[0, 2147483649\\) was requested, "
+ "but this can only generate numbers in \\[0, 2147483648\\)");
+}
+
+TEST(RandomTest, GeneratesNumbersWithinRange) {
+ const UInt32 kRange = 10000;
+ testing::internal::Random random(12345);
+ for (int i = 0; i < 10; i++) {
+ EXPECT_LT(random.Generate(kRange), kRange) << " for iteration " << i;
+ }
+
+ testing::internal::Random random2(testing::internal::Random::kMaxRange);
+ for (int i = 0; i < 10; i++) {
+ EXPECT_LT(random2.Generate(kRange), kRange) << " for iteration " << i;
+ }
+}
+
+TEST(RandomTest, RepeatsWhenReseeded) {
+ const int kSeed = 123;
+ const int kArraySize = 10;
+ const UInt32 kRange = 10000;
+ UInt32 values[kArraySize];
+
+ testing::internal::Random random(kSeed);
+ for (int i = 0; i < kArraySize; i++) {
+ values[i] = random.Generate(kRange);
+ }
+
+ random.Reseed(kSeed);
+ for (int i = 0; i < kArraySize; i++) {
+ EXPECT_EQ(values[i], random.Generate(kRange)) << " for iteration " << i;
+ }
+}
+
+// Tests STL container utilities.
+
+// Tests CountIf().
+
+static bool IsPositive(int n) { return n > 0; }
+
+TEST(ContainerUtilityTest, CountIf) {
+ std::vector<int> v;
+ EXPECT_EQ(0, CountIf(v, IsPositive)); // Works for an empty container.
+
+ v.push_back(-1);
+ v.push_back(0);
+ EXPECT_EQ(0, CountIf(v, IsPositive)); // Works when no value satisfies.
+
+ v.push_back(2);
+ v.push_back(-10);
+ v.push_back(10);
+ EXPECT_EQ(2, CountIf(v, IsPositive));
+}
+
+// Tests ForEach().
+
+static int g_sum = 0;
+static void Accumulate(int n) { g_sum += n; }
+
+TEST(ContainerUtilityTest, ForEach) {
+ std::vector<int> v;
+ g_sum = 0;
+ ForEach(v, Accumulate);
+ EXPECT_EQ(0, g_sum); // Works for an empty container;
+
+ g_sum = 0;
+ v.push_back(1);
+ ForEach(v, Accumulate);
+ EXPECT_EQ(1, g_sum); // Works for a container with one element.
+
+ g_sum = 0;
+ v.push_back(20);
+ v.push_back(300);
+ ForEach(v, Accumulate);
+ EXPECT_EQ(321, g_sum);
+}
+
+// Tests GetElementOr().
+TEST(ContainerUtilityTest, GetElementOr) {
+ std::vector<char> a;
+ EXPECT_EQ('x', GetElementOr(a, 0, 'x'));
+
+ a.push_back('a');
+ a.push_back('b');
+ EXPECT_EQ('a', GetElementOr(a, 0, 'x'));
+ EXPECT_EQ('b', GetElementOr(a, 1, 'x'));
+ EXPECT_EQ('x', GetElementOr(a, -2, 'x'));
+ EXPECT_EQ('x', GetElementOr(a, 2, 'x'));
+}
+
+TEST(ContainerUtilityDeathTest, ShuffleRange) {
+ std::vector<int> a;
+ a.push_back(0);
+ a.push_back(1);
+ a.push_back(2);
+ testing::internal::Random random(1);
+
+ EXPECT_DEATH_IF_SUPPORTED(
+ ShuffleRange(&random, -1, 1, &a),
+ "Invalid shuffle range start -1: must be in range \\[0, 3\\]");
+ EXPECT_DEATH_IF_SUPPORTED(
+ ShuffleRange(&random, 4, 4, &a),
+ "Invalid shuffle range start 4: must be in range \\[0, 3\\]");
+ EXPECT_DEATH_IF_SUPPORTED(
+ ShuffleRange(&random, 3, 2, &a),
+ "Invalid shuffle range finish 2: must be in range \\[3, 3\\]");
+ EXPECT_DEATH_IF_SUPPORTED(
+ ShuffleRange(&random, 3, 4, &a),
+ "Invalid shuffle range finish 4: must be in range \\[3, 3\\]");
+}
+
+class VectorShuffleTest : public Test {
+ protected:
+ static const int kVectorSize = 20;
+
+ VectorShuffleTest() : random_(1) {
+ for (int i = 0; i < kVectorSize; i++) {
+ vector_.push_back(i);
+ }
+ }
+
+ static bool VectorIsCorrupt(const TestingVector& vector) {
+ if (kVectorSize != static_cast<int>(vector.size())) {
+ return true;
+ }
+
+ bool found_in_vector[kVectorSize] = { false };
+ for (size_t i = 0; i < vector.size(); i++) {
+ const int e = vector[i];
+ if (e < 0 || e >= kVectorSize || found_in_vector[e]) {
+ return true;
+ }
+ found_in_vector[e] = true;
+ }
+
+ // Vector size is correct, elements' range is correct, no
+ // duplicate elements. Therefore no corruption has occurred.
+ return false;
+ }
+
+ static bool VectorIsNotCorrupt(const TestingVector& vector) {
+ return !VectorIsCorrupt(vector);
+ }
+
+ static bool RangeIsShuffled(const TestingVector& vector, int begin, int end) {
+ for (int i = begin; i < end; i++) {
+ if (i != vector[i]) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ static bool RangeIsUnshuffled(
+ const TestingVector& vector, int begin, int end) {
+ return !RangeIsShuffled(vector, begin, end);
+ }
+
+ static bool VectorIsShuffled(const TestingVector& vector) {
+ return RangeIsShuffled(vector, 0, static_cast<int>(vector.size()));
+ }
+
+ static bool VectorIsUnshuffled(const TestingVector& vector) {
+ return !VectorIsShuffled(vector);
+ }
+
+ testing::internal::Random random_;
+ TestingVector vector_;
+}; // class VectorShuffleTest
+
+const int VectorShuffleTest::kVectorSize;
+
+TEST_F(VectorShuffleTest, HandlesEmptyRange) {
+ // Tests an empty range at the beginning...
+ ShuffleRange(&random_, 0, 0, &vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ ASSERT_PRED1(VectorIsUnshuffled, vector_);
+
+ // ...in the middle...
+ ShuffleRange(&random_, kVectorSize/2, kVectorSize/2, &vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ ASSERT_PRED1(VectorIsUnshuffled, vector_);
+
+ // ...at the end...
+ ShuffleRange(&random_, kVectorSize - 1, kVectorSize - 1, &vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ ASSERT_PRED1(VectorIsUnshuffled, vector_);
+
+ // ...and past the end.
+ ShuffleRange(&random_, kVectorSize, kVectorSize, &vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ ASSERT_PRED1(VectorIsUnshuffled, vector_);
+}
+
+TEST_F(VectorShuffleTest, HandlesRangeOfSizeOne) {
+ // Tests a size one range at the beginning...
+ ShuffleRange(&random_, 0, 1, &vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ ASSERT_PRED1(VectorIsUnshuffled, vector_);
+
+ // ...in the middle...
+ ShuffleRange(&random_, kVectorSize/2, kVectorSize/2 + 1, &vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ ASSERT_PRED1(VectorIsUnshuffled, vector_);
+
+ // ...and at the end.
+ ShuffleRange(&random_, kVectorSize - 1, kVectorSize, &vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ ASSERT_PRED1(VectorIsUnshuffled, vector_);
+}
+
+// Because we use our own random number generator and a fixed seed,
+// we can guarantee that the following "random" tests will succeed.
+
+TEST_F(VectorShuffleTest, ShufflesEntireVector) {
+ Shuffle(&random_, &vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ EXPECT_FALSE(VectorIsUnshuffled(vector_)) << vector_;
+
+ // Tests the first and last elements in particular to ensure that
+ // there are no off-by-one problems in our shuffle algorithm.
+ EXPECT_NE(0, vector_[0]);
+ EXPECT_NE(kVectorSize - 1, vector_[kVectorSize - 1]);
+}
+
+TEST_F(VectorShuffleTest, ShufflesStartOfVector) {
+ const int kRangeSize = kVectorSize/2;
+
+ ShuffleRange(&random_, 0, kRangeSize, &vector_);
+
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ EXPECT_PRED3(RangeIsShuffled, vector_, 0, kRangeSize);
+ EXPECT_PRED3(RangeIsUnshuffled, vector_, kRangeSize, kVectorSize);
+}
+
+TEST_F(VectorShuffleTest, ShufflesEndOfVector) {
+ const int kRangeSize = kVectorSize / 2;
+ ShuffleRange(&random_, kRangeSize, kVectorSize, &vector_);
+
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ EXPECT_PRED3(RangeIsUnshuffled, vector_, 0, kRangeSize);
+ EXPECT_PRED3(RangeIsShuffled, vector_, kRangeSize, kVectorSize);
+}
+
+TEST_F(VectorShuffleTest, ShufflesMiddleOfVector) {
+ int kRangeSize = kVectorSize/3;
+ ShuffleRange(&random_, kRangeSize, 2*kRangeSize, &vector_);
+
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ EXPECT_PRED3(RangeIsUnshuffled, vector_, 0, kRangeSize);
+ EXPECT_PRED3(RangeIsShuffled, vector_, kRangeSize, 2*kRangeSize);
+ EXPECT_PRED3(RangeIsUnshuffled, vector_, 2*kRangeSize, kVectorSize);
+}
+
+TEST_F(VectorShuffleTest, ShufflesRepeatably) {
+ TestingVector vector2;
+ for (int i = 0; i < kVectorSize; i++) {
+ vector2.push_back(i);
+ }
+
+ random_.Reseed(1234);
+ Shuffle(&random_, &vector_);
+ random_.Reseed(1234);
+ Shuffle(&random_, &vector2);
+
+ ASSERT_PRED1(VectorIsNotCorrupt, vector_);
+ ASSERT_PRED1(VectorIsNotCorrupt, vector2);
+
+ for (int i = 0; i < kVectorSize; i++) {
+ EXPECT_EQ(vector_[i], vector2[i]) << " where i is " << i;
+ }
+}
+
+// Tests the size of the AssertHelper class.
+
+TEST(AssertHelperTest, AssertHelperIsSmall) {
+ // To avoid breaking clients that use lots of assertions in one
+ // function, we cannot grow the size of AssertHelper.
+ EXPECT_LE(sizeof(testing::internal::AssertHelper), sizeof(void*));
+}
+
+// Tests the String class.
+
+// Tests String's constructors.
+TEST(StringTest, Constructors) {
+ // Default ctor.
+ String s1;
+ // We aren't using EXPECT_EQ(NULL, s1.c_str()) because comparing
+ // pointers with NULL isn't supported on all platforms.
+ EXPECT_EQ(0U, s1.length());
+ EXPECT_TRUE(NULL == s1.c_str());
+
+ // Implicitly constructs from a C-string.
+ String s2 = "Hi";
+ EXPECT_EQ(2U, s2.length());
+ EXPECT_STREQ("Hi", s2.c_str());
+
+ // Constructs from a C-string and a length.
+ String s3("hello", 3);
+ EXPECT_EQ(3U, s3.length());
+ EXPECT_STREQ("hel", s3.c_str());
+
+ // The empty String should be created when String is constructed with
+ // a NULL pointer and length 0.
+ EXPECT_EQ(0U, String(NULL, 0).length());
+ EXPECT_FALSE(String(NULL, 0).c_str() == NULL);
+
+ // Constructs a String that contains '\0'.
+ String s4("a\0bcd", 4);
+ EXPECT_EQ(4U, s4.length());
+ EXPECT_EQ('a', s4.c_str()[0]);
+ EXPECT_EQ('\0', s4.c_str()[1]);
+ EXPECT_EQ('b', s4.c_str()[2]);
+ EXPECT_EQ('c', s4.c_str()[3]);
+
+ // Copy ctor where the source is NULL.
+ const String null_str;
+ String s5 = null_str;
+ EXPECT_TRUE(s5.c_str() == NULL);
+
+ // Copy ctor where the source isn't NULL.
+ String s6 = s3;
+ EXPECT_EQ(3U, s6.length());
+ EXPECT_STREQ("hel", s6.c_str());
+
+ // Copy ctor where the source contains '\0'.
+ String s7 = s4;
+ EXPECT_EQ(4U, s7.length());
+ EXPECT_EQ('a', s7.c_str()[0]);
+ EXPECT_EQ('\0', s7.c_str()[1]);
+ EXPECT_EQ('b', s7.c_str()[2]);
+ EXPECT_EQ('c', s7.c_str()[3]);
+}
+
+TEST(StringTest, ConvertsFromStdString) {
+ // An empty std::string.
+ const std::string src1("");
+ const String dest1 = src1;
+ EXPECT_EQ(0U, dest1.length());
+ EXPECT_STREQ("", dest1.c_str());
+
+ // A normal std::string.
+ const std::string src2("Hi");
+ const String dest2 = src2;
+ EXPECT_EQ(2U, dest2.length());
+ EXPECT_STREQ("Hi", dest2.c_str());
+
+ // An std::string with an embedded NUL character.
+ const char src3[] = "a\0b";
+ const String dest3 = std::string(src3, sizeof(src3));
+ EXPECT_EQ(sizeof(src3), dest3.length());
+ EXPECT_EQ('a', dest3.c_str()[0]);
+ EXPECT_EQ('\0', dest3.c_str()[1]);
+ EXPECT_EQ('b', dest3.c_str()[2]);
+}
+
+TEST(StringTest, ConvertsToStdString) {
+ // An empty String.
+ const String src1("");
+ const std::string dest1 = src1;
+ EXPECT_EQ("", dest1);
+
+ // A normal String.
+ const String src2("Hi");
+ const std::string dest2 = src2;
+ EXPECT_EQ("Hi", dest2);
+
+ // A String containing a '\0'.
+ const String src3("x\0y", 3);
+ const std::string dest3 = src3;
+ EXPECT_EQ(std::string("x\0y", 3), dest3);
+}
+
+#if GTEST_HAS_GLOBAL_STRING
+
+TEST(StringTest, ConvertsFromGlobalString) {
+ // An empty ::string.
+ const ::string src1("");
+ const String dest1 = src1;
+ EXPECT_EQ(0U, dest1.length());
+ EXPECT_STREQ("", dest1.c_str());
+
+ // A normal ::string.
+ const ::string src2("Hi");
+ const String dest2 = src2;
+ EXPECT_EQ(2U, dest2.length());
+ EXPECT_STREQ("Hi", dest2.c_str());
+
+ // An ::string with an embedded NUL character.
+ const char src3[] = "x\0y";
+ const String dest3 = ::string(src3, sizeof(src3));
+ EXPECT_EQ(sizeof(src3), dest3.length());
+ EXPECT_EQ('x', dest3.c_str()[0]);
+ EXPECT_EQ('\0', dest3.c_str()[1]);
+ EXPECT_EQ('y', dest3.c_str()[2]);
+}
+
+TEST(StringTest, ConvertsToGlobalString) {
+ // An empty String.
+ const String src1("");
+ const ::string dest1 = src1;
+ EXPECT_EQ("", dest1);
+
+ // A normal String.
+ const String src2("Hi");
+ const ::string dest2 = src2;
+ EXPECT_EQ("Hi", dest2);
+
+ const String src3("x\0y", 3);
+ const ::string dest3 = src3;
+ EXPECT_EQ(::string("x\0y", 3), dest3);
+}
+
+#endif // GTEST_HAS_GLOBAL_STRING
+
+// Tests String::ShowCStringQuoted().
+TEST(StringTest, ShowCStringQuoted) {
+ EXPECT_STREQ("(null)",
+ String::ShowCStringQuoted(NULL).c_str());
+ EXPECT_STREQ("\"\"",
+ String::ShowCStringQuoted("").c_str());
+ EXPECT_STREQ("\"foo\"",
+ String::ShowCStringQuoted("foo").c_str());
+}
+
+// Tests String::empty().
+TEST(StringTest, Empty) {
+ EXPECT_TRUE(String("").empty());
+ EXPECT_FALSE(String().empty());
+ EXPECT_FALSE(String(NULL).empty());
+ EXPECT_FALSE(String("a").empty());
+ EXPECT_FALSE(String("\0", 1).empty());
+}
+
+// Tests String::Compare().
+TEST(StringTest, Compare) {
+ // NULL vs NULL.
+ EXPECT_EQ(0, String().Compare(String()));
+
+ // NULL vs non-NULL.
+ EXPECT_EQ(-1, String().Compare(String("")));
+
+ // Non-NULL vs NULL.
+ EXPECT_EQ(1, String("").Compare(String()));
+
+ // The following covers non-NULL vs non-NULL.
+
+ // "" vs "".
+ EXPECT_EQ(0, String("").Compare(String("")));
+
+ // "" vs non-"".
+ EXPECT_EQ(-1, String("").Compare(String("\0", 1)));
+ EXPECT_EQ(-1, String("").Compare(" "));
+
+ // Non-"" vs "".
+ EXPECT_EQ(1, String("a").Compare(String("")));
+
+ // The following covers non-"" vs non-"".
+
+ // Same length and equal.
+ EXPECT_EQ(0, String("a").Compare(String("a")));
+
+ // Same length and different.
+ EXPECT_EQ(-1, String("a\0b", 3).Compare(String("a\0c", 3)));
+ EXPECT_EQ(1, String("b").Compare(String("a")));
+
+ // Different lengths.
+ EXPECT_EQ(-1, String("a").Compare(String("ab")));
+ EXPECT_EQ(-1, String("a").Compare(String("a\0", 2)));
+ EXPECT_EQ(1, String("abc").Compare(String("aacd")));
+}
+
+// Tests String::operator==().
+TEST(StringTest, Equals) {
+ const String null(NULL);
+ EXPECT_TRUE(null == NULL); // NOLINT
+ EXPECT_FALSE(null == ""); // NOLINT
+ EXPECT_FALSE(null == "bar"); // NOLINT
+
+ const String empty("");
+ EXPECT_FALSE(empty == NULL); // NOLINT
+ EXPECT_TRUE(empty == ""); // NOLINT
+ EXPECT_FALSE(empty == "bar"); // NOLINT
+
+ const String foo("foo");
+ EXPECT_FALSE(foo == NULL); // NOLINT
+ EXPECT_FALSE(foo == ""); // NOLINT
+ EXPECT_FALSE(foo == "bar"); // NOLINT
+ EXPECT_TRUE(foo == "foo"); // NOLINT
+
+ const String bar("x\0y", 3);
+ EXPECT_FALSE(bar == "x");
+}
+
+// Tests String::operator!=().
+TEST(StringTest, NotEquals) {
+ const String null(NULL);
+ EXPECT_FALSE(null != NULL); // NOLINT
+ EXPECT_TRUE(null != ""); // NOLINT
+ EXPECT_TRUE(null != "bar"); // NOLINT
+
+ const String empty("");
+ EXPECT_TRUE(empty != NULL); // NOLINT
+ EXPECT_FALSE(empty != ""); // NOLINT
+ EXPECT_TRUE(empty != "bar"); // NOLINT
+
+ const String foo("foo");
+ EXPECT_TRUE(foo != NULL); // NOLINT
+ EXPECT_TRUE(foo != ""); // NOLINT
+ EXPECT_TRUE(foo != "bar"); // NOLINT
+ EXPECT_FALSE(foo != "foo"); // NOLINT
+
+ const String bar("x\0y", 3);
+ EXPECT_TRUE(bar != "x");
+}
+
+// Tests String::length().
+TEST(StringTest, Length) {
+ EXPECT_EQ(0U, String().length());
+ EXPECT_EQ(0U, String("").length());
+ EXPECT_EQ(2U, String("ab").length());
+ EXPECT_EQ(3U, String("a\0b", 3).length());
+}
+
+// Tests String::EndsWith().
+TEST(StringTest, EndsWith) {
+ EXPECT_TRUE(String("foobar").EndsWith("bar"));
+ EXPECT_TRUE(String("foobar").EndsWith(""));
+ EXPECT_TRUE(String("").EndsWith(""));
+
+ EXPECT_FALSE(String("foobar").EndsWith("foo"));
+ EXPECT_FALSE(String("").EndsWith("foo"));
+}
+
+// Tests String::EndsWithCaseInsensitive().
+TEST(StringTest, EndsWithCaseInsensitive) {
+ EXPECT_TRUE(String("foobar").EndsWithCaseInsensitive("BAR"));
+ EXPECT_TRUE(String("foobaR").EndsWithCaseInsensitive("bar"));
+ EXPECT_TRUE(String("foobar").EndsWithCaseInsensitive(""));
+ EXPECT_TRUE(String("").EndsWithCaseInsensitive(""));
+
+ EXPECT_FALSE(String("Foobar").EndsWithCaseInsensitive("foo"));
+ EXPECT_FALSE(String("foobar").EndsWithCaseInsensitive("Foo"));
+ EXPECT_FALSE(String("").EndsWithCaseInsensitive("foo"));
+}
+
+// C++Builder's preprocessor is buggy; it fails to expand macros that
+// appear in macro parameters after wide char literals. Provide an alias
+// for NULL as a workaround.
+static const wchar_t* const kNull = NULL;
+
+// Tests String::CaseInsensitiveWideCStringEquals
+TEST(StringTest, CaseInsensitiveWideCStringEquals) {
+ EXPECT_TRUE(String::CaseInsensitiveWideCStringEquals(NULL, NULL));
+ EXPECT_FALSE(String::CaseInsensitiveWideCStringEquals(kNull, L""));
+ EXPECT_FALSE(String::CaseInsensitiveWideCStringEquals(L"", kNull));
+ EXPECT_FALSE(String::CaseInsensitiveWideCStringEquals(kNull, L"foobar"));
+ EXPECT_FALSE(String::CaseInsensitiveWideCStringEquals(L"foobar", kNull));
+ EXPECT_TRUE(String::CaseInsensitiveWideCStringEquals(L"foobar", L"foobar"));
+ EXPECT_TRUE(String::CaseInsensitiveWideCStringEquals(L"foobar", L"FOOBAR"));
+ EXPECT_TRUE(String::CaseInsensitiveWideCStringEquals(L"FOOBAR", L"foobar"));
+}
+
+// Tests that NULL can be assigned to a String.
+TEST(StringTest, CanBeAssignedNULL) {
+ const String src(NULL);
+ String dest;
+
+ dest = src;
+ EXPECT_STREQ(NULL, dest.c_str());
+}
+
+// Tests that the empty string "" can be assigned to a String.
+TEST(StringTest, CanBeAssignedEmpty) {
+ const String src("");
+ String dest;
+
+ dest = src;
+ EXPECT_STREQ("", dest.c_str());
+}
+
+// Tests that a non-empty string can be assigned to a String.
+TEST(StringTest, CanBeAssignedNonEmpty) {
+ const String src("hello");
+ String dest;
+ dest = src;
+ EXPECT_EQ(5U, dest.length());
+ EXPECT_STREQ("hello", dest.c_str());
+
+ const String src2("x\0y", 3);
+ String dest2;
+ dest2 = src2;
+ EXPECT_EQ(3U, dest2.length());
+ EXPECT_EQ('x', dest2.c_str()[0]);
+ EXPECT_EQ('\0', dest2.c_str()[1]);
+ EXPECT_EQ('y', dest2.c_str()[2]);
+}
+
+// Tests that a String can be assigned to itself.
+TEST(StringTest, CanBeAssignedSelf) {
+ String dest("hello");
+
+ // Use explicit function call notation here to suppress self-assign warning.
+ dest.operator=(dest);
+ EXPECT_STREQ("hello", dest.c_str());
+}
+
+// Sun Studio < 12 incorrectly rejects this code due to an overloading
+// ambiguity.
+#if !(defined(__SUNPRO_CC) && __SUNPRO_CC < 0x590)
+// Tests streaming a String.
+TEST(StringTest, Streams) {
+ EXPECT_EQ(StreamableToString(String()), "(null)");
+ EXPECT_EQ(StreamableToString(String("")), "");
+ EXPECT_EQ(StreamableToString(String("a\0b", 3)), "a\\0b");
+}
+#endif
+
+// Tests that String::Format() works.
+TEST(StringTest, FormatWorks) {
+ // Normal case: the format spec is valid, the arguments match the
+ // spec, and the result is < 4095 characters.
+ EXPECT_STREQ("Hello, 42", String::Format("%s, %d", "Hello", 42).c_str());
+
+ // Edge case: the result is 4095 characters.
+ char buffer[4096];
+ const size_t kSize = sizeof(buffer);
+ memset(buffer, 'a', kSize - 1);
+ buffer[kSize - 1] = '\0';
+ EXPECT_STREQ(buffer, String::Format("%s", buffer).c_str());
+
+ // The result needs to be 4096 characters, exceeding Format()'s limit.
+ EXPECT_STREQ("<formatting error or buffer exceeded>",
+ String::Format("x%s", buffer).c_str());
+
+#if GTEST_OS_LINUX
+ // On Linux, invalid format spec should lead to an error message.
+ // In other environment (e.g. MSVC on Windows), String::Format() may
+ // simply ignore a bad format spec, so this assertion is run on
+ // Linux only.
+ EXPECT_STREQ("<formatting error or buffer exceeded>",
+ String::Format("%").c_str());
+#endif
+}
+
+#if GTEST_OS_WINDOWS
+
+// Tests String::ShowWideCString().
+TEST(StringTest, ShowWideCString) {
+ EXPECT_STREQ("(null)",
+ String::ShowWideCString(NULL).c_str());
+ EXPECT_STREQ("", String::ShowWideCString(L"").c_str());
+ EXPECT_STREQ("foo", String::ShowWideCString(L"foo").c_str());
+}
+
+// Tests String::ShowWideCStringQuoted().
+TEST(StringTest, ShowWideCStringQuoted) {
+ EXPECT_STREQ("(null)",
+ String::ShowWideCStringQuoted(NULL).c_str());
+ EXPECT_STREQ("L\"\"",
+ String::ShowWideCStringQuoted(L"").c_str());
+ EXPECT_STREQ("L\"foo\"",
+ String::ShowWideCStringQuoted(L"foo").c_str());
+}
+
+# if GTEST_OS_WINDOWS_MOBILE
+TEST(StringTest, AnsiAndUtf16Null) {
+ EXPECT_EQ(NULL, String::AnsiToUtf16(NULL));
+ EXPECT_EQ(NULL, String::Utf16ToAnsi(NULL));
+}
+
+TEST(StringTest, AnsiAndUtf16ConvertBasic) {
+ const char* ansi = String::Utf16ToAnsi(L"str");
+ EXPECT_STREQ("str", ansi);
+ delete [] ansi;
+ const WCHAR* utf16 = String::AnsiToUtf16("str");
+ EXPECT_EQ(0, wcsncmp(L"str", utf16, 3));
+ delete [] utf16;
+}
+
+TEST(StringTest, AnsiAndUtf16ConvertPathChars) {
+ const char* ansi = String::Utf16ToAnsi(L".:\\ \"*?");
+ EXPECT_STREQ(".:\\ \"*?", ansi);
+ delete [] ansi;
+ const WCHAR* utf16 = String::AnsiToUtf16(".:\\ \"*?");
+ EXPECT_EQ(0, wcsncmp(L".:\\ \"*?", utf16, 3));
+ delete [] utf16;
+}
+# endif // GTEST_OS_WINDOWS_MOBILE
+
+#endif // GTEST_OS_WINDOWS
+
+// Tests TestProperty construction.
+TEST(TestPropertyTest, StringValue) {
+ TestProperty property("key", "1");
+ EXPECT_STREQ("key", property.key());
+ EXPECT_STREQ("1", property.value());
+}
+
+// Tests TestProperty replacing a value.
+TEST(TestPropertyTest, ReplaceStringValue) {
+ TestProperty property("key", "1");
+ EXPECT_STREQ("1", property.value());
+ property.SetValue("2");
+ EXPECT_STREQ("2", property.value());
+}
+
+// AddFatalFailure() and AddNonfatalFailure() must be stand-alone
+// functions (i.e. their definitions cannot be inlined at the call
+// sites), or C++Builder won't compile the code.
+static void AddFatalFailure() {
+ FAIL() << "Expected fatal failure.";
+}
+
+static void AddNonfatalFailure() {
+ ADD_FAILURE() << "Expected non-fatal failure.";
+}
+
+class ScopedFakeTestPartResultReporterTest : public Test {
+ public: // Must be public and not protected due to a bug in g++ 3.4.2.
+ enum FailureMode {
+ FATAL_FAILURE,
+ NONFATAL_FAILURE
+ };
+ static void AddFailure(FailureMode failure) {
+ if (failure == FATAL_FAILURE) {
+ AddFatalFailure();
+ } else {
+ AddNonfatalFailure();
+ }
+ }
+};
+
+// Tests that ScopedFakeTestPartResultReporter intercepts test
+// failures.
+TEST_F(ScopedFakeTestPartResultReporterTest, InterceptsTestFailures) {
+ TestPartResultArray results;
+ {
+ ScopedFakeTestPartResultReporter reporter(
+ ScopedFakeTestPartResultReporter::INTERCEPT_ONLY_CURRENT_THREAD,
+ &results);
+ AddFailure(NONFATAL_FAILURE);
+ AddFailure(FATAL_FAILURE);
+ }
+
+ EXPECT_EQ(2, results.size());
+ EXPECT_TRUE(results.GetTestPartResult(0).nonfatally_failed());
+ EXPECT_TRUE(results.GetTestPartResult(1).fatally_failed());
+}
+
+TEST_F(ScopedFakeTestPartResultReporterTest, DeprecatedConstructor) {
+ TestPartResultArray results;
+ {
+ // Tests, that the deprecated constructor still works.
+ ScopedFakeTestPartResultReporter reporter(&results);
+ AddFailure(NONFATAL_FAILURE);
+ }
+ EXPECT_EQ(1, results.size());
+}
+
+#if GTEST_IS_THREADSAFE
+
+class ScopedFakeTestPartResultReporterWithThreadsTest
+ : public ScopedFakeTestPartResultReporterTest {
+ protected:
+ static void AddFailureInOtherThread(FailureMode failure) {
+ ThreadWithParam<FailureMode> thread(&AddFailure, failure, NULL);
+ thread.Join();
+ }
+};
+
+TEST_F(ScopedFakeTestPartResultReporterWithThreadsTest,
+ InterceptsTestFailuresInAllThreads) {
+ TestPartResultArray results;
+ {
+ ScopedFakeTestPartResultReporter reporter(
+ ScopedFakeTestPartResultReporter::INTERCEPT_ALL_THREADS, &results);
+ AddFailure(NONFATAL_FAILURE);
+ AddFailure(FATAL_FAILURE);
+ AddFailureInOtherThread(NONFATAL_FAILURE);
+ AddFailureInOtherThread(FATAL_FAILURE);
+ }
+
+ EXPECT_EQ(4, results.size());
+ EXPECT_TRUE(results.GetTestPartResult(0).nonfatally_failed());
+ EXPECT_TRUE(results.GetTestPartResult(1).fatally_failed());
+ EXPECT_TRUE(results.GetTestPartResult(2).nonfatally_failed());
+ EXPECT_TRUE(results.GetTestPartResult(3).fatally_failed());
+}
+
+#endif // GTEST_IS_THREADSAFE
+
+// Tests EXPECT_FATAL_FAILURE{,ON_ALL_THREADS}. Makes sure that they
+// work even if the failure is generated in a called function rather than
+// the current context.
+
+typedef ScopedFakeTestPartResultReporterTest ExpectFatalFailureTest;
+
+TEST_F(ExpectFatalFailureTest, CatchesFatalFaliure) {
+ EXPECT_FATAL_FAILURE(AddFatalFailure(), "Expected fatal failure.");
+}
+
+#if GTEST_HAS_GLOBAL_STRING
+TEST_F(ExpectFatalFailureTest, AcceptsStringObject) {
+ EXPECT_FATAL_FAILURE(AddFatalFailure(), ::string("Expected fatal failure."));
+}
+#endif
+
+TEST_F(ExpectFatalFailureTest, AcceptsStdStringObject) {
+ EXPECT_FATAL_FAILURE(AddFatalFailure(),
+ ::std::string("Expected fatal failure."));
+}
+
+TEST_F(ExpectFatalFailureTest, CatchesFatalFailureOnAllThreads) {
+ // We have another test below to verify that the macro catches fatal
+ // failures generated on another thread.
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS(AddFatalFailure(),
+ "Expected fatal failure.");
+}
+
+#ifdef __BORLANDC__
+// Silences warnings: "Condition is always true"
+# pragma option push -w-ccc
+#endif
+
+// Tests that EXPECT_FATAL_FAILURE() can be used in a non-void
+// function even when the statement in it contains ASSERT_*.
+
+int NonVoidFunction() {
+ EXPECT_FATAL_FAILURE(ASSERT_TRUE(false), "");
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS(FAIL(), "");
+ return 0;
+}
+
+TEST_F(ExpectFatalFailureTest, CanBeUsedInNonVoidFunction) {
+ NonVoidFunction();
+}
+
+// Tests that EXPECT_FATAL_FAILURE(statement, ...) doesn't abort the
+// current function even though 'statement' generates a fatal failure.
+
+void DoesNotAbortHelper(bool* aborted) {
+ EXPECT_FATAL_FAILURE(ASSERT_TRUE(false), "");
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS(FAIL(), "");
+
+ *aborted = false;
+}
+
+#ifdef __BORLANDC__
+// Restores warnings after previous "#pragma option push" suppressed them.
+# pragma option pop
+#endif
+
+TEST_F(ExpectFatalFailureTest, DoesNotAbort) {
+ bool aborted = true;
+ DoesNotAbortHelper(&aborted);
+ EXPECT_FALSE(aborted);
+}
+
+// Tests that the EXPECT_FATAL_FAILURE{,_ON_ALL_THREADS} accepts a
+// statement that contains a macro which expands to code containing an
+// unprotected comma.
+
+static int global_var = 0;
+#define GTEST_USE_UNPROTECTED_COMMA_ global_var++, global_var++
+
+TEST_F(ExpectFatalFailureTest, AcceptsMacroThatExpandsToUnprotectedComma) {
+#ifndef __BORLANDC__
+ // ICE's in C++Builder.
+ EXPECT_FATAL_FAILURE({
+ GTEST_USE_UNPROTECTED_COMMA_;
+ AddFatalFailure();
+ }, "");
+#endif
+
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS({
+ GTEST_USE_UNPROTECTED_COMMA_;
+ AddFatalFailure();
+ }, "");
+}
+
+// Tests EXPECT_NONFATAL_FAILURE{,ON_ALL_THREADS}.
+
+typedef ScopedFakeTestPartResultReporterTest ExpectNonfatalFailureTest;
+
+TEST_F(ExpectNonfatalFailureTest, CatchesNonfatalFailure) {
+ EXPECT_NONFATAL_FAILURE(AddNonfatalFailure(),
+ "Expected non-fatal failure.");
+}
+
+#if GTEST_HAS_GLOBAL_STRING
+TEST_F(ExpectNonfatalFailureTest, AcceptsStringObject) {
+ EXPECT_NONFATAL_FAILURE(AddNonfatalFailure(),
+ ::string("Expected non-fatal failure."));
+}
+#endif
+
+TEST_F(ExpectNonfatalFailureTest, AcceptsStdStringObject) {
+ EXPECT_NONFATAL_FAILURE(AddNonfatalFailure(),
+ ::std::string("Expected non-fatal failure."));
+}
+
+TEST_F(ExpectNonfatalFailureTest, CatchesNonfatalFailureOnAllThreads) {
+ // We have another test below to verify that the macro catches
+ // non-fatal failures generated on another thread.
+ EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(AddNonfatalFailure(),
+ "Expected non-fatal failure.");
+}
+
+// Tests that the EXPECT_NONFATAL_FAILURE{,_ON_ALL_THREADS} accepts a
+// statement that contains a macro which expands to code containing an
+// unprotected comma.
+TEST_F(ExpectNonfatalFailureTest, AcceptsMacroThatExpandsToUnprotectedComma) {
+ EXPECT_NONFATAL_FAILURE({
+ GTEST_USE_UNPROTECTED_COMMA_;
+ AddNonfatalFailure();
+ }, "");
+
+ EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS({
+ GTEST_USE_UNPROTECTED_COMMA_;
+ AddNonfatalFailure();
+ }, "");
+}
+
+#if GTEST_IS_THREADSAFE
+
+typedef ScopedFakeTestPartResultReporterWithThreadsTest
+ ExpectFailureWithThreadsTest;
+
+TEST_F(ExpectFailureWithThreadsTest, ExpectFatalFailureOnAllThreads) {
+ EXPECT_FATAL_FAILURE_ON_ALL_THREADS(AddFailureInOtherThread(FATAL_FAILURE),
+ "Expected fatal failure.");
+}
+
+TEST_F(ExpectFailureWithThreadsTest, ExpectNonFatalFailureOnAllThreads) {
+ EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(
+ AddFailureInOtherThread(NONFATAL_FAILURE), "Expected non-fatal failure.");
+}
+
+#endif // GTEST_IS_THREADSAFE
+
+// Tests the TestProperty class.
+
+TEST(TestPropertyTest, ConstructorWorks) {
+ const TestProperty property("key", "value");
+ EXPECT_STREQ("key", property.key());
+ EXPECT_STREQ("value", property.value());
+}
+
+TEST(TestPropertyTest, SetValue) {
+ TestProperty property("key", "value_1");
+ EXPECT_STREQ("key", property.key());
+ property.SetValue("value_2");
+ EXPECT_STREQ("key", property.key());
+ EXPECT_STREQ("value_2", property.value());
+}
+
+// Tests the TestResult class
+
+// The test fixture for testing TestResult.
+class TestResultTest : public Test {
+ protected:
+ typedef std::vector<TestPartResult> TPRVector;
+
+ // We make use of 2 TestPartResult objects,
+ TestPartResult * pr1, * pr2;
+
+ // ... and 3 TestResult objects.
+ TestResult * r0, * r1, * r2;
+
+ virtual void SetUp() {
+ // pr1 is for success.
+ pr1 = new TestPartResult(TestPartResult::kSuccess,
+ "foo/bar.cc",
+ 10,
+ "Success!");
+
+ // pr2 is for fatal failure.
+ pr2 = new TestPartResult(TestPartResult::kFatalFailure,
+ "foo/bar.cc",
+ -1, // This line number means "unknown"
+ "Failure!");
+
+ // Creates the TestResult objects.
+ r0 = new TestResult();
+ r1 = new TestResult();
+ r2 = new TestResult();
+
+ // In order to test TestResult, we need to modify its internal
+ // state, in particular the TestPartResult vector it holds.
+ // test_part_results() returns a const reference to this vector.
+ // We cast it to a non-const object s.t. it can be modified (yes,
+ // this is a hack).
+ TPRVector* results1 = const_cast<TPRVector*>(
+ &TestResultAccessor::test_part_results(*r1));
+ TPRVector* results2 = const_cast<TPRVector*>(
+ &TestResultAccessor::test_part_results(*r2));
+
+ // r0 is an empty TestResult.
+
+ // r1 contains a single SUCCESS TestPartResult.
+ results1->push_back(*pr1);
+
+ // r2 contains a SUCCESS, and a FAILURE.
+ results2->push_back(*pr1);
+ results2->push_back(*pr2);
+ }
+
+ virtual void TearDown() {
+ delete pr1;
+ delete pr2;
+
+ delete r0;
+ delete r1;
+ delete r2;
+ }
+
+ // Helper that compares two two TestPartResults.
+ static void CompareTestPartResult(const TestPartResult& expected,
+ const TestPartResult& actual) {
+ EXPECT_EQ(expected.type(), actual.type());
+ EXPECT_STREQ(expected.file_name(), actual.file_name());
+ EXPECT_EQ(expected.line_number(), actual.line_number());
+ EXPECT_STREQ(expected.summary(), actual.summary());
+ EXPECT_STREQ(expected.message(), actual.message());
+ EXPECT_EQ(expected.passed(), actual.passed());
+ EXPECT_EQ(expected.failed(), actual.failed());
+ EXPECT_EQ(expected.nonfatally_failed(), actual.nonfatally_failed());
+ EXPECT_EQ(expected.fatally_failed(), actual.fatally_failed());
+ }
+};
+
+// Tests TestResult::total_part_count().
+TEST_F(TestResultTest, total_part_count) {
+ ASSERT_EQ(0, r0->total_part_count());
+ ASSERT_EQ(1, r1->total_part_count());
+ ASSERT_EQ(2, r2->total_part_count());
+}
+
+// Tests TestResult::Passed().
+TEST_F(TestResultTest, Passed) {
+ ASSERT_TRUE(r0->Passed());
+ ASSERT_TRUE(r1->Passed());
+ ASSERT_FALSE(r2->Passed());
+}
+
+// Tests TestResult::Failed().
+TEST_F(TestResultTest, Failed) {
+ ASSERT_FALSE(r0->Failed());
+ ASSERT_FALSE(r1->Failed());
+ ASSERT_TRUE(r2->Failed());
+}
+
+// Tests TestResult::GetTestPartResult().
+
+typedef TestResultTest TestResultDeathTest;
+
+TEST_F(TestResultDeathTest, GetTestPartResult) {
+ CompareTestPartResult(*pr1, r2->GetTestPartResult(0));
+ CompareTestPartResult(*pr2, r2->GetTestPartResult(1));
+ EXPECT_DEATH_IF_SUPPORTED(r2->GetTestPartResult(2), "");
+ EXPECT_DEATH_IF_SUPPORTED(r2->GetTestPartResult(-1), "");
+}
+
+// Tests TestResult has no properties when none are added.
+TEST(TestResultPropertyTest, NoPropertiesFoundWhenNoneAreAdded) {
+ TestResult test_result;
+ ASSERT_EQ(0, test_result.test_property_count());
+}
+
+// Tests TestResult has the expected property when added.
+TEST(TestResultPropertyTest, OnePropertyFoundWhenAdded) {
+ TestResult test_result;
+ TestProperty property("key_1", "1");
+ TestResultAccessor::RecordProperty(&test_result, property);
+ ASSERT_EQ(1, test_result.test_property_count());
+ const TestProperty& actual_property = test_result.GetTestProperty(0);
+ EXPECT_STREQ("key_1", actual_property.key());
+ EXPECT_STREQ("1", actual_property.value());
+}
+
+// Tests TestResult has multiple properties when added.
+TEST(TestResultPropertyTest, MultiplePropertiesFoundWhenAdded) {
+ TestResult test_result;
+ TestProperty property_1("key_1", "1");
+ TestProperty property_2("key_2", "2");
+ TestResultAccessor::RecordProperty(&test_result, property_1);
+ TestResultAccessor::RecordProperty(&test_result, property_2);
+ ASSERT_EQ(2, test_result.test_property_count());
+ const TestProperty& actual_property_1 = test_result.GetTestProperty(0);
+ EXPECT_STREQ("key_1", actual_property_1.key());
+ EXPECT_STREQ("1", actual_property_1.value());
+
+ const TestProperty& actual_property_2 = test_result.GetTestProperty(1);
+ EXPECT_STREQ("key_2", actual_property_2.key());
+ EXPECT_STREQ("2", actual_property_2.value());
+}
+
+// Tests TestResult::RecordProperty() overrides values for duplicate keys.
+TEST(TestResultPropertyTest, OverridesValuesForDuplicateKeys) {
+ TestResult test_result;
+ TestProperty property_1_1("key_1", "1");
+ TestProperty property_2_1("key_2", "2");
+ TestProperty property_1_2("key_1", "12");
+ TestProperty property_2_2("key_2", "22");
+ TestResultAccessor::RecordProperty(&test_result, property_1_1);
+ TestResultAccessor::RecordProperty(&test_result, property_2_1);
+ TestResultAccessor::RecordProperty(&test_result, property_1_2);
+ TestResultAccessor::RecordProperty(&test_result, property_2_2);
+
+ ASSERT_EQ(2, test_result.test_property_count());
+ const TestProperty& actual_property_1 = test_result.GetTestProperty(0);
+ EXPECT_STREQ("key_1", actual_property_1.key());
+ EXPECT_STREQ("12", actual_property_1.value());
+
+ const TestProperty& actual_property_2 = test_result.GetTestProperty(1);
+ EXPECT_STREQ("key_2", actual_property_2.key());
+ EXPECT_STREQ("22", actual_property_2.value());
+}
+
+// Tests TestResult::GetTestProperty().
+TEST(TestResultPropertyDeathTest, GetTestProperty) {
+ TestResult test_result;
+ TestProperty property_1("key_1", "1");
+ TestProperty property_2("key_2", "2");
+ TestProperty property_3("key_3", "3");
+ TestResultAccessor::RecordProperty(&test_result, property_1);
+ TestResultAccessor::RecordProperty(&test_result, property_2);
+ TestResultAccessor::RecordProperty(&test_result, property_3);
+
+ const TestProperty& fetched_property_1 = test_result.GetTestProperty(0);
+ const TestProperty& fetched_property_2 = test_result.GetTestProperty(1);
+ const TestProperty& fetched_property_3 = test_result.GetTestProperty(2);
+
+ EXPECT_STREQ("key_1", fetched_property_1.key());
+ EXPECT_STREQ("1", fetched_property_1.value());
+
+ EXPECT_STREQ("key_2", fetched_property_2.key());
+ EXPECT_STREQ("2", fetched_property_2.value());
+
+ EXPECT_STREQ("key_3", fetched_property_3.key());
+ EXPECT_STREQ("3", fetched_property_3.value());
+
+ EXPECT_DEATH_IF_SUPPORTED(test_result.GetTestProperty(3), "");
+ EXPECT_DEATH_IF_SUPPORTED(test_result.GetTestProperty(-1), "");
+}
+
+// When a property using a reserved key is supplied to this function, it tests
+// that a non-fatal failure is added, a fatal failure is not added, and that the
+// property is not recorded.
+void ExpectNonFatalFailureRecordingPropertyWithReservedKey(const char* key) {
+ TestResult test_result;
+ TestProperty property(key, "1");
+ EXPECT_NONFATAL_FAILURE(
+ TestResultAccessor::RecordProperty(&test_result, property),
+ "Reserved key");
+ ASSERT_EQ(0, test_result.test_property_count()) << "Not recorded";
+}
+
+// Attempting to recording a property with the Reserved literal "name"
+// should add a non-fatal failure and the property should not be recorded.
+TEST(TestResultPropertyTest, AddFailureWhenUsingReservedKeyCalledName) {
+ ExpectNonFatalFailureRecordingPropertyWithReservedKey("name");
+}
+
+// Attempting to recording a property with the Reserved literal "status"
+// should add a non-fatal failure and the property should not be recorded.
+TEST(TestResultPropertyTest, AddFailureWhenUsingReservedKeyCalledStatus) {
+ ExpectNonFatalFailureRecordingPropertyWithReservedKey("status");
+}
+
+// Attempting to recording a property with the Reserved literal "time"
+// should add a non-fatal failure and the property should not be recorded.
+TEST(TestResultPropertyTest, AddFailureWhenUsingReservedKeyCalledTime) {
+ ExpectNonFatalFailureRecordingPropertyWithReservedKey("time");
+}
+
+// Attempting to recording a property with the Reserved literal "classname"
+// should add a non-fatal failure and the property should not be recorded.
+TEST(TestResultPropertyTest, AddFailureWhenUsingReservedKeyCalledClassname) {
+ ExpectNonFatalFailureRecordingPropertyWithReservedKey("classname");
+}
+
+// Tests that GTestFlagSaver works on Windows and Mac.
+
+class GTestFlagSaverTest : public Test {
+ protected:
+ // Saves the Google Test flags such that we can restore them later, and
+ // then sets them to their default values. This will be called
+ // before the first test in this test case is run.
+ static void SetUpTestCase() {
+ saver_ = new GTestFlagSaver;
+
+ GTEST_FLAG(also_run_disabled_tests) = false;
+ GTEST_FLAG(break_on_failure) = false;
+ GTEST_FLAG(catch_exceptions) = false;
+ GTEST_FLAG(death_test_use_fork) = false;
+ GTEST_FLAG(color) = "auto";
+ GTEST_FLAG(filter) = "";
+ GTEST_FLAG(list_tests) = false;
+ GTEST_FLAG(output) = "";
+ GTEST_FLAG(print_time) = true;
+ GTEST_FLAG(random_seed) = 0;
+ GTEST_FLAG(repeat) = 1;
+ GTEST_FLAG(shuffle) = false;
+ GTEST_FLAG(stack_trace_depth) = kMaxStackTraceDepth;
+ GTEST_FLAG(stream_result_to) = "";
+ GTEST_FLAG(throw_on_failure) = false;
+ }
+
+ // Restores the Google Test flags that the tests have modified. This will
+ // be called after the last test in this test case is run.
+ static void TearDownTestCase() {
+ delete saver_;
+ saver_ = NULL;
+ }
+
+ // Verifies that the Google Test flags have their default values, and then
+ // modifies each of them.
+ void VerifyAndModifyFlags() {
+ EXPECT_FALSE(GTEST_FLAG(also_run_disabled_tests));
+ EXPECT_FALSE(GTEST_FLAG(break_on_failure));
+ EXPECT_FALSE(GTEST_FLAG(catch_exceptions));
+ EXPECT_STREQ("auto", GTEST_FLAG(color).c_str());
+ EXPECT_FALSE(GTEST_FLAG(death_test_use_fork));
+ EXPECT_STREQ("", GTEST_FLAG(filter).c_str());
+ EXPECT_FALSE(GTEST_FLAG(list_tests));
+ EXPECT_STREQ("", GTEST_FLAG(output).c_str());
+ EXPECT_TRUE(GTEST_FLAG(print_time));
+ EXPECT_EQ(0, GTEST_FLAG(random_seed));
+ EXPECT_EQ(1, GTEST_FLAG(repeat));
+ EXPECT_FALSE(GTEST_FLAG(shuffle));
+ EXPECT_EQ(kMaxStackTraceDepth, GTEST_FLAG(stack_trace_depth));
+ EXPECT_STREQ("", GTEST_FLAG(stream_result_to).c_str());
+ EXPECT_FALSE(GTEST_FLAG(throw_on_failure));
+
+ GTEST_FLAG(also_run_disabled_tests) = true;
+ GTEST_FLAG(break_on_failure) = true;
+ GTEST_FLAG(catch_exceptions) = true;
+ GTEST_FLAG(color) = "no";
+ GTEST_FLAG(death_test_use_fork) = true;
+ GTEST_FLAG(filter) = "abc";
+ GTEST_FLAG(list_tests) = true;
+ GTEST_FLAG(output) = "xml:foo.xml";
+ GTEST_FLAG(print_time) = false;
+ GTEST_FLAG(random_seed) = 1;
+ GTEST_FLAG(repeat) = 100;
+ GTEST_FLAG(shuffle) = true;
+ GTEST_FLAG(stack_trace_depth) = 1;
+ GTEST_FLAG(stream_result_to) = "localhost:1234";
+ GTEST_FLAG(throw_on_failure) = true;
+ }
+ private:
+ // For saving Google Test flags during this test case.
+ static GTestFlagSaver* saver_;
+};
+
+GTestFlagSaver* GTestFlagSaverTest::saver_ = NULL;
+
+// Google Test doesn't guarantee the order of tests. The following two
+// tests are designed to work regardless of their order.
+
+// Modifies the Google Test flags in the test body.
+TEST_F(GTestFlagSaverTest, ModifyGTestFlags) {
+ VerifyAndModifyFlags();
+}
+
+// Verifies that the Google Test flags in the body of the previous test were
+// restored to their original values.
+TEST_F(GTestFlagSaverTest, VerifyGTestFlags) {
+ VerifyAndModifyFlags();
+}
+
+// Sets an environment variable with the given name to the given
+// value. If the value argument is "", unsets the environment
+// variable. The caller must ensure that both arguments are not NULL.
+static void SetEnv(const char* name, const char* value) {
+#if GTEST_OS_WINDOWS_MOBILE
+ // Environment variables are not supported on Windows CE.
+ return;
+#elif defined(__BORLANDC__) || defined(__SunOS_5_8) || defined(__SunOS_5_9)
+ // C++Builder's putenv only stores a pointer to its parameter; we have to
+ // ensure that the string remains valid as long as it might be needed.
+ // We use an std::map to do so.
+ static std::map<String, String*> added_env;
+
+ // Because putenv stores a pointer to the string buffer, we can't delete the
+ // previous string (if present) until after it's replaced.
+ String *prev_env = NULL;
+ if (added_env.find(name) != added_env.end()) {
+ prev_env = added_env[name];
+ }
+ added_env[name] = new String((Message() << name << "=" << value).GetString());
+
+ // The standard signature of putenv accepts a 'char*' argument. Other
+ // implementations, like C++Builder's, accept a 'const char*'.
+ // We cast away the 'const' since that would work for both variants.
+ putenv(const_cast<char*>(added_env[name]->c_str()));
+ delete prev_env;
+#elif GTEST_OS_WINDOWS // If we are on Windows proper.
+ _putenv((Message() << name << "=" << value).GetString().c_str());
+#else
+ if (*value == '\0') {
+ unsetenv(name);
+ } else {
+ setenv(name, value, 1);
+ }
+#endif // GTEST_OS_WINDOWS_MOBILE
+}
+
+#if !GTEST_OS_WINDOWS_MOBILE
+// Environment variables are not supported on Windows CE.
+
+using testing::internal::Int32FromGTestEnv;
+
+// Tests Int32FromGTestEnv().
+
+// Tests that Int32FromGTestEnv() returns the default value when the
+// environment variable is not set.
+TEST(Int32FromGTestEnvTest, ReturnsDefaultWhenVariableIsNotSet) {
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "TEMP", "");
+ EXPECT_EQ(10, Int32FromGTestEnv("temp", 10));
+}
+
+// Tests that Int32FromGTestEnv() returns the default value when the
+// environment variable overflows as an Int32.
+TEST(Int32FromGTestEnvTest, ReturnsDefaultWhenValueOverflows) {
+ printf("(expecting 2 warnings)\n");
+
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "TEMP", "12345678987654321");
+ EXPECT_EQ(20, Int32FromGTestEnv("temp", 20));
+
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "TEMP", "-12345678987654321");
+ EXPECT_EQ(30, Int32FromGTestEnv("temp", 30));
+}
+
+// Tests that Int32FromGTestEnv() returns the default value when the
+// environment variable does not represent a valid decimal integer.
+TEST(Int32FromGTestEnvTest, ReturnsDefaultWhenValueIsInvalid) {
+ printf("(expecting 2 warnings)\n");
+
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "TEMP", "A1");
+ EXPECT_EQ(40, Int32FromGTestEnv("temp", 40));
+
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "TEMP", "12X");
+ EXPECT_EQ(50, Int32FromGTestEnv("temp", 50));
+}
+
+// Tests that Int32FromGTestEnv() parses and returns the value of the
+// environment variable when it represents a valid decimal integer in
+// the range of an Int32.
+TEST(Int32FromGTestEnvTest, ParsesAndReturnsValidValue) {
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "TEMP", "123");
+ EXPECT_EQ(123, Int32FromGTestEnv("temp", 0));
+
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "TEMP", "-321");
+ EXPECT_EQ(-321, Int32FromGTestEnv("temp", 0));
+}
+#endif // !GTEST_OS_WINDOWS_MOBILE
+
+// Tests ParseInt32Flag().
+
+// Tests that ParseInt32Flag() returns false and doesn't change the
+// output value when the flag has wrong format
+TEST(ParseInt32FlagTest, ReturnsFalseForInvalidFlag) {
+ Int32 value = 123;
+ EXPECT_FALSE(ParseInt32Flag("--a=100", "b", &value));
+ EXPECT_EQ(123, value);
+
+ EXPECT_FALSE(ParseInt32Flag("a=100", "a", &value));
+ EXPECT_EQ(123, value);
+}
+
+// Tests that ParseInt32Flag() returns false and doesn't change the
+// output value when the flag overflows as an Int32.
+TEST(ParseInt32FlagTest, ReturnsDefaultWhenValueOverflows) {
+ printf("(expecting 2 warnings)\n");
+
+ Int32 value = 123;
+ EXPECT_FALSE(ParseInt32Flag("--abc=12345678987654321", "abc", &value));
+ EXPECT_EQ(123, value);
+
+ EXPECT_FALSE(ParseInt32Flag("--abc=-12345678987654321", "abc", &value));
+ EXPECT_EQ(123, value);
+}
+
+// Tests that ParseInt32Flag() returns false and doesn't change the
+// output value when the flag does not represent a valid decimal
+// integer.
+TEST(ParseInt32FlagTest, ReturnsDefaultWhenValueIsInvalid) {
+ printf("(expecting 2 warnings)\n");
+
+ Int32 value = 123;
+ EXPECT_FALSE(ParseInt32Flag("--abc=A1", "abc", &value));
+ EXPECT_EQ(123, value);
+
+ EXPECT_FALSE(ParseInt32Flag("--abc=12X", "abc", &value));
+ EXPECT_EQ(123, value);
+}
+
+// Tests that ParseInt32Flag() parses the value of the flag and
+// returns true when the flag represents a valid decimal integer in
+// the range of an Int32.
+TEST(ParseInt32FlagTest, ParsesAndReturnsValidValue) {
+ Int32 value = 123;
+ EXPECT_TRUE(ParseInt32Flag("--" GTEST_FLAG_PREFIX_ "abc=456", "abc", &value));
+ EXPECT_EQ(456, value);
+
+ EXPECT_TRUE(ParseInt32Flag("--" GTEST_FLAG_PREFIX_ "abc=-789",
+ "abc", &value));
+ EXPECT_EQ(-789, value);
+}
+
+// Tests that Int32FromEnvOrDie() parses the value of the var or
+// returns the correct default.
+// Environment variables are not supported on Windows CE.
+#if !GTEST_OS_WINDOWS_MOBILE
+TEST(Int32FromEnvOrDieTest, ParsesAndReturnsValidValue) {
+ EXPECT_EQ(333, Int32FromEnvOrDie(GTEST_FLAG_PREFIX_UPPER_ "UnsetVar", 333));
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "UnsetVar", "123");
+ EXPECT_EQ(123, Int32FromEnvOrDie(GTEST_FLAG_PREFIX_UPPER_ "UnsetVar", 333));
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "UnsetVar", "-123");
+ EXPECT_EQ(-123, Int32FromEnvOrDie(GTEST_FLAG_PREFIX_UPPER_ "UnsetVar", 333));
+}
+#endif // !GTEST_OS_WINDOWS_MOBILE
+
+// Tests that Int32FromEnvOrDie() aborts with an error message
+// if the variable is not an Int32.
+TEST(Int32FromEnvOrDieDeathTest, AbortsOnFailure) {
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "VAR", "xxx");
+ EXPECT_DEATH_IF_SUPPORTED(
+ Int32FromEnvOrDie(GTEST_FLAG_PREFIX_UPPER_ "VAR", 123),
+ ".*");
+}
+
+// Tests that Int32FromEnvOrDie() aborts with an error message
+// if the variable cannot be represnted by an Int32.
+TEST(Int32FromEnvOrDieDeathTest, AbortsOnInt32Overflow) {
+ SetEnv(GTEST_FLAG_PREFIX_UPPER_ "VAR", "1234567891234567891234");
+ EXPECT_DEATH_IF_SUPPORTED(
+ Int32FromEnvOrDie(GTEST_FLAG_PREFIX_UPPER_ "VAR", 123),
+ ".*");
+}
+
+// Tests that ShouldRunTestOnShard() selects all tests
+// where there is 1 shard.
+TEST(ShouldRunTestOnShardTest, IsPartitionWhenThereIsOneShard) {
+ EXPECT_TRUE(ShouldRunTestOnShard(1, 0, 0));
+ EXPECT_TRUE(ShouldRunTestOnShard(1, 0, 1));
+ EXPECT_TRUE(ShouldRunTestOnShard(1, 0, 2));
+ EXPECT_TRUE(ShouldRunTestOnShard(1, 0, 3));
+ EXPECT_TRUE(ShouldRunTestOnShard(1, 0, 4));
+}
+
+class ShouldShardTest : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ index_var_ = GTEST_FLAG_PREFIX_UPPER_ "INDEX";
+ total_var_ = GTEST_FLAG_PREFIX_UPPER_ "TOTAL";
+ }
+
+ virtual void TearDown() {
+ SetEnv(index_var_, "");
+ SetEnv(total_var_, "");
+ }
+
+ const char* index_var_;
+ const char* total_var_;
+};
+
+// Tests that sharding is disabled if neither of the environment variables
+// are set.
+TEST_F(ShouldShardTest, ReturnsFalseWhenNeitherEnvVarIsSet) {
+ SetEnv(index_var_, "");
+ SetEnv(total_var_, "");
+
+ EXPECT_FALSE(ShouldShard(total_var_, index_var_, false));
+ EXPECT_FALSE(ShouldShard(total_var_, index_var_, true));
+}
+
+// Tests that sharding is not enabled if total_shards == 1.
+TEST_F(ShouldShardTest, ReturnsFalseWhenTotalShardIsOne) {
+ SetEnv(index_var_, "0");
+ SetEnv(total_var_, "1");
+ EXPECT_FALSE(ShouldShard(total_var_, index_var_, false));
+ EXPECT_FALSE(ShouldShard(total_var_, index_var_, true));
+}
+
+// Tests that sharding is enabled if total_shards > 1 and
+// we are not in a death test subprocess.
+// Environment variables are not supported on Windows CE.
+#if !GTEST_OS_WINDOWS_MOBILE
+TEST_F(ShouldShardTest, WorksWhenShardEnvVarsAreValid) {
+ SetEnv(index_var_, "4");
+ SetEnv(total_var_, "22");
+ EXPECT_TRUE(ShouldShard(total_var_, index_var_, false));
+ EXPECT_FALSE(ShouldShard(total_var_, index_var_, true));
+
+ SetEnv(index_var_, "8");
+ SetEnv(total_var_, "9");
+ EXPECT_TRUE(ShouldShard(total_var_, index_var_, false));
+ EXPECT_FALSE(ShouldShard(total_var_, index_var_, true));
+
+ SetEnv(index_var_, "0");
+ SetEnv(total_var_, "9");
+ EXPECT_TRUE(ShouldShard(total_var_, index_var_, false));
+ EXPECT_FALSE(ShouldShard(total_var_, index_var_, true));
+}
+#endif // !GTEST_OS_WINDOWS_MOBILE
+
+// Tests that we exit in error if the sharding values are not valid.
+
+typedef ShouldShardTest ShouldShardDeathTest;
+
+TEST_F(ShouldShardDeathTest, AbortsWhenShardingEnvVarsAreInvalid) {
+ SetEnv(index_var_, "4");
+ SetEnv(total_var_, "4");
+ EXPECT_DEATH_IF_SUPPORTED(ShouldShard(total_var_, index_var_, false), ".*");
+
+ SetEnv(index_var_, "4");
+ SetEnv(total_var_, "-2");
+ EXPECT_DEATH_IF_SUPPORTED(ShouldShard(total_var_, index_var_, false), ".*");
+
+ SetEnv(index_var_, "5");
+ SetEnv(total_var_, "");
+ EXPECT_DEATH_IF_SUPPORTED(ShouldShard(total_var_, index_var_, false), ".*");
+
+ SetEnv(index_var_, "");
+ SetEnv(total_var_, "5");
+ EXPECT_DEATH_IF_SUPPORTED(ShouldShard(total_var_, index_var_, false), ".*");
+}
+
+// Tests that ShouldRunTestOnShard is a partition when 5
+// shards are used.
+TEST(ShouldRunTestOnShardTest, IsPartitionWhenThereAreFiveShards) {
+ // Choose an arbitrary number of tests and shards.
+ const int num_tests = 17;
+ const int num_shards = 5;
+
+ // Check partitioning: each test should be on exactly 1 shard.
+ for (int test_id = 0; test_id < num_tests; test_id++) {
+ int prev_selected_shard_index = -1;
+ for (int shard_index = 0; shard_index < num_shards; shard_index++) {
+ if (ShouldRunTestOnShard(num_shards, shard_index, test_id)) {
+ if (prev_selected_shard_index < 0) {
+ prev_selected_shard_index = shard_index;
+ } else {
+ ADD_FAILURE() << "Shard " << prev_selected_shard_index << " and "
+ << shard_index << " are both selected to run test " << test_id;
+ }
+ }
+ }
+ }
+
+ // Check balance: This is not required by the sharding protocol, but is a
+ // desirable property for performance.
+ for (int shard_index = 0; shard_index < num_shards; shard_index++) {
+ int num_tests_on_shard = 0;
+ for (int test_id = 0; test_id < num_tests; test_id++) {
+ num_tests_on_shard +=
+ ShouldRunTestOnShard(num_shards, shard_index, test_id);
+ }
+ EXPECT_GE(num_tests_on_shard, num_tests / num_shards);
+ }
+}
+
+// For the same reason we are not explicitly testing everything in the
+// Test class, there are no separate tests for the following classes
+// (except for some trivial cases):
+//
+// TestCase, UnitTest, UnitTestResultPrinter.
+//
+// Similarly, there are no separate tests for the following macros:
+//
+// TEST, TEST_F, RUN_ALL_TESTS
+
+TEST(UnitTestTest, CanGetOriginalWorkingDir) {
+ ASSERT_TRUE(UnitTest::GetInstance()->original_working_dir() != NULL);
+ EXPECT_STRNE(UnitTest::GetInstance()->original_working_dir(), "");
+}
+
+// This group of tests is for predicate assertions (ASSERT_PRED*, etc)
+// of various arities. They do not attempt to be exhaustive. Rather,
+// view them as smoke tests that can be easily reviewed and verified.
+// A more complete set of tests for predicate assertions can be found
+// in gtest_pred_impl_unittest.cc.
+
+// First, some predicates and predicate-formatters needed by the tests.
+
+// Returns true iff the argument is an even number.
+bool IsEven(int n) {
+ return (n % 2) == 0;
+}
+
+// A functor that returns true iff the argument is an even number.
+struct IsEvenFunctor {
+ bool operator()(int n) { return IsEven(n); }
+};
+
+// A predicate-formatter function that asserts the argument is an even
+// number.
+AssertionResult AssertIsEven(const char* expr, int n) {
+ if (IsEven(n)) {
+ return AssertionSuccess();
+ }
+
+ Message msg;
+ msg << expr << " evaluates to " << n << ", which is not even.";
+ return AssertionFailure(msg);
+}
+
+// A predicate function that returns AssertionResult for use in
+// EXPECT/ASSERT_TRUE/FALSE.
+AssertionResult ResultIsEven(int n) {
+ if (IsEven(n))
+ return AssertionSuccess() << n << " is even";
+ else
+ return AssertionFailure() << n << " is odd";
+}
+
+// A predicate function that returns AssertionResult but gives no
+// explanation why it succeeds. Needed for testing that
+// EXPECT/ASSERT_FALSE handles such functions correctly.
+AssertionResult ResultIsEvenNoExplanation(int n) {
+ if (IsEven(n))
+ return AssertionSuccess();
+ else
+ return AssertionFailure() << n << " is odd";
+}
+
+// A predicate-formatter functor that asserts the argument is an even
+// number.
+struct AssertIsEvenFunctor {
+ AssertionResult operator()(const char* expr, int n) {
+ return AssertIsEven(expr, n);
+ }
+};
+
+// Returns true iff the sum of the arguments is an even number.
+bool SumIsEven2(int n1, int n2) {
+ return IsEven(n1 + n2);
+}
+
+// A functor that returns true iff the sum of the arguments is an even
+// number.
+struct SumIsEven3Functor {
+ bool operator()(int n1, int n2, int n3) {
+ return IsEven(n1 + n2 + n3);
+ }
+};
+
+// A predicate-formatter function that asserts the sum of the
+// arguments is an even number.
+AssertionResult AssertSumIsEven4(
+ const char* e1, const char* e2, const char* e3, const char* e4,
+ int n1, int n2, int n3, int n4) {
+ const int sum = n1 + n2 + n3 + n4;
+ if (IsEven(sum)) {
+ return AssertionSuccess();
+ }
+
+ Message msg;
+ msg << e1 << " + " << e2 << " + " << e3 << " + " << e4
+ << " (" << n1 << " + " << n2 << " + " << n3 << " + " << n4
+ << ") evaluates to " << sum << ", which is not even.";
+ return AssertionFailure(msg);
+}
+
+// A predicate-formatter functor that asserts the sum of the arguments
+// is an even number.
+struct AssertSumIsEven5Functor {
+ AssertionResult operator()(
+ const char* e1, const char* e2, const char* e3, const char* e4,
+ const char* e5, int n1, int n2, int n3, int n4, int n5) {
+ const int sum = n1 + n2 + n3 + n4 + n5;
+ if (IsEven(sum)) {
+ return AssertionSuccess();
+ }
+
+ Message msg;
+ msg << e1 << " + " << e2 << " + " << e3 << " + " << e4 << " + " << e5
+ << " ("
+ << n1 << " + " << n2 << " + " << n3 << " + " << n4 << " + " << n5
+ << ") evaluates to " << sum << ", which is not even.";
+ return AssertionFailure(msg);
+ }
+};
+
+
+// Tests unary predicate assertions.
+
+// Tests unary predicate assertions that don't use a custom formatter.
+TEST(Pred1Test, WithoutFormat) {
+ // Success cases.
+ EXPECT_PRED1(IsEvenFunctor(), 2) << "This failure is UNEXPECTED!";
+ ASSERT_PRED1(IsEven, 4);
+
+ // Failure cases.
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED1(IsEven, 5) << "This failure is expected.";
+ }, "This failure is expected.");
+ EXPECT_FATAL_FAILURE(ASSERT_PRED1(IsEvenFunctor(), 5),
+ "evaluates to false");
+}
+
+// Tests unary predicate assertions that use a custom formatter.
+TEST(Pred1Test, WithFormat) {
+ // Success cases.
+ EXPECT_PRED_FORMAT1(AssertIsEven, 2);
+ ASSERT_PRED_FORMAT1(AssertIsEvenFunctor(), 4)
+ << "This failure is UNEXPECTED!";
+
+ // Failure cases.
+ const int n = 5;
+ EXPECT_NONFATAL_FAILURE(EXPECT_PRED_FORMAT1(AssertIsEvenFunctor(), n),
+ "n evaluates to 5, which is not even.");
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT1(AssertIsEven, 5) << "This failure is expected.";
+ }, "This failure is expected.");
+}
+
+// Tests that unary predicate assertions evaluates their arguments
+// exactly once.
+TEST(Pred1Test, SingleEvaluationOnFailure) {
+ // A success case.
+ static int n = 0;
+ EXPECT_PRED1(IsEven, n++);
+ EXPECT_EQ(1, n) << "The argument is not evaluated exactly once.";
+
+ // A failure case.
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT1(AssertIsEvenFunctor(), n++)
+ << "This failure is expected.";
+ }, "This failure is expected.");
+ EXPECT_EQ(2, n) << "The argument is not evaluated exactly once.";
+}
+
+
+// Tests predicate assertions whose arity is >= 2.
+
+// Tests predicate assertions that don't use a custom formatter.
+TEST(PredTest, WithoutFormat) {
+ // Success cases.
+ ASSERT_PRED2(SumIsEven2, 2, 4) << "This failure is UNEXPECTED!";
+ EXPECT_PRED3(SumIsEven3Functor(), 4, 6, 8);
+
+ // Failure cases.
+ const int n1 = 1;
+ const int n2 = 2;
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED2(SumIsEven2, n1, n2) << "This failure is expected.";
+ }, "This failure is expected.");
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED3(SumIsEven3Functor(), 1, 2, 4);
+ }, "evaluates to false");
+}
+
+// Tests predicate assertions that use a custom formatter.
+TEST(PredTest, WithFormat) {
+ // Success cases.
+ ASSERT_PRED_FORMAT4(AssertSumIsEven4, 4, 6, 8, 10) <<
+ "This failure is UNEXPECTED!";
+ EXPECT_PRED_FORMAT5(AssertSumIsEven5Functor(), 2, 4, 6, 8, 10);
+
+ // Failure cases.
+ const int n1 = 1;
+ const int n2 = 2;
+ const int n3 = 4;
+ const int n4 = 6;
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT4(AssertSumIsEven4, n1, n2, n3, n4);
+ }, "evaluates to 13, which is not even.");
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT5(AssertSumIsEven5Functor(), 1, 2, 4, 6, 8)
+ << "This failure is expected.";
+ }, "This failure is expected.");
+}
+
+// Tests that predicate assertions evaluates their arguments
+// exactly once.
+TEST(PredTest, SingleEvaluationOnFailure) {
+ // A success case.
+ int n1 = 0;
+ int n2 = 0;
+ EXPECT_PRED2(SumIsEven2, n1++, n2++);
+ EXPECT_EQ(1, n1) << "Argument 1 is not evaluated exactly once.";
+ EXPECT_EQ(1, n2) << "Argument 2 is not evaluated exactly once.";
+
+ // Another success case.
+ n1 = n2 = 0;
+ int n3 = 0;
+ int n4 = 0;
+ int n5 = 0;
+ ASSERT_PRED_FORMAT5(AssertSumIsEven5Functor(),
+ n1++, n2++, n3++, n4++, n5++)
+ << "This failure is UNEXPECTED!";
+ EXPECT_EQ(1, n1) << "Argument 1 is not evaluated exactly once.";
+ EXPECT_EQ(1, n2) << "Argument 2 is not evaluated exactly once.";
+ EXPECT_EQ(1, n3) << "Argument 3 is not evaluated exactly once.";
+ EXPECT_EQ(1, n4) << "Argument 4 is not evaluated exactly once.";
+ EXPECT_EQ(1, n5) << "Argument 5 is not evaluated exactly once.";
+
+ // A failure case.
+ n1 = n2 = n3 = 0;
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED3(SumIsEven3Functor(), ++n1, n2++, n3++)
+ << "This failure is expected.";
+ }, "This failure is expected.");
+ EXPECT_EQ(1, n1) << "Argument 1 is not evaluated exactly once.";
+ EXPECT_EQ(1, n2) << "Argument 2 is not evaluated exactly once.";
+ EXPECT_EQ(1, n3) << "Argument 3 is not evaluated exactly once.";
+
+ // Another failure case.
+ n1 = n2 = n3 = n4 = 0;
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT4(AssertSumIsEven4, ++n1, n2++, n3++, n4++);
+ }, "evaluates to 1, which is not even.");
+ EXPECT_EQ(1, n1) << "Argument 1 is not evaluated exactly once.";
+ EXPECT_EQ(1, n2) << "Argument 2 is not evaluated exactly once.";
+ EXPECT_EQ(1, n3) << "Argument 3 is not evaluated exactly once.";
+ EXPECT_EQ(1, n4) << "Argument 4 is not evaluated exactly once.";
+}
+
+
+// Some helper functions for testing using overloaded/template
+// functions with ASSERT_PREDn and EXPECT_PREDn.
+
+bool IsPositive(double x) {
+ return x > 0;
+}
+
+template <typename T>
+bool IsNegative(T x) {
+ return x < 0;
+}
+
+template <typename T1, typename T2>
+bool GreaterThan(T1 x1, T2 x2) {
+ return x1 > x2;
+}
+
+// Tests that overloaded functions can be used in *_PRED* as long as
+// their types are explicitly specified.
+TEST(PredicateAssertionTest, AcceptsOverloadedFunction) {
+ // C++Builder requires C-style casts rather than static_cast.
+ EXPECT_PRED1((bool (*)(int))(IsPositive), 5); // NOLINT
+ ASSERT_PRED1((bool (*)(double))(IsPositive), 6.0); // NOLINT
+}
+
+// Tests that template functions can be used in *_PRED* as long as
+// their types are explicitly specified.
+TEST(PredicateAssertionTest, AcceptsTemplateFunction) {
+ EXPECT_PRED1(IsNegative<int>, -5);
+ // Makes sure that we can handle templates with more than one
+ // parameter.
+ ASSERT_PRED2((GreaterThan<int, int>), 5, 0);
+}
+
+
+// Some helper functions for testing using overloaded/template
+// functions with ASSERT_PRED_FORMATn and EXPECT_PRED_FORMATn.
+
+AssertionResult IsPositiveFormat(const char* /* expr */, int n) {
+ return n > 0 ? AssertionSuccess() :
+ AssertionFailure(Message() << "Failure");
+}
+
+AssertionResult IsPositiveFormat(const char* /* expr */, double x) {
+ return x > 0 ? AssertionSuccess() :
+ AssertionFailure(Message() << "Failure");
+}
+
+template <typename T>
+AssertionResult IsNegativeFormat(const char* /* expr */, T x) {
+ return x < 0 ? AssertionSuccess() :
+ AssertionFailure(Message() << "Failure");
+}
+
+template <typename T1, typename T2>
+AssertionResult EqualsFormat(const char* /* expr1 */, const char* /* expr2 */,
+ const T1& x1, const T2& x2) {
+ return x1 == x2 ? AssertionSuccess() :
+ AssertionFailure(Message() << "Failure");
+}
+
+// Tests that overloaded functions can be used in *_PRED_FORMAT*
+// without explicitly specifying their types.
+TEST(PredicateFormatAssertionTest, AcceptsOverloadedFunction) {
+ EXPECT_PRED_FORMAT1(IsPositiveFormat, 5);
+ ASSERT_PRED_FORMAT1(IsPositiveFormat, 6.0);
+}
+
+// Tests that template functions can be used in *_PRED_FORMAT* without
+// explicitly specifying their types.
+TEST(PredicateFormatAssertionTest, AcceptsTemplateFunction) {
+ EXPECT_PRED_FORMAT1(IsNegativeFormat, -5);
+ ASSERT_PRED_FORMAT2(EqualsFormat, 3, 3);
+}
+
+
+// Tests string assertions.
+
+// Tests ASSERT_STREQ with non-NULL arguments.
+TEST(StringAssertionTest, ASSERT_STREQ) {
+ const char * const p1 = "good";
+ ASSERT_STREQ(p1, p1);
+
+ // Let p2 have the same content as p1, but be at a different address.
+ const char p2[] = "good";
+ ASSERT_STREQ(p1, p2);
+
+ EXPECT_FATAL_FAILURE(ASSERT_STREQ("bad", "good"),
+ "Expected: \"bad\"");
+}
+
+// Tests ASSERT_STREQ with NULL arguments.
+TEST(StringAssertionTest, ASSERT_STREQ_Null) {
+ ASSERT_STREQ(static_cast<const char *>(NULL), NULL);
+ EXPECT_FATAL_FAILURE(ASSERT_STREQ(NULL, "non-null"),
+ "non-null");
+}
+
+// Tests ASSERT_STREQ with NULL arguments.
+TEST(StringAssertionTest, ASSERT_STREQ_Null2) {
+ EXPECT_FATAL_FAILURE(ASSERT_STREQ("non-null", NULL),
+ "non-null");
+}
+
+// Tests ASSERT_STRNE.
+TEST(StringAssertionTest, ASSERT_STRNE) {
+ ASSERT_STRNE("hi", "Hi");
+ ASSERT_STRNE("Hi", NULL);
+ ASSERT_STRNE(NULL, "Hi");
+ ASSERT_STRNE("", NULL);
+ ASSERT_STRNE(NULL, "");
+ ASSERT_STRNE("", "Hi");
+ ASSERT_STRNE("Hi", "");
+ EXPECT_FATAL_FAILURE(ASSERT_STRNE("Hi", "Hi"),
+ "\"Hi\" vs \"Hi\"");
+}
+
+// Tests ASSERT_STRCASEEQ.
+TEST(StringAssertionTest, ASSERT_STRCASEEQ) {
+ ASSERT_STRCASEEQ("hi", "Hi");
+ ASSERT_STRCASEEQ(static_cast<const char *>(NULL), NULL);
+
+ ASSERT_STRCASEEQ("", "");
+ EXPECT_FATAL_FAILURE(ASSERT_STRCASEEQ("Hi", "hi2"),
+ "(ignoring case)");
+}
+
+// Tests ASSERT_STRCASENE.
+TEST(StringAssertionTest, ASSERT_STRCASENE) {
+ ASSERT_STRCASENE("hi1", "Hi2");
+ ASSERT_STRCASENE("Hi", NULL);
+ ASSERT_STRCASENE(NULL, "Hi");
+ ASSERT_STRCASENE("", NULL);
+ ASSERT_STRCASENE(NULL, "");
+ ASSERT_STRCASENE("", "Hi");
+ ASSERT_STRCASENE("Hi", "");
+ EXPECT_FATAL_FAILURE(ASSERT_STRCASENE("Hi", "hi"),
+ "(ignoring case)");
+}
+
+// Tests *_STREQ on wide strings.
+TEST(StringAssertionTest, STREQ_Wide) {
+ // NULL strings.
+ ASSERT_STREQ(static_cast<const wchar_t *>(NULL), NULL);
+
+ // Empty strings.
+ ASSERT_STREQ(L"", L"");
+
+ // Non-null vs NULL.
+ EXPECT_NONFATAL_FAILURE(EXPECT_STREQ(L"non-null", NULL),
+ "non-null");
+
+ // Equal strings.
+ EXPECT_STREQ(L"Hi", L"Hi");
+
+ // Unequal strings.
+ EXPECT_NONFATAL_FAILURE(EXPECT_STREQ(L"abc", L"Abc"),
+ "Abc");
+
+ // Strings containing wide characters.
+ EXPECT_NONFATAL_FAILURE(EXPECT_STREQ(L"abc\x8119", L"abc\x8120"),
+ "abc");
+}
+
+// Tests *_STRNE on wide strings.
+TEST(StringAssertionTest, STRNE_Wide) {
+ // NULL strings.
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_STRNE(static_cast<const wchar_t *>(NULL), NULL);
+ }, "");
+
+ // Empty strings.
+ EXPECT_NONFATAL_FAILURE(EXPECT_STRNE(L"", L""),
+ "L\"\"");
+
+ // Non-null vs NULL.
+ ASSERT_STRNE(L"non-null", NULL);
+
+ // Equal strings.
+ EXPECT_NONFATAL_FAILURE(EXPECT_STRNE(L"Hi", L"Hi"),
+ "L\"Hi\"");
+
+ // Unequal strings.
+ EXPECT_STRNE(L"abc", L"Abc");
+
+ // Strings containing wide characters.
+ EXPECT_NONFATAL_FAILURE(EXPECT_STRNE(L"abc\x8119", L"abc\x8119"),
+ "abc");
+}
+
+// Tests for ::testing::IsSubstring().
+
+// Tests that IsSubstring() returns the correct result when the input
+// argument type is const char*.
+TEST(IsSubstringTest, ReturnsCorrectResultForCString) {
+ EXPECT_FALSE(IsSubstring("", "", NULL, "a"));
+ EXPECT_FALSE(IsSubstring("", "", "b", NULL));
+ EXPECT_FALSE(IsSubstring("", "", "needle", "haystack"));
+
+ EXPECT_TRUE(IsSubstring("", "", static_cast<const char*>(NULL), NULL));
+ EXPECT_TRUE(IsSubstring("", "", "needle", "two needles"));
+}
+
+// Tests that IsSubstring() returns the correct result when the input
+// argument type is const wchar_t*.
+TEST(IsSubstringTest, ReturnsCorrectResultForWideCString) {
+ EXPECT_FALSE(IsSubstring("", "", kNull, L"a"));
+ EXPECT_FALSE(IsSubstring("", "", L"b", kNull));
+ EXPECT_FALSE(IsSubstring("", "", L"needle", L"haystack"));
+
+ EXPECT_TRUE(IsSubstring("", "", static_cast<const wchar_t*>(NULL), NULL));
+ EXPECT_TRUE(IsSubstring("", "", L"needle", L"two needles"));
+}
+
+// Tests that IsSubstring() generates the correct message when the input
+// argument type is const char*.
+TEST(IsSubstringTest, GeneratesCorrectMessageForCString) {
+ EXPECT_STREQ("Value of: needle_expr\n"
+ " Actual: \"needle\"\n"
+ "Expected: a substring of haystack_expr\n"
+ "Which is: \"haystack\"",
+ IsSubstring("needle_expr", "haystack_expr",
+ "needle", "haystack").failure_message());
+}
+
+// Tests that IsSubstring returns the correct result when the input
+// argument type is ::std::string.
+TEST(IsSubstringTest, ReturnsCorrectResultsForStdString) {
+ EXPECT_TRUE(IsSubstring("", "", std::string("hello"), "ahellob"));
+ EXPECT_FALSE(IsSubstring("", "", "hello", std::string("world")));
+}
+
+#if GTEST_HAS_STD_WSTRING
+// Tests that IsSubstring returns the correct result when the input
+// argument type is ::std::wstring.
+TEST(IsSubstringTest, ReturnsCorrectResultForStdWstring) {
+ EXPECT_TRUE(IsSubstring("", "", ::std::wstring(L"needle"), L"two needles"));
+ EXPECT_FALSE(IsSubstring("", "", L"needle", ::std::wstring(L"haystack")));
+}
+
+// Tests that IsSubstring() generates the correct message when the input
+// argument type is ::std::wstring.
+TEST(IsSubstringTest, GeneratesCorrectMessageForWstring) {
+ EXPECT_STREQ("Value of: needle_expr\n"
+ " Actual: L\"needle\"\n"
+ "Expected: a substring of haystack_expr\n"
+ "Which is: L\"haystack\"",
+ IsSubstring(
+ "needle_expr", "haystack_expr",
+ ::std::wstring(L"needle"), L"haystack").failure_message());
+}
+
+#endif // GTEST_HAS_STD_WSTRING
+
+// Tests for ::testing::IsNotSubstring().
+
+// Tests that IsNotSubstring() returns the correct result when the input
+// argument type is const char*.
+TEST(IsNotSubstringTest, ReturnsCorrectResultForCString) {
+ EXPECT_TRUE(IsNotSubstring("", "", "needle", "haystack"));
+ EXPECT_FALSE(IsNotSubstring("", "", "needle", "two needles"));
+}
+
+// Tests that IsNotSubstring() returns the correct result when the input
+// argument type is const wchar_t*.
+TEST(IsNotSubstringTest, ReturnsCorrectResultForWideCString) {
+ EXPECT_TRUE(IsNotSubstring("", "", L"needle", L"haystack"));
+ EXPECT_FALSE(IsNotSubstring("", "", L"needle", L"two needles"));
+}
+
+// Tests that IsNotSubstring() generates the correct message when the input
+// argument type is const wchar_t*.
+TEST(IsNotSubstringTest, GeneratesCorrectMessageForWideCString) {
+ EXPECT_STREQ("Value of: needle_expr\n"
+ " Actual: L\"needle\"\n"
+ "Expected: not a substring of haystack_expr\n"
+ "Which is: L\"two needles\"",
+ IsNotSubstring(
+ "needle_expr", "haystack_expr",
+ L"needle", L"two needles").failure_message());
+}
+
+// Tests that IsNotSubstring returns the correct result when the input
+// argument type is ::std::string.
+TEST(IsNotSubstringTest, ReturnsCorrectResultsForStdString) {
+ EXPECT_FALSE(IsNotSubstring("", "", std::string("hello"), "ahellob"));
+ EXPECT_TRUE(IsNotSubstring("", "", "hello", std::string("world")));
+}
+
+// Tests that IsNotSubstring() generates the correct message when the input
+// argument type is ::std::string.
+TEST(IsNotSubstringTest, GeneratesCorrectMessageForStdString) {
+ EXPECT_STREQ("Value of: needle_expr\n"
+ " Actual: \"needle\"\n"
+ "Expected: not a substring of haystack_expr\n"
+ "Which is: \"two needles\"",
+ IsNotSubstring(
+ "needle_expr", "haystack_expr",
+ ::std::string("needle"), "two needles").failure_message());
+}
+
+#if GTEST_HAS_STD_WSTRING
+
+// Tests that IsNotSubstring returns the correct result when the input
+// argument type is ::std::wstring.
+TEST(IsNotSubstringTest, ReturnsCorrectResultForStdWstring) {
+ EXPECT_FALSE(
+ IsNotSubstring("", "", ::std::wstring(L"needle"), L"two needles"));
+ EXPECT_TRUE(IsNotSubstring("", "", L"needle", ::std::wstring(L"haystack")));
+}
+
+#endif // GTEST_HAS_STD_WSTRING
+
+// Tests floating-point assertions.
+
+template <typename RawType>
+class FloatingPointTest : public Test {
+ protected:
+
+ // Pre-calculated numbers to be used by the tests.
+ struct TestValues {
+ RawType close_to_positive_zero;
+ RawType close_to_negative_zero;
+ RawType further_from_negative_zero;
+
+ RawType close_to_one;
+ RawType further_from_one;
+
+ RawType infinity;
+ RawType close_to_infinity;
+ RawType further_from_infinity;
+
+ RawType nan1;
+ RawType nan2;
+ };
+
+ typedef typename testing::internal::FloatingPoint<RawType> Floating;
+ typedef typename Floating::Bits Bits;
+
+ virtual void SetUp() {
+ const size_t max_ulps = Floating::kMaxUlps;
+
+ // The bits that represent 0.0.
+ const Bits zero_bits = Floating(0).bits();
+
+ // Makes some numbers close to 0.0.
+ values_.close_to_positive_zero = Floating::ReinterpretBits(
+ zero_bits + max_ulps/2);
+ values_.close_to_negative_zero = -Floating::ReinterpretBits(
+ zero_bits + max_ulps - max_ulps/2);
+ values_.further_from_negative_zero = -Floating::ReinterpretBits(
+ zero_bits + max_ulps + 1 - max_ulps/2);
+
+ // The bits that represent 1.0.
+ const Bits one_bits = Floating(1).bits();
+
+ // Makes some numbers close to 1.0.
+ values_.close_to_one = Floating::ReinterpretBits(one_bits + max_ulps);
+ values_.further_from_one = Floating::ReinterpretBits(
+ one_bits + max_ulps + 1);
+
+ // +infinity.
+ values_.infinity = Floating::Infinity();
+
+ // The bits that represent +infinity.
+ const Bits infinity_bits = Floating(values_.infinity).bits();
+
+ // Makes some numbers close to infinity.
+ values_.close_to_infinity = Floating::ReinterpretBits(
+ infinity_bits - max_ulps);
+ values_.further_from_infinity = Floating::ReinterpretBits(
+ infinity_bits - max_ulps - 1);
+
+ // Makes some NAN's. Sets the most significant bit of the fraction so that
+ // our NaN's are quiet; trying to process a signaling NaN would raise an
+ // exception if our environment enables floating point exceptions.
+ values_.nan1 = Floating::ReinterpretBits(Floating::kExponentBitMask
+ | (static_cast<Bits>(1) << (Floating::kFractionBitCount - 1)) | 1);
+ values_.nan2 = Floating::ReinterpretBits(Floating::kExponentBitMask
+ | (static_cast<Bits>(1) << (Floating::kFractionBitCount - 1)) | 200);
+ }
+
+ void TestSize() {
+ EXPECT_EQ(sizeof(RawType), sizeof(Bits));
+ }
+
+ static TestValues values_;
+};
+
+template <typename RawType>
+typename FloatingPointTest<RawType>::TestValues
+ FloatingPointTest<RawType>::values_;
+
+// Instantiates FloatingPointTest for testing *_FLOAT_EQ.
+typedef FloatingPointTest<float> FloatTest;
+
+// Tests that the size of Float::Bits matches the size of float.
+TEST_F(FloatTest, Size) {
+ TestSize();
+}
+
+// Tests comparing with +0 and -0.
+TEST_F(FloatTest, Zeros) {
+ EXPECT_FLOAT_EQ(0.0, -0.0);
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(-0.0, 1.0),
+ "1.0");
+ EXPECT_FATAL_FAILURE(ASSERT_FLOAT_EQ(0.0, 1.5),
+ "1.5");
+}
+
+// Tests comparing numbers close to 0.
+//
+// This ensures that *_FLOAT_EQ handles the sign correctly and no
+// overflow occurs when comparing numbers whose absolute value is very
+// small.
+TEST_F(FloatTest, AlmostZeros) {
+ // In C++Builder, names within local classes (such as used by
+ // EXPECT_FATAL_FAILURE) cannot be resolved against static members of the
+ // scoping class. Use a static local alias as a workaround.
+ // We use the assignment syntax since some compilers, like Sun Studio,
+ // don't allow initializing references using construction syntax
+ // (parentheses).
+ static const FloatTest::TestValues& v = this->values_;
+
+ EXPECT_FLOAT_EQ(0.0, v.close_to_positive_zero);
+ EXPECT_FLOAT_EQ(-0.0, v.close_to_negative_zero);
+ EXPECT_FLOAT_EQ(v.close_to_positive_zero, v.close_to_negative_zero);
+
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_FLOAT_EQ(v.close_to_positive_zero,
+ v.further_from_negative_zero);
+ }, "v.further_from_negative_zero");
+}
+
+// Tests comparing numbers close to each other.
+TEST_F(FloatTest, SmallDiff) {
+ EXPECT_FLOAT_EQ(1.0, values_.close_to_one);
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(1.0, values_.further_from_one),
+ "values_.further_from_one");
+}
+
+// Tests comparing numbers far apart.
+TEST_F(FloatTest, LargeDiff) {
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(2.5, 3.0),
+ "3.0");
+}
+
+// Tests comparing with infinity.
+//
+// This ensures that no overflow occurs when comparing numbers whose
+// absolute value is very large.
+TEST_F(FloatTest, Infinity) {
+ EXPECT_FLOAT_EQ(values_.infinity, values_.close_to_infinity);
+ EXPECT_FLOAT_EQ(-values_.infinity, -values_.close_to_infinity);
+#if !GTEST_OS_SYMBIAN
+ // Nokia's STLport crashes if we try to output infinity or NaN.
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(values_.infinity, -values_.infinity),
+ "-values_.infinity");
+
+ // This is interesting as the representations of infinity and nan1
+ // are only 1 DLP apart.
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(values_.infinity, values_.nan1),
+ "values_.nan1");
+#endif // !GTEST_OS_SYMBIAN
+}
+
+// Tests that comparing with NAN always returns false.
+TEST_F(FloatTest, NaN) {
+#if !GTEST_OS_SYMBIAN
+// Nokia's STLport crashes if we try to output infinity or NaN.
+
+ // In C++Builder, names within local classes (such as used by
+ // EXPECT_FATAL_FAILURE) cannot be resolved against static members of the
+ // scoping class. Use a static local alias as a workaround.
+ // We use the assignment syntax since some compilers, like Sun Studio,
+ // don't allow initializing references using construction syntax
+ // (parentheses).
+ static const FloatTest::TestValues& v = this->values_;
+
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(v.nan1, v.nan1),
+ "v.nan1");
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(v.nan1, v.nan2),
+ "v.nan2");
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(1.0, v.nan1),
+ "v.nan1");
+
+ EXPECT_FATAL_FAILURE(ASSERT_FLOAT_EQ(v.nan1, v.infinity),
+ "v.infinity");
+#endif // !GTEST_OS_SYMBIAN
+}
+
+// Tests that *_FLOAT_EQ are reflexive.
+TEST_F(FloatTest, Reflexive) {
+ EXPECT_FLOAT_EQ(0.0, 0.0);
+ EXPECT_FLOAT_EQ(1.0, 1.0);
+ ASSERT_FLOAT_EQ(values_.infinity, values_.infinity);
+}
+
+// Tests that *_FLOAT_EQ are commutative.
+TEST_F(FloatTest, Commutative) {
+ // We already tested EXPECT_FLOAT_EQ(1.0, values_.close_to_one).
+ EXPECT_FLOAT_EQ(values_.close_to_one, 1.0);
+
+ // We already tested EXPECT_FLOAT_EQ(1.0, values_.further_from_one).
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(values_.further_from_one, 1.0),
+ "1.0");
+}
+
+// Tests EXPECT_NEAR.
+TEST_F(FloatTest, EXPECT_NEAR) {
+ EXPECT_NEAR(-1.0f, -1.1f, 0.2f);
+ EXPECT_NEAR(2.0f, 3.0f, 1.0f);
+ EXPECT_NONFATAL_FAILURE(EXPECT_NEAR(1.0f,1.5f, 0.25f), // NOLINT
+ "The difference between 1.0f and 1.5f is 0.5, "
+ "which exceeds 0.25f");
+ // To work around a bug in gcc 2.95.0, there is intentionally no
+ // space after the first comma in the previous line.
+}
+
+// Tests ASSERT_NEAR.
+TEST_F(FloatTest, ASSERT_NEAR) {
+ ASSERT_NEAR(-1.0f, -1.1f, 0.2f);
+ ASSERT_NEAR(2.0f, 3.0f, 1.0f);
+ EXPECT_FATAL_FAILURE(ASSERT_NEAR(1.0f,1.5f, 0.25f), // NOLINT
+ "The difference between 1.0f and 1.5f is 0.5, "
+ "which exceeds 0.25f");
+ // To work around a bug in gcc 2.95.0, there is intentionally no
+ // space after the first comma in the previous line.
+}
+
+// Tests the cases where FloatLE() should succeed.
+TEST_F(FloatTest, FloatLESucceeds) {
+ EXPECT_PRED_FORMAT2(FloatLE, 1.0f, 2.0f); // When val1 < val2,
+ ASSERT_PRED_FORMAT2(FloatLE, 1.0f, 1.0f); // val1 == val2,
+
+ // or when val1 is greater than, but almost equals to, val2.
+ EXPECT_PRED_FORMAT2(FloatLE, values_.close_to_positive_zero, 0.0f);
+}
+
+// Tests the cases where FloatLE() should fail.
+TEST_F(FloatTest, FloatLEFails) {
+ // When val1 is greater than val2 by a large margin,
+ EXPECT_NONFATAL_FAILURE(EXPECT_PRED_FORMAT2(FloatLE, 2.0f, 1.0f),
+ "(2.0f) <= (1.0f)");
+
+ // or by a small yet non-negligible margin,
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(FloatLE, values_.further_from_one, 1.0f);
+ }, "(values_.further_from_one) <= (1.0f)");
+
+#if !GTEST_OS_SYMBIAN && !defined(__BORLANDC__)
+ // Nokia's STLport crashes if we try to output infinity or NaN.
+ // C++Builder gives bad results for ordered comparisons involving NaNs
+ // due to compiler bugs.
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(FloatLE, values_.nan1, values_.infinity);
+ }, "(values_.nan1) <= (values_.infinity)");
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(FloatLE, -values_.infinity, values_.nan1);
+ }, "(-values_.infinity) <= (values_.nan1)");
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT2(FloatLE, values_.nan1, values_.nan1);
+ }, "(values_.nan1) <= (values_.nan1)");
+#endif // !GTEST_OS_SYMBIAN && !defined(__BORLANDC__)
+}
+
+// Instantiates FloatingPointTest for testing *_DOUBLE_EQ.
+typedef FloatingPointTest<double> DoubleTest;
+
+// Tests that the size of Double::Bits matches the size of double.
+TEST_F(DoubleTest, Size) {
+ TestSize();
+}
+
+// Tests comparing with +0 and -0.
+TEST_F(DoubleTest, Zeros) {
+ EXPECT_DOUBLE_EQ(0.0, -0.0);
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(-0.0, 1.0),
+ "1.0");
+ EXPECT_FATAL_FAILURE(ASSERT_DOUBLE_EQ(0.0, 1.0),
+ "1.0");
+}
+
+// Tests comparing numbers close to 0.
+//
+// This ensures that *_DOUBLE_EQ handles the sign correctly and no
+// overflow occurs when comparing numbers whose absolute value is very
+// small.
+TEST_F(DoubleTest, AlmostZeros) {
+ // In C++Builder, names within local classes (such as used by
+ // EXPECT_FATAL_FAILURE) cannot be resolved against static members of the
+ // scoping class. Use a static local alias as a workaround.
+ // We use the assignment syntax since some compilers, like Sun Studio,
+ // don't allow initializing references using construction syntax
+ // (parentheses).
+ static const DoubleTest::TestValues& v = this->values_;
+
+ EXPECT_DOUBLE_EQ(0.0, v.close_to_positive_zero);
+ EXPECT_DOUBLE_EQ(-0.0, v.close_to_negative_zero);
+ EXPECT_DOUBLE_EQ(v.close_to_positive_zero, v.close_to_negative_zero);
+
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_DOUBLE_EQ(v.close_to_positive_zero,
+ v.further_from_negative_zero);
+ }, "v.further_from_negative_zero");
+}
+
+// Tests comparing numbers close to each other.
+TEST_F(DoubleTest, SmallDiff) {
+ EXPECT_DOUBLE_EQ(1.0, values_.close_to_one);
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(1.0, values_.further_from_one),
+ "values_.further_from_one");
+}
+
+// Tests comparing numbers far apart.
+TEST_F(DoubleTest, LargeDiff) {
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(2.0, 3.0),
+ "3.0");
+}
+
+// Tests comparing with infinity.
+//
+// This ensures that no overflow occurs when comparing numbers whose
+// absolute value is very large.
+TEST_F(DoubleTest, Infinity) {
+ EXPECT_DOUBLE_EQ(values_.infinity, values_.close_to_infinity);
+ EXPECT_DOUBLE_EQ(-values_.infinity, -values_.close_to_infinity);
+#if !GTEST_OS_SYMBIAN
+ // Nokia's STLport crashes if we try to output infinity or NaN.
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(values_.infinity, -values_.infinity),
+ "-values_.infinity");
+
+ // This is interesting as the representations of infinity_ and nan1_
+ // are only 1 DLP apart.
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(values_.infinity, values_.nan1),
+ "values_.nan1");
+#endif // !GTEST_OS_SYMBIAN
+}
+
+// Tests that comparing with NAN always returns false.
+TEST_F(DoubleTest, NaN) {
+#if !GTEST_OS_SYMBIAN
+ // In C++Builder, names within local classes (such as used by
+ // EXPECT_FATAL_FAILURE) cannot be resolved against static members of the
+ // scoping class. Use a static local alias as a workaround.
+ // We use the assignment syntax since some compilers, like Sun Studio,
+ // don't allow initializing references using construction syntax
+ // (parentheses).
+ static const DoubleTest::TestValues& v = this->values_;
+
+ // Nokia's STLport crashes if we try to output infinity or NaN.
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(v.nan1, v.nan1),
+ "v.nan1");
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(v.nan1, v.nan2), "v.nan2");
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(1.0, v.nan1), "v.nan1");
+ EXPECT_FATAL_FAILURE(ASSERT_DOUBLE_EQ(v.nan1, v.infinity),
+ "v.infinity");
+#endif // !GTEST_OS_SYMBIAN
+}
+
+// Tests that *_DOUBLE_EQ are reflexive.
+TEST_F(DoubleTest, Reflexive) {
+ EXPECT_DOUBLE_EQ(0.0, 0.0);
+ EXPECT_DOUBLE_EQ(1.0, 1.0);
+#if !GTEST_OS_SYMBIAN
+ // Nokia's STLport crashes if we try to output infinity or NaN.
+ ASSERT_DOUBLE_EQ(values_.infinity, values_.infinity);
+#endif // !GTEST_OS_SYMBIAN
+}
+
+// Tests that *_DOUBLE_EQ are commutative.
+TEST_F(DoubleTest, Commutative) {
+ // We already tested EXPECT_DOUBLE_EQ(1.0, values_.close_to_one).
+ EXPECT_DOUBLE_EQ(values_.close_to_one, 1.0);
+
+ // We already tested EXPECT_DOUBLE_EQ(1.0, values_.further_from_one).
+ EXPECT_NONFATAL_FAILURE(EXPECT_DOUBLE_EQ(values_.further_from_one, 1.0),
+ "1.0");
+}
+
+// Tests EXPECT_NEAR.
+TEST_F(DoubleTest, EXPECT_NEAR) {
+ EXPECT_NEAR(-1.0, -1.1, 0.2);
+ EXPECT_NEAR(2.0, 3.0, 1.0);
+ EXPECT_NONFATAL_FAILURE(EXPECT_NEAR(1.0, 1.5, 0.25), // NOLINT
+ "The difference between 1.0 and 1.5 is 0.5, "
+ "which exceeds 0.25");
+ // To work around a bug in gcc 2.95.0, there is intentionally no
+ // space after the first comma in the previous statement.
+}
+
+// Tests ASSERT_NEAR.
+TEST_F(DoubleTest, ASSERT_NEAR) {
+ ASSERT_NEAR(-1.0, -1.1, 0.2);
+ ASSERT_NEAR(2.0, 3.0, 1.0);
+ EXPECT_FATAL_FAILURE(ASSERT_NEAR(1.0, 1.5, 0.25), // NOLINT
+ "The difference between 1.0 and 1.5 is 0.5, "
+ "which exceeds 0.25");
+ // To work around a bug in gcc 2.95.0, there is intentionally no
+ // space after the first comma in the previous statement.
+}
+
+// Tests the cases where DoubleLE() should succeed.
+TEST_F(DoubleTest, DoubleLESucceeds) {
+ EXPECT_PRED_FORMAT2(DoubleLE, 1.0, 2.0); // When val1 < val2,
+ ASSERT_PRED_FORMAT2(DoubleLE, 1.0, 1.0); // val1 == val2,
+
+ // or when val1 is greater than, but almost equals to, val2.
+ EXPECT_PRED_FORMAT2(DoubleLE, values_.close_to_positive_zero, 0.0);
+}
+
+// Tests the cases where DoubleLE() should fail.
+TEST_F(DoubleTest, DoubleLEFails) {
+ // When val1 is greater than val2 by a large margin,
+ EXPECT_NONFATAL_FAILURE(EXPECT_PRED_FORMAT2(DoubleLE, 2.0, 1.0),
+ "(2.0) <= (1.0)");
+
+ // or by a small yet non-negligible margin,
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(DoubleLE, values_.further_from_one, 1.0);
+ }, "(values_.further_from_one) <= (1.0)");
+
+#if !GTEST_OS_SYMBIAN && !defined(__BORLANDC__)
+ // Nokia's STLport crashes if we try to output infinity or NaN.
+ // C++Builder gives bad results for ordered comparisons involving NaNs
+ // due to compiler bugs.
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(DoubleLE, values_.nan1, values_.infinity);
+ }, "(values_.nan1) <= (values_.infinity)");
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_PRED_FORMAT2(DoubleLE, -values_.infinity, values_.nan1);
+ }, " (-values_.infinity) <= (values_.nan1)");
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_PRED_FORMAT2(DoubleLE, values_.nan1, values_.nan1);
+ }, "(values_.nan1) <= (values_.nan1)");
+#endif // !GTEST_OS_SYMBIAN && !defined(__BORLANDC__)
+}
+
+
+// Verifies that a test or test case whose name starts with DISABLED_ is
+// not run.
+
+// A test whose name starts with DISABLED_.
+// Should not run.
+TEST(DisabledTest, DISABLED_TestShouldNotRun) {
+ FAIL() << "Unexpected failure: Disabled test should not be run.";
+}
+
+// A test whose name does not start with DISABLED_.
+// Should run.
+TEST(DisabledTest, NotDISABLED_TestShouldRun) {
+ EXPECT_EQ(1, 1);
+}
+
+// A test case whose name starts with DISABLED_.
+// Should not run.
+TEST(DISABLED_TestCase, TestShouldNotRun) {
+ FAIL() << "Unexpected failure: Test in disabled test case should not be run.";
+}
+
+// A test case and test whose names start with DISABLED_.
+// Should not run.
+TEST(DISABLED_TestCase, DISABLED_TestShouldNotRun) {
+ FAIL() << "Unexpected failure: Test in disabled test case should not be run.";
+}
+
+// Check that when all tests in a test case are disabled, SetupTestCase() and
+// TearDownTestCase() are not called.
+class DisabledTestsTest : public Test {
+ protected:
+ static void SetUpTestCase() {
+ FAIL() << "Unexpected failure: All tests disabled in test case. "
+ "SetupTestCase() should not be called.";
+ }
+
+ static void TearDownTestCase() {
+ FAIL() << "Unexpected failure: All tests disabled in test case. "
+ "TearDownTestCase() should not be called.";
+ }
+};
+
+TEST_F(DisabledTestsTest, DISABLED_TestShouldNotRun_1) {
+ FAIL() << "Unexpected failure: Disabled test should not be run.";
+}
+
+TEST_F(DisabledTestsTest, DISABLED_TestShouldNotRun_2) {
+ FAIL() << "Unexpected failure: Disabled test should not be run.";
+}
+
+// Tests that disabled typed tests aren't run.
+
+#if GTEST_HAS_TYPED_TEST
+
+template <typename T>
+class TypedTest : public Test {
+};
+
+typedef testing::Types<int, double> NumericTypes;
+TYPED_TEST_CASE(TypedTest, NumericTypes);
+
+TYPED_TEST(TypedTest, DISABLED_ShouldNotRun) {
+ FAIL() << "Unexpected failure: Disabled typed test should not run.";
+}
+
+template <typename T>
+class DISABLED_TypedTest : public Test {
+};
+
+TYPED_TEST_CASE(DISABLED_TypedTest, NumericTypes);
+
+TYPED_TEST(DISABLED_TypedTest, ShouldNotRun) {
+ FAIL() << "Unexpected failure: Disabled typed test should not run.";
+}
+
+#endif // GTEST_HAS_TYPED_TEST
+
+// Tests that disabled type-parameterized tests aren't run.
+
+#if GTEST_HAS_TYPED_TEST_P
+
+template <typename T>
+class TypedTestP : public Test {
+};
+
+TYPED_TEST_CASE_P(TypedTestP);
+
+TYPED_TEST_P(TypedTestP, DISABLED_ShouldNotRun) {
+ FAIL() << "Unexpected failure: "
+ << "Disabled type-parameterized test should not run.";
+}
+
+REGISTER_TYPED_TEST_CASE_P(TypedTestP, DISABLED_ShouldNotRun);
+
+INSTANTIATE_TYPED_TEST_CASE_P(My, TypedTestP, NumericTypes);
+
+template <typename T>
+class DISABLED_TypedTestP : public Test {
+};
+
+TYPED_TEST_CASE_P(DISABLED_TypedTestP);
+
+TYPED_TEST_P(DISABLED_TypedTestP, ShouldNotRun) {
+ FAIL() << "Unexpected failure: "
+ << "Disabled type-parameterized test should not run.";
+}
+
+REGISTER_TYPED_TEST_CASE_P(DISABLED_TypedTestP, ShouldNotRun);
+
+INSTANTIATE_TYPED_TEST_CASE_P(My, DISABLED_TypedTestP, NumericTypes);
+
+#endif // GTEST_HAS_TYPED_TEST_P
+
+// Tests that assertion macros evaluate their arguments exactly once.
+
+class SingleEvaluationTest : public Test {
+ public: // Must be public and not protected due to a bug in g++ 3.4.2.
+ // This helper function is needed by the FailedASSERT_STREQ test
+ // below. It's public to work around C++Builder's bug with scoping local
+ // classes.
+ static void CompareAndIncrementCharPtrs() {
+ ASSERT_STREQ(p1_++, p2_++);
+ }
+
+ // This helper function is needed by the FailedASSERT_NE test below. It's
+ // public to work around C++Builder's bug with scoping local classes.
+ static void CompareAndIncrementInts() {
+ ASSERT_NE(a_++, b_++);
+ }
+
+ protected:
+ SingleEvaluationTest() {
+ p1_ = s1_;
+ p2_ = s2_;
+ a_ = 0;
+ b_ = 0;
+ }
+
+ static const char* const s1_;
+ static const char* const s2_;
+ static const char* p1_;
+ static const char* p2_;
+
+ static int a_;
+ static int b_;
+};
+
+const char* const SingleEvaluationTest::s1_ = "01234";
+const char* const SingleEvaluationTest::s2_ = "abcde";
+const char* SingleEvaluationTest::p1_;
+const char* SingleEvaluationTest::p2_;
+int SingleEvaluationTest::a_;
+int SingleEvaluationTest::b_;
+
+// Tests that when ASSERT_STREQ fails, it evaluates its arguments
+// exactly once.
+TEST_F(SingleEvaluationTest, FailedASSERT_STREQ) {
+ EXPECT_FATAL_FAILURE(SingleEvaluationTest::CompareAndIncrementCharPtrs(),
+ "p2_++");
+ EXPECT_EQ(s1_ + 1, p1_);
+ EXPECT_EQ(s2_ + 1, p2_);
+}
+
+// Tests that string assertion arguments are evaluated exactly once.
+TEST_F(SingleEvaluationTest, ASSERT_STR) {
+ // successful EXPECT_STRNE
+ EXPECT_STRNE(p1_++, p2_++);
+ EXPECT_EQ(s1_ + 1, p1_);
+ EXPECT_EQ(s2_ + 1, p2_);
+
+ // failed EXPECT_STRCASEEQ
+ EXPECT_NONFATAL_FAILURE(EXPECT_STRCASEEQ(p1_++, p2_++),
+ "ignoring case");
+ EXPECT_EQ(s1_ + 2, p1_);
+ EXPECT_EQ(s2_ + 2, p2_);
+}
+
+// Tests that when ASSERT_NE fails, it evaluates its arguments exactly
+// once.
+TEST_F(SingleEvaluationTest, FailedASSERT_NE) {
+ EXPECT_FATAL_FAILURE(SingleEvaluationTest::CompareAndIncrementInts(),
+ "(a_++) != (b_++)");
+ EXPECT_EQ(1, a_);
+ EXPECT_EQ(1, b_);
+}
+
+// Tests that assertion arguments are evaluated exactly once.
+TEST_F(SingleEvaluationTest, OtherCases) {
+ // successful EXPECT_TRUE
+ EXPECT_TRUE(0 == a_++); // NOLINT
+ EXPECT_EQ(1, a_);
+
+ // failed EXPECT_TRUE
+ EXPECT_NONFATAL_FAILURE(EXPECT_TRUE(-1 == a_++), "-1 == a_++");
+ EXPECT_EQ(2, a_);
+
+ // successful EXPECT_GT
+ EXPECT_GT(a_++, b_++);
+ EXPECT_EQ(3, a_);
+ EXPECT_EQ(1, b_);
+
+ // failed EXPECT_LT
+ EXPECT_NONFATAL_FAILURE(EXPECT_LT(a_++, b_++), "(a_++) < (b_++)");
+ EXPECT_EQ(4, a_);
+ EXPECT_EQ(2, b_);
+
+ // successful ASSERT_TRUE
+ ASSERT_TRUE(0 < a_++); // NOLINT
+ EXPECT_EQ(5, a_);
+
+ // successful ASSERT_GT
+ ASSERT_GT(a_++, b_++);
+ EXPECT_EQ(6, a_);
+ EXPECT_EQ(3, b_);
+}
+
+#if GTEST_HAS_EXCEPTIONS
+
+void ThrowAnInteger() {
+ throw 1;
+}
+
+// Tests that assertion arguments are evaluated exactly once.
+TEST_F(SingleEvaluationTest, ExceptionTests) {
+ // successful EXPECT_THROW
+ EXPECT_THROW({ // NOLINT
+ a_++;
+ ThrowAnInteger();
+ }, int);
+ EXPECT_EQ(1, a_);
+
+ // failed EXPECT_THROW, throws different
+ EXPECT_NONFATAL_FAILURE(EXPECT_THROW({ // NOLINT
+ a_++;
+ ThrowAnInteger();
+ }, bool), "throws a different type");
+ EXPECT_EQ(2, a_);
+
+ // failed EXPECT_THROW, throws nothing
+ EXPECT_NONFATAL_FAILURE(EXPECT_THROW(a_++, bool), "throws nothing");
+ EXPECT_EQ(3, a_);
+
+ // successful EXPECT_NO_THROW
+ EXPECT_NO_THROW(a_++);
+ EXPECT_EQ(4, a_);
+
+ // failed EXPECT_NO_THROW
+ EXPECT_NONFATAL_FAILURE(EXPECT_NO_THROW({ // NOLINT
+ a_++;
+ ThrowAnInteger();
+ }), "it throws");
+ EXPECT_EQ(5, a_);
+
+ // successful EXPECT_ANY_THROW
+ EXPECT_ANY_THROW({ // NOLINT
+ a_++;
+ ThrowAnInteger();
+ });
+ EXPECT_EQ(6, a_);
+
+ // failed EXPECT_ANY_THROW
+ EXPECT_NONFATAL_FAILURE(EXPECT_ANY_THROW(a_++), "it doesn't");
+ EXPECT_EQ(7, a_);
+}
+
+#endif // GTEST_HAS_EXCEPTIONS
+
+// Tests {ASSERT|EXPECT}_NO_FATAL_FAILURE.
+class NoFatalFailureTest : public Test {
+ protected:
+ void Succeeds() {}
+ void FailsNonFatal() {
+ ADD_FAILURE() << "some non-fatal failure";
+ }
+ void Fails() {
+ FAIL() << "some fatal failure";
+ }
+
+ void DoAssertNoFatalFailureOnFails() {
+ ASSERT_NO_FATAL_FAILURE(Fails());
+ ADD_FAILURE() << "shold not reach here.";
+ }
+
+ void DoExpectNoFatalFailureOnFails() {
+ EXPECT_NO_FATAL_FAILURE(Fails());
+ ADD_FAILURE() << "other failure";
+ }
+};
+
+TEST_F(NoFatalFailureTest, NoFailure) {
+ EXPECT_NO_FATAL_FAILURE(Succeeds());
+ ASSERT_NO_FATAL_FAILURE(Succeeds());
+}
+
+TEST_F(NoFatalFailureTest, NonFatalIsNoFailure) {
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_NO_FATAL_FAILURE(FailsNonFatal()),
+ "some non-fatal failure");
+ EXPECT_NONFATAL_FAILURE(
+ ASSERT_NO_FATAL_FAILURE(FailsNonFatal()),
+ "some non-fatal failure");
+}
+
+TEST_F(NoFatalFailureTest, AssertNoFatalFailureOnFatalFailure) {
+ TestPartResultArray gtest_failures;
+ {
+ ScopedFakeTestPartResultReporter gtest_reporter(>est_failures);
+ DoAssertNoFatalFailureOnFails();
+ }
+ ASSERT_EQ(2, gtest_failures.size());
+ EXPECT_EQ(TestPartResult::kFatalFailure,
+ gtest_failures.GetTestPartResult(0).type());
+ EXPECT_EQ(TestPartResult::kFatalFailure,
+ gtest_failures.GetTestPartResult(1).type());
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "some fatal failure",
+ gtest_failures.GetTestPartResult(0).message());
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "it does",
+ gtest_failures.GetTestPartResult(1).message());
+}
+
+TEST_F(NoFatalFailureTest, ExpectNoFatalFailureOnFatalFailure) {
+ TestPartResultArray gtest_failures;
+ {
+ ScopedFakeTestPartResultReporter gtest_reporter(>est_failures);
+ DoExpectNoFatalFailureOnFails();
+ }
+ ASSERT_EQ(3, gtest_failures.size());
+ EXPECT_EQ(TestPartResult::kFatalFailure,
+ gtest_failures.GetTestPartResult(0).type());
+ EXPECT_EQ(TestPartResult::kNonFatalFailure,
+ gtest_failures.GetTestPartResult(1).type());
+ EXPECT_EQ(TestPartResult::kNonFatalFailure,
+ gtest_failures.GetTestPartResult(2).type());
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "some fatal failure",
+ gtest_failures.GetTestPartResult(0).message());
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "it does",
+ gtest_failures.GetTestPartResult(1).message());
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "other failure",
+ gtest_failures.GetTestPartResult(2).message());
+}
+
+TEST_F(NoFatalFailureTest, MessageIsStreamable) {
+ TestPartResultArray gtest_failures;
+ {
+ ScopedFakeTestPartResultReporter gtest_reporter(>est_failures);
+ EXPECT_NO_FATAL_FAILURE(FAIL() << "foo") << "my message";
+ }
+ ASSERT_EQ(2, gtest_failures.size());
+ EXPECT_EQ(TestPartResult::kNonFatalFailure,
+ gtest_failures.GetTestPartResult(0).type());
+ EXPECT_EQ(TestPartResult::kNonFatalFailure,
+ gtest_failures.GetTestPartResult(1).type());
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "foo",
+ gtest_failures.GetTestPartResult(0).message());
+ EXPECT_PRED_FORMAT2(testing::IsSubstring, "my message",
+ gtest_failures.GetTestPartResult(1).message());
+}
+
+// Tests non-string assertions.
+
+// Tests EqFailure(), used for implementing *EQ* assertions.
+TEST(AssertionTest, EqFailure) {
+ const String foo_val("5"), bar_val("6");
+ const String msg1(
+ EqFailure("foo", "bar", foo_val, bar_val, false)
+ .failure_message());
+ EXPECT_STREQ(
+ "Value of: bar\n"
+ " Actual: 6\n"
+ "Expected: foo\n"
+ "Which is: 5",
+ msg1.c_str());
+
+ const String msg2(
+ EqFailure("foo", "6", foo_val, bar_val, false)
+ .failure_message());
+ EXPECT_STREQ(
+ "Value of: 6\n"
+ "Expected: foo\n"
+ "Which is: 5",
+ msg2.c_str());
+
+ const String msg3(
+ EqFailure("5", "bar", foo_val, bar_val, false)
+ .failure_message());
+ EXPECT_STREQ(
+ "Value of: bar\n"
+ " Actual: 6\n"
+ "Expected: 5",
+ msg3.c_str());
+
+ const String msg4(
+ EqFailure("5", "6", foo_val, bar_val, false).failure_message());
+ EXPECT_STREQ(
+ "Value of: 6\n"
+ "Expected: 5",
+ msg4.c_str());
+
+ const String msg5(
+ EqFailure("foo", "bar",
+ String("\"x\""), String("\"y\""),
+ true).failure_message());
+ EXPECT_STREQ(
+ "Value of: bar\n"
+ " Actual: \"y\"\n"
+ "Expected: foo (ignoring case)\n"
+ "Which is: \"x\"",
+ msg5.c_str());
+}
+
+// Tests AppendUserMessage(), used for implementing the *EQ* macros.
+TEST(AssertionTest, AppendUserMessage) {
+ const String foo("foo");
+
+ Message msg;
+ EXPECT_STREQ("foo",
+ AppendUserMessage(foo, msg).c_str());
+
+ msg << "bar";
+ EXPECT_STREQ("foo\nbar",
+ AppendUserMessage(foo, msg).c_str());
+}
+
+#ifdef __BORLANDC__
+// Silences warnings: "Condition is always true", "Unreachable code"
+# pragma option push -w-ccc -w-rch
+#endif
+
+// Tests ASSERT_TRUE.
+TEST(AssertionTest, ASSERT_TRUE) {
+ ASSERT_TRUE(2 > 1); // NOLINT
+ EXPECT_FATAL_FAILURE(ASSERT_TRUE(2 < 1),
+ "2 < 1");
+}
+
+// Tests ASSERT_TRUE(predicate) for predicates returning AssertionResult.
+TEST(AssertionTest, AssertTrueWithAssertionResult) {
+ ASSERT_TRUE(ResultIsEven(2));
+#ifndef __BORLANDC__
+ // ICE's in C++Builder.
+ EXPECT_FATAL_FAILURE(ASSERT_TRUE(ResultIsEven(3)),
+ "Value of: ResultIsEven(3)\n"
+ " Actual: false (3 is odd)\n"
+ "Expected: true");
+#endif
+ ASSERT_TRUE(ResultIsEvenNoExplanation(2));
+ EXPECT_FATAL_FAILURE(ASSERT_TRUE(ResultIsEvenNoExplanation(3)),
+ "Value of: ResultIsEvenNoExplanation(3)\n"
+ " Actual: false (3 is odd)\n"
+ "Expected: true");
+}
+
+// Tests ASSERT_FALSE.
+TEST(AssertionTest, ASSERT_FALSE) {
+ ASSERT_FALSE(2 < 1); // NOLINT
+ EXPECT_FATAL_FAILURE(ASSERT_FALSE(2 > 1),
+ "Value of: 2 > 1\n"
+ " Actual: true\n"
+ "Expected: false");
+}
+
+// Tests ASSERT_FALSE(predicate) for predicates returning AssertionResult.
+TEST(AssertionTest, AssertFalseWithAssertionResult) {
+ ASSERT_FALSE(ResultIsEven(3));
+#ifndef __BORLANDC__
+ // ICE's in C++Builder.
+ EXPECT_FATAL_FAILURE(ASSERT_FALSE(ResultIsEven(2)),
+ "Value of: ResultIsEven(2)\n"
+ " Actual: true (2 is even)\n"
+ "Expected: false");
+#endif
+ ASSERT_FALSE(ResultIsEvenNoExplanation(3));
+ EXPECT_FATAL_FAILURE(ASSERT_FALSE(ResultIsEvenNoExplanation(2)),
+ "Value of: ResultIsEvenNoExplanation(2)\n"
+ " Actual: true\n"
+ "Expected: false");
+}
+
+#ifdef __BORLANDC__
+// Restores warnings after previous "#pragma option push" supressed them
+# pragma option pop
+#endif
+
+// Tests using ASSERT_EQ on double values. The purpose is to make
+// sure that the specialization we did for integer and anonymous enums
+// isn't used for double arguments.
+TEST(ExpectTest, ASSERT_EQ_Double) {
+ // A success.
+ ASSERT_EQ(5.6, 5.6);
+
+ // A failure.
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(5.1, 5.2),
+ "5.1");
+}
+
+// Tests ASSERT_EQ.
+TEST(AssertionTest, ASSERT_EQ) {
+ ASSERT_EQ(5, 2 + 3);
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(5, 2*3),
+ "Value of: 2*3\n"
+ " Actual: 6\n"
+ "Expected: 5");
+}
+
+// Tests ASSERT_EQ(NULL, pointer).
+#if GTEST_CAN_COMPARE_NULL
+TEST(AssertionTest, ASSERT_EQ_NULL) {
+ // A success.
+ const char* p = NULL;
+ // Some older GCC versions may issue a spurious waring in this or the next
+ // assertion statement. This warning should not be suppressed with
+ // static_cast since the test verifies the ability to use bare NULL as the
+ // expected parameter to the macro.
+ ASSERT_EQ(NULL, p);
+
+ // A failure.
+ static int n = 0;
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(NULL, &n),
+ "Value of: &n\n");
+}
+#endif // GTEST_CAN_COMPARE_NULL
+
+// Tests ASSERT_EQ(0, non_pointer). Since the literal 0 can be
+// treated as a null pointer by the compiler, we need to make sure
+// that ASSERT_EQ(0, non_pointer) isn't interpreted by Google Test as
+// ASSERT_EQ(static_cast<void*>(NULL), non_pointer).
+TEST(ExpectTest, ASSERT_EQ_0) {
+ int n = 0;
+
+ // A success.
+ ASSERT_EQ(0, n);
+
+ // A failure.
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(0, 5.6),
+ "Expected: 0");
+}
+
+// Tests ASSERT_NE.
+TEST(AssertionTest, ASSERT_NE) {
+ ASSERT_NE(6, 7);
+ EXPECT_FATAL_FAILURE(ASSERT_NE('a', 'a'),
+ "Expected: ('a') != ('a'), "
+ "actual: 'a' (97, 0x61) vs 'a' (97, 0x61)");
+}
+
+// Tests ASSERT_LE.
+TEST(AssertionTest, ASSERT_LE) {
+ ASSERT_LE(2, 3);
+ ASSERT_LE(2, 2);
+ EXPECT_FATAL_FAILURE(ASSERT_LE(2, 0),
+ "Expected: (2) <= (0), actual: 2 vs 0");
+}
+
+// Tests ASSERT_LT.
+TEST(AssertionTest, ASSERT_LT) {
+ ASSERT_LT(2, 3);
+ EXPECT_FATAL_FAILURE(ASSERT_LT(2, 2),
+ "Expected: (2) < (2), actual: 2 vs 2");
+}
+
+// Tests ASSERT_GE.
+TEST(AssertionTest, ASSERT_GE) {
+ ASSERT_GE(2, 1);
+ ASSERT_GE(2, 2);
+ EXPECT_FATAL_FAILURE(ASSERT_GE(2, 3),
+ "Expected: (2) >= (3), actual: 2 vs 3");
+}
+
+// Tests ASSERT_GT.
+TEST(AssertionTest, ASSERT_GT) {
+ ASSERT_GT(2, 1);
+ EXPECT_FATAL_FAILURE(ASSERT_GT(2, 2),
+ "Expected: (2) > (2), actual: 2 vs 2");
+}
+
+#if GTEST_HAS_EXCEPTIONS
+
+void ThrowNothing() {}
+
+// Tests ASSERT_THROW.
+TEST(AssertionTest, ASSERT_THROW) {
+ ASSERT_THROW(ThrowAnInteger(), int);
+
+# ifndef __BORLANDC__
+
+ // ICE's in C++Builder 2007 and 2009.
+ EXPECT_FATAL_FAILURE(
+ ASSERT_THROW(ThrowAnInteger(), bool),
+ "Expected: ThrowAnInteger() throws an exception of type bool.\n"
+ " Actual: it throws a different type.");
+# endif
+
+ EXPECT_FATAL_FAILURE(
+ ASSERT_THROW(ThrowNothing(), bool),
+ "Expected: ThrowNothing() throws an exception of type bool.\n"
+ " Actual: it throws nothing.");
+}
+
+// Tests ASSERT_NO_THROW.
+TEST(AssertionTest, ASSERT_NO_THROW) {
+ ASSERT_NO_THROW(ThrowNothing());
+ EXPECT_FATAL_FAILURE(ASSERT_NO_THROW(ThrowAnInteger()),
+ "Expected: ThrowAnInteger() doesn't throw an exception."
+ "\n Actual: it throws.");
+}
+
+// Tests ASSERT_ANY_THROW.
+TEST(AssertionTest, ASSERT_ANY_THROW) {
+ ASSERT_ANY_THROW(ThrowAnInteger());
+ EXPECT_FATAL_FAILURE(
+ ASSERT_ANY_THROW(ThrowNothing()),
+ "Expected: ThrowNothing() throws an exception.\n"
+ " Actual: it doesn't.");
+}
+
+#endif // GTEST_HAS_EXCEPTIONS
+
+// Makes sure we deal with the precedence of <<. This test should
+// compile.
+TEST(AssertionTest, AssertPrecedence) {
+ ASSERT_EQ(1 < 2, true);
+ bool false_value = false;
+ ASSERT_EQ(true && false_value, false);
+}
+
+// A subroutine used by the following test.
+void TestEq1(int x) {
+ ASSERT_EQ(1, x);
+}
+
+// Tests calling a test subroutine that's not part of a fixture.
+TEST(AssertionTest, NonFixtureSubroutine) {
+ EXPECT_FATAL_FAILURE(TestEq1(2),
+ "Value of: x");
+}
+
+// An uncopyable class.
+class Uncopyable {
+ public:
+ explicit Uncopyable(int a_value) : value_(a_value) {}
+
+ int value() const { return value_; }
+ bool operator==(const Uncopyable& rhs) const {
+ return value() == rhs.value();
+ }
+ private:
+ // This constructor deliberately has no implementation, as we don't
+ // want this class to be copyable.
+ Uncopyable(const Uncopyable&); // NOLINT
+
+ int value_;
+};
+
+::std::ostream& operator<<(::std::ostream& os, const Uncopyable& value) {
+ return os << value.value();
+}
+
+
+bool IsPositiveUncopyable(const Uncopyable& x) {
+ return x.value() > 0;
+}
+
+// A subroutine used by the following test.
+void TestAssertNonPositive() {
+ Uncopyable y(-1);
+ ASSERT_PRED1(IsPositiveUncopyable, y);
+}
+// A subroutine used by the following test.
+void TestAssertEqualsUncopyable() {
+ Uncopyable x(5);
+ Uncopyable y(-1);
+ ASSERT_EQ(x, y);
+}
+
+// Tests that uncopyable objects can be used in assertions.
+TEST(AssertionTest, AssertWorksWithUncopyableObject) {
+ Uncopyable x(5);
+ ASSERT_PRED1(IsPositiveUncopyable, x);
+ ASSERT_EQ(x, x);
+ EXPECT_FATAL_FAILURE(TestAssertNonPositive(),
+ "IsPositiveUncopyable(y) evaluates to false, where\ny evaluates to -1");
+ EXPECT_FATAL_FAILURE(TestAssertEqualsUncopyable(),
+ "Value of: y\n Actual: -1\nExpected: x\nWhich is: 5");
+}
+
+// Tests that uncopyable objects can be used in expects.
+TEST(AssertionTest, ExpectWorksWithUncopyableObject) {
+ Uncopyable x(5);
+ EXPECT_PRED1(IsPositiveUncopyable, x);
+ Uncopyable y(-1);
+ EXPECT_NONFATAL_FAILURE(EXPECT_PRED1(IsPositiveUncopyable, y),
+ "IsPositiveUncopyable(y) evaluates to false, where\ny evaluates to -1");
+ EXPECT_EQ(x, x);
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(x, y),
+ "Value of: y\n Actual: -1\nExpected: x\nWhich is: 5");
+}
+
+enum NamedEnum {
+ kE1 = 0,
+ kE2 = 1
+};
+
+TEST(AssertionTest, NamedEnum) {
+ EXPECT_EQ(kE1, kE1);
+ EXPECT_LT(kE1, kE2);
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(kE1, kE2), "Which is: 0");
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(kE1, kE2), "Actual: 1");
+}
+
+// The version of gcc used in XCode 2.2 has a bug and doesn't allow
+// anonymous enums in assertions. Therefore the following test is not
+// done on Mac.
+// Sun Studio and HP aCC also reject this code.
+#if !GTEST_OS_MAC && !defined(__SUNPRO_CC) && !defined(__HP_aCC)
+
+// Tests using assertions with anonymous enums.
+enum {
+ kCaseA = -1,
+
+# if GTEST_OS_LINUX
+
+ // We want to test the case where the size of the anonymous enum is
+ // larger than sizeof(int), to make sure our implementation of the
+ // assertions doesn't truncate the enums. However, MSVC
+ // (incorrectly) doesn't allow an enum value to exceed the range of
+ // an int, so this has to be conditionally compiled.
+ //
+ // On Linux, kCaseB and kCaseA have the same value when truncated to
+ // int size. We want to test whether this will confuse the
+ // assertions.
+ kCaseB = testing::internal::kMaxBiggestInt,
+
+# else
+
+ kCaseB = INT_MAX,
+
+# endif // GTEST_OS_LINUX
+
+ kCaseC = 42
+};
+
+TEST(AssertionTest, AnonymousEnum) {
+# if GTEST_OS_LINUX
+
+ EXPECT_EQ(static_cast<int>(kCaseA), static_cast<int>(kCaseB));
+
+# endif // GTEST_OS_LINUX
+
+ EXPECT_EQ(kCaseA, kCaseA);
+ EXPECT_NE(kCaseA, kCaseB);
+ EXPECT_LT(kCaseA, kCaseB);
+ EXPECT_LE(kCaseA, kCaseB);
+ EXPECT_GT(kCaseB, kCaseA);
+ EXPECT_GE(kCaseA, kCaseA);
+ EXPECT_NONFATAL_FAILURE(EXPECT_GE(kCaseA, kCaseB),
+ "(kCaseA) >= (kCaseB)");
+ EXPECT_NONFATAL_FAILURE(EXPECT_GE(kCaseA, kCaseC),
+ "-1 vs 42");
+
+ ASSERT_EQ(kCaseA, kCaseA);
+ ASSERT_NE(kCaseA, kCaseB);
+ ASSERT_LT(kCaseA, kCaseB);
+ ASSERT_LE(kCaseA, kCaseB);
+ ASSERT_GT(kCaseB, kCaseA);
+ ASSERT_GE(kCaseA, kCaseA);
+
+# ifndef __BORLANDC__
+
+ // ICE's in C++Builder.
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(kCaseA, kCaseB),
+ "Value of: kCaseB");
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(kCaseA, kCaseC),
+ "Actual: 42");
+# endif
+
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(kCaseA, kCaseC),
+ "Which is: -1");
+}
+
+#endif // !GTEST_OS_MAC && !defined(__SUNPRO_CC)
+
+#if GTEST_OS_WINDOWS
+
+static HRESULT UnexpectedHRESULTFailure() {
+ return E_UNEXPECTED;
+}
+
+static HRESULT OkHRESULTSuccess() {
+ return S_OK;
+}
+
+static HRESULT FalseHRESULTSuccess() {
+ return S_FALSE;
+}
+
+// HRESULT assertion tests test both zero and non-zero
+// success codes as well as failure message for each.
+//
+// Windows CE doesn't support message texts.
+TEST(HRESULTAssertionTest, EXPECT_HRESULT_SUCCEEDED) {
+ EXPECT_HRESULT_SUCCEEDED(S_OK);
+ EXPECT_HRESULT_SUCCEEDED(S_FALSE);
+
+ EXPECT_NONFATAL_FAILURE(EXPECT_HRESULT_SUCCEEDED(UnexpectedHRESULTFailure()),
+ "Expected: (UnexpectedHRESULTFailure()) succeeds.\n"
+ " Actual: 0x8000FFFF");
+}
+
+TEST(HRESULTAssertionTest, ASSERT_HRESULT_SUCCEEDED) {
+ ASSERT_HRESULT_SUCCEEDED(S_OK);
+ ASSERT_HRESULT_SUCCEEDED(S_FALSE);
+
+ EXPECT_FATAL_FAILURE(ASSERT_HRESULT_SUCCEEDED(UnexpectedHRESULTFailure()),
+ "Expected: (UnexpectedHRESULTFailure()) succeeds.\n"
+ " Actual: 0x8000FFFF");
+}
+
+TEST(HRESULTAssertionTest, EXPECT_HRESULT_FAILED) {
+ EXPECT_HRESULT_FAILED(E_UNEXPECTED);
+
+ EXPECT_NONFATAL_FAILURE(EXPECT_HRESULT_FAILED(OkHRESULTSuccess()),
+ "Expected: (OkHRESULTSuccess()) fails.\n"
+ " Actual: 0x00000000");
+ EXPECT_NONFATAL_FAILURE(EXPECT_HRESULT_FAILED(FalseHRESULTSuccess()),
+ "Expected: (FalseHRESULTSuccess()) fails.\n"
+ " Actual: 0x00000001");
+}
+
+TEST(HRESULTAssertionTest, ASSERT_HRESULT_FAILED) {
+ ASSERT_HRESULT_FAILED(E_UNEXPECTED);
+
+# ifndef __BORLANDC__
+
+ // ICE's in C++Builder 2007 and 2009.
+ EXPECT_FATAL_FAILURE(ASSERT_HRESULT_FAILED(OkHRESULTSuccess()),
+ "Expected: (OkHRESULTSuccess()) fails.\n"
+ " Actual: 0x00000000");
+# endif
+
+ EXPECT_FATAL_FAILURE(ASSERT_HRESULT_FAILED(FalseHRESULTSuccess()),
+ "Expected: (FalseHRESULTSuccess()) fails.\n"
+ " Actual: 0x00000001");
+}
+
+// Tests that streaming to the HRESULT macros works.
+TEST(HRESULTAssertionTest, Streaming) {
+ EXPECT_HRESULT_SUCCEEDED(S_OK) << "unexpected failure";
+ ASSERT_HRESULT_SUCCEEDED(S_OK) << "unexpected failure";
+ EXPECT_HRESULT_FAILED(E_UNEXPECTED) << "unexpected failure";
+ ASSERT_HRESULT_FAILED(E_UNEXPECTED) << "unexpected failure";
+
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_HRESULT_SUCCEEDED(E_UNEXPECTED) << "expected failure",
+ "expected failure");
+
+# ifndef __BORLANDC__
+
+ // ICE's in C++Builder 2007 and 2009.
+ EXPECT_FATAL_FAILURE(
+ ASSERT_HRESULT_SUCCEEDED(E_UNEXPECTED) << "expected failure",
+ "expected failure");
+# endif
+
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_HRESULT_FAILED(S_OK) << "expected failure",
+ "expected failure");
+
+ EXPECT_FATAL_FAILURE(
+ ASSERT_HRESULT_FAILED(S_OK) << "expected failure",
+ "expected failure");
+}
+
+#endif // GTEST_OS_WINDOWS
+
+#ifdef __BORLANDC__
+// Silences warnings: "Condition is always true", "Unreachable code"
+# pragma option push -w-ccc -w-rch
+#endif
+
+// Tests that the assertion macros behave like single statements.
+TEST(AssertionSyntaxTest, BasicAssertionsBehavesLikeSingleStatement) {
+ if (AlwaysFalse())
+ ASSERT_TRUE(false) << "This should never be executed; "
+ "It's a compilation test only.";
+
+ if (AlwaysTrue())
+ EXPECT_FALSE(false);
+ else
+ ; // NOLINT
+
+ if (AlwaysFalse())
+ ASSERT_LT(1, 3);
+
+ if (AlwaysFalse())
+ ; // NOLINT
+ else
+ EXPECT_GT(3, 2) << "";
+}
+
+#if GTEST_HAS_EXCEPTIONS
+// Tests that the compiler will not complain about unreachable code in the
+// EXPECT_THROW/EXPECT_ANY_THROW/EXPECT_NO_THROW macros.
+TEST(ExpectThrowTest, DoesNotGenerateUnreachableCodeWarning) {
+ int n = 0;
+
+ EXPECT_THROW(throw 1, int);
+ EXPECT_NONFATAL_FAILURE(EXPECT_THROW(n++, int), "");
+ EXPECT_NONFATAL_FAILURE(EXPECT_THROW(throw 1, const char*), "");
+ EXPECT_NO_THROW(n++);
+ EXPECT_NONFATAL_FAILURE(EXPECT_NO_THROW(throw 1), "");
+ EXPECT_ANY_THROW(throw 1);
+ EXPECT_NONFATAL_FAILURE(EXPECT_ANY_THROW(n++), "");
+}
+
+TEST(AssertionSyntaxTest, ExceptionAssertionsBehavesLikeSingleStatement) {
+ if (AlwaysFalse())
+ EXPECT_THROW(ThrowNothing(), bool);
+
+ if (AlwaysTrue())
+ EXPECT_THROW(ThrowAnInteger(), int);
+ else
+ ; // NOLINT
+
+ if (AlwaysFalse())
+ EXPECT_NO_THROW(ThrowAnInteger());
+
+ if (AlwaysTrue())
+ EXPECT_NO_THROW(ThrowNothing());
+ else
+ ; // NOLINT
+
+ if (AlwaysFalse())
+ EXPECT_ANY_THROW(ThrowNothing());
+
+ if (AlwaysTrue())
+ EXPECT_ANY_THROW(ThrowAnInteger());
+ else
+ ; // NOLINT
+}
+#endif // GTEST_HAS_EXCEPTIONS
+
+TEST(AssertionSyntaxTest, NoFatalFailureAssertionsBehavesLikeSingleStatement) {
+ if (AlwaysFalse())
+ EXPECT_NO_FATAL_FAILURE(FAIL()) << "This should never be executed. "
+ << "It's a compilation test only.";
+ else
+ ; // NOLINT
+
+ if (AlwaysFalse())
+ ASSERT_NO_FATAL_FAILURE(FAIL()) << "";
+ else
+ ; // NOLINT
+
+ if (AlwaysTrue())
+ EXPECT_NO_FATAL_FAILURE(SUCCEED());
+ else
+ ; // NOLINT
+
+ if (AlwaysFalse())
+ ; // NOLINT
+ else
+ ASSERT_NO_FATAL_FAILURE(SUCCEED());
+}
+
+// Tests that the assertion macros work well with switch statements.
+TEST(AssertionSyntaxTest, WorksWithSwitch) {
+ switch (0) {
+ case 1:
+ break;
+ default:
+ ASSERT_TRUE(true);
+ }
+
+ switch (0)
+ case 0:
+ EXPECT_FALSE(false) << "EXPECT_FALSE failed in switch case";
+
+ // Binary assertions are implemented using a different code path
+ // than the Boolean assertions. Hence we test them separately.
+ switch (0) {
+ case 1:
+ default:
+ ASSERT_EQ(1, 1) << "ASSERT_EQ failed in default switch handler";
+ }
+
+ switch (0)
+ case 0:
+ EXPECT_NE(1, 2);
+}
+
+#if GTEST_HAS_EXCEPTIONS
+
+void ThrowAString() {
+ throw "String";
+}
+
+// Test that the exception assertion macros compile and work with const
+// type qualifier.
+TEST(AssertionSyntaxTest, WorksWithConst) {
+ ASSERT_THROW(ThrowAString(), const char*);
+
+ EXPECT_THROW(ThrowAString(), const char*);
+}
+
+#endif // GTEST_HAS_EXCEPTIONS
+
+} // namespace
+
+namespace testing {
+
+// Tests that Google Test tracks SUCCEED*.
+TEST(SuccessfulAssertionTest, SUCCEED) {
+ SUCCEED();
+ SUCCEED() << "OK";
+ EXPECT_EQ(2, GetUnitTestImpl()->current_test_result()->total_part_count());
+}
+
+// Tests that Google Test doesn't track successful EXPECT_*.
+TEST(SuccessfulAssertionTest, EXPECT) {
+ EXPECT_TRUE(true);
+ EXPECT_EQ(0, GetUnitTestImpl()->current_test_result()->total_part_count());
+}
+
+// Tests that Google Test doesn't track successful EXPECT_STR*.
+TEST(SuccessfulAssertionTest, EXPECT_STR) {
+ EXPECT_STREQ("", "");
+ EXPECT_EQ(0, GetUnitTestImpl()->current_test_result()->total_part_count());
+}
+
+// Tests that Google Test doesn't track successful ASSERT_*.
+TEST(SuccessfulAssertionTest, ASSERT) {
+ ASSERT_TRUE(true);
+ EXPECT_EQ(0, GetUnitTestImpl()->current_test_result()->total_part_count());
+}
+
+// Tests that Google Test doesn't track successful ASSERT_STR*.
+TEST(SuccessfulAssertionTest, ASSERT_STR) {
+ ASSERT_STREQ("", "");
+ EXPECT_EQ(0, GetUnitTestImpl()->current_test_result()->total_part_count());
+}
+
+} // namespace testing
+
+namespace {
+
+// Tests EXPECT_TRUE.
+TEST(ExpectTest, EXPECT_TRUE) {
+ EXPECT_TRUE(2 > 1); // NOLINT
+ EXPECT_NONFATAL_FAILURE(EXPECT_TRUE(2 < 1),
+ "Value of: 2 < 1\n"
+ " Actual: false\n"
+ "Expected: true");
+ EXPECT_NONFATAL_FAILURE(EXPECT_TRUE(2 > 3),
+ "2 > 3");
+}
+
+// Tests EXPECT_TRUE(predicate) for predicates returning AssertionResult.
+TEST(ExpectTest, ExpectTrueWithAssertionResult) {
+ EXPECT_TRUE(ResultIsEven(2));
+ EXPECT_NONFATAL_FAILURE(EXPECT_TRUE(ResultIsEven(3)),
+ "Value of: ResultIsEven(3)\n"
+ " Actual: false (3 is odd)\n"
+ "Expected: true");
+ EXPECT_TRUE(ResultIsEvenNoExplanation(2));
+ EXPECT_NONFATAL_FAILURE(EXPECT_TRUE(ResultIsEvenNoExplanation(3)),
+ "Value of: ResultIsEvenNoExplanation(3)\n"
+ " Actual: false (3 is odd)\n"
+ "Expected: true");
+}
+
+// Tests EXPECT_FALSE.
+TEST(ExpectTest, EXPECT_FALSE) {
+ EXPECT_FALSE(2 < 1); // NOLINT
+ EXPECT_NONFATAL_FAILURE(EXPECT_FALSE(2 > 1),
+ "Value of: 2 > 1\n"
+ " Actual: true\n"
+ "Expected: false");
+ EXPECT_NONFATAL_FAILURE(EXPECT_FALSE(2 < 3),
+ "2 < 3");
+}
+
+// Tests EXPECT_FALSE(predicate) for predicates returning AssertionResult.
+TEST(ExpectTest, ExpectFalseWithAssertionResult) {
+ EXPECT_FALSE(ResultIsEven(3));
+ EXPECT_NONFATAL_FAILURE(EXPECT_FALSE(ResultIsEven(2)),
+ "Value of: ResultIsEven(2)\n"
+ " Actual: true (2 is even)\n"
+ "Expected: false");
+ EXPECT_FALSE(ResultIsEvenNoExplanation(3));
+ EXPECT_NONFATAL_FAILURE(EXPECT_FALSE(ResultIsEvenNoExplanation(2)),
+ "Value of: ResultIsEvenNoExplanation(2)\n"
+ " Actual: true\n"
+ "Expected: false");
+}
+
+#ifdef __BORLANDC__
+// Restores warnings after previous "#pragma option push" supressed them
+# pragma option pop
+#endif
+
+// Tests EXPECT_EQ.
+TEST(ExpectTest, EXPECT_EQ) {
+ EXPECT_EQ(5, 2 + 3);
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(5, 2*3),
+ "Value of: 2*3\n"
+ " Actual: 6\n"
+ "Expected: 5");
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(5, 2 - 3),
+ "2 - 3");
+}
+
+// Tests using EXPECT_EQ on double values. The purpose is to make
+// sure that the specialization we did for integer and anonymous enums
+// isn't used for double arguments.
+TEST(ExpectTest, EXPECT_EQ_Double) {
+ // A success.
+ EXPECT_EQ(5.6, 5.6);
+
+ // A failure.
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(5.1, 5.2),
+ "5.1");
+}
+
+#if GTEST_CAN_COMPARE_NULL
+// Tests EXPECT_EQ(NULL, pointer).
+TEST(ExpectTest, EXPECT_EQ_NULL) {
+ // A success.
+ const char* p = NULL;
+ // Some older GCC versions may issue a spurious warning in this or the next
+ // assertion statement. This warning should not be suppressed with
+ // static_cast since the test verifies the ability to use bare NULL as the
+ // expected parameter to the macro.
+ EXPECT_EQ(NULL, p);
+
+ // A failure.
+ int n = 0;
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(NULL, &n),
+ "Value of: &n\n");
+}
+#endif // GTEST_CAN_COMPARE_NULL
+
+// Tests EXPECT_EQ(0, non_pointer). Since the literal 0 can be
+// treated as a null pointer by the compiler, we need to make sure
+// that EXPECT_EQ(0, non_pointer) isn't interpreted by Google Test as
+// EXPECT_EQ(static_cast<void*>(NULL), non_pointer).
+TEST(ExpectTest, EXPECT_EQ_0) {
+ int n = 0;
+
+ // A success.
+ EXPECT_EQ(0, n);
+
+ // A failure.
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(0, 5.6),
+ "Expected: 0");
+}
+
+// Tests EXPECT_NE.
+TEST(ExpectTest, EXPECT_NE) {
+ EXPECT_NE(6, 7);
+
+ EXPECT_NONFATAL_FAILURE(EXPECT_NE('a', 'a'),
+ "Expected: ('a') != ('a'), "
+ "actual: 'a' (97, 0x61) vs 'a' (97, 0x61)");
+ EXPECT_NONFATAL_FAILURE(EXPECT_NE(2, 2),
+ "2");
+ char* const p0 = NULL;
+ EXPECT_NONFATAL_FAILURE(EXPECT_NE(p0, p0),
+ "p0");
+ // Only way to get the Nokia compiler to compile the cast
+ // is to have a separate void* variable first. Putting
+ // the two casts on the same line doesn't work, neither does
+ // a direct C-style to char*.
+ void* pv1 = (void*)0x1234; // NOLINT
+ char* const p1 = reinterpret_cast<char*>(pv1);
+ EXPECT_NONFATAL_FAILURE(EXPECT_NE(p1, p1),
+ "p1");
+}
+
+// Tests EXPECT_LE.
+TEST(ExpectTest, EXPECT_LE) {
+ EXPECT_LE(2, 3);
+ EXPECT_LE(2, 2);
+ EXPECT_NONFATAL_FAILURE(EXPECT_LE(2, 0),
+ "Expected: (2) <= (0), actual: 2 vs 0");
+ EXPECT_NONFATAL_FAILURE(EXPECT_LE(1.1, 0.9),
+ "(1.1) <= (0.9)");
+}
+
+// Tests EXPECT_LT.
+TEST(ExpectTest, EXPECT_LT) {
+ EXPECT_LT(2, 3);
+ EXPECT_NONFATAL_FAILURE(EXPECT_LT(2, 2),
+ "Expected: (2) < (2), actual: 2 vs 2");
+ EXPECT_NONFATAL_FAILURE(EXPECT_LT(2, 1),
+ "(2) < (1)");
+}
+
+// Tests EXPECT_GE.
+TEST(ExpectTest, EXPECT_GE) {
+ EXPECT_GE(2, 1);
+ EXPECT_GE(2, 2);
+ EXPECT_NONFATAL_FAILURE(EXPECT_GE(2, 3),
+ "Expected: (2) >= (3), actual: 2 vs 3");
+ EXPECT_NONFATAL_FAILURE(EXPECT_GE(0.9, 1.1),
+ "(0.9) >= (1.1)");
+}
+
+// Tests EXPECT_GT.
+TEST(ExpectTest, EXPECT_GT) {
+ EXPECT_GT(2, 1);
+ EXPECT_NONFATAL_FAILURE(EXPECT_GT(2, 2),
+ "Expected: (2) > (2), actual: 2 vs 2");
+ EXPECT_NONFATAL_FAILURE(EXPECT_GT(2, 3),
+ "(2) > (3)");
+}
+
+#if GTEST_HAS_EXCEPTIONS
+
+// Tests EXPECT_THROW.
+TEST(ExpectTest, EXPECT_THROW) {
+ EXPECT_THROW(ThrowAnInteger(), int);
+ EXPECT_NONFATAL_FAILURE(EXPECT_THROW(ThrowAnInteger(), bool),
+ "Expected: ThrowAnInteger() throws an exception of "
+ "type bool.\n Actual: it throws a different type.");
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_THROW(ThrowNothing(), bool),
+ "Expected: ThrowNothing() throws an exception of type bool.\n"
+ " Actual: it throws nothing.");
+}
+
+// Tests EXPECT_NO_THROW.
+TEST(ExpectTest, EXPECT_NO_THROW) {
+ EXPECT_NO_THROW(ThrowNothing());
+ EXPECT_NONFATAL_FAILURE(EXPECT_NO_THROW(ThrowAnInteger()),
+ "Expected: ThrowAnInteger() doesn't throw an "
+ "exception.\n Actual: it throws.");
+}
+
+// Tests EXPECT_ANY_THROW.
+TEST(ExpectTest, EXPECT_ANY_THROW) {
+ EXPECT_ANY_THROW(ThrowAnInteger());
+ EXPECT_NONFATAL_FAILURE(
+ EXPECT_ANY_THROW(ThrowNothing()),
+ "Expected: ThrowNothing() throws an exception.\n"
+ " Actual: it doesn't.");
+}
+
+#endif // GTEST_HAS_EXCEPTIONS
+
+// Make sure we deal with the precedence of <<.
+TEST(ExpectTest, ExpectPrecedence) {
+ EXPECT_EQ(1 < 2, true);
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(true, true && false),
+ "Value of: true && false");
+}
+
+
+// Tests the StreamableToString() function.
+
+// Tests using StreamableToString() on a scalar.
+TEST(StreamableToStringTest, Scalar) {
+ EXPECT_STREQ("5", StreamableToString(5).c_str());
+}
+
+// Tests using StreamableToString() on a non-char pointer.
+TEST(StreamableToStringTest, Pointer) {
+ int n = 0;
+ int* p = &n;
+ EXPECT_STRNE("(null)", StreamableToString(p).c_str());
+}
+
+// Tests using StreamableToString() on a NULL non-char pointer.
+TEST(StreamableToStringTest, NullPointer) {
+ int* p = NULL;
+ EXPECT_STREQ("(null)", StreamableToString(p).c_str());
+}
+
+// Tests using StreamableToString() on a C string.
+TEST(StreamableToStringTest, CString) {
+ EXPECT_STREQ("Foo", StreamableToString("Foo").c_str());
+}
+
+// Tests using StreamableToString() on a NULL C string.
+TEST(StreamableToStringTest, NullCString) {
+ char* p = NULL;
+ EXPECT_STREQ("(null)", StreamableToString(p).c_str());
+}
+
+// Tests using streamable values as assertion messages.
+
+// Tests using std::string as an assertion message.
+TEST(StreamableTest, string) {
+ static const std::string str(
+ "This failure message is a std::string, and is expected.");
+ EXPECT_FATAL_FAILURE(FAIL() << str,
+ str.c_str());
+}
+
+// Tests that we can output strings containing embedded NULs.
+// Limited to Linux because we can only do this with std::string's.
+TEST(StreamableTest, stringWithEmbeddedNUL) {
+ static const char char_array_with_nul[] =
+ "Here's a NUL\0 and some more string";
+ static const std::string string_with_nul(char_array_with_nul,
+ sizeof(char_array_with_nul)
+ - 1); // drops the trailing NUL
+ EXPECT_FATAL_FAILURE(FAIL() << string_with_nul,
+ "Here's a NUL\\0 and some more string");
+}
+
+// Tests that we can output a NUL char.
+TEST(StreamableTest, NULChar) {
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ FAIL() << "A NUL" << '\0' << " and some more string";
+ }, "A NUL\\0 and some more string");
+}
+
+// Tests using int as an assertion message.
+TEST(StreamableTest, int) {
+ EXPECT_FATAL_FAILURE(FAIL() << 900913,
+ "900913");
+}
+
+// Tests using NULL char pointer as an assertion message.
+//
+// In MSVC, streaming a NULL char * causes access violation. Google Test
+// implemented a workaround (substituting "(null)" for NULL). This
+// tests whether the workaround works.
+TEST(StreamableTest, NullCharPtr) {
+ EXPECT_FATAL_FAILURE(FAIL() << static_cast<const char*>(NULL),
+ "(null)");
+}
+
+// Tests that basic IO manipulators (endl, ends, and flush) can be
+// streamed to testing::Message.
+TEST(StreamableTest, BasicIoManip) {
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ FAIL() << "Line 1." << std::endl
+ << "A NUL char " << std::ends << std::flush << " in line 2.";
+ }, "Line 1.\nA NUL char \\0 in line 2.");
+}
+
+// Tests the macros that haven't been covered so far.
+
+void AddFailureHelper(bool* aborted) {
+ *aborted = true;
+ ADD_FAILURE() << "Failure";
+ *aborted = false;
+}
+
+// Tests ADD_FAILURE.
+TEST(MacroTest, ADD_FAILURE) {
+ bool aborted = true;
+ EXPECT_NONFATAL_FAILURE(AddFailureHelper(&aborted),
+ "Failure");
+ EXPECT_FALSE(aborted);
+}
+
+// Tests ADD_FAILURE_AT.
+TEST(MacroTest, ADD_FAILURE_AT) {
+ // Verifies that ADD_FAILURE_AT does generate a nonfatal failure and
+ // the failure message contains the user-streamed part.
+ EXPECT_NONFATAL_FAILURE(ADD_FAILURE_AT("foo.cc", 42) << "Wrong!", "Wrong!");
+
+ // Verifies that the user-streamed part is optional.
+ EXPECT_NONFATAL_FAILURE(ADD_FAILURE_AT("foo.cc", 42), "Failed");
+
+ // Unfortunately, we cannot verify that the failure message contains
+ // the right file path and line number the same way, as
+ // EXPECT_NONFATAL_FAILURE() doesn't get to see the file path and
+ // line number. Instead, we do that in gtest_output_test_.cc.
+}
+
+// Tests FAIL.
+TEST(MacroTest, FAIL) {
+ EXPECT_FATAL_FAILURE(FAIL(),
+ "Failed");
+ EXPECT_FATAL_FAILURE(FAIL() << "Intentional failure.",
+ "Intentional failure.");
+}
+
+// Tests SUCCEED
+TEST(MacroTest, SUCCEED) {
+ SUCCEED();
+ SUCCEED() << "Explicit success.";
+}
+
+
+// Tests for EXPECT_EQ() and ASSERT_EQ().
+//
+// These tests fail *intentionally*, s.t. the failure messages can be
+// generated and tested.
+//
+// We have different tests for different argument types.
+
+// Tests using bool values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, Bool) {
+ EXPECT_EQ(true, true);
+ EXPECT_FATAL_FAILURE({
+ bool false_value = false;
+ ASSERT_EQ(false_value, true);
+ }, "Value of: true");
+}
+
+// Tests using int values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, Int) {
+ ASSERT_EQ(32, 32);
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(32, 33),
+ "33");
+}
+
+// Tests using time_t values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, Time_T) {
+ EXPECT_EQ(static_cast<time_t>(0),
+ static_cast<time_t>(0));
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(static_cast<time_t>(0),
+ static_cast<time_t>(1234)),
+ "1234");
+}
+
+// Tests using char values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, Char) {
+ ASSERT_EQ('z', 'z');
+ const char ch = 'b';
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ('\0', ch),
+ "ch");
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ('a', ch),
+ "ch");
+}
+
+// Tests using wchar_t values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, WideChar) {
+ EXPECT_EQ(L'b', L'b');
+
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(L'\0', L'x'),
+ "Value of: L'x'\n"
+ " Actual: L'x' (120, 0x78)\n"
+ "Expected: L'\0'\n"
+ "Which is: L'\0' (0, 0x0)");
+
+ static wchar_t wchar;
+ wchar = L'b';
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(L'a', wchar),
+ "wchar");
+ wchar = 0x8119;
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(static_cast<wchar_t>(0x8120), wchar),
+ "Value of: wchar");
+}
+
+// Tests using ::std::string values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, StdString) {
+ // Compares a const char* to an std::string that has identical
+ // content.
+ ASSERT_EQ("Test", ::std::string("Test"));
+
+ // Compares two identical std::strings.
+ static const ::std::string str1("A * in the middle");
+ static const ::std::string str2(str1);
+ EXPECT_EQ(str1, str2);
+
+ // Compares a const char* to an std::string that has different
+ // content
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ("Test", ::std::string("test")),
+ "::std::string(\"test\")");
+
+ // Compares an std::string to a char* that has different content.
+ char* const p1 = const_cast<char*>("foo");
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(::std::string("bar"), p1),
+ "p1");
+
+ // Compares two std::strings that have different contents, one of
+ // which having a NUL character in the middle. This should fail.
+ static ::std::string str3(str1);
+ str3.at(2) = '\0';
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(str1, str3),
+ "Value of: str3\n"
+ " Actual: \"A \\0 in the middle\"");
+}
+
+#if GTEST_HAS_STD_WSTRING
+
+// Tests using ::std::wstring values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, StdWideString) {
+ // Compares two identical std::wstrings.
+ const ::std::wstring wstr1(L"A * in the middle");
+ const ::std::wstring wstr2(wstr1);
+ ASSERT_EQ(wstr1, wstr2);
+
+ // Compares an std::wstring to a const wchar_t* that has identical
+ // content.
+ const wchar_t kTestX8119[] = { 'T', 'e', 's', 't', 0x8119, '\0' };
+ EXPECT_EQ(::std::wstring(kTestX8119), kTestX8119);
+
+ // Compares an std::wstring to a const wchar_t* that has different
+ // content.
+ const wchar_t kTestX8120[] = { 'T', 'e', 's', 't', 0x8120, '\0' };
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_EQ(::std::wstring(kTestX8119), kTestX8120);
+ }, "kTestX8120");
+
+ // Compares two std::wstrings that have different contents, one of
+ // which having a NUL character in the middle.
+ ::std::wstring wstr3(wstr1);
+ wstr3.at(2) = L'\0';
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(wstr1, wstr3),
+ "wstr3");
+
+ // Compares a wchar_t* to an std::wstring that has different
+ // content.
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_EQ(const_cast<wchar_t*>(L"foo"), ::std::wstring(L"bar"));
+ }, "");
+}
+
+#endif // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_GLOBAL_STRING
+// Tests using ::string values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, GlobalString) {
+ // Compares a const char* to a ::string that has identical content.
+ EXPECT_EQ("Test", ::string("Test"));
+
+ // Compares two identical ::strings.
+ const ::string str1("A * in the middle");
+ const ::string str2(str1);
+ ASSERT_EQ(str1, str2);
+
+ // Compares a ::string to a const char* that has different content.
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(::string("Test"), "test"),
+ "test");
+
+ // Compares two ::strings that have different contents, one of which
+ // having a NUL character in the middle.
+ ::string str3(str1);
+ str3.at(2) = '\0';
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(str1, str3),
+ "str3");
+
+ // Compares a ::string to a char* that has different content.
+ EXPECT_FATAL_FAILURE({ // NOLINT
+ ASSERT_EQ(::string("bar"), const_cast<char*>("foo"));
+ }, "");
+}
+
+#endif // GTEST_HAS_GLOBAL_STRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+
+// Tests using ::wstring values in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, GlobalWideString) {
+ // Compares two identical ::wstrings.
+ static const ::wstring wstr1(L"A * in the middle");
+ static const ::wstring wstr2(wstr1);
+ EXPECT_EQ(wstr1, wstr2);
+
+ // Compares a const wchar_t* to a ::wstring that has identical content.
+ const wchar_t kTestX8119[] = { 'T', 'e', 's', 't', 0x8119, '\0' };
+ ASSERT_EQ(kTestX8119, ::wstring(kTestX8119));
+
+ // Compares a const wchar_t* to a ::wstring that has different
+ // content.
+ const wchar_t kTestX8120[] = { 'T', 'e', 's', 't', 0x8120, '\0' };
+ EXPECT_NONFATAL_FAILURE({ // NOLINT
+ EXPECT_EQ(kTestX8120, ::wstring(kTestX8119));
+ }, "Test\\x8119");
+
+ // Compares a wchar_t* to a ::wstring that has different content.
+ wchar_t* const p1 = const_cast<wchar_t*>(L"foo");
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(p1, ::wstring(L"bar")),
+ "bar");
+
+ // Compares two ::wstrings that have different contents, one of which
+ // having a NUL character in the middle.
+ static ::wstring wstr3;
+ wstr3 = wstr1;
+ wstr3.at(2) = L'\0';
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(wstr1, wstr3),
+ "wstr3");
+}
+
+#endif // GTEST_HAS_GLOBAL_WSTRING
+
+// Tests using char pointers in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, CharPointer) {
+ char* const p0 = NULL;
+ // Only way to get the Nokia compiler to compile the cast
+ // is to have a separate void* variable first. Putting
+ // the two casts on the same line doesn't work, neither does
+ // a direct C-style to char*.
+ void* pv1 = (void*)0x1234; // NOLINT
+ void* pv2 = (void*)0xABC0; // NOLINT
+ char* const p1 = reinterpret_cast<char*>(pv1);
+ char* const p2 = reinterpret_cast<char*>(pv2);
+ ASSERT_EQ(p1, p1);
+
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(p0, p2),
+ "Value of: p2");
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(p1, p2),
+ "p2");
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(reinterpret_cast<char*>(0x1234),
+ reinterpret_cast<char*>(0xABC0)),
+ "ABC0");
+}
+
+// Tests using wchar_t pointers in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, WideCharPointer) {
+ wchar_t* const p0 = NULL;
+ // Only way to get the Nokia compiler to compile the cast
+ // is to have a separate void* variable first. Putting
+ // the two casts on the same line doesn't work, neither does
+ // a direct C-style to char*.
+ void* pv1 = (void*)0x1234; // NOLINT
+ void* pv2 = (void*)0xABC0; // NOLINT
+ wchar_t* const p1 = reinterpret_cast<wchar_t*>(pv1);
+ wchar_t* const p2 = reinterpret_cast<wchar_t*>(pv2);
+ EXPECT_EQ(p0, p0);
+
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(p0, p2),
+ "Value of: p2");
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(p1, p2),
+ "p2");
+ void* pv3 = (void*)0x1234; // NOLINT
+ void* pv4 = (void*)0xABC0; // NOLINT
+ const wchar_t* p3 = reinterpret_cast<const wchar_t*>(pv3);
+ const wchar_t* p4 = reinterpret_cast<const wchar_t*>(pv4);
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(p3, p4),
+ "p4");
+}
+
+// Tests using other types of pointers in {EXPECT|ASSERT}_EQ.
+TEST(EqAssertionTest, OtherPointer) {
+ ASSERT_EQ(static_cast<const int*>(NULL),
+ static_cast<const int*>(NULL));
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(static_cast<const int*>(NULL),
+ reinterpret_cast<const int*>(0x1234)),
+ "0x1234");
+}
+
+// A class that supports binary comparison operators but not streaming.
+class UnprintableChar {
+ public:
+ explicit UnprintableChar(char ch) : char_(ch) {}
+
+ bool operator==(const UnprintableChar& rhs) const {
+ return char_ == rhs.char_;
+ }
+ bool operator!=(const UnprintableChar& rhs) const {
+ return char_ != rhs.char_;
+ }
+ bool operator<(const UnprintableChar& rhs) const {
+ return char_ < rhs.char_;
+ }
+ bool operator<=(const UnprintableChar& rhs) const {
+ return char_ <= rhs.char_;
+ }
+ bool operator>(const UnprintableChar& rhs) const {
+ return char_ > rhs.char_;
+ }
+ bool operator>=(const UnprintableChar& rhs) const {
+ return char_ >= rhs.char_;
+ }
+
+ private:
+ char char_;
+};
+
+// Tests that ASSERT_EQ() and friends don't require the arguments to
+// be printable.
+TEST(ComparisonAssertionTest, AcceptsUnprintableArgs) {
+ const UnprintableChar x('x'), y('y');
+ ASSERT_EQ(x, x);
+ EXPECT_NE(x, y);
+ ASSERT_LT(x, y);
+ EXPECT_LE(x, y);
+ ASSERT_GT(y, x);
+ EXPECT_GE(x, x);
+
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(x, y), "1-byte object <78>");
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(x, y), "1-byte object <79>");
+ EXPECT_NONFATAL_FAILURE(EXPECT_LT(y, y), "1-byte object <79>");
+ EXPECT_NONFATAL_FAILURE(EXPECT_GT(x, y), "1-byte object <78>");
+ EXPECT_NONFATAL_FAILURE(EXPECT_GT(x, y), "1-byte object <79>");
+
+ // Code tested by EXPECT_FATAL_FAILURE cannot reference local
+ // variables, so we have to write UnprintableChar('x') instead of x.
+#ifndef __BORLANDC__
+ // ICE's in C++Builder.
+ EXPECT_FATAL_FAILURE(ASSERT_NE(UnprintableChar('x'), UnprintableChar('x')),
+ "1-byte object <78>");
+ EXPECT_FATAL_FAILURE(ASSERT_LE(UnprintableChar('y'), UnprintableChar('x')),
+ "1-byte object <78>");
+#endif
+ EXPECT_FATAL_FAILURE(ASSERT_LE(UnprintableChar('y'), UnprintableChar('x')),
+ "1-byte object <79>");
+ EXPECT_FATAL_FAILURE(ASSERT_GE(UnprintableChar('x'), UnprintableChar('y')),
+ "1-byte object <78>");
+ EXPECT_FATAL_FAILURE(ASSERT_GE(UnprintableChar('x'), UnprintableChar('y')),
+ "1-byte object <79>");
+}
+
+// Tests the FRIEND_TEST macro.
+
+// This class has a private member we want to test. We will test it
+// both in a TEST and in a TEST_F.
+class Foo {
+ public:
+ Foo() {}
+
+ private:
+ int Bar() const { return 1; }
+
+ // Declares the friend tests that can access the private member
+ // Bar().
+ FRIEND_TEST(FRIEND_TEST_Test, TEST);
+ FRIEND_TEST(FRIEND_TEST_Test2, TEST_F);
+};
+
+// Tests that the FRIEND_TEST declaration allows a TEST to access a
+// class's private members. This should compile.
+TEST(FRIEND_TEST_Test, TEST) {
+ ASSERT_EQ(1, Foo().Bar());
+}
+
+// The fixture needed to test using FRIEND_TEST with TEST_F.
+class FRIEND_TEST_Test2 : public Test {
+ protected:
+ Foo foo;
+};
+
+// Tests that the FRIEND_TEST declaration allows a TEST_F to access a
+// class's private members. This should compile.
+TEST_F(FRIEND_TEST_Test2, TEST_F) {
+ ASSERT_EQ(1, foo.Bar());
+}
+
+// Tests the life cycle of Test objects.
+
+// The test fixture for testing the life cycle of Test objects.
+//
+// This class counts the number of live test objects that uses this
+// fixture.
+class TestLifeCycleTest : public Test {
+ protected:
+ // Constructor. Increments the number of test objects that uses
+ // this fixture.
+ TestLifeCycleTest() { count_++; }
+
+ // Destructor. Decrements the number of test objects that uses this
+ // fixture.
+ ~TestLifeCycleTest() { count_--; }
+
+ // Returns the number of live test objects that uses this fixture.
+ int count() const { return count_; }
+
+ private:
+ static int count_;
+};
+
+int TestLifeCycleTest::count_ = 0;
+
+// Tests the life cycle of test objects.
+TEST_F(TestLifeCycleTest, Test1) {
+ // There should be only one test object in this test case that's
+ // currently alive.
+ ASSERT_EQ(1, count());
+}
+
+// Tests the life cycle of test objects.
+TEST_F(TestLifeCycleTest, Test2) {
+ // After Test1 is done and Test2 is started, there should still be
+ // only one live test object, as the object for Test1 should've been
+ // deleted.
+ ASSERT_EQ(1, count());
+}
+
+} // namespace
+
+// Tests that the copy constructor works when it is NOT optimized away by
+// the compiler.
+TEST(AssertionResultTest, CopyConstructorWorksWhenNotOptimied) {
+ // Checks that the copy constructor doesn't try to dereference NULL pointers
+ // in the source object.
+ AssertionResult r1 = AssertionSuccess();
+ AssertionResult r2 = r1;
+ // The following line is added to prevent the compiler from optimizing
+ // away the constructor call.
+ r1 << "abc";
+
+ AssertionResult r3 = r1;
+ EXPECT_EQ(static_cast<bool>(r3), static_cast<bool>(r1));
+ EXPECT_STREQ("abc", r1.message());
+}
+
+// Tests that AssertionSuccess and AssertionFailure construct
+// AssertionResult objects as expected.
+TEST(AssertionResultTest, ConstructionWorks) {
+ AssertionResult r1 = AssertionSuccess();
+ EXPECT_TRUE(r1);
+ EXPECT_STREQ("", r1.message());
+
+ AssertionResult r2 = AssertionSuccess() << "abc";
+ EXPECT_TRUE(r2);
+ EXPECT_STREQ("abc", r2.message());
+
+ AssertionResult r3 = AssertionFailure();
+ EXPECT_FALSE(r3);
+ EXPECT_STREQ("", r3.message());
+
+ AssertionResult r4 = AssertionFailure() << "def";
+ EXPECT_FALSE(r4);
+ EXPECT_STREQ("def", r4.message());
+
+ AssertionResult r5 = AssertionFailure(Message() << "ghi");
+ EXPECT_FALSE(r5);
+ EXPECT_STREQ("ghi", r5.message());
+}
+
+// Tests that the negation flips the predicate result but keeps the message.
+TEST(AssertionResultTest, NegationWorks) {
+ AssertionResult r1 = AssertionSuccess() << "abc";
+ EXPECT_FALSE(!r1);
+ EXPECT_STREQ("abc", (!r1).message());
+
+ AssertionResult r2 = AssertionFailure() << "def";
+ EXPECT_TRUE(!r2);
+ EXPECT_STREQ("def", (!r2).message());
+}
+
+TEST(AssertionResultTest, StreamingWorks) {
+ AssertionResult r = AssertionSuccess();
+ r << "abc" << 'd' << 0 << true;
+ EXPECT_STREQ("abcd0true", r.message());
+}
+
+TEST(AssertionResultTest, CanStreamOstreamManipulators) {
+ AssertionResult r = AssertionSuccess();
+ r << "Data" << std::endl << std::flush << std::ends << "Will be visible";
+ EXPECT_STREQ("Data\n\\0Will be visible", r.message());
+}
+
+// Tests streaming a user type whose definition and operator << are
+// both in the global namespace.
+class Base {
+ public:
+ explicit Base(int an_x) : x_(an_x) {}
+ int x() const { return x_; }
+ private:
+ int x_;
+};
+std::ostream& operator<<(std::ostream& os,
+ const Base& val) {
+ return os << val.x();
+}
+std::ostream& operator<<(std::ostream& os,
+ const Base* pointer) {
+ return os << "(" << pointer->x() << ")";
+}
+
+TEST(MessageTest, CanStreamUserTypeInGlobalNameSpace) {
+ Message msg;
+ Base a(1);
+
+ msg << a << &a; // Uses ::operator<<.
+ EXPECT_STREQ("1(1)", msg.GetString().c_str());
+}
+
+// Tests streaming a user type whose definition and operator<< are
+// both in an unnamed namespace.
+namespace {
+class MyTypeInUnnamedNameSpace : public Base {
+ public:
+ explicit MyTypeInUnnamedNameSpace(int an_x): Base(an_x) {}
+};
+std::ostream& operator<<(std::ostream& os,
+ const MyTypeInUnnamedNameSpace& val) {
+ return os << val.x();
+}
+std::ostream& operator<<(std::ostream& os,
+ const MyTypeInUnnamedNameSpace* pointer) {
+ return os << "(" << pointer->x() << ")";
+}
+} // namespace
+
+TEST(MessageTest, CanStreamUserTypeInUnnamedNameSpace) {
+ Message msg;
+ MyTypeInUnnamedNameSpace a(1);
+
+ msg << a << &a; // Uses <unnamed_namespace>::operator<<.
+ EXPECT_STREQ("1(1)", msg.GetString().c_str());
+}
+
+// Tests streaming a user type whose definition and operator<< are
+// both in a user namespace.
+namespace namespace1 {
+class MyTypeInNameSpace1 : public Base {
+ public:
+ explicit MyTypeInNameSpace1(int an_x): Base(an_x) {}
+};
+std::ostream& operator<<(std::ostream& os,
+ const MyTypeInNameSpace1& val) {
+ return os << val.x();
+}
+std::ostream& operator<<(std::ostream& os,
+ const MyTypeInNameSpace1* pointer) {
+ return os << "(" << pointer->x() << ")";
+}
+} // namespace namespace1
+
+TEST(MessageTest, CanStreamUserTypeInUserNameSpace) {
+ Message msg;
+ namespace1::MyTypeInNameSpace1 a(1);
+
+ msg << a << &a; // Uses namespace1::operator<<.
+ EXPECT_STREQ("1(1)", msg.GetString().c_str());
+}
+
+// Tests streaming a user type whose definition is in a user namespace
+// but whose operator<< is in the global namespace.
+namespace namespace2 {
+class MyTypeInNameSpace2 : public ::Base {
+ public:
+ explicit MyTypeInNameSpace2(int an_x): Base(an_x) {}
+};
+} // namespace namespace2
+std::ostream& operator<<(std::ostream& os,
+ const namespace2::MyTypeInNameSpace2& val) {
+ return os << val.x();
+}
+std::ostream& operator<<(std::ostream& os,
+ const namespace2::MyTypeInNameSpace2* pointer) {
+ return os << "(" << pointer->x() << ")";
+}
+
+TEST(MessageTest, CanStreamUserTypeInUserNameSpaceWithStreamOperatorInGlobal) {
+ Message msg;
+ namespace2::MyTypeInNameSpace2 a(1);
+
+ msg << a << &a; // Uses ::operator<<.
+ EXPECT_STREQ("1(1)", msg.GetString().c_str());
+}
+
+// Tests streaming NULL pointers to testing::Message.
+TEST(MessageTest, NullPointers) {
+ Message msg;
+ char* const p1 = NULL;
+ unsigned char* const p2 = NULL;
+ int* p3 = NULL;
+ double* p4 = NULL;
+ bool* p5 = NULL;
+ Message* p6 = NULL;
+
+ msg << p1 << p2 << p3 << p4 << p5 << p6;
+ ASSERT_STREQ("(null)(null)(null)(null)(null)(null)",
+ msg.GetString().c_str());
+}
+
+// Tests streaming wide strings to testing::Message.
+TEST(MessageTest, WideStrings) {
+ // Streams a NULL of type const wchar_t*.
+ const wchar_t* const_wstr = NULL;
+ EXPECT_STREQ("(null)",
+ (Message() << const_wstr).GetString().c_str());
+
+ // Streams a NULL of type wchar_t*.
+ wchar_t* wstr = NULL;
+ EXPECT_STREQ("(null)",
+ (Message() << wstr).GetString().c_str());
+
+ // Streams a non-NULL of type const wchar_t*.
+ const_wstr = L"abc\x8119";
+ EXPECT_STREQ("abc\xe8\x84\x99",
+ (Message() << const_wstr).GetString().c_str());
+
+ // Streams a non-NULL of type wchar_t*.
+ wstr = const_cast<wchar_t*>(const_wstr);
+ EXPECT_STREQ("abc\xe8\x84\x99",
+ (Message() << wstr).GetString().c_str());
+}
+
+
+// This line tests that we can define tests in the testing namespace.
+namespace testing {
+
+// Tests the TestInfo class.
+
+class TestInfoTest : public Test {
+ protected:
+ static const TestInfo* GetTestInfo(const char* test_name) {
+ const TestCase* const test_case = GetUnitTestImpl()->
+ GetTestCase("TestInfoTest", "", NULL, NULL);
+
+ for (int i = 0; i < test_case->total_test_count(); ++i) {
+ const TestInfo* const test_info = test_case->GetTestInfo(i);
+ if (strcmp(test_name, test_info->name()) == 0)
+ return test_info;
+ }
+ return NULL;
+ }
+
+ static const TestResult* GetTestResult(
+ const TestInfo* test_info) {
+ return test_info->result();
+ }
+};
+
+// Tests TestInfo::test_case_name() and TestInfo::name().
+TEST_F(TestInfoTest, Names) {
+ const TestInfo* const test_info = GetTestInfo("Names");
+
+ ASSERT_STREQ("TestInfoTest", test_info->test_case_name());
+ ASSERT_STREQ("Names", test_info->name());
+}
+
+// Tests TestInfo::result().
+TEST_F(TestInfoTest, result) {
+ const TestInfo* const test_info = GetTestInfo("result");
+
+ // Initially, there is no TestPartResult for this test.
+ ASSERT_EQ(0, GetTestResult(test_info)->total_part_count());
+
+ // After the previous assertion, there is still none.
+ ASSERT_EQ(0, GetTestResult(test_info)->total_part_count());
+}
+
+// Tests setting up and tearing down a test case.
+
+class SetUpTestCaseTest : public Test {
+ protected:
+ // This will be called once before the first test in this test case
+ // is run.
+ static void SetUpTestCase() {
+ printf("Setting up the test case . . .\n");
+
+ // Initializes some shared resource. In this simple example, we
+ // just create a C string. More complex stuff can be done if
+ // desired.
+ shared_resource_ = "123";
+
+ // Increments the number of test cases that have been set up.
+ counter_++;
+
+ // SetUpTestCase() should be called only once.
+ EXPECT_EQ(1, counter_);
+ }
+
+ // This will be called once after the last test in this test case is
+ // run.
+ static void TearDownTestCase() {
+ printf("Tearing down the test case . . .\n");
+
+ // Decrements the number of test cases that have been set up.
+ counter_--;
+
+ // TearDownTestCase() should be called only once.
+ EXPECT_EQ(0, counter_);
+
+ // Cleans up the shared resource.
+ shared_resource_ = NULL;
+ }
+
+ // This will be called before each test in this test case.
+ virtual void SetUp() {
+ // SetUpTestCase() should be called only once, so counter_ should
+ // always be 1.
+ EXPECT_EQ(1, counter_);
+ }
+
+ // Number of test cases that have been set up.
+ static int counter_;
+
+ // Some resource to be shared by all tests in this test case.
+ static const char* shared_resource_;
+};
+
+int SetUpTestCaseTest::counter_ = 0;
+const char* SetUpTestCaseTest::shared_resource_ = NULL;
+
+// A test that uses the shared resource.
+TEST_F(SetUpTestCaseTest, Test1) {
+ EXPECT_STRNE(NULL, shared_resource_);
+}
+
+// Another test that uses the shared resource.
+TEST_F(SetUpTestCaseTest, Test2) {
+ EXPECT_STREQ("123", shared_resource_);
+}
+
+// The InitGoogleTestTest test case tests testing::InitGoogleTest().
+
+// The Flags struct stores a copy of all Google Test flags.
+struct Flags {
+ // Constructs a Flags struct where each flag has its default value.
+ Flags() : also_run_disabled_tests(false),
+ break_on_failure(false),
+ catch_exceptions(false),
+ death_test_use_fork(false),
+ filter(""),
+ list_tests(false),
+ output(""),
+ print_time(true),
+ random_seed(0),
+ repeat(1),
+ shuffle(false),
+ stack_trace_depth(kMaxStackTraceDepth),
+ stream_result_to(""),
+ throw_on_failure(false) {}
+
+ // Factory methods.
+
+ // Creates a Flags struct where the gtest_also_run_disabled_tests flag has
+ // the given value.
+ static Flags AlsoRunDisabledTests(bool also_run_disabled_tests) {
+ Flags flags;
+ flags.also_run_disabled_tests = also_run_disabled_tests;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_break_on_failure flag has
+ // the given value.
+ static Flags BreakOnFailure(bool break_on_failure) {
+ Flags flags;
+ flags.break_on_failure = break_on_failure;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_catch_exceptions flag has
+ // the given value.
+ static Flags CatchExceptions(bool catch_exceptions) {
+ Flags flags;
+ flags.catch_exceptions = catch_exceptions;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_death_test_use_fork flag has
+ // the given value.
+ static Flags DeathTestUseFork(bool death_test_use_fork) {
+ Flags flags;
+ flags.death_test_use_fork = death_test_use_fork;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_filter flag has the given
+ // value.
+ static Flags Filter(const char* filter) {
+ Flags flags;
+ flags.filter = filter;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_list_tests flag has the
+ // given value.
+ static Flags ListTests(bool list_tests) {
+ Flags flags;
+ flags.list_tests = list_tests;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_output flag has the given
+ // value.
+ static Flags Output(const char* output) {
+ Flags flags;
+ flags.output = output;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_print_time flag has the given
+ // value.
+ static Flags PrintTime(bool print_time) {
+ Flags flags;
+ flags.print_time = print_time;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_random_seed flag has
+ // the given value.
+ static Flags RandomSeed(Int32 random_seed) {
+ Flags flags;
+ flags.random_seed = random_seed;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_repeat flag has the given
+ // value.
+ static Flags Repeat(Int32 repeat) {
+ Flags flags;
+ flags.repeat = repeat;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_shuffle flag has
+ // the given value.
+ static Flags Shuffle(bool shuffle) {
+ Flags flags;
+ flags.shuffle = shuffle;
+ return flags;
+ }
+
+ // Creates a Flags struct where the GTEST_FLAG(stack_trace_depth) flag has
+ // the given value.
+ static Flags StackTraceDepth(Int32 stack_trace_depth) {
+ Flags flags;
+ flags.stack_trace_depth = stack_trace_depth;
+ return flags;
+ }
+
+ // Creates a Flags struct where the GTEST_FLAG(stream_result_to) flag has
+ // the given value.
+ static Flags StreamResultTo(const char* stream_result_to) {
+ Flags flags;
+ flags.stream_result_to = stream_result_to;
+ return flags;
+ }
+
+ // Creates a Flags struct where the gtest_throw_on_failure flag has
+ // the given value.
+ static Flags ThrowOnFailure(bool throw_on_failure) {
+ Flags flags;
+ flags.throw_on_failure = throw_on_failure;
+ return flags;
+ }
+
+ // These fields store the flag values.
+ bool also_run_disabled_tests;
+ bool break_on_failure;
+ bool catch_exceptions;
+ bool death_test_use_fork;
+ const char* filter;
+ bool list_tests;
+ const char* output;
+ bool print_time;
+ Int32 random_seed;
+ Int32 repeat;
+ bool shuffle;
+ Int32 stack_trace_depth;
+ const char* stream_result_to;
+ bool throw_on_failure;
+};
+
+// Fixture for testing InitGoogleTest().
+class InitGoogleTestTest : public Test {
+ protected:
+ // Clears the flags before each test.
+ virtual void SetUp() {
+ GTEST_FLAG(also_run_disabled_tests) = false;
+ GTEST_FLAG(break_on_failure) = false;
+ GTEST_FLAG(catch_exceptions) = false;
+ GTEST_FLAG(death_test_use_fork) = false;
+ GTEST_FLAG(filter) = "";
+ GTEST_FLAG(list_tests) = false;
+ GTEST_FLAG(output) = "";
+ GTEST_FLAG(print_time) = true;
+ GTEST_FLAG(random_seed) = 0;
+ GTEST_FLAG(repeat) = 1;
+ GTEST_FLAG(shuffle) = false;
+ GTEST_FLAG(stack_trace_depth) = kMaxStackTraceDepth;
+ GTEST_FLAG(stream_result_to) = "";
+ GTEST_FLAG(throw_on_failure) = false;
+ }
+
+ // Asserts that two narrow or wide string arrays are equal.
+ template <typename CharType>
+ static void AssertStringArrayEq(size_t size1, CharType** array1,
+ size_t size2, CharType** array2) {
+ ASSERT_EQ(size1, size2) << " Array sizes different.";
+
+ for (size_t i = 0; i != size1; i++) {
+ ASSERT_STREQ(array1[i], array2[i]) << " where i == " << i;
+ }
+ }
+
+ // Verifies that the flag values match the expected values.
+ static void CheckFlags(const Flags& expected) {
+ EXPECT_EQ(expected.also_run_disabled_tests,
+ GTEST_FLAG(also_run_disabled_tests));
+ EXPECT_EQ(expected.break_on_failure, GTEST_FLAG(break_on_failure));
+ EXPECT_EQ(expected.catch_exceptions, GTEST_FLAG(catch_exceptions));
+ EXPECT_EQ(expected.death_test_use_fork, GTEST_FLAG(death_test_use_fork));
+ EXPECT_STREQ(expected.filter, GTEST_FLAG(filter).c_str());
+ EXPECT_EQ(expected.list_tests, GTEST_FLAG(list_tests));
+ EXPECT_STREQ(expected.output, GTEST_FLAG(output).c_str());
+ EXPECT_EQ(expected.print_time, GTEST_FLAG(print_time));
+ EXPECT_EQ(expected.random_seed, GTEST_FLAG(random_seed));
+ EXPECT_EQ(expected.repeat, GTEST_FLAG(repeat));
+ EXPECT_EQ(expected.shuffle, GTEST_FLAG(shuffle));
+ EXPECT_EQ(expected.stack_trace_depth, GTEST_FLAG(stack_trace_depth));
+ EXPECT_STREQ(expected.stream_result_to,
+ GTEST_FLAG(stream_result_to).c_str());
+ EXPECT_EQ(expected.throw_on_failure, GTEST_FLAG(throw_on_failure));
+ }
+
+ // Parses a command line (specified by argc1 and argv1), then
+ // verifies that the flag values are expected and that the
+ // recognized flags are removed from the command line.
+ template <typename CharType>
+ static void TestParsingFlags(int argc1, const CharType** argv1,
+ int argc2, const CharType** argv2,
+ const Flags& expected, bool should_print_help) {
+ const bool saved_help_flag = ::testing::internal::g_help_flag;
+ ::testing::internal::g_help_flag = false;
+
+#if GTEST_HAS_STREAM_REDIRECTION
+ CaptureStdout();
+#endif
+
+ // Parses the command line.
+ internal::ParseGoogleTestFlagsOnly(&argc1, const_cast<CharType**>(argv1));
+
+#if GTEST_HAS_STREAM_REDIRECTION
+ const String captured_stdout = GetCapturedStdout();
+#endif
+
+ // Verifies the flag values.
+ CheckFlags(expected);
+
+ // Verifies that the recognized flags are removed from the command
+ // line.
+ AssertStringArrayEq(argc1 + 1, argv1, argc2 + 1, argv2);
+
+ // ParseGoogleTestFlagsOnly should neither set g_help_flag nor print the
+ // help message for the flags it recognizes.
+ EXPECT_EQ(should_print_help, ::testing::internal::g_help_flag);
+
+#if GTEST_HAS_STREAM_REDIRECTION
+ const char* const expected_help_fragment =
+ "This program contains tests written using";
+ if (should_print_help) {
+ EXPECT_PRED_FORMAT2(IsSubstring, expected_help_fragment, captured_stdout);
+ } else {
+ EXPECT_PRED_FORMAT2(IsNotSubstring,
+ expected_help_fragment, captured_stdout);
+ }
+#endif // GTEST_HAS_STREAM_REDIRECTION
+
+ ::testing::internal::g_help_flag = saved_help_flag;
+ }
+
+ // This macro wraps TestParsingFlags s.t. the user doesn't need
+ // to specify the array sizes.
+
+#define GTEST_TEST_PARSING_FLAGS_(argv1, argv2, expected, should_print_help) \
+ TestParsingFlags(sizeof(argv1)/sizeof(*argv1) - 1, argv1, \
+ sizeof(argv2)/sizeof(*argv2) - 1, argv2, \
+ expected, should_print_help)
+};
+
+// Tests parsing an empty command line.
+TEST_F(InitGoogleTestTest, Empty) {
+ const char* argv[] = {
+ NULL
+ };
+
+ const char* argv2[] = {
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags(), false);
+}
+
+// Tests parsing a command line that has no flag.
+TEST_F(InitGoogleTestTest, NoFlag) {
+ const char* argv[] = {
+ "foo.exe",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags(), false);
+}
+
+// Tests parsing a bad --gtest_filter flag.
+TEST_F(InitGoogleTestTest, FilterBad) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_filter",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ "--gtest_filter",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Filter(""), true);
+}
+
+// Tests parsing an empty --gtest_filter flag.
+TEST_F(InitGoogleTestTest, FilterEmpty) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_filter=",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Filter(""), false);
+}
+
+// Tests parsing a non-empty --gtest_filter flag.
+TEST_F(InitGoogleTestTest, FilterNonEmpty) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_filter=abc",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Filter("abc"), false);
+}
+
+// Tests parsing --gtest_break_on_failure.
+TEST_F(InitGoogleTestTest, BreakOnFailureWithoutValue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_break_on_failure",
+ NULL
+};
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::BreakOnFailure(true), false);
+}
+
+// Tests parsing --gtest_break_on_failure=0.
+TEST_F(InitGoogleTestTest, BreakOnFailureFalse_0) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_break_on_failure=0",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::BreakOnFailure(false), false);
+}
+
+// Tests parsing --gtest_break_on_failure=f.
+TEST_F(InitGoogleTestTest, BreakOnFailureFalse_f) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_break_on_failure=f",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::BreakOnFailure(false), false);
+}
+
+// Tests parsing --gtest_break_on_failure=F.
+TEST_F(InitGoogleTestTest, BreakOnFailureFalse_F) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_break_on_failure=F",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::BreakOnFailure(false), false);
+}
+
+// Tests parsing a --gtest_break_on_failure flag that has a "true"
+// definition.
+TEST_F(InitGoogleTestTest, BreakOnFailureTrue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_break_on_failure=1",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::BreakOnFailure(true), false);
+}
+
+// Tests parsing --gtest_catch_exceptions.
+TEST_F(InitGoogleTestTest, CatchExceptions) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_catch_exceptions",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::CatchExceptions(true), false);
+}
+
+// Tests parsing --gtest_death_test_use_fork.
+TEST_F(InitGoogleTestTest, DeathTestUseFork) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_death_test_use_fork",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::DeathTestUseFork(true), false);
+}
+
+// Tests having the same flag twice with different values. The
+// expected behavior is that the one coming last takes precedence.
+TEST_F(InitGoogleTestTest, DuplicatedFlags) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_filter=a",
+ "--gtest_filter=b",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Filter("b"), false);
+}
+
+// Tests having an unrecognized flag on the command line.
+TEST_F(InitGoogleTestTest, UnrecognizedFlag) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_break_on_failure",
+ "bar", // Unrecognized by Google Test.
+ "--gtest_filter=b",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ "bar",
+ NULL
+ };
+
+ Flags flags;
+ flags.break_on_failure = true;
+ flags.filter = "b";
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, flags, false);
+}
+
+// Tests having a --gtest_list_tests flag
+TEST_F(InitGoogleTestTest, ListTestsFlag) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_list_tests",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::ListTests(true), false);
+}
+
+// Tests having a --gtest_list_tests flag with a "true" value
+TEST_F(InitGoogleTestTest, ListTestsTrue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_list_tests=1",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::ListTests(true), false);
+}
+
+// Tests having a --gtest_list_tests flag with a "false" value
+TEST_F(InitGoogleTestTest, ListTestsFalse) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_list_tests=0",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::ListTests(false), false);
+}
+
+// Tests parsing --gtest_list_tests=f.
+TEST_F(InitGoogleTestTest, ListTestsFalse_f) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_list_tests=f",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::ListTests(false), false);
+}
+
+// Tests parsing --gtest_list_tests=F.
+TEST_F(InitGoogleTestTest, ListTestsFalse_F) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_list_tests=F",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::ListTests(false), false);
+}
+
+// Tests parsing --gtest_output (invalid).
+TEST_F(InitGoogleTestTest, OutputEmpty) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_output",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ "--gtest_output",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags(), true);
+}
+
+// Tests parsing --gtest_output=xml
+TEST_F(InitGoogleTestTest, OutputXml) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_output=xml",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Output("xml"), false);
+}
+
+// Tests parsing --gtest_output=xml:file
+TEST_F(InitGoogleTestTest, OutputXmlFile) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_output=xml:file",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Output("xml:file"), false);
+}
+
+// Tests parsing --gtest_output=xml:directory/path/
+TEST_F(InitGoogleTestTest, OutputXmlDirectory) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_output=xml:directory/path/",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2,
+ Flags::Output("xml:directory/path/"), false);
+}
+
+// Tests having a --gtest_print_time flag
+TEST_F(InitGoogleTestTest, PrintTimeFlag) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_print_time",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::PrintTime(true), false);
+}
+
+// Tests having a --gtest_print_time flag with a "true" value
+TEST_F(InitGoogleTestTest, PrintTimeTrue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_print_time=1",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::PrintTime(true), false);
+}
+
+// Tests having a --gtest_print_time flag with a "false" value
+TEST_F(InitGoogleTestTest, PrintTimeFalse) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_print_time=0",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::PrintTime(false), false);
+}
+
+// Tests parsing --gtest_print_time=f.
+TEST_F(InitGoogleTestTest, PrintTimeFalse_f) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_print_time=f",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::PrintTime(false), false);
+}
+
+// Tests parsing --gtest_print_time=F.
+TEST_F(InitGoogleTestTest, PrintTimeFalse_F) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_print_time=F",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::PrintTime(false), false);
+}
+
+// Tests parsing --gtest_random_seed=number
+TEST_F(InitGoogleTestTest, RandomSeed) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_random_seed=1000",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::RandomSeed(1000), false);
+}
+
+// Tests parsing --gtest_repeat=number
+TEST_F(InitGoogleTestTest, Repeat) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_repeat=1000",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Repeat(1000), false);
+}
+
+// Tests having a --gtest_also_run_disabled_tests flag
+TEST_F(InitGoogleTestTest, AlsoRunDisabledTestsFlag) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_also_run_disabled_tests",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2,
+ Flags::AlsoRunDisabledTests(true), false);
+}
+
+// Tests having a --gtest_also_run_disabled_tests flag with a "true" value
+TEST_F(InitGoogleTestTest, AlsoRunDisabledTestsTrue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_also_run_disabled_tests=1",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2,
+ Flags::AlsoRunDisabledTests(true), false);
+}
+
+// Tests having a --gtest_also_run_disabled_tests flag with a "false" value
+TEST_F(InitGoogleTestTest, AlsoRunDisabledTestsFalse) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_also_run_disabled_tests=0",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2,
+ Flags::AlsoRunDisabledTests(false), false);
+}
+
+// Tests parsing --gtest_shuffle.
+TEST_F(InitGoogleTestTest, ShuffleWithoutValue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_shuffle",
+ NULL
+};
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Shuffle(true), false);
+}
+
+// Tests parsing --gtest_shuffle=0.
+TEST_F(InitGoogleTestTest, ShuffleFalse_0) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_shuffle=0",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Shuffle(false), false);
+}
+
+// Tests parsing a --gtest_shuffle flag that has a "true"
+// definition.
+TEST_F(InitGoogleTestTest, ShuffleTrue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_shuffle=1",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::Shuffle(true), false);
+}
+
+// Tests parsing --gtest_stack_trace_depth=number.
+TEST_F(InitGoogleTestTest, StackTraceDepth) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_stack_trace_depth=5",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::StackTraceDepth(5), false);
+}
+
+TEST_F(InitGoogleTestTest, StreamResultTo) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_stream_result_to=localhost:1234",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(
+ argv, argv2, Flags::StreamResultTo("localhost:1234"), false);
+}
+
+// Tests parsing --gtest_throw_on_failure.
+TEST_F(InitGoogleTestTest, ThrowOnFailureWithoutValue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_throw_on_failure",
+ NULL
+};
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::ThrowOnFailure(true), false);
+}
+
+// Tests parsing --gtest_throw_on_failure=0.
+TEST_F(InitGoogleTestTest, ThrowOnFailureFalse_0) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_throw_on_failure=0",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::ThrowOnFailure(false), false);
+}
+
+// Tests parsing a --gtest_throw_on_failure flag that has a "true"
+// definition.
+TEST_F(InitGoogleTestTest, ThrowOnFailureTrue) {
+ const char* argv[] = {
+ "foo.exe",
+ "--gtest_throw_on_failure=1",
+ NULL
+ };
+
+ const char* argv2[] = {
+ "foo.exe",
+ NULL
+ };
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, Flags::ThrowOnFailure(true), false);
+}
+
+#if GTEST_OS_WINDOWS
+// Tests parsing wide strings.
+TEST_F(InitGoogleTestTest, WideStrings) {
+ const wchar_t* argv[] = {
+ L"foo.exe",
+ L"--gtest_filter=Foo*",
+ L"--gtest_list_tests=1",
+ L"--gtest_break_on_failure",
+ L"--non_gtest_flag",
+ NULL
+ };
+
+ const wchar_t* argv2[] = {
+ L"foo.exe",
+ L"--non_gtest_flag",
+ NULL
+ };
+
+ Flags expected_flags;
+ expected_flags.break_on_failure = true;
+ expected_flags.filter = "Foo*";
+ expected_flags.list_tests = true;
+
+ GTEST_TEST_PARSING_FLAGS_(argv, argv2, expected_flags, false);
+}
+#endif // GTEST_OS_WINDOWS
+
+// Tests current_test_info() in UnitTest.
+class CurrentTestInfoTest : public Test {
+ protected:
+ // Tests that current_test_info() returns NULL before the first test in
+ // the test case is run.
+ static void SetUpTestCase() {
+ // There should be no tests running at this point.
+ const TestInfo* test_info =
+ UnitTest::GetInstance()->current_test_info();
+ EXPECT_TRUE(test_info == NULL)
+ << "There should be no tests running at this point.";
+ }
+
+ // Tests that current_test_info() returns NULL after the last test in
+ // the test case has run.
+ static void TearDownTestCase() {
+ const TestInfo* test_info =
+ UnitTest::GetInstance()->current_test_info();
+ EXPECT_TRUE(test_info == NULL)
+ << "There should be no tests running at this point.";
+ }
+};
+
+// Tests that current_test_info() returns TestInfo for currently running
+// test by checking the expected test name against the actual one.
+TEST_F(CurrentTestInfoTest, WorksForFirstTestInATestCase) {
+ const TestInfo* test_info =
+ UnitTest::GetInstance()->current_test_info();
+ ASSERT_TRUE(NULL != test_info)
+ << "There is a test running so we should have a valid TestInfo.";
+ EXPECT_STREQ("CurrentTestInfoTest", test_info->test_case_name())
+ << "Expected the name of the currently running test case.";
+ EXPECT_STREQ("WorksForFirstTestInATestCase", test_info->name())
+ << "Expected the name of the currently running test.";
+}
+
+// Tests that current_test_info() returns TestInfo for currently running
+// test by checking the expected test name against the actual one. We
+// use this test to see that the TestInfo object actually changed from
+// the previous invocation.
+TEST_F(CurrentTestInfoTest, WorksForSecondTestInATestCase) {
+ const TestInfo* test_info =
+ UnitTest::GetInstance()->current_test_info();
+ ASSERT_TRUE(NULL != test_info)
+ << "There is a test running so we should have a valid TestInfo.";
+ EXPECT_STREQ("CurrentTestInfoTest", test_info->test_case_name())
+ << "Expected the name of the currently running test case.";
+ EXPECT_STREQ("WorksForSecondTestInATestCase", test_info->name())
+ << "Expected the name of the currently running test.";
+}
+
+} // namespace testing
+
+// These two lines test that we can define tests in a namespace that
+// has the name "testing" and is nested in another namespace.
+namespace my_namespace {
+namespace testing {
+
+// Makes sure that TEST knows to use ::testing::Test instead of
+// ::my_namespace::testing::Test.
+class Test {};
+
+// Makes sure that an assertion knows to use ::testing::Message instead of
+// ::my_namespace::testing::Message.
+class Message {};
+
+// Makes sure that an assertion knows to use
+// ::testing::AssertionResult instead of
+// ::my_namespace::testing::AssertionResult.
+class AssertionResult {};
+
+// Tests that an assertion that should succeed works as expected.
+TEST(NestedTestingNamespaceTest, Success) {
+ EXPECT_EQ(1, 1) << "This shouldn't fail.";
+}
+
+// Tests that an assertion that should fail works as expected.
+TEST(NestedTestingNamespaceTest, Failure) {
+ EXPECT_FATAL_FAILURE(FAIL() << "This failure is expected.",
+ "This failure is expected.");
+}
+
+} // namespace testing
+} // namespace my_namespace
+
+// Tests that one can call superclass SetUp and TearDown methods--
+// that is, that they are not private.
+// No tests are based on this fixture; the test "passes" if it compiles
+// successfully.
+class ProtectedFixtureMethodsTest : public Test {
+ protected:
+ virtual void SetUp() {
+ Test::SetUp();
+ }
+ virtual void TearDown() {
+ Test::TearDown();
+ }
+};
+
+// StreamingAssertionsTest tests the streaming versions of a representative
+// sample of assertions.
+TEST(StreamingAssertionsTest, Unconditional) {
+ SUCCEED() << "expected success";
+ EXPECT_NONFATAL_FAILURE(ADD_FAILURE() << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(FAIL() << "expected failure",
+ "expected failure");
+}
+
+#ifdef __BORLANDC__
+// Silences warnings: "Condition is always true", "Unreachable code"
+# pragma option push -w-ccc -w-rch
+#endif
+
+TEST(StreamingAssertionsTest, Truth) {
+ EXPECT_TRUE(true) << "unexpected failure";
+ ASSERT_TRUE(true) << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_TRUE(false) << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_TRUE(false) << "expected failure",
+ "expected failure");
+}
+
+TEST(StreamingAssertionsTest, Truth2) {
+ EXPECT_FALSE(false) << "unexpected failure";
+ ASSERT_FALSE(false) << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_FALSE(true) << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_FALSE(true) << "expected failure",
+ "expected failure");
+}
+
+#ifdef __BORLANDC__
+// Restores warnings after previous "#pragma option push" supressed them
+# pragma option pop
+#endif
+
+TEST(StreamingAssertionsTest, IntegerEquals) {
+ EXPECT_EQ(1, 1) << "unexpected failure";
+ ASSERT_EQ(1, 1) << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_EQ(1, 2) << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_EQ(1, 2) << "expected failure",
+ "expected failure");
+}
+
+TEST(StreamingAssertionsTest, IntegerLessThan) {
+ EXPECT_LT(1, 2) << "unexpected failure";
+ ASSERT_LT(1, 2) << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_LT(2, 1) << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_LT(2, 1) << "expected failure",
+ "expected failure");
+}
+
+TEST(StreamingAssertionsTest, StringsEqual) {
+ EXPECT_STREQ("foo", "foo") << "unexpected failure";
+ ASSERT_STREQ("foo", "foo") << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_STREQ("foo", "bar") << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_STREQ("foo", "bar") << "expected failure",
+ "expected failure");
+}
+
+TEST(StreamingAssertionsTest, StringsNotEqual) {
+ EXPECT_STRNE("foo", "bar") << "unexpected failure";
+ ASSERT_STRNE("foo", "bar") << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_STRNE("foo", "foo") << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_STRNE("foo", "foo") << "expected failure",
+ "expected failure");
+}
+
+TEST(StreamingAssertionsTest, StringsEqualIgnoringCase) {
+ EXPECT_STRCASEEQ("foo", "FOO") << "unexpected failure";
+ ASSERT_STRCASEEQ("foo", "FOO") << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_STRCASEEQ("foo", "bar") << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_STRCASEEQ("foo", "bar") << "expected failure",
+ "expected failure");
+}
+
+TEST(StreamingAssertionsTest, StringNotEqualIgnoringCase) {
+ EXPECT_STRCASENE("foo", "bar") << "unexpected failure";
+ ASSERT_STRCASENE("foo", "bar") << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_STRCASENE("foo", "FOO") << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_STRCASENE("bar", "BAR") << "expected failure",
+ "expected failure");
+}
+
+TEST(StreamingAssertionsTest, FloatingPointEquals) {
+ EXPECT_FLOAT_EQ(1.0, 1.0) << "unexpected failure";
+ ASSERT_FLOAT_EQ(1.0, 1.0) << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_FLOAT_EQ(0.0, 1.0) << "expected failure",
+ "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_FLOAT_EQ(0.0, 1.0) << "expected failure",
+ "expected failure");
+}
+
+#if GTEST_HAS_EXCEPTIONS
+
+TEST(StreamingAssertionsTest, Throw) {
+ EXPECT_THROW(ThrowAnInteger(), int) << "unexpected failure";
+ ASSERT_THROW(ThrowAnInteger(), int) << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_THROW(ThrowAnInteger(), bool) <<
+ "expected failure", "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_THROW(ThrowAnInteger(), bool) <<
+ "expected failure", "expected failure");
+}
+
+TEST(StreamingAssertionsTest, NoThrow) {
+ EXPECT_NO_THROW(ThrowNothing()) << "unexpected failure";
+ ASSERT_NO_THROW(ThrowNothing()) << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_NO_THROW(ThrowAnInteger()) <<
+ "expected failure", "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_NO_THROW(ThrowAnInteger()) <<
+ "expected failure", "expected failure");
+}
+
+TEST(StreamingAssertionsTest, AnyThrow) {
+ EXPECT_ANY_THROW(ThrowAnInteger()) << "unexpected failure";
+ ASSERT_ANY_THROW(ThrowAnInteger()) << "unexpected failure";
+ EXPECT_NONFATAL_FAILURE(EXPECT_ANY_THROW(ThrowNothing()) <<
+ "expected failure", "expected failure");
+ EXPECT_FATAL_FAILURE(ASSERT_ANY_THROW(ThrowNothing()) <<
+ "expected failure", "expected failure");
+}
+
+#endif // GTEST_HAS_EXCEPTIONS
+
+// Tests that Google Test correctly decides whether to use colors in the output.
+
+TEST(ColoredOutputTest, UsesColorsWhenGTestColorFlagIsYes) {
+ GTEST_FLAG(color) = "yes";
+
+ SetEnv("TERM", "xterm"); // TERM supports colors.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+ EXPECT_TRUE(ShouldUseColor(false)); // Stdout is not a TTY.
+
+ SetEnv("TERM", "dumb"); // TERM doesn't support colors.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+ EXPECT_TRUE(ShouldUseColor(false)); // Stdout is not a TTY.
+}
+
+TEST(ColoredOutputTest, UsesColorsWhenGTestColorFlagIsAliasOfYes) {
+ SetEnv("TERM", "dumb"); // TERM doesn't support colors.
+
+ GTEST_FLAG(color) = "True";
+ EXPECT_TRUE(ShouldUseColor(false)); // Stdout is not a TTY.
+
+ GTEST_FLAG(color) = "t";
+ EXPECT_TRUE(ShouldUseColor(false)); // Stdout is not a TTY.
+
+ GTEST_FLAG(color) = "1";
+ EXPECT_TRUE(ShouldUseColor(false)); // Stdout is not a TTY.
+}
+
+TEST(ColoredOutputTest, UsesNoColorWhenGTestColorFlagIsNo) {
+ GTEST_FLAG(color) = "no";
+
+ SetEnv("TERM", "xterm"); // TERM supports colors.
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+ EXPECT_FALSE(ShouldUseColor(false)); // Stdout is not a TTY.
+
+ SetEnv("TERM", "dumb"); // TERM doesn't support colors.
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+ EXPECT_FALSE(ShouldUseColor(false)); // Stdout is not a TTY.
+}
+
+TEST(ColoredOutputTest, UsesNoColorWhenGTestColorFlagIsInvalid) {
+ SetEnv("TERM", "xterm"); // TERM supports colors.
+
+ GTEST_FLAG(color) = "F";
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ GTEST_FLAG(color) = "0";
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ GTEST_FLAG(color) = "unknown";
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+}
+
+TEST(ColoredOutputTest, UsesColorsWhenStdoutIsTty) {
+ GTEST_FLAG(color) = "auto";
+
+ SetEnv("TERM", "xterm"); // TERM supports colors.
+ EXPECT_FALSE(ShouldUseColor(false)); // Stdout is not a TTY.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+}
+
+TEST(ColoredOutputTest, UsesColorsWhenTermSupportsColors) {
+ GTEST_FLAG(color) = "auto";
+
+#if GTEST_OS_WINDOWS
+ // On Windows, we ignore the TERM variable as it's usually not set.
+
+ SetEnv("TERM", "dumb");
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "");
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "xterm");
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+#else
+ // On non-Windows platforms, we rely on TERM to determine if the
+ // terminal supports colors.
+
+ SetEnv("TERM", "dumb"); // TERM doesn't support colors.
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "emacs"); // TERM doesn't support colors.
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "vt100"); // TERM doesn't support colors.
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "xterm-mono"); // TERM doesn't support colors.
+ EXPECT_FALSE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "xterm"); // TERM supports colors.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "xterm-color"); // TERM supports colors.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "xterm-256color"); // TERM supports colors.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "screen"); // TERM supports colors.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "linux"); // TERM supports colors.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+
+ SetEnv("TERM", "cygwin"); // TERM supports colors.
+ EXPECT_TRUE(ShouldUseColor(true)); // Stdout is a TTY.
+#endif // GTEST_OS_WINDOWS
+}
+
+// Verifies that StaticAssertTypeEq works in a namespace scope.
+
+static bool dummy1 GTEST_ATTRIBUTE_UNUSED_ = StaticAssertTypeEq<bool, bool>();
+static bool dummy2 GTEST_ATTRIBUTE_UNUSED_ =
+ StaticAssertTypeEq<const int, const int>();
+
+// Verifies that StaticAssertTypeEq works in a class.
+
+template <typename T>
+class StaticAssertTypeEqTestHelper {
+ public:
+ StaticAssertTypeEqTestHelper() { StaticAssertTypeEq<bool, T>(); }
+};
+
+TEST(StaticAssertTypeEqTest, WorksInClass) {
+ StaticAssertTypeEqTestHelper<bool>();
+}
+
+// Verifies that StaticAssertTypeEq works inside a function.
+
+typedef int IntAlias;
+
+TEST(StaticAssertTypeEqTest, CompilesForEqualTypes) {
+ StaticAssertTypeEq<int, IntAlias>();
+ StaticAssertTypeEq<int*, IntAlias*>();
+}
+
+TEST(GetCurrentOsStackTraceExceptTopTest, ReturnsTheStackTrace) {
+ testing::UnitTest* const unit_test = testing::UnitTest::GetInstance();
+
+ // We don't have a stack walker in Google Test yet.
+ EXPECT_STREQ("", GetCurrentOsStackTraceExceptTop(unit_test, 0).c_str());
+ EXPECT_STREQ("", GetCurrentOsStackTraceExceptTop(unit_test, 1).c_str());
+}
+
+TEST(HasNonfatalFailureTest, ReturnsFalseWhenThereIsNoFailure) {
+ EXPECT_FALSE(HasNonfatalFailure());
+}
+
+static void FailFatally() { FAIL(); }
+
+TEST(HasNonfatalFailureTest, ReturnsFalseWhenThereIsOnlyFatalFailure) {
+ FailFatally();
+ const bool has_nonfatal_failure = HasNonfatalFailure();
+ ClearCurrentTestPartResults();
+ EXPECT_FALSE(has_nonfatal_failure);
+}
+
+TEST(HasNonfatalFailureTest, ReturnsTrueWhenThereIsNonfatalFailure) {
+ ADD_FAILURE();
+ const bool has_nonfatal_failure = HasNonfatalFailure();
+ ClearCurrentTestPartResults();
+ EXPECT_TRUE(has_nonfatal_failure);
+}
+
+TEST(HasNonfatalFailureTest, ReturnsTrueWhenThereAreFatalAndNonfatalFailures) {
+ FailFatally();
+ ADD_FAILURE();
+ const bool has_nonfatal_failure = HasNonfatalFailure();
+ ClearCurrentTestPartResults();
+ EXPECT_TRUE(has_nonfatal_failure);
+}
+
+// A wrapper for calling HasNonfatalFailure outside of a test body.
+static bool HasNonfatalFailureHelper() {
+ return testing::Test::HasNonfatalFailure();
+}
+
+TEST(HasNonfatalFailureTest, WorksOutsideOfTestBody) {
+ EXPECT_FALSE(HasNonfatalFailureHelper());
+}
+
+TEST(HasNonfatalFailureTest, WorksOutsideOfTestBody2) {
+ ADD_FAILURE();
+ const bool has_nonfatal_failure = HasNonfatalFailureHelper();
+ ClearCurrentTestPartResults();
+ EXPECT_TRUE(has_nonfatal_failure);
+}
+
+TEST(HasFailureTest, ReturnsFalseWhenThereIsNoFailure) {
+ EXPECT_FALSE(HasFailure());
+}
+
+TEST(HasFailureTest, ReturnsTrueWhenThereIsFatalFailure) {
+ FailFatally();
+ const bool has_failure = HasFailure();
+ ClearCurrentTestPartResults();
+ EXPECT_TRUE(has_failure);
+}
+
+TEST(HasFailureTest, ReturnsTrueWhenThereIsNonfatalFailure) {
+ ADD_FAILURE();
+ const bool has_failure = HasFailure();
+ ClearCurrentTestPartResults();
+ EXPECT_TRUE(has_failure);
+}
+
+TEST(HasFailureTest, ReturnsTrueWhenThereAreFatalAndNonfatalFailures) {
+ FailFatally();
+ ADD_FAILURE();
+ const bool has_failure = HasFailure();
+ ClearCurrentTestPartResults();
+ EXPECT_TRUE(has_failure);
+}
+
+// A wrapper for calling HasFailure outside of a test body.
+static bool HasFailureHelper() { return testing::Test::HasFailure(); }
+
+TEST(HasFailureTest, WorksOutsideOfTestBody) {
+ EXPECT_FALSE(HasFailureHelper());
+}
+
+TEST(HasFailureTest, WorksOutsideOfTestBody2) {
+ ADD_FAILURE();
+ const bool has_failure = HasFailureHelper();
+ ClearCurrentTestPartResults();
+ EXPECT_TRUE(has_failure);
+}
+
+class TestListener : public EmptyTestEventListener {
+ public:
+ TestListener() : on_start_counter_(NULL), is_destroyed_(NULL) {}
+ TestListener(int* on_start_counter, bool* is_destroyed)
+ : on_start_counter_(on_start_counter),
+ is_destroyed_(is_destroyed) {}
+
+ virtual ~TestListener() {
+ if (is_destroyed_)
+ *is_destroyed_ = true;
+ }
+
+ protected:
+ virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {
+ if (on_start_counter_ != NULL)
+ (*on_start_counter_)++;
+ }
+
+ private:
+ int* on_start_counter_;
+ bool* is_destroyed_;
+};
+
+// Tests the constructor.
+TEST(TestEventListenersTest, ConstructionWorks) {
+ TestEventListeners listeners;
+
+ EXPECT_TRUE(TestEventListenersAccessor::GetRepeater(&listeners) != NULL);
+ EXPECT_TRUE(listeners.default_result_printer() == NULL);
+ EXPECT_TRUE(listeners.default_xml_generator() == NULL);
+}
+
+// Tests that the TestEventListeners destructor deletes all the listeners it
+// owns.
+TEST(TestEventListenersTest, DestructionWorks) {
+ bool default_result_printer_is_destroyed = false;
+ bool default_xml_printer_is_destroyed = false;
+ bool extra_listener_is_destroyed = false;
+ TestListener* default_result_printer = new TestListener(
+ NULL, &default_result_printer_is_destroyed);
+ TestListener* default_xml_printer = new TestListener(
+ NULL, &default_xml_printer_is_destroyed);
+ TestListener* extra_listener = new TestListener(
+ NULL, &extra_listener_is_destroyed);
+
+ {
+ TestEventListeners listeners;
+ TestEventListenersAccessor::SetDefaultResultPrinter(&listeners,
+ default_result_printer);
+ TestEventListenersAccessor::SetDefaultXmlGenerator(&listeners,
+ default_xml_printer);
+ listeners.Append(extra_listener);
+ }
+ EXPECT_TRUE(default_result_printer_is_destroyed);
+ EXPECT_TRUE(default_xml_printer_is_destroyed);
+ EXPECT_TRUE(extra_listener_is_destroyed);
+}
+
+// Tests that a listener Append'ed to a TestEventListeners list starts
+// receiving events.
+TEST(TestEventListenersTest, Append) {
+ int on_start_counter = 0;
+ bool is_destroyed = false;
+ TestListener* listener = new TestListener(&on_start_counter, &is_destroyed);
+ {
+ TestEventListeners listeners;
+ listeners.Append(listener);
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+ EXPECT_EQ(1, on_start_counter);
+ }
+ EXPECT_TRUE(is_destroyed);
+}
+
+// Tests that listeners receive events in the order they were appended to
+// the list, except for *End requests, which must be received in the reverse
+// order.
+class SequenceTestingListener : public EmptyTestEventListener {
+ public:
+ SequenceTestingListener(std::vector<String>* vector, const char* id)
+ : vector_(vector), id_(id) {}
+
+ protected:
+ virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {
+ vector_->push_back(GetEventDescription("OnTestProgramStart"));
+ }
+
+ virtual void OnTestProgramEnd(const UnitTest& /*unit_test*/) {
+ vector_->push_back(GetEventDescription("OnTestProgramEnd"));
+ }
+
+ virtual void OnTestIterationStart(const UnitTest& /*unit_test*/,
+ int /*iteration*/) {
+ vector_->push_back(GetEventDescription("OnTestIterationStart"));
+ }
+
+ virtual void OnTestIterationEnd(const UnitTest& /*unit_test*/,
+ int /*iteration*/) {
+ vector_->push_back(GetEventDescription("OnTestIterationEnd"));
+ }
+
+ private:
+ String GetEventDescription(const char* method) {
+ Message message;
+ message << id_ << "." << method;
+ return message.GetString();
+ }
+
+ std::vector<String>* vector_;
+ const char* const id_;
+
+ GTEST_DISALLOW_COPY_AND_ASSIGN_(SequenceTestingListener);
+};
+
+TEST(EventListenerTest, AppendKeepsOrder) {
+ std::vector<String> vec;
+ TestEventListeners listeners;
+ listeners.Append(new SequenceTestingListener(&vec, "1st"));
+ listeners.Append(new SequenceTestingListener(&vec, "2nd"));
+ listeners.Append(new SequenceTestingListener(&vec, "3rd"));
+
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+ ASSERT_EQ(3U, vec.size());
+ EXPECT_STREQ("1st.OnTestProgramStart", vec[0].c_str());
+ EXPECT_STREQ("2nd.OnTestProgramStart", vec[1].c_str());
+ EXPECT_STREQ("3rd.OnTestProgramStart", vec[2].c_str());
+
+ vec.clear();
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramEnd(
+ *UnitTest::GetInstance());
+ ASSERT_EQ(3U, vec.size());
+ EXPECT_STREQ("3rd.OnTestProgramEnd", vec[0].c_str());
+ EXPECT_STREQ("2nd.OnTestProgramEnd", vec[1].c_str());
+ EXPECT_STREQ("1st.OnTestProgramEnd", vec[2].c_str());
+
+ vec.clear();
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestIterationStart(
+ *UnitTest::GetInstance(), 0);
+ ASSERT_EQ(3U, vec.size());
+ EXPECT_STREQ("1st.OnTestIterationStart", vec[0].c_str());
+ EXPECT_STREQ("2nd.OnTestIterationStart", vec[1].c_str());
+ EXPECT_STREQ("3rd.OnTestIterationStart", vec[2].c_str());
+
+ vec.clear();
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestIterationEnd(
+ *UnitTest::GetInstance(), 0);
+ ASSERT_EQ(3U, vec.size());
+ EXPECT_STREQ("3rd.OnTestIterationEnd", vec[0].c_str());
+ EXPECT_STREQ("2nd.OnTestIterationEnd", vec[1].c_str());
+ EXPECT_STREQ("1st.OnTestIterationEnd", vec[2].c_str());
+}
+
+// Tests that a listener removed from a TestEventListeners list stops receiving
+// events and is not deleted when the list is destroyed.
+TEST(TestEventListenersTest, Release) {
+ int on_start_counter = 0;
+ bool is_destroyed = false;
+ // Although Append passes the ownership of this object to the list,
+ // the following calls release it, and we need to delete it before the
+ // test ends.
+ TestListener* listener = new TestListener(&on_start_counter, &is_destroyed);
+ {
+ TestEventListeners listeners;
+ listeners.Append(listener);
+ EXPECT_EQ(listener, listeners.Release(listener));
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+ EXPECT_TRUE(listeners.Release(listener) == NULL);
+ }
+ EXPECT_EQ(0, on_start_counter);
+ EXPECT_FALSE(is_destroyed);
+ delete listener;
+}
+
+// Tests that no events are forwarded when event forwarding is disabled.
+TEST(EventListenerTest, SuppressEventForwarding) {
+ int on_start_counter = 0;
+ TestListener* listener = new TestListener(&on_start_counter, NULL);
+
+ TestEventListeners listeners;
+ listeners.Append(listener);
+ ASSERT_TRUE(TestEventListenersAccessor::EventForwardingEnabled(listeners));
+ TestEventListenersAccessor::SuppressEventForwarding(&listeners);
+ ASSERT_FALSE(TestEventListenersAccessor::EventForwardingEnabled(listeners));
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+ EXPECT_EQ(0, on_start_counter);
+}
+
+// Tests that events generated by Google Test are not forwarded in
+// death test subprocesses.
+TEST(EventListenerDeathTest, EventsNotForwardedInDeathTestSubprecesses) {
+ EXPECT_DEATH_IF_SUPPORTED({
+ GTEST_CHECK_(TestEventListenersAccessor::EventForwardingEnabled(
+ *GetUnitTestImpl()->listeners())) << "expected failure";},
+ "expected failure");
+}
+
+// Tests that a listener installed via SetDefaultResultPrinter() starts
+// receiving events and is returned via default_result_printer() and that
+// the previous default_result_printer is removed from the list and deleted.
+TEST(EventListenerTest, default_result_printer) {
+ int on_start_counter = 0;
+ bool is_destroyed = false;
+ TestListener* listener = new TestListener(&on_start_counter, &is_destroyed);
+
+ TestEventListeners listeners;
+ TestEventListenersAccessor::SetDefaultResultPrinter(&listeners, listener);
+
+ EXPECT_EQ(listener, listeners.default_result_printer());
+
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+
+ EXPECT_EQ(1, on_start_counter);
+
+ // Replacing default_result_printer with something else should remove it
+ // from the list and destroy it.
+ TestEventListenersAccessor::SetDefaultResultPrinter(&listeners, NULL);
+
+ EXPECT_TRUE(listeners.default_result_printer() == NULL);
+ EXPECT_TRUE(is_destroyed);
+
+ // After broadcasting an event the counter is still the same, indicating
+ // the listener is not in the list anymore.
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+ EXPECT_EQ(1, on_start_counter);
+}
+
+// Tests that the default_result_printer listener stops receiving events
+// when removed via Release and that is not owned by the list anymore.
+TEST(EventListenerTest, RemovingDefaultResultPrinterWorks) {
+ int on_start_counter = 0;
+ bool is_destroyed = false;
+ // Although Append passes the ownership of this object to the list,
+ // the following calls release it, and we need to delete it before the
+ // test ends.
+ TestListener* listener = new TestListener(&on_start_counter, &is_destroyed);
+ {
+ TestEventListeners listeners;
+ TestEventListenersAccessor::SetDefaultResultPrinter(&listeners, listener);
+
+ EXPECT_EQ(listener, listeners.Release(listener));
+ EXPECT_TRUE(listeners.default_result_printer() == NULL);
+ EXPECT_FALSE(is_destroyed);
+
+ // Broadcasting events now should not affect default_result_printer.
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+ EXPECT_EQ(0, on_start_counter);
+ }
+ // Destroying the list should not affect the listener now, too.
+ EXPECT_FALSE(is_destroyed);
+ delete listener;
+}
+
+// Tests that a listener installed via SetDefaultXmlGenerator() starts
+// receiving events and is returned via default_xml_generator() and that
+// the previous default_xml_generator is removed from the list and deleted.
+TEST(EventListenerTest, default_xml_generator) {
+ int on_start_counter = 0;
+ bool is_destroyed = false;
+ TestListener* listener = new TestListener(&on_start_counter, &is_destroyed);
+
+ TestEventListeners listeners;
+ TestEventListenersAccessor::SetDefaultXmlGenerator(&listeners, listener);
+
+ EXPECT_EQ(listener, listeners.default_xml_generator());
+
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+
+ EXPECT_EQ(1, on_start_counter);
+
+ // Replacing default_xml_generator with something else should remove it
+ // from the list and destroy it.
+ TestEventListenersAccessor::SetDefaultXmlGenerator(&listeners, NULL);
+
+ EXPECT_TRUE(listeners.default_xml_generator() == NULL);
+ EXPECT_TRUE(is_destroyed);
+
+ // After broadcasting an event the counter is still the same, indicating
+ // the listener is not in the list anymore.
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+ EXPECT_EQ(1, on_start_counter);
+}
+
+// Tests that the default_xml_generator listener stops receiving events
+// when removed via Release and that is not owned by the list anymore.
+TEST(EventListenerTest, RemovingDefaultXmlGeneratorWorks) {
+ int on_start_counter = 0;
+ bool is_destroyed = false;
+ // Although Append passes the ownership of this object to the list,
+ // the following calls release it, and we need to delete it before the
+ // test ends.
+ TestListener* listener = new TestListener(&on_start_counter, &is_destroyed);
+ {
+ TestEventListeners listeners;
+ TestEventListenersAccessor::SetDefaultXmlGenerator(&listeners, listener);
+
+ EXPECT_EQ(listener, listeners.Release(listener));
+ EXPECT_TRUE(listeners.default_xml_generator() == NULL);
+ EXPECT_FALSE(is_destroyed);
+
+ // Broadcasting events now should not affect default_xml_generator.
+ TestEventListenersAccessor::GetRepeater(&listeners)->OnTestProgramStart(
+ *UnitTest::GetInstance());
+ EXPECT_EQ(0, on_start_counter);
+ }
+ // Destroying the list should not affect the listener now, too.
+ EXPECT_FALSE(is_destroyed);
+ delete listener;
+}
+
+// Sanity tests to ensure that the alternative, verbose spellings of
+// some of the macros work. We don't test them thoroughly as that
+// would be quite involved. Since their implementations are
+// straightforward, and they are rarely used, we'll just rely on the
+// users to tell us when they are broken.
+GTEST_TEST(AlternativeNameTest, Works) { // GTEST_TEST is the same as TEST.
+ GTEST_SUCCEED() << "OK"; // GTEST_SUCCEED is the same as SUCCEED.
+
+ // GTEST_FAIL is the same as FAIL.
+ EXPECT_FATAL_FAILURE(GTEST_FAIL() << "An expected failure",
+ "An expected failure");
+
+ // GTEST_ASSERT_XY is the same as ASSERT_XY.
+
+ GTEST_ASSERT_EQ(0, 0);
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_EQ(0, 1) << "An expected failure",
+ "An expected failure");
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_EQ(1, 0) << "An expected failure",
+ "An expected failure");
+
+ GTEST_ASSERT_NE(0, 1);
+ GTEST_ASSERT_NE(1, 0);
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_NE(0, 0) << "An expected failure",
+ "An expected failure");
+
+ GTEST_ASSERT_LE(0, 0);
+ GTEST_ASSERT_LE(0, 1);
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_LE(1, 0) << "An expected failure",
+ "An expected failure");
+
+ GTEST_ASSERT_LT(0, 1);
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_LT(0, 0) << "An expected failure",
+ "An expected failure");
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_LT(1, 0) << "An expected failure",
+ "An expected failure");
+
+ GTEST_ASSERT_GE(0, 0);
+ GTEST_ASSERT_GE(1, 0);
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_GE(0, 1) << "An expected failure",
+ "An expected failure");
+
+ GTEST_ASSERT_GT(1, 0);
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_GT(0, 1) << "An expected failure",
+ "An expected failure");
+ EXPECT_FATAL_FAILURE(GTEST_ASSERT_GT(1, 1) << "An expected failure",
+ "An expected failure");
+}
+
+// Tests for internal utilities necessary for implementation of the universal
+// printing.
+// TODO(vladl@google.com): Find a better home for them.
+
+class ConversionHelperBase {};
+class ConversionHelperDerived : public ConversionHelperBase {};
+
+// Tests that IsAProtocolMessage<T>::value is a compile-time constant.
+TEST(IsAProtocolMessageTest, ValueIsCompileTimeConstant) {
+ GTEST_COMPILE_ASSERT_(IsAProtocolMessage<ProtocolMessage>::value,
+ const_true);
+ GTEST_COMPILE_ASSERT_(!IsAProtocolMessage<int>::value, const_false);
+}
+
+// Tests that IsAProtocolMessage<T>::value is true when T is
+// proto2::Message or a sub-class of it.
+TEST(IsAProtocolMessageTest, ValueIsTrueWhenTypeIsAProtocolMessage) {
+ EXPECT_TRUE(IsAProtocolMessage< ::proto2::Message>::value);
+ EXPECT_TRUE(IsAProtocolMessage<ProtocolMessage>::value);
+}
+
+// Tests that IsAProtocolMessage<T>::value is false when T is neither
+// ProtocolMessage nor a sub-class of it.
+TEST(IsAProtocolMessageTest, ValueIsFalseWhenTypeIsNotAProtocolMessage) {
+ EXPECT_FALSE(IsAProtocolMessage<int>::value);
+ EXPECT_FALSE(IsAProtocolMessage<const ConversionHelperBase>::value);
+}
+
+// Tests that CompileAssertTypesEqual compiles when the type arguments are
+// equal.
+TEST(CompileAssertTypesEqual, CompilesWhenTypesAreEqual) {
+ CompileAssertTypesEqual<void, void>();
+ CompileAssertTypesEqual<int*, int*>();
+}
+
+// Tests that RemoveReference does not affect non-reference types.
+TEST(RemoveReferenceTest, DoesNotAffectNonReferenceType) {
+ CompileAssertTypesEqual<int, RemoveReference<int>::type>();
+ CompileAssertTypesEqual<const char, RemoveReference<const char>::type>();
+}
+
+// Tests that RemoveReference removes reference from reference types.
+TEST(RemoveReferenceTest, RemovesReference) {
+ CompileAssertTypesEqual<int, RemoveReference<int&>::type>();
+ CompileAssertTypesEqual<const char, RemoveReference<const char&>::type>();
+}
+
+// Tests GTEST_REMOVE_REFERENCE_.
+
+template <typename T1, typename T2>
+void TestGTestRemoveReference() {
+ CompileAssertTypesEqual<T1, GTEST_REMOVE_REFERENCE_(T2)>();
+}
+
+TEST(RemoveReferenceTest, MacroVersion) {
+ TestGTestRemoveReference<int, int>();
+ TestGTestRemoveReference<const char, const char&>();
+}
+
+
+// Tests that RemoveConst does not affect non-const types.
+TEST(RemoveConstTest, DoesNotAffectNonConstType) {
+ CompileAssertTypesEqual<int, RemoveConst<int>::type>();
+ CompileAssertTypesEqual<char&, RemoveConst<char&>::type>();
+}
+
+// Tests that RemoveConst removes const from const types.
+TEST(RemoveConstTest, RemovesConst) {
+ CompileAssertTypesEqual<int, RemoveConst<const int>::type>();
+ CompileAssertTypesEqual<char[2], RemoveConst<const char[2]>::type>();
+ CompileAssertTypesEqual<char[2][3], RemoveConst<const char[2][3]>::type>();
+}
+
+// Tests GTEST_REMOVE_CONST_.
+
+template <typename T1, typename T2>
+void TestGTestRemoveConst() {
+ CompileAssertTypesEqual<T1, GTEST_REMOVE_CONST_(T2)>();
+}
+
+TEST(RemoveConstTest, MacroVersion) {
+ TestGTestRemoveConst<int, int>();
+ TestGTestRemoveConst<double&, double&>();
+ TestGTestRemoveConst<char, const char>();
+}
+
+// Tests GTEST_REMOVE_REFERENCE_AND_CONST_.
+
+template <typename T1, typename T2>
+void TestGTestRemoveReferenceAndConst() {
+ CompileAssertTypesEqual<T1, GTEST_REMOVE_REFERENCE_AND_CONST_(T2)>();
+}
+
+TEST(RemoveReferenceToConstTest, Works) {
+ TestGTestRemoveReferenceAndConst<int, int>();
+ TestGTestRemoveReferenceAndConst<double, double&>();
+ TestGTestRemoveReferenceAndConst<char, const char>();
+ TestGTestRemoveReferenceAndConst<char, const char&>();
+ TestGTestRemoveReferenceAndConst<const char*, const char*>();
+}
+
+// Tests that AddReference does not affect reference types.
+TEST(AddReferenceTest, DoesNotAffectReferenceType) {
+ CompileAssertTypesEqual<int&, AddReference<int&>::type>();
+ CompileAssertTypesEqual<const char&, AddReference<const char&>::type>();
+}
+
+// Tests that AddReference adds reference to non-reference types.
+TEST(AddReferenceTest, AddsReference) {
+ CompileAssertTypesEqual<int&, AddReference<int>::type>();
+ CompileAssertTypesEqual<const char&, AddReference<const char>::type>();
+}
+
+// Tests GTEST_ADD_REFERENCE_.
+
+template <typename T1, typename T2>
+void TestGTestAddReference() {
+ CompileAssertTypesEqual<T1, GTEST_ADD_REFERENCE_(T2)>();
+}
+
+TEST(AddReferenceTest, MacroVersion) {
+ TestGTestAddReference<int&, int>();
+ TestGTestAddReference<const char&, const char&>();
+}
+
+// Tests GTEST_REFERENCE_TO_CONST_.
+
+template <typename T1, typename T2>
+void TestGTestReferenceToConst() {
+ CompileAssertTypesEqual<T1, GTEST_REFERENCE_TO_CONST_(T2)>();
+}
+
+TEST(GTestReferenceToConstTest, Works) {
+ TestGTestReferenceToConst<const char&, char>();
+ TestGTestReferenceToConst<const int&, const int>();
+ TestGTestReferenceToConst<const double&, double>();
+ TestGTestReferenceToConst<const String&, const String&>();
+}
+
+// Tests that ImplicitlyConvertible<T1, T2>::value is a compile-time constant.
+TEST(ImplicitlyConvertibleTest, ValueIsCompileTimeConstant) {
+ GTEST_COMPILE_ASSERT_((ImplicitlyConvertible<int, int>::value), const_true);
+ GTEST_COMPILE_ASSERT_((!ImplicitlyConvertible<void*, int*>::value),
+ const_false);
+}
+
+// Tests that ImplicitlyConvertible<T1, T2>::value is true when T1 can
+// be implicitly converted to T2.
+TEST(ImplicitlyConvertibleTest, ValueIsTrueWhenConvertible) {
+ EXPECT_TRUE((ImplicitlyConvertible<int, double>::value));
+ EXPECT_TRUE((ImplicitlyConvertible<double, int>::value));
+ EXPECT_TRUE((ImplicitlyConvertible<int*, void*>::value));
+ EXPECT_TRUE((ImplicitlyConvertible<int*, const int*>::value));
+ EXPECT_TRUE((ImplicitlyConvertible<ConversionHelperDerived&,
+ const ConversionHelperBase&>::value));
+ EXPECT_TRUE((ImplicitlyConvertible<const ConversionHelperBase,
+ ConversionHelperBase>::value));
+}
+
+// Tests that ImplicitlyConvertible<T1, T2>::value is false when T1
+// cannot be implicitly converted to T2.
+TEST(ImplicitlyConvertibleTest, ValueIsFalseWhenNotConvertible) {
+ EXPECT_FALSE((ImplicitlyConvertible<double, int*>::value));
+ EXPECT_FALSE((ImplicitlyConvertible<void*, int*>::value));
+ EXPECT_FALSE((ImplicitlyConvertible<const int*, int*>::value));
+ EXPECT_FALSE((ImplicitlyConvertible<ConversionHelperBase&,
+ ConversionHelperDerived&>::value));
+}
+
+// Tests IsContainerTest.
+
+class NonContainer {};
+
+TEST(IsContainerTestTest, WorksForNonContainer) {
+ EXPECT_EQ(sizeof(IsNotContainer), sizeof(IsContainerTest<int>(0)));
+ EXPECT_EQ(sizeof(IsNotContainer), sizeof(IsContainerTest<char[5]>(0)));
+ EXPECT_EQ(sizeof(IsNotContainer), sizeof(IsContainerTest<NonContainer>(0)));
+}
+
+TEST(IsContainerTestTest, WorksForContainer) {
+ EXPECT_EQ(sizeof(IsContainer),
+ sizeof(IsContainerTest<std::vector<bool> >(0)));
+ EXPECT_EQ(sizeof(IsContainer),
+ sizeof(IsContainerTest<std::map<int, double> >(0)));
+}
+
+// Tests ArrayEq().
+
+TEST(ArrayEqTest, WorksForDegeneratedArrays) {
+ EXPECT_TRUE(ArrayEq(5, 5L));
+ EXPECT_FALSE(ArrayEq('a', 0));
+}
+
+TEST(ArrayEqTest, WorksForOneDimensionalArrays) {
+ const int a[] = { 0, 1 };
+ long b[] = { 0, 1 };
+ EXPECT_TRUE(ArrayEq(a, b));
+ EXPECT_TRUE(ArrayEq(a, 2, b));
+
+ b[0] = 2;
+ EXPECT_FALSE(ArrayEq(a, b));
+ EXPECT_FALSE(ArrayEq(a, 1, b));
+}
+
+TEST(ArrayEqTest, WorksForTwoDimensionalArrays) {
+ const char a[][3] = { "hi", "lo" };
+ const char b[][3] = { "hi", "lo" };
+ const char c[][3] = { "hi", "li" };
+
+ EXPECT_TRUE(ArrayEq(a, b));
+ EXPECT_TRUE(ArrayEq(a, 2, b));
+
+ EXPECT_FALSE(ArrayEq(a, c));
+ EXPECT_FALSE(ArrayEq(a, 2, c));
+}
+
+// Tests ArrayAwareFind().
+
+TEST(ArrayAwareFindTest, WorksForOneDimensionalArray) {
+ const char a[] = "hello";
+ EXPECT_EQ(a + 4, ArrayAwareFind(a, a + 5, 'o'));
+ EXPECT_EQ(a + 5, ArrayAwareFind(a, a + 5, 'x'));
+}
+
+TEST(ArrayAwareFindTest, WorksForTwoDimensionalArray) {
+ int a[][2] = { { 0, 1 }, { 2, 3 }, { 4, 5 } };
+ const int b[2] = { 2, 3 };
+ EXPECT_EQ(a + 1, ArrayAwareFind(a, a + 3, b));
+
+ const int c[2] = { 6, 7 };
+ EXPECT_EQ(a + 3, ArrayAwareFind(a, a + 3, c));
+}
+
+// Tests CopyArray().
+
+TEST(CopyArrayTest, WorksForDegeneratedArrays) {
+ int n = 0;
+ CopyArray('a', &n);
+ EXPECT_EQ('a', n);
+}
+
+TEST(CopyArrayTest, WorksForOneDimensionalArrays) {
+ const char a[3] = "hi";
+ int b[3];
+#ifndef __BORLANDC__ // C++Builder cannot compile some array size deductions.
+ CopyArray(a, &b);
+ EXPECT_TRUE(ArrayEq(a, b));
+#endif
+
+ int c[3];
+ CopyArray(a, 3, c);
+ EXPECT_TRUE(ArrayEq(a, c));
+}
+
+TEST(CopyArrayTest, WorksForTwoDimensionalArrays) {
+ const int a[2][3] = { { 0, 1, 2 }, { 3, 4, 5 } };
+ int b[2][3];
+#ifndef __BORLANDC__ // C++Builder cannot compile some array size deductions.
+ CopyArray(a, &b);
+ EXPECT_TRUE(ArrayEq(a, b));
+#endif
+
+ int c[2][3];
+ CopyArray(a, 2, c);
+ EXPECT_TRUE(ArrayEq(a, c));
+}
+
+// Tests NativeArray.
+
+TEST(NativeArrayTest, ConstructorFromArrayWorks) {
+ const int a[3] = { 0, 1, 2 };
+ NativeArray<int> na(a, 3, kReference);
+ EXPECT_EQ(3U, na.size());
+ EXPECT_EQ(a, na.begin());
+}
+
+TEST(NativeArrayTest, CreatesAndDeletesCopyOfArrayWhenAskedTo) {
+ typedef int Array[2];
+ Array* a = new Array[1];
+ (*a)[0] = 0;
+ (*a)[1] = 1;
+ NativeArray<int> na(*a, 2, kCopy);
+ EXPECT_NE(*a, na.begin());
+ delete[] a;
+ EXPECT_EQ(0, na.begin()[0]);
+ EXPECT_EQ(1, na.begin()[1]);
+
+ // We rely on the heap checker to verify that na deletes the copy of
+ // array.
+}
+
+TEST(NativeArrayTest, TypeMembersAreCorrect) {
+ StaticAssertTypeEq<char, NativeArray<char>::value_type>();
+ StaticAssertTypeEq<int[2], NativeArray<int[2]>::value_type>();
+
+ StaticAssertTypeEq<const char*, NativeArray<char>::const_iterator>();
+ StaticAssertTypeEq<const bool(*)[2], NativeArray<bool[2]>::const_iterator>();
+}
+
+TEST(NativeArrayTest, MethodsWork) {
+ const int a[3] = { 0, 1, 2 };
+ NativeArray<int> na(a, 3, kCopy);
+ ASSERT_EQ(3U, na.size());
+ EXPECT_EQ(3, na.end() - na.begin());
+
+ NativeArray<int>::const_iterator it = na.begin();
+ EXPECT_EQ(0, *it);
+ ++it;
+ EXPECT_EQ(1, *it);
+ it++;
+ EXPECT_EQ(2, *it);
+ ++it;
+ EXPECT_EQ(na.end(), it);
+
+ EXPECT_TRUE(na == na);
+
+ NativeArray<int> na2(a, 3, kReference);
+ EXPECT_TRUE(na == na2);
+
+ const int b1[3] = { 0, 1, 1 };
+ const int b2[4] = { 0, 1, 2, 3 };
+ EXPECT_FALSE(na == NativeArray<int>(b1, 3, kReference));
+ EXPECT_FALSE(na == NativeArray<int>(b2, 4, kCopy));
+}
+
+TEST(NativeArrayTest, WorksForTwoDimensionalArray) {
+ const char a[2][3] = { "hi", "lo" };
+ NativeArray<char[3]> na(a, 2, kReference);
+ ASSERT_EQ(2U, na.size());
+ EXPECT_EQ(a, na.begin());
+}
+
+// Tests SkipPrefix().
+
+TEST(SkipPrefixTest, SkipsWhenPrefixMatches) {
+ const char* const str = "hello";
+
+ const char* p = str;
+ EXPECT_TRUE(SkipPrefix("", &p));
+ EXPECT_EQ(str, p);
+
+ p = str;
+ EXPECT_TRUE(SkipPrefix("hell", &p));
+ EXPECT_EQ(str + 4, p);
+}
+
+TEST(SkipPrefixTest, DoesNotSkipWhenPrefixDoesNotMatch) {
+ const char* const str = "world";
+
+ const char* p = str;
+ EXPECT_FALSE(SkipPrefix("W", &p));
+ EXPECT_EQ(str, p);
+
+ p = str;
+ EXPECT_FALSE(SkipPrefix("world!", &p));
+ EXPECT_EQ(str, p);
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_xml_outfile1_test_.cc b/libvpx/third_party/googletest/src/test/gtest_xml_outfile1_test_.cc
new file mode 100644
index 0000000..531ced4
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_xml_outfile1_test_.cc
@@ -0,0 +1,49 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: keith.ray@gmail.com (Keith Ray)
+//
+// gtest_xml_outfile1_test_ writes some xml via TestProperty used by
+// gtest_xml_outfiles_test.py
+
+#include "gtest/gtest.h"
+
+class PropertyOne : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ RecordProperty("SetUpProp", 1);
+ }
+ virtual void TearDown() {
+ RecordProperty("TearDownProp", 1);
+ }
+};
+
+TEST_F(PropertyOne, TestSomeProperties) {
+ RecordProperty("TestSomeProperty", 1);
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_xml_outfile2_test_.cc b/libvpx/third_party/googletest/src/test/gtest_xml_outfile2_test_.cc
new file mode 100644
index 0000000..7b400b2
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_xml_outfile2_test_.cc
@@ -0,0 +1,49 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: keith.ray@gmail.com (Keith Ray)
+//
+// gtest_xml_outfile2_test_ writes some xml via TestProperty used by
+// gtest_xml_outfiles_test.py
+
+#include "gtest/gtest.h"
+
+class PropertyTwo : public testing::Test {
+ protected:
+ virtual void SetUp() {
+ RecordProperty("SetUpProp", 2);
+ }
+ virtual void TearDown() {
+ RecordProperty("TearDownProp", 2);
+ }
+};
+
+TEST_F(PropertyTwo, TestSomeProperties) {
+ RecordProperty("TestSomeProperty", 2);
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_xml_outfiles_test.py b/libvpx/third_party/googletest/src/test/gtest_xml_outfiles_test.py
new file mode 100755
index 0000000..0fe947f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_xml_outfiles_test.py
@@ -0,0 +1,132 @@
+#!/usr/bin/env python
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Unit test for the gtest_xml_output module."""
+
+__author__ = "keith.ray@gmail.com (Keith Ray)"
+
+import os
+from xml.dom import minidom, Node
+
+import gtest_test_utils
+import gtest_xml_test_utils
+
+
+GTEST_OUTPUT_SUBDIR = "xml_outfiles"
+GTEST_OUTPUT_1_TEST = "gtest_xml_outfile1_test_"
+GTEST_OUTPUT_2_TEST = "gtest_xml_outfile2_test_"
+
+EXPECTED_XML_1 = """<?xml version="1.0" encoding="UTF-8"?>
+<testsuites tests="1" failures="0" disabled="0" errors="0" time="*" name="AllTests">
+ <testsuite name="PropertyOne" tests="1" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="TestSomeProperties" status="run" time="*" classname="PropertyOne" SetUpProp="1" TestSomeProperty="1" TearDownProp="1" />
+ </testsuite>
+</testsuites>
+"""
+
+EXPECTED_XML_2 = """<?xml version="1.0" encoding="UTF-8"?>
+<testsuites tests="1" failures="0" disabled="0" errors="0" time="*" name="AllTests">
+ <testsuite name="PropertyTwo" tests="1" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="TestSomeProperties" status="run" time="*" classname="PropertyTwo" SetUpProp="2" TestSomeProperty="2" TearDownProp="2" />
+ </testsuite>
+</testsuites>
+"""
+
+
+class GTestXMLOutFilesTest(gtest_xml_test_utils.GTestXMLTestCase):
+ """Unit test for Google Test's XML output functionality."""
+
+ def setUp(self):
+ # We want the trailing '/' that the last "" provides in os.path.join, for
+ # telling Google Test to create an output directory instead of a single file
+ # for xml output.
+ self.output_dir_ = os.path.join(gtest_test_utils.GetTempDir(),
+ GTEST_OUTPUT_SUBDIR, "")
+ self.DeleteFilesAndDir()
+
+ def tearDown(self):
+ self.DeleteFilesAndDir()
+
+ def DeleteFilesAndDir(self):
+ try:
+ os.remove(os.path.join(self.output_dir_, GTEST_OUTPUT_1_TEST + ".xml"))
+ except os.error:
+ pass
+ try:
+ os.remove(os.path.join(self.output_dir_, GTEST_OUTPUT_2_TEST + ".xml"))
+ except os.error:
+ pass
+ try:
+ os.rmdir(self.output_dir_)
+ except os.error:
+ pass
+
+ def testOutfile1(self):
+ self._TestOutFile(GTEST_OUTPUT_1_TEST, EXPECTED_XML_1)
+
+ def testOutfile2(self):
+ self._TestOutFile(GTEST_OUTPUT_2_TEST, EXPECTED_XML_2)
+
+ def _TestOutFile(self, test_name, expected_xml):
+ gtest_prog_path = gtest_test_utils.GetTestExecutablePath(test_name)
+ command = [gtest_prog_path, "--gtest_output=xml:%s" % self.output_dir_]
+ p = gtest_test_utils.Subprocess(command,
+ working_dir=gtest_test_utils.GetTempDir())
+ self.assert_(p.exited)
+ self.assertEquals(0, p.exit_code)
+
+ # TODO(wan@google.com): libtool causes the built test binary to be
+ # named lt-gtest_xml_outfiles_test_ instead of
+ # gtest_xml_outfiles_test_. To account for this possibillity, we
+ # allow both names in the following code. We should remove this
+ # hack when Chandler Carruth's libtool replacement tool is ready.
+ output_file_name1 = test_name + ".xml"
+ output_file1 = os.path.join(self.output_dir_, output_file_name1)
+ output_file_name2 = 'lt-' + output_file_name1
+ output_file2 = os.path.join(self.output_dir_, output_file_name2)
+ self.assert_(os.path.isfile(output_file1) or os.path.isfile(output_file2),
+ output_file1)
+
+ expected = minidom.parseString(expected_xml)
+ if os.path.isfile(output_file1):
+ actual = minidom.parse(output_file1)
+ else:
+ actual = minidom.parse(output_file2)
+ self.NormalizeXml(actual.documentElement)
+ self.AssertEquivalentNodes(expected.documentElement,
+ actual.documentElement)
+ expected.unlink()
+ actual.unlink()
+
+
+if __name__ == "__main__":
+ os.environ["GTEST_STACK_TRACE_DEPTH"] = "0"
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_xml_output_unittest.py b/libvpx/third_party/googletest/src/test/gtest_xml_output_unittest.py
new file mode 100755
index 0000000..bdd5035
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_xml_output_unittest.py
@@ -0,0 +1,242 @@
+#!/usr/bin/env python
+#
+# Copyright 2006, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Unit test for the gtest_xml_output module"""
+
+__author__ = 'eefacm@gmail.com (Sean Mcafee)'
+
+import errno
+import os
+import sys
+from xml.dom import minidom, Node
+
+import gtest_test_utils
+import gtest_xml_test_utils
+
+
+GTEST_OUTPUT_FLAG = "--gtest_output"
+GTEST_DEFAULT_OUTPUT_FILE = "test_detail.xml"
+GTEST_PROGRAM_NAME = "gtest_xml_output_unittest_"
+
+SUPPORTS_STACK_TRACES = False
+
+if SUPPORTS_STACK_TRACES:
+ STACK_TRACE_TEMPLATE = "\nStack trace:\n*"
+else:
+ STACK_TRACE_TEMPLATE = ""
+
+EXPECTED_NON_EMPTY_XML = """<?xml version="1.0" encoding="UTF-8"?>
+<testsuites tests="23" failures="4" disabled="2" errors="0" time="*" name="AllTests">
+ <testsuite name="SuccessfulTest" tests="1" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="Succeeds" status="run" time="*" classname="SuccessfulTest"/>
+ </testsuite>
+ <testsuite name="FailedTest" tests="1" failures="1" disabled="0" errors="0" time="*">
+ <testcase name="Fails" status="run" time="*" classname="FailedTest">
+ <failure message="Value of: 2
Expected: 1" type=""><![CDATA[gtest_xml_output_unittest_.cc:*
+Value of: 2
+Expected: 1%(stack)s]]></failure>
+ </testcase>
+ </testsuite>
+ <testsuite name="MixedResultTest" tests="3" failures="1" disabled="1" errors="0" time="*">
+ <testcase name="Succeeds" status="run" time="*" classname="MixedResultTest"/>
+ <testcase name="Fails" status="run" time="*" classname="MixedResultTest">
+ <failure message="Value of: 2
Expected: 1" type=""><![CDATA[gtest_xml_output_unittest_.cc:*
+Value of: 2
+Expected: 1%(stack)s]]></failure>
+ <failure message="Value of: 3
Expected: 2" type=""><![CDATA[gtest_xml_output_unittest_.cc:*
+Value of: 3
+Expected: 2%(stack)s]]></failure>
+ </testcase>
+ <testcase name="DISABLED_test" status="notrun" time="*" classname="MixedResultTest"/>
+ </testsuite>
+ <testsuite name="XmlQuotingTest" tests="1" failures="1" disabled="0" errors="0" time="*">
+ <testcase name="OutputsCData" status="run" time="*" classname="XmlQuotingTest">
+ <failure message="Failed
XML output: <?xml encoding="utf-8"><top><![CDATA[cdata text]]></top>" type=""><![CDATA[gtest_xml_output_unittest_.cc:*
+Failed
+XML output: <?xml encoding="utf-8"><top><![CDATA[cdata text]]>]]><![CDATA[</top>%(stack)s]]></failure>
+ </testcase>
+ </testsuite>
+ <testsuite name="InvalidCharactersTest" tests="1" failures="1" disabled="0" errors="0" time="*">
+ <testcase name="InvalidCharactersInMessage" status="run" time="*" classname="InvalidCharactersTest">
+ <failure message="Failed
Invalid characters in brackets []" type=""><![CDATA[gtest_xml_output_unittest_.cc:*
+Failed
+Invalid characters in brackets []%(stack)s]]></failure>
+ </testcase>
+ </testsuite>
+ <testsuite name="DisabledTest" tests="1" failures="0" disabled="1" errors="0" time="*">
+ <testcase name="DISABLED_test_not_run" status="notrun" time="*" classname="DisabledTest"/>
+ </testsuite>
+ <testsuite name="PropertyRecordingTest" tests="4" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="OneProperty" status="run" time="*" classname="PropertyRecordingTest" key_1="1"/>
+ <testcase name="IntValuedProperty" status="run" time="*" classname="PropertyRecordingTest" key_int="1"/>
+ <testcase name="ThreeProperties" status="run" time="*" classname="PropertyRecordingTest" key_1="1" key_2="2" key_3="3"/>
+ <testcase name="TwoValuesForOneKeyUsesLastValue" status="run" time="*" classname="PropertyRecordingTest" key_1="2"/>
+ </testsuite>
+ <testsuite name="NoFixtureTest" tests="3" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="RecordProperty" status="run" time="*" classname="NoFixtureTest" key="1"/>
+ <testcase name="ExternalUtilityThatCallsRecordIntValuedProperty" status="run" time="*" classname="NoFixtureTest" key_for_utility_int="1"/>
+ <testcase name="ExternalUtilityThatCallsRecordStringValuedProperty" status="run" time="*" classname="NoFixtureTest" key_for_utility_string="1"/>
+ </testsuite>
+ <testsuite name="Single/ValueParamTest" tests="4" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="HasValueParamAttribute/0" value_param="33" status="run" time="*" classname="Single/ValueParamTest" />
+ <testcase name="HasValueParamAttribute/1" value_param="42" status="run" time="*" classname="Single/ValueParamTest" />
+ <testcase name="AnotherTestThatHasValueParamAttribute/0" value_param="33" status="run" time="*" classname="Single/ValueParamTest" />
+ <testcase name="AnotherTestThatHasValueParamAttribute/1" value_param="42" status="run" time="*" classname="Single/ValueParamTest" />
+ </testsuite>
+ <testsuite name="TypedTest/0" tests="1" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="HasTypeParamAttribute" type_param="*" status="run" time="*" classname="TypedTest/0" />
+ </testsuite>
+ <testsuite name="TypedTest/1" tests="1" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="HasTypeParamAttribute" type_param="*" status="run" time="*" classname="TypedTest/1" />
+ </testsuite>
+ <testsuite name="Single/TypeParameterizedTestCase/0" tests="1" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="HasTypeParamAttribute" type_param="*" status="run" time="*" classname="Single/TypeParameterizedTestCase/0" />
+ </testsuite>
+ <testsuite name="Single/TypeParameterizedTestCase/1" tests="1" failures="0" disabled="0" errors="0" time="*">
+ <testcase name="HasTypeParamAttribute" type_param="*" status="run" time="*" classname="Single/TypeParameterizedTestCase/1" />
+ </testsuite>
+</testsuites>""" % {'stack': STACK_TRACE_TEMPLATE}
+
+
+EXPECTED_EMPTY_XML = """<?xml version="1.0" encoding="UTF-8"?>
+<testsuites tests="0" failures="0" disabled="0" errors="0" time="*" name="AllTests">
+</testsuites>"""
+
+
+class GTestXMLOutputUnitTest(gtest_xml_test_utils.GTestXMLTestCase):
+ """
+ Unit test for Google Test's XML output functionality.
+ """
+
+ def testNonEmptyXmlOutput(self):
+ """
+ Runs a test program that generates a non-empty XML output, and
+ tests that the XML output is expected.
+ """
+ self._TestXmlOutput(GTEST_PROGRAM_NAME, EXPECTED_NON_EMPTY_XML, 1)
+
+ def testEmptyXmlOutput(self):
+ """
+ Runs a test program that generates an empty XML output, and
+ tests that the XML output is expected.
+ """
+
+ self._TestXmlOutput("gtest_no_test_unittest",
+ EXPECTED_EMPTY_XML, 0)
+
+ def testDefaultOutputFile(self):
+ """
+ Confirms that Google Test produces an XML output file with the expected
+ default name if no name is explicitly specified.
+ """
+ output_file = os.path.join(gtest_test_utils.GetTempDir(),
+ GTEST_DEFAULT_OUTPUT_FILE)
+ gtest_prog_path = gtest_test_utils.GetTestExecutablePath(
+ "gtest_no_test_unittest")
+ try:
+ os.remove(output_file)
+ except OSError, e:
+ if e.errno != errno.ENOENT:
+ raise
+
+ p = gtest_test_utils.Subprocess(
+ [gtest_prog_path, "%s=xml" % GTEST_OUTPUT_FLAG],
+ working_dir=gtest_test_utils.GetTempDir())
+ self.assert_(p.exited)
+ self.assertEquals(0, p.exit_code)
+ self.assert_(os.path.isfile(output_file))
+
+ def testSuppressedXmlOutput(self):
+ """
+ Tests that no XML file is generated if the default XML listener is
+ shut down before RUN_ALL_TESTS is invoked.
+ """
+
+ xml_path = os.path.join(gtest_test_utils.GetTempDir(),
+ GTEST_PROGRAM_NAME + "out.xml")
+ if os.path.isfile(xml_path):
+ os.remove(xml_path)
+
+ gtest_prog_path = gtest_test_utils.GetTestExecutablePath(GTEST_PROGRAM_NAME)
+
+ command = [gtest_prog_path,
+ "%s=xml:%s" % (GTEST_OUTPUT_FLAG, xml_path),
+ "--shut_down_xml"]
+ p = gtest_test_utils.Subprocess(command)
+ if p.terminated_by_signal:
+ self.assert_(False,
+ "%s was killed by signal %d" % (gtest_prog_name, p.signal))
+ else:
+ self.assert_(p.exited)
+ self.assertEquals(1, p.exit_code,
+ "'%s' exited with code %s, which doesn't match "
+ "the expected exit code %s."
+ % (command, p.exit_code, 1))
+
+ self.assert_(not os.path.isfile(xml_path))
+
+
+ def _TestXmlOutput(self, gtest_prog_name, expected_xml, expected_exit_code):
+ """
+ Asserts that the XML document generated by running the program
+ gtest_prog_name matches expected_xml, a string containing another
+ XML document. Furthermore, the program's exit code must be
+ expected_exit_code.
+ """
+ xml_path = os.path.join(gtest_test_utils.GetTempDir(),
+ gtest_prog_name + "out.xml")
+ gtest_prog_path = gtest_test_utils.GetTestExecutablePath(gtest_prog_name)
+
+ command = [gtest_prog_path, "%s=xml:%s" % (GTEST_OUTPUT_FLAG, xml_path)]
+ p = gtest_test_utils.Subprocess(command)
+ if p.terminated_by_signal:
+ self.assert_(False,
+ "%s was killed by signal %d" % (gtest_prog_name, p.signal))
+ else:
+ self.assert_(p.exited)
+ self.assertEquals(expected_exit_code, p.exit_code,
+ "'%s' exited with code %s, which doesn't match "
+ "the expected exit code %s."
+ % (command, p.exit_code, expected_exit_code))
+
+ expected = minidom.parseString(expected_xml)
+ actual = minidom.parse(xml_path)
+ self.NormalizeXml(actual.documentElement)
+ self.AssertEquivalentNodes(expected.documentElement,
+ actual.documentElement)
+ expected.unlink()
+ actual .unlink()
+
+
+
+if __name__ == '__main__':
+ os.environ['GTEST_STACK_TRACE_DEPTH'] = '1'
+ gtest_test_utils.Main()
diff --git a/libvpx/third_party/googletest/src/test/gtest_xml_output_unittest_.cc b/libvpx/third_party/googletest/src/test/gtest_xml_output_unittest_.cc
new file mode 100644
index 0000000..741a887
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_xml_output_unittest_.cc
@@ -0,0 +1,174 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// Author: eefacm@gmail.com (Sean Mcafee)
+
+// Unit test for Google Test XML output.
+//
+// A user can specify XML output in a Google Test program to run via
+// either the GTEST_OUTPUT environment variable or the --gtest_output
+// flag. This is used for testing such functionality.
+//
+// This program will be invoked from a Python unit test. Don't run it
+// directly.
+
+#include "gtest/gtest.h"
+
+using ::testing::InitGoogleTest;
+using ::testing::TestEventListeners;
+using ::testing::TestWithParam;
+using ::testing::UnitTest;
+using ::testing::Test;
+using ::testing::Types;
+using ::testing::Values;
+
+class SuccessfulTest : public Test {
+};
+
+TEST_F(SuccessfulTest, Succeeds) {
+ SUCCEED() << "This is a success.";
+ ASSERT_EQ(1, 1);
+}
+
+class FailedTest : public Test {
+};
+
+TEST_F(FailedTest, Fails) {
+ ASSERT_EQ(1, 2);
+}
+
+class DisabledTest : public Test {
+};
+
+TEST_F(DisabledTest, DISABLED_test_not_run) {
+ FAIL() << "Unexpected failure: Disabled test should not be run";
+}
+
+TEST(MixedResultTest, Succeeds) {
+ EXPECT_EQ(1, 1);
+ ASSERT_EQ(1, 1);
+}
+
+TEST(MixedResultTest, Fails) {
+ EXPECT_EQ(1, 2);
+ ASSERT_EQ(2, 3);
+}
+
+TEST(MixedResultTest, DISABLED_test) {
+ FAIL() << "Unexpected failure: Disabled test should not be run";
+}
+
+TEST(XmlQuotingTest, OutputsCData) {
+ FAIL() << "XML output: "
+ "<?xml encoding=\"utf-8\"><top><![CDATA[cdata text]]></top>";
+}
+
+// Helps to test that invalid characters produced by test code do not make
+// it into the XML file.
+TEST(InvalidCharactersTest, InvalidCharactersInMessage) {
+ FAIL() << "Invalid characters in brackets [\x1\x2]";
+}
+
+class PropertyRecordingTest : public Test {
+};
+
+TEST_F(PropertyRecordingTest, OneProperty) {
+ RecordProperty("key_1", "1");
+}
+
+TEST_F(PropertyRecordingTest, IntValuedProperty) {
+ RecordProperty("key_int", 1);
+}
+
+TEST_F(PropertyRecordingTest, ThreeProperties) {
+ RecordProperty("key_1", "1");
+ RecordProperty("key_2", "2");
+ RecordProperty("key_3", "3");
+}
+
+TEST_F(PropertyRecordingTest, TwoValuesForOneKeyUsesLastValue) {
+ RecordProperty("key_1", "1");
+ RecordProperty("key_1", "2");
+}
+
+TEST(NoFixtureTest, RecordProperty) {
+ RecordProperty("key", "1");
+}
+
+void ExternalUtilityThatCallsRecordProperty(const char* key, int value) {
+ testing::Test::RecordProperty(key, value);
+}
+
+void ExternalUtilityThatCallsRecordProperty(const char* key,
+ const char* value) {
+ testing::Test::RecordProperty(key, value);
+}
+
+TEST(NoFixtureTest, ExternalUtilityThatCallsRecordIntValuedProperty) {
+ ExternalUtilityThatCallsRecordProperty("key_for_utility_int", 1);
+}
+
+TEST(NoFixtureTest, ExternalUtilityThatCallsRecordStringValuedProperty) {
+ ExternalUtilityThatCallsRecordProperty("key_for_utility_string", "1");
+}
+
+// Verifies that the test parameter value is output in the 'value_param'
+// XML attribute for value-parameterized tests.
+class ValueParamTest : public TestWithParam<int> {};
+TEST_P(ValueParamTest, HasValueParamAttribute) {}
+TEST_P(ValueParamTest, AnotherTestThatHasValueParamAttribute) {}
+INSTANTIATE_TEST_CASE_P(Single, ValueParamTest, Values(33, 42));
+
+// Verifies that the type parameter name is output in the 'type_param'
+// XML attribute for typed tests.
+template <typename T> class TypedTest : public Test {};
+typedef Types<int, long> TypedTestTypes;
+TYPED_TEST_CASE(TypedTest, TypedTestTypes);
+TYPED_TEST(TypedTest, HasTypeParamAttribute) {}
+
+// Verifies that the type parameter name is output in the 'type_param'
+// XML attribute for type-parameterized tests.
+template <typename T> class TypeParameterizedTestCase : public Test {};
+TYPED_TEST_CASE_P(TypeParameterizedTestCase);
+TYPED_TEST_P(TypeParameterizedTestCase, HasTypeParamAttribute) {}
+REGISTER_TYPED_TEST_CASE_P(TypeParameterizedTestCase, HasTypeParamAttribute);
+typedef Types<int, long> TypeParameterizedTestCaseTypes;
+INSTANTIATE_TYPED_TEST_CASE_P(Single,
+ TypeParameterizedTestCase,
+ TypeParameterizedTestCaseTypes);
+
+int main(int argc, char** argv) {
+ InitGoogleTest(&argc, argv);
+
+ if (argc > 1 && strcmp(argv[1], "--shut_down_xml") == 0) {
+ TestEventListeners& listeners = UnitTest::GetInstance()->listeners();
+ delete listeners.Release(listeners.default_xml_generator());
+ }
+ return RUN_ALL_TESTS();
+}
diff --git a/libvpx/third_party/googletest/src/test/gtest_xml_test_utils.py b/libvpx/third_party/googletest/src/test/gtest_xml_test_utils.py
new file mode 100755
index 0000000..0f55c16
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/gtest_xml_test_utils.py
@@ -0,0 +1,179 @@
+#!/usr/bin/env python
+#
+# Copyright 2006, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""Unit test utilities for gtest_xml_output"""
+
+__author__ = 'eefacm@gmail.com (Sean Mcafee)'
+
+import re
+from xml.dom import minidom, Node
+
+import gtest_test_utils
+
+
+GTEST_OUTPUT_FLAG = "--gtest_output"
+GTEST_DEFAULT_OUTPUT_FILE = "test_detail.xml"
+
+class GTestXMLTestCase(gtest_test_utils.TestCase):
+ """
+ Base class for tests of Google Test's XML output functionality.
+ """
+
+
+ def AssertEquivalentNodes(self, expected_node, actual_node):
+ """
+ Asserts that actual_node (a DOM node object) is equivalent to
+ expected_node (another DOM node object), in that either both of
+ them are CDATA nodes and have the same value, or both are DOM
+ elements and actual_node meets all of the following conditions:
+
+ * It has the same tag name as expected_node.
+ * It has the same set of attributes as expected_node, each with
+ the same value as the corresponding attribute of expected_node.
+ Exceptions are any attribute named "time", which needs only be
+ convertible to a floating-point number and any attribute named
+ "type_param" which only has to be non-empty.
+ * It has an equivalent set of child nodes (including elements and
+ CDATA sections) as expected_node. Note that we ignore the
+ order of the children as they are not guaranteed to be in any
+ particular order.
+ """
+
+ if expected_node.nodeType == Node.CDATA_SECTION_NODE:
+ self.assertEquals(Node.CDATA_SECTION_NODE, actual_node.nodeType)
+ self.assertEquals(expected_node.nodeValue, actual_node.nodeValue)
+ return
+
+ self.assertEquals(Node.ELEMENT_NODE, actual_node.nodeType)
+ self.assertEquals(Node.ELEMENT_NODE, expected_node.nodeType)
+ self.assertEquals(expected_node.tagName, actual_node.tagName)
+
+ expected_attributes = expected_node.attributes
+ actual_attributes = actual_node .attributes
+ self.assertEquals(
+ expected_attributes.length, actual_attributes.length,
+ "attribute numbers differ in element " + actual_node.tagName)
+ for i in range(expected_attributes.length):
+ expected_attr = expected_attributes.item(i)
+ actual_attr = actual_attributes.get(expected_attr.name)
+ self.assert_(
+ actual_attr is not None,
+ "expected attribute %s not found in element %s" %
+ (expected_attr.name, actual_node.tagName))
+ self.assertEquals(expected_attr.value, actual_attr.value,
+ " values of attribute %s in element %s differ" %
+ (expected_attr.name, actual_node.tagName))
+
+ expected_children = self._GetChildren(expected_node)
+ actual_children = self._GetChildren(actual_node)
+ self.assertEquals(
+ len(expected_children), len(actual_children),
+ "number of child elements differ in element " + actual_node.tagName)
+ for child_id, child in expected_children.iteritems():
+ self.assert_(child_id in actual_children,
+ '<%s> is not in <%s> (in element %s)' %
+ (child_id, actual_children, actual_node.tagName))
+ self.AssertEquivalentNodes(child, actual_children[child_id])
+
+ identifying_attribute = {
+ "testsuites": "name",
+ "testsuite": "name",
+ "testcase": "name",
+ "failure": "message",
+ }
+
+ def _GetChildren(self, element):
+ """
+ Fetches all of the child nodes of element, a DOM Element object.
+ Returns them as the values of a dictionary keyed by the IDs of the
+ children. For <testsuites>, <testsuite> and <testcase> elements, the ID
+ is the value of their "name" attribute; for <failure> elements, it is
+ the value of the "message" attribute; CDATA sections and non-whitespace
+ text nodes are concatenated into a single CDATA section with ID
+ "detail". An exception is raised if any element other than the above
+ four is encountered, if two child elements with the same identifying
+ attributes are encountered, or if any other type of node is encountered.
+ """
+
+ children = {}
+ for child in element.childNodes:
+ if child.nodeType == Node.ELEMENT_NODE:
+ self.assert_(child.tagName in self.identifying_attribute,
+ "Encountered unknown element <%s>" % child.tagName)
+ childID = child.getAttribute(self.identifying_attribute[child.tagName])
+ self.assert_(childID not in children)
+ children[childID] = child
+ elif child.nodeType in [Node.TEXT_NODE, Node.CDATA_SECTION_NODE]:
+ if "detail" not in children:
+ if (child.nodeType == Node.CDATA_SECTION_NODE or
+ not child.nodeValue.isspace()):
+ children["detail"] = child.ownerDocument.createCDATASection(
+ child.nodeValue)
+ else:
+ children["detail"].nodeValue += child.nodeValue
+ else:
+ self.fail("Encountered unexpected node type %d" % child.nodeType)
+ return children
+
+ def NormalizeXml(self, element):
+ """
+ Normalizes Google Test's XML output to eliminate references to transient
+ information that may change from run to run.
+
+ * The "time" attribute of <testsuites>, <testsuite> and <testcase>
+ elements is replaced with a single asterisk, if it contains
+ only digit characters.
+ * The "type_param" attribute of <testcase> elements is replaced with a
+ single asterisk (if it sn non-empty) as it is the type name returned
+ by the compiler and is platform dependent.
+ * The line number reported in the first line of the "message"
+ attribute of <failure> elements is replaced with a single asterisk.
+ * The directory names in file paths are removed.
+ * The stack traces are removed.
+ """
+
+ if element.tagName in ("testsuites", "testsuite", "testcase"):
+ time = element.getAttributeNode("time")
+ time.value = re.sub(r"^\d+(\.\d+)?$", "*", time.value)
+ type_param = element.getAttributeNode("type_param")
+ if type_param and type_param.value:
+ type_param.value = "*"
+ elif element.tagName == "failure":
+ for child in element.childNodes:
+ if child.nodeType == Node.CDATA_SECTION_NODE:
+ # Removes the source line number.
+ cdata = re.sub(r"^.*[/\\](.*:)\d+\n", "\\1*\n", child.nodeValue)
+ # Removes the actual stack trace.
+ child.nodeValue = re.sub(r"\nStack trace:\n(.|\n)*",
+ "", cdata)
+ for child in element.childNodes:
+ if child.nodeType == Node.ELEMENT_NODE:
+ self.NormalizeXml(child)
diff --git a/libvpx/third_party/googletest/src/test/production.cc b/libvpx/third_party/googletest/src/test/production.cc
new file mode 100644
index 0000000..8b8a40b
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/production.cc
@@ -0,0 +1,36 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// This is part of the unit test for include/gtest/gtest_prod.h.
+
+#include "production.h"
+
+PrivateCode::PrivateCode() : x_(0) {}
diff --git a/libvpx/third_party/googletest/src/test/production.h b/libvpx/third_party/googletest/src/test/production.h
new file mode 100644
index 0000000..98fd5e4
--- /dev/null
+++ b/libvpx/third_party/googletest/src/test/production.h
@@ -0,0 +1,55 @@
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// This is part of the unit test for include/gtest/gtest_prod.h.
+
+#ifndef GTEST_TEST_PRODUCTION_H_
+#define GTEST_TEST_PRODUCTION_H_
+
+#include "gtest/gtest_prod.h"
+
+class PrivateCode {
+ public:
+ // Declares a friend test that does not use a fixture.
+ FRIEND_TEST(PrivateCodeTest, CanAccessPrivateMembers);
+
+ // Declares a friend test that uses a fixture.
+ FRIEND_TEST(PrivateCodeFixtureTest, CanAccessPrivateMembers);
+
+ PrivateCode();
+
+ int x() const { return x_; }
+ private:
+ void set_x(int an_x) { x_ = an_x; }
+ int x_;
+};
+
+#endif // GTEST_TEST_PRODUCTION_H_
diff --git a/libvpx/third_party/googletest/src/xcode/Config/DebugProject.xcconfig b/libvpx/third_party/googletest/src/xcode/Config/DebugProject.xcconfig
new file mode 100644
index 0000000..3d68157
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Config/DebugProject.xcconfig
@@ -0,0 +1,30 @@
+//
+// DebugProject.xcconfig
+//
+// These are Debug Configuration project settings for the gtest framework and
+// examples. It is set in the "Based On:" dropdown in the "Project" info
+// dialog.
+// This file is based on the Xcode Configuration files in:
+// http://code.google.com/p/google-toolbox-for-mac/
+//
+
+#include "General.xcconfig"
+
+// No optimization
+GCC_OPTIMIZATION_LEVEL = 0
+
+// Deployment postprocessing is what triggers Xcode to strip, turn it off
+DEPLOYMENT_POSTPROCESSING = NO
+
+// Dead code stripping off
+DEAD_CODE_STRIPPING = NO
+
+// Debug symbols should be on obviously
+GCC_GENERATE_DEBUGGING_SYMBOLS = YES
+
+// Define the DEBUG macro in all debug builds
+OTHER_CFLAGS = $(OTHER_CFLAGS) -DDEBUG=1
+
+// These are turned off to avoid STL incompatibilities with client code
+// // Turns on special C++ STL checks to "encourage" good STL use
+// GCC_PREPROCESSOR_DEFINITIONS = $(GCC_PREPROCESSOR_DEFINITIONS) _GLIBCXX_DEBUG_PEDANTIC _GLIBCXX_DEBUG _GLIBCPP_CONCEPT_CHECKS
diff --git a/libvpx/third_party/googletest/src/xcode/Config/FrameworkTarget.xcconfig b/libvpx/third_party/googletest/src/xcode/Config/FrameworkTarget.xcconfig
new file mode 100644
index 0000000..357b1c8
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Config/FrameworkTarget.xcconfig
@@ -0,0 +1,17 @@
+//
+// FrameworkTarget.xcconfig
+//
+// These are Framework target settings for the gtest framework and examples. It
+// is set in the "Based On:" dropdown in the "Target" info dialog.
+// This file is based on the Xcode Configuration files in:
+// http://code.google.com/p/google-toolbox-for-mac/
+//
+
+// Dynamic libs need to be position independent
+GCC_DYNAMIC_NO_PIC = NO
+
+// Dynamic libs should not have their external symbols stripped.
+STRIP_STYLE = non-global
+
+// Let the user install by specifying the $DSTROOT with xcodebuild
+SKIP_INSTALL = NO
diff --git a/libvpx/third_party/googletest/src/xcode/Config/General.xcconfig b/libvpx/third_party/googletest/src/xcode/Config/General.xcconfig
new file mode 100644
index 0000000..f23e322
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Config/General.xcconfig
@@ -0,0 +1,41 @@
+//
+// General.xcconfig
+//
+// These are General configuration settings for the gtest framework and
+// examples.
+// This file is based on the Xcode Configuration files in:
+// http://code.google.com/p/google-toolbox-for-mac/
+//
+
+// Build for PPC and Intel, 32- and 64-bit
+ARCHS = i386 x86_64 ppc ppc64
+
+// Zerolink prevents link warnings so turn it off
+ZERO_LINK = NO
+
+// Prebinding considered unhelpful in 10.3 and later
+PREBINDING = NO
+
+// Strictest warning policy
+WARNING_CFLAGS = -Wall -Werror -Wendif-labels -Wnewline-eof -Wno-sign-compare -Wshadow
+
+// Work around Xcode bugs by using external strip. See:
+// http://lists.apple.com/archives/Xcode-users/2006/Feb/msg00050.html
+SEPARATE_STRIP = YES
+
+// Force C99 dialect
+GCC_C_LANGUAGE_STANDARD = c99
+
+// not sure why apple defaults this on, but it's pretty risky
+ALWAYS_SEARCH_USER_PATHS = NO
+
+// Turn on position dependent code for most cases (overridden where appropriate)
+GCC_DYNAMIC_NO_PIC = YES
+
+// Default SDK and minimum OS version is 10.4
+SDKROOT = $(DEVELOPER_SDK_DIR)/MacOSX10.4u.sdk
+MACOSX_DEPLOYMENT_TARGET = 10.4
+GCC_VERSION = 4.0
+
+// VERSIONING BUILD SETTINGS (used in Info.plist)
+GTEST_VERSIONINFO_ABOUT = © 2008 Google Inc.
diff --git a/libvpx/third_party/googletest/src/xcode/Config/ReleaseProject.xcconfig b/libvpx/third_party/googletest/src/xcode/Config/ReleaseProject.xcconfig
new file mode 100644
index 0000000..5349f0a
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Config/ReleaseProject.xcconfig
@@ -0,0 +1,32 @@
+//
+// ReleaseProject.xcconfig
+//
+// These are Release Configuration project settings for the gtest framework
+// and examples. It is set in the "Based On:" dropdown in the "Project" info
+// dialog.
+// This file is based on the Xcode Configuration files in:
+// http://code.google.com/p/google-toolbox-for-mac/
+//
+
+#include "General.xcconfig"
+
+// subconfig/Release.xcconfig
+
+// Optimize for space and size (Apple recommendation)
+GCC_OPTIMIZATION_LEVEL = s
+
+// Deploment postprocessing is what triggers Xcode to strip
+DEPLOYMENT_POSTPROCESSING = YES
+
+// No symbols
+GCC_GENERATE_DEBUGGING_SYMBOLS = NO
+
+// Dead code strip does not affect ObjC code but can help for C
+DEAD_CODE_STRIPPING = YES
+
+// NDEBUG is used by things like assert.h, so define it for general compat.
+// ASSERT going away in release tends to create unused vars.
+OTHER_CFLAGS = $(OTHER_CFLAGS) -DNDEBUG=1 -Wno-unused-variable
+
+// When we strip we want to strip all symbols in release, but save externals.
+STRIP_STYLE = all
diff --git a/libvpx/third_party/googletest/src/xcode/Config/StaticLibraryTarget.xcconfig b/libvpx/third_party/googletest/src/xcode/Config/StaticLibraryTarget.xcconfig
new file mode 100644
index 0000000..3922fa5
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Config/StaticLibraryTarget.xcconfig
@@ -0,0 +1,18 @@
+//
+// StaticLibraryTarget.xcconfig
+//
+// These are static library target settings for libgtest.a. It
+// is set in the "Based On:" dropdown in the "Target" info dialog.
+// This file is based on the Xcode Configuration files in:
+// http://code.google.com/p/google-toolbox-for-mac/
+//
+
+// Static libs can be included in bundles so make them position independent
+GCC_DYNAMIC_NO_PIC = NO
+
+// Static libs should not have their internal globals or external symbols
+// stripped.
+STRIP_STYLE = debugging
+
+// Let the user install by specifying the $DSTROOT with xcodebuild
+SKIP_INSTALL = NO
diff --git a/libvpx/third_party/googletest/src/xcode/Config/TestTarget.xcconfig b/libvpx/third_party/googletest/src/xcode/Config/TestTarget.xcconfig
new file mode 100644
index 0000000..e6652ba
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Config/TestTarget.xcconfig
@@ -0,0 +1,8 @@
+//
+// TestTarget.xcconfig
+//
+// These are Test target settings for the gtest framework and examples. It
+// is set in the "Based On:" dropdown in the "Target" info dialog.
+
+PRODUCT_NAME = $(TARGET_NAME)
+HEADER_SEARCH_PATHS = ../include
diff --git a/libvpx/third_party/googletest/src/xcode/Resources/Info.plist b/libvpx/third_party/googletest/src/xcode/Resources/Info.plist
new file mode 100644
index 0000000..9dd28ea
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Resources/Info.plist
@@ -0,0 +1,30 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+ <key>CFBundleDevelopmentRegion</key>
+ <string>English</string>
+ <key>CFBundleExecutable</key>
+ <string>${EXECUTABLE_NAME}</string>
+ <key>CFBundleIconFile</key>
+ <string></string>
+ <key>CFBundleIdentifier</key>
+ <string>com.google.${PRODUCT_NAME}</string>
+ <key>CFBundleInfoDictionaryVersion</key>
+ <string>6.0</string>
+ <key>CFBundlePackageType</key>
+ <string>FMWK</string>
+ <key>CFBundleSignature</key>
+ <string>????</string>
+ <key>CFBundleVersion</key>
+ <string>GTEST_VERSIONINFO_LONG</string>
+ <key>CFBundleShortVersionString</key>
+ <string>GTEST_VERSIONINFO_SHORT</string>
+ <key>CFBundleGetInfoString</key>
+ <string>${PRODUCT_NAME} GTEST_VERSIONINFO_LONG, ${GTEST_VERSIONINFO_ABOUT}</string>
+ <key>NSHumanReadableCopyright</key>
+ <string>${GTEST_VERSIONINFO_ABOUT}</string>
+ <key>CSResourcesFileMapped</key>
+ <true/>
+</dict>
+</plist>
diff --git a/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/Info.plist b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/Info.plist
new file mode 100644
index 0000000..f3852ed
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/Info.plist
@@ -0,0 +1,28 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+ <key>CFBundleDevelopmentRegion</key>
+ <string>English</string>
+ <key>CFBundleExecutable</key>
+ <string>${EXECUTABLE_NAME}</string>
+ <key>CFBundleIconFile</key>
+ <string></string>
+ <key>CFBundleIdentifier</key>
+ <string>com.google.gtest.${PRODUCT_NAME:identifier}</string>
+ <key>CFBundleInfoDictionaryVersion</key>
+ <string>6.0</string>
+ <key>CFBundleName</key>
+ <string>${PRODUCT_NAME}</string>
+ <key>CFBundlePackageType</key>
+ <string>FMWK</string>
+ <key>CFBundleShortVersionString</key>
+ <string>1.0</string>
+ <key>CFBundleSignature</key>
+ <string>????</string>
+ <key>CFBundleVersion</key>
+ <string>1.0</string>
+ <key>CSResourcesFileMapped</key>
+ <true/>
+</dict>
+</plist>
diff --git a/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/WidgetFramework.xcodeproj/project.pbxproj b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/WidgetFramework.xcodeproj/project.pbxproj
new file mode 100644
index 0000000..497617e
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/WidgetFramework.xcodeproj/project.pbxproj
@@ -0,0 +1,457 @@
+// !$*UTF8*$!
+{
+ archiveVersion = 1;
+ classes = {
+ };
+ objectVersion = 42;
+ objects = {
+
+/* Begin PBXAggregateTarget section */
+ 4024D162113D7D2400C7059E /* Test */ = {
+ isa = PBXAggregateTarget;
+ buildConfigurationList = 4024D169113D7D4600C7059E /* Build configuration list for PBXAggregateTarget "Test" */;
+ buildPhases = (
+ 4024D161113D7D2400C7059E /* ShellScript */,
+ );
+ dependencies = (
+ 4024D166113D7D3100C7059E /* PBXTargetDependency */,
+ );
+ name = Test;
+ productName = TestAndBuild;
+ };
+ 4024D1E9113D83FF00C7059E /* TestAndBuild */ = {
+ isa = PBXAggregateTarget;
+ buildConfigurationList = 4024D1F0113D842B00C7059E /* Build configuration list for PBXAggregateTarget "TestAndBuild" */;
+ buildPhases = (
+ );
+ dependencies = (
+ 4024D1ED113D840900C7059E /* PBXTargetDependency */,
+ 4024D1EF113D840D00C7059E /* PBXTargetDependency */,
+ );
+ name = TestAndBuild;
+ productName = TestAndBuild;
+ };
+/* End PBXAggregateTarget section */
+
+/* Begin PBXBuildFile section */
+ 3B7EB1250E5AEE3500C7F239 /* widget.cc in Sources */ = {isa = PBXBuildFile; fileRef = 3B7EB1230E5AEE3500C7F239 /* widget.cc */; };
+ 3B7EB1260E5AEE3500C7F239 /* widget.h in Headers */ = {isa = PBXBuildFile; fileRef = 3B7EB1240E5AEE3500C7F239 /* widget.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 3B7EB1280E5AEE4600C7F239 /* widget_test.cc in Sources */ = {isa = PBXBuildFile; fileRef = 3B7EB1270E5AEE4600C7F239 /* widget_test.cc */; };
+ 3B7EB1480E5AF3B400C7F239 /* Widget.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 8D07F2C80486CC7A007CD1D0 /* Widget.framework */; };
+ 4024D188113D7D7800C7059E /* libgtest.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 4024D185113D7D5500C7059E /* libgtest.a */; };
+ 4024D189113D7D7A00C7059E /* libgtest_main.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 4024D183113D7D5500C7059E /* libgtest_main.a */; };
+/* End PBXBuildFile section */
+
+/* Begin PBXContainerItemProxy section */
+ 3B07BDF00E3F3FAE00647869 /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 8D07F2BC0486CC7A007CD1D0;
+ remoteInfo = gTestExample;
+ };
+ 4024D165113D7D3100C7059E /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 3B07BDE90E3F3F9E00647869;
+ remoteInfo = WidgetFrameworkTest;
+ };
+ 4024D1EC113D840900C7059E /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 8D07F2BC0486CC7A007CD1D0;
+ remoteInfo = WidgetFramework;
+ };
+ 4024D1EE113D840D00C7059E /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 4024D162113D7D2400C7059E;
+ remoteInfo = Test;
+ };
+/* End PBXContainerItemProxy section */
+
+/* Begin PBXFileReference section */
+ 3B07BDEA0E3F3F9E00647869 /* WidgetFrameworkTest */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = WidgetFrameworkTest; sourceTree = BUILT_PRODUCTS_DIR; };
+ 3B7EB1230E5AEE3500C7F239 /* widget.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = widget.cc; sourceTree = "<group>"; };
+ 3B7EB1240E5AEE3500C7F239 /* widget.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = widget.h; sourceTree = "<group>"; };
+ 3B7EB1270E5AEE4600C7F239 /* widget_test.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = widget_test.cc; sourceTree = "<group>"; };
+ 4024D183113D7D5500C7059E /* libgtest_main.a */ = {isa = PBXFileReference; lastKnownFileType = archive.ar; name = libgtest_main.a; path = /usr/local/lib/libgtest_main.a; sourceTree = "<absolute>"; };
+ 4024D185113D7D5500C7059E /* libgtest.a */ = {isa = PBXFileReference; lastKnownFileType = archive.ar; name = libgtest.a; path = /usr/local/lib/libgtest.a; sourceTree = "<absolute>"; };
+ 4024D1E2113D838200C7059E /* runtests.sh */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.script.sh; path = runtests.sh; sourceTree = "<group>"; };
+ 8D07F2C70486CC7A007CD1D0 /* Info.plist */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.plist; path = Info.plist; sourceTree = "<group>"; };
+ 8D07F2C80486CC7A007CD1D0 /* Widget.framework */ = {isa = PBXFileReference; explicitFileType = wrapper.framework; includeInIndex = 0; path = Widget.framework; sourceTree = BUILT_PRODUCTS_DIR; };
+/* End PBXFileReference section */
+
+/* Begin PBXFrameworksBuildPhase section */
+ 3B07BDE80E3F3F9E00647869 /* Frameworks */ = {
+ isa = PBXFrameworksBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 4024D189113D7D7A00C7059E /* libgtest_main.a in Frameworks */,
+ 4024D188113D7D7800C7059E /* libgtest.a in Frameworks */,
+ 3B7EB1480E5AF3B400C7F239 /* Widget.framework in Frameworks */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 8D07F2C30486CC7A007CD1D0 /* Frameworks */ = {
+ isa = PBXFrameworksBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXFrameworksBuildPhase section */
+
+/* Begin PBXGroup section */
+ 034768DDFF38A45A11DB9C8B /* Products */ = {
+ isa = PBXGroup;
+ children = (
+ 8D07F2C80486CC7A007CD1D0 /* Widget.framework */,
+ 3B07BDEA0E3F3F9E00647869 /* WidgetFrameworkTest */,
+ );
+ name = Products;
+ sourceTree = "<group>";
+ };
+ 0867D691FE84028FC02AAC07 /* gTestExample */ = {
+ isa = PBXGroup;
+ children = (
+ 4024D1E1113D836C00C7059E /* Scripts */,
+ 08FB77ACFE841707C02AAC07 /* Source */,
+ 089C1665FE841158C02AAC07 /* Resources */,
+ 3B07BE350E4094E400647869 /* Test */,
+ 0867D69AFE84028FC02AAC07 /* External Frameworks and Libraries */,
+ 034768DDFF38A45A11DB9C8B /* Products */,
+ );
+ name = gTestExample;
+ sourceTree = "<group>";
+ };
+ 0867D69AFE84028FC02AAC07 /* External Frameworks and Libraries */ = {
+ isa = PBXGroup;
+ children = (
+ 4024D183113D7D5500C7059E /* libgtest_main.a */,
+ 4024D185113D7D5500C7059E /* libgtest.a */,
+ );
+ name = "External Frameworks and Libraries";
+ sourceTree = "<group>";
+ };
+ 089C1665FE841158C02AAC07 /* Resources */ = {
+ isa = PBXGroup;
+ children = (
+ 8D07F2C70486CC7A007CD1D0 /* Info.plist */,
+ );
+ name = Resources;
+ sourceTree = "<group>";
+ };
+ 08FB77ACFE841707C02AAC07 /* Source */ = {
+ isa = PBXGroup;
+ children = (
+ 3B7EB1230E5AEE3500C7F239 /* widget.cc */,
+ 3B7EB1240E5AEE3500C7F239 /* widget.h */,
+ );
+ name = Source;
+ sourceTree = "<group>";
+ };
+ 3B07BE350E4094E400647869 /* Test */ = {
+ isa = PBXGroup;
+ children = (
+ 3B7EB1270E5AEE4600C7F239 /* widget_test.cc */,
+ );
+ name = Test;
+ sourceTree = "<group>";
+ };
+ 4024D1E1113D836C00C7059E /* Scripts */ = {
+ isa = PBXGroup;
+ children = (
+ 4024D1E2113D838200C7059E /* runtests.sh */,
+ );
+ name = Scripts;
+ sourceTree = "<group>";
+ };
+/* End PBXGroup section */
+
+/* Begin PBXHeadersBuildPhase section */
+ 8D07F2BD0486CC7A007CD1D0 /* Headers */ = {
+ isa = PBXHeadersBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 3B7EB1260E5AEE3500C7F239 /* widget.h in Headers */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXHeadersBuildPhase section */
+
+/* Begin PBXNativeTarget section */
+ 3B07BDE90E3F3F9E00647869 /* WidgetFrameworkTest */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 3B07BDF40E3F3FB600647869 /* Build configuration list for PBXNativeTarget "WidgetFrameworkTest" */;
+ buildPhases = (
+ 3B07BDE70E3F3F9E00647869 /* Sources */,
+ 3B07BDE80E3F3F9E00647869 /* Frameworks */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ 3B07BDF10E3F3FAE00647869 /* PBXTargetDependency */,
+ );
+ name = WidgetFrameworkTest;
+ productName = gTestExampleTest;
+ productReference = 3B07BDEA0E3F3F9E00647869 /* WidgetFrameworkTest */;
+ productType = "com.apple.product-type.tool";
+ };
+ 8D07F2BC0486CC7A007CD1D0 /* WidgetFramework */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 4FADC24208B4156D00ABE55E /* Build configuration list for PBXNativeTarget "WidgetFramework" */;
+ buildPhases = (
+ 8D07F2C10486CC7A007CD1D0 /* Sources */,
+ 8D07F2C30486CC7A007CD1D0 /* Frameworks */,
+ 8D07F2BD0486CC7A007CD1D0 /* Headers */,
+ 8D07F2BF0486CC7A007CD1D0 /* Resources */,
+ 8D07F2C50486CC7A007CD1D0 /* Rez */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ );
+ name = WidgetFramework;
+ productInstallPath = "$(HOME)/Library/Frameworks";
+ productName = gTestExample;
+ productReference = 8D07F2C80486CC7A007CD1D0 /* Widget.framework */;
+ productType = "com.apple.product-type.framework";
+ };
+/* End PBXNativeTarget section */
+
+/* Begin PBXProject section */
+ 0867D690FE84028FC02AAC07 /* Project object */ = {
+ isa = PBXProject;
+ buildConfigurationList = 4FADC24608B4156D00ABE55E /* Build configuration list for PBXProject "WidgetFramework" */;
+ compatibilityVersion = "Xcode 2.4";
+ hasScannedForEncodings = 1;
+ mainGroup = 0867D691FE84028FC02AAC07 /* gTestExample */;
+ productRefGroup = 034768DDFF38A45A11DB9C8B /* Products */;
+ projectDirPath = "";
+ projectRoot = "";
+ targets = (
+ 8D07F2BC0486CC7A007CD1D0 /* WidgetFramework */,
+ 3B07BDE90E3F3F9E00647869 /* WidgetFrameworkTest */,
+ 4024D162113D7D2400C7059E /* Test */,
+ 4024D1E9113D83FF00C7059E /* TestAndBuild */,
+ );
+ };
+/* End PBXProject section */
+
+/* Begin PBXResourcesBuildPhase section */
+ 8D07F2BF0486CC7A007CD1D0 /* Resources */ = {
+ isa = PBXResourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXResourcesBuildPhase section */
+
+/* Begin PBXRezBuildPhase section */
+ 8D07F2C50486CC7A007CD1D0 /* Rez */ = {
+ isa = PBXRezBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXRezBuildPhase section */
+
+/* Begin PBXShellScriptBuildPhase section */
+ 4024D161113D7D2400C7059E /* ShellScript */ = {
+ isa = PBXShellScriptBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ );
+ inputPaths = (
+ );
+ outputPaths = (
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ shellPath = /bin/sh;
+ shellScript = "/bin/bash $SRCROOT/runtests.sh $BUILT_PRODUCTS_DIR/WidgetFrameworkTest\n";
+ };
+/* End PBXShellScriptBuildPhase section */
+
+/* Begin PBXSourcesBuildPhase section */
+ 3B07BDE70E3F3F9E00647869 /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 3B7EB1280E5AEE4600C7F239 /* widget_test.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 8D07F2C10486CC7A007CD1D0 /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 3B7EB1250E5AEE3500C7F239 /* widget.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXSourcesBuildPhase section */
+
+/* Begin PBXTargetDependency section */
+ 3B07BDF10E3F3FAE00647869 /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 8D07F2BC0486CC7A007CD1D0 /* WidgetFramework */;
+ targetProxy = 3B07BDF00E3F3FAE00647869 /* PBXContainerItemProxy */;
+ };
+ 4024D166113D7D3100C7059E /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 3B07BDE90E3F3F9E00647869 /* WidgetFrameworkTest */;
+ targetProxy = 4024D165113D7D3100C7059E /* PBXContainerItemProxy */;
+ };
+ 4024D1ED113D840900C7059E /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 8D07F2BC0486CC7A007CD1D0 /* WidgetFramework */;
+ targetProxy = 4024D1EC113D840900C7059E /* PBXContainerItemProxy */;
+ };
+ 4024D1EF113D840D00C7059E /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 4024D162113D7D2400C7059E /* Test */;
+ targetProxy = 4024D1EE113D840D00C7059E /* PBXContainerItemProxy */;
+ };
+/* End PBXTargetDependency section */
+
+/* Begin XCBuildConfiguration section */
+ 3B07BDEC0E3F3F9F00647869 /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = WidgetFrameworkTest;
+ };
+ name = Debug;
+ };
+ 3B07BDED0E3F3F9F00647869 /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = WidgetFrameworkTest;
+ };
+ name = Release;
+ };
+ 4024D163113D7D2400C7059E /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = TestAndBuild;
+ };
+ name = Debug;
+ };
+ 4024D164113D7D2400C7059E /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = TestAndBuild;
+ };
+ name = Release;
+ };
+ 4024D1EA113D83FF00C7059E /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = TestAndBuild;
+ };
+ name = Debug;
+ };
+ 4024D1EB113D83FF00C7059E /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = TestAndBuild;
+ };
+ name = Release;
+ };
+ 4FADC24308B4156D00ABE55E /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ DYLIB_COMPATIBILITY_VERSION = 1;
+ DYLIB_CURRENT_VERSION = 1;
+ FRAMEWORK_VERSION = A;
+ INFOPLIST_FILE = Info.plist;
+ INSTALL_PATH = "@loader_path/../Frameworks";
+ PRODUCT_NAME = Widget;
+ };
+ name = Debug;
+ };
+ 4FADC24408B4156D00ABE55E /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ DYLIB_COMPATIBILITY_VERSION = 1;
+ DYLIB_CURRENT_VERSION = 1;
+ FRAMEWORK_VERSION = A;
+ INFOPLIST_FILE = Info.plist;
+ INSTALL_PATH = "@loader_path/../Frameworks";
+ PRODUCT_NAME = Widget;
+ };
+ name = Release;
+ };
+ 4FADC24708B4156D00ABE55E /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ GCC_VERSION = 4.0;
+ SDKROOT = /Developer/SDKs/MacOSX10.4u.sdk;
+ };
+ name = Debug;
+ };
+ 4FADC24808B4156D00ABE55E /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ GCC_VERSION = 4.0;
+ SDKROOT = /Developer/SDKs/MacOSX10.4u.sdk;
+ };
+ name = Release;
+ };
+/* End XCBuildConfiguration section */
+
+/* Begin XCConfigurationList section */
+ 3B07BDF40E3F3FB600647869 /* Build configuration list for PBXNativeTarget "WidgetFrameworkTest" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 3B07BDEC0E3F3F9F00647869 /* Debug */,
+ 3B07BDED0E3F3F9F00647869 /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 4024D169113D7D4600C7059E /* Build configuration list for PBXAggregateTarget "Test" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 4024D163113D7D2400C7059E /* Debug */,
+ 4024D164113D7D2400C7059E /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 4024D1F0113D842B00C7059E /* Build configuration list for PBXAggregateTarget "TestAndBuild" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 4024D1EA113D83FF00C7059E /* Debug */,
+ 4024D1EB113D83FF00C7059E /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 4FADC24208B4156D00ABE55E /* Build configuration list for PBXNativeTarget "WidgetFramework" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 4FADC24308B4156D00ABE55E /* Debug */,
+ 4FADC24408B4156D00ABE55E /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 4FADC24608B4156D00ABE55E /* Build configuration list for PBXProject "WidgetFramework" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 4FADC24708B4156D00ABE55E /* Debug */,
+ 4FADC24808B4156D00ABE55E /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+/* End XCConfigurationList section */
+ };
+ rootObject = 0867D690FE84028FC02AAC07 /* Project object */;
+}
diff --git a/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/runtests.sh b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/runtests.sh
new file mode 100755
index 0000000..4a0d413
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/runtests.sh
@@ -0,0 +1,62 @@
+#!/bin/bash
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# Executes the samples and tests for the Google Test Framework.
+
+# Help the dynamic linker find the path to the libraries.
+export DYLD_FRAMEWORK_PATH=$BUILT_PRODUCTS_DIR
+export DYLD_LIBRARY_PATH=$BUILT_PRODUCTS_DIR
+
+# Create some executables.
+test_executables=$@
+
+# Now execute each one in turn keeping track of how many succeeded and failed.
+succeeded=0
+failed=0
+failed_list=()
+for test in ${test_executables[*]}; do
+ "$test"
+ result=$?
+ if [ $result -eq 0 ]; then
+ succeeded=$(( $succeeded + 1 ))
+ else
+ failed=$(( failed + 1 ))
+ failed_list="$failed_list $test"
+ fi
+done
+
+# Report the successes and failures to the console.
+echo "Tests complete with $succeeded successes and $failed failures."
+if [ $failed -ne 0 ]; then
+ echo "The following tests failed:"
+ echo $failed_list
+fi
+exit $failed
diff --git a/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget.cc b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget.cc
new file mode 100644
index 0000000..bfc4e7f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget.cc
@@ -0,0 +1,63 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: preston.a.jackson@gmail.com (Preston Jackson)
+//
+// Google Test - FrameworkSample
+// widget.cc
+//
+
+// Widget is a very simple class used for demonstrating the use of gtest
+
+#include "widget.h"
+
+Widget::Widget(int number, const std::string& name)
+ : number_(number),
+ name_(name) {}
+
+Widget::~Widget() {}
+
+float Widget::GetFloatValue() const {
+ return number_;
+}
+
+int Widget::GetIntValue() const {
+ return static_cast<int>(number_);
+}
+
+std::string Widget::GetStringValue() const {
+ return name_;
+}
+
+void Widget::GetCharPtrValue(char* buffer, size_t max_size) const {
+ // Copy the char* representation of name_ into buffer, up to max_size.
+ strncpy(buffer, name_.c_str(), max_size-1);
+ buffer[max_size-1] = '\0';
+ return;
+}
diff --git a/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget.h b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget.h
new file mode 100644
index 0000000..0c55cdc
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget.h
@@ -0,0 +1,59 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: preston.a.jackson@gmail.com (Preston Jackson)
+//
+// Google Test - FrameworkSample
+// widget.h
+//
+
+// Widget is a very simple class used for demonstrating the use of gtest. It
+// simply stores two values a string and an integer, which are returned via
+// public accessors in multiple forms.
+
+#import <string>
+
+class Widget {
+ public:
+ Widget(int number, const std::string& name);
+ ~Widget();
+
+ // Public accessors to number data
+ float GetFloatValue() const;
+ int GetIntValue() const;
+
+ // Public accessors to the string data
+ std::string GetStringValue() const;
+ void GetCharPtrValue(char* buffer, size_t max_size) const;
+
+ private:
+ // Data members
+ float number_;
+ std::string name_;
+};
diff --git a/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget_test.cc b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget_test.cc
new file mode 100644
index 0000000..8725994
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Samples/FrameworkSample/widget_test.cc
@@ -0,0 +1,68 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+// * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+// * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: preston.a.jackson@gmail.com (Preston Jackson)
+//
+// Google Test - FrameworkSample
+// widget_test.cc
+//
+
+// This is a simple test file for the Widget class in the Widget.framework
+
+#include <string>
+#include "gtest/gtest.h"
+
+#include <Widget/widget.h>
+
+// This test verifies that the constructor sets the internal state of the
+// Widget class correctly.
+TEST(WidgetInitializerTest, TestConstructor) {
+ Widget widget(1.0f, "name");
+ EXPECT_FLOAT_EQ(1.0f, widget.GetFloatValue());
+ EXPECT_EQ(std::string("name"), widget.GetStringValue());
+}
+
+// This test verifies the conversion of the float and string values to int and
+// char*, respectively.
+TEST(WidgetInitializerTest, TestConversion) {
+ Widget widget(1.0f, "name");
+ EXPECT_EQ(1, widget.GetIntValue());
+
+ size_t max_size = 128;
+ char buffer[max_size];
+ widget.GetCharPtrValue(buffer, max_size);
+ EXPECT_STREQ("name", buffer);
+}
+
+// Use the Google Test main that is linked into the framework. It does something
+// like this:
+// int main(int argc, char** argv) {
+// testing::InitGoogleTest(&argc, argv);
+// return RUN_ALL_TESTS();
+// }
diff --git a/libvpx/third_party/googletest/src/xcode/Scripts/runtests.sh b/libvpx/third_party/googletest/src/xcode/Scripts/runtests.sh
new file mode 100755
index 0000000..3fc229f
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Scripts/runtests.sh
@@ -0,0 +1,65 @@
+#!/bin/bash
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# Executes the samples and tests for the Google Test Framework.
+
+# Help the dynamic linker find the path to the libraries.
+export DYLD_FRAMEWORK_PATH=$BUILT_PRODUCTS_DIR
+export DYLD_LIBRARY_PATH=$BUILT_PRODUCTS_DIR
+
+# Create some executables.
+test_executables=("$BUILT_PRODUCTS_DIR/gtest_unittest-framework"
+ "$BUILT_PRODUCTS_DIR/gtest_unittest"
+ "$BUILT_PRODUCTS_DIR/sample1_unittest-framework"
+ "$BUILT_PRODUCTS_DIR/sample1_unittest-static")
+
+# Now execute each one in turn keeping track of how many succeeded and failed.
+succeeded=0
+failed=0
+failed_list=()
+for test in ${test_executables[*]}; do
+ "$test"
+ result=$?
+ if [ $result -eq 0 ]; then
+ succeeded=$(( $succeeded + 1 ))
+ else
+ failed=$(( failed + 1 ))
+ failed_list="$failed_list $test"
+ fi
+done
+
+# Report the successes and failures to the console.
+echo "Tests complete with $succeeded successes and $failed failures."
+if [ $failed -ne 0 ]; then
+ echo "The following tests failed:"
+ echo $failed_list
+fi
+exit $failed
diff --git a/libvpx/third_party/googletest/src/xcode/Scripts/versiongenerate.py b/libvpx/third_party/googletest/src/xcode/Scripts/versiongenerate.py
new file mode 100755
index 0000000..81de8c9
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/Scripts/versiongenerate.py
@@ -0,0 +1,100 @@
+#!/usr/bin/env python
+#
+# Copyright 2008, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+"""A script to prepare version informtion for use the gtest Info.plist file.
+
+ This script extracts the version information from the configure.ac file and
+ uses it to generate a header file containing the same information. The
+ #defines in this header file will be included in during the generation of
+ the Info.plist of the framework, giving the correct value to the version
+ shown in the Finder.
+
+ This script makes the following assumptions (these are faults of the script,
+ not problems with the Autoconf):
+ 1. The AC_INIT macro will be contained within the first 1024 characters
+ of configure.ac
+ 2. The version string will be 3 integers separated by periods and will be
+ surrounded by squre brackets, "[" and "]" (e.g. [1.0.1]). The first
+ segment represents the major version, the second represents the minor
+ version and the third represents the fix version.
+ 3. No ")" character exists between the opening "(" and closing ")" of
+ AC_INIT, including in comments and character strings.
+"""
+
+import sys
+import re
+
+# Read the command line argument (the output directory for Version.h)
+if (len(sys.argv) < 3):
+ print "Usage: versiongenerate.py input_dir output_dir"
+ sys.exit(1)
+else:
+ input_dir = sys.argv[1]
+ output_dir = sys.argv[2]
+
+# Read the first 1024 characters of the configure.ac file
+config_file = open("%s/configure.ac" % input_dir, 'r')
+buffer_size = 1024
+opening_string = config_file.read(buffer_size)
+config_file.close()
+
+# Extract the version string from the AC_INIT macro
+# The following init_expression means:
+# Extract three integers separated by periods and surrounded by squre
+# brackets(e.g. "[1.0.1]") between "AC_INIT(" and ")". Do not be greedy
+# (*? is the non-greedy flag) since that would pull in everything between
+# the first "(" and the last ")" in the file.
+version_expression = re.compile(r"AC_INIT\(.*?\[(\d+)\.(\d+)\.(\d+)\].*?\)",
+ re.DOTALL)
+version_values = version_expression.search(opening_string)
+major_version = version_values.group(1)
+minor_version = version_values.group(2)
+fix_version = version_values.group(3)
+
+# Write the version information to a header file to be included in the
+# Info.plist file.
+file_data = """//
+// DO NOT MODIFY THIS FILE (but you can delete it)
+//
+// This file is autogenerated by the versiongenerate.py script. This script
+// is executed in a "Run Script" build phase when creating gtest.framework. This
+// header file is not used during compilation of C-source. Rather, it simply
+// defines some version strings for substitution in the Info.plist. Because of
+// this, we are not not restricted to C-syntax nor are we using include guards.
+//
+
+#define GTEST_VERSIONINFO_SHORT %s.%s
+#define GTEST_VERSIONINFO_LONG %s.%s.%s
+
+""" % (major_version, minor_version, major_version, minor_version, fix_version)
+version_file = open("%s/Version.h" % output_dir, 'w')
+version_file.write(file_data)
+version_file.close()
diff --git a/libvpx/third_party/googletest/src/xcode/gtest.xcodeproj/project.pbxproj b/libvpx/third_party/googletest/src/xcode/gtest.xcodeproj/project.pbxproj
new file mode 100644
index 0000000..74a7815
--- /dev/null
+++ b/libvpx/third_party/googletest/src/xcode/gtest.xcodeproj/project.pbxproj
@@ -0,0 +1,1084 @@
+// !$*UTF8*$!
+{
+ archiveVersion = 1;
+ classes = {
+ };
+ objectVersion = 42;
+ objects = {
+
+/* Begin PBXAggregateTarget section */
+ 3B238F5F0E828B5400846E11 /* Check */ = {
+ isa = PBXAggregateTarget;
+ buildConfigurationList = 3B238FA30E828BB600846E11 /* Build configuration list for PBXAggregateTarget "Check" */;
+ buildPhases = (
+ 3B238F5E0E828B5400846E11 /* ShellScript */,
+ );
+ dependencies = (
+ 40899F9D0FFA740F000B29AE /* PBXTargetDependency */,
+ 40C849F7101A43440083642A /* PBXTargetDependency */,
+ 4089A0980FFAD34A000B29AE /* PBXTargetDependency */,
+ 40C849F9101A43490083642A /* PBXTargetDependency */,
+ );
+ name = Check;
+ productName = Check;
+ };
+ 40C44ADC0E3798F4008FCC51 /* Version Info */ = {
+ isa = PBXAggregateTarget;
+ buildConfigurationList = 40C44AE40E379905008FCC51 /* Build configuration list for PBXAggregateTarget "Version Info" */;
+ buildPhases = (
+ 40C44ADB0E3798F4008FCC51 /* Generate Version.h */,
+ );
+ comments = "The generation of Version.h must be performed in its own target. Since the Info.plist is preprocessed before any of the other build phases in gtest, the Version.h file would not be ready if included as a build phase of that target.";
+ dependencies = (
+ );
+ name = "Version Info";
+ productName = Version.h;
+ };
+/* End PBXAggregateTarget section */
+
+/* Begin PBXBuildFile section */
+ 224A12A30E9EADCC00BD17FD /* gtest-test-part.h in Headers */ = {isa = PBXBuildFile; fileRef = 224A12A20E9EADCC00BD17FD /* gtest-test-part.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 3BF6F2A00E79B5AD000F2EEE /* gtest-type-util.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 3BF6F29F0E79B5AD000F2EEE /* gtest-type-util.h */; };
+ 3BF6F2A50E79B616000F2EEE /* gtest-typed-test.h in Headers */ = {isa = PBXBuildFile; fileRef = 3BF6F2A40E79B616000F2EEE /* gtest-typed-test.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 404884380E2F799B00CF7658 /* gtest-death-test.h in Headers */ = {isa = PBXBuildFile; fileRef = 404883DB0E2F799B00CF7658 /* gtest-death-test.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 404884390E2F799B00CF7658 /* gtest-message.h in Headers */ = {isa = PBXBuildFile; fileRef = 404883DC0E2F799B00CF7658 /* gtest-message.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 4048843A0E2F799B00CF7658 /* gtest-spi.h in Headers */ = {isa = PBXBuildFile; fileRef = 404883DD0E2F799B00CF7658 /* gtest-spi.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 4048843B0E2F799B00CF7658 /* gtest.h in Headers */ = {isa = PBXBuildFile; fileRef = 404883DE0E2F799B00CF7658 /* gtest.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 4048843C0E2F799B00CF7658 /* gtest_pred_impl.h in Headers */ = {isa = PBXBuildFile; fileRef = 404883DF0E2F799B00CF7658 /* gtest_pred_impl.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 4048843D0E2F799B00CF7658 /* gtest_prod.h in Headers */ = {isa = PBXBuildFile; fileRef = 404883E00E2F799B00CF7658 /* gtest_prod.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 404884500E2F799B00CF7658 /* README in Resources */ = {isa = PBXBuildFile; fileRef = 404883F60E2F799B00CF7658 /* README */; };
+ 404884A00E2F7BE600CF7658 /* gtest-death-test-internal.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 404883E20E2F799B00CF7658 /* gtest-death-test-internal.h */; };
+ 404884A10E2F7BE600CF7658 /* gtest-filepath.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 404883E30E2F799B00CF7658 /* gtest-filepath.h */; };
+ 404884A20E2F7BE600CF7658 /* gtest-internal.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 404883E40E2F799B00CF7658 /* gtest-internal.h */; };
+ 404884A30E2F7BE600CF7658 /* gtest-port.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 404883E50E2F799B00CF7658 /* gtest-port.h */; };
+ 404884A40E2F7BE600CF7658 /* gtest-string.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 404883E60E2F799B00CF7658 /* gtest-string.h */; };
+ 404884AC0E2F7CD900CF7658 /* CHANGES in Resources */ = {isa = PBXBuildFile; fileRef = 404884A90E2F7CD900CF7658 /* CHANGES */; };
+ 404884AD0E2F7CD900CF7658 /* CONTRIBUTORS in Resources */ = {isa = PBXBuildFile; fileRef = 404884AA0E2F7CD900CF7658 /* CONTRIBUTORS */; };
+ 404884AE0E2F7CD900CF7658 /* COPYING in Resources */ = {isa = PBXBuildFile; fileRef = 404884AB0E2F7CD900CF7658 /* COPYING */; };
+ 40899F3A0FFA70D4000B29AE /* gtest-all.cc in Sources */ = {isa = PBXBuildFile; fileRef = 224A12A10E9EADA700BD17FD /* gtest-all.cc */; };
+ 40899F500FFA7281000B29AE /* gtest-tuple.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 40899F4D0FFA7271000B29AE /* gtest-tuple.h */; };
+ 40899F530FFA72A0000B29AE /* gtest_unittest.cc in Sources */ = {isa = PBXBuildFile; fileRef = 3B238C120E7FE13C00846E11 /* gtest_unittest.cc */; };
+ 4089A0440FFAD1BE000B29AE /* sample1.cc in Sources */ = {isa = PBXBuildFile; fileRef = 4089A02C0FFACF7F000B29AE /* sample1.cc */; };
+ 4089A0460FFAD1BE000B29AE /* sample1_unittest.cc in Sources */ = {isa = PBXBuildFile; fileRef = 4089A02E0FFACF7F000B29AE /* sample1_unittest.cc */; };
+ 40C848FF101A21150083642A /* gtest-all.cc in Sources */ = {isa = PBXBuildFile; fileRef = 224A12A10E9EADA700BD17FD /* gtest-all.cc */; };
+ 40C84915101A21DF0083642A /* gtest_main.cc in Sources */ = {isa = PBXBuildFile; fileRef = 4048840D0E2F799B00CF7658 /* gtest_main.cc */; };
+ 40C84916101A235B0083642A /* libgtest_main.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 40C8490B101A217E0083642A /* libgtest_main.a */; };
+ 40C84921101A23AD0083642A /* libgtest_main.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 40C8490B101A217E0083642A /* libgtest_main.a */; };
+ 40C84978101A36540083642A /* libgtest_main.a in Resources */ = {isa = PBXBuildFile; fileRef = 40C8490B101A217E0083642A /* libgtest_main.a */; };
+ 40C84980101A36850083642A /* gtest_unittest.cc in Sources */ = {isa = PBXBuildFile; fileRef = 3B238C120E7FE13C00846E11 /* gtest_unittest.cc */; };
+ 40C84982101A36850083642A /* libgtest.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 40C848FA101A209C0083642A /* libgtest.a */; };
+ 40C84983101A36850083642A /* libgtest_main.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 40C8490B101A217E0083642A /* libgtest_main.a */; };
+ 40C8498F101A36A60083642A /* sample1.cc in Sources */ = {isa = PBXBuildFile; fileRef = 4089A02C0FFACF7F000B29AE /* sample1.cc */; };
+ 40C84990101A36A60083642A /* sample1_unittest.cc in Sources */ = {isa = PBXBuildFile; fileRef = 4089A02E0FFACF7F000B29AE /* sample1_unittest.cc */; };
+ 40C84992101A36A60083642A /* libgtest.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 40C848FA101A209C0083642A /* libgtest.a */; };
+ 40C84993101A36A60083642A /* libgtest_main.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 40C8490B101A217E0083642A /* libgtest_main.a */; };
+ 40C849A2101A37050083642A /* gtest.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 4539C8FF0EC27F6400A70F4C /* gtest.framework */; };
+ 40C849A4101A37150083642A /* gtest.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 4539C8FF0EC27F6400A70F4C /* gtest.framework */; };
+ 4539C9340EC280AE00A70F4C /* gtest-param-test.h in Headers */ = {isa = PBXBuildFile; fileRef = 4539C9330EC280AE00A70F4C /* gtest-param-test.h */; settings = {ATTRIBUTES = (Public, ); }; };
+ 4539C9380EC280E200A70F4C /* gtest-linked_ptr.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 4539C9350EC280E200A70F4C /* gtest-linked_ptr.h */; };
+ 4539C9390EC280E200A70F4C /* gtest-param-util-generated.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 4539C9360EC280E200A70F4C /* gtest-param-util-generated.h */; };
+ 4539C93A0EC280E200A70F4C /* gtest-param-util.h in Copy Headers Internal */ = {isa = PBXBuildFile; fileRef = 4539C9370EC280E200A70F4C /* gtest-param-util.h */; };
+ 4567C8181264FF71007740BE /* gtest-printers.h in Headers */ = {isa = PBXBuildFile; fileRef = 4567C8171264FF71007740BE /* gtest-printers.h */; settings = {ATTRIBUTES = (Public, ); }; };
+/* End PBXBuildFile section */
+
+/* Begin PBXContainerItemProxy section */
+ 40899F9C0FFA740F000B29AE /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40899F420FFA7184000B29AE;
+ remoteInfo = gtest_unittest;
+ };
+ 4089A0970FFAD34A000B29AE /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 4089A0120FFACEFC000B29AE;
+ remoteInfo = sample1_unittest;
+ };
+ 408BEC0F1046CFE900DEF522 /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C848F9101A209C0083642A;
+ remoteInfo = "gtest-static";
+ };
+ 40C44AE50E379922008FCC51 /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C44ADC0E3798F4008FCC51;
+ remoteInfo = Version.h;
+ };
+ 40C8497C101A36850083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C848F9101A209C0083642A;
+ remoteInfo = "gtest-static";
+ };
+ 40C8497E101A36850083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C8490A101A217E0083642A;
+ remoteInfo = "gtest_main-static";
+ };
+ 40C8498B101A36A60083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C848F9101A209C0083642A;
+ remoteInfo = "gtest-static";
+ };
+ 40C8498D101A36A60083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C8490A101A217E0083642A;
+ remoteInfo = "gtest_main-static";
+ };
+ 40C8499B101A36DC0083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C8490A101A217E0083642A;
+ remoteInfo = "gtest_main-static";
+ };
+ 40C8499D101A36E50083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 8D07F2BC0486CC7A007CD1D0;
+ remoteInfo = "gtest-framework";
+ };
+ 40C8499F101A36F10083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 8D07F2BC0486CC7A007CD1D0;
+ remoteInfo = "gtest-framework";
+ };
+ 40C849F6101A43440083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C8497A101A36850083642A;
+ remoteInfo = "gtest_unittest-static";
+ };
+ 40C849F8101A43490083642A /* PBXContainerItemProxy */ = {
+ isa = PBXContainerItemProxy;
+ containerPortal = 0867D690FE84028FC02AAC07 /* Project object */;
+ proxyType = 1;
+ remoteGlobalIDString = 40C84989101A36A60083642A;
+ remoteInfo = "sample1_unittest-static";
+ };
+/* End PBXContainerItemProxy section */
+
+/* Begin PBXCopyFilesBuildPhase section */
+ 404884A50E2F7C0400CF7658 /* Copy Headers Internal */ = {
+ isa = PBXCopyFilesBuildPhase;
+ buildActionMask = 2147483647;
+ dstPath = Headers/internal;
+ dstSubfolderSpec = 6;
+ files = (
+ 404884A00E2F7BE600CF7658 /* gtest-death-test-internal.h in Copy Headers Internal */,
+ 404884A10E2F7BE600CF7658 /* gtest-filepath.h in Copy Headers Internal */,
+ 404884A20E2F7BE600CF7658 /* gtest-internal.h in Copy Headers Internal */,
+ 4539C9380EC280E200A70F4C /* gtest-linked_ptr.h in Copy Headers Internal */,
+ 4539C9390EC280E200A70F4C /* gtest-param-util-generated.h in Copy Headers Internal */,
+ 4539C93A0EC280E200A70F4C /* gtest-param-util.h in Copy Headers Internal */,
+ 404884A30E2F7BE600CF7658 /* gtest-port.h in Copy Headers Internal */,
+ 404884A40E2F7BE600CF7658 /* gtest-string.h in Copy Headers Internal */,
+ 40899F500FFA7281000B29AE /* gtest-tuple.h in Copy Headers Internal */,
+ 3BF6F2A00E79B5AD000F2EEE /* gtest-type-util.h in Copy Headers Internal */,
+ );
+ name = "Copy Headers Internal";
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXCopyFilesBuildPhase section */
+
+/* Begin PBXFileReference section */
+ 224A12A10E9EADA700BD17FD /* gtest-all.cc */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.cpp.cpp; path = "gtest-all.cc"; sourceTree = "<group>"; };
+ 224A12A20E9EADCC00BD17FD /* gtest-test-part.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = "gtest-test-part.h"; sourceTree = "<group>"; };
+ 3B238C120E7FE13C00846E11 /* gtest_unittest.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = gtest_unittest.cc; sourceTree = "<group>"; };
+ 3B87D2100E96B92E000D1852 /* runtests.sh */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.script.sh; path = runtests.sh; sourceTree = "<group>"; };
+ 3BF6F29F0E79B5AD000F2EEE /* gtest-type-util.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-type-util.h"; sourceTree = "<group>"; };
+ 3BF6F2A40E79B616000F2EEE /* gtest-typed-test.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-typed-test.h"; sourceTree = "<group>"; };
+ 403EE37C0E377822004BD1E2 /* versiongenerate.py */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.script.python; path = versiongenerate.py; sourceTree = "<group>"; };
+ 404883DB0E2F799B00CF7658 /* gtest-death-test.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-death-test.h"; sourceTree = "<group>"; };
+ 404883DC0E2F799B00CF7658 /* gtest-message.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-message.h"; sourceTree = "<group>"; };
+ 404883DD0E2F799B00CF7658 /* gtest-spi.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-spi.h"; sourceTree = "<group>"; };
+ 404883DE0E2F799B00CF7658 /* gtest.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = gtest.h; sourceTree = "<group>"; };
+ 404883DF0E2F799B00CF7658 /* gtest_pred_impl.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = gtest_pred_impl.h; sourceTree = "<group>"; };
+ 404883E00E2F799B00CF7658 /* gtest_prod.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = gtest_prod.h; sourceTree = "<group>"; };
+ 404883E20E2F799B00CF7658 /* gtest-death-test-internal.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-death-test-internal.h"; sourceTree = "<group>"; };
+ 404883E30E2F799B00CF7658 /* gtest-filepath.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-filepath.h"; sourceTree = "<group>"; };
+ 404883E40E2F799B00CF7658 /* gtest-internal.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-internal.h"; sourceTree = "<group>"; };
+ 404883E50E2F799B00CF7658 /* gtest-port.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-port.h"; sourceTree = "<group>"; };
+ 404883E60E2F799B00CF7658 /* gtest-string.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-string.h"; sourceTree = "<group>"; };
+ 404883F60E2F799B00CF7658 /* README */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; name = README; path = ../README; sourceTree = SOURCE_ROOT; };
+ 4048840D0E2F799B00CF7658 /* gtest_main.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = gtest_main.cc; sourceTree = "<group>"; };
+ 404884A90E2F7CD900CF7658 /* CHANGES */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; name = CHANGES; path = ../CHANGES; sourceTree = SOURCE_ROOT; };
+ 404884AA0E2F7CD900CF7658 /* CONTRIBUTORS */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; name = CONTRIBUTORS; path = ../CONTRIBUTORS; sourceTree = SOURCE_ROOT; };
+ 404884AB0E2F7CD900CF7658 /* COPYING */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; name = COPYING; path = ../COPYING; sourceTree = SOURCE_ROOT; };
+ 40899F430FFA7184000B29AE /* gtest_unittest-framework */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = "gtest_unittest-framework"; sourceTree = BUILT_PRODUCTS_DIR; };
+ 40899F4D0FFA7271000B29AE /* gtest-tuple.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-tuple.h"; sourceTree = "<group>"; };
+ 40899FB30FFA7567000B29AE /* StaticLibraryTarget.xcconfig */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.xcconfig; path = StaticLibraryTarget.xcconfig; sourceTree = "<group>"; };
+ 4089A0130FFACEFC000B29AE /* sample1_unittest-framework */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = "sample1_unittest-framework"; sourceTree = BUILT_PRODUCTS_DIR; };
+ 4089A02C0FFACF7F000B29AE /* sample1.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = sample1.cc; sourceTree = "<group>"; };
+ 4089A02D0FFACF7F000B29AE /* sample1.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = sample1.h; sourceTree = "<group>"; };
+ 4089A02E0FFACF7F000B29AE /* sample1_unittest.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = sample1_unittest.cc; sourceTree = "<group>"; };
+ 40C848FA101A209C0083642A /* libgtest.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libgtest.a; sourceTree = BUILT_PRODUCTS_DIR; };
+ 40C8490B101A217E0083642A /* libgtest_main.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libgtest_main.a; sourceTree = BUILT_PRODUCTS_DIR; };
+ 40C84987101A36850083642A /* gtest_unittest */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = gtest_unittest; sourceTree = BUILT_PRODUCTS_DIR; };
+ 40C84997101A36A60083642A /* sample1_unittest-static */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = "sample1_unittest-static"; sourceTree = BUILT_PRODUCTS_DIR; };
+ 40D4CDF10E30E07400294801 /* DebugProject.xcconfig */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.xcconfig; path = DebugProject.xcconfig; sourceTree = "<group>"; };
+ 40D4CDF20E30E07400294801 /* FrameworkTarget.xcconfig */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.xcconfig; path = FrameworkTarget.xcconfig; sourceTree = "<group>"; };
+ 40D4CDF30E30E07400294801 /* General.xcconfig */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.xcconfig; path = General.xcconfig; sourceTree = "<group>"; };
+ 40D4CDF40E30E07400294801 /* ReleaseProject.xcconfig */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.xcconfig; path = ReleaseProject.xcconfig; sourceTree = "<group>"; };
+ 40D4CF510E30F5E200294801 /* Info.plist */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
+ 4539C8FF0EC27F6400A70F4C /* gtest.framework */ = {isa = PBXFileReference; explicitFileType = wrapper.framework; includeInIndex = 0; path = gtest.framework; sourceTree = BUILT_PRODUCTS_DIR; };
+ 4539C9330EC280AE00A70F4C /* gtest-param-test.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-param-test.h"; sourceTree = "<group>"; };
+ 4539C9350EC280E200A70F4C /* gtest-linked_ptr.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-linked_ptr.h"; sourceTree = "<group>"; };
+ 4539C9360EC280E200A70F4C /* gtest-param-util-generated.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-param-util-generated.h"; sourceTree = "<group>"; };
+ 4539C9370EC280E200A70F4C /* gtest-param-util.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-param-util.h"; sourceTree = "<group>"; };
+ 4567C8171264FF71007740BE /* gtest-printers.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "gtest-printers.h"; sourceTree = "<group>"; };
+/* End PBXFileReference section */
+
+/* Begin PBXFrameworksBuildPhase section */
+ 40899F410FFA7184000B29AE /* Frameworks */ = {
+ isa = PBXFrameworksBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40C849A4101A37150083642A /* gtest.framework in Frameworks */,
+ 40C84916101A235B0083642A /* libgtest_main.a in Frameworks */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 4089A0110FFACEFC000B29AE /* Frameworks */ = {
+ isa = PBXFrameworksBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40C849A2101A37050083642A /* gtest.framework in Frameworks */,
+ 40C84921101A23AD0083642A /* libgtest_main.a in Frameworks */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 40C84981101A36850083642A /* Frameworks */ = {
+ isa = PBXFrameworksBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40C84982101A36850083642A /* libgtest.a in Frameworks */,
+ 40C84983101A36850083642A /* libgtest_main.a in Frameworks */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 40C84991101A36A60083642A /* Frameworks */ = {
+ isa = PBXFrameworksBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40C84992101A36A60083642A /* libgtest.a in Frameworks */,
+ 40C84993101A36A60083642A /* libgtest_main.a in Frameworks */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXFrameworksBuildPhase section */
+
+/* Begin PBXGroup section */
+ 034768DDFF38A45A11DB9C8B /* Products */ = {
+ isa = PBXGroup;
+ children = (
+ 4539C8FF0EC27F6400A70F4C /* gtest.framework */,
+ 40C848FA101A209C0083642A /* libgtest.a */,
+ 40C8490B101A217E0083642A /* libgtest_main.a */,
+ 40899F430FFA7184000B29AE /* gtest_unittest-framework */,
+ 40C84987101A36850083642A /* gtest_unittest */,
+ 4089A0130FFACEFC000B29AE /* sample1_unittest-framework */,
+ 40C84997101A36A60083642A /* sample1_unittest-static */,
+ );
+ name = Products;
+ sourceTree = "<group>";
+ };
+ 0867D691FE84028FC02AAC07 /* gtest */ = {
+ isa = PBXGroup;
+ children = (
+ 40D4CDF00E30E07400294801 /* Config */,
+ 08FB77ACFE841707C02AAC07 /* Source */,
+ 40D4CF4E0E30F5E200294801 /* Resources */,
+ 403EE37B0E377822004BD1E2 /* Scripts */,
+ 034768DDFF38A45A11DB9C8B /* Products */,
+ );
+ name = gtest;
+ sourceTree = "<group>";
+ };
+ 08FB77ACFE841707C02AAC07 /* Source */ = {
+ isa = PBXGroup;
+ children = (
+ 404884A90E2F7CD900CF7658 /* CHANGES */,
+ 404884AA0E2F7CD900CF7658 /* CONTRIBUTORS */,
+ 404884AB0E2F7CD900CF7658 /* COPYING */,
+ 404883F60E2F799B00CF7658 /* README */,
+ 404883D90E2F799B00CF7658 /* include */,
+ 4089A02F0FFACF84000B29AE /* samples */,
+ 404884070E2F799B00CF7658 /* src */,
+ 3B238BF00E7FE13B00846E11 /* test */,
+ );
+ name = Source;
+ sourceTree = "<group>";
+ };
+ 3B238BF00E7FE13B00846E11 /* test */ = {
+ isa = PBXGroup;
+ children = (
+ 3B238C120E7FE13C00846E11 /* gtest_unittest.cc */,
+ );
+ name = test;
+ path = ../test;
+ sourceTree = SOURCE_ROOT;
+ };
+ 403EE37B0E377822004BD1E2 /* Scripts */ = {
+ isa = PBXGroup;
+ children = (
+ 403EE37C0E377822004BD1E2 /* versiongenerate.py */,
+ 3B87D2100E96B92E000D1852 /* runtests.sh */,
+ );
+ path = Scripts;
+ sourceTree = "<group>";
+ };
+ 404883D90E2F799B00CF7658 /* include */ = {
+ isa = PBXGroup;
+ children = (
+ 404883DA0E2F799B00CF7658 /* gtest */,
+ );
+ name = include;
+ path = ../include;
+ sourceTree = SOURCE_ROOT;
+ };
+ 404883DA0E2F799B00CF7658 /* gtest */ = {
+ isa = PBXGroup;
+ children = (
+ 404883E10E2F799B00CF7658 /* internal */,
+ 224A12A20E9EADCC00BD17FD /* gtest-test-part.h */,
+ 404883DB0E2F799B00CF7658 /* gtest-death-test.h */,
+ 404883DC0E2F799B00CF7658 /* gtest-message.h */,
+ 4539C9330EC280AE00A70F4C /* gtest-param-test.h */,
+ 4567C8171264FF71007740BE /* gtest-printers.h */,
+ 404883DD0E2F799B00CF7658 /* gtest-spi.h */,
+ 404883DE0E2F799B00CF7658 /* gtest.h */,
+ 404883DF0E2F799B00CF7658 /* gtest_pred_impl.h */,
+ 404883E00E2F799B00CF7658 /* gtest_prod.h */,
+ 3BF6F2A40E79B616000F2EEE /* gtest-typed-test.h */,
+ );
+ path = gtest;
+ sourceTree = "<group>";
+ };
+ 404883E10E2F799B00CF7658 /* internal */ = {
+ isa = PBXGroup;
+ children = (
+ 404883E20E2F799B00CF7658 /* gtest-death-test-internal.h */,
+ 404883E30E2F799B00CF7658 /* gtest-filepath.h */,
+ 404883E40E2F799B00CF7658 /* gtest-internal.h */,
+ 4539C9350EC280E200A70F4C /* gtest-linked_ptr.h */,
+ 4539C9360EC280E200A70F4C /* gtest-param-util-generated.h */,
+ 4539C9370EC280E200A70F4C /* gtest-param-util.h */,
+ 404883E50E2F799B00CF7658 /* gtest-port.h */,
+ 404883E60E2F799B00CF7658 /* gtest-string.h */,
+ 40899F4D0FFA7271000B29AE /* gtest-tuple.h */,
+ 3BF6F29F0E79B5AD000F2EEE /* gtest-type-util.h */,
+ );
+ path = internal;
+ sourceTree = "<group>";
+ };
+ 404884070E2F799B00CF7658 /* src */ = {
+ isa = PBXGroup;
+ children = (
+ 224A12A10E9EADA700BD17FD /* gtest-all.cc */,
+ 4048840D0E2F799B00CF7658 /* gtest_main.cc */,
+ );
+ name = src;
+ path = ../src;
+ sourceTree = SOURCE_ROOT;
+ };
+ 4089A02F0FFACF84000B29AE /* samples */ = {
+ isa = PBXGroup;
+ children = (
+ 4089A02C0FFACF7F000B29AE /* sample1.cc */,
+ 4089A02D0FFACF7F000B29AE /* sample1.h */,
+ 4089A02E0FFACF7F000B29AE /* sample1_unittest.cc */,
+ );
+ name = samples;
+ path = ../samples;
+ sourceTree = SOURCE_ROOT;
+ };
+ 40D4CDF00E30E07400294801 /* Config */ = {
+ isa = PBXGroup;
+ children = (
+ 40D4CDF10E30E07400294801 /* DebugProject.xcconfig */,
+ 40D4CDF20E30E07400294801 /* FrameworkTarget.xcconfig */,
+ 40D4CDF30E30E07400294801 /* General.xcconfig */,
+ 40D4CDF40E30E07400294801 /* ReleaseProject.xcconfig */,
+ 40899FB30FFA7567000B29AE /* StaticLibraryTarget.xcconfig */,
+ );
+ path = Config;
+ sourceTree = "<group>";
+ };
+ 40D4CF4E0E30F5E200294801 /* Resources */ = {
+ isa = PBXGroup;
+ children = (
+ 40D4CF510E30F5E200294801 /* Info.plist */,
+ );
+ path = Resources;
+ sourceTree = "<group>";
+ };
+/* End PBXGroup section */
+
+/* Begin PBXHeadersBuildPhase section */
+ 8D07F2BD0486CC7A007CD1D0 /* Headers */ = {
+ isa = PBXHeadersBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 404884380E2F799B00CF7658 /* gtest-death-test.h in Headers */,
+ 404884390E2F799B00CF7658 /* gtest-message.h in Headers */,
+ 4539C9340EC280AE00A70F4C /* gtest-param-test.h in Headers */,
+ 4567C8181264FF71007740BE /* gtest-printers.h in Headers */,
+ 3BF6F2A50E79B616000F2EEE /* gtest-typed-test.h in Headers */,
+ 4048843A0E2F799B00CF7658 /* gtest-spi.h in Headers */,
+ 4048843B0E2F799B00CF7658 /* gtest.h in Headers */,
+ 4048843C0E2F799B00CF7658 /* gtest_pred_impl.h in Headers */,
+ 4048843D0E2F799B00CF7658 /* gtest_prod.h in Headers */,
+ 224A12A30E9EADCC00BD17FD /* gtest-test-part.h in Headers */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXHeadersBuildPhase section */
+
+/* Begin PBXNativeTarget section */
+ 40899F420FFA7184000B29AE /* gtest_unittest-framework */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 40899F4A0FFA71BC000B29AE /* Build configuration list for PBXNativeTarget "gtest_unittest-framework" */;
+ buildPhases = (
+ 40899F400FFA7184000B29AE /* Sources */,
+ 40899F410FFA7184000B29AE /* Frameworks */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ 40C849A0101A36F10083642A /* PBXTargetDependency */,
+ );
+ name = "gtest_unittest-framework";
+ productName = gtest_unittest;
+ productReference = 40899F430FFA7184000B29AE /* gtest_unittest-framework */;
+ productType = "com.apple.product-type.tool";
+ };
+ 4089A0120FFACEFC000B29AE /* sample1_unittest-framework */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 4089A0240FFACF01000B29AE /* Build configuration list for PBXNativeTarget "sample1_unittest-framework" */;
+ buildPhases = (
+ 4089A0100FFACEFC000B29AE /* Sources */,
+ 4089A0110FFACEFC000B29AE /* Frameworks */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ 40C8499E101A36E50083642A /* PBXTargetDependency */,
+ );
+ name = "sample1_unittest-framework";
+ productName = sample1_unittest;
+ productReference = 4089A0130FFACEFC000B29AE /* sample1_unittest-framework */;
+ productType = "com.apple.product-type.tool";
+ };
+ 40C848F9101A209C0083642A /* gtest-static */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 40C84902101A212E0083642A /* Build configuration list for PBXNativeTarget "gtest-static" */;
+ buildPhases = (
+ 40C848F7101A209C0083642A /* Sources */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ );
+ name = "gtest-static";
+ productName = "gtest-static";
+ productReference = 40C848FA101A209C0083642A /* libgtest.a */;
+ productType = "com.apple.product-type.library.static";
+ };
+ 40C8490A101A217E0083642A /* gtest_main-static */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 40C84912101A21D20083642A /* Build configuration list for PBXNativeTarget "gtest_main-static" */;
+ buildPhases = (
+ 40C84908101A217E0083642A /* Sources */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ );
+ name = "gtest_main-static";
+ productName = "gtest_main-static";
+ productReference = 40C8490B101A217E0083642A /* libgtest_main.a */;
+ productType = "com.apple.product-type.library.static";
+ };
+ 40C8497A101A36850083642A /* gtest_unittest-static */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 40C84984101A36850083642A /* Build configuration list for PBXNativeTarget "gtest_unittest-static" */;
+ buildPhases = (
+ 40C8497F101A36850083642A /* Sources */,
+ 40C84981101A36850083642A /* Frameworks */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ 40C8497B101A36850083642A /* PBXTargetDependency */,
+ 40C8497D101A36850083642A /* PBXTargetDependency */,
+ );
+ name = "gtest_unittest-static";
+ productName = gtest_unittest;
+ productReference = 40C84987101A36850083642A /* gtest_unittest */;
+ productType = "com.apple.product-type.tool";
+ };
+ 40C84989101A36A60083642A /* sample1_unittest-static */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 40C84994101A36A60083642A /* Build configuration list for PBXNativeTarget "sample1_unittest-static" */;
+ buildPhases = (
+ 40C8498E101A36A60083642A /* Sources */,
+ 40C84991101A36A60083642A /* Frameworks */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ 40C8498A101A36A60083642A /* PBXTargetDependency */,
+ 40C8498C101A36A60083642A /* PBXTargetDependency */,
+ );
+ name = "sample1_unittest-static";
+ productName = sample1_unittest;
+ productReference = 40C84997101A36A60083642A /* sample1_unittest-static */;
+ productType = "com.apple.product-type.tool";
+ };
+ 8D07F2BC0486CC7A007CD1D0 /* gtest-framework */ = {
+ isa = PBXNativeTarget;
+ buildConfigurationList = 4FADC24208B4156D00ABE55E /* Build configuration list for PBXNativeTarget "gtest-framework" */;
+ buildPhases = (
+ 8D07F2C10486CC7A007CD1D0 /* Sources */,
+ 8D07F2BD0486CC7A007CD1D0 /* Headers */,
+ 404884A50E2F7C0400CF7658 /* Copy Headers Internal */,
+ 8D07F2BF0486CC7A007CD1D0 /* Resources */,
+ );
+ buildRules = (
+ );
+ dependencies = (
+ 40C44AE60E379922008FCC51 /* PBXTargetDependency */,
+ 408BEC101046CFE900DEF522 /* PBXTargetDependency */,
+ 40C8499C101A36DC0083642A /* PBXTargetDependency */,
+ );
+ name = "gtest-framework";
+ productInstallPath = "$(HOME)/Library/Frameworks";
+ productName = gtest;
+ productReference = 4539C8FF0EC27F6400A70F4C /* gtest.framework */;
+ productType = "com.apple.product-type.framework";
+ };
+/* End PBXNativeTarget section */
+
+/* Begin PBXProject section */
+ 0867D690FE84028FC02AAC07 /* Project object */ = {
+ isa = PBXProject;
+ buildConfigurationList = 4FADC24608B4156D00ABE55E /* Build configuration list for PBXProject "gtest" */;
+ compatibilityVersion = "Xcode 2.4";
+ hasScannedForEncodings = 1;
+ knownRegions = (
+ English,
+ Japanese,
+ French,
+ German,
+ en,
+ );
+ mainGroup = 0867D691FE84028FC02AAC07 /* gtest */;
+ productRefGroup = 034768DDFF38A45A11DB9C8B /* Products */;
+ projectDirPath = "";
+ projectRoot = "";
+ targets = (
+ 8D07F2BC0486CC7A007CD1D0 /* gtest-framework */,
+ 40C848F9101A209C0083642A /* gtest-static */,
+ 40C8490A101A217E0083642A /* gtest_main-static */,
+ 40899F420FFA7184000B29AE /* gtest_unittest-framework */,
+ 40C8497A101A36850083642A /* gtest_unittest-static */,
+ 4089A0120FFACEFC000B29AE /* sample1_unittest-framework */,
+ 40C84989101A36A60083642A /* sample1_unittest-static */,
+ 3B238F5F0E828B5400846E11 /* Check */,
+ 40C44ADC0E3798F4008FCC51 /* Version Info */,
+ );
+ };
+/* End PBXProject section */
+
+/* Begin PBXResourcesBuildPhase section */
+ 8D07F2BF0486CC7A007CD1D0 /* Resources */ = {
+ isa = PBXResourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 404884500E2F799B00CF7658 /* README in Resources */,
+ 404884AC0E2F7CD900CF7658 /* CHANGES in Resources */,
+ 404884AD0E2F7CD900CF7658 /* CONTRIBUTORS in Resources */,
+ 404884AE0E2F7CD900CF7658 /* COPYING in Resources */,
+ 40C84978101A36540083642A /* libgtest_main.a in Resources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXResourcesBuildPhase section */
+
+/* Begin PBXShellScriptBuildPhase section */
+ 3B238F5E0E828B5400846E11 /* ShellScript */ = {
+ isa = PBXShellScriptBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ );
+ inputPaths = (
+ );
+ outputPaths = (
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ shellPath = /bin/sh;
+ shellScript = "# Remember, this \"Run Script\" build phase will be executed from $SRCROOT\n/bin/bash Scripts/runtests.sh";
+ };
+ 40C44ADB0E3798F4008FCC51 /* Generate Version.h */ = {
+ isa = PBXShellScriptBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ );
+ inputPaths = (
+ "$(SRCROOT)/Scripts/versiongenerate.py",
+ "$(SRCROOT)/../configure.ac",
+ );
+ name = "Generate Version.h";
+ outputPaths = (
+ "$(PROJECT_TEMP_DIR)/Version.h",
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ shellPath = /bin/sh;
+ shellScript = "# Remember, this \"Run Script\" build phase will be executed from $SRCROOT\n/usr/bin/python Scripts/versiongenerate.py ../ $PROJECT_TEMP_DIR";
+ };
+/* End PBXShellScriptBuildPhase section */
+
+/* Begin PBXSourcesBuildPhase section */
+ 40899F400FFA7184000B29AE /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40899F530FFA72A0000B29AE /* gtest_unittest.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 4089A0100FFACEFC000B29AE /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 4089A0440FFAD1BE000B29AE /* sample1.cc in Sources */,
+ 4089A0460FFAD1BE000B29AE /* sample1_unittest.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 40C848F7101A209C0083642A /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40C848FF101A21150083642A /* gtest-all.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 40C84908101A217E0083642A /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40C84915101A21DF0083642A /* gtest_main.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 40C8497F101A36850083642A /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40C84980101A36850083642A /* gtest_unittest.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 40C8498E101A36A60083642A /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40C8498F101A36A60083642A /* sample1.cc in Sources */,
+ 40C84990101A36A60083642A /* sample1_unittest.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+ 8D07F2C10486CC7A007CD1D0 /* Sources */ = {
+ isa = PBXSourcesBuildPhase;
+ buildActionMask = 2147483647;
+ files = (
+ 40899F3A0FFA70D4000B29AE /* gtest-all.cc in Sources */,
+ );
+ runOnlyForDeploymentPostprocessing = 0;
+ };
+/* End PBXSourcesBuildPhase section */
+
+/* Begin PBXTargetDependency section */
+ 40899F9D0FFA740F000B29AE /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40899F420FFA7184000B29AE /* gtest_unittest-framework */;
+ targetProxy = 40899F9C0FFA740F000B29AE /* PBXContainerItemProxy */;
+ };
+ 4089A0980FFAD34A000B29AE /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 4089A0120FFACEFC000B29AE /* sample1_unittest-framework */;
+ targetProxy = 4089A0970FFAD34A000B29AE /* PBXContainerItemProxy */;
+ };
+ 408BEC101046CFE900DEF522 /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C848F9101A209C0083642A /* gtest-static */;
+ targetProxy = 408BEC0F1046CFE900DEF522 /* PBXContainerItemProxy */;
+ };
+ 40C44AE60E379922008FCC51 /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C44ADC0E3798F4008FCC51 /* Version Info */;
+ targetProxy = 40C44AE50E379922008FCC51 /* PBXContainerItemProxy */;
+ };
+ 40C8497B101A36850083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C848F9101A209C0083642A /* gtest-static */;
+ targetProxy = 40C8497C101A36850083642A /* PBXContainerItemProxy */;
+ };
+ 40C8497D101A36850083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C8490A101A217E0083642A /* gtest_main-static */;
+ targetProxy = 40C8497E101A36850083642A /* PBXContainerItemProxy */;
+ };
+ 40C8498A101A36A60083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C848F9101A209C0083642A /* gtest-static */;
+ targetProxy = 40C8498B101A36A60083642A /* PBXContainerItemProxy */;
+ };
+ 40C8498C101A36A60083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C8490A101A217E0083642A /* gtest_main-static */;
+ targetProxy = 40C8498D101A36A60083642A /* PBXContainerItemProxy */;
+ };
+ 40C8499C101A36DC0083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C8490A101A217E0083642A /* gtest_main-static */;
+ targetProxy = 40C8499B101A36DC0083642A /* PBXContainerItemProxy */;
+ };
+ 40C8499E101A36E50083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 8D07F2BC0486CC7A007CD1D0 /* gtest-framework */;
+ targetProxy = 40C8499D101A36E50083642A /* PBXContainerItemProxy */;
+ };
+ 40C849A0101A36F10083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 8D07F2BC0486CC7A007CD1D0 /* gtest-framework */;
+ targetProxy = 40C8499F101A36F10083642A /* PBXContainerItemProxy */;
+ };
+ 40C849F7101A43440083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C8497A101A36850083642A /* gtest_unittest-static */;
+ targetProxy = 40C849F6101A43440083642A /* PBXContainerItemProxy */;
+ };
+ 40C849F9101A43490083642A /* PBXTargetDependency */ = {
+ isa = PBXTargetDependency;
+ target = 40C84989101A36A60083642A /* sample1_unittest-static */;
+ targetProxy = 40C849F8101A43490083642A /* PBXContainerItemProxy */;
+ };
+/* End PBXTargetDependency section */
+
+/* Begin XCBuildConfiguration section */
+ 3B238F600E828B5400846E11 /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ COPY_PHASE_STRIP = NO;
+ GCC_DYNAMIC_NO_PIC = NO;
+ GCC_OPTIMIZATION_LEVEL = 0;
+ PRODUCT_NAME = Check;
+ };
+ name = Debug;
+ };
+ 3B238F610E828B5400846E11 /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ COPY_PHASE_STRIP = YES;
+ DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
+ GCC_ENABLE_FIX_AND_CONTINUE = NO;
+ PRODUCT_NAME = Check;
+ ZERO_LINK = NO;
+ };
+ name = Release;
+ };
+ 40899F450FFA7185000B29AE /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ HEADER_SEARCH_PATHS = ../;
+ PRODUCT_NAME = "gtest_unittest-framework";
+ };
+ name = Debug;
+ };
+ 40899F460FFA7185000B29AE /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ HEADER_SEARCH_PATHS = ../;
+ PRODUCT_NAME = "gtest_unittest-framework";
+ };
+ name = Release;
+ };
+ 4089A0150FFACEFD000B29AE /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = "sample1_unittest-framework";
+ };
+ name = Debug;
+ };
+ 4089A0160FFACEFD000B29AE /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = "sample1_unittest-framework";
+ };
+ name = Release;
+ };
+ 40C44ADF0E3798F4008FCC51 /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = gtest;
+ TARGET_NAME = gtest;
+ };
+ name = Debug;
+ };
+ 40C44AE00E3798F4008FCC51 /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = gtest;
+ TARGET_NAME = gtest;
+ };
+ name = Release;
+ };
+ 40C848FB101A209D0083642A /* Debug */ = {
+ isa = XCBuildConfiguration;
+ baseConfigurationReference = 40899FB30FFA7567000B29AE /* StaticLibraryTarget.xcconfig */;
+ buildSettings = {
+ GCC_INLINES_ARE_PRIVATE_EXTERN = YES;
+ GCC_SYMBOLS_PRIVATE_EXTERN = YES;
+ HEADER_SEARCH_PATHS = (
+ ../,
+ ../include/,
+ );
+ PRODUCT_NAME = gtest;
+ };
+ name = Debug;
+ };
+ 40C848FC101A209D0083642A /* Release */ = {
+ isa = XCBuildConfiguration;
+ baseConfigurationReference = 40899FB30FFA7567000B29AE /* StaticLibraryTarget.xcconfig */;
+ buildSettings = {
+ GCC_INLINES_ARE_PRIVATE_EXTERN = YES;
+ GCC_SYMBOLS_PRIVATE_EXTERN = YES;
+ HEADER_SEARCH_PATHS = (
+ ../,
+ ../include/,
+ );
+ PRODUCT_NAME = gtest;
+ };
+ name = Release;
+ };
+ 40C8490E101A217F0083642A /* Debug */ = {
+ isa = XCBuildConfiguration;
+ baseConfigurationReference = 40899FB30FFA7567000B29AE /* StaticLibraryTarget.xcconfig */;
+ buildSettings = {
+ HEADER_SEARCH_PATHS = (
+ ../,
+ ../include/,
+ );
+ PRODUCT_NAME = gtest_main;
+ };
+ name = Debug;
+ };
+ 40C8490F101A217F0083642A /* Release */ = {
+ isa = XCBuildConfiguration;
+ baseConfigurationReference = 40899FB30FFA7567000B29AE /* StaticLibraryTarget.xcconfig */;
+ buildSettings = {
+ HEADER_SEARCH_PATHS = (
+ ../,
+ ../include/,
+ );
+ PRODUCT_NAME = gtest_main;
+ };
+ name = Release;
+ };
+ 40C84985101A36850083642A /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ HEADER_SEARCH_PATHS = ../;
+ PRODUCT_NAME = gtest_unittest;
+ };
+ name = Debug;
+ };
+ 40C84986101A36850083642A /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ HEADER_SEARCH_PATHS = ../;
+ PRODUCT_NAME = gtest_unittest;
+ };
+ name = Release;
+ };
+ 40C84995101A36A60083642A /* Debug */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = "sample1_unittest-static";
+ };
+ name = Debug;
+ };
+ 40C84996101A36A60083642A /* Release */ = {
+ isa = XCBuildConfiguration;
+ buildSettings = {
+ PRODUCT_NAME = "sample1_unittest-static";
+ };
+ name = Release;
+ };
+ 4FADC24308B4156D00ABE55E /* Debug */ = {
+ isa = XCBuildConfiguration;
+ baseConfigurationReference = 40D4CDF20E30E07400294801 /* FrameworkTarget.xcconfig */;
+ buildSettings = {
+ DYLIB_COMPATIBILITY_VERSION = 1;
+ DYLIB_CURRENT_VERSION = 1;
+ HEADER_SEARCH_PATHS = (
+ ../,
+ ../include/,
+ );
+ INFOPLIST_FILE = Resources/Info.plist;
+ INFOPLIST_PREFIX_HEADER = "$(PROJECT_TEMP_DIR)/Version.h";
+ INFOPLIST_PREPROCESS = YES;
+ PRODUCT_NAME = gtest;
+ VERSIONING_SYSTEM = "apple-generic";
+ };
+ name = Debug;
+ };
+ 4FADC24408B4156D00ABE55E /* Release */ = {
+ isa = XCBuildConfiguration;
+ baseConfigurationReference = 40D4CDF20E30E07400294801 /* FrameworkTarget.xcconfig */;
+ buildSettings = {
+ DYLIB_COMPATIBILITY_VERSION = 1;
+ DYLIB_CURRENT_VERSION = 1;
+ HEADER_SEARCH_PATHS = (
+ ../,
+ ../include/,
+ );
+ INFOPLIST_FILE = Resources/Info.plist;
+ INFOPLIST_PREFIX_HEADER = "$(PROJECT_TEMP_DIR)/Version.h";
+ INFOPLIST_PREPROCESS = YES;
+ PRODUCT_NAME = gtest;
+ VERSIONING_SYSTEM = "apple-generic";
+ };
+ name = Release;
+ };
+ 4FADC24708B4156D00ABE55E /* Debug */ = {
+ isa = XCBuildConfiguration;
+ baseConfigurationReference = 40D4CDF10E30E07400294801 /* DebugProject.xcconfig */;
+ buildSettings = {
+ };
+ name = Debug;
+ };
+ 4FADC24808B4156D00ABE55E /* Release */ = {
+ isa = XCBuildConfiguration;
+ baseConfigurationReference = 40D4CDF40E30E07400294801 /* ReleaseProject.xcconfig */;
+ buildSettings = {
+ };
+ name = Release;
+ };
+/* End XCBuildConfiguration section */
+
+/* Begin XCConfigurationList section */
+ 3B238FA30E828BB600846E11 /* Build configuration list for PBXAggregateTarget "Check" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 3B238F600E828B5400846E11 /* Debug */,
+ 3B238F610E828B5400846E11 /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 40899F4A0FFA71BC000B29AE /* Build configuration list for PBXNativeTarget "gtest_unittest-framework" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 40899F450FFA7185000B29AE /* Debug */,
+ 40899F460FFA7185000B29AE /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 4089A0240FFACF01000B29AE /* Build configuration list for PBXNativeTarget "sample1_unittest-framework" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 4089A0150FFACEFD000B29AE /* Debug */,
+ 4089A0160FFACEFD000B29AE /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 40C44AE40E379905008FCC51 /* Build configuration list for PBXAggregateTarget "Version Info" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 40C44ADF0E3798F4008FCC51 /* Debug */,
+ 40C44AE00E3798F4008FCC51 /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 40C84902101A212E0083642A /* Build configuration list for PBXNativeTarget "gtest-static" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 40C848FB101A209D0083642A /* Debug */,
+ 40C848FC101A209D0083642A /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 40C84912101A21D20083642A /* Build configuration list for PBXNativeTarget "gtest_main-static" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 40C8490E101A217F0083642A /* Debug */,
+ 40C8490F101A217F0083642A /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 40C84984101A36850083642A /* Build configuration list for PBXNativeTarget "gtest_unittest-static" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 40C84985101A36850083642A /* Debug */,
+ 40C84986101A36850083642A /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 40C84994101A36A60083642A /* Build configuration list for PBXNativeTarget "sample1_unittest-static" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 40C84995101A36A60083642A /* Debug */,
+ 40C84996101A36A60083642A /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 4FADC24208B4156D00ABE55E /* Build configuration list for PBXNativeTarget "gtest-framework" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 4FADC24308B4156D00ABE55E /* Debug */,
+ 4FADC24408B4156D00ABE55E /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+ 4FADC24608B4156D00ABE55E /* Build configuration list for PBXProject "gtest" */ = {
+ isa = XCConfigurationList;
+ buildConfigurations = (
+ 4FADC24708B4156D00ABE55E /* Debug */,
+ 4FADC24808B4156D00ABE55E /* Release */,
+ );
+ defaultConfigurationIsVisible = 0;
+ defaultConfigurationName = Release;
+ };
+/* End XCConfigurationList section */
+ };
+ rootObject = 0867D690FE84028FC02AAC07 /* Project object */;
+}
diff --git a/libvpx/third_party/libyuv/README.webm b/libvpx/third_party/libyuv/README.webm
new file mode 100644
index 0000000..d3495ca
--- /dev/null
+++ b/libvpx/third_party/libyuv/README.webm
@@ -0,0 +1,17 @@
+Name: libyuv
+URL: http://code.google.com/p/libyuv/
+Version: 102
+License: BSD
+License File: LICENSE
+
+Description:
+libyuv is an open source project that includes YUV conversion and scaling
+functionality.
+
+The optimized scaler in libyuv is used in multiple resolution encoder example,
+which down-samples the original input video (f.g. 1280x720) a number of times
+in order to encode multiple resolution bit streams.
+
+Local Modifications:
+Modified the original scaler code from C++ to C to fit in our current build
+system. This is a temporal solution, and will be improved later.
\ No newline at end of file
diff --git a/libvpx/third_party/libyuv/include/libyuv/basic_types.h b/libvpx/third_party/libyuv/include/libyuv/basic_types.h
new file mode 100644
index 0000000..30504ce
--- /dev/null
+++ b/libvpx/third_party/libyuv/include/libyuv/basic_types.h
@@ -0,0 +1,73 @@
+/*
+ * Copyright (c) 2011 The LibYuv project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef INCLUDE_LIBYUV_BASIC_TYPES_H_
+#define INCLUDE_LIBYUV_BASIC_TYPES_H_
+
+#include <stddef.h> // for NULL, size_t
+
+#if !(defined(_MSC_VER) && (_MSC_VER < 1600))
+#include <stdint.h> // for uintptr_t
+#endif
+
+#ifndef INT_TYPES_DEFINED
+#define INT_TYPES_DEFINED
+#ifdef COMPILER_MSVC
+typedef unsigned __int64 uint64;
+typedef __int64 int64;
+#ifndef INT64_C
+#define INT64_C(x) x ## I64
+#endif
+#ifndef UINT64_C
+#define UINT64_C(x) x ## UI64
+#endif
+#define INT64_F "I64"
+#else // COMPILER_MSVC
+#ifdef __LP64__
+typedef unsigned long uint64;
+typedef long int64;
+#ifndef INT64_C
+#define INT64_C(x) x ## L
+#endif
+#ifndef UINT64_C
+#define UINT64_C(x) x ## UL
+#endif
+#define INT64_F "l"
+#else // __LP64__
+typedef unsigned long long uint64;
+typedef long long int64;
+#ifndef INT64_C
+#define INT64_C(x) x ## LL
+#endif
+#ifndef UINT64_C
+#define UINT64_C(x) x ## ULL
+#endif
+#define INT64_F "ll"
+#endif // __LP64__
+#endif // COMPILER_MSVC
+typedef unsigned int uint32;
+typedef int int32;
+typedef unsigned short uint16;
+typedef short int16;
+typedef unsigned char uint8;
+typedef char int8;
+#endif // INT_TYPES_DEFINED
+
+// Detect compiler is for x86 or x64.
+#if defined(__x86_64__) || defined(_M_X64) || \
+ defined(__i386__) || defined(_M_IX86)
+#define CPU_X86 1
+#endif
+
+#define ALIGNP(p, t) \
+ ((uint8*)((((uintptr_t)(p) + \
+ ((t)-1)) & ~((t)-1))))
+
+#endif // INCLUDE_LIBYUV_BASIC_TYPES_H_
diff --git a/libvpx/third_party/libyuv/include/libyuv/cpu_id.h b/libvpx/third_party/libyuv/include/libyuv/cpu_id.h
new file mode 100644
index 0000000..4a53b5b
--- /dev/null
+++ b/libvpx/third_party/libyuv/include/libyuv/cpu_id.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright (c) 2011 The LibYuv project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef INCLUDE_LIBYUV_CPU_ID_H_
+#define INCLUDE_LIBYUV_CPU_ID_H_
+
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
+
+// These flags are only valid on x86 processors
+static const int kCpuHasSSE2 = 1;
+static const int kCpuHasSSSE3 = 2;
+
+// These flags are only valid on ARM processors
+static const int kCpuHasNEON = 4;
+
+// Internal flag to indicate cpuid is initialized.
+static const int kCpuInitialized = 8;
+
+// Detect CPU has SSE2 etc.
+// test_flag parameter should be one of kCpuHas constants above
+// returns non-zero if instruction set is detected
+static __inline int TestCpuFlag(int test_flag) {
+ extern int cpu_info_;
+ extern int InitCpuFlags();
+ return (cpu_info_ ? cpu_info_ : InitCpuFlags()) & test_flag;
+}
+
+// For testing, allow CPU flags to be disabled.
+// ie MaskCpuFlags(~kCpuHasSSSE3) to disable SSSE3.
+// -1 to enable all cpu specific optimizations.
+// 0 to disable all cpu specific optimizations.
+void MaskCpuFlags(int enable_flags);
+
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
+
+#endif // INCLUDE_LIBYUV_CPU_ID_H_
diff --git a/libvpx/third_party/libyuv/include/libyuv/scale.h b/libvpx/third_party/libyuv/include/libyuv/scale.h
new file mode 100644
index 0000000..21fe360
--- /dev/null
+++ b/libvpx/third_party/libyuv/include/libyuv/scale.h
@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2011 The LibYuv project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef INCLUDE_LIBYUV_SCALE_H_
+#define INCLUDE_LIBYUV_SCALE_H_
+
+#include "third_party/libyuv/include/libyuv/basic_types.h"
+
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
+
+// Supported filtering
+typedef enum {
+ kFilterNone = 0, // Point sample; Fastest
+ kFilterBilinear = 1, // Faster than box, but lower quality scaling down.
+ kFilterBox = 2 // Highest quality
+}FilterMode;
+
+// Scales a YUV 4:2:0 image from the src width and height to the
+// dst width and height.
+// If filtering is kFilterNone, a simple nearest-neighbor algorithm is
+// used. This produces basic (blocky) quality at the fastest speed.
+// If filtering is kFilterBilinear, interpolation is used to produce a better
+// quality image, at the expense of speed.
+// If filtering is kFilterBox, averaging is used to produce ever better
+// quality image, at further expense of speed.
+// Returns 0 if successful.
+
+int I420Scale(const uint8* src_y, int src_stride_y,
+ const uint8* src_u, int src_stride_u,
+ const uint8* src_v, int src_stride_v,
+ int src_width, int src_height,
+ uint8* dst_y, int dst_stride_y,
+ uint8* dst_u, int dst_stride_u,
+ uint8* dst_v, int dst_stride_v,
+ int dst_width, int dst_height,
+ FilterMode filtering);
+
+// Legacy API. Deprecated
+int Scale(const uint8* src_y, const uint8* src_u, const uint8* src_v,
+ int src_stride_y, int src_stride_u, int src_stride_v,
+ int src_width, int src_height,
+ uint8* dst_y, uint8* dst_u, uint8* dst_v,
+ int dst_stride_y, int dst_stride_u, int dst_stride_v,
+ int dst_width, int dst_height,
+ int interpolate);
+
+// Legacy API. Deprecated
+int ScaleOffset(const uint8* src, int src_width, int src_height,
+ uint8* dst, int dst_width, int dst_height, int dst_yoffset,
+ int interpolate);
+
+// For testing, allow disabling of optimizations.
+void SetUseReferenceImpl(int use);
+
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
+
+#endif // INCLUDE_LIBYUV_SCALE_H_
diff --git a/libvpx/third_party/libyuv/source/cpu_id.c b/libvpx/third_party/libyuv/source/cpu_id.c
new file mode 100644
index 0000000..fccf3dd
--- /dev/null
+++ b/libvpx/third_party/libyuv/source/cpu_id.c
@@ -0,0 +1,81 @@
+/*
+ * Copyright (c) 2011 The LibYuv project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "third_party/libyuv/include/libyuv/cpu_id.h"
+
+#ifdef _MSC_VER
+#include <intrin.h>
+#endif
+#ifdef __ANDROID__
+#include <cpu-features.h>
+#endif
+
+#include "third_party/libyuv/include/libyuv/basic_types.h" // for CPU_X86
+
+// TODO(fbarchard): Use cpuid.h when gcc 4.4 is used on OSX and Linux.
+#if (defined(__pic__) || defined(__APPLE__)) && defined(__i386__)
+static inline void __cpuid(int cpu_info[4], int info_type) {
+ asm volatile (
+ "mov %%ebx, %%edi \n"
+ "cpuid \n"
+ "xchg %%edi, %%ebx \n"
+ : "=a"(cpu_info[0]), "=D"(cpu_info[1]), "=c"(cpu_info[2]), "=d"(cpu_info[3])
+ : "a"(info_type)
+ );
+}
+#elif defined(__i386__) || defined(__x86_64__)
+static inline void __cpuid(int cpu_info[4], int info_type) {
+ asm volatile (
+ "cpuid \n"
+ : "=a"(cpu_info[0]), "=b"(cpu_info[1]), "=c"(cpu_info[2]), "=d"(cpu_info[3])
+ : "a"(info_type)
+ );
+}
+#endif
+
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
+
+// CPU detect function for SIMD instruction sets.
+int cpu_info_ = 0;
+
+int InitCpuFlags() {
+#ifdef CPU_X86
+ int cpu_info[4];
+ __cpuid(cpu_info, 1);
+ cpu_info_ = (cpu_info[3] & 0x04000000 ? kCpuHasSSE2 : 0) |
+ (cpu_info[2] & 0x00000200 ? kCpuHasSSSE3 : 0) |
+ kCpuInitialized;
+#elif defined(__ANDROID__) && defined(__ARM_NEON__)
+ uint64_t features = android_getCpuFeatures();
+ cpu_info_ = ((features & ANDROID_CPU_ARM_FEATURE_NEON) ? kCpuHasNEON : 0) |
+ kCpuInitialized;
+#elif defined(__ARM_NEON__)
+ // gcc -mfpu=neon defines __ARM_NEON__
+ // Enable Neon if you want support for Neon and Arm, and use MaskCpuFlags
+ // to disable Neon on devices that do not have it.
+ cpu_info_ = kCpuHasNEON | kCpuInitialized;
+#else
+ cpu_info_ = kCpuInitialized;
+#endif
+ return cpu_info_;
+}
+
+void MaskCpuFlags(int enable_flags) {
+ InitCpuFlags();
+ cpu_info_ = (cpu_info_ & enable_flags) | kCpuInitialized;
+}
+
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
diff --git a/libvpx/third_party/libyuv/source/row.h b/libvpx/third_party/libyuv/source/row.h
new file mode 100644
index 0000000..eabe180
--- /dev/null
+++ b/libvpx/third_party/libyuv/source/row.h
@@ -0,0 +1,264 @@
+/*
+ * Copyright (c) 2011 The LibYuv project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef LIBYUV_SOURCE_ROW_H_
+#define LIBYUV_SOURCE_ROW_H_
+
+#include "third_party/libyuv/include/libyuv/basic_types.h"
+
+#define kMaxStride (2048 * 4)
+#define IS_ALIGNED(p, a) (!((uintptr_t)(p) & ((a) - 1)))
+
+#if defined(COVERAGE_ENABLED) || defined(TARGET_IPHONE_SIMULATOR)
+#define YUV_DISABLE_ASM
+#endif
+
+#if defined(__ARM_NEON__) && !defined(YUV_DISABLE_ASM)
+#define HAS_FASTCONVERTYUVTOARGBROW_NEON
+void FastConvertYUVToARGBRow_NEON(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+#define HAS_FASTCONVERTYUVTOBGRAROW_NEON
+void FastConvertYUVToBGRARow_NEON(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+#define HAS_FASTCONVERTYUVTOABGRROW_NEON
+void FastConvertYUVToABGRRow_NEON(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+#endif
+
+// The following are available on all x86 platforms
+#if (defined(_M_IX86) || defined(__x86_64__) || defined(__i386__)) && \
+ !defined(YUV_DISABLE_ASM)
+#define HAS_ABGRTOARGBROW_SSSE3
+#define HAS_BGRATOARGBROW_SSSE3
+#define HAS_BG24TOARGBROW_SSSE3
+#define HAS_RAWTOARGBROW_SSSE3
+#define HAS_RGB24TOYROW_SSSE3
+#define HAS_RAWTOYROW_SSSE3
+#define HAS_RGB24TOUVROW_SSSE3
+#define HAS_RAWTOUVROW_SSSE3
+#define HAS_ARGBTOYROW_SSSE3
+#define HAS_BGRATOYROW_SSSE3
+#define HAS_ABGRTOYROW_SSSE3
+#define HAS_ARGBTOUVROW_SSSE3
+#define HAS_BGRATOUVROW_SSSE3
+#define HAS_ABGRTOUVROW_SSSE3
+#define HAS_I400TOARGBROW_SSE2
+#define HAS_FASTCONVERTYTOARGBROW_SSE2
+#define HAS_FASTCONVERTYUVTOARGBROW_SSSE3
+#define HAS_FASTCONVERTYUVTOBGRAROW_SSSE3
+#define HAS_FASTCONVERTYUVTOABGRROW_SSSE3
+#define HAS_FASTCONVERTYUV444TOARGBROW_SSSE3
+#define HAS_REVERSE_ROW_SSSE3
+#endif
+
+// The following are available on Neon platforms
+#if defined(__ARM_NEON__) && !defined(YUV_DISABLE_ASM)
+#define HAS_REVERSE_ROW_NEON
+#endif
+
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
+
+#ifdef HAS_ARGBTOYROW_SSSE3
+void ARGBToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
+void BGRAToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
+void ABGRToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
+void ARGBToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+void BGRAToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+void ABGRToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+#endif
+#if defined(HAS_BG24TOARGBROW_SSSE3) && defined(HAS_ARGBTOYROW_SSSE3)
+#define HASRGB24TOYROW_SSSE3
+#endif
+#ifdef HASRGB24TOYROW_SSSE3
+void RGB24ToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
+void RAWToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
+void RGB24ToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+void RAWToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+#endif
+#ifdef HAS_REVERSE_ROW_SSSE3
+void ReverseRow_SSSE3(const uint8* src, uint8* dst, int width);
+#endif
+#ifdef HAS_REVERSE_ROW_NEON
+void ReverseRow_NEON(const uint8* src, uint8* dst, int width);
+#endif
+void ReverseRow_C(const uint8* src, uint8* dst, int width);
+
+void ARGBToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
+void BGRAToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
+void ABGRToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
+void RGB24ToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
+void RAWToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
+void ARGBToUVRow_C(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+void BGRAToUVRow_C(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+void ABGRToUVRow_C(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+void RGB24ToUVRow_C(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+void RAWToUVRow_C(const uint8* src_argb0, int src_stride_argb,
+ uint8* dst_u, uint8* dst_v, int width);
+
+#ifdef HAS_BG24TOARGBROW_SSSE3
+void ABGRToARGBRow_SSSE3(const uint8* src_abgr, uint8* dst_argb, int pix);
+void BGRAToARGBRow_SSSE3(const uint8* src_bgra, uint8* dst_argb, int pix);
+void BG24ToARGBRow_SSSE3(const uint8* src_bg24, uint8* dst_argb, int pix);
+void RAWToARGBRow_SSSE3(const uint8* src_bg24, uint8* dst_argb, int pix);
+#endif
+void ABGRToARGBRow_C(const uint8* src_abgr, uint8* dst_argb, int pix);
+void BGRAToARGBRow_C(const uint8* src_bgra, uint8* dst_argb, int pix);
+void BG24ToARGBRow_C(const uint8* src_bg24, uint8* dst_argb, int pix);
+void RAWToARGBRow_C(const uint8* src_bg24, uint8* dst_argb, int pix);
+
+#ifdef HAS_I400TOARGBROW_SSE2
+void I400ToARGBRow_SSE2(const uint8* src_y, uint8* dst_argb, int pix);
+#endif
+void I400ToARGBRow_C(const uint8* src_y, uint8* dst_argb, int pix);
+
+#if defined(_MSC_VER)
+#define SIMD_ALIGNED(var) __declspec(align(16)) var
+typedef __declspec(align(16)) signed char vec8[16];
+typedef __declspec(align(16)) unsigned char uvec8[16];
+typedef __declspec(align(16)) signed short vec16[8];
+#else // __GNUC__
+#define SIMD_ALIGNED(var) var __attribute__((aligned(16)))
+typedef signed char __attribute__((vector_size(16))) vec8;
+typedef unsigned char __attribute__((vector_size(16))) uvec8;
+typedef signed short __attribute__((vector_size(16))) vec16;
+#endif
+
+//extern "C"
+SIMD_ALIGNED(const int16 kCoefficientsRgbY[768][4]);
+//extern "C"
+SIMD_ALIGNED(const int16 kCoefficientsBgraY[768][4]);
+//extern "C"
+SIMD_ALIGNED(const int16 kCoefficientsAbgrY[768][4]);
+
+void FastConvertYUVToARGBRow_C(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUVToBGRARow_C(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUVToABGRRow_C(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUV444ToARGBRow_C(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYToARGBRow_C(const uint8* y_buf,
+ uint8* rgb_buf,
+ int width);
+
+#ifdef HAS_FASTCONVERTYUVTOARGBROW_SSE2
+void FastConvertYUVToARGBRow_SSE2(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUVToARGBRow4_SSE2(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUVToBGRARow_SSE2(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUVToABGRRow_SSE2(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUV444ToARGBRow_SSE2(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYToARGBRow_SSE2(const uint8* y_buf,
+ uint8* rgb_buf,
+ int width);
+#endif
+
+#ifdef HAS_FASTCONVERTYUVTOARGBROW_SSSE3
+void FastConvertYUVToARGBRow_SSSE3(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUVToBGRARow_SSSE3(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUVToABGRRow_SSSE3(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+void FastConvertYUV444ToARGBRow_SSSE3(const uint8* y_buf,
+ const uint8* u_buf,
+ const uint8* v_buf,
+ uint8* rgb_buf,
+ int width);
+
+#endif
+
+#ifdef HAS_FASTCONVERTYTOARGBROW_SSE2
+void FastConvertYToARGBRow_SSE2(const uint8* y_buf,
+ uint8* rgb_buf,
+ int width);
+
+#endif
+
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
+
+#endif // LIBYUV_SOURCE_ROW_H_
diff --git a/libvpx/third_party/libyuv/source/scale.c b/libvpx/third_party/libyuv/source/scale.c
new file mode 100644
index 0000000..c142a17
--- /dev/null
+++ b/libvpx/third_party/libyuv/source/scale.c
@@ -0,0 +1,3884 @@
+/*
+ * Copyright (c) 2011 The LibYuv project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "third_party/libyuv/include/libyuv/scale.h"
+
+#include <assert.h>
+#include <string.h>
+
+#include "third_party/libyuv/include/libyuv/cpu_id.h"
+#include "third_party/libyuv/source/row.h"
+
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
+
+/*
+ * Note: Defining YUV_DISABLE_ASM allows to use c version.
+ */
+//#define YUV_DISABLE_ASM
+
+#if defined(_MSC_VER)
+#define ALIGN16(var) __declspec(align(16)) var
+#else
+#define ALIGN16(var) var __attribute__((aligned(16)))
+#endif
+
+// Note: A Neon reference manual
+// http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0204j/CJAJIIGG.html
+// Note: Some SSE2 reference manuals
+// cpuvol1.pdf agner_instruction_tables.pdf 253666.pdf 253667.pdf
+
+// Set the following flag to true to revert to only
+// using the reference implementation ScalePlaneBox(), and
+// NOT the optimized versions. Useful for debugging and
+// when comparing the quality of the resulting YUV planes
+// as produced by the optimized and non-optimized versions.
+
+static int use_reference_impl_ = 0;
+
+void SetUseReferenceImpl(int use) {
+ use_reference_impl_ = use;
+}
+
+// ScaleRowDown2Int also used by planar functions
+
+/**
+ * NEON downscalers with interpolation.
+ *
+ * Provided by Fritz Koenig
+ *
+ */
+
+#if defined(__ARM_NEON__) && !defined(YUV_DISABLE_ASM)
+#define HAS_SCALEROWDOWN2_NEON
+void ScaleRowDown2_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ asm volatile (
+ "1: \n"
+ "vld2.u8 {q0,q1}, [%0]! \n" // load even pixels into q0, odd into q1
+ "vst1.u8 {q0}, [%1]! \n" // store even pixels
+ "subs %2, %2, #16 \n" // 16 processed per loop
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst), // %1
+ "+r"(dst_width) // %2
+ :
+ : "q0", "q1" // Clobber List
+ );
+}
+
+void ScaleRowDown2Int_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ asm volatile (
+ "add %1, %0 \n" // change the stride to row 2 pointer
+ "1: \n"
+ "vld1.u8 {q0,q1}, [%0]! \n" // load row 1 and post increment
+ "vld1.u8 {q2,q3}, [%1]! \n" // load row 2 and post increment
+ "vpaddl.u8 q0, q0 \n" // row 1 add adjacent
+ "vpaddl.u8 q1, q1 \n"
+ "vpadal.u8 q0, q2 \n" // row 2 add adjacent, add row 1 to row 2
+ "vpadal.u8 q1, q3 \n"
+ "vrshrn.u16 d0, q0, #2 \n" // downshift, round and pack
+ "vrshrn.u16 d1, q1, #2 \n"
+ "vst1.u8 {q0}, [%2]! \n"
+ "subs %3, %3, #16 \n" // 16 processed per loop
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(src_stride), // %1
+ "+r"(dst), // %2
+ "+r"(dst_width) // %3
+ :
+ : "q0", "q1", "q2", "q3" // Clobber List
+ );
+}
+
+#define HAS_SCALEROWDOWN4_NEON
+static void ScaleRowDown4_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "1: \n"
+ "vld2.u8 {d0, d1}, [%0]! \n"
+ "vtrn.u8 d1, d0 \n"
+ "vshrn.u16 d0, q0, #8 \n"
+ "vst1.u32 {d0[1]}, [%1]! \n"
+
+ "subs %2, #4 \n"
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ :
+ : "q0", "q1", "memory", "cc"
+ );
+}
+
+static void ScaleRowDown4Int_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "add r4, %0, %3 \n"
+ "add r5, r4, %3 \n"
+ "add %3, r5, %3 \n"
+ "1: \n"
+ "vld1.u8 {q0}, [%0]! \n" // load up 16x4 block of input data
+ "vld1.u8 {q1}, [r4]! \n"
+ "vld1.u8 {q2}, [r5]! \n"
+ "vld1.u8 {q3}, [%3]! \n"
+
+ "vpaddl.u8 q0, q0 \n"
+ "vpadal.u8 q0, q1 \n"
+ "vpadal.u8 q0, q2 \n"
+ "vpadal.u8 q0, q3 \n"
+
+ "vpaddl.u16 q0, q0 \n"
+
+ "vrshrn.u32 d0, q0, #4 \n" // divide by 16 w/rounding
+
+ "vmovn.u16 d0, q0 \n"
+ "vst1.u32 {d0[0]}, [%1]! \n"
+
+ "subs %2, #4 \n"
+ "bhi 1b \n"
+
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"(src_stride) // %3
+ : "r4", "r5", "q0", "q1", "q2", "q3", "memory", "cc"
+ );
+}
+
+#define HAS_SCALEROWDOWN34_NEON
+// Down scale from 4 to 3 pixels. Use the neon multilane read/write
+// to load up the every 4th pixel into a 4 different registers.
+// Point samples 32 pixels to 24 pixels.
+static void ScaleRowDown34_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "1: \n"
+ "vld4.u8 {d0, d1, d2, d3}, [%0]! \n" // src line 0
+ "vmov d2, d3 \n" // order needs to be d0, d1, d2
+ "vst3.u8 {d0, d1, d2}, [%1]! \n"
+ "subs %2, #24 \n"
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ :
+ : "d0", "d1", "d2", "d3", "memory", "cc"
+ );
+}
+
+static void ScaleRowDown34_0_Int_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "vmov.u8 d24, #3 \n"
+ "add %3, %0 \n"
+ "1: \n"
+ "vld4.u8 {d0, d1, d2, d3}, [%0]! \n" // src line 0
+ "vld4.u8 {d4, d5, d6, d7}, [%3]! \n" // src line 1
+
+ // filter src line 0 with src line 1
+ // expand chars to shorts to allow for room
+ // when adding lines together
+ "vmovl.u8 q8, d4 \n"
+ "vmovl.u8 q9, d5 \n"
+ "vmovl.u8 q10, d6 \n"
+ "vmovl.u8 q11, d7 \n"
+
+ // 3 * line_0 + line_1
+ "vmlal.u8 q8, d0, d24 \n"
+ "vmlal.u8 q9, d1, d24 \n"
+ "vmlal.u8 q10, d2, d24 \n"
+ "vmlal.u8 q11, d3, d24 \n"
+
+ // (3 * line_0 + line_1) >> 2
+ "vqrshrn.u16 d0, q8, #2 \n"
+ "vqrshrn.u16 d1, q9, #2 \n"
+ "vqrshrn.u16 d2, q10, #2 \n"
+ "vqrshrn.u16 d3, q11, #2 \n"
+
+ // a0 = (src[0] * 3 + s[1] * 1) >> 2
+ "vmovl.u8 q8, d1 \n"
+ "vmlal.u8 q8, d0, d24 \n"
+ "vqrshrn.u16 d0, q8, #2 \n"
+
+ // a1 = (src[1] * 1 + s[2] * 1) >> 1
+ "vrhadd.u8 d1, d1, d2 \n"
+
+ // a2 = (src[2] * 1 + s[3] * 3) >> 2
+ "vmovl.u8 q8, d2 \n"
+ "vmlal.u8 q8, d3, d24 \n"
+ "vqrshrn.u16 d2, q8, #2 \n"
+
+ "vst3.u8 {d0, d1, d2}, [%1]! \n"
+
+ "subs %2, #24 \n"
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width), // %2
+ "+r"(src_stride) // %3
+ :
+ : "q0", "q1", "q2", "q3", "q8", "q9", "q10", "q11", "d24", "memory", "cc"
+ );
+}
+
+static void ScaleRowDown34_1_Int_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "vmov.u8 d24, #3 \n"
+ "add %3, %0 \n"
+ "1: \n"
+ "vld4.u8 {d0, d1, d2, d3}, [%0]! \n" // src line 0
+ "vld4.u8 {d4, d5, d6, d7}, [%3]! \n" // src line 1
+
+ // average src line 0 with src line 1
+ "vrhadd.u8 q0, q0, q2 \n"
+ "vrhadd.u8 q1, q1, q3 \n"
+
+ // a0 = (src[0] * 3 + s[1] * 1) >> 2
+ "vmovl.u8 q3, d1 \n"
+ "vmlal.u8 q3, d0, d24 \n"
+ "vqrshrn.u16 d0, q3, #2 \n"
+
+ // a1 = (src[1] * 1 + s[2] * 1) >> 1
+ "vrhadd.u8 d1, d1, d2 \n"
+
+ // a2 = (src[2] * 1 + s[3] * 3) >> 2
+ "vmovl.u8 q3, d2 \n"
+ "vmlal.u8 q3, d3, d24 \n"
+ "vqrshrn.u16 d2, q3, #2 \n"
+
+ "vst3.u8 {d0, d1, d2}, [%1]! \n"
+
+ "subs %2, #24 \n"
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width), // %2
+ "+r"(src_stride) // %3
+ :
+ : "r4", "q0", "q1", "q2", "q3", "d24", "memory", "cc"
+ );
+}
+
+#define HAS_SCALEROWDOWN38_NEON
+const uint8 shuf38[16] __attribute__ ((aligned(16))) =
+ { 0, 3, 6, 8, 11, 14, 16, 19, 22, 24, 27, 30, 0, 0, 0, 0 };
+const uint8 shuf38_2[16] __attribute__ ((aligned(16))) =
+ { 0, 8, 16, 2, 10, 17, 4, 12, 18, 6, 14, 19, 0, 0, 0, 0 };
+const unsigned short mult38_div6[8] __attribute__ ((aligned(16))) =
+ { 65536 / 12, 65536 / 12, 65536 / 12, 65536 / 12,
+ 65536 / 12, 65536 / 12, 65536 / 12, 65536 / 12 };
+const unsigned short mult38_div9[8] __attribute__ ((aligned(16))) =
+ { 65536 / 18, 65536 / 18, 65536 / 18, 65536 / 18,
+ 65536 / 18, 65536 / 18, 65536 / 18, 65536 / 18 };
+
+// 32 -> 12
+static void ScaleRowDown38_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "vld1.u8 {q3}, [%3] \n"
+ "1: \n"
+ "vld1.u8 {d0, d1, d2, d3}, [%0]! \n"
+ "vtbl.u8 d4, {d0, d1, d2, d3}, d6 \n"
+ "vtbl.u8 d5, {d0, d1, d2, d3}, d7 \n"
+ "vst1.u8 {d4}, [%1]! \n"
+ "vst1.u32 {d5[0]}, [%1]! \n"
+ "subs %2, #12 \n"
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"(shuf38) // %3
+ : "d0", "d1", "d2", "d3", "d4", "d5", "memory", "cc"
+ );
+}
+
+// 32x3 -> 12x1
+static void ScaleRowDown38_3_Int_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "vld1.u16 {q13}, [%4] \n"
+ "vld1.u8 {q14}, [%5] \n"
+ "vld1.u8 {q15}, [%6] \n"
+ "add r4, %0, %3, lsl #1 \n"
+ "add %3, %0 \n"
+ "1: \n"
+
+ // d0 = 00 40 01 41 02 42 03 43
+ // d1 = 10 50 11 51 12 52 13 53
+ // d2 = 20 60 21 61 22 62 23 63
+ // d3 = 30 70 31 71 32 72 33 73
+ "vld4.u8 {d0, d1, d2, d3}, [%0]! \n"
+ "vld4.u8 {d4, d5, d6, d7}, [%3]! \n"
+ "vld4.u8 {d16, d17, d18, d19}, [r4]! \n"
+
+ // Shuffle the input data around to get align the data
+ // so adjacent data can be added. 0,1 - 2,3 - 4,5 - 6,7
+ // d0 = 00 10 01 11 02 12 03 13
+ // d1 = 40 50 41 51 42 52 43 53
+ "vtrn.u8 d0, d1 \n"
+ "vtrn.u8 d4, d5 \n"
+ "vtrn.u8 d16, d17 \n"
+
+ // d2 = 20 30 21 31 22 32 23 33
+ // d3 = 60 70 61 71 62 72 63 73
+ "vtrn.u8 d2, d3 \n"
+ "vtrn.u8 d6, d7 \n"
+ "vtrn.u8 d18, d19 \n"
+
+ // d0 = 00+10 01+11 02+12 03+13
+ // d2 = 40+50 41+51 42+52 43+53
+ "vpaddl.u8 q0, q0 \n"
+ "vpaddl.u8 q2, q2 \n"
+ "vpaddl.u8 q8, q8 \n"
+
+ // d3 = 60+70 61+71 62+72 63+73
+ "vpaddl.u8 d3, d3 \n"
+ "vpaddl.u8 d7, d7 \n"
+ "vpaddl.u8 d19, d19 \n"
+
+ // combine source lines
+ "vadd.u16 q0, q2 \n"
+ "vadd.u16 q0, q8 \n"
+ "vadd.u16 d4, d3, d7 \n"
+ "vadd.u16 d4, d19 \n"
+
+ // dst_ptr[3] = (s[6 + st * 0] + s[7 + st * 0]
+ // + s[6 + st * 1] + s[7 + st * 1]
+ // + s[6 + st * 2] + s[7 + st * 2]) / 6
+ "vqrdmulh.s16 q2, q13 \n"
+ "vmovn.u16 d4, q2 \n"
+
+ // Shuffle 2,3 reg around so that 2 can be added to the
+ // 0,1 reg and 3 can be added to the 4,5 reg. This
+ // requires expanding from u8 to u16 as the 0,1 and 4,5
+ // registers are already expanded. Then do transposes
+ // to get aligned.
+ // q2 = xx 20 xx 30 xx 21 xx 31 xx 22 xx 32 xx 23 xx 33
+ "vmovl.u8 q1, d2 \n"
+ "vmovl.u8 q3, d6 \n"
+ "vmovl.u8 q9, d18 \n"
+
+ // combine source lines
+ "vadd.u16 q1, q3 \n"
+ "vadd.u16 q1, q9 \n"
+
+ // d4 = xx 20 xx 30 xx 22 xx 32
+ // d5 = xx 21 xx 31 xx 23 xx 33
+ "vtrn.u32 d2, d3 \n"
+
+ // d4 = xx 20 xx 21 xx 22 xx 23
+ // d5 = xx 30 xx 31 xx 32 xx 33
+ "vtrn.u16 d2, d3 \n"
+
+ // 0+1+2, 3+4+5
+ "vadd.u16 q0, q1 \n"
+
+ // Need to divide, but can't downshift as the the value
+ // isn't a power of 2. So multiply by 65536 / n
+ // and take the upper 16 bits.
+ "vqrdmulh.s16 q0, q15 \n"
+
+ // Align for table lookup, vtbl requires registers to
+ // be adjacent
+ "vmov.u8 d2, d4 \n"
+
+ "vtbl.u8 d3, {d0, d1, d2}, d28 \n"
+ "vtbl.u8 d4, {d0, d1, d2}, d29 \n"
+
+ "vst1.u8 {d3}, [%1]! \n"
+ "vst1.u32 {d4[0]}, [%1]! \n"
+ "subs %2, #12 \n"
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width), // %2
+ "+r"(src_stride) // %3
+ : "r"(mult38_div6), // %4
+ "r"(shuf38_2), // %5
+ "r"(mult38_div9) // %6
+ : "r4", "q0", "q1", "q2", "q3", "q8", "q9",
+ "q13", "q14", "q15", "memory", "cc"
+ );
+}
+
+// 32x2 -> 12x1
+static void ScaleRowDown38_2_Int_NEON(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "vld1.u16 {q13}, [%4] \n"
+ "vld1.u8 {q14}, [%5] \n"
+ "add %3, %0 \n"
+ "1: \n"
+
+ // d0 = 00 40 01 41 02 42 03 43
+ // d1 = 10 50 11 51 12 52 13 53
+ // d2 = 20 60 21 61 22 62 23 63
+ // d3 = 30 70 31 71 32 72 33 73
+ "vld4.u8 {d0, d1, d2, d3}, [%0]! \n"
+ "vld4.u8 {d4, d5, d6, d7}, [%3]! \n"
+
+ // Shuffle the input data around to get align the data
+ // so adjacent data can be added. 0,1 - 2,3 - 4,5 - 6,7
+ // d0 = 00 10 01 11 02 12 03 13
+ // d1 = 40 50 41 51 42 52 43 53
+ "vtrn.u8 d0, d1 \n"
+ "vtrn.u8 d4, d5 \n"
+
+ // d2 = 20 30 21 31 22 32 23 33
+ // d3 = 60 70 61 71 62 72 63 73
+ "vtrn.u8 d2, d3 \n"
+ "vtrn.u8 d6, d7 \n"
+
+ // d0 = 00+10 01+11 02+12 03+13
+ // d2 = 40+50 41+51 42+52 43+53
+ "vpaddl.u8 q0, q0 \n"
+ "vpaddl.u8 q2, q2 \n"
+
+ // d3 = 60+70 61+71 62+72 63+73
+ "vpaddl.u8 d3, d3 \n"
+ "vpaddl.u8 d7, d7 \n"
+
+ // combine source lines
+ "vadd.u16 q0, q2 \n"
+ "vadd.u16 d4, d3, d7 \n"
+
+ // dst_ptr[3] = (s[6] + s[7] + s[6+st] + s[7+st]) / 4
+ "vqrshrn.u16 d4, q2, #2 \n"
+
+ // Shuffle 2,3 reg around so that 2 can be added to the
+ // 0,1 reg and 3 can be added to the 4,5 reg. This
+ // requires expanding from u8 to u16 as the 0,1 and 4,5
+ // registers are already expanded. Then do transposes
+ // to get aligned.
+ // q2 = xx 20 xx 30 xx 21 xx 31 xx 22 xx 32 xx 23 xx 33
+ "vmovl.u8 q1, d2 \n"
+ "vmovl.u8 q3, d6 \n"
+
+ // combine source lines
+ "vadd.u16 q1, q3 \n"
+
+ // d4 = xx 20 xx 30 xx 22 xx 32
+ // d5 = xx 21 xx 31 xx 23 xx 33
+ "vtrn.u32 d2, d3 \n"
+
+ // d4 = xx 20 xx 21 xx 22 xx 23
+ // d5 = xx 30 xx 31 xx 32 xx 33
+ "vtrn.u16 d2, d3 \n"
+
+ // 0+1+2, 3+4+5
+ "vadd.u16 q0, q1 \n"
+
+ // Need to divide, but can't downshift as the the value
+ // isn't a power of 2. So multiply by 65536 / n
+ // and take the upper 16 bits.
+ "vqrdmulh.s16 q0, q13 \n"
+
+ // Align for table lookup, vtbl requires registers to
+ // be adjacent
+ "vmov.u8 d2, d4 \n"
+
+ "vtbl.u8 d3, {d0, d1, d2}, d28 \n"
+ "vtbl.u8 d4, {d0, d1, d2}, d29 \n"
+
+ "vst1.u8 {d3}, [%1]! \n"
+ "vst1.u32 {d4[0]}, [%1]! \n"
+ "subs %2, #12 \n"
+ "bhi 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width), // %2
+ "+r"(src_stride) // %3
+ : "r"(mult38_div6), // %4
+ "r"(shuf38_2) // %5
+ : "q0", "q1", "q2", "q3", "q13", "q14", "memory", "cc"
+ );
+}
+
+/**
+ * SSE2 downscalers with interpolation.
+ *
+ * Provided by Frank Barchard (fbarchard@google.com)
+ *
+ */
+
+// Constants for SSE2 code
+#elif (defined(_M_IX86) || defined(__i386__) || defined(__x86_64__)) && \
+ !defined(YUV_DISABLE_ASM)
+#if defined(_MSC_VER)
+#define TALIGN16(t, var) __declspec(align(16)) t _ ## var
+#elif (defined(__APPLE__) || defined(__MINGW32__) || defined(__CYGWIN__)) && defined(__i386__)
+#define TALIGN16(t, var) t var __attribute__((aligned(16)))
+#else
+#define TALIGN16(t, var) t _ ## var __attribute__((aligned(16)))
+#endif
+
+#if (defined(__APPLE__) || defined(__MINGW32__) || defined(__CYGWIN__)) && \
+ defined(__i386__)
+#define DECLARE_FUNCTION(name) \
+ ".text \n" \
+ ".globl _" #name " \n" \
+"_" #name ": \n"
+#else
+#define DECLARE_FUNCTION(name) \
+ ".text \n" \
+ ".global " #name " \n" \
+#name ": \n"
+#endif
+
+
+// Offsets for source bytes 0 to 9
+//extern "C"
+TALIGN16(const uint8, shuf0[16]) =
+ { 0, 1, 3, 4, 5, 7, 8, 9, 128, 128, 128, 128, 128, 128, 128, 128 };
+
+// Offsets for source bytes 11 to 20 with 8 subtracted = 3 to 12.
+//extern "C"
+TALIGN16(const uint8, shuf1[16]) =
+ { 3, 4, 5, 7, 8, 9, 11, 12, 128, 128, 128, 128, 128, 128, 128, 128 };
+
+// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
+//extern "C"
+TALIGN16(const uint8, shuf2[16]) =
+ { 5, 7, 8, 9, 11, 12, 13, 15, 128, 128, 128, 128, 128, 128, 128, 128 };
+
+// Offsets for source bytes 0 to 10
+//extern "C"
+TALIGN16(const uint8, shuf01[16]) =
+ { 0, 1, 1, 2, 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10 };
+
+// Offsets for source bytes 10 to 21 with 8 subtracted = 3 to 13.
+//extern "C"
+TALIGN16(const uint8, shuf11[16]) =
+ { 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13 };
+
+// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
+//extern "C"
+TALIGN16(const uint8, shuf21[16]) =
+ { 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13, 13, 14, 14, 15 };
+
+// Coefficients for source bytes 0 to 10
+//extern "C"
+TALIGN16(const uint8, madd01[16]) =
+ { 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2 };
+
+// Coefficients for source bytes 10 to 21
+//extern "C"
+TALIGN16(const uint8, madd11[16]) =
+ { 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1 };
+
+// Coefficients for source bytes 21 to 31
+//extern "C"
+TALIGN16(const uint8, madd21[16]) =
+ { 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3 };
+
+// Coefficients for source bytes 21 to 31
+//extern "C"
+TALIGN16(const int16, round34[8]) =
+ { 2, 2, 2, 2, 2, 2, 2, 2 };
+
+//extern "C"
+TALIGN16(const uint8, shuf38a[16]) =
+ { 0, 3, 6, 8, 11, 14, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
+
+//extern "C"
+TALIGN16(const uint8, shuf38b[16]) =
+ { 128, 128, 128, 128, 128, 128, 0, 3, 6, 8, 11, 14, 128, 128, 128, 128 };
+
+// Arrange words 0,3,6 into 0,1,2
+//extern "C"
+TALIGN16(const uint8, shufac0[16]) =
+ { 0, 1, 6, 7, 12, 13, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
+
+// Arrange words 0,3,6 into 3,4,5
+//extern "C"
+TALIGN16(const uint8, shufac3[16]) =
+ { 128, 128, 128, 128, 128, 128, 0, 1, 6, 7, 12, 13, 128, 128, 128, 128 };
+
+// Scaling values for boxes of 3x3 and 2x3
+//extern "C"
+TALIGN16(const uint16, scaleac3[8]) =
+ { 65536 / 9, 65536 / 9, 65536 / 6, 65536 / 9, 65536 / 9, 65536 / 6, 0, 0 };
+
+// Arrange first value for pixels 0,1,2,3,4,5
+//extern "C"
+TALIGN16(const uint8, shufab0[16]) =
+ { 0, 128, 3, 128, 6, 128, 8, 128, 11, 128, 14, 128, 128, 128, 128, 128 };
+
+// Arrange second value for pixels 0,1,2,3,4,5
+//extern "C"
+TALIGN16(const uint8, shufab1[16]) =
+ { 1, 128, 4, 128, 7, 128, 9, 128, 12, 128, 15, 128, 128, 128, 128, 128 };
+
+// Arrange third value for pixels 0,1,2,3,4,5
+//extern "C"
+TALIGN16(const uint8, shufab2[16]) =
+ { 2, 128, 5, 128, 128, 128, 10, 128, 13, 128, 128, 128, 128, 128, 128, 128 };
+
+// Scaling values for boxes of 3x2 and 2x2
+//extern "C"
+TALIGN16(const uint16, scaleab2[8]) =
+ { 65536 / 3, 65536 / 3, 65536 / 2, 65536 / 3, 65536 / 3, 65536 / 2, 0, 0 };
+#endif
+
+#if defined(_M_IX86) && !defined(YUV_DISABLE_ASM)
+
+#define HAS_SCALEROWDOWN2_SSE2
+// Reads 32 pixels, throws half away and writes 16 pixels.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
+__declspec(naked)
+static void ScaleRowDown2_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ mov eax, [esp + 4] // src_ptr
+ // src_stride ignored
+ mov edx, [esp + 12] // dst_ptr
+ mov ecx, [esp + 16] // dst_width
+ pcmpeqb xmm5, xmm5 // generate mask 0x00ff00ff
+ psrlw xmm5, 8
+
+ wloop:
+ movdqa xmm0, [eax]
+ movdqa xmm1, [eax + 16]
+ lea eax, [eax + 32]
+ pand xmm0, xmm5
+ pand xmm1, xmm5
+ packuswb xmm0, xmm1
+ movdqa [edx], xmm0
+ lea edx, [edx + 16]
+ sub ecx, 16
+ ja wloop
+
+ ret
+ }
+}
+// Blends 32x2 rectangle to 16x1.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
+__declspec(naked)
+void ScaleRowDown2Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ push esi
+ mov eax, [esp + 4 + 4] // src_ptr
+ mov esi, [esp + 4 + 8] // src_stride
+ mov edx, [esp + 4 + 12] // dst_ptr
+ mov ecx, [esp + 4 + 16] // dst_width
+ pcmpeqb xmm5, xmm5 // generate mask 0x00ff00ff
+ psrlw xmm5, 8
+
+ wloop:
+ movdqa xmm0, [eax]
+ movdqa xmm1, [eax + 16]
+ movdqa xmm2, [eax + esi]
+ movdqa xmm3, [eax + esi + 16]
+ lea eax, [eax + 32]
+ pavgb xmm0, xmm2 // average rows
+ pavgb xmm1, xmm3
+
+ movdqa xmm2, xmm0 // average columns (32 to 16 pixels)
+ psrlw xmm0, 8
+ movdqa xmm3, xmm1
+ psrlw xmm1, 8
+ pand xmm2, xmm5
+ pand xmm3, xmm5
+ pavgw xmm0, xmm2
+ pavgw xmm1, xmm3
+ packuswb xmm0, xmm1
+
+ movdqa [edx], xmm0
+ lea edx, [edx + 16]
+ sub ecx, 16
+ ja wloop
+
+ pop esi
+ ret
+ }
+}
+
+#define HAS_SCALEROWDOWN4_SSE2
+// Point samples 32 pixels to 8 pixels.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
+__declspec(naked)
+static void ScaleRowDown4_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ // src_stride ignored
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ pcmpeqb xmm5, xmm5 // generate mask 0x000000ff
+ psrld xmm5, 24
+
+ wloop:
+ movdqa xmm0, [esi]
+ movdqa xmm1, [esi + 16]
+ lea esi, [esi + 32]
+ pand xmm0, xmm5
+ pand xmm1, xmm5
+ packuswb xmm0, xmm1
+ packuswb xmm0, xmm0
+ movq qword ptr [edi], xmm0
+ lea edi, [edi + 8]
+ sub ecx, 8
+ ja wloop
+
+ popad
+ ret
+ }
+}
+
+// Blends 32x4 rectangle to 8x1.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
+__declspec(naked)
+static void ScaleRowDown4Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ mov ebx, [esp + 32 + 8] // src_stride
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ pcmpeqb xmm7, xmm7 // generate mask 0x00ff00ff
+ psrlw xmm7, 8
+ lea edx, [ebx + ebx * 2] // src_stride * 3
+
+ wloop:
+ movdqa xmm0, [esi]
+ movdqa xmm1, [esi + 16]
+ movdqa xmm2, [esi + ebx]
+ movdqa xmm3, [esi + ebx + 16]
+ pavgb xmm0, xmm2 // average rows
+ pavgb xmm1, xmm3
+ movdqa xmm2, [esi + ebx * 2]
+ movdqa xmm3, [esi + ebx * 2 + 16]
+ movdqa xmm4, [esi + edx]
+ movdqa xmm5, [esi + edx + 16]
+ lea esi, [esi + 32]
+ pavgb xmm2, xmm4
+ pavgb xmm3, xmm5
+ pavgb xmm0, xmm2
+ pavgb xmm1, xmm3
+
+ movdqa xmm2, xmm0 // average columns (32 to 16 pixels)
+ psrlw xmm0, 8
+ movdqa xmm3, xmm1
+ psrlw xmm1, 8
+ pand xmm2, xmm7
+ pand xmm3, xmm7
+ pavgw xmm0, xmm2
+ pavgw xmm1, xmm3
+ packuswb xmm0, xmm1
+
+ movdqa xmm2, xmm0 // average columns (16 to 8 pixels)
+ psrlw xmm0, 8
+ pand xmm2, xmm7
+ pavgw xmm0, xmm2
+ packuswb xmm0, xmm0
+
+ movq qword ptr [edi], xmm0
+ lea edi, [edi + 8]
+ sub ecx, 8
+ ja wloop
+
+ popad
+ ret
+ }
+}
+
+#define HAS_SCALEROWDOWN8_SSE2
+// Point samples 32 pixels to 4 pixels.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 4 byte aligned.
+__declspec(naked)
+static void ScaleRowDown8_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ // src_stride ignored
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ pcmpeqb xmm5, xmm5 // generate mask isolating 1 src 8 bytes
+ psrlq xmm5, 56
+
+ wloop:
+ movdqa xmm0, [esi]
+ movdqa xmm1, [esi + 16]
+ lea esi, [esi + 32]
+ pand xmm0, xmm5
+ pand xmm1, xmm5
+ packuswb xmm0, xmm1 // 32->16
+ packuswb xmm0, xmm0 // 16->8
+ packuswb xmm0, xmm0 // 8->4
+ movd dword ptr [edi], xmm0
+ lea edi, [edi + 4]
+ sub ecx, 4
+ ja wloop
+
+ popad
+ ret
+ }
+}
+
+// Blends 32x8 rectangle to 4x1.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 4 byte aligned.
+__declspec(naked)
+static void ScaleRowDown8Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ mov ebx, [esp + 32 + 8] // src_stride
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ lea edx, [ebx + ebx * 2] // src_stride * 3
+ pxor xmm7, xmm7
+
+ wloop:
+ movdqa xmm0, [esi] // average 8 rows to 1
+ movdqa xmm1, [esi + 16]
+ movdqa xmm2, [esi + ebx]
+ movdqa xmm3, [esi + ebx + 16]
+ pavgb xmm0, xmm2
+ pavgb xmm1, xmm3
+ movdqa xmm2, [esi + ebx * 2]
+ movdqa xmm3, [esi + ebx * 2 + 16]
+ movdqa xmm4, [esi + edx]
+ movdqa xmm5, [esi + edx + 16]
+ lea ebp, [esi + ebx * 4]
+ lea esi, [esi + 32]
+ pavgb xmm2, xmm4
+ pavgb xmm3, xmm5
+ pavgb xmm0, xmm2
+ pavgb xmm1, xmm3
+
+ movdqa xmm2, [ebp]
+ movdqa xmm3, [ebp + 16]
+ movdqa xmm4, [ebp + ebx]
+ movdqa xmm5, [ebp + ebx + 16]
+ pavgb xmm2, xmm4
+ pavgb xmm3, xmm5
+ movdqa xmm4, [ebp + ebx * 2]
+ movdqa xmm5, [ebp + ebx * 2 + 16]
+ movdqa xmm6, [ebp + edx]
+ pavgb xmm4, xmm6
+ movdqa xmm6, [ebp + edx + 16]
+ pavgb xmm5, xmm6
+ pavgb xmm2, xmm4
+ pavgb xmm3, xmm5
+ pavgb xmm0, xmm2
+ pavgb xmm1, xmm3
+
+ psadbw xmm0, xmm7 // average 32 pixels to 4
+ psadbw xmm1, xmm7
+ pshufd xmm0, xmm0, 0xd8 // x1x0 -> xx01
+ pshufd xmm1, xmm1, 0x8d // x3x2 -> 32xx
+ por xmm0, xmm1 // -> 3201
+ psrlw xmm0, 3
+ packuswb xmm0, xmm0
+ packuswb xmm0, xmm0
+ movd dword ptr [edi], xmm0
+
+ lea edi, [edi + 4]
+ sub ecx, 4
+ ja wloop
+
+ popad
+ ret
+ }
+}
+
+#define HAS_SCALEROWDOWN34_SSSE3
+// Point samples 32 pixels to 24 pixels.
+// Produces three 8 byte values. For each 8 bytes, 16 bytes are read.
+// Then shuffled to do the scaling.
+
+// Note that movdqa+palign may be better than movdqu.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
+__declspec(naked)
+static void ScaleRowDown34_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ // src_stride ignored
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ movdqa xmm3, _shuf0
+ movdqa xmm4, _shuf1
+ movdqa xmm5, _shuf2
+
+ wloop:
+ movdqa xmm0, [esi]
+ movdqa xmm1, [esi + 16]
+ lea esi, [esi + 32]
+ movdqa xmm2, xmm1
+ palignr xmm1, xmm0, 8
+ pshufb xmm0, xmm3
+ pshufb xmm1, xmm4
+ pshufb xmm2, xmm5
+ movq qword ptr [edi], xmm0
+ movq qword ptr [edi + 8], xmm1
+ movq qword ptr [edi + 16], xmm2
+ lea edi, [edi + 24]
+ sub ecx, 24
+ ja wloop
+
+ popad
+ ret
+ }
+}
+
+// Blends 32x2 rectangle to 24x1
+// Produces three 8 byte values. For each 8 bytes, 16 bytes are read.
+// Then shuffled to do the scaling.
+
+// Register usage:
+// xmm0 src_row 0
+// xmm1 src_row 1
+// xmm2 shuf 0
+// xmm3 shuf 1
+// xmm4 shuf 2
+// xmm5 madd 0
+// xmm6 madd 1
+// xmm7 round34
+
+// Note that movdqa+palign may be better than movdqu.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
+__declspec(naked)
+static void ScaleRowDown34_1_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ mov ebx, [esp + 32 + 8] // src_stride
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ movdqa xmm2, _shuf01
+ movdqa xmm3, _shuf11
+ movdqa xmm4, _shuf21
+ movdqa xmm5, _madd01
+ movdqa xmm6, _madd11
+ movdqa xmm7, _round34
+
+ wloop:
+ movdqa xmm0, [esi] // pixels 0..7
+ movdqa xmm1, [esi+ebx]
+ pavgb xmm0, xmm1
+ pshufb xmm0, xmm2
+ pmaddubsw xmm0, xmm5
+ paddsw xmm0, xmm7
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edi], xmm0
+ movdqu xmm0, [esi+8] // pixels 8..15
+ movdqu xmm1, [esi+ebx+8]
+ pavgb xmm0, xmm1
+ pshufb xmm0, xmm3
+ pmaddubsw xmm0, xmm6
+ paddsw xmm0, xmm7
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edi+8], xmm0
+ movdqa xmm0, [esi+16] // pixels 16..23
+ movdqa xmm1, [esi+ebx+16]
+ lea esi, [esi+32]
+ pavgb xmm0, xmm1
+ pshufb xmm0, xmm4
+ movdqa xmm1, _madd21
+ pmaddubsw xmm0, xmm1
+ paddsw xmm0, xmm7
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edi+16], xmm0
+ lea edi, [edi+24]
+ sub ecx, 24
+ ja wloop
+
+ popad
+ ret
+ }
+}
+
+// Note that movdqa+palign may be better than movdqu.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
+__declspec(naked)
+static void ScaleRowDown34_0_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ mov ebx, [esp + 32 + 8] // src_stride
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ movdqa xmm2, _shuf01
+ movdqa xmm3, _shuf11
+ movdqa xmm4, _shuf21
+ movdqa xmm5, _madd01
+ movdqa xmm6, _madd11
+ movdqa xmm7, _round34
+
+ wloop:
+ movdqa xmm0, [esi] // pixels 0..7
+ movdqa xmm1, [esi+ebx]
+ pavgb xmm1, xmm0
+ pavgb xmm0, xmm1
+ pshufb xmm0, xmm2
+ pmaddubsw xmm0, xmm5
+ paddsw xmm0, xmm7
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edi], xmm0
+ movdqu xmm0, [esi+8] // pixels 8..15
+ movdqu xmm1, [esi+ebx+8]
+ pavgb xmm1, xmm0
+ pavgb xmm0, xmm1
+ pshufb xmm0, xmm3
+ pmaddubsw xmm0, xmm6
+ paddsw xmm0, xmm7
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edi+8], xmm0
+ movdqa xmm0, [esi+16] // pixels 16..23
+ movdqa xmm1, [esi+ebx+16]
+ lea esi, [esi+32]
+ pavgb xmm1, xmm0
+ pavgb xmm0, xmm1
+ pshufb xmm0, xmm4
+ movdqa xmm1, _madd21
+ pmaddubsw xmm0, xmm1
+ paddsw xmm0, xmm7
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edi+16], xmm0
+ lea edi, [edi+24]
+ sub ecx, 24
+ ja wloop
+
+ popad
+ ret
+ }
+}
+
+#define HAS_SCALEROWDOWN38_SSSE3
+// 3/8 point sampler
+
+// Scale 32 pixels to 12
+__declspec(naked)
+static void ScaleRowDown38_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ mov edx, [esp + 32 + 8] // src_stride
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ movdqa xmm4, _shuf38a
+ movdqa xmm5, _shuf38b
+
+ xloop:
+ movdqa xmm0, [esi] // 16 pixels -> 0,1,2,3,4,5
+ movdqa xmm1, [esi + 16] // 16 pixels -> 6,7,8,9,10,11
+ lea esi, [esi + 32]
+ pshufb xmm0, xmm4
+ pshufb xmm1, xmm5
+ paddusb xmm0, xmm1
+
+ movq qword ptr [edi], xmm0 // write 12 pixels
+ movhlps xmm1, xmm0
+ movd [edi + 8], xmm1
+ lea edi, [edi + 12]
+ sub ecx, 12
+ ja xloop
+
+ popad
+ ret
+ }
+}
+
+// Scale 16x3 pixels to 6x1 with interpolation
+__declspec(naked)
+static void ScaleRowDown38_3_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ mov edx, [esp + 32 + 8] // src_stride
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ movdqa xmm4, _shufac0
+ movdqa xmm5, _shufac3
+ movdqa xmm6, _scaleac3
+ pxor xmm7, xmm7
+
+ xloop:
+ movdqa xmm0, [esi] // sum up 3 rows into xmm0/1
+ movdqa xmm2, [esi + edx]
+ movhlps xmm1, xmm0
+ movhlps xmm3, xmm2
+ punpcklbw xmm0, xmm7
+ punpcklbw xmm1, xmm7
+ punpcklbw xmm2, xmm7
+ punpcklbw xmm3, xmm7
+ paddusw xmm0, xmm2
+ paddusw xmm1, xmm3
+ movdqa xmm2, [esi + edx * 2]
+ lea esi, [esi + 16]
+ movhlps xmm3, xmm2
+ punpcklbw xmm2, xmm7
+ punpcklbw xmm3, xmm7
+ paddusw xmm0, xmm2
+ paddusw xmm1, xmm3
+
+ movdqa xmm2, xmm0 // 8 pixels -> 0,1,2 of xmm2
+ psrldq xmm0, 2
+ paddusw xmm2, xmm0
+ psrldq xmm0, 2
+ paddusw xmm2, xmm0
+ pshufb xmm2, xmm4
+
+ movdqa xmm3, xmm1 // 8 pixels -> 3,4,5 of xmm2
+ psrldq xmm1, 2
+ paddusw xmm3, xmm1
+ psrldq xmm1, 2
+ paddusw xmm3, xmm1
+ pshufb xmm3, xmm5
+ paddusw xmm2, xmm3
+
+ pmulhuw xmm2, xmm6 // divide by 9,9,6, 9,9,6
+ packuswb xmm2, xmm2
+
+ movd [edi], xmm2 // write 6 pixels
+ pextrw eax, xmm2, 2
+ mov [edi + 4], ax
+ lea edi, [edi + 6]
+ sub ecx, 6
+ ja xloop
+
+ popad
+ ret
+ }
+}
+
+// Scale 16x2 pixels to 6x1 with interpolation
+__declspec(naked)
+static void ScaleRowDown38_2_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ mov edx, [esp + 32 + 8] // src_stride
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ movdqa xmm4, _shufab0
+ movdqa xmm5, _shufab1
+ movdqa xmm6, _shufab2
+ movdqa xmm7, _scaleab2
+
+ xloop:
+ movdqa xmm2, [esi] // average 2 rows into xmm2
+ pavgb xmm2, [esi + edx]
+ lea esi, [esi + 16]
+
+ movdqa xmm0, xmm2 // 16 pixels -> 0,1,2,3,4,5 of xmm0
+ pshufb xmm0, xmm4
+ movdqa xmm1, xmm2
+ pshufb xmm1, xmm5
+ paddusw xmm0, xmm1
+ pshufb xmm2, xmm6
+ paddusw xmm0, xmm2
+
+ pmulhuw xmm0, xmm7 // divide by 3,3,2, 3,3,2
+ packuswb xmm0, xmm0
+
+ movd [edi], xmm0 // write 6 pixels
+ pextrw eax, xmm0, 2
+ mov [edi + 4], ax
+ lea edi, [edi + 6]
+ sub ecx, 6
+ ja xloop
+
+ popad
+ ret
+ }
+}
+
+#define HAS_SCALEADDROWS_SSE2
+
+// Reads 8xN bytes and produces 16 shorts at a time.
+__declspec(naked)
+static void ScaleAddRows_SSE2(const uint8* src_ptr, int src_stride,
+ uint16* dst_ptr, int src_width,
+ int src_height) {
+ __asm {
+ pushad
+ mov esi, [esp + 32 + 4] // src_ptr
+ mov edx, [esp + 32 + 8] // src_stride
+ mov edi, [esp + 32 + 12] // dst_ptr
+ mov ecx, [esp + 32 + 16] // dst_width
+ mov ebx, [esp + 32 + 20] // height
+ pxor xmm5, xmm5
+ dec ebx
+
+ xloop:
+ // first row
+ movdqa xmm2, [esi]
+ lea eax, [esi + edx]
+ movhlps xmm3, xmm2
+ mov ebp, ebx
+ punpcklbw xmm2, xmm5
+ punpcklbw xmm3, xmm5
+
+ // sum remaining rows
+ yloop:
+ movdqa xmm0, [eax] // read 16 pixels
+ lea eax, [eax + edx] // advance to next row
+ movhlps xmm1, xmm0
+ punpcklbw xmm0, xmm5
+ punpcklbw xmm1, xmm5
+ paddusw xmm2, xmm0 // sum 16 words
+ paddusw xmm3, xmm1
+ sub ebp, 1
+ ja yloop
+
+ movdqa [edi], xmm2
+ movdqa [edi + 16], xmm3
+ lea edi, [edi + 32]
+ lea esi, [esi + 16]
+
+ sub ecx, 16
+ ja xloop
+
+ popad
+ ret
+ }
+}
+
+// Bilinear row filtering combines 16x2 -> 16x1. SSE2 version.
+#define HAS_SCALEFILTERROWS_SSE2
+__declspec(naked)
+static void ScaleFilterRows_SSE2(uint8* dst_ptr, const uint8* src_ptr,
+ int src_stride, int dst_width,
+ int source_y_fraction) {
+ __asm {
+ push esi
+ push edi
+ mov edi, [esp + 8 + 4] // dst_ptr
+ mov esi, [esp + 8 + 8] // src_ptr
+ mov edx, [esp + 8 + 12] // src_stride
+ mov ecx, [esp + 8 + 16] // dst_width
+ mov eax, [esp + 8 + 20] // source_y_fraction (0..255)
+ cmp eax, 0
+ je xloop1
+ cmp eax, 128
+ je xloop2
+
+ movd xmm6, eax // xmm6 = y fraction
+ punpcklwd xmm6, xmm6
+ pshufd xmm6, xmm6, 0
+ neg eax // xmm5 = 256 - y fraction
+ add eax, 256
+ movd xmm5, eax
+ punpcklwd xmm5, xmm5
+ pshufd xmm5, xmm5, 0
+ pxor xmm7, xmm7
+
+ xloop:
+ movdqa xmm0, [esi]
+ movdqa xmm2, [esi + edx]
+ lea esi, [esi + 16]
+ movdqa xmm1, xmm0
+ movdqa xmm3, xmm2
+ punpcklbw xmm0, xmm7
+ punpcklbw xmm2, xmm7
+ punpckhbw xmm1, xmm7
+ punpckhbw xmm3, xmm7
+ pmullw xmm0, xmm5 // scale row 0
+ pmullw xmm1, xmm5
+ pmullw xmm2, xmm6 // scale row 1
+ pmullw xmm3, xmm6
+ paddusw xmm0, xmm2 // sum rows
+ paddusw xmm1, xmm3
+ psrlw xmm0, 8
+ psrlw xmm1, 8
+ packuswb xmm0, xmm1
+ movdqa [edi], xmm0
+ lea edi, [edi + 16]
+ sub ecx, 16
+ ja xloop
+
+ mov al, [edi - 1]
+ mov [edi], al
+ pop edi
+ pop esi
+ ret
+
+ xloop1:
+ movdqa xmm0, [esi]
+ lea esi, [esi + 16]
+ movdqa [edi], xmm0
+ lea edi, [edi + 16]
+ sub ecx, 16
+ ja xloop1
+
+ mov al, [edi - 1]
+ mov [edi], al
+ pop edi
+ pop esi
+ ret
+
+ xloop2:
+ movdqa xmm0, [esi]
+ movdqa xmm2, [esi + edx]
+ lea esi, [esi + 16]
+ pavgb xmm0, xmm2
+ movdqa [edi], xmm0
+ lea edi, [edi + 16]
+ sub ecx, 16
+ ja xloop2
+
+ mov al, [edi - 1]
+ mov [edi], al
+ pop edi
+ pop esi
+ ret
+ }
+}
+
+// Bilinear row filtering combines 16x2 -> 16x1. SSSE3 version.
+#define HAS_SCALEFILTERROWS_SSSE3
+__declspec(naked)
+static void ScaleFilterRows_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
+ int src_stride, int dst_width,
+ int source_y_fraction) {
+ __asm {
+ push esi
+ push edi
+ mov edi, [esp + 8 + 4] // dst_ptr
+ mov esi, [esp + 8 + 8] // src_ptr
+ mov edx, [esp + 8 + 12] // src_stride
+ mov ecx, [esp + 8 + 16] // dst_width
+ mov eax, [esp + 8 + 20] // source_y_fraction (0..255)
+ cmp eax, 0
+ je xloop1
+ cmp eax, 128
+ je xloop2
+
+ shr eax, 1
+ mov ah,al
+ neg al
+ add al, 128
+ movd xmm5, eax
+ punpcklwd xmm5, xmm5
+ pshufd xmm5, xmm5, 0
+
+ xloop:
+ movdqa xmm0, [esi]
+ movdqa xmm2, [esi + edx]
+ lea esi, [esi + 16]
+ movdqa xmm1, xmm0
+ punpcklbw xmm0, xmm2
+ punpckhbw xmm1, xmm2
+ pmaddubsw xmm0, xmm5
+ pmaddubsw xmm1, xmm5
+ psrlw xmm0, 7
+ psrlw xmm1, 7
+ packuswb xmm0, xmm1
+ movdqa [edi], xmm0
+ lea edi, [edi + 16]
+ sub ecx, 16
+ ja xloop
+
+ mov al, [edi - 1]
+ mov [edi], al
+ pop edi
+ pop esi
+ ret
+
+ xloop1:
+ movdqa xmm0, [esi]
+ lea esi, [esi + 16]
+ movdqa [edi], xmm0
+ lea edi, [edi + 16]
+ sub ecx, 16
+ ja xloop1
+
+ mov al, [edi - 1]
+ mov [edi], al
+ pop edi
+ pop esi
+ ret
+
+ xloop2:
+ movdqa xmm0, [esi]
+ movdqa xmm2, [esi + edx]
+ lea esi, [esi + 16]
+ pavgb xmm0, xmm2
+ movdqa [edi], xmm0
+ lea edi, [edi + 16]
+ sub ecx, 16
+ ja xloop2
+
+ mov al, [edi - 1]
+ mov [edi], al
+ pop edi
+ pop esi
+ ret
+
+ }
+}
+
+// Note that movdqa+palign may be better than movdqu.
+// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
+__declspec(naked)
+static void ScaleFilterCols34_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
+ int dst_width) {
+ __asm {
+ mov edx, [esp + 4] // dst_ptr
+ mov eax, [esp + 8] // src_ptr
+ mov ecx, [esp + 12] // dst_width
+ movdqa xmm1, _round34
+ movdqa xmm2, _shuf01
+ movdqa xmm3, _shuf11
+ movdqa xmm4, _shuf21
+ movdqa xmm5, _madd01
+ movdqa xmm6, _madd11
+ movdqa xmm7, _madd21
+
+ wloop:
+ movdqa xmm0, [eax] // pixels 0..7
+ pshufb xmm0, xmm2
+ pmaddubsw xmm0, xmm5
+ paddsw xmm0, xmm1
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edx], xmm0
+ movdqu xmm0, [eax+8] // pixels 8..15
+ pshufb xmm0, xmm3
+ pmaddubsw xmm0, xmm6
+ paddsw xmm0, xmm1
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edx+8], xmm0
+ movdqa xmm0, [eax+16] // pixels 16..23
+ lea eax, [eax+32]
+ pshufb xmm0, xmm4
+ pmaddubsw xmm0, xmm7
+ paddsw xmm0, xmm1
+ psrlw xmm0, 2
+ packuswb xmm0, xmm0
+ movq qword ptr [edx+16], xmm0
+ lea edx, [edx+24]
+ sub ecx, 24
+ ja wloop
+ ret
+ }
+}
+
+#elif (defined(__x86_64__) || defined(__i386__)) && !defined(YUV_DISABLE_ASM)
+
+// GCC versions of row functions are verbatim conversions from Visual C.
+// Generated using gcc disassembly on Visual C object file:
+// objdump -D yuvscaler.obj >yuvscaler.txt
+#define HAS_SCALEROWDOWN2_SSE2
+static void ScaleRowDown2_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "pcmpeqb %%xmm5,%%xmm5 \n"
+ "psrlw $0x8,%%xmm5 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa 0x10(%0),%%xmm1 \n"
+ "lea 0x20(%0),%0 \n"
+ "pand %%xmm5,%%xmm0 \n"
+ "pand %%xmm5,%%xmm1 \n"
+ "packuswb %%xmm1,%%xmm0 \n"
+ "movdqa %%xmm0,(%1) \n"
+ "lea 0x10(%1),%1 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ :
+ : "memory", "cc"
+);
+}
+
+static void ScaleRowDown2Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "pcmpeqb %%xmm5,%%xmm5 \n"
+ "psrlw $0x8,%%xmm5 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa 0x10(%0),%%xmm1 \n"
+ "movdqa (%0,%3,1),%%xmm2 \n"
+ "movdqa 0x10(%0,%3,1),%%xmm3 \n"
+ "lea 0x20(%0),%0 \n"
+ "pavgb %%xmm2,%%xmm0 \n"
+ "pavgb %%xmm3,%%xmm1 \n"
+ "movdqa %%xmm0,%%xmm2 \n"
+ "psrlw $0x8,%%xmm0 \n"
+ "movdqa %%xmm1,%%xmm3 \n"
+ "psrlw $0x8,%%xmm1 \n"
+ "pand %%xmm5,%%xmm2 \n"
+ "pand %%xmm5,%%xmm3 \n"
+ "pavgw %%xmm2,%%xmm0 \n"
+ "pavgw %%xmm3,%%xmm1 \n"
+ "packuswb %%xmm1,%%xmm0 \n"
+ "movdqa %%xmm0,(%1) \n"
+ "lea 0x10(%1),%1 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"((intptr_t)(src_stride)) // %3
+ : "memory", "cc"
+);
+}
+
+#define HAS_SCALEROWDOWN4_SSE2
+static void ScaleRowDown4_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "pcmpeqb %%xmm5,%%xmm5 \n"
+ "psrld $0x18,%%xmm5 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa 0x10(%0),%%xmm1 \n"
+ "lea 0x20(%0),%0 \n"
+ "pand %%xmm5,%%xmm0 \n"
+ "pand %%xmm5,%%xmm1 \n"
+ "packuswb %%xmm1,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movq %%xmm0,(%1) \n"
+ "lea 0x8(%1),%1 \n"
+ "sub $0x8,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ :
+ : "memory", "cc"
+);
+}
+
+static void ScaleRowDown4Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ intptr_t temp = 0;
+ asm volatile (
+ "pcmpeqb %%xmm7,%%xmm7 \n"
+ "psrlw $0x8,%%xmm7 \n"
+ "lea (%4,%4,2),%3 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa 0x10(%0),%%xmm1 \n"
+ "movdqa (%0,%4,1),%%xmm2 \n"
+ "movdqa 0x10(%0,%4,1),%%xmm3 \n"
+ "pavgb %%xmm2,%%xmm0 \n"
+ "pavgb %%xmm3,%%xmm1 \n"
+ "movdqa (%0,%4,2),%%xmm2 \n"
+ "movdqa 0x10(%0,%4,2),%%xmm3 \n"
+ "movdqa (%0,%3,1),%%xmm4 \n"
+ "movdqa 0x10(%0,%3,1),%%xmm5 \n"
+ "lea 0x20(%0),%0 \n"
+ "pavgb %%xmm4,%%xmm2 \n"
+ "pavgb %%xmm2,%%xmm0 \n"
+ "pavgb %%xmm5,%%xmm3 \n"
+ "pavgb %%xmm3,%%xmm1 \n"
+ "movdqa %%xmm0,%%xmm2 \n"
+ "psrlw $0x8,%%xmm0 \n"
+ "movdqa %%xmm1,%%xmm3 \n"
+ "psrlw $0x8,%%xmm1 \n"
+ "pand %%xmm7,%%xmm2 \n"
+ "pand %%xmm7,%%xmm3 \n"
+ "pavgw %%xmm2,%%xmm0 \n"
+ "pavgw %%xmm3,%%xmm1 \n"
+ "packuswb %%xmm1,%%xmm0 \n"
+ "movdqa %%xmm0,%%xmm2 \n"
+ "psrlw $0x8,%%xmm0 \n"
+ "pand %%xmm7,%%xmm2 \n"
+ "pavgw %%xmm2,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movq %%xmm0,(%1) \n"
+ "lea 0x8(%1),%1 \n"
+ "sub $0x8,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width), // %2
+ "+r"(temp) // %3
+ : "r"((intptr_t)(src_stride)) // %4
+ : "memory", "cc"
+#if defined(__x86_64__)
+ , "xmm6", "xmm7"
+#endif
+);
+}
+
+#define HAS_SCALEROWDOWN8_SSE2
+static void ScaleRowDown8_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "pcmpeqb %%xmm5,%%xmm5 \n"
+ "psrlq $0x38,%%xmm5 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa 0x10(%0),%%xmm1 \n"
+ "lea 0x20(%0),%0 \n"
+ "pand %%xmm5,%%xmm0 \n"
+ "pand %%xmm5,%%xmm1 \n"
+ "packuswb %%xmm1,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movd %%xmm0,(%1) \n"
+ "lea 0x4(%1),%1 \n"
+ "sub $0x4,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ :
+ : "memory", "cc"
+);
+}
+
+#if defined(__i386__)
+void ScaleRowDown8Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ asm(
+ DECLARE_FUNCTION(ScaleRowDown8Int_SSE2)
+ "pusha \n"
+ "mov 0x24(%esp),%esi \n"
+ "mov 0x28(%esp),%ebx \n"
+ "mov 0x2c(%esp),%edi \n"
+ "mov 0x30(%esp),%ecx \n"
+ "lea (%ebx,%ebx,2),%edx \n"
+ "pxor %xmm7,%xmm7 \n"
+
+"1:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa 0x10(%esi),%xmm1 \n"
+ "movdqa (%esi,%ebx,1),%xmm2 \n"
+ "movdqa 0x10(%esi,%ebx,1),%xmm3 \n"
+ "pavgb %xmm2,%xmm0 \n"
+ "pavgb %xmm3,%xmm1 \n"
+ "movdqa (%esi,%ebx,2),%xmm2 \n"
+ "movdqa 0x10(%esi,%ebx,2),%xmm3 \n"
+ "movdqa (%esi,%edx,1),%xmm4 \n"
+ "movdqa 0x10(%esi,%edx,1),%xmm5 \n"
+ "lea (%esi,%ebx,4),%ebp \n"
+ "lea 0x20(%esi),%esi \n"
+ "pavgb %xmm4,%xmm2 \n"
+ "pavgb %xmm5,%xmm3 \n"
+ "pavgb %xmm2,%xmm0 \n"
+ "pavgb %xmm3,%xmm1 \n"
+ "movdqa 0x0(%ebp),%xmm2 \n"
+ "movdqa 0x10(%ebp),%xmm3 \n"
+ "movdqa 0x0(%ebp,%ebx,1),%xmm4 \n"
+ "movdqa 0x10(%ebp,%ebx,1),%xmm5 \n"
+ "pavgb %xmm4,%xmm2 \n"
+ "pavgb %xmm5,%xmm3 \n"
+ "movdqa 0x0(%ebp,%ebx,2),%xmm4 \n"
+ "movdqa 0x10(%ebp,%ebx,2),%xmm5 \n"
+ "movdqa 0x0(%ebp,%edx,1),%xmm6 \n"
+ "pavgb %xmm6,%xmm4 \n"
+ "movdqa 0x10(%ebp,%edx,1),%xmm6 \n"
+ "pavgb %xmm6,%xmm5 \n"
+ "pavgb %xmm4,%xmm2 \n"
+ "pavgb %xmm5,%xmm3 \n"
+ "pavgb %xmm2,%xmm0 \n"
+ "pavgb %xmm3,%xmm1 \n"
+ "psadbw %xmm7,%xmm0 \n"
+ "psadbw %xmm7,%xmm1 \n"
+ "pshufd $0xd8,%xmm0,%xmm0 \n"
+ "pshufd $0x8d,%xmm1,%xmm1 \n"
+ "por %xmm1,%xmm0 \n"
+ "psrlw $0x3,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "movd %xmm0,(%edi) \n"
+ "lea 0x4(%edi),%edi \n"
+ "sub $0x4,%ecx \n"
+ "ja 1b \n"
+ "popa \n"
+ "ret \n"
+);
+
+// fpic is used for magiccam plugin
+#if !defined(__PIC__)
+#define HAS_SCALEROWDOWN34_SSSE3
+void ScaleRowDown34_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ asm(
+ DECLARE_FUNCTION(ScaleRowDown34_SSSE3)
+ "pusha \n"
+ "mov 0x24(%esp),%esi \n"
+ "mov 0x2c(%esp),%edi \n"
+ "mov 0x30(%esp),%ecx \n"
+ "movdqa _shuf0,%xmm3 \n"
+ "movdqa _shuf1,%xmm4 \n"
+ "movdqa _shuf2,%xmm5 \n"
+
+"1:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa 0x10(%esi),%xmm2 \n"
+ "lea 0x20(%esi),%esi \n"
+ "movdqa %xmm2,%xmm1 \n"
+ "palignr $0x8,%xmm0,%xmm1 \n"
+ "pshufb %xmm3,%xmm0 \n"
+ "pshufb %xmm4,%xmm1 \n"
+ "pshufb %xmm5,%xmm2 \n"
+ "movq %xmm0,(%edi) \n"
+ "movq %xmm1,0x8(%edi) \n"
+ "movq %xmm2,0x10(%edi) \n"
+ "lea 0x18(%edi),%edi \n"
+ "sub $0x18,%ecx \n"
+ "ja 1b \n"
+ "popa \n"
+ "ret \n"
+);
+
+void ScaleRowDown34_1_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ asm(
+ DECLARE_FUNCTION(ScaleRowDown34_1_Int_SSSE3)
+ "pusha \n"
+ "mov 0x24(%esp),%esi \n"
+ "mov 0x28(%esp),%ebp \n"
+ "mov 0x2c(%esp),%edi \n"
+ "mov 0x30(%esp),%ecx \n"
+ "movdqa _shuf01,%xmm2 \n"
+ "movdqa _shuf11,%xmm3 \n"
+ "movdqa _shuf21,%xmm4 \n"
+ "movdqa _madd01,%xmm5 \n"
+ "movdqa _madd11,%xmm6 \n"
+ "movdqa _round34,%xmm7 \n"
+
+"1:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa (%esi,%ebp),%xmm1 \n"
+ "pavgb %xmm1,%xmm0 \n"
+ "pshufb %xmm2,%xmm0 \n"
+ "pmaddubsw %xmm5,%xmm0 \n"
+ "paddsw %xmm7,%xmm0 \n"
+ "psrlw $0x2,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "movq %xmm0,(%edi) \n"
+ "movdqu 0x8(%esi),%xmm0 \n"
+ "movdqu 0x8(%esi,%ebp),%xmm1 \n"
+ "pavgb %xmm1,%xmm0 \n"
+ "pshufb %xmm3,%xmm0 \n"
+ "pmaddubsw %xmm6,%xmm0 \n"
+ "paddsw %xmm7,%xmm0 \n"
+ "psrlw $0x2,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "movq %xmm0,0x8(%edi) \n"
+ "movdqa 0x10(%esi),%xmm0 \n"
+ "movdqa 0x10(%esi,%ebp),%xmm1 \n"
+ "lea 0x20(%esi),%esi \n"
+ "pavgb %xmm1,%xmm0 \n"
+ "pshufb %xmm4,%xmm0 \n"
+ "movdqa _madd21,%xmm1 \n"
+ "pmaddubsw %xmm1,%xmm0 \n"
+ "paddsw %xmm7,%xmm0 \n"
+ "psrlw $0x2,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "movq %xmm0,0x10(%edi) \n"
+ "lea 0x18(%edi),%edi \n"
+ "sub $0x18,%ecx \n"
+ "ja 1b \n"
+
+ "popa \n"
+ "ret \n"
+);
+
+void ScaleRowDown34_0_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ asm(
+ DECLARE_FUNCTION(ScaleRowDown34_0_Int_SSSE3)
+ "pusha \n"
+ "mov 0x24(%esp),%esi \n"
+ "mov 0x28(%esp),%ebp \n"
+ "mov 0x2c(%esp),%edi \n"
+ "mov 0x30(%esp),%ecx \n"
+ "movdqa _shuf01,%xmm2 \n"
+ "movdqa _shuf11,%xmm3 \n"
+ "movdqa _shuf21,%xmm4 \n"
+ "movdqa _madd01,%xmm5 \n"
+ "movdqa _madd11,%xmm6 \n"
+ "movdqa _round34,%xmm7 \n"
+
+"1:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa (%esi,%ebp,1),%xmm1 \n"
+ "pavgb %xmm0,%xmm1 \n"
+ "pavgb %xmm1,%xmm0 \n"
+ "pshufb %xmm2,%xmm0 \n"
+ "pmaddubsw %xmm5,%xmm0 \n"
+ "paddsw %xmm7,%xmm0 \n"
+ "psrlw $0x2,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "movq %xmm0,(%edi) \n"
+ "movdqu 0x8(%esi),%xmm0 \n"
+ "movdqu 0x8(%esi,%ebp,1),%xmm1 \n"
+ "pavgb %xmm0,%xmm1 \n"
+ "pavgb %xmm1,%xmm0 \n"
+ "pshufb %xmm3,%xmm0 \n"
+ "pmaddubsw %xmm6,%xmm0 \n"
+ "paddsw %xmm7,%xmm0 \n"
+ "psrlw $0x2,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "movq %xmm0,0x8(%edi) \n"
+ "movdqa 0x10(%esi),%xmm0 \n"
+ "movdqa 0x10(%esi,%ebp,1),%xmm1 \n"
+ "lea 0x20(%esi),%esi \n"
+ "pavgb %xmm0,%xmm1 \n"
+ "pavgb %xmm1,%xmm0 \n"
+ "pshufb %xmm4,%xmm0 \n"
+ "movdqa _madd21,%xmm1 \n"
+ "pmaddubsw %xmm1,%xmm0 \n"
+ "paddsw %xmm7,%xmm0 \n"
+ "psrlw $0x2,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "movq %xmm0,0x10(%edi) \n"
+ "lea 0x18(%edi),%edi \n"
+ "sub $0x18,%ecx \n"
+ "ja 1b \n"
+ "popa \n"
+ "ret \n"
+);
+
+#define HAS_SCALEROWDOWN38_SSSE3
+void ScaleRowDown38_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ asm(
+ DECLARE_FUNCTION(ScaleRowDown38_SSSE3)
+ "pusha \n"
+ "mov 0x24(%esp),%esi \n"
+ "mov 0x28(%esp),%edx \n"
+ "mov 0x2c(%esp),%edi \n"
+ "mov 0x30(%esp),%ecx \n"
+ "movdqa _shuf38a ,%xmm4 \n"
+ "movdqa _shuf38b ,%xmm5 \n"
+
+"1:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa 0x10(%esi),%xmm1 \n"
+ "lea 0x20(%esi),%esi \n"
+ "pshufb %xmm4,%xmm0 \n"
+ "pshufb %xmm5,%xmm1 \n"
+ "paddusb %xmm1,%xmm0 \n"
+ "movq %xmm0,(%edi) \n"
+ "movhlps %xmm0,%xmm1 \n"
+ "movd %xmm1,0x8(%edi) \n"
+ "lea 0xc(%edi),%edi \n"
+ "sub $0xc,%ecx \n"
+ "ja 1b \n"
+ "popa \n"
+ "ret \n"
+);
+
+void ScaleRowDown38_3_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ asm(
+ DECLARE_FUNCTION(ScaleRowDown38_3_Int_SSSE3)
+ "pusha \n"
+ "mov 0x24(%esp),%esi \n"
+ "mov 0x28(%esp),%edx \n"
+ "mov 0x2c(%esp),%edi \n"
+ "mov 0x30(%esp),%ecx \n"
+ "movdqa _shufac0,%xmm4 \n"
+ "movdqa _shufac3,%xmm5 \n"
+ "movdqa _scaleac3,%xmm6 \n"
+ "pxor %xmm7,%xmm7 \n"
+
+"1:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa (%esi,%edx,1),%xmm2 \n"
+ "movhlps %xmm0,%xmm1 \n"
+ "movhlps %xmm2,%xmm3 \n"
+ "punpcklbw %xmm7,%xmm0 \n"
+ "punpcklbw %xmm7,%xmm1 \n"
+ "punpcklbw %xmm7,%xmm2 \n"
+ "punpcklbw %xmm7,%xmm3 \n"
+ "paddusw %xmm2,%xmm0 \n"
+ "paddusw %xmm3,%xmm1 \n"
+ "movdqa (%esi,%edx,2),%xmm2 \n"
+ "lea 0x10(%esi),%esi \n"
+ "movhlps %xmm2,%xmm3 \n"
+ "punpcklbw %xmm7,%xmm2 \n"
+ "punpcklbw %xmm7,%xmm3 \n"
+ "paddusw %xmm2,%xmm0 \n"
+ "paddusw %xmm3,%xmm1 \n"
+ "movdqa %xmm0,%xmm2 \n"
+ "psrldq $0x2,%xmm0 \n"
+ "paddusw %xmm0,%xmm2 \n"
+ "psrldq $0x2,%xmm0 \n"
+ "paddusw %xmm0,%xmm2 \n"
+ "pshufb %xmm4,%xmm2 \n"
+ "movdqa %xmm1,%xmm3 \n"
+ "psrldq $0x2,%xmm1 \n"
+ "paddusw %xmm1,%xmm3 \n"
+ "psrldq $0x2,%xmm1 \n"
+ "paddusw %xmm1,%xmm3 \n"
+ "pshufb %xmm5,%xmm3 \n"
+ "paddusw %xmm3,%xmm2 \n"
+ "pmulhuw %xmm6,%xmm2 \n"
+ "packuswb %xmm2,%xmm2 \n"
+ "movd %xmm2,(%edi) \n"
+ "pextrw $0x2,%xmm2,%eax \n"
+ "mov %ax,0x4(%edi) \n"
+ "lea 0x6(%edi),%edi \n"
+ "sub $0x6,%ecx \n"
+ "ja 1b \n"
+ "popa \n"
+ "ret \n"
+);
+
+void ScaleRowDown38_2_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ asm(
+ DECLARE_FUNCTION(ScaleRowDown38_2_Int_SSSE3)
+ "pusha \n"
+ "mov 0x24(%esp),%esi \n"
+ "mov 0x28(%esp),%edx \n"
+ "mov 0x2c(%esp),%edi \n"
+ "mov 0x30(%esp),%ecx \n"
+ "movdqa _shufab0,%xmm4 \n"
+ "movdqa _shufab1,%xmm5 \n"
+ "movdqa _shufab2,%xmm6 \n"
+ "movdqa _scaleab2,%xmm7 \n"
+
+"1:"
+ "movdqa (%esi),%xmm2 \n"
+ "pavgb (%esi,%edx,1),%xmm2 \n"
+ "lea 0x10(%esi),%esi \n"
+ "movdqa %xmm2,%xmm0 \n"
+ "pshufb %xmm4,%xmm0 \n"
+ "movdqa %xmm2,%xmm1 \n"
+ "pshufb %xmm5,%xmm1 \n"
+ "paddusw %xmm1,%xmm0 \n"
+ "pshufb %xmm6,%xmm2 \n"
+ "paddusw %xmm2,%xmm0 \n"
+ "pmulhuw %xmm7,%xmm0 \n"
+ "packuswb %xmm0,%xmm0 \n"
+ "movd %xmm0,(%edi) \n"
+ "pextrw $0x2,%xmm0,%eax \n"
+ "mov %ax,0x4(%edi) \n"
+ "lea 0x6(%edi),%edi \n"
+ "sub $0x6,%ecx \n"
+ "ja 1b \n"
+ "popa \n"
+ "ret \n"
+);
+#endif // __PIC__
+
+#define HAS_SCALEADDROWS_SSE2
+void ScaleAddRows_SSE2(const uint8* src_ptr, int src_stride,
+ uint16* dst_ptr, int src_width,
+ int src_height);
+ asm(
+ DECLARE_FUNCTION(ScaleAddRows_SSE2)
+ "pusha \n"
+ "mov 0x24(%esp),%esi \n"
+ "mov 0x28(%esp),%edx \n"
+ "mov 0x2c(%esp),%edi \n"
+ "mov 0x30(%esp),%ecx \n"
+ "mov 0x34(%esp),%ebx \n"
+ "pxor %xmm5,%xmm5 \n"
+
+"1:"
+ "movdqa (%esi),%xmm2 \n"
+ "lea (%esi,%edx,1),%eax \n"
+ "movhlps %xmm2,%xmm3 \n"
+ "lea -0x1(%ebx),%ebp \n"
+ "punpcklbw %xmm5,%xmm2 \n"
+ "punpcklbw %xmm5,%xmm3 \n"
+
+"2:"
+ "movdqa (%eax),%xmm0 \n"
+ "lea (%eax,%edx,1),%eax \n"
+ "movhlps %xmm0,%xmm1 \n"
+ "punpcklbw %xmm5,%xmm0 \n"
+ "punpcklbw %xmm5,%xmm1 \n"
+ "paddusw %xmm0,%xmm2 \n"
+ "paddusw %xmm1,%xmm3 \n"
+ "sub $0x1,%ebp \n"
+ "ja 2b \n"
+
+ "movdqa %xmm2,(%edi) \n"
+ "movdqa %xmm3,0x10(%edi) \n"
+ "lea 0x20(%edi),%edi \n"
+ "lea 0x10(%esi),%esi \n"
+ "sub $0x10,%ecx \n"
+ "ja 1b \n"
+ "popa \n"
+ "ret \n"
+);
+
+// Bilinear row filtering combines 16x2 -> 16x1. SSE2 version
+#define HAS_SCALEFILTERROWS_SSE2
+void ScaleFilterRows_SSE2(uint8* dst_ptr,
+ const uint8* src_ptr, int src_stride,
+ int dst_width, int source_y_fraction);
+ asm(
+ DECLARE_FUNCTION(ScaleFilterRows_SSE2)
+ "push %esi \n"
+ "push %edi \n"
+ "mov 0xc(%esp),%edi \n"
+ "mov 0x10(%esp),%esi \n"
+ "mov 0x14(%esp),%edx \n"
+ "mov 0x18(%esp),%ecx \n"
+ "mov 0x1c(%esp),%eax \n"
+ "cmp $0x0,%eax \n"
+ "je 2f \n"
+ "cmp $0x80,%eax \n"
+ "je 3f \n"
+ "movd %eax,%xmm6 \n"
+ "punpcklwd %xmm6,%xmm6 \n"
+ "pshufd $0x0,%xmm6,%xmm6 \n"
+ "neg %eax \n"
+ "add $0x100,%eax \n"
+ "movd %eax,%xmm5 \n"
+ "punpcklwd %xmm5,%xmm5 \n"
+ "pshufd $0x0,%xmm5,%xmm5 \n"
+ "pxor %xmm7,%xmm7 \n"
+
+"1:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa (%esi,%edx,1),%xmm2 \n"
+ "lea 0x10(%esi),%esi \n"
+ "movdqa %xmm0,%xmm1 \n"
+ "movdqa %xmm2,%xmm3 \n"
+ "punpcklbw %xmm7,%xmm0 \n"
+ "punpcklbw %xmm7,%xmm2 \n"
+ "punpckhbw %xmm7,%xmm1 \n"
+ "punpckhbw %xmm7,%xmm3 \n"
+ "pmullw %xmm5,%xmm0 \n"
+ "pmullw %xmm5,%xmm1 \n"
+ "pmullw %xmm6,%xmm2 \n"
+ "pmullw %xmm6,%xmm3 \n"
+ "paddusw %xmm2,%xmm0 \n"
+ "paddusw %xmm3,%xmm1 \n"
+ "psrlw $0x8,%xmm0 \n"
+ "psrlw $0x8,%xmm1 \n"
+ "packuswb %xmm1,%xmm0 \n"
+ "movdqa %xmm0,(%edi) \n"
+ "lea 0x10(%edi),%edi \n"
+ "sub $0x10,%ecx \n"
+ "ja 1b \n"
+ "mov -0x1(%edi),%al \n"
+ "mov %al,(%edi) \n"
+ "pop %edi \n"
+ "pop %esi \n"
+ "ret \n"
+
+"2:"
+ "movdqa (%esi),%xmm0 \n"
+ "lea 0x10(%esi),%esi \n"
+ "movdqa %xmm0,(%edi) \n"
+ "lea 0x10(%edi),%edi \n"
+ "sub $0x10,%ecx \n"
+ "ja 2b \n"
+
+ "mov -0x1(%edi),%al \n"
+ "mov %al,(%edi) \n"
+ "pop %edi \n"
+ "pop %esi \n"
+ "ret \n"
+
+"3:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa (%esi,%edx,1),%xmm2 \n"
+ "lea 0x10(%esi),%esi \n"
+ "pavgb %xmm2,%xmm0 \n"
+ "movdqa %xmm0,(%edi) \n"
+ "lea 0x10(%edi),%edi \n"
+ "sub $0x10,%ecx \n"
+ "ja 3b \n"
+
+ "mov -0x1(%edi),%al \n"
+ "mov %al,(%edi) \n"
+ "pop %edi \n"
+ "pop %esi \n"
+ "ret \n"
+);
+
+// Bilinear row filtering combines 16x2 -> 16x1. SSSE3 version
+#define HAS_SCALEFILTERROWS_SSSE3
+void ScaleFilterRows_SSSE3(uint8* dst_ptr,
+ const uint8* src_ptr, int src_stride,
+ int dst_width, int source_y_fraction);
+ asm(
+ DECLARE_FUNCTION(ScaleFilterRows_SSSE3)
+ "push %esi \n"
+ "push %edi \n"
+ "mov 0xc(%esp),%edi \n"
+ "mov 0x10(%esp),%esi \n"
+ "mov 0x14(%esp),%edx \n"
+ "mov 0x18(%esp),%ecx \n"
+ "mov 0x1c(%esp),%eax \n"
+ "cmp $0x0,%eax \n"
+ "je 2f \n"
+ "cmp $0x80,%eax \n"
+ "je 3f \n"
+ "shr %eax \n"
+ "mov %al,%ah \n"
+ "neg %al \n"
+ "add $0x80,%al \n"
+ "movd %eax,%xmm5 \n"
+ "punpcklwd %xmm5,%xmm5 \n"
+ "pshufd $0x0,%xmm5,%xmm5 \n"
+
+"1:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa (%esi,%edx,1),%xmm2 \n"
+ "lea 0x10(%esi),%esi \n"
+ "movdqa %xmm0,%xmm1 \n"
+ "punpcklbw %xmm2,%xmm0 \n"
+ "punpckhbw %xmm2,%xmm1 \n"
+ "pmaddubsw %xmm5,%xmm0 \n"
+ "pmaddubsw %xmm5,%xmm1 \n"
+ "psrlw $0x7,%xmm0 \n"
+ "psrlw $0x7,%xmm1 \n"
+ "packuswb %xmm1,%xmm0 \n"
+ "movdqa %xmm0,(%edi) \n"
+ "lea 0x10(%edi),%edi \n"
+ "sub $0x10,%ecx \n"
+ "ja 1b \n"
+ "mov -0x1(%edi),%al \n"
+ "mov %al,(%edi) \n"
+ "pop %edi \n"
+ "pop %esi \n"
+ "ret \n"
+
+"2:"
+ "movdqa (%esi),%xmm0 \n"
+ "lea 0x10(%esi),%esi \n"
+ "movdqa %xmm0,(%edi) \n"
+ "lea 0x10(%edi),%edi \n"
+ "sub $0x10,%ecx \n"
+ "ja 2b \n"
+ "mov -0x1(%edi),%al \n"
+ "mov %al,(%edi) \n"
+ "pop %edi \n"
+ "pop %esi \n"
+ "ret \n"
+
+"3:"
+ "movdqa (%esi),%xmm0 \n"
+ "movdqa (%esi,%edx,1),%xmm2 \n"
+ "lea 0x10(%esi),%esi \n"
+ "pavgb %xmm2,%xmm0 \n"
+ "movdqa %xmm0,(%edi) \n"
+ "lea 0x10(%edi),%edi \n"
+ "sub $0x10,%ecx \n"
+ "ja 3b \n"
+ "mov -0x1(%edi),%al \n"
+ "mov %al,(%edi) \n"
+ "pop %edi \n"
+ "pop %esi \n"
+ "ret \n"
+);
+
+#elif defined(__x86_64__)
+static void ScaleRowDown8Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "lea (%3,%3,2),%%r10 \n"
+ "pxor %%xmm7,%%xmm7 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa 0x10(%0),%%xmm1 \n"
+ "movdqa (%0,%3,1),%%xmm2 \n"
+ "movdqa 0x10(%0,%3,1),%%xmm3 \n"
+ "pavgb %%xmm2,%%xmm0 \n"
+ "pavgb %%xmm3,%%xmm1 \n"
+ "movdqa (%0,%3,2),%%xmm2 \n"
+ "movdqa 0x10(%0,%3,2),%%xmm3 \n"
+ "movdqa (%0,%%r10,1),%%xmm4 \n"
+ "movdqa 0x10(%0,%%r10,1),%%xmm5 \n"
+ "lea (%0,%3,4),%%r11 \n"
+ "lea 0x20(%0),%0 \n"
+ "pavgb %%xmm4,%%xmm2 \n"
+ "pavgb %%xmm5,%%xmm3 \n"
+ "pavgb %%xmm2,%%xmm0 \n"
+ "pavgb %%xmm3,%%xmm1 \n"
+ "movdqa 0x0(%%r11),%%xmm2 \n"
+ "movdqa 0x10(%%r11),%%xmm3 \n"
+ "movdqa 0x0(%%r11,%3,1),%%xmm4 \n"
+ "movdqa 0x10(%%r11,%3,1),%%xmm5 \n"
+ "pavgb %%xmm4,%%xmm2 \n"
+ "pavgb %%xmm5,%%xmm3 \n"
+ "movdqa 0x0(%%r11,%3,2),%%xmm4 \n"
+ "movdqa 0x10(%%r11,%3,2),%%xmm5 \n"
+ "movdqa 0x0(%%r11,%%r10,1),%%xmm6 \n"
+ "pavgb %%xmm6,%%xmm4 \n"
+ "movdqa 0x10(%%r11,%%r10,1),%%xmm6 \n"
+ "pavgb %%xmm6,%%xmm5 \n"
+ "pavgb %%xmm4,%%xmm2 \n"
+ "pavgb %%xmm5,%%xmm3 \n"
+ "pavgb %%xmm2,%%xmm0 \n"
+ "pavgb %%xmm3,%%xmm1 \n"
+ "psadbw %%xmm7,%%xmm0 \n"
+ "psadbw %%xmm7,%%xmm1 \n"
+ "pshufd $0xd8,%%xmm0,%%xmm0 \n"
+ "pshufd $0x8d,%%xmm1,%%xmm1 \n"
+ "por %%xmm1,%%xmm0 \n"
+ "psrlw $0x3,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movd %%xmm0,(%1) \n"
+ "lea 0x4(%1),%1 \n"
+ "sub $0x4,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"((intptr_t)(src_stride)) // %3
+ : "memory", "cc", "r10", "r11", "xmm6", "xmm7"
+);
+}
+
+#define HAS_SCALEROWDOWN34_SSSE3
+static void ScaleRowDown34_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "movdqa (%3),%%xmm3 \n"
+ "movdqa (%4),%%xmm4 \n"
+ "movdqa (%5),%%xmm5 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa 0x10(%0),%%xmm2 \n"
+ "lea 0x20(%0),%0 \n"
+ "movdqa %%xmm2,%%xmm1 \n"
+ "palignr $0x8,%%xmm0,%%xmm1 \n"
+ "pshufb %%xmm3,%%xmm0 \n"
+ "pshufb %%xmm4,%%xmm1 \n"
+ "pshufb %%xmm5,%%xmm2 \n"
+ "movq %%xmm0,(%1) \n"
+ "movq %%xmm1,0x8(%1) \n"
+ "movq %%xmm2,0x10(%1) \n"
+ "lea 0x18(%1),%1 \n"
+ "sub $0x18,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"(_shuf0), // %3
+ "r"(_shuf1), // %4
+ "r"(_shuf2) // %5
+ : "memory", "cc"
+);
+}
+
+static void ScaleRowDown34_1_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "movdqa (%4),%%xmm2 \n" // _shuf01
+ "movdqa (%5),%%xmm3 \n" // _shuf11
+ "movdqa (%6),%%xmm4 \n" // _shuf21
+ "movdqa (%7),%%xmm5 \n" // _madd01
+ "movdqa (%8),%%xmm6 \n" // _madd11
+ "movdqa (%9),%%xmm7 \n" // _round34
+ "movdqa (%10),%%xmm8 \n" // _madd21
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa (%0,%3),%%xmm1 \n"
+ "pavgb %%xmm1,%%xmm0 \n"
+ "pshufb %%xmm2,%%xmm0 \n"
+ "pmaddubsw %%xmm5,%%xmm0 \n"
+ "paddsw %%xmm7,%%xmm0 \n"
+ "psrlw $0x2,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movq %%xmm0,(%1) \n"
+ "movdqu 0x8(%0),%%xmm0 \n"
+ "movdqu 0x8(%0,%3),%%xmm1 \n"
+ "pavgb %%xmm1,%%xmm0 \n"
+ "pshufb %%xmm3,%%xmm0 \n"
+ "pmaddubsw %%xmm6,%%xmm0 \n"
+ "paddsw %%xmm7,%%xmm0 \n"
+ "psrlw $0x2,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movq %%xmm0,0x8(%1) \n"
+ "movdqa 0x10(%0),%%xmm0 \n"
+ "movdqa 0x10(%0,%3),%%xmm1 \n"
+ "lea 0x20(%0),%0 \n"
+ "pavgb %%xmm1,%%xmm0 \n"
+ "pshufb %%xmm4,%%xmm0 \n"
+ "pmaddubsw %%xmm8,%%xmm0 \n"
+ "paddsw %%xmm7,%%xmm0 \n"
+ "psrlw $0x2,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movq %%xmm0,0x10(%1) \n"
+ "lea 0x18(%1),%1 \n"
+ "sub $0x18,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"((intptr_t)(src_stride)), // %3
+ "r"(_shuf01), // %4
+ "r"(_shuf11), // %5
+ "r"(_shuf21), // %6
+ "r"(_madd01), // %7
+ "r"(_madd11), // %8
+ "r"(_round34), // %9
+ "r"(_madd21) // %10
+ : "memory", "cc", "xmm6", "xmm7", "xmm8"
+);
+}
+
+static void ScaleRowDown34_0_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "movdqa (%4),%%xmm2 \n" // _shuf01
+ "movdqa (%5),%%xmm3 \n" // _shuf11
+ "movdqa (%6),%%xmm4 \n" // _shuf21
+ "movdqa (%7),%%xmm5 \n" // _madd01
+ "movdqa (%8),%%xmm6 \n" // _madd11
+ "movdqa (%9),%%xmm7 \n" // _round34
+ "movdqa (%10),%%xmm8 \n" // _madd21
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa (%0,%3,1),%%xmm1 \n"
+ "pavgb %%xmm0,%%xmm1 \n"
+ "pavgb %%xmm1,%%xmm0 \n"
+ "pshufb %%xmm2,%%xmm0 \n"
+ "pmaddubsw %%xmm5,%%xmm0 \n"
+ "paddsw %%xmm7,%%xmm0 \n"
+ "psrlw $0x2,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movq %%xmm0,(%1) \n"
+ "movdqu 0x8(%0),%%xmm0 \n"
+ "movdqu 0x8(%0,%3,1),%%xmm1 \n"
+ "pavgb %%xmm0,%%xmm1 \n"
+ "pavgb %%xmm1,%%xmm0 \n"
+ "pshufb %%xmm3,%%xmm0 \n"
+ "pmaddubsw %%xmm6,%%xmm0 \n"
+ "paddsw %%xmm7,%%xmm0 \n"
+ "psrlw $0x2,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movq %%xmm0,0x8(%1) \n"
+ "movdqa 0x10(%0),%%xmm0 \n"
+ "movdqa 0x10(%0,%3,1),%%xmm1 \n"
+ "lea 0x20(%0),%0 \n"
+ "pavgb %%xmm0,%%xmm1 \n"
+ "pavgb %%xmm1,%%xmm0 \n"
+ "pshufb %%xmm4,%%xmm0 \n"
+ "pmaddubsw %%xmm8,%%xmm0 \n"
+ "paddsw %%xmm7,%%xmm0 \n"
+ "psrlw $0x2,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movq %%xmm0,0x10(%1) \n"
+ "lea 0x18(%1),%1 \n"
+ "sub $0x18,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"((intptr_t)(src_stride)), // %3
+ "r"(_shuf01), // %4
+ "r"(_shuf11), // %5
+ "r"(_shuf21), // %6
+ "r"(_madd01), // %7
+ "r"(_madd11), // %8
+ "r"(_round34), // %9
+ "r"(_madd21) // %10
+ : "memory", "cc", "xmm6", "xmm7", "xmm8"
+);
+}
+
+#define HAS_SCALEROWDOWN38_SSSE3
+static void ScaleRowDown38_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "movdqa (%3),%%xmm4 \n"
+ "movdqa (%4),%%xmm5 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa 0x10(%0),%%xmm1 \n"
+ "lea 0x20(%0),%0 \n"
+ "pshufb %%xmm4,%%xmm0 \n"
+ "pshufb %%xmm5,%%xmm1 \n"
+ "paddusb %%xmm1,%%xmm0 \n"
+ "movq %%xmm0,(%1) \n"
+ "movhlps %%xmm0,%%xmm1 \n"
+ "movd %%xmm1,0x8(%1) \n"
+ "lea 0xc(%1),%1 \n"
+ "sub $0xc,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"(_shuf38a), // %3
+ "r"(_shuf38b) // %4
+ : "memory", "cc"
+);
+}
+
+static void ScaleRowDown38_3_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "movdqa (%4),%%xmm4 \n"
+ "movdqa (%5),%%xmm5 \n"
+ "movdqa (%6),%%xmm6 \n"
+ "pxor %%xmm7,%%xmm7 \n"
+"1:"
+ "movdqa (%0),%%xmm0 \n"
+ "movdqa (%0,%3,1),%%xmm2 \n"
+ "movhlps %%xmm0,%%xmm1 \n"
+ "movhlps %%xmm2,%%xmm3 \n"
+ "punpcklbw %%xmm7,%%xmm0 \n"
+ "punpcklbw %%xmm7,%%xmm1 \n"
+ "punpcklbw %%xmm7,%%xmm2 \n"
+ "punpcklbw %%xmm7,%%xmm3 \n"
+ "paddusw %%xmm2,%%xmm0 \n"
+ "paddusw %%xmm3,%%xmm1 \n"
+ "movdqa (%0,%3,2),%%xmm2 \n"
+ "lea 0x10(%0),%0 \n"
+ "movhlps %%xmm2,%%xmm3 \n"
+ "punpcklbw %%xmm7,%%xmm2 \n"
+ "punpcklbw %%xmm7,%%xmm3 \n"
+ "paddusw %%xmm2,%%xmm0 \n"
+ "paddusw %%xmm3,%%xmm1 \n"
+ "movdqa %%xmm0,%%xmm2 \n"
+ "psrldq $0x2,%%xmm0 \n"
+ "paddusw %%xmm0,%%xmm2 \n"
+ "psrldq $0x2,%%xmm0 \n"
+ "paddusw %%xmm0,%%xmm2 \n"
+ "pshufb %%xmm4,%%xmm2 \n"
+ "movdqa %%xmm1,%%xmm3 \n"
+ "psrldq $0x2,%%xmm1 \n"
+ "paddusw %%xmm1,%%xmm3 \n"
+ "psrldq $0x2,%%xmm1 \n"
+ "paddusw %%xmm1,%%xmm3 \n"
+ "pshufb %%xmm5,%%xmm3 \n"
+ "paddusw %%xmm3,%%xmm2 \n"
+ "pmulhuw %%xmm6,%%xmm2 \n"
+ "packuswb %%xmm2,%%xmm2 \n"
+ "movd %%xmm2,(%1) \n"
+ "pextrw $0x2,%%xmm2,%%eax \n"
+ "mov %%ax,0x4(%1) \n"
+ "lea 0x6(%1),%1 \n"
+ "sub $0x6,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"((intptr_t)(src_stride)), // %3
+ "r"(_shufac0), // %4
+ "r"(_shufac3), // %5
+ "r"(_scaleac3) // %6
+ : "memory", "cc", "rax", "xmm6", "xmm7"
+);
+}
+
+static void ScaleRowDown38_2_Int_SSSE3(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ asm volatile (
+ "movdqa (%4),%%xmm4 \n"
+ "movdqa (%5),%%xmm5 \n"
+ "movdqa (%6),%%xmm6 \n"
+ "movdqa (%7),%%xmm7 \n"
+"1:"
+ "movdqa (%0),%%xmm2 \n"
+ "pavgb (%0,%3,1),%%xmm2 \n"
+ "lea 0x10(%0),%0 \n"
+ "movdqa %%xmm2,%%xmm0 \n"
+ "pshufb %%xmm4,%%xmm0 \n"
+ "movdqa %%xmm2,%%xmm1 \n"
+ "pshufb %%xmm5,%%xmm1 \n"
+ "paddusw %%xmm1,%%xmm0 \n"
+ "pshufb %%xmm6,%%xmm2 \n"
+ "paddusw %%xmm2,%%xmm0 \n"
+ "pmulhuw %%xmm7,%%xmm0 \n"
+ "packuswb %%xmm0,%%xmm0 \n"
+ "movd %%xmm0,(%1) \n"
+ "pextrw $0x2,%%xmm0,%%eax \n"
+ "mov %%ax,0x4(%1) \n"
+ "lea 0x6(%1),%1 \n"
+ "sub $0x6,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"((intptr_t)(src_stride)), // %3
+ "r"(_shufab0), // %4
+ "r"(_shufab1), // %5
+ "r"(_shufab2), // %6
+ "r"(_scaleab2) // %7
+ : "memory", "cc", "rax", "xmm6", "xmm7"
+);
+}
+
+#define HAS_SCALEADDROWS_SSE2
+static void ScaleAddRows_SSE2(const uint8* src_ptr, int src_stride,
+ uint16* dst_ptr, int src_width,
+ int src_height) {
+ asm volatile (
+ "pxor %%xmm5,%%xmm5 \n"
+"1:"
+ "movdqa (%0),%%xmm2 \n"
+ "lea (%0,%4,1),%%r10 \n"
+ "movhlps %%xmm2,%%xmm3 \n"
+ "lea -0x1(%3),%%r11 \n"
+ "punpcklbw %%xmm5,%%xmm2 \n"
+ "punpcklbw %%xmm5,%%xmm3 \n"
+
+"2:"
+ "movdqa (%%r10),%%xmm0 \n"
+ "lea (%%r10,%4,1),%%r10 \n"
+ "movhlps %%xmm0,%%xmm1 \n"
+ "punpcklbw %%xmm5,%%xmm0 \n"
+ "punpcklbw %%xmm5,%%xmm1 \n"
+ "paddusw %%xmm0,%%xmm2 \n"
+ "paddusw %%xmm1,%%xmm3 \n"
+ "sub $0x1,%%r11 \n"
+ "ja 2b \n"
+
+ "movdqa %%xmm2,(%1) \n"
+ "movdqa %%xmm3,0x10(%1) \n"
+ "lea 0x20(%1),%1 \n"
+ "lea 0x10(%0),%0 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ : "+r"(src_ptr), // %0
+ "+r"(dst_ptr), // %1
+ "+r"(src_width), // %2
+ "+r"(src_height) // %3
+ : "r"((intptr_t)(src_stride)) // %4
+ : "memory", "cc", "r10", "r11"
+);
+}
+
+// Bilinear row filtering combines 16x2 -> 16x1. SSE2 version
+#define HAS_SCALEFILTERROWS_SSE2
+static void ScaleFilterRows_SSE2(uint8* dst_ptr,
+ const uint8* src_ptr, int src_stride,
+ int dst_width, int source_y_fraction) {
+ if (source_y_fraction == 0) {
+ asm volatile (
+ "1:"
+ "movdqa (%1),%%xmm0 \n"
+ "lea 0x10(%1),%1 \n"
+ "movdqa %%xmm0,(%0) \n"
+ "lea 0x10(%0),%0 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ "mov -0x1(%0),%%al \n"
+ "mov %%al,(%0) \n"
+ : "+r"(dst_ptr), // %0
+ "+r"(src_ptr), // %1
+ "+r"(dst_width) // %2
+ :
+ : "memory", "cc", "rax"
+ );
+ return;
+ } else if (source_y_fraction == 128) {
+ asm volatile (
+ "1:"
+ "movdqa (%1),%%xmm0 \n"
+ "movdqa (%1,%3,1),%%xmm2 \n"
+ "lea 0x10(%1),%1 \n"
+ "pavgb %%xmm2,%%xmm0 \n"
+ "movdqa %%xmm0,(%0) \n"
+ "lea 0x10(%0),%0 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ "mov -0x1(%0),%%al \n"
+ "mov %%al,(%0) \n"
+ : "+r"(dst_ptr), // %0
+ "+r"(src_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"((intptr_t)(src_stride)) // %3
+ : "memory", "cc", "rax"
+ );
+ return;
+ } else {
+ asm volatile (
+ "mov %3,%%eax \n"
+ "movd %%eax,%%xmm6 \n"
+ "punpcklwd %%xmm6,%%xmm6 \n"
+ "pshufd $0x0,%%xmm6,%%xmm6 \n"
+ "neg %%eax \n"
+ "add $0x100,%%eax \n"
+ "movd %%eax,%%xmm5 \n"
+ "punpcklwd %%xmm5,%%xmm5 \n"
+ "pshufd $0x0,%%xmm5,%%xmm5 \n"
+ "pxor %%xmm7,%%xmm7 \n"
+ "1:"
+ "movdqa (%1),%%xmm0 \n"
+ "movdqa (%1,%4,1),%%xmm2 \n"
+ "lea 0x10(%1),%1 \n"
+ "movdqa %%xmm0,%%xmm1 \n"
+ "movdqa %%xmm2,%%xmm3 \n"
+ "punpcklbw %%xmm7,%%xmm0 \n"
+ "punpcklbw %%xmm7,%%xmm2 \n"
+ "punpckhbw %%xmm7,%%xmm1 \n"
+ "punpckhbw %%xmm7,%%xmm3 \n"
+ "pmullw %%xmm5,%%xmm0 \n"
+ "pmullw %%xmm5,%%xmm1 \n"
+ "pmullw %%xmm6,%%xmm2 \n"
+ "pmullw %%xmm6,%%xmm3 \n"
+ "paddusw %%xmm2,%%xmm0 \n"
+ "paddusw %%xmm3,%%xmm1 \n"
+ "psrlw $0x8,%%xmm0 \n"
+ "psrlw $0x8,%%xmm1 \n"
+ "packuswb %%xmm1,%%xmm0 \n"
+ "movdqa %%xmm0,(%0) \n"
+ "lea 0x10(%0),%0 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ "mov -0x1(%0),%%al \n"
+ "mov %%al,(%0) \n"
+ : "+r"(dst_ptr), // %0
+ "+r"(src_ptr), // %1
+ "+r"(dst_width), // %2
+ "+r"(source_y_fraction) // %3
+ : "r"((intptr_t)(src_stride)) // %4
+ : "memory", "cc", "rax", "xmm6", "xmm7"
+ );
+ }
+ return;
+}
+
+// Bilinear row filtering combines 16x2 -> 16x1. SSSE3 version
+#define HAS_SCALEFILTERROWS_SSSE3
+static void ScaleFilterRows_SSSE3(uint8* dst_ptr,
+ const uint8* src_ptr, int src_stride,
+ int dst_width, int source_y_fraction) {
+ if (source_y_fraction == 0) {
+ asm volatile (
+ "1:"
+ "movdqa (%1),%%xmm0 \n"
+ "lea 0x10(%1),%1 \n"
+ "movdqa %%xmm0,(%0) \n"
+ "lea 0x10(%0),%0 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ "mov -0x1(%0),%%al \n"
+ "mov %%al,(%0) \n"
+ : "+r"(dst_ptr), // %0
+ "+r"(src_ptr), // %1
+ "+r"(dst_width) // %2
+ :
+ : "memory", "cc", "rax"
+ );
+ return;
+ } else if (source_y_fraction == 128) {
+ asm volatile (
+ "1:"
+ "movdqa (%1),%%xmm0 \n"
+ "movdqa (%1,%3,1),%%xmm2 \n"
+ "lea 0x10(%1),%1 \n"
+ "pavgb %%xmm2,%%xmm0 \n"
+ "movdqa %%xmm0,(%0) \n"
+ "lea 0x10(%0),%0 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ "mov -0x1(%0),%%al \n"
+ "mov %%al,(%0) \n"
+ : "+r"(dst_ptr), // %0
+ "+r"(src_ptr), // %1
+ "+r"(dst_width) // %2
+ : "r"((intptr_t)(src_stride)) // %3
+ : "memory", "cc", "rax"
+ );
+ return;
+ } else {
+ asm volatile (
+ "mov %3,%%eax \n"
+ "shr %%eax \n"
+ "mov %%al,%%ah \n"
+ "neg %%al \n"
+ "add $0x80,%%al \n"
+ "movd %%eax,%%xmm5 \n"
+ "punpcklwd %%xmm5,%%xmm5 \n"
+ "pshufd $0x0,%%xmm5,%%xmm5 \n"
+ "1:"
+ "movdqa (%1),%%xmm0 \n"
+ "movdqa (%1,%4,1),%%xmm2 \n"
+ "lea 0x10(%1),%1 \n"
+ "movdqa %%xmm0,%%xmm1 \n"
+ "punpcklbw %%xmm2,%%xmm0 \n"
+ "punpckhbw %%xmm2,%%xmm1 \n"
+ "pmaddubsw %%xmm5,%%xmm0 \n"
+ "pmaddubsw %%xmm5,%%xmm1 \n"
+ "psrlw $0x7,%%xmm0 \n"
+ "psrlw $0x7,%%xmm1 \n"
+ "packuswb %%xmm1,%%xmm0 \n"
+ "movdqa %%xmm0,(%0) \n"
+ "lea 0x10(%0),%0 \n"
+ "sub $0x10,%2 \n"
+ "ja 1b \n"
+ "mov -0x1(%0),%%al \n"
+ "mov %%al,(%0) \n"
+ : "+r"(dst_ptr), // %0
+ "+r"(src_ptr), // %1
+ "+r"(dst_width), // %2
+ "+r"(source_y_fraction) // %3
+ : "r"((intptr_t)(src_stride)) // %4
+ : "memory", "cc", "rax"
+ );
+ }
+ return;
+}
+#endif
+#endif
+
+// CPU agnostic row functions
+static void ScaleRowDown2_C(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ int x;
+ for (x = 0; x < dst_width; ++x) {
+ *dst++ = *src_ptr;
+ src_ptr += 2;
+ }
+}
+
+static void ScaleRowDown2Int_C(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ int x;
+ for (x = 0; x < dst_width; ++x) {
+ *dst++ = (src_ptr[0] + src_ptr[1] +
+ src_ptr[src_stride] + src_ptr[src_stride + 1] + 2) >> 2;
+ src_ptr += 2;
+ }
+}
+
+static void ScaleRowDown4_C(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ int x;
+ for (x = 0; x < dst_width; ++x) {
+ *dst++ = *src_ptr;
+ src_ptr += 4;
+ }
+}
+
+static void ScaleRowDown4Int_C(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ int x;
+ for (x = 0; x < dst_width; ++x) {
+ *dst++ = (src_ptr[0] + src_ptr[1] + src_ptr[2] + src_ptr[3] +
+ src_ptr[src_stride + 0] + src_ptr[src_stride + 1] +
+ src_ptr[src_stride + 2] + src_ptr[src_stride + 3] +
+ src_ptr[src_stride * 2 + 0] + src_ptr[src_stride * 2 + 1] +
+ src_ptr[src_stride * 2 + 2] + src_ptr[src_stride * 2 + 3] +
+ src_ptr[src_stride * 3 + 0] + src_ptr[src_stride * 3 + 1] +
+ src_ptr[src_stride * 3 + 2] + src_ptr[src_stride * 3 + 3] +
+ 8) >> 4;
+ src_ptr += 4;
+ }
+}
+
+// 640 output pixels is enough to allow 5120 input pixels with 1/8 scale down.
+// Keeping the total buffer under 4096 bytes avoids a stackcheck, saving 4% cpu.
+// The following 2 lines cause error on Windows.
+//static const int kMaxOutputWidth = 640;
+//static const int kMaxRow12 = 1280; //kMaxOutputWidth * 2;
+#define kMaxOutputWidth 640
+#define kMaxRow12 1280
+
+static void ScaleRowDown8_C(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ int x;
+ for (x = 0; x < dst_width; ++x) {
+ *dst++ = *src_ptr;
+ src_ptr += 8;
+ }
+}
+
+// Note calling code checks width is less than max and if not
+// uses ScaleRowDown8_C instead.
+static void ScaleRowDown8Int_C(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ ALIGN16(uint8 src_row[kMaxRow12 * 2]);
+ assert(dst_width <= kMaxOutputWidth);
+ ScaleRowDown4Int_C(src_ptr, src_stride, src_row, dst_width * 2);
+ ScaleRowDown4Int_C(src_ptr + src_stride * 4, src_stride,
+ src_row + kMaxOutputWidth,
+ dst_width * 2);
+ ScaleRowDown2Int_C(src_row, kMaxOutputWidth, dst, dst_width);
+}
+
+static void ScaleRowDown34_C(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ uint8* dend;
+ assert((dst_width % 3 == 0) && (dst_width > 0));
+ dend = dst + dst_width;
+ do {
+ dst[0] = src_ptr[0];
+ dst[1] = src_ptr[1];
+ dst[2] = src_ptr[3];
+ dst += 3;
+ src_ptr += 4;
+ } while (dst < dend);
+}
+
+// Filter rows 0 and 1 together, 3 : 1
+static void ScaleRowDown34_0_Int_C(const uint8* src_ptr, int src_stride,
+ uint8* d, int dst_width) {
+ uint8* dend;
+ const uint8* s;
+ const uint8* t;
+ assert((dst_width % 3 == 0) && (dst_width > 0));
+ dend = d + dst_width;
+ s = src_ptr;
+ t = src_ptr + src_stride;
+ do {
+ uint8 a0 = (s[0] * 3 + s[1] * 1 + 2) >> 2;
+ uint8 a1 = (s[1] * 1 + s[2] * 1 + 1) >> 1;
+ uint8 a2 = (s[2] * 1 + s[3] * 3 + 2) >> 2;
+ uint8 b0 = (t[0] * 3 + t[1] * 1 + 2) >> 2;
+ uint8 b1 = (t[1] * 1 + t[2] * 1 + 1) >> 1;
+ uint8 b2 = (t[2] * 1 + t[3] * 3 + 2) >> 2;
+ d[0] = (a0 * 3 + b0 + 2) >> 2;
+ d[1] = (a1 * 3 + b1 + 2) >> 2;
+ d[2] = (a2 * 3 + b2 + 2) >> 2;
+ d += 3;
+ s += 4;
+ t += 4;
+ } while (d < dend);
+}
+
+// Filter rows 1 and 2 together, 1 : 1
+static void ScaleRowDown34_1_Int_C(const uint8* src_ptr, int src_stride,
+ uint8* d, int dst_width) {
+ uint8* dend;
+ const uint8* s;
+ const uint8* t;
+ assert((dst_width % 3 == 0) && (dst_width > 0));
+ dend = d + dst_width;
+ s = src_ptr;
+ t = src_ptr + src_stride;
+ do {
+ uint8 a0 = (s[0] * 3 + s[1] * 1 + 2) >> 2;
+ uint8 a1 = (s[1] * 1 + s[2] * 1 + 1) >> 1;
+ uint8 a2 = (s[2] * 1 + s[3] * 3 + 2) >> 2;
+ uint8 b0 = (t[0] * 3 + t[1] * 1 + 2) >> 2;
+ uint8 b1 = (t[1] * 1 + t[2] * 1 + 1) >> 1;
+ uint8 b2 = (t[2] * 1 + t[3] * 3 + 2) >> 2;
+ d[0] = (a0 + b0 + 1) >> 1;
+ d[1] = (a1 + b1 + 1) >> 1;
+ d[2] = (a2 + b2 + 1) >> 1;
+ d += 3;
+ s += 4;
+ t += 4;
+ } while (d < dend);
+}
+
+#if defined(HAS_SCALEFILTERROWS_SSE2)
+// Filter row to 3/4
+static void ScaleFilterCols34_C(uint8* dst_ptr, const uint8* src_ptr,
+ int dst_width) {
+ uint8* dend;
+ const uint8* s;
+ assert((dst_width % 3 == 0) && (dst_width > 0));
+ dend = dst_ptr + dst_width;
+ s = src_ptr;
+ do {
+ dst_ptr[0] = (s[0] * 3 + s[1] * 1 + 2) >> 2;
+ dst_ptr[1] = (s[1] * 1 + s[2] * 1 + 1) >> 1;
+ dst_ptr[2] = (s[2] * 1 + s[3] * 3 + 2) >> 2;
+ dst_ptr += 3;
+ s += 4;
+ } while (dst_ptr < dend);
+}
+#endif
+
+static void ScaleFilterCols_C(uint8* dst_ptr, const uint8* src_ptr,
+ int dst_width, int dx) {
+ int x = 0;
+ int j;
+ for (j = 0; j < dst_width; ++j) {
+ int xi = x >> 16;
+ int xf1 = x & 0xffff;
+ int xf0 = 65536 - xf1;
+
+ *dst_ptr++ = (src_ptr[xi] * xf0 + src_ptr[xi + 1] * xf1) >> 16;
+ x += dx;
+ }
+}
+
+//Not work on Windows
+//static const int kMaxInputWidth = 2560;
+#define kMaxInputWidth 2560
+#if defined(HAS_SCALEFILTERROWS_SSE2)
+#define HAS_SCALEROWDOWN34_SSE2
+// Filter rows 0 and 1 together, 3 : 1
+static void ScaleRowDown34_0_Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ ALIGN16(uint8 row[kMaxInputWidth]);
+ assert((dst_width % 3 == 0) && (dst_width > 0));
+ ScaleFilterRows_SSE2(row, src_ptr, src_stride, dst_width * 4 / 3, 256 / 4);
+ ScaleFilterCols34_C(dst_ptr, row, dst_width);
+}
+
+// Filter rows 1 and 2 together, 1 : 1
+static void ScaleRowDown34_1_Int_SSE2(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ ALIGN16(uint8 row[kMaxInputWidth]);
+ assert((dst_width % 3 == 0) && (dst_width > 0));
+ ScaleFilterRows_SSE2(row, src_ptr, src_stride, dst_width * 4 / 3, 256 / 2);
+ ScaleFilterCols34_C(dst_ptr, row, dst_width);
+}
+#endif
+
+static void ScaleRowDown38_C(const uint8* src_ptr, int src_stride,
+ uint8* dst, int dst_width) {
+ int x;
+ assert(dst_width % 3 == 0);
+ for (x = 0; x < dst_width; x += 3) {
+ dst[0] = src_ptr[0];
+ dst[1] = src_ptr[3];
+ dst[2] = src_ptr[6];
+ dst += 3;
+ src_ptr += 8;
+ }
+}
+
+// 8x3 -> 3x1
+static void ScaleRowDown38_3_Int_C(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ int i;
+ assert((dst_width % 3 == 0) && (dst_width > 0));
+ for (i = 0; i < dst_width; i+=3) {
+ dst_ptr[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] +
+ src_ptr[src_stride + 0] + src_ptr[src_stride + 1] +
+ src_ptr[src_stride + 2] + src_ptr[src_stride * 2 + 0] +
+ src_ptr[src_stride * 2 + 1] + src_ptr[src_stride * 2 + 2]) *
+ (65536 / 9) >> 16;
+ dst_ptr[1] = (src_ptr[3] + src_ptr[4] + src_ptr[5] +
+ src_ptr[src_stride + 3] + src_ptr[src_stride + 4] +
+ src_ptr[src_stride + 5] + src_ptr[src_stride * 2 + 3] +
+ src_ptr[src_stride * 2 + 4] + src_ptr[src_stride * 2 + 5]) *
+ (65536 / 9) >> 16;
+ dst_ptr[2] = (src_ptr[6] + src_ptr[7] +
+ src_ptr[src_stride + 6] + src_ptr[src_stride + 7] +
+ src_ptr[src_stride * 2 + 6] + src_ptr[src_stride * 2 + 7]) *
+ (65536 / 6) >> 16;
+ src_ptr += 8;
+ dst_ptr += 3;
+ }
+}
+
+// 8x2 -> 3x1
+static void ScaleRowDown38_2_Int_C(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width) {
+ int i;
+ assert((dst_width % 3 == 0) && (dst_width > 0));
+ for (i = 0; i < dst_width; i+=3) {
+ dst_ptr[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] +
+ src_ptr[src_stride + 0] + src_ptr[src_stride + 1] +
+ src_ptr[src_stride + 2]) * (65536 / 6) >> 16;
+ dst_ptr[1] = (src_ptr[3] + src_ptr[4] + src_ptr[5] +
+ src_ptr[src_stride + 3] + src_ptr[src_stride + 4] +
+ src_ptr[src_stride + 5]) * (65536 / 6) >> 16;
+ dst_ptr[2] = (src_ptr[6] + src_ptr[7] +
+ src_ptr[src_stride + 6] + src_ptr[src_stride + 7]) *
+ (65536 / 4) >> 16;
+ src_ptr += 8;
+ dst_ptr += 3;
+ }
+}
+
+// C version 8x2 -> 8x1
+static void ScaleFilterRows_C(uint8* dst_ptr,
+ const uint8* src_ptr, int src_stride,
+ int dst_width, int source_y_fraction) {
+ int y1_fraction;
+ int y0_fraction;
+ const uint8* src_ptr1;
+ uint8* end;
+ assert(dst_width > 0);
+ y1_fraction = source_y_fraction;
+ y0_fraction = 256 - y1_fraction;
+ src_ptr1 = src_ptr + src_stride;
+ end = dst_ptr + dst_width;
+ do {
+ dst_ptr[0] = (src_ptr[0] * y0_fraction + src_ptr1[0] * y1_fraction) >> 8;
+ dst_ptr[1] = (src_ptr[1] * y0_fraction + src_ptr1[1] * y1_fraction) >> 8;
+ dst_ptr[2] = (src_ptr[2] * y0_fraction + src_ptr1[2] * y1_fraction) >> 8;
+ dst_ptr[3] = (src_ptr[3] * y0_fraction + src_ptr1[3] * y1_fraction) >> 8;
+ dst_ptr[4] = (src_ptr[4] * y0_fraction + src_ptr1[4] * y1_fraction) >> 8;
+ dst_ptr[5] = (src_ptr[5] * y0_fraction + src_ptr1[5] * y1_fraction) >> 8;
+ dst_ptr[6] = (src_ptr[6] * y0_fraction + src_ptr1[6] * y1_fraction) >> 8;
+ dst_ptr[7] = (src_ptr[7] * y0_fraction + src_ptr1[7] * y1_fraction) >> 8;
+ src_ptr += 8;
+ src_ptr1 += 8;
+ dst_ptr += 8;
+ } while (dst_ptr < end);
+ dst_ptr[0] = dst_ptr[-1];
+}
+
+void ScaleAddRows_C(const uint8* src_ptr, int src_stride,
+ uint16* dst_ptr, int src_width, int src_height) {
+ int x,y;
+ assert(src_width > 0);
+ assert(src_height > 0);
+ for (x = 0; x < src_width; ++x) {
+ const uint8* s = src_ptr + x;
+ int sum = 0;
+ for (y = 0; y < src_height; ++y) {
+ sum += s[0];
+ s += src_stride;
+ }
+ dst_ptr[x] = sum;
+ }
+}
+
+/**
+ * Scale plane, 1/2
+ *
+ * This is an optimized version for scaling down a plane to 1/2 of
+ * its original size.
+ *
+ */
+static void ScalePlaneDown2(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr,
+ FilterMode filtering) {
+ void (*ScaleRowDown2)(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ assert(IS_ALIGNED(src_width, 2));
+ assert(IS_ALIGNED(src_height, 2));
+
+#if defined(HAS_SCALEROWDOWN2_NEON)
+ if (TestCpuFlag(kCpuHasNEON) &&
+ IS_ALIGNED(dst_width, 16)) {
+ ScaleRowDown2 = filtering ? ScaleRowDown2Int_NEON : ScaleRowDown2_NEON;
+ } else
+#endif
+#if defined(HAS_SCALEROWDOWN2_SSE2)
+ if (TestCpuFlag(kCpuHasSSE2) &&
+ IS_ALIGNED(dst_width, 16) &&
+ IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) &&
+ IS_ALIGNED(dst_ptr, 16) && IS_ALIGNED(dst_stride, 16)) {
+ ScaleRowDown2 = filtering ? ScaleRowDown2Int_SSE2 : ScaleRowDown2_SSE2;
+ } else
+#endif
+ {
+ ScaleRowDown2 = filtering ? ScaleRowDown2Int_C : ScaleRowDown2_C;
+ }
+
+ {
+ int y;
+ for (y = 0; y < dst_height; ++y) {
+ ScaleRowDown2(src_ptr, src_stride, dst_ptr, dst_width);
+ src_ptr += (src_stride << 1);
+ dst_ptr += dst_stride;
+ }
+ }
+}
+
+/**
+ * Scale plane, 1/4
+ *
+ * This is an optimized version for scaling down a plane to 1/4 of
+ * its original size.
+ */
+static void ScalePlaneDown4(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr,
+ FilterMode filtering) {
+ void (*ScaleRowDown4)(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ assert(IS_ALIGNED(src_width, 4));
+ assert(IS_ALIGNED(src_height, 4));
+
+#if defined(HAS_SCALEROWDOWN4_NEON)
+ if (TestCpuFlag(kCpuHasNEON) &&
+ IS_ALIGNED(dst_width, 4)) {
+ ScaleRowDown4 = filtering ? ScaleRowDown4Int_NEON : ScaleRowDown4_NEON;
+ } else
+#endif
+#if defined(HAS_SCALEROWDOWN4_SSE2)
+ if (TestCpuFlag(kCpuHasSSE2) &&
+ IS_ALIGNED(dst_width, 8) &&
+ IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) &&
+ IS_ALIGNED(dst_ptr, 8) && IS_ALIGNED(dst_stride, 8)) {
+ ScaleRowDown4 = filtering ? ScaleRowDown4Int_SSE2 : ScaleRowDown4_SSE2;
+ } else
+#endif
+ {
+ ScaleRowDown4 = filtering ? ScaleRowDown4Int_C : ScaleRowDown4_C;
+ }
+
+ {
+ int y;
+ for (y = 0; y < dst_height; ++y) {
+ ScaleRowDown4(src_ptr, src_stride, dst_ptr, dst_width);
+ src_ptr += (src_stride << 2);
+ dst_ptr += dst_stride;
+ }
+ }
+}
+
+/**
+ * Scale plane, 1/8
+ *
+ * This is an optimized version for scaling down a plane to 1/8
+ * of its original size.
+ *
+ */
+static void ScalePlaneDown8(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr,
+ FilterMode filtering) {
+ void (*ScaleRowDown8)(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ assert(IS_ALIGNED(src_width, 8));
+ assert(IS_ALIGNED(src_height, 8));
+
+#if defined(HAS_SCALEROWDOWN8_SSE2)
+ if (TestCpuFlag(kCpuHasSSE2) &&
+ IS_ALIGNED(dst_width, 4) &&
+ IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) &&
+ IS_ALIGNED(dst_ptr, 4) && IS_ALIGNED(dst_stride, 4)) {
+ ScaleRowDown8 = filtering ? ScaleRowDown8Int_SSE2 : ScaleRowDown8_SSE2;
+ } else
+#endif
+ {
+ ScaleRowDown8 = filtering && (dst_width <= kMaxOutputWidth) ?
+ ScaleRowDown8Int_C : ScaleRowDown8_C;
+ }
+
+ {
+ int y;
+ for (y = 0; y < dst_height; ++y) {
+ ScaleRowDown8(src_ptr, src_stride, dst_ptr, dst_width);
+ src_ptr += (src_stride << 3);
+ dst_ptr += dst_stride;
+ }
+ }
+}
+
+/**
+ * Scale plane down, 3/4
+ *
+ * Provided by Frank Barchard (fbarchard@google.com)
+ *
+ */
+static void ScalePlaneDown34(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr,
+ FilterMode filtering) {
+ void (*ScaleRowDown34_0)(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ void (*ScaleRowDown34_1)(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ assert(dst_width % 3 == 0);
+#if defined(HAS_SCALEROWDOWN34_NEON)
+ if (TestCpuFlag(kCpuHasNEON) &&
+ (dst_width % 24 == 0)) {
+ if (!filtering) {
+ ScaleRowDown34_0 = ScaleRowDown34_NEON;
+ ScaleRowDown34_1 = ScaleRowDown34_NEON;
+ } else {
+ ScaleRowDown34_0 = ScaleRowDown34_0_Int_NEON;
+ ScaleRowDown34_1 = ScaleRowDown34_1_Int_NEON;
+ }
+ } else
+#endif
+
+#if defined(HAS_SCALEROWDOWN34_SSSE3)
+ if (TestCpuFlag(kCpuHasSSSE3) &&
+ (dst_width % 24 == 0) &&
+ IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) &&
+ IS_ALIGNED(dst_ptr, 8) && IS_ALIGNED(dst_stride, 8)) {
+ if (!filtering) {
+ ScaleRowDown34_0 = ScaleRowDown34_SSSE3;
+ ScaleRowDown34_1 = ScaleRowDown34_SSSE3;
+ } else {
+ ScaleRowDown34_0 = ScaleRowDown34_0_Int_SSSE3;
+ ScaleRowDown34_1 = ScaleRowDown34_1_Int_SSSE3;
+ }
+ } else
+#endif
+#if defined(HAS_SCALEROWDOWN34_SSE2)
+ if (TestCpuFlag(kCpuHasSSE2) &&
+ (dst_width % 24 == 0) && IS_ALIGNED(src_stride, 16) &&
+ IS_ALIGNED(dst_stride, 8) &&
+ IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(dst_ptr, 8) &&
+ filtering) {
+ ScaleRowDown34_0 = ScaleRowDown34_0_Int_SSE2;
+ ScaleRowDown34_1 = ScaleRowDown34_1_Int_SSE2;
+ } else
+#endif
+ {
+ if (!filtering) {
+ ScaleRowDown34_0 = ScaleRowDown34_C;
+ ScaleRowDown34_1 = ScaleRowDown34_C;
+ } else {
+ ScaleRowDown34_0 = ScaleRowDown34_0_Int_C;
+ ScaleRowDown34_1 = ScaleRowDown34_1_Int_C;
+ }
+ }
+ {
+ int src_row = 0;
+ int y;
+ for (y = 0; y < dst_height; ++y) {
+ switch (src_row) {
+ case 0:
+ ScaleRowDown34_0(src_ptr, src_stride, dst_ptr, dst_width);
+ break;
+
+ case 1:
+ ScaleRowDown34_1(src_ptr, src_stride, dst_ptr, dst_width);
+ break;
+
+ case 2:
+ ScaleRowDown34_0(src_ptr + src_stride, -src_stride,
+ dst_ptr, dst_width);
+ break;
+ }
+ ++src_row;
+ src_ptr += src_stride;
+ dst_ptr += dst_stride;
+ if (src_row >= 3) {
+ src_ptr += src_stride;
+ src_row = 0;
+ }
+ }
+}
+}
+
+/**
+ * Scale plane, 3/8
+ *
+ * This is an optimized version for scaling down a plane to 3/8
+ * of its original size.
+ *
+ * Reduces 16x3 to 6x1
+ */
+static void ScalePlaneDown38(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr,
+ FilterMode filtering) {
+ void (*ScaleRowDown38_3)(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ void (*ScaleRowDown38_2)(const uint8* src_ptr, int src_stride,
+ uint8* dst_ptr, int dst_width);
+ assert(dst_width % 3 == 0);
+#if defined(HAS_SCALEROWDOWN38_NEON)
+ if (TestCpuFlag(kCpuHasNEON) &&
+ (dst_width % 12 == 0)) {
+ if (!filtering) {
+ ScaleRowDown38_3 = ScaleRowDown38_NEON;
+ ScaleRowDown38_2 = ScaleRowDown38_NEON;
+ } else {
+ ScaleRowDown38_3 = ScaleRowDown38_3_Int_NEON;
+ ScaleRowDown38_2 = ScaleRowDown38_2_Int_NEON;
+ }
+ } else
+#endif
+
+#if defined(HAS_SCALEROWDOWN38_SSSE3)
+ if (TestCpuFlag(kCpuHasSSSE3) &&
+ (dst_width % 24 == 0) && IS_ALIGNED(src_stride, 16) &&
+ IS_ALIGNED(dst_stride, 8) &&
+ IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(dst_ptr, 8)) {
+ if (!filtering) {
+ ScaleRowDown38_3 = ScaleRowDown38_SSSE3;
+ ScaleRowDown38_2 = ScaleRowDown38_SSSE3;
+ } else {
+ ScaleRowDown38_3 = ScaleRowDown38_3_Int_SSSE3;
+ ScaleRowDown38_2 = ScaleRowDown38_2_Int_SSSE3;
+ }
+ } else
+#endif
+ {
+ if (!filtering) {
+ ScaleRowDown38_3 = ScaleRowDown38_C;
+ ScaleRowDown38_2 = ScaleRowDown38_C;
+ } else {
+ ScaleRowDown38_3 = ScaleRowDown38_3_Int_C;
+ ScaleRowDown38_2 = ScaleRowDown38_2_Int_C;
+ }
+ }
+ {
+ int src_row = 0;
+ int y;
+ for (y = 0; y < dst_height; ++y) {
+ switch (src_row) {
+ case 0:
+ case 1:
+ ScaleRowDown38_3(src_ptr, src_stride, dst_ptr, dst_width);
+ src_ptr += src_stride * 3;
+ ++src_row;
+ break;
+
+ case 2:
+ ScaleRowDown38_2(src_ptr, src_stride, dst_ptr, dst_width);
+ src_ptr += src_stride * 2;
+ src_row = 0;
+ break;
+ }
+ dst_ptr += dst_stride;
+ }
+}
+}
+
+__inline static uint32 SumBox(int iboxwidth, int iboxheight,
+ int src_stride, const uint8* src_ptr) {
+ int x, y;
+ uint32 sum;
+ assert(iboxwidth > 0);
+ assert(iboxheight > 0);
+ sum = 0u;
+ for (y = 0; y < iboxheight; ++y) {
+ for (x = 0; x < iboxwidth; ++x) {
+ sum += src_ptr[x];
+ }
+ src_ptr += src_stride;
+ }
+ return sum;
+}
+
+static void ScalePlaneBoxRow(int dst_width, int boxheight,
+ int dx, int src_stride,
+ const uint8* src_ptr, uint8* dst_ptr) {
+ int x = 0;
+ int i;
+ for (i = 0; i < dst_width; ++i) {
+ int ix = x >> 16;
+ int boxwidth;
+ x += dx;
+ boxwidth = (x >> 16) - ix;
+ *dst_ptr++ = SumBox(boxwidth, boxheight, src_stride, src_ptr + ix) /
+ (boxwidth * boxheight);
+ }
+}
+
+__inline static uint32 SumPixels(int iboxwidth, const uint16* src_ptr) {
+ uint32 sum;
+ int x;
+ assert(iboxwidth > 0);
+ sum = 0u;
+ for (x = 0; x < iboxwidth; ++x) {
+ sum += src_ptr[x];
+ }
+ return sum;
+}
+
+static void ScaleAddCols2_C(int dst_width, int boxheight, int dx,
+ const uint16* src_ptr, uint8* dst_ptr) {
+ int scaletbl[2];
+ int minboxwidth = (dx >> 16);
+ scaletbl[0] = 65536 / (minboxwidth * boxheight);
+ scaletbl[1] = 65536 / ((minboxwidth + 1) * boxheight);
+ {
+ int *scaleptr = scaletbl - minboxwidth;
+ int x = 0;
+ int i;
+ for (i = 0; i < dst_width; ++i) {
+ int ix = x >> 16;
+ int boxwidth;
+ x += dx;
+ boxwidth = (x >> 16) - ix;
+ *dst_ptr++ = SumPixels(boxwidth, src_ptr + ix) * scaleptr[boxwidth] >> 16;
+ }
+ }
+}
+
+static void ScaleAddCols1_C(int dst_width, int boxheight, int dx,
+ const uint16* src_ptr, uint8* dst_ptr) {
+ int boxwidth = (dx >> 16);
+ int scaleval = 65536 / (boxwidth * boxheight);
+ int x = 0;
+ int i;
+ for (i = 0; i < dst_width; ++i) {
+ *dst_ptr++ = SumPixels(boxwidth, src_ptr + x) * scaleval >> 16;
+ x += boxwidth;
+ }
+}
+
+/**
+ * Scale plane down to any dimensions, with interpolation.
+ * (boxfilter).
+ *
+ * Same method as SimpleScale, which is fixed point, outputting
+ * one pixel of destination using fixed point (16.16) to step
+ * through source, sampling a box of pixel with simple
+ * averaging.
+ */
+static void ScalePlaneBox(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr) {
+ int dx, dy;
+ assert(dst_width > 0);
+ assert(dst_height > 0);
+ dy = (src_height << 16) / dst_height;
+ dx = (src_width << 16) / dst_width;
+ if (!IS_ALIGNED(src_width, 16) || (src_width > kMaxInputWidth) ||
+ dst_height * 2 > src_height) {
+ uint8* dst = dst_ptr;
+ int dy = (src_height << 16) / dst_height;
+ int dx = (src_width << 16) / dst_width;
+ int y = 0;
+ int j;
+ for (j = 0; j < dst_height; ++j) {
+ int iy = y >> 16;
+ const uint8* const src = src_ptr + iy * src_stride;
+ int boxheight;
+ y += dy;
+ if (y > (src_height << 16)) {
+ y = (src_height << 16);
+ }
+ boxheight = (y >> 16) - iy;
+ ScalePlaneBoxRow(dst_width, boxheight,
+ dx, src_stride,
+ src, dst);
+
+ dst += dst_stride;
+ }
+ } else {
+ ALIGN16(uint16 row[kMaxInputWidth]);
+ void (*ScaleAddRows)(const uint8* src_ptr, int src_stride,
+ uint16* dst_ptr, int src_width, int src_height);
+ void (*ScaleAddCols)(int dst_width, int boxheight, int dx,
+ const uint16* src_ptr, uint8* dst_ptr);
+#if defined(HAS_SCALEADDROWS_SSE2)
+ if (TestCpuFlag(kCpuHasSSE2) &&
+ IS_ALIGNED(src_stride, 16) && IS_ALIGNED(src_ptr, 16) &&
+ IS_ALIGNED(src_width, 16)) {
+ ScaleAddRows = ScaleAddRows_SSE2;
+ } else
+#endif
+ {
+ ScaleAddRows = ScaleAddRows_C;
+ }
+ if (dx & 0xffff) {
+ ScaleAddCols = ScaleAddCols2_C;
+ } else {
+ ScaleAddCols = ScaleAddCols1_C;
+ }
+
+ {
+ int y = 0;
+ int j;
+ for (j = 0; j < dst_height; ++j) {
+ int iy = y >> 16;
+ const uint8* const src = src_ptr + iy * src_stride;
+ int boxheight;
+ y += dy;
+ if (y > (src_height << 16)) {
+ y = (src_height << 16);
+ }
+ boxheight = (y >> 16) - iy;
+ ScaleAddRows(src, src_stride, row, src_width, boxheight);
+ ScaleAddCols(dst_width, boxheight, dx, row, dst_ptr);
+ dst_ptr += dst_stride;
+ }
+ }
+ }
+}
+
+/**
+ * Scale plane to/from any dimensions, with interpolation.
+ */
+static void ScalePlaneBilinearSimple(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr) {
+ int i, j;
+ uint8* dst = dst_ptr;
+ int dx = (src_width << 16) / dst_width;
+ int dy = (src_height << 16) / dst_height;
+ int maxx = ((src_width - 1) << 16) - 1;
+ int maxy = ((src_height - 1) << 16) - 1;
+ int y = (dst_height < src_height) ? 32768 :
+ (src_height << 16) / dst_height - 32768;
+ for (i = 0; i < dst_height; ++i) {
+ int cy = (y < 0) ? 0 : y;
+ int yi = cy >> 16;
+ int yf = cy & 0xffff;
+ const uint8* const src = src_ptr + yi * src_stride;
+ int x = (dst_width < src_width) ? 32768 :
+ (src_width << 16) / dst_width - 32768;
+ for (j = 0; j < dst_width; ++j) {
+ int cx = (x < 0) ? 0 : x;
+ int xi = cx >> 16;
+ int xf = cx & 0xffff;
+ int r0 = (src[xi] * (65536 - xf) + src[xi + 1] * xf) >> 16;
+ int r1 = (src[xi + src_stride] * (65536 - xf) +
+ src[xi + src_stride + 1] * xf) >> 16;
+ *dst++ = (r0 * (65536 - yf) + r1 * yf) >> 16;
+ x += dx;
+ if (x > maxx)
+ x = maxx;
+ }
+ dst += dst_stride - dst_width;
+ y += dy;
+ if (y > maxy)
+ y = maxy;
+ }
+}
+
+/**
+ * Scale plane to/from any dimensions, with bilinear
+ * interpolation.
+ */
+static void ScalePlaneBilinear(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr) {
+ int dy;
+ int dx;
+ assert(dst_width > 0);
+ assert(dst_height > 0);
+ dy = (src_height << 16) / dst_height;
+ dx = (src_width << 16) / dst_width;
+ if (!IS_ALIGNED(src_width, 8) || (src_width > kMaxInputWidth)) {
+ ScalePlaneBilinearSimple(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src_ptr, dst_ptr);
+
+ } else {
+ ALIGN16(uint8 row[kMaxInputWidth + 1]);
+ void (*ScaleFilterRows)(uint8* dst_ptr, const uint8* src_ptr,
+ int src_stride,
+ int dst_width, int source_y_fraction);
+ void (*ScaleFilterCols)(uint8* dst_ptr, const uint8* src_ptr,
+ int dst_width, int dx);
+#if defined(HAS_SCALEFILTERROWS_SSSE3)
+ if (TestCpuFlag(kCpuHasSSSE3) &&
+ IS_ALIGNED(src_stride, 16) && IS_ALIGNED(src_ptr, 16) &&
+ IS_ALIGNED(src_width, 16)) {
+ ScaleFilterRows = ScaleFilterRows_SSSE3;
+ } else
+#endif
+#if defined(HAS_SCALEFILTERROWS_SSE2)
+ if (TestCpuFlag(kCpuHasSSE2) &&
+ IS_ALIGNED(src_stride, 16) && IS_ALIGNED(src_ptr, 16) &&
+ IS_ALIGNED(src_width, 16)) {
+ ScaleFilterRows = ScaleFilterRows_SSE2;
+ } else
+#endif
+ {
+ ScaleFilterRows = ScaleFilterRows_C;
+ }
+ ScaleFilterCols = ScaleFilterCols_C;
+
+ {
+ int y = 0;
+ int maxy = ((src_height - 1) << 16) - 1; // max is filter of last 2 rows.
+ int j;
+ for (j = 0; j < dst_height; ++j) {
+ int iy = y >> 16;
+ int fy = (y >> 8) & 255;
+ const uint8* const src = src_ptr + iy * src_stride;
+ ScaleFilterRows(row, src, src_stride, src_width, fy);
+ ScaleFilterCols(dst_ptr, row, dst_width, dx);
+ dst_ptr += dst_stride;
+ y += dy;
+ if (y > maxy) {
+ y = maxy;
+ }
+ }
+ }
+}
+}
+
+/**
+ * Scale plane to/from any dimensions, without interpolation.
+ * Fixed point math is used for performance: The upper 16 bits
+ * of x and dx is the integer part of the source position and
+ * the lower 16 bits are the fixed decimal part.
+ */
+static void ScalePlaneSimple(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr) {
+ uint8* dst = dst_ptr;
+ int dx = (src_width << 16) / dst_width;
+ int y;
+ for (y = 0; y < dst_height; ++y) {
+ const uint8* const src = src_ptr + (y * src_height / dst_height) *
+ src_stride;
+ // TODO(fbarchard): Round X coordinate by setting x=0x8000.
+ int x = 0;
+ int i;
+ for (i = 0; i < dst_width; ++i) {
+ *dst++ = src[x >> 16];
+ x += dx;
+ }
+ dst += dst_stride - dst_width;
+ }
+}
+
+/**
+ * Scale plane to/from any dimensions.
+ */
+static void ScalePlaneAnySize(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr,
+ FilterMode filtering) {
+ if (!filtering) {
+ ScalePlaneSimple(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src_ptr, dst_ptr);
+ } else {
+ // fall back to non-optimized version
+ ScalePlaneBilinear(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src_ptr, dst_ptr);
+ }
+}
+
+/**
+ * Scale plane down, any size
+ *
+ * This is an optimized version for scaling down a plane to any size.
+ * The current implementation is ~10 times faster compared to the
+ * reference implementation for e.g. XGA->LowResPAL
+ *
+ */
+static void ScalePlaneDown(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr,
+ FilterMode filtering) {
+ if (!filtering) {
+ ScalePlaneSimple(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src_ptr, dst_ptr);
+ } else if (filtering == kFilterBilinear || src_height * 2 > dst_height) {
+ // between 1/2x and 1x use bilinear
+ ScalePlaneBilinear(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src_ptr, dst_ptr);
+ } else {
+ ScalePlaneBox(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src_ptr, dst_ptr);
+ }
+}
+
+/**
+ * Copy plane, no scaling
+ *
+ * This simply copies the given plane without scaling.
+ * The current implementation is ~115 times faster
+ * compared to the reference implementation.
+ *
+ */
+static void CopyPlane(int src_width, int src_height,
+ int dst_width, int dst_height,
+ int src_stride, int dst_stride,
+ const uint8* src_ptr, uint8* dst_ptr) {
+ if (src_stride == src_width && dst_stride == dst_width) {
+ // All contiguous, so can use REALLY fast path.
+ memcpy(dst_ptr, src_ptr, src_width * src_height);
+ } else {
+ // Not all contiguous; must copy scanlines individually
+ const uint8* src = src_ptr;
+ uint8* dst = dst_ptr;
+ int i;
+ for (i = 0; i < src_height; ++i) {
+ memcpy(dst, src, src_width);
+ dst += dst_stride;
+ src += src_stride;
+ }
+ }
+}
+
+static void ScalePlane(const uint8* src, int src_stride,
+ int src_width, int src_height,
+ uint8* dst, int dst_stride,
+ int dst_width, int dst_height,
+ FilterMode filtering, int use_ref) {
+ // Use specialized scales to improve performance for common resolutions.
+ // For example, all the 1/2 scalings will use ScalePlaneDown2()
+ if (dst_width == src_width && dst_height == src_height) {
+ // Straight copy.
+ CopyPlane(src_width, src_height, dst_width, dst_height, src_stride,
+ dst_stride, src, dst);
+ } else if (dst_width <= src_width && dst_height <= src_height) {
+ // Scale down.
+ if (use_ref) {
+ // For testing, allow the optimized versions to be disabled.
+ ScalePlaneDown(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src, dst, filtering);
+ } else if (4 * dst_width == 3 * src_width &&
+ 4 * dst_height == 3 * src_height) {
+ // optimized, 3/4
+ ScalePlaneDown34(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src, dst, filtering);
+ } else if (2 * dst_width == src_width && 2 * dst_height == src_height) {
+ // optimized, 1/2
+ ScalePlaneDown2(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src, dst, filtering);
+ // 3/8 rounded up for odd sized chroma height.
+ } else if (8 * dst_width == 3 * src_width &&
+ dst_height == ((src_height * 3 + 7) / 8)) {
+ // optimized, 3/8
+ ScalePlaneDown38(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src, dst, filtering);
+ } else if (4 * dst_width == src_width && 4 * dst_height == src_height) {
+ // optimized, 1/4
+ ScalePlaneDown4(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src, dst, filtering);
+ } else if (8 * dst_width == src_width && 8 * dst_height == src_height) {
+ // optimized, 1/8
+ ScalePlaneDown8(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src, dst, filtering);
+ } else {
+ // Arbitrary downsample
+ ScalePlaneDown(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src, dst, filtering);
+ }
+ } else {
+ // Arbitrary scale up and/or down.
+ ScalePlaneAnySize(src_width, src_height, dst_width, dst_height,
+ src_stride, dst_stride, src, dst, filtering);
+ }
+}
+
+/**
+ * Scale a plane.
+ *
+ * This function in turn calls a scaling function
+ * suitable for handling the desired resolutions.
+ *
+ */
+
+int I420Scale(const uint8* src_y, int src_stride_y,
+ const uint8* src_u, int src_stride_u,
+ const uint8* src_v, int src_stride_v,
+ int src_width, int src_height,
+ uint8* dst_y, int dst_stride_y,
+ uint8* dst_u, int dst_stride_u,
+ uint8* dst_v, int dst_stride_v,
+ int dst_width, int dst_height,
+ FilterMode filtering) {
+ if (!src_y || !src_u || !src_v || src_width <= 0 || src_height == 0 ||
+ !dst_y || !dst_u || !dst_v || dst_width <= 0 || dst_height <= 0) {
+ return -1;
+ }
+ // Negative height means invert the image.
+ if (src_height < 0) {
+ int halfheight;
+ src_height = -src_height;
+ halfheight = (src_height + 1) >> 1;
+ src_y = src_y + (src_height - 1) * src_stride_y;
+ src_u = src_u + (halfheight - 1) * src_stride_u;
+ src_v = src_v + (halfheight - 1) * src_stride_v;
+ src_stride_y = -src_stride_y;
+ src_stride_u = -src_stride_u;
+ src_stride_v = -src_stride_v;
+ }
+ {
+ int src_halfwidth = (src_width + 1) >> 1;
+ int src_halfheight = (src_height + 1) >> 1;
+ int dst_halfwidth = (dst_width + 1) >> 1;
+ int dst_halfheight = (dst_height + 1) >> 1;
+
+ ScalePlane(src_y, src_stride_y, src_width, src_height,
+ dst_y, dst_stride_y, dst_width, dst_height,
+ filtering, use_reference_impl_);
+ ScalePlane(src_u, src_stride_u, src_halfwidth, src_halfheight,
+ dst_u, dst_stride_u, dst_halfwidth, dst_halfheight,
+ filtering, use_reference_impl_);
+ ScalePlane(src_v, src_stride_v, src_halfwidth, src_halfheight,
+ dst_v, dst_stride_v, dst_halfwidth, dst_halfheight,
+ filtering, use_reference_impl_);
+ }
+ return 0;
+}
+
+// Deprecated api
+int Scale(const uint8* src_y, const uint8* src_u, const uint8* src_v,
+ int src_stride_y, int src_stride_u, int src_stride_v,
+ int src_width, int src_height,
+ uint8* dst_y, uint8* dst_u, uint8* dst_v,
+ int dst_stride_y, int dst_stride_u, int dst_stride_v,
+ int dst_width, int dst_height,
+ int interpolate) {
+ if (!src_y || !src_u || !src_v || src_width <= 0 || src_height == 0 ||
+ !dst_y || !dst_u || !dst_v || dst_width <= 0 || dst_height <= 0) {
+ return -1;
+ }
+ // Negative height means invert the image.
+ if (src_height < 0) {
+ int halfheight;
+ src_height = -src_height;
+ halfheight = (src_height + 1) >> 1;
+ src_y = src_y + (src_height - 1) * src_stride_y;
+ src_u = src_u + (halfheight - 1) * src_stride_u;
+ src_v = src_v + (halfheight - 1) * src_stride_v;
+ src_stride_y = -src_stride_y;
+ src_stride_u = -src_stride_u;
+ src_stride_v = -src_stride_v;
+ }
+ {
+ int src_halfwidth = (src_width + 1) >> 1;
+ int src_halfheight = (src_height + 1) >> 1;
+ int dst_halfwidth = (dst_width + 1) >> 1;
+ int dst_halfheight = (dst_height + 1) >> 1;
+ FilterMode filtering = interpolate ? kFilterBox : kFilterNone;
+
+ ScalePlane(src_y, src_stride_y, src_width, src_height,
+ dst_y, dst_stride_y, dst_width, dst_height,
+ filtering, use_reference_impl_);
+ ScalePlane(src_u, src_stride_u, src_halfwidth, src_halfheight,
+ dst_u, dst_stride_u, dst_halfwidth, dst_halfheight,
+ filtering, use_reference_impl_);
+ ScalePlane(src_v, src_stride_v, src_halfwidth, src_halfheight,
+ dst_v, dst_stride_v, dst_halfwidth, dst_halfheight,
+ filtering, use_reference_impl_);
+ }
+ return 0;
+}
+
+// Deprecated api
+int ScaleOffset(const uint8* src, int src_width, int src_height,
+ uint8* dst, int dst_width, int dst_height, int dst_yoffset,
+ int interpolate) {
+ if (!src || src_width <= 0 || src_height <= 0 ||
+ !dst || dst_width <= 0 || dst_height <= 0 || dst_yoffset < 0 ||
+ dst_yoffset >= dst_height) {
+ return -1;
+ }
+ dst_yoffset = dst_yoffset & ~1; // chroma requires offset to multiple of 2.
+ {
+ int src_halfwidth = (src_width + 1) >> 1;
+ int src_halfheight = (src_height + 1) >> 1;
+ int dst_halfwidth = (dst_width + 1) >> 1;
+ int dst_halfheight = (dst_height + 1) >> 1;
+ int aheight = dst_height - dst_yoffset * 2; // actual output height
+ const uint8* const src_y = src;
+ const uint8* const src_u = src + src_width * src_height;
+ const uint8* const src_v = src + src_width * src_height +
+ src_halfwidth * src_halfheight;
+ uint8* dst_y = dst + dst_yoffset * dst_width;
+ uint8* dst_u = dst + dst_width * dst_height +
+ (dst_yoffset >> 1) * dst_halfwidth;
+ uint8* dst_v = dst + dst_width * dst_height + dst_halfwidth * dst_halfheight +
+ (dst_yoffset >> 1) * dst_halfwidth;
+ return Scale(src_y, src_u, src_v, src_width, src_halfwidth, src_halfwidth,
+ src_width, src_height, dst_y, dst_u, dst_v, dst_width,
+ dst_halfwidth, dst_halfwidth, dst_width, aheight, interpolate);
+ }
+}
+
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
diff --git a/libvpx/tools/author_first_release.sh b/libvpx/tools/author_first_release.sh
new file mode 100755
index 0000000..7b0b797
--- /dev/null
+++ b/libvpx/tools/author_first_release.sh
@@ -0,0 +1,15 @@
+#!/bin/bash
+##
+## List the release each author first contributed to.
+##
+## Usage: author_first_release.sh [TAGS]
+##
+## If the TAGS arguments are unspecified, all tags reported by `git tag`
+## will be considered.
+##
+tags=${@:-$(git tag)}
+for tag in $tags; do
+ git shortlog -n -e -s $tag |
+ cut -f2- |
+ awk "{print \"${tag#v}\t\"\$0}"
+done | sort -k2 | uniq -f2
diff --git a/libvpx/tools/ftfy.sh b/libvpx/tools/ftfy.sh
new file mode 100755
index 0000000..c5cfdea
--- /dev/null
+++ b/libvpx/tools/ftfy.sh
@@ -0,0 +1,160 @@
+#!/bin/sh
+self="$0"
+dirname_self=$(dirname "$self")
+
+usage() {
+ cat <<EOF >&2
+Usage: $self [option]
+
+This script applies a whitespace transformation to the commit at HEAD. If no
+options are given, then the modified files are left in the working tree.
+
+Options:
+ -h, --help Shows this message
+ -n, --dry-run Shows a diff of the changes to be made.
+ --amend Squashes the changes into the commit at HEAD
+ This option will also reformat the commit message.
+ --commit Creates a new commit containing only the whitespace changes
+ --msg-only Reformat the commit message only, ignore the patch itself.
+
+EOF
+ rm -f ${CLEAN_FILES}
+ exit 1
+}
+
+
+log() {
+ echo "${self##*/}: $@" >&2
+}
+
+
+vpx_style() {
+ astyle --style=bsd --min-conditional-indent=0 --break-blocks \
+ --pad-oper --pad-header --unpad-paren \
+ --align-pointer=name \
+ --indent-preprocessor --convert-tabs --indent-labels \
+ --suffix=none --quiet "$@"
+ sed -i "" 's/[[:space:]]\{1,\},/,/g' "$@"
+}
+
+
+apply() {
+ [ $INTERSECT_RESULT -ne 0 ] && patch -p1 < "$1"
+}
+
+
+commit() {
+ LAST_CHANGEID=$(git show | awk '/Change-Id:/{print $2}')
+ if [ -z "$LAST_CHANGEID" ]; then
+ log "HEAD doesn't have a Change-Id, unable to generate a new commit"
+ exit 1
+ fi
+
+ # Build a deterministic Change-Id from the parent's
+ NEW_CHANGEID=${LAST_CHANGEID}-styled
+ NEW_CHANGEID=I$(echo $NEW_CHANGEID | git hash-object --stdin)
+
+ # Commit, preserving authorship from the parent commit.
+ git commit -a -C HEAD > /dev/null
+ git commit --amend -F- << EOF
+Cosmetic: Fix whitespace in change ${LAST_CHANGEID:0:9}
+
+Change-Id: ${NEW_CHANGEID}
+EOF
+}
+
+
+show_commit_msg_diff() {
+ if [ $DIFF_MSG_RESULT -ne 0 ]; then
+ log "Modified commit message:"
+ diff -u "$ORIG_COMMIT_MSG" "$NEW_COMMIT_MSG" | tail -n +3
+ fi
+}
+
+
+amend() {
+ show_commit_msg_diff
+ if [ $DIFF_MSG_RESULT -ne 0 ] || [ $INTERSECT_RESULT -ne 0 ]; then
+ git commit -a --amend -F "$NEW_COMMIT_MSG"
+ fi
+}
+
+
+diff_msg() {
+ git log -1 --format=%B > "$ORIG_COMMIT_MSG"
+ "${dirname_self}"/wrap-commit-msg.py \
+ < "$ORIG_COMMIT_MSG" > "$NEW_COMMIT_MSG"
+ cmp -s "$ORIG_COMMIT_MSG" "$NEW_COMMIT_MSG"
+ DIFF_MSG_RESULT=$?
+}
+
+
+# Temporary files
+ORIG_DIFF=orig.diff.$$
+MODIFIED_DIFF=modified.diff.$$
+FINAL_DIFF=final.diff.$$
+ORIG_COMMIT_MSG=orig.commit-msg.$$
+NEW_COMMIT_MSG=new.commit-msg.$$
+CLEAN_FILES="${ORIG_DIFF} ${MODIFIED_DIFF} ${FINAL_DIFF}"
+CLEAN_FILES="${CLEAN_FILES} ${ORIG_COMMIT_MSG} ${NEW_COMMIT_MSG}"
+
+# Preconditions
+[ $# -lt 2 ] || usage
+
+# Check that astyle supports pad-header and align-pointer=name
+if ! astyle --pad-header --align-pointer=name < /dev/null; then
+ log "Install astyle v1.24 or newer"
+ exit 1
+fi
+
+if ! git diff --quiet HEAD; then
+ log "Working tree is dirty, commit your changes first"
+ exit 1
+fi
+
+# Need to be in the root
+cd "$(git rev-parse --show-toplevel)"
+
+# Collect the original diff
+git show > "${ORIG_DIFF}"
+
+# Apply the style guide on new and modified files and collect its diff
+for f in $(git diff HEAD^ --name-only -M90 --diff-filter=AM \
+ | grep '\.[ch]$'); do
+ case "$f" in
+ third_party/*) continue;;
+ nestegg/*) continue;;
+ esac
+ vpx_style "$f"
+done
+git diff --no-color --no-ext-diff > "${MODIFIED_DIFF}"
+
+# Intersect the two diffs
+"${dirname_self}"/intersect-diffs.py \
+ "${ORIG_DIFF}" "${MODIFIED_DIFF}" > "${FINAL_DIFF}"
+INTERSECT_RESULT=$?
+git reset --hard >/dev/null
+
+# Fixup the commit message
+diff_msg
+
+# Handle options
+if [ -n "$1" ]; then
+ case "$1" in
+ -h|--help) usage;;
+ -n|--dry-run) cat "${FINAL_DIFF}"; show_commit_msg_diff;;
+ --commit) apply "${FINAL_DIFF}"; commit;;
+ --amend) apply "${FINAL_DIFF}"; amend;;
+ --msg-only) amend;;
+ *) usage;;
+ esac
+else
+ apply "${FINAL_DIFF}"
+ if ! git diff --quiet; then
+ log "Formatting changes applied, verify and commit."
+ log "See also: http://www.webmproject.org/code/contribute/conventions/"
+ git diff --stat
+ fi
+fi
+
+rm -f ${CLEAN_FILES}
diff --git a/tools/gen_authors.sh b/libvpx/tools/gen_authors.sh
similarity index 100%
rename from tools/gen_authors.sh
rename to libvpx/tools/gen_authors.sh
diff --git a/libvpx/tools/intersect-diffs.py b/libvpx/tools/intersect-diffs.py
new file mode 100755
index 0000000..be9dea5
--- /dev/null
+++ b/libvpx/tools/intersect-diffs.py
@@ -0,0 +1,188 @@
+#!/usr/bin/env python
+## Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+"""Calculates the "intersection" of two unified diffs.
+
+Given two diffs, A and B, it finds all hunks in B that had non-context lines
+in A and prints them to stdout. This is useful to determine the hunks in B that
+are relevant to A. The resulting file can be applied with patch(1) on top of A.
+"""
+
+__author__ = "jkoleszar@google.com"
+
+import re
+import sys
+
+
+class DiffLines(object):
+ """A container for one half of a diff."""
+
+ def __init__(self, filename, offset, length):
+ self.filename = filename
+ self.offset = offset
+ self.length = length
+ self.lines = []
+ self.delta_line_nums = []
+
+ def Append(self, line):
+ l = len(self.lines)
+ if line[0] != " ":
+ self.delta_line_nums.append(self.offset + l)
+ self.lines.append(line[1:])
+ assert l+1 <= self.length
+
+ def Complete(self):
+ return len(self.lines) == self.length
+
+ def __contains__(self, item):
+ return item >= self.offset and item <= self.offset + self.length - 1
+
+
+class DiffHunk(object):
+ """A container for one diff hunk, consisting of two DiffLines."""
+
+ def __init__(self, header, file_a, file_b, start_a, len_a, start_b, len_b):
+ self.header = header
+ self.left = DiffLines(file_a, start_a, len_a)
+ self.right = DiffLines(file_b, start_b, len_b)
+ self.lines = []
+
+ def Append(self, line):
+ """Adds a line to the DiffHunk and its DiffLines children."""
+ if line[0] == "-":
+ self.left.Append(line)
+ elif line[0] == "+":
+ self.right.Append(line)
+ elif line[0] == " ":
+ self.left.Append(line)
+ self.right.Append(line)
+ else:
+ assert False, ("Unrecognized character at start of diff line "
+ "%r" % line[0])
+ self.lines.append(line)
+
+ def Complete(self):
+ return self.left.Complete() and self.right.Complete()
+
+ def __repr__(self):
+ return "DiffHunk(%s, %s, len %d)" % (
+ self.left.filename, self.right.filename,
+ max(self.left.length, self.right.length))
+
+
+def ParseDiffHunks(stream):
+ """Walk a file-like object, yielding DiffHunks as they're parsed."""
+
+ file_regex = re.compile(r"(\+\+\+|---) (\S+)")
+ range_regex = re.compile(r"@@ -(\d+)(,(\d+))? \+(\d+)(,(\d+))?")
+ hunk = None
+ while True:
+ line = stream.readline()
+ if not line:
+ break
+
+ if hunk is None:
+ # Parse file names
+ diff_file = file_regex.match(line)
+ if diff_file:
+ if line.startswith("---"):
+ a_line = line
+ a = diff_file.group(2)
+ continue
+ if line.startswith("+++"):
+ b_line = line
+ b = diff_file.group(2)
+ continue
+
+ # Parse offset/lengths
+ diffrange = range_regex.match(line)
+ if diffrange:
+ if diffrange.group(2):
+ start_a = int(diffrange.group(1))
+ len_a = int(diffrange.group(3))
+ else:
+ start_a = 1
+ len_a = int(diffrange.group(1))
+
+ if diffrange.group(5):
+ start_b = int(diffrange.group(4))
+ len_b = int(diffrange.group(6))
+ else:
+ start_b = 1
+ len_b = int(diffrange.group(4))
+
+ header = [a_line, b_line, line]
+ hunk = DiffHunk(header, a, b, start_a, len_a, start_b, len_b)
+ else:
+ # Add the current line to the hunk
+ hunk.Append(line)
+
+ # See if the whole hunk has been parsed. If so, yield it and prepare
+ # for the next hunk.
+ if hunk.Complete():
+ yield hunk
+ hunk = None
+
+ # Partial hunks are a parse error
+ assert hunk is None
+
+
+def FormatDiffHunks(hunks):
+ """Re-serialize a list of DiffHunks."""
+ r = []
+ last_header = None
+ for hunk in hunks:
+ this_header = hunk.header[0:2]
+ if last_header != this_header:
+ r.extend(hunk.header)
+ last_header = this_header
+ else:
+ r.extend(hunk.header[2])
+ r.extend(hunk.lines)
+ r.append("\n")
+ return "".join(r)
+
+
+def ZipHunks(rhs_hunks, lhs_hunks):
+ """Join two hunk lists on filename."""
+ for rhs_hunk in rhs_hunks:
+ rhs_file = rhs_hunk.right.filename.split("/")[1:]
+
+ for lhs_hunk in lhs_hunks:
+ lhs_file = lhs_hunk.left.filename.split("/")[1:]
+ if lhs_file != rhs_file:
+ continue
+ yield (rhs_hunk, lhs_hunk)
+
+
+def main():
+ old_hunks = [x for x in ParseDiffHunks(open(sys.argv[1], "r"))]
+ new_hunks = [x for x in ParseDiffHunks(open(sys.argv[2], "r"))]
+ out_hunks = []
+
+ # Join the right hand side of the older diff with the left hand side of the
+ # newer diff.
+ for old_hunk, new_hunk in ZipHunks(old_hunks, new_hunks):
+ if new_hunk in out_hunks:
+ continue
+ old_lines = old_hunk.right
+ new_lines = new_hunk.left
+
+ # Determine if this hunk overlaps any non-context line from the other
+ for i in old_lines.delta_line_nums:
+ if i in new_lines:
+ out_hunks.append(new_hunk)
+ break
+
+ if out_hunks:
+ print FormatDiffHunks(out_hunks)
+ sys.exit(1)
+
+if __name__ == "__main__":
+ main()
diff --git a/libvpx/tools/wrap-commit-msg.py b/libvpx/tools/wrap-commit-msg.py
new file mode 100755
index 0000000..d5b4b04
--- /dev/null
+++ b/libvpx/tools/wrap-commit-msg.py
@@ -0,0 +1,70 @@
+#!/usr/bin/env python
+## Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+"""Wraps paragraphs of text, preserving manual formatting
+
+This is like fold(1), but has the special convention of not modifying lines
+that start with whitespace. This allows you to intersperse blocks with
+special formatting, like code blocks, with written prose. The prose will
+be wordwrapped, and the manual formatting will be preserved.
+
+ * This won't handle the case of a bulleted (or ordered) list specially, so
+ manual wrapping must be done.
+
+Occasionally it's useful to put something with explicit formatting that
+doesn't look at all like a block of text inline.
+
+ indicator = has_leading_whitespace(line);
+ if (indicator)
+ preserve_formatting(line);
+
+The intent is that this docstring would make it through the transform
+and still be legible and presented as it is in the source. If additional
+cases are handled, update this doc to describe the effect.
+"""
+
+__author__ = "jkoleszar@google.com"
+import textwrap
+import sys
+
+def wrap(text):
+ if text:
+ return textwrap.fill(text, break_long_words=False) + '\n'
+ return ""
+
+
+def main(fileobj):
+ text = ""
+ output = ""
+ while True:
+ line = fileobj.readline()
+ if not line:
+ break
+
+ if line.lstrip() == line:
+ text += line
+ else:
+ output += wrap(text)
+ text=""
+ output += line
+ output += wrap(text)
+
+ # Replace the file or write to stdout.
+ if fileobj == sys.stdin:
+ fileobj = sys.stdout
+ else:
+ fileobj.seek(0)
+ fileobj.truncate(0)
+ fileobj.write(output)
+
+if __name__ == "__main__":
+ if len(sys.argv) > 1:
+ main(open(sys.argv[1], "r+"))
+ else:
+ main(sys.stdin)
diff --git a/libvpx/tools_common.c b/libvpx/tools_common.c
new file mode 100644
index 0000000..6f95028
--- /dev/null
+++ b/libvpx/tools_common.c
@@ -0,0 +1,30 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include <stdio.h>
+#include "tools_common.h"
+#if defined(_WIN32) || defined(__OS2__)
+#include <io.h>
+#include <fcntl.h>
+
+#ifdef __OS2__
+#define _setmode setmode
+#define _fileno fileno
+#define _O_BINARY O_BINARY
+#endif
+#endif
+
+FILE* set_binary_mode(FILE *stream)
+{
+ (void)stream;
+#if defined(_WIN32) || defined(__OS2__)
+ _setmode(_fileno(stream), _O_BINARY);
+#endif
+ return stream;
+}
diff --git a/tools_common.h b/libvpx/tools_common.h
similarity index 100%
rename from tools_common.h
rename to libvpx/tools_common.h
diff --git a/libvpx/usage.dox b/libvpx/usage.dox
new file mode 100644
index 0000000..92fd6b2
--- /dev/null
+++ b/libvpx/usage.dox
@@ -0,0 +1,208 @@
+/*!\page usage Usage
+
+ The vpx multi-format codec SDK provides a unified interface amongst its
+ supported codecs. This abstraction allows applications using this SDK to
+ easily support multiple video formats with minimal code duplication or
+ "special casing." This section describes the interface common to all codecs.
+ For codec-specific details, see the \ref codecs page.
+
+ The following sections are common to all codecs:
+ - \ref usage_types
+ - \ref usage_features
+ - \ref usage_init
+ - \ref usage_errors
+
+ Fore more information on decoder and encoder specific usage, see the
+ following pages:
+ \if decoder
+ - \subpage usage_decode
+ \endif
+ \if decoder
+ - \subpage usage_encode
+ \endif
+
+ \section usage_types Important Data Types
+ There are two important data structures to consider in this interface.
+
+ \subsection usage_ctxs Contexts
+ A context is a storage area allocated by the calling application that the
+ codec may write into to store details about a single instance of that codec.
+ Most of the context is implementation specific, and thus opaque to the
+ application. The context structure as seen by the application is of fixed
+ size, and thus can be allocated with automatic storage or dynamically
+ on the heap.
+
+ Most operations require an initialized codec context. Codec context
+ instances are codec specific. That is, the codec to be used for the encoded
+ video must be known at initialization time. See #vpx_codec_ctx_t for further
+ information.
+
+ \subsection usage_ifaces Interfaces
+ A codec interface is an opaque structure that controls how function calls
+ into the generic interface are dispatched to their codec-specific
+ implementations. Applications \ref MUSTNOT attempt to examine or override
+ this storage, as it contains internal implementation details likely to
+ change from release to release.
+
+ Each supported codec will expose an interface structure to the application
+ as an <code>extern</code> reference to a structure of the incomplete type
+ #vpx_codec_iface_t.
+
+ \section usage_features Features
+ Several "features" are defined that are optionally implemented by codec
+ algorithms. Indeed, the same algorithm may support different features on
+ different platforms. The purpose of defining these features is that when
+ they are implemented, they conform to a common interface. The features, or
+ capabilities, of an algorithm can be queried from it's interface by using
+ the vpx_codec_get_caps() method. Attempts to invoke features not supported
+ by an algorithm will generally result in #VPX_CODEC_INCAPABLE.
+
+ Currently defined features available in both encoders and decoders include:
+ - \subpage usage_xma
+
+ \if decoder
+ Currently defined decoder features include:
+ - \ref usage_cb
+ - \ref usage_postproc
+ \endif
+
+ \section usage_init Initialization
+ To initialize a codec instance, the address of the codec context
+ and interface structures are passed to an initialization function. Depending
+ on the \ref usage_features that the codec supports, the codec could be
+ initialized in different modes. Most notably, the application may choose to
+ use \ref usage_xma mode to gain fine grained control over how and where
+ memory is allocated for the codec.
+
+ To prevent cases of confusion where the ABI of the library changes,
+ the ABI is versioned. The ABI version number must be passed at
+ initialization time to ensure the application is using a header file that
+ matches the library. The current ABI version number is stored in the
+ preprocessor macros #VPX_CODEC_ABI_VERSION, #VPX_ENCODER_ABI_VERSION, and
+ #VPX_DECODER_ABI_VERSION. For convenience, each initialization function has
+ a wrapper macro that inserts the correct version number. These macros are
+ named like the initialization methods, but without the _ver suffix.
+
+
+ The available initialization methods are:
+ \if encoder - #vpx_codec_enc_init (calls vpx_codec_enc_init_ver()) \endif
+ \if multi-encoder - #vpx_codec_enc_init_multi (calls vpx_codec_enc_init_multi_ver()) \endif
+ \if decoder - #vpx_codec_dec_init (calls vpx_codec_dec_init_ver()) \endif
+
+
+
+ \section usage_errors Error Handling
+ Almost all codec functions return an error status of type #vpx_codec_err_t.
+ The semantics of how each error condition should be processed is clearly
+ defined in the definitions of each enumerated value. Error values can be
+ converted into ASCII strings with the vpx_codec_error() and
+ vpx_codec_err_to_string() methods. The difference between these two methods is
+ that vpx_codec_error() returns the error state from an initialized context,
+ whereas vpx_codec_err_to_string() can be used in cases where an error occurs
+ outside any context. The enumerated value returned from the last call can be
+ retrieved from the <code>err</code> member of the decoder context as well.
+ Finally, more detailed error information may be able to be obtained by using
+ the vpx_codec_error_detail() method. Not all errors produce detailed error
+ information.
+
+ In addition to error information, the codec library's build configuration
+ is available at runtime on some platforms. This information can be returned
+ by calling vpx_codec_build_config(), and is formatted as a base64 coded string
+ (comprised of characters in the set [a-z_a-Z0-9+/]). This information is not
+ useful to an application at runtime, but may be of use to vpx for support.
+
+
+ \section usage_deadline Deadline
+ Both the encoding and decoding functions have a <code>deadline</code>
+ parameter. This parameter indicates the amount of time, in microseconds
+ (us), that the application wants the codec to spend processing before
+ returning. This is a soft deadline -- that is, the semantics of the
+ requested operation take precedence over meeting the deadline. If, for
+ example, an application sets a <code>deadline</code> of 1000us, and the
+ frame takes 2000us to decode, the call to vpx_codec_decode() will return
+ after 2000us. In this case the deadline is not met, but the semantics of the
+ function are preserved. If, for the same frame, an application instead sets
+ a <code>deadline</code> of 5000us, the decoder will see that it has 3000us
+ remaining in its time slice when decoding completes. It could then choose to
+ run a set of \ref usage_postproc filters, and perhaps would return after
+ 4000us (instead of the allocated 5000us). In this case the deadline is met,
+ and the semantics of the call are preserved, as before.
+
+ The special value <code>0</code> is reserved to represent an infinite
+ deadline. In this case, the codec will perform as much processing as
+ possible to yield the highest quality frame.
+
+ By convention, the value <code>1</code> is used to mean "return as fast as
+ possible."
+
+*/
+
+
+/*! \page usage_xma External Memory Allocation
+ Applications that wish to have fine grained control over how and where
+ decoders allocate memory \ref MAY make use of the eXternal Memory Allocation
+ (XMA) interface. Not all codecs support the XMA \ref usage_features.
+
+ To use a decoder in XMA mode, the decoder \ref MUST be initialized with the
+ vpx_codec_xma_init_ver() function. The amount of memory a decoder needs to
+ allocate is heavily dependent on the size of the encoded video frames. The
+ size of the video must be known before requesting the decoder's memory map.
+ This stream information can be obtained with the vpx_codec_peek_stream_info()
+ function, which does not require a constructed decoder context. If the exact
+ stream is not known, a stream info structure can be created that reflects
+ the maximum size that the decoder instance is required to support.
+
+ Once the decoder instance has been initialized and the stream information
+ determined, the application calls the vpx_codec_get_mem_map() iterator
+ repeatedly to get a list of the memory segments requested by the decoder.
+ The iterator value should be initialized to NULL to request the first
+ element, and the function will return #VPX_CODEC_LIST_END to signal the end of
+ the list.
+
+ After each segment is identified, it must be passed to the codec through the
+ vpx_codec_set_mem_map() function. Segments \ref MUST be passed in the same
+ order as they are returned from vpx_codec_get_mem_map(), but there is no
+ requirement that vpx_codec_get_mem_map() must finish iterating before
+ vpx_codec_set_mem_map() is called. For instance, some applications may choose
+ to get a list of all requests, construct an optimal heap, and then set all
+ maps at once with one call. Other applications may set one map at a time,
+ allocating it immediately after it is returned from vpx_codec_get_mem_map().
+
+ After all segments have been set using vpx_codec_set_mem_map(), the codec may
+ be used as it would be in normal internal allocation mode.
+
+ \section usage_xma_seg_id Segment Identifiers
+ Each requested segment is identified by an identifier unique to
+ that decoder type. Some of these identifiers are private, while others are
+ enumerated for application use. Identifiers not enumerated publicly are
+ subject to change. Identifiers are non-consecutive.
+
+ \section usage_xma_seg_szalign Segment Size and Alignment
+ The sz (size) and align (alignment) parameters describe the required size
+ and alignment of the requested segment. Alignment will always be a power of
+ two. Applications \ref MUST honor the alignment requested. Failure to do so
+ could result in program crashes or may incur a speed penalty.
+
+ \section usage_xma_seg_flags Segment Flags
+ The flags member of the segment structure indicates any requirements or
+ desires of the codec for the particular segment. The #VPX_CODEC_MEM_ZERO flag
+ indicates that the segment \ref MUST be zeroed by the application prior to
+ passing it to the application. The #VPX_CODEC_MEM_WRONLY flag indicates that
+ the segment will only be written into by the decoder, not read. If this flag
+ is not set, the application \ref MUST insure that the memory segment is
+ readable. On some platforms, framebuffer memory is writable but not
+ readable, for example. The #VPX_CODEC_MEM_FAST flag indicates that the segment
+ will be frequently accessed, and that it should be placed into fast memory,
+ if any is available. The application \ref MAY choose to place other segments
+ in fast memory as well, but the most critical segments will be identified by
+ this flag.
+
+ \section usage_xma_seg_basedtor Segment Base Address and Destructor
+ For each requested memory segment, the application must determine the
+ address of a memory segment that meets the requirements of the codec. This
+ address is set in the <code>base</code> member of the #vpx_codec_mmap
+ structure. If the application requires processing when the segment is no
+ longer used by the codec (for instance to deallocate it or close an
+ associated file descriptor) the <code>dtor</code> and <code>priv</code>
+ members can be set.
+*/
diff --git a/libvpx/usage_cx.dox b/libvpx/usage_cx.dox
new file mode 100644
index 0000000..62f3e45
--- /dev/null
+++ b/libvpx/usage_cx.dox
@@ -0,0 +1,13 @@
+/*! \page usage_encode Encode
+
+ The vpx_codec_encode() function is at the core of the encode loop. It
+ processes raw images passed by the application, producing packets of
+ compressed data. The <code>deadline</code> parameter controls the amount
+ of time in microseconds the encoder should spend working on the frame. For
+ more information on the <code>deadline</code> parameter, see
+ \ref usage_deadline.
+
+
+ \ref samples
+
+*/
diff --git a/usage_dx.dox b/libvpx/usage_dx.dox
similarity index 100%
rename from usage_dx.dox
rename to libvpx/usage_dx.dox
diff --git a/libvpx/vp8/common/alloccommon.c b/libvpx/vp8/common/alloccommon.c
new file mode 100644
index 0000000..8af9e90
--- /dev/null
+++ b/libvpx/vp8/common/alloccommon.c
@@ -0,0 +1,190 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "blockd.h"
+#include "vpx_mem/vpx_mem.h"
+#include "onyxc_int.h"
+#include "findnearmv.h"
+#include "entropymode.h"
+#include "systemdependent.h"
+
+void vp8_de_alloc_frame_buffers(VP8_COMMON *oci)
+{
+ int i;
+ for (i = 0; i < NUM_YV12_BUFFERS; i++)
+ vp8_yv12_de_alloc_frame_buffer(&oci->yv12_fb[i]);
+
+ vp8_yv12_de_alloc_frame_buffer(&oci->temp_scale_frame);
+#if CONFIG_POSTPROC
+ vp8_yv12_de_alloc_frame_buffer(&oci->post_proc_buffer);
+ if (oci->post_proc_buffer_int_used)
+ vp8_yv12_de_alloc_frame_buffer(&oci->post_proc_buffer_int);
+
+ vpx_free(oci->pp_limits_buffer);
+ oci->pp_limits_buffer = NULL;
+#endif
+
+ vpx_free(oci->above_context);
+ vpx_free(oci->mip);
+#if CONFIG_ERROR_CONCEALMENT
+ vpx_free(oci->prev_mip);
+ oci->prev_mip = NULL;
+#endif
+
+ oci->above_context = NULL;
+ oci->mip = NULL;
+}
+
+int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height)
+{
+ int i;
+
+ vp8_de_alloc_frame_buffers(oci);
+
+ /* our internal buffers are always multiples of 16 */
+ if ((width & 0xf) != 0)
+ width += 16 - (width & 0xf);
+
+ if ((height & 0xf) != 0)
+ height += 16 - (height & 0xf);
+
+
+ for (i = 0; i < NUM_YV12_BUFFERS; i++)
+ {
+ oci->fb_idx_ref_cnt[i] = 0;
+ oci->yv12_fb[i].flags = 0;
+ if (vp8_yv12_alloc_frame_buffer(&oci->yv12_fb[i], width, height, VP8BORDERINPIXELS) < 0)
+ goto allocation_fail;
+ }
+
+ oci->new_fb_idx = 0;
+ oci->lst_fb_idx = 1;
+ oci->gld_fb_idx = 2;
+ oci->alt_fb_idx = 3;
+
+ oci->fb_idx_ref_cnt[0] = 1;
+ oci->fb_idx_ref_cnt[1] = 1;
+ oci->fb_idx_ref_cnt[2] = 1;
+ oci->fb_idx_ref_cnt[3] = 1;
+
+ if (vp8_yv12_alloc_frame_buffer(&oci->temp_scale_frame, width, 16, VP8BORDERINPIXELS) < 0)
+ goto allocation_fail;
+
+ oci->mb_rows = height >> 4;
+ oci->mb_cols = width >> 4;
+ oci->MBs = oci->mb_rows * oci->mb_cols;
+ oci->mode_info_stride = oci->mb_cols + 1;
+ oci->mip = vpx_calloc((oci->mb_cols + 1) * (oci->mb_rows + 1), sizeof(MODE_INFO));
+
+ if (!oci->mip)
+ goto allocation_fail;
+
+ oci->mi = oci->mip + oci->mode_info_stride + 1;
+
+ /* Allocation of previous mode info will be done in vp8_decode_frame()
+ * as it is a decoder only data */
+
+ oci->above_context = vpx_calloc(sizeof(ENTROPY_CONTEXT_PLANES) * oci->mb_cols, 1);
+
+ if (!oci->above_context)
+ goto allocation_fail;
+
+#if CONFIG_POSTPROC
+ if (vp8_yv12_alloc_frame_buffer(&oci->post_proc_buffer, width, height, VP8BORDERINPIXELS) < 0)
+ goto allocation_fail;
+
+ oci->post_proc_buffer_int_used = 0;
+ vpx_memset(&oci->postproc_state, 0, sizeof(oci->postproc_state));
+ vpx_memset(oci->post_proc_buffer.buffer_alloc, 128,
+ oci->post_proc_buffer.frame_size);
+
+ /* Allocate buffer to store post-processing filter coefficients.
+ *
+ * Note: Round up mb_cols to support SIMD reads
+ */
+ oci->pp_limits_buffer = vpx_memalign(16, 24 * ((oci->mb_cols + 1) & ~1));
+ if (!oci->pp_limits_buffer)
+ goto allocation_fail;
+#endif
+
+ return 0;
+
+allocation_fail:
+ vp8_de_alloc_frame_buffers(oci);
+ return 1;
+}
+
+void vp8_setup_version(VP8_COMMON *cm)
+{
+ switch (cm->version)
+ {
+ case 0:
+ cm->no_lpf = 0;
+ cm->filter_type = NORMAL_LOOPFILTER;
+ cm->use_bilinear_mc_filter = 0;
+ cm->full_pixel = 0;
+ break;
+ case 1:
+ cm->no_lpf = 0;
+ cm->filter_type = SIMPLE_LOOPFILTER;
+ cm->use_bilinear_mc_filter = 1;
+ cm->full_pixel = 0;
+ break;
+ case 2:
+ cm->no_lpf = 1;
+ cm->filter_type = NORMAL_LOOPFILTER;
+ cm->use_bilinear_mc_filter = 1;
+ cm->full_pixel = 0;
+ break;
+ case 3:
+ cm->no_lpf = 1;
+ cm->filter_type = SIMPLE_LOOPFILTER;
+ cm->use_bilinear_mc_filter = 1;
+ cm->full_pixel = 1;
+ break;
+ default:
+ /*4,5,6,7 are reserved for future use*/
+ cm->no_lpf = 0;
+ cm->filter_type = NORMAL_LOOPFILTER;
+ cm->use_bilinear_mc_filter = 0;
+ cm->full_pixel = 0;
+ break;
+ }
+}
+void vp8_create_common(VP8_COMMON *oci)
+{
+ vp8_machine_specific_config(oci);
+
+ vp8_init_mbmode_probs(oci);
+ vp8_default_bmode_probs(oci->fc.bmode_prob);
+
+ oci->mb_no_coeff_skip = 1;
+ oci->no_lpf = 0;
+ oci->filter_type = NORMAL_LOOPFILTER;
+ oci->use_bilinear_mc_filter = 0;
+ oci->full_pixel = 0;
+ oci->multi_token_partition = ONE_PARTITION;
+ oci->clr_type = REG_YUV;
+ oci->clamp_type = RECON_CLAMP_REQUIRED;
+
+ /* Initialize reference frame sign bias structure to defaults */
+ vpx_memset(oci->ref_frame_sign_bias, 0, sizeof(oci->ref_frame_sign_bias));
+
+ /* Default disable buffer to buffer copying */
+ oci->copy_buffer_to_gf = 0;
+ oci->copy_buffer_to_arf = 0;
+}
+
+void vp8_remove_common(VP8_COMMON *oci)
+{
+ vp8_de_alloc_frame_buffers(oci);
+}
diff --git a/vp8/common/alloccommon.h b/libvpx/vp8/common/alloccommon.h
similarity index 100%
rename from vp8/common/alloccommon.h
rename to libvpx/vp8/common/alloccommon.h
diff --git a/libvpx/vp8/common/arm/armv6/bilinearfilter_v6.asm b/libvpx/vp8/common/arm/armv6/bilinearfilter_v6.asm
new file mode 100644
index 0000000..9704b42
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/bilinearfilter_v6.asm
@@ -0,0 +1,237 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_filter_block2d_bil_first_pass_armv6|
+ EXPORT |vp8_filter_block2d_bil_second_pass_armv6|
+
+ AREA |.text|, CODE, READONLY ; name this block of code
+
+;-------------------------------------
+; r0 unsigned char *src_ptr,
+; r1 unsigned short *dst_ptr,
+; r2 unsigned int src_pitch,
+; r3 unsigned int height,
+; stack unsigned int width,
+; stack const short *vp8_filter
+;-------------------------------------
+; The output is transposed stroed in output array to make it easy for second pass filtering.
+|vp8_filter_block2d_bil_first_pass_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ ldr r11, [sp, #40] ; vp8_filter address
+ ldr r4, [sp, #36] ; width
+
+ mov r12, r3 ; outer-loop counter
+
+ add r7, r2, r4 ; preload next row
+ pld [r0, r7]
+
+ sub r2, r2, r4 ; src increment for height loop
+
+ ldr r5, [r11] ; load up filter coefficients
+
+ mov r3, r3, lsl #1 ; height*2
+ add r3, r3, #2 ; plus 2 to make output buffer 4-bit aligned since height is actually (height+1)
+
+ mov r11, r1 ; save dst_ptr for each row
+
+ cmp r5, #128 ; if filter coef = 128, then skip the filter
+ beq bil_null_1st_filter
+
+|bil_height_loop_1st_v6|
+ ldrb r6, [r0] ; load source data
+ ldrb r7, [r0, #1]
+ ldrb r8, [r0, #2]
+ mov lr, r4, lsr #2 ; 4-in-parellel loop counter
+
+|bil_width_loop_1st_v6|
+ ldrb r9, [r0, #3]
+ ldrb r10, [r0, #4]
+
+ pkhbt r6, r6, r7, lsl #16 ; src[1] | src[0]
+ pkhbt r7, r7, r8, lsl #16 ; src[2] | src[1]
+
+ smuad r6, r6, r5 ; apply the filter
+ pkhbt r8, r8, r9, lsl #16 ; src[3] | src[2]
+ smuad r7, r7, r5
+ pkhbt r9, r9, r10, lsl #16 ; src[4] | src[3]
+
+ smuad r8, r8, r5
+ smuad r9, r9, r5
+
+ add r0, r0, #4
+ subs lr, lr, #1
+
+ add r6, r6, #0x40 ; round_shift_and_clamp
+ add r7, r7, #0x40
+ usat r6, #16, r6, asr #7
+ usat r7, #16, r7, asr #7
+
+ strh r6, [r1], r3 ; result is transposed and stored
+
+ add r8, r8, #0x40 ; round_shift_and_clamp
+ strh r7, [r1], r3
+ add r9, r9, #0x40
+ usat r8, #16, r8, asr #7
+ usat r9, #16, r9, asr #7
+
+ strh r8, [r1], r3 ; result is transposed and stored
+
+ ldrneb r6, [r0] ; load source data
+ strh r9, [r1], r3
+
+ ldrneb r7, [r0, #1]
+ ldrneb r8, [r0, #2]
+
+ bne bil_width_loop_1st_v6
+
+ add r0, r0, r2 ; move to next input row
+ subs r12, r12, #1
+
+ add r9, r2, r4, lsl #1 ; adding back block width
+ pld [r0, r9] ; preload next row
+
+ add r11, r11, #2 ; move over to next column
+ mov r1, r11
+
+ bne bil_height_loop_1st_v6
+
+ ldmia sp!, {r4 - r11, pc}
+
+|bil_null_1st_filter|
+|bil_height_loop_null_1st|
+ mov lr, r4, lsr #2 ; loop counter
+
+|bil_width_loop_null_1st|
+ ldrb r6, [r0] ; load data
+ ldrb r7, [r0, #1]
+ ldrb r8, [r0, #2]
+ ldrb r9, [r0, #3]
+
+ strh r6, [r1], r3 ; store it to immediate buffer
+ add r0, r0, #4
+ strh r7, [r1], r3
+ subs lr, lr, #1
+ strh r8, [r1], r3
+ strh r9, [r1], r3
+
+ bne bil_width_loop_null_1st
+
+ subs r12, r12, #1
+ add r0, r0, r2 ; move to next input line
+ add r11, r11, #2 ; move over to next column
+ mov r1, r11
+
+ bne bil_height_loop_null_1st
+
+ ldmia sp!, {r4 - r11, pc}
+
+ ENDP ; |vp8_filter_block2d_bil_first_pass_armv6|
+
+
+;---------------------------------
+; r0 unsigned short *src_ptr,
+; r1 unsigned char *dst_ptr,
+; r2 int dst_pitch,
+; r3 unsigned int height,
+; stack unsigned int width,
+; stack const short *vp8_filter
+;---------------------------------
+|vp8_filter_block2d_bil_second_pass_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ ldr r11, [sp, #40] ; vp8_filter address
+ ldr r4, [sp, #36] ; width
+
+ ldr r5, [r11] ; load up filter coefficients
+ mov r12, r4 ; outer-loop counter = width, since we work on transposed data matrix
+ mov r11, r1
+
+ cmp r5, #128 ; if filter coef = 128, then skip the filter
+ beq bil_null_2nd_filter
+
+|bil_height_loop_2nd|
+ ldr r6, [r0] ; load the data
+ ldr r8, [r0, #4]
+ ldrh r10, [r0, #8]
+ mov lr, r3, lsr #2 ; loop counter
+
+|bil_width_loop_2nd|
+ pkhtb r7, r6, r8 ; src[1] | src[2]
+ pkhtb r9, r8, r10 ; src[3] | src[4]
+
+ smuad r6, r6, r5 ; apply filter
+ smuad r8, r8, r5 ; apply filter
+
+ subs lr, lr, #1
+
+ smuadx r7, r7, r5 ; apply filter
+ smuadx r9, r9, r5 ; apply filter
+
+ add r0, r0, #8
+
+ add r6, r6, #0x40 ; round_shift_and_clamp
+ add r7, r7, #0x40
+ usat r6, #8, r6, asr #7
+ usat r7, #8, r7, asr #7
+ strb r6, [r1], r2 ; the result is transposed back and stored
+
+ add r8, r8, #0x40 ; round_shift_and_clamp
+ strb r7, [r1], r2
+ add r9, r9, #0x40
+ usat r8, #8, r8, asr #7
+ usat r9, #8, r9, asr #7
+ strb r8, [r1], r2 ; the result is transposed back and stored
+
+ ldrne r6, [r0] ; load data
+ strb r9, [r1], r2
+ ldrne r8, [r0, #4]
+ ldrneh r10, [r0, #8]
+
+ bne bil_width_loop_2nd
+
+ subs r12, r12, #1
+ add r0, r0, #4 ; update src for next row
+ add r11, r11, #1
+ mov r1, r11
+
+ bne bil_height_loop_2nd
+ ldmia sp!, {r4 - r11, pc}
+
+|bil_null_2nd_filter|
+|bil_height_loop_null_2nd|
+ mov lr, r3, lsr #2
+
+|bil_width_loop_null_2nd|
+ ldr r6, [r0], #4 ; load data
+ subs lr, lr, #1
+ ldr r8, [r0], #4
+
+ strb r6, [r1], r2 ; store data
+ mov r7, r6, lsr #16
+ strb r7, [r1], r2
+ mov r9, r8, lsr #16
+ strb r8, [r1], r2
+ strb r9, [r1], r2
+
+ bne bil_width_loop_null_2nd
+
+ subs r12, r12, #1
+ add r0, r0, #4
+ add r11, r11, #1
+ mov r1, r11
+
+ bne bil_height_loop_null_2nd
+
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_filter_block2d_second_pass_armv6|
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/copymem16x16_v6.asm b/libvpx/vp8/common/arm/armv6/copymem16x16_v6.asm
new file mode 100644
index 0000000..abf048c
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/copymem16x16_v6.asm
@@ -0,0 +1,186 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_copy_mem16x16_v6|
+ ; ARM
+ ; REQUIRE8
+ ; PRESERVE8
+
+ AREA Block, CODE, READONLY ; name this block of code
+;void copy_mem16x16_v6( unsigned char *src, int src_stride, unsigned char *dst, int dst_stride)
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+|vp8_copy_mem16x16_v6| PROC
+ stmdb sp!, {r4 - r7}
+ ;push {r4-r7}
+
+ ;preload
+ pld [r0, #31] ; preload for next 16x16 block
+
+ ands r4, r0, #15
+ beq copy_mem16x16_fast
+
+ ands r4, r0, #7
+ beq copy_mem16x16_8
+
+ ands r4, r0, #3
+ beq copy_mem16x16_4
+
+ ;copy one byte each time
+ ldrb r4, [r0]
+ ldrb r5, [r0, #1]
+ ldrb r6, [r0, #2]
+ ldrb r7, [r0, #3]
+
+ mov r12, #16
+
+copy_mem16x16_1_loop
+ strb r4, [r2]
+ strb r5, [r2, #1]
+ strb r6, [r2, #2]
+ strb r7, [r2, #3]
+
+ ldrb r4, [r0, #4]
+ ldrb r5, [r0, #5]
+ ldrb r6, [r0, #6]
+ ldrb r7, [r0, #7]
+
+ subs r12, r12, #1
+
+ strb r4, [r2, #4]
+ strb r5, [r2, #5]
+ strb r6, [r2, #6]
+ strb r7, [r2, #7]
+
+ ldrb r4, [r0, #8]
+ ldrb r5, [r0, #9]
+ ldrb r6, [r0, #10]
+ ldrb r7, [r0, #11]
+
+ strb r4, [r2, #8]
+ strb r5, [r2, #9]
+ strb r6, [r2, #10]
+ strb r7, [r2, #11]
+
+ ldrb r4, [r0, #12]
+ ldrb r5, [r0, #13]
+ ldrb r6, [r0, #14]
+ ldrb r7, [r0, #15]
+
+ add r0, r0, r1
+
+ strb r4, [r2, #12]
+ strb r5, [r2, #13]
+ strb r6, [r2, #14]
+ strb r7, [r2, #15]
+
+ add r2, r2, r3
+
+ ldrneb r4, [r0]
+ ldrneb r5, [r0, #1]
+ ldrneb r6, [r0, #2]
+ ldrneb r7, [r0, #3]
+
+ pld [r0, #31] ; preload for next 16x16 block
+
+ bne copy_mem16x16_1_loop
+
+ ldmia sp!, {r4 - r7}
+ ;pop {r4-r7}
+ mov pc, lr
+
+;copy 4 bytes each time
+copy_mem16x16_4
+ ldr r4, [r0]
+ ldr r5, [r0, #4]
+ ldr r6, [r0, #8]
+ ldr r7, [r0, #12]
+
+ mov r12, #16
+
+copy_mem16x16_4_loop
+ subs r12, r12, #1
+ add r0, r0, r1
+
+ str r4, [r2]
+ str r5, [r2, #4]
+ str r6, [r2, #8]
+ str r7, [r2, #12]
+
+ add r2, r2, r3
+
+ ldrne r4, [r0]
+ ldrne r5, [r0, #4]
+ ldrne r6, [r0, #8]
+ ldrne r7, [r0, #12]
+
+ pld [r0, #31] ; preload for next 16x16 block
+
+ bne copy_mem16x16_4_loop
+
+ ldmia sp!, {r4 - r7}
+ ;pop {r4-r7}
+ mov pc, lr
+
+;copy 8 bytes each time
+copy_mem16x16_8
+ sub r1, r1, #16
+ sub r3, r3, #16
+
+ mov r12, #16
+
+copy_mem16x16_8_loop
+ ldmia r0!, {r4-r5}
+ ;ldm r0, {r4-r5}
+ ldmia r0!, {r6-r7}
+
+ add r0, r0, r1
+
+ stmia r2!, {r4-r5}
+ subs r12, r12, #1
+ ;stm r2, {r4-r5}
+ stmia r2!, {r6-r7}
+
+ add r2, r2, r3
+
+ pld [r0, #31] ; preload for next 16x16 block
+ bne copy_mem16x16_8_loop
+
+ ldmia sp!, {r4 - r7}
+ ;pop {r4-r7}
+ mov pc, lr
+
+;copy 16 bytes each time
+copy_mem16x16_fast
+ ;sub r1, r1, #16
+ ;sub r3, r3, #16
+
+ mov r12, #16
+
+copy_mem16x16_fast_loop
+ ldmia r0, {r4-r7}
+ ;ldm r0, {r4-r7}
+ add r0, r0, r1
+
+ subs r12, r12, #1
+ stmia r2, {r4-r7}
+ ;stm r2, {r4-r7}
+ add r2, r2, r3
+
+ pld [r0, #31] ; preload for next 16x16 block
+ bne copy_mem16x16_fast_loop
+
+ ldmia sp!, {r4 - r7}
+ ;pop {r4-r7}
+ mov pc, lr
+
+ ENDP ; |vp8_copy_mem16x16_v6|
+
+ END
diff --git a/vp8/common/arm/armv6/copymem8x4_v6.asm b/libvpx/vp8/common/arm/armv6/copymem8x4_v6.asm
similarity index 100%
rename from vp8/common/arm/armv6/copymem8x4_v6.asm
rename to libvpx/vp8/common/arm/armv6/copymem8x4_v6.asm
diff --git a/vp8/common/arm/armv6/copymem8x8_v6.asm b/libvpx/vp8/common/arm/armv6/copymem8x8_v6.asm
similarity index 100%
rename from vp8/common/arm/armv6/copymem8x8_v6.asm
rename to libvpx/vp8/common/arm/armv6/copymem8x8_v6.asm
diff --git a/libvpx/vp8/common/arm/armv6/dc_only_idct_add_v6.asm b/libvpx/vp8/common/arm/armv6/dc_only_idct_add_v6.asm
new file mode 100644
index 0000000..9aa659f
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/dc_only_idct_add_v6.asm
@@ -0,0 +1,70 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license and patent
+; grant that can be found in the LICENSE file in the root of the source
+; tree. All contributing project authors may be found in the AUTHORS
+; file in the root of the source tree.
+;
+
+ EXPORT |vp8_dc_only_idct_add_v6|
+
+ AREA |.text|, CODE, READONLY
+
+;void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr,
+; int pred_stride, unsigned char *dst_ptr,
+; int dst_stride)
+; r0 input_dc
+; r1 pred_ptr
+; r2 pred_stride
+; r3 dst_ptr
+; sp dst_stride
+
+|vp8_dc_only_idct_add_v6| PROC
+ stmdb sp!, {r4 - r7}
+
+ add r0, r0, #4 ; input_dc += 4
+ ldr r12, c0x0000FFFF
+ ldr r4, [r1], r2
+ and r0, r12, r0, asr #3 ; input_dc >> 3 + mask
+ ldr r6, [r1], r2
+ orr r0, r0, r0, lsl #16 ; a1 | a1
+
+ ldr r12, [sp, #16] ; dst stride
+
+ uxtab16 r5, r0, r4 ; a1+2 | a1+0
+ uxtab16 r4, r0, r4, ror #8 ; a1+3 | a1+1
+ uxtab16 r7, r0, r6
+ uxtab16 r6, r0, r6, ror #8
+ usat16 r5, #8, r5
+ usat16 r4, #8, r4
+ usat16 r7, #8, r7
+ usat16 r6, #8, r6
+ orr r5, r5, r4, lsl #8
+ orr r7, r7, r6, lsl #8
+ ldr r4, [r1], r2
+ str r5, [r3], r12
+ ldr r6, [r1]
+ str r7, [r3], r12
+
+ uxtab16 r5, r0, r4
+ uxtab16 r4, r0, r4, ror #8
+ uxtab16 r7, r0, r6
+ uxtab16 r6, r0, r6, ror #8
+ usat16 r5, #8, r5
+ usat16 r4, #8, r4
+ usat16 r7, #8, r7
+ usat16 r6, #8, r6
+ orr r5, r5, r4, lsl #8
+ orr r7, r7, r6, lsl #8
+ str r5, [r3], r12
+ str r7, [r3]
+
+ ldmia sp!, {r4 - r7}
+ bx lr
+
+ ENDP ; |vp8_dc_only_idct_add_v6|
+
+; Constant Pool
+c0x0000FFFF DCD 0x0000FFFF
+ END
diff --git a/libvpx/vp8/common/arm/armv6/dequant_idct_v6.asm b/libvpx/vp8/common/arm/armv6/dequant_idct_v6.asm
new file mode 100644
index 0000000..2510ad8
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/dequant_idct_v6.asm
@@ -0,0 +1,190 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license and patent
+; grant that can be found in the LICENSE file in the root of the source
+; tree. All contributing project authors may be found in the AUTHORS
+; file in the root of the source tree.
+;
+
+ EXPORT |vp8_dequant_idct_add_v6|
+
+ AREA |.text|, CODE, READONLY
+;void vp8_dequant_idct_v6(short *input, short *dq,
+; unsigned char *dest, int stride)
+; r0 = q
+; r1 = dq
+; r2 = dst
+; r3 = stride
+
+|vp8_dequant_idct_add_v6| PROC
+ stmdb sp!, {r4-r11, lr}
+
+ ldr r4, [r0] ;input
+ ldr r5, [r1], #4 ;dq
+
+ sub sp, sp, #4
+ str r3, [sp]
+
+ mov r12, #4
+
+vp8_dequant_add_loop
+ smulbb r6, r4, r5
+ smultt r7, r4, r5
+
+ ldr r4, [r0, #4] ;input
+ ldr r5, [r1], #4 ;dq
+
+ strh r6, [r0], #2
+ strh r7, [r0], #2
+
+ smulbb r6, r4, r5
+ smultt r7, r4, r5
+
+ subs r12, r12, #1
+
+ ldrne r4, [r0, #4]
+ ldrne r5, [r1], #4
+
+ strh r6, [r0], #2
+ strh r7, [r0], #2
+
+ bne vp8_dequant_add_loop
+
+ sub r0, r0, #32
+ mov r1, r0
+
+; short_idct4x4llm_v6_dual
+ ldr r3, cospi8sqrt2minus1
+ ldr r4, sinpi8sqrt2
+ ldr r6, [r0, #8]
+ mov r5, #2
+vp8_dequant_idct_loop1_v6
+ ldr r12, [r0, #24]
+ ldr r14, [r0, #16]
+ smulwt r9, r3, r6
+ smulwb r7, r3, r6
+ smulwt r10, r4, r6
+ smulwb r8, r4, r6
+ pkhbt r7, r7, r9, lsl #16
+ smulwt r11, r3, r12
+ pkhbt r8, r8, r10, lsl #16
+ uadd16 r6, r6, r7
+ smulwt r7, r4, r12
+ smulwb r9, r3, r12
+ smulwb r10, r4, r12
+ subs r5, r5, #1
+ pkhbt r9, r9, r11, lsl #16
+ ldr r11, [r0], #4
+ pkhbt r10, r10, r7, lsl #16
+ uadd16 r7, r12, r9
+ usub16 r7, r8, r7
+ uadd16 r6, r6, r10
+ uadd16 r10, r11, r14
+ usub16 r8, r11, r14
+ uadd16 r9, r10, r6
+ usub16 r10, r10, r6
+ uadd16 r6, r8, r7
+ usub16 r7, r8, r7
+ str r6, [r1, #8]
+ ldrne r6, [r0, #8]
+ str r7, [r1, #16]
+ str r10, [r1, #24]
+ str r9, [r1], #4
+ bne vp8_dequant_idct_loop1_v6
+
+ mov r5, #2
+ sub r0, r1, #8
+vp8_dequant_idct_loop2_v6
+ ldr r6, [r0], #4
+ ldr r7, [r0], #4
+ ldr r8, [r0], #4
+ ldr r9, [r0], #4
+ smulwt r1, r3, r6
+ smulwt r12, r4, r6
+ smulwt lr, r3, r8
+ smulwt r10, r4, r8
+ pkhbt r11, r8, r6, lsl #16
+ pkhbt r1, lr, r1, lsl #16
+ pkhbt r12, r10, r12, lsl #16
+ pkhtb r6, r6, r8, asr #16
+ uadd16 r6, r1, r6
+ pkhbt lr, r9, r7, lsl #16
+ uadd16 r10, r11, lr
+ usub16 lr, r11, lr
+ pkhtb r8, r7, r9, asr #16
+ subs r5, r5, #1
+ smulwt r1, r3, r8
+ smulwb r7, r3, r8
+ smulwt r11, r4, r8
+ smulwb r9, r4, r8
+ pkhbt r1, r7, r1, lsl #16
+ uadd16 r8, r1, r8
+ pkhbt r11, r9, r11, lsl #16
+ usub16 r1, r12, r8
+ uadd16 r8, r11, r6
+ ldr r9, c0x00040004
+ ldr r12, [sp] ; get stride from stack
+ uadd16 r6, r10, r8
+ usub16 r7, r10, r8
+ uadd16 r7, r7, r9
+ uadd16 r6, r6, r9
+ uadd16 r10, r14, r1
+ usub16 r1, r14, r1
+ uadd16 r10, r10, r9
+ uadd16 r1, r1, r9
+ ldr r11, [r2] ; load input from dst
+ mov r8, r7, asr #3
+ pkhtb r9, r8, r10, asr #19
+ mov r8, r1, asr #3
+ pkhtb r8, r8, r6, asr #19
+ uxtb16 lr, r11, ror #8
+ qadd16 r9, r9, lr
+ uxtb16 lr, r11
+ qadd16 r8, r8, lr
+ usat16 r9, #8, r9
+ usat16 r8, #8, r8
+ orr r9, r8, r9, lsl #8
+ ldr r11, [r2, r12] ; load input from dst
+ mov r7, r7, lsl #16
+ mov r1, r1, lsl #16
+ mov r10, r10, lsl #16
+ mov r6, r6, lsl #16
+ mov r7, r7, asr #3
+ pkhtb r7, r7, r10, asr #19
+ mov r1, r1, asr #3
+ pkhtb r1, r1, r6, asr #19
+ uxtb16 r8, r11, ror #8
+ qadd16 r7, r7, r8
+ uxtb16 r8, r11
+ qadd16 r1, r1, r8
+ usat16 r7, #8, r7
+ usat16 r1, #8, r1
+ orr r1, r1, r7, lsl #8
+ str r9, [r2], r12 ; store output to dst
+ str r1, [r2], r12 ; store output to dst
+ bne vp8_dequant_idct_loop2_v6
+
+; vpx_memset
+ sub r0, r0, #32
+ add sp, sp, #4
+
+ mov r12, #0
+ str r12, [r0]
+ str r12, [r0, #4]
+ str r12, [r0, #8]
+ str r12, [r0, #12]
+ str r12, [r0, #16]
+ str r12, [r0, #20]
+ str r12, [r0, #24]
+ str r12, [r0, #28]
+
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_dequant_idct_add_v6|
+
+; Constant Pool
+cospi8sqrt2minus1 DCD 0x00004E7B
+sinpi8sqrt2 DCD 0x00008A8C
+c0x00040004 DCD 0x00040004
+
+ END
diff --git a/vp8/decoder/arm/armv6/dequantize_v6.asm b/libvpx/vp8/common/arm/armv6/dequantize_v6.asm
similarity index 100%
rename from vp8/decoder/arm/armv6/dequantize_v6.asm
rename to libvpx/vp8/common/arm/armv6/dequantize_v6.asm
diff --git a/libvpx/vp8/common/arm/armv6/filter_v6.asm b/libvpx/vp8/common/arm/armv6/filter_v6.asm
new file mode 100644
index 0000000..1ba91dd
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/filter_v6.asm
@@ -0,0 +1,624 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_filter_block2d_first_pass_armv6|
+ EXPORT |vp8_filter_block2d_first_pass_16x16_armv6|
+ EXPORT |vp8_filter_block2d_first_pass_8x8_armv6|
+ EXPORT |vp8_filter_block2d_second_pass_armv6|
+ EXPORT |vp8_filter4_block2d_second_pass_armv6|
+ EXPORT |vp8_filter_block2d_first_pass_only_armv6|
+ EXPORT |vp8_filter_block2d_second_pass_only_armv6|
+
+ AREA |.text|, CODE, READONLY ; name this block of code
+;-------------------------------------
+; r0 unsigned char *src_ptr
+; r1 short *output_ptr
+; r2 unsigned int src_pixels_per_line
+; r3 unsigned int output_width
+; stack unsigned int output_height
+; stack const short *vp8_filter
+;-------------------------------------
+; vp8_filter the input and put in the output array. Apply the 6 tap FIR filter with
+; the output being a 2 byte value and the intput being a 1 byte value.
+|vp8_filter_block2d_first_pass_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ ldr r11, [sp, #40] ; vp8_filter address
+ ldr r7, [sp, #36] ; output height
+
+ sub r2, r2, r3 ; inside loop increments input array,
+ ; so the height loop only needs to add
+ ; r2 - width to the input pointer
+
+ mov r3, r3, lsl #1 ; multiply width by 2 because using shorts
+ add r12, r3, #16 ; square off the output
+ sub sp, sp, #4
+
+ ldr r4, [r11] ; load up packed filter coefficients
+ ldr r5, [r11, #4]
+ ldr r6, [r11, #8]
+
+ str r1, [sp] ; push destination to stack
+ mov r7, r7, lsl #16 ; height is top part of counter
+
+; six tap filter
+|height_loop_1st_6|
+ ldrb r8, [r0, #-2] ; load source data
+ ldrb r9, [r0, #-1]
+ ldrb r10, [r0], #2
+ orr r7, r7, r3, lsr #2 ; construct loop counter
+
+|width_loop_1st_6|
+ ldrb r11, [r0, #-1]
+
+ pkhbt lr, r8, r9, lsl #16 ; r9 | r8
+ pkhbt r8, r9, r10, lsl #16 ; r10 | r9
+
+ ldrb r9, [r0]
+
+ smuad lr, lr, r4 ; apply the filter
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+ smuad r8, r8, r4
+ pkhbt r11, r11, r9, lsl #16 ; r9 | r11
+
+ smlad lr, r10, r5, lr
+ ldrb r10, [r0, #1]
+ smlad r8, r11, r5, r8
+ ldrb r11, [r0, #2]
+
+ sub r7, r7, #1
+
+ pkhbt r9, r9, r10, lsl #16 ; r10 | r9
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+
+ smlad lr, r9, r6, lr
+ smlad r11, r10, r6, r8
+
+ ands r10, r7, #0xff ; test loop counter
+
+ add lr, lr, #0x40 ; round_shift_and_clamp
+ ldrneb r8, [r0, #-2] ; load data for next loop
+ usat lr, #8, lr, asr #7
+ add r11, r11, #0x40
+ ldrneb r9, [r0, #-1]
+ usat r11, #8, r11, asr #7
+
+ strh lr, [r1], r12 ; result is transposed and stored, which
+ ; will make second pass filtering easier.
+ ldrneb r10, [r0], #2
+ strh r11, [r1], r12
+
+ bne width_loop_1st_6
+
+ ldr r1, [sp] ; load and update dst address
+ subs r7, r7, #0x10000
+ add r0, r0, r2 ; move to next input line
+
+ add r1, r1, #2 ; move over to next column
+ str r1, [sp]
+
+ bne height_loop_1st_6
+
+ add sp, sp, #4
+ ldmia sp!, {r4 - r11, pc}
+
+ ENDP
+
+; --------------------------
+; 16x16 version
+; -----------------------------
+|vp8_filter_block2d_first_pass_16x16_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ ldr r11, [sp, #40] ; vp8_filter address
+ ldr r7, [sp, #36] ; output height
+
+ add r4, r2, #18 ; preload next low
+ pld [r0, r4]
+
+ sub r2, r2, r3 ; inside loop increments input array,
+ ; so the height loop only needs to add
+ ; r2 - width to the input pointer
+
+ mov r3, r3, lsl #1 ; multiply width by 2 because using shorts
+ add r12, r3, #16 ; square off the output
+ sub sp, sp, #4
+
+ ldr r4, [r11] ; load up packed filter coefficients
+ ldr r5, [r11, #4]
+ ldr r6, [r11, #8]
+
+ str r1, [sp] ; push destination to stack
+ mov r7, r7, lsl #16 ; height is top part of counter
+
+; six tap filter
+|height_loop_1st_16_6|
+ ldrb r8, [r0, #-2] ; load source data
+ ldrb r9, [r0, #-1]
+ ldrb r10, [r0], #2
+ orr r7, r7, r3, lsr #2 ; construct loop counter
+
+|width_loop_1st_16_6|
+ ldrb r11, [r0, #-1]
+
+ pkhbt lr, r8, r9, lsl #16 ; r9 | r8
+ pkhbt r8, r9, r10, lsl #16 ; r10 | r9
+
+ ldrb r9, [r0]
+
+ smuad lr, lr, r4 ; apply the filter
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+ smuad r8, r8, r4
+ pkhbt r11, r11, r9, lsl #16 ; r9 | r11
+
+ smlad lr, r10, r5, lr
+ ldrb r10, [r0, #1]
+ smlad r8, r11, r5, r8
+ ldrb r11, [r0, #2]
+
+ sub r7, r7, #1
+
+ pkhbt r9, r9, r10, lsl #16 ; r10 | r9
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+
+ smlad lr, r9, r6, lr
+ smlad r11, r10, r6, r8
+
+ ands r10, r7, #0xff ; test loop counter
+
+ add lr, lr, #0x40 ; round_shift_and_clamp
+ ldrneb r8, [r0, #-2] ; load data for next loop
+ usat lr, #8, lr, asr #7
+ add r11, r11, #0x40
+ ldrneb r9, [r0, #-1]
+ usat r11, #8, r11, asr #7
+
+ strh lr, [r1], r12 ; result is transposed and stored, which
+ ; will make second pass filtering easier.
+ ldrneb r10, [r0], #2
+ strh r11, [r1], r12
+
+ bne width_loop_1st_16_6
+
+ ldr r1, [sp] ; load and update dst address
+ subs r7, r7, #0x10000
+ add r0, r0, r2 ; move to next input line
+
+ add r11, r2, #34 ; adding back block width(=16)
+ pld [r0, r11] ; preload next low
+
+ add r1, r1, #2 ; move over to next column
+ str r1, [sp]
+
+ bne height_loop_1st_16_6
+
+ add sp, sp, #4
+ ldmia sp!, {r4 - r11, pc}
+
+ ENDP
+
+; --------------------------
+; 8x8 version
+; -----------------------------
+|vp8_filter_block2d_first_pass_8x8_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ ldr r11, [sp, #40] ; vp8_filter address
+ ldr r7, [sp, #36] ; output height
+
+ add r4, r2, #10 ; preload next low
+ pld [r0, r4]
+
+ sub r2, r2, r3 ; inside loop increments input array,
+ ; so the height loop only needs to add
+ ; r2 - width to the input pointer
+
+ mov r3, r3, lsl #1 ; multiply width by 2 because using shorts
+ add r12, r3, #16 ; square off the output
+ sub sp, sp, #4
+
+ ldr r4, [r11] ; load up packed filter coefficients
+ ldr r5, [r11, #4]
+ ldr r6, [r11, #8]
+
+ str r1, [sp] ; push destination to stack
+ mov r7, r7, lsl #16 ; height is top part of counter
+
+; six tap filter
+|height_loop_1st_8_6|
+ ldrb r8, [r0, #-2] ; load source data
+ ldrb r9, [r0, #-1]
+ ldrb r10, [r0], #2
+ orr r7, r7, r3, lsr #2 ; construct loop counter
+
+|width_loop_1st_8_6|
+ ldrb r11, [r0, #-1]
+
+ pkhbt lr, r8, r9, lsl #16 ; r9 | r8
+ pkhbt r8, r9, r10, lsl #16 ; r10 | r9
+
+ ldrb r9, [r0]
+
+ smuad lr, lr, r4 ; apply the filter
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+ smuad r8, r8, r4
+ pkhbt r11, r11, r9, lsl #16 ; r9 | r11
+
+ smlad lr, r10, r5, lr
+ ldrb r10, [r0, #1]
+ smlad r8, r11, r5, r8
+ ldrb r11, [r0, #2]
+
+ sub r7, r7, #1
+
+ pkhbt r9, r9, r10, lsl #16 ; r10 | r9
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+
+ smlad lr, r9, r6, lr
+ smlad r11, r10, r6, r8
+
+ ands r10, r7, #0xff ; test loop counter
+
+ add lr, lr, #0x40 ; round_shift_and_clamp
+ ldrneb r8, [r0, #-2] ; load data for next loop
+ usat lr, #8, lr, asr #7
+ add r11, r11, #0x40
+ ldrneb r9, [r0, #-1]
+ usat r11, #8, r11, asr #7
+
+ strh lr, [r1], r12 ; result is transposed and stored, which
+ ; will make second pass filtering easier.
+ ldrneb r10, [r0], #2
+ strh r11, [r1], r12
+
+ bne width_loop_1st_8_6
+
+ ldr r1, [sp] ; load and update dst address
+ subs r7, r7, #0x10000
+ add r0, r0, r2 ; move to next input line
+
+ add r11, r2, #18 ; adding back block width(=8)
+ pld [r0, r11] ; preload next low
+
+ add r1, r1, #2 ; move over to next column
+ str r1, [sp]
+
+ bne height_loop_1st_8_6
+
+ add sp, sp, #4
+ ldmia sp!, {r4 - r11, pc}
+
+ ENDP
+
+;---------------------------------
+; r0 short *src_ptr,
+; r1 unsigned char *output_ptr,
+; r2 unsigned int output_pitch,
+; r3 unsigned int cnt,
+; stack const short *vp8_filter
+;---------------------------------
+|vp8_filter_block2d_second_pass_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ ldr r11, [sp, #36] ; vp8_filter address
+ sub sp, sp, #4
+ mov r7, r3, lsl #16 ; height is top part of counter
+ str r1, [sp] ; push destination to stack
+
+ ldr r4, [r11] ; load up packed filter coefficients
+ ldr r5, [r11, #4]
+ ldr r6, [r11, #8]
+
+ pkhbt r12, r5, r4 ; pack the filter differently
+ pkhbt r11, r6, r5
+
+ sub r0, r0, #4 ; offset input buffer
+
+|height_loop_2nd|
+ ldr r8, [r0] ; load the data
+ ldr r9, [r0, #4]
+ orr r7, r7, r3, lsr #1 ; loop counter
+
+|width_loop_2nd|
+ smuad lr, r4, r8 ; apply filter
+ sub r7, r7, #1
+ smulbt r8, r4, r8
+
+ ldr r10, [r0, #8]
+
+ smlad lr, r5, r9, lr
+ smladx r8, r12, r9, r8
+
+ ldrh r9, [r0, #12]
+
+ smlad lr, r6, r10, lr
+ smladx r8, r11, r10, r8
+
+ add r0, r0, #4
+ smlatb r10, r6, r9, r8
+
+ add lr, lr, #0x40 ; round_shift_and_clamp
+ ands r8, r7, #0xff
+ usat lr, #8, lr, asr #7
+ add r10, r10, #0x40
+ strb lr, [r1], r2 ; the result is transposed back and stored
+ usat r10, #8, r10, asr #7
+
+ ldrne r8, [r0] ; load data for next loop
+ ldrne r9, [r0, #4]
+ strb r10, [r1], r2
+
+ bne width_loop_2nd
+
+ ldr r1, [sp] ; update dst for next loop
+ subs r7, r7, #0x10000
+ add r0, r0, #16 ; updata src for next loop
+ add r1, r1, #1
+ str r1, [sp]
+
+ bne height_loop_2nd
+
+ add sp, sp, #4
+ ldmia sp!, {r4 - r11, pc}
+
+ ENDP
+
+;---------------------------------
+; r0 short *src_ptr,
+; r1 unsigned char *output_ptr,
+; r2 unsigned int output_pitch,
+; r3 unsigned int cnt,
+; stack const short *vp8_filter
+;---------------------------------
+|vp8_filter4_block2d_second_pass_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ ldr r11, [sp, #36] ; vp8_filter address
+ mov r7, r3, lsl #16 ; height is top part of counter
+
+ ldr r4, [r11] ; load up packed filter coefficients
+ add lr, r1, r3 ; save final destination pointer
+ ldr r5, [r11, #4]
+ ldr r6, [r11, #8]
+
+ pkhbt r12, r5, r4 ; pack the filter differently
+ pkhbt r11, r6, r5
+ mov r4, #0x40 ; rounding factor (for smlad{x})
+
+|height_loop_2nd_4|
+ ldrd r8, [r0, #-4] ; load the data
+ orr r7, r7, r3, lsr #1 ; loop counter
+
+|width_loop_2nd_4|
+ ldr r10, [r0, #4]!
+ smladx r6, r9, r12, r4 ; apply filter
+ pkhbt r8, r9, r8
+ smlad r5, r8, r12, r4
+ pkhbt r8, r10, r9
+ smladx r6, r10, r11, r6
+ sub r7, r7, #1
+ smlad r5, r8, r11, r5
+
+ mov r8, r9 ; shift the data for the next loop
+ mov r9, r10
+
+ usat r6, #8, r6, asr #7 ; shift and clamp
+ usat r5, #8, r5, asr #7
+
+ strb r5, [r1], r2 ; the result is transposed back and stored
+ tst r7, #0xff
+ strb r6, [r1], r2
+
+ bne width_loop_2nd_4
+
+ subs r7, r7, #0x10000
+ add r0, r0, #16 ; update src for next loop
+ sub r1, lr, r7, lsr #16 ; update dst for next loop
+
+ bne height_loop_2nd_4
+
+ ldmia sp!, {r4 - r11, pc}
+
+ ENDP
+
+;------------------------------------
+; r0 unsigned char *src_ptr
+; r1 unsigned char *output_ptr,
+; r2 unsigned int src_pixels_per_line
+; r3 unsigned int cnt,
+; stack unsigned int output_pitch,
+; stack const short *vp8_filter
+;------------------------------------
+|vp8_filter_block2d_first_pass_only_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ add r7, r2, r3 ; preload next low
+ add r7, r7, #2
+ pld [r0, r7]
+
+ ldr r4, [sp, #36] ; output pitch
+ ldr r11, [sp, #40] ; HFilter address
+ sub sp, sp, #8
+
+ mov r7, r3
+ sub r2, r2, r3 ; inside loop increments input array,
+ ; so the height loop only needs to add
+ ; r2 - width to the input pointer
+
+ sub r4, r4, r3
+ str r4, [sp] ; save modified output pitch
+ str r2, [sp, #4]
+
+ mov r2, #0x40
+
+ ldr r4, [r11] ; load up packed filter coefficients
+ ldr r5, [r11, #4]
+ ldr r6, [r11, #8]
+
+; six tap filter
+|height_loop_1st_only_6|
+ ldrb r8, [r0, #-2] ; load data
+ ldrb r9, [r0, #-1]
+ ldrb r10, [r0], #2
+
+ mov r12, r3, lsr #1 ; loop counter
+
+|width_loop_1st_only_6|
+ ldrb r11, [r0, #-1]
+
+ pkhbt lr, r8, r9, lsl #16 ; r9 | r8
+ pkhbt r8, r9, r10, lsl #16 ; r10 | r9
+
+ ldrb r9, [r0]
+
+;; smuad lr, lr, r4
+ smlad lr, lr, r4, r2
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+;; smuad r8, r8, r4
+ smlad r8, r8, r4, r2
+ pkhbt r11, r11, r9, lsl #16 ; r9 | r11
+
+ smlad lr, r10, r5, lr
+ ldrb r10, [r0, #1]
+ smlad r8, r11, r5, r8
+ ldrb r11, [r0, #2]
+
+ subs r12, r12, #1
+
+ pkhbt r9, r9, r10, lsl #16 ; r10 | r9
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+
+ smlad lr, r9, r6, lr
+ smlad r10, r10, r6, r8
+
+;; add lr, lr, #0x40 ; round_shift_and_clamp
+ ldrneb r8, [r0, #-2] ; load data for next loop
+ usat lr, #8, lr, asr #7
+;; add r10, r10, #0x40
+ strb lr, [r1], #1 ; store the result
+ usat r10, #8, r10, asr #7
+
+ ldrneb r9, [r0, #-1]
+ strb r10, [r1], #1
+ ldrneb r10, [r0], #2
+
+ bne width_loop_1st_only_6
+
+ ldr lr, [sp] ; load back output pitch
+ ldr r12, [sp, #4] ; load back output pitch
+ subs r7, r7, #1
+ add r0, r0, r12 ; updata src for next loop
+
+ add r11, r12, r3 ; preload next low
+ add r11, r11, #2
+ pld [r0, r11]
+
+ add r1, r1, lr ; update dst for next loop
+
+ bne height_loop_1st_only_6
+
+ add sp, sp, #8
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_filter_block2d_first_pass_only_armv6|
+
+
+;------------------------------------
+; r0 unsigned char *src_ptr,
+; r1 unsigned char *output_ptr,
+; r2 unsigned int src_pixels_per_line
+; r3 unsigned int cnt,
+; stack unsigned int output_pitch,
+; stack const short *vp8_filter
+;------------------------------------
+|vp8_filter_block2d_second_pass_only_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+
+ ldr r11, [sp, #40] ; VFilter address
+ ldr r12, [sp, #36] ; output pitch
+
+ mov r7, r3, lsl #16 ; height is top part of counter
+ sub r0, r0, r2, lsl #1 ; need 6 elements for filtering, 2 before, 3 after
+
+ sub sp, sp, #8
+
+ ldr r4, [r11] ; load up packed filter coefficients
+ ldr r5, [r11, #4]
+ ldr r6, [r11, #8]
+
+ str r0, [sp] ; save r0 to stack
+ str r1, [sp, #4] ; save dst to stack
+
+; six tap filter
+|width_loop_2nd_only_6|
+ ldrb r8, [r0], r2 ; load data
+ orr r7, r7, r3 ; loop counter
+ ldrb r9, [r0], r2
+ ldrb r10, [r0], r2
+
+|height_loop_2nd_only_6|
+ ; filter first column in this inner loop, than, move to next colum.
+ ldrb r11, [r0], r2
+
+ pkhbt lr, r8, r9, lsl #16 ; r9 | r8
+ pkhbt r8, r9, r10, lsl #16 ; r10 | r9
+
+ ldrb r9, [r0], r2
+
+ smuad lr, lr, r4
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+ smuad r8, r8, r4
+ pkhbt r11, r11, r9, lsl #16 ; r9 | r11
+
+ smlad lr, r10, r5, lr
+ ldrb r10, [r0], r2
+ smlad r8, r11, r5, r8
+ ldrb r11, [r0]
+
+ sub r7, r7, #2
+ sub r0, r0, r2, lsl #2
+
+ pkhbt r9, r9, r10, lsl #16 ; r10 | r9
+ pkhbt r10, r10, r11, lsl #16 ; r11 | r10
+
+ smlad lr, r9, r6, lr
+ smlad r10, r10, r6, r8
+
+ ands r9, r7, #0xff
+
+ add lr, lr, #0x40 ; round_shift_and_clamp
+ ldrneb r8, [r0], r2 ; load data for next loop
+ usat lr, #8, lr, asr #7
+ add r10, r10, #0x40
+ strb lr, [r1], r12 ; store the result for the column
+ usat r10, #8, r10, asr #7
+
+ ldrneb r9, [r0], r2
+ strb r10, [r1], r12
+ ldrneb r10, [r0], r2
+
+ bne height_loop_2nd_only_6
+
+ ldr r0, [sp]
+ ldr r1, [sp, #4]
+ subs r7, r7, #0x10000
+ add r0, r0, #1 ; move to filter next column
+ str r0, [sp]
+ add r1, r1, #1
+ str r1, [sp, #4]
+
+ bne width_loop_2nd_only_6
+
+ add sp, sp, #8
+
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_filter_block2d_second_pass_only_armv6|
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/idct_blk_v6.c b/libvpx/vp8/common/arm/armv6/idct_blk_v6.c
new file mode 100644
index 0000000..6002c0f
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/idct_blk_v6.c
@@ -0,0 +1,115 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+
+
+void vp8_dequant_idct_add_y_block_v6(short *q, short *dq,
+ unsigned char *dst,
+ int stride, char *eobs)
+{
+ int i;
+
+ for (i = 0; i < 4; i++)
+ {
+ if (eobs[0] > 1)
+ vp8_dequant_idct_add_v6 (q, dq, dst, stride);
+ else if (eobs[0] == 1)
+ {
+ vp8_dc_only_idct_add_v6 (q[0]*dq[0], dst, stride, dst, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ if (eobs[1] > 1)
+ vp8_dequant_idct_add_v6 (q+16, dq, dst+4, stride);
+ else if (eobs[1] == 1)
+ {
+ vp8_dc_only_idct_add_v6 (q[16]*dq[0], dst+4, stride, dst+4, stride);
+ ((int *)(q+16))[0] = 0;
+ }
+
+ if (eobs[2] > 1)
+ vp8_dequant_idct_add_v6 (q+32, dq, dst+8, stride);
+ else if (eobs[2] == 1)
+ {
+ vp8_dc_only_idct_add_v6 (q[32]*dq[0], dst+8, stride, dst+8, stride);
+ ((int *)(q+32))[0] = 0;
+ }
+
+ if (eobs[3] > 1)
+ vp8_dequant_idct_add_v6 (q+48, dq, dst+12, stride);
+ else if (eobs[3] == 1)
+ {
+ vp8_dc_only_idct_add_v6 (q[48]*dq[0], dst+12, stride,dst+12,stride);
+ ((int *)(q+48))[0] = 0;
+ }
+
+ q += 64;
+ dst += 4*stride;
+ eobs += 4;
+ }
+}
+
+void vp8_dequant_idct_add_uv_block_v6(short *q, short *dq,
+ unsigned char *dstu,
+ unsigned char *dstv,
+ int stride, char *eobs)
+{
+ int i;
+
+ for (i = 0; i < 2; i++)
+ {
+ if (eobs[0] > 1)
+ vp8_dequant_idct_add_v6 (q, dq, dstu, stride);
+ else if (eobs[0] == 1)
+ {
+ vp8_dc_only_idct_add_v6 (q[0]*dq[0], dstu, stride, dstu, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ if (eobs[1] > 1)
+ vp8_dequant_idct_add_v6 (q+16, dq, dstu+4, stride);
+ else if (eobs[1] == 1)
+ {
+ vp8_dc_only_idct_add_v6 (q[16]*dq[0], dstu+4, stride,
+ dstu+4, stride);
+ ((int *)(q+16))[0] = 0;
+ }
+
+ q += 32;
+ dstu += 4*stride;
+ eobs += 2;
+ }
+
+ for (i = 0; i < 2; i++)
+ {
+ if (eobs[0] > 1)
+ vp8_dequant_idct_add_v6 (q, dq, dstv, stride);
+ else if (eobs[0] == 1)
+ {
+ vp8_dc_only_idct_add_v6 (q[0]*dq[0], dstv, stride, dstv, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ if (eobs[1] > 1)
+ vp8_dequant_idct_add_v6 (q+16, dq, dstv+4, stride);
+ else if (eobs[1] == 1)
+ {
+ vp8_dc_only_idct_add_v6 (q[16]*dq[0], dstv+4, stride,
+ dstv+4, stride);
+ ((int *)(q+16))[0] = 0;
+ }
+
+ q += 32;
+ dstv += 4*stride;
+ eobs += 2;
+ }
+}
diff --git a/libvpx/vp8/common/arm/armv6/idct_v6.asm b/libvpx/vp8/common/arm/armv6/idct_v6.asm
new file mode 100644
index 0000000..b4d44cb
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/idct_v6.asm
@@ -0,0 +1,202 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_short_idct4x4llm_v6_dual|
+
+ AREA |.text|, CODE, READONLY
+
+
+; void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch,
+; unsigned char *dst, int stride)
+; r0 short* input
+; r1 unsigned char* pred
+; r2 int pitch
+; r3 unsigned char* dst
+; sp int stride
+
+|vp8_short_idct4x4llm_v6_dual| PROC
+ stmdb sp!, {r4-r11, lr}
+
+ sub sp, sp, #4
+
+ mov r4, #0x00008A00 ; sin
+ orr r4, r4, #0x0000008C ; sinpi8sqrt2
+
+ mov r5, #0x00004E00 ; cos
+ orr r5, r5, #0x0000007B ; cospi8sqrt2minus1
+ orr r5, r5, #1<<31 ; loop counter on top bit
+
+loop1_dual
+ ldr r6, [r0, #(4*2)] ; i5 | i4
+ ldr r12, [r0, #(12*2)] ; i13|i12
+ ldr r14, [r0, #(8*2)] ; i9 | i8
+
+ smulbt r9, r5, r6 ; (ip[5] * cospi8sqrt2minus1) >> 16
+ smulbb r7, r5, r6 ; (ip[4] * cospi8sqrt2minus1) >> 16
+ smulwt r10, r4, r6 ; (ip[5] * sinpi8sqrt2) >> 16
+ smulwb r8, r4, r6 ; (ip[4] * sinpi8sqrt2) >> 16
+
+ smulbt r11, r5, r12 ; (ip[13] * cospi8sqrt2minus1) >> 16
+ pkhtb r7, r9, r7, asr #16 ; 5c | 4c
+ pkhbt r8, r8, r10, lsl #16 ; 5s | 4s
+ uadd16 r6, r6, r7 ; 5c+5 | 4c+4
+
+ smulwt r7, r4, r12 ; (ip[13] * sinpi8sqrt2) >> 16
+ smulbb r9, r5, r12 ; (ip[12] * cospi8sqrt2minus1) >> 16
+ smulwb r10, r4, r12 ; (ip[12] * sinpi8sqrt2) >> 16
+
+ subs r5, r5, #1<<31 ; i--
+
+ pkhtb r9, r11, r9, asr #16 ; 13c | 12c
+ ldr r11, [r0] ; i1 | i0
+ pkhbt r10, r10, r7, lsl #16 ; 13s | 12s
+ uadd16 r7, r12, r9 ; 13c+13 | 12c+12
+
+ usub16 r7, r8, r7 ; c
+ uadd16 r6, r6, r10 ; d
+ uadd16 r10, r11, r14 ; a
+ usub16 r8, r11, r14 ; b
+
+ uadd16 r9, r10, r6 ; a+d
+ usub16 r10, r10, r6 ; a-d
+ uadd16 r6, r8, r7 ; b+c
+ usub16 r7, r8, r7 ; b-c
+
+ ; use input buffer to store intermediate results
+ str r6, [r0, #(4*2)] ; o5 | o4
+ str r7, [r0, #(8*2)] ; o9 | o8
+ str r10,[r0, #(12*2)] ; o13|o12
+ str r9, [r0], #4 ; o1 | o0
+
+ bcs loop1_dual
+
+ sub r0, r0, #8 ; reset input/output
+ str r0, [sp]
+
+loop2_dual
+
+ ldr r6, [r0, #(4*2)] ; i5 | i4
+ ldr r12,[r0, #(2*2)] ; i3 | i2
+ ldr r14,[r0, #(6*2)] ; i7 | i6
+ ldr r0, [r0, #(0*2)] ; i1 | i0
+
+ smulbt r9, r5, r6 ; (ip[5] * cospi8sqrt2minus1) >> 16
+ smulbt r7, r5, r0 ; (ip[1] * cospi8sqrt2minus1) >> 16
+ smulwt r10, r4, r6 ; (ip[5] * sinpi8sqrt2) >> 16
+ smulwt r8, r4, r0 ; (ip[1] * sinpi8sqrt2) >> 16
+
+ pkhbt r11, r6, r0, lsl #16 ; i0 | i4
+ pkhtb r7, r7, r9, asr #16 ; 1c | 5c
+ pkhtb r0, r0, r6, asr #16 ; i1 | i5
+ pkhbt r8, r10, r8, lsl #16 ; 1s | 5s = temp1
+
+ uadd16 r0, r7, r0 ; 1c+1 | 5c+5 = temp2
+ pkhbt r9, r14, r12, lsl #16 ; i2 | i6
+ uadd16 r10, r11, r9 ; a
+ usub16 r9, r11, r9 ; b
+ pkhtb r6, r12, r14, asr #16 ; i3 | i7
+
+ subs r5, r5, #1<<31 ; i--
+
+ smulbt r7, r5, r6 ; (ip[3] * cospi8sqrt2minus1) >> 16
+ smulwt r11, r4, r6 ; (ip[3] * sinpi8sqrt2) >> 16
+ smulbb r12, r5, r6 ; (ip[7] * cospi8sqrt2minus1) >> 16
+ smulwb r14, r4, r6 ; (ip[7] * sinpi8sqrt2) >> 16
+
+ pkhtb r7, r7, r12, asr #16 ; 3c | 7c
+ pkhbt r11, r14, r11, lsl #16 ; 3s | 7s = temp1
+
+ uadd16 r6, r7, r6 ; 3c+3 | 7c+7 = temp2
+ usub16 r12, r8, r6 ; c (o1 | o5)
+ uadd16 r6, r11, r0 ; d (o3 | o7)
+ uadd16 r7, r10, r6 ; a+d
+
+ mov r8, #4 ; set up 4's
+ orr r8, r8, #0x40000 ; 4|4
+
+ usub16 r6, r10, r6 ; a-d
+ uadd16 r6, r6, r8 ; a-d+4, 3|7
+ uadd16 r7, r7, r8 ; a+d+4, 0|4
+ uadd16 r10, r9, r12 ; b+c
+ usub16 r0, r9, r12 ; b-c
+ uadd16 r10, r10, r8 ; b+c+4, 1|5
+ uadd16 r8, r0, r8 ; b-c+4, 2|6
+
+ ldr lr, [sp, #40] ; dst stride
+
+ ldrb r0, [r1] ; pred p0
+ ldrb r11, [r1, #1] ; pred p1
+ ldrb r12, [r1, #2] ; pred p2
+
+ add r0, r0, r7, asr #19 ; p0 + o0
+ add r11, r11, r10, asr #19 ; p1 + o1
+ add r12, r12, r8, asr #19 ; p2 + o2
+
+ usat r0, #8, r0 ; d0 = clip8(p0 + o0)
+ usat r11, #8, r11 ; d1 = clip8(p1 + o1)
+ usat r12, #8, r12 ; d2 = clip8(p2 + o2)
+
+ add r0, r0, r11, lsl #8 ; |--|--|d1|d0|
+
+ ldrb r11, [r1, #3] ; pred p3
+
+ add r0, r0, r12, lsl #16 ; |--|d2|d1|d0|
+
+ add r11, r11, r6, asr #19 ; p3 + o3
+
+ sxth r7, r7 ;
+ sxth r10, r10 ;
+
+ usat r11, #8, r11 ; d3 = clip8(p3 + o3)
+
+ sxth r8, r8 ;
+ sxth r6, r6 ;
+
+ add r0, r0, r11, lsl #24 ; |d3|d2|d1|d0|
+
+ ldrb r12, [r1, r2]! ; pred p4
+ str r0, [r3], lr
+ ldrb r11, [r1, #1] ; pred p5
+
+ add r12, r12, r7, asr #3 ; p4 + o4
+ add r11, r11, r10, asr #3 ; p5 + o5
+
+ usat r12, #8, r12 ; d4 = clip8(p4 + o4)
+ usat r11, #8, r11 ; d5 = clip8(p5 + o5)
+
+ ldrb r7, [r1, #2] ; pred p6
+ ldrb r10, [r1, #3] ; pred p6
+
+ add r12, r12, r11, lsl #8 ; |--|--|d5|d4|
+
+ add r7, r7, r8, asr #3 ; p6 + o6
+ add r10, r10, r6, asr #3 ; p7 + o7
+
+ ldr r0, [sp] ; load input pointer
+
+ usat r7, #8, r7 ; d6 = clip8(p6 + o6)
+ usat r10, #8, r10 ; d7 = clip8(p7 + o7)
+
+ add r12, r12, r7, lsl #16 ; |--|d6|d5|d4|
+ add r12, r12, r10, lsl #24 ; |d7|d6|d5|d4|
+
+ str r12, [r3], lr
+ add r0, r0, #16
+ add r1, r1, r2 ; pred + pitch
+
+ bcs loop2_dual
+
+ add sp, sp, #4 ; idct_output buffer
+ ldmia sp!, {r4 - r11, pc}
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/intra4x4_predict_v6.asm b/libvpx/vp8/common/arm/armv6/intra4x4_predict_v6.asm
new file mode 100644
index 0000000..c5ec824
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/intra4x4_predict_v6.asm
@@ -0,0 +1,611 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_intra4x4_predict_armv6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+
+;void vp8_intra4x4_predict_armv6(unsigned char *Above, unsigned char *yleft,
+; B_PREDICTION_MODE left_stride, int b_mode,
+; unsigned char *dst, int dst_stride,
+; unsigned char top_left)
+
+; r0: *Above
+; r1: *yleft
+; r2: left_stride
+; r3: b_mode
+; sp + #40: dst
+; sp + #44: dst_stride
+; sp + #48: top_left
+|vp8_intra4x4_predict_armv6| PROC
+ push {r4-r12, lr}
+
+ cmp r3, #10
+ addlt pc, pc, r3, lsl #2 ; position independent switch
+ pop {r4-r12, pc} ; default
+ b b_dc_pred
+ b b_tm_pred
+ b b_ve_pred
+ b b_he_pred
+ b b_ld_pred
+ b b_rd_pred
+ b b_vr_pred
+ b b_vl_pred
+ b b_hd_pred
+ b b_hu_pred
+
+b_dc_pred
+ ; load values
+ ldr r8, [r0] ; Above
+ ldrb r4, [r1], r2 ; Left[0]
+ mov r9, #0
+ ldrb r5, [r1], r2 ; Left[1]
+ ldrb r6, [r1], r2 ; Left[2]
+ usad8 r12, r8, r9
+ ldrb r7, [r1] ; Left[3]
+
+ ; calculate dc
+ add r4, r4, r5
+ add r4, r4, r6
+ add r4, r4, r7
+ add r4, r4, r12
+ add r4, r4, #4
+ ldr r0, [sp, #44] ; dst_stride
+ mov r12, r4, asr #3 ; (expected_dc + 4) >> 3
+
+ add r12, r12, r12, lsl #8
+ ldr r3, [sp, #40] ; dst
+ add r12, r12, r12, lsl #16
+
+ ; store values
+ str r12, [r3], r0
+ str r12, [r3], r0
+ str r12, [r3], r0
+ str r12, [r3]
+
+ pop {r4-r12, pc}
+
+b_tm_pred
+ ldr r8, [r0] ; Above
+ ldrb r9, [sp, #48] ; top_left
+ ldrb r4, [r1], r2 ; Left[0]
+ ldrb r5, [r1], r2 ; Left[1]
+ ldrb r6, [r1], r2 ; Left[2]
+ ldrb r7, [r1] ; Left[3]
+ ldr r0, [sp, #44] ; dst_stride
+ ldr r3, [sp, #40] ; dst
+
+ add r9, r9, r9, lsl #16 ; [tl|tl]
+ uxtb16 r10, r8 ; a[2|0]
+ uxtb16 r11, r8, ror #8 ; a[3|1]
+ ssub16 r10, r10, r9 ; a[2|0] - [tl|tl]
+ ssub16 r11, r11, r9 ; a[3|1] - [tl|tl]
+
+ add r4, r4, r4, lsl #16 ; l[0|0]
+ add r5, r5, r5, lsl #16 ; l[1|1]
+ add r6, r6, r6, lsl #16 ; l[2|2]
+ add r7, r7, r7, lsl #16 ; l[3|3]
+
+ sadd16 r1, r4, r10 ; l[0|0] + a[2|0] - [tl|tl]
+ sadd16 r2, r4, r11 ; l[0|0] + a[3|1] - [tl|tl]
+ usat16 r1, #8, r1
+ usat16 r2, #8, r2
+
+ sadd16 r4, r5, r10 ; l[1|1] + a[2|0] - [tl|tl]
+ sadd16 r5, r5, r11 ; l[1|1] + a[3|1] - [tl|tl]
+
+ add r12, r1, r2, lsl #8 ; [3|2|1|0]
+ str r12, [r3], r0
+
+ usat16 r4, #8, r4
+ usat16 r5, #8, r5
+
+ sadd16 r1, r6, r10 ; l[2|2] + a[2|0] - [tl|tl]
+ sadd16 r2, r6, r11 ; l[2|2] + a[3|1] - [tl|tl]
+
+ add r12, r4, r5, lsl #8 ; [3|2|1|0]
+ str r12, [r3], r0
+
+ usat16 r1, #8, r1
+ usat16 r2, #8, r2
+
+ sadd16 r4, r7, r10 ; l[3|3] + a[2|0] - [tl|tl]
+ sadd16 r5, r7, r11 ; l[3|3] + a[3|1] - [tl|tl]
+
+ add r12, r1, r2, lsl #8 ; [3|2|1|0]
+
+ usat16 r4, #8, r4
+ usat16 r5, #8, r5
+
+ str r12, [r3], r0
+
+ add r12, r4, r5, lsl #8 ; [3|2|1|0]
+ str r12, [r3]
+
+ pop {r4-r12, pc}
+
+b_ve_pred
+ ldr r8, [r0] ; a[3|2|1|0]
+ ldr r11, c00FF00FF
+ ldrb r9, [sp, #48] ; top_left
+ ldrb r10, [r0, #4] ; a[4]
+
+ ldr r0, c00020002
+
+ uxtb16 r4, r8 ; a[2|0]
+ uxtb16 r5, r8, ror #8 ; a[3|1]
+ ldr r2, [sp, #44] ; dst_stride
+ pkhbt r9, r9, r5, lsl #16 ; a[1|-1]
+
+ add r9, r9, r4, lsl #1 ;[a[1]+2*a[2] | tl+2*a[0] ]
+ uxtab16 r9, r9, r5 ;[a[1]+2*a[2]+a[3] | tl+2*a[0]+a[1] ]
+ ldr r3, [sp, #40] ; dst
+ uxtab16 r9, r9, r0 ;[a[1]+2*a[2]+a[3]+2| tl+2*a[0]+a[1]+2]
+
+ add r0, r0, r10, lsl #16 ;[a[4]+2 | 2]
+ add r0, r0, r4, asr #16 ;[a[4]+2 | a[2]+2]
+ add r0, r0, r5, lsl #1 ;[a[4]+2*a[3]+2 | a[2]+2*a[1]+2]
+ uadd16 r4, r4, r0 ;[a[4]+2*a[3]+a[2]+2|a[2]+2*a[1]+a[0]+2]
+
+ and r9, r11, r9, asr #2
+ and r4, r11, r4, asr #2
+ add r9, r9, r4, lsl #8
+
+ ; store values
+ str r9, [r3], r2
+ str r9, [r3], r2
+ str r9, [r3], r2
+ str r9, [r3]
+
+ pop {r4-r12, pc}
+
+
+b_he_pred
+ ldrb r4, [r1], r2 ; Left[0]
+ ldrb r8, [sp, #48] ; top_left
+ ldrb r5, [r1], r2 ; Left[1]
+ ldrb r6, [r1], r2 ; Left[2]
+ ldrb r7, [r1] ; Left[3]
+
+ add r8, r8, r4 ; tl + l[0]
+ add r9, r4, r5 ; l[0] + l[1]
+ add r10, r5, r6 ; l[1] + l[2]
+ add r11, r6, r7 ; l[2] + l[3]
+
+ mov r0, #2<<14
+
+ add r8, r8, r9 ; tl + 2*l[0] + l[1]
+ add r4, r9, r10 ; l[0] + 2*l[1] + l[2]
+ add r5, r10, r11 ; l[1] + 2*l[2] + l[3]
+ add r6, r11, r7, lsl #1 ; l[2] + 2*l[3] + l[3]
+
+
+ add r8, r0, r8, lsl #14 ; (tl + 2*l[0] + l[1])>>2 in top half
+ add r9, r0, r4, lsl #14 ; (l[0] + 2*l[1] + l[2])>>2 in top half
+ add r10,r0, r5, lsl #14 ; (l[1] + 2*l[2] + l[3])>>2 in top half
+ add r11,r0, r6, lsl #14 ; (l[2] + 2*l[3] + l[3])>>2 in top half
+
+ pkhtb r8, r8, r8, asr #16 ; l[-|0|-|0]
+ pkhtb r9, r9, r9, asr #16 ; l[-|1|-|1]
+ pkhtb r10, r10, r10, asr #16 ; l[-|2|-|2]
+ pkhtb r11, r11, r11, asr #16 ; l[-|3|-|3]
+
+ ldr r0, [sp, #44] ; dst_stride
+ ldr r3, [sp, #40] ; dst
+
+ add r8, r8, r8, lsl #8 ; l[0|0|0|0]
+ add r9, r9, r9, lsl #8 ; l[1|1|1|1]
+ add r10, r10, r10, lsl #8 ; l[2|2|2|2]
+ add r11, r11, r11, lsl #8 ; l[3|3|3|3]
+
+ ; store values
+ str r8, [r3], r0
+ str r9, [r3], r0
+ str r10, [r3], r0
+ str r11, [r3]
+
+ pop {r4-r12, pc}
+
+b_ld_pred
+ ldr r4, [r0] ; Above[0-3]
+ ldr r12, c00020002
+ ldr r5, [r0, #4] ; Above[4-7]
+ ldr lr, c00FF00FF
+
+ uxtb16 r6, r4 ; a[2|0]
+ uxtb16 r7, r4, ror #8 ; a[3|1]
+ uxtb16 r8, r5 ; a[6|4]
+ uxtb16 r9, r5, ror #8 ; a[7|5]
+ pkhtb r10, r6, r8 ; a[2|4]
+ pkhtb r11, r7, r9 ; a[3|5]
+
+ add r4, r6, r7, lsl #1 ; [a2+2*a3 | a0+2*a1]
+ add r4, r4, r10, ror #16 ; [a2+2*a3+a4 | a0+2*a1+a2]
+ uxtab16 r4, r4, r12 ; [a2+2*a3+a4+2 | a0+2*a1+a2+2]
+
+ add r5, r7, r10, ror #15 ; [a3+2*a4 | a1+2*a2]
+ add r5, r5, r11, ror #16 ; [a3+2*a4+a5 | a1+2*a2+a3]
+ uxtab16 r5, r5, r12 ; [a3+2*a4+a5+2 | a1+2*a2+a3+2]
+
+ pkhtb r7, r9, r8, asr #16
+ add r6, r8, r9, lsl #1 ; [a6+2*a7 | a4+2*a5]
+ uadd16 r6, r6, r7 ; [a6+2*a7+a7 | a4+2*a5+a6]
+ uxtab16 r6, r6, r12 ; [a6+2*a7+a7+2 | a4+2*a5+a6+2]
+
+ uxth r7, r9 ; [ a5]
+ add r7, r7, r8, asr #15 ; [ a5+2*a6]
+ add r7, r7, r9, asr #16 ; [ a5+2*a6+a7]
+ uxtah r7, r7, r12 ; [ a5+2*a6+a7+2]
+
+ ldr r0, [sp, #44] ; dst_stride
+ ldr r3, [sp, #40] ; dst
+
+ ; scale down
+ and r4, lr, r4, asr #2
+ and r5, lr, r5, asr #2
+ and r6, lr, r6, asr #2
+ mov r7, r7, asr #2
+
+ add r8, r4, r5, lsl #8 ; [3|2|1|0]
+ str r8, [r3], r0
+
+ mov r9, r8, lsr #8
+ add r9, r9, r6, lsl #24 ; [4|3|2|1]
+ str r9, [r3], r0
+
+ mov r10, r9, lsr #8
+ add r10, r10, r7, lsl #24 ; [5|4|3|2]
+ str r10, [r3], r0
+
+ mov r6, r6, lsr #16
+ mov r11, r10, lsr #8
+ add r11, r11, r6, lsl #24 ; [6|5|4|3]
+ str r11, [r3]
+
+ pop {r4-r12, pc}
+
+b_rd_pred
+ ldrb r7, [r1], r2 ; l[0] = pp[3]
+ ldr lr, [r0] ; Above = pp[8|7|6|5]
+ ldrb r8, [sp, #48] ; tl = pp[4]
+ ldrb r6, [r1], r2 ; l[1] = pp[2]
+ ldrb r5, [r1], r2 ; l[2] = pp[1]
+ ldrb r4, [r1], r2 ; l[3] = pp[0]
+
+
+ uxtb16 r9, lr ; p[7|5]
+ uxtb16 r10, lr, ror #8 ; p[8|6]
+ add r4, r4, r6, lsl #16 ; p[2|0]
+ add r5, r5, r7, lsl #16 ; p[3|1]
+ add r6, r6, r8, lsl #16 ; p[4|2]
+ pkhbt r7, r7, r9, lsl #16 ; p[5|3]
+ pkhbt r8, r8, r10, lsl #16 ; p[6|4]
+
+ ldr r12, c00020002
+ ldr lr, c00FF00FF
+
+ add r4, r4, r5, lsl #1 ; [p2+2*p3 | p0+2*p1]
+ add r4, r4, r6 ; [p2+2*p3+p4 | p0+2*p1+p2]
+ uxtab16 r4, r4, r12 ; [p2+2*p3+p4+2 | p0+2*p1+p2+2]
+
+ add r5, r5, r6, lsl #1 ; [p3+2*p4 | p1+2*p2]
+ add r5, r5, r7 ; [p3+2*p4+p5 | p1+2*p2+p3]
+ uxtab16 r5, r5, r12 ; [p3+2*p4+p5+2 | p1+2*p2+p3+2]
+
+ add r6, r7, r8, lsl #1 ; [p5+2*p6 | p3+2*p4]
+ add r6, r6, r9 ; [p5+2*p6+p7 | p3+2*p4+p5]
+ uxtab16 r6, r6, r12 ; [p5+2*p6+p7+2 | p3+2*p4+p5+2]
+
+ add r7, r8, r9, lsl #1 ; [p6+2*p7 | p4+2*p5]
+ add r7, r7, r10 ; [p6+2*p7+p8 | p4+2*p5+p6]
+ uxtab16 r7, r7, r12 ; [p6+2*p7+p8+2 | p4+2*p5+p6+2]
+
+ ldr r0, [sp, #44] ; dst_stride
+ ldr r3, [sp, #40] ; dst
+
+ ; scale down
+ and r7, lr, r7, asr #2
+ and r6, lr, r6, asr #2
+ and r5, lr, r5, asr #2
+ and r4, lr, r4, asr #2
+
+ add r8, r6, r7, lsl #8 ; [6|5|4|3]
+ str r8, [r3], r0
+
+ mov r9, r8, lsl #8 ; [5|4|3|-]
+ uxtab r9, r9, r4, ror #16 ; [5|4|3|2]
+ str r9, [r3], r0
+
+ mov r10, r9, lsl #8 ; [4|3|2|-]
+ uxtab r10, r10, r5 ; [4|3|2|1]
+ str r10, [r3], r0
+
+ mov r11, r10, lsl #8 ; [3|2|1|-]
+ uxtab r11, r11, r4 ; [3|2|1|0]
+ str r11, [r3]
+
+ pop {r4-r12, pc}
+
+b_vr_pred
+ ldrb r7, [r1], r2 ; l[0] = pp[3]
+ ldr lr, [r0] ; Above = pp[8|7|6|5]
+ ldrb r8, [sp, #48] ; tl = pp[4]
+ ldrb r6, [r1], r2 ; l[1] = pp[2]
+ ldrb r5, [r1], r2 ; l[2] = pp[1]
+ ldrb r4, [r1] ; l[3] = pp[0]
+
+ add r5, r5, r7, lsl #16 ; p[3|1]
+ add r6, r6, r8, lsl #16 ; p[4|2]
+ uxtb16 r9, lr ; p[7|5]
+ uxtb16 r10, lr, ror #8 ; p[8|6]
+ pkhbt r7, r7, r9, lsl #16 ; p[5|3]
+ pkhbt r8, r8, r10, lsl #16 ; p[6|4]
+
+ ldr r4, c00010001
+ ldr r12, c00020002
+ ldr lr, c00FF00FF
+
+ add r5, r5, r6, lsl #1 ; [p3+2*p4 | p1+2*p2]
+ add r5, r5, r7 ; [p3+2*p4+p5 | p1+2*p2+p3]
+ uxtab16 r5, r5, r12 ; [p3+2*p4+p5+2 | p1+2*p2+p3+2]
+
+ add r6, r6, r7, lsl #1 ; [p4+2*p5 | p2+2*p3]
+ add r6, r6, r8 ; [p4+2*p5+p6 | p2+2*p3+p4]
+ uxtab16 r6, r6, r12 ; [p4+2*p5+p6+2 | p2+2*p3+p4+2]
+
+ uadd16 r11, r8, r9 ; [p6+p7 | p4+p5]
+ uhadd16 r11, r11, r4 ; [(p6+p7+1)>>1 | (p4+p5+1)>>1]
+ ; [F|E]
+
+ add r7, r7, r8, lsl #1 ; [p5+2*p6 | p3+2*p4]
+ add r7, r7, r9 ; [p5+2*p6+p7 | p3+2*p4+p5]
+ uxtab16 r7, r7, r12 ; [p5+2*p6+p7+2 | p3+2*p4+p5+2]
+
+ uadd16 r2, r9, r10 ; [p7+p8 | p5+p6]
+ uhadd16 r2, r2, r4 ; [(p7+p8+1)>>1 | (p5+p6+1)>>1]
+ ; [J|I]
+
+ add r8, r8, r9, lsl #1 ; [p6+2*p7 | p4+2*p5]
+ add r8, r8, r10 ; [p6+2*p7+p8 | p4+2*p5+p6]
+ uxtab16 r8, r8, r12 ; [p6+2*p7+p8+2 | p4+2*p5+p6+2]
+
+ ldr r0, [sp, #44] ; dst_stride
+ ldr r3, [sp, #40] ; dst
+
+ ; scale down
+ and r5, lr, r5, asr #2 ; [B|A]
+ and r6, lr, r6, asr #2 ; [D|C]
+ and r7, lr, r7, asr #2 ; [H|G]
+ and r8, lr, r8, asr #2 ; [L|K]
+
+ add r12, r11, r2, lsl #8 ; [J|F|I|E]
+ str r12, [r3], r0
+
+ add r12, r7, r8, lsl #8 ; [L|H|K|G]
+ str r12, [r3], r0
+
+ pkhbt r2, r6, r2, lsl #16 ; [-|I|-|C]
+ add r2, r2, r11, lsl #8 ; [F|I|E|C]
+
+ pkhtb r12, r6, r5 ; [-|D|-|A]
+ pkhtb r10, r7, r5, asr #16 ; [-|H|-|B]
+ str r2, [r3], r0
+ add r12, r12, r10, lsl #8 ; [H|D|B|A]
+ str r12, [r3]
+
+ pop {r4-r12, pc}
+
+b_vl_pred
+ ldr r4, [r0] ; [3|2|1|0] = Above[0-3]
+ ldr r12, c00020002
+ ldr r5, [r0, #4] ; [7|6|5|4] = Above[4-7]
+ ldr lr, c00FF00FF
+ ldr r2, c00010001
+
+ mov r0, r4, lsr #16 ; [-|-|3|2]
+ add r0, r0, r5, lsl #16 ; [5|4|3|2]
+ uxtb16 r6, r4 ; [2|0]
+ uxtb16 r7, r4, ror #8 ; [3|1]
+ uxtb16 r8, r0 ; [4|2]
+ uxtb16 r9, r0, ror #8 ; [5|3]
+ uxtb16 r10, r5 ; [6|4]
+ uxtb16 r11, r5, ror #8 ; [7|5]
+
+ uadd16 r4, r6, r7 ; [p2+p3 | p0+p1]
+ uhadd16 r4, r4, r2 ; [(p2+p3+1)>>1 | (p0+p1+1)>>1]
+ ; [B|A]
+
+ add r5, r6, r7, lsl #1 ; [p2+2*p3 | p0+2*p1]
+ add r5, r5, r8 ; [p2+2*p3+p4 | p0+2*p1+p2]
+ uxtab16 r5, r5, r12 ; [p2+2*p3+p4+2 | p0+2*p1+p2+2]
+
+ uadd16 r6, r7, r8 ; [p3+p4 | p1+p2]
+ uhadd16 r6, r6, r2 ; [(p3+p4+1)>>1 | (p1+p2+1)>>1]
+ ; [F|E]
+
+ add r7, r7, r8, lsl #1 ; [p3+2*p4 | p1+2*p2]
+ add r7, r7, r9 ; [p3+2*p4+p5 | p1+2*p2+p3]
+ uxtab16 r7, r7, r12 ; [p3+2*p4+p5+2 | p1+2*p2+p3+2]
+
+ add r8, r8, r9, lsl #1 ; [p4+2*p5 | p2+2*p3]
+ add r8, r8, r10 ; [p4+2*p5+p6 | p2+2*p3+p4]
+ uxtab16 r8, r8, r12 ; [p4+2*p5+p6+2 | p2+2*p3+p4+2]
+
+ add r9, r9, r10, lsl #1 ; [p5+2*p6 | p3+2*p4]
+ add r9, r9, r11 ; [p5+2*p6+p7 | p3+2*p4+p5]
+ uxtab16 r9, r9, r12 ; [p5+2*p6+p7+2 | p3+2*p4+p5+2]
+
+ ldr r0, [sp, #44] ; dst_stride
+ ldr r3, [sp, #40] ; dst
+
+ ; scale down
+ and r5, lr, r5, asr #2 ; [D|C]
+ and r7, lr, r7, asr #2 ; [H|G]
+ and r8, lr, r8, asr #2 ; [I|D]
+ and r9, lr, r9, asr #2 ; [J|H]
+
+ add r10, r4, r6, lsl #8 ; [F|B|E|A]
+ str r10, [r3], r0
+
+ add r5, r5, r7, lsl #8 ; [H|C|G|D]
+ str r5, [r3], r0
+
+ pkhtb r12, r8, r4, asr #16 ; [-|I|-|B]
+ pkhtb r10, r9, r8 ; [-|J|-|D]
+
+ add r12, r6, r12, lsl #8 ; [I|F|B|E]
+ str r12, [r3], r0
+
+ add r10, r7, r10, lsl #8 ; [J|H|D|G]
+ str r10, [r3]
+
+ pop {r4-r12, pc}
+
+b_hd_pred
+ ldrb r7, [r1], r2 ; l[0] = pp[3]
+ ldr lr, [r0] ; Above = pp[8|7|6|5]
+ ldrb r8, [sp, #48] ; tl = pp[4]
+ ldrb r6, [r1], r2 ; l[1] = pp[2]
+ ldrb r5, [r1], r2 ; l[2] = pp[1]
+ ldrb r4, [r1] ; l[3] = pp[0]
+
+ uxtb16 r9, lr ; p[7|5]
+ uxtb16 r10, lr, ror #8 ; p[8|6]
+
+ add r4, r4, r5, lsl #16 ; p[1|0]
+ add r5, r5, r6, lsl #16 ; p[2|1]
+ add r6, r6, r7, lsl #16 ; p[3|2]
+ add r7, r7, r8, lsl #16 ; p[4|3]
+
+ ldr r12, c00020002
+ ldr lr, c00FF00FF
+ ldr r2, c00010001
+
+ pkhtb r8, r7, r9 ; p[4|5]
+ pkhtb r1, r9, r10 ; p[7|6]
+ pkhbt r10, r8, r10, lsl #16 ; p[6|5]
+
+ uadd16 r11, r4, r5 ; [p1+p2 | p0+p1]
+ uhadd16 r11, r11, r2 ; [(p1+p2+1)>>1 | (p0+p1+1)>>1]
+ ; [B|A]
+
+ add r4, r4, r5, lsl #1 ; [p1+2*p2 | p0+2*p1]
+ add r4, r4, r6 ; [p1+2*p2+p3 | p0+2*p1+p2]
+ uxtab16 r4, r4, r12 ; [p1+2*p2+p3+2 | p0+2*p1+p2+2]
+
+ uadd16 r0, r6, r7 ; [p3+p4 | p2+p3]
+ uhadd16 r0, r0, r2 ; [(p3+p4+1)>>1 | (p2+p3+1)>>1]
+ ; [F|E]
+
+ add r5, r6, r7, lsl #1 ; [p3+2*p4 | p2+2*p3]
+ add r5, r5, r8, ror #16 ; [p3+2*p4+p5 | p2+2*p3+p4]
+ uxtab16 r5, r5, r12 ; [p3+2*p4+p5+2 | p2+2*p3+p4+2]
+
+ add r6, r12, r8, ror #16 ; [p5+2 | p4+2]
+ add r6, r6, r10, lsl #1 ; [p5+2+2*p6 | p4+2+2*p5]
+ uxtab16 r6, r6, r1 ; [p5+2+2*p6+p7 | p4+2+2*p5+p6]
+
+ ; scale down
+ and r4, lr, r4, asr #2 ; [D|C]
+ and r5, lr, r5, asr #2 ; [H|G]
+ and r6, lr, r6, asr #2 ; [J|I]
+
+ ldr lr, [sp, #44] ; dst_stride
+ ldr r3, [sp, #40] ; dst
+
+ pkhtb r2, r0, r6 ; [-|F|-|I]
+ pkhtb r12, r6, r5, asr #16 ; [-|J|-|H]
+ add r12, r12, r2, lsl #8 ; [F|J|I|H]
+ add r2, r0, r5, lsl #8 ; [H|F|G|E]
+ mov r12, r12, ror #24 ; [J|I|H|F]
+ str r12, [r3], lr
+
+ mov r7, r11, asr #16 ; [-|-|-|B]
+ str r2, [r3], lr
+ add r7, r7, r0, lsl #16 ; [-|E|-|B]
+ add r7, r7, r4, asr #8 ; [-|E|D|B]
+ add r7, r7, r5, lsl #24 ; [G|E|D|B]
+ str r7, [r3], lr
+
+ add r5, r11, r4, lsl #8 ; [D|B|C|A]
+ str r5, [r3]
+
+ pop {r4-r12, pc}
+
+
+
+b_hu_pred
+ ldrb r4, [r1], r2 ; Left[0]
+ ldr r12, c00020002
+ ldrb r5, [r1], r2 ; Left[1]
+ ldr lr, c00FF00FF
+ ldrb r6, [r1], r2 ; Left[2]
+ ldr r2, c00010001
+ ldrb r7, [r1] ; Left[3]
+
+ add r4, r4, r5, lsl #16 ; [1|0]
+ add r5, r5, r6, lsl #16 ; [2|1]
+ add r9, r6, r7, lsl #16 ; [3|2]
+
+ uadd16 r8, r4, r5 ; [p1+p2 | p0+p1]
+ uhadd16 r8, r8, r2 ; [(p1+p2+1)>>1 | (p0+p1+1)>>1]
+ ; [B|A]
+
+ add r4, r4, r5, lsl #1 ; [p1+2*p2 | p0+2*p1]
+ add r4, r4, r9 ; [p1+2*p2+p3 | p0+2*p1+p2]
+ uxtab16 r4, r4, r12 ; [p1+2*p2+p3+2 | p0+2*p1+p2+2]
+ ldr r2, [sp, #44] ; dst_stride
+ ldr r3, [sp, #40] ; dst
+ and r4, lr, r4, asr #2 ; [D|C]
+
+ add r10, r6, r7 ; [p2+p3]
+ add r11, r10, r7, lsl #1 ; [p2+3*p3]
+ add r10, r10, #1
+ add r11, r11, #2
+ mov r10, r10, asr #1 ; [E]
+ mov r11, r11, asr #2 ; [F]
+
+ add r9, r7, r9, asr #8 ; [-|-|G|G]
+ add r0, r8, r4, lsl #8 ; [D|B|C|A]
+ add r7, r9, r9, lsl #16 ; [G|G|G|G]
+
+ str r0, [r3], r2
+
+ mov r1, r8, asr #16 ; [-|-|-|B]
+ add r1, r1, r4, asr #8 ; [-|-|D|B]
+ add r1, r1, r10, lsl #16 ; [-|E|D|B]
+ add r1, r1, r11, lsl #24 ; [F|E|D|B]
+ str r1, [r3], r2
+
+ add r10, r11, lsl #8 ; [-|-|F|E]
+ add r10, r10, r9, lsl #16 ; [G|G|F|E]
+ str r10, [r3], r2
+
+ str r7, [r3]
+
+ pop {r4-r12, pc}
+
+ ENDP
+
+; constants
+c00010001
+ DCD 0x00010001
+c00020002
+ DCD 0x00020002
+c00FF00FF
+ DCD 0x00FF00FF
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/iwalsh_v6.asm b/libvpx/vp8/common/arm/armv6/iwalsh_v6.asm
new file mode 100644
index 0000000..31ef09c
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/iwalsh_v6.asm
@@ -0,0 +1,136 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+ EXPORT |vp8_short_inv_walsh4x4_v6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA |.text|, CODE, READONLY ; name this block of code
+
+;short vp8_short_inv_walsh4x4_v6(short *input, short *mb_dqcoeff)
+|vp8_short_inv_walsh4x4_v6| PROC
+
+ stmdb sp!, {r4 - r12, lr}
+
+ ldr r2, [r0, #0] ; [1 | 0]
+ ldr r3, [r0, #4] ; [3 | 2]
+ ldr r4, [r0, #8] ; [5 | 4]
+ ldr r5, [r0, #12] ; [7 | 6]
+ ldr r6, [r0, #16] ; [9 | 8]
+ ldr r7, [r0, #20] ; [11 | 10]
+ ldr r8, [r0, #24] ; [13 | 12]
+ ldr r9, [r0, #28] ; [15 | 14]
+
+ qadd16 r10, r2, r8 ; a1 [1+13 | 0+12]
+ qadd16 r11, r4, r6 ; b1 [5+9 | 4+8]
+ qsub16 r12, r4, r6 ; c1 [5-9 | 4-8]
+ qsub16 lr, r2, r8 ; d1 [1-13 | 0-12]
+
+ qadd16 r2, r10, r11 ; a1 + b1 [1 | 0]
+ qadd16 r4, r12, lr ; c1 + d1 [5 | 4]
+ qsub16 r6, r10, r11 ; a1 - b1 [9 | 8]
+ qsub16 r8, lr, r12 ; d1 - c1 [13 | 12]
+
+ qadd16 r10, r3, r9 ; a1 [3+15 | 2+14]
+ qadd16 r11, r5, r7 ; b1 [7+11 | 6+10]
+ qsub16 r12, r5, r7 ; c1 [7-11 | 6-10]
+ qsub16 lr, r3, r9 ; d1 [3-15 | 2-14]
+
+ qadd16 r3, r10, r11 ; a1 + b1 [3 | 2]
+ qadd16 r5, r12, lr ; c1 + d1 [7 | 6]
+ qsub16 r7, r10, r11 ; a1 - b1 [11 | 10]
+ qsub16 r9, lr, r12 ; d1 - c1 [15 | 14]
+
+ ; first transform complete
+
+ qsubaddx r10, r2, r3 ; [c1|a1] [1-2 | 0+3]
+ qaddsubx r11, r2, r3 ; [b1|d1] [1+2 | 0-3]
+ qsubaddx r12, r4, r5 ; [c1|a1] [5-6 | 4+7]
+ qaddsubx lr, r4, r5 ; [b1|d1] [5+6 | 4-7]
+
+ qaddsubx r2, r10, r11 ; [b2|c2] [c1+d1 | a1-b1]
+ qaddsubx r3, r11, r10 ; [a2|d2] [b1+a1 | d1-c1]
+ ldr r10, c0x00030003
+ qaddsubx r4, r12, lr ; [b2|c2] [c1+d1 | a1-b1]
+ qaddsubx r5, lr, r12 ; [a2|d2] [b1+a1 | d1-c1]
+
+ qadd16 r2, r2, r10 ; [b2+3|c2+3]
+ qadd16 r3, r3, r10 ; [a2+3|d2+3]
+ qadd16 r4, r4, r10 ; [b2+3|c2+3]
+ qadd16 r5, r5, r10 ; [a2+3|d2+3]
+
+ asr r12, r3, #19 ; [0]
+ strh r12, [r1], #32
+ asr lr, r2, #19 ; [1]
+ strh lr, [r1], #32
+ sxth r2, r2
+ sxth r3, r3
+ asr r2, r2, #3 ; [2]
+ strh r2, [r1], #32
+ asr r3, r3, #3 ; [3]
+ strh r3, [r1], #32
+
+ asr r12, r5, #19 ; [4]
+ strh r12, [r1], #32
+ asr lr, r4, #19 ; [5]
+ strh lr, [r1], #32
+ sxth r4, r4
+ sxth r5, r5
+ asr r4, r4, #3 ; [6]
+ strh r4, [r1], #32
+ asr r5, r5, #3 ; [7]
+ strh r5, [r1], #32
+
+ qsubaddx r2, r6, r7 ; [c1|a1] [9-10 | 8+11]
+ qaddsubx r3, r6, r7 ; [b1|d1] [9+10 | 8-11]
+ qsubaddx r4, r8, r9 ; [c1|a1] [13-14 | 12+15]
+ qaddsubx r5, r8, r9 ; [b1|d1] [13+14 | 12-15]
+
+ qaddsubx r6, r2, r3 ; [b2|c2] [c1+d1 | a1-b1]
+ qaddsubx r7, r3, r2 ; [a2|d2] [b1+a1 | d1-c1]
+ qaddsubx r8, r4, r5 ; [b2|c2] [c1+d1 | a1-b1]
+ qaddsubx r9, r5, r4 ; [a2|d2] [b1+a1 | d1-c1]
+
+ qadd16 r6, r6, r10 ; [b2+3|c2+3]
+ qadd16 r7, r7, r10 ; [a2+3|d2+3]
+ qadd16 r8, r8, r10 ; [b2+3|c2+3]
+ qadd16 r9, r9, r10 ; [a2+3|d2+3]
+
+ asr r12, r7, #19 ; [8]
+ strh r12, [r1], #32
+ asr lr, r6, #19 ; [9]
+ strh lr, [r1], #32
+ sxth r6, r6
+ sxth r7, r7
+ asr r6, r6, #3 ; [10]
+ strh r6, [r1], #32
+ asr r7, r7, #3 ; [11]
+ strh r7, [r1], #32
+
+ asr r12, r9, #19 ; [12]
+ strh r12, [r1], #32
+ asr lr, r8, #19 ; [13]
+ strh lr, [r1], #32
+ sxth r8, r8
+ sxth r9, r9
+ asr r8, r8, #3 ; [14]
+ strh r8, [r1], #32
+ asr r9, r9, #3 ; [15]
+ strh r9, [r1], #32
+
+ ldmia sp!, {r4 - r12, pc}
+ ENDP ; |vp8_short_inv_walsh4x4_v6|
+
+
+; Constant Pool
+c0x00030003 DCD 0x00030003
+ END
diff --git a/libvpx/vp8/common/arm/armv6/loopfilter_v6.asm b/libvpx/vp8/common/arm/armv6/loopfilter_v6.asm
new file mode 100644
index 0000000..1cbbbcd
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/loopfilter_v6.asm
@@ -0,0 +1,1282 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_loop_filter_horizontal_edge_armv6|
+ EXPORT |vp8_mbloop_filter_horizontal_edge_armv6|
+ EXPORT |vp8_loop_filter_vertical_edge_armv6|
+ EXPORT |vp8_mbloop_filter_vertical_edge_armv6|
+
+ AREA |.text|, CODE, READONLY ; name this block of code
+
+ MACRO
+ TRANSPOSE_MATRIX $a0, $a1, $a2, $a3, $b0, $b1, $b2, $b3
+ ; input: $a0, $a1, $a2, $a3; output: $b0, $b1, $b2, $b3
+ ; a0: 03 02 01 00
+ ; a1: 13 12 11 10
+ ; a2: 23 22 21 20
+ ; a3: 33 32 31 30
+ ; b3 b2 b1 b0
+
+ uxtb16 $b1, $a1 ; xx 12 xx 10
+ uxtb16 $b0, $a0 ; xx 02 xx 00
+ uxtb16 $b3, $a3 ; xx 32 xx 30
+ uxtb16 $b2, $a2 ; xx 22 xx 20
+ orr $b1, $b0, $b1, lsl #8 ; 12 02 10 00
+ orr $b3, $b2, $b3, lsl #8 ; 32 22 30 20
+
+ uxtb16 $a1, $a1, ror #8 ; xx 13 xx 11
+ uxtb16 $a3, $a3, ror #8 ; xx 33 xx 31
+ uxtb16 $a0, $a0, ror #8 ; xx 03 xx 01
+ uxtb16 $a2, $a2, ror #8 ; xx 23 xx 21
+ orr $a0, $a0, $a1, lsl #8 ; 13 03 11 01
+ orr $a2, $a2, $a3, lsl #8 ; 33 23 31 21
+
+ pkhtb $b2, $b3, $b1, asr #16 ; 32 22 12 02 -- p1
+ pkhbt $b0, $b1, $b3, lsl #16 ; 30 20 10 00 -- p3
+
+ pkhtb $b3, $a2, $a0, asr #16 ; 33 23 13 03 -- p0
+ pkhbt $b1, $a0, $a2, lsl #16 ; 31 21 11 01 -- p2
+ MEND
+
+
+src RN r0
+pstep RN r1
+count RN r5
+
+;r0 unsigned char *src_ptr,
+;r1 int src_pixel_step,
+;r2 const char *blimit,
+;r3 const char *limit,
+;stack const char *thresh,
+;stack int count
+
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+|vp8_loop_filter_horizontal_edge_armv6| PROC
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+ stmdb sp!, {r4 - r11, lr}
+
+ sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
+ ldr count, [sp, #40] ; count for 8-in-parallel
+ ldr r6, [sp, #36] ; load thresh address
+ sub sp, sp, #16 ; create temp buffer
+
+ ldr r9, [src], pstep ; p3
+ ldrb r4, [r2] ; blimit
+ ldr r10, [src], pstep ; p2
+ ldrb r2, [r3] ; limit
+ ldr r11, [src], pstep ; p1
+ orr r4, r4, r4, lsl #8
+ ldrb r3, [r6] ; thresh
+ orr r2, r2, r2, lsl #8
+ mov count, count, lsl #1 ; 4-in-parallel
+ orr r4, r4, r4, lsl #16
+ orr r3, r3, r3, lsl #8
+ orr r2, r2, r2, lsl #16
+ orr r3, r3, r3, lsl #16
+
+|Hnext8|
+ ; vp8_filter_mask() function
+ ; calculate breakout conditions
+ ldr r12, [src], pstep ; p0
+
+ uqsub8 r6, r9, r10 ; p3 - p2
+ uqsub8 r7, r10, r9 ; p2 - p3
+ uqsub8 r8, r10, r11 ; p2 - p1
+ uqsub8 r10, r11, r10 ; p1 - p2
+
+ orr r6, r6, r7 ; abs (p3-p2)
+ orr r8, r8, r10 ; abs (p2-p1)
+ uqsub8 lr, r6, r2 ; compare to limit. lr: vp8_filter_mask
+ uqsub8 r8, r8, r2 ; compare to limit
+ uqsub8 r6, r11, r12 ; p1 - p0
+ orr lr, lr, r8
+ uqsub8 r7, r12, r11 ; p0 - p1
+ ldr r9, [src], pstep ; q0
+ ldr r10, [src], pstep ; q1
+ orr r6, r6, r7 ; abs (p1-p0)
+ uqsub8 r7, r6, r2 ; compare to limit
+ uqsub8 r8, r6, r3 ; compare to thresh -- save r8 for later
+ orr lr, lr, r7
+
+ uqsub8 r6, r11, r10 ; p1 - q1
+ uqsub8 r7, r10, r11 ; q1 - p1
+ uqsub8 r11, r12, r9 ; p0 - q0
+ uqsub8 r12, r9, r12 ; q0 - p0
+ orr r6, r6, r7 ; abs (p1-q1)
+ ldr r7, c0x7F7F7F7F
+ orr r12, r11, r12 ; abs (p0-q0)
+ ldr r11, [src], pstep ; q2
+ uqadd8 r12, r12, r12 ; abs (p0-q0) * 2
+ and r6, r7, r6, lsr #1 ; abs (p1-q1) / 2
+ uqsub8 r7, r9, r10 ; q0 - q1
+ uqadd8 r12, r12, r6 ; abs (p0-q0)*2 + abs (p1-q1)/2
+ uqsub8 r6, r10, r9 ; q1 - q0
+ uqsub8 r12, r12, r4 ; compare to flimit
+ uqsub8 r9, r11, r10 ; q2 - q1
+
+ orr lr, lr, r12
+
+ ldr r12, [src], pstep ; q3
+ uqsub8 r10, r10, r11 ; q1 - q2
+ orr r6, r7, r6 ; abs (q1-q0)
+ orr r10, r9, r10 ; abs (q2-q1)
+ uqsub8 r7, r6, r2 ; compare to limit
+ uqsub8 r10, r10, r2 ; compare to limit
+ uqsub8 r6, r6, r3 ; compare to thresh -- save r6 for later
+ orr lr, lr, r7
+ orr lr, lr, r10
+
+ uqsub8 r10, r12, r11 ; q3 - q2
+ uqsub8 r9, r11, r12 ; q2 - q3
+
+ mvn r11, #0 ; r11 == -1
+
+ orr r10, r10, r9 ; abs (q3-q2)
+ uqsub8 r10, r10, r2 ; compare to limit
+
+ mov r12, #0
+ orr lr, lr, r10
+ sub src, src, pstep, lsl #2
+
+ usub8 lr, r12, lr ; use usub8 instead of ssub8
+ sel lr, r11, r12 ; filter mask: lr
+
+ cmp lr, #0
+ beq hskip_filter ; skip filtering
+
+ sub src, src, pstep, lsl #1 ; move src pointer down by 6 lines
+
+ ;vp8_hevmask() function
+ ;calculate high edge variance
+ orr r10, r6, r8 ; calculate vp8_hevmask
+
+ ldr r7, [src], pstep ; p1
+
+ usub8 r10, r12, r10 ; use usub8 instead of ssub8
+ sel r6, r12, r11 ; obtain vp8_hevmask: r6
+
+ ;vp8_filter() function
+ ldr r8, [src], pstep ; p0
+ ldr r12, c0x80808080
+ ldr r9, [src], pstep ; q0
+ ldr r10, [src], pstep ; q1
+
+ eor r7, r7, r12 ; p1 offset to convert to a signed value
+ eor r8, r8, r12 ; p0 offset to convert to a signed value
+ eor r9, r9, r12 ; q0 offset to convert to a signed value
+ eor r10, r10, r12 ; q1 offset to convert to a signed value
+
+ str r9, [sp] ; store qs0 temporarily
+ str r8, [sp, #4] ; store ps0 temporarily
+ str r10, [sp, #8] ; store qs1 temporarily
+ str r7, [sp, #12] ; store ps1 temporarily
+
+ qsub8 r7, r7, r10 ; vp8_signed_char_clamp(ps1-qs1)
+ qsub8 r8, r9, r8 ; vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
+
+ and r7, r7, r6 ; vp8_filter (r7) &= hev
+
+ qadd8 r7, r7, r8
+ ldr r9, c0x03030303 ; r9 = 3 --modified for vp8
+
+ qadd8 r7, r7, r8
+ ldr r10, c0x04040404
+
+ qadd8 r7, r7, r8
+ and r7, r7, lr ; vp8_filter &= mask;
+
+ ;modify code for vp8 -- Filter1 = vp8_filter (r7)
+ qadd8 r8 , r7 , r9 ; Filter2 (r8) = vp8_signed_char_clamp(vp8_filter+3)
+ qadd8 r7 , r7 , r10 ; vp8_filter = vp8_signed_char_clamp(vp8_filter+4)
+
+ mov r9, #0
+ shadd8 r8 , r8 , r9 ; Filter2 >>= 3
+ shadd8 r7 , r7 , r9 ; vp8_filter >>= 3
+ shadd8 r8 , r8 , r9
+ shadd8 r7 , r7 , r9
+ shadd8 lr , r8 , r9 ; lr: Filter2
+ shadd8 r7 , r7 , r9 ; r7: filter
+
+ ;usub8 lr, r8, r10 ; s = (s==4)*-1
+ ;sel lr, r11, r9
+ ;usub8 r8, r10, r8
+ ;sel r8, r11, r9
+ ;and r8, r8, lr ; -1 for each element that equals 4
+
+ ;calculate output
+ ;qadd8 lr, r8, r7 ; u = vp8_signed_char_clamp(s + vp8_filter)
+
+ ldr r8, [sp] ; load qs0
+ ldr r9, [sp, #4] ; load ps0
+
+ ldr r10, c0x01010101
+
+ qsub8 r8 ,r8, r7 ; u = vp8_signed_char_clamp(qs0 - vp8_filter)
+ qadd8 r9, r9, lr ; u = vp8_signed_char_clamp(ps0 + Filter2)
+
+ ;end of modification for vp8
+
+ mov lr, #0
+ sadd8 r7, r7 , r10 ; vp8_filter += 1
+ shadd8 r7, r7, lr ; vp8_filter >>= 1
+
+ ldr r11, [sp, #12] ; load ps1
+ ldr r10, [sp, #8] ; load qs1
+
+ bic r7, r7, r6 ; vp8_filter &= ~hev
+ sub src, src, pstep, lsl #2
+
+ qadd8 r11, r11, r7 ; u = vp8_signed_char_clamp(ps1 + vp8_filter)
+ qsub8 r10, r10,r7 ; u = vp8_signed_char_clamp(qs1 - vp8_filter)
+
+ eor r11, r11, r12 ; *op1 = u^0x80
+ str r11, [src], pstep ; store op1
+ eor r9, r9, r12 ; *op0 = u^0x80
+ str r9, [src], pstep ; store op0 result
+ eor r8, r8, r12 ; *oq0 = u^0x80
+ str r8, [src], pstep ; store oq0 result
+ eor r10, r10, r12 ; *oq1 = u^0x80
+ str r10, [src], pstep ; store oq1
+
+ sub src, src, pstep, lsl #1
+
+|hskip_filter|
+ add src, src, #4
+ sub src, src, pstep, lsl #2
+
+ subs count, count, #1
+
+ ldrne r9, [src], pstep ; p3
+ ldrne r10, [src], pstep ; p2
+ ldrne r11, [src], pstep ; p1
+
+ bne Hnext8
+
+ add sp, sp, #16
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_loop_filter_horizontal_edge_armv6|
+
+
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+|vp8_mbloop_filter_horizontal_edge_armv6| PROC
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+ stmdb sp!, {r4 - r11, lr}
+
+ sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
+ ldr count, [sp, #40] ; count for 8-in-parallel
+ ldr r6, [sp, #36] ; load thresh address
+ sub sp, sp, #16 ; create temp buffer
+
+ ldr r9, [src], pstep ; p3
+ ldrb r4, [r2] ; blimit
+ ldr r10, [src], pstep ; p2
+ ldrb r2, [r3] ; limit
+ ldr r11, [src], pstep ; p1
+ orr r4, r4, r4, lsl #8
+ ldrb r3, [r6] ; thresh
+ orr r2, r2, r2, lsl #8
+ mov count, count, lsl #1 ; 4-in-parallel
+ orr r4, r4, r4, lsl #16
+ orr r3, r3, r3, lsl #8
+ orr r2, r2, r2, lsl #16
+ orr r3, r3, r3, lsl #16
+
+|MBHnext8|
+
+ ; vp8_filter_mask() function
+ ; calculate breakout conditions
+ ldr r12, [src], pstep ; p0
+
+ uqsub8 r6, r9, r10 ; p3 - p2
+ uqsub8 r7, r10, r9 ; p2 - p3
+ uqsub8 r8, r10, r11 ; p2 - p1
+ uqsub8 r10, r11, r10 ; p1 - p2
+
+ orr r6, r6, r7 ; abs (p3-p2)
+ orr r8, r8, r10 ; abs (p2-p1)
+ uqsub8 lr, r6, r2 ; compare to limit. lr: vp8_filter_mask
+ uqsub8 r8, r8, r2 ; compare to limit
+
+ uqsub8 r6, r11, r12 ; p1 - p0
+ orr lr, lr, r8
+ uqsub8 r7, r12, r11 ; p0 - p1
+ ldr r9, [src], pstep ; q0
+ ldr r10, [src], pstep ; q1
+ orr r6, r6, r7 ; abs (p1-p0)
+ uqsub8 r7, r6, r2 ; compare to limit
+ uqsub8 r8, r6, r3 ; compare to thresh -- save r8 for later
+ orr lr, lr, r7
+
+ uqsub8 r6, r11, r10 ; p1 - q1
+ uqsub8 r7, r10, r11 ; q1 - p1
+ uqsub8 r11, r12, r9 ; p0 - q0
+ uqsub8 r12, r9, r12 ; q0 - p0
+ orr r6, r6, r7 ; abs (p1-q1)
+ ldr r7, c0x7F7F7F7F
+ orr r12, r11, r12 ; abs (p0-q0)
+ ldr r11, [src], pstep ; q2
+ uqadd8 r12, r12, r12 ; abs (p0-q0) * 2
+ and r6, r7, r6, lsr #1 ; abs (p1-q1) / 2
+ uqsub8 r7, r9, r10 ; q0 - q1
+ uqadd8 r12, r12, r6 ; abs (p0-q0)*2 + abs (p1-q1)/2
+ uqsub8 r6, r10, r9 ; q1 - q0
+ uqsub8 r12, r12, r4 ; compare to flimit
+ uqsub8 r9, r11, r10 ; q2 - q1
+
+ orr lr, lr, r12
+
+ ldr r12, [src], pstep ; q3
+
+ uqsub8 r10, r10, r11 ; q1 - q2
+ orr r6, r7, r6 ; abs (q1-q0)
+ orr r10, r9, r10 ; abs (q2-q1)
+ uqsub8 r7, r6, r2 ; compare to limit
+ uqsub8 r10, r10, r2 ; compare to limit
+ uqsub8 r6, r6, r3 ; compare to thresh -- save r6 for later
+ orr lr, lr, r7
+ orr lr, lr, r10
+
+ uqsub8 r10, r12, r11 ; q3 - q2
+ uqsub8 r9, r11, r12 ; q2 - q3
+
+ mvn r11, #0 ; r11 == -1
+
+ orr r10, r10, r9 ; abs (q3-q2)
+ uqsub8 r10, r10, r2 ; compare to limit
+
+ mov r12, #0
+
+ orr lr, lr, r10
+
+ usub8 lr, r12, lr ; use usub8 instead of ssub8
+ sel lr, r11, r12 ; filter mask: lr
+
+ cmp lr, #0
+ beq mbhskip_filter ; skip filtering
+
+ ;vp8_hevmask() function
+ ;calculate high edge variance
+ sub src, src, pstep, lsl #2 ; move src pointer down by 6 lines
+ sub src, src, pstep, lsl #1
+
+ orr r10, r6, r8
+ ldr r7, [src], pstep ; p1
+
+ usub8 r10, r12, r10
+ sel r6, r12, r11 ; hev mask: r6
+
+ ;vp8_mbfilter() function
+ ;p2, q2 are only needed at the end. Don't need to load them in now.
+ ldr r8, [src], pstep ; p0
+ ldr r12, c0x80808080
+ ldr r9, [src], pstep ; q0
+ ldr r10, [src] ; q1
+
+ eor r7, r7, r12 ; ps1
+ eor r8, r8, r12 ; ps0
+ eor r9, r9, r12 ; qs0
+ eor r10, r10, r12 ; qs1
+
+ qsub8 r12, r9, r8 ; vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
+ str r7, [sp, #12] ; store ps1 temporarily
+ qsub8 r7, r7, r10 ; vp8_signed_char_clamp(ps1-qs1)
+ str r10, [sp, #8] ; store qs1 temporarily
+ qadd8 r7, r7, r12
+ str r9, [sp] ; store qs0 temporarily
+ qadd8 r7, r7, r12
+ str r8, [sp, #4] ; store ps0 temporarily
+ qadd8 r7, r7, r12 ; vp8_filter: r7
+
+ ldr r10, c0x03030303 ; r10 = 3 --modified for vp8
+ ldr r9, c0x04040404
+
+ and r7, r7, lr ; vp8_filter &= mask (lr is free)
+
+ mov r12, r7 ; Filter2: r12
+ and r12, r12, r6 ; Filter2 &= hev
+
+ ;modify code for vp8
+ ;save bottom 3 bits so that we round one side +4 and the other +3
+ qadd8 r8 , r12 , r9 ; Filter1 (r8) = vp8_signed_char_clamp(Filter2+4)
+ qadd8 r12 , r12 , r10 ; Filter2 (r12) = vp8_signed_char_clamp(Filter2+3)
+
+ mov r10, #0
+ shadd8 r8 , r8 , r10 ; Filter1 >>= 3
+ shadd8 r12 , r12 , r10 ; Filter2 >>= 3
+ shadd8 r8 , r8 , r10
+ shadd8 r12 , r12 , r10
+ shadd8 r8 , r8 , r10 ; r8: Filter1
+ shadd8 r12 , r12 , r10 ; r12: Filter2
+
+ ldr r9, [sp] ; load qs0
+ ldr r11, [sp, #4] ; load ps0
+
+ qsub8 r9 , r9, r8 ; qs0 = vp8_signed_char_clamp(qs0 - Filter1)
+ qadd8 r11, r11, r12 ; ps0 = vp8_signed_char_clamp(ps0 + Filter2)
+
+ ;save bottom 3 bits so that we round one side +4 and the other +3
+ ;and r8, r12, r10 ; s = Filter2 & 7 (s: r8)
+ ;qadd8 r12 , r12 , r9 ; Filter2 = vp8_signed_char_clamp(Filter2+4)
+ ;mov r10, #0
+ ;shadd8 r12 , r12 , r10 ; Filter2 >>= 3
+ ;usub8 lr, r8, r9 ; s = (s==4)*-1
+ ;sel lr, r11, r10
+ ;shadd8 r12 , r12 , r10
+ ;usub8 r8, r9, r8
+ ;sel r8, r11, r10
+ ;ldr r9, [sp] ; load qs0
+ ;ldr r11, [sp, #4] ; load ps0
+ ;shadd8 r12 , r12 , r10
+ ;and r8, r8, lr ; -1 for each element that equals 4
+ ;qadd8 r10, r8, r12 ; u = vp8_signed_char_clamp(s + Filter2)
+ ;qsub8 r9 , r9, r12 ; qs0 = vp8_signed_char_clamp(qs0 - Filter2)
+ ;qadd8 r11, r11, r10 ; ps0 = vp8_signed_char_clamp(ps0 + u)
+
+ ;end of modification for vp8
+
+ bic r12, r7, r6 ; vp8_filter &= ~hev ( r6 is free)
+ ;mov r12, r7
+
+ ;roughly 3/7th difference across boundary
+ mov lr, #0x1b ; 27
+ mov r7, #0x3f ; 63
+
+ sxtb16 r6, r12
+ sxtb16 r10, r12, ror #8
+ smlabb r8, r6, lr, r7
+ smlatb r6, r6, lr, r7
+ smlabb r7, r10, lr, r7
+ smultb r10, r10, lr
+ ssat r8, #8, r8, asr #7
+ ssat r6, #8, r6, asr #7
+ add r10, r10, #63
+ ssat r7, #8, r7, asr #7
+ ssat r10, #8, r10, asr #7
+
+ ldr lr, c0x80808080
+
+ pkhbt r6, r8, r6, lsl #16
+ pkhbt r10, r7, r10, lsl #16
+ uxtb16 r6, r6
+ uxtb16 r10, r10
+
+ sub src, src, pstep
+
+ orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
+
+ qsub8 r8, r9, r10 ; s = vp8_signed_char_clamp(qs0 - u)
+ qadd8 r10, r11, r10 ; s = vp8_signed_char_clamp(ps0 + u)
+ eor r8, r8, lr ; *oq0 = s^0x80
+ str r8, [src] ; store *oq0
+ sub src, src, pstep
+ eor r10, r10, lr ; *op0 = s^0x80
+ str r10, [src] ; store *op0
+
+ ;roughly 2/7th difference across boundary
+ mov lr, #0x12 ; 18
+ mov r7, #0x3f ; 63
+
+ sxtb16 r6, r12
+ sxtb16 r10, r12, ror #8
+ smlabb r8, r6, lr, r7
+ smlatb r6, r6, lr, r7
+ smlabb r9, r10, lr, r7
+ smlatb r10, r10, lr, r7
+ ssat r8, #8, r8, asr #7
+ ssat r6, #8, r6, asr #7
+ ssat r9, #8, r9, asr #7
+ ssat r10, #8, r10, asr #7
+
+ ldr lr, c0x80808080
+
+ pkhbt r6, r8, r6, lsl #16
+ pkhbt r10, r9, r10, lsl #16
+
+ ldr r9, [sp, #8] ; load qs1
+ ldr r11, [sp, #12] ; load ps1
+
+ uxtb16 r6, r6
+ uxtb16 r10, r10
+
+ sub src, src, pstep
+
+ orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
+
+ qadd8 r11, r11, r10 ; s = vp8_signed_char_clamp(ps1 + u)
+ qsub8 r8, r9, r10 ; s = vp8_signed_char_clamp(qs1 - u)
+ eor r11, r11, lr ; *op1 = s^0x80
+ str r11, [src], pstep ; store *op1
+ eor r8, r8, lr ; *oq1 = s^0x80
+ add src, src, pstep, lsl #1
+
+ mov r7, #0x3f ; 63
+
+ str r8, [src], pstep ; store *oq1
+
+ ;roughly 1/7th difference across boundary
+ mov lr, #0x9 ; 9
+ ldr r9, [src] ; load q2
+
+ sxtb16 r6, r12
+ sxtb16 r10, r12, ror #8
+ smlabb r8, r6, lr, r7
+ smlatb r6, r6, lr, r7
+ smlabb r12, r10, lr, r7
+ smlatb r10, r10, lr, r7
+ ssat r8, #8, r8, asr #7
+ ssat r6, #8, r6, asr #7
+ ssat r12, #8, r12, asr #7
+ ssat r10, #8, r10, asr #7
+
+ sub src, src, pstep, lsl #2
+
+ pkhbt r6, r8, r6, lsl #16
+ pkhbt r10, r12, r10, lsl #16
+
+ sub src, src, pstep
+ ldr lr, c0x80808080
+
+ ldr r11, [src] ; load p2
+
+ uxtb16 r6, r6
+ uxtb16 r10, r10
+
+ eor r9, r9, lr
+ eor r11, r11, lr
+
+ orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
+
+ qadd8 r8, r11, r10 ; s = vp8_signed_char_clamp(ps2 + u)
+ qsub8 r10, r9, r10 ; s = vp8_signed_char_clamp(qs2 - u)
+ eor r8, r8, lr ; *op2 = s^0x80
+ str r8, [src], pstep, lsl #2 ; store *op2
+ add src, src, pstep
+ eor r10, r10, lr ; *oq2 = s^0x80
+ str r10, [src], pstep, lsl #1 ; store *oq2
+
+|mbhskip_filter|
+ add src, src, #4
+ sub src, src, pstep, lsl #3
+ subs count, count, #1
+
+ ldrne r9, [src], pstep ; p3
+ ldrne r10, [src], pstep ; p2
+ ldrne r11, [src], pstep ; p1
+
+ bne MBHnext8
+
+ add sp, sp, #16
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_mbloop_filter_horizontal_edge_armv6|
+
+
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+|vp8_loop_filter_vertical_edge_armv6| PROC
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+ stmdb sp!, {r4 - r11, lr}
+
+ sub src, src, #4 ; move src pointer down by 4
+ ldr count, [sp, #40] ; count for 8-in-parallel
+ ldr r12, [sp, #36] ; load thresh address
+ sub sp, sp, #16 ; create temp buffer
+
+ ldr r6, [src], pstep ; load source data
+ ldrb r4, [r2] ; blimit
+ ldr r7, [src], pstep
+ ldrb r2, [r3] ; limit
+ ldr r8, [src], pstep
+ orr r4, r4, r4, lsl #8
+ ldrb r3, [r12] ; thresh
+ orr r2, r2, r2, lsl #8
+ ldr lr, [src], pstep
+ mov count, count, lsl #1 ; 4-in-parallel
+ orr r4, r4, r4, lsl #16
+ orr r3, r3, r3, lsl #8
+ orr r2, r2, r2, lsl #16
+ orr r3, r3, r3, lsl #16
+
+|Vnext8|
+
+ ; vp8_filter_mask() function
+ ; calculate breakout conditions
+ ; transpose the source data for 4-in-parallel operation
+ TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
+
+ uqsub8 r7, r9, r10 ; p3 - p2
+ uqsub8 r8, r10, r9 ; p2 - p3
+ uqsub8 r9, r10, r11 ; p2 - p1
+ uqsub8 r10, r11, r10 ; p1 - p2
+ orr r7, r7, r8 ; abs (p3-p2)
+ orr r10, r9, r10 ; abs (p2-p1)
+ uqsub8 lr, r7, r2 ; compare to limit. lr: vp8_filter_mask
+ uqsub8 r10, r10, r2 ; compare to limit
+
+ sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
+
+ orr lr, lr, r10
+
+ uqsub8 r6, r11, r12 ; p1 - p0
+ uqsub8 r7, r12, r11 ; p0 - p1
+ add src, src, #4 ; move src pointer up by 4
+ orr r6, r6, r7 ; abs (p1-p0)
+ str r11, [sp, #12] ; save p1
+ uqsub8 r10, r6, r2 ; compare to limit
+ uqsub8 r11, r6, r3 ; compare to thresh
+ orr lr, lr, r10
+
+ ; transpose uses 8 regs(r6 - r12 and lr). Need to save reg value now
+ ; transpose the source data for 4-in-parallel operation
+ ldr r6, [src], pstep ; load source data
+ str r11, [sp] ; push r11 to stack
+ ldr r7, [src], pstep
+ str r12, [sp, #4] ; save current reg before load q0 - q3 data
+ ldr r8, [src], pstep
+ str lr, [sp, #8]
+ ldr lr, [src], pstep
+
+ TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
+
+ ldr lr, [sp, #8] ; load back (f)limit accumulator
+
+ uqsub8 r6, r12, r11 ; q3 - q2
+ uqsub8 r7, r11, r12 ; q2 - q3
+ uqsub8 r12, r11, r10 ; q2 - q1
+ uqsub8 r11, r10, r11 ; q1 - q2
+ orr r6, r6, r7 ; abs (q3-q2)
+ orr r7, r12, r11 ; abs (q2-q1)
+ uqsub8 r6, r6, r2 ; compare to limit
+ uqsub8 r7, r7, r2 ; compare to limit
+ ldr r11, [sp, #4] ; load back p0
+ ldr r12, [sp, #12] ; load back p1
+ orr lr, lr, r6
+ orr lr, lr, r7
+
+ uqsub8 r6, r11, r9 ; p0 - q0
+ uqsub8 r7, r9, r11 ; q0 - p0
+ uqsub8 r8, r12, r10 ; p1 - q1
+ uqsub8 r11, r10, r12 ; q1 - p1
+ orr r6, r6, r7 ; abs (p0-q0)
+ ldr r7, c0x7F7F7F7F
+ orr r8, r8, r11 ; abs (p1-q1)
+ uqadd8 r6, r6, r6 ; abs (p0-q0) * 2
+ and r8, r7, r8, lsr #1 ; abs (p1-q1) / 2
+ uqsub8 r11, r10, r9 ; q1 - q0
+ uqadd8 r6, r8, r6 ; abs (p0-q0)*2 + abs (p1-q1)/2
+ uqsub8 r12, r9, r10 ; q0 - q1
+ uqsub8 r6, r6, r4 ; compare to flimit
+
+ orr r9, r11, r12 ; abs (q1-q0)
+ uqsub8 r8, r9, r2 ; compare to limit
+ uqsub8 r10, r9, r3 ; compare to thresh
+ orr lr, lr, r6
+ orr lr, lr, r8
+
+ mvn r11, #0 ; r11 == -1
+ mov r12, #0
+
+ usub8 lr, r12, lr
+ ldr r9, [sp] ; load the compared result
+ sel lr, r11, r12 ; filter mask: lr
+
+ cmp lr, #0
+ beq vskip_filter ; skip filtering
+
+ ;vp8_hevmask() function
+ ;calculate high edge variance
+
+ sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
+
+ orr r9, r9, r10
+
+ ldrh r7, [src, #-2]
+ ldrh r8, [src], pstep
+
+ usub8 r9, r12, r9
+ sel r6, r12, r11 ; hev mask: r6
+
+ ;vp8_filter() function
+ ; load soure data to r6, r11, r12, lr
+ ldrh r9, [src, #-2]
+ ldrh r10, [src], pstep
+
+ pkhbt r12, r7, r8, lsl #16
+
+ ldrh r7, [src, #-2]
+ ldrh r8, [src], pstep
+
+ pkhbt r11, r9, r10, lsl #16
+
+ ldrh r9, [src, #-2]
+ ldrh r10, [src], pstep
+
+ ; Transpose needs 8 regs(r6 - r12, and lr). Save r6 and lr first
+ str r6, [sp]
+ str lr, [sp, #4]
+
+ pkhbt r6, r7, r8, lsl #16
+ pkhbt lr, r9, r10, lsl #16
+
+ ;transpose r12, r11, r6, lr to r7, r8, r9, r10
+ TRANSPOSE_MATRIX r12, r11, r6, lr, r7, r8, r9, r10
+
+ ;load back hev_mask r6 and filter_mask lr
+ ldr r12, c0x80808080
+ ldr r6, [sp]
+ ldr lr, [sp, #4]
+
+ eor r7, r7, r12 ; p1 offset to convert to a signed value
+ eor r8, r8, r12 ; p0 offset to convert to a signed value
+ eor r9, r9, r12 ; q0 offset to convert to a signed value
+ eor r10, r10, r12 ; q1 offset to convert to a signed value
+
+ str r9, [sp] ; store qs0 temporarily
+ str r8, [sp, #4] ; store ps0 temporarily
+ str r10, [sp, #8] ; store qs1 temporarily
+ str r7, [sp, #12] ; store ps1 temporarily
+
+ qsub8 r7, r7, r10 ; vp8_signed_char_clamp(ps1-qs1)
+ qsub8 r8, r9, r8 ; vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
+
+ and r7, r7, r6 ; vp8_filter (r7) &= hev (r7 : filter)
+
+ qadd8 r7, r7, r8
+ ldr r9, c0x03030303 ; r9 = 3 --modified for vp8
+
+ qadd8 r7, r7, r8
+ ldr r10, c0x04040404
+
+ qadd8 r7, r7, r8
+ ;mvn r11, #0 ; r11 == -1
+
+ and r7, r7, lr ; vp8_filter &= mask
+
+ ;modify code for vp8 -- Filter1 = vp8_filter (r7)
+ qadd8 r8 , r7 , r9 ; Filter2 (r8) = vp8_signed_char_clamp(vp8_filter+3)
+ qadd8 r7 , r7 , r10 ; vp8_filter = vp8_signed_char_clamp(vp8_filter+4)
+
+ mov r9, #0
+ shadd8 r8 , r8 , r9 ; Filter2 >>= 3
+ shadd8 r7 , r7 , r9 ; vp8_filter >>= 3
+ shadd8 r8 , r8 , r9
+ shadd8 r7 , r7 , r9
+ shadd8 lr , r8 , r9 ; lr: filter2
+ shadd8 r7 , r7 , r9 ; r7: filter
+
+ ;usub8 lr, r8, r10 ; s = (s==4)*-1
+ ;sel lr, r11, r9
+ ;usub8 r8, r10, r8
+ ;sel r8, r11, r9
+ ;and r8, r8, lr ; -1 for each element that equals 4 -- r8: s
+
+ ;calculate output
+ ;qadd8 lr, r8, r7 ; u = vp8_signed_char_clamp(s + vp8_filter)
+
+ ldr r8, [sp] ; load qs0
+ ldr r9, [sp, #4] ; load ps0
+
+ ldr r10, c0x01010101
+
+ qsub8 r8, r8, r7 ; u = vp8_signed_char_clamp(qs0 - vp8_filter)
+ qadd8 r9, r9, lr ; u = vp8_signed_char_clamp(ps0 + Filter2)
+ ;end of modification for vp8
+
+ eor r8, r8, r12
+ eor r9, r9, r12
+
+ mov lr, #0
+
+ sadd8 r7, r7, r10
+ shadd8 r7, r7, lr
+
+ ldr r10, [sp, #8] ; load qs1
+ ldr r11, [sp, #12] ; load ps1
+
+ bic r7, r7, r6 ; r7: vp8_filter
+
+ qsub8 r10 , r10, r7 ; u = vp8_signed_char_clamp(qs1 - vp8_filter)
+ qadd8 r11, r11, r7 ; u = vp8_signed_char_clamp(ps1 + vp8_filter)
+ eor r10, r10, r12
+ eor r11, r11, r12
+
+ sub src, src, pstep, lsl #2
+
+ ;we can use TRANSPOSE_MATRIX macro to transpose output - input: q1, q0, p0, p1
+ ;output is b0, b1, b2, b3
+ ;b0: 03 02 01 00
+ ;b1: 13 12 11 10
+ ;b2: 23 22 21 20
+ ;b3: 33 32 31 30
+ ; p1 p0 q0 q1
+ ; (a3 a2 a1 a0)
+ TRANSPOSE_MATRIX r11, r9, r8, r10, r6, r7, r12, lr
+
+ strh r6, [src, #-2] ; store the result
+ mov r6, r6, lsr #16
+ strh r6, [src], pstep
+
+ strh r7, [src, #-2]
+ mov r7, r7, lsr #16
+ strh r7, [src], pstep
+
+ strh r12, [src, #-2]
+ mov r12, r12, lsr #16
+ strh r12, [src], pstep
+
+ strh lr, [src, #-2]
+ mov lr, lr, lsr #16
+ strh lr, [src], pstep
+
+|vskip_filter|
+ sub src, src, #4
+ subs count, count, #1
+
+ ldrne r6, [src], pstep ; load source data
+ ldrne r7, [src], pstep
+ ldrne r8, [src], pstep
+ ldrne lr, [src], pstep
+
+ bne Vnext8
+
+ add sp, sp, #16
+
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_loop_filter_vertical_edge_armv6|
+
+
+
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+|vp8_mbloop_filter_vertical_edge_armv6| PROC
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+ stmdb sp!, {r4 - r11, lr}
+
+ sub src, src, #4 ; move src pointer down by 4
+ ldr count, [sp, #40] ; count for 8-in-parallel
+ ldr r12, [sp, #36] ; load thresh address
+ pld [src, #23] ; preload for next block
+ sub sp, sp, #16 ; create temp buffer
+
+ ldr r6, [src], pstep ; load source data
+ ldrb r4, [r2] ; blimit
+ pld [src, #23]
+ ldr r7, [src], pstep
+ ldrb r2, [r3] ; limit
+ pld [src, #23]
+ ldr r8, [src], pstep
+ orr r4, r4, r4, lsl #8
+ ldrb r3, [r12] ; thresh
+ orr r2, r2, r2, lsl #8
+ pld [src, #23]
+ ldr lr, [src], pstep
+ mov count, count, lsl #1 ; 4-in-parallel
+ orr r4, r4, r4, lsl #16
+ orr r3, r3, r3, lsl #8
+ orr r2, r2, r2, lsl #16
+ orr r3, r3, r3, lsl #16
+
+|MBVnext8|
+ ; vp8_filter_mask() function
+ ; calculate breakout conditions
+ ; transpose the source data for 4-in-parallel operation
+ TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
+
+ uqsub8 r7, r9, r10 ; p3 - p2
+ uqsub8 r8, r10, r9 ; p2 - p3
+ uqsub8 r9, r10, r11 ; p2 - p1
+ uqsub8 r10, r11, r10 ; p1 - p2
+ orr r7, r7, r8 ; abs (p3-p2)
+ orr r10, r9, r10 ; abs (p2-p1)
+ uqsub8 lr, r7, r2 ; compare to limit. lr: vp8_filter_mask
+ uqsub8 r10, r10, r2 ; compare to limit
+
+ sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
+
+ orr lr, lr, r10
+
+ uqsub8 r6, r11, r12 ; p1 - p0
+ uqsub8 r7, r12, r11 ; p0 - p1
+ add src, src, #4 ; move src pointer up by 4
+ orr r6, r6, r7 ; abs (p1-p0)
+ str r11, [sp, #12] ; save p1
+ uqsub8 r10, r6, r2 ; compare to limit
+ uqsub8 r11, r6, r3 ; compare to thresh
+ orr lr, lr, r10
+
+ ; transpose uses 8 regs(r6 - r12 and lr). Need to save reg value now
+ ; transpose the source data for 4-in-parallel operation
+ ldr r6, [src], pstep ; load source data
+ str r11, [sp] ; push r11 to stack
+ ldr r7, [src], pstep
+ str r12, [sp, #4] ; save current reg before load q0 - q3 data
+ ldr r8, [src], pstep
+ str lr, [sp, #8]
+ ldr lr, [src], pstep
+
+
+ TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
+
+ ldr lr, [sp, #8] ; load back (f)limit accumulator
+
+ uqsub8 r6, r12, r11 ; q3 - q2
+ uqsub8 r7, r11, r12 ; q2 - q3
+ uqsub8 r12, r11, r10 ; q2 - q1
+ uqsub8 r11, r10, r11 ; q1 - q2
+ orr r6, r6, r7 ; abs (q3-q2)
+ orr r7, r12, r11 ; abs (q2-q1)
+ uqsub8 r6, r6, r2 ; compare to limit
+ uqsub8 r7, r7, r2 ; compare to limit
+ ldr r11, [sp, #4] ; load back p0
+ ldr r12, [sp, #12] ; load back p1
+ orr lr, lr, r6
+ orr lr, lr, r7
+
+ uqsub8 r6, r11, r9 ; p0 - q0
+ uqsub8 r7, r9, r11 ; q0 - p0
+ uqsub8 r8, r12, r10 ; p1 - q1
+ uqsub8 r11, r10, r12 ; q1 - p1
+ orr r6, r6, r7 ; abs (p0-q0)
+ ldr r7, c0x7F7F7F7F
+ orr r8, r8, r11 ; abs (p1-q1)
+ uqadd8 r6, r6, r6 ; abs (p0-q0) * 2
+ and r8, r7, r8, lsr #1 ; abs (p1-q1) / 2
+ uqsub8 r11, r10, r9 ; q1 - q0
+ uqadd8 r6, r8, r6 ; abs (p0-q0)*2 + abs (p1-q1)/2
+ uqsub8 r12, r9, r10 ; q0 - q1
+ uqsub8 r6, r6, r4 ; compare to flimit
+
+ orr r9, r11, r12 ; abs (q1-q0)
+ uqsub8 r8, r9, r2 ; compare to limit
+ uqsub8 r10, r9, r3 ; compare to thresh
+ orr lr, lr, r6
+ orr lr, lr, r8
+
+ mvn r11, #0 ; r11 == -1
+ mov r12, #0
+
+ usub8 lr, r12, lr
+ ldr r9, [sp] ; load the compared result
+ sel lr, r11, r12 ; filter mask: lr
+
+ cmp lr, #0
+ beq mbvskip_filter ; skip filtering
+
+
+
+ ;vp8_hevmask() function
+ ;calculate high edge variance
+
+ sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
+
+ orr r9, r9, r10
+
+ ldrh r7, [src, #-2]
+ ldrh r8, [src], pstep
+
+ usub8 r9, r12, r9
+ sel r6, r12, r11 ; hev mask: r6
+
+
+ ; vp8_mbfilter() function
+ ; p2, q2 are only needed at the end. Don't need to load them in now.
+ ; Transpose needs 8 regs(r6 - r12, and lr). Save r6 and lr first
+ ; load soure data to r6, r11, r12, lr
+ ldrh r9, [src, #-2]
+ ldrh r10, [src], pstep
+
+ pkhbt r12, r7, r8, lsl #16
+
+ ldrh r7, [src, #-2]
+ ldrh r8, [src], pstep
+
+ pkhbt r11, r9, r10, lsl #16
+
+ ldrh r9, [src, #-2]
+ ldrh r10, [src], pstep
+
+ str r6, [sp] ; save r6
+ str lr, [sp, #4] ; save lr
+
+ pkhbt r6, r7, r8, lsl #16
+ pkhbt lr, r9, r10, lsl #16
+
+ ;transpose r12, r11, r6, lr to p1, p0, q0, q1
+ TRANSPOSE_MATRIX r12, r11, r6, lr, r7, r8, r9, r10
+
+ ;load back hev_mask r6 and filter_mask lr
+ ldr r12, c0x80808080
+ ldr r6, [sp]
+ ldr lr, [sp, #4]
+
+ eor r7, r7, r12 ; ps1
+ eor r8, r8, r12 ; ps0
+ eor r9, r9, r12 ; qs0
+ eor r10, r10, r12 ; qs1
+
+ qsub8 r12, r9, r8 ; vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
+ str r7, [sp, #12] ; store ps1 temporarily
+ qsub8 r7, r7, r10 ; vp8_signed_char_clamp(ps1-qs1)
+ str r10, [sp, #8] ; store qs1 temporarily
+ qadd8 r7, r7, r12
+ str r9, [sp] ; store qs0 temporarily
+ qadd8 r7, r7, r12
+ str r8, [sp, #4] ; store ps0 temporarily
+ qadd8 r7, r7, r12 ; vp8_filter: r7
+
+ ldr r10, c0x03030303 ; r10 = 3 --modified for vp8
+ ldr r9, c0x04040404
+ ;mvn r11, #0 ; r11 == -1
+
+ and r7, r7, lr ; vp8_filter &= mask (lr is free)
+
+ mov r12, r7 ; Filter2: r12
+ and r12, r12, r6 ; Filter2 &= hev
+
+ ;modify code for vp8
+ ;save bottom 3 bits so that we round one side +4 and the other +3
+ qadd8 r8 , r12 , r9 ; Filter1 (r8) = vp8_signed_char_clamp(Filter2+4)
+ qadd8 r12 , r12 , r10 ; Filter2 (r12) = vp8_signed_char_clamp(Filter2+3)
+
+ mov r10, #0
+ shadd8 r8 , r8 , r10 ; Filter1 >>= 3
+ shadd8 r12 , r12 , r10 ; Filter2 >>= 3
+ shadd8 r8 , r8 , r10
+ shadd8 r12 , r12 , r10
+ shadd8 r8 , r8 , r10 ; r8: Filter1
+ shadd8 r12 , r12 , r10 ; r12: Filter2
+
+ ldr r9, [sp] ; load qs0
+ ldr r11, [sp, #4] ; load ps0
+
+ qsub8 r9 , r9, r8 ; qs0 = vp8_signed_char_clamp(qs0 - Filter1)
+ qadd8 r11, r11, r12 ; ps0 = vp8_signed_char_clamp(ps0 + Filter2)
+
+ ;save bottom 3 bits so that we round one side +4 and the other +3
+ ;and r8, r12, r10 ; s = Filter2 & 7 (s: r8)
+ ;qadd8 r12 , r12 , r9 ; Filter2 = vp8_signed_char_clamp(Filter2+4)
+ ;mov r10, #0
+ ;shadd8 r12 , r12 , r10 ; Filter2 >>= 3
+ ;usub8 lr, r8, r9 ; s = (s==4)*-1
+ ;sel lr, r11, r10
+ ;shadd8 r12 , r12 , r10
+ ;usub8 r8, r9, r8
+ ;sel r8, r11, r10
+ ;ldr r9, [sp] ; load qs0
+ ;ldr r11, [sp, #4] ; load ps0
+ ;shadd8 r12 , r12 , r10
+ ;and r8, r8, lr ; -1 for each element that equals 4
+ ;qadd8 r10, r8, r12 ; u = vp8_signed_char_clamp(s + Filter2)
+ ;qsub8 r9 , r9, r12 ; qs0 = vp8_signed_char_clamp(qs0 - Filter2)
+ ;qadd8 r11, r11, r10 ; ps0 = vp8_signed_char_clamp(ps0 + u)
+
+ ;end of modification for vp8
+
+ bic r12, r7, r6 ;vp8_filter &= ~hev ( r6 is free)
+ ;mov r12, r7
+
+ ;roughly 3/7th difference across boundary
+ mov lr, #0x1b ; 27
+ mov r7, #0x3f ; 63
+
+ sxtb16 r6, r12
+ sxtb16 r10, r12, ror #8
+ smlabb r8, r6, lr, r7
+ smlatb r6, r6, lr, r7
+ smlabb r7, r10, lr, r7
+ smultb r10, r10, lr
+ ssat r8, #8, r8, asr #7
+ ssat r6, #8, r6, asr #7
+ add r10, r10, #63
+ ssat r7, #8, r7, asr #7
+ ssat r10, #8, r10, asr #7
+
+ ldr lr, c0x80808080
+
+ pkhbt r6, r8, r6, lsl #16
+ pkhbt r10, r7, r10, lsl #16
+ uxtb16 r6, r6
+ uxtb16 r10, r10
+
+ sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
+
+ orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
+
+ qsub8 r8, r9, r10 ; s = vp8_signed_char_clamp(qs0 - u)
+ qadd8 r10, r11, r10 ; s = vp8_signed_char_clamp(ps0 + u)
+ eor r8, r8, lr ; *oq0 = s^0x80
+ eor r10, r10, lr ; *op0 = s^0x80
+
+ strb r10, [src, #-1] ; store op0 result
+ strb r8, [src], pstep ; store oq0 result
+ mov r10, r10, lsr #8
+ mov r8, r8, lsr #8
+ strb r10, [src, #-1]
+ strb r8, [src], pstep
+ mov r10, r10, lsr #8
+ mov r8, r8, lsr #8
+ strb r10, [src, #-1]
+ strb r8, [src], pstep
+ mov r10, r10, lsr #8
+ mov r8, r8, lsr #8
+ strb r10, [src, #-1]
+ strb r8, [src], pstep
+
+ ;roughly 2/7th difference across boundary
+ mov lr, #0x12 ; 18
+ mov r7, #0x3f ; 63
+
+ sxtb16 r6, r12
+ sxtb16 r10, r12, ror #8
+ smlabb r8, r6, lr, r7
+ smlatb r6, r6, lr, r7
+ smlabb r9, r10, lr, r7
+
+ smlatb r10, r10, lr, r7
+ ssat r8, #8, r8, asr #7
+ ssat r6, #8, r6, asr #7
+ ssat r9, #8, r9, asr #7
+ ssat r10, #8, r10, asr #7
+
+ sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
+
+ pkhbt r6, r8, r6, lsl #16
+ pkhbt r10, r9, r10, lsl #16
+
+ ldr r9, [sp, #8] ; load qs1
+ ldr r11, [sp, #12] ; load ps1
+ ldr lr, c0x80808080
+
+ uxtb16 r6, r6
+ uxtb16 r10, r10
+
+ add src, src, #2
+
+ orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
+
+ qsub8 r8, r9, r10 ; s = vp8_signed_char_clamp(qs1 - u)
+ qadd8 r10, r11, r10 ; s = vp8_signed_char_clamp(ps1 + u)
+ eor r8, r8, lr ; *oq1 = s^0x80
+ eor r10, r10, lr ; *op1 = s^0x80
+
+ ldrb r11, [src, #-5] ; load p2 for 1/7th difference across boundary
+ strb r10, [src, #-4] ; store op1
+ strb r8, [src, #-1] ; store oq1
+ ldrb r9, [src], pstep ; load q2 for 1/7th difference across boundary
+
+ mov r10, r10, lsr #8
+ mov r8, r8, lsr #8
+
+ ldrb r6, [src, #-5]
+ strb r10, [src, #-4]
+ strb r8, [src, #-1]
+ ldrb r7, [src], pstep
+
+ mov r10, r10, lsr #8
+ mov r8, r8, lsr #8
+ orr r11, r11, r6, lsl #8
+ orr r9, r9, r7, lsl #8
+
+ ldrb r6, [src, #-5]
+ strb r10, [src, #-4]
+ strb r8, [src, #-1]
+ ldrb r7, [src], pstep
+
+ mov r10, r10, lsr #8
+ mov r8, r8, lsr #8
+ orr r11, r11, r6, lsl #16
+ orr r9, r9, r7, lsl #16
+
+ ldrb r6, [src, #-5]
+ strb r10, [src, #-4]
+ strb r8, [src, #-1]
+ ldrb r7, [src], pstep
+ orr r11, r11, r6, lsl #24
+ orr r9, r9, r7, lsl #24
+
+ ;roughly 1/7th difference across boundary
+ eor r9, r9, lr
+ eor r11, r11, lr
+
+ mov lr, #0x9 ; 9
+ mov r7, #0x3f ; 63
+
+ sxtb16 r6, r12
+ sxtb16 r10, r12, ror #8
+ smlabb r8, r6, lr, r7
+ smlatb r6, r6, lr, r7
+ smlabb r12, r10, lr, r7
+ smlatb r10, r10, lr, r7
+ ssat r8, #8, r8, asr #7
+ ssat r6, #8, r6, asr #7
+ ssat r12, #8, r12, asr #7
+ ssat r10, #8, r10, asr #7
+
+ sub src, src, pstep, lsl #2
+
+ pkhbt r6, r8, r6, lsl #16
+ pkhbt r10, r12, r10, lsl #16
+
+ uxtb16 r6, r6
+ uxtb16 r10, r10
+
+ ldr lr, c0x80808080
+
+ orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
+
+ qadd8 r8, r11, r10 ; s = vp8_signed_char_clamp(ps2 + u)
+ qsub8 r10, r9, r10 ; s = vp8_signed_char_clamp(qs2 - u)
+ eor r8, r8, lr ; *op2 = s^0x80
+ eor r10, r10, lr ; *oq2 = s^0x80
+
+ strb r8, [src, #-5] ; store *op2
+ strb r10, [src], pstep ; store *oq2
+ mov r8, r8, lsr #8
+ mov r10, r10, lsr #8
+ strb r8, [src, #-5]
+ strb r10, [src], pstep
+ mov r8, r8, lsr #8
+ mov r10, r10, lsr #8
+ strb r8, [src, #-5]
+ strb r10, [src], pstep
+ mov r8, r8, lsr #8
+ mov r10, r10, lsr #8
+ strb r8, [src, #-5]
+ strb r10, [src], pstep
+
+ ;adjust src pointer for next loop
+ sub src, src, #2
+
+|mbvskip_filter|
+ sub src, src, #4
+ subs count, count, #1
+
+ pld [src, #23] ; preload for next block
+ ldrne r6, [src], pstep ; load source data
+ pld [src, #23]
+ ldrne r7, [src], pstep
+ pld [src, #23]
+ ldrne r8, [src], pstep
+ pld [src, #23]
+ ldrne lr, [src], pstep
+
+ bne MBVnext8
+
+ add sp, sp, #16
+
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_mbloop_filter_vertical_edge_armv6|
+
+; Constant Pool
+c0x80808080 DCD 0x80808080
+c0x03030303 DCD 0x03030303
+c0x04040404 DCD 0x04040404
+c0x01010101 DCD 0x01010101
+c0x7F7F7F7F DCD 0x7F7F7F7F
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/simpleloopfilter_v6.asm b/libvpx/vp8/common/arm/armv6/simpleloopfilter_v6.asm
new file mode 100644
index 0000000..5e00cf0
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/simpleloopfilter_v6.asm
@@ -0,0 +1,286 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_loop_filter_simple_horizontal_edge_armv6|
+ EXPORT |vp8_loop_filter_simple_vertical_edge_armv6|
+
+ AREA |.text|, CODE, READONLY ; name this block of code
+
+ MACRO
+ TRANSPOSE_MATRIX $a0, $a1, $a2, $a3, $b0, $b1, $b2, $b3
+ ; input: $a0, $a1, $a2, $a3; output: $b0, $b1, $b2, $b3
+ ; a0: 03 02 01 00
+ ; a1: 13 12 11 10
+ ; a2: 23 22 21 20
+ ; a3: 33 32 31 30
+ ; b3 b2 b1 b0
+
+ uxtb16 $b1, $a1 ; xx 12 xx 10
+ uxtb16 $b0, $a0 ; xx 02 xx 00
+ uxtb16 $b3, $a3 ; xx 32 xx 30
+ uxtb16 $b2, $a2 ; xx 22 xx 20
+ orr $b1, $b0, $b1, lsl #8 ; 12 02 10 00
+ orr $b3, $b2, $b3, lsl #8 ; 32 22 30 20
+
+ uxtb16 $a1, $a1, ror #8 ; xx 13 xx 11
+ uxtb16 $a3, $a3, ror #8 ; xx 33 xx 31
+ uxtb16 $a0, $a0, ror #8 ; xx 03 xx 01
+ uxtb16 $a2, $a2, ror #8 ; xx 23 xx 21
+ orr $a0, $a0, $a1, lsl #8 ; 13 03 11 01
+ orr $a2, $a2, $a3, lsl #8 ; 33 23 31 21
+
+ pkhtb $b2, $b3, $b1, asr #16 ; 32 22 12 02 -- p1
+ pkhbt $b0, $b1, $b3, lsl #16 ; 30 20 10 00 -- p3
+
+ pkhtb $b3, $a2, $a0, asr #16 ; 33 23 13 03 -- p0
+ pkhbt $b1, $a0, $a2, lsl #16 ; 31 21 11 01 -- p2
+ MEND
+
+
+
+src RN r0
+pstep RN r1
+
+;r0 unsigned char *src_ptr,
+;r1 int src_pixel_step,
+;r2 const char *blimit
+
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+|vp8_loop_filter_simple_horizontal_edge_armv6| PROC
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+ stmdb sp!, {r4 - r11, lr}
+
+ ldrb r12, [r2] ; blimit
+ ldr r3, [src, -pstep, lsl #1] ; p1
+ ldr r4, [src, -pstep] ; p0
+ ldr r5, [src] ; q0
+ ldr r6, [src, pstep] ; q1
+ orr r12, r12, r12, lsl #8 ; blimit
+ ldr r2, c0x80808080
+ orr r12, r12, r12, lsl #16 ; blimit
+ mov r9, #4 ; double the count. we're doing 4 at a time
+ mov lr, #0 ; need 0 in a couple places
+
+|simple_hnext8|
+ ; vp8_simple_filter_mask()
+
+ uqsub8 r7, r3, r6 ; p1 - q1
+ uqsub8 r8, r6, r3 ; q1 - p1
+ uqsub8 r10, r4, r5 ; p0 - q0
+ uqsub8 r11, r5, r4 ; q0 - p0
+ orr r8, r8, r7 ; abs(p1 - q1)
+ orr r10, r10, r11 ; abs(p0 - q0)
+ uqadd8 r10, r10, r10 ; abs(p0 - q0) * 2
+ uhadd8 r8, r8, lr ; abs(p1 - q2) >> 1
+ uqadd8 r10, r10, r8 ; abs(p0 - q0)*2 + abs(p1 - q1)/2
+ mvn r8, #0
+ usub8 r10, r12, r10 ; compare to flimit. usub8 sets GE flags
+ sel r10, r8, lr ; filter mask: F or 0
+ cmp r10, #0
+ beq simple_hskip_filter ; skip filtering if all masks are 0x00
+
+ ;vp8_simple_filter()
+
+ eor r3, r3, r2 ; p1 offset to convert to a signed value
+ eor r6, r6, r2 ; q1 offset to convert to a signed value
+ eor r4, r4, r2 ; p0 offset to convert to a signed value
+ eor r5, r5, r2 ; q0 offset to convert to a signed value
+
+ qsub8 r3, r3, r6 ; vp8_filter = p1 - q1
+ qsub8 r6, r5, r4 ; q0 - p0
+ qadd8 r3, r3, r6 ; += q0 - p0
+ ldr r7, c0x04040404
+ qadd8 r3, r3, r6 ; += q0 - p0
+ ldr r8, c0x03030303
+ qadd8 r3, r3, r6 ; vp8_filter = p1-q1 + 3*(q0-p0))
+ ;STALL
+ and r3, r3, r10 ; vp8_filter &= mask
+
+ qadd8 r7 , r3 , r7 ; Filter1 = vp8_filter + 4
+ qadd8 r8 , r3 , r8 ; Filter2 = vp8_filter + 3
+
+ shadd8 r7 , r7 , lr
+ shadd8 r8 , r8 , lr
+ shadd8 r7 , r7 , lr
+ shadd8 r8 , r8 , lr
+ shadd8 r7 , r7 , lr ; Filter1 >>= 3
+ shadd8 r8 , r8 , lr ; Filter2 >>= 3
+
+ qsub8 r5 ,r5, r7 ; u = q0 - Filter1
+ qadd8 r4, r4, r8 ; u = p0 + Filter2
+ eor r5, r5, r2 ; *oq0 = u^0x80
+ str r5, [src] ; store oq0 result
+ eor r4, r4, r2 ; *op0 = u^0x80
+ str r4, [src, -pstep] ; store op0 result
+
+|simple_hskip_filter|
+ subs r9, r9, #1
+ addne src, src, #4 ; next row
+
+ ldrne r3, [src, -pstep, lsl #1] ; p1
+ ldrne r4, [src, -pstep] ; p0
+ ldrne r5, [src] ; q0
+ ldrne r6, [src, pstep] ; q1
+
+ bne simple_hnext8
+
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_loop_filter_simple_horizontal_edge_armv6|
+
+
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+|vp8_loop_filter_simple_vertical_edge_armv6| PROC
+;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
+ stmdb sp!, {r4 - r11, lr}
+
+ ldrb r12, [r2] ; r12: blimit
+ ldr r2, c0x80808080
+ orr r12, r12, r12, lsl #8
+
+ ; load soure data to r7, r8, r9, r10
+ ldrh r3, [src, #-2]
+ pld [src, #23] ; preload for next block
+ ldrh r4, [src], pstep
+ orr r12, r12, r12, lsl #16
+
+ ldrh r5, [src, #-2]
+ pld [src, #23]
+ ldrh r6, [src], pstep
+
+ pkhbt r7, r3, r4, lsl #16
+
+ ldrh r3, [src, #-2]
+ pld [src, #23]
+ ldrh r4, [src], pstep
+
+ pkhbt r8, r5, r6, lsl #16
+
+ ldrh r5, [src, #-2]
+ pld [src, #23]
+ ldrh r6, [src], pstep
+ mov r11, #4 ; double the count. we're doing 4 at a time
+
+|simple_vnext8|
+ ; vp8_simple_filter_mask() function
+ pkhbt r9, r3, r4, lsl #16
+ pkhbt r10, r5, r6, lsl #16
+
+ ;transpose r7, r8, r9, r10 to r3, r4, r5, r6
+ TRANSPOSE_MATRIX r7, r8, r9, r10, r3, r4, r5, r6
+
+ uqsub8 r7, r3, r6 ; p1 - q1
+ uqsub8 r8, r6, r3 ; q1 - p1
+ uqsub8 r9, r4, r5 ; p0 - q0
+ uqsub8 r10, r5, r4 ; q0 - p0
+ orr r7, r7, r8 ; abs(p1 - q1)
+ orr r9, r9, r10 ; abs(p0 - q0)
+ mov r8, #0
+ uqadd8 r9, r9, r9 ; abs(p0 - q0) * 2
+ uhadd8 r7, r7, r8 ; abs(p1 - q1) / 2
+ uqadd8 r7, r7, r9 ; abs(p0 - q0)*2 + abs(p1 - q1)/2
+ mvn r10, #0 ; r10 == -1
+
+ usub8 r7, r12, r7 ; compare to flimit
+ sel lr, r10, r8 ; filter mask
+
+ cmp lr, #0
+ beq simple_vskip_filter ; skip filtering
+
+ ;vp8_simple_filter() function
+ eor r3, r3, r2 ; p1 offset to convert to a signed value
+ eor r6, r6, r2 ; q1 offset to convert to a signed value
+ eor r4, r4, r2 ; p0 offset to convert to a signed value
+ eor r5, r5, r2 ; q0 offset to convert to a signed value
+
+ qsub8 r3, r3, r6 ; vp8_filter = p1 - q1
+ qsub8 r6, r5, r4 ; q0 - p0
+
+ qadd8 r3, r3, r6 ; vp8_filter += q0 - p0
+ ldr r9, c0x03030303 ; r9 = 3
+
+ qadd8 r3, r3, r6 ; vp8_filter += q0 - p0
+ ldr r7, c0x04040404
+
+ qadd8 r3, r3, r6 ; vp8_filter = p1-q1 + 3*(q0-p0))
+ ;STALL
+ and r3, r3, lr ; vp8_filter &= mask
+
+ qadd8 r9 , r3 , r9 ; Filter2 = vp8_filter + 3
+ qadd8 r3 , r3 , r7 ; Filter1 = vp8_filter + 4
+
+ shadd8 r9 , r9 , r8
+ shadd8 r3 , r3 , r8
+ shadd8 r9 , r9 , r8
+ shadd8 r3 , r3 , r8
+ shadd8 r9 , r9 , r8 ; Filter2 >>= 3
+ shadd8 r3 , r3 , r8 ; Filter1 >>= 3
+
+ ;calculate output
+ sub src, src, pstep, lsl #2
+
+ qadd8 r4, r4, r9 ; u = p0 + Filter2
+ qsub8 r5, r5, r3 ; u = q0 - Filter1
+ eor r4, r4, r2 ; *op0 = u^0x80
+ eor r5, r5, r2 ; *oq0 = u^0x80
+
+ strb r4, [src, #-1] ; store the result
+ mov r4, r4, lsr #8
+ strb r5, [src], pstep
+ mov r5, r5, lsr #8
+
+ strb r4, [src, #-1]
+ mov r4, r4, lsr #8
+ strb r5, [src], pstep
+ mov r5, r5, lsr #8
+
+ strb r4, [src, #-1]
+ mov r4, r4, lsr #8
+ strb r5, [src], pstep
+ mov r5, r5, lsr #8
+
+ strb r4, [src, #-1]
+ strb r5, [src], pstep
+
+|simple_vskip_filter|
+ subs r11, r11, #1
+
+ ; load soure data to r7, r8, r9, r10
+ ldrneh r3, [src, #-2]
+ pld [src, #23] ; preload for next block
+ ldrneh r4, [src], pstep
+
+ ldrneh r5, [src, #-2]
+ pld [src, #23]
+ ldrneh r6, [src], pstep
+
+ pkhbt r7, r3, r4, lsl #16
+
+ ldrneh r3, [src, #-2]
+ pld [src, #23]
+ ldrneh r4, [src], pstep
+
+ pkhbt r8, r5, r6, lsl #16
+
+ ldrneh r5, [src, #-2]
+ pld [src, #23]
+ ldrneh r6, [src], pstep
+
+ bne simple_vnext8
+
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_loop_filter_simple_vertical_edge_armv6|
+
+; Constant Pool
+c0x80808080 DCD 0x80808080
+c0x03030303 DCD 0x03030303
+c0x04040404 DCD 0x04040404
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/sixtappredict8x4_v6.asm b/libvpx/vp8/common/arm/armv6/sixtappredict8x4_v6.asm
new file mode 100644
index 0000000..e81aef5
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/sixtappredict8x4_v6.asm
@@ -0,0 +1,273 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_sixtap_predict8x4_armv6|
+
+ AREA |.text|, CODE, READONLY ; name this block of code
+;-------------------------------------
+; r0 unsigned char *src_ptr,
+; r1 int src_pixels_per_line,
+; r2 int xoffset,
+; r3 int yoffset,
+; stack unsigned char *dst_ptr,
+; stack int dst_pitch
+;-------------------------------------
+;note: In first pass, store the result in transpose(8linesx9columns) on stack. Temporary stack size is 184.
+;Line width is 20 that is 9 short data plus 2 to make it 4bytes aligned. In second pass, load data from stack,
+;and the result is stored in transpose.
+|vp8_sixtap_predict8x4_armv6| PROC
+ stmdb sp!, {r4 - r11, lr}
+ str r3, [sp, #-184]! ;reserve space on stack for temporary storage, store yoffset
+
+ cmp r2, #0 ;skip first_pass filter if xoffset=0
+ add lr, sp, #4 ;point to temporary buffer
+ beq skip_firstpass_filter
+
+;first-pass filter
+ adr r12, filter8_coeff
+ sub r0, r0, r1, lsl #1
+
+ add r3, r1, #10 ; preload next low
+ pld [r0, r3]
+
+ add r2, r12, r2, lsl #4 ;calculate filter location
+ add r0, r0, #3 ;adjust src only for loading convinience
+
+ ldr r3, [r2] ; load up packed filter coefficients
+ ldr r4, [r2, #4]
+ ldr r5, [r2, #8]
+
+ mov r2, #0x90000 ; height=9 is top part of counter
+
+ sub r1, r1, #8
+
+|first_pass_hloop_v6|
+ ldrb r6, [r0, #-5] ; load source data
+ ldrb r7, [r0, #-4]
+ ldrb r8, [r0, #-3]
+ ldrb r9, [r0, #-2]
+ ldrb r10, [r0, #-1]
+
+ orr r2, r2, #0x4 ; construct loop counter. width=8=4x2
+
+ pkhbt r6, r6, r7, lsl #16 ; r7 | r6
+ pkhbt r7, r7, r8, lsl #16 ; r8 | r7
+
+ pkhbt r8, r8, r9, lsl #16 ; r9 | r8
+ pkhbt r9, r9, r10, lsl #16 ; r10 | r9
+
+|first_pass_wloop_v6|
+ smuad r11, r6, r3 ; vp8_filter[0], vp8_filter[1]
+ smuad r12, r7, r3
+
+ ldrb r6, [r0], #1
+
+ smlad r11, r8, r4, r11 ; vp8_filter[2], vp8_filter[3]
+ ldrb r7, [r0], #1
+ smlad r12, r9, r4, r12
+
+ pkhbt r10, r10, r6, lsl #16 ; r10 | r9
+ pkhbt r6, r6, r7, lsl #16 ; r11 | r10
+ smlad r11, r10, r5, r11 ; vp8_filter[4], vp8_filter[5]
+ smlad r12, r6, r5, r12
+
+ sub r2, r2, #1
+
+ add r11, r11, #0x40 ; round_shift_and_clamp
+ tst r2, #0xff ; test loop counter
+ usat r11, #8, r11, asr #7
+ add r12, r12, #0x40
+ strh r11, [lr], #20 ; result is transposed and stored, which
+ usat r12, #8, r12, asr #7
+
+ strh r12, [lr], #20
+
+ movne r11, r6
+ movne r12, r7
+
+ movne r6, r8
+ movne r7, r9
+ movne r8, r10
+ movne r9, r11
+ movne r10, r12
+
+ bne first_pass_wloop_v6
+
+ ;;add r9, ppl, #30 ; attempt to load 2 adjacent cache lines
+ ;;IF ARCHITECTURE=6
+ ;pld [src, ppl]
+ ;;pld [src, r9]
+ ;;ENDIF
+
+ subs r2, r2, #0x10000
+
+ sub lr, lr, #158
+
+ add r0, r0, r1 ; move to next input line
+
+ add r11, r1, #18 ; preload next low. adding back block width(=8), which is subtracted earlier
+ pld [r0, r11]
+
+ bne first_pass_hloop_v6
+
+;second pass filter
+secondpass_filter
+ ldr r3, [sp], #4 ; load back yoffset
+ ldr r0, [sp, #216] ; load dst address from stack 180+36
+ ldr r1, [sp, #220] ; load dst stride from stack 180+40
+
+ cmp r3, #0
+ beq skip_secondpass_filter
+
+ adr r12, filter8_coeff
+ add lr, r12, r3, lsl #4 ;calculate filter location
+
+ mov r2, #0x00080000
+
+ ldr r3, [lr] ; load up packed filter coefficients
+ ldr r4, [lr, #4]
+ ldr r5, [lr, #8]
+
+ pkhbt r12, r4, r3 ; pack the filter differently
+ pkhbt r11, r5, r4
+
+second_pass_hloop_v6
+ ldr r6, [sp] ; load the data
+ ldr r7, [sp, #4]
+
+ orr r2, r2, #2 ; loop counter
+
+second_pass_wloop_v6
+ smuad lr, r3, r6 ; apply filter
+ smulbt r10, r3, r6
+
+ ldr r8, [sp, #8]
+
+ smlad lr, r4, r7, lr
+ smladx r10, r12, r7, r10
+
+ ldrh r9, [sp, #12]
+
+ smlad lr, r5, r8, lr
+ smladx r10, r11, r8, r10
+
+ add sp, sp, #4
+ smlatb r10, r5, r9, r10
+
+ sub r2, r2, #1
+
+ add lr, lr, #0x40 ; round_shift_and_clamp
+ tst r2, #0xff
+ usat lr, #8, lr, asr #7
+ add r10, r10, #0x40
+ strb lr, [r0], r1 ; the result is transposed back and stored
+ usat r10, #8, r10, asr #7
+
+ strb r10, [r0],r1
+
+ movne r6, r7
+ movne r7, r8
+
+ bne second_pass_wloop_v6
+
+ subs r2, r2, #0x10000
+ add sp, sp, #12 ; updata src for next loop (20-8)
+ sub r0, r0, r1, lsl #2
+ add r0, r0, #1
+
+ bne second_pass_hloop_v6
+
+ add sp, sp, #20
+ ldmia sp!, {r4 - r11, pc}
+
+;--------------------
+skip_firstpass_filter
+ sub r0, r0, r1, lsl #1
+ sub r1, r1, #8
+ mov r2, #9
+
+skip_firstpass_hloop
+ ldrb r4, [r0], #1 ; load data
+ subs r2, r2, #1
+ ldrb r5, [r0], #1
+ strh r4, [lr], #20 ; store it to immediate buffer
+ ldrb r6, [r0], #1 ; load data
+ strh r5, [lr], #20
+ ldrb r7, [r0], #1
+ strh r6, [lr], #20
+ ldrb r8, [r0], #1
+ strh r7, [lr], #20
+ ldrb r9, [r0], #1
+ strh r8, [lr], #20
+ ldrb r10, [r0], #1
+ strh r9, [lr], #20
+ ldrb r11, [r0], #1
+ strh r10, [lr], #20
+ add r0, r0, r1 ; move to next input line
+ strh r11, [lr], #20
+
+ sub lr, lr, #158 ; move over to next column
+ bne skip_firstpass_hloop
+
+ b secondpass_filter
+
+;--------------------
+skip_secondpass_filter
+ mov r2, #8
+ add sp, sp, #4 ;start from src[0] instead of src[-2]
+
+skip_secondpass_hloop
+ ldr r6, [sp], #4
+ subs r2, r2, #1
+ ldr r8, [sp], #4
+
+ mov r7, r6, lsr #16 ; unpack
+ strb r6, [r0], r1
+ mov r9, r8, lsr #16
+ strb r7, [r0], r1
+ add sp, sp, #12 ; 20-8
+ strb r8, [r0], r1
+ strb r9, [r0], r1
+
+ sub r0, r0, r1, lsl #2
+ add r0, r0, #1
+
+ bne skip_secondpass_hloop
+
+ add sp, sp, #16 ; 180 - (160 +4)
+
+ ldmia sp!, {r4 - r11, pc}
+
+ ENDP
+
+;-----------------
+;One word each is reserved. Label filter_coeff can be used to access the data.
+;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
+filter8_coeff
+ DCD 0x00000000, 0x00000080, 0x00000000, 0x00000000
+ DCD 0xfffa0000, 0x000c007b, 0x0000ffff, 0x00000000
+ DCD 0xfff50002, 0x0024006c, 0x0001fff8, 0x00000000
+ DCD 0xfff70000, 0x0032005d, 0x0000fffa, 0x00000000
+ DCD 0xfff00003, 0x004d004d, 0x0003fff0, 0x00000000
+ DCD 0xfffa0000, 0x005d0032, 0x0000fff7, 0x00000000
+ DCD 0xfff80001, 0x006c0024, 0x0002fff5, 0x00000000
+ DCD 0xffff0000, 0x007b000c, 0x0000fffa, 0x00000000
+
+ ;DCD 0, 0, 128, 0, 0, 0
+ ;DCD 0, -6, 123, 12, -1, 0
+ ;DCD 2, -11, 108, 36, -8, 1
+ ;DCD 0, -9, 93, 50, -6, 0
+ ;DCD 3, -16, 77, 77, -16, 3
+ ;DCD 0, -6, 50, 93, -9, 0
+ ;DCD 1, -8, 36, 108, -11, 2
+ ;DCD 0, -1, 12, 123, -6, 0
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/vp8_sad16x16_armv6.asm b/libvpx/vp8/common/arm/armv6/vp8_sad16x16_armv6.asm
new file mode 100644
index 0000000..1b4f5cf
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/vp8_sad16x16_armv6.asm
@@ -0,0 +1,96 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_sad16x16_armv6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 const unsigned char *src_ptr
+; r1 int src_stride
+; r2 const unsigned char *ref_ptr
+; r3 int ref_stride
+; stack max_sad (not used)
+|vp8_sad16x16_armv6| PROC
+ stmfd sp!, {r4-r12, lr}
+
+ pld [r0, r1, lsl #0]
+ pld [r2, r3, lsl #0]
+ pld [r0, r1, lsl #1]
+ pld [r2, r3, lsl #1]
+
+ mov r4, #0 ; sad = 0;
+ mov r5, #8 ; loop count
+
+loop
+ ; 1st row
+ ldr r6, [r0, #0x0] ; load 4 src pixels (1A)
+ ldr r8, [r2, #0x0] ; load 4 ref pixels (1A)
+ ldr r7, [r0, #0x4] ; load 4 src pixels (1A)
+ ldr r9, [r2, #0x4] ; load 4 ref pixels (1A)
+ ldr r10, [r0, #0x8] ; load 4 src pixels (1B)
+ ldr r11, [r0, #0xC] ; load 4 src pixels (1B)
+
+ usada8 r4, r8, r6, r4 ; calculate sad for 4 pixels
+ usad8 r8, r7, r9 ; calculate sad for 4 pixels
+
+ ldr r12, [r2, #0x8] ; load 4 ref pixels (1B)
+ ldr lr, [r2, #0xC] ; load 4 ref pixels (1B)
+
+ add r0, r0, r1 ; set src pointer to next row
+ add r2, r2, r3 ; set dst pointer to next row
+
+ pld [r0, r1, lsl #1]
+ pld [r2, r3, lsl #1]
+
+ usada8 r4, r10, r12, r4 ; calculate sad for 4 pixels
+ usada8 r8, r11, lr, r8 ; calculate sad for 4 pixels
+
+ ldr r6, [r0, #0x0] ; load 4 src pixels (2A)
+ ldr r7, [r0, #0x4] ; load 4 src pixels (2A)
+ add r4, r4, r8 ; add partial sad values
+
+ ; 2nd row
+ ldr r8, [r2, #0x0] ; load 4 ref pixels (2A)
+ ldr r9, [r2, #0x4] ; load 4 ref pixels (2A)
+ ldr r10, [r0, #0x8] ; load 4 src pixels (2B)
+ ldr r11, [r0, #0xC] ; load 4 src pixels (2B)
+
+ usada8 r4, r6, r8, r4 ; calculate sad for 4 pixels
+ usad8 r8, r7, r9 ; calculate sad for 4 pixels
+
+ ldr r12, [r2, #0x8] ; load 4 ref pixels (2B)
+ ldr lr, [r2, #0xC] ; load 4 ref pixels (2B)
+
+ add r0, r0, r1 ; set src pointer to next row
+ add r2, r2, r3 ; set dst pointer to next row
+
+ usada8 r4, r10, r12, r4 ; calculate sad for 4 pixels
+ usada8 r8, r11, lr, r8 ; calculate sad for 4 pixels
+
+ pld [r0, r1, lsl #1]
+ pld [r2, r3, lsl #1]
+
+ subs r5, r5, #1 ; decrement loop counter
+ add r4, r4, r8 ; add partial sad values
+
+ bne loop
+
+ mov r0, r4 ; return sad
+ ldmfd sp!, {r4-r12, pc}
+
+ ENDP
+
+ END
+
diff --git a/libvpx/vp8/common/arm/armv6/vp8_variance16x16_armv6.asm b/libvpx/vp8/common/arm/armv6/vp8_variance16x16_armv6.asm
new file mode 100644
index 0000000..dc84c30
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/vp8_variance16x16_armv6.asm
@@ -0,0 +1,154 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_variance16x16_armv6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+|vp8_variance16x16_armv6| PROC
+
+ stmfd sp!, {r4-r12, lr}
+
+ pld [r0, r1, lsl #0]
+ pld [r2, r3, lsl #0]
+
+ mov r8, #0 ; initialize sum = 0
+ mov r11, #0 ; initialize sse = 0
+ mov r12, #16 ; set loop counter to 16 (=block height)
+
+loop
+ ; 1st 4 pixels
+ ldr r4, [r0, #0] ; load 4 src pixels
+ ldr r5, [r2, #0] ; load 4 ref pixels
+
+ mov lr, #0 ; constant zero
+
+ usub8 r6, r4, r5 ; calculate difference
+ pld [r0, r1, lsl #1]
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r9, r5, r4 ; calculate difference with reversed operands
+ pld [r2, r3, lsl #1]
+ sel r6, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+ ; calculate total sum
+ adds r8, r8, r4 ; add positive differences to sum
+ subs r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 2nd 4 pixels
+ ldr r4, [r0, #4] ; load 4 src pixels
+ ldr r5, [r2, #4] ; load 4 ref pixels
+ smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r9, r5, r4 ; calculate difference with reversed operands
+ sel r6, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 3rd 4 pixels
+ ldr r4, [r0, #8] ; load 4 src pixels
+ ldr r5, [r2, #8] ; load 4 ref pixels
+ smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r9, r5, r4 ; calculate difference with reversed operands
+ sel r6, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 4th 4 pixels
+ ldr r4, [r0, #12] ; load 4 src pixels
+ ldr r5, [r2, #12] ; load 4 ref pixels
+ smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ add r0, r0, r1 ; set src_ptr to next row
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r9, r5, r4 ; calculate difference with reversed operands
+ add r2, r2, r3 ; set dst_ptr to next row
+ sel r6, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+ smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
+
+
+ subs r12, r12, #1
+
+ bne loop
+
+ ; return stuff
+ ldr r6, [sp, #40] ; get address of sse
+ mul r0, r8, r8 ; sum * sum
+ str r11, [r6] ; store sse
+ sub r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
+
+ ldmfd sp!, {r4-r12, pc}
+
+ ENDP
+
+ END
+
diff --git a/libvpx/vp8/common/arm/armv6/vp8_variance8x8_armv6.asm b/libvpx/vp8/common/arm/armv6/vp8_variance8x8_armv6.asm
new file mode 100644
index 0000000..adc353d
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/vp8_variance8x8_armv6.asm
@@ -0,0 +1,101 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_variance8x8_armv6|
+
+ ARM
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+|vp8_variance8x8_armv6| PROC
+
+ push {r4-r10, lr}
+
+ pld [r0, r1, lsl #0]
+ pld [r2, r3, lsl #0]
+
+ mov r12, #8 ; set loop counter to 8 (=block height)
+ mov r4, #0 ; initialize sum = 0
+ mov r5, #0 ; initialize sse = 0
+
+loop
+ ; 1st 4 pixels
+ ldr r6, [r0, #0x0] ; load 4 src pixels
+ ldr r7, [r2, #0x0] ; load 4 ref pixels
+
+ mov lr, #0 ; constant zero
+
+ usub8 r8, r6, r7 ; calculate difference
+ pld [r0, r1, lsl #1]
+ sel r10, r8, lr ; select bytes with positive difference
+ usub8 r9, r7, r6 ; calculate difference with reversed operands
+ pld [r2, r3, lsl #1]
+ sel r8, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r6, r10, lr ; calculate sum of positive differences
+ usad8 r7, r8, lr ; calculate sum of negative differences
+ orr r8, r8, r10 ; differences of all 4 pixels
+ ; calculate total sum
+ add r4, r4, r6 ; add positive differences to sum
+ sub r4, r4, r7 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r7, r8 ; byte (two pixels) to halfwords
+ uxtb16 r10, r8, ror #8 ; another two pixels to halfwords
+ smlad r5, r7, r7, r5 ; dual signed multiply, add and accumulate (1)
+
+ ; 2nd 4 pixels
+ ldr r6, [r0, #0x4] ; load 4 src pixels
+ ldr r7, [r2, #0x4] ; load 4 ref pixels
+ smlad r5, r10, r10, r5 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r8, r6, r7 ; calculate difference
+ add r0, r0, r1 ; set src_ptr to next row
+ sel r10, r8, lr ; select bytes with positive difference
+ usub8 r9, r7, r6 ; calculate difference with reversed operands
+ add r2, r2, r3 ; set dst_ptr to next row
+ sel r8, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r6, r10, lr ; calculate sum of positive differences
+ usad8 r7, r8, lr ; calculate sum of negative differences
+ orr r8, r8, r10 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r4, r4, r6 ; add positive differences to sum
+ sub r4, r4, r7 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r7, r8 ; byte (two pixels) to halfwords
+ uxtb16 r10, r8, ror #8 ; another two pixels to halfwords
+ smlad r5, r7, r7, r5 ; dual signed multiply, add and accumulate (1)
+ subs r12, r12, #1 ; next row
+ smlad r5, r10, r10, r5 ; dual signed multiply, add and accumulate (2)
+
+ bne loop
+
+ ; return stuff
+ ldr r8, [sp, #32] ; get address of sse
+ mul r1, r4, r4 ; sum * sum
+ str r5, [r8] ; store sse
+ sub r0, r5, r1, ASR #6 ; return (sse - ((sum * sum) >> 6))
+
+ pop {r4-r10, pc}
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6.asm b/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6.asm
new file mode 100644
index 0000000..dd2ce68
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6.asm
@@ -0,0 +1,182 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_variance_halfpixvar16x16_h_armv6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+|vp8_variance_halfpixvar16x16_h_armv6| PROC
+
+ stmfd sp!, {r4-r12, lr}
+
+ pld [r0, r1, lsl #0]
+ pld [r2, r3, lsl #0]
+
+ mov r8, #0 ; initialize sum = 0
+ ldr r10, c80808080
+ mov r11, #0 ; initialize sse = 0
+ mov r12, #16 ; set loop counter to 16 (=block height)
+ mov lr, #0 ; constant zero
+loop
+ ; 1st 4 pixels
+ ldr r4, [r0, #0] ; load 4 src pixels
+ ldr r6, [r0, #1] ; load 4 src pixels with 1 byte offset
+ ldr r5, [r2, #0] ; load 4 ref pixels
+
+ ; bilinear interpolation
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+
+ usub8 r6, r4, r5 ; calculate difference
+ pld [r0, r1, lsl #1]
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ pld [r2, r3, lsl #1]
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+ ; calculate total sum
+ adds r8, r8, r4 ; add positive differences to sum
+ subs r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 2nd 4 pixels
+ ldr r4, [r0, #4] ; load 4 src pixels
+ ldr r6, [r0, #5] ; load 4 src pixels with 1 byte offset
+ ldr r5, [r2, #4] ; load 4 ref pixels
+
+ ; bilinear interpolation
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 3rd 4 pixels
+ ldr r4, [r0, #8] ; load 4 src pixels
+ ldr r6, [r0, #9] ; load 4 src pixels with 1 byte offset
+ ldr r5, [r2, #8] ; load 4 ref pixels
+
+ ; bilinear interpolation
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 4th 4 pixels
+ ldr r4, [r0, #12] ; load 4 src pixels
+ ldr r6, [r0, #13] ; load 4 src pixels with 1 byte offset
+ ldr r5, [r2, #12] ; load 4 ref pixels
+
+ ; bilinear interpolation
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ add r0, r0, r1 ; set src_ptr to next row
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ add r2, r2, r3 ; set dst_ptr to next row
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ subs r12, r12, #1
+
+ bne loop
+
+ ; return stuff
+ ldr r6, [sp, #40] ; get address of sse
+ mul r0, r8, r8 ; sum * sum
+ str r11, [r6] ; store sse
+ sub r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
+
+ ldmfd sp!, {r4-r12, pc}
+
+ ENDP
+
+c80808080
+ DCD 0x80808080
+
+ END
+
diff --git a/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6.asm b/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6.asm
new file mode 100644
index 0000000..f972d9b
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6.asm
@@ -0,0 +1,222 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_variance_halfpixvar16x16_hv_armv6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+|vp8_variance_halfpixvar16x16_hv_armv6| PROC
+
+ stmfd sp!, {r4-r12, lr}
+
+ pld [r0, r1, lsl #0]
+ pld [r2, r3, lsl #0]
+
+ mov r8, #0 ; initialize sum = 0
+ ldr r10, c80808080
+ mov r11, #0 ; initialize sse = 0
+ mov r12, #16 ; set loop counter to 16 (=block height)
+ mov lr, #0 ; constant zero
+loop
+ add r9, r0, r1 ; pointer to pixels on the next row
+ ; 1st 4 pixels
+ ldr r4, [r0, #0] ; load source pixels a, row N
+ ldr r6, [r0, #1] ; load source pixels b, row N
+ ldr r5, [r9, #0] ; load source pixels c, row N+1
+ ldr r7, [r9, #1] ; load source pixels d, row N+1
+
+ ; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+ ; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
+ mvn r7, r7
+ uhsub8 r5, r5, r7
+ eor r5, r5, r10
+ ; z = (x + y + 1) >> 1, interpolate half pixel values vertically
+ mvn r5, r5
+ uhsub8 r4, r4, r5
+ ldr r5, [r2, #0] ; load 4 ref pixels
+ eor r4, r4, r10
+
+ usub8 r6, r4, r5 ; calculate difference
+ pld [r0, r1, lsl #1]
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ pld [r2, r3, lsl #1]
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+ ; calculate total sum
+ adds r8, r8, r4 ; add positive differences to sum
+ subs r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 2nd 4 pixels
+ ldr r4, [r0, #4] ; load source pixels a, row N
+ ldr r6, [r0, #5] ; load source pixels b, row N
+ ldr r5, [r9, #4] ; load source pixels c, row N+1
+
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ ldr r7, [r9, #5] ; load source pixels d, row N+1
+
+ ; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+ ; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
+ mvn r7, r7
+ uhsub8 r5, r5, r7
+ eor r5, r5, r10
+ ; z = (x + y + 1) >> 1, interpolate half pixel values vertically
+ mvn r5, r5
+ uhsub8 r4, r4, r5
+ ldr r5, [r2, #4] ; load 4 ref pixels
+ eor r4, r4, r10
+
+ usub8 r6, r4, r5 ; calculate difference
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 3rd 4 pixels
+ ldr r4, [r0, #8] ; load source pixels a, row N
+ ldr r6, [r0, #9] ; load source pixels b, row N
+ ldr r5, [r9, #8] ; load source pixels c, row N+1
+
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ ldr r7, [r9, #9] ; load source pixels d, row N+1
+
+ ; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+ ; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
+ mvn r7, r7
+ uhsub8 r5, r5, r7
+ eor r5, r5, r10
+ ; z = (x + y + 1) >> 1, interpolate half pixel values vertically
+ mvn r5, r5
+ uhsub8 r4, r4, r5
+ ldr r5, [r2, #8] ; load 4 ref pixels
+ eor r4, r4, r10
+
+ usub8 r6, r4, r5 ; calculate difference
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 4th 4 pixels
+ ldr r4, [r0, #12] ; load source pixels a, row N
+ ldr r6, [r0, #13] ; load source pixels b, row N
+ ldr r5, [r9, #12] ; load source pixels c, row N+1
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+ ldr r7, [r9, #13] ; load source pixels d, row N+1
+
+ ; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+ ; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
+ mvn r7, r7
+ uhsub8 r5, r5, r7
+ eor r5, r5, r10
+ ; z = (x + y + 1) >> 1, interpolate half pixel values vertically
+ mvn r5, r5
+ uhsub8 r4, r4, r5
+ ldr r5, [r2, #12] ; load 4 ref pixels
+ eor r4, r4, r10
+
+ usub8 r6, r4, r5 ; calculate difference
+ add r0, r0, r1 ; set src_ptr to next row
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ add r2, r2, r3 ; set dst_ptr to next row
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+ subs r12, r12, #1
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ bne loop
+
+ ; return stuff
+ ldr r6, [sp, #40] ; get address of sse
+ mul r0, r8, r8 ; sum * sum
+ str r11, [r6] ; store sse
+ sub r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
+
+ ldmfd sp!, {r4-r12, pc}
+
+ ENDP
+
+c80808080
+ DCD 0x80808080
+
+ END
diff --git a/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6.asm b/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6.asm
new file mode 100644
index 0000000..f5da9c0
--- /dev/null
+++ b/libvpx/vp8/common/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6.asm
@@ -0,0 +1,184 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_variance_halfpixvar16x16_v_armv6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+|vp8_variance_halfpixvar16x16_v_armv6| PROC
+
+ stmfd sp!, {r4-r12, lr}
+
+ pld [r0, r1, lsl #0]
+ pld [r2, r3, lsl #0]
+
+ mov r8, #0 ; initialize sum = 0
+ ldr r10, c80808080
+ mov r11, #0 ; initialize sse = 0
+ mov r12, #16 ; set loop counter to 16 (=block height)
+ mov lr, #0 ; constant zero
+loop
+ add r9, r0, r1 ; set src pointer to next row
+ ; 1st 4 pixels
+ ldr r4, [r0, #0] ; load 4 src pixels
+ ldr r6, [r9, #0] ; load 4 src pixels from next row
+ ldr r5, [r2, #0] ; load 4 ref pixels
+
+ ; bilinear interpolation
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+
+ usub8 r6, r4, r5 ; calculate difference
+ pld [r0, r1, lsl #1]
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ pld [r2, r3, lsl #1]
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+ ; calculate total sum
+ adds r8, r8, r4 ; add positive differences to sum
+ subs r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 2nd 4 pixels
+ ldr r4, [r0, #4] ; load 4 src pixels
+ ldr r6, [r9, #4] ; load 4 src pixels from next row
+ ldr r5, [r2, #4] ; load 4 ref pixels
+
+ ; bilinear interpolation
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 3rd 4 pixels
+ ldr r4, [r0, #8] ; load 4 src pixels
+ ldr r6, [r9, #8] ; load 4 src pixels from next row
+ ldr r5, [r2, #8] ; load 4 ref pixels
+
+ ; bilinear interpolation
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+
+ ; 4th 4 pixels
+ ldr r4, [r0, #12] ; load 4 src pixels
+ ldr r6, [r9, #12] ; load 4 src pixels from next row
+ ldr r5, [r2, #12] ; load 4 ref pixels
+
+ ; bilinear interpolation
+ mvn r6, r6
+ uhsub8 r4, r4, r6
+ eor r4, r4, r10
+
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r6, r4, r5 ; calculate difference
+ add r0, r0, r1 ; set src_ptr to next row
+ sel r7, r6, lr ; select bytes with positive difference
+ usub8 r6, r5, r4 ; calculate difference with reversed operands
+ add r2, r2, r3 ; set dst_ptr to next row
+ sel r6, r6, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r4, r7, lr ; calculate sum of positive differences
+ usad8 r5, r6, lr ; calculate sum of negative differences
+ orr r6, r6, r7 ; differences of all 4 pixels
+
+ ; calculate total sum
+ add r8, r8, r4 ; add positive differences to sum
+ sub r8, r8, r5 ; substract negative differences from sum
+
+ ; calculate sse
+ uxtb16 r5, r6 ; byte (two pixels) to halfwords
+ uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
+ smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
+ smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
+
+
+ subs r12, r12, #1
+
+ bne loop
+
+ ; return stuff
+ ldr r6, [sp, #40] ; get address of sse
+ mul r0, r8, r8 ; sum * sum
+ str r11, [r6] ; store sse
+ sub r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
+
+ ldmfd sp!, {r4-r12, pc}
+
+ ENDP
+
+c80808080
+ DCD 0x80808080
+
+ END
+
diff --git a/libvpx/vp8/common/arm/bilinearfilter_arm.c b/libvpx/vp8/common/arm/bilinearfilter_arm.c
new file mode 100644
index 0000000..c63073c
--- /dev/null
+++ b/libvpx/vp8/common/arm/bilinearfilter_arm.c
@@ -0,0 +1,113 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include <math.h>
+#include "vp8/common/filter.h"
+#include "bilinearfilter_arm.h"
+
+void vp8_filter_block2d_bil_armv6
+(
+ unsigned char *src_ptr,
+ unsigned char *dst_ptr,
+ unsigned int src_pitch,
+ unsigned int dst_pitch,
+ const short *HFilter,
+ const short *VFilter,
+ int Width,
+ int Height
+)
+{
+ unsigned short FData[36*16]; /* Temp data buffer used in filtering */
+
+ /* First filter 1-D horizontally... */
+ vp8_filter_block2d_bil_first_pass_armv6(src_ptr, FData, src_pitch, Height + 1, Width, HFilter);
+
+ /* then 1-D vertically... */
+ vp8_filter_block2d_bil_second_pass_armv6(FData, dst_ptr, dst_pitch, Height, Width, VFilter);
+}
+
+
+void vp8_bilinear_predict4x4_armv6
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 4, 4);
+}
+
+void vp8_bilinear_predict8x8_armv6
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 8);
+}
+
+void vp8_bilinear_predict8x4_armv6
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 4);
+}
+
+void vp8_bilinear_predict16x16_armv6
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 16, 16);
+}
diff --git a/vp8/common/arm/bilinearfilter_arm.h b/libvpx/vp8/common/arm/bilinearfilter_arm.h
similarity index 100%
rename from vp8/common/arm/bilinearfilter_arm.h
rename to libvpx/vp8/common/arm/bilinearfilter_arm.h
diff --git a/libvpx/vp8/common/arm/dequantize_arm.c b/libvpx/vp8/common/arm/dequantize_arm.c
new file mode 100644
index 0000000..70e72aa
--- /dev/null
+++ b/libvpx/vp8/common/arm/dequantize_arm.c
@@ -0,0 +1,42 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vp8/common/blockd.h"
+
+#if HAVE_NEON
+extern void vp8_dequantize_b_loop_neon(short *Q, short *DQC, short *DQ);
+#endif
+
+#if HAVE_MEDIA
+extern void vp8_dequantize_b_loop_v6(short *Q, short *DQC, short *DQ);
+#endif
+
+#if HAVE_NEON
+
+void vp8_dequantize_b_neon(BLOCKD *d, short *DQC)
+{
+ short *DQ = d->dqcoeff;
+ short *Q = d->qcoeff;
+
+ vp8_dequantize_b_loop_neon(Q, DQC, DQ);
+}
+#endif
+
+#if HAVE_MEDIA
+void vp8_dequantize_b_v6(BLOCKD *d, short *DQC)
+{
+ short *DQ = d->dqcoeff;
+ short *Q = d->qcoeff;
+
+ vp8_dequantize_b_loop_v6(Q, DQC, DQ);
+}
+#endif
diff --git a/libvpx/vp8/common/arm/filter_arm.c b/libvpx/vp8/common/arm/filter_arm.c
new file mode 100644
index 0000000..148951a
--- /dev/null
+++ b/libvpx/vp8/common/arm/filter_arm.c
@@ -0,0 +1,221 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include <math.h>
+#include "vp8/common/filter.h"
+#include "vpx_ports/mem.h"
+
+extern void vp8_filter_block2d_first_pass_armv6
+(
+ unsigned char *src_ptr,
+ short *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int output_width,
+ unsigned int output_height,
+ const short *vp8_filter
+);
+
+// 8x8
+extern void vp8_filter_block2d_first_pass_8x8_armv6
+(
+ unsigned char *src_ptr,
+ short *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int output_width,
+ unsigned int output_height,
+ const short *vp8_filter
+);
+
+// 16x16
+extern void vp8_filter_block2d_first_pass_16x16_armv6
+(
+ unsigned char *src_ptr,
+ short *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int output_width,
+ unsigned int output_height,
+ const short *vp8_filter
+);
+
+extern void vp8_filter_block2d_second_pass_armv6
+(
+ short *src_ptr,
+ unsigned char *output_ptr,
+ unsigned int output_pitch,
+ unsigned int cnt,
+ const short *vp8_filter
+);
+
+extern void vp8_filter4_block2d_second_pass_armv6
+(
+ short *src_ptr,
+ unsigned char *output_ptr,
+ unsigned int output_pitch,
+ unsigned int cnt,
+ const short *vp8_filter
+);
+
+extern void vp8_filter_block2d_first_pass_only_armv6
+(
+ unsigned char *src_ptr,
+ unsigned char *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int cnt,
+ unsigned int output_pitch,
+ const short *vp8_filter
+);
+
+
+extern void vp8_filter_block2d_second_pass_only_armv6
+(
+ unsigned char *src_ptr,
+ unsigned char *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int cnt,
+ unsigned int output_pitch,
+ const short *vp8_filter
+);
+
+#if HAVE_MEDIA
+void vp8_sixtap_predict4x4_armv6
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+ DECLARE_ALIGNED_ARRAY(4, short, FData, 12*4); /* Temp data buffer used in filtering */
+
+
+ HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
+ VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
+
+ /* Vfilter is null. First pass only */
+ if (xoffset && !yoffset)
+ {
+ /*vp8_filter_block2d_first_pass_armv6 ( src_ptr, FData+2, src_pixels_per_line, 4, 4, HFilter );
+ vp8_filter_block2d_second_pass_armv6 ( FData+2, dst_ptr, dst_pitch, 4, VFilter );*/
+
+ vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, HFilter);
+ }
+ /* Hfilter is null. Second pass only */
+ else if (!xoffset && yoffset)
+ {
+ vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, VFilter);
+ }
+ else
+ {
+ /* Vfilter is a 4 tap filter */
+ if (yoffset & 0x1)
+ {
+ vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 4, 7, HFilter);
+ vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
+ }
+ /* Vfilter is 6 tap filter */
+ else
+ {
+ vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 4, 9, HFilter);
+ vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
+ }
+ }
+}
+
+void vp8_sixtap_predict8x8_armv6
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+ DECLARE_ALIGNED_ARRAY(4, short, FData, 16*8); /* Temp data buffer used in filtering */
+
+ HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
+ VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
+
+ if (xoffset && !yoffset)
+ {
+ vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, HFilter);
+ }
+ /* Hfilter is null. Second pass only */
+ else if (!xoffset && yoffset)
+ {
+ vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, VFilter);
+ }
+ else
+ {
+ if (yoffset & 0x1)
+ {
+ vp8_filter_block2d_first_pass_8x8_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 8, 11, HFilter);
+ vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
+ }
+ else
+ {
+ vp8_filter_block2d_first_pass_8x8_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 8, 13, HFilter);
+ vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
+ }
+ }
+}
+
+
+void vp8_sixtap_predict16x16_armv6
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+ DECLARE_ALIGNED_ARRAY(4, short, FData, 24*16); /* Temp data buffer used in filtering */
+
+ HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
+ VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
+
+ if (xoffset && !yoffset)
+ {
+ vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, HFilter);
+ }
+ /* Hfilter is null. Second pass only */
+ else if (!xoffset && yoffset)
+ {
+ vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, VFilter);
+ }
+ else
+ {
+ if (yoffset & 0x1)
+ {
+ vp8_filter_block2d_first_pass_16x16_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 16, 19, HFilter);
+ vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
+ }
+ else
+ {
+ vp8_filter_block2d_first_pass_16x16_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 16, 21, HFilter);
+ vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
+ }
+ }
+
+}
+#endif
diff --git a/libvpx/vp8/common/arm/loopfilter_arm.c b/libvpx/vp8/common/arm/loopfilter_arm.c
new file mode 100644
index 0000000..b8f9bd9
--- /dev/null
+++ b/libvpx/vp8/common/arm/loopfilter_arm.c
@@ -0,0 +1,181 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/common/loopfilter.h"
+#include "vp8/common/onyxc_int.h"
+
+#define prototype_loopfilter(sym) \
+ void sym(unsigned char *src, int pitch, const unsigned char *blimit,\
+ const unsigned char *limit, const unsigned char *thresh, int count)
+
+#if HAVE_MEDIA
+extern prototype_loopfilter(vp8_loop_filter_horizontal_edge_armv6);
+extern prototype_loopfilter(vp8_loop_filter_vertical_edge_armv6);
+extern prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_armv6);
+extern prototype_loopfilter(vp8_mbloop_filter_vertical_edge_armv6);
+#endif
+
+#if HAVE_NEON
+typedef void loopfilter_y_neon(unsigned char *src, int pitch,
+ unsigned char blimit, unsigned char limit, unsigned char thresh);
+typedef void loopfilter_uv_neon(unsigned char *u, int pitch,
+ unsigned char blimit, unsigned char limit, unsigned char thresh,
+ unsigned char *v);
+
+extern loopfilter_y_neon vp8_loop_filter_horizontal_edge_y_neon;
+extern loopfilter_y_neon vp8_loop_filter_vertical_edge_y_neon;
+extern loopfilter_y_neon vp8_mbloop_filter_horizontal_edge_y_neon;
+extern loopfilter_y_neon vp8_mbloop_filter_vertical_edge_y_neon;
+
+extern loopfilter_uv_neon vp8_loop_filter_horizontal_edge_uv_neon;
+extern loopfilter_uv_neon vp8_loop_filter_vertical_edge_uv_neon;
+extern loopfilter_uv_neon vp8_mbloop_filter_horizontal_edge_uv_neon;
+extern loopfilter_uv_neon vp8_mbloop_filter_vertical_edge_uv_neon;
+#endif
+
+#if HAVE_MEDIA
+/* ARMV6/MEDIA loopfilter functions*/
+/* Horizontal MB filtering */
+void vp8_loop_filter_mbh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_mbloop_filter_horizontal_edge_armv6(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_mbloop_filter_horizontal_edge_armv6(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_mbloop_filter_horizontal_edge_armv6(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+}
+
+/* Vertical MB Filtering */
+void vp8_loop_filter_mbv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_mbloop_filter_vertical_edge_armv6(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_mbloop_filter_vertical_edge_armv6(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_mbloop_filter_vertical_edge_armv6(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+}
+
+/* Horizontal B Filtering */
+void vp8_loop_filter_bh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_loop_filter_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_loop_filter_horizontal_edge_armv6(u_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_loop_filter_horizontal_edge_armv6(v_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+}
+
+void vp8_loop_filter_bhs_armv6(unsigned char *y_ptr, int y_stride,
+ const unsigned char *blimit)
+{
+ vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride, blimit);
+ vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride, blimit);
+ vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, blimit);
+}
+
+/* Vertical B Filtering */
+void vp8_loop_filter_bv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_loop_filter_vertical_edge_armv6(y_ptr + 4, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_vertical_edge_armv6(y_ptr + 8, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_vertical_edge_armv6(y_ptr + 12, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_loop_filter_vertical_edge_armv6(u_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_loop_filter_vertical_edge_armv6(v_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+}
+
+void vp8_loop_filter_bvs_armv6(unsigned char *y_ptr, int y_stride,
+ const unsigned char *blimit)
+{
+ vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 4, y_stride, blimit);
+ vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 8, y_stride, blimit);
+ vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 12, y_stride, blimit);
+}
+#endif
+
+#if HAVE_NEON
+/* NEON loopfilter functions */
+/* Horizontal MB filtering */
+void vp8_loop_filter_mbh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ unsigned char mblim = *lfi->mblim;
+ unsigned char lim = *lfi->lim;
+ unsigned char hev_thr = *lfi->hev_thr;
+ vp8_mbloop_filter_horizontal_edge_y_neon(y_ptr, y_stride, mblim, lim, hev_thr);
+
+ if (u_ptr)
+ vp8_mbloop_filter_horizontal_edge_uv_neon(u_ptr, uv_stride, mblim, lim, hev_thr, v_ptr);
+}
+
+/* Vertical MB Filtering */
+void vp8_loop_filter_mbv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ unsigned char mblim = *lfi->mblim;
+ unsigned char lim = *lfi->lim;
+ unsigned char hev_thr = *lfi->hev_thr;
+
+ vp8_mbloop_filter_vertical_edge_y_neon(y_ptr, y_stride, mblim, lim, hev_thr);
+
+ if (u_ptr)
+ vp8_mbloop_filter_vertical_edge_uv_neon(u_ptr, uv_stride, mblim, lim, hev_thr, v_ptr);
+}
+
+/* Horizontal B Filtering */
+void vp8_loop_filter_bh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ unsigned char blim = *lfi->blim;
+ unsigned char lim = *lfi->lim;
+ unsigned char hev_thr = *lfi->hev_thr;
+
+ vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 4 * y_stride, y_stride, blim, lim, hev_thr);
+ vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 8 * y_stride, y_stride, blim, lim, hev_thr);
+ vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 12 * y_stride, y_stride, blim, lim, hev_thr);
+
+ if (u_ptr)
+ vp8_loop_filter_horizontal_edge_uv_neon(u_ptr + 4 * uv_stride, uv_stride, blim, lim, hev_thr, v_ptr + 4 * uv_stride);
+}
+
+/* Vertical B Filtering */
+void vp8_loop_filter_bv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ unsigned char blim = *lfi->blim;
+ unsigned char lim = *lfi->lim;
+ unsigned char hev_thr = *lfi->hev_thr;
+
+ vp8_loop_filter_vertical_edge_y_neon(y_ptr + 4, y_stride, blim, lim, hev_thr);
+ vp8_loop_filter_vertical_edge_y_neon(y_ptr + 8, y_stride, blim, lim, hev_thr);
+ vp8_loop_filter_vertical_edge_y_neon(y_ptr + 12, y_stride, blim, lim, hev_thr);
+
+ if (u_ptr)
+ vp8_loop_filter_vertical_edge_uv_neon(u_ptr + 4, uv_stride, blim, lim, hev_thr, v_ptr + 4);
+}
+#endif
diff --git a/vp8/common/arm/neon/bilinearpredict16x16_neon.asm b/libvpx/vp8/common/arm/neon/bilinearpredict16x16_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/bilinearpredict16x16_neon.asm
rename to libvpx/vp8/common/arm/neon/bilinearpredict16x16_neon.asm
diff --git a/vp8/common/arm/neon/bilinearpredict4x4_neon.asm b/libvpx/vp8/common/arm/neon/bilinearpredict4x4_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/bilinearpredict4x4_neon.asm
rename to libvpx/vp8/common/arm/neon/bilinearpredict4x4_neon.asm
diff --git a/vp8/common/arm/neon/bilinearpredict8x4_neon.asm b/libvpx/vp8/common/arm/neon/bilinearpredict8x4_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/bilinearpredict8x4_neon.asm
rename to libvpx/vp8/common/arm/neon/bilinearpredict8x4_neon.asm
diff --git a/vp8/common/arm/neon/bilinearpredict8x8_neon.asm b/libvpx/vp8/common/arm/neon/bilinearpredict8x8_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/bilinearpredict8x8_neon.asm
rename to libvpx/vp8/common/arm/neon/bilinearpredict8x8_neon.asm
diff --git a/vp8/common/arm/neon/buildintrapredictorsmby_neon.asm b/libvpx/vp8/common/arm/neon/buildintrapredictorsmby_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/buildintrapredictorsmby_neon.asm
rename to libvpx/vp8/common/arm/neon/buildintrapredictorsmby_neon.asm
diff --git a/vp8/common/arm/neon/copymem16x16_neon.asm b/libvpx/vp8/common/arm/neon/copymem16x16_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/copymem16x16_neon.asm
rename to libvpx/vp8/common/arm/neon/copymem16x16_neon.asm
diff --git a/vp8/common/arm/neon/copymem8x4_neon.asm b/libvpx/vp8/common/arm/neon/copymem8x4_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/copymem8x4_neon.asm
rename to libvpx/vp8/common/arm/neon/copymem8x4_neon.asm
diff --git a/vp8/common/arm/neon/copymem8x8_neon.asm b/libvpx/vp8/common/arm/neon/copymem8x8_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/copymem8x8_neon.asm
rename to libvpx/vp8/common/arm/neon/copymem8x8_neon.asm
diff --git a/libvpx/vp8/common/arm/neon/dc_only_idct_add_neon.asm b/libvpx/vp8/common/arm/neon/dc_only_idct_add_neon.asm
new file mode 100644
index 0000000..79ff02c
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/dc_only_idct_add_neon.asm
@@ -0,0 +1,54 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license and patent
+; grant that can be found in the LICENSE file in the root of the source
+; tree. All contributing project authors may be found in the AUTHORS
+; file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_dc_only_idct_add_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+;void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr,
+; int pred_stride, unsigned char *dst_ptr,
+; int dst_stride)
+
+; r0 input_dc
+; r1 pred_ptr
+; r2 pred_stride
+; r3 dst_ptr
+; sp dst_stride
+
+|vp8_dc_only_idct_add_neon| PROC
+ add r0, r0, #4
+ asr r0, r0, #3
+ ldr r12, [sp]
+ vdup.16 q0, r0
+
+ vld1.32 {d2[0]}, [r1], r2
+ vld1.32 {d2[1]}, [r1], r2
+ vld1.32 {d4[0]}, [r1], r2
+ vld1.32 {d4[1]}, [r1]
+
+ vaddw.u8 q1, q0, d2
+ vaddw.u8 q2, q0, d4
+
+ vqmovun.s16 d2, q1
+ vqmovun.s16 d4, q2
+
+ vst1.32 {d2[0]}, [r3], r12
+ vst1.32 {d2[1]}, [r3], r12
+ vst1.32 {d4[0]}, [r3], r12
+ vst1.32 {d4[1]}, [r3]
+
+ bx lr
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/dequant_idct_neon.asm b/libvpx/vp8/common/arm/neon/dequant_idct_neon.asm
new file mode 100644
index 0000000..602cce6
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/dequant_idct_neon.asm
@@ -0,0 +1,131 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_dequant_idct_add_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+;void vp8_dequant_idct_add_neon(short *input, short *dq,
+; unsigned char *dest, int stride)
+; r0 short *input,
+; r1 short *dq,
+; r2 unsigned char *dest
+; r3 int stride
+
+|vp8_dequant_idct_add_neon| PROC
+ vld1.16 {q3, q4}, [r0]
+ vld1.16 {q5, q6}, [r1]
+
+ add r1, r2, r3 ; r1 = dest + stride
+ lsl r3, #1 ; 2x stride
+
+ vld1.32 {d14[0]}, [r2], r3
+ vld1.32 {d14[1]}, [r1], r3
+ vld1.32 {d15[0]}, [r2]
+ vld1.32 {d15[1]}, [r1]
+
+ adr r12, cospi8sqrt2minus1 ; pointer to the first constant
+
+ vmul.i16 q1, q3, q5 ;input for short_idct4x4llm_neon
+ vmul.i16 q2, q4, q6
+
+;|short_idct4x4llm_neon| PROC
+ vld1.16 {d0}, [r12]
+ vswp d3, d4 ;q2(vp[4] vp[12])
+
+ vqdmulh.s16 q3, q2, d0[2]
+ vqdmulh.s16 q4, q2, d0[0]
+
+ vqadd.s16 d12, d2, d3 ;a1
+ vqsub.s16 d13, d2, d3 ;b1
+
+ vshr.s16 q3, q3, #1
+ vshr.s16 q4, q4, #1
+
+ vqadd.s16 q3, q3, q2
+ vqadd.s16 q4, q4, q2
+
+ vqsub.s16 d10, d6, d9 ;c1
+ vqadd.s16 d11, d7, d8 ;d1
+
+ vqadd.s16 d2, d12, d11
+ vqadd.s16 d3, d13, d10
+ vqsub.s16 d4, d13, d10
+ vqsub.s16 d5, d12, d11
+
+ vtrn.32 d2, d4
+ vtrn.32 d3, d5
+ vtrn.16 d2, d3
+ vtrn.16 d4, d5
+
+; memset(input, 0, 32) -- 32bytes
+ vmov.i16 q14, #0
+
+ vswp d3, d4
+ vqdmulh.s16 q3, q2, d0[2]
+ vqdmulh.s16 q4, q2, d0[0]
+
+ vqadd.s16 d12, d2, d3 ;a1
+ vqsub.s16 d13, d2, d3 ;b1
+
+ vmov q15, q14
+
+ vshr.s16 q3, q3, #1
+ vshr.s16 q4, q4, #1
+
+ vqadd.s16 q3, q3, q2
+ vqadd.s16 q4, q4, q2
+
+ vqsub.s16 d10, d6, d9 ;c1
+ vqadd.s16 d11, d7, d8 ;d1
+
+ vqadd.s16 d2, d12, d11
+ vqadd.s16 d3, d13, d10
+ vqsub.s16 d4, d13, d10
+ vqsub.s16 d5, d12, d11
+
+ vst1.16 {q14, q15}, [r0]
+
+ vrshr.s16 d2, d2, #3
+ vrshr.s16 d3, d3, #3
+ vrshr.s16 d4, d4, #3
+ vrshr.s16 d5, d5, #3
+
+ vtrn.32 d2, d4
+ vtrn.32 d3, d5
+ vtrn.16 d2, d3
+ vtrn.16 d4, d5
+
+ vaddw.u8 q1, q1, d14
+ vaddw.u8 q2, q2, d15
+
+ sub r2, r2, r3
+ sub r1, r1, r3
+
+ vqmovun.s16 d0, q1
+ vqmovun.s16 d1, q2
+
+ vst1.32 {d0[0]}, [r2], r3
+ vst1.32 {d0[1]}, [r1], r3
+ vst1.32 {d1[0]}, [r2]
+ vst1.32 {d1[1]}, [r1]
+
+ bx lr
+
+ ENDP ; |vp8_dequant_idct_add_neon|
+
+; Constant Pool
+cospi8sqrt2minus1 DCD 0x4e7b4e7b
+sinpi8sqrt2 DCD 0x8a8c8a8c
+
+ END
diff --git a/vp8/decoder/arm/neon/dequantizeb_neon.asm b/libvpx/vp8/common/arm/neon/dequantizeb_neon.asm
similarity index 100%
rename from vp8/decoder/arm/neon/dequantizeb_neon.asm
rename to libvpx/vp8/common/arm/neon/dequantizeb_neon.asm
diff --git a/libvpx/vp8/common/arm/neon/idct_blk_neon.c b/libvpx/vp8/common/arm/neon/idct_blk_neon.c
new file mode 100644
index 0000000..ee7f223
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/idct_blk_neon.c
@@ -0,0 +1,96 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+
+/* place these declarations here because we don't want to maintain them
+ * outside of this scope
+ */
+void idct_dequant_full_2x_neon(short *q, short *dq,
+ unsigned char *dst, int stride);
+void idct_dequant_0_2x_neon(short *q, short dq,
+ unsigned char *dst, int stride);
+
+
+void vp8_dequant_idct_add_y_block_neon(short *q, short *dq,
+ unsigned char *dst,
+ int stride, char *eobs)
+{
+ int i;
+
+ for (i = 0; i < 4; i++)
+ {
+ if (((short *)(eobs))[0])
+ {
+ if (((short *)eobs)[0] & 0xfefe)
+ idct_dequant_full_2x_neon (q, dq, dst, stride);
+ else
+ idct_dequant_0_2x_neon (q, dq[0], dst, stride);
+ }
+
+ if (((short *)(eobs))[1])
+ {
+ if (((short *)eobs)[1] & 0xfefe)
+ idct_dequant_full_2x_neon (q+32, dq, dst+8, stride);
+ else
+ idct_dequant_0_2x_neon (q+32, dq[0], dst+8, stride);
+ }
+ q += 64;
+ dst += 4*stride;
+ eobs += 4;
+ }
+}
+
+void vp8_dequant_idct_add_uv_block_neon(short *q, short *dq,
+ unsigned char *dstu,
+ unsigned char *dstv,
+ int stride, char *eobs)
+{
+ if (((short *)(eobs))[0])
+ {
+ if (((short *)eobs)[0] & 0xfefe)
+ idct_dequant_full_2x_neon (q, dq, dstu, stride);
+ else
+ idct_dequant_0_2x_neon (q, dq[0], dstu, stride);
+ }
+
+ q += 32;
+ dstu += 4*stride;
+
+ if (((short *)(eobs))[1])
+ {
+ if (((short *)eobs)[1] & 0xfefe)
+ idct_dequant_full_2x_neon (q, dq, dstu, stride);
+ else
+ idct_dequant_0_2x_neon (q, dq[0], dstu, stride);
+ }
+
+ q += 32;
+
+ if (((short *)(eobs))[2])
+ {
+ if (((short *)eobs)[2] & 0xfefe)
+ idct_dequant_full_2x_neon (q, dq, dstv, stride);
+ else
+ idct_dequant_0_2x_neon (q, dq[0], dstv, stride);
+ }
+
+ q += 32;
+ dstv += 4*stride;
+
+ if (((short *)(eobs))[3])
+ {
+ if (((short *)eobs)[3] & 0xfefe)
+ idct_dequant_full_2x_neon (q, dq, dstv, stride);
+ else
+ idct_dequant_0_2x_neon (q, dq[0], dstv, stride);
+ }
+}
diff --git a/libvpx/vp8/common/arm/neon/idct_dequant_0_2x_neon.asm b/libvpx/vp8/common/arm/neon/idct_dequant_0_2x_neon.asm
new file mode 100644
index 0000000..6c29c55
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/idct_dequant_0_2x_neon.asm
@@ -0,0 +1,79 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license and patent
+; grant that can be found in the LICENSE file in the root of the source
+; tree. All contributing project authors may be found in the AUTHORS
+; file in the root of the source tree.
+;
+
+
+ EXPORT |idct_dequant_0_2x_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+;void idct_dequant_0_2x_neon(short *q, short dq,
+; unsigned char *dst, int stride);
+; r0 *q
+; r1 dq
+; r2 *dst
+; r3 stride
+|idct_dequant_0_2x_neon| PROC
+ push {r4, r5}
+
+ add r12, r2, #4
+ vld1.32 {d2[0]}, [r2], r3
+ vld1.32 {d8[0]}, [r12], r3
+ vld1.32 {d2[1]}, [r2], r3
+ vld1.32 {d8[1]}, [r12], r3
+ vld1.32 {d4[0]}, [r2], r3
+ vld1.32 {d10[0]}, [r12], r3
+ vld1.32 {d4[1]}, [r2], r3
+ vld1.32 {d10[1]}, [r12], r3
+
+ ldrh r12, [r0] ; lo q
+ ldrh r4, [r0, #32] ; hi q
+ mov r5, #0
+ strh r5, [r0]
+ strh r5, [r0, #32]
+
+ sxth r12, r12 ; lo
+ mul r0, r12, r1
+ add r0, r0, #4
+ asr r0, r0, #3
+ vdup.16 q0, r0
+ sxth r4, r4 ; hi
+ mul r0, r4, r1
+ add r0, r0, #4
+ asr r0, r0, #3
+ vdup.16 q3, r0
+
+ vaddw.u8 q1, q0, d2 ; lo
+ vaddw.u8 q2, q0, d4
+ vaddw.u8 q4, q3, d8 ; hi
+ vaddw.u8 q5, q3, d10
+
+ sub r2, r2, r3, lsl #2 ; dst - 4*stride
+ add r0, r2, #4
+
+ vqmovun.s16 d2, q1 ; lo
+ vqmovun.s16 d4, q2
+ vqmovun.s16 d8, q4 ; hi
+ vqmovun.s16 d10, q5
+
+ vst1.32 {d2[0]}, [r2], r3 ; lo
+ vst1.32 {d8[0]}, [r0], r3 ; hi
+ vst1.32 {d2[1]}, [r2], r3
+ vst1.32 {d8[1]}, [r0], r3
+ vst1.32 {d4[0]}, [r2], r3
+ vst1.32 {d10[0]}, [r0], r3
+ vst1.32 {d4[1]}, [r2]
+ vst1.32 {d10[1]}, [r0]
+
+ pop {r4, r5}
+ bx lr
+
+ ENDP ; |idct_dequant_0_2x_neon|
+ END
diff --git a/libvpx/vp8/common/arm/neon/idct_dequant_full_2x_neon.asm b/libvpx/vp8/common/arm/neon/idct_dequant_full_2x_neon.asm
new file mode 100644
index 0000000..d5dce63
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/idct_dequant_full_2x_neon.asm
@@ -0,0 +1,196 @@
+;
+; Copyright (c) 2010 The Webm project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |idct_dequant_full_2x_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+;void idct_dequant_full_2x_neon(short *q, short *dq,
+; unsigned char *dst, int stride);
+; r0 *q,
+; r1 *dq,
+; r2 *dst
+; r3 stride
+|idct_dequant_full_2x_neon| PROC
+ vld1.16 {q0, q1}, [r1] ; dq (same l/r)
+ vld1.16 {q2, q3}, [r0] ; l q
+ add r0, r0, #32
+ vld1.16 {q4, q5}, [r0] ; r q
+ add r12, r2, #4
+
+ ; interleave the predictors
+ vld1.32 {d28[0]}, [r2], r3 ; l pre
+ vld1.32 {d28[1]}, [r12], r3 ; r pre
+ vld1.32 {d29[0]}, [r2], r3
+ vld1.32 {d29[1]}, [r12], r3
+ vld1.32 {d30[0]}, [r2], r3
+ vld1.32 {d30[1]}, [r12], r3
+ vld1.32 {d31[0]}, [r2], r3
+ vld1.32 {d31[1]}, [r12]
+
+ adr r1, cospi8sqrt2minus1 ; pointer to the first constant
+
+ ; dequant: q[i] = q[i] * dq[i]
+ vmul.i16 q2, q2, q0
+ vmul.i16 q3, q3, q1
+ vmul.i16 q4, q4, q0
+ vmul.i16 q5, q5, q1
+
+ vld1.16 {d0}, [r1]
+
+ ; q2: l0r0 q3: l8r8
+ ; q4: l4r4 q5: l12r12
+ vswp d5, d8
+ vswp d7, d10
+
+ ; _CONSTANTS_ * 4,12 >> 16
+ ; q6: 4 * sinpi : c1/temp1
+ ; q7: 12 * sinpi : d1/temp2
+ ; q8: 4 * cospi
+ ; q9: 12 * cospi
+ vqdmulh.s16 q6, q4, d0[2] ; sinpi8sqrt2
+ vqdmulh.s16 q7, q5, d0[2]
+ vqdmulh.s16 q8, q4, d0[0] ; cospi8sqrt2minus1
+ vqdmulh.s16 q9, q5, d0[0]
+
+ vqadd.s16 q10, q2, q3 ; a1 = 0 + 8
+ vqsub.s16 q11, q2, q3 ; b1 = 0 - 8
+
+ ; vqdmulh only accepts signed values. this was a problem because
+ ; our constant had the high bit set, and was treated as a negative value.
+ ; vqdmulh also doubles the value before it shifts by 16. we need to
+ ; compensate for this. in the case of sinpi8sqrt2, the lowest bit is 0,
+ ; so we can shift the constant without losing precision. this avoids
+ ; shift again afterward, but also avoids the sign issue. win win!
+ ; for cospi8sqrt2minus1 the lowest bit is 1, so we lose precision if we
+ ; pre-shift it
+ vshr.s16 q8, q8, #1
+ vshr.s16 q9, q9, #1
+
+ ; q4: 4 + 4 * cospi : d1/temp1
+ ; q5: 12 + 12 * cospi : c1/temp2
+ vqadd.s16 q4, q4, q8
+ vqadd.s16 q5, q5, q9
+
+ ; c1 = temp1 - temp2
+ ; d1 = temp1 + temp2
+ vqsub.s16 q2, q6, q5
+ vqadd.s16 q3, q4, q7
+
+ ; [0]: a1+d1
+ ; [1]: b1+c1
+ ; [2]: b1-c1
+ ; [3]: a1-d1
+ vqadd.s16 q4, q10, q3
+ vqadd.s16 q5, q11, q2
+ vqsub.s16 q6, q11, q2
+ vqsub.s16 q7, q10, q3
+
+ ; rotate
+ vtrn.32 q4, q6
+ vtrn.32 q5, q7
+ vtrn.16 q4, q5
+ vtrn.16 q6, q7
+ ; idct loop 2
+ ; q4: l 0, 4, 8,12 r 0, 4, 8,12
+ ; q5: l 1, 5, 9,13 r 1, 5, 9,13
+ ; q6: l 2, 6,10,14 r 2, 6,10,14
+ ; q7: l 3, 7,11,15 r 3, 7,11,15
+
+ ; q8: 1 * sinpi : c1/temp1
+ ; q9: 3 * sinpi : d1/temp2
+ ; q10: 1 * cospi
+ ; q11: 3 * cospi
+ vqdmulh.s16 q8, q5, d0[2] ; sinpi8sqrt2
+ vqdmulh.s16 q9, q7, d0[2]
+ vqdmulh.s16 q10, q5, d0[0] ; cospi8sqrt2minus1
+ vqdmulh.s16 q11, q7, d0[0]
+
+ vqadd.s16 q2, q4, q6 ; a1 = 0 + 2
+ vqsub.s16 q3, q4, q6 ; b1 = 0 - 2
+
+ ; see note on shifting above
+ vshr.s16 q10, q10, #1
+ vshr.s16 q11, q11, #1
+
+ ; q10: 1 + 1 * cospi : d1/temp1
+ ; q11: 3 + 3 * cospi : c1/temp2
+ vqadd.s16 q10, q5, q10
+ vqadd.s16 q11, q7, q11
+
+ ; q8: c1 = temp1 - temp2
+ ; q9: d1 = temp1 + temp2
+ vqsub.s16 q8, q8, q11
+ vqadd.s16 q9, q10, q9
+
+ ; a1+d1
+ ; b1+c1
+ ; b1-c1
+ ; a1-d1
+ vqadd.s16 q4, q2, q9
+ vqadd.s16 q5, q3, q8
+ vqsub.s16 q6, q3, q8
+ vqsub.s16 q7, q2, q9
+
+ ; +4 >> 3 (rounding)
+ vrshr.s16 q4, q4, #3 ; lo
+ vrshr.s16 q5, q5, #3
+ vrshr.s16 q6, q6, #3 ; hi
+ vrshr.s16 q7, q7, #3
+
+ vtrn.32 q4, q6
+ vtrn.32 q5, q7
+ vtrn.16 q4, q5
+ vtrn.16 q6, q7
+
+ ; adding pre
+ ; input is still packed. pre was read interleaved
+ vaddw.u8 q4, q4, d28
+ vaddw.u8 q5, q5, d29
+ vaddw.u8 q6, q6, d30
+ vaddw.u8 q7, q7, d31
+
+ vmov.i16 q14, #0
+ vmov q15, q14
+ vst1.16 {q14, q15}, [r0] ; write over high input
+ sub r0, r0, #32
+ vst1.16 {q14, q15}, [r0] ; write over low input
+
+ sub r2, r2, r3, lsl #2 ; dst - 4*stride
+ add r1, r2, #4 ; hi
+
+ ;saturate and narrow
+ vqmovun.s16 d0, q4 ; lo
+ vqmovun.s16 d1, q5
+ vqmovun.s16 d2, q6 ; hi
+ vqmovun.s16 d3, q7
+
+ vst1.32 {d0[0]}, [r2], r3 ; lo
+ vst1.32 {d0[1]}, [r1], r3 ; hi
+ vst1.32 {d1[0]}, [r2], r3
+ vst1.32 {d1[1]}, [r1], r3
+ vst1.32 {d2[0]}, [r2], r3
+ vst1.32 {d2[1]}, [r1], r3
+ vst1.32 {d3[0]}, [r2]
+ vst1.32 {d3[1]}, [r1]
+
+ bx lr
+
+ ENDP ; |idct_dequant_full_2x_neon|
+
+; Constant Pool
+cospi8sqrt2minus1 DCD 0x4e7b
+; because the lowest bit in 0x8a8c is 0, we can pre-shift this
+sinpi8sqrt2 DCD 0x4546
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/iwalsh_neon.asm b/libvpx/vp8/common/arm/neon/iwalsh_neon.asm
new file mode 100644
index 0000000..e8ea2a6
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/iwalsh_neon.asm
@@ -0,0 +1,87 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+ EXPORT |vp8_short_inv_walsh4x4_neon|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA |.text|, CODE, READONLY ; name this block of code
+
+;short vp8_short_inv_walsh4x4_neon(short *input, short *mb_dqcoeff)
+|vp8_short_inv_walsh4x4_neon| PROC
+
+ ; read in all four lines of values: d0->d3
+ vld1.i16 {q0-q1}, [r0@128]
+
+ ; first for loop
+ vadd.s16 d4, d0, d3 ;a = [0] + [12]
+ vadd.s16 d6, d1, d2 ;b = [4] + [8]
+ vsub.s16 d5, d0, d3 ;d = [0] - [12]
+ vsub.s16 d7, d1, d2 ;c = [4] - [8]
+
+ vadd.s16 q0, q2, q3 ; a+b d+c
+ vsub.s16 q1, q2, q3 ; a-b d-c
+
+ vtrn.32 d0, d2 ;d0: 0 1 8 9
+ ;d2: 2 3 10 11
+ vtrn.32 d1, d3 ;d1: 4 5 12 13
+ ;d3: 6 7 14 15
+
+ vtrn.16 d0, d1 ;d0: 0 4 8 12
+ ;d1: 1 5 9 13
+ vtrn.16 d2, d3 ;d2: 2 6 10 14
+ ;d3: 3 7 11 15
+
+ ; second for loop
+
+ vadd.s16 d4, d0, d3 ;a = [0] + [3]
+ vadd.s16 d6, d1, d2 ;b = [1] + [2]
+ vsub.s16 d5, d0, d3 ;d = [0] - [3]
+ vsub.s16 d7, d1, d2 ;c = [1] - [2]
+
+ vmov.i16 q8, #3
+
+ vadd.s16 q0, q2, q3 ; a+b d+c
+ vsub.s16 q1, q2, q3 ; a-b d-c
+
+ vadd.i16 q0, q0, q8 ;e/f += 3
+ vadd.i16 q1, q1, q8 ;g/h += 3
+
+ vshr.s16 q0, q0, #3 ;e/f >> 3
+ vshr.s16 q1, q1, #3 ;g/h >> 3
+
+ mov r2, #64
+ add r3, r1, #32
+
+ vst1.i16 d0[0], [r1],r2
+ vst1.i16 d1[0], [r3],r2
+ vst1.i16 d2[0], [r1],r2
+ vst1.i16 d3[0], [r3],r2
+
+ vst1.i16 d0[1], [r1],r2
+ vst1.i16 d1[1], [r3],r2
+ vst1.i16 d2[1], [r1],r2
+ vst1.i16 d3[1], [r3],r2
+
+ vst1.i16 d0[2], [r1],r2
+ vst1.i16 d1[2], [r3],r2
+ vst1.i16 d2[2], [r1],r2
+ vst1.i16 d3[2], [r3],r2
+
+ vst1.i16 d0[3], [r1],r2
+ vst1.i16 d1[3], [r3],r2
+ vst1.i16 d2[3], [r1]
+ vst1.i16 d3[3], [r3]
+
+ bx lr
+ ENDP ; |vp8_short_inv_walsh4x4_neon|
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/loopfilter_neon.asm b/libvpx/vp8/common/arm/neon/loopfilter_neon.asm
new file mode 100644
index 0000000..e44be0a
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/loopfilter_neon.asm
@@ -0,0 +1,397 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_loop_filter_horizontal_edge_y_neon|
+ EXPORT |vp8_loop_filter_horizontal_edge_uv_neon|
+ EXPORT |vp8_loop_filter_vertical_edge_y_neon|
+ EXPORT |vp8_loop_filter_vertical_edge_uv_neon|
+ ARM
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *src
+; r1 int pitch
+; r2 unsigned char blimit
+; r3 unsigned char limit
+; sp unsigned char thresh,
+|vp8_loop_filter_horizontal_edge_y_neon| PROC
+ push {lr}
+ vdup.u8 q0, r2 ; duplicate blimit
+ vdup.u8 q1, r3 ; duplicate limit
+ sub r2, r0, r1, lsl #2 ; move src pointer down by 4 lines
+ ldr r3, [sp, #4] ; load thresh
+ add r12, r2, r1
+ add r1, r1, r1
+
+ vdup.u8 q2, r3 ; duplicate thresh
+
+ vld1.u8 {q3}, [r2@128], r1 ; p3
+ vld1.u8 {q4}, [r12@128], r1 ; p2
+ vld1.u8 {q5}, [r2@128], r1 ; p1
+ vld1.u8 {q6}, [r12@128], r1 ; p0
+ vld1.u8 {q7}, [r2@128], r1 ; q0
+ vld1.u8 {q8}, [r12@128], r1 ; q1
+ vld1.u8 {q9}, [r2@128] ; q2
+ vld1.u8 {q10}, [r12@128] ; q3
+
+ sub r2, r2, r1, lsl #1
+ sub r12, r12, r1, lsl #1
+
+ bl vp8_loop_filter_neon
+
+ vst1.u8 {q5}, [r2@128], r1 ; store op1
+ vst1.u8 {q6}, [r12@128], r1 ; store op0
+ vst1.u8 {q7}, [r2@128], r1 ; store oq0
+ vst1.u8 {q8}, [r12@128], r1 ; store oq1
+
+ pop {pc}
+ ENDP ; |vp8_loop_filter_horizontal_edge_y_neon|
+
+
+; r0 unsigned char *u,
+; r1 int pitch,
+; r2 unsigned char blimit
+; r3 unsigned char limit
+; sp unsigned char thresh,
+; sp+4 unsigned char *v
+|vp8_loop_filter_horizontal_edge_uv_neon| PROC
+ push {lr}
+ vdup.u8 q0, r2 ; duplicate blimit
+ vdup.u8 q1, r3 ; duplicate limit
+ ldr r12, [sp, #4] ; load thresh
+ ldr r2, [sp, #8] ; load v ptr
+ vdup.u8 q2, r12 ; duplicate thresh
+
+ sub r3, r0, r1, lsl #2 ; move u pointer down by 4 lines
+ sub r12, r2, r1, lsl #2 ; move v pointer down by 4 lines
+
+ vld1.u8 {d6}, [r3@64], r1 ; p3
+ vld1.u8 {d7}, [r12@64], r1 ; p3
+ vld1.u8 {d8}, [r3@64], r1 ; p2
+ vld1.u8 {d9}, [r12@64], r1 ; p2
+ vld1.u8 {d10}, [r3@64], r1 ; p1
+ vld1.u8 {d11}, [r12@64], r1 ; p1
+ vld1.u8 {d12}, [r3@64], r1 ; p0
+ vld1.u8 {d13}, [r12@64], r1 ; p0
+ vld1.u8 {d14}, [r3@64], r1 ; q0
+ vld1.u8 {d15}, [r12@64], r1 ; q0
+ vld1.u8 {d16}, [r3@64], r1 ; q1
+ vld1.u8 {d17}, [r12@64], r1 ; q1
+ vld1.u8 {d18}, [r3@64], r1 ; q2
+ vld1.u8 {d19}, [r12@64], r1 ; q2
+ vld1.u8 {d20}, [r3@64] ; q3
+ vld1.u8 {d21}, [r12@64] ; q3
+
+ bl vp8_loop_filter_neon
+
+ sub r0, r0, r1, lsl #1
+ sub r2, r2, r1, lsl #1
+
+ vst1.u8 {d10}, [r0@64], r1 ; store u op1
+ vst1.u8 {d11}, [r2@64], r1 ; store v op1
+ vst1.u8 {d12}, [r0@64], r1 ; store u op0
+ vst1.u8 {d13}, [r2@64], r1 ; store v op0
+ vst1.u8 {d14}, [r0@64], r1 ; store u oq0
+ vst1.u8 {d15}, [r2@64], r1 ; store v oq0
+ vst1.u8 {d16}, [r0@64] ; store u oq1
+ vst1.u8 {d17}, [r2@64] ; store v oq1
+
+ pop {pc}
+ ENDP ; |vp8_loop_filter_horizontal_edge_uv_neon|
+
+; void vp8_loop_filter_vertical_edge_y_neon(unsigned char *src, int pitch,
+; const signed char *flimit,
+; const signed char *limit,
+; const signed char *thresh,
+; int count)
+; r0 unsigned char *src
+; r1 int pitch
+; r2 unsigned char blimit
+; r3 unsigned char limit
+; sp unsigned char thresh,
+
+|vp8_loop_filter_vertical_edge_y_neon| PROC
+ push {lr}
+ vdup.u8 q0, r2 ; duplicate blimit
+ vdup.u8 q1, r3 ; duplicate limit
+ sub r2, r0, #4 ; src ptr down by 4 columns
+ add r1, r1, r1
+ ldr r3, [sp, #4] ; load thresh
+ add r12, r2, r1, asr #1
+
+ vld1.u8 {d6}, [r2], r1
+ vld1.u8 {d8}, [r12], r1
+ vld1.u8 {d10}, [r2], r1
+ vld1.u8 {d12}, [r12], r1
+ vld1.u8 {d14}, [r2], r1
+ vld1.u8 {d16}, [r12], r1
+ vld1.u8 {d18}, [r2], r1
+ vld1.u8 {d20}, [r12], r1
+
+ vld1.u8 {d7}, [r2], r1 ; load second 8-line src data
+ vld1.u8 {d9}, [r12], r1
+ vld1.u8 {d11}, [r2], r1
+ vld1.u8 {d13}, [r12], r1
+ vld1.u8 {d15}, [r2], r1
+ vld1.u8 {d17}, [r12], r1
+ vld1.u8 {d19}, [r2]
+ vld1.u8 {d21}, [r12]
+
+ ;transpose to 8x16 matrix
+ vtrn.32 q3, q7
+ vtrn.32 q4, q8
+ vtrn.32 q5, q9
+ vtrn.32 q6, q10
+
+ vdup.u8 q2, r3 ; duplicate thresh
+
+ vtrn.16 q3, q5
+ vtrn.16 q4, q6
+ vtrn.16 q7, q9
+ vtrn.16 q8, q10
+
+ vtrn.8 q3, q4
+ vtrn.8 q5, q6
+ vtrn.8 q7, q8
+ vtrn.8 q9, q10
+
+ bl vp8_loop_filter_neon
+
+ vswp d12, d11
+ vswp d16, d13
+
+ sub r0, r0, #2 ; dst ptr
+
+ vswp d14, d12
+ vswp d16, d15
+
+ add r12, r0, r1, asr #1
+
+ ;store op1, op0, oq0, oq1
+ vst4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r1
+ vst4.8 {d10[1], d11[1], d12[1], d13[1]}, [r12], r1
+ vst4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r1
+ vst4.8 {d10[3], d11[3], d12[3], d13[3]}, [r12], r1
+ vst4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r1
+ vst4.8 {d10[5], d11[5], d12[5], d13[5]}, [r12], r1
+ vst4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r1
+ vst4.8 {d10[7], d11[7], d12[7], d13[7]}, [r12], r1
+
+ vst4.8 {d14[0], d15[0], d16[0], d17[0]}, [r0], r1
+ vst4.8 {d14[1], d15[1], d16[1], d17[1]}, [r12], r1
+ vst4.8 {d14[2], d15[2], d16[2], d17[2]}, [r0], r1
+ vst4.8 {d14[3], d15[3], d16[3], d17[3]}, [r12], r1
+ vst4.8 {d14[4], d15[4], d16[4], d17[4]}, [r0], r1
+ vst4.8 {d14[5], d15[5], d16[5], d17[5]}, [r12], r1
+ vst4.8 {d14[6], d15[6], d16[6], d17[6]}, [r0]
+ vst4.8 {d14[7], d15[7], d16[7], d17[7]}, [r12]
+
+ pop {pc}
+ ENDP ; |vp8_loop_filter_vertical_edge_y_neon|
+
+; void vp8_loop_filter_vertical_edge_uv_neon(unsigned char *u, int pitch
+; const signed char *flimit,
+; const signed char *limit,
+; const signed char *thresh,
+; unsigned char *v)
+; r0 unsigned char *u,
+; r1 int pitch,
+; r2 unsigned char blimit
+; r3 unsigned char limit
+; sp unsigned char thresh,
+; sp+4 unsigned char *v
+|vp8_loop_filter_vertical_edge_uv_neon| PROC
+ push {lr}
+ vdup.u8 q0, r2 ; duplicate blimit
+ sub r12, r0, #4 ; move u pointer down by 4 columns
+ ldr r2, [sp, #8] ; load v ptr
+ vdup.u8 q1, r3 ; duplicate limit
+ sub r3, r2, #4 ; move v pointer down by 4 columns
+
+ vld1.u8 {d6}, [r12], r1 ;load u data
+ vld1.u8 {d7}, [r3], r1 ;load v data
+ vld1.u8 {d8}, [r12], r1
+ vld1.u8 {d9}, [r3], r1
+ vld1.u8 {d10}, [r12], r1
+ vld1.u8 {d11}, [r3], r1
+ vld1.u8 {d12}, [r12], r1
+ vld1.u8 {d13}, [r3], r1
+ vld1.u8 {d14}, [r12], r1
+ vld1.u8 {d15}, [r3], r1
+ vld1.u8 {d16}, [r12], r1
+ vld1.u8 {d17}, [r3], r1
+ vld1.u8 {d18}, [r12], r1
+ vld1.u8 {d19}, [r3], r1
+ vld1.u8 {d20}, [r12]
+ vld1.u8 {d21}, [r3]
+
+ ldr r12, [sp, #4] ; load thresh
+
+ ;transpose to 8x16 matrix
+ vtrn.32 q3, q7
+ vtrn.32 q4, q8
+ vtrn.32 q5, q9
+ vtrn.32 q6, q10
+
+ vdup.u8 q2, r12 ; duplicate thresh
+
+ vtrn.16 q3, q5
+ vtrn.16 q4, q6
+ vtrn.16 q7, q9
+ vtrn.16 q8, q10
+
+ vtrn.8 q3, q4
+ vtrn.8 q5, q6
+ vtrn.8 q7, q8
+ vtrn.8 q9, q10
+
+ bl vp8_loop_filter_neon
+
+ vswp d12, d11
+ vswp d16, d13
+ vswp d14, d12
+ vswp d16, d15
+
+ sub r0, r0, #2
+ sub r2, r2, #2
+
+ ;store op1, op0, oq0, oq1
+ vst4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r1
+ vst4.8 {d14[0], d15[0], d16[0], d17[0]}, [r2], r1
+ vst4.8 {d10[1], d11[1], d12[1], d13[1]}, [r0], r1
+ vst4.8 {d14[1], d15[1], d16[1], d17[1]}, [r2], r1
+ vst4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r1
+ vst4.8 {d14[2], d15[2], d16[2], d17[2]}, [r2], r1
+ vst4.8 {d10[3], d11[3], d12[3], d13[3]}, [r0], r1
+ vst4.8 {d14[3], d15[3], d16[3], d17[3]}, [r2], r1
+ vst4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r1
+ vst4.8 {d14[4], d15[4], d16[4], d17[4]}, [r2], r1
+ vst4.8 {d10[5], d11[5], d12[5], d13[5]}, [r0], r1
+ vst4.8 {d14[5], d15[5], d16[5], d17[5]}, [r2], r1
+ vst4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r1
+ vst4.8 {d14[6], d15[6], d16[6], d17[6]}, [r2], r1
+ vst4.8 {d10[7], d11[7], d12[7], d13[7]}, [r0]
+ vst4.8 {d14[7], d15[7], d16[7], d17[7]}, [r2]
+
+ pop {pc}
+ ENDP ; |vp8_loop_filter_vertical_edge_uv_neon|
+
+; void vp8_loop_filter_neon();
+; This is a helper function for the loopfilters. The invidual functions do the
+; necessary load, transpose (if necessary) and store.
+
+; r0-r3 PRESERVE
+; q0 flimit
+; q1 limit
+; q2 thresh
+; q3 p3
+; q4 p2
+; q5 p1
+; q6 p0
+; q7 q0
+; q8 q1
+; q9 q2
+; q10 q3
+|vp8_loop_filter_neon| PROC
+
+ ; vp8_filter_mask
+ vabd.u8 q11, q3, q4 ; abs(p3 - p2)
+ vabd.u8 q12, q4, q5 ; abs(p2 - p1)
+ vabd.u8 q13, q5, q6 ; abs(p1 - p0)
+ vabd.u8 q14, q8, q7 ; abs(q1 - q0)
+ vabd.u8 q3, q9, q8 ; abs(q2 - q1)
+ vabd.u8 q4, q10, q9 ; abs(q3 - q2)
+
+ vmax.u8 q11, q11, q12
+ vmax.u8 q12, q13, q14
+ vmax.u8 q3, q3, q4
+ vmax.u8 q15, q11, q12
+
+ vabd.u8 q9, q6, q7 ; abs(p0 - q0)
+
+ ; vp8_hevmask
+ vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
+ vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
+ vmax.u8 q15, q15, q3
+
+ vmov.u8 q10, #0x80 ; 0x80
+
+ vabd.u8 q2, q5, q8 ; a = abs(p1 - q1)
+ vqadd.u8 q9, q9, q9 ; b = abs(p0 - q0) * 2
+
+ vcge.u8 q15, q1, q15
+
+ ; vp8_filter() function
+ ; convert to signed
+ veor q7, q7, q10 ; qs0
+ vshr.u8 q2, q2, #1 ; a = a / 2
+ veor q6, q6, q10 ; ps0
+
+ veor q5, q5, q10 ; ps1
+ vqadd.u8 q9, q9, q2 ; a = b + a
+
+ veor q8, q8, q10 ; qs1
+
+ vmov.u8 q10, #3 ; #3
+
+ vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
+ vsubl.s8 q11, d15, d13
+
+ vcge.u8 q9, q0, q9 ; (a > flimit * 2 + limit) * -1
+
+ vmovl.u8 q4, d20
+
+ vqsub.s8 q1, q5, q8 ; vp8_filter = clamp(ps1-qs1)
+ vorr q14, q13, q14 ; vp8_hevmask
+
+ vmul.i16 q2, q2, q4 ; 3 * ( qs0 - ps0)
+ vmul.i16 q11, q11, q4
+
+ vand q1, q1, q14 ; vp8_filter &= hev
+ vand q15, q15, q9 ; vp8_filter_mask
+
+ vaddw.s8 q2, q2, d2
+ vaddw.s8 q11, q11, d3
+
+ vmov.u8 q9, #4 ; #4
+
+ ; vp8_filter = clamp(vp8_filter + 3 * ( qs0 - ps0))
+ vqmovn.s16 d2, q2
+ vqmovn.s16 d3, q11
+ vand q1, q1, q15 ; vp8_filter &= mask
+
+ vqadd.s8 q2, q1, q10 ; Filter2 = clamp(vp8_filter+3)
+ vqadd.s8 q1, q1, q9 ; Filter1 = clamp(vp8_filter+4)
+ vshr.s8 q2, q2, #3 ; Filter2 >>= 3
+ vshr.s8 q1, q1, #3 ; Filter1 >>= 3
+
+
+ vqadd.s8 q11, q6, q2 ; u = clamp(ps0 + Filter2)
+ vqsub.s8 q10, q7, q1 ; u = clamp(qs0 - Filter1)
+
+ ; outer tap adjustments: ++vp8_filter >> 1
+ vrshr.s8 q1, q1, #1
+ vbic q1, q1, q14 ; vp8_filter &= ~hev
+ vmov.u8 q0, #0x80 ; 0x80
+ vqadd.s8 q13, q5, q1 ; u = clamp(ps1 + vp8_filter)
+ vqsub.s8 q12, q8, q1 ; u = clamp(qs1 - vp8_filter)
+
+ veor q6, q11, q0 ; *op0 = u^0x80
+ veor q7, q10, q0 ; *oq0 = u^0x80
+ veor q5, q13, q0 ; *op1 = u^0x80
+ veor q8, q12, q0 ; *oq1 = u^0x80
+
+ bx lr
+ ENDP ; |vp8_loop_filter_horizontal_edge_y_neon|
+
+;-----------------
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm b/libvpx/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm
new file mode 100644
index 0000000..adf848b
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm
@@ -0,0 +1,117 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ ;EXPORT |vp8_loop_filter_simple_horizontal_edge_neon|
+ EXPORT |vp8_loop_filter_bhs_neon|
+ EXPORT |vp8_loop_filter_mbhs_neon|
+ ARM
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *s, PRESERVE
+; r1 int p, PRESERVE
+; q1 limit, PRESERVE
+
+|vp8_loop_filter_simple_horizontal_edge_neon| PROC
+
+ sub r3, r0, r1, lsl #1 ; move src pointer down by 2 lines
+
+ vld1.u8 {q7}, [r0@128], r1 ; q0
+ vld1.u8 {q5}, [r3@128], r1 ; p0
+ vld1.u8 {q8}, [r0@128] ; q1
+ vld1.u8 {q6}, [r3@128] ; p1
+
+ vabd.u8 q15, q6, q7 ; abs(p0 - q0)
+ vabd.u8 q14, q5, q8 ; abs(p1 - q1)
+
+ vqadd.u8 q15, q15, q15 ; abs(p0 - q0) * 2
+ vshr.u8 q14, q14, #1 ; abs(p1 - q1) / 2
+ vmov.u8 q0, #0x80 ; 0x80
+ vmov.s16 q13, #3
+ vqadd.u8 q15, q15, q14 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
+
+ veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
+ veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
+ veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
+ veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
+
+ vcge.u8 q15, q1, q15 ; (abs(p0 - q0)*2 + abs(p1-q1)/2 > limit)*-1
+
+ vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
+ vsubl.s8 q3, d15, d13
+
+ vqsub.s8 q4, q5, q8 ; q4: vp8_filter = vp8_signed_char_clamp(ps1-qs1)
+
+ vmul.s16 q2, q2, q13 ; 3 * ( qs0 - ps0)
+ vmul.s16 q3, q3, q13
+
+ vmov.u8 q10, #0x03 ; 0x03
+ vmov.u8 q9, #0x04 ; 0x04
+
+ vaddw.s8 q2, q2, d8 ; vp8_filter + 3 * ( qs0 - ps0)
+ vaddw.s8 q3, q3, d9
+
+ vqmovn.s16 d8, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
+ vqmovn.s16 d9, q3
+
+ vand q14, q4, q15 ; vp8_filter &= mask
+
+ vqadd.s8 q2, q14, q10 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
+ vqadd.s8 q3, q14, q9 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
+ vshr.s8 q2, q2, #3 ; Filter2 >>= 3
+ vshr.s8 q4, q3, #3 ; Filter1 >>= 3
+
+ sub r0, r0, r1
+
+ ;calculate output
+ vqadd.s8 q11, q6, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
+ vqsub.s8 q10, q7, q4 ; u = vp8_signed_char_clamp(qs0 - Filter1)
+
+ veor q6, q11, q0 ; *op0 = u^0x80
+ veor q7, q10, q0 ; *oq0 = u^0x80
+
+ vst1.u8 {q6}, [r3@128] ; store op0
+ vst1.u8 {q7}, [r0@128] ; store oq0
+
+ bx lr
+ ENDP ; |vp8_loop_filter_simple_horizontal_edge_neon|
+
+; r0 unsigned char *y
+; r1 int ystride
+; r2 const unsigned char *blimit
+
+|vp8_loop_filter_bhs_neon| PROC
+ push {r4, lr}
+ ldrb r3, [r2] ; load blim from mem
+ vdup.s8 q1, r3 ; duplicate blim
+
+ add r0, r0, r1, lsl #2 ; src = y_ptr + 4 * y_stride
+ bl vp8_loop_filter_simple_horizontal_edge_neon
+ ; vp8_loop_filter_simple_horizontal_edge_neon preserves r0, r1 and q1
+ add r0, r0, r1, lsl #2 ; src = y_ptr + 8* y_stride
+ bl vp8_loop_filter_simple_horizontal_edge_neon
+ add r0, r0, r1, lsl #2 ; src = y_ptr + 12 * y_stride
+ pop {r4, lr}
+ b vp8_loop_filter_simple_horizontal_edge_neon
+ ENDP ;|vp8_loop_filter_bhs_neon|
+
+; r0 unsigned char *y
+; r1 int ystride
+; r2 const unsigned char *blimit
+
+|vp8_loop_filter_mbhs_neon| PROC
+ ldrb r3, [r2] ; load blim from mem
+ vdup.s8 q1, r3 ; duplicate mblim
+ b vp8_loop_filter_simple_horizontal_edge_neon
+ ENDP ;|vp8_loop_filter_bhs_neon|
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm b/libvpx/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm
new file mode 100644
index 0000000..e690df2
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm
@@ -0,0 +1,154 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ ;EXPORT |vp8_loop_filter_simple_vertical_edge_neon|
+ EXPORT |vp8_loop_filter_bvs_neon|
+ EXPORT |vp8_loop_filter_mbvs_neon|
+ ARM
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *s, PRESERVE
+; r1 int p, PRESERVE
+; q1 limit, PRESERVE
+
+|vp8_loop_filter_simple_vertical_edge_neon| PROC
+ sub r0, r0, #2 ; move src pointer down by 2 columns
+ add r12, r1, r1
+ add r3, r0, r1
+
+ vld4.8 {d6[0], d7[0], d8[0], d9[0]}, [r0], r12
+ vld4.8 {d6[1], d7[1], d8[1], d9[1]}, [r3], r12
+ vld4.8 {d6[2], d7[2], d8[2], d9[2]}, [r0], r12
+ vld4.8 {d6[3], d7[3], d8[3], d9[3]}, [r3], r12
+ vld4.8 {d6[4], d7[4], d8[4], d9[4]}, [r0], r12
+ vld4.8 {d6[5], d7[5], d8[5], d9[5]}, [r3], r12
+ vld4.8 {d6[6], d7[6], d8[6], d9[6]}, [r0], r12
+ vld4.8 {d6[7], d7[7], d8[7], d9[7]}, [r3], r12
+
+ vld4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r12
+ vld4.8 {d10[1], d11[1], d12[1], d13[1]}, [r3], r12
+ vld4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r12
+ vld4.8 {d10[3], d11[3], d12[3], d13[3]}, [r3], r12
+ vld4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r12
+ vld4.8 {d10[5], d11[5], d12[5], d13[5]}, [r3], r12
+ vld4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r12
+ vld4.8 {d10[7], d11[7], d12[7], d13[7]}, [r3]
+
+ vswp d7, d10
+ vswp d12, d9
+
+ ;vp8_filter_mask() function
+ ;vp8_hevmask() function
+ sub r0, r0, r1, lsl #4
+ vabd.u8 q15, q5, q4 ; abs(p0 - q0)
+ vabd.u8 q14, q3, q6 ; abs(p1 - q1)
+
+ vqadd.u8 q15, q15, q15 ; abs(p0 - q0) * 2
+ vshr.u8 q14, q14, #1 ; abs(p1 - q1) / 2
+ vmov.u8 q0, #0x80 ; 0x80
+ vmov.s16 q11, #3
+ vqadd.u8 q15, q15, q14 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
+
+ veor q4, q4, q0 ; qs0: q0 offset to convert to a signed value
+ veor q5, q5, q0 ; ps0: p0 offset to convert to a signed value
+ veor q3, q3, q0 ; ps1: p1 offset to convert to a signed value
+ veor q6, q6, q0 ; qs1: q1 offset to convert to a signed value
+
+ vcge.u8 q15, q1, q15 ; abs(p0 - q0)*2 + abs(p1-q1)/2 > flimit*2 + limit)*-1
+
+ vsubl.s8 q2, d8, d10 ; ( qs0 - ps0)
+ vsubl.s8 q13, d9, d11
+
+ vqsub.s8 q14, q3, q6 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
+
+ vmul.s16 q2, q2, q11 ; 3 * ( qs0 - ps0)
+ vmul.s16 q13, q13, q11
+
+ vmov.u8 q11, #0x03 ; 0x03
+ vmov.u8 q12, #0x04 ; 0x04
+
+ vaddw.s8 q2, q2, d28 ; vp8_filter + 3 * ( qs0 - ps0)
+ vaddw.s8 q13, q13, d29
+
+ vqmovn.s16 d28, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
+ vqmovn.s16 d29, q13
+
+ add r0, r0, #1
+ add r3, r0, r1
+
+ vand q14, q14, q15 ; vp8_filter &= mask
+
+ vqadd.s8 q2, q14, q11 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
+ vqadd.s8 q3, q14, q12 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
+ vshr.s8 q2, q2, #3 ; Filter2 >>= 3
+ vshr.s8 q14, q3, #3 ; Filter1 >>= 3
+
+ ;calculate output
+ vqadd.s8 q11, q5, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
+ vqsub.s8 q10, q4, q14 ; u = vp8_signed_char_clamp(qs0 - Filter1)
+
+ veor q6, q11, q0 ; *op0 = u^0x80
+ veor q7, q10, q0 ; *oq0 = u^0x80
+ add r12, r1, r1
+ vswp d13, d14
+
+ ;store op1, op0, oq0, oq1
+ vst2.8 {d12[0], d13[0]}, [r0], r12
+ vst2.8 {d12[1], d13[1]}, [r3], r12
+ vst2.8 {d12[2], d13[2]}, [r0], r12
+ vst2.8 {d12[3], d13[3]}, [r3], r12
+ vst2.8 {d12[4], d13[4]}, [r0], r12
+ vst2.8 {d12[5], d13[5]}, [r3], r12
+ vst2.8 {d12[6], d13[6]}, [r0], r12
+ vst2.8 {d12[7], d13[7]}, [r3], r12
+ vst2.8 {d14[0], d15[0]}, [r0], r12
+ vst2.8 {d14[1], d15[1]}, [r3], r12
+ vst2.8 {d14[2], d15[2]}, [r0], r12
+ vst2.8 {d14[3], d15[3]}, [r3], r12
+ vst2.8 {d14[4], d15[4]}, [r0], r12
+ vst2.8 {d14[5], d15[5]}, [r3], r12
+ vst2.8 {d14[6], d15[6]}, [r0], r12
+ vst2.8 {d14[7], d15[7]}, [r3]
+
+ bx lr
+ ENDP ; |vp8_loop_filter_simple_vertical_edge_neon|
+
+; r0 unsigned char *y
+; r1 int ystride
+; r2 const unsigned char *blimit
+
+|vp8_loop_filter_bvs_neon| PROC
+ push {r4, lr}
+ ldrb r3, [r2] ; load blim from mem
+ mov r4, r0
+ add r0, r0, #4
+ vdup.s8 q1, r3 ; duplicate blim
+ bl vp8_loop_filter_simple_vertical_edge_neon
+ ; vp8_loop_filter_simple_vertical_edge_neon preserves r1 and q1
+ add r0, r4, #8
+ bl vp8_loop_filter_simple_vertical_edge_neon
+ add r0, r4, #12
+ pop {r4, lr}
+ b vp8_loop_filter_simple_vertical_edge_neon
+ ENDP ;|vp8_loop_filter_bvs_neon|
+
+; r0 unsigned char *y
+; r1 int ystride
+; r2 const unsigned char *blimit
+
+|vp8_loop_filter_mbvs_neon| PROC
+ ldrb r3, [r2] ; load mblim from mem
+ vdup.s8 q1, r3 ; duplicate mblim
+ b vp8_loop_filter_simple_vertical_edge_neon
+ ENDP ;|vp8_loop_filter_bvs_neon|
+ END
diff --git a/libvpx/vp8/common/arm/neon/mbloopfilter_neon.asm b/libvpx/vp8/common/arm/neon/mbloopfilter_neon.asm
new file mode 100644
index 0000000..f41c156
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/mbloopfilter_neon.asm
@@ -0,0 +1,469 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_mbloop_filter_horizontal_edge_y_neon|
+ EXPORT |vp8_mbloop_filter_horizontal_edge_uv_neon|
+ EXPORT |vp8_mbloop_filter_vertical_edge_y_neon|
+ EXPORT |vp8_mbloop_filter_vertical_edge_uv_neon|
+ ARM
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; void vp8_mbloop_filter_horizontal_edge_y_neon(unsigned char *src, int pitch,
+; const unsigned char *blimit,
+; const unsigned char *limit,
+; const unsigned char *thresh)
+; r0 unsigned char *src,
+; r1 int pitch,
+; r2 unsigned char blimit
+; r3 unsigned char limit
+; sp unsigned char thresh,
+|vp8_mbloop_filter_horizontal_edge_y_neon| PROC
+ push {lr}
+ add r1, r1, r1 ; double stride
+ ldr r12, [sp, #4] ; load thresh
+ sub r0, r0, r1, lsl #1 ; move src pointer down by 4 lines
+ vdup.u8 q2, r12 ; thresh
+ add r12, r0, r1, lsr #1 ; move src pointer up by 1 line
+
+ vld1.u8 {q3}, [r0@128], r1 ; p3
+ vld1.u8 {q4}, [r12@128], r1 ; p2
+ vld1.u8 {q5}, [r0@128], r1 ; p1
+ vld1.u8 {q6}, [r12@128], r1 ; p0
+ vld1.u8 {q7}, [r0@128], r1 ; q0
+ vld1.u8 {q8}, [r12@128], r1 ; q1
+ vld1.u8 {q9}, [r0@128], r1 ; q2
+ vld1.u8 {q10}, [r12@128], r1 ; q3
+
+ bl vp8_mbloop_filter_neon
+
+ sub r12, r12, r1, lsl #2
+ add r0, r12, r1, lsr #1
+
+ vst1.u8 {q4}, [r12@128],r1 ; store op2
+ vst1.u8 {q5}, [r0@128],r1 ; store op1
+ vst1.u8 {q6}, [r12@128], r1 ; store op0
+ vst1.u8 {q7}, [r0@128],r1 ; store oq0
+ vst1.u8 {q8}, [r12@128] ; store oq1
+ vst1.u8 {q9}, [r0@128] ; store oq2
+
+ pop {pc}
+ ENDP ; |vp8_mbloop_filter_horizontal_edge_y_neon|
+
+; void vp8_mbloop_filter_horizontal_edge_uv_neon(unsigned char *u, int pitch,
+; const unsigned char *blimit,
+; const unsigned char *limit,
+; const unsigned char *thresh,
+; unsigned char *v)
+; r0 unsigned char *u,
+; r1 int pitch,
+; r2 unsigned char blimit
+; r3 unsigned char limit
+; sp unsigned char thresh,
+; sp+4 unsigned char *v
+
+|vp8_mbloop_filter_horizontal_edge_uv_neon| PROC
+ push {lr}
+ ldr r12, [sp, #4] ; load thresh
+ sub r0, r0, r1, lsl #2 ; move u pointer down by 4 lines
+ vdup.u8 q2, r12 ; thresh
+ ldr r12, [sp, #8] ; load v ptr
+ sub r12, r12, r1, lsl #2 ; move v pointer down by 4 lines
+
+ vld1.u8 {d6}, [r0@64], r1 ; p3
+ vld1.u8 {d7}, [r12@64], r1 ; p3
+ vld1.u8 {d8}, [r0@64], r1 ; p2
+ vld1.u8 {d9}, [r12@64], r1 ; p2
+ vld1.u8 {d10}, [r0@64], r1 ; p1
+ vld1.u8 {d11}, [r12@64], r1 ; p1
+ vld1.u8 {d12}, [r0@64], r1 ; p0
+ vld1.u8 {d13}, [r12@64], r1 ; p0
+ vld1.u8 {d14}, [r0@64], r1 ; q0
+ vld1.u8 {d15}, [r12@64], r1 ; q0
+ vld1.u8 {d16}, [r0@64], r1 ; q1
+ vld1.u8 {d17}, [r12@64], r1 ; q1
+ vld1.u8 {d18}, [r0@64], r1 ; q2
+ vld1.u8 {d19}, [r12@64], r1 ; q2
+ vld1.u8 {d20}, [r0@64], r1 ; q3
+ vld1.u8 {d21}, [r12@64], r1 ; q3
+
+ bl vp8_mbloop_filter_neon
+
+ sub r0, r0, r1, lsl #3
+ sub r12, r12, r1, lsl #3
+
+ add r0, r0, r1
+ add r12, r12, r1
+
+ vst1.u8 {d8}, [r0@64], r1 ; store u op2
+ vst1.u8 {d9}, [r12@64], r1 ; store v op2
+ vst1.u8 {d10}, [r0@64], r1 ; store u op1
+ vst1.u8 {d11}, [r12@64], r1 ; store v op1
+ vst1.u8 {d12}, [r0@64], r1 ; store u op0
+ vst1.u8 {d13}, [r12@64], r1 ; store v op0
+ vst1.u8 {d14}, [r0@64], r1 ; store u oq0
+ vst1.u8 {d15}, [r12@64], r1 ; store v oq0
+ vst1.u8 {d16}, [r0@64], r1 ; store u oq1
+ vst1.u8 {d17}, [r12@64], r1 ; store v oq1
+ vst1.u8 {d18}, [r0@64], r1 ; store u oq2
+ vst1.u8 {d19}, [r12@64], r1 ; store v oq2
+
+ pop {pc}
+ ENDP ; |vp8_mbloop_filter_horizontal_edge_uv_neon|
+
+; void vp8_mbloop_filter_vertical_edge_y_neon(unsigned char *src, int pitch,
+; const unsigned char *blimit,
+; const unsigned char *limit,
+; const unsigned char *thresh)
+; r0 unsigned char *src,
+; r1 int pitch,
+; r2 unsigned char blimit
+; r3 unsigned char limit
+; sp unsigned char thresh,
+|vp8_mbloop_filter_vertical_edge_y_neon| PROC
+ push {lr}
+ ldr r12, [sp, #4] ; load thresh
+ sub r0, r0, #4 ; move src pointer down by 4 columns
+ vdup.s8 q2, r12 ; thresh
+ add r12, r0, r1, lsl #3 ; move src pointer down by 8 lines
+
+ vld1.u8 {d6}, [r0], r1 ; load first 8-line src data
+ vld1.u8 {d7}, [r12], r1 ; load second 8-line src data
+ vld1.u8 {d8}, [r0], r1
+ vld1.u8 {d9}, [r12], r1
+ vld1.u8 {d10}, [r0], r1
+ vld1.u8 {d11}, [r12], r1
+ vld1.u8 {d12}, [r0], r1
+ vld1.u8 {d13}, [r12], r1
+ vld1.u8 {d14}, [r0], r1
+ vld1.u8 {d15}, [r12], r1
+ vld1.u8 {d16}, [r0], r1
+ vld1.u8 {d17}, [r12], r1
+ vld1.u8 {d18}, [r0], r1
+ vld1.u8 {d19}, [r12], r1
+ vld1.u8 {d20}, [r0], r1
+ vld1.u8 {d21}, [r12], r1
+
+ ;transpose to 8x16 matrix
+ vtrn.32 q3, q7
+ vtrn.32 q4, q8
+ vtrn.32 q5, q9
+ vtrn.32 q6, q10
+
+ vtrn.16 q3, q5
+ vtrn.16 q4, q6
+ vtrn.16 q7, q9
+ vtrn.16 q8, q10
+
+ vtrn.8 q3, q4
+ vtrn.8 q5, q6
+ vtrn.8 q7, q8
+ vtrn.8 q9, q10
+
+ sub r0, r0, r1, lsl #3
+
+ bl vp8_mbloop_filter_neon
+
+ sub r12, r12, r1, lsl #3
+
+ ;transpose to 16x8 matrix
+ vtrn.32 q3, q7
+ vtrn.32 q4, q8
+ vtrn.32 q5, q9
+ vtrn.32 q6, q10
+
+ vtrn.16 q3, q5
+ vtrn.16 q4, q6
+ vtrn.16 q7, q9
+ vtrn.16 q8, q10
+
+ vtrn.8 q3, q4
+ vtrn.8 q5, q6
+ vtrn.8 q7, q8
+ vtrn.8 q9, q10
+
+ ;store op2, op1, op0, oq0, oq1, oq2
+ vst1.8 {d6}, [r0], r1
+ vst1.8 {d7}, [r12], r1
+ vst1.8 {d8}, [r0], r1
+ vst1.8 {d9}, [r12], r1
+ vst1.8 {d10}, [r0], r1
+ vst1.8 {d11}, [r12], r1
+ vst1.8 {d12}, [r0], r1
+ vst1.8 {d13}, [r12], r1
+ vst1.8 {d14}, [r0], r1
+ vst1.8 {d15}, [r12], r1
+ vst1.8 {d16}, [r0], r1
+ vst1.8 {d17}, [r12], r1
+ vst1.8 {d18}, [r0], r1
+ vst1.8 {d19}, [r12], r1
+ vst1.8 {d20}, [r0]
+ vst1.8 {d21}, [r12]
+
+ pop {pc}
+ ENDP ; |vp8_mbloop_filter_vertical_edge_y_neon|
+
+; void vp8_mbloop_filter_vertical_edge_uv_neon(unsigned char *u, int pitch,
+; const unsigned char *blimit,
+; const unsigned char *limit,
+; const unsigned char *thresh,
+; unsigned char *v)
+; r0 unsigned char *u,
+; r1 int pitch,
+; r2 const signed char *flimit,
+; r3 const signed char *limit,
+; sp const signed char *thresh,
+; sp+4 unsigned char *v
+|vp8_mbloop_filter_vertical_edge_uv_neon| PROC
+ push {lr}
+ ldr r12, [sp, #4] ; load thresh
+ sub r0, r0, #4 ; move u pointer down by 4 columns
+ vdup.u8 q2, r12 ; thresh
+ ldr r12, [sp, #8] ; load v ptr
+ sub r12, r12, #4 ; move v pointer down by 4 columns
+
+ vld1.u8 {d6}, [r0], r1 ;load u data
+ vld1.u8 {d7}, [r12], r1 ;load v data
+ vld1.u8 {d8}, [r0], r1
+ vld1.u8 {d9}, [r12], r1
+ vld1.u8 {d10}, [r0], r1
+ vld1.u8 {d11}, [r12], r1
+ vld1.u8 {d12}, [r0], r1
+ vld1.u8 {d13}, [r12], r1
+ vld1.u8 {d14}, [r0], r1
+ vld1.u8 {d15}, [r12], r1
+ vld1.u8 {d16}, [r0], r1
+ vld1.u8 {d17}, [r12], r1
+ vld1.u8 {d18}, [r0], r1
+ vld1.u8 {d19}, [r12], r1
+ vld1.u8 {d20}, [r0], r1
+ vld1.u8 {d21}, [r12], r1
+
+ ;transpose to 8x16 matrix
+ vtrn.32 q3, q7
+ vtrn.32 q4, q8
+ vtrn.32 q5, q9
+ vtrn.32 q6, q10
+
+ vtrn.16 q3, q5
+ vtrn.16 q4, q6
+ vtrn.16 q7, q9
+ vtrn.16 q8, q10
+
+ vtrn.8 q3, q4
+ vtrn.8 q5, q6
+ vtrn.8 q7, q8
+ vtrn.8 q9, q10
+
+ sub r0, r0, r1, lsl #3
+
+ bl vp8_mbloop_filter_neon
+
+ sub r12, r12, r1, lsl #3
+
+ ;transpose to 16x8 matrix
+ vtrn.32 q3, q7
+ vtrn.32 q4, q8
+ vtrn.32 q5, q9
+ vtrn.32 q6, q10
+
+ vtrn.16 q3, q5
+ vtrn.16 q4, q6
+ vtrn.16 q7, q9
+ vtrn.16 q8, q10
+
+ vtrn.8 q3, q4
+ vtrn.8 q5, q6
+ vtrn.8 q7, q8
+ vtrn.8 q9, q10
+
+ ;store op2, op1, op0, oq0, oq1, oq2
+ vst1.8 {d6}, [r0], r1
+ vst1.8 {d7}, [r12], r1
+ vst1.8 {d8}, [r0], r1
+ vst1.8 {d9}, [r12], r1
+ vst1.8 {d10}, [r0], r1
+ vst1.8 {d11}, [r12], r1
+ vst1.8 {d12}, [r0], r1
+ vst1.8 {d13}, [r12], r1
+ vst1.8 {d14}, [r0], r1
+ vst1.8 {d15}, [r12], r1
+ vst1.8 {d16}, [r0], r1
+ vst1.8 {d17}, [r12], r1
+ vst1.8 {d18}, [r0], r1
+ vst1.8 {d19}, [r12], r1
+ vst1.8 {d20}, [r0]
+ vst1.8 {d21}, [r12]
+
+ pop {pc}
+ ENDP ; |vp8_mbloop_filter_vertical_edge_uv_neon|
+
+; void vp8_mbloop_filter_neon()
+; This is a helper function for the macroblock loopfilters. The individual
+; functions do the necessary load, transpose (if necessary), preserve (if
+; necessary) and store.
+
+; r0,r1 PRESERVE
+; r2 mblimit
+; r3 limit
+
+; q2 thresh
+; q3 p3 PRESERVE
+; q4 p2
+; q5 p1
+; q6 p0
+; q7 q0
+; q8 q1
+; q9 q2
+; q10 q3 PRESERVE
+
+|vp8_mbloop_filter_neon| PROC
+
+ ; vp8_filter_mask
+ vabd.u8 q11, q3, q4 ; abs(p3 - p2)
+ vabd.u8 q12, q4, q5 ; abs(p2 - p1)
+ vabd.u8 q13, q5, q6 ; abs(p1 - p0)
+ vabd.u8 q14, q8, q7 ; abs(q1 - q0)
+ vabd.u8 q1, q9, q8 ; abs(q2 - q1)
+ vabd.u8 q0, q10, q9 ; abs(q3 - q2)
+
+ vmax.u8 q11, q11, q12
+ vmax.u8 q12, q13, q14
+ vmax.u8 q1, q1, q0
+ vmax.u8 q15, q11, q12
+
+ vabd.u8 q12, q6, q7 ; abs(p0 - q0)
+
+ ; vp8_hevmask
+ vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh) * -1
+ vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh) * -1
+ vmax.u8 q15, q15, q1
+
+ vdup.u8 q1, r3 ; limit
+ vdup.u8 q2, r2 ; mblimit
+
+ vmov.u8 q0, #0x80 ; 0x80
+
+ vcge.u8 q15, q1, q15
+
+ vabd.u8 q1, q5, q8 ; a = abs(p1 - q1)
+ vqadd.u8 q12, q12, q12 ; b = abs(p0 - q0) * 2
+ vmov.u16 q11, #3 ; #3
+
+ ; vp8_filter
+ ; convert to signed
+ veor q7, q7, q0 ; qs0
+ vshr.u8 q1, q1, #1 ; a = a / 2
+ veor q6, q6, q0 ; ps0
+ veor q5, q5, q0 ; ps1
+
+ vqadd.u8 q12, q12, q1 ; a = b + a
+
+ veor q8, q8, q0 ; qs1
+ veor q4, q4, q0 ; ps2
+ veor q9, q9, q0 ; qs2
+
+ vorr q14, q13, q14 ; vp8_hevmask
+
+ vcge.u8 q12, q2, q12 ; (a > flimit * 2 + limit) * -1
+
+ vsubl.s8 q2, d14, d12 ; qs0 - ps0
+ vsubl.s8 q13, d15, d13
+
+ vqsub.s8 q1, q5, q8 ; vp8_filter = clamp(ps1-qs1)
+
+ vmul.i16 q2, q2, q11 ; 3 * ( qs0 - ps0)
+
+ vand q15, q15, q12 ; vp8_filter_mask
+
+ vmul.i16 q13, q13, q11
+
+ vmov.u8 q12, #3 ; #3
+
+ vaddw.s8 q2, q2, d2 ; vp8_filter + 3 * ( qs0 - ps0)
+ vaddw.s8 q13, q13, d3
+
+ vmov.u8 q11, #4 ; #4
+
+ ; vp8_filter = clamp(vp8_filter + 3 * ( qs0 - ps0))
+ vqmovn.s16 d2, q2
+ vqmovn.s16 d3, q13
+
+ vand q1, q1, q15 ; vp8_filter &= mask
+
+ vmov.u16 q15, #63 ; #63
+
+ vand q13, q1, q14 ; Filter2 &= hev
+
+ vqadd.s8 q2, q13, q11 ; Filter1 = clamp(Filter2+4)
+ vqadd.s8 q13, q13, q12 ; Filter2 = clamp(Filter2+3)
+
+ vmov q0, q15
+
+ vshr.s8 q2, q2, #3 ; Filter1 >>= 3
+ vshr.s8 q13, q13, #3 ; Filter2 >>= 3
+
+ vmov q11, q15
+ vmov q12, q15
+
+ vqsub.s8 q7, q7, q2 ; qs0 = clamp(qs0 - Filter1)
+
+ vqadd.s8 q6, q6, q13 ; ps0 = clamp(ps0 + Filter2)
+
+ vbic q1, q1, q14 ; vp8_filter &= ~hev
+
+ ; roughly 1/7th difference across boundary
+ ; roughly 2/7th difference across boundary
+ ; roughly 3/7th difference across boundary
+
+ vmov.u8 d5, #9 ; #9
+ vmov.u8 d4, #18 ; #18
+
+ vmov q13, q15
+ vmov q14, q15
+
+ vmlal.s8 q0, d2, d5 ; 63 + Filter2 * 9
+ vmlal.s8 q11, d3, d5
+ vmov.u8 d5, #27 ; #27
+ vmlal.s8 q12, d2, d4 ; 63 + Filter2 * 18
+ vmlal.s8 q13, d3, d4
+ vmlal.s8 q14, d2, d5 ; 63 + Filter2 * 27
+ vmlal.s8 q15, d3, d5
+
+ vqshrn.s16 d0, q0, #7 ; u = clamp((63 + Filter2 * 9)>>7)
+ vqshrn.s16 d1, q11, #7
+ vqshrn.s16 d24, q12, #7 ; u = clamp((63 + Filter2 * 18)>>7)
+ vqshrn.s16 d25, q13, #7
+ vqshrn.s16 d28, q14, #7 ; u = clamp((63 + Filter2 * 27)>>7)
+ vqshrn.s16 d29, q15, #7
+
+ vmov.u8 q1, #0x80 ; 0x80
+
+ vqsub.s8 q11, q9, q0 ; s = clamp(qs2 - u)
+ vqadd.s8 q0, q4, q0 ; s = clamp(ps2 + u)
+ vqsub.s8 q13, q8, q12 ; s = clamp(qs1 - u)
+ vqadd.s8 q12, q5, q12 ; s = clamp(ps1 + u)
+ vqsub.s8 q15, q7, q14 ; s = clamp(qs0 - u)
+ vqadd.s8 q14, q6, q14 ; s = clamp(ps0 + u)
+
+ veor q9, q11, q1 ; *oq2 = s^0x80
+ veor q4, q0, q1 ; *op2 = s^0x80
+ veor q8, q13, q1 ; *oq1 = s^0x80
+ veor q5, q12, q1 ; *op2 = s^0x80
+ veor q7, q15, q1 ; *oq0 = s^0x80
+ veor q6, q14, q1 ; *op0 = s^0x80
+
+ bx lr
+ ENDP ; |vp8_mbloop_filter_neon|
+
+;-----------------
+
+ END
diff --git a/vp8/encoder/arm/neon/sad16_neon.asm b/libvpx/vp8/common/arm/neon/sad16_neon.asm
similarity index 100%
rename from vp8/encoder/arm/neon/sad16_neon.asm
rename to libvpx/vp8/common/arm/neon/sad16_neon.asm
diff --git a/vp8/encoder/arm/neon/sad8_neon.asm b/libvpx/vp8/common/arm/neon/sad8_neon.asm
similarity index 100%
rename from vp8/encoder/arm/neon/sad8_neon.asm
rename to libvpx/vp8/common/arm/neon/sad8_neon.asm
diff --git a/vp8/common/arm/neon/save_neon_reg.asm b/libvpx/vp8/common/arm/neon/save_reg_neon.asm
similarity index 100%
rename from vp8/common/arm/neon/save_neon_reg.asm
rename to libvpx/vp8/common/arm/neon/save_reg_neon.asm
diff --git a/libvpx/vp8/common/arm/neon/shortidct4x4llm_neon.asm b/libvpx/vp8/common/arm/neon/shortidct4x4llm_neon.asm
new file mode 100644
index 0000000..67d2ab0
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/shortidct4x4llm_neon.asm
@@ -0,0 +1,139 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_short_idct4x4llm_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+;*************************************************************
+;void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch,
+; unsigned char *dst, int stride)
+;r0 short * input
+;r1 short * pred
+;r2 int pitch
+;r3 unsigned char dst
+;sp int stride
+;*************************************************************
+
+; static const int cospi8sqrt2minus1=20091;
+; static const int sinpi8sqrt2 =35468;
+; static const int rounding = 0;
+
+; Optimization note: The resulted data from dequantization are signed
+; 13-bit data that is in the range of [-4096, 4095]. This allows to
+; use "vqdmulh"(neon) instruction since it won't go out of range
+; (13+16+1=30bits<32bits). This instruction gives the high half
+; result of the multiplication that is needed in IDCT.
+
+|vp8_short_idct4x4llm_neon| PROC
+ adr r12, idct_coeff
+ vld1.16 {q1, q2}, [r0]
+ vld1.16 {d0}, [r12]
+
+ vswp d3, d4 ;q2(vp[4] vp[12])
+ ldr r0, [sp] ; stride
+
+ vqdmulh.s16 q3, q2, d0[2]
+ vqdmulh.s16 q4, q2, d0[0]
+
+ vqadd.s16 d12, d2, d3 ;a1
+ vqsub.s16 d13, d2, d3 ;b1
+
+ vshr.s16 q3, q3, #1
+ vshr.s16 q4, q4, #1
+
+ vqadd.s16 q3, q3, q2 ;modify since sinpi8sqrt2 > 65536/2 (negtive number)
+ vqadd.s16 q4, q4, q2
+
+ ;d6 - c1:temp1
+ ;d7 - d1:temp2
+ ;d8 - d1:temp1
+ ;d9 - c1:temp2
+
+ vqsub.s16 d10, d6, d9 ;c1
+ vqadd.s16 d11, d7, d8 ;d1
+
+ vqadd.s16 d2, d12, d11
+ vqadd.s16 d3, d13, d10
+ vqsub.s16 d4, d13, d10
+ vqsub.s16 d5, d12, d11
+
+ vtrn.32 d2, d4
+ vtrn.32 d3, d5
+ vtrn.16 d2, d3
+ vtrn.16 d4, d5
+
+ vswp d3, d4
+
+ vqdmulh.s16 q3, q2, d0[2]
+ vqdmulh.s16 q4, q2, d0[0]
+
+ vqadd.s16 d12, d2, d3 ;a1
+ vqsub.s16 d13, d2, d3 ;b1
+
+ vshr.s16 q3, q3, #1
+ vshr.s16 q4, q4, #1
+
+ vqadd.s16 q3, q3, q2 ;modify since sinpi8sqrt2 > 65536/2 (negtive number)
+ vqadd.s16 q4, q4, q2
+
+ vqsub.s16 d10, d6, d9 ;c1
+ vqadd.s16 d11, d7, d8 ;d1
+
+ vqadd.s16 d2, d12, d11
+ vqadd.s16 d3, d13, d10
+ vqsub.s16 d4, d13, d10
+ vqsub.s16 d5, d12, d11
+
+ vrshr.s16 d2, d2, #3
+ vrshr.s16 d3, d3, #3
+ vrshr.s16 d4, d4, #3
+ vrshr.s16 d5, d5, #3
+
+ vtrn.32 d2, d4
+ vtrn.32 d3, d5
+ vtrn.16 d2, d3
+ vtrn.16 d4, d5
+
+ ; load prediction data
+ vld1.32 d6[0], [r1], r2
+ vld1.32 d6[1], [r1], r2
+ vld1.32 d7[0], [r1], r2
+ vld1.32 d7[1], [r1], r2
+
+ ; add prediction and residual
+ vaddw.u8 q1, q1, d6
+ vaddw.u8 q2, q2, d7
+
+ vqmovun.s16 d1, q1
+ vqmovun.s16 d2, q2
+
+ ; store to destination
+ vst1.32 d1[0], [r3], r0
+ vst1.32 d1[1], [r3], r0
+ vst1.32 d2[0], [r3], r0
+ vst1.32 d2[1], [r3], r0
+
+ bx lr
+
+ ENDP
+
+;-----------------
+
+idct_coeff
+ DCD 0x4e7b4e7b, 0x8a8c8a8c
+
+;20091, 20091, 35468, 35468
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/sixtappredict16x16_neon.asm b/libvpx/vp8/common/arm/neon/sixtappredict16x16_neon.asm
new file mode 100644
index 0000000..9fdafd3
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/sixtappredict16x16_neon.asm
@@ -0,0 +1,490 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_sixtap_predict16x16_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+filter16_coeff
+ DCD 0, 0, 128, 0, 0, 0, 0, 0
+ DCD 0, -6, 123, 12, -1, 0, 0, 0
+ DCD 2, -11, 108, 36, -8, 1, 0, 0
+ DCD 0, -9, 93, 50, -6, 0, 0, 0
+ DCD 3, -16, 77, 77, -16, 3, 0, 0
+ DCD 0, -6, 50, 93, -9, 0, 0, 0
+ DCD 1, -8, 36, 108, -11, 2, 0, 0
+ DCD 0, -1, 12, 123, -6, 0, 0, 0
+
+; r0 unsigned char *src_ptr,
+; r1 int src_pixels_per_line,
+; r2 int xoffset,
+; r3 int yoffset,
+; r4 unsigned char *dst_ptr,
+; stack(r5) int dst_pitch
+
+;Note: To take advantage of 8-bit mulplication instruction in NEON. First apply abs() to
+; filter coeffs to make them u8. Then, use vmlsl for negtive coeffs. After multiplication,
+; the result can be negtive. So, I treat the result as s16. But, since it is also possible
+; that the result can be a large positive number (> 2^15-1), which could be confused as a
+; negtive number. To avoid that error, apply filter coeffs in the order of 0, 1, 4 ,5 ,2,
+; which ensures that the result stays in s16 range. Finally, saturated add the result by
+; applying 3rd filter coeff. Same applys to other filter functions.
+
+|vp8_sixtap_predict16x16_neon| PROC
+ push {r4-r5, lr}
+
+ adr r12, filter16_coeff
+ ldr r4, [sp, #12] ;load parameters from stack
+ ldr r5, [sp, #16] ;load parameters from stack
+
+ cmp r2, #0 ;skip first_pass filter if xoffset=0
+ beq secondpass_filter16x16_only
+
+ add r2, r12, r2, lsl #5 ;calculate filter location
+
+ cmp r3, #0 ;skip second_pass filter if yoffset=0
+
+ vld1.s32 {q14, q15}, [r2] ;load first_pass filter
+
+ beq firstpass_filter16x16_only
+
+ sub sp, sp, #336 ;reserve space on stack for temporary storage
+ mov lr, sp
+
+ vabs.s32 q12, q14
+ vabs.s32 q13, q15
+
+ mov r2, #7 ;loop counter
+ sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
+ sub r0, r0, r1, lsl #1
+
+ vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
+ vdup.8 d1, d24[4]
+ vdup.8 d2, d25[0]
+ vdup.8 d3, d25[4]
+ vdup.8 d4, d26[0]
+ vdup.8 d5, d26[4]
+
+;First Pass: output_height lines x output_width columns (21x16)
+filt_blk2d_fp16x16_loop_neon
+ vld1.u8 {d6, d7, d8}, [r0], r1 ;load src data
+ vld1.u8 {d9, d10, d11}, [r0], r1
+ vld1.u8 {d12, d13, d14}, [r0], r1
+
+ pld [r0]
+ pld [r0, r1]
+ pld [r0, r1, lsl #1]
+
+ vmull.u8 q8, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q9, d7, d0
+ vmull.u8 q10, d9, d0
+ vmull.u8 q11, d10, d0
+ vmull.u8 q12, d12, d0
+ vmull.u8 q13, d13, d0
+
+ vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
+ vext.8 d29, d9, d10, #1
+ vext.8 d30, d12, d13, #1
+
+ vmlsl.u8 q8, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q10, d29, d1
+ vmlsl.u8 q12, d30, d1
+
+ vext.8 d28, d7, d8, #1
+ vext.8 d29, d10, d11, #1
+ vext.8 d30, d13, d14, #1
+
+ vmlsl.u8 q9, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q11, d29, d1
+ vmlsl.u8 q13, d30, d1
+
+ vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
+ vext.8 d29, d9, d10, #4
+ vext.8 d30, d12, d13, #4
+
+ vmlsl.u8 q8, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q10, d29, d4
+ vmlsl.u8 q12, d30, d4
+
+ vext.8 d28, d7, d8, #4
+ vext.8 d29, d10, d11, #4
+ vext.8 d30, d13, d14, #4
+
+ vmlsl.u8 q9, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q11, d29, d4
+ vmlsl.u8 q13, d30, d4
+
+ vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d29, d9, d10, #5
+ vext.8 d30, d12, d13, #5
+
+ vmlal.u8 q8, d28, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q10, d29, d5
+ vmlal.u8 q12, d30, d5
+
+ vext.8 d28, d7, d8, #5
+ vext.8 d29, d10, d11, #5
+ vext.8 d30, d13, d14, #5
+
+ vmlal.u8 q9, d28, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q11, d29, d5
+ vmlal.u8 q13, d30, d5
+
+ vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
+ vext.8 d29, d9, d10, #2
+ vext.8 d30, d12, d13, #2
+
+ vmlal.u8 q8, d28, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q10, d29, d2
+ vmlal.u8 q12, d30, d2
+
+ vext.8 d28, d7, d8, #2
+ vext.8 d29, d10, d11, #2
+ vext.8 d30, d13, d14, #2
+
+ vmlal.u8 q9, d28, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q11, d29, d2
+ vmlal.u8 q13, d30, d2
+
+ vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
+ vext.8 d29, d9, d10, #3
+ vext.8 d30, d12, d13, #3
+
+ vext.8 d15, d7, d8, #3
+ vext.8 d31, d10, d11, #3
+ vext.8 d6, d13, d14, #3
+
+ vmull.u8 q4, d28, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q5, d29, d3
+ vmull.u8 q6, d30, d3
+
+ vqadd.s16 q8, q4 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q10, q5
+ vqadd.s16 q12, q6
+
+ vmull.u8 q6, d15, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q7, d31, d3
+ vmull.u8 q3, d6, d3
+
+ subs r2, r2, #1
+
+ vqadd.s16 q9, q6
+ vqadd.s16 q11, q7
+ vqadd.s16 q13, q3
+
+ vqrshrun.s16 d6, q8, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d7, q9, #7
+ vqrshrun.s16 d8, q10, #7
+ vqrshrun.s16 d9, q11, #7
+ vqrshrun.s16 d10, q12, #7
+ vqrshrun.s16 d11, q13, #7
+
+ vst1.u8 {d6, d7, d8}, [lr]! ;store result
+ vst1.u8 {d9, d10, d11}, [lr]!
+
+ bne filt_blk2d_fp16x16_loop_neon
+
+;Second pass: 16x16
+;secondpass_filter - do first 8-columns and then second 8-columns
+ add r3, r12, r3, lsl #5
+ sub lr, lr, #336
+
+ vld1.s32 {q5, q6}, [r3] ;load second_pass filter
+ mov r3, #2 ;loop counter
+
+ vabs.s32 q7, q5
+ vabs.s32 q8, q6
+
+ mov r2, #16
+
+ vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
+ vdup.8 d1, d14[4]
+ vdup.8 d2, d15[0]
+ vdup.8 d3, d15[4]
+ vdup.8 d4, d16[0]
+ vdup.8 d5, d16[4]
+
+filt_blk2d_sp16x16_outloop_neon
+ vld1.u8 {d18}, [lr], r2 ;load src data
+ vld1.u8 {d19}, [lr], r2
+ vld1.u8 {d20}, [lr], r2
+ vld1.u8 {d21}, [lr], r2
+ mov r12, #4 ;loop counter
+ vld1.u8 {d22}, [lr], r2
+
+secondpass_inner_loop_neon
+ vld1.u8 {d23}, [lr], r2 ;load src data
+ vld1.u8 {d24}, [lr], r2
+ vld1.u8 {d25}, [lr], r2
+ vld1.u8 {d26}, [lr], r2
+
+ vmull.u8 q3, d18, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q4, d19, d0
+ vmull.u8 q5, d20, d0
+ vmull.u8 q6, d21, d0
+
+ vmlsl.u8 q3, d19, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q4, d20, d1
+ vmlsl.u8 q5, d21, d1
+ vmlsl.u8 q6, d22, d1
+
+ vmlsl.u8 q3, d22, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q4, d23, d4
+ vmlsl.u8 q5, d24, d4
+ vmlsl.u8 q6, d25, d4
+
+ vmlal.u8 q3, d20, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q4, d21, d2
+ vmlal.u8 q5, d22, d2
+ vmlal.u8 q6, d23, d2
+
+ vmlal.u8 q3, d23, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q4, d24, d5
+ vmlal.u8 q5, d25, d5
+ vmlal.u8 q6, d26, d5
+
+ vmull.u8 q7, d21, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q8, d22, d3
+ vmull.u8 q9, d23, d3
+ vmull.u8 q10, d24, d3
+
+ subs r12, r12, #1
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d7, q8, #7
+ vqrshrun.s16 d8, q9, #7
+ vqrshrun.s16 d9, q10, #7
+
+ vst1.u8 {d6}, [r4], r5 ;store result
+ vmov q9, q11
+ vst1.u8 {d7}, [r4], r5
+ vmov q10, q12
+ vst1.u8 {d8}, [r4], r5
+ vmov d22, d26
+ vst1.u8 {d9}, [r4], r5
+
+ bne secondpass_inner_loop_neon
+
+ subs r3, r3, #1
+ sub lr, lr, #336
+ add lr, lr, #8
+
+ sub r4, r4, r5, lsl #4
+ add r4, r4, #8
+
+ bne filt_blk2d_sp16x16_outloop_neon
+
+ add sp, sp, #336
+ pop {r4-r5,pc}
+
+;--------------------
+firstpass_filter16x16_only
+ vabs.s32 q12, q14
+ vabs.s32 q13, q15
+
+ mov r2, #8 ;loop counter
+ sub r0, r0, #2 ;move srcptr back to (column-2)
+
+ vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
+ vdup.8 d1, d24[4]
+ vdup.8 d2, d25[0]
+ vdup.8 d3, d25[4]
+ vdup.8 d4, d26[0]
+ vdup.8 d5, d26[4]
+
+;First Pass: output_height lines x output_width columns (16x16)
+filt_blk2d_fpo16x16_loop_neon
+ vld1.u8 {d6, d7, d8}, [r0], r1 ;load src data
+ vld1.u8 {d9, d10, d11}, [r0], r1
+
+ pld [r0]
+ pld [r0, r1]
+
+ vmull.u8 q6, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q7, d7, d0
+ vmull.u8 q8, d9, d0
+ vmull.u8 q9, d10, d0
+
+ vext.8 d20, d6, d7, #1 ;construct src_ptr[-1]
+ vext.8 d21, d9, d10, #1
+ vext.8 d22, d7, d8, #1
+ vext.8 d23, d10, d11, #1
+ vext.8 d24, d6, d7, #4 ;construct src_ptr[2]
+ vext.8 d25, d9, d10, #4
+ vext.8 d26, d7, d8, #4
+ vext.8 d27, d10, d11, #4
+ vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d29, d9, d10, #5
+
+ vmlsl.u8 q6, d20, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q8, d21, d1
+ vmlsl.u8 q7, d22, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q9, d23, d1
+ vmlsl.u8 q6, d24, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q8, d25, d4
+ vmlsl.u8 q7, d26, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q9, d27, d4
+ vmlal.u8 q6, d28, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q8, d29, d5
+
+ vext.8 d20, d7, d8, #5
+ vext.8 d21, d10, d11, #5
+ vext.8 d22, d6, d7, #2 ;construct src_ptr[0]
+ vext.8 d23, d9, d10, #2
+ vext.8 d24, d7, d8, #2
+ vext.8 d25, d10, d11, #2
+
+ vext.8 d26, d6, d7, #3 ;construct src_ptr[1]
+ vext.8 d27, d9, d10, #3
+ vext.8 d28, d7, d8, #3
+ vext.8 d29, d10, d11, #3
+
+ vmlal.u8 q7, d20, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q9, d21, d5
+ vmlal.u8 q6, d22, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q8, d23, d2
+ vmlal.u8 q7, d24, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q9, d25, d2
+
+ vmull.u8 q10, d26, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q11, d27, d3
+ vmull.u8 q12, d28, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q15, d29, d3
+
+ vqadd.s16 q6, q10 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q11
+ vqadd.s16 q7, q12
+ vqadd.s16 q9, q15
+
+ subs r2, r2, #1
+
+ vqrshrun.s16 d6, q6, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d7, q7, #7
+ vqrshrun.s16 d8, q8, #7
+ vqrshrun.s16 d9, q9, #7
+
+ vst1.u8 {q3}, [r4], r5 ;store result
+ vst1.u8 {q4}, [r4], r5
+
+ bne filt_blk2d_fpo16x16_loop_neon
+
+ pop {r4-r5,pc}
+
+;--------------------
+secondpass_filter16x16_only
+;Second pass: 16x16
+ add r3, r12, r3, lsl #5
+ sub r0, r0, r1, lsl #1
+
+ vld1.s32 {q5, q6}, [r3] ;load second_pass filter
+ mov r3, #2 ;loop counter
+
+ vabs.s32 q7, q5
+ vabs.s32 q8, q6
+
+ vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
+ vdup.8 d1, d14[4]
+ vdup.8 d2, d15[0]
+ vdup.8 d3, d15[4]
+ vdup.8 d4, d16[0]
+ vdup.8 d5, d16[4]
+
+filt_blk2d_spo16x16_outloop_neon
+ vld1.u8 {d18}, [r0], r1 ;load src data
+ vld1.u8 {d19}, [r0], r1
+ vld1.u8 {d20}, [r0], r1
+ vld1.u8 {d21}, [r0], r1
+ mov r12, #4 ;loop counter
+ vld1.u8 {d22}, [r0], r1
+
+secondpass_only_inner_loop_neon
+ vld1.u8 {d23}, [r0], r1 ;load src data
+ vld1.u8 {d24}, [r0], r1
+ vld1.u8 {d25}, [r0], r1
+ vld1.u8 {d26}, [r0], r1
+
+ vmull.u8 q3, d18, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q4, d19, d0
+ vmull.u8 q5, d20, d0
+ vmull.u8 q6, d21, d0
+
+ vmlsl.u8 q3, d19, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q4, d20, d1
+ vmlsl.u8 q5, d21, d1
+ vmlsl.u8 q6, d22, d1
+
+ vmlsl.u8 q3, d22, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q4, d23, d4
+ vmlsl.u8 q5, d24, d4
+ vmlsl.u8 q6, d25, d4
+
+ vmlal.u8 q3, d20, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q4, d21, d2
+ vmlal.u8 q5, d22, d2
+ vmlal.u8 q6, d23, d2
+
+ vmlal.u8 q3, d23, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q4, d24, d5
+ vmlal.u8 q5, d25, d5
+ vmlal.u8 q6, d26, d5
+
+ vmull.u8 q7, d21, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q8, d22, d3
+ vmull.u8 q9, d23, d3
+ vmull.u8 q10, d24, d3
+
+ subs r12, r12, #1
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d7, q8, #7
+ vqrshrun.s16 d8, q9, #7
+ vqrshrun.s16 d9, q10, #7
+
+ vst1.u8 {d6}, [r4], r5 ;store result
+ vmov q9, q11
+ vst1.u8 {d7}, [r4], r5
+ vmov q10, q12
+ vst1.u8 {d8}, [r4], r5
+ vmov d22, d26
+ vst1.u8 {d9}, [r4], r5
+
+ bne secondpass_only_inner_loop_neon
+
+ subs r3, r3, #1
+ sub r0, r0, r1, lsl #4
+ sub r0, r0, r1, lsl #2
+ sub r0, r0, r1
+ add r0, r0, #8
+
+ sub r4, r4, r5, lsl #4
+ add r4, r4, #8
+
+ bne filt_blk2d_spo16x16_outloop_neon
+
+ pop {r4-r5,pc}
+
+ ENDP
+
+;-----------------
+ END
diff --git a/libvpx/vp8/common/arm/neon/sixtappredict4x4_neon.asm b/libvpx/vp8/common/arm/neon/sixtappredict4x4_neon.asm
new file mode 100644
index 0000000..a4222bc
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/sixtappredict4x4_neon.asm
@@ -0,0 +1,422 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_sixtap_predict4x4_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+filter4_coeff
+ DCD 0, 0, 128, 0, 0, 0, 0, 0
+ DCD 0, -6, 123, 12, -1, 0, 0, 0
+ DCD 2, -11, 108, 36, -8, 1, 0, 0
+ DCD 0, -9, 93, 50, -6, 0, 0, 0
+ DCD 3, -16, 77, 77, -16, 3, 0, 0
+ DCD 0, -6, 50, 93, -9, 0, 0, 0
+ DCD 1, -8, 36, 108, -11, 2, 0, 0
+ DCD 0, -1, 12, 123, -6, 0, 0, 0
+
+; r0 unsigned char *src_ptr,
+; r1 int src_pixels_per_line,
+; r2 int xoffset,
+; r3 int yoffset,
+; stack(r4) unsigned char *dst_ptr,
+; stack(lr) int dst_pitch
+
+|vp8_sixtap_predict4x4_neon| PROC
+ push {r4, lr}
+
+ adr r12, filter4_coeff
+ ldr r4, [sp, #8] ;load parameters from stack
+ ldr lr, [sp, #12] ;load parameters from stack
+
+ cmp r2, #0 ;skip first_pass filter if xoffset=0
+ beq secondpass_filter4x4_only
+
+ add r2, r12, r2, lsl #5 ;calculate filter location
+
+ cmp r3, #0 ;skip second_pass filter if yoffset=0
+ vld1.s32 {q14, q15}, [r2] ;load first_pass filter
+
+ beq firstpass_filter4x4_only
+
+ vabs.s32 q12, q14 ;get abs(filer_parameters)
+ vabs.s32 q13, q15
+
+ sub r0, r0, #2 ;go back 2 columns of src data
+ sub r0, r0, r1, lsl #1 ;go back 2 lines of src data
+
+;First pass: output_height lines x output_width columns (9x4)
+ vld1.u8 {q3}, [r0], r1 ;load first 4-line src data
+ vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
+ vld1.u8 {q4}, [r0], r1
+ vdup.8 d1, d24[4]
+ vld1.u8 {q5}, [r0], r1
+ vdup.8 d2, d25[0]
+ vld1.u8 {q6}, [r0], r1
+ vdup.8 d3, d25[4]
+ vdup.8 d4, d26[0]
+ vdup.8 d5, d26[4]
+
+ pld [r0]
+ pld [r0, r1]
+ pld [r0, r1, lsl #1]
+
+ vext.8 d18, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d19, d8, d9, #5
+ vext.8 d20, d10, d11, #5
+ vext.8 d21, d12, d13, #5
+
+ vswp d7, d8 ;discard 2nd half data after src_ptr[3] is done
+ vswp d11, d12
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[3])
+ vzip.32 d20, d21
+ vmull.u8 q7, d18, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmull.u8 q8, d20, d5
+
+ vmov q4, q3 ;keep original src data in q4 q6
+ vmov q6, q5
+
+ vzip.32 d6, d7 ;construct src_ptr[-2], and put 2-line data together
+ vzip.32 d10, d11
+ vshr.u64 q9, q4, #8 ;construct src_ptr[-1]
+ vshr.u64 q10, q6, #8
+ vmlal.u8 q7, d6, d0 ;+(src_ptr[-2] * vp8_filter[0])
+ vmlal.u8 q8, d10, d0
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[-1])
+ vzip.32 d20, d21
+ vshr.u64 q3, q4, #32 ;construct src_ptr[2]
+ vshr.u64 q5, q6, #32
+ vmlsl.u8 q7, d18, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q8, d20, d1
+
+ vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[2])
+ vzip.32 d10, d11
+ vshr.u64 q9, q4, #16 ;construct src_ptr[0]
+ vshr.u64 q10, q6, #16
+ vmlsl.u8 q7, d6, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q8, d10, d4
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[0])
+ vzip.32 d20, d21
+ vshr.u64 q3, q4, #24 ;construct src_ptr[1]
+ vshr.u64 q5, q6, #24
+ vmlal.u8 q7, d18, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q8, d20, d2
+
+ vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[1])
+ vzip.32 d10, d11
+ vmull.u8 q9, d6, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q10, d10, d3
+
+ vld1.u8 {q3}, [r0], r1 ;load rest 5-line src data
+ vld1.u8 {q4}, [r0], r1
+
+ vqadd.s16 q7, q9 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q10
+
+ vld1.u8 {q5}, [r0], r1
+ vld1.u8 {q6}, [r0], r1
+
+ vqrshrun.s16 d27, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d28, q8, #7
+
+ ;First Pass on rest 5-line data
+ vld1.u8 {q11}, [r0], r1
+
+ vext.8 d18, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d19, d8, d9, #5
+ vext.8 d20, d10, d11, #5
+ vext.8 d21, d12, d13, #5
+
+ vswp d7, d8 ;discard 2nd half data after src_ptr[3] is done
+ vswp d11, d12
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[3])
+ vzip.32 d20, d21
+ vext.8 d31, d22, d23, #5 ;construct src_ptr[3]
+ vmull.u8 q7, d18, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmull.u8 q8, d20, d5
+ vmull.u8 q12, d31, d5 ;(src_ptr[3] * vp8_filter[5])
+
+ vmov q4, q3 ;keep original src data in q4 q6
+ vmov q6, q5
+
+ vzip.32 d6, d7 ;construct src_ptr[-2], and put 2-line data together
+ vzip.32 d10, d11
+ vshr.u64 q9, q4, #8 ;construct src_ptr[-1]
+ vshr.u64 q10, q6, #8
+
+ vmlal.u8 q7, d6, d0 ;+(src_ptr[-2] * vp8_filter[0])
+ vmlal.u8 q8, d10, d0
+ vmlal.u8 q12, d22, d0 ;(src_ptr[-2] * vp8_filter[0])
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[-1])
+ vzip.32 d20, d21
+ vshr.u64 q3, q4, #32 ;construct src_ptr[2]
+ vshr.u64 q5, q6, #32
+ vext.8 d31, d22, d23, #1 ;construct src_ptr[-1]
+
+ vmlsl.u8 q7, d18, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q8, d20, d1
+ vmlsl.u8 q12, d31, d1 ;-(src_ptr[-1] * vp8_filter[1])
+
+ vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[2])
+ vzip.32 d10, d11
+ vshr.u64 q9, q4, #16 ;construct src_ptr[0]
+ vshr.u64 q10, q6, #16
+ vext.8 d31, d22, d23, #4 ;construct src_ptr[2]
+
+ vmlsl.u8 q7, d6, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q8, d10, d4
+ vmlsl.u8 q12, d31, d4 ;-(src_ptr[2] * vp8_filter[4])
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[0])
+ vzip.32 d20, d21
+ vshr.u64 q3, q4, #24 ;construct src_ptr[1]
+ vshr.u64 q5, q6, #24
+ vext.8 d31, d22, d23, #2 ;construct src_ptr[0]
+
+ vmlal.u8 q7, d18, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q8, d20, d2
+ vmlal.u8 q12, d31, d2 ;(src_ptr[0] * vp8_filter[2])
+
+ vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[1])
+ vzip.32 d10, d11
+ vext.8 d31, d22, d23, #3 ;construct src_ptr[1]
+ vmull.u8 q9, d6, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q10, d10, d3
+ vmull.u8 q11, d31, d3 ;(src_ptr[1] * vp8_filter[3])
+
+ add r3, r12, r3, lsl #5
+
+ vqadd.s16 q7, q9 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q10
+ vqadd.s16 q12, q11
+
+ vext.8 d23, d27, d28, #4
+ vld1.s32 {q5, q6}, [r3] ;load second_pass filter
+
+ vqrshrun.s16 d29, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d30, q8, #7
+ vqrshrun.s16 d31, q12, #7
+
+;Second pass: 4x4
+ vabs.s32 q7, q5
+ vabs.s32 q8, q6
+
+ vext.8 d24, d28, d29, #4
+ vext.8 d25, d29, d30, #4
+ vext.8 d26, d30, d31, #4
+
+ vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
+ vdup.8 d1, d14[4]
+ vdup.8 d2, d15[0]
+ vdup.8 d3, d15[4]
+ vdup.8 d4, d16[0]
+ vdup.8 d5, d16[4]
+
+ vmull.u8 q3, d27, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q4, d28, d0
+
+ vmull.u8 q5, d25, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmull.u8 q6, d26, d5
+
+ vmlsl.u8 q3, d29, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q4, d30, d4
+
+ vmlsl.u8 q5, d23, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q6, d24, d1
+
+ vmlal.u8 q3, d28, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q4, d29, d2
+
+ vmlal.u8 q5, d24, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmlal.u8 q6, d25, d3
+
+ add r0, r4, lr
+ add r1, r0, lr
+ add r2, r1, lr
+
+ vqadd.s16 q5, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q6, q4
+
+ vqrshrun.s16 d3, q5, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d4, q6, #7
+
+ vst1.32 {d3[0]}, [r4] ;store result
+ vst1.32 {d3[1]}, [r0]
+ vst1.32 {d4[0]}, [r1]
+ vst1.32 {d4[1]}, [r2]
+
+ pop {r4, pc}
+
+
+;---------------------
+firstpass_filter4x4_only
+ vabs.s32 q12, q14 ;get abs(filer_parameters)
+ vabs.s32 q13, q15
+
+ sub r0, r0, #2 ;go back 2 columns of src data
+
+;First pass: output_height lines x output_width columns (4x4)
+ vld1.u8 {q3}, [r0], r1 ;load first 4-line src data
+ vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
+ vld1.u8 {q4}, [r0], r1
+ vdup.8 d1, d24[4]
+ vld1.u8 {q5}, [r0], r1
+ vdup.8 d2, d25[0]
+ vld1.u8 {q6}, [r0], r1
+
+ vdup.8 d3, d25[4]
+ vdup.8 d4, d26[0]
+ vdup.8 d5, d26[4]
+
+ vext.8 d18, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d19, d8, d9, #5
+ vext.8 d20, d10, d11, #5
+ vext.8 d21, d12, d13, #5
+
+ vswp d7, d8 ;discard 2nd half data after src_ptr[3] is done
+ vswp d11, d12
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[3])
+ vzip.32 d20, d21
+ vmull.u8 q7, d18, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmull.u8 q8, d20, d5
+
+ vmov q4, q3 ;keep original src data in q4 q6
+ vmov q6, q5
+
+ vzip.32 d6, d7 ;construct src_ptr[-2], and put 2-line data together
+ vzip.32 d10, d11
+ vshr.u64 q9, q4, #8 ;construct src_ptr[-1]
+ vshr.u64 q10, q6, #8
+ vmlal.u8 q7, d6, d0 ;+(src_ptr[-2] * vp8_filter[0])
+ vmlal.u8 q8, d10, d0
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[-1])
+ vzip.32 d20, d21
+ vshr.u64 q3, q4, #32 ;construct src_ptr[2]
+ vshr.u64 q5, q6, #32
+ vmlsl.u8 q7, d18, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q8, d20, d1
+
+ vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[2])
+ vzip.32 d10, d11
+ vshr.u64 q9, q4, #16 ;construct src_ptr[0]
+ vshr.u64 q10, q6, #16
+ vmlsl.u8 q7, d6, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q8, d10, d4
+
+ vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[0])
+ vzip.32 d20, d21
+ vshr.u64 q3, q4, #24 ;construct src_ptr[1]
+ vshr.u64 q5, q6, #24
+ vmlal.u8 q7, d18, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q8, d20, d2
+
+ vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[1])
+ vzip.32 d10, d11
+ vmull.u8 q9, d6, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q10, d10, d3
+
+ add r0, r4, lr
+ add r1, r0, lr
+ add r2, r1, lr
+
+ vqadd.s16 q7, q9 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q10
+
+ vqrshrun.s16 d27, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d28, q8, #7
+
+ vst1.32 {d27[0]}, [r4] ;store result
+ vst1.32 {d27[1]}, [r0]
+ vst1.32 {d28[0]}, [r1]
+ vst1.32 {d28[1]}, [r2]
+
+ pop {r4, pc}
+
+
+;---------------------
+secondpass_filter4x4_only
+ sub r0, r0, r1, lsl #1
+ add r3, r12, r3, lsl #5
+
+ vld1.32 {d27[0]}, [r0], r1 ;load src data
+ vld1.s32 {q5, q6}, [r3] ;load second_pass filter
+ vld1.32 {d27[1]}, [r0], r1
+ vabs.s32 q7, q5
+ vld1.32 {d28[0]}, [r0], r1
+ vabs.s32 q8, q6
+ vld1.32 {d28[1]}, [r0], r1
+ vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
+ vld1.32 {d29[0]}, [r0], r1
+ vdup.8 d1, d14[4]
+ vld1.32 {d29[1]}, [r0], r1
+ vdup.8 d2, d15[0]
+ vld1.32 {d30[0]}, [r0], r1
+ vdup.8 d3, d15[4]
+ vld1.32 {d30[1]}, [r0], r1
+ vdup.8 d4, d16[0]
+ vld1.32 {d31[0]}, [r0], r1
+ vdup.8 d5, d16[4]
+
+ vext.8 d23, d27, d28, #4
+ vext.8 d24, d28, d29, #4
+ vext.8 d25, d29, d30, #4
+ vext.8 d26, d30, d31, #4
+
+ vmull.u8 q3, d27, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q4, d28, d0
+
+ vmull.u8 q5, d25, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmull.u8 q6, d26, d5
+
+ vmlsl.u8 q3, d29, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q4, d30, d4
+
+ vmlsl.u8 q5, d23, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q6, d24, d1
+
+ vmlal.u8 q3, d28, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q4, d29, d2
+
+ vmlal.u8 q5, d24, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmlal.u8 q6, d25, d3
+
+ add r0, r4, lr
+ add r1, r0, lr
+ add r2, r1, lr
+
+ vqadd.s16 q5, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q6, q4
+
+ vqrshrun.s16 d3, q5, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d4, q6, #7
+
+ vst1.32 {d3[0]}, [r4] ;store result
+ vst1.32 {d3[1]}, [r0]
+ vst1.32 {d4[0]}, [r1]
+ vst1.32 {d4[1]}, [r2]
+
+ pop {r4, pc}
+
+ ENDP
+
+;-----------------
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/sixtappredict8x4_neon.asm b/libvpx/vp8/common/arm/neon/sixtappredict8x4_neon.asm
new file mode 100644
index 0000000..a57ec01
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/sixtappredict8x4_neon.asm
@@ -0,0 +1,473 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_sixtap_predict8x4_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+filter8_coeff
+ DCD 0, 0, 128, 0, 0, 0, 0, 0
+ DCD 0, -6, 123, 12, -1, 0, 0, 0
+ DCD 2, -11, 108, 36, -8, 1, 0, 0
+ DCD 0, -9, 93, 50, -6, 0, 0, 0
+ DCD 3, -16, 77, 77, -16, 3, 0, 0
+ DCD 0, -6, 50, 93, -9, 0, 0, 0
+ DCD 1, -8, 36, 108, -11, 2, 0, 0
+ DCD 0, -1, 12, 123, -6, 0, 0, 0
+
+; r0 unsigned char *src_ptr,
+; r1 int src_pixels_per_line,
+; r2 int xoffset,
+; r3 int yoffset,
+; r4 unsigned char *dst_ptr,
+; stack(r5) int dst_pitch
+
+|vp8_sixtap_predict8x4_neon| PROC
+ push {r4-r5, lr}
+
+ adr r12, filter8_coeff
+ ldr r4, [sp, #12] ;load parameters from stack
+ ldr r5, [sp, #16] ;load parameters from stack
+
+ cmp r2, #0 ;skip first_pass filter if xoffset=0
+ beq secondpass_filter8x4_only
+
+ add r2, r12, r2, lsl #5 ;calculate filter location
+
+ cmp r3, #0 ;skip second_pass filter if yoffset=0
+
+ vld1.s32 {q14, q15}, [r2] ;load first_pass filter
+
+ beq firstpass_filter8x4_only
+
+ sub sp, sp, #32 ;reserve space on stack for temporary storage
+ vabs.s32 q12, q14
+ vabs.s32 q13, q15
+
+ sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
+ mov lr, sp
+ sub r0, r0, r1, lsl #1
+
+ vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
+ vdup.8 d1, d24[4]
+ vdup.8 d2, d25[0]
+
+;First pass: output_height lines x output_width columns (9x8)
+ vld1.u8 {q3}, [r0], r1 ;load src data
+ vdup.8 d3, d25[4]
+ vld1.u8 {q4}, [r0], r1
+ vdup.8 d4, d26[0]
+ vld1.u8 {q5}, [r0], r1
+ vdup.8 d5, d26[4]
+ vld1.u8 {q6}, [r0], r1
+
+ pld [r0]
+ pld [r0, r1]
+ pld [r0, r1, lsl #1]
+
+ vmull.u8 q7, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q8, d8, d0
+ vmull.u8 q9, d10, d0
+ vmull.u8 q10, d12, d0
+
+ vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
+ vext.8 d29, d8, d9, #1
+ vext.8 d30, d10, d11, #1
+ vext.8 d31, d12, d13, #1
+
+ vmlsl.u8 q7, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q8, d29, d1
+ vmlsl.u8 q9, d30, d1
+ vmlsl.u8 q10, d31, d1
+
+ vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
+ vext.8 d29, d8, d9, #4
+ vext.8 d30, d10, d11, #4
+ vext.8 d31, d12, d13, #4
+
+ vmlsl.u8 q7, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q8, d29, d4
+ vmlsl.u8 q9, d30, d4
+ vmlsl.u8 q10, d31, d4
+
+ vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
+ vext.8 d29, d8, d9, #2
+ vext.8 d30, d10, d11, #2
+ vext.8 d31, d12, d13, #2
+
+ vmlal.u8 q7, d28, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q8, d29, d2
+ vmlal.u8 q9, d30, d2
+ vmlal.u8 q10, d31, d2
+
+ vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d29, d8, d9, #5
+ vext.8 d30, d10, d11, #5
+ vext.8 d31, d12, d13, #5
+
+ vmlal.u8 q7, d28, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q8, d29, d5
+ vmlal.u8 q9, d30, d5
+ vmlal.u8 q10, d31, d5
+
+ vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
+ vext.8 d29, d8, d9, #3
+ vext.8 d30, d10, d11, #3
+ vext.8 d31, d12, d13, #3
+
+ vmull.u8 q3, d28, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q4, d29, d3
+ vmull.u8 q5, d30, d3
+ vmull.u8 q6, d31, d3
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vld1.u8 {q3}, [r0], r1 ;load src data
+
+ vqrshrun.s16 d22, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d23, q8, #7
+ vqrshrun.s16 d24, q9, #7
+ vqrshrun.s16 d25, q10, #7
+
+ vld1.u8 {q4}, [r0], r1
+ vst1.u8 {d22}, [lr]! ;store result
+ vld1.u8 {q5}, [r0], r1
+ vst1.u8 {d23}, [lr]!
+ vld1.u8 {q6}, [r0], r1
+ vst1.u8 {d24}, [lr]!
+ vld1.u8 {q7}, [r0], r1
+ vst1.u8 {d25}, [lr]!
+
+ ;first_pass filtering on the rest 5-line data
+ vmull.u8 q8, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q9, d8, d0
+ vmull.u8 q10, d10, d0
+ vmull.u8 q11, d12, d0
+ vmull.u8 q12, d14, d0
+
+ vext.8 d27, d6, d7, #1 ;construct src_ptr[-1]
+ vext.8 d28, d8, d9, #1
+ vext.8 d29, d10, d11, #1
+ vext.8 d30, d12, d13, #1
+ vext.8 d31, d14, d15, #1
+
+ vmlsl.u8 q8, d27, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q9, d28, d1
+ vmlsl.u8 q10, d29, d1
+ vmlsl.u8 q11, d30, d1
+ vmlsl.u8 q12, d31, d1
+
+ vext.8 d27, d6, d7, #4 ;construct src_ptr[2]
+ vext.8 d28, d8, d9, #4
+ vext.8 d29, d10, d11, #4
+ vext.8 d30, d12, d13, #4
+ vext.8 d31, d14, d15, #4
+
+ vmlsl.u8 q8, d27, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q9, d28, d4
+ vmlsl.u8 q10, d29, d4
+ vmlsl.u8 q11, d30, d4
+ vmlsl.u8 q12, d31, d4
+
+ vext.8 d27, d6, d7, #2 ;construct src_ptr[0]
+ vext.8 d28, d8, d9, #2
+ vext.8 d29, d10, d11, #2
+ vext.8 d30, d12, d13, #2
+ vext.8 d31, d14, d15, #2
+
+ vmlal.u8 q8, d27, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q9, d28, d2
+ vmlal.u8 q10, d29, d2
+ vmlal.u8 q11, d30, d2
+ vmlal.u8 q12, d31, d2
+
+ vext.8 d27, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d28, d8, d9, #5
+ vext.8 d29, d10, d11, #5
+ vext.8 d30, d12, d13, #5
+ vext.8 d31, d14, d15, #5
+
+ vmlal.u8 q8, d27, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q9, d28, d5
+ vmlal.u8 q10, d29, d5
+ vmlal.u8 q11, d30, d5
+ vmlal.u8 q12, d31, d5
+
+ vext.8 d27, d6, d7, #3 ;construct src_ptr[1]
+ vext.8 d28, d8, d9, #3
+ vext.8 d29, d10, d11, #3
+ vext.8 d30, d12, d13, #3
+ vext.8 d31, d14, d15, #3
+
+ vmull.u8 q3, d27, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q4, d28, d3
+ vmull.u8 q5, d29, d3
+ vmull.u8 q6, d30, d3
+ vmull.u8 q7, d31, d3
+
+ vqadd.s16 q8, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q9, q4
+ vqadd.s16 q10, q5
+ vqadd.s16 q11, q6
+ vqadd.s16 q12, q7
+
+ vqrshrun.s16 d26, q8, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d27, q9, #7
+ vqrshrun.s16 d28, q10, #7
+ vqrshrun.s16 d29, q11, #7 ;load intermediate data from stack
+ vqrshrun.s16 d30, q12, #7
+
+;Second pass: 8x4
+;secondpass_filter
+ add r3, r12, r3, lsl #5
+ sub lr, lr, #32
+
+ vld1.s32 {q5, q6}, [r3] ;load second_pass filter
+ vld1.u8 {q11}, [lr]!
+
+ vabs.s32 q7, q5
+ vabs.s32 q8, q6
+
+ vld1.u8 {q12}, [lr]!
+
+ vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
+ vdup.8 d1, d14[4]
+ vdup.8 d2, d15[0]
+ vdup.8 d3, d15[4]
+ vdup.8 d4, d16[0]
+ vdup.8 d5, d16[4]
+
+ vmull.u8 q3, d22, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q4, d23, d0
+ vmull.u8 q5, d24, d0
+ vmull.u8 q6, d25, d0
+
+ vmlsl.u8 q3, d23, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q4, d24, d1
+ vmlsl.u8 q5, d25, d1
+ vmlsl.u8 q6, d26, d1
+
+ vmlsl.u8 q3, d26, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q4, d27, d4
+ vmlsl.u8 q5, d28, d4
+ vmlsl.u8 q6, d29, d4
+
+ vmlal.u8 q3, d24, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q4, d25, d2
+ vmlal.u8 q5, d26, d2
+ vmlal.u8 q6, d27, d2
+
+ vmlal.u8 q3, d27, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q4, d28, d5
+ vmlal.u8 q5, d29, d5
+ vmlal.u8 q6, d30, d5
+
+ vmull.u8 q7, d25, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q8, d26, d3
+ vmull.u8 q9, d27, d3
+ vmull.u8 q10, d28, d3
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d7, q8, #7
+ vqrshrun.s16 d8, q9, #7
+ vqrshrun.s16 d9, q10, #7
+
+ vst1.u8 {d6}, [r4], r5 ;store result
+ vst1.u8 {d7}, [r4], r5
+ vst1.u8 {d8}, [r4], r5
+ vst1.u8 {d9}, [r4], r5
+
+ add sp, sp, #32
+ pop {r4-r5,pc}
+
+;--------------------
+firstpass_filter8x4_only
+ vabs.s32 q12, q14
+ vabs.s32 q13, q15
+
+ sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
+ vld1.u8 {q3}, [r0], r1 ;load src data
+
+ vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
+ vld1.u8 {q4}, [r0], r1
+ vdup.8 d1, d24[4]
+ vld1.u8 {q5}, [r0], r1
+ vdup.8 d2, d25[0]
+ vld1.u8 {q6}, [r0], r1
+ vdup.8 d3, d25[4]
+ vdup.8 d4, d26[0]
+ vdup.8 d5, d26[4]
+
+;First pass: output_height lines x output_width columns (4x8)
+ pld [r0]
+ pld [r0, r1]
+ pld [r0, r1, lsl #1]
+
+ vmull.u8 q7, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q8, d8, d0
+ vmull.u8 q9, d10, d0
+ vmull.u8 q10, d12, d0
+
+ vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
+ vext.8 d29, d8, d9, #1
+ vext.8 d30, d10, d11, #1
+ vext.8 d31, d12, d13, #1
+
+ vmlsl.u8 q7, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q8, d29, d1
+ vmlsl.u8 q9, d30, d1
+ vmlsl.u8 q10, d31, d1
+
+ vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
+ vext.8 d29, d8, d9, #4
+ vext.8 d30, d10, d11, #4
+ vext.8 d31, d12, d13, #4
+
+ vmlsl.u8 q7, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q8, d29, d4
+ vmlsl.u8 q9, d30, d4
+ vmlsl.u8 q10, d31, d4
+
+ vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
+ vext.8 d29, d8, d9, #2
+ vext.8 d30, d10, d11, #2
+ vext.8 d31, d12, d13, #2
+
+ vmlal.u8 q7, d28, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q8, d29, d2
+ vmlal.u8 q9, d30, d2
+ vmlal.u8 q10, d31, d2
+
+ vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d29, d8, d9, #5
+ vext.8 d30, d10, d11, #5
+ vext.8 d31, d12, d13, #5
+
+ vmlal.u8 q7, d28, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q8, d29, d5
+ vmlal.u8 q9, d30, d5
+ vmlal.u8 q10, d31, d5
+
+ vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
+ vext.8 d29, d8, d9, #3
+ vext.8 d30, d10, d11, #3
+ vext.8 d31, d12, d13, #3
+
+ vmull.u8 q3, d28, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q4, d29, d3
+ vmull.u8 q5, d30, d3
+ vmull.u8 q6, d31, d3
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vqrshrun.s16 d22, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d23, q8, #7
+ vqrshrun.s16 d24, q9, #7
+ vqrshrun.s16 d25, q10, #7
+
+ vst1.u8 {d22}, [r4], r5 ;store result
+ vst1.u8 {d23}, [r4], r5
+ vst1.u8 {d24}, [r4], r5
+ vst1.u8 {d25}, [r4], r5
+
+ pop {r4-r5,pc}
+
+;---------------------
+secondpass_filter8x4_only
+;Second pass: 8x4
+ add r3, r12, r3, lsl #5
+ sub r0, r0, r1, lsl #1
+ vld1.s32 {q5, q6}, [r3] ;load second_pass filter
+ vabs.s32 q7, q5
+ vabs.s32 q8, q6
+
+ vld1.u8 {d22}, [r0], r1
+ vld1.u8 {d23}, [r0], r1
+ vld1.u8 {d24}, [r0], r1
+ vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
+ vld1.u8 {d25}, [r0], r1
+ vdup.8 d1, d14[4]
+ vld1.u8 {d26}, [r0], r1
+ vdup.8 d2, d15[0]
+ vld1.u8 {d27}, [r0], r1
+ vdup.8 d3, d15[4]
+ vld1.u8 {d28}, [r0], r1
+ vdup.8 d4, d16[0]
+ vld1.u8 {d29}, [r0], r1
+ vdup.8 d5, d16[4]
+ vld1.u8 {d30}, [r0], r1
+
+ vmull.u8 q3, d22, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q4, d23, d0
+ vmull.u8 q5, d24, d0
+ vmull.u8 q6, d25, d0
+
+ vmlsl.u8 q3, d23, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q4, d24, d1
+ vmlsl.u8 q5, d25, d1
+ vmlsl.u8 q6, d26, d1
+
+ vmlsl.u8 q3, d26, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q4, d27, d4
+ vmlsl.u8 q5, d28, d4
+ vmlsl.u8 q6, d29, d4
+
+ vmlal.u8 q3, d24, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q4, d25, d2
+ vmlal.u8 q5, d26, d2
+ vmlal.u8 q6, d27, d2
+
+ vmlal.u8 q3, d27, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q4, d28, d5
+ vmlal.u8 q5, d29, d5
+ vmlal.u8 q6, d30, d5
+
+ vmull.u8 q7, d25, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q8, d26, d3
+ vmull.u8 q9, d27, d3
+ vmull.u8 q10, d28, d3
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d7, q8, #7
+ vqrshrun.s16 d8, q9, #7
+ vqrshrun.s16 d9, q10, #7
+
+ vst1.u8 {d6}, [r4], r5 ;store result
+ vst1.u8 {d7}, [r4], r5
+ vst1.u8 {d8}, [r4], r5
+ vst1.u8 {d9}, [r4], r5
+
+ pop {r4-r5,pc}
+
+ ENDP
+
+;-----------------
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/sixtappredict8x8_neon.asm b/libvpx/vp8/common/arm/neon/sixtappredict8x8_neon.asm
new file mode 100644
index 0000000..00ed5ae
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/sixtappredict8x8_neon.asm
@@ -0,0 +1,524 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_sixtap_predict8x8_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+filter8_coeff
+ DCD 0, 0, 128, 0, 0, 0, 0, 0
+ DCD 0, -6, 123, 12, -1, 0, 0, 0
+ DCD 2, -11, 108, 36, -8, 1, 0, 0
+ DCD 0, -9, 93, 50, -6, 0, 0, 0
+ DCD 3, -16, 77, 77, -16, 3, 0, 0
+ DCD 0, -6, 50, 93, -9, 0, 0, 0
+ DCD 1, -8, 36, 108, -11, 2, 0, 0
+ DCD 0, -1, 12, 123, -6, 0, 0, 0
+
+; r0 unsigned char *src_ptr,
+; r1 int src_pixels_per_line,
+; r2 int xoffset,
+; r3 int yoffset,
+; stack(r4) unsigned char *dst_ptr,
+; stack(r5) int dst_pitch
+
+|vp8_sixtap_predict8x8_neon| PROC
+ push {r4-r5, lr}
+
+ adr r12, filter8_coeff
+
+ ldr r4, [sp, #12] ;load parameters from stack
+ ldr r5, [sp, #16] ;load parameters from stack
+
+ cmp r2, #0 ;skip first_pass filter if xoffset=0
+ beq secondpass_filter8x8_only
+
+ add r2, r12, r2, lsl #5 ;calculate filter location
+
+ cmp r3, #0 ;skip second_pass filter if yoffset=0
+
+ vld1.s32 {q14, q15}, [r2] ;load first_pass filter
+
+ beq firstpass_filter8x8_only
+
+ sub sp, sp, #64 ;reserve space on stack for temporary storage
+ mov lr, sp
+
+ vabs.s32 q12, q14
+ vabs.s32 q13, q15
+
+ mov r2, #2 ;loop counter
+ sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
+ sub r0, r0, r1, lsl #1
+
+ vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
+ vdup.8 d1, d24[4]
+ vdup.8 d2, d25[0]
+
+;First pass: output_height lines x output_width columns (13x8)
+ vld1.u8 {q3}, [r0], r1 ;load src data
+ vdup.8 d3, d25[4]
+ vld1.u8 {q4}, [r0], r1
+ vdup.8 d4, d26[0]
+ vld1.u8 {q5}, [r0], r1
+ vdup.8 d5, d26[4]
+ vld1.u8 {q6}, [r0], r1
+
+filt_blk2d_fp8x8_loop_neon
+ pld [r0]
+ pld [r0, r1]
+ pld [r0, r1, lsl #1]
+
+ vmull.u8 q7, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q8, d8, d0
+ vmull.u8 q9, d10, d0
+ vmull.u8 q10, d12, d0
+
+ vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
+ vext.8 d29, d8, d9, #1
+ vext.8 d30, d10, d11, #1
+ vext.8 d31, d12, d13, #1
+
+ vmlsl.u8 q7, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q8, d29, d1
+ vmlsl.u8 q9, d30, d1
+ vmlsl.u8 q10, d31, d1
+
+ vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
+ vext.8 d29, d8, d9, #4
+ vext.8 d30, d10, d11, #4
+ vext.8 d31, d12, d13, #4
+
+ vmlsl.u8 q7, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q8, d29, d4
+ vmlsl.u8 q9, d30, d4
+ vmlsl.u8 q10, d31, d4
+
+ vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
+ vext.8 d29, d8, d9, #2
+ vext.8 d30, d10, d11, #2
+ vext.8 d31, d12, d13, #2
+
+ vmlal.u8 q7, d28, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q8, d29, d2
+ vmlal.u8 q9, d30, d2
+ vmlal.u8 q10, d31, d2
+
+ vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d29, d8, d9, #5
+ vext.8 d30, d10, d11, #5
+ vext.8 d31, d12, d13, #5
+
+ vmlal.u8 q7, d28, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q8, d29, d5
+ vmlal.u8 q9, d30, d5
+ vmlal.u8 q10, d31, d5
+
+ vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
+ vext.8 d29, d8, d9, #3
+ vext.8 d30, d10, d11, #3
+ vext.8 d31, d12, d13, #3
+
+ vmull.u8 q3, d28, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q4, d29, d3
+ vmull.u8 q5, d30, d3
+ vmull.u8 q6, d31, d3
+
+ subs r2, r2, #1
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vld1.u8 {q3}, [r0], r1 ;load src data
+
+ vqrshrun.s16 d22, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d23, q8, #7
+ vqrshrun.s16 d24, q9, #7
+ vqrshrun.s16 d25, q10, #7
+
+ vst1.u8 {d22}, [lr]! ;store result
+ vld1.u8 {q4}, [r0], r1
+ vst1.u8 {d23}, [lr]!
+ vld1.u8 {q5}, [r0], r1
+ vst1.u8 {d24}, [lr]!
+ vld1.u8 {q6}, [r0], r1
+ vst1.u8 {d25}, [lr]!
+
+ bne filt_blk2d_fp8x8_loop_neon
+
+ ;first_pass filtering on the rest 5-line data
+ ;vld1.u8 {q3}, [r0], r1 ;load src data
+ ;vld1.u8 {q4}, [r0], r1
+ ;vld1.u8 {q5}, [r0], r1
+ ;vld1.u8 {q6}, [r0], r1
+ vld1.u8 {q7}, [r0], r1
+
+ vmull.u8 q8, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q9, d8, d0
+ vmull.u8 q10, d10, d0
+ vmull.u8 q11, d12, d0
+ vmull.u8 q12, d14, d0
+
+ vext.8 d27, d6, d7, #1 ;construct src_ptr[-1]
+ vext.8 d28, d8, d9, #1
+ vext.8 d29, d10, d11, #1
+ vext.8 d30, d12, d13, #1
+ vext.8 d31, d14, d15, #1
+
+ vmlsl.u8 q8, d27, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q9, d28, d1
+ vmlsl.u8 q10, d29, d1
+ vmlsl.u8 q11, d30, d1
+ vmlsl.u8 q12, d31, d1
+
+ vext.8 d27, d6, d7, #4 ;construct src_ptr[2]
+ vext.8 d28, d8, d9, #4
+ vext.8 d29, d10, d11, #4
+ vext.8 d30, d12, d13, #4
+ vext.8 d31, d14, d15, #4
+
+ vmlsl.u8 q8, d27, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q9, d28, d4
+ vmlsl.u8 q10, d29, d4
+ vmlsl.u8 q11, d30, d4
+ vmlsl.u8 q12, d31, d4
+
+ vext.8 d27, d6, d7, #2 ;construct src_ptr[0]
+ vext.8 d28, d8, d9, #2
+ vext.8 d29, d10, d11, #2
+ vext.8 d30, d12, d13, #2
+ vext.8 d31, d14, d15, #2
+
+ vmlal.u8 q8, d27, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q9, d28, d2
+ vmlal.u8 q10, d29, d2
+ vmlal.u8 q11, d30, d2
+ vmlal.u8 q12, d31, d2
+
+ vext.8 d27, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d28, d8, d9, #5
+ vext.8 d29, d10, d11, #5
+ vext.8 d30, d12, d13, #5
+ vext.8 d31, d14, d15, #5
+
+ vmlal.u8 q8, d27, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q9, d28, d5
+ vmlal.u8 q10, d29, d5
+ vmlal.u8 q11, d30, d5
+ vmlal.u8 q12, d31, d5
+
+ vext.8 d27, d6, d7, #3 ;construct src_ptr[1]
+ vext.8 d28, d8, d9, #3
+ vext.8 d29, d10, d11, #3
+ vext.8 d30, d12, d13, #3
+ vext.8 d31, d14, d15, #3
+
+ vmull.u8 q3, d27, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q4, d28, d3
+ vmull.u8 q5, d29, d3
+ vmull.u8 q6, d30, d3
+ vmull.u8 q7, d31, d3
+
+ vqadd.s16 q8, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q9, q4
+ vqadd.s16 q10, q5
+ vqadd.s16 q11, q6
+ vqadd.s16 q12, q7
+
+ add r3, r12, r3, lsl #5
+
+ vqrshrun.s16 d26, q8, #7 ;shift/round/saturate to u8
+ sub lr, lr, #64
+ vqrshrun.s16 d27, q9, #7
+ vld1.u8 {q9}, [lr]! ;load intermediate data from stack
+ vqrshrun.s16 d28, q10, #7
+ vld1.u8 {q10}, [lr]!
+
+ vld1.s32 {q5, q6}, [r3] ;load second_pass filter
+
+ vqrshrun.s16 d29, q11, #7
+ vld1.u8 {q11}, [lr]!
+
+ vabs.s32 q7, q5
+ vabs.s32 q8, q6
+
+ vqrshrun.s16 d30, q12, #7
+ vld1.u8 {q12}, [lr]!
+
+;Second pass: 8x8
+ mov r3, #2 ;loop counter
+
+ vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
+ vdup.8 d1, d14[4]
+ vdup.8 d2, d15[0]
+ vdup.8 d3, d15[4]
+ vdup.8 d4, d16[0]
+ vdup.8 d5, d16[4]
+
+filt_blk2d_sp8x8_loop_neon
+ vmull.u8 q3, d18, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q4, d19, d0
+ vmull.u8 q5, d20, d0
+ vmull.u8 q6, d21, d0
+
+ vmlsl.u8 q3, d19, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q4, d20, d1
+ vmlsl.u8 q5, d21, d1
+ vmlsl.u8 q6, d22, d1
+
+ vmlsl.u8 q3, d22, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q4, d23, d4
+ vmlsl.u8 q5, d24, d4
+ vmlsl.u8 q6, d25, d4
+
+ vmlal.u8 q3, d20, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q4, d21, d2
+ vmlal.u8 q5, d22, d2
+ vmlal.u8 q6, d23, d2
+
+ vmlal.u8 q3, d23, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q4, d24, d5
+ vmlal.u8 q5, d25, d5
+ vmlal.u8 q6, d26, d5
+
+ vmull.u8 q7, d21, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q8, d22, d3
+ vmull.u8 q9, d23, d3
+ vmull.u8 q10, d24, d3
+
+ subs r3, r3, #1
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d7, q8, #7
+ vqrshrun.s16 d8, q9, #7
+ vqrshrun.s16 d9, q10, #7
+
+ vmov q9, q11
+ vst1.u8 {d6}, [r4], r5 ;store result
+ vmov q10, q12
+ vst1.u8 {d7}, [r4], r5
+ vmov q11, q13
+ vst1.u8 {d8}, [r4], r5
+ vmov q12, q14
+ vst1.u8 {d9}, [r4], r5
+ vmov d26, d30
+
+ bne filt_blk2d_sp8x8_loop_neon
+
+ add sp, sp, #64
+ pop {r4-r5,pc}
+
+;---------------------
+firstpass_filter8x8_only
+ ;add r2, r12, r2, lsl #5 ;calculate filter location
+ ;vld1.s32 {q14, q15}, [r2] ;load first_pass filter
+ vabs.s32 q12, q14
+ vabs.s32 q13, q15
+
+ mov r2, #2 ;loop counter
+ sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
+
+ vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
+ vdup.8 d1, d24[4]
+ vdup.8 d2, d25[0]
+ vdup.8 d3, d25[4]
+ vdup.8 d4, d26[0]
+ vdup.8 d5, d26[4]
+
+;First pass: output_height lines x output_width columns (8x8)
+filt_blk2d_fpo8x8_loop_neon
+ vld1.u8 {q3}, [r0], r1 ;load src data
+ vld1.u8 {q4}, [r0], r1
+ vld1.u8 {q5}, [r0], r1
+ vld1.u8 {q6}, [r0], r1
+
+ pld [r0]
+ pld [r0, r1]
+ pld [r0, r1, lsl #1]
+
+ vmull.u8 q7, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q8, d8, d0
+ vmull.u8 q9, d10, d0
+ vmull.u8 q10, d12, d0
+
+ vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
+ vext.8 d29, d8, d9, #1
+ vext.8 d30, d10, d11, #1
+ vext.8 d31, d12, d13, #1
+
+ vmlsl.u8 q7, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q8, d29, d1
+ vmlsl.u8 q9, d30, d1
+ vmlsl.u8 q10, d31, d1
+
+ vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
+ vext.8 d29, d8, d9, #4
+ vext.8 d30, d10, d11, #4
+ vext.8 d31, d12, d13, #4
+
+ vmlsl.u8 q7, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q8, d29, d4
+ vmlsl.u8 q9, d30, d4
+ vmlsl.u8 q10, d31, d4
+
+ vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
+ vext.8 d29, d8, d9, #2
+ vext.8 d30, d10, d11, #2
+ vext.8 d31, d12, d13, #2
+
+ vmlal.u8 q7, d28, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q8, d29, d2
+ vmlal.u8 q9, d30, d2
+ vmlal.u8 q10, d31, d2
+
+ vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
+ vext.8 d29, d8, d9, #5
+ vext.8 d30, d10, d11, #5
+ vext.8 d31, d12, d13, #5
+
+ vmlal.u8 q7, d28, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q8, d29, d5
+ vmlal.u8 q9, d30, d5
+ vmlal.u8 q10, d31, d5
+
+ vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
+ vext.8 d29, d8, d9, #3
+ vext.8 d30, d10, d11, #3
+ vext.8 d31, d12, d13, #3
+
+ vmull.u8 q3, d28, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q4, d29, d3
+ vmull.u8 q5, d30, d3
+ vmull.u8 q6, d31, d3
+ ;
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ subs r2, r2, #1
+
+ vqrshrun.s16 d22, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d23, q8, #7
+ vqrshrun.s16 d24, q9, #7
+ vqrshrun.s16 d25, q10, #7
+
+ vst1.u8 {d22}, [r4], r5 ;store result
+ vst1.u8 {d23}, [r4], r5
+ vst1.u8 {d24}, [r4], r5
+ vst1.u8 {d25}, [r4], r5
+
+ bne filt_blk2d_fpo8x8_loop_neon
+
+ pop {r4-r5,pc}
+
+;---------------------
+secondpass_filter8x8_only
+ sub r0, r0, r1, lsl #1
+ add r3, r12, r3, lsl #5
+
+ vld1.u8 {d18}, [r0], r1 ;load src data
+ vld1.s32 {q5, q6}, [r3] ;load second_pass filter
+ vld1.u8 {d19}, [r0], r1
+ vabs.s32 q7, q5
+ vld1.u8 {d20}, [r0], r1
+ vabs.s32 q8, q6
+ vld1.u8 {d21}, [r0], r1
+ mov r3, #2 ;loop counter
+ vld1.u8 {d22}, [r0], r1
+ vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
+ vld1.u8 {d23}, [r0], r1
+ vdup.8 d1, d14[4]
+ vld1.u8 {d24}, [r0], r1
+ vdup.8 d2, d15[0]
+ vld1.u8 {d25}, [r0], r1
+ vdup.8 d3, d15[4]
+ vld1.u8 {d26}, [r0], r1
+ vdup.8 d4, d16[0]
+ vld1.u8 {d27}, [r0], r1
+ vdup.8 d5, d16[4]
+ vld1.u8 {d28}, [r0], r1
+ vld1.u8 {d29}, [r0], r1
+ vld1.u8 {d30}, [r0], r1
+
+;Second pass: 8x8
+filt_blk2d_spo8x8_loop_neon
+ vmull.u8 q3, d18, d0 ;(src_ptr[-2] * vp8_filter[0])
+ vmull.u8 q4, d19, d0
+ vmull.u8 q5, d20, d0
+ vmull.u8 q6, d21, d0
+
+ vmlsl.u8 q3, d19, d1 ;-(src_ptr[-1] * vp8_filter[1])
+ vmlsl.u8 q4, d20, d1
+ vmlsl.u8 q5, d21, d1
+ vmlsl.u8 q6, d22, d1
+
+ vmlsl.u8 q3, d22, d4 ;-(src_ptr[2] * vp8_filter[4])
+ vmlsl.u8 q4, d23, d4
+ vmlsl.u8 q5, d24, d4
+ vmlsl.u8 q6, d25, d4
+
+ vmlal.u8 q3, d20, d2 ;(src_ptr[0] * vp8_filter[2])
+ vmlal.u8 q4, d21, d2
+ vmlal.u8 q5, d22, d2
+ vmlal.u8 q6, d23, d2
+
+ vmlal.u8 q3, d23, d5 ;(src_ptr[3] * vp8_filter[5])
+ vmlal.u8 q4, d24, d5
+ vmlal.u8 q5, d25, d5
+ vmlal.u8 q6, d26, d5
+
+ vmull.u8 q7, d21, d3 ;(src_ptr[1] * vp8_filter[3])
+ vmull.u8 q8, d22, d3
+ vmull.u8 q9, d23, d3
+ vmull.u8 q10, d24, d3
+
+ subs r3, r3, #1
+
+ vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
+ vqadd.s16 q8, q4
+ vqadd.s16 q9, q5
+ vqadd.s16 q10, q6
+
+ vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
+ vqrshrun.s16 d7, q8, #7
+ vqrshrun.s16 d8, q9, #7
+ vqrshrun.s16 d9, q10, #7
+
+ vmov q9, q11
+ vst1.u8 {d6}, [r4], r5 ;store result
+ vmov q10, q12
+ vst1.u8 {d7}, [r4], r5
+ vmov q11, q13
+ vst1.u8 {d8}, [r4], r5
+ vmov q12, q14
+ vst1.u8 {d9}, [r4], r5
+ vmov d26, d30
+
+ bne filt_blk2d_spo8x8_loop_neon
+
+ pop {r4-r5,pc}
+
+ ENDP
+
+;-----------------
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/variance_neon.asm b/libvpx/vp8/common/arm/neon/variance_neon.asm
new file mode 100644
index 0000000..e3b4832
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/variance_neon.asm
@@ -0,0 +1,276 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_variance16x16_neon|
+ EXPORT |vp8_variance16x8_neon|
+ EXPORT |vp8_variance8x16_neon|
+ EXPORT |vp8_variance8x8_neon|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+|vp8_variance16x16_neon| PROC
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+ mov r12, #8
+
+variance16x16_neon_loop
+ vld1.8 {q0}, [r0], r1 ;Load up source and reference
+ vld1.8 {q2}, [r2], r3
+ vld1.8 {q1}, [r0], r1
+ vld1.8 {q3}, [r2], r3
+
+ vsubl.u8 q11, d0, d4 ;calculate diff
+ vsubl.u8 q12, d1, d5
+ vsubl.u8 q13, d2, d6
+ vsubl.u8 q14, d3, d7
+
+ ;VPADAL adds adjacent pairs of elements of a vector, and accumulates
+ ;the results into the elements of the destination vector. The explanation
+ ;in ARM guide is wrong.
+ vpadal.s16 q8, q11 ;calculate sum
+ vmlal.s16 q9, d22, d22 ;calculate sse
+ vmlal.s16 q10, d23, d23
+
+ subs r12, r12, #1
+
+ vpadal.s16 q8, q12
+ vmlal.s16 q9, d24, d24
+ vmlal.s16 q10, d25, d25
+ vpadal.s16 q8, q13
+ vmlal.s16 q9, d26, d26
+ vmlal.s16 q10, d27, d27
+ vpadal.s16 q8, q14
+ vmlal.s16 q9, d28, d28
+ vmlal.s16 q10, d29, d29
+
+ bne variance16x16_neon_loop
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ ldr r12, [sp] ;load *sse from stack
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ ;vmov.32 r0, d0[0] ;this instruction costs a lot
+ ;vmov.32 r1, d1[0]
+ ;mul r0, r0, r0
+ ;str r1, [r12]
+ ;sub r0, r1, r0, lsr #8
+
+ ; while sum is signed, sum * sum is always positive and must be treated as
+ ; unsigned to avoid propagating the sign bit.
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [r12] ;store sse
+ vshr.u32 d10, d10, #8
+ vsub.u32 d0, d1, d10
+
+ vmov.32 r0, d0[0] ;return
+ bx lr
+
+ ENDP
+
+;================================
+;unsigned int vp8_variance16x8_c(
+; unsigned char *src_ptr,
+; int source_stride,
+; unsigned char *ref_ptr,
+; int recon_stride,
+; unsigned int *sse)
+|vp8_variance16x8_neon| PROC
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+ mov r12, #4
+
+variance16x8_neon_loop
+ vld1.8 {q0}, [r0], r1 ;Load up source and reference
+ vld1.8 {q2}, [r2], r3
+ vld1.8 {q1}, [r0], r1
+ vld1.8 {q3}, [r2], r3
+
+ vsubl.u8 q11, d0, d4 ;calculate diff
+ vsubl.u8 q12, d1, d5
+ vsubl.u8 q13, d2, d6
+ vsubl.u8 q14, d3, d7
+
+ vpadal.s16 q8, q11 ;calculate sum
+ vmlal.s16 q9, d22, d22 ;calculate sse
+ vmlal.s16 q10, d23, d23
+
+ subs r12, r12, #1
+
+ vpadal.s16 q8, q12
+ vmlal.s16 q9, d24, d24
+ vmlal.s16 q10, d25, d25
+ vpadal.s16 q8, q13
+ vmlal.s16 q9, d26, d26
+ vmlal.s16 q10, d27, d27
+ vpadal.s16 q8, q14
+ vmlal.s16 q9, d28, d28
+ vmlal.s16 q10, d29, d29
+
+ bne variance16x8_neon_loop
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ ldr r12, [sp] ;load *sse from stack
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [r12] ;store sse
+ vshr.u32 d10, d10, #7
+ vsub.u32 d0, d1, d10
+
+ vmov.32 r0, d0[0] ;return
+ bx lr
+
+ ENDP
+
+;=================================
+;unsigned int vp8_variance8x16_c(
+; unsigned char *src_ptr,
+; int source_stride,
+; unsigned char *ref_ptr,
+; int recon_stride,
+; unsigned int *sse)
+
+|vp8_variance8x16_neon| PROC
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+ mov r12, #8
+
+variance8x16_neon_loop
+ vld1.8 {d0}, [r0], r1 ;Load up source and reference
+ vld1.8 {d4}, [r2], r3
+ vld1.8 {d2}, [r0], r1
+ vld1.8 {d6}, [r2], r3
+
+ vsubl.u8 q11, d0, d4 ;calculate diff
+ vsubl.u8 q12, d2, d6
+
+ vpadal.s16 q8, q11 ;calculate sum
+ vmlal.s16 q9, d22, d22 ;calculate sse
+ vmlal.s16 q10, d23, d23
+
+ subs r12, r12, #1
+
+ vpadal.s16 q8, q12
+ vmlal.s16 q9, d24, d24
+ vmlal.s16 q10, d25, d25
+
+ bne variance8x16_neon_loop
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ ldr r12, [sp] ;load *sse from stack
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [r12] ;store sse
+ vshr.u32 d10, d10, #7
+ vsub.u32 d0, d1, d10
+
+ vmov.32 r0, d0[0] ;return
+ bx lr
+
+ ENDP
+
+;==================================
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+|vp8_variance8x8_neon| PROC
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+ mov r12, #2
+
+variance8x8_neon_loop
+ vld1.8 {d0}, [r0], r1 ;Load up source and reference
+ vld1.8 {d4}, [r2], r3
+ vld1.8 {d1}, [r0], r1
+ vld1.8 {d5}, [r2], r3
+ vld1.8 {d2}, [r0], r1
+ vld1.8 {d6}, [r2], r3
+ vld1.8 {d3}, [r0], r1
+ vld1.8 {d7}, [r2], r3
+
+ vsubl.u8 q11, d0, d4 ;calculate diff
+ vsubl.u8 q12, d1, d5
+ vsubl.u8 q13, d2, d6
+ vsubl.u8 q14, d3, d7
+
+ vpadal.s16 q8, q11 ;calculate sum
+ vmlal.s16 q9, d22, d22 ;calculate sse
+ vmlal.s16 q10, d23, d23
+
+ subs r12, r12, #1
+
+ vpadal.s16 q8, q12
+ vmlal.s16 q9, d24, d24
+ vmlal.s16 q10, d25, d25
+ vpadal.s16 q8, q13
+ vmlal.s16 q9, d26, d26
+ vmlal.s16 q10, d27, d27
+ vpadal.s16 q8, q14
+ vmlal.s16 q9, d28, d28
+ vmlal.s16 q10, d29, d29
+
+ bne variance8x8_neon_loop
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ ldr r12, [sp] ;load *sse from stack
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [r12] ;store sse
+ vshr.u32 d10, d10, #6
+ vsub.u32 d0, d1, d10
+
+ vmov.32 r0, d0[0] ;return
+ bx lr
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/vp8_subpixelvariance16x16_neon.asm b/libvpx/vp8/common/arm/neon/vp8_subpixelvariance16x16_neon.asm
new file mode 100644
index 0000000..e7a3ed1
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/vp8_subpixelvariance16x16_neon.asm
@@ -0,0 +1,423 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+bilinear_taps_coeff
+ DCD 128, 0, 112, 16, 96, 32, 80, 48, 64, 64, 48, 80, 32, 96, 16, 112
+
+;-----------------
+
+ EXPORT |vp8_sub_pixel_variance16x16_neon_func|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+; r0 unsigned char *src_ptr,
+; r1 int src_pixels_per_line,
+; r2 int xoffset,
+; r3 int yoffset,
+; stack(r4) unsigned char *dst_ptr,
+; stack(r5) int dst_pixels_per_line,
+; stack(r6) unsigned int *sse
+;note: most of the code is copied from bilinear_predict16x16_neon and vp8_variance16x16_neon.
+
+|vp8_sub_pixel_variance16x16_neon_func| PROC
+ push {r4-r6, lr}
+
+ adr r12, bilinear_taps_coeff
+ ldr r4, [sp, #16] ;load *dst_ptr from stack
+ ldr r5, [sp, #20] ;load dst_pixels_per_line from stack
+ ldr r6, [sp, #24] ;load *sse from stack
+
+ cmp r2, #0 ;skip first_pass filter if xoffset=0
+ beq secondpass_bfilter16x16_only
+
+ add r2, r12, r2, lsl #3 ;calculate filter location
+
+ cmp r3, #0 ;skip second_pass filter if yoffset=0
+
+ vld1.s32 {d31}, [r2] ;load first_pass filter
+
+ beq firstpass_bfilter16x16_only
+
+ sub sp, sp, #272 ;reserve space on stack for temporary storage
+ vld1.u8 {d2, d3, d4}, [r0], r1 ;load src data
+ mov lr, sp
+ vld1.u8 {d5, d6, d7}, [r0], r1
+
+ mov r2, #3 ;loop counter
+ vld1.u8 {d8, d9, d10}, [r0], r1
+
+ vdup.8 d0, d31[0] ;first_pass filter (d0 d1)
+ vld1.u8 {d11, d12, d13}, [r0], r1
+
+ vdup.8 d1, d31[4]
+
+;First Pass: output_height lines x output_width columns (17x16)
+vp8e_filt_blk2d_fp16x16_loop_neon
+ pld [r0]
+ pld [r0, r1]
+ pld [r0, r1, lsl #1]
+
+ vmull.u8 q7, d2, d0 ;(src_ptr[0] * Filter[0])
+ vmull.u8 q8, d3, d0
+ vmull.u8 q9, d5, d0
+ vmull.u8 q10, d6, d0
+ vmull.u8 q11, d8, d0
+ vmull.u8 q12, d9, d0
+ vmull.u8 q13, d11, d0
+ vmull.u8 q14, d12, d0
+
+ vext.8 d2, d2, d3, #1 ;construct src_ptr[1]
+ vext.8 d5, d5, d6, #1
+ vext.8 d8, d8, d9, #1
+ vext.8 d11, d11, d12, #1
+
+ vmlal.u8 q7, d2, d1 ;(src_ptr[0] * Filter[1])
+ vmlal.u8 q9, d5, d1
+ vmlal.u8 q11, d8, d1
+ vmlal.u8 q13, d11, d1
+
+ vext.8 d3, d3, d4, #1
+ vext.8 d6, d6, d7, #1
+ vext.8 d9, d9, d10, #1
+ vext.8 d12, d12, d13, #1
+
+ vmlal.u8 q8, d3, d1 ;(src_ptr[0] * Filter[1])
+ vmlal.u8 q10, d6, d1
+ vmlal.u8 q12, d9, d1
+ vmlal.u8 q14, d12, d1
+
+ subs r2, r2, #1
+
+ vqrshrn.u16 d14, q7, #7 ;shift/round/saturate to u8
+ vqrshrn.u16 d15, q8, #7
+ vqrshrn.u16 d16, q9, #7
+ vqrshrn.u16 d17, q10, #7
+ vqrshrn.u16 d18, q11, #7
+ vqrshrn.u16 d19, q12, #7
+ vqrshrn.u16 d20, q13, #7
+
+ vld1.u8 {d2, d3, d4}, [r0], r1 ;load src data
+ vqrshrn.u16 d21, q14, #7
+ vld1.u8 {d5, d6, d7}, [r0], r1
+
+ vst1.u8 {d14, d15, d16, d17}, [lr]! ;store result
+ vld1.u8 {d8, d9, d10}, [r0], r1
+ vst1.u8 {d18, d19, d20, d21}, [lr]!
+ vld1.u8 {d11, d12, d13}, [r0], r1
+
+ bne vp8e_filt_blk2d_fp16x16_loop_neon
+
+;First-pass filtering for rest 5 lines
+ vld1.u8 {d14, d15, d16}, [r0], r1
+
+ vmull.u8 q9, d2, d0 ;(src_ptr[0] * Filter[0])
+ vmull.u8 q10, d3, d0
+ vmull.u8 q11, d5, d0
+ vmull.u8 q12, d6, d0
+ vmull.u8 q13, d8, d0
+ vmull.u8 q14, d9, d0
+
+ vext.8 d2, d2, d3, #1 ;construct src_ptr[1]
+ vext.8 d5, d5, d6, #1
+ vext.8 d8, d8, d9, #1
+
+ vmlal.u8 q9, d2, d1 ;(src_ptr[0] * Filter[1])
+ vmlal.u8 q11, d5, d1
+ vmlal.u8 q13, d8, d1
+
+ vext.8 d3, d3, d4, #1
+ vext.8 d6, d6, d7, #1
+ vext.8 d9, d9, d10, #1
+
+ vmlal.u8 q10, d3, d1 ;(src_ptr[0] * Filter[1])
+ vmlal.u8 q12, d6, d1
+ vmlal.u8 q14, d9, d1
+
+ vmull.u8 q1, d11, d0
+ vmull.u8 q2, d12, d0
+ vmull.u8 q3, d14, d0
+ vmull.u8 q4, d15, d0
+
+ vext.8 d11, d11, d12, #1 ;construct src_ptr[1]
+ vext.8 d14, d14, d15, #1
+
+ vmlal.u8 q1, d11, d1 ;(src_ptr[0] * Filter[1])
+ vmlal.u8 q3, d14, d1
+
+ vext.8 d12, d12, d13, #1
+ vext.8 d15, d15, d16, #1
+
+ vmlal.u8 q2, d12, d1 ;(src_ptr[0] * Filter[1])
+ vmlal.u8 q4, d15, d1
+
+ vqrshrn.u16 d10, q9, #7 ;shift/round/saturate to u8
+ vqrshrn.u16 d11, q10, #7
+ vqrshrn.u16 d12, q11, #7
+ vqrshrn.u16 d13, q12, #7
+ vqrshrn.u16 d14, q13, #7
+ vqrshrn.u16 d15, q14, #7
+ vqrshrn.u16 d16, q1, #7
+ vqrshrn.u16 d17, q2, #7
+ vqrshrn.u16 d18, q3, #7
+ vqrshrn.u16 d19, q4, #7
+
+ vst1.u8 {d10, d11, d12, d13}, [lr]! ;store result
+ vst1.u8 {d14, d15, d16, d17}, [lr]!
+ vst1.u8 {d18, d19}, [lr]!
+
+;Second pass: 16x16
+;secondpass_filter
+ add r3, r12, r3, lsl #3
+ sub lr, lr, #272
+
+ vld1.u32 {d31}, [r3] ;load second_pass filter
+
+ sub sp, sp, #256
+ mov r3, sp
+
+ vld1.u8 {d22, d23}, [lr]! ;load src data
+
+ vdup.8 d0, d31[0] ;second_pass filter parameters (d0 d1)
+ vdup.8 d1, d31[4]
+ mov r12, #4 ;loop counter
+
+vp8e_filt_blk2d_sp16x16_loop_neon
+ vld1.u8 {d24, d25}, [lr]!
+ vmull.u8 q1, d22, d0 ;(src_ptr[0] * Filter[0])
+ vld1.u8 {d26, d27}, [lr]!
+ vmull.u8 q2, d23, d0
+ vld1.u8 {d28, d29}, [lr]!
+ vmull.u8 q3, d24, d0
+ vld1.u8 {d30, d31}, [lr]!
+
+ vmull.u8 q4, d25, d0
+ vmull.u8 q5, d26, d0
+ vmull.u8 q6, d27, d0
+ vmull.u8 q7, d28, d0
+ vmull.u8 q8, d29, d0
+
+ vmlal.u8 q1, d24, d1 ;(src_ptr[pixel_step] * Filter[1])
+ vmlal.u8 q2, d25, d1
+ vmlal.u8 q3, d26, d1
+ vmlal.u8 q4, d27, d1
+ vmlal.u8 q5, d28, d1
+ vmlal.u8 q6, d29, d1
+ vmlal.u8 q7, d30, d1
+ vmlal.u8 q8, d31, d1
+
+ subs r12, r12, #1
+
+ vqrshrn.u16 d2, q1, #7 ;shift/round/saturate to u8
+ vqrshrn.u16 d3, q2, #7
+ vqrshrn.u16 d4, q3, #7
+ vqrshrn.u16 d5, q4, #7
+ vqrshrn.u16 d6, q5, #7
+ vqrshrn.u16 d7, q6, #7
+ vqrshrn.u16 d8, q7, #7
+ vqrshrn.u16 d9, q8, #7
+
+ vst1.u8 {d2, d3}, [r3]! ;store result
+ vst1.u8 {d4, d5}, [r3]!
+ vst1.u8 {d6, d7}, [r3]!
+ vmov q11, q15
+ vst1.u8 {d8, d9}, [r3]!
+
+ bne vp8e_filt_blk2d_sp16x16_loop_neon
+
+ b sub_pixel_variance16x16_neon
+
+;--------------------
+firstpass_bfilter16x16_only
+ mov r2, #4 ;loop counter
+ sub sp, sp, #528 ;reserve space on stack for temporary storage
+ vdup.8 d0, d31[0] ;first_pass filter (d0 d1)
+ vdup.8 d1, d31[4]
+ mov r3, sp
+
+;First Pass: output_height lines x output_width columns (16x16)
+vp8e_filt_blk2d_fpo16x16_loop_neon
+ vld1.u8 {d2, d3, d4}, [r0], r1 ;load src data
+ vld1.u8 {d5, d6, d7}, [r0], r1
+ vld1.u8 {d8, d9, d10}, [r0], r1
+ vld1.u8 {d11, d12, d13}, [r0], r1
+
+ pld [r0]
+ pld [r0, r1]
+ pld [r0, r1, lsl #1]
+
+ vmull.u8 q7, d2, d0 ;(src_ptr[0] * Filter[0])
+ vmull.u8 q8, d3, d0
+ vmull.u8 q9, d5, d0
+ vmull.u8 q10, d6, d0
+ vmull.u8 q11, d8, d0
+ vmull.u8 q12, d9, d0
+ vmull.u8 q13, d11, d0
+ vmull.u8 q14, d12, d0
+
+ vext.8 d2, d2, d3, #1 ;construct src_ptr[1]
+ vext.8 d5, d5, d6, #1
+ vext.8 d8, d8, d9, #1
+ vext.8 d11, d11, d12, #1
+
+ vmlal.u8 q7, d2, d1 ;(src_ptr[0] * Filter[1])
+ vmlal.u8 q9, d5, d1
+ vmlal.u8 q11, d8, d1
+ vmlal.u8 q13, d11, d1
+
+ vext.8 d3, d3, d4, #1
+ vext.8 d6, d6, d7, #1
+ vext.8 d9, d9, d10, #1
+ vext.8 d12, d12, d13, #1
+
+ vmlal.u8 q8, d3, d1 ;(src_ptr[0] * Filter[1])
+ vmlal.u8 q10, d6, d1
+ vmlal.u8 q12, d9, d1
+ vmlal.u8 q14, d12, d1
+
+ subs r2, r2, #1
+
+ vqrshrn.u16 d14, q7, #7 ;shift/round/saturate to u8
+ vqrshrn.u16 d15, q8, #7
+ vqrshrn.u16 d16, q9, #7
+ vqrshrn.u16 d17, q10, #7
+ vqrshrn.u16 d18, q11, #7
+ vqrshrn.u16 d19, q12, #7
+ vqrshrn.u16 d20, q13, #7
+ vst1.u8 {d14, d15}, [r3]! ;store result
+ vqrshrn.u16 d21, q14, #7
+
+ vst1.u8 {d16, d17}, [r3]!
+ vst1.u8 {d18, d19}, [r3]!
+ vst1.u8 {d20, d21}, [r3]!
+
+ bne vp8e_filt_blk2d_fpo16x16_loop_neon
+
+ b sub_pixel_variance16x16_neon
+
+;---------------------
+secondpass_bfilter16x16_only
+;Second pass: 16x16
+;secondpass_filter
+ sub sp, sp, #528 ;reserve space on stack for temporary storage
+ add r3, r12, r3, lsl #3
+ mov r12, #4 ;loop counter
+ vld1.u32 {d31}, [r3] ;load second_pass filter
+ vld1.u8 {d22, d23}, [r0], r1 ;load src data
+ mov r3, sp
+
+ vdup.8 d0, d31[0] ;second_pass filter parameters (d0 d1)
+ vdup.8 d1, d31[4]
+
+vp8e_filt_blk2d_spo16x16_loop_neon
+ vld1.u8 {d24, d25}, [r0], r1
+ vmull.u8 q1, d22, d0 ;(src_ptr[0] * Filter[0])
+ vld1.u8 {d26, d27}, [r0], r1
+ vmull.u8 q2, d23, d0
+ vld1.u8 {d28, d29}, [r0], r1
+ vmull.u8 q3, d24, d0
+ vld1.u8 {d30, d31}, [r0], r1
+
+ vmull.u8 q4, d25, d0
+ vmull.u8 q5, d26, d0
+ vmull.u8 q6, d27, d0
+ vmull.u8 q7, d28, d0
+ vmull.u8 q8, d29, d0
+
+ vmlal.u8 q1, d24, d1 ;(src_ptr[pixel_step] * Filter[1])
+ vmlal.u8 q2, d25, d1
+ vmlal.u8 q3, d26, d1
+ vmlal.u8 q4, d27, d1
+ vmlal.u8 q5, d28, d1
+ vmlal.u8 q6, d29, d1
+ vmlal.u8 q7, d30, d1
+ vmlal.u8 q8, d31, d1
+
+ vqrshrn.u16 d2, q1, #7 ;shift/round/saturate to u8
+ vqrshrn.u16 d3, q2, #7
+ vqrshrn.u16 d4, q3, #7
+ vqrshrn.u16 d5, q4, #7
+ vqrshrn.u16 d6, q5, #7
+ vqrshrn.u16 d7, q6, #7
+ vqrshrn.u16 d8, q7, #7
+ vqrshrn.u16 d9, q8, #7
+
+ vst1.u8 {d2, d3}, [r3]! ;store result
+ subs r12, r12, #1
+ vst1.u8 {d4, d5}, [r3]!
+ vmov q11, q15
+ vst1.u8 {d6, d7}, [r3]!
+ vst1.u8 {d8, d9}, [r3]!
+
+ bne vp8e_filt_blk2d_spo16x16_loop_neon
+
+ b sub_pixel_variance16x16_neon
+
+;----------------------------
+;variance16x16
+sub_pixel_variance16x16_neon
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+ sub r3, r3, #256
+ mov r12, #8
+
+sub_pixel_variance16x16_neon_loop
+ vld1.8 {q0}, [r3]! ;Load up source and reference
+ vld1.8 {q2}, [r4], r5
+ vld1.8 {q1}, [r3]!
+ vld1.8 {q3}, [r4], r5
+
+ vsubl.u8 q11, d0, d4 ;diff
+ vsubl.u8 q12, d1, d5
+ vsubl.u8 q13, d2, d6
+ vsubl.u8 q14, d3, d7
+
+ vpadal.s16 q8, q11 ;sum
+ vmlal.s16 q9, d22, d22 ;sse
+ vmlal.s16 q10, d23, d23
+
+ subs r12, r12, #1
+
+ vpadal.s16 q8, q12
+ vmlal.s16 q9, d24, d24
+ vmlal.s16 q10, d25, d25
+ vpadal.s16 q8, q13
+ vmlal.s16 q9, d26, d26
+ vmlal.s16 q10, d27, d27
+ vpadal.s16 q8, q14
+ vmlal.s16 q9, d28, d28
+ vmlal.s16 q10, d29, d29
+
+ bne sub_pixel_variance16x16_neon_loop
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [r6] ;store sse
+ vshr.u32 d10, d10, #8
+ vsub.u32 d0, d1, d10
+
+ add sp, sp, #528
+ vmov.32 r0, d0[0] ;return
+
+ pop {r4-r6,pc}
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/vp8_subpixelvariance16x16s_neon.asm b/libvpx/vp8/common/arm/neon/vp8_subpixelvariance16x16s_neon.asm
new file mode 100644
index 0000000..155be4f
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/vp8_subpixelvariance16x16s_neon.asm
@@ -0,0 +1,572 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_variance_halfpixvar16x16_h_neon|
+ EXPORT |vp8_variance_halfpixvar16x16_v_neon|
+ EXPORT |vp8_variance_halfpixvar16x16_hv_neon|
+ EXPORT |vp8_sub_pixel_variance16x16s_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+;================================================
+;unsigned int vp8_variance_halfpixvar16x16_h_neon
+;(
+; unsigned char *src_ptr, r0
+; int src_pixels_per_line, r1
+; unsigned char *dst_ptr, r2
+; int dst_pixels_per_line, r3
+; unsigned int *sse
+;);
+;================================================
+|vp8_variance_halfpixvar16x16_h_neon| PROC
+ push {lr}
+
+ mov r12, #4 ;loop counter
+ ldr lr, [sp, #4] ;load *sse from stack
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+;First Pass: output_height lines x output_width columns (16x16)
+vp8_filt_fpo16x16s_4_0_loop_neon
+ vld1.u8 {d0, d1, d2, d3}, [r0], r1 ;load src data
+ vld1.8 {q11}, [r2], r3
+ vld1.u8 {d4, d5, d6, d7}, [r0], r1
+ vld1.8 {q12}, [r2], r3
+ vld1.u8 {d8, d9, d10, d11}, [r0], r1
+ vld1.8 {q13}, [r2], r3
+ vld1.u8 {d12, d13, d14, d15}, [r0], r1
+
+ ;pld [r0]
+ ;pld [r0, r1]
+ ;pld [r0, r1, lsl #1]
+
+ vext.8 q1, q0, q1, #1 ;construct src_ptr[1]
+ vext.8 q3, q2, q3, #1
+ vext.8 q5, q4, q5, #1
+ vext.8 q7, q6, q7, #1
+
+ vrhadd.u8 q0, q0, q1 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
+ vld1.8 {q14}, [r2], r3
+ vrhadd.u8 q1, q2, q3
+ vrhadd.u8 q2, q4, q5
+ vrhadd.u8 q3, q6, q7
+
+ vsubl.u8 q4, d0, d22 ;diff
+ vsubl.u8 q5, d1, d23
+ vsubl.u8 q6, d2, d24
+ vsubl.u8 q7, d3, d25
+ vsubl.u8 q0, d4, d26
+ vsubl.u8 q1, d5, d27
+ vsubl.u8 q2, d6, d28
+ vsubl.u8 q3, d7, d29
+
+ vpadal.s16 q8, q4 ;sum
+ vmlal.s16 q9, d8, d8 ;sse
+ vmlal.s16 q10, d9, d9
+
+ subs r12, r12, #1
+
+ vpadal.s16 q8, q5
+ vmlal.s16 q9, d10, d10
+ vmlal.s16 q10, d11, d11
+ vpadal.s16 q8, q6
+ vmlal.s16 q9, d12, d12
+ vmlal.s16 q10, d13, d13
+ vpadal.s16 q8, q7
+ vmlal.s16 q9, d14, d14
+ vmlal.s16 q10, d15, d15
+
+ vpadal.s16 q8, q0 ;sum
+ vmlal.s16 q9, d0, d0 ;sse
+ vmlal.s16 q10, d1, d1
+ vpadal.s16 q8, q1
+ vmlal.s16 q9, d2, d2
+ vmlal.s16 q10, d3, d3
+ vpadal.s16 q8, q2
+ vmlal.s16 q9, d4, d4
+ vmlal.s16 q10, d5, d5
+ vpadal.s16 q8, q3
+ vmlal.s16 q9, d6, d6
+ vmlal.s16 q10, d7, d7
+
+ bne vp8_filt_fpo16x16s_4_0_loop_neon
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [lr] ;store sse
+ vshr.u32 d10, d10, #8
+ vsub.u32 d0, d1, d10
+
+ vmov.32 r0, d0[0] ;return
+ pop {pc}
+ ENDP
+
+;================================================
+;unsigned int vp8_variance_halfpixvar16x16_v_neon
+;(
+; unsigned char *src_ptr, r0
+; int src_pixels_per_line, r1
+; unsigned char *dst_ptr, r2
+; int dst_pixels_per_line, r3
+; unsigned int *sse
+;);
+;================================================
+|vp8_variance_halfpixvar16x16_v_neon| PROC
+ push {lr}
+
+ mov r12, #4 ;loop counter
+
+ vld1.u8 {q0}, [r0], r1 ;load src data
+ ldr lr, [sp, #4] ;load *sse from stack
+
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+vp8_filt_spo16x16s_0_4_loop_neon
+ vld1.u8 {q2}, [r0], r1
+ vld1.8 {q1}, [r2], r3
+ vld1.u8 {q4}, [r0], r1
+ vld1.8 {q3}, [r2], r3
+ vld1.u8 {q6}, [r0], r1
+ vld1.8 {q5}, [r2], r3
+ vld1.u8 {q15}, [r0], r1
+
+ vrhadd.u8 q0, q0, q2
+ vld1.8 {q7}, [r2], r3
+ vrhadd.u8 q2, q2, q4
+ vrhadd.u8 q4, q4, q6
+ vrhadd.u8 q6, q6, q15
+
+ vsubl.u8 q11, d0, d2 ;diff
+ vsubl.u8 q12, d1, d3
+ vsubl.u8 q13, d4, d6
+ vsubl.u8 q14, d5, d7
+ vsubl.u8 q0, d8, d10
+ vsubl.u8 q1, d9, d11
+ vsubl.u8 q2, d12, d14
+ vsubl.u8 q3, d13, d15
+
+ vpadal.s16 q8, q11 ;sum
+ vmlal.s16 q9, d22, d22 ;sse
+ vmlal.s16 q10, d23, d23
+
+ subs r12, r12, #1
+
+ vpadal.s16 q8, q12
+ vmlal.s16 q9, d24, d24
+ vmlal.s16 q10, d25, d25
+ vpadal.s16 q8, q13
+ vmlal.s16 q9, d26, d26
+ vmlal.s16 q10, d27, d27
+ vpadal.s16 q8, q14
+ vmlal.s16 q9, d28, d28
+ vmlal.s16 q10, d29, d29
+
+ vpadal.s16 q8, q0 ;sum
+ vmlal.s16 q9, d0, d0 ;sse
+ vmlal.s16 q10, d1, d1
+ vpadal.s16 q8, q1
+ vmlal.s16 q9, d2, d2
+ vmlal.s16 q10, d3, d3
+ vpadal.s16 q8, q2
+ vmlal.s16 q9, d4, d4
+ vmlal.s16 q10, d5, d5
+
+ vmov q0, q15
+
+ vpadal.s16 q8, q3
+ vmlal.s16 q9, d6, d6
+ vmlal.s16 q10, d7, d7
+
+ bne vp8_filt_spo16x16s_0_4_loop_neon
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [lr] ;store sse
+ vshr.u32 d10, d10, #8
+ vsub.u32 d0, d1, d10
+
+ vmov.32 r0, d0[0] ;return
+ pop {pc}
+ ENDP
+
+;================================================
+;unsigned int vp8_variance_halfpixvar16x16_hv_neon
+;(
+; unsigned char *src_ptr, r0
+; int src_pixels_per_line, r1
+; unsigned char *dst_ptr, r2
+; int dst_pixels_per_line, r3
+; unsigned int *sse
+;);
+;================================================
+|vp8_variance_halfpixvar16x16_hv_neon| PROC
+ push {lr}
+
+ vld1.u8 {d0, d1, d2, d3}, [r0], r1 ;load src data
+
+ ldr lr, [sp, #4] ;load *sse from stack
+ vmov.i8 q13, #0 ;q8 - sum
+ vext.8 q1, q0, q1, #1 ;construct src_ptr[1]
+
+ vmov.i8 q14, #0 ;q9, q10 - sse
+ vmov.i8 q15, #0
+
+ mov r12, #4 ;loop counter
+ vrhadd.u8 q0, q0, q1 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
+
+;First Pass: output_height lines x output_width columns (17x16)
+vp8_filt16x16s_4_4_loop_neon
+ vld1.u8 {d4, d5, d6, d7}, [r0], r1
+ vld1.u8 {d8, d9, d10, d11}, [r0], r1
+ vld1.u8 {d12, d13, d14, d15}, [r0], r1
+ vld1.u8 {d16, d17, d18, d19}, [r0], r1
+
+ ;pld [r0]
+ ;pld [r0, r1]
+ ;pld [r0, r1, lsl #1]
+
+ vext.8 q3, q2, q3, #1 ;construct src_ptr[1]
+ vext.8 q5, q4, q5, #1
+ vext.8 q7, q6, q7, #1
+ vext.8 q9, q8, q9, #1
+
+ vrhadd.u8 q1, q2, q3 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
+ vrhadd.u8 q2, q4, q5
+ vrhadd.u8 q3, q6, q7
+ vrhadd.u8 q4, q8, q9
+
+ vld1.8 {q5}, [r2], r3
+ vrhadd.u8 q0, q0, q1
+ vld1.8 {q6}, [r2], r3
+ vrhadd.u8 q1, q1, q2
+ vld1.8 {q7}, [r2], r3
+ vrhadd.u8 q2, q2, q3
+ vld1.8 {q8}, [r2], r3
+ vrhadd.u8 q3, q3, q4
+
+ vsubl.u8 q9, d0, d10 ;diff
+ vsubl.u8 q10, d1, d11
+ vsubl.u8 q11, d2, d12
+ vsubl.u8 q12, d3, d13
+
+ vsubl.u8 q0, d4, d14 ;diff
+ vsubl.u8 q1, d5, d15
+ vsubl.u8 q5, d6, d16
+ vsubl.u8 q6, d7, d17
+
+ vpadal.s16 q13, q9 ;sum
+ vmlal.s16 q14, d18, d18 ;sse
+ vmlal.s16 q15, d19, d19
+
+ vpadal.s16 q13, q10 ;sum
+ vmlal.s16 q14, d20, d20 ;sse
+ vmlal.s16 q15, d21, d21
+
+ vpadal.s16 q13, q11 ;sum
+ vmlal.s16 q14, d22, d22 ;sse
+ vmlal.s16 q15, d23, d23
+
+ vpadal.s16 q13, q12 ;sum
+ vmlal.s16 q14, d24, d24 ;sse
+ vmlal.s16 q15, d25, d25
+
+ subs r12, r12, #1
+
+ vpadal.s16 q13, q0 ;sum
+ vmlal.s16 q14, d0, d0 ;sse
+ vmlal.s16 q15, d1, d1
+
+ vpadal.s16 q13, q1 ;sum
+ vmlal.s16 q14, d2, d2 ;sse
+ vmlal.s16 q15, d3, d3
+
+ vpadal.s16 q13, q5 ;sum
+ vmlal.s16 q14, d10, d10 ;sse
+ vmlal.s16 q15, d11, d11
+
+ vmov q0, q4
+
+ vpadal.s16 q13, q6 ;sum
+ vmlal.s16 q14, d12, d12 ;sse
+ vmlal.s16 q15, d13, d13
+
+ bne vp8_filt16x16s_4_4_loop_neon
+
+ vadd.u32 q15, q14, q15 ;accumulate sse
+ vpaddl.s32 q0, q13 ;accumulate sum
+
+ vpaddl.u32 q1, q15
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [lr] ;store sse
+ vshr.u32 d10, d10, #8
+ vsub.u32 d0, d1, d10
+
+ vmov.32 r0, d0[0] ;return
+ pop {pc}
+ ENDP
+
+;==============================
+; r0 unsigned char *src_ptr,
+; r1 int src_pixels_per_line,
+; r2 int xoffset,
+; r3 int yoffset,
+; stack unsigned char *dst_ptr,
+; stack int dst_pixels_per_line,
+; stack unsigned int *sse
+;note: in vp8_find_best_half_pixel_step()(called when 8<Speed<15), and first call of vp8_find_best_sub_pixel_step()
+;(called when speed<=8). xoffset/yoffset can only be 4 or 0, which means either by pass the filter,
+;or filter coeff is {64, 64}. This simplified program only works in this situation.
+;note: It happens that both xoffset and yoffset are zero. This can be handled in c code later.
+
+|vp8_sub_pixel_variance16x16s_neon| PROC
+ push {r4, lr}
+
+ ldr r4, [sp, #8] ;load *dst_ptr from stack
+ ldr r12, [sp, #12] ;load dst_pixels_per_line from stack
+ ldr lr, [sp, #16] ;load *sse from stack
+
+ cmp r2, #0 ;skip first_pass filter if xoffset=0
+ beq secondpass_bfilter16x16s_only
+
+ cmp r3, #0 ;skip second_pass filter if yoffset=0
+ beq firstpass_bfilter16x16s_only
+
+ vld1.u8 {d0, d1, d2, d3}, [r0], r1 ;load src data
+ sub sp, sp, #256 ;reserve space on stack for temporary storage
+ vext.8 q1, q0, q1, #1 ;construct src_ptr[1]
+ mov r3, sp
+ mov r2, #4 ;loop counter
+ vrhadd.u8 q0, q0, q1 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
+
+;First Pass: output_height lines x output_width columns (17x16)
+vp8e_filt_blk2d_fp16x16s_loop_neon
+ vld1.u8 {d4, d5, d6, d7}, [r0], r1
+ vld1.u8 {d8, d9, d10, d11}, [r0], r1
+ vld1.u8 {d12, d13, d14, d15}, [r0], r1
+ vld1.u8 {d16, d17, d18, d19}, [r0], r1
+
+ ;pld [r0]
+ ;pld [r0, r1]
+ ;pld [r0, r1, lsl #1]
+
+ vext.8 q3, q2, q3, #1 ;construct src_ptr[1]
+ vext.8 q5, q4, q5, #1
+ vext.8 q7, q6, q7, #1
+ vext.8 q9, q8, q9, #1
+
+ vrhadd.u8 q1, q2, q3 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
+ vrhadd.u8 q2, q4, q5
+ vrhadd.u8 q3, q6, q7
+ vrhadd.u8 q4, q8, q9
+
+ vrhadd.u8 q0, q0, q1
+ vrhadd.u8 q1, q1, q2
+ vrhadd.u8 q2, q2, q3
+ vrhadd.u8 q3, q3, q4
+
+ subs r2, r2, #1
+ vst1.u8 {d0, d1 ,d2, d3}, [r3]! ;store result
+ vmov q0, q4
+ vst1.u8 {d4, d5, d6, d7}, [r3]!
+
+ bne vp8e_filt_blk2d_fp16x16s_loop_neon
+
+ b sub_pixel_variance16x16s_neon
+
+;--------------------
+firstpass_bfilter16x16s_only
+ mov r2, #2 ;loop counter
+ sub sp, sp, #256 ;reserve space on stack for temporary storage
+ mov r3, sp
+
+;First Pass: output_height lines x output_width columns (16x16)
+vp8e_filt_blk2d_fpo16x16s_loop_neon
+ vld1.u8 {d0, d1, d2, d3}, [r0], r1 ;load src data
+ vld1.u8 {d4, d5, d6, d7}, [r0], r1
+ vld1.u8 {d8, d9, d10, d11}, [r0], r1
+ vld1.u8 {d12, d13, d14, d15}, [r0], r1
+
+ ;pld [r0]
+ ;pld [r0, r1]
+ ;pld [r0, r1, lsl #1]
+
+ vext.8 q1, q0, q1, #1 ;construct src_ptr[1]
+ vld1.u8 {d16, d17, d18, d19}, [r0], r1
+ vext.8 q3, q2, q3, #1
+ vld1.u8 {d20, d21, d22, d23}, [r0], r1
+ vext.8 q5, q4, q5, #1
+ vld1.u8 {d24, d25, d26, d27}, [r0], r1
+ vext.8 q7, q6, q7, #1
+ vld1.u8 {d28, d29, d30, d31}, [r0], r1
+ vext.8 q9, q8, q9, #1
+ vext.8 q11, q10, q11, #1
+ vext.8 q13, q12, q13, #1
+ vext.8 q15, q14, q15, #1
+
+ vrhadd.u8 q0, q0, q1 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
+ vrhadd.u8 q1, q2, q3
+ vrhadd.u8 q2, q4, q5
+ vrhadd.u8 q3, q6, q7
+ vrhadd.u8 q4, q8, q9
+ vrhadd.u8 q5, q10, q11
+ vrhadd.u8 q6, q12, q13
+ vrhadd.u8 q7, q14, q15
+
+ subs r2, r2, #1
+
+ vst1.u8 {d0, d1, d2, d3}, [r3]! ;store result
+ vst1.u8 {d4, d5, d6, d7}, [r3]!
+ vst1.u8 {d8, d9, d10, d11}, [r3]!
+ vst1.u8 {d12, d13, d14, d15}, [r3]!
+
+ bne vp8e_filt_blk2d_fpo16x16s_loop_neon
+
+ b sub_pixel_variance16x16s_neon
+
+;---------------------
+secondpass_bfilter16x16s_only
+ sub sp, sp, #256 ;reserve space on stack for temporary storage
+
+ mov r2, #2 ;loop counter
+ vld1.u8 {d0, d1}, [r0], r1 ;load src data
+ mov r3, sp
+
+vp8e_filt_blk2d_spo16x16s_loop_neon
+ vld1.u8 {d2, d3}, [r0], r1
+ vld1.u8 {d4, d5}, [r0], r1
+ vld1.u8 {d6, d7}, [r0], r1
+ vld1.u8 {d8, d9}, [r0], r1
+
+ vrhadd.u8 q0, q0, q1
+ vld1.u8 {d10, d11}, [r0], r1
+ vrhadd.u8 q1, q1, q2
+ vld1.u8 {d12, d13}, [r0], r1
+ vrhadd.u8 q2, q2, q3
+ vld1.u8 {d14, d15}, [r0], r1
+ vrhadd.u8 q3, q3, q4
+ vld1.u8 {d16, d17}, [r0], r1
+ vrhadd.u8 q4, q4, q5
+ vrhadd.u8 q5, q5, q6
+ vrhadd.u8 q6, q6, q7
+ vrhadd.u8 q7, q7, q8
+
+ subs r2, r2, #1
+
+ vst1.u8 {d0, d1, d2, d3}, [r3]! ;store result
+ vmov q0, q8
+ vst1.u8 {d4, d5, d6, d7}, [r3]!
+ vst1.u8 {d8, d9, d10, d11}, [r3]! ;store result
+ vst1.u8 {d12, d13, d14, d15}, [r3]!
+
+ bne vp8e_filt_blk2d_spo16x16s_loop_neon
+
+ b sub_pixel_variance16x16s_neon
+
+;----------------------------
+;variance16x16
+sub_pixel_variance16x16s_neon
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+ sub r3, r3, #256
+ mov r2, #4
+
+sub_pixel_variance16x16s_neon_loop
+ vld1.8 {q0}, [r3]! ;Load up source and reference
+ vld1.8 {q1}, [r4], r12
+ vld1.8 {q2}, [r3]!
+ vld1.8 {q3}, [r4], r12
+ vld1.8 {q4}, [r3]!
+ vld1.8 {q5}, [r4], r12
+ vld1.8 {q6}, [r3]!
+ vld1.8 {q7}, [r4], r12
+
+ vsubl.u8 q11, d0, d2 ;diff
+ vsubl.u8 q12, d1, d3
+ vsubl.u8 q13, d4, d6
+ vsubl.u8 q14, d5, d7
+ vsubl.u8 q0, d8, d10
+ vsubl.u8 q1, d9, d11
+ vsubl.u8 q2, d12, d14
+ vsubl.u8 q3, d13, d15
+
+ vpadal.s16 q8, q11 ;sum
+ vmlal.s16 q9, d22, d22 ;sse
+ vmlal.s16 q10, d23, d23
+
+ subs r2, r2, #1
+
+ vpadal.s16 q8, q12
+ vmlal.s16 q9, d24, d24
+ vmlal.s16 q10, d25, d25
+ vpadal.s16 q8, q13
+ vmlal.s16 q9, d26, d26
+ vmlal.s16 q10, d27, d27
+ vpadal.s16 q8, q14
+ vmlal.s16 q9, d28, d28
+ vmlal.s16 q10, d29, d29
+
+ vpadal.s16 q8, q0 ;sum
+ vmlal.s16 q9, d0, d0 ;sse
+ vmlal.s16 q10, d1, d1
+ vpadal.s16 q8, q1
+ vmlal.s16 q9, d2, d2
+ vmlal.s16 q10, d3, d3
+ vpadal.s16 q8, q2
+ vmlal.s16 q9, d4, d4
+ vmlal.s16 q10, d5, d5
+ vpadal.s16 q8, q3
+ vmlal.s16 q9, d6, d6
+ vmlal.s16 q10, d7, d7
+
+ bne sub_pixel_variance16x16s_neon_loop
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [lr] ;store sse
+ vshr.u32 d10, d10, #8
+ vsub.u32 d0, d1, d10
+
+ add sp, sp, #256
+ vmov.32 r0, d0[0] ;return
+
+ pop {r4, pc}
+ ENDP
+
+ END
diff --git a/libvpx/vp8/common/arm/neon/vp8_subpixelvariance8x8_neon.asm b/libvpx/vp8/common/arm/neon/vp8_subpixelvariance8x8_neon.asm
new file mode 100644
index 0000000..f6b6847
--- /dev/null
+++ b/libvpx/vp8/common/arm/neon/vp8_subpixelvariance8x8_neon.asm
@@ -0,0 +1,222 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_sub_pixel_variance8x8_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+; r0 unsigned char *src_ptr,
+; r1 int src_pixels_per_line,
+; r2 int xoffset,
+; r3 int yoffset,
+; stack(r4) unsigned char *dst_ptr,
+; stack(r5) int dst_pixels_per_line,
+; stack(r6) unsigned int *sse
+;note: most of the code is copied from bilinear_predict8x8_neon and vp8_variance8x8_neon.
+
+|vp8_sub_pixel_variance8x8_neon| PROC
+ push {r4-r5, lr}
+
+ adr r12, bilinear_taps_coeff
+ ldr r4, [sp, #12] ;load *dst_ptr from stack
+ ldr r5, [sp, #16] ;load dst_pixels_per_line from stack
+ ldr lr, [sp, #20] ;load *sse from stack
+
+ cmp r2, #0 ;skip first_pass filter if xoffset=0
+ beq skip_firstpass_filter
+
+;First pass: output_height lines x output_width columns (9x8)
+ add r2, r12, r2, lsl #3 ;calculate filter location
+
+ vld1.u8 {q1}, [r0], r1 ;load src data
+ vld1.u32 {d31}, [r2] ;load first_pass filter
+ vld1.u8 {q2}, [r0], r1
+ vdup.8 d0, d31[0] ;first_pass filter (d0 d1)
+ vld1.u8 {q3}, [r0], r1
+ vdup.8 d1, d31[4]
+ vld1.u8 {q4}, [r0], r1
+
+ vmull.u8 q6, d2, d0 ;(src_ptr[0] * Filter[0])
+ vmull.u8 q7, d4, d0
+ vmull.u8 q8, d6, d0
+ vmull.u8 q9, d8, d0
+
+ vext.8 d3, d2, d3, #1 ;construct src_ptr[-1]
+ vext.8 d5, d4, d5, #1
+ vext.8 d7, d6, d7, #1
+ vext.8 d9, d8, d9, #1
+
+ vmlal.u8 q6, d3, d1 ;(src_ptr[1] * Filter[1])
+ vmlal.u8 q7, d5, d1
+ vmlal.u8 q8, d7, d1
+ vmlal.u8 q9, d9, d1
+
+ vld1.u8 {q1}, [r0], r1 ;load src data
+ vqrshrn.u16 d22, q6, #7 ;shift/round/saturate to u8
+ vld1.u8 {q2}, [r0], r1
+ vqrshrn.u16 d23, q7, #7
+ vld1.u8 {q3}, [r0], r1
+ vqrshrn.u16 d24, q8, #7
+ vld1.u8 {q4}, [r0], r1
+ vqrshrn.u16 d25, q9, #7
+
+ ;first_pass filtering on the rest 5-line data
+ vld1.u8 {q5}, [r0], r1
+
+ vmull.u8 q6, d2, d0 ;(src_ptr[0] * Filter[0])
+ vmull.u8 q7, d4, d0
+ vmull.u8 q8, d6, d0
+ vmull.u8 q9, d8, d0
+ vmull.u8 q10, d10, d0
+
+ vext.8 d3, d2, d3, #1 ;construct src_ptr[-1]
+ vext.8 d5, d4, d5, #1
+ vext.8 d7, d6, d7, #1
+ vext.8 d9, d8, d9, #1
+ vext.8 d11, d10, d11, #1
+
+ vmlal.u8 q6, d3, d1 ;(src_ptr[1] * Filter[1])
+ vmlal.u8 q7, d5, d1
+ vmlal.u8 q8, d7, d1
+ vmlal.u8 q9, d9, d1
+ vmlal.u8 q10, d11, d1
+
+ vqrshrn.u16 d26, q6, #7 ;shift/round/saturate to u8
+ vqrshrn.u16 d27, q7, #7
+ vqrshrn.u16 d28, q8, #7
+ vqrshrn.u16 d29, q9, #7
+ vqrshrn.u16 d30, q10, #7
+
+;Second pass: 8x8
+secondpass_filter
+ cmp r3, #0 ;skip second_pass filter if yoffset=0
+ ;skip_secondpass_filter
+ beq sub_pixel_variance8x8_neon
+
+ add r3, r12, r3, lsl #3
+
+ vld1.u32 {d31}, [r3] ;load second_pass filter
+
+ vdup.8 d0, d31[0] ;second_pass filter parameters (d0 d1)
+ vdup.8 d1, d31[4]
+
+ vmull.u8 q1, d22, d0 ;(src_ptr[0] * Filter[0])
+ vmull.u8 q2, d23, d0
+ vmull.u8 q3, d24, d0
+ vmull.u8 q4, d25, d0
+ vmull.u8 q5, d26, d0
+ vmull.u8 q6, d27, d0
+ vmull.u8 q7, d28, d0
+ vmull.u8 q8, d29, d0
+
+ vmlal.u8 q1, d23, d1 ;(src_ptr[pixel_step] * Filter[1])
+ vmlal.u8 q2, d24, d1
+ vmlal.u8 q3, d25, d1
+ vmlal.u8 q4, d26, d1
+ vmlal.u8 q5, d27, d1
+ vmlal.u8 q6, d28, d1
+ vmlal.u8 q7, d29, d1
+ vmlal.u8 q8, d30, d1
+
+ vqrshrn.u16 d22, q1, #7 ;shift/round/saturate to u8
+ vqrshrn.u16 d23, q2, #7
+ vqrshrn.u16 d24, q3, #7
+ vqrshrn.u16 d25, q4, #7
+ vqrshrn.u16 d26, q5, #7
+ vqrshrn.u16 d27, q6, #7
+ vqrshrn.u16 d28, q7, #7
+ vqrshrn.u16 d29, q8, #7
+
+ b sub_pixel_variance8x8_neon
+
+;--------------------
+skip_firstpass_filter
+ vld1.u8 {d22}, [r0], r1 ;load src data
+ vld1.u8 {d23}, [r0], r1
+ vld1.u8 {d24}, [r0], r1
+ vld1.u8 {d25}, [r0], r1
+ vld1.u8 {d26}, [r0], r1
+ vld1.u8 {d27}, [r0], r1
+ vld1.u8 {d28}, [r0], r1
+ vld1.u8 {d29}, [r0], r1
+ vld1.u8 {d30}, [r0], r1
+
+ b secondpass_filter
+
+;----------------------
+;vp8_variance8x8_neon
+sub_pixel_variance8x8_neon
+ vmov.i8 q8, #0 ;q8 - sum
+ vmov.i8 q9, #0 ;q9, q10 - sse
+ vmov.i8 q10, #0
+
+ mov r12, #2
+
+sub_pixel_variance8x8_neon_loop
+ vld1.8 {d0}, [r4], r5 ;load dst data
+ subs r12, r12, #1
+ vld1.8 {d1}, [r4], r5
+ vld1.8 {d2}, [r4], r5
+ vsubl.u8 q4, d22, d0 ;calculate diff
+ vld1.8 {d3}, [r4], r5
+
+ vsubl.u8 q5, d23, d1
+ vsubl.u8 q6, d24, d2
+
+ vpadal.s16 q8, q4 ;sum
+ vmlal.s16 q9, d8, d8 ;sse
+ vmlal.s16 q10, d9, d9
+
+ vsubl.u8 q7, d25, d3
+
+ vpadal.s16 q8, q5
+ vmlal.s16 q9, d10, d10
+ vmlal.s16 q10, d11, d11
+
+ vmov q11, q13
+
+ vpadal.s16 q8, q6
+ vmlal.s16 q9, d12, d12
+ vmlal.s16 q10, d13, d13
+
+ vmov q12, q14
+
+ vpadal.s16 q8, q7
+ vmlal.s16 q9, d14, d14
+ vmlal.s16 q10, d15, d15
+
+ bne sub_pixel_variance8x8_neon_loop
+
+ vadd.u32 q10, q9, q10 ;accumulate sse
+ vpaddl.s32 q0, q8 ;accumulate sum
+
+ vpaddl.u32 q1, q10
+ vadd.s64 d0, d0, d1
+ vadd.u64 d1, d2, d3
+
+ vmull.s32 q5, d0, d0
+ vst1.32 {d1[0]}, [lr] ;store sse
+ vshr.u32 d10, d10, #6
+ vsub.u32 d0, d1, d10
+
+ vmov.32 r0, d0[0] ;return
+ pop {r4-r5, pc}
+
+ ENDP
+
+;-----------------
+
+bilinear_taps_coeff
+ DCD 128, 0, 112, 16, 96, 32, 80, 48, 64, 64, 48, 80, 32, 96, 16, 112
+
+ END
diff --git a/libvpx/vp8/common/arm/reconintra_arm.c b/libvpx/vp8/common/arm/reconintra_arm.c
new file mode 100644
index 0000000..121e090
--- /dev/null
+++ b/libvpx/vp8/common/arm/reconintra_arm.c
@@ -0,0 +1,58 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/common/blockd.h"
+#include "vpx_mem/vpx_mem.h"
+
+#if HAVE_NEON
+extern void vp8_build_intra_predictors_mby_neon_func(
+ unsigned char *y_buffer,
+ unsigned char *ypred_ptr,
+ int y_stride,
+ int mode,
+ int Up,
+ int Left);
+
+void vp8_build_intra_predictors_mby_neon(MACROBLOCKD *x)
+{
+ unsigned char *y_buffer = x->dst.y_buffer;
+ unsigned char *ypred_ptr = x->predictor;
+ int y_stride = x->dst.y_stride;
+ int mode = x->mode_info_context->mbmi.mode;
+ int Up = x->up_available;
+ int Left = x->left_available;
+
+ vp8_build_intra_predictors_mby_neon_func(y_buffer, ypred_ptr, y_stride, mode, Up, Left);
+}
+
+extern void vp8_build_intra_predictors_mby_s_neon_func(
+ unsigned char *y_buffer,
+ unsigned char *ypred_ptr,
+ int y_stride,
+ int mode,
+ int Up,
+ int Left);
+
+void vp8_build_intra_predictors_mby_s_neon(MACROBLOCKD *x)
+{
+ unsigned char *y_buffer = x->dst.y_buffer;
+ unsigned char *ypred_ptr = x->predictor;
+ int y_stride = x->dst.y_stride;
+ int mode = x->mode_info_context->mbmi.mode;
+ int Up = x->up_available;
+ int Left = x->left_available;
+
+ vp8_build_intra_predictors_mby_s_neon_func(y_buffer, ypred_ptr, y_stride, mode, Up, Left);
+}
+
+#endif
diff --git a/libvpx/vp8/common/arm/variance_arm.c b/libvpx/vp8/common/arm/variance_arm.c
new file mode 100644
index 0000000..891d767
--- /dev/null
+++ b/libvpx/vp8/common/arm/variance_arm.c
@@ -0,0 +1,132 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/common/variance.h"
+#include "vp8/common/filter.h"
+
+#if HAVE_MEDIA
+#include "vp8/common/arm/bilinearfilter_arm.h"
+
+unsigned int vp8_sub_pixel_variance8x8_armv6
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ unsigned short first_pass[10*8];
+ unsigned char second_pass[8*8];
+ const short *HFilter, *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ vp8_filter_block2d_bil_first_pass_armv6(src_ptr, first_pass,
+ src_pixels_per_line,
+ 9, 8, HFilter);
+ vp8_filter_block2d_bil_second_pass_armv6(first_pass, second_pass,
+ 8, 8, 8, VFilter);
+
+ return vp8_variance8x8_armv6(second_pass, 8, dst_ptr,
+ dst_pixels_per_line, sse);
+}
+
+unsigned int vp8_sub_pixel_variance16x16_armv6
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ unsigned short first_pass[36*16];
+ unsigned char second_pass[20*16];
+ const short *HFilter, *VFilter;
+ unsigned int var;
+
+ if (xoffset == 4 && yoffset == 0)
+ {
+ var = vp8_variance_halfpixvar16x16_h_armv6(src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, sse);
+ }
+ else if (xoffset == 0 && yoffset == 4)
+ {
+ var = vp8_variance_halfpixvar16x16_v_armv6(src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, sse);
+ }
+ else if (xoffset == 4 && yoffset == 4)
+ {
+ var = vp8_variance_halfpixvar16x16_hv_armv6(src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, sse);
+ }
+ else
+ {
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ vp8_filter_block2d_bil_first_pass_armv6(src_ptr, first_pass,
+ src_pixels_per_line,
+ 17, 16, HFilter);
+ vp8_filter_block2d_bil_second_pass_armv6(first_pass, second_pass,
+ 16, 16, 16, VFilter);
+
+ var = vp8_variance16x16_armv6(second_pass, 16, dst_ptr,
+ dst_pixels_per_line, sse);
+ }
+ return var;
+}
+
+#endif /* HAVE_MEDIA */
+
+
+#if HAVE_NEON
+
+extern unsigned int vp8_sub_pixel_variance16x16_neon_func
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+);
+
+unsigned int vp8_sub_pixel_variance16x16_neon
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ if (xoffset == 4 && yoffset == 0)
+ return vp8_variance_halfpixvar16x16_h_neon(src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
+ else if (xoffset == 0 && yoffset == 4)
+ return vp8_variance_halfpixvar16x16_v_neon(src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
+ else if (xoffset == 4 && yoffset == 4)
+ return vp8_variance_halfpixvar16x16_hv_neon(src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
+ else
+ return vp8_sub_pixel_variance16x16_neon_func(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
+}
+
+#endif
diff --git a/libvpx/vp8/common/asm_com_offsets.c b/libvpx/vp8/common/asm_com_offsets.c
new file mode 100644
index 0000000..ae22b5f
--- /dev/null
+++ b/libvpx/vp8/common/asm_com_offsets.c
@@ -0,0 +1,71 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx/vpx_codec.h"
+#include "vpx_ports/asm_offsets.h"
+#include "vpx_scale/yv12config.h"
+#include "vp8/common/blockd.h"
+
+#if CONFIG_POSTPROC
+#include "postproc.h"
+#endif /* CONFIG_POSTPROC */
+
+BEGIN
+
+/* vpx_scale */
+DEFINE(yv12_buffer_config_y_width, offsetof(YV12_BUFFER_CONFIG, y_width));
+DEFINE(yv12_buffer_config_y_height, offsetof(YV12_BUFFER_CONFIG, y_height));
+DEFINE(yv12_buffer_config_y_stride, offsetof(YV12_BUFFER_CONFIG, y_stride));
+DEFINE(yv12_buffer_config_uv_width, offsetof(YV12_BUFFER_CONFIG, uv_width));
+DEFINE(yv12_buffer_config_uv_height, offsetof(YV12_BUFFER_CONFIG, uv_height));
+DEFINE(yv12_buffer_config_uv_stride, offsetof(YV12_BUFFER_CONFIG, uv_stride));
+DEFINE(yv12_buffer_config_y_buffer, offsetof(YV12_BUFFER_CONFIG, y_buffer));
+DEFINE(yv12_buffer_config_u_buffer, offsetof(YV12_BUFFER_CONFIG, u_buffer));
+DEFINE(yv12_buffer_config_v_buffer, offsetof(YV12_BUFFER_CONFIG, v_buffer));
+DEFINE(yv12_buffer_config_border, offsetof(YV12_BUFFER_CONFIG, border));
+DEFINE(VP8BORDERINPIXELS_VAL, VP8BORDERINPIXELS);
+
+#if CONFIG_POSTPROC
+/* mfqe.c / filter_by_weight */
+DEFINE(MFQE_PRECISION_VAL, MFQE_PRECISION);
+#endif /* CONFIG_POSTPROC */
+
+END
+
+/* add asserts for any offset that is not supported by assembly code */
+/* add asserts for any size that is not supported by assembly code */
+
+#if HAVE_MEDIA
+/* switch case in vp8_intra4x4_predict_armv6 is based on these enumerated values */
+ct_assert(B_DC_PRED, B_DC_PRED == 0);
+ct_assert(B_TM_PRED, B_TM_PRED == 1);
+ct_assert(B_VE_PRED, B_VE_PRED == 2);
+ct_assert(B_HE_PRED, B_HE_PRED == 3);
+ct_assert(B_LD_PRED, B_LD_PRED == 4);
+ct_assert(B_RD_PRED, B_RD_PRED == 5);
+ct_assert(B_VR_PRED, B_VR_PRED == 6);
+ct_assert(B_VL_PRED, B_VL_PRED == 7);
+ct_assert(B_HD_PRED, B_HD_PRED == 8);
+ct_assert(B_HU_PRED, B_HU_PRED == 9);
+#endif
+
+#if HAVE_NEON
+/* vp8_yv12_extend_frame_borders_neon makes several assumptions based on this */
+ct_assert(VP8BORDERINPIXELS_VAL, VP8BORDERINPIXELS == 32)
+#endif
+
+#if HAVE_SSE2
+#if CONFIG_POSTPROC
+/* vp8_filter_by_weight16x16 and 8x8 */
+ct_assert(MFQE_PRECISION_VAL, MFQE_PRECISION == 4)
+#endif /* CONFIG_POSTPROC */
+#endif /* HAVE_SSE2 */
diff --git a/vp8/common/blockd.c b/libvpx/vp8/common/blockd.c
similarity index 100%
rename from vp8/common/blockd.c
rename to libvpx/vp8/common/blockd.c
diff --git a/libvpx/vp8/common/blockd.h b/libvpx/vp8/common/blockd.h
new file mode 100644
index 0000000..f7ff577
--- /dev/null
+++ b/libvpx/vp8/common/blockd.h
@@ -0,0 +1,300 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_BLOCKD_H
+#define __INC_BLOCKD_H
+
+void vpx_log(const char *format, ...);
+
+#include "vpx_config.h"
+#include "vpx_scale/yv12config.h"
+#include "mv.h"
+#include "treecoder.h"
+#include "vpx_ports/mem.h"
+
+/*#define DCPRED 1*/
+#define DCPREDSIMTHRESH 0
+#define DCPREDCNTTHRESH 3
+
+#define MB_FEATURE_TREE_PROBS 3
+#define MAX_MB_SEGMENTS 4
+
+#define MAX_REF_LF_DELTAS 4
+#define MAX_MODE_LF_DELTAS 4
+
+/* Segment Feature Masks */
+#define SEGMENT_DELTADATA 0
+#define SEGMENT_ABSDATA 1
+
+typedef struct
+{
+ int r, c;
+} POS;
+
+#define PLANE_TYPE_Y_NO_DC 0
+#define PLANE_TYPE_Y2 1
+#define PLANE_TYPE_UV 2
+#define PLANE_TYPE_Y_WITH_DC 3
+
+
+typedef char ENTROPY_CONTEXT;
+typedef struct
+{
+ ENTROPY_CONTEXT y1[4];
+ ENTROPY_CONTEXT u[2];
+ ENTROPY_CONTEXT v[2];
+ ENTROPY_CONTEXT y2;
+} ENTROPY_CONTEXT_PLANES;
+
+extern const unsigned char vp8_block2left[25];
+extern const unsigned char vp8_block2above[25];
+
+#define VP8_COMBINEENTROPYCONTEXTS( Dest, A, B) \
+ Dest = (A)+(B);
+
+
+typedef enum
+{
+ KEY_FRAME = 0,
+ INTER_FRAME = 1
+} FRAME_TYPE;
+
+typedef enum
+{
+ DC_PRED, /* average of above and left pixels */
+ V_PRED, /* vertical prediction */
+ H_PRED, /* horizontal prediction */
+ TM_PRED, /* Truemotion prediction */
+ B_PRED, /* block based prediction, each block has its own prediction mode */
+
+ NEARESTMV,
+ NEARMV,
+ ZEROMV,
+ NEWMV,
+ SPLITMV,
+
+ MB_MODE_COUNT
+} MB_PREDICTION_MODE;
+
+/* Macroblock level features */
+typedef enum
+{
+ MB_LVL_ALT_Q = 0, /* Use alternate Quantizer .... */
+ MB_LVL_ALT_LF = 1, /* Use alternate loop filter value... */
+ MB_LVL_MAX = 2 /* Number of MB level features supported */
+
+} MB_LVL_FEATURES;
+
+/* Segment Feature Masks */
+#define SEGMENT_ALTQ 0x01
+#define SEGMENT_ALT_LF 0x02
+
+#define VP8_YMODES (B_PRED + 1)
+#define VP8_UV_MODES (TM_PRED + 1)
+
+#define VP8_MVREFS (1 + SPLITMV - NEARESTMV)
+
+typedef enum
+{
+ B_DC_PRED, /* average of above and left pixels */
+ B_TM_PRED,
+
+ B_VE_PRED, /* vertical prediction */
+ B_HE_PRED, /* horizontal prediction */
+
+ B_LD_PRED,
+ B_RD_PRED,
+
+ B_VR_PRED,
+ B_VL_PRED,
+ B_HD_PRED,
+ B_HU_PRED,
+
+ LEFT4X4,
+ ABOVE4X4,
+ ZERO4X4,
+ NEW4X4,
+
+ B_MODE_COUNT
+} B_PREDICTION_MODE;
+
+#define VP8_BINTRAMODES (B_HU_PRED + 1) /* 10 */
+#define VP8_SUBMVREFS (1 + NEW4X4 - LEFT4X4)
+
+/* For keyframes, intra block modes are predicted by the (already decoded)
+ modes for the Y blocks to the left and above us; for interframes, there
+ is a single probability table. */
+
+union b_mode_info
+{
+ B_PREDICTION_MODE as_mode;
+ int_mv mv;
+};
+
+typedef enum
+{
+ INTRA_FRAME = 0,
+ LAST_FRAME = 1,
+ GOLDEN_FRAME = 2,
+ ALTREF_FRAME = 3,
+ MAX_REF_FRAMES = 4
+} MV_REFERENCE_FRAME;
+
+typedef struct
+{
+ uint8_t mode, uv_mode;
+ uint8_t ref_frame;
+ uint8_t is_4x4;
+ int_mv mv;
+
+ uint8_t partitioning;
+ uint8_t mb_skip_coeff; /* does this mb has coefficients at all, 1=no coefficients, 0=need decode tokens */
+ uint8_t need_to_clamp_mvs;
+ uint8_t segment_id; /* Which set of segmentation parameters should be used for this MB */
+} MB_MODE_INFO;
+
+typedef struct modeinfo
+{
+ MB_MODE_INFO mbmi;
+ union b_mode_info bmi[16];
+} MODE_INFO;
+
+#if CONFIG_MULTI_RES_ENCODING
+/* The mb-level information needed to be stored for higher-resolution encoder */
+typedef struct
+{
+ MB_PREDICTION_MODE mode;
+ MV_REFERENCE_FRAME ref_frame;
+ int_mv mv;
+ int dissim; /* dissimilarity level of the macroblock */
+} LOWER_RES_MB_INFO;
+
+/* The frame-level information needed to be stored for higher-resolution
+ * encoder */
+typedef struct
+{
+ FRAME_TYPE frame_type;
+ int is_frame_dropped;
+ /* The frame number of each reference frames */
+ unsigned int low_res_ref_frames[MAX_REF_FRAMES];
+ LOWER_RES_MB_INFO *mb_info;
+} LOWER_RES_FRAME_INFO;
+#endif
+
+typedef struct blockd
+{
+ short *qcoeff;
+ short *dqcoeff;
+ unsigned char *predictor;
+ short *dequant;
+
+ int offset;
+ char *eob;
+
+ union b_mode_info bmi;
+} BLOCKD;
+
+typedef void (*vp8_subpix_fn_t)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+typedef struct macroblockd
+{
+ DECLARE_ALIGNED(16, unsigned char, predictor[384]);
+ DECLARE_ALIGNED(16, short, qcoeff[400]);
+ DECLARE_ALIGNED(16, short, dqcoeff[400]);
+ DECLARE_ALIGNED(16, char, eobs[25]);
+
+ DECLARE_ALIGNED(16, short, dequant_y1[16]);
+ DECLARE_ALIGNED(16, short, dequant_y1_dc[16]);
+ DECLARE_ALIGNED(16, short, dequant_y2[16]);
+ DECLARE_ALIGNED(16, short, dequant_uv[16]);
+
+ /* 16 Y blocks, 4 U, 4 V, 1 DC 2nd order block, each with 16 entries. */
+ BLOCKD block[25];
+ int fullpixel_mask;
+
+ YV12_BUFFER_CONFIG pre; /* Filtered copy of previous frame reconstruction */
+ YV12_BUFFER_CONFIG dst;
+
+ MODE_INFO *mode_info_context;
+ int mode_info_stride;
+
+ FRAME_TYPE frame_type;
+
+ int up_available;
+ int left_available;
+
+ unsigned char *recon_above[3];
+ unsigned char *recon_left[3];
+ int recon_left_stride[2];
+
+ /* Y,U,V,Y2 */
+ ENTROPY_CONTEXT_PLANES *above_context;
+ ENTROPY_CONTEXT_PLANES *left_context;
+
+ /* 0 indicates segmentation at MB level is not enabled. Otherwise the individual bits indicate which features are active. */
+ unsigned char segmentation_enabled;
+
+ /* 0 (do not update) 1 (update) the macroblock segmentation map. */
+ unsigned char update_mb_segmentation_map;
+
+ /* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
+ unsigned char update_mb_segmentation_data;
+
+ /* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
+ unsigned char mb_segement_abs_delta;
+
+ /* Per frame flags that define which MB level features (such as quantizer or loop filter level) */
+ /* are enabled and when enabled the proabilities used to decode the per MB flags in MB_MODE_INFO */
+ vp8_prob mb_segment_tree_probs[MB_FEATURE_TREE_PROBS]; /* Probability Tree used to code Segment number */
+
+ signed char segment_feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS]; /* Segment parameters */
+
+ /* mode_based Loop filter adjustment */
+ unsigned char mode_ref_lf_delta_enabled;
+ unsigned char mode_ref_lf_delta_update;
+
+ /* Delta values have the range +/- MAX_LOOP_FILTER */
+ signed char last_ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
+ signed char ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
+ signed char last_mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
+ signed char mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
+
+ /* Distance of MB away from frame edges */
+ int mb_to_left_edge;
+ int mb_to_right_edge;
+ int mb_to_top_edge;
+ int mb_to_bottom_edge;
+
+
+
+ vp8_subpix_fn_t subpixel_predict;
+ vp8_subpix_fn_t subpixel_predict8x4;
+ vp8_subpix_fn_t subpixel_predict8x8;
+ vp8_subpix_fn_t subpixel_predict16x16;
+
+ void *current_bc;
+
+ int corrupted;
+
+#if ARCH_X86 || ARCH_X86_64
+ /* This is an intermediate buffer currently used in sub-pixel motion search
+ * to keep a copy of the reference area. This buffer can be used for other
+ * purpose.
+ */
+ DECLARE_ALIGNED(32, unsigned char, y_buf[22*32]);
+#endif
+} MACROBLOCKD;
+
+
+extern void vp8_build_block_doffsets(MACROBLOCKD *x);
+extern void vp8_setup_block_dptrs(MACROBLOCKD *x);
+
+#endif /* __INC_BLOCKD_H */
diff --git a/libvpx/vp8/common/coefupdateprobs.h b/libvpx/vp8/common/coefupdateprobs.h
new file mode 100644
index 0000000..9e194dc
--- /dev/null
+++ b/libvpx/vp8/common/coefupdateprobs.h
@@ -0,0 +1,185 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/* Update probabilities for the nodes in the token entropy tree.
+ Generated file included by entropy.c */
+
+const vp8_prob vp8_coef_update_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES] =
+{
+ {
+ {
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {176, 246, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {223, 241, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {249, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 244, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {234, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 246, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {239, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {251, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {251, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 254, 253, 255, 254, 255, 255, 255, 255, 255, 255, },
+ {250, 255, 254, 255, 254, 255, 255, 255, 255, 255, 255, },
+ {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ },
+ {
+ {
+ {217, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {225, 252, 241, 253, 255, 255, 254, 255, 255, 255, 255, },
+ {234, 250, 241, 250, 253, 255, 253, 254, 255, 255, 255, },
+ },
+ {
+ {255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {223, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {238, 253, 254, 254, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {249, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {247, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {252, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ },
+ {
+ {
+ {186, 251, 250, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {234, 251, 244, 254, 255, 255, 255, 255, 255, 255, 255, },
+ {251, 251, 243, 253, 254, 255, 254, 255, 255, 255, 255, },
+ },
+ {
+ {255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {236, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {251, 253, 253, 254, 254, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {254, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {254, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ },
+ {
+ {
+ {248, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {250, 254, 252, 254, 255, 255, 255, 255, 255, 255, 255, },
+ {248, 254, 249, 253, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {246, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {252, 254, 251, 254, 254, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 254, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {248, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {253, 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {245, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {253, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 251, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {252, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 252, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {249, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 255, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ {
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
+ },
+ },
+};
diff --git a/libvpx/vp8/common/common.h b/libvpx/vp8/common/common.h
new file mode 100644
index 0000000..2cc1c54
--- /dev/null
+++ b/libvpx/vp8/common/common.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef common_h
+#define common_h 1
+
+#include <assert.h>
+
+/* Interface header for common constant data structures and lookup tables */
+
+#include "vpx_mem/vpx_mem.h"
+
+/* Only need this for fixed-size arrays, for structs just assign. */
+
+#define vp8_copy( Dest, Src) { \
+ assert( sizeof( Dest) == sizeof( Src)); \
+ vpx_memcpy( Dest, Src, sizeof( Src)); \
+ }
+
+/* Use this for variably-sized arrays. */
+
+#define vp8_copy_array( Dest, Src, N) { \
+ assert( sizeof( *Dest) == sizeof( *Src)); \
+ vpx_memcpy( Dest, Src, N * sizeof( *Src)); \
+ }
+
+#define vp8_zero( Dest) vpx_memset( &Dest, 0, sizeof( Dest));
+
+#define vp8_zero_array( Dest, N) vpx_memset( Dest, 0, N * sizeof( *Dest));
+
+
+#endif /* common_h */
diff --git a/vp8/common/context.c b/libvpx/vp8/common/context.c
similarity index 100%
rename from vp8/common/context.c
rename to libvpx/vp8/common/context.c
diff --git a/libvpx/vp8/common/debugmodes.c b/libvpx/vp8/common/debugmodes.c
new file mode 100644
index 0000000..46064e6
--- /dev/null
+++ b/libvpx/vp8/common/debugmodes.c
@@ -0,0 +1,157 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdio.h>
+#include "blockd.h"
+
+
+void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols, int frame)
+{
+
+ int mb_row;
+ int mb_col;
+ int mb_index = 0;
+ FILE *mvs = fopen("mvs.stt", "a");
+
+ /* print out the macroblock Y modes */
+ mb_index = 0;
+ fprintf(mvs, "Mb Modes for Frame %d\n", frame);
+
+ for (mb_row = 0; mb_row < rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < cols; mb_col++)
+ {
+
+ fprintf(mvs, "%2d ", mi[mb_index].mbmi.mode);
+
+ mb_index++;
+ }
+
+ fprintf(mvs, "\n");
+ mb_index++;
+ }
+
+ fprintf(mvs, "\n");
+
+ mb_index = 0;
+ fprintf(mvs, "Mb mv ref for Frame %d\n", frame);
+
+ for (mb_row = 0; mb_row < rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < cols; mb_col++)
+ {
+
+ fprintf(mvs, "%2d ", mi[mb_index].mbmi.ref_frame);
+
+ mb_index++;
+ }
+
+ fprintf(mvs, "\n");
+ mb_index++;
+ }
+
+ fprintf(mvs, "\n");
+
+ /* print out the macroblock UV modes */
+ mb_index = 0;
+ fprintf(mvs, "UV Modes for Frame %d\n", frame);
+
+ for (mb_row = 0; mb_row < rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < cols; mb_col++)
+ {
+
+ fprintf(mvs, "%2d ", mi[mb_index].mbmi.uv_mode);
+
+ mb_index++;
+ }
+
+ mb_index++;
+ fprintf(mvs, "\n");
+ }
+
+ fprintf(mvs, "\n");
+
+ /* print out the block modes */
+ mb_index = 0;
+ fprintf(mvs, "Mbs for Frame %d\n", frame);
+ {
+ int b_row;
+
+ for (b_row = 0; b_row < 4 * rows; b_row++)
+ {
+ int b_col;
+ int bindex;
+
+ for (b_col = 0; b_col < 4 * cols; b_col++)
+ {
+ mb_index = (b_row >> 2) * (cols + 1) + (b_col >> 2);
+ bindex = (b_row & 3) * 4 + (b_col & 3);
+
+ if (mi[mb_index].mbmi.mode == B_PRED)
+ fprintf(mvs, "%2d ", mi[mb_index].bmi[bindex].as_mode);
+ else
+ fprintf(mvs, "xx ");
+
+ }
+
+ fprintf(mvs, "\n");
+ }
+ }
+ fprintf(mvs, "\n");
+
+ /* print out the macroblock mvs */
+ mb_index = 0;
+ fprintf(mvs, "MVs for Frame %d\n", frame);
+
+ for (mb_row = 0; mb_row < rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < cols; mb_col++)
+ {
+ fprintf(mvs, "%5d:%-5d", mi[mb_index].mbmi.mv.as_mv.row / 2, mi[mb_index].mbmi.mv.as_mv.col / 2);
+
+ mb_index++;
+ }
+
+ mb_index++;
+ fprintf(mvs, "\n");
+ }
+
+ fprintf(mvs, "\n");
+
+
+ /* print out the block modes */
+ mb_index = 0;
+ fprintf(mvs, "MVs for Frame %d\n", frame);
+ {
+ int b_row;
+
+ for (b_row = 0; b_row < 4 * rows; b_row++)
+ {
+ int b_col;
+ int bindex;
+
+ for (b_col = 0; b_col < 4 * cols; b_col++)
+ {
+ mb_index = (b_row >> 2) * (cols + 1) + (b_col >> 2);
+ bindex = (b_row & 3) * 4 + (b_col & 3);
+ fprintf(mvs, "%3d:%-3d ", mi[mb_index].bmi[bindex].mv.as_mv.row, mi[mb_index].bmi[bindex].mv.as_mv.col);
+
+ }
+
+ fprintf(mvs, "\n");
+ }
+ }
+ fprintf(mvs, "\n");
+
+
+ fclose(mvs);
+}
diff --git a/libvpx/vp8/common/default_coef_probs.h b/libvpx/vp8/common/default_coef_probs.h
new file mode 100644
index 0000000..0d19563
--- /dev/null
+++ b/libvpx/vp8/common/default_coef_probs.h
@@ -0,0 +1,188 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+*/
+
+
+/*Generated file, included by entropy.c*/
+
+
+static const vp8_prob default_coef_probs [BLOCK_TYPES]
+ [COEF_BANDS]
+ [PREV_COEF_CONTEXTS]
+ [ENTROPY_NODES] =
+{
+ { /* Block Type ( 0 ) */
+ { /* Coeff Band ( 0 )*/
+ { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 1 )*/
+ { 253, 136, 254, 255, 228, 219, 128, 128, 128, 128, 128 },
+ { 189, 129, 242, 255, 227, 213, 255, 219, 128, 128, 128 },
+ { 106, 126, 227, 252, 214, 209, 255, 255, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 2 )*/
+ { 1, 98, 248, 255, 236, 226, 255, 255, 128, 128, 128 },
+ { 181, 133, 238, 254, 221, 234, 255, 154, 128, 128, 128 },
+ { 78, 134, 202, 247, 198, 180, 255, 219, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 3 )*/
+ { 1, 185, 249, 255, 243, 255, 128, 128, 128, 128, 128 },
+ { 184, 150, 247, 255, 236, 224, 128, 128, 128, 128, 128 },
+ { 77, 110, 216, 255, 236, 230, 128, 128, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 4 )*/
+ { 1, 101, 251, 255, 241, 255, 128, 128, 128, 128, 128 },
+ { 170, 139, 241, 252, 236, 209, 255, 255, 128, 128, 128 },
+ { 37, 116, 196, 243, 228, 255, 255, 255, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 5 )*/
+ { 1, 204, 254, 255, 245, 255, 128, 128, 128, 128, 128 },
+ { 207, 160, 250, 255, 238, 128, 128, 128, 128, 128, 128 },
+ { 102, 103, 231, 255, 211, 171, 128, 128, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 6 )*/
+ { 1, 152, 252, 255, 240, 255, 128, 128, 128, 128, 128 },
+ { 177, 135, 243, 255, 234, 225, 128, 128, 128, 128, 128 },
+ { 80, 129, 211, 255, 194, 224, 128, 128, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 7 )*/
+ { 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 246, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 255, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
+ }
+ },
+ { /* Block Type ( 1 ) */
+ { /* Coeff Band ( 0 )*/
+ { 198, 35, 237, 223, 193, 187, 162, 160, 145, 155, 62 },
+ { 131, 45, 198, 221, 172, 176, 220, 157, 252, 221, 1 },
+ { 68, 47, 146, 208, 149, 167, 221, 162, 255, 223, 128 }
+ },
+ { /* Coeff Band ( 1 )*/
+ { 1, 149, 241, 255, 221, 224, 255, 255, 128, 128, 128 },
+ { 184, 141, 234, 253, 222, 220, 255, 199, 128, 128, 128 },
+ { 81, 99, 181, 242, 176, 190, 249, 202, 255, 255, 128 }
+ },
+ { /* Coeff Band ( 2 )*/
+ { 1, 129, 232, 253, 214, 197, 242, 196, 255, 255, 128 },
+ { 99, 121, 210, 250, 201, 198, 255, 202, 128, 128, 128 },
+ { 23, 91, 163, 242, 170, 187, 247, 210, 255, 255, 128 }
+ },
+ { /* Coeff Band ( 3 )*/
+ { 1, 200, 246, 255, 234, 255, 128, 128, 128, 128, 128 },
+ { 109, 178, 241, 255, 231, 245, 255, 255, 128, 128, 128 },
+ { 44, 130, 201, 253, 205, 192, 255, 255, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 4 )*/
+ { 1, 132, 239, 251, 219, 209, 255, 165, 128, 128, 128 },
+ { 94, 136, 225, 251, 218, 190, 255, 255, 128, 128, 128 },
+ { 22, 100, 174, 245, 186, 161, 255, 199, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 5 )*/
+ { 1, 182, 249, 255, 232, 235, 128, 128, 128, 128, 128 },
+ { 124, 143, 241, 255, 227, 234, 128, 128, 128, 128, 128 },
+ { 35, 77, 181, 251, 193, 211, 255, 205, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 6 )*/
+ { 1, 157, 247, 255, 236, 231, 255, 255, 128, 128, 128 },
+ { 121, 141, 235, 255, 225, 227, 255, 255, 128, 128, 128 },
+ { 45, 99, 188, 251, 195, 217, 255, 224, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 7 )*/
+ { 1, 1, 251, 255, 213, 255, 128, 128, 128, 128, 128 },
+ { 203, 1, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
+ { 137, 1, 177, 255, 224, 255, 128, 128, 128, 128, 128 }
+ }
+ },
+ { /* Block Type ( 2 ) */
+ { /* Coeff Band ( 0 )*/
+ { 253, 9, 248, 251, 207, 208, 255, 192, 128, 128, 128 },
+ { 175, 13, 224, 243, 193, 185, 249, 198, 255, 255, 128 },
+ { 73, 17, 171, 221, 161, 179, 236, 167, 255, 234, 128 }
+ },
+ { /* Coeff Band ( 1 )*/
+ { 1, 95, 247, 253, 212, 183, 255, 255, 128, 128, 128 },
+ { 239, 90, 244, 250, 211, 209, 255, 255, 128, 128, 128 },
+ { 155, 77, 195, 248, 188, 195, 255, 255, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 2 )*/
+ { 1, 24, 239, 251, 218, 219, 255, 205, 128, 128, 128 },
+ { 201, 51, 219, 255, 196, 186, 128, 128, 128, 128, 128 },
+ { 69, 46, 190, 239, 201, 218, 255, 228, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 3 )*/
+ { 1, 191, 251, 255, 255, 128, 128, 128, 128, 128, 128 },
+ { 223, 165, 249, 255, 213, 255, 128, 128, 128, 128, 128 },
+ { 141, 124, 248, 255, 255, 128, 128, 128, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 4 )*/
+ { 1, 16, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
+ { 190, 36, 230, 255, 236, 255, 128, 128, 128, 128, 128 },
+ { 149, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 5 )*/
+ { 1, 226, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 247, 192, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 240, 128, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 6 )*/
+ { 1, 134, 252, 255, 255, 128, 128, 128, 128, 128, 128 },
+ { 213, 62, 250, 255, 255, 128, 128, 128, 128, 128, 128 },
+ { 55, 93, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 7 )*/
+ { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
+ }
+ },
+ { /* Block Type ( 3 ) */
+ { /* Coeff Band ( 0 )*/
+ { 202, 24, 213, 235, 186, 191, 220, 160, 240, 175, 255 },
+ { 126, 38, 182, 232, 169, 184, 228, 174, 255, 187, 128 },
+ { 61, 46, 138, 219, 151, 178, 240, 170, 255, 216, 128 }
+ },
+ { /* Coeff Band ( 1 )*/
+ { 1, 112, 230, 250, 199, 191, 247, 159, 255, 255, 128 },
+ { 166, 109, 228, 252, 211, 215, 255, 174, 128, 128, 128 },
+ { 39, 77, 162, 232, 172, 180, 245, 178, 255, 255, 128 }
+ },
+ { /* Coeff Band ( 2 )*/
+ { 1, 52, 220, 246, 198, 199, 249, 220, 255, 255, 128 },
+ { 124, 74, 191, 243, 183, 193, 250, 221, 255, 255, 128 },
+ { 24, 71, 130, 219, 154, 170, 243, 182, 255, 255, 128 }
+ },
+ { /* Coeff Band ( 3 )*/
+ { 1, 182, 225, 249, 219, 240, 255, 224, 128, 128, 128 },
+ { 149, 150, 226, 252, 216, 205, 255, 171, 128, 128, 128 },
+ { 28, 108, 170, 242, 183, 194, 254, 223, 255, 255, 128 }
+ },
+ { /* Coeff Band ( 4 )*/
+ { 1, 81, 230, 252, 204, 203, 255, 192, 128, 128, 128 },
+ { 123, 102, 209, 247, 188, 196, 255, 233, 128, 128, 128 },
+ { 20, 95, 153, 243, 164, 173, 255, 203, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 5 )*/
+ { 1, 222, 248, 255, 216, 213, 128, 128, 128, 128, 128 },
+ { 168, 175, 246, 252, 235, 205, 255, 255, 128, 128, 128 },
+ { 47, 116, 215, 255, 211, 212, 255, 255, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 6 )*/
+ { 1, 121, 236, 253, 212, 214, 255, 255, 128, 128, 128 },
+ { 141, 84, 213, 252, 201, 202, 255, 219, 128, 128, 128 },
+ { 42, 80, 160, 240, 162, 185, 255, 205, 128, 128, 128 }
+ },
+ { /* Coeff Band ( 7 )*/
+ { 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 244, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
+ { 238, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
+ }
+ }
+};
diff --git a/libvpx/vp8/common/dequantize.c b/libvpx/vp8/common/dequantize.c
new file mode 100644
index 0000000..8eda486
--- /dev/null
+++ b/libvpx/vp8/common/dequantize.c
@@ -0,0 +1,43 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/common/blockd.h"
+#include "vpx_mem/vpx_mem.h"
+
+void vp8_dequantize_b_c(BLOCKD *d, short *DQC)
+{
+ int i;
+ short *DQ = d->dqcoeff;
+ short *Q = d->qcoeff;
+
+ for (i = 0; i < 16; i++)
+ {
+ DQ[i] = Q[i] * DQC[i];
+ }
+}
+
+void vp8_dequant_idct_add_c(short *input, short *dq,
+ unsigned char *dest, int stride)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ input[i] = dq[i] * input[i];
+ }
+
+ vp8_short_idct4x4llm_c(input, dest, stride, dest, stride);
+
+ vpx_memset(input, 0, 32);
+
+}
diff --git a/libvpx/vp8/common/entropy.c b/libvpx/vp8/common/entropy.c
new file mode 100644
index 0000000..8c046a4
--- /dev/null
+++ b/libvpx/vp8/common/entropy.c
@@ -0,0 +1,189 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "entropy.h"
+#include "blockd.h"
+#include "onyxc_int.h"
+#include "vpx_mem/vpx_mem.h"
+
+#include "coefupdateprobs.h"
+
+DECLARE_ALIGNED(16, const unsigned char, vp8_norm[256]) =
+{
+ 0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4,
+ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
+ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+};
+
+DECLARE_ALIGNED(16, const unsigned char, vp8_coef_bands[16]) =
+{ 0, 1, 2, 3, 6, 4, 5, 6, 6, 6, 6, 6, 6, 6, 6, 7};
+
+DECLARE_ALIGNED(16, const unsigned char,
+ vp8_prev_token_class[MAX_ENTROPY_TOKENS]) =
+{ 0, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 0};
+
+DECLARE_ALIGNED(16, const int, vp8_default_zig_zag1d[16]) =
+{
+ 0, 1, 4, 8,
+ 5, 2, 3, 6,
+ 9, 12, 13, 10,
+ 7, 11, 14, 15,
+};
+
+DECLARE_ALIGNED(16, const short, vp8_default_inv_zig_zag[16]) =
+{
+ 1, 2, 6, 7,
+ 3, 5, 8, 13,
+ 4, 9, 12, 14,
+ 10, 11, 15, 16
+};
+
+/* vp8_default_zig_zag_mask generated with:
+
+ void vp8_init_scan_order_mask()
+ {
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ vp8_default_zig_zag_mask[vp8_default_zig_zag1d[i]] = 1 << i;
+ }
+
+ }
+*/
+DECLARE_ALIGNED(16, const short, vp8_default_zig_zag_mask[16]) =
+{
+ 1, 2, 32, 64,
+ 4, 16, 128, 4096,
+ 8, 256, 2048, 8192,
+ 512, 1024, 16384, -32768
+};
+
+const int vp8_mb_feature_data_bits[MB_LVL_MAX] = {7, 6};
+
+/* Array indices are identical to previously-existing CONTEXT_NODE indices */
+
+const vp8_tree_index vp8_coef_tree[ 22] = /* corresponding _CONTEXT_NODEs */
+{
+ -DCT_EOB_TOKEN, 2, /* 0 = EOB */
+ -ZERO_TOKEN, 4, /* 1 = ZERO */
+ -ONE_TOKEN, 6, /* 2 = ONE */
+ 8, 12, /* 3 = LOW_VAL */
+ -TWO_TOKEN, 10, /* 4 = TWO */
+ -THREE_TOKEN, -FOUR_TOKEN, /* 5 = THREE */
+ 14, 16, /* 6 = HIGH_LOW */
+ -DCT_VAL_CATEGORY1, -DCT_VAL_CATEGORY2, /* 7 = CAT_ONE */
+ 18, 20, /* 8 = CAT_THREEFOUR */
+ -DCT_VAL_CATEGORY3, -DCT_VAL_CATEGORY4, /* 9 = CAT_THREE */
+ -DCT_VAL_CATEGORY5, -DCT_VAL_CATEGORY6 /* 10 = CAT_FIVE */
+};
+
+/* vp8_coef_encodings generated with:
+ vp8_tokens_from_tree(vp8_coef_encodings, vp8_coef_tree);
+*/
+vp8_token vp8_coef_encodings[MAX_ENTROPY_TOKENS] =
+{
+ {2, 2},
+ {6, 3},
+ {28, 5},
+ {58, 6},
+ {59, 6},
+ {60, 6},
+ {61, 6},
+ {124, 7},
+ {125, 7},
+ {126, 7},
+ {127, 7},
+ {0, 1}
+};
+
+/* Trees for extra bits. Probabilities are constant and
+ do not depend on previously encoded bits */
+
+static const vp8_prob Pcat1[] = { 159};
+static const vp8_prob Pcat2[] = { 165, 145};
+static const vp8_prob Pcat3[] = { 173, 148, 140};
+static const vp8_prob Pcat4[] = { 176, 155, 140, 135};
+static const vp8_prob Pcat5[] = { 180, 157, 141, 134, 130};
+static const vp8_prob Pcat6[] =
+{ 254, 254, 243, 230, 196, 177, 153, 140, 133, 130, 129};
+
+
+/* tree index tables generated with:
+
+ void init_bit_tree(vp8_tree_index *p, int n)
+ {
+ int i = 0;
+
+ while (++i < n)
+ {
+ p[0] = p[1] = i << 1;
+ p += 2;
+ }
+
+ p[0] = p[1] = 0;
+ }
+
+ void init_bit_trees()
+ {
+ init_bit_tree(cat1, 1);
+ init_bit_tree(cat2, 2);
+ init_bit_tree(cat3, 3);
+ init_bit_tree(cat4, 4);
+ init_bit_tree(cat5, 5);
+ init_bit_tree(cat6, 11);
+ }
+*/
+
+static const vp8_tree_index cat1[2] = { 0, 0 };
+static const vp8_tree_index cat2[4] = { 2, 2, 0, 0 };
+static const vp8_tree_index cat3[6] = { 2, 2, 4, 4, 0, 0 };
+static const vp8_tree_index cat4[8] = { 2, 2, 4, 4, 6, 6, 0, 0 };
+static const vp8_tree_index cat5[10] = { 2, 2, 4, 4, 6, 6, 8, 8, 0, 0 };
+static const vp8_tree_index cat6[22] = { 2, 2, 4, 4, 6, 6, 8, 8, 10, 10, 12, 12,
+ 14, 14, 16, 16, 18, 18, 20, 20, 0, 0 };
+
+const vp8_extra_bit_struct vp8_extra_bits[12] =
+{
+ { 0, 0, 0, 0},
+ { 0, 0, 0, 1},
+ { 0, 0, 0, 2},
+ { 0, 0, 0, 3},
+ { 0, 0, 0, 4},
+ { cat1, Pcat1, 1, 5},
+ { cat2, Pcat2, 2, 7},
+ { cat3, Pcat3, 3, 11},
+ { cat4, Pcat4, 4, 19},
+ { cat5, Pcat5, 5, 35},
+ { cat6, Pcat6, 11, 67},
+ { 0, 0, 0, 0}
+};
+
+#include "default_coef_probs.h"
+
+void vp8_default_coef_probs(VP8_COMMON *pc)
+{
+ vpx_memcpy(pc->fc.coef_probs, default_coef_probs,
+ sizeof(default_coef_probs));
+}
+
diff --git a/libvpx/vp8/common/entropy.h b/libvpx/vp8/common/entropy.h
new file mode 100644
index 0000000..5389bc1
--- /dev/null
+++ b/libvpx/vp8/common/entropy.h
@@ -0,0 +1,101 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_ENTROPY_H
+#define __INC_ENTROPY_H
+
+#include "treecoder.h"
+#include "blockd.h"
+
+/* Coefficient token alphabet */
+
+#define ZERO_TOKEN 0 /* 0 Extra Bits 0+0 */
+#define ONE_TOKEN 1 /* 1 Extra Bits 0+1 */
+#define TWO_TOKEN 2 /* 2 Extra Bits 0+1 */
+#define THREE_TOKEN 3 /* 3 Extra Bits 0+1 */
+#define FOUR_TOKEN 4 /* 4 Extra Bits 0+1 */
+#define DCT_VAL_CATEGORY1 5 /* 5-6 Extra Bits 1+1 */
+#define DCT_VAL_CATEGORY2 6 /* 7-10 Extra Bits 2+1 */
+#define DCT_VAL_CATEGORY3 7 /* 11-18 Extra Bits 3+1 */
+#define DCT_VAL_CATEGORY4 8 /* 19-34 Extra Bits 4+1 */
+#define DCT_VAL_CATEGORY5 9 /* 35-66 Extra Bits 5+1 */
+#define DCT_VAL_CATEGORY6 10 /* 67+ Extra Bits 11+1 */
+#define DCT_EOB_TOKEN 11 /* EOB Extra Bits 0+0 */
+
+#define MAX_ENTROPY_TOKENS 12
+#define ENTROPY_NODES 11
+
+extern const vp8_tree_index vp8_coef_tree[];
+
+extern const struct vp8_token_struct vp8_coef_encodings[MAX_ENTROPY_TOKENS];
+
+typedef struct
+{
+ vp8_tree_p tree;
+ const vp8_prob *prob;
+ int Len;
+ int base_val;
+} vp8_extra_bit_struct;
+
+extern const vp8_extra_bit_struct vp8_extra_bits[12]; /* indexed by token value */
+
+#define PROB_UPDATE_BASELINE_COST 7
+
+#define MAX_PROB 255
+#define DCT_MAX_VALUE 2048
+
+
+/* Coefficients are predicted via a 3-dimensional probability table. */
+
+/* Outside dimension. 0 = Y no DC, 1 = Y2, 2 = UV, 3 = Y with DC */
+
+#define BLOCK_TYPES 4
+
+/* Middle dimension is a coarsening of the coefficient's
+ position within the 4x4 DCT. */
+
+#define COEF_BANDS 8
+extern DECLARE_ALIGNED(16, const unsigned char, vp8_coef_bands[16]);
+
+/* Inside dimension is 3-valued measure of nearby complexity, that is,
+ the extent to which nearby coefficients are nonzero. For the first
+ coefficient (DC, unless block type is 0), we look at the (already encoded)
+ blocks above and to the left of the current block. The context index is
+ then the number (0,1,or 2) of these blocks having nonzero coefficients.
+ After decoding a coefficient, the measure is roughly the size of the
+ most recently decoded coefficient (0 for 0, 1 for 1, 2 for >1).
+ Note that the intuitive meaning of this measure changes as coefficients
+ are decoded, e.g., prior to the first token, a zero means that my neighbors
+ are empty while, after the first token, because of the use of end-of-block,
+ a zero means we just decoded a zero and hence guarantees that a non-zero
+ coefficient will appear later in this block. However, this shift
+ in meaning is perfectly OK because our context depends also on the
+ coefficient band (and since zigzag positions 0, 1, and 2 are in
+ distinct bands). */
+
+/*# define DC_TOKEN_CONTEXTS 3*/ /* 00, 0!0, !0!0 */
+# define PREV_COEF_CONTEXTS 3
+
+extern DECLARE_ALIGNED(16, const unsigned char, vp8_prev_token_class[MAX_ENTROPY_TOKENS]);
+
+extern const vp8_prob vp8_coef_update_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES];
+
+
+struct VP8Common;
+void vp8_default_coef_probs(struct VP8Common *);
+
+extern DECLARE_ALIGNED(16, const int, vp8_default_zig_zag1d[16]);
+extern DECLARE_ALIGNED(16, const short, vp8_default_inv_zig_zag[16]);
+extern DECLARE_ALIGNED(16, const short, vp8_default_zig_zag_mask[16]);
+extern const int vp8_mb_feature_data_bits[MB_LVL_MAX];
+
+void vp8_coef_tree_initialize(void);
+#endif
diff --git a/libvpx/vp8/common/entropymode.c b/libvpx/vp8/common/entropymode.c
new file mode 100644
index 0000000..091e4c7
--- /dev/null
+++ b/libvpx/vp8/common/entropymode.c
@@ -0,0 +1,171 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#define USE_PREBUILT_TABLES
+
+#include "entropymode.h"
+#include "entropy.h"
+#include "vpx_mem/vpx_mem.h"
+
+#include "vp8_entropymodedata.h"
+
+int vp8_mv_cont(const int_mv *l, const int_mv *a)
+{
+ int lez = (l->as_int == 0);
+ int aez = (a->as_int == 0);
+ int lea = (l->as_int == a->as_int);
+
+ if (lea && lez)
+ return SUBMVREF_LEFT_ABOVE_ZED;
+
+ if (lea)
+ return SUBMVREF_LEFT_ABOVE_SAME;
+
+ if (aez)
+ return SUBMVREF_ABOVE_ZED;
+
+ if (lez)
+ return SUBMVREF_LEFT_ZED;
+
+ return SUBMVREF_NORMAL;
+}
+
+static const vp8_prob sub_mv_ref_prob [VP8_SUBMVREFS-1] = { 180, 162, 25};
+
+const vp8_prob vp8_sub_mv_ref_prob2 [SUBMVREF_COUNT][VP8_SUBMVREFS-1] =
+{
+ { 147, 136, 18 },
+ { 106, 145, 1 },
+ { 179, 121, 1 },
+ { 223, 1 , 34 },
+ { 208, 1 , 1 }
+};
+
+
+
+const vp8_mbsplit vp8_mbsplits [VP8_NUMMBSPLITS] =
+{
+ {
+ 0, 0, 0, 0,
+ 0, 0, 0, 0,
+ 1, 1, 1, 1,
+ 1, 1, 1, 1,
+ },
+ {
+ 0, 0, 1, 1,
+ 0, 0, 1, 1,
+ 0, 0, 1, 1,
+ 0, 0, 1, 1,
+ },
+ {
+ 0, 0, 1, 1,
+ 0, 0, 1, 1,
+ 2, 2, 3, 3,
+ 2, 2, 3, 3,
+ },
+ {
+ 0, 1, 2, 3,
+ 4, 5, 6, 7,
+ 8, 9, 10, 11,
+ 12, 13, 14, 15,
+ }
+};
+
+const int vp8_mbsplit_count [VP8_NUMMBSPLITS] = { 2, 2, 4, 16};
+
+const vp8_prob vp8_mbsplit_probs [VP8_NUMMBSPLITS-1] = { 110, 111, 150};
+
+
+/* Array indices are identical to previously-existing INTRAMODECONTEXTNODES. */
+
+const vp8_tree_index vp8_bmode_tree[18] = /* INTRAMODECONTEXTNODE value */
+{
+ -B_DC_PRED, 2, /* 0 = DC_NODE */
+ -B_TM_PRED, 4, /* 1 = TM_NODE */
+ -B_VE_PRED, 6, /* 2 = VE_NODE */
+ 8, 12, /* 3 = COM_NODE */
+ -B_HE_PRED, 10, /* 4 = HE_NODE */
+ -B_RD_PRED, -B_VR_PRED, /* 5 = RD_NODE */
+ -B_LD_PRED, 14, /* 6 = LD_NODE */
+ -B_VL_PRED, 16, /* 7 = VL_NODE */
+ -B_HD_PRED, -B_HU_PRED /* 8 = HD_NODE */
+};
+
+/* Again, these trees use the same probability indices as their
+ explicitly-programmed predecessors. */
+
+const vp8_tree_index vp8_ymode_tree[8] =
+{
+ -DC_PRED, 2,
+ 4, 6,
+ -V_PRED, -H_PRED,
+ -TM_PRED, -B_PRED
+};
+
+const vp8_tree_index vp8_kf_ymode_tree[8] =
+{
+ -B_PRED, 2,
+ 4, 6,
+ -DC_PRED, -V_PRED,
+ -H_PRED, -TM_PRED
+};
+
+const vp8_tree_index vp8_uv_mode_tree[6] =
+{
+ -DC_PRED, 2,
+ -V_PRED, 4,
+ -H_PRED, -TM_PRED
+};
+
+const vp8_tree_index vp8_mbsplit_tree[6] =
+{
+ -3, 2,
+ -2, 4,
+ -0, -1
+};
+
+const vp8_tree_index vp8_mv_ref_tree[8] =
+{
+ -ZEROMV, 2,
+ -NEARESTMV, 4,
+ -NEARMV, 6,
+ -NEWMV, -SPLITMV
+};
+
+const vp8_tree_index vp8_sub_mv_ref_tree[6] =
+{
+ -LEFT4X4, 2,
+ -ABOVE4X4, 4,
+ -ZERO4X4, -NEW4X4
+};
+
+const vp8_tree_index vp8_small_mvtree [14] =
+{
+ 2, 8,
+ 4, 6,
+ -0, -1,
+ -2, -3,
+ 10, 12,
+ -4, -5,
+ -6, -7
+};
+
+void vp8_init_mbmode_probs(VP8_COMMON *x)
+{
+ vpx_memcpy(x->fc.ymode_prob, vp8_ymode_prob, sizeof(vp8_ymode_prob));
+ vpx_memcpy(x->fc.uv_mode_prob, vp8_uv_mode_prob, sizeof(vp8_uv_mode_prob));
+ vpx_memcpy(x->fc.sub_mv_ref_prob, sub_mv_ref_prob, sizeof(sub_mv_ref_prob));
+}
+
+void vp8_default_bmode_probs(vp8_prob p [VP8_BINTRAMODES-1])
+{
+ vpx_memcpy(p, vp8_bmode_prob, sizeof(vp8_bmode_prob));
+}
+
diff --git a/libvpx/vp8/common/entropymode.h b/libvpx/vp8/common/entropymode.h
new file mode 100644
index 0000000..1df0f64
--- /dev/null
+++ b/libvpx/vp8/common/entropymode.h
@@ -0,0 +1,80 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_ENTROPYMODE_H
+#define __INC_ENTROPYMODE_H
+
+#include "onyxc_int.h"
+#include "treecoder.h"
+
+typedef enum
+{
+ SUBMVREF_NORMAL,
+ SUBMVREF_LEFT_ZED,
+ SUBMVREF_ABOVE_ZED,
+ SUBMVREF_LEFT_ABOVE_SAME,
+ SUBMVREF_LEFT_ABOVE_ZED
+} sumvfref_t;
+
+typedef int vp8_mbsplit[16];
+
+#define VP8_NUMMBSPLITS 4
+
+extern const vp8_mbsplit vp8_mbsplits [VP8_NUMMBSPLITS];
+
+extern const int vp8_mbsplit_count [VP8_NUMMBSPLITS]; /* # of subsets */
+
+extern const vp8_prob vp8_mbsplit_probs [VP8_NUMMBSPLITS-1];
+
+extern int vp8_mv_cont(const int_mv *l, const int_mv *a);
+#define SUBMVREF_COUNT 5
+extern const vp8_prob vp8_sub_mv_ref_prob2 [SUBMVREF_COUNT][VP8_SUBMVREFS-1];
+
+
+extern const unsigned int vp8_kf_default_bmode_counts [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES];
+
+
+extern const vp8_tree_index vp8_bmode_tree[];
+
+extern const vp8_tree_index vp8_ymode_tree[];
+extern const vp8_tree_index vp8_kf_ymode_tree[];
+extern const vp8_tree_index vp8_uv_mode_tree[];
+
+extern const vp8_tree_index vp8_mbsplit_tree[];
+extern const vp8_tree_index vp8_mv_ref_tree[];
+extern const vp8_tree_index vp8_sub_mv_ref_tree[];
+
+extern const struct vp8_token_struct vp8_bmode_encodings[VP8_BINTRAMODES];
+extern const struct vp8_token_struct vp8_ymode_encodings[VP8_YMODES];
+extern const struct vp8_token_struct vp8_kf_ymode_encodings[VP8_YMODES];
+extern const struct vp8_token_struct vp8_uv_mode_encodings[VP8_UV_MODES];
+extern const struct vp8_token_struct vp8_mbsplit_encodings[VP8_NUMMBSPLITS];
+
+/* Inter mode values do not start at zero */
+
+extern const struct vp8_token_struct vp8_mv_ref_encoding_array[VP8_MVREFS];
+extern const struct vp8_token_struct vp8_sub_mv_ref_encoding_array[VP8_SUBMVREFS];
+
+extern const vp8_tree_index vp8_small_mvtree[];
+
+extern const struct vp8_token_struct vp8_small_mvencodings[8];
+
+/* Key frame default mode probs */
+extern const vp8_prob vp8_kf_bmode_prob[VP8_BINTRAMODES][VP8_BINTRAMODES]
+[VP8_BINTRAMODES-1];
+extern const vp8_prob vp8_kf_uv_mode_prob[VP8_UV_MODES-1];
+extern const vp8_prob vp8_kf_ymode_prob[VP8_YMODES-1];
+
+void vp8_init_mbmode_probs(VP8_COMMON *x);
+void vp8_default_bmode_probs(vp8_prob dest [VP8_BINTRAMODES-1]);
+void vp8_kf_default_bmode_probs(vp8_prob dest [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES-1]);
+
+#endif
diff --git a/vp8/common/entropymv.c b/libvpx/vp8/common/entropymv.c
similarity index 100%
rename from vp8/common/entropymv.c
rename to libvpx/vp8/common/entropymv.c
diff --git a/libvpx/vp8/common/entropymv.h b/libvpx/vp8/common/entropymv.h
new file mode 100644
index 0000000..2db1e38
--- /dev/null
+++ b/libvpx/vp8/common/entropymv.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_ENTROPYMV_H
+#define __INC_ENTROPYMV_H
+
+#include "treecoder.h"
+
+enum
+{
+ mv_max = 1023, /* max absolute value of a MV component */
+ MVvals = (2 * mv_max) + 1, /* # possible values "" */
+ mvfp_max = 255, /* max absolute value of a full pixel MV component */
+ MVfpvals = (2 * mvfp_max) +1, /* # possible full pixel MV values */
+
+ mvlong_width = 10, /* Large MVs have 9 bit magnitudes */
+ mvnum_short = 8, /* magnitudes 0 through 7 */
+
+ /* probability offsets for coding each MV component */
+
+ mvpis_short = 0, /* short (<= 7) vs long (>= 8) */
+ MVPsign, /* sign for non-zero */
+ MVPshort, /* 8 short values = 7-position tree */
+
+ MVPbits = MVPshort + mvnum_short - 1, /* mvlong_width long value bits */
+ MVPcount = MVPbits + mvlong_width /* (with independent probabilities) */
+};
+
+typedef struct mv_context
+{
+ vp8_prob prob[MVPcount]; /* often come in row, col pairs */
+} MV_CONTEXT;
+
+extern const MV_CONTEXT vp8_mv_update_probs[2], vp8_default_mv_context[2];
+
+#endif
diff --git a/libvpx/vp8/common/extend.c b/libvpx/vp8/common/extend.c
new file mode 100644
index 0000000..c9bdd21
--- /dev/null
+++ b/libvpx/vp8/common/extend.c
@@ -0,0 +1,188 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "extend.h"
+#include "vpx_mem/vpx_mem.h"
+
+
+static void copy_and_extend_plane
+(
+ unsigned char *s, /* source */
+ int sp, /* source pitch */
+ unsigned char *d, /* destination */
+ int dp, /* destination pitch */
+ int h, /* height */
+ int w, /* width */
+ int et, /* extend top border */
+ int el, /* extend left border */
+ int eb, /* extend bottom border */
+ int er /* extend right border */
+)
+{
+ int i;
+ unsigned char *src_ptr1, *src_ptr2;
+ unsigned char *dest_ptr1, *dest_ptr2;
+ int linesize;
+
+ /* copy the left and right most columns out */
+ src_ptr1 = s;
+ src_ptr2 = s + w - 1;
+ dest_ptr1 = d - el;
+ dest_ptr2 = d + w;
+
+ for (i = 0; i < h; i++)
+ {
+ vpx_memset(dest_ptr1, src_ptr1[0], el);
+ vpx_memcpy(dest_ptr1 + el, src_ptr1, w);
+ vpx_memset(dest_ptr2, src_ptr2[0], er);
+ src_ptr1 += sp;
+ src_ptr2 += sp;
+ dest_ptr1 += dp;
+ dest_ptr2 += dp;
+ }
+
+ /* Now copy the top and bottom lines into each line of the respective
+ * borders
+ */
+ src_ptr1 = d - el;
+ src_ptr2 = d + dp * (h - 1) - el;
+ dest_ptr1 = d + dp * (-et) - el;
+ dest_ptr2 = d + dp * (h) - el;
+ linesize = el + er + w;
+
+ for (i = 0; i < et; i++)
+ {
+ vpx_memcpy(dest_ptr1, src_ptr1, linesize);
+ dest_ptr1 += dp;
+ }
+
+ for (i = 0; i < eb; i++)
+ {
+ vpx_memcpy(dest_ptr2, src_ptr2, linesize);
+ dest_ptr2 += dp;
+ }
+}
+
+
+void vp8_copy_and_extend_frame(YV12_BUFFER_CONFIG *src,
+ YV12_BUFFER_CONFIG *dst)
+{
+ int et = dst->border;
+ int el = dst->border;
+ int eb = dst->border + dst->y_height - src->y_height;
+ int er = dst->border + dst->y_width - src->y_width;
+
+ copy_and_extend_plane(src->y_buffer, src->y_stride,
+ dst->y_buffer, dst->y_stride,
+ src->y_height, src->y_width,
+ et, el, eb, er);
+
+ et = dst->border >> 1;
+ el = dst->border >> 1;
+ eb = (dst->border >> 1) + dst->uv_height - src->uv_height;
+ er = (dst->border >> 1) + dst->uv_width - src->uv_width;
+
+ copy_and_extend_plane(src->u_buffer, src->uv_stride,
+ dst->u_buffer, dst->uv_stride,
+ src->uv_height, src->uv_width,
+ et, el, eb, er);
+
+ copy_and_extend_plane(src->v_buffer, src->uv_stride,
+ dst->v_buffer, dst->uv_stride,
+ src->uv_height, src->uv_width,
+ et, el, eb, er);
+}
+
+
+void vp8_copy_and_extend_frame_with_rect(YV12_BUFFER_CONFIG *src,
+ YV12_BUFFER_CONFIG *dst,
+ int srcy, int srcx,
+ int srch, int srcw)
+{
+ int et = dst->border;
+ int el = dst->border;
+ int eb = dst->border + dst->y_height - src->y_height;
+ int er = dst->border + dst->y_width - src->y_width;
+ int src_y_offset = srcy * src->y_stride + srcx;
+ int dst_y_offset = srcy * dst->y_stride + srcx;
+ int src_uv_offset = ((srcy * src->uv_stride) >> 1) + (srcx >> 1);
+ int dst_uv_offset = ((srcy * dst->uv_stride) >> 1) + (srcx >> 1);
+
+ /* If the side is not touching the bounder then don't extend. */
+ if (srcy)
+ et = 0;
+ if (srcx)
+ el = 0;
+ if (srcy + srch != src->y_height)
+ eb = 0;
+ if (srcx + srcw != src->y_width)
+ er = 0;
+
+ copy_and_extend_plane(src->y_buffer + src_y_offset,
+ src->y_stride,
+ dst->y_buffer + dst_y_offset,
+ dst->y_stride,
+ srch, srcw,
+ et, el, eb, er);
+
+ et = (et + 1) >> 1;
+ el = (el + 1) >> 1;
+ eb = (eb + 1) >> 1;
+ er = (er + 1) >> 1;
+ srch = (srch + 1) >> 1;
+ srcw = (srcw + 1) >> 1;
+
+ copy_and_extend_plane(src->u_buffer + src_uv_offset,
+ src->uv_stride,
+ dst->u_buffer + dst_uv_offset,
+ dst->uv_stride,
+ srch, srcw,
+ et, el, eb, er);
+
+ copy_and_extend_plane(src->v_buffer + src_uv_offset,
+ src->uv_stride,
+ dst->v_buffer + dst_uv_offset,
+ dst->uv_stride,
+ srch, srcw,
+ et, el, eb, er);
+}
+
+
+/* note the extension is only for the last row, for intra prediction purpose */
+void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf,
+ unsigned char *YPtr,
+ unsigned char *UPtr,
+ unsigned char *VPtr)
+{
+ int i;
+
+ YPtr += ybf->y_stride * 14;
+ UPtr += ybf->uv_stride * 6;
+ VPtr += ybf->uv_stride * 6;
+
+ for (i = 0; i < 4; i++)
+ {
+ YPtr[i] = YPtr[-1];
+ UPtr[i] = UPtr[-1];
+ VPtr[i] = VPtr[-1];
+ }
+
+ YPtr += ybf->y_stride;
+ UPtr += ybf->uv_stride;
+ VPtr += ybf->uv_stride;
+
+ for (i = 0; i < 4; i++)
+ {
+ YPtr[i] = YPtr[-1];
+ UPtr[i] = UPtr[-1];
+ VPtr[i] = VPtr[-1];
+ }
+}
diff --git a/libvpx/vp8/common/extend.h b/libvpx/vp8/common/extend.h
new file mode 100644
index 0000000..74a0b17
--- /dev/null
+++ b/libvpx/vp8/common/extend.h
@@ -0,0 +1,25 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_EXTEND_H
+#define __INC_EXTEND_H
+
+#include "vpx_scale/yv12config.h"
+
+void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf, unsigned char *YPtr, unsigned char *UPtr, unsigned char *VPtr);
+void vp8_copy_and_extend_frame(YV12_BUFFER_CONFIG *src,
+ YV12_BUFFER_CONFIG *dst);
+void vp8_copy_and_extend_frame_with_rect(YV12_BUFFER_CONFIG *src,
+ YV12_BUFFER_CONFIG *dst,
+ int srcy, int srcx,
+ int srch, int srcw);
+
+#endif
diff --git a/libvpx/vp8/common/filter.c b/libvpx/vp8/common/filter.c
new file mode 100644
index 0000000..1901ea3
--- /dev/null
+++ b/libvpx/vp8/common/filter.c
@@ -0,0 +1,494 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdlib.h>
+#include "filter.h"
+#include "vpx_ports/mem.h"
+
+DECLARE_ALIGNED(16, const short, vp8_bilinear_filters[8][2]) =
+{
+ { 128, 0 },
+ { 112, 16 },
+ { 96, 32 },
+ { 80, 48 },
+ { 64, 64 },
+ { 48, 80 },
+ { 32, 96 },
+ { 16, 112 }
+};
+
+DECLARE_ALIGNED(16, const short, vp8_sub_pel_filters[8][6]) =
+{
+
+ { 0, 0, 128, 0, 0, 0 }, /* note that 1/8 pel positions are just as per alpha -0.5 bicubic */
+ { 0, -6, 123, 12, -1, 0 },
+ { 2, -11, 108, 36, -8, 1 }, /* New 1/4 pel 6 tap filter */
+ { 0, -9, 93, 50, -6, 0 },
+ { 3, -16, 77, 77, -16, 3 }, /* New 1/2 pel 6 tap filter */
+ { 0, -6, 50, 93, -9, 0 },
+ { 1, -8, 36, 108, -11, 2 }, /* New 1/4 pel 6 tap filter */
+ { 0, -1, 12, 123, -6, 0 },
+};
+
+static void filter_block2d_first_pass
+(
+ unsigned char *src_ptr,
+ int *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+)
+{
+ unsigned int i, j;
+ int Temp;
+
+ for (i = 0; i < output_height; i++)
+ {
+ for (j = 0; j < output_width; j++)
+ {
+ Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
+ ((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
+ ((int)src_ptr[0] * vp8_filter[2]) +
+ ((int)src_ptr[pixel_step] * vp8_filter[3]) +
+ ((int)src_ptr[2*pixel_step] * vp8_filter[4]) +
+ ((int)src_ptr[3*pixel_step] * vp8_filter[5]) +
+ (VP8_FILTER_WEIGHT >> 1); /* Rounding */
+
+ /* Normalize back to 0-255 */
+ Temp = Temp >> VP8_FILTER_SHIFT;
+
+ if (Temp < 0)
+ Temp = 0;
+ else if (Temp > 255)
+ Temp = 255;
+
+ output_ptr[j] = Temp;
+ src_ptr++;
+ }
+
+ /* Next row... */
+ src_ptr += src_pixels_per_line - output_width;
+ output_ptr += output_width;
+ }
+}
+
+static void filter_block2d_second_pass
+(
+ int *src_ptr,
+ unsigned char *output_ptr,
+ int output_pitch,
+ unsigned int src_pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+)
+{
+ unsigned int i, j;
+ int Temp;
+
+ for (i = 0; i < output_height; i++)
+ {
+ for (j = 0; j < output_width; j++)
+ {
+ /* Apply filter */
+ Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
+ ((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
+ ((int)src_ptr[0] * vp8_filter[2]) +
+ ((int)src_ptr[pixel_step] * vp8_filter[3]) +
+ ((int)src_ptr[2*pixel_step] * vp8_filter[4]) +
+ ((int)src_ptr[3*pixel_step] * vp8_filter[5]) +
+ (VP8_FILTER_WEIGHT >> 1); /* Rounding */
+
+ /* Normalize back to 0-255 */
+ Temp = Temp >> VP8_FILTER_SHIFT;
+
+ if (Temp < 0)
+ Temp = 0;
+ else if (Temp > 255)
+ Temp = 255;
+
+ output_ptr[j] = (unsigned char)Temp;
+ src_ptr++;
+ }
+
+ /* Start next row */
+ src_ptr += src_pixels_per_line - output_width;
+ output_ptr += output_pitch;
+ }
+}
+
+
+static void filter_block2d
+(
+ unsigned char *src_ptr,
+ unsigned char *output_ptr,
+ unsigned int src_pixels_per_line,
+ int output_pitch,
+ const short *HFilter,
+ const short *VFilter
+)
+{
+ int FData[9*4]; /* Temp data buffer used in filtering */
+
+ /* First filter 1-D horizontally... */
+ filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 9, 4, HFilter);
+
+ /* then filter verticaly... */
+ filter_block2d_second_pass(FData + 8, output_ptr, output_pitch, 4, 4, 4, 4, VFilter);
+}
+
+
+void vp8_sixtap_predict4x4_c
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
+ VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
+
+ filter_block2d(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter);
+}
+void vp8_sixtap_predict8x8_c
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+ int FData[13*16]; /* Temp data buffer used in filtering */
+
+ HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
+ VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
+
+ /* First filter 1-D horizontally... */
+ filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 13, 8, HFilter);
+
+
+ /* then filter verticaly... */
+ filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 8, 8, VFilter);
+
+}
+
+void vp8_sixtap_predict8x4_c
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+ int FData[13*16]; /* Temp data buffer used in filtering */
+
+ HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
+ VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
+
+ /* First filter 1-D horizontally... */
+ filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 9, 8, HFilter);
+
+
+ /* then filter verticaly... */
+ filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 4, 8, VFilter);
+
+}
+
+void vp8_sixtap_predict16x16_c
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+ int FData[21*24]; /* Temp data buffer used in filtering */
+
+
+ HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
+ VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
+
+ /* First filter 1-D horizontally... */
+ filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 21, 16, HFilter);
+
+ /* then filter verticaly... */
+ filter_block2d_second_pass(FData + 32, dst_ptr, dst_pitch, 16, 16, 16, 16, VFilter);
+
+}
+
+
+/****************************************************************************
+ *
+ * ROUTINE : filter_block2d_bil_first_pass
+ *
+ * INPUTS : UINT8 *src_ptr : Pointer to source block.
+ * UINT32 src_stride : Stride of source block.
+ * UINT32 height : Block height.
+ * UINT32 width : Block width.
+ * INT32 *vp8_filter : Array of 2 bi-linear filter taps.
+ *
+ * OUTPUTS : INT32 *dst_ptr : Pointer to filtered block.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block
+ * in the horizontal direction to produce the filtered output
+ * block. Used to implement first-pass of 2-D separable filter.
+ *
+ * SPECIAL NOTES : Produces INT32 output to retain precision for next pass.
+ * Two filter taps should sum to VP8_FILTER_WEIGHT.
+ *
+ ****************************************************************************/
+static void filter_block2d_bil_first_pass
+(
+ unsigned char *src_ptr,
+ unsigned short *dst_ptr,
+ unsigned int src_stride,
+ unsigned int height,
+ unsigned int width,
+ const short *vp8_filter
+)
+{
+ unsigned int i, j;
+
+ for (i = 0; i < height; i++)
+ {
+ for (j = 0; j < width; j++)
+ {
+ /* Apply bilinear filter */
+ dst_ptr[j] = (((int)src_ptr[0] * vp8_filter[0]) +
+ ((int)src_ptr[1] * vp8_filter[1]) +
+ (VP8_FILTER_WEIGHT / 2)) >> VP8_FILTER_SHIFT;
+ src_ptr++;
+ }
+
+ /* Next row... */
+ src_ptr += src_stride - width;
+ dst_ptr += width;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : filter_block2d_bil_second_pass
+ *
+ * INPUTS : INT32 *src_ptr : Pointer to source block.
+ * UINT32 dst_pitch : Destination block pitch.
+ * UINT32 height : Block height.
+ * UINT32 width : Block width.
+ * INT32 *vp8_filter : Array of 2 bi-linear filter taps.
+ *
+ * OUTPUTS : UINT16 *dst_ptr : Pointer to filtered block.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block
+ * in the vertical direction to produce the filtered output
+ * block. Used to implement second-pass of 2-D separable filter.
+ *
+ * SPECIAL NOTES : Requires 32-bit input as produced by filter_block2d_bil_first_pass.
+ * Two filter taps should sum to VP8_FILTER_WEIGHT.
+ *
+ ****************************************************************************/
+static void filter_block2d_bil_second_pass
+(
+ unsigned short *src_ptr,
+ unsigned char *dst_ptr,
+ int dst_pitch,
+ unsigned int height,
+ unsigned int width,
+ const short *vp8_filter
+)
+{
+ unsigned int i, j;
+ int Temp;
+
+ for (i = 0; i < height; i++)
+ {
+ for (j = 0; j < width; j++)
+ {
+ /* Apply filter */
+ Temp = ((int)src_ptr[0] * vp8_filter[0]) +
+ ((int)src_ptr[width] * vp8_filter[1]) +
+ (VP8_FILTER_WEIGHT / 2);
+ dst_ptr[j] = (unsigned int)(Temp >> VP8_FILTER_SHIFT);
+ src_ptr++;
+ }
+
+ /* Next row... */
+ dst_ptr += dst_pitch;
+ }
+}
+
+
+/****************************************************************************
+ *
+ * ROUTINE : filter_block2d_bil
+ *
+ * INPUTS : UINT8 *src_ptr : Pointer to source block.
+ * UINT32 src_pitch : Stride of source block.
+ * UINT32 dst_pitch : Stride of destination block.
+ * INT32 *HFilter : Array of 2 horizontal filter taps.
+ * INT32 *VFilter : Array of 2 vertical filter taps.
+ * INT32 Width : Block width
+ * INT32 Height : Block height
+ *
+ * OUTPUTS : UINT16 *dst_ptr : Pointer to filtered block.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 2-D filters an input block by applying a 2-tap
+ * bi-linear filter horizontally followed by a 2-tap
+ * bi-linear filter vertically on the result.
+ *
+ * SPECIAL NOTES : The largest block size can be handled here is 16x16
+ *
+ ****************************************************************************/
+static void filter_block2d_bil
+(
+ unsigned char *src_ptr,
+ unsigned char *dst_ptr,
+ unsigned int src_pitch,
+ unsigned int dst_pitch,
+ const short *HFilter,
+ const short *VFilter,
+ int Width,
+ int Height
+)
+{
+
+ unsigned short FData[17*16]; /* Temp data buffer used in filtering */
+
+ /* First filter 1-D horizontally... */
+ filter_block2d_bil_first_pass(src_ptr, FData, src_pitch, Height + 1, Width, HFilter);
+
+ /* then 1-D vertically... */
+ filter_block2d_bil_second_pass(FData, dst_ptr, dst_pitch, Height, Width, VFilter);
+}
+
+
+void vp8_bilinear_predict4x4_c
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+#if 0
+ {
+ int i;
+ unsigned char temp1[16];
+ unsigned char temp2[16];
+
+ bilinear_predict4x4_mmx(src_ptr, src_pixels_per_line, xoffset, yoffset, temp1, 4);
+ filter_block2d_bil(src_ptr, temp2, src_pixels_per_line, 4, HFilter, VFilter, 4, 4);
+
+ for (i = 0; i < 16; i++)
+ {
+ if (temp1[i] != temp2[i])
+ {
+ bilinear_predict4x4_mmx(src_ptr, src_pixels_per_line, xoffset, yoffset, temp1, 4);
+ filter_block2d_bil(src_ptr, temp2, src_pixels_per_line, 4, HFilter, VFilter, 4, 4);
+ }
+ }
+ }
+#endif
+ filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 4, 4);
+
+}
+
+void vp8_bilinear_predict8x8_c
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 8);
+
+}
+
+void vp8_bilinear_predict8x4_c
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 4);
+
+}
+
+void vp8_bilinear_predict16x16_c
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ const short *HFilter;
+ const short *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 16, 16);
+}
diff --git a/libvpx/vp8/common/filter.h b/libvpx/vp8/common/filter.h
new file mode 100644
index 0000000..b7591f2
--- /dev/null
+++ b/libvpx/vp8/common/filter.h
@@ -0,0 +1,22 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef FILTER_H
+#define FILTER_H
+
+#define BLOCK_HEIGHT_WIDTH 4
+#define VP8_FILTER_WEIGHT 128
+#define VP8_FILTER_SHIFT 7
+
+extern const short vp8_bilinear_filters[8][2];
+extern const short vp8_sub_pel_filters[8][6];
+
+#endif
diff --git a/libvpx/vp8/common/findnearmv.c b/libvpx/vp8/common/findnearmv.c
new file mode 100644
index 0000000..e8ee40f
--- /dev/null
+++ b/libvpx/vp8/common/findnearmv.c
@@ -0,0 +1,193 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "findnearmv.h"
+
+const unsigned char vp8_mbsplit_offset[4][16] = {
+ { 0, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
+ { 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
+ { 0, 2, 8, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
+ { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}
+};
+
+/* Predict motion vectors using those from already-decoded nearby blocks.
+ Note that we only consider one 4x4 subblock from each candidate 16x16
+ macroblock. */
+void vp8_find_near_mvs
+(
+ MACROBLOCKD *xd,
+ const MODE_INFO *here,
+ int_mv *nearest,
+ int_mv *nearby,
+ int_mv *best_mv,
+ int cnt[4],
+ int refframe,
+ int *ref_frame_sign_bias
+)
+{
+ const MODE_INFO *above = here - xd->mode_info_stride;
+ const MODE_INFO *left = here - 1;
+ const MODE_INFO *aboveleft = above - 1;
+ int_mv near_mvs[4];
+ int_mv *mv = near_mvs;
+ int *cntx = cnt;
+ enum {CNT_INTRA, CNT_NEAREST, CNT_NEAR, CNT_SPLITMV};
+
+ /* Zero accumulators */
+ mv[0].as_int = mv[1].as_int = mv[2].as_int = 0;
+ cnt[0] = cnt[1] = cnt[2] = cnt[3] = 0;
+
+ /* Process above */
+ if (above->mbmi.ref_frame != INTRA_FRAME)
+ {
+ if (above->mbmi.mv.as_int)
+ {
+ (++mv)->as_int = above->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[above->mbmi.ref_frame], refframe, mv, ref_frame_sign_bias);
+ ++cntx;
+ }
+
+ *cntx += 2;
+ }
+
+ /* Process left */
+ if (left->mbmi.ref_frame != INTRA_FRAME)
+ {
+ if (left->mbmi.mv.as_int)
+ {
+ int_mv this_mv;
+
+ this_mv.as_int = left->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[left->mbmi.ref_frame], refframe, &this_mv, ref_frame_sign_bias);
+
+ if (this_mv.as_int != mv->as_int)
+ {
+ (++mv)->as_int = this_mv.as_int;
+ ++cntx;
+ }
+
+ *cntx += 2;
+ }
+ else
+ cnt[CNT_INTRA] += 2;
+ }
+
+ /* Process above left */
+ if (aboveleft->mbmi.ref_frame != INTRA_FRAME)
+ {
+ if (aboveleft->mbmi.mv.as_int)
+ {
+ int_mv this_mv;
+
+ this_mv.as_int = aboveleft->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[aboveleft->mbmi.ref_frame], refframe, &this_mv, ref_frame_sign_bias);
+
+ if (this_mv.as_int != mv->as_int)
+ {
+ (++mv)->as_int = this_mv.as_int;
+ ++cntx;
+ }
+
+ *cntx += 1;
+ }
+ else
+ cnt[CNT_INTRA] += 1;
+ }
+
+ /* If we have three distinct MV's ... */
+ if (cnt[CNT_SPLITMV])
+ {
+ /* See if above-left MV can be merged with NEAREST */
+ if (mv->as_int == near_mvs[CNT_NEAREST].as_int)
+ cnt[CNT_NEAREST] += 1;
+ }
+
+ cnt[CNT_SPLITMV] = ((above->mbmi.mode == SPLITMV)
+ + (left->mbmi.mode == SPLITMV)) * 2
+ + (aboveleft->mbmi.mode == SPLITMV);
+
+ /* Swap near and nearest if necessary */
+ if (cnt[CNT_NEAR] > cnt[CNT_NEAREST])
+ {
+ int tmp;
+ tmp = cnt[CNT_NEAREST];
+ cnt[CNT_NEAREST] = cnt[CNT_NEAR];
+ cnt[CNT_NEAR] = tmp;
+ tmp = near_mvs[CNT_NEAREST].as_int;
+ near_mvs[CNT_NEAREST].as_int = near_mvs[CNT_NEAR].as_int;
+ near_mvs[CNT_NEAR].as_int = tmp;
+ }
+
+ /* Use near_mvs[0] to store the "best" MV */
+ if (cnt[CNT_NEAREST] >= cnt[CNT_INTRA])
+ near_mvs[CNT_INTRA] = near_mvs[CNT_NEAREST];
+
+ /* Set up return values */
+ best_mv->as_int = near_mvs[0].as_int;
+ nearest->as_int = near_mvs[CNT_NEAREST].as_int;
+ nearby->as_int = near_mvs[CNT_NEAR].as_int;
+}
+
+
+static void invert_and_clamp_mvs(int_mv *inv, int_mv *src, MACROBLOCKD *xd)
+{
+ inv->as_mv.row = src->as_mv.row * -1;
+ inv->as_mv.col = src->as_mv.col * -1;
+ vp8_clamp_mv2(inv, xd);
+ vp8_clamp_mv2(src, xd);
+}
+
+
+int vp8_find_near_mvs_bias
+(
+ MACROBLOCKD *xd,
+ const MODE_INFO *here,
+ int_mv mode_mv_sb[2][MB_MODE_COUNT],
+ int_mv best_mv_sb[2],
+ int cnt[4],
+ int refframe,
+ int *ref_frame_sign_bias
+)
+{
+ int sign_bias = ref_frame_sign_bias[refframe];
+
+ vp8_find_near_mvs(xd,
+ here,
+ &mode_mv_sb[sign_bias][NEARESTMV],
+ &mode_mv_sb[sign_bias][NEARMV],
+ &best_mv_sb[sign_bias],
+ cnt,
+ refframe,
+ ref_frame_sign_bias);
+
+ invert_and_clamp_mvs(&mode_mv_sb[!sign_bias][NEARESTMV],
+ &mode_mv_sb[sign_bias][NEARESTMV], xd);
+ invert_and_clamp_mvs(&mode_mv_sb[!sign_bias][NEARMV],
+ &mode_mv_sb[sign_bias][NEARMV], xd);
+ invert_and_clamp_mvs(&best_mv_sb[!sign_bias],
+ &best_mv_sb[sign_bias], xd);
+
+ return sign_bias;
+}
+
+
+vp8_prob *vp8_mv_ref_probs(
+ vp8_prob p[VP8_MVREFS-1], const int near_mv_ref_ct[4]
+)
+{
+ p[0] = vp8_mode_contexts [near_mv_ref_ct[0]] [0];
+ p[1] = vp8_mode_contexts [near_mv_ref_ct[1]] [1];
+ p[2] = vp8_mode_contexts [near_mv_ref_ct[2]] [2];
+ p[3] = vp8_mode_contexts [near_mv_ref_ct[3]] [3];
+ /*p[3] = vp8_mode_contexts [near_mv_ref_ct[1] + near_mv_ref_ct[2] + near_mv_ref_ct[3]] [3];*/
+ return p;
+}
+
diff --git a/libvpx/vp8/common/findnearmv.h b/libvpx/vp8/common/findnearmv.h
new file mode 100644
index 0000000..06ef060
--- /dev/null
+++ b/libvpx/vp8/common/findnearmv.h
@@ -0,0 +1,182 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_FINDNEARMV_H
+#define __INC_FINDNEARMV_H
+
+#include "mv.h"
+#include "blockd.h"
+#include "modecont.h"
+#include "treecoder.h"
+
+
+static void mv_bias(int refmb_ref_frame_sign_bias, int refframe, int_mv *mvp,
+ const int *ref_frame_sign_bias)
+{
+ if (refmb_ref_frame_sign_bias != ref_frame_sign_bias[refframe])
+ {
+ mvp->as_mv.row *= -1;
+ mvp->as_mv.col *= -1;
+ }
+}
+
+#define LEFT_TOP_MARGIN (16 << 3)
+#define RIGHT_BOTTOM_MARGIN (16 << 3)
+static void vp8_clamp_mv2(int_mv *mv, const MACROBLOCKD *xd)
+{
+ if (mv->as_mv.col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN))
+ mv->as_mv.col = xd->mb_to_left_edge - LEFT_TOP_MARGIN;
+ else if (mv->as_mv.col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN)
+ mv->as_mv.col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN;
+
+ if (mv->as_mv.row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN))
+ mv->as_mv.row = xd->mb_to_top_edge - LEFT_TOP_MARGIN;
+ else if (mv->as_mv.row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN)
+ mv->as_mv.row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN;
+}
+
+static void vp8_clamp_mv(int_mv *mv, int mb_to_left_edge, int mb_to_right_edge,
+ int mb_to_top_edge, int mb_to_bottom_edge)
+{
+ mv->as_mv.col = (mv->as_mv.col < mb_to_left_edge) ?
+ mb_to_left_edge : mv->as_mv.col;
+ mv->as_mv.col = (mv->as_mv.col > mb_to_right_edge) ?
+ mb_to_right_edge : mv->as_mv.col;
+ mv->as_mv.row = (mv->as_mv.row < mb_to_top_edge) ?
+ mb_to_top_edge : mv->as_mv.row;
+ mv->as_mv.row = (mv->as_mv.row > mb_to_bottom_edge) ?
+ mb_to_bottom_edge : mv->as_mv.row;
+}
+static unsigned int vp8_check_mv_bounds(int_mv *mv, int mb_to_left_edge,
+ int mb_to_right_edge, int mb_to_top_edge,
+ int mb_to_bottom_edge)
+{
+ unsigned int need_to_clamp;
+ need_to_clamp = (mv->as_mv.col < mb_to_left_edge);
+ need_to_clamp |= (mv->as_mv.col > mb_to_right_edge);
+ need_to_clamp |= (mv->as_mv.row < mb_to_top_edge);
+ need_to_clamp |= (mv->as_mv.row > mb_to_bottom_edge);
+ return need_to_clamp;
+}
+
+void vp8_find_near_mvs
+(
+ MACROBLOCKD *xd,
+ const MODE_INFO *here,
+ int_mv *nearest, int_mv *nearby, int_mv *best,
+ int near_mv_ref_cts[4],
+ int refframe,
+ int *ref_frame_sign_bias
+);
+
+
+int vp8_find_near_mvs_bias
+(
+ MACROBLOCKD *xd,
+ const MODE_INFO *here,
+ int_mv mode_mv_sb[2][MB_MODE_COUNT],
+ int_mv best_mv_sb[2],
+ int cnt[4],
+ int refframe,
+ int *ref_frame_sign_bias
+);
+
+
+vp8_prob *vp8_mv_ref_probs(
+ vp8_prob p[VP8_MVREFS-1], const int near_mv_ref_ct[4]
+);
+
+extern const unsigned char vp8_mbsplit_offset[4][16];
+
+
+static int left_block_mv(const MODE_INFO *cur_mb, int b)
+{
+ if (!(b & 3))
+ {
+ /* On L edge, get from MB to left of us */
+ --cur_mb;
+
+ if(cur_mb->mbmi.mode != SPLITMV)
+ return cur_mb->mbmi.mv.as_int;
+ b += 4;
+ }
+
+ return (cur_mb->bmi + b - 1)->mv.as_int;
+}
+
+static int above_block_mv(const MODE_INFO *cur_mb, int b, int mi_stride)
+{
+ if (!(b >> 2))
+ {
+ /* On top edge, get from MB above us */
+ cur_mb -= mi_stride;
+
+ if(cur_mb->mbmi.mode != SPLITMV)
+ return cur_mb->mbmi.mv.as_int;
+ b += 16;
+ }
+
+ return (cur_mb->bmi + b - 4)->mv.as_int;
+}
+static B_PREDICTION_MODE left_block_mode(const MODE_INFO *cur_mb, int b)
+{
+ if (!(b & 3))
+ {
+ /* On L edge, get from MB to left of us */
+ --cur_mb;
+ switch (cur_mb->mbmi.mode)
+ {
+ case B_PRED:
+ return (cur_mb->bmi + b + 3)->as_mode;
+ case DC_PRED:
+ return B_DC_PRED;
+ case V_PRED:
+ return B_VE_PRED;
+ case H_PRED:
+ return B_HE_PRED;
+ case TM_PRED:
+ return B_TM_PRED;
+ default:
+ return B_DC_PRED;
+ }
+ }
+
+ return (cur_mb->bmi + b - 1)->as_mode;
+}
+
+static B_PREDICTION_MODE above_block_mode(const MODE_INFO *cur_mb, int b, int mi_stride)
+{
+ if (!(b >> 2))
+ {
+ /* On top edge, get from MB above us */
+ cur_mb -= mi_stride;
+
+ switch (cur_mb->mbmi.mode)
+ {
+ case B_PRED:
+ return (cur_mb->bmi + b + 12)->as_mode;
+ case DC_PRED:
+ return B_DC_PRED;
+ case V_PRED:
+ return B_VE_PRED;
+ case H_PRED:
+ return B_HE_PRED;
+ case TM_PRED:
+ return B_TM_PRED;
+ default:
+ return B_DC_PRED;
+ }
+ }
+
+ return (cur_mb->bmi + b - 4)->as_mode;
+}
+
+#endif
diff --git a/libvpx/vp8/common/generic/systemdependent.c b/libvpx/vp8/common/generic/systemdependent.c
new file mode 100644
index 0000000..5a6ac7b
--- /dev/null
+++ b/libvpx/vp8/common/generic/systemdependent.c
@@ -0,0 +1,97 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#if ARCH_ARM
+#include "vpx_ports/arm.h"
+#elif ARCH_X86 || ARCH_X86_64
+#include "vpx_ports/x86.h"
+#endif
+#include "vp8/common/onyxc_int.h"
+
+#if CONFIG_MULTITHREAD
+#if HAVE_UNISTD_H && !defined(__OS2__)
+#include <unistd.h>
+#elif defined(_WIN32)
+#include <windows.h>
+typedef void (WINAPI *PGNSI)(LPSYSTEM_INFO);
+#elif defined(__OS2__)
+#define INCL_DOS
+#define INCL_DOSSPINLOCK
+#include <os2.h>
+#endif
+#endif
+
+#if CONFIG_MULTITHREAD
+static int get_cpu_count()
+{
+ int core_count = 16;
+
+#if HAVE_UNISTD_H && !defined(__OS2__)
+#if defined(_SC_NPROCESSORS_ONLN)
+ core_count = sysconf(_SC_NPROCESSORS_ONLN);
+#elif defined(_SC_NPROC_ONLN)
+ core_count = sysconf(_SC_NPROC_ONLN);
+#endif
+#elif defined(_WIN32)
+ {
+ PGNSI pGNSI;
+ SYSTEM_INFO sysinfo;
+
+ /* Call GetNativeSystemInfo if supported or
+ * GetSystemInfo otherwise. */
+
+ pGNSI = (PGNSI) GetProcAddress(
+ GetModuleHandle(TEXT("kernel32.dll")), "GetNativeSystemInfo");
+ if (pGNSI != NULL)
+ pGNSI(&sysinfo);
+ else
+ GetSystemInfo(&sysinfo);
+
+ core_count = sysinfo.dwNumberOfProcessors;
+ }
+#elif defined(__OS2__)
+ {
+ ULONG proc_id;
+ ULONG status;
+
+ core_count = 0;
+ for (proc_id = 1; ; proc_id++)
+ {
+ if (DosGetProcessorStatus(proc_id, &status))
+ break;
+
+ if (status == PROC_ONLINE)
+ core_count++;
+ }
+ }
+#else
+ /* other platforms */
+#endif
+
+ return core_count > 0 ? core_count : 1;
+}
+#endif
+
+
+void vp8_machine_specific_config(VP8_COMMON *ctx)
+{
+#if CONFIG_MULTITHREAD
+ ctx->processor_core_count = get_cpu_count();
+#endif /* CONFIG_MULTITHREAD */
+
+#if ARCH_ARM
+ ctx->cpu_caps = arm_cpu_caps();
+#elif ARCH_X86 || ARCH_X86_64
+ ctx->cpu_caps = x86_simd_caps();
+#endif
+}
diff --git a/vp8/common/header.h b/libvpx/vp8/common/header.h
similarity index 100%
rename from vp8/common/header.h
rename to libvpx/vp8/common/header.h
diff --git a/libvpx/vp8/common/idct_blk.c b/libvpx/vp8/common/idct_blk.c
new file mode 100644
index 0000000..0b058c7
--- /dev/null
+++ b/libvpx/vp8/common/idct_blk.c
@@ -0,0 +1,89 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+
+void vp8_dequant_idct_add_c(short *input, short *dq,
+ unsigned char *dest, int stride);
+void vp8_dc_only_idct_add_c(short input_dc, unsigned char * pred,
+ int pred_stride, unsigned char *dst_ptr,
+ int dst_stride);
+
+void vp8_dequant_idct_add_y_block_c
+ (short *q, short *dq,
+ unsigned char *dst, int stride, char *eobs)
+{
+ int i, j;
+
+ for (i = 0; i < 4; i++)
+ {
+ for (j = 0; j < 4; j++)
+ {
+ if (*eobs++ > 1)
+ vp8_dequant_idct_add_c (q, dq, dst, stride);
+ else
+ {
+ vp8_dc_only_idct_add_c (q[0]*dq[0], dst, stride, dst, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ q += 16;
+ dst += 4;
+ }
+
+ dst += 4*stride - 16;
+ }
+}
+
+void vp8_dequant_idct_add_uv_block_c
+ (short *q, short *dq,
+ unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
+{
+ int i, j;
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 2; j++)
+ {
+ if (*eobs++ > 1)
+ vp8_dequant_idct_add_c (q, dq, dstu, stride);
+ else
+ {
+ vp8_dc_only_idct_add_c (q[0]*dq[0], dstu, stride, dstu, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ q += 16;
+ dstu += 4;
+ }
+
+ dstu += 4*stride - 8;
+ }
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 2; j++)
+ {
+ if (*eobs++ > 1)
+ vp8_dequant_idct_add_c (q, dq, dstv, stride);
+ else
+ {
+ vp8_dc_only_idct_add_c (q[0]*dq[0], dstv, stride, dstv, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ q += 16;
+ dstv += 4;
+ }
+
+ dstv += 4*stride - 8;
+ }
+}
diff --git a/libvpx/vp8/common/idctllm.c b/libvpx/vp8/common/idctllm.c
new file mode 100644
index 0000000..47af52f
--- /dev/null
+++ b/libvpx/vp8/common/idctllm.c
@@ -0,0 +1,204 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/****************************************************************************
+ * Notes:
+ *
+ * This implementation makes use of 16 bit fixed point verio of two multiply
+ * constants:
+ * 1. sqrt(2) * cos (pi/8)
+ * 2. sqrt(2) * sin (pi/8)
+ * Becuase the first constant is bigger than 1, to maintain the same 16 bit
+ * fixed point precision as the second one, we use a trick of
+ * x * a = x + x*(a-1)
+ * so
+ * x * sqrt(2) * cos (pi/8) = x + x * (sqrt(2) *cos(pi/8)-1).
+ **************************************************************************/
+static const int cospi8sqrt2minus1 = 20091;
+static const int sinpi8sqrt2 = 35468;
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred_ptr,
+ int pred_stride, unsigned char *dst_ptr,
+ int dst_stride)
+{
+ int i;
+ int r, c;
+ int a1, b1, c1, d1;
+ short output[16];
+ short *ip = input;
+ short *op = output;
+ int temp1, temp2;
+ int shortpitch = 4;
+
+ for (i = 0; i < 4; i++)
+ {
+ a1 = ip[0] + ip[8];
+ b1 = ip[0] - ip[8];
+
+ temp1 = (ip[4] * sinpi8sqrt2) >> 16;
+ temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
+ c1 = temp1 - temp2;
+
+ temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
+ temp2 = (ip[12] * sinpi8sqrt2) >> 16;
+ d1 = temp1 + temp2;
+
+ op[shortpitch*0] = a1 + d1;
+ op[shortpitch*3] = a1 - d1;
+
+ op[shortpitch*1] = b1 + c1;
+ op[shortpitch*2] = b1 - c1;
+
+ ip++;
+ op++;
+ }
+
+ ip = output;
+ op = output;
+
+ for (i = 0; i < 4; i++)
+ {
+ a1 = ip[0] + ip[2];
+ b1 = ip[0] - ip[2];
+
+ temp1 = (ip[1] * sinpi8sqrt2) >> 16;
+ temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
+ c1 = temp1 - temp2;
+
+ temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
+ temp2 = (ip[3] * sinpi8sqrt2) >> 16;
+ d1 = temp1 + temp2;
+
+
+ op[0] = (a1 + d1 + 4) >> 3;
+ op[3] = (a1 - d1 + 4) >> 3;
+
+ op[1] = (b1 + c1 + 4) >> 3;
+ op[2] = (b1 - c1 + 4) >> 3;
+
+ ip += shortpitch;
+ op += shortpitch;
+ }
+
+ ip = output;
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ int a = ip[c] + pred_ptr[c] ;
+
+ if (a < 0)
+ a = 0;
+
+ if (a > 255)
+ a = 255;
+
+ dst_ptr[c] = (unsigned char) a ;
+ }
+ ip += 4;
+ dst_ptr += dst_stride;
+ pred_ptr += pred_stride;
+ }
+}
+
+void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr,
+ int pred_stride, unsigned char *dst_ptr,
+ int dst_stride)
+{
+ int a1 = ((input_dc + 4) >> 3);
+ int r, c;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ int a = a1 + pred_ptr[c] ;
+
+ if (a < 0)
+ a = 0;
+
+ if (a > 255)
+ a = 255;
+
+ dst_ptr[c] = (unsigned char) a ;
+ }
+
+ dst_ptr += dst_stride;
+ pred_ptr += pred_stride;
+ }
+
+}
+
+void vp8_short_inv_walsh4x4_c(short *input, short *mb_dqcoeff)
+{
+ short output[16];
+ int i;
+ int a1, b1, c1, d1;
+ int a2, b2, c2, d2;
+ short *ip = input;
+ short *op = output;
+
+ for (i = 0; i < 4; i++)
+ {
+ a1 = ip[0] + ip[12];
+ b1 = ip[4] + ip[8];
+ c1 = ip[4] - ip[8];
+ d1 = ip[0] - ip[12];
+
+ op[0] = a1 + b1;
+ op[4] = c1 + d1;
+ op[8] = a1 - b1;
+ op[12] = d1 - c1;
+ ip++;
+ op++;
+ }
+
+ ip = output;
+ op = output;
+
+ for (i = 0; i < 4; i++)
+ {
+ a1 = ip[0] + ip[3];
+ b1 = ip[1] + ip[2];
+ c1 = ip[1] - ip[2];
+ d1 = ip[0] - ip[3];
+
+ a2 = a1 + b1;
+ b2 = c1 + d1;
+ c2 = a1 - b1;
+ d2 = d1 - c1;
+
+ op[0] = (a2 + 3) >> 3;
+ op[1] = (b2 + 3) >> 3;
+ op[2] = (c2 + 3) >> 3;
+ op[3] = (d2 + 3) >> 3;
+
+ ip += 4;
+ op += 4;
+ }
+
+ for(i = 0; i < 16; i++)
+ {
+ mb_dqcoeff[i * 16] = output[i];
+ }
+}
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *mb_dqcoeff)
+{
+ int i;
+ int a1;
+
+ a1 = ((input[0] + 3) >> 3);
+ for(i = 0; i < 16; i++)
+ {
+ mb_dqcoeff[i * 16] = a1;
+ }
+}
diff --git a/libvpx/vp8/common/invtrans.h b/libvpx/vp8/common/invtrans.h
new file mode 100644
index 0000000..d048665
--- /dev/null
+++ b/libvpx/vp8/common/invtrans.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_INVTRANS_H
+#define __INC_INVTRANS_H
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "blockd.h"
+#include "onyxc_int.h"
+
+#if CONFIG_MULTITHREAD
+#include "vpx_mem/vpx_mem.h"
+#endif
+
+static void eob_adjust(char *eobs, short *diff)
+{
+ /* eob adjust.... the idct can only skip if both the dc and eob are zero */
+ int js;
+ for(js = 0; js < 16; js++)
+ {
+ if((eobs[js] == 0) && (diff[0] != 0))
+ eobs[js]++;
+ diff+=16;
+ }
+}
+
+static void vp8_inverse_transform_mby(MACROBLOCKD *xd)
+{
+ short *DQC = xd->dequant_y1;
+
+ if (xd->mode_info_context->mbmi.mode != SPLITMV)
+ {
+ /* do 2nd order transform on the dc block */
+ if (xd->eobs[24] > 1)
+ {
+ vp8_short_inv_walsh4x4
+ (&xd->block[24].dqcoeff[0], xd->qcoeff);
+ }
+ else
+ {
+ vp8_short_inv_walsh4x4_1
+ (&xd->block[24].dqcoeff[0], xd->qcoeff);
+ }
+ eob_adjust(xd->eobs, xd->qcoeff);
+
+ DQC = xd->dequant_y1_dc;
+ }
+ vp8_dequant_idct_add_y_block
+ (xd->qcoeff, DQC,
+ xd->dst.y_buffer,
+ xd->dst.y_stride, xd->eobs);
+}
+#endif
diff --git a/libvpx/vp8/common/loopfilter.c b/libvpx/vp8/common/loopfilter.c
new file mode 100644
index 0000000..41b4f12
--- /dev/null
+++ b/libvpx/vp8/common/loopfilter.c
@@ -0,0 +1,679 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "loopfilter.h"
+#include "onyxc_int.h"
+#include "vpx_mem/vpx_mem.h"
+
+typedef unsigned char uc;
+
+static void lf_init_lut(loop_filter_info_n *lfi)
+{
+ int filt_lvl;
+
+ for (filt_lvl = 0; filt_lvl <= MAX_LOOP_FILTER; filt_lvl++)
+ {
+ if (filt_lvl >= 40)
+ {
+ lfi->hev_thr_lut[KEY_FRAME][filt_lvl] = 2;
+ lfi->hev_thr_lut[INTER_FRAME][filt_lvl] = 3;
+ }
+ else if (filt_lvl >= 20)
+ {
+ lfi->hev_thr_lut[KEY_FRAME][filt_lvl] = 1;
+ lfi->hev_thr_lut[INTER_FRAME][filt_lvl] = 2;
+ }
+ else if (filt_lvl >= 15)
+ {
+ lfi->hev_thr_lut[KEY_FRAME][filt_lvl] = 1;
+ lfi->hev_thr_lut[INTER_FRAME][filt_lvl] = 1;
+ }
+ else
+ {
+ lfi->hev_thr_lut[KEY_FRAME][filt_lvl] = 0;
+ lfi->hev_thr_lut[INTER_FRAME][filt_lvl] = 0;
+ }
+ }
+
+ lfi->mode_lf_lut[DC_PRED] = 1;
+ lfi->mode_lf_lut[V_PRED] = 1;
+ lfi->mode_lf_lut[H_PRED] = 1;
+ lfi->mode_lf_lut[TM_PRED] = 1;
+ lfi->mode_lf_lut[B_PRED] = 0;
+
+ lfi->mode_lf_lut[ZEROMV] = 1;
+ lfi->mode_lf_lut[NEARESTMV] = 2;
+ lfi->mode_lf_lut[NEARMV] = 2;
+ lfi->mode_lf_lut[NEWMV] = 2;
+ lfi->mode_lf_lut[SPLITMV] = 3;
+
+}
+
+void vp8_loop_filter_update_sharpness(loop_filter_info_n *lfi,
+ int sharpness_lvl)
+{
+ int i;
+
+ /* For each possible value for the loop filter fill out limits */
+ for (i = 0; i <= MAX_LOOP_FILTER; i++)
+ {
+ int filt_lvl = i;
+ int block_inside_limit = 0;
+
+ /* Set loop filter paramaeters that control sharpness. */
+ block_inside_limit = filt_lvl >> (sharpness_lvl > 0);
+ block_inside_limit = block_inside_limit >> (sharpness_lvl > 4);
+
+ if (sharpness_lvl > 0)
+ {
+ if (block_inside_limit > (9 - sharpness_lvl))
+ block_inside_limit = (9 - sharpness_lvl);
+ }
+
+ if (block_inside_limit < 1)
+ block_inside_limit = 1;
+
+ vpx_memset(lfi->lim[i], block_inside_limit, SIMD_WIDTH);
+ vpx_memset(lfi->blim[i], (2 * filt_lvl + block_inside_limit),
+ SIMD_WIDTH);
+ vpx_memset(lfi->mblim[i], (2 * (filt_lvl + 2) + block_inside_limit),
+ SIMD_WIDTH);
+ }
+}
+
+void vp8_loop_filter_init(VP8_COMMON *cm)
+{
+ loop_filter_info_n *lfi = &cm->lf_info;
+ int i;
+
+ /* init limits for given sharpness*/
+ vp8_loop_filter_update_sharpness(lfi, cm->sharpness_level);
+ cm->last_sharpness_level = cm->sharpness_level;
+
+ /* init LUT for lvl and hev thr picking */
+ lf_init_lut(lfi);
+
+ /* init hev threshold const vectors */
+ for(i = 0; i < 4 ; i++)
+ {
+ vpx_memset(lfi->hev_thr[i], i, SIMD_WIDTH);
+ }
+}
+
+void vp8_loop_filter_frame_init(VP8_COMMON *cm,
+ MACROBLOCKD *mbd,
+ int default_filt_lvl)
+{
+ int seg, /* segment number */
+ ref, /* index in ref_lf_deltas */
+ mode; /* index in mode_lf_deltas */
+
+ loop_filter_info_n *lfi = &cm->lf_info;
+
+ /* update limits if sharpness has changed */
+ if(cm->last_sharpness_level != cm->sharpness_level)
+ {
+ vp8_loop_filter_update_sharpness(lfi, cm->sharpness_level);
+ cm->last_sharpness_level = cm->sharpness_level;
+ }
+
+ for(seg = 0; seg < MAX_MB_SEGMENTS; seg++)
+ {
+ int lvl_seg = default_filt_lvl;
+ int lvl_ref, lvl_mode;
+
+ /* Note the baseline filter values for each segment */
+ if (mbd->segmentation_enabled)
+ {
+ /* Abs value */
+ if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
+ {
+ lvl_seg = mbd->segment_feature_data[MB_LVL_ALT_LF][seg];
+ }
+ else /* Delta Value */
+ {
+ lvl_seg += mbd->segment_feature_data[MB_LVL_ALT_LF][seg];
+ lvl_seg = (lvl_seg > 0) ? ((lvl_seg > 63) ? 63: lvl_seg) : 0;
+ }
+ }
+
+ if (!mbd->mode_ref_lf_delta_enabled)
+ {
+ /* we could get rid of this if we assume that deltas are set to
+ * zero when not in use; encoder always uses deltas
+ */
+ vpx_memset(lfi->lvl[seg][0], lvl_seg, 4 * 4 );
+ continue;
+ }
+
+ lvl_ref = lvl_seg;
+
+ /* INTRA_FRAME */
+ ref = INTRA_FRAME;
+
+ /* Apply delta for reference frame */
+ lvl_ref += mbd->ref_lf_deltas[ref];
+
+ /* Apply delta for Intra modes */
+ mode = 0; /* B_PRED */
+ /* Only the split mode BPRED has a further special case */
+ lvl_mode = lvl_ref + mbd->mode_lf_deltas[mode];
+ lvl_mode = (lvl_mode > 0) ? (lvl_mode > 63 ? 63 : lvl_mode) : 0; /* clamp */
+
+ lfi->lvl[seg][ref][mode] = lvl_mode;
+
+ mode = 1; /* all the rest of Intra modes */
+ lvl_mode = (lvl_ref > 0) ? (lvl_ref > 63 ? 63 : lvl_ref) : 0; /* clamp */
+ lfi->lvl[seg][ref][mode] = lvl_mode;
+
+ /* LAST, GOLDEN, ALT */
+ for(ref = 1; ref < MAX_REF_FRAMES; ref++)
+ {
+ int lvl_ref = lvl_seg;
+
+ /* Apply delta for reference frame */
+ lvl_ref += mbd->ref_lf_deltas[ref];
+
+ /* Apply delta for Inter modes */
+ for (mode = 1; mode < 4; mode++)
+ {
+ lvl_mode = lvl_ref + mbd->mode_lf_deltas[mode];
+ lvl_mode = (lvl_mode > 0) ? (lvl_mode > 63 ? 63 : lvl_mode) : 0; /* clamp */
+
+ lfi->lvl[seg][ref][mode] = lvl_mode;
+ }
+ }
+ }
+}
+
+
+void vp8_loop_filter_row_normal(VP8_COMMON *cm, MODE_INFO *mode_info_context,
+ int mb_row, int post_ystride, int post_uvstride,
+ unsigned char *y_ptr, unsigned char *u_ptr,
+ unsigned char *v_ptr)
+{
+ int mb_col;
+ int filter_level;
+ loop_filter_info_n *lfi_n = &cm->lf_info;
+ loop_filter_info lfi;
+ FRAME_TYPE frame_type = cm->frame_type;
+
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+ int skip_lf = (mode_info_context->mbmi.mode != B_PRED &&
+ mode_info_context->mbmi.mode != SPLITMV &&
+ mode_info_context->mbmi.mb_skip_coeff);
+
+ const int mode_index = lfi_n->mode_lf_lut[mode_info_context->mbmi.mode];
+ const int seg = mode_info_context->mbmi.segment_id;
+ const int ref_frame = mode_info_context->mbmi.ref_frame;
+
+ filter_level = lfi_n->lvl[seg][ref_frame][mode_index];
+
+ if (filter_level)
+ {
+ const int hev_index = lfi_n->hev_thr_lut[frame_type][filter_level];
+ lfi.mblim = lfi_n->mblim[filter_level];
+ lfi.blim = lfi_n->blim[filter_level];
+ lfi.lim = lfi_n->lim[filter_level];
+ lfi.hev_thr = lfi_n->hev_thr[hev_index];
+
+ if (mb_col > 0)
+ vp8_loop_filter_mbv
+ (y_ptr, u_ptr, v_ptr, post_ystride, post_uvstride, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bv
+ (y_ptr, u_ptr, v_ptr, post_ystride, post_uvstride, &lfi);
+
+ /* don't apply across umv border */
+ if (mb_row > 0)
+ vp8_loop_filter_mbh
+ (y_ptr, u_ptr, v_ptr, post_ystride, post_uvstride, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bh
+ (y_ptr, u_ptr, v_ptr, post_ystride, post_uvstride, &lfi);
+ }
+
+ y_ptr += 16;
+ u_ptr += 8;
+ v_ptr += 8;
+
+ mode_info_context++; /* step to next MB */
+ }
+
+}
+
+void vp8_loop_filter_row_simple(VP8_COMMON *cm, MODE_INFO *mode_info_context,
+ int mb_row, int post_ystride, int post_uvstride,
+ unsigned char *y_ptr, unsigned char *u_ptr,
+ unsigned char *v_ptr)
+{
+ int mb_col;
+ int filter_level;
+ loop_filter_info_n *lfi_n = &cm->lf_info;
+
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+ int skip_lf = (mode_info_context->mbmi.mode != B_PRED &&
+ mode_info_context->mbmi.mode != SPLITMV &&
+ mode_info_context->mbmi.mb_skip_coeff);
+
+ const int mode_index = lfi_n->mode_lf_lut[mode_info_context->mbmi.mode];
+ const int seg = mode_info_context->mbmi.segment_id;
+ const int ref_frame = mode_info_context->mbmi.ref_frame;
+
+ filter_level = lfi_n->lvl[seg][ref_frame][mode_index];
+
+ if (filter_level)
+ {
+ if (mb_col > 0)
+ vp8_loop_filter_simple_mbv
+ (y_ptr, post_ystride, lfi_n->mblim[filter_level]);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bv
+ (y_ptr, post_ystride, lfi_n->blim[filter_level]);
+
+ /* don't apply across umv border */
+ if (mb_row > 0)
+ vp8_loop_filter_simple_mbh
+ (y_ptr, post_ystride, lfi_n->mblim[filter_level]);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bh
+ (y_ptr, post_ystride, lfi_n->blim[filter_level]);
+ }
+
+ y_ptr += 16;
+ u_ptr += 8;
+ v_ptr += 8;
+
+ mode_info_context++; /* step to next MB */
+ }
+
+}
+void vp8_loop_filter_frame(VP8_COMMON *cm,
+ MACROBLOCKD *mbd,
+ int frame_type)
+{
+ YV12_BUFFER_CONFIG *post = cm->frame_to_show;
+ loop_filter_info_n *lfi_n = &cm->lf_info;
+ loop_filter_info lfi;
+
+ int mb_row;
+ int mb_col;
+ int mb_rows = cm->mb_rows;
+ int mb_cols = cm->mb_cols;
+
+ int filter_level;
+
+ unsigned char *y_ptr, *u_ptr, *v_ptr;
+
+ /* Point at base of Mb MODE_INFO list */
+ const MODE_INFO *mode_info_context = cm->mi;
+ int post_y_stride = post->y_stride;
+ int post_uv_stride = post->uv_stride;
+
+ /* Initialize the loop filter for this frame. */
+ vp8_loop_filter_frame_init(cm, mbd, cm->filter_level);
+
+ /* Set up the buffer pointers */
+ y_ptr = post->y_buffer;
+ u_ptr = post->u_buffer;
+ v_ptr = post->v_buffer;
+
+ /* vp8_filter each macro block */
+ if (cm->filter_type == NORMAL_LOOPFILTER)
+ {
+ for (mb_row = 0; mb_row < mb_rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < mb_cols; mb_col++)
+ {
+ int skip_lf = (mode_info_context->mbmi.mode != B_PRED &&
+ mode_info_context->mbmi.mode != SPLITMV &&
+ mode_info_context->mbmi.mb_skip_coeff);
+
+ const int mode_index = lfi_n->mode_lf_lut[mode_info_context->mbmi.mode];
+ const int seg = mode_info_context->mbmi.segment_id;
+ const int ref_frame = mode_info_context->mbmi.ref_frame;
+
+ filter_level = lfi_n->lvl[seg][ref_frame][mode_index];
+
+ if (filter_level)
+ {
+ const int hev_index = lfi_n->hev_thr_lut[frame_type][filter_level];
+ lfi.mblim = lfi_n->mblim[filter_level];
+ lfi.blim = lfi_n->blim[filter_level];
+ lfi.lim = lfi_n->lim[filter_level];
+ lfi.hev_thr = lfi_n->hev_thr[hev_index];
+
+ if (mb_col > 0)
+ vp8_loop_filter_mbv
+ (y_ptr, u_ptr, v_ptr, post_y_stride, post_uv_stride, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bv
+ (y_ptr, u_ptr, v_ptr, post_y_stride, post_uv_stride, &lfi);
+
+ /* don't apply across umv border */
+ if (mb_row > 0)
+ vp8_loop_filter_mbh
+ (y_ptr, u_ptr, v_ptr, post_y_stride, post_uv_stride, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bh
+ (y_ptr, u_ptr, v_ptr, post_y_stride, post_uv_stride, &lfi);
+ }
+
+ y_ptr += 16;
+ u_ptr += 8;
+ v_ptr += 8;
+
+ mode_info_context++; /* step to next MB */
+ }
+ y_ptr += post_y_stride * 16 - post->y_width;
+ u_ptr += post_uv_stride * 8 - post->uv_width;
+ v_ptr += post_uv_stride * 8 - post->uv_width;
+
+ mode_info_context++; /* Skip border mb */
+
+ }
+ }
+ else /* SIMPLE_LOOPFILTER */
+ {
+ for (mb_row = 0; mb_row < mb_rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < mb_cols; mb_col++)
+ {
+ int skip_lf = (mode_info_context->mbmi.mode != B_PRED &&
+ mode_info_context->mbmi.mode != SPLITMV &&
+ mode_info_context->mbmi.mb_skip_coeff);
+
+ const int mode_index = lfi_n->mode_lf_lut[mode_info_context->mbmi.mode];
+ const int seg = mode_info_context->mbmi.segment_id;
+ const int ref_frame = mode_info_context->mbmi.ref_frame;
+
+ filter_level = lfi_n->lvl[seg][ref_frame][mode_index];
+ if (filter_level)
+ {
+ const unsigned char * mblim = lfi_n->mblim[filter_level];
+ const unsigned char * blim = lfi_n->blim[filter_level];
+
+ if (mb_col > 0)
+ vp8_loop_filter_simple_mbv
+ (y_ptr, post_y_stride, mblim);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bv
+ (y_ptr, post_y_stride, blim);
+
+ /* don't apply across umv border */
+ if (mb_row > 0)
+ vp8_loop_filter_simple_mbh
+ (y_ptr, post_y_stride, mblim);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bh
+ (y_ptr, post_y_stride, blim);
+ }
+
+ y_ptr += 16;
+ u_ptr += 8;
+ v_ptr += 8;
+
+ mode_info_context++; /* step to next MB */
+ }
+ y_ptr += post_y_stride * 16 - post->y_width;
+ u_ptr += post_uv_stride * 8 - post->uv_width;
+ v_ptr += post_uv_stride * 8 - post->uv_width;
+
+ mode_info_context++; /* Skip border mb */
+
+ }
+ }
+}
+
+void vp8_loop_filter_frame_yonly
+(
+ VP8_COMMON *cm,
+ MACROBLOCKD *mbd,
+ int default_filt_lvl
+)
+{
+ YV12_BUFFER_CONFIG *post = cm->frame_to_show;
+
+ unsigned char *y_ptr;
+ int mb_row;
+ int mb_col;
+
+ loop_filter_info_n *lfi_n = &cm->lf_info;
+ loop_filter_info lfi;
+
+ int filter_level;
+ FRAME_TYPE frame_type = cm->frame_type;
+
+ /* Point at base of Mb MODE_INFO list */
+ const MODE_INFO *mode_info_context = cm->mi;
+
+#if 0
+ if(default_filt_lvl == 0) /* no filter applied */
+ return;
+#endif
+
+ /* Initialize the loop filter for this frame. */
+ vp8_loop_filter_frame_init( cm, mbd, default_filt_lvl);
+
+ /* Set up the buffer pointers */
+ y_ptr = post->y_buffer;
+
+ /* vp8_filter each macro block */
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+ int skip_lf = (mode_info_context->mbmi.mode != B_PRED &&
+ mode_info_context->mbmi.mode != SPLITMV &&
+ mode_info_context->mbmi.mb_skip_coeff);
+
+ const int mode_index = lfi_n->mode_lf_lut[mode_info_context->mbmi.mode];
+ const int seg = mode_info_context->mbmi.segment_id;
+ const int ref_frame = mode_info_context->mbmi.ref_frame;
+
+ filter_level = lfi_n->lvl[seg][ref_frame][mode_index];
+
+ if (filter_level)
+ {
+ if (cm->filter_type == NORMAL_LOOPFILTER)
+ {
+ const int hev_index = lfi_n->hev_thr_lut[frame_type][filter_level];
+ lfi.mblim = lfi_n->mblim[filter_level];
+ lfi.blim = lfi_n->blim[filter_level];
+ lfi.lim = lfi_n->lim[filter_level];
+ lfi.hev_thr = lfi_n->hev_thr[hev_index];
+
+ if (mb_col > 0)
+ vp8_loop_filter_mbv
+ (y_ptr, 0, 0, post->y_stride, 0, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bv
+ (y_ptr, 0, 0, post->y_stride, 0, &lfi);
+
+ /* don't apply across umv border */
+ if (mb_row > 0)
+ vp8_loop_filter_mbh
+ (y_ptr, 0, 0, post->y_stride, 0, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bh
+ (y_ptr, 0, 0, post->y_stride, 0, &lfi);
+ }
+ else
+ {
+ if (mb_col > 0)
+ vp8_loop_filter_simple_mbv
+ (y_ptr, post->y_stride, lfi_n->mblim[filter_level]);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bv
+ (y_ptr, post->y_stride, lfi_n->blim[filter_level]);
+
+ /* don't apply across umv border */
+ if (mb_row > 0)
+ vp8_loop_filter_simple_mbh
+ (y_ptr, post->y_stride, lfi_n->mblim[filter_level]);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bh
+ (y_ptr, post->y_stride, lfi_n->blim[filter_level]);
+ }
+ }
+
+ y_ptr += 16;
+ mode_info_context ++; /* step to next MB */
+
+ }
+
+ y_ptr += post->y_stride * 16 - post->y_width;
+ mode_info_context ++; /* Skip border mb */
+ }
+
+}
+
+void vp8_loop_filter_partial_frame
+(
+ VP8_COMMON *cm,
+ MACROBLOCKD *mbd,
+ int default_filt_lvl
+)
+{
+ YV12_BUFFER_CONFIG *post = cm->frame_to_show;
+
+ unsigned char *y_ptr;
+ int mb_row;
+ int mb_col;
+ int mb_cols = post->y_width >> 4;
+ int mb_rows = post->y_height >> 4;
+
+ int linestocopy, i;
+
+ loop_filter_info_n *lfi_n = &cm->lf_info;
+ loop_filter_info lfi;
+
+ int filter_level;
+ int alt_flt_enabled = mbd->segmentation_enabled;
+ FRAME_TYPE frame_type = cm->frame_type;
+
+ const MODE_INFO *mode_info_context;
+
+ int lvl_seg[MAX_MB_SEGMENTS];
+
+ /* number of MB rows to use in partial filtering */
+ linestocopy = mb_rows / PARTIAL_FRAME_FRACTION;
+ linestocopy = linestocopy ? linestocopy << 4 : 16; /* 16 lines per MB */
+
+ /* Note the baseline filter values for each segment */
+ /* See vp8_loop_filter_frame_init. Rather than call that for each change
+ * to default_filt_lvl, copy the relevant calculation here.
+ */
+ if (alt_flt_enabled)
+ {
+ for (i = 0; i < MAX_MB_SEGMENTS; i++)
+ { /* Abs value */
+ if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
+ {
+ lvl_seg[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i];
+ }
+ /* Delta Value */
+ else
+ {
+ lvl_seg[i] = default_filt_lvl
+ + mbd->segment_feature_data[MB_LVL_ALT_LF][i];
+ lvl_seg[i] = (lvl_seg[i] > 0) ?
+ ((lvl_seg[i] > 63) ? 63: lvl_seg[i]) : 0;
+ }
+ }
+ }
+
+ /* Set up the buffer pointers; partial image starts at ~middle of frame */
+ y_ptr = post->y_buffer + ((post->y_height >> 5) * 16) * post->y_stride;
+ mode_info_context = cm->mi + (post->y_height >> 5) * (mb_cols + 1);
+
+ /* vp8_filter each macro block */
+ for (mb_row = 0; mb_row<(linestocopy >> 4); mb_row++)
+ {
+ for (mb_col = 0; mb_col < mb_cols; mb_col++)
+ {
+ int skip_lf = (mode_info_context->mbmi.mode != B_PRED &&
+ mode_info_context->mbmi.mode != SPLITMV &&
+ mode_info_context->mbmi.mb_skip_coeff);
+
+ if (alt_flt_enabled)
+ filter_level = lvl_seg[mode_info_context->mbmi.segment_id];
+ else
+ filter_level = default_filt_lvl;
+
+ if (filter_level)
+ {
+ if (cm->filter_type == NORMAL_LOOPFILTER)
+ {
+ const int hev_index = lfi_n->hev_thr_lut[frame_type][filter_level];
+ lfi.mblim = lfi_n->mblim[filter_level];
+ lfi.blim = lfi_n->blim[filter_level];
+ lfi.lim = lfi_n->lim[filter_level];
+ lfi.hev_thr = lfi_n->hev_thr[hev_index];
+
+ if (mb_col > 0)
+ vp8_loop_filter_mbv
+ (y_ptr, 0, 0, post->y_stride, 0, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bv
+ (y_ptr, 0, 0, post->y_stride, 0, &lfi);
+
+ vp8_loop_filter_mbh
+ (y_ptr, 0, 0, post->y_stride, 0, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bh
+ (y_ptr, 0, 0, post->y_stride, 0, &lfi);
+ }
+ else
+ {
+ if (mb_col > 0)
+ vp8_loop_filter_simple_mbv
+ (y_ptr, post->y_stride, lfi_n->mblim[filter_level]);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bv
+ (y_ptr, post->y_stride, lfi_n->blim[filter_level]);
+
+ vp8_loop_filter_simple_mbh
+ (y_ptr, post->y_stride, lfi_n->mblim[filter_level]);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bh
+ (y_ptr, post->y_stride, lfi_n->blim[filter_level]);
+ }
+ }
+
+ y_ptr += 16;
+ mode_info_context += 1; /* step to next MB */
+ }
+
+ y_ptr += post->y_stride * 16 - post->y_width;
+ mode_info_context += 1; /* Skip border mb */
+ }
+}
diff --git a/libvpx/vp8/common/loopfilter.h b/libvpx/vp8/common/loopfilter.h
new file mode 100644
index 0000000..b3af2d6
--- /dev/null
+++ b/libvpx/vp8/common/loopfilter.h
@@ -0,0 +1,105 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef loopfilter_h
+#define loopfilter_h
+
+#include "vpx_ports/mem.h"
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+
+#define MAX_LOOP_FILTER 63
+/* fraction of total macroblock rows to be used in fast filter level picking */
+/* has to be > 2 */
+#define PARTIAL_FRAME_FRACTION 8
+
+typedef enum
+{
+ NORMAL_LOOPFILTER = 0,
+ SIMPLE_LOOPFILTER = 1
+} LOOPFILTERTYPE;
+
+#if ARCH_ARM
+#define SIMD_WIDTH 1
+#else
+#define SIMD_WIDTH 16
+#endif
+
+/* Need to align this structure so when it is declared and
+ * passed it can be loaded into vector registers.
+ */
+typedef struct
+{
+ DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, mblim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
+ DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, blim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
+ DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, lim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
+ DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, hev_thr[4][SIMD_WIDTH]);
+ unsigned char lvl[4][4][4];
+ unsigned char hev_thr_lut[2][MAX_LOOP_FILTER + 1];
+ unsigned char mode_lf_lut[10];
+} loop_filter_info_n;
+
+typedef struct loop_filter_info
+{
+ const unsigned char * mblim;
+ const unsigned char * blim;
+ const unsigned char * lim;
+ const unsigned char * hev_thr;
+} loop_filter_info;
+
+
+typedef void loop_filter_uvfunction
+(
+ unsigned char *u, /* source pointer */
+ int p, /* pitch */
+ const unsigned char *blimit,
+ const unsigned char *limit,
+ const unsigned char *thresh,
+ unsigned char *v
+);
+
+/* assorted loopfilter functions which get used elsewhere */
+struct VP8Common;
+struct macroblockd;
+struct modeinfo;
+
+void vp8_loop_filter_init(struct VP8Common *cm);
+
+void vp8_loop_filter_frame_init(struct VP8Common *cm,
+ struct macroblockd *mbd,
+ int default_filt_lvl);
+
+void vp8_loop_filter_frame(struct VP8Common *cm, struct macroblockd *mbd,
+ int frame_type);
+
+void vp8_loop_filter_partial_frame(struct VP8Common *cm,
+ struct macroblockd *mbd,
+ int default_filt_lvl);
+
+void vp8_loop_filter_frame_yonly(struct VP8Common *cm,
+ struct macroblockd *mbd,
+ int default_filt_lvl);
+
+void vp8_loop_filter_update_sharpness(loop_filter_info_n *lfi,
+ int sharpness_lvl);
+
+void vp8_loop_filter_row_normal(struct VP8Common *cm,
+ struct modeinfo *mode_info_context,
+ int mb_row, int post_ystride, int post_uvstride,
+ unsigned char *y_ptr, unsigned char *u_ptr,
+ unsigned char *v_ptr);
+
+void vp8_loop_filter_row_simple(struct VP8Common *cm,
+ struct modeinfo *mode_info_context,
+ int mb_row, int post_ystride, int post_uvstride,
+ unsigned char *y_ptr, unsigned char *u_ptr,
+ unsigned char *v_ptr);
+#endif
diff --git a/libvpx/vp8/common/loopfilter_filters.c b/libvpx/vp8/common/loopfilter_filters.c
new file mode 100644
index 0000000..8235f6e
--- /dev/null
+++ b/libvpx/vp8/common/loopfilter_filters.c
@@ -0,0 +1,430 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdlib.h>
+#include "loopfilter.h"
+#include "onyxc_int.h"
+
+typedef unsigned char uc;
+
+static signed char vp8_signed_char_clamp(int t)
+{
+ t = (t < -128 ? -128 : t);
+ t = (t > 127 ? 127 : t);
+ return (signed char) t;
+}
+
+
+/* should we apply any filter at all ( 11111111 yes, 00000000 no) */
+static signed char vp8_filter_mask(uc limit, uc blimit,
+ uc p3, uc p2, uc p1, uc p0,
+ uc q0, uc q1, uc q2, uc q3)
+{
+ signed char mask = 0;
+ mask |= (abs(p3 - p2) > limit);
+ mask |= (abs(p2 - p1) > limit);
+ mask |= (abs(p1 - p0) > limit);
+ mask |= (abs(q1 - q0) > limit);
+ mask |= (abs(q2 - q1) > limit);
+ mask |= (abs(q3 - q2) > limit);
+ mask |= (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 > blimit);
+ return mask - 1;
+}
+
+/* is there high variance internal edge ( 11111111 yes, 00000000 no) */
+static signed char vp8_hevmask(uc thresh, uc p1, uc p0, uc q0, uc q1)
+{
+ signed char hev = 0;
+ hev |= (abs(p1 - p0) > thresh) * -1;
+ hev |= (abs(q1 - q0) > thresh) * -1;
+ return hev;
+}
+
+static void vp8_filter(signed char mask, uc hev, uc *op1,
+ uc *op0, uc *oq0, uc *oq1)
+
+{
+ signed char ps0, qs0;
+ signed char ps1, qs1;
+ signed char vp8_filter, Filter1, Filter2;
+ signed char u;
+
+ ps1 = (signed char) * op1 ^ 0x80;
+ ps0 = (signed char) * op0 ^ 0x80;
+ qs0 = (signed char) * oq0 ^ 0x80;
+ qs1 = (signed char) * oq1 ^ 0x80;
+
+ /* add outer taps if we have high edge variance */
+ vp8_filter = vp8_signed_char_clamp(ps1 - qs1);
+ vp8_filter &= hev;
+
+ /* inner taps */
+ vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0));
+ vp8_filter &= mask;
+
+ /* save bottom 3 bits so that we round one side +4 and the other +3
+ * if it equals 4 we'll set to adjust by -1 to account for the fact
+ * we'd round 3 the other way
+ */
+ Filter1 = vp8_signed_char_clamp(vp8_filter + 4);
+ Filter2 = vp8_signed_char_clamp(vp8_filter + 3);
+ Filter1 >>= 3;
+ Filter2 >>= 3;
+ u = vp8_signed_char_clamp(qs0 - Filter1);
+ *oq0 = u ^ 0x80;
+ u = vp8_signed_char_clamp(ps0 + Filter2);
+ *op0 = u ^ 0x80;
+ vp8_filter = Filter1;
+
+ /* outer tap adjustments */
+ vp8_filter += 1;
+ vp8_filter >>= 1;
+ vp8_filter &= ~hev;
+
+ u = vp8_signed_char_clamp(qs1 - vp8_filter);
+ *oq1 = u ^ 0x80;
+ u = vp8_signed_char_clamp(ps1 + vp8_filter);
+ *op1 = u ^ 0x80;
+
+}
+void vp8_loop_filter_horizontal_edge_c
+(
+ unsigned char *s,
+ int p, /* pitch */
+ const unsigned char *blimit,
+ const unsigned char *limit,
+ const unsigned char *thresh,
+ int count
+)
+{
+ int hev = 0; /* high edge variance */
+ signed char mask = 0;
+ int i = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+ do
+ {
+ mask = vp8_filter_mask(limit[0], blimit[0],
+ s[-4*p], s[-3*p], s[-2*p], s[-1*p],
+ s[0*p], s[1*p], s[2*p], s[3*p]);
+
+ hev = vp8_hevmask(thresh[0], s[-2*p], s[-1*p], s[0*p], s[1*p]);
+
+ vp8_filter(mask, hev, s - 2 * p, s - 1 * p, s, s + 1 * p);
+
+ ++s;
+ }
+ while (++i < count * 8);
+}
+
+void vp8_loop_filter_vertical_edge_c
+(
+ unsigned char *s,
+ int p,
+ const unsigned char *blimit,
+ const unsigned char *limit,
+ const unsigned char *thresh,
+ int count
+)
+{
+ int hev = 0; /* high edge variance */
+ signed char mask = 0;
+ int i = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+ do
+ {
+ mask = vp8_filter_mask(limit[0], blimit[0],
+ s[-4], s[-3], s[-2], s[-1], s[0], s[1], s[2], s[3]);
+
+ hev = vp8_hevmask(thresh[0], s[-2], s[-1], s[0], s[1]);
+
+ vp8_filter(mask, hev, s - 2, s - 1, s, s + 1);
+
+ s += p;
+ }
+ while (++i < count * 8);
+}
+
+static void vp8_mbfilter(signed char mask, uc hev,
+ uc *op2, uc *op1, uc *op0, uc *oq0, uc *oq1, uc *oq2)
+{
+ signed char s, u;
+ signed char vp8_filter, Filter1, Filter2;
+ signed char ps2 = (signed char) * op2 ^ 0x80;
+ signed char ps1 = (signed char) * op1 ^ 0x80;
+ signed char ps0 = (signed char) * op0 ^ 0x80;
+ signed char qs0 = (signed char) * oq0 ^ 0x80;
+ signed char qs1 = (signed char) * oq1 ^ 0x80;
+ signed char qs2 = (signed char) * oq2 ^ 0x80;
+
+ /* add outer taps if we have high edge variance */
+ vp8_filter = vp8_signed_char_clamp(ps1 - qs1);
+ vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0));
+ vp8_filter &= mask;
+
+ Filter2 = vp8_filter;
+ Filter2 &= hev;
+
+ /* save bottom 3 bits so that we round one side +4 and the other +3 */
+ Filter1 = vp8_signed_char_clamp(Filter2 + 4);
+ Filter2 = vp8_signed_char_clamp(Filter2 + 3);
+ Filter1 >>= 3;
+ Filter2 >>= 3;
+ qs0 = vp8_signed_char_clamp(qs0 - Filter1);
+ ps0 = vp8_signed_char_clamp(ps0 + Filter2);
+
+
+ /* only apply wider filter if not high edge variance */
+ vp8_filter &= ~hev;
+ Filter2 = vp8_filter;
+
+ /* roughly 3/7th difference across boundary */
+ u = vp8_signed_char_clamp((63 + Filter2 * 27) >> 7);
+ s = vp8_signed_char_clamp(qs0 - u);
+ *oq0 = s ^ 0x80;
+ s = vp8_signed_char_clamp(ps0 + u);
+ *op0 = s ^ 0x80;
+
+ /* roughly 2/7th difference across boundary */
+ u = vp8_signed_char_clamp((63 + Filter2 * 18) >> 7);
+ s = vp8_signed_char_clamp(qs1 - u);
+ *oq1 = s ^ 0x80;
+ s = vp8_signed_char_clamp(ps1 + u);
+ *op1 = s ^ 0x80;
+
+ /* roughly 1/7th difference across boundary */
+ u = vp8_signed_char_clamp((63 + Filter2 * 9) >> 7);
+ s = vp8_signed_char_clamp(qs2 - u);
+ *oq2 = s ^ 0x80;
+ s = vp8_signed_char_clamp(ps2 + u);
+ *op2 = s ^ 0x80;
+}
+
+void vp8_mbloop_filter_horizontal_edge_c
+(
+ unsigned char *s,
+ int p,
+ const unsigned char *blimit,
+ const unsigned char *limit,
+ const unsigned char *thresh,
+ int count
+)
+{
+ signed char hev = 0; /* high edge variance */
+ signed char mask = 0;
+ int i = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+ do
+ {
+
+ mask = vp8_filter_mask(limit[0], blimit[0],
+ s[-4*p], s[-3*p], s[-2*p], s[-1*p],
+ s[0*p], s[1*p], s[2*p], s[3*p]);
+
+ hev = vp8_hevmask(thresh[0], s[-2*p], s[-1*p], s[0*p], s[1*p]);
+
+ vp8_mbfilter(mask, hev, s - 3 * p, s - 2 * p, s - 1 * p, s, s + 1 * p, s + 2 * p);
+
+ ++s;
+ }
+ while (++i < count * 8);
+
+}
+
+
+void vp8_mbloop_filter_vertical_edge_c
+(
+ unsigned char *s,
+ int p,
+ const unsigned char *blimit,
+ const unsigned char *limit,
+ const unsigned char *thresh,
+ int count
+)
+{
+ signed char hev = 0; /* high edge variance */
+ signed char mask = 0;
+ int i = 0;
+
+ do
+ {
+
+ mask = vp8_filter_mask(limit[0], blimit[0],
+ s[-4], s[-3], s[-2], s[-1], s[0], s[1], s[2], s[3]);
+
+ hev = vp8_hevmask(thresh[0], s[-2], s[-1], s[0], s[1]);
+
+ vp8_mbfilter(mask, hev, s - 3, s - 2, s - 1, s, s + 1, s + 2);
+
+ s += p;
+ }
+ while (++i < count * 8);
+
+}
+
+/* should we apply any filter at all ( 11111111 yes, 00000000 no) */
+static signed char vp8_simple_filter_mask(uc blimit, uc p1, uc p0, uc q0, uc q1)
+{
+/* Why does this cause problems for win32?
+ * error C2143: syntax error : missing ';' before 'type'
+ * (void) limit;
+ */
+ signed char mask = (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 <= blimit) * -1;
+ return mask;
+}
+
+static void vp8_simple_filter(signed char mask, uc *op1, uc *op0, uc *oq0, uc *oq1)
+{
+ signed char vp8_filter, Filter1, Filter2;
+ signed char p1 = (signed char) * op1 ^ 0x80;
+ signed char p0 = (signed char) * op0 ^ 0x80;
+ signed char q0 = (signed char) * oq0 ^ 0x80;
+ signed char q1 = (signed char) * oq1 ^ 0x80;
+ signed char u;
+
+ vp8_filter = vp8_signed_char_clamp(p1 - q1);
+ vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (q0 - p0));
+ vp8_filter &= mask;
+
+ /* save bottom 3 bits so that we round one side +4 and the other +3 */
+ Filter1 = vp8_signed_char_clamp(vp8_filter + 4);
+ Filter1 >>= 3;
+ u = vp8_signed_char_clamp(q0 - Filter1);
+ *oq0 = u ^ 0x80;
+
+ Filter2 = vp8_signed_char_clamp(vp8_filter + 3);
+ Filter2 >>= 3;
+ u = vp8_signed_char_clamp(p0 + Filter2);
+ *op0 = u ^ 0x80;
+}
+
+void vp8_loop_filter_simple_horizontal_edge_c
+(
+ unsigned char *s,
+ int p,
+ const unsigned char *blimit
+)
+{
+ signed char mask = 0;
+ int i = 0;
+
+ do
+ {
+ mask = vp8_simple_filter_mask(blimit[0], s[-2*p], s[-1*p], s[0*p], s[1*p]);
+ vp8_simple_filter(mask, s - 2 * p, s - 1 * p, s, s + 1 * p);
+ ++s;
+ }
+ while (++i < 16);
+}
+
+void vp8_loop_filter_simple_vertical_edge_c
+(
+ unsigned char *s,
+ int p,
+ const unsigned char *blimit
+)
+{
+ signed char mask = 0;
+ int i = 0;
+
+ do
+ {
+ mask = vp8_simple_filter_mask(blimit[0], s[-2], s[-1], s[0], s[1]);
+ vp8_simple_filter(mask, s - 2, s - 1, s, s + 1);
+ s += p;
+ }
+ while (++i < 16);
+
+}
+
+/* Horizontal MB filtering */
+void vp8_loop_filter_mbh_c(unsigned char *y_ptr, unsigned char *u_ptr,
+ unsigned char *v_ptr, int y_stride, int uv_stride,
+ loop_filter_info *lfi)
+{
+ vp8_mbloop_filter_horizontal_edge_c(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_mbloop_filter_horizontal_edge_c(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_mbloop_filter_horizontal_edge_c(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+}
+
+/* Vertical MB Filtering */
+void vp8_loop_filter_mbv_c(unsigned char *y_ptr, unsigned char *u_ptr,
+ unsigned char *v_ptr, int y_stride, int uv_stride,
+ loop_filter_info *lfi)
+{
+ vp8_mbloop_filter_vertical_edge_c(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_mbloop_filter_vertical_edge_c(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_mbloop_filter_vertical_edge_c(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+}
+
+/* Horizontal B Filtering */
+void vp8_loop_filter_bh_c(unsigned char *y_ptr, unsigned char *u_ptr,
+ unsigned char *v_ptr, int y_stride, int uv_stride,
+ loop_filter_info *lfi)
+{
+ vp8_loop_filter_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_loop_filter_horizontal_edge_c(u_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_loop_filter_horizontal_edge_c(v_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+}
+
+void vp8_loop_filter_bhs_c(unsigned char *y_ptr, int y_stride,
+ const unsigned char *blimit)
+{
+ vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, blimit);
+ vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, blimit);
+ vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, blimit);
+}
+
+/* Vertical B Filtering */
+void vp8_loop_filter_bv_c(unsigned char *y_ptr, unsigned char *u_ptr,
+ unsigned char *v_ptr, int y_stride, int uv_stride,
+ loop_filter_info *lfi)
+{
+ vp8_loop_filter_vertical_edge_c(y_ptr + 4, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_vertical_edge_c(y_ptr + 8, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_vertical_edge_c(y_ptr + 12, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_loop_filter_vertical_edge_c(u_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_loop_filter_vertical_edge_c(v_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+}
+
+void vp8_loop_filter_bvs_c(unsigned char *y_ptr, int y_stride,
+ const unsigned char *blimit)
+{
+ vp8_loop_filter_simple_vertical_edge_c(y_ptr + 4, y_stride, blimit);
+ vp8_loop_filter_simple_vertical_edge_c(y_ptr + 8, y_stride, blimit);
+ vp8_loop_filter_simple_vertical_edge_c(y_ptr + 12, y_stride, blimit);
+}
diff --git a/libvpx/vp8/common/mbpitch.c b/libvpx/vp8/common/mbpitch.c
new file mode 100644
index 0000000..32e1b66
--- /dev/null
+++ b/libvpx/vp8/common/mbpitch.c
@@ -0,0 +1,68 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "blockd.h"
+
+void vp8_setup_block_dptrs(MACROBLOCKD *x)
+{
+ int r, c;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ x->block[r*4+c].predictor = x->predictor + r * 4 * 16 + c * 4;
+ }
+ }
+
+ for (r = 0; r < 2; r++)
+ {
+ for (c = 0; c < 2; c++)
+ {
+ x->block[16+r*2+c].predictor = x->predictor + 256 + r * 4 * 8 + c * 4;
+
+ }
+ }
+
+ for (r = 0; r < 2; r++)
+ {
+ for (c = 0; c < 2; c++)
+ {
+ x->block[20+r*2+c].predictor = x->predictor + 320 + r * 4 * 8 + c * 4;
+
+ }
+ }
+
+ for (r = 0; r < 25; r++)
+ {
+ x->block[r].qcoeff = x->qcoeff + r * 16;
+ x->block[r].dqcoeff = x->dqcoeff + r * 16;
+ x->block[r].eob = x->eobs + r;
+ }
+}
+
+void vp8_build_block_doffsets(MACROBLOCKD *x)
+{
+ int block;
+
+ for (block = 0; block < 16; block++) /* y blocks */
+ {
+ x->block[block].offset =
+ (block >> 2) * 4 * x->dst.y_stride + (block & 3) * 4;
+ }
+
+ for (block = 16; block < 20; block++) /* U and V blocks */
+ {
+ x->block[block+4].offset =
+ x->block[block].offset =
+ ((block - 16) >> 1) * 4 * x->dst.uv_stride + (block & 1) * 4;
+ }
+}
diff --git a/libvpx/vp8/common/mfqe.c b/libvpx/vp8/common/mfqe.c
new file mode 100644
index 0000000..3dff150
--- /dev/null
+++ b/libvpx/vp8/common/mfqe.c
@@ -0,0 +1,385 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/* MFQE: Multiframe Quality Enhancement
+ * In rate limited situations keyframes may cause significant visual artifacts
+ * commonly referred to as "popping." This file implements a postproccesing
+ * algorithm which blends data from the preceeding frame when there is no
+ * motion and the q from the previous frame is lower which indicates that it is
+ * higher quality.
+ */
+
+#include "postproc.h"
+#include "variance.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vpx_rtcd.h"
+#include "vpx_scale/yv12config.h"
+
+#include <limits.h>
+#include <stdlib.h>
+
+static void filter_by_weight(unsigned char *src, int src_stride,
+ unsigned char *dst, int dst_stride,
+ int block_size, int src_weight)
+{
+ int dst_weight = (1 << MFQE_PRECISION) - src_weight;
+ int rounding_bit = 1 << (MFQE_PRECISION - 1);
+ int r, c;
+
+ for (r = 0; r < block_size; r++)
+ {
+ for (c = 0; c < block_size; c++)
+ {
+ dst[c] = (src[c] * src_weight +
+ dst[c] * dst_weight +
+ rounding_bit) >> MFQE_PRECISION;
+ }
+ src += src_stride;
+ dst += dst_stride;
+ }
+}
+
+void vp8_filter_by_weight16x16_c(unsigned char *src, int src_stride,
+ unsigned char *dst, int dst_stride,
+ int src_weight)
+{
+ filter_by_weight(src, src_stride, dst, dst_stride, 16, src_weight);
+}
+
+void vp8_filter_by_weight8x8_c(unsigned char *src, int src_stride,
+ unsigned char *dst, int dst_stride,
+ int src_weight)
+{
+ filter_by_weight(src, src_stride, dst, dst_stride, 8, src_weight);
+}
+
+void vp8_filter_by_weight4x4_c(unsigned char *src, int src_stride,
+ unsigned char *dst, int dst_stride,
+ int src_weight)
+{
+ filter_by_weight(src, src_stride, dst, dst_stride, 4, src_weight);
+}
+
+static void apply_ifactor(unsigned char *y_src,
+ int y_src_stride,
+ unsigned char *y_dst,
+ int y_dst_stride,
+ unsigned char *u_src,
+ unsigned char *v_src,
+ int uv_src_stride,
+ unsigned char *u_dst,
+ unsigned char *v_dst,
+ int uv_dst_stride,
+ int block_size,
+ int src_weight)
+{
+ if (block_size == 16)
+ {
+ vp8_filter_by_weight16x16(y_src, y_src_stride, y_dst, y_dst_stride, src_weight);
+ vp8_filter_by_weight8x8(u_src, uv_src_stride, u_dst, uv_dst_stride, src_weight);
+ vp8_filter_by_weight8x8(v_src, uv_src_stride, v_dst, uv_dst_stride, src_weight);
+ }
+ else /* if (block_size == 8) */
+ {
+ vp8_filter_by_weight8x8(y_src, y_src_stride, y_dst, y_dst_stride, src_weight);
+ vp8_filter_by_weight4x4(u_src, uv_src_stride, u_dst, uv_dst_stride, src_weight);
+ vp8_filter_by_weight4x4(v_src, uv_src_stride, v_dst, uv_dst_stride, src_weight);
+ }
+}
+
+static unsigned int int_sqrt(unsigned int x)
+{
+ unsigned int y = x;
+ unsigned int guess;
+ int p = 1;
+ while (y>>=1) p++;
+ p>>=1;
+
+ guess=0;
+ while (p>=0)
+ {
+ guess |= (1<<p);
+ if (x<guess*guess)
+ guess -= (1<<p);
+ p--;
+ }
+ /* choose between guess or guess+1 */
+ return guess+(guess*guess+guess+1<=x);
+}
+
+#define USE_SSD
+static void multiframe_quality_enhance_block
+(
+ int blksize, /* Currently only values supported are 16, 8 */
+ int qcurr,
+ int qprev,
+ unsigned char *y,
+ unsigned char *u,
+ unsigned char *v,
+ int y_stride,
+ int uv_stride,
+ unsigned char *yd,
+ unsigned char *ud,
+ unsigned char *vd,
+ int yd_stride,
+ int uvd_stride
+)
+{
+ static const unsigned char VP8_ZEROS[16]=
+ {
+ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+ };
+ int uvblksize = blksize >> 1;
+ int qdiff = qcurr - qprev;
+
+ int i;
+ unsigned char *up;
+ unsigned char *udp;
+ unsigned char *vp;
+ unsigned char *vdp;
+
+ unsigned int act, actd, sad, usad, vsad, sse, thr, thrsq, actrisk;
+
+ if (blksize == 16)
+ {
+ actd = (vp8_variance16x16(yd, yd_stride, VP8_ZEROS, 0, &sse)+128)>>8;
+ act = (vp8_variance16x16(y, y_stride, VP8_ZEROS, 0, &sse)+128)>>8;
+#ifdef USE_SSD
+ sad = (vp8_variance16x16(y, y_stride, yd, yd_stride, &sse));
+ sad = (sse + 128)>>8;
+ usad = (vp8_variance8x8(u, uv_stride, ud, uvd_stride, &sse));
+ usad = (sse + 32)>>6;
+ vsad = (vp8_variance8x8(v, uv_stride, vd, uvd_stride, &sse));
+ vsad = (sse + 32)>>6;
+#else
+ sad = (vp8_sad16x16(y, y_stride, yd, yd_stride, UINT_MAX) + 128) >> 8;
+ usad = (vp8_sad8x8(u, uv_stride, ud, uvd_stride, UINT_MAX) + 32) >> 6;
+ vsad = (vp8_sad8x8(v, uv_stride, vd, uvd_stride, UINT_MAX)+ 32) >> 6;
+#endif
+ }
+ else /* if (blksize == 8) */
+ {
+ actd = (vp8_variance8x8(yd, yd_stride, VP8_ZEROS, 0, &sse)+32)>>6;
+ act = (vp8_variance8x8(y, y_stride, VP8_ZEROS, 0, &sse)+32)>>6;
+#ifdef USE_SSD
+ sad = (vp8_variance8x8(y, y_stride, yd, yd_stride, &sse));
+ sad = (sse + 32)>>6;
+ usad = (vp8_variance4x4(u, uv_stride, ud, uvd_stride, &sse));
+ usad = (sse + 8)>>4;
+ vsad = (vp8_variance4x4(v, uv_stride, vd, uvd_stride, &sse));
+ vsad = (sse + 8)>>4;
+#else
+ sad = (vp8_sad8x8(y, y_stride, yd, yd_stride, UINT_MAX) + 32) >> 6;
+ usad = (vp8_sad4x4(u, uv_stride, ud, uvd_stride, UINT_MAX) + 8) >> 4;
+ vsad = (vp8_sad4x4(v, uv_stride, vd, uvd_stride, UINT_MAX) + 8) >> 4;
+#endif
+ }
+
+ actrisk = (actd > act * 5);
+
+ /* thr = qdiff/16 + log2(act) + log4(qprev) */
+ thr = (qdiff >> 4);
+ while (actd >>= 1) thr++;
+ while (qprev >>= 2) thr++;
+
+#ifdef USE_SSD
+ thrsq = thr * thr;
+ if (sad < thrsq &&
+ /* additional checks for color mismatch and excessive addition of
+ * high-frequencies */
+ 4 * usad < thrsq && 4 * vsad < thrsq && !actrisk)
+#else
+ if (sad < thr &&
+ /* additional checks for color mismatch and excessive addition of
+ * high-frequencies */
+ 2 * usad < thr && 2 * vsad < thr && !actrisk)
+#endif
+ {
+ int ifactor;
+#ifdef USE_SSD
+ /* TODO: optimize this later to not need sqr root */
+ sad = int_sqrt(sad);
+#endif
+ ifactor = (sad << MFQE_PRECISION) / thr;
+ ifactor >>= (qdiff >> 5);
+
+ if (ifactor)
+ {
+ apply_ifactor(y, y_stride, yd, yd_stride,
+ u, v, uv_stride,
+ ud, vd, uvd_stride,
+ blksize, ifactor);
+ }
+ }
+ else /* else implicitly copy from previous frame */
+ {
+ if (blksize == 16)
+ {
+ vp8_copy_mem16x16(y, y_stride, yd, yd_stride);
+ vp8_copy_mem8x8(u, uv_stride, ud, uvd_stride);
+ vp8_copy_mem8x8(v, uv_stride, vd, uvd_stride);
+ }
+ else /* if (blksize == 8) */
+ {
+ vp8_copy_mem8x8(y, y_stride, yd, yd_stride);
+ for (up = u, udp = ud, i = 0; i < uvblksize; ++i, up += uv_stride, udp += uvd_stride)
+ vpx_memcpy(udp, up, uvblksize);
+ for (vp = v, vdp = vd, i = 0; i < uvblksize; ++i, vp += uv_stride, vdp += uvd_stride)
+ vpx_memcpy(vdp, vp, uvblksize);
+ }
+ }
+}
+
+static int qualify_inter_mb(const MODE_INFO *mode_info_context, int *map)
+{
+ if (mode_info_context->mbmi.mb_skip_coeff)
+ map[0] = map[1] = map[2] = map[3] = 1;
+ else if (mode_info_context->mbmi.mode==SPLITMV)
+ {
+ static int ndx[4][4] =
+ {
+ {0, 1, 4, 5},
+ {2, 3, 6, 7},
+ {8, 9, 12, 13},
+ {10, 11, 14, 15}
+ };
+ int i, j;
+ for (i=0; i<4; ++i)
+ {
+ map[i] = 1;
+ for (j=0; j<4 && map[j]; ++j)
+ map[i] &= (mode_info_context->bmi[ndx[i][j]].mv.as_mv.row <= 2 &&
+ mode_info_context->bmi[ndx[i][j]].mv.as_mv.col <= 2);
+ }
+ }
+ else
+ {
+ map[0] = map[1] = map[2] = map[3] =
+ (mode_info_context->mbmi.mode > B_PRED &&
+ abs(mode_info_context->mbmi.mv.as_mv.row) <= 2 &&
+ abs(mode_info_context->mbmi.mv.as_mv.col) <= 2);
+ }
+ return (map[0]+map[1]+map[2]+map[3]);
+}
+
+void vp8_multiframe_quality_enhance
+(
+ VP8_COMMON *cm
+)
+{
+ YV12_BUFFER_CONFIG *show = cm->frame_to_show;
+ YV12_BUFFER_CONFIG *dest = &cm->post_proc_buffer;
+
+ FRAME_TYPE frame_type = cm->frame_type;
+ /* Point at base of Mb MODE_INFO list has motion vectors etc */
+ const MODE_INFO *mode_info_context = cm->mi;
+ int mb_row;
+ int mb_col;
+ int totmap, map[4];
+ int qcurr = cm->base_qindex;
+ int qprev = cm->postproc_state.last_base_qindex;
+
+ unsigned char *y_ptr, *u_ptr, *v_ptr;
+ unsigned char *yd_ptr, *ud_ptr, *vd_ptr;
+
+ /* Set up the buffer pointers */
+ y_ptr = show->y_buffer;
+ u_ptr = show->u_buffer;
+ v_ptr = show->v_buffer;
+ yd_ptr = dest->y_buffer;
+ ud_ptr = dest->u_buffer;
+ vd_ptr = dest->v_buffer;
+
+ /* postprocess each macro block */
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+ /* if motion is high there will likely be no benefit */
+ if (frame_type == INTER_FRAME) totmap = qualify_inter_mb(mode_info_context, map);
+ else totmap = (frame_type == KEY_FRAME ? 4 : 0);
+ if (totmap)
+ {
+ if (totmap < 4)
+ {
+ int i, j;
+ for (i=0; i<2; ++i)
+ for (j=0; j<2; ++j)
+ {
+ if (map[i*2+j])
+ {
+ multiframe_quality_enhance_block(8, qcurr, qprev,
+ y_ptr + 8*(i*show->y_stride+j),
+ u_ptr + 4*(i*show->uv_stride+j),
+ v_ptr + 4*(i*show->uv_stride+j),
+ show->y_stride,
+ show->uv_stride,
+ yd_ptr + 8*(i*dest->y_stride+j),
+ ud_ptr + 4*(i*dest->uv_stride+j),
+ vd_ptr + 4*(i*dest->uv_stride+j),
+ dest->y_stride,
+ dest->uv_stride);
+ }
+ else
+ {
+ /* copy a 8x8 block */
+ int k;
+ unsigned char *up = u_ptr + 4*(i*show->uv_stride+j);
+ unsigned char *udp = ud_ptr + 4*(i*dest->uv_stride+j);
+ unsigned char *vp = v_ptr + 4*(i*show->uv_stride+j);
+ unsigned char *vdp = vd_ptr + 4*(i*dest->uv_stride+j);
+ vp8_copy_mem8x8(y_ptr + 8*(i*show->y_stride+j), show->y_stride,
+ yd_ptr + 8*(i*dest->y_stride+j), dest->y_stride);
+ for (k = 0; k < 4; ++k, up += show->uv_stride, udp += dest->uv_stride,
+ vp += show->uv_stride, vdp += dest->uv_stride)
+ {
+ vpx_memcpy(udp, up, 4);
+ vpx_memcpy(vdp, vp, 4);
+ }
+ }
+ }
+ }
+ else /* totmap = 4 */
+ {
+ multiframe_quality_enhance_block(16, qcurr, qprev, y_ptr,
+ u_ptr, v_ptr,
+ show->y_stride,
+ show->uv_stride,
+ yd_ptr, ud_ptr, vd_ptr,
+ dest->y_stride,
+ dest->uv_stride);
+ }
+ }
+ else
+ {
+ vp8_copy_mem16x16(y_ptr, show->y_stride, yd_ptr, dest->y_stride);
+ vp8_copy_mem8x8(u_ptr, show->uv_stride, ud_ptr, dest->uv_stride);
+ vp8_copy_mem8x8(v_ptr, show->uv_stride, vd_ptr, dest->uv_stride);
+ }
+ y_ptr += 16;
+ u_ptr += 8;
+ v_ptr += 8;
+ yd_ptr += 16;
+ ud_ptr += 8;
+ vd_ptr += 8;
+ mode_info_context++; /* step to next MB */
+ }
+
+ y_ptr += show->y_stride * 16 - 16 * cm->mb_cols;
+ u_ptr += show->uv_stride * 8 - 8 * cm->mb_cols;
+ v_ptr += show->uv_stride * 8 - 8 * cm->mb_cols;
+ yd_ptr += dest->y_stride * 16 - 16 * cm->mb_cols;
+ ud_ptr += dest->uv_stride * 8 - 8 * cm->mb_cols;
+ vd_ptr += dest->uv_stride * 8 - 8 * cm->mb_cols;
+
+ mode_info_context++; /* Skip border mb */
+ }
+}
diff --git a/libvpx/vp8/common/mips/dspr2/dequantize_dspr2.c b/libvpx/vp8/common/mips/dspr2/dequantize_dspr2.c
new file mode 100644
index 0000000..6823325
--- /dev/null
+++ b/libvpx/vp8/common/mips/dspr2/dequantize_dspr2.c
@@ -0,0 +1,33 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx_mem/vpx_mem.h"
+
+#if HAVE_DSPR2
+void vp8_dequant_idct_add_dspr2(short *input, short *dq,
+ unsigned char *dest, int stride)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ input[i] = dq[i] * input[i];
+ }
+
+ vp8_short_idct4x4llm_dspr2(input, dest, stride, dest, stride);
+
+ vpx_memset(input, 0, 32);
+
+}
+
+#endif
diff --git a/libvpx/vp8/common/mips/dspr2/filter_dspr2.c b/libvpx/vp8/common/mips/dspr2/filter_dspr2.c
new file mode 100644
index 0000000..71fdcd7
--- /dev/null
+++ b/libvpx/vp8/common/mips/dspr2/filter_dspr2.c
@@ -0,0 +1,2823 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdlib.h>
+#include "vpx_rtcd.h"
+#include "vpx_ports/mem.h"
+
+#if HAVE_DSPR2
+#define CROP_WIDTH 256
+unsigned char ff_cropTbl[256 + 2 * CROP_WIDTH];
+
+static const unsigned short sub_pel_filterss[8][3] =
+{
+ { 0, 0, 0},
+ { 0, 0x0601, 0x7b0c},
+ { 0x0201, 0x0b08, 0x6c24},
+ { 0, 0x0906, 0x5d32},
+ { 0x0303, 0x1010, 0x4d4d},
+ { 0, 0x0609, 0x325d},
+ { 0x0102, 0x080b, 0x246c},
+ { 0, 0x0106, 0x0c7b},
+};
+
+
+static const int sub_pel_filters_int[8][3] =
+{
+ { 0, 0, 0},
+ { 0x0000fffa, 0x007b000c, 0xffff0000},
+ { 0x0002fff5, 0x006c0024, 0xfff80001},
+ { 0x0000fff7, 0x005d0032, 0xfffa0000},
+ { 0x0003fff0, 0x004d004d, 0xfff00003},
+ { 0x0000fffa, 0x0032005d, 0xfff70000},
+ { 0x0001fff8, 0x0024006c, 0xfff50002},
+ { 0x0000ffff, 0x000c007b, 0xfffa0000},
+};
+
+
+static const int sub_pel_filters_inv[8][3] =
+{
+ { 0, 0, 0},
+ { 0xfffa0000, 0x000c007b, 0x0000ffff},
+ { 0xfff50002, 0x0024006c, 0x0001fff8},
+ { 0xfff70000, 0x0032005d, 0x0000fffa},
+ { 0xfff00003, 0x004d004d, 0x0003fff0},
+ { 0xfffa0000, 0x005d0032, 0x0000fff7},
+ { 0xfff80001, 0x006c0024, 0x0002fff5},
+ { 0xffff0000, 0x007b000c, 0x0000fffa},
+};
+
+
+static const int sub_pel_filters_int_tap_4[8][2] =
+{
+ { 0, 0},
+ { 0xfffa007b, 0x000cffff},
+ { 0, 0},
+ { 0xfff7005d, 0x0032fffa},
+ { 0, 0},
+ { 0xfffa0032, 0x005dfff7},
+ { 0, 0},
+ { 0xffff000c, 0x007bfffa},
+};
+
+
+static const int sub_pel_filters_inv_tap_4[8][2] =
+{
+ { 0, 0},
+ { 0x007bfffa, 0xffff000c},
+ { 0, 0},
+ { 0x005dfff7, 0xfffa0032},
+ { 0, 0},
+ { 0x0032fffa, 0xfff7005d},
+ { 0, 0},
+ { 0x000cffff, 0xfffa007b},
+};
+
+inline void prefetch_load(unsigned char *src)
+{
+ __asm__ __volatile__ (
+ "pref 0, 0(%[src]) \n\t"
+ :
+ : [src] "r" (src)
+ );
+}
+
+
+inline void prefetch_store(unsigned char *dst)
+{
+ __asm__ __volatile__ (
+ "pref 1, 0(%[dst]) \n\t"
+ :
+ : [dst] "r" (dst)
+ );
+}
+
+void dsputil_static_init(void)
+{
+ int i;
+
+ for (i = 0; i < 256; i++) ff_cropTbl[i + CROP_WIDTH] = i;
+
+ for (i = 0; i < CROP_WIDTH; i++)
+ {
+ ff_cropTbl[i] = 0;
+ ff_cropTbl[i + CROP_WIDTH + 256] = 255;
+ }
+}
+
+void vp8_filter_block2d_first_pass_4
+(
+ unsigned char *RESTRICT src_ptr,
+ unsigned char *RESTRICT dst_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int output_height,
+ int xoffset,
+ int pitch
+)
+{
+ unsigned int i;
+ int Temp1, Temp2, Temp3, Temp4;
+
+ unsigned int vector4a = 64;
+ int vector1b, vector2b, vector3b;
+ unsigned int tp1, tp2, tn1, tn2;
+ unsigned int p1, p2, p3;
+ unsigned int n1, n2, n3;
+ unsigned char *cm = ff_cropTbl + CROP_WIDTH;
+
+ vector3b = sub_pel_filters_inv[xoffset][2];
+
+ /* if (xoffset == 0) we don't need any filtering */
+ if (vector3b == 0)
+ {
+ for (i = 0; i < output_height; i++)
+ {
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + src_pixels_per_line);
+ dst_ptr[0] = src_ptr[0];
+ dst_ptr[1] = src_ptr[1];
+ dst_ptr[2] = src_ptr[2];
+ dst_ptr[3] = src_ptr[3];
+
+ /* next row... */
+ src_ptr += src_pixels_per_line;
+ dst_ptr += 4;
+ }
+ }
+ else
+ {
+ if (vector3b > 65536)
+ {
+ /* 6 tap filter */
+
+ vector1b = sub_pel_filters_inv[xoffset][0];
+ vector2b = sub_pel_filters_inv[xoffset][1];
+
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + src_pixels_per_line);
+
+ for (i = output_height; i--;)
+ {
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "ulw %[tp1], -2(%[src_ptr]) \n\t"
+ "ulw %[tp2], 2(%[src_ptr]) \n\t"
+
+ /* even 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p1], %[tp1] \n\t"
+ "preceu.ph.qbl %[p2], %[tp1] \n\t"
+ "preceu.ph.qbr %[p3], %[tp2] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[p3], %[vector3b] \n\t"
+
+ /* even 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[p1], %[tp2] \n\t"
+ "balign %[tp2], %[tp1], 3 \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[p1], %[vector3b] \n\t"
+
+ /* odd 1. pixel */
+ "ulw %[tn2], 3(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n1], %[tp2] \n\t"
+ "preceu.ph.qbl %[n2], %[tp2] \n\t"
+ "preceu.ph.qbr %[n3], %[tn2] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[n3], %[vector3b] \n\t"
+
+ /* even 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[n1], %[tn2] \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[n1], %[vector3b] \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ /* clamp */
+ "lbux %[tp1], %[Temp1](%[cm]) \n\t"
+ "lbux %[tn1], %[Temp2](%[cm]) \n\t"
+ "lbux %[tp2], %[Temp3](%[cm]) \n\t"
+ "lbux %[n2], %[Temp4](%[cm]) \n\t"
+
+ /* store bytes */
+ "sb %[tp1], 0(%[dst_ptr]) \n\t"
+ "sb %[tn1], 1(%[dst_ptr]) \n\t"
+ "sb %[tp2], 2(%[dst_ptr]) \n\t"
+ "sb %[n2], 3(%[dst_ptr]) \n\t"
+
+ : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn1] "=&r" (tn1),
+ [tn2] "=&r" (tn2), [p1] "=&r" (p1), [p2] "=&r" (p2),
+ [p3] "=&r" (p3), [n1] "=&r" (n1), [n2] "=&r" (n2),
+ [n3] "=&r" (n3), [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector4a] "r" (vector4a), [cm] "r" (cm), [dst_ptr] "r" (dst_ptr),
+ [vector3b] "r" (vector3b), [src_ptr] "r" (src_ptr)
+ );
+
+ /* Next row... */
+ src_ptr += src_pixels_per_line;
+ dst_ptr += pitch;
+ }
+ }
+ else
+ {
+ /* 4 tap filter */
+
+ vector1b = sub_pel_filters_inv_tap_4[xoffset][0];
+ vector2b = sub_pel_filters_inv_tap_4[xoffset][1];
+
+ for (i = output_height; i--;)
+ {
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "ulw %[tp1], -1(%[src_ptr]) \n\t"
+ "ulw %[tp2], 3(%[src_ptr]) \n\t"
+
+ /* even 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p1], %[tp1] \n\t"
+ "preceu.ph.qbl %[p2], %[tp1] \n\t"
+ "preceu.ph.qbr %[p3], %[tp2] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
+
+ /* even 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+
+ /* odd 1. pixel */
+ "srl %[tn1], %[tp2], 8 \n\t"
+ "balign %[tp2], %[tp1], 3 \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n1], %[tp2] \n\t"
+ "preceu.ph.qbl %[n2], %[tp2] \n\t"
+ "preceu.ph.qbr %[n3], %[tn1] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
+
+ /* odd 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ /* clamp and store results */
+ "lbux %[tp1], %[Temp1](%[cm]) \n\t"
+ "lbux %[tn1], %[Temp2](%[cm]) \n\t"
+ "lbux %[tp2], %[Temp3](%[cm]) \n\t"
+ "sb %[tp1], 0(%[dst_ptr]) \n\t"
+ "sb %[tn1], 1(%[dst_ptr]) \n\t"
+ "lbux %[n2], %[Temp4](%[cm]) \n\t"
+ "sb %[tp2], 2(%[dst_ptr]) \n\t"
+ "sb %[n2], 3(%[dst_ptr]) \n\t"
+
+ : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn1] "=&r" (tn1),
+ [p1] "=&r" (p1), [p2] "=&r" (p2), [p3] "=&r" (p3),
+ [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector4a] "r" (vector4a), [cm] "r" (cm), [dst_ptr] "r" (dst_ptr),
+ [src_ptr] "r" (src_ptr)
+ );
+ /* Next row... */
+ src_ptr += src_pixels_per_line;
+ dst_ptr += pitch;
+ }
+ }
+ }
+}
+
+void vp8_filter_block2d_first_pass_8_all
+(
+ unsigned char *RESTRICT src_ptr,
+ unsigned char *RESTRICT dst_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int output_height,
+ int xoffset,
+ int pitch
+)
+{
+ unsigned int i;
+ int Temp1, Temp2, Temp3, Temp4;
+
+ unsigned int vector4a = 64;
+ unsigned int vector1b, vector2b, vector3b;
+ unsigned int tp1, tp2, tn1, tn2;
+ unsigned int p1, p2, p3, p4;
+ unsigned int n1, n2, n3, n4;
+
+ unsigned char *cm = ff_cropTbl + CROP_WIDTH;
+
+ /* if (xoffset == 0) we don't need any filtering */
+ if (xoffset == 0)
+ {
+ for (i = 0; i < output_height; i++)
+ {
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + src_pixels_per_line);
+
+ dst_ptr[0] = src_ptr[0];
+ dst_ptr[1] = src_ptr[1];
+ dst_ptr[2] = src_ptr[2];
+ dst_ptr[3] = src_ptr[3];
+ dst_ptr[4] = src_ptr[4];
+ dst_ptr[5] = src_ptr[5];
+ dst_ptr[6] = src_ptr[6];
+ dst_ptr[7] = src_ptr[7];
+
+ /* next row... */
+ src_ptr += src_pixels_per_line;
+ dst_ptr += 8;
+ }
+ }
+ else
+ {
+ vector3b = sub_pel_filters_inv[xoffset][2];
+
+ if (vector3b > 65536)
+ {
+ /* 6 tap filter */
+
+ vector1b = sub_pel_filters_inv[xoffset][0];
+ vector2b = sub_pel_filters_inv[xoffset][1];
+
+ for (i = output_height; i--;)
+ {
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + src_pixels_per_line);
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "ulw %[tp1], -2(%[src_ptr]) \n\t"
+ "ulw %[tp2], 2(%[src_ptr]) \n\t"
+
+ /* even 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p1], %[tp1] \n\t"
+ "preceu.ph.qbl %[p2], %[tp1] \n\t"
+ "preceu.ph.qbr %[p3], %[tp2] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[p3], %[vector3b] \n\t"
+
+ /* even 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[p1], %[tp2] \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[p1], %[vector3b] \n\t"
+
+ "balign %[tp2], %[tp1], 3 \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+ "ulw %[tn2], 3(%[src_ptr]) \n\t"
+
+ /* odd 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n1], %[tp2] \n\t"
+ "preceu.ph.qbl %[n2], %[tp2] \n\t"
+ "preceu.ph.qbr %[n3], %[tn2] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[n3], %[vector3b] \n\t"
+
+ /* odd 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[n1], %[tn2] \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[n1], %[vector3b] \n\t"
+ "ulw %[tp1], 6(%[src_ptr]) \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p2], %[tp1] \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn2] "=&r" (tn2),
+ [p1] "=&r" (p1), [p2] "=&r" (p2), [p3] "=&r" (p3),
+ [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector4a] "r" (vector4a), [vector3b] "r" (vector3b),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ dst_ptr[0] = cm[Temp1];
+ dst_ptr[1] = cm[Temp2];
+ dst_ptr[2] = cm[Temp3];
+ dst_ptr[3] = cm[Temp4];
+
+ /* next 4 pixels */
+ __asm__ __volatile__ (
+ /* even 3. pixel */
+ "dpa.w.ph $ac3, %[p3], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector3b] \n\t"
+
+ /* even 4. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[p4], %[tp1] \n\t"
+ "dpa.w.ph $ac2, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[p4], %[vector3b] \n\t"
+
+ "ulw %[tn1], 7(%[src_ptr]) \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+
+ /* odd 3. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n2], %[tn1] \n\t"
+ "dpa.w.ph $ac3, %[n3], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector3b] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+
+ /* odd 4. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[n4], %[tn1] \n\t"
+ "dpa.w.ph $ac2, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[n4], %[vector3b] \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ : [tn1] "=&r" (tn1), [n2] "=&r" (n2),
+ [p4] "=&r" (p4), [n4] "=&r" (n4),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
+ : [tp1] "r" (tp1), [vector1b] "r" (vector1b), [p2] "r" (p2),
+ [vector2b] "r" (vector2b), [n1] "r" (n1), [p1] "r" (p1),
+ [vector4a] "r" (vector4a), [vector3b] "r" (vector3b),
+ [p3] "r" (p3), [n3] "r" (n3), [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ dst_ptr[4] = cm[Temp1];
+ dst_ptr[5] = cm[Temp2];
+ dst_ptr[6] = cm[Temp3];
+ dst_ptr[7] = cm[Temp4];
+
+ src_ptr += src_pixels_per_line;
+ dst_ptr += pitch;
+ }
+ }
+ else
+ {
+ /* 4 tap filter */
+
+ vector1b = sub_pel_filters_inv_tap_4[xoffset][0];
+ vector2b = sub_pel_filters_inv_tap_4[xoffset][1];
+
+ for (i = output_height; i--;)
+ {
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + src_pixels_per_line);
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "ulw %[tp1], -1(%[src_ptr]) \n\t"
+
+ /* even 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p1], %[tp1] \n\t"
+ "preceu.ph.qbl %[p2], %[tp1] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
+
+ "ulw %[tp2], 3(%[src_ptr]) \n\t"
+
+ /* even 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbr %[p3], %[tp2] \n\t"
+ "preceu.ph.qbl %[p4], %[tp2] \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+
+ "balign %[tp2], %[tp1], 3 \n\t"
+
+ /* odd 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n1], %[tp2] \n\t"
+ "preceu.ph.qbl %[n2], %[tp2] \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+
+ "ulw %[tn2], 4(%[src_ptr]) \n\t"
+
+ /* odd 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbr %[n3], %[tn2] \n\t"
+ "preceu.ph.qbl %[n4], %[tn2] \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
+ "ulw %[tp1], 7(%[src_ptr]) \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2),
+ [tn2] "=&r" (tn2), [p1] "=&r" (p1), [p2] "=&r" (p2),
+ [p3] "=&r" (p3), [p4] "=&r" (p4), [n1] "=&r" (n1),
+ [n2] "=&r" (n2), [n3] "=&r" (n3), [n4] "=&r" (n4),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ dst_ptr[0] = cm[Temp1];
+ dst_ptr[1] = cm[Temp2];
+ dst_ptr[2] = cm[Temp3];
+ dst_ptr[3] = cm[Temp4];
+
+ /* next 4 pixels */
+ __asm__ __volatile__ (
+ /* even 3. pixel */
+ "dpa.w.ph $ac3, %[p3], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p4], %[vector2b] \n\t"
+
+ /* even 4. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbr %[p2], %[tp1] \n\t"
+ "dpa.w.ph $ac2, %[p4], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector2b] \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+
+ /* odd 3. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "dpa.w.ph $ac3, %[n3], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n4], %[vector2b] \n\t"
+ "ulw %[tn1], 8(%[src_ptr]) \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+
+ /* odd 4. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbr %[n2], %[tn1] \n\t"
+ "dpa.w.ph $ac2, %[n4], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector2b] \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ : [tn1] "=&r" (tn1), [p2] "=&r" (p2), [n2] "=&r" (n2),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
+ : [tp1] "r" (tp1), [p3] "r" (p3), [p4] "r" (p4),
+ [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr),
+ [n3] "r" (n3), [n4] "r" (n4)
+ );
+
+ /* clamp and store results */
+ dst_ptr[4] = cm[Temp1];
+ dst_ptr[5] = cm[Temp2];
+ dst_ptr[6] = cm[Temp3];
+ dst_ptr[7] = cm[Temp4];
+
+ /* next row... */
+ src_ptr += src_pixels_per_line;
+ dst_ptr += pitch;
+ }
+ }
+ }
+}
+
+
+void vp8_filter_block2d_first_pass16_6tap
+(
+ unsigned char *RESTRICT src_ptr,
+ unsigned char *RESTRICT dst_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int output_height,
+ int xoffset,
+ int pitch
+)
+{
+ unsigned int i;
+ int Temp1, Temp2, Temp3, Temp4;
+
+ unsigned int vector4a;
+ unsigned int vector1b, vector2b, vector3b;
+ unsigned int tp1, tp2, tn1, tn2;
+ unsigned int p1, p2, p3, p4;
+ unsigned int n1, n2, n3, n4;
+ unsigned char *cm = ff_cropTbl + CROP_WIDTH;
+
+ vector1b = sub_pel_filters_inv[xoffset][0];
+ vector2b = sub_pel_filters_inv[xoffset][1];
+ vector3b = sub_pel_filters_inv[xoffset][2];
+ vector4a = 64;
+
+ for (i = output_height; i--;)
+ {
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + src_pixels_per_line);
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "ulw %[tp1], -2(%[src_ptr]) \n\t"
+ "ulw %[tp2], 2(%[src_ptr]) \n\t"
+
+ /* even 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p1], %[tp1] \n\t"
+ "preceu.ph.qbl %[p2], %[tp1] \n\t"
+ "preceu.ph.qbr %[p3], %[tp2] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[p3], %[vector3b] \n\t"
+
+ /* even 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[p1], %[tp2] \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[p1], %[vector3b] \n\t"
+
+ "balign %[tp2], %[tp1], 3 \n\t"
+ "ulw %[tn2], 3(%[src_ptr]) \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+
+ /* odd 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n1], %[tp2] \n\t"
+ "preceu.ph.qbl %[n2], %[tp2] \n\t"
+ "preceu.ph.qbr %[n3], %[tn2] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[n3], %[vector3b] \n\t"
+
+ /* odd 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[n1], %[tn2] \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[n1], %[vector3b] \n\t"
+ "ulw %[tp1], 6(%[src_ptr]) \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p2], %[tp1] \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn2] "=&r" (tn2),
+ [p1] "=&r" (p1), [p2] "=&r" (p2), [p3] "=&r" (p3),
+ [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector4a] "r" (vector4a), [vector3b] "r" (vector3b),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ dst_ptr[0] = cm[Temp1];
+ dst_ptr[1] = cm[Temp2];
+ dst_ptr[2] = cm[Temp3];
+ dst_ptr[3] = cm[Temp4];
+
+ /* next 4 pixels */
+ __asm__ __volatile__ (
+ /* even 3. pixel */
+ "dpa.w.ph $ac3, %[p3], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector3b] \n\t"
+
+ /* even 4. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[p4], %[tp1] \n\t"
+ "dpa.w.ph $ac2, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[p4], %[vector3b] \n\t"
+ "ulw %[tn1], 7(%[src_ptr]) \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+
+ /* odd 3. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n2], %[tn1] \n\t"
+ "dpa.w.ph $ac3, %[n3], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector3b] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+
+ /* odd 4. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[n4], %[tn1] \n\t"
+ "dpa.w.ph $ac2, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[n4], %[vector3b] \n\t"
+ "ulw %[tp2], 10(%[src_ptr]) \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p1], %[tp2] \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ : [tn1] "=&r" (tn1), [tp2] "=&r" (tp2), [n2] "=&r" (n2),
+ [p4] "=&r" (p4), [n4] "=&r" (n4),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [tp1] "r" (tp1), [n1] "r" (n1), [p1] "r" (p1),
+ [vector4a] "r" (vector4a), [p2] "r" (p2), [vector3b] "r" (vector3b),
+ [p3] "r" (p3), [n3] "r" (n3), [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ dst_ptr[4] = cm[Temp1];
+ dst_ptr[5] = cm[Temp2];
+ dst_ptr[6] = cm[Temp3];
+ dst_ptr[7] = cm[Temp4];
+
+ /* next 4 pixels */
+ __asm__ __volatile__ (
+ /* even 5. pixel */
+ "dpa.w.ph $ac3, %[p2], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p4], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector3b] \n\t"
+
+ /* even 6. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[p3], %[tp2] \n\t"
+ "dpa.w.ph $ac2, %[p4], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p1], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector3b] \n\t"
+
+ "ulw %[tn1], 11(%[src_ptr]) \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+
+ /* odd 5. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n1], %[tn1] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n4], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector3b] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+
+ /* odd 6. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[n3], %[tn1] \n\t"
+ "dpa.w.ph $ac2, %[n4], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n1], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector3b] \n\t"
+ "ulw %[tp1], 14(%[src_ptr]) \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[p4], %[tp1] \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ : [tn1] "=&r" (tn1), [tp1] "=&r" (tp1),
+ [n1] "=&r" (n1), [p3] "=&r" (p3), [n3] "=&r" (n3),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [tp2] "r" (tp2), [p2] "r" (p2), [n2] "r" (n2),
+ [p4] "r" (p4), [n4] "r" (n4), [p1] "r" (p1), [src_ptr] "r" (src_ptr),
+ [vector4a] "r" (vector4a), [vector3b] "r" (vector3b)
+ );
+
+ /* clamp and store results */
+ dst_ptr[8] = cm[Temp1];
+ dst_ptr[9] = cm[Temp2];
+ dst_ptr[10] = cm[Temp3];
+ dst_ptr[11] = cm[Temp4];
+
+ /* next 4 pixels */
+ __asm__ __volatile__ (
+ /* even 7. pixel */
+ "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p3], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[p4], %[vector3b] \n\t"
+
+ /* even 8. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[p2], %[tp1] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p4], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector3b] \n\t"
+ "ulw %[tn1], 15(%[src_ptr]) \n\t"
+ "extp %[Temp1], $ac3, 9 \n\t"
+
+ /* odd 7. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "preceu.ph.qbr %[n4], %[tn1] \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n3], %[vector2b] \n\t"
+ "dpa.w.ph $ac3, %[n4], %[vector3b] \n\t"
+ "extp %[Temp3], $ac2, 9 \n\t"
+
+ /* odd 8. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "preceu.ph.qbl %[n2], %[tn1] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n4], %[vector2b] \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector3b] \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+ "extp %[Temp4], $ac2, 9 \n\t"
+
+ /* clamp and store results */
+ "lbux %[tp1], %[Temp1](%[cm]) \n\t"
+ "lbux %[tn1], %[Temp2](%[cm]) \n\t"
+ "lbux %[p2], %[Temp3](%[cm]) \n\t"
+ "sb %[tp1], 12(%[dst_ptr]) \n\t"
+ "sb %[tn1], 13(%[dst_ptr]) \n\t"
+ "lbux %[n2], %[Temp4](%[cm]) \n\t"
+ "sb %[p2], 14(%[dst_ptr]) \n\t"
+ "sb %[n2], 15(%[dst_ptr]) \n\t"
+
+ : [tn1] "=&r" (tn1), [p2] "=&r" (p2), [n2] "=&r" (n2), [n4] "=&r" (n4),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [tp1] "r" (tp1), [p4] "r" (p4), [n1] "r" (n1), [p1] "r" (p1),
+ [vector4a] "r" (vector4a), [vector3b] "r" (vector3b), [p3] "r" (p3),
+ [n3] "r" (n3), [src_ptr] "r" (src_ptr),
+ [cm] "r" (cm), [dst_ptr] "r" (dst_ptr)
+ );
+
+ src_ptr += src_pixels_per_line;
+ dst_ptr += pitch;
+ }
+}
+
+
+void vp8_filter_block2d_first_pass16_0
+(
+ unsigned char *RESTRICT src_ptr,
+ unsigned char *RESTRICT output_ptr,
+ unsigned int src_pixels_per_line
+)
+{
+ int Temp1, Temp2, Temp3, Temp4;
+ int i;
+
+ /* prefetch src_ptr data to cache memory */
+ prefetch_store(output_ptr + 32);
+
+ /* copy memory from src buffer to dst buffer */
+ for (i = 0; i < 7; i++)
+ {
+ __asm__ __volatile__ (
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "ulw %[Temp3], 8(%[src_ptr]) \n\t"
+ "ulw %[Temp4], 12(%[src_ptr]) \n\t"
+ "sw %[Temp1], 0(%[output_ptr]) \n\t"
+ "sw %[Temp2], 4(%[output_ptr]) \n\t"
+ "sw %[Temp3], 8(%[output_ptr]) \n\t"
+ "sw %[Temp4], 12(%[output_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [Temp3] "=&r" (Temp3),
+ [Temp4] "=&r" (Temp4), [src_ptr] "+r" (src_ptr)
+ : [src_pixels_per_line] "r" (src_pixels_per_line),
+ [output_ptr] "r" (output_ptr)
+ );
+
+ __asm__ __volatile__ (
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "ulw %[Temp3], 8(%[src_ptr]) \n\t"
+ "ulw %[Temp4], 12(%[src_ptr]) \n\t"
+ "sw %[Temp1], 16(%[output_ptr]) \n\t"
+ "sw %[Temp2], 20(%[output_ptr]) \n\t"
+ "sw %[Temp3], 24(%[output_ptr]) \n\t"
+ "sw %[Temp4], 28(%[output_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [Temp3] "=&r" (Temp3),
+ [Temp4] "=&r" (Temp4), [src_ptr] "+r" (src_ptr)
+ : [src_pixels_per_line] "r" (src_pixels_per_line),
+ [output_ptr] "r" (output_ptr)
+ );
+
+ __asm__ __volatile__ (
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "ulw %[Temp3], 8(%[src_ptr]) \n\t"
+ "ulw %[Temp4], 12(%[src_ptr]) \n\t"
+ "sw %[Temp1], 32(%[output_ptr]) \n\t"
+ "sw %[Temp2], 36(%[output_ptr]) \n\t"
+ "sw %[Temp3], 40(%[output_ptr]) \n\t"
+ "sw %[Temp4], 44(%[output_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [Temp3] "=&r" (Temp3),
+ [Temp4] "=&r" (Temp4), [src_ptr] "+r" (src_ptr)
+ : [src_pixels_per_line] "r" (src_pixels_per_line),
+ [output_ptr] "r" (output_ptr)
+ );
+
+ output_ptr += 48;
+ }
+}
+
+
+void vp8_filter_block2d_first_pass16_4tap
+(
+ unsigned char *RESTRICT src_ptr,
+ unsigned char *RESTRICT output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int output_width,
+ unsigned int output_height,
+ int xoffset,
+ int yoffset,
+ unsigned char *RESTRICT dst_ptr,
+ int pitch
+)
+{
+ unsigned int i, j;
+ int Temp1, Temp2, Temp3, Temp4;
+
+ unsigned int vector4a;
+ int vector1b, vector2b;
+ unsigned int tp1, tp2, tp3, tn1;
+ unsigned int p1, p2, p3;
+ unsigned int n1, n2, n3;
+ unsigned char *cm = ff_cropTbl + CROP_WIDTH;
+
+ vector4a = 64;
+
+ vector1b = sub_pel_filters_inv_tap_4[xoffset][0];
+ vector2b = sub_pel_filters_inv_tap_4[xoffset][1];
+
+ /* if (yoffset == 0) don't need temp buffer, data will be stored in dst_ptr */
+ if (yoffset == 0)
+ {
+ output_height -= 5;
+ src_ptr += (src_pixels_per_line + src_pixels_per_line);
+
+ for (i = output_height; i--;)
+ {
+ __asm__ __volatile__ (
+ "ulw %[tp3], -1(%[src_ptr]) \n\t"
+ : [tp3] "=&r" (tp3)
+ : [src_ptr] "r" (src_ptr)
+ );
+
+ /* processing 4 adjacent pixels */
+ for (j = 0; j < 16; j += 4)
+ {
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "ulw %[tp2], 3(%[src_ptr]) \n\t"
+ "move %[tp1], %[tp3] \n\t"
+
+ /* even 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "mthi $0, $ac3 \n\t"
+ "move %[tp3], %[tp2] \n\t"
+ "preceu.ph.qbr %[p1], %[tp1] \n\t"
+ "preceu.ph.qbl %[p2], %[tp1] \n\t"
+ "preceu.ph.qbr %[p3], %[tp2] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
+
+ /* even 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "mthi $0, $ac2 \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
+ "extr.w %[Temp1], $ac3, 7 \n\t"
+
+ /* odd 1. pixel */
+ "ulw %[tn1], 4(%[src_ptr]) \n\t"
+ "balign %[tp2], %[tp1], 3 \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "mthi $0, $ac3 \n\t"
+ "preceu.ph.qbr %[n1], %[tp2] \n\t"
+ "preceu.ph.qbl %[n2], %[tp2] \n\t"
+ "preceu.ph.qbr %[n3], %[tn1] \n\t"
+ "extr.w %[Temp3], $ac2, 7 \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
+
+ /* odd 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "mthi $0, $ac2 \n\t"
+ "extr.w %[Temp2], $ac3, 7 \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
+ "extr.w %[Temp4], $ac2, 7 \n\t"
+
+ /* clamp and store results */
+ "lbux %[tp1], %[Temp1](%[cm]) \n\t"
+ "lbux %[tn1], %[Temp2](%[cm]) \n\t"
+ "lbux %[tp2], %[Temp3](%[cm]) \n\t"
+ "sb %[tp1], 0(%[dst_ptr]) \n\t"
+ "sb %[tn1], 1(%[dst_ptr]) \n\t"
+ "lbux %[n2], %[Temp4](%[cm]) \n\t"
+ "sb %[tp2], 2(%[dst_ptr]) \n\t"
+ "sb %[n2], 3(%[dst_ptr]) \n\t"
+
+ : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tp3] "=&r" (tp3),
+ [tn1] "=&r" (tn1), [p1] "=&r" (p1), [p2] "=&r" (p2),
+ [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [p3] "=&r" (p3),
+ [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector4a] "r" (vector4a), [cm] "r" (cm), [dst_ptr] "r" (dst_ptr),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ src_ptr += 4;
+ }
+
+ /* Next row... */
+ src_ptr += src_pixels_per_line - 16;
+ dst_ptr += pitch;
+ }
+ }
+ else
+ {
+ for (i = output_height; i--;)
+ {
+ /* processing 4 adjacent pixels */
+ for (j = 0; j < 16; j += 4)
+ {
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "ulw %[tp1], -1(%[src_ptr]) \n\t"
+ "ulw %[tp2], 3(%[src_ptr]) \n\t"
+
+ /* even 1. pixel */
+ "mtlo %[vector4a], $ac3 \n\t"
+ "mthi $0, $ac3 \n\t"
+ "preceu.ph.qbr %[p1], %[tp1] \n\t"
+ "preceu.ph.qbl %[p2], %[tp1] \n\t"
+ "preceu.ph.qbr %[p3], %[tp2] \n\t"
+ "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
+
+ /* even 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "mthi $0, $ac2 \n\t"
+ "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
+ "extr.w %[Temp1], $ac3, 7 \n\t"
+
+ /* odd 1. pixel */
+ "ulw %[tn1], 4(%[src_ptr]) \n\t"
+ "balign %[tp2], %[tp1], 3 \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "mthi $0, $ac3 \n\t"
+ "preceu.ph.qbr %[n1], %[tp2] \n\t"
+ "preceu.ph.qbl %[n2], %[tp2] \n\t"
+ "preceu.ph.qbr %[n3], %[tn1] \n\t"
+ "extr.w %[Temp3], $ac2, 7 \n\t"
+ "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
+ "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
+
+ /* odd 2. pixel */
+ "mtlo %[vector4a], $ac2 \n\t"
+ "mthi $0, $ac2 \n\t"
+ "extr.w %[Temp2], $ac3, 7 \n\t"
+ "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
+ "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
+ "extr.w %[Temp4], $ac2, 7 \n\t"
+
+ /* clamp and store results */
+ "lbux %[tp1], %[Temp1](%[cm]) \n\t"
+ "lbux %[tn1], %[Temp2](%[cm]) \n\t"
+ "lbux %[tp2], %[Temp3](%[cm]) \n\t"
+ "sb %[tp1], 0(%[output_ptr]) \n\t"
+ "sb %[tn1], 1(%[output_ptr]) \n\t"
+ "lbux %[n2], %[Temp4](%[cm]) \n\t"
+ "sb %[tp2], 2(%[output_ptr]) \n\t"
+ "sb %[n2], 3(%[output_ptr]) \n\t"
+
+ : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn1] "=&r" (tn1),
+ [p1] "=&r" (p1), [p2] "=&r" (p2), [p3] "=&r" (p3),
+ [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
+ [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector4a] "r" (vector4a), [cm] "r" (cm),
+ [output_ptr] "r" (output_ptr), [src_ptr] "r" (src_ptr)
+ );
+
+ src_ptr += 4;
+ }
+
+ /* next row... */
+ src_ptr += src_pixels_per_line;
+ output_ptr += output_width;
+ }
+ }
+}
+
+
+void vp8_filter_block2d_second_pass4
+(
+ unsigned char *RESTRICT src_ptr,
+ unsigned char *RESTRICT output_ptr,
+ int output_pitch,
+ int yoffset
+)
+{
+ unsigned int i;
+
+ int Temp1, Temp2, Temp3, Temp4;
+ unsigned int vector1b, vector2b, vector3b, vector4a;
+
+ unsigned char src_ptr_l2;
+ unsigned char src_ptr_l1;
+ unsigned char src_ptr_0;
+ unsigned char src_ptr_r1;
+ unsigned char src_ptr_r2;
+ unsigned char src_ptr_r3;
+
+ unsigned char *cm = ff_cropTbl + CROP_WIDTH;
+
+ vector4a = 64;
+
+ /* load filter coefficients */
+ vector1b = sub_pel_filterss[yoffset][0];
+ vector2b = sub_pel_filterss[yoffset][2];
+ vector3b = sub_pel_filterss[yoffset][1];
+
+ if (vector1b)
+ {
+ /* 6 tap filter */
+
+ for (i = 2; i--;)
+ {
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr);
+
+ /* do not allow compiler to reorder instructions */
+ __asm__ __volatile__ (
+ ".set noreorder \n\t"
+ :
+ :
+ );
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l2], -8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 12(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 13(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 10(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 14(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 11(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 15(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp3], $ac0, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp4], $ac1, 9 \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
+ [src_ptr_l2] "=&r" (src_ptr_l2), [src_ptr_r3] "=&r" (src_ptr_r3)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ output_ptr[0] = cm[Temp1];
+ output_ptr[1] = cm[Temp2];
+ output_ptr[2] = cm[Temp3];
+ output_ptr[3] = cm[Temp4];
+
+ output_ptr += output_pitch;
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l2], -4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], 0(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 12(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 16(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], 1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 13(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 17(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], 2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 10(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 14(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 18(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], 3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 11(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 15(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 19(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp3], $ac0, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp4], $ac1, 9 \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
+ [src_ptr_l2] "=&r" (src_ptr_l2), [src_ptr_r3] "=&r" (src_ptr_r3)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ output_ptr[0] = cm[Temp1];
+ output_ptr[1] = cm[Temp2];
+ output_ptr[2] = cm[Temp3];
+ output_ptr[3] = cm[Temp4];
+
+ src_ptr += 8;
+ output_ptr += output_pitch;
+ }
+ }
+ else
+ {
+ /* 4 tap filter */
+
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr);
+
+ for (i = 2; i--;)
+ {
+ /* do not allow compiler to reorder instructions */
+ __asm__ __volatile__ (
+ ".set noreorder \n\t"
+ :
+ :
+ );
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l1], -4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 8(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 9(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 10(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 11(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp3], $ac0, 9 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp4], $ac1, 9 \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ output_ptr[0] = cm[Temp1];
+ output_ptr[1] = cm[Temp2];
+ output_ptr[2] = cm[Temp3];
+ output_ptr[3] = cm[Temp4];
+
+ output_ptr += output_pitch;
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l1], 0(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 12(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], 1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 13(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], 2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 10(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 14(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], 3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 11(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 15(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp3], $ac0, 9 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp4], $ac1, 9 \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ output_ptr[0] = cm[Temp1];
+ output_ptr[1] = cm[Temp2];
+ output_ptr[2] = cm[Temp3];
+ output_ptr[3] = cm[Temp4];
+
+ src_ptr += 8;
+ output_ptr += output_pitch;
+ }
+ }
+}
+
+
+void vp8_filter_block2d_second_pass_8
+(
+ unsigned char *RESTRICT src_ptr,
+ unsigned char *RESTRICT output_ptr,
+ int output_pitch,
+ unsigned int output_height,
+ unsigned int output_width,
+ unsigned int yoffset
+)
+{
+ unsigned int i;
+
+ int Temp1, Temp2, Temp3, Temp4, Temp5, Temp6, Temp7, Temp8;
+ unsigned int vector1b, vector2b, vector3b, vector4a;
+
+ unsigned char src_ptr_l2;
+ unsigned char src_ptr_l1;
+ unsigned char src_ptr_0;
+ unsigned char src_ptr_r1;
+ unsigned char src_ptr_r2;
+ unsigned char src_ptr_r3;
+ unsigned char *cm = ff_cropTbl + CROP_WIDTH;
+
+ vector4a = 64;
+
+ vector1b = sub_pel_filterss[yoffset][0];
+ vector2b = sub_pel_filterss[yoffset][2];
+ vector3b = sub_pel_filterss[yoffset][1];
+
+ if (vector1b)
+ {
+ /* 6 tap filter */
+
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr);
+
+ for (i = output_height; i--;)
+ {
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l2], -16(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 16(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 24(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -15(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 17(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 25(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -14(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 10(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 18(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 26(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -13(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 11(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 19(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 27(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp3], $ac0, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [Temp3] "=&r" (Temp3),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
+ [src_ptr_l2] "=&r" (src_ptr_l2), [src_ptr_r3] "=&r" (src_ptr_r3)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l2], -12(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 12(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 20(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 28(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp4], $ac1, 9 \n\t"
+
+ "lbu %[src_ptr_l2], -11(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 13(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 21(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 29(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp5], $ac2, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -10(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 14(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 22(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 30(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp6], $ac3, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 15(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 23(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 31(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp7], $ac0, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp8], $ac1, 9 \n\t"
+
+ : [Temp4] "=&r" (Temp4), [Temp5] "=&r" (Temp5),
+ [Temp6] "=&r" (Temp6), [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
+ [src_ptr_l2] "=&r" (src_ptr_l2),[src_ptr_r3] "=&r" (src_ptr_r3)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ output_ptr[0] = cm[Temp1];
+ output_ptr[1] = cm[Temp2];
+ output_ptr[2] = cm[Temp3];
+ output_ptr[3] = cm[Temp4];
+ output_ptr[4] = cm[Temp5];
+ output_ptr[5] = cm[Temp6];
+ output_ptr[6] = cm[Temp7];
+ output_ptr[7] = cm[Temp8];
+
+ src_ptr += 8;
+ output_ptr += output_pitch;
+ }
+ }
+ else
+ {
+ /* 4 tap filter */
+
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr);
+
+ for (i = output_height; i--;)
+ {
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l1], -8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 16(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ : [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
+ );
+
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l1], -7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 17(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ : [Temp1] "=r" (Temp1),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
+ );
+
+ src_ptr_l1 = src_ptr[-6];
+ src_ptr_0 = src_ptr[2];
+ src_ptr_r1 = src_ptr[10];
+ src_ptr_r2 = src_ptr[18];
+
+ __asm__ __volatile__ (
+ "mtlo %[vector4a], $ac0 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+
+ : [Temp2] "=r" (Temp2)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
+ [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
+ [vector4a] "r" (vector4a)
+ );
+
+ src_ptr_l1 = src_ptr[-5];
+ src_ptr_0 = src_ptr[3];
+ src_ptr_r1 = src_ptr[11];
+ src_ptr_r2 = src_ptr[19];
+
+ __asm__ __volatile__ (
+ "mtlo %[vector4a], $ac1 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp3], $ac0, 9 \n\t"
+
+ : [Temp3] "=r" (Temp3)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
+ [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
+ [vector4a] "r" (vector4a)
+ );
+
+ src_ptr_l1 = src_ptr[-4];
+ src_ptr_0 = src_ptr[4];
+ src_ptr_r1 = src_ptr[12];
+ src_ptr_r2 = src_ptr[20];
+
+ __asm__ __volatile__ (
+ "mtlo %[vector4a], $ac2 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp4], $ac1, 9 \n\t"
+
+ : [Temp4] "=r" (Temp4)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
+ [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
+ [vector4a] "r" (vector4a)
+ );
+
+ src_ptr_l1 = src_ptr[-3];
+ src_ptr_0 = src_ptr[5];
+ src_ptr_r1 = src_ptr[13];
+ src_ptr_r2 = src_ptr[21];
+
+ __asm__ __volatile__ (
+ "mtlo %[vector4a], $ac3 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp5], $ac2, 9 \n\t"
+
+ : [Temp5] "=&r" (Temp5)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
+ [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
+ [vector4a] "r" (vector4a)
+ );
+
+ src_ptr_l1 = src_ptr[-2];
+ src_ptr_0 = src_ptr[6];
+ src_ptr_r1 = src_ptr[14];
+ src_ptr_r2 = src_ptr[22];
+
+ __asm__ __volatile__ (
+ "mtlo %[vector4a], $ac0 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp6], $ac3, 9 \n\t"
+
+ : [Temp6] "=r" (Temp6)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
+ [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
+ [vector4a] "r" (vector4a)
+ );
+
+ src_ptr_l1 = src_ptr[-1];
+ src_ptr_0 = src_ptr[7];
+ src_ptr_r1 = src_ptr[15];
+ src_ptr_r2 = src_ptr[23];
+
+ __asm__ __volatile__ (
+ "mtlo %[vector4a], $ac1 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp7], $ac0, 9 \n\t"
+ "extp %[Temp8], $ac1, 9 \n\t"
+
+ : [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
+ [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
+ [vector4a] "r" (vector4a)
+ );
+
+ /* clamp and store results */
+ output_ptr[0] = cm[Temp1];
+ output_ptr[1] = cm[Temp2];
+ output_ptr[2] = cm[Temp3];
+ output_ptr[3] = cm[Temp4];
+ output_ptr[4] = cm[Temp5];
+ output_ptr[5] = cm[Temp6];
+ output_ptr[6] = cm[Temp7];
+ output_ptr[7] = cm[Temp8];
+
+ src_ptr += 8;
+ output_ptr += output_pitch;
+ }
+ }
+}
+
+
+void vp8_filter_block2d_second_pass161
+(
+ unsigned char *RESTRICT src_ptr,
+ unsigned char *RESTRICT output_ptr,
+ int output_pitch,
+ const unsigned short *vp8_filter
+)
+{
+ unsigned int i, j;
+
+ int Temp1, Temp2, Temp3, Temp4, Temp5, Temp6, Temp7, Temp8;
+ unsigned int vector4a;
+ unsigned int vector1b, vector2b, vector3b;
+
+ unsigned char src_ptr_l2;
+ unsigned char src_ptr_l1;
+ unsigned char src_ptr_0;
+ unsigned char src_ptr_r1;
+ unsigned char src_ptr_r2;
+ unsigned char src_ptr_r3;
+ unsigned char *cm = ff_cropTbl + CROP_WIDTH;
+
+ vector4a = 64;
+
+ vector1b = vp8_filter[0];
+ vector2b = vp8_filter[2];
+ vector3b = vp8_filter[1];
+
+ if (vector1b == 0)
+ {
+ /* 4 tap filter */
+
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + 16);
+
+ for (i = 16; i--;)
+ {
+ /* unrolling for loop */
+ for (j = 0; j < 16; j += 8)
+ {
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l1], -16(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 16(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 32(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -15(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 17(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 33(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -14(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 18(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 34(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp2], $ac3, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -13(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 19(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 35(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp3], $ac1, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -12(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 20(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 36(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+ "extp %[Temp4], $ac3, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -11(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 21(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 37(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp5], $ac2, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -10(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 22(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 38(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp6], $ac3, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l1], -9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 23(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 39(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp7], $ac1, 9 \n\t"
+
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp8], $ac3, 9 \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4),
+ [Temp5] "=&r" (Temp5), [Temp6] "=&r" (Temp6),
+ [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
+ : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
+ [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ output_ptr[j] = cm[Temp1];
+ output_ptr[j + 1] = cm[Temp2];
+ output_ptr[j + 2] = cm[Temp3];
+ output_ptr[j + 3] = cm[Temp4];
+ output_ptr[j + 4] = cm[Temp5];
+ output_ptr[j + 5] = cm[Temp6];
+ output_ptr[j + 6] = cm[Temp7];
+ output_ptr[j + 7] = cm[Temp8];
+
+ src_ptr += 8;
+ }
+
+ output_ptr += output_pitch;
+ }
+ }
+ else
+ {
+ /* 4 tap filter */
+
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + 16);
+
+ /* unroll for loop */
+ for (i = 16; i--;)
+ {
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l2], -32(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -16(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 16(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 32(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 48(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -31(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -15(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 17(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 33(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 49(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -30(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -14(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 18(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 34(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 50(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp2], $ac0, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -29(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -13(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 19(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 35(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 51(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp3], $ac1, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -28(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -12(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 20(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 36(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 52(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+ "extp %[Temp4], $ac3, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -27(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -11(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 21(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 37(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 53(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp5], $ac2, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -26(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -10(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 22(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 38(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 54(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp6], $ac0, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -25(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 23(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 39(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 55(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp7], $ac1, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp8], $ac3, 9 \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4),
+ [Temp5] "=&r" (Temp5), [Temp6] "=&r" (Temp6),
+ [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
+ [src_ptr_l2] "=&r" (src_ptr_l2),[src_ptr_r3] "=&r" (src_ptr_r3)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ /* clamp and store results */
+ output_ptr[0] = cm[Temp1];
+ output_ptr[1] = cm[Temp2];
+ output_ptr[2] = cm[Temp3];
+ output_ptr[3] = cm[Temp4];
+ output_ptr[4] = cm[Temp5];
+ output_ptr[5] = cm[Temp6];
+ output_ptr[6] = cm[Temp7];
+ output_ptr[7] = cm[Temp8];
+
+ /* apply filter with vectors pairs */
+ __asm__ __volatile__ (
+ "lbu %[src_ptr_l2], -24(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 8(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 24(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 40(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 56(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -23(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -7(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 9(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 25(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 41(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 57(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp1], $ac2, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -22(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -6(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 10(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 26(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 42(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 58(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp2], $ac0, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -21(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -5(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 11(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 27(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 43(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 59(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp3], $ac1, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -20(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -4(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 12(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 28(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 44(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 60(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac2 \n\t"
+ "extp %[Temp4], $ac3, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -19(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -3(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 13(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 29(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 45(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 61(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac0 \n\t"
+ "extp %[Temp5], $ac2, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -18(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -2(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 14(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 30(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 46(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 62(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac1 \n\t"
+ "extp %[Temp6], $ac0, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
+
+ "lbu %[src_ptr_l2], -17(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_l1], -1(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_0], 15(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r1], 31(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r2], 47(%[src_ptr]) \n\t"
+ "lbu %[src_ptr_r3], 63(%[src_ptr]) \n\t"
+ "mtlo %[vector4a], $ac3 \n\t"
+ "extp %[Temp7], $ac1, 9 \n\t"
+
+ "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
+ "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
+ "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
+ "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
+ "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
+ "extp %[Temp8], $ac3, 9 \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
+ [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4),
+ [Temp5] "=&r" (Temp5), [Temp6] "=&r" (Temp6),
+ [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8),
+ [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
+ [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
+ [src_ptr_l2] "=&r" (src_ptr_l2), [src_ptr_r3] "=&r" (src_ptr_r3)
+ : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
+ [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
+ [src_ptr] "r" (src_ptr)
+ );
+
+ src_ptr += 16;
+ output_ptr[8] = cm[Temp1];
+ output_ptr[9] = cm[Temp2];
+ output_ptr[10] = cm[Temp3];
+ output_ptr[11] = cm[Temp4];
+ output_ptr[12] = cm[Temp5];
+ output_ptr[13] = cm[Temp6];
+ output_ptr[14] = cm[Temp7];
+ output_ptr[15] = cm[Temp8];
+
+ output_ptr += output_pitch;
+ }
+ }
+}
+
+
+void vp8_sixtap_predict4x4_dspr2
+(
+ unsigned char *RESTRICT src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *RESTRICT dst_ptr,
+ int dst_pitch
+)
+{
+ unsigned char FData[9 * 4]; /* Temp data bufffer used in filtering */
+ unsigned int pos = 16;
+
+ /* bit positon for extract from acc */
+ __asm__ __volatile__ (
+ "wrdsp %[pos], 1 \n\t"
+ :
+ : [pos] "r" (pos)
+ );
+
+ if (yoffset)
+ {
+ /* First filter 1-D horizontally... */
+ vp8_filter_block2d_first_pass_4(src_ptr - (2 * src_pixels_per_line), FData,
+ src_pixels_per_line, 9, xoffset, 4);
+ /* then filter verticaly... */
+ vp8_filter_block2d_second_pass4(FData + 8, dst_ptr, dst_pitch, yoffset);
+ }
+ else
+ /* if (yoffsset == 0) vp8_filter_block2d_first_pass save data to dst_ptr */
+ vp8_filter_block2d_first_pass_4(src_ptr, dst_ptr, src_pixels_per_line,
+ 4, xoffset, dst_pitch);
+}
+
+
+void vp8_sixtap_predict8x8_dspr2
+(
+ unsigned char *RESTRICT src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *RESTRICT dst_ptr,
+ int dst_pitch
+)
+{
+
+ unsigned char FData[13 * 8]; /* Temp data bufffer used in filtering */
+ unsigned int pos, Temp1, Temp2;
+
+ pos = 16;
+
+ /* bit positon for extract from acc */
+ __asm__ __volatile__ (
+ "wrdsp %[pos], 1 \n\t"
+ :
+ : [pos] "r" (pos)
+ );
+
+ if (yoffset)
+ {
+
+ src_ptr = src_ptr - (2 * src_pixels_per_line);
+
+ if (xoffset)
+ /* filter 1-D horizontally... */
+ vp8_filter_block2d_first_pass_8_all(src_ptr, FData, src_pixels_per_line,
+ 13, xoffset, 8);
+
+ else
+ {
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + 2 * src_pixels_per_line);
+
+ __asm__ __volatile__ (
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 0(%[FData]) \n\t"
+ "sw %[Temp2], 4(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 8(%[FData]) \n\t"
+ "sw %[Temp2], 12(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 16(%[FData]) \n\t"
+ "sw %[Temp2], 20(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 24(%[FData]) \n\t"
+ "sw %[Temp2], 28(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 32(%[FData]) \n\t"
+ "sw %[Temp2], 36(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 40(%[FData]) \n\t"
+ "sw %[Temp2], 44(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 48(%[FData]) \n\t"
+ "sw %[Temp2], 52(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 56(%[FData]) \n\t"
+ "sw %[Temp2], 60(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 64(%[FData]) \n\t"
+ "sw %[Temp2], 68(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 72(%[FData]) \n\t"
+ "sw %[Temp2], 76(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 80(%[FData]) \n\t"
+ "sw %[Temp2], 84(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 88(%[FData]) \n\t"
+ "sw %[Temp2], 92(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 96(%[FData]) \n\t"
+ "sw %[Temp2], 100(%[FData]) \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2)
+ : [FData] "r" (FData), [src_ptr] "r" (src_ptr),
+ [src_pixels_per_line] "r" (src_pixels_per_line)
+ );
+ }
+
+ /* filter verticaly... */
+ vp8_filter_block2d_second_pass_8(FData + 16, dst_ptr, dst_pitch, 8, 8, yoffset);
+ }
+
+ /* if (yoffsset == 0) vp8_filter_block2d_first_pass save data to dst_ptr */
+ else
+ {
+ if (xoffset)
+ vp8_filter_block2d_first_pass_8_all(src_ptr, dst_ptr, src_pixels_per_line,
+ 8, xoffset, dst_pitch);
+
+ else
+ {
+ /* copy from src buffer to dst buffer */
+ __asm__ __volatile__ (
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 0(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 4(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 8(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 12(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 16(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 20(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 24(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 28(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 32(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 36(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 40(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 44(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 48(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 52(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 56(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 60(%[dst_ptr]) \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2)
+ : [dst_ptr] "r" (dst_ptr), [src_ptr] "r" (src_ptr),
+ [src_pixels_per_line] "r" (src_pixels_per_line)
+ );
+ }
+ }
+}
+
+
+void vp8_sixtap_predict8x4_dspr2
+(
+ unsigned char *RESTRICT src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *RESTRICT dst_ptr,
+ int dst_pitch
+)
+{
+ unsigned char FData[9 * 8]; /* Temp data bufffer used in filtering */
+ unsigned int pos, Temp1, Temp2;
+
+ pos = 16;
+
+ /* bit positon for extract from acc */
+ __asm__ __volatile__ (
+ "wrdsp %[pos], 1 \n\t"
+ :
+ : [pos] "r" (pos)
+ );
+
+ if (yoffset)
+ {
+
+ src_ptr = src_ptr - (2 * src_pixels_per_line);
+
+ if (xoffset)
+ /* filter 1-D horizontally... */
+ vp8_filter_block2d_first_pass_8_all(src_ptr, FData, src_pixels_per_line,
+ 9, xoffset, 8);
+
+ else
+ {
+ /* prefetch src_ptr data to cache memory */
+ prefetch_load(src_ptr + 2 * src_pixels_per_line);
+
+ __asm__ __volatile__ (
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 0(%[FData]) \n\t"
+ "sw %[Temp2], 4(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 8(%[FData]) \n\t"
+ "sw %[Temp2], 12(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 16(%[FData]) \n\t"
+ "sw %[Temp2], 20(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 24(%[FData]) \n\t"
+ "sw %[Temp2], 28(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 32(%[FData]) \n\t"
+ "sw %[Temp2], 36(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 40(%[FData]) \n\t"
+ "sw %[Temp2], 44(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 48(%[FData]) \n\t"
+ "sw %[Temp2], 52(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 56(%[FData]) \n\t"
+ "sw %[Temp2], 60(%[FData]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 64(%[FData]) \n\t"
+ "sw %[Temp2], 68(%[FData]) \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2)
+ : [FData] "r" (FData), [src_ptr] "r" (src_ptr),
+ [src_pixels_per_line] "r" (src_pixels_per_line)
+ );
+ }
+
+ /* filter verticaly... */
+ vp8_filter_block2d_second_pass_8(FData + 16, dst_ptr, dst_pitch, 4, 8, yoffset);
+ }
+
+ /* if (yoffsset == 0) vp8_filter_block2d_first_pass save data to dst_ptr */
+ else
+ {
+ if (xoffset)
+ vp8_filter_block2d_first_pass_8_all(src_ptr, dst_ptr, src_pixels_per_line,
+ 4, xoffset, dst_pitch);
+
+ else
+ {
+ /* copy from src buffer to dst buffer */
+ __asm__ __volatile__ (
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 0(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 4(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 8(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 12(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 16(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 20(%[dst_ptr]) \n\t"
+ "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
+
+ "ulw %[Temp1], 0(%[src_ptr]) \n\t"
+ "ulw %[Temp2], 4(%[src_ptr]) \n\t"
+ "sw %[Temp1], 24(%[dst_ptr]) \n\t"
+ "sw %[Temp2], 28(%[dst_ptr]) \n\t"
+
+ : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2)
+ : [dst_ptr] "r" (dst_ptr), [src_ptr] "r" (src_ptr),
+ [src_pixels_per_line] "r" (src_pixels_per_line)
+ );
+ }
+ }
+}
+
+
+void vp8_sixtap_predict16x16_dspr2
+(
+ unsigned char *RESTRICT src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *RESTRICT dst_ptr,
+ int dst_pitch
+)
+{
+ const unsigned short *VFilter;
+ unsigned char FData[21 * 16]; /* Temp data bufffer used in filtering */
+ unsigned int pos;
+
+ VFilter = sub_pel_filterss[yoffset];
+
+ pos = 16;
+
+ /* bit positon for extract from acc */
+ __asm__ __volatile__ (
+ "wrdsp %[pos], 1 \n\t"
+ :
+ : [pos] "r" (pos)
+ );
+
+ if (yoffset)
+ {
+
+ src_ptr = src_ptr - (2 * src_pixels_per_line);
+
+ switch (xoffset)
+ {
+ /* filter 1-D horizontally... */
+ case 2:
+ case 4:
+ case 6:
+ /* 6 tap filter */
+ vp8_filter_block2d_first_pass16_6tap(src_ptr, FData, src_pixels_per_line,
+ 21, xoffset, 16);
+ break;
+
+ case 0:
+ /* only copy buffer */
+ vp8_filter_block2d_first_pass16_0(src_ptr, FData, src_pixels_per_line);
+ break;
+
+ case 1:
+ case 3:
+ case 5:
+ case 7:
+ /* 4 tap filter */
+ vp8_filter_block2d_first_pass16_4tap(src_ptr, FData, src_pixels_per_line, 16,
+ 21, xoffset, yoffset, dst_ptr, dst_pitch);
+ break;
+ }
+
+ /* filter verticaly... */
+ vp8_filter_block2d_second_pass161(FData + 32, dst_ptr, dst_pitch, VFilter);
+ }
+ else
+ {
+ /* if (yoffsset == 0) vp8_filter_block2d_first_pass save data to dst_ptr */
+ switch (xoffset)
+ {
+ case 2:
+ case 4:
+ case 6:
+ /* 6 tap filter */
+ vp8_filter_block2d_first_pass16_6tap(src_ptr, dst_ptr, src_pixels_per_line,
+ 16, xoffset, dst_pitch);
+ break;
+
+ case 1:
+ case 3:
+ case 5:
+ case 7:
+ /* 4 tap filter */
+ vp8_filter_block2d_first_pass16_4tap(src_ptr, dst_ptr, src_pixels_per_line, 16,
+ 21, xoffset, yoffset, dst_ptr, dst_pitch);
+ break;
+ }
+ }
+}
+
+#endif
diff --git a/libvpx/vp8/common/mips/dspr2/idct_blk_dspr2.c b/libvpx/vp8/common/mips/dspr2/idct_blk_dspr2.c
new file mode 100644
index 0000000..1e0ebd1
--- /dev/null
+++ b/libvpx/vp8/common/mips/dspr2/idct_blk_dspr2.c
@@ -0,0 +1,88 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+
+#if HAVE_DSPR2
+
+void vp8_dequant_idct_add_y_block_dspr2
+(short *q, short *dq,
+ unsigned char *dst, int stride, char *eobs)
+{
+ int i, j;
+
+ for (i = 0; i < 4; i++)
+ {
+ for (j = 0; j < 4; j++)
+ {
+ if (*eobs++ > 1)
+ vp8_dequant_idct_add_dspr2(q, dq, dst, stride);
+ else
+ {
+ vp8_dc_only_idct_add_dspr2(q[0]*dq[0], dst, stride, dst, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ q += 16;
+ dst += 4;
+ }
+
+ dst += 4 * stride - 16;
+ }
+}
+
+void vp8_dequant_idct_add_uv_block_dspr2
+(short *q, short *dq,
+ unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
+{
+ int i, j;
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 2; j++)
+ {
+ if (*eobs++ > 1)
+ vp8_dequant_idct_add_dspr2(q, dq, dstu, stride);
+ else
+ {
+ vp8_dc_only_idct_add_dspr2(q[0]*dq[0], dstu, stride, dstu, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ q += 16;
+ dstu += 4;
+ }
+
+ dstu += 4 * stride - 8;
+ }
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 2; j++)
+ {
+ if (*eobs++ > 1)
+ vp8_dequant_idct_add_dspr2(q, dq, dstv, stride);
+ else
+ {
+ vp8_dc_only_idct_add_dspr2(q[0]*dq[0], dstv, stride, dstv, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ q += 16;
+ dstv += 4;
+ }
+
+ dstv += 4 * stride - 8;
+ }
+}
+
+#endif
+
diff --git a/libvpx/vp8/common/mips/dspr2/idctllm_dspr2.c b/libvpx/vp8/common/mips/dspr2/idctllm_dspr2.c
new file mode 100644
index 0000000..25b7936
--- /dev/null
+++ b/libvpx/vp8/common/mips/dspr2/idctllm_dspr2.c
@@ -0,0 +1,369 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_rtcd.h"
+
+#if HAVE_DSPR2
+#define CROP_WIDTH 256
+
+/******************************************************************************
+ * Notes:
+ *
+ * This implementation makes use of 16 bit fixed point version of two multiply
+ * constants:
+ * 1. sqrt(2) * cos (pi/8)
+ * 2. sqrt(2) * sin (pi/8)
+ * Since the first constant is bigger than 1, to maintain the same 16 bit
+ * fixed point precision as the second one, we use a trick of
+ * x * a = x + x*(a-1)
+ * so
+ * x * sqrt(2) * cos (pi/8) = x + x * (sqrt(2) *cos(pi/8)-1).
+ ****************************************************************************/
+extern unsigned char ff_cropTbl[256 + 2 * CROP_WIDTH];
+static const int cospi8sqrt2minus1 = 20091;
+static const int sinpi8sqrt2 = 35468;
+
+inline void prefetch_load_short(short *src)
+{
+ __asm__ __volatile__ (
+ "pref 0, 0(%[src]) \n\t"
+ :
+ : [src] "r" (src)
+ );
+}
+
+void vp8_short_idct4x4llm_dspr2(short *input, unsigned char *pred_ptr,
+ int pred_stride, unsigned char *dst_ptr,
+ int dst_stride)
+{
+ int r, c;
+ int a1, b1, c1, d1;
+ short output[16];
+ short *ip = input;
+ short *op = output;
+ int temp1, temp2;
+ int shortpitch = 4;
+
+ int c2, d2;
+ int temp3, temp4;
+ unsigned char *cm = ff_cropTbl + CROP_WIDTH;
+
+ /* prepare data for load */
+ prefetch_load_short(ip + 8);
+
+ /* first loop is unrolled */
+ a1 = ip[0] + ip[8];
+ b1 = ip[0] - ip[8];
+
+ temp1 = (ip[4] * sinpi8sqrt2) >> 16;
+ temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
+ c1 = temp1 - temp2;
+
+ temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
+ temp2 = (ip[12] * sinpi8sqrt2) >> 16;
+ d1 = temp1 + temp2;
+
+ temp3 = (ip[5] * sinpi8sqrt2) >> 16;
+ temp4 = ip[13] + ((ip[13] * cospi8sqrt2minus1) >> 16);
+ c2 = temp3 - temp4;
+
+ temp3 = ip[5] + ((ip[5] * cospi8sqrt2minus1) >> 16);
+ temp4 = (ip[13] * sinpi8sqrt2) >> 16;
+ d2 = temp3 + temp4;
+
+ op[0] = a1 + d1;
+ op[12] = a1 - d1;
+ op[4] = b1 + c1;
+ op[8] = b1 - c1;
+
+ a1 = ip[1] + ip[9];
+ b1 = ip[1] - ip[9];
+
+ op[1] = a1 + d2;
+ op[13] = a1 - d2;
+ op[5] = b1 + c2;
+ op[9] = b1 - c2;
+
+ a1 = ip[2] + ip[10];
+ b1 = ip[2] - ip[10];
+
+ temp1 = (ip[6] * sinpi8sqrt2) >> 16;
+ temp2 = ip[14] + ((ip[14] * cospi8sqrt2minus1) >> 16);
+ c1 = temp1 - temp2;
+
+ temp1 = ip[6] + ((ip[6] * cospi8sqrt2minus1) >> 16);
+ temp2 = (ip[14] * sinpi8sqrt2) >> 16;
+ d1 = temp1 + temp2;
+
+ temp3 = (ip[7] * sinpi8sqrt2) >> 16;
+ temp4 = ip[15] + ((ip[15] * cospi8sqrt2minus1) >> 16);
+ c2 = temp3 - temp4;
+
+ temp3 = ip[7] + ((ip[7] * cospi8sqrt2minus1) >> 16);
+ temp4 = (ip[15] * sinpi8sqrt2) >> 16;
+ d2 = temp3 + temp4;
+
+ op[2] = a1 + d1;
+ op[14] = a1 - d1;
+ op[6] = b1 + c1;
+ op[10] = b1 - c1;
+
+ a1 = ip[3] + ip[11];
+ b1 = ip[3] - ip[11];
+
+ op[3] = a1 + d2;
+ op[15] = a1 - d2;
+ op[7] = b1 + c2;
+ op[11] = b1 - c2;
+
+ ip = output;
+
+ /* prepare data for load */
+ prefetch_load_short(ip + shortpitch);
+
+ /* second loop is unrolled */
+ a1 = ip[0] + ip[2];
+ b1 = ip[0] - ip[2];
+
+ temp1 = (ip[1] * sinpi8sqrt2) >> 16;
+ temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
+ c1 = temp1 - temp2;
+
+ temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
+ temp2 = (ip[3] * sinpi8sqrt2) >> 16;
+ d1 = temp1 + temp2;
+
+ temp3 = (ip[5] * sinpi8sqrt2) >> 16;
+ temp4 = ip[7] + ((ip[7] * cospi8sqrt2minus1) >> 16);
+ c2 = temp3 - temp4;
+
+ temp3 = ip[5] + ((ip[5] * cospi8sqrt2minus1) >> 16);
+ temp4 = (ip[7] * sinpi8sqrt2) >> 16;
+ d2 = temp3 + temp4;
+
+ op[0] = (a1 + d1 + 4) >> 3;
+ op[3] = (a1 - d1 + 4) >> 3;
+ op[1] = (b1 + c1 + 4) >> 3;
+ op[2] = (b1 - c1 + 4) >> 3;
+
+ a1 = ip[4] + ip[6];
+ b1 = ip[4] - ip[6];
+
+ op[4] = (a1 + d2 + 4) >> 3;
+ op[7] = (a1 - d2 + 4) >> 3;
+ op[5] = (b1 + c2 + 4) >> 3;
+ op[6] = (b1 - c2 + 4) >> 3;
+
+ a1 = ip[8] + ip[10];
+ b1 = ip[8] - ip[10];
+
+ temp1 = (ip[9] * sinpi8sqrt2) >> 16;
+ temp2 = ip[11] + ((ip[11] * cospi8sqrt2minus1) >> 16);
+ c1 = temp1 - temp2;
+
+ temp1 = ip[9] + ((ip[9] * cospi8sqrt2minus1) >> 16);
+ temp2 = (ip[11] * sinpi8sqrt2) >> 16;
+ d1 = temp1 + temp2;
+
+ temp3 = (ip[13] * sinpi8sqrt2) >> 16;
+ temp4 = ip[15] + ((ip[15] * cospi8sqrt2minus1) >> 16);
+ c2 = temp3 - temp4;
+
+ temp3 = ip[13] + ((ip[13] * cospi8sqrt2minus1) >> 16);
+ temp4 = (ip[15] * sinpi8sqrt2) >> 16;
+ d2 = temp3 + temp4;
+
+ op[8] = (a1 + d1 + 4) >> 3;
+ op[11] = (a1 - d1 + 4) >> 3;
+ op[9] = (b1 + c1 + 4) >> 3;
+ op[10] = (b1 - c1 + 4) >> 3;
+
+ a1 = ip[12] + ip[14];
+ b1 = ip[12] - ip[14];
+
+ op[12] = (a1 + d2 + 4) >> 3;
+ op[15] = (a1 - d2 + 4) >> 3;
+ op[13] = (b1 + c2 + 4) >> 3;
+ op[14] = (b1 - c2 + 4) >> 3;
+
+ ip = output;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ short a = ip[c] + pred_ptr[c] ;
+ dst_ptr[c] = cm[a] ;
+ }
+
+ ip += 4;
+ dst_ptr += dst_stride;
+ pred_ptr += pred_stride;
+ }
+}
+
+void vp8_dc_only_idct_add_dspr2(short input_dc, unsigned char *pred_ptr, int pred_stride, unsigned char *dst_ptr, int dst_stride)
+{
+ int a1;
+ int i, absa1;
+ int t2, vector_a1, vector_a;
+
+ /* a1 = ((input_dc + 4) >> 3); */
+ __asm__ __volatile__ (
+ "addi %[a1], %[input_dc], 4 \n\t"
+ "sra %[a1], %[a1], 3 \n\t"
+ : [a1] "=r" (a1)
+ : [input_dc] "r" (input_dc)
+ );
+
+ if (a1 < 0)
+ {
+ /* use quad-byte
+ * input and output memory are four byte aligned
+ */
+ __asm__ __volatile__ (
+ "abs %[absa1], %[a1] \n\t"
+ "replv.qb %[vector_a1], %[absa1] \n\t"
+ : [absa1] "=r" (absa1), [vector_a1] "=r" (vector_a1)
+ : [a1] "r" (a1)
+ );
+
+ /* use (a1 - predptr[c]) instead a1 + predptr[c] */
+ for (i = 4; i--;)
+ {
+ __asm__ __volatile__ (
+ "lw %[t2], 0(%[pred_ptr]) \n\t"
+ "add %[pred_ptr], %[pred_ptr], %[pred_stride] \n\t"
+ "subu_s.qb %[vector_a], %[t2], %[vector_a1] \n\t"
+ "sw %[vector_a], 0(%[dst_ptr]) \n\t"
+ "add %[dst_ptr], %[dst_ptr], %[dst_stride] \n\t"
+ : [t2] "=&r" (t2), [vector_a] "=&r" (vector_a),
+ [dst_ptr] "+&r" (dst_ptr), [pred_ptr] "+&r" (pred_ptr)
+ : [dst_stride] "r" (dst_stride), [pred_stride] "r" (pred_stride), [vector_a1] "r" (vector_a1)
+ );
+ }
+ }
+ else
+ {
+ /* use quad-byte
+ * input and output memory are four byte aligned
+ */
+ __asm__ __volatile__ (
+ "replv.qb %[vector_a1], %[a1] \n\t"
+ : [vector_a1] "=r" (vector_a1)
+ : [a1] "r" (a1)
+ );
+
+ for (i = 4; i--;)
+ {
+ __asm__ __volatile__ (
+ "lw %[t2], 0(%[pred_ptr]) \n\t"
+ "add %[pred_ptr], %[pred_ptr], %[pred_stride] \n\t"
+ "addu_s.qb %[vector_a], %[vector_a1], %[t2] \n\t"
+ "sw %[vector_a], 0(%[dst_ptr]) \n\t"
+ "add %[dst_ptr], %[dst_ptr], %[dst_stride] \n\t"
+ : [t2] "=&r" (t2), [vector_a] "=&r" (vector_a),
+ [dst_ptr] "+&r" (dst_ptr), [pred_ptr] "+&r" (pred_ptr)
+ : [dst_stride] "r" (dst_stride), [pred_stride] "r" (pred_stride), [vector_a1] "r" (vector_a1)
+ );
+ }
+ }
+
+}
+
+void vp8_short_inv_walsh4x4_dspr2(short *input, short *mb_dqcoeff)
+{
+ short output[16];
+ int i;
+ int a1, b1, c1, d1;
+ int a2, b2, c2, d2;
+ short *ip = input;
+ short *op = output;
+
+ prefetch_load_short(ip);
+
+ for (i = 4; i--;)
+ {
+ a1 = ip[0] + ip[12];
+ b1 = ip[4] + ip[8];
+ c1 = ip[4] - ip[8];
+ d1 = ip[0] - ip[12];
+
+ op[0] = a1 + b1;
+ op[4] = c1 + d1;
+ op[8] = a1 - b1;
+ op[12] = d1 - c1;
+
+ ip++;
+ op++;
+ }
+
+ ip = output;
+ op = output;
+
+ prefetch_load_short(ip);
+
+ for (i = 4; i--;)
+ {
+ a1 = ip[0] + ip[3] + 3;
+ b1 = ip[1] + ip[2];
+ c1 = ip[1] - ip[2];
+ d1 = ip[0] - ip[3] + 3;
+
+ a2 = a1 + b1;
+ b2 = d1 + c1;
+ c2 = a1 - b1;
+ d2 = d1 - c1;
+
+ op[0] = a2 >> 3;
+ op[1] = b2 >> 3;
+ op[2] = c2 >> 3;
+ op[3] = d2 >> 3;
+
+ ip += 4;
+ op += 4;
+ }
+
+ for (i = 0; i < 16; i++)
+ {
+ mb_dqcoeff[i * 16] = output[i];
+ }
+}
+
+void vp8_short_inv_walsh4x4_1_dspr2(short *input, short *mb_dqcoeff)
+{
+ int a1;
+
+ a1 = ((input[0] + 3) >> 3);
+
+ __asm__ __volatile__ (
+ "sh %[a1], 0(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 32(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 64(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 96(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 128(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 160(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 192(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 224(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 256(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 288(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 320(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 352(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 384(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 416(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 448(%[mb_dqcoeff]) \n\t"
+ "sh %[a1], 480(%[mb_dqcoeff]) \n\t"
+
+ :
+ : [a1] "r" (a1), [mb_dqcoeff] "r" (mb_dqcoeff)
+ );
+}
+
+#endif
diff --git a/libvpx/vp8/common/mips/dspr2/loopfilter_filters_dspr2.c b/libvpx/vp8/common/mips/dspr2/loopfilter_filters_dspr2.c
new file mode 100644
index 0000000..b8e5e4d
--- /dev/null
+++ b/libvpx/vp8/common/mips/dspr2/loopfilter_filters_dspr2.c
@@ -0,0 +1,2622 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdlib.h>
+#include "vpx_rtcd.h"
+#include "vp8/common/onyxc_int.h"
+
+#if HAVE_DSPR2
+typedef unsigned char uc;
+
+/* prefetch data for load */
+inline void prefetch_load_lf(unsigned char *src)
+{
+ __asm__ __volatile__ (
+ "pref 0, 0(%[src]) \n\t"
+ :
+ : [src] "r" (src)
+ );
+}
+
+
+/* prefetch data for store */
+inline void prefetch_store_lf(unsigned char *dst)
+{
+ __asm__ __volatile__ (
+ "pref 1, 0(%[dst]) \n\t"
+ :
+ : [dst] "r" (dst)
+ );
+}
+
+/* processing 4 pixels at the same time
+ * compute hev and mask in the same function
+ */
+static __inline void vp8_filter_mask_vec_mips
+(
+ uint32_t limit,
+ uint32_t flimit,
+ uint32_t p1,
+ uint32_t p0,
+ uint32_t p3,
+ uint32_t p2,
+ uint32_t q0,
+ uint32_t q1,
+ uint32_t q2,
+ uint32_t q3,
+ uint32_t thresh,
+ uint32_t *hev,
+ uint32_t *mask
+)
+{
+ uint32_t c, r, r3, r_k;
+ uint32_t s1, s2, s3;
+ uint32_t ones = 0xFFFFFFFF;
+ uint32_t hev1;
+
+ __asm__ __volatile__ (
+ /* mask |= (abs(p3 - p2) > limit) */
+ "subu_s.qb %[c], %[p3], %[p2] \n\t"
+ "subu_s.qb %[r_k], %[p2], %[p3] \n\t"
+ "or %[r_k], %[r_k], %[c] \n\t"
+ "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
+ "or %[r], $0, %[c] \n\t"
+
+ /* mask |= (abs(p2 - p1) > limit) */
+ "subu_s.qb %[c], %[p2], %[p1] \n\t"
+ "subu_s.qb %[r_k], %[p1], %[p2] \n\t"
+ "or %[r_k], %[r_k], %[c] \n\t"
+ "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
+ "or %[r], %[r], %[c] \n\t"
+
+ /* mask |= (abs(p1 - p0) > limit)
+ * hev |= (abs(p1 - p0) > thresh)
+ */
+ "subu_s.qb %[c], %[p1], %[p0] \n\t"
+ "subu_s.qb %[r_k], %[p0], %[p1] \n\t"
+ "or %[r_k], %[r_k], %[c] \n\t"
+ "cmpgu.lt.qb %[c], %[thresh], %[r_k] \n\t"
+ "or %[r3], $0, %[c] \n\t"
+ "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
+ "or %[r], %[r], %[c] \n\t"
+
+ /* mask |= (abs(q1 - q0) > limit)
+ * hev |= (abs(q1 - q0) > thresh)
+ */
+ "subu_s.qb %[c], %[q1], %[q0] \n\t"
+ "subu_s.qb %[r_k], %[q0], %[q1] \n\t"
+ "or %[r_k], %[r_k], %[c] \n\t"
+ "cmpgu.lt.qb %[c], %[thresh], %[r_k] \n\t"
+ "or %[r3], %[r3], %[c] \n\t"
+ "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
+ "or %[r], %[r], %[c] \n\t"
+
+ /* mask |= (abs(q2 - q1) > limit) */
+ "subu_s.qb %[c], %[q2], %[q1] \n\t"
+ "subu_s.qb %[r_k], %[q1], %[q2] \n\t"
+ "or %[r_k], %[r_k], %[c] \n\t"
+ "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
+ "or %[r], %[r], %[c] \n\t"
+ "sll %[r3], %[r3], 24 \n\t"
+
+ /* mask |= (abs(q3 - q2) > limit) */
+ "subu_s.qb %[c], %[q3], %[q2] \n\t"
+ "subu_s.qb %[r_k], %[q2], %[q3] \n\t"
+ "or %[r_k], %[r_k], %[c] \n\t"
+ "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
+ "or %[r], %[r], %[c] \n\t"
+
+ : [c] "=&r" (c), [r_k] "=&r" (r_k),
+ [r] "=&r" (r), [r3] "=&r" (r3)
+ : [limit] "r" (limit), [p3] "r" (p3), [p2] "r" (p2),
+ [p1] "r" (p1), [p0] "r" (p0), [q1] "r" (q1), [q0] "r" (q0),
+ [q2] "r" (q2), [q3] "r" (q3), [thresh] "r" (thresh)
+ );
+
+ __asm__ __volatile__ (
+ /* abs(p0 - q0) */
+ "subu_s.qb %[c], %[p0], %[q0] \n\t"
+ "subu_s.qb %[r_k], %[q0], %[p0] \n\t"
+ "wrdsp %[r3] \n\t"
+ "or %[s1], %[r_k], %[c] \n\t"
+
+ /* abs(p1 - q1) */
+ "subu_s.qb %[c], %[p1], %[q1] \n\t"
+ "addu_s.qb %[s3], %[s1], %[s1] \n\t"
+ "pick.qb %[hev1], %[ones], $0 \n\t"
+ "subu_s.qb %[r_k], %[q1], %[p1] \n\t"
+ "or %[s2], %[r_k], %[c] \n\t"
+
+ /* abs(p0 - q0) * 2 + abs(p1 - q1) / 2 > flimit * 2 + limit */
+ "shrl.qb %[s2], %[s2], 1 \n\t"
+ "addu_s.qb %[s1], %[s2], %[s3] \n\t"
+ "cmpgu.lt.qb %[c], %[flimit], %[s1] \n\t"
+ "or %[r], %[r], %[c] \n\t"
+ "sll %[r], %[r], 24 \n\t"
+
+ "wrdsp %[r] \n\t"
+ "pick.qb %[s2], $0, %[ones] \n\t"
+
+ : [c] "=&r" (c), [r_k] "=&r" (r_k), [s1] "=&r" (s1), [hev1] "=&r" (hev1),
+ [s2] "=&r" (s2), [r] "+r" (r), [s3] "=&r" (s3)
+ : [p0] "r" (p0), [q0] "r" (q0), [p1] "r" (p1), [r3] "r" (r3),
+ [q1] "r" (q1), [ones] "r" (ones), [flimit] "r" (flimit)
+ );
+
+ *hev = hev1;
+ *mask = s2;
+}
+
+
+/* inputs & outputs are quad-byte vectors */
+static __inline void vp8_filter_mips
+(
+ uint32_t mask,
+ uint32_t hev,
+ uint32_t *ps1,
+ uint32_t *ps0,
+ uint32_t *qs0,
+ uint32_t *qs1
+)
+{
+ int32_t vp8_filter_l, vp8_filter_r;
+ int32_t Filter1_l, Filter1_r, Filter2_l, Filter2_r;
+ int32_t subr_r, subr_l;
+ uint32_t t1, t2, HWM, t3;
+ uint32_t hev_l, hev_r, mask_l, mask_r, invhev_l, invhev_r;
+
+ int32_t vps1, vps0, vqs0, vqs1;
+ int32_t vps1_l, vps1_r, vps0_l, vps0_r, vqs0_l, vqs0_r, vqs1_l, vqs1_r;
+ uint32_t N128;
+
+ N128 = 0x80808080;
+ t1 = 0x03000300;
+ t2 = 0x04000400;
+ t3 = 0x01000100;
+ HWM = 0xFF00FF00;
+
+ vps0 = (*ps0) ^ N128;
+ vps1 = (*ps1) ^ N128;
+ vqs0 = (*qs0) ^ N128;
+ vqs1 = (*qs1) ^ N128;
+
+ /* use halfword pairs instead quad-bytes because of accuracy */
+ vps0_l = vps0 & HWM;
+ vps0_r = vps0 << 8;
+ vps0_r = vps0_r & HWM;
+
+ vps1_l = vps1 & HWM;
+ vps1_r = vps1 << 8;
+ vps1_r = vps1_r & HWM;
+
+ vqs0_l = vqs0 & HWM;
+ vqs0_r = vqs0 << 8;
+ vqs0_r = vqs0_r & HWM;
+
+ vqs1_l = vqs1 & HWM;
+ vqs1_r = vqs1 << 8;
+ vqs1_r = vqs1_r & HWM;
+
+ mask_l = mask & HWM;
+ mask_r = mask << 8;
+ mask_r = mask_r & HWM;
+
+ hev_l = hev & HWM;
+ hev_r = hev << 8;
+ hev_r = hev_r & HWM;
+
+ __asm__ __volatile__ (
+ /* vp8_filter = vp8_signed_char_clamp(ps1 - qs1); */
+ "subq_s.ph %[vp8_filter_l], %[vps1_l], %[vqs1_l] \n\t"
+ "subq_s.ph %[vp8_filter_r], %[vps1_r], %[vqs1_r] \n\t"
+
+ /* qs0 - ps0 */
+ "subq_s.ph %[subr_l], %[vqs0_l], %[vps0_l] \n\t"
+ "subq_s.ph %[subr_r], %[vqs0_r], %[vps0_r] \n\t"
+
+ /* vp8_filter &= hev; */
+ "and %[vp8_filter_l], %[vp8_filter_l], %[hev_l] \n\t"
+ "and %[vp8_filter_r], %[vp8_filter_r], %[hev_r] \n\t"
+
+ /* vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0)); */
+ "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
+ "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
+ "xor %[invhev_l], %[hev_l], %[HWM] \n\t"
+ "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
+ "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
+ "xor %[invhev_r], %[hev_r], %[HWM] \n\t"
+ "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
+ "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
+
+ /* vp8_filter &= mask; */
+ "and %[vp8_filter_l], %[vp8_filter_l], %[mask_l] \n\t"
+ "and %[vp8_filter_r], %[vp8_filter_r], %[mask_r] \n\t"
+
+ : [vp8_filter_l] "=&r" (vp8_filter_l), [vp8_filter_r] "=&r" (vp8_filter_r),
+ [subr_l] "=&r" (subr_l), [subr_r] "=&r" (subr_r),
+ [invhev_l] "=&r" (invhev_l), [invhev_r] "=&r" (invhev_r)
+
+ : [vps0_l] "r" (vps0_l), [vps0_r] "r" (vps0_r), [vps1_l] "r" (vps1_l),
+ [vps1_r] "r" (vps1_r), [vqs0_l] "r" (vqs0_l), [vqs0_r] "r" (vqs0_r),
+ [vqs1_l] "r" (vqs1_l), [vqs1_r] "r" (vqs1_r),
+ [mask_l] "r" (mask_l), [mask_r] "r" (mask_r),
+ [hev_l] "r" (hev_l), [hev_r] "r" (hev_r),
+ [HWM] "r" (HWM)
+ );
+
+ /* save bottom 3 bits so that we round one side +4 and the other +3 */
+ __asm__ __volatile__ (
+ /* Filter2 = vp8_signed_char_clamp(vp8_filter + 3) >>= 3; */
+ "addq_s.ph %[Filter1_l], %[vp8_filter_l], %[t2] \n\t"
+ "addq_s.ph %[Filter1_r], %[vp8_filter_r], %[t2] \n\t"
+
+ /* Filter1 = vp8_signed_char_clamp(vp8_filter + 4) >>= 3; */
+ "addq_s.ph %[Filter2_l], %[vp8_filter_l], %[t1] \n\t"
+ "addq_s.ph %[Filter2_r], %[vp8_filter_r], %[t1] \n\t"
+ "shra.ph %[Filter1_r], %[Filter1_r], 3 \n\t"
+ "shra.ph %[Filter1_l], %[Filter1_l], 3 \n\t"
+
+ "shra.ph %[Filter2_l], %[Filter2_l], 3 \n\t"
+ "shra.ph %[Filter2_r], %[Filter2_r], 3 \n\t"
+
+ "and %[Filter1_l], %[Filter1_l], %[HWM] \n\t"
+ "and %[Filter1_r], %[Filter1_r], %[HWM] \n\t"
+
+ /* vps0 = vp8_signed_char_clamp(ps0 + Filter2); */
+ "addq_s.ph %[vps0_l], %[vps0_l], %[Filter2_l] \n\t"
+ "addq_s.ph %[vps0_r], %[vps0_r], %[Filter2_r] \n\t"
+
+ /* vqs0 = vp8_signed_char_clamp(qs0 - Filter1); */
+ "subq_s.ph %[vqs0_l], %[vqs0_l], %[Filter1_l] \n\t"
+ "subq_s.ph %[vqs0_r], %[vqs0_r], %[Filter1_r] \n\t"
+
+ : [Filter1_l] "=&r" (Filter1_l), [Filter1_r] "=&r" (Filter1_r),
+ [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r),
+ [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
+ [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)
+
+ : [t1] "r" (t1), [t2] "r" (t2),
+ [vp8_filter_l] "r" (vp8_filter_l), [vp8_filter_r] "r" (vp8_filter_r),
+ [HWM] "r" (HWM)
+ );
+
+ __asm__ __volatile__ (
+ /* (vp8_filter += 1) >>= 1 */
+ "addqh.ph %[Filter1_l], %[Filter1_l], %[t3] \n\t"
+ "addqh.ph %[Filter1_r], %[Filter1_r], %[t3] \n\t"
+
+ /* vp8_filter &= ~hev; */
+ "and %[Filter1_l], %[Filter1_l], %[invhev_l] \n\t"
+ "and %[Filter1_r], %[Filter1_r], %[invhev_r] \n\t"
+
+ /* vps1 = vp8_signed_char_clamp(ps1 + vp8_filter); */
+ "addq_s.ph %[vps1_l], %[vps1_l], %[Filter1_l] \n\t"
+ "addq_s.ph %[vps1_r], %[vps1_r], %[Filter1_r] \n\t"
+
+ /* vqs1 = vp8_signed_char_clamp(qs1 - vp8_filter); */
+ "subq_s.ph %[vqs1_l], %[vqs1_l], %[Filter1_l] \n\t"
+ "subq_s.ph %[vqs1_r], %[vqs1_r], %[Filter1_r] \n\t"
+
+ : [Filter1_l] "+r" (Filter1_l), [Filter1_r] "+r" (Filter1_r),
+ [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r),
+ [vqs1_l] "+r" (vqs1_l), [vqs1_r] "+r" (vqs1_r)
+
+ : [t3] "r" (t3), [invhev_l] "r" (invhev_l), [invhev_r] "r" (invhev_r)
+ );
+
+ /* Create quad-bytes from halfword pairs */
+ vqs0_l = vqs0_l & HWM;
+ vqs1_l = vqs1_l & HWM;
+ vps0_l = vps0_l & HWM;
+ vps1_l = vps1_l & HWM;
+
+ __asm__ __volatile__ (
+ "shrl.ph %[vqs0_r], %[vqs0_r], 8 \n\t"
+ "shrl.ph %[vps0_r], %[vps0_r], 8 \n\t"
+ "shrl.ph %[vqs1_r], %[vqs1_r], 8 \n\t"
+ "shrl.ph %[vps1_r], %[vps1_r], 8 \n\t"
+
+ : [vps1_r] "+r" (vps1_r), [vqs1_r] "+r" (vqs1_r),
+ [vps0_r] "+r" (vps0_r), [vqs0_r] "+r" (vqs0_r)
+ :
+ );
+
+ vqs0 = vqs0_l | vqs0_r;
+ vqs1 = vqs1_l | vqs1_r;
+ vps0 = vps0_l | vps0_r;
+ vps1 = vps1_l | vps1_r;
+
+ *ps0 = vps0 ^ N128;
+ *ps1 = vps1 ^ N128;
+ *qs0 = vqs0 ^ N128;
+ *qs1 = vqs1 ^ N128;
+}
+
+void vp8_loop_filter_horizontal_edge_mips
+(
+ unsigned char *s,
+ int p,
+ unsigned int flimit,
+ unsigned int limit,
+ unsigned int thresh,
+ int count
+)
+{
+ uint32_t mask;
+ uint32_t hev;
+ uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
+ unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
+
+ mask = 0;
+ hev = 0;
+ p1 = 0;
+ p2 = 0;
+ p3 = 0;
+ p4 = 0;
+
+ /* prefetch data for store */
+ prefetch_store_lf(s);
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+
+ sm1 = s - (p << 2);
+ s0 = s - p - p - p;
+ s1 = s - p - p ;
+ s2 = s - p;
+ s3 = s;
+ s4 = s + p;
+ s5 = s + p + p;
+ s6 = s + p + p + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood */
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ }
+ }
+
+ sm1 += 4;
+ s0 += 4;
+ s1 += 4;
+ s2 += 4;
+ s3 += 4;
+ s4 += 4;
+ s5 += 4;
+ s6 += 4;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood */
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ }
+ }
+
+ sm1 += 4;
+ s0 += 4;
+ s1 += 4;
+ s2 += 4;
+ s3 += 4;
+ s4 += 4;
+ s5 += 4;
+ s6 += 4;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood */
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ }
+ }
+
+ sm1 += 4;
+ s0 += 4;
+ s1 += 4;
+ s2 += 4;
+ s3 += 4;
+ s4 += 4;
+ s5 += 4;
+ s6 += 4;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood */
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ }
+ }
+}
+
+void vp8_loop_filter_uvhorizontal_edge_mips
+(
+ unsigned char *s,
+ int p,
+ unsigned int flimit,
+ unsigned int limit,
+ unsigned int thresh,
+ int count
+)
+{
+ uint32_t mask;
+ uint32_t hev;
+ uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
+ unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
+
+ mask = 0;
+ hev = 0;
+ p1 = 0;
+ p2 = 0;
+ p3 = 0;
+ p4 = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+
+ sm1 = s - (p << 2);
+ s0 = s - p - p - p;
+ s1 = s - p - p ;
+ s2 = s - p;
+ s3 = s;
+ s4 = s + p;
+ s5 = s + p + p;
+ s6 = s + p + p + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood */
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ }
+ }
+
+ sm1 += 4;
+ s0 += 4;
+ s1 += 4;
+ s2 += 4;
+ s3 += 4;
+ s4 += 4;
+ s5 += 4;
+ s6 += 4;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood */
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ }
+ }
+}
+
+void vp8_loop_filter_vertical_edge_mips
+(
+ unsigned char *s,
+ int p,
+ const unsigned int flimit,
+ const unsigned int limit,
+ const unsigned int thresh,
+ int count
+)
+{
+ int i;
+ uint32_t mask, hev;
+ uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
+ unsigned char *s1, *s2, *s3, *s4;
+ uint32_t prim1, prim2, sec3, sec4, prim3, prim4;
+
+ hev = 0;
+ mask = 0;
+ i = 0;
+ pm1 = 0;
+ p0 = 0;
+ p1 = 0;
+ p2 = 0;
+ p3 = 0;
+ p4 = 0;
+ p5 = 0;
+ p6 = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+
+ /* apply filter on 4 pixesl at the same time */
+ do
+ {
+
+ /* prefetch data for store */
+ prefetch_store_lf(s + p);
+
+ s1 = s;
+ s2 = s + p;
+ s3 = s2 + p;
+ s4 = s3 + p;
+ s = s4 + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p2 = *((uint32_t *)(s1 - 4));
+ p6 = *((uint32_t *)(s1));
+ p1 = *((uint32_t *)(s2 - 4));
+ p5 = *((uint32_t *)(s2));
+ p0 = *((uint32_t *)(s3 - 4));
+ p4 = *((uint32_t *)(s3));
+ pm1 = *((uint32_t *)(s4 - 4));
+ p3 = *((uint32_t *)(s4));
+
+ /* transpose pm1, p0, p1, p2 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
+ "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
+ "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
+ "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
+
+ "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
+ "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
+ "append %[p1], %[sec3], 16 \n\t"
+ "append %[pm1], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* transpose p3, p4, p5, p6 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
+ "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
+ "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
+ "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
+
+ "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
+ "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
+ "append %[p5], %[sec3], 16 \n\t"
+ "append %[p3], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood
+ * don't use transpose on output data
+ * because memory isn't aligned
+ */
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s4]) \n\t"
+ "sb %[p3], 0(%[s4]) \n\t"
+ "sb %[p2], -1(%[s4]) \n\t"
+ "sb %[p1], -2(%[s4]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s3]) \n\t"
+ "sb %[p3], 0(%[s3]) \n\t"
+ "sb %[p2], -1(%[s3]) \n\t"
+ "sb %[p1], -2(%[s3]) \n\t"
+ : [p1] "+r" (p1)
+ : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s2]) \n\t"
+ "sb %[p3], 0(%[s2]) \n\t"
+ "sb %[p2], -1(%[s2]) \n\t"
+ "sb %[p1], -2(%[s2]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s1]) \n\t"
+ "sb %[p3], 0(%[s1]) \n\t"
+ "sb %[p2], -1(%[s1]) \n\t"
+ "sb %[p1], -2(%[s1]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+ }
+ }
+
+ s1 = s;
+ s2 = s + p;
+ s3 = s2 + p;
+ s4 = s3 + p;
+ s = s4 + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p2 = *((uint32_t *)(s1 - 4));
+ p6 = *((uint32_t *)(s1));
+ p1 = *((uint32_t *)(s2 - 4));
+ p5 = *((uint32_t *)(s2));
+ p0 = *((uint32_t *)(s3 - 4));
+ p4 = *((uint32_t *)(s3));
+ pm1 = *((uint32_t *)(s4 - 4));
+ p3 = *((uint32_t *)(s4));
+
+ /* transpose pm1, p0, p1, p2 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
+ "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
+ "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
+ "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
+
+ "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
+ "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
+ "append %[p1], %[sec3], 16 \n\t"
+ "append %[pm1], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* transpose p3, p4, p5, p6 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
+ "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
+ "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
+ "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
+
+ "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
+ "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
+ "append %[p5], %[sec3], 16 \n\t"
+ "append %[p3], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood
+ * don't use transpose on output data
+ * because memory isn't aligned
+ */
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s4]) \n\t"
+ "sb %[p3], 0(%[s4]) \n\t"
+ "sb %[p2], -1(%[s4]) \n\t"
+ "sb %[p1], -2(%[s4]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s3]) \n\t"
+ "sb %[p3], 0(%[s3]) \n\t"
+ "sb %[p2], -1(%[s3]) \n\t"
+ "sb %[p1], -2(%[s3]) \n\t"
+ : [p1] "+r" (p1)
+ : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s2]) \n\t"
+ "sb %[p3], 0(%[s2]) \n\t"
+ "sb %[p2], -1(%[s2]) \n\t"
+ "sb %[p1], -2(%[s2]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s1]) \n\t"
+ "sb %[p3], 0(%[s1]) \n\t"
+ "sb %[p2], -1(%[s1]) \n\t"
+ "sb %[p1], -2(%[s1]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+ }
+ }
+
+ i += 8;
+ }
+
+ while (i < count);
+}
+
+void vp8_loop_filter_uvvertical_edge_mips
+(
+ unsigned char *s,
+ int p,
+ unsigned int flimit,
+ unsigned int limit,
+ unsigned int thresh,
+ int count
+)
+{
+ uint32_t mask, hev;
+ uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
+ unsigned char *s1, *s2, *s3, *s4;
+ uint32_t prim1, prim2, sec3, sec4, prim3, prim4;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+
+ /* apply filter on 4 pixesl at the same time */
+
+ s1 = s;
+ s2 = s + p;
+ s3 = s2 + p;
+ s4 = s3 + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p2 = *((uint32_t *)(s1 - 4));
+ p6 = *((uint32_t *)(s1));
+ p1 = *((uint32_t *)(s2 - 4));
+ p5 = *((uint32_t *)(s2));
+ p0 = *((uint32_t *)(s3 - 4));
+ p4 = *((uint32_t *)(s3));
+ pm1 = *((uint32_t *)(s4 - 4));
+ p3 = *((uint32_t *)(s4));
+
+ /* transpose pm1, p0, p1, p2 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
+ "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
+ "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
+ "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
+
+ "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
+ "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
+ "append %[p1], %[sec3], 16 \n\t"
+ "append %[pm1], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* transpose p3, p4, p5, p6 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
+ "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
+ "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
+ "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
+
+ "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
+ "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
+ "append %[p5], %[sec3], 16 \n\t"
+ "append %[p3], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood
+ * don't use transpose on output data
+ * because memory isn't aligned
+ */
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s4]) \n\t"
+ "sb %[p3], 0(%[s4]) \n\t"
+ "sb %[p2], -1(%[s4]) \n\t"
+ "sb %[p1], -2(%[s4]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s3]) \n\t"
+ "sb %[p3], 0(%[s3]) \n\t"
+ "sb %[p2], -1(%[s3]) \n\t"
+ "sb %[p1], -2(%[s3]) \n\t"
+ : [p1] "+r" (p1)
+ : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s2]) \n\t"
+ "sb %[p3], 0(%[s2]) \n\t"
+ "sb %[p2], -1(%[s2]) \n\t"
+ "sb %[p1], -2(%[s2]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s1]) \n\t"
+ "sb %[p3], 0(%[s1]) \n\t"
+ "sb %[p2], -1(%[s1]) \n\t"
+ "sb %[p1], -2(%[s1]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1), [p2] "r" (p2), [p1] "r" (p1)
+ );
+ }
+ }
+
+ s1 = s4 + p;
+ s2 = s1 + p;
+ s3 = s2 + p;
+ s4 = s3 + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p2 = *((uint32_t *)(s1 - 4));
+ p6 = *((uint32_t *)(s1));
+ p1 = *((uint32_t *)(s2 - 4));
+ p5 = *((uint32_t *)(s2));
+ p0 = *((uint32_t *)(s3 - 4));
+ p4 = *((uint32_t *)(s3));
+ pm1 = *((uint32_t *)(s4 - 4));
+ p3 = *((uint32_t *)(s4));
+
+ /* transpose pm1, p0, p1, p2 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
+ "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
+ "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
+ "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
+
+ "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
+ "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
+ "append %[p1], %[sec3], 16 \n\t"
+ "append %[pm1], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* transpose p3, p4, p5, p6 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
+ "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
+ "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
+ "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
+
+ "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
+ "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
+ "append %[p5], %[sec3], 16 \n\t"
+ "append %[p3], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
+
+ /* unpack processed 4x4 neighborhood
+ * don't use transpose on output data
+ * because memory isn't aligned
+ */
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s4]) \n\t"
+ "sb %[p3], 0(%[s4]) \n\t"
+ "sb %[p2], -1(%[s4]) \n\t"
+ "sb %[p1], -2(%[s4]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s3]) \n\t"
+ "sb %[p3], 0(%[s3]) \n\t"
+ "sb %[p2], -1(%[s3]) \n\t"
+ "sb %[p1], -2(%[s3]) \n\t"
+ : [p1] "+r" (p1)
+ : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s2]) \n\t"
+ "sb %[p3], 0(%[s2]) \n\t"
+ "sb %[p2], -1(%[s2]) \n\t"
+ "sb %[p1], -2(%[s2]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p4], 1(%[s1]) \n\t"
+ "sb %[p3], 0(%[s1]) \n\t"
+ "sb %[p2], -1(%[s1]) \n\t"
+ "sb %[p1], -2(%[s1]) \n\t"
+ :
+ : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
+ [p2] "r" (p2), [p1] "r" (p1)
+ );
+ }
+ }
+}
+
+/* inputs & outputs are quad-byte vectors */
+static __inline void vp8_mbfilter_mips
+(
+ uint32_t mask,
+ uint32_t hev,
+ uint32_t *ps2,
+ uint32_t *ps1,
+ uint32_t *ps0,
+ uint32_t *qs0,
+ uint32_t *qs1,
+ uint32_t *qs2
+)
+{
+ int32_t vps2, vps1, vps0, vqs0, vqs1, vqs2;
+ int32_t vps2_l, vps1_l, vps0_l, vqs0_l, vqs1_l, vqs2_l;
+ int32_t vps2_r, vps1_r, vps0_r, vqs0_r, vqs1_r, vqs2_r;
+ uint32_t HWM, vp8_filter_l, vp8_filter_r, mask_l, mask_r, hev_l, hev_r, subr_r, subr_l;
+ uint32_t Filter2_l, Filter2_r, t1, t2, Filter1_l, Filter1_r, invhev_l, invhev_r;
+ uint32_t N128, R63;
+ uint32_t u1_l, u1_r, u2_l, u2_r, u3_l, u3_r;
+
+ R63 = 0x003F003F;
+ HWM = 0xFF00FF00;
+ N128 = 0x80808080;
+ t1 = 0x03000300;
+ t2 = 0x04000400;
+
+ vps0 = (*ps0) ^ N128;
+ vps1 = (*ps1) ^ N128;
+ vps2 = (*ps2) ^ N128;
+ vqs0 = (*qs0) ^ N128;
+ vqs1 = (*qs1) ^ N128;
+ vqs2 = (*qs2) ^ N128;
+
+ /* use halfword pairs instead quad-bytes because of accuracy */
+ vps0_l = vps0 & HWM;
+ vps0_r = vps0 << 8;
+ vps0_r = vps0_r & HWM;
+
+ vqs0_l = vqs0 & HWM;
+ vqs0_r = vqs0 << 8;
+ vqs0_r = vqs0_r & HWM;
+
+ vps1_l = vps1 & HWM;
+ vps1_r = vps1 << 8;
+ vps1_r = vps1_r & HWM;
+
+ vqs1_l = vqs1 & HWM;
+ vqs1_r = vqs1 << 8;
+ vqs1_r = vqs1_r & HWM;
+
+ vqs2_l = vqs2 & HWM;
+ vqs2_r = vqs2 << 8;
+ vqs2_r = vqs2_r & HWM;
+
+ __asm__ __volatile__ (
+ /* qs0 - ps0 */
+ "subq_s.ph %[subr_l], %[vqs0_l], %[vps0_l] \n\t"
+ "subq_s.ph %[subr_r], %[vqs0_r], %[vps0_r] \n\t"
+
+ /* vp8_filter = vp8_signed_char_clamp(ps1 - qs1); */
+ "subq_s.ph %[vp8_filter_l], %[vps1_l], %[vqs1_l] \n\t"
+ "subq_s.ph %[vp8_filter_r], %[vps1_r], %[vqs1_r] \n\t"
+
+ : [vp8_filter_l] "=&r" (vp8_filter_l), [vp8_filter_r] "=r" (vp8_filter_r),
+ [subr_l] "=&r" (subr_l), [subr_r] "=&r" (subr_r)
+ : [vps0_l] "r" (vps0_l), [vps0_r] "r" (vps0_r), [vps1_l] "r" (vps1_l),
+ [vps1_r] "r" (vps1_r), [vqs0_l] "r" (vqs0_l), [vqs0_r] "r" (vqs0_r),
+ [vqs1_l] "r" (vqs1_l), [vqs1_r] "r" (vqs1_r)
+ );
+
+ vps2_l = vps2 & HWM;
+ vps2_r = vps2 << 8;
+ vps2_r = vps2_r & HWM;
+
+ /* add outer taps if we have high edge variance */
+ __asm__ __volatile__ (
+ /* vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0)); */
+ "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
+ "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
+ "and %[mask_l], %[HWM], %[mask] \n\t"
+ "sll %[mask_r], %[mask], 8 \n\t"
+ "and %[mask_r], %[HWM], %[mask_r] \n\t"
+ "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
+ "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
+ "and %[hev_l], %[HWM], %[hev] \n\t"
+ "sll %[hev_r], %[hev], 8 \n\t"
+ "and %[hev_r], %[HWM], %[hev_r] \n\t"
+ "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
+ "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
+
+ /* vp8_filter &= mask; */
+ "and %[vp8_filter_l], %[vp8_filter_l], %[mask_l] \n\t"
+ "and %[vp8_filter_r], %[vp8_filter_r], %[mask_r] \n\t"
+
+ /* Filter2 = vp8_filter & hev; */
+ "and %[Filter2_l], %[vp8_filter_l], %[hev_l] \n\t"
+ "and %[Filter2_r], %[vp8_filter_r], %[hev_r] \n\t"
+
+ : [vp8_filter_l] "+r" (vp8_filter_l), [vp8_filter_r] "+r" (vp8_filter_r),
+ [hev_l] "=&r" (hev_l), [hev_r] "=&r" (hev_r),
+ [mask_l] "=&r" (mask_l), [mask_r] "=&r" (mask_r),
+ [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r)
+ : [subr_l] "r" (subr_l), [subr_r] "r" (subr_r),
+ [HWM] "r" (HWM), [hev] "r" (hev), [mask] "r" (mask)
+ );
+
+ /* save bottom 3 bits so that we round one side +4 and the other +3 */
+ __asm__ __volatile__ (
+ /* Filter1 = vp8_signed_char_clamp(Filter2 + 4) >>= 3; */
+ "addq_s.ph %[Filter1_l], %[Filter2_l], %[t2] \n\t"
+ "xor %[invhev_l], %[hev_l], %[HWM] \n\t"
+ "addq_s.ph %[Filter1_r], %[Filter2_r], %[t2] \n\t"
+
+ /* Filter2 = vp8_signed_char_clamp(Filter2 + 3) >>= 3; */
+ "addq_s.ph %[Filter2_l], %[Filter2_l], %[t1] \n\t"
+ "addq_s.ph %[Filter2_r], %[Filter2_r], %[t1] \n\t"
+
+ "shra.ph %[Filter1_l], %[Filter1_l], 3 \n\t"
+ "shra.ph %[Filter1_r], %[Filter1_r], 3 \n\t"
+
+ "shra.ph %[Filter2_l], %[Filter2_l], 3 \n\t"
+ "shra.ph %[Filter2_r], %[Filter2_r], 3 \n\t"
+ "and %[Filter1_l], %[Filter1_l], %[HWM] \n\t"
+ "and %[Filter1_r], %[Filter1_r], %[HWM] \n\t"
+ "xor %[invhev_r], %[hev_r], %[HWM] \n\t"
+
+ /* qs0 = vp8_signed_char_clamp(qs0 - Filter1); */
+ "subq_s.ph %[vqs0_l], %[vqs0_l], %[Filter1_l] \n\t"
+ "subq_s.ph %[vqs0_r], %[vqs0_r], %[Filter1_r] \n\t"
+
+ /* ps0 = vp8_signed_char_clamp(ps0 + Filter2); */
+ "addq_s.ph %[vps0_l], %[vps0_l], %[Filter2_l] \n\t"
+ "addq_s.ph %[vps0_r], %[vps0_r], %[Filter2_r] \n\t"
+
+ : [invhev_l] "=&r" (invhev_l), [invhev_r] "=&r" (invhev_r),
+ [Filter1_l] "=&r" (Filter1_l), [Filter1_r] "=&r" (Filter1_r),
+ [Filter2_l] "+r" (Filter2_l), [Filter2_r] "+r" (Filter2_r),
+ [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
+ [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)
+ : [t1] "r" (t1), [t2] "r" (t2), [HWM] "r" (HWM),
+ [hev_l] "r" (hev_l), [hev_r] "r" (hev_r)
+ );
+
+ /* only apply wider filter if not high edge variance */
+ __asm__ __volatile__ (
+ /* vp8_filter &= ~hev; */
+ "and %[Filter2_l], %[vp8_filter_l], %[invhev_l] \n\t"
+ "and %[Filter2_r], %[vp8_filter_r], %[invhev_r] \n\t"
+
+ "shra.ph %[Filter2_l], %[Filter2_l], 8 \n\t"
+ "shra.ph %[Filter2_r], %[Filter2_r], 8 \n\t"
+
+ : [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r)
+ : [vp8_filter_l] "r" (vp8_filter_l), [vp8_filter_r] "r" (vp8_filter_r),
+ [invhev_l] "r" (invhev_l), [invhev_r] "r" (invhev_r)
+ );
+
+ /* roughly 3/7th difference across boundary */
+ __asm__ __volatile__ (
+ "shll.ph %[u3_l], %[Filter2_l], 3 \n\t"
+ "shll.ph %[u3_r], %[Filter2_r], 3 \n\t"
+
+ "addq.ph %[u3_l], %[u3_l], %[Filter2_l] \n\t"
+ "addq.ph %[u3_r], %[u3_r], %[Filter2_r] \n\t"
+
+ "shll.ph %[u2_l], %[u3_l], 1 \n\t"
+ "shll.ph %[u2_r], %[u3_r], 1 \n\t"
+
+ "addq.ph %[u1_l], %[u3_l], %[u2_l] \n\t"
+ "addq.ph %[u1_r], %[u3_r], %[u2_r] \n\t"
+
+ "addq.ph %[u2_l], %[u2_l], %[R63] \n\t"
+ "addq.ph %[u2_r], %[u2_r], %[R63] \n\t"
+
+ "addq.ph %[u3_l], %[u3_l], %[R63] \n\t"
+ "addq.ph %[u3_r], %[u3_r], %[R63] \n\t"
+
+ /* vp8_signed_char_clamp((63 + Filter2 * 27) >> 7)
+ * vp8_signed_char_clamp((63 + Filter2 * 18) >> 7)
+ */
+ "addq.ph %[u1_l], %[u1_l], %[R63] \n\t"
+ "addq.ph %[u1_r], %[u1_r], %[R63] \n\t"
+ "shra.ph %[u1_l], %[u1_l], 7 \n\t"
+ "shra.ph %[u1_r], %[u1_r], 7 \n\t"
+ "shra.ph %[u2_l], %[u2_l], 7 \n\t"
+ "shra.ph %[u2_r], %[u2_r], 7 \n\t"
+ "shll.ph %[u1_l], %[u1_l], 8 \n\t"
+ "shll.ph %[u1_r], %[u1_r], 8 \n\t"
+ "shll.ph %[u2_l], %[u2_l], 8 \n\t"
+ "shll.ph %[u2_r], %[u2_r], 8 \n\t"
+
+ /* vqs0 = vp8_signed_char_clamp(qs0 - u); */
+ "subq_s.ph %[vqs0_l], %[vqs0_l], %[u1_l] \n\t"
+ "subq_s.ph %[vqs0_r], %[vqs0_r], %[u1_r] \n\t"
+
+ /* vps0 = vp8_signed_char_clamp(ps0 + u); */
+ "addq_s.ph %[vps0_l], %[vps0_l], %[u1_l] \n\t"
+ "addq_s.ph %[vps0_r], %[vps0_r], %[u1_r] \n\t"
+
+ : [u1_l] "=&r" (u1_l), [u1_r] "=&r" (u1_r), [u2_l] "=&r" (u2_l),
+ [u2_r] "=&r" (u2_r), [u3_l] "=&r" (u3_l), [u3_r] "=&r" (u3_r),
+ [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
+ [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)
+ : [R63] "r" (R63),
+ [Filter2_l] "r" (Filter2_l), [Filter2_r] "r" (Filter2_r)
+ );
+
+ __asm__ __volatile__ (
+ /* vqs1 = vp8_signed_char_clamp(qs1 - u); */
+ "subq_s.ph %[vqs1_l], %[vqs1_l], %[u2_l] \n\t"
+ "addq_s.ph %[vps1_l], %[vps1_l], %[u2_l] \n\t"
+
+ /* vps1 = vp8_signed_char_clamp(ps1 + u); */
+ "addq_s.ph %[vps1_r], %[vps1_r], %[u2_r] \n\t"
+ "subq_s.ph %[vqs1_r], %[vqs1_r], %[u2_r] \n\t"
+
+ : [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r),
+ [vqs1_l] "+r" (vqs1_l), [vqs1_r] "+r" (vqs1_r)
+ : [u2_l] "r" (u2_l), [u2_r] "r" (u2_r)
+ );
+
+ /* roughly 1/7th difference across boundary */
+ __asm__ __volatile__ (
+ /* u = vp8_signed_char_clamp((63 + Filter2 * 9) >> 7); */
+ "shra.ph %[u3_l], %[u3_l], 7 \n\t"
+ "shra.ph %[u3_r], %[u3_r], 7 \n\t"
+ "shll.ph %[u3_l], %[u3_l], 8 \n\t"
+ "shll.ph %[u3_r], %[u3_r], 8 \n\t"
+
+ /* vqs2 = vp8_signed_char_clamp(qs2 - u); */
+ "subq_s.ph %[vqs2_l], %[vqs2_l], %[u3_l] \n\t"
+ "subq_s.ph %[vqs2_r], %[vqs2_r], %[u3_r] \n\t"
+
+ /* vps2 = vp8_signed_char_clamp(ps2 + u); */
+ "addq_s.ph %[vps2_l], %[vps2_l], %[u3_l] \n\t"
+ "addq_s.ph %[vps2_r], %[vps2_r], %[u3_r] \n\t"
+
+ : [u3_l] "+r" (u3_l), [u3_r] "+r" (u3_r), [vps2_l] "+r" (vps2_l),
+ [vps2_r] "+r" (vps2_r), [vqs2_l] "+r" (vqs2_l), [vqs2_r] "+r" (vqs2_r)
+ :
+ );
+
+ /* Create quad-bytes from halfword pairs */
+ __asm__ __volatile__ (
+ "and %[vqs0_l], %[vqs0_l], %[HWM] \n\t"
+ "shrl.ph %[vqs0_r], %[vqs0_r], 8 \n\t"
+
+ "and %[vps0_l], %[vps0_l], %[HWM] \n\t"
+ "shrl.ph %[vps0_r], %[vps0_r], 8 \n\t"
+
+ "and %[vqs1_l], %[vqs1_l], %[HWM] \n\t"
+ "shrl.ph %[vqs1_r], %[vqs1_r], 8 \n\t"
+
+ "and %[vps1_l], %[vps1_l], %[HWM] \n\t"
+ "shrl.ph %[vps1_r], %[vps1_r], 8 \n\t"
+
+ "and %[vqs2_l], %[vqs2_l], %[HWM] \n\t"
+ "shrl.ph %[vqs2_r], %[vqs2_r], 8 \n\t"
+
+ "and %[vps2_l], %[vps2_l], %[HWM] \n\t"
+ "shrl.ph %[vps2_r], %[vps2_r], 8 \n\t"
+
+ "or %[vqs0_r], %[vqs0_l], %[vqs0_r] \n\t"
+ "or %[vps0_r], %[vps0_l], %[vps0_r] \n\t"
+ "or %[vqs1_r], %[vqs1_l], %[vqs1_r] \n\t"
+ "or %[vps1_r], %[vps1_l], %[vps1_r] \n\t"
+ "or %[vqs2_r], %[vqs2_l], %[vqs2_r] \n\t"
+ "or %[vps2_r], %[vps2_l], %[vps2_r] \n\t"
+
+ : [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r), [vqs1_l] "+r" (vqs1_l),
+ [vqs1_r] "+r" (vqs1_r), [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
+ [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r), [vqs2_l] "+r" (vqs2_l),
+ [vqs2_r] "+r" (vqs2_r), [vps2_r] "+r" (vps2_r), [vps2_l] "+r" (vps2_l)
+ : [HWM] "r" (HWM)
+ );
+
+ *ps0 = vps0_r ^ N128;
+ *ps1 = vps1_r ^ N128;
+ *ps2 = vps2_r ^ N128;
+ *qs0 = vqs0_r ^ N128;
+ *qs1 = vqs1_r ^ N128;
+ *qs2 = vqs2_r ^ N128;
+}
+
+void vp8_mbloop_filter_horizontal_edge_mips
+(
+ unsigned char *s,
+ int p,
+ unsigned int flimit,
+ unsigned int limit,
+ unsigned int thresh,
+ int count
+)
+{
+ int i;
+ uint32_t mask, hev;
+ uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
+ unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
+
+ mask = 0;
+ hev = 0;
+ i = 0;
+ p1 = 0;
+ p2 = 0;
+ p3 = 0;
+ p4 = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+
+ sm1 = s - (p << 2);
+ s0 = s - p - p - p;
+ s1 = s - p - p;
+ s2 = s - p;
+ s3 = s;
+ s4 = s + p;
+ s5 = s + p + p;
+ s6 = s + p + p + p;
+
+ /* prefetch data for load */
+ prefetch_load_lf(s + p);
+
+ /* apply filter on 4 pixesl at the same time */
+ do
+ {
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
+
+ /* unpack processed 4x4 neighborhood
+ * memory is 4 byte aligned
+ */
+ *((uint32_t *)s0) = p0;
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ *((uint32_t *)s5) = p5;
+ }
+ }
+
+ sm1 += 4;
+ s0 += 4;
+ s1 += 4;
+ s2 += 4;
+ s3 += 4;
+ s4 += 4;
+ s5 += 4;
+ s6 += 4;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
+
+ /* unpack processed 4x4 neighborhood
+ * memory is 4 byte aligned
+ */
+ *((uint32_t *)s0) = p0;
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ *((uint32_t *)s5) = p5;
+ }
+ }
+
+ sm1 += 4;
+ s0 += 4;
+ s1 += 4;
+ s2 += 4;
+ s3 += 4;
+ s4 += 4;
+ s5 += 4;
+ s6 += 4;
+
+ i += 8;
+ }
+
+ while (i < count);
+}
+
+void vp8_mbloop_filter_uvhorizontal_edge_mips
+(
+ unsigned char *s,
+ int p,
+ unsigned int flimit,
+ unsigned int limit,
+ unsigned int thresh,
+ int count
+)
+{
+ uint32_t mask, hev;
+ uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
+ unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
+
+ mask = 0;
+ hev = 0;
+ p1 = 0;
+ p2 = 0;
+ p3 = 0;
+ p4 = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+
+ sm1 = s - (p << 2);
+ s0 = s - p - p - p;
+ s1 = s - p - p;
+ s2 = s - p;
+ s3 = s;
+ s4 = s + p;
+ s5 = s + p + p;
+ s6 = s + p + p + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ /* if mask == 0 do filtering is not needed */
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ if (mask)
+ {
+ /* filtering */
+ vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
+
+ /* unpack processed 4x4 neighborhood
+ * memory is 4 byte aligned
+ */
+ *((uint32_t *)s0) = p0;
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ *((uint32_t *)s5) = p5;
+ }
+ }
+
+ sm1 += 4;
+ s0 += 4;
+ s1 += 4;
+ s2 += 4;
+ s3 += 4;
+ s4 += 4;
+ s5 += 4;
+ s6 += 4;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p1 = *((uint32_t *)(s1));
+ p2 = *((uint32_t *)(s2));
+ p3 = *((uint32_t *)(s3));
+ p4 = *((uint32_t *)(s4));
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ pm1 = *((uint32_t *)(sm1));
+ p0 = *((uint32_t *)(s0));
+ p5 = *((uint32_t *)(s5));
+ p6 = *((uint32_t *)(s6));
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
+
+ /* unpack processed 4x4 neighborhood
+ * memory is 4 byte aligned
+ */
+ *((uint32_t *)s0) = p0;
+ *((uint32_t *)s1) = p1;
+ *((uint32_t *)s2) = p2;
+ *((uint32_t *)s3) = p3;
+ *((uint32_t *)s4) = p4;
+ *((uint32_t *)s5) = p5;
+ }
+ }
+}
+
+
+void vp8_mbloop_filter_vertical_edge_mips
+(
+ unsigned char *s,
+ int p,
+ unsigned int flimit,
+ unsigned int limit,
+ unsigned int thresh,
+ int count
+)
+{
+
+ int i;
+ uint32_t mask, hev;
+ uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
+ unsigned char *s1, *s2, *s3, *s4;
+ uint32_t prim1, prim2, sec3, sec4, prim3, prim4;
+
+ mask = 0;
+ hev = 0;
+ i = 0;
+ pm1 = 0;
+ p0 = 0;
+ p1 = 0;
+ p2 = 0;
+ p3 = 0;
+ p4 = 0;
+ p5 = 0;
+ p6 = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+
+ /* apply filter on 4 pixesl at the same time */
+ do
+ {
+ s1 = s;
+ s2 = s + p;
+ s3 = s2 + p;
+ s4 = s3 + p;
+ s = s4 + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p2 = *((uint32_t *)(s1 - 4));
+ p6 = *((uint32_t *)(s1));
+ p1 = *((uint32_t *)(s2 - 4));
+ p5 = *((uint32_t *)(s2));
+ p0 = *((uint32_t *)(s3 - 4));
+ p4 = *((uint32_t *)(s3));
+ pm1 = *((uint32_t *)(s4 - 4));
+ p3 = *((uint32_t *)(s4));
+
+ /* transpose pm1, p0, p1, p2 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
+ "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
+ "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
+ "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
+
+ "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
+ "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
+ "append %[p1], %[sec3], 16 \n\t"
+ "append %[pm1], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* transpose p3, p4, p5, p6 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
+ "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
+ "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
+ "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
+
+ "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
+ "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
+ "append %[p5], %[sec3], 16 \n\t"
+ "append %[p3], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
+
+ /* don't use transpose on output data
+ * because memory isn't aligned
+ */
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s4]) \n\t"
+ "sb %[p4], 1(%[s4]) \n\t"
+ "sb %[p3], 0(%[s4]) \n\t"
+ "sb %[p2], -1(%[s4]) \n\t"
+ "sb %[p1], -2(%[s4]) \n\t"
+ "sb %[p0], -3(%[s4]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s3]) \n\t"
+ "sb %[p4], 1(%[s3]) \n\t"
+ "sb %[p3], 0(%[s3]) \n\t"
+ "sb %[p2], -1(%[s3]) \n\t"
+ "sb %[p1], -2(%[s3]) \n\t"
+ "sb %[p0], -3(%[s3]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s2]) \n\t"
+ "sb %[p4], 1(%[s2]) \n\t"
+ "sb %[p3], 0(%[s2]) \n\t"
+ "sb %[p2], -1(%[s2]) \n\t"
+ "sb %[p1], -2(%[s2]) \n\t"
+ "sb %[p0], -3(%[s2]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s1]) \n\t"
+ "sb %[p4], 1(%[s1]) \n\t"
+ "sb %[p3], 0(%[s1]) \n\t"
+ "sb %[p2], -1(%[s1]) \n\t"
+ "sb %[p1], -2(%[s1]) \n\t"
+ "sb %[p0], -3(%[s1]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+ }
+ }
+
+ i += 4;
+ }
+
+ while (i < count);
+}
+
+void vp8_mbloop_filter_uvvertical_edge_mips
+(
+ unsigned char *s,
+ int p,
+ unsigned int flimit,
+ unsigned int limit,
+ unsigned int thresh,
+ int count
+)
+{
+ uint32_t mask, hev;
+ uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
+ unsigned char *s1, *s2, *s3, *s4;
+ uint32_t prim1, prim2, sec3, sec4, prim3, prim4;
+
+ mask = 0;
+ hev = 0;
+ pm1 = 0;
+ p0 = 0;
+ p1 = 0;
+ p2 = 0;
+ p3 = 0;
+ p4 = 0;
+ p5 = 0;
+ p6 = 0;
+
+ /* loop filter designed to work using chars so that we can make maximum use
+ * of 8 bit simd instructions.
+ */
+
+ /* apply filter on 4 pixesl at the same time */
+
+ s1 = s;
+ s2 = s + p;
+ s3 = s2 + p;
+ s4 = s3 + p;
+
+ /* prefetch data for load */
+ prefetch_load_lf(s + 2 * p);
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p2 = *((uint32_t *)(s1 - 4));
+ p6 = *((uint32_t *)(s1));
+ p1 = *((uint32_t *)(s2 - 4));
+ p5 = *((uint32_t *)(s2));
+ p0 = *((uint32_t *)(s3 - 4));
+ p4 = *((uint32_t *)(s3));
+ pm1 = *((uint32_t *)(s4 - 4));
+ p3 = *((uint32_t *)(s4));
+
+ /* transpose pm1, p0, p1, p2 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
+ "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
+ "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
+ "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
+
+ "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
+ "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
+ "append %[p1], %[sec3], 16 \n\t"
+ "append %[pm1], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* transpose p3, p4, p5, p6 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
+ "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
+ "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
+ "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
+
+ "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
+ "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
+ "append %[p5], %[sec3], 16 \n\t"
+ "append %[p3], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
+ thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
+
+ /* don't use transpose on output data
+ * because memory isn't aligned
+ */
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s4]) \n\t"
+ "sb %[p4], 1(%[s4]) \n\t"
+ "sb %[p3], 0(%[s4]) \n\t"
+ "sb %[p2], -1(%[s4]) \n\t"
+ "sb %[p1], -2(%[s4]) \n\t"
+ "sb %[p0], -3(%[s4]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s3]) \n\t"
+ "sb %[p4], 1(%[s3]) \n\t"
+ "sb %[p3], 0(%[s3]) \n\t"
+ "sb %[p2], -1(%[s3]) \n\t"
+ "sb %[p1], -2(%[s3]) \n\t"
+ "sb %[p0], -3(%[s3]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s2]) \n\t"
+ "sb %[p4], 1(%[s2]) \n\t"
+ "sb %[p3], 0(%[s2]) \n\t"
+ "sb %[p2], -1(%[s2]) \n\t"
+ "sb %[p1], -2(%[s2]) \n\t"
+ "sb %[p0], -3(%[s2]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s1]) \n\t"
+ "sb %[p4], 1(%[s1]) \n\t"
+ "sb %[p3], 0(%[s1]) \n\t"
+ "sb %[p2], -1(%[s1]) \n\t"
+ "sb %[p1], -2(%[s1]) \n\t"
+ "sb %[p0], -3(%[s1]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+ }
+ }
+
+ s1 = s4 + p;
+ s2 = s1 + p;
+ s3 = s2 + p;
+ s4 = s3 + p;
+
+ /* load quad-byte vectors
+ * memory is 4 byte aligned
+ */
+ p2 = *((uint32_t *)(s1 - 4));
+ p6 = *((uint32_t *)(s1));
+ p1 = *((uint32_t *)(s2 - 4));
+ p5 = *((uint32_t *)(s2));
+ p0 = *((uint32_t *)(s3 - 4));
+ p4 = *((uint32_t *)(s3));
+ pm1 = *((uint32_t *)(s4 - 4));
+ p3 = *((uint32_t *)(s4));
+
+ /* transpose pm1, p0, p1, p2 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
+ "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
+ "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
+ "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
+
+ "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
+ "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
+ "append %[p1], %[sec3], 16 \n\t"
+ "append %[pm1], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* transpose p3, p4, p5, p6 */
+ __asm__ __volatile__ (
+ "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
+ "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
+ "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
+ "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
+
+ "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
+ "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
+ "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
+ "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
+
+ "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
+ "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
+ "append %[p5], %[sec3], 16 \n\t"
+ "append %[p3], %[sec4], 16 \n\t"
+
+ : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
+ [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
+ [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
+ :
+ );
+
+ /* if (p1 - p4 == 0) and (p2 - p3 == 0)
+ * mask will be zero and filtering is not needed
+ */
+ if (!(((p1 - p4) == 0) && ((p2 - p3) == 0)))
+ {
+
+ vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6, thresh, &hev, &mask);
+
+ /* if mask == 0 do filtering is not needed */
+ if (mask)
+ {
+ /* filtering */
+ vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
+
+ /* don't use transpose on output data
+ * because memory isn't aligned
+ */
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s4]) \n\t"
+ "sb %[p4], 1(%[s4]) \n\t"
+ "sb %[p3], 0(%[s4]) \n\t"
+ "sb %[p2], -1(%[s4]) \n\t"
+ "sb %[p1], -2(%[s4]) \n\t"
+ "sb %[p0], -3(%[s4]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s3]) \n\t"
+ "sb %[p4], 1(%[s3]) \n\t"
+ "sb %[p3], 0(%[s3]) \n\t"
+ "sb %[p2], -1(%[s3]) \n\t"
+ "sb %[p1], -2(%[s3]) \n\t"
+ "sb %[p0], -3(%[s3]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s2]) \n\t"
+ "sb %[p4], 1(%[s2]) \n\t"
+ "sb %[p3], 0(%[s2]) \n\t"
+ "sb %[p2], -1(%[s2]) \n\t"
+ "sb %[p1], -2(%[s2]) \n\t"
+ "sb %[p0], -3(%[s2]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+
+ __asm__ __volatile__ (
+ "srl %[p5], %[p5], 8 \n\t"
+ "srl %[p4], %[p4], 8 \n\t"
+ "srl %[p3], %[p3], 8 \n\t"
+ "srl %[p2], %[p2], 8 \n\t"
+ "srl %[p1], %[p1], 8 \n\t"
+ "srl %[p0], %[p0], 8 \n\t"
+ : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
+ [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
+ :
+ );
+
+ __asm__ __volatile__ (
+ "sb %[p5], 2(%[s1]) \n\t"
+ "sb %[p4], 1(%[s1]) \n\t"
+ "sb %[p3], 0(%[s1]) \n\t"
+ "sb %[p2], -1(%[s1]) \n\t"
+ "sb %[p1], -2(%[s1]) \n\t"
+ "sb %[p0], -3(%[s1]) \n\t"
+ :
+ : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
+ [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
+ );
+ }
+ }
+}
+
+/* Horizontal MB filtering */
+void vp8_loop_filter_mbh_dspr2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ unsigned int thresh_vec, flimit_vec, limit_vec;
+ unsigned char thresh, flimit, limit, flimit_temp;
+
+ /* use direct value instead pointers */
+ limit = *(lfi->lim);
+ flimit_temp = *(lfi->mblim);
+ thresh = *(lfi->hev_thr);
+ flimit = flimit_temp;
+
+ /* create quad-byte */
+ __asm__ __volatile__ (
+ "replv.qb %[thresh_vec], %[thresh] \n\t"
+ "replv.qb %[flimit_vec], %[flimit] \n\t"
+ "replv.qb %[limit_vec], %[limit] \n\t"
+ : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
+ : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
+ );
+
+ vp8_mbloop_filter_horizontal_edge_mips(y_ptr, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
+
+ if (u_ptr)
+ {
+ vp8_mbloop_filter_uvhorizontal_edge_mips(u_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
+ }
+
+ if (v_ptr)
+ {
+ vp8_mbloop_filter_uvhorizontal_edge_mips(v_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
+ }
+}
+
+
+/* Vertical MB Filtering */
+void vp8_loop_filter_mbv_dspr2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ unsigned int thresh_vec, flimit_vec, limit_vec;
+ unsigned char thresh, flimit, limit, flimit_temp;
+
+ /* use direct value instead pointers */
+ limit = *(lfi->lim);
+ flimit_temp = *(lfi->mblim);
+ thresh = *(lfi->hev_thr);
+ flimit = flimit_temp;
+
+ /* create quad-byte */
+ __asm__ __volatile__ (
+ "replv.qb %[thresh_vec], %[thresh] \n\t"
+ "replv.qb %[flimit_vec], %[flimit] \n\t"
+ "replv.qb %[limit_vec], %[limit] \n\t"
+ : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
+ : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
+ );
+
+ vp8_mbloop_filter_vertical_edge_mips(y_ptr, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
+
+ if (u_ptr)
+ vp8_mbloop_filter_uvvertical_edge_mips(u_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
+
+ if (v_ptr)
+ vp8_mbloop_filter_uvvertical_edge_mips(v_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
+}
+
+
+/* Horizontal B Filtering */
+void vp8_loop_filter_bh_dspr2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ unsigned int thresh_vec, flimit_vec, limit_vec;
+ unsigned char thresh, flimit, limit, flimit_temp;
+
+ /* use direct value instead pointers */
+ limit = *(lfi->lim);
+ flimit_temp = *(lfi->blim);
+ thresh = *(lfi->hev_thr);
+ flimit = flimit_temp;
+
+ /* create quad-byte */
+ __asm__ __volatile__ (
+ "replv.qb %[thresh_vec], %[thresh] \n\t"
+ "replv.qb %[flimit_vec], %[flimit] \n\t"
+ "replv.qb %[limit_vec], %[limit] \n\t"
+ : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
+ : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
+ );
+
+ vp8_loop_filter_horizontal_edge_mips(y_ptr + 4 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
+ vp8_loop_filter_horizontal_edge_mips(y_ptr + 8 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
+ vp8_loop_filter_horizontal_edge_mips(y_ptr + 12 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
+
+ if (u_ptr)
+ vp8_loop_filter_uvhorizontal_edge_mips(u_ptr + 4 * uv_stride, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
+
+ if (v_ptr)
+ vp8_loop_filter_uvhorizontal_edge_mips(v_ptr + 4 * uv_stride, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
+}
+
+
+/* Vertical B Filtering */
+void vp8_loop_filter_bv_dspr2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ unsigned int thresh_vec, flimit_vec, limit_vec;
+ unsigned char thresh, flimit, limit, flimit_temp;
+
+ /* use direct value instead pointers */
+ limit = *(lfi->lim);
+ flimit_temp = *(lfi->blim);
+ thresh = *(lfi->hev_thr);
+ flimit = flimit_temp;
+
+ /* create quad-byte */
+ __asm__ __volatile__ (
+ "replv.qb %[thresh_vec], %[thresh] \n\t"
+ "replv.qb %[flimit_vec], %[flimit] \n\t"
+ "replv.qb %[limit_vec], %[limit] \n\t"
+ : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
+ : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
+ );
+
+ vp8_loop_filter_vertical_edge_mips(y_ptr + 4, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
+ vp8_loop_filter_vertical_edge_mips(y_ptr + 8, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
+ vp8_loop_filter_vertical_edge_mips(y_ptr + 12, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
+
+ if (u_ptr)
+ vp8_loop_filter_uvvertical_edge_mips(u_ptr + 4, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
+
+ if (v_ptr)
+ vp8_loop_filter_uvvertical_edge_mips(v_ptr + 4, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
+}
+
+#endif
diff --git a/libvpx/vp8/common/mips/dspr2/reconinter_dspr2.c b/libvpx/vp8/common/mips/dspr2/reconinter_dspr2.c
new file mode 100644
index 0000000..a5239a3
--- /dev/null
+++ b/libvpx/vp8/common/mips/dspr2/reconinter_dspr2.c
@@ -0,0 +1,121 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx/vpx_integer.h"
+
+#if HAVE_DSPR2
+inline void prefetch_load_int(unsigned char *src)
+{
+ __asm__ __volatile__ (
+ "pref 0, 0(%[src]) \n\t"
+ :
+ : [src] "r" (src)
+ );
+}
+
+
+__inline void vp8_copy_mem16x16_dspr2(
+ unsigned char *RESTRICT src,
+ int src_stride,
+ unsigned char *RESTRICT dst,
+ int dst_stride)
+{
+ int r;
+ unsigned int a0, a1, a2, a3;
+
+ for (r = 16; r--;)
+ {
+ /* load src data in cache memory */
+ prefetch_load_int(src + src_stride);
+
+ /* use unaligned memory load and store */
+ __asm__ __volatile__ (
+ "ulw %[a0], 0(%[src]) \n\t"
+ "ulw %[a1], 4(%[src]) \n\t"
+ "ulw %[a2], 8(%[src]) \n\t"
+ "ulw %[a3], 12(%[src]) \n\t"
+ "sw %[a0], 0(%[dst]) \n\t"
+ "sw %[a1], 4(%[dst]) \n\t"
+ "sw %[a2], 8(%[dst]) \n\t"
+ "sw %[a3], 12(%[dst]) \n\t"
+ : [a0] "=&r" (a0), [a1] "=&r" (a1),
+ [a2] "=&r" (a2), [a3] "=&r" (a3)
+ : [src] "r" (src), [dst] "r" (dst)
+ );
+
+ src += src_stride;
+ dst += dst_stride;
+ }
+}
+
+
+__inline void vp8_copy_mem8x8_dspr2(
+ unsigned char *RESTRICT src,
+ int src_stride,
+ unsigned char *RESTRICT dst,
+ int dst_stride)
+{
+ int r;
+ unsigned int a0, a1;
+
+ /* load src data in cache memory */
+ prefetch_load_int(src + src_stride);
+
+ for (r = 8; r--;)
+ {
+ /* use unaligned memory load and store */
+ __asm__ __volatile__ (
+ "ulw %[a0], 0(%[src]) \n\t"
+ "ulw %[a1], 4(%[src]) \n\t"
+ "sw %[a0], 0(%[dst]) \n\t"
+ "sw %[a1], 4(%[dst]) \n\t"
+ : [a0] "=&r" (a0), [a1] "=&r" (a1)
+ : [src] "r" (src), [dst] "r" (dst)
+ );
+
+ src += src_stride;
+ dst += dst_stride;
+ }
+}
+
+
+__inline void vp8_copy_mem8x4_dspr2(
+ unsigned char *RESTRICT src,
+ int src_stride,
+ unsigned char *RESTRICT dst,
+ int dst_stride)
+{
+ int r;
+ unsigned int a0, a1;
+
+ /* load src data in cache memory */
+ prefetch_load_int(src + src_stride);
+
+ for (r = 4; r--;)
+ {
+ /* use unaligned memory load and store */
+ __asm__ __volatile__ (
+ "ulw %[a0], 0(%[src]) \n\t"
+ "ulw %[a1], 4(%[src]) \n\t"
+ "sw %[a0], 0(%[dst]) \n\t"
+ "sw %[a1], 4(%[dst]) \n\t"
+ : [a0] "=&r" (a0), [a1] "=&r" (a1)
+ : [src] "r" (src), [dst] "r" (dst)
+ );
+
+ src += src_stride;
+ dst += dst_stride;
+ }
+}
+
+#endif
diff --git a/vp8/common/modecont.c b/libvpx/vp8/common/modecont.c
similarity index 100%
rename from vp8/common/modecont.c
rename to libvpx/vp8/common/modecont.c
diff --git a/vp8/common/modecont.h b/libvpx/vp8/common/modecont.h
similarity index 100%
rename from vp8/common/modecont.h
rename to libvpx/vp8/common/modecont.h
diff --git a/libvpx/vp8/common/mv.h b/libvpx/vp8/common/mv.h
new file mode 100644
index 0000000..b3f919d
--- /dev/null
+++ b/libvpx/vp8/common/mv.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_MV_H
+#define __INC_MV_H
+#include "vpx/vpx_integer.h"
+
+typedef struct
+{
+ short row;
+ short col;
+} MV;
+
+typedef union int_mv
+{
+ uint32_t as_int;
+ MV as_mv;
+} int_mv; /* facilitates faster equality tests and copies */
+
+#endif
diff --git a/libvpx/vp8/common/onyx.h b/libvpx/vp8/common/onyx.h
new file mode 100644
index 0000000..766b4ea
--- /dev/null
+++ b/libvpx/vp8/common/onyx.h
@@ -0,0 +1,269 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_VP8_H
+#define __INC_VP8_H
+
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+
+#include "vpx_config.h"
+#include "vpx/internal/vpx_codec_internal.h"
+#include "vpx/vp8cx.h"
+#include "vpx/vpx_encoder.h"
+#include "vpx_scale/yv12config.h"
+#include "ppflags.h"
+
+ struct VP8_COMP;
+
+ /* Create/destroy static data structures. */
+
+ typedef enum
+ {
+ NORMAL = 0,
+ FOURFIVE = 1,
+ THREEFIVE = 2,
+ ONETWO = 3
+
+ } VPX_SCALING;
+
+ typedef enum
+ {
+ USAGE_STREAM_FROM_SERVER = 0x0,
+ USAGE_LOCAL_FILE_PLAYBACK = 0x1,
+ USAGE_CONSTRAINED_QUALITY = 0x2
+ } END_USAGE;
+
+
+ typedef enum
+ {
+ MODE_REALTIME = 0x0,
+ MODE_GOODQUALITY = 0x1,
+ MODE_BESTQUALITY = 0x2,
+ MODE_FIRSTPASS = 0x3,
+ MODE_SECONDPASS = 0x4,
+ MODE_SECONDPASS_BEST = 0x5
+ } MODE;
+
+ typedef enum
+ {
+ FRAMEFLAGS_KEY = 1,
+ FRAMEFLAGS_GOLDEN = 2,
+ FRAMEFLAGS_ALTREF = 4
+ } FRAMETYPE_FLAGS;
+
+
+#include <assert.h>
+ static void Scale2Ratio(int mode, int *hr, int *hs)
+ {
+ switch (mode)
+ {
+ case NORMAL:
+ *hr = 1;
+ *hs = 1;
+ break;
+ case FOURFIVE:
+ *hr = 4;
+ *hs = 5;
+ break;
+ case THREEFIVE:
+ *hr = 3;
+ *hs = 5;
+ break;
+ case ONETWO:
+ *hr = 1;
+ *hs = 2;
+ break;
+ default:
+ *hr = 1;
+ *hs = 1;
+ assert(0);
+ break;
+ }
+ }
+
+ typedef struct
+ {
+ /* 4 versions of bitstream defined:
+ * 0 best quality/slowest decode, 3 lowest quality/fastest decode
+ */
+ int Version;
+ int Width;
+ int Height;
+ struct vpx_rational timebase;
+ unsigned int target_bandwidth; /* kilobits per second */
+
+ /* parameter used for applying pre processing blur: recommendation 0 */
+ int noise_sensitivity;
+
+ /* parameter used for sharpening output: recommendation 0: */
+ int Sharpness;
+ int cpu_used;
+ unsigned int rc_max_intra_bitrate_pct;
+
+ /* mode ->
+ *(0)=Realtime/Live Encoding. This mode is optimized for realtim
+ * encoding (for example, capturing a television signal or feed
+ * from a live camera). ( speed setting controls how fast )
+ *(1)=Good Quality Fast Encoding. The encoder balances quality with
+ * the amount of time it takes to encode the output. ( speed
+ * setting controls how fast )
+ *(2)=One Pass - Best Quality. The encoder places priority on the
+ * quality of the output over encoding speed. The output is
+ * compressed at the highest possible quality. This option takes
+ * the longest amount of time to encode. ( speed setting ignored
+ * )
+ *(3)=Two Pass - First Pass. The encoder generates a file of
+ * statistics for use in the second encoding pass. ( speed
+ * setting controls how fast )
+ *(4)=Two Pass - Second Pass. The encoder uses the statistics that
+ * were generated in the first encoding pass to create the
+ * compressed output. ( speed setting controls how fast )
+ *(5)=Two Pass - Second Pass Best. The encoder uses the statistics
+ * that were generated in the first encoding pass to create the
+ * compressed output using the highest possible quality, and
+ * taking a longer amount of time to encode.. ( speed setting
+ * ignored )
+ */
+ int Mode;
+
+ /* Key Framing Operations */
+ int auto_key; /* automatically detect cut scenes */
+ int key_freq; /* maximum distance to key frame. */
+
+ /* lagged compression (if allow_lag == 0 lag_in_frames is ignored) */
+ int allow_lag;
+ int lag_in_frames; /* how many frames lag before we start encoding */
+
+ /*
+ * DATARATE CONTROL OPTIONS
+ */
+
+ int end_usage; /* vbr or cbr */
+
+ /* buffer targeting aggressiveness */
+ int under_shoot_pct;
+ int over_shoot_pct;
+
+ /* buffering parameters */
+ int64_t starting_buffer_level;
+ int64_t optimal_buffer_level;
+ int64_t maximum_buffer_size;
+
+ int64_t starting_buffer_level_in_ms;
+ int64_t optimal_buffer_level_in_ms;
+ int64_t maximum_buffer_size_in_ms;
+
+ /* controlling quality */
+ int fixed_q;
+ int worst_allowed_q;
+ int best_allowed_q;
+ int cq_level;
+
+ /* allow internal resizing */
+ int allow_spatial_resampling;
+ int resample_down_water_mark;
+ int resample_up_water_mark;
+
+ /* allow internal frame rate alterations */
+ int allow_df;
+ int drop_frames_water_mark;
+
+ /* two pass datarate control */
+ int two_pass_vbrbias;
+ int two_pass_vbrmin_section;
+ int two_pass_vbrmax_section;
+
+ /*
+ * END DATARATE CONTROL OPTIONS
+ */
+
+ /* these parameters aren't to be used in final build don't use!!! */
+ int play_alternate;
+ int alt_freq;
+ int alt_q;
+ int key_q;
+ int gold_q;
+
+
+ int multi_threaded; /* how many threads to run the encoder on */
+ int token_partitions; /* how many token partitions to create */
+
+ /* early breakout threshold: for video conf recommend 800 */
+ int encode_breakout;
+
+ /* Bitfield defining the error resiliency features to enable.
+ * Can provide decodable frames after losses in previous
+ * frames and decodable partitions after losses in the same frame.
+ */
+ unsigned int error_resilient_mode;
+
+ int arnr_max_frames;
+ int arnr_strength;
+ int arnr_type;
+
+ struct vpx_fixed_buf two_pass_stats_in;
+ struct vpx_codec_pkt_list *output_pkt_list;
+
+ vp8e_tuning tuning;
+
+ /* Temporal scaling parameters */
+ unsigned int number_of_layers;
+ unsigned int target_bitrate[VPX_TS_MAX_PERIODICITY];
+ unsigned int rate_decimator[VPX_TS_MAX_PERIODICITY];
+ unsigned int periodicity;
+ unsigned int layer_id[VPX_TS_MAX_PERIODICITY];
+
+#if CONFIG_MULTI_RES_ENCODING
+ /* Number of total resolutions encoded */
+ unsigned int mr_total_resolutions;
+
+ /* Current encoder ID */
+ unsigned int mr_encoder_id;
+
+ /* Down-sampling factor */
+ vpx_rational_t mr_down_sampling_factor;
+
+ /* Memory location to store low-resolution encoder's mode info */
+ void* mr_low_res_mode_info;
+#endif
+ } VP8_CONFIG;
+
+
+ void vp8_initialize();
+
+ struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf);
+ void vp8_remove_compressor(struct VP8_COMP* *comp);
+
+ void vp8_init_config(struct VP8_COMP* onyx, VP8_CONFIG *oxcf);
+ void vp8_change_config(struct VP8_COMP* onyx, VP8_CONFIG *oxcf);
+
+ int vp8_receive_raw_frame(struct VP8_COMP* comp, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, int64_t time_stamp, int64_t end_time_stamp);
+ int vp8_get_compressed_data(struct VP8_COMP* comp, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, unsigned char *dest_end, int64_t *time_stamp, int64_t *time_end, int flush);
+ int vp8_get_preview_raw_frame(struct VP8_COMP* comp, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags);
+
+ int vp8_use_as_reference(struct VP8_COMP* comp, int ref_frame_flags);
+ int vp8_update_reference(struct VP8_COMP* comp, int ref_frame_flags);
+ int vp8_get_reference(struct VP8_COMP* comp, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd);
+ int vp8_set_reference(struct VP8_COMP* comp, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd);
+ int vp8_update_entropy(struct VP8_COMP* comp, int update);
+ int vp8_set_roimap(struct VP8_COMP* comp, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4]);
+ int vp8_set_active_map(struct VP8_COMP* comp, unsigned char *map, unsigned int rows, unsigned int cols);
+ int vp8_set_internal_size(struct VP8_COMP* comp, VPX_SCALING horiz_mode, VPX_SCALING vert_mode);
+ int vp8_get_quantizer(struct VP8_COMP* c);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/libvpx/vp8/common/onyxc_int.h b/libvpx/vp8/common/onyxc_int.h
new file mode 100644
index 0000000..5325bac
--- /dev/null
+++ b/libvpx/vp8/common/onyxc_int.h
@@ -0,0 +1,179 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_VP8C_INT_H
+#define __INC_VP8C_INT_H
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx/internal/vpx_codec_internal.h"
+#include "loopfilter.h"
+#include "entropymv.h"
+#include "entropy.h"
+#if CONFIG_POSTPROC
+#include "postproc.h"
+#endif
+
+/*#ifdef PACKET_TESTING*/
+#include "header.h"
+/*#endif*/
+
+#define MINQ 0
+#define MAXQ 127
+#define QINDEX_RANGE (MAXQ + 1)
+
+#define NUM_YV12_BUFFERS 4
+
+#define MAX_PARTITIONS 9
+
+typedef struct frame_contexts
+{
+ vp8_prob bmode_prob [VP8_BINTRAMODES-1];
+ vp8_prob ymode_prob [VP8_YMODES-1]; /* interframe intra mode probs */
+ vp8_prob uv_mode_prob [VP8_UV_MODES-1];
+ vp8_prob sub_mv_ref_prob [VP8_SUBMVREFS-1];
+ vp8_prob coef_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES];
+ MV_CONTEXT mvc[2];
+} FRAME_CONTEXT;
+
+typedef enum
+{
+ ONE_PARTITION = 0,
+ TWO_PARTITION = 1,
+ FOUR_PARTITION = 2,
+ EIGHT_PARTITION = 3
+} TOKEN_PARTITION;
+
+typedef enum
+{
+ RECON_CLAMP_REQUIRED = 0,
+ RECON_CLAMP_NOTREQUIRED = 1
+} CLAMP_TYPE;
+
+typedef struct VP8Common
+
+{
+ struct vpx_internal_error_info error;
+
+ DECLARE_ALIGNED(16, short, Y1dequant[QINDEX_RANGE][2]);
+ DECLARE_ALIGNED(16, short, Y2dequant[QINDEX_RANGE][2]);
+ DECLARE_ALIGNED(16, short, UVdequant[QINDEX_RANGE][2]);
+
+ int Width;
+ int Height;
+ int horiz_scale;
+ int vert_scale;
+
+ YUV_TYPE clr_type;
+ CLAMP_TYPE clamp_type;
+
+ YV12_BUFFER_CONFIG *frame_to_show;
+
+ YV12_BUFFER_CONFIG yv12_fb[NUM_YV12_BUFFERS];
+ int fb_idx_ref_cnt[NUM_YV12_BUFFERS];
+ int new_fb_idx, lst_fb_idx, gld_fb_idx, alt_fb_idx;
+
+ YV12_BUFFER_CONFIG temp_scale_frame;
+
+#if CONFIG_POSTPROC
+ YV12_BUFFER_CONFIG post_proc_buffer;
+ YV12_BUFFER_CONFIG post_proc_buffer_int;
+ int post_proc_buffer_int_used;
+ unsigned char *pp_limits_buffer; /* post-processing filter coefficients */
+#endif
+
+ FRAME_TYPE last_frame_type; /* Save last frame's frame type for motion search. */
+ FRAME_TYPE frame_type;
+
+ int show_frame;
+
+ int frame_flags;
+ int MBs;
+ int mb_rows;
+ int mb_cols;
+ int mode_info_stride;
+
+ /* profile settings */
+ int mb_no_coeff_skip;
+ int no_lpf;
+ int use_bilinear_mc_filter;
+ int full_pixel;
+
+ int base_qindex;
+
+ int y1dc_delta_q;
+ int y2dc_delta_q;
+ int y2ac_delta_q;
+ int uvdc_delta_q;
+ int uvac_delta_q;
+
+ unsigned int frames_since_golden;
+ unsigned int frames_till_alt_ref_frame;
+
+ /* We allocate a MODE_INFO struct for each macroblock, together with
+ an extra row on top and column on the left to simplify prediction. */
+
+ MODE_INFO *mip; /* Base of allocated array */
+ MODE_INFO *mi; /* Corresponds to upper left visible macroblock */
+#if CONFIG_ERROR_CONCEALMENT
+ MODE_INFO *prev_mip; /* MODE_INFO array 'mip' from last decoded frame */
+ MODE_INFO *prev_mi; /* 'mi' from last frame (points into prev_mip) */
+#endif
+
+ LOOPFILTERTYPE filter_type;
+
+ loop_filter_info_n lf_info;
+
+ int filter_level;
+ int last_sharpness_level;
+ int sharpness_level;
+
+ int refresh_last_frame; /* Two state 0 = NO, 1 = YES */
+ int refresh_golden_frame; /* Two state 0 = NO, 1 = YES */
+ int refresh_alt_ref_frame; /* Two state 0 = NO, 1 = YES */
+
+ int copy_buffer_to_gf; /* 0 none, 1 Last to GF, 2 ARF to GF */
+ int copy_buffer_to_arf; /* 0 none, 1 Last to ARF, 2 GF to ARF */
+
+ int refresh_entropy_probs; /* Two state 0 = NO, 1 = YES */
+
+ int ref_frame_sign_bias[MAX_REF_FRAMES]; /* Two state 0, 1 */
+
+ /* Y,U,V,Y2 */
+ ENTROPY_CONTEXT_PLANES *above_context; /* row of context for each plane */
+ ENTROPY_CONTEXT_PLANES left_context; /* (up to) 4 contexts "" */
+
+ FRAME_CONTEXT lfc; /* last frame entropy */
+ FRAME_CONTEXT fc; /* this frame entropy */
+
+ unsigned int current_video_frame;
+
+ int near_boffset[3];
+ int version;
+
+ TOKEN_PARTITION multi_token_partition;
+
+#ifdef PACKET_TESTING
+ VP8_HEADER oh;
+#endif
+ double bitrate;
+ double framerate;
+
+#if CONFIG_MULTITHREAD
+ int processor_core_count;
+#endif
+#if CONFIG_POSTPROC
+ struct postproc_state postproc_state;
+#endif
+ int cpu_caps;
+} VP8_COMMON;
+
+#endif
diff --git a/libvpx/vp8/common/onyxd.h b/libvpx/vp8/common/onyxd.h
new file mode 100644
index 0000000..fd7e051
--- /dev/null
+++ b/libvpx/vp8/common/onyxd.h
@@ -0,0 +1,68 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_VP8D_H
+#define __INC_VP8D_H
+
+
+/* Create/destroy static data structures. */
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+#include "vpx_scale/yv12config.h"
+#include "ppflags.h"
+#include "vpx_ports/mem.h"
+#include "vpx/vpx_codec.h"
+#include "vpx/vp8.h"
+
+ struct VP8D_COMP;
+
+ typedef struct
+ {
+ int Width;
+ int Height;
+ int Version;
+ int postprocess;
+ int max_threads;
+ int error_concealment;
+ int input_fragments;
+ } VP8D_CONFIG;
+
+ typedef enum
+ {
+ VP8D_OK = 0
+ } VP8D_SETTING;
+
+ void vp8dx_initialize(void);
+
+ void vp8dx_set_setting(struct VP8D_COMP* comp, VP8D_SETTING oxst, int x);
+
+ int vp8dx_get_setting(struct VP8D_COMP* comp, VP8D_SETTING oxst);
+
+ int vp8dx_receive_compressed_data(struct VP8D_COMP* comp,
+ size_t size, const uint8_t *dest,
+ int64_t time_stamp);
+ int vp8dx_get_raw_frame(struct VP8D_COMP* comp, YV12_BUFFER_CONFIG *sd, int64_t *time_stamp, int64_t *time_end_stamp, vp8_ppflags_t *flags);
+
+ vpx_codec_err_t vp8dx_get_reference(struct VP8D_COMP* comp, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd);
+ vpx_codec_err_t vp8dx_set_reference(struct VP8D_COMP* comp, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd);
+
+ struct VP8D_COMP* vp8dx_create_decompressor(VP8D_CONFIG *oxcf);
+
+ void vp8dx_remove_decompressor(struct VP8D_COMP* comp);
+
+#ifdef __cplusplus
+}
+#endif
+
+
+#endif
diff --git a/libvpx/vp8/common/postproc.c b/libvpx/vp8/common/postproc.c
new file mode 100644
index 0000000..80fa530
--- /dev/null
+++ b/libvpx/vp8/common/postproc.c
@@ -0,0 +1,1206 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx_scale/yv12config.h"
+#include "postproc.h"
+#include "common.h"
+#include "vpx_scale/vpxscale.h"
+#include "systemdependent.h"
+
+#include <limits.h>
+#include <math.h>
+#include <stdlib.h>
+#include <stdio.h>
+
+#define RGB_TO_YUV(t) \
+ ( (0.257*(float)(t>>16)) + (0.504*(float)(t>>8&0xff)) + (0.098*(float)(t&0xff)) + 16), \
+ (-(0.148*(float)(t>>16)) - (0.291*(float)(t>>8&0xff)) + (0.439*(float)(t&0xff)) + 128), \
+ ( (0.439*(float)(t>>16)) - (0.368*(float)(t>>8&0xff)) - (0.071*(float)(t&0xff)) + 128)
+
+/* global constants */
+#if CONFIG_POSTPROC_VISUALIZER
+static const unsigned char MB_PREDICTION_MODE_colors[MB_MODE_COUNT][3] =
+{
+ { RGB_TO_YUV(0x98FB98) }, /* PaleGreen */
+ { RGB_TO_YUV(0x00FF00) }, /* Green */
+ { RGB_TO_YUV(0xADFF2F) }, /* GreenYellow */
+ { RGB_TO_YUV(0x228B22) }, /* ForestGreen */
+ { RGB_TO_YUV(0x006400) }, /* DarkGreen */
+ { RGB_TO_YUV(0x98F5FF) }, /* Cadet Blue */
+ { RGB_TO_YUV(0x6CA6CD) }, /* Sky Blue */
+ { RGB_TO_YUV(0x00008B) }, /* Dark blue */
+ { RGB_TO_YUV(0x551A8B) }, /* Purple */
+ { RGB_TO_YUV(0xFF0000) } /* Red */
+};
+
+static const unsigned char B_PREDICTION_MODE_colors[B_MODE_COUNT][3] =
+{
+ { RGB_TO_YUV(0x6633ff) }, /* Purple */
+ { RGB_TO_YUV(0xcc33ff) }, /* Magenta */
+ { RGB_TO_YUV(0xff33cc) }, /* Pink */
+ { RGB_TO_YUV(0xff3366) }, /* Coral */
+ { RGB_TO_YUV(0x3366ff) }, /* Blue */
+ { RGB_TO_YUV(0xed00f5) }, /* Dark Blue */
+ { RGB_TO_YUV(0x2e00b8) }, /* Dark Purple */
+ { RGB_TO_YUV(0xff6633) }, /* Orange */
+ { RGB_TO_YUV(0x33ccff) }, /* Light Blue */
+ { RGB_TO_YUV(0x8ab800) }, /* Green */
+ { RGB_TO_YUV(0xffcc33) }, /* Light Orange */
+ { RGB_TO_YUV(0x33ffcc) }, /* Aqua */
+ { RGB_TO_YUV(0x66ff33) }, /* Light Green */
+ { RGB_TO_YUV(0xccff33) }, /* Yellow */
+};
+
+static const unsigned char MV_REFERENCE_FRAME_colors[MAX_REF_FRAMES][3] =
+{
+ { RGB_TO_YUV(0x00ff00) }, /* Blue */
+ { RGB_TO_YUV(0x0000ff) }, /* Green */
+ { RGB_TO_YUV(0xffff00) }, /* Yellow */
+ { RGB_TO_YUV(0xff0000) }, /* Red */
+};
+#endif
+
+static const short kernel5[] =
+{
+ 1, 1, 4, 1, 1
+};
+
+const short vp8_rv[] =
+{
+ 8, 5, 2, 2, 8, 12, 4, 9, 8, 3,
+ 0, 3, 9, 0, 0, 0, 8, 3, 14, 4,
+ 10, 1, 11, 14, 1, 14, 9, 6, 12, 11,
+ 8, 6, 10, 0, 0, 8, 9, 0, 3, 14,
+ 8, 11, 13, 4, 2, 9, 0, 3, 9, 6,
+ 1, 2, 3, 14, 13, 1, 8, 2, 9, 7,
+ 3, 3, 1, 13, 13, 6, 6, 5, 2, 7,
+ 11, 9, 11, 8, 7, 3, 2, 0, 13, 13,
+ 14, 4, 12, 5, 12, 10, 8, 10, 13, 10,
+ 4, 14, 4, 10, 0, 8, 11, 1, 13, 7,
+ 7, 14, 6, 14, 13, 2, 13, 5, 4, 4,
+ 0, 10, 0, 5, 13, 2, 12, 7, 11, 13,
+ 8, 0, 4, 10, 7, 2, 7, 2, 2, 5,
+ 3, 4, 7, 3, 3, 14, 14, 5, 9, 13,
+ 3, 14, 3, 6, 3, 0, 11, 8, 13, 1,
+ 13, 1, 12, 0, 10, 9, 7, 6, 2, 8,
+ 5, 2, 13, 7, 1, 13, 14, 7, 6, 7,
+ 9, 6, 10, 11, 7, 8, 7, 5, 14, 8,
+ 4, 4, 0, 8, 7, 10, 0, 8, 14, 11,
+ 3, 12, 5, 7, 14, 3, 14, 5, 2, 6,
+ 11, 12, 12, 8, 0, 11, 13, 1, 2, 0,
+ 5, 10, 14, 7, 8, 0, 4, 11, 0, 8,
+ 0, 3, 10, 5, 8, 0, 11, 6, 7, 8,
+ 10, 7, 13, 9, 2, 5, 1, 5, 10, 2,
+ 4, 3, 5, 6, 10, 8, 9, 4, 11, 14,
+ 0, 10, 0, 5, 13, 2, 12, 7, 11, 13,
+ 8, 0, 4, 10, 7, 2, 7, 2, 2, 5,
+ 3, 4, 7, 3, 3, 14, 14, 5, 9, 13,
+ 3, 14, 3, 6, 3, 0, 11, 8, 13, 1,
+ 13, 1, 12, 0, 10, 9, 7, 6, 2, 8,
+ 5, 2, 13, 7, 1, 13, 14, 7, 6, 7,
+ 9, 6, 10, 11, 7, 8, 7, 5, 14, 8,
+ 4, 4, 0, 8, 7, 10, 0, 8, 14, 11,
+ 3, 12, 5, 7, 14, 3, 14, 5, 2, 6,
+ 11, 12, 12, 8, 0, 11, 13, 1, 2, 0,
+ 5, 10, 14, 7, 8, 0, 4, 11, 0, 8,
+ 0, 3, 10, 5, 8, 0, 11, 6, 7, 8,
+ 10, 7, 13, 9, 2, 5, 1, 5, 10, 2,
+ 4, 3, 5, 6, 10, 8, 9, 4, 11, 14,
+ 3, 8, 3, 7, 8, 5, 11, 4, 12, 3,
+ 11, 9, 14, 8, 14, 13, 4, 3, 1, 2,
+ 14, 6, 5, 4, 4, 11, 4, 6, 2, 1,
+ 5, 8, 8, 12, 13, 5, 14, 10, 12, 13,
+ 0, 9, 5, 5, 11, 10, 13, 9, 10, 13,
+};
+
+extern void vp8_blit_text(const char *msg, unsigned char *address, const int pitch);
+extern void vp8_blit_line(int x0, int x1, int y0, int y1, unsigned char *image, const int pitch);
+/***********************************************************************************************************
+ */
+void vp8_post_proc_down_and_across_mb_row_c
+(
+ unsigned char *src_ptr,
+ unsigned char *dst_ptr,
+ int src_pixels_per_line,
+ int dst_pixels_per_line,
+ int cols,
+ unsigned char *f,
+ int size
+)
+{
+ unsigned char *p_src, *p_dst;
+ int row;
+ int col;
+ unsigned char v;
+ unsigned char d[4];
+
+ for (row = 0; row < size; row++)
+ {
+ /* post_proc_down for one row */
+ p_src = src_ptr;
+ p_dst = dst_ptr;
+
+ for (col = 0; col < cols; col++)
+ {
+ unsigned char p_above2 = p_src[col - 2 * src_pixels_per_line];
+ unsigned char p_above1 = p_src[col - src_pixels_per_line];
+ unsigned char p_below1 = p_src[col + src_pixels_per_line];
+ unsigned char p_below2 = p_src[col + 2 * src_pixels_per_line];
+
+ v = p_src[col];
+
+ if ((abs(v - p_above2) < f[col]) && (abs(v - p_above1) < f[col])
+ && (abs(v - p_below1) < f[col]) && (abs(v - p_below2) < f[col]))
+ {
+ unsigned char k1, k2, k3;
+ k1 = (p_above2 + p_above1 + 1) >> 1;
+ k2 = (p_below2 + p_below1 + 1) >> 1;
+ k3 = (k1 + k2 + 1) >> 1;
+ v = (k3 + v + 1) >> 1;
+ }
+
+ p_dst[col] = v;
+ }
+
+ /* now post_proc_across */
+ p_src = dst_ptr;
+ p_dst = dst_ptr;
+
+ p_src[-2] = p_src[-1] = p_src[0];
+ p_src[cols] = p_src[cols + 1] = p_src[cols - 1];
+
+ for (col = 0; col < cols; col++)
+ {
+ v = p_src[col];
+
+ if ((abs(v - p_src[col - 2]) < f[col])
+ && (abs(v - p_src[col - 1]) < f[col])
+ && (abs(v - p_src[col + 1]) < f[col])
+ && (abs(v - p_src[col + 2]) < f[col]))
+ {
+ unsigned char k1, k2, k3;
+ k1 = (p_src[col - 2] + p_src[col - 1] + 1) >> 1;
+ k2 = (p_src[col + 2] + p_src[col + 1] + 1) >> 1;
+ k3 = (k1 + k2 + 1) >> 1;
+ v = (k3 + v + 1) >> 1;
+ }
+
+ d[col & 3] = v;
+
+ if (col >= 2)
+ p_dst[col - 2] = d[(col - 2) & 3];
+ }
+
+ /* handle the last two pixels */
+ p_dst[col - 2] = d[(col - 2) & 3];
+ p_dst[col - 1] = d[(col - 1) & 3];
+
+ /* next row */
+ src_ptr += src_pixels_per_line;
+ dst_ptr += dst_pixels_per_line;
+ }
+}
+
+static int q2mbl(int x)
+{
+ if (x < 20) x = 20;
+
+ x = 50 + (x - 50) * 10 / 8;
+ return x * x / 3;
+}
+void vp8_mbpost_proc_across_ip_c(unsigned char *src, int pitch, int rows, int cols, int flimit)
+{
+ int r, c, i;
+
+ unsigned char *s = src;
+ unsigned char d[16];
+
+ for (r = 0; r < rows; r++)
+ {
+ int sumsq = 0;
+ int sum = 0;
+
+ for (i = -8; i<0; i++)
+ s[i]=s[0];
+
+ /* 17 avoids valgrind warning - we buffer values in c in d
+ * and only write them when we've read 8 ahead...
+ */
+ for (i = cols; i<cols+17; i++)
+ s[i]=s[cols-1];
+
+ for (i = -8; i <= 6; i++)
+ {
+ sumsq += s[i] * s[i];
+ sum += s[i];
+ d[i+8] = 0;
+ }
+
+ for (c = 0; c < cols + 8; c++)
+ {
+ int x = s[c+7] - s[c-8];
+ int y = s[c+7] + s[c-8];
+
+ sum += x;
+ sumsq += x * y;
+
+ d[c&15] = s[c];
+
+ if (sumsq * 15 - sum * sum < flimit)
+ {
+ d[c&15] = (8 + sum + s[c]) >> 4;
+ }
+
+ s[c-8] = d[(c-8)&15];
+ }
+
+ s += pitch;
+ }
+}
+
+
+void vp8_mbpost_proc_down_c(unsigned char *dst, int pitch, int rows, int cols, int flimit)
+{
+ int r, c, i;
+ const short *rv3 = &vp8_rv[63&rand()];
+
+ for (c = 0; c < cols; c++ )
+ {
+ unsigned char *s = &dst[c];
+ int sumsq = 0;
+ int sum = 0;
+ unsigned char d[16];
+ const short *rv2 = rv3 + ((c * 17) & 127);
+
+ for (i = -8; i < 0; i++)
+ s[i*pitch]=s[0];
+
+ /* 17 avoids valgrind warning - we buffer values in c in d
+ * and only write them when we've read 8 ahead...
+ */
+ for (i = rows; i < rows+17; i++)
+ s[i*pitch]=s[(rows-1)*pitch];
+
+ for (i = -8; i <= 6; i++)
+ {
+ sumsq += s[i*pitch] * s[i*pitch];
+ sum += s[i*pitch];
+ }
+
+ for (r = 0; r < rows + 8; r++)
+ {
+ sumsq += s[7*pitch] * s[ 7*pitch] - s[-8*pitch] * s[-8*pitch];
+ sum += s[7*pitch] - s[-8*pitch];
+ d[r&15] = s[0];
+
+ if (sumsq * 15 - sum * sum < flimit)
+ {
+ d[r&15] = (rv2[r&127] + sum + s[0]) >> 4;
+ }
+
+ s[-8*pitch] = d[(r-8)&15];
+ s += pitch;
+ }
+ }
+}
+
+static void vp8_de_mblock(YV12_BUFFER_CONFIG *post,
+ int q)
+{
+ vp8_mbpost_proc_across_ip(post->y_buffer, post->y_stride, post->y_height,
+ post->y_width, q2mbl(q));
+ vp8_mbpost_proc_down(post->y_buffer, post->y_stride, post->y_height,
+ post->y_width, q2mbl(q));
+}
+
+void vp8_deblock(VP8_COMMON *cm,
+ YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *post,
+ int q,
+ int low_var_thresh,
+ int flag)
+{
+ double level = 6.0e-05 * q * q * q - .0067 * q * q + .306 * q + .0065;
+ int ppl = (int)(level + .5);
+
+ const MODE_INFO *mode_info_context = cm->mi;
+ int mbr, mbc;
+
+ /* The pixel thresholds are adjusted according to if or not the macroblock
+ * is a skipped block. */
+ unsigned char *ylimits = cm->pp_limits_buffer;
+ unsigned char *uvlimits = cm->pp_limits_buffer + 16 * cm->mb_cols;
+ (void) low_var_thresh;
+ (void) flag;
+
+ if (ppl > 0)
+ {
+ for (mbr = 0; mbr < cm->mb_rows; mbr++)
+ {
+ unsigned char *ylptr = ylimits;
+ unsigned char *uvlptr = uvlimits;
+ for (mbc = 0; mbc < cm->mb_cols; mbc++)
+ {
+ unsigned char mb_ppl;
+
+ if (mode_info_context->mbmi.mb_skip_coeff)
+ mb_ppl = (unsigned char)ppl >> 1;
+ else
+ mb_ppl = (unsigned char)ppl;
+
+ vpx_memset(ylptr, mb_ppl, 16);
+ vpx_memset(uvlptr, mb_ppl, 8);
+
+ ylptr += 16;
+ uvlptr += 8;
+ mode_info_context++;
+ }
+ mode_info_context++;
+
+ vp8_post_proc_down_and_across_mb_row(
+ source->y_buffer + 16 * mbr * source->y_stride,
+ post->y_buffer + 16 * mbr * post->y_stride, source->y_stride,
+ post->y_stride, source->y_width, ylimits, 16);
+
+ vp8_post_proc_down_and_across_mb_row(
+ source->u_buffer + 8 * mbr * source->uv_stride,
+ post->u_buffer + 8 * mbr * post->uv_stride, source->uv_stride,
+ post->uv_stride, source->uv_width, uvlimits, 8);
+ vp8_post_proc_down_and_across_mb_row(
+ source->v_buffer + 8 * mbr * source->uv_stride,
+ post->v_buffer + 8 * mbr * post->uv_stride, source->uv_stride,
+ post->uv_stride, source->uv_width, uvlimits, 8);
+ }
+ } else
+ {
+ vp8_yv12_copy_frame(source, post);
+ }
+}
+
+#if !(CONFIG_TEMPORAL_DENOISING)
+void vp8_de_noise(VP8_COMMON *cm,
+ YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *post,
+ int q,
+ int low_var_thresh,
+ int flag)
+{
+ double level = 6.0e-05 * q * q * q - .0067 * q * q + .306 * q + .0065;
+ int ppl = (int)(level + .5);
+ int mb_rows = source->y_width >> 4;
+ int mb_cols = source->y_height >> 4;
+ unsigned char *limits = cm->pp_limits_buffer;;
+ int mbr, mbc;
+ (void) post;
+ (void) low_var_thresh;
+ (void) flag;
+
+ vpx_memset(limits, (unsigned char)ppl, 16 * mb_cols);
+
+ /* TODO: The original code don't filter the 2 outer rows and columns. */
+ for (mbr = 0; mbr < mb_rows; mbr++)
+ {
+ vp8_post_proc_down_and_across_mb_row(
+ source->y_buffer + 16 * mbr * source->y_stride,
+ source->y_buffer + 16 * mbr * source->y_stride,
+ source->y_stride, source->y_stride, source->y_width, limits, 16);
+
+ vp8_post_proc_down_and_across_mb_row(
+ source->u_buffer + 8 * mbr * source->uv_stride,
+ source->u_buffer + 8 * mbr * source->uv_stride,
+ source->uv_stride, source->uv_stride, source->uv_width, limits, 8);
+ vp8_post_proc_down_and_across_mb_row(
+ source->v_buffer + 8 * mbr * source->uv_stride,
+ source->v_buffer + 8 * mbr * source->uv_stride,
+ source->uv_stride, source->uv_stride, source->uv_width, limits, 8);
+ }
+}
+#endif
+
+double vp8_gaussian(double sigma, double mu, double x)
+{
+ return 1 / (sigma * sqrt(2.0 * 3.14159265)) *
+ (exp(-(x - mu) * (x - mu) / (2 * sigma * sigma)));
+}
+
+static void fillrd(struct postproc_state *state, int q, int a)
+{
+ char char_dist[300];
+
+ double sigma;
+ int ai = a, qi = q, i;
+
+ vp8_clear_system_state();
+
+
+ sigma = ai + .5 + .6 * (63 - qi) / 63.0;
+
+ /* set up a lookup table of 256 entries that matches
+ * a gaussian distribution with sigma determined by q.
+ */
+ {
+ double i;
+ int next, j;
+
+ next = 0;
+
+ for (i = -32; i < 32; i++)
+ {
+ int a = (int)(.5 + 256 * vp8_gaussian(sigma, 0, i));
+
+ if (a)
+ {
+ for (j = 0; j < a; j++)
+ {
+ char_dist[next+j] = (char) i;
+ }
+
+ next = next + j;
+ }
+
+ }
+
+ for (; next < 256; next++)
+ char_dist[next] = 0;
+
+ }
+
+ for (i = 0; i < 3072; i++)
+ {
+ state->noise[i] = char_dist[rand() & 0xff];
+ }
+
+ for (i = 0; i < 16; i++)
+ {
+ state->blackclamp[i] = -char_dist[0];
+ state->whiteclamp[i] = -char_dist[0];
+ state->bothclamp[i] = -2 * char_dist[0];
+ }
+
+ state->last_q = q;
+ state->last_noise = a;
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : plane_add_noise_c
+ *
+ * INPUTS : unsigned char *Start starting address of buffer to add gaussian
+ * noise to
+ * unsigned int Width width of plane
+ * unsigned int Height height of plane
+ * int Pitch distance between subsequent lines of frame
+ * int q quantizer used to determine amount of noise
+ * to add
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void.
+ *
+ * FUNCTION : adds gaussian noise to a plane of pixels
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+void vp8_plane_add_noise_c(unsigned char *Start, char *noise,
+ char blackclamp[16],
+ char whiteclamp[16],
+ char bothclamp[16],
+ unsigned int Width, unsigned int Height, int Pitch)
+{
+ unsigned int i, j;
+
+ for (i = 0; i < Height; i++)
+ {
+ unsigned char *Pos = Start + i * Pitch;
+ char *Ref = (char *)(noise + (rand() & 0xff));
+
+ for (j = 0; j < Width; j++)
+ {
+ if (Pos[j] < blackclamp[0])
+ Pos[j] = blackclamp[0];
+
+ if (Pos[j] > 255 + whiteclamp[0])
+ Pos[j] = 255 + whiteclamp[0];
+
+ Pos[j] += Ref[j];
+ }
+ }
+}
+
+/* Blend the macro block with a solid colored square. Leave the
+ * edges unblended to give distinction to macro blocks in areas
+ * filled with the same color block.
+ */
+void vp8_blend_mb_inner_c (unsigned char *y, unsigned char *u, unsigned char *v,
+ int y1, int u1, int v1, int alpha, int stride)
+{
+ int i, j;
+ int y1_const = y1*((1<<16)-alpha);
+ int u1_const = u1*((1<<16)-alpha);
+ int v1_const = v1*((1<<16)-alpha);
+
+ y += 2*stride + 2;
+ for (i = 0; i < 12; i++)
+ {
+ for (j = 0; j < 12; j++)
+ {
+ y[j] = (y[j]*alpha + y1_const)>>16;
+ }
+ y += stride;
+ }
+
+ stride >>= 1;
+
+ u += stride + 1;
+ v += stride + 1;
+
+ for (i = 0; i < 6; i++)
+ {
+ for (j = 0; j < 6; j++)
+ {
+ u[j] = (u[j]*alpha + u1_const)>>16;
+ v[j] = (v[j]*alpha + v1_const)>>16;
+ }
+ u += stride;
+ v += stride;
+ }
+}
+
+/* Blend only the edge of the macro block. Leave center
+ * unblended to allow for other visualizations to be layered.
+ */
+void vp8_blend_mb_outer_c (unsigned char *y, unsigned char *u, unsigned char *v,
+ int y1, int u1, int v1, int alpha, int stride)
+{
+ int i, j;
+ int y1_const = y1*((1<<16)-alpha);
+ int u1_const = u1*((1<<16)-alpha);
+ int v1_const = v1*((1<<16)-alpha);
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 16; j++)
+ {
+ y[j] = (y[j]*alpha + y1_const)>>16;
+ }
+ y += stride;
+ }
+
+ for (i = 0; i < 12; i++)
+ {
+ y[0] = (y[0]*alpha + y1_const)>>16;
+ y[1] = (y[1]*alpha + y1_const)>>16;
+ y[14] = (y[14]*alpha + y1_const)>>16;
+ y[15] = (y[15]*alpha + y1_const)>>16;
+ y += stride;
+ }
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 16; j++)
+ {
+ y[j] = (y[j]*alpha + y1_const)>>16;
+ }
+ y += stride;
+ }
+
+ stride >>= 1;
+
+ for (j = 0; j < 8; j++)
+ {
+ u[j] = (u[j]*alpha + u1_const)>>16;
+ v[j] = (v[j]*alpha + v1_const)>>16;
+ }
+ u += stride;
+ v += stride;
+
+ for (i = 0; i < 6; i++)
+ {
+ u[0] = (u[0]*alpha + u1_const)>>16;
+ v[0] = (v[0]*alpha + v1_const)>>16;
+
+ u[7] = (u[7]*alpha + u1_const)>>16;
+ v[7] = (v[7]*alpha + v1_const)>>16;
+
+ u += stride;
+ v += stride;
+ }
+
+ for (j = 0; j < 8; j++)
+ {
+ u[j] = (u[j]*alpha + u1_const)>>16;
+ v[j] = (v[j]*alpha + v1_const)>>16;
+ }
+}
+
+void vp8_blend_b_c (unsigned char *y, unsigned char *u, unsigned char *v,
+ int y1, int u1, int v1, int alpha, int stride)
+{
+ int i, j;
+ int y1_const = y1*((1<<16)-alpha);
+ int u1_const = u1*((1<<16)-alpha);
+ int v1_const = v1*((1<<16)-alpha);
+
+ for (i = 0; i < 4; i++)
+ {
+ for (j = 0; j < 4; j++)
+ {
+ y[j] = (y[j]*alpha + y1_const)>>16;
+ }
+ y += stride;
+ }
+
+ stride >>= 1;
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 2; j++)
+ {
+ u[j] = (u[j]*alpha + u1_const)>>16;
+ v[j] = (v[j]*alpha + v1_const)>>16;
+ }
+ u += stride;
+ v += stride;
+ }
+}
+
+static void constrain_line (int x0, int *x1, int y0, int *y1, int width, int height)
+{
+ int dx;
+ int dy;
+
+ if (*x1 > width)
+ {
+ dx = *x1 - x0;
+ dy = *y1 - y0;
+
+ *x1 = width;
+ if (dx)
+ *y1 = ((width-x0)*dy)/dx + y0;
+ }
+ if (*x1 < 0)
+ {
+ dx = *x1 - x0;
+ dy = *y1 - y0;
+
+ *x1 = 0;
+ if (dx)
+ *y1 = ((0-x0)*dy)/dx + y0;
+ }
+ if (*y1 > height)
+ {
+ dx = *x1 - x0;
+ dy = *y1 - y0;
+
+ *y1 = height;
+ if (dy)
+ *x1 = ((height-y0)*dx)/dy + x0;
+ }
+ if (*y1 < 0)
+ {
+ dx = *x1 - x0;
+ dy = *y1 - y0;
+
+ *y1 = 0;
+ if (dy)
+ *x1 = ((0-y0)*dx)/dy + x0;
+ }
+}
+
+#if CONFIG_POSTPROC
+int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *ppflags)
+{
+ int q = oci->filter_level * 10 / 6;
+ int flags = ppflags->post_proc_flag;
+ int deblock_level = ppflags->deblocking_level;
+ int noise_level = ppflags->noise_level;
+
+ if (!oci->frame_to_show)
+ return -1;
+
+ if (q > 63)
+ q = 63;
+
+ if (!flags)
+ {
+ *dest = *oci->frame_to_show;
+
+ /* handle problem with extending borders */
+ dest->y_width = oci->Width;
+ dest->y_height = oci->Height;
+ dest->uv_height = dest->y_height / 2;
+ oci->postproc_state.last_base_qindex = oci->base_qindex;
+ oci->postproc_state.last_frame_valid = 1;
+ return 0;
+ }
+
+ /* Allocate post_proc_buffer_int if needed */
+ if ((flags & VP8D_MFQE) && !oci->post_proc_buffer_int_used)
+ {
+ if ((flags & VP8D_DEBLOCK) || (flags & VP8D_DEMACROBLOCK))
+ {
+ int width = (oci->Width + 15) & ~15;
+ int height = (oci->Height + 15) & ~15;
+
+ if (vp8_yv12_alloc_frame_buffer(&oci->post_proc_buffer_int,
+ width, height, VP8BORDERINPIXELS))
+ vpx_internal_error(&oci->error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate MFQE framebuffer");
+
+ oci->post_proc_buffer_int_used = 1;
+
+ /* insure that postproc is set to all 0's so that post proc
+ * doesn't pull random data in from edge
+ */
+ vpx_memset((&oci->post_proc_buffer_int)->buffer_alloc,128,(&oci->post_proc_buffer)->frame_size);
+
+ }
+ }
+
+ vp8_clear_system_state();
+
+ if ((flags & VP8D_MFQE) &&
+ oci->postproc_state.last_frame_valid &&
+ oci->current_video_frame >= 2 &&
+ oci->postproc_state.last_base_qindex < 60 &&
+ oci->base_qindex - oci->postproc_state.last_base_qindex >= 20)
+ {
+ vp8_multiframe_quality_enhance(oci);
+ if (((flags & VP8D_DEBLOCK) || (flags & VP8D_DEMACROBLOCK)) &&
+ oci->post_proc_buffer_int_used)
+ {
+ vp8_yv12_copy_frame(&oci->post_proc_buffer, &oci->post_proc_buffer_int);
+ if (flags & VP8D_DEMACROBLOCK)
+ {
+ vp8_deblock(oci, &oci->post_proc_buffer_int, &oci->post_proc_buffer,
+ q + (deblock_level - 5) * 10, 1, 0);
+ vp8_de_mblock(&oci->post_proc_buffer,
+ q + (deblock_level - 5) * 10);
+ }
+ else if (flags & VP8D_DEBLOCK)
+ {
+ vp8_deblock(oci, &oci->post_proc_buffer_int, &oci->post_proc_buffer,
+ q, 1, 0);
+ }
+ }
+ /* Move partially towards the base q of the previous frame */
+ oci->postproc_state.last_base_qindex = (3*oci->postproc_state.last_base_qindex + oci->base_qindex)>>2;
+ }
+ else if (flags & VP8D_DEMACROBLOCK)
+ {
+ vp8_deblock(oci, oci->frame_to_show, &oci->post_proc_buffer,
+ q + (deblock_level - 5) * 10, 1, 0);
+ vp8_de_mblock(&oci->post_proc_buffer, q + (deblock_level - 5) * 10);
+
+ oci->postproc_state.last_base_qindex = oci->base_qindex;
+ }
+ else if (flags & VP8D_DEBLOCK)
+ {
+ vp8_deblock(oci, oci->frame_to_show, &oci->post_proc_buffer,
+ q, 1, 0);
+ oci->postproc_state.last_base_qindex = oci->base_qindex;
+ }
+ else
+ {
+ vp8_yv12_copy_frame(oci->frame_to_show, &oci->post_proc_buffer);
+ oci->postproc_state.last_base_qindex = oci->base_qindex;
+ }
+ oci->postproc_state.last_frame_valid = 1;
+
+ if (flags & VP8D_ADDNOISE)
+ {
+ if (oci->postproc_state.last_q != q
+ || oci->postproc_state.last_noise != noise_level)
+ {
+ fillrd(&oci->postproc_state, 63 - q, noise_level);
+ }
+
+ vp8_plane_add_noise
+ (oci->post_proc_buffer.y_buffer,
+ oci->postproc_state.noise,
+ oci->postproc_state.blackclamp,
+ oci->postproc_state.whiteclamp,
+ oci->postproc_state.bothclamp,
+ oci->post_proc_buffer.y_width, oci->post_proc_buffer.y_height,
+ oci->post_proc_buffer.y_stride);
+ }
+
+#if CONFIG_POSTPROC_VISUALIZER
+ if (flags & VP8D_DEBUG_TXT_FRAME_INFO)
+ {
+ char message[512];
+ sprintf(message, "F%1dG%1dQ%3dF%3dP%d_s%dx%d",
+ (oci->frame_type == KEY_FRAME),
+ oci->refresh_golden_frame,
+ oci->base_qindex,
+ oci->filter_level,
+ flags,
+ oci->mb_cols, oci->mb_rows);
+ vp8_blit_text(message, oci->post_proc_buffer.y_buffer, oci->post_proc_buffer.y_stride);
+ }
+
+ if (flags & VP8D_DEBUG_TXT_MBLK_MODES)
+ {
+ int i, j;
+ unsigned char *y_ptr;
+ YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
+ int mb_rows = post->y_height >> 4;
+ int mb_cols = post->y_width >> 4;
+ int mb_index = 0;
+ MODE_INFO *mi = oci->mi;
+
+ y_ptr = post->y_buffer + 4 * post->y_stride + 4;
+
+ /* vp8_filter each macro block */
+ for (i = 0; i < mb_rows; i++)
+ {
+ for (j = 0; j < mb_cols; j++)
+ {
+ char zz[4];
+
+ sprintf(zz, "%c", mi[mb_index].mbmi.mode + 'a');
+
+ vp8_blit_text(zz, y_ptr, post->y_stride);
+ mb_index ++;
+ y_ptr += 16;
+ }
+
+ mb_index ++; /* border */
+ y_ptr += post->y_stride * 16 - post->y_width;
+
+ }
+ }
+
+ if (flags & VP8D_DEBUG_TXT_DC_DIFF)
+ {
+ int i, j;
+ unsigned char *y_ptr;
+ YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
+ int mb_rows = post->y_height >> 4;
+ int mb_cols = post->y_width >> 4;
+ int mb_index = 0;
+ MODE_INFO *mi = oci->mi;
+
+ y_ptr = post->y_buffer + 4 * post->y_stride + 4;
+
+ /* vp8_filter each macro block */
+ for (i = 0; i < mb_rows; i++)
+ {
+ for (j = 0; j < mb_cols; j++)
+ {
+ char zz[4];
+ int dc_diff = !(mi[mb_index].mbmi.mode != B_PRED &&
+ mi[mb_index].mbmi.mode != SPLITMV &&
+ mi[mb_index].mbmi.mb_skip_coeff);
+
+ if (oci->frame_type == KEY_FRAME)
+ sprintf(zz, "a");
+ else
+ sprintf(zz, "%c", dc_diff + '0');
+
+ vp8_blit_text(zz, y_ptr, post->y_stride);
+ mb_index ++;
+ y_ptr += 16;
+ }
+
+ mb_index ++; /* border */
+ y_ptr += post->y_stride * 16 - post->y_width;
+
+ }
+ }
+
+ if (flags & VP8D_DEBUG_TXT_RATE_INFO)
+ {
+ char message[512];
+ sprintf(message, "Bitrate: %10.2f frame_rate: %10.2f ", oci->bitrate, oci->framerate);
+ vp8_blit_text(message, oci->post_proc_buffer.y_buffer, oci->post_proc_buffer.y_stride);
+ }
+
+ /* Draw motion vectors */
+ if ((flags & VP8D_DEBUG_DRAW_MV) && ppflags->display_mv_flag)
+ {
+ YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
+ int width = post->y_width;
+ int height = post->y_height;
+ unsigned char *y_buffer = oci->post_proc_buffer.y_buffer;
+ int y_stride = oci->post_proc_buffer.y_stride;
+ MODE_INFO *mi = oci->mi;
+ int x0, y0;
+
+ for (y0 = 0; y0 < height; y0 += 16)
+ {
+ for (x0 = 0; x0 < width; x0 += 16)
+ {
+ int x1, y1;
+
+ if (!(ppflags->display_mv_flag & (1<<mi->mbmi.mode)))
+ {
+ mi++;
+ continue;
+ }
+
+ if (mi->mbmi.mode == SPLITMV)
+ {
+ switch (mi->mbmi.partitioning)
+ {
+ case 0 : /* mv_top_bottom */
+ {
+ union b_mode_info *bmi = &mi->bmi[0];
+ MV *mv = &bmi->mv.as_mv;
+
+ x1 = x0 + 8 + (mv->col >> 3);
+ y1 = y0 + 4 + (mv->row >> 3);
+
+ constrain_line (x0+8, &x1, y0+4, &y1, width, height);
+ vp8_blit_line (x0+8, x1, y0+4, y1, y_buffer, y_stride);
+
+ bmi = &mi->bmi[8];
+
+ x1 = x0 + 8 + (mv->col >> 3);
+ y1 = y0 +12 + (mv->row >> 3);
+
+ constrain_line (x0+8, &x1, y0+12, &y1, width, height);
+ vp8_blit_line (x0+8, x1, y0+12, y1, y_buffer, y_stride);
+
+ break;
+ }
+ case 1 : /* mv_left_right */
+ {
+ union b_mode_info *bmi = &mi->bmi[0];
+ MV *mv = &bmi->mv.as_mv;
+
+ x1 = x0 + 4 + (mv->col >> 3);
+ y1 = y0 + 8 + (mv->row >> 3);
+
+ constrain_line (x0+4, &x1, y0+8, &y1, width, height);
+ vp8_blit_line (x0+4, x1, y0+8, y1, y_buffer, y_stride);
+
+ bmi = &mi->bmi[2];
+
+ x1 = x0 +12 + (mv->col >> 3);
+ y1 = y0 + 8 + (mv->row >> 3);
+
+ constrain_line (x0+12, &x1, y0+8, &y1, width, height);
+ vp8_blit_line (x0+12, x1, y0+8, y1, y_buffer, y_stride);
+
+ break;
+ }
+ case 2 : /* mv_quarters */
+ {
+ union b_mode_info *bmi = &mi->bmi[0];
+ MV *mv = &bmi->mv.as_mv;
+
+ x1 = x0 + 4 + (mv->col >> 3);
+ y1 = y0 + 4 + (mv->row >> 3);
+
+ constrain_line (x0+4, &x1, y0+4, &y1, width, height);
+ vp8_blit_line (x0+4, x1, y0+4, y1, y_buffer, y_stride);
+
+ bmi = &mi->bmi[2];
+
+ x1 = x0 +12 + (mv->col >> 3);
+ y1 = y0 + 4 + (mv->row >> 3);
+
+ constrain_line (x0+12, &x1, y0+4, &y1, width, height);
+ vp8_blit_line (x0+12, x1, y0+4, y1, y_buffer, y_stride);
+
+ bmi = &mi->bmi[8];
+
+ x1 = x0 + 4 + (mv->col >> 3);
+ y1 = y0 +12 + (mv->row >> 3);
+
+ constrain_line (x0+4, &x1, y0+12, &y1, width, height);
+ vp8_blit_line (x0+4, x1, y0+12, y1, y_buffer, y_stride);
+
+ bmi = &mi->bmi[10];
+
+ x1 = x0 +12 + (mv->col >> 3);
+ y1 = y0 +12 + (mv->row >> 3);
+
+ constrain_line (x0+12, &x1, y0+12, &y1, width, height);
+ vp8_blit_line (x0+12, x1, y0+12, y1, y_buffer, y_stride);
+ break;
+ }
+ default :
+ {
+ union b_mode_info *bmi = mi->bmi;
+ int bx0, by0;
+
+ for (by0 = y0; by0 < (y0+16); by0 += 4)
+ {
+ for (bx0 = x0; bx0 < (x0+16); bx0 += 4)
+ {
+ MV *mv = &bmi->mv.as_mv;
+
+ x1 = bx0 + 2 + (mv->col >> 3);
+ y1 = by0 + 2 + (mv->row >> 3);
+
+ constrain_line (bx0+2, &x1, by0+2, &y1, width, height);
+ vp8_blit_line (bx0+2, x1, by0+2, y1, y_buffer, y_stride);
+
+ bmi++;
+ }
+ }
+ }
+ }
+ }
+ else if (mi->mbmi.mode >= NEARESTMV)
+ {
+ MV *mv = &mi->mbmi.mv.as_mv;
+ const int lx0 = x0 + 8;
+ const int ly0 = y0 + 8;
+
+ x1 = lx0 + (mv->col >> 3);
+ y1 = ly0 + (mv->row >> 3);
+
+ if (x1 != lx0 && y1 != ly0)
+ {
+ constrain_line (lx0, &x1, ly0-1, &y1, width, height);
+ vp8_blit_line (lx0, x1, ly0-1, y1, y_buffer, y_stride);
+
+ constrain_line (lx0, &x1, ly0+1, &y1, width, height);
+ vp8_blit_line (lx0, x1, ly0+1, y1, y_buffer, y_stride);
+ }
+ else
+ vp8_blit_line (lx0, x1, ly0, y1, y_buffer, y_stride);
+ }
+
+ mi++;
+ }
+ mi++;
+ }
+ }
+
+ /* Color in block modes */
+ if ((flags & VP8D_DEBUG_CLR_BLK_MODES)
+ && (ppflags->display_mb_modes_flag || ppflags->display_b_modes_flag))
+ {
+ int y, x;
+ YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
+ int width = post->y_width;
+ int height = post->y_height;
+ unsigned char *y_ptr = oci->post_proc_buffer.y_buffer;
+ unsigned char *u_ptr = oci->post_proc_buffer.u_buffer;
+ unsigned char *v_ptr = oci->post_proc_buffer.v_buffer;
+ int y_stride = oci->post_proc_buffer.y_stride;
+ MODE_INFO *mi = oci->mi;
+
+ for (y = 0; y < height; y += 16)
+ {
+ for (x = 0; x < width; x += 16)
+ {
+ int Y = 0, U = 0, V = 0;
+
+ if (mi->mbmi.mode == B_PRED &&
+ ((ppflags->display_mb_modes_flag & B_PRED) || ppflags->display_b_modes_flag))
+ {
+ int by, bx;
+ unsigned char *yl, *ul, *vl;
+ union b_mode_info *bmi = mi->bmi;
+
+ yl = y_ptr + x;
+ ul = u_ptr + (x>>1);
+ vl = v_ptr + (x>>1);
+
+ for (by = 0; by < 16; by += 4)
+ {
+ for (bx = 0; bx < 16; bx += 4)
+ {
+ if ((ppflags->display_b_modes_flag & (1<<mi->mbmi.mode))
+ || (ppflags->display_mb_modes_flag & B_PRED))
+ {
+ Y = B_PREDICTION_MODE_colors[bmi->as_mode][0];
+ U = B_PREDICTION_MODE_colors[bmi->as_mode][1];
+ V = B_PREDICTION_MODE_colors[bmi->as_mode][2];
+
+ vp8_blend_b
+ (yl+bx, ul+(bx>>1), vl+(bx>>1), Y, U, V, 0xc000, y_stride);
+ }
+ bmi++;
+ }
+
+ yl += y_stride*4;
+ ul += y_stride*1;
+ vl += y_stride*1;
+ }
+ }
+ else if (ppflags->display_mb_modes_flag & (1<<mi->mbmi.mode))
+ {
+ Y = MB_PREDICTION_MODE_colors[mi->mbmi.mode][0];
+ U = MB_PREDICTION_MODE_colors[mi->mbmi.mode][1];
+ V = MB_PREDICTION_MODE_colors[mi->mbmi.mode][2];
+
+ vp8_blend_mb_inner
+ (y_ptr+x, u_ptr+(x>>1), v_ptr+(x>>1), Y, U, V, 0xc000, y_stride);
+ }
+
+ mi++;
+ }
+ y_ptr += y_stride*16;
+ u_ptr += y_stride*4;
+ v_ptr += y_stride*4;
+
+ mi++;
+ }
+ }
+
+ /* Color in frame reference blocks */
+ if ((flags & VP8D_DEBUG_CLR_FRM_REF_BLKS) && ppflags->display_ref_frame_flag)
+ {
+ int y, x;
+ YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
+ int width = post->y_width;
+ int height = post->y_height;
+ unsigned char *y_ptr = oci->post_proc_buffer.y_buffer;
+ unsigned char *u_ptr = oci->post_proc_buffer.u_buffer;
+ unsigned char *v_ptr = oci->post_proc_buffer.v_buffer;
+ int y_stride = oci->post_proc_buffer.y_stride;
+ MODE_INFO *mi = oci->mi;
+
+ for (y = 0; y < height; y += 16)
+ {
+ for (x = 0; x < width; x +=16)
+ {
+ int Y = 0, U = 0, V = 0;
+
+ if (ppflags->display_ref_frame_flag & (1<<mi->mbmi.ref_frame))
+ {
+ Y = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][0];
+ U = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][1];
+ V = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][2];
+
+ vp8_blend_mb_outer
+ (y_ptr+x, u_ptr+(x>>1), v_ptr+(x>>1), Y, U, V, 0xc000, y_stride);
+ }
+
+ mi++;
+ }
+ y_ptr += y_stride*16;
+ u_ptr += y_stride*4;
+ v_ptr += y_stride*4;
+
+ mi++;
+ }
+ }
+#endif
+
+ *dest = oci->post_proc_buffer;
+
+ /* handle problem with extending borders */
+ dest->y_width = oci->Width;
+ dest->y_height = oci->Height;
+ dest->uv_height = dest->y_height / 2;
+ return 0;
+}
+#endif
diff --git a/libvpx/vp8/common/postproc.h b/libvpx/vp8/common/postproc.h
new file mode 100644
index 0000000..495a2c9
--- /dev/null
+++ b/libvpx/vp8/common/postproc.h
@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef POSTPROC_H
+#define POSTPROC_H
+
+#include "vpx_ports/mem.h"
+struct postproc_state
+{
+ int last_q;
+ int last_noise;
+ char noise[3072];
+ int last_base_qindex;
+ int last_frame_valid;
+ DECLARE_ALIGNED(16, char, blackclamp[16]);
+ DECLARE_ALIGNED(16, char, whiteclamp[16]);
+ DECLARE_ALIGNED(16, char, bothclamp[16]);
+};
+#include "onyxc_int.h"
+#include "ppflags.h"
+int vp8_post_proc_frame(struct VP8Common *oci, YV12_BUFFER_CONFIG *dest,
+ vp8_ppflags_t *flags);
+
+
+void vp8_de_noise(struct VP8Common *oci,
+ YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *post,
+ int q,
+ int low_var_thresh,
+ int flag);
+
+void vp8_deblock(struct VP8Common *oci,
+ YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *post,
+ int q,
+ int low_var_thresh,
+ int flag);
+
+#define MFQE_PRECISION 4
+
+void vp8_multiframe_quality_enhance(struct VP8Common *cm);
+#endif
diff --git a/vp8/common/ppc/copy_altivec.asm b/libvpx/vp8/common/ppc/copy_altivec.asm
similarity index 100%
rename from vp8/common/ppc/copy_altivec.asm
rename to libvpx/vp8/common/ppc/copy_altivec.asm
diff --git a/vp8/common/ppc/filter_altivec.asm b/libvpx/vp8/common/ppc/filter_altivec.asm
similarity index 100%
rename from vp8/common/ppc/filter_altivec.asm
rename to libvpx/vp8/common/ppc/filter_altivec.asm
diff --git a/vp8/common/ppc/filter_bilinear_altivec.asm b/libvpx/vp8/common/ppc/filter_bilinear_altivec.asm
similarity index 100%
rename from vp8/common/ppc/filter_bilinear_altivec.asm
rename to libvpx/vp8/common/ppc/filter_bilinear_altivec.asm
diff --git a/vp8/common/ppc/idctllm_altivec.asm b/libvpx/vp8/common/ppc/idctllm_altivec.asm
similarity index 100%
rename from vp8/common/ppc/idctllm_altivec.asm
rename to libvpx/vp8/common/ppc/idctllm_altivec.asm
diff --git a/libvpx/vp8/common/ppc/loopfilter_altivec.c b/libvpx/vp8/common/ppc/loopfilter_altivec.c
new file mode 100644
index 0000000..71bf6e2
--- /dev/null
+++ b/libvpx/vp8/common/ppc/loopfilter_altivec.c
@@ -0,0 +1,135 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "loopfilter.h"
+#include "onyxc_int.h"
+
+typedef void loop_filter_function_y_ppc
+(
+ unsigned char *s, // source pointer
+ int p, // pitch
+ const signed char *flimit,
+ const signed char *limit,
+ const signed char *thresh
+);
+
+typedef void loop_filter_function_uv_ppc
+(
+ unsigned char *u, // source pointer
+ unsigned char *v, // source pointer
+ int p, // pitch
+ const signed char *flimit,
+ const signed char *limit,
+ const signed char *thresh
+);
+
+typedef void loop_filter_function_s_ppc
+(
+ unsigned char *s, // source pointer
+ int p, // pitch
+ const signed char *flimit
+);
+
+loop_filter_function_y_ppc mbloop_filter_horizontal_edge_y_ppc;
+loop_filter_function_y_ppc mbloop_filter_vertical_edge_y_ppc;
+loop_filter_function_y_ppc loop_filter_horizontal_edge_y_ppc;
+loop_filter_function_y_ppc loop_filter_vertical_edge_y_ppc;
+
+loop_filter_function_uv_ppc mbloop_filter_horizontal_edge_uv_ppc;
+loop_filter_function_uv_ppc mbloop_filter_vertical_edge_uv_ppc;
+loop_filter_function_uv_ppc loop_filter_horizontal_edge_uv_ppc;
+loop_filter_function_uv_ppc loop_filter_vertical_edge_uv_ppc;
+
+loop_filter_function_s_ppc loop_filter_simple_horizontal_edge_ppc;
+loop_filter_function_s_ppc loop_filter_simple_vertical_edge_ppc;
+
+// Horizontal MB filtering
+void loop_filter_mbh_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ mbloop_filter_horizontal_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr);
+
+ if (u_ptr)
+ mbloop_filter_horizontal_edge_uv_ppc(u_ptr, v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr);
+}
+
+void loop_filter_mbhs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ (void)u_ptr;
+ (void)v_ptr;
+ (void)uv_stride;
+ loop_filter_simple_horizontal_edge_ppc(y_ptr, y_stride, lfi->mbflim);
+}
+
+// Vertical MB Filtering
+void loop_filter_mbv_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ mbloop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr);
+
+ if (u_ptr)
+ mbloop_filter_vertical_edge_uv_ppc(u_ptr, v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr);
+}
+
+void loop_filter_mbvs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ (void)u_ptr;
+ (void)v_ptr;
+ (void)uv_stride;
+ loop_filter_simple_vertical_edge_ppc(y_ptr, y_stride, lfi->mbflim);
+}
+
+// Horizontal B Filtering
+void loop_filter_bh_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ // These should all be done at once with one call, instead of 3
+ loop_filter_horizontal_edge_y_ppc(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
+ loop_filter_horizontal_edge_y_ppc(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
+ loop_filter_horizontal_edge_y_ppc(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
+
+ if (u_ptr)
+ loop_filter_horizontal_edge_uv_ppc(u_ptr + 4 * uv_stride, v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr);
+}
+
+void loop_filter_bhs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ (void)u_ptr;
+ (void)v_ptr;
+ (void)uv_stride;
+ loop_filter_simple_horizontal_edge_ppc(y_ptr + 4 * y_stride, y_stride, lfi->flim);
+ loop_filter_simple_horizontal_edge_ppc(y_ptr + 8 * y_stride, y_stride, lfi->flim);
+ loop_filter_simple_horizontal_edge_ppc(y_ptr + 12 * y_stride, y_stride, lfi->flim);
+}
+
+// Vertical B Filtering
+void loop_filter_bv_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ loop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->flim, lfi->lim, lfi->thr);
+
+ if (u_ptr)
+ loop_filter_vertical_edge_uv_ppc(u_ptr + 4, v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr);
+}
+
+void loop_filter_bvs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ (void)u_ptr;
+ (void)v_ptr;
+ (void)uv_stride;
+ loop_filter_simple_vertical_edge_ppc(y_ptr + 4, y_stride, lfi->flim);
+ loop_filter_simple_vertical_edge_ppc(y_ptr + 8, y_stride, lfi->flim);
+ loop_filter_simple_vertical_edge_ppc(y_ptr + 12, y_stride, lfi->flim);
+}
diff --git a/vp8/common/ppc/loopfilter_filters_altivec.asm b/libvpx/vp8/common/ppc/loopfilter_filters_altivec.asm
similarity index 100%
rename from vp8/common/ppc/loopfilter_filters_altivec.asm
rename to libvpx/vp8/common/ppc/loopfilter_filters_altivec.asm
diff --git a/vp8/common/ppc/platform_altivec.asm b/libvpx/vp8/common/ppc/platform_altivec.asm
similarity index 100%
rename from vp8/common/ppc/platform_altivec.asm
rename to libvpx/vp8/common/ppc/platform_altivec.asm
diff --git a/vp8/common/ppc/recon_altivec.asm b/libvpx/vp8/common/ppc/recon_altivec.asm
similarity index 100%
rename from vp8/common/ppc/recon_altivec.asm
rename to libvpx/vp8/common/ppc/recon_altivec.asm
diff --git a/vp8/encoder/ppc/sad_altivec.asm b/libvpx/vp8/common/ppc/sad_altivec.asm
similarity index 100%
rename from vp8/encoder/ppc/sad_altivec.asm
rename to libvpx/vp8/common/ppc/sad_altivec.asm
diff --git a/libvpx/vp8/common/ppc/systemdependent.c b/libvpx/vp8/common/ppc/systemdependent.c
new file mode 100644
index 0000000..87f4cac
--- /dev/null
+++ b/libvpx/vp8/common/ppc/systemdependent.c
@@ -0,0 +1,170 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "subpixel.h"
+#include "loopfilter.h"
+#include "recon.h"
+#include "idct.h"
+#include "onyxc_int.h"
+
+void (*vp8_short_idct4x4)(short *input, short *output, int pitch);
+void (*vp8_short_idct4x4_1)(short *input, short *output, int pitch);
+void (*vp8_dc_only_idct)(short input_dc, short *output, int pitch);
+
+extern void (*vp8_post_proc_down_and_across_mb_row)(
+ unsigned char *src_ptr,
+ unsigned char *dst_ptr,
+ int src_pixels_per_line,
+ int dst_pixels_per_line,
+ int cols,
+ unsigned char *f,
+ int size
+);
+
+extern void (*vp8_mbpost_proc_down)(unsigned char *dst, int pitch, int rows, int cols, int flimit);
+extern void vp8_mbpost_proc_down_c(unsigned char *dst, int pitch, int rows, int cols, int flimit);
+extern void (*vp8_mbpost_proc_across_ip)(unsigned char *src, int pitch, int rows, int cols, int flimit);
+extern void vp8_mbpost_proc_across_ip_c(unsigned char *src, int pitch, int rows, int cols, int flimit);
+
+extern void vp8_post_proc_down_and_across_mb_row_c
+(
+ unsigned char *src_ptr,
+ unsigned char *dst_ptr,
+ int src_pixels_per_line,
+ int dst_pixels_per_line,
+ int cols,
+ unsigned char *f,
+ int size
+);
+void vp8_plane_add_noise_c(unsigned char *Start, unsigned int Width, unsigned int Height, int Pitch, int q, int a);
+
+extern copy_mem_block_function *vp8_copy_mem16x16;
+extern copy_mem_block_function *vp8_copy_mem8x8;
+extern copy_mem_block_function *vp8_copy_mem8x4;
+
+// PPC
+extern subpixel_predict_function sixtap_predict_ppc;
+extern subpixel_predict_function sixtap_predict8x4_ppc;
+extern subpixel_predict_function sixtap_predict8x8_ppc;
+extern subpixel_predict_function sixtap_predict16x16_ppc;
+extern subpixel_predict_function bilinear_predict4x4_ppc;
+extern subpixel_predict_function bilinear_predict8x4_ppc;
+extern subpixel_predict_function bilinear_predict8x8_ppc;
+extern subpixel_predict_function bilinear_predict16x16_ppc;
+
+extern copy_mem_block_function copy_mem16x16_ppc;
+
+void recon_b_ppc(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
+void recon2b_ppc(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
+void recon4b_ppc(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
+
+extern void short_idct4x4llm_ppc(short *input, short *output, int pitch);
+
+// Generic C
+extern subpixel_predict_function vp8_sixtap_predict_c;
+extern subpixel_predict_function vp8_sixtap_predict8x4_c;
+extern subpixel_predict_function vp8_sixtap_predict8x8_c;
+extern subpixel_predict_function vp8_sixtap_predict16x16_c;
+extern subpixel_predict_function vp8_bilinear_predict4x4_c;
+extern subpixel_predict_function vp8_bilinear_predict8x4_c;
+extern subpixel_predict_function vp8_bilinear_predict8x8_c;
+extern subpixel_predict_function vp8_bilinear_predict16x16_c;
+
+extern copy_mem_block_function vp8_copy_mem16x16_c;
+extern copy_mem_block_function vp8_copy_mem8x8_c;
+extern copy_mem_block_function vp8_copy_mem8x4_c;
+
+void vp8_recon_b_c(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
+void vp8_recon2b_c(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
+void vp8_recon4b_c(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
+
+extern void vp8_short_idct4x4llm_1_c(short *input, short *output, int pitch);
+extern void vp8_short_idct4x4llm_c(short *input, short *output, int pitch);
+extern void vp8_dc_only_idct_c(short input_dc, short *output, int pitch);
+
+// PPC
+extern loop_filter_block_function loop_filter_mbv_ppc;
+extern loop_filter_block_function loop_filter_bv_ppc;
+extern loop_filter_block_function loop_filter_mbh_ppc;
+extern loop_filter_block_function loop_filter_bh_ppc;
+
+extern loop_filter_block_function loop_filter_mbvs_ppc;
+extern loop_filter_block_function loop_filter_bvs_ppc;
+extern loop_filter_block_function loop_filter_mbhs_ppc;
+extern loop_filter_block_function loop_filter_bhs_ppc;
+
+// Generic C
+extern loop_filter_block_function vp8_loop_filter_mbv_c;
+extern loop_filter_block_function vp8_loop_filter_bv_c;
+extern loop_filter_block_function vp8_loop_filter_mbh_c;
+extern loop_filter_block_function vp8_loop_filter_bh_c;
+
+extern loop_filter_block_function vp8_loop_filter_mbvs_c;
+extern loop_filter_block_function vp8_loop_filter_bvs_c;
+extern loop_filter_block_function vp8_loop_filter_mbhs_c;
+extern loop_filter_block_function vp8_loop_filter_bhs_c;
+
+extern loop_filter_block_function *vp8_lf_mbvfull;
+extern loop_filter_block_function *vp8_lf_mbhfull;
+extern loop_filter_block_function *vp8_lf_bvfull;
+extern loop_filter_block_function *vp8_lf_bhfull;
+
+extern loop_filter_block_function *vp8_lf_mbvsimple;
+extern loop_filter_block_function *vp8_lf_mbhsimple;
+extern loop_filter_block_function *vp8_lf_bvsimple;
+extern loop_filter_block_function *vp8_lf_bhsimple;
+
+void vp8_clear_c(void)
+{
+}
+
+void vp8_machine_specific_config(void)
+{
+ // Pure C:
+ vp8_clear_system_state = vp8_clear_c;
+ vp8_recon_b = vp8_recon_b_c;
+ vp8_recon4b = vp8_recon4b_c;
+ vp8_recon2b = vp8_recon2b_c;
+
+ vp8_bilinear_predict16x16 = bilinear_predict16x16_ppc;
+ vp8_bilinear_predict8x8 = bilinear_predict8x8_ppc;
+ vp8_bilinear_predict8x4 = bilinear_predict8x4_ppc;
+ vp8_bilinear_predict = bilinear_predict4x4_ppc;
+
+ vp8_sixtap_predict16x16 = sixtap_predict16x16_ppc;
+ vp8_sixtap_predict8x8 = sixtap_predict8x8_ppc;
+ vp8_sixtap_predict8x4 = sixtap_predict8x4_ppc;
+ vp8_sixtap_predict = sixtap_predict_ppc;
+
+ vp8_short_idct4x4_1 = vp8_short_idct4x4llm_1_c;
+ vp8_short_idct4x4 = short_idct4x4llm_ppc;
+ vp8_dc_only_idct = vp8_dc_only_idct_c;
+
+ vp8_lf_mbvfull = loop_filter_mbv_ppc;
+ vp8_lf_bvfull = loop_filter_bv_ppc;
+ vp8_lf_mbhfull = loop_filter_mbh_ppc;
+ vp8_lf_bhfull = loop_filter_bh_ppc;
+
+ vp8_lf_mbvsimple = loop_filter_mbvs_ppc;
+ vp8_lf_bvsimple = loop_filter_bvs_ppc;
+ vp8_lf_mbhsimple = loop_filter_mbhs_ppc;
+ vp8_lf_bhsimple = loop_filter_bhs_ppc;
+
+ vp8_post_proc_down_and_across_mb_row = vp8_post_proc_down_and_across_mb_row_c;
+ vp8_mbpost_proc_down = vp8_mbpost_proc_down_c;
+ vp8_mbpost_proc_across_ip = vp8_mbpost_proc_across_ip_c;
+ vp8_plane_add_noise = vp8_plane_add_noise_c;
+
+ vp8_copy_mem16x16 = copy_mem16x16_ppc;
+ vp8_copy_mem8x8 = vp8_copy_mem8x8_c;
+ vp8_copy_mem8x4 = vp8_copy_mem8x4_c;
+
+}
diff --git a/libvpx/vp8/common/ppc/variance_altivec.asm b/libvpx/vp8/common/ppc/variance_altivec.asm
new file mode 100644
index 0000000..fb8d5bb
--- /dev/null
+++ b/libvpx/vp8/common/ppc/variance_altivec.asm
@@ -0,0 +1,375 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ .globl vp8_get8x8var_ppc
+ .globl vp8_get16x16var_ppc
+ .globl vp8_mse16x16_ppc
+ .globl vp8_variance16x16_ppc
+ .globl vp8_variance16x8_ppc
+ .globl vp8_variance8x16_ppc
+ .globl vp8_variance8x8_ppc
+ .globl vp8_variance4x4_ppc
+
+.macro load_aligned_16 V R O
+ lvsl v3, 0, \R ;# permutate value for alignment
+
+ lvx v1, 0, \R
+ lvx v2, \O, \R
+
+ vperm \V, v1, v2, v3
+.endm
+
+.macro prologue
+ mfspr r11, 256 ;# get old VRSAVE
+ oris r12, r11, 0xffc0
+ mtspr 256, r12 ;# set VRSAVE
+
+ stwu r1, -32(r1) ;# create space on the stack
+
+ li r10, 16 ;# load offset and loop counter
+
+ vspltisw v7, 0 ;# zero for merging
+ vspltisw v8, 0 ;# zero out total to start
+ vspltisw v9, 0 ;# zero out total for dif^2
+.endm
+
+.macro epilogue
+ addi r1, r1, 32 ;# recover stack
+
+ mtspr 256, r11 ;# reset old VRSAVE
+.endm
+
+.macro compute_sum_sse
+ ;# Compute sum first. Unpack to so signed subract
+ ;# can be used. Only have a half word signed
+ ;# subract. Do high, then low.
+ vmrghb v2, v7, v4
+ vmrghb v3, v7, v5
+ vsubshs v2, v2, v3
+ vsum4shs v8, v2, v8
+
+ vmrglb v2, v7, v4
+ vmrglb v3, v7, v5
+ vsubshs v2, v2, v3
+ vsum4shs v8, v2, v8
+
+ ;# Now compute sse.
+ vsububs v2, v4, v5
+ vsububs v3, v5, v4
+ vor v2, v2, v3
+
+ vmsumubm v9, v2, v2, v9
+.endm
+
+.macro variance_16 DS loop_label store_sum
+\loop_label:
+ ;# only one of the inputs should need to be aligned.
+ load_aligned_16 v4, r3, r10
+ load_aligned_16 v5, r5, r10
+
+ ;# move onto the next line
+ add r3, r3, r4
+ add r5, r5, r6
+
+ compute_sum_sse
+
+ bdnz \loop_label
+
+ vsumsws v8, v8, v7
+ vsumsws v9, v9, v7
+
+ stvx v8, 0, r1
+ lwz r3, 12(r1)
+
+ stvx v9, 0, r1
+ lwz r4, 12(r1)
+
+.if \store_sum
+ stw r3, 0(r8) ;# sum
+.endif
+ stw r4, 0(r7) ;# sse
+
+ mullw r3, r3, r3 ;# sum*sum
+ srlwi r3, r3, \DS ;# (sum*sum) >> DS
+ subf r3, r3, r4 ;# sse - ((sum*sum) >> DS)
+.endm
+
+.macro variance_8 DS loop_label store_sum
+\loop_label:
+ ;# only one of the inputs should need to be aligned.
+ load_aligned_16 v4, r3, r10
+ load_aligned_16 v5, r5, r10
+
+ ;# move onto the next line
+ add r3, r3, r4
+ add r5, r5, r6
+
+ ;# only one of the inputs should need to be aligned.
+ load_aligned_16 v6, r3, r10
+ load_aligned_16 v0, r5, r10
+
+ ;# move onto the next line
+ add r3, r3, r4
+ add r5, r5, r6
+
+ vmrghb v4, v4, v6
+ vmrghb v5, v5, v0
+
+ compute_sum_sse
+
+ bdnz \loop_label
+
+ vsumsws v8, v8, v7
+ vsumsws v9, v9, v7
+
+ stvx v8, 0, r1
+ lwz r3, 12(r1)
+
+ stvx v9, 0, r1
+ lwz r4, 12(r1)
+
+.if \store_sum
+ stw r3, 0(r8) ;# sum
+.endif
+ stw r4, 0(r7) ;# sse
+
+ mullw r3, r3, r3 ;# sum*sum
+ srlwi r3, r3, \DS ;# (sum*sum) >> 8
+ subf r3, r3, r4 ;# sse - ((sum*sum) >> 8)
+.endm
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int source_stride
+;# r5 unsigned char *ref_ptr
+;# r6 int recon_stride
+;# r7 unsigned int *SSE
+;# r8 int *Sum
+;#
+;# r3 return value
+vp8_get8x8var_ppc:
+
+ prologue
+
+ li r9, 4
+ mtctr r9
+
+ variance_8 6, get8x8var_loop, 1
+
+ epilogue
+
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int source_stride
+;# r5 unsigned char *ref_ptr
+;# r6 int recon_stride
+;# r7 unsigned int *SSE
+;# r8 int *Sum
+;#
+;# r3 return value
+vp8_get16x16var_ppc:
+
+ prologue
+
+ mtctr r10
+
+ variance_16 8, get16x16var_loop, 1
+
+ epilogue
+
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int source_stride
+;# r5 unsigned char *ref_ptr
+;# r6 int recon_stride
+;# r7 unsigned int *sse
+;#
+;# r 3 return value
+vp8_mse16x16_ppc:
+ prologue
+
+ mtctr r10
+
+mse16x16_loop:
+ ;# only one of the inputs should need to be aligned.
+ load_aligned_16 v4, r3, r10
+ load_aligned_16 v5, r5, r10
+
+ ;# move onto the next line
+ add r3, r3, r4
+ add r5, r5, r6
+
+ ;# Now compute sse.
+ vsububs v2, v4, v5
+ vsububs v3, v5, v4
+ vor v2, v2, v3
+
+ vmsumubm v9, v2, v2, v9
+
+ bdnz mse16x16_loop
+
+ vsumsws v9, v9, v7
+
+ stvx v9, 0, r1
+ lwz r3, 12(r1)
+
+ stvx v9, 0, r1
+ lwz r3, 12(r1)
+
+ stw r3, 0(r7) ;# sse
+
+ epilogue
+
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int source_stride
+;# r5 unsigned char *ref_ptr
+;# r6 int recon_stride
+;# r7 unsigned int *sse
+;#
+;# r3 return value
+vp8_variance16x16_ppc:
+
+ prologue
+
+ mtctr r10
+
+ variance_16 8, variance16x16_loop, 0
+
+ epilogue
+
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int source_stride
+;# r5 unsigned char *ref_ptr
+;# r6 int recon_stride
+;# r7 unsigned int *sse
+;#
+;# r3 return value
+vp8_variance16x8_ppc:
+
+ prologue
+
+ li r9, 8
+ mtctr r9
+
+ variance_16 7, variance16x8_loop, 0
+
+ epilogue
+
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int source_stride
+;# r5 unsigned char *ref_ptr
+;# r6 int recon_stride
+;# r7 unsigned int *sse
+;#
+;# r3 return value
+vp8_variance8x16_ppc:
+
+ prologue
+
+ li r9, 8
+ mtctr r9
+
+ variance_8 7, variance8x16_loop, 0
+
+ epilogue
+
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int source_stride
+;# r5 unsigned char *ref_ptr
+;# r6 int recon_stride
+;# r7 unsigned int *sse
+;#
+;# r3 return value
+vp8_variance8x8_ppc:
+
+ prologue
+
+ li r9, 4
+ mtctr r9
+
+ variance_8 6, variance8x8_loop, 0
+
+ epilogue
+
+ blr
+
+.macro transfer_4x4 I P
+ lwz r0, 0(\I)
+ add \I, \I, \P
+
+ lwz r10,0(\I)
+ add \I, \I, \P
+
+ lwz r8, 0(\I)
+ add \I, \I, \P
+
+ lwz r9, 0(\I)
+
+ stw r0, 0(r1)
+ stw r10, 4(r1)
+ stw r8, 8(r1)
+ stw r9, 12(r1)
+.endm
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int source_stride
+;# r5 unsigned char *ref_ptr
+;# r6 int recon_stride
+;# r7 unsigned int *sse
+;#
+;# r3 return value
+vp8_variance4x4_ppc:
+
+ prologue
+
+ transfer_4x4 r3, r4
+ lvx v4, 0, r1
+
+ transfer_4x4 r5, r6
+ lvx v5, 0, r1
+
+ compute_sum_sse
+
+ vsumsws v8, v8, v7
+ vsumsws v9, v9, v7
+
+ stvx v8, 0, r1
+ lwz r3, 12(r1)
+
+ stvx v9, 0, r1
+ lwz r4, 12(r1)
+
+ stw r4, 0(r7) ;# sse
+
+ mullw r3, r3, r3 ;# sum*sum
+ srlwi r3, r3, 4 ;# (sum*sum) >> 4
+ subf r3, r3, r4 ;# sse - ((sum*sum) >> 4)
+
+ epilogue
+
+ blr
diff --git a/libvpx/vp8/common/ppc/variance_subpixel_altivec.asm b/libvpx/vp8/common/ppc/variance_subpixel_altivec.asm
new file mode 100644
index 0000000..2308373
--- /dev/null
+++ b/libvpx/vp8/common/ppc/variance_subpixel_altivec.asm
@@ -0,0 +1,865 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ .globl vp8_sub_pixel_variance4x4_ppc
+ .globl vp8_sub_pixel_variance8x8_ppc
+ .globl vp8_sub_pixel_variance8x16_ppc
+ .globl vp8_sub_pixel_variance16x8_ppc
+ .globl vp8_sub_pixel_variance16x16_ppc
+
+.macro load_c V, LABEL, OFF, R0, R1
+ lis \R0, \LABEL@ha
+ la \R1, \LABEL@l(\R0)
+ lvx \V, \OFF, \R1
+.endm
+
+.macro load_vfilter V0, V1
+ load_c \V0, vfilter_b, r6, r12, r10
+
+ addi r6, r6, 16
+ lvx \V1, r6, r10
+.endm
+
+.macro HProlog jump_label
+ ;# load up horizontal filter
+ slwi. r5, r5, 4 ;# index into horizontal filter array
+
+ ;# index to the next set of vectors in the row.
+ li r10, 16
+
+ ;# downshift by 7 ( divide by 128 ) at the end
+ vspltish v19, 7
+
+ ;# If there isn't any filtering to be done for the horizontal, then
+ ;# just skip to the second pass.
+ beq \jump_label
+
+ load_c v20, hfilter_b, r5, r12, r0
+
+ ;# setup constants
+ ;# v14 permutation value for alignment
+ load_c v28, b_hperm_b, 0, r12, r0
+
+ ;# index to the next set of vectors in the row.
+ li r12, 32
+
+ ;# rounding added in on the multiply
+ vspltisw v21, 8
+ vspltisw v18, 3
+ vslw v18, v21, v18 ;# 0x00000040000000400000004000000040
+
+ slwi. r6, r6, 5 ;# index into vertical filter array
+.endm
+
+;# Filters a horizontal line
+;# expects:
+;# r3 src_ptr
+;# r4 pitch
+;# r10 16
+;# r12 32
+;# v17 perm intput
+;# v18 rounding
+;# v19 shift
+;# v20 filter taps
+;# v21 tmp
+;# v22 tmp
+;# v23 tmp
+;# v24 tmp
+;# v25 tmp
+;# v26 tmp
+;# v27 tmp
+;# v28 perm output
+;#
+
+.macro hfilter_8 V, hp, lp, increment_counter
+ lvsl v17, 0, r3 ;# permutate value for alignment
+
+ ;# input to filter is 9 bytes wide, output is 8 bytes.
+ lvx v21, 0, r3
+ lvx v22, r10, r3
+
+.if \increment_counter
+ add r3, r3, r4
+.endif
+ vperm v21, v21, v22, v17
+
+ vperm v24, v21, v21, \hp ;# v20 = 0123 1234 2345 3456
+ vperm v25, v21, v21, \lp ;# v21 = 4567 5678 6789 789A
+
+ vmsummbm v24, v20, v24, v18
+ vmsummbm v25, v20, v25, v18
+
+ vpkswus v24, v24, v25 ;# v24 = 0 4 8 C 1 5 9 D (16-bit)
+
+ vsrh v24, v24, v19 ;# divide v0, v1 by 128
+
+ vpkuhus \V, v24, v24 ;# \V = scrambled 8-bit result
+.endm
+
+.macro vfilter_16 P0 P1
+ vmuleub v22, \P0, v20 ;# 64 + 4 positive taps
+ vadduhm v22, v18, v22
+ vmuloub v23, \P0, v20
+ vadduhm v23, v18, v23
+
+ vmuleub v24, \P1, v21
+ vadduhm v22, v22, v24 ;# Re = evens, saturation unnecessary
+ vmuloub v25, \P1, v21
+ vadduhm v23, v23, v25 ;# Ro = odds
+
+ vsrh v22, v22, v19 ;# divide by 128
+ vsrh v23, v23, v19 ;# v16 v17 = evens, odds
+ vmrghh \P0, v22, v23 ;# v18 v19 = 16-bit result in order
+ vmrglh v23, v22, v23
+ vpkuhus \P0, \P0, v23 ;# P0 = 8-bit result
+.endm
+
+.macro compute_sum_sse src, ref, sum, sse, t1, t2, z0
+ ;# Compute sum first. Unpack to so signed subract
+ ;# can be used. Only have a half word signed
+ ;# subract. Do high, then low.
+ vmrghb \t1, \z0, \src
+ vmrghb \t2, \z0, \ref
+ vsubshs \t1, \t1, \t2
+ vsum4shs \sum, \t1, \sum
+
+ vmrglb \t1, \z0, \src
+ vmrglb \t2, \z0, \ref
+ vsubshs \t1, \t1, \t2
+ vsum4shs \sum, \t1, \sum
+
+ ;# Now compute sse.
+ vsububs \t1, \src, \ref
+ vsububs \t2, \ref, \src
+ vor \t1, \t1, \t2
+
+ vmsumubm \sse, \t1, \t1, \sse
+.endm
+
+.macro variance_final sum, sse, z0, DS
+ vsumsws \sum, \sum, \z0
+ vsumsws \sse, \sse, \z0
+
+ stvx \sum, 0, r1
+ lwz r3, 12(r1)
+
+ stvx \sse, 0, r1
+ lwz r4, 12(r1)
+
+ stw r4, 0(r9) ;# sse
+
+ mullw r3, r3, r3 ;# sum*sum
+ srlwi r3, r3, \DS ;# (sum*sum) >> 8
+ subf r3, r3, r4 ;# sse - ((sum*sum) >> 8)
+.endm
+
+.macro compute_sum_sse_16 V, increment_counter
+ load_and_align_16 v16, r7, r8, \increment_counter
+ compute_sum_sse \V, v16, v18, v19, v20, v21, v23
+.endm
+
+.macro load_and_align_16 V, R, P, increment_counter
+ lvsl v17, 0, \R ;# permutate value for alignment
+
+ ;# input to filter is 21 bytes wide, output is 16 bytes.
+ ;# input will can span three vectors if not aligned correctly.
+ lvx v21, 0, \R
+ lvx v22, r10, \R
+
+.if \increment_counter
+ add \R, \R, \P
+.endif
+
+ vperm \V, v21, v22, v17
+.endm
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int src_pixels_per_line
+;# r5 int xoffset
+;# r6 int yoffset
+;# r7 unsigned char *dst_ptr
+;# r8 int dst_pixels_per_line
+;# r9 unsigned int *sse
+;#
+;# r3 return value
+vp8_sub_pixel_variance4x4_ppc:
+ mfspr r11, 256 ;# get old VRSAVE
+ oris r12, r11, 0xf830
+ ori r12, r12, 0xfff8
+ mtspr 256, r12 ;# set VRSAVE
+
+ stwu r1,-32(r1) ;# create space on the stack
+
+ HProlog second_pass_4x4_pre_copy_b
+
+ ;# Load up permutation constants
+ load_c v10, b_0123_b, 0, r12, r0
+ load_c v11, b_4567_b, 0, r12, r0
+
+ hfilter_8 v0, v10, v11, 1
+ hfilter_8 v1, v10, v11, 1
+ hfilter_8 v2, v10, v11, 1
+ hfilter_8 v3, v10, v11, 1
+
+ ;# Finished filtering main horizontal block. If there is no
+ ;# vertical filtering, jump to storing the data. Otherwise
+ ;# load up and filter the additional line that is needed
+ ;# for the vertical filter.
+ beq compute_sum_sse_4x4_b
+
+ hfilter_8 v4, v10, v11, 0
+
+ b second_pass_4x4_b
+
+second_pass_4x4_pre_copy_b:
+ slwi r6, r6, 5 ;# index into vertical filter array
+
+ load_and_align_16 v0, r3, r4, 1
+ load_and_align_16 v1, r3, r4, 1
+ load_and_align_16 v2, r3, r4, 1
+ load_and_align_16 v3, r3, r4, 1
+ load_and_align_16 v4, r3, r4, 0
+
+second_pass_4x4_b:
+ vspltish v20, 8
+ vspltish v18, 3
+ vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
+
+ load_vfilter v20, v21
+
+ vfilter_16 v0, v1
+ vfilter_16 v1, v2
+ vfilter_16 v2, v3
+ vfilter_16 v3, v4
+
+compute_sum_sse_4x4_b:
+ vspltish v18, 0 ;# sum
+ vspltish v19, 0 ;# sse
+ vspltish v23, 0 ;# unpack
+ li r10, 16
+
+ load_and_align_16 v4, r7, r8, 1
+ load_and_align_16 v5, r7, r8, 1
+ load_and_align_16 v6, r7, r8, 1
+ load_and_align_16 v7, r7, r8, 1
+
+ vmrghb v0, v0, v1
+ vmrghb v1, v2, v3
+
+ vmrghb v2, v4, v5
+ vmrghb v3, v6, v7
+
+ load_c v10, b_hilo_b, 0, r12, r0
+
+ vperm v0, v0, v1, v10
+ vperm v1, v2, v3, v10
+
+ compute_sum_sse v0, v1, v18, v19, v20, v21, v23
+
+ variance_final v18, v19, v23, 4
+
+ addi r1, r1, 32 ;# recover stack
+ mtspr 256, r11 ;# reset old VRSAVE
+
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int src_pixels_per_line
+;# r5 int xoffset
+;# r6 int yoffset
+;# r7 unsigned char *dst_ptr
+;# r8 int dst_pixels_per_line
+;# r9 unsigned int *sse
+;#
+;# r3 return value
+vp8_sub_pixel_variance8x8_ppc:
+ mfspr r11, 256 ;# get old VRSAVE
+ oris r12, r11, 0xfff0
+ ori r12, r12, 0xffff
+ mtspr 256, r12 ;# set VRSAVE
+
+ stwu r1,-32(r1) ;# create space on the stack
+
+ HProlog second_pass_8x8_pre_copy_b
+
+ ;# Load up permutation constants
+ load_c v10, b_0123_b, 0, r12, r0
+ load_c v11, b_4567_b, 0, r12, r0
+
+ hfilter_8 v0, v10, v11, 1
+ hfilter_8 v1, v10, v11, 1
+ hfilter_8 v2, v10, v11, 1
+ hfilter_8 v3, v10, v11, 1
+ hfilter_8 v4, v10, v11, 1
+ hfilter_8 v5, v10, v11, 1
+ hfilter_8 v6, v10, v11, 1
+ hfilter_8 v7, v10, v11, 1
+
+ ;# Finished filtering main horizontal block. If there is no
+ ;# vertical filtering, jump to storing the data. Otherwise
+ ;# load up and filter the additional line that is needed
+ ;# for the vertical filter.
+ beq compute_sum_sse_8x8_b
+
+ hfilter_8 v8, v10, v11, 0
+
+ b second_pass_8x8_b
+
+second_pass_8x8_pre_copy_b:
+ slwi. r6, r6, 5 ;# index into vertical filter array
+
+ load_and_align_16 v0, r3, r4, 1
+ load_and_align_16 v1, r3, r4, 1
+ load_and_align_16 v2, r3, r4, 1
+ load_and_align_16 v3, r3, r4, 1
+ load_and_align_16 v4, r3, r4, 1
+ load_and_align_16 v5, r3, r4, 1
+ load_and_align_16 v6, r3, r4, 1
+ load_and_align_16 v7, r3, r4, 1
+ load_and_align_16 v8, r3, r4, 0
+
+ beq compute_sum_sse_8x8_b
+
+second_pass_8x8_b:
+ vspltish v20, 8
+ vspltish v18, 3
+ vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
+
+ load_vfilter v20, v21
+
+ vfilter_16 v0, v1
+ vfilter_16 v1, v2
+ vfilter_16 v2, v3
+ vfilter_16 v3, v4
+ vfilter_16 v4, v5
+ vfilter_16 v5, v6
+ vfilter_16 v6, v7
+ vfilter_16 v7, v8
+
+compute_sum_sse_8x8_b:
+ vspltish v18, 0 ;# sum
+ vspltish v19, 0 ;# sse
+ vspltish v23, 0 ;# unpack
+ li r10, 16
+
+ vmrghb v0, v0, v1
+ vmrghb v1, v2, v3
+ vmrghb v2, v4, v5
+ vmrghb v3, v6, v7
+
+ load_and_align_16 v4, r7, r8, 1
+ load_and_align_16 v5, r7, r8, 1
+ load_and_align_16 v6, r7, r8, 1
+ load_and_align_16 v7, r7, r8, 1
+ load_and_align_16 v8, r7, r8, 1
+ load_and_align_16 v9, r7, r8, 1
+ load_and_align_16 v10, r7, r8, 1
+ load_and_align_16 v11, r7, r8, 0
+
+ vmrghb v4, v4, v5
+ vmrghb v5, v6, v7
+ vmrghb v6, v8, v9
+ vmrghb v7, v10, v11
+
+ compute_sum_sse v0, v4, v18, v19, v20, v21, v23
+ compute_sum_sse v1, v5, v18, v19, v20, v21, v23
+ compute_sum_sse v2, v6, v18, v19, v20, v21, v23
+ compute_sum_sse v3, v7, v18, v19, v20, v21, v23
+
+ variance_final v18, v19, v23, 6
+
+ addi r1, r1, 32 ;# recover stack
+ mtspr 256, r11 ;# reset old VRSAVE
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int src_pixels_per_line
+;# r5 int xoffset
+;# r6 int yoffset
+;# r7 unsigned char *dst_ptr
+;# r8 int dst_pixels_per_line
+;# r9 unsigned int *sse
+;#
+;# r3 return value
+vp8_sub_pixel_variance8x16_ppc:
+ mfspr r11, 256 ;# get old VRSAVE
+ oris r12, r11, 0xffff
+ ori r12, r12, 0xfffc
+ mtspr 256, r12 ;# set VRSAVE
+
+ stwu r1,-32(r1) ;# create space on the stack
+
+ HProlog second_pass_8x16_pre_copy_b
+
+ ;# Load up permutation constants
+ load_c v29, b_0123_b, 0, r12, r0
+ load_c v30, b_4567_b, 0, r12, r0
+
+ hfilter_8 v0, v29, v30, 1
+ hfilter_8 v1, v29, v30, 1
+ hfilter_8 v2, v29, v30, 1
+ hfilter_8 v3, v29, v30, 1
+ hfilter_8 v4, v29, v30, 1
+ hfilter_8 v5, v29, v30, 1
+ hfilter_8 v6, v29, v30, 1
+ hfilter_8 v7, v29, v30, 1
+ hfilter_8 v8, v29, v30, 1
+ hfilter_8 v9, v29, v30, 1
+ hfilter_8 v10, v29, v30, 1
+ hfilter_8 v11, v29, v30, 1
+ hfilter_8 v12, v29, v30, 1
+ hfilter_8 v13, v29, v30, 1
+ hfilter_8 v14, v29, v30, 1
+ hfilter_8 v15, v29, v30, 1
+
+ ;# Finished filtering main horizontal block. If there is no
+ ;# vertical filtering, jump to storing the data. Otherwise
+ ;# load up and filter the additional line that is needed
+ ;# for the vertical filter.
+ beq compute_sum_sse_8x16_b
+
+ hfilter_8 v16, v29, v30, 0
+
+ b second_pass_8x16_b
+
+second_pass_8x16_pre_copy_b:
+ slwi. r6, r6, 5 ;# index into vertical filter array
+
+ load_and_align_16 v0, r3, r4, 1
+ load_and_align_16 v1, r3, r4, 1
+ load_and_align_16 v2, r3, r4, 1
+ load_and_align_16 v3, r3, r4, 1
+ load_and_align_16 v4, r3, r4, 1
+ load_and_align_16 v5, r3, r4, 1
+ load_and_align_16 v6, r3, r4, 1
+ load_and_align_16 v7, r3, r4, 1
+ load_and_align_16 v8, r3, r4, 1
+ load_and_align_16 v9, r3, r4, 1
+ load_and_align_16 v10, r3, r4, 1
+ load_and_align_16 v11, r3, r4, 1
+ load_and_align_16 v12, r3, r4, 1
+ load_and_align_16 v13, r3, r4, 1
+ load_and_align_16 v14, r3, r4, 1
+ load_and_align_16 v15, r3, r4, 1
+ load_and_align_16 v16, r3, r4, 0
+
+ beq compute_sum_sse_8x16_b
+
+second_pass_8x16_b:
+ vspltish v20, 8
+ vspltish v18, 3
+ vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
+
+ load_vfilter v20, v21
+
+ vfilter_16 v0, v1
+ vfilter_16 v1, v2
+ vfilter_16 v2, v3
+ vfilter_16 v3, v4
+ vfilter_16 v4, v5
+ vfilter_16 v5, v6
+ vfilter_16 v6, v7
+ vfilter_16 v7, v8
+ vfilter_16 v8, v9
+ vfilter_16 v9, v10
+ vfilter_16 v10, v11
+ vfilter_16 v11, v12
+ vfilter_16 v12, v13
+ vfilter_16 v13, v14
+ vfilter_16 v14, v15
+ vfilter_16 v15, v16
+
+compute_sum_sse_8x16_b:
+ vspltish v18, 0 ;# sum
+ vspltish v19, 0 ;# sse
+ vspltish v23, 0 ;# unpack
+ li r10, 16
+
+ vmrghb v0, v0, v1
+ vmrghb v1, v2, v3
+ vmrghb v2, v4, v5
+ vmrghb v3, v6, v7
+ vmrghb v4, v8, v9
+ vmrghb v5, v10, v11
+ vmrghb v6, v12, v13
+ vmrghb v7, v14, v15
+
+ load_and_align_16 v8, r7, r8, 1
+ load_and_align_16 v9, r7, r8, 1
+ load_and_align_16 v10, r7, r8, 1
+ load_and_align_16 v11, r7, r8, 1
+ load_and_align_16 v12, r7, r8, 1
+ load_and_align_16 v13, r7, r8, 1
+ load_and_align_16 v14, r7, r8, 1
+ load_and_align_16 v15, r7, r8, 1
+
+ vmrghb v8, v8, v9
+ vmrghb v9, v10, v11
+ vmrghb v10, v12, v13
+ vmrghb v11, v14, v15
+
+ compute_sum_sse v0, v8, v18, v19, v20, v21, v23
+ compute_sum_sse v1, v9, v18, v19, v20, v21, v23
+ compute_sum_sse v2, v10, v18, v19, v20, v21, v23
+ compute_sum_sse v3, v11, v18, v19, v20, v21, v23
+
+ load_and_align_16 v8, r7, r8, 1
+ load_and_align_16 v9, r7, r8, 1
+ load_and_align_16 v10, r7, r8, 1
+ load_and_align_16 v11, r7, r8, 1
+ load_and_align_16 v12, r7, r8, 1
+ load_and_align_16 v13, r7, r8, 1
+ load_and_align_16 v14, r7, r8, 1
+ load_and_align_16 v15, r7, r8, 0
+
+ vmrghb v8, v8, v9
+ vmrghb v9, v10, v11
+ vmrghb v10, v12, v13
+ vmrghb v11, v14, v15
+
+ compute_sum_sse v4, v8, v18, v19, v20, v21, v23
+ compute_sum_sse v5, v9, v18, v19, v20, v21, v23
+ compute_sum_sse v6, v10, v18, v19, v20, v21, v23
+ compute_sum_sse v7, v11, v18, v19, v20, v21, v23
+
+ variance_final v18, v19, v23, 7
+
+ addi r1, r1, 32 ;# recover stack
+ mtspr 256, r11 ;# reset old VRSAVE
+ blr
+
+;# Filters a horizontal line
+;# expects:
+;# r3 src_ptr
+;# r4 pitch
+;# r10 16
+;# r12 32
+;# v17 perm intput
+;# v18 rounding
+;# v19 shift
+;# v20 filter taps
+;# v21 tmp
+;# v22 tmp
+;# v23 tmp
+;# v24 tmp
+;# v25 tmp
+;# v26 tmp
+;# v27 tmp
+;# v28 perm output
+;#
+.macro hfilter_16 V, increment_counter
+
+ lvsl v17, 0, r3 ;# permutate value for alignment
+
+ ;# input to filter is 21 bytes wide, output is 16 bytes.
+ ;# input will can span three vectors if not aligned correctly.
+ lvx v21, 0, r3
+ lvx v22, r10, r3
+ lvx v23, r12, r3
+
+.if \increment_counter
+ add r3, r3, r4
+.endif
+ vperm v21, v21, v22, v17
+ vperm v22, v22, v23, v17 ;# v8 v9 = 21 input pixels left-justified
+
+ ;# set 0
+ vmsummbm v24, v20, v21, v18 ;# taps times elements
+
+ ;# set 1
+ vsldoi v23, v21, v22, 1
+ vmsummbm v25, v20, v23, v18
+
+ ;# set 2
+ vsldoi v23, v21, v22, 2
+ vmsummbm v26, v20, v23, v18
+
+ ;# set 3
+ vsldoi v23, v21, v22, 3
+ vmsummbm v27, v20, v23, v18
+
+ vpkswus v24, v24, v25 ;# v24 = 0 4 8 C 1 5 9 D (16-bit)
+ vpkswus v25, v26, v27 ;# v25 = 2 6 A E 3 7 B F
+
+ vsrh v24, v24, v19 ;# divide v0, v1 by 128
+ vsrh v25, v25, v19
+
+ vpkuhus \V, v24, v25 ;# \V = scrambled 8-bit result
+ vperm \V, \V, v0, v28 ;# \V = correctly-ordered result
+.endm
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int src_pixels_per_line
+;# r5 int xoffset
+;# r6 int yoffset
+;# r7 unsigned char *dst_ptr
+;# r8 int dst_pixels_per_line
+;# r9 unsigned int *sse
+;#
+;# r3 return value
+vp8_sub_pixel_variance16x8_ppc:
+ mfspr r11, 256 ;# get old VRSAVE
+ oris r12, r11, 0xffff
+ ori r12, r12, 0xfff8
+ mtspr 256, r12 ;# set VRSAVE
+
+ stwu r1, -32(r1) ;# create space on the stack
+
+ HProlog second_pass_16x8_pre_copy_b
+
+ hfilter_16 v0, 1
+ hfilter_16 v1, 1
+ hfilter_16 v2, 1
+ hfilter_16 v3, 1
+ hfilter_16 v4, 1
+ hfilter_16 v5, 1
+ hfilter_16 v6, 1
+ hfilter_16 v7, 1
+
+ ;# Finished filtering main horizontal block. If there is no
+ ;# vertical filtering, jump to storing the data. Otherwise
+ ;# load up and filter the additional line that is needed
+ ;# for the vertical filter.
+ beq compute_sum_sse_16x8_b
+
+ hfilter_16 v8, 0
+
+ b second_pass_16x8_b
+
+second_pass_16x8_pre_copy_b:
+ slwi. r6, r6, 5 ;# index into vertical filter array
+
+ load_and_align_16 v0, r3, r4, 1
+ load_and_align_16 v1, r3, r4, 1
+ load_and_align_16 v2, r3, r4, 1
+ load_and_align_16 v3, r3, r4, 1
+ load_and_align_16 v4, r3, r4, 1
+ load_and_align_16 v5, r3, r4, 1
+ load_and_align_16 v6, r3, r4, 1
+ load_and_align_16 v7, r3, r4, 1
+ load_and_align_16 v8, r3, r4, 1
+
+ beq compute_sum_sse_16x8_b
+
+second_pass_16x8_b:
+ vspltish v20, 8
+ vspltish v18, 3
+ vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
+
+ load_vfilter v20, v21
+
+ vfilter_16 v0, v1
+ vfilter_16 v1, v2
+ vfilter_16 v2, v3
+ vfilter_16 v3, v4
+ vfilter_16 v4, v5
+ vfilter_16 v5, v6
+ vfilter_16 v6, v7
+ vfilter_16 v7, v8
+
+compute_sum_sse_16x8_b:
+ vspltish v18, 0 ;# sum
+ vspltish v19, 0 ;# sse
+ vspltish v23, 0 ;# unpack
+ li r10, 16
+
+ compute_sum_sse_16 v0, 1
+ compute_sum_sse_16 v1, 1
+ compute_sum_sse_16 v2, 1
+ compute_sum_sse_16 v3, 1
+ compute_sum_sse_16 v4, 1
+ compute_sum_sse_16 v5, 1
+ compute_sum_sse_16 v6, 1
+ compute_sum_sse_16 v7, 0
+
+ variance_final v18, v19, v23, 7
+
+ addi r1, r1, 32 ;# recover stack
+
+ mtspr 256, r11 ;# reset old VRSAVE
+
+ blr
+
+ .align 2
+;# r3 unsigned char *src_ptr
+;# r4 int src_pixels_per_line
+;# r5 int xoffset
+;# r6 int yoffset
+;# r7 unsigned char *dst_ptr
+;# r8 int dst_pixels_per_line
+;# r9 unsigned int *sse
+;#
+;# r3 return value
+vp8_sub_pixel_variance16x16_ppc:
+ mfspr r11, 256 ;# get old VRSAVE
+ oris r12, r11, 0xffff
+ ori r12, r12, 0xfff8
+ mtspr 256, r12 ;# set VRSAVE
+
+ stwu r1, -32(r1) ;# create space on the stack
+
+ HProlog second_pass_16x16_pre_copy_b
+
+ hfilter_16 v0, 1
+ hfilter_16 v1, 1
+ hfilter_16 v2, 1
+ hfilter_16 v3, 1
+ hfilter_16 v4, 1
+ hfilter_16 v5, 1
+ hfilter_16 v6, 1
+ hfilter_16 v7, 1
+ hfilter_16 v8, 1
+ hfilter_16 v9, 1
+ hfilter_16 v10, 1
+ hfilter_16 v11, 1
+ hfilter_16 v12, 1
+ hfilter_16 v13, 1
+ hfilter_16 v14, 1
+ hfilter_16 v15, 1
+
+ ;# Finished filtering main horizontal block. If there is no
+ ;# vertical filtering, jump to storing the data. Otherwise
+ ;# load up and filter the additional line that is needed
+ ;# for the vertical filter.
+ beq compute_sum_sse_16x16_b
+
+ hfilter_16 v16, 0
+
+ b second_pass_16x16_b
+
+second_pass_16x16_pre_copy_b:
+ slwi. r6, r6, 5 ;# index into vertical filter array
+
+ load_and_align_16 v0, r3, r4, 1
+ load_and_align_16 v1, r3, r4, 1
+ load_and_align_16 v2, r3, r4, 1
+ load_and_align_16 v3, r3, r4, 1
+ load_and_align_16 v4, r3, r4, 1
+ load_and_align_16 v5, r3, r4, 1
+ load_and_align_16 v6, r3, r4, 1
+ load_and_align_16 v7, r3, r4, 1
+ load_and_align_16 v8, r3, r4, 1
+ load_and_align_16 v9, r3, r4, 1
+ load_and_align_16 v10, r3, r4, 1
+ load_and_align_16 v11, r3, r4, 1
+ load_and_align_16 v12, r3, r4, 1
+ load_and_align_16 v13, r3, r4, 1
+ load_and_align_16 v14, r3, r4, 1
+ load_and_align_16 v15, r3, r4, 1
+ load_and_align_16 v16, r3, r4, 0
+
+ beq compute_sum_sse_16x16_b
+
+second_pass_16x16_b:
+ vspltish v20, 8
+ vspltish v18, 3
+ vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
+
+ load_vfilter v20, v21
+
+ vfilter_16 v0, v1
+ vfilter_16 v1, v2
+ vfilter_16 v2, v3
+ vfilter_16 v3, v4
+ vfilter_16 v4, v5
+ vfilter_16 v5, v6
+ vfilter_16 v6, v7
+ vfilter_16 v7, v8
+ vfilter_16 v8, v9
+ vfilter_16 v9, v10
+ vfilter_16 v10, v11
+ vfilter_16 v11, v12
+ vfilter_16 v12, v13
+ vfilter_16 v13, v14
+ vfilter_16 v14, v15
+ vfilter_16 v15, v16
+
+compute_sum_sse_16x16_b:
+ vspltish v18, 0 ;# sum
+ vspltish v19, 0 ;# sse
+ vspltish v23, 0 ;# unpack
+ li r10, 16
+
+ compute_sum_sse_16 v0, 1
+ compute_sum_sse_16 v1, 1
+ compute_sum_sse_16 v2, 1
+ compute_sum_sse_16 v3, 1
+ compute_sum_sse_16 v4, 1
+ compute_sum_sse_16 v5, 1
+ compute_sum_sse_16 v6, 1
+ compute_sum_sse_16 v7, 1
+ compute_sum_sse_16 v8, 1
+ compute_sum_sse_16 v9, 1
+ compute_sum_sse_16 v10, 1
+ compute_sum_sse_16 v11, 1
+ compute_sum_sse_16 v12, 1
+ compute_sum_sse_16 v13, 1
+ compute_sum_sse_16 v14, 1
+ compute_sum_sse_16 v15, 0
+
+ variance_final v18, v19, v23, 8
+
+ addi r1, r1, 32 ;# recover stack
+
+ mtspr 256, r11 ;# reset old VRSAVE
+
+ blr
+
+ .data
+
+ .align 4
+hfilter_b:
+ .byte 128, 0, 0, 0,128, 0, 0, 0,128, 0, 0, 0,128, 0, 0, 0
+ .byte 112, 16, 0, 0,112, 16, 0, 0,112, 16, 0, 0,112, 16, 0, 0
+ .byte 96, 32, 0, 0, 96, 32, 0, 0, 96, 32, 0, 0, 96, 32, 0, 0
+ .byte 80, 48, 0, 0, 80, 48, 0, 0, 80, 48, 0, 0, 80, 48, 0, 0
+ .byte 64, 64, 0, 0, 64, 64, 0, 0, 64, 64, 0, 0, 64, 64, 0, 0
+ .byte 48, 80, 0, 0, 48, 80, 0, 0, 48, 80, 0, 0, 48, 80, 0, 0
+ .byte 32, 96, 0, 0, 32, 96, 0, 0, 32, 96, 0, 0, 32, 96, 0, 0
+ .byte 16,112, 0, 0, 16,112, 0, 0, 16,112, 0, 0, 16,112, 0, 0
+
+ .align 4
+vfilter_b:
+ .byte 128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128
+ .byte 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+ .byte 112,112,112,112,112,112,112,112,112,112,112,112,112,112,112,112
+ .byte 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16
+ .byte 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96
+ .byte 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32
+ .byte 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80
+ .byte 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48
+ .byte 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
+ .byte 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
+ .byte 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48
+ .byte 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80
+ .byte 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32
+ .byte 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96
+ .byte 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16
+ .byte 112,112,112,112,112,112,112,112,112,112,112,112,112,112,112,112
+
+ .align 4
+b_hperm_b:
+ .byte 0, 4, 8, 12, 1, 5, 9, 13, 2, 6, 10, 14, 3, 7, 11, 15
+
+ .align 4
+b_0123_b:
+ .byte 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6
+
+ .align 4
+b_4567_b:
+ .byte 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10
+
+b_hilo_b:
+ .byte 0, 1, 2, 3, 4, 5, 6, 7, 16, 17, 18, 19, 20, 21, 22, 23
diff --git a/libvpx/vp8/common/ppflags.h b/libvpx/vp8/common/ppflags.h
new file mode 100644
index 0000000..665e21f
--- /dev/null
+++ b/libvpx/vp8/common/ppflags.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_PPFLAGS_H
+#define __INC_PPFLAGS_H
+enum
+{
+ VP8D_NOFILTERING = 0,
+ VP8D_DEBLOCK = 1<<0,
+ VP8D_DEMACROBLOCK = 1<<1,
+ VP8D_ADDNOISE = 1<<2,
+ VP8D_DEBUG_TXT_FRAME_INFO = 1<<3,
+ VP8D_DEBUG_TXT_MBLK_MODES = 1<<4,
+ VP8D_DEBUG_TXT_DC_DIFF = 1<<5,
+ VP8D_DEBUG_TXT_RATE_INFO = 1<<6,
+ VP8D_DEBUG_DRAW_MV = 1<<7,
+ VP8D_DEBUG_CLR_BLK_MODES = 1<<8,
+ VP8D_DEBUG_CLR_FRM_REF_BLKS = 1<<9,
+ VP8D_MFQE = 1<<10
+};
+
+typedef struct
+{
+ int post_proc_flag;
+ int deblocking_level;
+ int noise_level;
+ int display_ref_frame_flag;
+ int display_mb_modes_flag;
+ int display_b_modes_flag;
+ int display_mv_flag;
+} vp8_ppflags_t;
+
+#endif
diff --git a/vp8/common/pragmas.h b/libvpx/vp8/common/pragmas.h
similarity index 100%
rename from vp8/common/pragmas.h
rename to libvpx/vp8/common/pragmas.h
diff --git a/libvpx/vp8/common/quant_common.c b/libvpx/vp8/common/quant_common.c
new file mode 100644
index 0000000..05f9210
--- /dev/null
+++ b/libvpx/vp8/common/quant_common.c
@@ -0,0 +1,135 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "quant_common.h"
+
+static const int dc_qlookup[QINDEX_RANGE] =
+{
+ 4, 5, 6, 7, 8, 9, 10, 10, 11, 12, 13, 14, 15, 16, 17, 17,
+ 18, 19, 20, 20, 21, 21, 22, 22, 23, 23, 24, 25, 25, 26, 27, 28,
+ 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43,
+ 44, 45, 46, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58,
+ 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74,
+ 75, 76, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89,
+ 91, 93, 95, 96, 98, 100, 101, 102, 104, 106, 108, 110, 112, 114, 116, 118,
+ 122, 124, 126, 128, 130, 132, 134, 136, 138, 140, 143, 145, 148, 151, 154, 157,
+};
+
+static const int ac_qlookup[QINDEX_RANGE] =
+{
+ 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
+ 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,
+ 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,
+ 52, 53, 54, 55, 56, 57, 58, 60, 62, 64, 66, 68, 70, 72, 74, 76,
+ 78, 80, 82, 84, 86, 88, 90, 92, 94, 96, 98, 100, 102, 104, 106, 108,
+ 110, 112, 114, 116, 119, 122, 125, 128, 131, 134, 137, 140, 143, 146, 149, 152,
+ 155, 158, 161, 164, 167, 170, 173, 177, 181, 185, 189, 193, 197, 201, 205, 209,
+ 213, 217, 221, 225, 229, 234, 239, 245, 249, 254, 259, 264, 269, 274, 279, 284,
+};
+
+
+int vp8_dc_quant(int QIndex, int Delta)
+{
+ int retval;
+
+ QIndex = QIndex + Delta;
+
+ if (QIndex > 127)
+ QIndex = 127;
+ else if (QIndex < 0)
+ QIndex = 0;
+
+ retval = dc_qlookup[ QIndex ];
+ return retval;
+}
+
+int vp8_dc2quant(int QIndex, int Delta)
+{
+ int retval;
+
+ QIndex = QIndex + Delta;
+
+ if (QIndex > 127)
+ QIndex = 127;
+ else if (QIndex < 0)
+ QIndex = 0;
+
+ retval = dc_qlookup[ QIndex ] * 2;
+ return retval;
+
+}
+int vp8_dc_uv_quant(int QIndex, int Delta)
+{
+ int retval;
+
+ QIndex = QIndex + Delta;
+
+ if (QIndex > 127)
+ QIndex = 127;
+ else if (QIndex < 0)
+ QIndex = 0;
+
+ retval = dc_qlookup[ QIndex ];
+
+ if (retval > 132)
+ retval = 132;
+
+ return retval;
+}
+
+int vp8_ac_yquant(int QIndex)
+{
+ int retval;
+
+ if (QIndex > 127)
+ QIndex = 127;
+ else if (QIndex < 0)
+ QIndex = 0;
+
+ retval = ac_qlookup[ QIndex ];
+ return retval;
+}
+
+int vp8_ac2quant(int QIndex, int Delta)
+{
+ int retval;
+
+ QIndex = QIndex + Delta;
+
+ if (QIndex > 127)
+ QIndex = 127;
+ else if (QIndex < 0)
+ QIndex = 0;
+
+ /* For all x in [0..284], x*155/100 is bitwise equal to (x*101581) >> 16.
+ * The smallest precision for that is '(x*6349) >> 12' but 16 is a good
+ * word size. */
+ retval = (ac_qlookup[ QIndex ] * 101581) >> 16;
+
+ if (retval < 8)
+ retval = 8;
+
+ return retval;
+}
+int vp8_ac_uv_quant(int QIndex, int Delta)
+{
+ int retval;
+
+ QIndex = QIndex + Delta;
+
+ if (QIndex > 127)
+ QIndex = 127;
+ else if (QIndex < 0)
+ QIndex = 0;
+
+ retval = ac_qlookup[ QIndex ];
+ return retval;
+}
diff --git a/vp8/common/quant_common.h b/libvpx/vp8/common/quant_common.h
similarity index 100%
rename from vp8/common/quant_common.h
rename to libvpx/vp8/common/quant_common.h
diff --git a/libvpx/vp8/common/reconinter.c b/libvpx/vp8/common/reconinter.c
new file mode 100644
index 0000000..3da3bc7
--- /dev/null
+++ b/libvpx/vp8/common/reconinter.c
@@ -0,0 +1,595 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <limits.h>
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx/vpx_integer.h"
+#include "blockd.h"
+#include "reconinter.h"
+#if CONFIG_RUNTIME_CPU_DETECT
+#include "onyxc_int.h"
+#endif
+
+void vp8_copy_mem16x16_c(
+ unsigned char *src,
+ int src_stride,
+ unsigned char *dst,
+ int dst_stride)
+{
+
+ int r;
+
+ for (r = 0; r < 16; r++)
+ {
+#if !(CONFIG_FAST_UNALIGNED)
+ dst[0] = src[0];
+ dst[1] = src[1];
+ dst[2] = src[2];
+ dst[3] = src[3];
+ dst[4] = src[4];
+ dst[5] = src[5];
+ dst[6] = src[6];
+ dst[7] = src[7];
+ dst[8] = src[8];
+ dst[9] = src[9];
+ dst[10] = src[10];
+ dst[11] = src[11];
+ dst[12] = src[12];
+ dst[13] = src[13];
+ dst[14] = src[14];
+ dst[15] = src[15];
+
+#else
+ ((uint32_t *)dst)[0] = ((uint32_t *)src)[0] ;
+ ((uint32_t *)dst)[1] = ((uint32_t *)src)[1] ;
+ ((uint32_t *)dst)[2] = ((uint32_t *)src)[2] ;
+ ((uint32_t *)dst)[3] = ((uint32_t *)src)[3] ;
+
+#endif
+ src += src_stride;
+ dst += dst_stride;
+
+ }
+
+}
+
+void vp8_copy_mem8x8_c(
+ unsigned char *src,
+ int src_stride,
+ unsigned char *dst,
+ int dst_stride)
+{
+ int r;
+
+ for (r = 0; r < 8; r++)
+ {
+#if !(CONFIG_FAST_UNALIGNED)
+ dst[0] = src[0];
+ dst[1] = src[1];
+ dst[2] = src[2];
+ dst[3] = src[3];
+ dst[4] = src[4];
+ dst[5] = src[5];
+ dst[6] = src[6];
+ dst[7] = src[7];
+#else
+ ((uint32_t *)dst)[0] = ((uint32_t *)src)[0] ;
+ ((uint32_t *)dst)[1] = ((uint32_t *)src)[1] ;
+#endif
+ src += src_stride;
+ dst += dst_stride;
+
+ }
+
+}
+
+void vp8_copy_mem8x4_c(
+ unsigned char *src,
+ int src_stride,
+ unsigned char *dst,
+ int dst_stride)
+{
+ int r;
+
+ for (r = 0; r < 4; r++)
+ {
+#if !(CONFIG_FAST_UNALIGNED)
+ dst[0] = src[0];
+ dst[1] = src[1];
+ dst[2] = src[2];
+ dst[3] = src[3];
+ dst[4] = src[4];
+ dst[5] = src[5];
+ dst[6] = src[6];
+ dst[7] = src[7];
+#else
+ ((uint32_t *)dst)[0] = ((uint32_t *)src)[0] ;
+ ((uint32_t *)dst)[1] = ((uint32_t *)src)[1] ;
+#endif
+ src += src_stride;
+ dst += dst_stride;
+
+ }
+
+}
+
+
+void vp8_build_inter_predictors_b(BLOCKD *d, int pitch, unsigned char *base_pre, int pre_stride, vp8_subpix_fn_t sppf)
+{
+ int r;
+ unsigned char *pred_ptr = d->predictor;
+ unsigned char *ptr;
+ ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride + (d->bmi.mv.as_mv.col >> 3);
+
+ if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
+ {
+ sppf(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, pred_ptr, pitch);
+ }
+ else
+ {
+ for (r = 0; r < 4; r++)
+ {
+#if !(CONFIG_FAST_UNALIGNED)
+ pred_ptr[0] = ptr[0];
+ pred_ptr[1] = ptr[1];
+ pred_ptr[2] = ptr[2];
+ pred_ptr[3] = ptr[3];
+#else
+ *(uint32_t *)pred_ptr = *(uint32_t *)ptr ;
+#endif
+ pred_ptr += pitch;
+ ptr += pre_stride;
+ }
+ }
+}
+
+static void build_inter_predictors4b(MACROBLOCKD *x, BLOCKD *d, unsigned char *dst, int dst_stride, unsigned char *base_pre, int pre_stride)
+{
+ unsigned char *ptr;
+ ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride + (d->bmi.mv.as_mv.col >> 3);
+
+ if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
+ {
+ x->subpixel_predict8x8(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst, dst_stride);
+ }
+ else
+ {
+ vp8_copy_mem8x8(ptr, pre_stride, dst, dst_stride);
+ }
+}
+
+static void build_inter_predictors2b(MACROBLOCKD *x, BLOCKD *d, unsigned char *dst, int dst_stride, unsigned char *base_pre, int pre_stride)
+{
+ unsigned char *ptr;
+ ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride + (d->bmi.mv.as_mv.col >> 3);
+
+ if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
+ {
+ x->subpixel_predict8x4(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst, dst_stride);
+ }
+ else
+ {
+ vp8_copy_mem8x4(ptr, pre_stride, dst, dst_stride);
+ }
+}
+
+static void build_inter_predictors_b(BLOCKD *d, unsigned char *dst, int dst_stride, unsigned char *base_pre, int pre_stride, vp8_subpix_fn_t sppf)
+{
+ int r;
+ unsigned char *ptr;
+ ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride + (d->bmi.mv.as_mv.col >> 3);
+
+ if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
+ {
+ sppf(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst, dst_stride);
+ }
+ else
+ {
+ for (r = 0; r < 4; r++)
+ {
+#if !(CONFIG_FAST_UNALIGNED)
+ dst[0] = ptr[0];
+ dst[1] = ptr[1];
+ dst[2] = ptr[2];
+ dst[3] = ptr[3];
+#else
+ *(uint32_t *)dst = *(uint32_t *)ptr ;
+#endif
+ dst += dst_stride;
+ ptr += pre_stride;
+ }
+ }
+}
+
+
+/*encoder only*/
+void vp8_build_inter16x16_predictors_mbuv(MACROBLOCKD *x)
+{
+ unsigned char *uptr, *vptr;
+ unsigned char *upred_ptr = &x->predictor[256];
+ unsigned char *vpred_ptr = &x->predictor[320];
+
+ int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
+ int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
+ int offset;
+ int pre_stride = x->pre.uv_stride;
+
+ /* calc uv motion vectors */
+ mv_row += 1 | (mv_row >> (sizeof(int) * CHAR_BIT - 1));
+ mv_col += 1 | (mv_col >> (sizeof(int) * CHAR_BIT - 1));
+ mv_row /= 2;
+ mv_col /= 2;
+ mv_row &= x->fullpixel_mask;
+ mv_col &= x->fullpixel_mask;
+
+ offset = (mv_row >> 3) * pre_stride + (mv_col >> 3);
+ uptr = x->pre.u_buffer + offset;
+ vptr = x->pre.v_buffer + offset;
+
+ if ((mv_row | mv_col) & 7)
+ {
+ x->subpixel_predict8x8(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr, 8);
+ x->subpixel_predict8x8(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr, 8);
+ }
+ else
+ {
+ vp8_copy_mem8x8(uptr, pre_stride, upred_ptr, 8);
+ vp8_copy_mem8x8(vptr, pre_stride, vpred_ptr, 8);
+ }
+}
+
+/*encoder only*/
+void vp8_build_inter4x4_predictors_mbuv(MACROBLOCKD *x)
+{
+ int i, j;
+ int pre_stride = x->pre.uv_stride;
+ unsigned char *base_pre;
+
+ /* build uv mvs */
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 2; j++)
+ {
+ int yoffset = i * 8 + j * 2;
+ int uoffset = 16 + i * 2 + j;
+ int voffset = 20 + i * 2 + j;
+
+ int temp;
+
+ temp = x->block[yoffset ].bmi.mv.as_mv.row
+ + x->block[yoffset+1].bmi.mv.as_mv.row
+ + x->block[yoffset+4].bmi.mv.as_mv.row
+ + x->block[yoffset+5].bmi.mv.as_mv.row;
+
+ temp += 4 + ((temp >> (sizeof(int) * CHAR_BIT - 1)) << 3);
+
+ x->block[uoffset].bmi.mv.as_mv.row = (temp / 8) & x->fullpixel_mask;
+
+ temp = x->block[yoffset ].bmi.mv.as_mv.col
+ + x->block[yoffset+1].bmi.mv.as_mv.col
+ + x->block[yoffset+4].bmi.mv.as_mv.col
+ + x->block[yoffset+5].bmi.mv.as_mv.col;
+
+ temp += 4 + ((temp >> (sizeof(int) * CHAR_BIT - 1)) << 3);
+
+ x->block[uoffset].bmi.mv.as_mv.col = (temp / 8) & x->fullpixel_mask;
+
+ x->block[voffset].bmi.mv.as_int = x->block[uoffset].bmi.mv.as_int;
+ }
+ }
+
+ base_pre = x->pre.u_buffer;
+ for (i = 16; i < 20; i += 2)
+ {
+ BLOCKD *d0 = &x->block[i];
+ BLOCKD *d1 = &x->block[i+1];
+
+ if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
+ build_inter_predictors2b(x, d0, d0->predictor, 8, base_pre, pre_stride);
+ else
+ {
+ vp8_build_inter_predictors_b(d0, 8, base_pre, pre_stride, x->subpixel_predict);
+ vp8_build_inter_predictors_b(d1, 8, base_pre, pre_stride, x->subpixel_predict);
+ }
+ }
+
+ base_pre = x->pre.v_buffer;
+ for (i = 20; i < 24; i += 2)
+ {
+ BLOCKD *d0 = &x->block[i];
+ BLOCKD *d1 = &x->block[i+1];
+
+ if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
+ build_inter_predictors2b(x, d0, d0->predictor, 8, base_pre, pre_stride);
+ else
+ {
+ vp8_build_inter_predictors_b(d0, 8, base_pre, pre_stride, x->subpixel_predict);
+ vp8_build_inter_predictors_b(d1, 8, base_pre, pre_stride, x->subpixel_predict);
+ }
+ }
+}
+
+
+/*encoder only*/
+void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x,
+ unsigned char *dst_y,
+ int dst_ystride)
+{
+ unsigned char *ptr_base;
+ unsigned char *ptr;
+ int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
+ int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
+ int pre_stride = x->pre.y_stride;
+
+ ptr_base = x->pre.y_buffer;
+ ptr = ptr_base + (mv_row >> 3) * pre_stride + (mv_col >> 3);
+
+ if ((mv_row | mv_col) & 7)
+ {
+ x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7,
+ dst_y, dst_ystride);
+ }
+ else
+ {
+ vp8_copy_mem16x16(ptr, pre_stride, dst_y,
+ dst_ystride);
+ }
+}
+
+static void clamp_mv_to_umv_border(MV *mv, const MACROBLOCKD *xd)
+{
+ /* If the MV points so far into the UMV border that no visible pixels
+ * are used for reconstruction, the subpel part of the MV can be
+ * discarded and the MV limited to 16 pixels with equivalent results.
+ *
+ * This limit kicks in at 19 pixels for the top and left edges, for
+ * the 16 pixels plus 3 taps right of the central pixel when subpel
+ * filtering. The bottom and right edges use 16 pixels plus 2 pixels
+ * left of the central pixel when filtering.
+ */
+ if (mv->col < (xd->mb_to_left_edge - (19 << 3)))
+ mv->col = xd->mb_to_left_edge - (16 << 3);
+ else if (mv->col > xd->mb_to_right_edge + (18 << 3))
+ mv->col = xd->mb_to_right_edge + (16 << 3);
+
+ if (mv->row < (xd->mb_to_top_edge - (19 << 3)))
+ mv->row = xd->mb_to_top_edge - (16 << 3);
+ else if (mv->row > xd->mb_to_bottom_edge + (18 << 3))
+ mv->row = xd->mb_to_bottom_edge + (16 << 3);
+}
+
+/* A version of the above function for chroma block MVs.*/
+static void clamp_uvmv_to_umv_border(MV *mv, const MACROBLOCKD *xd)
+{
+ mv->col = (2*mv->col < (xd->mb_to_left_edge - (19 << 3))) ?
+ (xd->mb_to_left_edge - (16 << 3)) >> 1 : mv->col;
+ mv->col = (2*mv->col > xd->mb_to_right_edge + (18 << 3)) ?
+ (xd->mb_to_right_edge + (16 << 3)) >> 1 : mv->col;
+
+ mv->row = (2*mv->row < (xd->mb_to_top_edge - (19 << 3))) ?
+ (xd->mb_to_top_edge - (16 << 3)) >> 1 : mv->row;
+ mv->row = (2*mv->row > xd->mb_to_bottom_edge + (18 << 3)) ?
+ (xd->mb_to_bottom_edge + (16 << 3)) >> 1 : mv->row;
+}
+
+void vp8_build_inter16x16_predictors_mb(MACROBLOCKD *x,
+ unsigned char *dst_y,
+ unsigned char *dst_u,
+ unsigned char *dst_v,
+ int dst_ystride,
+ int dst_uvstride)
+{
+ int offset;
+ unsigned char *ptr;
+ unsigned char *uptr, *vptr;
+
+ int_mv _16x16mv;
+
+ unsigned char *ptr_base = x->pre.y_buffer;
+ int pre_stride = x->pre.y_stride;
+
+ _16x16mv.as_int = x->mode_info_context->mbmi.mv.as_int;
+
+ if (x->mode_info_context->mbmi.need_to_clamp_mvs)
+ {
+ clamp_mv_to_umv_border(&_16x16mv.as_mv, x);
+ }
+
+ ptr = ptr_base + ( _16x16mv.as_mv.row >> 3) * pre_stride + (_16x16mv.as_mv.col >> 3);
+
+ if ( _16x16mv.as_int & 0x00070007)
+ {
+ x->subpixel_predict16x16(ptr, pre_stride, _16x16mv.as_mv.col & 7, _16x16mv.as_mv.row & 7, dst_y, dst_ystride);
+ }
+ else
+ {
+ vp8_copy_mem16x16(ptr, pre_stride, dst_y, dst_ystride);
+ }
+
+ /* calc uv motion vectors */
+ _16x16mv.as_mv.row += 1 | (_16x16mv.as_mv.row >> (sizeof(int) * CHAR_BIT - 1));
+ _16x16mv.as_mv.col += 1 | (_16x16mv.as_mv.col >> (sizeof(int) * CHAR_BIT - 1));
+ _16x16mv.as_mv.row /= 2;
+ _16x16mv.as_mv.col /= 2;
+ _16x16mv.as_mv.row &= x->fullpixel_mask;
+ _16x16mv.as_mv.col &= x->fullpixel_mask;
+
+ pre_stride >>= 1;
+ offset = ( _16x16mv.as_mv.row >> 3) * pre_stride + (_16x16mv.as_mv.col >> 3);
+ uptr = x->pre.u_buffer + offset;
+ vptr = x->pre.v_buffer + offset;
+
+ if ( _16x16mv.as_int & 0x00070007)
+ {
+ x->subpixel_predict8x8(uptr, pre_stride, _16x16mv.as_mv.col & 7, _16x16mv.as_mv.row & 7, dst_u, dst_uvstride);
+ x->subpixel_predict8x8(vptr, pre_stride, _16x16mv.as_mv.col & 7, _16x16mv.as_mv.row & 7, dst_v, dst_uvstride);
+ }
+ else
+ {
+ vp8_copy_mem8x8(uptr, pre_stride, dst_u, dst_uvstride);
+ vp8_copy_mem8x8(vptr, pre_stride, dst_v, dst_uvstride);
+ }
+}
+
+static void build_inter4x4_predictors_mb(MACROBLOCKD *x)
+{
+ int i;
+ unsigned char *base_dst = x->dst.y_buffer;
+ unsigned char *base_pre = x->pre.y_buffer;
+
+ if (x->mode_info_context->mbmi.partitioning < 3)
+ {
+ BLOCKD *b;
+ int dst_stride = x->dst.y_stride;
+
+ x->block[ 0].bmi = x->mode_info_context->bmi[ 0];
+ x->block[ 2].bmi = x->mode_info_context->bmi[ 2];
+ x->block[ 8].bmi = x->mode_info_context->bmi[ 8];
+ x->block[10].bmi = x->mode_info_context->bmi[10];
+ if (x->mode_info_context->mbmi.need_to_clamp_mvs)
+ {
+ clamp_mv_to_umv_border(&x->block[ 0].bmi.mv.as_mv, x);
+ clamp_mv_to_umv_border(&x->block[ 2].bmi.mv.as_mv, x);
+ clamp_mv_to_umv_border(&x->block[ 8].bmi.mv.as_mv, x);
+ clamp_mv_to_umv_border(&x->block[10].bmi.mv.as_mv, x);
+ }
+
+ b = &x->block[ 0];
+ build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre, dst_stride);
+ b = &x->block[ 2];
+ build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre, dst_stride);
+ b = &x->block[ 8];
+ build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre, dst_stride);
+ b = &x->block[10];
+ build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre, dst_stride);
+ }
+ else
+ {
+ for (i = 0; i < 16; i += 2)
+ {
+ BLOCKD *d0 = &x->block[i];
+ BLOCKD *d1 = &x->block[i+1];
+ int dst_stride = x->dst.y_stride;
+
+ x->block[i+0].bmi = x->mode_info_context->bmi[i+0];
+ x->block[i+1].bmi = x->mode_info_context->bmi[i+1];
+ if (x->mode_info_context->mbmi.need_to_clamp_mvs)
+ {
+ clamp_mv_to_umv_border(&x->block[i+0].bmi.mv.as_mv, x);
+ clamp_mv_to_umv_border(&x->block[i+1].bmi.mv.as_mv, x);
+ }
+
+ if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
+ build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride);
+ else
+ {
+ build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
+ build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
+ }
+
+ }
+
+ }
+ base_dst = x->dst.u_buffer;
+ base_pre = x->pre.u_buffer;
+ for (i = 16; i < 20; i += 2)
+ {
+ BLOCKD *d0 = &x->block[i];
+ BLOCKD *d1 = &x->block[i+1];
+ int dst_stride = x->dst.uv_stride;
+
+ /* Note: uv mvs already clamped in build_4x4uvmvs() */
+
+ if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
+ build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride);
+ else
+ {
+ build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
+ build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
+ }
+ }
+
+ base_dst = x->dst.v_buffer;
+ base_pre = x->pre.v_buffer;
+ for (i = 20; i < 24; i += 2)
+ {
+ BLOCKD *d0 = &x->block[i];
+ BLOCKD *d1 = &x->block[i+1];
+ int dst_stride = x->dst.uv_stride;
+
+ /* Note: uv mvs already clamped in build_4x4uvmvs() */
+
+ if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
+ build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride);
+ else
+ {
+ build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
+ build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
+ }
+ }
+}
+
+static
+void build_4x4uvmvs(MACROBLOCKD *x)
+{
+ int i, j;
+
+ for (i = 0; i < 2; i++)
+ {
+ for (j = 0; j < 2; j++)
+ {
+ int yoffset = i * 8 + j * 2;
+ int uoffset = 16 + i * 2 + j;
+ int voffset = 20 + i * 2 + j;
+
+ int temp;
+
+ temp = x->mode_info_context->bmi[yoffset + 0].mv.as_mv.row
+ + x->mode_info_context->bmi[yoffset + 1].mv.as_mv.row
+ + x->mode_info_context->bmi[yoffset + 4].mv.as_mv.row
+ + x->mode_info_context->bmi[yoffset + 5].mv.as_mv.row;
+
+ temp += 4 + ((temp >> (sizeof(int) * CHAR_BIT - 1)) << 3);
+
+ x->block[uoffset].bmi.mv.as_mv.row = (temp / 8) & x->fullpixel_mask;
+
+ temp = x->mode_info_context->bmi[yoffset + 0].mv.as_mv.col
+ + x->mode_info_context->bmi[yoffset + 1].mv.as_mv.col
+ + x->mode_info_context->bmi[yoffset + 4].mv.as_mv.col
+ + x->mode_info_context->bmi[yoffset + 5].mv.as_mv.col;
+
+ temp += 4 + ((temp >> (sizeof(int) * CHAR_BIT - 1)) << 3);
+
+ x->block[uoffset].bmi.mv.as_mv.col = (temp / 8) & x->fullpixel_mask;
+
+ if (x->mode_info_context->mbmi.need_to_clamp_mvs)
+ clamp_uvmv_to_umv_border(&x->block[uoffset].bmi.mv.as_mv, x);
+
+ x->block[voffset].bmi.mv.as_int = x->block[uoffset].bmi.mv.as_int;
+ }
+ }
+}
+
+void vp8_build_inter_predictors_mb(MACROBLOCKD *xd)
+{
+ if (xd->mode_info_context->mbmi.mode != SPLITMV)
+ {
+ vp8_build_inter16x16_predictors_mb(xd, xd->dst.y_buffer,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.y_stride, xd->dst.uv_stride);
+ }
+ else
+ {
+ build_4x4uvmvs(xd);
+ build_inter4x4_predictors_mb(xd);
+ }
+}
diff --git a/libvpx/vp8/common/reconinter.h b/libvpx/vp8/common/reconinter.h
new file mode 100644
index 0000000..233c02e
--- /dev/null
+++ b/libvpx/vp8/common/reconinter.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_RECONINTER_H
+#define __INC_RECONINTER_H
+
+extern void vp8_build_inter_predictors_mb(MACROBLOCKD *x);
+extern void vp8_build_inter16x16_predictors_mb(MACROBLOCKD *x,
+ unsigned char *dst_y,
+ unsigned char *dst_u,
+ unsigned char *dst_v,
+ int dst_ystride,
+ int dst_uvstride);
+
+
+extern void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x,
+ unsigned char *dst_y,
+ int dst_ystride);
+extern void vp8_build_inter_predictors_b(BLOCKD *d, int pitch,
+ unsigned char *base_pre,
+ int pre_stride,
+ vp8_subpix_fn_t sppf);
+
+extern void vp8_build_inter16x16_predictors_mbuv(MACROBLOCKD *x);
+extern void vp8_build_inter4x4_predictors_mbuv(MACROBLOCKD *x);
+
+#endif
diff --git a/libvpx/vp8/common/reconintra.c b/libvpx/vp8/common/reconintra.c
new file mode 100644
index 0000000..4067a68
--- /dev/null
+++ b/libvpx/vp8/common/reconintra.c
@@ -0,0 +1,288 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx_mem/vpx_mem.h"
+#include "blockd.h"
+
+void vp8_build_intra_predictors_mby_s_c(MACROBLOCKD *x,
+ unsigned char * yabove_row,
+ unsigned char * yleft,
+ int left_stride,
+ unsigned char * ypred_ptr,
+ int y_stride)
+{
+ unsigned char yleft_col[16];
+ unsigned char ytop_left = yabove_row[-1];
+ int r, c, i;
+
+ for (i = 0; i < 16; i++)
+ {
+ yleft_col[i] = yleft[i* left_stride];
+ }
+
+ /* for Y */
+ switch (x->mode_info_context->mbmi.mode)
+ {
+ case DC_PRED:
+ {
+ int expected_dc;
+ int i;
+ int shift;
+ int average = 0;
+
+
+ if (x->up_available || x->left_available)
+ {
+ if (x->up_available)
+ {
+ for (i = 0; i < 16; i++)
+ {
+ average += yabove_row[i];
+ }
+ }
+
+ if (x->left_available)
+ {
+
+ for (i = 0; i < 16; i++)
+ {
+ average += yleft_col[i];
+ }
+
+ }
+
+
+
+ shift = 3 + x->up_available + x->left_available;
+ expected_dc = (average + (1 << (shift - 1))) >> shift;
+ }
+ else
+ {
+ expected_dc = 128;
+ }
+
+ /*vpx_memset(ypred_ptr, expected_dc, 256);*/
+ for (r = 0; r < 16; r++)
+ {
+ vpx_memset(ypred_ptr, expected_dc, 16);
+ ypred_ptr += y_stride;
+ }
+ }
+ break;
+ case V_PRED:
+ {
+
+ for (r = 0; r < 16; r++)
+ {
+
+ ((int *)ypred_ptr)[0] = ((int *)yabove_row)[0];
+ ((int *)ypred_ptr)[1] = ((int *)yabove_row)[1];
+ ((int *)ypred_ptr)[2] = ((int *)yabove_row)[2];
+ ((int *)ypred_ptr)[3] = ((int *)yabove_row)[3];
+ ypred_ptr += y_stride;
+ }
+ }
+ break;
+ case H_PRED:
+ {
+
+ for (r = 0; r < 16; r++)
+ {
+
+ vpx_memset(ypred_ptr, yleft_col[r], 16);
+ ypred_ptr += y_stride;
+ }
+
+ }
+ break;
+ case TM_PRED:
+ {
+
+ for (r = 0; r < 16; r++)
+ {
+ for (c = 0; c < 16; c++)
+ {
+ int pred = yleft_col[r] + yabove_row[ c] - ytop_left;
+
+ if (pred < 0)
+ pred = 0;
+
+ if (pred > 255)
+ pred = 255;
+
+ ypred_ptr[c] = pred;
+ }
+
+ ypred_ptr += y_stride;
+ }
+
+ }
+ break;
+ case B_PRED:
+ case NEARESTMV:
+ case NEARMV:
+ case ZEROMV:
+ case NEWMV:
+ case SPLITMV:
+ case MB_MODE_COUNT:
+ break;
+ }
+}
+
+void vp8_build_intra_predictors_mbuv_s_c(MACROBLOCKD *x,
+ unsigned char * uabove_row,
+ unsigned char * vabove_row,
+ unsigned char * uleft,
+ unsigned char * vleft,
+ int left_stride,
+ unsigned char * upred_ptr,
+ unsigned char * vpred_ptr,
+ int pred_stride)
+{
+ unsigned char uleft_col[8];
+ unsigned char utop_left = uabove_row[-1];
+ unsigned char vleft_col[8];
+ unsigned char vtop_left = vabove_row[-1];
+
+ int i, j;
+
+ for (i = 0; i < 8; i++)
+ {
+ uleft_col[i] = uleft [i* left_stride];
+ vleft_col[i] = vleft [i* left_stride];
+ }
+
+ switch (x->mode_info_context->mbmi.uv_mode)
+ {
+ case DC_PRED:
+ {
+ int expected_udc;
+ int expected_vdc;
+ int i;
+ int shift;
+ int Uaverage = 0;
+ int Vaverage = 0;
+
+ if (x->up_available)
+ {
+ for (i = 0; i < 8; i++)
+ {
+ Uaverage += uabove_row[i];
+ Vaverage += vabove_row[i];
+ }
+ }
+
+ if (x->left_available)
+ {
+ for (i = 0; i < 8; i++)
+ {
+ Uaverage += uleft_col[i];
+ Vaverage += vleft_col[i];
+ }
+ }
+
+ if (!x->up_available && !x->left_available)
+ {
+ expected_udc = 128;
+ expected_vdc = 128;
+ }
+ else
+ {
+ shift = 2 + x->up_available + x->left_available;
+ expected_udc = (Uaverage + (1 << (shift - 1))) >> shift;
+ expected_vdc = (Vaverage + (1 << (shift - 1))) >> shift;
+ }
+
+
+ /*vpx_memset(upred_ptr,expected_udc,64);*/
+ /*vpx_memset(vpred_ptr,expected_vdc,64);*/
+ for (i = 0; i < 8; i++)
+ {
+ vpx_memset(upred_ptr, expected_udc, 8);
+ vpx_memset(vpred_ptr, expected_vdc, 8);
+ upred_ptr += pred_stride;
+ vpred_ptr += pred_stride;
+ }
+ }
+ break;
+ case V_PRED:
+ {
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ vpx_memcpy(upred_ptr, uabove_row, 8);
+ vpx_memcpy(vpred_ptr, vabove_row, 8);
+ upred_ptr += pred_stride;
+ vpred_ptr += pred_stride;
+ }
+
+ }
+ break;
+ case H_PRED:
+ {
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ vpx_memset(upred_ptr, uleft_col[i], 8);
+ vpx_memset(vpred_ptr, vleft_col[i], 8);
+ upred_ptr += pred_stride;
+ vpred_ptr += pred_stride;
+ }
+ }
+
+ break;
+ case TM_PRED:
+ {
+ int i;
+
+ for (i = 0; i < 8; i++)
+ {
+ for (j = 0; j < 8; j++)
+ {
+ int predu = uleft_col[i] + uabove_row[j] - utop_left;
+ int predv = vleft_col[i] + vabove_row[j] - vtop_left;
+
+ if (predu < 0)
+ predu = 0;
+
+ if (predu > 255)
+ predu = 255;
+
+ if (predv < 0)
+ predv = 0;
+
+ if (predv > 255)
+ predv = 255;
+
+ upred_ptr[j] = predu;
+ vpred_ptr[j] = predv;
+ }
+
+ upred_ptr += pred_stride;
+ vpred_ptr += pred_stride;
+ }
+
+ }
+ break;
+ case B_PRED:
+ case NEARESTMV:
+ case NEARMV:
+ case ZEROMV:
+ case NEWMV:
+ case SPLITMV:
+ case MB_MODE_COUNT:
+ break;
+ }
+}
diff --git a/libvpx/vp8/common/reconintra4x4.c b/libvpx/vp8/common/reconintra4x4.c
new file mode 100644
index 0000000..7bb8d0a
--- /dev/null
+++ b/libvpx/vp8/common/reconintra4x4.c
@@ -0,0 +1,297 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "blockd.h"
+
+void vp8_intra4x4_predict_c(unsigned char *Above,
+ unsigned char *yleft, int left_stride,
+ B_PREDICTION_MODE b_mode,
+ unsigned char *dst, int dst_stride,
+ unsigned char top_left)
+{
+ int i, r, c;
+
+ unsigned char Left[4];
+ Left[0] = yleft[0];
+ Left[1] = yleft[left_stride];
+ Left[2] = yleft[2 * left_stride];
+ Left[3] = yleft[3 * left_stride];
+
+ switch (b_mode)
+ {
+ case B_DC_PRED:
+ {
+ int expected_dc = 0;
+
+ for (i = 0; i < 4; i++)
+ {
+ expected_dc += Above[i];
+ expected_dc += Left[i];
+ }
+
+ expected_dc = (expected_dc + 4) >> 3;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ dst[c] = expected_dc;
+ }
+
+ dst += dst_stride;
+ }
+ }
+ break;
+ case B_TM_PRED:
+ {
+ /* prediction similar to true_motion prediction */
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ int pred = Above[c] - top_left + Left[r];
+
+ if (pred < 0)
+ pred = 0;
+
+ if (pred > 255)
+ pred = 255;
+
+ dst[c] = pred;
+ }
+
+ dst += dst_stride;
+ }
+ }
+ break;
+
+ case B_VE_PRED:
+ {
+
+ unsigned int ap[4];
+ ap[0] = (top_left + 2 * Above[0] + Above[1] + 2) >> 2;
+ ap[1] = (Above[0] + 2 * Above[1] + Above[2] + 2) >> 2;
+ ap[2] = (Above[1] + 2 * Above[2] + Above[3] + 2) >> 2;
+ ap[3] = (Above[2] + 2 * Above[3] + Above[4] + 2) >> 2;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+
+ dst[c] = ap[c];
+ }
+
+ dst += dst_stride;
+ }
+
+ }
+ break;
+
+
+ case B_HE_PRED:
+ {
+
+ unsigned int lp[4];
+ lp[0] = (top_left + 2 * Left[0] + Left[1] + 2) >> 2;
+ lp[1] = (Left[0] + 2 * Left[1] + Left[2] + 2) >> 2;
+ lp[2] = (Left[1] + 2 * Left[2] + Left[3] + 2) >> 2;
+ lp[3] = (Left[2] + 2 * Left[3] + Left[3] + 2) >> 2;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ dst[c] = lp[r];
+ }
+
+ dst += dst_stride;
+ }
+ }
+ break;
+ case B_LD_PRED:
+ {
+ unsigned char *ptr = Above;
+ dst[0 * dst_stride + 0] = (ptr[0] + ptr[1] * 2 + ptr[2] + 2) >> 2;
+ dst[0 * dst_stride + 1] =
+ dst[1 * dst_stride + 0] = (ptr[1] + ptr[2] * 2 + ptr[3] + 2) >> 2;
+ dst[0 * dst_stride + 2] =
+ dst[1 * dst_stride + 1] =
+ dst[2 * dst_stride + 0] = (ptr[2] + ptr[3] * 2 + ptr[4] + 2) >> 2;
+ dst[0 * dst_stride + 3] =
+ dst[1 * dst_stride + 2] =
+ dst[2 * dst_stride + 1] =
+ dst[3 * dst_stride + 0] = (ptr[3] + ptr[4] * 2 + ptr[5] + 2) >> 2;
+ dst[1 * dst_stride + 3] =
+ dst[2 * dst_stride + 2] =
+ dst[3 * dst_stride + 1] = (ptr[4] + ptr[5] * 2 + ptr[6] + 2) >> 2;
+ dst[2 * dst_stride + 3] =
+ dst[3 * dst_stride + 2] = (ptr[5] + ptr[6] * 2 + ptr[7] + 2) >> 2;
+ dst[3 * dst_stride + 3] = (ptr[6] + ptr[7] * 2 + ptr[7] + 2) >> 2;
+
+ }
+ break;
+ case B_RD_PRED:
+ {
+
+ unsigned char pp[9];
+
+ pp[0] = Left[3];
+ pp[1] = Left[2];
+ pp[2] = Left[1];
+ pp[3] = Left[0];
+ pp[4] = top_left;
+ pp[5] = Above[0];
+ pp[6] = Above[1];
+ pp[7] = Above[2];
+ pp[8] = Above[3];
+
+ dst[3 * dst_stride + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
+ dst[3 * dst_stride + 1] =
+ dst[2 * dst_stride + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
+ dst[3 * dst_stride + 2] =
+ dst[2 * dst_stride + 1] =
+ dst[1 * dst_stride + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
+ dst[3 * dst_stride + 3] =
+ dst[2 * dst_stride + 2] =
+ dst[1 * dst_stride + 1] =
+ dst[0 * dst_stride + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
+ dst[2 * dst_stride + 3] =
+ dst[1 * dst_stride + 2] =
+ dst[0 * dst_stride + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
+ dst[1 * dst_stride + 3] =
+ dst[0 * dst_stride + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
+ dst[0 * dst_stride + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
+
+ }
+ break;
+ case B_VR_PRED:
+ {
+
+ unsigned char pp[9];
+
+ pp[0] = Left[3];
+ pp[1] = Left[2];
+ pp[2] = Left[1];
+ pp[3] = Left[0];
+ pp[4] = top_left;
+ pp[5] = Above[0];
+ pp[6] = Above[1];
+ pp[7] = Above[2];
+ pp[8] = Above[3];
+
+
+ dst[3 * dst_stride + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
+ dst[2 * dst_stride + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
+ dst[3 * dst_stride + 1] =
+ dst[1 * dst_stride + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
+ dst[2 * dst_stride + 1] =
+ dst[0 * dst_stride + 0] = (pp[4] + pp[5] + 1) >> 1;
+ dst[3 * dst_stride + 2] =
+ dst[1 * dst_stride + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
+ dst[2 * dst_stride + 2] =
+ dst[0 * dst_stride + 1] = (pp[5] + pp[6] + 1) >> 1;
+ dst[3 * dst_stride + 3] =
+ dst[1 * dst_stride + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
+ dst[2 * dst_stride + 3] =
+ dst[0 * dst_stride + 2] = (pp[6] + pp[7] + 1) >> 1;
+ dst[1 * dst_stride + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
+ dst[0 * dst_stride + 3] = (pp[7] + pp[8] + 1) >> 1;
+
+ }
+ break;
+ case B_VL_PRED:
+ {
+
+ unsigned char *pp = Above;
+
+ dst[0 * dst_stride + 0] = (pp[0] + pp[1] + 1) >> 1;
+ dst[1 * dst_stride + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
+ dst[2 * dst_stride + 0] =
+ dst[0 * dst_stride + 1] = (pp[1] + pp[2] + 1) >> 1;
+ dst[1 * dst_stride + 1] =
+ dst[3 * dst_stride + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
+ dst[2 * dst_stride + 1] =
+ dst[0 * dst_stride + 2] = (pp[2] + pp[3] + 1) >> 1;
+ dst[3 * dst_stride + 1] =
+ dst[1 * dst_stride + 2] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
+ dst[0 * dst_stride + 3] =
+ dst[2 * dst_stride + 2] = (pp[3] + pp[4] + 1) >> 1;
+ dst[1 * dst_stride + 3] =
+ dst[3 * dst_stride + 2] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
+ dst[2 * dst_stride + 3] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
+ dst[3 * dst_stride + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
+ }
+ break;
+
+ case B_HD_PRED:
+ {
+ unsigned char pp[9];
+ pp[0] = Left[3];
+ pp[1] = Left[2];
+ pp[2] = Left[1];
+ pp[3] = Left[0];
+ pp[4] = top_left;
+ pp[5] = Above[0];
+ pp[6] = Above[1];
+ pp[7] = Above[2];
+ pp[8] = Above[3];
+
+
+ dst[3 * dst_stride + 0] = (pp[0] + pp[1] + 1) >> 1;
+ dst[3 * dst_stride + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
+ dst[2 * dst_stride + 0] =
+ dst[3 * dst_stride + 2] = (pp[1] + pp[2] + 1) >> 1;
+ dst[2 * dst_stride + 1] =
+ dst[3 * dst_stride + 3] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
+ dst[2 * dst_stride + 2] =
+ dst[1 * dst_stride + 0] = (pp[2] + pp[3] + 1) >> 1;
+ dst[2 * dst_stride + 3] =
+ dst[1 * dst_stride + 1] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
+ dst[1 * dst_stride + 2] =
+ dst[0 * dst_stride + 0] = (pp[3] + pp[4] + 1) >> 1;
+ dst[1 * dst_stride + 3] =
+ dst[0 * dst_stride + 1] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
+ dst[0 * dst_stride + 2] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
+ dst[0 * dst_stride + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
+ }
+ break;
+
+
+ case B_HU_PRED:
+ {
+ unsigned char *pp = Left;
+ dst[0 * dst_stride + 0] = (pp[0] + pp[1] + 1) >> 1;
+ dst[0 * dst_stride + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
+ dst[0 * dst_stride + 2] =
+ dst[1 * dst_stride + 0] = (pp[1] + pp[2] + 1) >> 1;
+ dst[0 * dst_stride + 3] =
+ dst[1 * dst_stride + 1] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
+ dst[1 * dst_stride + 2] =
+ dst[2 * dst_stride + 0] = (pp[2] + pp[3] + 1) >> 1;
+ dst[1 * dst_stride + 3] =
+ dst[2 * dst_stride + 1] = (pp[2] + pp[3] * 2 + pp[3] + 2) >> 2;
+ dst[2 * dst_stride + 2] =
+ dst[2 * dst_stride + 3] =
+ dst[3 * dst_stride + 0] =
+ dst[3 * dst_stride + 1] =
+ dst[3 * dst_stride + 2] =
+ dst[3 * dst_stride + 3] = pp[3];
+ }
+ break;
+
+ default:
+ break;
+
+ }
+}
diff --git a/libvpx/vp8/common/reconintra4x4.h b/libvpx/vp8/common/reconintra4x4.h
new file mode 100644
index 0000000..d2b0d43
--- /dev/null
+++ b/libvpx/vp8/common/reconintra4x4.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_RECONINTRA4x4_H
+#define __INC_RECONINTRA4x4_H
+#include "vp8/common/blockd.h"
+
+static void intra_prediction_down_copy(MACROBLOCKD *xd,
+ unsigned char *above_right_src)
+{
+ int dst_stride = xd->dst.y_stride;
+ unsigned char *above_right_dst = xd->dst.y_buffer - dst_stride + 16;
+
+ unsigned int *src_ptr = (unsigned int *)above_right_src;
+ unsigned int *dst_ptr0 = (unsigned int *)(above_right_dst + 4 * dst_stride);
+ unsigned int *dst_ptr1 = (unsigned int *)(above_right_dst + 8 * dst_stride);
+ unsigned int *dst_ptr2 = (unsigned int *)(above_right_dst + 12 * dst_stride);
+
+ *dst_ptr0 = *src_ptr;
+ *dst_ptr1 = *src_ptr;
+ *dst_ptr2 = *src_ptr;
+}
+
+#endif
diff --git a/libvpx/vp8/common/rtcd.c b/libvpx/vp8/common/rtcd.c
new file mode 100644
index 0000000..01dad46
--- /dev/null
+++ b/libvpx/vp8/common/rtcd.c
@@ -0,0 +1,105 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include "vpx_config.h"
+#define RTCD_C
+#include "vpx_rtcd.h"
+
+#if CONFIG_MULTITHREAD && defined(_WIN32)
+#include <windows.h>
+#include <stdlib.h>
+static void once(void (*func)(void))
+{
+ static CRITICAL_SECTION *lock;
+ static LONG waiters;
+ static int done;
+ void *lock_ptr = &lock;
+
+ /* If the initialization is complete, return early. This isn't just an
+ * optimization, it prevents races on the destruction of the global
+ * lock.
+ */
+ if(done)
+ return;
+
+ InterlockedIncrement(&waiters);
+
+ /* Get a lock. We create one and try to make it the one-true-lock,
+ * throwing it away if we lost the race.
+ */
+
+ {
+ /* Scope to protect access to new_lock */
+ CRITICAL_SECTION *new_lock = malloc(sizeof(CRITICAL_SECTION));
+ InitializeCriticalSection(new_lock);
+ if (InterlockedCompareExchangePointer(lock_ptr, new_lock, NULL) != NULL)
+ {
+ DeleteCriticalSection(new_lock);
+ free(new_lock);
+ }
+ }
+
+ /* At this point, we have a lock that can be synchronized on. We don't
+ * care which thread actually performed the allocation.
+ */
+
+ EnterCriticalSection(lock);
+
+ if (!done)
+ {
+ func();
+ done = 1;
+ }
+
+ LeaveCriticalSection(lock);
+
+ /* Last one out should free resources. The destructed objects are
+ * protected by checking if(done) above.
+ */
+ if(!InterlockedDecrement(&waiters))
+ {
+ DeleteCriticalSection(lock);
+ free(lock);
+ lock = NULL;
+ }
+}
+
+
+#elif CONFIG_MULTITHREAD && HAVE_PTHREAD_H
+#include <pthread.h>
+static void once(void (*func)(void))
+{
+ static pthread_once_t lock = PTHREAD_ONCE_INIT;
+ pthread_once(&lock, func);
+}
+
+
+#else
+/* No-op version that performs no synchronization. vpx_rtcd() is idempotent,
+ * so as long as your platform provides atomic loads/stores of pointers
+ * no synchronization is strictly necessary.
+ */
+
+static void once(void (*func)(void))
+{
+ static int done;
+
+ if(!done)
+ {
+ func();
+ done = 1;
+ }
+}
+#endif
+
+
+void vpx_rtcd()
+{
+ once(setup_rtcd_internal);
+}
diff --git a/libvpx/vp8/common/rtcd_defs.sh b/libvpx/vp8/common/rtcd_defs.sh
new file mode 100644
index 0000000..0f950f8
--- /dev/null
+++ b/libvpx/vp8/common/rtcd_defs.sh
@@ -0,0 +1,568 @@
+common_forward_decls() {
+cat <<EOF
+#include "vp8/common/blockd.h"
+
+struct blockd;
+struct macroblockd;
+struct loop_filter_info;
+
+/* Encoder forward decls */
+struct block;
+struct macroblock;
+struct variance_vtable;
+union int_mv;
+struct yv12_buffer_config;
+EOF
+}
+forward_decls common_forward_decls
+
+#
+# Dequant
+#
+prototype void vp8_dequantize_b "struct blockd*, short *dqc"
+specialize vp8_dequantize_b mmx media neon
+vp8_dequantize_b_media=vp8_dequantize_b_v6
+
+prototype void vp8_dequant_idct_add "short *input, short *dq, unsigned char *output, int stride"
+specialize vp8_dequant_idct_add mmx media neon dspr2
+vp8_dequant_idct_add_media=vp8_dequant_idct_add_v6
+vp8_dequant_idct_add_dspr2=vp8_dequant_idct_add_dspr2
+
+prototype void vp8_dequant_idct_add_y_block "short *q, short *dq, unsigned char *dst, int stride, char *eobs"
+specialize vp8_dequant_idct_add_y_block mmx sse2 media neon dspr2
+vp8_dequant_idct_add_y_block_media=vp8_dequant_idct_add_y_block_v6
+vp8_dequant_idct_add_y_block_dspr2=vp8_dequant_idct_add_y_block_dspr2
+
+prototype void vp8_dequant_idct_add_uv_block "short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs"
+specialize vp8_dequant_idct_add_uv_block mmx sse2 media neon dspr2
+vp8_dequant_idct_add_uv_block_media=vp8_dequant_idct_add_uv_block_v6
+vp8_dequant_idct_add_y_block_dspr2=vp8_dequant_idct_add_y_block_dspr2
+
+#
+# Loopfilter
+#
+prototype void vp8_loop_filter_mbv "unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi"
+specialize vp8_loop_filter_mbv mmx sse2 media neon dspr2
+vp8_loop_filter_mbv_media=vp8_loop_filter_mbv_armv6
+vp8_loop_filter_mbv_dspr2=vp8_loop_filter_mbv_dspr2
+
+prototype void vp8_loop_filter_bv "unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi"
+specialize vp8_loop_filter_bv mmx sse2 media neon dspr2
+vp8_loop_filter_bv_media=vp8_loop_filter_bv_armv6
+vp8_loop_filter_bv_dspr2=vp8_loop_filter_bv_dspr2
+
+prototype void vp8_loop_filter_mbh "unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi"
+specialize vp8_loop_filter_mbh mmx sse2 media neon dspr2
+vp8_loop_filter_mbh_media=vp8_loop_filter_mbh_armv6
+vp8_loop_filter_mbh_dspr2=vp8_loop_filter_mbh_dspr2
+
+prototype void vp8_loop_filter_bh "unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi"
+specialize vp8_loop_filter_bh mmx sse2 media neon dspr2
+vp8_loop_filter_bh_media=vp8_loop_filter_bh_armv6
+vp8_loop_filter_bh_dspr2=vp8_loop_filter_bh_dspr2
+
+
+prototype void vp8_loop_filter_simple_mbv "unsigned char *y, int ystride, const unsigned char *blimit"
+specialize vp8_loop_filter_simple_mbv mmx sse2 media neon
+vp8_loop_filter_simple_mbv_c=vp8_loop_filter_simple_vertical_edge_c
+vp8_loop_filter_simple_mbv_mmx=vp8_loop_filter_simple_vertical_edge_mmx
+vp8_loop_filter_simple_mbv_sse2=vp8_loop_filter_simple_vertical_edge_sse2
+vp8_loop_filter_simple_mbv_media=vp8_loop_filter_simple_vertical_edge_armv6
+vp8_loop_filter_simple_mbv_neon=vp8_loop_filter_mbvs_neon
+
+prototype void vp8_loop_filter_simple_mbh "unsigned char *y, int ystride, const unsigned char *blimit"
+specialize vp8_loop_filter_simple_mbh mmx sse2 media neon
+vp8_loop_filter_simple_mbh_c=vp8_loop_filter_simple_horizontal_edge_c
+vp8_loop_filter_simple_mbh_mmx=vp8_loop_filter_simple_horizontal_edge_mmx
+vp8_loop_filter_simple_mbh_sse2=vp8_loop_filter_simple_horizontal_edge_sse2
+vp8_loop_filter_simple_mbh_media=vp8_loop_filter_simple_horizontal_edge_armv6
+vp8_loop_filter_simple_mbh_neon=vp8_loop_filter_mbhs_neon
+
+prototype void vp8_loop_filter_simple_bv "unsigned char *y, int ystride, const unsigned char *blimit"
+specialize vp8_loop_filter_simple_bv mmx sse2 media neon
+vp8_loop_filter_simple_bv_c=vp8_loop_filter_bvs_c
+vp8_loop_filter_simple_bv_mmx=vp8_loop_filter_bvs_mmx
+vp8_loop_filter_simple_bv_sse2=vp8_loop_filter_bvs_sse2
+vp8_loop_filter_simple_bv_media=vp8_loop_filter_bvs_armv6
+vp8_loop_filter_simple_bv_neon=vp8_loop_filter_bvs_neon
+
+prototype void vp8_loop_filter_simple_bh "unsigned char *y, int ystride, const unsigned char *blimit"
+specialize vp8_loop_filter_simple_bh mmx sse2 media neon
+vp8_loop_filter_simple_bh_c=vp8_loop_filter_bhs_c
+vp8_loop_filter_simple_bh_mmx=vp8_loop_filter_bhs_mmx
+vp8_loop_filter_simple_bh_sse2=vp8_loop_filter_bhs_sse2
+vp8_loop_filter_simple_bh_media=vp8_loop_filter_bhs_armv6
+vp8_loop_filter_simple_bh_neon=vp8_loop_filter_bhs_neon
+
+#
+# IDCT
+#
+#idct16
+prototype void vp8_short_idct4x4llm "short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride"
+specialize vp8_short_idct4x4llm mmx media neon dspr2
+vp8_short_idct4x4llm_media=vp8_short_idct4x4llm_v6_dual
+vp8_short_idct4x4llm_dspr2=vp8_short_idct4x4llm_dspr2
+
+#iwalsh1
+prototype void vp8_short_inv_walsh4x4_1 "short *input, short *output"
+specialize vp8_short_inv_walsh4x4_1 dspr2
+vp8_short_inv_walsh4x4_1_dspr2=vp8_short_inv_walsh4x4_1_dspr2
+# no asm yet
+
+#iwalsh16
+prototype void vp8_short_inv_walsh4x4 "short *input, short *output"
+specialize vp8_short_inv_walsh4x4 mmx sse2 media neon dspr2
+vp8_short_inv_walsh4x4_media=vp8_short_inv_walsh4x4_v6
+vp8_short_inv_walsh4x4_dspr2=vp8_short_inv_walsh4x4_dspr2
+
+#idct1_scalar_add
+prototype void vp8_dc_only_idct_add "short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride"
+specialize vp8_dc_only_idct_add mmx media neon dspr2
+vp8_dc_only_idct_add_media=vp8_dc_only_idct_add_v6
+vp8_dc_only_idct_add_dspr2=vp8_dc_only_idct_add_dspr2
+
+#
+# RECON
+#
+prototype void vp8_copy_mem16x16 "unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch"
+specialize vp8_copy_mem16x16 mmx sse2 media neon dspr2
+vp8_copy_mem16x16_media=vp8_copy_mem16x16_v6
+vp8_copy_mem16x16_dspr2=vp8_copy_mem16x16_dspr2
+
+prototype void vp8_copy_mem8x8 "unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch"
+specialize vp8_copy_mem8x8 mmx media neon dspr2
+vp8_copy_mem8x8_media=vp8_copy_mem8x8_v6
+vp8_copy_mem8x8_dspr2=vp8_copy_mem8x8_dspr2
+
+prototype void vp8_copy_mem8x4 "unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch"
+specialize vp8_copy_mem8x4 mmx media neon dspr2
+vp8_copy_mem8x4_media=vp8_copy_mem8x4_v6
+vp8_copy_mem8x4_dspr2=vp8_copy_mem8x4_dspr2
+
+prototype void vp8_build_intra_predictors_mby_s "struct macroblockd *x, unsigned char * yabove_row, unsigned char * yleft, int left_stride, unsigned char * ypred_ptr, int y_stride"
+specialize vp8_build_intra_predictors_mby_s sse2 ssse3
+#TODO: fix assembly for neon
+
+prototype void vp8_build_intra_predictors_mbuv_s "struct macroblockd *x, unsigned char * uabove_row, unsigned char * vabove_row, unsigned char *uleft, unsigned char *vleft, int left_stride, unsigned char * upred_ptr, unsigned char * vpred_ptr, int pred_stride"
+specialize vp8_build_intra_predictors_mbuv_s sse2 ssse3
+
+prototype void vp8_intra4x4_predict "unsigned char *Above, unsigned char *yleft, int left_stride, B_PREDICTION_MODE b_mode, unsigned char *dst, int dst_stride, unsigned char top_left"
+specialize vp8_intra4x4_predict media
+vp8_intra4x4_predict_media=vp8_intra4x4_predict_armv6
+
+#
+# Postproc
+#
+if [ "$CONFIG_POSTPROC" = "yes" ]; then
+ prototype void vp8_mbpost_proc_down "unsigned char *dst, int pitch, int rows, int cols,int flimit"
+ specialize vp8_mbpost_proc_down mmx sse2
+ vp8_mbpost_proc_down_sse2=vp8_mbpost_proc_down_xmm
+
+ prototype void vp8_mbpost_proc_across_ip "unsigned char *dst, int pitch, int rows, int cols,int flimit"
+ specialize vp8_mbpost_proc_across_ip sse2
+ vp8_mbpost_proc_across_ip_sse2=vp8_mbpost_proc_across_ip_xmm
+
+ prototype void vp8_post_proc_down_and_across_mb_row "unsigned char *src, unsigned char *dst, int src_pitch, int dst_pitch, int cols, unsigned char *flimits, int size"
+ specialize vp8_post_proc_down_and_across_mb_row sse2
+
+ prototype void vp8_plane_add_noise "unsigned char *s, char *noise, char blackclamp[16], char whiteclamp[16], char bothclamp[16], unsigned int w, unsigned int h, int pitch"
+ specialize vp8_plane_add_noise mmx sse2
+ vp8_plane_add_noise_sse2=vp8_plane_add_noise_wmt
+
+ prototype void vp8_blend_mb_inner "unsigned char *y, unsigned char *u, unsigned char *v, int y1, int u1, int v1, int alpha, int stride"
+ # no asm yet
+
+ prototype void vp8_blend_mb_outer "unsigned char *y, unsigned char *u, unsigned char *v, int y1, int u1, int v1, int alpha, int stride"
+ # no asm yet
+
+ prototype void vp8_blend_b "unsigned char *y, unsigned char *u, unsigned char *v, int y1, int u1, int v1, int alpha, int stride"
+ # no asm yet
+
+ prototype void vp8_filter_by_weight16x16 "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride, int src_weight"
+ specialize vp8_filter_by_weight16x16 sse2
+
+ prototype void vp8_filter_by_weight8x8 "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride, int src_weight"
+ specialize vp8_filter_by_weight8x8 sse2
+
+ prototype void vp8_filter_by_weight4x4 "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride, int src_weight"
+ # no asm yet
+fi
+
+#
+# Subpixel
+#
+prototype void vp8_sixtap_predict16x16 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
+specialize vp8_sixtap_predict16x16 mmx sse2 ssse3 media neon dspr2
+vp8_sixtap_predict16x16_media=vp8_sixtap_predict16x16_armv6
+vp8_sixtap_predict16x16_dspr2=vp8_sixtap_predict16x16_dspr2
+
+prototype void vp8_sixtap_predict8x8 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
+specialize vp8_sixtap_predict8x8 mmx sse2 ssse3 media neon dspr2
+vp8_sixtap_predict8x8_media=vp8_sixtap_predict8x8_armv6
+vp8_sixtap_predict8x8_dspr2=vp8_sixtap_predict8x8_dspr2
+
+prototype void vp8_sixtap_predict8x4 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
+specialize vp8_sixtap_predict8x4 mmx sse2 ssse3 media neon dspr2
+vp8_sixtap_predict8x4_media=vp8_sixtap_predict8x4_armv6
+vp8_sixtap_predict8x4_dspr2=vp8_sixtap_predict8x4_dspr2
+
+prototype void vp8_sixtap_predict4x4 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
+specialize vp8_sixtap_predict4x4 mmx ssse3 media neon dspr2
+vp8_sixtap_predict4x4_media=vp8_sixtap_predict4x4_armv6
+vp8_sixtap_predict4x4_dspr2=vp8_sixtap_predict4x4_dspr2
+
+prototype void vp8_bilinear_predict16x16 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
+specialize vp8_bilinear_predict16x16 mmx sse2 ssse3 media neon
+vp8_bilinear_predict16x16_media=vp8_bilinear_predict16x16_armv6
+
+prototype void vp8_bilinear_predict8x8 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
+specialize vp8_bilinear_predict8x8 mmx sse2 ssse3 media neon
+vp8_bilinear_predict8x8_media=vp8_bilinear_predict8x8_armv6
+
+prototype void vp8_bilinear_predict8x4 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
+specialize vp8_bilinear_predict8x4 mmx media neon
+vp8_bilinear_predict8x4_media=vp8_bilinear_predict8x4_armv6
+
+prototype void vp8_bilinear_predict4x4 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
+specialize vp8_bilinear_predict4x4 mmx media neon
+vp8_bilinear_predict4x4_media=vp8_bilinear_predict4x4_armv6
+
+#
+# Whole-pixel Variance
+#
+prototype unsigned int vp8_variance4x4 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_variance4x4 mmx sse2
+vp8_variance4x4_sse2=vp8_variance4x4_wmt
+
+prototype unsigned int vp8_variance8x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_variance8x8 mmx sse2 media neon
+vp8_variance8x8_sse2=vp8_variance8x8_wmt
+vp8_variance8x8_media=vp8_variance8x8_armv6
+
+prototype unsigned int vp8_variance8x16 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_variance8x16 mmx sse2 neon
+vp8_variance8x16_sse2=vp8_variance8x16_wmt
+
+prototype unsigned int vp8_variance16x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_variance16x8 mmx sse2 neon
+vp8_variance16x8_sse2=vp8_variance16x8_wmt
+
+prototype unsigned int vp8_variance16x16 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_variance16x16 mmx sse2 media neon
+vp8_variance16x16_sse2=vp8_variance16x16_wmt
+vp8_variance16x16_media=vp8_variance16x16_armv6
+
+#
+# Sub-pixel Variance
+#
+prototype unsigned int vp8_sub_pixel_variance4x4 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
+specialize vp8_sub_pixel_variance4x4 mmx sse2
+vp8_sub_pixel_variance4x4_sse2=vp8_sub_pixel_variance4x4_wmt
+
+prototype unsigned int vp8_sub_pixel_variance8x8 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
+specialize vp8_sub_pixel_variance8x8 mmx sse2 media neon
+vp8_sub_pixel_variance8x8_sse2=vp8_sub_pixel_variance8x8_wmt
+vp8_sub_pixel_variance8x8_media=vp8_sub_pixel_variance8x8_armv6
+
+prototype unsigned int vp8_sub_pixel_variance8x16 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
+specialize vp8_sub_pixel_variance8x16 mmx sse2
+vp8_sub_pixel_variance8x16_sse2=vp8_sub_pixel_variance8x16_wmt
+
+prototype unsigned int vp8_sub_pixel_variance16x8 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
+specialize vp8_sub_pixel_variance16x8 mmx sse2 ssse3
+vp8_sub_pixel_variance16x8_sse2=vp8_sub_pixel_variance16x8_wmt
+
+prototype unsigned int vp8_sub_pixel_variance16x16 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
+specialize vp8_sub_pixel_variance16x16 mmx sse2 ssse3 media neon
+vp8_sub_pixel_variance16x16_sse2=vp8_sub_pixel_variance16x16_wmt
+vp8_sub_pixel_variance16x16_media=vp8_sub_pixel_variance16x16_armv6
+
+prototype unsigned int vp8_variance_halfpixvar16x16_h "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_variance_halfpixvar16x16_h mmx sse2 media neon
+vp8_variance_halfpixvar16x16_h_sse2=vp8_variance_halfpixvar16x16_h_wmt
+vp8_variance_halfpixvar16x16_h_media=vp8_variance_halfpixvar16x16_h_armv6
+
+prototype unsigned int vp8_variance_halfpixvar16x16_v "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_variance_halfpixvar16x16_v mmx sse2 media neon
+vp8_variance_halfpixvar16x16_v_sse2=vp8_variance_halfpixvar16x16_v_wmt
+vp8_variance_halfpixvar16x16_v_media=vp8_variance_halfpixvar16x16_v_armv6
+
+prototype unsigned int vp8_variance_halfpixvar16x16_hv "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_variance_halfpixvar16x16_hv mmx sse2 media neon
+vp8_variance_halfpixvar16x16_hv_sse2=vp8_variance_halfpixvar16x16_hv_wmt
+vp8_variance_halfpixvar16x16_hv_media=vp8_variance_halfpixvar16x16_hv_armv6
+
+#
+# Single block SAD
+#
+prototype unsigned int vp8_sad4x4 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad"
+specialize vp8_sad4x4 mmx sse2 neon
+vp8_sad4x4_sse2=vp8_sad4x4_wmt
+
+prototype unsigned int vp8_sad8x8 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad"
+specialize vp8_sad8x8 mmx sse2 neon
+vp8_sad8x8_sse2=vp8_sad8x8_wmt
+
+prototype unsigned int vp8_sad8x16 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad"
+specialize vp8_sad8x16 mmx sse2 neon
+vp8_sad8x16_sse2=vp8_sad8x16_wmt
+
+prototype unsigned int vp8_sad16x8 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad"
+specialize vp8_sad16x8 mmx sse2 neon
+vp8_sad16x8_sse2=vp8_sad16x8_wmt
+
+prototype unsigned int vp8_sad16x16 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad"
+specialize vp8_sad16x16 mmx sse2 sse3 media neon
+vp8_sad16x16_sse2=vp8_sad16x16_wmt
+vp8_sad16x16_media=vp8_sad16x16_armv6
+
+#
+# Multi-block SAD, comparing a reference to N blocks 1 pixel apart horizontally
+#
+prototype void vp8_sad4x4x3 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
+specialize vp8_sad4x4x3 sse3
+
+prototype void vp8_sad8x8x3 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
+specialize vp8_sad8x8x3 sse3
+
+prototype void vp8_sad8x16x3 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
+specialize vp8_sad8x16x3 sse3
+
+prototype void vp8_sad16x8x3 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
+specialize vp8_sad16x8x3 sse3 ssse3
+
+prototype void vp8_sad16x16x3 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
+specialize vp8_sad16x16x3 sse3 ssse3
+
+# Note the only difference in the following prototypes is that they return into
+# an array of short
+prototype void vp8_sad4x4x8 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
+specialize vp8_sad4x4x8 sse4_1
+vp8_sad4x4x8_sse4_1=vp8_sad4x4x8_sse4
+
+prototype void vp8_sad8x8x8 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
+specialize vp8_sad8x8x8 sse4_1
+vp8_sad8x8x8_sse4_1=vp8_sad8x8x8_sse4
+
+prototype void vp8_sad8x16x8 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
+specialize vp8_sad8x16x8 sse4_1
+vp8_sad8x16x8_sse4_1=vp8_sad8x16x8_sse4
+
+prototype void vp8_sad16x8x8 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
+specialize vp8_sad16x8x8 sse4_1
+vp8_sad16x8x8_sse4_1=vp8_sad16x8x8_sse4
+
+prototype void vp8_sad16x16x8 "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
+specialize vp8_sad16x16x8 sse4_1
+vp8_sad16x16x8_sse4_1=vp8_sad16x16x8_sse4
+
+#
+# Multi-block SAD, comparing a reference to N independent blocks
+#
+prototype void vp8_sad4x4x4d "const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array"
+specialize vp8_sad4x4x4d sse3
+
+prototype void vp8_sad8x8x4d "const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array"
+specialize vp8_sad8x8x4d sse3
+
+prototype void vp8_sad8x16x4d "const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array"
+specialize vp8_sad8x16x4d sse3
+
+prototype void vp8_sad16x8x4d "const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array"
+specialize vp8_sad16x8x4d sse3
+
+prototype void vp8_sad16x16x4d "const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array"
+specialize vp8_sad16x16x4d sse3
+
+#
+# Encoder functions below this point.
+#
+if [ "$CONFIG_VP8_ENCODER" = "yes" ]; then
+
+#
+# Sum of squares (vector)
+#
+prototype unsigned int vp8_get_mb_ss "const short *"
+specialize vp8_get_mb_ss mmx sse2
+
+#
+# SSE (Sum Squared Error)
+#
+prototype unsigned int vp8_sub_pixel_mse16x16 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
+specialize vp8_sub_pixel_mse16x16 mmx sse2
+vp8_sub_pixel_mse16x16_sse2=vp8_sub_pixel_mse16x16_wmt
+
+prototype unsigned int vp8_mse16x16 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
+specialize vp8_mse16x16 mmx sse2 media neon
+vp8_mse16x16_sse2=vp8_mse16x16_wmt
+vp8_mse16x16_media=vp8_mse16x16_armv6
+
+prototype unsigned int vp8_get4x4sse_cs "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride"
+specialize vp8_get4x4sse_cs mmx neon
+
+#
+# Block copy
+#
+case $arch in
+ x86*)
+ prototype void vp8_copy32xn "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, int n"
+ specialize vp8_copy32xn sse2 sse3
+ ;;
+esac
+
+#
+# Structured Similarity (SSIM)
+#
+if [ "$CONFIG_INTERNAL_STATS" = "yes" ]; then
+ [ $arch = "x86_64" ] && sse2_on_x86_64=sse2
+
+ prototype void vp8_ssim_parms_8x8 "unsigned char *s, int sp, unsigned char *r, int rp, unsigned long *sum_s, unsigned long *sum_r, unsigned long *sum_sq_s, unsigned long *sum_sq_r, unsigned long *sum_sxr"
+ specialize vp8_ssim_parms_8x8 $sse2_on_x86_64
+
+ prototype void vp8_ssim_parms_16x16 "unsigned char *s, int sp, unsigned char *r, int rp, unsigned long *sum_s, unsigned long *sum_r, unsigned long *sum_sq_s, unsigned long *sum_sq_r, unsigned long *sum_sxr"
+ specialize vp8_ssim_parms_16x16 $sse2_on_x86_64
+fi
+
+#
+# Forward DCT
+#
+prototype void vp8_short_fdct4x4 "short *input, short *output, int pitch"
+specialize vp8_short_fdct4x4 mmx sse2 media neon
+vp8_short_fdct4x4_media=vp8_short_fdct4x4_armv6
+
+prototype void vp8_short_fdct8x4 "short *input, short *output, int pitch"
+specialize vp8_short_fdct8x4 mmx sse2 media neon
+vp8_short_fdct8x4_media=vp8_short_fdct8x4_armv6
+
+prototype void vp8_short_walsh4x4 "short *input, short *output, int pitch"
+specialize vp8_short_walsh4x4 sse2 media neon
+vp8_short_walsh4x4_media=vp8_short_walsh4x4_armv6
+
+#
+# Quantizer
+#
+prototype void vp8_regular_quantize_b "struct block *, struct blockd *"
+specialize vp8_regular_quantize_b sse2 sse4_1
+vp8_regular_quantize_b_sse4_1=vp8_regular_quantize_b_sse4
+
+prototype void vp8_fast_quantize_b "struct block *, struct blockd *"
+specialize vp8_fast_quantize_b sse2 ssse3 media neon
+vp8_fast_quantize_b_media=vp8_fast_quantize_b_armv6
+
+prototype void vp8_regular_quantize_b_pair "struct block *b1, struct block *b2, struct blockd *d1, struct blockd *d2"
+# no asm yet
+
+prototype void vp8_fast_quantize_b_pair "struct block *b1, struct block *b2, struct blockd *d1, struct blockd *d2"
+specialize vp8_fast_quantize_b_pair neon
+
+prototype void vp8_quantize_mb "struct macroblock *"
+specialize vp8_quantize_mb neon
+
+prototype void vp8_quantize_mby "struct macroblock *"
+specialize vp8_quantize_mby neon
+
+prototype void vp8_quantize_mbuv "struct macroblock *"
+specialize vp8_quantize_mbuv neon
+
+#
+# Block subtraction
+#
+prototype int vp8_block_error "short *coeff, short *dqcoeff"
+specialize vp8_block_error mmx sse2
+vp8_block_error_sse2=vp8_block_error_xmm
+
+prototype int vp8_mbblock_error "struct macroblock *mb, int dc"
+specialize vp8_mbblock_error mmx sse2
+vp8_mbblock_error_sse2=vp8_mbblock_error_xmm
+
+prototype int vp8_mbuverror "struct macroblock *mb"
+specialize vp8_mbuverror mmx sse2
+vp8_mbuverror_sse2=vp8_mbuverror_xmm
+
+prototype void vp8_subtract_b "struct block *be, struct blockd *bd, int pitch"
+specialize vp8_subtract_b mmx sse2 media neon
+vp8_subtract_b_media=vp8_subtract_b_armv6
+
+prototype void vp8_subtract_mby "short *diff, unsigned char *src, int src_stride, unsigned char *pred, int pred_stride"
+specialize vp8_subtract_mby mmx sse2 media neon
+vp8_subtract_mby_media=vp8_subtract_mby_armv6
+
+prototype void vp8_subtract_mbuv "short *diff, unsigned char *usrc, unsigned char *vsrc, int src_stride, unsigned char *upred, unsigned char *vpred, int pred_stride"
+specialize vp8_subtract_mbuv mmx sse2 media neon
+vp8_subtract_mbuv_media=vp8_subtract_mbuv_armv6
+
+#
+# Motion search
+#
+prototype int vp8_full_search_sad "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, int sad_per_bit, int distance, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"
+specialize vp8_full_search_sad sse3 sse4_1
+vp8_full_search_sad_sse3=vp8_full_search_sadx3
+vp8_full_search_sad_sse4_1=vp8_full_search_sadx8
+
+prototype int vp8_refining_search_sad "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, int sad_per_bit, int distance, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"
+specialize vp8_refining_search_sad sse3
+vp8_refining_search_sad_sse3=vp8_refining_search_sadx4
+
+prototype int vp8_diamond_search_sad "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, union int_mv *best_mv, int search_param, int sad_per_bit, int *num00, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"
+vp8_diamond_search_sad_sse3=vp8_diamond_search_sadx4
+
+#
+# Alt-ref Noise Reduction (ARNR)
+#
+if [ "$CONFIG_REALTIME_ONLY" != "yes" ]; then
+ prototype void vp8_temporal_filter_apply "unsigned char *frame1, unsigned int stride, unsigned char *frame2, unsigned int block_size, int strength, int filter_weight, unsigned int *accumulator, unsigned short *count"
+ specialize vp8_temporal_filter_apply sse2
+fi
+
+#
+# Pick Loopfilter
+#
+prototype void vp8_yv12_copy_partial_frame "struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc"
+specialize vp8_yv12_copy_partial_frame neon
+
+#
+# Denoiser filter
+#
+if [ "$CONFIG_TEMPORAL_DENOISING" = "yes" ]; then
+ prototype int vp8_denoiser_filter "struct yv12_buffer_config* mc_running_avg, struct yv12_buffer_config* running_avg, struct macroblock* signal, unsigned int motion_magnitude2, int y_offset, int uv_offset"
+ specialize vp8_denoiser_filter sse2
+fi
+
+# End of encoder only functions
+fi
+
+# Scaler functions
+if [ "CONFIG_SPATIAL_RESAMPLING" != "yes" ]; then
+ prototype void vp8_horizontal_line_4_5_scale "const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width"
+ prototype void vp8_vertical_band_4_5_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_last_vertical_band_4_5_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_horizontal_line_2_3_scale "const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width"
+ prototype void vp8_vertical_band_2_3_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_last_vertical_band_2_3_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_horizontal_line_3_5_scale "const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width"
+ prototype void vp8_vertical_band_3_5_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_last_vertical_band_3_5_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_horizontal_line_3_4_scale "const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width"
+ prototype void vp8_vertical_band_3_4_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_last_vertical_band_3_4_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_horizontal_line_1_2_scale "const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width"
+ prototype void vp8_vertical_band_1_2_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_last_vertical_band_1_2_scale "unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_horizontal_line_5_4_scale "const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width"
+ prototype void vp8_vertical_band_5_4_scale "unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_horizontal_line_5_3_scale "const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width"
+ prototype void vp8_vertical_band_5_3_scale "unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_horizontal_line_2_1_scale "const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width"
+ prototype void vp8_vertical_band_2_1_scale "unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+ prototype void vp8_vertical_band_2_1_scale_i "unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width"
+fi
+
+prototype void vp8_yv12_extend_frame_borders "struct yv12_buffer_config *ybf"
+specialize vp8_yv12_extend_frame_borders neon
+
+prototype void vp8_yv12_copy_frame "struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc"
+specialize vp8_yv12_copy_frame neon
+
+prototype void vp8_yv12_copy_y "struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc"
+specialize vp8_yv12_copy_y neon
+
diff --git a/libvpx/vp8/common/sad_c.c b/libvpx/vp8/common/sad_c.c
new file mode 100644
index 0000000..5f36fc9
--- /dev/null
+++ b/libvpx/vp8/common/sad_c.c
@@ -0,0 +1,302 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <limits.h>
+#include <stdlib.h>
+#include "vpx_config.h"
+#include "vpx/vpx_integer.h"
+
+static unsigned int sad_mx_n_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int max_sad, int m, int n)
+{
+ int r, c;
+ unsigned int sad = 0;
+
+ for (r = 0; r < n; r++)
+ {
+ for (c = 0; c < m; c++)
+ {
+ sad += abs(src_ptr[c] - ref_ptr[c]);
+ }
+
+ if (sad > max_sad)
+ break;
+
+ src_ptr += src_stride;
+ ref_ptr += ref_stride;
+ }
+
+ return sad;
+}
+
+/* max_sad is provided as an optional optimization point. Alternative
+ * implementations of these functions are not required to check it.
+ */
+
+unsigned int vp8_sad16x16_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int max_sad)
+{
+ return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 16, 16);
+}
+
+unsigned int vp8_sad8x8_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int max_sad)
+{
+ return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 8, 8);
+}
+
+unsigned int vp8_sad16x8_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int max_sad)
+{
+ return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 16, 8);
+
+}
+
+unsigned int vp8_sad8x16_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int max_sad)
+{
+ return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 8, 16);
+}
+
+unsigned int vp8_sad4x4_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int max_sad)
+{
+ return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 4, 4);
+}
+
+void vp8_sad16x16x3_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+}
+
+void vp8_sad16x16x8_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned short *sad_array)
+{
+ sad_array[0] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+ sad_array[3] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
+ sad_array[4] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
+ sad_array[5] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
+ sad_array[6] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
+ sad_array[7] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
+}
+
+void vp8_sad16x8x3_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+}
+
+void vp8_sad16x8x8_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned short *sad_array)
+{
+ sad_array[0] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+ sad_array[3] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
+ sad_array[4] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
+ sad_array[5] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
+ sad_array[6] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
+ sad_array[7] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
+}
+
+void vp8_sad8x8x3_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+}
+
+void vp8_sad8x8x8_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned short *sad_array)
+{
+ sad_array[0] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+ sad_array[3] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
+ sad_array[4] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
+ sad_array[5] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
+ sad_array[6] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
+ sad_array[7] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
+}
+
+void vp8_sad8x16x3_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+}
+
+void vp8_sad8x16x8_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned short *sad_array)
+{
+ sad_array[0] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+ sad_array[3] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
+ sad_array[4] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
+ sad_array[5] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
+ sad_array[6] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
+ sad_array[7] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
+}
+
+void vp8_sad4x4x3_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+}
+
+void vp8_sad4x4x8_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char *ref_ptr, int ref_stride,
+ unsigned short *sad_array)
+{
+ sad_array[0] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
+ sad_array[1] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
+ sad_array[2] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
+ sad_array[3] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
+ sad_array[4] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
+ sad_array[5] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
+ sad_array[6] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
+ sad_array[7] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
+}
+
+void vp8_sad16x16x4d_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char * const ref_ptr[], int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
+ sad_array[3] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
+}
+
+void vp8_sad16x8x4d_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char * const ref_ptr[], int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
+ sad_array[3] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
+}
+
+void vp8_sad8x8x4d_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char * const ref_ptr[], int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
+ sad_array[3] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
+}
+
+void vp8_sad8x16x4d_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char * const ref_ptr[], int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
+ sad_array[3] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
+}
+
+void vp8_sad4x4x4d_c(const unsigned char *src_ptr, int src_stride,
+ const unsigned char * const ref_ptr[], int ref_stride,
+ unsigned int *sad_array)
+{
+ sad_array[0] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
+ sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
+ sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
+ sad_array[3] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
+}
+
+/* Copy 2 macroblocks to a buffer */
+void vp8_copy32xn_c(unsigned char *src_ptr, int src_stride,
+ unsigned char *dst_ptr, int dst_stride,
+ int height)
+{
+ int r;
+
+ for (r = 0; r < height; r++)
+ {
+#if !(CONFIG_FAST_UNALIGNED)
+ dst_ptr[0] = src_ptr[0];
+ dst_ptr[1] = src_ptr[1];
+ dst_ptr[2] = src_ptr[2];
+ dst_ptr[3] = src_ptr[3];
+ dst_ptr[4] = src_ptr[4];
+ dst_ptr[5] = src_ptr[5];
+ dst_ptr[6] = src_ptr[6];
+ dst_ptr[7] = src_ptr[7];
+ dst_ptr[8] = src_ptr[8];
+ dst_ptr[9] = src_ptr[9];
+ dst_ptr[10] = src_ptr[10];
+ dst_ptr[11] = src_ptr[11];
+ dst_ptr[12] = src_ptr[12];
+ dst_ptr[13] = src_ptr[13];
+ dst_ptr[14] = src_ptr[14];
+ dst_ptr[15] = src_ptr[15];
+ dst_ptr[16] = src_ptr[16];
+ dst_ptr[17] = src_ptr[17];
+ dst_ptr[18] = src_ptr[18];
+ dst_ptr[19] = src_ptr[19];
+ dst_ptr[20] = src_ptr[20];
+ dst_ptr[21] = src_ptr[21];
+ dst_ptr[22] = src_ptr[22];
+ dst_ptr[23] = src_ptr[23];
+ dst_ptr[24] = src_ptr[24];
+ dst_ptr[25] = src_ptr[25];
+ dst_ptr[26] = src_ptr[26];
+ dst_ptr[27] = src_ptr[27];
+ dst_ptr[28] = src_ptr[28];
+ dst_ptr[29] = src_ptr[29];
+ dst_ptr[30] = src_ptr[30];
+ dst_ptr[31] = src_ptr[31];
+#else
+ ((uint32_t *)dst_ptr)[0] = ((uint32_t *)src_ptr)[0] ;
+ ((uint32_t *)dst_ptr)[1] = ((uint32_t *)src_ptr)[1] ;
+ ((uint32_t *)dst_ptr)[2] = ((uint32_t *)src_ptr)[2] ;
+ ((uint32_t *)dst_ptr)[3] = ((uint32_t *)src_ptr)[3] ;
+ ((uint32_t *)dst_ptr)[4] = ((uint32_t *)src_ptr)[4] ;
+ ((uint32_t *)dst_ptr)[5] = ((uint32_t *)src_ptr)[5] ;
+ ((uint32_t *)dst_ptr)[6] = ((uint32_t *)src_ptr)[6] ;
+ ((uint32_t *)dst_ptr)[7] = ((uint32_t *)src_ptr)[7] ;
+#endif
+ src_ptr += src_stride;
+ dst_ptr += dst_stride;
+
+ }
+}
diff --git a/libvpx/vp8/common/setupintrarecon.c b/libvpx/vp8/common/setupintrarecon.c
new file mode 100644
index 0000000..60afe51
--- /dev/null
+++ b/libvpx/vp8/common/setupintrarecon.c
@@ -0,0 +1,39 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "setupintrarecon.h"
+#include "vpx_mem/vpx_mem.h"
+
+void vp8_setup_intra_recon(YV12_BUFFER_CONFIG *ybf)
+{
+ int i;
+
+ /* set up frame new frame for intra coded blocks */
+ vpx_memset(ybf->y_buffer - 1 - ybf->y_stride, 127, ybf->y_width + 5);
+ for (i = 0; i < ybf->y_height; i++)
+ ybf->y_buffer[ybf->y_stride *i - 1] = (unsigned char) 129;
+
+ vpx_memset(ybf->u_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
+ for (i = 0; i < ybf->uv_height; i++)
+ ybf->u_buffer[ybf->uv_stride *i - 1] = (unsigned char) 129;
+
+ vpx_memset(ybf->v_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
+ for (i = 0; i < ybf->uv_height; i++)
+ ybf->v_buffer[ybf->uv_stride *i - 1] = (unsigned char) 129;
+
+}
+
+void vp8_setup_intra_recon_top_line(YV12_BUFFER_CONFIG *ybf)
+{
+ vpx_memset(ybf->y_buffer - 1 - ybf->y_stride, 127, ybf->y_width + 5);
+ vpx_memset(ybf->u_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
+ vpx_memset(ybf->v_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
+}
diff --git a/libvpx/vp8/common/setupintrarecon.h b/libvpx/vp8/common/setupintrarecon.h
new file mode 100644
index 0000000..e515c3a
--- /dev/null
+++ b/libvpx/vp8/common/setupintrarecon.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_scale/yv12config.h"
+extern void vp8_setup_intra_recon(YV12_BUFFER_CONFIG *ybf);
+extern void vp8_setup_intra_recon_top_line(YV12_BUFFER_CONFIG *ybf);
+
+static
+void setup_intra_recon_left(unsigned char *y_buffer,
+ unsigned char *u_buffer,
+ unsigned char *v_buffer,
+ int y_stride,
+ int uv_stride)
+{
+ int i;
+
+ for (i = 0; i < 16; i++)
+ y_buffer[y_stride *i] = (unsigned char) 129;
+
+ for (i = 0; i < 8; i++)
+ u_buffer[uv_stride *i] = (unsigned char) 129;
+
+ for (i = 0; i < 8; i++)
+ v_buffer[uv_stride *i] = (unsigned char) 129;
+}
diff --git a/vp8/common/swapyv12buffer.c b/libvpx/vp8/common/swapyv12buffer.c
similarity index 100%
rename from vp8/common/swapyv12buffer.c
rename to libvpx/vp8/common/swapyv12buffer.c
diff --git a/vp8/common/swapyv12buffer.h b/libvpx/vp8/common/swapyv12buffer.h
similarity index 100%
rename from vp8/common/swapyv12buffer.h
rename to libvpx/vp8/common/swapyv12buffer.h
diff --git a/libvpx/vp8/common/systemdependent.h b/libvpx/vp8/common/systemdependent.h
new file mode 100644
index 0000000..f99c4bb
--- /dev/null
+++ b/libvpx/vp8/common/systemdependent.h
@@ -0,0 +1,21 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#if ARCH_X86 || ARCH_X86_64
+void vpx_reset_mmx_state(void);
+#define vp8_clear_system_state() vpx_reset_mmx_state()
+#else
+#define vp8_clear_system_state()
+#endif
+
+struct VP8Common;
+void vp8_machine_specific_config(struct VP8Common *);
diff --git a/vp8/common/textblit.c b/libvpx/vp8/common/textblit.c
similarity index 100%
rename from vp8/common/textblit.c
rename to libvpx/vp8/common/textblit.c
diff --git a/libvpx/vp8/common/threading.h b/libvpx/vp8/common/threading.h
new file mode 100644
index 0000000..ed9e3e6
--- /dev/null
+++ b/libvpx/vp8/common/threading.h
@@ -0,0 +1,186 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef _PTHREAD_EMULATION
+#define _PTHREAD_EMULATION
+
+#if CONFIG_OS_SUPPORT && CONFIG_MULTITHREAD
+
+/* Thread management macros */
+#ifdef _WIN32
+/* Win32 */
+#include <process.h>
+#include <windows.h>
+#define THREAD_FUNCTION DWORD WINAPI
+#define THREAD_FUNCTION_RETURN DWORD
+#define THREAD_SPECIFIC_INDEX DWORD
+#define pthread_t HANDLE
+#define pthread_attr_t DWORD
+#define pthread_create(thhandle,attr,thfunc,tharg) (int)((*thhandle=(HANDLE)_beginthreadex(NULL,0,(unsigned int (__stdcall *)(void *))thfunc,tharg,0,NULL))==NULL)
+#define pthread_join(thread, result) ((WaitForSingleObject((thread),INFINITE)!=WAIT_OBJECT_0) || !CloseHandle(thread))
+#define pthread_detach(thread) if(thread!=NULL)CloseHandle(thread)
+#define thread_sleep(nms) Sleep(nms)
+#define pthread_cancel(thread) terminate_thread(thread,0)
+#define ts_key_create(ts_key, destructor) {ts_key = TlsAlloc();};
+#define pthread_getspecific(ts_key) TlsGetValue(ts_key)
+#define pthread_setspecific(ts_key, value) TlsSetValue(ts_key, (void *)value)
+#define pthread_self() GetCurrentThreadId()
+
+#elif defined(__OS2__)
+/* OS/2 */
+#define INCL_DOS
+#include <os2.h>
+
+#include <stdlib.h>
+#define THREAD_FUNCTION void
+#define THREAD_FUNCTION_RETURN void
+#define THREAD_SPECIFIC_INDEX PULONG
+#define pthread_t TID
+#define pthread_attr_t ULONG
+#define pthread_create(thhandle,attr,thfunc,tharg) \
+ ((int)((*(thhandle)=_beginthread(thfunc,NULL,1024*1024,tharg))==-1))
+#define pthread_join(thread, result) ((int)DosWaitThread(&(thread),0))
+#define pthread_detach(thread) 0
+#define thread_sleep(nms) DosSleep(nms)
+#define pthread_cancel(thread) DosKillThread(thread)
+#define ts_key_create(ts_key, destructor) \
+ DosAllocThreadLocalMemory(1, &(ts_key));
+#define pthread_getspecific(ts_key) ((void *)(*(ts_key)))
+#define pthread_setspecific(ts_key, value) (*(ts_key)=(ULONG)(value))
+#define pthread_self() _gettid()
+#else
+#ifdef __APPLE__
+#include <mach/mach_init.h>
+#include <mach/semaphore.h>
+#include <mach/task.h>
+#include <time.h>
+#include <unistd.h>
+
+#else
+#include <semaphore.h>
+#endif
+
+#include <pthread.h>
+/* pthreads */
+/* Nearly everything is already defined */
+#define THREAD_FUNCTION void *
+#define THREAD_FUNCTION_RETURN void *
+#define THREAD_SPECIFIC_INDEX pthread_key_t
+#define ts_key_create(ts_key, destructor) pthread_key_create (&(ts_key), destructor);
+#endif
+
+/* Syncrhronization macros: Win32 and Pthreads */
+#ifdef _WIN32
+#define sem_t HANDLE
+#define pause(voidpara) __asm PAUSE
+#define sem_init(sem, sem_attr1, sem_init_value) (int)((*sem = CreateSemaphore(NULL,0,32768,NULL))==NULL)
+#define sem_wait(sem) (int)(WAIT_OBJECT_0 != WaitForSingleObject(*sem,INFINITE))
+#define sem_post(sem) ReleaseSemaphore(*sem,1,NULL)
+#define sem_destroy(sem) if(*sem)((int)(CloseHandle(*sem))==TRUE)
+#define thread_sleep(nms) Sleep(nms)
+
+#elif defined(__OS2__)
+typedef struct
+{
+ HEV event;
+ HMTX wait_mutex;
+ HMTX count_mutex;
+ int count;
+} sem_t;
+
+static inline int sem_init(sem_t *sem, int pshared, unsigned int value)
+{
+ DosCreateEventSem(NULL, &sem->event, pshared ? DC_SEM_SHARED : 0,
+ value > 0 ? TRUE : FALSE);
+ DosCreateMutexSem(NULL, &sem->wait_mutex, 0, FALSE);
+ DosCreateMutexSem(NULL, &sem->count_mutex, 0, FALSE);
+
+ sem->count = value;
+
+ return 0;
+}
+
+static inline int sem_wait(sem_t * sem)
+{
+ DosRequestMutexSem(sem->wait_mutex, -1);
+
+ DosWaitEventSem(sem->event, -1);
+
+ DosRequestMutexSem(sem->count_mutex, -1);
+
+ sem->count--;
+ if (sem->count == 0)
+ {
+ ULONG post_count;
+
+ DosResetEventSem(sem->event, &post_count);
+ }
+
+ DosReleaseMutexSem(sem->count_mutex);
+
+ DosReleaseMutexSem(sem->wait_mutex);
+
+ return 0;
+}
+
+static inline int sem_post(sem_t * sem)
+{
+ DosRequestMutexSem(sem->count_mutex, -1);
+
+ if (sem->count < 32768)
+ {
+ sem->count++;
+ DosPostEventSem(sem->event);
+ }
+
+ DosReleaseMutexSem(sem->count_mutex);
+
+ return 0;
+}
+
+static inline int sem_destroy(sem_t * sem)
+{
+ DosCloseEventSem(sem->event);
+ DosCloseMutexSem(sem->wait_mutex);
+ DosCloseMutexSem(sem->count_mutex);
+
+ return 0;
+}
+
+#define thread_sleep(nms) DosSleep(nms)
+
+#else
+
+#ifdef __APPLE__
+#define sem_t semaphore_t
+#define sem_init(X,Y,Z) semaphore_create(mach_task_self(), X, SYNC_POLICY_FIFO, Z)
+#define sem_wait(sem) (semaphore_wait(*sem) )
+#define sem_post(sem) semaphore_signal(*sem)
+#define sem_destroy(sem) semaphore_destroy(mach_task_self(),*sem)
+#define thread_sleep(nms) /* { struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} */
+#else
+#include <unistd.h>
+#include <sched.h>
+#define thread_sleep(nms) sched_yield();/* {struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} */
+#endif
+/* Not Windows. Assume pthreads */
+
+#endif
+
+#if ARCH_X86 || ARCH_X86_64
+#include "vpx_ports/x86.h"
+#else
+#define x86_pause_hint()
+#endif
+
+#endif /* CONFIG_OS_SUPPORT && CONFIG_MULTITHREAD */
+
+#endif
diff --git a/vp8/common/treecoder.c b/libvpx/vp8/common/treecoder.c
similarity index 100%
rename from vp8/common/treecoder.c
rename to libvpx/vp8/common/treecoder.c
diff --git a/vp8/common/treecoder.h b/libvpx/vp8/common/treecoder.h
similarity index 100%
rename from vp8/common/treecoder.h
rename to libvpx/vp8/common/treecoder.h
diff --git a/libvpx/vp8/common/variance.h b/libvpx/vp8/common/variance.h
new file mode 100644
index 0000000..01193b8
--- /dev/null
+++ b/libvpx/vp8/common/variance.h
@@ -0,0 +1,115 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef VARIANCE_H
+#define VARIANCE_H
+
+#include "vpx_config.h"
+
+typedef unsigned int(*vp8_sad_fn_t)(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int ref_stride,
+ unsigned int max_sad);
+
+typedef void (*vp8_copy32xn_fn_t)(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int ref_stride,
+ int n);
+
+typedef void (*vp8_sad_multi_fn_t)(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int ref_stride,
+ unsigned int *sad_array);
+
+typedef void (*vp8_sad_multi1_fn_t)
+ (
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int ref_stride,
+ unsigned short *sad_array
+ );
+
+typedef void (*vp8_sad_multi_d_fn_t)
+ (
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char * const ref_ptr[],
+ int ref_stride,
+ unsigned int *sad_array
+ );
+
+typedef unsigned int (*vp8_variance_fn_t)
+ (
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int ref_stride,
+ unsigned int *sse
+ );
+
+typedef unsigned int (*vp8_subpixvariance_fn_t)
+ (
+ const unsigned char *src_ptr,
+ int source_stride,
+ int xoffset,
+ int yoffset,
+ const unsigned char *ref_ptr,
+ int Refstride,
+ unsigned int *sse
+ );
+
+typedef void (*vp8_ssimpf_fn_t)
+ (
+ unsigned char *s,
+ int sp,
+ unsigned char *r,
+ int rp,
+ unsigned long *sum_s,
+ unsigned long *sum_r,
+ unsigned long *sum_sq_s,
+ unsigned long *sum_sq_r,
+ unsigned long *sum_sxr
+ );
+
+typedef unsigned int (*vp8_getmbss_fn_t)(const short *);
+
+typedef unsigned int (*vp8_get16x16prederror_fn_t)
+ (
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int ref_stride
+ );
+
+typedef struct variance_vtable
+{
+ vp8_sad_fn_t sdf;
+ vp8_variance_fn_t vf;
+ vp8_subpixvariance_fn_t svf;
+ vp8_variance_fn_t svf_halfpix_h;
+ vp8_variance_fn_t svf_halfpix_v;
+ vp8_variance_fn_t svf_halfpix_hv;
+ vp8_sad_multi_fn_t sdx3f;
+ vp8_sad_multi1_fn_t sdx8f;
+ vp8_sad_multi_d_fn_t sdx4df;
+#if ARCH_X86 || ARCH_X86_64
+ vp8_copy32xn_fn_t copymem;
+#endif
+} vp8_variance_fn_ptr_t;
+
+#endif
diff --git a/libvpx/vp8/common/variance_c.c b/libvpx/vp8/common/variance_c.c
new file mode 100644
index 0000000..da08aff
--- /dev/null
+++ b/libvpx/vp8/common/variance_c.c
@@ -0,0 +1,458 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "variance.h"
+#include "filter.h"
+
+
+unsigned int vp8_get_mb_ss_c
+(
+ const short *src_ptr
+)
+{
+ unsigned int i = 0, sum = 0;
+
+ do
+ {
+ sum += (src_ptr[i] * src_ptr[i]);
+ i++;
+ }
+ while (i < 256);
+
+ return sum;
+}
+
+
+static void variance(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ int w,
+ int h,
+ unsigned int *sse,
+ int *sum)
+{
+ int i, j;
+ int diff;
+
+ *sum = 0;
+ *sse = 0;
+
+ for (i = 0; i < h; i++)
+ {
+ for (j = 0; j < w; j++)
+ {
+ diff = src_ptr[j] - ref_ptr[j];
+ *sum += diff;
+ *sse += diff * diff;
+ }
+
+ src_ptr += source_stride;
+ ref_ptr += recon_stride;
+ }
+}
+
+
+unsigned int vp8_variance16x16_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+
+ variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, &var, &avg);
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 8));
+}
+
+unsigned int vp8_variance8x16_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+
+ variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 16, &var, &avg);
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 7));
+}
+
+unsigned int vp8_variance16x8_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+
+ variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 8, &var, &avg);
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 7));
+}
+
+
+unsigned int vp8_variance8x8_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+
+ variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 8, &var, &avg);
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 6));
+}
+
+unsigned int vp8_variance4x4_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+
+ variance(src_ptr, source_stride, ref_ptr, recon_stride, 4, 4, &var, &avg);
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 4));
+}
+
+
+unsigned int vp8_mse16x16_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+ variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, &var, &avg);
+ *sse = var;
+ return var;
+}
+
+
+/****************************************************************************
+ *
+ * ROUTINE : filter_block2d_bil_first_pass
+ *
+ * INPUTS : UINT8 *src_ptr : Pointer to source block.
+ * UINT32 src_pixels_per_line : Stride of input block.
+ * UINT32 pixel_step : Offset between filter input samples (see notes).
+ * UINT32 output_height : Input block height.
+ * UINT32 output_width : Input block width.
+ * INT32 *vp8_filter : Array of 2 bi-linear filter taps.
+ *
+ * OUTPUTS : INT32 *output_ptr : Pointer to filtered block.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block in
+ * either horizontal or vertical direction to produce the
+ * filtered output block. Used to implement first-pass
+ * of 2-D separable filter.
+ *
+ * SPECIAL NOTES : Produces INT32 output to retain precision for next pass.
+ * Two filter taps should sum to VP8_FILTER_WEIGHT.
+ * pixel_step defines whether the filter is applied
+ * horizontally (pixel_step=1) or vertically (pixel_step=stride).
+ * It defines the offset required to move from one input
+ * to the next.
+ *
+ ****************************************************************************/
+static void var_filter_block2d_bil_first_pass
+(
+ const unsigned char *src_ptr,
+ unsigned short *output_ptr,
+ unsigned int src_pixels_per_line,
+ int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+)
+{
+ unsigned int i, j;
+
+ for (i = 0; i < output_height; i++)
+ {
+ for (j = 0; j < output_width; j++)
+ {
+ /* Apply bilinear filter */
+ output_ptr[j] = (((int)src_ptr[0] * vp8_filter[0]) +
+ ((int)src_ptr[pixel_step] * vp8_filter[1]) +
+ (VP8_FILTER_WEIGHT / 2)) >> VP8_FILTER_SHIFT;
+ src_ptr++;
+ }
+
+ /* Next row... */
+ src_ptr += src_pixels_per_line - output_width;
+ output_ptr += output_width;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : filter_block2d_bil_second_pass
+ *
+ * INPUTS : INT32 *src_ptr : Pointer to source block.
+ * UINT32 src_pixels_per_line : Stride of input block.
+ * UINT32 pixel_step : Offset between filter input samples (see notes).
+ * UINT32 output_height : Input block height.
+ * UINT32 output_width : Input block width.
+ * INT32 *vp8_filter : Array of 2 bi-linear filter taps.
+ *
+ * OUTPUTS : UINT16 *output_ptr : Pointer to filtered block.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block in
+ * either horizontal or vertical direction to produce the
+ * filtered output block. Used to implement second-pass
+ * of 2-D separable filter.
+ *
+ * SPECIAL NOTES : Requires 32-bit input as produced by filter_block2d_bil_first_pass.
+ * Two filter taps should sum to VP8_FILTER_WEIGHT.
+ * pixel_step defines whether the filter is applied
+ * horizontally (pixel_step=1) or vertically (pixel_step=stride).
+ * It defines the offset required to move from one input
+ * to the next.
+ *
+ ****************************************************************************/
+static void var_filter_block2d_bil_second_pass
+(
+ const unsigned short *src_ptr,
+ unsigned char *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+)
+{
+ unsigned int i, j;
+ int Temp;
+
+ for (i = 0; i < output_height; i++)
+ {
+ for (j = 0; j < output_width; j++)
+ {
+ /* Apply filter */
+ Temp = ((int)src_ptr[0] * vp8_filter[0]) +
+ ((int)src_ptr[pixel_step] * vp8_filter[1]) +
+ (VP8_FILTER_WEIGHT / 2);
+ output_ptr[j] = (unsigned int)(Temp >> VP8_FILTER_SHIFT);
+ src_ptr++;
+ }
+
+ /* Next row... */
+ src_ptr += src_pixels_per_line - output_width;
+ output_ptr += output_width;
+ }
+}
+
+
+unsigned int vp8_sub_pixel_variance4x4_c
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ unsigned char temp2[20*16];
+ const short *HFilter, *VFilter;
+ unsigned short FData3[5*4]; /* Temp data bufffer used in filtering */
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ /* First filter 1d Horizontal */
+ var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 5, 4, HFilter);
+
+ /* Now filter Verticaly */
+ var_filter_block2d_bil_second_pass(FData3, temp2, 4, 4, 4, 4, VFilter);
+
+ return vp8_variance4x4_c(temp2, 4, dst_ptr, dst_pixels_per_line, sse);
+}
+
+
+unsigned int vp8_sub_pixel_variance8x8_c
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ unsigned short FData3[9*8]; /* Temp data bufffer used in filtering */
+ unsigned char temp2[20*16];
+ const short *HFilter, *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 9, 8, HFilter);
+ var_filter_block2d_bil_second_pass(FData3, temp2, 8, 8, 8, 8, VFilter);
+
+ return vp8_variance8x8_c(temp2, 8, dst_ptr, dst_pixels_per_line, sse);
+}
+
+unsigned int vp8_sub_pixel_variance16x16_c
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ unsigned short FData3[17*16]; /* Temp data bufffer used in filtering */
+ unsigned char temp2[20*16];
+ const short *HFilter, *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 17, 16, HFilter);
+ var_filter_block2d_bil_second_pass(FData3, temp2, 16, 16, 16, 16, VFilter);
+
+ return vp8_variance16x16_c(temp2, 16, dst_ptr, dst_pixels_per_line, sse);
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_h_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 4, 0,
+ ref_ptr, recon_stride, sse);
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_v_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 0, 4,
+ ref_ptr, recon_stride, sse);
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_hv_c(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 4, 4,
+ ref_ptr, recon_stride, sse);
+}
+
+
+unsigned int vp8_sub_pixel_mse16x16_c
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ vp8_sub_pixel_variance16x16_c(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
+ return *sse;
+}
+
+unsigned int vp8_sub_pixel_variance16x8_c
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ unsigned short FData3[16*9]; /* Temp data bufffer used in filtering */
+ unsigned char temp2[20*16];
+ const short *HFilter, *VFilter;
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+ var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 9, 16, HFilter);
+ var_filter_block2d_bil_second_pass(FData3, temp2, 16, 16, 8, 16, VFilter);
+
+ return vp8_variance16x8_c(temp2, 16, dst_ptr, dst_pixels_per_line, sse);
+}
+
+unsigned int vp8_sub_pixel_variance8x16_c
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ unsigned short FData3[9*16]; /* Temp data bufffer used in filtering */
+ unsigned char temp2[20*16];
+ const short *HFilter, *VFilter;
+
+
+ HFilter = vp8_bilinear_filters[xoffset];
+ VFilter = vp8_bilinear_filters[yoffset];
+
+
+ var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 17, 8, HFilter);
+ var_filter_block2d_bil_second_pass(FData3, temp2, 8, 8, 16, 8, VFilter);
+
+ return vp8_variance8x16_c(temp2, 8, dst_ptr, dst_pixels_per_line, sse);
+}
diff --git a/libvpx/vp8/common/vp8_entropymodedata.h b/libvpx/vp8/common/vp8_entropymodedata.h
new file mode 100644
index 0000000..13e9a92
--- /dev/null
+++ b/libvpx/vp8/common/vp8_entropymodedata.h
@@ -0,0 +1,242 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+*/
+
+
+/*Generated file, included by entropymode.c*/
+
+
+const struct vp8_token_struct vp8_bmode_encodings[VP8_BINTRAMODES] =
+{
+ { 0, 1 },
+ { 2, 2 },
+ { 6, 3 },
+ { 28, 5 },
+ { 30, 5 },
+ { 58, 6 },
+ { 59, 6 },
+ { 62, 6 },
+ { 126, 7 },
+ { 127, 7 }
+};
+
+const struct vp8_token_struct vp8_ymode_encodings[VP8_YMODES] =
+{
+ { 0, 1 },
+ { 4, 3 },
+ { 5, 3 },
+ { 6, 3 },
+ { 7, 3 }
+};
+
+const struct vp8_token_struct vp8_kf_ymode_encodings[VP8_YMODES] =
+{
+ { 4, 3 },
+ { 5, 3 },
+ { 6, 3 },
+ { 7, 3 },
+ { 0, 1 }
+};
+
+const struct vp8_token_struct vp8_uv_mode_encodings[VP8_UV_MODES] =
+{
+ { 0, 1 },
+ { 2, 2 },
+ { 6, 3 },
+ { 7, 3 }
+};
+
+const struct vp8_token_struct vp8_mbsplit_encodings[VP8_NUMMBSPLITS] =
+{
+ { 6, 3 },
+ { 7, 3 },
+ { 2, 2 },
+ { 0, 1 }
+};
+
+const struct vp8_token_struct vp8_mv_ref_encoding_array[VP8_MVREFS] =
+{
+ { 2, 2 },
+ { 6, 3 },
+ { 0, 1 },
+ { 14, 4 },
+ { 15, 4 }
+};
+
+const struct vp8_token_struct vp8_sub_mv_ref_encoding_array[VP8_SUBMVREFS] =
+{
+ { 0, 1 },
+ { 2, 2 },
+ { 6, 3 },
+ { 7, 3 }
+};
+
+const struct vp8_token_struct vp8_small_mvencodings[8] =
+{
+ { 0, 3 },
+ { 1, 3 },
+ { 2, 3 },
+ { 3, 3 },
+ { 4, 3 },
+ { 5, 3 },
+ { 6, 3 },
+ { 7, 3 }
+};
+
+const vp8_prob vp8_ymode_prob[VP8_YMODES-1] =
+{
+ 112, 86, 140, 37
+};
+
+const vp8_prob vp8_kf_ymode_prob[VP8_YMODES-1] =
+{
+ 145, 156, 163, 128
+};
+
+const vp8_prob vp8_uv_mode_prob[VP8_UV_MODES-1] =
+{
+ 162, 101, 204
+};
+
+const vp8_prob vp8_kf_uv_mode_prob[VP8_UV_MODES-1] =
+{
+ 142, 114, 183
+};
+
+const vp8_prob vp8_bmode_prob[VP8_BINTRAMODES-1] =
+{
+ 120, 90, 79, 133, 87, 85, 80, 111, 151
+};
+
+
+
+const vp8_prob vp8_kf_bmode_prob
+[VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES-1] =
+{
+ {
+ { 231, 120, 48, 89, 115, 113, 120, 152, 112 },
+ { 152, 179, 64, 126, 170, 118, 46, 70, 95 },
+ { 175, 69, 143, 80, 85, 82, 72, 155, 103 },
+ { 56, 58, 10, 171, 218, 189, 17, 13, 152 },
+ { 144, 71, 10, 38, 171, 213, 144, 34, 26 },
+ { 114, 26, 17, 163, 44, 195, 21, 10, 173 },
+ { 121, 24, 80, 195, 26, 62, 44, 64, 85 },
+ { 170, 46, 55, 19, 136, 160, 33, 206, 71 },
+ { 63, 20, 8, 114, 114, 208, 12, 9, 226 },
+ { 81, 40, 11, 96, 182, 84, 29, 16, 36 }
+ },
+ {
+ { 134, 183, 89, 137, 98, 101, 106, 165, 148 },
+ { 72, 187, 100, 130, 157, 111, 32, 75, 80 },
+ { 66, 102, 167, 99, 74, 62, 40, 234, 128 },
+ { 41, 53, 9, 178, 241, 141, 26, 8, 107 },
+ { 104, 79, 12, 27, 217, 255, 87, 17, 7 },
+ { 74, 43, 26, 146, 73, 166, 49, 23, 157 },
+ { 65, 38, 105, 160, 51, 52, 31, 115, 128 },
+ { 87, 68, 71, 44, 114, 51, 15, 186, 23 },
+ { 47, 41, 14, 110, 182, 183, 21, 17, 194 },
+ { 66, 45, 25, 102, 197, 189, 23, 18, 22 }
+ },
+ {
+ { 88, 88, 147, 150, 42, 46, 45, 196, 205 },
+ { 43, 97, 183, 117, 85, 38, 35, 179, 61 },
+ { 39, 53, 200, 87, 26, 21, 43, 232, 171 },
+ { 56, 34, 51, 104, 114, 102, 29, 93, 77 },
+ { 107, 54, 32, 26, 51, 1, 81, 43, 31 },
+ { 39, 28, 85, 171, 58, 165, 90, 98, 64 },
+ { 34, 22, 116, 206, 23, 34, 43, 166, 73 },
+ { 68, 25, 106, 22, 64, 171, 36, 225, 114 },
+ { 34, 19, 21, 102, 132, 188, 16, 76, 124 },
+ { 62, 18, 78, 95, 85, 57, 50, 48, 51 }
+ },
+ {
+ { 193, 101, 35, 159, 215, 111, 89, 46, 111 },
+ { 60, 148, 31, 172, 219, 228, 21, 18, 111 },
+ { 112, 113, 77, 85, 179, 255, 38, 120, 114 },
+ { 40, 42, 1, 196, 245, 209, 10, 25, 109 },
+ { 100, 80, 8, 43, 154, 1, 51, 26, 71 },
+ { 88, 43, 29, 140, 166, 213, 37, 43, 154 },
+ { 61, 63, 30, 155, 67, 45, 68, 1, 209 },
+ { 142, 78, 78, 16, 255, 128, 34, 197, 171 },
+ { 41, 40, 5, 102, 211, 183, 4, 1, 221 },
+ { 51, 50, 17, 168, 209, 192, 23, 25, 82 }
+ },
+ {
+ { 125, 98, 42, 88, 104, 85, 117, 175, 82 },
+ { 95, 84, 53, 89, 128, 100, 113, 101, 45 },
+ { 75, 79, 123, 47, 51, 128, 81, 171, 1 },
+ { 57, 17, 5, 71, 102, 57, 53, 41, 49 },
+ { 115, 21, 2, 10, 102, 255, 166, 23, 6 },
+ { 38, 33, 13, 121, 57, 73, 26, 1, 85 },
+ { 41, 10, 67, 138, 77, 110, 90, 47, 114 },
+ { 101, 29, 16, 10, 85, 128, 101, 196, 26 },
+ { 57, 18, 10, 102, 102, 213, 34, 20, 43 },
+ { 117, 20, 15, 36, 163, 128, 68, 1, 26 }
+ },
+ {
+ { 138, 31, 36, 171, 27, 166, 38, 44, 229 },
+ { 67, 87, 58, 169, 82, 115, 26, 59, 179 },
+ { 63, 59, 90, 180, 59, 166, 93, 73, 154 },
+ { 40, 40, 21, 116, 143, 209, 34, 39, 175 },
+ { 57, 46, 22, 24, 128, 1, 54, 17, 37 },
+ { 47, 15, 16, 183, 34, 223, 49, 45, 183 },
+ { 46, 17, 33, 183, 6, 98, 15, 32, 183 },
+ { 65, 32, 73, 115, 28, 128, 23, 128, 205 },
+ { 40, 3, 9, 115, 51, 192, 18, 6, 223 },
+ { 87, 37, 9, 115, 59, 77, 64, 21, 47 }
+ },
+ {
+ { 104, 55, 44, 218, 9, 54, 53, 130, 226 },
+ { 64, 90, 70, 205, 40, 41, 23, 26, 57 },
+ { 54, 57, 112, 184, 5, 41, 38, 166, 213 },
+ { 30, 34, 26, 133, 152, 116, 10, 32, 134 },
+ { 75, 32, 12, 51, 192, 255, 160, 43, 51 },
+ { 39, 19, 53, 221, 26, 114, 32, 73, 255 },
+ { 31, 9, 65, 234, 2, 15, 1, 118, 73 },
+ { 88, 31, 35, 67, 102, 85, 55, 186, 85 },
+ { 56, 21, 23, 111, 59, 205, 45, 37, 192 },
+ { 55, 38, 70, 124, 73, 102, 1, 34, 98 }
+ },
+ {
+ { 102, 61, 71, 37, 34, 53, 31, 243, 192 },
+ { 69, 60, 71, 38, 73, 119, 28, 222, 37 },
+ { 68, 45, 128, 34, 1, 47, 11, 245, 171 },
+ { 62, 17, 19, 70, 146, 85, 55, 62, 70 },
+ { 75, 15, 9, 9, 64, 255, 184, 119, 16 },
+ { 37, 43, 37, 154, 100, 163, 85, 160, 1 },
+ { 63, 9, 92, 136, 28, 64, 32, 201, 85 },
+ { 86, 6, 28, 5, 64, 255, 25, 248, 1 },
+ { 56, 8, 17, 132, 137, 255, 55, 116, 128 },
+ { 58, 15, 20, 82, 135, 57, 26, 121, 40 }
+ },
+ {
+ { 164, 50, 31, 137, 154, 133, 25, 35, 218 },
+ { 51, 103, 44, 131, 131, 123, 31, 6, 158 },
+ { 86, 40, 64, 135, 148, 224, 45, 183, 128 },
+ { 22, 26, 17, 131, 240, 154, 14, 1, 209 },
+ { 83, 12, 13, 54, 192, 255, 68, 47, 28 },
+ { 45, 16, 21, 91, 64, 222, 7, 1, 197 },
+ { 56, 21, 39, 155, 60, 138, 23, 102, 213 },
+ { 85, 26, 85, 85, 128, 128, 32, 146, 171 },
+ { 18, 11, 7, 63, 144, 171, 4, 4, 246 },
+ { 35, 27, 10, 146, 174, 171, 12, 26, 128 }
+ },
+ {
+ { 190, 80, 35, 99, 180, 80, 126, 54, 45 },
+ { 85, 126, 47, 87, 176, 51, 41, 20, 32 },
+ { 101, 75, 128, 139, 118, 146, 116, 128, 85 },
+ { 56, 41, 15, 176, 236, 85, 37, 9, 62 },
+ { 146, 36, 19, 30, 171, 255, 97, 27, 20 },
+ { 71, 30, 17, 119, 118, 255, 17, 18, 138 },
+ { 101, 38, 60, 138, 55, 70, 43, 26, 142 },
+ { 138, 45, 61, 62, 219, 1, 81, 188, 64 },
+ { 32, 41, 20, 117, 151, 142, 20, 21, 163 },
+ { 112, 19, 12, 61, 195, 128, 48, 4, 24 }
+ }
+};
diff --git a/libvpx/vp8/common/x86/dequantize_mmx.asm b/libvpx/vp8/common/x86/dequantize_mmx.asm
new file mode 100644
index 0000000..4e551f0
--- /dev/null
+++ b/libvpx/vp8/common/x86/dequantize_mmx.asm
@@ -0,0 +1,258 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+
+;void vp8_dequantize_b_impl_mmx(short *sq, short *dq, short *q)
+global sym(vp8_dequantize_b_impl_mmx) PRIVATE
+sym(vp8_dequantize_b_impl_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 3
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;sq
+ mov rdi, arg(1) ;dq
+ mov rax, arg(2) ;q
+
+ movq mm1, [rsi]
+ pmullw mm1, [rax+0] ; mm4 *= kernel 0 modifiers.
+ movq [rdi], mm1
+
+ movq mm1, [rsi+8]
+ pmullw mm1, [rax+8] ; mm4 *= kernel 0 modifiers.
+ movq [rdi+8], mm1
+
+ movq mm1, [rsi+16]
+ pmullw mm1, [rax+16] ; mm4 *= kernel 0 modifiers.
+ movq [rdi+16], mm1
+
+ movq mm1, [rsi+24]
+ pmullw mm1, [rax+24] ; mm4 *= kernel 0 modifiers.
+ movq [rdi+24], mm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void dequant_idct_add_mmx(
+;short *input, 0
+;short *dq, 1
+;unsigned char *dest, 2
+;int stride) 3
+global sym(vp8_dequant_idct_add_mmx) PRIVATE
+sym(vp8_dequant_idct_add_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ GET_GOT rbx
+ push rdi
+ ; end prolog
+
+ mov rax, arg(0) ;input
+ mov rdx, arg(1) ;dq
+
+
+ movq mm0, [rax ]
+ pmullw mm0, [rdx]
+
+ movq mm1, [rax +8]
+ pmullw mm1, [rdx +8]
+
+ movq mm2, [rax+16]
+ pmullw mm2, [rdx+16]
+
+ movq mm3, [rax+24]
+ pmullw mm3, [rdx+24]
+
+ mov rdx, arg(2) ;dest
+
+ pxor mm7, mm7
+
+
+ movq [rax], mm7
+ movq [rax+8], mm7
+
+ movq [rax+16],mm7
+ movq [rax+24],mm7
+
+
+ movsxd rdi, dword ptr arg(3) ;stride
+
+ psubw mm0, mm2 ; b1= 0-2
+ paddw mm2, mm2 ;
+
+ movq mm5, mm1
+ paddw mm2, mm0 ; a1 =0+2
+
+ pmulhw mm5, [GLOBAL(x_s1sqr2)];
+ paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
+
+ movq mm7, mm3 ;
+ pmulhw mm7, [GLOBAL(x_c1sqr2less1)];
+
+ paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
+ psubw mm7, mm5 ; c1
+
+ movq mm5, mm1
+ movq mm4, mm3
+
+ pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
+ paddw mm5, mm1
+
+ pmulhw mm3, [GLOBAL(x_s1sqr2)]
+ paddw mm3, mm4
+
+ paddw mm3, mm5 ; d1
+ movq mm6, mm2 ; a1
+
+ movq mm4, mm0 ; b1
+ paddw mm2, mm3 ;0
+
+ paddw mm4, mm7 ;1
+ psubw mm0, mm7 ;2
+
+ psubw mm6, mm3 ;3
+
+ movq mm1, mm2 ; 03 02 01 00
+ movq mm3, mm4 ; 23 22 21 20
+
+ punpcklwd mm1, mm0 ; 11 01 10 00
+ punpckhwd mm2, mm0 ; 13 03 12 02
+
+ punpcklwd mm3, mm6 ; 31 21 30 20
+ punpckhwd mm4, mm6 ; 33 23 32 22
+
+ movq mm0, mm1 ; 11 01 10 00
+ movq mm5, mm2 ; 13 03 12 02
+
+ punpckldq mm0, mm3 ; 30 20 10 00
+ punpckhdq mm1, mm3 ; 31 21 11 01
+
+ punpckldq mm2, mm4 ; 32 22 12 02
+ punpckhdq mm5, mm4 ; 33 23 13 03
+
+ movq mm3, mm5 ; 33 23 13 03
+
+ psubw mm0, mm2 ; b1= 0-2
+ paddw mm2, mm2 ;
+
+ movq mm5, mm1
+ paddw mm2, mm0 ; a1 =0+2
+
+ pmulhw mm5, [GLOBAL(x_s1sqr2)];
+ paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
+
+ movq mm7, mm3 ;
+ pmulhw mm7, [GLOBAL(x_c1sqr2less1)];
+
+ paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
+ psubw mm7, mm5 ; c1
+
+ movq mm5, mm1
+ movq mm4, mm3
+
+ pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
+ paddw mm5, mm1
+
+ pmulhw mm3, [GLOBAL(x_s1sqr2)]
+ paddw mm3, mm4
+
+ paddw mm3, mm5 ; d1
+ paddw mm0, [GLOBAL(fours)]
+
+ paddw mm2, [GLOBAL(fours)]
+ movq mm6, mm2 ; a1
+
+ movq mm4, mm0 ; b1
+ paddw mm2, mm3 ;0
+
+ paddw mm4, mm7 ;1
+ psubw mm0, mm7 ;2
+
+ psubw mm6, mm3 ;3
+ psraw mm2, 3
+
+ psraw mm0, 3
+ psraw mm4, 3
+
+ psraw mm6, 3
+
+ movq mm1, mm2 ; 03 02 01 00
+ movq mm3, mm4 ; 23 22 21 20
+
+ punpcklwd mm1, mm0 ; 11 01 10 00
+ punpckhwd mm2, mm0 ; 13 03 12 02
+
+ punpcklwd mm3, mm6 ; 31 21 30 20
+ punpckhwd mm4, mm6 ; 33 23 32 22
+
+ movq mm0, mm1 ; 11 01 10 00
+ movq mm5, mm2 ; 13 03 12 02
+
+ punpckldq mm0, mm3 ; 30 20 10 00
+ punpckhdq mm1, mm3 ; 31 21 11 01
+
+ punpckldq mm2, mm4 ; 32 22 12 02
+ punpckhdq mm5, mm4 ; 33 23 13 03
+
+ pxor mm7, mm7
+
+ movd mm4, [rdx]
+ punpcklbw mm4, mm7
+ paddsw mm0, mm4
+ packuswb mm0, mm7
+ movd [rdx], mm0
+
+ movd mm4, [rdx+rdi]
+ punpcklbw mm4, mm7
+ paddsw mm1, mm4
+ packuswb mm1, mm7
+ movd [rdx+rdi], mm1
+
+ movd mm4, [rdx+2*rdi]
+ punpcklbw mm4, mm7
+ paddsw mm2, mm4
+ packuswb mm2, mm7
+ movd [rdx+rdi*2], mm2
+
+ add rdx, rdi
+
+ movd mm4, [rdx+2*rdi]
+ punpcklbw mm4, mm7
+ paddsw mm5, mm4
+ packuswb mm5, mm7
+ movd [rdx+rdi*2], mm5
+
+ ; begin epilog
+ pop rdi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+x_s1sqr2:
+ times 4 dw 0x8A8C
+align 16
+x_c1sqr2less1:
+ times 4 dw 0x4E7B
+align 16
+fours:
+ times 4 dw 0x0004
diff --git a/libvpx/vp8/common/x86/filter_x86.c b/libvpx/vp8/common/x86/filter_x86.c
new file mode 100644
index 0000000..ebab814
--- /dev/null
+++ b/libvpx/vp8/common/x86/filter_x86.c
@@ -0,0 +1,35 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_ports/mem.h"
+
+DECLARE_ALIGNED(16, const short, vp8_bilinear_filters_x86_4[8][8]) =
+{
+ { 128, 128, 128, 128, 0, 0, 0, 0 },
+ { 112, 112, 112, 112, 16, 16, 16, 16 },
+ { 96, 96, 96, 96, 32, 32, 32, 32 },
+ { 80, 80, 80, 80, 48, 48, 48, 48 },
+ { 64, 64, 64, 64, 64, 64, 64, 64 },
+ { 48, 48, 48, 48, 80, 80, 80, 80 },
+ { 32, 32, 32, 32, 96, 96, 96, 96 },
+ { 16, 16, 16, 16, 112, 112, 112, 112 }
+};
+
+DECLARE_ALIGNED(16, const short, vp8_bilinear_filters_x86_8[8][16]) =
+{
+ { 128, 128, 128, 128, 128, 128, 128, 128, 0, 0, 0, 0, 0, 0, 0, 0 },
+ { 112, 112, 112, 112, 112, 112, 112, 112, 16, 16, 16, 16, 16, 16, 16, 16 },
+ { 96, 96, 96, 96, 96, 96, 96, 96, 32, 32, 32, 32, 32, 32, 32, 32 },
+ { 80, 80, 80, 80, 80, 80, 80, 80, 48, 48, 48, 48, 48, 48, 48, 48 },
+ { 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64 },
+ { 48, 48, 48, 48, 48, 48, 48, 48, 80, 80, 80, 80, 80, 80, 80, 80 },
+ { 32, 32, 32, 32, 32, 32, 32, 32, 96, 96, 96, 96, 96, 96, 96, 96 },
+ { 16, 16, 16, 16, 16, 16, 16, 16, 112, 112, 112, 112, 112, 112, 112, 112 }
+};
diff --git a/libvpx/vp8/common/x86/filter_x86.h b/libvpx/vp8/common/x86/filter_x86.h
new file mode 100644
index 0000000..efcc4dc
--- /dev/null
+++ b/libvpx/vp8/common/x86/filter_x86.h
@@ -0,0 +1,19 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef FILTER_X86_H
+#define FILTER_X86_H
+
+/* x86 assembly specific copy of vp8/common/filter.c:vp8_bilinear_filters with
+ * duplicated values */
+extern const short vp8_bilinear_filters_x86_4[8][8]; /* duplicated 4x */
+extern const short vp8_bilinear_filters_x86_8[8][16]; /* duplicated 8x */
+
+#endif /* FILTER_X86_H */
diff --git a/libvpx/vp8/common/x86/idct_blk_mmx.c b/libvpx/vp8/common/x86/idct_blk_mmx.c
new file mode 100644
index 0000000..4adf3f5
--- /dev/null
+++ b/libvpx/vp8/common/x86/idct_blk_mmx.c
@@ -0,0 +1,127 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/common/blockd.h"
+
+extern void vp8_dequantize_b_impl_mmx(short *sq, short *dq, short *q);
+
+void vp8_dequantize_b_mmx(BLOCKD *d, short *DQC)
+{
+ short *sq = (short *) d->qcoeff;
+ short *dq = (short *) d->dqcoeff;
+
+ vp8_dequantize_b_impl_mmx(sq, dq, DQC);
+}
+
+void vp8_dequant_idct_add_y_block_mmx
+ (short *q, short *dq,
+ unsigned char *dst, int stride, char *eobs)
+{
+ int i;
+
+ for (i = 0; i < 4; i++)
+ {
+ if (eobs[0] > 1)
+ vp8_dequant_idct_add_mmx (q, dq, dst, stride);
+ else if (eobs[0] == 1)
+ {
+ vp8_dc_only_idct_add_mmx (q[0]*dq[0], dst, stride, dst, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ if (eobs[1] > 1)
+ vp8_dequant_idct_add_mmx (q+16, dq, dst+4, stride);
+ else if (eobs[1] == 1)
+ {
+ vp8_dc_only_idct_add_mmx (q[16]*dq[0], dst+4, stride,
+ dst+4, stride);
+ ((int *)(q+16))[0] = 0;
+ }
+
+ if (eobs[2] > 1)
+ vp8_dequant_idct_add_mmx (q+32, dq, dst+8, stride);
+ else if (eobs[2] == 1)
+ {
+ vp8_dc_only_idct_add_mmx (q[32]*dq[0], dst+8, stride,
+ dst+8, stride);
+ ((int *)(q+32))[0] = 0;
+ }
+
+ if (eobs[3] > 1)
+ vp8_dequant_idct_add_mmx (q+48, dq, dst+12, stride);
+ else if (eobs[3] == 1)
+ {
+ vp8_dc_only_idct_add_mmx (q[48]*dq[0], dst+12, stride,
+ dst+12, stride);
+ ((int *)(q+48))[0] = 0;
+ }
+
+ q += 64;
+ dst += 4*stride;
+ eobs += 4;
+ }
+}
+
+void vp8_dequant_idct_add_uv_block_mmx
+ (short *q, short *dq,
+ unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
+{
+ int i;
+
+ for (i = 0; i < 2; i++)
+ {
+ if (eobs[0] > 1)
+ vp8_dequant_idct_add_mmx (q, dq, dstu, stride);
+ else if (eobs[0] == 1)
+ {
+ vp8_dc_only_idct_add_mmx (q[0]*dq[0], dstu, stride, dstu, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ if (eobs[1] > 1)
+ vp8_dequant_idct_add_mmx (q+16, dq, dstu+4, stride);
+ else if (eobs[1] == 1)
+ {
+ vp8_dc_only_idct_add_mmx (q[16]*dq[0], dstu+4, stride,
+ dstu+4, stride);
+ ((int *)(q+16))[0] = 0;
+ }
+
+ q += 32;
+ dstu += 4*stride;
+ eobs += 2;
+ }
+
+ for (i = 0; i < 2; i++)
+ {
+ if (eobs[0] > 1)
+ vp8_dequant_idct_add_mmx (q, dq, dstv, stride);
+ else if (eobs[0] == 1)
+ {
+ vp8_dc_only_idct_add_mmx (q[0]*dq[0], dstv, stride, dstv, stride);
+ ((int *)q)[0] = 0;
+ }
+
+ if (eobs[1] > 1)
+ vp8_dequant_idct_add_mmx (q+16, dq, dstv+4, stride);
+ else if (eobs[1] == 1)
+ {
+ vp8_dc_only_idct_add_mmx (q[16]*dq[0], dstv+4, stride,
+ dstv+4, stride);
+ ((int *)(q+16))[0] = 0;
+ }
+
+ q += 32;
+ dstv += 4*stride;
+ eobs += 2;
+ }
+}
diff --git a/libvpx/vp8/common/x86/idct_blk_sse2.c b/libvpx/vp8/common/x86/idct_blk_sse2.c
new file mode 100644
index 0000000..056e052
--- /dev/null
+++ b/libvpx/vp8/common/x86/idct_blk_sse2.c
@@ -0,0 +1,89 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+
+void vp8_idct_dequant_0_2x_sse2
+ (short *q, short *dq ,
+ unsigned char *dst, int dst_stride);
+void vp8_idct_dequant_full_2x_sse2
+ (short *q, short *dq ,
+ unsigned char *dst, int dst_stride);
+
+void vp8_dequant_idct_add_y_block_sse2
+ (short *q, short *dq,
+ unsigned char *dst, int stride, char *eobs)
+{
+ int i;
+
+ for (i = 0; i < 4; i++)
+ {
+ if (((short *)(eobs))[0])
+ {
+ if (((short *)(eobs))[0] & 0xfefe)
+ vp8_idct_dequant_full_2x_sse2 (q, dq, dst, stride);
+ else
+ vp8_idct_dequant_0_2x_sse2 (q, dq, dst, stride);
+ }
+ if (((short *)(eobs))[1])
+ {
+ if (((short *)(eobs))[1] & 0xfefe)
+ vp8_idct_dequant_full_2x_sse2 (q+32, dq, dst+8, stride);
+ else
+ vp8_idct_dequant_0_2x_sse2 (q+32, dq, dst+8, stride);
+ }
+ q += 64;
+ dst += stride*4;
+ eobs += 4;
+ }
+}
+
+void vp8_dequant_idct_add_uv_block_sse2
+ (short *q, short *dq,
+ unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
+{
+ if (((short *)(eobs))[0])
+ {
+ if (((short *)(eobs))[0] & 0xfefe)
+ vp8_idct_dequant_full_2x_sse2 (q, dq, dstu, stride);
+ else
+ vp8_idct_dequant_0_2x_sse2 (q, dq, dstu, stride);
+ }
+ q += 32;
+ dstu += stride*4;
+
+ if (((short *)(eobs))[1])
+ {
+ if (((short *)(eobs))[1] & 0xfefe)
+ vp8_idct_dequant_full_2x_sse2 (q, dq, dstu, stride);
+ else
+ vp8_idct_dequant_0_2x_sse2 (q, dq, dstu, stride);
+ }
+ q += 32;
+
+ if (((short *)(eobs))[2])
+ {
+ if (((short *)(eobs))[2] & 0xfefe)
+ vp8_idct_dequant_full_2x_sse2 (q, dq, dstv, stride);
+ else
+ vp8_idct_dequant_0_2x_sse2 (q, dq, dstv, stride);
+ }
+ q += 32;
+ dstv += stride*4;
+
+ if (((short *)(eobs))[3])
+ {
+ if (((short *)(eobs))[3] & 0xfefe)
+ vp8_idct_dequant_full_2x_sse2 (q, dq, dstv, stride);
+ else
+ vp8_idct_dequant_0_2x_sse2 (q, dq, dstv, stride);
+ }
+}
diff --git a/libvpx/vp8/common/x86/idctllm_mmx.asm b/libvpx/vp8/common/x86/idctllm_mmx.asm
new file mode 100644
index 0000000..96fa2c6
--- /dev/null
+++ b/libvpx/vp8/common/x86/idctllm_mmx.asm
@@ -0,0 +1,295 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+; /****************************************************************************
+; * Notes:
+; *
+; * This implementation makes use of 16 bit fixed point version of two multiply
+; * constants:
+; * 1. sqrt(2) * cos (pi/8)
+; * 2. sqrt(2) * sin (pi/8)
+; * Because the first constant is bigger than 1, to maintain the same 16 bit
+; * fixed point precision as the second one, we use a trick of
+; * x * a = x + x*(a-1)
+; * so
+; * x * sqrt(2) * cos (pi/8) = x + x * (sqrt(2) *cos(pi/8)-1).
+; *
+; * For the second constant, because of the 16bit version is 35468, which
+; * is bigger than 32768, in signed 16 bit multiply, it becomes a negative
+; * number.
+; * (x * (unsigned)35468 >> 16) = x * (signed)35468 >> 16 + x
+; *
+; **************************************************************************/
+
+
+;void vp8_short_idct4x4llm_mmx(short *input, unsigned char *pred,
+;int pitch, unsigned char *dest,int stride)
+global sym(vp8_short_idct4x4llm_mmx) PRIVATE
+sym(vp8_short_idct4x4llm_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rax, arg(0) ;input
+ mov rsi, arg(1) ;pred
+
+ movq mm0, [rax ]
+ movq mm1, [rax+ 8]
+ movq mm2, [rax+16]
+ movq mm3, [rax+24]
+
+%if 0
+ pxor mm7, mm7
+ movq [rax], mm7
+ movq [rax+8], mm7
+ movq [rax+16],mm7
+ movq [rax+24],mm7
+%endif
+ movsxd rax, dword ptr arg(2) ;pitch
+ mov rdx, arg(3) ;dest
+ movsxd rdi, dword ptr arg(4) ;stride
+
+
+ psubw mm0, mm2 ; b1= 0-2
+ paddw mm2, mm2 ;
+
+ movq mm5, mm1
+ paddw mm2, mm0 ; a1 =0+2
+
+ pmulhw mm5, [GLOBAL(x_s1sqr2)];
+ paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
+
+ movq mm7, mm3 ;
+ pmulhw mm7, [GLOBAL(x_c1sqr2less1)];
+
+ paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
+ psubw mm7, mm5 ; c1
+
+ movq mm5, mm1
+ movq mm4, mm3
+
+ pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
+ paddw mm5, mm1
+
+ pmulhw mm3, [GLOBAL(x_s1sqr2)]
+ paddw mm3, mm4
+
+ paddw mm3, mm5 ; d1
+ movq mm6, mm2 ; a1
+
+ movq mm4, mm0 ; b1
+ paddw mm2, mm3 ;0
+
+ paddw mm4, mm7 ;1
+ psubw mm0, mm7 ;2
+
+ psubw mm6, mm3 ;3
+
+ movq mm1, mm2 ; 03 02 01 00
+ movq mm3, mm4 ; 23 22 21 20
+
+ punpcklwd mm1, mm0 ; 11 01 10 00
+ punpckhwd mm2, mm0 ; 13 03 12 02
+
+ punpcklwd mm3, mm6 ; 31 21 30 20
+ punpckhwd mm4, mm6 ; 33 23 32 22
+
+ movq mm0, mm1 ; 11 01 10 00
+ movq mm5, mm2 ; 13 03 12 02
+
+ punpckldq mm0, mm3 ; 30 20 10 00
+ punpckhdq mm1, mm3 ; 31 21 11 01
+
+ punpckldq mm2, mm4 ; 32 22 12 02
+ punpckhdq mm5, mm4 ; 33 23 13 03
+
+ movq mm3, mm5 ; 33 23 13 03
+
+ psubw mm0, mm2 ; b1= 0-2
+ paddw mm2, mm2 ;
+
+ movq mm5, mm1
+ paddw mm2, mm0 ; a1 =0+2
+
+ pmulhw mm5, [GLOBAL(x_s1sqr2)];
+ paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
+
+ movq mm7, mm3 ;
+ pmulhw mm7, [GLOBAL(x_c1sqr2less1)];
+
+ paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
+ psubw mm7, mm5 ; c1
+
+ movq mm5, mm1
+ movq mm4, mm3
+
+ pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
+ paddw mm5, mm1
+
+ pmulhw mm3, [GLOBAL(x_s1sqr2)]
+ paddw mm3, mm4
+
+ paddw mm3, mm5 ; d1
+ paddw mm0, [GLOBAL(fours)]
+
+ paddw mm2, [GLOBAL(fours)]
+ movq mm6, mm2 ; a1
+
+ movq mm4, mm0 ; b1
+ paddw mm2, mm3 ;0
+
+ paddw mm4, mm7 ;1
+ psubw mm0, mm7 ;2
+
+ psubw mm6, mm3 ;3
+ psraw mm2, 3
+
+ psraw mm0, 3
+ psraw mm4, 3
+
+ psraw mm6, 3
+
+ movq mm1, mm2 ; 03 02 01 00
+ movq mm3, mm4 ; 23 22 21 20
+
+ punpcklwd mm1, mm0 ; 11 01 10 00
+ punpckhwd mm2, mm0 ; 13 03 12 02
+
+ punpcklwd mm3, mm6 ; 31 21 30 20
+ punpckhwd mm4, mm6 ; 33 23 32 22
+
+ movq mm0, mm1 ; 11 01 10 00
+ movq mm5, mm2 ; 13 03 12 02
+
+ punpckldq mm0, mm3 ; 30 20 10 00
+ punpckhdq mm1, mm3 ; 31 21 11 01
+
+ punpckldq mm2, mm4 ; 32 22 12 02
+ punpckhdq mm5, mm4 ; 33 23 13 03
+
+ pxor mm7, mm7
+
+ movd mm4, [rsi]
+ punpcklbw mm4, mm7
+ paddsw mm0, mm4
+ packuswb mm0, mm7
+ movd [rdx], mm0
+
+ movd mm4, [rsi+rax]
+ punpcklbw mm4, mm7
+ paddsw mm1, mm4
+ packuswb mm1, mm7
+ movd [rdx+rdi], mm1
+
+ movd mm4, [rsi+2*rax]
+ punpcklbw mm4, mm7
+ paddsw mm2, mm4
+ packuswb mm2, mm7
+ movd [rdx+rdi*2], mm2
+
+ add rdx, rdi
+ add rsi, rax
+
+ movd mm4, [rsi+2*rax]
+ punpcklbw mm4, mm7
+ paddsw mm5, mm4
+ packuswb mm5, mm7
+ movd [rdx+rdi*2], mm5
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_dc_only_idct_add_mmx(
+;short input_dc,
+;unsigned char *pred_ptr,
+;int pred_stride,
+;unsigned char *dst_ptr,
+;int stride)
+global sym(vp8_dc_only_idct_add_mmx) PRIVATE
+sym(vp8_dc_only_idct_add_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ ; end prolog
+
+ movd mm5, arg(0) ;input_dc
+ mov rax, arg(1) ;pred_ptr
+ movsxd rdx, dword ptr arg(2) ;pred_stride
+
+ pxor mm0, mm0
+
+ paddw mm5, [GLOBAL(fours)]
+ lea rcx, [rdx + rdx*2]
+
+ psraw mm5, 3
+
+ punpcklwd mm5, mm5
+
+ punpckldq mm5, mm5
+
+ movd mm1, [rax]
+ movd mm2, [rax+rdx]
+ movd mm3, [rax+2*rdx]
+ movd mm4, [rax+rcx]
+
+ mov rax, arg(3) ;d -- destination
+ movsxd rdx, dword ptr arg(4) ;dst_stride
+
+ punpcklbw mm1, mm0
+ paddsw mm1, mm5
+ packuswb mm1, mm0 ; pack and unpack to saturate
+ lea rcx, [rdx + rdx*2]
+
+ punpcklbw mm2, mm0
+ paddsw mm2, mm5
+ packuswb mm2, mm0 ; pack and unpack to saturate
+
+ punpcklbw mm3, mm0
+ paddsw mm3, mm5
+ packuswb mm3, mm0 ; pack and unpack to saturate
+
+ punpcklbw mm4, mm0
+ paddsw mm4, mm5
+ packuswb mm4, mm0 ; pack and unpack to saturate
+
+ movd [rax], mm1
+ movd [rax+rdx], mm2
+ movd [rax+2*rdx], mm3
+ movd [rax+rcx], mm4
+
+ ; begin epilog
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+x_s1sqr2:
+ times 4 dw 0x8A8C
+align 16
+x_c1sqr2less1:
+ times 4 dw 0x4E7B
+align 16
+fours:
+ times 4 dw 0x0004
diff --git a/libvpx/vp8/common/x86/idctllm_sse2.asm b/libvpx/vp8/common/x86/idctllm_sse2.asm
new file mode 100644
index 0000000..bf8e2c4
--- /dev/null
+++ b/libvpx/vp8/common/x86/idctllm_sse2.asm
@@ -0,0 +1,708 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void vp8_idct_dequant_0_2x_sse2
+; (
+; short *qcoeff - 0
+; short *dequant - 1
+; unsigned char *dst - 2
+; int dst_stride - 3
+; )
+
+global sym(vp8_idct_dequant_0_2x_sse2) PRIVATE
+sym(vp8_idct_dequant_0_2x_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ GET_GOT rbx
+ ; end prolog
+
+ mov rdx, arg(1) ; dequant
+ mov rax, arg(0) ; qcoeff
+
+ movd xmm4, [rax]
+ movd xmm5, [rdx]
+
+ pinsrw xmm4, [rax+32], 4
+ pinsrw xmm5, [rdx], 4
+
+ pmullw xmm4, xmm5
+
+ ; Zero out xmm5, for use unpacking
+ pxor xmm5, xmm5
+
+ ; clear coeffs
+ movd [rax], xmm5
+ movd [rax+32], xmm5
+;pshufb
+ mov rax, arg(2) ; dst
+ movsxd rdx, dword ptr arg(3) ; dst_stride
+
+ pshuflw xmm4, xmm4, 00000000b
+ pshufhw xmm4, xmm4, 00000000b
+
+ lea rcx, [rdx + rdx*2]
+ paddw xmm4, [GLOBAL(fours)]
+
+ psraw xmm4, 3
+
+ movq xmm0, [rax]
+ movq xmm1, [rax+rdx]
+ movq xmm2, [rax+2*rdx]
+ movq xmm3, [rax+rcx]
+
+ punpcklbw xmm0, xmm5
+ punpcklbw xmm1, xmm5
+ punpcklbw xmm2, xmm5
+ punpcklbw xmm3, xmm5
+
+
+ ; Add to predict buffer
+ paddw xmm0, xmm4
+ paddw xmm1, xmm4
+ paddw xmm2, xmm4
+ paddw xmm3, xmm4
+
+ ; pack up before storing
+ packuswb xmm0, xmm5
+ packuswb xmm1, xmm5
+ packuswb xmm2, xmm5
+ packuswb xmm3, xmm5
+
+ ; store blocks back out
+ movq [rax], xmm0
+ movq [rax + rdx], xmm1
+
+ lea rax, [rax + 2*rdx]
+
+ movq [rax], xmm2
+ movq [rax + rdx], xmm3
+
+ ; begin epilog
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_idct_dequant_full_2x_sse2
+; (
+; short *qcoeff - 0
+; short *dequant - 1
+; unsigned char *dst - 2
+; int dst_stride - 3
+; )
+global sym(vp8_idct_dequant_full_2x_sse2) PRIVATE
+sym(vp8_idct_dequant_full_2x_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ; special case when 2 blocks have 0 or 1 coeffs
+ ; dc is set as first coeff, so no need to load qcoeff
+ mov rax, arg(0) ; qcoeff
+ mov rdx, arg(1) ; dequant
+ mov rdi, arg(2) ; dst
+
+
+ ; Zero out xmm7, for use unpacking
+ pxor xmm7, xmm7
+
+
+ ; note the transpose of xmm1 and xmm2, necessary for shuffle
+ ; to spit out sensicle data
+ movdqa xmm0, [rax]
+ movdqa xmm2, [rax+16]
+ movdqa xmm1, [rax+32]
+ movdqa xmm3, [rax+48]
+
+ ; Clear out coeffs
+ movdqa [rax], xmm7
+ movdqa [rax+16], xmm7
+ movdqa [rax+32], xmm7
+ movdqa [rax+48], xmm7
+
+ ; dequantize qcoeff buffer
+ pmullw xmm0, [rdx]
+ pmullw xmm2, [rdx+16]
+ pmullw xmm1, [rdx]
+ pmullw xmm3, [rdx+16]
+ movsxd rdx, dword ptr arg(3) ; dst_stride
+
+ ; repack so block 0 row x and block 1 row x are together
+ movdqa xmm4, xmm0
+ punpckldq xmm0, xmm1
+ punpckhdq xmm4, xmm1
+
+ pshufd xmm0, xmm0, 11011000b
+ pshufd xmm1, xmm4, 11011000b
+
+ movdqa xmm4, xmm2
+ punpckldq xmm2, xmm3
+ punpckhdq xmm4, xmm3
+
+ pshufd xmm2, xmm2, 11011000b
+ pshufd xmm3, xmm4, 11011000b
+
+ ; first pass
+ psubw xmm0, xmm2 ; b1 = 0-2
+ paddw xmm2, xmm2 ;
+
+ movdqa xmm5, xmm1
+ paddw xmm2, xmm0 ; a1 = 0+2
+
+ pmulhw xmm5, [GLOBAL(x_s1sqr2)]
+ lea rcx, [rdx + rdx*2] ;dst_stride * 3
+ paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2)
+
+ movdqa xmm7, xmm3
+ pmulhw xmm7, [GLOBAL(x_c1sqr2less1)]
+
+ paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2)
+ psubw xmm7, xmm5 ; c1
+
+ movdqa xmm5, xmm1
+ movdqa xmm4, xmm3
+
+ pmulhw xmm5, [GLOBAL(x_c1sqr2less1)]
+ paddw xmm5, xmm1
+
+ pmulhw xmm3, [GLOBAL(x_s1sqr2)]
+ paddw xmm3, xmm4
+
+ paddw xmm3, xmm5 ; d1
+ movdqa xmm6, xmm2 ; a1
+
+ movdqa xmm4, xmm0 ; b1
+ paddw xmm2, xmm3 ;0
+
+ paddw xmm4, xmm7 ;1
+ psubw xmm0, xmm7 ;2
+
+ psubw xmm6, xmm3 ;3
+
+ ; transpose for the second pass
+ movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000
+ punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000
+ punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100
+
+ movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008
+ punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008
+ punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108
+
+
+ movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000
+ punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000
+ punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002
+
+ movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100
+ punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100
+ punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102
+
+
+ movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000
+ punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000
+ punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001
+
+ movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002
+ punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002
+ punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003
+
+ pshufd xmm0, xmm2, 11011000b
+ pshufd xmm2, xmm1, 11011000b
+
+ pshufd xmm1, xmm5, 11011000b
+ pshufd xmm3, xmm7, 11011000b
+
+ ; second pass
+ psubw xmm0, xmm2 ; b1 = 0-2
+ paddw xmm2, xmm2
+
+ movdqa xmm5, xmm1
+ paddw xmm2, xmm0 ; a1 = 0+2
+
+ pmulhw xmm5, [GLOBAL(x_s1sqr2)]
+ paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2)
+
+ movdqa xmm7, xmm3
+ pmulhw xmm7, [GLOBAL(x_c1sqr2less1)]
+
+ paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2)
+ psubw xmm7, xmm5 ; c1
+
+ movdqa xmm5, xmm1
+ movdqa xmm4, xmm3
+
+ pmulhw xmm5, [GLOBAL(x_c1sqr2less1)]
+ paddw xmm5, xmm1
+
+ pmulhw xmm3, [GLOBAL(x_s1sqr2)]
+ paddw xmm3, xmm4
+
+ paddw xmm3, xmm5 ; d1
+ paddw xmm0, [GLOBAL(fours)]
+
+ paddw xmm2, [GLOBAL(fours)]
+ movdqa xmm6, xmm2 ; a1
+
+ movdqa xmm4, xmm0 ; b1
+ paddw xmm2, xmm3 ;0
+
+ paddw xmm4, xmm7 ;1
+ psubw xmm0, xmm7 ;2
+
+ psubw xmm6, xmm3 ;3
+ psraw xmm2, 3
+
+ psraw xmm0, 3
+ psraw xmm4, 3
+
+ psraw xmm6, 3
+
+ ; transpose to save
+ movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000
+ punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000
+ punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100
+
+ movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008
+ punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008
+ punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108
+
+
+ movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000
+ punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000
+ punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002
+
+ movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100
+ punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100
+ punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102
+
+
+ movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000
+ punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000
+ punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001
+
+ movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002
+ punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002
+ punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003
+
+ pshufd xmm0, xmm2, 11011000b
+ pshufd xmm2, xmm1, 11011000b
+
+ pshufd xmm1, xmm5, 11011000b
+ pshufd xmm3, xmm7, 11011000b
+
+ pxor xmm7, xmm7
+
+ ; Load up predict blocks
+ movq xmm4, [rdi]
+ movq xmm5, [rdi+rdx]
+
+ punpcklbw xmm4, xmm7
+ punpcklbw xmm5, xmm7
+
+ paddw xmm0, xmm4
+ paddw xmm1, xmm5
+
+ movq xmm4, [rdi+2*rdx]
+ movq xmm5, [rdi+rcx]
+
+ punpcklbw xmm4, xmm7
+ punpcklbw xmm5, xmm7
+
+ paddw xmm2, xmm4
+ paddw xmm3, xmm5
+
+.finish:
+
+ ; pack up before storing
+ packuswb xmm0, xmm7
+ packuswb xmm1, xmm7
+ packuswb xmm2, xmm7
+ packuswb xmm3, xmm7
+
+ ; store blocks back out
+ movq [rdi], xmm0
+ movq [rdi + rdx], xmm1
+ movq [rdi + rdx*2], xmm2
+ movq [rdi + rcx], xmm3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_idct_dequant_dc_0_2x_sse2
+; (
+; short *qcoeff - 0
+; short *dequant - 1
+; unsigned char *dst - 2
+; int dst_stride - 3
+; short *dc - 4
+; )
+global sym(vp8_idct_dequant_dc_0_2x_sse2) PRIVATE
+sym(vp8_idct_dequant_dc_0_2x_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rdi
+ ; end prolog
+
+ ; special case when 2 blocks have 0 or 1 coeffs
+ ; dc is set as first coeff, so no need to load qcoeff
+ mov rax, arg(0) ; qcoeff
+
+ mov rdi, arg(2) ; dst
+ mov rdx, arg(4) ; dc
+
+ ; Zero out xmm5, for use unpacking
+ pxor xmm5, xmm5
+
+ ; load up 2 dc words here == 2*16 = doubleword
+ movd xmm4, [rdx]
+
+ movsxd rdx, dword ptr arg(3) ; dst_stride
+ lea rcx, [rdx + rdx*2]
+ ; Load up predict blocks
+ movq xmm0, [rdi]
+ movq xmm1, [rdi+rdx*1]
+ movq xmm2, [rdi+rdx*2]
+ movq xmm3, [rdi+rcx]
+
+ ; Duplicate and expand dc across
+ punpcklwd xmm4, xmm4
+ punpckldq xmm4, xmm4
+
+ ; Rounding to dequant and downshift
+ paddw xmm4, [GLOBAL(fours)]
+ psraw xmm4, 3
+
+ ; Predict buffer needs to be expanded from bytes to words
+ punpcklbw xmm0, xmm5
+ punpcklbw xmm1, xmm5
+ punpcklbw xmm2, xmm5
+ punpcklbw xmm3, xmm5
+
+ ; Add to predict buffer
+ paddw xmm0, xmm4
+ paddw xmm1, xmm4
+ paddw xmm2, xmm4
+ paddw xmm3, xmm4
+
+ ; pack up before storing
+ packuswb xmm0, xmm5
+ packuswb xmm1, xmm5
+ packuswb xmm2, xmm5
+ packuswb xmm3, xmm5
+
+ ; store blocks back out
+ movq [rdi], xmm0
+ movq [rdi + rdx], xmm1
+ movq [rdi + rdx*2], xmm2
+ movq [rdi + rcx], xmm3
+
+ ; begin epilog
+ pop rdi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+;void vp8_idct_dequant_dc_full_2x_sse2
+; (
+; short *qcoeff - 0
+; short *dequant - 1
+; unsigned char *dst - 2
+; int dst_stride - 3
+; short *dc - 4
+; )
+global sym(vp8_idct_dequant_dc_full_2x_sse2) PRIVATE
+sym(vp8_idct_dequant_dc_full_2x_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rdi
+ ; end prolog
+
+ ; special case when 2 blocks have 0 or 1 coeffs
+ ; dc is set as first coeff, so no need to load qcoeff
+ mov rax, arg(0) ; qcoeff
+ mov rdx, arg(1) ; dequant
+
+ mov rdi, arg(2) ; dst
+
+ ; Zero out xmm7, for use unpacking
+ pxor xmm7, xmm7
+
+
+ ; note the transpose of xmm1 and xmm2, necessary for shuffle
+ ; to spit out sensicle data
+ movdqa xmm0, [rax]
+ movdqa xmm2, [rax+16]
+ movdqa xmm1, [rax+32]
+ movdqa xmm3, [rax+48]
+
+ ; Clear out coeffs
+ movdqa [rax], xmm7
+ movdqa [rax+16], xmm7
+ movdqa [rax+32], xmm7
+ movdqa [rax+48], xmm7
+
+ ; dequantize qcoeff buffer
+ pmullw xmm0, [rdx]
+ pmullw xmm2, [rdx+16]
+ pmullw xmm1, [rdx]
+ pmullw xmm3, [rdx+16]
+
+ ; DC component
+ mov rdx, arg(4)
+
+ ; repack so block 0 row x and block 1 row x are together
+ movdqa xmm4, xmm0
+ punpckldq xmm0, xmm1
+ punpckhdq xmm4, xmm1
+
+ pshufd xmm0, xmm0, 11011000b
+ pshufd xmm1, xmm4, 11011000b
+
+ movdqa xmm4, xmm2
+ punpckldq xmm2, xmm3
+ punpckhdq xmm4, xmm3
+
+ pshufd xmm2, xmm2, 11011000b
+ pshufd xmm3, xmm4, 11011000b
+
+ ; insert DC component
+ pinsrw xmm0, [rdx], 0
+ pinsrw xmm0, [rdx+2], 4
+
+ ; first pass
+ psubw xmm0, xmm2 ; b1 = 0-2
+ paddw xmm2, xmm2 ;
+
+ movdqa xmm5, xmm1
+ paddw xmm2, xmm0 ; a1 = 0+2
+
+ pmulhw xmm5, [GLOBAL(x_s1sqr2)]
+ paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2)
+
+ movdqa xmm7, xmm3
+ pmulhw xmm7, [GLOBAL(x_c1sqr2less1)]
+
+ paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2)
+ psubw xmm7, xmm5 ; c1
+
+ movdqa xmm5, xmm1
+ movdqa xmm4, xmm3
+
+ pmulhw xmm5, [GLOBAL(x_c1sqr2less1)]
+ paddw xmm5, xmm1
+
+ pmulhw xmm3, [GLOBAL(x_s1sqr2)]
+ paddw xmm3, xmm4
+
+ paddw xmm3, xmm5 ; d1
+ movdqa xmm6, xmm2 ; a1
+
+ movdqa xmm4, xmm0 ; b1
+ paddw xmm2, xmm3 ;0
+
+ paddw xmm4, xmm7 ;1
+ psubw xmm0, xmm7 ;2
+
+ psubw xmm6, xmm3 ;3
+
+ ; transpose for the second pass
+ movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000
+ punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000
+ punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100
+
+ movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008
+ punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008
+ punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108
+
+
+ movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000
+ punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000
+ punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002
+
+ movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100
+ punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100
+ punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102
+
+
+ movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000
+ punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000
+ punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001
+
+ movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002
+ punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002
+ punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003
+
+ pshufd xmm0, xmm2, 11011000b
+ pshufd xmm2, xmm1, 11011000b
+
+ pshufd xmm1, xmm5, 11011000b
+ pshufd xmm3, xmm7, 11011000b
+
+ ; second pass
+ psubw xmm0, xmm2 ; b1 = 0-2
+ paddw xmm2, xmm2
+
+ movdqa xmm5, xmm1
+ paddw xmm2, xmm0 ; a1 = 0+2
+
+ pmulhw xmm5, [GLOBAL(x_s1sqr2)]
+ paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2)
+
+ movdqa xmm7, xmm3
+ pmulhw xmm7, [GLOBAL(x_c1sqr2less1)]
+
+ paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2)
+ psubw xmm7, xmm5 ; c1
+
+ movdqa xmm5, xmm1
+ movdqa xmm4, xmm3
+
+ pmulhw xmm5, [GLOBAL(x_c1sqr2less1)]
+ paddw xmm5, xmm1
+
+ pmulhw xmm3, [GLOBAL(x_s1sqr2)]
+ paddw xmm3, xmm4
+
+ paddw xmm3, xmm5 ; d1
+ paddw xmm0, [GLOBAL(fours)]
+
+ paddw xmm2, [GLOBAL(fours)]
+ movdqa xmm6, xmm2 ; a1
+
+ movdqa xmm4, xmm0 ; b1
+ paddw xmm2, xmm3 ;0
+
+ paddw xmm4, xmm7 ;1
+ psubw xmm0, xmm7 ;2
+
+ psubw xmm6, xmm3 ;3
+ psraw xmm2, 3
+
+ psraw xmm0, 3
+ psraw xmm4, 3
+
+ psraw xmm6, 3
+
+ ; transpose to save
+ movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000
+ punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000
+ punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100
+
+ movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008
+ punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008
+ punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108
+
+
+ movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000
+ punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000
+ punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002
+
+ movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100
+ punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100
+ punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102
+
+
+ movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000
+ punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000
+ punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001
+
+ movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002
+ punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002
+ punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003
+
+ pshufd xmm0, xmm2, 11011000b
+ pshufd xmm2, xmm1, 11011000b
+
+ pshufd xmm1, xmm5, 11011000b
+ pshufd xmm3, xmm7, 11011000b
+
+ pxor xmm7, xmm7
+
+ ; Load up predict blocks
+ movsxd rdx, dword ptr arg(3) ; dst_stride
+ movq xmm4, [rdi]
+ movq xmm5, [rdi+rdx]
+ lea rcx, [rdx + rdx*2]
+
+ punpcklbw xmm4, xmm7
+ punpcklbw xmm5, xmm7
+
+ paddw xmm0, xmm4
+ paddw xmm1, xmm5
+
+ movq xmm4, [rdi+rdx*2]
+ movq xmm5, [rdi+rcx]
+
+ punpcklbw xmm4, xmm7
+ punpcklbw xmm5, xmm7
+
+ paddw xmm2, xmm4
+ paddw xmm3, xmm5
+
+.finish:
+
+ ; pack up before storing
+ packuswb xmm0, xmm7
+ packuswb xmm1, xmm7
+ packuswb xmm2, xmm7
+ packuswb xmm3, xmm7
+
+ ; Load destination stride before writing out,
+ ; doesn't need to persist
+ movsxd rdx, dword ptr arg(3) ; dst_stride
+
+ ; store blocks back out
+ movq [rdi], xmm0
+ movq [rdi + rdx], xmm1
+
+ lea rdi, [rdi + 2*rdx]
+
+ movq [rdi], xmm2
+ movq [rdi + rdx], xmm3
+
+
+ ; begin epilog
+ pop rdi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+fours:
+ times 8 dw 0x0004
+align 16
+x_s1sqr2:
+ times 8 dw 0x8A8C
+align 16
+x_c1sqr2less1:
+ times 8 dw 0x4E7B
diff --git a/libvpx/vp8/common/x86/iwalsh_mmx.asm b/libvpx/vp8/common/x86/iwalsh_mmx.asm
new file mode 100644
index 0000000..4aac094
--- /dev/null
+++ b/libvpx/vp8/common/x86/iwalsh_mmx.asm
@@ -0,0 +1,140 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void vp8_short_inv_walsh4x4_mmx(short *input, short *output)
+global sym(vp8_short_inv_walsh4x4_mmx) PRIVATE
+sym(vp8_short_inv_walsh4x4_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 2
+ ; end prolog
+
+ mov rdx, arg(0)
+ mov rax, 30003h
+
+ movq mm0, [rdx + 0] ;ip[0]
+ movq mm1, [rdx + 8] ;ip[4]
+ movd mm7, rax
+
+ movq mm2, [rdx + 16] ;ip[8]
+ movq mm3, [rdx + 24] ;ip[12]
+ punpcklwd mm7, mm7 ;0003000300030003h
+ mov rdx, arg(1)
+
+ movq mm4, mm0
+ movq mm5, mm1
+
+ paddw mm4, mm3 ;ip[0] + ip[12] aka al
+ paddw mm5, mm2 ;ip[4] + ip[8] aka bl
+
+ movq mm6, mm4 ;temp al
+ paddw mm4, mm5 ;al + bl
+ psubw mm6, mm5 ;al - bl
+
+ psubw mm0, mm3 ;ip[0] - ip[12] aka d1
+ psubw mm1, mm2 ;ip[4] - ip[8] aka c1
+
+ movq mm5, mm0 ;temp dl
+ paddw mm0, mm1 ;dl + cl
+ psubw mm5, mm1 ;dl - cl
+
+ ; 03 02 01 00
+ ; 13 12 11 10
+ ; 23 22 21 20
+ ; 33 32 31 30
+
+ movq mm3, mm4 ; 03 02 01 00
+ punpcklwd mm4, mm0 ; 11 01 10 00
+ punpckhwd mm3, mm0 ; 13 03 12 02
+
+ movq mm1, mm6 ; 23 22 21 20
+ punpcklwd mm6, mm5 ; 31 21 30 20
+ punpckhwd mm1, mm5 ; 33 23 32 22
+
+ movq mm0, mm4 ; 11 01 10 00
+ movq mm2, mm3 ; 13 03 12 02
+
+ punpckldq mm0, mm6 ; 30 20 10 00 aka ip[0]
+ punpckhdq mm4, mm6 ; 31 21 11 01 aka ip[4]
+
+ punpckldq mm2, mm1 ; 32 22 12 02 aka ip[8]
+ punpckhdq mm3, mm1 ; 33 23 13 03 aka ip[12]
+;~~~~~~~~~~~~~~~~~~~~~
+ movq mm1, mm0
+ movq mm5, mm4
+ paddw mm1, mm3 ;ip[0] + ip[12] aka al
+ paddw mm5, mm2 ;ip[4] + ip[8] aka bl
+
+ movq mm6, mm1 ;temp al
+ paddw mm1, mm5 ;al + bl
+ psubw mm6, mm5 ;al - bl
+ paddw mm1, mm7
+ paddw mm6, mm7
+ psraw mm1, 3
+ psraw mm6, 3
+
+ psubw mm0, mm3 ;ip[0] - ip[12] aka d1
+ psubw mm4, mm2 ;ip[4] - ip[8] aka c1
+
+ movq mm5, mm0 ;temp dl
+ paddw mm0, mm4 ;dl + cl
+ psubw mm5, mm4 ;dl - cl
+ paddw mm0, mm7
+ paddw mm5, mm7
+ psraw mm0, 3
+ psraw mm5, 3
+;~~~~~~~~~~~~~~~~~~~~~
+
+ movd eax, mm1
+ movd ecx, mm0
+ psrlq mm0, 32
+ psrlq mm1, 32
+ mov word ptr[rdx+32*0], ax
+ mov word ptr[rdx+32*1], cx
+ shr eax, 16
+ shr ecx, 16
+ mov word ptr[rdx+32*4], ax
+ mov word ptr[rdx+32*5], cx
+ movd eax, mm1
+ movd ecx, mm0
+ mov word ptr[rdx+32*8], ax
+ mov word ptr[rdx+32*9], cx
+ shr eax, 16
+ shr ecx, 16
+ mov word ptr[rdx+32*12], ax
+ mov word ptr[rdx+32*13], cx
+
+ movd eax, mm6
+ movd ecx, mm5
+ psrlq mm5, 32
+ psrlq mm6, 32
+ mov word ptr[rdx+32*2], ax
+ mov word ptr[rdx+32*3], cx
+ shr eax, 16
+ shr ecx, 16
+ mov word ptr[rdx+32*6], ax
+ mov word ptr[rdx+32*7], cx
+ movd eax, mm6
+ movd ecx, mm5
+ mov word ptr[rdx+32*10], ax
+ mov word ptr[rdx+32*11], cx
+ shr eax, 16
+ shr ecx, 16
+ mov word ptr[rdx+32*14], ax
+ mov word ptr[rdx+32*15], cx
+
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
diff --git a/libvpx/vp8/common/x86/iwalsh_sse2.asm b/libvpx/vp8/common/x86/iwalsh_sse2.asm
new file mode 100644
index 0000000..06e86a8
--- /dev/null
+++ b/libvpx/vp8/common/x86/iwalsh_sse2.asm
@@ -0,0 +1,121 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void vp8_short_inv_walsh4x4_sse2(short *input, short *output)
+global sym(vp8_short_inv_walsh4x4_sse2) PRIVATE
+sym(vp8_short_inv_walsh4x4_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 2
+ ; end prolog
+
+ mov rcx, arg(0)
+ mov rdx, arg(1)
+ mov rax, 30003h
+
+ movdqa xmm0, [rcx + 0] ;ip[4] ip[0]
+ movdqa xmm1, [rcx + 16] ;ip[12] ip[8]
+
+
+ pshufd xmm2, xmm1, 4eh ;ip[8] ip[12]
+ movdqa xmm3, xmm0 ;ip[4] ip[0]
+
+ paddw xmm0, xmm2 ;ip[4]+ip[8] ip[0]+ip[12] aka b1 a1
+ psubw xmm3, xmm2 ;ip[4]-ip[8] ip[0]-ip[12] aka c1 d1
+
+ movdqa xmm4, xmm0
+ punpcklqdq xmm0, xmm3 ;d1 a1
+ punpckhqdq xmm4, xmm3 ;c1 b1
+
+ movdqa xmm1, xmm4 ;c1 b1
+ paddw xmm4, xmm0 ;dl+cl a1+b1 aka op[4] op[0]
+ psubw xmm0, xmm1 ;d1-c1 a1-b1 aka op[12] op[8]
+
+ ;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ ; 13 12 11 10 03 02 01 00
+ ;
+ ; 33 32 31 30 23 22 21 20
+ ;
+ movdqa xmm3, xmm4 ; 13 12 11 10 03 02 01 00
+ punpcklwd xmm4, xmm0 ; 23 03 22 02 21 01 20 00
+ punpckhwd xmm3, xmm0 ; 33 13 32 12 31 11 30 10
+ movdqa xmm1, xmm4 ; 23 03 22 02 21 01 20 00
+ punpcklwd xmm4, xmm3 ; 31 21 11 01 30 20 10 00
+ punpckhwd xmm1, xmm3 ; 33 23 13 03 32 22 12 02
+ ;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ movd xmm0, eax
+ pshufd xmm2, xmm1, 4eh ;ip[8] ip[12]
+ movdqa xmm3, xmm4 ;ip[4] ip[0]
+
+ pshufd xmm0, xmm0, 0 ;03 03 03 03 03 03 03 03
+
+ paddw xmm4, xmm2 ;ip[4]+ip[8] ip[0]+ip[12] aka b1 a1
+ psubw xmm3, xmm2 ;ip[4]-ip[8] ip[0]-ip[12] aka c1 d1
+
+ movdqa xmm5, xmm4
+ punpcklqdq xmm4, xmm3 ;d1 a1
+ punpckhqdq xmm5, xmm3 ;c1 b1
+
+ movdqa xmm1, xmm5 ;c1 b1
+ paddw xmm5, xmm4 ;dl+cl a1+b1 aka op[4] op[0]
+ psubw xmm4, xmm1 ;d1-c1 a1-b1 aka op[12] op[8]
+
+ paddw xmm5, xmm0
+ paddw xmm4, xmm0
+ psraw xmm5, 3
+ psraw xmm4, 3
+
+ movd eax, xmm5
+ movd ecx, xmm4
+ psrldq xmm5, 4
+ psrldq xmm4, 4
+ mov word ptr[rdx+32*0], ax
+ mov word ptr[rdx+32*2], cx
+ shr eax, 16
+ shr ecx, 16
+ mov word ptr[rdx+32*4], ax
+ mov word ptr[rdx+32*6], cx
+ movd eax, xmm5
+ movd ecx, xmm4
+ psrldq xmm5, 4
+ psrldq xmm4, 4
+ mov word ptr[rdx+32*8], ax
+ mov word ptr[rdx+32*10], cx
+ shr eax, 16
+ shr ecx, 16
+ mov word ptr[rdx+32*12], ax
+ mov word ptr[rdx+32*14], cx
+
+ movd eax, xmm5
+ movd ecx, xmm4
+ psrldq xmm5, 4
+ psrldq xmm4, 4
+ mov word ptr[rdx+32*1], ax
+ mov word ptr[rdx+32*3], cx
+ shr eax, 16
+ shr ecx, 16
+ mov word ptr[rdx+32*5], ax
+ mov word ptr[rdx+32*7], cx
+ movd eax, xmm5
+ movd ecx, xmm4
+ mov word ptr[rdx+32*9], ax
+ mov word ptr[rdx+32*11], cx
+ shr eax, 16
+ shr ecx, 16
+ mov word ptr[rdx+32*13], ax
+ mov word ptr[rdx+32*15], cx
+
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/common/x86/loopfilter_block_sse2.asm b/libvpx/vp8/common/x86/loopfilter_block_sse2.asm
new file mode 100644
index 0000000..1c445ef
--- /dev/null
+++ b/libvpx/vp8/common/x86/loopfilter_block_sse2.asm
@@ -0,0 +1,813 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%macro LF_ABS 2
+ ; %1 value not preserved
+ ; %2 value preserved
+ ; output in %1
+ movdqa scratch1, %2 ; v2
+
+ psubusb scratch1, %1 ; v2 - v1
+ psubusb %1, %2 ; v1 - v2
+ por %1, scratch1 ; abs(v2 - v1)
+%endmacro
+
+%macro LF_FILTER_HEV_MASK 8-9
+
+ LF_ABS %1, %2 ; abs(p3 - p2)
+ LF_ABS %2, %3 ; abs(p2 - p1)
+ pmaxub %1, %2 ; accumulate mask
+%if %0 == 8
+ movdqa scratch2, %3 ; save p1
+ LF_ABS scratch2, %4 ; abs(p1 - p0)
+%endif
+ LF_ABS %4, %5 ; abs(p0 - q0)
+ LF_ABS %5, %6 ; abs(q0 - q1)
+%if %0 == 8
+ pmaxub %5, scratch2 ; accumulate hev
+%else
+ pmaxub %5, %9
+%endif
+ pmaxub %1, %5 ; accumulate mask
+
+ LF_ABS %3, %6 ; abs(p1 - q1)
+ LF_ABS %6, %7 ; abs(q1 - q2)
+ pmaxub %1, %6 ; accumulate mask
+ LF_ABS %7, %8 ; abs(q2 - q3)
+ pmaxub %1, %7 ; accumulate mask
+
+ paddusb %4, %4 ; 2 * abs(p0 - q0)
+ pand %3, [GLOBAL(tfe)]
+ psrlw %3, 1 ; abs(p1 - q1) / 2
+ paddusb %4, %3 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
+
+ psubusb %1, [limit]
+ psubusb %4, [blimit]
+ por %1, %4
+ pcmpeqb %1, zero ; mask
+
+ psubusb %5, [thresh]
+ pcmpeqb %5, zero ; ~hev
+%endmacro
+
+%macro LF_FILTER 6
+ ; %1-%4: p1-q1
+ ; %5: mask
+ ; %6: hev
+
+ movdqa scratch2, %6 ; save hev
+
+ pxor %1, [GLOBAL(t80)] ; ps1
+ pxor %4, [GLOBAL(t80)] ; qs1
+ movdqa scratch1, %1
+ psubsb scratch1, %4 ; signed_char_clamp(ps1 - qs1)
+ pandn scratch2, scratch1 ; vp8_filter &= hev
+
+ pxor %2, [GLOBAL(t80)] ; ps0
+ pxor %3, [GLOBAL(t80)] ; qs0
+ movdqa scratch1, %3
+ psubsb scratch1, %2 ; qs0 - ps0
+ paddsb scratch2, scratch1 ; vp8_filter += (qs0 - ps0)
+ paddsb scratch2, scratch1 ; vp8_filter += (qs0 - ps0)
+ paddsb scratch2, scratch1 ; vp8_filter += (qs0 - ps0)
+ pand %5, scratch2 ; &= mask
+
+ movdqa scratch2, %5
+ paddsb %5, [GLOBAL(t4)] ; Filter1
+ paddsb scratch2, [GLOBAL(t3)] ; Filter2
+
+ ; Filter1 >> 3
+ movdqa scratch1, zero
+ pcmpgtb scratch1, %5
+ psrlw %5, 3
+ pand scratch1, [GLOBAL(te0)]
+ pand %5, [GLOBAL(t1f)]
+ por %5, scratch1
+
+ psubsb %3, %5 ; qs0 - Filter1
+ pxor %3, [GLOBAL(t80)]
+
+ ; Filter2 >> 3
+ movdqa scratch1, zero
+ pcmpgtb scratch1, scratch2
+ psrlw scratch2, 3
+ pand scratch1, [GLOBAL(te0)]
+ pand scratch2, [GLOBAL(t1f)]
+ por scratch2, scratch1
+
+ paddsb %2, scratch2 ; ps0 + Filter2
+ pxor %2, [GLOBAL(t80)]
+
+ ; outer tap adjustments
+ paddsb %5, [GLOBAL(t1)]
+ movdqa scratch1, zero
+ pcmpgtb scratch1, %5
+ psrlw %5, 1
+ pand scratch1, [GLOBAL(t80)]
+ pand %5, [GLOBAL(t7f)]
+ por %5, scratch1
+ pand %5, %6 ; vp8_filter &= ~hev
+
+ psubsb %4, %5 ; qs1 - vp8_filter
+ pxor %4, [GLOBAL(t80)]
+
+ paddsb %1, %5 ; ps1 + vp8_filter
+ pxor %1, [GLOBAL(t80)]
+%endmacro
+
+;void vp8_loop_filter_bh_y_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh
+;)
+global sym(vp8_loop_filter_bh_y_sse2) PRIVATE
+sym(vp8_loop_filter_bh_y_sse2):
+
+%ifidn __OUTPUT_FORMAT__,x64
+ %define src rcx ; src_ptr
+ %define stride rdx ; src_pixel_step
+ %define blimit r8
+ %define limit r9
+ %define thresh r10
+
+ %define spp rax
+ %define stride3 r11
+ %define stride5 r12
+ %define stride7 r13
+
+ push rbp
+ mov rbp, rsp
+ push r12
+ push r13
+ mov thresh, arg(4)
+%else
+ %define src rdi ; src_ptr
+ %define stride rsi ; src_pixel_step
+ %define blimit rdx
+ %define limit rcx
+ %define thresh r8
+
+ %define spp rax
+ %define stride3 r9
+ %define stride5 r10
+ %define stride7 r11
+%endif
+
+ %define scratch1 xmm5
+ %define scratch2 xmm6
+ %define zero xmm7
+
+ %define i0 [src]
+ %define i1 [spp]
+ %define i2 [src + 2 * stride]
+ %define i3 [spp + 2 * stride]
+ %define i4 [src + 4 * stride]
+ %define i5 [spp + 4 * stride]
+ %define i6 [src + 2 * stride3]
+ %define i7 [spp + 2 * stride3]
+ %define i8 [src + 8 * stride]
+ %define i9 [spp + 8 * stride]
+ %define i10 [src + 2 * stride5]
+ %define i11 [spp + 2 * stride5]
+ %define i12 [src + 4 * stride3]
+ %define i13 [spp + 4 * stride3]
+ %define i14 [src + 2 * stride7]
+ %define i15 [spp + 2 * stride7]
+
+ ; prep work
+ lea spp, [src + stride]
+ lea stride3, [stride + 2 * stride]
+ lea stride5, [stride3 + 2 * stride]
+ lea stride7, [stride3 + 4 * stride]
+ pxor zero, zero
+
+ ; load the first set into registers
+ movdqa xmm0, i0
+ movdqa xmm1, i1
+ movdqa xmm2, i2
+ movdqa xmm3, i3
+ movdqa xmm4, i4
+ movdqa xmm8, i5
+ movdqa xmm9, i6 ; q2, will contain abs(p1-p0)
+ movdqa xmm10, i7
+LF_FILTER_HEV_MASK xmm0, xmm1, xmm2, xmm3, xmm4, xmm8, xmm9, xmm10
+
+ movdqa xmm1, i2
+ movdqa xmm2, i3
+ movdqa xmm3, i4
+ movdqa xmm8, i5
+LF_FILTER xmm1, xmm2, xmm3, xmm8, xmm0, xmm4
+ movdqa i2, xmm1
+ movdqa i3, xmm2
+
+; second set
+ movdqa i4, xmm3
+ movdqa i5, xmm8
+
+ movdqa xmm0, i6
+ movdqa xmm1, i7
+ movdqa xmm2, i8
+ movdqa xmm4, i9
+ movdqa xmm10, i10 ; q2, will contain abs(p1-p0)
+ movdqa xmm11, i11
+LF_FILTER_HEV_MASK xmm3, xmm8, xmm0, xmm1, xmm2, xmm4, xmm10, xmm11, xmm9
+
+ movdqa xmm0, i6
+ movdqa xmm1, i7
+ movdqa xmm4, i8
+ movdqa xmm8, i9
+LF_FILTER xmm0, xmm1, xmm4, xmm8, xmm3, xmm2
+ movdqa i6, xmm0
+ movdqa i7, xmm1
+
+; last set
+ movdqa i8, xmm4
+ movdqa i9, xmm8
+
+ movdqa xmm0, i10
+ movdqa xmm1, i11
+ movdqa xmm2, i12
+ movdqa xmm3, i13
+ movdqa xmm9, i14 ; q2, will contain abs(p1-p0)
+ movdqa xmm11, i15
+LF_FILTER_HEV_MASK xmm4, xmm8, xmm0, xmm1, xmm2, xmm3, xmm9, xmm11, xmm10
+
+ movdqa xmm0, i10
+ movdqa xmm1, i11
+ movdqa xmm3, i12
+ movdqa xmm8, i13
+LF_FILTER xmm0, xmm1, xmm3, xmm8, xmm4, xmm2
+ movdqa i10, xmm0
+ movdqa i11, xmm1
+ movdqa i12, xmm3
+ movdqa i13, xmm8
+
+%ifidn __OUTPUT_FORMAT__,x64
+ pop r13
+ pop r12
+ pop rbp
+%endif
+
+ ret
+
+
+;void vp8_loop_filter_bv_y_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh
+;)
+
+global sym(vp8_loop_filter_bv_y_sse2) PRIVATE
+sym(vp8_loop_filter_bv_y_sse2):
+
+%ifidn __OUTPUT_FORMAT__,x64
+ %define src rcx ; src_ptr
+ %define stride rdx ; src_pixel_step
+ %define blimit r8
+ %define limit r9
+ %define thresh r10
+
+ %define spp rax
+ %define stride3 r11
+ %define stride5 r12
+ %define stride7 r13
+
+ push rbp
+ mov rbp, rsp
+ SAVE_XMM 15
+ push r12
+ push r13
+ mov thresh, arg(4)
+%else
+ %define src rdi
+ %define stride rsi
+ %define blimit rdx
+ %define limit rcx
+ %define thresh r8
+
+ %define spp rax
+ %define stride3 r9
+ %define stride5 r10
+ %define stride7 r11
+%endif
+
+ %define scratch1 xmm5
+ %define scratch2 xmm6
+ %define zero xmm7
+
+ %define s0 [src]
+ %define s1 [spp]
+ %define s2 [src + 2 * stride]
+ %define s3 [spp + 2 * stride]
+ %define s4 [src + 4 * stride]
+ %define s5 [spp + 4 * stride]
+ %define s6 [src + 2 * stride3]
+ %define s7 [spp + 2 * stride3]
+ %define s8 [src + 8 * stride]
+ %define s9 [spp + 8 * stride]
+ %define s10 [src + 2 * stride5]
+ %define s11 [spp + 2 * stride5]
+ %define s12 [src + 4 * stride3]
+ %define s13 [spp + 4 * stride3]
+ %define s14 [src + 2 * stride7]
+ %define s15 [spp + 2 * stride7]
+
+ %define i0 [rsp]
+ %define i1 [rsp + 16]
+ %define i2 [rsp + 32]
+ %define i3 [rsp + 48]
+ %define i4 [rsp + 64]
+ %define i5 [rsp + 80]
+ %define i6 [rsp + 96]
+ %define i7 [rsp + 112]
+ %define i8 [rsp + 128]
+ %define i9 [rsp + 144]
+ %define i10 [rsp + 160]
+ %define i11 [rsp + 176]
+ %define i12 [rsp + 192]
+ %define i13 [rsp + 208]
+ %define i14 [rsp + 224]
+ %define i15 [rsp + 240]
+
+ ALIGN_STACK 16, rax
+
+ ; reserve stack space
+ %define temp_storage 0 ; size is 256 (16*16)
+ %define stack_size 256
+ sub rsp, stack_size
+
+ ; prep work
+ lea spp, [src + stride]
+ lea stride3, [stride + 2 * stride]
+ lea stride5, [stride3 + 2 * stride]
+ lea stride7, [stride3 + 4 * stride]
+
+ ; 8-f
+ movdqa xmm0, s8
+ movdqa xmm1, xmm0
+ punpcklbw xmm0, s9 ; 80 90
+ punpckhbw xmm1, s9 ; 88 98
+
+ movdqa xmm2, s10
+ movdqa xmm3, xmm2
+ punpcklbw xmm2, s11 ; a0 b0
+ punpckhbw xmm3, s11 ; a8 b8
+
+ movdqa xmm4, xmm0
+ punpcklwd xmm0, xmm2 ; 80 90 a0 b0
+ punpckhwd xmm4, xmm2 ; 84 94 a4 b4
+
+ movdqa xmm2, xmm1
+ punpcklwd xmm1, xmm3 ; 88 98 a8 b8
+ punpckhwd xmm2, xmm3 ; 8c 9c ac bc
+
+ ; using xmm[0124]
+ ; work on next 4 rows
+
+ movdqa xmm3, s12
+ movdqa xmm5, xmm3
+ punpcklbw xmm3, s13 ; c0 d0
+ punpckhbw xmm5, s13 ; c8 d8
+
+ movdqa xmm6, s14
+ movdqa xmm7, xmm6
+ punpcklbw xmm6, s15 ; e0 f0
+ punpckhbw xmm7, s15 ; e8 f8
+
+ movdqa xmm8, xmm3
+ punpcklwd xmm3, xmm6 ; c0 d0 e0 f0
+ punpckhwd xmm8, xmm6 ; c4 d4 e4 f4
+
+ movdqa xmm6, xmm5
+ punpcklwd xmm5, xmm7 ; c8 d8 e8 f8
+ punpckhwd xmm6, xmm7 ; cc dc ec fc
+
+ ; pull the third and fourth sets together
+
+ movdqa xmm7, xmm0
+ punpckldq xmm0, xmm3 ; 80 90 a0 b0 c0 d0 e0 f0
+ punpckhdq xmm7, xmm3 ; 82 92 a2 b2 c2 d2 e2 f2
+
+ movdqa xmm3, xmm4
+ punpckldq xmm4, xmm8 ; 84 94 a4 b4 c4 d4 e4 f4
+ punpckhdq xmm3, xmm8 ; 86 96 a6 b6 c6 d6 e6 f6
+
+ movdqa xmm8, xmm1
+ punpckldq xmm1, xmm5 ; 88 88 a8 b8 c8 d8 e8 f8
+ punpckhdq xmm8, xmm5 ; 8a 9a aa ba ca da ea fa
+
+ movdqa xmm5, xmm2
+ punpckldq xmm2, xmm6 ; 8c 9c ac bc cc dc ec fc
+ punpckhdq xmm5, xmm6 ; 8e 9e ae be ce de ee fe
+
+ ; save the calculations. we only have 15 registers ...
+ movdqa i0, xmm0
+ movdqa i1, xmm7
+ movdqa i2, xmm4
+ movdqa i3, xmm3
+ movdqa i4, xmm1
+ movdqa i5, xmm8
+ movdqa i6, xmm2
+ movdqa i7, xmm5
+
+ ; 0-7
+ movdqa xmm0, s0
+ movdqa xmm1, xmm0
+ punpcklbw xmm0, s1 ; 00 10
+ punpckhbw xmm1, s1 ; 08 18
+
+ movdqa xmm2, s2
+ movdqa xmm3, xmm2
+ punpcklbw xmm2, s3 ; 20 30
+ punpckhbw xmm3, s3 ; 28 38
+
+ movdqa xmm4, xmm0
+ punpcklwd xmm0, xmm2 ; 00 10 20 30
+ punpckhwd xmm4, xmm2 ; 04 14 24 34
+
+ movdqa xmm2, xmm1
+ punpcklwd xmm1, xmm3 ; 08 18 28 38
+ punpckhwd xmm2, xmm3 ; 0c 1c 2c 3c
+
+ ; using xmm[0124]
+ ; work on next 4 rows
+
+ movdqa xmm3, s4
+ movdqa xmm5, xmm3
+ punpcklbw xmm3, s5 ; 40 50
+ punpckhbw xmm5, s5 ; 48 58
+
+ movdqa xmm6, s6
+ movdqa xmm7, xmm6
+ punpcklbw xmm6, s7 ; 60 70
+ punpckhbw xmm7, s7 ; 68 78
+
+ movdqa xmm8, xmm3
+ punpcklwd xmm3, xmm6 ; 40 50 60 70
+ punpckhwd xmm8, xmm6 ; 44 54 64 74
+
+ movdqa xmm6, xmm5
+ punpcklwd xmm5, xmm7 ; 48 58 68 78
+ punpckhwd xmm6, xmm7 ; 4c 5c 6c 7c
+
+ ; pull the first two sets together
+
+ movdqa xmm7, xmm0
+ punpckldq xmm0, xmm3 ; 00 10 20 30 40 50 60 70
+ punpckhdq xmm7, xmm3 ; 02 12 22 32 42 52 62 72
+
+ movdqa xmm3, xmm4
+ punpckldq xmm4, xmm8 ; 04 14 24 34 44 54 64 74
+ punpckhdq xmm3, xmm8 ; 06 16 26 36 46 56 66 76
+
+ movdqa xmm8, xmm1
+ punpckldq xmm1, xmm5 ; 08 18 28 38 48 58 68 78
+ punpckhdq xmm8, xmm5 ; 0a 1a 2a 3a 4a 5a 6a 7a
+
+ movdqa xmm5, xmm2
+ punpckldq xmm2, xmm6 ; 0c 1c 2c 3c 4c 5c 6c 7c
+ punpckhdq xmm5, xmm6 ; 0e 1e 2e 3e 4e 5e 6e 7e
+ ; final combination
+
+ movdqa xmm6, xmm0
+ punpcklqdq xmm0, i0
+ punpckhqdq xmm6, i0
+
+ movdqa xmm9, xmm7
+ punpcklqdq xmm7, i1
+ punpckhqdq xmm9, i1
+
+ movdqa xmm10, xmm4
+ punpcklqdq xmm4, i2
+ punpckhqdq xmm10, i2
+
+ movdqa xmm11, xmm3
+ punpcklqdq xmm3, i3
+ punpckhqdq xmm11, i3
+
+ movdqa xmm12, xmm1
+ punpcklqdq xmm1, i4
+ punpckhqdq xmm12, i4
+
+ movdqa xmm13, xmm8
+ punpcklqdq xmm8, i5
+ punpckhqdq xmm13, i5
+
+ movdqa xmm14, xmm2
+ punpcklqdq xmm2, i6
+ punpckhqdq xmm14, i6
+
+ movdqa xmm15, xmm5
+ punpcklqdq xmm5, i7
+ punpckhqdq xmm15, i7
+
+ movdqa i0, xmm0
+ movdqa i1, xmm6
+ movdqa i2, xmm7
+ movdqa i3, xmm9
+ movdqa i4, xmm4
+ movdqa i5, xmm10
+ movdqa i6, xmm3
+ movdqa i7, xmm11
+ movdqa i8, xmm1
+ movdqa i9, xmm12
+ movdqa i10, xmm8
+ movdqa i11, xmm13
+ movdqa i12, xmm2
+ movdqa i13, xmm14
+ movdqa i14, xmm5
+ movdqa i15, xmm15
+
+; TRANSPOSED DATA AVAILABLE ON THE STACK
+
+ movdqa xmm12, xmm6
+ movdqa xmm13, xmm7
+
+ pxor zero, zero
+
+LF_FILTER_HEV_MASK xmm0, xmm12, xmm13, xmm9, xmm4, xmm10, xmm3, xmm11
+
+ movdqa xmm1, i2
+ movdqa xmm2, i3
+ movdqa xmm8, i4
+ movdqa xmm9, i5
+LF_FILTER xmm1, xmm2, xmm8, xmm9, xmm0, xmm4
+ movdqa i2, xmm1
+ movdqa i3, xmm2
+
+; second set
+ movdqa i4, xmm8
+ movdqa i5, xmm9
+
+ movdqa xmm0, i6
+ movdqa xmm1, i7
+ movdqa xmm2, i8
+ movdqa xmm4, i9
+ movdqa xmm10, i10 ; q2, will contain abs(p1-p0)
+ movdqa xmm11, i11
+LF_FILTER_HEV_MASK xmm8, xmm9, xmm0, xmm1, xmm2, xmm4, xmm10, xmm11, xmm3
+
+ movdqa xmm0, i6
+ movdqa xmm1, i7
+ movdqa xmm3, i8
+ movdqa xmm4, i9
+LF_FILTER xmm0, xmm1, xmm3, xmm4, xmm8, xmm2
+ movdqa i6, xmm0
+ movdqa i7, xmm1
+
+; last set
+ movdqa i8, xmm3
+ movdqa i9, xmm4
+
+ movdqa xmm0, i10
+ movdqa xmm1, i11
+ movdqa xmm2, i12
+ movdqa xmm8, i13
+ movdqa xmm9, i14 ; q2, will contain abs(p1-p0)
+ movdqa xmm11, i15
+LF_FILTER_HEV_MASK xmm3, xmm4, xmm0, xmm1, xmm2, xmm8, xmm9, xmm11, xmm10
+
+ movdqa xmm0, i10
+ movdqa xmm1, i11
+ movdqa xmm4, i12
+ movdqa xmm8, i13
+LF_FILTER xmm0, xmm1, xmm4, xmm8, xmm3, xmm2
+ movdqa i10, xmm0
+ movdqa i11, xmm1
+ movdqa i12, xmm4
+ movdqa i13, xmm8
+
+
+; RESHUFFLE AND WRITE OUT
+ ; 8-f
+ movdqa xmm0, i8
+ movdqa xmm1, xmm0
+ punpcklbw xmm0, i9 ; 80 90
+ punpckhbw xmm1, i9 ; 88 98
+
+ movdqa xmm2, i10
+ movdqa xmm3, xmm2
+ punpcklbw xmm2, i11 ; a0 b0
+ punpckhbw xmm3, i11 ; a8 b8
+
+ movdqa xmm4, xmm0
+ punpcklwd xmm0, xmm2 ; 80 90 a0 b0
+ punpckhwd xmm4, xmm2 ; 84 94 a4 b4
+
+ movdqa xmm2, xmm1
+ punpcklwd xmm1, xmm3 ; 88 98 a8 b8
+ punpckhwd xmm2, xmm3 ; 8c 9c ac bc
+
+ ; using xmm[0124]
+ ; work on next 4 rows
+
+ movdqa xmm3, i12
+ movdqa xmm5, xmm3
+ punpcklbw xmm3, i13 ; c0 d0
+ punpckhbw xmm5, i13 ; c8 d8
+
+ movdqa xmm6, i14
+ movdqa xmm7, xmm6
+ punpcklbw xmm6, i15 ; e0 f0
+ punpckhbw xmm7, i15 ; e8 f8
+
+ movdqa xmm8, xmm3
+ punpcklwd xmm3, xmm6 ; c0 d0 e0 f0
+ punpckhwd xmm8, xmm6 ; c4 d4 e4 f4
+
+ movdqa xmm6, xmm5
+ punpcklwd xmm5, xmm7 ; c8 d8 e8 f8
+ punpckhwd xmm6, xmm7 ; cc dc ec fc
+
+ ; pull the third and fourth sets together
+
+ movdqa xmm7, xmm0
+ punpckldq xmm0, xmm3 ; 80 90 a0 b0 c0 d0 e0 f0
+ punpckhdq xmm7, xmm3 ; 82 92 a2 b2 c2 d2 e2 f2
+
+ movdqa xmm3, xmm4
+ punpckldq xmm4, xmm8 ; 84 94 a4 b4 c4 d4 e4 f4
+ punpckhdq xmm3, xmm8 ; 86 96 a6 b6 c6 d6 e6 f6
+
+ movdqa xmm8, xmm1
+ punpckldq xmm1, xmm5 ; 88 88 a8 b8 c8 d8 e8 f8
+ punpckhdq xmm8, xmm5 ; 8a 9a aa ba ca da ea fa
+
+ movdqa xmm5, xmm2
+ punpckldq xmm2, xmm6 ; 8c 9c ac bc cc dc ec fc
+ punpckhdq xmm5, xmm6 ; 8e 9e ae be ce de ee fe
+
+ ; save the calculations. we only have 15 registers ...
+ movdqa i8, xmm0
+ movdqa i9, xmm7
+ movdqa i10, xmm4
+ movdqa i11, xmm3
+ movdqa i12, xmm1
+ movdqa i13, xmm8
+ movdqa i14, xmm2
+ movdqa i15, xmm5
+
+ ; 0-7
+ movdqa xmm0, i0
+ movdqa xmm1, xmm0
+ punpcklbw xmm0, i1 ; 00 10
+ punpckhbw xmm1, i1 ; 08 18
+
+ movdqa xmm2, i2
+ movdqa xmm3, xmm2
+ punpcklbw xmm2, i3 ; 20 30
+ punpckhbw xmm3, i3 ; 28 38
+
+ movdqa xmm4, xmm0
+ punpcklwd xmm0, xmm2 ; 00 10 20 30
+ punpckhwd xmm4, xmm2 ; 04 14 24 34
+
+ movdqa xmm2, xmm1
+ punpcklwd xmm1, xmm3 ; 08 18 28 38
+ punpckhwd xmm2, xmm3 ; 0c 1c 2c 3c
+
+ ; using xmm[0124]
+ ; work on next 4 rows
+
+ movdqa xmm3, i4
+ movdqa xmm5, xmm3
+ punpcklbw xmm3, i5 ; 40 50
+ punpckhbw xmm5, i5 ; 48 58
+
+ movdqa xmm6, i6
+ movdqa xmm7, xmm6
+ punpcklbw xmm6, i7 ; 60 70
+ punpckhbw xmm7, i7 ; 68 78
+
+ movdqa xmm8, xmm3
+ punpcklwd xmm3, xmm6 ; 40 50 60 70
+ punpckhwd xmm8, xmm6 ; 44 54 64 74
+
+ movdqa xmm6, xmm5
+ punpcklwd xmm5, xmm7 ; 48 58 68 78
+ punpckhwd xmm6, xmm7 ; 4c 5c 6c 7c
+
+ ; pull the first two sets together
+
+ movdqa xmm7, xmm0
+ punpckldq xmm0, xmm3 ; 00 10 20 30 40 50 60 70
+ punpckhdq xmm7, xmm3 ; 02 12 22 32 42 52 62 72
+
+ movdqa xmm3, xmm4
+ punpckldq xmm4, xmm8 ; 04 14 24 34 44 54 64 74
+ punpckhdq xmm3, xmm8 ; 06 16 26 36 46 56 66 76
+
+ movdqa xmm8, xmm1
+ punpckldq xmm1, xmm5 ; 08 18 28 38 48 58 68 78
+ punpckhdq xmm8, xmm5 ; 0a 1a 2a 3a 4a 5a 6a 7a
+
+ movdqa xmm5, xmm2
+ punpckldq xmm2, xmm6 ; 0c 1c 2c 3c 4c 5c 6c 7c
+ punpckhdq xmm5, xmm6 ; 0e 1e 2e 3e 4e 5e 6e 7e
+ ; final combination
+
+ movdqa xmm6, xmm0
+ punpcklqdq xmm0, i8
+ punpckhqdq xmm6, i8
+
+ movdqa xmm9, xmm7
+ punpcklqdq xmm7, i9
+ punpckhqdq xmm9, i9
+
+ movdqa xmm10, xmm4
+ punpcklqdq xmm4, i10
+ punpckhqdq xmm10, i10
+
+ movdqa xmm11, xmm3
+ punpcklqdq xmm3, i11
+ punpckhqdq xmm11, i11
+
+ movdqa xmm12, xmm1
+ punpcklqdq xmm1, i12
+ punpckhqdq xmm12, i12
+
+ movdqa xmm13, xmm8
+ punpcklqdq xmm8, i13
+ punpckhqdq xmm13, i13
+
+ movdqa xmm14, xmm2
+ punpcklqdq xmm2, i14
+ punpckhqdq xmm14, i14
+
+ movdqa xmm15, xmm5
+ punpcklqdq xmm5, i15
+ punpckhqdq xmm15, i15
+
+ movdqa s0, xmm0
+ movdqa s1, xmm6
+ movdqa s2, xmm7
+ movdqa s3, xmm9
+ movdqa s4, xmm4
+ movdqa s5, xmm10
+ movdqa s6, xmm3
+ movdqa s7, xmm11
+ movdqa s8, xmm1
+ movdqa s9, xmm12
+ movdqa s10, xmm8
+ movdqa s11, xmm13
+ movdqa s12, xmm2
+ movdqa s13, xmm14
+ movdqa s14, xmm5
+ movdqa s15, xmm15
+
+ ; free stack space
+ add rsp, stack_size
+
+ ; un-ALIGN_STACK
+ pop rsp
+
+%ifidn __OUTPUT_FORMAT__,x64
+ pop r13
+ pop r12
+ RESTORE_XMM
+ pop rbp
+%endif
+
+ ret
+
+SECTION_RODATA
+align 16
+te0:
+ times 16 db 0xe0
+align 16
+t7f:
+ times 16 db 0x7f
+align 16
+tfe:
+ times 16 db 0xfe
+align 16
+t1f:
+ times 16 db 0x1f
+align 16
+t80:
+ times 16 db 0x80
+align 16
+t1:
+ times 16 db 0x01
+align 16
+t3:
+ times 16 db 0x03
+align 16
+t4:
+ times 16 db 0x04
diff --git a/libvpx/vp8/common/x86/loopfilter_mmx.asm b/libvpx/vp8/common/x86/loopfilter_mmx.asm
new file mode 100644
index 0000000..f388d24
--- /dev/null
+++ b/libvpx/vp8/common/x86/loopfilter_mmx.asm
@@ -0,0 +1,1753 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+
+;void vp8_loop_filter_horizontal_edge_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+; int count
+;)
+global sym(vp8_loop_filter_horizontal_edge_mmx) PRIVATE
+sym(vp8_loop_filter_horizontal_edge_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 32 ; reserve 32 bytes
+ %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
+ %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
+
+ movsxd rcx, dword ptr arg(5) ;count
+.next8_h:
+ mov rdx, arg(3) ;limit
+ movq mm7, [rdx]
+ mov rdi, rsi ; rdi points to row +1 for indirect addressing
+ add rdi, rax
+
+ ; calculate breakout conditions
+ movq mm2, [rdi+2*rax] ; q3
+ movq mm1, [rsi+2*rax] ; q2
+ movq mm6, mm1 ; q2
+ psubusb mm1, mm2 ; q2-=q3
+ psubusb mm2, mm6 ; q3-=q2
+ por mm1, mm2 ; abs(q3-q2)
+ psubusb mm1, mm7 ;
+
+
+ movq mm4, [rsi+rax] ; q1
+ movq mm3, mm4 ; q1
+ psubusb mm4, mm6 ; q1-=q2
+ psubusb mm6, mm3 ; q2-=q1
+ por mm4, mm6 ; abs(q2-q1)
+
+ psubusb mm4, mm7
+ por mm1, mm4
+
+ movq mm4, [rsi] ; q0
+ movq mm0, mm4 ; q0
+ psubusb mm4, mm3 ; q0-=q1
+ psubusb mm3, mm0 ; q1-=q0
+ por mm4, mm3 ; abs(q0-q1)
+ movq t0, mm4 ; save to t0
+ psubusb mm4, mm7
+ por mm1, mm4
+
+
+ neg rax ; negate pitch to deal with above border
+
+ movq mm2, [rsi+4*rax] ; p3
+ movq mm4, [rdi+4*rax] ; p2
+ movq mm5, mm4 ; p2
+ psubusb mm4, mm2 ; p2-=p3
+ psubusb mm2, mm5 ; p3-=p2
+ por mm4, mm2 ; abs(p3 - p2)
+ psubusb mm4, mm7
+ por mm1, mm4
+
+
+ movq mm4, [rsi+2*rax] ; p1
+ movq mm3, mm4 ; p1
+ psubusb mm4, mm5 ; p1-=p2
+ psubusb mm5, mm3 ; p2-=p1
+ por mm4, mm5 ; abs(p2 - p1)
+ psubusb mm4, mm7
+ por mm1, mm4
+
+ movq mm2, mm3 ; p1
+
+ movq mm4, [rsi+rax] ; p0
+ movq mm5, mm4 ; p0
+ psubusb mm4, mm3 ; p0-=p1
+ psubusb mm3, mm5 ; p1-=p0
+ por mm4, mm3 ; abs(p1 - p0)
+ movq t1, mm4 ; save to t1
+ psubusb mm4, mm7
+ por mm1, mm4
+
+ movq mm3, [rdi] ; q1
+ movq mm4, mm3 ; q1
+ psubusb mm3, mm2 ; q1-=p1
+ psubusb mm2, mm4 ; p1-=q1
+ por mm2, mm3 ; abs(p1-q1)
+ pand mm2, [GLOBAL(tfe)] ; set lsb of each byte to zero
+ psrlw mm2, 1 ; abs(p1-q1)/2
+
+ movq mm6, mm5 ; p0
+ movq mm3, [rsi] ; q0
+ psubusb mm5, mm3 ; p0-=q0
+ psubusb mm3, mm6 ; q0-=p0
+ por mm5, mm3 ; abs(p0 - q0)
+ paddusb mm5, mm5 ; abs(p0-q0)*2
+ paddusb mm5, mm2 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+
+ mov rdx, arg(2) ;blimit ; get blimit
+ movq mm7, [rdx] ; blimit
+
+ psubusb mm5, mm7 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ por mm1, mm5
+ pxor mm5, mm5
+ pcmpeqb mm1, mm5 ; mask mm1
+
+ ; calculate high edge variance
+ mov rdx, arg(4) ;thresh ; get thresh
+ movq mm7, [rdx] ;
+ movq mm4, t0 ; get abs (q1 - q0)
+ psubusb mm4, mm7
+ movq mm3, t1 ; get abs (p1 - p0)
+ psubusb mm3, mm7
+ paddb mm4, mm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
+
+ pcmpeqb mm4, mm5
+
+ pcmpeqb mm5, mm5
+ pxor mm4, mm5
+
+
+ ; start work on filters
+ movq mm2, [rsi+2*rax] ; p1
+ movq mm7, [rdi] ; q1
+ pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
+ pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
+ psubsb mm2, mm7 ; p1 - q1
+ pand mm2, mm4 ; high var mask (hvm)(p1 - q1)
+ pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
+ pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
+ movq mm3, mm0 ; q0
+ psubsb mm0, mm6 ; q0 - p0
+ paddsb mm2, mm0 ; 1 * (q0 - p0) + hvm(p1 - q1)
+ paddsb mm2, mm0 ; 2 * (q0 - p0) + hvm(p1 - q1)
+ paddsb mm2, mm0 ; 3 * (q0 - p0) + hvm(p1 - q1)
+ pand mm1, mm2 ; mask filter values we don't care about
+ movq mm2, mm1
+ paddsb mm1, [GLOBAL(t4)] ; 3* (q0 - p0) + hvm(p1 - q1) + 4
+ paddsb mm2, [GLOBAL(t3)] ; 3* (q0 - p0) + hvm(p1 - q1) + 3
+
+ pxor mm0, mm0 ;
+ pxor mm5, mm5
+ punpcklbw mm0, mm2 ;
+ punpckhbw mm5, mm2 ;
+ psraw mm0, 11 ;
+ psraw mm5, 11
+ packsswb mm0, mm5
+ movq mm2, mm0 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3;
+
+ pxor mm0, mm0 ; 0
+ movq mm5, mm1 ; abcdefgh
+ punpcklbw mm0, mm1 ; e0f0g0h0
+ psraw mm0, 11 ; sign extended shift right by 3
+ pxor mm1, mm1 ; 0
+ punpckhbw mm1, mm5 ; a0b0c0d0
+ psraw mm1, 11 ; sign extended shift right by 3
+ movq mm5, mm0 ; save results
+
+ packsswb mm0, mm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3
+ paddsw mm5, [GLOBAL(ones)]
+ paddsw mm1, [GLOBAL(ones)]
+ psraw mm5, 1 ; partial shifted one more time for 2nd tap
+ psraw mm1, 1 ; partial shifted one more time for 2nd tap
+ packsswb mm5, mm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4
+ pandn mm4, mm5 ; high edge variance additive
+
+ paddsb mm6, mm2 ; p0+= p0 add
+ pxor mm6, [GLOBAL(t80)] ; unoffset
+ movq [rsi+rax], mm6 ; write back
+
+ movq mm6, [rsi+2*rax] ; p1
+ pxor mm6, [GLOBAL(t80)] ; reoffset
+ paddsb mm6, mm4 ; p1+= p1 add
+ pxor mm6, [GLOBAL(t80)] ; unoffset
+ movq [rsi+2*rax], mm6 ; write back
+
+ psubsb mm3, mm0 ; q0-= q0 add
+ pxor mm3, [GLOBAL(t80)] ; unoffset
+ movq [rsi], mm3 ; write back
+
+ psubsb mm7, mm4 ; q1-= q1 add
+ pxor mm7, [GLOBAL(t80)] ; unoffset
+ movq [rdi], mm7 ; write back
+
+ add rsi,8
+ neg rax
+ dec rcx
+ jnz .next8_h
+
+ add rsp, 32
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_loop_filter_vertical_edge_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+; int count
+;)
+global sym(vp8_loop_filter_vertical_edge_mmx) PRIVATE
+sym(vp8_loop_filter_vertical_edge_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 64 ; reserve 64 bytes
+ %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
+ %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
+ %define srct [rsp + 32] ;__declspec(align(16)) char srct[32];
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
+
+ lea rsi, [rsi + rax*4 - 4]
+
+ movsxd rcx, dword ptr arg(5) ;count
+.next8_v:
+ mov rdi, rsi ; rdi points to row +1 for indirect addressing
+ add rdi, rax
+
+
+ ;transpose
+ movq mm6, [rsi+2*rax] ; 67 66 65 64 63 62 61 60
+ movq mm7, mm6 ; 77 76 75 74 73 72 71 70
+
+ punpckhbw mm7, [rdi+2*rax] ; 77 67 76 66 75 65 74 64
+ punpcklbw mm6, [rdi+2*rax] ; 73 63 72 62 71 61 70 60
+
+ movq mm4, [rsi] ; 47 46 45 44 43 42 41 40
+ movq mm5, mm4 ; 47 46 45 44 43 42 41 40
+
+ punpckhbw mm5, [rsi+rax] ; 57 47 56 46 55 45 54 44
+ punpcklbw mm4, [rsi+rax] ; 53 43 52 42 51 41 50 40
+
+ movq mm3, mm5 ; 57 47 56 46 55 45 54 44
+ punpckhwd mm5, mm7 ; 77 67 57 47 76 66 56 46
+
+ punpcklwd mm3, mm7 ; 75 65 55 45 74 64 54 44
+ movq mm2, mm4 ; 53 43 52 42 51 41 50 40
+
+ punpckhwd mm4, mm6 ; 73 63 53 43 72 62 52 42
+ punpcklwd mm2, mm6 ; 71 61 51 41 70 60 50 40
+
+ neg rax
+ movq mm6, [rsi+rax*2] ; 27 26 25 24 23 22 21 20
+
+ movq mm1, mm6 ; 27 26 25 24 23 22 21 20
+ punpckhbw mm6, [rsi+rax] ; 37 27 36 36 35 25 34 24
+
+ punpcklbw mm1, [rsi+rax] ; 33 23 32 22 31 21 30 20
+ movq mm7, [rsi+rax*4]; ; 07 06 05 04 03 02 01 00
+
+ punpckhbw mm7, [rdi+rax*4] ; 17 07 16 06 15 05 14 04
+ movq mm0, mm7 ; 17 07 16 06 15 05 14 04
+
+ punpckhwd mm7, mm6 ; 37 27 17 07 36 26 16 06
+ punpcklwd mm0, mm6 ; 35 25 15 05 34 24 14 04
+
+ movq mm6, mm7 ; 37 27 17 07 36 26 16 06
+ punpckhdq mm7, mm5 ; 77 67 57 47 37 27 17 07 = q3
+
+ punpckldq mm6, mm5 ; 76 66 56 46 36 26 16 06 = q2
+
+ movq mm5, mm6 ; 76 66 56 46 36 26 16 06
+ psubusb mm5, mm7 ; q2-q3
+
+ psubusb mm7, mm6 ; q3-q2
+ por mm7, mm5; ; mm7=abs (q3-q2)
+
+ movq mm5, mm0 ; 35 25 15 05 34 24 14 04
+ punpckhdq mm5, mm3 ; 75 65 55 45 35 25 15 05 = q1
+
+ punpckldq mm0, mm3 ; 74 64 54 44 34 24 15 04 = q0
+ movq mm3, mm5 ; 75 65 55 45 35 25 15 05 = q1
+
+ psubusb mm3, mm6 ; q1-q2
+ psubusb mm6, mm5 ; q2-q1
+
+ por mm6, mm3 ; mm6=abs(q2-q1)
+ lea rdx, srct
+
+ movq [rdx+24], mm5 ; save q1
+ movq [rdx+16], mm0 ; save q0
+
+ movq mm3, [rsi+rax*4] ; 07 06 05 04 03 02 01 00
+ punpcklbw mm3, [rdi+rax*4] ; 13 03 12 02 11 01 10 00
+
+ movq mm0, mm3 ; 13 03 12 02 11 01 10 00
+ punpcklwd mm0, mm1 ; 31 21 11 01 30 20 10 00
+
+ punpckhwd mm3, mm1 ; 33 23 13 03 32 22 12 02
+ movq mm1, mm0 ; 31 21 11 01 30 20 10 00
+
+ punpckldq mm0, mm2 ; 70 60 50 40 30 20 10 00 =p3
+ punpckhdq mm1, mm2 ; 71 61 51 41 31 21 11 01 =p2
+
+ movq mm2, mm1 ; 71 61 51 41 31 21 11 01 =p2
+ psubusb mm2, mm0 ; p2-p3
+
+ psubusb mm0, mm1 ; p3-p2
+ por mm0, mm2 ; mm0=abs(p3-p2)
+
+ movq mm2, mm3 ; 33 23 13 03 32 22 12 02
+ punpckldq mm2, mm4 ; 72 62 52 42 32 22 12 02 = p1
+
+ punpckhdq mm3, mm4 ; 73 63 53 43 33 23 13 03 = p0
+ movq [rdx+8], mm3 ; save p0
+
+ movq [rdx], mm2 ; save p1
+ movq mm5, mm2 ; mm5 = p1
+
+ psubusb mm2, mm1 ; p1-p2
+ psubusb mm1, mm5 ; p2-p1
+
+ por mm1, mm2 ; mm1=abs(p2-p1)
+ mov rdx, arg(3) ;limit
+
+ movq mm4, [rdx] ; mm4 = limit
+ psubusb mm7, mm4
+
+ psubusb mm0, mm4
+ psubusb mm1, mm4
+
+ psubusb mm6, mm4
+ por mm7, mm6
+
+ por mm0, mm1
+ por mm0, mm7 ; abs(q3-q2) > limit || abs(p3-p2) > limit ||abs(p2-p1) > limit || abs(q2-q1) > limit
+
+ movq mm1, mm5 ; p1
+
+ movq mm7, mm3 ; mm3=mm7=p0
+ psubusb mm7, mm5 ; p0 - p1
+
+ psubusb mm5, mm3 ; p1 - p0
+ por mm5, mm7 ; abs(p1-p0)
+
+ movq t0, mm5 ; save abs(p1-p0)
+ lea rdx, srct
+
+ psubusb mm5, mm4
+ por mm0, mm5 ; mm0=mask
+
+ movq mm5, [rdx+16] ; mm5=q0
+ movq mm7, [rdx+24] ; mm7=q1
+
+ movq mm6, mm5 ; mm6=q0
+ movq mm2, mm7 ; q1
+ psubusb mm5, mm7 ; q0-q1
+
+ psubusb mm7, mm6 ; q1-q0
+ por mm7, mm5 ; abs(q1-q0)
+
+ movq t1, mm7 ; save abs(q1-q0)
+ psubusb mm7, mm4
+
+ por mm0, mm7 ; mask
+
+ movq mm5, mm2 ; q1
+ psubusb mm5, mm1 ; q1-=p1
+ psubusb mm1, mm2 ; p1-=q1
+ por mm5, mm1 ; abs(p1-q1)
+ pand mm5, [GLOBAL(tfe)] ; set lsb of each byte to zero
+ psrlw mm5, 1 ; abs(p1-q1)/2
+
+ mov rdx, arg(2) ;blimit ;
+
+ movq mm4, [rdx] ;blimit
+ movq mm1, mm3 ; mm1=mm3=p0
+
+ movq mm7, mm6 ; mm7=mm6=q0
+ psubusb mm1, mm7 ; p0-q0
+
+ psubusb mm7, mm3 ; q0-p0
+ por mm1, mm7 ; abs(q0-p0)
+ paddusb mm1, mm1 ; abs(q0-p0)*2
+ paddusb mm1, mm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+
+ psubusb mm1, mm4 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ por mm1, mm0; ; mask
+
+ pxor mm0, mm0
+ pcmpeqb mm1, mm0
+
+ ; calculate high edge variance
+ mov rdx, arg(4) ;thresh ; get thresh
+ movq mm7, [rdx]
+ ;
+ movq mm4, t0 ; get abs (q1 - q0)
+ psubusb mm4, mm7
+
+ movq mm3, t1 ; get abs (p1 - p0)
+ psubusb mm3, mm7
+
+ por mm4, mm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
+ pcmpeqb mm4, mm0
+
+ pcmpeqb mm0, mm0
+ pxor mm4, mm0
+
+
+
+ ; start work on filters
+ lea rdx, srct
+
+ movq mm2, [rdx] ; p1
+ movq mm7, [rdx+24] ; q1
+
+ movq mm6, [rdx+8] ; p0
+ movq mm0, [rdx+16] ; q0
+
+ pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
+ pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
+
+ psubsb mm2, mm7 ; p1 - q1
+ pand mm2, mm4 ; high var mask (hvm)(p1 - q1)
+
+ pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
+ pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
+
+ movq mm3, mm0 ; q0
+ psubsb mm0, mm6 ; q0 - p0
+
+ paddsb mm2, mm0 ; 1 * (q0 - p0) + hvm(p1 - q1)
+ paddsb mm2, mm0 ; 2 * (q0 - p0) + hvm(p1 - q1)
+
+ paddsb mm2, mm0 ; 3 * (q0 - p0) + hvm(p1 - q1)
+ pand mm1, mm2 ; mask filter values we don't care about
+
+ movq mm2, mm1
+ paddsb mm1, [GLOBAL(t4)] ; 3* (q0 - p0) + hvm(p1 - q1) + 4
+
+ paddsb mm2, [GLOBAL(t3)] ; 3* (q0 - p0) + hvm(p1 - q1) + 3
+ pxor mm0, mm0 ;
+
+ pxor mm5, mm5
+ punpcklbw mm0, mm2 ;
+
+ punpckhbw mm5, mm2 ;
+ psraw mm0, 11 ;
+
+ psraw mm5, 11
+ packsswb mm0, mm5
+
+ movq mm2, mm0 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3;
+
+ pxor mm0, mm0 ; 0
+ movq mm5, mm1 ; abcdefgh
+
+ punpcklbw mm0, mm1 ; e0f0g0h0
+ psraw mm0, 11 ; sign extended shift right by 3
+
+ pxor mm1, mm1 ; 0
+ punpckhbw mm1, mm5 ; a0b0c0d0
+
+ psraw mm1, 11 ; sign extended shift right by 3
+ movq mm5, mm0 ; save results
+
+ packsswb mm0, mm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3
+ paddsw mm5, [GLOBAL(ones)]
+
+ paddsw mm1, [GLOBAL(ones)]
+ psraw mm5, 1 ; partial shifted one more time for 2nd tap
+
+ psraw mm1, 1 ; partial shifted one more time for 2nd tap
+ packsswb mm5, mm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4
+
+ pandn mm4, mm5 ; high edge variance additive
+
+ paddsb mm6, mm2 ; p0+= p0 add
+ pxor mm6, [GLOBAL(t80)] ; unoffset
+
+ ; mm6=p0 ;
+ movq mm1, [rdx] ; p1
+ pxor mm1, [GLOBAL(t80)] ; reoffset
+
+ paddsb mm1, mm4 ; p1+= p1 add
+ pxor mm1, [GLOBAL(t80)] ; unoffset
+ ; mm6 = p0 mm1 = p1
+
+ psubsb mm3, mm0 ; q0-= q0 add
+ pxor mm3, [GLOBAL(t80)] ; unoffset
+
+ ; mm3 = q0
+ psubsb mm7, mm4 ; q1-= q1 add
+ pxor mm7, [GLOBAL(t80)] ; unoffset
+ ; mm7 = q1
+
+ ; tranpose and write back
+ ; mm1 = 72 62 52 42 32 22 12 02
+ ; mm6 = 73 63 53 43 33 23 13 03
+ ; mm3 = 74 64 54 44 34 24 14 04
+ ; mm7 = 75 65 55 45 35 25 15 05
+
+ movq mm2, mm1 ; 72 62 52 42 32 22 12 02
+ punpcklbw mm2, mm6 ; 33 32 23 22 13 12 03 02
+
+ movq mm4, mm3 ; 74 64 54 44 34 24 14 04
+ punpckhbw mm1, mm6 ; 73 72 63 62 53 52 43 42
+
+ punpcklbw mm4, mm7 ; 35 34 25 24 15 14 05 04
+ punpckhbw mm3, mm7 ; 75 74 65 64 55 54 45 44
+
+ movq mm6, mm2 ; 33 32 23 22 13 12 03 02
+ punpcklwd mm2, mm4 ; 15 14 13 12 05 04 03 02
+
+ punpckhwd mm6, mm4 ; 35 34 33 32 25 24 23 22
+ movq mm5, mm1 ; 73 72 63 62 53 52 43 42
+
+ punpcklwd mm1, mm3 ; 55 54 53 52 45 44 43 42
+ punpckhwd mm5, mm3 ; 75 74 73 72 65 64 63 62
+
+
+ ; mm2 = 15 14 13 12 05 04 03 02
+ ; mm6 = 35 34 33 32 25 24 23 22
+ ; mm5 = 55 54 53 52 45 44 43 42
+ ; mm1 = 75 74 73 72 65 64 63 62
+
+
+
+ movd [rsi+rax*4+2], mm2
+ psrlq mm2, 32
+
+ movd [rdi+rax*4+2], mm2
+ movd [rsi+rax*2+2], mm6
+
+ psrlq mm6, 32
+ movd [rsi+rax+2],mm6
+
+ movd [rsi+2], mm1
+ psrlq mm1, 32
+
+ movd [rdi+2], mm1
+ neg rax
+
+ movd [rdi+rax+2],mm5
+ psrlq mm5, 32
+
+ movd [rdi+rax*2+2], mm5
+
+ lea rsi, [rsi+rax*8]
+ dec rcx
+ jnz .next8_v
+
+ add rsp, 64
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_mbloop_filter_horizontal_edge_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+; int count
+;)
+global sym(vp8_mbloop_filter_horizontal_edge_mmx) PRIVATE
+sym(vp8_mbloop_filter_horizontal_edge_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 32 ; reserve 32 bytes
+ %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
+ %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
+
+ movsxd rcx, dword ptr arg(5) ;count
+.next8_mbh:
+ mov rdx, arg(3) ;limit
+ movq mm7, [rdx]
+ mov rdi, rsi ; rdi points to row +1 for indirect addressing
+ add rdi, rax
+
+ ; calculate breakout conditions
+ movq mm2, [rdi+2*rax] ; q3
+
+ movq mm1, [rsi+2*rax] ; q2
+ movq mm6, mm1 ; q2
+ psubusb mm1, mm2 ; q2-=q3
+ psubusb mm2, mm6 ; q3-=q2
+ por mm1, mm2 ; abs(q3-q2)
+ psubusb mm1, mm7
+
+
+ ; mm1 = abs(q3-q2), mm6 =q2, mm7 = limit
+ movq mm4, [rsi+rax] ; q1
+ movq mm3, mm4 ; q1
+ psubusb mm4, mm6 ; q1-=q2
+ psubusb mm6, mm3 ; q2-=q1
+ por mm4, mm6 ; abs(q2-q1)
+ psubusb mm4, mm7
+ por mm1, mm4
+
+
+ ; mm1 = mask, mm3=q1, mm7 = limit
+
+ movq mm4, [rsi] ; q0
+ movq mm0, mm4 ; q0
+ psubusb mm4, mm3 ; q0-=q1
+ psubusb mm3, mm0 ; q1-=q0
+ por mm4, mm3 ; abs(q0-q1)
+ movq t0, mm4 ; save to t0
+ psubusb mm4, mm7
+ por mm1, mm4
+
+
+ ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1)
+
+ neg rax ; negate pitch to deal with above border
+
+ movq mm2, [rsi+4*rax] ; p3
+ movq mm4, [rdi+4*rax] ; p2
+ movq mm5, mm4 ; p2
+ psubusb mm4, mm2 ; p2-=p3
+ psubusb mm2, mm5 ; p3-=p2
+ por mm4, mm2 ; abs(p3 - p2)
+ psubusb mm4, mm7
+ por mm1, mm4
+ ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1)
+
+ movq mm4, [rsi+2*rax] ; p1
+ movq mm3, mm4 ; p1
+ psubusb mm4, mm5 ; p1-=p2
+ psubusb mm5, mm3 ; p2-=p1
+ por mm4, mm5 ; abs(p2 - p1)
+ psubusb mm4, mm7
+ por mm1, mm4
+
+ movq mm2, mm3 ; p1
+
+
+ ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1)
+
+ movq mm4, [rsi+rax] ; p0
+ movq mm5, mm4 ; p0
+ psubusb mm4, mm3 ; p0-=p1
+ psubusb mm3, mm5 ; p1-=p0
+ por mm4, mm3 ; abs(p1 - p0)
+ movq t1, mm4 ; save to t1
+ psubusb mm4, mm7
+ por mm1, mm4
+ ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1) t1 = abs(p1-p0)
+ ; mm5 = p0
+ movq mm3, [rdi] ; q1
+ movq mm4, mm3 ; q1
+ psubusb mm3, mm2 ; q1-=p1
+ psubusb mm2, mm4 ; p1-=q1
+ por mm2, mm3 ; abs(p1-q1)
+ pand mm2, [GLOBAL(tfe)] ; set lsb of each byte to zero
+ psrlw mm2, 1 ; abs(p1-q1)/2
+
+ movq mm6, mm5 ; p0
+ movq mm3, mm0 ; q0
+ psubusb mm5, mm3 ; p0-=q0
+ psubusb mm3, mm6 ; q0-=p0
+ por mm5, mm3 ; abs(p0 - q0)
+ paddusb mm5, mm5 ; abs(p0-q0)*2
+ paddusb mm5, mm2 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+
+ mov rdx, arg(2) ;blimit ; get blimit
+ movq mm7, [rdx] ; blimit
+
+ psubusb mm5, mm7 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ por mm1, mm5
+ pxor mm5, mm5
+ pcmpeqb mm1, mm5 ; mask mm1
+
+ ; mm1 = mask, mm0=q0, mm7 = blimit, t0 = abs(q0-q1) t1 = abs(p1-p0)
+ ; mm6 = p0,
+
+ ; calculate high edge variance
+ mov rdx, arg(4) ;thresh ; get thresh
+ movq mm7, [rdx] ;
+ movq mm4, t0 ; get abs (q1 - q0)
+ psubusb mm4, mm7
+ movq mm3, t1 ; get abs (p1 - p0)
+ psubusb mm3, mm7
+ paddb mm4, mm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
+
+ pcmpeqb mm4, mm5
+
+ pcmpeqb mm5, mm5
+ pxor mm4, mm5
+
+
+
+ ; mm1 = mask, mm0=q0, mm7 = thresh, t0 = abs(q0-q1) t1 = abs(p1-p0)
+ ; mm6 = p0, mm4=hev
+ ; start work on filters
+ movq mm2, [rsi+2*rax] ; p1
+ movq mm7, [rdi] ; q1
+ pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
+ pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
+ psubsb mm2, mm7 ; p1 - q1
+
+ pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
+ pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
+ movq mm3, mm0 ; q0
+ psubsb mm0, mm6 ; q0 - p0
+ paddsb mm2, mm0 ; 1 * (q0 - p0) + (p1 - q1)
+ paddsb mm2, mm0 ; 2 * (q0 - p0)
+ paddsb mm2, mm0 ; 3 * (q0 - p0) + (p1 - q1)
+ pand mm1, mm2 ; mask filter values we don't care about
+
+
+ ; mm1 = vp8_filter, mm4=hev, mm6=ps0, mm3=qs0
+ movq mm2, mm1 ; vp8_filter
+ pand mm2, mm4; ; Filter2 = vp8_filter & hev
+
+ movq mm5, mm2 ;
+ paddsb mm5, [GLOBAL(t3)];
+
+ pxor mm0, mm0 ; 0
+ pxor mm7, mm7 ; 0
+
+ punpcklbw mm0, mm5 ; e0f0g0h0
+ psraw mm0, 11 ; sign extended shift right by 3
+ punpckhbw mm7, mm5 ; a0b0c0d0
+ psraw mm7, 11 ; sign extended shift right by 3
+ packsswb mm0, mm7 ; Filter2 >>=3;
+
+ movq mm5, mm0 ; Filter2
+
+ paddsb mm2, [GLOBAL(t4)] ; vp8_signed_char_clamp(Filter2 + 4)
+ pxor mm0, mm0 ; 0
+ pxor mm7, mm7 ; 0
+
+ punpcklbw mm0, mm2 ; e0f0g0h0
+ psraw mm0, 11 ; sign extended shift right by 3
+ punpckhbw mm7, mm2 ; a0b0c0d0
+ psraw mm7, 11 ; sign extended shift right by 3
+ packsswb mm0, mm7 ; Filter2 >>=3;
+
+ ; mm0= filter2 mm1 = vp8_filter, mm3 =qs0 mm5=s mm4 =hev mm6=ps0
+ psubsb mm3, mm0 ; qs0 =qs0 - filter1
+ paddsb mm6, mm5 ; ps0 =ps0 + Fitler2
+
+ ; mm1=vp8_filter, mm3=qs0, mm4 =hev mm6=ps0
+ ; vp8_filter &= ~hev;
+ ; Filter2 = vp8_filter;
+ pandn mm4, mm1 ; vp8_filter&=~hev
+
+
+ ; mm3=qs0, mm4=filter2, mm6=ps0
+
+ ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7);
+ ; s = vp8_signed_char_clamp(qs0 - u);
+ ; *oq0 = s^0x80;
+ ; s = vp8_signed_char_clamp(ps0 + u);
+ ; *op0 = s^0x80;
+ pxor mm0, mm0
+
+ pxor mm1, mm1
+ pxor mm2, mm2
+ punpcklbw mm1, mm4
+ punpckhbw mm2, mm4
+ pmulhw mm1, [GLOBAL(s27)]
+ pmulhw mm2, [GLOBAL(s27)]
+ paddw mm1, [GLOBAL(s63)]
+ paddw mm2, [GLOBAL(s63)]
+ psraw mm1, 7
+ psraw mm2, 7
+ packsswb mm1, mm2
+
+ psubsb mm3, mm1
+ paddsb mm6, mm1
+
+ pxor mm3, [GLOBAL(t80)]
+ pxor mm6, [GLOBAL(t80)]
+ movq [rsi+rax], mm6
+ movq [rsi], mm3
+
+ ; roughly 2/7th difference across boundary
+ ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7);
+ ; s = vp8_signed_char_clamp(qs1 - u);
+ ; *oq1 = s^0x80;
+ ; s = vp8_signed_char_clamp(ps1 + u);
+ ; *op1 = s^0x80;
+ pxor mm1, mm1
+ pxor mm2, mm2
+ punpcklbw mm1, mm4
+ punpckhbw mm2, mm4
+ pmulhw mm1, [GLOBAL(s18)]
+ pmulhw mm2, [GLOBAL(s18)]
+ paddw mm1, [GLOBAL(s63)]
+ paddw mm2, [GLOBAL(s63)]
+ psraw mm1, 7
+ psraw mm2, 7
+ packsswb mm1, mm2
+
+ movq mm3, [rdi]
+ movq mm6, [rsi+rax*2] ; p1
+
+ pxor mm3, [GLOBAL(t80)]
+ pxor mm6, [GLOBAL(t80)]
+
+ paddsb mm6, mm1
+ psubsb mm3, mm1
+
+ pxor mm6, [GLOBAL(t80)]
+ pxor mm3, [GLOBAL(t80)]
+ movq [rdi], mm3
+ movq [rsi+rax*2], mm6
+
+ ; roughly 1/7th difference across boundary
+ ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7);
+ ; s = vp8_signed_char_clamp(qs2 - u);
+ ; *oq2 = s^0x80;
+ ; s = vp8_signed_char_clamp(ps2 + u);
+ ; *op2 = s^0x80;
+ pxor mm1, mm1
+ pxor mm2, mm2
+ punpcklbw mm1, mm4
+ punpckhbw mm2, mm4
+ pmulhw mm1, [GLOBAL(s9)]
+ pmulhw mm2, [GLOBAL(s9)]
+ paddw mm1, [GLOBAL(s63)]
+ paddw mm2, [GLOBAL(s63)]
+ psraw mm1, 7
+ psraw mm2, 7
+ packsswb mm1, mm2
+
+
+ movq mm6, [rdi+rax*4]
+ neg rax
+ movq mm3, [rdi+rax ]
+
+ pxor mm6, [GLOBAL(t80)]
+ pxor mm3, [GLOBAL(t80)]
+
+ paddsb mm6, mm1
+ psubsb mm3, mm1
+
+ pxor mm6, [GLOBAL(t80)]
+ pxor mm3, [GLOBAL(t80)]
+ movq [rdi+rax ], mm3
+ neg rax
+ movq [rdi+rax*4], mm6
+
+;EARLY_BREAK_OUT:
+ neg rax
+ add rsi,8
+ dec rcx
+ jnz .next8_mbh
+
+ add rsp, 32
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_mbloop_filter_vertical_edge_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+; int count
+;)
+global sym(vp8_mbloop_filter_vertical_edge_mmx) PRIVATE
+sym(vp8_mbloop_filter_vertical_edge_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 96 ; reserve 96 bytes
+ %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
+ %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
+ %define srct [rsp + 32] ;__declspec(align(16)) char srct[64];
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
+
+ lea rsi, [rsi + rax*4 - 4]
+
+ movsxd rcx, dword ptr arg(5) ;count
+.next8_mbv:
+ lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
+
+ ;transpose
+ movq mm0, [rdi+2*rax] ; 77 76 75 74 73 72 71 70
+ movq mm6, [rsi+2*rax] ; 67 66 65 64 63 62 61 60
+
+ movq mm7, mm6 ; 77 76 75 74 73 72 71 70
+ punpckhbw mm7, mm0 ; 77 67 76 66 75 65 74 64
+
+ punpcklbw mm6, mm0 ; 73 63 72 62 71 61 70 60
+ movq mm0, [rsi+rax] ; 57 56 55 54 53 52 51 50
+
+ movq mm4, [rsi] ; 47 46 45 44 43 42 41 40
+ movq mm5, mm4 ; 47 46 45 44 43 42 41 40
+
+ punpckhbw mm5, mm0 ; 57 47 56 46 55 45 54 44
+ punpcklbw mm4, mm0 ; 53 43 52 42 51 41 50 40
+
+ movq mm3, mm5 ; 57 47 56 46 55 45 54 44
+ punpckhwd mm5, mm7 ; 77 67 57 47 76 66 56 46
+
+ punpcklwd mm3, mm7 ; 75 65 55 45 74 64 54 44
+ movq mm2, mm4 ; 53 43 52 42 51 41 50 40
+
+ punpckhwd mm4, mm6 ; 73 63 53 43 72 62 52 42
+ punpcklwd mm2, mm6 ; 71 61 51 41 70 60 50 40
+
+ neg rax
+
+ movq mm7, [rsi+rax] ; 37 36 35 34 33 32 31 30
+ movq mm6, [rsi+rax*2] ; 27 26 25 24 23 22 21 20
+
+ movq mm1, mm6 ; 27 26 25 24 23 22 21 20
+ punpckhbw mm6, mm7 ; 37 27 36 36 35 25 34 24
+
+ punpcklbw mm1, mm7 ; 33 23 32 22 31 21 30 20
+
+ movq mm7, [rsi+rax*4]; ; 07 06 05 04 03 02 01 00
+ punpckhbw mm7, [rdi+rax*4] ; 17 07 16 06 15 05 14 04
+
+ movq mm0, mm7 ; 17 07 16 06 15 05 14 04
+ punpckhwd mm7, mm6 ; 37 27 17 07 36 26 16 06
+
+ punpcklwd mm0, mm6 ; 35 25 15 05 34 24 14 04
+ movq mm6, mm7 ; 37 27 17 07 36 26 16 06
+
+ punpckhdq mm7, mm5 ; 77 67 57 47 37 27 17 07 = q3
+ punpckldq mm6, mm5 ; 76 66 56 46 36 26 16 06 = q2
+
+ lea rdx, srct
+ movq mm5, mm6 ; 76 66 56 46 36 26 16 06
+
+ movq [rdx+56], mm7
+ psubusb mm5, mm7 ; q2-q3
+
+
+ movq [rdx+48], mm6
+ psubusb mm7, mm6 ; q3-q2
+
+ por mm7, mm5; ; mm7=abs (q3-q2)
+ movq mm5, mm0 ; 35 25 15 05 34 24 14 04
+
+ punpckhdq mm5, mm3 ; 75 65 55 45 35 25 15 05 = q1
+ punpckldq mm0, mm3 ; 74 64 54 44 34 24 15 04 = q0
+
+ movq mm3, mm5 ; 75 65 55 45 35 25 15 05 = q1
+ psubusb mm3, mm6 ; q1-q2
+
+ psubusb mm6, mm5 ; q2-q1
+ por mm6, mm3 ; mm6=abs(q2-q1)
+
+ movq [rdx+40], mm5 ; save q1
+ movq [rdx+32], mm0 ; save q0
+
+ movq mm3, [rsi+rax*4] ; 07 06 05 04 03 02 01 00
+ punpcklbw mm3, [rdi+rax*4] ; 13 03 12 02 11 01 10 00
+
+ movq mm0, mm3 ; 13 03 12 02 11 01 10 00
+ punpcklwd mm0, mm1 ; 31 21 11 01 30 20 10 00
+
+ punpckhwd mm3, mm1 ; 33 23 13 03 32 22 12 02
+ movq mm1, mm0 ; 31 21 11 01 30 20 10 00
+
+ punpckldq mm0, mm2 ; 70 60 50 40 30 20 10 00 =p3
+ punpckhdq mm1, mm2 ; 71 61 51 41 31 21 11 01 =p2
+
+ movq [rdx], mm0 ; save p3
+ movq [rdx+8], mm1 ; save p2
+
+ movq mm2, mm1 ; 71 61 51 41 31 21 11 01 =p2
+ psubusb mm2, mm0 ; p2-p3
+
+ psubusb mm0, mm1 ; p3-p2
+ por mm0, mm2 ; mm0=abs(p3-p2)
+
+ movq mm2, mm3 ; 33 23 13 03 32 22 12 02
+ punpckldq mm2, mm4 ; 72 62 52 42 32 22 12 02 = p1
+
+ punpckhdq mm3, mm4 ; 73 63 53 43 33 23 13 03 = p0
+ movq [rdx+24], mm3 ; save p0
+
+ movq [rdx+16], mm2 ; save p1
+ movq mm5, mm2 ; mm5 = p1
+
+ psubusb mm2, mm1 ; p1-p2
+ psubusb mm1, mm5 ; p2-p1
+
+ por mm1, mm2 ; mm1=abs(p2-p1)
+ mov rdx, arg(3) ;limit
+
+ movq mm4, [rdx] ; mm4 = limit
+ psubusb mm7, mm4 ; abs(q3-q2) > limit
+
+ psubusb mm0, mm4 ; abs(p3-p2) > limit
+ psubusb mm1, mm4 ; abs(p2-p1) > limit
+
+ psubusb mm6, mm4 ; abs(q2-q1) > limit
+ por mm7, mm6 ; or
+
+ por mm0, mm1 ;
+ por mm0, mm7 ; abs(q3-q2) > limit || abs(p3-p2) > limit ||abs(p2-p1) > limit || abs(q2-q1) > limit
+
+ movq mm1, mm5 ; p1
+
+ movq mm7, mm3 ; mm3=mm7=p0
+ psubusb mm7, mm5 ; p0 - p1
+
+ psubusb mm5, mm3 ; p1 - p0
+ por mm5, mm7 ; abs(p1-p0)
+
+ movq t0, mm5 ; save abs(p1-p0)
+ lea rdx, srct
+
+ psubusb mm5, mm4 ; mm5 = abs(p1-p0) > limit
+ por mm0, mm5 ; mm0=mask
+
+ movq mm5, [rdx+32] ; mm5=q0
+ movq mm7, [rdx+40] ; mm7=q1
+
+ movq mm6, mm5 ; mm6=q0
+ movq mm2, mm7 ; q1
+ psubusb mm5, mm7 ; q0-q1
+
+ psubusb mm7, mm6 ; q1-q0
+ por mm7, mm5 ; abs(q1-q0)
+
+ movq t1, mm7 ; save abs(q1-q0)
+ psubusb mm7, mm4 ; mm7=abs(q1-q0)> limit
+
+ por mm0, mm7 ; mask
+
+ movq mm5, mm2 ; q1
+ psubusb mm5, mm1 ; q1-=p1
+ psubusb mm1, mm2 ; p1-=q1
+ por mm5, mm1 ; abs(p1-q1)
+ pand mm5, [GLOBAL(tfe)] ; set lsb of each byte to zero
+ psrlw mm5, 1 ; abs(p1-q1)/2
+
+ mov rdx, arg(2) ;blimit ;
+
+ movq mm4, [rdx] ;blimit
+ movq mm1, mm3 ; mm1=mm3=p0
+
+ movq mm7, mm6 ; mm7=mm6=q0
+ psubusb mm1, mm7 ; p0-q0
+
+ psubusb mm7, mm3 ; q0-p0
+ por mm1, mm7 ; abs(q0-p0)
+ paddusb mm1, mm1 ; abs(q0-p0)*2
+ paddusb mm1, mm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+
+ psubusb mm1, mm4 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ por mm1, mm0; ; mask
+
+ pxor mm0, mm0
+ pcmpeqb mm1, mm0
+
+ ; calculate high edge variance
+ mov rdx, arg(4) ;thresh ; get thresh
+ movq mm7, [rdx]
+ ;
+ movq mm4, t0 ; get abs (q1 - q0)
+ psubusb mm4, mm7 ; abs(q1 - q0) > thresh
+
+ movq mm3, t1 ; get abs (p1 - p0)
+ psubusb mm3, mm7 ; abs(p1 - p0)> thresh
+
+ por mm4, mm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
+ pcmpeqb mm4, mm0
+
+ pcmpeqb mm0, mm0
+ pxor mm4, mm0
+
+
+
+
+ ; start work on filters
+ lea rdx, srct
+
+ ; start work on filters
+ movq mm2, [rdx+16] ; p1
+ movq mm7, [rdx+40] ; q1
+ pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
+ pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
+ psubsb mm2, mm7 ; p1 - q1
+
+ movq mm6, [rdx+24] ; p0
+ movq mm0, [rdx+32] ; q0
+ pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
+ pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
+
+ movq mm3, mm0 ; q0
+ psubsb mm0, mm6 ; q0 - p0
+ paddsb mm2, mm0 ; 1 * (q0 - p0) + (p1 - q1)
+ paddsb mm2, mm0 ; 2 * (q0 - p0)
+ paddsb mm2, mm0 ; 3 * (q0 - p0) + (p1 - q1)
+ pand mm1, mm2 ; mask filter values we don't care about
+
+ ; mm1 = vp8_filter, mm4=hev, mm6=ps0, mm3=qs0
+ movq mm2, mm1 ; vp8_filter
+ pand mm2, mm4; ; Filter2 = vp8_filter & hev
+
+ movq mm5, mm2 ;
+ paddsb mm5, [GLOBAL(t3)];
+
+ pxor mm0, mm0 ; 0
+ pxor mm7, mm7 ; 0
+
+ punpcklbw mm0, mm5 ; e0f0g0h0
+ psraw mm0, 11 ; sign extended shift right by 3
+ punpckhbw mm7, mm5 ; a0b0c0d0
+ psraw mm7, 11 ; sign extended shift right by 3
+ packsswb mm0, mm7 ; Filter2 >>=3;
+
+ movq mm5, mm0 ; Filter2
+
+ paddsb mm2, [GLOBAL(t4)] ; vp8_signed_char_clamp(Filter2 + 4)
+ pxor mm0, mm0 ; 0
+ pxor mm7, mm7 ; 0
+
+ punpcklbw mm0, mm2 ; e0f0g0h0
+ psraw mm0, 11 ; sign extended shift right by 3
+ punpckhbw mm7, mm2 ; a0b0c0d0
+ psraw mm7, 11 ; sign extended shift right by 3
+ packsswb mm0, mm7 ; Filter2 >>=3;
+
+ ; mm0= filter2 mm1 = vp8_filter, mm3 =qs0 mm5=s mm4 =hev mm6=ps0
+ psubsb mm3, mm0 ; qs0 =qs0 - filter1
+ paddsb mm6, mm5 ; ps0 =ps0 + Fitler2
+
+ ; mm1=vp8_filter, mm3=qs0, mm4 =hev mm6=ps0
+ ; vp8_filter &= ~hev;
+ ; Filter2 = vp8_filter;
+ pandn mm4, mm1 ; vp8_filter&=~hev
+
+
+ ; mm3=qs0, mm4=filter2, mm6=ps0
+
+ ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7);
+ ; s = vp8_signed_char_clamp(qs0 - u);
+ ; *oq0 = s^0x80;
+ ; s = vp8_signed_char_clamp(ps0 + u);
+ ; *op0 = s^0x80;
+ pxor mm0, mm0
+
+ pxor mm1, mm1
+ pxor mm2, mm2
+ punpcklbw mm1, mm4
+ punpckhbw mm2, mm4
+ pmulhw mm1, [GLOBAL(s27)]
+ pmulhw mm2, [GLOBAL(s27)]
+ paddw mm1, [GLOBAL(s63)]
+ paddw mm2, [GLOBAL(s63)]
+ psraw mm1, 7
+ psraw mm2, 7
+ packsswb mm1, mm2
+
+ psubsb mm3, mm1
+ paddsb mm6, mm1
+
+ pxor mm3, [GLOBAL(t80)]
+ pxor mm6, [GLOBAL(t80)]
+ movq [rdx+24], mm6
+ movq [rdx+32], mm3
+
+ ; roughly 2/7th difference across boundary
+ ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7);
+ ; s = vp8_signed_char_clamp(qs1 - u);
+ ; *oq1 = s^0x80;
+ ; s = vp8_signed_char_clamp(ps1 + u);
+ ; *op1 = s^0x80;
+ pxor mm1, mm1
+ pxor mm2, mm2
+ punpcklbw mm1, mm4
+ punpckhbw mm2, mm4
+ pmulhw mm1, [GLOBAL(s18)]
+ pmulhw mm2, [GLOBAL(s18)]
+ paddw mm1, [GLOBAL(s63)]
+ paddw mm2, [GLOBAL(s63)]
+ psraw mm1, 7
+ psraw mm2, 7
+ packsswb mm1, mm2
+
+ movq mm3, [rdx + 40]
+ movq mm6, [rdx + 16] ; p1
+ pxor mm3, [GLOBAL(t80)]
+ pxor mm6, [GLOBAL(t80)]
+
+ paddsb mm6, mm1
+ psubsb mm3, mm1
+
+ pxor mm6, [GLOBAL(t80)]
+ pxor mm3, [GLOBAL(t80)]
+ movq [rdx + 40], mm3
+ movq [rdx + 16], mm6
+
+ ; roughly 1/7th difference across boundary
+ ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7);
+ ; s = vp8_signed_char_clamp(qs2 - u);
+ ; *oq2 = s^0x80;
+ ; s = vp8_signed_char_clamp(ps2 + u);
+ ; *op2 = s^0x80;
+ pxor mm1, mm1
+ pxor mm2, mm2
+ punpcklbw mm1, mm4
+ punpckhbw mm2, mm4
+ pmulhw mm1, [GLOBAL(s9)]
+ pmulhw mm2, [GLOBAL(s9)]
+ paddw mm1, [GLOBAL(s63)]
+ paddw mm2, [GLOBAL(s63)]
+ psraw mm1, 7
+ psraw mm2, 7
+ packsswb mm1, mm2
+
+ movq mm6, [rdx+ 8]
+ movq mm3, [rdx+48]
+
+ pxor mm6, [GLOBAL(t80)]
+ pxor mm3, [GLOBAL(t80)]
+
+ paddsb mm6, mm1
+ psubsb mm3, mm1
+
+ pxor mm6, [GLOBAL(t80)] ; mm6 = 71 61 51 41 31 21 11 01
+ pxor mm3, [GLOBAL(t80)] ; mm3 = 76 66 56 46 36 26 15 06
+
+ ; tranpose and write back
+ movq mm0, [rdx] ; mm0 = 70 60 50 40 30 20 10 00
+ movq mm1, mm0 ; mm0 = 70 60 50 40 30 20 10 00
+
+ punpcklbw mm0, mm6 ; mm0 = 31 30 21 20 11 10 01 00
+ punpckhbw mm1, mm6 ; mm3 = 71 70 61 60 51 50 41 40
+
+ movq mm2, [rdx+16] ; mm2 = 72 62 52 42 32 22 12 02
+ movq mm6, mm2 ; mm3 = 72 62 52 42 32 22 12 02
+
+ punpcklbw mm2, [rdx+24] ; mm2 = 33 32 23 22 13 12 03 02
+ punpckhbw mm6, [rdx+24] ; mm3 = 73 72 63 62 53 52 43 42
+
+ movq mm5, mm0 ; mm5 = 31 30 21 20 11 10 01 00
+ punpcklwd mm0, mm2 ; mm0 = 13 12 11 10 03 02 01 00
+
+ punpckhwd mm5, mm2 ; mm5 = 33 32 31 30 23 22 21 20
+ movq mm4, mm1 ; mm4 = 71 70 61 60 51 50 41 40
+
+ punpcklwd mm1, mm6 ; mm1 = 53 52 51 50 43 42 41 40
+ punpckhwd mm4, mm6 ; mm4 = 73 72 71 70 63 62 61 60
+
+ movq mm2, [rdx+32] ; mm2 = 74 64 54 44 34 24 14 04
+ punpcklbw mm2, [rdx+40] ; mm2 = 35 34 25 24 15 14 05 04
+
+ movq mm6, mm3 ; mm6 = 76 66 56 46 36 26 15 06
+ punpcklbw mm6, [rdx+56] ; mm6 = 37 36 27 26 17 16 07 06
+
+ movq mm7, mm2 ; mm7 = 35 34 25 24 15 14 05 04
+ punpcklwd mm2, mm6 ; mm2 = 17 16 15 14 07 06 05 04
+
+ punpckhwd mm7, mm6 ; mm7 = 37 36 35 34 27 26 25 24
+ movq mm6, mm0 ; mm6 = 13 12 11 10 03 02 01 00
+
+ punpckldq mm0, mm2 ; mm0 = 07 06 05 04 03 02 01 00
+ punpckhdq mm6, mm2 ; mm6 = 17 16 15 14 13 12 11 10
+
+ movq [rsi+rax*4], mm0 ; write out
+ movq [rdi+rax*4], mm6 ; write out
+
+ movq mm0, mm5 ; mm0 = 33 32 31 30 23 22 21 20
+ punpckldq mm0, mm7 ; mm0 = 27 26 25 24 23 22 20 20
+
+ punpckhdq mm5, mm7 ; mm5 = 37 36 35 34 33 32 31 30
+ movq [rsi+rax*2], mm0 ; write out
+
+ movq [rdi+rax*2], mm5 ; write out
+ movq mm2, [rdx+32] ; mm2 = 74 64 54 44 34 24 14 04
+
+ punpckhbw mm2, [rdx+40] ; mm2 = 75 74 65 64 54 54 45 44
+ punpckhbw mm3, [rdx+56] ; mm3 = 77 76 67 66 57 56 47 46
+
+ movq mm5, mm2 ; mm5 = 75 74 65 64 54 54 45 44
+ punpcklwd mm2, mm3 ; mm2 = 57 56 55 54 47 46 45 44
+
+ punpckhwd mm5, mm3 ; mm5 = 77 76 75 74 67 66 65 64
+ movq mm0, mm1 ; mm0= 53 52 51 50 43 42 41 40
+
+ movq mm3, mm4 ; mm4 = 73 72 71 70 63 62 61 60
+ punpckldq mm0, mm2 ; mm0 = 47 46 45 44 43 42 41 40
+
+ punpckhdq mm1, mm2 ; mm1 = 57 56 55 54 53 52 51 50
+ movq [rsi], mm0 ; write out
+
+ movq [rdi], mm1 ; write out
+ neg rax
+
+ punpckldq mm3, mm5 ; mm3 = 67 66 65 64 63 62 61 60
+ punpckhdq mm4, mm5 ; mm4 = 77 76 75 74 73 72 71 60
+
+ movq [rsi+rax*2], mm3
+ movq [rdi+rax*2], mm4
+
+ lea rsi, [rsi+rax*8]
+ dec rcx
+
+ jnz .next8_mbv
+
+ add rsp, 96
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_loop_filter_simple_horizontal_edge_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit
+;)
+global sym(vp8_loop_filter_simple_horizontal_edge_mmx) PRIVATE
+sym(vp8_loop_filter_simple_horizontal_edge_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 3
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
+
+ mov rcx, 2 ; count
+.nexts8_h:
+ mov rdx, arg(2) ;blimit ; get blimit
+ movq mm3, [rdx] ;
+
+ mov rdi, rsi ; rdi points to row +1 for indirect addressing
+ add rdi, rax
+ neg rax
+
+ ; calculate mask
+ movq mm1, [rsi+2*rax] ; p1
+ movq mm0, [rdi] ; q1
+ movq mm2, mm1
+ movq mm7, mm0
+ movq mm4, mm0
+ psubusb mm0, mm1 ; q1-=p1
+ psubusb mm1, mm4 ; p1-=q1
+ por mm1, mm0 ; abs(p1-q1)
+ pand mm1, [GLOBAL(tfe)] ; set lsb of each byte to zero
+ psrlw mm1, 1 ; abs(p1-q1)/2
+
+ movq mm5, [rsi+rax] ; p0
+ movq mm4, [rsi] ; q0
+ movq mm0, mm4 ; q0
+ movq mm6, mm5 ; p0
+ psubusb mm5, mm4 ; p0-=q0
+ psubusb mm4, mm6 ; q0-=p0
+ por mm5, mm4 ; abs(p0 - q0)
+ paddusb mm5, mm5 ; abs(p0-q0)*2
+ paddusb mm5, mm1 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+
+ psubusb mm5, mm3 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ pxor mm3, mm3
+ pcmpeqb mm5, mm3
+
+ ; start work on filters
+ pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
+ pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
+ psubsb mm2, mm7 ; p1 - q1
+
+ pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
+ pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
+ movq mm3, mm0 ; q0
+ psubsb mm0, mm6 ; q0 - p0
+ paddsb mm2, mm0 ; p1 - q1 + 1 * (q0 - p0)
+ paddsb mm2, mm0 ; p1 - q1 + 2 * (q0 - p0)
+ paddsb mm2, mm0 ; p1 - q1 + 3 * (q0 - p0)
+ pand mm5, mm2 ; mask filter values we don't care about
+
+ ; do + 4 side
+ paddsb mm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
+
+ movq mm0, mm5 ; get a copy of filters
+ psllw mm0, 8 ; shift left 8
+ psraw mm0, 3 ; arithmetic shift right 11
+ psrlw mm0, 8
+ movq mm1, mm5 ; get a copy of filters
+ psraw mm1, 11 ; arithmetic shift right 11
+ psllw mm1, 8 ; shift left 8 to put it back
+
+ por mm0, mm1 ; put the two together to get result
+
+ psubsb mm3, mm0 ; q0-= q0 add
+ pxor mm3, [GLOBAL(t80)] ; unoffset
+ movq [rsi], mm3 ; write back
+
+
+ ; now do +3 side
+ psubsb mm5, [GLOBAL(t1s)] ; +3 instead of +4
+
+ movq mm0, mm5 ; get a copy of filters
+ psllw mm0, 8 ; shift left 8
+ psraw mm0, 3 ; arithmetic shift right 11
+ psrlw mm0, 8
+ psraw mm5, 11 ; arithmetic shift right 11
+ psllw mm5, 8 ; shift left 8 to put it back
+ por mm0, mm5 ; put the two together to get result
+
+
+ paddsb mm6, mm0 ; p0+= p0 add
+ pxor mm6, [GLOBAL(t80)] ; unoffset
+ movq [rsi+rax], mm6 ; write back
+
+ add rsi,8
+ neg rax
+ dec rcx
+ jnz .nexts8_h
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_loop_filter_simple_vertical_edge_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit
+;)
+global sym(vp8_loop_filter_simple_vertical_edge_mmx) PRIVATE
+sym(vp8_loop_filter_simple_vertical_edge_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 3
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 32 ; reserve 32 bytes
+ %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
+ %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
+
+ lea rsi, [rsi + rax*4- 2]; ;
+ mov rcx, 2 ; count
+.nexts8_v:
+
+ lea rdi, [rsi + rax];
+ movd mm0, [rdi + rax * 2] ; xx xx xx xx 73 72 71 70
+
+ movd mm6, [rsi + rax * 2] ; xx xx xx xx 63 62 61 60
+ punpcklbw mm6, mm0 ; 73 63 72 62 71 61 70 60
+
+ movd mm0, [rsi + rax] ; xx xx xx xx 53 52 51 50
+ movd mm4, [rsi] ; xx xx xx xx 43 42 41 40
+
+ punpcklbw mm4, mm0 ; 53 43 52 42 51 41 50 40
+ movq mm5, mm4 ; 53 43 52 42 51 41 50 40
+
+ punpcklwd mm4, mm6 ; 71 61 51 41 70 60 50 40
+ punpckhwd mm5, mm6 ; 73 63 53 43 72 62 52 42
+
+ neg rax
+
+ movd mm7, [rsi + rax] ; xx xx xx xx 33 32 31 30
+ movd mm6, [rsi + rax * 2] ; xx xx xx xx 23 22 21 20
+
+ punpcklbw mm6, mm7 ; 33 23 32 22 31 21 30 20
+ movd mm1, [rdi + rax * 4] ; xx xx xx xx 13 12 11 10
+
+ movd mm0, [rsi + rax * 4] ; xx xx xx xx 03 02 01 00
+ punpcklbw mm0, mm1 ; 13 03 12 02 11 01 10 00
+
+ movq mm2, mm0 ; 13 03 12 02 11 01 10 00
+ punpcklwd mm0, mm6 ; 31 21 11 01 30 20 10 00
+
+ punpckhwd mm2, mm6 ; 33 23 13 03 32 22 12 02
+ movq mm1, mm0 ; 13 03 12 02 11 01 10 00
+
+ punpckldq mm0, mm4 ; 70 60 50 40 30 20 10 00 = p1
+ movq mm3, mm2 ; 33 23 13 03 32 22 12 02
+
+ punpckhdq mm1, mm4 ; 71 61 51 41 31 21 11 01 = p0
+ punpckldq mm2, mm5 ; 72 62 52 42 32 22 12 02 = q0
+
+ punpckhdq mm3, mm5 ; 73 63 53 43 33 23 13 03 = q1
+
+
+ ; calculate mask
+ movq mm6, mm0 ; p1
+ movq mm7, mm3 ; q1
+ psubusb mm7, mm6 ; q1-=p1
+ psubusb mm6, mm3 ; p1-=q1
+ por mm6, mm7 ; abs(p1-q1)
+ pand mm6, [GLOBAL(tfe)] ; set lsb of each byte to zero
+ psrlw mm6, 1 ; abs(p1-q1)/2
+
+ movq mm5, mm1 ; p0
+ movq mm4, mm2 ; q0
+
+ psubusb mm5, mm2 ; p0-=q0
+ psubusb mm4, mm1 ; q0-=p0
+
+ por mm5, mm4 ; abs(p0 - q0)
+ paddusb mm5, mm5 ; abs(p0-q0)*2
+ paddusb mm5, mm6 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+
+ mov rdx, arg(2) ;blimit ; get blimit
+ movq mm7, [rdx]
+
+ psubusb mm5, mm7 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ pxor mm7, mm7
+ pcmpeqb mm5, mm7 ; mm5 = mask
+
+ ; start work on filters
+ movq t0, mm0
+ movq t1, mm3
+
+ pxor mm0, [GLOBAL(t80)] ; p1 offset to convert to signed values
+ pxor mm3, [GLOBAL(t80)] ; q1 offset to convert to signed values
+
+ psubsb mm0, mm3 ; p1 - q1
+ movq mm6, mm1 ; p0
+
+ movq mm7, mm2 ; q0
+ pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
+
+ pxor mm7, [GLOBAL(t80)] ; offset to convert to signed values
+ movq mm3, mm7 ; offseted ; q0
+
+ psubsb mm7, mm6 ; q0 - p0
+ paddsb mm0, mm7 ; p1 - q1 + 1 * (q0 - p0)
+
+ paddsb mm0, mm7 ; p1 - q1 + 2 * (q0 - p0)
+ paddsb mm0, mm7 ; p1 - q1 + 3 * (q0 - p0)
+
+ pand mm5, mm0 ; mask filter values we don't care about
+
+ paddsb mm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
+
+ movq mm0, mm5 ; get a copy of filters
+ psllw mm0, 8 ; shift left 8
+ psraw mm0, 3 ; arithmetic shift right 11
+ psrlw mm0, 8
+
+ movq mm7, mm5 ; get a copy of filters
+ psraw mm7, 11 ; arithmetic shift right 11
+ psllw mm7, 8 ; shift left 8 to put it back
+
+ por mm0, mm7 ; put the two together to get result
+
+ psubsb mm3, mm0 ; q0-= q0sz add
+ pxor mm3, [GLOBAL(t80)] ; unoffset
+
+ ; now do +3 side
+ psubsb mm5, [GLOBAL(t1s)] ; +3 instead of +4
+
+ movq mm0, mm5 ; get a copy of filters
+ psllw mm0, 8 ; shift left 8
+ psraw mm0, 3 ; arithmetic shift right 11
+ psrlw mm0, 8
+
+ psraw mm5, 11 ; arithmetic shift right 11
+ psllw mm5, 8 ; shift left 8 to put it back
+ por mm0, mm5 ; put the two together to get result
+
+ paddsb mm6, mm0 ; p0+= p0 add
+ pxor mm6, [GLOBAL(t80)] ; unoffset
+
+
+ movq mm0, t0
+ movq mm4, t1
+
+ ; mm0 = 70 60 50 40 30 20 10 00
+ ; mm6 = 71 61 51 41 31 21 11 01
+ ; mm3 = 72 62 52 42 32 22 12 02
+ ; mm4 = 73 63 53 43 33 23 13 03
+ ; transpose back to write out
+
+ movq mm1, mm0 ;
+ punpcklbw mm0, mm6 ; 31 30 21 20 11 10 01 00
+
+ punpckhbw mm1, mm6 ; 71 70 61 60 51 50 41 40
+ movq mm2, mm3 ;
+
+ punpcklbw mm2, mm4 ; 33 32 23 22 13 12 03 02
+ movq mm5, mm1 ; 71 70 61 60 51 50 41 40
+
+ punpckhbw mm3, mm4 ; 73 72 63 62 53 52 43 42
+ movq mm6, mm0 ; 31 30 21 20 11 10 01 00
+
+ punpcklwd mm0, mm2 ; 13 12 11 10 03 02 01 00
+ punpckhwd mm6, mm2 ; 33 32 31 30 23 22 21 20
+
+ movd [rsi+rax*4], mm0 ; write 03 02 01 00
+ punpcklwd mm1, mm3 ; 53 52 51 50 43 42 41 40
+
+ psrlq mm0, 32 ; xx xx xx xx 13 12 11 10
+ punpckhwd mm5, mm3 ; 73 72 71 70 63 62 61 60
+
+ movd [rdi+rax*4], mm0 ; write 13 12 11 10
+ movd [rsi+rax*2], mm6 ; write 23 22 21 20
+
+ psrlq mm6, 32 ; 33 32 31 30
+ movd [rsi], mm1 ; write 43 42 41 40
+
+ movd [rsi + rax], mm6 ; write 33 32 31 30
+ neg rax
+
+ movd [rsi + rax*2], mm5 ; write 63 62 61 60
+ psrlq mm1, 32 ; 53 52 51 50
+
+ movd [rdi], mm1 ; write out 53 52 51 50
+ psrlq mm5, 32 ; 73 72 71 70
+
+ movd [rdi + rax*2], mm5 ; write 73 72 71 70
+
+ lea rsi, [rsi+rax*8] ; next 8
+
+ dec rcx
+ jnz .nexts8_v
+
+ add rsp, 32
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+
+;void fast_loop_filter_vertical_edges_mmx(unsigned char *y_ptr,
+; int y_stride,
+; loop_filter_info *lfi)
+;{
+;
+;
+; vp8_loop_filter_simple_vertical_edge_mmx(y_ptr+4, y_stride, lfi->flim,lfi->lim,lfi->thr,2);
+; vp8_loop_filter_simple_vertical_edge_mmx(y_ptr+8, y_stride, lfi->flim,lfi->lim,lfi->thr,2);
+; vp8_loop_filter_simple_vertical_edge_mmx(y_ptr+12, y_stride, lfi->flim,lfi->lim,lfi->thr,2);
+;}
+
+SECTION_RODATA
+align 16
+tfe:
+ times 8 db 0xfe
+align 16
+t80:
+ times 8 db 0x80
+align 16
+t1s:
+ times 8 db 0x01
+align 16
+t3:
+ times 8 db 0x03
+align 16
+t4:
+ times 8 db 0x04
+align 16
+ones:
+ times 4 dw 0x0001
+align 16
+s27:
+ times 4 dw 0x1b00
+align 16
+s18:
+ times 4 dw 0x1200
+align 16
+s9:
+ times 4 dw 0x0900
+align 16
+s63:
+ times 4 dw 0x003f
diff --git a/libvpx/vp8/common/x86/loopfilter_sse2.asm b/libvpx/vp8/common/x86/loopfilter_sse2.asm
new file mode 100644
index 0000000..a66753b
--- /dev/null
+++ b/libvpx/vp8/common/x86/loopfilter_sse2.asm
@@ -0,0 +1,1640 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+%define _t0 0
+%define _t1 _t0 + 16
+%define _p3 _t1 + 16
+%define _p2 _p3 + 16
+%define _p1 _p2 + 16
+%define _p0 _p1 + 16
+%define _q0 _p0 + 16
+%define _q1 _q0 + 16
+%define _q2 _q1 + 16
+%define _q3 _q2 + 16
+%define lf_var_size 160
+
+; Use of pmaxub instead of psubusb to compute filter mask was seen
+; in ffvp8
+
+%macro LFH_FILTER_AND_HEV_MASK 1
+%if %1
+ movdqa xmm2, [rdi+2*rax] ; q3
+ movdqa xmm1, [rsi+2*rax] ; q2
+ movdqa xmm4, [rsi+rax] ; q1
+ movdqa xmm5, [rsi] ; q0
+ neg rax ; negate pitch to deal with above border
+%else
+ movlps xmm2, [rsi + rcx*2] ; q3
+ movlps xmm1, [rsi + rcx] ; q2
+ movlps xmm4, [rsi] ; q1
+ movlps xmm5, [rsi + rax] ; q0
+
+ movhps xmm2, [rdi + rcx*2]
+ movhps xmm1, [rdi + rcx]
+ movhps xmm4, [rdi]
+ movhps xmm5, [rdi + rax]
+
+ lea rsi, [rsi + rax*4]
+ lea rdi, [rdi + rax*4]
+
+ movdqa [rsp+_q2], xmm1 ; store q2
+ movdqa [rsp+_q1], xmm4 ; store q1
+%endif
+ movdqa xmm7, [rdx] ;limit
+
+ movdqa xmm6, xmm1 ; q2
+ movdqa xmm3, xmm4 ; q1
+
+ psubusb xmm1, xmm2 ; q2-=q3
+ psubusb xmm2, xmm6 ; q3-=q2
+
+ psubusb xmm4, xmm6 ; q1-=q2
+ psubusb xmm6, xmm3 ; q2-=q1
+
+ por xmm4, xmm6 ; abs(q2-q1)
+ por xmm1, xmm2 ; abs(q3-q2)
+
+ movdqa xmm0, xmm5 ; q0
+ pmaxub xmm1, xmm4
+
+ psubusb xmm5, xmm3 ; q0-=q1
+ psubusb xmm3, xmm0 ; q1-=q0
+
+ por xmm5, xmm3 ; abs(q0-q1)
+ movdqa [rsp+_t0], xmm5 ; save to t0
+
+ pmaxub xmm1, xmm5
+
+%if %1
+ movdqa xmm2, [rsi+4*rax] ; p3
+ movdqa xmm4, [rdi+4*rax] ; p2
+ movdqa xmm6, [rsi+2*rax] ; p1
+%else
+ movlps xmm2, [rsi + rax] ; p3
+ movlps xmm4, [rsi] ; p2
+ movlps xmm6, [rsi + rcx] ; p1
+
+ movhps xmm2, [rdi + rax]
+ movhps xmm4, [rdi]
+ movhps xmm6, [rdi + rcx]
+
+ movdqa [rsp+_p2], xmm4 ; store p2
+ movdqa [rsp+_p1], xmm6 ; store p1
+%endif
+
+ movdqa xmm5, xmm4 ; p2
+ movdqa xmm3, xmm6 ; p1
+
+ psubusb xmm4, xmm2 ; p2-=p3
+ psubusb xmm2, xmm5 ; p3-=p2
+
+ psubusb xmm3, xmm5 ; p1-=p2
+ pmaxub xmm1, xmm4 ; abs(p3 - p2)
+
+ psubusb xmm5, xmm6 ; p2-=p1
+ pmaxub xmm1, xmm2 ; abs(p3 - p2)
+
+ pmaxub xmm1, xmm5 ; abs(p2 - p1)
+ movdqa xmm2, xmm6 ; p1
+
+ pmaxub xmm1, xmm3 ; abs(p2 - p1)
+%if %1
+ movdqa xmm4, [rsi+rax] ; p0
+ movdqa xmm3, [rdi] ; q1
+%else
+ movlps xmm4, [rsi + rcx*2] ; p0
+ movhps xmm4, [rdi + rcx*2]
+ movdqa xmm3, [rsp+_q1] ; q1
+%endif
+
+ movdqa xmm5, xmm4 ; p0
+ psubusb xmm4, xmm6 ; p0-=p1
+
+ psubusb xmm6, xmm5 ; p1-=p0
+
+ por xmm6, xmm4 ; abs(p1 - p0)
+ mov rdx, arg(2) ; get blimit
+
+ movdqa [rsp+_t1], xmm6 ; save to t1
+
+ movdqa xmm4, xmm3 ; q1
+ pmaxub xmm1, xmm6
+
+ psubusb xmm3, xmm2 ; q1-=p1
+ psubusb xmm2, xmm4 ; p1-=q1
+
+ psubusb xmm1, xmm7
+ por xmm2, xmm3 ; abs(p1-q1)
+
+ movdqa xmm7, [rdx] ; blimit
+ mov rdx, arg(4) ; hev get thresh
+
+ movdqa xmm3, xmm0 ; q0
+ pand xmm2, [GLOBAL(tfe)] ; set lsb of each byte to zero
+
+ movdqa xmm6, xmm5 ; p0
+ psrlw xmm2, 1 ; abs(p1-q1)/2
+
+ psubusb xmm5, xmm3 ; p0-=q0
+ psubusb xmm3, xmm6 ; q0-=p0
+ por xmm5, xmm3 ; abs(p0 - q0)
+
+ paddusb xmm5, xmm5 ; abs(p0-q0)*2
+
+ movdqa xmm4, [rsp+_t0] ; hev get abs (q1 - q0)
+ movdqa xmm3, [rsp+_t1] ; get abs (p1 - p0)
+
+ paddusb xmm5, xmm2 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+
+ movdqa xmm2, [rdx] ; hev
+
+ psubusb xmm5, xmm7 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ psubusb xmm4, xmm2 ; hev
+
+ psubusb xmm3, xmm2 ; hev
+ por xmm1, xmm5
+
+ pxor xmm7, xmm7
+ paddb xmm4, xmm3 ; hev abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
+
+ pcmpeqb xmm4, xmm5 ; hev
+ pcmpeqb xmm3, xmm3 ; hev
+
+ pcmpeqb xmm1, xmm7 ; mask xmm1
+ pxor xmm4, xmm3 ; hev
+%endmacro
+
+%macro B_FILTER 1
+ movdqa xmm3, [GLOBAL(t80)]
+%if %1 == 0
+ movdqa xmm2, [rsp+_p1] ; p1
+ movdqa xmm7, [rsp+_q1] ; q1
+%elif %1 == 1
+ movdqa xmm2, [rsi+2*rax] ; p1
+ movdqa xmm7, [rdi] ; q1
+%elif %1 == 2
+ movdqa xmm2, [rsp+_p1] ; p1
+ movdqa xmm6, [rsp+_p0] ; p0
+ movdqa xmm0, [rsp+_q0] ; q0
+ movdqa xmm7, [rsp+_q1] ; q1
+%endif
+
+ pxor xmm2, xmm3 ; p1 offset to convert to signed values
+ pxor xmm7, xmm3 ; q1 offset to convert to signed values
+
+ psubsb xmm2, xmm7 ; p1 - q1
+ pxor xmm6, xmm3 ; offset to convert to signed values
+
+ pand xmm2, xmm4 ; high var mask (hvm)(p1 - q1)
+ pxor xmm0, xmm3 ; offset to convert to signed values
+
+ movdqa xmm3, xmm0 ; q0
+ psubsb xmm0, xmm6 ; q0 - p0
+ paddsb xmm2, xmm0 ; 1 * (q0 - p0) + hvm(p1 - q1)
+ paddsb xmm2, xmm0 ; 2 * (q0 - p0) + hvm(p1 - q1)
+ paddsb xmm2, xmm0 ; 3 * (q0 - p0) + hvm(p1 - q1)
+ pand xmm1, xmm2 ; mask filter values we don't care about
+
+ movdqa xmm2, xmm1
+ paddsb xmm1, [GLOBAL(t4)] ; 3* (q0 - p0) + hvm(p1 - q1) + 4
+ paddsb xmm2, [GLOBAL(t3)] ; 3* (q0 - p0) + hvm(p1 - q1) + 3
+
+ punpckhbw xmm5, xmm2 ; axbxcxdx
+ punpcklbw xmm2, xmm2 ; exfxgxhx
+
+ punpcklbw xmm0, xmm1 ; exfxgxhx
+ psraw xmm5, 11 ; sign extended shift right by 3
+
+ punpckhbw xmm1, xmm1 ; axbxcxdx
+ psraw xmm2, 11 ; sign extended shift right by 3
+
+ packsswb xmm2, xmm5 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3;
+ psraw xmm0, 11 ; sign extended shift right by 3
+
+ psraw xmm1, 11 ; sign extended shift right by 3
+ movdqa xmm5, xmm0 ; save results
+
+ packsswb xmm0, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3
+
+ paddsb xmm6, xmm2 ; p0+= p0 add
+
+ movdqa xmm2, [GLOBAL(ones)]
+ paddsw xmm5, xmm2
+ paddsw xmm1, xmm2
+ psraw xmm5, 1 ; partial shifted one more time for 2nd tap
+ psraw xmm1, 1 ; partial shifted one more time for 2nd tap
+ packsswb xmm5, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4
+ movdqa xmm2, [GLOBAL(t80)]
+
+%if %1 == 0
+ movdqa xmm1, [rsp+_p1] ; p1
+ lea rsi, [rsi + rcx*2]
+ lea rdi, [rdi + rcx*2]
+%elif %1 == 1
+ movdqa xmm1, [rsi+2*rax] ; p1
+%elif %1 == 2
+ movdqa xmm1, [rsp+_p1] ; p1
+%endif
+
+ pandn xmm4, xmm5 ; high edge variance additive
+ pxor xmm6, xmm2 ; unoffset
+
+ pxor xmm1, xmm2 ; reoffset
+ psubsb xmm3, xmm0 ; q0-= q0 add
+
+ paddsb xmm1, xmm4 ; p1+= p1 add
+ pxor xmm3, xmm2 ; unoffset
+
+ pxor xmm1, xmm2 ; unoffset
+ psubsb xmm7, xmm4 ; q1-= q1 add
+
+ pxor xmm7, xmm2 ; unoffset
+%if %1 == 0
+ movq [rsi], xmm6 ; p0
+ movhps [rdi], xmm6
+ movq [rsi + rax], xmm1 ; p1
+ movhps [rdi + rax], xmm1
+ movq [rsi + rcx], xmm3 ; q0
+ movhps [rdi + rcx], xmm3
+ movq [rsi + rcx*2], xmm7 ; q1
+ movhps [rdi + rcx*2], xmm7
+%elif %1 == 1
+ movdqa [rsi+rax], xmm6 ; write back
+ movdqa [rsi+2*rax], xmm1 ; write back
+ movdqa [rsi], xmm3 ; write back
+ movdqa [rdi], xmm7 ; write back
+%endif
+
+%endmacro
+
+%if ABI_IS_32BIT
+
+;void vp8_loop_filter_horizontal_edge_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+;)
+global sym(vp8_loop_filter_horizontal_edge_sse2) PRIVATE
+sym(vp8_loop_filter_horizontal_edge_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, lf_var_size
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step
+
+ mov rdx, arg(3) ;limit
+
+ lea rdi, [rsi+rax] ; rdi points to row +1 for indirect addressing
+
+ ; calculate breakout conditions and high edge variance
+ LFH_FILTER_AND_HEV_MASK 1
+ ; filter and write back the result
+ B_FILTER 1
+
+ add rsp, lf_var_size
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+%endif
+
+;void vp8_loop_filter_horizontal_edge_uv_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+; int count
+;)
+global sym(vp8_loop_filter_horizontal_edge_uv_sse2) PRIVATE
+sym(vp8_loop_filter_horizontal_edge_uv_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, lf_var_size
+
+ mov rsi, arg(0) ; u
+ mov rdi, arg(5) ; v
+ movsxd rax, dword ptr arg(1) ; src_pixel_step
+ mov rcx, rax
+ neg rax ; negate pitch to deal with above border
+
+ mov rdx, arg(3) ;limit
+
+ lea rsi, [rsi + rcx]
+ lea rdi, [rdi + rcx]
+
+ ; calculate breakout conditions and high edge variance
+ LFH_FILTER_AND_HEV_MASK 0
+ ; filter and write back the result
+ B_FILTER 0
+
+ add rsp, lf_var_size
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+%macro MB_FILTER_AND_WRITEBACK 1
+ movdqa xmm3, [GLOBAL(t80)]
+%if %1 == 0
+ movdqa xmm2, [rsp+_p1] ; p1
+ movdqa xmm7, [rsp+_q1] ; q1
+%elif %1 == 1
+ movdqa xmm2, [rsi+2*rax] ; p1
+ movdqa xmm7, [rdi] ; q1
+
+ mov rcx, rax
+ neg rcx
+%elif %1 == 2
+ movdqa xmm2, [rsp+_p1] ; p1
+ movdqa xmm6, [rsp+_p0] ; p0
+ movdqa xmm0, [rsp+_q0] ; q0
+ movdqa xmm7, [rsp+_q1] ; q1
+%endif
+
+ pxor xmm2, xmm3 ; p1 offset to convert to signed values
+ pxor xmm7, xmm3 ; q1 offset to convert to signed values
+ pxor xmm6, xmm3 ; offset to convert to signed values
+ pxor xmm0, xmm3 ; offset to convert to signed values
+
+ psubsb xmm2, xmm7 ; p1 - q1
+
+ movdqa xmm3, xmm0 ; q0
+ psubsb xmm0, xmm6 ; q0 - p0
+ paddsb xmm2, xmm0 ; 1 * (q0 - p0) + (p1 - q1)
+ paddsb xmm2, xmm0 ; 2 * (q0 - p0)
+ paddsb xmm2, xmm0 ; 3 * (q0 - p0) + (p1 - q1)
+ pand xmm1, xmm2 ; mask filter values we don't care about
+
+ movdqa xmm2, xmm1 ; vp8_filter
+
+ pand xmm2, xmm4 ; Filter2 = vp8_filter & hev
+ pxor xmm0, xmm0
+
+ pandn xmm4, xmm1 ; vp8_filter&=~hev
+ pxor xmm1, xmm1
+
+ punpcklbw xmm0, xmm4 ; Filter 2 (hi)
+ punpckhbw xmm1, xmm4 ; Filter 2 (lo)
+
+ movdqa xmm5, xmm2
+
+ movdqa xmm4, [GLOBAL(s9)]
+ paddsb xmm5, [GLOBAL(t3)] ; vp8_signed_char_clamp(Filter2 + 3)
+ paddsb xmm2, [GLOBAL(t4)] ; vp8_signed_char_clamp(Filter2 + 4)
+
+ pmulhw xmm1, xmm4 ; Filter 2 (lo) * 9
+ pmulhw xmm0, xmm4 ; Filter 2 (hi) * 9
+
+ punpckhbw xmm7, xmm5 ; axbxcxdx
+ punpcklbw xmm5, xmm5 ; exfxgxhx
+
+ psraw xmm7, 11 ; sign extended shift right by 3
+
+ psraw xmm5, 11 ; sign extended shift right by 3
+ punpckhbw xmm4, xmm2 ; axbxcxdx
+
+ punpcklbw xmm2, xmm2 ; exfxgxhx
+ psraw xmm4, 11 ; sign extended shift right by 3
+
+ packsswb xmm5, xmm7 ; Filter2 >>=3;
+ psraw xmm2, 11 ; sign extended shift right by 3
+
+ packsswb xmm2, xmm4 ; Filter1 >>=3;
+
+ paddsb xmm6, xmm5 ; ps0 =ps0 + Fitler2
+
+ psubsb xmm3, xmm2 ; qs0 =qs0 - Filter1
+ movdqa xmm7, xmm1
+
+ movdqa xmm4, [GLOBAL(s63)]
+ movdqa xmm5, xmm0
+ movdqa xmm2, xmm5
+ paddw xmm0, xmm4 ; Filter 2 (hi) * 9 + 63
+ paddw xmm1, xmm4 ; Filter 2 (lo) * 9 + 63
+ movdqa xmm4, xmm7
+
+ paddw xmm5, xmm5 ; Filter 2 (hi) * 18
+
+ paddw xmm7, xmm7 ; Filter 2 (lo) * 18
+ paddw xmm5, xmm0 ; Filter 2 (hi) * 27 + 63
+
+ paddw xmm7, xmm1 ; Filter 2 (lo) * 27 + 63
+ paddw xmm2, xmm0 ; Filter 2 (hi) * 18 + 63
+ psraw xmm0, 7 ; (Filter 2 (hi) * 9 + 63) >> 7
+
+ paddw xmm4, xmm1 ; Filter 2 (lo) * 18 + 63
+ psraw xmm1, 7 ; (Filter 2 (lo) * 9 + 63) >> 7
+ psraw xmm2, 7 ; (Filter 2 (hi) * 18 + 63) >> 7
+
+ packsswb xmm0, xmm1 ; u1 = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
+
+ psraw xmm4, 7 ; (Filter 2 (lo) * 18 + 63) >> 7
+ psraw xmm5, 7 ; (Filter 2 (hi) * 27 + 63) >> 7
+ psraw xmm7, 7 ; (Filter 2 (lo) * 27 + 63) >> 7
+
+ packsswb xmm5, xmm7 ; u3 = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
+ packsswb xmm2, xmm4 ; u2 = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
+ movdqa xmm7, [GLOBAL(t80)]
+
+%if %1 == 0
+ movdqa xmm1, [rsp+_q1] ; q1
+ movdqa xmm4, [rsp+_p1] ; p1
+ lea rsi, [rsi+rcx*2]
+ lea rdi, [rdi+rcx*2]
+
+%elif %1 == 1
+ movdqa xmm1, [rdi] ; q1
+ movdqa xmm4, [rsi+rax*2] ; p1
+%elif %1 == 2
+ movdqa xmm4, [rsp+_p1] ; p1
+ movdqa xmm1, [rsp+_q1] ; q1
+%endif
+
+ pxor xmm1, xmm7
+ pxor xmm4, xmm7
+
+ psubsb xmm3, xmm5 ; sq = vp8_signed_char_clamp(qs0 - u3)
+ paddsb xmm6, xmm5 ; sp = vp8_signed_char_clamp(ps0 - u3)
+ psubsb xmm1, xmm2 ; sq = vp8_signed_char_clamp(qs1 - u2)
+ paddsb xmm4, xmm2 ; sp = vp8_signed_char_clamp(ps1 - u2)
+
+%if %1 == 1
+ movdqa xmm2, [rdi+rax*4] ; p2
+ movdqa xmm5, [rdi+rcx] ; q2
+%else
+ movdqa xmm2, [rsp+_p2] ; p2
+ movdqa xmm5, [rsp+_q2] ; q2
+%endif
+
+ pxor xmm1, xmm7 ; *oq1 = sq^0x80;
+ pxor xmm4, xmm7 ; *op1 = sp^0x80;
+ pxor xmm2, xmm7
+ pxor xmm5, xmm7
+ paddsb xmm2, xmm0 ; sp = vp8_signed_char_clamp(ps2 - u)
+ psubsb xmm5, xmm0 ; sq = vp8_signed_char_clamp(qs2 - u)
+ pxor xmm2, xmm7 ; *op2 = sp^0x80;
+ pxor xmm5, xmm7 ; *oq2 = sq^0x80;
+ pxor xmm3, xmm7 ; *oq0 = sq^0x80
+ pxor xmm6, xmm7 ; *oq0 = sp^0x80
+%if %1 == 0
+ movq [rsi], xmm6 ; p0
+ movhps [rdi], xmm6
+ movq [rsi + rcx], xmm3 ; q0
+ movhps [rdi + rcx], xmm3
+ lea rdx, [rcx + rcx*2]
+ movq [rsi+rcx*2], xmm1 ; q1
+ movhps [rdi+rcx*2], xmm1
+
+ movq [rsi + rax], xmm4 ; p1
+ movhps [rdi + rax], xmm4
+
+ movq [rsi+rax*2], xmm2 ; p2
+ movhps [rdi+rax*2], xmm2
+
+ movq [rsi+rdx], xmm5 ; q2
+ movhps [rdi+rdx], xmm5
+%elif %1 == 1
+ movdqa [rdi+rcx], xmm5 ; q2
+ movdqa [rdi], xmm1 ; q1
+ movdqa [rsi], xmm3 ; q0
+ movdqa [rsi+rax ], xmm6 ; p0
+ movdqa [rsi+rax*2], xmm4 ; p1
+ movdqa [rdi+rax*4], xmm2 ; p2
+%elif %1 == 2
+ movdqa [rsp+_p1], xmm4 ; p1
+ movdqa [rsp+_p0], xmm6 ; p0
+ movdqa [rsp+_q0], xmm3 ; q0
+ movdqa [rsp+_q1], xmm1 ; q1
+%endif
+
+%endmacro
+
+
+;void vp8_mbloop_filter_horizontal_edge_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+;)
+global sym(vp8_mbloop_filter_horizontal_edge_sse2) PRIVATE
+sym(vp8_mbloop_filter_horizontal_edge_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, lf_var_size
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step
+ mov rdx, arg(3) ;limit
+
+ lea rdi, [rsi+rax] ; rdi points to row +1 for indirect addressing
+
+ ; calculate breakout conditions and high edge variance
+ LFH_FILTER_AND_HEV_MASK 1
+ ; filter and write back the results
+ MB_FILTER_AND_WRITEBACK 1
+
+ add rsp, lf_var_size
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_mbloop_filter_horizontal_edge_uv_sse2
+;(
+; unsigned char *u,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+; unsigned char *v
+;)
+global sym(vp8_mbloop_filter_horizontal_edge_uv_sse2) PRIVATE
+sym(vp8_mbloop_filter_horizontal_edge_uv_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, lf_var_size
+
+ mov rsi, arg(0) ; u
+ mov rdi, arg(5) ; v
+ movsxd rax, dword ptr arg(1) ; src_pixel_step
+ mov rcx, rax
+ neg rax ; negate pitch to deal with above border
+ mov rdx, arg(3) ;limit
+
+ lea rsi, [rsi + rcx]
+ lea rdi, [rdi + rcx]
+
+ ; calculate breakout conditions and high edge variance
+ LFH_FILTER_AND_HEV_MASK 0
+ ; filter and write back the results
+ MB_FILTER_AND_WRITEBACK 0
+
+ add rsp, lf_var_size
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+%macro TRANSPOSE_16X8 2
+ movq xmm4, [rsi] ; xx xx xx xx xx xx xx xx 07 06 05 04 03 02 01 00
+ movq xmm1, [rdi] ; xx xx xx xx xx xx xx xx 17 16 15 14 13 12 11 10
+ movq xmm0, [rsi+2*rax] ; xx xx xx xx xx xx xx xx 27 26 25 24 23 22 21 20
+ movq xmm7, [rdi+2*rax] ; xx xx xx xx xx xx xx xx 37 36 35 34 33 32 31 30
+ movq xmm5, [rsi+4*rax] ; xx xx xx xx xx xx xx xx 47 46 45 44 43 42 41 40
+ movq xmm2, [rdi+4*rax] ; xx xx xx xx xx xx xx xx 57 56 55 54 53 52 51 50
+
+ punpcklbw xmm4, xmm1 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00
+
+ movq xmm1, [rdi+2*rcx] ; xx xx xx xx xx xx xx xx 77 76 75 74 73 72 71 70
+
+ movdqa xmm3, xmm4 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00
+ punpcklbw xmm0, xmm7 ; 37 27 36 36 35 25 34 24 33 23 32 22 31 21 30 20
+
+ movq xmm7, [rsi+2*rcx] ; xx xx xx xx xx xx xx xx 67 66 65 64 63 62 61 60
+
+ punpcklbw xmm5, xmm2 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40
+%if %1
+ lea rsi, [rsi+rax*8]
+ lea rdi, [rdi+rax*8]
+%else
+ mov rsi, arg(5) ; v_ptr
+%endif
+
+ movdqa xmm6, xmm5 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40
+ punpcklbw xmm7, xmm1 ; 77 67 76 66 75 65 74 64 73 63 72 62 71 61 70 60
+ punpcklwd xmm5, xmm7 ; 73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40
+ punpckhwd xmm6, xmm7 ; 77 67 57 47 76 66 56 46 75 65 55 45 74 64 54 44
+ punpcklwd xmm3, xmm0 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00
+
+%if %1 == 0
+ lea rdi, [rsi + rax - 4] ; rdi points to row +1 for indirect addressing
+ lea rsi, [rsi - 4]
+%endif
+
+ movdqa xmm2, xmm3 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00
+ punpckhwd xmm4, xmm0 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04
+
+ movdqa xmm7, xmm4 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04
+ punpckhdq xmm3, xmm5 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
+
+ punpckhdq xmm7, xmm6 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06
+
+ punpckldq xmm4, xmm6 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04
+
+ punpckldq xmm2, xmm5 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
+
+ movdqa [rsp+_t0], xmm2 ; save to free XMM2
+
+ movq xmm2, [rsi] ; xx xx xx xx xx xx xx xx 87 86 85 84 83 82 81 80
+ movq xmm6, [rdi] ; xx xx xx xx xx xx xx xx 97 96 95 94 93 92 91 90
+ movq xmm0, [rsi+2*rax] ; xx xx xx xx xx xx xx xx a7 a6 a5 a4 a3 a2 a1 a0
+ movq xmm5, [rdi+2*rax] ; xx xx xx xx xx xx xx xx b7 b6 b5 b4 b3 b2 b1 b0
+ movq xmm1, [rsi+4*rax] ; xx xx xx xx xx xx xx xx c7 c6 c5 c4 c3 c2 c1 c0
+
+ punpcklbw xmm2, xmm6 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80
+
+ movq xmm6, [rdi+4*rax] ; xx xx xx xx xx xx xx xx d7 d6 d5 d4 d3 d2 d1 d0
+
+ punpcklbw xmm0, xmm5 ; b7 a7 b6 a6 b5 a5 b4 a4 b3 a3 b2 a2 b1 a1 b0 a0
+
+ movq xmm5, [rsi+2*rcx] ; xx xx xx xx xx xx xx xx e7 e6 e5 e4 e3 e2 e1 e0
+
+ punpcklbw xmm1, xmm6 ; d7 c7 d6 c6 d5 c5 d4 c4 d3 c3 d2 c2 d1 e1 d0 c0
+
+ movq xmm6, [rdi+2*rcx] ; xx xx xx xx xx xx xx xx f7 f6 f5 f4 f3 f2 f1 f0
+
+ punpcklbw xmm5, xmm6 ; f7 e7 f6 e6 f5 e5 f4 e4 f3 e3 f2 e2 f1 e1 f0 e0
+
+ movdqa xmm6, xmm1 ;
+ punpckhwd xmm6, xmm5 ; f7 e7 d7 c7 f6 e6 d6 c6 f5 e5 d5 c5 f4 e4 d4 c4
+
+ punpcklwd xmm1, xmm5 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0
+ movdqa xmm5, xmm2 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80
+
+ punpcklwd xmm5, xmm0 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80
+
+ punpckhwd xmm2, xmm0 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84
+
+ movdqa xmm0, xmm5
+ punpckldq xmm0, xmm1 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80
+
+ punpckhdq xmm5, xmm1 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82
+ movdqa xmm1, xmm2 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84
+
+ punpckldq xmm1, xmm6 ; f5 e5 d5 c5 b5 a5 95 85 f4 e4 d4 c4 b4 a4 94 84
+
+ punpckhdq xmm2, xmm6 ; f7 e7 d7 c7 b7 a7 97 87 f6 e6 d6 c6 b6 a6 96 86
+ movdqa xmm6, xmm7 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06
+
+ punpcklqdq xmm6, xmm2 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06
+
+ punpckhqdq xmm7, xmm2 ; f7 e7 d7 c7 b7 a7 97 87 77 67 57 47 37 27 17 07
+
+%if %2 == 0
+ movdqa [rsp+_q3], xmm7 ; save 7
+ movdqa [rsp+_q2], xmm6 ; save 6
+%endif
+ movdqa xmm2, xmm3 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
+ punpckhqdq xmm3, xmm5 ; f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
+ punpcklqdq xmm2, xmm5 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
+ movdqa [rsp+_p1], xmm2 ; save 2
+
+ movdqa xmm5, xmm4 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04
+ punpcklqdq xmm4, xmm1 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
+ movdqa [rsp+_p0], xmm3 ; save 3
+
+ punpckhqdq xmm5, xmm1 ; f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05
+
+ movdqa [rsp+_q0], xmm4 ; save 4
+ movdqa [rsp+_q1], xmm5 ; save 5
+ movdqa xmm1, [rsp+_t0]
+
+ movdqa xmm2, xmm1 ;
+ punpckhqdq xmm1, xmm0 ; f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01
+ punpcklqdq xmm2, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
+
+%if %2 == 0
+ movdqa [rsp+_p2], xmm1
+ movdqa [rsp+_p3], xmm2
+%endif
+
+%endmacro
+
+%macro LFV_FILTER_MASK_HEV_MASK 0
+ movdqa xmm0, xmm6 ; q2
+ psubusb xmm0, xmm7 ; q2-q3
+
+ psubusb xmm7, xmm6 ; q3-q2
+ movdqa xmm4, xmm5 ; q1
+
+ por xmm7, xmm0 ; abs (q3-q2)
+ psubusb xmm4, xmm6 ; q1-q2
+
+ movdqa xmm0, xmm1
+ psubusb xmm6, xmm5 ; q2-q1
+
+ por xmm6, xmm4 ; abs (q2-q1)
+ psubusb xmm0, xmm2 ; p2 - p3;
+
+ psubusb xmm2, xmm1 ; p3 - p2;
+ por xmm0, xmm2 ; abs(p2-p3)
+
+ movdqa xmm5, [rsp+_p1] ; p1
+ pmaxub xmm0, xmm7
+
+ movdqa xmm2, xmm5 ; p1
+ psubusb xmm5, xmm1 ; p1-p2
+ psubusb xmm1, xmm2 ; p2-p1
+
+ movdqa xmm7, xmm3 ; p0
+ psubusb xmm7, xmm2 ; p0-p1
+
+ por xmm1, xmm5 ; abs(p2-p1)
+ pmaxub xmm0, xmm6
+
+ pmaxub xmm0, xmm1
+ movdqa xmm1, xmm2 ; p1
+
+ psubusb xmm2, xmm3 ; p1-p0
+
+ por xmm2, xmm7 ; abs(p1-p0)
+
+ pmaxub xmm0, xmm2
+
+ movdqa xmm5, [rsp+_q0] ; q0
+ movdqa xmm7, [rsp+_q1] ; q1
+
+ mov rdx, arg(3) ; limit
+
+ movdqa xmm6, xmm5 ; q0
+ movdqa xmm4, xmm7 ; q1
+
+ psubusb xmm5, xmm7 ; q0-q1
+ psubusb xmm7, xmm6 ; q1-q0
+
+ por xmm7, xmm5 ; abs(q1-q0)
+
+ pmaxub xmm0, xmm7
+
+ psubusb xmm0, [rdx] ; limit
+
+ mov rdx, arg(2) ; blimit
+ movdqa xmm5, xmm4 ; q1
+
+ psubusb xmm5, xmm1 ; q1-=p1
+ psubusb xmm1, xmm4 ; p1-=q1
+
+ por xmm5, xmm1 ; abs(p1-q1)
+ movdqa xmm1, xmm3 ; p0
+
+ pand xmm5, [GLOBAL(tfe)] ; set lsb of each byte to zero
+ psubusb xmm1, xmm6 ; p0-q0
+
+ movdqa xmm4, [rdx] ; blimit
+ mov rdx, arg(4) ; get thresh
+
+ psrlw xmm5, 1 ; abs(p1-q1)/2
+ psubusb xmm6, xmm3 ; q0-p0
+
+ por xmm1, xmm6 ; abs(q0-p0)
+ paddusb xmm1, xmm1 ; abs(q0-p0)*2
+ movdqa xmm3, [rdx]
+
+ paddusb xmm1, xmm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+ psubusb xmm2, xmm3 ; abs(q1 - q0) > thresh
+
+ psubusb xmm7, xmm3 ; abs(p1 - p0)> thresh
+
+ psubusb xmm1, xmm4 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ por xmm2, xmm7 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
+
+ por xmm1, xmm0 ; mask
+ pcmpeqb xmm2, xmm0
+
+ pxor xmm0, xmm0
+ pcmpeqb xmm4, xmm4
+
+ pcmpeqb xmm1, xmm0
+ pxor xmm4, xmm2
+%endmacro
+
+%macro BV_TRANSPOSE 0
+ ; xmm1 = f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
+ ; xmm6 = f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
+ ; xmm3 = f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
+ ; xmm7 = f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05
+ movdqa xmm2, xmm1 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
+ punpcklbw xmm2, xmm6 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
+
+ movdqa xmm4, xmm3 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
+ punpckhbw xmm1, xmm6 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
+
+ punpcklbw xmm4, xmm7 ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04
+
+ punpckhbw xmm3, xmm7 ; f5 f4 e5 e4 d5 d4 c5 c4 b5 b4 a5 a4 95 94 85 84
+
+ movdqa xmm6, xmm2 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
+ punpcklwd xmm2, xmm4 ; 35 34 33 32 25 24 23 22 15 14 13 12 05 04 03 02
+
+ punpckhwd xmm6, xmm4 ; 75 74 73 72 65 64 63 62 55 54 53 52 45 44 43 42
+ movdqa xmm5, xmm1 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
+
+ punpcklwd xmm1, xmm3 ; b5 b4 b3 b2 a5 a4 a3 a2 95 94 93 92 85 84 83 82
+
+ punpckhwd xmm5, xmm3 ; f5 f4 f3 f2 e5 e4 e3 e2 d5 d4 d3 d2 c5 c4 c3 c2
+ ; xmm2 = 35 34 33 32 25 24 23 22 15 14 13 12 05 04 03 02
+ ; xmm6 = 75 74 73 72 65 64 63 62 55 54 53 52 45 44 43 42
+ ; xmm1 = b5 b4 b3 b2 a5 a4 a3 a2 95 94 93 92 85 84 83 82
+ ; xmm5 = f5 f4 f3 f2 e5 e4 e3 e2 d5 d4 d3 d2 c5 c4 c3 c2
+%endmacro
+
+%macro BV_WRITEBACK 2
+ movd [rsi+2], %1
+ movd [rsi+4*rax+2], %2
+ psrldq %1, 4
+ psrldq %2, 4
+ movd [rdi+2], %1
+ movd [rdi+4*rax+2], %2
+ psrldq %1, 4
+ psrldq %2, 4
+ movd [rsi+2*rax+2], %1
+ movd [rsi+2*rcx+2], %2
+ psrldq %1, 4
+ psrldq %2, 4
+ movd [rdi+2*rax+2], %1
+ movd [rdi+2*rcx+2], %2
+%endmacro
+
+%if ABI_IS_32BIT
+
+;void vp8_loop_filter_vertical_edge_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+;)
+global sym(vp8_loop_filter_vertical_edge_sse2) PRIVATE
+sym(vp8_loop_filter_vertical_edge_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, lf_var_size
+
+ mov rsi, arg(0) ; src_ptr
+ movsxd rax, dword ptr arg(1) ; src_pixel_step
+
+ lea rsi, [rsi - 4]
+ lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
+ lea rcx, [rax*2+rax]
+
+ ;transpose 16x8 to 8x16, and store the 8-line result on stack.
+ TRANSPOSE_16X8 1, 1
+
+ ; calculate filter mask and high edge variance
+ LFV_FILTER_MASK_HEV_MASK
+
+ ; start work on filters
+ B_FILTER 2
+
+ ; tranpose and write back - only work on q1, q0, p0, p1
+ BV_TRANSPOSE
+ ; store 16-line result
+
+ lea rdx, [rax]
+ neg rdx
+
+ BV_WRITEBACK xmm1, xmm5
+
+ lea rsi, [rsi+rdx*8]
+ lea rdi, [rdi+rdx*8]
+ BV_WRITEBACK xmm2, xmm6
+
+ add rsp, lf_var_size
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+%endif
+
+;void vp8_loop_filter_vertical_edge_uv_sse2
+;(
+; unsigned char *u,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+; unsigned char *v
+;)
+global sym(vp8_loop_filter_vertical_edge_uv_sse2) PRIVATE
+sym(vp8_loop_filter_vertical_edge_uv_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, lf_var_size
+
+ mov rsi, arg(0) ; u_ptr
+ movsxd rax, dword ptr arg(1) ; src_pixel_step
+
+ lea rsi, [rsi - 4]
+ lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
+ lea rcx, [rax+2*rax]
+
+ ;transpose 16x8 to 8x16, and store the 8-line result on stack.
+ TRANSPOSE_16X8 0, 1
+
+ ; calculate filter mask and high edge variance
+ LFV_FILTER_MASK_HEV_MASK
+
+ ; start work on filters
+ B_FILTER 2
+
+ ; tranpose and write back - only work on q1, q0, p0, p1
+ BV_TRANSPOSE
+
+ lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
+
+ ; store 16-line result
+ BV_WRITEBACK xmm1, xmm5
+
+ mov rsi, arg(0) ; u_ptr
+ lea rsi, [rsi - 4]
+ lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
+ BV_WRITEBACK xmm2, xmm6
+
+ add rsp, lf_var_size
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+%macro MBV_TRANSPOSE 0
+ movdqa xmm0, [rsp+_p3] ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
+ movdqa xmm1, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
+
+ punpcklbw xmm0, xmm2 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00
+ punpckhbw xmm1, xmm2 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80
+
+ movdqa xmm7, [rsp+_p1] ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
+ movdqa xmm6, xmm7 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
+
+ punpcklbw xmm7, [rsp+_p0] ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
+ punpckhbw xmm6, [rsp+_p0] ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
+
+ movdqa xmm3, xmm0 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00
+ punpcklwd xmm0, xmm7 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00
+
+ punpckhwd xmm3, xmm7 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40
+ movdqa xmm4, xmm1 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80
+
+ punpcklwd xmm1, xmm6 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80
+ punpckhwd xmm4, xmm6 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0
+
+ movdqa xmm7, [rsp+_q0] ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
+ punpcklbw xmm7, [rsp+_q1] ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04
+
+ movdqa xmm6, xmm5 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06
+ punpcklbw xmm6, [rsp+_q3] ; 77 76 67 66 57 56 47 46 37 36 27 26 17 16 07 06
+
+ movdqa xmm2, xmm7 ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04
+ punpcklwd xmm7, xmm6 ; 37 36 35 34 27 26 25 24 17 16 15 14 07 06 05 04
+
+ punpckhwd xmm2, xmm6 ; 77 76 75 74 67 66 65 64 57 56 55 54 47 46 45 44
+ movdqa xmm6, xmm0 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00
+
+ punpckldq xmm0, xmm7 ; 17 16 15 14 13 12 11 10 07 06 05 04 03 02 01 00
+ punpckhdq xmm6, xmm7 ; 37 36 35 34 33 32 31 30 27 26 25 24 23 22 21 20
+%endmacro
+
+%macro MBV_WRITEBACK_1 0
+ movq [rsi], xmm0
+ movhps [rdi], xmm0
+
+ movq [rsi+2*rax], xmm6
+ movhps [rdi+2*rax], xmm6
+
+ movdqa xmm0, xmm3 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40
+ punpckldq xmm0, xmm2 ; 57 56 55 54 53 52 51 50 47 46 45 44 43 42 41 40
+ punpckhdq xmm3, xmm2 ; 77 76 75 74 73 72 71 70 67 66 65 64 63 62 61 60
+
+ movq [rsi+4*rax], xmm0
+ movhps [rdi+4*rax], xmm0
+
+ movq [rsi+2*rcx], xmm3
+ movhps [rdi+2*rcx], xmm3
+
+ movdqa xmm7, [rsp+_q0] ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
+ punpckhbw xmm7, [rsp+_q1] ; f5 f4 e5 e4 d5 d4 c5 c4 b5 b4 a5 a4 95 94 85 84
+ punpckhbw xmm5, [rsp+_q3] ; f7 f6 e7 e6 d7 d6 c7 c6 b7 b6 a7 a6 97 96 87 86
+
+ movdqa xmm0, xmm7
+ punpcklwd xmm0, xmm5 ; b7 b6 b4 b4 a7 a6 a5 a4 97 96 95 94 87 86 85 84
+ punpckhwd xmm7, xmm5 ; f7 f6 f5 f4 e7 e6 e5 e4 d7 d6 d5 d4 c7 c6 c5 c4
+
+ movdqa xmm5, xmm1 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80
+ punpckldq xmm1, xmm0 ; 97 96 95 94 93 92 91 90 87 86 85 83 84 82 81 80
+ punpckhdq xmm5, xmm0 ; b7 b6 b5 b4 b3 b2 b1 b0 a7 a6 a5 a4 a3 a2 a1 a0
+%endmacro
+
+%macro MBV_WRITEBACK_2 0
+ movq [rsi], xmm1
+ movhps [rdi], xmm1
+
+ movq [rsi+2*rax], xmm5
+ movhps [rdi+2*rax], xmm5
+
+ movdqa xmm1, xmm4 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0
+ punpckldq xmm1, xmm7 ; d7 d6 d5 d4 d3 d2 d1 d0 c7 c6 c5 c4 c3 c2 c1 c0
+ punpckhdq xmm4, xmm7 ; f7 f6 f4 f4 f3 f2 f1 f0 e7 e6 e5 e4 e3 e2 e1 e0
+
+ movq [rsi+4*rax], xmm1
+ movhps [rdi+4*rax], xmm1
+
+ movq [rsi+2*rcx], xmm4
+ movhps [rdi+2*rcx], xmm4
+%endmacro
+
+
+;void vp8_mbloop_filter_vertical_edge_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+;)
+global sym(vp8_mbloop_filter_vertical_edge_sse2) PRIVATE
+sym(vp8_mbloop_filter_vertical_edge_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, lf_var_size
+
+ mov rsi, arg(0) ; src_ptr
+ movsxd rax, dword ptr arg(1) ; src_pixel_step
+
+ lea rsi, [rsi - 4]
+ lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
+ lea rcx, [rax*2+rax]
+
+ ; Transpose
+ TRANSPOSE_16X8 1, 0
+
+ ; calculate filter mask and high edge variance
+ LFV_FILTER_MASK_HEV_MASK
+
+ neg rax
+ ; start work on filters
+ MB_FILTER_AND_WRITEBACK 2
+
+ lea rsi, [rsi+rax*8]
+ lea rdi, [rdi+rax*8]
+
+ ; transpose and write back
+ MBV_TRANSPOSE
+
+ neg rax
+
+ MBV_WRITEBACK_1
+
+
+ lea rsi, [rsi+rax*8]
+ lea rdi, [rdi+rax*8]
+ MBV_WRITEBACK_2
+
+ add rsp, lf_var_size
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_mbloop_filter_vertical_edge_uv_sse2
+;(
+; unsigned char *u,
+; int src_pixel_step,
+; const char *blimit,
+; const char *limit,
+; const char *thresh,
+; unsigned char *v
+;)
+global sym(vp8_mbloop_filter_vertical_edge_uv_sse2) PRIVATE
+sym(vp8_mbloop_filter_vertical_edge_uv_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, lf_var_size
+
+ mov rsi, arg(0) ; u_ptr
+ movsxd rax, dword ptr arg(1) ; src_pixel_step
+
+ lea rsi, [rsi - 4]
+ lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
+ lea rcx, [rax+2*rax]
+
+ ; Transpose
+ TRANSPOSE_16X8 0, 0
+
+ ; calculate filter mask and high edge variance
+ LFV_FILTER_MASK_HEV_MASK
+
+ ; start work on filters
+ MB_FILTER_AND_WRITEBACK 2
+
+ ; transpose and write back
+ MBV_TRANSPOSE
+
+ mov rsi, arg(0) ;u_ptr
+ lea rsi, [rsi - 4]
+ lea rdi, [rsi + rax]
+ MBV_WRITEBACK_1
+ mov rsi, arg(5) ;v_ptr
+ lea rsi, [rsi - 4]
+ lea rdi, [rsi + rax]
+ MBV_WRITEBACK_2
+
+ add rsp, lf_var_size
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_loop_filter_simple_horizontal_edge_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+;)
+global sym(vp8_loop_filter_simple_horizontal_edge_sse2) PRIVATE
+sym(vp8_loop_filter_simple_horizontal_edge_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 3
+ SAVE_XMM 7
+ GET_GOT rbx
+ ; end prolog
+
+ mov rcx, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
+ movdqa xmm6, [GLOBAL(tfe)]
+ lea rdx, [rcx + rax]
+ neg rax
+
+ ; calculate mask
+ movdqa xmm0, [rdx] ; q1
+ mov rdx, arg(2) ;blimit
+ movdqa xmm1, [rcx+2*rax] ; p1
+
+ movdqa xmm2, xmm1
+ movdqa xmm3, xmm0
+
+ psubusb xmm0, xmm1 ; q1-=p1
+ psubusb xmm1, xmm3 ; p1-=q1
+ por xmm1, xmm0 ; abs(p1-q1)
+ pand xmm1, xmm6 ; set lsb of each byte to zero
+ psrlw xmm1, 1 ; abs(p1-q1)/2
+
+ movdqa xmm7, XMMWORD PTR [rdx]
+
+ movdqa xmm5, [rcx+rax] ; p0
+ movdqa xmm4, [rcx] ; q0
+ movdqa xmm0, xmm4 ; q0
+ movdqa xmm6, xmm5 ; p0
+ psubusb xmm5, xmm4 ; p0-=q0
+ psubusb xmm4, xmm6 ; q0-=p0
+ por xmm5, xmm4 ; abs(p0 - q0)
+
+ movdqa xmm4, [GLOBAL(t80)]
+
+ paddusb xmm5, xmm5 ; abs(p0-q0)*2
+ paddusb xmm5, xmm1 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+ psubusb xmm5, xmm7 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ pxor xmm7, xmm7
+ pcmpeqb xmm5, xmm7
+
+
+ ; start work on filters
+ pxor xmm2, xmm4 ; p1 offset to convert to signed values
+ pxor xmm3, xmm4 ; q1 offset to convert to signed values
+ psubsb xmm2, xmm3 ; p1 - q1
+
+ pxor xmm6, xmm4 ; offset to convert to signed values
+ pxor xmm0, xmm4 ; offset to convert to signed values
+ movdqa xmm3, xmm0 ; q0
+ psubsb xmm0, xmm6 ; q0 - p0
+ paddsb xmm2, xmm0 ; p1 - q1 + 1 * (q0 - p0)
+ paddsb xmm2, xmm0 ; p1 - q1 + 2 * (q0 - p0)
+ paddsb xmm2, xmm0 ; p1 - q1 + 3 * (q0 - p0)
+ pand xmm5, xmm2 ; mask filter values we don't care about
+
+ movdqa xmm0, xmm5
+ paddsb xmm5, [GLOBAL(t3)] ; 3* (q0 - p0) + (p1 - q1) + 4
+ paddsb xmm0, [GLOBAL(t4)] ; +3 instead of +4
+
+ movdqa xmm1, [GLOBAL(te0)]
+ movdqa xmm2, [GLOBAL(t1f)]
+
+; pxor xmm7, xmm7
+ pcmpgtb xmm7, xmm0 ;save sign
+ pand xmm7, xmm1 ;preserve the upper 3 bits
+ psrlw xmm0, 3
+ pand xmm0, xmm2 ;clear out upper 3 bits
+ por xmm0, xmm7 ;add sign
+ psubsb xmm3, xmm0 ; q0-= q0sz add
+
+ pxor xmm7, xmm7
+ pcmpgtb xmm7, xmm5 ;save sign
+ pand xmm7, xmm1 ;preserve the upper 3 bits
+ psrlw xmm5, 3
+ pand xmm5, xmm2 ;clear out upper 3 bits
+ por xmm5, xmm7 ;add sign
+ paddsb xmm6, xmm5 ; p0+= p0 add
+
+ pxor xmm3, xmm4 ; unoffset
+ movdqa [rcx], xmm3 ; write back
+
+ pxor xmm6, xmm4 ; unoffset
+ movdqa [rcx+rax], xmm6 ; write back
+
+ ; begin epilog
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_loop_filter_simple_vertical_edge_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixel_step,
+; const char *blimit,
+;)
+global sym(vp8_loop_filter_simple_vertical_edge_sse2) PRIVATE
+sym(vp8_loop_filter_simple_vertical_edge_sse2):
+ push rbp ; save old base pointer value.
+ mov rbp, rsp ; set new base pointer value.
+ SHADOW_ARGS_TO_STACK 3
+ SAVE_XMM 7
+ GET_GOT rbx ; save callee-saved reg
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 32 ; reserve 32 bytes
+ %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16];
+ %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16];
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
+
+ lea rsi, [rsi - 2 ]
+ lea rdi, [rsi + rax]
+ lea rdx, [rsi + rax*4]
+ lea rcx, [rdx + rax]
+
+ movd xmm0, [rsi] ; (high 96 bits unused) 03 02 01 00
+ movd xmm1, [rdx] ; (high 96 bits unused) 43 42 41 40
+ movd xmm2, [rdi] ; 13 12 11 10
+ movd xmm3, [rcx] ; 53 52 51 50
+ punpckldq xmm0, xmm1 ; (high 64 bits unused) 43 42 41 40 03 02 01 00
+ punpckldq xmm2, xmm3 ; 53 52 51 50 13 12 11 10
+
+ movd xmm4, [rsi + rax*2] ; 23 22 21 20
+ movd xmm5, [rdx + rax*2] ; 63 62 61 60
+ movd xmm6, [rdi + rax*2] ; 33 32 31 30
+ movd xmm7, [rcx + rax*2] ; 73 72 71 70
+ punpckldq xmm4, xmm5 ; 63 62 61 60 23 22 21 20
+ punpckldq xmm6, xmm7 ; 73 72 71 70 33 32 31 30
+
+ punpcklbw xmm0, xmm2 ; 53 43 52 42 51 41 50 40 13 03 12 02 11 01 10 00
+ punpcklbw xmm4, xmm6 ; 73 63 72 62 71 61 70 60 33 23 32 22 31 21 30 20
+
+ movdqa xmm1, xmm0
+ punpcklwd xmm0, xmm4 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00
+ punpckhwd xmm1, xmm4 ; 73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40
+
+ movdqa xmm2, xmm0
+ punpckldq xmm0, xmm1 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
+ punpckhdq xmm2, xmm1 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
+
+ lea rsi, [rsi + rax*8]
+ lea rdi, [rsi + rax]
+ lea rdx, [rsi + rax*4]
+ lea rcx, [rdx + rax]
+
+ movd xmm4, [rsi] ; 83 82 81 80
+ movd xmm1, [rdx] ; c3 c2 c1 c0
+ movd xmm6, [rdi] ; 93 92 91 90
+ movd xmm3, [rcx] ; d3 d2 d1 d0
+ punpckldq xmm4, xmm1 ; c3 c2 c1 c0 83 82 81 80
+ punpckldq xmm6, xmm3 ; d3 d2 d1 d0 93 92 91 90
+
+ movd xmm1, [rsi + rax*2] ; a3 a2 a1 a0
+ movd xmm5, [rdx + rax*2] ; e3 e2 e1 e0
+ movd xmm3, [rdi + rax*2] ; b3 b2 b1 b0
+ movd xmm7, [rcx + rax*2] ; f3 f2 f1 f0
+ punpckldq xmm1, xmm5 ; e3 e2 e1 e0 a3 a2 a1 a0
+ punpckldq xmm3, xmm7 ; f3 f2 f1 f0 b3 b2 b1 b0
+
+ punpcklbw xmm4, xmm6 ; d3 c3 d2 c2 d1 c1 d0 c0 93 83 92 82 91 81 90 80
+ punpcklbw xmm1, xmm3 ; f3 e3 f2 e2 f1 e1 f0 e0 b3 a3 b2 a2 b1 a1 b0 a0
+
+ movdqa xmm7, xmm4
+ punpcklwd xmm4, xmm1 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80
+ punpckhwd xmm7, xmm1 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0
+
+ movdqa xmm6, xmm4
+ punpckldq xmm4, xmm7 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80
+ punpckhdq xmm6, xmm7 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82
+
+ movdqa xmm1, xmm0
+ movdqa xmm3, xmm2
+
+ punpcklqdq xmm0, xmm4 ; p1 f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
+ punpckhqdq xmm1, xmm4 ; p0 f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01
+ punpcklqdq xmm2, xmm6 ; q0 f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
+ punpckhqdq xmm3, xmm6 ; q1 f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
+
+ mov rdx, arg(2) ;blimit
+
+ ; calculate mask
+ movdqa xmm6, xmm0 ; p1
+ movdqa xmm7, xmm3 ; q1
+ psubusb xmm7, xmm0 ; q1-=p1
+ psubusb xmm6, xmm3 ; p1-=q1
+ por xmm6, xmm7 ; abs(p1-q1)
+ pand xmm6, [GLOBAL(tfe)] ; set lsb of each byte to zero
+ psrlw xmm6, 1 ; abs(p1-q1)/2
+
+ movdqa xmm7, [rdx]
+
+ movdqa xmm5, xmm1 ; p0
+ movdqa xmm4, xmm2 ; q0
+ psubusb xmm5, xmm2 ; p0-=q0
+ psubusb xmm4, xmm1 ; q0-=p0
+ por xmm5, xmm4 ; abs(p0 - q0)
+ paddusb xmm5, xmm5 ; abs(p0-q0)*2
+ paddusb xmm5, xmm6 ; abs (p0 - q0) *2 + abs(p1-q1)/2
+
+ movdqa xmm4, [GLOBAL(t80)]
+
+ psubusb xmm5, xmm7 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > blimit
+ pxor xmm7, xmm7
+ pcmpeqb xmm5, xmm7 ; mm5 = mask
+
+ ; start work on filters
+ movdqa t0, xmm0
+ movdqa t1, xmm3
+
+ pxor xmm0, xmm4 ; p1 offset to convert to signed values
+ pxor xmm3, xmm4 ; q1 offset to convert to signed values
+ psubsb xmm0, xmm3 ; p1 - q1
+
+ pxor xmm1, xmm4 ; offset to convert to signed values
+ pxor xmm2, xmm4 ; offset to convert to signed values
+
+ movdqa xmm3, xmm2 ; offseted ; q0
+ psubsb xmm2, xmm1 ; q0 - p0
+ paddsb xmm0, xmm2 ; p1 - q1 + 1 * (q0 - p0)
+ paddsb xmm0, xmm2 ; p1 - q1 + 2 * (q0 - p0)
+ paddsb xmm0, xmm2 ; p1 - q1 + 3 * (q0 - p0)
+ pand xmm5, xmm0 ; mask filter values we don't care about
+
+ movdqa xmm0, xmm5
+ paddsb xmm5, [GLOBAL(t3)] ; 3* (q0 - p0) + (p1 - q1) + 4
+ paddsb xmm0, [GLOBAL(t4)] ; +3 instead of +4
+
+ movdqa xmm6, [GLOBAL(te0)]
+ movdqa xmm2, [GLOBAL(t1f)]
+
+; pxor xmm7, xmm7
+ pcmpgtb xmm7, xmm0 ;save sign
+ pand xmm7, xmm6 ;preserve the upper 3 bits
+ psrlw xmm0, 3
+ pand xmm0, xmm2 ;clear out upper 3 bits
+ por xmm0, xmm7 ;add sign
+ psubsb xmm3, xmm0 ; q0-= q0sz add
+
+ pxor xmm7, xmm7
+ pcmpgtb xmm7, xmm5 ;save sign
+ pand xmm7, xmm6 ;preserve the upper 3 bits
+ psrlw xmm5, 3
+ pand xmm5, xmm2 ;clear out upper 3 bits
+ por xmm5, xmm7 ;add sign
+ paddsb xmm1, xmm5 ; p0+= p0 add
+
+ pxor xmm3, xmm4 ; unoffset q0
+ pxor xmm1, xmm4 ; unoffset p0
+
+ movdqa xmm0, t0 ; p1
+ movdqa xmm4, t1 ; q1
+
+ ; write out order: xmm0 xmm2 xmm1 xmm3
+ lea rdx, [rsi + rax*4]
+
+ ; transpose back to write out
+ ; p1 f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
+ ; p0 f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01
+ ; q0 f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
+ ; q1 f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
+ movdqa xmm6, xmm0
+ punpcklbw xmm0, xmm1 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00
+ punpckhbw xmm6, xmm1 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80
+
+ movdqa xmm5, xmm3
+ punpcklbw xmm3, xmm4 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
+ punpckhbw xmm5, xmm4 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
+
+ movdqa xmm2, xmm0
+ punpcklwd xmm0, xmm3 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00
+ punpckhwd xmm2, xmm3 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40
+
+ movdqa xmm3, xmm6
+ punpcklwd xmm6, xmm5 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80
+ punpckhwd xmm3, xmm5 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0
+
+ movd [rsi], xmm6 ; write the second 8-line result
+ movd [rdx], xmm3
+ psrldq xmm6, 4
+ psrldq xmm3, 4
+ movd [rdi], xmm6
+ movd [rcx], xmm3
+ psrldq xmm6, 4
+ psrldq xmm3, 4
+ movd [rsi + rax*2], xmm6
+ movd [rdx + rax*2], xmm3
+ psrldq xmm6, 4
+ psrldq xmm3, 4
+ movd [rdi + rax*2], xmm6
+ movd [rcx + rax*2], xmm3
+
+ neg rax
+ lea rsi, [rsi + rax*8]
+ neg rax
+ lea rdi, [rsi + rax]
+ lea rdx, [rsi + rax*4]
+ lea rcx, [rdx + rax]
+
+ movd [rsi], xmm0 ; write the first 8-line result
+ movd [rdx], xmm2
+ psrldq xmm0, 4
+ psrldq xmm2, 4
+ movd [rdi], xmm0
+ movd [rcx], xmm2
+ psrldq xmm0, 4
+ psrldq xmm2, 4
+ movd [rsi + rax*2], xmm0
+ movd [rdx + rax*2], xmm2
+ psrldq xmm0, 4
+ psrldq xmm2, 4
+ movd [rdi + rax*2], xmm0
+ movd [rcx + rax*2], xmm2
+
+ add rsp, 32
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+tfe:
+ times 16 db 0xfe
+align 16
+t80:
+ times 16 db 0x80
+align 16
+t1s:
+ times 16 db 0x01
+align 16
+t3:
+ times 16 db 0x03
+align 16
+t4:
+ times 16 db 0x04
+align 16
+ones:
+ times 8 dw 0x0001
+align 16
+s9:
+ times 8 dw 0x0900
+align 16
+s63:
+ times 8 dw 0x003f
+align 16
+te0:
+ times 16 db 0xe0
+align 16
+t1f:
+ times 16 db 0x1f
diff --git a/libvpx/vp8/common/x86/loopfilter_x86.c b/libvpx/vp8/common/x86/loopfilter_x86.c
new file mode 100644
index 0000000..6586004
--- /dev/null
+++ b/libvpx/vp8/common/x86/loopfilter_x86.c
@@ -0,0 +1,198 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vp8/common/loopfilter.h"
+
+#define prototype_loopfilter(sym) \
+ void sym(unsigned char *src, int pitch, const unsigned char *blimit,\
+ const unsigned char *limit, const unsigned char *thresh, int count)
+
+#define prototype_loopfilter_nc(sym) \
+ void sym(unsigned char *src, int pitch, const unsigned char *blimit,\
+ const unsigned char *limit, const unsigned char *thresh)
+
+#define prototype_simple_loopfilter(sym) \
+ void sym(unsigned char *y, int ystride, const unsigned char *blimit)
+
+prototype_loopfilter(vp8_mbloop_filter_vertical_edge_mmx);
+prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_mmx);
+prototype_loopfilter(vp8_loop_filter_vertical_edge_mmx);
+prototype_loopfilter(vp8_loop_filter_horizontal_edge_mmx);
+prototype_simple_loopfilter(vp8_loop_filter_simple_horizontal_edge_mmx);
+prototype_simple_loopfilter(vp8_loop_filter_simple_vertical_edge_mmx);
+
+#if HAVE_SSE2 && ARCH_X86_64
+prototype_loopfilter(vp8_loop_filter_bv_y_sse2);
+prototype_loopfilter(vp8_loop_filter_bh_y_sse2);
+#else
+prototype_loopfilter_nc(vp8_loop_filter_vertical_edge_sse2);
+prototype_loopfilter_nc(vp8_loop_filter_horizontal_edge_sse2);
+#endif
+prototype_loopfilter_nc(vp8_mbloop_filter_vertical_edge_sse2);
+prototype_loopfilter_nc(vp8_mbloop_filter_horizontal_edge_sse2);
+
+extern loop_filter_uvfunction vp8_loop_filter_horizontal_edge_uv_sse2;
+extern loop_filter_uvfunction vp8_loop_filter_vertical_edge_uv_sse2;
+extern loop_filter_uvfunction vp8_mbloop_filter_horizontal_edge_uv_sse2;
+extern loop_filter_uvfunction vp8_mbloop_filter_vertical_edge_uv_sse2;
+
+#if HAVE_MMX
+/* Horizontal MB filtering */
+void vp8_loop_filter_mbh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_mbloop_filter_horizontal_edge_mmx(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_mbloop_filter_horizontal_edge_mmx(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_mbloop_filter_horizontal_edge_mmx(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+}
+
+
+/* Vertical MB Filtering */
+void vp8_loop_filter_mbv_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_mbloop_filter_vertical_edge_mmx(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_mbloop_filter_vertical_edge_mmx(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_mbloop_filter_vertical_edge_mmx(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
+}
+
+
+/* Horizontal B Filtering */
+void vp8_loop_filter_bh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_loop_filter_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_loop_filter_horizontal_edge_mmx(u_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_loop_filter_horizontal_edge_mmx(v_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+}
+
+
+void vp8_loop_filter_bhs_mmx(unsigned char *y_ptr, int y_stride, const unsigned char *blimit)
+{
+ vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, blimit);
+ vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, blimit);
+ vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, blimit);
+}
+
+
+/* Vertical B Filtering */
+void vp8_loop_filter_bv_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_loop_filter_vertical_edge_mmx(y_ptr + 4, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_vertical_edge_mmx(y_ptr + 8, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+ vp8_loop_filter_vertical_edge_mmx(y_ptr + 12, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+
+ if (u_ptr)
+ vp8_loop_filter_vertical_edge_mmx(u_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+
+ if (v_ptr)
+ vp8_loop_filter_vertical_edge_mmx(v_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
+}
+
+
+void vp8_loop_filter_bvs_mmx(unsigned char *y_ptr, int y_stride, const unsigned char *blimit)
+{
+ vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 4, y_stride, blimit);
+ vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 8, y_stride, blimit);
+ vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 12, y_stride, blimit);
+}
+#endif
+
+
+/* Horizontal MB filtering */
+#if HAVE_SSE2
+void vp8_loop_filter_mbh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_mbloop_filter_horizontal_edge_sse2(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr);
+
+ if (u_ptr)
+ vp8_mbloop_filter_horizontal_edge_uv_sse2(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, v_ptr);
+}
+
+
+/* Vertical MB Filtering */
+void vp8_loop_filter_mbv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+ vp8_mbloop_filter_vertical_edge_sse2(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr);
+
+ if (u_ptr)
+ vp8_mbloop_filter_vertical_edge_uv_sse2(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, v_ptr);
+}
+
+
+/* Horizontal B Filtering */
+void vp8_loop_filter_bh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+#if ARCH_X86_64
+ vp8_loop_filter_bh_y_sse2(y_ptr, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+#else
+ vp8_loop_filter_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
+ vp8_loop_filter_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
+ vp8_loop_filter_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
+#endif
+
+ if (u_ptr)
+ vp8_loop_filter_horizontal_edge_uv_sse2(u_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, v_ptr + 4 * uv_stride);
+}
+
+
+void vp8_loop_filter_bhs_sse2(unsigned char *y_ptr, int y_stride, const unsigned char *blimit)
+{
+ vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride, blimit);
+ vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride, blimit);
+ vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, blimit);
+}
+
+
+/* Vertical B Filtering */
+void vp8_loop_filter_bv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
+ int y_stride, int uv_stride, loop_filter_info *lfi)
+{
+#if ARCH_X86_64
+ vp8_loop_filter_bv_y_sse2(y_ptr, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
+#else
+ vp8_loop_filter_vertical_edge_sse2(y_ptr + 4, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
+ vp8_loop_filter_vertical_edge_sse2(y_ptr + 8, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
+ vp8_loop_filter_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
+#endif
+
+ if (u_ptr)
+ vp8_loop_filter_vertical_edge_uv_sse2(u_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, v_ptr + 4);
+}
+
+
+void vp8_loop_filter_bvs_sse2(unsigned char *y_ptr, int y_stride, const unsigned char *blimit)
+{
+ vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 4, y_stride, blimit);
+ vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 8, y_stride, blimit);
+ vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 12, y_stride, blimit);
+}
+
+#endif
diff --git a/libvpx/vp8/common/x86/mfqe_sse2.asm b/libvpx/vp8/common/x86/mfqe_sse2.asm
new file mode 100644
index 0000000..c1d2174
--- /dev/null
+++ b/libvpx/vp8/common/x86/mfqe_sse2.asm
@@ -0,0 +1,281 @@
+;
+; Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void vp8_filter_by_weight16x16_sse2
+;(
+; unsigned char *src,
+; int src_stride,
+; unsigned char *dst,
+; int dst_stride,
+; int src_weight
+;)
+global sym(vp8_filter_by_weight16x16_sse2) PRIVATE
+sym(vp8_filter_by_weight16x16_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movd xmm0, arg(4) ; src_weight
+ pshuflw xmm0, xmm0, 0x0 ; replicate to all low words
+ punpcklqdq xmm0, xmm0 ; replicate to all hi words
+
+ movdqa xmm1, [GLOBAL(tMFQE)]
+ psubw xmm1, xmm0 ; dst_weight
+
+ mov rax, arg(0) ; src
+ mov rsi, arg(1) ; src_stride
+ mov rdx, arg(2) ; dst
+ mov rdi, arg(3) ; dst_stride
+
+ mov rcx, 16 ; loop count
+ pxor xmm6, xmm6
+
+.combine
+ movdqa xmm2, [rax]
+ movdqa xmm4, [rdx]
+ add rax, rsi
+
+ ; src * src_weight
+ movdqa xmm3, xmm2
+ punpcklbw xmm2, xmm6
+ punpckhbw xmm3, xmm6
+ pmullw xmm2, xmm0
+ pmullw xmm3, xmm0
+
+ ; dst * dst_weight
+ movdqa xmm5, xmm4
+ punpcklbw xmm4, xmm6
+ punpckhbw xmm5, xmm6
+ pmullw xmm4, xmm1
+ pmullw xmm5, xmm1
+
+ ; sum, round and shift
+ paddw xmm2, xmm4
+ paddw xmm3, xmm5
+ paddw xmm2, [GLOBAL(tMFQE_round)]
+ paddw xmm3, [GLOBAL(tMFQE_round)]
+ psrlw xmm2, 4
+ psrlw xmm3, 4
+
+ packuswb xmm2, xmm3
+ movdqa [rdx], xmm2
+ add rdx, rdi
+
+ dec rcx
+ jnz .combine
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+
+ ret
+
+;void vp8_filter_by_weight8x8_sse2
+;(
+; unsigned char *src,
+; int src_stride,
+; unsigned char *dst,
+; int dst_stride,
+; int src_weight
+;)
+global sym(vp8_filter_by_weight8x8_sse2) PRIVATE
+sym(vp8_filter_by_weight8x8_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movd xmm0, arg(4) ; src_weight
+ pshuflw xmm0, xmm0, 0x0 ; replicate to all low words
+ punpcklqdq xmm0, xmm0 ; replicate to all hi words
+
+ movdqa xmm1, [GLOBAL(tMFQE)]
+ psubw xmm1, xmm0 ; dst_weight
+
+ mov rax, arg(0) ; src
+ mov rsi, arg(1) ; src_stride
+ mov rdx, arg(2) ; dst
+ mov rdi, arg(3) ; dst_stride
+
+ mov rcx, 8 ; loop count
+ pxor xmm4, xmm4
+
+.combine
+ movq xmm2, [rax]
+ movq xmm3, [rdx]
+ add rax, rsi
+
+ ; src * src_weight
+ punpcklbw xmm2, xmm4
+ pmullw xmm2, xmm0
+
+ ; dst * dst_weight
+ punpcklbw xmm3, xmm4
+ pmullw xmm3, xmm1
+
+ ; sum, round and shift
+ paddw xmm2, xmm3
+ paddw xmm2, [GLOBAL(tMFQE_round)]
+ psrlw xmm2, 4
+
+ packuswb xmm2, xmm4
+ movq [rdx], xmm2
+ add rdx, rdi
+
+ dec rcx
+ jnz .combine
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+
+ ret
+
+;void vp8_variance_and_sad_16x16_sse2 | arg
+;(
+; unsigned char *src1, 0
+; int stride1, 1
+; unsigned char *src2, 2
+; int stride2, 3
+; unsigned int *variance, 4
+; unsigned int *sad, 5
+;)
+global sym(vp8_variance_and_sad_16x16_sse2) PRIVATE
+sym(vp8_variance_and_sad_16x16_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rax, arg(0) ; src1
+ mov rcx, arg(1) ; stride1
+ mov rdx, arg(2) ; src2
+ mov rdi, arg(3) ; stride2
+
+ mov rsi, 16 ; block height
+
+ ; Prep accumulator registers
+ pxor xmm3, xmm3 ; SAD
+ pxor xmm4, xmm4 ; sum of src2
+ pxor xmm5, xmm5 ; sum of src2^2
+
+ ; Because we're working with the actual output frames
+ ; we can't depend on any kind of data alignment.
+.accumulate
+ movdqa xmm0, [rax] ; src1
+ movdqa xmm1, [rdx] ; src2
+ add rax, rcx ; src1 + stride1
+ add rdx, rdi ; src2 + stride2
+
+ ; SAD(src1, src2)
+ psadbw xmm0, xmm1
+ paddusw xmm3, xmm0
+
+ ; SUM(src2)
+ pxor xmm2, xmm2
+ psadbw xmm2, xmm1 ; sum src2 by misusing SAD against 0
+ paddusw xmm4, xmm2
+
+ ; pmaddubsw would be ideal if it took two unsigned values. instead,
+ ; it expects a signed and an unsigned value. so instead we zero extend
+ ; and operate on words.
+ pxor xmm2, xmm2
+ movdqa xmm0, xmm1
+ punpcklbw xmm0, xmm2
+ punpckhbw xmm1, xmm2
+ pmaddwd xmm0, xmm0
+ pmaddwd xmm1, xmm1
+ paddd xmm5, xmm0
+ paddd xmm5, xmm1
+
+ sub rsi, 1
+ jnz .accumulate
+
+ ; phaddd only operates on adjacent double words.
+ ; Finalize SAD and store
+ movdqa xmm0, xmm3
+ psrldq xmm0, 8
+ paddusw xmm0, xmm3
+ paddd xmm0, [GLOBAL(t128)]
+ psrld xmm0, 8
+
+ mov rax, arg(5)
+ movd [rax], xmm0
+
+ ; Accumulate sum of src2
+ movdqa xmm0, xmm4
+ psrldq xmm0, 8
+ paddusw xmm0, xmm4
+ ; Square src2. Ignore high value
+ pmuludq xmm0, xmm0
+ psrld xmm0, 8
+
+ ; phaddw could be used to sum adjacent values but we want
+ ; all the values summed. promote to doubles, accumulate,
+ ; shift and sum
+ pxor xmm2, xmm2
+ movdqa xmm1, xmm5
+ punpckldq xmm1, xmm2
+ punpckhdq xmm5, xmm2
+ paddd xmm1, xmm5
+ movdqa xmm2, xmm1
+ psrldq xmm1, 8
+ paddd xmm1, xmm2
+
+ psubd xmm1, xmm0
+
+ ; (variance + 128) >> 8
+ paddd xmm1, [GLOBAL(t128)]
+ psrld xmm1, 8
+ mov rax, arg(4)
+
+ movd [rax], xmm1
+
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+t128:
+ ddq 128
+align 16
+tMFQE: ; 1 << MFQE_PRECISION
+ times 8 dw 0x10
+align 16
+tMFQE_round: ; 1 << (MFQE_PRECISION - 1)
+ times 8 dw 0x08
+
diff --git a/libvpx/vp8/common/x86/postproc_mmx.asm b/libvpx/vp8/common/x86/postproc_mmx.asm
new file mode 100644
index 0000000..966c586
--- /dev/null
+++ b/libvpx/vp8/common/x86/postproc_mmx.asm
@@ -0,0 +1,314 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%define VP8_FILTER_WEIGHT 128
+%define VP8_FILTER_SHIFT 7
+
+;void vp8_mbpost_proc_down_mmx(unsigned char *dst,
+; int pitch, int rows, int cols,int flimit)
+extern sym(vp8_rv)
+global sym(vp8_mbpost_proc_down_mmx) PRIVATE
+sym(vp8_mbpost_proc_down_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 136
+
+ ; unsigned char d[16][8] at [rsp]
+ ; create flimit2 at [rsp+128]
+ mov eax, dword ptr arg(4) ;flimit
+ mov [rsp+128], eax
+ mov [rsp+128+4], eax
+%define flimit2 [rsp+128]
+
+%if ABI_IS_32BIT=0
+ lea r8, [GLOBAL(sym(vp8_rv))]
+%endif
+
+ ;rows +=8;
+ add dword ptr arg(2), 8
+
+ ;for(c=0; c<cols; c+=4)
+.loop_col:
+ mov rsi, arg(0) ;s
+ pxor mm0, mm0 ;
+
+ movsxd rax, dword ptr arg(1) ;pitch ;
+
+ ; this copies the last row down into the border 8 rows
+ mov rdi, rsi
+ mov rdx, arg(2)
+ sub rdx, 9
+ imul rdx, rax
+ lea rdi, [rdi+rdx]
+ movq mm1, QWORD ptr[rdi] ; first row
+ mov rcx, 8
+.init_borderd ; initialize borders
+ lea rdi, [rdi + rax]
+ movq [rdi], xmm1
+
+ dec rcx
+ jne .init_borderd
+
+ neg rax ; rax = -pitch
+
+ ; this copies the first row up into the border 8 rows
+ mov rdi, rsi
+ movq mm1, QWORD ptr[rdi] ; first row
+ mov rcx, 8
+.init_border ; initialize borders
+ lea rdi, [rdi + rax]
+ movq [rdi], mm1
+
+ dec rcx
+ jne .init_border
+
+
+ lea rsi, [rsi + rax*8]; ; rdi = s[-pitch*8]
+ neg rax
+
+
+ pxor mm5, mm5
+ pxor mm6, mm6 ;
+
+ pxor mm7, mm7 ;
+ mov rdi, rsi
+
+ mov rcx, 15 ;
+
+.loop_initvar:
+ movd mm1, DWORD PTR [rdi];
+ punpcklbw mm1, mm0 ;
+
+ paddw mm5, mm1 ;
+ pmullw mm1, mm1 ;
+
+ movq mm2, mm1 ;
+ punpcklwd mm1, mm0 ;
+
+ punpckhwd mm2, mm0 ;
+ paddd mm6, mm1 ;
+
+ paddd mm7, mm2 ;
+ lea rdi, [rdi+rax] ;
+
+ dec rcx
+ jne .loop_initvar
+ ;save the var and sum
+ xor rdx, rdx
+.loop_row:
+ movd mm1, DWORD PTR [rsi] ; [s-pitch*8]
+ movd mm2, DWORD PTR [rdi] ; [s+pitch*7]
+
+ punpcklbw mm1, mm0
+ punpcklbw mm2, mm0
+
+ paddw mm5, mm2
+ psubw mm5, mm1
+
+ pmullw mm2, mm2
+ movq mm4, mm2
+
+ punpcklwd mm2, mm0
+ punpckhwd mm4, mm0
+
+ paddd mm6, mm2
+ paddd mm7, mm4
+
+ pmullw mm1, mm1
+ movq mm2, mm1
+
+ punpcklwd mm1, mm0
+ psubd mm6, mm1
+
+ punpckhwd mm2, mm0
+ psubd mm7, mm2
+
+
+ movq mm3, mm6
+ pslld mm3, 4
+
+ psubd mm3, mm6
+ movq mm1, mm5
+
+ movq mm4, mm5
+ pmullw mm1, mm1
+
+ pmulhw mm4, mm4
+ movq mm2, mm1
+
+ punpcklwd mm1, mm4
+ punpckhwd mm2, mm4
+
+ movq mm4, mm7
+ pslld mm4, 4
+
+ psubd mm4, mm7
+
+ psubd mm3, mm1
+ psubd mm4, mm2
+
+ psubd mm3, flimit2
+ psubd mm4, flimit2
+
+ psrad mm3, 31
+ psrad mm4, 31
+
+ packssdw mm3, mm4
+ packsswb mm3, mm0
+
+ movd mm1, DWORD PTR [rsi+rax*8]
+
+ movq mm2, mm1
+ punpcklbw mm1, mm0
+
+ paddw mm1, mm5
+ mov rcx, rdx
+
+ and rcx, 127
+%if ABI_IS_32BIT=1 && CONFIG_PIC=1
+ push rax
+ lea rax, [GLOBAL(sym(vp8_rv))]
+ movq mm4, [rax + rcx*2] ;vp8_rv[rcx*2]
+ pop rax
+%elif ABI_IS_32BIT=0
+ movq mm4, [r8 + rcx*2] ;vp8_rv[rcx*2]
+%else
+ movq mm4, [sym(vp8_rv) + rcx*2]
+%endif
+ paddw mm1, mm4
+ ;paddw xmm1, eight8s
+ psraw mm1, 4
+
+ packuswb mm1, mm0
+ pand mm1, mm3
+
+ pandn mm3, mm2
+ por mm1, mm3
+
+ and rcx, 15
+ movd DWORD PTR [rsp+rcx*4], mm1 ;d[rcx*4]
+
+ mov rcx, rdx
+ sub rcx, 8
+
+ and rcx, 15
+ movd mm1, DWORD PTR [rsp+rcx*4] ;d[rcx*4]
+
+ movd [rsi], mm1
+ lea rsi, [rsi+rax]
+
+ lea rdi, [rdi+rax]
+ add rdx, 1
+
+ cmp edx, dword arg(2) ;rows
+ jl .loop_row
+
+
+ add dword arg(0), 4 ; s += 4
+ sub dword arg(3), 4 ; cols -= 4
+ cmp dword arg(3), 0
+ jg .loop_col
+
+ add rsp, 136
+ pop rsp
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+%undef flimit2
+
+
+;void vp8_plane_add_noise_mmx (unsigned char *Start, unsigned char *noise,
+; unsigned char blackclamp[16],
+; unsigned char whiteclamp[16],
+; unsigned char bothclamp[16],
+; unsigned int Width, unsigned int Height, int Pitch)
+extern sym(rand)
+global sym(vp8_plane_add_noise_mmx) PRIVATE
+sym(vp8_plane_add_noise_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 8
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+.addnoise_loop:
+ call sym(rand) WRT_PLT
+ mov rcx, arg(1) ;noise
+ and rax, 0xff
+ add rcx, rax
+
+ ; we rely on the fact that the clamping vectors are stored contiguously
+ ; in black/white/both order. Note that we have to reload this here because
+ ; rdx could be trashed by rand()
+ mov rdx, arg(2) ; blackclamp
+
+
+ mov rdi, rcx
+ movsxd rcx, dword arg(5) ;[Width]
+ mov rsi, arg(0) ;Pos
+ xor rax,rax
+
+.addnoise_nextset:
+ movq mm1,[rsi+rax] ; get the source
+
+ psubusb mm1, [rdx] ;blackclamp ; clamp both sides so we don't outrange adding noise
+ paddusb mm1, [rdx+32] ;bothclamp
+ psubusb mm1, [rdx+16] ;whiteclamp
+
+ movq mm2,[rdi+rax] ; get the noise for this line
+ paddb mm1,mm2 ; add it in
+ movq [rsi+rax],mm1 ; store the result
+
+ add rax,8 ; move to the next line
+
+ cmp rax, rcx
+ jl .addnoise_nextset
+
+ movsxd rax, dword arg(7) ; Pitch
+ add arg(0), rax ; Start += Pitch
+ sub dword arg(6), 1 ; Height -= 1
+ jg .addnoise_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+SECTION_RODATA
+align 16
+Blur:
+ times 16 dw 16
+ times 8 dw 64
+ times 16 dw 16
+ times 8 dw 0
+
+rd:
+ times 4 dw 0x40
diff --git a/libvpx/vp8/common/x86/postproc_sse2.asm b/libvpx/vp8/common/x86/postproc_sse2.asm
new file mode 100644
index 0000000..00f84a3
--- /dev/null
+++ b/libvpx/vp8/common/x86/postproc_sse2.asm
@@ -0,0 +1,721 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;macro in deblock functions
+%macro FIRST_2_ROWS 0
+ movdqa xmm4, xmm0
+ movdqa xmm6, xmm0
+ movdqa xmm5, xmm1
+ pavgb xmm5, xmm3
+
+ ;calculate absolute value
+ psubusb xmm4, xmm1
+ psubusb xmm1, xmm0
+ psubusb xmm6, xmm3
+ psubusb xmm3, xmm0
+ paddusb xmm4, xmm1
+ paddusb xmm6, xmm3
+
+ ;get threshold
+ movdqa xmm2, flimit
+ pxor xmm1, xmm1
+ movdqa xmm7, xmm2
+
+ ;get mask
+ psubusb xmm2, xmm4
+ psubusb xmm7, xmm6
+ pcmpeqb xmm2, xmm1
+ pcmpeqb xmm7, xmm1
+ por xmm7, xmm2
+%endmacro
+
+%macro SECOND_2_ROWS 0
+ movdqa xmm6, xmm0
+ movdqa xmm4, xmm0
+ movdqa xmm2, xmm1
+ pavgb xmm1, xmm3
+
+ ;calculate absolute value
+ psubusb xmm6, xmm2
+ psubusb xmm2, xmm0
+ psubusb xmm4, xmm3
+ psubusb xmm3, xmm0
+ paddusb xmm6, xmm2
+ paddusb xmm4, xmm3
+
+ pavgb xmm5, xmm1
+
+ ;get threshold
+ movdqa xmm2, flimit
+ pxor xmm1, xmm1
+ movdqa xmm3, xmm2
+
+ ;get mask
+ psubusb xmm2, xmm6
+ psubusb xmm3, xmm4
+ pcmpeqb xmm2, xmm1
+ pcmpeqb xmm3, xmm1
+
+ por xmm7, xmm2
+ por xmm7, xmm3
+
+ pavgb xmm5, xmm0
+
+ ;decide if or not to use filtered value
+ pand xmm0, xmm7
+ pandn xmm7, xmm5
+ paddusb xmm0, xmm7
+%endmacro
+
+%macro UPDATE_FLIMIT 0
+ movdqa xmm2, XMMWORD PTR [rbx]
+ movdqa [rsp], xmm2
+ add rbx, 16
+%endmacro
+
+;void vp8_post_proc_down_and_across_mb_row_sse2
+;(
+; unsigned char *src_ptr,
+; unsigned char *dst_ptr,
+; int src_pixels_per_line,
+; int dst_pixels_per_line,
+; int cols,
+; int *flimits,
+; int size
+;)
+global sym(vp8_post_proc_down_and_across_mb_row_sse2) PRIVATE
+sym(vp8_post_proc_down_and_across_mb_row_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ push rbx
+ push rsi
+ push rdi
+ ; end prolog
+ ALIGN_STACK 16, rax
+ sub rsp, 16
+
+ ; put flimit on stack
+ mov rbx, arg(5) ;flimits ptr
+ UPDATE_FLIMIT
+
+%define flimit [rsp]
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(1) ;dst_ptr
+
+ movsxd rax, DWORD PTR arg(2) ;src_pixels_per_line
+ movsxd rcx, DWORD PTR arg(6) ;rows in a macroblock
+.nextrow:
+ xor rdx, rdx ;col
+.nextcol:
+ ;load current and next 2 rows
+ movdqu xmm0, XMMWORD PTR [rsi]
+ movdqu xmm1, XMMWORD PTR [rsi + rax]
+ movdqu xmm3, XMMWORD PTR [rsi + 2*rax]
+
+ FIRST_2_ROWS
+
+ ;load above 2 rows
+ neg rax
+ movdqu xmm1, XMMWORD PTR [rsi + 2*rax]
+ movdqu xmm3, XMMWORD PTR [rsi + rax]
+
+ SECOND_2_ROWS
+
+ movdqu XMMWORD PTR [rdi], xmm0
+
+ neg rax ; positive stride
+ add rsi, 16
+ add rdi, 16
+
+ add rdx, 16
+ cmp edx, dword arg(4) ;cols
+ jge .downdone
+ UPDATE_FLIMIT
+ jmp .nextcol
+
+.downdone:
+ ; done with the all cols, start the across filtering in place
+ sub rsi, rdx
+ sub rdi, rdx
+
+ mov rbx, arg(5) ; flimits
+ UPDATE_FLIMIT
+
+ ; dup the first byte into the left border 8 times
+ movq mm1, [rdi]
+ punpcklbw mm1, mm1
+ punpcklwd mm1, mm1
+ punpckldq mm1, mm1
+ mov rdx, -8
+ movq [rdi+rdx], mm1
+
+ ; dup the last byte into the right border
+ movsxd rdx, dword arg(4)
+ movq mm1, [rdi + rdx + -1]
+ punpcklbw mm1, mm1
+ punpcklwd mm1, mm1
+ punpckldq mm1, mm1
+ movq [rdi+rdx], mm1
+
+ xor rdx, rdx
+ movq mm0, QWORD PTR [rdi-16];
+ movq mm1, QWORD PTR [rdi-8];
+
+.acrossnextcol:
+ movdqu xmm0, XMMWORD PTR [rdi + rdx]
+ movdqu xmm1, XMMWORD PTR [rdi + rdx -2]
+ movdqu xmm3, XMMWORD PTR [rdi + rdx -1]
+
+ FIRST_2_ROWS
+
+ movdqu xmm1, XMMWORD PTR [rdi + rdx +1]
+ movdqu xmm3, XMMWORD PTR [rdi + rdx +2]
+
+ SECOND_2_ROWS
+
+ movq QWORD PTR [rdi+rdx-16], mm0 ; store previous 8 bytes
+ movq QWORD PTR [rdi+rdx-8], mm1 ; store previous 8 bytes
+ movdq2q mm0, xmm0
+ psrldq xmm0, 8
+ movdq2q mm1, xmm0
+
+ add rdx, 16
+ cmp edx, dword arg(4) ;cols
+ jge .acrossdone
+ UPDATE_FLIMIT
+ jmp .acrossnextcol
+
+.acrossdone
+ ; last 16 pixels
+ movq QWORD PTR [rdi+rdx-16], mm0
+
+ cmp edx, dword arg(4)
+ jne .throw_last_8
+ movq QWORD PTR [rdi+rdx-8], mm1
+.throw_last_8:
+ ; done with this rwo
+ add rsi,rax ;next src line
+ mov eax, dword arg(3) ;dst_pixels_per_line
+ add rdi,rax ;next destination
+ mov eax, dword arg(2) ;src_pixels_per_line
+
+ mov rbx, arg(5) ;flimits
+ UPDATE_FLIMIT
+
+ dec rcx ;decrement count
+ jnz .nextrow ;next row
+
+ add rsp, 16
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ pop rbx
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+%undef flimit
+
+;void vp8_mbpost_proc_down_xmm(unsigned char *dst,
+; int pitch, int rows, int cols,int flimit)
+extern sym(vp8_rv)
+global sym(vp8_mbpost_proc_down_xmm) PRIVATE
+sym(vp8_mbpost_proc_down_xmm):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 128+16
+
+ ; unsigned char d[16][8] at [rsp]
+ ; create flimit2 at [rsp+128]
+ mov eax, dword ptr arg(4) ;flimit
+ mov [rsp+128], eax
+ mov [rsp+128+4], eax
+ mov [rsp+128+8], eax
+ mov [rsp+128+12], eax
+%define flimit4 [rsp+128]
+
+%if ABI_IS_32BIT=0
+ lea r8, [GLOBAL(sym(vp8_rv))]
+%endif
+
+ ;rows +=8;
+ add dword arg(2), 8
+
+ ;for(c=0; c<cols; c+=8)
+.loop_col:
+ mov rsi, arg(0) ; s
+ pxor xmm0, xmm0 ;
+
+ movsxd rax, dword ptr arg(1) ;pitch ;
+
+ ; this copies the last row down into the border 8 rows
+ mov rdi, rsi
+ mov rdx, arg(2)
+ sub rdx, 9
+ imul rdx, rax
+ lea rdi, [rdi+rdx]
+ movq xmm1, QWORD ptr[rdi] ; first row
+ mov rcx, 8
+.init_borderd ; initialize borders
+ lea rdi, [rdi + rax]
+ movq [rdi], xmm1
+
+ dec rcx
+ jne .init_borderd
+
+ neg rax ; rax = -pitch
+
+ ; this copies the first row up into the border 8 rows
+ mov rdi, rsi
+ movq xmm1, QWORD ptr[rdi] ; first row
+ mov rcx, 8
+.init_border ; initialize borders
+ lea rdi, [rdi + rax]
+ movq [rdi], xmm1
+
+ dec rcx
+ jne .init_border
+
+
+
+ lea rsi, [rsi + rax*8]; ; rdi = s[-pitch*8]
+ neg rax
+
+ pxor xmm5, xmm5
+ pxor xmm6, xmm6 ;
+
+ pxor xmm7, xmm7 ;
+ mov rdi, rsi
+
+ mov rcx, 15 ;
+
+.loop_initvar:
+ movq xmm1, QWORD PTR [rdi];
+ punpcklbw xmm1, xmm0 ;
+
+ paddw xmm5, xmm1 ;
+ pmullw xmm1, xmm1 ;
+
+ movdqa xmm2, xmm1 ;
+ punpcklwd xmm1, xmm0 ;
+
+ punpckhwd xmm2, xmm0 ;
+ paddd xmm6, xmm1 ;
+
+ paddd xmm7, xmm2 ;
+ lea rdi, [rdi+rax] ;
+
+ dec rcx
+ jne .loop_initvar
+ ;save the var and sum
+ xor rdx, rdx
+.loop_row:
+ movq xmm1, QWORD PTR [rsi] ; [s-pitch*8]
+ movq xmm2, QWORD PTR [rdi] ; [s+pitch*7]
+
+ punpcklbw xmm1, xmm0
+ punpcklbw xmm2, xmm0
+
+ paddw xmm5, xmm2
+ psubw xmm5, xmm1
+
+ pmullw xmm2, xmm2
+ movdqa xmm4, xmm2
+
+ punpcklwd xmm2, xmm0
+ punpckhwd xmm4, xmm0
+
+ paddd xmm6, xmm2
+ paddd xmm7, xmm4
+
+ pmullw xmm1, xmm1
+ movdqa xmm2, xmm1
+
+ punpcklwd xmm1, xmm0
+ psubd xmm6, xmm1
+
+ punpckhwd xmm2, xmm0
+ psubd xmm7, xmm2
+
+
+ movdqa xmm3, xmm6
+ pslld xmm3, 4
+
+ psubd xmm3, xmm6
+ movdqa xmm1, xmm5
+
+ movdqa xmm4, xmm5
+ pmullw xmm1, xmm1
+
+ pmulhw xmm4, xmm4
+ movdqa xmm2, xmm1
+
+ punpcklwd xmm1, xmm4
+ punpckhwd xmm2, xmm4
+
+ movdqa xmm4, xmm7
+ pslld xmm4, 4
+
+ psubd xmm4, xmm7
+
+ psubd xmm3, xmm1
+ psubd xmm4, xmm2
+
+ psubd xmm3, flimit4
+ psubd xmm4, flimit4
+
+ psrad xmm3, 31
+ psrad xmm4, 31
+
+ packssdw xmm3, xmm4
+ packsswb xmm3, xmm0
+
+ movq xmm1, QWORD PTR [rsi+rax*8]
+
+ movq xmm2, xmm1
+ punpcklbw xmm1, xmm0
+
+ paddw xmm1, xmm5
+ mov rcx, rdx
+
+ and rcx, 127
+%if ABI_IS_32BIT=1 && CONFIG_PIC=1
+ push rax
+ lea rax, [GLOBAL(sym(vp8_rv))]
+ movdqu xmm4, [rax + rcx*2] ;vp8_rv[rcx*2]
+ pop rax
+%elif ABI_IS_32BIT=0
+ movdqu xmm4, [r8 + rcx*2] ;vp8_rv[rcx*2]
+%else
+ movdqu xmm4, [sym(vp8_rv) + rcx*2]
+%endif
+
+ paddw xmm1, xmm4
+ ;paddw xmm1, eight8s
+ psraw xmm1, 4
+
+ packuswb xmm1, xmm0
+ pand xmm1, xmm3
+
+ pandn xmm3, xmm2
+ por xmm1, xmm3
+
+ and rcx, 15
+ movq QWORD PTR [rsp + rcx*8], xmm1 ;d[rcx*8]
+
+ mov rcx, rdx
+ sub rcx, 8
+
+ and rcx, 15
+ movq mm0, [rsp + rcx*8] ;d[rcx*8]
+
+ movq [rsi], mm0
+ lea rsi, [rsi+rax]
+
+ lea rdi, [rdi+rax]
+ add rdx, 1
+
+ cmp edx, dword arg(2) ;rows
+ jl .loop_row
+
+ add dword arg(0), 8 ; s += 8
+ sub dword arg(3), 8 ; cols -= 8
+ cmp dword arg(3), 0
+ jg .loop_col
+
+ add rsp, 128+16
+ pop rsp
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+%undef flimit4
+
+
+;void vp8_mbpost_proc_across_ip_xmm(unsigned char *src,
+; int pitch, int rows, int cols,int flimit)
+global sym(vp8_mbpost_proc_across_ip_xmm) PRIVATE
+sym(vp8_mbpost_proc_across_ip_xmm):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 16
+
+ ; create flimit4 at [rsp]
+ mov eax, dword ptr arg(4) ;flimit
+ mov [rsp], eax
+ mov [rsp+4], eax
+ mov [rsp+8], eax
+ mov [rsp+12], eax
+%define flimit4 [rsp]
+
+
+ ;for(r=0;r<rows;r++)
+.ip_row_loop:
+
+ xor rdx, rdx ;sumsq=0;
+ xor rcx, rcx ;sum=0;
+ mov rsi, arg(0); s
+
+
+ ; dup the first byte into the left border 8 times
+ movq mm1, [rsi]
+ punpcklbw mm1, mm1
+ punpcklwd mm1, mm1
+ punpckldq mm1, mm1
+
+ mov rdi, -8
+ movq [rsi+rdi], mm1
+
+ ; dup the last byte into the right border
+ movsxd rdx, dword arg(3)
+ movq mm1, [rsi + rdx + -1]
+ punpcklbw mm1, mm1
+ punpcklwd mm1, mm1
+ punpckldq mm1, mm1
+ movq [rsi+rdx], mm1
+
+.ip_var_loop:
+ ;for(i=-8;i<=6;i++)
+ ;{
+ ; sumsq += s[i]*s[i];
+ ; sum += s[i];
+ ;}
+ movzx eax, byte [rsi+rdi]
+ add ecx, eax
+ mul al
+ add edx, eax
+ add rdi, 1
+ cmp rdi, 6
+ jle .ip_var_loop
+
+
+ ;mov rax, sumsq
+ ;movd xmm7, rax
+ movd xmm7, edx
+
+ ;mov rax, sum
+ ;movd xmm6, rax
+ movd xmm6, ecx
+
+ mov rsi, arg(0) ;s
+ xor rcx, rcx
+
+ movsxd rdx, dword arg(3) ;cols
+ add rdx, 8
+ pxor mm0, mm0
+ pxor mm1, mm1
+
+ pxor xmm0, xmm0
+.nextcol4:
+
+ movd xmm1, DWORD PTR [rsi+rcx-8] ; -8 -7 -6 -5
+ movd xmm2, DWORD PTR [rsi+rcx+7] ; +7 +8 +9 +10
+
+ punpcklbw xmm1, xmm0 ; expanding
+ punpcklbw xmm2, xmm0 ; expanding
+
+ punpcklwd xmm1, xmm0 ; expanding to dwords
+ punpcklwd xmm2, xmm0 ; expanding to dwords
+
+ psubd xmm2, xmm1 ; 7--8 8--7 9--6 10--5
+ paddd xmm1, xmm1 ; -8*2 -7*2 -6*2 -5*2
+
+ paddd xmm1, xmm2 ; 7+-8 8+-7 9+-6 10+-5
+ pmaddwd xmm1, xmm2 ; squared of 7+-8 8+-7 9+-6 10+-5
+
+ paddd xmm6, xmm2
+ paddd xmm7, xmm1
+
+ pshufd xmm6, xmm6, 0 ; duplicate the last ones
+ pshufd xmm7, xmm7, 0 ; duplicate the last ones
+
+ psrldq xmm1, 4 ; 8--7 9--6 10--5 0000
+ psrldq xmm2, 4 ; 8--7 9--6 10--5 0000
+
+ pshufd xmm3, xmm1, 3 ; 0000 8--7 8--7 8--7 squared
+ pshufd xmm4, xmm2, 3 ; 0000 8--7 8--7 8--7 squared
+
+ paddd xmm6, xmm4
+ paddd xmm7, xmm3
+
+ pshufd xmm3, xmm1, 01011111b ; 0000 0000 9--6 9--6 squared
+ pshufd xmm4, xmm2, 01011111b ; 0000 0000 9--6 9--6 squared
+
+ paddd xmm7, xmm3
+ paddd xmm6, xmm4
+
+ pshufd xmm3, xmm1, 10111111b ; 0000 0000 8--7 8--7 squared
+ pshufd xmm4, xmm2, 10111111b ; 0000 0000 8--7 8--7 squared
+
+ paddd xmm7, xmm3
+ paddd xmm6, xmm4
+
+ movdqa xmm3, xmm6
+ pmaddwd xmm3, xmm3
+
+ movdqa xmm5, xmm7
+ pslld xmm5, 4
+
+ psubd xmm5, xmm7
+ psubd xmm5, xmm3
+
+ psubd xmm5, flimit4
+ psrad xmm5, 31
+
+ packssdw xmm5, xmm0
+ packsswb xmm5, xmm0
+
+ movd xmm1, DWORD PTR [rsi+rcx]
+ movq xmm2, xmm1
+
+ punpcklbw xmm1, xmm0
+ punpcklwd xmm1, xmm0
+
+ paddd xmm1, xmm6
+ paddd xmm1, [GLOBAL(four8s)]
+
+ psrad xmm1, 4
+ packssdw xmm1, xmm0
+
+ packuswb xmm1, xmm0
+ pand xmm1, xmm5
+
+ pandn xmm5, xmm2
+ por xmm5, xmm1
+
+ movd [rsi+rcx-8], mm0
+ movq mm0, mm1
+
+ movdq2q mm1, xmm5
+ psrldq xmm7, 12
+
+ psrldq xmm6, 12
+ add rcx, 4
+
+ cmp rcx, rdx
+ jl .nextcol4
+
+ ;s+=pitch;
+ movsxd rax, dword arg(1)
+ add arg(0), rax
+
+ sub dword arg(2), 1 ;rows-=1
+ cmp dword arg(2), 0
+ jg .ip_row_loop
+
+ add rsp, 16
+ pop rsp
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+%undef flimit4
+
+
+;void vp8_plane_add_noise_wmt (unsigned char *Start, unsigned char *noise,
+; unsigned char blackclamp[16],
+; unsigned char whiteclamp[16],
+; unsigned char bothclamp[16],
+; unsigned int Width, unsigned int Height, int Pitch)
+extern sym(rand)
+global sym(vp8_plane_add_noise_wmt) PRIVATE
+sym(vp8_plane_add_noise_wmt):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 8
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+.addnoise_loop:
+ call sym(rand) WRT_PLT
+ mov rcx, arg(1) ;noise
+ and rax, 0xff
+ add rcx, rax
+
+ ; we rely on the fact that the clamping vectors are stored contiguously
+ ; in black/white/both order. Note that we have to reload this here because
+ ; rdx could be trashed by rand()
+ mov rdx, arg(2) ; blackclamp
+
+
+ mov rdi, rcx
+ movsxd rcx, dword arg(5) ;[Width]
+ mov rsi, arg(0) ;Pos
+ xor rax,rax
+
+.addnoise_nextset:
+ movdqu xmm1,[rsi+rax] ; get the source
+
+ psubusb xmm1, [rdx] ;blackclamp ; clamp both sides so we don't outrange adding noise
+ paddusb xmm1, [rdx+32] ;bothclamp
+ psubusb xmm1, [rdx+16] ;whiteclamp
+
+ movdqu xmm2,[rdi+rax] ; get the noise for this line
+ paddb xmm1,xmm2 ; add it in
+ movdqu [rsi+rax],xmm1 ; store the result
+
+ add rax,16 ; move to the next line
+
+ cmp rax, rcx
+ jl .addnoise_nextset
+
+ movsxd rax, dword arg(7) ; Pitch
+ add arg(0), rax ; Start += Pitch
+ sub dword arg(6), 1 ; Height -= 1
+ jg .addnoise_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+SECTION_RODATA
+align 16
+four8s:
+ times 4 dd 8
diff --git a/libvpx/vp8/common/x86/postproc_x86.c b/libvpx/vp8/common/x86/postproc_x86.c
new file mode 100644
index 0000000..3ec0106
--- /dev/null
+++ b/libvpx/vp8/common/x86/postproc_x86.c
@@ -0,0 +1,24 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+/* On Android NDK, rand is inlined function, but postproc needs rand symbol */
+#if defined(__ANDROID__)
+#define rand __rand
+#include <stdlib.h>
+#undef rand
+
+extern int rand(void)
+{
+ return __rand();
+}
+#else
+/* ISO C forbids an empty translation unit. */
+int vp8_unused;
+#endif
diff --git a/libvpx/vp8/common/x86/recon_mmx.asm b/libvpx/vp8/common/x86/recon_mmx.asm
new file mode 100644
index 0000000..15e9871
--- /dev/null
+++ b/libvpx/vp8/common/x86/recon_mmx.asm
@@ -0,0 +1,274 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+
+;void copy_mem8x8_mmx(
+; unsigned char *src,
+; int src_stride,
+; unsigned char *dst,
+; int dst_stride
+; )
+global sym(vp8_copy_mem8x8_mmx) PRIVATE
+sym(vp8_copy_mem8x8_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src;
+ movq mm0, [rsi]
+
+ movsxd rax, dword ptr arg(1) ;src_stride;
+ mov rdi, arg(2) ;dst;
+
+ movq mm1, [rsi+rax]
+ movq mm2, [rsi+rax*2]
+
+ movsxd rcx, dword ptr arg(3) ;dst_stride
+ lea rsi, [rsi+rax*2]
+
+ movq [rdi], mm0
+ add rsi, rax
+
+ movq [rdi+rcx], mm1
+ movq [rdi+rcx*2], mm2
+
+
+ lea rdi, [rdi+rcx*2]
+ movq mm3, [rsi]
+
+ add rdi, rcx
+ movq mm4, [rsi+rax]
+
+ movq mm5, [rsi+rax*2]
+ movq [rdi], mm3
+
+ lea rsi, [rsi+rax*2]
+ movq [rdi+rcx], mm4
+
+ movq [rdi+rcx*2], mm5
+ lea rdi, [rdi+rcx*2]
+
+ movq mm0, [rsi+rax]
+ movq mm1, [rsi+rax*2]
+
+ movq [rdi+rcx], mm0
+ movq [rdi+rcx*2],mm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void copy_mem8x4_mmx(
+; unsigned char *src,
+; int src_stride,
+; unsigned char *dst,
+; int dst_stride
+; )
+global sym(vp8_copy_mem8x4_mmx) PRIVATE
+sym(vp8_copy_mem8x4_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src;
+ movq mm0, [rsi]
+
+ movsxd rax, dword ptr arg(1) ;src_stride;
+ mov rdi, arg(2) ;dst;
+
+ movq mm1, [rsi+rax]
+ movq mm2, [rsi+rax*2]
+
+ movsxd rcx, dword ptr arg(3) ;dst_stride
+ lea rsi, [rsi+rax*2]
+
+ movq [rdi], mm0
+ movq [rdi+rcx], mm1
+
+ movq [rdi+rcx*2], mm2
+ lea rdi, [rdi+rcx*2]
+
+ movq mm3, [rsi+rax]
+ movq [rdi+rcx], mm3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void copy_mem16x16_mmx(
+; unsigned char *src,
+; int src_stride,
+; unsigned char *dst,
+; int dst_stride
+; )
+global sym(vp8_copy_mem16x16_mmx) PRIVATE
+sym(vp8_copy_mem16x16_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src;
+ movsxd rax, dword ptr arg(1) ;src_stride;
+
+ mov rdi, arg(2) ;dst;
+ movsxd rcx, dword ptr arg(3) ;dst_stride
+
+ movq mm0, [rsi]
+ movq mm3, [rsi+8];
+
+ movq mm1, [rsi+rax]
+ movq mm4, [rsi+rax+8]
+
+ movq mm2, [rsi+rax*2]
+ movq mm5, [rsi+rax*2+8]
+
+ lea rsi, [rsi+rax*2]
+ add rsi, rax
+
+ movq [rdi], mm0
+ movq [rdi+8], mm3
+
+ movq [rdi+rcx], mm1
+ movq [rdi+rcx+8], mm4
+
+ movq [rdi+rcx*2], mm2
+ movq [rdi+rcx*2+8], mm5
+
+ lea rdi, [rdi+rcx*2]
+ add rdi, rcx
+
+ movq mm0, [rsi]
+ movq mm3, [rsi+8];
+
+ movq mm1, [rsi+rax]
+ movq mm4, [rsi+rax+8]
+
+ movq mm2, [rsi+rax*2]
+ movq mm5, [rsi+rax*2+8]
+
+ lea rsi, [rsi+rax*2]
+ add rsi, rax
+
+ movq [rdi], mm0
+ movq [rdi+8], mm3
+
+ movq [rdi+rcx], mm1
+ movq [rdi+rcx+8], mm4
+
+ movq [rdi+rcx*2], mm2
+ movq [rdi+rcx*2+8], mm5
+
+ lea rdi, [rdi+rcx*2]
+ add rdi, rcx
+
+ movq mm0, [rsi]
+ movq mm3, [rsi+8];
+
+ movq mm1, [rsi+rax]
+ movq mm4, [rsi+rax+8]
+
+ movq mm2, [rsi+rax*2]
+ movq mm5, [rsi+rax*2+8]
+
+ lea rsi, [rsi+rax*2]
+ add rsi, rax
+
+ movq [rdi], mm0
+ movq [rdi+8], mm3
+
+ movq [rdi+rcx], mm1
+ movq [rdi+rcx+8], mm4
+
+ movq [rdi+rcx*2], mm2
+ movq [rdi+rcx*2+8], mm5
+
+ lea rdi, [rdi+rcx*2]
+ add rdi, rcx
+
+ movq mm0, [rsi]
+ movq mm3, [rsi+8];
+
+ movq mm1, [rsi+rax]
+ movq mm4, [rsi+rax+8]
+
+ movq mm2, [rsi+rax*2]
+ movq mm5, [rsi+rax*2+8]
+
+ lea rsi, [rsi+rax*2]
+ add rsi, rax
+
+ movq [rdi], mm0
+ movq [rdi+8], mm3
+
+ movq [rdi+rcx], mm1
+ movq [rdi+rcx+8], mm4
+
+ movq [rdi+rcx*2], mm2
+ movq [rdi+rcx*2+8], mm5
+
+ lea rdi, [rdi+rcx*2]
+ add rdi, rcx
+
+ movq mm0, [rsi]
+ movq mm3, [rsi+8];
+
+ movq mm1, [rsi+rax]
+ movq mm4, [rsi+rax+8]
+
+ movq mm2, [rsi+rax*2]
+ movq mm5, [rsi+rax*2+8]
+
+ lea rsi, [rsi+rax*2]
+ add rsi, rax
+
+ movq [rdi], mm0
+ movq [rdi+8], mm3
+
+ movq [rdi+rcx], mm1
+ movq [rdi+rcx+8], mm4
+
+ movq [rdi+rcx*2], mm2
+ movq [rdi+rcx*2+8], mm5
+
+ lea rdi, [rdi+rcx*2]
+ add rdi, rcx
+
+ movq mm0, [rsi]
+ movq mm3, [rsi+8];
+
+ movq [rdi], mm0
+ movq [rdi+8], mm3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/common/x86/recon_sse2.asm b/libvpx/vp8/common/x86/recon_sse2.asm
new file mode 100644
index 0000000..fe77450
--- /dev/null
+++ b/libvpx/vp8/common/x86/recon_sse2.asm
@@ -0,0 +1,1080 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void copy_mem16x16_sse2(
+; unsigned char *src,
+; int src_stride,
+; unsigned char *dst,
+; int dst_stride
+; )
+global sym(vp8_copy_mem16x16_sse2) PRIVATE
+sym(vp8_copy_mem16x16_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src;
+ movdqu xmm0, [rsi]
+
+ movsxd rax, dword ptr arg(1) ;src_stride;
+ mov rdi, arg(2) ;dst;
+
+ movdqu xmm1, [rsi+rax]
+ movdqu xmm2, [rsi+rax*2]
+
+ movsxd rcx, dword ptr arg(3) ;dst_stride
+ lea rsi, [rsi+rax*2]
+
+ movdqa [rdi], xmm0
+ add rsi, rax
+
+ movdqa [rdi+rcx], xmm1
+ movdqa [rdi+rcx*2],xmm2
+
+ lea rdi, [rdi+rcx*2]
+ movdqu xmm3, [rsi]
+
+ add rdi, rcx
+ movdqu xmm4, [rsi+rax]
+
+ movdqu xmm5, [rsi+rax*2]
+ lea rsi, [rsi+rax*2]
+
+ movdqa [rdi], xmm3
+ add rsi, rax
+
+ movdqa [rdi+rcx], xmm4
+ movdqa [rdi+rcx*2],xmm5
+
+ lea rdi, [rdi+rcx*2]
+ movdqu xmm0, [rsi]
+
+ add rdi, rcx
+ movdqu xmm1, [rsi+rax]
+
+ movdqu xmm2, [rsi+rax*2]
+ lea rsi, [rsi+rax*2]
+
+ movdqa [rdi], xmm0
+ add rsi, rax
+
+ movdqa [rdi+rcx], xmm1
+
+ movdqa [rdi+rcx*2], xmm2
+ movdqu xmm3, [rsi]
+
+ movdqu xmm4, [rsi+rax]
+ lea rdi, [rdi+rcx*2]
+
+ add rdi, rcx
+ movdqu xmm5, [rsi+rax*2]
+
+ lea rsi, [rsi+rax*2]
+ movdqa [rdi], xmm3
+
+ add rsi, rax
+ movdqa [rdi+rcx], xmm4
+
+ movdqa [rdi+rcx*2],xmm5
+ movdqu xmm0, [rsi]
+
+ lea rdi, [rdi+rcx*2]
+ movdqu xmm1, [rsi+rax]
+
+ add rdi, rcx
+ movdqu xmm2, [rsi+rax*2]
+
+ lea rsi, [rsi+rax*2]
+ movdqa [rdi], xmm0
+
+ movdqa [rdi+rcx], xmm1
+ movdqa [rdi+rcx*2],xmm2
+
+ movdqu xmm3, [rsi+rax]
+ lea rdi, [rdi+rcx*2]
+
+ movdqa [rdi+rcx], xmm3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_intra_pred_uv_dc_mmx2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride,
+; )
+global sym(vp8_intra_pred_uv_dc_mmx2) PRIVATE
+sym(vp8_intra_pred_uv_dc_mmx2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ ; from top
+ mov rdi, arg(2) ;above;
+ mov rsi, arg(3) ;left;
+ movsxd rax, dword ptr arg(4) ;left_stride;
+ pxor mm0, mm0
+ movq mm1, [rdi]
+ lea rdi, [rax*3]
+ psadbw mm1, mm0
+ ; from left
+ movzx ecx, byte [rsi]
+ movzx edx, byte [rsi+rax*1]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+
+ movzx edx, byte [rsi+rdi]
+ lea rsi, [rsi+rax*4]
+ add ecx, edx
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+
+ ; add up
+ pextrw edx, mm1, 0x0
+ lea edx, [edx+ecx+8]
+ sar edx, 4
+ movd mm1, edx
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ pshufw mm1, mm1, 0x0
+ mov rdi, arg(0) ;dst;
+ packuswb mm1, mm1
+
+ ; write out
+ lea rax, [rcx*3]
+ lea rdx, [rdi+rcx*4]
+
+ movq [rdi ], mm1
+ movq [rdi+rcx ], mm1
+ movq [rdi+rcx*2], mm1
+ movq [rdi+rax ], mm1
+ movq [rdx ], mm1
+ movq [rdx+rcx ], mm1
+ movq [rdx+rcx*2], mm1
+ movq [rdx+rax ], mm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_uv_dctop_mmx2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride,
+; )
+global sym(vp8_intra_pred_uv_dctop_mmx2) PRIVATE
+sym(vp8_intra_pred_uv_dctop_mmx2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ;arg(3), arg(4) not used
+
+ ; from top
+ mov rsi, arg(2) ;above;
+ pxor mm0, mm0
+ movq mm1, [rsi]
+ psadbw mm1, mm0
+
+ ; add up
+ paddw mm1, [GLOBAL(dc_4)]
+ psraw mm1, 3
+ pshufw mm1, mm1, 0x0
+ packuswb mm1, mm1
+
+ ; write out
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ lea rax, [rcx*3]
+
+ movq [rdi ], mm1
+ movq [rdi+rcx ], mm1
+ movq [rdi+rcx*2], mm1
+ movq [rdi+rax ], mm1
+ lea rdi, [rdi+rcx*4]
+ movq [rdi ], mm1
+ movq [rdi+rcx ], mm1
+ movq [rdi+rcx*2], mm1
+ movq [rdi+rax ], mm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_uv_dcleft_mmx2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride,
+; )
+global sym(vp8_intra_pred_uv_dcleft_mmx2) PRIVATE
+sym(vp8_intra_pred_uv_dcleft_mmx2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ ;arg(2) not used
+
+ ; from left
+ mov rsi, arg(3) ;left;
+ movsxd rax, dword ptr arg(4) ;left_stride;
+ lea rdi, [rax*3]
+ movzx ecx, byte [rsi]
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ lea edx, [ecx+edx+4]
+
+ ; add up
+ shr edx, 3
+ movd mm1, edx
+ pshufw mm1, mm1, 0x0
+ packuswb mm1, mm1
+
+ ; write out
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ lea rax, [rcx*3]
+
+ movq [rdi ], mm1
+ movq [rdi+rcx ], mm1
+ movq [rdi+rcx*2], mm1
+ movq [rdi+rax ], mm1
+ lea rdi, [rdi+rcx*4]
+ movq [rdi ], mm1
+ movq [rdi+rcx ], mm1
+ movq [rdi+rcx*2], mm1
+ movq [rdi+rax ], mm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_uv_dc128_mmx(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride,
+; )
+global sym(vp8_intra_pred_uv_dc128_mmx) PRIVATE
+sym(vp8_intra_pred_uv_dc128_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ ; end prolog
+
+ ;arg(2), arg(3), arg(4) not used
+
+ ; write out
+ movq mm1, [GLOBAL(dc_128)]
+ mov rax, arg(0) ;dst;
+ movsxd rdx, dword ptr arg(1) ;dst_stride
+ lea rcx, [rdx*3]
+
+ movq [rax ], mm1
+ movq [rax+rdx ], mm1
+ movq [rax+rdx*2], mm1
+ movq [rax+rcx ], mm1
+ lea rax, [rax+rdx*4]
+ movq [rax ], mm1
+ movq [rax+rdx ], mm1
+ movq [rax+rdx*2], mm1
+ movq [rax+rcx ], mm1
+
+ ; begin epilog
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_uv_tm_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride,
+; )
+%macro vp8_intra_pred_uv_tm 1
+global sym(vp8_intra_pred_uv_tm_%1) PRIVATE
+sym(vp8_intra_pred_uv_tm_%1):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ; read top row
+ mov edx, 4
+ mov rsi, arg(2) ;above
+ movsxd rax, dword ptr arg(4) ;left_stride;
+ pxor xmm0, xmm0
+%ifidn %1, ssse3
+ movdqa xmm2, [GLOBAL(dc_1024)]
+%endif
+ movq xmm1, [rsi]
+ punpcklbw xmm1, xmm0
+
+ ; set up left ptrs ans subtract topleft
+ movd xmm3, [rsi-1]
+ mov rsi, arg(3) ;left;
+%ifidn %1, sse2
+ punpcklbw xmm3, xmm0
+ pshuflw xmm3, xmm3, 0x0
+ punpcklqdq xmm3, xmm3
+%else
+ pshufb xmm3, xmm2
+%endif
+ psubw xmm1, xmm3
+
+ ; set up dest ptrs
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+
+.vp8_intra_pred_uv_tm_%1_loop:
+ movd xmm3, [rsi]
+ movd xmm5, [rsi+rax]
+%ifidn %1, sse2
+ punpcklbw xmm3, xmm0
+ punpcklbw xmm5, xmm0
+ pshuflw xmm3, xmm3, 0x0
+ pshuflw xmm5, xmm5, 0x0
+ punpcklqdq xmm3, xmm3
+ punpcklqdq xmm5, xmm5
+%else
+ pshufb xmm3, xmm2
+ pshufb xmm5, xmm2
+%endif
+ paddw xmm3, xmm1
+ paddw xmm5, xmm1
+ packuswb xmm3, xmm5
+ movq [rdi ], xmm3
+ movhps[rdi+rcx], xmm3
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rcx*2]
+ dec edx
+ jnz .vp8_intra_pred_uv_tm_%1_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+%endmacro
+
+vp8_intra_pred_uv_tm sse2
+vp8_intra_pred_uv_tm ssse3
+
+;void vp8_intra_pred_uv_ve_mmx(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride,
+; )
+global sym(vp8_intra_pred_uv_ve_mmx) PRIVATE
+sym(vp8_intra_pred_uv_ve_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ ; end prolog
+
+ ; arg(3), arg(4) not used
+
+ ; read from top
+ mov rax, arg(2) ;src;
+
+ movq mm1, [rax]
+
+ ; write out
+ mov rax, arg(0) ;dst;
+ movsxd rdx, dword ptr arg(1) ;dst_stride
+ lea rcx, [rdx*3]
+
+ movq [rax ], mm1
+ movq [rax+rdx ], mm1
+ movq [rax+rdx*2], mm1
+ movq [rax+rcx ], mm1
+ lea rax, [rax+rdx*4]
+ movq [rax ], mm1
+ movq [rax+rdx ], mm1
+ movq [rax+rdx*2], mm1
+ movq [rax+rcx ], mm1
+
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_uv_ho_mmx2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride
+; )
+%macro vp8_intra_pred_uv_ho 1
+global sym(vp8_intra_pred_uv_ho_%1) PRIVATE
+sym(vp8_intra_pred_uv_ho_%1):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+%ifidn %1, ssse3
+%ifndef GET_GOT_SAVE_ARG
+ push rbx
+%endif
+ GET_GOT rbx
+%endif
+ ; end prolog
+
+ ;arg(2) not used
+
+ ; read from left and write out
+%ifidn %1, mmx2
+ mov edx, 4
+%endif
+ mov rsi, arg(3) ;left
+ movsxd rax, dword ptr arg(4) ;left_stride;
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+%ifidn %1, ssse3
+ lea rdx, [rcx*3]
+ movdqa xmm2, [GLOBAL(dc_00001111)]
+ lea rbx, [rax*3]
+%endif
+
+%ifidn %1, mmx2
+.vp8_intra_pred_uv_ho_%1_loop:
+ movd mm0, [rsi]
+ movd mm1, [rsi+rax]
+ punpcklbw mm0, mm0
+ punpcklbw mm1, mm1
+ pshufw mm0, mm0, 0x0
+ pshufw mm1, mm1, 0x0
+ movq [rdi ], mm0
+ movq [rdi+rcx], mm1
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rcx*2]
+ dec edx
+ jnz .vp8_intra_pred_uv_ho_%1_loop
+%else
+ movd xmm0, [rsi]
+ movd xmm3, [rsi+rax]
+ movd xmm1, [rsi+rax*2]
+ movd xmm4, [rsi+rbx]
+ punpcklbw xmm0, xmm3
+ punpcklbw xmm1, xmm4
+ pshufb xmm0, xmm2
+ pshufb xmm1, xmm2
+ movq [rdi ], xmm0
+ movhps [rdi+rcx], xmm0
+ movq [rdi+rcx*2], xmm1
+ movhps [rdi+rdx], xmm1
+ lea rsi, [rsi+rax*4]
+ lea rdi, [rdi+rcx*4]
+ movd xmm0, [rsi]
+ movd xmm3, [rsi+rax]
+ movd xmm1, [rsi+rax*2]
+ movd xmm4, [rsi+rbx]
+ punpcklbw xmm0, xmm3
+ punpcklbw xmm1, xmm4
+ pshufb xmm0, xmm2
+ pshufb xmm1, xmm2
+ movq [rdi ], xmm0
+ movhps [rdi+rcx], xmm0
+ movq [rdi+rcx*2], xmm1
+ movhps [rdi+rdx], xmm1
+%endif
+
+ ; begin epilog
+%ifidn %1, ssse3
+ RESTORE_GOT
+%ifndef GET_GOT_SAVE_ARG
+ pop rbx
+%endif
+%endif
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+%endmacro
+
+vp8_intra_pred_uv_ho mmx2
+vp8_intra_pred_uv_ho ssse3
+
+;void vp8_intra_pred_y_dc_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride
+; )
+global sym(vp8_intra_pred_y_dc_sse2) PRIVATE
+sym(vp8_intra_pred_y_dc_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ ; from top
+ mov rdi, arg(2) ;above
+ mov rsi, arg(3) ;left
+ movsxd rax, dword ptr arg(4) ;left_stride;
+
+ pxor xmm0, xmm0
+ movdqa xmm1, [rdi]
+ psadbw xmm1, xmm0
+ movq xmm2, xmm1
+ punpckhqdq xmm1, xmm1
+ paddw xmm1, xmm2
+
+ ; from left
+ lea rdi, [rax*3]
+
+ movzx ecx, byte [rsi]
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+
+ ; add up
+ pextrw edx, xmm1, 0x0
+ lea edx, [edx+ecx+16]
+ sar edx, 5
+ movd xmm1, edx
+ ; FIXME use pshufb for ssse3 version
+ pshuflw xmm1, xmm1, 0x0
+ punpcklqdq xmm1, xmm1
+ packuswb xmm1, xmm1
+
+ ; write out
+ mov rsi, 2
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ lea rax, [rcx*3]
+
+.label
+ movdqa [rdi ], xmm1
+ movdqa [rdi+rcx ], xmm1
+ movdqa [rdi+rcx*2], xmm1
+ movdqa [rdi+rax ], xmm1
+ lea rdi, [rdi+rcx*4]
+ movdqa [rdi ], xmm1
+ movdqa [rdi+rcx ], xmm1
+ movdqa [rdi+rcx*2], xmm1
+ movdqa [rdi+rax ], xmm1
+ lea rdi, [rdi+rcx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_dctop_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride
+; )
+global sym(vp8_intra_pred_y_dctop_sse2) PRIVATE
+sym(vp8_intra_pred_y_dctop_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ GET_GOT rbx
+ ; end prolog
+
+ ;arg(3), arg(4) not used
+
+ ; from top
+ mov rcx, arg(2) ;above;
+ pxor xmm0, xmm0
+ movdqa xmm1, [rcx]
+ psadbw xmm1, xmm0
+ movdqa xmm2, xmm1
+ punpckhqdq xmm1, xmm1
+ paddw xmm1, xmm2
+
+ ; add up
+ paddw xmm1, [GLOBAL(dc_8)]
+ psraw xmm1, 4
+ ; FIXME use pshufb for ssse3 version
+ pshuflw xmm1, xmm1, 0x0
+ punpcklqdq xmm1, xmm1
+ packuswb xmm1, xmm1
+
+ ; write out
+ mov rsi, 2
+ mov rdx, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ lea rax, [rcx*3]
+
+.label
+ movdqa [rdx ], xmm1
+ movdqa [rdx+rcx ], xmm1
+ movdqa [rdx+rcx*2], xmm1
+ movdqa [rdx+rax ], xmm1
+ lea rdx, [rdx+rcx*4]
+ movdqa [rdx ], xmm1
+ movdqa [rdx+rcx ], xmm1
+ movdqa [rdx+rcx*2], xmm1
+ movdqa [rdx+rax ], xmm1
+ lea rdx, [rdx+rcx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ RESTORE_GOT
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_dcleft_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride
+; )
+global sym(vp8_intra_pred_y_dcleft_sse2) PRIVATE
+sym(vp8_intra_pred_y_dcleft_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ ;arg(2) not used
+
+ ; from left
+ mov rsi, arg(3) ;left;
+ movsxd rax, dword ptr arg(4) ;left_stride;
+
+ lea rdi, [rax*3]
+ movzx ecx, byte [rsi]
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ lea edx, [ecx+edx+8]
+
+ ; add up
+ shr edx, 4
+ movd xmm1, edx
+ ; FIXME use pshufb for ssse3 version
+ pshuflw xmm1, xmm1, 0x0
+ punpcklqdq xmm1, xmm1
+ packuswb xmm1, xmm1
+
+ ; write out
+ mov rsi, 2
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ lea rax, [rcx*3]
+
+.label
+ movdqa [rdi ], xmm1
+ movdqa [rdi+rcx ], xmm1
+ movdqa [rdi+rcx*2], xmm1
+ movdqa [rdi+rax ], xmm1
+ lea rdi, [rdi+rcx*4]
+ movdqa [rdi ], xmm1
+ movdqa [rdi+rcx ], xmm1
+ movdqa [rdi+rcx*2], xmm1
+ movdqa [rdi+rax ], xmm1
+ lea rdi, [rdi+rcx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_dc128_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride
+; )
+global sym(vp8_intra_pred_y_dc128_sse2) PRIVATE
+sym(vp8_intra_pred_y_dc128_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ GET_GOT rbx
+ ; end prolog
+
+ ;arg(2), arg(3), arg(4) not used
+
+ ; write out
+ mov rsi, 2
+ movdqa xmm1, [GLOBAL(dc_128)]
+ mov rax, arg(0) ;dst;
+ movsxd rdx, dword ptr arg(1) ;dst_stride
+ lea rcx, [rdx*3]
+
+.label
+ movdqa [rax ], xmm1
+ movdqa [rax+rdx ], xmm1
+ movdqa [rax+rdx*2], xmm1
+ movdqa [rax+rcx ], xmm1
+ lea rax, [rax+rdx*4]
+ movdqa [rax ], xmm1
+ movdqa [rax+rdx ], xmm1
+ movdqa [rax+rdx*2], xmm1
+ movdqa [rax+rcx ], xmm1
+ lea rax, [rax+rdx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ RESTORE_GOT
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_tm_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride
+; )
+%macro vp8_intra_pred_y_tm 1
+global sym(vp8_intra_pred_y_tm_%1) PRIVATE
+sym(vp8_intra_pred_y_tm_%1):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ GET_GOT rbx
+ ; end prolog
+
+ ; read top row
+ mov edx, 8
+ mov rsi, arg(2) ;above
+ movsxd rax, dword ptr arg(4) ;left_stride;
+ pxor xmm0, xmm0
+%ifidn %1, ssse3
+ movdqa xmm3, [GLOBAL(dc_1024)]
+%endif
+ movdqa xmm1, [rsi]
+ movdqa xmm2, xmm1
+ punpcklbw xmm1, xmm0
+ punpckhbw xmm2, xmm0
+
+ ; set up left ptrs ans subtract topleft
+ movd xmm4, [rsi-1]
+ mov rsi, arg(3) ;left
+%ifidn %1, sse2
+ punpcklbw xmm4, xmm0
+ pshuflw xmm4, xmm4, 0x0
+ punpcklqdq xmm4, xmm4
+%else
+ pshufb xmm4, xmm3
+%endif
+ psubw xmm1, xmm4
+ psubw xmm2, xmm4
+
+ ; set up dest ptrs
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+vp8_intra_pred_y_tm_%1_loop:
+ movd xmm4, [rsi]
+ movd xmm5, [rsi+rax]
+%ifidn %1, sse2
+ punpcklbw xmm4, xmm0
+ punpcklbw xmm5, xmm0
+ pshuflw xmm4, xmm4, 0x0
+ pshuflw xmm5, xmm5, 0x0
+ punpcklqdq xmm4, xmm4
+ punpcklqdq xmm5, xmm5
+%else
+ pshufb xmm4, xmm3
+ pshufb xmm5, xmm3
+%endif
+ movdqa xmm6, xmm4
+ movdqa xmm7, xmm5
+ paddw xmm4, xmm1
+ paddw xmm6, xmm2
+ paddw xmm5, xmm1
+ paddw xmm7, xmm2
+ packuswb xmm4, xmm6
+ packuswb xmm5, xmm7
+ movdqa [rdi ], xmm4
+ movdqa [rdi+rcx], xmm5
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rcx*2]
+ dec edx
+ jnz vp8_intra_pred_y_tm_%1_loop
+
+ ; begin epilog
+ RESTORE_GOT
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+%endmacro
+
+vp8_intra_pred_y_tm sse2
+vp8_intra_pred_y_tm ssse3
+
+;void vp8_intra_pred_y_ve_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride
+; )
+global sym(vp8_intra_pred_y_ve_sse2) PRIVATE
+sym(vp8_intra_pred_y_ve_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ ; end prolog
+
+ ;arg(3), arg(4) not used
+
+ mov rax, arg(2) ;above;
+ mov rsi, 2
+ movsxd rdx, dword ptr arg(1) ;dst_stride
+
+ ; read from top
+ movdqa xmm1, [rax]
+
+ ; write out
+ mov rax, arg(0) ;dst;
+ lea rcx, [rdx*3]
+
+.label
+ movdqa [rax ], xmm1
+ movdqa [rax+rdx ], xmm1
+ movdqa [rax+rdx*2], xmm1
+ movdqa [rax+rcx ], xmm1
+ lea rax, [rax+rdx*4]
+ movdqa [rax ], xmm1
+ movdqa [rax+rdx ], xmm1
+ movdqa [rax+rdx*2], xmm1
+ movdqa [rax+rcx ], xmm1
+ lea rax, [rax+rdx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_ho_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *above,
+; unsigned char *left,
+; int left_stride,
+; )
+global sym(vp8_intra_pred_y_ho_sse2) PRIVATE
+sym(vp8_intra_pred_y_ho_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ ;arg(2) not used
+
+ ; read from left and write out
+ mov edx, 8
+ mov rsi, arg(3) ;left;
+ movsxd rax, dword ptr arg(4) ;left_stride;
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+
+vp8_intra_pred_y_ho_sse2_loop:
+ movd xmm0, [rsi]
+ movd xmm1, [rsi+rax]
+ ; FIXME use pshufb for ssse3 version
+ punpcklbw xmm0, xmm0
+ punpcklbw xmm1, xmm1
+ pshuflw xmm0, xmm0, 0x0
+ pshuflw xmm1, xmm1, 0x0
+ punpcklqdq xmm0, xmm0
+ punpcklqdq xmm1, xmm1
+ movdqa [rdi ], xmm0
+ movdqa [rdi+rcx], xmm1
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rcx*2]
+ dec edx
+ jnz vp8_intra_pred_y_ho_sse2_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+dc_128:
+ times 16 db 128
+dc_4:
+ times 4 dw 4
+align 16
+dc_8:
+ times 8 dw 8
+align 16
+dc_1024:
+ times 8 dw 0x400
+align 16
+dc_00001111:
+ times 8 db 0
+ times 8 db 1
diff --git a/libvpx/vp8/common/x86/recon_wrapper_sse2.c b/libvpx/vp8/common/x86/recon_wrapper_sse2.c
new file mode 100644
index 0000000..b482faa
--- /dev/null
+++ b/libvpx/vp8/common/x86/recon_wrapper_sse2.c
@@ -0,0 +1,186 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/blockd.h"
+
+#define build_intra_predictors_mbuv_prototype(sym) \
+ void sym(unsigned char *dst, int dst_stride, \
+ const unsigned char *above, \
+ const unsigned char *left, int left_stride)
+typedef build_intra_predictors_mbuv_prototype((*build_intra_predictors_mbuv_fn_t));
+
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_dc_mmx2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_dctop_mmx2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_dcleft_mmx2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_dc128_mmx);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_ho_mmx2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_ho_ssse3);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_ve_mmx);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_tm_sse2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_uv_tm_ssse3);
+
+static void vp8_build_intra_predictors_mbuv_x86(MACROBLOCKD *x,
+ unsigned char * uabove_row,
+ unsigned char * vabove_row,
+ unsigned char *dst_u,
+ unsigned char *dst_v,
+ int dst_stride,
+ unsigned char * uleft,
+ unsigned char * vleft,
+ int left_stride,
+ build_intra_predictors_mbuv_fn_t tm_func,
+ build_intra_predictors_mbuv_fn_t ho_func)
+{
+ int mode = x->mode_info_context->mbmi.uv_mode;
+ build_intra_predictors_mbuv_fn_t fn;
+
+ switch (mode) {
+ case V_PRED: fn = vp8_intra_pred_uv_ve_mmx; break;
+ case H_PRED: fn = ho_func; break;
+ case TM_PRED: fn = tm_func; break;
+ case DC_PRED:
+ if (x->up_available) {
+ if (x->left_available) {
+ fn = vp8_intra_pred_uv_dc_mmx2; break;
+ } else {
+ fn = vp8_intra_pred_uv_dctop_mmx2; break;
+ }
+ } else if (x->left_available) {
+ fn = vp8_intra_pred_uv_dcleft_mmx2; break;
+ } else {
+ fn = vp8_intra_pred_uv_dc128_mmx; break;
+ }
+ break;
+ default: return;
+ }
+
+ fn(dst_u, dst_stride, uabove_row, uleft, left_stride);
+ fn(dst_v, dst_stride, vabove_row, vleft, left_stride);
+}
+
+void vp8_build_intra_predictors_mbuv_s_sse2(MACROBLOCKD *x,
+ unsigned char * uabove_row,
+ unsigned char * vabove_row,
+ unsigned char * uleft,
+ unsigned char * vleft,
+ int left_stride,
+ unsigned char * upred_ptr,
+ unsigned char * vpred_ptr,
+ int pred_stride)
+{
+ vp8_build_intra_predictors_mbuv_x86(x,
+ uabove_row, vabove_row,
+ upred_ptr,
+ vpred_ptr, pred_stride,
+ uleft,
+ vleft,
+ left_stride,
+ vp8_intra_pred_uv_tm_sse2,
+ vp8_intra_pred_uv_ho_mmx2);
+}
+
+void vp8_build_intra_predictors_mbuv_s_ssse3(MACROBLOCKD *x,
+ unsigned char * uabove_row,
+ unsigned char * vabove_row,
+ unsigned char * uleft,
+ unsigned char * vleft,
+ int left_stride,
+ unsigned char * upred_ptr,
+ unsigned char * vpred_ptr,
+ int pred_stride)
+{
+ vp8_build_intra_predictors_mbuv_x86(x,
+ uabove_row, vabove_row,
+ upred_ptr,
+ vpred_ptr, pred_stride,
+ uleft,
+ vleft,
+ left_stride,
+ vp8_intra_pred_uv_tm_ssse3,
+ vp8_intra_pred_uv_ho_ssse3);
+}
+
+#define build_intra_predictors_mby_prototype(sym) \
+ void sym(unsigned char *dst, int dst_stride, \
+ const unsigned char *above, \
+ const unsigned char *left, int left_stride)
+typedef build_intra_predictors_mby_prototype((*build_intra_predictors_mby_fn_t));
+
+extern build_intra_predictors_mby_prototype(vp8_intra_pred_y_dc_sse2);
+extern build_intra_predictors_mby_prototype(vp8_intra_pred_y_dctop_sse2);
+extern build_intra_predictors_mby_prototype(vp8_intra_pred_y_dcleft_sse2);
+extern build_intra_predictors_mby_prototype(vp8_intra_pred_y_dc128_sse2);
+extern build_intra_predictors_mby_prototype(vp8_intra_pred_y_ho_sse2);
+extern build_intra_predictors_mby_prototype(vp8_intra_pred_y_ve_sse2);
+extern build_intra_predictors_mby_prototype(vp8_intra_pred_y_tm_sse2);
+extern build_intra_predictors_mby_prototype(vp8_intra_pred_y_tm_ssse3);
+
+static void vp8_build_intra_predictors_mby_x86(MACROBLOCKD *x,
+ unsigned char * yabove_row,
+ unsigned char *dst_y,
+ int dst_stride,
+ unsigned char * yleft,
+ int left_stride,
+ build_intra_predictors_mby_fn_t tm_func)
+{
+ int mode = x->mode_info_context->mbmi.mode;
+ build_intra_predictors_mbuv_fn_t fn;
+
+ switch (mode) {
+ case V_PRED: fn = vp8_intra_pred_y_ve_sse2; break;
+ case H_PRED: fn = vp8_intra_pred_y_ho_sse2; break;
+ case TM_PRED: fn = tm_func; break;
+ case DC_PRED:
+ if (x->up_available) {
+ if (x->left_available) {
+ fn = vp8_intra_pred_y_dc_sse2; break;
+ } else {
+ fn = vp8_intra_pred_y_dctop_sse2; break;
+ }
+ } else if (x->left_available) {
+ fn = vp8_intra_pred_y_dcleft_sse2; break;
+ } else {
+ fn = vp8_intra_pred_y_dc128_sse2; break;
+ }
+ break;
+ default: return;
+ }
+
+ fn(dst_y, dst_stride, yabove_row, yleft, left_stride);
+ return;
+}
+
+void vp8_build_intra_predictors_mby_s_sse2(MACROBLOCKD *x,
+ unsigned char * yabove_row,
+ unsigned char * yleft,
+ int left_stride,
+ unsigned char * ypred_ptr,
+ int y_stride)
+{
+ vp8_build_intra_predictors_mby_x86(x, yabove_row, ypred_ptr,
+ y_stride, yleft, left_stride,
+ vp8_intra_pred_y_tm_sse2);
+}
+
+void vp8_build_intra_predictors_mby_s_ssse3(MACROBLOCKD *x,
+ unsigned char * yabove_row,
+ unsigned char * yleft,
+ int left_stride,
+ unsigned char * ypred_ptr,
+ int y_stride)
+{
+ vp8_build_intra_predictors_mby_x86(x, yabove_row, ypred_ptr,
+ y_stride, yleft, left_stride,
+ vp8_intra_pred_y_tm_ssse3);
+
+}
diff --git a/libvpx/vp8/common/x86/sad_mmx.asm b/libvpx/vp8/common/x86/sad_mmx.asm
new file mode 100644
index 0000000..592112f
--- /dev/null
+++ b/libvpx/vp8/common/x86/sad_mmx.asm
@@ -0,0 +1,427 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+global sym(vp8_sad16x16_mmx) PRIVATE
+global sym(vp8_sad8x16_mmx) PRIVATE
+global sym(vp8_sad8x8_mmx) PRIVATE
+global sym(vp8_sad4x4_mmx) PRIVATE
+global sym(vp8_sad16x8_mmx) PRIVATE
+
+;unsigned int vp8_sad16x16_mmx(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+sym(vp8_sad16x16_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ lea rcx, [rsi+rax*8]
+
+ lea rcx, [rcx+rax*8]
+ pxor mm7, mm7
+
+ pxor mm6, mm6
+
+.x16x16sad_mmx_loop:
+
+ movq mm0, QWORD PTR [rsi]
+ movq mm2, QWORD PTR [rsi+8]
+
+ movq mm1, QWORD PTR [rdi]
+ movq mm3, QWORD PTR [rdi+8]
+
+ movq mm4, mm0
+ movq mm5, mm2
+
+ psubusb mm0, mm1
+ psubusb mm1, mm4
+
+ psubusb mm2, mm3
+ psubusb mm3, mm5
+
+ por mm0, mm1
+ por mm2, mm3
+
+ movq mm1, mm0
+ movq mm3, mm2
+
+ punpcklbw mm0, mm6
+ punpcklbw mm2, mm6
+
+ punpckhbw mm1, mm6
+ punpckhbw mm3, mm6
+
+ paddw mm0, mm2
+ paddw mm1, mm3
+
+
+ lea rsi, [rsi+rax]
+ add rdi, rdx
+
+ paddw mm7, mm0
+ paddw mm7, mm1
+
+ cmp rsi, rcx
+ jne .x16x16sad_mmx_loop
+
+
+ movq mm0, mm7
+
+ punpcklwd mm0, mm6
+ punpckhwd mm7, mm6
+
+ paddw mm0, mm7
+ movq mm7, mm0
+
+
+ psrlq mm0, 32
+ paddw mm7, mm0
+
+ movq rax, mm7
+
+ pop rdi
+ pop rsi
+ mov rsp, rbp
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;unsigned int vp8_sad8x16_mmx(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+sym(vp8_sad8x16_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ lea rcx, [rsi+rax*8]
+
+ lea rcx, [rcx+rax*8]
+ pxor mm7, mm7
+
+ pxor mm6, mm6
+
+.x8x16sad_mmx_loop:
+
+ movq mm0, QWORD PTR [rsi]
+ movq mm1, QWORD PTR [rdi]
+
+ movq mm2, mm0
+ psubusb mm0, mm1
+
+ psubusb mm1, mm2
+ por mm0, mm1
+
+ movq mm2, mm0
+ punpcklbw mm0, mm6
+
+ punpckhbw mm2, mm6
+ lea rsi, [rsi+rax]
+
+ add rdi, rdx
+ paddw mm7, mm0
+
+ paddw mm7, mm2
+ cmp rsi, rcx
+
+ jne .x8x16sad_mmx_loop
+
+ movq mm0, mm7
+ punpcklwd mm0, mm6
+
+ punpckhwd mm7, mm6
+ paddw mm0, mm7
+
+ movq mm7, mm0
+ psrlq mm0, 32
+
+ paddw mm7, mm0
+ movq rax, mm7
+
+ pop rdi
+ pop rsi
+ mov rsp, rbp
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;unsigned int vp8_sad8x8_mmx(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+sym(vp8_sad8x8_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ lea rcx, [rsi+rax*8]
+ pxor mm7, mm7
+
+ pxor mm6, mm6
+
+.x8x8sad_mmx_loop:
+
+ movq mm0, QWORD PTR [rsi]
+ movq mm1, QWORD PTR [rdi]
+
+ movq mm2, mm0
+ psubusb mm0, mm1
+
+ psubusb mm1, mm2
+ por mm0, mm1
+
+ movq mm2, mm0
+ punpcklbw mm0, mm6
+
+ punpckhbw mm2, mm6
+ paddw mm0, mm2
+
+ lea rsi, [rsi+rax]
+ add rdi, rdx
+
+ paddw mm7, mm0
+ cmp rsi, rcx
+
+ jne .x8x8sad_mmx_loop
+
+ movq mm0, mm7
+ punpcklwd mm0, mm6
+
+ punpckhwd mm7, mm6
+ paddw mm0, mm7
+
+ movq mm7, mm0
+ psrlq mm0, 32
+
+ paddw mm7, mm0
+ movq rax, mm7
+
+ pop rdi
+ pop rsi
+ mov rsp, rbp
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;unsigned int vp8_sad4x4_mmx(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+sym(vp8_sad4x4_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ movd mm0, DWORD PTR [rsi]
+ movd mm1, DWORD PTR [rdi]
+
+ movd mm2, DWORD PTR [rsi+rax]
+ movd mm3, DWORD PTR [rdi+rdx]
+
+ punpcklbw mm0, mm2
+ punpcklbw mm1, mm3
+
+ movq mm2, mm0
+ psubusb mm0, mm1
+
+ psubusb mm1, mm2
+ por mm0, mm1
+
+ movq mm2, mm0
+ pxor mm3, mm3
+
+ punpcklbw mm0, mm3
+ punpckhbw mm2, mm3
+
+ paddw mm0, mm2
+
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rdx*2]
+
+ movd mm4, DWORD PTR [rsi]
+ movd mm5, DWORD PTR [rdi]
+
+ movd mm6, DWORD PTR [rsi+rax]
+ movd mm7, DWORD PTR [rdi+rdx]
+
+ punpcklbw mm4, mm6
+ punpcklbw mm5, mm7
+
+ movq mm6, mm4
+ psubusb mm4, mm5
+
+ psubusb mm5, mm6
+ por mm4, mm5
+
+ movq mm5, mm4
+ punpcklbw mm4, mm3
+
+ punpckhbw mm5, mm3
+ paddw mm4, mm5
+
+ paddw mm0, mm4
+ movq mm1, mm0
+
+ punpcklwd mm0, mm3
+ punpckhwd mm1, mm3
+
+ paddw mm0, mm1
+ movq mm1, mm0
+
+ psrlq mm0, 32
+ paddw mm0, mm1
+
+ movq rax, mm0
+
+ pop rdi
+ pop rsi
+ mov rsp, rbp
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;unsigned int vp8_sad16x8_mmx(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+sym(vp8_sad16x8_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ lea rcx, [rsi+rax*8]
+ pxor mm7, mm7
+
+ pxor mm6, mm6
+
+.x16x8sad_mmx_loop:
+
+ movq mm0, [rsi]
+ movq mm1, [rdi]
+
+ movq mm2, [rsi+8]
+ movq mm3, [rdi+8]
+
+ movq mm4, mm0
+ movq mm5, mm2
+
+ psubusb mm0, mm1
+ psubusb mm1, mm4
+
+ psubusb mm2, mm3
+ psubusb mm3, mm5
+
+ por mm0, mm1
+ por mm2, mm3
+
+ movq mm1, mm0
+ movq mm3, mm2
+
+ punpcklbw mm0, mm6
+ punpckhbw mm1, mm6
+
+ punpcklbw mm2, mm6
+ punpckhbw mm3, mm6
+
+
+ paddw mm0, mm2
+ paddw mm1, mm3
+
+ paddw mm0, mm1
+ lea rsi, [rsi+rax]
+
+ add rdi, rdx
+ paddw mm7, mm0
+
+ cmp rsi, rcx
+ jne .x16x8sad_mmx_loop
+
+ movq mm0, mm7
+ punpcklwd mm0, mm6
+
+ punpckhwd mm7, mm6
+ paddw mm0, mm7
+
+ movq mm7, mm0
+ psrlq mm0, 32
+
+ paddw mm7, mm0
+ movq rax, mm7
+
+ pop rdi
+ pop rsi
+ mov rsp, rbp
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/common/x86/sad_sse2.asm b/libvpx/vp8/common/x86/sad_sse2.asm
new file mode 100644
index 0000000..8d86abc
--- /dev/null
+++ b/libvpx/vp8/common/x86/sad_sse2.asm
@@ -0,0 +1,410 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;unsigned int vp8_sad16x16_wmt(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+global sym(vp8_sad16x16_wmt) PRIVATE
+sym(vp8_sad16x16_wmt):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ SAVE_XMM 6
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ lea rcx, [rsi+rax*8]
+
+ lea rcx, [rcx+rax*8]
+ pxor xmm6, xmm6
+
+.x16x16sad_wmt_loop:
+
+ movq xmm0, QWORD PTR [rsi]
+ movq xmm2, QWORD PTR [rsi+8]
+
+ movq xmm1, QWORD PTR [rdi]
+ movq xmm3, QWORD PTR [rdi+8]
+
+ movq xmm4, QWORD PTR [rsi+rax]
+ movq xmm5, QWORD PTR [rdi+rdx]
+
+
+ punpcklbw xmm0, xmm2
+ punpcklbw xmm1, xmm3
+
+ psadbw xmm0, xmm1
+ movq xmm2, QWORD PTR [rsi+rax+8]
+
+ movq xmm3, QWORD PTR [rdi+rdx+8]
+ lea rsi, [rsi+rax*2]
+
+ lea rdi, [rdi+rdx*2]
+ punpcklbw xmm4, xmm2
+
+ punpcklbw xmm5, xmm3
+ psadbw xmm4, xmm5
+
+ paddw xmm6, xmm0
+ paddw xmm6, xmm4
+
+ cmp rsi, rcx
+ jne .x16x16sad_wmt_loop
+
+ movq xmm0, xmm6
+ psrldq xmm6, 8
+
+ paddw xmm0, xmm6
+ movq rax, xmm0
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;unsigned int vp8_sad8x16_wmt(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int max_sad)
+global sym(vp8_sad8x16_wmt) PRIVATE
+sym(vp8_sad8x16_wmt):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rbx, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ lea rcx, [rsi+rbx*8]
+
+ lea rcx, [rcx+rbx*8]
+ pxor mm7, mm7
+
+.x8x16sad_wmt_loop:
+
+ movq rax, mm7
+ cmp eax, arg(4)
+ ja .x8x16sad_wmt_early_exit
+
+ movq mm0, QWORD PTR [rsi]
+ movq mm1, QWORD PTR [rdi]
+
+ movq mm2, QWORD PTR [rsi+rbx]
+ movq mm3, QWORD PTR [rdi+rdx]
+
+ psadbw mm0, mm1
+ psadbw mm2, mm3
+
+ lea rsi, [rsi+rbx*2]
+ lea rdi, [rdi+rdx*2]
+
+ paddw mm7, mm0
+ paddw mm7, mm2
+
+ cmp rsi, rcx
+ jne .x8x16sad_wmt_loop
+
+ movq rax, mm7
+
+.x8x16sad_wmt_early_exit:
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ pop rbx
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;unsigned int vp8_sad8x8_wmt(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+global sym(vp8_sad8x8_wmt) PRIVATE
+sym(vp8_sad8x8_wmt):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rbx, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ lea rcx, [rsi+rbx*8]
+ pxor mm7, mm7
+
+.x8x8sad_wmt_loop:
+
+ movq rax, mm7
+ cmp eax, arg(4)
+ ja .x8x8sad_wmt_early_exit
+
+ movq mm0, QWORD PTR [rsi]
+ movq mm1, QWORD PTR [rdi]
+
+ psadbw mm0, mm1
+ lea rsi, [rsi+rbx]
+
+ add rdi, rdx
+ paddw mm7, mm0
+
+ cmp rsi, rcx
+ jne .x8x8sad_wmt_loop
+
+ movq rax, mm7
+.x8x8sad_wmt_early_exit:
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ pop rbx
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;unsigned int vp8_sad4x4_wmt(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+global sym(vp8_sad4x4_wmt) PRIVATE
+sym(vp8_sad4x4_wmt):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ movd mm0, DWORD PTR [rsi]
+ movd mm1, DWORD PTR [rdi]
+
+ movd mm2, DWORD PTR [rsi+rax]
+ movd mm3, DWORD PTR [rdi+rdx]
+
+ punpcklbw mm0, mm2
+ punpcklbw mm1, mm3
+
+ psadbw mm0, mm1
+ lea rsi, [rsi+rax*2]
+
+ lea rdi, [rdi+rdx*2]
+ movd mm4, DWORD PTR [rsi]
+
+ movd mm5, DWORD PTR [rdi]
+ movd mm6, DWORD PTR [rsi+rax]
+
+ movd mm7, DWORD PTR [rdi+rdx]
+ punpcklbw mm4, mm6
+
+ punpcklbw mm5, mm7
+ psadbw mm4, mm5
+
+ paddw mm0, mm4
+ movq rax, mm0
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;unsigned int vp8_sad16x8_wmt(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride)
+global sym(vp8_sad16x8_wmt) PRIVATE
+sym(vp8_sad16x8_wmt):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rbx, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ lea rcx, [rsi+rbx*8]
+ pxor mm7, mm7
+
+.x16x8sad_wmt_loop:
+
+ movq rax, mm7
+ cmp eax, arg(4)
+ ja .x16x8sad_wmt_early_exit
+
+ movq mm0, QWORD PTR [rsi]
+ movq mm2, QWORD PTR [rsi+8]
+
+ movq mm1, QWORD PTR [rdi]
+ movq mm3, QWORD PTR [rdi+8]
+
+ movq mm4, QWORD PTR [rsi+rbx]
+ movq mm5, QWORD PTR [rdi+rdx]
+
+ psadbw mm0, mm1
+ psadbw mm2, mm3
+
+ movq mm1, QWORD PTR [rsi+rbx+8]
+ movq mm3, QWORD PTR [rdi+rdx+8]
+
+ psadbw mm4, mm5
+ psadbw mm1, mm3
+
+ lea rsi, [rsi+rbx*2]
+ lea rdi, [rdi+rdx*2]
+
+ paddw mm0, mm2
+ paddw mm4, mm1
+
+ paddw mm7, mm0
+ paddw mm7, mm4
+
+ cmp rsi, rcx
+ jne .x16x8sad_wmt_loop
+
+ movq rax, mm7
+
+.x16x8sad_wmt_early_exit:
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ pop rbx
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_copy32xn_sse2(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *dst_ptr,
+; int dst_stride,
+; int height);
+global sym(vp8_copy32xn_sse2) PRIVATE
+sym(vp8_copy32xn_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;dst_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;dst_stride
+ movsxd rcx, dword ptr arg(4) ;height
+
+.block_copy_sse2_loopx4:
+ movdqu xmm0, XMMWORD PTR [rsi]
+ movdqu xmm1, XMMWORD PTR [rsi + 16]
+ movdqu xmm2, XMMWORD PTR [rsi + rax]
+ movdqu xmm3, XMMWORD PTR [rsi + rax + 16]
+
+ lea rsi, [rsi+rax*2]
+
+ movdqu xmm4, XMMWORD PTR [rsi]
+ movdqu xmm5, XMMWORD PTR [rsi + 16]
+ movdqu xmm6, XMMWORD PTR [rsi + rax]
+ movdqu xmm7, XMMWORD PTR [rsi + rax + 16]
+
+ lea rsi, [rsi+rax*2]
+
+ movdqa XMMWORD PTR [rdi], xmm0
+ movdqa XMMWORD PTR [rdi + 16], xmm1
+ movdqa XMMWORD PTR [rdi + rdx], xmm2
+ movdqa XMMWORD PTR [rdi + rdx + 16], xmm3
+
+ lea rdi, [rdi+rdx*2]
+
+ movdqa XMMWORD PTR [rdi], xmm4
+ movdqa XMMWORD PTR [rdi + 16], xmm5
+ movdqa XMMWORD PTR [rdi + rdx], xmm6
+ movdqa XMMWORD PTR [rdi + rdx + 16], xmm7
+
+ lea rdi, [rdi+rdx*2]
+
+ sub rcx, 4
+ cmp rcx, 4
+ jge .block_copy_sse2_loopx4
+
+ cmp rcx, 0
+ je .copy_is_done
+
+.block_copy_sse2_loop:
+ movdqu xmm0, XMMWORD PTR [rsi]
+ movdqu xmm1, XMMWORD PTR [rsi + 16]
+ lea rsi, [rsi+rax]
+
+ movdqa XMMWORD PTR [rdi], xmm0
+ movdqa XMMWORD PTR [rdi + 16], xmm1
+ lea rdi, [rdi+rdx]
+
+ sub rcx, 1
+ jne .block_copy_sse2_loop
+
+.copy_is_done:
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/common/x86/sad_sse3.asm b/libvpx/vp8/common/x86/sad_sse3.asm
new file mode 100644
index 0000000..f90a589
--- /dev/null
+++ b/libvpx/vp8/common/x86/sad_sse3.asm
@@ -0,0 +1,960 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%macro STACK_FRAME_CREATE_X3 0
+%if ABI_IS_32BIT
+ %define src_ptr rsi
+ %define src_stride rax
+ %define ref_ptr rdi
+ %define ref_stride rdx
+ %define end_ptr rcx
+ %define ret_var rbx
+ %define result_ptr arg(4)
+ %define max_sad arg(4)
+ %define height dword ptr arg(4)
+ push rbp
+ mov rbp, rsp
+ push rsi
+ push rdi
+ push rbx
+
+ mov rsi, arg(0) ; src_ptr
+ mov rdi, arg(2) ; ref_ptr
+
+ movsxd rax, dword ptr arg(1) ; src_stride
+ movsxd rdx, dword ptr arg(3) ; ref_stride
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ SAVE_XMM 7, u
+ %define src_ptr rcx
+ %define src_stride rdx
+ %define ref_ptr r8
+ %define ref_stride r9
+ %define end_ptr r10
+ %define ret_var r11
+ %define result_ptr [rsp+xmm_stack_space+8+4*8]
+ %define max_sad [rsp+xmm_stack_space+8+4*8]
+ %define height dword ptr [rsp+xmm_stack_space+8+4*8]
+ %else
+ %define src_ptr rdi
+ %define src_stride rsi
+ %define ref_ptr rdx
+ %define ref_stride rcx
+ %define end_ptr r9
+ %define ret_var r10
+ %define result_ptr r8
+ %define max_sad r8
+ %define height r8
+ %endif
+%endif
+
+%endmacro
+
+%macro STACK_FRAME_DESTROY_X3 0
+ %define src_ptr
+ %define src_stride
+ %define ref_ptr
+ %define ref_stride
+ %define end_ptr
+ %define ret_var
+ %define result_ptr
+ %define max_sad
+ %define height
+
+%if ABI_IS_32BIT
+ pop rbx
+ pop rdi
+ pop rsi
+ pop rbp
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ RESTORE_XMM
+ %endif
+%endif
+ ret
+%endmacro
+
+%macro STACK_FRAME_CREATE_X4 0
+%if ABI_IS_32BIT
+ %define src_ptr rsi
+ %define src_stride rax
+ %define r0_ptr rcx
+ %define r1_ptr rdx
+ %define r2_ptr rbx
+ %define r3_ptr rdi
+ %define ref_stride rbp
+ %define result_ptr arg(4)
+ push rbp
+ mov rbp, rsp
+ push rsi
+ push rdi
+ push rbx
+
+ push rbp
+ mov rdi, arg(2) ; ref_ptr_base
+
+ LOAD_X4_ADDRESSES rdi, rcx, rdx, rax, rdi
+
+ mov rsi, arg(0) ; src_ptr
+
+ movsxd rbx, dword ptr arg(1) ; src_stride
+ movsxd rbp, dword ptr arg(3) ; ref_stride
+
+ xchg rbx, rax
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ SAVE_XMM 7, u
+ %define src_ptr rcx
+ %define src_stride rdx
+ %define r0_ptr rsi
+ %define r1_ptr r10
+ %define r2_ptr r11
+ %define r3_ptr r8
+ %define ref_stride r9
+ %define result_ptr [rsp+xmm_stack_space+16+4*8]
+ push rsi
+
+ LOAD_X4_ADDRESSES r8, r0_ptr, r1_ptr, r2_ptr, r3_ptr
+ %else
+ %define src_ptr rdi
+ %define src_stride rsi
+ %define r0_ptr r9
+ %define r1_ptr r10
+ %define r2_ptr r11
+ %define r3_ptr rdx
+ %define ref_stride rcx
+ %define result_ptr r8
+
+ LOAD_X4_ADDRESSES rdx, r0_ptr, r1_ptr, r2_ptr, r3_ptr
+
+ %endif
+%endif
+%endmacro
+
+%macro STACK_FRAME_DESTROY_X4 0
+ %define src_ptr
+ %define src_stride
+ %define r0_ptr
+ %define r1_ptr
+ %define r2_ptr
+ %define r3_ptr
+ %define ref_stride
+ %define result_ptr
+
+%if ABI_IS_32BIT
+ pop rbx
+ pop rdi
+ pop rsi
+ pop rbp
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ pop rsi
+ RESTORE_XMM
+ %endif
+%endif
+ ret
+%endmacro
+
+%macro PROCESS_16X2X3 5
+%if %1==0
+ movdqa xmm0, XMMWORD PTR [%2]
+ lddqu xmm5, XMMWORD PTR [%3]
+ lddqu xmm6, XMMWORD PTR [%3+1]
+ lddqu xmm7, XMMWORD PTR [%3+2]
+
+ psadbw xmm5, xmm0
+ psadbw xmm6, xmm0
+ psadbw xmm7, xmm0
+%else
+ movdqa xmm0, XMMWORD PTR [%2]
+ lddqu xmm1, XMMWORD PTR [%3]
+ lddqu xmm2, XMMWORD PTR [%3+1]
+ lddqu xmm3, XMMWORD PTR [%3+2]
+
+ psadbw xmm1, xmm0
+ psadbw xmm2, xmm0
+ psadbw xmm3, xmm0
+
+ paddw xmm5, xmm1
+ paddw xmm6, xmm2
+ paddw xmm7, xmm3
+%endif
+ movdqa xmm0, XMMWORD PTR [%2+%4]
+ lddqu xmm1, XMMWORD PTR [%3+%5]
+ lddqu xmm2, XMMWORD PTR [%3+%5+1]
+ lddqu xmm3, XMMWORD PTR [%3+%5+2]
+
+%if %1==0 || %1==1
+ lea %2, [%2+%4*2]
+ lea %3, [%3+%5*2]
+%endif
+
+ psadbw xmm1, xmm0
+ psadbw xmm2, xmm0
+ psadbw xmm3, xmm0
+
+ paddw xmm5, xmm1
+ paddw xmm6, xmm2
+ paddw xmm7, xmm3
+%endmacro
+
+%macro PROCESS_8X2X3 5
+%if %1==0
+ movq mm0, QWORD PTR [%2]
+ movq mm5, QWORD PTR [%3]
+ movq mm6, QWORD PTR [%3+1]
+ movq mm7, QWORD PTR [%3+2]
+
+ psadbw mm5, mm0
+ psadbw mm6, mm0
+ psadbw mm7, mm0
+%else
+ movq mm0, QWORD PTR [%2]
+ movq mm1, QWORD PTR [%3]
+ movq mm2, QWORD PTR [%3+1]
+ movq mm3, QWORD PTR [%3+2]
+
+ psadbw mm1, mm0
+ psadbw mm2, mm0
+ psadbw mm3, mm0
+
+ paddw mm5, mm1
+ paddw mm6, mm2
+ paddw mm7, mm3
+%endif
+ movq mm0, QWORD PTR [%2+%4]
+ movq mm1, QWORD PTR [%3+%5]
+ movq mm2, QWORD PTR [%3+%5+1]
+ movq mm3, QWORD PTR [%3+%5+2]
+
+%if %1==0 || %1==1
+ lea %2, [%2+%4*2]
+ lea %3, [%3+%5*2]
+%endif
+
+ psadbw mm1, mm0
+ psadbw mm2, mm0
+ psadbw mm3, mm0
+
+ paddw mm5, mm1
+ paddw mm6, mm2
+ paddw mm7, mm3
+%endmacro
+
+%macro LOAD_X4_ADDRESSES 5
+ mov %2, [%1+REG_SZ_BYTES*0]
+ mov %3, [%1+REG_SZ_BYTES*1]
+
+ mov %4, [%1+REG_SZ_BYTES*2]
+ mov %5, [%1+REG_SZ_BYTES*3]
+%endmacro
+
+%macro PROCESS_16X2X4 8
+%if %1==0
+ movdqa xmm0, XMMWORD PTR [%2]
+ lddqu xmm4, XMMWORD PTR [%3]
+ lddqu xmm5, XMMWORD PTR [%4]
+ lddqu xmm6, XMMWORD PTR [%5]
+ lddqu xmm7, XMMWORD PTR [%6]
+
+ psadbw xmm4, xmm0
+ psadbw xmm5, xmm0
+ psadbw xmm6, xmm0
+ psadbw xmm7, xmm0
+%else
+ movdqa xmm0, XMMWORD PTR [%2]
+ lddqu xmm1, XMMWORD PTR [%3]
+ lddqu xmm2, XMMWORD PTR [%4]
+ lddqu xmm3, XMMWORD PTR [%5]
+
+ psadbw xmm1, xmm0
+ psadbw xmm2, xmm0
+ psadbw xmm3, xmm0
+
+ paddw xmm4, xmm1
+ lddqu xmm1, XMMWORD PTR [%6]
+ paddw xmm5, xmm2
+ paddw xmm6, xmm3
+
+ psadbw xmm1, xmm0
+ paddw xmm7, xmm1
+%endif
+ movdqa xmm0, XMMWORD PTR [%2+%7]
+ lddqu xmm1, XMMWORD PTR [%3+%8]
+ lddqu xmm2, XMMWORD PTR [%4+%8]
+ lddqu xmm3, XMMWORD PTR [%5+%8]
+
+ psadbw xmm1, xmm0
+ psadbw xmm2, xmm0
+ psadbw xmm3, xmm0
+
+ paddw xmm4, xmm1
+ lddqu xmm1, XMMWORD PTR [%6+%8]
+ paddw xmm5, xmm2
+ paddw xmm6, xmm3
+
+%if %1==0 || %1==1
+ lea %2, [%2+%7*2]
+ lea %3, [%3+%8*2]
+
+ lea %4, [%4+%8*2]
+ lea %5, [%5+%8*2]
+
+ lea %6, [%6+%8*2]
+%endif
+ psadbw xmm1, xmm0
+ paddw xmm7, xmm1
+
+%endmacro
+
+%macro PROCESS_8X2X4 8
+%if %1==0
+ movq mm0, QWORD PTR [%2]
+ movq mm4, QWORD PTR [%3]
+ movq mm5, QWORD PTR [%4]
+ movq mm6, QWORD PTR [%5]
+ movq mm7, QWORD PTR [%6]
+
+ psadbw mm4, mm0
+ psadbw mm5, mm0
+ psadbw mm6, mm0
+ psadbw mm7, mm0
+%else
+ movq mm0, QWORD PTR [%2]
+ movq mm1, QWORD PTR [%3]
+ movq mm2, QWORD PTR [%4]
+ movq mm3, QWORD PTR [%5]
+
+ psadbw mm1, mm0
+ psadbw mm2, mm0
+ psadbw mm3, mm0
+
+ paddw mm4, mm1
+ movq mm1, QWORD PTR [%6]
+ paddw mm5, mm2
+ paddw mm6, mm3
+
+ psadbw mm1, mm0
+ paddw mm7, mm1
+%endif
+ movq mm0, QWORD PTR [%2+%7]
+ movq mm1, QWORD PTR [%3+%8]
+ movq mm2, QWORD PTR [%4+%8]
+ movq mm3, QWORD PTR [%5+%8]
+
+ psadbw mm1, mm0
+ psadbw mm2, mm0
+ psadbw mm3, mm0
+
+ paddw mm4, mm1
+ movq mm1, QWORD PTR [%6+%8]
+ paddw mm5, mm2
+ paddw mm6, mm3
+
+%if %1==0 || %1==1
+ lea %2, [%2+%7*2]
+ lea %3, [%3+%8*2]
+
+ lea %4, [%4+%8*2]
+ lea %5, [%5+%8*2]
+
+ lea %6, [%6+%8*2]
+%endif
+ psadbw mm1, mm0
+ paddw mm7, mm1
+
+%endmacro
+
+;void int vp8_sad16x16x3_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad16x16x3_sse3) PRIVATE
+sym(vp8_sad16x16x3_sse3):
+
+ STACK_FRAME_CREATE_X3
+
+ PROCESS_16X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride
+
+ mov rcx, result_ptr
+
+ movq xmm0, xmm5
+ psrldq xmm5, 8
+
+ paddw xmm0, xmm5
+ movd [rcx], xmm0
+;-
+ movq xmm0, xmm6
+ psrldq xmm6, 8
+
+ paddw xmm0, xmm6
+ movd [rcx+4], xmm0
+;-
+ movq xmm0, xmm7
+ psrldq xmm7, 8
+
+ paddw xmm0, xmm7
+ movd [rcx+8], xmm0
+
+ STACK_FRAME_DESTROY_X3
+
+;void int vp8_sad16x8x3_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad16x8x3_sse3) PRIVATE
+sym(vp8_sad16x8x3_sse3):
+
+ STACK_FRAME_CREATE_X3
+
+ PROCESS_16X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_16X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride
+
+ mov rcx, result_ptr
+
+ movq xmm0, xmm5
+ psrldq xmm5, 8
+
+ paddw xmm0, xmm5
+ movd [rcx], xmm0
+;-
+ movq xmm0, xmm6
+ psrldq xmm6, 8
+
+ paddw xmm0, xmm6
+ movd [rcx+4], xmm0
+;-
+ movq xmm0, xmm7
+ psrldq xmm7, 8
+
+ paddw xmm0, xmm7
+ movd [rcx+8], xmm0
+
+ STACK_FRAME_DESTROY_X3
+
+;void int vp8_sad8x16x3_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad8x16x3_sse3) PRIVATE
+sym(vp8_sad8x16x3_sse3):
+
+ STACK_FRAME_CREATE_X3
+
+ PROCESS_8X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride
+
+ mov rcx, result_ptr
+
+ punpckldq mm5, mm6
+
+ movq [rcx], mm5
+ movd [rcx+8], mm7
+
+ STACK_FRAME_DESTROY_X3
+
+;void int vp8_sad8x8x3_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad8x8x3_sse3) PRIVATE
+sym(vp8_sad8x8x3_sse3):
+
+ STACK_FRAME_CREATE_X3
+
+ PROCESS_8X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
+ PROCESS_8X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride
+
+ mov rcx, result_ptr
+
+ punpckldq mm5, mm6
+
+ movq [rcx], mm5
+ movd [rcx+8], mm7
+
+ STACK_FRAME_DESTROY_X3
+
+;void int vp8_sad4x4x3_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad4x4x3_sse3) PRIVATE
+sym(vp8_sad4x4x3_sse3):
+
+ STACK_FRAME_CREATE_X3
+
+ movd mm0, DWORD PTR [src_ptr]
+ movd mm1, DWORD PTR [ref_ptr]
+
+ movd mm2, DWORD PTR [src_ptr+src_stride]
+ movd mm3, DWORD PTR [ref_ptr+ref_stride]
+
+ punpcklbw mm0, mm2
+ punpcklbw mm1, mm3
+
+ movd mm4, DWORD PTR [ref_ptr+1]
+ movd mm5, DWORD PTR [ref_ptr+2]
+
+ movd mm2, DWORD PTR [ref_ptr+ref_stride+1]
+ movd mm3, DWORD PTR [ref_ptr+ref_stride+2]
+
+ psadbw mm1, mm0
+
+ punpcklbw mm4, mm2
+ punpcklbw mm5, mm3
+
+ psadbw mm4, mm0
+ psadbw mm5, mm0
+
+ lea src_ptr, [src_ptr+src_stride*2]
+ lea ref_ptr, [ref_ptr+ref_stride*2]
+
+ movd mm0, DWORD PTR [src_ptr]
+ movd mm2, DWORD PTR [ref_ptr]
+
+ movd mm3, DWORD PTR [src_ptr+src_stride]
+ movd mm6, DWORD PTR [ref_ptr+ref_stride]
+
+ punpcklbw mm0, mm3
+ punpcklbw mm2, mm6
+
+ movd mm3, DWORD PTR [ref_ptr+1]
+ movd mm7, DWORD PTR [ref_ptr+2]
+
+ psadbw mm2, mm0
+
+ paddw mm1, mm2
+
+ movd mm2, DWORD PTR [ref_ptr+ref_stride+1]
+ movd mm6, DWORD PTR [ref_ptr+ref_stride+2]
+
+ punpcklbw mm3, mm2
+ punpcklbw mm7, mm6
+
+ psadbw mm3, mm0
+ psadbw mm7, mm0
+
+ paddw mm3, mm4
+ paddw mm7, mm5
+
+ mov rcx, result_ptr
+
+ punpckldq mm1, mm3
+
+ movq [rcx], mm1
+ movd [rcx+8], mm7
+
+ STACK_FRAME_DESTROY_X3
+
+;unsigned int vp8_sad16x16_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int max_sad)
+;%define lddqu movdqu
+global sym(vp8_sad16x16_sse3) PRIVATE
+sym(vp8_sad16x16_sse3):
+
+ STACK_FRAME_CREATE_X3
+
+ mov end_ptr, 4
+ pxor xmm7, xmm7
+
+.vp8_sad16x16_sse3_loop:
+ movdqa xmm0, XMMWORD PTR [src_ptr]
+ movdqu xmm1, XMMWORD PTR [ref_ptr]
+ movdqa xmm2, XMMWORD PTR [src_ptr+src_stride]
+ movdqu xmm3, XMMWORD PTR [ref_ptr+ref_stride]
+
+ lea src_ptr, [src_ptr+src_stride*2]
+ lea ref_ptr, [ref_ptr+ref_stride*2]
+
+ movdqa xmm4, XMMWORD PTR [src_ptr]
+ movdqu xmm5, XMMWORD PTR [ref_ptr]
+ movdqa xmm6, XMMWORD PTR [src_ptr+src_stride]
+
+ psadbw xmm0, xmm1
+
+ movdqu xmm1, XMMWORD PTR [ref_ptr+ref_stride]
+
+ psadbw xmm2, xmm3
+ psadbw xmm4, xmm5
+ psadbw xmm6, xmm1
+
+ lea src_ptr, [src_ptr+src_stride*2]
+ lea ref_ptr, [ref_ptr+ref_stride*2]
+
+ paddw xmm7, xmm0
+ paddw xmm7, xmm2
+ paddw xmm7, xmm4
+ paddw xmm7, xmm6
+
+ sub end_ptr, 1
+ jne .vp8_sad16x16_sse3_loop
+
+ movq xmm0, xmm7
+ psrldq xmm7, 8
+ paddw xmm0, xmm7
+ movq rax, xmm0
+
+ STACK_FRAME_DESTROY_X3
+
+;void vp8_copy32xn_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *dst_ptr,
+; int dst_stride,
+; int height);
+global sym(vp8_copy32xn_sse3) PRIVATE
+sym(vp8_copy32xn_sse3):
+
+ STACK_FRAME_CREATE_X3
+
+.block_copy_sse3_loopx4:
+ lea end_ptr, [src_ptr+src_stride*2]
+
+ movdqu xmm0, XMMWORD PTR [src_ptr]
+ movdqu xmm1, XMMWORD PTR [src_ptr + 16]
+ movdqu xmm2, XMMWORD PTR [src_ptr + src_stride]
+ movdqu xmm3, XMMWORD PTR [src_ptr + src_stride + 16]
+ movdqu xmm4, XMMWORD PTR [end_ptr]
+ movdqu xmm5, XMMWORD PTR [end_ptr + 16]
+ movdqu xmm6, XMMWORD PTR [end_ptr + src_stride]
+ movdqu xmm7, XMMWORD PTR [end_ptr + src_stride + 16]
+
+ lea src_ptr, [src_ptr+src_stride*4]
+
+ lea end_ptr, [ref_ptr+ref_stride*2]
+
+ movdqa XMMWORD PTR [ref_ptr], xmm0
+ movdqa XMMWORD PTR [ref_ptr + 16], xmm1
+ movdqa XMMWORD PTR [ref_ptr + ref_stride], xmm2
+ movdqa XMMWORD PTR [ref_ptr + ref_stride + 16], xmm3
+ movdqa XMMWORD PTR [end_ptr], xmm4
+ movdqa XMMWORD PTR [end_ptr + 16], xmm5
+ movdqa XMMWORD PTR [end_ptr + ref_stride], xmm6
+ movdqa XMMWORD PTR [end_ptr + ref_stride + 16], xmm7
+
+ lea ref_ptr, [ref_ptr+ref_stride*4]
+
+ sub height, 4
+ cmp height, 4
+ jge .block_copy_sse3_loopx4
+
+ ;Check to see if there is more rows need to be copied.
+ cmp height, 0
+ je .copy_is_done
+
+.block_copy_sse3_loop:
+ movdqu xmm0, XMMWORD PTR [src_ptr]
+ movdqu xmm1, XMMWORD PTR [src_ptr + 16]
+ lea src_ptr, [src_ptr+src_stride]
+
+ movdqa XMMWORD PTR [ref_ptr], xmm0
+ movdqa XMMWORD PTR [ref_ptr + 16], xmm1
+ lea ref_ptr, [ref_ptr+ref_stride]
+
+ sub height, 1
+ jne .block_copy_sse3_loop
+
+.copy_is_done:
+ STACK_FRAME_DESTROY_X3
+
+;void vp8_sad16x16x4d_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr_base,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad16x16x4d_sse3) PRIVATE
+sym(vp8_sad16x16x4d_sse3):
+
+ STACK_FRAME_CREATE_X4
+
+ PROCESS_16X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+
+%if ABI_IS_32BIT
+ pop rbp
+%endif
+ mov rcx, result_ptr
+
+ movq xmm0, xmm4
+ psrldq xmm4, 8
+
+ paddw xmm0, xmm4
+ movd [rcx], xmm0
+;-
+ movq xmm0, xmm5
+ psrldq xmm5, 8
+
+ paddw xmm0, xmm5
+ movd [rcx+4], xmm0
+;-
+ movq xmm0, xmm6
+ psrldq xmm6, 8
+
+ paddw xmm0, xmm6
+ movd [rcx+8], xmm0
+;-
+ movq xmm0, xmm7
+ psrldq xmm7, 8
+
+ paddw xmm0, xmm7
+ movd [rcx+12], xmm0
+
+ STACK_FRAME_DESTROY_X4
+
+;void vp8_sad16x8x4d_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr_base,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad16x8x4d_sse3) PRIVATE
+sym(vp8_sad16x8x4d_sse3):
+
+ STACK_FRAME_CREATE_X4
+
+ PROCESS_16X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_16X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+
+%if ABI_IS_32BIT
+ pop rbp
+%endif
+ mov rcx, result_ptr
+
+ movq xmm0, xmm4
+ psrldq xmm4, 8
+
+ paddw xmm0, xmm4
+ movd [rcx], xmm0
+;-
+ movq xmm0, xmm5
+ psrldq xmm5, 8
+
+ paddw xmm0, xmm5
+ movd [rcx+4], xmm0
+;-
+ movq xmm0, xmm6
+ psrldq xmm6, 8
+
+ paddw xmm0, xmm6
+ movd [rcx+8], xmm0
+;-
+ movq xmm0, xmm7
+ psrldq xmm7, 8
+
+ paddw xmm0, xmm7
+ movd [rcx+12], xmm0
+
+ STACK_FRAME_DESTROY_X4
+
+;void int vp8_sad8x16x4d_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad8x16x4d_sse3) PRIVATE
+sym(vp8_sad8x16x4d_sse3):
+
+ STACK_FRAME_CREATE_X4
+
+ PROCESS_8X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+
+%if ABI_IS_32BIT
+ pop rbp
+%endif
+ mov rcx, result_ptr
+
+ punpckldq mm4, mm5
+ punpckldq mm6, mm7
+
+ movq [rcx], mm4
+ movq [rcx+8], mm6
+
+ STACK_FRAME_DESTROY_X4
+
+;void int vp8_sad8x8x4d_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad8x8x4d_sse3) PRIVATE
+sym(vp8_sad8x8x4d_sse3):
+
+ STACK_FRAME_CREATE_X4
+
+ PROCESS_8X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+ PROCESS_8X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
+
+%if ABI_IS_32BIT
+ pop rbp
+%endif
+ mov rcx, result_ptr
+
+ punpckldq mm4, mm5
+ punpckldq mm6, mm7
+
+ movq [rcx], mm4
+ movq [rcx+8], mm6
+
+ STACK_FRAME_DESTROY_X4
+
+;void int vp8_sad4x4x4d_sse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad4x4x4d_sse3) PRIVATE
+sym(vp8_sad4x4x4d_sse3):
+
+ STACK_FRAME_CREATE_X4
+
+ movd mm0, DWORD PTR [src_ptr]
+ movd mm1, DWORD PTR [r0_ptr]
+
+ movd mm2, DWORD PTR [src_ptr+src_stride]
+ movd mm3, DWORD PTR [r0_ptr+ref_stride]
+
+ punpcklbw mm0, mm2
+ punpcklbw mm1, mm3
+
+ movd mm4, DWORD PTR [r1_ptr]
+ movd mm5, DWORD PTR [r2_ptr]
+
+ movd mm6, DWORD PTR [r3_ptr]
+ movd mm2, DWORD PTR [r1_ptr+ref_stride]
+
+ movd mm3, DWORD PTR [r2_ptr+ref_stride]
+ movd mm7, DWORD PTR [r3_ptr+ref_stride]
+
+ psadbw mm1, mm0
+
+ punpcklbw mm4, mm2
+ punpcklbw mm5, mm3
+
+ punpcklbw mm6, mm7
+ psadbw mm4, mm0
+
+ psadbw mm5, mm0
+ psadbw mm6, mm0
+
+
+
+ lea src_ptr, [src_ptr+src_stride*2]
+ lea r0_ptr, [r0_ptr+ref_stride*2]
+
+ lea r1_ptr, [r1_ptr+ref_stride*2]
+ lea r2_ptr, [r2_ptr+ref_stride*2]
+
+ lea r3_ptr, [r3_ptr+ref_stride*2]
+
+ movd mm0, DWORD PTR [src_ptr]
+ movd mm2, DWORD PTR [r0_ptr]
+
+ movd mm3, DWORD PTR [src_ptr+src_stride]
+ movd mm7, DWORD PTR [r0_ptr+ref_stride]
+
+ punpcklbw mm0, mm3
+ punpcklbw mm2, mm7
+
+ movd mm3, DWORD PTR [r1_ptr]
+ movd mm7, DWORD PTR [r2_ptr]
+
+ psadbw mm2, mm0
+%if ABI_IS_32BIT
+ mov rax, rbp
+
+ pop rbp
+%define ref_stride rax
+%endif
+ mov rsi, result_ptr
+
+ paddw mm1, mm2
+ movd [rsi], mm1
+
+ movd mm2, DWORD PTR [r1_ptr+ref_stride]
+ movd mm1, DWORD PTR [r2_ptr+ref_stride]
+
+ punpcklbw mm3, mm2
+ punpcklbw mm7, mm1
+
+ psadbw mm3, mm0
+ psadbw mm7, mm0
+
+ movd mm2, DWORD PTR [r3_ptr]
+ movd mm1, DWORD PTR [r3_ptr+ref_stride]
+
+ paddw mm3, mm4
+ paddw mm7, mm5
+
+ movd [rsi+4], mm3
+ punpcklbw mm2, mm1
+
+ movd [rsi+8], mm7
+ psadbw mm2, mm0
+
+ paddw mm2, mm6
+ movd [rsi+12], mm2
+
+
+ STACK_FRAME_DESTROY_X4
+
diff --git a/libvpx/vp8/common/x86/sad_sse4.asm b/libvpx/vp8/common/x86/sad_sse4.asm
new file mode 100644
index 0000000..f7fccd7
--- /dev/null
+++ b/libvpx/vp8/common/x86/sad_sse4.asm
@@ -0,0 +1,353 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%macro PROCESS_16X2X8 1
+%if %1
+ movdqa xmm0, XMMWORD PTR [rsi]
+ movq xmm1, MMWORD PTR [rdi]
+ movq xmm3, MMWORD PTR [rdi+8]
+ movq xmm2, MMWORD PTR [rdi+16]
+ punpcklqdq xmm1, xmm3
+ punpcklqdq xmm3, xmm2
+
+ movdqa xmm2, xmm1
+ mpsadbw xmm1, xmm0, 0x0
+ mpsadbw xmm2, xmm0, 0x5
+
+ psrldq xmm0, 8
+
+ movdqa xmm4, xmm3
+ mpsadbw xmm3, xmm0, 0x0
+ mpsadbw xmm4, xmm0, 0x5
+
+ paddw xmm1, xmm2
+ paddw xmm1, xmm3
+ paddw xmm1, xmm4
+%else
+ movdqa xmm0, XMMWORD PTR [rsi]
+ movq xmm5, MMWORD PTR [rdi]
+ movq xmm3, MMWORD PTR [rdi+8]
+ movq xmm2, MMWORD PTR [rdi+16]
+ punpcklqdq xmm5, xmm3
+ punpcklqdq xmm3, xmm2
+
+ movdqa xmm2, xmm5
+ mpsadbw xmm5, xmm0, 0x0
+ mpsadbw xmm2, xmm0, 0x5
+
+ psrldq xmm0, 8
+
+ movdqa xmm4, xmm3
+ mpsadbw xmm3, xmm0, 0x0
+ mpsadbw xmm4, xmm0, 0x5
+
+ paddw xmm5, xmm2
+ paddw xmm5, xmm3
+ paddw xmm5, xmm4
+
+ paddw xmm1, xmm5
+%endif
+ movdqa xmm0, XMMWORD PTR [rsi + rax]
+ movq xmm5, MMWORD PTR [rdi+ rdx]
+ movq xmm3, MMWORD PTR [rdi+ rdx+8]
+ movq xmm2, MMWORD PTR [rdi+ rdx+16]
+ punpcklqdq xmm5, xmm3
+ punpcklqdq xmm3, xmm2
+
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rdx*2]
+
+ movdqa xmm2, xmm5
+ mpsadbw xmm5, xmm0, 0x0
+ mpsadbw xmm2, xmm0, 0x5
+
+ psrldq xmm0, 8
+ movdqa xmm4, xmm3
+ mpsadbw xmm3, xmm0, 0x0
+ mpsadbw xmm4, xmm0, 0x5
+
+ paddw xmm5, xmm2
+ paddw xmm5, xmm3
+ paddw xmm5, xmm4
+
+ paddw xmm1, xmm5
+%endmacro
+
+%macro PROCESS_8X2X8 1
+%if %1
+ movq xmm0, MMWORD PTR [rsi]
+ movq xmm1, MMWORD PTR [rdi]
+ movq xmm3, MMWORD PTR [rdi+8]
+ punpcklqdq xmm1, xmm3
+
+ movdqa xmm2, xmm1
+ mpsadbw xmm1, xmm0, 0x0
+ mpsadbw xmm2, xmm0, 0x5
+ paddw xmm1, xmm2
+%else
+ movq xmm0, MMWORD PTR [rsi]
+ movq xmm5, MMWORD PTR [rdi]
+ movq xmm3, MMWORD PTR [rdi+8]
+ punpcklqdq xmm5, xmm3
+
+ movdqa xmm2, xmm5
+ mpsadbw xmm5, xmm0, 0x0
+ mpsadbw xmm2, xmm0, 0x5
+ paddw xmm5, xmm2
+
+ paddw xmm1, xmm5
+%endif
+ movq xmm0, MMWORD PTR [rsi + rax]
+ movq xmm5, MMWORD PTR [rdi+ rdx]
+ movq xmm3, MMWORD PTR [rdi+ rdx+8]
+ punpcklqdq xmm5, xmm3
+
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rdx*2]
+
+ movdqa xmm2, xmm5
+ mpsadbw xmm5, xmm0, 0x0
+ mpsadbw xmm2, xmm0, 0x5
+ paddw xmm5, xmm2
+
+ paddw xmm1, xmm5
+%endmacro
+
+%macro PROCESS_4X2X8 1
+%if %1
+ movd xmm0, [rsi]
+ movq xmm1, MMWORD PTR [rdi]
+ movq xmm3, MMWORD PTR [rdi+8]
+ punpcklqdq xmm1, xmm3
+
+ mpsadbw xmm1, xmm0, 0x0
+%else
+ movd xmm0, [rsi]
+ movq xmm5, MMWORD PTR [rdi]
+ movq xmm3, MMWORD PTR [rdi+8]
+ punpcklqdq xmm5, xmm3
+
+ mpsadbw xmm5, xmm0, 0x0
+
+ paddw xmm1, xmm5
+%endif
+ movd xmm0, [rsi + rax]
+ movq xmm5, MMWORD PTR [rdi+ rdx]
+ movq xmm3, MMWORD PTR [rdi+ rdx+8]
+ punpcklqdq xmm5, xmm3
+
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rdx*2]
+
+ mpsadbw xmm5, xmm0, 0x0
+
+ paddw xmm1, xmm5
+%endmacro
+
+
+;void vp8_sad16x16x8_sse4(
+; const unsigned char *src_ptr,
+; int src_stride,
+; const unsigned char *ref_ptr,
+; int ref_stride,
+; unsigned short *sad_array);
+global sym(vp8_sad16x16x8_sse4) PRIVATE
+sym(vp8_sad16x16x8_sse4):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ PROCESS_16X2X8 1
+ PROCESS_16X2X8 0
+ PROCESS_16X2X8 0
+ PROCESS_16X2X8 0
+ PROCESS_16X2X8 0
+ PROCESS_16X2X8 0
+ PROCESS_16X2X8 0
+ PROCESS_16X2X8 0
+
+ mov rdi, arg(4) ;Results
+ movdqa XMMWORD PTR [rdi], xmm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_sad16x8x8_sse4(
+; const unsigned char *src_ptr,
+; int src_stride,
+; const unsigned char *ref_ptr,
+; int ref_stride,
+; unsigned short *sad_array
+;);
+global sym(vp8_sad16x8x8_sse4) PRIVATE
+sym(vp8_sad16x8x8_sse4):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ PROCESS_16X2X8 1
+ PROCESS_16X2X8 0
+ PROCESS_16X2X8 0
+ PROCESS_16X2X8 0
+
+ mov rdi, arg(4) ;Results
+ movdqa XMMWORD PTR [rdi], xmm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_sad8x8x8_sse4(
+; const unsigned char *src_ptr,
+; int src_stride,
+; const unsigned char *ref_ptr,
+; int ref_stride,
+; unsigned short *sad_array
+;);
+global sym(vp8_sad8x8x8_sse4) PRIVATE
+sym(vp8_sad8x8x8_sse4):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ PROCESS_8X2X8 1
+ PROCESS_8X2X8 0
+ PROCESS_8X2X8 0
+ PROCESS_8X2X8 0
+
+ mov rdi, arg(4) ;Results
+ movdqa XMMWORD PTR [rdi], xmm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_sad8x16x8_sse4(
+; const unsigned char *src_ptr,
+; int src_stride,
+; const unsigned char *ref_ptr,
+; int ref_stride,
+; unsigned short *sad_array
+;);
+global sym(vp8_sad8x16x8_sse4) PRIVATE
+sym(vp8_sad8x16x8_sse4):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ PROCESS_8X2X8 1
+ PROCESS_8X2X8 0
+ PROCESS_8X2X8 0
+ PROCESS_8X2X8 0
+ PROCESS_8X2X8 0
+ PROCESS_8X2X8 0
+ PROCESS_8X2X8 0
+ PROCESS_8X2X8 0
+ mov rdi, arg(4) ;Results
+ movdqa XMMWORD PTR [rdi], xmm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_sad4x4x8_c(
+; const unsigned char *src_ptr,
+; int src_stride,
+; const unsigned char *ref_ptr,
+; int ref_stride,
+; unsigned short *sad_array
+;);
+global sym(vp8_sad4x4x8_sse4) PRIVATE
+sym(vp8_sad4x4x8_sse4):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ PROCESS_4X2X8 1
+ PROCESS_4X2X8 0
+
+ mov rdi, arg(4) ;Results
+ movdqa XMMWORD PTR [rdi], xmm1
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+
+
diff --git a/libvpx/vp8/common/x86/sad_ssse3.asm b/libvpx/vp8/common/x86/sad_ssse3.asm
new file mode 100644
index 0000000..278fc06
--- /dev/null
+++ b/libvpx/vp8/common/x86/sad_ssse3.asm
@@ -0,0 +1,370 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%macro PROCESS_16X2X3 1
+%if %1
+ movdqa xmm0, XMMWORD PTR [rsi]
+ lddqu xmm5, XMMWORD PTR [rdi]
+ lddqu xmm6, XMMWORD PTR [rdi+1]
+ lddqu xmm7, XMMWORD PTR [rdi+2]
+
+ psadbw xmm5, xmm0
+ psadbw xmm6, xmm0
+ psadbw xmm7, xmm0
+%else
+ movdqa xmm0, XMMWORD PTR [rsi]
+ lddqu xmm1, XMMWORD PTR [rdi]
+ lddqu xmm2, XMMWORD PTR [rdi+1]
+ lddqu xmm3, XMMWORD PTR [rdi+2]
+
+ psadbw xmm1, xmm0
+ psadbw xmm2, xmm0
+ psadbw xmm3, xmm0
+
+ paddw xmm5, xmm1
+ paddw xmm6, xmm2
+ paddw xmm7, xmm3
+%endif
+ movdqa xmm0, XMMWORD PTR [rsi+rax]
+ lddqu xmm1, XMMWORD PTR [rdi+rdx]
+ lddqu xmm2, XMMWORD PTR [rdi+rdx+1]
+ lddqu xmm3, XMMWORD PTR [rdi+rdx+2]
+
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rdx*2]
+
+ psadbw xmm1, xmm0
+ psadbw xmm2, xmm0
+ psadbw xmm3, xmm0
+
+ paddw xmm5, xmm1
+ paddw xmm6, xmm2
+ paddw xmm7, xmm3
+%endmacro
+
+%macro PROCESS_16X2X3_OFFSET 2
+%if %1
+ movdqa xmm0, XMMWORD PTR [rsi]
+ movdqa xmm4, XMMWORD PTR [rdi]
+ movdqa xmm7, XMMWORD PTR [rdi+16]
+
+ movdqa xmm5, xmm7
+ palignr xmm5, xmm4, %2
+
+ movdqa xmm6, xmm7
+ palignr xmm6, xmm4, (%2+1)
+
+ palignr xmm7, xmm4, (%2+2)
+
+ psadbw xmm5, xmm0
+ psadbw xmm6, xmm0
+ psadbw xmm7, xmm0
+%else
+ movdqa xmm0, XMMWORD PTR [rsi]
+ movdqa xmm4, XMMWORD PTR [rdi]
+ movdqa xmm3, XMMWORD PTR [rdi+16]
+
+ movdqa xmm1, xmm3
+ palignr xmm1, xmm4, %2
+
+ movdqa xmm2, xmm3
+ palignr xmm2, xmm4, (%2+1)
+
+ palignr xmm3, xmm4, (%2+2)
+
+ psadbw xmm1, xmm0
+ psadbw xmm2, xmm0
+ psadbw xmm3, xmm0
+
+ paddw xmm5, xmm1
+ paddw xmm6, xmm2
+ paddw xmm7, xmm3
+%endif
+ movdqa xmm0, XMMWORD PTR [rsi+rax]
+ movdqa xmm4, XMMWORD PTR [rdi+rdx]
+ movdqa xmm3, XMMWORD PTR [rdi+rdx+16]
+
+ movdqa xmm1, xmm3
+ palignr xmm1, xmm4, %2
+
+ movdqa xmm2, xmm3
+ palignr xmm2, xmm4, (%2+1)
+
+ palignr xmm3, xmm4, (%2+2)
+
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rdx*2]
+
+ psadbw xmm1, xmm0
+ psadbw xmm2, xmm0
+ psadbw xmm3, xmm0
+
+ paddw xmm5, xmm1
+ paddw xmm6, xmm2
+ paddw xmm7, xmm3
+%endmacro
+
+%macro PROCESS_16X16X3_OFFSET 2
+%2_aligned_by_%1:
+
+ sub rdi, %1
+
+ PROCESS_16X2X3_OFFSET 1, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+
+ jmp %2_store_off
+
+%endmacro
+
+%macro PROCESS_16X8X3_OFFSET 2
+%2_aligned_by_%1:
+
+ sub rdi, %1
+
+ PROCESS_16X2X3_OFFSET 1, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+ PROCESS_16X2X3_OFFSET 0, %1
+
+ jmp %2_store_off
+
+%endmacro
+
+;void int vp8_sad16x16x3_ssse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad16x16x3_ssse3) PRIVATE
+sym(vp8_sad16x16x3_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ push rsi
+ push rdi
+ push rcx
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ mov rdx, 0xf
+ and rdx, rdi
+
+ jmp .vp8_sad16x16x3_ssse3_skiptable
+.vp8_sad16x16x3_ssse3_jumptable:
+ dd .vp8_sad16x16x3_ssse3_aligned_by_0 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_1 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_2 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_3 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_4 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_5 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_6 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_7 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_8 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_9 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_10 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_11 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_12 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_13 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_14 - .vp8_sad16x16x3_ssse3_do_jump
+ dd .vp8_sad16x16x3_ssse3_aligned_by_15 - .vp8_sad16x16x3_ssse3_do_jump
+.vp8_sad16x16x3_ssse3_skiptable:
+
+ call .vp8_sad16x16x3_ssse3_do_jump
+.vp8_sad16x16x3_ssse3_do_jump:
+ pop rcx ; get the address of do_jump
+ mov rax, .vp8_sad16x16x3_ssse3_jumptable - .vp8_sad16x16x3_ssse3_do_jump
+ add rax, rcx ; get the absolute address of vp8_sad16x16x3_ssse3_jumptable
+
+ movsxd rax, dword [rax + 4*rdx] ; get the 32 bit offset from the jumptable
+ add rcx, rax
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ jmp rcx
+
+ PROCESS_16X16X3_OFFSET 0, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 1, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 2, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 3, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 4, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 5, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 6, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 7, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 8, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 9, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 10, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 11, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 12, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 13, .vp8_sad16x16x3_ssse3
+ PROCESS_16X16X3_OFFSET 14, .vp8_sad16x16x3_ssse3
+
+.vp8_sad16x16x3_ssse3_aligned_by_15:
+ PROCESS_16X2X3 1
+ PROCESS_16X2X3 0
+ PROCESS_16X2X3 0
+ PROCESS_16X2X3 0
+ PROCESS_16X2X3 0
+ PROCESS_16X2X3 0
+ PROCESS_16X2X3 0
+ PROCESS_16X2X3 0
+
+.vp8_sad16x16x3_ssse3_store_off:
+ mov rdi, arg(4) ;Results
+
+ movq xmm0, xmm5
+ psrldq xmm5, 8
+
+ paddw xmm0, xmm5
+ movd [rdi], xmm0
+;-
+ movq xmm0, xmm6
+ psrldq xmm6, 8
+
+ paddw xmm0, xmm6
+ movd [rdi+4], xmm0
+;-
+ movq xmm0, xmm7
+ psrldq xmm7, 8
+
+ paddw xmm0, xmm7
+ movd [rdi+8], xmm0
+
+ ; begin epilog
+ pop rcx
+ pop rdi
+ pop rsi
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void int vp8_sad16x8x3_ssse3(
+; unsigned char *src_ptr,
+; int src_stride,
+; unsigned char *ref_ptr,
+; int ref_stride,
+; int *results)
+global sym(vp8_sad16x8x3_ssse3) PRIVATE
+sym(vp8_sad16x8x3_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
+ push rsi
+ push rdi
+ push rcx
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;ref_ptr
+
+ mov rdx, 0xf
+ and rdx, rdi
+
+ jmp .vp8_sad16x8x3_ssse3_skiptable
+.vp8_sad16x8x3_ssse3_jumptable:
+ dd .vp8_sad16x8x3_ssse3_aligned_by_0 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_1 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_2 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_3 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_4 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_5 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_6 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_7 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_8 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_9 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_10 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_11 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_12 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_13 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_14 - .vp8_sad16x8x3_ssse3_do_jump
+ dd .vp8_sad16x8x3_ssse3_aligned_by_15 - .vp8_sad16x8x3_ssse3_do_jump
+.vp8_sad16x8x3_ssse3_skiptable:
+
+ call .vp8_sad16x8x3_ssse3_do_jump
+.vp8_sad16x8x3_ssse3_do_jump:
+ pop rcx ; get the address of do_jump
+ mov rax, .vp8_sad16x8x3_ssse3_jumptable - .vp8_sad16x8x3_ssse3_do_jump
+ add rax, rcx ; get the absolute address of vp8_sad16x8x3_ssse3_jumptable
+
+ movsxd rax, dword [rax + 4*rdx] ; get the 32 bit offset from the jumptable
+ add rcx, rax
+
+ movsxd rax, dword ptr arg(1) ;src_stride
+ movsxd rdx, dword ptr arg(3) ;ref_stride
+
+ jmp rcx
+
+ PROCESS_16X8X3_OFFSET 0, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 1, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 2, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 3, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 4, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 5, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 6, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 7, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 8, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 9, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 10, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 11, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 12, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 13, .vp8_sad16x8x3_ssse3
+ PROCESS_16X8X3_OFFSET 14, .vp8_sad16x8x3_ssse3
+
+.vp8_sad16x8x3_ssse3_aligned_by_15:
+
+ PROCESS_16X2X3 1
+ PROCESS_16X2X3 0
+ PROCESS_16X2X3 0
+ PROCESS_16X2X3 0
+
+.vp8_sad16x8x3_ssse3_store_off:
+ mov rdi, arg(4) ;Results
+
+ movq xmm0, xmm5
+ psrldq xmm5, 8
+
+ paddw xmm0, xmm5
+ movd [rdi], xmm0
+;-
+ movq xmm0, xmm6
+ psrldq xmm6, 8
+
+ paddw xmm0, xmm6
+ movd [rdi+4], xmm0
+;-
+ movq xmm0, xmm7
+ psrldq xmm7, 8
+
+ paddw xmm0, xmm7
+ movd [rdi+8], xmm0
+
+ ; begin epilog
+ pop rcx
+ pop rdi
+ pop rsi
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/common/x86/subpixel_mmx.asm b/libvpx/vp8/common/x86/subpixel_mmx.asm
new file mode 100644
index 0000000..47dd452
--- /dev/null
+++ b/libvpx/vp8/common/x86/subpixel_mmx.asm
@@ -0,0 +1,702 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+extern sym(vp8_bilinear_filters_x86_8)
+
+
+%define BLOCK_HEIGHT_WIDTH 4
+%define vp8_filter_weight 128
+%define VP8_FILTER_SHIFT 7
+
+
+;void vp8_filter_block1d_h6_mmx
+;(
+; unsigned char *src_ptr,
+; unsigned short *output_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned int pixel_step,
+; unsigned int output_height,
+; unsigned int output_width,
+; short * vp8_filter
+;)
+global sym(vp8_filter_block1d_h6_mmx) PRIVATE
+sym(vp8_filter_block1d_h6_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdx, arg(6) ;vp8_filter
+
+ movq mm1, [rdx + 16] ; do both the negative taps first!!!
+ movq mm2, [rdx + 32] ;
+ movq mm6, [rdx + 48] ;
+ movq mm7, [rdx + 64] ;
+
+ mov rdi, arg(1) ;output_ptr
+ mov rsi, arg(0) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;output_height
+ movsxd rax, dword ptr arg(5) ;output_width ; destination pitch?
+ pxor mm0, mm0 ; mm0 = 00000000
+
+.nextrow:
+ movq mm3, [rsi-2] ; mm3 = p-2..p5
+ movq mm4, mm3 ; mm4 = p-2..p5
+ psrlq mm3, 8 ; mm3 = p-1..p5
+ punpcklbw mm3, mm0 ; mm3 = p-1..p2
+ pmullw mm3, mm1 ; mm3 *= kernel 1 modifiers.
+
+ movq mm5, mm4 ; mm5 = p-2..p5
+ punpckhbw mm4, mm0 ; mm5 = p2..p5
+ pmullw mm4, mm7 ; mm5 *= kernel 4 modifiers
+ paddsw mm3, mm4 ; mm3 += mm5
+
+ movq mm4, mm5 ; mm4 = p-2..p5;
+ psrlq mm5, 16 ; mm5 = p0..p5;
+ punpcklbw mm5, mm0 ; mm5 = p0..p3
+ pmullw mm5, mm2 ; mm5 *= kernel 2 modifiers
+ paddsw mm3, mm5 ; mm3 += mm5
+
+ movq mm5, mm4 ; mm5 = p-2..p5
+ psrlq mm4, 24 ; mm4 = p1..p5
+ punpcklbw mm4, mm0 ; mm4 = p1..p4
+ pmullw mm4, mm6 ; mm5 *= kernel 3 modifiers
+ paddsw mm3, mm4 ; mm3 += mm5
+
+ ; do outer positive taps
+ movd mm4, [rsi+3]
+ punpcklbw mm4, mm0 ; mm5 = p3..p6
+ pmullw mm4, [rdx+80] ; mm5 *= kernel 0 modifiers
+ paddsw mm3, mm4 ; mm3 += mm5
+
+ punpcklbw mm5, mm0 ; mm5 = p-2..p1
+ pmullw mm5, [rdx] ; mm5 *= kernel 5 modifiers
+ paddsw mm3, mm5 ; mm3 += mm5
+
+ paddsw mm3, [GLOBAL(rd)] ; mm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; mm3 /= 128
+ packuswb mm3, mm0 ; pack and unpack to saturate
+ punpcklbw mm3, mm0 ;
+
+ movq [rdi], mm3 ; store the results in the destination
+
+%if ABI_IS_32BIT
+ add rsi, dword ptr arg(2) ;src_pixels_per_line ; next line
+ add rdi, rax;
+%else
+ movsxd r8, dword ptr arg(2) ;src_pixels_per_line
+ add rdi, rax;
+
+ add rsi, r8 ; next line
+%endif
+
+ dec rcx ; decrement count
+ jnz .nextrow ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_filter_block1dc_v6_mmx
+;(
+; short *src_ptr,
+; unsigned char *output_ptr,
+; int output_pitch,
+; unsigned int pixels_per_line,
+; unsigned int pixel_step,
+; unsigned int output_height,
+; unsigned int output_width,
+; short * vp8_filter
+;)
+global sym(vp8_filter_block1dc_v6_mmx) PRIVATE
+sym(vp8_filter_block1dc_v6_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 8
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movq mm5, [GLOBAL(rd)]
+ push rbx
+ mov rbx, arg(7) ;vp8_filter
+ movq mm1, [rbx + 16] ; do both the negative taps first!!!
+ movq mm2, [rbx + 32] ;
+ movq mm6, [rbx + 48] ;
+ movq mm7, [rbx + 64] ;
+
+ movsxd rdx, dword ptr arg(3) ;pixels_per_line
+ mov rdi, arg(1) ;output_ptr
+ mov rsi, arg(0) ;src_ptr
+ sub rsi, rdx
+ sub rsi, rdx
+ movsxd rcx, DWORD PTR arg(5) ;output_height
+ movsxd rax, DWORD PTR arg(2) ;output_pitch ; destination pitch?
+ pxor mm0, mm0 ; mm0 = 00000000
+
+
+.nextrow_cv:
+ movq mm3, [rsi+rdx] ; mm3 = p0..p8 = row -1
+ pmullw mm3, mm1 ; mm3 *= kernel 1 modifiers.
+
+
+ movq mm4, [rsi + 4*rdx] ; mm4 = p0..p3 = row 2
+ pmullw mm4, mm7 ; mm4 *= kernel 4 modifiers.
+ paddsw mm3, mm4 ; mm3 += mm4
+
+ movq mm4, [rsi + 2*rdx] ; mm4 = p0..p3 = row 0
+ pmullw mm4, mm2 ; mm4 *= kernel 2 modifiers.
+ paddsw mm3, mm4 ; mm3 += mm4
+
+ movq mm4, [rsi] ; mm4 = p0..p3 = row -2
+ pmullw mm4, [rbx] ; mm4 *= kernel 0 modifiers.
+ paddsw mm3, mm4 ; mm3 += mm4
+
+
+ add rsi, rdx ; move source forward 1 line to avoid 3 * pitch
+ movq mm4, [rsi + 2*rdx] ; mm4 = p0..p3 = row 1
+ pmullw mm4, mm6 ; mm4 *= kernel 3 modifiers.
+ paddsw mm3, mm4 ; mm3 += mm4
+
+ movq mm4, [rsi + 4*rdx] ; mm4 = p0..p3 = row 3
+ pmullw mm4, [rbx +80] ; mm4 *= kernel 3 modifiers.
+ paddsw mm3, mm4 ; mm3 += mm4
+
+
+ paddsw mm3, mm5 ; mm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; mm3 /= 128
+ packuswb mm3, mm0 ; pack and saturate
+
+ movd [rdi],mm3 ; store the results in the destination
+ ; the subsequent iterations repeat 3 out of 4 of these reads. Since the
+ ; recon block should be in cache this shouldn't cost much. Its obviously
+ ; avoidable!!!.
+ lea rdi, [rdi+rax] ;
+ dec rcx ; decrement count
+ jnz .nextrow_cv ; next row
+
+ pop rbx
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void bilinear_predict8x8_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; int xoffset,
+; int yoffset,
+; unsigned char *dst_ptr,
+; int dst_pitch
+;)
+global sym(vp8_bilinear_predict8x8_mmx) PRIVATE
+sym(vp8_bilinear_predict8x8_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ;const short *HFilter = vp8_bilinear_filters_x86_8[xoffset];
+ ;const short *VFilter = vp8_bilinear_filters_x86_8[yoffset];
+
+ movsxd rax, dword ptr arg(2) ;xoffset
+ mov rdi, arg(4) ;dst_ptr ;
+
+ shl rax, 5 ; offset * 32
+ lea rcx, [GLOBAL(sym(vp8_bilinear_filters_x86_8))]
+
+ add rax, rcx ; HFilter
+ mov rsi, arg(0) ;src_ptr ;
+
+ movsxd rdx, dword ptr arg(5) ;dst_pitch
+ movq mm1, [rax] ;
+
+ movq mm2, [rax+16] ;
+ movsxd rax, dword ptr arg(3) ;yoffset
+
+ pxor mm0, mm0 ;
+
+ shl rax, 5 ; offset*32
+ add rax, rcx ; VFilter
+
+ lea rcx, [rdi+rdx*8] ;
+ movsxd rdx, dword ptr arg(1) ;src_pixels_per_line ;
+
+
+
+ ; get the first horizontal line done ;
+ movq mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ movq mm4, mm3 ; make a copy of current line
+
+ punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
+ punpckhbw mm4, mm0 ;
+
+ pmullw mm3, mm1 ;
+ pmullw mm4, mm1 ;
+
+ movq mm5, [rsi+1] ;
+ movq mm6, mm5 ;
+
+ punpcklbw mm5, mm0 ;
+ punpckhbw mm6, mm0 ;
+
+ pmullw mm5, mm2 ;
+ pmullw mm6, mm2 ;
+
+ paddw mm3, mm5 ;
+ paddw mm4, mm6 ;
+
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw mm4, [GLOBAL(rd)] ;
+ psraw mm4, VP8_FILTER_SHIFT ;
+
+ movq mm7, mm3 ;
+ packuswb mm7, mm4 ;
+
+ add rsi, rdx ; next line
+.next_row_8x8:
+ movq mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ movq mm4, mm3 ; make a copy of current line
+
+ punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
+ punpckhbw mm4, mm0 ;
+
+ pmullw mm3, mm1 ;
+ pmullw mm4, mm1 ;
+
+ movq mm5, [rsi+1] ;
+ movq mm6, mm5 ;
+
+ punpcklbw mm5, mm0 ;
+ punpckhbw mm6, mm0 ;
+
+ pmullw mm5, mm2 ;
+ pmullw mm6, mm2 ;
+
+ paddw mm3, mm5 ;
+ paddw mm4, mm6 ;
+
+ movq mm5, mm7 ;
+ movq mm6, mm7 ;
+
+ punpcklbw mm5, mm0 ;
+ punpckhbw mm6, mm0
+
+ pmullw mm5, [rax] ;
+ pmullw mm6, [rax] ;
+
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw mm4, [GLOBAL(rd)] ;
+ psraw mm4, VP8_FILTER_SHIFT ;
+
+ movq mm7, mm3 ;
+ packuswb mm7, mm4 ;
+
+
+ pmullw mm3, [rax+16] ;
+ pmullw mm4, [rax+16] ;
+
+ paddw mm3, mm5 ;
+ paddw mm4, mm6 ;
+
+
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw mm4, [GLOBAL(rd)] ;
+ psraw mm4, VP8_FILTER_SHIFT ;
+
+ packuswb mm3, mm4
+
+ movq [rdi], mm3 ; store the results in the destination
+
+%if ABI_IS_32BIT
+ add rsi, rdx ; next line
+ add rdi, dword ptr arg(5) ;dst_pitch ;
+%else
+ movsxd r8, dword ptr arg(5) ;dst_pitch
+ add rsi, rdx ; next line
+ add rdi, r8 ;dst_pitch
+%endif
+ cmp rdi, rcx ;
+ jne .next_row_8x8
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void bilinear_predict8x4_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; int xoffset,
+; int yoffset,
+; unsigned char *dst_ptr,
+; int dst_pitch
+;)
+global sym(vp8_bilinear_predict8x4_mmx) PRIVATE
+sym(vp8_bilinear_predict8x4_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ;const short *HFilter = vp8_bilinear_filters_x86_8[xoffset];
+ ;const short *VFilter = vp8_bilinear_filters_x86_8[yoffset];
+
+ movsxd rax, dword ptr arg(2) ;xoffset
+ mov rdi, arg(4) ;dst_ptr ;
+
+ lea rcx, [GLOBAL(sym(vp8_bilinear_filters_x86_8))]
+ shl rax, 5
+
+ mov rsi, arg(0) ;src_ptr ;
+ add rax, rcx
+
+ movsxd rdx, dword ptr arg(5) ;dst_pitch
+ movq mm1, [rax] ;
+
+ movq mm2, [rax+16] ;
+ movsxd rax, dword ptr arg(3) ;yoffset
+
+ pxor mm0, mm0 ;
+ shl rax, 5
+
+ add rax, rcx
+ lea rcx, [rdi+rdx*4] ;
+
+ movsxd rdx, dword ptr arg(1) ;src_pixels_per_line ;
+
+ ; get the first horizontal line done ;
+ movq mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ movq mm4, mm3 ; make a copy of current line
+
+ punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
+ punpckhbw mm4, mm0 ;
+
+ pmullw mm3, mm1 ;
+ pmullw mm4, mm1 ;
+
+ movq mm5, [rsi+1] ;
+ movq mm6, mm5 ;
+
+ punpcklbw mm5, mm0 ;
+ punpckhbw mm6, mm0 ;
+
+ pmullw mm5, mm2 ;
+ pmullw mm6, mm2 ;
+
+ paddw mm3, mm5 ;
+ paddw mm4, mm6 ;
+
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw mm4, [GLOBAL(rd)] ;
+ psraw mm4, VP8_FILTER_SHIFT ;
+
+ movq mm7, mm3 ;
+ packuswb mm7, mm4 ;
+
+ add rsi, rdx ; next line
+.next_row_8x4:
+ movq mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ movq mm4, mm3 ; make a copy of current line
+
+ punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
+ punpckhbw mm4, mm0 ;
+
+ pmullw mm3, mm1 ;
+ pmullw mm4, mm1 ;
+
+ movq mm5, [rsi+1] ;
+ movq mm6, mm5 ;
+
+ punpcklbw mm5, mm0 ;
+ punpckhbw mm6, mm0 ;
+
+ pmullw mm5, mm2 ;
+ pmullw mm6, mm2 ;
+
+ paddw mm3, mm5 ;
+ paddw mm4, mm6 ;
+
+ movq mm5, mm7 ;
+ movq mm6, mm7 ;
+
+ punpcklbw mm5, mm0 ;
+ punpckhbw mm6, mm0
+
+ pmullw mm5, [rax] ;
+ pmullw mm6, [rax] ;
+
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw mm4, [GLOBAL(rd)] ;
+ psraw mm4, VP8_FILTER_SHIFT ;
+
+ movq mm7, mm3 ;
+ packuswb mm7, mm4 ;
+
+
+ pmullw mm3, [rax+16] ;
+ pmullw mm4, [rax+16] ;
+
+ paddw mm3, mm5 ;
+ paddw mm4, mm6 ;
+
+
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw mm4, [GLOBAL(rd)] ;
+ psraw mm4, VP8_FILTER_SHIFT ;
+
+ packuswb mm3, mm4
+
+ movq [rdi], mm3 ; store the results in the destination
+
+%if ABI_IS_32BIT
+ add rsi, rdx ; next line
+ add rdi, dword ptr arg(5) ;dst_pitch ;
+%else
+ movsxd r8, dword ptr arg(5) ;dst_pitch
+ add rsi, rdx ; next line
+ add rdi, r8
+%endif
+ cmp rdi, rcx ;
+ jne .next_row_8x4
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void bilinear_predict4x4_mmx
+;(
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; int xoffset,
+; int yoffset,
+; unsigned char *dst_ptr,
+; int dst_pitch
+;)
+global sym(vp8_bilinear_predict4x4_mmx) PRIVATE
+sym(vp8_bilinear_predict4x4_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ;const short *HFilter = vp8_bilinear_filters_x86_8[xoffset];
+ ;const short *VFilter = vp8_bilinear_filters_x86_8[yoffset];
+
+ movsxd rax, dword ptr arg(2) ;xoffset
+ mov rdi, arg(4) ;dst_ptr ;
+
+ lea rcx, [GLOBAL(sym(vp8_bilinear_filters_x86_8))]
+ shl rax, 5
+
+ add rax, rcx ; HFilter
+ mov rsi, arg(0) ;src_ptr ;
+
+ movsxd rdx, dword ptr arg(5) ;ldst_pitch
+ movq mm1, [rax] ;
+
+ movq mm2, [rax+16] ;
+ movsxd rax, dword ptr arg(3) ;yoffset
+
+ pxor mm0, mm0 ;
+ shl rax, 5
+
+ add rax, rcx
+ lea rcx, [rdi+rdx*4] ;
+
+ movsxd rdx, dword ptr arg(1) ;src_pixels_per_line ;
+
+ ; get the first horizontal line done ;
+ movd mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
+
+ pmullw mm3, mm1 ;
+ movd mm5, [rsi+1] ;
+
+ punpcklbw mm5, mm0 ;
+ pmullw mm5, mm2 ;
+
+ paddw mm3, mm5 ;
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ movq mm7, mm3 ;
+ packuswb mm7, mm0 ;
+
+ add rsi, rdx ; next line
+.next_row_4x4:
+ movd mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
+
+ pmullw mm3, mm1 ;
+ movd mm5, [rsi+1] ;
+
+ punpcklbw mm5, mm0 ;
+ pmullw mm5, mm2 ;
+
+ paddw mm3, mm5 ;
+
+ movq mm5, mm7 ;
+ punpcklbw mm5, mm0 ;
+
+ pmullw mm5, [rax] ;
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+ movq mm7, mm3 ;
+
+ packuswb mm7, mm0 ;
+
+ pmullw mm3, [rax+16] ;
+ paddw mm3, mm5 ;
+
+
+ paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ packuswb mm3, mm0
+ movd [rdi], mm3 ; store the results in the destination
+
+%if ABI_IS_32BIT
+ add rsi, rdx ; next line
+ add rdi, dword ptr arg(5) ;dst_pitch ;
+%else
+ movsxd r8, dword ptr arg(5) ;dst_pitch ;
+ add rsi, rdx ; next line
+ add rdi, r8
+%endif
+
+ cmp rdi, rcx ;
+ jne .next_row_4x4
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+
+SECTION_RODATA
+align 16
+rd:
+ times 4 dw 0x40
+
+align 16
+global HIDDEN_DATA(sym(vp8_six_tap_mmx))
+sym(vp8_six_tap_mmx):
+ times 8 dw 0
+ times 8 dw 0
+ times 8 dw 128
+ times 8 dw 0
+ times 8 dw 0
+ times 8 dw 0
+
+ times 8 dw 0
+ times 8 dw -6
+ times 8 dw 123
+ times 8 dw 12
+ times 8 dw -1
+ times 8 dw 0
+
+ times 8 dw 2
+ times 8 dw -11
+ times 8 dw 108
+ times 8 dw 36
+ times 8 dw -8
+ times 8 dw 1
+
+ times 8 dw 0
+ times 8 dw -9
+ times 8 dw 93
+ times 8 dw 50
+ times 8 dw -6
+ times 8 dw 0
+
+ times 8 dw 3
+ times 8 dw -16
+ times 8 dw 77
+ times 8 dw 77
+ times 8 dw -16
+ times 8 dw 3
+
+ times 8 dw 0
+ times 8 dw -6
+ times 8 dw 50
+ times 8 dw 93
+ times 8 dw -9
+ times 8 dw 0
+
+ times 8 dw 1
+ times 8 dw -8
+ times 8 dw 36
+ times 8 dw 108
+ times 8 dw -11
+ times 8 dw 2
+
+ times 8 dw 0
+ times 8 dw -1
+ times 8 dw 12
+ times 8 dw 123
+ times 8 dw -6
+ times 8 dw 0
+
+
diff --git a/libvpx/vp8/common/x86/subpixel_sse2.asm b/libvpx/vp8/common/x86/subpixel_sse2.asm
new file mode 100644
index 0000000..69f8d10
--- /dev/null
+++ b/libvpx/vp8/common/x86/subpixel_sse2.asm
@@ -0,0 +1,1372 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+extern sym(vp8_bilinear_filters_x86_8)
+
+%define BLOCK_HEIGHT_WIDTH 4
+%define VP8_FILTER_WEIGHT 128
+%define VP8_FILTER_SHIFT 7
+
+
+;/************************************************************************************
+; Notes: filter_block1d_h6 applies a 6 tap filter horizontally to the input pixels. The
+; input pixel array has output_height rows. This routine assumes that output_height is an
+; even number. This function handles 8 pixels in horizontal direction, calculating ONE
+; rows each iteration to take advantage of the 128 bits operations.
+;*************************************************************************************/
+;void vp8_filter_block1d8_h6_sse2
+;(
+; unsigned char *src_ptr,
+; unsigned short *output_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned int pixel_step,
+; unsigned int output_height,
+; unsigned int output_width,
+; short *vp8_filter
+;)
+global sym(vp8_filter_block1d8_h6_sse2) PRIVATE
+sym(vp8_filter_block1d8_h6_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdx, arg(6) ;vp8_filter
+ mov rsi, arg(0) ;src_ptr
+
+ mov rdi, arg(1) ;output_ptr
+
+ movsxd rcx, dword ptr arg(4) ;output_height
+ movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(5) ;output_width
+%endif
+ pxor xmm0, xmm0 ; clear xmm0 for unpack
+
+.filter_block1d8_h6_rowloop:
+ movq xmm3, MMWORD PTR [rsi - 2]
+ movq xmm1, MMWORD PTR [rsi + 6]
+
+ prefetcht2 [rsi+rax-2]
+
+ pslldq xmm1, 8
+ por xmm1, xmm3
+
+ movdqa xmm4, xmm1
+ movdqa xmm5, xmm1
+
+ movdqa xmm6, xmm1
+ movdqa xmm7, xmm1
+
+ punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
+ psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
+
+ pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
+ punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
+
+ psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
+ pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
+
+
+ punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
+ psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
+
+ pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
+
+ punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
+ psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
+
+ pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
+
+ punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
+ psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
+
+
+ pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
+
+ punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
+ pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
+
+
+ paddsw xmm4, xmm7
+ paddsw xmm4, xmm5
+
+ paddsw xmm4, xmm3
+ paddsw xmm4, xmm6
+
+ paddsw xmm4, xmm1
+ paddsw xmm4, [GLOBAL(rd)]
+
+ psraw xmm4, 7
+
+ packuswb xmm4, xmm0
+ punpcklbw xmm4, xmm0
+
+ movdqa XMMWORD Ptr [rdi], xmm4
+ lea rsi, [rsi + rax]
+
+%if ABI_IS_32BIT
+ add rdi, DWORD Ptr arg(5) ;[output_width]
+%else
+ add rdi, r8
+%endif
+ dec rcx
+
+ jnz .filter_block1d8_h6_rowloop ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_filter_block1d16_h6_sse2
+;(
+; unsigned char *src_ptr,
+; unsigned short *output_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned int pixel_step,
+; unsigned int output_height,
+; unsigned int output_width,
+; short *vp8_filter
+;)
+;/************************************************************************************
+; Notes: filter_block1d_h6 applies a 6 tap filter horizontally to the input pixels. The
+; input pixel array has output_height rows. This routine assumes that output_height is an
+; even number. This function handles 8 pixels in horizontal direction, calculating ONE
+; rows each iteration to take advantage of the 128 bits operations.
+;*************************************************************************************/
+global sym(vp8_filter_block1d16_h6_sse2) PRIVATE
+sym(vp8_filter_block1d16_h6_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdx, arg(6) ;vp8_filter
+ mov rsi, arg(0) ;src_ptr
+
+ mov rdi, arg(1) ;output_ptr
+
+ movsxd rcx, dword ptr arg(4) ;output_height
+ movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(5) ;output_width
+%endif
+
+ pxor xmm0, xmm0 ; clear xmm0 for unpack
+
+.filter_block1d16_h6_sse2_rowloop:
+ movq xmm3, MMWORD PTR [rsi - 2]
+ movq xmm1, MMWORD PTR [rsi + 6]
+
+ movq xmm2, MMWORD PTR [rsi +14]
+ pslldq xmm2, 8
+
+ por xmm2, xmm1
+ prefetcht2 [rsi+rax-2]
+
+ pslldq xmm1, 8
+ por xmm1, xmm3
+
+ movdqa xmm4, xmm1
+ movdqa xmm5, xmm1
+
+ movdqa xmm6, xmm1
+ movdqa xmm7, xmm1
+
+ punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
+ psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
+
+ pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
+ punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
+
+ psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
+ pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
+
+
+ punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
+ psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
+
+ pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
+
+ punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
+ psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
+
+ pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
+
+ punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
+ psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
+
+
+ pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
+
+ punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
+ pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
+
+ paddsw xmm4, xmm7
+ paddsw xmm4, xmm5
+
+ paddsw xmm4, xmm3
+ paddsw xmm4, xmm6
+
+ paddsw xmm4, xmm1
+ paddsw xmm4, [GLOBAL(rd)]
+
+ psraw xmm4, 7
+
+ packuswb xmm4, xmm0
+ punpcklbw xmm4, xmm0
+
+ movdqa XMMWORD Ptr [rdi], xmm4
+
+ movdqa xmm3, xmm2
+ movdqa xmm4, xmm2
+
+ movdqa xmm5, xmm2
+ movdqa xmm6, xmm2
+
+ movdqa xmm7, xmm2
+
+ punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
+ psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
+
+ pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
+ punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
+
+ psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
+ pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
+
+
+ punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
+ psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
+
+ pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
+
+ punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
+ psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
+
+ pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
+
+ punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
+ psrldq xmm2, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
+
+ pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
+
+ punpcklbw xmm2, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
+ pmullw xmm2, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
+
+
+ paddsw xmm4, xmm7
+ paddsw xmm4, xmm5
+
+ paddsw xmm4, xmm3
+ paddsw xmm4, xmm6
+
+ paddsw xmm4, xmm2
+ paddsw xmm4, [GLOBAL(rd)]
+
+ psraw xmm4, 7
+
+ packuswb xmm4, xmm0
+ punpcklbw xmm4, xmm0
+
+ movdqa XMMWORD Ptr [rdi+16], xmm4
+
+ lea rsi, [rsi + rax]
+%if ABI_IS_32BIT
+ add rdi, DWORD Ptr arg(5) ;[output_width]
+%else
+ add rdi, r8
+%endif
+
+ dec rcx
+ jnz .filter_block1d16_h6_sse2_rowloop ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_filter_block1d8_v6_sse2
+;(
+; short *src_ptr,
+; unsigned char *output_ptr,
+; int dst_ptich,
+; unsigned int pixels_per_line,
+; unsigned int pixel_step,
+; unsigned int output_height,
+; unsigned int output_width,
+; short * vp8_filter
+;)
+;/************************************************************************************
+; Notes: filter_block1d8_v6 applies a 6 tap filter vertically to the input pixels. The
+; input pixel array has output_height rows.
+;*************************************************************************************/
+global sym(vp8_filter_block1d8_v6_sse2) PRIVATE
+sym(vp8_filter_block1d8_v6_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 8
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rax, arg(7) ;vp8_filter
+ movsxd rdx, dword ptr arg(3) ;pixels_per_line
+
+ mov rdi, arg(1) ;output_ptr
+ mov rsi, arg(0) ;src_ptr
+
+ sub rsi, rdx
+ sub rsi, rdx
+
+ movsxd rcx, DWORD PTR arg(5) ;[output_height]
+ pxor xmm0, xmm0 ; clear xmm0
+
+ movdqa xmm7, XMMWORD PTR [GLOBAL(rd)]
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(2) ; dst_ptich
+%endif
+
+.vp8_filter_block1d8_v6_sse2_loop:
+ movdqa xmm1, XMMWORD PTR [rsi]
+ pmullw xmm1, [rax]
+
+ movdqa xmm2, XMMWORD PTR [rsi + rdx]
+ pmullw xmm2, [rax + 16]
+
+ movdqa xmm3, XMMWORD PTR [rsi + rdx * 2]
+ pmullw xmm3, [rax + 32]
+
+ movdqa xmm5, XMMWORD PTR [rsi + rdx * 4]
+ pmullw xmm5, [rax + 64]
+
+ add rsi, rdx
+ movdqa xmm4, XMMWORD PTR [rsi + rdx * 2]
+
+ pmullw xmm4, [rax + 48]
+ movdqa xmm6, XMMWORD PTR [rsi + rdx * 4]
+
+ pmullw xmm6, [rax + 80]
+
+ paddsw xmm2, xmm5
+ paddsw xmm2, xmm3
+
+ paddsw xmm2, xmm1
+ paddsw xmm2, xmm4
+
+ paddsw xmm2, xmm6
+ paddsw xmm2, xmm7
+
+ psraw xmm2, 7
+ packuswb xmm2, xmm0 ; pack and saturate
+
+ movq QWORD PTR [rdi], xmm2 ; store the results in the destination
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(2) ;[dst_ptich]
+%else
+ add rdi, r8
+%endif
+ dec rcx ; decrement count
+ jnz .vp8_filter_block1d8_v6_sse2_loop ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_filter_block1d16_v6_sse2
+;(
+; unsigned short *src_ptr,
+; unsigned char *output_ptr,
+; int dst_ptich,
+; unsigned int pixels_per_line,
+; unsigned int pixel_step,
+; unsigned int output_height,
+; unsigned int output_width,
+; const short *vp8_filter
+;)
+;/************************************************************************************
+; Notes: filter_block1d16_v6 applies a 6 tap filter vertically to the input pixels. The
+; input pixel array has output_height rows.
+;*************************************************************************************/
+global sym(vp8_filter_block1d16_v6_sse2) PRIVATE
+sym(vp8_filter_block1d16_v6_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 8
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rax, arg(7) ;vp8_filter
+ movsxd rdx, dword ptr arg(3) ;pixels_per_line
+
+ mov rdi, arg(1) ;output_ptr
+ mov rsi, arg(0) ;src_ptr
+
+ sub rsi, rdx
+ sub rsi, rdx
+
+ movsxd rcx, DWORD PTR arg(5) ;[output_height]
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(2) ; dst_ptich
+%endif
+
+.vp8_filter_block1d16_v6_sse2_loop:
+; The order for adding 6-tap is 2 5 3 1 4 6. Read in data in that order.
+ movdqa xmm1, XMMWORD PTR [rsi + rdx] ; line 2
+ movdqa xmm2, XMMWORD PTR [rsi + rdx + 16]
+ pmullw xmm1, [rax + 16]
+ pmullw xmm2, [rax + 16]
+
+ movdqa xmm3, XMMWORD PTR [rsi + rdx * 4] ; line 5
+ movdqa xmm4, XMMWORD PTR [rsi + rdx * 4 + 16]
+ pmullw xmm3, [rax + 64]
+ pmullw xmm4, [rax + 64]
+
+ movdqa xmm5, XMMWORD PTR [rsi + rdx * 2] ; line 3
+ movdqa xmm6, XMMWORD PTR [rsi + rdx * 2 + 16]
+ pmullw xmm5, [rax + 32]
+ pmullw xmm6, [rax + 32]
+
+ movdqa xmm7, XMMWORD PTR [rsi] ; line 1
+ movdqa xmm0, XMMWORD PTR [rsi + 16]
+ pmullw xmm7, [rax]
+ pmullw xmm0, [rax]
+
+ paddsw xmm1, xmm3
+ paddsw xmm2, xmm4
+ paddsw xmm1, xmm5
+ paddsw xmm2, xmm6
+ paddsw xmm1, xmm7
+ paddsw xmm2, xmm0
+
+ add rsi, rdx
+
+ movdqa xmm3, XMMWORD PTR [rsi + rdx * 2] ; line 4
+ movdqa xmm4, XMMWORD PTR [rsi + rdx * 2 + 16]
+ pmullw xmm3, [rax + 48]
+ pmullw xmm4, [rax + 48]
+
+ movdqa xmm5, XMMWORD PTR [rsi + rdx * 4] ; line 6
+ movdqa xmm6, XMMWORD PTR [rsi + rdx * 4 + 16]
+ pmullw xmm5, [rax + 80]
+ pmullw xmm6, [rax + 80]
+
+ movdqa xmm7, XMMWORD PTR [GLOBAL(rd)]
+ pxor xmm0, xmm0 ; clear xmm0
+
+ paddsw xmm1, xmm3
+ paddsw xmm2, xmm4
+ paddsw xmm1, xmm5
+ paddsw xmm2, xmm6
+
+ paddsw xmm1, xmm7
+ paddsw xmm2, xmm7
+
+ psraw xmm1, 7
+ psraw xmm2, 7
+
+ packuswb xmm1, xmm2 ; pack and saturate
+ movdqa XMMWORD PTR [rdi], xmm1 ; store the results in the destination
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(2) ;[dst_ptich]
+%else
+ add rdi, r8
+%endif
+ dec rcx ; decrement count
+ jnz .vp8_filter_block1d16_v6_sse2_loop ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_filter_block1d8_h6_only_sse2
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned char *output_ptr,
+; int dst_ptich,
+; unsigned int output_height,
+; const short *vp8_filter
+;)
+; First-pass filter only when yoffset==0
+global sym(vp8_filter_block1d8_h6_only_sse2) PRIVATE
+sym(vp8_filter_block1d8_h6_only_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdx, arg(5) ;vp8_filter
+ mov rsi, arg(0) ;src_ptr
+
+ mov rdi, arg(2) ;output_ptr
+
+ movsxd rcx, dword ptr arg(4) ;output_height
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line ; Pitch for Source
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(3) ;dst_ptich
+%endif
+ pxor xmm0, xmm0 ; clear xmm0 for unpack
+
+.filter_block1d8_h6_only_rowloop:
+ movq xmm3, MMWORD PTR [rsi - 2]
+ movq xmm1, MMWORD PTR [rsi + 6]
+
+ prefetcht2 [rsi+rax-2]
+
+ pslldq xmm1, 8
+ por xmm1, xmm3
+
+ movdqa xmm4, xmm1
+ movdqa xmm5, xmm1
+
+ movdqa xmm6, xmm1
+ movdqa xmm7, xmm1
+
+ punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
+ psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
+
+ pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
+ punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
+
+ psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
+ pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
+
+
+ punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
+ psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
+
+ pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
+
+ punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
+ psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
+
+ pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
+
+ punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
+ psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
+
+
+ pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
+
+ punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
+ pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
+
+
+ paddsw xmm4, xmm7
+ paddsw xmm4, xmm5
+
+ paddsw xmm4, xmm3
+ paddsw xmm4, xmm6
+
+ paddsw xmm4, xmm1
+ paddsw xmm4, [GLOBAL(rd)]
+
+ psraw xmm4, 7
+
+ packuswb xmm4, xmm0
+
+ movq QWORD PTR [rdi], xmm4 ; store the results in the destination
+ lea rsi, [rsi + rax]
+
+%if ABI_IS_32BIT
+ add rdi, DWORD Ptr arg(3) ;dst_ptich
+%else
+ add rdi, r8
+%endif
+ dec rcx
+
+ jnz .filter_block1d8_h6_only_rowloop ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_filter_block1d16_h6_only_sse2
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned char *output_ptr,
+; int dst_ptich,
+; unsigned int output_height,
+; const short *vp8_filter
+;)
+; First-pass filter only when yoffset==0
+global sym(vp8_filter_block1d16_h6_only_sse2) PRIVATE
+sym(vp8_filter_block1d16_h6_only_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdx, arg(5) ;vp8_filter
+ mov rsi, arg(0) ;src_ptr
+
+ mov rdi, arg(2) ;output_ptr
+
+ movsxd rcx, dword ptr arg(4) ;output_height
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line ; Pitch for Source
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(3) ;dst_ptich
+%endif
+
+ pxor xmm0, xmm0 ; clear xmm0 for unpack
+
+.filter_block1d16_h6_only_sse2_rowloop:
+ movq xmm3, MMWORD PTR [rsi - 2]
+ movq xmm1, MMWORD PTR [rsi + 6]
+
+ movq xmm2, MMWORD PTR [rsi +14]
+ pslldq xmm2, 8
+
+ por xmm2, xmm1
+ prefetcht2 [rsi+rax-2]
+
+ pslldq xmm1, 8
+ por xmm1, xmm3
+
+ movdqa xmm4, xmm1
+ movdqa xmm5, xmm1
+
+ movdqa xmm6, xmm1
+ movdqa xmm7, xmm1
+
+ punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
+ psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
+
+ pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
+ punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
+
+ psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
+ pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
+
+ punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
+ psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
+
+ pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
+
+ punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
+ psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
+
+ pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
+
+ punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
+ psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
+
+ pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
+
+ punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
+ pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
+
+ paddsw xmm4, xmm7
+ paddsw xmm4, xmm5
+
+ paddsw xmm4, xmm3
+ paddsw xmm4, xmm6
+
+ paddsw xmm4, xmm1
+ paddsw xmm4, [GLOBAL(rd)]
+
+ psraw xmm4, 7
+
+ packuswb xmm4, xmm0 ; lower 8 bytes
+
+ movq QWORD Ptr [rdi], xmm4 ; store the results in the destination
+
+ movdqa xmm3, xmm2
+ movdqa xmm4, xmm2
+
+ movdqa xmm5, xmm2
+ movdqa xmm6, xmm2
+
+ movdqa xmm7, xmm2
+
+ punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
+ psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
+
+ pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
+ punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
+
+ psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
+ pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
+
+ punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
+ psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
+
+ pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
+
+ punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
+ psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
+
+ pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
+
+ punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
+ psrldq xmm2, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
+
+ pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
+
+ punpcklbw xmm2, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
+ pmullw xmm2, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
+
+ paddsw xmm4, xmm7
+ paddsw xmm4, xmm5
+
+ paddsw xmm4, xmm3
+ paddsw xmm4, xmm6
+
+ paddsw xmm4, xmm2
+ paddsw xmm4, [GLOBAL(rd)]
+
+ psraw xmm4, 7
+
+ packuswb xmm4, xmm0 ; higher 8 bytes
+
+ movq QWORD Ptr [rdi+8], xmm4 ; store the results in the destination
+
+ lea rsi, [rsi + rax]
+%if ABI_IS_32BIT
+ add rdi, DWORD Ptr arg(3) ;dst_ptich
+%else
+ add rdi, r8
+%endif
+
+ dec rcx
+ jnz .filter_block1d16_h6_only_sse2_rowloop ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_filter_block1d8_v6_only_sse2
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned char *output_ptr,
+; int dst_ptich,
+; unsigned int output_height,
+; const short *vp8_filter
+;)
+; Second-pass filter only when xoffset==0
+global sym(vp8_filter_block1d8_v6_only_sse2) PRIVATE
+sym(vp8_filter_block1d8_v6_only_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;output_ptr
+
+ movsxd rcx, dword ptr arg(4) ;output_height
+ movsxd rdx, dword ptr arg(1) ;src_pixels_per_line
+
+ mov rax, arg(5) ;vp8_filter
+
+ pxor xmm0, xmm0 ; clear xmm0
+
+ movdqa xmm7, XMMWORD PTR [GLOBAL(rd)]
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(3) ; dst_ptich
+%endif
+
+.vp8_filter_block1d8_v6_only_sse2_loop:
+ movq xmm1, MMWORD PTR [rsi]
+ movq xmm2, MMWORD PTR [rsi + rdx]
+ movq xmm3, MMWORD PTR [rsi + rdx * 2]
+ movq xmm5, MMWORD PTR [rsi + rdx * 4]
+ add rsi, rdx
+ movq xmm4, MMWORD PTR [rsi + rdx * 2]
+ movq xmm6, MMWORD PTR [rsi + rdx * 4]
+
+ punpcklbw xmm1, xmm0
+ pmullw xmm1, [rax]
+
+ punpcklbw xmm2, xmm0
+ pmullw xmm2, [rax + 16]
+
+ punpcklbw xmm3, xmm0
+ pmullw xmm3, [rax + 32]
+
+ punpcklbw xmm5, xmm0
+ pmullw xmm5, [rax + 64]
+
+ punpcklbw xmm4, xmm0
+ pmullw xmm4, [rax + 48]
+
+ punpcklbw xmm6, xmm0
+ pmullw xmm6, [rax + 80]
+
+ paddsw xmm2, xmm5
+ paddsw xmm2, xmm3
+
+ paddsw xmm2, xmm1
+ paddsw xmm2, xmm4
+
+ paddsw xmm2, xmm6
+ paddsw xmm2, xmm7
+
+ psraw xmm2, 7
+ packuswb xmm2, xmm0 ; pack and saturate
+
+ movq QWORD PTR [rdi], xmm2 ; store the results in the destination
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(3) ;[dst_ptich]
+%else
+ add rdi, r8
+%endif
+ dec rcx ; decrement count
+ jnz .vp8_filter_block1d8_v6_only_sse2_loop ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_unpack_block1d16_h6_sse2
+;(
+; unsigned char *src_ptr,
+; unsigned short *output_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned int output_height,
+; unsigned int output_width
+;)
+global sym(vp8_unpack_block1d16_h6_sse2) PRIVATE
+sym(vp8_unpack_block1d16_h6_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(1) ;output_ptr
+
+ movsxd rcx, dword ptr arg(3) ;output_height
+ movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source
+
+ pxor xmm0, xmm0 ; clear xmm0 for unpack
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(4) ;output_width ; Pitch for Source
+%endif
+
+.unpack_block1d16_h6_sse2_rowloop:
+ movq xmm1, MMWORD PTR [rsi] ; 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1 -2
+ movq xmm3, MMWORD PTR [rsi+8] ; make copy of xmm1
+
+ punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
+ punpcklbw xmm1, xmm0
+
+ movdqa XMMWORD Ptr [rdi], xmm1
+ movdqa XMMWORD Ptr [rdi + 16], xmm3
+
+ lea rsi, [rsi + rax]
+%if ABI_IS_32BIT
+ add rdi, DWORD Ptr arg(4) ;[output_width]
+%else
+ add rdi, r8
+%endif
+ dec rcx
+ jnz .unpack_block1d16_h6_sse2_rowloop ; next row
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_bilinear_predict16x16_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; int xoffset,
+; int yoffset,
+; unsigned char *dst_ptr,
+; int dst_pitch
+;)
+extern sym(vp8_bilinear_filters_x86_8)
+global sym(vp8_bilinear_predict16x16_sse2) PRIVATE
+sym(vp8_bilinear_predict16x16_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ;const short *HFilter = vp8_bilinear_filters_x86_8[xoffset]
+ ;const short *VFilter = vp8_bilinear_filters_x86_8[yoffset]
+
+ lea rcx, [GLOBAL(sym(vp8_bilinear_filters_x86_8))]
+ movsxd rax, dword ptr arg(2) ;xoffset
+
+ cmp rax, 0 ;skip first_pass filter if xoffset=0
+ je .b16x16_sp_only
+
+ shl rax, 5
+ add rax, rcx ;HFilter
+
+ mov rdi, arg(4) ;dst_ptr
+ mov rsi, arg(0) ;src_ptr
+ movsxd rdx, dword ptr arg(5) ;dst_pitch
+
+ movdqa xmm1, [rax]
+ movdqa xmm2, [rax+16]
+
+ movsxd rax, dword ptr arg(3) ;yoffset
+
+ cmp rax, 0 ;skip second_pass filter if yoffset=0
+ je .b16x16_fp_only
+
+ shl rax, 5
+ add rax, rcx ;VFilter
+
+ lea rcx, [rdi+rdx*8]
+ lea rcx, [rcx+rdx*8]
+ movsxd rdx, dword ptr arg(1) ;src_pixels_per_line
+
+ pxor xmm0, xmm0
+
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(5) ;dst_pitch
+%endif
+ ; get the first horizontal line done
+ movdqu xmm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ movdqa xmm4, xmm3 ; make a copy of current line
+
+ punpcklbw xmm3, xmm0 ; xx 00 01 02 03 04 05 06
+ punpckhbw xmm4, xmm0
+
+ pmullw xmm3, xmm1
+ pmullw xmm4, xmm1
+
+ movdqu xmm5, [rsi+1]
+ movdqa xmm6, xmm5
+
+ punpcklbw xmm5, xmm0
+ punpckhbw xmm6, xmm0
+
+ pmullw xmm5, xmm2
+ pmullw xmm6, xmm2
+
+ paddw xmm3, xmm5
+ paddw xmm4, xmm6
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw xmm4, [GLOBAL(rd)]
+ psraw xmm4, VP8_FILTER_SHIFT
+
+ movdqa xmm7, xmm3
+ packuswb xmm7, xmm4
+
+ add rsi, rdx ; next line
+.next_row:
+ movdqu xmm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ movdqa xmm4, xmm3 ; make a copy of current line
+
+ punpcklbw xmm3, xmm0 ; xx 00 01 02 03 04 05 06
+ punpckhbw xmm4, xmm0
+
+ pmullw xmm3, xmm1
+ pmullw xmm4, xmm1
+
+ movdqu xmm5, [rsi+1]
+ movdqa xmm6, xmm5
+
+ punpcklbw xmm5, xmm0
+ punpckhbw xmm6, xmm0
+
+ pmullw xmm5, xmm2
+ pmullw xmm6, xmm2
+
+ paddw xmm3, xmm5
+ paddw xmm4, xmm6
+
+ movdqa xmm5, xmm7
+ movdqa xmm6, xmm7
+
+ punpcklbw xmm5, xmm0
+ punpckhbw xmm6, xmm0
+
+ pmullw xmm5, [rax]
+ pmullw xmm6, [rax]
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw xmm4, [GLOBAL(rd)]
+ psraw xmm4, VP8_FILTER_SHIFT
+
+ movdqa xmm7, xmm3
+ packuswb xmm7, xmm4
+
+ pmullw xmm3, [rax+16]
+ pmullw xmm4, [rax+16]
+
+ paddw xmm3, xmm5
+ paddw xmm4, xmm6
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw xmm4, [GLOBAL(rd)]
+ psraw xmm4, VP8_FILTER_SHIFT
+
+ packuswb xmm3, xmm4
+ movdqa [rdi], xmm3 ; store the results in the destination
+
+ add rsi, rdx ; next line
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(5) ;dst_pitch
+%else
+ add rdi, r8
+%endif
+
+ cmp rdi, rcx
+ jne .next_row
+
+ jmp .done
+
+.b16x16_sp_only:
+ movsxd rax, dword ptr arg(3) ;yoffset
+ shl rax, 5
+ add rax, rcx ;VFilter
+
+ mov rdi, arg(4) ;dst_ptr
+ mov rsi, arg(0) ;src_ptr
+ movsxd rdx, dword ptr arg(5) ;dst_pitch
+
+ movdqa xmm1, [rax]
+ movdqa xmm2, [rax+16]
+
+ lea rcx, [rdi+rdx*8]
+ lea rcx, [rcx+rdx*8]
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line
+
+ pxor xmm0, xmm0
+
+ ; get the first horizontal line done
+ movdqu xmm7, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+
+ add rsi, rax ; next line
+.next_row_spo:
+ movdqu xmm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+
+ movdqa xmm5, xmm7
+ movdqa xmm6, xmm7
+
+ movdqa xmm4, xmm3 ; make a copy of current line
+ movdqa xmm7, xmm3
+
+ punpcklbw xmm5, xmm0
+ punpckhbw xmm6, xmm0
+ punpcklbw xmm3, xmm0 ; xx 00 01 02 03 04 05 06
+ punpckhbw xmm4, xmm0
+
+ pmullw xmm5, xmm1
+ pmullw xmm6, xmm1
+ pmullw xmm3, xmm2
+ pmullw xmm4, xmm2
+
+ paddw xmm3, xmm5
+ paddw xmm4, xmm6
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw xmm4, [GLOBAL(rd)]
+ psraw xmm4, VP8_FILTER_SHIFT
+
+ packuswb xmm3, xmm4
+ movdqa [rdi], xmm3 ; store the results in the destination
+
+ add rsi, rax ; next line
+ add rdi, rdx ;dst_pitch
+ cmp rdi, rcx
+ jne .next_row_spo
+
+ jmp .done
+
+.b16x16_fp_only:
+ lea rcx, [rdi+rdx*8]
+ lea rcx, [rcx+rdx*8]
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line
+ pxor xmm0, xmm0
+
+.next_row_fpo:
+ movdqu xmm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
+ movdqa xmm4, xmm3 ; make a copy of current line
+
+ punpcklbw xmm3, xmm0 ; xx 00 01 02 03 04 05 06
+ punpckhbw xmm4, xmm0
+
+ pmullw xmm3, xmm1
+ pmullw xmm4, xmm1
+
+ movdqu xmm5, [rsi+1]
+ movdqa xmm6, xmm5
+
+ punpcklbw xmm5, xmm0
+ punpckhbw xmm6, xmm0
+
+ pmullw xmm5, xmm2
+ pmullw xmm6, xmm2
+
+ paddw xmm3, xmm5
+ paddw xmm4, xmm6
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw xmm4, [GLOBAL(rd)]
+ psraw xmm4, VP8_FILTER_SHIFT
+
+ packuswb xmm3, xmm4
+ movdqa [rdi], xmm3 ; store the results in the destination
+
+ add rsi, rax ; next line
+ add rdi, rdx ; dst_pitch
+ cmp rdi, rcx
+ jne .next_row_fpo
+
+.done:
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_bilinear_predict8x8_sse2
+;(
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; int xoffset,
+; int yoffset,
+; unsigned char *dst_ptr,
+; int dst_pitch
+;)
+global sym(vp8_bilinear_predict8x8_sse2) PRIVATE
+sym(vp8_bilinear_predict8x8_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 144 ; reserve 144 bytes
+
+ ;const short *HFilter = vp8_bilinear_filters_x86_8[xoffset]
+ ;const short *VFilter = vp8_bilinear_filters_x86_8[yoffset]
+ lea rcx, [GLOBAL(sym(vp8_bilinear_filters_x86_8))]
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rdx, dword ptr arg(1) ;src_pixels_per_line
+
+ ;Read 9-line unaligned data in and put them on stack. This gives a big
+ ;performance boost.
+ movdqu xmm0, [rsi]
+ lea rax, [rdx + rdx*2]
+ movdqu xmm1, [rsi+rdx]
+ movdqu xmm2, [rsi+rdx*2]
+ add rsi, rax
+ movdqu xmm3, [rsi]
+ movdqu xmm4, [rsi+rdx]
+ movdqu xmm5, [rsi+rdx*2]
+ add rsi, rax
+ movdqu xmm6, [rsi]
+ movdqu xmm7, [rsi+rdx]
+
+ movdqa XMMWORD PTR [rsp], xmm0
+
+ movdqu xmm0, [rsi+rdx*2]
+
+ movdqa XMMWORD PTR [rsp+16], xmm1
+ movdqa XMMWORD PTR [rsp+32], xmm2
+ movdqa XMMWORD PTR [rsp+48], xmm3
+ movdqa XMMWORD PTR [rsp+64], xmm4
+ movdqa XMMWORD PTR [rsp+80], xmm5
+ movdqa XMMWORD PTR [rsp+96], xmm6
+ movdqa XMMWORD PTR [rsp+112], xmm7
+ movdqa XMMWORD PTR [rsp+128], xmm0
+
+ movsxd rax, dword ptr arg(2) ;xoffset
+ shl rax, 5
+ add rax, rcx ;HFilter
+
+ mov rdi, arg(4) ;dst_ptr
+ movsxd rdx, dword ptr arg(5) ;dst_pitch
+
+ movdqa xmm1, [rax]
+ movdqa xmm2, [rax+16]
+
+ movsxd rax, dword ptr arg(3) ;yoffset
+ shl rax, 5
+ add rax, rcx ;VFilter
+
+ lea rcx, [rdi+rdx*8]
+
+ movdqa xmm5, [rax]
+ movdqa xmm6, [rax+16]
+
+ pxor xmm0, xmm0
+
+ ; get the first horizontal line done
+ movdqa xmm3, XMMWORD PTR [rsp]
+ movdqa xmm4, xmm3 ; make a copy of current line
+ psrldq xmm4, 1
+
+ punpcklbw xmm3, xmm0 ; 00 01 02 03 04 05 06 07
+ punpcklbw xmm4, xmm0 ; 01 02 03 04 05 06 07 08
+
+ pmullw xmm3, xmm1
+ pmullw xmm4, xmm2
+
+ paddw xmm3, xmm4
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ movdqa xmm7, xmm3
+ add rsp, 16 ; next line
+.next_row8x8:
+ movdqa xmm3, XMMWORD PTR [rsp] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
+ movdqa xmm4, xmm3 ; make a copy of current line
+ psrldq xmm4, 1
+
+ punpcklbw xmm3, xmm0 ; 00 01 02 03 04 05 06 07
+ punpcklbw xmm4, xmm0 ; 01 02 03 04 05 06 07 08
+
+ pmullw xmm3, xmm1
+ pmullw xmm4, xmm2
+
+ paddw xmm3, xmm4
+ pmullw xmm7, xmm5
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ movdqa xmm4, xmm3
+
+ pmullw xmm3, xmm6
+ paddw xmm3, xmm7
+
+ movdqa xmm7, xmm4
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ packuswb xmm3, xmm0
+ movq [rdi], xmm3 ; store the results in the destination
+
+ add rsp, 16 ; next line
+ add rdi, rdx
+
+ cmp rdi, rcx
+ jne .next_row8x8
+
+ ;add rsp, 144
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+SECTION_RODATA
+align 16
+rd:
+ times 8 dw 0x40
diff --git a/libvpx/vp8/common/x86/subpixel_ssse3.asm b/libvpx/vp8/common/x86/subpixel_ssse3.asm
new file mode 100644
index 0000000..13bcaf6
--- /dev/null
+++ b/libvpx/vp8/common/x86/subpixel_ssse3.asm
@@ -0,0 +1,1507 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%define BLOCK_HEIGHT_WIDTH 4
+%define VP8_FILTER_WEIGHT 128
+%define VP8_FILTER_SHIFT 7
+
+
+;/************************************************************************************
+; Notes: filter_block1d_h6 applies a 6 tap filter horizontally to the input pixels. The
+; input pixel array has output_height rows. This routine assumes that output_height is an
+; even number. This function handles 8 pixels in horizontal direction, calculating ONE
+; rows each iteration to take advantage of the 128 bits operations.
+;
+; This is an implementation of some of the SSE optimizations first seen in ffvp8
+;
+;*************************************************************************************/
+;void vp8_filter_block1d8_h6_ssse3
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned char *output_ptr,
+; unsigned int output_pitch,
+; unsigned int output_height,
+; unsigned int vp8_filter_index
+;)
+global sym(vp8_filter_block1d8_h6_ssse3) PRIVATE
+sym(vp8_filter_block1d8_h6_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movsxd rdx, DWORD PTR arg(5) ;table index
+ xor rsi, rsi
+ shl rdx, 4
+
+ movdqa xmm7, [GLOBAL(rd)]
+
+ lea rax, [GLOBAL(k0_k5)]
+ add rax, rdx
+ mov rdi, arg(2) ;output_ptr
+
+ cmp esi, DWORD PTR [rax]
+ je vp8_filter_block1d8_h4_ssse3
+
+ movdqa xmm4, XMMWORD PTR [rax] ;k0_k5
+ movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line
+ movsxd rcx, dword ptr arg(4) ;output_height
+
+ movsxd rdx, dword ptr arg(3) ;output_pitch
+
+ sub rdi, rdx
+;xmm3 free
+.filter_block1d8_h6_rowloop_ssse3:
+ movq xmm0, MMWORD PTR [rsi - 2] ; -2 -1 0 1 2 3 4 5
+
+ movq xmm2, MMWORD PTR [rsi + 3] ; 3 4 5 6 7 8 9 10
+
+ punpcklbw xmm0, xmm2 ; -2 3 -1 4 0 5 1 6 2 7 3 8 4 9 5 10
+
+ movdqa xmm1, xmm0
+ pmaddubsw xmm0, xmm4
+
+ movdqa xmm2, xmm1
+ pshufb xmm1, [GLOBAL(shuf2bfrom1)]
+
+ pshufb xmm2, [GLOBAL(shuf3bfrom1)]
+ pmaddubsw xmm1, xmm5
+
+ lea rdi, [rdi + rdx]
+ pmaddubsw xmm2, xmm6
+
+ lea rsi, [rsi + rax]
+ dec rcx
+
+ paddsw xmm0, xmm1
+ paddsw xmm2, xmm7
+
+ paddsw xmm0, xmm2
+
+ psraw xmm0, 7
+
+ packuswb xmm0, xmm0
+
+ movq MMWORD Ptr [rdi], xmm0
+ jnz .filter_block1d8_h6_rowloop_ssse3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+vp8_filter_block1d8_h4_ssse3:
+ movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
+
+ movdqa xmm3, XMMWORD PTR [GLOBAL(shuf2bfrom1)]
+ movdqa xmm4, XMMWORD PTR [GLOBAL(shuf3bfrom1)]
+
+ mov rsi, arg(0) ;src_ptr
+
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line
+ movsxd rcx, dword ptr arg(4) ;output_height
+
+ movsxd rdx, dword ptr arg(3) ;output_pitch
+
+ sub rdi, rdx
+
+.filter_block1d8_h4_rowloop_ssse3:
+ movq xmm0, MMWORD PTR [rsi - 2] ; -2 -1 0 1 2 3 4 5
+
+ movq xmm1, MMWORD PTR [rsi + 3] ; 3 4 5 6 7 8 9 10
+
+ punpcklbw xmm0, xmm1 ; -2 3 -1 4 0 5 1 6 2 7 3 8 4 9 5 10
+
+ movdqa xmm2, xmm0
+ pshufb xmm0, xmm3
+
+ pshufb xmm2, xmm4
+ pmaddubsw xmm0, xmm5
+
+ lea rdi, [rdi + rdx]
+ pmaddubsw xmm2, xmm6
+
+ lea rsi, [rsi + rax]
+ dec rcx
+
+ paddsw xmm0, xmm7
+
+ paddsw xmm0, xmm2
+
+ psraw xmm0, 7
+
+ packuswb xmm0, xmm0
+
+ movq MMWORD Ptr [rdi], xmm0
+
+ jnz .filter_block1d8_h4_rowloop_ssse3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+;void vp8_filter_block1d16_h6_ssse3
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned char *output_ptr,
+; unsigned int output_pitch,
+; unsigned int output_height,
+; unsigned int vp8_filter_index
+;)
+global sym(vp8_filter_block1d16_h6_ssse3) PRIVATE
+sym(vp8_filter_block1d16_h6_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movsxd rdx, DWORD PTR arg(5) ;table index
+ xor rsi, rsi
+ shl rdx, 4 ;
+
+ lea rax, [GLOBAL(k0_k5)]
+ add rax, rdx
+
+ mov rdi, arg(2) ;output_ptr
+
+ mov rsi, arg(0) ;src_ptr
+
+ movdqa xmm4, XMMWORD PTR [rax] ;k0_k5
+ movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
+
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line
+ movsxd rcx, dword ptr arg(4) ;output_height
+ movsxd rdx, dword ptr arg(3) ;output_pitch
+
+.filter_block1d16_h6_rowloop_ssse3:
+ movq xmm0, MMWORD PTR [rsi - 2] ; -2 -1 0 1 2 3 4 5
+
+ movq xmm3, MMWORD PTR [rsi + 3] ; 3 4 5 6 7 8 9 10
+
+ punpcklbw xmm0, xmm3 ; -2 3 -1 4 0 5 1 6 2 7 3 8 4 9 5 10
+
+ movdqa xmm1, xmm0
+ pmaddubsw xmm0, xmm4
+
+ movdqa xmm2, xmm1
+ pshufb xmm1, [GLOBAL(shuf2bfrom1)]
+
+ pshufb xmm2, [GLOBAL(shuf3bfrom1)]
+ movq xmm3, MMWORD PTR [rsi + 6]
+
+ pmaddubsw xmm1, xmm5
+ movq xmm7, MMWORD PTR [rsi + 11]
+
+ pmaddubsw xmm2, xmm6
+ punpcklbw xmm3, xmm7
+
+ paddsw xmm0, xmm1
+ movdqa xmm1, xmm3
+
+ pmaddubsw xmm3, xmm4
+ paddsw xmm0, xmm2
+
+ movdqa xmm2, xmm1
+ paddsw xmm0, [GLOBAL(rd)]
+
+ pshufb xmm1, [GLOBAL(shuf2bfrom1)]
+ pshufb xmm2, [GLOBAL(shuf3bfrom1)]
+
+ psraw xmm0, 7
+ pmaddubsw xmm1, xmm5
+
+ pmaddubsw xmm2, xmm6
+ packuswb xmm0, xmm0
+
+ lea rsi, [rsi + rax]
+ paddsw xmm3, xmm1
+
+ paddsw xmm3, xmm2
+
+ paddsw xmm3, [GLOBAL(rd)]
+
+ psraw xmm3, 7
+
+ packuswb xmm3, xmm3
+
+ punpcklqdq xmm0, xmm3
+
+ movdqa XMMWORD Ptr [rdi], xmm0
+
+ lea rdi, [rdi + rdx]
+ dec rcx
+ jnz .filter_block1d16_h6_rowloop_ssse3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_filter_block1d4_h6_ssse3
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pixels_per_line,
+; unsigned char *output_ptr,
+; unsigned int output_pitch,
+; unsigned int output_height,
+; unsigned int vp8_filter_index
+;)
+global sym(vp8_filter_block1d4_h6_ssse3) PRIVATE
+sym(vp8_filter_block1d4_h6_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movsxd rdx, DWORD PTR arg(5) ;table index
+ xor rsi, rsi
+ shl rdx, 4 ;
+
+ lea rax, [GLOBAL(k0_k5)]
+ add rax, rdx
+ movdqa xmm7, [GLOBAL(rd)]
+
+ cmp esi, DWORD PTR [rax]
+ je .vp8_filter_block1d4_h4_ssse3
+
+ movdqa xmm4, XMMWORD PTR [rax] ;k0_k5
+ movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;output_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line
+ movsxd rcx, dword ptr arg(4) ;output_height
+
+ movsxd rdx, dword ptr arg(3) ;output_pitch
+
+;xmm3 free
+.filter_block1d4_h6_rowloop_ssse3:
+ movdqu xmm0, XMMWORD PTR [rsi - 2]
+
+ movdqa xmm1, xmm0
+ pshufb xmm0, [GLOBAL(shuf1b)]
+
+ movdqa xmm2, xmm1
+ pshufb xmm1, [GLOBAL(shuf2b)]
+ pmaddubsw xmm0, xmm4
+ pshufb xmm2, [GLOBAL(shuf3b)]
+ pmaddubsw xmm1, xmm5
+
+;--
+ pmaddubsw xmm2, xmm6
+
+ lea rsi, [rsi + rax]
+;--
+ paddsw xmm0, xmm1
+ paddsw xmm0, xmm7
+ pxor xmm1, xmm1
+ paddsw xmm0, xmm2
+ psraw xmm0, 7
+ packuswb xmm0, xmm0
+
+ movd DWORD PTR [rdi], xmm0
+
+ add rdi, rdx
+ dec rcx
+ jnz .filter_block1d4_h6_rowloop_ssse3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+.vp8_filter_block1d4_h4_ssse3:
+ movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
+ movdqa xmm0, XMMWORD PTR [GLOBAL(shuf2b)]
+ movdqa xmm3, XMMWORD PTR [GLOBAL(shuf3b)]
+
+ mov rsi, arg(0) ;src_ptr
+ mov rdi, arg(2) ;output_ptr
+ movsxd rax, dword ptr arg(1) ;src_pixels_per_line
+ movsxd rcx, dword ptr arg(4) ;output_height
+
+ movsxd rdx, dword ptr arg(3) ;output_pitch
+
+.filter_block1d4_h4_rowloop_ssse3:
+ movdqu xmm1, XMMWORD PTR [rsi - 2]
+
+ movdqa xmm2, xmm1
+ pshufb xmm1, xmm0 ;;[GLOBAL(shuf2b)]
+ pshufb xmm2, xmm3 ;;[GLOBAL(shuf3b)]
+ pmaddubsw xmm1, xmm5
+
+;--
+ pmaddubsw xmm2, xmm6
+
+ lea rsi, [rsi + rax]
+;--
+ paddsw xmm1, xmm7
+ paddsw xmm1, xmm2
+ psraw xmm1, 7
+ packuswb xmm1, xmm1
+
+ movd DWORD PTR [rdi], xmm1
+
+ add rdi, rdx
+ dec rcx
+ jnz .filter_block1d4_h4_rowloop_ssse3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+
+;void vp8_filter_block1d16_v6_ssse3
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pitch,
+; unsigned char *output_ptr,
+; unsigned int out_pitch,
+; unsigned int output_height,
+; unsigned int vp8_filter_index
+;)
+global sym(vp8_filter_block1d16_v6_ssse3) PRIVATE
+sym(vp8_filter_block1d16_v6_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movsxd rdx, DWORD PTR arg(5) ;table index
+ xor rsi, rsi
+ shl rdx, 4 ;
+
+ lea rax, [GLOBAL(k0_k5)]
+ add rax, rdx
+
+ cmp esi, DWORD PTR [rax]
+ je .vp8_filter_block1d16_v4_ssse3
+
+ movdqa xmm5, XMMWORD PTR [rax] ;k0_k5
+ movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rdx, DWORD PTR arg(1) ;pixels_per_line
+ mov rdi, arg(2) ;output_ptr
+
+%if ABI_IS_32BIT=0
+ movsxd r8, DWORD PTR arg(3) ;out_pitch
+%endif
+ mov rax, rsi
+ movsxd rcx, DWORD PTR arg(4) ;output_height
+ add rax, rdx
+
+
+.vp8_filter_block1d16_v6_ssse3_loop:
+ movq xmm1, MMWORD PTR [rsi] ;A
+ movq xmm2, MMWORD PTR [rsi + rdx] ;B
+ movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C
+ movq xmm4, MMWORD PTR [rax + rdx * 2] ;D
+ movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E
+
+ punpcklbw xmm2, xmm4 ;B D
+ punpcklbw xmm3, xmm0 ;C E
+
+ movq xmm0, MMWORD PTR [rax + rdx * 4] ;F
+
+ pmaddubsw xmm3, xmm6
+ punpcklbw xmm1, xmm0 ;A F
+ pmaddubsw xmm2, xmm7
+ pmaddubsw xmm1, xmm5
+
+ paddsw xmm2, xmm3
+ paddsw xmm2, xmm1
+ paddsw xmm2, [GLOBAL(rd)]
+ psraw xmm2, 7
+ packuswb xmm2, xmm2
+
+ movq MMWORD PTR [rdi], xmm2 ;store the results
+
+ movq xmm1, MMWORD PTR [rsi + 8] ;A
+ movq xmm2, MMWORD PTR [rsi + rdx + 8] ;B
+ movq xmm3, MMWORD PTR [rsi + rdx * 2 + 8] ;C
+ movq xmm4, MMWORD PTR [rax + rdx * 2 + 8] ;D
+ movq xmm0, MMWORD PTR [rsi + rdx * 4 + 8] ;E
+
+ punpcklbw xmm2, xmm4 ;B D
+ punpcklbw xmm3, xmm0 ;C E
+
+ movq xmm0, MMWORD PTR [rax + rdx * 4 + 8] ;F
+ pmaddubsw xmm3, xmm6
+ punpcklbw xmm1, xmm0 ;A F
+ pmaddubsw xmm2, xmm7
+ pmaddubsw xmm1, xmm5
+
+ add rsi, rdx
+ add rax, rdx
+;--
+;--
+ paddsw xmm2, xmm3
+ paddsw xmm2, xmm1
+ paddsw xmm2, [GLOBAL(rd)]
+ psraw xmm2, 7
+ packuswb xmm2, xmm2
+
+ movq MMWORD PTR [rdi+8], xmm2
+
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(3) ;out_pitch
+%else
+ add rdi, r8
+%endif
+ dec rcx
+ jnz .vp8_filter_block1d16_v6_ssse3_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+.vp8_filter_block1d16_v4_ssse3:
+ movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rdx, DWORD PTR arg(1) ;pixels_per_line
+ mov rdi, arg(2) ;output_ptr
+
+%if ABI_IS_32BIT=0
+ movsxd r8, DWORD PTR arg(3) ;out_pitch
+%endif
+ mov rax, rsi
+ movsxd rcx, DWORD PTR arg(4) ;output_height
+ add rax, rdx
+
+.vp8_filter_block1d16_v4_ssse3_loop:
+ movq xmm2, MMWORD PTR [rsi + rdx] ;B
+ movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C
+ movq xmm4, MMWORD PTR [rax + rdx * 2] ;D
+ movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E
+
+ punpcklbw xmm2, xmm4 ;B D
+ punpcklbw xmm3, xmm0 ;C E
+
+ pmaddubsw xmm3, xmm6
+ pmaddubsw xmm2, xmm7
+ movq xmm5, MMWORD PTR [rsi + rdx + 8] ;B
+ movq xmm1, MMWORD PTR [rsi + rdx * 2 + 8] ;C
+ movq xmm4, MMWORD PTR [rax + rdx * 2 + 8] ;D
+ movq xmm0, MMWORD PTR [rsi + rdx * 4 + 8] ;E
+
+ paddsw xmm2, [GLOBAL(rd)]
+ paddsw xmm2, xmm3
+ psraw xmm2, 7
+ packuswb xmm2, xmm2
+
+ punpcklbw xmm5, xmm4 ;B D
+ punpcklbw xmm1, xmm0 ;C E
+
+ pmaddubsw xmm1, xmm6
+ pmaddubsw xmm5, xmm7
+
+ movdqa xmm4, [GLOBAL(rd)]
+ add rsi, rdx
+ add rax, rdx
+;--
+;--
+ paddsw xmm5, xmm1
+ paddsw xmm5, xmm4
+ psraw xmm5, 7
+ packuswb xmm5, xmm5
+
+ punpcklqdq xmm2, xmm5
+
+ movdqa XMMWORD PTR [rdi], xmm2
+
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(3) ;out_pitch
+%else
+ add rdi, r8
+%endif
+ dec rcx
+ jnz .vp8_filter_block1d16_v4_ssse3_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_filter_block1d8_v6_ssse3
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pitch,
+; unsigned char *output_ptr,
+; unsigned int out_pitch,
+; unsigned int output_height,
+; unsigned int vp8_filter_index
+;)
+global sym(vp8_filter_block1d8_v6_ssse3) PRIVATE
+sym(vp8_filter_block1d8_v6_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movsxd rdx, DWORD PTR arg(5) ;table index
+ xor rsi, rsi
+ shl rdx, 4 ;
+
+ lea rax, [GLOBAL(k0_k5)]
+ add rax, rdx
+
+ movsxd rdx, DWORD PTR arg(1) ;pixels_per_line
+ mov rdi, arg(2) ;output_ptr
+%if ABI_IS_32BIT=0
+ movsxd r8, DWORD PTR arg(3) ; out_pitch
+%endif
+ movsxd rcx, DWORD PTR arg(4) ;[output_height]
+
+ cmp esi, DWORD PTR [rax]
+ je .vp8_filter_block1d8_v4_ssse3
+
+ movdqa xmm5, XMMWORD PTR [rax] ;k0_k5
+ movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3
+
+ mov rsi, arg(0) ;src_ptr
+
+ mov rax, rsi
+ add rax, rdx
+
+.vp8_filter_block1d8_v6_ssse3_loop:
+ movq xmm1, MMWORD PTR [rsi] ;A
+ movq xmm2, MMWORD PTR [rsi + rdx] ;B
+ movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C
+ movq xmm4, MMWORD PTR [rax + rdx * 2] ;D
+ movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E
+
+ punpcklbw xmm2, xmm4 ;B D
+ punpcklbw xmm3, xmm0 ;C E
+
+ movq xmm0, MMWORD PTR [rax + rdx * 4] ;F
+ movdqa xmm4, [GLOBAL(rd)]
+
+ pmaddubsw xmm3, xmm6
+ punpcklbw xmm1, xmm0 ;A F
+ pmaddubsw xmm2, xmm7
+ pmaddubsw xmm1, xmm5
+ add rsi, rdx
+ add rax, rdx
+;--
+;--
+ paddsw xmm2, xmm3
+ paddsw xmm2, xmm1
+ paddsw xmm2, xmm4
+ psraw xmm2, 7
+ packuswb xmm2, xmm2
+
+ movq MMWORD PTR [rdi], xmm2
+
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(3) ;[out_pitch]
+%else
+ add rdi, r8
+%endif
+ dec rcx
+ jnz .vp8_filter_block1d8_v6_ssse3_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+.vp8_filter_block1d8_v4_ssse3:
+ movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4
+ movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3
+ movdqa xmm5, [GLOBAL(rd)]
+
+ mov rsi, arg(0) ;src_ptr
+
+ mov rax, rsi
+ add rax, rdx
+
+.vp8_filter_block1d8_v4_ssse3_loop:
+ movq xmm2, MMWORD PTR [rsi + rdx] ;B
+ movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C
+ movq xmm4, MMWORD PTR [rax + rdx * 2] ;D
+ movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E
+
+ punpcklbw xmm2, xmm4 ;B D
+ punpcklbw xmm3, xmm0 ;C E
+
+ pmaddubsw xmm3, xmm6
+ pmaddubsw xmm2, xmm7
+ add rsi, rdx
+ add rax, rdx
+;--
+;--
+ paddsw xmm2, xmm3
+ paddsw xmm2, xmm5
+ psraw xmm2, 7
+ packuswb xmm2, xmm2
+
+ movq MMWORD PTR [rdi], xmm2
+
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(3) ;[out_pitch]
+%else
+ add rdi, r8
+%endif
+ dec rcx
+ jnz .vp8_filter_block1d8_v4_ssse3_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+;void vp8_filter_block1d4_v6_ssse3
+;(
+; unsigned char *src_ptr,
+; unsigned int src_pitch,
+; unsigned char *output_ptr,
+; unsigned int out_pitch,
+; unsigned int output_height,
+; unsigned int vp8_filter_index
+;)
+global sym(vp8_filter_block1d4_v6_ssse3) PRIVATE
+sym(vp8_filter_block1d4_v6_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movsxd rdx, DWORD PTR arg(5) ;table index
+ xor rsi, rsi
+ shl rdx, 4 ;
+
+ lea rax, [GLOBAL(k0_k5)]
+ add rax, rdx
+
+ movsxd rdx, DWORD PTR arg(1) ;pixels_per_line
+ mov rdi, arg(2) ;output_ptr
+%if ABI_IS_32BIT=0
+ movsxd r8, DWORD PTR arg(3) ; out_pitch
+%endif
+ movsxd rcx, DWORD PTR arg(4) ;[output_height]
+
+ cmp esi, DWORD PTR [rax]
+ je .vp8_filter_block1d4_v4_ssse3
+
+ movq mm5, MMWORD PTR [rax] ;k0_k5
+ movq mm6, MMWORD PTR [rax+256] ;k2_k4
+ movq mm7, MMWORD PTR [rax+128] ;k1_k3
+
+ mov rsi, arg(0) ;src_ptr
+
+ mov rax, rsi
+ add rax, rdx
+
+.vp8_filter_block1d4_v6_ssse3_loop:
+ movd mm1, DWORD PTR [rsi] ;A
+ movd mm2, DWORD PTR [rsi + rdx] ;B
+ movd mm3, DWORD PTR [rsi + rdx * 2] ;C
+ movd mm4, DWORD PTR [rax + rdx * 2] ;D
+ movd mm0, DWORD PTR [rsi + rdx * 4] ;E
+
+ punpcklbw mm2, mm4 ;B D
+ punpcklbw mm3, mm0 ;C E
+
+ movd mm0, DWORD PTR [rax + rdx * 4] ;F
+
+ movq mm4, [GLOBAL(rd)]
+
+ pmaddubsw mm3, mm6
+ punpcklbw mm1, mm0 ;A F
+ pmaddubsw mm2, mm7
+ pmaddubsw mm1, mm5
+ add rsi, rdx
+ add rax, rdx
+;--
+;--
+ paddsw mm2, mm3
+ paddsw mm2, mm1
+ paddsw mm2, mm4
+ psraw mm2, 7
+ packuswb mm2, mm2
+
+ movd DWORD PTR [rdi], mm2
+
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(3) ;[out_pitch]
+%else
+ add rdi, r8
+%endif
+ dec rcx
+ jnz .vp8_filter_block1d4_v6_ssse3_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+.vp8_filter_block1d4_v4_ssse3:
+ movq mm6, MMWORD PTR [rax+256] ;k2_k4
+ movq mm7, MMWORD PTR [rax+128] ;k1_k3
+ movq mm5, MMWORD PTR [GLOBAL(rd)]
+
+ mov rsi, arg(0) ;src_ptr
+
+ mov rax, rsi
+ add rax, rdx
+
+.vp8_filter_block1d4_v4_ssse3_loop:
+ movd mm2, DWORD PTR [rsi + rdx] ;B
+ movd mm3, DWORD PTR [rsi + rdx * 2] ;C
+ movd mm4, DWORD PTR [rax + rdx * 2] ;D
+ movd mm0, DWORD PTR [rsi + rdx * 4] ;E
+
+ punpcklbw mm2, mm4 ;B D
+ punpcklbw mm3, mm0 ;C E
+
+ pmaddubsw mm3, mm6
+ pmaddubsw mm2, mm7
+ add rsi, rdx
+ add rax, rdx
+;--
+;--
+ paddsw mm2, mm3
+ paddsw mm2, mm5
+ psraw mm2, 7
+ packuswb mm2, mm2
+
+ movd DWORD PTR [rdi], mm2
+
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(3) ;[out_pitch]
+%else
+ add rdi, r8
+%endif
+ dec rcx
+ jnz .vp8_filter_block1d4_v4_ssse3_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_bilinear_predict16x16_ssse3
+;(
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; int xoffset,
+; int yoffset,
+; unsigned char *dst_ptr,
+; int dst_pitch
+;)
+global sym(vp8_bilinear_predict16x16_ssse3) PRIVATE
+sym(vp8_bilinear_predict16x16_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ lea rcx, [GLOBAL(vp8_bilinear_filters_ssse3)]
+ movsxd rax, dword ptr arg(2) ; xoffset
+
+ cmp rax, 0 ; skip first_pass filter if xoffset=0
+ je .b16x16_sp_only
+
+ shl rax, 4
+ lea rax, [rax + rcx] ; HFilter
+
+ mov rdi, arg(4) ; dst_ptr
+ mov rsi, arg(0) ; src_ptr
+ movsxd rdx, dword ptr arg(5) ; dst_pitch
+
+ movdqa xmm1, [rax]
+
+ movsxd rax, dword ptr arg(3) ; yoffset
+
+ cmp rax, 0 ; skip second_pass filter if yoffset=0
+ je .b16x16_fp_only
+
+ shl rax, 4
+ lea rax, [rax + rcx] ; VFilter
+
+ lea rcx, [rdi+rdx*8]
+ lea rcx, [rcx+rdx*8]
+ movsxd rdx, dword ptr arg(1) ; src_pixels_per_line
+
+ movdqa xmm2, [rax]
+
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(5) ; dst_pitch
+%endif
+ movq xmm3, [rsi] ; 00 01 02 03 04 05 06 07
+ movq xmm5, [rsi+1] ; 01 02 03 04 05 06 07 08
+
+ punpcklbw xmm3, xmm5 ; 00 01 01 02 02 03 03 04 04 05 05 06 06 07 07 08
+ movq xmm4, [rsi+8] ; 08 09 10 11 12 13 14 15
+
+ movq xmm5, [rsi+9] ; 09 10 11 12 13 14 15 16
+
+ lea rsi, [rsi + rdx] ; next line
+
+ pmaddubsw xmm3, xmm1 ; 00 02 04 06 08 10 12 14
+
+ punpcklbw xmm4, xmm5 ; 08 09 09 10 10 11 11 12 12 13 13 14 14 15 15 16
+ pmaddubsw xmm4, xmm1 ; 01 03 05 07 09 11 13 15
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ paddw xmm4, [GLOBAL(rd)] ; xmm4 += round value
+ psraw xmm4, VP8_FILTER_SHIFT ; xmm4 /= 128
+
+ movdqa xmm7, xmm3
+ packuswb xmm7, xmm4 ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
+
+.next_row:
+ movq xmm6, [rsi] ; 00 01 02 03 04 05 06 07
+ movq xmm5, [rsi+1] ; 01 02 03 04 05 06 07 08
+
+ punpcklbw xmm6, xmm5
+ movq xmm4, [rsi+8] ; 08 09 10 11 12 13 14 15
+
+ movq xmm5, [rsi+9] ; 09 10 11 12 13 14 15 16
+ lea rsi, [rsi + rdx] ; next line
+
+ pmaddubsw xmm6, xmm1
+
+ punpcklbw xmm4, xmm5
+ pmaddubsw xmm4, xmm1
+
+ paddw xmm6, [GLOBAL(rd)] ; xmm6 += round value
+ psraw xmm6, VP8_FILTER_SHIFT ; xmm6 /= 128
+
+ paddw xmm4, [GLOBAL(rd)] ; xmm4 += round value
+ psraw xmm4, VP8_FILTER_SHIFT ; xmm4 /= 128
+
+ packuswb xmm6, xmm4
+ movdqa xmm5, xmm7
+
+ punpcklbw xmm5, xmm6
+ pmaddubsw xmm5, xmm2
+
+ punpckhbw xmm7, xmm6
+ pmaddubsw xmm7, xmm2
+
+ paddw xmm5, [GLOBAL(rd)] ; xmm5 += round value
+ psraw xmm5, VP8_FILTER_SHIFT ; xmm5 /= 128
+
+ paddw xmm7, [GLOBAL(rd)] ; xmm7 += round value
+ psraw xmm7, VP8_FILTER_SHIFT ; xmm7 /= 128
+
+ packuswb xmm5, xmm7
+ movdqa xmm7, xmm6
+
+ movdqa [rdi], xmm5 ; store the results in the destination
+%if ABI_IS_32BIT
+ add rdi, DWORD PTR arg(5) ; dst_pitch
+%else
+ add rdi, r8
+%endif
+
+ cmp rdi, rcx
+ jne .next_row
+
+ jmp .done
+
+.b16x16_sp_only:
+ movsxd rax, dword ptr arg(3) ; yoffset
+ shl rax, 4
+ lea rax, [rax + rcx] ; VFilter
+
+ mov rdi, arg(4) ; dst_ptr
+ mov rsi, arg(0) ; src_ptr
+ movsxd rdx, dword ptr arg(5) ; dst_pitch
+
+ movdqa xmm1, [rax] ; VFilter
+
+ lea rcx, [rdi+rdx*8]
+ lea rcx, [rcx+rdx*8]
+ movsxd rax, dword ptr arg(1) ; src_pixels_per_line
+
+ ; get the first horizontal line done
+ movq xmm4, [rsi] ; load row 0
+ movq xmm2, [rsi + 8] ; load row 0
+
+ lea rsi, [rsi + rax] ; next line
+.next_row_sp:
+ movq xmm3, [rsi] ; load row + 1
+ movq xmm5, [rsi + 8] ; load row + 1
+
+ punpcklbw xmm4, xmm3
+ punpcklbw xmm2, xmm5
+
+ pmaddubsw xmm4, xmm1
+ movq xmm7, [rsi + rax] ; load row + 2
+
+ pmaddubsw xmm2, xmm1
+ movq xmm6, [rsi + rax + 8] ; load row + 2
+
+ punpcklbw xmm3, xmm7
+ punpcklbw xmm5, xmm6
+
+ pmaddubsw xmm3, xmm1
+ paddw xmm4, [GLOBAL(rd)]
+
+ pmaddubsw xmm5, xmm1
+ paddw xmm2, [GLOBAL(rd)]
+
+ psraw xmm4, VP8_FILTER_SHIFT
+ psraw xmm2, VP8_FILTER_SHIFT
+
+ packuswb xmm4, xmm2
+ paddw xmm3, [GLOBAL(rd)]
+
+ movdqa [rdi], xmm4 ; store row 0
+ paddw xmm5, [GLOBAL(rd)]
+
+ psraw xmm3, VP8_FILTER_SHIFT
+ psraw xmm5, VP8_FILTER_SHIFT
+
+ packuswb xmm3, xmm5
+ movdqa xmm4, xmm7
+
+ movdqa [rdi + rdx],xmm3 ; store row 1
+ lea rsi, [rsi + 2*rax]
+
+ movdqa xmm2, xmm6
+ lea rdi, [rdi + 2*rdx]
+
+ cmp rdi, rcx
+ jne .next_row_sp
+
+ jmp .done
+
+.b16x16_fp_only:
+ lea rcx, [rdi+rdx*8]
+ lea rcx, [rcx+rdx*8]
+ movsxd rax, dword ptr arg(1) ; src_pixels_per_line
+
+.next_row_fp:
+ movq xmm2, [rsi] ; 00 01 02 03 04 05 06 07
+ movq xmm4, [rsi+1] ; 01 02 03 04 05 06 07 08
+
+ punpcklbw xmm2, xmm4
+ movq xmm3, [rsi+8] ; 08 09 10 11 12 13 14 15
+
+ pmaddubsw xmm2, xmm1
+ movq xmm4, [rsi+9] ; 09 10 11 12 13 14 15 16
+
+ lea rsi, [rsi + rax] ; next line
+ punpcklbw xmm3, xmm4
+
+ pmaddubsw xmm3, xmm1
+ movq xmm5, [rsi]
+
+ paddw xmm2, [GLOBAL(rd)]
+ movq xmm7, [rsi+1]
+
+ movq xmm6, [rsi+8]
+ psraw xmm2, VP8_FILTER_SHIFT
+
+ punpcklbw xmm5, xmm7
+ movq xmm7, [rsi+9]
+
+ paddw xmm3, [GLOBAL(rd)]
+ pmaddubsw xmm5, xmm1
+
+ psraw xmm3, VP8_FILTER_SHIFT
+ punpcklbw xmm6, xmm7
+
+ packuswb xmm2, xmm3
+ pmaddubsw xmm6, xmm1
+
+ movdqa [rdi], xmm2 ; store the results in the destination
+ paddw xmm5, [GLOBAL(rd)]
+
+ lea rdi, [rdi + rdx] ; dst_pitch
+ psraw xmm5, VP8_FILTER_SHIFT
+
+ paddw xmm6, [GLOBAL(rd)]
+ psraw xmm6, VP8_FILTER_SHIFT
+
+ packuswb xmm5, xmm6
+ lea rsi, [rsi + rax] ; next line
+
+ movdqa [rdi], xmm5 ; store the results in the destination
+ lea rdi, [rdi + rdx] ; dst_pitch
+
+ cmp rdi, rcx
+
+ jne .next_row_fp
+
+.done:
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_bilinear_predict8x8_ssse3
+;(
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; int xoffset,
+; int yoffset,
+; unsigned char *dst_ptr,
+; int dst_pitch
+;)
+global sym(vp8_bilinear_predict8x8_ssse3) PRIVATE
+sym(vp8_bilinear_predict8x8_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ ALIGN_STACK 16, rax
+ sub rsp, 144 ; reserve 144 bytes
+
+ lea rcx, [GLOBAL(vp8_bilinear_filters_ssse3)]
+
+ mov rsi, arg(0) ;src_ptr
+ movsxd rdx, dword ptr arg(1) ;src_pixels_per_line
+
+ ;Read 9-line unaligned data in and put them on stack. This gives a big
+ ;performance boost.
+ movdqu xmm0, [rsi]
+ lea rax, [rdx + rdx*2]
+ movdqu xmm1, [rsi+rdx]
+ movdqu xmm2, [rsi+rdx*2]
+ add rsi, rax
+ movdqu xmm3, [rsi]
+ movdqu xmm4, [rsi+rdx]
+ movdqu xmm5, [rsi+rdx*2]
+ add rsi, rax
+ movdqu xmm6, [rsi]
+ movdqu xmm7, [rsi+rdx]
+
+ movdqa XMMWORD PTR [rsp], xmm0
+
+ movdqu xmm0, [rsi+rdx*2]
+
+ movdqa XMMWORD PTR [rsp+16], xmm1
+ movdqa XMMWORD PTR [rsp+32], xmm2
+ movdqa XMMWORD PTR [rsp+48], xmm3
+ movdqa XMMWORD PTR [rsp+64], xmm4
+ movdqa XMMWORD PTR [rsp+80], xmm5
+ movdqa XMMWORD PTR [rsp+96], xmm6
+ movdqa XMMWORD PTR [rsp+112], xmm7
+ movdqa XMMWORD PTR [rsp+128], xmm0
+
+ movsxd rax, dword ptr arg(2) ; xoffset
+ cmp rax, 0 ; skip first_pass filter if xoffset=0
+ je .b8x8_sp_only
+
+ shl rax, 4
+ add rax, rcx ; HFilter
+
+ mov rdi, arg(4) ; dst_ptr
+ movsxd rdx, dword ptr arg(5) ; dst_pitch
+
+ movdqa xmm0, [rax]
+
+ movsxd rax, dword ptr arg(3) ; yoffset
+ cmp rax, 0 ; skip second_pass filter if yoffset=0
+ je .b8x8_fp_only
+
+ shl rax, 4
+ lea rax, [rax + rcx] ; VFilter
+
+ lea rcx, [rdi+rdx*8]
+
+ movdqa xmm1, [rax]
+
+ ; get the first horizontal line done
+ movdqa xmm3, [rsp] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
+ movdqa xmm5, xmm3 ; 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 xx
+
+ psrldq xmm5, 1
+ lea rsp, [rsp + 16] ; next line
+
+ punpcklbw xmm3, xmm5 ; 00 01 01 02 02 03 03 04 04 05 05 06 06 07 07 08
+ pmaddubsw xmm3, xmm0 ; 00 02 04 06 08 10 12 14
+
+ paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
+ psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
+
+ movdqa xmm7, xmm3
+ packuswb xmm7, xmm7 ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
+
+.next_row:
+ movdqa xmm6, [rsp] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
+ lea rsp, [rsp + 16] ; next line
+
+ movdqa xmm5, xmm6
+
+ psrldq xmm5, 1
+
+ punpcklbw xmm6, xmm5
+ pmaddubsw xmm6, xmm0
+
+ paddw xmm6, [GLOBAL(rd)] ; xmm6 += round value
+ psraw xmm6, VP8_FILTER_SHIFT ; xmm6 /= 128
+
+ packuswb xmm6, xmm6
+
+ punpcklbw xmm7, xmm6
+ pmaddubsw xmm7, xmm1
+
+ paddw xmm7, [GLOBAL(rd)] ; xmm7 += round value
+ psraw xmm7, VP8_FILTER_SHIFT ; xmm7 /= 128
+
+ packuswb xmm7, xmm7
+
+ movq [rdi], xmm7 ; store the results in the destination
+ lea rdi, [rdi + rdx]
+
+ movdqa xmm7, xmm6
+
+ cmp rdi, rcx
+ jne .next_row
+
+ jmp .done8x8
+
+.b8x8_sp_only:
+ movsxd rax, dword ptr arg(3) ; yoffset
+ shl rax, 4
+ lea rax, [rax + rcx] ; VFilter
+
+ mov rdi, arg(4) ;dst_ptr
+ movsxd rdx, dword ptr arg(5) ; dst_pitch
+
+ movdqa xmm0, [rax] ; VFilter
+
+ movq xmm1, XMMWORD PTR [rsp]
+ movq xmm2, XMMWORD PTR [rsp+16]
+
+ movq xmm3, XMMWORD PTR [rsp+32]
+ punpcklbw xmm1, xmm2
+
+ movq xmm4, XMMWORD PTR [rsp+48]
+ punpcklbw xmm2, xmm3
+
+ movq xmm5, XMMWORD PTR [rsp+64]
+ punpcklbw xmm3, xmm4
+
+ movq xmm6, XMMWORD PTR [rsp+80]
+ punpcklbw xmm4, xmm5
+
+ movq xmm7, XMMWORD PTR [rsp+96]
+ punpcklbw xmm5, xmm6
+
+ pmaddubsw xmm1, xmm0
+ pmaddubsw xmm2, xmm0
+
+ pmaddubsw xmm3, xmm0
+ pmaddubsw xmm4, xmm0
+
+ pmaddubsw xmm5, xmm0
+ punpcklbw xmm6, xmm7
+
+ pmaddubsw xmm6, xmm0
+ paddw xmm1, [GLOBAL(rd)]
+
+ paddw xmm2, [GLOBAL(rd)]
+ psraw xmm1, VP8_FILTER_SHIFT
+
+ paddw xmm3, [GLOBAL(rd)]
+ psraw xmm2, VP8_FILTER_SHIFT
+
+ paddw xmm4, [GLOBAL(rd)]
+ psraw xmm3, VP8_FILTER_SHIFT
+
+ paddw xmm5, [GLOBAL(rd)]
+ psraw xmm4, VP8_FILTER_SHIFT
+
+ paddw xmm6, [GLOBAL(rd)]
+ psraw xmm5, VP8_FILTER_SHIFT
+
+ psraw xmm6, VP8_FILTER_SHIFT
+ packuswb xmm1, xmm1
+
+ packuswb xmm2, xmm2
+ movq [rdi], xmm1
+
+ packuswb xmm3, xmm3
+ movq [rdi+rdx], xmm2
+
+ packuswb xmm4, xmm4
+ movq xmm1, XMMWORD PTR [rsp+112]
+
+ lea rdi, [rdi + 2*rdx]
+ movq xmm2, XMMWORD PTR [rsp+128]
+
+ packuswb xmm5, xmm5
+ movq [rdi], xmm3
+
+ packuswb xmm6, xmm6
+ movq [rdi+rdx], xmm4
+
+ lea rdi, [rdi + 2*rdx]
+ punpcklbw xmm7, xmm1
+
+ movq [rdi], xmm5
+ pmaddubsw xmm7, xmm0
+
+ movq [rdi+rdx], xmm6
+ punpcklbw xmm1, xmm2
+
+ pmaddubsw xmm1, xmm0
+ paddw xmm7, [GLOBAL(rd)]
+
+ psraw xmm7, VP8_FILTER_SHIFT
+ paddw xmm1, [GLOBAL(rd)]
+
+ psraw xmm1, VP8_FILTER_SHIFT
+ packuswb xmm7, xmm7
+
+ packuswb xmm1, xmm1
+ lea rdi, [rdi + 2*rdx]
+
+ movq [rdi], xmm7
+
+ movq [rdi+rdx], xmm1
+ lea rsp, [rsp + 144]
+
+ jmp .done8x8
+
+.b8x8_fp_only:
+ lea rcx, [rdi+rdx*8]
+
+.next_row_fp:
+ movdqa xmm1, XMMWORD PTR [rsp]
+ movdqa xmm3, XMMWORD PTR [rsp+16]
+
+ movdqa xmm2, xmm1
+ movdqa xmm5, XMMWORD PTR [rsp+32]
+
+ psrldq xmm2, 1
+ movdqa xmm7, XMMWORD PTR [rsp+48]
+
+ movdqa xmm4, xmm3
+ psrldq xmm4, 1
+
+ movdqa xmm6, xmm5
+ psrldq xmm6, 1
+
+ punpcklbw xmm1, xmm2
+ pmaddubsw xmm1, xmm0
+
+ punpcklbw xmm3, xmm4
+ pmaddubsw xmm3, xmm0
+
+ punpcklbw xmm5, xmm6
+ pmaddubsw xmm5, xmm0
+
+ movdqa xmm2, xmm7
+ psrldq xmm2, 1
+
+ punpcklbw xmm7, xmm2
+ pmaddubsw xmm7, xmm0
+
+ paddw xmm1, [GLOBAL(rd)]
+ psraw xmm1, VP8_FILTER_SHIFT
+
+ paddw xmm3, [GLOBAL(rd)]
+ psraw xmm3, VP8_FILTER_SHIFT
+
+ paddw xmm5, [GLOBAL(rd)]
+ psraw xmm5, VP8_FILTER_SHIFT
+
+ paddw xmm7, [GLOBAL(rd)]
+ psraw xmm7, VP8_FILTER_SHIFT
+
+ packuswb xmm1, xmm1
+ packuswb xmm3, xmm3
+
+ packuswb xmm5, xmm5
+ movq [rdi], xmm1
+
+ packuswb xmm7, xmm7
+ movq [rdi+rdx], xmm3
+
+ lea rdi, [rdi + 2*rdx]
+ movq [rdi], xmm5
+
+ lea rsp, [rsp + 4*16]
+ movq [rdi+rdx], xmm7
+
+ lea rdi, [rdi + 2*rdx]
+ cmp rdi, rcx
+
+ jne .next_row_fp
+
+ lea rsp, [rsp + 16]
+
+.done8x8:
+ ;add rsp, 144
+ pop rsp
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+shuf1b:
+ db 0, 5, 1, 6, 2, 7, 3, 8, 4, 9, 5, 10, 6, 11, 7, 12
+shuf2b:
+ db 2, 4, 3, 5, 4, 6, 5, 7, 6, 8, 7, 9, 8, 10, 9, 11
+shuf3b:
+ db 1, 3, 2, 4, 3, 5, 4, 6, 5, 7, 6, 8, 7, 9, 8, 10
+
+align 16
+shuf2bfrom1:
+ db 4, 8, 6, 1, 8, 3, 1, 5, 3, 7, 5, 9, 7,11, 9,13
+align 16
+shuf3bfrom1:
+ db 2, 6, 4, 8, 6, 1, 8, 3, 1, 5, 3, 7, 5, 9, 7,11
+
+align 16
+rd:
+ times 8 dw 0x40
+
+align 16
+k0_k5:
+ times 8 db 0, 0 ;placeholder
+ times 8 db 0, 0
+ times 8 db 2, 1
+ times 8 db 0, 0
+ times 8 db 3, 3
+ times 8 db 0, 0
+ times 8 db 1, 2
+ times 8 db 0, 0
+k1_k3:
+ times 8 db 0, 0 ;placeholder
+ times 8 db -6, 12
+ times 8 db -11, 36
+ times 8 db -9, 50
+ times 8 db -16, 77
+ times 8 db -6, 93
+ times 8 db -8, 108
+ times 8 db -1, 123
+k2_k4:
+ times 8 db 128, 0 ;placeholder
+ times 8 db 123, -1
+ times 8 db 108, -8
+ times 8 db 93, -6
+ times 8 db 77, -16
+ times 8 db 50, -9
+ times 8 db 36, -11
+ times 8 db 12, -6
+align 16
+vp8_bilinear_filters_ssse3:
+ times 8 db 128, 0
+ times 8 db 112, 16
+ times 8 db 96, 32
+ times 8 db 80, 48
+ times 8 db 64, 64
+ times 8 db 48, 80
+ times 8 db 32, 96
+ times 8 db 16, 112
+
diff --git a/libvpx/vp8/common/x86/variance_impl_mmx.asm b/libvpx/vp8/common/x86/variance_impl_mmx.asm
new file mode 100644
index 0000000..d9120d0
--- /dev/null
+++ b/libvpx/vp8/common/x86/variance_impl_mmx.asm
@@ -0,0 +1,851 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;unsigned int vp8_get_mb_ss_mmx( short *src_ptr )
+global sym(vp8_get_mb_ss_mmx) PRIVATE
+sym(vp8_get_mb_ss_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ sub rsp, 8
+ ; end prolog
+
+ mov rax, arg(0) ;src_ptr
+ mov rcx, 16
+ pxor mm4, mm4
+
+.NEXTROW:
+ movq mm0, [rax]
+ movq mm1, [rax+8]
+ movq mm2, [rax+16]
+ movq mm3, [rax+24]
+ pmaddwd mm0, mm0
+ pmaddwd mm1, mm1
+ pmaddwd mm2, mm2
+ pmaddwd mm3, mm3
+
+ paddd mm4, mm0
+ paddd mm4, mm1
+ paddd mm4, mm2
+ paddd mm4, mm3
+
+ add rax, 32
+ dec rcx
+ ja .NEXTROW
+ movq QWORD PTR [rsp], mm4
+
+ ;return sum[0]+sum[1];
+ movsxd rax, dword ptr [rsp]
+ movsxd rcx, dword ptr [rsp+4]
+ add rax, rcx
+
+
+ ; begin epilog
+ add rsp, 8
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;unsigned int vp8_get8x8var_mmx
+;(
+; unsigned char *src_ptr,
+; int source_stride,
+; unsigned char *ref_ptr,
+; int recon_stride,
+; unsigned int *SSE,
+; int *Sum
+;)
+global sym(vp8_get8x8var_mmx) PRIVATE
+sym(vp8_get8x8var_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ push rsi
+ push rdi
+ push rbx
+ sub rsp, 16
+ ; end prolog
+
+
+ pxor mm5, mm5 ; Blank mmx6
+ pxor mm6, mm6 ; Blank mmx7
+ pxor mm7, mm7 ; Blank mmx7
+
+ mov rax, arg(0) ;[src_ptr] ; Load base addresses
+ mov rbx, arg(2) ;[ref_ptr]
+ movsxd rcx, dword ptr arg(1) ;[source_stride]
+ movsxd rdx, dword ptr arg(3) ;[recon_stride]
+
+ ; Row 1
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ movq mm2, mm0 ; Take copies
+ movq mm3, mm1 ; Take copies
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ punpckhbw mm2, mm6 ; unpack to higher prrcision
+ punpckhbw mm3, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ psubsw mm2, mm3 ; A-B (high order) to MM2
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+ paddw mm5, mm2 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ pmaddwd mm2, mm2 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+ paddd mm7, mm2 ; accumulate in mm7
+
+
+ ; Row 2
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm2, mm0 ; Take copies
+ movq mm3, mm1 ; Take copies
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ punpckhbw mm2, mm6 ; unpack to higher prrcision
+ punpckhbw mm3, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ psubsw mm2, mm3 ; A-B (high order) to MM2
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+ paddw mm5, mm2 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ pmaddwd mm2, mm2 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+ paddd mm7, mm2 ; accumulate in mm7
+
+ ; Row 3
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm2, mm0 ; Take copies
+ movq mm3, mm1 ; Take copies
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ punpckhbw mm2, mm6 ; unpack to higher prrcision
+ punpckhbw mm3, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ psubsw mm2, mm3 ; A-B (high order) to MM2
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+ paddw mm5, mm2 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ pmaddwd mm2, mm2 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+ paddd mm7, mm2 ; accumulate in mm7
+
+ ; Row 4
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm2, mm0 ; Take copies
+ movq mm3, mm1 ; Take copies
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ punpckhbw mm2, mm6 ; unpack to higher prrcision
+ punpckhbw mm3, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ psubsw mm2, mm3 ; A-B (high order) to MM2
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+ paddw mm5, mm2 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ pmaddwd mm2, mm2 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+ paddd mm7, mm2 ; accumulate in mm7
+
+ ; Row 5
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm2, mm0 ; Take copies
+ movq mm3, mm1 ; Take copies
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ punpckhbw mm2, mm6 ; unpack to higher prrcision
+ punpckhbw mm3, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ psubsw mm2, mm3 ; A-B (high order) to MM2
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+ paddw mm5, mm2 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ pmaddwd mm2, mm2 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ ; movq mm4, [rbx + rdx]
+ paddd mm7, mm0 ; accumulate in mm7
+ paddd mm7, mm2 ; accumulate in mm7
+
+ ; Row 6
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm2, mm0 ; Take copies
+ movq mm3, mm1 ; Take copies
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ punpckhbw mm2, mm6 ; unpack to higher prrcision
+ punpckhbw mm3, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ psubsw mm2, mm3 ; A-B (high order) to MM2
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+ paddw mm5, mm2 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ pmaddwd mm2, mm2 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+ paddd mm7, mm2 ; accumulate in mm7
+
+ ; Row 7
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm2, mm0 ; Take copies
+ movq mm3, mm1 ; Take copies
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ punpckhbw mm2, mm6 ; unpack to higher prrcision
+ punpckhbw mm3, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ psubsw mm2, mm3 ; A-B (high order) to MM2
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+ paddw mm5, mm2 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ pmaddwd mm2, mm2 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+ paddd mm7, mm2 ; accumulate in mm7
+
+ ; Row 8
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm2, mm0 ; Take copies
+ movq mm3, mm1 ; Take copies
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ punpckhbw mm2, mm6 ; unpack to higher prrcision
+ punpckhbw mm3, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ psubsw mm2, mm3 ; A-B (high order) to MM2
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+ paddw mm5, mm2 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ pmaddwd mm2, mm2 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ paddd mm7, mm0 ; accumulate in mm7
+ paddd mm7, mm2 ; accumulate in mm7
+
+ ; Now accumulate the final results.
+ movq QWORD PTR [rsp+8], mm5 ; copy back accumulated results into normal memory
+ movq QWORD PTR [rsp], mm7 ; copy back accumulated results into normal memory
+ movsx rdx, WORD PTR [rsp+8]
+ movsx rcx, WORD PTR [rsp+10]
+ movsx rbx, WORD PTR [rsp+12]
+ movsx rax, WORD PTR [rsp+14]
+ add rdx, rcx
+ add rbx, rax
+ add rdx, rbx ;XSum
+ movsxd rax, DWORD PTR [rsp]
+ movsxd rcx, DWORD PTR [rsp+4]
+ add rax, rcx ;XXSum
+ mov rsi, arg(4) ;SSE
+ mov rdi, arg(5) ;Sum
+ mov dword ptr [rsi], eax
+ mov dword ptr [rdi], edx
+ xor rax, rax ; return 0
+
+
+ ; begin epilog
+ add rsp, 16
+ pop rbx
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+
+;unsigned int
+;vp8_get4x4var_mmx
+;(
+; unsigned char *src_ptr,
+; int source_stride,
+; unsigned char *ref_ptr,
+; int recon_stride,
+; unsigned int *SSE,
+; int *Sum
+;)
+global sym(vp8_get4x4var_mmx) PRIVATE
+sym(vp8_get4x4var_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ push rsi
+ push rdi
+ push rbx
+ sub rsp, 16
+ ; end prolog
+
+
+ pxor mm5, mm5 ; Blank mmx6
+ pxor mm6, mm6 ; Blank mmx7
+ pxor mm7, mm7 ; Blank mmx7
+
+ mov rax, arg(0) ;[src_ptr] ; Load base addresses
+ mov rbx, arg(2) ;[ref_ptr]
+ movsxd rcx, dword ptr arg(1) ;[source_stride]
+ movsxd rdx, dword ptr arg(3) ;[recon_stride]
+
+ ; Row 1
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ paddw mm5, mm0 ; accumulate differences in mm5
+ pmaddwd mm0, mm0 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+
+
+ ; Row 2
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ paddw mm5, mm0 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+
+ ; Row 3
+ movq mm0, [rax] ; Copy eight bytes to mm0
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ paddw mm5, mm0 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movq mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+
+ ; Row 4
+ movq mm0, [rax] ; Copy eight bytes to mm0
+
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+
+ paddw mm5, mm0 ; accumulate differences in mm5
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ paddd mm7, mm0 ; accumulate in mm7
+
+
+ ; Now accumulate the final results.
+ movq QWORD PTR [rsp+8], mm5 ; copy back accumulated results into normal memory
+ movq QWORD PTR [rsp], mm7 ; copy back accumulated results into normal memory
+ movsx rdx, WORD PTR [rsp+8]
+ movsx rcx, WORD PTR [rsp+10]
+ movsx rbx, WORD PTR [rsp+12]
+ movsx rax, WORD PTR [rsp+14]
+ add rdx, rcx
+ add rbx, rax
+ add rdx, rbx ;XSum
+ movsxd rax, DWORD PTR [rsp]
+ movsxd rcx, DWORD PTR [rsp+4]
+ add rax, rcx ;XXSum
+ mov rsi, arg(4) ;SSE
+ mov rdi, arg(5) ;Sum
+ mov dword ptr [rsi], eax
+ mov dword ptr [rdi], edx
+ xor rax, rax ; return 0
+
+
+ ; begin epilog
+ add rsp, 16
+ pop rbx
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+
+;unsigned int
+;vp8_get4x4sse_cs_mmx
+;(
+; unsigned char *src_ptr,
+; int source_stride,
+; unsigned char *ref_ptr,
+; int recon_stride
+;)
+global sym(vp8_get4x4sse_cs_mmx) PRIVATE
+sym(vp8_get4x4sse_cs_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ push rbx
+ ; end prolog
+
+
+ pxor mm6, mm6 ; Blank mmx7
+ pxor mm7, mm7 ; Blank mmx7
+
+ mov rax, arg(0) ;[src_ptr] ; Load base addresses
+ mov rbx, arg(2) ;[ref_ptr]
+ movsxd rcx, dword ptr arg(1) ;[source_stride]
+ movsxd rdx, dword ptr arg(3) ;[recon_stride]
+ ; Row 1
+ movd mm0, [rax] ; Copy eight bytes to mm0
+ movd mm1, [rbx] ; Copy eight bytes to mm1
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ pmaddwd mm0, mm0 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movd mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+
+ ; Row 2
+ movd mm0, [rax] ; Copy eight bytes to mm0
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ pmaddwd mm0, mm0 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movd mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+
+ ; Row 3
+ movd mm0, [rax] ; Copy eight bytes to mm0
+ punpcklbw mm1, mm6
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+
+ pmaddwd mm0, mm0 ; square and accumulate
+ add rbx,rdx ; Inc pointer into ref data
+ add rax,rcx ; Inc pointer into the new data
+ movd mm1, [rbx] ; Copy eight bytes to mm1
+ paddd mm7, mm0 ; accumulate in mm7
+
+ ; Row 4
+ movd mm0, [rax] ; Copy eight bytes to mm0
+ punpcklbw mm0, mm6 ; unpack to higher prrcision
+ punpcklbw mm1, mm6
+ psubsw mm0, mm1 ; A-B (low order) to MM0
+ pmaddwd mm0, mm0 ; square and accumulate
+ paddd mm7, mm0 ; accumulate in mm7
+
+ movq mm0, mm7 ;
+ psrlq mm7, 32
+
+ paddd mm0, mm7
+ movq rax, mm0
+
+
+ ; begin epilog
+ pop rbx
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+%define mmx_filter_shift 7
+
+;void vp8_filter_block2d_bil4x4_var_mmx
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned short *HFilter,
+; unsigned short *VFilter,
+; int *sum,
+; unsigned int *sumsquared
+;)
+global sym(vp8_filter_block2d_bil4x4_var_mmx) PRIVATE
+sym(vp8_filter_block2d_bil4x4_var_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 8
+ GET_GOT rbx
+ push rsi
+ push rdi
+ sub rsp, 16
+ ; end prolog
+
+
+ pxor mm6, mm6 ;
+ pxor mm7, mm7 ;
+
+ mov rax, arg(4) ;HFilter ;
+ mov rdx, arg(5) ;VFilter ;
+
+ mov rsi, arg(0) ;ref_ptr ;
+ mov rdi, arg(2) ;src_ptr ;
+
+ mov rcx, 4 ;
+ pxor mm0, mm0 ;
+
+ movd mm1, [rsi] ;
+ movd mm3, [rsi+1] ;
+
+ punpcklbw mm1, mm0 ;
+ pmullw mm1, [rax] ;
+
+ punpcklbw mm3, mm0 ;
+ pmullw mm3, [rax+8] ;
+
+ paddw mm1, mm3 ;
+ paddw mm1, [GLOBAL(mmx_bi_rd)] ;
+
+ psraw mm1, mmx_filter_shift ;
+ movq mm5, mm1
+
+%if ABI_IS_32BIT
+ add rsi, dword ptr arg(1) ;ref_pixels_per_line ;
+%else
+ movsxd r8, dword ptr arg(1) ;ref_pixels_per_line ;
+ add rsi, r8
+%endif
+
+.filter_block2d_bil4x4_var_mmx_loop:
+
+ movd mm1, [rsi] ;
+ movd mm3, [rsi+1] ;
+
+ punpcklbw mm1, mm0 ;
+ pmullw mm1, [rax] ;
+
+ punpcklbw mm3, mm0 ;
+ pmullw mm3, [rax+8] ;
+
+ paddw mm1, mm3 ;
+ paddw mm1, [GLOBAL(mmx_bi_rd)] ;
+
+ psraw mm1, mmx_filter_shift ;
+ movq mm3, mm5 ;
+
+ movq mm5, mm1 ;
+ pmullw mm3, [rdx] ;
+
+ pmullw mm1, [rdx+8] ;
+ paddw mm1, mm3 ;
+
+
+ paddw mm1, [GLOBAL(mmx_bi_rd)] ;
+ psraw mm1, mmx_filter_shift ;
+
+ movd mm3, [rdi] ;
+ punpcklbw mm3, mm0 ;
+
+ psubw mm1, mm3 ;
+ paddw mm6, mm1 ;
+
+ pmaddwd mm1, mm1 ;
+ paddd mm7, mm1 ;
+
+%if ABI_IS_32BIT
+ add rsi, dword ptr arg(1) ;ref_pixels_per_line ;
+ add rdi, dword ptr arg(3) ;src_pixels_per_line ;
+%else
+ movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line
+ add rsi, r8
+ add rdi, r9
+%endif
+ sub rcx, 1 ;
+ jnz .filter_block2d_bil4x4_var_mmx_loop ;
+
+
+ pxor mm3, mm3 ;
+ pxor mm2, mm2 ;
+
+ punpcklwd mm2, mm6 ;
+ punpckhwd mm3, mm6 ;
+
+ paddd mm2, mm3 ;
+ movq mm6, mm2 ;
+
+ psrlq mm6, 32 ;
+ paddd mm2, mm6 ;
+
+ psrad mm2, 16 ;
+ movq mm4, mm7 ;
+
+ psrlq mm4, 32 ;
+ paddd mm4, mm7 ;
+
+ mov rdi, arg(6) ;sum
+ mov rsi, arg(7) ;sumsquared
+
+ movd dword ptr [rdi], mm2 ;
+ movd dword ptr [rsi], mm4 ;
+
+
+
+ ; begin epilog
+ add rsp, 16
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+
+
+;void vp8_filter_block2d_bil_var_mmx
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; unsigned short *HFilter,
+; unsigned short *VFilter,
+; int *sum,
+; unsigned int *sumsquared
+;)
+global sym(vp8_filter_block2d_bil_var_mmx) PRIVATE
+sym(vp8_filter_block2d_bil_var_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 9
+ GET_GOT rbx
+ push rsi
+ push rdi
+ sub rsp, 16
+ ; end prolog
+
+ pxor mm6, mm6 ;
+ pxor mm7, mm7 ;
+ mov rax, arg(5) ;HFilter ;
+
+ mov rdx, arg(6) ;VFilter ;
+ mov rsi, arg(0) ;ref_ptr ;
+
+ mov rdi, arg(2) ;src_ptr ;
+ movsxd rcx, dword ptr arg(4) ;Height ;
+
+ pxor mm0, mm0 ;
+ movq mm1, [rsi] ;
+
+ movq mm3, [rsi+1] ;
+ movq mm2, mm1 ;
+
+ movq mm4, mm3 ;
+ punpcklbw mm1, mm0 ;
+
+ punpckhbw mm2, mm0 ;
+ pmullw mm1, [rax] ;
+
+ pmullw mm2, [rax] ;
+ punpcklbw mm3, mm0 ;
+
+ punpckhbw mm4, mm0 ;
+ pmullw mm3, [rax+8] ;
+
+ pmullw mm4, [rax+8] ;
+ paddw mm1, mm3 ;
+
+ paddw mm2, mm4 ;
+ paddw mm1, [GLOBAL(mmx_bi_rd)] ;
+
+ psraw mm1, mmx_filter_shift ;
+ paddw mm2, [GLOBAL(mmx_bi_rd)] ;
+
+ psraw mm2, mmx_filter_shift ;
+ movq mm5, mm1
+
+ packuswb mm5, mm2 ;
+%if ABI_IS_32BIT
+ add rsi, dword ptr arg(1) ;ref_pixels_per_line
+%else
+ movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
+ add rsi, r8
+%endif
+
+.filter_block2d_bil_var_mmx_loop:
+
+ movq mm1, [rsi] ;
+ movq mm3, [rsi+1] ;
+
+ movq mm2, mm1 ;
+ movq mm4, mm3 ;
+
+ punpcklbw mm1, mm0 ;
+ punpckhbw mm2, mm0 ;
+
+ pmullw mm1, [rax] ;
+ pmullw mm2, [rax] ;
+
+ punpcklbw mm3, mm0 ;
+ punpckhbw mm4, mm0 ;
+
+ pmullw mm3, [rax+8] ;
+ pmullw mm4, [rax+8] ;
+
+ paddw mm1, mm3 ;
+ paddw mm2, mm4 ;
+
+ paddw mm1, [GLOBAL(mmx_bi_rd)] ;
+ psraw mm1, mmx_filter_shift ;
+
+ paddw mm2, [GLOBAL(mmx_bi_rd)] ;
+ psraw mm2, mmx_filter_shift ;
+
+ movq mm3, mm5 ;
+ movq mm4, mm5 ;
+
+ punpcklbw mm3, mm0 ;
+ punpckhbw mm4, mm0 ;
+
+ movq mm5, mm1 ;
+ packuswb mm5, mm2 ;
+
+ pmullw mm3, [rdx] ;
+ pmullw mm4, [rdx] ;
+
+ pmullw mm1, [rdx+8] ;
+ pmullw mm2, [rdx+8] ;
+
+ paddw mm1, mm3 ;
+ paddw mm2, mm4 ;
+
+ paddw mm1, [GLOBAL(mmx_bi_rd)] ;
+ paddw mm2, [GLOBAL(mmx_bi_rd)] ;
+
+ psraw mm1, mmx_filter_shift ;
+ psraw mm2, mmx_filter_shift ;
+
+ movq mm3, [rdi] ;
+ movq mm4, mm3 ;
+
+ punpcklbw mm3, mm0 ;
+ punpckhbw mm4, mm0 ;
+
+ psubw mm1, mm3 ;
+ psubw mm2, mm4 ;
+
+ paddw mm6, mm1 ;
+ pmaddwd mm1, mm1 ;
+
+ paddw mm6, mm2 ;
+ pmaddwd mm2, mm2 ;
+
+ paddd mm7, mm1 ;
+ paddd mm7, mm2 ;
+
+%if ABI_IS_32BIT
+ add rsi, dword ptr arg(1) ;ref_pixels_per_line ;
+ add rdi, dword ptr arg(3) ;src_pixels_per_line ;
+%else
+ movsxd r8, dword ptr arg(1) ;ref_pixels_per_line ;
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line ;
+ add rsi, r8
+ add rdi, r9
+%endif
+ sub rcx, 1 ;
+ jnz .filter_block2d_bil_var_mmx_loop ;
+
+
+ pxor mm3, mm3 ;
+ pxor mm2, mm2 ;
+
+ punpcklwd mm2, mm6 ;
+ punpckhwd mm3, mm6 ;
+
+ paddd mm2, mm3 ;
+ movq mm6, mm2 ;
+
+ psrlq mm6, 32 ;
+ paddd mm2, mm6 ;
+
+ psrad mm2, 16 ;
+ movq mm4, mm7 ;
+
+ psrlq mm4, 32 ;
+ paddd mm4, mm7 ;
+
+ mov rdi, arg(7) ;sum
+ mov rsi, arg(8) ;sumsquared
+
+ movd dword ptr [rdi], mm2 ;
+ movd dword ptr [rsi], mm4 ;
+
+ ; begin epilog
+ add rsp, 16
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+SECTION_RODATA
+;short mmx_bi_rd[4] = { 64, 64, 64, 64};
+align 16
+mmx_bi_rd:
+ times 4 dw 64
diff --git a/libvpx/vp8/common/x86/variance_impl_sse2.asm b/libvpx/vp8/common/x86/variance_impl_sse2.asm
new file mode 100644
index 0000000..761433c
--- /dev/null
+++ b/libvpx/vp8/common/x86/variance_impl_sse2.asm
@@ -0,0 +1,1359 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%define xmm_filter_shift 7
+
+;unsigned int vp8_get_mb_ss_sse2
+;(
+; short *src_ptr
+;)
+global sym(vp8_get_mb_ss_sse2) PRIVATE
+sym(vp8_get_mb_ss_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 1
+ GET_GOT rbx
+ push rsi
+ push rdi
+ sub rsp, 16
+ ; end prolog
+
+
+ mov rax, arg(0) ;[src_ptr]
+ mov rcx, 8
+ pxor xmm4, xmm4
+
+.NEXTROW:
+ movdqa xmm0, [rax]
+ movdqa xmm1, [rax+16]
+ movdqa xmm2, [rax+32]
+ movdqa xmm3, [rax+48]
+ pmaddwd xmm0, xmm0
+ pmaddwd xmm1, xmm1
+ pmaddwd xmm2, xmm2
+ pmaddwd xmm3, xmm3
+
+ paddd xmm0, xmm1
+ paddd xmm2, xmm3
+ paddd xmm4, xmm0
+ paddd xmm4, xmm2
+
+ add rax, 0x40
+ dec rcx
+ ja .NEXTROW
+
+ movdqa xmm3,xmm4
+ psrldq xmm4,8
+ paddd xmm4,xmm3
+ movdqa xmm3,xmm4
+ psrldq xmm4,4
+ paddd xmm4,xmm3
+ movq rax,xmm4
+
+
+ ; begin epilog
+ add rsp, 16
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;unsigned int vp8_get16x16var_sse2
+;(
+; unsigned char * src_ptr,
+; int source_stride,
+; unsigned char * ref_ptr,
+; int recon_stride,
+; unsigned int * SSE,
+; int * Sum
+;)
+global sym(vp8_get16x16var_sse2) PRIVATE
+sym(vp8_get16x16var_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ push rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;[src_ptr]
+ mov rdi, arg(2) ;[ref_ptr]
+
+ movsxd rax, DWORD PTR arg(1) ;[source_stride]
+ movsxd rdx, DWORD PTR arg(3) ;[recon_stride]
+
+ ; Prefetch data
+ lea rcx, [rax+rax*2]
+ prefetcht0 [rsi]
+ prefetcht0 [rsi+rax]
+ prefetcht0 [rsi+rax*2]
+ prefetcht0 [rsi+rcx]
+ lea rbx, [rsi+rax*4]
+ prefetcht0 [rbx]
+ prefetcht0 [rbx+rax]
+ prefetcht0 [rbx+rax*2]
+ prefetcht0 [rbx+rcx]
+
+ lea rcx, [rdx+rdx*2]
+ prefetcht0 [rdi]
+ prefetcht0 [rdi+rdx]
+ prefetcht0 [rdi+rdx*2]
+ prefetcht0 [rdi+rcx]
+ lea rbx, [rdi+rdx*4]
+ prefetcht0 [rbx]
+ prefetcht0 [rbx+rdx]
+ prefetcht0 [rbx+rdx*2]
+ prefetcht0 [rbx+rcx]
+
+ pxor xmm0, xmm0 ; clear xmm0 for unpack
+ pxor xmm7, xmm7 ; clear xmm7 for accumulating diffs
+
+ pxor xmm6, xmm6 ; clear xmm6 for accumulating sse
+ mov rcx, 16
+
+.var16loop:
+ movdqu xmm1, XMMWORD PTR [rsi]
+ movdqu xmm2, XMMWORD PTR [rdi]
+
+ prefetcht0 [rsi+rax*8]
+ prefetcht0 [rdi+rdx*8]
+
+ movdqa xmm3, xmm1
+ movdqa xmm4, xmm2
+
+
+ punpcklbw xmm1, xmm0
+ punpckhbw xmm3, xmm0
+
+ punpcklbw xmm2, xmm0
+ punpckhbw xmm4, xmm0
+
+
+ psubw xmm1, xmm2
+ psubw xmm3, xmm4
+
+ paddw xmm7, xmm1
+ pmaddwd xmm1, xmm1
+
+ paddw xmm7, xmm3
+ pmaddwd xmm3, xmm3
+
+ paddd xmm6, xmm1
+ paddd xmm6, xmm3
+
+ add rsi, rax
+ add rdi, rdx
+
+ sub rcx, 1
+ jnz .var16loop
+
+
+ movdqa xmm1, xmm6
+ pxor xmm6, xmm6
+
+ pxor xmm5, xmm5
+ punpcklwd xmm6, xmm7
+
+ punpckhwd xmm5, xmm7
+ psrad xmm5, 16
+
+ psrad xmm6, 16
+ paddd xmm6, xmm5
+
+ movdqa xmm2, xmm1
+ punpckldq xmm1, xmm0
+
+ punpckhdq xmm2, xmm0
+ movdqa xmm7, xmm6
+
+ paddd xmm1, xmm2
+ punpckldq xmm6, xmm0
+
+ punpckhdq xmm7, xmm0
+ paddd xmm6, xmm7
+
+ movdqa xmm2, xmm1
+ movdqa xmm7, xmm6
+
+ psrldq xmm1, 8
+ psrldq xmm6, 8
+
+ paddd xmm7, xmm6
+ paddd xmm1, xmm2
+
+ mov rax, arg(5) ;[Sum]
+ mov rdi, arg(4) ;[SSE]
+
+ movd DWORD PTR [rax], xmm7
+ movd DWORD PTR [rdi], xmm1
+
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ pop rbx
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+
+
+;unsigned int vp8_get8x8var_sse2
+;(
+; unsigned char * src_ptr,
+; int source_stride,
+; unsigned char * ref_ptr,
+; int recon_stride,
+; unsigned int * SSE,
+; int * Sum
+;)
+global sym(vp8_get8x8var_sse2) PRIVATE
+sym(vp8_get8x8var_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 6
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ sub rsp, 16
+ ; end prolog
+
+ mov rsi, arg(0) ;[src_ptr]
+ mov rdi, arg(2) ;[ref_ptr]
+
+ movsxd rax, DWORD PTR arg(1) ;[source_stride]
+ movsxd rdx, DWORD PTR arg(3) ;[recon_stride]
+
+ pxor xmm0, xmm0 ; clear xmm0 for unpack
+ pxor xmm7, xmm7 ; clear xmm7 for accumulating diffs
+
+ movq xmm1, QWORD PTR [rsi]
+ movq xmm2, QWORD PTR [rdi]
+
+ punpcklbw xmm1, xmm0
+ punpcklbw xmm2, xmm0
+
+ psubsw xmm1, xmm2
+ paddw xmm7, xmm1
+
+ pmaddwd xmm1, xmm1
+
+ movq xmm2, QWORD PTR[rsi + rax]
+ movq xmm3, QWORD PTR[rdi + rdx]
+
+ punpcklbw xmm2, xmm0
+ punpcklbw xmm3, xmm0
+
+ psubsw xmm2, xmm3
+ paddw xmm7, xmm2
+
+ pmaddwd xmm2, xmm2
+ paddd xmm1, xmm2
+
+
+ movq xmm2, QWORD PTR[rsi + rax * 2]
+ movq xmm3, QWORD PTR[rdi + rdx * 2]
+
+ punpcklbw xmm2, xmm0
+ punpcklbw xmm3, xmm0
+
+ psubsw xmm2, xmm3
+ paddw xmm7, xmm2
+
+ pmaddwd xmm2, xmm2
+ paddd xmm1, xmm2
+
+
+ lea rsi, [rsi + rax * 2]
+ lea rdi, [rdi + rdx * 2]
+ movq xmm2, QWORD PTR[rsi + rax]
+ movq xmm3, QWORD PTR[rdi + rdx]
+
+ punpcklbw xmm2, xmm0
+ punpcklbw xmm3, xmm0
+
+ psubsw xmm2, xmm3
+ paddw xmm7, xmm2
+
+ pmaddwd xmm2, xmm2
+ paddd xmm1, xmm2
+
+ movq xmm2, QWORD PTR[rsi + rax *2]
+ movq xmm3, QWORD PTR[rdi + rdx *2]
+
+ punpcklbw xmm2, xmm0
+ punpcklbw xmm3, xmm0
+
+ psubsw xmm2, xmm3
+ paddw xmm7, xmm2
+
+ pmaddwd xmm2, xmm2
+ paddd xmm1, xmm2
+
+
+ lea rsi, [rsi + rax * 2]
+ lea rdi, [rdi + rdx * 2]
+
+
+ movq xmm2, QWORD PTR[rsi + rax]
+ movq xmm3, QWORD PTR[rdi + rdx]
+
+ punpcklbw xmm2, xmm0
+ punpcklbw xmm3, xmm0
+
+ psubsw xmm2, xmm3
+ paddw xmm7, xmm2
+
+ pmaddwd xmm2, xmm2
+ paddd xmm1, xmm2
+
+ movq xmm2, QWORD PTR[rsi + rax *2]
+ movq xmm3, QWORD PTR[rdi + rdx *2]
+
+ punpcklbw xmm2, xmm0
+ punpcklbw xmm3, xmm0
+
+ psubsw xmm2, xmm3
+ paddw xmm7, xmm2
+
+ pmaddwd xmm2, xmm2
+ paddd xmm1, xmm2
+
+
+ lea rsi, [rsi + rax * 2]
+ lea rdi, [rdi + rdx * 2]
+
+ movq xmm2, QWORD PTR[rsi + rax]
+ movq xmm3, QWORD PTR[rdi + rdx]
+
+ punpcklbw xmm2, xmm0
+ punpcklbw xmm3, xmm0
+
+ psubsw xmm2, xmm3
+ paddw xmm7, xmm2
+
+ pmaddwd xmm2, xmm2
+ paddd xmm1, xmm2
+
+
+ movdqa xmm6, xmm7
+ punpcklwd xmm6, xmm0
+
+ punpckhwd xmm7, xmm0
+ movdqa xmm2, xmm1
+
+ paddw xmm6, xmm7
+ punpckldq xmm1, xmm0
+
+ punpckhdq xmm2, xmm0
+ movdqa xmm7, xmm6
+
+ paddd xmm1, xmm2
+ punpckldq xmm6, xmm0
+
+ punpckhdq xmm7, xmm0
+ paddw xmm6, xmm7
+
+ movdqa xmm2, xmm1
+ movdqa xmm7, xmm6
+
+ psrldq xmm1, 8
+ psrldq xmm6, 8
+
+ paddw xmm7, xmm6
+ paddd xmm1, xmm2
+
+ mov rax, arg(5) ;[Sum]
+ mov rdi, arg(4) ;[SSE]
+
+ movq rdx, xmm7
+ movsx rcx, dx
+
+ mov dword ptr [rax], ecx
+ movd DWORD PTR [rdi], xmm1
+
+ ; begin epilog
+ add rsp, 16
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_filter_block2d_bil_var_sse2
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; int xoffset,
+; int yoffset,
+; int *sum,
+; unsigned int *sumsquared;;
+;
+;)
+global sym(vp8_filter_block2d_bil_var_sse2) PRIVATE
+sym(vp8_filter_block2d_bil_var_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 9
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ push rbx
+ ; end prolog
+
+ pxor xmm6, xmm6 ;
+ pxor xmm7, xmm7 ;
+
+ lea rsi, [GLOBAL(xmm_bi_rd)] ; rounding
+ movdqa xmm4, XMMWORD PTR [rsi]
+
+ lea rcx, [GLOBAL(vp8_bilinear_filters_sse2)]
+ movsxd rax, dword ptr arg(5) ; xoffset
+
+ cmp rax, 0 ; skip first_pass filter if xoffset=0
+ je filter_block2d_bil_var_sse2_sp_only
+
+ shl rax, 5 ; point to filter coeff with xoffset
+ lea rax, [rax + rcx] ; HFilter
+
+ movsxd rdx, dword ptr arg(6) ; yoffset
+
+ cmp rdx, 0 ; skip second_pass filter if yoffset=0
+ je filter_block2d_bil_var_sse2_fp_only
+
+ shl rdx, 5
+ lea rdx, [rdx + rcx] ; VFilter
+
+ mov rsi, arg(0) ;ref_ptr
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+
+ pxor xmm0, xmm0 ;
+ movq xmm1, QWORD PTR [rsi] ;
+ movq xmm3, QWORD PTR [rsi+1] ;
+
+ punpcklbw xmm1, xmm0 ;
+ pmullw xmm1, [rax] ;
+ punpcklbw xmm3, xmm0
+ pmullw xmm3, [rax+16] ;
+
+ paddw xmm1, xmm3 ;
+ paddw xmm1, xmm4 ;
+ psraw xmm1, xmm_filter_shift ;
+ movdqa xmm5, xmm1
+
+ movsxd rbx, dword ptr arg(1) ;ref_pixels_per_line
+ lea rsi, [rsi + rbx]
+%if ABI_IS_32BIT=0
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line
+%endif
+
+filter_block2d_bil_var_sse2_loop:
+ movq xmm1, QWORD PTR [rsi] ;
+ movq xmm3, QWORD PTR [rsi+1] ;
+
+ punpcklbw xmm1, xmm0 ;
+ pmullw xmm1, [rax] ;
+ punpcklbw xmm3, xmm0 ;
+ pmullw xmm3, [rax+16] ;
+
+ paddw xmm1, xmm3 ;
+ paddw xmm1, xmm4 ;
+ psraw xmm1, xmm_filter_shift ;
+
+ movdqa xmm3, xmm5 ;
+ movdqa xmm5, xmm1 ;
+
+ pmullw xmm3, [rdx] ;
+ pmullw xmm1, [rdx+16] ;
+ paddw xmm1, xmm3 ;
+ paddw xmm1, xmm4 ;
+ psraw xmm1, xmm_filter_shift ;
+
+ movq xmm3, QWORD PTR [rdi] ;
+ punpcklbw xmm3, xmm0 ;
+
+ psubw xmm1, xmm3 ;
+ paddw xmm6, xmm1 ;
+
+ pmaddwd xmm1, xmm1 ;
+ paddd xmm7, xmm1 ;
+
+ lea rsi, [rsi + rbx] ;ref_pixels_per_line
+%if ABI_IS_32BIT
+ add rdi, dword ptr arg(3) ;src_pixels_per_line
+%else
+ lea rdi, [rdi + r9]
+%endif
+
+ sub rcx, 1 ;
+ jnz filter_block2d_bil_var_sse2_loop ;
+
+ jmp filter_block2d_bil_variance
+
+filter_block2d_bil_var_sse2_sp_only:
+ movsxd rdx, dword ptr arg(6) ; yoffset
+
+ cmp rdx, 0 ; skip all if both xoffset=0 and yoffset=0
+ je filter_block2d_bil_var_sse2_full_pixel
+
+ shl rdx, 5
+ lea rdx, [rdx + rcx] ; VFilter
+
+ mov rsi, arg(0) ;ref_ptr
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+
+ pxor xmm0, xmm0 ;
+ movq xmm1, QWORD PTR [rsi] ;
+ punpcklbw xmm1, xmm0 ;
+
+ movsxd rbx, dword ptr arg(3) ;src_pixels_per_line
+ lea rsi, [rsi + rax]
+
+filter_block2d_bil_sp_only_loop:
+ movq xmm3, QWORD PTR [rsi] ;
+ punpcklbw xmm3, xmm0 ;
+ movdqa xmm5, xmm3
+
+ pmullw xmm1, [rdx] ;
+ pmullw xmm3, [rdx+16] ;
+ paddw xmm1, xmm3 ;
+ paddw xmm1, xmm4 ;
+ psraw xmm1, xmm_filter_shift ;
+
+ movq xmm3, QWORD PTR [rdi] ;
+ punpcklbw xmm3, xmm0 ;
+
+ psubw xmm1, xmm3 ;
+ paddw xmm6, xmm1 ;
+
+ pmaddwd xmm1, xmm1 ;
+ paddd xmm7, xmm1 ;
+
+ movdqa xmm1, xmm5 ;
+ lea rsi, [rsi + rax] ;ref_pixels_per_line
+ lea rdi, [rdi + rbx] ;src_pixels_per_line
+
+ sub rcx, 1 ;
+ jnz filter_block2d_bil_sp_only_loop ;
+
+ jmp filter_block2d_bil_variance
+
+filter_block2d_bil_var_sse2_full_pixel:
+ mov rsi, arg(0) ;ref_ptr
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd rbx, dword ptr arg(3) ;src_pixels_per_line
+ pxor xmm0, xmm0 ;
+
+filter_block2d_bil_full_pixel_loop:
+ movq xmm1, QWORD PTR [rsi] ;
+ punpcklbw xmm1, xmm0 ;
+
+ movq xmm2, QWORD PTR [rdi] ;
+ punpcklbw xmm2, xmm0 ;
+
+ psubw xmm1, xmm2 ;
+ paddw xmm6, xmm1 ;
+
+ pmaddwd xmm1, xmm1 ;
+ paddd xmm7, xmm1 ;
+
+ lea rsi, [rsi + rax] ;ref_pixels_per_line
+ lea rdi, [rdi + rbx] ;src_pixels_per_line
+
+ sub rcx, 1 ;
+ jnz filter_block2d_bil_full_pixel_loop ;
+
+ jmp filter_block2d_bil_variance
+
+filter_block2d_bil_var_sse2_fp_only:
+ mov rsi, arg(0) ;ref_ptr
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+ movsxd rdx, dword ptr arg(1) ;ref_pixels_per_line
+
+ pxor xmm0, xmm0 ;
+ movsxd rbx, dword ptr arg(3) ;src_pixels_per_line
+
+filter_block2d_bil_fp_only_loop:
+ movq xmm1, QWORD PTR [rsi] ;
+ movq xmm3, QWORD PTR [rsi+1] ;
+
+ punpcklbw xmm1, xmm0 ;
+ pmullw xmm1, [rax] ;
+ punpcklbw xmm3, xmm0 ;
+ pmullw xmm3, [rax+16] ;
+
+ paddw xmm1, xmm3 ;
+ paddw xmm1, xmm4 ;
+ psraw xmm1, xmm_filter_shift ;
+
+ movq xmm3, QWORD PTR [rdi] ;
+ punpcklbw xmm3, xmm0 ;
+
+ psubw xmm1, xmm3 ;
+ paddw xmm6, xmm1 ;
+
+ pmaddwd xmm1, xmm1 ;
+ paddd xmm7, xmm1 ;
+ lea rsi, [rsi + rdx]
+ lea rdi, [rdi + rbx] ;src_pixels_per_line
+
+ sub rcx, 1 ;
+ jnz filter_block2d_bil_fp_only_loop ;
+
+ jmp filter_block2d_bil_variance
+
+filter_block2d_bil_variance:
+ movdq2q mm6, xmm6 ;
+ movdq2q mm7, xmm7 ;
+
+ psrldq xmm6, 8
+ psrldq xmm7, 8
+
+ movdq2q mm2, xmm6
+ movdq2q mm3, xmm7
+
+ paddw mm6, mm2
+ paddd mm7, mm3
+
+ pxor mm3, mm3 ;
+ pxor mm2, mm2 ;
+
+ punpcklwd mm2, mm6 ;
+ punpckhwd mm3, mm6 ;
+
+ paddd mm2, mm3 ;
+ movq mm6, mm2 ;
+
+ psrlq mm6, 32 ;
+ paddd mm2, mm6 ;
+
+ psrad mm2, 16 ;
+ movq mm4, mm7 ;
+
+ psrlq mm4, 32 ;
+ paddd mm4, mm7 ;
+
+ mov rsi, arg(7) ; sum
+ mov rdi, arg(8) ; sumsquared
+
+ movd [rsi], mm2 ; xsum
+ movd [rdi], mm4 ; xxsum
+
+ ; begin epilog
+ pop rbx
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_half_horiz_vert_variance8x_h_sse2
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; int *sum,
+; unsigned int *sumsquared
+;)
+global sym(vp8_half_horiz_vert_variance8x_h_sse2) PRIVATE
+sym(vp8_half_horiz_vert_variance8x_h_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line
+%endif
+
+ pxor xmm6, xmm6 ; error accumulator
+ pxor xmm7, xmm7 ; sse eaccumulator
+ mov rsi, arg(0) ;ref_ptr ;
+
+ mov rdi, arg(2) ;src_ptr ;
+ movsxd rcx, dword ptr arg(4) ;Height ;
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+
+ pxor xmm0, xmm0 ;
+
+ movq xmm5, QWORD PTR [rsi] ; xmm5 = s0,s1,s2..s8
+ movq xmm3, QWORD PTR [rsi+1] ; xmm3 = s1,s2,s3..s9
+ pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3) horizontal line 1
+
+%if ABI_IS_32BIT
+ add rsi, dword ptr arg(1) ;ref_pixels_per_line ; next source
+%else
+ add rsi, r8
+%endif
+
+vp8_half_horiz_vert_variance8x_h_1:
+
+ movq xmm1, QWORD PTR [rsi] ;
+ movq xmm2, QWORD PTR [rsi+1] ;
+ pavgb xmm1, xmm2 ; xmm1 = avg(xmm1,xmm3) horizontal line i+1
+
+ pavgb xmm5, xmm1 ; xmm = vertical average of the above
+ punpcklbw xmm5, xmm0 ; xmm5 = words of above
+
+ movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d8
+ punpcklbw xmm3, xmm0 ; xmm3 = words of above
+
+ psubw xmm5, xmm3 ; xmm5 -= xmm3
+ paddw xmm6, xmm5 ; xmm6 += accumulated column differences
+ pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
+ paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
+
+ movdqa xmm5, xmm1 ; save xmm1 for use on the next row
+
+%if ABI_IS_32BIT
+ add esi, dword ptr arg(1) ;ref_pixels_per_line ; next source
+ add edi, dword ptr arg(3) ;src_pixels_per_line ; next destination
+%else
+ add rsi, r8
+ add rdi, r9
+%endif
+
+ sub rcx, 1 ;
+ jnz vp8_half_horiz_vert_variance8x_h_1 ;
+
+ movdq2q mm6, xmm6 ;
+ movdq2q mm7, xmm7 ;
+
+ psrldq xmm6, 8
+ psrldq xmm7, 8
+
+ movdq2q mm2, xmm6
+ movdq2q mm3, xmm7
+
+ paddw mm6, mm2
+ paddd mm7, mm3
+
+ pxor mm3, mm3 ;
+ pxor mm2, mm2 ;
+
+ punpcklwd mm2, mm6 ;
+ punpckhwd mm3, mm6 ;
+
+ paddd mm2, mm3 ;
+ movq mm6, mm2 ;
+
+ psrlq mm6, 32 ;
+ paddd mm2, mm6 ;
+
+ psrad mm2, 16 ;
+ movq mm4, mm7 ;
+
+ psrlq mm4, 32 ;
+ paddd mm4, mm7 ;
+
+ mov rsi, arg(5) ; sum
+ mov rdi, arg(6) ; sumsquared
+
+ movd [rsi], mm2 ;
+ movd [rdi], mm4 ;
+
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_half_horiz_vert_variance16x_h_sse2
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; int *sum,
+; unsigned int *sumsquared
+;)
+global sym(vp8_half_horiz_vert_variance16x_h_sse2) PRIVATE
+sym(vp8_half_horiz_vert_variance16x_h_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ pxor xmm6, xmm6 ; error accumulator
+ pxor xmm7, xmm7 ; sse eaccumulator
+ mov rsi, arg(0) ;ref_ptr ;
+
+ mov rdi, arg(2) ;src_ptr ;
+ movsxd rcx, dword ptr arg(4) ;Height ;
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd rdx, dword ptr arg(3) ;src_pixels_per_line
+
+ pxor xmm0, xmm0 ;
+
+ movdqu xmm5, XMMWORD PTR [rsi]
+ movdqu xmm3, XMMWORD PTR [rsi+1]
+ pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3) horizontal line 1
+
+ lea rsi, [rsi + rax]
+
+vp8_half_horiz_vert_variance16x_h_1:
+ movdqu xmm1, XMMWORD PTR [rsi] ;
+ movdqu xmm2, XMMWORD PTR [rsi+1] ;
+ pavgb xmm1, xmm2 ; xmm1 = avg(xmm1,xmm3) horizontal line i+1
+
+ pavgb xmm5, xmm1 ; xmm = vertical average of the above
+
+ movdqa xmm4, xmm5
+ punpcklbw xmm5, xmm0 ; xmm5 = words of above
+ punpckhbw xmm4, xmm0
+
+ movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d7
+ punpcklbw xmm3, xmm0 ; xmm3 = words of above
+ psubw xmm5, xmm3 ; xmm5 -= xmm3
+
+ movq xmm3, QWORD PTR [rdi+8]
+ punpcklbw xmm3, xmm0
+ psubw xmm4, xmm3
+
+ paddw xmm6, xmm5 ; xmm6 += accumulated column differences
+ paddw xmm6, xmm4
+ pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
+ pmaddwd xmm4, xmm4
+ paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
+ paddd xmm7, xmm4
+
+ movdqa xmm5, xmm1 ; save xmm1 for use on the next row
+
+ lea rsi, [rsi + rax]
+ lea rdi, [rdi + rdx]
+
+ sub rcx, 1 ;
+ jnz vp8_half_horiz_vert_variance16x_h_1 ;
+
+ pxor xmm1, xmm1
+ pxor xmm5, xmm5
+
+ punpcklwd xmm0, xmm6
+ punpckhwd xmm1, xmm6
+ psrad xmm0, 16
+ psrad xmm1, 16
+ paddd xmm0, xmm1
+ movdqa xmm1, xmm0
+
+ movdqa xmm6, xmm7
+ punpckldq xmm6, xmm5
+ punpckhdq xmm7, xmm5
+ paddd xmm6, xmm7
+
+ punpckldq xmm0, xmm5
+ punpckhdq xmm1, xmm5
+ paddd xmm0, xmm1
+
+ movdqa xmm7, xmm6
+ movdqa xmm1, xmm0
+
+ psrldq xmm7, 8
+ psrldq xmm1, 8
+
+ paddd xmm6, xmm7
+ paddd xmm0, xmm1
+
+ mov rsi, arg(5) ;[Sum]
+ mov rdi, arg(6) ;[SSE]
+
+ movd [rsi], xmm0
+ movd [rdi], xmm6
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_half_vert_variance8x_h_sse2
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; int *sum,
+; unsigned int *sumsquared
+;)
+global sym(vp8_half_vert_variance8x_h_sse2) PRIVATE
+sym(vp8_half_vert_variance8x_h_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line
+%endif
+
+ pxor xmm6, xmm6 ; error accumulator
+ pxor xmm7, xmm7 ; sse eaccumulator
+ mov rsi, arg(0) ;ref_ptr ;
+
+ mov rdi, arg(2) ;src_ptr ;
+ movsxd rcx, dword ptr arg(4) ;Height ;
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+
+ pxor xmm0, xmm0 ;
+vp8_half_vert_variance8x_h_1:
+ movq xmm5, QWORD PTR [rsi] ; xmm5 = s0,s1,s2..s8
+ movq xmm3, QWORD PTR [rsi+rax] ; xmm3 = s1,s2,s3..s9
+
+ pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3)
+ punpcklbw xmm5, xmm0 ; xmm5 = words of above
+
+ movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d8
+ punpcklbw xmm3, xmm0 ; xmm3 = words of above
+
+ psubw xmm5, xmm3 ; xmm5 -= xmm3
+ paddw xmm6, xmm5 ; xmm6 += accumulated column differences
+ pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
+ paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
+
+%if ABI_IS_32BIT
+ add esi, dword ptr arg(1) ;ref_pixels_per_line ; next source
+ add edi, dword ptr arg(3) ;src_pixels_per_line ; next destination
+%else
+ add rsi, r8
+ add rdi, r9
+%endif
+
+ sub rcx, 1 ;
+ jnz vp8_half_vert_variance8x_h_1 ;
+
+ movdq2q mm6, xmm6 ;
+ movdq2q mm7, xmm7 ;
+
+ psrldq xmm6, 8
+ psrldq xmm7, 8
+
+ movdq2q mm2, xmm6
+ movdq2q mm3, xmm7
+
+ paddw mm6, mm2
+ paddd mm7, mm3
+
+ pxor mm3, mm3 ;
+ pxor mm2, mm2 ;
+
+ punpcklwd mm2, mm6 ;
+ punpckhwd mm3, mm6 ;
+
+ paddd mm2, mm3 ;
+ movq mm6, mm2 ;
+
+ psrlq mm6, 32 ;
+ paddd mm2, mm6 ;
+
+ psrad mm2, 16 ;
+ movq mm4, mm7 ;
+
+ psrlq mm4, 32 ;
+ paddd mm4, mm7 ;
+
+ mov rsi, arg(5) ; sum
+ mov rdi, arg(6) ; sumsquared
+
+ movd [rsi], mm2 ;
+ movd [rdi], mm4 ;
+
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_half_vert_variance16x_h_sse2
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; int *sum,
+; unsigned int *sumsquared
+;)
+global sym(vp8_half_vert_variance16x_h_sse2) PRIVATE
+sym(vp8_half_vert_variance16x_h_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ pxor xmm6, xmm6 ; error accumulator
+ pxor xmm7, xmm7 ; sse eaccumulator
+ mov rsi, arg(0) ;ref_ptr
+
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd rdx, dword ptr arg(3) ;src_pixels_per_line
+
+ movdqu xmm5, XMMWORD PTR [rsi]
+ lea rsi, [rsi + rax ]
+ pxor xmm0, xmm0
+
+vp8_half_vert_variance16x_h_1:
+ movdqu xmm3, XMMWORD PTR [rsi]
+
+ pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3)
+ movdqa xmm4, xmm5
+ punpcklbw xmm5, xmm0
+ punpckhbw xmm4, xmm0
+
+ movq xmm2, QWORD PTR [rdi]
+ punpcklbw xmm2, xmm0
+ psubw xmm5, xmm2
+ movq xmm2, QWORD PTR [rdi+8]
+ punpcklbw xmm2, xmm0
+ psubw xmm4, xmm2
+
+ paddw xmm6, xmm5 ; xmm6 += accumulated column differences
+ paddw xmm6, xmm4
+ pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
+ pmaddwd xmm4, xmm4
+ paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
+ paddd xmm7, xmm4
+
+ movdqa xmm5, xmm3
+
+ lea rsi, [rsi + rax]
+ lea rdi, [rdi + rdx]
+
+ sub rcx, 1
+ jnz vp8_half_vert_variance16x_h_1
+
+ pxor xmm1, xmm1
+ pxor xmm5, xmm5
+
+ punpcklwd xmm0, xmm6
+ punpckhwd xmm1, xmm6
+ psrad xmm0, 16
+ psrad xmm1, 16
+ paddd xmm0, xmm1
+ movdqa xmm1, xmm0
+
+ movdqa xmm6, xmm7
+ punpckldq xmm6, xmm5
+ punpckhdq xmm7, xmm5
+ paddd xmm6, xmm7
+
+ punpckldq xmm0, xmm5
+ punpckhdq xmm1, xmm5
+ paddd xmm0, xmm1
+
+ movdqa xmm7, xmm6
+ movdqa xmm1, xmm0
+
+ psrldq xmm7, 8
+ psrldq xmm1, 8
+
+ paddd xmm6, xmm7
+ paddd xmm0, xmm1
+
+ mov rsi, arg(5) ;[Sum]
+ mov rdi, arg(6) ;[SSE]
+
+ movd [rsi], xmm0
+ movd [rdi], xmm6
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_half_horiz_variance8x_h_sse2
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; int *sum,
+; unsigned int *sumsquared
+;)
+global sym(vp8_half_horiz_variance8x_h_sse2) PRIVATE
+sym(vp8_half_horiz_variance8x_h_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+%if ABI_IS_32BIT=0
+ movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line
+%endif
+
+ pxor xmm6, xmm6 ; error accumulator
+ pxor xmm7, xmm7 ; sse eaccumulator
+ mov rsi, arg(0) ;ref_ptr ;
+
+ mov rdi, arg(2) ;src_ptr ;
+ movsxd rcx, dword ptr arg(4) ;Height ;
+
+ pxor xmm0, xmm0 ;
+vp8_half_horiz_variance8x_h_1:
+ movq xmm5, QWORD PTR [rsi] ; xmm5 = s0,s1,s2..s8
+ movq xmm3, QWORD PTR [rsi+1] ; xmm3 = s1,s2,s3..s9
+
+ pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3)
+ punpcklbw xmm5, xmm0 ; xmm5 = words of above
+
+ movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d8
+ punpcklbw xmm3, xmm0 ; xmm3 = words of above
+
+ psubw xmm5, xmm3 ; xmm5 -= xmm3
+ paddw xmm6, xmm5 ; xmm6 += accumulated column differences
+ pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
+ paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
+
+%if ABI_IS_32BIT
+ add esi, dword ptr arg(1) ;ref_pixels_per_line ; next source
+ add edi, dword ptr arg(3) ;src_pixels_per_line ; next destination
+%else
+ add rsi, r8
+ add rdi, r9
+%endif
+ sub rcx, 1 ;
+ jnz vp8_half_horiz_variance8x_h_1 ;
+
+ movdq2q mm6, xmm6 ;
+ movdq2q mm7, xmm7 ;
+
+ psrldq xmm6, 8
+ psrldq xmm7, 8
+
+ movdq2q mm2, xmm6
+ movdq2q mm3, xmm7
+
+ paddw mm6, mm2
+ paddd mm7, mm3
+
+ pxor mm3, mm3 ;
+ pxor mm2, mm2 ;
+
+ punpcklwd mm2, mm6 ;
+ punpckhwd mm3, mm6 ;
+
+ paddd mm2, mm3 ;
+ movq mm6, mm2 ;
+
+ psrlq mm6, 32 ;
+ paddd mm2, mm6 ;
+
+ psrad mm2, 16 ;
+ movq mm4, mm7 ;
+
+ psrlq mm4, 32 ;
+ paddd mm4, mm7 ;
+
+ mov rsi, arg(5) ; sum
+ mov rdi, arg(6) ; sumsquared
+
+ movd [rsi], mm2 ;
+ movd [rdi], mm4 ;
+
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_half_horiz_variance16x_h_sse2
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; int *sum,
+; unsigned int *sumsquared
+;)
+global sym(vp8_half_horiz_variance16x_h_sse2) PRIVATE
+sym(vp8_half_horiz_variance16x_h_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ pxor xmm6, xmm6 ; error accumulator
+ pxor xmm7, xmm7 ; sse eaccumulator
+ mov rsi, arg(0) ;ref_ptr ;
+
+ mov rdi, arg(2) ;src_ptr ;
+ movsxd rcx, dword ptr arg(4) ;Height ;
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd rdx, dword ptr arg(3) ;src_pixels_per_line
+
+ pxor xmm0, xmm0 ;
+
+vp8_half_horiz_variance16x_h_1:
+ movdqu xmm5, XMMWORD PTR [rsi] ; xmm5 = s0,s1,s2..s15
+ movdqu xmm3, XMMWORD PTR [rsi+1] ; xmm3 = s1,s2,s3..s16
+
+ pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3)
+ movdqa xmm1, xmm5
+ punpcklbw xmm5, xmm0 ; xmm5 = words of above
+ punpckhbw xmm1, xmm0
+
+ movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d7
+ punpcklbw xmm3, xmm0 ; xmm3 = words of above
+ movq xmm2, QWORD PTR [rdi+8]
+ punpcklbw xmm2, xmm0
+
+ psubw xmm5, xmm3 ; xmm5 -= xmm3
+ psubw xmm1, xmm2
+ paddw xmm6, xmm5 ; xmm6 += accumulated column differences
+ paddw xmm6, xmm1
+ pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
+ pmaddwd xmm1, xmm1
+ paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
+ paddd xmm7, xmm1
+
+ lea rsi, [rsi + rax]
+ lea rdi, [rdi + rdx]
+
+ sub rcx, 1 ;
+ jnz vp8_half_horiz_variance16x_h_1 ;
+
+ pxor xmm1, xmm1
+ pxor xmm5, xmm5
+
+ punpcklwd xmm0, xmm6
+ punpckhwd xmm1, xmm6
+ psrad xmm0, 16
+ psrad xmm1, 16
+ paddd xmm0, xmm1
+ movdqa xmm1, xmm0
+
+ movdqa xmm6, xmm7
+ punpckldq xmm6, xmm5
+ punpckhdq xmm7, xmm5
+ paddd xmm6, xmm7
+
+ punpckldq xmm0, xmm5
+ punpckhdq xmm1, xmm5
+ paddd xmm0, xmm1
+
+ movdqa xmm7, xmm6
+ movdqa xmm1, xmm0
+
+ psrldq xmm7, 8
+ psrldq xmm1, 8
+
+ paddd xmm6, xmm7
+ paddd xmm0, xmm1
+
+ mov rsi, arg(5) ;[Sum]
+ mov rdi, arg(6) ;[SSE]
+
+ movd [rsi], xmm0
+ movd [rdi], xmm6
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+; short xmm_bi_rd[8] = { 64, 64, 64, 64,64, 64, 64, 64};
+align 16
+xmm_bi_rd:
+ times 8 dw 64
+align 16
+vp8_bilinear_filters_sse2:
+ dw 128, 128, 128, 128, 128, 128, 128, 128, 0, 0, 0, 0, 0, 0, 0, 0
+ dw 112, 112, 112, 112, 112, 112, 112, 112, 16, 16, 16, 16, 16, 16, 16, 16
+ dw 96, 96, 96, 96, 96, 96, 96, 96, 32, 32, 32, 32, 32, 32, 32, 32
+ dw 80, 80, 80, 80, 80, 80, 80, 80, 48, 48, 48, 48, 48, 48, 48, 48
+ dw 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
+ dw 48, 48, 48, 48, 48, 48, 48, 48, 80, 80, 80, 80, 80, 80, 80, 80
+ dw 32, 32, 32, 32, 32, 32, 32, 32, 96, 96, 96, 96, 96, 96, 96, 96
+ dw 16, 16, 16, 16, 16, 16, 16, 16, 112, 112, 112, 112, 112, 112, 112, 112
diff --git a/libvpx/vp8/common/x86/variance_impl_ssse3.asm b/libvpx/vp8/common/x86/variance_impl_ssse3.asm
new file mode 100644
index 0000000..686b4a9
--- /dev/null
+++ b/libvpx/vp8/common/x86/variance_impl_ssse3.asm
@@ -0,0 +1,364 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%define xmm_filter_shift 7
+
+
+;void vp8_filter_block2d_bil_var_ssse3
+;(
+; unsigned char *ref_ptr,
+; int ref_pixels_per_line,
+; unsigned char *src_ptr,
+; int src_pixels_per_line,
+; unsigned int Height,
+; int xoffset,
+; int yoffset,
+; int *sum,
+; unsigned int *sumsquared;;
+;
+;)
+;Note: The filter coefficient at offset=0 is 128. Since the second register
+;for Pmaddubsw is signed bytes, we must calculate zero offset seperately.
+global sym(vp8_filter_block2d_bil_var_ssse3) PRIVATE
+sym(vp8_filter_block2d_bil_var_ssse3):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 9
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ pxor xmm6, xmm6
+ pxor xmm7, xmm7
+
+ lea rcx, [GLOBAL(vp8_bilinear_filters_ssse3)]
+ movsxd rax, dword ptr arg(5) ; xoffset
+
+ cmp rax, 0 ; skip first_pass filter if xoffset=0
+ je .filter_block2d_bil_var_ssse3_sp_only
+
+ shl rax, 4 ; point to filter coeff with xoffset
+ lea rax, [rax + rcx] ; HFilter
+
+ movsxd rdx, dword ptr arg(6) ; yoffset
+
+ cmp rdx, 0 ; skip second_pass filter if yoffset=0
+ je .filter_block2d_bil_var_ssse3_fp_only
+
+ shl rdx, 4
+ lea rdx, [rdx + rcx] ; VFilter
+
+ mov rsi, arg(0) ;ref_ptr
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+
+ movdqu xmm0, XMMWORD PTR [rsi]
+ movdqu xmm1, XMMWORD PTR [rsi+1]
+ movdqa xmm2, xmm0
+
+ punpcklbw xmm0, xmm1
+ punpckhbw xmm2, xmm1
+ pmaddubsw xmm0, [rax]
+ pmaddubsw xmm2, [rax]
+
+ paddw xmm0, [GLOBAL(xmm_bi_rd)]
+ paddw xmm2, [GLOBAL(xmm_bi_rd)]
+ psraw xmm0, xmm_filter_shift
+ psraw xmm2, xmm_filter_shift
+
+ packuswb xmm0, xmm2
+
+%if ABI_IS_32BIT
+ add rsi, dword ptr arg(1) ;ref_pixels_per_line
+%else
+ movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line
+ lea rsi, [rsi + r8]
+%endif
+
+.filter_block2d_bil_var_ssse3_loop:
+ movdqu xmm1, XMMWORD PTR [rsi]
+ movdqu xmm2, XMMWORD PTR [rsi+1]
+ movdqa xmm3, xmm1
+
+ punpcklbw xmm1, xmm2
+ punpckhbw xmm3, xmm2
+ pmaddubsw xmm1, [rax]
+ pmaddubsw xmm3, [rax]
+
+ paddw xmm1, [GLOBAL(xmm_bi_rd)]
+ paddw xmm3, [GLOBAL(xmm_bi_rd)]
+ psraw xmm1, xmm_filter_shift
+ psraw xmm3, xmm_filter_shift
+ packuswb xmm1, xmm3
+
+ movdqa xmm2, xmm0
+ movdqa xmm0, xmm1
+ movdqa xmm3, xmm2
+
+ punpcklbw xmm2, xmm1
+ punpckhbw xmm3, xmm1
+ pmaddubsw xmm2, [rdx]
+ pmaddubsw xmm3, [rdx]
+
+ paddw xmm2, [GLOBAL(xmm_bi_rd)]
+ paddw xmm3, [GLOBAL(xmm_bi_rd)]
+ psraw xmm2, xmm_filter_shift
+ psraw xmm3, xmm_filter_shift
+
+ movq xmm1, QWORD PTR [rdi]
+ pxor xmm4, xmm4
+ punpcklbw xmm1, xmm4
+ movq xmm5, QWORD PTR [rdi+8]
+ punpcklbw xmm5, xmm4
+
+ psubw xmm2, xmm1
+ psubw xmm3, xmm5
+ paddw xmm6, xmm2
+ paddw xmm6, xmm3
+ pmaddwd xmm2, xmm2
+ pmaddwd xmm3, xmm3
+ paddd xmm7, xmm2
+ paddd xmm7, xmm3
+
+%if ABI_IS_32BIT
+ add rsi, dword ptr arg(1) ;ref_pixels_per_line
+ add rdi, dword ptr arg(3) ;src_pixels_per_line
+%else
+ lea rsi, [rsi + r8]
+ lea rdi, [rdi + r9]
+%endif
+
+ sub rcx, 1
+ jnz .filter_block2d_bil_var_ssse3_loop
+
+ jmp .filter_block2d_bil_variance
+
+.filter_block2d_bil_var_ssse3_sp_only:
+ movsxd rdx, dword ptr arg(6) ; yoffset
+
+ cmp rdx, 0 ; Both xoffset =0 and yoffset=0
+ je .filter_block2d_bil_var_ssse3_full_pixel
+
+ shl rdx, 4
+ lea rdx, [rdx + rcx] ; VFilter
+
+ mov rsi, arg(0) ;ref_ptr
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+
+ movdqu xmm1, XMMWORD PTR [rsi]
+ movdqa xmm0, xmm1
+
+%if ABI_IS_32BIT=0
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line
+%endif
+
+ lea rsi, [rsi + rax]
+
+.filter_block2d_bil_sp_only_loop:
+ movdqu xmm3, XMMWORD PTR [rsi]
+ movdqa xmm2, xmm1
+ movdqa xmm0, xmm3
+
+ punpcklbw xmm1, xmm3
+ punpckhbw xmm2, xmm3
+ pmaddubsw xmm1, [rdx]
+ pmaddubsw xmm2, [rdx]
+
+ paddw xmm1, [GLOBAL(xmm_bi_rd)]
+ paddw xmm2, [GLOBAL(xmm_bi_rd)]
+ psraw xmm1, xmm_filter_shift
+ psraw xmm2, xmm_filter_shift
+
+ movq xmm3, QWORD PTR [rdi]
+ pxor xmm4, xmm4
+ punpcklbw xmm3, xmm4
+ movq xmm5, QWORD PTR [rdi+8]
+ punpcklbw xmm5, xmm4
+
+ psubw xmm1, xmm3
+ psubw xmm2, xmm5
+ paddw xmm6, xmm1
+ paddw xmm6, xmm2
+ pmaddwd xmm1, xmm1
+ pmaddwd xmm2, xmm2
+ paddd xmm7, xmm1
+ paddd xmm7, xmm2
+
+ movdqa xmm1, xmm0
+ lea rsi, [rsi + rax] ;ref_pixels_per_line
+
+%if ABI_IS_32BIT
+ add rdi, dword ptr arg(3) ;src_pixels_per_line
+%else
+ lea rdi, [rdi + r9]
+%endif
+
+ sub rcx, 1
+ jnz .filter_block2d_bil_sp_only_loop
+
+ jmp .filter_block2d_bil_variance
+
+.filter_block2d_bil_var_ssse3_full_pixel:
+ mov rsi, arg(0) ;ref_ptr
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+ movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
+ movsxd rdx, dword ptr arg(3) ;src_pixels_per_line
+ pxor xmm0, xmm0
+
+.filter_block2d_bil_full_pixel_loop:
+ movq xmm1, QWORD PTR [rsi]
+ punpcklbw xmm1, xmm0
+ movq xmm2, QWORD PTR [rsi+8]
+ punpcklbw xmm2, xmm0
+
+ movq xmm3, QWORD PTR [rdi]
+ punpcklbw xmm3, xmm0
+ movq xmm4, QWORD PTR [rdi+8]
+ punpcklbw xmm4, xmm0
+
+ psubw xmm1, xmm3
+ psubw xmm2, xmm4
+ paddw xmm6, xmm1
+ paddw xmm6, xmm2
+ pmaddwd xmm1, xmm1
+ pmaddwd xmm2, xmm2
+ paddd xmm7, xmm1
+ paddd xmm7, xmm2
+
+ lea rsi, [rsi + rax] ;ref_pixels_per_line
+ lea rdi, [rdi + rdx] ;src_pixels_per_line
+ sub rcx, 1
+ jnz .filter_block2d_bil_full_pixel_loop
+
+ jmp .filter_block2d_bil_variance
+
+.filter_block2d_bil_var_ssse3_fp_only:
+ mov rsi, arg(0) ;ref_ptr
+ mov rdi, arg(2) ;src_ptr
+ movsxd rcx, dword ptr arg(4) ;Height
+ movsxd rdx, dword ptr arg(1) ;ref_pixels_per_line
+
+ pxor xmm0, xmm0
+
+%if ABI_IS_32BIT=0
+ movsxd r9, dword ptr arg(3) ;src_pixels_per_line
+%endif
+
+.filter_block2d_bil_fp_only_loop:
+ movdqu xmm1, XMMWORD PTR [rsi]
+ movdqu xmm2, XMMWORD PTR [rsi+1]
+ movdqa xmm3, xmm1
+
+ punpcklbw xmm1, xmm2
+ punpckhbw xmm3, xmm2
+ pmaddubsw xmm1, [rax]
+ pmaddubsw xmm3, [rax]
+
+ paddw xmm1, [GLOBAL(xmm_bi_rd)]
+ paddw xmm3, [GLOBAL(xmm_bi_rd)]
+ psraw xmm1, xmm_filter_shift
+ psraw xmm3, xmm_filter_shift
+
+ movq xmm2, XMMWORD PTR [rdi]
+ pxor xmm4, xmm4
+ punpcklbw xmm2, xmm4
+ movq xmm5, QWORD PTR [rdi+8]
+ punpcklbw xmm5, xmm4
+
+ psubw xmm1, xmm2
+ psubw xmm3, xmm5
+ paddw xmm6, xmm1
+ paddw xmm6, xmm3
+ pmaddwd xmm1, xmm1
+ pmaddwd xmm3, xmm3
+ paddd xmm7, xmm1
+ paddd xmm7, xmm3
+
+ lea rsi, [rsi + rdx]
+%if ABI_IS_32BIT
+ add rdi, dword ptr arg(3) ;src_pixels_per_line
+%else
+ lea rdi, [rdi + r9]
+%endif
+
+ sub rcx, 1
+ jnz .filter_block2d_bil_fp_only_loop
+
+ jmp .filter_block2d_bil_variance
+
+.filter_block2d_bil_variance:
+ pxor xmm0, xmm0
+ pxor xmm1, xmm1
+ pxor xmm5, xmm5
+
+ punpcklwd xmm0, xmm6
+ punpckhwd xmm1, xmm6
+ psrad xmm0, 16
+ psrad xmm1, 16
+ paddd xmm0, xmm1
+ movdqa xmm1, xmm0
+
+ movdqa xmm6, xmm7
+ punpckldq xmm6, xmm5
+ punpckhdq xmm7, xmm5
+ paddd xmm6, xmm7
+
+ punpckldq xmm0, xmm5
+ punpckhdq xmm1, xmm5
+ paddd xmm0, xmm1
+
+ movdqa xmm7, xmm6
+ movdqa xmm1, xmm0
+
+ psrldq xmm7, 8
+ psrldq xmm1, 8
+
+ paddd xmm6, xmm7
+ paddd xmm0, xmm1
+
+ mov rsi, arg(7) ;[Sum]
+ mov rdi, arg(8) ;[SSE]
+
+ movd [rsi], xmm0
+ movd [rdi], xmm6
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+SECTION_RODATA
+align 16
+xmm_bi_rd:
+ times 8 dw 64
+align 16
+vp8_bilinear_filters_ssse3:
+ times 8 db 128, 0
+ times 8 db 112, 16
+ times 8 db 96, 32
+ times 8 db 80, 48
+ times 8 db 64, 64
+ times 8 db 48, 80
+ times 8 db 32, 96
+ times 8 db 16, 112
diff --git a/libvpx/vp8/common/x86/variance_mmx.c b/libvpx/vp8/common/x86/variance_mmx.c
new file mode 100644
index 0000000..0c4dd4a
--- /dev/null
+++ b/libvpx/vp8/common/x86/variance_mmx.c
@@ -0,0 +1,398 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vp8/common/variance.h"
+#include "vp8/common/pragmas.h"
+#include "vpx_ports/mem.h"
+#include "vp8/common/x86/filter_x86.h"
+
+extern void filter_block1d_h6_mmx
+(
+ const unsigned char *src_ptr,
+ unsigned short *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ short *filter
+);
+extern void filter_block1d_v6_mmx
+(
+ const short *src_ptr,
+ unsigned char *output_ptr,
+ unsigned int pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ short *filter
+);
+
+extern unsigned int vp8_get_mb_ss_mmx(const short *src_ptr);
+extern unsigned int vp8_get8x8var_mmx
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *SSE,
+ int *Sum
+);
+extern unsigned int vp8_get4x4var_mmx
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *SSE,
+ int *Sum
+);
+extern void vp8_filter_block2d_bil4x4_var_mmx
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ const short *HFilter,
+ const short *VFilter,
+ int *sum,
+ unsigned int *sumsquared
+);
+extern void vp8_filter_block2d_bil_var_mmx
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ const short *HFilter,
+ const short *VFilter,
+ int *sum,
+ unsigned int *sumsquared
+);
+
+
+unsigned int vp8_variance4x4_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+ vp8_get4x4var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 4));
+
+}
+
+unsigned int vp8_variance8x8_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+ vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
+ *sse = var;
+
+ return (var - ((unsigned int)(avg * avg) >> 6));
+
+}
+
+unsigned int vp8_mse16x16_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int sse0, sse1, sse2, sse3, var;
+ int sum0, sum1, sum2, sum3;
+
+
+ vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
+ vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
+ vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse2, &sum2) ;
+ vp8_get8x8var_mmx(src_ptr + 8 * source_stride + 8, source_stride, ref_ptr + 8 * recon_stride + 8, recon_stride, &sse3, &sum3);
+
+ var = sse0 + sse1 + sse2 + sse3;
+ *sse = var;
+ return var;
+}
+
+
+unsigned int vp8_variance16x16_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int sse0, sse1, sse2, sse3, var;
+ int sum0, sum1, sum2, sum3, avg;
+
+
+ vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
+ vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
+ vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse2, &sum2) ;
+ vp8_get8x8var_mmx(src_ptr + 8 * source_stride + 8, source_stride, ref_ptr + 8 * recon_stride + 8, recon_stride, &sse3, &sum3);
+
+ var = sse0 + sse1 + sse2 + sse3;
+ avg = sum0 + sum1 + sum2 + sum3;
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 8));
+}
+
+unsigned int vp8_variance16x8_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int sse0, sse1, var;
+ int sum0, sum1, avg;
+
+ vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
+ vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
+
+ var = sse0 + sse1;
+ avg = sum0 + sum1;
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 7));
+
+}
+
+
+unsigned int vp8_variance8x16_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int sse0, sse1, var;
+ int sum0, sum1, avg;
+
+ vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
+ vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse1, &sum1) ;
+
+ var = sse0 + sse1;
+ avg = sum0 + sum1;
+ *sse = var;
+
+ return (var - ((unsigned int)(avg * avg) >> 7));
+
+}
+
+
+unsigned int vp8_sub_pixel_variance4x4_mmx
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse)
+
+{
+ int xsum;
+ unsigned int xxsum;
+ vp8_filter_block2d_bil4x4_var_mmx(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line,
+ vp8_bilinear_filters_x86_4[xoffset], vp8_bilinear_filters_x86_4[yoffset],
+ &xsum, &xxsum
+ );
+ *sse = xxsum;
+ return (xxsum - ((unsigned int)(xsum * xsum) >> 4));
+}
+
+
+unsigned int vp8_sub_pixel_variance8x8_mmx
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+
+ int xsum;
+ unsigned int xxsum;
+ vp8_filter_block2d_bil_var_mmx(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ vp8_bilinear_filters_x86_4[xoffset], vp8_bilinear_filters_x86_4[yoffset],
+ &xsum, &xxsum
+ );
+ *sse = xxsum;
+ return (xxsum - ((unsigned int)(xsum * xsum) >> 6));
+}
+
+unsigned int vp8_sub_pixel_variance16x16_mmx
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+
+ int xsum0, xsum1;
+ unsigned int xxsum0, xxsum1;
+
+
+ vp8_filter_block2d_bil_var_mmx(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ vp8_bilinear_filters_x86_4[xoffset], vp8_bilinear_filters_x86_4[yoffset],
+ &xsum0, &xxsum0
+ );
+
+
+ vp8_filter_block2d_bil_var_mmx(
+ src_ptr + 8, src_pixels_per_line,
+ dst_ptr + 8, dst_pixels_per_line, 16,
+ vp8_bilinear_filters_x86_4[xoffset], vp8_bilinear_filters_x86_4[yoffset],
+ &xsum1, &xxsum1
+ );
+
+ xsum0 += xsum1;
+ xxsum0 += xxsum1;
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 8));
+
+
+}
+
+unsigned int vp8_sub_pixel_mse16x16_mmx(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ vp8_sub_pixel_variance16x16_mmx(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
+ return *sse;
+}
+
+unsigned int vp8_sub_pixel_variance16x8_mmx
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ int xsum0, xsum1;
+ unsigned int xxsum0, xxsum1;
+
+
+ vp8_filter_block2d_bil_var_mmx(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ vp8_bilinear_filters_x86_4[xoffset], vp8_bilinear_filters_x86_4[yoffset],
+ &xsum0, &xxsum0
+ );
+
+
+ vp8_filter_block2d_bil_var_mmx(
+ src_ptr + 8, src_pixels_per_line,
+ dst_ptr + 8, dst_pixels_per_line, 8,
+ vp8_bilinear_filters_x86_4[xoffset], vp8_bilinear_filters_x86_4[yoffset],
+ &xsum1, &xxsum1
+ );
+
+ xsum0 += xsum1;
+ xxsum0 += xxsum1;
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 7));
+}
+
+unsigned int vp8_sub_pixel_variance8x16_mmx
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ int xsum;
+ unsigned int xxsum;
+ vp8_filter_block2d_bil_var_mmx(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ vp8_bilinear_filters_x86_4[xoffset], vp8_bilinear_filters_x86_4[yoffset],
+ &xsum, &xxsum
+ );
+ *sse = xxsum;
+ return (xxsum - ((unsigned int)(xsum * xsum) >> 7));
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_h_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ return vp8_sub_pixel_variance16x16_mmx(src_ptr, source_stride, 4, 0,
+ ref_ptr, recon_stride, sse);
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_v_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ return vp8_sub_pixel_variance16x16_mmx(src_ptr, source_stride, 0, 4,
+ ref_ptr, recon_stride, sse);
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_hv_mmx(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ return vp8_sub_pixel_variance16x16_mmx(src_ptr, source_stride, 4, 4,
+ ref_ptr, recon_stride, sse);
+}
diff --git a/libvpx/vp8/common/x86/variance_sse2.c b/libvpx/vp8/common/x86/variance_sse2.c
new file mode 100644
index 0000000..afd6429
--- /dev/null
+++ b/libvpx/vp8/common/x86/variance_sse2.c
@@ -0,0 +1,558 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vp8/common/variance.h"
+#include "vp8/common/pragmas.h"
+#include "vpx_ports/mem.h"
+#include "vp8/common/x86/filter_x86.h"
+
+extern void filter_block1d_h6_mmx(const unsigned char *src_ptr, unsigned short *output_ptr, unsigned int src_pixels_per_line, unsigned int pixel_step, unsigned int output_height, unsigned int output_width, short *filter);
+extern void filter_block1d_v6_mmx(const short *src_ptr, unsigned char *output_ptr, unsigned int pixels_per_line, unsigned int pixel_step, unsigned int output_height, unsigned int output_width, short *filter);
+extern void filter_block1d8_h6_sse2(const unsigned char *src_ptr, unsigned short *output_ptr, unsigned int src_pixels_per_line, unsigned int pixel_step, unsigned int output_height, unsigned int output_width, short *filter);
+extern void filter_block1d8_v6_sse2(const short *src_ptr, unsigned char *output_ptr, unsigned int pixels_per_line, unsigned int pixel_step, unsigned int output_height, unsigned int output_width, short *filter);
+
+extern void vp8_filter_block2d_bil4x4_var_mmx
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ const short *HFilter,
+ const short *VFilter,
+ int *sum,
+ unsigned int *sumsquared
+);
+
+extern unsigned int vp8_get4x4var_mmx
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *SSE,
+ int *Sum
+);
+
+unsigned int vp8_get_mb_ss_sse2
+(
+ const short *src_ptr
+);
+unsigned int vp8_get16x16var_sse2
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *SSE,
+ int *Sum
+);
+unsigned int vp8_get8x8var_sse2
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *SSE,
+ int *Sum
+);
+void vp8_filter_block2d_bil_var_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int xoffset,
+ int yoffset,
+ int *sum,
+ unsigned int *sumsquared
+);
+void vp8_half_horiz_vert_variance8x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+void vp8_half_horiz_vert_variance16x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+void vp8_half_horiz_variance8x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+void vp8_half_horiz_variance16x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+void vp8_half_vert_variance8x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+void vp8_half_vert_variance16x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+
+unsigned int vp8_variance4x4_wmt(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+ vp8_get4x4var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 4));
+
+}
+
+unsigned int vp8_variance8x8_wmt
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int var;
+ int avg;
+
+ vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 6));
+
+}
+
+
+unsigned int vp8_variance16x16_wmt
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int sse0;
+ int sum0;
+
+
+ vp8_get16x16var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
+ *sse = sse0;
+ return (sse0 - ((unsigned int)(sum0 * sum0) >> 8));
+}
+unsigned int vp8_mse16x16_wmt(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+
+ unsigned int sse0;
+ int sum0;
+ vp8_get16x16var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
+ *sse = sse0;
+ return sse0;
+
+}
+
+
+unsigned int vp8_variance16x8_wmt
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int sse0, sse1, var;
+ int sum0, sum1, avg;
+
+ vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
+ vp8_get8x8var_sse2(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
+
+ var = sse0 + sse1;
+ avg = sum0 + sum1;
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 7));
+
+}
+
+unsigned int vp8_variance8x16_wmt
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *sse)
+{
+ unsigned int sse0, sse1, var;
+ int sum0, sum1, avg;
+
+ vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
+ vp8_get8x8var_sse2(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse1, &sum1) ;
+
+ var = sse0 + sse1;
+ avg = sum0 + sum1;
+ *sse = var;
+ return (var - ((unsigned int)(avg * avg) >> 7));
+
+}
+
+unsigned int vp8_sub_pixel_variance4x4_wmt
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ int xsum;
+ unsigned int xxsum;
+ vp8_filter_block2d_bil4x4_var_mmx(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line,
+ vp8_bilinear_filters_x86_4[xoffset], vp8_bilinear_filters_x86_4[yoffset],
+ &xsum, &xxsum
+ );
+ *sse = xxsum;
+ return (xxsum - ((unsigned int)(xsum * xsum) >> 4));
+}
+
+
+unsigned int vp8_sub_pixel_variance8x8_wmt
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ int xsum;
+ unsigned int xxsum;
+
+ if (xoffset == 4 && yoffset == 0)
+ {
+ vp8_half_horiz_variance8x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum, &xxsum);
+ }
+ else if (xoffset == 0 && yoffset == 4)
+ {
+ vp8_half_vert_variance8x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum, &xxsum);
+ }
+ else if (xoffset == 4 && yoffset == 4)
+ {
+ vp8_half_horiz_vert_variance8x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum, &xxsum);
+ }
+ else
+ {
+ vp8_filter_block2d_bil_var_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ xoffset, yoffset,
+ &xsum, &xxsum);
+ }
+
+ *sse = xxsum;
+ return (xxsum - ((unsigned int)(xsum * xsum) >> 6));
+}
+
+unsigned int vp8_sub_pixel_variance16x16_wmt
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ int xsum0, xsum1;
+ unsigned int xxsum0, xxsum1;
+
+
+ /* note we could avoid these if statements if the calling function
+ * just called the appropriate functions inside.
+ */
+ if (xoffset == 4 && yoffset == 0)
+ {
+ vp8_half_horiz_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+ }
+ else if (xoffset == 0 && yoffset == 4)
+ {
+ vp8_half_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+ }
+ else if (xoffset == 4 && yoffset == 4)
+ {
+ vp8_half_horiz_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+ }
+ else
+ {
+ vp8_filter_block2d_bil_var_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ xoffset, yoffset,
+ &xsum0, &xxsum0
+ );
+
+ vp8_filter_block2d_bil_var_sse2(
+ src_ptr + 8, src_pixels_per_line,
+ dst_ptr + 8, dst_pixels_per_line, 16,
+ xoffset, yoffset,
+ &xsum1, &xxsum1
+ );
+ xsum0 += xsum1;
+ xxsum0 += xxsum1;
+ }
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 8));
+}
+
+unsigned int vp8_sub_pixel_mse16x16_wmt(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ vp8_sub_pixel_variance16x16_wmt(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
+ return *sse;
+}
+
+unsigned int vp8_sub_pixel_variance16x8_wmt
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+
+)
+{
+ int xsum0, xsum1;
+ unsigned int xxsum0, xxsum1;
+
+ if (xoffset == 4 && yoffset == 0)
+ {
+ vp8_half_horiz_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum0, &xxsum0);
+ }
+ else if (xoffset == 0 && yoffset == 4)
+ {
+ vp8_half_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum0, &xxsum0);
+ }
+ else if (xoffset == 4 && yoffset == 4)
+ {
+ vp8_half_horiz_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum0, &xxsum0);
+ }
+ else
+ {
+ vp8_filter_block2d_bil_var_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ xoffset, yoffset,
+ &xsum0, &xxsum0);
+
+ vp8_filter_block2d_bil_var_sse2(
+ src_ptr + 8, src_pixels_per_line,
+ dst_ptr + 8, dst_pixels_per_line, 8,
+ xoffset, yoffset,
+ &xsum1, &xxsum1);
+ xsum0 += xsum1;
+ xxsum0 += xxsum1;
+ }
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 7));
+}
+
+unsigned int vp8_sub_pixel_variance8x16_wmt
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ int xsum;
+ unsigned int xxsum;
+
+ if (xoffset == 4 && yoffset == 0)
+ {
+ vp8_half_horiz_variance8x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum, &xxsum);
+ }
+ else if (xoffset == 0 && yoffset == 4)
+ {
+ vp8_half_vert_variance8x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum, &xxsum);
+ }
+ else if (xoffset == 4 && yoffset == 4)
+ {
+ vp8_half_horiz_vert_variance8x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum, &xxsum);
+ }
+ else
+ {
+ vp8_filter_block2d_bil_var_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ xoffset, yoffset,
+ &xsum, &xxsum);
+ }
+
+ *sse = xxsum;
+ return (xxsum - ((unsigned int)(xsum * xsum) >> 7));
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_h_wmt(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse)
+{
+ int xsum0;
+ unsigned int xxsum0;
+
+ vp8_half_horiz_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 8));
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_v_wmt(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse)
+{
+ int xsum0;
+ unsigned int xxsum0;
+ vp8_half_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 8));
+}
+
+
+unsigned int vp8_variance_halfpixvar16x16_hv_wmt(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse)
+{
+ int xsum0;
+ unsigned int xxsum0;
+
+ vp8_half_horiz_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 8));
+}
diff --git a/libvpx/vp8/common/x86/variance_ssse3.c b/libvpx/vp8/common/x86/variance_ssse3.c
new file mode 100644
index 0000000..ba2055c
--- /dev/null
+++ b/libvpx/vp8/common/x86/variance_ssse3.c
@@ -0,0 +1,166 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vp8/common/variance.h"
+#include "vp8/common/pragmas.h"
+#include "vpx_ports/mem.h"
+
+extern unsigned int vp8_get16x16var_sse2
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride,
+ unsigned int *SSE,
+ int *Sum
+);
+extern void vp8_half_horiz_vert_variance16x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+extern void vp8_half_horiz_variance16x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+extern void vp8_half_vert_variance16x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
+extern void vp8_filter_block2d_bil_var_ssse3
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int xoffset,
+ int yoffset,
+ int *sum,
+ unsigned int *sumsquared
+);
+
+unsigned int vp8_sub_pixel_variance16x16_ssse3
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+)
+{
+ int xsum0;
+ unsigned int xxsum0;
+
+ /* note we could avoid these if statements if the calling function
+ * just called the appropriate functions inside.
+ */
+ if (xoffset == 4 && yoffset == 0)
+ {
+ vp8_half_horiz_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+ }
+ else if (xoffset == 0 && yoffset == 4)
+ {
+ vp8_half_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+ }
+ else if (xoffset == 4 && yoffset == 4)
+ {
+ vp8_half_horiz_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ &xsum0, &xxsum0);
+ }
+ else
+ {
+ vp8_filter_block2d_bil_var_ssse3(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 16,
+ xoffset, yoffset,
+ &xsum0, &xxsum0);
+ }
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 8));
+}
+
+unsigned int vp8_sub_pixel_variance16x8_ssse3
+(
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ const unsigned char *dst_ptr,
+ int dst_pixels_per_line,
+ unsigned int *sse
+
+)
+{
+ int xsum0;
+ unsigned int xxsum0;
+
+ if (xoffset == 4 && yoffset == 0)
+ {
+ vp8_half_horiz_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum0, &xxsum0);
+ }
+ else if (xoffset == 0 && yoffset == 4)
+ {
+ vp8_half_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum0, &xxsum0);
+ }
+ else if (xoffset == 4 && yoffset == 4)
+ {
+ vp8_half_horiz_vert_variance16x_h_sse2(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ &xsum0, &xxsum0);
+ }
+ else
+ {
+ vp8_filter_block2d_bil_var_ssse3(
+ src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pixels_per_line, 8,
+ xoffset, yoffset,
+ &xsum0, &xxsum0);
+ }
+
+ *sse = xxsum0;
+ return (xxsum0 - ((unsigned int)(xsum0 * xsum0) >> 7));
+}
diff --git a/libvpx/vp8/common/x86/vp8_asm_stubs.c b/libvpx/vp8/common/x86/vp8_asm_stubs.c
new file mode 100644
index 0000000..3437a23
--- /dev/null
+++ b/libvpx/vp8/common/x86/vp8_asm_stubs.c
@@ -0,0 +1,629 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx_ports/mem.h"
+#include "filter_x86.h"
+
+extern const short vp8_six_tap_mmx[8][6*8];
+
+extern void vp8_filter_block1d_h6_mmx
+(
+ unsigned char *src_ptr,
+ unsigned short *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+);
+extern void vp8_filter_block1dc_v6_mmx
+(
+ unsigned short *src_ptr,
+ unsigned char *output_ptr,
+ int output_pitch,
+ unsigned int pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+);
+extern void vp8_filter_block1d8_h6_sse2
+(
+ unsigned char *src_ptr,
+ unsigned short *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+);
+extern void vp8_filter_block1d16_h6_sse2
+(
+ unsigned char *src_ptr,
+ unsigned short *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+);
+extern void vp8_filter_block1d8_v6_sse2
+(
+ unsigned short *src_ptr,
+ unsigned char *output_ptr,
+ int dst_ptich,
+ unsigned int pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+);
+extern void vp8_filter_block1d16_v6_sse2
+(
+ unsigned short *src_ptr,
+ unsigned char *output_ptr,
+ int dst_ptich,
+ unsigned int pixels_per_line,
+ unsigned int pixel_step,
+ unsigned int output_height,
+ unsigned int output_width,
+ const short *vp8_filter
+);
+extern void vp8_unpack_block1d16_h6_sse2
+(
+ unsigned char *src_ptr,
+ unsigned short *output_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned int output_height,
+ unsigned int output_width
+);
+extern void vp8_filter_block1d8_h6_only_sse2
+(
+ unsigned char *src_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned char *output_ptr,
+ int dst_ptich,
+ unsigned int output_height,
+ const short *vp8_filter
+);
+extern void vp8_filter_block1d16_h6_only_sse2
+(
+ unsigned char *src_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned char *output_ptr,
+ int dst_ptich,
+ unsigned int output_height,
+ const short *vp8_filter
+);
+extern void vp8_filter_block1d8_v6_only_sse2
+(
+ unsigned char *src_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned char *output_ptr,
+ int dst_ptich,
+ unsigned int output_height,
+ const short *vp8_filter
+);
+
+
+#if HAVE_MMX
+void vp8_sixtap_predict4x4_mmx
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 16*16); /* Temp data bufffer used in filtering */
+ const short *HFilter, *VFilter;
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 9, 8, HFilter);
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1dc_v6_mmx(FData2 + 8, dst_ptr, dst_pitch, 8, 4 , 4, 4, VFilter);
+
+}
+
+
+void vp8_sixtap_predict16x16_mmx
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 24*24); /* Temp data bufffer used in filtering */
+
+ const short *HFilter, *VFilter;
+
+
+ HFilter = vp8_six_tap_mmx[xoffset];
+
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 21, 32, HFilter);
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 4, FData2 + 4, src_pixels_per_line, 1, 21, 32, HFilter);
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 8, FData2 + 8, src_pixels_per_line, 1, 21, 32, HFilter);
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 12, FData2 + 12, src_pixels_per_line, 1, 21, 32, HFilter);
+
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1dc_v6_mmx(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, 16, VFilter);
+ vp8_filter_block1dc_v6_mmx(FData2 + 36, dst_ptr + 4, dst_pitch, 32, 16 , 16, 16, VFilter);
+ vp8_filter_block1dc_v6_mmx(FData2 + 40, dst_ptr + 8, dst_pitch, 32, 16 , 16, 16, VFilter);
+ vp8_filter_block1dc_v6_mmx(FData2 + 44, dst_ptr + 12, dst_pitch, 32, 16 , 16, 16, VFilter);
+
+}
+
+
+void vp8_sixtap_predict8x8_mmx
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); /* Temp data bufffer used in filtering */
+
+ const short *HFilter, *VFilter;
+
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 13, 16, HFilter);
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 4, FData2 + 4, src_pixels_per_line, 1, 13, 16, HFilter);
+
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1dc_v6_mmx(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 8, 8, VFilter);
+ vp8_filter_block1dc_v6_mmx(FData2 + 20, dst_ptr + 4, dst_pitch, 16, 8 , 8, 8, VFilter);
+
+}
+
+
+void vp8_sixtap_predict8x4_mmx
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); /* Temp data bufffer used in filtering */
+
+ const short *HFilter, *VFilter;
+
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 9, 16, HFilter);
+ vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 4, FData2 + 4, src_pixels_per_line, 1, 9, 16, HFilter);
+
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1dc_v6_mmx(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 4, 8, VFilter);
+ vp8_filter_block1dc_v6_mmx(FData2 + 20, dst_ptr + 4, dst_pitch, 16, 8 , 4, 8, VFilter);
+
+}
+
+
+
+void vp8_bilinear_predict16x16_mmx
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ vp8_bilinear_predict8x8_mmx(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pitch);
+ vp8_bilinear_predict8x8_mmx(src_ptr + 8, src_pixels_per_line, xoffset, yoffset, dst_ptr + 8, dst_pitch);
+ vp8_bilinear_predict8x8_mmx(src_ptr + 8 * src_pixels_per_line, src_pixels_per_line, xoffset, yoffset, dst_ptr + dst_pitch * 8, dst_pitch);
+ vp8_bilinear_predict8x8_mmx(src_ptr + 8 * src_pixels_per_line + 8, src_pixels_per_line, xoffset, yoffset, dst_ptr + dst_pitch * 8 + 8, dst_pitch);
+}
+#endif
+
+
+#if HAVE_SSE2
+void vp8_sixtap_predict16x16_sse2
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+
+)
+{
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 24*24); /* Temp data bufffer used in filtering */
+
+ const short *HFilter, *VFilter;
+
+ if (xoffset)
+ {
+ if (yoffset)
+ {
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d16_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 21, 32, HFilter);
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1d16_v6_sse2(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, dst_pitch, VFilter);
+ }
+ else
+ {
+ /* First-pass only */
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d16_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 16, HFilter);
+ }
+ }
+ else
+ {
+ /* Second-pass only */
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_unpack_block1d16_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 21, 32);
+ vp8_filter_block1d16_v6_sse2(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, dst_pitch, VFilter);
+ }
+}
+
+
+void vp8_sixtap_predict8x8_sse2
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); /* Temp data bufffer used in filtering */
+ const short *HFilter, *VFilter;
+
+ if (xoffset)
+ {
+ if (yoffset)
+ {
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 13, 16, HFilter);
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 8, dst_pitch, VFilter);
+ }
+ else
+ {
+ /* First-pass only */
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d8_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 8, HFilter);
+ }
+ }
+ else
+ {
+ /* Second-pass only */
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1d8_v6_only_sse2(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 8, VFilter);
+ }
+}
+
+
+void vp8_sixtap_predict8x4_sse2
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); /* Temp data bufffer used in filtering */
+ const short *HFilter, *VFilter;
+
+ if (xoffset)
+ {
+ if (yoffset)
+ {
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 9, 16, HFilter);
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 4, dst_pitch, VFilter);
+ }
+ else
+ {
+ /* First-pass only */
+ HFilter = vp8_six_tap_mmx[xoffset];
+ vp8_filter_block1d8_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 4, HFilter);
+ }
+ }
+ else
+ {
+ /* Second-pass only */
+ VFilter = vp8_six_tap_mmx[yoffset];
+ vp8_filter_block1d8_v6_only_sse2(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 4, VFilter);
+ }
+}
+
+#endif
+
+#if HAVE_SSSE3
+
+extern void vp8_filter_block1d8_h6_ssse3
+(
+ unsigned char *src_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned char *output_ptr,
+ unsigned int output_pitch,
+ unsigned int output_height,
+ unsigned int vp8_filter_index
+);
+
+extern void vp8_filter_block1d16_h6_ssse3
+(
+ unsigned char *src_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned char *output_ptr,
+ unsigned int output_pitch,
+ unsigned int output_height,
+ unsigned int vp8_filter_index
+);
+
+extern void vp8_filter_block1d16_v6_ssse3
+(
+ unsigned char *src_ptr,
+ unsigned int src_pitch,
+ unsigned char *output_ptr,
+ unsigned int out_pitch,
+ unsigned int output_height,
+ unsigned int vp8_filter_index
+);
+
+extern void vp8_filter_block1d8_v6_ssse3
+(
+ unsigned char *src_ptr,
+ unsigned int src_pitch,
+ unsigned char *output_ptr,
+ unsigned int out_pitch,
+ unsigned int output_height,
+ unsigned int vp8_filter_index
+);
+
+extern void vp8_filter_block1d4_h6_ssse3
+(
+ unsigned char *src_ptr,
+ unsigned int src_pixels_per_line,
+ unsigned char *output_ptr,
+ unsigned int output_pitch,
+ unsigned int output_height,
+ unsigned int vp8_filter_index
+);
+
+extern void vp8_filter_block1d4_v6_ssse3
+(
+ unsigned char *src_ptr,
+ unsigned int src_pitch,
+ unsigned char *output_ptr,
+ unsigned int out_pitch,
+ unsigned int output_height,
+ unsigned int vp8_filter_index
+);
+
+void vp8_sixtap_predict16x16_ssse3
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+
+)
+{
+ DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 24*24);
+
+ if (xoffset)
+ {
+ if (yoffset)
+ {
+ vp8_filter_block1d16_h6_ssse3(src_ptr - (2 * src_pixels_per_line),
+ src_pixels_per_line, FData2,
+ 16, 21, xoffset);
+ vp8_filter_block1d16_v6_ssse3(FData2 , 16, dst_ptr, dst_pitch,
+ 16, yoffset);
+ }
+ else
+ {
+ /* First-pass only */
+ vp8_filter_block1d16_h6_ssse3(src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pitch, 16, xoffset);
+ }
+ }
+ else
+ {
+ if (yoffset)
+ {
+ /* Second-pass only */
+ vp8_filter_block1d16_v6_ssse3(src_ptr - (2 * src_pixels_per_line),
+ src_pixels_per_line,
+ dst_ptr, dst_pitch, 16, yoffset);
+ }
+ else
+ {
+ /* ssse3 second-pass only function couldn't handle (xoffset==0 &&
+ * yoffset==0) case correctly. Add copy function here to guarantee
+ * six-tap function handles all possible offsets. */
+ vp8_copy_mem16x16(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch);
+ }
+ }
+}
+
+void vp8_sixtap_predict8x8_ssse3
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 256);
+
+ if (xoffset)
+ {
+ if (yoffset)
+ {
+ vp8_filter_block1d8_h6_ssse3(src_ptr - (2 * src_pixels_per_line),
+ src_pixels_per_line, FData2,
+ 8, 13, xoffset);
+ vp8_filter_block1d8_v6_ssse3(FData2, 8, dst_ptr, dst_pitch,
+ 8, yoffset);
+ }
+ else
+ {
+ vp8_filter_block1d8_h6_ssse3(src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pitch, 8, xoffset);
+ }
+ }
+ else
+ {
+ if (yoffset)
+ {
+ /* Second-pass only */
+ vp8_filter_block1d8_v6_ssse3(src_ptr - (2 * src_pixels_per_line),
+ src_pixels_per_line,
+ dst_ptr, dst_pitch, 8, yoffset);
+ }
+ else
+ {
+ /* ssse3 second-pass only function couldn't handle (xoffset==0 &&
+ * yoffset==0) case correctly. Add copy function here to guarantee
+ * six-tap function handles all possible offsets. */
+ vp8_copy_mem8x8(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch);
+ }
+ }
+}
+
+
+void vp8_sixtap_predict8x4_ssse3
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 256);
+
+ if (xoffset)
+ {
+ if (yoffset)
+ {
+ vp8_filter_block1d8_h6_ssse3(src_ptr - (2 * src_pixels_per_line),
+ src_pixels_per_line, FData2,
+ 8, 9, xoffset);
+ vp8_filter_block1d8_v6_ssse3(FData2, 8, dst_ptr, dst_pitch,
+ 4, yoffset);
+ }
+ else
+ {
+ /* First-pass only */
+ vp8_filter_block1d8_h6_ssse3(src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pitch, 4, xoffset);
+ }
+ }
+ else
+ {
+ if (yoffset)
+ {
+ /* Second-pass only */
+ vp8_filter_block1d8_v6_ssse3(src_ptr - (2 * src_pixels_per_line),
+ src_pixels_per_line,
+ dst_ptr, dst_pitch, 4, yoffset);
+ }
+ else
+ {
+ /* ssse3 second-pass only function couldn't handle (xoffset==0 &&
+ * yoffset==0) case correctly. Add copy function here to guarantee
+ * six-tap function handles all possible offsets. */
+ vp8_copy_mem8x4(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch);
+ }
+ }
+}
+
+void vp8_sixtap_predict4x4_ssse3
+(
+ unsigned char *src_ptr,
+ int src_pixels_per_line,
+ int xoffset,
+ int yoffset,
+ unsigned char *dst_ptr,
+ int dst_pitch
+)
+{
+ DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 4*9);
+
+ if (xoffset)
+ {
+ if (yoffset)
+ {
+ vp8_filter_block1d4_h6_ssse3(src_ptr - (2 * src_pixels_per_line),
+ src_pixels_per_line,
+ FData2, 4, 9, xoffset);
+ vp8_filter_block1d4_v6_ssse3(FData2, 4, dst_ptr, dst_pitch,
+ 4, yoffset);
+ }
+ else
+ {
+ vp8_filter_block1d4_h6_ssse3(src_ptr, src_pixels_per_line,
+ dst_ptr, dst_pitch, 4, xoffset);
+ }
+ }
+ else
+ {
+ if (yoffset)
+ {
+ vp8_filter_block1d4_v6_ssse3(src_ptr - (2 * src_pixels_per_line),
+ src_pixels_per_line,
+ dst_ptr, dst_pitch, 4, yoffset);
+ }
+ else
+ {
+ /* ssse3 second-pass only function couldn't handle (xoffset==0 &&
+ * yoffset==0) case correctly. Add copy function here to guarantee
+ * six-tap function handles all possible offsets. */
+ int r;
+
+ for (r = 0; r < 4; r++)
+ {
+ #if !(CONFIG_FAST_UNALIGNED)
+ dst_ptr[0] = src_ptr[0];
+ dst_ptr[1] = src_ptr[1];
+ dst_ptr[2] = src_ptr[2];
+ dst_ptr[3] = src_ptr[3];
+ #else
+ *(uint32_t *)dst_ptr = *(uint32_t *)src_ptr ;
+ #endif
+ dst_ptr += dst_pitch;
+ src_ptr += src_pixels_per_line;
+ }
+ }
+ }
+}
+
+#endif
diff --git a/libvpx/vp8/decoder/asm_dec_offsets.c b/libvpx/vp8/decoder/asm_dec_offsets.c
new file mode 100644
index 0000000..842a0d5
--- /dev/null
+++ b/libvpx/vp8/decoder/asm_dec_offsets.c
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_ports/asm_offsets.h"
+#include "onyxd_int.h"
+
+BEGIN
+
+DEFINE(bool_decoder_user_buffer_end, offsetof(BOOL_DECODER, user_buffer_end));
+DEFINE(bool_decoder_user_buffer, offsetof(BOOL_DECODER, user_buffer));
+DEFINE(bool_decoder_value, offsetof(BOOL_DECODER, value));
+DEFINE(bool_decoder_count, offsetof(BOOL_DECODER, count));
+DEFINE(bool_decoder_range, offsetof(BOOL_DECODER, range));
+
+END
+
+/* add asserts for any offset that is not supported by assembly code */
+/* add asserts for any size that is not supported by assembly code */
diff --git a/libvpx/vp8/decoder/dboolhuff.c b/libvpx/vp8/decoder/dboolhuff.c
new file mode 100644
index 0000000..7e7b05a
--- /dev/null
+++ b/libvpx/vp8/decoder/dboolhuff.c
@@ -0,0 +1,52 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "dboolhuff.h"
+#include "vpx_ports/mem.h"
+#include "vpx_mem/vpx_mem.h"
+
+int vp8dx_start_decode(BOOL_DECODER *br,
+ const unsigned char *source,
+ unsigned int source_sz)
+{
+ br->user_buffer_end = source+source_sz;
+ br->user_buffer = source;
+ br->value = 0;
+ br->count = -8;
+ br->range = 255;
+
+ if (source_sz && !source)
+ return 1;
+
+ /* Populate the buffer */
+ vp8dx_bool_decoder_fill(br);
+
+ return 0;
+}
+
+
+void vp8dx_bool_decoder_fill(BOOL_DECODER *br)
+{
+ const unsigned char *bufptr;
+ const unsigned char *bufend;
+ VP8_BD_VALUE value;
+ int count;
+ bufend = br->user_buffer_end;
+ bufptr = br->user_buffer;
+ value = br->value;
+ count = br->count;
+
+ VP8DX_BOOL_DECODER_FILL(count, value, bufptr, bufend);
+
+ br->user_buffer = bufptr;
+ br->value = value;
+ br->count = count;
+}
diff --git a/libvpx/vp8/decoder/dboolhuff.h b/libvpx/vp8/decoder/dboolhuff.h
new file mode 100644
index 0000000..1a08c05
--- /dev/null
+++ b/libvpx/vp8/decoder/dboolhuff.h
@@ -0,0 +1,154 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef DBOOLHUFF_H
+#define DBOOLHUFF_H
+#include <stddef.h>
+#include <limits.h>
+#include "vpx_config.h"
+#include "vpx_ports/mem.h"
+#include "vpx/vpx_integer.h"
+
+typedef size_t VP8_BD_VALUE;
+
+# define VP8_BD_VALUE_SIZE ((int)sizeof(VP8_BD_VALUE)*CHAR_BIT)
+/*This is meant to be a large, positive constant that can still be efficiently
+ loaded as an immediate (on platforms like ARM, for example).
+ Even relatively modest values like 100 would work fine.*/
+# define VP8_LOTS_OF_BITS (0x40000000)
+
+typedef struct
+{
+ const unsigned char *user_buffer_end;
+ const unsigned char *user_buffer;
+ VP8_BD_VALUE value;
+ int count;
+ unsigned int range;
+} BOOL_DECODER;
+
+DECLARE_ALIGNED(16, extern const unsigned char, vp8_norm[256]);
+
+int vp8dx_start_decode(BOOL_DECODER *br,
+ const unsigned char *source,
+ unsigned int source_sz);
+
+void vp8dx_bool_decoder_fill(BOOL_DECODER *br);
+
+/*The refill loop is used in several places, so define it in a macro to make
+ sure they're all consistent.
+ An inline function would be cleaner, but has a significant penalty, because
+ multiple BOOL_DECODER fields must be modified, and the compiler is not smart
+ enough to eliminate the stores to those fields and the subsequent reloads
+ from them when inlining the function.*/
+#define VP8DX_BOOL_DECODER_FILL(_count,_value,_bufptr,_bufend) \
+ do \
+ { \
+ int shift = VP8_BD_VALUE_SIZE - 8 - ((_count) + 8); \
+ int loop_end, x; \
+ size_t bits_left = ((_bufend)-(_bufptr))*CHAR_BIT; \
+ \
+ x = (int)(shift + CHAR_BIT - bits_left); \
+ loop_end = 0; \
+ if(x >= 0) \
+ { \
+ (_count) += VP8_LOTS_OF_BITS; \
+ loop_end = x; \
+ if(!bits_left) break; \
+ } \
+ while(shift >= loop_end) \
+ { \
+ (_count) += CHAR_BIT; \
+ (_value) |= (VP8_BD_VALUE)*(_bufptr)++ << shift; \
+ shift -= CHAR_BIT; \
+ } \
+ } \
+ while(0) \
+
+
+static int vp8dx_decode_bool(BOOL_DECODER *br, int probability) {
+ unsigned int bit = 0;
+ VP8_BD_VALUE value;
+ unsigned int split;
+ VP8_BD_VALUE bigsplit;
+ int count;
+ unsigned int range;
+
+ split = 1 + (((br->range - 1) * probability) >> 8);
+
+ if(br->count < 0)
+ vp8dx_bool_decoder_fill(br);
+
+ value = br->value;
+ count = br->count;
+
+ bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8);
+
+ range = split;
+
+ if (value >= bigsplit)
+ {
+ range = br->range - split;
+ value = value - bigsplit;
+ bit = 1;
+ }
+
+ {
+ register unsigned int shift = vp8_norm[range];
+ range <<= shift;
+ value <<= shift;
+ count -= shift;
+ }
+ br->value = value;
+ br->count = count;
+ br->range = range;
+
+ return bit;
+}
+
+static int vp8_decode_value(BOOL_DECODER *br, int bits)
+{
+ int z = 0;
+ int bit;
+
+ for (bit = bits - 1; bit >= 0; bit--)
+ {
+ z |= (vp8dx_decode_bool(br, 0x80) << bit);
+ }
+
+ return z;
+}
+
+static int vp8dx_bool_error(BOOL_DECODER *br)
+{
+ /* Check if we have reached the end of the buffer.
+ *
+ * Variable 'count' stores the number of bits in the 'value' buffer, minus
+ * 8. The top byte is part of the algorithm, and the remainder is buffered
+ * to be shifted into it. So if count == 8, the top 16 bits of 'value' are
+ * occupied, 8 for the algorithm and 8 in the buffer.
+ *
+ * When reading a byte from the user's buffer, count is filled with 8 and
+ * one byte is filled into the value buffer. When we reach the end of the
+ * data, count is additionally filled with VP8_LOTS_OF_BITS. So when
+ * count == VP8_LOTS_OF_BITS - 1, the user's data has been exhausted.
+ */
+ if ((br->count > VP8_BD_VALUE_SIZE) && (br->count < VP8_LOTS_OF_BITS))
+ {
+ /* We have tried to decode bits after the end of
+ * stream was encountered.
+ */
+ return 1;
+ }
+
+ /* No error. */
+ return 0;
+}
+#endif
diff --git a/libvpx/vp8/decoder/decodemv.c b/libvpx/vp8/decoder/decodemv.c
new file mode 100644
index 0000000..8027a07
--- /dev/null
+++ b/libvpx/vp8/decoder/decodemv.c
@@ -0,0 +1,668 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "treereader.h"
+#include "vp8/common/entropymv.h"
+#include "vp8/common/entropymode.h"
+#include "onyxd_int.h"
+#include "vp8/common/findnearmv.h"
+
+#if CONFIG_DEBUG
+#include <assert.h>
+#endif
+static B_PREDICTION_MODE read_bmode(vp8_reader *bc, const vp8_prob *p)
+{
+ const int i = vp8_treed_read(bc, vp8_bmode_tree, p);
+
+ return (B_PREDICTION_MODE)i;
+}
+
+static MB_PREDICTION_MODE read_ymode(vp8_reader *bc, const vp8_prob *p)
+{
+ const int i = vp8_treed_read(bc, vp8_ymode_tree, p);
+
+ return (MB_PREDICTION_MODE)i;
+}
+
+static MB_PREDICTION_MODE read_kf_ymode(vp8_reader *bc, const vp8_prob *p)
+{
+ const int i = vp8_treed_read(bc, vp8_kf_ymode_tree, p);
+
+ return (MB_PREDICTION_MODE)i;
+}
+
+static MB_PREDICTION_MODE read_uv_mode(vp8_reader *bc, const vp8_prob *p)
+{
+ const int i = vp8_treed_read(bc, vp8_uv_mode_tree, p);
+
+ return (MB_PREDICTION_MODE)i;
+}
+
+static void read_kf_modes(VP8D_COMP *pbi, MODE_INFO *mi)
+{
+ vp8_reader *const bc = & pbi->mbc[8];
+ const int mis = pbi->common.mode_info_stride;
+
+ mi->mbmi.ref_frame = INTRA_FRAME;
+ mi->mbmi.mode = read_kf_ymode(bc, vp8_kf_ymode_prob);
+
+ if (mi->mbmi.mode == B_PRED)
+ {
+ int i = 0;
+ mi->mbmi.is_4x4 = 1;
+
+ do
+ {
+ const B_PREDICTION_MODE A = above_block_mode(mi, i, mis);
+ const B_PREDICTION_MODE L = left_block_mode(mi, i);
+
+ mi->bmi[i].as_mode =
+ read_bmode(bc, vp8_kf_bmode_prob [A] [L]);
+ }
+ while (++i < 16);
+ }
+
+ mi->mbmi.uv_mode = read_uv_mode(bc, vp8_kf_uv_mode_prob);
+}
+
+static int read_mvcomponent(vp8_reader *r, const MV_CONTEXT *mvc)
+{
+ const vp8_prob *const p = (const vp8_prob *) mvc;
+ int x = 0;
+
+ if (vp8_read(r, p [mvpis_short])) /* Large */
+ {
+ int i = 0;
+
+ do
+ {
+ x += vp8_read(r, p [MVPbits + i]) << i;
+ }
+ while (++i < 3);
+
+ i = mvlong_width - 1; /* Skip bit 3, which is sometimes implicit */
+
+ do
+ {
+ x += vp8_read(r, p [MVPbits + i]) << i;
+ }
+ while (--i > 3);
+
+ if (!(x & 0xFFF0) || vp8_read(r, p [MVPbits + 3]))
+ x += 8;
+ }
+ else /* small */
+ x = vp8_treed_read(r, vp8_small_mvtree, p + MVPshort);
+
+ if (x && vp8_read(r, p [MVPsign]))
+ x = -x;
+
+ return x;
+}
+
+static void read_mv(vp8_reader *r, MV *mv, const MV_CONTEXT *mvc)
+{
+ mv->row = (short)(read_mvcomponent(r, mvc) << 1);
+ mv->col = (short)(read_mvcomponent(r, ++mvc) << 1);
+}
+
+
+static void read_mvcontexts(vp8_reader *bc, MV_CONTEXT *mvc)
+{
+ int i = 0;
+
+ do
+ {
+ const vp8_prob *up = vp8_mv_update_probs[i].prob;
+ vp8_prob *p = (vp8_prob *)(mvc + i);
+ vp8_prob *const pstop = p + MVPcount;
+
+ do
+ {
+ if (vp8_read(bc, *up++))
+ {
+ const vp8_prob x = (vp8_prob)vp8_read_literal(bc, 7);
+
+ *p = x ? x << 1 : 1;
+ }
+ }
+ while (++p < pstop);
+ }
+ while (++i < 2);
+}
+
+static const unsigned char mbsplit_fill_count[4] = {8, 8, 4, 1};
+static const unsigned char mbsplit_fill_offset[4][16] = {
+ { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15},
+ { 0, 1, 4, 5, 8, 9, 12, 13, 2, 3, 6, 7, 10, 11, 14, 15},
+ { 0, 1, 4, 5, 2, 3, 6, 7, 8, 9, 12, 13, 10, 11, 14, 15},
+ { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}
+};
+
+
+static void mb_mode_mv_init(VP8D_COMP *pbi)
+{
+ vp8_reader *const bc = & pbi->mbc[8];
+ MV_CONTEXT *const mvc = pbi->common.fc.mvc;
+
+#if CONFIG_ERROR_CONCEALMENT
+ /* Default is that no macroblock is corrupt, therefore we initialize
+ * mvs_corrupt_from_mb to something very big, which we can be sure is
+ * outside the frame. */
+ pbi->mvs_corrupt_from_mb = UINT_MAX;
+#endif
+ /* Read the mb_no_coeff_skip flag */
+ pbi->common.mb_no_coeff_skip = (int)vp8_read_bit(bc);
+
+ pbi->prob_skip_false = 0;
+ if (pbi->common.mb_no_coeff_skip)
+ pbi->prob_skip_false = (vp8_prob)vp8_read_literal(bc, 8);
+
+ if(pbi->common.frame_type != KEY_FRAME)
+ {
+ pbi->prob_intra = (vp8_prob)vp8_read_literal(bc, 8);
+ pbi->prob_last = (vp8_prob)vp8_read_literal(bc, 8);
+ pbi->prob_gf = (vp8_prob)vp8_read_literal(bc, 8);
+
+ if (vp8_read_bit(bc))
+ {
+ int i = 0;
+
+ do
+ {
+ pbi->common.fc.ymode_prob[i] =
+ (vp8_prob) vp8_read_literal(bc, 8);
+ }
+ while (++i < 4);
+ }
+
+ if (vp8_read_bit(bc))
+ {
+ int i = 0;
+
+ do
+ {
+ pbi->common.fc.uv_mode_prob[i] =
+ (vp8_prob) vp8_read_literal(bc, 8);
+ }
+ while (++i < 3);
+ }
+
+ read_mvcontexts(bc, mvc);
+ }
+}
+
+const vp8_prob vp8_sub_mv_ref_prob3 [8][VP8_SUBMVREFS-1] =
+{
+ { 147, 136, 18 }, /* SUBMVREF_NORMAL */
+ { 223, 1 , 34 }, /* SUBMVREF_LEFT_ABOVE_SAME */
+ { 106, 145, 1 }, /* SUBMVREF_LEFT_ZED */
+ { 208, 1 , 1 }, /* SUBMVREF_LEFT_ABOVE_ZED */
+ { 179, 121, 1 }, /* SUBMVREF_ABOVE_ZED */
+ { 223, 1 , 34 }, /* SUBMVREF_LEFT_ABOVE_SAME */
+ { 179, 121, 1 }, /* SUBMVREF_ABOVE_ZED */
+ { 208, 1 , 1 } /* SUBMVREF_LEFT_ABOVE_ZED */
+};
+
+static
+const vp8_prob * get_sub_mv_ref_prob(const int left, const int above)
+{
+ int lez = (left == 0);
+ int aez = (above == 0);
+ int lea = (left == above);
+ const vp8_prob * prob;
+
+ prob = vp8_sub_mv_ref_prob3[(aez << 2) |
+ (lez << 1) |
+ (lea)];
+
+ return prob;
+}
+
+static void decode_split_mv(vp8_reader *const bc, MODE_INFO *mi,
+ const MODE_INFO *left_mb, const MODE_INFO *above_mb,
+ MB_MODE_INFO *mbmi, int_mv best_mv,
+ MV_CONTEXT *const mvc, int mb_to_left_edge,
+ int mb_to_right_edge, int mb_to_top_edge,
+ int mb_to_bottom_edge)
+{
+ int s; /* split configuration (16x8, 8x16, 8x8, 4x4) */
+ int num_p; /* number of partitions in the split configuration
+ (see vp8_mbsplit_count) */
+ int j = 0;
+
+ s = 3;
+ num_p = 16;
+ if( vp8_read(bc, 110) )
+ {
+ s = 2;
+ num_p = 4;
+ if( vp8_read(bc, 111) )
+ {
+ s = vp8_read(bc, 150);
+ num_p = 2;
+ }
+ }
+
+ do /* for each subset j */
+ {
+ int_mv leftmv, abovemv;
+ int_mv blockmv;
+ int k; /* first block in subset j */
+
+ const vp8_prob *prob;
+ k = vp8_mbsplit_offset[s][j];
+
+ if (!(k & 3))
+ {
+ /* On L edge, get from MB to left of us */
+ if(left_mb->mbmi.mode != SPLITMV)
+ leftmv.as_int = left_mb->mbmi.mv.as_int;
+ else
+ leftmv.as_int = (left_mb->bmi + k + 4 - 1)->mv.as_int;
+ }
+ else
+ leftmv.as_int = (mi->bmi + k - 1)->mv.as_int;
+
+ if (!(k >> 2))
+ {
+ /* On top edge, get from MB above us */
+ if(above_mb->mbmi.mode != SPLITMV)
+ abovemv.as_int = above_mb->mbmi.mv.as_int;
+ else
+ abovemv.as_int = (above_mb->bmi + k + 16 - 4)->mv.as_int;
+ }
+ else
+ abovemv.as_int = (mi->bmi + k - 4)->mv.as_int;
+
+ prob = get_sub_mv_ref_prob(leftmv.as_int, abovemv.as_int);
+
+ if( vp8_read(bc, prob[0]) )
+ {
+ if( vp8_read(bc, prob[1]) )
+ {
+ blockmv.as_int = 0;
+ if( vp8_read(bc, prob[2]) )
+ {
+ blockmv.as_mv.row = read_mvcomponent(bc, &mvc[0]) << 1;
+ blockmv.as_mv.row += best_mv.as_mv.row;
+ blockmv.as_mv.col = read_mvcomponent(bc, &mvc[1]) << 1;
+ blockmv.as_mv.col += best_mv.as_mv.col;
+ }
+ }
+ else
+ {
+ blockmv.as_int = abovemv.as_int;
+ }
+ }
+ else
+ {
+ blockmv.as_int = leftmv.as_int;
+ }
+
+ mbmi->need_to_clamp_mvs |= vp8_check_mv_bounds(&blockmv,
+ mb_to_left_edge,
+ mb_to_right_edge,
+ mb_to_top_edge,
+ mb_to_bottom_edge);
+
+ {
+ /* Fill (uniform) modes, mvs of jth subset.
+ Must do it here because ensuing subsets can
+ refer back to us via "left" or "above". */
+ const unsigned char *fill_offset;
+ unsigned int fill_count = mbsplit_fill_count[s];
+
+ fill_offset = &mbsplit_fill_offset[s]
+ [(unsigned char)j * mbsplit_fill_count[s]];
+
+ do {
+ mi->bmi[ *fill_offset].mv.as_int = blockmv.as_int;
+ fill_offset++;
+ }while (--fill_count);
+ }
+
+ }
+ while (++j < num_p);
+
+ mbmi->partitioning = s;
+}
+
+static void read_mb_modes_mv(VP8D_COMP *pbi, MODE_INFO *mi, MB_MODE_INFO *mbmi)
+{
+ vp8_reader *const bc = & pbi->mbc[8];
+ mbmi->ref_frame = (MV_REFERENCE_FRAME) vp8_read(bc, pbi->prob_intra);
+ if (mbmi->ref_frame) /* inter MB */
+ {
+ enum {CNT_INTRA, CNT_NEAREST, CNT_NEAR, CNT_SPLITMV};
+ int cnt[4];
+ int *cntx = cnt;
+ int_mv near_mvs[4];
+ int_mv *nmv = near_mvs;
+ const int mis = pbi->mb.mode_info_stride;
+ const MODE_INFO *above = mi - mis;
+ const MODE_INFO *left = mi - 1;
+ const MODE_INFO *aboveleft = above - 1;
+ int *ref_frame_sign_bias = pbi->common.ref_frame_sign_bias;
+
+ mbmi->need_to_clamp_mvs = 0;
+
+ if (vp8_read(bc, pbi->prob_last))
+ {
+ mbmi->ref_frame =
+ (MV_REFERENCE_FRAME)((int)(2 + vp8_read(bc, pbi->prob_gf)));
+ }
+
+ /* Zero accumulators */
+ nmv[0].as_int = nmv[1].as_int = nmv[2].as_int = 0;
+ cnt[0] = cnt[1] = cnt[2] = cnt[3] = 0;
+
+ /* Process above */
+ if (above->mbmi.ref_frame != INTRA_FRAME)
+ {
+ if (above->mbmi.mv.as_int)
+ {
+ (++nmv)->as_int = above->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[above->mbmi.ref_frame],
+ mbmi->ref_frame, nmv, ref_frame_sign_bias);
+ ++cntx;
+ }
+
+ *cntx += 2;
+ }
+
+ /* Process left */
+ if (left->mbmi.ref_frame != INTRA_FRAME)
+ {
+ if (left->mbmi.mv.as_int)
+ {
+ int_mv this_mv;
+
+ this_mv.as_int = left->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[left->mbmi.ref_frame],
+ mbmi->ref_frame, &this_mv, ref_frame_sign_bias);
+
+ if (this_mv.as_int != nmv->as_int)
+ {
+ (++nmv)->as_int = this_mv.as_int;
+ ++cntx;
+ }
+
+ *cntx += 2;
+ }
+ else
+ cnt[CNT_INTRA] += 2;
+ }
+
+ /* Process above left */
+ if (aboveleft->mbmi.ref_frame != INTRA_FRAME)
+ {
+ if (aboveleft->mbmi.mv.as_int)
+ {
+ int_mv this_mv;
+
+ this_mv.as_int = aboveleft->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[aboveleft->mbmi.ref_frame],
+ mbmi->ref_frame, &this_mv, ref_frame_sign_bias);
+
+ if (this_mv.as_int != nmv->as_int)
+ {
+ (++nmv)->as_int = this_mv.as_int;
+ ++cntx;
+ }
+
+ *cntx += 1;
+ }
+ else
+ cnt[CNT_INTRA] += 1;
+ }
+
+ if( vp8_read(bc, vp8_mode_contexts [cnt[CNT_INTRA]] [0]) )
+ {
+
+ /* If we have three distinct MV's ... */
+ /* See if above-left MV can be merged with NEAREST */
+ cnt[CNT_NEAREST] += ( (cnt[CNT_SPLITMV] > 0) &
+ (nmv->as_int == near_mvs[CNT_NEAREST].as_int));
+
+ /* Swap near and nearest if necessary */
+ if (cnt[CNT_NEAR] > cnt[CNT_NEAREST])
+ {
+ int tmp;
+ tmp = cnt[CNT_NEAREST];
+ cnt[CNT_NEAREST] = cnt[CNT_NEAR];
+ cnt[CNT_NEAR] = tmp;
+ tmp = near_mvs[CNT_NEAREST].as_int;
+ near_mvs[CNT_NEAREST].as_int = near_mvs[CNT_NEAR].as_int;
+ near_mvs[CNT_NEAR].as_int = tmp;
+ }
+
+ if( vp8_read(bc, vp8_mode_contexts [cnt[CNT_NEAREST]] [1]) )
+ {
+
+ if( vp8_read(bc, vp8_mode_contexts [cnt[CNT_NEAR]] [2]) )
+ {
+ int mb_to_top_edge;
+ int mb_to_bottom_edge;
+ int mb_to_left_edge;
+ int mb_to_right_edge;
+ MV_CONTEXT *const mvc = pbi->common.fc.mvc;
+ int near_index;
+
+ mb_to_top_edge = pbi->mb.mb_to_top_edge;
+ mb_to_bottom_edge = pbi->mb.mb_to_bottom_edge;
+ mb_to_top_edge -= LEFT_TOP_MARGIN;
+ mb_to_bottom_edge += RIGHT_BOTTOM_MARGIN;
+ mb_to_right_edge = pbi->mb.mb_to_right_edge;
+ mb_to_right_edge += RIGHT_BOTTOM_MARGIN;
+ mb_to_left_edge = pbi->mb.mb_to_left_edge;
+ mb_to_left_edge -= LEFT_TOP_MARGIN;
+
+ /* Use near_mvs[0] to store the "best" MV */
+ near_index = CNT_INTRA +
+ (cnt[CNT_NEAREST] >= cnt[CNT_INTRA]);
+
+ vp8_clamp_mv2(&near_mvs[near_index], &pbi->mb);
+
+ cnt[CNT_SPLITMV] = ((above->mbmi.mode == SPLITMV)
+ + (left->mbmi.mode == SPLITMV)) * 2
+ + (aboveleft->mbmi.mode == SPLITMV);
+
+ if( vp8_read(bc, vp8_mode_contexts [cnt[CNT_SPLITMV]] [3]) )
+ {
+ decode_split_mv(bc, mi, left, above,
+ mbmi,
+ near_mvs[near_index],
+ mvc, mb_to_left_edge,
+ mb_to_right_edge,
+ mb_to_top_edge,
+ mb_to_bottom_edge);
+ mbmi->mv.as_int = mi->bmi[15].mv.as_int;
+ mbmi->mode = SPLITMV;
+ mbmi->is_4x4 = 1;
+ }
+ else
+ {
+ int_mv *const mbmi_mv = & mbmi->mv;
+ read_mv(bc, &mbmi_mv->as_mv, (const MV_CONTEXT *) mvc);
+ mbmi_mv->as_mv.row += near_mvs[near_index].as_mv.row;
+ mbmi_mv->as_mv.col += near_mvs[near_index].as_mv.col;
+
+ /* Don't need to check this on NEARMV and NEARESTMV
+ * modes since those modes clamp the MV. The NEWMV mode
+ * does not, so signal to the prediction stage whether
+ * special handling may be required.
+ */
+ mbmi->need_to_clamp_mvs =
+ vp8_check_mv_bounds(mbmi_mv, mb_to_left_edge,
+ mb_to_right_edge,
+ mb_to_top_edge,
+ mb_to_bottom_edge);
+ mbmi->mode = NEWMV;
+ }
+ }
+ else
+ {
+ mbmi->mode = NEARMV;
+ vp8_clamp_mv2(&near_mvs[CNT_NEAR], &pbi->mb);
+ mbmi->mv.as_int = near_mvs[CNT_NEAR].as_int;
+ }
+ }
+ else
+ {
+ mbmi->mode = NEARESTMV;
+ vp8_clamp_mv2(&near_mvs[CNT_NEAREST], &pbi->mb);
+ mbmi->mv.as_int = near_mvs[CNT_NEAREST].as_int;
+ }
+ }
+ else
+ {
+ mbmi->mode = ZEROMV;
+ mbmi->mv.as_int = 0;
+ }
+
+#if CONFIG_ERROR_CONCEALMENT
+ if(pbi->ec_enabled && (mbmi->mode != SPLITMV))
+ {
+ mi->bmi[ 0].mv.as_int =
+ mi->bmi[ 1].mv.as_int =
+ mi->bmi[ 2].mv.as_int =
+ mi->bmi[ 3].mv.as_int =
+ mi->bmi[ 4].mv.as_int =
+ mi->bmi[ 5].mv.as_int =
+ mi->bmi[ 6].mv.as_int =
+ mi->bmi[ 7].mv.as_int =
+ mi->bmi[ 8].mv.as_int =
+ mi->bmi[ 9].mv.as_int =
+ mi->bmi[10].mv.as_int =
+ mi->bmi[11].mv.as_int =
+ mi->bmi[12].mv.as_int =
+ mi->bmi[13].mv.as_int =
+ mi->bmi[14].mv.as_int =
+ mi->bmi[15].mv.as_int = mbmi->mv.as_int;
+ }
+#endif
+ }
+ else
+ {
+ /* required for left and above block mv */
+ mbmi->mv.as_int = 0;
+
+ /* MB is intra coded */
+ if ((mbmi->mode = read_ymode(bc, pbi->common.fc.ymode_prob)) == B_PRED)
+ {
+ int j = 0;
+ mbmi->is_4x4 = 1;
+ do
+ {
+ mi->bmi[j].as_mode = read_bmode(bc, pbi->common.fc.bmode_prob);
+ }
+ while (++j < 16);
+ }
+
+ mbmi->uv_mode = read_uv_mode(bc, pbi->common.fc.uv_mode_prob);
+ }
+
+}
+
+static void read_mb_features(vp8_reader *r, MB_MODE_INFO *mi, MACROBLOCKD *x)
+{
+ /* Is segmentation enabled */
+ if (x->segmentation_enabled && x->update_mb_segmentation_map)
+ {
+ /* If so then read the segment id. */
+ if (vp8_read(r, x->mb_segment_tree_probs[0]))
+ mi->segment_id =
+ (unsigned char)(2 + vp8_read(r, x->mb_segment_tree_probs[2]));
+ else
+ mi->segment_id =
+ (unsigned char)(vp8_read(r, x->mb_segment_tree_probs[1]));
+ }
+}
+
+static void decode_mb_mode_mvs(VP8D_COMP *pbi, MODE_INFO *mi,
+ MB_MODE_INFO *mbmi)
+{
+ /* Read the Macroblock segmentation map if it is being updated explicitly
+ * this frame (reset to 0 above by default)
+ * By default on a key frame reset all MBs to segment 0
+ */
+ if (pbi->mb.update_mb_segmentation_map)
+ read_mb_features(&pbi->mbc[8], &mi->mbmi, &pbi->mb);
+ else if(pbi->common.frame_type == KEY_FRAME)
+ mi->mbmi.segment_id = 0;
+
+ /* Read the macroblock coeff skip flag if this feature is in use,
+ * else default to 0 */
+ if (pbi->common.mb_no_coeff_skip)
+ mi->mbmi.mb_skip_coeff = vp8_read(&pbi->mbc[8], pbi->prob_skip_false);
+ else
+ mi->mbmi.mb_skip_coeff = 0;
+
+ mi->mbmi.is_4x4 = 0;
+ if(pbi->common.frame_type == KEY_FRAME)
+ read_kf_modes(pbi, mi);
+ else
+ read_mb_modes_mv(pbi, mi, &mi->mbmi);
+
+}
+
+void vp8_decode_mode_mvs(VP8D_COMP *pbi)
+{
+ MODE_INFO *mi = pbi->common.mi;
+ int mb_row = -1;
+ int mb_to_right_edge_start;
+
+ mb_mode_mv_init(pbi);
+
+ pbi->mb.mb_to_top_edge = 0;
+ pbi->mb.mb_to_bottom_edge = ((pbi->common.mb_rows - 1) * 16) << 3;
+ mb_to_right_edge_start = ((pbi->common.mb_cols - 1) * 16) << 3;
+
+ while (++mb_row < pbi->common.mb_rows)
+ {
+ int mb_col = -1;
+
+ pbi->mb.mb_to_left_edge = 0;
+ pbi->mb.mb_to_right_edge = mb_to_right_edge_start;
+
+ while (++mb_col < pbi->common.mb_cols)
+ {
+#if CONFIG_ERROR_CONCEALMENT
+ int mb_num = mb_row * pbi->common.mb_cols + mb_col;
+#endif
+
+ decode_mb_mode_mvs(pbi, mi, &mi->mbmi);
+
+#if CONFIG_ERROR_CONCEALMENT
+ /* look for corruption. set mvs_corrupt_from_mb to the current
+ * mb_num if the frame is corrupt from this macroblock. */
+ if (vp8dx_bool_error(&pbi->mbc[8]) && mb_num <
+ (int)pbi->mvs_corrupt_from_mb)
+ {
+ pbi->mvs_corrupt_from_mb = mb_num;
+ /* no need to continue since the partition is corrupt from
+ * here on.
+ */
+ return;
+ }
+#endif
+
+ pbi->mb.mb_to_left_edge -= (16 << 3);
+ pbi->mb.mb_to_right_edge -= (16 << 3);
+ mi++; /* next macroblock */
+ }
+ pbi->mb.mb_to_top_edge -= (16 << 3);
+ pbi->mb.mb_to_bottom_edge -= (16 << 3);
+
+ mi++; /* skip left predictor each row */
+ }
+}
diff --git a/vp8/decoder/decodemv.h b/libvpx/vp8/decoder/decodemv.h
similarity index 100%
rename from vp8/decoder/decodemv.h
rename to libvpx/vp8/decoder/decodemv.h
diff --git a/vp8/decoder/decoderthreading.h b/libvpx/vp8/decoder/decoderthreading.h
similarity index 100%
rename from vp8/decoder/decoderthreading.h
rename to libvpx/vp8/decoder/decoderthreading.h
diff --git a/libvpx/vp8/decoder/decodframe.c b/libvpx/vp8/decoder/decodframe.c
new file mode 100644
index 0000000..a4a00f6
--- /dev/null
+++ b/libvpx/vp8/decoder/decodframe.c
@@ -0,0 +1,1385 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "onyxd_int.h"
+#include "vp8/common/header.h"
+#include "vp8/common/reconintra4x4.h"
+#include "vp8/common/reconinter.h"
+#include "detokenize.h"
+#include "vp8/common/invtrans.h"
+#include "vp8/common/alloccommon.h"
+#include "vp8/common/entropymode.h"
+#include "vp8/common/quant_common.h"
+#include "vpx_scale/vpxscale.h"
+#include "vp8/common/setupintrarecon.h"
+
+#include "decodemv.h"
+#include "vp8/common/extend.h"
+#if CONFIG_ERROR_CONCEALMENT
+#include "error_concealment.h"
+#endif
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/threading.h"
+#include "decoderthreading.h"
+#include "dboolhuff.h"
+
+#include <assert.h>
+#include <stdio.h>
+
+void vp8cx_init_de_quantizer(VP8D_COMP *pbi)
+{
+ int Q;
+ VP8_COMMON *const pc = & pbi->common;
+
+ for (Q = 0; Q < QINDEX_RANGE; Q++)
+ {
+ pc->Y1dequant[Q][0] = (short)vp8_dc_quant(Q, pc->y1dc_delta_q);
+ pc->Y2dequant[Q][0] = (short)vp8_dc2quant(Q, pc->y2dc_delta_q);
+ pc->UVdequant[Q][0] = (short)vp8_dc_uv_quant(Q, pc->uvdc_delta_q);
+
+ pc->Y1dequant[Q][1] = (short)vp8_ac_yquant(Q);
+ pc->Y2dequant[Q][1] = (short)vp8_ac2quant(Q, pc->y2ac_delta_q);
+ pc->UVdequant[Q][1] = (short)vp8_ac_uv_quant(Q, pc->uvac_delta_q);
+ }
+}
+
+void vp8_mb_init_dequantizer(VP8D_COMP *pbi, MACROBLOCKD *xd)
+{
+ int i;
+ int QIndex;
+ MB_MODE_INFO *mbmi = &xd->mode_info_context->mbmi;
+ VP8_COMMON *const pc = & pbi->common;
+
+ /* Decide whether to use the default or alternate baseline Q value. */
+ if (xd->segmentation_enabled)
+ {
+ /* Abs Value */
+ if (xd->mb_segement_abs_delta == SEGMENT_ABSDATA)
+ QIndex = xd->segment_feature_data[MB_LVL_ALT_Q][mbmi->segment_id];
+
+ /* Delta Value */
+ else
+ {
+ QIndex = pc->base_qindex + xd->segment_feature_data[MB_LVL_ALT_Q][mbmi->segment_id];
+ QIndex = (QIndex >= 0) ? ((QIndex <= MAXQ) ? QIndex : MAXQ) : 0; /* Clamp to valid range */
+ }
+ }
+ else
+ QIndex = pc->base_qindex;
+
+ /* Set up the macroblock dequant constants */
+ xd->dequant_y1_dc[0] = 1;
+ xd->dequant_y1[0] = pc->Y1dequant[QIndex][0];
+ xd->dequant_y2[0] = pc->Y2dequant[QIndex][0];
+ xd->dequant_uv[0] = pc->UVdequant[QIndex][0];
+
+ for (i = 1; i < 16; i++)
+ {
+ xd->dequant_y1_dc[i] =
+ xd->dequant_y1[i] = pc->Y1dequant[QIndex][1];
+ xd->dequant_y2[i] = pc->Y2dequant[QIndex][1];
+ xd->dequant_uv[i] = pc->UVdequant[QIndex][1];
+ }
+}
+
+static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
+ unsigned int mb_idx)
+{
+ MB_PREDICTION_MODE mode;
+ int i;
+#if CONFIG_ERROR_CONCEALMENT
+ int corruption_detected = 0;
+#endif
+
+ if (xd->mode_info_context->mbmi.mb_skip_coeff)
+ {
+ vp8_reset_mb_tokens_context(xd);
+ }
+ else if (!vp8dx_bool_error(xd->current_bc))
+ {
+ int eobtotal;
+ eobtotal = vp8_decode_mb_tokens(pbi, xd);
+
+ /* Special case: Force the loopfilter to skip when eobtotal is zero */
+ xd->mode_info_context->mbmi.mb_skip_coeff = (eobtotal==0);
+ }
+
+ mode = xd->mode_info_context->mbmi.mode;
+
+ if (xd->segmentation_enabled)
+ vp8_mb_init_dequantizer(pbi, xd);
+
+
+#if CONFIG_ERROR_CONCEALMENT
+
+ if(pbi->ec_active)
+ {
+ int throw_residual;
+ /* When we have independent partitions we can apply residual even
+ * though other partitions within the frame are corrupt.
+ */
+ throw_residual = (!pbi->independent_partitions &&
+ pbi->frame_corrupt_residual);
+ throw_residual = (throw_residual || vp8dx_bool_error(xd->current_bc));
+
+ if ((mb_idx >= pbi->mvs_corrupt_from_mb || throw_residual))
+ {
+ /* MB with corrupt residuals or corrupt mode/motion vectors.
+ * Better to use the predictor as reconstruction.
+ */
+ pbi->frame_corrupt_residual = 1;
+ vpx_memset(xd->qcoeff, 0, sizeof(xd->qcoeff));
+ vp8_conceal_corrupt_mb(xd);
+
+
+ corruption_detected = 1;
+
+ /* force idct to be skipped for B_PRED and use the
+ * prediction only for reconstruction
+ * */
+ vpx_memset(xd->eobs, 0, 25);
+ }
+ }
+#endif
+
+ /* do prediction */
+ if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
+ {
+ vp8_build_intra_predictors_mbuv_s(xd,
+ xd->recon_above[1],
+ xd->recon_above[2],
+ xd->recon_left[1],
+ xd->recon_left[2],
+ xd->recon_left_stride[1],
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride);
+
+ if (mode != B_PRED)
+ {
+ vp8_build_intra_predictors_mby_s(xd,
+ xd->recon_above[0],
+ xd->recon_left[0],
+ xd->recon_left_stride[0],
+ xd->dst.y_buffer,
+ xd->dst.y_stride);
+ }
+ else
+ {
+ short *DQC = xd->dequant_y1;
+ int dst_stride = xd->dst.y_stride;
+
+ /* clear out residual eob info */
+ if(xd->mode_info_context->mbmi.mb_skip_coeff)
+ vpx_memset(xd->eobs, 0, 25);
+
+ intra_prediction_down_copy(xd, xd->recon_above[0] + 16);
+
+ for (i = 0; i < 16; i++)
+ {
+ BLOCKD *b = &xd->block[i];
+ unsigned char *dst = xd->dst.y_buffer + b->offset;
+ B_PREDICTION_MODE b_mode =
+ xd->mode_info_context->bmi[i].as_mode;
+ unsigned char *Above = dst - dst_stride;
+ unsigned char *yleft = dst - 1;
+ int left_stride = dst_stride;
+ unsigned char top_left = Above[-1];
+
+ vp8_intra4x4_predict(Above, yleft, left_stride, b_mode,
+ dst, dst_stride, top_left);
+
+ if (xd->eobs[i])
+ {
+ if (xd->eobs[i] > 1)
+ {
+ vp8_dequant_idct_add(b->qcoeff, DQC, dst, dst_stride);
+ }
+ else
+ {
+ vp8_dc_only_idct_add
+ (b->qcoeff[0] * DQC[0],
+ dst, dst_stride,
+ dst, dst_stride);
+ ((int *)b->qcoeff)[0] = 0;
+ }
+ }
+ }
+ }
+ }
+ else
+ {
+ vp8_build_inter_predictors_mb(xd);
+ }
+
+
+#if CONFIG_ERROR_CONCEALMENT
+ if (corruption_detected)
+ {
+ return;
+ }
+#endif
+
+ if(!xd->mode_info_context->mbmi.mb_skip_coeff)
+ {
+ /* dequantization and idct */
+ if (mode != B_PRED)
+ {
+ short *DQC = xd->dequant_y1;
+
+ if (mode != SPLITMV)
+ {
+ BLOCKD *b = &xd->block[24];
+
+ /* do 2nd order transform on the dc block */
+ if (xd->eobs[24] > 1)
+ {
+ vp8_dequantize_b(b, xd->dequant_y2);
+
+ vp8_short_inv_walsh4x4(&b->dqcoeff[0],
+ xd->qcoeff);
+ ((int *)b->qcoeff)[0] = 0;
+ ((int *)b->qcoeff)[1] = 0;
+ ((int *)b->qcoeff)[2] = 0;
+ ((int *)b->qcoeff)[3] = 0;
+ ((int *)b->qcoeff)[4] = 0;
+ ((int *)b->qcoeff)[5] = 0;
+ ((int *)b->qcoeff)[6] = 0;
+ ((int *)b->qcoeff)[7] = 0;
+ }
+ else
+ {
+ b->dqcoeff[0] = b->qcoeff[0] * xd->dequant_y2[0];
+ vp8_short_inv_walsh4x4_1(&b->dqcoeff[0],
+ xd->qcoeff);
+ ((int *)b->qcoeff)[0] = 0;
+ }
+
+ /* override the dc dequant constant in order to preserve the
+ * dc components
+ */
+ DQC = xd->dequant_y1_dc;
+ }
+
+ vp8_dequant_idct_add_y_block
+ (xd->qcoeff, DQC,
+ xd->dst.y_buffer,
+ xd->dst.y_stride, xd->eobs);
+ }
+
+ vp8_dequant_idct_add_uv_block
+ (xd->qcoeff+16*16, xd->dequant_uv,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride, xd->eobs+16);
+ }
+}
+
+static int get_delta_q(vp8_reader *bc, int prev, int *q_update)
+{
+ int ret_val = 0;
+
+ if (vp8_read_bit(bc))
+ {
+ ret_val = vp8_read_literal(bc, 4);
+
+ if (vp8_read_bit(bc))
+ ret_val = -ret_val;
+ }
+
+ /* Trigger a quantizer update if the delta-q value has changed */
+ if (ret_val != prev)
+ *q_update = 1;
+
+ return ret_val;
+}
+
+#ifdef PACKET_TESTING
+#include <stdio.h>
+FILE *vpxlog = 0;
+#endif
+
+static void yv12_extend_frame_top_c(YV12_BUFFER_CONFIG *ybf)
+{
+ int i;
+ unsigned char *src_ptr1;
+ unsigned char *dest_ptr1;
+
+ unsigned int Border;
+ int plane_stride;
+
+ /***********/
+ /* Y Plane */
+ /***********/
+ Border = ybf->border;
+ plane_stride = ybf->y_stride;
+ src_ptr1 = ybf->y_buffer - Border;
+ dest_ptr1 = src_ptr1 - (Border * plane_stride);
+
+ for (i = 0; i < (int)Border; i++)
+ {
+ vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
+ dest_ptr1 += plane_stride;
+ }
+
+
+ /***********/
+ /* U Plane */
+ /***********/
+ plane_stride = ybf->uv_stride;
+ Border /= 2;
+ src_ptr1 = ybf->u_buffer - Border;
+ dest_ptr1 = src_ptr1 - (Border * plane_stride);
+
+ for (i = 0; i < (int)(Border); i++)
+ {
+ vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
+ dest_ptr1 += plane_stride;
+ }
+
+ /***********/
+ /* V Plane */
+ /***********/
+
+ src_ptr1 = ybf->v_buffer - Border;
+ dest_ptr1 = src_ptr1 - (Border * plane_stride);
+
+ for (i = 0; i < (int)(Border); i++)
+ {
+ vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
+ dest_ptr1 += plane_stride;
+ }
+}
+
+static void yv12_extend_frame_bottom_c(YV12_BUFFER_CONFIG *ybf)
+{
+ int i;
+ unsigned char *src_ptr1, *src_ptr2;
+ unsigned char *dest_ptr2;
+
+ unsigned int Border;
+ int plane_stride;
+ int plane_height;
+
+ /***********/
+ /* Y Plane */
+ /***********/
+ Border = ybf->border;
+ plane_stride = ybf->y_stride;
+ plane_height = ybf->y_height;
+
+ src_ptr1 = ybf->y_buffer - Border;
+ src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
+ dest_ptr2 = src_ptr2 + plane_stride;
+
+ for (i = 0; i < (int)Border; i++)
+ {
+ vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
+ dest_ptr2 += plane_stride;
+ }
+
+
+ /***********/
+ /* U Plane */
+ /***********/
+ plane_stride = ybf->uv_stride;
+ plane_height = ybf->uv_height;
+ Border /= 2;
+
+ src_ptr1 = ybf->u_buffer - Border;
+ src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
+ dest_ptr2 = src_ptr2 + plane_stride;
+
+ for (i = 0; i < (int)(Border); i++)
+ {
+ vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
+ dest_ptr2 += plane_stride;
+ }
+
+ /***********/
+ /* V Plane */
+ /***********/
+
+ src_ptr1 = ybf->v_buffer - Border;
+ src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
+ dest_ptr2 = src_ptr2 + plane_stride;
+
+ for (i = 0; i < (int)(Border); i++)
+ {
+ vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
+ dest_ptr2 += plane_stride;
+ }
+}
+
+static void yv12_extend_frame_left_right_c(YV12_BUFFER_CONFIG *ybf,
+ unsigned char *y_src,
+ unsigned char *u_src,
+ unsigned char *v_src)
+{
+ int i;
+ unsigned char *src_ptr1, *src_ptr2;
+ unsigned char *dest_ptr1, *dest_ptr2;
+
+ unsigned int Border;
+ int plane_stride;
+ int plane_height;
+ int plane_width;
+
+ /***********/
+ /* Y Plane */
+ /***********/
+ Border = ybf->border;
+ plane_stride = ybf->y_stride;
+ plane_height = 16;
+ plane_width = ybf->y_width;
+
+ /* copy the left and right most columns out */
+ src_ptr1 = y_src;
+ src_ptr2 = src_ptr1 + plane_width - 1;
+ dest_ptr1 = src_ptr1 - Border;
+ dest_ptr2 = src_ptr2 + 1;
+
+ for (i = 0; i < plane_height; i++)
+ {
+ vpx_memset(dest_ptr1, src_ptr1[0], Border);
+ vpx_memset(dest_ptr2, src_ptr2[0], Border);
+ src_ptr1 += plane_stride;
+ src_ptr2 += plane_stride;
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+ /***********/
+ /* U Plane */
+ /***********/
+ plane_stride = ybf->uv_stride;
+ plane_height = 8;
+ plane_width = ybf->uv_width;
+ Border /= 2;
+
+ /* copy the left and right most columns out */
+ src_ptr1 = u_src;
+ src_ptr2 = src_ptr1 + plane_width - 1;
+ dest_ptr1 = src_ptr1 - Border;
+ dest_ptr2 = src_ptr2 + 1;
+
+ for (i = 0; i < plane_height; i++)
+ {
+ vpx_memset(dest_ptr1, src_ptr1[0], Border);
+ vpx_memset(dest_ptr2, src_ptr2[0], Border);
+ src_ptr1 += plane_stride;
+ src_ptr2 += plane_stride;
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+ /***********/
+ /* V Plane */
+ /***********/
+
+ /* copy the left and right most columns out */
+ src_ptr1 = v_src;
+ src_ptr2 = src_ptr1 + plane_width - 1;
+ dest_ptr1 = src_ptr1 - Border;
+ dest_ptr2 = src_ptr2 + 1;
+
+ for (i = 0; i < plane_height; i++)
+ {
+ vpx_memset(dest_ptr1, src_ptr1[0], Border);
+ vpx_memset(dest_ptr2, src_ptr2[0], Border);
+ src_ptr1 += plane_stride;
+ src_ptr2 += plane_stride;
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+}
+
+static void decode_mb_rows(VP8D_COMP *pbi)
+{
+ VP8_COMMON *const pc = & pbi->common;
+ MACROBLOCKD *const xd = & pbi->mb;
+
+ MODE_INFO *lf_mic = xd->mode_info_context;
+
+ int ibc = 0;
+ int num_part = 1 << pc->multi_token_partition;
+
+ int recon_yoffset, recon_uvoffset;
+ int mb_row, mb_col;
+ int mb_idx = 0;
+
+ YV12_BUFFER_CONFIG *yv12_fb_new = pbi->dec_fb_ref[INTRA_FRAME];
+
+ int recon_y_stride = yv12_fb_new->y_stride;
+ int recon_uv_stride = yv12_fb_new->uv_stride;
+
+ unsigned char *ref_buffer[MAX_REF_FRAMES][3];
+ unsigned char *dst_buffer[3];
+ unsigned char *lf_dst[3];
+ unsigned char *eb_dst[3];
+ int i;
+ int ref_fb_corrupted[MAX_REF_FRAMES];
+
+ ref_fb_corrupted[INTRA_FRAME] = 0;
+
+ for(i = 1; i < MAX_REF_FRAMES; i++)
+ {
+ YV12_BUFFER_CONFIG *this_fb = pbi->dec_fb_ref[i];
+
+ ref_buffer[i][0] = this_fb->y_buffer;
+ ref_buffer[i][1] = this_fb->u_buffer;
+ ref_buffer[i][2] = this_fb->v_buffer;
+
+ ref_fb_corrupted[i] = this_fb->corrupted;
+ }
+
+ /* Set up the buffer pointers */
+ eb_dst[0] = lf_dst[0] = dst_buffer[0] = yv12_fb_new->y_buffer;
+ eb_dst[1] = lf_dst[1] = dst_buffer[1] = yv12_fb_new->u_buffer;
+ eb_dst[2] = lf_dst[2] = dst_buffer[2] = yv12_fb_new->v_buffer;
+
+ xd->up_available = 0;
+
+ /* Initialize the loop filter for this frame. */
+ if(pc->filter_level)
+ vp8_loop_filter_frame_init(pc, xd, pc->filter_level);
+
+ vp8_setup_intra_recon_top_line(yv12_fb_new);
+
+ /* Decode the individual macro block */
+ for (mb_row = 0; mb_row < pc->mb_rows; mb_row++)
+ {
+ if (num_part > 1)
+ {
+ xd->current_bc = & pbi->mbc[ibc];
+ ibc++;
+
+ if (ibc == num_part)
+ ibc = 0;
+ }
+
+ recon_yoffset = mb_row * recon_y_stride * 16;
+ recon_uvoffset = mb_row * recon_uv_stride * 8;
+
+ /* reset contexts */
+ xd->above_context = pc->above_context;
+ vpx_memset(xd->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ xd->left_available = 0;
+
+ xd->mb_to_top_edge = -((mb_row * 16)) << 3;
+ xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3;
+
+ xd->recon_above[0] = dst_buffer[0] + recon_yoffset;
+ xd->recon_above[1] = dst_buffer[1] + recon_uvoffset;
+ xd->recon_above[2] = dst_buffer[2] + recon_uvoffset;
+
+ xd->recon_left[0] = xd->recon_above[0] - 1;
+ xd->recon_left[1] = xd->recon_above[1] - 1;
+ xd->recon_left[2] = xd->recon_above[2] - 1;
+
+ xd->recon_above[0] -= xd->dst.y_stride;
+ xd->recon_above[1] -= xd->dst.uv_stride;
+ xd->recon_above[2] -= xd->dst.uv_stride;
+
+ /* TODO: move to outside row loop */
+ xd->recon_left_stride[0] = xd->dst.y_stride;
+ xd->recon_left_stride[1] = xd->dst.uv_stride;
+
+ setup_intra_recon_left(xd->recon_left[0], xd->recon_left[1],
+ xd->recon_left[2], xd->dst.y_stride,
+ xd->dst.uv_stride);
+
+ for (mb_col = 0; mb_col < pc->mb_cols; mb_col++)
+ {
+ /* Distance of Mb to the various image edges.
+ * These are specified to 8th pel as they are always compared to values
+ * that are in 1/8th pel units
+ */
+ xd->mb_to_left_edge = -((mb_col * 16) << 3);
+ xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3;
+
+#if CONFIG_ERROR_CONCEALMENT
+ {
+ int corrupt_residual = (!pbi->independent_partitions &&
+ pbi->frame_corrupt_residual) ||
+ vp8dx_bool_error(xd->current_bc);
+ if (pbi->ec_active &&
+ xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME &&
+ corrupt_residual)
+ {
+ /* We have an intra block with corrupt coefficients, better to
+ * conceal with an inter block. Interpolate MVs from neighboring
+ * MBs.
+ *
+ * Note that for the first mb with corrupt residual in a frame,
+ * we might not discover that before decoding the residual. That
+ * happens after this check, and therefore no inter concealment
+ * will be done.
+ */
+ vp8_interpolate_motion(xd,
+ mb_row, mb_col,
+ pc->mb_rows, pc->mb_cols,
+ pc->mode_info_stride);
+ }
+ }
+#endif
+
+ xd->dst.y_buffer = dst_buffer[0] + recon_yoffset;
+ xd->dst.u_buffer = dst_buffer[1] + recon_uvoffset;
+ xd->dst.v_buffer = dst_buffer[2] + recon_uvoffset;
+
+ xd->pre.y_buffer = ref_buffer[xd->mode_info_context->mbmi.ref_frame][0] + recon_yoffset;
+ xd->pre.u_buffer = ref_buffer[xd->mode_info_context->mbmi.ref_frame][1] + recon_uvoffset;
+ xd->pre.v_buffer = ref_buffer[xd->mode_info_context->mbmi.ref_frame][2] + recon_uvoffset;
+
+ /* propagate errors from reference frames */
+ xd->corrupted |= ref_fb_corrupted[xd->mode_info_context->mbmi.ref_frame];
+
+ decode_macroblock(pbi, xd, mb_idx);
+
+ mb_idx++;
+ xd->left_available = 1;
+
+ /* check if the boolean decoder has suffered an error */
+ xd->corrupted |= vp8dx_bool_error(xd->current_bc);
+
+ xd->recon_above[0] += 16;
+ xd->recon_above[1] += 8;
+ xd->recon_above[2] += 8;
+ xd->recon_left[0] += 16;
+ xd->recon_left[1] += 8;
+ xd->recon_left[2] += 8;
+
+ recon_yoffset += 16;
+ recon_uvoffset += 8;
+
+ ++xd->mode_info_context; /* next mb */
+
+ xd->above_context++;
+ }
+
+ /* adjust to the next row of mbs */
+ vp8_extend_mb_row(yv12_fb_new, xd->dst.y_buffer + 16,
+ xd->dst.u_buffer + 8, xd->dst.v_buffer + 8);
+
+ ++xd->mode_info_context; /* skip prediction column */
+ xd->up_available = 1;
+
+ if(pc->filter_level)
+ {
+ if(mb_row > 0)
+ {
+ if (pc->filter_type == NORMAL_LOOPFILTER)
+ vp8_loop_filter_row_normal(pc, lf_mic, mb_row-1,
+ recon_y_stride, recon_uv_stride,
+ lf_dst[0], lf_dst[1], lf_dst[2]);
+ else
+ vp8_loop_filter_row_simple(pc, lf_mic, mb_row-1,
+ recon_y_stride, recon_uv_stride,
+ lf_dst[0], lf_dst[1], lf_dst[2]);
+
+ if(mb_row > 1)
+ {
+ yv12_extend_frame_left_right_c(yv12_fb_new,
+ eb_dst[0],
+ eb_dst[1],
+ eb_dst[2]);
+
+ eb_dst[0] += recon_y_stride * 16;
+ eb_dst[1] += recon_uv_stride * 8;
+ eb_dst[2] += recon_uv_stride * 8;
+
+ if(mb_row == 2)
+ yv12_extend_frame_top_c(yv12_fb_new);
+
+ }
+
+ lf_dst[0] += recon_y_stride * 16;
+ lf_dst[1] += recon_uv_stride * 8;
+ lf_dst[2] += recon_uv_stride * 8;
+ lf_mic += pc->mb_cols;
+ lf_mic++; /* Skip border mb */
+ }
+ }
+ else
+ {
+ if(mb_row > 0)
+ {
+ /**/
+ yv12_extend_frame_left_right_c(yv12_fb_new,
+ eb_dst[0],
+ eb_dst[1],
+ eb_dst[2]);
+
+ eb_dst[0] += recon_y_stride * 16;
+ eb_dst[1] += recon_uv_stride * 8;
+ eb_dst[2] += recon_uv_stride * 8;
+
+ if(mb_row == 1)
+ yv12_extend_frame_top_c(yv12_fb_new);
+ }
+ }
+ }
+
+ if(pc->filter_level)
+ {
+ if (pc->filter_type == NORMAL_LOOPFILTER)
+ vp8_loop_filter_row_normal(pc, lf_mic, mb_row-1, recon_y_stride,
+ recon_uv_stride, lf_dst[0], lf_dst[1],
+ lf_dst[2]);
+ else
+ vp8_loop_filter_row_simple(pc, lf_mic, mb_row-1, recon_y_stride,
+ recon_uv_stride, lf_dst[0], lf_dst[1],
+ lf_dst[2]);
+
+ yv12_extend_frame_left_right_c(yv12_fb_new,
+ eb_dst[0],
+ eb_dst[1],
+ eb_dst[2]);
+ eb_dst[0] += recon_y_stride * 16;
+ eb_dst[1] += recon_uv_stride * 8;
+ eb_dst[2] += recon_uv_stride * 8;
+ }
+ yv12_extend_frame_left_right_c(yv12_fb_new,
+ eb_dst[0],
+ eb_dst[1],
+ eb_dst[2]);
+
+ yv12_extend_frame_bottom_c(yv12_fb_new);
+
+}
+
+static unsigned int read_partition_size(const unsigned char *cx_size)
+{
+ const unsigned int size =
+ cx_size[0] + (cx_size[1] << 8) + (cx_size[2] << 16);
+ return size;
+}
+
+static int read_is_valid(const unsigned char *start,
+ size_t len,
+ const unsigned char *end)
+{
+ return (start + len > start && start + len <= end);
+}
+
+static unsigned int read_available_partition_size(
+ VP8D_COMP *pbi,
+ const unsigned char *token_part_sizes,
+ const unsigned char *fragment_start,
+ const unsigned char *first_fragment_end,
+ const unsigned char *fragment_end,
+ int i,
+ int num_part)
+{
+ VP8_COMMON* pc = &pbi->common;
+ const unsigned char *partition_size_ptr = token_part_sizes + i * 3;
+ unsigned int partition_size = 0;
+ ptrdiff_t bytes_left = fragment_end - fragment_start;
+ /* Calculate the length of this partition. The last partition
+ * size is implicit. If the partition size can't be read, then
+ * either use the remaining data in the buffer (for EC mode)
+ * or throw an error.
+ */
+ if (i < num_part - 1)
+ {
+ if (read_is_valid(partition_size_ptr, 3, first_fragment_end))
+ partition_size = read_partition_size(partition_size_ptr);
+ else if (pbi->ec_active)
+ partition_size = (unsigned int)bytes_left;
+ else
+ vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
+ "Truncated partition size data");
+ }
+ else
+ partition_size = (unsigned int)bytes_left;
+
+ /* Validate the calculated partition length. If the buffer
+ * described by the partition can't be fully read, then restrict
+ * it to the portion that can be (for EC mode) or throw an error.
+ */
+ if (!read_is_valid(fragment_start, partition_size, fragment_end))
+ {
+ if (pbi->ec_active)
+ partition_size = (unsigned int)bytes_left;
+ else
+ vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
+ "Truncated packet or corrupt partition "
+ "%d length", i + 1);
+ }
+ return partition_size;
+}
+
+
+static void setup_token_decoder(VP8D_COMP *pbi,
+ const unsigned char* token_part_sizes)
+{
+ vp8_reader *bool_decoder = &pbi->mbc[0];
+ unsigned int partition_idx;
+ unsigned int fragment_idx;
+ unsigned int num_token_partitions;
+ const unsigned char *first_fragment_end = pbi->fragments[0] +
+ pbi->fragment_sizes[0];
+
+ TOKEN_PARTITION multi_token_partition =
+ (TOKEN_PARTITION)vp8_read_literal(&pbi->mbc[8], 2);
+ if (!vp8dx_bool_error(&pbi->mbc[8]))
+ pbi->common.multi_token_partition = multi_token_partition;
+ num_token_partitions = 1 << pbi->common.multi_token_partition;
+
+ /* Check for partitions within the fragments and unpack the fragments
+ * so that each fragment pointer points to its corresponding partition. */
+ for (fragment_idx = 0; fragment_idx < pbi->num_fragments; ++fragment_idx)
+ {
+ unsigned int fragment_size = pbi->fragment_sizes[fragment_idx];
+ const unsigned char *fragment_end = pbi->fragments[fragment_idx] +
+ fragment_size;
+ /* Special case for handling the first partition since we have already
+ * read its size. */
+ if (fragment_idx == 0)
+ {
+ /* Size of first partition + token partition sizes element */
+ ptrdiff_t ext_first_part_size = token_part_sizes -
+ pbi->fragments[0] + 3 * (num_token_partitions - 1);
+ fragment_size -= (unsigned int)ext_first_part_size;
+ if (fragment_size > 0)
+ {
+ pbi->fragment_sizes[0] = (unsigned int)ext_first_part_size;
+ /* The fragment contains an additional partition. Move to
+ * next. */
+ fragment_idx++;
+ pbi->fragments[fragment_idx] = pbi->fragments[0] +
+ pbi->fragment_sizes[0];
+ }
+ }
+ /* Split the chunk into partitions read from the bitstream */
+ while (fragment_size > 0)
+ {
+ ptrdiff_t partition_size = read_available_partition_size(
+ pbi,
+ token_part_sizes,
+ pbi->fragments[fragment_idx],
+ first_fragment_end,
+ fragment_end,
+ fragment_idx - 1,
+ num_token_partitions);
+ pbi->fragment_sizes[fragment_idx] = (unsigned int)partition_size;
+ fragment_size -= (unsigned int)partition_size;
+ assert(fragment_idx <= num_token_partitions);
+ if (fragment_size > 0)
+ {
+ /* The fragment contains an additional partition.
+ * Move to next. */
+ fragment_idx++;
+ pbi->fragments[fragment_idx] =
+ pbi->fragments[fragment_idx - 1] + partition_size;
+ }
+ }
+ }
+
+ pbi->num_fragments = num_token_partitions + 1;
+
+ for (partition_idx = 1; partition_idx < pbi->num_fragments; ++partition_idx)
+ {
+ if (vp8dx_start_decode(bool_decoder,
+ pbi->fragments[partition_idx],
+ pbi->fragment_sizes[partition_idx]))
+ vpx_internal_error(&pbi->common.error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate bool decoder %d",
+ partition_idx);
+
+ bool_decoder++;
+ }
+
+#if CONFIG_MULTITHREAD
+ /* Clamp number of decoder threads */
+ if (pbi->decoding_thread_count > num_token_partitions - 1)
+ pbi->decoding_thread_count = num_token_partitions - 1;
+#endif
+}
+
+
+static void init_frame(VP8D_COMP *pbi)
+{
+ VP8_COMMON *const pc = & pbi->common;
+ MACROBLOCKD *const xd = & pbi->mb;
+
+ if (pc->frame_type == KEY_FRAME)
+ {
+ /* Various keyframe initializations */
+ vpx_memcpy(pc->fc.mvc, vp8_default_mv_context, sizeof(vp8_default_mv_context));
+
+ vp8_init_mbmode_probs(pc);
+
+ vp8_default_coef_probs(pc);
+
+ /* reset the segment feature data to 0 with delta coding (Default state). */
+ vpx_memset(xd->segment_feature_data, 0, sizeof(xd->segment_feature_data));
+ xd->mb_segement_abs_delta = SEGMENT_DELTADATA;
+
+ /* reset the mode ref deltasa for loop filter */
+ vpx_memset(xd->ref_lf_deltas, 0, sizeof(xd->ref_lf_deltas));
+ vpx_memset(xd->mode_lf_deltas, 0, sizeof(xd->mode_lf_deltas));
+
+ /* All buffers are implicitly updated on key frames. */
+ pc->refresh_golden_frame = 1;
+ pc->refresh_alt_ref_frame = 1;
+ pc->copy_buffer_to_gf = 0;
+ pc->copy_buffer_to_arf = 0;
+
+ /* Note that Golden and Altref modes cannot be used on a key frame so
+ * ref_frame_sign_bias[] is undefined and meaningless
+ */
+ pc->ref_frame_sign_bias[GOLDEN_FRAME] = 0;
+ pc->ref_frame_sign_bias[ALTREF_FRAME] = 0;
+ }
+ else
+ {
+ /* To enable choice of different interploation filters */
+ if (!pc->use_bilinear_mc_filter)
+ {
+ xd->subpixel_predict = vp8_sixtap_predict4x4;
+ xd->subpixel_predict8x4 = vp8_sixtap_predict8x4;
+ xd->subpixel_predict8x8 = vp8_sixtap_predict8x8;
+ xd->subpixel_predict16x16 = vp8_sixtap_predict16x16;
+ }
+ else
+ {
+ xd->subpixel_predict = vp8_bilinear_predict4x4;
+ xd->subpixel_predict8x4 = vp8_bilinear_predict8x4;
+ xd->subpixel_predict8x8 = vp8_bilinear_predict8x8;
+ xd->subpixel_predict16x16 = vp8_bilinear_predict16x16;
+ }
+
+ if (pbi->decoded_key_frame && pbi->ec_enabled && !pbi->ec_active)
+ pbi->ec_active = 1;
+ }
+
+ xd->left_context = &pc->left_context;
+ xd->mode_info_context = pc->mi;
+ xd->frame_type = pc->frame_type;
+ xd->mode_info_context->mbmi.mode = DC_PRED;
+ xd->mode_info_stride = pc->mode_info_stride;
+ xd->corrupted = 0; /* init without corruption */
+
+ xd->fullpixel_mask = 0xffffffff;
+ if(pc->full_pixel)
+ xd->fullpixel_mask = 0xfffffff8;
+
+}
+
+int vp8_decode_frame(VP8D_COMP *pbi)
+{
+ vp8_reader *const bc = & pbi->mbc[8];
+ VP8_COMMON *const pc = & pbi->common;
+ MACROBLOCKD *const xd = & pbi->mb;
+ const unsigned char *data = pbi->fragments[0];
+ const unsigned char *data_end = data + pbi->fragment_sizes[0];
+ ptrdiff_t first_partition_length_in_bytes;
+
+ int i, j, k, l;
+ const int *const mb_feature_data_bits = vp8_mb_feature_data_bits;
+ int corrupt_tokens = 0;
+ int prev_independent_partitions = pbi->independent_partitions;
+
+ YV12_BUFFER_CONFIG *yv12_fb_new = pbi->dec_fb_ref[INTRA_FRAME];
+
+ /* start with no corruption of current frame */
+ xd->corrupted = 0;
+ yv12_fb_new->corrupted = 0;
+
+ if (data_end - data < 3)
+ {
+ if (!pbi->ec_active)
+ {
+ vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
+ "Truncated packet");
+ }
+
+ /* Declare the missing frame as an inter frame since it will
+ be handled as an inter frame when we have estimated its
+ motion vectors. */
+ pc->frame_type = INTER_FRAME;
+ pc->version = 0;
+ pc->show_frame = 1;
+ first_partition_length_in_bytes = 0;
+ }
+ else
+ {
+ pc->frame_type = (FRAME_TYPE)(data[0] & 1);
+ pc->version = (data[0] >> 1) & 7;
+ pc->show_frame = (data[0] >> 4) & 1;
+ first_partition_length_in_bytes =
+ (data[0] | (data[1] << 8) | (data[2] << 16)) >> 5;
+
+ if (!pbi->ec_active && (data + first_partition_length_in_bytes > data_end
+ || data + first_partition_length_in_bytes < data))
+ vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
+ "Truncated packet or corrupt partition 0 length");
+
+ data += 3;
+
+ vp8_setup_version(pc);
+
+
+ if (pc->frame_type == KEY_FRAME)
+ {
+ /* vet via sync code */
+ /* When error concealment is enabled we should only check the sync
+ * code if we have enough bits available
+ */
+ if (!pbi->ec_active || data + 3 < data_end)
+ {
+ if (data[0] != 0x9d || data[1] != 0x01 || data[2] != 0x2a)
+ vpx_internal_error(&pc->error, VPX_CODEC_UNSUP_BITSTREAM,
+ "Invalid frame sync code");
+ }
+
+ /* If error concealment is enabled we should only parse the new size
+ * if we have enough data. Otherwise we will end up with the wrong
+ * size.
+ */
+ if (!pbi->ec_active || data + 6 < data_end)
+ {
+ pc->Width = (data[3] | (data[4] << 8)) & 0x3fff;
+ pc->horiz_scale = data[4] >> 6;
+ pc->Height = (data[5] | (data[6] << 8)) & 0x3fff;
+ pc->vert_scale = data[6] >> 6;
+ }
+ data += 7;
+
+ }
+ else
+ {
+ vpx_memcpy(&xd->pre, yv12_fb_new, sizeof(YV12_BUFFER_CONFIG));
+ vpx_memcpy(&xd->dst, yv12_fb_new, sizeof(YV12_BUFFER_CONFIG));
+ }
+ }
+ if ((!pbi->decoded_key_frame && pc->frame_type != KEY_FRAME))
+ {
+ return -1;
+ }
+
+ init_frame(pbi);
+
+ if (vp8dx_start_decode(bc, data, (unsigned int)(data_end - data)))
+ vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate bool decoder 0");
+ if (pc->frame_type == KEY_FRAME) {
+ pc->clr_type = (YUV_TYPE)vp8_read_bit(bc);
+ pc->clamp_type = (CLAMP_TYPE)vp8_read_bit(bc);
+ }
+
+ /* Is segmentation enabled */
+ xd->segmentation_enabled = (unsigned char)vp8_read_bit(bc);
+
+ if (xd->segmentation_enabled)
+ {
+ /* Signal whether or not the segmentation map is being explicitly updated this frame. */
+ xd->update_mb_segmentation_map = (unsigned char)vp8_read_bit(bc);
+ xd->update_mb_segmentation_data = (unsigned char)vp8_read_bit(bc);
+
+ if (xd->update_mb_segmentation_data)
+ {
+ xd->mb_segement_abs_delta = (unsigned char)vp8_read_bit(bc);
+
+ vpx_memset(xd->segment_feature_data, 0, sizeof(xd->segment_feature_data));
+
+ /* For each segmentation feature (Quant and loop filter level) */
+ for (i = 0; i < MB_LVL_MAX; i++)
+ {
+ for (j = 0; j < MAX_MB_SEGMENTS; j++)
+ {
+ /* Frame level data */
+ if (vp8_read_bit(bc))
+ {
+ xd->segment_feature_data[i][j] = (signed char)vp8_read_literal(bc, mb_feature_data_bits[i]);
+
+ if (vp8_read_bit(bc))
+ xd->segment_feature_data[i][j] = -xd->segment_feature_data[i][j];
+ }
+ else
+ xd->segment_feature_data[i][j] = 0;
+ }
+ }
+ }
+
+ if (xd->update_mb_segmentation_map)
+ {
+ /* Which macro block level features are enabled */
+ vpx_memset(xd->mb_segment_tree_probs, 255, sizeof(xd->mb_segment_tree_probs));
+
+ /* Read the probs used to decode the segment id for each macro block. */
+ for (i = 0; i < MB_FEATURE_TREE_PROBS; i++)
+ {
+ /* If not explicitly set value is defaulted to 255 by memset above */
+ if (vp8_read_bit(bc))
+ xd->mb_segment_tree_probs[i] = (vp8_prob)vp8_read_literal(bc, 8);
+ }
+ }
+ }
+ else
+ {
+ /* No segmentation updates on this frame */
+ xd->update_mb_segmentation_map = 0;
+ xd->update_mb_segmentation_data = 0;
+ }
+
+ /* Read the loop filter level and type */
+ pc->filter_type = (LOOPFILTERTYPE) vp8_read_bit(bc);
+ pc->filter_level = vp8_read_literal(bc, 6);
+ pc->sharpness_level = vp8_read_literal(bc, 3);
+
+ /* Read in loop filter deltas applied at the MB level based on mode or ref frame. */
+ xd->mode_ref_lf_delta_update = 0;
+ xd->mode_ref_lf_delta_enabled = (unsigned char)vp8_read_bit(bc);
+
+ if (xd->mode_ref_lf_delta_enabled)
+ {
+ /* Do the deltas need to be updated */
+ xd->mode_ref_lf_delta_update = (unsigned char)vp8_read_bit(bc);
+
+ if (xd->mode_ref_lf_delta_update)
+ {
+ /* Send update */
+ for (i = 0; i < MAX_REF_LF_DELTAS; i++)
+ {
+ if (vp8_read_bit(bc))
+ {
+ /*sign = vp8_read_bit( bc );*/
+ xd->ref_lf_deltas[i] = (signed char)vp8_read_literal(bc, 6);
+
+ if (vp8_read_bit(bc)) /* Apply sign */
+ xd->ref_lf_deltas[i] = xd->ref_lf_deltas[i] * -1;
+ }
+ }
+
+ /* Send update */
+ for (i = 0; i < MAX_MODE_LF_DELTAS; i++)
+ {
+ if (vp8_read_bit(bc))
+ {
+ /*sign = vp8_read_bit( bc );*/
+ xd->mode_lf_deltas[i] = (signed char)vp8_read_literal(bc, 6);
+
+ if (vp8_read_bit(bc)) /* Apply sign */
+ xd->mode_lf_deltas[i] = xd->mode_lf_deltas[i] * -1;
+ }
+ }
+ }
+ }
+
+ setup_token_decoder(pbi, data + first_partition_length_in_bytes);
+
+ xd->current_bc = &pbi->mbc[0];
+
+ /* Read the default quantizers. */
+ {
+ int Q, q_update;
+
+ Q = vp8_read_literal(bc, 7); /* AC 1st order Q = default */
+ pc->base_qindex = Q;
+ q_update = 0;
+ pc->y1dc_delta_q = get_delta_q(bc, pc->y1dc_delta_q, &q_update);
+ pc->y2dc_delta_q = get_delta_q(bc, pc->y2dc_delta_q, &q_update);
+ pc->y2ac_delta_q = get_delta_q(bc, pc->y2ac_delta_q, &q_update);
+ pc->uvdc_delta_q = get_delta_q(bc, pc->uvdc_delta_q, &q_update);
+ pc->uvac_delta_q = get_delta_q(bc, pc->uvac_delta_q, &q_update);
+
+ if (q_update)
+ vp8cx_init_de_quantizer(pbi);
+
+ /* MB level dequantizer setup */
+ vp8_mb_init_dequantizer(pbi, &pbi->mb);
+ }
+
+ /* Determine if the golden frame or ARF buffer should be updated and how.
+ * For all non key frames the GF and ARF refresh flags and sign bias
+ * flags must be set explicitly.
+ */
+ if (pc->frame_type != KEY_FRAME)
+ {
+ /* Should the GF or ARF be updated from the current frame */
+ pc->refresh_golden_frame = vp8_read_bit(bc);
+#if CONFIG_ERROR_CONCEALMENT
+ /* Assume we shouldn't refresh golden if the bit is missing */
+ xd->corrupted |= vp8dx_bool_error(bc);
+ if (pbi->ec_active && xd->corrupted)
+ pc->refresh_golden_frame = 0;
+#endif
+
+ pc->refresh_alt_ref_frame = vp8_read_bit(bc);
+#if CONFIG_ERROR_CONCEALMENT
+ /* Assume we shouldn't refresh altref if the bit is missing */
+ xd->corrupted |= vp8dx_bool_error(bc);
+ if (pbi->ec_active && xd->corrupted)
+ pc->refresh_alt_ref_frame = 0;
+#endif
+
+ /* Buffer to buffer copy flags. */
+ pc->copy_buffer_to_gf = 0;
+
+ if (!pc->refresh_golden_frame)
+ pc->copy_buffer_to_gf = vp8_read_literal(bc, 2);
+
+#if CONFIG_ERROR_CONCEALMENT
+ /* Assume we shouldn't copy to the golden if the bit is missing */
+ xd->corrupted |= vp8dx_bool_error(bc);
+ if (pbi->ec_active && xd->corrupted)
+ pc->copy_buffer_to_gf = 0;
+#endif
+
+ pc->copy_buffer_to_arf = 0;
+
+ if (!pc->refresh_alt_ref_frame)
+ pc->copy_buffer_to_arf = vp8_read_literal(bc, 2);
+
+#if CONFIG_ERROR_CONCEALMENT
+ /* Assume we shouldn't copy to the alt-ref if the bit is missing */
+ xd->corrupted |= vp8dx_bool_error(bc);
+ if (pbi->ec_active && xd->corrupted)
+ pc->copy_buffer_to_arf = 0;
+#endif
+
+
+ pc->ref_frame_sign_bias[GOLDEN_FRAME] = vp8_read_bit(bc);
+ pc->ref_frame_sign_bias[ALTREF_FRAME] = vp8_read_bit(bc);
+ }
+
+ pc->refresh_entropy_probs = vp8_read_bit(bc);
+#if CONFIG_ERROR_CONCEALMENT
+ /* Assume we shouldn't refresh the probabilities if the bit is
+ * missing */
+ xd->corrupted |= vp8dx_bool_error(bc);
+ if (pbi->ec_active && xd->corrupted)
+ pc->refresh_entropy_probs = 0;
+#endif
+ if (pc->refresh_entropy_probs == 0)
+ {
+ vpx_memcpy(&pc->lfc, &pc->fc, sizeof(pc->fc));
+ }
+
+ pc->refresh_last_frame = pc->frame_type == KEY_FRAME || vp8_read_bit(bc);
+
+#if CONFIG_ERROR_CONCEALMENT
+ /* Assume we should refresh the last frame if the bit is missing */
+ xd->corrupted |= vp8dx_bool_error(bc);
+ if (pbi->ec_active && xd->corrupted)
+ pc->refresh_last_frame = 1;
+#endif
+
+ if (0)
+ {
+ FILE *z = fopen("decodestats.stt", "a");
+ fprintf(z, "%6d F:%d,G:%d,A:%d,L:%d,Q:%d\n",
+ pc->current_video_frame,
+ pc->frame_type,
+ pc->refresh_golden_frame,
+ pc->refresh_alt_ref_frame,
+ pc->refresh_last_frame,
+ pc->base_qindex);
+ fclose(z);
+ }
+
+ {
+ pbi->independent_partitions = 1;
+
+ /* read coef probability tree */
+ for (i = 0; i < BLOCK_TYPES; i++)
+ for (j = 0; j < COEF_BANDS; j++)
+ for (k = 0; k < PREV_COEF_CONTEXTS; k++)
+ for (l = 0; l < ENTROPY_NODES; l++)
+ {
+
+ vp8_prob *const p = pc->fc.coef_probs [i][j][k] + l;
+
+ if (vp8_read(bc, vp8_coef_update_probs [i][j][k][l]))
+ {
+ *p = (vp8_prob)vp8_read_literal(bc, 8);
+
+ }
+ if (k > 0 && *p != pc->fc.coef_probs[i][j][k-1][l])
+ pbi->independent_partitions = 0;
+
+ }
+ }
+
+ /* clear out the coeff buffer */
+ vpx_memset(xd->qcoeff, 0, sizeof(xd->qcoeff));
+
+ vp8_decode_mode_mvs(pbi);
+
+#if CONFIG_ERROR_CONCEALMENT
+ if (pbi->ec_active &&
+ pbi->mvs_corrupt_from_mb < (unsigned int)pc->mb_cols * pc->mb_rows)
+ {
+ /* Motion vectors are missing in this frame. We will try to estimate
+ * them and then continue decoding the frame as usual */
+ vp8_estimate_missing_mvs(pbi);
+ }
+#endif
+
+ vpx_memset(pc->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES) * pc->mb_cols);
+ pbi->frame_corrupt_residual = 0;
+
+#if CONFIG_MULTITHREAD
+ if (pbi->b_multithreaded_rd && pc->multi_token_partition != ONE_PARTITION)
+ {
+ unsigned int i;
+ vp8mt_decode_mb_rows(pbi, xd);
+ vp8_yv12_extend_frame_borders(yv12_fb_new);
+ for (i = 0; i < pbi->decoding_thread_count; ++i)
+ corrupt_tokens |= pbi->mb_row_di[i].mbd.corrupted;
+ }
+ else
+#endif
+ {
+ decode_mb_rows(pbi);
+ corrupt_tokens |= xd->corrupted;
+ }
+
+ /* Collect information about decoder corruption. */
+ /* 1. Check first boolean decoder for errors. */
+ yv12_fb_new->corrupted = vp8dx_bool_error(bc);
+ /* 2. Check the macroblock information */
+ yv12_fb_new->corrupted |= corrupt_tokens;
+
+ if (!pbi->decoded_key_frame)
+ {
+ if (pc->frame_type == KEY_FRAME &&
+ !yv12_fb_new->corrupted)
+ pbi->decoded_key_frame = 1;
+ else
+ vpx_internal_error(&pbi->common.error, VPX_CODEC_CORRUPT_FRAME,
+ "A stream must start with a complete key frame");
+ }
+
+ /* vpx_log("Decoder: Frame Decoded, Size Roughly:%d bytes \n",bc->pos+pbi->bc2.pos); */
+
+ if (pc->refresh_entropy_probs == 0)
+ {
+ vpx_memcpy(&pc->fc, &pc->lfc, sizeof(pc->fc));
+ pbi->independent_partitions = prev_independent_partitions;
+ }
+
+#ifdef PACKET_TESTING
+ {
+ FILE *f = fopen("decompressor.VP8", "ab");
+ unsigned int size = pbi->bc2.pos + pbi->bc.pos + 8;
+ fwrite((void *) &size, 4, 1, f);
+ fwrite((void *) pbi->Source, size, 1, f);
+ fclose(f);
+ }
+#endif
+
+ return 0;
+}
diff --git a/libvpx/vp8/decoder/detokenize.c b/libvpx/vp8/decoder/detokenize.c
new file mode 100644
index 0000000..452ff6c
--- /dev/null
+++ b/libvpx/vp8/decoder/detokenize.c
@@ -0,0 +1,245 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/common/blockd.h"
+#include "onyxd_int.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vpx_ports/mem.h"
+#include "detokenize.h"
+
+void vp8_reset_mb_tokens_context(MACROBLOCKD *x)
+{
+ ENTROPY_CONTEXT *a_ctx = ((ENTROPY_CONTEXT *)x->above_context);
+ ENTROPY_CONTEXT *l_ctx = ((ENTROPY_CONTEXT *)x->left_context);
+
+ vpx_memset(a_ctx, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1);
+ vpx_memset(l_ctx, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1);
+
+ /* Clear entropy contexts for Y2 blocks */
+ if (!x->mode_info_context->mbmi.is_4x4)
+ {
+ a_ctx[8] = l_ctx[8] = 0;
+ }
+}
+
+/*
+ ------------------------------------------------------------------------------
+ Residual decoding (Paragraph 13.2 / 13.3)
+*/
+static const uint8_t kBands[16 + 1] = {
+ 0, 1, 2, 3, 6, 4, 5, 6, 6, 6, 6, 6, 6, 6, 6, 7,
+ 0 /* extra entry as sentinel */
+};
+
+static const uint8_t kCat3[] = { 173, 148, 140, 0 };
+static const uint8_t kCat4[] = { 176, 155, 140, 135, 0 };
+static const uint8_t kCat5[] = { 180, 157, 141, 134, 130, 0 };
+static const uint8_t kCat6[] =
+ { 254, 254, 243, 230, 196, 177, 153, 140, 133, 130, 129, 0 };
+static const uint8_t* const kCat3456[] = { kCat3, kCat4, kCat5, kCat6 };
+static const uint8_t kZigzag[16] = {
+ 0, 1, 4, 8, 5, 2, 3, 6, 9, 12, 13, 10, 7, 11, 14, 15
+};
+
+#define VP8GetBit vp8dx_decode_bool
+#define NUM_PROBAS 11
+#define NUM_CTX 3
+
+/* for const-casting */
+typedef const uint8_t (*ProbaArray)[NUM_CTX][NUM_PROBAS];
+
+static int GetSigned(BOOL_DECODER *br, int value_to_sign)
+{
+ int split = (br->range + 1) >> 1;
+ VP8_BD_VALUE bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8);
+ int v;
+
+ if(br->count < 0)
+ vp8dx_bool_decoder_fill(br);
+
+ if ( br->value < bigsplit )
+ {
+ br->range = split;
+ v= value_to_sign;
+ }
+ else
+ {
+ br->range = br->range-split;
+ br->value = br->value-bigsplit;
+ v = -value_to_sign;
+ }
+ br->range +=br->range;
+ br->value +=br->value;
+ br->count--;
+
+ return v;
+}
+/*
+ Returns the position of the last non-zero coeff plus one
+ (and 0 if there's no coeff at all)
+*/
+static int GetCoeffs(BOOL_DECODER *br, ProbaArray prob,
+ int ctx, int n, int16_t* out)
+{
+ const uint8_t* p = prob[n][ctx];
+ if (!VP8GetBit(br, p[0]))
+ { /* first EOB is more a 'CBP' bit. */
+ return 0;
+ }
+ while (1)
+ {
+ ++n;
+ if (!VP8GetBit(br, p[1]))
+ {
+ p = prob[kBands[n]][0];
+ }
+ else
+ { /* non zero coeff */
+ int v, j;
+ if (!VP8GetBit(br, p[2]))
+ {
+ p = prob[kBands[n]][1];
+ v = 1;
+ }
+ else
+ {
+ if (!VP8GetBit(br, p[3]))
+ {
+ if (!VP8GetBit(br, p[4]))
+ {
+ v = 2;
+ }
+ else
+ {
+ v = 3 + VP8GetBit(br, p[5]);
+ }
+ }
+ else
+ {
+ if (!VP8GetBit(br, p[6]))
+ {
+ if (!VP8GetBit(br, p[7]))
+ {
+ v = 5 + VP8GetBit(br, 159);
+ } else
+ {
+ v = 7 + 2 * VP8GetBit(br, 165);
+ v += VP8GetBit(br, 145);
+ }
+ }
+ else
+ {
+ const uint8_t* tab;
+ const int bit1 = VP8GetBit(br, p[8]);
+ const int bit0 = VP8GetBit(br, p[9 + bit1]);
+ const int cat = 2 * bit1 + bit0;
+ v = 0;
+ for (tab = kCat3456[cat]; *tab; ++tab)
+ {
+ v += v + VP8GetBit(br, *tab);
+ }
+ v += 3 + (8 << cat);
+ }
+ }
+ p = prob[kBands[n]][2];
+ }
+ j = kZigzag[n - 1];
+
+ out[j] = GetSigned(br, v);
+
+ if (n == 16 || !VP8GetBit(br, p[0]))
+ { /* EOB */
+ return n;
+ }
+ }
+ if (n == 16)
+ {
+ return 16;
+ }
+ }
+}
+
+int vp8_decode_mb_tokens(VP8D_COMP *dx, MACROBLOCKD *x)
+{
+ BOOL_DECODER *bc = x->current_bc;
+ const FRAME_CONTEXT * const fc = &dx->common.fc;
+ char *eobs = x->eobs;
+
+ int i;
+ int nonzeros;
+ int eobtotal = 0;
+
+ short *qcoeff_ptr;
+ ProbaArray coef_probs;
+ ENTROPY_CONTEXT *a_ctx = ((ENTROPY_CONTEXT *)x->above_context);
+ ENTROPY_CONTEXT *l_ctx = ((ENTROPY_CONTEXT *)x->left_context);
+ ENTROPY_CONTEXT *a;
+ ENTROPY_CONTEXT *l;
+ int skip_dc = 0;
+
+ qcoeff_ptr = &x->qcoeff[0];
+
+ if (!x->mode_info_context->mbmi.is_4x4)
+ {
+ a = a_ctx + 8;
+ l = l_ctx + 8;
+
+ coef_probs = fc->coef_probs [1];
+
+ nonzeros = GetCoeffs(bc, coef_probs, (*a + *l), 0, qcoeff_ptr + 24 * 16);
+ *a = *l = (nonzeros > 0);
+
+ eobs[24] = nonzeros;
+ eobtotal += nonzeros - 16;
+
+ coef_probs = fc->coef_probs [0];
+ skip_dc = 1;
+ }
+ else
+ {
+ coef_probs = fc->coef_probs [3];
+ skip_dc = 0;
+ }
+
+ for (i = 0; i < 16; ++i)
+ {
+ a = a_ctx + (i&3);
+ l = l_ctx + ((i&0xc)>>2);
+
+ nonzeros = GetCoeffs(bc, coef_probs, (*a + *l), skip_dc, qcoeff_ptr);
+ *a = *l = (nonzeros > 0);
+
+ nonzeros += skip_dc;
+ eobs[i] = nonzeros;
+ eobtotal += nonzeros;
+ qcoeff_ptr += 16;
+ }
+
+ coef_probs = fc->coef_probs [2];
+
+ a_ctx += 4;
+ l_ctx += 4;
+ for (i = 16; i < 24; ++i)
+ {
+ a = a_ctx + ((i > 19)<<1) + (i&1);
+ l = l_ctx + ((i > 19)<<1) + ((i&3)>1);
+
+ nonzeros = GetCoeffs(bc, coef_probs, (*a + *l), 0, qcoeff_ptr);
+ *a = *l = (nonzeros > 0);
+
+ eobs[i] = nonzeros;
+ eobtotal += nonzeros;
+ qcoeff_ptr += 16;
+ }
+
+ return eobtotal;
+}
+
diff --git a/vp8/decoder/detokenize.h b/libvpx/vp8/decoder/detokenize.h
similarity index 100%
rename from vp8/decoder/detokenize.h
rename to libvpx/vp8/decoder/detokenize.h
diff --git a/libvpx/vp8/decoder/ec_types.h b/libvpx/vp8/decoder/ec_types.h
new file mode 100644
index 0000000..ccb5ddb
--- /dev/null
+++ b/libvpx/vp8/decoder/ec_types.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef VP8_DEC_EC_TYPES_H
+#define VP8_DEC_EC_TYPES_H
+
+#define MAX_OVERLAPS 16
+
+
+
+/* The area (pixel area in Q6) the block pointed to by bmi overlaps
+ * another block with.
+ */
+typedef struct
+{
+ int overlap;
+ union b_mode_info *bmi;
+} OVERLAP_NODE;
+
+/* Structure to keep track of overlapping blocks on a block level. */
+typedef struct
+{
+ /* TODO(holmer): This array should be exchanged for a linked list */
+ OVERLAP_NODE overlaps[MAX_OVERLAPS];
+} B_OVERLAP;
+
+/* Structure used to hold all the overlaps of a macroblock. The overlaps of a
+ * macroblock is further divided into block overlaps.
+ */
+typedef struct
+{
+ B_OVERLAP overlaps[16];
+} MB_OVERLAP;
+
+/* Structure for keeping track of motion vectors and which reference frame they
+ * refer to. Used for motion vector interpolation.
+ */
+typedef struct
+{
+ MV mv;
+ MV_REFERENCE_FRAME ref_frame;
+} EC_BLOCK;
+
+#endif /* VP8_DEC_EC_TYPES_H */
diff --git a/libvpx/vp8/decoder/error_concealment.c b/libvpx/vp8/decoder/error_concealment.c
new file mode 100644
index 0000000..8b2e32b
--- /dev/null
+++ b/libvpx/vp8/decoder/error_concealment.c
@@ -0,0 +1,598 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "error_concealment.h"
+#include "onyxd_int.h"
+#include "decodemv.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/findnearmv.h"
+
+#include <assert.h>
+
+#define MIN(x,y) (((x)<(y))?(x):(y))
+#define MAX(x,y) (((x)>(y))?(x):(y))
+
+#define FLOOR(x,q) ((x) & -(1 << (q)))
+
+#define NUM_NEIGHBORS 20
+
+typedef struct ec_position
+{
+ int row;
+ int col;
+} EC_POS;
+
+/*
+ * Regenerate the table in Matlab with:
+ * x = meshgrid((1:4), (1:4));
+ * y = meshgrid((1:4), (1:4))';
+ * W = round((1./(sqrt(x.^2 + y.^2))*2^7));
+ * W(1,1) = 0;
+ */
+static const int weights_q7[5][5] = {
+ { 0, 128, 64, 43, 32 },
+ {128, 91, 57, 40, 31 },
+ { 64, 57, 45, 36, 29 },
+ { 43, 40, 36, 30, 26 },
+ { 32, 31, 29, 26, 23 }
+};
+
+int vp8_alloc_overlap_lists(VP8D_COMP *pbi)
+{
+ if (pbi->overlaps != NULL)
+ {
+ vpx_free(pbi->overlaps);
+ pbi->overlaps = NULL;
+ }
+
+ pbi->overlaps = vpx_calloc(pbi->common.mb_rows * pbi->common.mb_cols,
+ sizeof(MB_OVERLAP));
+
+ if (pbi->overlaps == NULL)
+ return -1;
+
+ return 0;
+}
+
+void vp8_de_alloc_overlap_lists(VP8D_COMP *pbi)
+{
+ vpx_free(pbi->overlaps);
+ pbi->overlaps = NULL;
+}
+
+/* Inserts a new overlap area value to the list of overlaps of a block */
+static void assign_overlap(OVERLAP_NODE* overlaps,
+ union b_mode_info *bmi,
+ int overlap)
+{
+ int i;
+ if (overlap <= 0)
+ return;
+ /* Find and assign to the next empty overlap node in the list of overlaps.
+ * Empty is defined as bmi == NULL */
+ for (i = 0; i < MAX_OVERLAPS; i++)
+ {
+ if (overlaps[i].bmi == NULL)
+ {
+ overlaps[i].bmi = bmi;
+ overlaps[i].overlap = overlap;
+ break;
+ }
+ }
+}
+
+/* Calculates the overlap area between two 4x4 squares, where the first
+ * square has its upper-left corner at (b1_row, b1_col) and the second
+ * square has its upper-left corner at (b2_row, b2_col). Doesn't
+ * properly handle squares which do not overlap.
+ */
+static int block_overlap(int b1_row, int b1_col, int b2_row, int b2_col)
+{
+ const int int_top = MAX(b1_row, b2_row); // top
+ const int int_left = MAX(b1_col, b2_col); // left
+ /* Since each block is 4x4 pixels, adding 4 (Q3) to the left/top edge
+ * gives us the right/bottom edge.
+ */
+ const int int_right = MIN(b1_col + (4<<3), b2_col + (4<<3)); // right
+ const int int_bottom = MIN(b1_row + (4<<3), b2_row + (4<<3)); // bottom
+ return (int_bottom - int_top) * (int_right - int_left);
+}
+
+/* Calculates the overlap area for all blocks in a macroblock at position
+ * (mb_row, mb_col) in macroblocks, which are being overlapped by a given
+ * overlapping block at position (new_row, new_col) (in pixels, Q3). The
+ * first block being overlapped in the macroblock has position (first_blk_row,
+ * first_blk_col) in blocks relative the upper-left corner of the image.
+ */
+static void calculate_overlaps_mb(B_OVERLAP *b_overlaps, union b_mode_info *bmi,
+ int new_row, int new_col,
+ int mb_row, int mb_col,
+ int first_blk_row, int first_blk_col)
+{
+ /* Find the blocks within this MB (defined by mb_row, mb_col) which are
+ * overlapped by bmi and calculate and assign overlap for each of those
+ * blocks. */
+
+ /* Block coordinates relative the upper-left block */
+ const int rel_ol_blk_row = first_blk_row - mb_row * 4;
+ const int rel_ol_blk_col = first_blk_col - mb_col * 4;
+ /* If the block partly overlaps any previous MB, these coordinates
+ * can be < 0. We don't want to access blocks in previous MBs.
+ */
+ const int blk_idx = MAX(rel_ol_blk_row,0) * 4 + MAX(rel_ol_blk_col,0);
+ /* Upper left overlapping block */
+ B_OVERLAP *b_ol_ul = &(b_overlaps[blk_idx]);
+
+ /* Calculate and assign overlaps for all blocks in this MB
+ * which the motion compensated block overlaps
+ */
+ /* Avoid calculating overlaps for blocks in later MBs */
+ int end_row = MIN(4 + mb_row * 4 - first_blk_row, 2);
+ int end_col = MIN(4 + mb_col * 4 - first_blk_col, 2);
+ int row, col;
+
+ /* Check if new_row and new_col are evenly divisible by 4 (Q3),
+ * and if so we shouldn't check neighboring blocks
+ */
+ if (new_row >= 0 && (new_row & 0x1F) == 0)
+ end_row = 1;
+ if (new_col >= 0 && (new_col & 0x1F) == 0)
+ end_col = 1;
+
+ /* Check if the overlapping block partly overlaps a previous MB
+ * and if so, we're overlapping fewer blocks in this MB.
+ */
+ if (new_row < (mb_row*16)<<3)
+ end_row = 1;
+ if (new_col < (mb_col*16)<<3)
+ end_col = 1;
+
+ for (row = 0; row < end_row; ++row)
+ {
+ for (col = 0; col < end_col; ++col)
+ {
+ /* input in Q3, result in Q6 */
+ const int overlap = block_overlap(new_row, new_col,
+ (((first_blk_row + row) *
+ 4) << 3),
+ (((first_blk_col + col) *
+ 4) << 3));
+ assign_overlap(b_ol_ul[row * 4 + col].overlaps, bmi, overlap);
+ }
+ }
+}
+
+void vp8_calculate_overlaps(MB_OVERLAP *overlap_ul,
+ int mb_rows, int mb_cols,
+ union b_mode_info *bmi,
+ int b_row, int b_col)
+{
+ MB_OVERLAP *mb_overlap;
+ int row, col, rel_row, rel_col;
+ int new_row, new_col;
+ int end_row, end_col;
+ int overlap_b_row, overlap_b_col;
+ int overlap_mb_row, overlap_mb_col;
+
+ /* mb subpixel position */
+ row = (4 * b_row) << 3; /* Q3 */
+ col = (4 * b_col) << 3; /* Q3 */
+
+ /* reverse compensate for motion */
+ new_row = row - bmi->mv.as_mv.row;
+ new_col = col - bmi->mv.as_mv.col;
+
+ if (new_row >= ((16*mb_rows) << 3) || new_col >= ((16*mb_cols) << 3))
+ {
+ /* the new block ended up outside the frame */
+ return;
+ }
+
+ if (new_row <= (-4 << 3) || new_col <= (-4 << 3))
+ {
+ /* outside the frame */
+ return;
+ }
+ /* overlapping block's position in blocks */
+ overlap_b_row = FLOOR(new_row / 4, 3) >> 3;
+ overlap_b_col = FLOOR(new_col / 4, 3) >> 3;
+
+ /* overlapping block's MB position in MBs
+ * operations are done in Q3
+ */
+ overlap_mb_row = FLOOR((overlap_b_row << 3) / 4, 3) >> 3;
+ overlap_mb_col = FLOOR((overlap_b_col << 3) / 4, 3) >> 3;
+
+ end_row = MIN(mb_rows - overlap_mb_row, 2);
+ end_col = MIN(mb_cols - overlap_mb_col, 2);
+
+ /* Don't calculate overlap for MBs we don't overlap */
+ /* Check if the new block row starts at the last block row of the MB */
+ if (abs(new_row - ((16*overlap_mb_row) << 3)) < ((3*4) << 3))
+ end_row = 1;
+ /* Check if the new block col starts at the last block col of the MB */
+ if (abs(new_col - ((16*overlap_mb_col) << 3)) < ((3*4) << 3))
+ end_col = 1;
+
+ /* find the MB(s) this block is overlapping */
+ for (rel_row = 0; rel_row < end_row; ++rel_row)
+ {
+ for (rel_col = 0; rel_col < end_col; ++rel_col)
+ {
+ if (overlap_mb_row + rel_row < 0 ||
+ overlap_mb_col + rel_col < 0)
+ continue;
+ mb_overlap = overlap_ul + (overlap_mb_row + rel_row) * mb_cols +
+ overlap_mb_col + rel_col;
+
+ calculate_overlaps_mb(mb_overlap->overlaps, bmi,
+ new_row, new_col,
+ overlap_mb_row + rel_row,
+ overlap_mb_col + rel_col,
+ overlap_b_row + rel_row,
+ overlap_b_col + rel_col);
+ }
+ }
+}
+
+/* Estimates a motion vector given the overlapping blocks' motion vectors.
+ * Filters out all overlapping blocks which do not refer to the correct
+ * reference frame type.
+ */
+static void estimate_mv(const OVERLAP_NODE *overlaps, union b_mode_info *bmi)
+{
+ int i;
+ int overlap_sum = 0;
+ int row_acc = 0;
+ int col_acc = 0;
+
+ bmi->mv.as_int = 0;
+ for (i=0; i < MAX_OVERLAPS; ++i)
+ {
+ if (overlaps[i].bmi == NULL)
+ break;
+ col_acc += overlaps[i].overlap * overlaps[i].bmi->mv.as_mv.col;
+ row_acc += overlaps[i].overlap * overlaps[i].bmi->mv.as_mv.row;
+ overlap_sum += overlaps[i].overlap;
+ }
+ if (overlap_sum > 0)
+ {
+ /* Q9 / Q6 = Q3 */
+ bmi->mv.as_mv.col = col_acc / overlap_sum;
+ bmi->mv.as_mv.row = row_acc / overlap_sum;
+ }
+ else
+ {
+ bmi->mv.as_mv.col = 0;
+ bmi->mv.as_mv.row = 0;
+ }
+}
+
+/* Estimates all motion vectors for a macroblock given the lists of
+ * overlaps for each block. Decides whether or not the MVs must be clamped.
+ */
+static void estimate_mb_mvs(const B_OVERLAP *block_overlaps,
+ MODE_INFO *mi,
+ int mb_to_left_edge,
+ int mb_to_right_edge,
+ int mb_to_top_edge,
+ int mb_to_bottom_edge)
+{
+ int row, col;
+ int non_zero_count = 0;
+ MV * const filtered_mv = &(mi->mbmi.mv.as_mv);
+ union b_mode_info * const bmi = mi->bmi;
+ filtered_mv->col = 0;
+ filtered_mv->row = 0;
+ mi->mbmi.need_to_clamp_mvs = 0;
+ for (row = 0; row < 4; ++row)
+ {
+ int this_b_to_top_edge = mb_to_top_edge + ((row*4)<<3);
+ int this_b_to_bottom_edge = mb_to_bottom_edge - ((row*4)<<3);
+ for (col = 0; col < 4; ++col)
+ {
+ int i = row * 4 + col;
+ int this_b_to_left_edge = mb_to_left_edge + ((col*4)<<3);
+ int this_b_to_right_edge = mb_to_right_edge - ((col*4)<<3);
+ /* Estimate vectors for all blocks which are overlapped by this */
+ /* type. Interpolate/extrapolate the rest of the block's MVs */
+ estimate_mv(block_overlaps[i].overlaps, &(bmi[i]));
+ mi->mbmi.need_to_clamp_mvs |= vp8_check_mv_bounds(
+ &bmi[i].mv,
+ this_b_to_left_edge,
+ this_b_to_right_edge,
+ this_b_to_top_edge,
+ this_b_to_bottom_edge);
+ if (bmi[i].mv.as_int != 0)
+ {
+ ++non_zero_count;
+ filtered_mv->col += bmi[i].mv.as_mv.col;
+ filtered_mv->row += bmi[i].mv.as_mv.row;
+ }
+ }
+ }
+ if (non_zero_count > 0)
+ {
+ filtered_mv->col /= non_zero_count;
+ filtered_mv->row /= non_zero_count;
+ }
+}
+
+static void calc_prev_mb_overlaps(MB_OVERLAP *overlaps, MODE_INFO *prev_mi,
+ int mb_row, int mb_col,
+ int mb_rows, int mb_cols)
+{
+ int sub_row;
+ int sub_col;
+ for (sub_row = 0; sub_row < 4; ++sub_row)
+ {
+ for (sub_col = 0; sub_col < 4; ++sub_col)
+ {
+ vp8_calculate_overlaps(
+ overlaps, mb_rows, mb_cols,
+ &(prev_mi->bmi[sub_row * 4 + sub_col]),
+ 4 * mb_row + sub_row,
+ 4 * mb_col + sub_col);
+ }
+ }
+}
+
+/* Estimate all missing motion vectors. This function does the same as the one
+ * above, but has different input arguments. */
+static void estimate_missing_mvs(MB_OVERLAP *overlaps,
+ MODE_INFO *mi, MODE_INFO *prev_mi,
+ int mb_rows, int mb_cols,
+ unsigned int first_corrupt)
+{
+ int mb_row, mb_col;
+ vpx_memset(overlaps, 0, sizeof(MB_OVERLAP) * mb_rows * mb_cols);
+ /* First calculate the overlaps for all blocks */
+ for (mb_row = 0; mb_row < mb_rows; ++mb_row)
+ {
+ for (mb_col = 0; mb_col < mb_cols; ++mb_col)
+ {
+ /* We're only able to use blocks referring to the last frame
+ * when extrapolating new vectors.
+ */
+ if (prev_mi->mbmi.ref_frame == LAST_FRAME)
+ {
+ calc_prev_mb_overlaps(overlaps, prev_mi,
+ mb_row, mb_col,
+ mb_rows, mb_cols);
+ }
+ ++prev_mi;
+ }
+ ++prev_mi;
+ }
+
+ mb_row = first_corrupt / mb_cols;
+ mb_col = first_corrupt - mb_row * mb_cols;
+ mi += mb_row*(mb_cols + 1) + mb_col;
+ /* Go through all macroblocks in the current image with missing MVs
+ * and calculate new MVs using the overlaps.
+ */
+ for (; mb_row < mb_rows; ++mb_row)
+ {
+ int mb_to_top_edge = -((mb_row * 16)) << 3;
+ int mb_to_bottom_edge = ((mb_rows - 1 - mb_row) * 16) << 3;
+ for (; mb_col < mb_cols; ++mb_col)
+ {
+ int mb_to_left_edge = -((mb_col * 16) << 3);
+ int mb_to_right_edge = ((mb_cols - 1 - mb_col) * 16) << 3;
+ const B_OVERLAP *block_overlaps =
+ overlaps[mb_row*mb_cols + mb_col].overlaps;
+ mi->mbmi.ref_frame = LAST_FRAME;
+ mi->mbmi.mode = SPLITMV;
+ mi->mbmi.uv_mode = DC_PRED;
+ mi->mbmi.partitioning = 3;
+ mi->mbmi.segment_id = 0;
+ estimate_mb_mvs(block_overlaps,
+ mi,
+ mb_to_left_edge,
+ mb_to_right_edge,
+ mb_to_top_edge,
+ mb_to_bottom_edge);
+ ++mi;
+ }
+ mb_col = 0;
+ ++mi;
+ }
+}
+
+void vp8_estimate_missing_mvs(VP8D_COMP *pbi)
+{
+ VP8_COMMON * const pc = &pbi->common;
+ estimate_missing_mvs(pbi->overlaps,
+ pc->mi, pc->prev_mi,
+ pc->mb_rows, pc->mb_cols,
+ pbi->mvs_corrupt_from_mb);
+}
+
+static void assign_neighbor(EC_BLOCK *neighbor, MODE_INFO *mi, int block_idx)
+{
+ assert(mi->mbmi.ref_frame < MAX_REF_FRAMES);
+ neighbor->ref_frame = mi->mbmi.ref_frame;
+ neighbor->mv = mi->bmi[block_idx].mv.as_mv;
+}
+
+/* Finds the neighboring blocks of a macroblocks. In the general case
+ * 20 blocks are found. If a fewer number of blocks are found due to
+ * image boundaries, those positions in the EC_BLOCK array are left "empty".
+ * The neighbors are enumerated with the upper-left neighbor as the first
+ * element, the second element refers to the neighbor to right of the previous
+ * neighbor, and so on. The last element refers to the neighbor below the first
+ * neighbor.
+ */
+static void find_neighboring_blocks(MODE_INFO *mi,
+ EC_BLOCK *neighbors,
+ int mb_row, int mb_col,
+ int mb_rows, int mb_cols,
+ int mi_stride)
+{
+ int i = 0;
+ int j;
+ if (mb_row > 0)
+ {
+ /* upper left */
+ if (mb_col > 0)
+ assign_neighbor(&neighbors[i], mi - mi_stride - 1, 15);
+ ++i;
+ /* above */
+ for (j = 12; j < 16; ++j, ++i)
+ assign_neighbor(&neighbors[i], mi - mi_stride, j);
+ }
+ else
+ i += 5;
+ if (mb_col < mb_cols - 1)
+ {
+ /* upper right */
+ if (mb_row > 0)
+ assign_neighbor(&neighbors[i], mi - mi_stride + 1, 12);
+ ++i;
+ /* right */
+ for (j = 0; j <= 12; j += 4, ++i)
+ assign_neighbor(&neighbors[i], mi + 1, j);
+ }
+ else
+ i += 5;
+ if (mb_row < mb_rows - 1)
+ {
+ /* lower right */
+ if (mb_col < mb_cols - 1)
+ assign_neighbor(&neighbors[i], mi + mi_stride + 1, 0);
+ ++i;
+ /* below */
+ for (j = 0; j < 4; ++j, ++i)
+ assign_neighbor(&neighbors[i], mi + mi_stride, j);
+ }
+ else
+ i += 5;
+ if (mb_col > 0)
+ {
+ /* lower left */
+ if (mb_row < mb_rows - 1)
+ assign_neighbor(&neighbors[i], mi + mi_stride - 1, 4);
+ ++i;
+ /* left */
+ for (j = 3; j < 16; j += 4, ++i)
+ {
+ assign_neighbor(&neighbors[i], mi - 1, j);
+ }
+ }
+ else
+ i += 5;
+ assert(i == 20);
+}
+
+/* Interpolates all motion vectors for a macroblock from the neighboring blocks'
+ * motion vectors.
+ */
+static void interpolate_mvs(MACROBLOCKD *mb,
+ EC_BLOCK *neighbors,
+ MV_REFERENCE_FRAME dom_ref_frame)
+{
+ int row, col, i;
+ MODE_INFO * const mi = mb->mode_info_context;
+ /* Table with the position of the neighboring blocks relative the position
+ * of the upper left block of the current MB. Starting with the upper left
+ * neighbor and going to the right.
+ */
+ const EC_POS neigh_pos[NUM_NEIGHBORS] = {
+ {-1,-1}, {-1,0}, {-1,1}, {-1,2}, {-1,3},
+ {-1,4}, {0,4}, {1,4}, {2,4}, {3,4},
+ {4,4}, {4,3}, {4,2}, {4,1}, {4,0},
+ {4,-1}, {3,-1}, {2,-1}, {1,-1}, {0,-1}
+ };
+ mi->mbmi.need_to_clamp_mvs = 0;
+ for (row = 0; row < 4; ++row)
+ {
+ int mb_to_top_edge = mb->mb_to_top_edge + ((row*4)<<3);
+ int mb_to_bottom_edge = mb->mb_to_bottom_edge - ((row*4)<<3);
+ for (col = 0; col < 4; ++col)
+ {
+ int mb_to_left_edge = mb->mb_to_left_edge + ((col*4)<<3);
+ int mb_to_right_edge = mb->mb_to_right_edge - ((col*4)<<3);
+ int w_sum = 0;
+ int mv_row_sum = 0;
+ int mv_col_sum = 0;
+ int_mv * const mv = &(mi->bmi[row*4 + col].mv);
+ mv->as_int = 0;
+ for (i = 0; i < NUM_NEIGHBORS; ++i)
+ {
+ /* Calculate the weighted sum of neighboring MVs referring
+ * to the dominant frame type.
+ */
+ const int w = weights_q7[abs(row - neigh_pos[i].row)]
+ [abs(col - neigh_pos[i].col)];
+ if (neighbors[i].ref_frame != dom_ref_frame)
+ continue;
+ w_sum += w;
+ /* Q7 * Q3 = Q10 */
+ mv_row_sum += w*neighbors[i].mv.row;
+ mv_col_sum += w*neighbors[i].mv.col;
+ }
+ if (w_sum > 0)
+ {
+ /* Avoid division by zero.
+ * Normalize with the sum of the coefficients
+ * Q3 = Q10 / Q7
+ */
+ mv->as_mv.row = mv_row_sum / w_sum;
+ mv->as_mv.col = mv_col_sum / w_sum;
+ mi->mbmi.need_to_clamp_mvs |= vp8_check_mv_bounds(
+ mv,
+ mb_to_left_edge,
+ mb_to_right_edge,
+ mb_to_top_edge,
+ mb_to_bottom_edge);
+ }
+ }
+ }
+}
+
+void vp8_interpolate_motion(MACROBLOCKD *mb,
+ int mb_row, int mb_col,
+ int mb_rows, int mb_cols,
+ int mi_stride)
+{
+ /* Find relevant neighboring blocks */
+ EC_BLOCK neighbors[NUM_NEIGHBORS];
+ int i;
+ /* Initialize the array. MAX_REF_FRAMES is interpreted as "doesn't exist" */
+ for (i = 0; i < NUM_NEIGHBORS; ++i)
+ {
+ neighbors[i].ref_frame = MAX_REF_FRAMES;
+ neighbors[i].mv.row = neighbors[i].mv.col = 0;
+ }
+ find_neighboring_blocks(mb->mode_info_context,
+ neighbors,
+ mb_row, mb_col,
+ mb_rows, mb_cols,
+ mb->mode_info_stride);
+ /* Interpolate MVs for the missing blocks from the surrounding
+ * blocks which refer to the last frame. */
+ interpolate_mvs(mb, neighbors, LAST_FRAME);
+
+ mb->mode_info_context->mbmi.ref_frame = LAST_FRAME;
+ mb->mode_info_context->mbmi.mode = SPLITMV;
+ mb->mode_info_context->mbmi.uv_mode = DC_PRED;
+ mb->mode_info_context->mbmi.partitioning = 3;
+ mb->mode_info_context->mbmi.segment_id = 0;
+}
+
+void vp8_conceal_corrupt_mb(MACROBLOCKD *xd)
+{
+ /* This macroblock has corrupt residual, use the motion compensated
+ image (predictor) for concealment */
+
+ /* The build predictor functions now output directly into the dst buffer,
+ * so the copies are no longer necessary */
+
+}
diff --git a/libvpx/vp8/decoder/error_concealment.h b/libvpx/vp8/decoder/error_concealment.h
new file mode 100644
index 0000000..65ae9d9
--- /dev/null
+++ b/libvpx/vp8/decoder/error_concealment.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef ERROR_CONCEALMENT_H
+#define ERROR_CONCEALMENT_H
+
+#include "onyxd_int.h"
+#include "ec_types.h"
+
+/* Allocate memory for the overlap lists */
+int vp8_alloc_overlap_lists(VP8D_COMP *pbi);
+
+/* Deallocate the overlap lists */
+void vp8_de_alloc_overlap_lists(VP8D_COMP *pbi);
+
+/* Estimate all missing motion vectors. */
+void vp8_estimate_missing_mvs(VP8D_COMP *pbi);
+
+/* Functions for spatial MV interpolation */
+
+/* Interpolates all motion vectors for a macroblock mb at position
+ * (mb_row, mb_col). */
+void vp8_interpolate_motion(MACROBLOCKD *mb,
+ int mb_row, int mb_col,
+ int mb_rows, int mb_cols,
+ int mi_stride);
+
+/* Conceal a macroblock with corrupt residual.
+ * Copies the prediction signal to the reconstructed image.
+ */
+void vp8_conceal_corrupt_mb(MACROBLOCKD *xd);
+
+#endif
diff --git a/libvpx/vp8/decoder/onyxd_if.c b/libvpx/vp8/decoder/onyxd_if.c
new file mode 100644
index 0000000..8d6871b
--- /dev/null
+++ b/libvpx/vp8/decoder/onyxd_if.c
@@ -0,0 +1,522 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/common/onyxc_int.h"
+#if CONFIG_POSTPROC
+#include "vp8/common/postproc.h"
+#endif
+#include "vp8/common/onyxd.h"
+#include "onyxd_int.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/alloccommon.h"
+#include "vp8/common/loopfilter.h"
+#include "vp8/common/swapyv12buffer.h"
+#include "vp8/common/threading.h"
+#include "decoderthreading.h"
+#include <stdio.h>
+#include <assert.h>
+
+#include "vp8/common/quant_common.h"
+#include "vpx_scale/vpxscale.h"
+#include "vp8/common/systemdependent.h"
+#include "vpx_ports/vpx_timer.h"
+#include "detokenize.h"
+#if CONFIG_ERROR_CONCEALMENT
+#include "error_concealment.h"
+#endif
+#if ARCH_ARM
+#include "vpx_ports/arm.h"
+#endif
+
+extern void vp8_init_loop_filter(VP8_COMMON *cm);
+extern void vp8cx_init_de_quantizer(VP8D_COMP *pbi);
+static int get_free_fb (VP8_COMMON *cm);
+static void ref_cnt_fb (int *buf, int *idx, int new_idx);
+
+struct VP8D_COMP * vp8dx_create_decompressor(VP8D_CONFIG *oxcf)
+{
+ VP8D_COMP *pbi = vpx_memalign(32, sizeof(VP8D_COMP));
+
+ if (!pbi)
+ return NULL;
+
+ vpx_memset(pbi, 0, sizeof(VP8D_COMP));
+
+ if (setjmp(pbi->common.error.jmp))
+ {
+ pbi->common.error.setjmp = 0;
+ vp8dx_remove_decompressor(pbi);
+ return 0;
+ }
+
+ pbi->common.error.setjmp = 1;
+
+ vp8_create_common(&pbi->common);
+
+ pbi->common.current_video_frame = 0;
+ pbi->ready_for_new_data = 1;
+
+#if CONFIG_MULTITHREAD
+ pbi->max_threads = oxcf->max_threads;
+ vp8_decoder_create_threads(pbi);
+#endif
+
+ /* vp8cx_init_de_quantizer() is first called here. Add check in frame_init_dequantizer() to avoid
+ * unnecessary calling of vp8cx_init_de_quantizer() for every frame.
+ */
+ vp8cx_init_de_quantizer(pbi);
+
+ vp8_loop_filter_init(&pbi->common);
+
+ pbi->common.error.setjmp = 0;
+
+#if CONFIG_ERROR_CONCEALMENT
+ pbi->ec_enabled = oxcf->error_concealment;
+ pbi->overlaps = NULL;
+#else
+ pbi->ec_enabled = 0;
+#endif
+ /* Error concealment is activated after a key frame has been
+ * decoded without errors when error concealment is enabled.
+ */
+ pbi->ec_active = 0;
+
+ pbi->decoded_key_frame = 0;
+
+ pbi->input_fragments = oxcf->input_fragments;
+ pbi->num_fragments = 0;
+
+ /* Independent partitions is activated when a frame updates the
+ * token probability table to have equal probabilities over the
+ * PREV_COEF context.
+ */
+ pbi->independent_partitions = 0;
+
+ vp8_setup_block_dptrs(&pbi->mb);
+
+ return pbi;
+}
+
+
+void vp8dx_remove_decompressor(VP8D_COMP *pbi)
+{
+ if (!pbi)
+ return;
+
+#if CONFIG_MULTITHREAD
+ if (pbi->b_multithreaded_rd)
+ vp8mt_de_alloc_temp_buffers(pbi, pbi->common.mb_rows);
+ vp8_decoder_remove_threads(pbi);
+#endif
+#if CONFIG_ERROR_CONCEALMENT
+ vp8_de_alloc_overlap_lists(pbi);
+#endif
+ vp8_remove_common(&pbi->common);
+ vpx_free(pbi);
+}
+
+
+vpx_codec_err_t vp8dx_get_reference(VP8D_COMP *pbi, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd)
+{
+ VP8_COMMON *cm = &pbi->common;
+ int ref_fb_idx;
+
+ if (ref_frame_flag == VP8_LAST_FRAME)
+ ref_fb_idx = cm->lst_fb_idx;
+ else if (ref_frame_flag == VP8_GOLD_FRAME)
+ ref_fb_idx = cm->gld_fb_idx;
+ else if (ref_frame_flag == VP8_ALTR_FRAME)
+ ref_fb_idx = cm->alt_fb_idx;
+ else{
+ vpx_internal_error(&pbi->common.error, VPX_CODEC_ERROR,
+ "Invalid reference frame");
+ return pbi->common.error.error_code;
+ }
+
+ if(cm->yv12_fb[ref_fb_idx].y_height != sd->y_height ||
+ cm->yv12_fb[ref_fb_idx].y_width != sd->y_width ||
+ cm->yv12_fb[ref_fb_idx].uv_height != sd->uv_height ||
+ cm->yv12_fb[ref_fb_idx].uv_width != sd->uv_width){
+ vpx_internal_error(&pbi->common.error, VPX_CODEC_ERROR,
+ "Incorrect buffer dimensions");
+ }
+ else
+ vp8_yv12_copy_frame(&cm->yv12_fb[ref_fb_idx], sd);
+
+ return pbi->common.error.error_code;
+}
+
+
+vpx_codec_err_t vp8dx_set_reference(VP8D_COMP *pbi, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd)
+{
+ VP8_COMMON *cm = &pbi->common;
+ int *ref_fb_ptr = NULL;
+ int free_fb;
+
+ if (ref_frame_flag == VP8_LAST_FRAME)
+ ref_fb_ptr = &cm->lst_fb_idx;
+ else if (ref_frame_flag == VP8_GOLD_FRAME)
+ ref_fb_ptr = &cm->gld_fb_idx;
+ else if (ref_frame_flag == VP8_ALTR_FRAME)
+ ref_fb_ptr = &cm->alt_fb_idx;
+ else{
+ vpx_internal_error(&pbi->common.error, VPX_CODEC_ERROR,
+ "Invalid reference frame");
+ return pbi->common.error.error_code;
+ }
+
+ if(cm->yv12_fb[*ref_fb_ptr].y_height != sd->y_height ||
+ cm->yv12_fb[*ref_fb_ptr].y_width != sd->y_width ||
+ cm->yv12_fb[*ref_fb_ptr].uv_height != sd->uv_height ||
+ cm->yv12_fb[*ref_fb_ptr].uv_width != sd->uv_width){
+ vpx_internal_error(&pbi->common.error, VPX_CODEC_ERROR,
+ "Incorrect buffer dimensions");
+ }
+ else{
+ /* Find an empty frame buffer. */
+ free_fb = get_free_fb(cm);
+ /* Decrease fb_idx_ref_cnt since it will be increased again in
+ * ref_cnt_fb() below. */
+ cm->fb_idx_ref_cnt[free_fb]--;
+
+ /* Manage the reference counters and copy image. */
+ ref_cnt_fb (cm->fb_idx_ref_cnt, ref_fb_ptr, free_fb);
+ vp8_yv12_copy_frame(sd, &cm->yv12_fb[*ref_fb_ptr]);
+ }
+
+ return pbi->common.error.error_code;
+}
+
+/*For ARM NEON, d8-d15 are callee-saved registers, and need to be saved by us.*/
+#if HAVE_NEON
+extern void vp8_push_neon(int64_t *store);
+extern void vp8_pop_neon(int64_t *store);
+#endif
+
+static int get_free_fb (VP8_COMMON *cm)
+{
+ int i;
+ for (i = 0; i < NUM_YV12_BUFFERS; i++)
+ if (cm->fb_idx_ref_cnt[i] == 0)
+ break;
+
+ assert(i < NUM_YV12_BUFFERS);
+ cm->fb_idx_ref_cnt[i] = 1;
+ return i;
+}
+
+static void ref_cnt_fb (int *buf, int *idx, int new_idx)
+{
+ if (buf[*idx] > 0)
+ buf[*idx]--;
+
+ *idx = new_idx;
+
+ buf[new_idx]++;
+}
+
+/* If any buffer copy / swapping is signalled it should be done here. */
+static int swap_frame_buffers (VP8_COMMON *cm)
+{
+ int err = 0;
+
+ /* The alternate reference frame or golden frame can be updated
+ * using the new, last, or golden/alt ref frame. If it
+ * is updated using the newly decoded frame it is a refresh.
+ * An update using the last or golden/alt ref frame is a copy.
+ */
+ if (cm->copy_buffer_to_arf)
+ {
+ int new_fb = 0;
+
+ if (cm->copy_buffer_to_arf == 1)
+ new_fb = cm->lst_fb_idx;
+ else if (cm->copy_buffer_to_arf == 2)
+ new_fb = cm->gld_fb_idx;
+ else
+ err = -1;
+
+ ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->alt_fb_idx, new_fb);
+ }
+
+ if (cm->copy_buffer_to_gf)
+ {
+ int new_fb = 0;
+
+ if (cm->copy_buffer_to_gf == 1)
+ new_fb = cm->lst_fb_idx;
+ else if (cm->copy_buffer_to_gf == 2)
+ new_fb = cm->alt_fb_idx;
+ else
+ err = -1;
+
+ ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->gld_fb_idx, new_fb);
+ }
+
+ if (cm->refresh_golden_frame)
+ ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->gld_fb_idx, cm->new_fb_idx);
+
+ if (cm->refresh_alt_ref_frame)
+ ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->alt_fb_idx, cm->new_fb_idx);
+
+ if (cm->refresh_last_frame)
+ {
+ ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->lst_fb_idx, cm->new_fb_idx);
+
+ cm->frame_to_show = &cm->yv12_fb[cm->lst_fb_idx];
+ }
+ else
+ cm->frame_to_show = &cm->yv12_fb[cm->new_fb_idx];
+
+ cm->fb_idx_ref_cnt[cm->new_fb_idx]--;
+
+ return err;
+}
+
+int vp8dx_receive_compressed_data(VP8D_COMP *pbi, size_t size,
+ const uint8_t *source,
+ int64_t time_stamp)
+{
+#if HAVE_NEON
+ int64_t dx_store_reg[8];
+#endif
+ VP8_COMMON *cm = &pbi->common;
+ int retcode = -1;
+
+ pbi->common.error.error_code = VPX_CODEC_OK;
+
+ if (pbi->num_fragments == 0)
+ {
+ /* New frame, reset fragment pointers and sizes */
+ vpx_memset((void*)pbi->fragments, 0, sizeof(pbi->fragments));
+ vpx_memset(pbi->fragment_sizes, 0, sizeof(pbi->fragment_sizes));
+ }
+ if (pbi->input_fragments && !(source == NULL && size == 0))
+ {
+ /* Store a pointer to this fragment and return. We haven't
+ * received the complete frame yet, so we will wait with decoding.
+ */
+ assert(pbi->num_fragments < MAX_PARTITIONS);
+ pbi->fragments[pbi->num_fragments] = source;
+ pbi->fragment_sizes[pbi->num_fragments] = size;
+ pbi->num_fragments++;
+ if (pbi->num_fragments > (1 << EIGHT_PARTITION) + 1)
+ {
+ pbi->common.error.error_code = VPX_CODEC_UNSUP_BITSTREAM;
+ pbi->common.error.setjmp = 0;
+ pbi->num_fragments = 0;
+ return -1;
+ }
+ return 0;
+ }
+
+ if (!pbi->input_fragments)
+ {
+ pbi->fragments[0] = source;
+ pbi->fragment_sizes[0] = size;
+ pbi->num_fragments = 1;
+ }
+ assert(pbi->common.multi_token_partition <= EIGHT_PARTITION);
+ if (pbi->num_fragments == 0)
+ {
+ pbi->num_fragments = 1;
+ pbi->fragments[0] = NULL;
+ pbi->fragment_sizes[0] = 0;
+ }
+
+ if (!pbi->ec_active &&
+ pbi->num_fragments <= 1 && pbi->fragment_sizes[0] == 0)
+ {
+ /* If error concealment is disabled we won't signal missing frames
+ * to the decoder.
+ */
+ if (cm->fb_idx_ref_cnt[cm->lst_fb_idx] > 1)
+ {
+ /* The last reference shares buffer with another reference
+ * buffer. Move it to its own buffer before setting it as
+ * corrupt, otherwise we will make multiple buffers corrupt.
+ */
+ const int prev_idx = cm->lst_fb_idx;
+ cm->fb_idx_ref_cnt[prev_idx]--;
+ cm->lst_fb_idx = get_free_fb(cm);
+ vp8_yv12_copy_frame(&cm->yv12_fb[prev_idx],
+ &cm->yv12_fb[cm->lst_fb_idx]);
+ }
+ /* This is used to signal that we are missing frames.
+ * We do not know if the missing frame(s) was supposed to update
+ * any of the reference buffers, but we act conservative and
+ * mark only the last buffer as corrupted.
+ */
+ cm->yv12_fb[cm->lst_fb_idx].corrupted = 1;
+
+ /* Signal that we have no frame to show. */
+ cm->show_frame = 0;
+
+ pbi->num_fragments = 0;
+
+ /* Nothing more to do. */
+ return 0;
+ }
+
+#if HAVE_NEON
+#if CONFIG_RUNTIME_CPU_DETECT
+ if (cm->cpu_caps & HAS_NEON)
+#endif
+ {
+ vp8_push_neon(dx_store_reg);
+ }
+#endif
+
+ cm->new_fb_idx = get_free_fb (cm);
+
+ /* setup reference frames for vp8_decode_frame */
+ pbi->dec_fb_ref[INTRA_FRAME] = &cm->yv12_fb[cm->new_fb_idx];
+ pbi->dec_fb_ref[LAST_FRAME] = &cm->yv12_fb[cm->lst_fb_idx];
+ pbi->dec_fb_ref[GOLDEN_FRAME] = &cm->yv12_fb[cm->gld_fb_idx];
+ pbi->dec_fb_ref[ALTREF_FRAME] = &cm->yv12_fb[cm->alt_fb_idx];
+
+ if (setjmp(pbi->common.error.jmp))
+ {
+ /* We do not know if the missing frame(s) was supposed to update
+ * any of the reference buffers, but we act conservative and
+ * mark only the last buffer as corrupted.
+ */
+ cm->yv12_fb[cm->lst_fb_idx].corrupted = 1;
+
+ if (cm->fb_idx_ref_cnt[cm->new_fb_idx] > 0)
+ cm->fb_idx_ref_cnt[cm->new_fb_idx]--;
+
+ goto decode_exit;
+ }
+
+ pbi->common.error.setjmp = 1;
+
+ retcode = vp8_decode_frame(pbi);
+
+ if (retcode < 0)
+ {
+ if (cm->fb_idx_ref_cnt[cm->new_fb_idx] > 0)
+ cm->fb_idx_ref_cnt[cm->new_fb_idx]--;
+
+ pbi->common.error.error_code = VPX_CODEC_ERROR;
+ goto decode_exit;
+ }
+
+ if (swap_frame_buffers (cm))
+ {
+ pbi->common.error.error_code = VPX_CODEC_ERROR;
+ goto decode_exit;
+ }
+
+ vp8_clear_system_state();
+
+#if CONFIG_ERROR_CONCEALMENT
+ /* swap the mode infos to storage for future error concealment */
+ if (pbi->ec_enabled && pbi->common.prev_mi)
+ {
+ MODE_INFO* tmp = pbi->common.prev_mi;
+ int row, col;
+ pbi->common.prev_mi = pbi->common.mi;
+ pbi->common.mi = tmp;
+
+ /* Propagate the segment_ids to the next frame */
+ for (row = 0; row < pbi->common.mb_rows; ++row)
+ {
+ for (col = 0; col < pbi->common.mb_cols; ++col)
+ {
+ const int i = row*pbi->common.mode_info_stride + col;
+ pbi->common.mi[i].mbmi.segment_id =
+ pbi->common.prev_mi[i].mbmi.segment_id;
+ }
+ }
+ }
+#endif
+
+ if (cm->show_frame)
+ cm->current_video_frame++;
+
+ pbi->ready_for_new_data = 0;
+ pbi->last_time_stamp = time_stamp;
+
+decode_exit:
+#if HAVE_NEON
+#if CONFIG_RUNTIME_CPU_DETECT
+ if (cm->cpu_caps & HAS_NEON)
+#endif
+ {
+ vp8_pop_neon(dx_store_reg);
+ }
+#endif
+
+ pbi->common.error.setjmp = 0;
+ pbi->num_fragments = 0;
+ return retcode;
+}
+int vp8dx_get_raw_frame(VP8D_COMP *pbi, YV12_BUFFER_CONFIG *sd, int64_t *time_stamp, int64_t *time_end_stamp, vp8_ppflags_t *flags)
+{
+ int ret = -1;
+
+ if (pbi->ready_for_new_data == 1)
+ return ret;
+
+ /* ie no raw frame to show!!! */
+ if (pbi->common.show_frame == 0)
+ return ret;
+
+ pbi->ready_for_new_data = 1;
+ *time_stamp = pbi->last_time_stamp;
+ *time_end_stamp = 0;
+
+ sd->clrtype = pbi->common.clr_type;
+#if CONFIG_POSTPROC
+ ret = vp8_post_proc_frame(&pbi->common, sd, flags);
+#else
+
+ if (pbi->common.frame_to_show)
+ {
+ *sd = *pbi->common.frame_to_show;
+ sd->y_width = pbi->common.Width;
+ sd->y_height = pbi->common.Height;
+ sd->uv_height = pbi->common.Height / 2;
+ ret = 0;
+ }
+ else
+ {
+ ret = -1;
+ }
+
+#endif /*!CONFIG_POSTPROC*/
+ vp8_clear_system_state();
+ return ret;
+}
+
+
+/* This function as written isn't decoder specific, but the encoder has
+ * much faster ways of computing this, so it's ok for it to live in a
+ * decode specific file.
+ */
+int vp8dx_references_buffer( VP8_COMMON *oci, int ref_frame )
+{
+ const MODE_INFO *mi = oci->mi;
+ int mb_row, mb_col;
+
+ for (mb_row = 0; mb_row < oci->mb_rows; mb_row++)
+ {
+ for (mb_col = 0; mb_col < oci->mb_cols; mb_col++,mi++)
+ {
+ if( mi->mbmi.ref_frame == ref_frame)
+ return 1;
+ }
+ mi++;
+ }
+ return 0;
+
+}
diff --git a/libvpx/vp8/decoder/onyxd_int.h b/libvpx/vp8/decoder/onyxd_int.h
new file mode 100644
index 0000000..0063beb
--- /dev/null
+++ b/libvpx/vp8/decoder/onyxd_int.h
@@ -0,0 +1,124 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_VP8D_INT_H
+#define __INC_VP8D_INT_H
+#include "vpx_config.h"
+#include "vp8/common/onyxd.h"
+#include "treereader.h"
+#include "vp8/common/onyxc_int.h"
+#include "vp8/common/threading.h"
+
+#if CONFIG_ERROR_CONCEALMENT
+#include "ec_types.h"
+#endif
+
+typedef struct
+{
+ int ithread;
+ void *ptr1;
+ void *ptr2;
+} DECODETHREAD_DATA;
+
+typedef struct
+{
+ MACROBLOCKD mbd;
+} MB_ROW_DEC;
+
+typedef struct VP8D_COMP
+{
+ DECLARE_ALIGNED(16, MACROBLOCKD, mb);
+
+ YV12_BUFFER_CONFIG *dec_fb_ref[NUM_YV12_BUFFERS];
+
+ DECLARE_ALIGNED(16, VP8_COMMON, common);
+
+ /* the last partition will be used for the modes/mvs */
+ vp8_reader mbc[MAX_PARTITIONS];
+
+ VP8D_CONFIG oxcf;
+
+
+ const unsigned char *fragments[MAX_PARTITIONS];
+ unsigned int fragment_sizes[MAX_PARTITIONS];
+ unsigned int num_fragments;
+
+#if CONFIG_MULTITHREAD
+ /* variable for threading */
+
+ volatile int b_multithreaded_rd;
+ int max_threads;
+ int current_mb_col_main;
+ unsigned int decoding_thread_count;
+ int allocated_decoding_thread_count;
+
+ int mt_baseline_filter_level[MAX_MB_SEGMENTS];
+ int sync_range;
+ int *mt_current_mb_col; /* Each row remembers its already decoded column. */
+
+ unsigned char **mt_yabove_row; /* mb_rows x width */
+ unsigned char **mt_uabove_row;
+ unsigned char **mt_vabove_row;
+ unsigned char **mt_yleft_col; /* mb_rows x 16 */
+ unsigned char **mt_uleft_col; /* mb_rows x 8 */
+ unsigned char **mt_vleft_col; /* mb_rows x 8 */
+
+ MB_ROW_DEC *mb_row_di;
+ DECODETHREAD_DATA *de_thread_data;
+
+ pthread_t *h_decoding_thread;
+ sem_t *h_event_start_decoding;
+ sem_t h_event_end_decoding;
+ /* end of threading data */
+#endif
+
+ int64_t last_time_stamp;
+ int ready_for_new_data;
+
+ vp8_prob prob_intra;
+ vp8_prob prob_last;
+ vp8_prob prob_gf;
+ vp8_prob prob_skip_false;
+
+#if CONFIG_ERROR_CONCEALMENT
+ MB_OVERLAP *overlaps;
+ /* the mb num from which modes and mvs (first partition) are corrupt */
+ unsigned int mvs_corrupt_from_mb;
+#endif
+ int ec_enabled;
+ int ec_active;
+ int input_fragments;
+ int decoded_key_frame;
+ int independent_partitions;
+ int frame_corrupt_residual;
+
+} VP8D_COMP;
+
+int vp8_decode_frame(VP8D_COMP *cpi);
+
+#if CONFIG_DEBUG
+#define CHECK_MEM_ERROR(lval,expr) do {\
+ lval = (expr); \
+ if(!lval) \
+ vpx_internal_error(&pbi->common.error, VPX_CODEC_MEM_ERROR,\
+ "Failed to allocate "#lval" at %s:%d", \
+ __FILE__,__LINE__);\
+ } while(0)
+#else
+#define CHECK_MEM_ERROR(lval,expr) do {\
+ lval = (expr); \
+ if(!lval) \
+ vpx_internal_error(&pbi->common.error, VPX_CODEC_MEM_ERROR,\
+ "Failed to allocate "#lval);\
+ } while(0)
+#endif
+
+#endif
diff --git a/libvpx/vp8/decoder/threading.c b/libvpx/vp8/decoder/threading.c
new file mode 100644
index 0000000..88c06be
--- /dev/null
+++ b/libvpx/vp8/decoder/threading.c
@@ -0,0 +1,918 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#if !defined(WIN32) && CONFIG_OS_SUPPORT == 1
+# include <unistd.h>
+#endif
+#include "onyxd_int.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/threading.h"
+
+#include "vp8/common/loopfilter.h"
+#include "vp8/common/extend.h"
+#include "vpx_ports/vpx_timer.h"
+#include "detokenize.h"
+#include "vp8/common/reconintra4x4.h"
+#include "vp8/common/reconinter.h"
+#include "vp8/common/setupintrarecon.h"
+#if CONFIG_ERROR_CONCEALMENT
+#include "error_concealment.h"
+#endif
+
+#define CALLOC_ARRAY(p, n) CHECK_MEM_ERROR((p), vpx_calloc(sizeof(*(p)), (n)))
+#define CALLOC_ARRAY_ALIGNED(p, n, algn) do { \
+ CHECK_MEM_ERROR((p), vpx_memalign((algn), sizeof(*(p)) * (n))); \
+ memset((p), 0, (n) * sizeof(*(p))); \
+} while (0)
+
+
+extern void vp8_mb_init_dequantizer(VP8D_COMP *pbi, MACROBLOCKD *xd);
+
+static void setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC *mbrd, int count)
+{
+ VP8_COMMON *const pc = & pbi->common;
+ int i;
+
+ for (i = 0; i < count; i++)
+ {
+ MACROBLOCKD *mbd = &mbrd[i].mbd;
+ mbd->subpixel_predict = xd->subpixel_predict;
+ mbd->subpixel_predict8x4 = xd->subpixel_predict8x4;
+ mbd->subpixel_predict8x8 = xd->subpixel_predict8x8;
+ mbd->subpixel_predict16x16 = xd->subpixel_predict16x16;
+
+ mbd->mode_info_context = pc->mi + pc->mode_info_stride * (i + 1);
+ mbd->mode_info_stride = pc->mode_info_stride;
+
+ mbd->frame_type = pc->frame_type;
+ mbd->pre = xd->pre;
+ mbd->dst = xd->dst;
+
+ mbd->segmentation_enabled = xd->segmentation_enabled;
+ mbd->mb_segement_abs_delta = xd->mb_segement_abs_delta;
+ vpx_memcpy(mbd->segment_feature_data, xd->segment_feature_data, sizeof(xd->segment_feature_data));
+
+ /*signed char ref_lf_deltas[MAX_REF_LF_DELTAS];*/
+ vpx_memcpy(mbd->ref_lf_deltas, xd->ref_lf_deltas, sizeof(xd->ref_lf_deltas));
+ /*signed char mode_lf_deltas[MAX_MODE_LF_DELTAS];*/
+ vpx_memcpy(mbd->mode_lf_deltas, xd->mode_lf_deltas, sizeof(xd->mode_lf_deltas));
+ /*unsigned char mode_ref_lf_delta_enabled;
+ unsigned char mode_ref_lf_delta_update;*/
+ mbd->mode_ref_lf_delta_enabled = xd->mode_ref_lf_delta_enabled;
+ mbd->mode_ref_lf_delta_update = xd->mode_ref_lf_delta_update;
+
+ mbd->current_bc = &pbi->mbc[0];
+
+ vpx_memcpy(mbd->dequant_y1_dc, xd->dequant_y1_dc, sizeof(xd->dequant_y1_dc));
+ vpx_memcpy(mbd->dequant_y1, xd->dequant_y1, sizeof(xd->dequant_y1));
+ vpx_memcpy(mbd->dequant_y2, xd->dequant_y2, sizeof(xd->dequant_y2));
+ vpx_memcpy(mbd->dequant_uv, xd->dequant_uv, sizeof(xd->dequant_uv));
+
+ mbd->fullpixel_mask = 0xffffffff;
+
+ if (pc->full_pixel)
+ mbd->fullpixel_mask = 0xfffffff8;
+
+ }
+
+ for (i = 0; i < pc->mb_rows; i++)
+ pbi->mt_current_mb_col[i] = -1;
+}
+
+static void mt_decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
+ unsigned int mb_idx)
+{
+ MB_PREDICTION_MODE mode;
+ int i;
+#if CONFIG_ERROR_CONCEALMENT
+ int corruption_detected = 0;
+#endif
+
+ if (xd->mode_info_context->mbmi.mb_skip_coeff)
+ {
+ vp8_reset_mb_tokens_context(xd);
+ }
+ else if (!vp8dx_bool_error(xd->current_bc))
+ {
+ int eobtotal;
+ eobtotal = vp8_decode_mb_tokens(pbi, xd);
+
+ /* Special case: Force the loopfilter to skip when eobtotal is zero */
+ xd->mode_info_context->mbmi.mb_skip_coeff = (eobtotal==0);
+ }
+
+ mode = xd->mode_info_context->mbmi.mode;
+
+ if (xd->segmentation_enabled)
+ vp8_mb_init_dequantizer(pbi, xd);
+
+
+#if CONFIG_ERROR_CONCEALMENT
+
+ if(pbi->ec_active)
+ {
+ int throw_residual;
+ /* When we have independent partitions we can apply residual even
+ * though other partitions within the frame are corrupt.
+ */
+ throw_residual = (!pbi->independent_partitions &&
+ pbi->frame_corrupt_residual);
+ throw_residual = (throw_residual || vp8dx_bool_error(xd->current_bc));
+
+ if ((mb_idx >= pbi->mvs_corrupt_from_mb || throw_residual))
+ {
+ /* MB with corrupt residuals or corrupt mode/motion vectors.
+ * Better to use the predictor as reconstruction.
+ */
+ pbi->frame_corrupt_residual = 1;
+ vpx_memset(xd->qcoeff, 0, sizeof(xd->qcoeff));
+ vp8_conceal_corrupt_mb(xd);
+
+
+ corruption_detected = 1;
+
+ /* force idct to be skipped for B_PRED and use the
+ * prediction only for reconstruction
+ * */
+ vpx_memset(xd->eobs, 0, 25);
+ }
+ }
+#endif
+
+ /* do prediction */
+ if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
+ {
+ vp8_build_intra_predictors_mbuv_s(xd,
+ xd->recon_above[1],
+ xd->recon_above[2],
+ xd->recon_left[1],
+ xd->recon_left[2],
+ xd->recon_left_stride[1],
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride);
+
+ if (mode != B_PRED)
+ {
+ vp8_build_intra_predictors_mby_s(xd,
+ xd->recon_above[0],
+ xd->recon_left[0],
+ xd->recon_left_stride[0],
+ xd->dst.y_buffer,
+ xd->dst.y_stride);
+ }
+ else
+ {
+ short *DQC = xd->dequant_y1;
+ int dst_stride = xd->dst.y_stride;
+
+ /* clear out residual eob info */
+ if(xd->mode_info_context->mbmi.mb_skip_coeff)
+ vpx_memset(xd->eobs, 0, 25);
+
+ intra_prediction_down_copy(xd, xd->recon_above[0] + 16);
+
+ for (i = 0; i < 16; i++)
+ {
+ BLOCKD *b = &xd->block[i];
+ unsigned char *dst = xd->dst.y_buffer + b->offset;
+ B_PREDICTION_MODE b_mode =
+ xd->mode_info_context->bmi[i].as_mode;
+ unsigned char *Above;
+ unsigned char *yleft;
+ int left_stride;
+ unsigned char top_left;
+
+ /*Caution: For some b_mode, it needs 8 pixels (4 above + 4 above-right).*/
+ if (i < 4 && pbi->common.filter_level)
+ Above = xd->recon_above[0] + b->offset;
+ else
+ Above = dst - dst_stride;
+
+ if (i%4==0 && pbi->common.filter_level)
+ {
+ yleft = xd->recon_left[0] + i;
+ left_stride = 1;
+ }
+ else
+ {
+ yleft = dst - 1;
+ left_stride = dst_stride;
+ }
+
+ if ((i==4 || i==8 || i==12) && pbi->common.filter_level)
+ top_left = *(xd->recon_left[0] + i - 1);
+ else
+ top_left = Above[-1];
+
+ vp8_intra4x4_predict(Above, yleft, left_stride,
+ b_mode, dst, dst_stride, top_left);
+
+ if (xd->eobs[i] )
+ {
+ if (xd->eobs[i] > 1)
+ {
+ vp8_dequant_idct_add(b->qcoeff, DQC, dst, dst_stride);
+ }
+ else
+ {
+ vp8_dc_only_idct_add(b->qcoeff[0] * DQC[0],
+ dst, dst_stride, dst, dst_stride);
+ ((int *)b->qcoeff)[0] = 0;
+ }
+ }
+ }
+ }
+ }
+ else
+ {
+ vp8_build_inter_predictors_mb(xd);
+ }
+
+
+#if CONFIG_ERROR_CONCEALMENT
+ if (corruption_detected)
+ {
+ return;
+ }
+#endif
+
+ if(!xd->mode_info_context->mbmi.mb_skip_coeff)
+ {
+ /* dequantization and idct */
+ if (mode != B_PRED)
+ {
+ short *DQC = xd->dequant_y1;
+
+ if (mode != SPLITMV)
+ {
+ BLOCKD *b = &xd->block[24];
+
+ /* do 2nd order transform on the dc block */
+ if (xd->eobs[24] > 1)
+ {
+ vp8_dequantize_b(b, xd->dequant_y2);
+
+ vp8_short_inv_walsh4x4(&b->dqcoeff[0],
+ xd->qcoeff);
+ ((int *)b->qcoeff)[0] = 0;
+ ((int *)b->qcoeff)[1] = 0;
+ ((int *)b->qcoeff)[2] = 0;
+ ((int *)b->qcoeff)[3] = 0;
+ ((int *)b->qcoeff)[4] = 0;
+ ((int *)b->qcoeff)[5] = 0;
+ ((int *)b->qcoeff)[6] = 0;
+ ((int *)b->qcoeff)[7] = 0;
+ }
+ else
+ {
+ b->dqcoeff[0] = b->qcoeff[0] * xd->dequant_y2[0];
+ vp8_short_inv_walsh4x4_1(&b->dqcoeff[0],
+ xd->qcoeff);
+ ((int *)b->qcoeff)[0] = 0;
+ }
+
+ /* override the dc dequant constant in order to preserve the
+ * dc components
+ */
+ DQC = xd->dequant_y1_dc;
+ }
+
+ vp8_dequant_idct_add_y_block
+ (xd->qcoeff, DQC,
+ xd->dst.y_buffer,
+ xd->dst.y_stride, xd->eobs);
+ }
+
+ vp8_dequant_idct_add_uv_block
+ (xd->qcoeff+16*16, xd->dequant_uv,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride, xd->eobs+16);
+ }
+}
+
+static void mt_decode_mb_rows(VP8D_COMP *pbi, MACROBLOCKD *xd, int start_mb_row)
+{
+ volatile const int *last_row_current_mb_col;
+ volatile int *current_mb_col;
+ int mb_row;
+ VP8_COMMON *pc = &pbi->common;
+ const int nsync = pbi->sync_range;
+ const int first_row_no_sync_above = pc->mb_cols + nsync;
+ int num_part = 1 << pbi->common.multi_token_partition;
+ int last_mb_row = start_mb_row;
+
+ YV12_BUFFER_CONFIG *yv12_fb_new = pbi->dec_fb_ref[INTRA_FRAME];
+ YV12_BUFFER_CONFIG *yv12_fb_lst = pbi->dec_fb_ref[LAST_FRAME];
+
+ int recon_y_stride = yv12_fb_new->y_stride;
+ int recon_uv_stride = yv12_fb_new->uv_stride;
+
+ unsigned char *ref_buffer[MAX_REF_FRAMES][3];
+ unsigned char *dst_buffer[3];
+ int i;
+ int ref_fb_corrupted[MAX_REF_FRAMES];
+
+ ref_fb_corrupted[INTRA_FRAME] = 0;
+
+ for(i = 1; i < MAX_REF_FRAMES; i++)
+ {
+ YV12_BUFFER_CONFIG *this_fb = pbi->dec_fb_ref[i];
+
+ ref_buffer[i][0] = this_fb->y_buffer;
+ ref_buffer[i][1] = this_fb->u_buffer;
+ ref_buffer[i][2] = this_fb->v_buffer;
+
+ ref_fb_corrupted[i] = this_fb->corrupted;
+ }
+
+ dst_buffer[0] = yv12_fb_new->y_buffer;
+ dst_buffer[1] = yv12_fb_new->u_buffer;
+ dst_buffer[2] = yv12_fb_new->v_buffer;
+
+ xd->up_available = (start_mb_row != 0);
+
+ for (mb_row = start_mb_row; mb_row < pc->mb_rows; mb_row += (pbi->decoding_thread_count + 1))
+ {
+ int i;
+ int recon_yoffset, recon_uvoffset;
+ int mb_col;
+ int filter_level;
+ loop_filter_info_n *lfi_n = &pc->lf_info;
+
+ /* save last row processed by this thread */
+ last_mb_row = mb_row;
+ /* select bool coder for current partition */
+ xd->current_bc = &pbi->mbc[mb_row%num_part];
+
+ if (mb_row > 0)
+ last_row_current_mb_col = &pbi->mt_current_mb_col[mb_row -1];
+ else
+ last_row_current_mb_col = &first_row_no_sync_above;
+
+ current_mb_col = &pbi->mt_current_mb_col[mb_row];
+
+ recon_yoffset = mb_row * recon_y_stride * 16;
+ recon_uvoffset = mb_row * recon_uv_stride * 8;
+
+ /* reset contexts */
+ xd->above_context = pc->above_context;
+ vpx_memset(xd->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ xd->left_available = 0;
+
+ xd->mb_to_top_edge = -((mb_row * 16)) << 3;
+ xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3;
+
+ if (pbi->common.filter_level)
+ {
+ xd->recon_above[0] = pbi->mt_yabove_row[mb_row] + 0*16 +32;
+ xd->recon_above[1] = pbi->mt_uabove_row[mb_row] + 0*8 +16;
+ xd->recon_above[2] = pbi->mt_vabove_row[mb_row] + 0*8 +16;
+
+ xd->recon_left[0] = pbi->mt_yleft_col[mb_row];
+ xd->recon_left[1] = pbi->mt_uleft_col[mb_row];
+ xd->recon_left[2] = pbi->mt_vleft_col[mb_row];
+
+ /* TODO: move to outside row loop */
+ xd->recon_left_stride[0] = 1;
+ xd->recon_left_stride[1] = 1;
+ }
+ else
+ {
+ xd->recon_above[0] = dst_buffer[0] + recon_yoffset;
+ xd->recon_above[1] = dst_buffer[1] + recon_uvoffset;
+ xd->recon_above[2] = dst_buffer[2] + recon_uvoffset;
+
+ xd->recon_left[0] = xd->recon_above[0] - 1;
+ xd->recon_left[1] = xd->recon_above[1] - 1;
+ xd->recon_left[2] = xd->recon_above[2] - 1;
+
+ xd->recon_above[0] -= xd->dst.y_stride;
+ xd->recon_above[1] -= xd->dst.uv_stride;
+ xd->recon_above[2] -= xd->dst.uv_stride;
+
+ /* TODO: move to outside row loop */
+ xd->recon_left_stride[0] = xd->dst.y_stride;
+ xd->recon_left_stride[1] = xd->dst.uv_stride;
+
+ setup_intra_recon_left(xd->recon_left[0], xd->recon_left[1],
+ xd->recon_left[2], xd->dst.y_stride,
+ xd->dst.uv_stride);
+ }
+
+ for (mb_col = 0; mb_col < pc->mb_cols; mb_col++)
+ {
+ *current_mb_col = mb_col - 1;
+
+ if ((mb_col & (nsync - 1)) == 0)
+ {
+ while (mb_col > (*last_row_current_mb_col - nsync))
+ {
+ x86_pause_hint();
+ thread_sleep(0);
+ }
+ }
+
+ /* Distance of MB to the various image edges.
+ * These are specified to 8th pel as they are always
+ * compared to values that are in 1/8th pel units.
+ */
+ xd->mb_to_left_edge = -((mb_col * 16) << 3);
+ xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3;
+
+ #if CONFIG_ERROR_CONCEALMENT
+ {
+ int corrupt_residual =
+ (!pbi->independent_partitions &&
+ pbi->frame_corrupt_residual) ||
+ vp8dx_bool_error(xd->current_bc);
+ if (pbi->ec_active &&
+ (xd->mode_info_context->mbmi.ref_frame ==
+ INTRA_FRAME) &&
+ corrupt_residual)
+ {
+ /* We have an intra block with corrupt
+ * coefficients, better to conceal with an inter
+ * block.
+ * Interpolate MVs from neighboring MBs
+ *
+ * Note that for the first mb with corrupt
+ * residual in a frame, we might not discover
+ * that before decoding the residual. That
+ * happens after this check, and therefore no
+ * inter concealment will be done.
+ */
+ vp8_interpolate_motion(xd,
+ mb_row, mb_col,
+ pc->mb_rows, pc->mb_cols,
+ pc->mode_info_stride);
+ }
+ }
+ #endif
+
+
+ xd->dst.y_buffer = dst_buffer[0] + recon_yoffset;
+ xd->dst.u_buffer = dst_buffer[1] + recon_uvoffset;
+ xd->dst.v_buffer = dst_buffer[2] + recon_uvoffset;
+
+ xd->pre.y_buffer = ref_buffer[xd->mode_info_context->mbmi.ref_frame][0] + recon_yoffset;
+ xd->pre.u_buffer = ref_buffer[xd->mode_info_context->mbmi.ref_frame][1] + recon_uvoffset;
+ xd->pre.v_buffer = ref_buffer[xd->mode_info_context->mbmi.ref_frame][2] + recon_uvoffset;
+
+ /* propagate errors from reference frames */
+ xd->corrupted |= ref_fb_corrupted[xd->mode_info_context->mbmi.ref_frame];
+
+ mt_decode_macroblock(pbi, xd, 0);
+
+ xd->left_available = 1;
+
+ /* check if the boolean decoder has suffered an error */
+ xd->corrupted |= vp8dx_bool_error(xd->current_bc);
+
+ xd->recon_above[0] += 16;
+ xd->recon_above[1] += 8;
+ xd->recon_above[2] += 8;
+
+ if (!pbi->common.filter_level)
+ {
+ xd->recon_left[0] += 16;
+ xd->recon_left[1] += 8;
+ xd->recon_left[2] += 8;
+ }
+
+ if (pbi->common.filter_level)
+ {
+ int skip_lf = (xd->mode_info_context->mbmi.mode != B_PRED &&
+ xd->mode_info_context->mbmi.mode != SPLITMV &&
+ xd->mode_info_context->mbmi.mb_skip_coeff);
+
+ const int mode_index = lfi_n->mode_lf_lut[xd->mode_info_context->mbmi.mode];
+ const int seg = xd->mode_info_context->mbmi.segment_id;
+ const int ref_frame = xd->mode_info_context->mbmi.ref_frame;
+
+ filter_level = lfi_n->lvl[seg][ref_frame][mode_index];
+
+ if( mb_row != pc->mb_rows-1 )
+ {
+ /* Save decoded MB last row data for next-row decoding */
+ vpx_memcpy((pbi->mt_yabove_row[mb_row + 1] + 32 + mb_col*16), (xd->dst.y_buffer + 15 * recon_y_stride), 16);
+ vpx_memcpy((pbi->mt_uabove_row[mb_row + 1] + 16 + mb_col*8), (xd->dst.u_buffer + 7 * recon_uv_stride), 8);
+ vpx_memcpy((pbi->mt_vabove_row[mb_row + 1] + 16 + mb_col*8), (xd->dst.v_buffer + 7 * recon_uv_stride), 8);
+ }
+
+ /* save left_col for next MB decoding */
+ if(mb_col != pc->mb_cols-1)
+ {
+ MODE_INFO *next = xd->mode_info_context +1;
+
+ if (next->mbmi.ref_frame == INTRA_FRAME)
+ {
+ for (i = 0; i < 16; i++)
+ pbi->mt_yleft_col[mb_row][i] = xd->dst.y_buffer [i* recon_y_stride + 15];
+ for (i = 0; i < 8; i++)
+ {
+ pbi->mt_uleft_col[mb_row][i] = xd->dst.u_buffer [i* recon_uv_stride + 7];
+ pbi->mt_vleft_col[mb_row][i] = xd->dst.v_buffer [i* recon_uv_stride + 7];
+ }
+ }
+ }
+
+ /* loopfilter on this macroblock. */
+ if (filter_level)
+ {
+ if(pc->filter_type == NORMAL_LOOPFILTER)
+ {
+ loop_filter_info lfi;
+ FRAME_TYPE frame_type = pc->frame_type;
+ const int hev_index = lfi_n->hev_thr_lut[frame_type][filter_level];
+ lfi.mblim = lfi_n->mblim[filter_level];
+ lfi.blim = lfi_n->blim[filter_level];
+ lfi.lim = lfi_n->lim[filter_level];
+ lfi.hev_thr = lfi_n->hev_thr[hev_index];
+
+ if (mb_col > 0)
+ vp8_loop_filter_mbv
+ (xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bv
+ (xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi);
+
+ /* don't apply across umv border */
+ if (mb_row > 0)
+ vp8_loop_filter_mbh
+ (xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi);
+
+ if (!skip_lf)
+ vp8_loop_filter_bh
+ (xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi);
+ }
+ else
+ {
+ if (mb_col > 0)
+ vp8_loop_filter_simple_mbv
+ (xd->dst.y_buffer, recon_y_stride, lfi_n->mblim[filter_level]);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bv
+ (xd->dst.y_buffer, recon_y_stride, lfi_n->blim[filter_level]);
+
+ /* don't apply across umv border */
+ if (mb_row > 0)
+ vp8_loop_filter_simple_mbh
+ (xd->dst.y_buffer, recon_y_stride, lfi_n->mblim[filter_level]);
+
+ if (!skip_lf)
+ vp8_loop_filter_simple_bh
+ (xd->dst.y_buffer, recon_y_stride, lfi_n->blim[filter_level]);
+ }
+ }
+
+ }
+
+ recon_yoffset += 16;
+ recon_uvoffset += 8;
+
+ ++xd->mode_info_context; /* next mb */
+
+ xd->above_context++;
+ }
+
+ /* adjust to the next row of mbs */
+ if (pbi->common.filter_level)
+ {
+ if(mb_row != pc->mb_rows-1)
+ {
+ int lasty = yv12_fb_lst->y_width + VP8BORDERINPIXELS;
+ int lastuv = (yv12_fb_lst->y_width>>1) + (VP8BORDERINPIXELS>>1);
+
+ for (i = 0; i < 4; i++)
+ {
+ pbi->mt_yabove_row[mb_row +1][lasty + i] = pbi->mt_yabove_row[mb_row +1][lasty -1];
+ pbi->mt_uabove_row[mb_row +1][lastuv + i] = pbi->mt_uabove_row[mb_row +1][lastuv -1];
+ pbi->mt_vabove_row[mb_row +1][lastuv + i] = pbi->mt_vabove_row[mb_row +1][lastuv -1];
+ }
+ }
+ }
+ else
+ vp8_extend_mb_row(yv12_fb_new, xd->dst.y_buffer + 16,
+ xd->dst.u_buffer + 8, xd->dst.v_buffer + 8);
+
+ /* last MB of row is ready just after extension is done */
+ *current_mb_col = mb_col + nsync;
+
+ ++xd->mode_info_context; /* skip prediction column */
+ xd->up_available = 1;
+
+ /* since we have multithread */
+ xd->mode_info_context += xd->mode_info_stride * pbi->decoding_thread_count;
+ }
+
+ /* signal end of frame decoding if this thread processed the last mb_row */
+ if (last_mb_row == (pc->mb_rows - 1))
+ sem_post(&pbi->h_event_end_decoding);
+
+}
+
+
+static THREAD_FUNCTION thread_decoding_proc(void *p_data)
+{
+ int ithread = ((DECODETHREAD_DATA *)p_data)->ithread;
+ VP8D_COMP *pbi = (VP8D_COMP *)(((DECODETHREAD_DATA *)p_data)->ptr1);
+ MB_ROW_DEC *mbrd = (MB_ROW_DEC *)(((DECODETHREAD_DATA *)p_data)->ptr2);
+ ENTROPY_CONTEXT_PLANES mb_row_left_context;
+
+ while (1)
+ {
+ if (pbi->b_multithreaded_rd == 0)
+ break;
+
+ if (sem_wait(&pbi->h_event_start_decoding[ithread]) == 0)
+ {
+ if (pbi->b_multithreaded_rd == 0)
+ break;
+ else
+ {
+ MACROBLOCKD *xd = &mbrd->mbd;
+ xd->left_context = &mb_row_left_context;
+
+ mt_decode_mb_rows(pbi, xd, ithread+1);
+ }
+ }
+ }
+
+ return 0 ;
+}
+
+
+void vp8_decoder_create_threads(VP8D_COMP *pbi)
+{
+ int core_count = 0;
+ unsigned int ithread;
+
+ pbi->b_multithreaded_rd = 0;
+ pbi->allocated_decoding_thread_count = 0;
+
+ /* limit decoding threads to the max number of token partitions */
+ core_count = (pbi->max_threads > 8) ? 8 : pbi->max_threads;
+
+ /* limit decoding threads to the available cores */
+ if (core_count > pbi->common.processor_core_count)
+ core_count = pbi->common.processor_core_count;
+
+ if (core_count > 1)
+ {
+ pbi->b_multithreaded_rd = 1;
+ pbi->decoding_thread_count = core_count - 1;
+
+ CALLOC_ARRAY(pbi->h_decoding_thread, pbi->decoding_thread_count);
+ CALLOC_ARRAY(pbi->h_event_start_decoding, pbi->decoding_thread_count);
+ CALLOC_ARRAY_ALIGNED(pbi->mb_row_di, pbi->decoding_thread_count, 32);
+ CALLOC_ARRAY(pbi->de_thread_data, pbi->decoding_thread_count);
+
+ for (ithread = 0; ithread < pbi->decoding_thread_count; ithread++)
+ {
+ sem_init(&pbi->h_event_start_decoding[ithread], 0, 0);
+
+ vp8_setup_block_dptrs(&pbi->mb_row_di[ithread].mbd);
+
+ pbi->de_thread_data[ithread].ithread = ithread;
+ pbi->de_thread_data[ithread].ptr1 = (void *)pbi;
+ pbi->de_thread_data[ithread].ptr2 = (void *) &pbi->mb_row_di[ithread];
+
+ pthread_create(&pbi->h_decoding_thread[ithread], 0, thread_decoding_proc, (&pbi->de_thread_data[ithread]));
+ }
+
+ sem_init(&pbi->h_event_end_decoding, 0, 0);
+
+ pbi->allocated_decoding_thread_count = pbi->decoding_thread_count;
+ }
+}
+
+
+void vp8mt_de_alloc_temp_buffers(VP8D_COMP *pbi, int mb_rows)
+{
+ int i;
+
+ if (pbi->b_multithreaded_rd)
+ {
+ vpx_free(pbi->mt_current_mb_col);
+ pbi->mt_current_mb_col = NULL ;
+
+ /* Free above_row buffers. */
+ if (pbi->mt_yabove_row)
+ {
+ for (i=0; i< mb_rows; i++)
+ {
+ vpx_free(pbi->mt_yabove_row[i]);
+ pbi->mt_yabove_row[i] = NULL ;
+ }
+ vpx_free(pbi->mt_yabove_row);
+ pbi->mt_yabove_row = NULL ;
+ }
+
+ if (pbi->mt_uabove_row)
+ {
+ for (i=0; i< mb_rows; i++)
+ {
+ vpx_free(pbi->mt_uabove_row[i]);
+ pbi->mt_uabove_row[i] = NULL ;
+ }
+ vpx_free(pbi->mt_uabove_row);
+ pbi->mt_uabove_row = NULL ;
+ }
+
+ if (pbi->mt_vabove_row)
+ {
+ for (i=0; i< mb_rows; i++)
+ {
+ vpx_free(pbi->mt_vabove_row[i]);
+ pbi->mt_vabove_row[i] = NULL ;
+ }
+ vpx_free(pbi->mt_vabove_row);
+ pbi->mt_vabove_row = NULL ;
+ }
+
+ /* Free left_col buffers. */
+ if (pbi->mt_yleft_col)
+ {
+ for (i=0; i< mb_rows; i++)
+ {
+ vpx_free(pbi->mt_yleft_col[i]);
+ pbi->mt_yleft_col[i] = NULL ;
+ }
+ vpx_free(pbi->mt_yleft_col);
+ pbi->mt_yleft_col = NULL ;
+ }
+
+ if (pbi->mt_uleft_col)
+ {
+ for (i=0; i< mb_rows; i++)
+ {
+ vpx_free(pbi->mt_uleft_col[i]);
+ pbi->mt_uleft_col[i] = NULL ;
+ }
+ vpx_free(pbi->mt_uleft_col);
+ pbi->mt_uleft_col = NULL ;
+ }
+
+ if (pbi->mt_vleft_col)
+ {
+ for (i=0; i< mb_rows; i++)
+ {
+ vpx_free(pbi->mt_vleft_col[i]);
+ pbi->mt_vleft_col[i] = NULL ;
+ }
+ vpx_free(pbi->mt_vleft_col);
+ pbi->mt_vleft_col = NULL ;
+ }
+ }
+}
+
+
+void vp8mt_alloc_temp_buffers(VP8D_COMP *pbi, int width, int prev_mb_rows)
+{
+ VP8_COMMON *const pc = & pbi->common;
+ int i;
+ int uv_width;
+
+ if (pbi->b_multithreaded_rd)
+ {
+ vp8mt_de_alloc_temp_buffers(pbi, prev_mb_rows);
+
+ /* our internal buffers are always multiples of 16 */
+ if ((width & 0xf) != 0)
+ width += 16 - (width & 0xf);
+
+ if (width < 640) pbi->sync_range = 1;
+ else if (width <= 1280) pbi->sync_range = 8;
+ else if (width <= 2560) pbi->sync_range =16;
+ else pbi->sync_range = 32;
+
+ uv_width = width >>1;
+
+ /* Allocate an int for each mb row. */
+ CALLOC_ARRAY(pbi->mt_current_mb_col, pc->mb_rows);
+
+ /* Allocate memory for above_row buffers. */
+ CALLOC_ARRAY(pbi->mt_yabove_row, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
+ CHECK_MEM_ERROR(pbi->mt_yabove_row[i], vpx_memalign(16,sizeof(unsigned char) * (width + (VP8BORDERINPIXELS<<1))));
+
+ CALLOC_ARRAY(pbi->mt_uabove_row, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
+ CHECK_MEM_ERROR(pbi->mt_uabove_row[i], vpx_memalign(16,sizeof(unsigned char) * (uv_width + VP8BORDERINPIXELS)));
+
+ CALLOC_ARRAY(pbi->mt_vabove_row, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
+ CHECK_MEM_ERROR(pbi->mt_vabove_row[i], vpx_memalign(16,sizeof(unsigned char) * (uv_width + VP8BORDERINPIXELS)));
+
+ /* Allocate memory for left_col buffers. */
+ CALLOC_ARRAY(pbi->mt_yleft_col, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
+ CHECK_MEM_ERROR(pbi->mt_yleft_col[i], vpx_calloc(sizeof(unsigned char) * 16, 1));
+
+ CALLOC_ARRAY(pbi->mt_uleft_col, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
+ CHECK_MEM_ERROR(pbi->mt_uleft_col[i], vpx_calloc(sizeof(unsigned char) * 8, 1));
+
+ CALLOC_ARRAY(pbi->mt_vleft_col, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
+ CHECK_MEM_ERROR(pbi->mt_vleft_col[i], vpx_calloc(sizeof(unsigned char) * 8, 1));
+ }
+}
+
+
+void vp8_decoder_remove_threads(VP8D_COMP *pbi)
+{
+ /* shutdown MB Decoding thread; */
+ if (pbi->b_multithreaded_rd)
+ {
+ int i;
+
+ pbi->b_multithreaded_rd = 0;
+
+ /* allow all threads to exit */
+ for (i = 0; i < pbi->allocated_decoding_thread_count; i++)
+ {
+ sem_post(&pbi->h_event_start_decoding[i]);
+ pthread_join(pbi->h_decoding_thread[i], NULL);
+ }
+
+ for (i = 0; i < pbi->allocated_decoding_thread_count; i++)
+ {
+ sem_destroy(&pbi->h_event_start_decoding[i]);
+ }
+
+ sem_destroy(&pbi->h_event_end_decoding);
+
+ vpx_free(pbi->h_decoding_thread);
+ pbi->h_decoding_thread = NULL;
+
+ vpx_free(pbi->h_event_start_decoding);
+ pbi->h_event_start_decoding = NULL;
+
+ vpx_free(pbi->mb_row_di);
+ pbi->mb_row_di = NULL ;
+
+ vpx_free(pbi->de_thread_data);
+ pbi->de_thread_data = NULL;
+ }
+}
+
+void vp8mt_decode_mb_rows( VP8D_COMP *pbi, MACROBLOCKD *xd)
+{
+ VP8_COMMON *pc = &pbi->common;
+ unsigned int i;
+ int j;
+
+ int filter_level = pc->filter_level;
+ YV12_BUFFER_CONFIG *yv12_fb_new = pbi->dec_fb_ref[INTRA_FRAME];
+
+ if (filter_level)
+ {
+ /* Set above_row buffer to 127 for decoding first MB row */
+ vpx_memset(pbi->mt_yabove_row[0] + VP8BORDERINPIXELS-1, 127, yv12_fb_new->y_width + 5);
+ vpx_memset(pbi->mt_uabove_row[0] + (VP8BORDERINPIXELS>>1)-1, 127, (yv12_fb_new->y_width>>1) +5);
+ vpx_memset(pbi->mt_vabove_row[0] + (VP8BORDERINPIXELS>>1)-1, 127, (yv12_fb_new->y_width>>1) +5);
+
+ for (j=1; j<pc->mb_rows; j++)
+ {
+ vpx_memset(pbi->mt_yabove_row[j] + VP8BORDERINPIXELS-1, (unsigned char)129, 1);
+ vpx_memset(pbi->mt_uabove_row[j] + (VP8BORDERINPIXELS>>1)-1, (unsigned char)129, 1);
+ vpx_memset(pbi->mt_vabove_row[j] + (VP8BORDERINPIXELS>>1)-1, (unsigned char)129, 1);
+ }
+
+ /* Set left_col to 129 initially */
+ for (j=0; j<pc->mb_rows; j++)
+ {
+ vpx_memset(pbi->mt_yleft_col[j], (unsigned char)129, 16);
+ vpx_memset(pbi->mt_uleft_col[j], (unsigned char)129, 8);
+ vpx_memset(pbi->mt_vleft_col[j], (unsigned char)129, 8);
+ }
+
+ /* Initialize the loop filter for this frame. */
+ vp8_loop_filter_frame_init(pc, &pbi->mb, filter_level);
+ }
+ else
+ vp8_setup_intra_recon_top_line(yv12_fb_new);
+
+ setup_decoding_thread_data(pbi, xd, pbi->mb_row_di, pbi->decoding_thread_count);
+
+ for (i = 0; i < pbi->decoding_thread_count; i++)
+ sem_post(&pbi->h_event_start_decoding[i]);
+
+ mt_decode_mb_rows(pbi, xd, 0);
+
+ sem_wait(&pbi->h_event_end_decoding); /* add back for each frame */
+}
diff --git a/libvpx/vp8/decoder/treereader.h b/libvpx/vp8/decoder/treereader.h
new file mode 100644
index 0000000..238ff85
--- /dev/null
+++ b/libvpx/vp8/decoder/treereader.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef tree_reader_h
+#define tree_reader_h 1
+
+#include "vp8/common/treecoder.h"
+
+#include "dboolhuff.h"
+
+typedef BOOL_DECODER vp8_reader;
+
+#define vp8_read vp8dx_decode_bool
+#define vp8_read_literal vp8_decode_value
+#define vp8_read_bit( R) vp8_read( R, vp8_prob_half)
+
+
+/* Intent of tree data structure is to make decoding trivial. */
+
+static int vp8_treed_read(
+ vp8_reader *const r, /* !!! must return a 0 or 1 !!! */
+ vp8_tree t,
+ const vp8_prob *const p
+)
+{
+ register vp8_tree_index i = 0;
+
+ while ((i = t[ i + vp8_read(r, p[i>>1])]) > 0) ;
+
+ return -i;
+}
+
+#endif /* tree_reader_h */
diff --git a/libvpx/vp8/encoder/arm/armv5te/boolhuff_armv5te.asm b/libvpx/vp8/encoder/arm/armv5te/boolhuff_armv5te.asm
new file mode 100644
index 0000000..a644a00
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv5te/boolhuff_armv5te.asm
@@ -0,0 +1,310 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_start_encode|
+ EXPORT |vp8_encode_bool|
+ EXPORT |vp8_stop_encode|
+ EXPORT |vp8_encode_value|
+ IMPORT |vp8_validate_buffer_arm|
+
+ INCLUDE asm_enc_offsets.asm
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA |.text|, CODE, READONLY
+
+ ; macro for validating write buffer position
+ ; needs vp8_writer in r0
+ ; start shall not be in r1
+ MACRO
+ VALIDATE_POS $start, $pos
+ push {r0-r3, r12, lr} ; rest of regs are preserved by subroutine call
+ ldr r2, [r0, #vp8_writer_buffer_end]
+ ldr r3, [r0, #vp8_writer_error]
+ mov r1, $pos
+ mov r0, $start
+ bl vp8_validate_buffer_arm
+ pop {r0-r3, r12, lr}
+ MEND
+
+; r0 BOOL_CODER *br
+; r1 unsigned char *source
+; r2 unsigned char *source_end
+|vp8_start_encode| PROC
+ str r2, [r0, #vp8_writer_buffer_end]
+ mov r12, #0
+ mov r3, #255
+ mvn r2, #23
+ str r12, [r0, #vp8_writer_lowvalue]
+ str r3, [r0, #vp8_writer_range]
+ str r2, [r0, #vp8_writer_count]
+ str r12, [r0, #vp8_writer_pos]
+ str r1, [r0, #vp8_writer_buffer]
+ bx lr
+ ENDP
+
+; r0 BOOL_CODER *br
+; r1 int bit
+; r2 int probability
+|vp8_encode_bool| PROC
+ push {r4-r10, lr}
+
+ mov r4, r2
+
+ ldr r2, [r0, #vp8_writer_lowvalue]
+ ldr r5, [r0, #vp8_writer_range]
+ ldr r3, [r0, #vp8_writer_count]
+
+ sub r7, r5, #1 ; range-1
+
+ cmp r1, #0
+ mul r6, r4, r7 ; ((range-1) * probability)
+
+ mov r7, #1
+ add r4, r7, r6, lsr #8 ; 1 + (((range-1) * probability) >> 8)
+
+ addne r2, r2, r4 ; if (bit) lowvalue += split
+ subne r4, r5, r4 ; if (bit) range = range-split
+
+ ; Counting the leading zeros is used to normalize range.
+ clz r6, r4
+ sub r6, r6, #24 ; shift
+
+ ; Flag is set on the sum of count. This flag is used later
+ ; to determine if count >= 0
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi token_count_lt_zero ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset = shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl token_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos-1
+ b token_zero_while_start
+token_zero_while_loop
+ mov r9, #0
+ strb r9, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+token_zero_while_start
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r1, [r7, r4]
+ cmpge r1, #0xff
+ beq token_zero_while_loop
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r9, [r7, r4] ; w->buffer[x]
+ add r9, r9, #1
+ strb r9, [r7, r4] ; w->buffer[x] + 1
+token_high_bit_not_set
+ rsb r4, r6, #24 ; 24-offset
+ ldr r9, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos] ; w->pos
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r1, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r1, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r9, r1 ; validate_buffer at pos
+
+ strb r7, [r9, r4] ; w->buffer[w->pos++]
+
+token_count_lt_zero
+ lsl r2, r2, r6 ; lowvalue <<= shift
+
+ str r2, [r0, #vp8_writer_lowvalue]
+ str r5, [r0, #vp8_writer_range]
+ str r3, [r0, #vp8_writer_count]
+ pop {r4-r10, pc}
+ ENDP
+
+; r0 BOOL_CODER *br
+|vp8_stop_encode| PROC
+ push {r4-r10, lr}
+
+ ldr r2, [r0, #vp8_writer_lowvalue]
+ ldr r5, [r0, #vp8_writer_range]
+ ldr r3, [r0, #vp8_writer_count]
+
+ mov r10, #32
+
+stop_encode_loop
+ sub r7, r5, #1 ; range-1
+
+ mov r4, r7, lsl #7 ; ((range-1) * 128)
+
+ mov r7, #1
+ add r4, r7, r4, lsr #8 ; 1 + (((range-1) * 128) >> 8)
+
+ ; Counting the leading zeros is used to normalize range.
+ clz r6, r4
+ sub r6, r6, #24 ; shift
+
+ ; Flag is set on the sum of count. This flag is used later
+ ; to determine if count >= 0
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi token_count_lt_zero_se ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset = shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl token_high_bit_not_set_se
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos-1
+ b token_zero_while_start_se
+token_zero_while_loop_se
+ mov r9, #0
+ strb r9, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+token_zero_while_start_se
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r1, [r7, r4]
+ cmpge r1, #0xff
+ beq token_zero_while_loop_se
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r9, [r7, r4] ; w->buffer[x]
+ add r9, r9, #1
+ strb r9, [r7, r4] ; w->buffer[x] + 1
+token_high_bit_not_set_se
+ rsb r4, r6, #24 ; 24-offset
+ ldr r9, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos] ; w->pos
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r1, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r1, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r9, r1 ; validate_buffer at pos
+
+ strb r7, [r9, r4] ; w->buffer[w->pos++]
+
+token_count_lt_zero_se
+ lsl r2, r2, r6 ; lowvalue <<= shift
+
+ subs r10, r10, #1
+ bne stop_encode_loop
+
+ str r2, [r0, #vp8_writer_lowvalue]
+ str r5, [r0, #vp8_writer_range]
+ str r3, [r0, #vp8_writer_count]
+ pop {r4-r10, pc}
+
+ ENDP
+
+; r0 BOOL_CODER *br
+; r1 int data
+; r2 int bits
+|vp8_encode_value| PROC
+ push {r4-r12, lr}
+
+ mov r10, r2
+
+ ldr r2, [r0, #vp8_writer_lowvalue]
+ ldr r5, [r0, #vp8_writer_range]
+ ldr r3, [r0, #vp8_writer_count]
+
+ rsb r4, r10, #32 ; 32-n
+
+ ; v is kept in r1 during the token pack loop
+ lsl r1, r1, r4 ; r1 = v << 32 - n
+
+encode_value_loop
+ sub r7, r5, #1 ; range-1
+
+ ; Decisions are made based on the bit value shifted
+ ; off of v, so set a flag here based on this.
+ ; This value is refered to as "bb"
+ lsls r1, r1, #1 ; bit = v >> n
+ mov r4, r7, lsl #7 ; ((range-1) * 128)
+
+ mov r7, #1
+ add r4, r7, r4, lsr #8 ; 1 + (((range-1) * 128) >> 8)
+
+ addcs r2, r2, r4 ; if (bit) lowvalue += split
+ subcs r4, r5, r4 ; if (bit) range = range-split
+
+ ; Counting the leading zeros is used to normalize range.
+ clz r6, r4
+ sub r6, r6, #24 ; shift
+
+ ; Flag is set on the sum of count. This flag is used later
+ ; to determine if count >= 0
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi token_count_lt_zero_ev ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset = shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl token_high_bit_not_set_ev
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos-1
+ b token_zero_while_start_ev
+token_zero_while_loop_ev
+ mov r9, #0
+ strb r9, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+token_zero_while_start_ev
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r11, [r7, r4]
+ cmpge r11, #0xff
+ beq token_zero_while_loop_ev
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r9, [r7, r4] ; w->buffer[x]
+ add r9, r9, #1
+ strb r9, [r7, r4] ; w->buffer[x] + 1
+token_high_bit_not_set_ev
+ rsb r4, r6, #24 ; 24-offset
+ ldr r9, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos] ; w->pos
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r11, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r11, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r9, r11 ; validate_buffer at pos
+
+ strb r7, [r9, r4] ; w->buffer[w->pos++]
+
+token_count_lt_zero_ev
+ lsl r2, r2, r6 ; lowvalue <<= shift
+
+ subs r10, r10, #1
+ bne encode_value_loop
+
+ str r2, [r0, #vp8_writer_lowvalue]
+ str r5, [r0, #vp8_writer_range]
+ str r3, [r0, #vp8_writer_count]
+ pop {r4-r12, pc}
+ ENDP
+
+ END
diff --git a/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_armv5.asm b/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_armv5.asm
new file mode 100644
index 0000000..a1cd467
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_armv5.asm
@@ -0,0 +1,317 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8cx_pack_tokens_armv5|
+ IMPORT |vp8_validate_buffer_arm|
+
+ INCLUDE asm_enc_offsets.asm
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA |.text|, CODE, READONLY
+
+
+ ; macro for validating write buffer position
+ ; needs vp8_writer in r0
+ ; start shall not be in r1
+ MACRO
+ VALIDATE_POS $start, $pos
+ push {r0-r3, r12, lr} ; rest of regs are preserved by subroutine call
+ ldr r2, [r0, #vp8_writer_buffer_end]
+ ldr r3, [r0, #vp8_writer_error]
+ mov r1, $pos
+ mov r0, $start
+ bl vp8_validate_buffer_arm
+ pop {r0-r3, r12, lr}
+ MEND
+
+
+; r0 vp8_writer *w
+; r1 const TOKENEXTRA *p
+; r2 int xcount
+; r3 vp8_coef_encodings
+; s0 vp8_extra_bits
+; s1 vp8_coef_tree
+|vp8cx_pack_tokens_armv5| PROC
+ push {r4-r12, lr}
+ sub sp, sp, #16
+
+ ; Add size of xcount * sizeof (TOKENEXTRA) to get stop
+ ; sizeof (TOKENEXTRA) is 8
+ add r2, r1, r2, lsl #3 ; stop = p + xcount*sizeof(TOKENEXTRA)
+ str r2, [sp, #0]
+ str r3, [sp, #8] ; save vp8_coef_encodings
+ ldr r2, [r0, #vp8_writer_lowvalue]
+ ldr r5, [r0, #vp8_writer_range]
+ ldr r3, [r0, #vp8_writer_count]
+ b check_p_lt_stop
+
+while_p_lt_stop
+ ldrb r6, [r1, #tokenextra_token] ; t
+ ldr r4, [sp, #8] ; vp8_coef_encodings
+ mov lr, #0
+ add r4, r4, r6, lsl #3 ; a = vp8_coef_encodings + t
+ ldr r9, [r1, #tokenextra_context_tree] ; pp
+
+ ldrb r7, [r1, #tokenextra_skip_eob_node]
+
+ ldr r6, [r4, #vp8_token_value] ; v
+ ldr r8, [r4, #vp8_token_len] ; n
+
+ ; vp8 specific skip_eob_node
+ cmp r7, #0
+ movne lr, #2 ; i = 2
+ subne r8, r8, #1 ; --n
+
+ rsb r4, r8, #32 ; 32-n
+ ldr r10, [sp, #60] ; vp8_coef_tree
+
+ ; v is kept in r12 during the token pack loop
+ lsl r12, r6, r4 ; r12 = v << 32 - n
+
+; loop start
+token_loop
+ ldrb r4, [r9, lr, asr #1] ; pp [i>>1]
+ sub r7, r5, #1 ; range-1
+
+ ; Decisions are made based on the bit value shifted
+ ; off of v, so set a flag here based on this.
+ ; This value is refered to as "bb"
+ lsls r12, r12, #1 ; bb = v >> n
+ mul r6, r4, r7 ; ((range-1) * pp[i>>1]))
+
+ ; bb can only be 0 or 1. So only execute this statement
+ ; if bb == 1, otherwise it will act like i + 0
+ addcs lr, lr, #1 ; i + bb
+
+ mov r7, #1
+ ldrsb lr, [r10, lr] ; i = vp8_coef_tree[i+bb]
+ add r4, r7, r6, lsr #8 ; 1 + (((range-1) * pp[i>>1]) >> 8)
+
+ addcs r2, r2, r4 ; if (bb) lowvalue += split
+ subcs r4, r5, r4 ; if (bb) range = range-split
+
+ ; Counting the leading zeros is used to normalize range.
+ clz r6, r4
+ sub r6, r6, #24 ; shift
+
+ ; Flag is set on the sum of count. This flag is used later
+ ; to determine if count >= 0
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi token_count_lt_zero ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset = shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl token_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos-1
+ b token_zero_while_start
+token_zero_while_loop
+ mov r10, #0
+ strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+token_zero_while_start
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r11, [r7, r4]
+ cmpge r11, #0xff
+ beq token_zero_while_loop
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r10, [r7, r4] ; w->buffer[x]
+ add r10, r10, #1
+ strb r10, [r7, r4] ; w->buffer[x] + 1
+token_high_bit_not_set
+ rsb r4, r6, #24 ; 24-offset
+ ldr r10, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos] ; w->pos
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r11, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r11, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r10, r11 ; validate_buffer at pos
+
+ strb r7, [r10, r4] ; w->buffer[w->pos++]
+
+ ; r10 is used earlier in the loop, but r10 is used as
+ ; temp variable here. So after r10 is used, reload
+ ; vp8_coef_tree_dcd into r10
+ ldr r10, [sp, #60] ; vp8_coef_tree
+
+token_count_lt_zero
+ lsl r2, r2, r6 ; lowvalue <<= shift
+
+ subs r8, r8, #1 ; --n
+ bne token_loop
+
+ ldrb r6, [r1, #tokenextra_token] ; t
+ ldr r7, [sp, #56] ; vp8_extra_bits
+ ; Add t * sizeof (vp8_extra_bit_struct) to get the desired
+ ; element. Here vp8_extra_bit_struct == 16
+ add r12, r7, r6, lsl #4 ; b = vp8_extra_bits + t
+
+ ldr r4, [r12, #vp8_extra_bit_struct_base_val]
+ cmp r4, #0
+ beq skip_extra_bits
+
+; if( b->base_val)
+ ldr r8, [r12, #vp8_extra_bit_struct_len] ; L
+ ldrsh lr, [r1, #tokenextra_extra] ; e = p->Extra
+ cmp r8, #0 ; if( L)
+ beq no_extra_bits
+
+ ldr r9, [r12, #vp8_extra_bit_struct_prob]
+ asr r7, lr, #1 ; v=e>>1
+
+ ldr r10, [r12, #vp8_extra_bit_struct_tree]
+ str r10, [sp, #4] ; b->tree
+
+ rsb r4, r8, #32
+ lsl r12, r7, r4
+
+ mov lr, #0 ; i = 0
+
+extra_bits_loop
+ ldrb r4, [r9, lr, asr #1] ; pp[i>>1]
+ sub r7, r5, #1 ; range-1
+ lsls r12, r12, #1 ; v >> n
+ mul r6, r4, r7 ; (range-1) * pp[i>>1]
+ addcs lr, lr, #1 ; i + bb
+
+ mov r7, #1
+ ldrsb lr, [r10, lr] ; i = b->tree[i+bb]
+ add r4, r7, r6, lsr #8 ; split = 1 + (((range-1) * pp[i>>1]) >> 8)
+
+ addcs r2, r2, r4 ; if (bb) lowvalue += split
+ subcs r4, r5, r4 ; if (bb) range = range-split
+
+ clz r6, r4
+ sub r6, r6, #24
+
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi extra_count_lt_zero ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset= shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl extra_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos - 1
+ b extra_zero_while_start
+extra_zero_while_loop
+ mov r10, #0
+ strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+extra_zero_while_start
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r11, [r7, r4]
+ cmpge r11, #0xff
+ beq extra_zero_while_loop
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r10, [r7, r4]
+ add r10, r10, #1
+ strb r10, [r7, r4]
+extra_high_bit_not_set
+ rsb r4, r6, #24 ; 24-offset
+ ldr r10, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos]
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r11, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r11, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r10, r11 ; validate_buffer at pos
+
+ strb r7, [r10, r4] ; w->buffer[w->pos++]=(lowvalue >> (24-offset))
+ ldr r10, [sp, #4] ; b->tree
+extra_count_lt_zero
+ lsl r2, r2, r6
+
+ subs r8, r8, #1 ; --n
+ bne extra_bits_loop ; while (n)
+
+no_extra_bits
+ ldr lr, [r1, #4] ; e = p->Extra
+ add r4, r5, #1 ; range + 1
+ tst lr, #1
+ lsr r4, r4, #1 ; split = (range + 1) >> 1
+ addne r2, r2, r4 ; lowvalue += split
+ subne r4, r5, r4 ; range = range-split
+ tst r2, #0x80000000 ; lowvalue & 0x80000000
+ lsl r5, r4, #1 ; range <<= 1
+ beq end_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos]
+ mov r7, #0
+ sub r4, r4, #1
+ b end_zero_while_start
+end_zero_while_loop
+ strb r7, [r6, r4]
+ sub r4, r4, #1 ; x--
+end_zero_while_start
+ cmp r4, #0
+ ldrge r6, [r0, #vp8_writer_buffer]
+ ldrb r12, [r6, r4]
+ cmpge r12, #0xff
+ beq end_zero_while_loop
+
+ ldr r6, [r0, #vp8_writer_buffer]
+ ldrb r7, [r6, r4]
+ add r7, r7, #1
+ strb r7, [r6, r4]
+end_high_bit_not_set
+ adds r3, r3, #1 ; ++count
+ lsl r2, r2, #1 ; lowvalue <<= 1
+ bne end_count_zero
+
+ ldr r4, [r0, #vp8_writer_pos]
+ mvn r3, #7
+ ldr r7, [r0, #vp8_writer_buffer]
+ lsr r6, r2, #24 ; lowvalue >> 24
+ add r12, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r12, [r0, #vp8_writer_pos]
+
+ VALIDATE_POS r7, r12 ; validate_buffer at pos
+
+ strb r6, [r7, r4]
+end_count_zero
+skip_extra_bits
+ add r1, r1, #TOKENEXTRA_SZ ; ++p
+check_p_lt_stop
+ ldr r4, [sp, #0] ; stop
+ cmp r1, r4 ; while( p < stop)
+ bcc while_p_lt_stop
+
+ str r2, [r0, #vp8_writer_lowvalue]
+ str r5, [r0, #vp8_writer_range]
+ str r3, [r0, #vp8_writer_count]
+ add sp, sp, #16
+ pop {r4-r12, pc}
+ ENDP
+
+ END
diff --git a/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_mbrow_armv5.asm b/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_mbrow_armv5.asm
new file mode 100644
index 0000000..1fa5e6c
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_mbrow_armv5.asm
@@ -0,0 +1,352 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8cx_pack_mb_row_tokens_armv5|
+ IMPORT |vp8_validate_buffer_arm|
+
+ INCLUDE asm_enc_offsets.asm
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA |.text|, CODE, READONLY
+
+
+ ; macro for validating write buffer position
+ ; needs vp8_writer in r0
+ ; start shall not be in r1
+ MACRO
+ VALIDATE_POS $start, $pos
+ push {r0-r3, r12, lr} ; rest of regs are preserved by subroutine call
+ ldr r2, [r0, #vp8_writer_buffer_end]
+ ldr r3, [r0, #vp8_writer_error]
+ mov r1, $pos
+ mov r0, $start
+ bl vp8_validate_buffer_arm
+ pop {r0-r3, r12, lr}
+ MEND
+
+; r0 VP8_COMP *cpi
+; r1 vp8_writer *w
+; r2 vp8_coef_encodings
+; r3 vp8_extra_bits
+; s0 vp8_coef_tree
+
+|vp8cx_pack_mb_row_tokens_armv5| PROC
+ push {r4-r12, lr}
+ sub sp, sp, #24
+
+ ; Compute address of cpi->common.mb_rows
+ ldr r4, _VP8_COMP_common_
+ ldr r6, _VP8_COMMON_MBrows_
+ add r4, r0, r4
+
+ ldr r5, [r4, r6] ; load up mb_rows
+
+ str r2, [sp, #20] ; save vp8_coef_encodings
+ str r5, [sp, #12] ; save mb_rows
+ str r3, [sp, #8] ; save vp8_extra_bits
+
+ ldr r4, _VP8_COMP_tplist_
+ add r4, r0, r4
+ ldr r7, [r4, #0] ; dereference cpi->tp_list
+
+ mov r0, r1 ; keep same as other loops
+
+ ldr r2, [r0, #vp8_writer_lowvalue]
+ ldr r5, [r0, #vp8_writer_range]
+ ldr r3, [r0, #vp8_writer_count]
+
+mb_row_loop
+
+ ldr r1, [r7, #tokenlist_start]
+ ldr r9, [r7, #tokenlist_stop]
+ str r9, [sp, #0] ; save stop for later comparison
+ str r7, [sp, #16] ; tokenlist address for next time
+
+ b check_p_lt_stop
+
+ ; actuall work gets done here!
+
+while_p_lt_stop
+ ldrb r6, [r1, #tokenextra_token] ; t
+ ldr r4, [sp, #20] ; vp8_coef_encodings
+ mov lr, #0
+ add r4, r4, r6, lsl #3 ; a = vp8_coef_encodings + t
+ ldr r9, [r1, #tokenextra_context_tree] ; pp
+
+ ldrb r7, [r1, #tokenextra_skip_eob_node]
+
+ ldr r6, [r4, #vp8_token_value] ; v
+ ldr r8, [r4, #vp8_token_len] ; n
+
+ ; vp8 specific skip_eob_node
+ cmp r7, #0
+ movne lr, #2 ; i = 2
+ subne r8, r8, #1 ; --n
+
+ rsb r4, r8, #32 ; 32-n
+ ldr r10, [sp, #64] ; vp8_coef_tree
+
+ ; v is kept in r12 during the token pack loop
+ lsl r12, r6, r4 ; r12 = v << 32 - n
+
+; loop start
+token_loop
+ ldrb r4, [r9, lr, asr #1] ; pp [i>>1]
+ sub r7, r5, #1 ; range-1
+
+ ; Decisions are made based on the bit value shifted
+ ; off of v, so set a flag here based on this.
+ ; This value is refered to as "bb"
+ lsls r12, r12, #1 ; bb = v >> n
+ mul r6, r4, r7 ; ((range-1) * pp[i>>1]))
+
+ ; bb can only be 0 or 1. So only execute this statement
+ ; if bb == 1, otherwise it will act like i + 0
+ addcs lr, lr, #1 ; i + bb
+
+ mov r7, #1
+ ldrsb lr, [r10, lr] ; i = vp8_coef_tree[i+bb]
+ add r4, r7, r6, lsr #8 ; 1 + (((range-1) * pp[i>>1]) >> 8)
+
+ addcs r2, r2, r4 ; if (bb) lowvalue += split
+ subcs r4, r5, r4 ; if (bb) range = range-split
+
+ ; Counting the leading zeros is used to normalize range.
+ clz r6, r4
+ sub r6, r6, #24 ; shift
+
+ ; Flag is set on the sum of count. This flag is used later
+ ; to determine if count >= 0
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi token_count_lt_zero ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset = shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl token_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos-1
+ b token_zero_while_start
+token_zero_while_loop
+ mov r10, #0
+ strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+token_zero_while_start
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r11, [r7, r4]
+ cmpge r11, #0xff
+ beq token_zero_while_loop
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r10, [r7, r4] ; w->buffer[x]
+ add r10, r10, #1
+ strb r10, [r7, r4] ; w->buffer[x] + 1
+token_high_bit_not_set
+ rsb r4, r6, #24 ; 24-offset
+ ldr r10, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos] ; w->pos
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r11, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r11, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r10, r11 ; validate_buffer at pos
+
+ strb r7, [r10, r4] ; w->buffer[w->pos++]
+
+ ; r10 is used earlier in the loop, but r10 is used as
+ ; temp variable here. So after r10 is used, reload
+ ; vp8_coef_tree_dcd into r10
+ ldr r10, [sp, #64] ; vp8_coef_tree
+
+token_count_lt_zero
+ lsl r2, r2, r6 ; lowvalue <<= shift
+
+ subs r8, r8, #1 ; --n
+ bne token_loop
+
+ ldrb r6, [r1, #tokenextra_token] ; t
+ ldr r7, [sp, #8] ; vp8_extra_bits
+ ; Add t * sizeof (vp8_extra_bit_struct) to get the desired
+ ; element. Here vp8_extra_bit_struct == 16
+ add r12, r7, r6, lsl #4 ; b = vp8_extra_bits + t
+
+ ldr r4, [r12, #vp8_extra_bit_struct_base_val]
+ cmp r4, #0
+ beq skip_extra_bits
+
+; if( b->base_val)
+ ldr r8, [r12, #vp8_extra_bit_struct_len] ; L
+ ldrsh lr, [r1, #tokenextra_extra] ; e = p->Extra
+ cmp r8, #0 ; if( L)
+ beq no_extra_bits
+
+ ldr r9, [r12, #vp8_extra_bit_struct_prob]
+ asr r7, lr, #1 ; v=e>>1
+
+ ldr r10, [r12, #vp8_extra_bit_struct_tree]
+ str r10, [sp, #4] ; b->tree
+
+ rsb r4, r8, #32
+ lsl r12, r7, r4
+
+ mov lr, #0 ; i = 0
+
+extra_bits_loop
+ ldrb r4, [r9, lr, asr #1] ; pp[i>>1]
+ sub r7, r5, #1 ; range-1
+ lsls r12, r12, #1 ; v >> n
+ mul r6, r4, r7 ; (range-1) * pp[i>>1]
+ addcs lr, lr, #1 ; i + bb
+
+ mov r7, #1
+ ldrsb lr, [r10, lr] ; i = b->tree[i+bb]
+ add r4, r7, r6, lsr #8 ; split = 1 + (((range-1) * pp[i>>1]) >> 8)
+
+ addcs r2, r2, r4 ; if (bb) lowvalue += split
+ subcs r4, r5, r4 ; if (bb) range = range-split
+
+ clz r6, r4
+ sub r6, r6, #24
+
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi extra_count_lt_zero ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset= shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl extra_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos - 1
+ b extra_zero_while_start
+extra_zero_while_loop
+ mov r10, #0
+ strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+extra_zero_while_start
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r11, [r7, r4]
+ cmpge r11, #0xff
+ beq extra_zero_while_loop
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r10, [r7, r4]
+ add r10, r10, #1
+ strb r10, [r7, r4]
+extra_high_bit_not_set
+ rsb r4, r6, #24 ; 24-offset
+ ldr r10, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos]
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r11, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r11, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r10, r11 ; validate_buffer at pos
+
+ strb r7, [r10, r4] ; w->buffer[w->pos++]=(lowvalue >> (24-offset))
+ ldr r10, [sp, #4] ; b->tree
+extra_count_lt_zero
+ lsl r2, r2, r6
+
+ subs r8, r8, #1 ; --n
+ bne extra_bits_loop ; while (n)
+
+no_extra_bits
+ ldr lr, [r1, #4] ; e = p->Extra
+ add r4, r5, #1 ; range + 1
+ tst lr, #1
+ lsr r4, r4, #1 ; split = (range + 1) >> 1
+ addne r2, r2, r4 ; lowvalue += split
+ subne r4, r5, r4 ; range = range-split
+ tst r2, #0x80000000 ; lowvalue & 0x80000000
+ lsl r5, r4, #1 ; range <<= 1
+ beq end_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos]
+ mov r7, #0
+ sub r4, r4, #1
+ b end_zero_while_start
+end_zero_while_loop
+ strb r7, [r6, r4]
+ sub r4, r4, #1 ; x--
+end_zero_while_start
+ cmp r4, #0
+ ldrge r6, [r0, #vp8_writer_buffer]
+ ldrb r12, [r6, r4]
+ cmpge r12, #0xff
+ beq end_zero_while_loop
+
+ ldr r6, [r0, #vp8_writer_buffer]
+ ldrb r7, [r6, r4]
+ add r7, r7, #1
+ strb r7, [r6, r4]
+end_high_bit_not_set
+ adds r3, r3, #1 ; ++count
+ lsl r2, r2, #1 ; lowvalue <<= 1
+ bne end_count_zero
+
+ ldr r4, [r0, #vp8_writer_pos]
+ mvn r3, #7
+ ldr r7, [r0, #vp8_writer_buffer]
+ lsr r6, r2, #24 ; lowvalue >> 24
+ add r12, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r12, [r0, #vp8_writer_pos]
+
+ VALIDATE_POS r7, r12 ; validate_buffer at pos
+
+ strb r6, [r7, r4]
+end_count_zero
+skip_extra_bits
+ add r1, r1, #TOKENEXTRA_SZ ; ++p
+check_p_lt_stop
+ ldr r4, [sp, #0] ; stop
+ cmp r1, r4 ; while( p < stop)
+ bcc while_p_lt_stop
+
+ ldr r6, [sp, #12] ; mb_rows
+ ldr r7, [sp, #16] ; tokenlist address
+ subs r6, r6, #1
+ add r7, r7, #TOKENLIST_SZ ; next element in the array
+ str r6, [sp, #12]
+ bne mb_row_loop
+
+ str r2, [r0, #vp8_writer_lowvalue]
+ str r5, [r0, #vp8_writer_range]
+ str r3, [r0, #vp8_writer_count]
+ add sp, sp, #24
+ pop {r4-r12, pc}
+ ENDP
+
+_VP8_COMP_common_
+ DCD vp8_comp_common
+_VP8_COMMON_MBrows_
+ DCD vp8_common_mb_rows
+_VP8_COMP_tplist_
+ DCD vp8_comp_tplist
+
+ END
diff --git a/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_partitions_armv5.asm b/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_partitions_armv5.asm
new file mode 100644
index 0000000..90a98fe
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv5te/vp8_packtokens_partitions_armv5.asm
@@ -0,0 +1,471 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8cx_pack_tokens_into_partitions_armv5|
+ IMPORT |vp8_validate_buffer_arm|
+
+ INCLUDE asm_enc_offsets.asm
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA |.text|, CODE, READONLY
+
+ ; macro for validating write buffer position
+ ; needs vp8_writer in r0
+ ; start shall not be in r1
+ MACRO
+ VALIDATE_POS $start, $pos
+ push {r0-r3, r12, lr} ; rest of regs are preserved by subroutine call
+ ldr r2, [r0, #vp8_writer_buffer_end]
+ ldr r3, [r0, #vp8_writer_error]
+ mov r1, $pos
+ mov r0, $start
+ bl vp8_validate_buffer_arm
+ pop {r0-r3, r12, lr}
+ MEND
+
+; r0 VP8_COMP *cpi
+; r1 unsigned char *cx_data
+; r2 const unsigned char *cx_data_end
+; r3 int num_part
+; s0 vp8_coef_encodings
+; s1 vp8_extra_bits,
+; s2 const vp8_tree_index *
+
+|vp8cx_pack_tokens_into_partitions_armv5| PROC
+ push {r4-r12, lr}
+ sub sp, sp, #40
+
+ ; Compute address of cpi->common.mb_rows
+ ldr r4, _VP8_COMP_common_
+ ldr r6, _VP8_COMMON_MBrows_
+ add r4, r0, r4
+
+ ldr r5, [r4, r6] ; load up mb_rows
+
+ str r5, [sp, #36] ; save mb_rows
+ str r1, [sp, #24] ; save ptr = cx_data
+ str r3, [sp, #20] ; save num_part
+ str r2, [sp, #8] ; save cx_data_end
+
+ ldr r4, _VP8_COMP_tplist_
+ add r4, r0, r4
+ ldr r7, [r4, #0] ; dereference cpi->tp_list
+ str r7, [sp, #32] ; store start of cpi->tp_list
+
+ ldr r11, _VP8_COMP_bc_ ; load up vp8_writer out of cpi
+ add r0, r0, r11
+
+ mov r11, #0
+ str r11, [sp, #28] ; i
+
+numparts_loop
+ ldr r2, _vp8_writer_sz_ ; load up sizeof(vp8_writer)
+ add r0, r2 ; bc[i + 1]
+
+ ldr r10, [sp, #24] ; ptr
+ ldr r5, [sp, #36] ; move mb_rows to the counting section
+ subs r5, r5, r11 ; move start point with each partition
+ ; mb_rows starts at i
+ str r5, [sp, #12]
+
+ ; Reset all of the VP8 Writer data for each partition that
+ ; is processed.
+ ; start_encode
+
+ ldr r3, [sp, #8]
+ str r3, [r0, #vp8_writer_buffer_end]
+
+ mov r2, #0 ; vp8_writer_lowvalue
+ mov r5, #255 ; vp8_writer_range
+ mvn r3, #23 ; vp8_writer_count
+
+ str r2, [r0, #vp8_writer_pos]
+ str r10, [r0, #vp8_writer_buffer]
+
+ ble end_partition ; if (mb_rows <= 0) end partition
+
+mb_row_loop
+
+ ldr r1, [r7, #tokenlist_start]
+ ldr r9, [r7, #tokenlist_stop]
+ str r9, [sp, #0] ; save stop for later comparison
+ str r7, [sp, #16] ; tokenlist address for next time
+
+ b check_p_lt_stop
+
+ ; actual work gets done here!
+
+while_p_lt_stop
+ ldrb r6, [r1, #tokenextra_token] ; t
+ ldr r4, [sp, #80] ; vp8_coef_encodings
+ mov lr, #0
+ add r4, r4, r6, lsl #3 ; a = vp8_coef_encodings + t
+ ldr r9, [r1, #tokenextra_context_tree] ; pp
+
+ ldrb r7, [r1, #tokenextra_skip_eob_node]
+
+ ldr r6, [r4, #vp8_token_value] ; v
+ ldr r8, [r4, #vp8_token_len] ; n
+
+ ; vp8 specific skip_eob_node
+ cmp r7, #0
+ movne lr, #2 ; i = 2
+ subne r8, r8, #1 ; --n
+
+ rsb r4, r8, #32 ; 32-n
+ ldr r10, [sp, #88] ; vp8_coef_tree
+
+ ; v is kept in r12 during the token pack loop
+ lsl r12, r6, r4 ; r12 = v << 32 - n
+
+; loop start
+token_loop
+ ldrb r4, [r9, lr, asr #1] ; pp [i>>1]
+ sub r7, r5, #1 ; range-1
+
+ ; Decisions are made based on the bit value shifted
+ ; off of v, so set a flag here based on this.
+ ; This value is refered to as "bb"
+ lsls r12, r12, #1 ; bb = v >> n
+ mul r6, r4, r7 ; ((range-1) * pp[i>>1]))
+
+ ; bb can only be 0 or 1. So only execute this statement
+ ; if bb == 1, otherwise it will act like i + 0
+ addcs lr, lr, #1 ; i + bb
+
+ mov r7, #1
+ ldrsb lr, [r10, lr] ; i = vp8_coef_tree[i+bb]
+ add r4, r7, r6, lsr #8 ; 1 + (((range-1) * pp[i>>1]) >> 8)
+
+ addcs r2, r2, r4 ; if (bb) lowvalue += split
+ subcs r4, r5, r4 ; if (bb) range = range-split
+
+ ; Counting the leading zeros is used to normalize range.
+ clz r6, r4
+ sub r6, r6, #24 ; shift
+
+ ; Flag is set on the sum of count. This flag is used later
+ ; to determine if count >= 0
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi token_count_lt_zero ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset = shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl token_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos-1
+ b token_zero_while_start
+token_zero_while_loop
+ mov r10, #0
+ strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+token_zero_while_start
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r11, [r7, r4]
+ cmpge r11, #0xff
+ beq token_zero_while_loop
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r10, [r7, r4] ; w->buffer[x]
+ add r10, r10, #1
+ strb r10, [r7, r4] ; w->buffer[x] + 1
+token_high_bit_not_set
+ rsb r4, r6, #24 ; 24-offset
+ ldr r10, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos] ; w->pos
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r11, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r11, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r10, r11 ; validate_buffer at pos
+
+ strb r7, [r10, r4] ; w->buffer[w->pos++]
+
+ ; r10 is used earlier in the loop, but r10 is used as
+ ; temp variable here. So after r10 is used, reload
+ ; vp8_coef_tree_dcd into r10
+ ldr r10, [sp, #88] ; vp8_coef_tree
+
+token_count_lt_zero
+ lsl r2, r2, r6 ; lowvalue <<= shift
+
+ subs r8, r8, #1 ; --n
+ bne token_loop
+
+ ldrb r6, [r1, #tokenextra_token] ; t
+ ldr r7, [sp, #84] ; vp8_extra_bits
+ ; Add t * sizeof (vp8_extra_bit_struct) to get the desired
+ ; element. Here vp8_extra_bit_struct == 16
+ add r12, r7, r6, lsl #4 ; b = vp8_extra_bits + t
+
+ ldr r4, [r12, #vp8_extra_bit_struct_base_val]
+ cmp r4, #0
+ beq skip_extra_bits
+
+; if( b->base_val)
+ ldr r8, [r12, #vp8_extra_bit_struct_len] ; L
+ ldrsh lr, [r1, #tokenextra_extra] ; e = p->Extra
+ cmp r8, #0 ; if( L)
+ beq no_extra_bits
+
+ ldr r9, [r12, #vp8_extra_bit_struct_prob]
+ asr r7, lr, #1 ; v=e>>1
+
+ ldr r10, [r12, #vp8_extra_bit_struct_tree]
+ str r10, [sp, #4] ; b->tree
+
+ rsb r4, r8, #32
+ lsl r12, r7, r4
+
+ mov lr, #0 ; i = 0
+
+extra_bits_loop
+ ldrb r4, [r9, lr, asr #1] ; pp[i>>1]
+ sub r7, r5, #1 ; range-1
+ lsls r12, r12, #1 ; v >> n
+ mul r6, r4, r7 ; (range-1) * pp[i>>1]
+ addcs lr, lr, #1 ; i + bb
+
+ mov r7, #1
+ ldrsb lr, [r10, lr] ; i = b->tree[i+bb]
+ add r4, r7, r6, lsr #8 ; split = 1 + (((range-1) * pp[i>>1]) >> 8)
+
+ addcs r2, r2, r4 ; if (bb) lowvalue += split
+ subcs r4, r5, r4 ; if (bb) range = range-split
+
+ clz r6, r4
+ sub r6, r6, #24
+
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi extra_count_lt_zero ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset= shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl extra_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos - 1
+ b extra_zero_while_start
+extra_zero_while_loop
+ mov r10, #0
+ strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+extra_zero_while_start
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r11, [r7, r4]
+ cmpge r11, #0xff
+ beq extra_zero_while_loop
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r10, [r7, r4]
+ add r10, r10, #1
+ strb r10, [r7, r4]
+extra_high_bit_not_set
+ rsb r4, r6, #24 ; 24-offset
+ ldr r10, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos]
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r11, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r11, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r10, r11 ; validate_buffer at pos
+
+ strb r7, [r10, r4] ; w->buffer[w->pos++]=(lowvalue >> (24-offset))
+ ldr r10, [sp, #4] ; b->tree
+extra_count_lt_zero
+ lsl r2, r2, r6
+
+ subs r8, r8, #1 ; --n
+ bne extra_bits_loop ; while (n)
+
+no_extra_bits
+ ldr lr, [r1, #4] ; e = p->Extra
+ add r4, r5, #1 ; range + 1
+ tst lr, #1
+ lsr r4, r4, #1 ; split = (range + 1) >> 1
+ addne r2, r2, r4 ; lowvalue += split
+ subne r4, r5, r4 ; range = range-split
+ tst r2, #0x80000000 ; lowvalue & 0x80000000
+ lsl r5, r4, #1 ; range <<= 1
+ beq end_high_bit_not_set
+
+ ldr r4, [r0, #vp8_writer_pos]
+ mov r7, #0
+ sub r4, r4, #1
+ b end_zero_while_start
+end_zero_while_loop
+ strb r7, [r6, r4]
+ sub r4, r4, #1 ; x--
+end_zero_while_start
+ cmp r4, #0
+ ldrge r6, [r0, #vp8_writer_buffer]
+ ldrb r12, [r6, r4]
+ cmpge r12, #0xff
+ beq end_zero_while_loop
+
+ ldr r6, [r0, #vp8_writer_buffer]
+ ldrb r7, [r6, r4]
+ add r7, r7, #1
+ strb r7, [r6, r4]
+end_high_bit_not_set
+ adds r3, r3, #1 ; ++count
+ lsl r2, r2, #1 ; lowvalue <<= 1
+ bne end_count_zero
+
+ ldr r4, [r0, #vp8_writer_pos]
+ mvn r3, #7 ; count = -8
+ ldr r7, [r0, #vp8_writer_buffer]
+ lsr r6, r2, #24 ; lowvalue >> 24
+ add r12, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r12, [r0, #vp8_writer_pos]
+
+ VALIDATE_POS r7, r12 ; validate_buffer at pos
+
+ strb r6, [r7, r4]
+end_count_zero
+skip_extra_bits
+ add r1, r1, #TOKENEXTRA_SZ ; ++p
+check_p_lt_stop
+ ldr r4, [sp, #0] ; stop
+ cmp r1, r4 ; while( p < stop)
+ bcc while_p_lt_stop
+
+ ldr r10, [sp, #20] ; num_parts
+ mov r1, #TOKENLIST_SZ
+ mul r1, r10, r1
+
+ ldr r6, [sp, #12] ; mb_rows
+ ldr r7, [sp, #16] ; tokenlist address
+ subs r6, r6, r10
+ add r7, r7, r1 ; next element in the array
+ str r6, [sp, #12]
+ bgt mb_row_loop
+
+end_partition
+ mov r12, #32
+
+stop_encode_loop
+ sub r7, r5, #1 ; range-1
+
+ mov r4, r7, lsl #7 ; ((range-1) * 128)
+
+ mov r7, #1
+ add r4, r7, r4, lsr #8 ; 1 + (((range-1) * 128) >> 8)
+
+ ; Counting the leading zeros is used to normalize range.
+ clz r6, r4
+ sub r6, r6, #24 ; shift
+
+ ; Flag is set on the sum of count. This flag is used later
+ ; to determine if count >= 0
+ adds r3, r3, r6 ; count += shift
+ lsl r5, r4, r6 ; range <<= shift
+ bmi token_count_lt_zero_se ; if(count >= 0)
+
+ sub r6, r6, r3 ; offset = shift - count
+ sub r4, r6, #1 ; offset-1
+ lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
+ bpl token_high_bit_not_set_se
+
+ ldr r4, [r0, #vp8_writer_pos] ; x
+ sub r4, r4, #1 ; x = w->pos-1
+ b token_zero_while_start_se
+token_zero_while_loop_se
+ mov r10, #0
+ strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
+ sub r4, r4, #1 ; x--
+token_zero_while_start_se
+ cmp r4, #0
+ ldrge r7, [r0, #vp8_writer_buffer]
+ ldrb r11, [r7, r4]
+ cmpge r11, #0xff
+ beq token_zero_while_loop_se
+
+ ldr r7, [r0, #vp8_writer_buffer]
+ ldrb r10, [r7, r4] ; w->buffer[x]
+ add r10, r10, #1
+ strb r10, [r7, r4] ; w->buffer[x] + 1
+token_high_bit_not_set_se
+ rsb r4, r6, #24 ; 24-offset
+ ldr r10, [r0, #vp8_writer_buffer]
+ lsr r7, r2, r4 ; lowvalue >> (24-offset)
+ ldr r4, [r0, #vp8_writer_pos] ; w->pos
+ lsl r2, r2, r6 ; lowvalue <<= offset
+ mov r6, r3 ; shift = count
+ add r11, r4, #1 ; w->pos++
+ bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
+ str r11, [r0, #vp8_writer_pos]
+ sub r3, r3, #8 ; count -= 8
+
+ VALIDATE_POS r10, r11 ; validate_buffer at pos
+
+ strb r7, [r10, r4] ; w->buffer[w->pos++]
+
+token_count_lt_zero_se
+ lsl r2, r2, r6 ; lowvalue <<= shift
+
+ subs r12, r12, #1
+ bne stop_encode_loop
+
+ ldr r4, [r0, #vp8_writer_pos] ; w->pos
+ ldr r12, [sp, #24] ; ptr
+ add r12, r12, r4 ; ptr += w->pos
+ str r12, [sp, #24]
+
+ ldr r11, [sp, #28] ; i
+ ldr r10, [sp, #20] ; num_parts
+
+ add r11, r11, #1 ; i++
+ str r11, [sp, #28]
+
+ ldr r7, [sp, #32] ; cpi->tp_list[i]
+ mov r1, #TOKENLIST_SZ
+ add r7, r7, r1 ; next element in cpi->tp_list
+ str r7, [sp, #32] ; cpi->tp_list[i+1]
+
+ cmp r10, r11
+ bgt numparts_loop
+
+ add sp, sp, #40
+ pop {r4-r12, pc}
+ ENDP
+
+_VP8_COMP_common_
+ DCD vp8_comp_common
+_VP8_COMMON_MBrows_
+ DCD vp8_common_mb_rows
+_VP8_COMP_tplist_
+ DCD vp8_comp_tplist
+_VP8_COMP_bc_
+ DCD vp8_comp_bc
+_vp8_writer_sz_
+ DCD vp8_writer_sz
+
+ END
diff --git a/libvpx/vp8/encoder/arm/armv6/vp8_fast_quantize_b_armv6.asm b/libvpx/vp8/encoder/arm/armv6/vp8_fast_quantize_b_armv6.asm
new file mode 100644
index 0000000..d61f5d9
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv6/vp8_fast_quantize_b_armv6.asm
@@ -0,0 +1,225 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_fast_quantize_b_armv6|
+
+ INCLUDE asm_enc_offsets.asm
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 BLOCK *b
+; r1 BLOCKD *d
+|vp8_fast_quantize_b_armv6| PROC
+ stmfd sp!, {r1, r4-r11, lr}
+
+ ldr r3, [r0, #vp8_block_coeff] ; coeff
+ ldr r4, [r0, #vp8_block_quant_fast] ; quant_fast
+ ldr r5, [r0, #vp8_block_round] ; round
+ ldr r6, [r1, #vp8_blockd_qcoeff] ; qcoeff
+ ldr r7, [r1, #vp8_blockd_dqcoeff] ; dqcoeff
+ ldr r8, [r1, #vp8_blockd_dequant] ; dequant
+
+ ldr r2, loop_count ; loop_count=0x1000000. 'lsls' instruction
+ ; is used to update the counter so that
+ ; it can be used to mark nonzero
+ ; quantized coefficient pairs.
+
+ mov r1, #0 ; flags for quantized coeffs
+
+ ; PART 1: quantization and dequantization loop
+loop
+ ldr r9, [r3], #4 ; [z1 | z0]
+ ldr r10, [r5], #4 ; [r1 | r0]
+ ldr r11, [r4], #4 ; [q1 | q0]
+
+ ssat16 lr, #1, r9 ; [sz1 | sz0]
+ eor r9, r9, lr ; [z1 ^ sz1 | z0 ^ sz0]
+ ssub16 r9, r9, lr ; x = (z ^ sz) - sz
+ sadd16 r9, r9, r10 ; [x1+r1 | x0+r0]
+
+ ldr r12, [r3], #4 ; [z3 | z2]
+
+ smulbb r0, r9, r11 ; [(x0+r0)*q0]
+ smultt r9, r9, r11 ; [(x1+r1)*q1]
+
+ ldr r10, [r5], #4 ; [r3 | r2]
+
+ ssat16 r11, #1, r12 ; [sz3 | sz2]
+ eor r12, r12, r11 ; [z3 ^ sz3 | z2 ^ sz2]
+ pkhtb r0, r9, r0, asr #16 ; [y1 | y0]
+ ldr r9, [r4], #4 ; [q3 | q2]
+ ssub16 r12, r12, r11 ; x = (z ^ sz) - sz
+
+ sadd16 r12, r12, r10 ; [x3+r3 | x2+r2]
+
+ eor r0, r0, lr ; [(y1 ^ sz1) | (y0 ^ sz0)]
+
+ smulbb r10, r12, r9 ; [(x2+r2)*q2]
+ smultt r12, r12, r9 ; [(x3+r3)*q3]
+
+ ssub16 r0, r0, lr ; x = (y ^ sz) - sz
+
+ cmp r0, #0 ; check if zero
+ orrne r1, r1, r2, lsr #24 ; add flag for nonzero coeffs
+
+ str r0, [r6], #4 ; *qcoeff++ = x
+ ldr r9, [r8], #4 ; [dq1 | dq0]
+
+ pkhtb r10, r12, r10, asr #16 ; [y3 | y2]
+ eor r10, r10, r11 ; [(y3 ^ sz3) | (y2 ^ sz2)]
+ ssub16 r10, r10, r11 ; x = (y ^ sz) - sz
+
+ cmp r10, #0 ; check if zero
+ orrne r1, r1, r2, lsr #23 ; add flag for nonzero coeffs
+
+ str r10, [r6], #4 ; *qcoeff++ = x
+ ldr r11, [r8], #4 ; [dq3 | dq2]
+
+ smulbb r12, r0, r9 ; [x0*dq0]
+ smultt r0, r0, r9 ; [x1*dq1]
+
+ smulbb r9, r10, r11 ; [x2*dq2]
+ smultt r10, r10, r11 ; [x3*dq3]
+
+ lsls r2, r2, #2 ; update loop counter
+ strh r12, [r7, #0] ; dqcoeff[0] = [x0*dq0]
+ strh r0, [r7, #2] ; dqcoeff[1] = [x1*dq1]
+ strh r9, [r7, #4] ; dqcoeff[2] = [x2*dq2]
+ strh r10, [r7, #6] ; dqcoeff[3] = [x3*dq3]
+ add r7, r7, #8 ; dqcoeff += 8
+ bne loop
+
+ ; PART 2: check position for eob...
+ ldr r11, [sp, #0] ; restore BLOCKD pointer
+ mov lr, #0 ; init eob
+ cmp r1, #0 ; coeffs after quantization?
+ ldr r12, [r11, #vp8_blockd_eob]
+ beq end ; skip eob calculations if all zero
+
+ ldr r0, [r11, #vp8_blockd_qcoeff]
+
+ ; check shortcut for nonzero qcoeffs
+ tst r1, #0x80
+ bne quant_coeff_15_14
+ tst r1, #0x20
+ bne quant_coeff_13_11
+ tst r1, #0x8
+ bne quant_coeff_12_7
+ tst r1, #0x40
+ bne quant_coeff_10_9
+ tst r1, #0x10
+ bne quant_coeff_8_3
+ tst r1, #0x2
+ bne quant_coeff_6_5
+ tst r1, #0x4
+ bne quant_coeff_4_2
+ b quant_coeff_1_0
+
+quant_coeff_15_14
+ ldrh r2, [r0, #30] ; rc=15, i=15
+ mov lr, #16
+ cmp r2, #0
+ bne end
+
+ ldrh r3, [r0, #28] ; rc=14, i=14
+ mov lr, #15
+ cmp r3, #0
+ bne end
+
+quant_coeff_13_11
+ ldrh r2, [r0, #22] ; rc=11, i=13
+ mov lr, #14
+ cmp r2, #0
+ bne end
+
+quant_coeff_12_7
+ ldrh r3, [r0, #14] ; rc=7, i=12
+ mov lr, #13
+ cmp r3, #0
+ bne end
+
+ ldrh r2, [r0, #20] ; rc=10, i=11
+ mov lr, #12
+ cmp r2, #0
+ bne end
+
+quant_coeff_10_9
+ ldrh r3, [r0, #26] ; rc=13, i=10
+ mov lr, #11
+ cmp r3, #0
+ bne end
+
+ ldrh r2, [r0, #24] ; rc=12, i=9
+ mov lr, #10
+ cmp r2, #0
+ bne end
+
+quant_coeff_8_3
+ ldrh r3, [r0, #18] ; rc=9, i=8
+ mov lr, #9
+ cmp r3, #0
+ bne end
+
+ ldrh r2, [r0, #12] ; rc=6, i=7
+ mov lr, #8
+ cmp r2, #0
+ bne end
+
+quant_coeff_6_5
+ ldrh r3, [r0, #6] ; rc=3, i=6
+ mov lr, #7
+ cmp r3, #0
+ bne end
+
+ ldrh r2, [r0, #4] ; rc=2, i=5
+ mov lr, #6
+ cmp r2, #0
+ bne end
+
+quant_coeff_4_2
+ ldrh r3, [r0, #10] ; rc=5, i=4
+ mov lr, #5
+ cmp r3, #0
+ bne end
+
+ ldrh r2, [r0, #16] ; rc=8, i=3
+ mov lr, #4
+ cmp r2, #0
+ bne end
+
+ ldrh r3, [r0, #8] ; rc=4, i=2
+ mov lr, #3
+ cmp r3, #0
+ bne end
+
+quant_coeff_1_0
+ ldrh r2, [r0, #2] ; rc=1, i=1
+ mov lr, #2
+ cmp r2, #0
+ bne end
+
+ mov lr, #1 ; rc=0, i=0
+
+end
+ strb lr, [r12]
+ ldmfd sp!, {r1, r4-r11, pc}
+
+ ENDP
+
+loop_count
+ DCD 0x1000000
+
+ END
+
diff --git a/libvpx/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm b/libvpx/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm
new file mode 100644
index 0000000..000805d
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm
@@ -0,0 +1,138 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_mse16x16_armv6|
+
+ ARM
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+;
+;note: Based on vp8_variance16x16_armv6. In this function, sum is never used.
+; So, we can remove this part of calculation.
+
+|vp8_mse16x16_armv6| PROC
+
+ push {r4-r9, lr}
+
+ pld [r0, r1, lsl #0]
+ pld [r2, r3, lsl #0]
+
+ mov r12, #16 ; set loop counter to 16 (=block height)
+ mov r4, #0 ; initialize sse = 0
+
+loop
+ ; 1st 4 pixels
+ ldr r5, [r0, #0x0] ; load 4 src pixels
+ ldr r6, [r2, #0x0] ; load 4 ref pixels
+
+ mov lr, #0 ; constant zero
+
+ usub8 r8, r5, r6 ; calculate difference
+ pld [r0, r1, lsl #1]
+ sel r7, r8, lr ; select bytes with positive difference
+ usub8 r9, r6, r5 ; calculate difference with reversed operands
+ pld [r2, r3, lsl #1]
+ sel r8, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r5, r7, lr ; calculate sum of positive differences
+ usad8 r6, r8, lr ; calculate sum of negative differences
+ orr r8, r8, r7 ; differences of all 4 pixels
+
+ ldr r5, [r0, #0x4] ; load 4 src pixels
+
+ ; calculate sse
+ uxtb16 r6, r8 ; byte (two pixels) to halfwords
+ uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
+ smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
+
+ ; 2nd 4 pixels
+ ldr r6, [r2, #0x4] ; load 4 ref pixels
+ smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r8, r5, r6 ; calculate difference
+ sel r7, r8, lr ; select bytes with positive difference
+ usub8 r9, r6, r5 ; calculate difference with reversed operands
+ sel r8, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r5, r7, lr ; calculate sum of positive differences
+ usad8 r6, r8, lr ; calculate sum of negative differences
+ orr r8, r8, r7 ; differences of all 4 pixels
+ ldr r5, [r0, #0x8] ; load 4 src pixels
+ ; calculate sse
+ uxtb16 r6, r8 ; byte (two pixels) to halfwords
+ uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
+ smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
+
+ ; 3rd 4 pixels
+ ldr r6, [r2, #0x8] ; load 4 ref pixels
+ smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r8, r5, r6 ; calculate difference
+ sel r7, r8, lr ; select bytes with positive difference
+ usub8 r9, r6, r5 ; calculate difference with reversed operands
+ sel r8, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r5, r7, lr ; calculate sum of positive differences
+ usad8 r6, r8, lr ; calculate sum of negative differences
+ orr r8, r8, r7 ; differences of all 4 pixels
+
+ ldr r5, [r0, #0xc] ; load 4 src pixels
+
+ ; calculate sse
+ uxtb16 r6, r8 ; byte (two pixels) to halfwords
+ uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
+ smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
+
+ ; 4th 4 pixels
+ ldr r6, [r2, #0xc] ; load 4 ref pixels
+ smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
+
+ usub8 r8, r5, r6 ; calculate difference
+ add r0, r0, r1 ; set src_ptr to next row
+ sel r7, r8, lr ; select bytes with positive difference
+ usub8 r9, r6, r5 ; calculate difference with reversed operands
+ add r2, r2, r3 ; set dst_ptr to next row
+ sel r8, r9, lr ; select bytes with negative difference
+
+ ; calculate partial sums
+ usad8 r5, r7, lr ; calculate sum of positive differences
+ usad8 r6, r8, lr ; calculate sum of negative differences
+ orr r8, r8, r7 ; differences of all 4 pixels
+
+ subs r12, r12, #1 ; next row
+
+ ; calculate sse
+ uxtb16 r6, r8 ; byte (two pixels) to halfwords
+ uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
+ smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
+ smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
+
+ bne loop
+
+ ; return stuff
+ ldr r1, [sp, #28] ; get address of sse
+ mov r0, r4 ; return sse
+ str r4, [r1] ; store sse
+
+ pop {r4-r9, pc}
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/encoder/arm/armv6/vp8_short_fdct4x4_armv6.asm b/libvpx/vp8/encoder/arm/armv6/vp8_short_fdct4x4_armv6.asm
new file mode 100644
index 0000000..8034c1d
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv6/vp8_short_fdct4x4_armv6.asm
@@ -0,0 +1,262 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+ EXPORT |vp8_short_fdct4x4_armv6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA |.text|, CODE, READONLY
+; void vp8_short_fdct4x4_c(short *input, short *output, int pitch)
+|vp8_short_fdct4x4_armv6| PROC
+
+ stmfd sp!, {r4 - r12, lr}
+
+ ; PART 1
+
+ ; coeffs 0-3
+ ldrd r4, r5, [r0] ; [i1 | i0] [i3 | i2]
+
+ ldr r10, c7500
+ ldr r11, c14500
+ ldr r12, c0x22a453a0 ; [2217*4 | 5352*4]
+ ldr lr, c0x00080008
+ ror r5, r5, #16 ; [i2 | i3]
+
+ qadd16 r6, r4, r5 ; [i1+i2 | i0+i3] = [b1 | a1] without shift
+ qsub16 r7, r4, r5 ; [i1-i2 | i0-i3] = [c1 | d1] without shift
+
+ add r0, r0, r2 ; update input pointer
+
+ qadd16 r7, r7, r7 ; 2*[c1|d1] --> we can use smlad and smlsd
+ ; with 2217*4 and 5352*4 without losing the
+ ; sign bit (overflow)
+
+ smuad r4, r6, lr ; o0 = (i1+i2)*8 + (i0+i3)*8
+ smusd r5, r6, lr ; o2 = (i1+i2)*8 - (i0+i3)*8
+
+ smlad r6, r7, r12, r11 ; o1 = (c1 * 2217 + d1 * 5352 + 14500)
+ smlsdx r7, r7, r12, r10 ; o3 = (d1 * 2217 - c1 * 5352 + 7500)
+
+ ldrd r8, r9, [r0] ; [i5 | i4] [i7 | i6]
+
+ pkhbt r3, r4, r6, lsl #4 ; [o1 | o0], keep in register for PART 2
+ pkhbt r6, r5, r7, lsl #4 ; [o3 | o2]
+
+ str r6, [r1, #4]
+
+ ; coeffs 4-7
+ ror r9, r9, #16 ; [i6 | i7]
+
+ qadd16 r6, r8, r9 ; [i5+i6 | i4+i7] = [b1 | a1] without shift
+ qsub16 r7, r8, r9 ; [i5-i6 | i4-i7] = [c1 | d1] without shift
+
+ add r0, r0, r2 ; update input pointer
+
+ qadd16 r7, r7, r7 ; 2x[c1|d1] --> we can use smlad and smlsd
+ ; with 2217*4 and 5352*4 without losing the
+ ; sign bit (overflow)
+
+ smuad r9, r6, lr ; o4 = (i5+i6)*8 + (i4+i7)*8
+ smusd r8, r6, lr ; o6 = (i5+i6)*8 - (i4+i7)*8
+
+ smlad r6, r7, r12, r11 ; o5 = (c1 * 2217 + d1 * 5352 + 14500)
+ smlsdx r7, r7, r12, r10 ; o7 = (d1 * 2217 - c1 * 5352 + 7500)
+
+ ldrd r4, r5, [r0] ; [i9 | i8] [i11 | i10]
+
+ pkhbt r9, r9, r6, lsl #4 ; [o5 | o4], keep in register for PART 2
+ pkhbt r6, r8, r7, lsl #4 ; [o7 | o6]
+
+ str r6, [r1, #12]
+
+ ; coeffs 8-11
+ ror r5, r5, #16 ; [i10 | i11]
+
+ qadd16 r6, r4, r5 ; [i9+i10 | i8+i11]=[b1 | a1] without shift
+ qsub16 r7, r4, r5 ; [i9-i10 | i8-i11]=[c1 | d1] without shift
+
+ add r0, r0, r2 ; update input pointer
+
+ qadd16 r7, r7, r7 ; 2x[c1|d1] --> we can use smlad and smlsd
+ ; with 2217*4 and 5352*4 without losing the
+ ; sign bit (overflow)
+
+ smuad r2, r6, lr ; o8 = (i9+i10)*8 + (i8+i11)*8
+ smusd r8, r6, lr ; o10 = (i9+i10)*8 - (i8+i11)*8
+
+ smlad r6, r7, r12, r11 ; o9 = (c1 * 2217 + d1 * 5352 + 14500)
+ smlsdx r7, r7, r12, r10 ; o11 = (d1 * 2217 - c1 * 5352 + 7500)
+
+ ldrd r4, r5, [r0] ; [i13 | i12] [i15 | i14]
+
+ pkhbt r2, r2, r6, lsl #4 ; [o9 | o8], keep in register for PART 2
+ pkhbt r6, r8, r7, lsl #4 ; [o11 | o10]
+
+ str r6, [r1, #20]
+
+ ; coeffs 12-15
+ ror r5, r5, #16 ; [i14 | i15]
+
+ qadd16 r6, r4, r5 ; [i13+i14 | i12+i15]=[b1|a1] without shift
+ qsub16 r7, r4, r5 ; [i13-i14 | i12-i15]=[c1|d1] without shift
+
+ qadd16 r7, r7, r7 ; 2x[c1|d1] --> we can use smlad and smlsd
+ ; with 2217*4 and 5352*4 without losing the
+ ; sign bit (overflow)
+
+ smuad r4, r6, lr ; o12 = (i13+i14)*8 + (i12+i15)*8
+ smusd r5, r6, lr ; o14 = (i13+i14)*8 - (i12+i15)*8
+
+ smlad r6, r7, r12, r11 ; o13 = (c1 * 2217 + d1 * 5352 + 14500)
+ smlsdx r7, r7, r12, r10 ; o15 = (d1 * 2217 - c1 * 5352 + 7500)
+
+ pkhbt r0, r4, r6, lsl #4 ; [o13 | o12], keep in register for PART 2
+ pkhbt r6, r5, r7, lsl #4 ; [o15 | o14]
+
+ str r6, [r1, #28]
+
+
+ ; PART 2 -------------------------------------------------
+ ldr r11, c12000
+ ldr r10, c51000
+ ldr lr, c0x00070007
+
+ qadd16 r4, r3, r0 ; a1 = [i1+i13 | i0+i12]
+ qadd16 r5, r9, r2 ; b1 = [i5+i9 | i4+i8]
+ qsub16 r6, r9, r2 ; c1 = [i5-i9 | i4-i8]
+ qsub16 r7, r3, r0 ; d1 = [i1-i13 | i0-i12]
+
+ qadd16 r4, r4, lr ; a1 + 7
+
+ add r0, r11, #0x10000 ; add (d!=0)
+
+ qadd16 r2, r4, r5 ; a1 + b1 + 7
+ qsub16 r3, r4, r5 ; a1 - b1 + 7
+
+ ldr r12, c0x08a914e8 ; [2217 | 5352]
+
+ lsl r8, r2, #16 ; prepare bottom halfword for scaling
+ asr r2, r2, #4 ; scale top halfword
+ lsl r9, r3, #16 ; prepare bottom halfword for scaling
+ asr r3, r3, #4 ; scale top halfword
+ pkhtb r4, r2, r8, asr #20 ; pack and scale bottom halfword
+ pkhtb r5, r3, r9, asr #20 ; pack and scale bottom halfword
+
+ smulbt r2, r6, r12 ; [ ------ | c1*2217]
+ str r4, [r1, #0] ; [ o1 | o0]
+ smultt r3, r6, r12 ; [c1*2217 | ------ ]
+ str r5, [r1, #16] ; [ o9 | o8]
+
+ smlabb r8, r7, r12, r2 ; [ ------ | d1*5352]
+ smlatb r9, r7, r12, r3 ; [d1*5352 | ------ ]
+
+ smulbb r2, r6, r12 ; [ ------ | c1*5352]
+ smultb r3, r6, r12 ; [c1*5352 | ------ ]
+
+ lsls r6, r7, #16 ; d1 != 0 ?
+ addeq r8, r8, r11 ; c1_b*2217+d1_b*5352+12000 + (d==0)
+ addne r8, r8, r0 ; c1_b*2217+d1_b*5352+12000 + (d!=0)
+ asrs r6, r7, #16
+ addeq r9, r9, r11 ; c1_t*2217+d1_t*5352+12000 + (d==0)
+ addne r9, r9, r0 ; c1_t*2217+d1_t*5352+12000 + (d!=0)
+
+ smlabt r4, r7, r12, r10 ; [ ------ | d1*2217] + 51000
+ smlatt r5, r7, r12, r10 ; [d1*2217 | ------ ] + 51000
+
+ pkhtb r9, r9, r8, asr #16
+
+ sub r4, r4, r2
+ sub r5, r5, r3
+
+ ldr r3, [r1, #4] ; [i3 | i2]
+
+ pkhtb r5, r5, r4, asr #16 ; [o13|o12]
+
+ str r9, [r1, #8] ; [o5 | 04]
+
+ ldr r9, [r1, #12] ; [i7 | i6]
+ ldr r8, [r1, #28] ; [i15|i14]
+ ldr r2, [r1, #20] ; [i11|i10]
+ str r5, [r1, #24] ; [o13|o12]
+
+ qadd16 r4, r3, r8 ; a1 = [i3+i15 | i2+i14]
+ qadd16 r5, r9, r2 ; b1 = [i7+i11 | i6+i10]
+
+ qadd16 r4, r4, lr ; a1 + 7
+
+ qsub16 r6, r9, r2 ; c1 = [i7-i11 | i6-i10]
+ qadd16 r2, r4, r5 ; a1 + b1 + 7
+ qsub16 r7, r3, r8 ; d1 = [i3-i15 | i2-i14]
+ qsub16 r3, r4, r5 ; a1 - b1 + 7
+
+ lsl r8, r2, #16 ; prepare bottom halfword for scaling
+ asr r2, r2, #4 ; scale top halfword
+ lsl r9, r3, #16 ; prepare bottom halfword for scaling
+ asr r3, r3, #4 ; scale top halfword
+ pkhtb r4, r2, r8, asr #20 ; pack and scale bottom halfword
+ pkhtb r5, r3, r9, asr #20 ; pack and scale bottom halfword
+
+ smulbt r2, r6, r12 ; [ ------ | c1*2217]
+ str r4, [r1, #4] ; [ o3 | o2]
+ smultt r3, r6, r12 ; [c1*2217 | ------ ]
+ str r5, [r1, #20] ; [ o11 | o10]
+
+ smlabb r8, r7, r12, r2 ; [ ------ | d1*5352]
+ smlatb r9, r7, r12, r3 ; [d1*5352 | ------ ]
+
+ smulbb r2, r6, r12 ; [ ------ | c1*5352]
+ smultb r3, r6, r12 ; [c1*5352 | ------ ]
+
+ lsls r6, r7, #16 ; d1 != 0 ?
+ addeq r8, r8, r11 ; c1_b*2217+d1_b*5352+12000 + (d==0)
+ addne r8, r8, r0 ; c1_b*2217+d1_b*5352+12000 + (d!=0)
+
+ asrs r6, r7, #16
+ addeq r9, r9, r11 ; c1_t*2217+d1_t*5352+12000 + (d==0)
+ addne r9, r9, r0 ; c1_t*2217+d1_t*5352+12000 + (d!=0)
+
+ smlabt r4, r7, r12, r10 ; [ ------ | d1*2217] + 51000
+ smlatt r5, r7, r12, r10 ; [d1*2217 | ------ ] + 51000
+
+ pkhtb r9, r9, r8, asr #16
+
+ sub r4, r4, r2
+ sub r5, r5, r3
+
+ str r9, [r1, #12] ; [o7 | o6]
+ pkhtb r5, r5, r4, asr #16 ; [o15|o14]
+
+ str r5, [r1, #28] ; [o15|o14]
+
+ ldmfd sp!, {r4 - r12, pc}
+
+ ENDP
+
+; Used constants
+c7500
+ DCD 7500
+c14500
+ DCD 14500
+c0x22a453a0
+ DCD 0x22a453a0
+c0x00080008
+ DCD 0x00080008
+c12000
+ DCD 12000
+c51000
+ DCD 51000
+c0x00070007
+ DCD 0x00070007
+c0x08a914e8
+ DCD 0x08a914e8
+
+ END
diff --git a/libvpx/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm b/libvpx/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm
new file mode 100644
index 0000000..f329f8f
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm
@@ -0,0 +1,272 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_subtract_mby_armv6|
+ EXPORT |vp8_subtract_mbuv_armv6|
+ EXPORT |vp8_subtract_b_armv6|
+
+ INCLUDE asm_enc_offsets.asm
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+; r0 BLOCK *be
+; r1 BLOCKD *bd
+; r2 int pitch
+|vp8_subtract_b_armv6| PROC
+
+ stmfd sp!, {r4-r9}
+
+ ldr r4, [r0, #vp8_block_base_src]
+ ldr r5, [r0, #vp8_block_src]
+ ldr r6, [r0, #vp8_block_src_diff]
+
+ ldr r3, [r4]
+ ldr r7, [r0, #vp8_block_src_stride]
+ add r3, r3, r5 ; src = *base_src + src
+ ldr r8, [r1, #vp8_blockd_predictor]
+
+ mov r9, #4 ; loop count
+
+loop_block
+
+ ldr r0, [r3], r7 ; src
+ ldr r1, [r8], r2 ; pred
+
+ uxtb16 r4, r0 ; [s2 | s0]
+ uxtb16 r5, r1 ; [p2 | p0]
+ uxtb16 r0, r0, ror #8 ; [s3 | s1]
+ uxtb16 r1, r1, ror #8 ; [p3 | p1]
+
+ usub16 r4, r4, r5 ; [d2 | d0]
+ usub16 r5, r0, r1 ; [d3 | d1]
+
+ subs r9, r9, #1 ; decrement loop counter
+
+ pkhbt r0, r4, r5, lsl #16 ; [d1 | d0]
+ pkhtb r1, r5, r4, asr #16 ; [d3 | d2]
+
+ str r0, [r6, #0] ; diff
+ str r1, [r6, #4] ; diff
+
+ add r6, r6, r2, lsl #1 ; update diff pointer
+ bne loop_block
+
+ ldmfd sp!, {r4-r9}
+ mov pc, lr
+
+ ENDP
+
+
+; r0 short *diff
+; r1 unsigned char *usrc
+; r2 unsigned char *vsrc
+; r3 int src_stride
+; sp unsigned char *upred
+; sp unsigned char *vpred
+; sp int pred_stride
+|vp8_subtract_mbuv_armv6| PROC
+
+ stmfd sp!, {r4-r11}
+
+ add r0, r0, #512 ; set *diff point to Cb
+ mov r4, #8 ; loop count
+ ldr r5, [sp, #32] ; upred
+ ldr r12, [sp, #40] ; pred_stride
+
+ ; Subtract U block
+loop_u
+ ldr r6, [r1] ; usrc (A)
+ ldr r7, [r5] ; upred (A)
+
+ uxtb16 r8, r6 ; [s2 | s0] (A)
+ uxtb16 r9, r7 ; [p2 | p0] (A)
+ uxtb16 r10, r6, ror #8 ; [s3 | s1] (A)
+ uxtb16 r11, r7, ror #8 ; [p3 | p1] (A)
+
+ usub16 r6, r8, r9 ; [d2 | d0] (A)
+ usub16 r7, r10, r11 ; [d3 | d1] (A)
+
+ ldr r10, [r1, #4] ; usrc (B)
+ ldr r11, [r5, #4] ; upred (B)
+
+ pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (A)
+ pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (A)
+
+ str r8, [r0], #4 ; diff (A)
+ uxtb16 r8, r10 ; [s2 | s0] (B)
+ str r9, [r0], #4 ; diff (A)
+
+ uxtb16 r9, r11 ; [p2 | p0] (B)
+ uxtb16 r10, r10, ror #8 ; [s3 | s1] (B)
+ uxtb16 r11, r11, ror #8 ; [p3 | p1] (B)
+
+ usub16 r6, r8, r9 ; [d2 | d0] (B)
+ usub16 r7, r10, r11 ; [d3 | d1] (B)
+
+ add r1, r1, r3 ; update usrc pointer
+ add r5, r5, r12 ; update upred pointer
+
+ pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (B)
+ pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (B)
+
+ str r8, [r0], #4 ; diff (B)
+ subs r4, r4, #1 ; update loop counter
+ str r9, [r0], #4 ; diff (B)
+
+ bne loop_u
+
+ ldr r5, [sp, #36] ; vpred
+ mov r4, #8 ; loop count
+
+ ; Subtract V block
+loop_v
+ ldr r6, [r2] ; vsrc (A)
+ ldr r7, [r5] ; vpred (A)
+
+ uxtb16 r8, r6 ; [s2 | s0] (A)
+ uxtb16 r9, r7 ; [p2 | p0] (A)
+ uxtb16 r10, r6, ror #8 ; [s3 | s1] (A)
+ uxtb16 r11, r7, ror #8 ; [p3 | p1] (A)
+
+ usub16 r6, r8, r9 ; [d2 | d0] (A)
+ usub16 r7, r10, r11 ; [d3 | d1] (A)
+
+ ldr r10, [r2, #4] ; vsrc (B)
+ ldr r11, [r5, #4] ; vpred (B)
+
+ pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (A)
+ pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (A)
+
+ str r8, [r0], #4 ; diff (A)
+ uxtb16 r8, r10 ; [s2 | s0] (B)
+ str r9, [r0], #4 ; diff (A)
+
+ uxtb16 r9, r11 ; [p2 | p0] (B)
+ uxtb16 r10, r10, ror #8 ; [s3 | s1] (B)
+ uxtb16 r11, r11, ror #8 ; [p3 | p1] (B)
+
+ usub16 r6, r8, r9 ; [d2 | d0] (B)
+ usub16 r7, r10, r11 ; [d3 | d1] (B)
+
+ add r2, r2, r3 ; update vsrc pointer
+ add r5, r5, r12 ; update vpred pointer
+
+ pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (B)
+ pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (B)
+
+ str r8, [r0], #4 ; diff (B)
+ subs r4, r4, #1 ; update loop counter
+ str r9, [r0], #4 ; diff (B)
+
+ bne loop_v
+
+ ldmfd sp!, {r4-r11}
+ bx lr
+
+ ENDP
+
+
+; r0 short *diff
+; r1 unsigned char *src
+; r2 int src_stride
+; r3 unsigned char *pred
+; sp int pred_stride
+|vp8_subtract_mby_armv6| PROC
+
+ stmfd sp!, {r4-r11}
+ ldr r12, [sp, #32] ; pred_stride
+ mov r4, #16
+loop
+ ldr r6, [r1] ; src (A)
+ ldr r7, [r3] ; pred (A)
+
+ uxtb16 r8, r6 ; [s2 | s0] (A)
+ uxtb16 r9, r7 ; [p2 | p0] (A)
+ uxtb16 r10, r6, ror #8 ; [s3 | s1] (A)
+ uxtb16 r11, r7, ror #8 ; [p3 | p1] (A)
+
+ usub16 r6, r8, r9 ; [d2 | d0] (A)
+ usub16 r7, r10, r11 ; [d3 | d1] (A)
+
+ ldr r10, [r1, #4] ; src (B)
+ ldr r11, [r3, #4] ; pred (B)
+
+ pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (A)
+ pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (A)
+
+ str r8, [r0], #4 ; diff (A)
+ uxtb16 r8, r10 ; [s2 | s0] (B)
+ str r9, [r0], #4 ; diff (A)
+
+ uxtb16 r9, r11 ; [p2 | p0] (B)
+ uxtb16 r10, r10, ror #8 ; [s3 | s1] (B)
+ uxtb16 r11, r11, ror #8 ; [p3 | p1] (B)
+
+ usub16 r6, r8, r9 ; [d2 | d0] (B)
+ usub16 r7, r10, r11 ; [d3 | d1] (B)
+
+ ldr r10, [r1, #8] ; src (C)
+ ldr r11, [r3, #8] ; pred (C)
+
+ pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (B)
+ pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (B)
+
+ str r8, [r0], #4 ; diff (B)
+ uxtb16 r8, r10 ; [s2 | s0] (C)
+ str r9, [r0], #4 ; diff (B)
+
+ uxtb16 r9, r11 ; [p2 | p0] (C)
+ uxtb16 r10, r10, ror #8 ; [s3 | s1] (C)
+ uxtb16 r11, r11, ror #8 ; [p3 | p1] (C)
+
+ usub16 r6, r8, r9 ; [d2 | d0] (C)
+ usub16 r7, r10, r11 ; [d3 | d1] (C)
+
+ ldr r10, [r1, #12] ; src (D)
+ ldr r11, [r3, #12] ; pred (D)
+
+ pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (C)
+ pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (C)
+
+ str r8, [r0], #4 ; diff (C)
+ uxtb16 r8, r10 ; [s2 | s0] (D)
+ str r9, [r0], #4 ; diff (C)
+
+ uxtb16 r9, r11 ; [p2 | p0] (D)
+ uxtb16 r10, r10, ror #8 ; [s3 | s1] (D)
+ uxtb16 r11, r11, ror #8 ; [p3 | p1] (D)
+
+ usub16 r6, r8, r9 ; [d2 | d0] (D)
+ usub16 r7, r10, r11 ; [d3 | d1] (D)
+
+ add r1, r1, r2 ; update src pointer
+ add r3, r3, r12 ; update pred pointer
+
+ pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (D)
+ pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (D)
+
+ str r8, [r0], #4 ; diff (D)
+ subs r4, r4, #1 ; update loop counter
+ str r9, [r0], #4 ; diff (D)
+
+ bne loop
+
+ ldmfd sp!, {r4-r11}
+ bx lr
+
+ ENDP
+
+ END
+
diff --git a/libvpx/vp8/encoder/arm/armv6/walsh_v6.asm b/libvpx/vp8/encoder/arm/armv6/walsh_v6.asm
new file mode 100644
index 0000000..5eaf3f2
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/armv6/walsh_v6.asm
@@ -0,0 +1,212 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+ EXPORT |vp8_short_walsh4x4_armv6|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA |.text|, CODE, READONLY ; name this block of code
+
+;short vp8_short_walsh4x4_armv6(short *input, short *output, int pitch)
+; r0 short *input,
+; r1 short *output,
+; r2 int pitch
+|vp8_short_walsh4x4_armv6| PROC
+
+ stmdb sp!, {r4 - r11, lr}
+
+ ldrd r4, r5, [r0], r2
+ ldr lr, c00040004
+ ldrd r6, r7, [r0], r2
+
+ ; 0-3
+ qadd16 r3, r4, r5 ; [d1|a1] [1+3 | 0+2]
+ qsub16 r4, r4, r5 ; [c1|b1] [1-3 | 0-2]
+
+ ldrd r8, r9, [r0], r2
+ ; 4-7
+ qadd16 r5, r6, r7 ; [d1|a1] [5+7 | 4+6]
+ qsub16 r6, r6, r7 ; [c1|b1] [5-7 | 4-6]
+
+ ldrd r10, r11, [r0]
+ ; 8-11
+ qadd16 r7, r8, r9 ; [d1|a1] [9+11 | 8+10]
+ qsub16 r8, r8, r9 ; [c1|b1] [9-11 | 8-10]
+
+ ; 12-15
+ qadd16 r9, r10, r11 ; [d1|a1] [13+15 | 12+14]
+ qsub16 r10, r10, r11 ; [c1|b1] [13-15 | 12-14]
+
+
+ lsls r2, r3, #16
+ smuad r11, r3, lr ; A0 = a1<<2 + d1<<2
+ addne r11, r11, #1 ; A0 += (a1!=0)
+
+ lsls r2, r7, #16
+ smuad r12, r7, lr ; C0 = a1<<2 + d1<<2
+ addne r12, r12, #1 ; C0 += (a1!=0)
+
+ add r0, r11, r12 ; a1_0 = A0 + C0
+ sub r11, r11, r12 ; b1_0 = A0 - C0
+
+ lsls r2, r5, #16
+ smuad r12, r5, lr ; B0 = a1<<2 + d1<<2
+ addne r12, r12, #1 ; B0 += (a1!=0)
+
+ lsls r2, r9, #16
+ smuad r2, r9, lr ; D0 = a1<<2 + d1<<2
+ addne r2, r2, #1 ; D0 += (a1!=0)
+
+ add lr, r12, r2 ; d1_0 = B0 + D0
+ sub r12, r12, r2 ; c1_0 = B0 - D0
+
+ ; op[0,4,8,12]
+ adds r2, r0, lr ; a2 = a1_0 + d1_0
+ addmi r2, r2, #1 ; += a2 < 0
+ add r2, r2, #3 ; += 3
+ subs r0, r0, lr ; d2 = a1_0 - d1_0
+ mov r2, r2, asr #3 ; >> 3
+ strh r2, [r1] ; op[0]
+
+ addmi r0, r0, #1 ; += a2 < 0
+ add r0, r0, #3 ; += 3
+ ldr lr, c00040004
+ mov r0, r0, asr #3 ; >> 3
+ strh r0, [r1, #24] ; op[12]
+
+ adds r2, r11, r12 ; b2 = b1_0 + c1_0
+ addmi r2, r2, #1 ; += a2 < 0
+ add r2, r2, #3 ; += 3
+ subs r0, r11, r12 ; c2 = b1_0 - c1_0
+ mov r2, r2, asr #3 ; >> 3
+ strh r2, [r1, #8] ; op[4]
+
+ addmi r0, r0, #1 ; += a2 < 0
+ add r0, r0, #3 ; += 3
+ smusd r3, r3, lr ; A3 = a1<<2 - d1<<2
+ smusd r7, r7, lr ; C3 = a1<<2 - d1<<2
+ mov r0, r0, asr #3 ; >> 3
+ strh r0, [r1, #16] ; op[8]
+
+
+ ; op[3,7,11,15]
+ add r0, r3, r7 ; a1_3 = A3 + C3
+ sub r3, r3, r7 ; b1_3 = A3 - C3
+
+ smusd r5, r5, lr ; B3 = a1<<2 - d1<<2
+ smusd r9, r9, lr ; D3 = a1<<2 - d1<<2
+ add r7, r5, r9 ; d1_3 = B3 + D3
+ sub r5, r5, r9 ; c1_3 = B3 - D3
+
+ adds r2, r0, r7 ; a2 = a1_3 + d1_3
+ addmi r2, r2, #1 ; += a2 < 0
+ add r2, r2, #3 ; += 3
+ adds r9, r3, r5 ; b2 = b1_3 + c1_3
+ mov r2, r2, asr #3 ; >> 3
+ strh r2, [r1, #6] ; op[3]
+
+ addmi r9, r9, #1 ; += a2 < 0
+ add r9, r9, #3 ; += 3
+ subs r2, r3, r5 ; c2 = b1_3 - c1_3
+ mov r9, r9, asr #3 ; >> 3
+ strh r9, [r1, #14] ; op[7]
+
+ addmi r2, r2, #1 ; += a2 < 0
+ add r2, r2, #3 ; += 3
+ subs r9, r0, r7 ; d2 = a1_3 - d1_3
+ mov r2, r2, asr #3 ; >> 3
+ strh r2, [r1, #22] ; op[11]
+
+ addmi r9, r9, #1 ; += a2 < 0
+ add r9, r9, #3 ; += 3
+ smuad r3, r4, lr ; A1 = b1<<2 + c1<<2
+ smuad r5, r8, lr ; C1 = b1<<2 + c1<<2
+ mov r9, r9, asr #3 ; >> 3
+ strh r9, [r1, #30] ; op[15]
+
+ ; op[1,5,9,13]
+ add r0, r3, r5 ; a1_1 = A1 + C1
+ sub r3, r3, r5 ; b1_1 = A1 - C1
+
+ smuad r7, r6, lr ; B1 = b1<<2 + c1<<2
+ smuad r9, r10, lr ; D1 = b1<<2 + c1<<2
+ add r5, r7, r9 ; d1_1 = B1 + D1
+ sub r7, r7, r9 ; c1_1 = B1 - D1
+
+ adds r2, r0, r5 ; a2 = a1_1 + d1_1
+ addmi r2, r2, #1 ; += a2 < 0
+ add r2, r2, #3 ; += 3
+ adds r9, r3, r7 ; b2 = b1_1 + c1_1
+ mov r2, r2, asr #3 ; >> 3
+ strh r2, [r1, #2] ; op[1]
+
+ addmi r9, r9, #1 ; += a2 < 0
+ add r9, r9, #3 ; += 3
+ subs r2, r3, r7 ; c2 = b1_1 - c1_1
+ mov r9, r9, asr #3 ; >> 3
+ strh r9, [r1, #10] ; op[5]
+
+ addmi r2, r2, #1 ; += a2 < 0
+ add r2, r2, #3 ; += 3
+ subs r9, r0, r5 ; d2 = a1_1 - d1_1
+ mov r2, r2, asr #3 ; >> 3
+ strh r2, [r1, #18] ; op[9]
+
+ addmi r9, r9, #1 ; += a2 < 0
+ add r9, r9, #3 ; += 3
+ smusd r4, r4, lr ; A2 = b1<<2 - c1<<2
+ smusd r8, r8, lr ; C2 = b1<<2 - c1<<2
+ mov r9, r9, asr #3 ; >> 3
+ strh r9, [r1, #26] ; op[13]
+
+
+ ; op[2,6,10,14]
+ add r11, r4, r8 ; a1_2 = A2 + C2
+ sub r12, r4, r8 ; b1_2 = A2 - C2
+
+ smusd r6, r6, lr ; B2 = b1<<2 - c1<<2
+ smusd r10, r10, lr ; D2 = b1<<2 - c1<<2
+ add r4, r6, r10 ; d1_2 = B2 + D2
+ sub r8, r6, r10 ; c1_2 = B2 - D2
+
+ adds r2, r11, r4 ; a2 = a1_2 + d1_2
+ addmi r2, r2, #1 ; += a2 < 0
+ add r2, r2, #3 ; += 3
+ adds r9, r12, r8 ; b2 = b1_2 + c1_2
+ mov r2, r2, asr #3 ; >> 3
+ strh r2, [r1, #4] ; op[2]
+
+ addmi r9, r9, #1 ; += a2 < 0
+ add r9, r9, #3 ; += 3
+ subs r2, r12, r8 ; c2 = b1_2 - c1_2
+ mov r9, r9, asr #3 ; >> 3
+ strh r9, [r1, #12] ; op[6]
+
+ addmi r2, r2, #1 ; += a2 < 0
+ add r2, r2, #3 ; += 3
+ subs r9, r11, r4 ; d2 = a1_2 - d1_2
+ mov r2, r2, asr #3 ; >> 3
+ strh r2, [r1, #20] ; op[10]
+
+ addmi r9, r9, #1 ; += a2 < 0
+ add r9, r9, #3 ; += 3
+ mov r9, r9, asr #3 ; >> 3
+ strh r9, [r1, #28] ; op[14]
+
+
+ ldmia sp!, {r4 - r11, pc}
+ ENDP ; |vp8_short_walsh4x4_armv6|
+
+c00040004
+ DCD 0x00040004
+
+ END
diff --git a/libvpx/vp8/encoder/arm/boolhuff_arm.c b/libvpx/vp8/encoder/arm/boolhuff_arm.c
new file mode 100644
index 0000000..17a941b
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/boolhuff_arm.c
@@ -0,0 +1,41 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/encoder/boolhuff.h"
+#include "vpx/internal/vpx_codec_internal.h"
+
+const unsigned int vp8_prob_cost[256] =
+{
+ 2047, 2047, 1791, 1641, 1535, 1452, 1385, 1328, 1279, 1235, 1196, 1161, 1129, 1099, 1072, 1046,
+ 1023, 1000, 979, 959, 940, 922, 905, 889, 873, 858, 843, 829, 816, 803, 790, 778,
+ 767, 755, 744, 733, 723, 713, 703, 693, 684, 675, 666, 657, 649, 641, 633, 625,
+ 617, 609, 602, 594, 587, 580, 573, 567, 560, 553, 547, 541, 534, 528, 522, 516,
+ 511, 505, 499, 494, 488, 483, 477, 472, 467, 462, 457, 452, 447, 442, 437, 433,
+ 428, 424, 419, 415, 410, 406, 401, 397, 393, 389, 385, 381, 377, 373, 369, 365,
+ 361, 357, 353, 349, 346, 342, 338, 335, 331, 328, 324, 321, 317, 314, 311, 307,
+ 304, 301, 297, 294, 291, 288, 285, 281, 278, 275, 272, 269, 266, 263, 260, 257,
+ 255, 252, 249, 246, 243, 240, 238, 235, 232, 229, 227, 224, 221, 219, 216, 214,
+ 211, 208, 206, 203, 201, 198, 196, 194, 191, 189, 186, 184, 181, 179, 177, 174,
+ 172, 170, 168, 165, 163, 161, 159, 156, 154, 152, 150, 148, 145, 143, 141, 139,
+ 137, 135, 133, 131, 129, 127, 125, 123, 121, 119, 117, 115, 113, 111, 109, 107,
+ 105, 103, 101, 99, 97, 95, 93, 92, 90, 88, 86, 84, 82, 81, 79, 77,
+ 75, 73, 72, 70, 68, 66, 65, 63, 61, 60, 58, 56, 55, 53, 51, 50,
+ 48, 46, 45, 43, 41, 40, 38, 37, 35, 33, 32, 30, 29, 27, 25, 24,
+ 22, 21, 19, 18, 16, 15, 13, 12, 10, 9, 7, 6, 4, 3, 1, 1
+};
+
+int vp8_validate_buffer_arm(const unsigned char *start,
+ size_t len,
+ const unsigned char *end,
+ struct vpx_internal_error_info *error)
+{
+ return validate_buffer(start, len, end, error);
+}
diff --git a/libvpx/vp8/encoder/arm/dct_arm.c b/libvpx/vp8/encoder/arm/dct_arm.c
new file mode 100644
index 0000000..af0fb27
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/dct_arm.c
@@ -0,0 +1,22 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+
+#if HAVE_MEDIA
+
+void vp8_short_fdct8x4_armv6(short *input, short *output, int pitch)
+{
+ vp8_short_fdct4x4_armv6(input, output, pitch);
+ vp8_short_fdct4x4_armv6(input + 4, output + 16, pitch);
+}
+
+#endif /* HAVE_MEDIA */
diff --git a/libvpx/vp8/encoder/arm/neon/fastquantizeb_neon.asm b/libvpx/vp8/encoder/arm/neon/fastquantizeb_neon.asm
new file mode 100644
index 0000000..1430588
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/neon/fastquantizeb_neon.asm
@@ -0,0 +1,258 @@
+;
+; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_fast_quantize_b_neon|
+ EXPORT |vp8_fast_quantize_b_pair_neon|
+
+ INCLUDE asm_enc_offsets.asm
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=4
+
+;vp8_fast_quantize_b_pair_neon(BLOCK *b1, BLOCK *b2, BLOCKD *d1, BLOCKD *d2);
+|vp8_fast_quantize_b_pair_neon| PROC
+
+ stmfd sp!, {r4-r9}
+ vstmdb sp!, {q4-q7}
+
+ ldr r4, [r0, #vp8_block_coeff]
+ ldr r5, [r0, #vp8_block_quant_fast]
+ ldr r6, [r0, #vp8_block_round]
+
+ vld1.16 {q0, q1}, [r4@128] ; load z
+
+ ldr r7, [r2, #vp8_blockd_qcoeff]
+
+ vabs.s16 q4, q0 ; calculate x = abs(z)
+ vabs.s16 q5, q1
+
+ ;right shift 15 to get sign, all 0 if it is positive, all 1 if it is negative
+ vshr.s16 q2, q0, #15 ; sz
+ vshr.s16 q3, q1, #15
+
+ vld1.s16 {q6, q7}, [r6@128] ; load round_ptr [0-15]
+ vld1.s16 {q8, q9}, [r5@128] ; load quant_ptr [0-15]
+
+ ldr r4, [r1, #vp8_block_coeff]
+
+ vadd.s16 q4, q6 ; x + Round
+ vadd.s16 q5, q7
+
+ vld1.16 {q0, q1}, [r4@128] ; load z2
+
+ vqdmulh.s16 q4, q8 ; y = ((Round+abs(z)) * Quant) >> 16
+ vqdmulh.s16 q5, q9
+
+ vabs.s16 q10, q0 ; calculate x2 = abs(z_2)
+ vabs.s16 q11, q1
+ vshr.s16 q12, q0, #15 ; sz2
+ vshr.s16 q13, q1, #15
+
+ ;modify data to have its original sign
+ veor.s16 q4, q2 ; y^sz
+ veor.s16 q5, q3
+
+ vadd.s16 q10, q6 ; x2 + Round
+ vadd.s16 q11, q7
+
+ ldr r8, [r2, #vp8_blockd_dequant]
+
+ vqdmulh.s16 q10, q8 ; y2 = ((Round+abs(z)) * Quant) >> 16
+ vqdmulh.s16 q11, q9
+
+ vshr.s16 q4, #1 ; right shift 1 after vqdmulh
+ vshr.s16 q5, #1
+
+ vld1.s16 {q6, q7}, [r8@128] ;load dequant_ptr[i]
+
+ vsub.s16 q4, q2 ; x1=(y^sz)-sz = (y^sz)-(-1) (2's complement)
+ vsub.s16 q5, q3
+
+ vshr.s16 q10, #1 ; right shift 1 after vqdmulh
+ vshr.s16 q11, #1
+
+ ldr r9, [r2, #vp8_blockd_dqcoeff]
+
+ veor.s16 q10, q12 ; y2^sz2
+ veor.s16 q11, q13
+
+ vst1.s16 {q4, q5}, [r7] ; store: qcoeff = x1
+
+
+ vsub.s16 q10, q12 ; x2=(y^sz)-sz = (y^sz)-(-1) (2's complement)
+ vsub.s16 q11, q13
+
+ ldr r6, [r3, #vp8_blockd_qcoeff]
+
+ vmul.s16 q2, q6, q4 ; x * Dequant
+ vmul.s16 q3, q7, q5
+
+ adr r0, inv_zig_zag ; load ptr of inverse zigzag table
+
+ vceq.s16 q8, q8 ; set q8 to all 1
+
+ vst1.s16 {q10, q11}, [r6] ; store: qcoeff = x2
+
+ vmul.s16 q12, q6, q10 ; x2 * Dequant
+ vmul.s16 q13, q7, q11
+
+ vld1.16 {q6, q7}, [r0@128] ; load inverse scan order
+
+ vtst.16 q14, q4, q8 ; now find eob
+ vtst.16 q15, q5, q8 ; non-zero element is set to all 1
+
+ vst1.s16 {q2, q3}, [r9] ; store dqcoeff = x * Dequant
+
+ ldr r7, [r3, #vp8_blockd_dqcoeff]
+
+ vand q0, q6, q14 ; get all valid numbers from scan array
+ vand q1, q7, q15
+
+ vst1.s16 {q12, q13}, [r7] ; store dqcoeff = x * Dequant
+
+ vtst.16 q2, q10, q8 ; now find eob
+ vtst.16 q3, q11, q8 ; non-zero element is set to all 1
+
+ vmax.u16 q0, q0, q1 ; find maximum value in q0, q1
+
+ vand q10, q6, q2 ; get all valid numbers from scan array
+ vand q11, q7, q3
+ vmax.u16 q10, q10, q11 ; find maximum value in q10, q11
+
+ vmax.u16 d0, d0, d1
+ vmax.u16 d20, d20, d21
+ vmovl.u16 q0, d0
+ vmovl.u16 q10, d20
+
+ vmax.u32 d0, d0, d1
+ vmax.u32 d20, d20, d21
+ vpmax.u32 d0, d0, d0
+ vpmax.u32 d20, d20, d20
+
+ ldr r4, [r2, #vp8_blockd_eob]
+ ldr r5, [r3, #vp8_blockd_eob]
+
+ vst1.8 {d0[0]}, [r4] ; store eob
+ vst1.8 {d20[0]}, [r5] ; store eob
+
+ vldmia sp!, {q4-q7}
+ ldmfd sp!, {r4-r9}
+ bx lr
+
+ ENDP
+
+;void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d)
+|vp8_fast_quantize_b_neon| PROC
+
+ stmfd sp!, {r4-r7}
+
+ ldr r3, [r0, #vp8_block_coeff]
+ ldr r4, [r0, #vp8_block_quant_fast]
+ ldr r5, [r0, #vp8_block_round]
+
+ vld1.16 {q0, q1}, [r3@128] ; load z
+ vorr.s16 q14, q0, q1 ; check if all zero (step 1)
+ ldr r6, [r1, #vp8_blockd_qcoeff]
+ ldr r7, [r1, #vp8_blockd_dqcoeff]
+ vorr.s16 d28, d28, d29 ; check if all zero (step 2)
+
+ vabs.s16 q12, q0 ; calculate x = abs(z)
+ vabs.s16 q13, q1
+
+ ;right shift 15 to get sign, all 0 if it is positive, all 1 if it is negative
+ vshr.s16 q2, q0, #15 ; sz
+ vmov r2, r3, d28 ; check if all zero (step 3)
+ vshr.s16 q3, q1, #15
+
+ vld1.s16 {q14, q15}, [r5@128]; load round_ptr [0-15]
+ vld1.s16 {q8, q9}, [r4@128] ; load quant_ptr [0-15]
+
+ vadd.s16 q12, q14 ; x + Round
+ vadd.s16 q13, q15
+
+ adr r0, inv_zig_zag ; load ptr of inverse zigzag table
+
+ vqdmulh.s16 q12, q8 ; y = ((Round+abs(z)) * Quant) >> 16
+ vqdmulh.s16 q13, q9
+
+ vld1.16 {q10, q11}, [r0@128]; load inverse scan order
+
+ vceq.s16 q8, q8 ; set q8 to all 1
+
+ ldr r4, [r1, #vp8_blockd_dequant]
+
+ vshr.s16 q12, #1 ; right shift 1 after vqdmulh
+ vshr.s16 q13, #1
+
+ ldr r5, [r1, #vp8_blockd_eob]
+
+ orr r2, r2, r3 ; check if all zero (step 4)
+ cmp r2, #0 ; check if all zero (step 5)
+ beq zero_output ; check if all zero (step 6)
+
+ ;modify data to have its original sign
+ veor.s16 q12, q2 ; y^sz
+ veor.s16 q13, q3
+
+ vsub.s16 q12, q2 ; x1=(y^sz)-sz = (y^sz)-(-1) (2's complement)
+ vsub.s16 q13, q3
+
+ vld1.s16 {q2, q3}, [r4@128] ; load dequant_ptr[i]
+
+ vtst.16 q14, q12, q8 ; now find eob
+ vtst.16 q15, q13, q8 ; non-zero element is set to all 1
+
+ vst1.s16 {q12, q13}, [r6@128]; store: qcoeff = x1
+
+ vand q10, q10, q14 ; get all valid numbers from scan array
+ vand q11, q11, q15
+
+
+ vmax.u16 q0, q10, q11 ; find maximum value in q0, q1
+ vmax.u16 d0, d0, d1
+ vmovl.u16 q0, d0
+
+ vmul.s16 q2, q12 ; x * Dequant
+ vmul.s16 q3, q13
+
+ vmax.u32 d0, d0, d1
+ vpmax.u32 d0, d0, d0
+
+ vst1.s16 {q2, q3}, [r7@128] ; store dqcoeff = x * Dequant
+
+ vst1.8 {d0[0]}, [r5] ; store eob
+
+ ldmfd sp!, {r4-r7}
+ bx lr
+
+zero_output
+ strb r2, [r5] ; store eob
+ vst1.s16 {q0, q1}, [r6@128] ; qcoeff = 0
+ vst1.s16 {q0, q1}, [r7@128] ; dqcoeff = 0
+
+ ldmfd sp!, {r4-r7}
+ bx lr
+
+ ENDP
+
+; default inverse zigzag table is defined in vp8/common/entropy.c
+ ALIGN 16 ; enable use of @128 bit aligned loads
+inv_zig_zag
+ DCW 0x0001, 0x0002, 0x0006, 0x0007
+ DCW 0x0003, 0x0005, 0x0008, 0x000d
+ DCW 0x0004, 0x0009, 0x000c, 0x000e
+ DCW 0x000a, 0x000b, 0x000f, 0x0010
+
+ END
+
diff --git a/libvpx/vp8/encoder/arm/neon/picklpf_arm.c b/libvpx/vp8/encoder/arm/neon/picklpf_arm.c
new file mode 100644
index 0000000..ec8071e
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/neon/picklpf_arm.c
@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vp8/common/loopfilter.h"
+#include "vpx_scale/yv12config.h"
+
+extern void vp8_memcpy_partial_neon(unsigned char *dst_ptr,
+ unsigned char *src_ptr,
+ int sz);
+
+
+void vp8_yv12_copy_partial_frame_neon(YV12_BUFFER_CONFIG *src_ybc,
+ YV12_BUFFER_CONFIG *dst_ybc)
+{
+ unsigned char *src_y, *dst_y;
+ int yheight;
+ int ystride;
+ int yoffset;
+ int linestocopy;
+
+ yheight = src_ybc->y_height;
+ ystride = src_ybc->y_stride;
+
+ /* number of MB rows to use in partial filtering */
+ linestocopy = (yheight >> 4) / PARTIAL_FRAME_FRACTION;
+ linestocopy = linestocopy ? linestocopy << 4 : 16; /* 16 lines per MB */
+
+ /* Copy extra 4 so that full filter context is available if filtering done
+ * on the copied partial frame and not original. Partial filter does mb
+ * filtering for top row also, which can modify3 pixels above.
+ */
+ linestocopy += 4;
+ /* partial image starts at ~middle of frame (macroblock border) */
+ yoffset = ystride * (((yheight >> 5) * 16) - 4);
+ src_y = src_ybc->y_buffer + yoffset;
+ dst_y = dst_ybc->y_buffer + yoffset;
+
+ vp8_memcpy_partial_neon(dst_y, src_y, ystride * linestocopy);
+}
diff --git a/libvpx/vp8/encoder/arm/neon/shortfdct_neon.asm b/libvpx/vp8/encoder/arm/neon/shortfdct_neon.asm
new file mode 100644
index 0000000..09dd011
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/neon/shortfdct_neon.asm
@@ -0,0 +1,221 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_short_fdct4x4_neon|
+ EXPORT |vp8_short_fdct8x4_neon|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=4
+
+
+ ALIGN 16 ; enable use of @128 bit aligned loads
+coeff
+ DCW 5352, 5352, 5352, 5352
+ DCW 2217, 2217, 2217, 2217
+ DCD 14500, 14500, 14500, 14500
+ DCD 7500, 7500, 7500, 7500
+ DCD 12000, 12000, 12000, 12000
+ DCD 51000, 51000, 51000, 51000
+
+;void vp8_short_fdct4x4_c(short *input, short *output, int pitch)
+|vp8_short_fdct4x4_neon| PROC
+
+ ; Part one
+ vld1.16 {d0}, [r0@64], r2
+ adr r12, coeff
+ vld1.16 {d1}, [r0@64], r2
+ vld1.16 {q8}, [r12@128]! ; d16=5352, d17=2217
+ vld1.16 {d2}, [r0@64], r2
+ vld1.32 {q9, q10}, [r12@128]! ; q9=14500, q10=7500
+ vld1.16 {d3}, [r0@64], r2
+
+ ; transpose d0=ip[0], d1=ip[1], d2=ip[2], d3=ip[3]
+ vtrn.32 d0, d2
+ vtrn.32 d1, d3
+ vld1.32 {q11,q12}, [r12@128] ; q11=12000, q12=51000
+ vtrn.16 d0, d1
+ vtrn.16 d2, d3
+
+ vadd.s16 d4, d0, d3 ; a1 = ip[0] + ip[3]
+ vadd.s16 d5, d1, d2 ; b1 = ip[1] + ip[2]
+ vsub.s16 d6, d1, d2 ; c1 = ip[1] - ip[2]
+ vsub.s16 d7, d0, d3 ; d1 = ip[0] - ip[3]
+
+ vshl.s16 q2, q2, #3 ; (a1, b1) << 3
+ vshl.s16 q3, q3, #3 ; (c1, d1) << 3
+
+ vadd.s16 d0, d4, d5 ; op[0] = a1 + b1
+ vsub.s16 d2, d4, d5 ; op[2] = a1 - b1
+
+ vmlal.s16 q9, d7, d16 ; d1*5352 + 14500
+ vmlal.s16 q10, d7, d17 ; d1*2217 + 7500
+ vmlal.s16 q9, d6, d17 ; c1*2217 + d1*5352 + 14500
+ vmlsl.s16 q10, d6, d16 ; d1*2217 - c1*5352 + 7500
+
+ vshrn.s32 d1, q9, #12 ; op[1] = (c1*2217 + d1*5352 + 14500)>>12
+ vshrn.s32 d3, q10, #12 ; op[3] = (d1*2217 - c1*5352 + 7500)>>12
+
+
+ ; Part two
+
+ ; transpose d0=ip[0], d1=ip[4], d2=ip[8], d3=ip[12]
+ vtrn.32 d0, d2
+ vtrn.32 d1, d3
+ vtrn.16 d0, d1
+ vtrn.16 d2, d3
+
+ vmov.s16 d26, #7
+
+ vadd.s16 d4, d0, d3 ; a1 = ip[0] + ip[12]
+ vadd.s16 d5, d1, d2 ; b1 = ip[4] + ip[8]
+ vsub.s16 d6, d1, d2 ; c1 = ip[4] - ip[8]
+ vadd.s16 d4, d4, d26 ; a1 + 7
+ vsub.s16 d7, d0, d3 ; d1 = ip[0] - ip[12]
+
+ vadd.s16 d0, d4, d5 ; op[0] = a1 + b1 + 7
+ vsub.s16 d2, d4, d5 ; op[8] = a1 - b1 + 7
+
+ vmlal.s16 q11, d7, d16 ; d1*5352 + 12000
+ vmlal.s16 q12, d7, d17 ; d1*2217 + 51000
+
+ vceq.s16 d4, d7, #0
+
+ vshr.s16 d0, d0, #4
+ vshr.s16 d2, d2, #4
+
+ vmlal.s16 q11, d6, d17 ; c1*2217 + d1*5352 + 12000
+ vmlsl.s16 q12, d6, d16 ; d1*2217 - c1*5352 + 51000
+
+ vmvn.s16 d4, d4
+ vshrn.s32 d1, q11, #16 ; op[4] = (c1*2217 + d1*5352 + 12000)>>16
+ vsub.s16 d1, d1, d4 ; op[4] += (d1!=0)
+ vshrn.s32 d3, q12, #16 ; op[12]= (d1*2217 - c1*5352 + 51000)>>16
+
+ vst1.16 {q0, q1}, [r1@128]
+
+ bx lr
+
+ ENDP
+
+;void vp8_short_fdct8x4_c(short *input, short *output, int pitch)
+|vp8_short_fdct8x4_neon| PROC
+
+ ; Part one
+
+ vld1.16 {q0}, [r0@128], r2
+ adr r12, coeff
+ vld1.16 {q1}, [r0@128], r2
+ vld1.16 {q8}, [r12@128]! ; d16=5352, d17=2217
+ vld1.16 {q2}, [r0@128], r2
+ vld1.32 {q9, q10}, [r12@128]! ; q9=14500, q10=7500
+ vld1.16 {q3}, [r0@128], r2
+
+ ; transpose q0=ip[0], q1=ip[1], q2=ip[2], q3=ip[3]
+ vtrn.32 q0, q2 ; [A0|B0]
+ vtrn.32 q1, q3 ; [A1|B1]
+ vtrn.16 q0, q1 ; [A2|B2]
+ vtrn.16 q2, q3 ; [A3|B3]
+
+ vadd.s16 q11, q0, q3 ; a1 = ip[0] + ip[3]
+ vadd.s16 q12, q1, q2 ; b1 = ip[1] + ip[2]
+ vsub.s16 q13, q1, q2 ; c1 = ip[1] - ip[2]
+ vsub.s16 q14, q0, q3 ; d1 = ip[0] - ip[3]
+
+ vshl.s16 q11, q11, #3 ; a1 << 3
+ vshl.s16 q12, q12, #3 ; b1 << 3
+ vshl.s16 q13, q13, #3 ; c1 << 3
+ vshl.s16 q14, q14, #3 ; d1 << 3
+
+ vadd.s16 q0, q11, q12 ; [A0 | B0] = a1 + b1
+ vsub.s16 q2, q11, q12 ; [A2 | B2] = a1 - b1
+
+ vmov.s16 q11, q9 ; 14500
+ vmov.s16 q12, q10 ; 7500
+
+ vmlal.s16 q9, d28, d16 ; A[1] = d1*5352 + 14500
+ vmlal.s16 q10, d28, d17 ; A[3] = d1*2217 + 7500
+ vmlal.s16 q11, d29, d16 ; B[1] = d1*5352 + 14500
+ vmlal.s16 q12, d29, d17 ; B[3] = d1*2217 + 7500
+
+ vmlal.s16 q9, d26, d17 ; A[1] = c1*2217 + d1*5352 + 14500
+ vmlsl.s16 q10, d26, d16 ; A[3] = d1*2217 - c1*5352 + 7500
+ vmlal.s16 q11, d27, d17 ; B[1] = c1*2217 + d1*5352 + 14500
+ vmlsl.s16 q12, d27, d16 ; B[3] = d1*2217 - c1*5352 + 7500
+
+ vshrn.s32 d2, q9, #12 ; A[1] = (c1*2217 + d1*5352 + 14500)>>12
+ vshrn.s32 d6, q10, #12 ; A[3] = (d1*2217 - c1*5352 + 7500)>>12
+ vshrn.s32 d3, q11, #12 ; B[1] = (c1*2217 + d1*5352 + 14500)>>12
+ vshrn.s32 d7, q12, #12 ; B[3] = (d1*2217 - c1*5352 + 7500)>>12
+
+
+ ; Part two
+ vld1.32 {q9,q10}, [r12@128] ; q9=12000, q10=51000
+
+ ; transpose q0=ip[0], q1=ip[4], q2=ip[8], q3=ip[12]
+ vtrn.32 q0, q2 ; q0=[A0 | B0]
+ vtrn.32 q1, q3 ; q1=[A4 | B4]
+ vtrn.16 q0, q1 ; q2=[A8 | B8]
+ vtrn.16 q2, q3 ; q3=[A12|B12]
+
+ vmov.s16 q15, #7
+
+ vadd.s16 q11, q0, q3 ; a1 = ip[0] + ip[12]
+ vadd.s16 q12, q1, q2 ; b1 = ip[4] + ip[8]
+ vadd.s16 q11, q11, q15 ; a1 + 7
+ vsub.s16 q13, q1, q2 ; c1 = ip[4] - ip[8]
+ vsub.s16 q14, q0, q3 ; d1 = ip[0] - ip[12]
+
+ vadd.s16 q0, q11, q12 ; a1 + b1 + 7
+ vsub.s16 q1, q11, q12 ; a1 - b1 + 7
+
+ vmov.s16 q11, q9 ; 12000
+ vmov.s16 q12, q10 ; 51000
+
+ vshr.s16 d0, d0, #4 ; A[0] = (a1 + b1 + 7)>>4
+ vshr.s16 d4, d1, #4 ; B[0] = (a1 + b1 + 7)>>4
+ vshr.s16 d2, d2, #4 ; A[8] = (a1 + b1 + 7)>>4
+ vshr.s16 d6, d3, #4 ; B[8] = (a1 + b1 + 7)>>4
+
+
+ vmlal.s16 q9, d28, d16 ; A[4] = d1*5352 + 12000
+ vmlal.s16 q10, d28, d17 ; A[12] = d1*2217 + 51000
+ vmlal.s16 q11, d29, d16 ; B[4] = d1*5352 + 12000
+ vmlal.s16 q12, d29, d17 ; B[12] = d1*2217 + 51000
+
+ vceq.s16 q14, q14, #0
+
+ vmlal.s16 q9, d26, d17 ; A[4] = c1*2217 + d1*5352 + 12000
+ vmlsl.s16 q10, d26, d16 ; A[12] = d1*2217 - c1*5352 + 51000
+ vmlal.s16 q11, d27, d17 ; B[4] = c1*2217 + d1*5352 + 12000
+ vmlsl.s16 q12, d27, d16 ; B[12] = d1*2217 - c1*5352 + 51000
+
+ vmvn.s16 q14, q14
+
+ vshrn.s32 d1, q9, #16 ; A[4] = (c1*2217 + d1*5352 + 12000)>>16
+ vshrn.s32 d3, q10, #16 ; A[12]= (d1*2217 - c1*5352 + 51000)>>16
+ vsub.s16 d1, d1, d28 ; A[4] += (d1!=0)
+
+ vshrn.s32 d5, q11, #16 ; B[4] = (c1*2217 + d1*5352 + 12000)>>16
+ vshrn.s32 d7, q12, #16 ; B[12]= (d1*2217 - c1*5352 + 51000)>>16
+ vsub.s16 d5, d5, d29 ; B[4] += (d1!=0)
+
+ vst1.16 {q0, q1}, [r1@128]! ; block A
+ vst1.16 {q2, q3}, [r1@128]! ; block B
+
+ bx lr
+
+ ENDP
+
+ END
+
diff --git a/libvpx/vp8/encoder/arm/neon/subtract_neon.asm b/libvpx/vp8/encoder/arm/neon/subtract_neon.asm
new file mode 100644
index 0000000..91a328c
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/neon/subtract_neon.asm
@@ -0,0 +1,199 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+ EXPORT |vp8_subtract_b_neon|
+ EXPORT |vp8_subtract_mby_neon|
+ EXPORT |vp8_subtract_mbuv_neon|
+
+ INCLUDE asm_enc_offsets.asm
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+;void vp8_subtract_b_neon(BLOCK *be, BLOCKD *bd, int pitch)
+|vp8_subtract_b_neon| PROC
+
+ stmfd sp!, {r4-r7}
+
+ ldr r3, [r0, #vp8_block_base_src]
+ ldr r4, [r0, #vp8_block_src]
+ ldr r5, [r0, #vp8_block_src_diff]
+ ldr r3, [r3]
+ ldr r6, [r0, #vp8_block_src_stride]
+ add r3, r3, r4 ; src = *base_src + src
+ ldr r7, [r1, #vp8_blockd_predictor]
+
+ vld1.8 {d0}, [r3], r6 ;load src
+ vld1.8 {d1}, [r7], r2 ;load pred
+ vld1.8 {d2}, [r3], r6
+ vld1.8 {d3}, [r7], r2
+ vld1.8 {d4}, [r3], r6
+ vld1.8 {d5}, [r7], r2
+ vld1.8 {d6}, [r3], r6
+ vld1.8 {d7}, [r7], r2
+
+ vsubl.u8 q10, d0, d1
+ vsubl.u8 q11, d2, d3
+ vsubl.u8 q12, d4, d5
+ vsubl.u8 q13, d6, d7
+
+ mov r2, r2, lsl #1
+
+ vst1.16 {d20}, [r5], r2 ;store diff
+ vst1.16 {d22}, [r5], r2
+ vst1.16 {d24}, [r5], r2
+ vst1.16 {d26}, [r5], r2
+
+ ldmfd sp!, {r4-r7}
+ bx lr
+
+ ENDP
+
+
+;==========================================
+;void vp8_subtract_mby_neon(short *diff, unsigned char *src, int src_stride
+; unsigned char *pred, int pred_stride)
+|vp8_subtract_mby_neon| PROC
+ push {r4-r7}
+ mov r12, #4
+ ldr r4, [sp, #16] ; pred_stride
+ mov r6, #32 ; "diff" stride x2
+ add r5, r0, #16 ; second diff pointer
+
+subtract_mby_loop
+ vld1.8 {q0}, [r1], r2 ;load src
+ vld1.8 {q1}, [r3], r4 ;load pred
+ vld1.8 {q2}, [r1], r2
+ vld1.8 {q3}, [r3], r4
+ vld1.8 {q4}, [r1], r2
+ vld1.8 {q5}, [r3], r4
+ vld1.8 {q6}, [r1], r2
+ vld1.8 {q7}, [r3], r4
+
+ vsubl.u8 q8, d0, d2
+ vsubl.u8 q9, d1, d3
+ vsubl.u8 q10, d4, d6
+ vsubl.u8 q11, d5, d7
+ vsubl.u8 q12, d8, d10
+ vsubl.u8 q13, d9, d11
+ vsubl.u8 q14, d12, d14
+ vsubl.u8 q15, d13, d15
+
+ vst1.16 {q8}, [r0], r6 ;store diff
+ vst1.16 {q9}, [r5], r6
+ vst1.16 {q10}, [r0], r6
+ vst1.16 {q11}, [r5], r6
+ vst1.16 {q12}, [r0], r6
+ vst1.16 {q13}, [r5], r6
+ vst1.16 {q14}, [r0], r6
+ vst1.16 {q15}, [r5], r6
+
+ subs r12, r12, #1
+ bne subtract_mby_loop
+
+ pop {r4-r7}
+ bx lr
+ ENDP
+
+;=================================
+;void vp8_subtract_mbuv_c(short *diff, unsigned char *usrc, unsigned char *vsrc,
+; int src_stride, unsigned char *upred,
+; unsigned char *vpred, int pred_stride)
+
+|vp8_subtract_mbuv_neon| PROC
+ push {r4-r7}
+ ldr r4, [sp, #16] ; upred
+ ldr r5, [sp, #20] ; vpred
+ ldr r6, [sp, #24] ; pred_stride
+ add r0, r0, #512 ; short *udiff = diff + 256;
+ mov r12, #32 ; "diff" stride x2
+ add r7, r0, #16 ; second diff pointer
+
+;u
+ vld1.8 {d0}, [r1], r3 ;load usrc
+ vld1.8 {d1}, [r4], r6 ;load upred
+ vld1.8 {d2}, [r1], r3
+ vld1.8 {d3}, [r4], r6
+ vld1.8 {d4}, [r1], r3
+ vld1.8 {d5}, [r4], r6
+ vld1.8 {d6}, [r1], r3
+ vld1.8 {d7}, [r4], r6
+ vld1.8 {d8}, [r1], r3
+ vld1.8 {d9}, [r4], r6
+ vld1.8 {d10}, [r1], r3
+ vld1.8 {d11}, [r4], r6
+ vld1.8 {d12}, [r1], r3
+ vld1.8 {d13}, [r4], r6
+ vld1.8 {d14}, [r1], r3
+ vld1.8 {d15}, [r4], r6
+
+ vsubl.u8 q8, d0, d1
+ vsubl.u8 q9, d2, d3
+ vsubl.u8 q10, d4, d5
+ vsubl.u8 q11, d6, d7
+ vsubl.u8 q12, d8, d9
+ vsubl.u8 q13, d10, d11
+ vsubl.u8 q14, d12, d13
+ vsubl.u8 q15, d14, d15
+
+ vst1.16 {q8}, [r0], r12 ;store diff
+ vst1.16 {q9}, [r7], r12
+ vst1.16 {q10}, [r0], r12
+ vst1.16 {q11}, [r7], r12
+ vst1.16 {q12}, [r0], r12
+ vst1.16 {q13}, [r7], r12
+ vst1.16 {q14}, [r0], r12
+ vst1.16 {q15}, [r7], r12
+
+;v
+ vld1.8 {d0}, [r2], r3 ;load vsrc
+ vld1.8 {d1}, [r5], r6 ;load vpred
+ vld1.8 {d2}, [r2], r3
+ vld1.8 {d3}, [r5], r6
+ vld1.8 {d4}, [r2], r3
+ vld1.8 {d5}, [r5], r6
+ vld1.8 {d6}, [r2], r3
+ vld1.8 {d7}, [r5], r6
+ vld1.8 {d8}, [r2], r3
+ vld1.8 {d9}, [r5], r6
+ vld1.8 {d10}, [r2], r3
+ vld1.8 {d11}, [r5], r6
+ vld1.8 {d12}, [r2], r3
+ vld1.8 {d13}, [r5], r6
+ vld1.8 {d14}, [r2], r3
+ vld1.8 {d15}, [r5], r6
+
+ vsubl.u8 q8, d0, d1
+ vsubl.u8 q9, d2, d3
+ vsubl.u8 q10, d4, d5
+ vsubl.u8 q11, d6, d7
+ vsubl.u8 q12, d8, d9
+ vsubl.u8 q13, d10, d11
+ vsubl.u8 q14, d12, d13
+ vsubl.u8 q15, d14, d15
+
+ vst1.16 {q8}, [r0], r12 ;store diff
+ vst1.16 {q9}, [r7], r12
+ vst1.16 {q10}, [r0], r12
+ vst1.16 {q11}, [r7], r12
+ vst1.16 {q12}, [r0], r12
+ vst1.16 {q13}, [r7], r12
+ vst1.16 {q14}, [r0], r12
+ vst1.16 {q15}, [r7], r12
+
+ pop {r4-r7}
+ bx lr
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/encoder/arm/neon/vp8_memcpy_neon.asm b/libvpx/vp8/encoder/arm/neon/vp8_memcpy_neon.asm
new file mode 100644
index 0000000..5b9f11e
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/neon/vp8_memcpy_neon.asm
@@ -0,0 +1,70 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_memcpy_partial_neon|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+;=========================================
+;this is not a full memcpy function!!!
+;void vp8_memcpy_partial_neon(unsigned char *dst_ptr, unsigned char *src_ptr,
+; int sz);
+|vp8_memcpy_partial_neon| PROC
+ ;pld [r1] ;preload pred data
+ ;pld [r1, #128]
+ ;pld [r1, #256]
+ ;pld [r1, #384]
+
+ mov r12, r2, lsr #8 ;copy 256 bytes data at one time
+
+memcpy_neon_loop
+ vld1.8 {q0, q1}, [r1]! ;load src data
+ subs r12, r12, #1
+ vld1.8 {q2, q3}, [r1]!
+ vst1.8 {q0, q1}, [r0]! ;copy to dst_ptr
+ vld1.8 {q4, q5}, [r1]!
+ vst1.8 {q2, q3}, [r0]!
+ vld1.8 {q6, q7}, [r1]!
+ vst1.8 {q4, q5}, [r0]!
+ vld1.8 {q8, q9}, [r1]!
+ vst1.8 {q6, q7}, [r0]!
+ vld1.8 {q10, q11}, [r1]!
+ vst1.8 {q8, q9}, [r0]!
+ vld1.8 {q12, q13}, [r1]!
+ vst1.8 {q10, q11}, [r0]!
+ vld1.8 {q14, q15}, [r1]!
+ vst1.8 {q12, q13}, [r0]!
+ vst1.8 {q14, q15}, [r0]!
+
+ ;pld [r1] ;preload pred data -- need to adjust for real device
+ ;pld [r1, #128]
+ ;pld [r1, #256]
+ ;pld [r1, #384]
+
+ bne memcpy_neon_loop
+
+ ands r3, r2, #0xff ;extra copy
+ beq done_copy_neon_loop
+
+extra_copy_neon_loop
+ vld1.8 {q0}, [r1]! ;load src data
+ subs r3, r3, #16
+ vst1.8 {q0}, [r0]!
+ bne extra_copy_neon_loop
+
+done_copy_neon_loop
+ bx lr
+ ENDP
+
+ END
diff --git a/libvpx/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm b/libvpx/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm
new file mode 100644
index 0000000..55edbf5
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm
@@ -0,0 +1,116 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_mse16x16_neon|
+ EXPORT |vp8_get4x4sse_cs_neon|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+;============================
+; r0 unsigned char *src_ptr
+; r1 int source_stride
+; r2 unsigned char *ref_ptr
+; r3 int recon_stride
+; stack unsigned int *sse
+;note: in this function, sum is never used. So, we can remove this part of calculation
+;from vp8_variance().
+
+|vp8_mse16x16_neon| PROC
+ vmov.i8 q7, #0 ;q7, q8, q9, q10 - sse
+ vmov.i8 q8, #0
+ vmov.i8 q9, #0
+ vmov.i8 q10, #0
+
+ mov r12, #8
+
+mse16x16_neon_loop
+ vld1.8 {q0}, [r0], r1 ;Load up source and reference
+ vld1.8 {q2}, [r2], r3
+ vld1.8 {q1}, [r0], r1
+ vld1.8 {q3}, [r2], r3
+
+ vsubl.u8 q11, d0, d4
+ vsubl.u8 q12, d1, d5
+ vsubl.u8 q13, d2, d6
+ vsubl.u8 q14, d3, d7
+
+ vmlal.s16 q7, d22, d22
+ vmlal.s16 q8, d23, d23
+
+ subs r12, r12, #1
+
+ vmlal.s16 q9, d24, d24
+ vmlal.s16 q10, d25, d25
+ vmlal.s16 q7, d26, d26
+ vmlal.s16 q8, d27, d27
+ vmlal.s16 q9, d28, d28
+ vmlal.s16 q10, d29, d29
+
+ bne mse16x16_neon_loop
+
+ vadd.u32 q7, q7, q8
+ vadd.u32 q9, q9, q10
+
+ ldr r12, [sp] ;load *sse from stack
+
+ vadd.u32 q10, q7, q9
+ vpaddl.u32 q1, q10
+ vadd.u64 d0, d2, d3
+
+ vst1.32 {d0[0]}, [r12]
+ vmov.32 r0, d0[0]
+
+ bx lr
+
+ ENDP
+
+
+;=============================
+; r0 unsigned char *src_ptr,
+; r1 int source_stride,
+; r2 unsigned char *ref_ptr,
+; r3 int recon_stride
+|vp8_get4x4sse_cs_neon| PROC
+ vld1.8 {d0}, [r0], r1 ;Load up source and reference
+ vld1.8 {d4}, [r2], r3
+ vld1.8 {d1}, [r0], r1
+ vld1.8 {d5}, [r2], r3
+ vld1.8 {d2}, [r0], r1
+ vld1.8 {d6}, [r2], r3
+ vld1.8 {d3}, [r0], r1
+ vld1.8 {d7}, [r2], r3
+
+ vsubl.u8 q11, d0, d4
+ vsubl.u8 q12, d1, d5
+ vsubl.u8 q13, d2, d6
+ vsubl.u8 q14, d3, d7
+
+ vmull.s16 q7, d22, d22
+ vmull.s16 q8, d24, d24
+ vmull.s16 q9, d26, d26
+ vmull.s16 q10, d28, d28
+
+ vadd.u32 q7, q7, q8
+ vadd.u32 q9, q9, q10
+ vadd.u32 q9, q7, q9
+
+ vpaddl.u32 q1, q9
+ vadd.u64 d0, d2, d3
+
+ vmov.32 r0, d0[0]
+ bx lr
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm b/libvpx/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm
new file mode 100644
index 0000000..2226629
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm
@@ -0,0 +1,103 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_short_walsh4x4_neon|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+;void vp8_short_walsh4x4_neon(short *input, short *output, int pitch)
+; r0 short *input,
+; r1 short *output,
+; r2 int pitch
+|vp8_short_walsh4x4_neon| PROC
+
+ vld1.16 {d0}, [r0@64], r2 ; load input
+ vld1.16 {d1}, [r0@64], r2
+ vld1.16 {d2}, [r0@64], r2
+ vld1.16 {d3}, [r0@64]
+
+ ;First for-loop
+ ;transpose d0, d1, d2, d3. Then, d0=ip[0], d1=ip[1], d2=ip[2], d3=ip[3]
+ vtrn.32 d0, d2
+ vtrn.32 d1, d3
+
+ vmov.s32 q15, #3 ; add 3 to all values
+
+ vtrn.16 d0, d1
+ vtrn.16 d2, d3
+
+ vadd.s16 d4, d0, d2 ; ip[0] + ip[2]
+ vadd.s16 d5, d1, d3 ; ip[1] + ip[3]
+ vsub.s16 d6, d1, d3 ; ip[1] - ip[3]
+ vsub.s16 d7, d0, d2 ; ip[0] - ip[2]
+
+ vshl.s16 d4, d4, #2 ; a1 = (ip[0] + ip[2]) << 2
+ vshl.s16 d5, d5, #2 ; d1 = (ip[1] + ip[3]) << 2
+ vshl.s16 d6, d6, #2 ; c1 = (ip[1] - ip[3]) << 2
+ vceq.s16 d16, d4, #0 ; a1 == 0
+ vshl.s16 d7, d7, #2 ; b1 = (ip[0] - ip[2]) << 2
+
+ vadd.s16 d0, d4, d5 ; a1 + d1
+ vmvn d16, d16 ; a1 != 0
+ vsub.s16 d3, d4, d5 ; op[3] = a1 - d1
+ vadd.s16 d1, d7, d6 ; op[1] = b1 + c1
+ vsub.s16 d2, d7, d6 ; op[2] = b1 - c1
+ vsub.s16 d0, d0, d16 ; op[0] = a1 + d1 + (a1 != 0)
+
+ ;Second for-loop
+ ;transpose d0, d1, d2, d3, Then, d0=ip[0], d1=ip[4], d2=ip[8], d3=ip[12]
+ vtrn.32 d1, d3
+ vtrn.32 d0, d2
+ vtrn.16 d2, d3
+ vtrn.16 d0, d1
+
+ vaddl.s16 q8, d0, d2 ; a1 = ip[0]+ip[8]
+ vaddl.s16 q9, d1, d3 ; d1 = ip[4]+ip[12]
+ vsubl.s16 q10, d1, d3 ; c1 = ip[4]-ip[12]
+ vsubl.s16 q11, d0, d2 ; b1 = ip[0]-ip[8]
+
+ vadd.s32 q0, q8, q9 ; a2 = a1 + d1
+ vadd.s32 q1, q11, q10 ; b2 = b1 + c1
+ vsub.s32 q2, q11, q10 ; c2 = b1 - c1
+ vsub.s32 q3, q8, q9 ; d2 = a1 - d1
+
+ vclt.s32 q8, q0, #0
+ vclt.s32 q9, q1, #0
+ vclt.s32 q10, q2, #0
+ vclt.s32 q11, q3, #0
+
+ ; subtract -1 (or 0)
+ vsub.s32 q0, q0, q8 ; a2 += a2 < 0
+ vsub.s32 q1, q1, q9 ; b2 += b2 < 0
+ vsub.s32 q2, q2, q10 ; c2 += c2 < 0
+ vsub.s32 q3, q3, q11 ; d2 += d2 < 0
+
+ vadd.s32 q8, q0, q15 ; a2 + 3
+ vadd.s32 q9, q1, q15 ; b2 + 3
+ vadd.s32 q10, q2, q15 ; c2 + 3
+ vadd.s32 q11, q3, q15 ; d2 + 3
+
+ ; vrshrn? would add 1 << 3-1 = 2
+ vshrn.s32 d0, q8, #3
+ vshrn.s32 d1, q9, #3
+ vshrn.s32 d2, q10, #3
+ vshrn.s32 d3, q11, #3
+
+ vst1.16 {q0, q1}, [r1@128]
+
+ bx lr
+
+ ENDP
+
+ END
diff --git a/libvpx/vp8/encoder/arm/quantize_arm.c b/libvpx/vp8/encoder/arm/quantize_arm.c
new file mode 100644
index 0000000..8999e34
--- /dev/null
+++ b/libvpx/vp8/encoder/arm/quantize_arm.c
@@ -0,0 +1,64 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/encoder/block.h"
+#include <math.h>
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/encoder/quantize.h"
+#include "vp8/common/entropy.h"
+
+
+#if HAVE_NEON
+
+/* vp8_quantize_mbX functions here differs from corresponding ones in
+ * quantize.c only by using quantize_b_pair function pointer instead of
+ * the regular quantize_b function pointer */
+void vp8_quantize_mby_neon(MACROBLOCK *x)
+{
+ int i;
+ int has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED
+ && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
+
+ for (i = 0; i < 16; i+=2)
+ x->quantize_b_pair(&x->block[i], &x->block[i+1],
+ &x->e_mbd.block[i], &x->e_mbd.block[i+1]);
+
+ if(has_2nd_order)
+ x->quantize_b(&x->block[24], &x->e_mbd.block[24]);
+}
+
+void vp8_quantize_mb_neon(MACROBLOCK *x)
+{
+ int i;
+ int has_2nd_order=(x->e_mbd.mode_info_context->mbmi.mode != B_PRED
+ && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
+
+ for (i = 0; i < 24; i+=2)
+ x->quantize_b_pair(&x->block[i], &x->block[i+1],
+ &x->e_mbd.block[i], &x->e_mbd.block[i+1]);
+
+ if (has_2nd_order)
+ x->quantize_b(&x->block[24], &x->e_mbd.block[24]);
+}
+
+
+void vp8_quantize_mbuv_neon(MACROBLOCK *x)
+{
+ int i;
+
+ for (i = 16; i < 24; i+=2)
+ x->quantize_b_pair(&x->block[i], &x->block[i+1],
+ &x->e_mbd.block[i], &x->e_mbd.block[i+1]);
+}
+
+#endif /* HAVE_NEON */
diff --git a/libvpx/vp8/encoder/asm_enc_offsets.c b/libvpx/vp8/encoder/asm_enc_offsets.c
new file mode 100644
index 0000000..a4169b3
--- /dev/null
+++ b/libvpx/vp8/encoder/asm_enc_offsets.c
@@ -0,0 +1,93 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_ports/asm_offsets.h"
+#include "vpx_config.h"
+#include "block.h"
+#include "vp8/common/blockd.h"
+#include "onyx_int.h"
+#include "treewriter.h"
+#include "tokenize.h"
+
+BEGIN
+
+/* regular quantize */
+DEFINE(vp8_block_coeff, offsetof(BLOCK, coeff));
+DEFINE(vp8_block_zbin, offsetof(BLOCK, zbin));
+DEFINE(vp8_block_round, offsetof(BLOCK, round));
+DEFINE(vp8_block_quant, offsetof(BLOCK, quant));
+DEFINE(vp8_block_quant_fast, offsetof(BLOCK, quant_fast));
+DEFINE(vp8_block_zbin_extra, offsetof(BLOCK, zbin_extra));
+DEFINE(vp8_block_zrun_zbin_boost, offsetof(BLOCK, zrun_zbin_boost));
+DEFINE(vp8_block_quant_shift, offsetof(BLOCK, quant_shift));
+
+DEFINE(vp8_blockd_qcoeff, offsetof(BLOCKD, qcoeff));
+DEFINE(vp8_blockd_dequant, offsetof(BLOCKD, dequant));
+DEFINE(vp8_blockd_dqcoeff, offsetof(BLOCKD, dqcoeff));
+DEFINE(vp8_blockd_eob, offsetof(BLOCKD, eob));
+
+/* subtract */
+DEFINE(vp8_block_base_src, offsetof(BLOCK, base_src));
+DEFINE(vp8_block_src, offsetof(BLOCK, src));
+DEFINE(vp8_block_src_diff, offsetof(BLOCK, src_diff));
+DEFINE(vp8_block_src_stride, offsetof(BLOCK, src_stride));
+
+DEFINE(vp8_blockd_predictor, offsetof(BLOCKD, predictor));
+
+/* pack tokens */
+DEFINE(vp8_writer_lowvalue, offsetof(vp8_writer, lowvalue));
+DEFINE(vp8_writer_range, offsetof(vp8_writer, range));
+DEFINE(vp8_writer_count, offsetof(vp8_writer, count));
+DEFINE(vp8_writer_pos, offsetof(vp8_writer, pos));
+DEFINE(vp8_writer_buffer, offsetof(vp8_writer, buffer));
+DEFINE(vp8_writer_buffer_end, offsetof(vp8_writer, buffer_end));
+DEFINE(vp8_writer_error, offsetof(vp8_writer, error));
+
+DEFINE(tokenextra_token, offsetof(TOKENEXTRA, Token));
+DEFINE(tokenextra_extra, offsetof(TOKENEXTRA, Extra));
+DEFINE(tokenextra_context_tree, offsetof(TOKENEXTRA, context_tree));
+DEFINE(tokenextra_skip_eob_node, offsetof(TOKENEXTRA, skip_eob_node));
+DEFINE(TOKENEXTRA_SZ, sizeof(TOKENEXTRA));
+
+DEFINE(vp8_extra_bit_struct_sz, sizeof(vp8_extra_bit_struct));
+
+DEFINE(vp8_token_value, offsetof(vp8_token, value));
+DEFINE(vp8_token_len, offsetof(vp8_token, Len));
+
+DEFINE(vp8_extra_bit_struct_tree, offsetof(vp8_extra_bit_struct, tree));
+DEFINE(vp8_extra_bit_struct_prob, offsetof(vp8_extra_bit_struct, prob));
+DEFINE(vp8_extra_bit_struct_len, offsetof(vp8_extra_bit_struct, Len));
+DEFINE(vp8_extra_bit_struct_base_val, offsetof(vp8_extra_bit_struct, base_val));
+
+DEFINE(vp8_comp_tplist, offsetof(VP8_COMP, tplist));
+DEFINE(vp8_comp_common, offsetof(VP8_COMP, common));
+DEFINE(vp8_comp_bc , offsetof(VP8_COMP, bc));
+DEFINE(vp8_writer_sz , sizeof(vp8_writer));
+
+DEFINE(tokenlist_start, offsetof(TOKENLIST, start));
+DEFINE(tokenlist_stop, offsetof(TOKENLIST, stop));
+DEFINE(TOKENLIST_SZ, sizeof(TOKENLIST));
+
+DEFINE(vp8_common_mb_rows, offsetof(VP8_COMMON, mb_rows));
+
+END
+
+/* add asserts for any offset that is not supported by assembly code
+ * add asserts for any size that is not supported by assembly code
+
+ * These are used in vp8cx_pack_tokens. They are hard coded so if their sizes
+ * change they will have to be adjusted.
+ */
+
+#if HAVE_EDSP
+ct_assert(TOKENEXTRA_SZ, sizeof(TOKENEXTRA) == 8)
+ct_assert(vp8_extra_bit_struct_sz, sizeof(vp8_extra_bit_struct) == 16)
+#endif
diff --git a/libvpx/vp8/encoder/bitstream.c b/libvpx/vp8/encoder/bitstream.c
new file mode 100644
index 0000000..f84ae68
--- /dev/null
+++ b/libvpx/vp8/encoder/bitstream.c
@@ -0,0 +1,1732 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/common/header.h"
+#include "encodemv.h"
+#include "vp8/common/entropymode.h"
+#include "vp8/common/findnearmv.h"
+#include "mcomp.h"
+#include "vp8/common/systemdependent.h"
+#include <assert.h>
+#include <stdio.h>
+#include <limits.h>
+#include "vp8/common/pragmas.h"
+#include "vpx/vpx_encoder.h"
+#include "vpx_mem/vpx_mem.h"
+#include "bitstream.h"
+
+#include "defaultcoefcounts.h"
+#include "vp8/common/common.h"
+
+const int vp8cx_base_skip_false_prob[128] =
+{
+ 255, 255, 255, 255, 255, 255, 255, 255,
+ 255, 255, 255, 255, 255, 255, 255, 255,
+ 255, 255, 255, 255, 255, 255, 255, 255,
+ 255, 255, 255, 255, 255, 255, 255, 255,
+ 255, 255, 255, 255, 255, 255, 255, 255,
+ 255, 255, 255, 255, 255, 255, 255, 255,
+ 255, 255, 255, 255, 255, 255, 255, 255,
+ 251, 248, 244, 240, 236, 232, 229, 225,
+ 221, 217, 213, 208, 204, 199, 194, 190,
+ 187, 183, 179, 175, 172, 168, 164, 160,
+ 157, 153, 149, 145, 142, 138, 134, 130,
+ 127, 124, 120, 117, 114, 110, 107, 104,
+ 101, 98, 95, 92, 89, 86, 83, 80,
+ 77, 74, 71, 68, 65, 62, 59, 56,
+ 53, 50, 47, 44, 41, 38, 35, 32,
+ 30, 28, 26, 24, 22, 20, 18, 16,
+};
+
+#if defined(SECTIONBITS_OUTPUT)
+unsigned __int64 Sectionbits[500];
+#endif
+
+#ifdef ENTROPY_STATS
+int intra_mode_stats[10][10][10];
+static unsigned int tree_update_hist [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES] [2];
+extern unsigned int active_section;
+#endif
+
+#ifdef MODE_STATS
+int count_mb_seg[4] = { 0, 0, 0, 0 };
+#endif
+
+
+static void update_mode(
+ vp8_writer *const w,
+ int n,
+ vp8_token tok [/* n */],
+ vp8_tree tree,
+ vp8_prob Pnew [/* n-1 */],
+ vp8_prob Pcur [/* n-1 */],
+ unsigned int bct [/* n-1 */] [2],
+ const unsigned int num_events[/* n */]
+)
+{
+ unsigned int new_b = 0, old_b = 0;
+ int i = 0;
+
+ vp8_tree_probs_from_distribution(
+ n--, tok, tree,
+ Pnew, bct, num_events,
+ 256, 1
+ );
+
+ do
+ {
+ new_b += vp8_cost_branch(bct[i], Pnew[i]);
+ old_b += vp8_cost_branch(bct[i], Pcur[i]);
+ }
+ while (++i < n);
+
+ if (new_b + (n << 8) < old_b)
+ {
+ int i = 0;
+
+ vp8_write_bit(w, 1);
+
+ do
+ {
+ const vp8_prob p = Pnew[i];
+
+ vp8_write_literal(w, Pcur[i] = p ? p : 1, 8);
+ }
+ while (++i < n);
+ }
+ else
+ vp8_write_bit(w, 0);
+}
+
+static void update_mbintra_mode_probs(VP8_COMP *cpi)
+{
+ VP8_COMMON *const x = & cpi->common;
+
+ vp8_writer *const w = cpi->bc;
+
+ {
+ vp8_prob Pnew [VP8_YMODES-1];
+ unsigned int bct [VP8_YMODES-1] [2];
+
+ update_mode(
+ w, VP8_YMODES, vp8_ymode_encodings, vp8_ymode_tree,
+ Pnew, x->fc.ymode_prob, bct, (unsigned int *)cpi->mb.ymode_count
+ );
+ }
+ {
+ vp8_prob Pnew [VP8_UV_MODES-1];
+ unsigned int bct [VP8_UV_MODES-1] [2];
+
+ update_mode(
+ w, VP8_UV_MODES, vp8_uv_mode_encodings, vp8_uv_mode_tree,
+ Pnew, x->fc.uv_mode_prob, bct, (unsigned int *)cpi->mb.uv_mode_count
+ );
+ }
+}
+
+static void write_ymode(vp8_writer *bc, int m, const vp8_prob *p)
+{
+ vp8_write_token(bc, vp8_ymode_tree, p, vp8_ymode_encodings + m);
+}
+
+static void kfwrite_ymode(vp8_writer *bc, int m, const vp8_prob *p)
+{
+ vp8_write_token(bc, vp8_kf_ymode_tree, p, vp8_kf_ymode_encodings + m);
+}
+
+static void write_uv_mode(vp8_writer *bc, int m, const vp8_prob *p)
+{
+ vp8_write_token(bc, vp8_uv_mode_tree, p, vp8_uv_mode_encodings + m);
+}
+
+
+static void write_bmode(vp8_writer *bc, int m, const vp8_prob *p)
+{
+ vp8_write_token(bc, vp8_bmode_tree, p, vp8_bmode_encodings + m);
+}
+
+static void write_split(vp8_writer *bc, int x)
+{
+ vp8_write_token(
+ bc, vp8_mbsplit_tree, vp8_mbsplit_probs, vp8_mbsplit_encodings + x
+ );
+}
+
+void vp8_pack_tokens_c(vp8_writer *w, const TOKENEXTRA *p, int xcount)
+{
+ const TOKENEXTRA *stop = p + xcount;
+ unsigned int split;
+ unsigned int shift;
+ int count = w->count;
+ unsigned int range = w->range;
+ unsigned int lowvalue = w->lowvalue;
+
+ while (p < stop)
+ {
+ const int t = p->Token;
+ vp8_token *a = vp8_coef_encodings + t;
+ const vp8_extra_bit_struct *b = vp8_extra_bits + t;
+ int i = 0;
+ const unsigned char *pp = p->context_tree;
+ int v = a->value;
+ int n = a->Len;
+
+ if (p->skip_eob_node)
+ {
+ n--;
+ i = 2;
+ }
+
+ do
+ {
+ const int bb = (v >> --n) & 1;
+ split = 1 + (((range - 1) * pp[i>>1]) >> 8);
+ i = vp8_coef_tree[i+bb];
+
+ if (bb)
+ {
+ lowvalue += split;
+ range = range - split;
+ }
+ else
+ {
+ range = split;
+ }
+
+ shift = vp8_norm[range];
+ range <<= shift;
+ count += shift;
+
+ if (count >= 0)
+ {
+ int offset = shift - count;
+
+ if ((lowvalue << (offset - 1)) & 0x80000000)
+ {
+ int x = w->pos - 1;
+
+ while (x >= 0 && w->buffer[x] == 0xff)
+ {
+ w->buffer[x] = (unsigned char)0;
+ x--;
+ }
+
+ w->buffer[x] += 1;
+ }
+
+ validate_buffer(w->buffer + w->pos,
+ 1,
+ w->buffer_end,
+ w->error);
+
+ w->buffer[w->pos++] = (lowvalue >> (24 - offset));
+ lowvalue <<= offset;
+ shift = count;
+ lowvalue &= 0xffffff;
+ count -= 8 ;
+ }
+
+ lowvalue <<= shift;
+ }
+ while (n);
+
+
+ if (b->base_val)
+ {
+ const int e = p->Extra, L = b->Len;
+
+ if (L)
+ {
+ const unsigned char *pp = b->prob;
+ int v = e >> 1;
+ int n = L; /* number of bits in v, assumed nonzero */
+ int i = 0;
+
+ do
+ {
+ const int bb = (v >> --n) & 1;
+ split = 1 + (((range - 1) * pp[i>>1]) >> 8);
+ i = b->tree[i+bb];
+
+ if (bb)
+ {
+ lowvalue += split;
+ range = range - split;
+ }
+ else
+ {
+ range = split;
+ }
+
+ shift = vp8_norm[range];
+ range <<= shift;
+ count += shift;
+
+ if (count >= 0)
+ {
+ int offset = shift - count;
+
+ if ((lowvalue << (offset - 1)) & 0x80000000)
+ {
+ int x = w->pos - 1;
+
+ while (x >= 0 && w->buffer[x] == 0xff)
+ {
+ w->buffer[x] = (unsigned char)0;
+ x--;
+ }
+
+ w->buffer[x] += 1;
+ }
+
+ validate_buffer(w->buffer + w->pos,
+ 1,
+ w->buffer_end,
+ w->error);
+
+ w->buffer[w->pos++] = (lowvalue >> (24 - offset));
+ lowvalue <<= offset;
+ shift = count;
+ lowvalue &= 0xffffff;
+ count -= 8 ;
+ }
+
+ lowvalue <<= shift;
+ }
+ while (n);
+ }
+
+
+ {
+
+ split = (range + 1) >> 1;
+
+ if (e & 1)
+ {
+ lowvalue += split;
+ range = range - split;
+ }
+ else
+ {
+ range = split;
+ }
+
+ range <<= 1;
+
+ if ((lowvalue & 0x80000000))
+ {
+ int x = w->pos - 1;
+
+ while (x >= 0 && w->buffer[x] == 0xff)
+ {
+ w->buffer[x] = (unsigned char)0;
+ x--;
+ }
+
+ w->buffer[x] += 1;
+
+ }
+
+ lowvalue <<= 1;
+
+ if (!++count)
+ {
+ count = -8;
+
+ validate_buffer(w->buffer + w->pos,
+ 1,
+ w->buffer_end,
+ w->error);
+
+ w->buffer[w->pos++] = (lowvalue >> 24);
+ lowvalue &= 0xffffff;
+ }
+ }
+
+ }
+
+ ++p;
+ }
+
+ w->count = count;
+ w->lowvalue = lowvalue;
+ w->range = range;
+
+}
+
+static void write_partition_size(unsigned char *cx_data, int size)
+{
+ signed char csize;
+
+ csize = size & 0xff;
+ *cx_data = csize;
+ csize = (size >> 8) & 0xff;
+ *(cx_data + 1) = csize;
+ csize = (size >> 16) & 0xff;
+ *(cx_data + 2) = csize;
+
+}
+
+static void pack_tokens_into_partitions_c(VP8_COMP *cpi, unsigned char *cx_data,
+ unsigned char * cx_data_end,
+ int num_part)
+{
+
+ int i;
+ unsigned char *ptr = cx_data;
+ unsigned char *ptr_end = cx_data_end;
+ vp8_writer * w;
+
+ for (i = 0; i < num_part; i++)
+ {
+ int mb_row;
+
+ w = cpi->bc + i + 1;
+
+ vp8_start_encode(w, ptr, ptr_end);
+
+ for (mb_row = i; mb_row < cpi->common.mb_rows; mb_row += num_part)
+ {
+ const TOKENEXTRA *p = cpi->tplist[mb_row].start;
+ const TOKENEXTRA *stop = cpi->tplist[mb_row].stop;
+ int tokens = (int)(stop - p);
+
+ vp8_pack_tokens_c(w, p, tokens);
+ }
+
+ vp8_stop_encode(w);
+ ptr += w->pos;
+ }
+}
+
+
+static void pack_mb_row_tokens_c(VP8_COMP *cpi, vp8_writer *w)
+{
+ int mb_row;
+
+ for (mb_row = 0; mb_row < cpi->common.mb_rows; mb_row++)
+ {
+ const TOKENEXTRA *p = cpi->tplist[mb_row].start;
+ const TOKENEXTRA *stop = cpi->tplist[mb_row].stop;
+ int tokens = (int)(stop - p);
+
+ vp8_pack_tokens_c(w, p, tokens);
+ }
+
+}
+
+static void write_mv_ref
+(
+ vp8_writer *w, MB_PREDICTION_MODE m, const vp8_prob *p
+)
+{
+#if CONFIG_DEBUG
+ assert(NEARESTMV <= m && m <= SPLITMV);
+#endif
+ vp8_write_token(w, vp8_mv_ref_tree, p,
+ vp8_mv_ref_encoding_array - NEARESTMV + m);
+}
+
+static void write_sub_mv_ref
+(
+ vp8_writer *w, B_PREDICTION_MODE m, const vp8_prob *p
+)
+{
+#if CONFIG_DEBUG
+ assert(LEFT4X4 <= m && m <= NEW4X4);
+#endif
+ vp8_write_token(w, vp8_sub_mv_ref_tree, p,
+ vp8_sub_mv_ref_encoding_array - LEFT4X4 + m);
+}
+
+static void write_mv
+(
+ vp8_writer *w, const MV *mv, const int_mv *ref, const MV_CONTEXT *mvc
+)
+{
+ MV e;
+ e.row = mv->row - ref->as_mv.row;
+ e.col = mv->col - ref->as_mv.col;
+
+ vp8_encode_motion_vector(w, &e, mvc);
+}
+
+static void write_mb_features(vp8_writer *w, const MB_MODE_INFO *mi, const MACROBLOCKD *x)
+{
+ /* Encode the MB segment id. */
+ if (x->segmentation_enabled && x->update_mb_segmentation_map)
+ {
+ switch (mi->segment_id)
+ {
+ case 0:
+ vp8_write(w, 0, x->mb_segment_tree_probs[0]);
+ vp8_write(w, 0, x->mb_segment_tree_probs[1]);
+ break;
+ case 1:
+ vp8_write(w, 0, x->mb_segment_tree_probs[0]);
+ vp8_write(w, 1, x->mb_segment_tree_probs[1]);
+ break;
+ case 2:
+ vp8_write(w, 1, x->mb_segment_tree_probs[0]);
+ vp8_write(w, 0, x->mb_segment_tree_probs[2]);
+ break;
+ case 3:
+ vp8_write(w, 1, x->mb_segment_tree_probs[0]);
+ vp8_write(w, 1, x->mb_segment_tree_probs[2]);
+ break;
+
+ /* TRAP.. This should not happen */
+ default:
+ vp8_write(w, 0, x->mb_segment_tree_probs[0]);
+ vp8_write(w, 0, x->mb_segment_tree_probs[1]);
+ break;
+ }
+ }
+}
+void vp8_convert_rfct_to_prob(VP8_COMP *const cpi)
+{
+ const int *const rfct = cpi->count_mb_ref_frame_usage;
+ const int rf_intra = rfct[INTRA_FRAME];
+ const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
+
+ /* Calculate the probabilities used to code the ref frame based on usage */
+ if (!(cpi->prob_intra_coded = rf_intra * 255 / (rf_intra + rf_inter)))
+ cpi->prob_intra_coded = 1;
+
+ cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
+
+ if (!cpi->prob_last_coded)
+ cpi->prob_last_coded = 1;
+
+ cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
+ ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
+
+ if (!cpi->prob_gf_coded)
+ cpi->prob_gf_coded = 1;
+
+}
+
+static void pack_inter_mode_mvs(VP8_COMP *const cpi)
+{
+ VP8_COMMON *const pc = & cpi->common;
+ vp8_writer *const w = cpi->bc;
+ const MV_CONTEXT *mvc = pc->fc.mvc;
+
+
+ MODE_INFO *m = pc->mi;
+ const int mis = pc->mode_info_stride;
+ int mb_row = -1;
+
+ int prob_skip_false = 0;
+
+ cpi->mb.partition_info = cpi->mb.pi;
+
+ vp8_convert_rfct_to_prob(cpi);
+
+#ifdef ENTROPY_STATS
+ active_section = 1;
+#endif
+
+ if (pc->mb_no_coeff_skip)
+ {
+ int total_mbs = pc->mb_rows * pc->mb_cols;
+
+ prob_skip_false = (total_mbs - cpi->mb.skip_true_count ) * 256 / total_mbs;
+
+ if (prob_skip_false <= 1)
+ prob_skip_false = 1;
+
+ if (prob_skip_false > 255)
+ prob_skip_false = 255;
+
+ cpi->prob_skip_false = prob_skip_false;
+ vp8_write_literal(w, prob_skip_false, 8);
+ }
+
+ vp8_write_literal(w, cpi->prob_intra_coded, 8);
+ vp8_write_literal(w, cpi->prob_last_coded, 8);
+ vp8_write_literal(w, cpi->prob_gf_coded, 8);
+
+ update_mbintra_mode_probs(cpi);
+
+ vp8_write_mvprobs(cpi);
+
+ while (++mb_row < pc->mb_rows)
+ {
+ int mb_col = -1;
+
+ while (++mb_col < pc->mb_cols)
+ {
+ const MB_MODE_INFO *const mi = & m->mbmi;
+ const MV_REFERENCE_FRAME rf = mi->ref_frame;
+ const MB_PREDICTION_MODE mode = mi->mode;
+
+ MACROBLOCKD *xd = &cpi->mb.e_mbd;
+
+ /* Distance of Mb to the various image edges.
+ * These specified to 8th pel as they are always compared to MV
+ * values that are in 1/8th pel units
+ */
+ xd->mb_to_left_edge = -((mb_col * 16) << 3);
+ xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3;
+ xd->mb_to_top_edge = -((mb_row * 16)) << 3;
+ xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3;
+
+#ifdef ENTROPY_STATS
+ active_section = 9;
+#endif
+
+ if (cpi->mb.e_mbd.update_mb_segmentation_map)
+ write_mb_features(w, mi, &cpi->mb.e_mbd);
+
+ if (pc->mb_no_coeff_skip)
+ vp8_encode_bool(w, m->mbmi.mb_skip_coeff, prob_skip_false);
+
+ if (rf == INTRA_FRAME)
+ {
+ vp8_write(w, 0, cpi->prob_intra_coded);
+#ifdef ENTROPY_STATS
+ active_section = 6;
+#endif
+ write_ymode(w, mode, pc->fc.ymode_prob);
+
+ if (mode == B_PRED)
+ {
+ int j = 0;
+
+ do
+ write_bmode(w, m->bmi[j].as_mode, pc->fc.bmode_prob);
+ while (++j < 16);
+ }
+
+ write_uv_mode(w, mi->uv_mode, pc->fc.uv_mode_prob);
+ }
+ else /* inter coded */
+ {
+ int_mv best_mv;
+ vp8_prob mv_ref_p [VP8_MVREFS-1];
+
+ vp8_write(w, 1, cpi->prob_intra_coded);
+
+ if (rf == LAST_FRAME)
+ vp8_write(w, 0, cpi->prob_last_coded);
+ else
+ {
+ vp8_write(w, 1, cpi->prob_last_coded);
+ vp8_write(w, (rf == GOLDEN_FRAME) ? 0 : 1, cpi->prob_gf_coded);
+ }
+
+ {
+ int_mv n1, n2;
+ int ct[4];
+
+ vp8_find_near_mvs(xd, m, &n1, &n2, &best_mv, ct, rf, cpi->common.ref_frame_sign_bias);
+ vp8_clamp_mv2(&best_mv, xd);
+
+ vp8_mv_ref_probs(mv_ref_p, ct);
+
+#ifdef ENTROPY_STATS
+ accum_mv_refs(mode, ct);
+#endif
+
+ }
+
+#ifdef ENTROPY_STATS
+ active_section = 3;
+#endif
+
+ write_mv_ref(w, mode, mv_ref_p);
+
+ switch (mode) /* new, split require MVs */
+ {
+ case NEWMV:
+
+#ifdef ENTROPY_STATS
+ active_section = 5;
+#endif
+
+ write_mv(w, &mi->mv.as_mv, &best_mv, mvc);
+ break;
+
+ case SPLITMV:
+ {
+ int j = 0;
+
+#ifdef MODE_STATS
+ ++count_mb_seg [mi->partitioning];
+#endif
+
+ write_split(w, mi->partitioning);
+
+ do
+ {
+ B_PREDICTION_MODE blockmode;
+ int_mv blockmv;
+ const int *const L = vp8_mbsplits [mi->partitioning];
+ int k = -1; /* first block in subset j */
+ int mv_contz;
+ int_mv leftmv, abovemv;
+
+ blockmode = cpi->mb.partition_info->bmi[j].mode;
+ blockmv = cpi->mb.partition_info->bmi[j].mv;
+#if CONFIG_DEBUG
+ while (j != L[++k])
+ if (k >= 16)
+ assert(0);
+#else
+ while (j != L[++k]);
+#endif
+ leftmv.as_int = left_block_mv(m, k);
+ abovemv.as_int = above_block_mv(m, k, mis);
+ mv_contz = vp8_mv_cont(&leftmv, &abovemv);
+
+ write_sub_mv_ref(w, blockmode, vp8_sub_mv_ref_prob2 [mv_contz]);
+
+ if (blockmode == NEW4X4)
+ {
+#ifdef ENTROPY_STATS
+ active_section = 11;
+#endif
+ write_mv(w, &blockmv.as_mv, &best_mv, (const MV_CONTEXT *) mvc);
+ }
+ }
+ while (++j < cpi->mb.partition_info->count);
+ }
+ break;
+ default:
+ break;
+ }
+ }
+
+ ++m;
+ cpi->mb.partition_info++;
+ }
+
+ ++m; /* skip L prediction border */
+ cpi->mb.partition_info++;
+ }
+}
+
+
+static void write_kfmodes(VP8_COMP *cpi)
+{
+ vp8_writer *const bc = cpi->bc;
+ const VP8_COMMON *const c = & cpi->common;
+ /* const */
+ MODE_INFO *m = c->mi;
+
+ int mb_row = -1;
+ int prob_skip_false = 0;
+
+ if (c->mb_no_coeff_skip)
+ {
+ int total_mbs = c->mb_rows * c->mb_cols;
+
+ prob_skip_false = (total_mbs - cpi->mb.skip_true_count ) * 256 / total_mbs;
+
+ if (prob_skip_false <= 1)
+ prob_skip_false = 1;
+
+ if (prob_skip_false >= 255)
+ prob_skip_false = 255;
+
+ cpi->prob_skip_false = prob_skip_false;
+ vp8_write_literal(bc, prob_skip_false, 8);
+ }
+
+ while (++mb_row < c->mb_rows)
+ {
+ int mb_col = -1;
+
+ while (++mb_col < c->mb_cols)
+ {
+ const int ym = m->mbmi.mode;
+
+ if (cpi->mb.e_mbd.update_mb_segmentation_map)
+ write_mb_features(bc, &m->mbmi, &cpi->mb.e_mbd);
+
+ if (c->mb_no_coeff_skip)
+ vp8_encode_bool(bc, m->mbmi.mb_skip_coeff, prob_skip_false);
+
+ kfwrite_ymode(bc, ym, vp8_kf_ymode_prob);
+
+ if (ym == B_PRED)
+ {
+ const int mis = c->mode_info_stride;
+ int i = 0;
+
+ do
+ {
+ const B_PREDICTION_MODE A = above_block_mode(m, i, mis);
+ const B_PREDICTION_MODE L = left_block_mode(m, i);
+ const int bm = m->bmi[i].as_mode;
+
+#ifdef ENTROPY_STATS
+ ++intra_mode_stats [A] [L] [bm];
+#endif
+
+ write_bmode(bc, bm, vp8_kf_bmode_prob [A] [L]);
+ }
+ while (++i < 16);
+ }
+
+ write_uv_mode(bc, (m++)->mbmi.uv_mode, vp8_kf_uv_mode_prob);
+ }
+
+ m++; /* skip L prediction border */
+ }
+}
+
+#if 0
+/* This function is used for debugging probability trees. */
+static void print_prob_tree(vp8_prob
+ coef_probs[BLOCK_TYPES][COEF_BANDS][PREV_COEF_CONTEXTS][ENTROPY_NODES])
+{
+ /* print coef probability tree */
+ int i,j,k,l;
+ FILE* f = fopen("enc_tree_probs.txt", "a");
+ fprintf(f, "{\n");
+ for (i = 0; i < BLOCK_TYPES; i++)
+ {
+ fprintf(f, " {\n");
+ for (j = 0; j < COEF_BANDS; j++)
+ {
+ fprintf(f, " {\n");
+ for (k = 0; k < PREV_COEF_CONTEXTS; k++)
+ {
+ fprintf(f, " {");
+ for (l = 0; l < ENTROPY_NODES; l++)
+ {
+ fprintf(f, "%3u, ",
+ (unsigned int)(coef_probs [i][j][k][l]));
+ }
+ fprintf(f, " }\n");
+ }
+ fprintf(f, " }\n");
+ }
+ fprintf(f, " }\n");
+ }
+ fprintf(f, "}\n");
+ fclose(f);
+}
+#endif
+
+static void sum_probs_over_prev_coef_context(
+ const unsigned int probs[PREV_COEF_CONTEXTS][MAX_ENTROPY_TOKENS],
+ unsigned int* out)
+{
+ int i, j;
+ for (i=0; i < MAX_ENTROPY_TOKENS; ++i)
+ {
+ for (j=0; j < PREV_COEF_CONTEXTS; ++j)
+ {
+ const unsigned int tmp = out[i];
+ out[i] += probs[j][i];
+ /* check for wrap */
+ if (out[i] < tmp)
+ out[i] = UINT_MAX;
+ }
+ }
+}
+
+static int prob_update_savings(const unsigned int *ct,
+ const vp8_prob oldp, const vp8_prob newp,
+ const vp8_prob upd)
+{
+ const int old_b = vp8_cost_branch(ct, oldp);
+ const int new_b = vp8_cost_branch(ct, newp);
+ const int update_b = 8 +
+ ((vp8_cost_one(upd) - vp8_cost_zero(upd)) >> 8);
+
+ return old_b - new_b - update_b;
+}
+
+static int independent_coef_context_savings(VP8_COMP *cpi)
+{
+ MACROBLOCK *const x = & cpi->mb;
+ int savings = 0;
+ int i = 0;
+ do
+ {
+ int j = 0;
+ do
+ {
+ int k = 0;
+ unsigned int prev_coef_count_sum[MAX_ENTROPY_TOKENS] = {0};
+ int prev_coef_savings[MAX_ENTROPY_TOKENS] = {0};
+ const unsigned int (*probs)[MAX_ENTROPY_TOKENS];
+ /* Calculate new probabilities given the constraint that
+ * they must be equal over the prev coef contexts
+ */
+
+ probs = (const unsigned int (*)[MAX_ENTROPY_TOKENS])
+ x->coef_counts[i][j];
+
+ /* Reset to default probabilities at key frames */
+ if (cpi->common.frame_type == KEY_FRAME)
+ probs = default_coef_counts[i][j];
+
+ sum_probs_over_prev_coef_context(probs, prev_coef_count_sum);
+
+ do
+ {
+ /* at every context */
+
+ /* calc probs and branch cts for this frame only */
+ int t = 0; /* token/prob index */
+
+ vp8_tree_probs_from_distribution(
+ MAX_ENTROPY_TOKENS, vp8_coef_encodings, vp8_coef_tree,
+ cpi->frame_coef_probs[i][j][k],
+ cpi->frame_branch_ct [i][j][k],
+ prev_coef_count_sum,
+ 256, 1);
+
+ do
+ {
+ const unsigned int *ct = cpi->frame_branch_ct [i][j][k][t];
+ const vp8_prob newp = cpi->frame_coef_probs [i][j][k][t];
+ const vp8_prob oldp = cpi->common.fc.coef_probs [i][j][k][t];
+ const vp8_prob upd = vp8_coef_update_probs [i][j][k][t];
+ const int s = prob_update_savings(ct, oldp, newp, upd);
+
+ if (cpi->common.frame_type != KEY_FRAME ||
+ (cpi->common.frame_type == KEY_FRAME && newp != oldp))
+ prev_coef_savings[t] += s;
+ }
+ while (++t < ENTROPY_NODES);
+ }
+ while (++k < PREV_COEF_CONTEXTS);
+ k = 0;
+ do
+ {
+ /* We only update probabilities if we can save bits, except
+ * for key frames where we have to update all probabilities
+ * to get the equal probabilities across the prev coef
+ * contexts.
+ */
+ if (prev_coef_savings[k] > 0 ||
+ cpi->common.frame_type == KEY_FRAME)
+ savings += prev_coef_savings[k];
+ }
+ while (++k < ENTROPY_NODES);
+ }
+ while (++j < COEF_BANDS);
+ }
+ while (++i < BLOCK_TYPES);
+ return savings;
+}
+
+static int default_coef_context_savings(VP8_COMP *cpi)
+{
+ MACROBLOCK *const x = & cpi->mb;
+ int savings = 0;
+ int i = 0;
+ do
+ {
+ int j = 0;
+ do
+ {
+ int k = 0;
+ do
+ {
+ /* at every context */
+
+ /* calc probs and branch cts for this frame only */
+ int t = 0; /* token/prob index */
+
+ vp8_tree_probs_from_distribution(
+ MAX_ENTROPY_TOKENS, vp8_coef_encodings, vp8_coef_tree,
+ cpi->frame_coef_probs [i][j][k],
+ cpi->frame_branch_ct [i][j][k],
+ x->coef_counts [i][j][k],
+ 256, 1
+ );
+
+ do
+ {
+ const unsigned int *ct = cpi->frame_branch_ct [i][j][k][t];
+ const vp8_prob newp = cpi->frame_coef_probs [i][j][k][t];
+ const vp8_prob oldp = cpi->common.fc.coef_probs [i][j][k][t];
+ const vp8_prob upd = vp8_coef_update_probs [i][j][k][t];
+ const int s = prob_update_savings(ct, oldp, newp, upd);
+
+ if (s > 0)
+ {
+ savings += s;
+ }
+ }
+ while (++t < ENTROPY_NODES);
+ }
+ while (++k < PREV_COEF_CONTEXTS);
+ }
+ while (++j < COEF_BANDS);
+ }
+ while (++i < BLOCK_TYPES);
+ return savings;
+}
+
+void vp8_calc_ref_frame_costs(int *ref_frame_cost,
+ int prob_intra,
+ int prob_last,
+ int prob_garf
+ )
+{
+ ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(prob_intra);
+ ref_frame_cost[LAST_FRAME] = vp8_cost_one(prob_intra)
+ + vp8_cost_zero(prob_last);
+ ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(prob_intra)
+ + vp8_cost_one(prob_last)
+ + vp8_cost_zero(prob_garf);
+ ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(prob_intra)
+ + vp8_cost_one(prob_last)
+ + vp8_cost_one(prob_garf);
+
+}
+
+int vp8_estimate_entropy_savings(VP8_COMP *cpi)
+{
+ int savings = 0;
+
+ const int *const rfct = cpi->count_mb_ref_frame_usage;
+ const int rf_intra = rfct[INTRA_FRAME];
+ const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
+ int new_intra, new_last, new_garf, oldtotal, newtotal;
+ int ref_frame_cost[MAX_REF_FRAMES];
+
+ vp8_clear_system_state();
+
+ if (cpi->common.frame_type != KEY_FRAME)
+ {
+ if (!(new_intra = rf_intra * 255 / (rf_intra + rf_inter)))
+ new_intra = 1;
+
+ new_last = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
+
+ new_garf = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
+ ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
+
+
+ vp8_calc_ref_frame_costs(ref_frame_cost,new_intra,new_last,new_garf);
+
+ newtotal =
+ rfct[INTRA_FRAME] * ref_frame_cost[INTRA_FRAME] +
+ rfct[LAST_FRAME] * ref_frame_cost[LAST_FRAME] +
+ rfct[GOLDEN_FRAME] * ref_frame_cost[GOLDEN_FRAME] +
+ rfct[ALTREF_FRAME] * ref_frame_cost[ALTREF_FRAME];
+
+
+ /* old costs */
+ vp8_calc_ref_frame_costs(ref_frame_cost,cpi->prob_intra_coded,
+ cpi->prob_last_coded,cpi->prob_gf_coded);
+
+ oldtotal =
+ rfct[INTRA_FRAME] * ref_frame_cost[INTRA_FRAME] +
+ rfct[LAST_FRAME] * ref_frame_cost[LAST_FRAME] +
+ rfct[GOLDEN_FRAME] * ref_frame_cost[GOLDEN_FRAME] +
+ rfct[ALTREF_FRAME] * ref_frame_cost[ALTREF_FRAME];
+
+ savings += (oldtotal - newtotal) / 256;
+ }
+
+
+ if (cpi->oxcf.error_resilient_mode & VPX_ERROR_RESILIENT_PARTITIONS)
+ savings += independent_coef_context_savings(cpi);
+ else
+ savings += default_coef_context_savings(cpi);
+
+
+ return savings;
+}
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+int vp8_update_coef_context(VP8_COMP *cpi)
+{
+ int savings = 0;
+
+
+ if (cpi->common.frame_type == KEY_FRAME)
+ {
+ /* Reset to default counts/probabilities at key frames */
+ vp8_copy(cpi->coef_counts, default_coef_counts);
+ }
+
+ if (cpi->oxcf.error_resilient_mode & VPX_ERROR_RESILIENT_PARTITIONS)
+ savings += independent_coef_context_savings(cpi);
+ else
+ savings += default_coef_context_savings(cpi);
+
+ return savings;
+}
+#endif
+
+void vp8_update_coef_probs(VP8_COMP *cpi)
+{
+ int i = 0;
+#if !(CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ vp8_writer *const w = cpi->bc;
+#endif
+ int savings = 0;
+
+ vp8_clear_system_state();
+
+ do
+ {
+ int j = 0;
+
+ do
+ {
+ int k = 0;
+ int prev_coef_savings[ENTROPY_NODES] = {0};
+ if (cpi->oxcf.error_resilient_mode & VPX_ERROR_RESILIENT_PARTITIONS)
+ {
+ for (k = 0; k < PREV_COEF_CONTEXTS; ++k)
+ {
+ int t; /* token/prob index */
+ for (t = 0; t < ENTROPY_NODES; ++t)
+ {
+ const unsigned int *ct = cpi->frame_branch_ct [i][j]
+ [k][t];
+ const vp8_prob newp = cpi->frame_coef_probs[i][j][k][t];
+ const vp8_prob oldp = cpi->common.fc.coef_probs[i][j]
+ [k][t];
+ const vp8_prob upd = vp8_coef_update_probs[i][j][k][t];
+
+ prev_coef_savings[t] +=
+ prob_update_savings(ct, oldp, newp, upd);
+ }
+ }
+ k = 0;
+ }
+ do
+ {
+ /* note: use result from vp8_estimate_entropy_savings, so no
+ * need to call vp8_tree_probs_from_distribution here.
+ */
+
+ /* at every context */
+
+ /* calc probs and branch cts for this frame only */
+ int t = 0; /* token/prob index */
+
+ do
+ {
+ const vp8_prob newp = cpi->frame_coef_probs [i][j][k][t];
+
+ vp8_prob *Pold = cpi->common.fc.coef_probs [i][j][k] + t;
+ const vp8_prob upd = vp8_coef_update_probs [i][j][k][t];
+
+ int s = prev_coef_savings[t];
+ int u = 0;
+
+ if (!(cpi->oxcf.error_resilient_mode &
+ VPX_ERROR_RESILIENT_PARTITIONS))
+ {
+ s = prob_update_savings(
+ cpi->frame_branch_ct [i][j][k][t],
+ *Pold, newp, upd);
+ }
+
+ if (s > 0)
+ u = 1;
+
+ /* Force updates on key frames if the new is different,
+ * so that we can be sure we end up with equal probabilities
+ * over the prev coef contexts.
+ */
+ if ((cpi->oxcf.error_resilient_mode &
+ VPX_ERROR_RESILIENT_PARTITIONS) &&
+ cpi->common.frame_type == KEY_FRAME && newp != *Pold)
+ u = 1;
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ cpi->update_probs[i][j][k][t] = u;
+#else
+ vp8_write(w, u, upd);
+#endif
+
+
+#ifdef ENTROPY_STATS
+ ++ tree_update_hist [i][j][k][t] [u];
+#endif
+
+ if (u)
+ {
+ /* send/use new probability */
+
+ *Pold = newp;
+#if !(CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ vp8_write_literal(w, newp, 8);
+#endif
+
+ savings += s;
+
+ }
+
+ }
+ while (++t < ENTROPY_NODES);
+
+ /* Accum token counts for generation of default statistics */
+#ifdef ENTROPY_STATS
+ t = 0;
+
+ do
+ {
+ context_counters [i][j][k][t] += cpi->coef_counts [i][j][k][t];
+ }
+ while (++t < MAX_ENTROPY_TOKENS);
+
+#endif
+
+ }
+ while (++k < PREV_COEF_CONTEXTS);
+ }
+ while (++j < COEF_BANDS);
+ }
+ while (++i < BLOCK_TYPES);
+
+}
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+static void pack_coef_probs(VP8_COMP *cpi)
+{
+ int i = 0;
+ vp8_writer *const w = cpi->bc;
+
+ do
+ {
+ int j = 0;
+
+ do
+ {
+ int k = 0;
+
+ do
+ {
+ int t = 0; /* token/prob index */
+
+ do
+ {
+ const vp8_prob newp = cpi->common.fc.coef_probs [i][j][k][t];
+ const vp8_prob upd = vp8_coef_update_probs [i][j][k][t];
+
+ const char u = cpi->update_probs[i][j][k][t] ;
+
+ vp8_write(w, u, upd);
+
+ if (u)
+ {
+ /* send/use new probability */
+ vp8_write_literal(w, newp, 8);
+ }
+ }
+ while (++t < ENTROPY_NODES);
+ }
+ while (++k < PREV_COEF_CONTEXTS);
+ }
+ while (++j < COEF_BANDS);
+ }
+ while (++i < BLOCK_TYPES);
+}
+#endif
+
+#ifdef PACKET_TESTING
+FILE *vpxlogc = 0;
+#endif
+
+static void put_delta_q(vp8_writer *bc, int delta_q)
+{
+ if (delta_q != 0)
+ {
+ vp8_write_bit(bc, 1);
+ vp8_write_literal(bc, abs(delta_q), 4);
+
+ if (delta_q < 0)
+ vp8_write_bit(bc, 1);
+ else
+ vp8_write_bit(bc, 0);
+ }
+ else
+ vp8_write_bit(bc, 0);
+}
+
+void vp8_pack_bitstream(VP8_COMP *cpi, unsigned char *dest, unsigned char * dest_end, unsigned long *size)
+{
+ int i, j;
+ VP8_HEADER oh;
+ VP8_COMMON *const pc = & cpi->common;
+ vp8_writer *const bc = cpi->bc;
+ MACROBLOCKD *const xd = & cpi->mb.e_mbd;
+ int extra_bytes_packed = 0;
+
+ unsigned char *cx_data = dest;
+ unsigned char *cx_data_end = dest_end;
+ const int *mb_feature_data_bits;
+
+ oh.show_frame = (int) pc->show_frame;
+ oh.type = (int)pc->frame_type;
+ oh.version = pc->version;
+ oh.first_partition_length_in_bytes = 0;
+
+ mb_feature_data_bits = vp8_mb_feature_data_bits;
+
+ bc[0].error = &pc->error;
+
+ validate_buffer(cx_data, 3, cx_data_end, &cpi->common.error);
+ cx_data += 3;
+
+#if defined(SECTIONBITS_OUTPUT)
+ Sectionbits[active_section = 1] += sizeof(VP8_HEADER) * 8 * 256;
+#endif
+
+ /* every keyframe send startcode, width, height, scale factor, clamp
+ * and color type
+ */
+ if (oh.type == KEY_FRAME)
+ {
+ int v;
+
+ validate_buffer(cx_data, 7, cx_data_end, &cpi->common.error);
+
+ /* Start / synch code */
+ cx_data[0] = 0x9D;
+ cx_data[1] = 0x01;
+ cx_data[2] = 0x2a;
+
+ v = (pc->horiz_scale << 14) | pc->Width;
+ cx_data[3] = v;
+ cx_data[4] = v >> 8;
+
+ v = (pc->vert_scale << 14) | pc->Height;
+ cx_data[5] = v;
+ cx_data[6] = v >> 8;
+
+
+ extra_bytes_packed = 7;
+ cx_data += extra_bytes_packed ;
+
+ vp8_start_encode(bc, cx_data, cx_data_end);
+
+ /* signal clr type */
+ vp8_write_bit(bc, pc->clr_type);
+ vp8_write_bit(bc, pc->clamp_type);
+
+ }
+ else
+ vp8_start_encode(bc, cx_data, cx_data_end);
+
+
+ /* Signal whether or not Segmentation is enabled */
+ vp8_write_bit(bc, xd->segmentation_enabled);
+
+ /* Indicate which features are enabled */
+ if (xd->segmentation_enabled)
+ {
+ /* Signal whether or not the segmentation map is being updated. */
+ vp8_write_bit(bc, xd->update_mb_segmentation_map);
+ vp8_write_bit(bc, xd->update_mb_segmentation_data);
+
+ if (xd->update_mb_segmentation_data)
+ {
+ signed char Data;
+
+ vp8_write_bit(bc, xd->mb_segement_abs_delta);
+
+ /* For each segmentation feature (Quant and loop filter level) */
+ for (i = 0; i < MB_LVL_MAX; i++)
+ {
+ /* For each of the segments */
+ for (j = 0; j < MAX_MB_SEGMENTS; j++)
+ {
+ Data = xd->segment_feature_data[i][j];
+
+ /* Frame level data */
+ if (Data)
+ {
+ vp8_write_bit(bc, 1);
+
+ if (Data < 0)
+ {
+ Data = - Data;
+ vp8_write_literal(bc, Data, mb_feature_data_bits[i]);
+ vp8_write_bit(bc, 1);
+ }
+ else
+ {
+ vp8_write_literal(bc, Data, mb_feature_data_bits[i]);
+ vp8_write_bit(bc, 0);
+ }
+ }
+ else
+ vp8_write_bit(bc, 0);
+ }
+ }
+ }
+
+ if (xd->update_mb_segmentation_map)
+ {
+ /* Write the probs used to decode the segment id for each mb */
+ for (i = 0; i < MB_FEATURE_TREE_PROBS; i++)
+ {
+ int Data = xd->mb_segment_tree_probs[i];
+
+ if (Data != 255)
+ {
+ vp8_write_bit(bc, 1);
+ vp8_write_literal(bc, Data, 8);
+ }
+ else
+ vp8_write_bit(bc, 0);
+ }
+ }
+ }
+
+ vp8_write_bit(bc, pc->filter_type);
+ vp8_write_literal(bc, pc->filter_level, 6);
+ vp8_write_literal(bc, pc->sharpness_level, 3);
+
+ /* Write out loop filter deltas applied at the MB level based on mode
+ * or ref frame (if they are enabled).
+ */
+ vp8_write_bit(bc, xd->mode_ref_lf_delta_enabled);
+
+ if (xd->mode_ref_lf_delta_enabled)
+ {
+ /* Do the deltas need to be updated */
+ int send_update = xd->mode_ref_lf_delta_update
+ || cpi->oxcf.error_resilient_mode;
+
+ vp8_write_bit(bc, send_update);
+ if (send_update)
+ {
+ int Data;
+
+ /* Send update */
+ for (i = 0; i < MAX_REF_LF_DELTAS; i++)
+ {
+ Data = xd->ref_lf_deltas[i];
+
+ /* Frame level data */
+ if (xd->ref_lf_deltas[i] != xd->last_ref_lf_deltas[i]
+ || cpi->oxcf.error_resilient_mode)
+ {
+ xd->last_ref_lf_deltas[i] = xd->ref_lf_deltas[i];
+ vp8_write_bit(bc, 1);
+
+ if (Data > 0)
+ {
+ vp8_write_literal(bc, (Data & 0x3F), 6);
+ vp8_write_bit(bc, 0); /* sign */
+ }
+ else
+ {
+ Data = -Data;
+ vp8_write_literal(bc, (Data & 0x3F), 6);
+ vp8_write_bit(bc, 1); /* sign */
+ }
+ }
+ else
+ vp8_write_bit(bc, 0);
+ }
+
+ /* Send update */
+ for (i = 0; i < MAX_MODE_LF_DELTAS; i++)
+ {
+ Data = xd->mode_lf_deltas[i];
+
+ if (xd->mode_lf_deltas[i] != xd->last_mode_lf_deltas[i]
+ || cpi->oxcf.error_resilient_mode)
+ {
+ xd->last_mode_lf_deltas[i] = xd->mode_lf_deltas[i];
+ vp8_write_bit(bc, 1);
+
+ if (Data > 0)
+ {
+ vp8_write_literal(bc, (Data & 0x3F), 6);
+ vp8_write_bit(bc, 0); /* sign */
+ }
+ else
+ {
+ Data = -Data;
+ vp8_write_literal(bc, (Data & 0x3F), 6);
+ vp8_write_bit(bc, 1); /* sign */
+ }
+ }
+ else
+ vp8_write_bit(bc, 0);
+ }
+ }
+ }
+
+ /* signal here is multi token partition is enabled */
+ vp8_write_literal(bc, pc->multi_token_partition, 2);
+
+ /* Frame Qbaseline quantizer index */
+ vp8_write_literal(bc, pc->base_qindex, 7);
+
+ /* Transmit Dc, Second order and Uv quantizer delta information */
+ put_delta_q(bc, pc->y1dc_delta_q);
+ put_delta_q(bc, pc->y2dc_delta_q);
+ put_delta_q(bc, pc->y2ac_delta_q);
+ put_delta_q(bc, pc->uvdc_delta_q);
+ put_delta_q(bc, pc->uvac_delta_q);
+
+ /* When there is a key frame all reference buffers are updated using
+ * the new key frame
+ */
+ if (pc->frame_type != KEY_FRAME)
+ {
+ /* Should the GF or ARF be updated using the transmitted frame
+ * or buffer
+ */
+ vp8_write_bit(bc, pc->refresh_golden_frame);
+ vp8_write_bit(bc, pc->refresh_alt_ref_frame);
+
+ /* If not being updated from current frame should either GF or ARF
+ * be updated from another buffer
+ */
+ if (!pc->refresh_golden_frame)
+ vp8_write_literal(bc, pc->copy_buffer_to_gf, 2);
+
+ if (!pc->refresh_alt_ref_frame)
+ vp8_write_literal(bc, pc->copy_buffer_to_arf, 2);
+
+ /* Indicate reference frame sign bias for Golden and ARF frames
+ * (always 0 for last frame buffer)
+ */
+ vp8_write_bit(bc, pc->ref_frame_sign_bias[GOLDEN_FRAME]);
+ vp8_write_bit(bc, pc->ref_frame_sign_bias[ALTREF_FRAME]);
+ }
+
+#if !(CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ if (cpi->oxcf.error_resilient_mode & VPX_ERROR_RESILIENT_PARTITIONS)
+ {
+ if (pc->frame_type == KEY_FRAME)
+ pc->refresh_entropy_probs = 1;
+ else
+ pc->refresh_entropy_probs = 0;
+ }
+#endif
+
+ vp8_write_bit(bc, pc->refresh_entropy_probs);
+
+ if (pc->frame_type != KEY_FRAME)
+ vp8_write_bit(bc, pc->refresh_last_frame);
+
+#ifdef ENTROPY_STATS
+
+ if (pc->frame_type == INTER_FRAME)
+ active_section = 0;
+ else
+ active_section = 7;
+
+#endif
+
+ vp8_clear_system_state();
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ pack_coef_probs(cpi);
+#else
+ if (pc->refresh_entropy_probs == 0)
+ {
+ /* save a copy for later refresh */
+ vpx_memcpy(&cpi->common.lfc, &cpi->common.fc, sizeof(cpi->common.fc));
+ }
+
+ vp8_update_coef_probs(cpi);
+#endif
+
+#ifdef ENTROPY_STATS
+ active_section = 2;
+#endif
+
+ /* Write out the mb_no_coeff_skip flag */
+ vp8_write_bit(bc, pc->mb_no_coeff_skip);
+
+ if (pc->frame_type == KEY_FRAME)
+ {
+ write_kfmodes(cpi);
+
+#ifdef ENTROPY_STATS
+ active_section = 8;
+#endif
+ }
+ else
+ {
+ pack_inter_mode_mvs(cpi);
+
+#ifdef ENTROPY_STATS
+ active_section = 1;
+#endif
+ }
+
+ vp8_stop_encode(bc);
+
+ cx_data += bc->pos;
+
+ oh.first_partition_length_in_bytes = cpi->bc->pos;
+
+ /* update frame tag */
+ {
+ int v = (oh.first_partition_length_in_bytes << 5) |
+ (oh.show_frame << 4) |
+ (oh.version << 1) |
+ oh.type;
+
+ dest[0] = v;
+ dest[1] = v >> 8;
+ dest[2] = v >> 16;
+ }
+
+ *size = VP8_HEADER_SIZE + extra_bytes_packed + cpi->bc->pos;
+
+ cpi->partition_sz[0] = *size;
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ {
+ const int num_part = (1 << pc->multi_token_partition);
+ unsigned char * dp = cpi->partition_d[0] + cpi->partition_sz[0];
+
+ if (num_part > 1)
+ {
+ /* write token part sizes (all but last) if more than 1 */
+ validate_buffer(dp, 3 * (num_part - 1), cpi->partition_d_end[0],
+ &pc->error);
+
+ cpi->partition_sz[0] += 3*(num_part-1);
+
+ for(i = 1; i < num_part; i++)
+ {
+ write_partition_size(dp, cpi->partition_sz[i]);
+ dp += 3;
+ }
+ }
+
+ if (!cpi->output_partition)
+ {
+ /* concatenate partition buffers */
+ for(i = 0; i < num_part; i++)
+ {
+ vpx_memmove(dp, cpi->partition_d[i+1], cpi->partition_sz[i+1]);
+ cpi->partition_d[i+1] = dp;
+ dp += cpi->partition_sz[i+1];
+ }
+ }
+
+ /* update total size */
+ *size = 0;
+ for(i = 0; i < num_part+1; i++)
+ {
+ *size += cpi->partition_sz[i];
+ }
+ }
+#else
+ if (pc->multi_token_partition != ONE_PARTITION)
+ {
+ int num_part = 1 << pc->multi_token_partition;
+
+ /* partition size table at the end of first partition */
+ cpi->partition_sz[0] += 3 * (num_part - 1);
+ *size += 3 * (num_part - 1);
+
+ validate_buffer(cx_data, 3 * (num_part - 1), cx_data_end,
+ &pc->error);
+
+ for(i = 1; i < num_part + 1; i++)
+ {
+ cpi->bc[i].error = &pc->error;
+ }
+
+ pack_tokens_into_partitions(cpi, cx_data + 3 * (num_part - 1),
+ cx_data_end, num_part);
+
+ for(i = 1; i < num_part; i++)
+ {
+ cpi->partition_sz[i] = cpi->bc[i].pos;
+ write_partition_size(cx_data, cpi->partition_sz[i]);
+ cx_data += 3;
+ *size += cpi->partition_sz[i]; /* add to total */
+ }
+
+ /* add last partition to total size */
+ cpi->partition_sz[i] = cpi->bc[i].pos;
+ *size += cpi->partition_sz[i];
+ }
+ else
+ {
+ bc[1].error = &pc->error;
+
+ vp8_start_encode(&cpi->bc[1], cx_data, cx_data_end);
+
+#if CONFIG_MULTITHREAD
+ if (cpi->b_multi_threaded)
+ pack_mb_row_tokens(cpi, &cpi->bc[1]);
+ else
+#endif
+ pack_tokens(&cpi->bc[1], cpi->tok, cpi->tok_count);
+
+ vp8_stop_encode(&cpi->bc[1]);
+
+ *size += cpi->bc[1].pos;
+ cpi->partition_sz[1] = cpi->bc[1].pos;
+ }
+#endif
+}
+
+#ifdef ENTROPY_STATS
+void print_tree_update_probs()
+{
+ int i, j, k, l;
+ FILE *f = fopen("context.c", "a");
+ int Sum;
+ fprintf(f, "\n/* Update probabilities for token entropy tree. */\n\n");
+ fprintf(f, "const vp8_prob tree_update_probs[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES] = {\n");
+
+ for (i = 0; i < BLOCK_TYPES; i++)
+ {
+ fprintf(f, " { \n");
+
+ for (j = 0; j < COEF_BANDS; j++)
+ {
+ fprintf(f, " {\n");
+
+ for (k = 0; k < PREV_COEF_CONTEXTS; k++)
+ {
+ fprintf(f, " {");
+
+ for (l = 0; l < ENTROPY_NODES; l++)
+ {
+ Sum = tree_update_hist[i][j][k][l][0] + tree_update_hist[i][j][k][l][1];
+
+ if (Sum > 0)
+ {
+ if (((tree_update_hist[i][j][k][l][0] * 255) / Sum) > 0)
+ fprintf(f, "%3ld, ", (tree_update_hist[i][j][k][l][0] * 255) / Sum);
+ else
+ fprintf(f, "%3ld, ", 1);
+ }
+ else
+ fprintf(f, "%3ld, ", 128);
+ }
+
+ fprintf(f, "},\n");
+ }
+
+ fprintf(f, " },\n");
+ }
+
+ fprintf(f, " },\n");
+ }
+
+ fprintf(f, "};\n");
+ fclose(f);
+}
+#endif
diff --git a/libvpx/vp8/encoder/bitstream.h b/libvpx/vp8/encoder/bitstream.h
new file mode 100644
index 0000000..455a94f
--- /dev/null
+++ b/libvpx/vp8/encoder/bitstream.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_BITSTREAM_H
+#define __INC_BITSTREAM_H
+
+#if HAVE_EDSP
+void vp8cx_pack_tokens_armv5(vp8_writer *w, const TOKENEXTRA *p, int xcount,
+ const vp8_token *,
+ const vp8_extra_bit_struct *,
+ const vp8_tree_index *);
+void vp8cx_pack_tokens_into_partitions_armv5(VP8_COMP *,
+ unsigned char * cx_data,
+ const unsigned char *cx_data_end,
+ int num_parts,
+ const vp8_token *,
+ const vp8_extra_bit_struct *,
+ const vp8_tree_index *);
+void vp8cx_pack_mb_row_tokens_armv5(VP8_COMP *cpi, vp8_writer *w,
+ const vp8_token *,
+ const vp8_extra_bit_struct *,
+ const vp8_tree_index *);
+# define pack_tokens(a,b,c) \
+ vp8cx_pack_tokens_armv5(a,b,c,vp8_coef_encodings,vp8_extra_bits,vp8_coef_tree)
+# define pack_tokens_into_partitions(a,b,c,d) \
+ vp8cx_pack_tokens_into_partitions_armv5(a,b,c,d,vp8_coef_encodings,vp8_extra_bits,vp8_coef_tree)
+# define pack_mb_row_tokens(a,b) \
+ vp8cx_pack_mb_row_tokens_armv5(a,b,vp8_coef_encodings,vp8_extra_bits,vp8_coef_tree)
+#else
+
+void vp8_pack_tokens_c(vp8_writer *w, const TOKENEXTRA *p, int xcount);
+
+# define pack_tokens(a,b,c) vp8_pack_tokens_c(a,b,c)
+# define pack_tokens_into_partitions(a,b,c,d) pack_tokens_into_partitions_c(a,b,c,d)
+# define pack_mb_row_tokens(a,b) pack_mb_row_tokens_c(a,b)
+#endif
+
+#endif
diff --git a/libvpx/vp8/encoder/block.h b/libvpx/vp8/encoder/block.h
new file mode 100644
index 0000000..f9d63eb
--- /dev/null
+++ b/libvpx/vp8/encoder/block.h
@@ -0,0 +1,148 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_BLOCK_H
+#define __INC_BLOCK_H
+
+#include "vp8/common/onyx.h"
+#include "vp8/common/blockd.h"
+#include "vp8/common/entropymv.h"
+#include "vp8/common/entropy.h"
+#include "vpx_ports/mem.h"
+
+/* motion search site */
+typedef struct
+{
+ MV mv;
+ int offset;
+} search_site;
+
+typedef struct block
+{
+ /* 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries */
+ short *src_diff;
+ short *coeff;
+
+ /* 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries */
+ short *quant;
+ short *quant_fast;
+ unsigned char *quant_shift;
+ short *zbin;
+ short *zrun_zbin_boost;
+ short *round;
+
+ /* Zbin Over Quant value */
+ short zbin_extra;
+
+ unsigned char **base_src;
+ int src;
+ int src_stride;
+} BLOCK;
+
+typedef struct
+{
+ int count;
+ struct
+ {
+ B_PREDICTION_MODE mode;
+ int_mv mv;
+ } bmi[16];
+} PARTITION_INFO;
+
+typedef struct macroblock
+{
+ DECLARE_ALIGNED(16, short, src_diff[400]); /* 25 blocks Y,U,V,Y2 */
+ DECLARE_ALIGNED(16, short, coeff[400]); /* 25 blocks Y,U,V,Y2 */
+ DECLARE_ALIGNED(16, unsigned char, thismb[256]);
+
+ unsigned char *thismb_ptr;
+ /* 16 Y, 4 U, 4 V, 1 DC 2nd order block */
+ BLOCK block[25];
+
+ YV12_BUFFER_CONFIG src;
+
+ MACROBLOCKD e_mbd;
+ PARTITION_INFO *partition_info; /* work pointer */
+ PARTITION_INFO *pi; /* Corresponds to upper left visible macroblock */
+ PARTITION_INFO *pip; /* Base of allocated array */
+
+ int ref_frame_cost[MAX_REF_FRAMES];
+
+ search_site *ss;
+ int ss_count;
+ int searches_per_step;
+
+ int errorperbit;
+ int sadperbit16;
+ int sadperbit4;
+ int rddiv;
+ int rdmult;
+ unsigned int * mb_activity_ptr;
+ int * mb_norm_activity_ptr;
+ signed int act_zbin_adj;
+ signed int last_act_zbin_adj;
+
+ int *mvcost[2];
+ int *mvsadcost[2];
+ int (*mbmode_cost)[MB_MODE_COUNT];
+ int (*intra_uv_mode_cost)[MB_MODE_COUNT];
+ int (*bmode_costs)[10][10];
+ int *inter_bmode_costs;
+ int (*token_costs)[COEF_BANDS][PREV_COEF_CONTEXTS]
+ [MAX_ENTROPY_TOKENS];
+
+ /* These define limits to motion vector components to prevent
+ * them from extending outside the UMV borders.
+ */
+ int mv_col_min;
+ int mv_col_max;
+ int mv_row_min;
+ int mv_row_max;
+
+ int skip;
+
+ unsigned int encode_breakout;
+
+ signed char *gf_active_ptr;
+
+ unsigned char *active_ptr;
+ MV_CONTEXT *mvc;
+
+ int optimize;
+ int q_index;
+
+#if CONFIG_TEMPORAL_DENOISING
+ MB_PREDICTION_MODE best_sse_inter_mode;
+ int_mv best_sse_mv;
+ MV_REFERENCE_FRAME best_reference_frame;
+ MV_REFERENCE_FRAME best_zeromv_reference_frame;
+ unsigned char need_to_clamp_best_mvs;
+#endif
+
+ int skip_true_count;
+ unsigned int coef_counts [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [MAX_ENTROPY_TOKENS];
+ unsigned int MVcount [2] [MVvals]; /* (row,col) MV cts this frame */
+ int ymode_count [VP8_YMODES]; /* intra MB type cts this frame */
+ int uv_mode_count[VP8_UV_MODES]; /* intra MB type cts this frame */
+ int64_t prediction_error;
+ int64_t intra_error;
+
+
+ void (*short_fdct4x4)(short *input, short *output, int pitch);
+ void (*short_fdct8x4)(short *input, short *output, int pitch);
+ void (*short_walsh4x4)(short *input, short *output, int pitch);
+ void (*quantize_b)(BLOCK *b, BLOCKD *d);
+ void (*quantize_b_pair)(BLOCK *b1, BLOCK *b2, BLOCKD *d0, BLOCKD *d1);
+
+} MACROBLOCK;
+
+
+#endif
diff --git a/libvpx/vp8/encoder/boolhuff.c b/libvpx/vp8/encoder/boolhuff.c
new file mode 100644
index 0000000..74770a2
--- /dev/null
+++ b/libvpx/vp8/encoder/boolhuff.c
@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "boolhuff.h"
+
+#if defined(SECTIONBITS_OUTPUT)
+unsigned __int64 Sectionbits[500];
+
+#endif
+
+#ifdef ENTROPY_STATS
+unsigned int active_section = 0;
+#endif
+
+const unsigned int vp8_prob_cost[256] =
+{
+ 2047, 2047, 1791, 1641, 1535, 1452, 1385, 1328, 1279, 1235, 1196, 1161, 1129, 1099, 1072, 1046,
+ 1023, 1000, 979, 959, 940, 922, 905, 889, 873, 858, 843, 829, 816, 803, 790, 778,
+ 767, 755, 744, 733, 723, 713, 703, 693, 684, 675, 666, 657, 649, 641, 633, 625,
+ 617, 609, 602, 594, 587, 580, 573, 567, 560, 553, 547, 541, 534, 528, 522, 516,
+ 511, 505, 499, 494, 488, 483, 477, 472, 467, 462, 457, 452, 447, 442, 437, 433,
+ 428, 424, 419, 415, 410, 406, 401, 397, 393, 389, 385, 381, 377, 373, 369, 365,
+ 361, 357, 353, 349, 346, 342, 338, 335, 331, 328, 324, 321, 317, 314, 311, 307,
+ 304, 301, 297, 294, 291, 288, 285, 281, 278, 275, 272, 269, 266, 263, 260, 257,
+ 255, 252, 249, 246, 243, 240, 238, 235, 232, 229, 227, 224, 221, 219, 216, 214,
+ 211, 208, 206, 203, 201, 198, 196, 194, 191, 189, 186, 184, 181, 179, 177, 174,
+ 172, 170, 168, 165, 163, 161, 159, 156, 154, 152, 150, 148, 145, 143, 141, 139,
+ 137, 135, 133, 131, 129, 127, 125, 123, 121, 119, 117, 115, 113, 111, 109, 107,
+ 105, 103, 101, 99, 97, 95, 93, 92, 90, 88, 86, 84, 82, 81, 79, 77,
+ 75, 73, 72, 70, 68, 66, 65, 63, 61, 60, 58, 56, 55, 53, 51, 50,
+ 48, 46, 45, 43, 41, 40, 38, 37, 35, 33, 32, 30, 29, 27, 25, 24,
+ 22, 21, 19, 18, 16, 15, 13, 12, 10, 9, 7, 6, 4, 3, 1, 1
+};
+
+void vp8_start_encode(BOOL_CODER *br, unsigned char *source, unsigned char *source_end)
+{
+
+ br->lowvalue = 0;
+ br->range = 255;
+ br->count = -24;
+ br->buffer = source;
+ br->buffer_end = source_end;
+ br->pos = 0;
+}
+
+void vp8_stop_encode(BOOL_CODER *br)
+{
+ int i;
+
+ for (i = 0; i < 32; i++)
+ vp8_encode_bool(br, 0, 128);
+}
+
+
+void vp8_encode_value(BOOL_CODER *br, int data, int bits)
+{
+ int bit;
+
+ for (bit = bits - 1; bit >= 0; bit--)
+ vp8_encode_bool(br, (1 & (data >> bit)), 0x80);
+
+}
diff --git a/libvpx/vp8/encoder/boolhuff.h b/libvpx/vp8/encoder/boolhuff.h
new file mode 100644
index 0000000..8309063
--- /dev/null
+++ b/libvpx/vp8/encoder/boolhuff.h
@@ -0,0 +1,128 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/****************************************************************************
+*
+* Module Title : boolhuff.h
+*
+* Description : Bool Coder header file.
+*
+****************************************************************************/
+#ifndef __INC_BOOLHUFF_H
+#define __INC_BOOLHUFF_H
+
+#include "vpx_ports/mem.h"
+#include "vpx/internal/vpx_codec_internal.h"
+
+typedef struct
+{
+ unsigned int lowvalue;
+ unsigned int range;
+ int count;
+ unsigned int pos;
+ unsigned char *buffer;
+ unsigned char *buffer_end;
+ struct vpx_internal_error_info *error;
+
+ /* Variables used to track bit costs without outputing to the bitstream */
+ unsigned int measure_cost;
+ unsigned long bit_counter;
+} BOOL_CODER;
+
+extern void vp8_start_encode(BOOL_CODER *bc, unsigned char *buffer, unsigned char *buffer_end);
+
+extern void vp8_encode_value(BOOL_CODER *br, int data, int bits);
+extern void vp8_stop_encode(BOOL_CODER *bc);
+extern const unsigned int vp8_prob_cost[256];
+
+
+DECLARE_ALIGNED(16, extern const unsigned char, vp8_norm[256]);
+
+static int validate_buffer(const unsigned char *start,
+ size_t len,
+ const unsigned char *end,
+ struct vpx_internal_error_info *error)
+{
+ if (start + len > start && start + len < end)
+ return 1;
+ else
+ vpx_internal_error(error, VPX_CODEC_CORRUPT_FRAME,
+ "Truncated packet or corrupt partition ");
+
+ return 0;
+}
+static void vp8_encode_bool(BOOL_CODER *br, int bit, int probability)
+{
+ unsigned int split;
+ int count = br->count;
+ unsigned int range = br->range;
+ unsigned int lowvalue = br->lowvalue;
+ register unsigned int shift;
+
+#ifdef ENTROPY_STATS
+#if defined(SECTIONBITS_OUTPUT)
+
+ if (bit)
+ Sectionbits[active_section] += vp8_prob_cost[255-probability];
+ else
+ Sectionbits[active_section] += vp8_prob_cost[probability];
+
+#endif
+#endif
+
+ split = 1 + (((range - 1) * probability) >> 8);
+
+ range = split;
+
+ if (bit)
+ {
+ lowvalue += split;
+ range = br->range - split;
+ }
+
+ shift = vp8_norm[range];
+
+ range <<= shift;
+ count += shift;
+
+ if (count >= 0)
+ {
+ int offset = shift - count;
+
+ if ((lowvalue << (offset - 1)) & 0x80000000)
+ {
+ int x = br->pos - 1;
+
+ while (x >= 0 && br->buffer[x] == 0xff)
+ {
+ br->buffer[x] = (unsigned char)0;
+ x--;
+ }
+
+ br->buffer[x] += 1;
+ }
+
+ validate_buffer(br->buffer + br->pos, 1, br->buffer_end, br->error);
+ br->buffer[br->pos++] = (lowvalue >> (24 - offset));
+
+ lowvalue <<= offset;
+ shift = count;
+ lowvalue &= 0xffffff;
+ count -= 8 ;
+ }
+
+ lowvalue <<= shift;
+ br->count = count;
+ br->lowvalue = lowvalue;
+ br->range = range;
+}
+
+#endif
diff --git a/vp8/encoder/dct.c b/libvpx/vp8/encoder/dct.c
similarity index 100%
rename from vp8/encoder/dct.c
rename to libvpx/vp8/encoder/dct.c
diff --git a/libvpx/vp8/encoder/dct_value_cost.h b/libvpx/vp8/encoder/dct_value_cost.h
new file mode 100644
index 0000000..e892765
--- /dev/null
+++ b/libvpx/vp8/encoder/dct_value_cost.h
@@ -0,0 +1,358 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+/* Generated file, included by tokenize.c */
+/* Values generated by fill_value_tokens() */
+
+static const short dct_value_cost[2048*2] =
+{
+ 8285, 8277, 8267, 8259, 8253, 8245, 8226, 8218, 8212, 8204, 8194, 8186,
+ 8180, 8172, 8150, 8142, 8136, 8128, 8118, 8110, 8104, 8096, 8077, 8069,
+ 8063, 8055, 8045, 8037, 8031, 8023, 7997, 7989, 7983, 7975, 7965, 7957,
+ 7951, 7943, 7924, 7916, 7910, 7902, 7892, 7884, 7878, 7870, 7848, 7840,
+ 7834, 7826, 7816, 7808, 7802, 7794, 7775, 7767, 7761, 7753, 7743, 7735,
+ 7729, 7721, 7923, 7915, 7909, 7901, 7891, 7883, 7877, 7869, 7850, 7842,
+ 7836, 7828, 7818, 7810, 7804, 7796, 7774, 7766, 7760, 7752, 7742, 7734,
+ 7728, 7720, 7701, 7693, 7687, 7679, 7669, 7661, 7655, 7647, 7621, 7613,
+ 7607, 7599, 7589, 7581, 7575, 7567, 7548, 7540, 7534, 7526, 7516, 7508,
+ 7502, 7494, 7472, 7464, 7458, 7450, 7440, 7432, 7426, 7418, 7399, 7391,
+ 7385, 7377, 7367, 7359, 7353, 7345, 7479, 7471, 7465, 7457, 7447, 7439,
+ 7433, 7425, 7406, 7398, 7392, 7384, 7374, 7366, 7360, 7352, 7330, 7322,
+ 7316, 7308, 7298, 7290, 7284, 7276, 7257, 7249, 7243, 7235, 7225, 7217,
+ 7211, 7203, 7177, 7169, 7163, 7155, 7145, 7137, 7131, 7123, 7104, 7096,
+ 7090, 7082, 7072, 7064, 7058, 7050, 7028, 7020, 7014, 7006, 6996, 6988,
+ 6982, 6974, 6955, 6947, 6941, 6933, 6923, 6915, 6909, 6901, 7632, 7624,
+ 7618, 7610, 7600, 7592, 7586, 7578, 7559, 7551, 7545, 7537, 7527, 7519,
+ 7513, 7505, 7483, 7475, 7469, 7461, 7451, 7443, 7437, 7429, 7410, 7402,
+ 7396, 7388, 7378, 7370, 7364, 7356, 7330, 7322, 7316, 7308, 7298, 7290,
+ 7284, 7276, 7257, 7249, 7243, 7235, 7225, 7217, 7211, 7203, 7181, 7173,
+ 7167, 7159, 7149, 7141, 7135, 7127, 7108, 7100, 7094, 7086, 7076, 7068,
+ 7062, 7054, 7188, 7180, 7174, 7166, 7156, 7148, 7142, 7134, 7115, 7107,
+ 7101, 7093, 7083, 7075, 7069, 7061, 7039, 7031, 7025, 7017, 7007, 6999,
+ 6993, 6985, 6966, 6958, 6952, 6944, 6934, 6926, 6920, 6912, 6886, 6878,
+ 6872, 6864, 6854, 6846, 6840, 6832, 6813, 6805, 6799, 6791, 6781, 6773,
+ 6767, 6759, 6737, 6729, 6723, 6715, 6705, 6697, 6691, 6683, 6664, 6656,
+ 6650, 6642, 6632, 6624, 6618, 6610, 6812, 6804, 6798, 6790, 6780, 6772,
+ 6766, 6758, 6739, 6731, 6725, 6717, 6707, 6699, 6693, 6685, 6663, 6655,
+ 6649, 6641, 6631, 6623, 6617, 6609, 6590, 6582, 6576, 6568, 6558, 6550,
+ 6544, 6536, 6510, 6502, 6496, 6488, 6478, 6470, 6464, 6456, 6437, 6429,
+ 6423, 6415, 6405, 6397, 6391, 6383, 6361, 6353, 6347, 6339, 6329, 6321,
+ 6315, 6307, 6288, 6280, 6274, 6266, 6256, 6248, 6242, 6234, 6368, 6360,
+ 6354, 6346, 6336, 6328, 6322, 6314, 6295, 6287, 6281, 6273, 6263, 6255,
+ 6249, 6241, 6219, 6211, 6205, 6197, 6187, 6179, 6173, 6165, 6146, 6138,
+ 6132, 6124, 6114, 6106, 6100, 6092, 6066, 6058, 6052, 6044, 6034, 6026,
+ 6020, 6012, 5993, 5985, 5979, 5971, 5961, 5953, 5947, 5939, 5917, 5909,
+ 5903, 5895, 5885, 5877, 5871, 5863, 5844, 5836, 5830, 5822, 5812, 5804,
+ 5798, 5790, 6697, 6689, 6683, 6675, 6665, 6657, 6651, 6643, 6624, 6616,
+ 6610, 6602, 6592, 6584, 6578, 6570, 6548, 6540, 6534, 6526, 6516, 6508,
+ 6502, 6494, 6475, 6467, 6461, 6453, 6443, 6435, 6429, 6421, 6395, 6387,
+ 6381, 6373, 6363, 6355, 6349, 6341, 6322, 6314, 6308, 6300, 6290, 6282,
+ 6276, 6268, 6246, 6238, 6232, 6224, 6214, 6206, 6200, 6192, 6173, 6165,
+ 6159, 6151, 6141, 6133, 6127, 6119, 6253, 6245, 6239, 6231, 6221, 6213,
+ 6207, 6199, 6180, 6172, 6166, 6158, 6148, 6140, 6134, 6126, 6104, 6096,
+ 6090, 6082, 6072, 6064, 6058, 6050, 6031, 6023, 6017, 6009, 5999, 5991,
+ 5985, 5977, 5951, 5943, 5937, 5929, 5919, 5911, 5905, 5897, 5878, 5870,
+ 5864, 5856, 5846, 5838, 5832, 5824, 5802, 5794, 5788, 5780, 5770, 5762,
+ 5756, 5748, 5729, 5721, 5715, 5707, 5697, 5689, 5683, 5675, 5877, 5869,
+ 5863, 5855, 5845, 5837, 5831, 5823, 5804, 5796, 5790, 5782, 5772, 5764,
+ 5758, 5750, 5728, 5720, 5714, 5706, 5696, 5688, 5682, 5674, 5655, 5647,
+ 5641, 5633, 5623, 5615, 5609, 5601, 5575, 5567, 5561, 5553, 5543, 5535,
+ 5529, 5521, 5502, 5494, 5488, 5480, 5470, 5462, 5456, 5448, 5426, 5418,
+ 5412, 5404, 5394, 5386, 5380, 5372, 5353, 5345, 5339, 5331, 5321, 5313,
+ 5307, 5299, 5433, 5425, 5419, 5411, 5401, 5393, 5387, 5379, 5360, 5352,
+ 5346, 5338, 5328, 5320, 5314, 5306, 5284, 5276, 5270, 5262, 5252, 5244,
+ 5238, 5230, 5211, 5203, 5197, 5189, 5179, 5171, 5165, 5157, 5131, 5123,
+ 5117, 5109, 5099, 5091, 5085, 5077, 5058, 5050, 5044, 5036, 5026, 5018,
+ 5012, 5004, 4982, 4974, 4968, 4960, 4950, 4942, 4936, 4928, 4909, 4901,
+ 4895, 4887, 4877, 4869, 4863, 4855, 5586, 5578, 5572, 5564, 5554, 5546,
+ 5540, 5532, 5513, 5505, 5499, 5491, 5481, 5473, 5467, 5459, 5437, 5429,
+ 5423, 5415, 5405, 5397, 5391, 5383, 5364, 5356, 5350, 5342, 5332, 5324,
+ 5318, 5310, 5284, 5276, 5270, 5262, 5252, 5244, 5238, 5230, 5211, 5203,
+ 5197, 5189, 5179, 5171, 5165, 5157, 5135, 5127, 5121, 5113, 5103, 5095,
+ 5089, 5081, 5062, 5054, 5048, 5040, 5030, 5022, 5016, 5008, 5142, 5134,
+ 5128, 5120, 5110, 5102, 5096, 5088, 5069, 5061, 5055, 5047, 5037, 5029,
+ 5023, 5015, 4993, 4985, 4979, 4971, 4961, 4953, 4947, 4939, 4920, 4912,
+ 4906, 4898, 4888, 4880, 4874, 4866, 4840, 4832, 4826, 4818, 4808, 4800,
+ 4794, 4786, 4767, 4759, 4753, 4745, 4735, 4727, 4721, 4713, 4691, 4683,
+ 4677, 4669, 4659, 4651, 4645, 4637, 4618, 4610, 4604, 4596, 4586, 4578,
+ 4572, 4564, 4766, 4758, 4752, 4744, 4734, 4726, 4720, 4712, 4693, 4685,
+ 4679, 4671, 4661, 4653, 4647, 4639, 4617, 4609, 4603, 4595, 4585, 4577,
+ 4571, 4563, 4544, 4536, 4530, 4522, 4512, 4504, 4498, 4490, 4464, 4456,
+ 4450, 4442, 4432, 4424, 4418, 4410, 4391, 4383, 4377, 4369, 4359, 4351,
+ 4345, 4337, 4315, 4307, 4301, 4293, 4283, 4275, 4269, 4261, 4242, 4234,
+ 4228, 4220, 4210, 4202, 4196, 4188, 4322, 4314, 4308, 4300, 4290, 4282,
+ 4276, 4268, 4249, 4241, 4235, 4227, 4217, 4209, 4203, 4195, 4173, 4165,
+ 4159, 4151, 4141, 4133, 4127, 4119, 4100, 4092, 4086, 4078, 4068, 4060,
+ 4054, 4046, 4020, 4012, 4006, 3998, 3988, 3980, 3974, 3966, 3947, 3939,
+ 3933, 3925, 3915, 3907, 3901, 3893, 3871, 3863, 3857, 3849, 3839, 3831,
+ 3825, 3817, 3798, 3790, 3784, 3776, 3766, 3758, 3752, 3744, 6697, 6689,
+ 6683, 6675, 6665, 6657, 6651, 6643, 6624, 6616, 6610, 6602, 6592, 6584,
+ 6578, 6570, 6548, 6540, 6534, 6526, 6516, 6508, 6502, 6494, 6475, 6467,
+ 6461, 6453, 6443, 6435, 6429, 6421, 6395, 6387, 6381, 6373, 6363, 6355,
+ 6349, 6341, 6322, 6314, 6308, 6300, 6290, 6282, 6276, 6268, 6246, 6238,
+ 6232, 6224, 6214, 6206, 6200, 6192, 6173, 6165, 6159, 6151, 6141, 6133,
+ 6127, 6119, 6253, 6245, 6239, 6231, 6221, 6213, 6207, 6199, 6180, 6172,
+ 6166, 6158, 6148, 6140, 6134, 6126, 6104, 6096, 6090, 6082, 6072, 6064,
+ 6058, 6050, 6031, 6023, 6017, 6009, 5999, 5991, 5985, 5977, 5951, 5943,
+ 5937, 5929, 5919, 5911, 5905, 5897, 5878, 5870, 5864, 5856, 5846, 5838,
+ 5832, 5824, 5802, 5794, 5788, 5780, 5770, 5762, 5756, 5748, 5729, 5721,
+ 5715, 5707, 5697, 5689, 5683, 5675, 5877, 5869, 5863, 5855, 5845, 5837,
+ 5831, 5823, 5804, 5796, 5790, 5782, 5772, 5764, 5758, 5750, 5728, 5720,
+ 5714, 5706, 5696, 5688, 5682, 5674, 5655, 5647, 5641, 5633, 5623, 5615,
+ 5609, 5601, 5575, 5567, 5561, 5553, 5543, 5535, 5529, 5521, 5502, 5494,
+ 5488, 5480, 5470, 5462, 5456, 5448, 5426, 5418, 5412, 5404, 5394, 5386,
+ 5380, 5372, 5353, 5345, 5339, 5331, 5321, 5313, 5307, 5299, 5433, 5425,
+ 5419, 5411, 5401, 5393, 5387, 5379, 5360, 5352, 5346, 5338, 5328, 5320,
+ 5314, 5306, 5284, 5276, 5270, 5262, 5252, 5244, 5238, 5230, 5211, 5203,
+ 5197, 5189, 5179, 5171, 5165, 5157, 5131, 5123, 5117, 5109, 5099, 5091,
+ 5085, 5077, 5058, 5050, 5044, 5036, 5026, 5018, 5012, 5004, 4982, 4974,
+ 4968, 4960, 4950, 4942, 4936, 4928, 4909, 4901, 4895, 4887, 4877, 4869,
+ 4863, 4855, 5586, 5578, 5572, 5564, 5554, 5546, 5540, 5532, 5513, 5505,
+ 5499, 5491, 5481, 5473, 5467, 5459, 5437, 5429, 5423, 5415, 5405, 5397,
+ 5391, 5383, 5364, 5356, 5350, 5342, 5332, 5324, 5318, 5310, 5284, 5276,
+ 5270, 5262, 5252, 5244, 5238, 5230, 5211, 5203, 5197, 5189, 5179, 5171,
+ 5165, 5157, 5135, 5127, 5121, 5113, 5103, 5095, 5089, 5081, 5062, 5054,
+ 5048, 5040, 5030, 5022, 5016, 5008, 5142, 5134, 5128, 5120, 5110, 5102,
+ 5096, 5088, 5069, 5061, 5055, 5047, 5037, 5029, 5023, 5015, 4993, 4985,
+ 4979, 4971, 4961, 4953, 4947, 4939, 4920, 4912, 4906, 4898, 4888, 4880,
+ 4874, 4866, 4840, 4832, 4826, 4818, 4808, 4800, 4794, 4786, 4767, 4759,
+ 4753, 4745, 4735, 4727, 4721, 4713, 4691, 4683, 4677, 4669, 4659, 4651,
+ 4645, 4637, 4618, 4610, 4604, 4596, 4586, 4578, 4572, 4564, 4766, 4758,
+ 4752, 4744, 4734, 4726, 4720, 4712, 4693, 4685, 4679, 4671, 4661, 4653,
+ 4647, 4639, 4617, 4609, 4603, 4595, 4585, 4577, 4571, 4563, 4544, 4536,
+ 4530, 4522, 4512, 4504, 4498, 4490, 4464, 4456, 4450, 4442, 4432, 4424,
+ 4418, 4410, 4391, 4383, 4377, 4369, 4359, 4351, 4345, 4337, 4315, 4307,
+ 4301, 4293, 4283, 4275, 4269, 4261, 4242, 4234, 4228, 4220, 4210, 4202,
+ 4196, 4188, 4322, 4314, 4308, 4300, 4290, 4282, 4276, 4268, 4249, 4241,
+ 4235, 4227, 4217, 4209, 4203, 4195, 4173, 4165, 4159, 4151, 4141, 4133,
+ 4127, 4119, 4100, 4092, 4086, 4078, 4068, 4060, 4054, 4046, 4020, 4012,
+ 4006, 3998, 3988, 3980, 3974, 3966, 3947, 3939, 3933, 3925, 3915, 3907,
+ 3901, 3893, 3871, 3863, 3857, 3849, 3839, 3831, 3825, 3817, 3798, 3790,
+ 3784, 3776, 3766, 3758, 3752, 3744, 4651, 4643, 4637, 4629, 4619, 4611,
+ 4605, 4597, 4578, 4570, 4564, 4556, 4546, 4538, 4532, 4524, 4502, 4494,
+ 4488, 4480, 4470, 4462, 4456, 4448, 4429, 4421, 4415, 4407, 4397, 4389,
+ 4383, 4375, 4349, 4341, 4335, 4327, 4317, 4309, 4303, 4295, 4276, 4268,
+ 4262, 4254, 4244, 4236, 4230, 4222, 4200, 4192, 4186, 4178, 4168, 4160,
+ 4154, 4146, 4127, 4119, 4113, 4105, 4095, 4087, 4081, 4073, 4207, 4199,
+ 4193, 4185, 4175, 4167, 4161, 4153, 4134, 4126, 4120, 4112, 4102, 4094,
+ 4088, 4080, 4058, 4050, 4044, 4036, 4026, 4018, 4012, 4004, 3985, 3977,
+ 3971, 3963, 3953, 3945, 3939, 3931, 3905, 3897, 3891, 3883, 3873, 3865,
+ 3859, 3851, 3832, 3824, 3818, 3810, 3800, 3792, 3786, 3778, 3756, 3748,
+ 3742, 3734, 3724, 3716, 3710, 3702, 3683, 3675, 3669, 3661, 3651, 3643,
+ 3637, 3629, 3831, 3823, 3817, 3809, 3799, 3791, 3785, 3777, 3758, 3750,
+ 3744, 3736, 3726, 3718, 3712, 3704, 3682, 3674, 3668, 3660, 3650, 3642,
+ 3636, 3628, 3609, 3601, 3595, 3587, 3577, 3569, 3563, 3555, 3529, 3521,
+ 3515, 3507, 3497, 3489, 3483, 3475, 3456, 3448, 3442, 3434, 3424, 3416,
+ 3410, 3402, 3380, 3372, 3366, 3358, 3348, 3340, 3334, 3326, 3307, 3299,
+ 3293, 3285, 3275, 3267, 3261, 3253, 3387, 3379, 3373, 3365, 3355, 3347,
+ 3341, 3333, 3314, 3306, 3300, 3292, 3282, 3274, 3268, 3260, 3238, 3230,
+ 3224, 3216, 3206, 3198, 3192, 3184, 3165, 3157, 3151, 3143, 3133, 3125,
+ 3119, 3111, 3085, 3077, 3071, 3063, 3053, 3045, 3039, 3031, 3012, 3004,
+ 2998, 2990, 2980, 2972, 2966, 2958, 2936, 2928, 2922, 2914, 2904, 2896,
+ 2890, 2882, 2863, 2855, 2849, 2841, 2831, 2823, 2817, 2809, 3540, 3532,
+ 3526, 3518, 3508, 3500, 3494, 3486, 3467, 3459, 3453, 3445, 3435, 3427,
+ 3421, 3413, 3391, 3383, 3377, 3369, 3359, 3351, 3345, 3337, 3318, 3310,
+ 3304, 3296, 3286, 3278, 3272, 3264, 3238, 3230, 3224, 3216, 3206, 3198,
+ 3192, 3184, 3165, 3157, 3151, 3143, 3133, 3125, 3119, 3111, 3089, 3081,
+ 3075, 3067, 3057, 3049, 3043, 3035, 3016, 3008, 3002, 2994, 2984, 2976,
+ 2970, 2962, 3096, 3088, 3082, 3074, 3064, 3056, 3050, 3042, 3023, 3015,
+ 3009, 3001, 2991, 2983, 2977, 2969, 2947, 2939, 2933, 2925, 2915, 2907,
+ 2901, 2893, 2874, 2866, 2860, 2852, 2842, 2834, 2828, 2820, 2794, 2786,
+ 2780, 2772, 2762, 2754, 2748, 2740, 2721, 2713, 2707, 2699, 2689, 2681,
+ 2675, 2667, 2645, 2637, 2631, 2623, 2613, 2605, 2599, 2591, 2572, 2564,
+ 2558, 2550, 2540, 2532, 2526, 2518, 2720, 2712, 2706, 2698, 2688, 2680,
+ 2674, 2666, 2647, 2639, 2633, 2625, 2615, 2607, 2601, 2593, 2571, 2563,
+ 2557, 2549, 2539, 2531, 2525, 2517, 2498, 2490, 2484, 2476, 2466, 2458,
+ 2452, 2444, 2418, 2410, 2404, 2396, 2386, 2378, 2372, 2364, 2345, 2337,
+ 2331, 2323, 2313, 2305, 2299, 2291, 2269, 2261, 2255, 2247, 2237, 2229,
+ 2223, 2215, 2196, 2188, 2182, 2174, 2164, 2156, 2150, 2142, 2276, 2268,
+ 2262, 2254, 2244, 2236, 2230, 2222, 2203, 2195, 2189, 2181, 2171, 2163,
+ 2157, 2149, 2127, 2119, 2113, 2105, 2095, 2087, 2081, 2073, 2054, 2046,
+ 2040, 2032, 2022, 2014, 2008, 2000, 1974, 1966, 1960, 1952, 1942, 1934,
+ 1928, 1920, 1901, 1893, 1887, 1879, 1869, 1861, 1855, 1847, 1825, 1817,
+ 1811, 1803, 1793, 1785, 1779, 1771, 1752, 1744, 1738, 1730, 1720, 1712,
+ 1706, 1698, 1897, 1883, 1860, 1846, 1819, 1805, 1782, 1768, 1723, 1709,
+ 1686, 1672, 1645, 1631, 1608, 1594, 1574, 1560, 1537, 1523, 1496, 1482,
+ 1459, 1445, 1400, 1386, 1363, 1349, 1322, 1308, 1285, 1271, 1608, 1565,
+ 1535, 1492, 1446, 1403, 1373, 1330, 1312, 1269, 1239, 1196, 1150, 1107,
+ 1077, 1034, 1291, 1218, 1171, 1098, 1015, 942, 895, 822, 953, 850,
+ 729, 626, 618, 431, 257, 257, 257, 257, 0, 255, 255, 255,
+ 255, 429, 616, 624, 727, 848, 951, 820, 893, 940, 1013, 1096,
+ 1169, 1216, 1289, 1032, 1075, 1105, 1148, 1194, 1237, 1267, 1310, 1328,
+ 1371, 1401, 1444, 1490, 1533, 1563, 1606, 1269, 1283, 1306, 1320, 1347,
+ 1361, 1384, 1398, 1443, 1457, 1480, 1494, 1521, 1535, 1558, 1572, 1592,
+ 1606, 1629, 1643, 1670, 1684, 1707, 1721, 1766, 1780, 1803, 1817, 1844,
+ 1858, 1881, 1895, 1696, 1704, 1710, 1718, 1728, 1736, 1742, 1750, 1769,
+ 1777, 1783, 1791, 1801, 1809, 1815, 1823, 1845, 1853, 1859, 1867, 1877,
+ 1885, 1891, 1899, 1918, 1926, 1932, 1940, 1950, 1958, 1964, 1972, 1998,
+ 2006, 2012, 2020, 2030, 2038, 2044, 2052, 2071, 2079, 2085, 2093, 2103,
+ 2111, 2117, 2125, 2147, 2155, 2161, 2169, 2179, 2187, 2193, 2201, 2220,
+ 2228, 2234, 2242, 2252, 2260, 2266, 2274, 2140, 2148, 2154, 2162, 2172,
+ 2180, 2186, 2194, 2213, 2221, 2227, 2235, 2245, 2253, 2259, 2267, 2289,
+ 2297, 2303, 2311, 2321, 2329, 2335, 2343, 2362, 2370, 2376, 2384, 2394,
+ 2402, 2408, 2416, 2442, 2450, 2456, 2464, 2474, 2482, 2488, 2496, 2515,
+ 2523, 2529, 2537, 2547, 2555, 2561, 2569, 2591, 2599, 2605, 2613, 2623,
+ 2631, 2637, 2645, 2664, 2672, 2678, 2686, 2696, 2704, 2710, 2718, 2516,
+ 2524, 2530, 2538, 2548, 2556, 2562, 2570, 2589, 2597, 2603, 2611, 2621,
+ 2629, 2635, 2643, 2665, 2673, 2679, 2687, 2697, 2705, 2711, 2719, 2738,
+ 2746, 2752, 2760, 2770, 2778, 2784, 2792, 2818, 2826, 2832, 2840, 2850,
+ 2858, 2864, 2872, 2891, 2899, 2905, 2913, 2923, 2931, 2937, 2945, 2967,
+ 2975, 2981, 2989, 2999, 3007, 3013, 3021, 3040, 3048, 3054, 3062, 3072,
+ 3080, 3086, 3094, 2960, 2968, 2974, 2982, 2992, 3000, 3006, 3014, 3033,
+ 3041, 3047, 3055, 3065, 3073, 3079, 3087, 3109, 3117, 3123, 3131, 3141,
+ 3149, 3155, 3163, 3182, 3190, 3196, 3204, 3214, 3222, 3228, 3236, 3262,
+ 3270, 3276, 3284, 3294, 3302, 3308, 3316, 3335, 3343, 3349, 3357, 3367,
+ 3375, 3381, 3389, 3411, 3419, 3425, 3433, 3443, 3451, 3457, 3465, 3484,
+ 3492, 3498, 3506, 3516, 3524, 3530, 3538, 2807, 2815, 2821, 2829, 2839,
+ 2847, 2853, 2861, 2880, 2888, 2894, 2902, 2912, 2920, 2926, 2934, 2956,
+ 2964, 2970, 2978, 2988, 2996, 3002, 3010, 3029, 3037, 3043, 3051, 3061,
+ 3069, 3075, 3083, 3109, 3117, 3123, 3131, 3141, 3149, 3155, 3163, 3182,
+ 3190, 3196, 3204, 3214, 3222, 3228, 3236, 3258, 3266, 3272, 3280, 3290,
+ 3298, 3304, 3312, 3331, 3339, 3345, 3353, 3363, 3371, 3377, 3385, 3251,
+ 3259, 3265, 3273, 3283, 3291, 3297, 3305, 3324, 3332, 3338, 3346, 3356,
+ 3364, 3370, 3378, 3400, 3408, 3414, 3422, 3432, 3440, 3446, 3454, 3473,
+ 3481, 3487, 3495, 3505, 3513, 3519, 3527, 3553, 3561, 3567, 3575, 3585,
+ 3593, 3599, 3607, 3626, 3634, 3640, 3648, 3658, 3666, 3672, 3680, 3702,
+ 3710, 3716, 3724, 3734, 3742, 3748, 3756, 3775, 3783, 3789, 3797, 3807,
+ 3815, 3821, 3829, 3627, 3635, 3641, 3649, 3659, 3667, 3673, 3681, 3700,
+ 3708, 3714, 3722, 3732, 3740, 3746, 3754, 3776, 3784, 3790, 3798, 3808,
+ 3816, 3822, 3830, 3849, 3857, 3863, 3871, 3881, 3889, 3895, 3903, 3929,
+ 3937, 3943, 3951, 3961, 3969, 3975, 3983, 4002, 4010, 4016, 4024, 4034,
+ 4042, 4048, 4056, 4078, 4086, 4092, 4100, 4110, 4118, 4124, 4132, 4151,
+ 4159, 4165, 4173, 4183, 4191, 4197, 4205, 4071, 4079, 4085, 4093, 4103,
+ 4111, 4117, 4125, 4144, 4152, 4158, 4166, 4176, 4184, 4190, 4198, 4220,
+ 4228, 4234, 4242, 4252, 4260, 4266, 4274, 4293, 4301, 4307, 4315, 4325,
+ 4333, 4339, 4347, 4373, 4381, 4387, 4395, 4405, 4413, 4419, 4427, 4446,
+ 4454, 4460, 4468, 4478, 4486, 4492, 4500, 4522, 4530, 4536, 4544, 4554,
+ 4562, 4568, 4576, 4595, 4603, 4609, 4617, 4627, 4635, 4641, 4649, 3742,
+ 3750, 3756, 3764, 3774, 3782, 3788, 3796, 3815, 3823, 3829, 3837, 3847,
+ 3855, 3861, 3869, 3891, 3899, 3905, 3913, 3923, 3931, 3937, 3945, 3964,
+ 3972, 3978, 3986, 3996, 4004, 4010, 4018, 4044, 4052, 4058, 4066, 4076,
+ 4084, 4090, 4098, 4117, 4125, 4131, 4139, 4149, 4157, 4163, 4171, 4193,
+ 4201, 4207, 4215, 4225, 4233, 4239, 4247, 4266, 4274, 4280, 4288, 4298,
+ 4306, 4312, 4320, 4186, 4194, 4200, 4208, 4218, 4226, 4232, 4240, 4259,
+ 4267, 4273, 4281, 4291, 4299, 4305, 4313, 4335, 4343, 4349, 4357, 4367,
+ 4375, 4381, 4389, 4408, 4416, 4422, 4430, 4440, 4448, 4454, 4462, 4488,
+ 4496, 4502, 4510, 4520, 4528, 4534, 4542, 4561, 4569, 4575, 4583, 4593,
+ 4601, 4607, 4615, 4637, 4645, 4651, 4659, 4669, 4677, 4683, 4691, 4710,
+ 4718, 4724, 4732, 4742, 4750, 4756, 4764, 4562, 4570, 4576, 4584, 4594,
+ 4602, 4608, 4616, 4635, 4643, 4649, 4657, 4667, 4675, 4681, 4689, 4711,
+ 4719, 4725, 4733, 4743, 4751, 4757, 4765, 4784, 4792, 4798, 4806, 4816,
+ 4824, 4830, 4838, 4864, 4872, 4878, 4886, 4896, 4904, 4910, 4918, 4937,
+ 4945, 4951, 4959, 4969, 4977, 4983, 4991, 5013, 5021, 5027, 5035, 5045,
+ 5053, 5059, 5067, 5086, 5094, 5100, 5108, 5118, 5126, 5132, 5140, 5006,
+ 5014, 5020, 5028, 5038, 5046, 5052, 5060, 5079, 5087, 5093, 5101, 5111,
+ 5119, 5125, 5133, 5155, 5163, 5169, 5177, 5187, 5195, 5201, 5209, 5228,
+ 5236, 5242, 5250, 5260, 5268, 5274, 5282, 5308, 5316, 5322, 5330, 5340,
+ 5348, 5354, 5362, 5381, 5389, 5395, 5403, 5413, 5421, 5427, 5435, 5457,
+ 5465, 5471, 5479, 5489, 5497, 5503, 5511, 5530, 5538, 5544, 5552, 5562,
+ 5570, 5576, 5584, 4853, 4861, 4867, 4875, 4885, 4893, 4899, 4907, 4926,
+ 4934, 4940, 4948, 4958, 4966, 4972, 4980, 5002, 5010, 5016, 5024, 5034,
+ 5042, 5048, 5056, 5075, 5083, 5089, 5097, 5107, 5115, 5121, 5129, 5155,
+ 5163, 5169, 5177, 5187, 5195, 5201, 5209, 5228, 5236, 5242, 5250, 5260,
+ 5268, 5274, 5282, 5304, 5312, 5318, 5326, 5336, 5344, 5350, 5358, 5377,
+ 5385, 5391, 5399, 5409, 5417, 5423, 5431, 5297, 5305, 5311, 5319, 5329,
+ 5337, 5343, 5351, 5370, 5378, 5384, 5392, 5402, 5410, 5416, 5424, 5446,
+ 5454, 5460, 5468, 5478, 5486, 5492, 5500, 5519, 5527, 5533, 5541, 5551,
+ 5559, 5565, 5573, 5599, 5607, 5613, 5621, 5631, 5639, 5645, 5653, 5672,
+ 5680, 5686, 5694, 5704, 5712, 5718, 5726, 5748, 5756, 5762, 5770, 5780,
+ 5788, 5794, 5802, 5821, 5829, 5835, 5843, 5853, 5861, 5867, 5875, 5673,
+ 5681, 5687, 5695, 5705, 5713, 5719, 5727, 5746, 5754, 5760, 5768, 5778,
+ 5786, 5792, 5800, 5822, 5830, 5836, 5844, 5854, 5862, 5868, 5876, 5895,
+ 5903, 5909, 5917, 5927, 5935, 5941, 5949, 5975, 5983, 5989, 5997, 6007,
+ 6015, 6021, 6029, 6048, 6056, 6062, 6070, 6080, 6088, 6094, 6102, 6124,
+ 6132, 6138, 6146, 6156, 6164, 6170, 6178, 6197, 6205, 6211, 6219, 6229,
+ 6237, 6243, 6251, 6117, 6125, 6131, 6139, 6149, 6157, 6163, 6171, 6190,
+ 6198, 6204, 6212, 6222, 6230, 6236, 6244, 6266, 6274, 6280, 6288, 6298,
+ 6306, 6312, 6320, 6339, 6347, 6353, 6361, 6371, 6379, 6385, 6393, 6419,
+ 6427, 6433, 6441, 6451, 6459, 6465, 6473, 6492, 6500, 6506, 6514, 6524,
+ 6532, 6538, 6546, 6568, 6576, 6582, 6590, 6600, 6608, 6614, 6622, 6641,
+ 6649, 6655, 6663, 6673, 6681, 6687, 6695, 3742, 3750, 3756, 3764, 3774,
+ 3782, 3788, 3796, 3815, 3823, 3829, 3837, 3847, 3855, 3861, 3869, 3891,
+ 3899, 3905, 3913, 3923, 3931, 3937, 3945, 3964, 3972, 3978, 3986, 3996,
+ 4004, 4010, 4018, 4044, 4052, 4058, 4066, 4076, 4084, 4090, 4098, 4117,
+ 4125, 4131, 4139, 4149, 4157, 4163, 4171, 4193, 4201, 4207, 4215, 4225,
+ 4233, 4239, 4247, 4266, 4274, 4280, 4288, 4298, 4306, 4312, 4320, 4186,
+ 4194, 4200, 4208, 4218, 4226, 4232, 4240, 4259, 4267, 4273, 4281, 4291,
+ 4299, 4305, 4313, 4335, 4343, 4349, 4357, 4367, 4375, 4381, 4389, 4408,
+ 4416, 4422, 4430, 4440, 4448, 4454, 4462, 4488, 4496, 4502, 4510, 4520,
+ 4528, 4534, 4542, 4561, 4569, 4575, 4583, 4593, 4601, 4607, 4615, 4637,
+ 4645, 4651, 4659, 4669, 4677, 4683, 4691, 4710, 4718, 4724, 4732, 4742,
+ 4750, 4756, 4764, 4562, 4570, 4576, 4584, 4594, 4602, 4608, 4616, 4635,
+ 4643, 4649, 4657, 4667, 4675, 4681, 4689, 4711, 4719, 4725, 4733, 4743,
+ 4751, 4757, 4765, 4784, 4792, 4798, 4806, 4816, 4824, 4830, 4838, 4864,
+ 4872, 4878, 4886, 4896, 4904, 4910, 4918, 4937, 4945, 4951, 4959, 4969,
+ 4977, 4983, 4991, 5013, 5021, 5027, 5035, 5045, 5053, 5059, 5067, 5086,
+ 5094, 5100, 5108, 5118, 5126, 5132, 5140, 5006, 5014, 5020, 5028, 5038,
+ 5046, 5052, 5060, 5079, 5087, 5093, 5101, 5111, 5119, 5125, 5133, 5155,
+ 5163, 5169, 5177, 5187, 5195, 5201, 5209, 5228, 5236, 5242, 5250, 5260,
+ 5268, 5274, 5282, 5308, 5316, 5322, 5330, 5340, 5348, 5354, 5362, 5381,
+ 5389, 5395, 5403, 5413, 5421, 5427, 5435, 5457, 5465, 5471, 5479, 5489,
+ 5497, 5503, 5511, 5530, 5538, 5544, 5552, 5562, 5570, 5576, 5584, 4853,
+ 4861, 4867, 4875, 4885, 4893, 4899, 4907, 4926, 4934, 4940, 4948, 4958,
+ 4966, 4972, 4980, 5002, 5010, 5016, 5024, 5034, 5042, 5048, 5056, 5075,
+ 5083, 5089, 5097, 5107, 5115, 5121, 5129, 5155, 5163, 5169, 5177, 5187,
+ 5195, 5201, 5209, 5228, 5236, 5242, 5250, 5260, 5268, 5274, 5282, 5304,
+ 5312, 5318, 5326, 5336, 5344, 5350, 5358, 5377, 5385, 5391, 5399, 5409,
+ 5417, 5423, 5431, 5297, 5305, 5311, 5319, 5329, 5337, 5343, 5351, 5370,
+ 5378, 5384, 5392, 5402, 5410, 5416, 5424, 5446, 5454, 5460, 5468, 5478,
+ 5486, 5492, 5500, 5519, 5527, 5533, 5541, 5551, 5559, 5565, 5573, 5599,
+ 5607, 5613, 5621, 5631, 5639, 5645, 5653, 5672, 5680, 5686, 5694, 5704,
+ 5712, 5718, 5726, 5748, 5756, 5762, 5770, 5780, 5788, 5794, 5802, 5821,
+ 5829, 5835, 5843, 5853, 5861, 5867, 5875, 5673, 5681, 5687, 5695, 5705,
+ 5713, 5719, 5727, 5746, 5754, 5760, 5768, 5778, 5786, 5792, 5800, 5822,
+ 5830, 5836, 5844, 5854, 5862, 5868, 5876, 5895, 5903, 5909, 5917, 5927,
+ 5935, 5941, 5949, 5975, 5983, 5989, 5997, 6007, 6015, 6021, 6029, 6048,
+ 6056, 6062, 6070, 6080, 6088, 6094, 6102, 6124, 6132, 6138, 6146, 6156,
+ 6164, 6170, 6178, 6197, 6205, 6211, 6219, 6229, 6237, 6243, 6251, 6117,
+ 6125, 6131, 6139, 6149, 6157, 6163, 6171, 6190, 6198, 6204, 6212, 6222,
+ 6230, 6236, 6244, 6266, 6274, 6280, 6288, 6298, 6306, 6312, 6320, 6339,
+ 6347, 6353, 6361, 6371, 6379, 6385, 6393, 6419, 6427, 6433, 6441, 6451,
+ 6459, 6465, 6473, 6492, 6500, 6506, 6514, 6524, 6532, 6538, 6546, 6568,
+ 6576, 6582, 6590, 6600, 6608, 6614, 6622, 6641, 6649, 6655, 6663, 6673,
+ 6681, 6687, 6695, 5788, 5796, 5802, 5810, 5820, 5828, 5834, 5842, 5861,
+ 5869, 5875, 5883, 5893, 5901, 5907, 5915, 5937, 5945, 5951, 5959, 5969,
+ 5977, 5983, 5991, 6010, 6018, 6024, 6032, 6042, 6050, 6056, 6064, 6090,
+ 6098, 6104, 6112, 6122, 6130, 6136, 6144, 6163, 6171, 6177, 6185, 6195,
+ 6203, 6209, 6217, 6239, 6247, 6253, 6261, 6271, 6279, 6285, 6293, 6312,
+ 6320, 6326, 6334, 6344, 6352, 6358, 6366, 6232, 6240, 6246, 6254, 6264,
+ 6272, 6278, 6286, 6305, 6313, 6319, 6327, 6337, 6345, 6351, 6359, 6381,
+ 6389, 6395, 6403, 6413, 6421, 6427, 6435, 6454, 6462, 6468, 6476, 6486,
+ 6494, 6500, 6508, 6534, 6542, 6548, 6556, 6566, 6574, 6580, 6588, 6607,
+ 6615, 6621, 6629, 6639, 6647, 6653, 6661, 6683, 6691, 6697, 6705, 6715,
+ 6723, 6729, 6737, 6756, 6764, 6770, 6778, 6788, 6796, 6802, 6810, 6608,
+ 6616, 6622, 6630, 6640, 6648, 6654, 6662, 6681, 6689, 6695, 6703, 6713,
+ 6721, 6727, 6735, 6757, 6765, 6771, 6779, 6789, 6797, 6803, 6811, 6830,
+ 6838, 6844, 6852, 6862, 6870, 6876, 6884, 6910, 6918, 6924, 6932, 6942,
+ 6950, 6956, 6964, 6983, 6991, 6997, 7005, 7015, 7023, 7029, 7037, 7059,
+ 7067, 7073, 7081, 7091, 7099, 7105, 7113, 7132, 7140, 7146, 7154, 7164,
+ 7172, 7178, 7186, 7052, 7060, 7066, 7074, 7084, 7092, 7098, 7106, 7125,
+ 7133, 7139, 7147, 7157, 7165, 7171, 7179, 7201, 7209, 7215, 7223, 7233,
+ 7241, 7247, 7255, 7274, 7282, 7288, 7296, 7306, 7314, 7320, 7328, 7354,
+ 7362, 7368, 7376, 7386, 7394, 7400, 7408, 7427, 7435, 7441, 7449, 7459,
+ 7467, 7473, 7481, 7503, 7511, 7517, 7525, 7535, 7543, 7549, 7557, 7576,
+ 7584, 7590, 7598, 7608, 7616, 7622, 7630, 6899, 6907, 6913, 6921, 6931,
+ 6939, 6945, 6953, 6972, 6980, 6986, 6994, 7004, 7012, 7018, 7026, 7048,
+ 7056, 7062, 7070, 7080, 7088, 7094, 7102, 7121, 7129, 7135, 7143, 7153,
+ 7161, 7167, 7175, 7201, 7209, 7215, 7223, 7233, 7241, 7247, 7255, 7274,
+ 7282, 7288, 7296, 7306, 7314, 7320, 7328, 7350, 7358, 7364, 7372, 7382,
+ 7390, 7396, 7404, 7423, 7431, 7437, 7445, 7455, 7463, 7469, 7477, 7343,
+ 7351, 7357, 7365, 7375, 7383, 7389, 7397, 7416, 7424, 7430, 7438, 7448,
+ 7456, 7462, 7470, 7492, 7500, 7506, 7514, 7524, 7532, 7538, 7546, 7565,
+ 7573, 7579, 7587, 7597, 7605, 7611, 7619, 7645, 7653, 7659, 7667, 7677,
+ 7685, 7691, 7699, 7718, 7726, 7732, 7740, 7750, 7758, 7764, 7772, 7794,
+ 7802, 7808, 7816, 7826, 7834, 7840, 7848, 7867, 7875, 7881, 7889, 7899,
+ 7907, 7913, 7921, 7719, 7727, 7733, 7741, 7751, 7759, 7765, 7773, 7792,
+ 7800, 7806, 7814, 7824, 7832, 7838, 7846, 7868, 7876, 7882, 7890, 7900,
+ 7908, 7914, 7922, 7941, 7949, 7955, 7963, 7973, 7981, 7987, 7995, 8021,
+ 8029, 8035, 8043, 8053, 8061, 8067, 8075, 8094, 8102, 8108, 8116, 8126,
+ 8134, 8140, 8148, 8170, 8178, 8184, 8192, 8202, 8210, 8216, 8224, 8243,
+ 8251, 8257, 8265, 8275
+};
diff --git a/libvpx/vp8/encoder/dct_value_tokens.h b/libvpx/vp8/encoder/dct_value_tokens.h
new file mode 100644
index 0000000..ef08eed
--- /dev/null
+++ b/libvpx/vp8/encoder/dct_value_tokens.h
@@ -0,0 +1,699 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+/* Generated file, included by tokenize.c */
+/* Values generated by fill_value_tokens() */
+
+static const TOKENVALUE dct_value_tokens[2048*2] =
+{
+ {10, 3963}, {10, 3961}, {10, 3959}, {10, 3957}, {10, 3955}, {10, 3953},
+ {10, 3951}, {10, 3949}, {10, 3947}, {10, 3945}, {10, 3943}, {10, 3941},
+ {10, 3939}, {10, 3937}, {10, 3935}, {10, 3933}, {10, 3931}, {10, 3929},
+ {10, 3927}, {10, 3925}, {10, 3923}, {10, 3921}, {10, 3919}, {10, 3917},
+ {10, 3915}, {10, 3913}, {10, 3911}, {10, 3909}, {10, 3907}, {10, 3905},
+ {10, 3903}, {10, 3901}, {10, 3899}, {10, 3897}, {10, 3895}, {10, 3893},
+ {10, 3891}, {10, 3889}, {10, 3887}, {10, 3885}, {10, 3883}, {10, 3881},
+ {10, 3879}, {10, 3877}, {10, 3875}, {10, 3873}, {10, 3871}, {10, 3869},
+ {10, 3867}, {10, 3865}, {10, 3863}, {10, 3861}, {10, 3859}, {10, 3857},
+ {10, 3855}, {10, 3853}, {10, 3851}, {10, 3849}, {10, 3847}, {10, 3845},
+ {10, 3843}, {10, 3841}, {10, 3839}, {10, 3837}, {10, 3835}, {10, 3833},
+ {10, 3831}, {10, 3829}, {10, 3827}, {10, 3825}, {10, 3823}, {10, 3821},
+ {10, 3819}, {10, 3817}, {10, 3815}, {10, 3813}, {10, 3811}, {10, 3809},
+ {10, 3807}, {10, 3805}, {10, 3803}, {10, 3801}, {10, 3799}, {10, 3797},
+ {10, 3795}, {10, 3793}, {10, 3791}, {10, 3789}, {10, 3787}, {10, 3785},
+ {10, 3783}, {10, 3781}, {10, 3779}, {10, 3777}, {10, 3775}, {10, 3773},
+ {10, 3771}, {10, 3769}, {10, 3767}, {10, 3765}, {10, 3763}, {10, 3761},
+ {10, 3759}, {10, 3757}, {10, 3755}, {10, 3753}, {10, 3751}, {10, 3749},
+ {10, 3747}, {10, 3745}, {10, 3743}, {10, 3741}, {10, 3739}, {10, 3737},
+ {10, 3735}, {10, 3733}, {10, 3731}, {10, 3729}, {10, 3727}, {10, 3725},
+ {10, 3723}, {10, 3721}, {10, 3719}, {10, 3717}, {10, 3715}, {10, 3713},
+ {10, 3711}, {10, 3709}, {10, 3707}, {10, 3705}, {10, 3703}, {10, 3701},
+ {10, 3699}, {10, 3697}, {10, 3695}, {10, 3693}, {10, 3691}, {10, 3689},
+ {10, 3687}, {10, 3685}, {10, 3683}, {10, 3681}, {10, 3679}, {10, 3677},
+ {10, 3675}, {10, 3673}, {10, 3671}, {10, 3669}, {10, 3667}, {10, 3665},
+ {10, 3663}, {10, 3661}, {10, 3659}, {10, 3657}, {10, 3655}, {10, 3653},
+ {10, 3651}, {10, 3649}, {10, 3647}, {10, 3645}, {10, 3643}, {10, 3641},
+ {10, 3639}, {10, 3637}, {10, 3635}, {10, 3633}, {10, 3631}, {10, 3629},
+ {10, 3627}, {10, 3625}, {10, 3623}, {10, 3621}, {10, 3619}, {10, 3617},
+ {10, 3615}, {10, 3613}, {10, 3611}, {10, 3609}, {10, 3607}, {10, 3605},
+ {10, 3603}, {10, 3601}, {10, 3599}, {10, 3597}, {10, 3595}, {10, 3593},
+ {10, 3591}, {10, 3589}, {10, 3587}, {10, 3585}, {10, 3583}, {10, 3581},
+ {10, 3579}, {10, 3577}, {10, 3575}, {10, 3573}, {10, 3571}, {10, 3569},
+ {10, 3567}, {10, 3565}, {10, 3563}, {10, 3561}, {10, 3559}, {10, 3557},
+ {10, 3555}, {10, 3553}, {10, 3551}, {10, 3549}, {10, 3547}, {10, 3545},
+ {10, 3543}, {10, 3541}, {10, 3539}, {10, 3537}, {10, 3535}, {10, 3533},
+ {10, 3531}, {10, 3529}, {10, 3527}, {10, 3525}, {10, 3523}, {10, 3521},
+ {10, 3519}, {10, 3517}, {10, 3515}, {10, 3513}, {10, 3511}, {10, 3509},
+ {10, 3507}, {10, 3505}, {10, 3503}, {10, 3501}, {10, 3499}, {10, 3497},
+ {10, 3495}, {10, 3493}, {10, 3491}, {10, 3489}, {10, 3487}, {10, 3485},
+ {10, 3483}, {10, 3481}, {10, 3479}, {10, 3477}, {10, 3475}, {10, 3473},
+ {10, 3471}, {10, 3469}, {10, 3467}, {10, 3465}, {10, 3463}, {10, 3461},
+ {10, 3459}, {10, 3457}, {10, 3455}, {10, 3453}, {10, 3451}, {10, 3449},
+ {10, 3447}, {10, 3445}, {10, 3443}, {10, 3441}, {10, 3439}, {10, 3437},
+ {10, 3435}, {10, 3433}, {10, 3431}, {10, 3429}, {10, 3427}, {10, 3425},
+ {10, 3423}, {10, 3421}, {10, 3419}, {10, 3417}, {10, 3415}, {10, 3413},
+ {10, 3411}, {10, 3409}, {10, 3407}, {10, 3405}, {10, 3403}, {10, 3401},
+ {10, 3399}, {10, 3397}, {10, 3395}, {10, 3393}, {10, 3391}, {10, 3389},
+ {10, 3387}, {10, 3385}, {10, 3383}, {10, 3381}, {10, 3379}, {10, 3377},
+ {10, 3375}, {10, 3373}, {10, 3371}, {10, 3369}, {10, 3367}, {10, 3365},
+ {10, 3363}, {10, 3361}, {10, 3359}, {10, 3357}, {10, 3355}, {10, 3353},
+ {10, 3351}, {10, 3349}, {10, 3347}, {10, 3345}, {10, 3343}, {10, 3341},
+ {10, 3339}, {10, 3337}, {10, 3335}, {10, 3333}, {10, 3331}, {10, 3329},
+ {10, 3327}, {10, 3325}, {10, 3323}, {10, 3321}, {10, 3319}, {10, 3317},
+ {10, 3315}, {10, 3313}, {10, 3311}, {10, 3309}, {10, 3307}, {10, 3305},
+ {10, 3303}, {10, 3301}, {10, 3299}, {10, 3297}, {10, 3295}, {10, 3293},
+ {10, 3291}, {10, 3289}, {10, 3287}, {10, 3285}, {10, 3283}, {10, 3281},
+ {10, 3279}, {10, 3277}, {10, 3275}, {10, 3273}, {10, 3271}, {10, 3269},
+ {10, 3267}, {10, 3265}, {10, 3263}, {10, 3261}, {10, 3259}, {10, 3257},
+ {10, 3255}, {10, 3253}, {10, 3251}, {10, 3249}, {10, 3247}, {10, 3245},
+ {10, 3243}, {10, 3241}, {10, 3239}, {10, 3237}, {10, 3235}, {10, 3233},
+ {10, 3231}, {10, 3229}, {10, 3227}, {10, 3225}, {10, 3223}, {10, 3221},
+ {10, 3219}, {10, 3217}, {10, 3215}, {10, 3213}, {10, 3211}, {10, 3209},
+ {10, 3207}, {10, 3205}, {10, 3203}, {10, 3201}, {10, 3199}, {10, 3197},
+ {10, 3195}, {10, 3193}, {10, 3191}, {10, 3189}, {10, 3187}, {10, 3185},
+ {10, 3183}, {10, 3181}, {10, 3179}, {10, 3177}, {10, 3175}, {10, 3173},
+ {10, 3171}, {10, 3169}, {10, 3167}, {10, 3165}, {10, 3163}, {10, 3161},
+ {10, 3159}, {10, 3157}, {10, 3155}, {10, 3153}, {10, 3151}, {10, 3149},
+ {10, 3147}, {10, 3145}, {10, 3143}, {10, 3141}, {10, 3139}, {10, 3137},
+ {10, 3135}, {10, 3133}, {10, 3131}, {10, 3129}, {10, 3127}, {10, 3125},
+ {10, 3123}, {10, 3121}, {10, 3119}, {10, 3117}, {10, 3115}, {10, 3113},
+ {10, 3111}, {10, 3109}, {10, 3107}, {10, 3105}, {10, 3103}, {10, 3101},
+ {10, 3099}, {10, 3097}, {10, 3095}, {10, 3093}, {10, 3091}, {10, 3089},
+ {10, 3087}, {10, 3085}, {10, 3083}, {10, 3081}, {10, 3079}, {10, 3077},
+ {10, 3075}, {10, 3073}, {10, 3071}, {10, 3069}, {10, 3067}, {10, 3065},
+ {10, 3063}, {10, 3061}, {10, 3059}, {10, 3057}, {10, 3055}, {10, 3053},
+ {10, 3051}, {10, 3049}, {10, 3047}, {10, 3045}, {10, 3043}, {10, 3041},
+ {10, 3039}, {10, 3037}, {10, 3035}, {10, 3033}, {10, 3031}, {10, 3029},
+ {10, 3027}, {10, 3025}, {10, 3023}, {10, 3021}, {10, 3019}, {10, 3017},
+ {10, 3015}, {10, 3013}, {10, 3011}, {10, 3009}, {10, 3007}, {10, 3005},
+ {10, 3003}, {10, 3001}, {10, 2999}, {10, 2997}, {10, 2995}, {10, 2993},
+ {10, 2991}, {10, 2989}, {10, 2987}, {10, 2985}, {10, 2983}, {10, 2981},
+ {10, 2979}, {10, 2977}, {10, 2975}, {10, 2973}, {10, 2971}, {10, 2969},
+ {10, 2967}, {10, 2965}, {10, 2963}, {10, 2961}, {10, 2959}, {10, 2957},
+ {10, 2955}, {10, 2953}, {10, 2951}, {10, 2949}, {10, 2947}, {10, 2945},
+ {10, 2943}, {10, 2941}, {10, 2939}, {10, 2937}, {10, 2935}, {10, 2933},
+ {10, 2931}, {10, 2929}, {10, 2927}, {10, 2925}, {10, 2923}, {10, 2921},
+ {10, 2919}, {10, 2917}, {10, 2915}, {10, 2913}, {10, 2911}, {10, 2909},
+ {10, 2907}, {10, 2905}, {10, 2903}, {10, 2901}, {10, 2899}, {10, 2897},
+ {10, 2895}, {10, 2893}, {10, 2891}, {10, 2889}, {10, 2887}, {10, 2885},
+ {10, 2883}, {10, 2881}, {10, 2879}, {10, 2877}, {10, 2875}, {10, 2873},
+ {10, 2871}, {10, 2869}, {10, 2867}, {10, 2865}, {10, 2863}, {10, 2861},
+ {10, 2859}, {10, 2857}, {10, 2855}, {10, 2853}, {10, 2851}, {10, 2849},
+ {10, 2847}, {10, 2845}, {10, 2843}, {10, 2841}, {10, 2839}, {10, 2837},
+ {10, 2835}, {10, 2833}, {10, 2831}, {10, 2829}, {10, 2827}, {10, 2825},
+ {10, 2823}, {10, 2821}, {10, 2819}, {10, 2817}, {10, 2815}, {10, 2813},
+ {10, 2811}, {10, 2809}, {10, 2807}, {10, 2805}, {10, 2803}, {10, 2801},
+ {10, 2799}, {10, 2797}, {10, 2795}, {10, 2793}, {10, 2791}, {10, 2789},
+ {10, 2787}, {10, 2785}, {10, 2783}, {10, 2781}, {10, 2779}, {10, 2777},
+ {10, 2775}, {10, 2773}, {10, 2771}, {10, 2769}, {10, 2767}, {10, 2765},
+ {10, 2763}, {10, 2761}, {10, 2759}, {10, 2757}, {10, 2755}, {10, 2753},
+ {10, 2751}, {10, 2749}, {10, 2747}, {10, 2745}, {10, 2743}, {10, 2741},
+ {10, 2739}, {10, 2737}, {10, 2735}, {10, 2733}, {10, 2731}, {10, 2729},
+ {10, 2727}, {10, 2725}, {10, 2723}, {10, 2721}, {10, 2719}, {10, 2717},
+ {10, 2715}, {10, 2713}, {10, 2711}, {10, 2709}, {10, 2707}, {10, 2705},
+ {10, 2703}, {10, 2701}, {10, 2699}, {10, 2697}, {10, 2695}, {10, 2693},
+ {10, 2691}, {10, 2689}, {10, 2687}, {10, 2685}, {10, 2683}, {10, 2681},
+ {10, 2679}, {10, 2677}, {10, 2675}, {10, 2673}, {10, 2671}, {10, 2669},
+ {10, 2667}, {10, 2665}, {10, 2663}, {10, 2661}, {10, 2659}, {10, 2657},
+ {10, 2655}, {10, 2653}, {10, 2651}, {10, 2649}, {10, 2647}, {10, 2645},
+ {10, 2643}, {10, 2641}, {10, 2639}, {10, 2637}, {10, 2635}, {10, 2633},
+ {10, 2631}, {10, 2629}, {10, 2627}, {10, 2625}, {10, 2623}, {10, 2621},
+ {10, 2619}, {10, 2617}, {10, 2615}, {10, 2613}, {10, 2611}, {10, 2609},
+ {10, 2607}, {10, 2605}, {10, 2603}, {10, 2601}, {10, 2599}, {10, 2597},
+ {10, 2595}, {10, 2593}, {10, 2591}, {10, 2589}, {10, 2587}, {10, 2585},
+ {10, 2583}, {10, 2581}, {10, 2579}, {10, 2577}, {10, 2575}, {10, 2573},
+ {10, 2571}, {10, 2569}, {10, 2567}, {10, 2565}, {10, 2563}, {10, 2561},
+ {10, 2559}, {10, 2557}, {10, 2555}, {10, 2553}, {10, 2551}, {10, 2549},
+ {10, 2547}, {10, 2545}, {10, 2543}, {10, 2541}, {10, 2539}, {10, 2537},
+ {10, 2535}, {10, 2533}, {10, 2531}, {10, 2529}, {10, 2527}, {10, 2525},
+ {10, 2523}, {10, 2521}, {10, 2519}, {10, 2517}, {10, 2515}, {10, 2513},
+ {10, 2511}, {10, 2509}, {10, 2507}, {10, 2505}, {10, 2503}, {10, 2501},
+ {10, 2499}, {10, 2497}, {10, 2495}, {10, 2493}, {10, 2491}, {10, 2489},
+ {10, 2487}, {10, 2485}, {10, 2483}, {10, 2481}, {10, 2479}, {10, 2477},
+ {10, 2475}, {10, 2473}, {10, 2471}, {10, 2469}, {10, 2467}, {10, 2465},
+ {10, 2463}, {10, 2461}, {10, 2459}, {10, 2457}, {10, 2455}, {10, 2453},
+ {10, 2451}, {10, 2449}, {10, 2447}, {10, 2445}, {10, 2443}, {10, 2441},
+ {10, 2439}, {10, 2437}, {10, 2435}, {10, 2433}, {10, 2431}, {10, 2429},
+ {10, 2427}, {10, 2425}, {10, 2423}, {10, 2421}, {10, 2419}, {10, 2417},
+ {10, 2415}, {10, 2413}, {10, 2411}, {10, 2409}, {10, 2407}, {10, 2405},
+ {10, 2403}, {10, 2401}, {10, 2399}, {10, 2397}, {10, 2395}, {10, 2393},
+ {10, 2391}, {10, 2389}, {10, 2387}, {10, 2385}, {10, 2383}, {10, 2381},
+ {10, 2379}, {10, 2377}, {10, 2375}, {10, 2373}, {10, 2371}, {10, 2369},
+ {10, 2367}, {10, 2365}, {10, 2363}, {10, 2361}, {10, 2359}, {10, 2357},
+ {10, 2355}, {10, 2353}, {10, 2351}, {10, 2349}, {10, 2347}, {10, 2345},
+ {10, 2343}, {10, 2341}, {10, 2339}, {10, 2337}, {10, 2335}, {10, 2333},
+ {10, 2331}, {10, 2329}, {10, 2327}, {10, 2325}, {10, 2323}, {10, 2321},
+ {10, 2319}, {10, 2317}, {10, 2315}, {10, 2313}, {10, 2311}, {10, 2309},
+ {10, 2307}, {10, 2305}, {10, 2303}, {10, 2301}, {10, 2299}, {10, 2297},
+ {10, 2295}, {10, 2293}, {10, 2291}, {10, 2289}, {10, 2287}, {10, 2285},
+ {10, 2283}, {10, 2281}, {10, 2279}, {10, 2277}, {10, 2275}, {10, 2273},
+ {10, 2271}, {10, 2269}, {10, 2267}, {10, 2265}, {10, 2263}, {10, 2261},
+ {10, 2259}, {10, 2257}, {10, 2255}, {10, 2253}, {10, 2251}, {10, 2249},
+ {10, 2247}, {10, 2245}, {10, 2243}, {10, 2241}, {10, 2239}, {10, 2237},
+ {10, 2235}, {10, 2233}, {10, 2231}, {10, 2229}, {10, 2227}, {10, 2225},
+ {10, 2223}, {10, 2221}, {10, 2219}, {10, 2217}, {10, 2215}, {10, 2213},
+ {10, 2211}, {10, 2209}, {10, 2207}, {10, 2205}, {10, 2203}, {10, 2201},
+ {10, 2199}, {10, 2197}, {10, 2195}, {10, 2193}, {10, 2191}, {10, 2189},
+ {10, 2187}, {10, 2185}, {10, 2183}, {10, 2181}, {10, 2179}, {10, 2177},
+ {10, 2175}, {10, 2173}, {10, 2171}, {10, 2169}, {10, 2167}, {10, 2165},
+ {10, 2163}, {10, 2161}, {10, 2159}, {10, 2157}, {10, 2155}, {10, 2153},
+ {10, 2151}, {10, 2149}, {10, 2147}, {10, 2145}, {10, 2143}, {10, 2141},
+ {10, 2139}, {10, 2137}, {10, 2135}, {10, 2133}, {10, 2131}, {10, 2129},
+ {10, 2127}, {10, 2125}, {10, 2123}, {10, 2121}, {10, 2119}, {10, 2117},
+ {10, 2115}, {10, 2113}, {10, 2111}, {10, 2109}, {10, 2107}, {10, 2105},
+ {10, 2103}, {10, 2101}, {10, 2099}, {10, 2097}, {10, 2095}, {10, 2093},
+ {10, 2091}, {10, 2089}, {10, 2087}, {10, 2085}, {10, 2083}, {10, 2081},
+ {10, 2079}, {10, 2077}, {10, 2075}, {10, 2073}, {10, 2071}, {10, 2069},
+ {10, 2067}, {10, 2065}, {10, 2063}, {10, 2061}, {10, 2059}, {10, 2057},
+ {10, 2055}, {10, 2053}, {10, 2051}, {10, 2049}, {10, 2047}, {10, 2045},
+ {10, 2043}, {10, 2041}, {10, 2039}, {10, 2037}, {10, 2035}, {10, 2033},
+ {10, 2031}, {10, 2029}, {10, 2027}, {10, 2025}, {10, 2023}, {10, 2021},
+ {10, 2019}, {10, 2017}, {10, 2015}, {10, 2013}, {10, 2011}, {10, 2009},
+ {10, 2007}, {10, 2005}, {10, 2003}, {10, 2001}, {10, 1999}, {10, 1997},
+ {10, 1995}, {10, 1993}, {10, 1991}, {10, 1989}, {10, 1987}, {10, 1985},
+ {10, 1983}, {10, 1981}, {10, 1979}, {10, 1977}, {10, 1975}, {10, 1973},
+ {10, 1971}, {10, 1969}, {10, 1967}, {10, 1965}, {10, 1963}, {10, 1961},
+ {10, 1959}, {10, 1957}, {10, 1955}, {10, 1953}, {10, 1951}, {10, 1949},
+ {10, 1947}, {10, 1945}, {10, 1943}, {10, 1941}, {10, 1939}, {10, 1937},
+ {10, 1935}, {10, 1933}, {10, 1931}, {10, 1929}, {10, 1927}, {10, 1925},
+ {10, 1923}, {10, 1921}, {10, 1919}, {10, 1917}, {10, 1915}, {10, 1913},
+ {10, 1911}, {10, 1909}, {10, 1907}, {10, 1905}, {10, 1903}, {10, 1901},
+ {10, 1899}, {10, 1897}, {10, 1895}, {10, 1893}, {10, 1891}, {10, 1889},
+ {10, 1887}, {10, 1885}, {10, 1883}, {10, 1881}, {10, 1879}, {10, 1877},
+ {10, 1875}, {10, 1873}, {10, 1871}, {10, 1869}, {10, 1867}, {10, 1865},
+ {10, 1863}, {10, 1861}, {10, 1859}, {10, 1857}, {10, 1855}, {10, 1853},
+ {10, 1851}, {10, 1849}, {10, 1847}, {10, 1845}, {10, 1843}, {10, 1841},
+ {10, 1839}, {10, 1837}, {10, 1835}, {10, 1833}, {10, 1831}, {10, 1829},
+ {10, 1827}, {10, 1825}, {10, 1823}, {10, 1821}, {10, 1819}, {10, 1817},
+ {10, 1815}, {10, 1813}, {10, 1811}, {10, 1809}, {10, 1807}, {10, 1805},
+ {10, 1803}, {10, 1801}, {10, 1799}, {10, 1797}, {10, 1795}, {10, 1793},
+ {10, 1791}, {10, 1789}, {10, 1787}, {10, 1785}, {10, 1783}, {10, 1781},
+ {10, 1779}, {10, 1777}, {10, 1775}, {10, 1773}, {10, 1771}, {10, 1769},
+ {10, 1767}, {10, 1765}, {10, 1763}, {10, 1761}, {10, 1759}, {10, 1757},
+ {10, 1755}, {10, 1753}, {10, 1751}, {10, 1749}, {10, 1747}, {10, 1745},
+ {10, 1743}, {10, 1741}, {10, 1739}, {10, 1737}, {10, 1735}, {10, 1733},
+ {10, 1731}, {10, 1729}, {10, 1727}, {10, 1725}, {10, 1723}, {10, 1721},
+ {10, 1719}, {10, 1717}, {10, 1715}, {10, 1713}, {10, 1711}, {10, 1709},
+ {10, 1707}, {10, 1705}, {10, 1703}, {10, 1701}, {10, 1699}, {10, 1697},
+ {10, 1695}, {10, 1693}, {10, 1691}, {10, 1689}, {10, 1687}, {10, 1685},
+ {10, 1683}, {10, 1681}, {10, 1679}, {10, 1677}, {10, 1675}, {10, 1673},
+ {10, 1671}, {10, 1669}, {10, 1667}, {10, 1665}, {10, 1663}, {10, 1661},
+ {10, 1659}, {10, 1657}, {10, 1655}, {10, 1653}, {10, 1651}, {10, 1649},
+ {10, 1647}, {10, 1645}, {10, 1643}, {10, 1641}, {10, 1639}, {10, 1637},
+ {10, 1635}, {10, 1633}, {10, 1631}, {10, 1629}, {10, 1627}, {10, 1625},
+ {10, 1623}, {10, 1621}, {10, 1619}, {10, 1617}, {10, 1615}, {10, 1613},
+ {10, 1611}, {10, 1609}, {10, 1607}, {10, 1605}, {10, 1603}, {10, 1601},
+ {10, 1599}, {10, 1597}, {10, 1595}, {10, 1593}, {10, 1591}, {10, 1589},
+ {10, 1587}, {10, 1585}, {10, 1583}, {10, 1581}, {10, 1579}, {10, 1577},
+ {10, 1575}, {10, 1573}, {10, 1571}, {10, 1569}, {10, 1567}, {10, 1565},
+ {10, 1563}, {10, 1561}, {10, 1559}, {10, 1557}, {10, 1555}, {10, 1553},
+ {10, 1551}, {10, 1549}, {10, 1547}, {10, 1545}, {10, 1543}, {10, 1541},
+ {10, 1539}, {10, 1537}, {10, 1535}, {10, 1533}, {10, 1531}, {10, 1529},
+ {10, 1527}, {10, 1525}, {10, 1523}, {10, 1521}, {10, 1519}, {10, 1517},
+ {10, 1515}, {10, 1513}, {10, 1511}, {10, 1509}, {10, 1507}, {10, 1505},
+ {10, 1503}, {10, 1501}, {10, 1499}, {10, 1497}, {10, 1495}, {10, 1493},
+ {10, 1491}, {10, 1489}, {10, 1487}, {10, 1485}, {10, 1483}, {10, 1481},
+ {10, 1479}, {10, 1477}, {10, 1475}, {10, 1473}, {10, 1471}, {10, 1469},
+ {10, 1467}, {10, 1465}, {10, 1463}, {10, 1461}, {10, 1459}, {10, 1457},
+ {10, 1455}, {10, 1453}, {10, 1451}, {10, 1449}, {10, 1447}, {10, 1445},
+ {10, 1443}, {10, 1441}, {10, 1439}, {10, 1437}, {10, 1435}, {10, 1433},
+ {10, 1431}, {10, 1429}, {10, 1427}, {10, 1425}, {10, 1423}, {10, 1421},
+ {10, 1419}, {10, 1417}, {10, 1415}, {10, 1413}, {10, 1411}, {10, 1409},
+ {10, 1407}, {10, 1405}, {10, 1403}, {10, 1401}, {10, 1399}, {10, 1397},
+ {10, 1395}, {10, 1393}, {10, 1391}, {10, 1389}, {10, 1387}, {10, 1385},
+ {10, 1383}, {10, 1381}, {10, 1379}, {10, 1377}, {10, 1375}, {10, 1373},
+ {10, 1371}, {10, 1369}, {10, 1367}, {10, 1365}, {10, 1363}, {10, 1361},
+ {10, 1359}, {10, 1357}, {10, 1355}, {10, 1353}, {10, 1351}, {10, 1349},
+ {10, 1347}, {10, 1345}, {10, 1343}, {10, 1341}, {10, 1339}, {10, 1337},
+ {10, 1335}, {10, 1333}, {10, 1331}, {10, 1329}, {10, 1327}, {10, 1325},
+ {10, 1323}, {10, 1321}, {10, 1319}, {10, 1317}, {10, 1315}, {10, 1313},
+ {10, 1311}, {10, 1309}, {10, 1307}, {10, 1305}, {10, 1303}, {10, 1301},
+ {10, 1299}, {10, 1297}, {10, 1295}, {10, 1293}, {10, 1291}, {10, 1289},
+ {10, 1287}, {10, 1285}, {10, 1283}, {10, 1281}, {10, 1279}, {10, 1277},
+ {10, 1275}, {10, 1273}, {10, 1271}, {10, 1269}, {10, 1267}, {10, 1265},
+ {10, 1263}, {10, 1261}, {10, 1259}, {10, 1257}, {10, 1255}, {10, 1253},
+ {10, 1251}, {10, 1249}, {10, 1247}, {10, 1245}, {10, 1243}, {10, 1241},
+ {10, 1239}, {10, 1237}, {10, 1235}, {10, 1233}, {10, 1231}, {10, 1229},
+ {10, 1227}, {10, 1225}, {10, 1223}, {10, 1221}, {10, 1219}, {10, 1217},
+ {10, 1215}, {10, 1213}, {10, 1211}, {10, 1209}, {10, 1207}, {10, 1205},
+ {10, 1203}, {10, 1201}, {10, 1199}, {10, 1197}, {10, 1195}, {10, 1193},
+ {10, 1191}, {10, 1189}, {10, 1187}, {10, 1185}, {10, 1183}, {10, 1181},
+ {10, 1179}, {10, 1177}, {10, 1175}, {10, 1173}, {10, 1171}, {10, 1169},
+ {10, 1167}, {10, 1165}, {10, 1163}, {10, 1161}, {10, 1159}, {10, 1157},
+ {10, 1155}, {10, 1153}, {10, 1151}, {10, 1149}, {10, 1147}, {10, 1145},
+ {10, 1143}, {10, 1141}, {10, 1139}, {10, 1137}, {10, 1135}, {10, 1133},
+ {10, 1131}, {10, 1129}, {10, 1127}, {10, 1125}, {10, 1123}, {10, 1121},
+ {10, 1119}, {10, 1117}, {10, 1115}, {10, 1113}, {10, 1111}, {10, 1109},
+ {10, 1107}, {10, 1105}, {10, 1103}, {10, 1101}, {10, 1099}, {10, 1097},
+ {10, 1095}, {10, 1093}, {10, 1091}, {10, 1089}, {10, 1087}, {10, 1085},
+ {10, 1083}, {10, 1081}, {10, 1079}, {10, 1077}, {10, 1075}, {10, 1073},
+ {10, 1071}, {10, 1069}, {10, 1067}, {10, 1065}, {10, 1063}, {10, 1061},
+ {10, 1059}, {10, 1057}, {10, 1055}, {10, 1053}, {10, 1051}, {10, 1049},
+ {10, 1047}, {10, 1045}, {10, 1043}, {10, 1041}, {10, 1039}, {10, 1037},
+ {10, 1035}, {10, 1033}, {10, 1031}, {10, 1029}, {10, 1027}, {10, 1025},
+ {10, 1023}, {10, 1021}, {10, 1019}, {10, 1017}, {10, 1015}, {10, 1013},
+ {10, 1011}, {10, 1009}, {10, 1007}, {10, 1005}, {10, 1003}, {10, 1001},
+ {10, 999}, {10, 997}, {10, 995}, {10, 993}, {10, 991}, {10, 989},
+ {10, 987}, {10, 985}, {10, 983}, {10, 981}, {10, 979}, {10, 977},
+ {10, 975}, {10, 973}, {10, 971}, {10, 969}, {10, 967}, {10, 965},
+ {10, 963}, {10, 961}, {10, 959}, {10, 957}, {10, 955}, {10, 953},
+ {10, 951}, {10, 949}, {10, 947}, {10, 945}, {10, 943}, {10, 941},
+ {10, 939}, {10, 937}, {10, 935}, {10, 933}, {10, 931}, {10, 929},
+ {10, 927}, {10, 925}, {10, 923}, {10, 921}, {10, 919}, {10, 917},
+ {10, 915}, {10, 913}, {10, 911}, {10, 909}, {10, 907}, {10, 905},
+ {10, 903}, {10, 901}, {10, 899}, {10, 897}, {10, 895}, {10, 893},
+ {10, 891}, {10, 889}, {10, 887}, {10, 885}, {10, 883}, {10, 881},
+ {10, 879}, {10, 877}, {10, 875}, {10, 873}, {10, 871}, {10, 869},
+ {10, 867}, {10, 865}, {10, 863}, {10, 861}, {10, 859}, {10, 857},
+ {10, 855}, {10, 853}, {10, 851}, {10, 849}, {10, 847}, {10, 845},
+ {10, 843}, {10, 841}, {10, 839}, {10, 837}, {10, 835}, {10, 833},
+ {10, 831}, {10, 829}, {10, 827}, {10, 825}, {10, 823}, {10, 821},
+ {10, 819}, {10, 817}, {10, 815}, {10, 813}, {10, 811}, {10, 809},
+ {10, 807}, {10, 805}, {10, 803}, {10, 801}, {10, 799}, {10, 797},
+ {10, 795}, {10, 793}, {10, 791}, {10, 789}, {10, 787}, {10, 785},
+ {10, 783}, {10, 781}, {10, 779}, {10, 777}, {10, 775}, {10, 773},
+ {10, 771}, {10, 769}, {10, 767}, {10, 765}, {10, 763}, {10, 761},
+ {10, 759}, {10, 757}, {10, 755}, {10, 753}, {10, 751}, {10, 749},
+ {10, 747}, {10, 745}, {10, 743}, {10, 741}, {10, 739}, {10, 737},
+ {10, 735}, {10, 733}, {10, 731}, {10, 729}, {10, 727}, {10, 725},
+ {10, 723}, {10, 721}, {10, 719}, {10, 717}, {10, 715}, {10, 713},
+ {10, 711}, {10, 709}, {10, 707}, {10, 705}, {10, 703}, {10, 701},
+ {10, 699}, {10, 697}, {10, 695}, {10, 693}, {10, 691}, {10, 689},
+ {10, 687}, {10, 685}, {10, 683}, {10, 681}, {10, 679}, {10, 677},
+ {10, 675}, {10, 673}, {10, 671}, {10, 669}, {10, 667}, {10, 665},
+ {10, 663}, {10, 661}, {10, 659}, {10, 657}, {10, 655}, {10, 653},
+ {10, 651}, {10, 649}, {10, 647}, {10, 645}, {10, 643}, {10, 641},
+ {10, 639}, {10, 637}, {10, 635}, {10, 633}, {10, 631}, {10, 629},
+ {10, 627}, {10, 625}, {10, 623}, {10, 621}, {10, 619}, {10, 617},
+ {10, 615}, {10, 613}, {10, 611}, {10, 609}, {10, 607}, {10, 605},
+ {10, 603}, {10, 601}, {10, 599}, {10, 597}, {10, 595}, {10, 593},
+ {10, 591}, {10, 589}, {10, 587}, {10, 585}, {10, 583}, {10, 581},
+ {10, 579}, {10, 577}, {10, 575}, {10, 573}, {10, 571}, {10, 569},
+ {10, 567}, {10, 565}, {10, 563}, {10, 561}, {10, 559}, {10, 557},
+ {10, 555}, {10, 553}, {10, 551}, {10, 549}, {10, 547}, {10, 545},
+ {10, 543}, {10, 541}, {10, 539}, {10, 537}, {10, 535}, {10, 533},
+ {10, 531}, {10, 529}, {10, 527}, {10, 525}, {10, 523}, {10, 521},
+ {10, 519}, {10, 517}, {10, 515}, {10, 513}, {10, 511}, {10, 509},
+ {10, 507}, {10, 505}, {10, 503}, {10, 501}, {10, 499}, {10, 497},
+ {10, 495}, {10, 493}, {10, 491}, {10, 489}, {10, 487}, {10, 485},
+ {10, 483}, {10, 481}, {10, 479}, {10, 477}, {10, 475}, {10, 473},
+ {10, 471}, {10, 469}, {10, 467}, {10, 465}, {10, 463}, {10, 461},
+ {10, 459}, {10, 457}, {10, 455}, {10, 453}, {10, 451}, {10, 449},
+ {10, 447}, {10, 445}, {10, 443}, {10, 441}, {10, 439}, {10, 437},
+ {10, 435}, {10, 433}, {10, 431}, {10, 429}, {10, 427}, {10, 425},
+ {10, 423}, {10, 421}, {10, 419}, {10, 417}, {10, 415}, {10, 413},
+ {10, 411}, {10, 409}, {10, 407}, {10, 405}, {10, 403}, {10, 401},
+ {10, 399}, {10, 397}, {10, 395}, {10, 393}, {10, 391}, {10, 389},
+ {10, 387}, {10, 385}, {10, 383}, {10, 381}, {10, 379}, {10, 377},
+ {10, 375}, {10, 373}, {10, 371}, {10, 369}, {10, 367}, {10, 365},
+ {10, 363}, {10, 361}, {10, 359}, {10, 357}, {10, 355}, {10, 353},
+ {10, 351}, {10, 349}, {10, 347}, {10, 345}, {10, 343}, {10, 341},
+ {10, 339}, {10, 337}, {10, 335}, {10, 333}, {10, 331}, {10, 329},
+ {10, 327}, {10, 325}, {10, 323}, {10, 321}, {10, 319}, {10, 317},
+ {10, 315}, {10, 313}, {10, 311}, {10, 309}, {10, 307}, {10, 305},
+ {10, 303}, {10, 301}, {10, 299}, {10, 297}, {10, 295}, {10, 293},
+ {10, 291}, {10, 289}, {10, 287}, {10, 285}, {10, 283}, {10, 281},
+ {10, 279}, {10, 277}, {10, 275}, {10, 273}, {10, 271}, {10, 269},
+ {10, 267}, {10, 265}, {10, 263}, {10, 261}, {10, 259}, {10, 257},
+ {10, 255}, {10, 253}, {10, 251}, {10, 249}, {10, 247}, {10, 245},
+ {10, 243}, {10, 241}, {10, 239}, {10, 237}, {10, 235}, {10, 233},
+ {10, 231}, {10, 229}, {10, 227}, {10, 225}, {10, 223}, {10, 221},
+ {10, 219}, {10, 217}, {10, 215}, {10, 213}, {10, 211}, {10, 209},
+ {10, 207}, {10, 205}, {10, 203}, {10, 201}, {10, 199}, {10, 197},
+ {10, 195}, {10, 193}, {10, 191}, {10, 189}, {10, 187}, {10, 185},
+ {10, 183}, {10, 181}, {10, 179}, {10, 177}, {10, 175}, {10, 173},
+ {10, 171}, {10, 169}, {10, 167}, {10, 165}, {10, 163}, {10, 161},
+ {10, 159}, {10, 157}, {10, 155}, {10, 153}, {10, 151}, {10, 149},
+ {10, 147}, {10, 145}, {10, 143}, {10, 141}, {10, 139}, {10, 137},
+ {10, 135}, {10, 133}, {10, 131}, {10, 129}, {10, 127}, {10, 125},
+ {10, 123}, {10, 121}, {10, 119}, {10, 117}, {10, 115}, {10, 113},
+ {10, 111}, {10, 109}, {10, 107}, {10, 105}, {10, 103}, {10, 101},
+ {10, 99}, {10, 97}, {10, 95}, {10, 93}, {10, 91}, {10, 89},
+ {10, 87}, {10, 85}, {10, 83}, {10, 81}, {10, 79}, {10, 77},
+ {10, 75}, {10, 73}, {10, 71}, {10, 69}, {10, 67}, {10, 65},
+ {10, 63}, {10, 61}, {10, 59}, {10, 57}, {10, 55}, {10, 53},
+ {10, 51}, {10, 49}, {10, 47}, {10, 45}, {10, 43}, {10, 41},
+ {10, 39}, {10, 37}, {10, 35}, {10, 33}, {10, 31}, {10, 29},
+ {10, 27}, {10, 25}, {10, 23}, {10, 21}, {10, 19}, {10, 17},
+ {10, 15}, {10, 13}, {10, 11}, {10, 9}, {10, 7}, {10, 5},
+ {10, 3}, {10, 1}, {9, 63}, {9, 61}, {9, 59}, {9, 57},
+ {9, 55}, {9, 53}, {9, 51}, {9, 49}, {9, 47}, {9, 45},
+ {9, 43}, {9, 41}, {9, 39}, {9, 37}, {9, 35}, {9, 33},
+ {9, 31}, {9, 29}, {9, 27}, {9, 25}, {9, 23}, {9, 21},
+ {9, 19}, {9, 17}, {9, 15}, {9, 13}, {9, 11}, {9, 9},
+ {9, 7}, {9, 5}, {9, 3}, {9, 1}, {8, 31}, {8, 29},
+ {8, 27}, {8, 25}, {8, 23}, {8, 21}, {8, 19}, {8, 17},
+ {8, 15}, {8, 13}, {8, 11}, {8, 9}, {8, 7}, {8, 5},
+ {8, 3}, {8, 1}, {7, 15}, {7, 13}, {7, 11}, {7, 9},
+ {7, 7}, {7, 5}, {7, 3}, {7, 1}, {6, 7}, {6, 5},
+ {6, 3}, {6, 1}, {5, 3}, {5, 1}, {4, 1}, {3, 1},
+ {2, 1}, {1, 1}, {0, 0}, {1, 0}, {2, 0}, {3, 0},
+ {4, 0}, {5, 0}, {5, 2}, {6, 0}, {6, 2}, {6, 4},
+ {6, 6}, {7, 0}, {7, 2}, {7, 4}, {7, 6}, {7, 8},
+ {7, 10}, {7, 12}, {7, 14}, {8, 0}, {8, 2}, {8, 4},
+ {8, 6}, {8, 8}, {8, 10}, {8, 12}, {8, 14}, {8, 16},
+ {8, 18}, {8, 20}, {8, 22}, {8, 24}, {8, 26}, {8, 28},
+ {8, 30}, {9, 0}, {9, 2}, {9, 4}, {9, 6}, {9, 8},
+ {9, 10}, {9, 12}, {9, 14}, {9, 16}, {9, 18}, {9, 20},
+ {9, 22}, {9, 24}, {9, 26}, {9, 28}, {9, 30}, {9, 32},
+ {9, 34}, {9, 36}, {9, 38}, {9, 40}, {9, 42}, {9, 44},
+ {9, 46}, {9, 48}, {9, 50}, {9, 52}, {9, 54}, {9, 56},
+ {9, 58}, {9, 60}, {9, 62}, {10, 0}, {10, 2}, {10, 4},
+ {10, 6}, {10, 8}, {10, 10}, {10, 12}, {10, 14}, {10, 16},
+ {10, 18}, {10, 20}, {10, 22}, {10, 24}, {10, 26}, {10, 28},
+ {10, 30}, {10, 32}, {10, 34}, {10, 36}, {10, 38}, {10, 40},
+ {10, 42}, {10, 44}, {10, 46}, {10, 48}, {10, 50}, {10, 52},
+ {10, 54}, {10, 56}, {10, 58}, {10, 60}, {10, 62}, {10, 64},
+ {10, 66}, {10, 68}, {10, 70}, {10, 72}, {10, 74}, {10, 76},
+ {10, 78}, {10, 80}, {10, 82}, {10, 84}, {10, 86}, {10, 88},
+ {10, 90}, {10, 92}, {10, 94}, {10, 96}, {10, 98}, {10, 100},
+ {10, 102}, {10, 104}, {10, 106}, {10, 108}, {10, 110}, {10, 112},
+ {10, 114}, {10, 116}, {10, 118}, {10, 120}, {10, 122}, {10, 124},
+ {10, 126}, {10, 128}, {10, 130}, {10, 132}, {10, 134}, {10, 136},
+ {10, 138}, {10, 140}, {10, 142}, {10, 144}, {10, 146}, {10, 148},
+ {10, 150}, {10, 152}, {10, 154}, {10, 156}, {10, 158}, {10, 160},
+ {10, 162}, {10, 164}, {10, 166}, {10, 168}, {10, 170}, {10, 172},
+ {10, 174}, {10, 176}, {10, 178}, {10, 180}, {10, 182}, {10, 184},
+ {10, 186}, {10, 188}, {10, 190}, {10, 192}, {10, 194}, {10, 196},
+ {10, 198}, {10, 200}, {10, 202}, {10, 204}, {10, 206}, {10, 208},
+ {10, 210}, {10, 212}, {10, 214}, {10, 216}, {10, 218}, {10, 220},
+ {10, 222}, {10, 224}, {10, 226}, {10, 228}, {10, 230}, {10, 232},
+ {10, 234}, {10, 236}, {10, 238}, {10, 240}, {10, 242}, {10, 244},
+ {10, 246}, {10, 248}, {10, 250}, {10, 252}, {10, 254}, {10, 256},
+ {10, 258}, {10, 260}, {10, 262}, {10, 264}, {10, 266}, {10, 268},
+ {10, 270}, {10, 272}, {10, 274}, {10, 276}, {10, 278}, {10, 280},
+ {10, 282}, {10, 284}, {10, 286}, {10, 288}, {10, 290}, {10, 292},
+ {10, 294}, {10, 296}, {10, 298}, {10, 300}, {10, 302}, {10, 304},
+ {10, 306}, {10, 308}, {10, 310}, {10, 312}, {10, 314}, {10, 316},
+ {10, 318}, {10, 320}, {10, 322}, {10, 324}, {10, 326}, {10, 328},
+ {10, 330}, {10, 332}, {10, 334}, {10, 336}, {10, 338}, {10, 340},
+ {10, 342}, {10, 344}, {10, 346}, {10, 348}, {10, 350}, {10, 352},
+ {10, 354}, {10, 356}, {10, 358}, {10, 360}, {10, 362}, {10, 364},
+ {10, 366}, {10, 368}, {10, 370}, {10, 372}, {10, 374}, {10, 376},
+ {10, 378}, {10, 380}, {10, 382}, {10, 384}, {10, 386}, {10, 388},
+ {10, 390}, {10, 392}, {10, 394}, {10, 396}, {10, 398}, {10, 400},
+ {10, 402}, {10, 404}, {10, 406}, {10, 408}, {10, 410}, {10, 412},
+ {10, 414}, {10, 416}, {10, 418}, {10, 420}, {10, 422}, {10, 424},
+ {10, 426}, {10, 428}, {10, 430}, {10, 432}, {10, 434}, {10, 436},
+ {10, 438}, {10, 440}, {10, 442}, {10, 444}, {10, 446}, {10, 448},
+ {10, 450}, {10, 452}, {10, 454}, {10, 456}, {10, 458}, {10, 460},
+ {10, 462}, {10, 464}, {10, 466}, {10, 468}, {10, 470}, {10, 472},
+ {10, 474}, {10, 476}, {10, 478}, {10, 480}, {10, 482}, {10, 484},
+ {10, 486}, {10, 488}, {10, 490}, {10, 492}, {10, 494}, {10, 496},
+ {10, 498}, {10, 500}, {10, 502}, {10, 504}, {10, 506}, {10, 508},
+ {10, 510}, {10, 512}, {10, 514}, {10, 516}, {10, 518}, {10, 520},
+ {10, 522}, {10, 524}, {10, 526}, {10, 528}, {10, 530}, {10, 532},
+ {10, 534}, {10, 536}, {10, 538}, {10, 540}, {10, 542}, {10, 544},
+ {10, 546}, {10, 548}, {10, 550}, {10, 552}, {10, 554}, {10, 556},
+ {10, 558}, {10, 560}, {10, 562}, {10, 564}, {10, 566}, {10, 568},
+ {10, 570}, {10, 572}, {10, 574}, {10, 576}, {10, 578}, {10, 580},
+ {10, 582}, {10, 584}, {10, 586}, {10, 588}, {10, 590}, {10, 592},
+ {10, 594}, {10, 596}, {10, 598}, {10, 600}, {10, 602}, {10, 604},
+ {10, 606}, {10, 608}, {10, 610}, {10, 612}, {10, 614}, {10, 616},
+ {10, 618}, {10, 620}, {10, 622}, {10, 624}, {10, 626}, {10, 628},
+ {10, 630}, {10, 632}, {10, 634}, {10, 636}, {10, 638}, {10, 640},
+ {10, 642}, {10, 644}, {10, 646}, {10, 648}, {10, 650}, {10, 652},
+ {10, 654}, {10, 656}, {10, 658}, {10, 660}, {10, 662}, {10, 664},
+ {10, 666}, {10, 668}, {10, 670}, {10, 672}, {10, 674}, {10, 676},
+ {10, 678}, {10, 680}, {10, 682}, {10, 684}, {10, 686}, {10, 688},
+ {10, 690}, {10, 692}, {10, 694}, {10, 696}, {10, 698}, {10, 700},
+ {10, 702}, {10, 704}, {10, 706}, {10, 708}, {10, 710}, {10, 712},
+ {10, 714}, {10, 716}, {10, 718}, {10, 720}, {10, 722}, {10, 724},
+ {10, 726}, {10, 728}, {10, 730}, {10, 732}, {10, 734}, {10, 736},
+ {10, 738}, {10, 740}, {10, 742}, {10, 744}, {10, 746}, {10, 748},
+ {10, 750}, {10, 752}, {10, 754}, {10, 756}, {10, 758}, {10, 760},
+ {10, 762}, {10, 764}, {10, 766}, {10, 768}, {10, 770}, {10, 772},
+ {10, 774}, {10, 776}, {10, 778}, {10, 780}, {10, 782}, {10, 784},
+ {10, 786}, {10, 788}, {10, 790}, {10, 792}, {10, 794}, {10, 796},
+ {10, 798}, {10, 800}, {10, 802}, {10, 804}, {10, 806}, {10, 808},
+ {10, 810}, {10, 812}, {10, 814}, {10, 816}, {10, 818}, {10, 820},
+ {10, 822}, {10, 824}, {10, 826}, {10, 828}, {10, 830}, {10, 832},
+ {10, 834}, {10, 836}, {10, 838}, {10, 840}, {10, 842}, {10, 844},
+ {10, 846}, {10, 848}, {10, 850}, {10, 852}, {10, 854}, {10, 856},
+ {10, 858}, {10, 860}, {10, 862}, {10, 864}, {10, 866}, {10, 868},
+ {10, 870}, {10, 872}, {10, 874}, {10, 876}, {10, 878}, {10, 880},
+ {10, 882}, {10, 884}, {10, 886}, {10, 888}, {10, 890}, {10, 892},
+ {10, 894}, {10, 896}, {10, 898}, {10, 900}, {10, 902}, {10, 904},
+ {10, 906}, {10, 908}, {10, 910}, {10, 912}, {10, 914}, {10, 916},
+ {10, 918}, {10, 920}, {10, 922}, {10, 924}, {10, 926}, {10, 928},
+ {10, 930}, {10, 932}, {10, 934}, {10, 936}, {10, 938}, {10, 940},
+ {10, 942}, {10, 944}, {10, 946}, {10, 948}, {10, 950}, {10, 952},
+ {10, 954}, {10, 956}, {10, 958}, {10, 960}, {10, 962}, {10, 964},
+ {10, 966}, {10, 968}, {10, 970}, {10, 972}, {10, 974}, {10, 976},
+ {10, 978}, {10, 980}, {10, 982}, {10, 984}, {10, 986}, {10, 988},
+ {10, 990}, {10, 992}, {10, 994}, {10, 996}, {10, 998}, {10, 1000},
+ {10, 1002}, {10, 1004}, {10, 1006}, {10, 1008}, {10, 1010}, {10, 1012},
+ {10, 1014}, {10, 1016}, {10, 1018}, {10, 1020}, {10, 1022}, {10, 1024},
+ {10, 1026}, {10, 1028}, {10, 1030}, {10, 1032}, {10, 1034}, {10, 1036},
+ {10, 1038}, {10, 1040}, {10, 1042}, {10, 1044}, {10, 1046}, {10, 1048},
+ {10, 1050}, {10, 1052}, {10, 1054}, {10, 1056}, {10, 1058}, {10, 1060},
+ {10, 1062}, {10, 1064}, {10, 1066}, {10, 1068}, {10, 1070}, {10, 1072},
+ {10, 1074}, {10, 1076}, {10, 1078}, {10, 1080}, {10, 1082}, {10, 1084},
+ {10, 1086}, {10, 1088}, {10, 1090}, {10, 1092}, {10, 1094}, {10, 1096},
+ {10, 1098}, {10, 1100}, {10, 1102}, {10, 1104}, {10, 1106}, {10, 1108},
+ {10, 1110}, {10, 1112}, {10, 1114}, {10, 1116}, {10, 1118}, {10, 1120},
+ {10, 1122}, {10, 1124}, {10, 1126}, {10, 1128}, {10, 1130}, {10, 1132},
+ {10, 1134}, {10, 1136}, {10, 1138}, {10, 1140}, {10, 1142}, {10, 1144},
+ {10, 1146}, {10, 1148}, {10, 1150}, {10, 1152}, {10, 1154}, {10, 1156},
+ {10, 1158}, {10, 1160}, {10, 1162}, {10, 1164}, {10, 1166}, {10, 1168},
+ {10, 1170}, {10, 1172}, {10, 1174}, {10, 1176}, {10, 1178}, {10, 1180},
+ {10, 1182}, {10, 1184}, {10, 1186}, {10, 1188}, {10, 1190}, {10, 1192},
+ {10, 1194}, {10, 1196}, {10, 1198}, {10, 1200}, {10, 1202}, {10, 1204},
+ {10, 1206}, {10, 1208}, {10, 1210}, {10, 1212}, {10, 1214}, {10, 1216},
+ {10, 1218}, {10, 1220}, {10, 1222}, {10, 1224}, {10, 1226}, {10, 1228},
+ {10, 1230}, {10, 1232}, {10, 1234}, {10, 1236}, {10, 1238}, {10, 1240},
+ {10, 1242}, {10, 1244}, {10, 1246}, {10, 1248}, {10, 1250}, {10, 1252},
+ {10, 1254}, {10, 1256}, {10, 1258}, {10, 1260}, {10, 1262}, {10, 1264},
+ {10, 1266}, {10, 1268}, {10, 1270}, {10, 1272}, {10, 1274}, {10, 1276},
+ {10, 1278}, {10, 1280}, {10, 1282}, {10, 1284}, {10, 1286}, {10, 1288},
+ {10, 1290}, {10, 1292}, {10, 1294}, {10, 1296}, {10, 1298}, {10, 1300},
+ {10, 1302}, {10, 1304}, {10, 1306}, {10, 1308}, {10, 1310}, {10, 1312},
+ {10, 1314}, {10, 1316}, {10, 1318}, {10, 1320}, {10, 1322}, {10, 1324},
+ {10, 1326}, {10, 1328}, {10, 1330}, {10, 1332}, {10, 1334}, {10, 1336},
+ {10, 1338}, {10, 1340}, {10, 1342}, {10, 1344}, {10, 1346}, {10, 1348},
+ {10, 1350}, {10, 1352}, {10, 1354}, {10, 1356}, {10, 1358}, {10, 1360},
+ {10, 1362}, {10, 1364}, {10, 1366}, {10, 1368}, {10, 1370}, {10, 1372},
+ {10, 1374}, {10, 1376}, {10, 1378}, {10, 1380}, {10, 1382}, {10, 1384},
+ {10, 1386}, {10, 1388}, {10, 1390}, {10, 1392}, {10, 1394}, {10, 1396},
+ {10, 1398}, {10, 1400}, {10, 1402}, {10, 1404}, {10, 1406}, {10, 1408},
+ {10, 1410}, {10, 1412}, {10, 1414}, {10, 1416}, {10, 1418}, {10, 1420},
+ {10, 1422}, {10, 1424}, {10, 1426}, {10, 1428}, {10, 1430}, {10, 1432},
+ {10, 1434}, {10, 1436}, {10, 1438}, {10, 1440}, {10, 1442}, {10, 1444},
+ {10, 1446}, {10, 1448}, {10, 1450}, {10, 1452}, {10, 1454}, {10, 1456},
+ {10, 1458}, {10, 1460}, {10, 1462}, {10, 1464}, {10, 1466}, {10, 1468},
+ {10, 1470}, {10, 1472}, {10, 1474}, {10, 1476}, {10, 1478}, {10, 1480},
+ {10, 1482}, {10, 1484}, {10, 1486}, {10, 1488}, {10, 1490}, {10, 1492},
+ {10, 1494}, {10, 1496}, {10, 1498}, {10, 1500}, {10, 1502}, {10, 1504},
+ {10, 1506}, {10, 1508}, {10, 1510}, {10, 1512}, {10, 1514}, {10, 1516},
+ {10, 1518}, {10, 1520}, {10, 1522}, {10, 1524}, {10, 1526}, {10, 1528},
+ {10, 1530}, {10, 1532}, {10, 1534}, {10, 1536}, {10, 1538}, {10, 1540},
+ {10, 1542}, {10, 1544}, {10, 1546}, {10, 1548}, {10, 1550}, {10, 1552},
+ {10, 1554}, {10, 1556}, {10, 1558}, {10, 1560}, {10, 1562}, {10, 1564},
+ {10, 1566}, {10, 1568}, {10, 1570}, {10, 1572}, {10, 1574}, {10, 1576},
+ {10, 1578}, {10, 1580}, {10, 1582}, {10, 1584}, {10, 1586}, {10, 1588},
+ {10, 1590}, {10, 1592}, {10, 1594}, {10, 1596}, {10, 1598}, {10, 1600},
+ {10, 1602}, {10, 1604}, {10, 1606}, {10, 1608}, {10, 1610}, {10, 1612},
+ {10, 1614}, {10, 1616}, {10, 1618}, {10, 1620}, {10, 1622}, {10, 1624},
+ {10, 1626}, {10, 1628}, {10, 1630}, {10, 1632}, {10, 1634}, {10, 1636},
+ {10, 1638}, {10, 1640}, {10, 1642}, {10, 1644}, {10, 1646}, {10, 1648},
+ {10, 1650}, {10, 1652}, {10, 1654}, {10, 1656}, {10, 1658}, {10, 1660},
+ {10, 1662}, {10, 1664}, {10, 1666}, {10, 1668}, {10, 1670}, {10, 1672},
+ {10, 1674}, {10, 1676}, {10, 1678}, {10, 1680}, {10, 1682}, {10, 1684},
+ {10, 1686}, {10, 1688}, {10, 1690}, {10, 1692}, {10, 1694}, {10, 1696},
+ {10, 1698}, {10, 1700}, {10, 1702}, {10, 1704}, {10, 1706}, {10, 1708},
+ {10, 1710}, {10, 1712}, {10, 1714}, {10, 1716}, {10, 1718}, {10, 1720},
+ {10, 1722}, {10, 1724}, {10, 1726}, {10, 1728}, {10, 1730}, {10, 1732},
+ {10, 1734}, {10, 1736}, {10, 1738}, {10, 1740}, {10, 1742}, {10, 1744},
+ {10, 1746}, {10, 1748}, {10, 1750}, {10, 1752}, {10, 1754}, {10, 1756},
+ {10, 1758}, {10, 1760}, {10, 1762}, {10, 1764}, {10, 1766}, {10, 1768},
+ {10, 1770}, {10, 1772}, {10, 1774}, {10, 1776}, {10, 1778}, {10, 1780},
+ {10, 1782}, {10, 1784}, {10, 1786}, {10, 1788}, {10, 1790}, {10, 1792},
+ {10, 1794}, {10, 1796}, {10, 1798}, {10, 1800}, {10, 1802}, {10, 1804},
+ {10, 1806}, {10, 1808}, {10, 1810}, {10, 1812}, {10, 1814}, {10, 1816},
+ {10, 1818}, {10, 1820}, {10, 1822}, {10, 1824}, {10, 1826}, {10, 1828},
+ {10, 1830}, {10, 1832}, {10, 1834}, {10, 1836}, {10, 1838}, {10, 1840},
+ {10, 1842}, {10, 1844}, {10, 1846}, {10, 1848}, {10, 1850}, {10, 1852},
+ {10, 1854}, {10, 1856}, {10, 1858}, {10, 1860}, {10, 1862}, {10, 1864},
+ {10, 1866}, {10, 1868}, {10, 1870}, {10, 1872}, {10, 1874}, {10, 1876},
+ {10, 1878}, {10, 1880}, {10, 1882}, {10, 1884}, {10, 1886}, {10, 1888},
+ {10, 1890}, {10, 1892}, {10, 1894}, {10, 1896}, {10, 1898}, {10, 1900},
+ {10, 1902}, {10, 1904}, {10, 1906}, {10, 1908}, {10, 1910}, {10, 1912},
+ {10, 1914}, {10, 1916}, {10, 1918}, {10, 1920}, {10, 1922}, {10, 1924},
+ {10, 1926}, {10, 1928}, {10, 1930}, {10, 1932}, {10, 1934}, {10, 1936},
+ {10, 1938}, {10, 1940}, {10, 1942}, {10, 1944}, {10, 1946}, {10, 1948},
+ {10, 1950}, {10, 1952}, {10, 1954}, {10, 1956}, {10, 1958}, {10, 1960},
+ {10, 1962}, {10, 1964}, {10, 1966}, {10, 1968}, {10, 1970}, {10, 1972},
+ {10, 1974}, {10, 1976}, {10, 1978}, {10, 1980}, {10, 1982}, {10, 1984},
+ {10, 1986}, {10, 1988}, {10, 1990}, {10, 1992}, {10, 1994}, {10, 1996},
+ {10, 1998}, {10, 2000}, {10, 2002}, {10, 2004}, {10, 2006}, {10, 2008},
+ {10, 2010}, {10, 2012}, {10, 2014}, {10, 2016}, {10, 2018}, {10, 2020},
+ {10, 2022}, {10, 2024}, {10, 2026}, {10, 2028}, {10, 2030}, {10, 2032},
+ {10, 2034}, {10, 2036}, {10, 2038}, {10, 2040}, {10, 2042}, {10, 2044},
+ {10, 2046}, {10, 2048}, {10, 2050}, {10, 2052}, {10, 2054}, {10, 2056},
+ {10, 2058}, {10, 2060}, {10, 2062}, {10, 2064}, {10, 2066}, {10, 2068},
+ {10, 2070}, {10, 2072}, {10, 2074}, {10, 2076}, {10, 2078}, {10, 2080},
+ {10, 2082}, {10, 2084}, {10, 2086}, {10, 2088}, {10, 2090}, {10, 2092},
+ {10, 2094}, {10, 2096}, {10, 2098}, {10, 2100}, {10, 2102}, {10, 2104},
+ {10, 2106}, {10, 2108}, {10, 2110}, {10, 2112}, {10, 2114}, {10, 2116},
+ {10, 2118}, {10, 2120}, {10, 2122}, {10, 2124}, {10, 2126}, {10, 2128},
+ {10, 2130}, {10, 2132}, {10, 2134}, {10, 2136}, {10, 2138}, {10, 2140},
+ {10, 2142}, {10, 2144}, {10, 2146}, {10, 2148}, {10, 2150}, {10, 2152},
+ {10, 2154}, {10, 2156}, {10, 2158}, {10, 2160}, {10, 2162}, {10, 2164},
+ {10, 2166}, {10, 2168}, {10, 2170}, {10, 2172}, {10, 2174}, {10, 2176},
+ {10, 2178}, {10, 2180}, {10, 2182}, {10, 2184}, {10, 2186}, {10, 2188},
+ {10, 2190}, {10, 2192}, {10, 2194}, {10, 2196}, {10, 2198}, {10, 2200},
+ {10, 2202}, {10, 2204}, {10, 2206}, {10, 2208}, {10, 2210}, {10, 2212},
+ {10, 2214}, {10, 2216}, {10, 2218}, {10, 2220}, {10, 2222}, {10, 2224},
+ {10, 2226}, {10, 2228}, {10, 2230}, {10, 2232}, {10, 2234}, {10, 2236},
+ {10, 2238}, {10, 2240}, {10, 2242}, {10, 2244}, {10, 2246}, {10, 2248},
+ {10, 2250}, {10, 2252}, {10, 2254}, {10, 2256}, {10, 2258}, {10, 2260},
+ {10, 2262}, {10, 2264}, {10, 2266}, {10, 2268}, {10, 2270}, {10, 2272},
+ {10, 2274}, {10, 2276}, {10, 2278}, {10, 2280}, {10, 2282}, {10, 2284},
+ {10, 2286}, {10, 2288}, {10, 2290}, {10, 2292}, {10, 2294}, {10, 2296},
+ {10, 2298}, {10, 2300}, {10, 2302}, {10, 2304}, {10, 2306}, {10, 2308},
+ {10, 2310}, {10, 2312}, {10, 2314}, {10, 2316}, {10, 2318}, {10, 2320},
+ {10, 2322}, {10, 2324}, {10, 2326}, {10, 2328}, {10, 2330}, {10, 2332},
+ {10, 2334}, {10, 2336}, {10, 2338}, {10, 2340}, {10, 2342}, {10, 2344},
+ {10, 2346}, {10, 2348}, {10, 2350}, {10, 2352}, {10, 2354}, {10, 2356},
+ {10, 2358}, {10, 2360}, {10, 2362}, {10, 2364}, {10, 2366}, {10, 2368},
+ {10, 2370}, {10, 2372}, {10, 2374}, {10, 2376}, {10, 2378}, {10, 2380},
+ {10, 2382}, {10, 2384}, {10, 2386}, {10, 2388}, {10, 2390}, {10, 2392},
+ {10, 2394}, {10, 2396}, {10, 2398}, {10, 2400}, {10, 2402}, {10, 2404},
+ {10, 2406}, {10, 2408}, {10, 2410}, {10, 2412}, {10, 2414}, {10, 2416},
+ {10, 2418}, {10, 2420}, {10, 2422}, {10, 2424}, {10, 2426}, {10, 2428},
+ {10, 2430}, {10, 2432}, {10, 2434}, {10, 2436}, {10, 2438}, {10, 2440},
+ {10, 2442}, {10, 2444}, {10, 2446}, {10, 2448}, {10, 2450}, {10, 2452},
+ {10, 2454}, {10, 2456}, {10, 2458}, {10, 2460}, {10, 2462}, {10, 2464},
+ {10, 2466}, {10, 2468}, {10, 2470}, {10, 2472}, {10, 2474}, {10, 2476},
+ {10, 2478}, {10, 2480}, {10, 2482}, {10, 2484}, {10, 2486}, {10, 2488},
+ {10, 2490}, {10, 2492}, {10, 2494}, {10, 2496}, {10, 2498}, {10, 2500},
+ {10, 2502}, {10, 2504}, {10, 2506}, {10, 2508}, {10, 2510}, {10, 2512},
+ {10, 2514}, {10, 2516}, {10, 2518}, {10, 2520}, {10, 2522}, {10, 2524},
+ {10, 2526}, {10, 2528}, {10, 2530}, {10, 2532}, {10, 2534}, {10, 2536},
+ {10, 2538}, {10, 2540}, {10, 2542}, {10, 2544}, {10, 2546}, {10, 2548},
+ {10, 2550}, {10, 2552}, {10, 2554}, {10, 2556}, {10, 2558}, {10, 2560},
+ {10, 2562}, {10, 2564}, {10, 2566}, {10, 2568}, {10, 2570}, {10, 2572},
+ {10, 2574}, {10, 2576}, {10, 2578}, {10, 2580}, {10, 2582}, {10, 2584},
+ {10, 2586}, {10, 2588}, {10, 2590}, {10, 2592}, {10, 2594}, {10, 2596},
+ {10, 2598}, {10, 2600}, {10, 2602}, {10, 2604}, {10, 2606}, {10, 2608},
+ {10, 2610}, {10, 2612}, {10, 2614}, {10, 2616}, {10, 2618}, {10, 2620},
+ {10, 2622}, {10, 2624}, {10, 2626}, {10, 2628}, {10, 2630}, {10, 2632},
+ {10, 2634}, {10, 2636}, {10, 2638}, {10, 2640}, {10, 2642}, {10, 2644},
+ {10, 2646}, {10, 2648}, {10, 2650}, {10, 2652}, {10, 2654}, {10, 2656},
+ {10, 2658}, {10, 2660}, {10, 2662}, {10, 2664}, {10, 2666}, {10, 2668},
+ {10, 2670}, {10, 2672}, {10, 2674}, {10, 2676}, {10, 2678}, {10, 2680},
+ {10, 2682}, {10, 2684}, {10, 2686}, {10, 2688}, {10, 2690}, {10, 2692},
+ {10, 2694}, {10, 2696}, {10, 2698}, {10, 2700}, {10, 2702}, {10, 2704},
+ {10, 2706}, {10, 2708}, {10, 2710}, {10, 2712}, {10, 2714}, {10, 2716},
+ {10, 2718}, {10, 2720}, {10, 2722}, {10, 2724}, {10, 2726}, {10, 2728},
+ {10, 2730}, {10, 2732}, {10, 2734}, {10, 2736}, {10, 2738}, {10, 2740},
+ {10, 2742}, {10, 2744}, {10, 2746}, {10, 2748}, {10, 2750}, {10, 2752},
+ {10, 2754}, {10, 2756}, {10, 2758}, {10, 2760}, {10, 2762}, {10, 2764},
+ {10, 2766}, {10, 2768}, {10, 2770}, {10, 2772}, {10, 2774}, {10, 2776},
+ {10, 2778}, {10, 2780}, {10, 2782}, {10, 2784}, {10, 2786}, {10, 2788},
+ {10, 2790}, {10, 2792}, {10, 2794}, {10, 2796}, {10, 2798}, {10, 2800},
+ {10, 2802}, {10, 2804}, {10, 2806}, {10, 2808}, {10, 2810}, {10, 2812},
+ {10, 2814}, {10, 2816}, {10, 2818}, {10, 2820}, {10, 2822}, {10, 2824},
+ {10, 2826}, {10, 2828}, {10, 2830}, {10, 2832}, {10, 2834}, {10, 2836},
+ {10, 2838}, {10, 2840}, {10, 2842}, {10, 2844}, {10, 2846}, {10, 2848},
+ {10, 2850}, {10, 2852}, {10, 2854}, {10, 2856}, {10, 2858}, {10, 2860},
+ {10, 2862}, {10, 2864}, {10, 2866}, {10, 2868}, {10, 2870}, {10, 2872},
+ {10, 2874}, {10, 2876}, {10, 2878}, {10, 2880}, {10, 2882}, {10, 2884},
+ {10, 2886}, {10, 2888}, {10, 2890}, {10, 2892}, {10, 2894}, {10, 2896},
+ {10, 2898}, {10, 2900}, {10, 2902}, {10, 2904}, {10, 2906}, {10, 2908},
+ {10, 2910}, {10, 2912}, {10, 2914}, {10, 2916}, {10, 2918}, {10, 2920},
+ {10, 2922}, {10, 2924}, {10, 2926}, {10, 2928}, {10, 2930}, {10, 2932},
+ {10, 2934}, {10, 2936}, {10, 2938}, {10, 2940}, {10, 2942}, {10, 2944},
+ {10, 2946}, {10, 2948}, {10, 2950}, {10, 2952}, {10, 2954}, {10, 2956},
+ {10, 2958}, {10, 2960}, {10, 2962}, {10, 2964}, {10, 2966}, {10, 2968},
+ {10, 2970}, {10, 2972}, {10, 2974}, {10, 2976}, {10, 2978}, {10, 2980},
+ {10, 2982}, {10, 2984}, {10, 2986}, {10, 2988}, {10, 2990}, {10, 2992},
+ {10, 2994}, {10, 2996}, {10, 2998}, {10, 3000}, {10, 3002}, {10, 3004},
+ {10, 3006}, {10, 3008}, {10, 3010}, {10, 3012}, {10, 3014}, {10, 3016},
+ {10, 3018}, {10, 3020}, {10, 3022}, {10, 3024}, {10, 3026}, {10, 3028},
+ {10, 3030}, {10, 3032}, {10, 3034}, {10, 3036}, {10, 3038}, {10, 3040},
+ {10, 3042}, {10, 3044}, {10, 3046}, {10, 3048}, {10, 3050}, {10, 3052},
+ {10, 3054}, {10, 3056}, {10, 3058}, {10, 3060}, {10, 3062}, {10, 3064},
+ {10, 3066}, {10, 3068}, {10, 3070}, {10, 3072}, {10, 3074}, {10, 3076},
+ {10, 3078}, {10, 3080}, {10, 3082}, {10, 3084}, {10, 3086}, {10, 3088},
+ {10, 3090}, {10, 3092}, {10, 3094}, {10, 3096}, {10, 3098}, {10, 3100},
+ {10, 3102}, {10, 3104}, {10, 3106}, {10, 3108}, {10, 3110}, {10, 3112},
+ {10, 3114}, {10, 3116}, {10, 3118}, {10, 3120}, {10, 3122}, {10, 3124},
+ {10, 3126}, {10, 3128}, {10, 3130}, {10, 3132}, {10, 3134}, {10, 3136},
+ {10, 3138}, {10, 3140}, {10, 3142}, {10, 3144}, {10, 3146}, {10, 3148},
+ {10, 3150}, {10, 3152}, {10, 3154}, {10, 3156}, {10, 3158}, {10, 3160},
+ {10, 3162}, {10, 3164}, {10, 3166}, {10, 3168}, {10, 3170}, {10, 3172},
+ {10, 3174}, {10, 3176}, {10, 3178}, {10, 3180}, {10, 3182}, {10, 3184},
+ {10, 3186}, {10, 3188}, {10, 3190}, {10, 3192}, {10, 3194}, {10, 3196},
+ {10, 3198}, {10, 3200}, {10, 3202}, {10, 3204}, {10, 3206}, {10, 3208},
+ {10, 3210}, {10, 3212}, {10, 3214}, {10, 3216}, {10, 3218}, {10, 3220},
+ {10, 3222}, {10, 3224}, {10, 3226}, {10, 3228}, {10, 3230}, {10, 3232},
+ {10, 3234}, {10, 3236}, {10, 3238}, {10, 3240}, {10, 3242}, {10, 3244},
+ {10, 3246}, {10, 3248}, {10, 3250}, {10, 3252}, {10, 3254}, {10, 3256},
+ {10, 3258}, {10, 3260}, {10, 3262}, {10, 3264}, {10, 3266}, {10, 3268},
+ {10, 3270}, {10, 3272}, {10, 3274}, {10, 3276}, {10, 3278}, {10, 3280},
+ {10, 3282}, {10, 3284}, {10, 3286}, {10, 3288}, {10, 3290}, {10, 3292},
+ {10, 3294}, {10, 3296}, {10, 3298}, {10, 3300}, {10, 3302}, {10, 3304},
+ {10, 3306}, {10, 3308}, {10, 3310}, {10, 3312}, {10, 3314}, {10, 3316},
+ {10, 3318}, {10, 3320}, {10, 3322}, {10, 3324}, {10, 3326}, {10, 3328},
+ {10, 3330}, {10, 3332}, {10, 3334}, {10, 3336}, {10, 3338}, {10, 3340},
+ {10, 3342}, {10, 3344}, {10, 3346}, {10, 3348}, {10, 3350}, {10, 3352},
+ {10, 3354}, {10, 3356}, {10, 3358}, {10, 3360}, {10, 3362}, {10, 3364},
+ {10, 3366}, {10, 3368}, {10, 3370}, {10, 3372}, {10, 3374}, {10, 3376},
+ {10, 3378}, {10, 3380}, {10, 3382}, {10, 3384}, {10, 3386}, {10, 3388},
+ {10, 3390}, {10, 3392}, {10, 3394}, {10, 3396}, {10, 3398}, {10, 3400},
+ {10, 3402}, {10, 3404}, {10, 3406}, {10, 3408}, {10, 3410}, {10, 3412},
+ {10, 3414}, {10, 3416}, {10, 3418}, {10, 3420}, {10, 3422}, {10, 3424},
+ {10, 3426}, {10, 3428}, {10, 3430}, {10, 3432}, {10, 3434}, {10, 3436},
+ {10, 3438}, {10, 3440}, {10, 3442}, {10, 3444}, {10, 3446}, {10, 3448},
+ {10, 3450}, {10, 3452}, {10, 3454}, {10, 3456}, {10, 3458}, {10, 3460},
+ {10, 3462}, {10, 3464}, {10, 3466}, {10, 3468}, {10, 3470}, {10, 3472},
+ {10, 3474}, {10, 3476}, {10, 3478}, {10, 3480}, {10, 3482}, {10, 3484},
+ {10, 3486}, {10, 3488}, {10, 3490}, {10, 3492}, {10, 3494}, {10, 3496},
+ {10, 3498}, {10, 3500}, {10, 3502}, {10, 3504}, {10, 3506}, {10, 3508},
+ {10, 3510}, {10, 3512}, {10, 3514}, {10, 3516}, {10, 3518}, {10, 3520},
+ {10, 3522}, {10, 3524}, {10, 3526}, {10, 3528}, {10, 3530}, {10, 3532},
+ {10, 3534}, {10, 3536}, {10, 3538}, {10, 3540}, {10, 3542}, {10, 3544},
+ {10, 3546}, {10, 3548}, {10, 3550}, {10, 3552}, {10, 3554}, {10, 3556},
+ {10, 3558}, {10, 3560}, {10, 3562}, {10, 3564}, {10, 3566}, {10, 3568},
+ {10, 3570}, {10, 3572}, {10, 3574}, {10, 3576}, {10, 3578}, {10, 3580},
+ {10, 3582}, {10, 3584}, {10, 3586}, {10, 3588}, {10, 3590}, {10, 3592},
+ {10, 3594}, {10, 3596}, {10, 3598}, {10, 3600}, {10, 3602}, {10, 3604},
+ {10, 3606}, {10, 3608}, {10, 3610}, {10, 3612}, {10, 3614}, {10, 3616},
+ {10, 3618}, {10, 3620}, {10, 3622}, {10, 3624}, {10, 3626}, {10, 3628},
+ {10, 3630}, {10, 3632}, {10, 3634}, {10, 3636}, {10, 3638}, {10, 3640},
+ {10, 3642}, {10, 3644}, {10, 3646}, {10, 3648}, {10, 3650}, {10, 3652},
+ {10, 3654}, {10, 3656}, {10, 3658}, {10, 3660}, {10, 3662}, {10, 3664},
+ {10, 3666}, {10, 3668}, {10, 3670}, {10, 3672}, {10, 3674}, {10, 3676},
+ {10, 3678}, {10, 3680}, {10, 3682}, {10, 3684}, {10, 3686}, {10, 3688},
+ {10, 3690}, {10, 3692}, {10, 3694}, {10, 3696}, {10, 3698}, {10, 3700},
+ {10, 3702}, {10, 3704}, {10, 3706}, {10, 3708}, {10, 3710}, {10, 3712},
+ {10, 3714}, {10, 3716}, {10, 3718}, {10, 3720}, {10, 3722}, {10, 3724},
+ {10, 3726}, {10, 3728}, {10, 3730}, {10, 3732}, {10, 3734}, {10, 3736},
+ {10, 3738}, {10, 3740}, {10, 3742}, {10, 3744}, {10, 3746}, {10, 3748},
+ {10, 3750}, {10, 3752}, {10, 3754}, {10, 3756}, {10, 3758}, {10, 3760},
+ {10, 3762}, {10, 3764}, {10, 3766}, {10, 3768}, {10, 3770}, {10, 3772},
+ {10, 3774}, {10, 3776}, {10, 3778}, {10, 3780}, {10, 3782}, {10, 3784},
+ {10, 3786}, {10, 3788}, {10, 3790}, {10, 3792}, {10, 3794}, {10, 3796},
+ {10, 3798}, {10, 3800}, {10, 3802}, {10, 3804}, {10, 3806}, {10, 3808},
+ {10, 3810}, {10, 3812}, {10, 3814}, {10, 3816}, {10, 3818}, {10, 3820},
+ {10, 3822}, {10, 3824}, {10, 3826}, {10, 3828}, {10, 3830}, {10, 3832},
+ {10, 3834}, {10, 3836}, {10, 3838}, {10, 3840}, {10, 3842}, {10, 3844},
+ {10, 3846}, {10, 3848}, {10, 3850}, {10, 3852}, {10, 3854}, {10, 3856},
+ {10, 3858}, {10, 3860}, {10, 3862}, {10, 3864}, {10, 3866}, {10, 3868},
+ {10, 3870}, {10, 3872}, {10, 3874}, {10, 3876}, {10, 3878}, {10, 3880},
+ {10, 3882}, {10, 3884}, {10, 3886}, {10, 3888}, {10, 3890}, {10, 3892},
+ {10, 3894}, {10, 3896}, {10, 3898}, {10, 3900}, {10, 3902}, {10, 3904},
+ {10, 3906}, {10, 3908}, {10, 3910}, {10, 3912}, {10, 3914}, {10, 3916},
+ {10, 3918}, {10, 3920}, {10, 3922}, {10, 3924}, {10, 3926}, {10, 3928},
+ {10, 3930}, {10, 3932}, {10, 3934}, {10, 3936}, {10, 3938}, {10, 3940},
+ {10, 3942}, {10, 3944}, {10, 3946}, {10, 3948}, {10, 3950}, {10, 3952},
+ {10, 3954}, {10, 3956}, {10, 3958}, {10, 3960}
+};
diff --git a/libvpx/vp8/encoder/defaultcoefcounts.h b/libvpx/vp8/encoder/defaultcoefcounts.h
new file mode 100644
index 0000000..2c0f3dd
--- /dev/null
+++ b/libvpx/vp8/encoder/defaultcoefcounts.h
@@ -0,0 +1,223 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+/* Generated file, included by entropy.c */
+
+static const unsigned int default_coef_counts[BLOCK_TYPES]
+ [COEF_BANDS]
+ [PREV_COEF_CONTEXTS]
+ [MAX_ENTROPY_TOKENS] =
+{
+
+ {
+ /* Block Type ( 0 ) */
+ {
+ /* Coeff Band ( 0 ) */
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ },
+ {
+ /* Coeff Band ( 1 ) */
+ {30190, 26544, 225, 24, 4, 0, 0, 0, 0, 0, 0, 4171593,},
+ {26846, 25157, 1241, 130, 26, 6, 1, 0, 0, 0, 0, 149987,},
+ {10484, 9538, 1006, 160, 36, 18, 0, 0, 0, 0, 0, 15104,},
+ },
+ {
+ /* Coeff Band ( 2 ) */
+ {25842, 40456, 1126, 83, 11, 2, 0, 0, 0, 0, 0, 0,},
+ {9338, 8010, 512, 73, 7, 3, 2, 0, 0, 0, 0, 43294,},
+ {1047, 751, 149, 31, 13, 6, 1, 0, 0, 0, 0, 879,},
+ },
+ {
+ /* Coeff Band ( 3 ) */
+ {26136, 9826, 252, 13, 0, 0, 0, 0, 0, 0, 0, 0,},
+ {8134, 5574, 191, 14, 2, 0, 0, 0, 0, 0, 0, 35302,},
+ { 605, 677, 116, 9, 1, 0, 0, 0, 0, 0, 0, 611,},
+ },
+ {
+ /* Coeff Band ( 4 ) */
+ {10263, 15463, 283, 17, 0, 0, 0, 0, 0, 0, 0, 0,},
+ {2773, 2191, 128, 9, 2, 2, 0, 0, 0, 0, 0, 10073,},
+ { 134, 125, 32, 4, 0, 2, 0, 0, 0, 0, 0, 50,},
+ },
+ {
+ /* Coeff Band ( 5 ) */
+ {10483, 2663, 23, 1, 0, 0, 0, 0, 0, 0, 0, 0,},
+ {2137, 1251, 27, 1, 1, 0, 0, 0, 0, 0, 0, 14362,},
+ { 116, 156, 14, 2, 1, 0, 0, 0, 0, 0, 0, 190,},
+ },
+ {
+ /* Coeff Band ( 6 ) */
+ {40977, 27614, 412, 28, 0, 0, 0, 0, 0, 0, 0, 0,},
+ {6113, 5213, 261, 22, 3, 0, 0, 0, 0, 0, 0, 26164,},
+ { 382, 312, 50, 14, 2, 0, 0, 0, 0, 0, 0, 345,},
+ },
+ {
+ /* Coeff Band ( 7 ) */
+ { 0, 26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 0, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 319,},
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8,},
+ },
+ },
+ {
+ /* Block Type ( 1 ) */
+ {
+ /* Coeff Band ( 0 ) */
+ {3268, 19382, 1043, 250, 93, 82, 49, 26, 17, 8, 25, 82289,},
+ {8758, 32110, 5436, 1832, 827, 668, 420, 153, 24, 0, 3, 52914,},
+ {9337, 23725, 8487, 3954, 2107, 1836, 1069, 399, 59, 0, 0, 18620,},
+ },
+ {
+ /* Coeff Band ( 1 ) */
+ {12419, 8420, 452, 62, 9, 1, 0, 0, 0, 0, 0, 0,},
+ {11715, 8705, 693, 92, 15, 7, 2, 0, 0, 0, 0, 53988,},
+ {7603, 8585, 2306, 778, 270, 145, 39, 5, 0, 0, 0, 9136,},
+ },
+ {
+ /* Coeff Band ( 2 ) */
+ {15938, 14335, 1207, 184, 55, 13, 4, 1, 0, 0, 0, 0,},
+ {7415, 6829, 1138, 244, 71, 26, 7, 0, 0, 0, 0, 9980,},
+ {1580, 1824, 655, 241, 89, 46, 10, 2, 0, 0, 0, 429,},
+ },
+ {
+ /* Coeff Band ( 3 ) */
+ {19453, 5260, 201, 19, 0, 0, 0, 0, 0, 0, 0, 0,},
+ {9173, 3758, 213, 22, 1, 1, 0, 0, 0, 0, 0, 9820,},
+ {1689, 1277, 276, 51, 17, 4, 0, 0, 0, 0, 0, 679,},
+ },
+ {
+ /* Coeff Band ( 4 ) */
+ {12076, 10667, 620, 85, 19, 9, 5, 0, 0, 0, 0, 0,},
+ {4665, 3625, 423, 55, 19, 9, 0, 0, 0, 0, 0, 5127,},
+ { 415, 440, 143, 34, 20, 7, 2, 0, 0, 0, 0, 101,},
+ },
+ {
+ /* Coeff Band ( 5 ) */
+ {12183, 4846, 115, 11, 1, 0, 0, 0, 0, 0, 0, 0,},
+ {4226, 3149, 177, 21, 2, 0, 0, 0, 0, 0, 0, 7157,},
+ { 375, 621, 189, 51, 11, 4, 1, 0, 0, 0, 0, 198,},
+ },
+ {
+ /* Coeff Band ( 6 ) */
+ {61658, 37743, 1203, 94, 10, 3, 0, 0, 0, 0, 0, 0,},
+ {15514, 11563, 903, 111, 14, 5, 0, 0, 0, 0, 0, 25195,},
+ { 929, 1077, 291, 78, 14, 7, 1, 0, 0, 0, 0, 507,},
+ },
+ {
+ /* Coeff Band ( 7 ) */
+ { 0, 990, 15, 3, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 0, 412, 13, 0, 0, 0, 0, 0, 0, 0, 0, 1641,},
+ { 0, 18, 7, 1, 0, 0, 0, 0, 0, 0, 0, 30,},
+ },
+ },
+ {
+ /* Block Type ( 2 ) */
+ {
+ /* Coeff Band ( 0 ) */
+ { 953, 24519, 628, 120, 28, 12, 4, 0, 0, 0, 0, 2248798,},
+ {1525, 25654, 2647, 617, 239, 143, 42, 5, 0, 0, 0, 66837,},
+ {1180, 11011, 3001, 1237, 532, 448, 239, 54, 5, 0, 0, 7122,},
+ },
+ {
+ /* Coeff Band ( 1 ) */
+ {1356, 2220, 67, 10, 4, 1, 0, 0, 0, 0, 0, 0,},
+ {1450, 2544, 102, 18, 4, 3, 0, 0, 0, 0, 0, 57063,},
+ {1182, 2110, 470, 130, 41, 21, 0, 0, 0, 0, 0, 6047,},
+ },
+ {
+ /* Coeff Band ( 2 ) */
+ { 370, 3378, 200, 30, 5, 4, 1, 0, 0, 0, 0, 0,},
+ { 293, 1006, 131, 29, 11, 0, 0, 0, 0, 0, 0, 5404,},
+ { 114, 387, 98, 23, 4, 8, 1, 0, 0, 0, 0, 236,},
+ },
+ {
+ /* Coeff Band ( 3 ) */
+ { 579, 194, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 395, 213, 5, 1, 0, 0, 0, 0, 0, 0, 0, 4157,},
+ { 119, 122, 4, 0, 0, 0, 0, 0, 0, 0, 0, 300,},
+ },
+ {
+ /* Coeff Band ( 4 ) */
+ { 38, 557, 19, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 21, 114, 12, 1, 0, 0, 0, 0, 0, 0, 0, 427,},
+ { 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7,},
+ },
+ {
+ /* Coeff Band ( 5 ) */
+ { 52, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 18, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 652,},
+ { 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 30,},
+ },
+ {
+ /* Coeff Band ( 6 ) */
+ { 640, 569, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 25, 77, 2, 0, 0, 0, 0, 0, 0, 0, 0, 517,},
+ { 4, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3,},
+ },
+ {
+ /* Coeff Band ( 7 ) */
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ },
+ },
+ {
+ /* Block Type ( 3 ) */
+ {
+ /* Coeff Band ( 0 ) */
+ {2506, 20161, 2707, 767, 261, 178, 107, 30, 14, 3, 0, 100694,},
+ {8806, 36478, 8817, 3268, 1280, 850, 401, 114, 42, 0, 0, 58572,},
+ {11003, 27214, 11798, 5716, 2482, 2072, 1048, 175, 32, 0, 0, 19284,},
+ },
+ {
+ /* Coeff Band ( 1 ) */
+ {9738, 11313, 959, 205, 70, 18, 11, 1, 0, 0, 0, 0,},
+ {12628, 15085, 1507, 273, 52, 19, 9, 0, 0, 0, 0, 54280,},
+ {10701, 15846, 5561, 1926, 813, 570, 249, 36, 0, 0, 0, 6460,},
+ },
+ {
+ /* Coeff Band ( 2 ) */
+ {6781, 22539, 2784, 634, 182, 123, 20, 4, 0, 0, 0, 0,},
+ {6263, 11544, 2649, 790, 259, 168, 27, 5, 0, 0, 0, 20539,},
+ {3109, 4075, 2031, 896, 457, 386, 158, 29, 0, 0, 0, 1138,},
+ },
+ {
+ /* Coeff Band ( 3 ) */
+ {11515, 4079, 465, 73, 5, 14, 2, 0, 0, 0, 0, 0,},
+ {9361, 5834, 650, 96, 24, 8, 4, 0, 0, 0, 0, 22181,},
+ {4343, 3974, 1360, 415, 132, 96, 14, 1, 0, 0, 0, 1267,},
+ },
+ {
+ /* Coeff Band ( 4 ) */
+ {4787, 9297, 823, 168, 44, 12, 4, 0, 0, 0, 0, 0,},
+ {3619, 4472, 719, 198, 60, 31, 3, 0, 0, 0, 0, 8401,},
+ {1157, 1175, 483, 182, 88, 31, 8, 0, 0, 0, 0, 268,},
+ },
+ {
+ /* Coeff Band ( 5 ) */
+ {8299, 1226, 32, 5, 1, 0, 0, 0, 0, 0, 0, 0,},
+ {3502, 1568, 57, 4, 1, 1, 0, 0, 0, 0, 0, 9811,},
+ {1055, 1070, 166, 29, 6, 1, 0, 0, 0, 0, 0, 527,},
+ },
+ {
+ /* Coeff Band ( 6 ) */
+ {27414, 27927, 1989, 347, 69, 26, 0, 0, 0, 0, 0, 0,},
+ {5876, 10074, 1574, 341, 91, 24, 4, 0, 0, 0, 0, 21954,},
+ {1571, 2171, 778, 324, 124, 65, 16, 0, 0, 0, 0, 979,},
+ },
+ {
+ /* Coeff Band ( 7 ) */
+ { 0, 29, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
+ { 0, 23, 0, 0, 0, 0, 0, 0, 0, 0, 0, 459,},
+ { 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 13,},
+ },
+ },
+};
diff --git a/libvpx/vp8/encoder/denoising.c b/libvpx/vp8/encoder/denoising.c
new file mode 100644
index 0000000..c0dd7c1
--- /dev/null
+++ b/libvpx/vp8/encoder/denoising.c
@@ -0,0 +1,309 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "denoising.h"
+
+#include "vp8/common/reconinter.h"
+#include "vpx/vpx_integer.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vpx_rtcd.h"
+
+static const unsigned int NOISE_MOTION_THRESHOLD = 25 * 25;
+/* SSE_DIFF_THRESHOLD is selected as ~95% confidence assuming
+ * var(noise) ~= 100.
+ */
+static const unsigned int SSE_DIFF_THRESHOLD = 16 * 16 * 20;
+static const unsigned int SSE_THRESHOLD = 16 * 16 * 40;
+
+/*
+ * The filter function was modified to reduce the computational complexity.
+ * Step 1:
+ * Instead of applying tap coefficients for each pixel, we calculated the
+ * pixel adjustments vs. pixel diff value ahead of time.
+ * adjustment = filtered_value - current_raw
+ * = (filter_coefficient * diff + 128) >> 8
+ * where
+ * filter_coefficient = (255 << 8) / (256 + ((absdiff * 330) >> 3));
+ * filter_coefficient += filter_coefficient /
+ * (3 + motion_magnitude_adjustment);
+ * filter_coefficient is clamped to 0 ~ 255.
+ *
+ * Step 2:
+ * The adjustment vs. diff curve becomes flat very quick when diff increases.
+ * This allowed us to use only several levels to approximate the curve without
+ * changing the filtering algorithm too much.
+ * The adjustments were further corrected by checking the motion magnitude.
+ * The levels used are:
+ * diff adjustment w/o motion correction adjustment w/ motion correction
+ * [-255, -16] -6 -7
+ * [-15, -8] -4 -5
+ * [-7, -4] -3 -4
+ * [-3, 3] diff diff
+ * [4, 7] 3 4
+ * [8, 15] 4 5
+ * [16, 255] 6 7
+ */
+
+int vp8_denoiser_filter_c(YV12_BUFFER_CONFIG *mc_running_avg,
+ YV12_BUFFER_CONFIG *running_avg, MACROBLOCK *signal,
+ unsigned int motion_magnitude, int y_offset,
+ int uv_offset)
+{
+ unsigned char *sig = signal->thismb;
+ int sig_stride = 16;
+ unsigned char *mc_running_avg_y = mc_running_avg->y_buffer + y_offset;
+ int mc_avg_y_stride = mc_running_avg->y_stride;
+ unsigned char *running_avg_y = running_avg->y_buffer + y_offset;
+ int avg_y_stride = running_avg->y_stride;
+ int r, c, i;
+ int sum_diff = 0;
+ int adj_val[3] = {3, 4, 6};
+
+ /* If motion_magnitude is small, making the denoiser more aggressive by
+ * increasing the adjustment for each level. */
+ if (motion_magnitude <= MOTION_MAGNITUDE_THRESHOLD)
+ {
+ for (i = 0; i < 3; i++)
+ adj_val[i] += 1;
+ }
+
+ for (r = 0; r < 16; ++r)
+ {
+ for (c = 0; c < 16; ++c)
+ {
+ int diff = 0;
+ int adjustment = 0;
+ int absdiff = 0;
+
+ diff = mc_running_avg_y[c] - sig[c];
+ absdiff = abs(diff);
+
+ /* When |diff| < 4, use pixel value from last denoised raw. */
+ if (absdiff <= 3)
+ {
+ running_avg_y[c] = mc_running_avg_y[c];
+ sum_diff += diff;
+ }
+ else
+ {
+ if (absdiff >= 4 && absdiff <= 7)
+ adjustment = adj_val[0];
+ else if (absdiff >= 8 && absdiff <= 15)
+ adjustment = adj_val[1];
+ else
+ adjustment = adj_val[2];
+
+ if (diff > 0)
+ {
+ if ((sig[c] + adjustment) > 255)
+ running_avg_y[c] = 255;
+ else
+ running_avg_y[c] = sig[c] + adjustment;
+
+ sum_diff += adjustment;
+ }
+ else
+ {
+ if ((sig[c] - adjustment) < 0)
+ running_avg_y[c] = 0;
+ else
+ running_avg_y[c] = sig[c] - adjustment;
+
+ sum_diff -= adjustment;
+ }
+ }
+ }
+
+ /* Update pointers for next iteration. */
+ sig += sig_stride;
+ mc_running_avg_y += mc_avg_y_stride;
+ running_avg_y += avg_y_stride;
+ }
+
+ if (abs(sum_diff) > SUM_DIFF_THRESHOLD)
+ return COPY_BLOCK;
+
+ vp8_copy_mem16x16(running_avg->y_buffer + y_offset, avg_y_stride,
+ signal->thismb, sig_stride);
+ return FILTER_BLOCK;
+}
+
+int vp8_denoiser_allocate(VP8_DENOISER *denoiser, int width, int height)
+{
+ int i;
+ assert(denoiser);
+
+ /* don't need one for intra start at 1 */
+ for (i = 1; i < MAX_REF_FRAMES; i++)
+ {
+ denoiser->yv12_running_avg[i].flags = 0;
+
+ if (vp8_yv12_alloc_frame_buffer(&(denoiser->yv12_running_avg[i]), width,
+ height, VP8BORDERINPIXELS)
+ < 0)
+ {
+ vp8_denoiser_free(denoiser);
+ return 1;
+ }
+ vpx_memset(denoiser->yv12_running_avg[i].buffer_alloc, 0,
+ denoiser->yv12_running_avg[i].frame_size);
+
+ }
+ denoiser->yv12_mc_running_avg.flags = 0;
+
+ if (vp8_yv12_alloc_frame_buffer(&(denoiser->yv12_mc_running_avg), width,
+ height, VP8BORDERINPIXELS) < 0)
+ {
+ vp8_denoiser_free(denoiser);
+ return 1;
+ }
+
+ vpx_memset(denoiser->yv12_mc_running_avg.buffer_alloc, 0,
+ denoiser->yv12_mc_running_avg.frame_size);
+ return 0;
+}
+
+void vp8_denoiser_free(VP8_DENOISER *denoiser)
+{
+ int i;
+ assert(denoiser);
+
+ /* we don't have one for intra ref frame */
+ for (i = 1; i < MAX_REF_FRAMES ; i++)
+ {
+ vp8_yv12_de_alloc_frame_buffer(&denoiser->yv12_running_avg[i]);
+ }
+ vp8_yv12_de_alloc_frame_buffer(&denoiser->yv12_mc_running_avg);
+}
+
+
+void vp8_denoiser_denoise_mb(VP8_DENOISER *denoiser,
+ MACROBLOCK *x,
+ unsigned int best_sse,
+ unsigned int zero_mv_sse,
+ int recon_yoffset,
+ int recon_uvoffset)
+{
+ int mv_row;
+ int mv_col;
+ unsigned int motion_magnitude2;
+
+ MV_REFERENCE_FRAME frame = x->best_reference_frame;
+ MV_REFERENCE_FRAME zero_frame = x->best_zeromv_reference_frame;
+
+ enum vp8_denoiser_decision decision = FILTER_BLOCK;
+
+ if (zero_frame)
+ {
+ YV12_BUFFER_CONFIG *src = &denoiser->yv12_running_avg[frame];
+ YV12_BUFFER_CONFIG *dst = &denoiser->yv12_mc_running_avg;
+ YV12_BUFFER_CONFIG saved_pre,saved_dst;
+ MB_MODE_INFO saved_mbmi;
+ MACROBLOCKD *filter_xd = &x->e_mbd;
+ MB_MODE_INFO *mbmi = &filter_xd->mode_info_context->mbmi;
+ int mv_col;
+ int mv_row;
+ int sse_diff = zero_mv_sse - best_sse;
+
+ saved_mbmi = *mbmi;
+
+ /* Use the best MV for the compensation. */
+ mbmi->ref_frame = x->best_reference_frame;
+ mbmi->mode = x->best_sse_inter_mode;
+ mbmi->mv = x->best_sse_mv;
+ mbmi->need_to_clamp_mvs = x->need_to_clamp_best_mvs;
+ mv_col = x->best_sse_mv.as_mv.col;
+ mv_row = x->best_sse_mv.as_mv.row;
+
+ if (frame == INTRA_FRAME ||
+ ((unsigned int)(mv_row *mv_row + mv_col *mv_col)
+ <= NOISE_MOTION_THRESHOLD &&
+ sse_diff < (int)SSE_DIFF_THRESHOLD))
+ {
+ /*
+ * Handle intra blocks as referring to last frame with zero motion
+ * and let the absolute pixel difference affect the filter factor.
+ * Also consider small amount of motion as being random walk due
+ * to noise, if it doesn't mean that we get a much bigger error.
+ * Note that any changes to the mode info only affects the
+ * denoising.
+ */
+ mbmi->ref_frame =
+ x->best_zeromv_reference_frame;
+
+ src = &denoiser->yv12_running_avg[zero_frame];
+
+ mbmi->mode = ZEROMV;
+ mbmi->mv.as_int = 0;
+ x->best_sse_inter_mode = ZEROMV;
+ x->best_sse_mv.as_int = 0;
+ best_sse = zero_mv_sse;
+ }
+
+ saved_pre = filter_xd->pre;
+ saved_dst = filter_xd->dst;
+
+ /* Compensate the running average. */
+ filter_xd->pre.y_buffer = src->y_buffer + recon_yoffset;
+ filter_xd->pre.u_buffer = src->u_buffer + recon_uvoffset;
+ filter_xd->pre.v_buffer = src->v_buffer + recon_uvoffset;
+ /* Write the compensated running average to the destination buffer. */
+ filter_xd->dst.y_buffer = dst->y_buffer + recon_yoffset;
+ filter_xd->dst.u_buffer = dst->u_buffer + recon_uvoffset;
+ filter_xd->dst.v_buffer = dst->v_buffer + recon_uvoffset;
+
+ if (!x->skip)
+ {
+ vp8_build_inter_predictors_mb(filter_xd);
+ }
+ else
+ {
+ vp8_build_inter16x16_predictors_mb(filter_xd,
+ filter_xd->dst.y_buffer,
+ filter_xd->dst.u_buffer,
+ filter_xd->dst.v_buffer,
+ filter_xd->dst.y_stride,
+ filter_xd->dst.uv_stride);
+ }
+ filter_xd->pre = saved_pre;
+ filter_xd->dst = saved_dst;
+ *mbmi = saved_mbmi;
+
+ }
+
+ mv_row = x->best_sse_mv.as_mv.row;
+ mv_col = x->best_sse_mv.as_mv.col;
+ motion_magnitude2 = mv_row * mv_row + mv_col * mv_col;
+ if (best_sse > SSE_THRESHOLD || motion_magnitude2
+ > 8 * NOISE_MOTION_THRESHOLD)
+ {
+ decision = COPY_BLOCK;
+ }
+
+ if (decision == FILTER_BLOCK)
+ {
+ /* Filter. */
+ decision = vp8_denoiser_filter(&denoiser->yv12_mc_running_avg,
+ &denoiser->yv12_running_avg[LAST_FRAME],
+ x,
+ motion_magnitude2,
+ recon_yoffset, recon_uvoffset);
+ }
+ if (decision == COPY_BLOCK)
+ {
+ /* No filtering of this block; it differs too much from the predictor,
+ * or the motion vector magnitude is considered too big.
+ */
+ vp8_copy_mem16x16(
+ x->thismb, 16,
+ denoiser->yv12_running_avg[LAST_FRAME].y_buffer + recon_yoffset,
+ denoiser->yv12_running_avg[LAST_FRAME].y_stride);
+ }
+}
diff --git a/libvpx/vp8/encoder/denoising.h b/libvpx/vp8/encoder/denoising.h
new file mode 100644
index 0000000..b025f5c
--- /dev/null
+++ b/libvpx/vp8/encoder/denoising.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef VP8_ENCODER_DENOISING_H_
+#define VP8_ENCODER_DENOISING_H_
+
+#include "block.h"
+
+#define SUM_DIFF_THRESHOLD (16 * 16 * 2)
+#define MOTION_MAGNITUDE_THRESHOLD (8*3)
+
+enum vp8_denoiser_decision
+{
+ COPY_BLOCK,
+ FILTER_BLOCK
+};
+
+typedef struct vp8_denoiser
+{
+ YV12_BUFFER_CONFIG yv12_running_avg[MAX_REF_FRAMES];
+ YV12_BUFFER_CONFIG yv12_mc_running_avg;
+} VP8_DENOISER;
+
+int vp8_denoiser_allocate(VP8_DENOISER *denoiser, int width, int height);
+
+void vp8_denoiser_free(VP8_DENOISER *denoiser);
+
+void vp8_denoiser_denoise_mb(VP8_DENOISER *denoiser,
+ MACROBLOCK *x,
+ unsigned int best_sse,
+ unsigned int zero_mv_sse,
+ int recon_yoffset,
+ int recon_uvoffset);
+
+#endif /* VP8_ENCODER_DENOISING_H_ */
diff --git a/libvpx/vp8/encoder/encodeframe.c b/libvpx/vp8/encoder/encodeframe.c
new file mode 100644
index 0000000..8828dd9
--- /dev/null
+++ b/libvpx/vp8/encoder/encodeframe.c
@@ -0,0 +1,1393 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "encodemb.h"
+#include "encodemv.h"
+#include "vp8/common/common.h"
+#include "onyx_int.h"
+#include "vp8/common/extend.h"
+#include "vp8/common/entropymode.h"
+#include "vp8/common/quant_common.h"
+#include "segmentation.h"
+#include "vp8/common/setupintrarecon.h"
+#include "encodeintra.h"
+#include "vp8/common/reconinter.h"
+#include "rdopt.h"
+#include "pickinter.h"
+#include "vp8/common/findnearmv.h"
+#include <stdio.h>
+#include <limits.h>
+#include "vp8/common/invtrans.h"
+#include "vpx_ports/vpx_timer.h"
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+#include "bitstream.h"
+#endif
+#include "encodeframe.h"
+
+extern void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) ;
+extern void vp8_calc_ref_frame_costs(int *ref_frame_cost,
+ int prob_intra,
+ int prob_last,
+ int prob_garf
+ );
+extern void vp8_convert_rfct_to_prob(VP8_COMP *const cpi);
+extern void vp8cx_initialize_me_consts(VP8_COMP *cpi, int QIndex);
+extern void vp8_auto_select_speed(VP8_COMP *cpi);
+extern void vp8cx_init_mbrthread_data(VP8_COMP *cpi,
+ MACROBLOCK *x,
+ MB_ROW_COMP *mbr_ei,
+ int mb_row,
+ int count);
+static void adjust_act_zbin( VP8_COMP *cpi, MACROBLOCK *x );
+
+#ifdef MODE_STATS
+unsigned int inter_y_modes[10] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+unsigned int inter_uv_modes[4] = {0, 0, 0, 0};
+unsigned int inter_b_modes[15] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+unsigned int y_modes[5] = {0, 0, 0, 0, 0};
+unsigned int uv_modes[4] = {0, 0, 0, 0};
+unsigned int b_modes[14] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+#endif
+
+
+/* activity_avg must be positive, or flat regions could get a zero weight
+ * (infinite lambda), which confounds analysis.
+ * This also avoids the need for divide by zero checks in
+ * vp8_activity_masking().
+ */
+#define VP8_ACTIVITY_AVG_MIN (64)
+
+/* This is used as a reference when computing the source variance for the
+ * purposes of activity masking.
+ * Eventually this should be replaced by custom no-reference routines,
+ * which will be faster.
+ */
+static const unsigned char VP8_VAR_OFFS[16]=
+{
+ 128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128
+};
+
+
+/* Original activity measure from Tim T's code. */
+static unsigned int tt_activity_measure( VP8_COMP *cpi, MACROBLOCK *x )
+{
+ unsigned int act;
+ unsigned int sse;
+ /* TODO: This could also be done over smaller areas (8x8), but that would
+ * require extensive changes elsewhere, as lambda is assumed to be fixed
+ * over an entire MB in most of the code.
+ * Another option is to compute four 8x8 variances, and pick a single
+ * lambda using a non-linear combination (e.g., the smallest, or second
+ * smallest, etc.).
+ */
+ act = vp8_variance16x16(x->src.y_buffer,
+ x->src.y_stride, VP8_VAR_OFFS, 0, &sse);
+ act = act<<4;
+
+ /* If the region is flat, lower the activity some more. */
+ if (act < 8<<12)
+ act = act < 5<<12 ? act : 5<<12;
+
+ return act;
+}
+
+/* Stub for alternative experimental activity measures. */
+static unsigned int alt_activity_measure( VP8_COMP *cpi,
+ MACROBLOCK *x, int use_dc_pred )
+{
+ return vp8_encode_intra(cpi,x, use_dc_pred);
+}
+
+
+/* Measure the activity of the current macroblock
+ * What we measure here is TBD so abstracted to this function
+ */
+#define ALT_ACT_MEASURE 1
+static unsigned int mb_activity_measure( VP8_COMP *cpi, MACROBLOCK *x,
+ int mb_row, int mb_col)
+{
+ unsigned int mb_activity;
+
+ if ( ALT_ACT_MEASURE )
+ {
+ int use_dc_pred = (mb_col || mb_row) && (!mb_col || !mb_row);
+
+ /* Or use and alternative. */
+ mb_activity = alt_activity_measure( cpi, x, use_dc_pred );
+ }
+ else
+ {
+ /* Original activity measure from Tim T's code. */
+ mb_activity = tt_activity_measure( cpi, x );
+ }
+
+ if ( mb_activity < VP8_ACTIVITY_AVG_MIN )
+ mb_activity = VP8_ACTIVITY_AVG_MIN;
+
+ return mb_activity;
+}
+
+/* Calculate an "average" mb activity value for the frame */
+#define ACT_MEDIAN 0
+static void calc_av_activity( VP8_COMP *cpi, int64_t activity_sum )
+{
+#if ACT_MEDIAN
+ /* Find median: Simple n^2 algorithm for experimentation */
+ {
+ unsigned int median;
+ unsigned int i,j;
+ unsigned int * sortlist;
+ unsigned int tmp;
+
+ /* Create a list to sort to */
+ CHECK_MEM_ERROR(sortlist,
+ vpx_calloc(sizeof(unsigned int),
+ cpi->common.MBs));
+
+ /* Copy map to sort list */
+ vpx_memcpy( sortlist, cpi->mb_activity_map,
+ sizeof(unsigned int) * cpi->common.MBs );
+
+
+ /* Ripple each value down to its correct position */
+ for ( i = 1; i < cpi->common.MBs; i ++ )
+ {
+ for ( j = i; j > 0; j -- )
+ {
+ if ( sortlist[j] < sortlist[j-1] )
+ {
+ /* Swap values */
+ tmp = sortlist[j-1];
+ sortlist[j-1] = sortlist[j];
+ sortlist[j] = tmp;
+ }
+ else
+ break;
+ }
+ }
+
+ /* Even number MBs so estimate median as mean of two either side. */
+ median = ( 1 + sortlist[cpi->common.MBs >> 1] +
+ sortlist[(cpi->common.MBs >> 1) + 1] ) >> 1;
+
+ cpi->activity_avg = median;
+
+ vpx_free(sortlist);
+ }
+#else
+ /* Simple mean for now */
+ cpi->activity_avg = (unsigned int)(activity_sum/cpi->common.MBs);
+#endif
+
+ if (cpi->activity_avg < VP8_ACTIVITY_AVG_MIN)
+ cpi->activity_avg = VP8_ACTIVITY_AVG_MIN;
+
+ /* Experimental code: return fixed value normalized for several clips */
+ if ( ALT_ACT_MEASURE )
+ cpi->activity_avg = 100000;
+}
+
+#define USE_ACT_INDEX 0
+#define OUTPUT_NORM_ACT_STATS 0
+
+#if USE_ACT_INDEX
+/* Calculate and activity index for each mb */
+static void calc_activity_index( VP8_COMP *cpi, MACROBLOCK *x )
+{
+ VP8_COMMON *const cm = & cpi->common;
+ int mb_row, mb_col;
+
+ int64_t act;
+ int64_t a;
+ int64_t b;
+
+#if OUTPUT_NORM_ACT_STATS
+ FILE *f = fopen("norm_act.stt", "a");
+ fprintf(f, "\n%12d\n", cpi->activity_avg );
+#endif
+
+ /* Reset pointers to start of activity map */
+ x->mb_activity_ptr = cpi->mb_activity_map;
+
+ /* Calculate normalized mb activity number. */
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
+ {
+ /* for each macroblock col in image */
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+ /* Read activity from the map */
+ act = *(x->mb_activity_ptr);
+
+ /* Calculate a normalized activity number */
+ a = act + 4*cpi->activity_avg;
+ b = 4*act + cpi->activity_avg;
+
+ if ( b >= a )
+ *(x->activity_ptr) = (int)((b + (a>>1))/a) - 1;
+ else
+ *(x->activity_ptr) = 1 - (int)((a + (b>>1))/b);
+
+#if OUTPUT_NORM_ACT_STATS
+ fprintf(f, " %6d", *(x->mb_activity_ptr));
+#endif
+ /* Increment activity map pointers */
+ x->mb_activity_ptr++;
+ }
+
+#if OUTPUT_NORM_ACT_STATS
+ fprintf(f, "\n");
+#endif
+
+ }
+
+#if OUTPUT_NORM_ACT_STATS
+ fclose(f);
+#endif
+
+}
+#endif
+
+/* Loop through all MBs. Note activity of each, average activity and
+ * calculate a normalized activity for each
+ */
+static void build_activity_map( VP8_COMP *cpi )
+{
+ MACROBLOCK *const x = & cpi->mb;
+ MACROBLOCKD *xd = &x->e_mbd;
+ VP8_COMMON *const cm = & cpi->common;
+
+#if ALT_ACT_MEASURE
+ YV12_BUFFER_CONFIG *new_yv12 = &cm->yv12_fb[cm->new_fb_idx];
+ int recon_yoffset;
+ int recon_y_stride = new_yv12->y_stride;
+#endif
+
+ int mb_row, mb_col;
+ unsigned int mb_activity;
+ int64_t activity_sum = 0;
+
+ /* for each macroblock row in image */
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
+ {
+#if ALT_ACT_MEASURE
+ /* reset above block coeffs */
+ xd->up_available = (mb_row != 0);
+ recon_yoffset = (mb_row * recon_y_stride * 16);
+#endif
+ /* for each macroblock col in image */
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+#if ALT_ACT_MEASURE
+ xd->dst.y_buffer = new_yv12->y_buffer + recon_yoffset;
+ xd->left_available = (mb_col != 0);
+ recon_yoffset += 16;
+#endif
+ /* Copy current mb to a buffer */
+ vp8_copy_mem16x16(x->src.y_buffer, x->src.y_stride, x->thismb, 16);
+
+ /* measure activity */
+ mb_activity = mb_activity_measure( cpi, x, mb_row, mb_col );
+
+ /* Keep frame sum */
+ activity_sum += mb_activity;
+
+ /* Store MB level activity details. */
+ *x->mb_activity_ptr = mb_activity;
+
+ /* Increment activity map pointer */
+ x->mb_activity_ptr++;
+
+ /* adjust to the next column of source macroblocks */
+ x->src.y_buffer += 16;
+ }
+
+
+ /* adjust to the next row of mbs */
+ x->src.y_buffer += 16 * x->src.y_stride - 16 * cm->mb_cols;
+
+#if ALT_ACT_MEASURE
+ /* extend the recon for intra prediction */
+ vp8_extend_mb_row(new_yv12, xd->dst.y_buffer + 16,
+ xd->dst.u_buffer + 8, xd->dst.v_buffer + 8);
+#endif
+
+ }
+
+ /* Calculate an "average" MB activity */
+ calc_av_activity(cpi, activity_sum);
+
+#if USE_ACT_INDEX
+ /* Calculate an activity index number of each mb */
+ calc_activity_index( cpi, x );
+#endif
+
+}
+
+/* Macroblock activity masking */
+void vp8_activity_masking(VP8_COMP *cpi, MACROBLOCK *x)
+{
+#if USE_ACT_INDEX
+ x->rdmult += *(x->mb_activity_ptr) * (x->rdmult >> 2);
+ x->errorperbit = x->rdmult * 100 /(110 * x->rddiv);
+ x->errorperbit += (x->errorperbit==0);
+#else
+ int64_t a;
+ int64_t b;
+ int64_t act = *(x->mb_activity_ptr);
+
+ /* Apply the masking to the RD multiplier. */
+ a = act + (2*cpi->activity_avg);
+ b = (2*act) + cpi->activity_avg;
+
+ x->rdmult = (unsigned int)(((int64_t)x->rdmult*b + (a>>1))/a);
+ x->errorperbit = x->rdmult * 100 /(110 * x->rddiv);
+ x->errorperbit += (x->errorperbit==0);
+#endif
+
+ /* Activity based Zbin adjustment */
+ adjust_act_zbin(cpi, x);
+}
+
+static
+void encode_mb_row(VP8_COMP *cpi,
+ VP8_COMMON *cm,
+ int mb_row,
+ MACROBLOCK *x,
+ MACROBLOCKD *xd,
+ TOKENEXTRA **tp,
+ int *segment_counts,
+ int *totalrate)
+{
+ int recon_yoffset, recon_uvoffset;
+ int mb_col;
+ int ref_fb_idx = cm->lst_fb_idx;
+ int dst_fb_idx = cm->new_fb_idx;
+ int recon_y_stride = cm->yv12_fb[ref_fb_idx].y_stride;
+ int recon_uv_stride = cm->yv12_fb[ref_fb_idx].uv_stride;
+ int map_index = (mb_row * cpi->common.mb_cols);
+
+#if (CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ const int num_part = (1 << cm->multi_token_partition);
+ TOKENEXTRA * tp_start = cpi->tok;
+ vp8_writer *w;
+#endif
+
+#if CONFIG_MULTITHREAD
+ const int nsync = cpi->mt_sync_range;
+ const int rightmost_col = cm->mb_cols + nsync;
+ volatile const int *last_row_current_mb_col;
+ volatile int *current_mb_col = &cpi->mt_current_mb_col[mb_row];
+
+ if ((cpi->b_multi_threaded != 0) && (mb_row != 0))
+ last_row_current_mb_col = &cpi->mt_current_mb_col[mb_row - 1];
+ else
+ last_row_current_mb_col = &rightmost_col;
+#endif
+
+#if (CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ if(num_part > 1)
+ w= &cpi->bc[1 + (mb_row % num_part)];
+ else
+ w = &cpi->bc[1];
+#endif
+
+ /* reset above block coeffs */
+ xd->above_context = cm->above_context;
+
+ xd->up_available = (mb_row != 0);
+ recon_yoffset = (mb_row * recon_y_stride * 16);
+ recon_uvoffset = (mb_row * recon_uv_stride * 8);
+
+ cpi->tplist[mb_row].start = *tp;
+ /* printf("Main mb_row = %d\n", mb_row); */
+
+ /* Distance of Mb to the top & bottom edges, specified in 1/8th pel
+ * units as they are always compared to values that are in 1/8th pel
+ */
+ xd->mb_to_top_edge = -((mb_row * 16) << 3);
+ xd->mb_to_bottom_edge = ((cm->mb_rows - 1 - mb_row) * 16) << 3;
+
+ /* Set up limit values for vertical motion vector components
+ * to prevent them extending beyond the UMV borders
+ */
+ x->mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 16));
+ x->mv_row_max = ((cm->mb_rows - 1 - mb_row) * 16)
+ + (VP8BORDERINPIXELS - 16);
+
+ /* Set the mb activity pointer to the start of the row. */
+ x->mb_activity_ptr = &cpi->mb_activity_map[map_index];
+
+ /* for each macroblock col in image */
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+
+#if (CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ *tp = cpi->tok;
+#endif
+ /* Distance of Mb to the left & right edges, specified in
+ * 1/8th pel units as they are always compared to values
+ * that are in 1/8th pel units
+ */
+ xd->mb_to_left_edge = -((mb_col * 16) << 3);
+ xd->mb_to_right_edge = ((cm->mb_cols - 1 - mb_col) * 16) << 3;
+
+ /* Set up limit values for horizontal motion vector components
+ * to prevent them extending beyond the UMV borders
+ */
+ x->mv_col_min = -((mb_col * 16) + (VP8BORDERINPIXELS - 16));
+ x->mv_col_max = ((cm->mb_cols - 1 - mb_col) * 16)
+ + (VP8BORDERINPIXELS - 16);
+
+ xd->dst.y_buffer = cm->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset;
+ xd->dst.u_buffer = cm->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset;
+ xd->dst.v_buffer = cm->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset;
+ xd->left_available = (mb_col != 0);
+
+ x->rddiv = cpi->RDDIV;
+ x->rdmult = cpi->RDMULT;
+
+ /* Copy current mb to a buffer */
+ vp8_copy_mem16x16(x->src.y_buffer, x->src.y_stride, x->thismb, 16);
+
+#if CONFIG_MULTITHREAD
+ if (cpi->b_multi_threaded != 0)
+ {
+ *current_mb_col = mb_col - 1; /* set previous MB done */
+
+ if ((mb_col & (nsync - 1)) == 0)
+ {
+ while (mb_col > (*last_row_current_mb_col - nsync))
+ {
+ x86_pause_hint();
+ thread_sleep(0);
+ }
+ }
+ }
+#endif
+
+ if(cpi->oxcf.tuning == VP8_TUNE_SSIM)
+ vp8_activity_masking(cpi, x);
+
+ /* Is segmentation enabled */
+ /* MB level adjustment to quantizer */
+ if (xd->segmentation_enabled)
+ {
+ /* Code to set segment id in xd->mbmi.segment_id for current MB
+ * (with range checking)
+ */
+ if (cpi->segmentation_map[map_index+mb_col] <= 3)
+ xd->mode_info_context->mbmi.segment_id = cpi->segmentation_map[map_index+mb_col];
+ else
+ xd->mode_info_context->mbmi.segment_id = 0;
+
+ vp8cx_mb_init_quantizer(cpi, x, 1);
+ }
+ else
+ /* Set to Segment 0 by default */
+ xd->mode_info_context->mbmi.segment_id = 0;
+
+ x->active_ptr = cpi->active_map + map_index + mb_col;
+
+ if (cm->frame_type == KEY_FRAME)
+ {
+ *totalrate += vp8cx_encode_intra_macroblock(cpi, x, tp);
+#ifdef MODE_STATS
+ y_modes[xd->mbmi.mode] ++;
+#endif
+ }
+ else
+ {
+ *totalrate += vp8cx_encode_inter_macroblock(cpi, x, tp, recon_yoffset, recon_uvoffset, mb_row, mb_col);
+
+#ifdef MODE_STATS
+ inter_y_modes[xd->mbmi.mode] ++;
+
+ if (xd->mbmi.mode == SPLITMV)
+ {
+ int b;
+
+ for (b = 0; b < xd->mbmi.partition_count; b++)
+ {
+ inter_b_modes[x->partition->bmi[b].mode] ++;
+ }
+ }
+
+#endif
+
+ /* Special case code for cyclic refresh
+ * If cyclic update enabled then copy xd->mbmi.segment_id; (which
+ * may have been updated based on mode during
+ * vp8cx_encode_inter_macroblock()) back into the global
+ * segmentation map
+ */
+ if ((cpi->current_layer == 0) &&
+ (cpi->cyclic_refresh_mode_enabled && xd->segmentation_enabled))
+ {
+ cpi->segmentation_map[map_index+mb_col] = xd->mode_info_context->mbmi.segment_id;
+
+ /* If the block has been refreshed mark it as clean (the
+ * magnitude of the -ve influences how long it will be before
+ * we consider another refresh):
+ * Else if it was coded (last frame 0,0) and has not already
+ * been refreshed then mark it as a candidate for cleanup
+ * next time (marked 0) else mark it as dirty (1).
+ */
+ if (xd->mode_info_context->mbmi.segment_id)
+ cpi->cyclic_refresh_map[map_index+mb_col] = -1;
+ else if ((xd->mode_info_context->mbmi.mode == ZEROMV) && (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME))
+ {
+ if (cpi->cyclic_refresh_map[map_index+mb_col] == 1)
+ cpi->cyclic_refresh_map[map_index+mb_col] = 0;
+ }
+ else
+ cpi->cyclic_refresh_map[map_index+mb_col] = 1;
+
+ }
+ }
+
+ cpi->tplist[mb_row].stop = *tp;
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ /* pack tokens for this MB */
+ {
+ int tok_count = *tp - tp_start;
+ pack_tokens(w, tp_start, tok_count);
+ }
+#endif
+ /* Increment pointer into gf usage flags structure. */
+ x->gf_active_ptr++;
+
+ /* Increment the activity mask pointers. */
+ x->mb_activity_ptr++;
+
+ /* adjust to the next column of macroblocks */
+ x->src.y_buffer += 16;
+ x->src.u_buffer += 8;
+ x->src.v_buffer += 8;
+
+ recon_yoffset += 16;
+ recon_uvoffset += 8;
+
+ /* Keep track of segment usage */
+ segment_counts[xd->mode_info_context->mbmi.segment_id] ++;
+
+ /* skip to next mb */
+ xd->mode_info_context++;
+ x->partition_info++;
+ xd->above_context++;
+ }
+
+ /* extend the recon for intra prediction */
+ vp8_extend_mb_row( &cm->yv12_fb[dst_fb_idx],
+ xd->dst.y_buffer + 16,
+ xd->dst.u_buffer + 8,
+ xd->dst.v_buffer + 8);
+
+#if CONFIG_MULTITHREAD
+ if (cpi->b_multi_threaded != 0)
+ *current_mb_col = rightmost_col;
+#endif
+
+ /* this is to account for the border */
+ xd->mode_info_context++;
+ x->partition_info++;
+}
+
+static void init_encode_frame_mb_context(VP8_COMP *cpi)
+{
+ MACROBLOCK *const x = & cpi->mb;
+ VP8_COMMON *const cm = & cpi->common;
+ MACROBLOCKD *const xd = & x->e_mbd;
+
+ /* GF active flags data structure */
+ x->gf_active_ptr = (signed char *)cpi->gf_active_flags;
+
+ /* Activity map pointer */
+ x->mb_activity_ptr = cpi->mb_activity_map;
+
+ x->act_zbin_adj = 0;
+
+ x->partition_info = x->pi;
+
+ xd->mode_info_context = cm->mi;
+ xd->mode_info_stride = cm->mode_info_stride;
+
+ xd->frame_type = cm->frame_type;
+
+ /* reset intra mode contexts */
+ if (cm->frame_type == KEY_FRAME)
+ vp8_init_mbmode_probs(cm);
+
+ /* Copy data over into macro block data structures. */
+ x->src = * cpi->Source;
+ xd->pre = cm->yv12_fb[cm->lst_fb_idx];
+ xd->dst = cm->yv12_fb[cm->new_fb_idx];
+
+ /* set up frame for intra coded blocks */
+ vp8_setup_intra_recon(&cm->yv12_fb[cm->new_fb_idx]);
+
+ vp8_build_block_offsets(x);
+
+ xd->mode_info_context->mbmi.mode = DC_PRED;
+ xd->mode_info_context->mbmi.uv_mode = DC_PRED;
+
+ xd->left_context = &cm->left_context;
+
+ vp8_zero(cpi->count_mb_ref_frame_usage)
+
+ x->mvc = cm->fc.mvc;
+
+ vpx_memset(cm->above_context, 0,
+ sizeof(ENTROPY_CONTEXT_PLANES) * cm->mb_cols);
+
+ /* Special case treatment when GF and ARF are not sensible options
+ * for reference
+ */
+ if (cpi->ref_frame_flags == VP8_LAST_FRAME)
+ vp8_calc_ref_frame_costs(x->ref_frame_cost,
+ cpi->prob_intra_coded,255,128);
+ else if ((cpi->oxcf.number_of_layers > 1) &&
+ (cpi->ref_frame_flags == VP8_GOLD_FRAME))
+ vp8_calc_ref_frame_costs(x->ref_frame_cost,
+ cpi->prob_intra_coded,1,255);
+ else if ((cpi->oxcf.number_of_layers > 1) &&
+ (cpi->ref_frame_flags == VP8_ALTR_FRAME))
+ vp8_calc_ref_frame_costs(x->ref_frame_cost,
+ cpi->prob_intra_coded,1,1);
+ else
+ vp8_calc_ref_frame_costs(x->ref_frame_cost,
+ cpi->prob_intra_coded,
+ cpi->prob_last_coded,
+ cpi->prob_gf_coded);
+
+ xd->fullpixel_mask = 0xffffffff;
+ if(cm->full_pixel)
+ xd->fullpixel_mask = 0xfffffff8;
+
+ vp8_zero(x->coef_counts);
+ vp8_zero(x->ymode_count);
+ vp8_zero(x->uv_mode_count)
+ x->prediction_error = 0;
+ x->intra_error = 0;
+}
+
+static void sum_coef_counts(MACROBLOCK *x, MACROBLOCK *x_thread)
+{
+ int i = 0;
+ do
+ {
+ int j = 0;
+ do
+ {
+ int k = 0;
+ do
+ {
+ /* at every context */
+
+ /* calc probs and branch cts for this frame only */
+ int t = 0; /* token/prob index */
+
+ do
+ {
+ x->coef_counts [i][j][k][t] +=
+ x_thread->coef_counts [i][j][k][t];
+ }
+ while (++t < ENTROPY_NODES);
+ }
+ while (++k < PREV_COEF_CONTEXTS);
+ }
+ while (++j < COEF_BANDS);
+ }
+ while (++i < BLOCK_TYPES);
+}
+
+void vp8_encode_frame(VP8_COMP *cpi)
+{
+ int mb_row;
+ MACROBLOCK *const x = & cpi->mb;
+ VP8_COMMON *const cm = & cpi->common;
+ MACROBLOCKD *const xd = & x->e_mbd;
+ TOKENEXTRA *tp = cpi->tok;
+ int segment_counts[MAX_MB_SEGMENTS];
+ int totalrate;
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ BOOL_CODER * bc = &cpi->bc[1]; /* bc[0] is for control partition */
+ const int num_part = (1 << cm->multi_token_partition);
+#endif
+
+ vpx_memset(segment_counts, 0, sizeof(segment_counts));
+ totalrate = 0;
+
+ if (cpi->compressor_speed == 2)
+ {
+ if (cpi->oxcf.cpu_used < 0)
+ cpi->Speed = -(cpi->oxcf.cpu_used);
+ else
+ vp8_auto_select_speed(cpi);
+ }
+
+ /* Functions setup for all frame types so we can use MC in AltRef */
+ if(!cm->use_bilinear_mc_filter)
+ {
+ xd->subpixel_predict = vp8_sixtap_predict4x4;
+ xd->subpixel_predict8x4 = vp8_sixtap_predict8x4;
+ xd->subpixel_predict8x8 = vp8_sixtap_predict8x8;
+ xd->subpixel_predict16x16 = vp8_sixtap_predict16x16;
+ }
+ else
+ {
+ xd->subpixel_predict = vp8_bilinear_predict4x4;
+ xd->subpixel_predict8x4 = vp8_bilinear_predict8x4;
+ xd->subpixel_predict8x8 = vp8_bilinear_predict8x8;
+ xd->subpixel_predict16x16 = vp8_bilinear_predict16x16;
+ }
+
+ cpi->mb.skip_true_count = 0;
+ cpi->tok_count = 0;
+
+#if 0
+ /* Experimental code */
+ cpi->frame_distortion = 0;
+ cpi->last_mb_distortion = 0;
+#endif
+
+ xd->mode_info_context = cm->mi;
+
+ vp8_zero(cpi->mb.MVcount);
+
+ vp8cx_frame_init_quantizer(cpi);
+
+ vp8_initialize_rd_consts(cpi,
+ vp8_dc_quant(cm->base_qindex, cm->y1dc_delta_q));
+
+ vp8cx_initialize_me_consts(cpi, cm->base_qindex);
+
+ if(cpi->oxcf.tuning == VP8_TUNE_SSIM)
+ {
+ /* Initialize encode frame context. */
+ init_encode_frame_mb_context(cpi);
+
+ /* Build a frame level activity map */
+ build_activity_map(cpi);
+ }
+
+ /* re-init encode frame context. */
+ init_encode_frame_mb_context(cpi);
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ {
+ int i;
+ for(i = 0; i < num_part; i++)
+ {
+ vp8_start_encode(&bc[i], cpi->partition_d[i + 1],
+ cpi->partition_d_end[i + 1]);
+ bc[i].error = &cm->error;
+ }
+ }
+
+#endif
+
+ {
+ struct vpx_usec_timer emr_timer;
+ vpx_usec_timer_start(&emr_timer);
+
+#if CONFIG_MULTITHREAD
+ if (cpi->b_multi_threaded)
+ {
+ int i;
+
+ vp8cx_init_mbrthread_data(cpi, x, cpi->mb_row_ei, 1, cpi->encoding_thread_count);
+
+ for (i = 0; i < cm->mb_rows; i++)
+ cpi->mt_current_mb_col[i] = -1;
+
+ for (i = 0; i < cpi->encoding_thread_count; i++)
+ {
+ sem_post(&cpi->h_event_start_encoding[i]);
+ }
+
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row += (cpi->encoding_thread_count + 1))
+ {
+ vp8_zero(cm->left_context)
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ tp = cpi->tok;
+#else
+ tp = cpi->tok + mb_row * (cm->mb_cols * 16 * 24);
+#endif
+
+ encode_mb_row(cpi, cm, mb_row, x, xd, &tp, segment_counts, &totalrate);
+
+ /* adjust to the next row of mbs */
+ x->src.y_buffer += 16 * x->src.y_stride * (cpi->encoding_thread_count + 1) - 16 * cm->mb_cols;
+ x->src.u_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols;
+ x->src.v_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols;
+
+ xd->mode_info_context += xd->mode_info_stride * cpi->encoding_thread_count;
+ x->partition_info += xd->mode_info_stride * cpi->encoding_thread_count;
+ x->gf_active_ptr += cm->mb_cols * cpi->encoding_thread_count;
+
+ if(mb_row == cm->mb_rows - 1)
+ {
+ sem_post(&cpi->h_event_end_encoding); /* signal frame encoding end */
+ }
+ }
+
+ sem_wait(&cpi->h_event_end_encoding); /* wait for other threads to finish */
+
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row ++)
+ {
+ cpi->tok_count += (unsigned int)
+ (cpi->tplist[mb_row].stop - cpi->tplist[mb_row].start);
+ }
+
+ if (xd->segmentation_enabled)
+ {
+ int i, j;
+
+ if (xd->segmentation_enabled)
+ {
+
+ for (i = 0; i < cpi->encoding_thread_count; i++)
+ {
+ for (j = 0; j < 4; j++)
+ segment_counts[j] += cpi->mb_row_ei[i].segment_counts[j];
+ }
+ }
+ }
+
+ for (i = 0; i < cpi->encoding_thread_count; i++)
+ {
+ int mode_count;
+ int mv_vals;
+ totalrate += cpi->mb_row_ei[i].totalrate;
+
+ cpi->mb.skip_true_count += cpi->mb_row_ei[i].mb.skip_true_count;
+
+ for(mode_count = 0; mode_count < VP8_YMODES; mode_count++)
+ cpi->mb.ymode_count[mode_count] +=
+ cpi->mb_row_ei[i].mb.ymode_count[mode_count];
+
+ for(mode_count = 0; mode_count < VP8_UV_MODES; mode_count++)
+ cpi->mb.uv_mode_count[mode_count] +=
+ cpi->mb_row_ei[i].mb.uv_mode_count[mode_count];
+
+ for(mv_vals = 0; mv_vals < MVvals; mv_vals++)
+ {
+ cpi->mb.MVcount[0][mv_vals] +=
+ cpi->mb_row_ei[i].mb.MVcount[0][mv_vals];
+ cpi->mb.MVcount[1][mv_vals] +=
+ cpi->mb_row_ei[i].mb.MVcount[1][mv_vals];
+ }
+
+ cpi->mb.prediction_error +=
+ cpi->mb_row_ei[i].mb.prediction_error;
+ cpi->mb.intra_error += cpi->mb_row_ei[i].mb.intra_error;
+
+ /* add up counts for each thread */
+ sum_coef_counts(x, &cpi->mb_row_ei[i].mb);
+ }
+
+ }
+ else
+#endif
+ {
+
+ /* for each macroblock row in image */
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
+ {
+ vp8_zero(cm->left_context)
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ tp = cpi->tok;
+#endif
+
+ encode_mb_row(cpi, cm, mb_row, x, xd, &tp, segment_counts, &totalrate);
+
+ /* adjust to the next row of mbs */
+ x->src.y_buffer += 16 * x->src.y_stride - 16 * cm->mb_cols;
+ x->src.u_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols;
+ x->src.v_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols;
+ }
+
+ cpi->tok_count = (unsigned int)(tp - cpi->tok);
+ }
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ {
+ int i;
+ for(i = 0; i < num_part; i++)
+ {
+ vp8_stop_encode(&bc[i]);
+ cpi->partition_sz[i+1] = bc[i].pos;
+ }
+ }
+#endif
+
+ vpx_usec_timer_mark(&emr_timer);
+ cpi->time_encode_mb_row += vpx_usec_timer_elapsed(&emr_timer);
+ }
+
+
+ // Work out the segment probabilities if segmentation is enabled
+ // and needs to be updated
+ if (xd->segmentation_enabled && xd->update_mb_segmentation_map)
+ {
+ int tot_count;
+ int i;
+
+ /* Set to defaults */
+ vpx_memset(xd->mb_segment_tree_probs, 255 , sizeof(xd->mb_segment_tree_probs));
+
+ tot_count = segment_counts[0] + segment_counts[1] + segment_counts[2] + segment_counts[3];
+
+ if (tot_count)
+ {
+ xd->mb_segment_tree_probs[0] = ((segment_counts[0] + segment_counts[1]) * 255) / tot_count;
+
+ tot_count = segment_counts[0] + segment_counts[1];
+
+ if (tot_count > 0)
+ {
+ xd->mb_segment_tree_probs[1] = (segment_counts[0] * 255) / tot_count;
+ }
+
+ tot_count = segment_counts[2] + segment_counts[3];
+
+ if (tot_count > 0)
+ xd->mb_segment_tree_probs[2] = (segment_counts[2] * 255) / tot_count;
+
+ /* Zero probabilities not allowed */
+ for (i = 0; i < MB_FEATURE_TREE_PROBS; i ++)
+ {
+ if (xd->mb_segment_tree_probs[i] == 0)
+ xd->mb_segment_tree_probs[i] = 1;
+ }
+ }
+ }
+
+ /* projected_frame_size in units of BYTES */
+ cpi->projected_frame_size = totalrate >> 8;
+
+ /* Make a note of the percentage MBs coded Intra. */
+ if (cm->frame_type == KEY_FRAME)
+ {
+ cpi->this_frame_percent_intra = 100;
+ }
+ else
+ {
+ int tot_modes;
+
+ tot_modes = cpi->count_mb_ref_frame_usage[INTRA_FRAME]
+ + cpi->count_mb_ref_frame_usage[LAST_FRAME]
+ + cpi->count_mb_ref_frame_usage[GOLDEN_FRAME]
+ + cpi->count_mb_ref_frame_usage[ALTREF_FRAME];
+
+ if (tot_modes)
+ cpi->this_frame_percent_intra = cpi->count_mb_ref_frame_usage[INTRA_FRAME] * 100 / tot_modes;
+
+ }
+
+#if ! CONFIG_REALTIME_ONLY
+ /* Adjust the projected reference frame usage probability numbers to
+ * reflect what we have just seen. This may be useful when we make
+ * multiple iterations of the recode loop rather than continuing to use
+ * values from the previous frame.
+ */
+ if ((cm->frame_type != KEY_FRAME) && ((cpi->oxcf.number_of_layers > 1) ||
+ (!cm->refresh_alt_ref_frame && !cm->refresh_golden_frame)))
+ {
+ vp8_convert_rfct_to_prob(cpi);
+ }
+#endif
+}
+void vp8_setup_block_ptrs(MACROBLOCK *x)
+{
+ int r, c;
+ int i;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ x->block[r*4+c].src_diff = x->src_diff + r * 4 * 16 + c * 4;
+ }
+ }
+
+ for (r = 0; r < 2; r++)
+ {
+ for (c = 0; c < 2; c++)
+ {
+ x->block[16 + r*2+c].src_diff = x->src_diff + 256 + r * 4 * 8 + c * 4;
+ }
+ }
+
+
+ for (r = 0; r < 2; r++)
+ {
+ for (c = 0; c < 2; c++)
+ {
+ x->block[20 + r*2+c].src_diff = x->src_diff + 320 + r * 4 * 8 + c * 4;
+ }
+ }
+
+ x->block[24].src_diff = x->src_diff + 384;
+
+
+ for (i = 0; i < 25; i++)
+ {
+ x->block[i].coeff = x->coeff + i * 16;
+ }
+}
+
+void vp8_build_block_offsets(MACROBLOCK *x)
+{
+ int block = 0;
+ int br, bc;
+
+ vp8_build_block_doffsets(&x->e_mbd);
+
+ /* y blocks */
+ x->thismb_ptr = &x->thismb[0];
+ for (br = 0; br < 4; br++)
+ {
+ for (bc = 0; bc < 4; bc++)
+ {
+ BLOCK *this_block = &x->block[block];
+ this_block->base_src = &x->thismb_ptr;
+ this_block->src_stride = 16;
+ this_block->src = 4 * br * 16 + 4 * bc;
+ ++block;
+ }
+ }
+
+ /* u blocks */
+ for (br = 0; br < 2; br++)
+ {
+ for (bc = 0; bc < 2; bc++)
+ {
+ BLOCK *this_block = &x->block[block];
+ this_block->base_src = &x->src.u_buffer;
+ this_block->src_stride = x->src.uv_stride;
+ this_block->src = 4 * br * this_block->src_stride + 4 * bc;
+ ++block;
+ }
+ }
+
+ /* v blocks */
+ for (br = 0; br < 2; br++)
+ {
+ for (bc = 0; bc < 2; bc++)
+ {
+ BLOCK *this_block = &x->block[block];
+ this_block->base_src = &x->src.v_buffer;
+ this_block->src_stride = x->src.uv_stride;
+ this_block->src = 4 * br * this_block->src_stride + 4 * bc;
+ ++block;
+ }
+ }
+}
+
+static void sum_intra_stats(VP8_COMP *cpi, MACROBLOCK *x)
+{
+ const MACROBLOCKD *xd = & x->e_mbd;
+ const MB_PREDICTION_MODE m = xd->mode_info_context->mbmi.mode;
+ const MB_PREDICTION_MODE uvm = xd->mode_info_context->mbmi.uv_mode;
+
+#ifdef MODE_STATS
+ const int is_key = cpi->common.frame_type == KEY_FRAME;
+
+ ++ (is_key ? uv_modes : inter_uv_modes)[uvm];
+
+ if (m == B_PRED)
+ {
+ unsigned int *const bct = is_key ? b_modes : inter_b_modes;
+
+ int b = 0;
+
+ do
+ {
+ ++ bct[xd->block[b].bmi.mode];
+ }
+ while (++b < 16);
+ }
+
+#endif
+
+ ++x->ymode_count[m];
+ ++x->uv_mode_count[uvm];
+
+}
+
+/* Experimental stub function to create a per MB zbin adjustment based on
+ * some previously calculated measure of MB activity.
+ */
+static void adjust_act_zbin( VP8_COMP *cpi, MACROBLOCK *x )
+{
+#if USE_ACT_INDEX
+ x->act_zbin_adj = *(x->mb_activity_ptr);
+#else
+ int64_t a;
+ int64_t b;
+ int64_t act = *(x->mb_activity_ptr);
+
+ /* Apply the masking to the RD multiplier. */
+ a = act + 4*cpi->activity_avg;
+ b = 4*act + cpi->activity_avg;
+
+ if ( act > cpi->activity_avg )
+ x->act_zbin_adj = (int)(((int64_t)b + (a>>1))/a) - 1;
+ else
+ x->act_zbin_adj = 1 - (int)(((int64_t)a + (b>>1))/b);
+#endif
+}
+
+int vp8cx_encode_intra_macroblock(VP8_COMP *cpi, MACROBLOCK *x,
+ TOKENEXTRA **t)
+{
+ MACROBLOCKD *xd = &x->e_mbd;
+ int rate;
+
+ if (cpi->sf.RD && cpi->compressor_speed != 2)
+ vp8_rd_pick_intra_mode(x, &rate);
+ else
+ vp8_pick_intra_mode(x, &rate);
+
+ if(cpi->oxcf.tuning == VP8_TUNE_SSIM)
+ {
+ adjust_act_zbin( cpi, x );
+ vp8_update_zbin_extra(cpi, x);
+ }
+
+ if (x->e_mbd.mode_info_context->mbmi.mode == B_PRED)
+ vp8_encode_intra4x4mby(x);
+ else
+ vp8_encode_intra16x16mby(x);
+
+ vp8_encode_intra16x16mbuv(x);
+
+ sum_intra_stats(cpi, x);
+
+ vp8_tokenize_mb(cpi, x, t);
+
+ if (xd->mode_info_context->mbmi.mode != B_PRED)
+ vp8_inverse_transform_mby(xd);
+
+ vp8_dequant_idct_add_uv_block
+ (xd->qcoeff+16*16, xd->dequant_uv,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride, xd->eobs+16);
+ return rate;
+}
+#ifdef SPEEDSTATS
+extern int cnt_pm;
+#endif
+
+extern void vp8_fix_contexts(MACROBLOCKD *x);
+
+int vp8cx_encode_inter_macroblock
+(
+ VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t,
+ int recon_yoffset, int recon_uvoffset,
+ int mb_row, int mb_col
+)
+{
+ MACROBLOCKD *const xd = &x->e_mbd;
+ int intra_error = 0;
+ int rate;
+ int distortion;
+
+ x->skip = 0;
+
+ if (xd->segmentation_enabled)
+ x->encode_breakout = cpi->segment_encode_breakout[xd->mode_info_context->mbmi.segment_id];
+ else
+ x->encode_breakout = cpi->oxcf.encode_breakout;
+
+#if CONFIG_TEMPORAL_DENOISING
+ /* Reset the best sse mode/mv for each macroblock. */
+ x->best_reference_frame = INTRA_FRAME;
+ x->best_zeromv_reference_frame = INTRA_FRAME;
+ x->best_sse_inter_mode = 0;
+ x->best_sse_mv.as_int = 0;
+ x->need_to_clamp_best_mvs = 0;
+#endif
+
+ if (cpi->sf.RD)
+ {
+ int zbin_mode_boost_enabled = cpi->zbin_mode_boost_enabled;
+
+ /* Are we using the fast quantizer for the mode selection? */
+ if(cpi->sf.use_fastquant_for_pick)
+ {
+ cpi->mb.quantize_b = vp8_fast_quantize_b;
+ cpi->mb.quantize_b_pair = vp8_fast_quantize_b_pair;
+
+ /* the fast quantizer does not use zbin_extra, so
+ * do not recalculate */
+ cpi->zbin_mode_boost_enabled = 0;
+ }
+ vp8_rd_pick_inter_mode(cpi, x, recon_yoffset, recon_uvoffset, &rate,
+ &distortion, &intra_error);
+
+ /* switch back to the regular quantizer for the encode */
+ if (cpi->sf.improved_quant)
+ {
+ cpi->mb.quantize_b = vp8_regular_quantize_b;
+ cpi->mb.quantize_b_pair = vp8_regular_quantize_b_pair;
+ }
+
+ /* restore cpi->zbin_mode_boost_enabled */
+ cpi->zbin_mode_boost_enabled = zbin_mode_boost_enabled;
+
+ }
+ else
+ {
+ vp8_pick_inter_mode(cpi, x, recon_yoffset, recon_uvoffset, &rate,
+ &distortion, &intra_error, mb_row, mb_col);
+ }
+
+ x->prediction_error += distortion;
+ x->intra_error += intra_error;
+
+ if(cpi->oxcf.tuning == VP8_TUNE_SSIM)
+ {
+ /* Adjust the zbin based on this MB rate. */
+ adjust_act_zbin( cpi, x );
+ }
+
+#if 0
+ /* Experimental RD code */
+ cpi->frame_distortion += distortion;
+ cpi->last_mb_distortion = distortion;
+#endif
+
+ /* MB level adjutment to quantizer setup */
+ if (xd->segmentation_enabled)
+ {
+ /* If cyclic update enabled */
+ if (cpi->current_layer == 0 && cpi->cyclic_refresh_mode_enabled)
+ {
+ /* Clear segment_id back to 0 if not coded (last frame 0,0) */
+ if ((xd->mode_info_context->mbmi.segment_id == 1) &&
+ ((xd->mode_info_context->mbmi.ref_frame != LAST_FRAME) || (xd->mode_info_context->mbmi.mode != ZEROMV)))
+ {
+ xd->mode_info_context->mbmi.segment_id = 0;
+
+ /* segment_id changed, so update */
+ vp8cx_mb_init_quantizer(cpi, x, 1);
+ }
+ }
+ }
+
+ {
+ /* Experimental code. Special case for gf and arf zeromv modes.
+ * Increase zbin size to supress noise
+ */
+ cpi->zbin_mode_boost = 0;
+ if (cpi->zbin_mode_boost_enabled)
+ {
+ if ( xd->mode_info_context->mbmi.ref_frame != INTRA_FRAME )
+ {
+ if (xd->mode_info_context->mbmi.mode == ZEROMV)
+ {
+ if (xd->mode_info_context->mbmi.ref_frame != LAST_FRAME)
+ cpi->zbin_mode_boost = GF_ZEROMV_ZBIN_BOOST;
+ else
+ cpi->zbin_mode_boost = LF_ZEROMV_ZBIN_BOOST;
+ }
+ else if (xd->mode_info_context->mbmi.mode == SPLITMV)
+ cpi->zbin_mode_boost = 0;
+ else
+ cpi->zbin_mode_boost = MV_ZBIN_BOOST;
+ }
+ }
+
+ /* The fast quantizer doesn't use zbin_extra, only do so with
+ * the regular quantizer. */
+ if (cpi->sf.improved_quant)
+ vp8_update_zbin_extra(cpi, x);
+ }
+
+ cpi->count_mb_ref_frame_usage[xd->mode_info_context->mbmi.ref_frame] ++;
+
+ if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
+ {
+ vp8_encode_intra16x16mbuv(x);
+
+ if (xd->mode_info_context->mbmi.mode == B_PRED)
+ {
+ vp8_encode_intra4x4mby(x);
+ }
+ else
+ {
+ vp8_encode_intra16x16mby(x);
+ }
+
+ sum_intra_stats(cpi, x);
+ }
+ else
+ {
+ int ref_fb_idx;
+
+ if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME)
+ ref_fb_idx = cpi->common.lst_fb_idx;
+ else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME)
+ ref_fb_idx = cpi->common.gld_fb_idx;
+ else
+ ref_fb_idx = cpi->common.alt_fb_idx;
+
+ xd->pre.y_buffer = cpi->common.yv12_fb[ref_fb_idx].y_buffer + recon_yoffset;
+ xd->pre.u_buffer = cpi->common.yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset;
+ xd->pre.v_buffer = cpi->common.yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset;
+
+ if (!x->skip)
+ {
+ vp8_encode_inter16x16(x);
+ }
+ else
+ vp8_build_inter16x16_predictors_mb(xd, xd->dst.y_buffer,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.y_stride, xd->dst.uv_stride);
+
+ }
+
+ if (!x->skip)
+ {
+ vp8_tokenize_mb(cpi, x, t);
+
+ if (xd->mode_info_context->mbmi.mode != B_PRED)
+ vp8_inverse_transform_mby(xd);
+
+ vp8_dequant_idct_add_uv_block
+ (xd->qcoeff+16*16, xd->dequant_uv,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride, xd->eobs+16);
+ }
+ else
+ {
+ /* always set mb_skip_coeff as it is needed by the loopfilter */
+ xd->mode_info_context->mbmi.mb_skip_coeff = 1;
+
+ if (cpi->common.mb_no_coeff_skip)
+ {
+ x->skip_true_count ++;
+ vp8_fix_contexts(xd);
+ }
+ else
+ {
+ vp8_stuff_mb(cpi, x, t);
+ }
+ }
+
+ return rate;
+}
diff --git a/libvpx/vp8/encoder/encodeframe.h b/libvpx/vp8/encoder/encodeframe.h
new file mode 100644
index 0000000..4dd6ba0
--- /dev/null
+++ b/libvpx/vp8/encoder/encodeframe.h
@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#ifndef ENCODEFRAME_H
+#define ENCODEFRAME_H
+extern void vp8_activity_masking(VP8_COMP *cpi, MACROBLOCK *x);
+
+extern void vp8_build_block_offsets(MACROBLOCK *x);
+
+extern void vp8_setup_block_ptrs(MACROBLOCK *x);
+
+extern void vp8_encode_frame(VP8_COMP *cpi);
+
+extern int vp8cx_encode_inter_macroblock(VP8_COMP *cpi, MACROBLOCK *x,
+ TOKENEXTRA **t,
+ int recon_yoffset, int recon_uvoffset,
+ int mb_row, int mb_col);
+
+extern int vp8cx_encode_intra_macroblock(VP8_COMP *cpi, MACROBLOCK *x,
+ TOKENEXTRA **t);
+#endif
diff --git a/libvpx/vp8/encoder/encodeintra.c b/libvpx/vp8/encoder/encodeintra.c
new file mode 100644
index 0000000..340dd63
--- /dev/null
+++ b/libvpx/vp8/encoder/encodeintra.c
@@ -0,0 +1,138 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "quantize.h"
+#include "vp8/common/reconintra4x4.h"
+#include "encodemb.h"
+#include "vp8/common/invtrans.h"
+#include "encodeintra.h"
+
+
+int vp8_encode_intra(VP8_COMP *cpi, MACROBLOCK *x, int use_dc_pred)
+{
+
+ int i;
+ int intra_pred_var = 0;
+ (void) cpi;
+
+ if (use_dc_pred)
+ {
+ x->e_mbd.mode_info_context->mbmi.mode = DC_PRED;
+ x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
+ x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
+
+ vp8_encode_intra16x16mby(x);
+
+ vp8_inverse_transform_mby(&x->e_mbd);
+ }
+ else
+ {
+ for (i = 0; i < 16; i++)
+ {
+ x->e_mbd.block[i].bmi.as_mode = B_DC_PRED;
+ vp8_encode_intra4x4block(x, i);
+ }
+ }
+
+ intra_pred_var = vp8_get_mb_ss(x->src_diff);
+
+ return intra_pred_var;
+}
+
+void vp8_encode_intra4x4block(MACROBLOCK *x, int ib)
+{
+ BLOCKD *b = &x->e_mbd.block[ib];
+ BLOCK *be = &x->block[ib];
+ int dst_stride = x->e_mbd.dst.y_stride;
+ unsigned char *dst = x->e_mbd.dst.y_buffer + b->offset;
+ unsigned char *Above = dst - dst_stride;
+ unsigned char *yleft = dst - 1;
+ unsigned char top_left = Above[-1];
+
+ vp8_intra4x4_predict(Above, yleft, dst_stride, b->bmi.as_mode,
+ b->predictor, 16, top_left);
+
+ vp8_subtract_b(be, b, 16);
+
+ x->short_fdct4x4(be->src_diff, be->coeff, 32);
+
+ x->quantize_b(be, b);
+
+ if (*b->eob > 1)
+ {
+ vp8_short_idct4x4llm(b->dqcoeff, b->predictor, 16, dst, dst_stride);
+ }
+ else
+ {
+ vp8_dc_only_idct_add(b->dqcoeff[0], b->predictor, 16, dst, dst_stride);
+ }
+}
+
+void vp8_encode_intra4x4mby(MACROBLOCK *mb)
+{
+ int i;
+
+ MACROBLOCKD *xd = &mb->e_mbd;
+ intra_prediction_down_copy(xd, xd->dst.y_buffer - xd->dst.y_stride + 16);
+
+ for (i = 0; i < 16; i++)
+ vp8_encode_intra4x4block(mb, i);
+ return;
+}
+
+void vp8_encode_intra16x16mby(MACROBLOCK *x)
+{
+ BLOCK *b = &x->block[0];
+ MACROBLOCKD *xd = &x->e_mbd;
+
+ vp8_build_intra_predictors_mby_s(xd,
+ xd->dst.y_buffer - xd->dst.y_stride,
+ xd->dst.y_buffer - 1,
+ xd->dst.y_stride,
+ xd->dst.y_buffer,
+ xd->dst.y_stride);
+
+ vp8_subtract_mby(x->src_diff, *(b->base_src),
+ b->src_stride, xd->dst.y_buffer, xd->dst.y_stride);
+
+ vp8_transform_intra_mby(x);
+
+ vp8_quantize_mby(x);
+
+ if (x->optimize)
+ vp8_optimize_mby(x);
+}
+
+void vp8_encode_intra16x16mbuv(MACROBLOCK *x)
+{
+ MACROBLOCKD *xd = &x->e_mbd;
+
+ vp8_build_intra_predictors_mbuv_s(xd, xd->dst.u_buffer - xd->dst.uv_stride,
+ xd->dst.v_buffer - xd->dst.uv_stride,
+ xd->dst.u_buffer - 1,
+ xd->dst.v_buffer - 1,
+ xd->dst.uv_stride,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride);
+
+ vp8_subtract_mbuv(x->src_diff, x->src.u_buffer,
+ x->src.v_buffer, x->src.uv_stride, xd->dst.u_buffer,
+ xd->dst.v_buffer, xd->dst.uv_stride);
+
+ vp8_transform_mbuv(x);
+
+ vp8_quantize_mbuv(x);
+
+ if (x->optimize)
+ vp8_optimize_mbuv(x);
+}
diff --git a/libvpx/vp8/encoder/encodeintra.h b/libvpx/vp8/encoder/encodeintra.h
new file mode 100644
index 0000000..be2141f
--- /dev/null
+++ b/libvpx/vp8/encoder/encodeintra.h
@@ -0,0 +1,21 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef _ENCODEINTRA_H_
+#define _ENCODEINTRA_H_
+#include "onyx_int.h"
+
+int vp8_encode_intra(VP8_COMP *cpi, MACROBLOCK *x, int use_dc_pred);
+void vp8_encode_intra16x16mby(MACROBLOCK *x);
+void vp8_encode_intra16x16mbuv(MACROBLOCK *x);
+void vp8_encode_intra4x4mby(MACROBLOCK *mb);
+void vp8_encode_intra4x4block(MACROBLOCK *x, int ib);
+#endif
diff --git a/libvpx/vp8/encoder/encodemb.c b/libvpx/vp8/encoder/encodemb.c
new file mode 100644
index 0000000..7d494f2
--- /dev/null
+++ b/libvpx/vp8/encoder/encodemb.c
@@ -0,0 +1,648 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "encodemb.h"
+#include "vp8/common/reconinter.h"
+#include "quantize.h"
+#include "tokenize.h"
+#include "vp8/common/invtrans.h"
+#include "vpx_mem/vpx_mem.h"
+#include "rdopt.h"
+
+void vp8_subtract_b_c(BLOCK *be, BLOCKD *bd, int pitch)
+{
+ unsigned char *src_ptr = (*(be->base_src) + be->src);
+ short *diff_ptr = be->src_diff;
+ unsigned char *pred_ptr = bd->predictor;
+ int src_stride = be->src_stride;
+
+ int r, c;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ diff_ptr[c] = src_ptr[c] - pred_ptr[c];
+ }
+
+ diff_ptr += pitch;
+ pred_ptr += pitch;
+ src_ptr += src_stride;
+ }
+}
+
+void vp8_subtract_mbuv_c(short *diff, unsigned char *usrc, unsigned char *vsrc,
+ int src_stride, unsigned char *upred,
+ unsigned char *vpred, int pred_stride)
+{
+ short *udiff = diff + 256;
+ short *vdiff = diff + 320;
+
+ int r, c;
+
+ for (r = 0; r < 8; r++)
+ {
+ for (c = 0; c < 8; c++)
+ {
+ udiff[c] = usrc[c] - upred[c];
+ }
+
+ udiff += 8;
+ upred += pred_stride;
+ usrc += src_stride;
+ }
+
+ for (r = 0; r < 8; r++)
+ {
+ for (c = 0; c < 8; c++)
+ {
+ vdiff[c] = vsrc[c] - vpred[c];
+ }
+
+ vdiff += 8;
+ vpred += pred_stride;
+ vsrc += src_stride;
+ }
+}
+
+void vp8_subtract_mby_c(short *diff, unsigned char *src, int src_stride,
+ unsigned char *pred, int pred_stride)
+{
+ int r, c;
+
+ for (r = 0; r < 16; r++)
+ {
+ for (c = 0; c < 16; c++)
+ {
+ diff[c] = src[c] - pred[c];
+ }
+
+ diff += 16;
+ pred += pred_stride;
+ src += src_stride;
+ }
+}
+
+static void vp8_subtract_mb(MACROBLOCK *x)
+{
+ BLOCK *b = &x->block[0];
+
+ vp8_subtract_mby(x->src_diff, *(b->base_src),
+ b->src_stride, x->e_mbd.dst.y_buffer, x->e_mbd.dst.y_stride);
+ vp8_subtract_mbuv(x->src_diff, x->src.u_buffer,
+ x->src.v_buffer, x->src.uv_stride, x->e_mbd.dst.u_buffer,
+ x->e_mbd.dst.v_buffer, x->e_mbd.dst.uv_stride);
+}
+
+static void build_dcblock(MACROBLOCK *x)
+{
+ short *src_diff_ptr = &x->src_diff[384];
+ int i;
+
+ for (i = 0; i < 16; i++)
+ {
+ src_diff_ptr[i] = x->coeff[i * 16];
+ }
+}
+
+void vp8_transform_mbuv(MACROBLOCK *x)
+{
+ int i;
+
+ for (i = 16; i < 24; i += 2)
+ {
+ x->short_fdct8x4(&x->block[i].src_diff[0],
+ &x->block[i].coeff[0], 16);
+ }
+}
+
+
+void vp8_transform_intra_mby(MACROBLOCK *x)
+{
+ int i;
+
+ for (i = 0; i < 16; i += 2)
+ {
+ x->short_fdct8x4(&x->block[i].src_diff[0],
+ &x->block[i].coeff[0], 32);
+ }
+
+ /* build dc block from 16 y dc values */
+ build_dcblock(x);
+
+ /* do 2nd order transform on the dc block */
+ x->short_walsh4x4(&x->block[24].src_diff[0],
+ &x->block[24].coeff[0], 8);
+
+}
+
+
+static void transform_mb(MACROBLOCK *x)
+{
+ int i;
+
+ for (i = 0; i < 16; i += 2)
+ {
+ x->short_fdct8x4(&x->block[i].src_diff[0],
+ &x->block[i].coeff[0], 32);
+ }
+
+ /* build dc block from 16 y dc values */
+ if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV)
+ build_dcblock(x);
+
+ for (i = 16; i < 24; i += 2)
+ {
+ x->short_fdct8x4(&x->block[i].src_diff[0],
+ &x->block[i].coeff[0], 16);
+ }
+
+ /* do 2nd order transform on the dc block */
+ if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV)
+ x->short_walsh4x4(&x->block[24].src_diff[0],
+ &x->block[24].coeff[0], 8);
+
+}
+
+
+static void transform_mby(MACROBLOCK *x)
+{
+ int i;
+
+ for (i = 0; i < 16; i += 2)
+ {
+ x->short_fdct8x4(&x->block[i].src_diff[0],
+ &x->block[i].coeff[0], 32);
+ }
+
+ /* build dc block from 16 y dc values */
+ if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV)
+ {
+ build_dcblock(x);
+ x->short_walsh4x4(&x->block[24].src_diff[0],
+ &x->block[24].coeff[0], 8);
+ }
+}
+
+
+
+#define RDTRUNC(RM,DM,R,D) ( (128+(R)*(RM)) & 0xFF )
+
+typedef struct vp8_token_state vp8_token_state;
+
+struct vp8_token_state{
+ int rate;
+ int error;
+ signed char next;
+ signed char token;
+ short qc;
+};
+
+/* TODO: experiments to find optimal multiple numbers */
+#define Y1_RD_MULT 4
+#define UV_RD_MULT 2
+#define Y2_RD_MULT 16
+
+static const int plane_rd_mult[4]=
+{
+ Y1_RD_MULT,
+ Y2_RD_MULT,
+ UV_RD_MULT,
+ Y1_RD_MULT
+};
+
+static void optimize_b(MACROBLOCK *mb, int ib, int type,
+ ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l)
+{
+ BLOCK *b;
+ BLOCKD *d;
+ vp8_token_state tokens[17][2];
+ unsigned best_mask[2];
+ const short *dequant_ptr;
+ const short *coeff_ptr;
+ short *qcoeff_ptr;
+ short *dqcoeff_ptr;
+ int eob;
+ int i0;
+ int rc;
+ int x;
+ int sz = 0;
+ int next;
+ int rdmult;
+ int rddiv;
+ int final_eob;
+ int rd_cost0;
+ int rd_cost1;
+ int rate0;
+ int rate1;
+ int error0;
+ int error1;
+ int t0;
+ int t1;
+ int best;
+ int band;
+ int pt;
+ int i;
+ int err_mult = plane_rd_mult[type];
+
+ b = &mb->block[ib];
+ d = &mb->e_mbd.block[ib];
+
+ /* Enable this to test the effect of RDO as a replacement for the dynamic
+ * zero bin instead of an augmentation of it.
+ */
+#if 0
+ vp8_strict_quantize_b(b, d);
+#endif
+
+ dequant_ptr = d->dequant;
+ coeff_ptr = b->coeff;
+ qcoeff_ptr = d->qcoeff;
+ dqcoeff_ptr = d->dqcoeff;
+ i0 = !type;
+ eob = *d->eob;
+
+ /* Now set up a Viterbi trellis to evaluate alternative roundings. */
+ rdmult = mb->rdmult * err_mult;
+ if(mb->e_mbd.mode_info_context->mbmi.ref_frame==INTRA_FRAME)
+ rdmult = (rdmult * 9)>>4;
+
+ rddiv = mb->rddiv;
+ best_mask[0] = best_mask[1] = 0;
+ /* Initialize the sentinel node of the trellis. */
+ tokens[eob][0].rate = 0;
+ tokens[eob][0].error = 0;
+ tokens[eob][0].next = 16;
+ tokens[eob][0].token = DCT_EOB_TOKEN;
+ tokens[eob][0].qc = 0;
+ *(tokens[eob] + 1) = *(tokens[eob] + 0);
+ next = eob;
+ for (i = eob; i-- > i0;)
+ {
+ int base_bits;
+ int d2;
+ int dx;
+
+ rc = vp8_default_zig_zag1d[i];
+ x = qcoeff_ptr[rc];
+ /* Only add a trellis state for non-zero coefficients. */
+ if (x)
+ {
+ int shortcut=0;
+ error0 = tokens[next][0].error;
+ error1 = tokens[next][1].error;
+ /* Evaluate the first possibility for this state. */
+ rate0 = tokens[next][0].rate;
+ rate1 = tokens[next][1].rate;
+ t0 = (vp8_dct_value_tokens_ptr + x)->Token;
+ /* Consider both possible successor states. */
+ if (next < 16)
+ {
+ band = vp8_coef_bands[i + 1];
+ pt = vp8_prev_token_class[t0];
+ rate0 +=
+ mb->token_costs[type][band][pt][tokens[next][0].token];
+ rate1 +=
+ mb->token_costs[type][band][pt][tokens[next][1].token];
+ }
+ rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0);
+ rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1);
+ if (rd_cost0 == rd_cost1)
+ {
+ rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0);
+ rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1);
+ }
+ /* And pick the best. */
+ best = rd_cost1 < rd_cost0;
+ base_bits = *(vp8_dct_value_cost_ptr + x);
+ dx = dqcoeff_ptr[rc] - coeff_ptr[rc];
+ d2 = dx*dx;
+ tokens[i][0].rate = base_bits + (best ? rate1 : rate0);
+ tokens[i][0].error = d2 + (best ? error1 : error0);
+ tokens[i][0].next = next;
+ tokens[i][0].token = t0;
+ tokens[i][0].qc = x;
+ best_mask[0] |= best << i;
+ /* Evaluate the second possibility for this state. */
+ rate0 = tokens[next][0].rate;
+ rate1 = tokens[next][1].rate;
+
+ if((abs(x)*dequant_ptr[rc]>abs(coeff_ptr[rc])) &&
+ (abs(x)*dequant_ptr[rc]<abs(coeff_ptr[rc])+dequant_ptr[rc]))
+ shortcut = 1;
+ else
+ shortcut = 0;
+
+ if(shortcut)
+ {
+ sz = -(x < 0);
+ x -= 2*sz + 1;
+ }
+
+ /* Consider both possible successor states. */
+ if (!x)
+ {
+ /* If we reduced this coefficient to zero, check to see if
+ * we need to move the EOB back here.
+ */
+ t0 = tokens[next][0].token == DCT_EOB_TOKEN ?
+ DCT_EOB_TOKEN : ZERO_TOKEN;
+ t1 = tokens[next][1].token == DCT_EOB_TOKEN ?
+ DCT_EOB_TOKEN : ZERO_TOKEN;
+ }
+ else
+ {
+ t0=t1 = (vp8_dct_value_tokens_ptr + x)->Token;
+ }
+ if (next < 16)
+ {
+ band = vp8_coef_bands[i + 1];
+ if(t0!=DCT_EOB_TOKEN)
+ {
+ pt = vp8_prev_token_class[t0];
+ rate0 += mb->token_costs[type][band][pt][
+ tokens[next][0].token];
+ }
+ if(t1!=DCT_EOB_TOKEN)
+ {
+ pt = vp8_prev_token_class[t1];
+ rate1 += mb->token_costs[type][band][pt][
+ tokens[next][1].token];
+ }
+ }
+
+ rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0);
+ rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1);
+ if (rd_cost0 == rd_cost1)
+ {
+ rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0);
+ rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1);
+ }
+ /* And pick the best. */
+ best = rd_cost1 < rd_cost0;
+ base_bits = *(vp8_dct_value_cost_ptr + x);
+
+ if(shortcut)
+ {
+ dx -= (dequant_ptr[rc] + sz) ^ sz;
+ d2 = dx*dx;
+ }
+ tokens[i][1].rate = base_bits + (best ? rate1 : rate0);
+ tokens[i][1].error = d2 + (best ? error1 : error0);
+ tokens[i][1].next = next;
+ tokens[i][1].token =best?t1:t0;
+ tokens[i][1].qc = x;
+ best_mask[1] |= best << i;
+ /* Finally, make this the new head of the trellis. */
+ next = i;
+ }
+ /* There's no choice to make for a zero coefficient, so we don't
+ * add a new trellis node, but we do need to update the costs.
+ */
+ else
+ {
+ band = vp8_coef_bands[i + 1];
+ t0 = tokens[next][0].token;
+ t1 = tokens[next][1].token;
+ /* Update the cost of each path if we're past the EOB token. */
+ if (t0 != DCT_EOB_TOKEN)
+ {
+ tokens[next][0].rate += mb->token_costs[type][band][0][t0];
+ tokens[next][0].token = ZERO_TOKEN;
+ }
+ if (t1 != DCT_EOB_TOKEN)
+ {
+ tokens[next][1].rate += mb->token_costs[type][band][0][t1];
+ tokens[next][1].token = ZERO_TOKEN;
+ }
+ /* Don't update next, because we didn't add a new node. */
+ }
+ }
+
+ /* Now pick the best path through the whole trellis. */
+ band = vp8_coef_bands[i + 1];
+ VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
+ rate0 = tokens[next][0].rate;
+ rate1 = tokens[next][1].rate;
+ error0 = tokens[next][0].error;
+ error1 = tokens[next][1].error;
+ t0 = tokens[next][0].token;
+ t1 = tokens[next][1].token;
+ rate0 += mb->token_costs[type][band][pt][t0];
+ rate1 += mb->token_costs[type][band][pt][t1];
+ rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0);
+ rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1);
+ if (rd_cost0 == rd_cost1)
+ {
+ rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0);
+ rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1);
+ }
+ best = rd_cost1 < rd_cost0;
+ final_eob = i0 - 1;
+ for (i = next; i < eob; i = next)
+ {
+ x = tokens[i][best].qc;
+ if (x)
+ final_eob = i;
+ rc = vp8_default_zig_zag1d[i];
+ qcoeff_ptr[rc] = x;
+ dqcoeff_ptr[rc] = x * dequant_ptr[rc];
+ next = tokens[i][best].next;
+ best = (best_mask[best] >> i) & 1;
+ }
+ final_eob++;
+
+ *a = *l = (final_eob != !type);
+ *d->eob = (char)final_eob;
+}
+static void check_reset_2nd_coeffs(MACROBLOCKD *x, int type,
+ ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l)
+{
+ int sum=0;
+ int i;
+ BLOCKD *bd = &x->block[24];
+
+ if(bd->dequant[0]>=35 && bd->dequant[1]>=35)
+ return;
+
+ for(i=0;i<(*bd->eob);i++)
+ {
+ int coef = bd->dqcoeff[vp8_default_zig_zag1d[i]];
+ sum+= (coef>=0)?coef:-coef;
+ if(sum>=35)
+ return;
+ }
+ /**************************************************************************
+ our inverse hadamard transform effectively is weighted sum of all 16 inputs
+ with weight either 1 or -1. It has a last stage scaling of (sum+3)>>3. And
+ dc only idct is (dc+4)>>3. So if all the sums are between -35 and 29, the
+ output after inverse wht and idct will be all zero. A sum of absolute value
+ smaller than 35 guarantees all 16 different (+1/-1) weighted sums in wht
+ fall between -35 and +35.
+ **************************************************************************/
+ if(sum < 35)
+ {
+ for(i=0;i<(*bd->eob);i++)
+ {
+ int rc = vp8_default_zig_zag1d[i];
+ bd->qcoeff[rc]=0;
+ bd->dqcoeff[rc]=0;
+ }
+ *bd->eob = 0;
+ *a = *l = (*bd->eob != !type);
+ }
+}
+
+static void optimize_mb(MACROBLOCK *x)
+{
+ int b;
+ int type;
+ int has_2nd_order;
+
+ ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT *ta;
+ ENTROPY_CONTEXT *tl;
+
+ vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta = (ENTROPY_CONTEXT *)&t_above;
+ tl = (ENTROPY_CONTEXT *)&t_left;
+
+ has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED
+ && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
+ type = has_2nd_order ? PLANE_TYPE_Y_NO_DC : PLANE_TYPE_Y_WITH_DC;
+
+ for (b = 0; b < 16; b++)
+ {
+ optimize_b(x, b, type,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+ }
+
+ for (b = 16; b < 24; b++)
+ {
+ optimize_b(x, b, PLANE_TYPE_UV,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+ }
+
+ if (has_2nd_order)
+ {
+ b=24;
+ optimize_b(x, b, PLANE_TYPE_Y2,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+ check_reset_2nd_coeffs(&x->e_mbd, PLANE_TYPE_Y2,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+ }
+}
+
+
+void vp8_optimize_mby(MACROBLOCK *x)
+{
+ int b;
+ int type;
+ int has_2nd_order;
+
+ ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT *ta;
+ ENTROPY_CONTEXT *tl;
+
+ if (!x->e_mbd.above_context)
+ return;
+
+ if (!x->e_mbd.left_context)
+ return;
+
+ vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta = (ENTROPY_CONTEXT *)&t_above;
+ tl = (ENTROPY_CONTEXT *)&t_left;
+
+ has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED
+ && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
+ type = has_2nd_order ? PLANE_TYPE_Y_NO_DC : PLANE_TYPE_Y_WITH_DC;
+
+ for (b = 0; b < 16; b++)
+ {
+ optimize_b(x, b, type,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+ }
+
+
+ if (has_2nd_order)
+ {
+ b=24;
+ optimize_b(x, b, PLANE_TYPE_Y2,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+ check_reset_2nd_coeffs(&x->e_mbd, PLANE_TYPE_Y2,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+ }
+}
+
+void vp8_optimize_mbuv(MACROBLOCK *x)
+{
+ int b;
+ ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT *ta;
+ ENTROPY_CONTEXT *tl;
+
+ if (!x->e_mbd.above_context)
+ return;
+
+ if (!x->e_mbd.left_context)
+ return;
+
+ vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta = (ENTROPY_CONTEXT *)&t_above;
+ tl = (ENTROPY_CONTEXT *)&t_left;
+
+ for (b = 16; b < 24; b++)
+ {
+ optimize_b(x, b, PLANE_TYPE_UV,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+ }
+}
+
+void vp8_encode_inter16x16(MACROBLOCK *x)
+{
+ vp8_build_inter_predictors_mb(&x->e_mbd);
+
+ vp8_subtract_mb(x);
+
+ transform_mb(x);
+
+ vp8_quantize_mb(x);
+
+ if (x->optimize)
+ optimize_mb(x);
+}
+
+/* this funciton is used by first pass only */
+void vp8_encode_inter16x16y(MACROBLOCK *x)
+{
+ BLOCK *b = &x->block[0];
+
+ vp8_build_inter16x16_predictors_mby(&x->e_mbd, x->e_mbd.dst.y_buffer,
+ x->e_mbd.dst.y_stride);
+
+ vp8_subtract_mby(x->src_diff, *(b->base_src),
+ b->src_stride, x->e_mbd.dst.y_buffer, x->e_mbd.dst.y_stride);
+
+ transform_mby(x);
+
+ vp8_quantize_mby(x);
+
+ vp8_inverse_transform_mby(&x->e_mbd);
+}
diff --git a/libvpx/vp8/encoder/encodemb.h b/libvpx/vp8/encoder/encodemb.h
new file mode 100644
index 0000000..6badf7d
--- /dev/null
+++ b/libvpx/vp8/encoder/encodemb.h
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_ENCODEMB_H
+#define __INC_ENCODEMB_H
+
+#include "onyx_int.h"
+void vp8_encode_inter16x16(MACROBLOCK *x);
+
+void vp8_build_dcblock(MACROBLOCK *b);
+void vp8_transform_mb(MACROBLOCK *mb);
+void vp8_transform_mbuv(MACROBLOCK *x);
+void vp8_transform_intra_mby(MACROBLOCK *x);
+
+void vp8_optimize_mby(MACROBLOCK *x);
+void vp8_optimize_mbuv(MACROBLOCK *x);
+void vp8_encode_inter16x16y(MACROBLOCK *x);
+#endif
diff --git a/libvpx/vp8/encoder/encodemv.c b/libvpx/vp8/encoder/encodemv.c
new file mode 100644
index 0000000..0c43d06
--- /dev/null
+++ b/libvpx/vp8/encoder/encodemv.c
@@ -0,0 +1,380 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/common/common.h"
+#include "encodemv.h"
+#include "vp8/common/entropymode.h"
+#include "vp8/common/systemdependent.h"
+
+#include <math.h>
+
+#ifdef ENTROPY_STATS
+extern unsigned int active_section;
+#endif
+
+static void encode_mvcomponent(
+ vp8_writer *const w,
+ const int v,
+ const struct mv_context *mvc
+)
+{
+ const vp8_prob *p = mvc->prob;
+ const int x = v < 0 ? -v : v;
+
+ if (x < mvnum_short) /* Small */
+ {
+ vp8_write(w, 0, p [mvpis_short]);
+ vp8_treed_write(w, vp8_small_mvtree, p + MVPshort, x, 3);
+
+ if (!x)
+ return; /* no sign bit */
+ }
+ else /* Large */
+ {
+ int i = 0;
+
+ vp8_write(w, 1, p [mvpis_short]);
+
+ do
+ vp8_write(w, (x >> i) & 1, p [MVPbits + i]);
+
+ while (++i < 3);
+
+ i = mvlong_width - 1; /* Skip bit 3, which is sometimes implicit */
+
+ do
+ vp8_write(w, (x >> i) & 1, p [MVPbits + i]);
+
+ while (--i > 3);
+
+ if (x & 0xFFF0)
+ vp8_write(w, (x >> 3) & 1, p [MVPbits + 3]);
+ }
+
+ vp8_write(w, v < 0, p [MVPsign]);
+}
+#if 0
+static int max_mv_r = 0;
+static int max_mv_c = 0;
+#endif
+void vp8_encode_motion_vector(vp8_writer *w, const MV *mv, const MV_CONTEXT *mvc)
+{
+
+#if 0
+ {
+ if (abs(mv->row >> 1) > max_mv_r)
+ {
+ FILE *f = fopen("maxmv.stt", "a");
+ max_mv_r = abs(mv->row >> 1);
+ fprintf(f, "New Mv Row Max %6d\n", (mv->row >> 1));
+
+ if ((abs(mv->row) / 2) != max_mv_r)
+ fprintf(f, "MV Row conversion error %6d\n", abs(mv->row) / 2);
+
+ fclose(f);
+ }
+
+ if (abs(mv->col >> 1) > max_mv_c)
+ {
+ FILE *f = fopen("maxmv.stt", "a");
+ fprintf(f, "New Mv Col Max %6d\n", (mv->col >> 1));
+ max_mv_c = abs(mv->col >> 1);
+ fclose(f);
+ }
+ }
+#endif
+
+ encode_mvcomponent(w, mv->row >> 1, &mvc[0]);
+ encode_mvcomponent(w, mv->col >> 1, &mvc[1]);
+}
+
+
+static unsigned int cost_mvcomponent(const int v, const struct mv_context *mvc)
+{
+ const vp8_prob *p = mvc->prob;
+ const int x = v;
+ unsigned int cost;
+
+ if (x < mvnum_short)
+ {
+ cost = vp8_cost_zero(p [mvpis_short])
+ + vp8_treed_cost(vp8_small_mvtree, p + MVPshort, x, 3);
+
+ if (!x)
+ return cost;
+ }
+ else
+ {
+ int i = 0;
+ cost = vp8_cost_one(p [mvpis_short]);
+
+ do
+ cost += vp8_cost_bit(p [MVPbits + i], (x >> i) & 1);
+
+ while (++i < 3);
+
+ i = mvlong_width - 1; /* Skip bit 3, which is sometimes implicit */
+
+ do
+ cost += vp8_cost_bit(p [MVPbits + i], (x >> i) & 1);
+
+ while (--i > 3);
+
+ if (x & 0xFFF0)
+ cost += vp8_cost_bit(p [MVPbits + 3], (x >> 3) & 1);
+ }
+
+ return cost; /* + vp8_cost_bit( p [MVPsign], v < 0); */
+}
+
+void vp8_build_component_cost_table(int *mvcost[2], const MV_CONTEXT *mvc, int mvc_flag[2])
+{
+ int i = 1;
+ unsigned int cost0 = 0;
+ unsigned int cost1 = 0;
+
+ vp8_clear_system_state();
+
+ i = 1;
+
+ if (mvc_flag[0])
+ {
+ mvcost [0] [0] = cost_mvcomponent(0, &mvc[0]);
+
+ do
+ {
+ cost0 = cost_mvcomponent(i, &mvc[0]);
+
+ mvcost [0] [i] = cost0 + vp8_cost_zero(mvc[0].prob[MVPsign]);
+ mvcost [0] [-i] = cost0 + vp8_cost_one(mvc[0].prob[MVPsign]);
+ }
+ while (++i <= mv_max);
+ }
+
+ i = 1;
+
+ if (mvc_flag[1])
+ {
+ mvcost [1] [0] = cost_mvcomponent(0, &mvc[1]);
+
+ do
+ {
+ cost1 = cost_mvcomponent(i, &mvc[1]);
+
+ mvcost [1] [i] = cost1 + vp8_cost_zero(mvc[1].prob[MVPsign]);
+ mvcost [1] [-i] = cost1 + vp8_cost_one(mvc[1].prob[MVPsign]);
+ }
+ while (++i <= mv_max);
+ }
+}
+
+
+/* Motion vector probability table update depends on benefit.
+ * Small correction allows for the fact that an update to an MV probability
+ * may have benefit in subsequent frames as well as the current one.
+ */
+#define MV_PROB_UPDATE_CORRECTION -1
+
+
+static void calc_prob(vp8_prob *p, const unsigned int ct[2])
+{
+ const unsigned int tot = ct[0] + ct[1];
+
+ if (tot)
+ {
+ const vp8_prob x = ((ct[0] * 255) / tot) & -2;
+ *p = x ? x : 1;
+ }
+}
+
+static void update(
+ vp8_writer *const w,
+ const unsigned int ct[2],
+ vp8_prob *const cur_p,
+ const vp8_prob new_p,
+ const vp8_prob update_p,
+ int *updated
+)
+{
+ const int cur_b = vp8_cost_branch(ct, *cur_p);
+ const int new_b = vp8_cost_branch(ct, new_p);
+ const int cost = 7 + MV_PROB_UPDATE_CORRECTION + ((vp8_cost_one(update_p) - vp8_cost_zero(update_p) + 128) >> 8);
+
+ if (cur_b - new_b > cost)
+ {
+ *cur_p = new_p;
+ vp8_write(w, 1, update_p);
+ vp8_write_literal(w, new_p >> 1, 7);
+ *updated = 1;
+
+ }
+ else
+ vp8_write(w, 0, update_p);
+}
+
+static void write_component_probs(
+ vp8_writer *const w,
+ struct mv_context *cur_mvc,
+ const struct mv_context *default_mvc_,
+ const struct mv_context *update_mvc,
+ const unsigned int events [MVvals],
+ unsigned int rc,
+ int *updated
+)
+{
+ vp8_prob *Pcur = cur_mvc->prob;
+ const vp8_prob *default_mvc = default_mvc_->prob;
+ const vp8_prob *Pupdate = update_mvc->prob;
+ unsigned int is_short_ct[2], sign_ct[2];
+
+ unsigned int bit_ct [mvlong_width] [2];
+
+ unsigned int short_ct [mvnum_short];
+ unsigned int short_bct [mvnum_short-1] [2];
+
+ vp8_prob Pnew [MVPcount];
+
+ (void) rc;
+ vp8_copy_array(Pnew, default_mvc, MVPcount);
+
+ vp8_zero(is_short_ct)
+ vp8_zero(sign_ct)
+ vp8_zero(bit_ct)
+ vp8_zero(short_ct)
+ vp8_zero(short_bct)
+
+
+ /* j=0 */
+ {
+ const int c = events [mv_max];
+
+ is_short_ct [0] += c; /* Short vector */
+ short_ct [0] += c; /* Magnitude distribution */
+ }
+
+ /* j: 1 ~ mv_max (1023) */
+ {
+ int j = 1;
+
+ do
+ {
+ const int c1 = events [mv_max + j]; /* positive */
+ const int c2 = events [mv_max - j]; /* negative */
+ const int c = c1 + c2;
+ int a = j;
+
+ sign_ct [0] += c1;
+ sign_ct [1] += c2;
+
+ if (a < mvnum_short)
+ {
+ is_short_ct [0] += c; /* Short vector */
+ short_ct [a] += c; /* Magnitude distribution */
+ }
+ else
+ {
+ int k = mvlong_width - 1;
+ is_short_ct [1] += c; /* Long vector */
+
+ /* bit 3 not always encoded. */
+ do
+ bit_ct [k] [(a >> k) & 1] += c;
+
+ while (--k >= 0);
+ }
+ }
+ while (++j <= mv_max);
+ }
+
+ calc_prob(Pnew + mvpis_short, is_short_ct);
+
+ calc_prob(Pnew + MVPsign, sign_ct);
+
+ {
+ vp8_prob p [mvnum_short - 1]; /* actually only need branch ct */
+ int j = 0;
+
+ vp8_tree_probs_from_distribution(
+ 8, vp8_small_mvencodings, vp8_small_mvtree,
+ p, short_bct, short_ct,
+ 256, 1
+ );
+
+ do
+ calc_prob(Pnew + MVPshort + j, short_bct[j]);
+
+ while (++j < mvnum_short - 1);
+ }
+
+ {
+ int j = 0;
+
+ do
+ calc_prob(Pnew + MVPbits + j, bit_ct[j]);
+
+ while (++j < mvlong_width);
+ }
+
+ update(w, is_short_ct, Pcur + mvpis_short, Pnew[mvpis_short], *Pupdate++, updated);
+
+ update(w, sign_ct, Pcur + MVPsign, Pnew[MVPsign], *Pupdate++, updated);
+
+ {
+ const vp8_prob *const new_p = Pnew + MVPshort;
+ vp8_prob *const cur_p = Pcur + MVPshort;
+
+ int j = 0;
+
+ do
+
+ update(w, short_bct[j], cur_p + j, new_p[j], *Pupdate++, updated);
+
+ while (++j < mvnum_short - 1);
+ }
+
+ {
+ const vp8_prob *const new_p = Pnew + MVPbits;
+ vp8_prob *const cur_p = Pcur + MVPbits;
+
+ int j = 0;
+
+ do
+
+ update(w, bit_ct[j], cur_p + j, new_p[j], *Pupdate++, updated);
+
+ while (++j < mvlong_width);
+ }
+}
+
+void vp8_write_mvprobs(VP8_COMP *cpi)
+{
+ vp8_writer *const w = cpi->bc;
+ MV_CONTEXT *mvc = cpi->common.fc.mvc;
+ int flags[2] = {0, 0};
+#ifdef ENTROPY_STATS
+ active_section = 4;
+#endif
+ write_component_probs(
+ w, &mvc[0], &vp8_default_mv_context[0], &vp8_mv_update_probs[0],
+ cpi->mb.MVcount[0], 0, &flags[0]
+ );
+ write_component_probs(
+ w, &mvc[1], &vp8_default_mv_context[1], &vp8_mv_update_probs[1],
+ cpi->mb.MVcount[1], 1, &flags[1]
+ );
+
+ if (flags[0] || flags[1])
+ vp8_build_component_cost_table(cpi->mb.mvcost, (const MV_CONTEXT *) cpi->common.fc.mvc, flags);
+
+#ifdef ENTROPY_STATS
+ active_section = 5;
+#endif
+}
diff --git a/libvpx/vp8/encoder/encodemv.h b/libvpx/vp8/encoder/encodemv.h
new file mode 100644
index 0000000..a6116c1
--- /dev/null
+++ b/libvpx/vp8/encoder/encodemv.h
@@ -0,0 +1,21 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_ENCODEMV_H
+#define __INC_ENCODEMV_H
+
+#include "onyx_int.h"
+
+void vp8_write_mvprobs(VP8_COMP *);
+void vp8_encode_motion_vector(vp8_writer *, const MV *, const MV_CONTEXT *);
+void vp8_build_component_cost_table(int *mvcost[2], const MV_CONTEXT *mvc, int mvc_flag[2]);
+
+#endif
diff --git a/libvpx/vp8/encoder/ethreading.c b/libvpx/vp8/encoder/ethreading.c
new file mode 100644
index 0000000..39340f2
--- /dev/null
+++ b/libvpx/vp8/encoder/ethreading.c
@@ -0,0 +1,642 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "onyx_int.h"
+#include "vp8/common/threading.h"
+#include "vp8/common/common.h"
+#include "vp8/common/extend.h"
+#include "bitstream.h"
+#include "encodeframe.h"
+
+#if CONFIG_MULTITHREAD
+
+extern void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x, int ok_to_skip);
+
+extern void vp8_loopfilter_frame(VP8_COMP *cpi, VP8_COMMON *cm);
+
+static THREAD_FUNCTION thread_loopfilter(void *p_data)
+{
+ VP8_COMP *cpi = (VP8_COMP *)(((LPFTHREAD_DATA *)p_data)->ptr1);
+ VP8_COMMON *cm = &cpi->common;
+
+ while (1)
+ {
+ if (cpi->b_multi_threaded == 0)
+ break;
+
+ if (sem_wait(&cpi->h_event_start_lpf) == 0)
+ {
+ if (cpi->b_multi_threaded == 0) /* we're shutting down */
+ break;
+
+ vp8_loopfilter_frame(cpi, cm);
+
+ sem_post(&cpi->h_event_end_lpf);
+ }
+ }
+
+ return 0;
+}
+
+static
+THREAD_FUNCTION thread_encoding_proc(void *p_data)
+{
+ int ithread = ((ENCODETHREAD_DATA *)p_data)->ithread;
+ VP8_COMP *cpi = (VP8_COMP *)(((ENCODETHREAD_DATA *)p_data)->ptr1);
+ MB_ROW_COMP *mbri = (MB_ROW_COMP *)(((ENCODETHREAD_DATA *)p_data)->ptr2);
+ ENTROPY_CONTEXT_PLANES mb_row_left_context;
+
+ while (1)
+ {
+ if (cpi->b_multi_threaded == 0)
+ break;
+
+ if (sem_wait(&cpi->h_event_start_encoding[ithread]) == 0)
+ {
+ const int nsync = cpi->mt_sync_range;
+ VP8_COMMON *cm = &cpi->common;
+ int mb_row;
+ MACROBLOCK *x = &mbri->mb;
+ MACROBLOCKD *xd = &x->e_mbd;
+ TOKENEXTRA *tp ;
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ TOKENEXTRA *tp_start = cpi->tok + (1 + ithread) * (16 * 24);
+ const int num_part = (1 << cm->multi_token_partition);
+#endif
+
+ int *segment_counts = mbri->segment_counts;
+ int *totalrate = &mbri->totalrate;
+
+ if (cpi->b_multi_threaded == 0) /* we're shutting down */
+ break;
+
+ for (mb_row = ithread + 1; mb_row < cm->mb_rows; mb_row += (cpi->encoding_thread_count + 1))
+ {
+
+ int recon_yoffset, recon_uvoffset;
+ int mb_col;
+ int ref_fb_idx = cm->lst_fb_idx;
+ int dst_fb_idx = cm->new_fb_idx;
+ int recon_y_stride = cm->yv12_fb[ref_fb_idx].y_stride;
+ int recon_uv_stride = cm->yv12_fb[ref_fb_idx].uv_stride;
+ int map_index = (mb_row * cm->mb_cols);
+ volatile const int *last_row_current_mb_col;
+ volatile int *current_mb_col = &cpi->mt_current_mb_col[mb_row];
+
+#if (CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ vp8_writer *w = &cpi->bc[1 + (mb_row % num_part)];
+#else
+ tp = cpi->tok + (mb_row * (cm->mb_cols * 16 * 24));
+ cpi->tplist[mb_row].start = tp;
+#endif
+
+ last_row_current_mb_col = &cpi->mt_current_mb_col[mb_row - 1];
+
+ /* reset above block coeffs */
+ xd->above_context = cm->above_context;
+ xd->left_context = &mb_row_left_context;
+
+ vp8_zero(mb_row_left_context);
+
+ xd->up_available = (mb_row != 0);
+ recon_yoffset = (mb_row * recon_y_stride * 16);
+ recon_uvoffset = (mb_row * recon_uv_stride * 8);
+
+ /* Set the mb activity pointer to the start of the row. */
+ x->mb_activity_ptr = &cpi->mb_activity_map[map_index];
+
+ /* for each macroblock col in image */
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+ *current_mb_col = mb_col - 1;
+
+ if ((mb_col & (nsync - 1)) == 0)
+ {
+ while (mb_col > (*last_row_current_mb_col - nsync))
+ {
+ x86_pause_hint();
+ thread_sleep(0);
+ }
+ }
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ tp = tp_start;
+#endif
+
+ /* Distance of Mb to the various image edges.
+ * These specified to 8th pel as they are always compared
+ * to values that are in 1/8th pel units
+ */
+ xd->mb_to_left_edge = -((mb_col * 16) << 3);
+ xd->mb_to_right_edge = ((cm->mb_cols - 1 - mb_col) * 16) << 3;
+ xd->mb_to_top_edge = -((mb_row * 16) << 3);
+ xd->mb_to_bottom_edge = ((cm->mb_rows - 1 - mb_row) * 16) << 3;
+
+ /* Set up limit values for motion vectors used to prevent
+ * them extending outside the UMV borders
+ */
+ x->mv_col_min = -((mb_col * 16) + (VP8BORDERINPIXELS - 16));
+ x->mv_col_max = ((cm->mb_cols - 1 - mb_col) * 16) + (VP8BORDERINPIXELS - 16);
+ x->mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 16));
+ x->mv_row_max = ((cm->mb_rows - 1 - mb_row) * 16) + (VP8BORDERINPIXELS - 16);
+
+ xd->dst.y_buffer = cm->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset;
+ xd->dst.u_buffer = cm->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset;
+ xd->dst.v_buffer = cm->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset;
+ xd->left_available = (mb_col != 0);
+
+ x->rddiv = cpi->RDDIV;
+ x->rdmult = cpi->RDMULT;
+
+ /* Copy current mb to a buffer */
+ vp8_copy_mem16x16(x->src.y_buffer, x->src.y_stride, x->thismb, 16);
+
+ if (cpi->oxcf.tuning == VP8_TUNE_SSIM)
+ vp8_activity_masking(cpi, x);
+
+ /* Is segmentation enabled */
+ /* MB level adjustment to quantizer */
+ if (xd->segmentation_enabled)
+ {
+ /* Code to set segment id in xd->mbmi.segment_id for
+ * current MB (with range checking)
+ */
+ if (cpi->segmentation_map[map_index + mb_col] <= 3)
+ xd->mode_info_context->mbmi.segment_id = cpi->segmentation_map[map_index + mb_col];
+ else
+ xd->mode_info_context->mbmi.segment_id = 0;
+
+ vp8cx_mb_init_quantizer(cpi, x, 1);
+ }
+ else
+ /* Set to Segment 0 by default */
+ xd->mode_info_context->mbmi.segment_id = 0;
+
+ x->active_ptr = cpi->active_map + map_index + mb_col;
+
+ if (cm->frame_type == KEY_FRAME)
+ {
+ *totalrate += vp8cx_encode_intra_macroblock(cpi, x, &tp);
+#ifdef MODE_STATS
+ y_modes[xd->mbmi.mode] ++;
+#endif
+ }
+ else
+ {
+ *totalrate += vp8cx_encode_inter_macroblock(cpi, x, &tp, recon_yoffset, recon_uvoffset, mb_row, mb_col);
+
+#ifdef MODE_STATS
+ inter_y_modes[xd->mbmi.mode] ++;
+
+ if (xd->mbmi.mode == SPLITMV)
+ {
+ int b;
+
+ for (b = 0; b < xd->mbmi.partition_count; b++)
+ {
+ inter_b_modes[x->partition->bmi[b].mode] ++;
+ }
+ }
+
+#endif
+
+ /* Special case code for cyclic refresh
+ * If cyclic update enabled then copy
+ * xd->mbmi.segment_id; (which may have been updated
+ * based on mode during
+ * vp8cx_encode_inter_macroblock()) back into the
+ * global segmentation map
+ */
+ if (cpi->cyclic_refresh_mode_enabled && xd->segmentation_enabled)
+ {
+ const MB_MODE_INFO * mbmi = &xd->mode_info_context->mbmi;
+ cpi->segmentation_map[map_index + mb_col] = mbmi->segment_id;
+
+ /* If the block has been refreshed mark it as clean
+ * (the magnitude of the -ve influences how long it
+ * will be before we consider another refresh):
+ * Else if it was coded (last frame 0,0) and has
+ * not already been refreshed then mark it as a
+ * candidate for cleanup next time (marked 0) else
+ * mark it as dirty (1).
+ */
+ if (mbmi->segment_id)
+ cpi->cyclic_refresh_map[map_index + mb_col] = -1;
+ else if ((mbmi->mode == ZEROMV) && (mbmi->ref_frame == LAST_FRAME))
+ {
+ if (cpi->cyclic_refresh_map[map_index + mb_col] == 1)
+ cpi->cyclic_refresh_map[map_index + mb_col] = 0;
+ }
+ else
+ cpi->cyclic_refresh_map[map_index + mb_col] = 1;
+
+ }
+ }
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ /* pack tokens for this MB */
+ {
+ int tok_count = tp - tp_start;
+ pack_tokens(w, tp_start, tok_count);
+ }
+#else
+ cpi->tplist[mb_row].stop = tp;
+#endif
+ /* Increment pointer into gf usage flags structure. */
+ x->gf_active_ptr++;
+
+ /* Increment the activity mask pointers. */
+ x->mb_activity_ptr++;
+
+ /* adjust to the next column of macroblocks */
+ x->src.y_buffer += 16;
+ x->src.u_buffer += 8;
+ x->src.v_buffer += 8;
+
+ recon_yoffset += 16;
+ recon_uvoffset += 8;
+
+ /* Keep track of segment usage */
+ segment_counts[xd->mode_info_context->mbmi.segment_id]++;
+
+ /* skip to next mb */
+ xd->mode_info_context++;
+ x->partition_info++;
+ xd->above_context++;
+ }
+
+ vp8_extend_mb_row( &cm->yv12_fb[dst_fb_idx],
+ xd->dst.y_buffer + 16,
+ xd->dst.u_buffer + 8,
+ xd->dst.v_buffer + 8);
+
+ *current_mb_col = mb_col + nsync;
+
+ /* this is to account for the border */
+ xd->mode_info_context++;
+ x->partition_info++;
+
+ x->src.y_buffer += 16 * x->src.y_stride * (cpi->encoding_thread_count + 1) - 16 * cm->mb_cols;
+ x->src.u_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols;
+ x->src.v_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols;
+
+ xd->mode_info_context += xd->mode_info_stride * cpi->encoding_thread_count;
+ x->partition_info += xd->mode_info_stride * cpi->encoding_thread_count;
+ x->gf_active_ptr += cm->mb_cols * cpi->encoding_thread_count;
+
+ if (mb_row == cm->mb_rows - 1)
+ {
+ sem_post(&cpi->h_event_end_encoding); /* signal frame encoding end */
+ }
+ }
+ }
+ }
+
+ /* printf("exit thread %d\n", ithread); */
+ return 0;
+}
+
+static void setup_mbby_copy(MACROBLOCK *mbdst, MACROBLOCK *mbsrc)
+{
+
+ MACROBLOCK *x = mbsrc;
+ MACROBLOCK *z = mbdst;
+ int i;
+
+ z->ss = x->ss;
+ z->ss_count = x->ss_count;
+ z->searches_per_step = x->searches_per_step;
+ z->errorperbit = x->errorperbit;
+
+ z->sadperbit16 = x->sadperbit16;
+ z->sadperbit4 = x->sadperbit4;
+
+ /*
+ z->mv_col_min = x->mv_col_min;
+ z->mv_col_max = x->mv_col_max;
+ z->mv_row_min = x->mv_row_min;
+ z->mv_row_max = x->mv_row_max;
+ */
+
+ z->short_fdct4x4 = x->short_fdct4x4;
+ z->short_fdct8x4 = x->short_fdct8x4;
+ z->short_walsh4x4 = x->short_walsh4x4;
+ z->quantize_b = x->quantize_b;
+ z->quantize_b_pair = x->quantize_b_pair;
+ z->optimize = x->optimize;
+
+ /*
+ z->mvc = x->mvc;
+ z->src.y_buffer = x->src.y_buffer;
+ z->src.u_buffer = x->src.u_buffer;
+ z->src.v_buffer = x->src.v_buffer;
+ */
+
+ z->mvcost[0] = x->mvcost[0];
+ z->mvcost[1] = x->mvcost[1];
+ z->mvsadcost[0] = x->mvsadcost[0];
+ z->mvsadcost[1] = x->mvsadcost[1];
+
+ z->token_costs = x->token_costs;
+ z->inter_bmode_costs = x->inter_bmode_costs;
+ z->mbmode_cost = x->mbmode_cost;
+ z->intra_uv_mode_cost = x->intra_uv_mode_cost;
+ z->bmode_costs = x->bmode_costs;
+
+ for (i = 0; i < 25; i++)
+ {
+ z->block[i].quant = x->block[i].quant;
+ z->block[i].quant_fast = x->block[i].quant_fast;
+ z->block[i].quant_shift = x->block[i].quant_shift;
+ z->block[i].zbin = x->block[i].zbin;
+ z->block[i].zrun_zbin_boost = x->block[i].zrun_zbin_boost;
+ z->block[i].round = x->block[i].round;
+ z->block[i].src_stride = x->block[i].src_stride;
+ }
+
+ z->q_index = x->q_index;
+ z->act_zbin_adj = x->act_zbin_adj;
+ z->last_act_zbin_adj = x->last_act_zbin_adj;
+
+ {
+ MACROBLOCKD *xd = &x->e_mbd;
+ MACROBLOCKD *zd = &z->e_mbd;
+
+ /*
+ zd->mode_info_context = xd->mode_info_context;
+ zd->mode_info = xd->mode_info;
+
+ zd->mode_info_stride = xd->mode_info_stride;
+ zd->frame_type = xd->frame_type;
+ zd->up_available = xd->up_available ;
+ zd->left_available = xd->left_available;
+ zd->left_context = xd->left_context;
+ zd->last_frame_dc = xd->last_frame_dc;
+ zd->last_frame_dccons = xd->last_frame_dccons;
+ zd->gold_frame_dc = xd->gold_frame_dc;
+ zd->gold_frame_dccons = xd->gold_frame_dccons;
+ zd->mb_to_left_edge = xd->mb_to_left_edge;
+ zd->mb_to_right_edge = xd->mb_to_right_edge;
+ zd->mb_to_top_edge = xd->mb_to_top_edge ;
+ zd->mb_to_bottom_edge = xd->mb_to_bottom_edge;
+ zd->gf_active_ptr = xd->gf_active_ptr;
+ zd->frames_since_golden = xd->frames_since_golden;
+ zd->frames_till_alt_ref_frame = xd->frames_till_alt_ref_frame;
+ */
+ zd->subpixel_predict = xd->subpixel_predict;
+ zd->subpixel_predict8x4 = xd->subpixel_predict8x4;
+ zd->subpixel_predict8x8 = xd->subpixel_predict8x8;
+ zd->subpixel_predict16x16 = xd->subpixel_predict16x16;
+ zd->segmentation_enabled = xd->segmentation_enabled;
+ zd->mb_segement_abs_delta = xd->mb_segement_abs_delta;
+ vpx_memcpy(zd->segment_feature_data, xd->segment_feature_data,
+ sizeof(xd->segment_feature_data));
+
+ vpx_memcpy(zd->dequant_y1_dc, xd->dequant_y1_dc,
+ sizeof(xd->dequant_y1_dc));
+ vpx_memcpy(zd->dequant_y1, xd->dequant_y1, sizeof(xd->dequant_y1));
+ vpx_memcpy(zd->dequant_y2, xd->dequant_y2, sizeof(xd->dequant_y2));
+ vpx_memcpy(zd->dequant_uv, xd->dequant_uv, sizeof(xd->dequant_uv));
+
+#if 1
+ /*TODO: Remove dequant from BLOCKD. This is a temporary solution until
+ * the quantizer code uses a passed in pointer to the dequant constants.
+ * This will also require modifications to the x86 and neon assembly.
+ * */
+ for (i = 0; i < 16; i++)
+ zd->block[i].dequant = zd->dequant_y1;
+ for (i = 16; i < 24; i++)
+ zd->block[i].dequant = zd->dequant_uv;
+ zd->block[24].dequant = zd->dequant_y2;
+#endif
+ }
+}
+
+void vp8cx_init_mbrthread_data(VP8_COMP *cpi,
+ MACROBLOCK *x,
+ MB_ROW_COMP *mbr_ei,
+ int mb_row,
+ int count
+ )
+{
+
+ VP8_COMMON *const cm = & cpi->common;
+ MACROBLOCKD *const xd = & x->e_mbd;
+ int i;
+ (void) mb_row;
+
+ for (i = 0; i < count; i++)
+ {
+ MACROBLOCK *mb = & mbr_ei[i].mb;
+ MACROBLOCKD *mbd = &mb->e_mbd;
+
+ mbd->subpixel_predict = xd->subpixel_predict;
+ mbd->subpixel_predict8x4 = xd->subpixel_predict8x4;
+ mbd->subpixel_predict8x8 = xd->subpixel_predict8x8;
+ mbd->subpixel_predict16x16 = xd->subpixel_predict16x16;
+ mb->gf_active_ptr = x->gf_active_ptr;
+
+ vpx_memset(mbr_ei[i].segment_counts, 0, sizeof(mbr_ei[i].segment_counts));
+ mbr_ei[i].totalrate = 0;
+
+ mb->partition_info = x->pi + x->e_mbd.mode_info_stride * (i + 1);
+
+ mbd->mode_info_context = cm->mi + x->e_mbd.mode_info_stride * (i + 1);
+ mbd->mode_info_stride = cm->mode_info_stride;
+
+ mbd->frame_type = cm->frame_type;
+
+ mb->src = * cpi->Source;
+ mbd->pre = cm->yv12_fb[cm->lst_fb_idx];
+ mbd->dst = cm->yv12_fb[cm->new_fb_idx];
+
+ mb->src.y_buffer += 16 * x->src.y_stride * (i + 1);
+ mb->src.u_buffer += 8 * x->src.uv_stride * (i + 1);
+ mb->src.v_buffer += 8 * x->src.uv_stride * (i + 1);
+
+ vp8_build_block_offsets(mb);
+
+ mbd->left_context = &cm->left_context;
+ mb->mvc = cm->fc.mvc;
+
+ setup_mbby_copy(&mbr_ei[i].mb, x);
+
+ mbd->fullpixel_mask = 0xffffffff;
+ if(cm->full_pixel)
+ mbd->fullpixel_mask = 0xfffffff8;
+
+ vp8_zero(mb->coef_counts);
+ vp8_zero(x->ymode_count);
+ mb->skip_true_count = 0;
+ vp8_zero(mb->MVcount);
+ mb->prediction_error = 0;
+ mb->intra_error = 0;
+ }
+}
+
+int vp8cx_create_encoder_threads(VP8_COMP *cpi)
+{
+ const VP8_COMMON * cm = &cpi->common;
+
+ cpi->b_multi_threaded = 0;
+ cpi->encoding_thread_count = 0;
+ cpi->b_lpf_running = 0;
+
+ if (cm->processor_core_count > 1 && cpi->oxcf.multi_threaded > 1)
+ {
+ int ithread;
+ int th_count = cpi->oxcf.multi_threaded - 1;
+ int rc = 0;
+
+ /* don't allocate more threads than cores available */
+ if (cpi->oxcf.multi_threaded > cm->processor_core_count)
+ th_count = cm->processor_core_count - 1;
+
+ /* we have th_count + 1 (main) threads processing one row each */
+ /* no point to have more threads than the sync range allows */
+ if(th_count > ((cm->mb_cols / cpi->mt_sync_range) - 1))
+ {
+ th_count = (cm->mb_cols / cpi->mt_sync_range) - 1;
+ }
+
+ if(th_count == 0)
+ return 0;
+
+ CHECK_MEM_ERROR(cpi->h_encoding_thread,
+ vpx_malloc(sizeof(pthread_t) * th_count));
+ CHECK_MEM_ERROR(cpi->h_event_start_encoding,
+ vpx_malloc(sizeof(sem_t) * th_count));
+ CHECK_MEM_ERROR(cpi->mb_row_ei,
+ vpx_memalign(32, sizeof(MB_ROW_COMP) * th_count));
+ vpx_memset(cpi->mb_row_ei, 0, sizeof(MB_ROW_COMP) * th_count);
+ CHECK_MEM_ERROR(cpi->en_thread_data,
+ vpx_malloc(sizeof(ENCODETHREAD_DATA) * th_count));
+
+ sem_init(&cpi->h_event_end_encoding, 0, 0);
+
+ cpi->b_multi_threaded = 1;
+ cpi->encoding_thread_count = th_count;
+
+ /*
+ printf("[VP8:] multi_threaded encoding is enabled with %d threads\n\n",
+ (cpi->encoding_thread_count +1));
+ */
+
+ for (ithread = 0; ithread < th_count; ithread++)
+ {
+ ENCODETHREAD_DATA *ethd = &cpi->en_thread_data[ithread];
+
+ /* Setup block ptrs and offsets */
+ vp8_setup_block_ptrs(&cpi->mb_row_ei[ithread].mb);
+ vp8_setup_block_dptrs(&cpi->mb_row_ei[ithread].mb.e_mbd);
+
+ sem_init(&cpi->h_event_start_encoding[ithread], 0, 0);
+
+ ethd->ithread = ithread;
+ ethd->ptr1 = (void *)cpi;
+ ethd->ptr2 = (void *)&cpi->mb_row_ei[ithread];
+
+ rc = pthread_create(&cpi->h_encoding_thread[ithread], 0,
+ thread_encoding_proc, ethd);
+ if(rc)
+ break;
+ }
+
+ if(rc)
+ {
+ /* shutdown other threads */
+ cpi->b_multi_threaded = 0;
+ for(--ithread; ithread >= 0; ithread--)
+ {
+ pthread_join(cpi->h_encoding_thread[ithread], 0);
+ sem_destroy(&cpi->h_event_start_encoding[ithread]);
+ }
+ sem_destroy(&cpi->h_event_end_encoding);
+
+ /* free thread related resources */
+ vpx_free(cpi->h_event_start_encoding);
+ vpx_free(cpi->h_encoding_thread);
+ vpx_free(cpi->mb_row_ei);
+ vpx_free(cpi->en_thread_data);
+
+ return -1;
+ }
+
+
+ {
+ LPFTHREAD_DATA * lpfthd = &cpi->lpf_thread_data;
+
+ sem_init(&cpi->h_event_start_lpf, 0, 0);
+ sem_init(&cpi->h_event_end_lpf, 0, 0);
+
+ lpfthd->ptr1 = (void *)cpi;
+ rc = pthread_create(&cpi->h_filter_thread, 0, thread_loopfilter,
+ lpfthd);
+
+ if(rc)
+ {
+ /* shutdown other threads */
+ cpi->b_multi_threaded = 0;
+ for(--ithread; ithread >= 0; ithread--)
+ {
+ sem_post(&cpi->h_event_start_encoding[ithread]);
+ pthread_join(cpi->h_encoding_thread[ithread], 0);
+ sem_destroy(&cpi->h_event_start_encoding[ithread]);
+ }
+ sem_destroy(&cpi->h_event_end_encoding);
+ sem_destroy(&cpi->h_event_end_lpf);
+ sem_destroy(&cpi->h_event_start_lpf);
+
+ /* free thread related resources */
+ vpx_free(cpi->h_event_start_encoding);
+ vpx_free(cpi->h_encoding_thread);
+ vpx_free(cpi->mb_row_ei);
+ vpx_free(cpi->en_thread_data);
+
+ return -2;
+ }
+ }
+ }
+ return 0;
+}
+
+void vp8cx_remove_encoder_threads(VP8_COMP *cpi)
+{
+ if (cpi->b_multi_threaded)
+ {
+ /* shutdown other threads */
+ cpi->b_multi_threaded = 0;
+ {
+ int i;
+
+ for (i = 0; i < cpi->encoding_thread_count; i++)
+ {
+ sem_post(&cpi->h_event_start_encoding[i]);
+ pthread_join(cpi->h_encoding_thread[i], 0);
+
+ sem_destroy(&cpi->h_event_start_encoding[i]);
+ }
+
+ sem_post(&cpi->h_event_start_lpf);
+ pthread_join(cpi->h_filter_thread, 0);
+ }
+
+ sem_destroy(&cpi->h_event_end_encoding);
+ sem_destroy(&cpi->h_event_end_lpf);
+ sem_destroy(&cpi->h_event_start_lpf);
+
+ /* free thread related resources */
+ vpx_free(cpi->h_event_start_encoding);
+ vpx_free(cpi->h_encoding_thread);
+ vpx_free(cpi->mb_row_ei);
+ vpx_free(cpi->en_thread_data);
+ }
+}
+#endif
diff --git a/libvpx/vp8/encoder/firstpass.c b/libvpx/vp8/encoder/firstpass.c
new file mode 100644
index 0000000..b668c8f
--- /dev/null
+++ b/libvpx/vp8/encoder/firstpass.c
@@ -0,0 +1,3360 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include <math.h>
+#include <limits.h>
+#include <stdio.h>
+
+#include "block.h"
+#include "onyx_int.h"
+#include "vp8/common/variance.h"
+#include "encodeintra.h"
+#include "vp8/common/setupintrarecon.h"
+#include "vp8/common/systemdependent.h"
+#include "mcomp.h"
+#include "firstpass.h"
+#include "vpx_scale/vpxscale.h"
+#include "encodemb.h"
+#include "vp8/common/extend.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/swapyv12buffer.h"
+#include "rdopt.h"
+#include "vp8/common/quant_common.h"
+#include "encodemv.h"
+#include "encodeframe.h"
+
+/* #define OUTPUT_FPF 1 */
+
+extern void vp8cx_frame_init_quantizer(VP8_COMP *cpi);
+extern void vp8_set_mbmode_and_mvs(MACROBLOCK *x, MB_PREDICTION_MODE mb, int_mv *mv);
+extern void vp8_alloc_compressor_data(VP8_COMP *cpi);
+
+#define GFQ_ADJUSTMENT vp8_gf_boost_qadjustment[Q]
+extern int vp8_kf_boost_qadjustment[QINDEX_RANGE];
+
+extern const int vp8_gf_boost_qadjustment[QINDEX_RANGE];
+
+#define IIFACTOR 1.5
+#define IIKFACTOR1 1.40
+#define IIKFACTOR2 1.5
+#define RMAX 14.0
+#define GF_RMAX 48.0
+
+#define KF_MB_INTRA_MIN 300
+#define GF_MB_INTRA_MIN 200
+
+#define DOUBLE_DIVIDE_CHECK(X) ((X)<0?(X)-.000001:(X)+.000001)
+
+#define POW1 (double)cpi->oxcf.two_pass_vbrbias/100.0
+#define POW2 (double)cpi->oxcf.two_pass_vbrbias/100.0
+
+#define NEW_BOOST 1
+
+static int vscale_lookup[7] = {0, 1, 1, 2, 2, 3, 3};
+static int hscale_lookup[7] = {0, 0, 1, 1, 2, 2, 3};
+
+
+static const int cq_level[QINDEX_RANGE] =
+{
+ 0,0,1,1,2,3,3,4,4,5,6,6,7,8,8,9,
+ 9,10,11,11,12,13,13,14,15,15,16,17,17,18,19,20,
+ 20,21,22,22,23,24,24,25,26,27,27,28,29,30,30,31,
+ 32,33,33,34,35,36,36,37,38,39,39,40,41,42,42,43,
+ 44,45,46,46,47,48,49,50,50,51,52,53,54,55,55,56,
+ 57,58,59,60,60,61,62,63,64,65,66,67,67,68,69,70,
+ 71,72,73,74,75,75,76,77,78,79,80,81,82,83,84,85,
+ 86,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100
+};
+
+static void find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame);
+
+/* Resets the first pass file to the given position using a relative seek
+ * from the current position
+ */
+static void reset_fpf_position(VP8_COMP *cpi, FIRSTPASS_STATS *Position)
+{
+ cpi->twopass.stats_in = Position;
+}
+
+static int lookup_next_frame_stats(VP8_COMP *cpi, FIRSTPASS_STATS *next_frame)
+{
+ if (cpi->twopass.stats_in >= cpi->twopass.stats_in_end)
+ return EOF;
+
+ *next_frame = *cpi->twopass.stats_in;
+ return 1;
+}
+
+/* Read frame stats at an offset from the current position */
+static int read_frame_stats( VP8_COMP *cpi,
+ FIRSTPASS_STATS *frame_stats,
+ int offset )
+{
+ FIRSTPASS_STATS * fps_ptr = cpi->twopass.stats_in;
+
+ /* Check legality of offset */
+ if ( offset >= 0 )
+ {
+ if ( &fps_ptr[offset] >= cpi->twopass.stats_in_end )
+ return EOF;
+ }
+ else if ( offset < 0 )
+ {
+ if ( &fps_ptr[offset] < cpi->twopass.stats_in_start )
+ return EOF;
+ }
+
+ *frame_stats = fps_ptr[offset];
+ return 1;
+}
+
+static int input_stats(VP8_COMP *cpi, FIRSTPASS_STATS *fps)
+{
+ if (cpi->twopass.stats_in >= cpi->twopass.stats_in_end)
+ return EOF;
+
+ *fps = *cpi->twopass.stats_in;
+ cpi->twopass.stats_in =
+ (void*)((char *)cpi->twopass.stats_in + sizeof(FIRSTPASS_STATS));
+ return 1;
+}
+
+static void output_stats(const VP8_COMP *cpi,
+ struct vpx_codec_pkt_list *pktlist,
+ FIRSTPASS_STATS *stats)
+{
+ struct vpx_codec_cx_pkt pkt;
+ pkt.kind = VPX_CODEC_STATS_PKT;
+ pkt.data.twopass_stats.buf = stats;
+ pkt.data.twopass_stats.sz = sizeof(FIRSTPASS_STATS);
+ vpx_codec_pkt_list_add(pktlist, &pkt);
+
+/* TEMP debug code */
+#if OUTPUT_FPF
+
+ {
+ FILE *fpfile;
+ fpfile = fopen("firstpass.stt", "a");
+
+ fprintf(fpfile, "%12.0f %12.0f %12.0f %12.4f %12.4f %12.4f %12.4f"
+ " %12.4f %12.4f %12.4f %12.4f %12.4f %12.4f %12.4f %12.4f"
+ " %12.0f %12.0f %12.4f\n",
+ stats->frame,
+ stats->intra_error,
+ stats->coded_error,
+ stats->ssim_weighted_pred_err,
+ stats->pcnt_inter,
+ stats->pcnt_motion,
+ stats->pcnt_second_ref,
+ stats->pcnt_neutral,
+ stats->MVr,
+ stats->mvr_abs,
+ stats->MVc,
+ stats->mvc_abs,
+ stats->MVrv,
+ stats->MVcv,
+ stats->mv_in_out_count,
+ stats->new_mv_count,
+ stats->count,
+ stats->duration);
+ fclose(fpfile);
+ }
+#endif
+}
+
+static void zero_stats(FIRSTPASS_STATS *section)
+{
+ section->frame = 0.0;
+ section->intra_error = 0.0;
+ section->coded_error = 0.0;
+ section->ssim_weighted_pred_err = 0.0;
+ section->pcnt_inter = 0.0;
+ section->pcnt_motion = 0.0;
+ section->pcnt_second_ref = 0.0;
+ section->pcnt_neutral = 0.0;
+ section->MVr = 0.0;
+ section->mvr_abs = 0.0;
+ section->MVc = 0.0;
+ section->mvc_abs = 0.0;
+ section->MVrv = 0.0;
+ section->MVcv = 0.0;
+ section->mv_in_out_count = 0.0;
+ section->new_mv_count = 0.0;
+ section->count = 0.0;
+ section->duration = 1.0;
+}
+
+static void accumulate_stats(FIRSTPASS_STATS *section, FIRSTPASS_STATS *frame)
+{
+ section->frame += frame->frame;
+ section->intra_error += frame->intra_error;
+ section->coded_error += frame->coded_error;
+ section->ssim_weighted_pred_err += frame->ssim_weighted_pred_err;
+ section->pcnt_inter += frame->pcnt_inter;
+ section->pcnt_motion += frame->pcnt_motion;
+ section->pcnt_second_ref += frame->pcnt_second_ref;
+ section->pcnt_neutral += frame->pcnt_neutral;
+ section->MVr += frame->MVr;
+ section->mvr_abs += frame->mvr_abs;
+ section->MVc += frame->MVc;
+ section->mvc_abs += frame->mvc_abs;
+ section->MVrv += frame->MVrv;
+ section->MVcv += frame->MVcv;
+ section->mv_in_out_count += frame->mv_in_out_count;
+ section->new_mv_count += frame->new_mv_count;
+ section->count += frame->count;
+ section->duration += frame->duration;
+}
+
+static void subtract_stats(FIRSTPASS_STATS *section, FIRSTPASS_STATS *frame)
+{
+ section->frame -= frame->frame;
+ section->intra_error -= frame->intra_error;
+ section->coded_error -= frame->coded_error;
+ section->ssim_weighted_pred_err -= frame->ssim_weighted_pred_err;
+ section->pcnt_inter -= frame->pcnt_inter;
+ section->pcnt_motion -= frame->pcnt_motion;
+ section->pcnt_second_ref -= frame->pcnt_second_ref;
+ section->pcnt_neutral -= frame->pcnt_neutral;
+ section->MVr -= frame->MVr;
+ section->mvr_abs -= frame->mvr_abs;
+ section->MVc -= frame->MVc;
+ section->mvc_abs -= frame->mvc_abs;
+ section->MVrv -= frame->MVrv;
+ section->MVcv -= frame->MVcv;
+ section->mv_in_out_count -= frame->mv_in_out_count;
+ section->new_mv_count -= frame->new_mv_count;
+ section->count -= frame->count;
+ section->duration -= frame->duration;
+}
+
+static void avg_stats(FIRSTPASS_STATS *section)
+{
+ if (section->count < 1.0)
+ return;
+
+ section->intra_error /= section->count;
+ section->coded_error /= section->count;
+ section->ssim_weighted_pred_err /= section->count;
+ section->pcnt_inter /= section->count;
+ section->pcnt_second_ref /= section->count;
+ section->pcnt_neutral /= section->count;
+ section->pcnt_motion /= section->count;
+ section->MVr /= section->count;
+ section->mvr_abs /= section->count;
+ section->MVc /= section->count;
+ section->mvc_abs /= section->count;
+ section->MVrv /= section->count;
+ section->MVcv /= section->count;
+ section->mv_in_out_count /= section->count;
+ section->duration /= section->count;
+}
+
+/* Calculate a modified Error used in distributing bits between easier
+ * and harder frames
+ */
+static double calculate_modified_err(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
+{
+ double av_err = ( cpi->twopass.total_stats.ssim_weighted_pred_err /
+ cpi->twopass.total_stats.count );
+ double this_err = this_frame->ssim_weighted_pred_err;
+ double modified_err;
+
+ if (this_err > av_err)
+ modified_err = av_err * pow((this_err / DOUBLE_DIVIDE_CHECK(av_err)), POW1);
+ else
+ modified_err = av_err * pow((this_err / DOUBLE_DIVIDE_CHECK(av_err)), POW2);
+
+ return modified_err;
+}
+
+static const double weight_table[256] = {
+0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000,
+0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000,
+0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000,
+0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000,
+0.020000, 0.031250, 0.062500, 0.093750, 0.125000, 0.156250, 0.187500, 0.218750,
+0.250000, 0.281250, 0.312500, 0.343750, 0.375000, 0.406250, 0.437500, 0.468750,
+0.500000, 0.531250, 0.562500, 0.593750, 0.625000, 0.656250, 0.687500, 0.718750,
+0.750000, 0.781250, 0.812500, 0.843750, 0.875000, 0.906250, 0.937500, 0.968750,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
+1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000
+};
+
+static double simple_weight(YV12_BUFFER_CONFIG *source)
+{
+ int i, j;
+
+ unsigned char *src = source->y_buffer;
+ double sum_weights = 0.0;
+
+ /* Loop throught the Y plane raw examining levels and creating a weight
+ * for the image
+ */
+ i = source->y_height;
+ do
+ {
+ j = source->y_width;
+ do
+ {
+ sum_weights += weight_table[ *src];
+ src++;
+ }while(--j);
+ src -= source->y_width;
+ src += source->y_stride;
+ }while(--i);
+
+ sum_weights /= (source->y_height * source->y_width);
+
+ return sum_weights;
+}
+
+
+/* This function returns the current per frame maximum bitrate target */
+static int frame_max_bits(VP8_COMP *cpi)
+{
+ /* Max allocation for a single frame based on the max section guidelines
+ * passed in and how many bits are left
+ */
+ int max_bits;
+
+ /* For CBR we need to also consider buffer fullness.
+ * If we are running below the optimal level then we need to gradually
+ * tighten up on max_bits.
+ */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ double buffer_fullness_ratio = (double)cpi->buffer_level / DOUBLE_DIVIDE_CHECK((double)cpi->oxcf.optimal_buffer_level);
+
+ /* For CBR base this on the target average bits per frame plus the
+ * maximum sedction rate passed in by the user
+ */
+ max_bits = (int)(cpi->av_per_frame_bandwidth * ((double)cpi->oxcf.two_pass_vbrmax_section / 100.0));
+
+ /* If our buffer is below the optimum level */
+ if (buffer_fullness_ratio < 1.0)
+ {
+ /* The lower of max_bits / 4 or cpi->av_per_frame_bandwidth / 4. */
+ int min_max_bits = ((cpi->av_per_frame_bandwidth >> 2) < (max_bits >> 2)) ? cpi->av_per_frame_bandwidth >> 2 : max_bits >> 2;
+
+ max_bits = (int)(max_bits * buffer_fullness_ratio);
+
+ /* Lowest value we will set ... which should allow the buffer to
+ * refill.
+ */
+ if (max_bits < min_max_bits)
+ max_bits = min_max_bits;
+ }
+ }
+ /* VBR */
+ else
+ {
+ /* For VBR base this on the bits and frames left plus the
+ * two_pass_vbrmax_section rate passed in by the user
+ */
+ max_bits = (int)(((double)cpi->twopass.bits_left / (cpi->twopass.total_stats.count - (double)cpi->common.current_video_frame)) * ((double)cpi->oxcf.two_pass_vbrmax_section / 100.0));
+ }
+
+ /* Trap case where we are out of bits */
+ if (max_bits < 0)
+ max_bits = 0;
+
+ return max_bits;
+}
+
+void vp8_init_first_pass(VP8_COMP *cpi)
+{
+ zero_stats(&cpi->twopass.total_stats);
+}
+
+void vp8_end_first_pass(VP8_COMP *cpi)
+{
+ output_stats(cpi, cpi->output_pkt_list, &cpi->twopass.total_stats);
+}
+
+static void zz_motion_search( VP8_COMP *cpi, MACROBLOCK * x,
+ YV12_BUFFER_CONFIG * raw_buffer,
+ int * raw_motion_err,
+ YV12_BUFFER_CONFIG * recon_buffer,
+ int * best_motion_err, int recon_yoffset)
+{
+ MACROBLOCKD * const xd = & x->e_mbd;
+ BLOCK *b = &x->block[0];
+ BLOCKD *d = &x->e_mbd.block[0];
+
+ unsigned char *src_ptr = (*(b->base_src) + b->src);
+ int src_stride = b->src_stride;
+ unsigned char *raw_ptr;
+ int raw_stride = raw_buffer->y_stride;
+ unsigned char *ref_ptr;
+ int ref_stride = x->e_mbd.pre.y_stride;
+
+ /* Set up pointers for this macro block raw buffer */
+ raw_ptr = (unsigned char *)(raw_buffer->y_buffer + recon_yoffset
+ + d->offset);
+ vp8_mse16x16 ( src_ptr, src_stride, raw_ptr, raw_stride,
+ (unsigned int *)(raw_motion_err));
+
+ /* Set up pointers for this macro block recon buffer */
+ xd->pre.y_buffer = recon_buffer->y_buffer + recon_yoffset;
+ ref_ptr = (unsigned char *)(xd->pre.y_buffer + d->offset );
+ vp8_mse16x16 ( src_ptr, src_stride, ref_ptr, ref_stride,
+ (unsigned int *)(best_motion_err));
+}
+
+static void first_pass_motion_search(VP8_COMP *cpi, MACROBLOCK *x,
+ int_mv *ref_mv, MV *best_mv,
+ YV12_BUFFER_CONFIG *recon_buffer,
+ int *best_motion_err, int recon_yoffset )
+{
+ MACROBLOCKD *const xd = & x->e_mbd;
+ BLOCK *b = &x->block[0];
+ BLOCKD *d = &x->e_mbd.block[0];
+ int num00;
+
+ int_mv tmp_mv;
+ int_mv ref_mv_full;
+
+ int tmp_err;
+ int step_param = 3; /* Dont search over full range for first pass */
+ int further_steps = (MAX_MVSEARCH_STEPS - 1) - step_param;
+ int n;
+ vp8_variance_fn_ptr_t v_fn_ptr = cpi->fn_ptr[BLOCK_16X16];
+ int new_mv_mode_penalty = 256;
+
+ /* override the default variance function to use MSE */
+ v_fn_ptr.vf = vp8_mse16x16;
+
+ /* Set up pointers for this macro block recon buffer */
+ xd->pre.y_buffer = recon_buffer->y_buffer + recon_yoffset;
+
+ /* Initial step/diamond search centred on best mv */
+ tmp_mv.as_int = 0;
+ ref_mv_full.as_mv.col = ref_mv->as_mv.col>>3;
+ ref_mv_full.as_mv.row = ref_mv->as_mv.row>>3;
+ tmp_err = cpi->diamond_search_sad(x, b, d, &ref_mv_full, &tmp_mv, step_param,
+ x->sadperbit16, &num00, &v_fn_ptr,
+ x->mvcost, ref_mv);
+ if ( tmp_err < INT_MAX-new_mv_mode_penalty )
+ tmp_err += new_mv_mode_penalty;
+
+ if (tmp_err < *best_motion_err)
+ {
+ *best_motion_err = tmp_err;
+ best_mv->row = tmp_mv.as_mv.row;
+ best_mv->col = tmp_mv.as_mv.col;
+ }
+
+ /* Further step/diamond searches as necessary */
+ n = num00;
+ num00 = 0;
+
+ while (n < further_steps)
+ {
+ n++;
+
+ if (num00)
+ num00--;
+ else
+ {
+ tmp_err = cpi->diamond_search_sad(x, b, d, &ref_mv_full, &tmp_mv,
+ step_param + n, x->sadperbit16,
+ &num00, &v_fn_ptr, x->mvcost,
+ ref_mv);
+ if ( tmp_err < INT_MAX-new_mv_mode_penalty )
+ tmp_err += new_mv_mode_penalty;
+
+ if (tmp_err < *best_motion_err)
+ {
+ *best_motion_err = tmp_err;
+ best_mv->row = tmp_mv.as_mv.row;
+ best_mv->col = tmp_mv.as_mv.col;
+ }
+ }
+ }
+}
+
+void vp8_first_pass(VP8_COMP *cpi)
+{
+ int mb_row, mb_col;
+ MACROBLOCK *const x = & cpi->mb;
+ VP8_COMMON *const cm = & cpi->common;
+ MACROBLOCKD *const xd = & x->e_mbd;
+
+ int recon_yoffset, recon_uvoffset;
+ YV12_BUFFER_CONFIG *lst_yv12 = &cm->yv12_fb[cm->lst_fb_idx];
+ YV12_BUFFER_CONFIG *new_yv12 = &cm->yv12_fb[cm->new_fb_idx];
+ YV12_BUFFER_CONFIG *gld_yv12 = &cm->yv12_fb[cm->gld_fb_idx];
+ int recon_y_stride = lst_yv12->y_stride;
+ int recon_uv_stride = lst_yv12->uv_stride;
+ int64_t intra_error = 0;
+ int64_t coded_error = 0;
+
+ int sum_mvr = 0, sum_mvc = 0;
+ int sum_mvr_abs = 0, sum_mvc_abs = 0;
+ int sum_mvrs = 0, sum_mvcs = 0;
+ int mvcount = 0;
+ int intercount = 0;
+ int second_ref_count = 0;
+ int intrapenalty = 256;
+ int neutral_count = 0;
+ int new_mv_count = 0;
+ int sum_in_vectors = 0;
+ uint32_t lastmv_as_int = 0;
+
+ int_mv zero_ref_mv;
+
+ zero_ref_mv.as_int = 0;
+
+ vp8_clear_system_state();
+
+ x->src = * cpi->Source;
+ xd->pre = *lst_yv12;
+ xd->dst = *new_yv12;
+
+ x->partition_info = x->pi;
+
+ xd->mode_info_context = cm->mi;
+
+ if(!cm->use_bilinear_mc_filter)
+ {
+ xd->subpixel_predict = vp8_sixtap_predict4x4;
+ xd->subpixel_predict8x4 = vp8_sixtap_predict8x4;
+ xd->subpixel_predict8x8 = vp8_sixtap_predict8x8;
+ xd->subpixel_predict16x16 = vp8_sixtap_predict16x16;
+ }
+ else
+ {
+ xd->subpixel_predict = vp8_bilinear_predict4x4;
+ xd->subpixel_predict8x4 = vp8_bilinear_predict8x4;
+ xd->subpixel_predict8x8 = vp8_bilinear_predict8x8;
+ xd->subpixel_predict16x16 = vp8_bilinear_predict16x16;
+ }
+
+ vp8_build_block_offsets(x);
+
+ /* set up frame new frame for intra coded blocks */
+ vp8_setup_intra_recon(new_yv12);
+ vp8cx_frame_init_quantizer(cpi);
+
+ /* Initialise the MV cost table to the defaults */
+ {
+ int flag[2] = {1, 1};
+ vp8_initialize_rd_consts(cpi, vp8_dc_quant(cm->base_qindex, cm->y1dc_delta_q));
+ vpx_memcpy(cm->fc.mvc, vp8_default_mv_context, sizeof(vp8_default_mv_context));
+ vp8_build_component_cost_table(cpi->mb.mvcost, (const MV_CONTEXT *) cm->fc.mvc, flag);
+ }
+
+ /* for each macroblock row in image */
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
+ {
+ int_mv best_ref_mv;
+
+ best_ref_mv.as_int = 0;
+
+ /* reset above block coeffs */
+ xd->up_available = (mb_row != 0);
+ recon_yoffset = (mb_row * recon_y_stride * 16);
+ recon_uvoffset = (mb_row * recon_uv_stride * 8);
+
+ /* Set up limit values for motion vectors to prevent them extending
+ * outside the UMV borders
+ */
+ x->mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 16));
+ x->mv_row_max = ((cm->mb_rows - 1 - mb_row) * 16) + (VP8BORDERINPIXELS - 16);
+
+
+ /* for each macroblock col in image */
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+ int this_error;
+ int gf_motion_error = INT_MAX;
+ int use_dc_pred = (mb_col || mb_row) && (!mb_col || !mb_row);
+
+ xd->dst.y_buffer = new_yv12->y_buffer + recon_yoffset;
+ xd->dst.u_buffer = new_yv12->u_buffer + recon_uvoffset;
+ xd->dst.v_buffer = new_yv12->v_buffer + recon_uvoffset;
+ xd->left_available = (mb_col != 0);
+
+ /* Copy current mb to a buffer */
+ vp8_copy_mem16x16(x->src.y_buffer, x->src.y_stride, x->thismb, 16);
+
+ /* do intra 16x16 prediction */
+ this_error = vp8_encode_intra(cpi, x, use_dc_pred);
+
+ /* "intrapenalty" below deals with situations where the intra
+ * and inter error scores are very low (eg a plain black frame)
+ * We do not have special cases in first pass for 0,0 and
+ * nearest etc so all inter modes carry an overhead cost
+ * estimate fot the mv. When the error score is very low this
+ * causes us to pick all or lots of INTRA modes and throw lots
+ * of key frames. This penalty adds a cost matching that of a
+ * 0,0 mv to the intra case.
+ */
+ this_error += intrapenalty;
+
+ /* Cumulative intra error total */
+ intra_error += (int64_t)this_error;
+
+ /* Set up limit values for motion vectors to prevent them
+ * extending outside the UMV borders
+ */
+ x->mv_col_min = -((mb_col * 16) + (VP8BORDERINPIXELS - 16));
+ x->mv_col_max = ((cm->mb_cols - 1 - mb_col) * 16) + (VP8BORDERINPIXELS - 16);
+
+ /* Other than for the first frame do a motion search */
+ if (cm->current_video_frame > 0)
+ {
+ BLOCKD *d = &x->e_mbd.block[0];
+ MV tmp_mv = {0, 0};
+ int tmp_err;
+ int motion_error = INT_MAX;
+ int raw_motion_error = INT_MAX;
+
+ /* Simple 0,0 motion with no mv overhead */
+ zz_motion_search( cpi, x, cpi->last_frame_unscaled_source,
+ &raw_motion_error, lst_yv12, &motion_error,
+ recon_yoffset );
+ d->bmi.mv.as_mv.row = 0;
+ d->bmi.mv.as_mv.col = 0;
+
+ if (raw_motion_error < cpi->oxcf.encode_breakout)
+ goto skip_motion_search;
+
+ /* Test last reference frame using the previous best mv as the
+ * starting point (best reference) for the search
+ */
+ first_pass_motion_search(cpi, x, &best_ref_mv,
+ &d->bmi.mv.as_mv, lst_yv12,
+ &motion_error, recon_yoffset);
+
+ /* If the current best reference mv is not centred on 0,0
+ * then do a 0,0 based search as well
+ */
+ if (best_ref_mv.as_int)
+ {
+ tmp_err = INT_MAX;
+ first_pass_motion_search(cpi, x, &zero_ref_mv, &tmp_mv,
+ lst_yv12, &tmp_err, recon_yoffset);
+
+ if ( tmp_err < motion_error )
+ {
+ motion_error = tmp_err;
+ d->bmi.mv.as_mv.row = tmp_mv.row;
+ d->bmi.mv.as_mv.col = tmp_mv.col;
+ }
+ }
+
+ /* Experimental search in a second reference frame ((0,0)
+ * based only)
+ */
+ if (cm->current_video_frame > 1)
+ {
+ first_pass_motion_search(cpi, x, &zero_ref_mv, &tmp_mv, gld_yv12, &gf_motion_error, recon_yoffset);
+
+ if ((gf_motion_error < motion_error) && (gf_motion_error < this_error))
+ {
+ second_ref_count++;
+ }
+
+ /* Reset to last frame as reference buffer */
+ xd->pre.y_buffer = lst_yv12->y_buffer + recon_yoffset;
+ xd->pre.u_buffer = lst_yv12->u_buffer + recon_uvoffset;
+ xd->pre.v_buffer = lst_yv12->v_buffer + recon_uvoffset;
+ }
+
+skip_motion_search:
+ /* Intra assumed best */
+ best_ref_mv.as_int = 0;
+
+ if (motion_error <= this_error)
+ {
+ /* Keep a count of cases where the inter and intra were
+ * very close and very low. This helps with scene cut
+ * detection for example in cropped clips with black bars
+ * at the sides or top and bottom.
+ */
+ if( (((this_error-intrapenalty) * 9) <=
+ (motion_error*10)) &&
+ (this_error < (2*intrapenalty)) )
+ {
+ neutral_count++;
+ }
+
+ d->bmi.mv.as_mv.row <<= 3;
+ d->bmi.mv.as_mv.col <<= 3;
+ this_error = motion_error;
+ vp8_set_mbmode_and_mvs(x, NEWMV, &d->bmi.mv);
+ vp8_encode_inter16x16y(x);
+ sum_mvr += d->bmi.mv.as_mv.row;
+ sum_mvr_abs += abs(d->bmi.mv.as_mv.row);
+ sum_mvc += d->bmi.mv.as_mv.col;
+ sum_mvc_abs += abs(d->bmi.mv.as_mv.col);
+ sum_mvrs += d->bmi.mv.as_mv.row * d->bmi.mv.as_mv.row;
+ sum_mvcs += d->bmi.mv.as_mv.col * d->bmi.mv.as_mv.col;
+ intercount++;
+
+ best_ref_mv.as_int = d->bmi.mv.as_int;
+
+ /* Was the vector non-zero */
+ if (d->bmi.mv.as_int)
+ {
+ mvcount++;
+
+ /* Was it different from the last non zero vector */
+ if ( d->bmi.mv.as_int != lastmv_as_int )
+ new_mv_count++;
+ lastmv_as_int = d->bmi.mv.as_int;
+
+ /* Does the Row vector point inwards or outwards */
+ if (mb_row < cm->mb_rows / 2)
+ {
+ if (d->bmi.mv.as_mv.row > 0)
+ sum_in_vectors--;
+ else if (d->bmi.mv.as_mv.row < 0)
+ sum_in_vectors++;
+ }
+ else if (mb_row > cm->mb_rows / 2)
+ {
+ if (d->bmi.mv.as_mv.row > 0)
+ sum_in_vectors++;
+ else if (d->bmi.mv.as_mv.row < 0)
+ sum_in_vectors--;
+ }
+
+ /* Does the Row vector point inwards or outwards */
+ if (mb_col < cm->mb_cols / 2)
+ {
+ if (d->bmi.mv.as_mv.col > 0)
+ sum_in_vectors--;
+ else if (d->bmi.mv.as_mv.col < 0)
+ sum_in_vectors++;
+ }
+ else if (mb_col > cm->mb_cols / 2)
+ {
+ if (d->bmi.mv.as_mv.col > 0)
+ sum_in_vectors++;
+ else if (d->bmi.mv.as_mv.col < 0)
+ sum_in_vectors--;
+ }
+ }
+ }
+ }
+
+ coded_error += (int64_t)this_error;
+
+ /* adjust to the next column of macroblocks */
+ x->src.y_buffer += 16;
+ x->src.u_buffer += 8;
+ x->src.v_buffer += 8;
+
+ recon_yoffset += 16;
+ recon_uvoffset += 8;
+ }
+
+ /* adjust to the next row of mbs */
+ x->src.y_buffer += 16 * x->src.y_stride - 16 * cm->mb_cols;
+ x->src.u_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols;
+ x->src.v_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols;
+
+ /* extend the recon for intra prediction */
+ vp8_extend_mb_row(new_yv12, xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8);
+ vp8_clear_system_state();
+ }
+
+ vp8_clear_system_state();
+ {
+ double weight = 0.0;
+
+ FIRSTPASS_STATS fps;
+
+ fps.frame = cm->current_video_frame ;
+ fps.intra_error = (double)(intra_error >> 8);
+ fps.coded_error = (double)(coded_error >> 8);
+ weight = simple_weight(cpi->Source);
+
+
+ if (weight < 0.1)
+ weight = 0.1;
+
+ fps.ssim_weighted_pred_err = fps.coded_error * weight;
+
+ fps.pcnt_inter = 0.0;
+ fps.pcnt_motion = 0.0;
+ fps.MVr = 0.0;
+ fps.mvr_abs = 0.0;
+ fps.MVc = 0.0;
+ fps.mvc_abs = 0.0;
+ fps.MVrv = 0.0;
+ fps.MVcv = 0.0;
+ fps.mv_in_out_count = 0.0;
+ fps.new_mv_count = 0.0;
+ fps.count = 1.0;
+
+ fps.pcnt_inter = 1.0 * (double)intercount / cm->MBs;
+ fps.pcnt_second_ref = 1.0 * (double)second_ref_count / cm->MBs;
+ fps.pcnt_neutral = 1.0 * (double)neutral_count / cm->MBs;
+
+ if (mvcount > 0)
+ {
+ fps.MVr = (double)sum_mvr / (double)mvcount;
+ fps.mvr_abs = (double)sum_mvr_abs / (double)mvcount;
+ fps.MVc = (double)sum_mvc / (double)mvcount;
+ fps.mvc_abs = (double)sum_mvc_abs / (double)mvcount;
+ fps.MVrv = ((double)sum_mvrs - (fps.MVr * fps.MVr / (double)mvcount)) / (double)mvcount;
+ fps.MVcv = ((double)sum_mvcs - (fps.MVc * fps.MVc / (double)mvcount)) / (double)mvcount;
+ fps.mv_in_out_count = (double)sum_in_vectors / (double)(mvcount * 2);
+ fps.new_mv_count = new_mv_count;
+
+ fps.pcnt_motion = 1.0 * (double)mvcount / cpi->common.MBs;
+ }
+
+ /* TODO: handle the case when duration is set to 0, or something less
+ * than the full time between subsequent cpi->source_time_stamps
+ */
+ fps.duration = (double)(cpi->source->ts_end
+ - cpi->source->ts_start);
+
+ /* don't want to do output stats with a stack variable! */
+ memcpy(&cpi->twopass.this_frame_stats,
+ &fps,
+ sizeof(FIRSTPASS_STATS));
+ output_stats(cpi, cpi->output_pkt_list, &cpi->twopass.this_frame_stats);
+ accumulate_stats(&cpi->twopass.total_stats, &fps);
+ }
+
+ /* Copy the previous Last Frame into the GF buffer if specific
+ * conditions for doing so are met
+ */
+ if ((cm->current_video_frame > 0) &&
+ (cpi->twopass.this_frame_stats.pcnt_inter > 0.20) &&
+ ((cpi->twopass.this_frame_stats.intra_error / cpi->twopass.this_frame_stats.coded_error) > 2.0))
+ {
+ vp8_yv12_copy_frame(lst_yv12, gld_yv12);
+ }
+
+ /* swap frame pointers so last frame refers to the frame we just
+ * compressed
+ */
+ vp8_swap_yv12_buffer(lst_yv12, new_yv12);
+ vp8_yv12_extend_frame_borders(lst_yv12);
+
+ /* Special case for the first frame. Copy into the GF buffer as a
+ * second reference.
+ */
+ if (cm->current_video_frame == 0)
+ {
+ vp8_yv12_copy_frame(lst_yv12, gld_yv12);
+ }
+
+
+ /* use this to see what the first pass reconstruction looks like */
+ if (0)
+ {
+ char filename[512];
+ FILE *recon_file;
+ sprintf(filename, "enc%04d.yuv", (int) cm->current_video_frame);
+
+ if (cm->current_video_frame == 0)
+ recon_file = fopen(filename, "wb");
+ else
+ recon_file = fopen(filename, "ab");
+
+ (void) fwrite(lst_yv12->buffer_alloc, lst_yv12->frame_size, 1,
+ recon_file);
+ fclose(recon_file);
+ }
+
+ cm->current_video_frame++;
+
+}
+extern const int vp8_bits_per_mb[2][QINDEX_RANGE];
+
+/* Estimate a cost per mb attributable to overheads such as the coding of
+ * modes and motion vectors.
+ * Currently simplistic in its assumptions for testing.
+ */
+
+static double bitcost( double prob )
+{
+ return -(log( prob ) / log( 2.0 ));
+}
+static int64_t estimate_modemvcost(VP8_COMP *cpi,
+ FIRSTPASS_STATS * fpstats)
+{
+ int mv_cost;
+ int mode_cost;
+
+ double av_pct_inter = fpstats->pcnt_inter / fpstats->count;
+ double av_pct_motion = fpstats->pcnt_motion / fpstats->count;
+ double av_intra = (1.0 - av_pct_inter);
+
+ double zz_cost;
+ double motion_cost;
+ double intra_cost;
+
+ zz_cost = bitcost(av_pct_inter - av_pct_motion);
+ motion_cost = bitcost(av_pct_motion);
+ intra_cost = bitcost(av_intra);
+
+ /* Estimate of extra bits per mv overhead for mbs
+ * << 9 is the normalization to the (bits * 512) used in vp8_bits_per_mb
+ */
+ mv_cost = ((int)(fpstats->new_mv_count / fpstats->count) * 8) << 9;
+
+ /* Crude estimate of overhead cost from modes
+ * << 9 is the normalization to (bits * 512) used in vp8_bits_per_mb
+ */
+ mode_cost =
+ (int)( ( ((av_pct_inter - av_pct_motion) * zz_cost) +
+ (av_pct_motion * motion_cost) +
+ (av_intra * intra_cost) ) * cpi->common.MBs ) << 9;
+
+ return mv_cost + mode_cost;
+}
+
+static double calc_correction_factor( double err_per_mb,
+ double err_devisor,
+ double pt_low,
+ double pt_high,
+ int Q )
+{
+ double power_term;
+ double error_term = err_per_mb / err_devisor;
+ double correction_factor;
+
+ /* Adjustment based on Q to power term. */
+ power_term = pt_low + (Q * 0.01);
+ power_term = (power_term > pt_high) ? pt_high : power_term;
+
+ /* Adjustments to error term */
+ /* TBD */
+
+ /* Calculate correction factor */
+ correction_factor = pow(error_term, power_term);
+
+ /* Clip range */
+ correction_factor =
+ (correction_factor < 0.05)
+ ? 0.05 : (correction_factor > 5.0) ? 5.0 : correction_factor;
+
+ return correction_factor;
+}
+
+static int estimate_max_q(VP8_COMP *cpi,
+ FIRSTPASS_STATS * fpstats,
+ int section_target_bandwitdh,
+ int overhead_bits )
+{
+ int Q;
+ int num_mbs = cpi->common.MBs;
+ int target_norm_bits_per_mb;
+
+ double section_err = (fpstats->coded_error / fpstats->count);
+ double err_per_mb = section_err / num_mbs;
+ double err_correction_factor;
+ double speed_correction = 1.0;
+ int overhead_bits_per_mb;
+
+ if (section_target_bandwitdh <= 0)
+ return cpi->twopass.maxq_max_limit; /* Highest value allowed */
+
+ target_norm_bits_per_mb =
+ (section_target_bandwitdh < (1 << 20))
+ ? (512 * section_target_bandwitdh) / num_mbs
+ : 512 * (section_target_bandwitdh / num_mbs);
+
+ /* Calculate a corrective factor based on a rolling ratio of bits spent
+ * vs target bits
+ */
+ if ((cpi->rolling_target_bits > 0) &&
+ (cpi->active_worst_quality < cpi->worst_quality))
+ {
+ double rolling_ratio;
+
+ rolling_ratio = (double)cpi->rolling_actual_bits /
+ (double)cpi->rolling_target_bits;
+
+ if (rolling_ratio < 0.95)
+ cpi->twopass.est_max_qcorrection_factor -= 0.005;
+ else if (rolling_ratio > 1.05)
+ cpi->twopass.est_max_qcorrection_factor += 0.005;
+
+ cpi->twopass.est_max_qcorrection_factor =
+ (cpi->twopass.est_max_qcorrection_factor < 0.1)
+ ? 0.1
+ : (cpi->twopass.est_max_qcorrection_factor > 10.0)
+ ? 10.0 : cpi->twopass.est_max_qcorrection_factor;
+ }
+
+ /* Corrections for higher compression speed settings
+ * (reduced compression expected)
+ */
+ if ((cpi->compressor_speed == 3) || (cpi->compressor_speed == 1))
+ {
+ if (cpi->oxcf.cpu_used <= 5)
+ speed_correction = 1.04 + (cpi->oxcf.cpu_used * 0.04);
+ else
+ speed_correction = 1.25;
+ }
+
+ /* Estimate of overhead bits per mb */
+ /* Correction to overhead bits for min allowed Q. */
+ overhead_bits_per_mb = overhead_bits / num_mbs;
+ overhead_bits_per_mb = (int)(overhead_bits_per_mb *
+ pow( 0.98, (double)cpi->twopass.maxq_min_limit ));
+
+ /* Try and pick a max Q that will be high enough to encode the
+ * content at the given rate.
+ */
+ for (Q = cpi->twopass.maxq_min_limit; Q < cpi->twopass.maxq_max_limit; Q++)
+ {
+ int bits_per_mb_at_this_q;
+
+ /* Error per MB based correction factor */
+ err_correction_factor =
+ calc_correction_factor(err_per_mb, 150.0, 0.40, 0.90, Q);
+
+ bits_per_mb_at_this_q =
+ vp8_bits_per_mb[INTER_FRAME][Q] + overhead_bits_per_mb;
+
+ bits_per_mb_at_this_q = (int)(.5 + err_correction_factor
+ * speed_correction * cpi->twopass.est_max_qcorrection_factor
+ * cpi->twopass.section_max_qfactor
+ * (double)bits_per_mb_at_this_q);
+
+ /* Mode and motion overhead */
+ /* As Q rises in real encode loop rd code will force overhead down
+ * We make a crude adjustment for this here as *.98 per Q step.
+ */
+ overhead_bits_per_mb = (int)((double)overhead_bits_per_mb * 0.98);
+
+ if (bits_per_mb_at_this_q <= target_norm_bits_per_mb)
+ break;
+ }
+
+ /* Restriction on active max q for constrained quality mode. */
+ if ( (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
+ (Q < cpi->cq_target_quality) )
+ {
+ Q = cpi->cq_target_quality;
+ }
+
+ /* Adjust maxq_min_limit and maxq_max_limit limits based on
+ * average q observed in clip for non kf/gf.arf frames
+ * Give average a chance to settle though.
+ */
+ if ( (cpi->ni_frames >
+ ((int)cpi->twopass.total_stats.count >> 8)) &&
+ (cpi->ni_frames > 150) )
+ {
+ cpi->twopass.maxq_max_limit = ((cpi->ni_av_qi + 32) < cpi->worst_quality)
+ ? (cpi->ni_av_qi + 32) : cpi->worst_quality;
+ cpi->twopass.maxq_min_limit = ((cpi->ni_av_qi - 32) > cpi->best_quality)
+ ? (cpi->ni_av_qi - 32) : cpi->best_quality;
+ }
+
+ return Q;
+}
+
+/* For cq mode estimate a cq level that matches the observed
+ * complexity and data rate.
+ */
+static int estimate_cq( VP8_COMP *cpi,
+ FIRSTPASS_STATS * fpstats,
+ int section_target_bandwitdh,
+ int overhead_bits )
+{
+ int Q;
+ int num_mbs = cpi->common.MBs;
+ int target_norm_bits_per_mb;
+
+ double section_err = (fpstats->coded_error / fpstats->count);
+ double err_per_mb = section_err / num_mbs;
+ double err_correction_factor;
+ double speed_correction = 1.0;
+ double clip_iiratio;
+ double clip_iifactor;
+ int overhead_bits_per_mb;
+
+ if (0)
+ {
+ FILE *f = fopen("epmp.stt", "a");
+ fprintf(f, "%10.2f\n", err_per_mb );
+ fclose(f);
+ }
+
+ target_norm_bits_per_mb = (section_target_bandwitdh < (1 << 20))
+ ? (512 * section_target_bandwitdh) / num_mbs
+ : 512 * (section_target_bandwitdh / num_mbs);
+
+ /* Estimate of overhead bits per mb */
+ overhead_bits_per_mb = overhead_bits / num_mbs;
+
+ /* Corrections for higher compression speed settings
+ * (reduced compression expected)
+ */
+ if ((cpi->compressor_speed == 3) || (cpi->compressor_speed == 1))
+ {
+ if (cpi->oxcf.cpu_used <= 5)
+ speed_correction = 1.04 + (cpi->oxcf.cpu_used * 0.04);
+ else
+ speed_correction = 1.25;
+ }
+
+ /* II ratio correction factor for clip as a whole */
+ clip_iiratio = cpi->twopass.total_stats.intra_error /
+ DOUBLE_DIVIDE_CHECK(cpi->twopass.total_stats.coded_error);
+ clip_iifactor = 1.0 - ((clip_iiratio - 10.0) * 0.025);
+ if (clip_iifactor < 0.80)
+ clip_iifactor = 0.80;
+
+ /* Try and pick a Q that can encode the content at the given rate. */
+ for (Q = 0; Q < MAXQ; Q++)
+ {
+ int bits_per_mb_at_this_q;
+
+ /* Error per MB based correction factor */
+ err_correction_factor =
+ calc_correction_factor(err_per_mb, 100.0, 0.40, 0.90, Q);
+
+ bits_per_mb_at_this_q =
+ vp8_bits_per_mb[INTER_FRAME][Q] + overhead_bits_per_mb;
+
+ bits_per_mb_at_this_q =
+ (int)( .5 + err_correction_factor *
+ speed_correction *
+ clip_iifactor *
+ (double)bits_per_mb_at_this_q);
+
+ /* Mode and motion overhead */
+ /* As Q rises in real encode loop rd code will force overhead down
+ * We make a crude adjustment for this here as *.98 per Q step.
+ */
+ overhead_bits_per_mb = (int)((double)overhead_bits_per_mb * 0.98);
+
+ if (bits_per_mb_at_this_q <= target_norm_bits_per_mb)
+ break;
+ }
+
+ /* Clip value to range "best allowed to (worst allowed - 1)" */
+ Q = cq_level[Q];
+ if ( Q >= cpi->worst_quality )
+ Q = cpi->worst_quality - 1;
+ if ( Q < cpi->best_quality )
+ Q = cpi->best_quality;
+
+ return Q;
+}
+
+static int estimate_q(VP8_COMP *cpi, double section_err, int section_target_bandwitdh)
+{
+ int Q;
+ int num_mbs = cpi->common.MBs;
+ int target_norm_bits_per_mb;
+
+ double err_per_mb = section_err / num_mbs;
+ double err_correction_factor;
+ double speed_correction = 1.0;
+
+ target_norm_bits_per_mb = (section_target_bandwitdh < (1 << 20)) ? (512 * section_target_bandwitdh) / num_mbs : 512 * (section_target_bandwitdh / num_mbs);
+
+ /* Corrections for higher compression speed settings
+ * (reduced compression expected)
+ */
+ if ((cpi->compressor_speed == 3) || (cpi->compressor_speed == 1))
+ {
+ if (cpi->oxcf.cpu_used <= 5)
+ speed_correction = 1.04 + (cpi->oxcf.cpu_used * 0.04);
+ else
+ speed_correction = 1.25;
+ }
+
+ /* Try and pick a Q that can encode the content at the given rate. */
+ for (Q = 0; Q < MAXQ; Q++)
+ {
+ int bits_per_mb_at_this_q;
+
+ /* Error per MB based correction factor */
+ err_correction_factor =
+ calc_correction_factor(err_per_mb, 150.0, 0.40, 0.90, Q);
+
+ bits_per_mb_at_this_q =
+ (int)( .5 + ( err_correction_factor *
+ speed_correction *
+ cpi->twopass.est_max_qcorrection_factor *
+ (double)vp8_bits_per_mb[INTER_FRAME][Q] / 1.0 ) );
+
+ if (bits_per_mb_at_this_q <= target_norm_bits_per_mb)
+ break;
+ }
+
+ return Q;
+}
+
+/* Estimate a worst case Q for a KF group */
+static int estimate_kf_group_q(VP8_COMP *cpi, double section_err, int section_target_bandwitdh, double group_iiratio)
+{
+ int Q;
+ int num_mbs = cpi->common.MBs;
+ int target_norm_bits_per_mb = (512 * section_target_bandwitdh) / num_mbs;
+ int bits_per_mb_at_this_q;
+
+ double err_per_mb = section_err / num_mbs;
+ double err_correction_factor;
+ double speed_correction = 1.0;
+ double current_spend_ratio = 1.0;
+
+ double pow_highq = (POW1 < 0.6) ? POW1 + 0.3 : 0.90;
+ double pow_lowq = (POW1 < 0.7) ? POW1 + 0.1 : 0.80;
+
+ double iiratio_correction_factor = 1.0;
+
+ double combined_correction_factor;
+
+ /* Trap special case where the target is <= 0 */
+ if (target_norm_bits_per_mb <= 0)
+ return MAXQ * 2;
+
+ /* Calculate a corrective factor based on a rolling ratio of bits spent
+ * vs target bits
+ * This is clamped to the range 0.1 to 10.0
+ */
+ if (cpi->long_rolling_target_bits <= 0)
+ current_spend_ratio = 10.0;
+ else
+ {
+ current_spend_ratio = (double)cpi->long_rolling_actual_bits / (double)cpi->long_rolling_target_bits;
+ current_spend_ratio = (current_spend_ratio > 10.0) ? 10.0 : (current_spend_ratio < 0.1) ? 0.1 : current_spend_ratio;
+ }
+
+ /* Calculate a correction factor based on the quality of prediction in
+ * the sequence as indicated by intra_inter error score ratio (IIRatio)
+ * The idea here is to favour subsampling in the hardest sections vs
+ * the easyest.
+ */
+ iiratio_correction_factor = 1.0 - ((group_iiratio - 6.0) * 0.1);
+
+ if (iiratio_correction_factor < 0.5)
+ iiratio_correction_factor = 0.5;
+
+ /* Corrections for higher compression speed settings
+ * (reduced compression expected)
+ */
+ if ((cpi->compressor_speed == 3) || (cpi->compressor_speed == 1))
+ {
+ if (cpi->oxcf.cpu_used <= 5)
+ speed_correction = 1.04 + (cpi->oxcf.cpu_used * 0.04);
+ else
+ speed_correction = 1.25;
+ }
+
+ /* Combine the various factors calculated above */
+ combined_correction_factor = speed_correction * iiratio_correction_factor * current_spend_ratio;
+
+ /* Try and pick a Q that should be high enough to encode the content at
+ * the given rate.
+ */
+ for (Q = 0; Q < MAXQ; Q++)
+ {
+ /* Error per MB based correction factor */
+ err_correction_factor =
+ calc_correction_factor(err_per_mb, 150.0, pow_lowq, pow_highq, Q);
+
+ bits_per_mb_at_this_q =
+ (int)(.5 + ( err_correction_factor *
+ combined_correction_factor *
+ (double)vp8_bits_per_mb[INTER_FRAME][Q]) );
+
+ if (bits_per_mb_at_this_q <= target_norm_bits_per_mb)
+ break;
+ }
+
+ /* If we could not hit the target even at Max Q then estimate what Q
+ * would have been required
+ */
+ while ((bits_per_mb_at_this_q > target_norm_bits_per_mb) && (Q < (MAXQ * 2)))
+ {
+
+ bits_per_mb_at_this_q = (int)(0.96 * bits_per_mb_at_this_q);
+ Q++;
+ }
+
+ if (0)
+ {
+ FILE *f = fopen("estkf_q.stt", "a");
+ fprintf(f, "%8d %8d %8d %8.2f %8.3f %8.2f %8.3f %8.3f %8.3f %8d\n", cpi->common.current_video_frame, bits_per_mb_at_this_q,
+ target_norm_bits_per_mb, err_per_mb, err_correction_factor,
+ current_spend_ratio, group_iiratio, iiratio_correction_factor,
+ (double)cpi->buffer_level / (double)cpi->oxcf.optimal_buffer_level, Q);
+ fclose(f);
+ }
+
+ return Q;
+}
+
+extern void vp8_new_frame_rate(VP8_COMP *cpi, double framerate);
+
+void vp8_init_second_pass(VP8_COMP *cpi)
+{
+ FIRSTPASS_STATS this_frame;
+ FIRSTPASS_STATS *start_pos;
+
+ double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100);
+
+ zero_stats(&cpi->twopass.total_stats);
+ zero_stats(&cpi->twopass.total_left_stats);
+
+ if (!cpi->twopass.stats_in_end)
+ return;
+
+ cpi->twopass.total_stats = *cpi->twopass.stats_in_end;
+ cpi->twopass.total_left_stats = cpi->twopass.total_stats;
+
+ /* each frame can have a different duration, as the frame rate in the
+ * source isn't guaranteed to be constant. The frame rate prior to
+ * the first frame encoded in the second pass is a guess. However the
+ * sum duration is not. Its calculated based on the actual durations of
+ * all frames from the first pass.
+ */
+ vp8_new_frame_rate(cpi, 10000000.0 * cpi->twopass.total_stats.count / cpi->twopass.total_stats.duration);
+
+ cpi->output_frame_rate = cpi->frame_rate;
+ cpi->twopass.bits_left = (int64_t)(cpi->twopass.total_stats.duration * cpi->oxcf.target_bandwidth / 10000000.0) ;
+ cpi->twopass.bits_left -= (int64_t)(cpi->twopass.total_stats.duration * two_pass_min_rate / 10000000.0);
+
+ /* Calculate a minimum intra value to be used in determining the IIratio
+ * scores used in the second pass. We have this minimum to make sure
+ * that clips that are static but "low complexity" in the intra domain
+ * are still boosted appropriately for KF/GF/ARF
+ */
+ cpi->twopass.kf_intra_err_min = KF_MB_INTRA_MIN * cpi->common.MBs;
+ cpi->twopass.gf_intra_err_min = GF_MB_INTRA_MIN * cpi->common.MBs;
+
+ /* Scan the first pass file and calculate an average Intra / Inter error
+ * score ratio for the sequence
+ */
+ {
+ double sum_iiratio = 0.0;
+ double IIRatio;
+
+ start_pos = cpi->twopass.stats_in; /* Note starting "file" position */
+
+ while (input_stats(cpi, &this_frame) != EOF)
+ {
+ IIRatio = this_frame.intra_error / DOUBLE_DIVIDE_CHECK(this_frame.coded_error);
+ IIRatio = (IIRatio < 1.0) ? 1.0 : (IIRatio > 20.0) ? 20.0 : IIRatio;
+ sum_iiratio += IIRatio;
+ }
+
+ cpi->twopass.avg_iiratio = sum_iiratio / DOUBLE_DIVIDE_CHECK((double)cpi->twopass.total_stats.count);
+
+ /* Reset file position */
+ reset_fpf_position(cpi, start_pos);
+ }
+
+ /* Scan the first pass file and calculate a modified total error based
+ * upon the bias/power function used to allocate bits
+ */
+ {
+ start_pos = cpi->twopass.stats_in; /* Note starting "file" position */
+
+ cpi->twopass.modified_error_total = 0.0;
+ cpi->twopass.modified_error_used = 0.0;
+
+ while (input_stats(cpi, &this_frame) != EOF)
+ {
+ cpi->twopass.modified_error_total += calculate_modified_err(cpi, &this_frame);
+ }
+ cpi->twopass.modified_error_left = cpi->twopass.modified_error_total;
+
+ reset_fpf_position(cpi, start_pos); /* Reset file position */
+
+ }
+}
+
+void vp8_end_second_pass(VP8_COMP *cpi)
+{
+}
+
+/* This function gives and estimate of how badly we believe the prediction
+ * quality is decaying from frame to frame.
+ */
+static double get_prediction_decay_rate(VP8_COMP *cpi, FIRSTPASS_STATS *next_frame)
+{
+ double prediction_decay_rate;
+ double motion_decay;
+ double motion_pct = next_frame->pcnt_motion;
+
+ /* Initial basis is the % mbs inter coded */
+ prediction_decay_rate = next_frame->pcnt_inter;
+
+ /* High % motion -> somewhat higher decay rate */
+ motion_decay = (1.0 - (motion_pct / 20.0));
+ if (motion_decay < prediction_decay_rate)
+ prediction_decay_rate = motion_decay;
+
+ /* Adjustment to decay rate based on speed of motion */
+ {
+ double this_mv_rabs;
+ double this_mv_cabs;
+ double distance_factor;
+
+ this_mv_rabs = fabs(next_frame->mvr_abs * motion_pct);
+ this_mv_cabs = fabs(next_frame->mvc_abs * motion_pct);
+
+ distance_factor = sqrt((this_mv_rabs * this_mv_rabs) +
+ (this_mv_cabs * this_mv_cabs)) / 250.0;
+ distance_factor = ((distance_factor > 1.0)
+ ? 0.0 : (1.0 - distance_factor));
+ if (distance_factor < prediction_decay_rate)
+ prediction_decay_rate = distance_factor;
+ }
+
+ return prediction_decay_rate;
+}
+
+/* Function to test for a condition where a complex transition is followed
+ * by a static section. For example in slide shows where there is a fade
+ * between slides. This is to help with more optimal kf and gf positioning.
+ */
+static int detect_transition_to_still(
+ VP8_COMP *cpi,
+ int frame_interval,
+ int still_interval,
+ double loop_decay_rate,
+ double decay_accumulator )
+{
+ int trans_to_still = 0;
+
+ /* Break clause to detect very still sections after motion
+ * For example a static image after a fade or other transition
+ * instead of a clean scene cut.
+ */
+ if ( (frame_interval > MIN_GF_INTERVAL) &&
+ (loop_decay_rate >= 0.999) &&
+ (decay_accumulator < 0.9) )
+ {
+ int j;
+ FIRSTPASS_STATS * position = cpi->twopass.stats_in;
+ FIRSTPASS_STATS tmp_next_frame;
+ double decay_rate;
+
+ /* Look ahead a few frames to see if static condition persists... */
+ for ( j = 0; j < still_interval; j++ )
+ {
+ if (EOF == input_stats(cpi, &tmp_next_frame))
+ break;
+
+ decay_rate = get_prediction_decay_rate(cpi, &tmp_next_frame);
+ if ( decay_rate < 0.999 )
+ break;
+ }
+ /* Reset file position */
+ reset_fpf_position(cpi, position);
+
+ /* Only if it does do we signal a transition to still */
+ if ( j == still_interval )
+ trans_to_still = 1;
+ }
+
+ return trans_to_still;
+}
+
+/* This function detects a flash through the high relative pcnt_second_ref
+ * score in the frame following a flash frame. The offset passed in should
+ * reflect this
+ */
+static int detect_flash( VP8_COMP *cpi, int offset )
+{
+ FIRSTPASS_STATS next_frame;
+
+ int flash_detected = 0;
+
+ /* Read the frame data. */
+ /* The return is 0 (no flash detected) if not a valid frame */
+ if ( read_frame_stats(cpi, &next_frame, offset) != EOF )
+ {
+ /* What we are looking for here is a situation where there is a
+ * brief break in prediction (such as a flash) but subsequent frames
+ * are reasonably well predicted by an earlier (pre flash) frame.
+ * The recovery after a flash is indicated by a high pcnt_second_ref
+ * comapred to pcnt_inter.
+ */
+ if ( (next_frame.pcnt_second_ref > next_frame.pcnt_inter) &&
+ (next_frame.pcnt_second_ref >= 0.5 ) )
+ {
+ flash_detected = 1;
+
+ /*if (1)
+ {
+ FILE *f = fopen("flash.stt", "a");
+ fprintf(f, "%8.0f %6.2f %6.2f\n",
+ next_frame.frame,
+ next_frame.pcnt_inter,
+ next_frame.pcnt_second_ref);
+ fclose(f);
+ }*/
+ }
+ }
+
+ return flash_detected;
+}
+
+/* Update the motion related elements to the GF arf boost calculation */
+static void accumulate_frame_motion_stats(
+ VP8_COMP *cpi,
+ FIRSTPASS_STATS * this_frame,
+ double * this_frame_mv_in_out,
+ double * mv_in_out_accumulator,
+ double * abs_mv_in_out_accumulator,
+ double * mv_ratio_accumulator )
+{
+ double this_frame_mvr_ratio;
+ double this_frame_mvc_ratio;
+ double motion_pct;
+
+ /* Accumulate motion stats. */
+ motion_pct = this_frame->pcnt_motion;
+
+ /* Accumulate Motion In/Out of frame stats */
+ *this_frame_mv_in_out = this_frame->mv_in_out_count * motion_pct;
+ *mv_in_out_accumulator += this_frame->mv_in_out_count * motion_pct;
+ *abs_mv_in_out_accumulator +=
+ fabs(this_frame->mv_in_out_count * motion_pct);
+
+ /* Accumulate a measure of how uniform (or conversely how random)
+ * the motion field is. (A ratio of absmv / mv)
+ */
+ if (motion_pct > 0.05)
+ {
+ this_frame_mvr_ratio = fabs(this_frame->mvr_abs) /
+ DOUBLE_DIVIDE_CHECK(fabs(this_frame->MVr));
+
+ this_frame_mvc_ratio = fabs(this_frame->mvc_abs) /
+ DOUBLE_DIVIDE_CHECK(fabs(this_frame->MVc));
+
+ *mv_ratio_accumulator +=
+ (this_frame_mvr_ratio < this_frame->mvr_abs)
+ ? (this_frame_mvr_ratio * motion_pct)
+ : this_frame->mvr_abs * motion_pct;
+
+ *mv_ratio_accumulator +=
+ (this_frame_mvc_ratio < this_frame->mvc_abs)
+ ? (this_frame_mvc_ratio * motion_pct)
+ : this_frame->mvc_abs * motion_pct;
+
+ }
+}
+
+/* Calculate a baseline boost number for the current frame. */
+static double calc_frame_boost(
+ VP8_COMP *cpi,
+ FIRSTPASS_STATS * this_frame,
+ double this_frame_mv_in_out )
+{
+ double frame_boost;
+
+ /* Underlying boost factor is based on inter intra error ratio */
+ if (this_frame->intra_error > cpi->twopass.gf_intra_err_min)
+ frame_boost = (IIFACTOR * this_frame->intra_error /
+ DOUBLE_DIVIDE_CHECK(this_frame->coded_error));
+ else
+ frame_boost = (IIFACTOR * cpi->twopass.gf_intra_err_min /
+ DOUBLE_DIVIDE_CHECK(this_frame->coded_error));
+
+ /* Increase boost for frames where new data coming into frame
+ * (eg zoom out). Slightly reduce boost if there is a net balance
+ * of motion out of the frame (zoom in).
+ * The range for this_frame_mv_in_out is -1.0 to +1.0
+ */
+ if (this_frame_mv_in_out > 0.0)
+ frame_boost += frame_boost * (this_frame_mv_in_out * 2.0);
+ /* In extreme case boost is halved */
+ else
+ frame_boost += frame_boost * (this_frame_mv_in_out / 2.0);
+
+ /* Clip to maximum */
+ if (frame_boost > GF_RMAX)
+ frame_boost = GF_RMAX;
+
+ return frame_boost;
+}
+
+#if NEW_BOOST
+static int calc_arf_boost(
+ VP8_COMP *cpi,
+ int offset,
+ int f_frames,
+ int b_frames,
+ int *f_boost,
+ int *b_boost )
+{
+ FIRSTPASS_STATS this_frame;
+
+ int i;
+ double boost_score = 0.0;
+ double mv_ratio_accumulator = 0.0;
+ double decay_accumulator = 1.0;
+ double this_frame_mv_in_out = 0.0;
+ double mv_in_out_accumulator = 0.0;
+ double abs_mv_in_out_accumulator = 0.0;
+ double r;
+ int flash_detected = 0;
+
+ /* Search forward from the proposed arf/next gf position */
+ for ( i = 0; i < f_frames; i++ )
+ {
+ if ( read_frame_stats(cpi, &this_frame, (i+offset)) == EOF )
+ break;
+
+ /* Update the motion related elements to the boost calculation */
+ accumulate_frame_motion_stats( cpi, &this_frame,
+ &this_frame_mv_in_out, &mv_in_out_accumulator,
+ &abs_mv_in_out_accumulator, &mv_ratio_accumulator );
+
+ /* Calculate the baseline boost number for this frame */
+ r = calc_frame_boost( cpi, &this_frame, this_frame_mv_in_out );
+
+ /* We want to discount the the flash frame itself and the recovery
+ * frame that follows as both will have poor scores.
+ */
+ flash_detected = detect_flash(cpi, (i+offset)) ||
+ detect_flash(cpi, (i+offset+1));
+
+ /* Cumulative effect of prediction quality decay */
+ if ( !flash_detected )
+ {
+ decay_accumulator =
+ decay_accumulator *
+ get_prediction_decay_rate(cpi, &this_frame);
+ decay_accumulator =
+ decay_accumulator < 0.1 ? 0.1 : decay_accumulator;
+ }
+ boost_score += (decay_accumulator * r);
+
+ /* Break out conditions. */
+ if ( (!flash_detected) &&
+ ((mv_ratio_accumulator > 100.0) ||
+ (abs_mv_in_out_accumulator > 3.0) ||
+ (mv_in_out_accumulator < -2.0) ) )
+ {
+ break;
+ }
+ }
+
+ *f_boost = (int)(boost_score * 100.0) >> 4;
+
+ /* Reset for backward looking loop */
+ boost_score = 0.0;
+ mv_ratio_accumulator = 0.0;
+ decay_accumulator = 1.0;
+ this_frame_mv_in_out = 0.0;
+ mv_in_out_accumulator = 0.0;
+ abs_mv_in_out_accumulator = 0.0;
+
+ /* Search forward from the proposed arf/next gf position */
+ for ( i = -1; i >= -b_frames; i-- )
+ {
+ if ( read_frame_stats(cpi, &this_frame, (i+offset)) == EOF )
+ break;
+
+ /* Update the motion related elements to the boost calculation */
+ accumulate_frame_motion_stats( cpi, &this_frame,
+ &this_frame_mv_in_out, &mv_in_out_accumulator,
+ &abs_mv_in_out_accumulator, &mv_ratio_accumulator );
+
+ /* Calculate the baseline boost number for this frame */
+ r = calc_frame_boost( cpi, &this_frame, this_frame_mv_in_out );
+
+ /* We want to discount the the flash frame itself and the recovery
+ * frame that follows as both will have poor scores.
+ */
+ flash_detected = detect_flash(cpi, (i+offset)) ||
+ detect_flash(cpi, (i+offset+1));
+
+ /* Cumulative effect of prediction quality decay */
+ if ( !flash_detected )
+ {
+ decay_accumulator =
+ decay_accumulator *
+ get_prediction_decay_rate(cpi, &this_frame);
+ decay_accumulator =
+ decay_accumulator < 0.1 ? 0.1 : decay_accumulator;
+ }
+
+ boost_score += (decay_accumulator * r);
+
+ /* Break out conditions. */
+ if ( (!flash_detected) &&
+ ((mv_ratio_accumulator > 100.0) ||
+ (abs_mv_in_out_accumulator > 3.0) ||
+ (mv_in_out_accumulator < -2.0) ) )
+ {
+ break;
+ }
+ }
+ *b_boost = (int)(boost_score * 100.0) >> 4;
+
+ return (*f_boost + *b_boost);
+}
+#endif
+
+/* Analyse and define a gf/arf group . */
+static void define_gf_group(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
+{
+ FIRSTPASS_STATS next_frame;
+ FIRSTPASS_STATS *start_pos;
+ int i;
+ double r;
+ double boost_score = 0.0;
+ double old_boost_score = 0.0;
+ double gf_group_err = 0.0;
+ double gf_first_frame_err = 0.0;
+ double mod_frame_err = 0.0;
+
+ double mv_ratio_accumulator = 0.0;
+ double decay_accumulator = 1.0;
+
+ double loop_decay_rate = 1.00; /* Starting decay rate */
+
+ double this_frame_mv_in_out = 0.0;
+ double mv_in_out_accumulator = 0.0;
+ double abs_mv_in_out_accumulator = 0.0;
+ double mod_err_per_mb_accumulator = 0.0;
+
+ int max_bits = frame_max_bits(cpi); /* Max for a single frame */
+
+ unsigned int allow_alt_ref =
+ cpi->oxcf.play_alternate && cpi->oxcf.lag_in_frames;
+
+ int alt_boost = 0;
+ int f_boost = 0;
+ int b_boost = 0;
+ int flash_detected;
+
+ cpi->twopass.gf_group_bits = 0;
+ cpi->twopass.gf_decay_rate = 0;
+
+ vp8_clear_system_state();
+
+ start_pos = cpi->twopass.stats_in;
+
+ vpx_memset(&next_frame, 0, sizeof(next_frame)); /* assure clean */
+
+ /* Load stats for the current frame. */
+ mod_frame_err = calculate_modified_err(cpi, this_frame);
+
+ /* Note the error of the frame at the start of the group (this will be
+ * the GF frame error if we code a normal gf
+ */
+ gf_first_frame_err = mod_frame_err;
+
+ /* Special treatment if the current frame is a key frame (which is also
+ * a gf). If it is then its error score (and hence bit allocation) need
+ * to be subtracted out from the calculation for the GF group
+ */
+ if (cpi->common.frame_type == KEY_FRAME)
+ gf_group_err -= gf_first_frame_err;
+
+ /* Scan forward to try and work out how many frames the next gf group
+ * should contain and what level of boost is appropriate for the GF
+ * or ARF that will be coded with the group
+ */
+ i = 0;
+
+ while (((i < cpi->twopass.static_scene_max_gf_interval) ||
+ ((cpi->twopass.frames_to_key - i) < MIN_GF_INTERVAL)) &&
+ (i < cpi->twopass.frames_to_key))
+ {
+ i++;
+
+ /* Accumulate error score of frames in this gf group */
+ mod_frame_err = calculate_modified_err(cpi, this_frame);
+
+ gf_group_err += mod_frame_err;
+
+ mod_err_per_mb_accumulator +=
+ mod_frame_err / DOUBLE_DIVIDE_CHECK((double)cpi->common.MBs);
+
+ if (EOF == input_stats(cpi, &next_frame))
+ break;
+
+ /* Test for the case where there is a brief flash but the prediction
+ * quality back to an earlier frame is then restored.
+ */
+ flash_detected = detect_flash(cpi, 0);
+
+ /* Update the motion related elements to the boost calculation */
+ accumulate_frame_motion_stats( cpi, &next_frame,
+ &this_frame_mv_in_out, &mv_in_out_accumulator,
+ &abs_mv_in_out_accumulator, &mv_ratio_accumulator );
+
+ /* Calculate a baseline boost number for this frame */
+ r = calc_frame_boost( cpi, &next_frame, this_frame_mv_in_out );
+
+ /* Cumulative effect of prediction quality decay */
+ if ( !flash_detected )
+ {
+ loop_decay_rate = get_prediction_decay_rate(cpi, &next_frame);
+ decay_accumulator = decay_accumulator * loop_decay_rate;
+ decay_accumulator =
+ decay_accumulator < 0.1 ? 0.1 : decay_accumulator;
+ }
+ boost_score += (decay_accumulator * r);
+
+ /* Break clause to detect very still sections after motion
+ * For example a staic image after a fade or other transition.
+ */
+ if ( detect_transition_to_still( cpi, i, 5,
+ loop_decay_rate,
+ decay_accumulator ) )
+ {
+ allow_alt_ref = 0;
+ boost_score = old_boost_score;
+ break;
+ }
+
+ /* Break out conditions. */
+ if (
+ /* Break at cpi->max_gf_interval unless almost totally static */
+ (i >= cpi->max_gf_interval && (decay_accumulator < 0.995)) ||
+ (
+ /* Dont break out with a very short interval */
+ (i > MIN_GF_INTERVAL) &&
+ /* Dont break out very close to a key frame */
+ ((cpi->twopass.frames_to_key - i) >= MIN_GF_INTERVAL) &&
+ ((boost_score > 20.0) || (next_frame.pcnt_inter < 0.75)) &&
+ (!flash_detected) &&
+ ((mv_ratio_accumulator > 100.0) ||
+ (abs_mv_in_out_accumulator > 3.0) ||
+ (mv_in_out_accumulator < -2.0) ||
+ ((boost_score - old_boost_score) < 2.0))
+ ) )
+ {
+ boost_score = old_boost_score;
+ break;
+ }
+
+ vpx_memcpy(this_frame, &next_frame, sizeof(*this_frame));
+
+ old_boost_score = boost_score;
+ }
+
+ cpi->twopass.gf_decay_rate =
+ (i > 0) ? (int)(100.0 * (1.0 - decay_accumulator)) / i : 0;
+
+ /* When using CBR apply additional buffer related upper limits */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ double max_boost;
+
+ /* For cbr apply buffer related limits */
+ if (cpi->drop_frames_allowed)
+ {
+ int64_t df_buffer_level = cpi->oxcf.drop_frames_water_mark *
+ (cpi->oxcf.optimal_buffer_level / 100);
+
+ if (cpi->buffer_level > df_buffer_level)
+ max_boost = ((double)((cpi->buffer_level - df_buffer_level) * 2 / 3) * 16.0) / DOUBLE_DIVIDE_CHECK((double)cpi->av_per_frame_bandwidth);
+ else
+ max_boost = 0.0;
+ }
+ else if (cpi->buffer_level > 0)
+ {
+ max_boost = ((double)(cpi->buffer_level * 2 / 3) * 16.0) / DOUBLE_DIVIDE_CHECK((double)cpi->av_per_frame_bandwidth);
+ }
+ else
+ {
+ max_boost = 0.0;
+ }
+
+ if (boost_score > max_boost)
+ boost_score = max_boost;
+ }
+
+ /* Dont allow conventional gf too near the next kf */
+ if ((cpi->twopass.frames_to_key - i) < MIN_GF_INTERVAL)
+ {
+ while (i < cpi->twopass.frames_to_key)
+ {
+ i++;
+
+ if (EOF == input_stats(cpi, this_frame))
+ break;
+
+ if (i < cpi->twopass.frames_to_key)
+ {
+ mod_frame_err = calculate_modified_err(cpi, this_frame);
+ gf_group_err += mod_frame_err;
+ }
+ }
+ }
+
+ cpi->gfu_boost = (int)(boost_score * 100.0) >> 4;
+
+#if NEW_BOOST
+ /* Alterrnative boost calculation for alt ref */
+ alt_boost = calc_arf_boost( cpi, 0, (i-1), (i-1), &f_boost, &b_boost );
+#endif
+
+ /* Should we use the alternate refernce frame */
+ if (allow_alt_ref &&
+ (i >= MIN_GF_INTERVAL) &&
+ /* dont use ARF very near next kf */
+ (i <= (cpi->twopass.frames_to_key - MIN_GF_INTERVAL)) &&
+#if NEW_BOOST
+ ((next_frame.pcnt_inter > 0.75) ||
+ (next_frame.pcnt_second_ref > 0.5)) &&
+ ((mv_in_out_accumulator / (double)i > -0.2) ||
+ (mv_in_out_accumulator > -2.0)) &&
+ (b_boost > 100) &&
+ (f_boost > 100) )
+#else
+ (next_frame.pcnt_inter > 0.75) &&
+ ((mv_in_out_accumulator / (double)i > -0.2) ||
+ (mv_in_out_accumulator > -2.0)) &&
+ (cpi->gfu_boost > 100) &&
+ (cpi->twopass.gf_decay_rate <=
+ (ARF_DECAY_THRESH + (cpi->gfu_boost / 200))) )
+#endif
+ {
+ int Boost;
+ int allocation_chunks;
+ int Q = (cpi->oxcf.fixed_q < 0)
+ ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
+ int tmp_q;
+ int arf_frame_bits = 0;
+ int group_bits;
+
+#if NEW_BOOST
+ cpi->gfu_boost = alt_boost;
+#endif
+
+ /* Estimate the bits to be allocated to the group as a whole */
+ if ((cpi->twopass.kf_group_bits > 0) &&
+ (cpi->twopass.kf_group_error_left > 0))
+ {
+ group_bits = (int)((double)cpi->twopass.kf_group_bits *
+ (gf_group_err / (double)cpi->twopass.kf_group_error_left));
+ }
+ else
+ group_bits = 0;
+
+ /* Boost for arf frame */
+#if NEW_BOOST
+ Boost = (alt_boost * GFQ_ADJUSTMENT) / 100;
+#else
+ Boost = (cpi->gfu_boost * 3 * GFQ_ADJUSTMENT) / (2 * 100);
+#endif
+ Boost += (i * 50);
+
+ /* Set max and minimum boost and hence minimum allocation */
+ if (Boost > ((cpi->baseline_gf_interval + 1) * 200))
+ Boost = ((cpi->baseline_gf_interval + 1) * 200);
+ else if (Boost < 125)
+ Boost = 125;
+
+ allocation_chunks = (i * 100) + Boost;
+
+ /* Normalize Altboost and allocations chunck down to prevent overflow */
+ while (Boost > 1000)
+ {
+ Boost /= 2;
+ allocation_chunks /= 2;
+ }
+
+ /* Calculate the number of bits to be spent on the arf based on the
+ * boost number
+ */
+ arf_frame_bits = (int)((double)Boost * (group_bits /
+ (double)allocation_chunks));
+
+ /* Estimate if there are enough bits available to make worthwhile use
+ * of an arf.
+ */
+ tmp_q = estimate_q(cpi, mod_frame_err, (int)arf_frame_bits);
+
+ /* Only use an arf if it is likely we will be able to code
+ * it at a lower Q than the surrounding frames.
+ */
+ if (tmp_q < cpi->worst_quality)
+ {
+ int half_gf_int;
+ int frames_after_arf;
+ int frames_bwd = cpi->oxcf.arnr_max_frames - 1;
+ int frames_fwd = cpi->oxcf.arnr_max_frames - 1;
+
+ cpi->source_alt_ref_pending = 1;
+
+ /*
+ * For alt ref frames the error score for the end frame of the
+ * group (the alt ref frame) should not contribute to the group
+ * total and hence the number of bit allocated to the group.
+ * Rather it forms part of the next group (it is the GF at the
+ * start of the next group)
+ * gf_group_err -= mod_frame_err;
+ *
+ * For alt ref frames alt ref frame is technically part of the
+ * GF frame for the next group but we always base the error
+ * calculation and bit allocation on the current group of frames.
+ *
+ * Set the interval till the next gf or arf.
+ * For ARFs this is the number of frames to be coded before the
+ * future frame that is coded as an ARF.
+ * The future frame itself is part of the next group
+ */
+ cpi->baseline_gf_interval = i;
+
+ /*
+ * Define the arnr filter width for this group of frames:
+ * We only filter frames that lie within a distance of half
+ * the GF interval from the ARF frame. We also have to trap
+ * cases where the filter extends beyond the end of clip.
+ * Note: this_frame->frame has been updated in the loop
+ * so it now points at the ARF frame.
+ */
+ half_gf_int = cpi->baseline_gf_interval >> 1;
+ frames_after_arf = (int)(cpi->twopass.total_stats.count -
+ this_frame->frame - 1);
+
+ switch (cpi->oxcf.arnr_type)
+ {
+ case 1: /* Backward filter */
+ frames_fwd = 0;
+ if (frames_bwd > half_gf_int)
+ frames_bwd = half_gf_int;
+ break;
+
+ case 2: /* Forward filter */
+ if (frames_fwd > half_gf_int)
+ frames_fwd = half_gf_int;
+ if (frames_fwd > frames_after_arf)
+ frames_fwd = frames_after_arf;
+ frames_bwd = 0;
+ break;
+
+ case 3: /* Centered filter */
+ default:
+ frames_fwd >>= 1;
+ if (frames_fwd > frames_after_arf)
+ frames_fwd = frames_after_arf;
+ if (frames_fwd > half_gf_int)
+ frames_fwd = half_gf_int;
+
+ frames_bwd = frames_fwd;
+
+ /* For even length filter there is one more frame backward
+ * than forward: e.g. len=6 ==> bbbAff, len=7 ==> bbbAfff.
+ */
+ if (frames_bwd < half_gf_int)
+ frames_bwd += (cpi->oxcf.arnr_max_frames+1) & 0x1;
+ break;
+ }
+
+ cpi->active_arnr_frames = frames_bwd + 1 + frames_fwd;
+ }
+ else
+ {
+ cpi->source_alt_ref_pending = 0;
+ cpi->baseline_gf_interval = i;
+ }
+ }
+ else
+ {
+ cpi->source_alt_ref_pending = 0;
+ cpi->baseline_gf_interval = i;
+ }
+
+ /*
+ * Now decide how many bits should be allocated to the GF group as a
+ * proportion of those remaining in the kf group.
+ * The final key frame group in the clip is treated as a special case
+ * where cpi->twopass.kf_group_bits is tied to cpi->twopass.bits_left.
+ * This is also important for short clips where there may only be one
+ * key frame.
+ */
+ if (cpi->twopass.frames_to_key >= (int)(cpi->twopass.total_stats.count -
+ cpi->common.current_video_frame))
+ {
+ cpi->twopass.kf_group_bits =
+ (cpi->twopass.bits_left > 0) ? cpi->twopass.bits_left : 0;
+ }
+
+ /* Calculate the bits to be allocated to the group as a whole */
+ if ((cpi->twopass.kf_group_bits > 0) &&
+ (cpi->twopass.kf_group_error_left > 0))
+ {
+ cpi->twopass.gf_group_bits =
+ (int)((double)cpi->twopass.kf_group_bits *
+ (gf_group_err / (double)cpi->twopass.kf_group_error_left));
+ }
+ else
+ cpi->twopass.gf_group_bits = 0;
+
+ cpi->twopass.gf_group_bits = (int)(
+ (cpi->twopass.gf_group_bits < 0)
+ ? 0
+ : (cpi->twopass.gf_group_bits > cpi->twopass.kf_group_bits)
+ ? cpi->twopass.kf_group_bits : cpi->twopass.gf_group_bits);
+
+ /* Clip cpi->twopass.gf_group_bits based on user supplied data rate
+ * variability limit (cpi->oxcf.two_pass_vbrmax_section)
+ */
+ if (cpi->twopass.gf_group_bits > max_bits * cpi->baseline_gf_interval)
+ cpi->twopass.gf_group_bits = max_bits * cpi->baseline_gf_interval;
+
+ /* Reset the file position */
+ reset_fpf_position(cpi, start_pos);
+
+ /* Update the record of error used so far (only done once per gf group) */
+ cpi->twopass.modified_error_used += gf_group_err;
+
+ /* Assign bits to the arf or gf. */
+ for (i = 0; i <= (cpi->source_alt_ref_pending && cpi->common.frame_type != KEY_FRAME); i++) {
+ int Boost;
+ int allocation_chunks;
+ int Q = (cpi->oxcf.fixed_q < 0) ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
+ int gf_bits;
+
+ /* For ARF frames */
+ if (cpi->source_alt_ref_pending && i == 0)
+ {
+#if NEW_BOOST
+ Boost = (alt_boost * GFQ_ADJUSTMENT) / 100;
+#else
+ Boost = (cpi->gfu_boost * 3 * GFQ_ADJUSTMENT) / (2 * 100);
+#endif
+ Boost += (cpi->baseline_gf_interval * 50);
+
+ /* Set max and minimum boost and hence minimum allocation */
+ if (Boost > ((cpi->baseline_gf_interval + 1) * 200))
+ Boost = ((cpi->baseline_gf_interval + 1) * 200);
+ else if (Boost < 125)
+ Boost = 125;
+
+ allocation_chunks =
+ ((cpi->baseline_gf_interval + 1) * 100) + Boost;
+ }
+ /* Else for standard golden frames */
+ else
+ {
+ /* boost based on inter / intra ratio of subsequent frames */
+ Boost = (cpi->gfu_boost * GFQ_ADJUSTMENT) / 100;
+
+ /* Set max and minimum boost and hence minimum allocation */
+ if (Boost > (cpi->baseline_gf_interval * 150))
+ Boost = (cpi->baseline_gf_interval * 150);
+ else if (Boost < 125)
+ Boost = 125;
+
+ allocation_chunks =
+ (cpi->baseline_gf_interval * 100) + (Boost - 100);
+ }
+
+ /* Normalize Altboost and allocations chunck down to prevent overflow */
+ while (Boost > 1000)
+ {
+ Boost /= 2;
+ allocation_chunks /= 2;
+ }
+
+ /* Calculate the number of bits to be spent on the gf or arf based on
+ * the boost number
+ */
+ gf_bits = (int)((double)Boost *
+ (cpi->twopass.gf_group_bits /
+ (double)allocation_chunks));
+
+ /* If the frame that is to be boosted is simpler than the average for
+ * the gf/arf group then use an alternative calculation
+ * based on the error score of the frame itself
+ */
+ if (mod_frame_err < gf_group_err / (double)cpi->baseline_gf_interval)
+ {
+ double alt_gf_grp_bits;
+ int alt_gf_bits;
+
+ alt_gf_grp_bits =
+ (double)cpi->twopass.kf_group_bits *
+ (mod_frame_err * (double)cpi->baseline_gf_interval) /
+ DOUBLE_DIVIDE_CHECK((double)cpi->twopass.kf_group_error_left);
+
+ alt_gf_bits = (int)((double)Boost * (alt_gf_grp_bits /
+ (double)allocation_chunks));
+
+ if (gf_bits > alt_gf_bits)
+ {
+ gf_bits = alt_gf_bits;
+ }
+ }
+ /* Else if it is harder than other frames in the group make sure it at
+ * least receives an allocation in keeping with its relative error
+ * score, otherwise it may be worse off than an "un-boosted" frame
+ */
+ else
+ {
+ int alt_gf_bits =
+ (int)((double)cpi->twopass.kf_group_bits *
+ mod_frame_err /
+ DOUBLE_DIVIDE_CHECK((double)cpi->twopass.kf_group_error_left));
+
+ if (alt_gf_bits > gf_bits)
+ {
+ gf_bits = alt_gf_bits;
+ }
+ }
+
+ /* Apply an additional limit for CBR */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ if (cpi->twopass.gf_bits > (int)(cpi->buffer_level >> 1))
+ cpi->twopass.gf_bits = (int)(cpi->buffer_level >> 1);
+ }
+
+ /* Dont allow a negative value for gf_bits */
+ if (gf_bits < 0)
+ gf_bits = 0;
+
+ /* Add in minimum for a frame */
+ gf_bits += cpi->min_frame_bandwidth;
+
+ if (i == 0)
+ {
+ cpi->twopass.gf_bits = gf_bits;
+ }
+ if (i == 1 || (!cpi->source_alt_ref_pending && (cpi->common.frame_type != KEY_FRAME)))
+ {
+ /* Per frame bit target for this frame */
+ cpi->per_frame_bandwidth = gf_bits;
+ }
+ }
+
+ {
+ /* Adjust KF group bits and error remainin */
+ cpi->twopass.kf_group_error_left -= (int64_t)gf_group_err;
+ cpi->twopass.kf_group_bits -= cpi->twopass.gf_group_bits;
+
+ if (cpi->twopass.kf_group_bits < 0)
+ cpi->twopass.kf_group_bits = 0;
+
+ /* Note the error score left in the remaining frames of the group.
+ * For normal GFs we want to remove the error score for the first
+ * frame of the group (except in Key frame case where this has
+ * already happened)
+ */
+ if (!cpi->source_alt_ref_pending && cpi->common.frame_type != KEY_FRAME)
+ cpi->twopass.gf_group_error_left = (int)(gf_group_err -
+ gf_first_frame_err);
+ else
+ cpi->twopass.gf_group_error_left = (int) gf_group_err;
+
+ cpi->twopass.gf_group_bits -= cpi->twopass.gf_bits - cpi->min_frame_bandwidth;
+
+ if (cpi->twopass.gf_group_bits < 0)
+ cpi->twopass.gf_group_bits = 0;
+
+ /* This condition could fail if there are two kfs very close together
+ * despite (MIN_GF_INTERVAL) and would cause a devide by 0 in the
+ * calculation of cpi->twopass.alt_extra_bits.
+ */
+ if ( cpi->baseline_gf_interval >= 3 )
+ {
+#if NEW_BOOST
+ int boost = (cpi->source_alt_ref_pending)
+ ? b_boost : cpi->gfu_boost;
+#else
+ int boost = cpi->gfu_boost;
+#endif
+ if ( boost >= 150 )
+ {
+ int pct_extra;
+
+ pct_extra = (boost - 100) / 50;
+ pct_extra = (pct_extra > 20) ? 20 : pct_extra;
+
+ cpi->twopass.alt_extra_bits =
+ (cpi->twopass.gf_group_bits * pct_extra) / 100;
+ cpi->twopass.gf_group_bits -= cpi->twopass.alt_extra_bits;
+ cpi->twopass.alt_extra_bits /=
+ ((cpi->baseline_gf_interval-1)>>1);
+ }
+ else
+ cpi->twopass.alt_extra_bits = 0;
+ }
+ else
+ cpi->twopass.alt_extra_bits = 0;
+ }
+
+ /* Adjustments based on a measure of complexity of the section */
+ if (cpi->common.frame_type != KEY_FRAME)
+ {
+ FIRSTPASS_STATS sectionstats;
+ double Ratio;
+
+ zero_stats(§ionstats);
+ reset_fpf_position(cpi, start_pos);
+
+ for (i = 0 ; i < cpi->baseline_gf_interval ; i++)
+ {
+ input_stats(cpi, &next_frame);
+ accumulate_stats(§ionstats, &next_frame);
+ }
+
+ avg_stats(§ionstats);
+
+ cpi->twopass.section_intra_rating = (unsigned int)
+ (sectionstats.intra_error /
+ DOUBLE_DIVIDE_CHECK(sectionstats.coded_error));
+
+ Ratio = sectionstats.intra_error / DOUBLE_DIVIDE_CHECK(sectionstats.coded_error);
+ cpi->twopass.section_max_qfactor = 1.0 - ((Ratio - 10.0) * 0.025);
+
+ if (cpi->twopass.section_max_qfactor < 0.80)
+ cpi->twopass.section_max_qfactor = 0.80;
+
+ reset_fpf_position(cpi, start_pos);
+ }
+}
+
+/* Allocate bits to a normal frame that is neither a gf an arf or a key frame. */
+static void assign_std_frame_bits(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
+{
+ int target_frame_size;
+
+ double modified_err;
+ double err_fraction;
+
+ int max_bits = frame_max_bits(cpi); /* Max for a single frame */
+
+ /* Calculate modified prediction error used in bit allocation */
+ modified_err = calculate_modified_err(cpi, this_frame);
+
+ /* What portion of the remaining GF group error is used by this frame */
+ if (cpi->twopass.gf_group_error_left > 0)
+ err_fraction = modified_err / cpi->twopass.gf_group_error_left;
+ else
+ err_fraction = 0.0;
+
+ /* How many of those bits available for allocation should we give it? */
+ target_frame_size = (int)((double)cpi->twopass.gf_group_bits * err_fraction);
+
+ /* Clip to target size to 0 - max_bits (or cpi->twopass.gf_group_bits)
+ * at the top end.
+ */
+ if (target_frame_size < 0)
+ target_frame_size = 0;
+ else
+ {
+ if (target_frame_size > max_bits)
+ target_frame_size = max_bits;
+
+ if (target_frame_size > cpi->twopass.gf_group_bits)
+ target_frame_size = cpi->twopass.gf_group_bits;
+ }
+
+ /* Adjust error and bits remaining */
+ cpi->twopass.gf_group_error_left -= (int)modified_err;
+ cpi->twopass.gf_group_bits -= target_frame_size;
+
+ if (cpi->twopass.gf_group_bits < 0)
+ cpi->twopass.gf_group_bits = 0;
+
+ /* Add in the minimum number of bits that is set aside for every frame. */
+ target_frame_size += cpi->min_frame_bandwidth;
+
+ /* Every other frame gets a few extra bits */
+ if ( (cpi->common.frames_since_golden & 0x01) &&
+ (cpi->frames_till_gf_update_due > 0) )
+ {
+ target_frame_size += cpi->twopass.alt_extra_bits;
+ }
+
+ /* Per frame bit target for this frame */
+ cpi->per_frame_bandwidth = target_frame_size;
+}
+
+void vp8_second_pass(VP8_COMP *cpi)
+{
+ int tmp_q;
+ int frames_left = (int)(cpi->twopass.total_stats.count - cpi->common.current_video_frame);
+
+ FIRSTPASS_STATS this_frame = {0};
+ FIRSTPASS_STATS this_frame_copy;
+
+ double this_frame_intra_error;
+ double this_frame_coded_error;
+
+ int overhead_bits;
+
+ if (!cpi->twopass.stats_in)
+ {
+ return ;
+ }
+
+ vp8_clear_system_state();
+
+ if (EOF == input_stats(cpi, &this_frame))
+ return;
+
+ this_frame_intra_error = this_frame.intra_error;
+ this_frame_coded_error = this_frame.coded_error;
+
+ /* keyframe and section processing ! */
+ if (cpi->twopass.frames_to_key == 0)
+ {
+ /* Define next KF group and assign bits to it */
+ vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
+ find_next_key_frame(cpi, &this_frame_copy);
+
+ /* Special case: Error error_resilient_mode mode does not make much
+ * sense for two pass but with its current meaning but this code is
+ * designed to stop outlandish behaviour if someone does set it when
+ * using two pass. It effectively disables GF groups. This is
+ * temporary code till we decide what should really happen in this
+ * case.
+ */
+ if (cpi->oxcf.error_resilient_mode)
+ {
+ cpi->twopass.gf_group_bits = (int)cpi->twopass.kf_group_bits;
+ cpi->twopass.gf_group_error_left =
+ (int)cpi->twopass.kf_group_error_left;
+ cpi->baseline_gf_interval = cpi->twopass.frames_to_key;
+ cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
+ cpi->source_alt_ref_pending = 0;
+ }
+
+ }
+
+ /* Is this a GF / ARF (Note that a KF is always also a GF) */
+ if (cpi->frames_till_gf_update_due == 0)
+ {
+ /* Define next gf group and assign bits to it */
+ vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
+ define_gf_group(cpi, &this_frame_copy);
+
+ /* If we are going to code an altref frame at the end of the group
+ * and the current frame is not a key frame.... If the previous
+ * group used an arf this frame has already benefited from that arf
+ * boost and it should not be given extra bits If the previous
+ * group was NOT coded using arf we may want to apply some boost to
+ * this GF as well
+ */
+ if (cpi->source_alt_ref_pending && (cpi->common.frame_type != KEY_FRAME))
+ {
+ /* Assign a standard frames worth of bits from those allocated
+ * to the GF group
+ */
+ int bak = cpi->per_frame_bandwidth;
+ vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
+ assign_std_frame_bits(cpi, &this_frame_copy);
+ cpi->per_frame_bandwidth = bak;
+ }
+ }
+
+ /* Otherwise this is an ordinary frame */
+ else
+ {
+ /* Special case: Error error_resilient_mode mode does not make much
+ * sense for two pass but with its current meaning but this code is
+ * designed to stop outlandish behaviour if someone does set it
+ * when using two pass. It effectively disables GF groups. This is
+ * temporary code till we decide what should really happen in this
+ * case.
+ */
+ if (cpi->oxcf.error_resilient_mode)
+ {
+ cpi->frames_till_gf_update_due = cpi->twopass.frames_to_key;
+
+ if (cpi->common.frame_type != KEY_FRAME)
+ {
+ /* Assign bits from those allocated to the GF group */
+ vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
+ assign_std_frame_bits(cpi, &this_frame_copy);
+ }
+ }
+ else
+ {
+ /* Assign bits from those allocated to the GF group */
+ vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
+ assign_std_frame_bits(cpi, &this_frame_copy);
+ }
+ }
+
+ /* Keep a globally available copy of this and the next frame's iiratio. */
+ cpi->twopass.this_iiratio = (unsigned int)(this_frame_intra_error /
+ DOUBLE_DIVIDE_CHECK(this_frame_coded_error));
+ {
+ FIRSTPASS_STATS next_frame;
+ if ( lookup_next_frame_stats(cpi, &next_frame) != EOF )
+ {
+ cpi->twopass.next_iiratio = (unsigned int)(next_frame.intra_error /
+ DOUBLE_DIVIDE_CHECK(next_frame.coded_error));
+ }
+ }
+
+ /* Set nominal per second bandwidth for this frame */
+ cpi->target_bandwidth = (int)
+ (cpi->per_frame_bandwidth * cpi->output_frame_rate);
+ if (cpi->target_bandwidth < 0)
+ cpi->target_bandwidth = 0;
+
+
+ /* Account for mv, mode and other overheads. */
+ overhead_bits = (int)estimate_modemvcost(
+ cpi, &cpi->twopass.total_left_stats );
+
+ /* Special case code for first frame. */
+ if (cpi->common.current_video_frame == 0)
+ {
+ cpi->twopass.est_max_qcorrection_factor = 1.0;
+
+ /* Set a cq_level in constrained quality mode. */
+ if ( cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY )
+ {
+ int est_cq;
+
+ est_cq =
+ estimate_cq( cpi,
+ &cpi->twopass.total_left_stats,
+ (int)(cpi->twopass.bits_left / frames_left),
+ overhead_bits );
+
+ cpi->cq_target_quality = cpi->oxcf.cq_level;
+ if ( est_cq > cpi->cq_target_quality )
+ cpi->cq_target_quality = est_cq;
+ }
+
+ /* guess at maxq needed in 2nd pass */
+ cpi->twopass.maxq_max_limit = cpi->worst_quality;
+ cpi->twopass.maxq_min_limit = cpi->best_quality;
+
+ tmp_q = estimate_max_q(
+ cpi,
+ &cpi->twopass.total_left_stats,
+ (int)(cpi->twopass.bits_left / frames_left),
+ overhead_bits );
+
+ /* Limit the maxq value returned subsequently.
+ * This increases the risk of overspend or underspend if the initial
+ * estimate for the clip is bad, but helps prevent excessive
+ * variation in Q, especially near the end of a clip
+ * where for example a small overspend may cause Q to crash
+ */
+ cpi->twopass.maxq_max_limit = ((tmp_q + 32) < cpi->worst_quality)
+ ? (tmp_q + 32) : cpi->worst_quality;
+ cpi->twopass.maxq_min_limit = ((tmp_q - 32) > cpi->best_quality)
+ ? (tmp_q - 32) : cpi->best_quality;
+
+ cpi->active_worst_quality = tmp_q;
+ cpi->ni_av_qi = tmp_q;
+ }
+
+ /* The last few frames of a clip almost always have to few or too many
+ * bits and for the sake of over exact rate control we dont want to make
+ * radical adjustments to the allowed quantizer range just to use up a
+ * few surplus bits or get beneath the target rate.
+ */
+ else if ( (cpi->common.current_video_frame <
+ (((unsigned int)cpi->twopass.total_stats.count * 255)>>8)) &&
+ ((cpi->common.current_video_frame + cpi->baseline_gf_interval) <
+ (unsigned int)cpi->twopass.total_stats.count) )
+ {
+ if (frames_left < 1)
+ frames_left = 1;
+
+ tmp_q = estimate_max_q(
+ cpi,
+ &cpi->twopass.total_left_stats,
+ (int)(cpi->twopass.bits_left / frames_left),
+ overhead_bits );
+
+ /* Move active_worst_quality but in a damped way */
+ if (tmp_q > cpi->active_worst_quality)
+ cpi->active_worst_quality ++;
+ else if (tmp_q < cpi->active_worst_quality)
+ cpi->active_worst_quality --;
+
+ cpi->active_worst_quality =
+ ((cpi->active_worst_quality * 3) + tmp_q + 2) / 4;
+ }
+
+ cpi->twopass.frames_to_key --;
+
+ /* Update the total stats remaining sturcture */
+ subtract_stats(&cpi->twopass.total_left_stats, &this_frame );
+}
+
+
+static int test_candidate_kf(VP8_COMP *cpi, FIRSTPASS_STATS *last_frame, FIRSTPASS_STATS *this_frame, FIRSTPASS_STATS *next_frame)
+{
+ int is_viable_kf = 0;
+
+ /* Does the frame satisfy the primary criteria of a key frame
+ * If so, then examine how well it predicts subsequent frames
+ */
+ if ((this_frame->pcnt_second_ref < 0.10) &&
+ (next_frame->pcnt_second_ref < 0.10) &&
+ ((this_frame->pcnt_inter < 0.05) ||
+ (
+ ((this_frame->pcnt_inter - this_frame->pcnt_neutral) < .25) &&
+ ((this_frame->intra_error / DOUBLE_DIVIDE_CHECK(this_frame->coded_error)) < 2.5) &&
+ ((fabs(last_frame->coded_error - this_frame->coded_error) / DOUBLE_DIVIDE_CHECK(this_frame->coded_error) > .40) ||
+ (fabs(last_frame->intra_error - this_frame->intra_error) / DOUBLE_DIVIDE_CHECK(this_frame->intra_error) > .40) ||
+ ((next_frame->intra_error / DOUBLE_DIVIDE_CHECK(next_frame->coded_error)) > 3.5)
+ )
+ )
+ )
+ )
+ {
+ int i;
+ FIRSTPASS_STATS *start_pos;
+
+ FIRSTPASS_STATS local_next_frame;
+
+ double boost_score = 0.0;
+ double old_boost_score = 0.0;
+ double decay_accumulator = 1.0;
+ double next_iiratio;
+
+ vpx_memcpy(&local_next_frame, next_frame, sizeof(*next_frame));
+
+ /* Note the starting file position so we can reset to it */
+ start_pos = cpi->twopass.stats_in;
+
+ /* Examine how well the key frame predicts subsequent frames */
+ for (i = 0 ; i < 16; i++)
+ {
+ next_iiratio = (IIKFACTOR1 * local_next_frame.intra_error / DOUBLE_DIVIDE_CHECK(local_next_frame.coded_error)) ;
+
+ if (next_iiratio > RMAX)
+ next_iiratio = RMAX;
+
+ /* Cumulative effect of decay in prediction quality */
+ if (local_next_frame.pcnt_inter > 0.85)
+ decay_accumulator = decay_accumulator * local_next_frame.pcnt_inter;
+ else
+ decay_accumulator = decay_accumulator * ((0.85 + local_next_frame.pcnt_inter) / 2.0);
+
+ /* Keep a running total */
+ boost_score += (decay_accumulator * next_iiratio);
+
+ /* Test various breakout clauses */
+ if ((local_next_frame.pcnt_inter < 0.05) ||
+ (next_iiratio < 1.5) ||
+ (((local_next_frame.pcnt_inter -
+ local_next_frame.pcnt_neutral) < 0.20) &&
+ (next_iiratio < 3.0)) ||
+ ((boost_score - old_boost_score) < 0.5) ||
+ (local_next_frame.intra_error < 200)
+ )
+ {
+ break;
+ }
+
+ old_boost_score = boost_score;
+
+ /* Get the next frame details */
+ if (EOF == input_stats(cpi, &local_next_frame))
+ break;
+ }
+
+ /* If there is tolerable prediction for at least the next 3 frames
+ * then break out else discard this pottential key frame and move on
+ */
+ if (boost_score > 5.0 && (i > 3))
+ is_viable_kf = 1;
+ else
+ {
+ /* Reset the file position */
+ reset_fpf_position(cpi, start_pos);
+
+ is_viable_kf = 0;
+ }
+ }
+
+ return is_viable_kf;
+}
+static void find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
+{
+ int i,j;
+ FIRSTPASS_STATS last_frame;
+ FIRSTPASS_STATS first_frame;
+ FIRSTPASS_STATS next_frame;
+ FIRSTPASS_STATS *start_position;
+
+ double decay_accumulator = 1.0;
+ double boost_score = 0;
+ double old_boost_score = 0.0;
+ double loop_decay_rate;
+
+ double kf_mod_err = 0.0;
+ double kf_group_err = 0.0;
+ double kf_group_intra_err = 0.0;
+ double kf_group_coded_err = 0.0;
+ double recent_loop_decay[8] = {1.0,1.0,1.0,1.0,1.0,1.0,1.0,1.0};
+
+ vpx_memset(&next_frame, 0, sizeof(next_frame));
+
+ vp8_clear_system_state();
+ start_position = cpi->twopass.stats_in;
+
+ cpi->common.frame_type = KEY_FRAME;
+
+ /* is this a forced key frame by interval */
+ cpi->this_key_frame_forced = cpi->next_key_frame_forced;
+
+ /* Clear the alt ref active flag as this can never be active on a key
+ * frame
+ */
+ cpi->source_alt_ref_active = 0;
+
+ /* Kf is always a gf so clear frames till next gf counter */
+ cpi->frames_till_gf_update_due = 0;
+
+ cpi->twopass.frames_to_key = 1;
+
+ /* Take a copy of the initial frame details */
+ vpx_memcpy(&first_frame, this_frame, sizeof(*this_frame));
+
+ cpi->twopass.kf_group_bits = 0;
+ cpi->twopass.kf_group_error_left = 0;
+
+ kf_mod_err = calculate_modified_err(cpi, this_frame);
+
+ /* find the next keyframe */
+ i = 0;
+ while (cpi->twopass.stats_in < cpi->twopass.stats_in_end)
+ {
+ /* Accumulate kf group error */
+ kf_group_err += calculate_modified_err(cpi, this_frame);
+
+ /* These figures keep intra and coded error counts for all frames
+ * including key frames in the group. The effect of the key frame
+ * itself can be subtracted out using the first_frame data
+ * collected above
+ */
+ kf_group_intra_err += this_frame->intra_error;
+ kf_group_coded_err += this_frame->coded_error;
+
+ /* load a the next frame's stats */
+ vpx_memcpy(&last_frame, this_frame, sizeof(*this_frame));
+ input_stats(cpi, this_frame);
+
+ /* Provided that we are not at the end of the file... */
+ if (cpi->oxcf.auto_key
+ && lookup_next_frame_stats(cpi, &next_frame) != EOF)
+ {
+ /* Normal scene cut check */
+ if ( ( i >= MIN_GF_INTERVAL ) &&
+ test_candidate_kf(cpi, &last_frame, this_frame, &next_frame) )
+ {
+ break;
+ }
+
+ /* How fast is prediction quality decaying */
+ loop_decay_rate = get_prediction_decay_rate(cpi, &next_frame);
+
+ /* We want to know something about the recent past... rather than
+ * as used elsewhere where we are concened with decay in prediction
+ * quality since the last GF or KF.
+ */
+ recent_loop_decay[i%8] = loop_decay_rate;
+ decay_accumulator = 1.0;
+ for (j = 0; j < 8; j++)
+ {
+ decay_accumulator = decay_accumulator * recent_loop_decay[j];
+ }
+
+ /* Special check for transition or high motion followed by a
+ * static scene.
+ */
+ if ( detect_transition_to_still( cpi, i,
+ (cpi->key_frame_frequency-i),
+ loop_decay_rate,
+ decay_accumulator ) )
+ {
+ break;
+ }
+
+
+ /* Step on to the next frame */
+ cpi->twopass.frames_to_key ++;
+
+ /* If we don't have a real key frame within the next two
+ * forcekeyframeevery intervals then break out of the loop.
+ */
+ if (cpi->twopass.frames_to_key >= 2 *(int)cpi->key_frame_frequency)
+ break;
+ } else
+ cpi->twopass.frames_to_key ++;
+
+ i++;
+ }
+
+ /* If there is a max kf interval set by the user we must obey it.
+ * We already breakout of the loop above at 2x max.
+ * This code centers the extra kf if the actual natural
+ * interval is between 1x and 2x
+ */
+ if (cpi->oxcf.auto_key
+ && cpi->twopass.frames_to_key > (int)cpi->key_frame_frequency )
+ {
+ FIRSTPASS_STATS *current_pos = cpi->twopass.stats_in;
+ FIRSTPASS_STATS tmp_frame;
+
+ cpi->twopass.frames_to_key /= 2;
+
+ /* Copy first frame details */
+ vpx_memcpy(&tmp_frame, &first_frame, sizeof(first_frame));
+
+ /* Reset to the start of the group */
+ reset_fpf_position(cpi, start_position);
+
+ kf_group_err = 0;
+ kf_group_intra_err = 0;
+ kf_group_coded_err = 0;
+
+ /* Rescan to get the correct error data for the forced kf group */
+ for( i = 0; i < cpi->twopass.frames_to_key; i++ )
+ {
+ /* Accumulate kf group errors */
+ kf_group_err += calculate_modified_err(cpi, &tmp_frame);
+ kf_group_intra_err += tmp_frame.intra_error;
+ kf_group_coded_err += tmp_frame.coded_error;
+
+ /* Load a the next frame's stats */
+ input_stats(cpi, &tmp_frame);
+ }
+
+ /* Reset to the start of the group */
+ reset_fpf_position(cpi, current_pos);
+
+ cpi->next_key_frame_forced = 1;
+ }
+ else
+ cpi->next_key_frame_forced = 0;
+
+ /* Special case for the last frame of the file */
+ if (cpi->twopass.stats_in >= cpi->twopass.stats_in_end)
+ {
+ /* Accumulate kf group error */
+ kf_group_err += calculate_modified_err(cpi, this_frame);
+
+ /* These figures keep intra and coded error counts for all frames
+ * including key frames in the group. The effect of the key frame
+ * itself can be subtracted out using the first_frame data
+ * collected above
+ */
+ kf_group_intra_err += this_frame->intra_error;
+ kf_group_coded_err += this_frame->coded_error;
+ }
+
+ /* Calculate the number of bits that should be assigned to the kf group. */
+ if ((cpi->twopass.bits_left > 0) && (cpi->twopass.modified_error_left > 0.0))
+ {
+ /* Max for a single normal frame (not key frame) */
+ int max_bits = frame_max_bits(cpi);
+
+ /* Maximum bits for the kf group */
+ int64_t max_grp_bits;
+
+ /* Default allocation based on bits left and relative
+ * complexity of the section
+ */
+ cpi->twopass.kf_group_bits = (int64_t)( cpi->twopass.bits_left *
+ ( kf_group_err /
+ cpi->twopass.modified_error_left ));
+
+ /* Clip based on maximum per frame rate defined by the user. */
+ max_grp_bits = (int64_t)max_bits * (int64_t)cpi->twopass.frames_to_key;
+ if (cpi->twopass.kf_group_bits > max_grp_bits)
+ cpi->twopass.kf_group_bits = max_grp_bits;
+
+ /* Additional special case for CBR if buffer is getting full. */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ int64_t opt_buffer_lvl = cpi->oxcf.optimal_buffer_level;
+ int64_t buffer_lvl = cpi->buffer_level;
+
+ /* If the buffer is near or above the optimal and this kf group is
+ * not being allocated much then increase the allocation a bit.
+ */
+ if (buffer_lvl >= opt_buffer_lvl)
+ {
+ int64_t high_water_mark = (opt_buffer_lvl +
+ cpi->oxcf.maximum_buffer_size) >> 1;
+
+ int64_t av_group_bits;
+
+ /* Av bits per frame * number of frames */
+ av_group_bits = (int64_t)cpi->av_per_frame_bandwidth *
+ (int64_t)cpi->twopass.frames_to_key;
+
+ /* We are at or above the maximum. */
+ if (cpi->buffer_level >= high_water_mark)
+ {
+ int64_t min_group_bits;
+
+ min_group_bits = av_group_bits +
+ (int64_t)(buffer_lvl -
+ high_water_mark);
+
+ if (cpi->twopass.kf_group_bits < min_group_bits)
+ cpi->twopass.kf_group_bits = min_group_bits;
+ }
+ /* We are above optimal but below the maximum */
+ else if (cpi->twopass.kf_group_bits < av_group_bits)
+ {
+ int64_t bits_below_av = av_group_bits -
+ cpi->twopass.kf_group_bits;
+
+ cpi->twopass.kf_group_bits +=
+ (int64_t)((double)bits_below_av *
+ (double)(buffer_lvl - opt_buffer_lvl) /
+ (double)(high_water_mark - opt_buffer_lvl));
+ }
+ }
+ }
+ }
+ else
+ cpi->twopass.kf_group_bits = 0;
+
+ /* Reset the first pass file position */
+ reset_fpf_position(cpi, start_position);
+
+ /* determine how big to make this keyframe based on how well the
+ * subsequent frames use inter blocks
+ */
+ decay_accumulator = 1.0;
+ boost_score = 0.0;
+ loop_decay_rate = 1.00; /* Starting decay rate */
+
+ for (i = 0 ; i < cpi->twopass.frames_to_key ; i++)
+ {
+ double r;
+
+ if (EOF == input_stats(cpi, &next_frame))
+ break;
+
+ if (next_frame.intra_error > cpi->twopass.kf_intra_err_min)
+ r = (IIKFACTOR2 * next_frame.intra_error /
+ DOUBLE_DIVIDE_CHECK(next_frame.coded_error));
+ else
+ r = (IIKFACTOR2 * cpi->twopass.kf_intra_err_min /
+ DOUBLE_DIVIDE_CHECK(next_frame.coded_error));
+
+ if (r > RMAX)
+ r = RMAX;
+
+ /* How fast is prediction quality decaying */
+ loop_decay_rate = get_prediction_decay_rate(cpi, &next_frame);
+
+ decay_accumulator = decay_accumulator * loop_decay_rate;
+ decay_accumulator = decay_accumulator < 0.1 ? 0.1 : decay_accumulator;
+
+ boost_score += (decay_accumulator * r);
+
+ if ((i > MIN_GF_INTERVAL) &&
+ ((boost_score - old_boost_score) < 1.0))
+ {
+ break;
+ }
+
+ old_boost_score = boost_score;
+ }
+
+ if (1)
+ {
+ FIRSTPASS_STATS sectionstats;
+ double Ratio;
+
+ zero_stats(§ionstats);
+ reset_fpf_position(cpi, start_position);
+
+ for (i = 0 ; i < cpi->twopass.frames_to_key ; i++)
+ {
+ input_stats(cpi, &next_frame);
+ accumulate_stats(§ionstats, &next_frame);
+ }
+
+ avg_stats(§ionstats);
+
+ cpi->twopass.section_intra_rating = (unsigned int)
+ (sectionstats.intra_error
+ / DOUBLE_DIVIDE_CHECK(sectionstats.coded_error));
+
+ Ratio = sectionstats.intra_error / DOUBLE_DIVIDE_CHECK(sectionstats.coded_error);
+ cpi->twopass.section_max_qfactor = 1.0 - ((Ratio - 10.0) * 0.025);
+
+ if (cpi->twopass.section_max_qfactor < 0.80)
+ cpi->twopass.section_max_qfactor = 0.80;
+ }
+
+ /* When using CBR apply additional buffer fullness related upper limits */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ double max_boost;
+
+ if (cpi->drop_frames_allowed)
+ {
+ int df_buffer_level = (int)(cpi->oxcf.drop_frames_water_mark
+ * (cpi->oxcf.optimal_buffer_level / 100));
+
+ if (cpi->buffer_level > df_buffer_level)
+ max_boost = ((double)((cpi->buffer_level - df_buffer_level) * 2 / 3) * 16.0) / DOUBLE_DIVIDE_CHECK((double)cpi->av_per_frame_bandwidth);
+ else
+ max_boost = 0.0;
+ }
+ else if (cpi->buffer_level > 0)
+ {
+ max_boost = ((double)(cpi->buffer_level * 2 / 3) * 16.0) / DOUBLE_DIVIDE_CHECK((double)cpi->av_per_frame_bandwidth);
+ }
+ else
+ {
+ max_boost = 0.0;
+ }
+
+ if (boost_score > max_boost)
+ boost_score = max_boost;
+ }
+
+ /* Reset the first pass file position */
+ reset_fpf_position(cpi, start_position);
+
+ /* Work out how many bits to allocate for the key frame itself */
+ if (1)
+ {
+ int kf_boost = (int)boost_score;
+ int allocation_chunks;
+ int Counter = cpi->twopass.frames_to_key;
+ int alt_kf_bits;
+ YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx];
+ /* Min boost based on kf interval */
+#if 0
+
+ while ((kf_boost < 48) && (Counter > 0))
+ {
+ Counter -= 2;
+ kf_boost ++;
+ }
+
+#endif
+
+ if (kf_boost < 48)
+ {
+ kf_boost += ((Counter + 1) >> 1);
+
+ if (kf_boost > 48) kf_boost = 48;
+ }
+
+ /* bigger frame sizes need larger kf boosts, smaller frames smaller
+ * boosts...
+ */
+ if ((lst_yv12->y_width * lst_yv12->y_height) > (320 * 240))
+ kf_boost += 2 * (lst_yv12->y_width * lst_yv12->y_height) / (320 * 240);
+ else if ((lst_yv12->y_width * lst_yv12->y_height) < (320 * 240))
+ kf_boost -= 4 * (320 * 240) / (lst_yv12->y_width * lst_yv12->y_height);
+
+ /* Min KF boost */
+ kf_boost = (int)((double)kf_boost * 100.0) >> 4; /* Scale 16 to 100 */
+ if (kf_boost < 250)
+ kf_boost = 250;
+
+ /*
+ * We do three calculations for kf size.
+ * The first is based on the error score for the whole kf group.
+ * The second (optionaly) on the key frames own error if this is
+ * smaller than the average for the group.
+ * The final one insures that the frame receives at least the
+ * allocation it would have received based on its own error score vs
+ * the error score remaining
+ * Special case if the sequence appears almost totaly static
+ * as measured by the decay accumulator. In this case we want to
+ * spend almost all of the bits on the key frame.
+ * cpi->twopass.frames_to_key-1 because key frame itself is taken
+ * care of by kf_boost.
+ */
+ if ( decay_accumulator >= 0.99 )
+ {
+ allocation_chunks =
+ ((cpi->twopass.frames_to_key - 1) * 10) + kf_boost;
+ }
+ else
+ {
+ allocation_chunks =
+ ((cpi->twopass.frames_to_key - 1) * 100) + kf_boost;
+ }
+
+ /* Normalize Altboost and allocations chunck down to prevent overflow */
+ while (kf_boost > 1000)
+ {
+ kf_boost /= 2;
+ allocation_chunks /= 2;
+ }
+
+ cpi->twopass.kf_group_bits = (cpi->twopass.kf_group_bits < 0) ? 0 : cpi->twopass.kf_group_bits;
+
+ /* Calculate the number of bits to be spent on the key frame */
+ cpi->twopass.kf_bits = (int)((double)kf_boost * ((double)cpi->twopass.kf_group_bits / (double)allocation_chunks));
+
+ /* Apply an additional limit for CBR */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ if (cpi->twopass.kf_bits > (int)((3 * cpi->buffer_level) >> 2))
+ cpi->twopass.kf_bits = (int)((3 * cpi->buffer_level) >> 2);
+ }
+
+ /* If the key frame is actually easier than the average for the
+ * kf group (which does sometimes happen... eg a blank intro frame)
+ * Then use an alternate calculation based on the kf error score
+ * which should give a smaller key frame.
+ */
+ if (kf_mod_err < kf_group_err / cpi->twopass.frames_to_key)
+ {
+ double alt_kf_grp_bits =
+ ((double)cpi->twopass.bits_left *
+ (kf_mod_err * (double)cpi->twopass.frames_to_key) /
+ DOUBLE_DIVIDE_CHECK(cpi->twopass.modified_error_left));
+
+ alt_kf_bits = (int)((double)kf_boost *
+ (alt_kf_grp_bits / (double)allocation_chunks));
+
+ if (cpi->twopass.kf_bits > alt_kf_bits)
+ {
+ cpi->twopass.kf_bits = alt_kf_bits;
+ }
+ }
+ /* Else if it is much harder than other frames in the group make sure
+ * it at least receives an allocation in keeping with its relative
+ * error score
+ */
+ else
+ {
+ alt_kf_bits =
+ (int)((double)cpi->twopass.bits_left *
+ (kf_mod_err /
+ DOUBLE_DIVIDE_CHECK(cpi->twopass.modified_error_left)));
+
+ if (alt_kf_bits > cpi->twopass.kf_bits)
+ {
+ cpi->twopass.kf_bits = alt_kf_bits;
+ }
+ }
+
+ cpi->twopass.kf_group_bits -= cpi->twopass.kf_bits;
+ /* Add in the minimum frame allowance */
+ cpi->twopass.kf_bits += cpi->min_frame_bandwidth;
+
+ /* Peer frame bit target for this frame */
+ cpi->per_frame_bandwidth = cpi->twopass.kf_bits;
+
+ /* Convert to a per second bitrate */
+ cpi->target_bandwidth = (int)(cpi->twopass.kf_bits *
+ cpi->output_frame_rate);
+ }
+
+ /* Note the total error score of the kf group minus the key frame itself */
+ cpi->twopass.kf_group_error_left = (int)(kf_group_err - kf_mod_err);
+
+ /* Adjust the count of total modified error left. The count of bits left
+ * is adjusted elsewhere based on real coded frame sizes
+ */
+ cpi->twopass.modified_error_left -= kf_group_err;
+
+ if (cpi->oxcf.allow_spatial_resampling)
+ {
+ int resample_trigger = 0;
+ int last_kf_resampled = 0;
+ int kf_q;
+ int scale_val = 0;
+ int hr, hs, vr, vs;
+ int new_width = cpi->oxcf.Width;
+ int new_height = cpi->oxcf.Height;
+
+ int projected_buffer_level = (int)cpi->buffer_level;
+ int tmp_q;
+
+ double projected_bits_perframe;
+ double group_iiratio = (kf_group_intra_err - first_frame.intra_error) / (kf_group_coded_err - first_frame.coded_error);
+ double err_per_frame = kf_group_err / cpi->twopass.frames_to_key;
+ double bits_per_frame;
+ double av_bits_per_frame;
+ double effective_size_ratio;
+
+ if ((cpi->common.Width != cpi->oxcf.Width) || (cpi->common.Height != cpi->oxcf.Height))
+ last_kf_resampled = 1;
+
+ /* Set back to unscaled by defaults */
+ cpi->common.horiz_scale = NORMAL;
+ cpi->common.vert_scale = NORMAL;
+
+ /* Calculate Average bits per frame. */
+ av_bits_per_frame = cpi->oxcf.target_bandwidth / DOUBLE_DIVIDE_CHECK((double)cpi->frame_rate);
+
+ /* CBR... Use the clip average as the target for deciding resample */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ bits_per_frame = av_bits_per_frame;
+ }
+
+ /* In VBR we want to avoid downsampling in easy section unless we
+ * are under extreme pressure So use the larger of target bitrate
+ * for this section or average bitrate for sequence
+ */
+ else
+ {
+ /* This accounts for how hard the section is... */
+ bits_per_frame = (double)
+ (cpi->twopass.kf_group_bits / cpi->twopass.frames_to_key);
+
+ /* Dont turn to resampling in easy sections just because they
+ * have been assigned a small number of bits
+ */
+ if (bits_per_frame < av_bits_per_frame)
+ bits_per_frame = av_bits_per_frame;
+ }
+
+ /* bits_per_frame should comply with our minimum */
+ if (bits_per_frame < (cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100))
+ bits_per_frame = (cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100);
+
+ /* Work out if spatial resampling is necessary */
+ kf_q = estimate_kf_group_q(cpi, err_per_frame,
+ (int)bits_per_frame, group_iiratio);
+
+ /* If we project a required Q higher than the maximum allowed Q then
+ * make a guess at the actual size of frames in this section
+ */
+ projected_bits_perframe = bits_per_frame;
+ tmp_q = kf_q;
+
+ while (tmp_q > cpi->worst_quality)
+ {
+ projected_bits_perframe *= 1.04;
+ tmp_q--;
+ }
+
+ /* Guess at buffer level at the end of the section */
+ projected_buffer_level = (int)
+ (cpi->buffer_level - (int)
+ ((projected_bits_perframe - av_bits_per_frame) *
+ cpi->twopass.frames_to_key));
+
+ if (0)
+ {
+ FILE *f = fopen("Subsamle.stt", "a");
+ fprintf(f, " %8d %8d %8d %8d %12.0f %8d %8d %8d\n", cpi->common.current_video_frame, kf_q, cpi->common.horiz_scale, cpi->common.vert_scale, kf_group_err / cpi->twopass.frames_to_key, (int)(cpi->twopass.kf_group_bits / cpi->twopass.frames_to_key), new_height, new_width);
+ fclose(f);
+ }
+
+ /* The trigger for spatial resampling depends on the various
+ * parameters such as whether we are streaming (CBR) or VBR.
+ */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ /* Trigger resample if we are projected to fall below down
+ * sample level or resampled last time and are projected to
+ * remain below the up sample level
+ */
+ if ((projected_buffer_level < (cpi->oxcf.resample_down_water_mark * cpi->oxcf.optimal_buffer_level / 100)) ||
+ (last_kf_resampled && (projected_buffer_level < (cpi->oxcf.resample_up_water_mark * cpi->oxcf.optimal_buffer_level / 100))))
+ resample_trigger = 1;
+ else
+ resample_trigger = 0;
+ }
+ else
+ {
+ int64_t clip_bits = (int64_t)(cpi->twopass.total_stats.count * cpi->oxcf.target_bandwidth / DOUBLE_DIVIDE_CHECK((double)cpi->frame_rate));
+ int64_t over_spend = cpi->oxcf.starting_buffer_level - cpi->buffer_level;
+
+ /* If triggered last time the threshold for triggering again is
+ * reduced:
+ *
+ * Projected Q higher than allowed and Overspend > 5% of total
+ * bits
+ */
+ if ((last_kf_resampled && (kf_q > cpi->worst_quality)) ||
+ ((kf_q > cpi->worst_quality) &&
+ (over_spend > clip_bits / 20)))
+ resample_trigger = 1;
+ else
+ resample_trigger = 0;
+
+ }
+
+ if (resample_trigger)
+ {
+ while ((kf_q >= cpi->worst_quality) && (scale_val < 6))
+ {
+ scale_val ++;
+
+ cpi->common.vert_scale = vscale_lookup[scale_val];
+ cpi->common.horiz_scale = hscale_lookup[scale_val];
+
+ Scale2Ratio(cpi->common.horiz_scale, &hr, &hs);
+ Scale2Ratio(cpi->common.vert_scale, &vr, &vs);
+
+ new_width = ((hs - 1) + (cpi->oxcf.Width * hr)) / hs;
+ new_height = ((vs - 1) + (cpi->oxcf.Height * vr)) / vs;
+
+ /* Reducing the area to 1/4 does not reduce the complexity
+ * (err_per_frame) to 1/4... effective_sizeratio attempts
+ * to provide a crude correction for this
+ */
+ effective_size_ratio = (double)(new_width * new_height) / (double)(cpi->oxcf.Width * cpi->oxcf.Height);
+ effective_size_ratio = (1.0 + (3.0 * effective_size_ratio)) / 4.0;
+
+ /* Now try again and see what Q we get with the smaller
+ * image size
+ */
+ kf_q = estimate_kf_group_q(cpi,
+ err_per_frame * effective_size_ratio,
+ (int)bits_per_frame, group_iiratio);
+
+ if (0)
+ {
+ FILE *f = fopen("Subsamle.stt", "a");
+ fprintf(f, "******** %8d %8d %8d %12.0f %8d %8d %8d\n", kf_q, cpi->common.horiz_scale, cpi->common.vert_scale, kf_group_err / cpi->twopass.frames_to_key, (int)(cpi->twopass.kf_group_bits / cpi->twopass.frames_to_key), new_height, new_width);
+ fclose(f);
+ }
+ }
+ }
+
+ if ((cpi->common.Width != new_width) || (cpi->common.Height != new_height))
+ {
+ cpi->common.Width = new_width;
+ cpi->common.Height = new_height;
+ vp8_alloc_compressor_data(cpi);
+ }
+ }
+}
diff --git a/vp8/encoder/firstpass.h b/libvpx/vp8/encoder/firstpass.h
similarity index 100%
rename from vp8/encoder/firstpass.h
rename to libvpx/vp8/encoder/firstpass.h
diff --git a/libvpx/vp8/encoder/lookahead.c b/libvpx/vp8/encoder/lookahead.c
new file mode 100644
index 0000000..ce2ce08
--- /dev/null
+++ b/libvpx/vp8/encoder/lookahead.c
@@ -0,0 +1,230 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#include <assert.h>
+#include <stdlib.h>
+#include "vpx_config.h"
+#include "lookahead.h"
+#include "vp8/common/extend.h"
+
+#define MAX_LAG_BUFFERS (CONFIG_REALTIME_ONLY? 1 : 25)
+
+struct lookahead_ctx
+{
+ unsigned int max_sz; /* Absolute size of the queue */
+ unsigned int sz; /* Number of buffers currently in the queue */
+ unsigned int read_idx; /* Read index */
+ unsigned int write_idx; /* Write index */
+ struct lookahead_entry *buf; /* Buffer list */
+};
+
+
+/* Return the buffer at the given absolute index and increment the index */
+static struct lookahead_entry *
+pop(struct lookahead_ctx *ctx,
+ unsigned int *idx)
+{
+ unsigned int index = *idx;
+ struct lookahead_entry *buf = ctx->buf + index;
+
+ assert(index < ctx->max_sz);
+ if(++index >= ctx->max_sz)
+ index -= ctx->max_sz;
+ *idx = index;
+ return buf;
+}
+
+
+void
+vp8_lookahead_destroy(struct lookahead_ctx *ctx)
+{
+ if(ctx)
+ {
+ if(ctx->buf)
+ {
+ unsigned int i;
+
+ for(i = 0; i < ctx->max_sz; i++)
+ vp8_yv12_de_alloc_frame_buffer(&ctx->buf[i].img);
+ free(ctx->buf);
+ }
+ free(ctx);
+ }
+}
+
+
+struct lookahead_ctx*
+vp8_lookahead_init(unsigned int width,
+ unsigned int height,
+ unsigned int depth)
+{
+ struct lookahead_ctx *ctx = NULL;
+ unsigned int i;
+
+ /* Clamp the lookahead queue depth */
+ if(depth < 1)
+ depth = 1;
+ else if(depth > MAX_LAG_BUFFERS)
+ depth = MAX_LAG_BUFFERS;
+
+ /* Keep last frame in lookahead buffer by increasing depth by 1.*/
+ depth += 1;
+
+ /* Align the buffer dimensions */
+ width = (width + 15) & ~15;
+ height = (height + 15) & ~15;
+
+ /* Allocate the lookahead structures */
+ ctx = calloc(1, sizeof(*ctx));
+ if(ctx)
+ {
+ ctx->max_sz = depth;
+ ctx->buf = calloc(depth, sizeof(*ctx->buf));
+ if(!ctx->buf)
+ goto bail;
+ for(i=0; i<depth; i++)
+ if (vp8_yv12_alloc_frame_buffer(&ctx->buf[i].img,
+ width, height, VP8BORDERINPIXELS))
+ goto bail;
+ }
+ return ctx;
+bail:
+ vp8_lookahead_destroy(ctx);
+ return NULL;
+}
+
+
+int
+vp8_lookahead_push(struct lookahead_ctx *ctx,
+ YV12_BUFFER_CONFIG *src,
+ int64_t ts_start,
+ int64_t ts_end,
+ unsigned int flags,
+ unsigned char *active_map)
+{
+ struct lookahead_entry* buf;
+ int row, col, active_end;
+ int mb_rows = (src->y_height + 15) >> 4;
+ int mb_cols = (src->y_width + 15) >> 4;
+
+ if(ctx->sz + 2 > ctx->max_sz)
+ return 1;
+ ctx->sz++;
+ buf = pop(ctx, &ctx->write_idx);
+
+ /* Only do this partial copy if the following conditions are all met:
+ * 1. Lookahead queue has has size of 1.
+ * 2. Active map is provided.
+ * 3. This is not a key frame, golden nor altref frame.
+ */
+ if (ctx->max_sz == 1 && active_map && !flags)
+ {
+ for (row = 0; row < mb_rows; ++row)
+ {
+ col = 0;
+
+ while (1)
+ {
+ /* Find the first active macroblock in this row. */
+ for (; col < mb_cols; ++col)
+ {
+ if (active_map[col])
+ break;
+ }
+
+ /* No more active macroblock in this row. */
+ if (col == mb_cols)
+ break;
+
+ /* Find the end of active region in this row. */
+ active_end = col;
+
+ for (; active_end < mb_cols; ++active_end)
+ {
+ if (!active_map[active_end])
+ break;
+ }
+
+ /* Only copy this active region. */
+ vp8_copy_and_extend_frame_with_rect(src, &buf->img,
+ row << 4,
+ col << 4, 16,
+ (active_end - col) << 4);
+
+ /* Start again from the end of this active region. */
+ col = active_end;
+ }
+
+ active_map += mb_cols;
+ }
+ }
+ else
+ {
+ vp8_copy_and_extend_frame(src, &buf->img);
+ }
+ buf->ts_start = ts_start;
+ buf->ts_end = ts_end;
+ buf->flags = flags;
+ return 0;
+}
+
+
+struct lookahead_entry*
+vp8_lookahead_pop(struct lookahead_ctx *ctx,
+ int drain)
+{
+ struct lookahead_entry* buf = NULL;
+
+ if(ctx->sz && (drain || ctx->sz == ctx->max_sz - 1))
+ {
+ buf = pop(ctx, &ctx->read_idx);
+ ctx->sz--;
+ }
+ return buf;
+}
+
+
+struct lookahead_entry*
+vp8_lookahead_peek(struct lookahead_ctx *ctx,
+ unsigned int index,
+ int direction)
+{
+ struct lookahead_entry* buf = NULL;
+
+ if (direction == PEEK_FORWARD)
+ {
+ assert(index < ctx->max_sz - 1);
+ if(index < ctx->sz)
+ {
+ index += ctx->read_idx;
+ if(index >= ctx->max_sz)
+ index -= ctx->max_sz;
+ buf = ctx->buf + index;
+ }
+ }
+ else if (direction == PEEK_BACKWARD)
+ {
+ assert(index == 1);
+
+ if(ctx->read_idx == 0)
+ index = ctx->max_sz - 1;
+ else
+ index = ctx->read_idx - index;
+ buf = ctx->buf + index;
+ }
+
+ return buf;
+}
+
+
+unsigned int
+vp8_lookahead_depth(struct lookahead_ctx *ctx)
+{
+ return ctx->sz;
+}
diff --git a/libvpx/vp8/encoder/lookahead.h b/libvpx/vp8/encoder/lookahead.h
new file mode 100644
index 0000000..cf56b75
--- /dev/null
+++ b/libvpx/vp8/encoder/lookahead.h
@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+#ifndef LOOKAHEAD_H
+#define LOOKAHEAD_H
+#include "vpx_scale/yv12config.h"
+#include "vpx/vpx_integer.h"
+
+struct lookahead_entry
+{
+ YV12_BUFFER_CONFIG img;
+ int64_t ts_start;
+ int64_t ts_end;
+ unsigned int flags;
+};
+
+
+struct lookahead_ctx;
+
+/**\brief Initializes the lookahead stage
+ *
+ * The lookahead stage is a queue of frame buffers on which some analysis
+ * may be done when buffers are enqueued.
+ *
+ *
+ */
+struct lookahead_ctx* vp8_lookahead_init(unsigned int width,
+ unsigned int height,
+ unsigned int depth
+ );
+
+
+/**\brief Destroys the lookahead stage
+ *
+ */
+void vp8_lookahead_destroy(struct lookahead_ctx *ctx);
+
+
+/**\brief Enqueue a source buffer
+ *
+ * This function will copy the source image into a new framebuffer with
+ * the expected stride/border.
+ *
+ * If active_map is non-NULL and there is only one frame in the queue, then copy
+ * only active macroblocks.
+ *
+ * \param[in] ctx Pointer to the lookahead context
+ * \param[in] src Pointer to the image to enqueue
+ * \param[in] ts_start Timestamp for the start of this frame
+ * \param[in] ts_end Timestamp for the end of this frame
+ * \param[in] flags Flags set on this frame
+ * \param[in] active_map Map that specifies which macroblock is active
+ */
+int
+vp8_lookahead_push(struct lookahead_ctx *ctx,
+ YV12_BUFFER_CONFIG *src,
+ int64_t ts_start,
+ int64_t ts_end,
+ unsigned int flags,
+ unsigned char *active_map);
+
+
+/**\brief Get the next source buffer to encode
+ *
+ *
+ * \param[in] ctx Pointer to the lookahead context
+ * \param[in] drain Flag indicating the buffer should be drained
+ * (return a buffer regardless of the current queue depth)
+ *
+ * \retval NULL, if drain set and queue is empty
+ * \retval NULL, if drain not set and queue not of the configured depth
+ *
+ */
+struct lookahead_entry*
+vp8_lookahead_pop(struct lookahead_ctx *ctx,
+ int drain);
+
+
+#define PEEK_FORWARD 1
+#define PEEK_BACKWARD -1
+/**\brief Get a future source buffer to encode
+ *
+ * \param[in] ctx Pointer to the lookahead context
+ * \param[in] index Index of the frame to be returned, 0 == next frame
+ *
+ * \retval NULL, if no buffer exists at the specified index
+ *
+ */
+struct lookahead_entry*
+vp8_lookahead_peek(struct lookahead_ctx *ctx,
+ unsigned int index,
+ int direction);
+
+
+/**\brief Get the number of frames currently in the lookahead queue
+ *
+ * \param[in] ctx Pointer to the lookahead context
+ */
+unsigned int
+vp8_lookahead_depth(struct lookahead_ctx *ctx);
+
+
+#endif
diff --git a/libvpx/vp8/encoder/mcomp.c b/libvpx/vp8/encoder/mcomp.c
new file mode 100644
index 0000000..b08c7a5
--- /dev/null
+++ b/libvpx/vp8/encoder/mcomp.c
@@ -0,0 +1,2026 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "onyx_int.h"
+#include "mcomp.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vpx_config.h"
+#include <stdio.h>
+#include <limits.h>
+#include <math.h>
+#include "vp8/common/findnearmv.h"
+
+#ifdef ENTROPY_STATS
+static int mv_ref_ct [31] [4] [2];
+static int mv_mode_cts [4] [2];
+#endif
+
+int vp8_mv_bit_cost(int_mv *mv, int_mv *ref, int *mvcost[2], int Weight)
+{
+ /* MV costing is based on the distribution of vectors in the previous
+ * frame and as such will tend to over state the cost of vectors. In
+ * addition coding a new vector can have a knock on effect on the cost
+ * of subsequent vectors and the quality of prediction from NEAR and
+ * NEAREST for subsequent blocks. The "Weight" parameter allows, to a
+ * limited extent, for some account to be taken of these factors.
+ */
+ return ((mvcost[0][(mv->as_mv.row - ref->as_mv.row) >> 1] + mvcost[1][(mv->as_mv.col - ref->as_mv.col) >> 1]) * Weight) >> 7;
+}
+
+static int mv_err_cost(int_mv *mv, int_mv *ref, int *mvcost[2], int error_per_bit)
+{
+ /* Ignore mv costing if mvcost is NULL */
+ if (mvcost)
+ return ((mvcost[0][(mv->as_mv.row - ref->as_mv.row) >> 1] +
+ mvcost[1][(mv->as_mv.col - ref->as_mv.col) >> 1])
+ * error_per_bit + 128) >> 8;
+ return 0;
+}
+
+static int mvsad_err_cost(int_mv *mv, int_mv *ref, int *mvsadcost[2], int error_per_bit)
+{
+ /* Calculate sad error cost on full pixel basis. */
+ /* Ignore mv costing if mvsadcost is NULL */
+ if (mvsadcost)
+ return ((mvsadcost[0][(mv->as_mv.row - ref->as_mv.row)] +
+ mvsadcost[1][(mv->as_mv.col - ref->as_mv.col)])
+ * error_per_bit + 128) >> 8;
+ return 0;
+}
+
+void vp8_init_dsmotion_compensation(MACROBLOCK *x, int stride)
+{
+ int Len;
+ int search_site_count = 0;
+
+
+ /* Generate offsets for 4 search sites per step. */
+ Len = MAX_FIRST_STEP;
+ x->ss[search_site_count].mv.col = 0;
+ x->ss[search_site_count].mv.row = 0;
+ x->ss[search_site_count].offset = 0;
+ search_site_count++;
+
+ while (Len > 0)
+ {
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = 0;
+ x->ss[search_site_count].mv.row = -Len;
+ x->ss[search_site_count].offset = -Len * stride;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = 0;
+ x->ss[search_site_count].mv.row = Len;
+ x->ss[search_site_count].offset = Len * stride;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = -Len;
+ x->ss[search_site_count].mv.row = 0;
+ x->ss[search_site_count].offset = -Len;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = Len;
+ x->ss[search_site_count].mv.row = 0;
+ x->ss[search_site_count].offset = Len;
+ search_site_count++;
+
+ /* Contract. */
+ Len /= 2;
+ }
+
+ x->ss_count = search_site_count;
+ x->searches_per_step = 4;
+}
+
+void vp8_init3smotion_compensation(MACROBLOCK *x, int stride)
+{
+ int Len;
+ int search_site_count = 0;
+
+ /* Generate offsets for 8 search sites per step. */
+ Len = MAX_FIRST_STEP;
+ x->ss[search_site_count].mv.col = 0;
+ x->ss[search_site_count].mv.row = 0;
+ x->ss[search_site_count].offset = 0;
+ search_site_count++;
+
+ while (Len > 0)
+ {
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = 0;
+ x->ss[search_site_count].mv.row = -Len;
+ x->ss[search_site_count].offset = -Len * stride;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = 0;
+ x->ss[search_site_count].mv.row = Len;
+ x->ss[search_site_count].offset = Len * stride;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = -Len;
+ x->ss[search_site_count].mv.row = 0;
+ x->ss[search_site_count].offset = -Len;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = Len;
+ x->ss[search_site_count].mv.row = 0;
+ x->ss[search_site_count].offset = Len;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = -Len;
+ x->ss[search_site_count].mv.row = -Len;
+ x->ss[search_site_count].offset = -Len * stride - Len;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = Len;
+ x->ss[search_site_count].mv.row = -Len;
+ x->ss[search_site_count].offset = -Len * stride + Len;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = -Len;
+ x->ss[search_site_count].mv.row = Len;
+ x->ss[search_site_count].offset = Len * stride - Len;
+ search_site_count++;
+
+ /* Compute offsets for search sites. */
+ x->ss[search_site_count].mv.col = Len;
+ x->ss[search_site_count].mv.row = Len;
+ x->ss[search_site_count].offset = Len * stride + Len;
+ search_site_count++;
+
+
+ /* Contract. */
+ Len /= 2;
+ }
+
+ x->ss_count = search_site_count;
+ x->searches_per_step = 8;
+}
+
+/*
+ * To avoid the penalty for crossing cache-line read, preload the reference
+ * area in a small buffer, which is aligned to make sure there won't be crossing
+ * cache-line read while reading from this buffer. This reduced the cpu
+ * cycles spent on reading ref data in sub-pixel filter functions.
+ * TODO: Currently, since sub-pixel search range here is -3 ~ 3, copy 22 rows x
+ * 32 cols area that is enough for 16x16 macroblock. Later, for SPLITMV, we
+ * could reduce the area.
+ */
+
+/* estimated cost of a motion vector (r,c) */
+#define MVC(r,c) (mvcost ? ((mvcost[0][(r)-rr] + mvcost[1][(c) - rc]) * error_per_bit + 128 )>>8 : 0)
+/* pointer to predictor base of a motionvector */
+#define PRE(r,c) (y + (((r)>>2) * y_stride + ((c)>>2) -(offset)))
+/* convert motion vector component to offset for svf calc */
+#define SP(x) (((x)&3)<<1)
+/* returns subpixel variance error function. */
+#define DIST(r,c) vfp->svf( PRE(r,c), y_stride, SP(c),SP(r), z,b->src_stride,&sse)
+#define IFMVCV(r,c,s,e) if ( c >= minc && c <= maxc && r >= minr && r <= maxr) s else e;
+/* returns distortion + motion vector cost */
+#define ERR(r,c) (MVC(r,c)+DIST(r,c))
+/* checks if (r,c) has better score than previous best */
+#define CHECK_BETTER(v,r,c) IFMVCV(r,c,{thismse = DIST(r,c); if((v = (MVC(r,c)+thismse)) < besterr) { besterr = v; br=r; bc=c; *distortion = thismse; *sse1 = sse; }}, v=UINT_MAX;)
+
+int vp8_find_best_sub_pixel_step_iteratively(MACROBLOCK *x, BLOCK *b, BLOCKD *d,
+ int_mv *bestmv, int_mv *ref_mv,
+ int error_per_bit,
+ const vp8_variance_fn_ptr_t *vfp,
+ int *mvcost[2], int *distortion,
+ unsigned int *sse1)
+{
+ unsigned char *z = (*(b->base_src) + b->src);
+
+ int rr = ref_mv->as_mv.row >> 1, rc = ref_mv->as_mv.col >> 1;
+ int br = bestmv->as_mv.row << 2, bc = bestmv->as_mv.col << 2;
+ int tr = br, tc = bc;
+ unsigned int besterr;
+ unsigned int left, right, up, down, diag;
+ unsigned int sse;
+ unsigned int whichdir;
+ unsigned int halfiters = 4;
+ unsigned int quarteriters = 4;
+ int thismse;
+
+ int minc = MAX(x->mv_col_min << 2, (ref_mv->as_mv.col >> 1) - ((1 << mvlong_width) - 1));
+ int maxc = MIN(x->mv_col_max << 2, (ref_mv->as_mv.col >> 1) + ((1 << mvlong_width) - 1));
+ int minr = MAX(x->mv_row_min << 2, (ref_mv->as_mv.row >> 1) - ((1 << mvlong_width) - 1));
+ int maxr = MIN(x->mv_row_max << 2, (ref_mv->as_mv.row >> 1) + ((1 << mvlong_width) - 1));
+
+ int y_stride;
+ int offset;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+
+
+#if ARCH_X86 || ARCH_X86_64
+ MACROBLOCKD *xd = &x->e_mbd;
+ unsigned char *y0 = base_pre + d->offset + (bestmv->as_mv.row) * pre_stride + bestmv->as_mv.col;
+ unsigned char *y;
+ int buf_r1, buf_r2, buf_c1, buf_c2;
+
+ /* Clamping to avoid out-of-range data access */
+ buf_r1 = ((bestmv->as_mv.row - 3) < x->mv_row_min)?(bestmv->as_mv.row - x->mv_row_min):3;
+ buf_r2 = ((bestmv->as_mv.row + 3) > x->mv_row_max)?(x->mv_row_max - bestmv->as_mv.row):3;
+ buf_c1 = ((bestmv->as_mv.col - 3) < x->mv_col_min)?(bestmv->as_mv.col - x->mv_col_min):3;
+ buf_c2 = ((bestmv->as_mv.col + 3) > x->mv_col_max)?(x->mv_col_max - bestmv->as_mv.col):3;
+ y_stride = 32;
+
+ /* Copy to intermediate buffer before searching. */
+ vfp->copymem(y0 - buf_c1 - pre_stride*buf_r1, pre_stride, xd->y_buf, y_stride, 16+buf_r1+buf_r2);
+ y = xd->y_buf + y_stride*buf_r1 +buf_c1;
+#else
+ unsigned char *y = base_pre + d->offset + (bestmv->as_mv.row) * pre_stride + bestmv->as_mv.col;
+ y_stride = pre_stride;
+#endif
+
+ offset = (bestmv->as_mv.row) * y_stride + bestmv->as_mv.col;
+
+ /* central mv */
+ bestmv->as_mv.row <<= 3;
+ bestmv->as_mv.col <<= 3;
+
+ /* calculate central point error */
+ besterr = vfp->vf(y, y_stride, z, b->src_stride, sse1);
+ *distortion = besterr;
+ besterr += mv_err_cost(bestmv, ref_mv, mvcost, error_per_bit);
+
+ /* TODO: Each subsequent iteration checks at least one point in common
+ * with the last iteration could be 2 ( if diag selected)
+ */
+ while (--halfiters)
+ {
+ /* 1/2 pel */
+ CHECK_BETTER(left, tr, tc - 2);
+ CHECK_BETTER(right, tr, tc + 2);
+ CHECK_BETTER(up, tr - 2, tc);
+ CHECK_BETTER(down, tr + 2, tc);
+
+ whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
+
+ switch (whichdir)
+ {
+ case 0:
+ CHECK_BETTER(diag, tr - 2, tc - 2);
+ break;
+ case 1:
+ CHECK_BETTER(diag, tr - 2, tc + 2);
+ break;
+ case 2:
+ CHECK_BETTER(diag, tr + 2, tc - 2);
+ break;
+ case 3:
+ CHECK_BETTER(diag, tr + 2, tc + 2);
+ break;
+ }
+
+ /* no reason to check the same one again. */
+ if (tr == br && tc == bc)
+ break;
+
+ tr = br;
+ tc = bc;
+ }
+
+ /* TODO: Each subsequent iteration checks at least one point in common
+ * with the last iteration could be 2 ( if diag selected)
+ */
+
+ /* 1/4 pel */
+ while (--quarteriters)
+ {
+ CHECK_BETTER(left, tr, tc - 1);
+ CHECK_BETTER(right, tr, tc + 1);
+ CHECK_BETTER(up, tr - 1, tc);
+ CHECK_BETTER(down, tr + 1, tc);
+
+ whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
+
+ switch (whichdir)
+ {
+ case 0:
+ CHECK_BETTER(diag, tr - 1, tc - 1);
+ break;
+ case 1:
+ CHECK_BETTER(diag, tr - 1, tc + 1);
+ break;
+ case 2:
+ CHECK_BETTER(diag, tr + 1, tc - 1);
+ break;
+ case 3:
+ CHECK_BETTER(diag, tr + 1, tc + 1);
+ break;
+ }
+
+ /* no reason to check the same one again. */
+ if (tr == br && tc == bc)
+ break;
+
+ tr = br;
+ tc = bc;
+ }
+
+ bestmv->as_mv.row = br << 1;
+ bestmv->as_mv.col = bc << 1;
+
+ if ((abs(bestmv->as_mv.col - ref_mv->as_mv.col) > (MAX_FULL_PEL_VAL<<3)) ||
+ (abs(bestmv->as_mv.row - ref_mv->as_mv.row) > (MAX_FULL_PEL_VAL<<3)))
+ return INT_MAX;
+
+ return besterr;
+}
+#undef MVC
+#undef PRE
+#undef SP
+#undef DIST
+#undef IFMVCV
+#undef ERR
+#undef CHECK_BETTER
+
+int vp8_find_best_sub_pixel_step(MACROBLOCK *x, BLOCK *b, BLOCKD *d,
+ int_mv *bestmv, int_mv *ref_mv,
+ int error_per_bit,
+ const vp8_variance_fn_ptr_t *vfp,
+ int *mvcost[2], int *distortion,
+ unsigned int *sse1)
+{
+ int bestmse = INT_MAX;
+ int_mv startmv;
+ int_mv this_mv;
+ unsigned char *z = (*(b->base_src) + b->src);
+ int left, right, up, down, diag;
+ unsigned int sse;
+ int whichdir ;
+ int thismse;
+ int y_stride;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+
+#if ARCH_X86 || ARCH_X86_64
+ MACROBLOCKD *xd = &x->e_mbd;
+ unsigned char *y0 = base_pre + d->offset + (bestmv->as_mv.row) * pre_stride + bestmv->as_mv.col;
+ unsigned char *y;
+
+ y_stride = 32;
+ /* Copy 18 rows x 32 cols area to intermediate buffer before searching. */
+ vfp->copymem(y0 - 1 - pre_stride, pre_stride, xd->y_buf, y_stride, 18);
+ y = xd->y_buf + y_stride + 1;
+#else
+ unsigned char *y = base_pre + d->offset + (bestmv->as_mv.row) * pre_stride + bestmv->as_mv.col;
+ y_stride = pre_stride;
+#endif
+
+ /* central mv */
+ bestmv->as_mv.row <<= 3;
+ bestmv->as_mv.col <<= 3;
+ startmv = *bestmv;
+
+ /* calculate central point error */
+ bestmse = vfp->vf(y, y_stride, z, b->src_stride, sse1);
+ *distortion = bestmse;
+ bestmse += mv_err_cost(bestmv, ref_mv, mvcost, error_per_bit);
+
+ /* go left then right and check error */
+ this_mv.as_mv.row = startmv.as_mv.row;
+ this_mv.as_mv.col = ((startmv.as_mv.col - 8) | 4);
+ thismse = vfp->svf_halfpix_h(y - 1, y_stride, z, b->src_stride, &sse);
+ left = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (left < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = left;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ this_mv.as_mv.col += 8;
+ thismse = vfp->svf_halfpix_h(y, y_stride, z, b->src_stride, &sse);
+ right = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (right < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = right;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ /* go up then down and check error */
+ this_mv.as_mv.col = startmv.as_mv.col;
+ this_mv.as_mv.row = ((startmv.as_mv.row - 8) | 4);
+ thismse = vfp->svf_halfpix_v(y - y_stride, y_stride, z, b->src_stride, &sse);
+ up = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (up < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = up;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ this_mv.as_mv.row += 8;
+ thismse = vfp->svf_halfpix_v(y, y_stride, z, b->src_stride, &sse);
+ down = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (down < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = down;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+
+ /* now check 1 more diagonal */
+ whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
+ this_mv = startmv;
+
+ switch (whichdir)
+ {
+ case 0:
+ this_mv.as_mv.col = (this_mv.as_mv.col - 8) | 4;
+ this_mv.as_mv.row = (this_mv.as_mv.row - 8) | 4;
+ thismse = vfp->svf_halfpix_hv(y - 1 - y_stride, y_stride, z, b->src_stride, &sse);
+ break;
+ case 1:
+ this_mv.as_mv.col += 4;
+ this_mv.as_mv.row = (this_mv.as_mv.row - 8) | 4;
+ thismse = vfp->svf_halfpix_hv(y - y_stride, y_stride, z, b->src_stride, &sse);
+ break;
+ case 2:
+ this_mv.as_mv.col = (this_mv.as_mv.col - 8) | 4;
+ this_mv.as_mv.row += 4;
+ thismse = vfp->svf_halfpix_hv(y - 1, y_stride, z, b->src_stride, &sse);
+ break;
+ case 3:
+ default:
+ this_mv.as_mv.col += 4;
+ this_mv.as_mv.row += 4;
+ thismse = vfp->svf_halfpix_hv(y, y_stride, z, b->src_stride, &sse);
+ break;
+ }
+
+ diag = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (diag < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = diag;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+
+ /* time to check quarter pels. */
+ if (bestmv->as_mv.row < startmv.as_mv.row)
+ y -= y_stride;
+
+ if (bestmv->as_mv.col < startmv.as_mv.col)
+ y--;
+
+ startmv = *bestmv;
+
+
+
+ /* go left then right and check error */
+ this_mv.as_mv.row = startmv.as_mv.row;
+
+ if (startmv.as_mv.col & 7)
+ {
+ this_mv.as_mv.col = startmv.as_mv.col - 2;
+ thismse = vfp->svf(y, y_stride, this_mv.as_mv.col & 7, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ }
+ else
+ {
+ this_mv.as_mv.col = (startmv.as_mv.col - 8) | 6;
+ thismse = vfp->svf(y - 1, y_stride, 6, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ }
+
+ left = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (left < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = left;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ this_mv.as_mv.col += 4;
+ thismse = vfp->svf(y, y_stride, this_mv.as_mv.col & 7, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ right = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (right < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = right;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ /* go up then down and check error */
+ this_mv.as_mv.col = startmv.as_mv.col;
+
+ if (startmv.as_mv.row & 7)
+ {
+ this_mv.as_mv.row = startmv.as_mv.row - 2;
+ thismse = vfp->svf(y, y_stride, this_mv.as_mv.col & 7, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ }
+ else
+ {
+ this_mv.as_mv.row = (startmv.as_mv.row - 8) | 6;
+ thismse = vfp->svf(y - y_stride, y_stride, this_mv.as_mv.col & 7, 6, z, b->src_stride, &sse);
+ }
+
+ up = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (up < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = up;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ this_mv.as_mv.row += 4;
+ thismse = vfp->svf(y, y_stride, this_mv.as_mv.col & 7, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ down = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (down < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = down;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+
+ /* now check 1 more diagonal */
+ whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
+
+ this_mv = startmv;
+
+ switch (whichdir)
+ {
+ case 0:
+
+ if (startmv.as_mv.row & 7)
+ {
+ this_mv.as_mv.row -= 2;
+
+ if (startmv.as_mv.col & 7)
+ {
+ this_mv.as_mv.col -= 2;
+ thismse = vfp->svf(y, y_stride, this_mv.as_mv.col & 7, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ }
+ else
+ {
+ this_mv.as_mv.col = (startmv.as_mv.col - 8) | 6;
+ thismse = vfp->svf(y - 1, y_stride, 6, this_mv.as_mv.row & 7, z, b->src_stride, &sse);;
+ }
+ }
+ else
+ {
+ this_mv.as_mv.row = (startmv.as_mv.row - 8) | 6;
+
+ if (startmv.as_mv.col & 7)
+ {
+ this_mv.as_mv.col -= 2;
+ thismse = vfp->svf(y - y_stride, y_stride, this_mv.as_mv.col & 7, 6, z, b->src_stride, &sse);
+ }
+ else
+ {
+ this_mv.as_mv.col = (startmv.as_mv.col - 8) | 6;
+ thismse = vfp->svf(y - y_stride - 1, y_stride, 6, 6, z, b->src_stride, &sse);
+ }
+ }
+
+ break;
+ case 1:
+ this_mv.as_mv.col += 2;
+
+ if (startmv.as_mv.row & 7)
+ {
+ this_mv.as_mv.row -= 2;
+ thismse = vfp->svf(y, y_stride, this_mv.as_mv.col & 7, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ }
+ else
+ {
+ this_mv.as_mv.row = (startmv.as_mv.row - 8) | 6;
+ thismse = vfp->svf(y - y_stride, y_stride, this_mv.as_mv.col & 7, 6, z, b->src_stride, &sse);
+ }
+
+ break;
+ case 2:
+ this_mv.as_mv.row += 2;
+
+ if (startmv.as_mv.col & 7)
+ {
+ this_mv.as_mv.col -= 2;
+ thismse = vfp->svf(y, y_stride, this_mv.as_mv.col & 7, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ }
+ else
+ {
+ this_mv.as_mv.col = (startmv.as_mv.col - 8) | 6;
+ thismse = vfp->svf(y - 1, y_stride, 6, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ }
+
+ break;
+ case 3:
+ this_mv.as_mv.col += 2;
+ this_mv.as_mv.row += 2;
+ thismse = vfp->svf(y, y_stride, this_mv.as_mv.col & 7, this_mv.as_mv.row & 7, z, b->src_stride, &sse);
+ break;
+ }
+
+ diag = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (diag < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = diag;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ return bestmse;
+}
+
+int vp8_find_best_half_pixel_step(MACROBLOCK *x, BLOCK *b, BLOCKD *d,
+ int_mv *bestmv, int_mv *ref_mv,
+ int error_per_bit,
+ const vp8_variance_fn_ptr_t *vfp,
+ int *mvcost[2], int *distortion,
+ unsigned int *sse1)
+{
+ int bestmse = INT_MAX;
+ int_mv startmv;
+ int_mv this_mv;
+ unsigned char *z = (*(b->base_src) + b->src);
+ int left, right, up, down, diag;
+ unsigned int sse;
+ int whichdir ;
+ int thismse;
+ int y_stride;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+
+#if ARCH_X86 || ARCH_X86_64
+ MACROBLOCKD *xd = &x->e_mbd;
+ unsigned char *y0 = base_pre + d->offset + (bestmv->as_mv.row) * pre_stride + bestmv->as_mv.col;
+ unsigned char *y;
+
+ y_stride = 32;
+ /* Copy 18 rows x 32 cols area to intermediate buffer before searching. */
+ vfp->copymem(y0 - 1 - pre_stride, pre_stride, xd->y_buf, y_stride, 18);
+ y = xd->y_buf + y_stride + 1;
+#else
+ unsigned char *y = base_pre + d->offset + (bestmv->as_mv.row) * pre_stride + bestmv->as_mv.col;
+ y_stride = pre_stride;
+#endif
+
+ /* central mv */
+ bestmv->as_mv.row <<= 3;
+ bestmv->as_mv.col <<= 3;
+ startmv = *bestmv;
+
+ /* calculate central point error */
+ bestmse = vfp->vf(y, y_stride, z, b->src_stride, sse1);
+ *distortion = bestmse;
+ bestmse += mv_err_cost(bestmv, ref_mv, mvcost, error_per_bit);
+
+ /* go left then right and check error */
+ this_mv.as_mv.row = startmv.as_mv.row;
+ this_mv.as_mv.col = ((startmv.as_mv.col - 8) | 4);
+ thismse = vfp->svf_halfpix_h(y - 1, y_stride, z, b->src_stride, &sse);
+ left = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (left < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = left;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ this_mv.as_mv.col += 8;
+ thismse = vfp->svf_halfpix_h(y, y_stride, z, b->src_stride, &sse);
+ right = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (right < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = right;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ /* go up then down and check error */
+ this_mv.as_mv.col = startmv.as_mv.col;
+ this_mv.as_mv.row = ((startmv.as_mv.row - 8) | 4);
+ thismse = vfp->svf_halfpix_v(y - y_stride, y_stride, z, b->src_stride, &sse);
+ up = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (up < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = up;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ this_mv.as_mv.row += 8;
+ thismse = vfp->svf_halfpix_v(y, y_stride, z, b->src_stride, &sse);
+ down = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (down < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = down;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ /* now check 1 more diagonal - */
+ whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
+ this_mv = startmv;
+
+ switch (whichdir)
+ {
+ case 0:
+ this_mv.as_mv.col = (this_mv.as_mv.col - 8) | 4;
+ this_mv.as_mv.row = (this_mv.as_mv.row - 8) | 4;
+ thismse = vfp->svf_halfpix_hv(y - 1 - y_stride, y_stride, z, b->src_stride, &sse);
+ break;
+ case 1:
+ this_mv.as_mv.col += 4;
+ this_mv.as_mv.row = (this_mv.as_mv.row - 8) | 4;
+ thismse = vfp->svf_halfpix_hv(y - y_stride, y_stride, z, b->src_stride, &sse);
+ break;
+ case 2:
+ this_mv.as_mv.col = (this_mv.as_mv.col - 8) | 4;
+ this_mv.as_mv.row += 4;
+ thismse = vfp->svf_halfpix_hv(y - 1, y_stride, z, b->src_stride, &sse);
+ break;
+ case 3:
+ default:
+ this_mv.as_mv.col += 4;
+ this_mv.as_mv.row += 4;
+ thismse = vfp->svf_halfpix_hv(y, y_stride, z, b->src_stride, &sse);
+ break;
+ }
+
+ diag = thismse + mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
+
+ if (diag < bestmse)
+ {
+ *bestmv = this_mv;
+ bestmse = diag;
+ *distortion = thismse;
+ *sse1 = sse;
+ }
+
+ return bestmse;
+}
+
+#define CHECK_BOUNDS(range) \
+{\
+ all_in = 1;\
+ all_in &= ((br-range) >= x->mv_row_min);\
+ all_in &= ((br+range) <= x->mv_row_max);\
+ all_in &= ((bc-range) >= x->mv_col_min);\
+ all_in &= ((bc+range) <= x->mv_col_max);\
+}
+
+#define CHECK_POINT \
+{\
+ if (this_mv.as_mv.col < x->mv_col_min) continue;\
+ if (this_mv.as_mv.col > x->mv_col_max) continue;\
+ if (this_mv.as_mv.row < x->mv_row_min) continue;\
+ if (this_mv.as_mv.row > x->mv_row_max) continue;\
+}
+
+#define CHECK_BETTER \
+{\
+ if (thissad < bestsad)\
+ {\
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv, mvsadcost, sad_per_bit);\
+ if (thissad < bestsad)\
+ {\
+ bestsad = thissad;\
+ best_site = i;\
+ }\
+ }\
+}
+
+static const MV next_chkpts[6][3] =
+{
+ {{ -2, 0}, { -1, -2}, {1, -2}},
+ {{ -1, -2}, {1, -2}, {2, 0}},
+ {{1, -2}, {2, 0}, {1, 2}},
+ {{2, 0}, {1, 2}, { -1, 2}},
+ {{1, 2}, { -1, 2}, { -2, 0}},
+ {{ -1, 2}, { -2, 0}, { -1, -2}}
+};
+
+int vp8_hex_search
+(
+ MACROBLOCK *x,
+ BLOCK *b,
+ BLOCKD *d,
+ int_mv *ref_mv,
+ int_mv *best_mv,
+ int search_param,
+ int sad_per_bit,
+ const vp8_variance_fn_ptr_t *vfp,
+ int *mvsadcost[2],
+ int *mvcost[2],
+ int_mv *center_mv
+)
+{
+ MV hex[6] = { { -1, -2}, {1, -2}, {2, 0}, {1, 2}, { -1, 2}, { -2, 0} } ;
+ MV neighbors[4] = {{0, -1}, { -1, 0}, {1, 0}, {0, 1}} ;
+ int i, j;
+
+ unsigned char *what = (*(b->base_src) + b->src);
+ int what_stride = b->src_stride;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+
+ int in_what_stride = pre_stride;
+ int br, bc;
+ int_mv this_mv;
+ unsigned int bestsad;
+ unsigned int thissad;
+ unsigned char *base_offset;
+ unsigned char *this_offset;
+ int k = -1;
+ int all_in;
+ int best_site = -1;
+ int hex_range = 127;
+ int dia_range = 8;
+
+ int_mv fcenter_mv;
+ fcenter_mv.as_mv.row = center_mv->as_mv.row >> 3;
+ fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
+
+ /* adjust ref_mv to make sure it is within MV range */
+ vp8_clamp_mv(ref_mv, x->mv_col_min, x->mv_col_max, x->mv_row_min, x->mv_row_max);
+ br = ref_mv->as_mv.row;
+ bc = ref_mv->as_mv.col;
+
+ /* Work out the start point for the search */
+ base_offset = (unsigned char *)(base_pre + d->offset);
+ this_offset = base_offset + (br * (pre_stride)) + bc;
+ this_mv.as_mv.row = br;
+ this_mv.as_mv.col = bc;
+ bestsad = vfp->sdf(what, what_stride, this_offset, in_what_stride, UINT_MAX)
+ + mvsad_err_cost(&this_mv, &fcenter_mv, mvsadcost, sad_per_bit);
+
+#if CONFIG_MULTI_RES_ENCODING
+ /* Lower search range based on prediction info */
+ if (search_param >= 6) goto cal_neighbors;
+ else if (search_param >= 5) hex_range = 4;
+ else if (search_param >= 4) hex_range = 6;
+ else if (search_param >= 3) hex_range = 15;
+ else if (search_param >= 2) hex_range = 31;
+ else if (search_param >= 1) hex_range = 63;
+
+ dia_range = 8;
+#endif
+
+ /* hex search */
+ CHECK_BOUNDS(2)
+
+ if(all_in)
+ {
+ for (i = 0; i < 6; i++)
+ {
+ this_mv.as_mv.row = br + hex[i].row;
+ this_mv.as_mv.col = bc + hex[i].col;
+ this_offset = base_offset + (this_mv.as_mv.row * in_what_stride) + this_mv.as_mv.col;
+ thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
+ CHECK_BETTER
+ }
+ }else
+ {
+ for (i = 0; i < 6; i++)
+ {
+ this_mv.as_mv.row = br + hex[i].row;
+ this_mv.as_mv.col = bc + hex[i].col;
+ CHECK_POINT
+ this_offset = base_offset + (this_mv.as_mv.row * in_what_stride) + this_mv.as_mv.col;
+ thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
+ CHECK_BETTER
+ }
+ }
+
+ if (best_site == -1)
+ goto cal_neighbors;
+ else
+ {
+ br += hex[best_site].row;
+ bc += hex[best_site].col;
+ k = best_site;
+ }
+
+ for (j = 1; j < hex_range; j++)
+ {
+ best_site = -1;
+ CHECK_BOUNDS(2)
+
+ if(all_in)
+ {
+ for (i = 0; i < 3; i++)
+ {
+ this_mv.as_mv.row = br + next_chkpts[k][i].row;
+ this_mv.as_mv.col = bc + next_chkpts[k][i].col;
+ this_offset = base_offset + (this_mv.as_mv.row * (in_what_stride)) + this_mv.as_mv.col;
+ thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
+ CHECK_BETTER
+ }
+ }else
+ {
+ for (i = 0; i < 3; i++)
+ {
+ this_mv.as_mv.row = br + next_chkpts[k][i].row;
+ this_mv.as_mv.col = bc + next_chkpts[k][i].col;
+ CHECK_POINT
+ this_offset = base_offset + (this_mv.as_mv.row * (in_what_stride)) + this_mv.as_mv.col;
+ thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
+ CHECK_BETTER
+ }
+ }
+
+ if (best_site == -1)
+ break;
+ else
+ {
+ br += next_chkpts[k][best_site].row;
+ bc += next_chkpts[k][best_site].col;
+ k += 5 + best_site;
+ if (k >= 12) k -= 12;
+ else if (k >= 6) k -= 6;
+ }
+ }
+
+ /* check 4 1-away neighbors */
+cal_neighbors:
+ for (j = 0; j < dia_range; j++)
+ {
+ best_site = -1;
+ CHECK_BOUNDS(1)
+
+ if(all_in)
+ {
+ for (i = 0; i < 4; i++)
+ {
+ this_mv.as_mv.row = br + neighbors[i].row;
+ this_mv.as_mv.col = bc + neighbors[i].col;
+ this_offset = base_offset + (this_mv.as_mv.row * (in_what_stride)) + this_mv.as_mv.col;
+ thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
+ CHECK_BETTER
+ }
+ }else
+ {
+ for (i = 0; i < 4; i++)
+ {
+ this_mv.as_mv.row = br + neighbors[i].row;
+ this_mv.as_mv.col = bc + neighbors[i].col;
+ CHECK_POINT
+ this_offset = base_offset + (this_mv.as_mv.row * (in_what_stride)) + this_mv.as_mv.col;
+ thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
+ CHECK_BETTER
+ }
+ }
+
+ if (best_site == -1)
+ break;
+ else
+ {
+ br += neighbors[best_site].row;
+ bc += neighbors[best_site].col;
+ }
+ }
+
+ best_mv->as_mv.row = br;
+ best_mv->as_mv.col = bc;
+
+ return bestsad;
+}
+#undef CHECK_BOUNDS
+#undef CHECK_POINT
+#undef CHECK_BETTER
+
+int vp8_diamond_search_sad_c
+(
+ MACROBLOCK *x,
+ BLOCK *b,
+ BLOCKD *d,
+ int_mv *ref_mv,
+ int_mv *best_mv,
+ int search_param,
+ int sad_per_bit,
+ int *num00,
+ vp8_variance_fn_ptr_t *fn_ptr,
+ int *mvcost[2],
+ int_mv *center_mv
+)
+{
+ int i, j, step;
+
+ unsigned char *what = (*(b->base_src) + b->src);
+ int what_stride = b->src_stride;
+ unsigned char *in_what;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+ int in_what_stride = pre_stride;
+ unsigned char *best_address;
+
+ int tot_steps;
+ int_mv this_mv;
+
+ unsigned int bestsad;
+ unsigned int thissad;
+ int best_site = 0;
+ int last_site = 0;
+
+ int ref_row;
+ int ref_col;
+ int this_row_offset;
+ int this_col_offset;
+ search_site *ss;
+
+ unsigned char *check_here;
+
+ int *mvsadcost[2];
+ int_mv fcenter_mv;
+
+ mvsadcost[0] = x->mvsadcost[0];
+ mvsadcost[1] = x->mvsadcost[1];
+ fcenter_mv.as_mv.row = center_mv->as_mv.row >> 3;
+ fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
+
+ vp8_clamp_mv(ref_mv, x->mv_col_min, x->mv_col_max, x->mv_row_min, x->mv_row_max);
+ ref_row = ref_mv->as_mv.row;
+ ref_col = ref_mv->as_mv.col;
+ *num00 = 0;
+ best_mv->as_mv.row = ref_row;
+ best_mv->as_mv.col = ref_col;
+
+ /* Work out the start point for the search */
+ in_what = (unsigned char *)(base_pre + d->offset + (ref_row * pre_stride) + ref_col);
+ best_address = in_what;
+
+ /* Check the starting position */
+ bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride, UINT_MAX)
+ + mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
+
+ /* search_param determines the length of the initial step and hence
+ * the number of iterations 0 = initial step (MAX_FIRST_STEP) pel :
+ * 1 = (MAX_FIRST_STEP/2) pel, 2 = (MAX_FIRST_STEP/4) pel... etc.
+ */
+ ss = &x->ss[search_param * x->searches_per_step];
+ tot_steps = (x->ss_count / x->searches_per_step) - search_param;
+
+ i = 1;
+
+ for (step = 0; step < tot_steps ; step++)
+ {
+ for (j = 0 ; j < x->searches_per_step ; j++)
+ {
+ /* Trap illegal vectors */
+ this_row_offset = best_mv->as_mv.row + ss[i].mv.row;
+ this_col_offset = best_mv->as_mv.col + ss[i].mv.col;
+
+ if ((this_col_offset > x->mv_col_min) && (this_col_offset < x->mv_col_max) &&
+ (this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
+
+ {
+ check_here = ss[i].offset + best_address;
+ thissad = fn_ptr->sdf(what, what_stride, check_here, in_what_stride, bestsad);
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.row = this_row_offset;
+ this_mv.as_mv.col = this_col_offset;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_site = i;
+ }
+ }
+ }
+
+ i++;
+ }
+
+ if (best_site != last_site)
+ {
+ best_mv->as_mv.row += ss[best_site].mv.row;
+ best_mv->as_mv.col += ss[best_site].mv.col;
+ best_address += ss[best_site].offset;
+ last_site = best_site;
+ }
+ else if (best_address == in_what)
+ (*num00)++;
+ }
+
+ this_mv.as_mv.row = best_mv->as_mv.row << 3;
+ this_mv.as_mv.col = best_mv->as_mv.col << 3;
+
+ return fn_ptr->vf(what, what_stride, best_address, in_what_stride, &thissad)
+ + mv_err_cost(&this_mv, center_mv, mvcost, x->errorperbit);
+}
+
+int vp8_diamond_search_sadx4
+(
+ MACROBLOCK *x,
+ BLOCK *b,
+ BLOCKD *d,
+ int_mv *ref_mv,
+ int_mv *best_mv,
+ int search_param,
+ int sad_per_bit,
+ int *num00,
+ vp8_variance_fn_ptr_t *fn_ptr,
+ int *mvcost[2],
+ int_mv *center_mv
+)
+{
+ int i, j, step;
+
+ unsigned char *what = (*(b->base_src) + b->src);
+ int what_stride = b->src_stride;
+ unsigned char *in_what;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+ int in_what_stride = pre_stride;
+ unsigned char *best_address;
+
+ int tot_steps;
+ int_mv this_mv;
+
+ unsigned int bestsad;
+ unsigned int thissad;
+ int best_site = 0;
+ int last_site = 0;
+
+ int ref_row;
+ int ref_col;
+ int this_row_offset;
+ int this_col_offset;
+ search_site *ss;
+
+ unsigned char *check_here;
+
+ int *mvsadcost[2];
+ int_mv fcenter_mv;
+
+ mvsadcost[0] = x->mvsadcost[0];
+ mvsadcost[1] = x->mvsadcost[1];
+ fcenter_mv.as_mv.row = center_mv->as_mv.row >> 3;
+ fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
+
+ vp8_clamp_mv(ref_mv, x->mv_col_min, x->mv_col_max, x->mv_row_min, x->mv_row_max);
+ ref_row = ref_mv->as_mv.row;
+ ref_col = ref_mv->as_mv.col;
+ *num00 = 0;
+ best_mv->as_mv.row = ref_row;
+ best_mv->as_mv.col = ref_col;
+
+ /* Work out the start point for the search */
+ in_what = (unsigned char *)(base_pre + d->offset + (ref_row * pre_stride) + ref_col);
+ best_address = in_what;
+
+ /* Check the starting position */
+ bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride, UINT_MAX)
+ + mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
+
+ /* search_param determines the length of the initial step and hence the
+ * number of iterations 0 = initial step (MAX_FIRST_STEP) pel : 1 =
+ * (MAX_FIRST_STEP/2) pel, 2 = (MAX_FIRST_STEP/4) pel... etc.
+ */
+ ss = &x->ss[search_param * x->searches_per_step];
+ tot_steps = (x->ss_count / x->searches_per_step) - search_param;
+
+ i = 1;
+
+ for (step = 0; step < tot_steps ; step++)
+ {
+ int all_in = 1, t;
+
+ /* To know if all neighbor points are within the bounds, 4 bounds
+ * checking are enough instead of checking 4 bounds for each
+ * points.
+ */
+ all_in &= ((best_mv->as_mv.row + ss[i].mv.row)> x->mv_row_min);
+ all_in &= ((best_mv->as_mv.row + ss[i+1].mv.row) < x->mv_row_max);
+ all_in &= ((best_mv->as_mv.col + ss[i+2].mv.col) > x->mv_col_min);
+ all_in &= ((best_mv->as_mv.col + ss[i+3].mv.col) < x->mv_col_max);
+
+ if (all_in)
+ {
+ unsigned int sad_array[4];
+
+ for (j = 0 ; j < x->searches_per_step ; j += 4)
+ {
+ const unsigned char *block_offset[4];
+
+ for (t = 0; t < 4; t++)
+ block_offset[t] = ss[i+t].offset + best_address;
+
+ fn_ptr->sdx4df(what, what_stride, block_offset, in_what_stride, sad_array);
+
+ for (t = 0; t < 4; t++, i++)
+ {
+ if (sad_array[t] < bestsad)
+ {
+ this_mv.as_mv.row = best_mv->as_mv.row + ss[i].mv.row;
+ this_mv.as_mv.col = best_mv->as_mv.col + ss[i].mv.col;
+ sad_array[t] += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (sad_array[t] < bestsad)
+ {
+ bestsad = sad_array[t];
+ best_site = i;
+ }
+ }
+ }
+ }
+ }
+ else
+ {
+ for (j = 0 ; j < x->searches_per_step ; j++)
+ {
+ /* Trap illegal vectors */
+ this_row_offset = best_mv->as_mv.row + ss[i].mv.row;
+ this_col_offset = best_mv->as_mv.col + ss[i].mv.col;
+
+ if ((this_col_offset > x->mv_col_min) && (this_col_offset < x->mv_col_max) &&
+ (this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
+ {
+ check_here = ss[i].offset + best_address;
+ thissad = fn_ptr->sdf(what, what_stride, check_here, in_what_stride, bestsad);
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.row = this_row_offset;
+ this_mv.as_mv.col = this_col_offset;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_site = i;
+ }
+ }
+ }
+ i++;
+ }
+ }
+
+ if (best_site != last_site)
+ {
+ best_mv->as_mv.row += ss[best_site].mv.row;
+ best_mv->as_mv.col += ss[best_site].mv.col;
+ best_address += ss[best_site].offset;
+ last_site = best_site;
+ }
+ else if (best_address == in_what)
+ (*num00)++;
+ }
+
+ this_mv.as_mv.row = best_mv->as_mv.row << 3;
+ this_mv.as_mv.col = best_mv->as_mv.col << 3;
+
+ return fn_ptr->vf(what, what_stride, best_address, in_what_stride, &thissad)
+ + mv_err_cost(&this_mv, center_mv, mvcost, x->errorperbit);
+}
+
+int vp8_full_search_sad_c(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
+ int sad_per_bit, int distance,
+ vp8_variance_fn_ptr_t *fn_ptr, int *mvcost[2],
+ int_mv *center_mv)
+{
+ unsigned char *what = (*(b->base_src) + b->src);
+ int what_stride = b->src_stride;
+ unsigned char *in_what;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+ int in_what_stride = pre_stride;
+ int mv_stride = pre_stride;
+ unsigned char *bestaddress;
+ int_mv *best_mv = &d->bmi.mv;
+ int_mv this_mv;
+ unsigned int bestsad;
+ unsigned int thissad;
+ int r, c;
+
+ unsigned char *check_here;
+
+ int ref_row = ref_mv->as_mv.row;
+ int ref_col = ref_mv->as_mv.col;
+
+ int row_min = ref_row - distance;
+ int row_max = ref_row + distance;
+ int col_min = ref_col - distance;
+ int col_max = ref_col + distance;
+
+ int *mvsadcost[2];
+ int_mv fcenter_mv;
+
+ mvsadcost[0] = x->mvsadcost[0];
+ mvsadcost[1] = x->mvsadcost[1];
+ fcenter_mv.as_mv.row = center_mv->as_mv.row >> 3;
+ fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
+
+ /* Work out the mid point for the search */
+ in_what = base_pre + d->offset;
+ bestaddress = in_what + (ref_row * pre_stride) + ref_col;
+
+ best_mv->as_mv.row = ref_row;
+ best_mv->as_mv.col = ref_col;
+
+ /* Baseline value at the centre */
+ bestsad = fn_ptr->sdf(what, what_stride, bestaddress,
+ in_what_stride, UINT_MAX)
+ + mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
+
+ /* Apply further limits to prevent us looking using vectors that
+ * stretch beyiond the UMV border
+ */
+ if (col_min < x->mv_col_min)
+ col_min = x->mv_col_min;
+
+ if (col_max > x->mv_col_max)
+ col_max = x->mv_col_max;
+
+ if (row_min < x->mv_row_min)
+ row_min = x->mv_row_min;
+
+ if (row_max > x->mv_row_max)
+ row_max = x->mv_row_max;
+
+ for (r = row_min; r < row_max ; r++)
+ {
+ this_mv.as_mv.row = r;
+ check_here = r * mv_stride + in_what + col_min;
+
+ for (c = col_min; c < col_max; c++)
+ {
+ thissad = fn_ptr->sdf(what, what_stride, check_here, in_what_stride, bestsad);
+
+ this_mv.as_mv.col = c;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_mv->as_mv.row = r;
+ best_mv->as_mv.col = c;
+ bestaddress = check_here;
+ }
+
+ check_here++;
+ }
+ }
+
+ this_mv.as_mv.row = best_mv->as_mv.row << 3;
+ this_mv.as_mv.col = best_mv->as_mv.col << 3;
+
+ return fn_ptr->vf(what, what_stride, bestaddress, in_what_stride, &thissad)
+ + mv_err_cost(&this_mv, center_mv, mvcost, x->errorperbit);
+}
+
+int vp8_full_search_sadx3(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
+ int sad_per_bit, int distance,
+ vp8_variance_fn_ptr_t *fn_ptr, int *mvcost[2],
+ int_mv *center_mv)
+{
+ unsigned char *what = (*(b->base_src) + b->src);
+ int what_stride = b->src_stride;
+ unsigned char *in_what;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+ int in_what_stride = pre_stride;
+ int mv_stride = pre_stride;
+ unsigned char *bestaddress;
+ int_mv *best_mv = &d->bmi.mv;
+ int_mv this_mv;
+ unsigned int bestsad;
+ unsigned int thissad;
+ int r, c;
+
+ unsigned char *check_here;
+
+ int ref_row = ref_mv->as_mv.row;
+ int ref_col = ref_mv->as_mv.col;
+
+ int row_min = ref_row - distance;
+ int row_max = ref_row + distance;
+ int col_min = ref_col - distance;
+ int col_max = ref_col + distance;
+
+ unsigned int sad_array[3];
+
+ int *mvsadcost[2];
+ int_mv fcenter_mv;
+
+ mvsadcost[0] = x->mvsadcost[0];
+ mvsadcost[1] = x->mvsadcost[1];
+ fcenter_mv.as_mv.row = center_mv->as_mv.row >> 3;
+ fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
+
+ /* Work out the mid point for the search */
+ in_what = base_pre + d->offset;
+ bestaddress = in_what + (ref_row * pre_stride) + ref_col;
+
+ best_mv->as_mv.row = ref_row;
+ best_mv->as_mv.col = ref_col;
+
+ /* Baseline value at the centre */
+ bestsad = fn_ptr->sdf(what, what_stride, bestaddress,
+ in_what_stride, UINT_MAX)
+ + mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
+
+ /* Apply further limits to prevent us looking using vectors that stretch
+ * beyond the UMV border
+ */
+ if (col_min < x->mv_col_min)
+ col_min = x->mv_col_min;
+
+ if (col_max > x->mv_col_max)
+ col_max = x->mv_col_max;
+
+ if (row_min < x->mv_row_min)
+ row_min = x->mv_row_min;
+
+ if (row_max > x->mv_row_max)
+ row_max = x->mv_row_max;
+
+ for (r = row_min; r < row_max ; r++)
+ {
+ this_mv.as_mv.row = r;
+ check_here = r * mv_stride + in_what + col_min;
+ c = col_min;
+
+ while ((c + 2) < col_max)
+ {
+ int i;
+
+ fn_ptr->sdx3f(what, what_stride, check_here, in_what_stride, sad_array);
+
+ for (i = 0; i < 3; i++)
+ {
+ thissad = sad_array[i];
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.col = c;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_mv->as_mv.row = r;
+ best_mv->as_mv.col = c;
+ bestaddress = check_here;
+ }
+ }
+
+ check_here++;
+ c++;
+ }
+ }
+
+ while (c < col_max)
+ {
+ thissad = fn_ptr->sdf(what, what_stride, check_here, in_what_stride, bestsad);
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.col = c;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_mv->as_mv.row = r;
+ best_mv->as_mv.col = c;
+ bestaddress = check_here;
+ }
+ }
+
+ check_here ++;
+ c ++;
+ }
+
+ }
+
+ this_mv.as_mv.row = best_mv->as_mv.row << 3;
+ this_mv.as_mv.col = best_mv->as_mv.col << 3;
+
+ return fn_ptr->vf(what, what_stride, bestaddress, in_what_stride, &thissad)
+ + mv_err_cost(&this_mv, center_mv, mvcost, x->errorperbit);
+}
+
+int vp8_full_search_sadx8(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
+ int sad_per_bit, int distance,
+ vp8_variance_fn_ptr_t *fn_ptr, int *mvcost[2],
+ int_mv *center_mv)
+{
+ unsigned char *what = (*(b->base_src) + b->src);
+ int what_stride = b->src_stride;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+ unsigned char *in_what;
+ int in_what_stride = pre_stride;
+ int mv_stride = pre_stride;
+ unsigned char *bestaddress;
+ int_mv *best_mv = &d->bmi.mv;
+ int_mv this_mv;
+ unsigned int bestsad;
+ unsigned int thissad;
+ int r, c;
+
+ unsigned char *check_here;
+
+ int ref_row = ref_mv->as_mv.row;
+ int ref_col = ref_mv->as_mv.col;
+
+ int row_min = ref_row - distance;
+ int row_max = ref_row + distance;
+ int col_min = ref_col - distance;
+ int col_max = ref_col + distance;
+
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, sad_array8, 8);
+ unsigned int sad_array[3];
+
+ int *mvsadcost[2];
+ int_mv fcenter_mv;
+
+ mvsadcost[0] = x->mvsadcost[0];
+ mvsadcost[1] = x->mvsadcost[1];
+ fcenter_mv.as_mv.row = center_mv->as_mv.row >> 3;
+ fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
+
+ /* Work out the mid point for the search */
+ in_what = base_pre + d->offset;
+ bestaddress = in_what + (ref_row * pre_stride) + ref_col;
+
+ best_mv->as_mv.row = ref_row;
+ best_mv->as_mv.col = ref_col;
+
+ /* Baseline value at the centre */
+ bestsad = fn_ptr->sdf(what, what_stride,
+ bestaddress, in_what_stride, UINT_MAX)
+ + mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
+
+ /* Apply further limits to prevent us looking using vectors that stretch
+ * beyond the UMV border
+ */
+ if (col_min < x->mv_col_min)
+ col_min = x->mv_col_min;
+
+ if (col_max > x->mv_col_max)
+ col_max = x->mv_col_max;
+
+ if (row_min < x->mv_row_min)
+ row_min = x->mv_row_min;
+
+ if (row_max > x->mv_row_max)
+ row_max = x->mv_row_max;
+
+ for (r = row_min; r < row_max ; r++)
+ {
+ this_mv.as_mv.row = r;
+ check_here = r * mv_stride + in_what + col_min;
+ c = col_min;
+
+ while ((c + 7) < col_max)
+ {
+ int i;
+
+ fn_ptr->sdx8f(what, what_stride, check_here, in_what_stride, sad_array8);
+
+ for (i = 0; i < 8; i++)
+ {
+ thissad = sad_array8[i];
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.col = c;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_mv->as_mv.row = r;
+ best_mv->as_mv.col = c;
+ bestaddress = check_here;
+ }
+ }
+
+ check_here++;
+ c++;
+ }
+ }
+
+ while ((c + 2) < col_max)
+ {
+ int i;
+
+ fn_ptr->sdx3f(what, what_stride, check_here , in_what_stride, sad_array);
+
+ for (i = 0; i < 3; i++)
+ {
+ thissad = sad_array[i];
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.col = c;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_mv->as_mv.row = r;
+ best_mv->as_mv.col = c;
+ bestaddress = check_here;
+ }
+ }
+
+ check_here++;
+ c++;
+ }
+ }
+
+ while (c < col_max)
+ {
+ thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.col = c;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
+ mvsadcost, sad_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_mv->as_mv.row = r;
+ best_mv->as_mv.col = c;
+ bestaddress = check_here;
+ }
+ }
+
+ check_here ++;
+ c ++;
+ }
+ }
+
+ this_mv.as_mv.row = best_mv->as_mv.row << 3;
+ this_mv.as_mv.col = best_mv->as_mv.col << 3;
+
+ return fn_ptr->vf(what, what_stride, bestaddress, in_what_stride, &thissad)
+ + mv_err_cost(&this_mv, center_mv, mvcost, x->errorperbit);
+}
+
+int vp8_refining_search_sad_c(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
+ int error_per_bit, int search_range,
+ vp8_variance_fn_ptr_t *fn_ptr, int *mvcost[2],
+ int_mv *center_mv)
+{
+ MV neighbors[4] = {{-1, 0}, {0, -1}, {0, 1}, {1, 0}};
+ int i, j;
+ short this_row_offset, this_col_offset;
+
+ int what_stride = b->src_stride;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+ int in_what_stride = pre_stride;
+ unsigned char *what = (*(b->base_src) + b->src);
+ unsigned char *best_address = (unsigned char *)(base_pre + d->offset +
+ (ref_mv->as_mv.row * pre_stride) + ref_mv->as_mv.col);
+ unsigned char *check_here;
+ int_mv this_mv;
+ unsigned int bestsad;
+ unsigned int thissad;
+
+ int *mvsadcost[2];
+ int_mv fcenter_mv;
+
+ mvsadcost[0] = x->mvsadcost[0];
+ mvsadcost[1] = x->mvsadcost[1];
+ fcenter_mv.as_mv.row = center_mv->as_mv.row >> 3;
+ fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
+
+ bestsad = fn_ptr->sdf(what, what_stride, best_address,
+ in_what_stride, UINT_MAX)
+ + mvsad_err_cost(ref_mv, &fcenter_mv, mvsadcost, error_per_bit);
+
+ for (i=0; i<search_range; i++)
+ {
+ int best_site = -1;
+
+ for (j = 0 ; j < 4 ; j++)
+ {
+ this_row_offset = ref_mv->as_mv.row + neighbors[j].row;
+ this_col_offset = ref_mv->as_mv.col + neighbors[j].col;
+
+ if ((this_col_offset > x->mv_col_min) && (this_col_offset < x->mv_col_max) &&
+ (this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
+ {
+ check_here = (neighbors[j].row)*in_what_stride + neighbors[j].col + best_address;
+ thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.row = this_row_offset;
+ this_mv.as_mv.col = this_col_offset;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv, mvsadcost, error_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_site = j;
+ }
+ }
+ }
+ }
+
+ if (best_site == -1)
+ break;
+ else
+ {
+ ref_mv->as_mv.row += neighbors[best_site].row;
+ ref_mv->as_mv.col += neighbors[best_site].col;
+ best_address += (neighbors[best_site].row)*in_what_stride + neighbors[best_site].col;
+ }
+ }
+
+ this_mv.as_mv.row = ref_mv->as_mv.row << 3;
+ this_mv.as_mv.col = ref_mv->as_mv.col << 3;
+
+ return fn_ptr->vf(what, what_stride, best_address, in_what_stride, &thissad)
+ + mv_err_cost(&this_mv, center_mv, mvcost, x->errorperbit);
+}
+
+int vp8_refining_search_sadx4(MACROBLOCK *x, BLOCK *b, BLOCKD *d,
+ int_mv *ref_mv, int error_per_bit,
+ int search_range, vp8_variance_fn_ptr_t *fn_ptr,
+ int *mvcost[2], int_mv *center_mv)
+{
+ MV neighbors[4] = {{-1, 0}, {0, -1}, {0, 1}, {1, 0}};
+ int i, j;
+ short this_row_offset, this_col_offset;
+
+ int what_stride = b->src_stride;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+ int in_what_stride = pre_stride;
+ unsigned char *what = (*(b->base_src) + b->src);
+ unsigned char *best_address = (unsigned char *)(base_pre + d->offset +
+ (ref_mv->as_mv.row * pre_stride) + ref_mv->as_mv.col);
+ unsigned char *check_here;
+ int_mv this_mv;
+ unsigned int bestsad;
+ unsigned int thissad;
+
+ int *mvsadcost[2];
+ int_mv fcenter_mv;
+
+ mvsadcost[0] = x->mvsadcost[0];
+ mvsadcost[1] = x->mvsadcost[1];
+ fcenter_mv.as_mv.row = center_mv->as_mv.row >> 3;
+ fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
+
+ bestsad = fn_ptr->sdf(what, what_stride, best_address,
+ in_what_stride, UINT_MAX)
+ + mvsad_err_cost(ref_mv, &fcenter_mv, mvsadcost, error_per_bit);
+
+ for (i=0; i<search_range; i++)
+ {
+ int best_site = -1;
+ int all_in = 1;
+
+ all_in &= ((ref_mv->as_mv.row - 1) > x->mv_row_min);
+ all_in &= ((ref_mv->as_mv.row + 1) < x->mv_row_max);
+ all_in &= ((ref_mv->as_mv.col - 1) > x->mv_col_min);
+ all_in &= ((ref_mv->as_mv.col + 1) < x->mv_col_max);
+
+ if(all_in)
+ {
+ unsigned int sad_array[4];
+ const unsigned char *block_offset[4];
+ block_offset[0] = best_address - in_what_stride;
+ block_offset[1] = best_address - 1;
+ block_offset[2] = best_address + 1;
+ block_offset[3] = best_address + in_what_stride;
+
+ fn_ptr->sdx4df(what, what_stride, block_offset, in_what_stride, sad_array);
+
+ for (j = 0; j < 4; j++)
+ {
+ if (sad_array[j] < bestsad)
+ {
+ this_mv.as_mv.row = ref_mv->as_mv.row + neighbors[j].row;
+ this_mv.as_mv.col = ref_mv->as_mv.col + neighbors[j].col;
+ sad_array[j] += mvsad_err_cost(&this_mv, &fcenter_mv, mvsadcost, error_per_bit);
+
+ if (sad_array[j] < bestsad)
+ {
+ bestsad = sad_array[j];
+ best_site = j;
+ }
+ }
+ }
+ }
+ else
+ {
+ for (j = 0 ; j < 4 ; j++)
+ {
+ this_row_offset = ref_mv->as_mv.row + neighbors[j].row;
+ this_col_offset = ref_mv->as_mv.col + neighbors[j].col;
+
+ if ((this_col_offset > x->mv_col_min) && (this_col_offset < x->mv_col_max) &&
+ (this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
+ {
+ check_here = (neighbors[j].row)*in_what_stride + neighbors[j].col + best_address;
+ thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
+
+ if (thissad < bestsad)
+ {
+ this_mv.as_mv.row = this_row_offset;
+ this_mv.as_mv.col = this_col_offset;
+ thissad += mvsad_err_cost(&this_mv, &fcenter_mv, mvsadcost, error_per_bit);
+
+ if (thissad < bestsad)
+ {
+ bestsad = thissad;
+ best_site = j;
+ }
+ }
+ }
+ }
+ }
+
+ if (best_site == -1)
+ break;
+ else
+ {
+ ref_mv->as_mv.row += neighbors[best_site].row;
+ ref_mv->as_mv.col += neighbors[best_site].col;
+ best_address += (neighbors[best_site].row)*in_what_stride + neighbors[best_site].col;
+ }
+ }
+
+ this_mv.as_mv.row = ref_mv->as_mv.row << 3;
+ this_mv.as_mv.col = ref_mv->as_mv.col << 3;
+
+ return fn_ptr->vf(what, what_stride, best_address, in_what_stride, &thissad)
+ + mv_err_cost(&this_mv, center_mv, mvcost, x->errorperbit);
+}
+
+#ifdef ENTROPY_STATS
+void print_mode_context(void)
+{
+ FILE *f = fopen("modecont.c", "w");
+ int i, j;
+
+ fprintf(f, "#include \"entropy.h\"\n");
+ fprintf(f, "const int vp8_mode_contexts[6][4] =\n");
+ fprintf(f, "{\n");
+
+ for (j = 0; j < 6; j++)
+ {
+ fprintf(f, " { /* %d */\n", j);
+ fprintf(f, " ");
+
+ for (i = 0; i < 4; i++)
+ {
+ int overal_prob;
+ int this_prob;
+ int count;
+
+ /* Overall probs */
+ count = mv_mode_cts[i][0] + mv_mode_cts[i][1];
+
+ if (count)
+ overal_prob = 256 * mv_mode_cts[i][0] / count;
+ else
+ overal_prob = 128;
+
+ if (overal_prob == 0)
+ overal_prob = 1;
+
+ /* context probs */
+ count = mv_ref_ct[j][i][0] + mv_ref_ct[j][i][1];
+
+ if (count)
+ this_prob = 256 * mv_ref_ct[j][i][0] / count;
+ else
+ this_prob = 128;
+
+ if (this_prob == 0)
+ this_prob = 1;
+
+ fprintf(f, "%5d, ", this_prob);
+ }
+
+ fprintf(f, " },\n");
+ }
+
+ fprintf(f, "};\n");
+ fclose(f);
+}
+
+/* MV ref count ENTROPY_STATS stats code */
+#ifdef ENTROPY_STATS
+void init_mv_ref_counts()
+{
+ vpx_memset(mv_ref_ct, 0, sizeof(mv_ref_ct));
+ vpx_memset(mv_mode_cts, 0, sizeof(mv_mode_cts));
+}
+
+void accum_mv_refs(MB_PREDICTION_MODE m, const int ct[4])
+{
+ if (m == ZEROMV)
+ {
+ ++mv_ref_ct [ct[0]] [0] [0];
+ ++mv_mode_cts[0][0];
+ }
+ else
+ {
+ ++mv_ref_ct [ct[0]] [0] [1];
+ ++mv_mode_cts[0][1];
+
+ if (m == NEARESTMV)
+ {
+ ++mv_ref_ct [ct[1]] [1] [0];
+ ++mv_mode_cts[1][0];
+ }
+ else
+ {
+ ++mv_ref_ct [ct[1]] [1] [1];
+ ++mv_mode_cts[1][1];
+
+ if (m == NEARMV)
+ {
+ ++mv_ref_ct [ct[2]] [2] [0];
+ ++mv_mode_cts[2][0];
+ }
+ else
+ {
+ ++mv_ref_ct [ct[2]] [2] [1];
+ ++mv_mode_cts[2][1];
+
+ if (m == NEWMV)
+ {
+ ++mv_ref_ct [ct[3]] [3] [0];
+ ++mv_mode_cts[3][0];
+ }
+ else
+ {
+ ++mv_ref_ct [ct[3]] [3] [1];
+ ++mv_mode_cts[3][1];
+ }
+ }
+ }
+ }
+}
+
+#endif/* END MV ref count ENTROPY_STATS stats code */
+
+#endif
diff --git a/libvpx/vp8/encoder/mcomp.h b/libvpx/vp8/encoder/mcomp.h
new file mode 100644
index 0000000..890113f
--- /dev/null
+++ b/libvpx/vp8/encoder/mcomp.h
@@ -0,0 +1,107 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_MCOMP_H
+#define __INC_MCOMP_H
+
+#include "block.h"
+#include "vp8/common/variance.h"
+
+#ifdef ENTROPY_STATS
+extern void init_mv_ref_counts();
+extern void accum_mv_refs(MB_PREDICTION_MODE, const int near_mv_ref_cts[4]);
+#endif
+
+
+/* The maximum number of steps in a step search given the largest allowed
+ * initial step
+ */
+#define MAX_MVSEARCH_STEPS 8
+
+/* Max full pel mv specified in 1 pel units */
+#define MAX_FULL_PEL_VAL ((1 << (MAX_MVSEARCH_STEPS)) - 1)
+
+/* Maximum size of the first step in full pel units */
+#define MAX_FIRST_STEP (1 << (MAX_MVSEARCH_STEPS-1))
+
+extern void print_mode_context(void);
+extern int vp8_mv_bit_cost(int_mv *mv, int_mv *ref, int *mvcost[2], int Weight);
+extern void vp8_init_dsmotion_compensation(MACROBLOCK *x, int stride);
+extern void vp8_init3smotion_compensation(MACROBLOCK *x, int stride);
+
+
+extern int vp8_hex_search
+(
+ MACROBLOCK *x,
+ BLOCK *b,
+ BLOCKD *d,
+ int_mv *ref_mv,
+ int_mv *best_mv,
+ int search_param,
+ int error_per_bit,
+ const vp8_variance_fn_ptr_t *vf,
+ int *mvsadcost[2],
+ int *mvcost[2],
+ int_mv *center_mv
+);
+
+typedef int (fractional_mv_step_fp)
+ (MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *bestmv, int_mv *ref_mv,
+ int error_per_bit, const vp8_variance_fn_ptr_t *vfp, int *mvcost[2],
+ int *distortion, unsigned int *sse);
+
+extern fractional_mv_step_fp vp8_find_best_sub_pixel_step_iteratively;
+extern fractional_mv_step_fp vp8_find_best_sub_pixel_step;
+extern fractional_mv_step_fp vp8_find_best_half_pixel_step;
+extern fractional_mv_step_fp vp8_skip_fractional_mv_step;
+
+typedef int (*vp8_full_search_fn_t)
+ (
+ MACROBLOCK *x,
+ BLOCK *b,
+ BLOCKD *d,
+ int_mv *ref_mv,
+ int sad_per_bit,
+ int distance,
+ vp8_variance_fn_ptr_t *fn_ptr,
+ int *mvcost[2],
+ int_mv *center_mv
+ );
+
+typedef int (*vp8_refining_search_fn_t)
+ (
+ MACROBLOCK *x,
+ BLOCK *b,
+ BLOCKD *d,
+ int_mv *ref_mv,
+ int sad_per_bit,
+ int distance,
+ vp8_variance_fn_ptr_t *fn_ptr,
+ int *mvcost[2],
+ int_mv *center_mv
+ );
+
+typedef int (*vp8_diamond_search_fn_t)
+ (
+ MACROBLOCK *x,
+ BLOCK *b,
+ BLOCKD *d,
+ int_mv *ref_mv,
+ int_mv *best_mv,
+ int search_param,
+ int sad_per_bit,
+ int *num00,
+ vp8_variance_fn_ptr_t *fn_ptr,
+ int *mvcost[2],
+ int_mv *center_mv
+ );
+
+#endif
diff --git a/libvpx/vp8/encoder/modecosts.c b/libvpx/vp8/encoder/modecosts.c
new file mode 100644
index 0000000..c61563c
--- /dev/null
+++ b/libvpx/vp8/encoder/modecosts.c
@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/common/blockd.h"
+#include "onyx_int.h"
+#include "treewriter.h"
+#include "vp8/common/entropymode.h"
+
+
+void vp8_init_mode_costs(VP8_COMP *c)
+{
+ VP8_COMMON *x = &c->common;
+ struct rd_costs_struct *rd_costs = &c->rd_costs;
+
+ {
+ const vp8_tree_p T = vp8_bmode_tree;
+
+ int i = 0;
+
+ do
+ {
+ int j = 0;
+
+ do
+ {
+ vp8_cost_tokens(rd_costs->bmode_costs[i][j],
+ vp8_kf_bmode_prob[i][j], T);
+ }
+ while (++j < VP8_BINTRAMODES);
+ }
+ while (++i < VP8_BINTRAMODES);
+
+ vp8_cost_tokens(rd_costs->inter_bmode_costs, x->fc.bmode_prob, T);
+ }
+ vp8_cost_tokens(rd_costs->inter_bmode_costs, x->fc.sub_mv_ref_prob,
+ vp8_sub_mv_ref_tree);
+
+ vp8_cost_tokens(rd_costs->mbmode_cost[1], x->fc.ymode_prob, vp8_ymode_tree);
+ vp8_cost_tokens(rd_costs->mbmode_cost[0], vp8_kf_ymode_prob,
+ vp8_kf_ymode_tree);
+
+ vp8_cost_tokens(rd_costs->intra_uv_mode_cost[1], x->fc.uv_mode_prob,
+ vp8_uv_mode_tree);
+ vp8_cost_tokens(rd_costs->intra_uv_mode_cost[0], vp8_kf_uv_mode_prob,
+ vp8_uv_mode_tree);
+}
diff --git a/vp8/encoder/modecosts.h b/libvpx/vp8/encoder/modecosts.h
similarity index 100%
rename from vp8/encoder/modecosts.h
rename to libvpx/vp8/encoder/modecosts.h
diff --git a/libvpx/vp8/encoder/mr_dissim.c b/libvpx/vp8/encoder/mr_dissim.c
new file mode 100644
index 0000000..71218cc
--- /dev/null
+++ b/libvpx/vp8/encoder/mr_dissim.c
@@ -0,0 +1,236 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <limits.h>
+#include "vpx_config.h"
+#include "onyx_int.h"
+#include "mr_dissim.h"
+#include "vpx_mem/vpx_mem.h"
+#include "rdopt.h"
+
+void vp8_cal_low_res_mb_cols(VP8_COMP *cpi)
+{
+ int low_res_w;
+
+ /* Support arbitrary down-sampling factor */
+ unsigned int iw = cpi->oxcf.Width*cpi->oxcf.mr_down_sampling_factor.den
+ + cpi->oxcf.mr_down_sampling_factor.num - 1;
+
+ low_res_w = iw/cpi->oxcf.mr_down_sampling_factor.num;
+ cpi->mr_low_res_mb_cols = ((low_res_w + 15) >> 4);
+}
+
+#define GET_MV(x) \
+if(x->mbmi.ref_frame !=INTRA_FRAME) \
+{ \
+ mvx[cnt] = x->mbmi.mv.as_mv.row; \
+ mvy[cnt] = x->mbmi.mv.as_mv.col; \
+ cnt++; \
+}
+
+#define GET_MV_SIGN(x) \
+if(x->mbmi.ref_frame !=INTRA_FRAME) \
+{ \
+ mvx[cnt] = x->mbmi.mv.as_mv.row; \
+ mvy[cnt] = x->mbmi.mv.as_mv.col; \
+ if (cm->ref_frame_sign_bias[x->mbmi.ref_frame] \
+ != cm->ref_frame_sign_bias[tmp->mbmi.ref_frame]) \
+ { \
+ mvx[cnt] *= -1; \
+ mvy[cnt] *= -1; \
+ } \
+ cnt++; \
+}
+
+void vp8_cal_dissimilarity(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+ int i;
+
+ /* Note: The first row & first column in mip are outside the frame, which
+ * were initialized to all 0.(ref_frame, mode, mv...)
+ * Their ref_frame = 0 means they won't be counted in the following
+ * calculation.
+ */
+ if (cpi->oxcf.mr_total_resolutions >1
+ && cpi->oxcf.mr_encoder_id < (cpi->oxcf.mr_total_resolutions - 1))
+ {
+ /* Store info for show/no-show frames for supporting alt_ref.
+ * If parent frame is alt_ref, child has one too.
+ */
+ LOWER_RES_FRAME_INFO* store_info
+ = (LOWER_RES_FRAME_INFO*)cpi->oxcf.mr_low_res_mode_info;
+
+ store_info->frame_type = cm->frame_type;
+
+ if(cm->frame_type != KEY_FRAME)
+ {
+ store_info->is_frame_dropped = 0;
+ for (i = 1; i < MAX_REF_FRAMES; i++)
+ store_info->low_res_ref_frames[i] = cpi->current_ref_frames[i];
+ }
+
+ if(cm->frame_type != KEY_FRAME)
+ {
+ int mb_row;
+ int mb_col;
+ /* Point to beginning of allocated MODE_INFO arrays. */
+ MODE_INFO *tmp = cm->mip + cm->mode_info_stride;
+ LOWER_RES_MB_INFO* store_mode_info = store_info->mb_info;
+
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row ++)
+ {
+ tmp++;
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col ++)
+ {
+ int dissim = INT_MAX;
+
+ if(tmp->mbmi.ref_frame !=INTRA_FRAME)
+ {
+ int mvx[8];
+ int mvy[8];
+ int mmvx;
+ int mmvy;
+ int cnt=0;
+ const MODE_INFO *here = tmp;
+ const MODE_INFO *above = here - cm->mode_info_stride;
+ const MODE_INFO *left = here - 1;
+ const MODE_INFO *aboveleft = above - 1;
+ const MODE_INFO *aboveright = NULL;
+ const MODE_INFO *right = NULL;
+ const MODE_INFO *belowleft = NULL;
+ const MODE_INFO *below = NULL;
+ const MODE_INFO *belowright = NULL;
+
+ /* If alternate reference frame is used, we have to
+ * check sign of MV. */
+ if(cpi->oxcf.play_alternate)
+ {
+ /* Gather mv of neighboring MBs */
+ GET_MV_SIGN(above)
+ GET_MV_SIGN(left)
+ GET_MV_SIGN(aboveleft)
+
+ if(mb_col < (cm->mb_cols-1))
+ {
+ right = here + 1;
+ aboveright = above + 1;
+ GET_MV_SIGN(right)
+ GET_MV_SIGN(aboveright)
+ }
+
+ if(mb_row < (cm->mb_rows-1))
+ {
+ below = here + cm->mode_info_stride;
+ belowleft = below - 1;
+ GET_MV_SIGN(below)
+ GET_MV_SIGN(belowleft)
+ }
+
+ if(mb_col < (cm->mb_cols-1)
+ && mb_row < (cm->mb_rows-1))
+ {
+ belowright = below + 1;
+ GET_MV_SIGN(belowright)
+ }
+ }else
+ {
+ /* No alt_ref and gather mv of neighboring MBs */
+ GET_MV(above)
+ GET_MV(left)
+ GET_MV(aboveleft)
+
+ if(mb_col < (cm->mb_cols-1))
+ {
+ right = here + 1;
+ aboveright = above + 1;
+ GET_MV(right)
+ GET_MV(aboveright)
+ }
+
+ if(mb_row < (cm->mb_rows-1))
+ {
+ below = here + cm->mode_info_stride;
+ belowleft = below - 1;
+ GET_MV(below)
+ GET_MV(belowleft)
+ }
+
+ if(mb_col < (cm->mb_cols-1)
+ && mb_row < (cm->mb_rows-1))
+ {
+ belowright = below + 1;
+ GET_MV(belowright)
+ }
+ }
+
+ if (cnt > 0)
+ {
+ int max_mvx = mvx[0];
+ int min_mvx = mvx[0];
+ int max_mvy = mvy[0];
+ int min_mvy = mvy[0];
+ int i;
+
+ if (cnt > 1)
+ {
+ for (i=1; i< cnt; i++)
+ {
+ if (mvx[i] > max_mvx) max_mvx = mvx[i];
+ else if (mvx[i] < min_mvx) min_mvx = mvx[i];
+ if (mvy[i] > max_mvy) max_mvy = mvy[i];
+ else if (mvy[i] < min_mvy) min_mvy = mvy[i];
+ }
+ }
+
+ mmvx = MAX(abs(min_mvx - here->mbmi.mv.as_mv.row),
+ abs(max_mvx - here->mbmi.mv.as_mv.row));
+ mmvy = MAX(abs(min_mvy - here->mbmi.mv.as_mv.col),
+ abs(max_mvy - here->mbmi.mv.as_mv.col));
+ dissim = MAX(mmvx, mmvy);
+ }
+ }
+
+ /* Store mode info for next resolution encoding */
+ store_mode_info->mode = tmp->mbmi.mode;
+ store_mode_info->ref_frame = tmp->mbmi.ref_frame;
+ store_mode_info->mv.as_int = tmp->mbmi.mv.as_int;
+ store_mode_info->dissim = dissim;
+ tmp++;
+ store_mode_info++;
+ }
+ }
+ }
+ }
+}
+
+/* This function is called only when this frame is dropped at current
+ resolution level. */
+void vp8_store_drop_frame_info(VP8_COMP *cpi)
+{
+ /* If the frame is dropped in lower-resolution encoding, this information
+ is passed to higher resolution level so that the encoder knows there
+ is no mode & motion info available.
+ */
+ if (cpi->oxcf.mr_total_resolutions >1
+ && cpi->oxcf.mr_encoder_id < (cpi->oxcf.mr_total_resolutions - 1))
+ {
+ /* Store info for show/no-show frames for supporting alt_ref.
+ * If parent frame is alt_ref, child has one too.
+ */
+ LOWER_RES_FRAME_INFO* store_info
+ = (LOWER_RES_FRAME_INFO*)cpi->oxcf.mr_low_res_mode_info;
+
+ /* Set frame_type to be INTER_FRAME since we won't drop key frame. */
+ store_info->frame_type = INTER_FRAME;
+ store_info->is_frame_dropped = 1;
+ }
+}
diff --git a/libvpx/vp8/encoder/mr_dissim.h b/libvpx/vp8/encoder/mr_dissim.h
new file mode 100644
index 0000000..f8cb135
--- /dev/null
+++ b/libvpx/vp8/encoder/mr_dissim.h
@@ -0,0 +1,20 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_MR_DISSIM_H
+#define __INC_MR_DISSIM_H
+#include "vpx_config.h"
+
+extern void vp8_cal_low_res_mb_cols(VP8_COMP *cpi);
+extern void vp8_cal_dissimilarity(VP8_COMP *cpi);
+extern void vp8_store_drop_frame_info(VP8_COMP *cpi);
+
+#endif
diff --git a/libvpx/vp8/encoder/onyx_if.c b/libvpx/vp8/encoder/onyx_if.c
new file mode 100644
index 0000000..c7d81b1
--- /dev/null
+++ b/libvpx/vp8/encoder/onyx_if.c
@@ -0,0 +1,5568 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vp8/common/onyxc_int.h"
+#include "vp8/common/blockd.h"
+#include "onyx_int.h"
+#include "vp8/common/systemdependent.h"
+#include "quantize.h"
+#include "vp8/common/alloccommon.h"
+#include "mcomp.h"
+#include "firstpass.h"
+#include "psnr.h"
+#include "vpx_scale/vpxscale.h"
+#include "vp8/common/extend.h"
+#include "ratectrl.h"
+#include "vp8/common/quant_common.h"
+#include "segmentation.h"
+#if CONFIG_POSTPROC
+#include "vp8/common/postproc.h"
+#endif
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/swapyv12buffer.h"
+#include "vp8/common/threading.h"
+#include "vpx_ports/vpx_timer.h"
+#if ARCH_ARM
+#include "vpx_ports/arm.h"
+#endif
+#if CONFIG_MULTI_RES_ENCODING
+#include "mr_dissim.h"
+#endif
+#include "encodeframe.h"
+
+#include <math.h>
+#include <stdio.h>
+#include <limits.h>
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+extern int vp8_update_coef_context(VP8_COMP *cpi);
+extern void vp8_update_coef_probs(VP8_COMP *cpi);
+#endif
+
+extern void vp8cx_pick_filter_level_fast(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi);
+extern void vp8cx_set_alt_lf_level(VP8_COMP *cpi, int filt_val);
+extern void vp8cx_pick_filter_level(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi);
+
+extern void vp8_deblock_frame(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *post, int filt_lvl, int low_var_thresh, int flag);
+extern void print_parms(VP8_CONFIG *ocf, char *filenam);
+extern unsigned int vp8_get_processor_freq();
+extern void print_tree_update_probs();
+extern int vp8cx_create_encoder_threads(VP8_COMP *cpi);
+extern void vp8cx_remove_encoder_threads(VP8_COMP *cpi);
+
+int vp8_estimate_entropy_savings(VP8_COMP *cpi);
+
+int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest);
+
+extern void vp8_temporal_filter_prepare_c(VP8_COMP *cpi, int distance);
+
+static void set_default_lf_deltas(VP8_COMP *cpi);
+
+extern const int vp8_gf_interval_table[101];
+
+#if CONFIG_INTERNAL_STATS
+#include "math.h"
+
+extern double vp8_calc_ssim
+(
+ YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *dest,
+ int lumamask,
+ double *weight
+);
+
+
+extern double vp8_calc_ssimg
+(
+ YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *dest,
+ double *ssim_y,
+ double *ssim_u,
+ double *ssim_v
+);
+
+
+#endif
+
+
+#ifdef OUTPUT_YUV_SRC
+FILE *yuv_file;
+#endif
+
+#if 0
+FILE *framepsnr;
+FILE *kf_list;
+FILE *keyfile;
+#endif
+
+#if 0
+extern int skip_true_count;
+extern int skip_false_count;
+#endif
+
+
+#ifdef ENTROPY_STATS
+extern int intra_mode_stats[10][10][10];
+#endif
+
+#ifdef SPEEDSTATS
+unsigned int frames_at_speed[16] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+unsigned int tot_pm = 0;
+unsigned int cnt_pm = 0;
+unsigned int tot_ef = 0;
+unsigned int cnt_ef = 0;
+#endif
+
+#ifdef MODE_STATS
+extern unsigned __int64 Sectionbits[50];
+extern int y_modes[5] ;
+extern int uv_modes[4] ;
+extern int b_modes[10] ;
+
+extern int inter_y_modes[10] ;
+extern int inter_uv_modes[4] ;
+extern unsigned int inter_b_modes[15];
+#endif
+
+extern const int vp8_bits_per_mb[2][QINDEX_RANGE];
+
+extern const int qrounding_factors[129];
+extern const int qzbin_factors[129];
+extern void vp8cx_init_quantizer(VP8_COMP *cpi);
+extern const int vp8cx_base_skip_false_prob[128];
+
+/* Tables relating active max Q to active min Q */
+static const unsigned char kf_low_motion_minq[QINDEX_RANGE] =
+{
+ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+ 0,0,0,0,1,1,1,1,1,1,1,1,2,2,2,2,
+ 3,3,3,3,3,3,4,4,4,5,5,5,5,5,6,6,
+ 6,6,7,7,8,8,8,8,9,9,10,10,10,10,11,11,
+ 11,11,12,12,13,13,13,13,14,14,15,15,15,15,16,16,
+ 16,16,17,17,18,18,18,18,19,20,20,21,21,22,23,23
+};
+static const unsigned char kf_high_motion_minq[QINDEX_RANGE] =
+{
+ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+ 1,1,1,1,1,1,1,1,2,2,2,2,3,3,3,3,
+ 3,3,3,3,4,4,4,4,5,5,5,5,5,5,6,6,
+ 6,6,7,7,8,8,8,8,9,9,10,10,10,10,11,11,
+ 11,11,12,12,13,13,13,13,14,14,15,15,15,15,16,16,
+ 16,16,17,17,18,18,18,18,19,19,20,20,20,20,21,21,
+ 21,21,22,22,23,23,24,25,25,26,26,27,28,28,29,30
+};
+static const unsigned char gf_low_motion_minq[QINDEX_RANGE] =
+{
+ 0,0,0,0,1,1,1,1,1,1,1,1,2,2,2,2,
+ 3,3,3,3,4,4,4,4,5,5,5,5,6,6,6,6,
+ 7,7,7,7,8,8,8,8,9,9,9,9,10,10,10,10,
+ 11,11,12,12,13,13,14,14,15,15,16,16,17,17,18,18,
+ 19,19,20,20,21,21,22,22,23,23,24,24,25,25,26,26,
+ 27,27,28,28,29,29,30,30,31,31,32,32,33,33,34,34,
+ 35,35,36,36,37,37,38,38,39,39,40,40,41,41,42,42,
+ 43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58
+};
+static const unsigned char gf_mid_motion_minq[QINDEX_RANGE] =
+{
+ 0,0,0,0,1,1,1,1,1,1,2,2,3,3,3,4,
+ 4,4,5,5,5,6,6,6,7,7,7,8,8,8,9,9,
+ 9,10,10,10,10,11,11,11,12,12,12,12,13,13,13,14,
+ 14,14,15,15,16,16,17,17,18,18,19,19,20,20,21,21,
+ 22,22,23,23,24,24,25,25,26,26,27,27,28,28,29,29,
+ 30,30,31,31,32,32,33,33,34,34,35,35,36,36,37,37,
+ 38,39,39,40,40,41,41,42,42,43,43,44,45,46,47,48,
+ 49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64
+};
+static const unsigned char gf_high_motion_minq[QINDEX_RANGE] =
+{
+ 0,0,0,0,1,1,1,1,1,2,2,2,3,3,3,4,
+ 4,4,5,5,5,6,6,6,7,7,7,8,8,8,9,9,
+ 9,10,10,10,11,11,12,12,13,13,14,14,15,15,16,16,
+ 17,17,18,18,19,19,20,20,21,21,22,22,23,23,24,24,
+ 25,25,26,26,27,27,28,28,29,29,30,30,31,31,32,32,
+ 33,33,34,34,35,35,36,36,37,37,38,38,39,39,40,40,
+ 41,41,42,42,43,44,45,46,47,48,49,50,51,52,53,54,
+ 55,56,57,58,59,60,62,64,66,68,70,72,74,76,78,80
+};
+static const unsigned char inter_minq[QINDEX_RANGE] =
+{
+ 0,0,1,1,2,3,3,4,4,5,6,6,7,8,8,9,
+ 9,10,11,11,12,13,13,14,15,15,16,17,17,18,19,20,
+ 20,21,22,22,23,24,24,25,26,27,27,28,29,30,30,31,
+ 32,33,33,34,35,36,36,37,38,39,39,40,41,42,42,43,
+ 44,45,46,46,47,48,49,50,50,51,52,53,54,55,55,56,
+ 57,58,59,60,60,61,62,63,64,65,66,67,67,68,69,70,
+ 71,72,73,74,75,75,76,77,78,79,80,81,82,83,84,85,
+ 86,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100
+};
+
+#ifdef PACKET_TESTING
+extern FILE *vpxlogc;
+#endif
+
+static void save_layer_context(VP8_COMP *cpi)
+{
+ LAYER_CONTEXT *lc = &cpi->layer_context[cpi->current_layer];
+
+ /* Save layer dependent coding state */
+ lc->target_bandwidth = cpi->target_bandwidth;
+ lc->starting_buffer_level = cpi->oxcf.starting_buffer_level;
+ lc->optimal_buffer_level = cpi->oxcf.optimal_buffer_level;
+ lc->maximum_buffer_size = cpi->oxcf.maximum_buffer_size;
+ lc->starting_buffer_level_in_ms = cpi->oxcf.starting_buffer_level_in_ms;
+ lc->optimal_buffer_level_in_ms = cpi->oxcf.optimal_buffer_level_in_ms;
+ lc->maximum_buffer_size_in_ms = cpi->oxcf.maximum_buffer_size_in_ms;
+ lc->buffer_level = cpi->buffer_level;
+ lc->bits_off_target = cpi->bits_off_target;
+ lc->total_actual_bits = cpi->total_actual_bits;
+ lc->worst_quality = cpi->worst_quality;
+ lc->active_worst_quality = cpi->active_worst_quality;
+ lc->best_quality = cpi->best_quality;
+ lc->active_best_quality = cpi->active_best_quality;
+ lc->ni_av_qi = cpi->ni_av_qi;
+ lc->ni_tot_qi = cpi->ni_tot_qi;
+ lc->ni_frames = cpi->ni_frames;
+ lc->avg_frame_qindex = cpi->avg_frame_qindex;
+ lc->rate_correction_factor = cpi->rate_correction_factor;
+ lc->key_frame_rate_correction_factor = cpi->key_frame_rate_correction_factor;
+ lc->gf_rate_correction_factor = cpi->gf_rate_correction_factor;
+ lc->zbin_over_quant = cpi->zbin_over_quant;
+ lc->inter_frame_target = cpi->inter_frame_target;
+ lc->total_byte_count = cpi->total_byte_count;
+ lc->filter_level = cpi->common.filter_level;
+
+ lc->last_frame_percent_intra = cpi->last_frame_percent_intra;
+
+ memcpy (lc->count_mb_ref_frame_usage,
+ cpi->count_mb_ref_frame_usage,
+ sizeof(cpi->count_mb_ref_frame_usage));
+}
+
+static void restore_layer_context(VP8_COMP *cpi, const int layer)
+{
+ LAYER_CONTEXT *lc = &cpi->layer_context[layer];
+
+ /* Restore layer dependent coding state */
+ cpi->current_layer = layer;
+ cpi->target_bandwidth = lc->target_bandwidth;
+ cpi->oxcf.target_bandwidth = lc->target_bandwidth;
+ cpi->oxcf.starting_buffer_level = lc->starting_buffer_level;
+ cpi->oxcf.optimal_buffer_level = lc->optimal_buffer_level;
+ cpi->oxcf.maximum_buffer_size = lc->maximum_buffer_size;
+ cpi->oxcf.starting_buffer_level_in_ms = lc->starting_buffer_level_in_ms;
+ cpi->oxcf.optimal_buffer_level_in_ms = lc->optimal_buffer_level_in_ms;
+ cpi->oxcf.maximum_buffer_size_in_ms = lc->maximum_buffer_size_in_ms;
+ cpi->buffer_level = lc->buffer_level;
+ cpi->bits_off_target = lc->bits_off_target;
+ cpi->total_actual_bits = lc->total_actual_bits;
+ cpi->active_worst_quality = lc->active_worst_quality;
+ cpi->active_best_quality = lc->active_best_quality;
+ cpi->ni_av_qi = lc->ni_av_qi;
+ cpi->ni_tot_qi = lc->ni_tot_qi;
+ cpi->ni_frames = lc->ni_frames;
+ cpi->avg_frame_qindex = lc->avg_frame_qindex;
+ cpi->rate_correction_factor = lc->rate_correction_factor;
+ cpi->key_frame_rate_correction_factor = lc->key_frame_rate_correction_factor;
+ cpi->gf_rate_correction_factor = lc->gf_rate_correction_factor;
+ cpi->zbin_over_quant = lc->zbin_over_quant;
+ cpi->inter_frame_target = lc->inter_frame_target;
+ cpi->total_byte_count = lc->total_byte_count;
+ cpi->common.filter_level = lc->filter_level;
+
+ cpi->last_frame_percent_intra = lc->last_frame_percent_intra;
+
+ memcpy (cpi->count_mb_ref_frame_usage,
+ lc->count_mb_ref_frame_usage,
+ sizeof(cpi->count_mb_ref_frame_usage));
+}
+
+static void setup_features(VP8_COMP *cpi)
+{
+ // If segmentation enabled set the update flags
+ if ( cpi->mb.e_mbd.segmentation_enabled )
+ {
+ cpi->mb.e_mbd.update_mb_segmentation_map = 1;
+ cpi->mb.e_mbd.update_mb_segmentation_data = 1;
+ }
+ else
+ {
+ cpi->mb.e_mbd.update_mb_segmentation_map = 0;
+ cpi->mb.e_mbd.update_mb_segmentation_data = 0;
+ }
+
+ cpi->mb.e_mbd.mode_ref_lf_delta_enabled = 0;
+ cpi->mb.e_mbd.mode_ref_lf_delta_update = 0;
+ vpx_memset(cpi->mb.e_mbd.ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
+ vpx_memset(cpi->mb.e_mbd.mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
+ vpx_memset(cpi->mb.e_mbd.last_ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
+ vpx_memset(cpi->mb.e_mbd.last_mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
+
+ set_default_lf_deltas(cpi);
+
+}
+
+
+static void dealloc_raw_frame_buffers(VP8_COMP *cpi);
+
+
+static void dealloc_compressor_data(VP8_COMP *cpi)
+{
+ vpx_free(cpi->tplist);
+ cpi->tplist = NULL;
+
+ /* Delete last frame MV storage buffers */
+ vpx_free(cpi->lfmv);
+ cpi->lfmv = 0;
+
+ vpx_free(cpi->lf_ref_frame_sign_bias);
+ cpi->lf_ref_frame_sign_bias = 0;
+
+ vpx_free(cpi->lf_ref_frame);
+ cpi->lf_ref_frame = 0;
+
+ /* Delete sementation map */
+ vpx_free(cpi->segmentation_map);
+ cpi->segmentation_map = 0;
+
+ vpx_free(cpi->active_map);
+ cpi->active_map = 0;
+
+ vp8_de_alloc_frame_buffers(&cpi->common);
+
+ vp8_yv12_de_alloc_frame_buffer(&cpi->pick_lf_lvl_frame);
+ vp8_yv12_de_alloc_frame_buffer(&cpi->scaled_source);
+ dealloc_raw_frame_buffers(cpi);
+
+ vpx_free(cpi->tok);
+ cpi->tok = 0;
+
+ /* Structure used to monitor GF usage */
+ vpx_free(cpi->gf_active_flags);
+ cpi->gf_active_flags = 0;
+
+ /* Activity mask based per mb zbin adjustments */
+ vpx_free(cpi->mb_activity_map);
+ cpi->mb_activity_map = 0;
+
+ vpx_free(cpi->mb.pip);
+ cpi->mb.pip = 0;
+
+#if CONFIG_MULTITHREAD
+ vpx_free(cpi->mt_current_mb_col);
+ cpi->mt_current_mb_col = NULL;
+#endif
+}
+
+static void enable_segmentation(VP8_COMP *cpi)
+{
+ /* Set the appropriate feature bit */
+ cpi->mb.e_mbd.segmentation_enabled = 1;
+ cpi->mb.e_mbd.update_mb_segmentation_map = 1;
+ cpi->mb.e_mbd.update_mb_segmentation_data = 1;
+}
+static void disable_segmentation(VP8_COMP *cpi)
+{
+ /* Clear the appropriate feature bit */
+ cpi->mb.e_mbd.segmentation_enabled = 0;
+}
+
+/* Valid values for a segment are 0 to 3
+ * Segmentation map is arrange as [Rows][Columns]
+ */
+static void set_segmentation_map(VP8_COMP *cpi, unsigned char *segmentation_map)
+{
+ /* Copy in the new segmentation map */
+ vpx_memcpy(cpi->segmentation_map, segmentation_map, (cpi->common.mb_rows * cpi->common.mb_cols));
+
+ /* Signal that the map should be updated. */
+ cpi->mb.e_mbd.update_mb_segmentation_map = 1;
+ cpi->mb.e_mbd.update_mb_segmentation_data = 1;
+}
+
+/* The values given for each segment can be either deltas (from the default
+ * value chosen for the frame) or absolute values.
+ *
+ * Valid range for abs values is:
+ * (0-127 for MB_LVL_ALT_Q), (0-63 for SEGMENT_ALT_LF)
+ * Valid range for delta values are:
+ * (+/-127 for MB_LVL_ALT_Q), (+/-63 for SEGMENT_ALT_LF)
+ *
+ * abs_delta = SEGMENT_DELTADATA (deltas)
+ * abs_delta = SEGMENT_ABSDATA (use the absolute values given).
+ *
+ */
+static void set_segment_data(VP8_COMP *cpi, signed char *feature_data, unsigned char abs_delta)
+{
+ cpi->mb.e_mbd.mb_segement_abs_delta = abs_delta;
+ vpx_memcpy(cpi->segment_feature_data, feature_data, sizeof(cpi->segment_feature_data));
+}
+
+
+static void segmentation_test_function(VP8_COMP *cpi)
+{
+ unsigned char *seg_map;
+ signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
+
+ // Create a temporary map for segmentation data.
+ CHECK_MEM_ERROR(seg_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
+
+ // Set the segmentation Map
+ set_segmentation_map(cpi, seg_map);
+
+ // Activate segmentation.
+ enable_segmentation(cpi);
+
+ // Set up the quant segment data
+ feature_data[MB_LVL_ALT_Q][0] = 0;
+ feature_data[MB_LVL_ALT_Q][1] = 4;
+ feature_data[MB_LVL_ALT_Q][2] = 0;
+ feature_data[MB_LVL_ALT_Q][3] = 0;
+ // Set up the loop segment data
+ feature_data[MB_LVL_ALT_LF][0] = 0;
+ feature_data[MB_LVL_ALT_LF][1] = 0;
+ feature_data[MB_LVL_ALT_LF][2] = 0;
+ feature_data[MB_LVL_ALT_LF][3] = 0;
+
+ // Initialise the feature data structure
+ // SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
+ set_segment_data(cpi, &feature_data[0][0], SEGMENT_DELTADATA);
+
+ // Delete sementation map
+ vpx_free(seg_map);
+
+ seg_map = 0;
+}
+
+/* A simple function to cyclically refresh the background at a lower Q */
+static void cyclic_background_refresh(VP8_COMP *cpi, int Q, int lf_adjustment)
+{
+ unsigned char *seg_map = cpi->segmentation_map;
+ signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
+ int i;
+ int block_count = cpi->cyclic_refresh_mode_max_mbs_perframe;
+ int mbs_in_frame = cpi->common.mb_rows * cpi->common.mb_cols;
+
+ cpi->cyclic_refresh_q = Q / 2;
+
+ // Set every macroblock to be eligible for update.
+ // For key frame this will reset seg map to 0.
+ vpx_memset(cpi->segmentation_map, 0, mbs_in_frame);
+
+ if (cpi->common.frame_type != KEY_FRAME)
+ {
+ /* Cycle through the macro_block rows */
+ /* MB loop to set local segmentation map */
+ i = cpi->cyclic_refresh_mode_index;
+ assert(i < mbs_in_frame);
+ do
+ {
+ /* If the MB is as a candidate for clean up then mark it for
+ * possible boost/refresh (segment 1) The segment id may get
+ * reset to 0 later if the MB gets coded anything other than
+ * last frame 0,0 as only (last frame 0,0) MBs are eligable for
+ * refresh : that is to say Mbs likely to be background blocks.
+ */
+ if (cpi->cyclic_refresh_map[i] == 0)
+ {
+ seg_map[i] = 1;
+ block_count --;
+ }
+ else if (cpi->cyclic_refresh_map[i] < 0)
+ cpi->cyclic_refresh_map[i]++;
+
+ i++;
+ if (i == mbs_in_frame)
+ i = 0;
+
+ }
+ while(block_count && i != cpi->cyclic_refresh_mode_index);
+
+ cpi->cyclic_refresh_mode_index = i;
+ }
+
+ /* Activate segmentation. */
+ cpi->mb.e_mbd.update_mb_segmentation_map = 1;
+ cpi->mb.e_mbd.update_mb_segmentation_data = 1;
+ enable_segmentation(cpi);
+
+ /* Set up the quant segment data */
+ feature_data[MB_LVL_ALT_Q][0] = 0;
+ feature_data[MB_LVL_ALT_Q][1] = (cpi->cyclic_refresh_q - Q);
+ feature_data[MB_LVL_ALT_Q][2] = 0;
+ feature_data[MB_LVL_ALT_Q][3] = 0;
+
+ /* Set up the loop segment data */
+ feature_data[MB_LVL_ALT_LF][0] = 0;
+ feature_data[MB_LVL_ALT_LF][1] = lf_adjustment;
+ feature_data[MB_LVL_ALT_LF][2] = 0;
+ feature_data[MB_LVL_ALT_LF][3] = 0;
+
+ /* Initialise the feature data structure */
+ set_segment_data(cpi, &feature_data[0][0], SEGMENT_DELTADATA);
+
+}
+
+static void set_default_lf_deltas(VP8_COMP *cpi)
+{
+ cpi->mb.e_mbd.mode_ref_lf_delta_enabled = 1;
+ cpi->mb.e_mbd.mode_ref_lf_delta_update = 1;
+
+ vpx_memset(cpi->mb.e_mbd.ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
+ vpx_memset(cpi->mb.e_mbd.mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
+
+ /* Test of ref frame deltas */
+ cpi->mb.e_mbd.ref_lf_deltas[INTRA_FRAME] = 2;
+ cpi->mb.e_mbd.ref_lf_deltas[LAST_FRAME] = 0;
+ cpi->mb.e_mbd.ref_lf_deltas[GOLDEN_FRAME] = -2;
+ cpi->mb.e_mbd.ref_lf_deltas[ALTREF_FRAME] = -2;
+
+ cpi->mb.e_mbd.mode_lf_deltas[0] = 4; /* BPRED */
+
+ if(cpi->oxcf.Mode == MODE_REALTIME)
+ cpi->mb.e_mbd.mode_lf_deltas[1] = -12; /* Zero */
+ else
+ cpi->mb.e_mbd.mode_lf_deltas[1] = -2; /* Zero */
+
+ cpi->mb.e_mbd.mode_lf_deltas[2] = 2; /* New mv */
+ cpi->mb.e_mbd.mode_lf_deltas[3] = 4; /* Split mv */
+}
+
+/* Convenience macros for mapping speed and mode into a continuous
+ * range
+ */
+#define GOOD(x) (x+1)
+#define RT(x) (x+7)
+
+static int speed_map(int speed, const int *map)
+{
+ int res;
+
+ do
+ {
+ res = *map++;
+ } while(speed >= *map++);
+ return res;
+}
+
+static const int thresh_mult_map_znn[] = {
+ /* map common to zero, nearest, and near */
+ 0, GOOD(2), 1500, GOOD(3), 2000, RT(0), 1000, RT(2), 2000, INT_MAX
+};
+
+static const int thresh_mult_map_vhpred[] = {
+ 1000, GOOD(2), 1500, GOOD(3), 2000, RT(0), 1000, RT(1), 2000,
+ RT(7), INT_MAX, INT_MAX
+};
+
+static const int thresh_mult_map_bpred[] = {
+ 2000, GOOD(0), 2500, GOOD(2), 5000, GOOD(3), 7500, RT(0), 2500, RT(1), 5000,
+ RT(6), INT_MAX, INT_MAX
+};
+
+static const int thresh_mult_map_tm[] = {
+ 1000, GOOD(2), 1500, GOOD(3), 2000, RT(0), 0, RT(1), 1000, RT(2), 2000,
+ RT(7), INT_MAX, INT_MAX
+};
+
+static const int thresh_mult_map_new1[] = {
+ 1000, GOOD(2), 2000, RT(0), 2000, INT_MAX
+};
+
+static const int thresh_mult_map_new2[] = {
+ 1000, GOOD(2), 2000, GOOD(3), 2500, GOOD(5), 4000, RT(0), 2000, RT(2), 2500,
+ RT(5), 4000, INT_MAX
+};
+
+static const int thresh_mult_map_split1[] = {
+ 2500, GOOD(0), 1700, GOOD(2), 10000, GOOD(3), 25000, GOOD(4), INT_MAX,
+ RT(0), 5000, RT(1), 10000, RT(2), 25000, RT(3), INT_MAX, INT_MAX
+};
+
+static const int thresh_mult_map_split2[] = {
+ 5000, GOOD(0), 4500, GOOD(2), 20000, GOOD(3), 50000, GOOD(4), INT_MAX,
+ RT(0), 10000, RT(1), 20000, RT(2), 50000, RT(3), INT_MAX, INT_MAX
+};
+
+static const int mode_check_freq_map_zn2[] = {
+ /* {zero,nearest}{2,3} */
+ 0, RT(10), 1<<1, RT(11), 1<<2, RT(12), 1<<3, INT_MAX
+};
+
+static const int mode_check_freq_map_vhbpred[] = {
+ 0, GOOD(5), 2, RT(0), 0, RT(3), 2, RT(5), 4, INT_MAX
+};
+
+static const int mode_check_freq_map_near2[] = {
+ 0, GOOD(5), 2, RT(0), 0, RT(3), 2, RT(10), 1<<2, RT(11), 1<<3, RT(12), 1<<4,
+ INT_MAX
+};
+
+static const int mode_check_freq_map_new1[] = {
+ 0, RT(10), 1<<1, RT(11), 1<<2, RT(12), 1<<3, INT_MAX
+};
+
+static const int mode_check_freq_map_new2[] = {
+ 0, GOOD(5), 4, RT(0), 0, RT(3), 4, RT(10), 1<<3, RT(11), 1<<4, RT(12), 1<<5,
+ INT_MAX
+};
+
+static const int mode_check_freq_map_split1[] = {
+ 0, GOOD(2), 2, GOOD(3), 7, RT(1), 2, RT(2), 7, INT_MAX
+};
+
+static const int mode_check_freq_map_split2[] = {
+ 0, GOOD(1), 2, GOOD(2), 4, GOOD(3), 15, RT(1), 4, RT(2), 15, INT_MAX
+};
+
+void vp8_set_speed_features(VP8_COMP *cpi)
+{
+ SPEED_FEATURES *sf = &cpi->sf;
+ int Mode = cpi->compressor_speed;
+ int Speed = cpi->Speed;
+ int i;
+ VP8_COMMON *cm = &cpi->common;
+ int last_improved_quant = sf->improved_quant;
+ int ref_frames;
+
+ /* Initialise default mode frequency sampling variables */
+ for (i = 0; i < MAX_MODES; i ++)
+ {
+ cpi->mode_check_freq[i] = 0;
+ cpi->mode_test_hit_counts[i] = 0;
+ cpi->mode_chosen_counts[i] = 0;
+ }
+
+ cpi->mbs_tested_so_far = 0;
+
+ /* best quality defaults */
+ sf->RD = 1;
+ sf->search_method = NSTEP;
+ sf->improved_quant = 1;
+ sf->improved_dct = 1;
+ sf->auto_filter = 1;
+ sf->recode_loop = 1;
+ sf->quarter_pixel_search = 1;
+ sf->half_pixel_search = 1;
+ sf->iterative_sub_pixel = 1;
+ sf->optimize_coefficients = 1;
+ sf->use_fastquant_for_pick = 0;
+ sf->no_skip_block4x4_search = 1;
+
+ sf->first_step = 0;
+ sf->max_step_search_steps = MAX_MVSEARCH_STEPS;
+ sf->improved_mv_pred = 1;
+
+ /* default thresholds to 0 */
+ for (i = 0; i < MAX_MODES; i++)
+ sf->thresh_mult[i] = 0;
+
+ /* Count enabled references */
+ ref_frames = 1;
+ if (cpi->ref_frame_flags & VP8_LAST_FRAME)
+ ref_frames++;
+ if (cpi->ref_frame_flags & VP8_GOLD_FRAME)
+ ref_frames++;
+ if (cpi->ref_frame_flags & VP8_ALTR_FRAME)
+ ref_frames++;
+
+ /* Convert speed to continuous range, with clamping */
+ if (Mode == 0)
+ Speed = 0;
+ else if (Mode == 2)
+ Speed = RT(Speed);
+ else
+ {
+ if (Speed > 5)
+ Speed = 5;
+ Speed = GOOD(Speed);
+ }
+
+ sf->thresh_mult[THR_ZERO1] =
+ sf->thresh_mult[THR_NEAREST1] =
+ sf->thresh_mult[THR_NEAR1] =
+ sf->thresh_mult[THR_DC] = 0; /* always */
+
+ sf->thresh_mult[THR_ZERO2] =
+ sf->thresh_mult[THR_ZERO3] =
+ sf->thresh_mult[THR_NEAREST2] =
+ sf->thresh_mult[THR_NEAREST3] =
+ sf->thresh_mult[THR_NEAR2] =
+ sf->thresh_mult[THR_NEAR3] = speed_map(Speed, thresh_mult_map_znn);
+
+ sf->thresh_mult[THR_V_PRED] =
+ sf->thresh_mult[THR_H_PRED] = speed_map(Speed, thresh_mult_map_vhpred);
+ sf->thresh_mult[THR_B_PRED] = speed_map(Speed, thresh_mult_map_bpred);
+ sf->thresh_mult[THR_TM] = speed_map(Speed, thresh_mult_map_tm);
+ sf->thresh_mult[THR_NEW1] = speed_map(Speed, thresh_mult_map_new1);
+ sf->thresh_mult[THR_NEW2] =
+ sf->thresh_mult[THR_NEW3] = speed_map(Speed, thresh_mult_map_new2);
+ sf->thresh_mult[THR_SPLIT1] = speed_map(Speed, thresh_mult_map_split1);
+ sf->thresh_mult[THR_SPLIT2] =
+ sf->thresh_mult[THR_SPLIT3] = speed_map(Speed, thresh_mult_map_split2);
+
+ cpi->mode_check_freq[THR_ZERO1] =
+ cpi->mode_check_freq[THR_NEAREST1] =
+ cpi->mode_check_freq[THR_NEAR1] =
+ cpi->mode_check_freq[THR_TM] =
+ cpi->mode_check_freq[THR_DC] = 0; /* always */
+
+ cpi->mode_check_freq[THR_ZERO2] =
+ cpi->mode_check_freq[THR_ZERO3] =
+ cpi->mode_check_freq[THR_NEAREST2] =
+ cpi->mode_check_freq[THR_NEAREST3] = speed_map(Speed,
+ mode_check_freq_map_zn2);
+
+ cpi->mode_check_freq[THR_NEAR2] =
+ cpi->mode_check_freq[THR_NEAR3] = speed_map(Speed,
+ mode_check_freq_map_near2);
+
+ cpi->mode_check_freq[THR_V_PRED] =
+ cpi->mode_check_freq[THR_H_PRED] =
+ cpi->mode_check_freq[THR_B_PRED] = speed_map(Speed,
+ mode_check_freq_map_vhbpred);
+ cpi->mode_check_freq[THR_NEW1] = speed_map(Speed,
+ mode_check_freq_map_new1);
+ cpi->mode_check_freq[THR_NEW2] =
+ cpi->mode_check_freq[THR_NEW3] = speed_map(Speed,
+ mode_check_freq_map_new2);
+ cpi->mode_check_freq[THR_SPLIT1] = speed_map(Speed,
+ mode_check_freq_map_split1);
+ cpi->mode_check_freq[THR_SPLIT2] =
+ cpi->mode_check_freq[THR_SPLIT3] = speed_map(Speed,
+ mode_check_freq_map_split2);
+ Speed = cpi->Speed;
+ switch (Mode)
+ {
+#if !(CONFIG_REALTIME_ONLY)
+ case 0: /* best quality mode */
+ sf->first_step = 0;
+ sf->max_step_search_steps = MAX_MVSEARCH_STEPS;
+ break;
+ case 1:
+ case 3:
+ if (Speed > 0)
+ {
+ /* Disable coefficient optimization above speed 0 */
+ sf->optimize_coefficients = 0;
+ sf->use_fastquant_for_pick = 1;
+ sf->no_skip_block4x4_search = 0;
+
+ sf->first_step = 1;
+ }
+
+ if (Speed > 2)
+ {
+ sf->improved_quant = 0;
+ sf->improved_dct = 0;
+
+ /* Only do recode loop on key frames, golden frames and
+ * alt ref frames
+ */
+ sf->recode_loop = 2;
+
+ }
+
+ if (Speed > 3)
+ {
+ sf->auto_filter = 1;
+ sf->recode_loop = 0; /* recode loop off */
+ sf->RD = 0; /* Turn rd off */
+
+ }
+
+ if (Speed > 4)
+ {
+ sf->auto_filter = 0; /* Faster selection of loop filter */
+ }
+
+ break;
+#endif
+ case 2:
+ sf->optimize_coefficients = 0;
+ sf->recode_loop = 0;
+ sf->auto_filter = 1;
+ sf->iterative_sub_pixel = 1;
+ sf->search_method = NSTEP;
+
+ if (Speed > 0)
+ {
+ sf->improved_quant = 0;
+ sf->improved_dct = 0;
+
+ sf->use_fastquant_for_pick = 1;
+ sf->no_skip_block4x4_search = 0;
+ sf->first_step = 1;
+ }
+
+ if (Speed > 2)
+ sf->auto_filter = 0; /* Faster selection of loop filter */
+
+ if (Speed > 3)
+ {
+ sf->RD = 0;
+ sf->auto_filter = 1;
+ }
+
+ if (Speed > 4)
+ {
+ sf->auto_filter = 0; /* Faster selection of loop filter */
+ sf->search_method = HEX;
+ sf->iterative_sub_pixel = 0;
+ }
+
+ if (Speed > 6)
+ {
+ unsigned int sum = 0;
+ unsigned int total_mbs = cm->MBs;
+ int i, thresh;
+ unsigned int total_skip;
+
+ int min = 2000;
+
+ if (cpi->oxcf.encode_breakout > 2000)
+ min = cpi->oxcf.encode_breakout;
+
+ min >>= 7;
+
+ for (i = 0; i < min; i++)
+ {
+ sum += cpi->error_bins[i];
+ }
+
+ total_skip = sum;
+ sum = 0;
+
+ /* i starts from 2 to make sure thresh started from 2048 */
+ for (; i < 1024; i++)
+ {
+ sum += cpi->error_bins[i];
+
+ if (10 * sum >= (unsigned int)(cpi->Speed - 6)*(total_mbs - total_skip))
+ break;
+ }
+
+ i--;
+ thresh = (i << 7);
+
+ if (thresh < 2000)
+ thresh = 2000;
+
+ if (ref_frames > 1)
+ {
+ sf->thresh_mult[THR_NEW1 ] = thresh;
+ sf->thresh_mult[THR_NEAREST1 ] = thresh >> 1;
+ sf->thresh_mult[THR_NEAR1 ] = thresh >> 1;
+ }
+
+ if (ref_frames > 2)
+ {
+ sf->thresh_mult[THR_NEW2] = thresh << 1;
+ sf->thresh_mult[THR_NEAREST2 ] = thresh;
+ sf->thresh_mult[THR_NEAR2 ] = thresh;
+ }
+
+ if (ref_frames > 3)
+ {
+ sf->thresh_mult[THR_NEW3] = thresh << 1;
+ sf->thresh_mult[THR_NEAREST3 ] = thresh;
+ sf->thresh_mult[THR_NEAR3 ] = thresh;
+ }
+
+ sf->improved_mv_pred = 0;
+ }
+
+ if (Speed > 8)
+ sf->quarter_pixel_search = 0;
+
+ if(cm->version == 0)
+ {
+ cm->filter_type = NORMAL_LOOPFILTER;
+
+ if (Speed >= 14)
+ cm->filter_type = SIMPLE_LOOPFILTER;
+ }
+ else
+ {
+ cm->filter_type = SIMPLE_LOOPFILTER;
+ }
+
+ /* This has a big hit on quality. Last resort */
+ if (Speed >= 15)
+ sf->half_pixel_search = 0;
+
+ vpx_memset(cpi->error_bins, 0, sizeof(cpi->error_bins));
+
+ }; /* switch */
+
+ /* Slow quant, dct and trellis not worthwhile for first pass
+ * so make sure they are always turned off.
+ */
+ if ( cpi->pass == 1 )
+ {
+ sf->improved_quant = 0;
+ sf->optimize_coefficients = 0;
+ sf->improved_dct = 0;
+ }
+
+ if (cpi->sf.search_method == NSTEP)
+ {
+ vp8_init3smotion_compensation(&cpi->mb, cm->yv12_fb[cm->lst_fb_idx].y_stride);
+ }
+ else if (cpi->sf.search_method == DIAMOND)
+ {
+ vp8_init_dsmotion_compensation(&cpi->mb, cm->yv12_fb[cm->lst_fb_idx].y_stride);
+ }
+
+ if (cpi->sf.improved_dct)
+ {
+ cpi->mb.short_fdct8x4 = vp8_short_fdct8x4;
+ cpi->mb.short_fdct4x4 = vp8_short_fdct4x4;
+ }
+ else
+ {
+ /* No fast FDCT defined for any platform at this time. */
+ cpi->mb.short_fdct8x4 = vp8_short_fdct8x4;
+ cpi->mb.short_fdct4x4 = vp8_short_fdct4x4;
+ }
+
+ cpi->mb.short_walsh4x4 = vp8_short_walsh4x4;
+
+ if (cpi->sf.improved_quant)
+ {
+ cpi->mb.quantize_b = vp8_regular_quantize_b;
+ cpi->mb.quantize_b_pair = vp8_regular_quantize_b_pair;
+ }
+ else
+ {
+ cpi->mb.quantize_b = vp8_fast_quantize_b;
+ cpi->mb.quantize_b_pair = vp8_fast_quantize_b_pair;
+ }
+ if (cpi->sf.improved_quant != last_improved_quant)
+ vp8cx_init_quantizer(cpi);
+
+ if (cpi->sf.iterative_sub_pixel == 1)
+ {
+ cpi->find_fractional_mv_step = vp8_find_best_sub_pixel_step_iteratively;
+ }
+ else if (cpi->sf.quarter_pixel_search)
+ {
+ cpi->find_fractional_mv_step = vp8_find_best_sub_pixel_step;
+ }
+ else if (cpi->sf.half_pixel_search)
+ {
+ cpi->find_fractional_mv_step = vp8_find_best_half_pixel_step;
+ }
+ else
+ {
+ cpi->find_fractional_mv_step = vp8_skip_fractional_mv_step;
+ }
+
+ if (cpi->sf.optimize_coefficients == 1 && cpi->pass!=1)
+ cpi->mb.optimize = 1;
+ else
+ cpi->mb.optimize = 0;
+
+ if (cpi->common.full_pixel)
+ cpi->find_fractional_mv_step = vp8_skip_fractional_mv_step;
+
+#ifdef SPEEDSTATS
+ frames_at_speed[cpi->Speed]++;
+#endif
+}
+#undef GOOD
+#undef RT
+
+static void alloc_raw_frame_buffers(VP8_COMP *cpi)
+{
+#if VP8_TEMPORAL_ALT_REF
+ int width = (cpi->oxcf.Width + 15) & ~15;
+ int height = (cpi->oxcf.Height + 15) & ~15;
+#endif
+
+ cpi->lookahead = vp8_lookahead_init(cpi->oxcf.Width, cpi->oxcf.Height,
+ cpi->oxcf.lag_in_frames);
+ if(!cpi->lookahead)
+ vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate lag buffers");
+
+#if VP8_TEMPORAL_ALT_REF
+
+ if (vp8_yv12_alloc_frame_buffer(&cpi->alt_ref_buffer,
+ width, height, VP8BORDERINPIXELS))
+ vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate altref buffer");
+
+#endif
+}
+
+
+static void dealloc_raw_frame_buffers(VP8_COMP *cpi)
+{
+#if VP8_TEMPORAL_ALT_REF
+ vp8_yv12_de_alloc_frame_buffer(&cpi->alt_ref_buffer);
+#endif
+ vp8_lookahead_destroy(cpi->lookahead);
+}
+
+
+static int vp8_alloc_partition_data(VP8_COMP *cpi)
+{
+ vpx_free(cpi->mb.pip);
+
+ cpi->mb.pip = vpx_calloc((cpi->common.mb_cols + 1) *
+ (cpi->common.mb_rows + 1),
+ sizeof(PARTITION_INFO));
+ if(!cpi->mb.pip)
+ return 1;
+
+ cpi->mb.pi = cpi->mb.pip + cpi->common.mode_info_stride + 1;
+
+ return 0;
+}
+
+void vp8_alloc_compressor_data(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = & cpi->common;
+
+ int width = cm->Width;
+ int height = cm->Height;
+
+ if (vp8_alloc_frame_buffers(cm, width, height))
+ vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate frame buffers");
+
+ if (vp8_alloc_partition_data(cpi))
+ vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate partition data");
+
+
+ if ((width & 0xf) != 0)
+ width += 16 - (width & 0xf);
+
+ if ((height & 0xf) != 0)
+ height += 16 - (height & 0xf);
+
+
+ if (vp8_yv12_alloc_frame_buffer(&cpi->pick_lf_lvl_frame,
+ width, height, VP8BORDERINPIXELS))
+ vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate last frame buffer");
+
+ if (vp8_yv12_alloc_frame_buffer(&cpi->scaled_source,
+ width, height, VP8BORDERINPIXELS))
+ vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate scaled source buffer");
+
+ vpx_free(cpi->tok);
+
+ {
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ unsigned int tokens = 8 * 24 * 16; /* one MB for each thread */
+#else
+ unsigned int tokens = cm->mb_rows * cm->mb_cols * 24 * 16;
+#endif
+ CHECK_MEM_ERROR(cpi->tok, vpx_calloc(tokens, sizeof(*cpi->tok)));
+ }
+
+ /* Data used for real time vc mode to see if gf needs refreshing */
+ cpi->inter_zz_count = 0;
+ cpi->zeromv_count = 0;
+ cpi->gf_bad_count = 0;
+ cpi->gf_update_recommended = 0;
+
+
+ /* Structures used to monitor GF usage */
+ vpx_free(cpi->gf_active_flags);
+ CHECK_MEM_ERROR(cpi->gf_active_flags,
+ vpx_calloc(sizeof(*cpi->gf_active_flags),
+ cm->mb_rows * cm->mb_cols));
+ cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
+
+ vpx_free(cpi->mb_activity_map);
+ CHECK_MEM_ERROR(cpi->mb_activity_map,
+ vpx_calloc(sizeof(*cpi->mb_activity_map),
+ cm->mb_rows * cm->mb_cols));
+
+ /* allocate memory for storing last frame's MVs for MV prediction. */
+ vpx_free(cpi->lfmv);
+ CHECK_MEM_ERROR(cpi->lfmv, vpx_calloc((cm->mb_rows+2) * (cm->mb_cols+2),
+ sizeof(*cpi->lfmv)));
+ vpx_free(cpi->lf_ref_frame_sign_bias);
+ CHECK_MEM_ERROR(cpi->lf_ref_frame_sign_bias,
+ vpx_calloc((cm->mb_rows+2) * (cm->mb_cols+2),
+ sizeof(*cpi->lf_ref_frame_sign_bias)));
+ vpx_free(cpi->lf_ref_frame);
+ CHECK_MEM_ERROR(cpi->lf_ref_frame,
+ vpx_calloc((cm->mb_rows+2) * (cm->mb_cols+2),
+ sizeof(*cpi->lf_ref_frame)));
+
+ /* Create the encoder segmentation map and set all entries to 0 */
+ vpx_free(cpi->segmentation_map);
+ CHECK_MEM_ERROR(cpi->segmentation_map,
+ vpx_calloc(cm->mb_rows * cm->mb_cols,
+ sizeof(*cpi->segmentation_map)));
+ cpi->cyclic_refresh_mode_index = 0;
+ vpx_free(cpi->active_map);
+ CHECK_MEM_ERROR(cpi->active_map,
+ vpx_calloc(cm->mb_rows * cm->mb_cols,
+ sizeof(*cpi->active_map)));
+ vpx_memset(cpi->active_map , 1, (cm->mb_rows * cm->mb_cols));
+
+#if CONFIG_MULTITHREAD
+ if (width < 640)
+ cpi->mt_sync_range = 1;
+ else if (width <= 1280)
+ cpi->mt_sync_range = 4;
+ else if (width <= 2560)
+ cpi->mt_sync_range = 8;
+ else
+ cpi->mt_sync_range = 16;
+
+ if (cpi->oxcf.multi_threaded > 1)
+ {
+ vpx_free(cpi->mt_current_mb_col);
+ CHECK_MEM_ERROR(cpi->mt_current_mb_col,
+ vpx_malloc(sizeof(*cpi->mt_current_mb_col) * cm->mb_rows));
+ }
+
+#endif
+
+ vpx_free(cpi->tplist);
+ CHECK_MEM_ERROR(cpi->tplist, vpx_malloc(sizeof(TOKENLIST) * cm->mb_rows));
+}
+
+
+/* Quant MOD */
+static const int q_trans[] =
+{
+ 0, 1, 2, 3, 4, 5, 7, 8,
+ 9, 10, 12, 13, 15, 17, 18, 19,
+ 20, 21, 23, 24, 25, 26, 27, 28,
+ 29, 30, 31, 33, 35, 37, 39, 41,
+ 43, 45, 47, 49, 51, 53, 55, 57,
+ 59, 61, 64, 67, 70, 73, 76, 79,
+ 82, 85, 88, 91, 94, 97, 100, 103,
+ 106, 109, 112, 115, 118, 121, 124, 127,
+};
+
+int vp8_reverse_trans(int x)
+{
+ int i;
+
+ for (i = 0; i < 64; i++)
+ if (q_trans[i] >= x)
+ return i;
+
+ return 63;
+}
+void vp8_new_frame_rate(VP8_COMP *cpi, double framerate)
+{
+ if(framerate < .1)
+ framerate = 30;
+
+ cpi->frame_rate = framerate;
+ cpi->output_frame_rate = framerate;
+ cpi->per_frame_bandwidth = (int)(cpi->oxcf.target_bandwidth /
+ cpi->output_frame_rate);
+ cpi->av_per_frame_bandwidth = cpi->per_frame_bandwidth;
+ cpi->min_frame_bandwidth = (int)(cpi->av_per_frame_bandwidth *
+ cpi->oxcf.two_pass_vbrmin_section / 100);
+
+ /* Set Maximum gf/arf interval */
+ cpi->max_gf_interval = ((int)(cpi->output_frame_rate / 2.0) + 2);
+
+ if(cpi->max_gf_interval < 12)
+ cpi->max_gf_interval = 12;
+
+ /* Extended interval for genuinely static scenes */
+ cpi->twopass.static_scene_max_gf_interval = cpi->key_frame_frequency >> 1;
+
+ /* Special conditions when altr ref frame enabled in lagged compress mode */
+ if (cpi->oxcf.play_alternate && cpi->oxcf.lag_in_frames)
+ {
+ if (cpi->max_gf_interval > cpi->oxcf.lag_in_frames - 1)
+ cpi->max_gf_interval = cpi->oxcf.lag_in_frames - 1;
+
+ if (cpi->twopass.static_scene_max_gf_interval > cpi->oxcf.lag_in_frames - 1)
+ cpi->twopass.static_scene_max_gf_interval = cpi->oxcf.lag_in_frames - 1;
+ }
+
+ if ( cpi->max_gf_interval > cpi->twopass.static_scene_max_gf_interval )
+ cpi->max_gf_interval = cpi->twopass.static_scene_max_gf_interval;
+}
+
+
+static int
+rescale(int val, int num, int denom)
+{
+ int64_t llnum = num;
+ int64_t llden = denom;
+ int64_t llval = val;
+
+ return (int)(llval * llnum / llden);
+}
+
+
+static void init_config(VP8_COMP *cpi, VP8_CONFIG *oxcf)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ cpi->oxcf = *oxcf;
+
+ cpi->auto_gold = 1;
+ cpi->auto_adjust_gold_quantizer = 1;
+
+ cm->version = oxcf->Version;
+ vp8_setup_version(cm);
+
+ /* frame rate is not available on the first frame, as it's derived from
+ * the observed timestamps. The actual value used here doesn't matter
+ * too much, as it will adapt quickly. If the reciprocal of the timebase
+ * seems like a reasonable framerate, then use that as a guess, otherwise
+ * use 30.
+ */
+ cpi->frame_rate = (double)(oxcf->timebase.den) /
+ (double)(oxcf->timebase.num);
+
+ if (cpi->frame_rate > 180)
+ cpi->frame_rate = 30;
+
+ cpi->ref_frame_rate = cpi->frame_rate;
+
+ /* change includes all joint functionality */
+ vp8_change_config(cpi, oxcf);
+
+ /* Initialize active best and worst q and average q values. */
+ cpi->active_worst_quality = cpi->oxcf.worst_allowed_q;
+ cpi->active_best_quality = cpi->oxcf.best_allowed_q;
+ cpi->avg_frame_qindex = cpi->oxcf.worst_allowed_q;
+
+ /* Initialise the starting buffer levels */
+ cpi->buffer_level = cpi->oxcf.starting_buffer_level;
+ cpi->bits_off_target = cpi->oxcf.starting_buffer_level;
+
+ cpi->rolling_target_bits = cpi->av_per_frame_bandwidth;
+ cpi->rolling_actual_bits = cpi->av_per_frame_bandwidth;
+ cpi->long_rolling_target_bits = cpi->av_per_frame_bandwidth;
+ cpi->long_rolling_actual_bits = cpi->av_per_frame_bandwidth;
+
+ cpi->total_actual_bits = 0;
+ cpi->total_target_vs_actual = 0;
+
+ /* Temporal scalabilty */
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ unsigned int i;
+ double prev_layer_frame_rate=0;
+
+ for (i=0; i<cpi->oxcf.number_of_layers; i++)
+ {
+ LAYER_CONTEXT *lc = &cpi->layer_context[i];
+
+ /* Layer configuration */
+ lc->frame_rate =
+ cpi->output_frame_rate / cpi->oxcf.rate_decimator[i];
+ lc->target_bandwidth = cpi->oxcf.target_bitrate[i] * 1000;
+
+ lc->starting_buffer_level_in_ms = oxcf->starting_buffer_level;
+ lc->optimal_buffer_level_in_ms = oxcf->optimal_buffer_level;
+ lc->maximum_buffer_size_in_ms = oxcf->maximum_buffer_size;
+
+ lc->starting_buffer_level =
+ rescale((int)(oxcf->starting_buffer_level),
+ lc->target_bandwidth, 1000);
+
+ if (oxcf->optimal_buffer_level == 0)
+ lc->optimal_buffer_level = lc->target_bandwidth / 8;
+ else
+ lc->optimal_buffer_level =
+ rescale((int)(oxcf->optimal_buffer_level),
+ lc->target_bandwidth, 1000);
+
+ if (oxcf->maximum_buffer_size == 0)
+ lc->maximum_buffer_size = lc->target_bandwidth / 8;
+ else
+ lc->maximum_buffer_size =
+ rescale((int)oxcf->maximum_buffer_size,
+ lc->target_bandwidth, 1000);
+
+ /* Work out the average size of a frame within this layer */
+ if (i > 0)
+ lc->avg_frame_size_for_layer =
+ (int)((cpi->oxcf.target_bitrate[i] -
+ cpi->oxcf.target_bitrate[i-1]) * 1000 /
+ (lc->frame_rate - prev_layer_frame_rate));
+
+ lc->active_worst_quality = cpi->oxcf.worst_allowed_q;
+ lc->active_best_quality = cpi->oxcf.best_allowed_q;
+ lc->avg_frame_qindex = cpi->oxcf.worst_allowed_q;
+
+ lc->buffer_level = lc->starting_buffer_level;
+ lc->bits_off_target = lc->starting_buffer_level;
+
+ lc->total_actual_bits = 0;
+ lc->ni_av_qi = 0;
+ lc->ni_tot_qi = 0;
+ lc->ni_frames = 0;
+ lc->rate_correction_factor = 1.0;
+ lc->key_frame_rate_correction_factor = 1.0;
+ lc->gf_rate_correction_factor = 1.0;
+ lc->inter_frame_target = 0;
+
+ prev_layer_frame_rate = lc->frame_rate;
+ }
+ }
+
+#if VP8_TEMPORAL_ALT_REF
+ {
+ int i;
+
+ cpi->fixed_divide[0] = 0;
+
+ for (i = 1; i < 512; i++)
+ cpi->fixed_divide[i] = 0x80000 / i;
+ }
+#endif
+}
+
+static void update_layer_contexts (VP8_COMP *cpi)
+{
+ VP8_CONFIG *oxcf = &cpi->oxcf;
+
+ /* Update snapshots of the layer contexts to reflect new parameters */
+ if (oxcf->number_of_layers > 1)
+ {
+ unsigned int i;
+ double prev_layer_frame_rate=0;
+
+ for (i=0; i<oxcf->number_of_layers; i++)
+ {
+ LAYER_CONTEXT *lc = &cpi->layer_context[i];
+
+ lc->frame_rate =
+ cpi->ref_frame_rate / oxcf->rate_decimator[i];
+ lc->target_bandwidth = oxcf->target_bitrate[i] * 1000;
+
+ lc->starting_buffer_level = rescale(
+ (int)oxcf->starting_buffer_level_in_ms,
+ lc->target_bandwidth, 1000);
+
+ if (oxcf->optimal_buffer_level == 0)
+ lc->optimal_buffer_level = lc->target_bandwidth / 8;
+ else
+ lc->optimal_buffer_level = rescale(
+ (int)oxcf->optimal_buffer_level_in_ms,
+ lc->target_bandwidth, 1000);
+
+ if (oxcf->maximum_buffer_size == 0)
+ lc->maximum_buffer_size = lc->target_bandwidth / 8;
+ else
+ lc->maximum_buffer_size = rescale(
+ (int)oxcf->maximum_buffer_size_in_ms,
+ lc->target_bandwidth, 1000);
+
+ /* Work out the average size of a frame within this layer */
+ if (i > 0)
+ lc->avg_frame_size_for_layer =
+ (int)((oxcf->target_bitrate[i] -
+ oxcf->target_bitrate[i-1]) * 1000 /
+ (lc->frame_rate - prev_layer_frame_rate));
+
+ prev_layer_frame_rate = lc->frame_rate;
+ }
+ }
+}
+
+void vp8_change_config(VP8_COMP *cpi, VP8_CONFIG *oxcf)
+{
+ VP8_COMMON *cm = &cpi->common;
+ int last_w, last_h;
+
+ if (!cpi)
+ return;
+
+ if (!oxcf)
+ return;
+
+#if CONFIG_MULTITHREAD
+ /* wait for the last picture loopfilter thread done */
+ if (cpi->b_lpf_running)
+ {
+ sem_wait(&cpi->h_event_end_lpf);
+ cpi->b_lpf_running = 0;
+ }
+#endif
+
+ if (cm->version != oxcf->Version)
+ {
+ cm->version = oxcf->Version;
+ vp8_setup_version(cm);
+ }
+
+ last_w = cpi->oxcf.Width;
+ last_h = cpi->oxcf.Height;
+
+ cpi->oxcf = *oxcf;
+
+ switch (cpi->oxcf.Mode)
+ {
+
+ case MODE_REALTIME:
+ cpi->pass = 0;
+ cpi->compressor_speed = 2;
+
+ if (cpi->oxcf.cpu_used < -16)
+ {
+ cpi->oxcf.cpu_used = -16;
+ }
+
+ if (cpi->oxcf.cpu_used > 16)
+ cpi->oxcf.cpu_used = 16;
+
+ break;
+
+ case MODE_GOODQUALITY:
+ cpi->pass = 0;
+ cpi->compressor_speed = 1;
+
+ if (cpi->oxcf.cpu_used < -5)
+ {
+ cpi->oxcf.cpu_used = -5;
+ }
+
+ if (cpi->oxcf.cpu_used > 5)
+ cpi->oxcf.cpu_used = 5;
+
+ break;
+
+ case MODE_BESTQUALITY:
+ cpi->pass = 0;
+ cpi->compressor_speed = 0;
+ break;
+
+ case MODE_FIRSTPASS:
+ cpi->pass = 1;
+ cpi->compressor_speed = 1;
+ break;
+ case MODE_SECONDPASS:
+ cpi->pass = 2;
+ cpi->compressor_speed = 1;
+
+ if (cpi->oxcf.cpu_used < -5)
+ {
+ cpi->oxcf.cpu_used = -5;
+ }
+
+ if (cpi->oxcf.cpu_used > 5)
+ cpi->oxcf.cpu_used = 5;
+
+ break;
+ case MODE_SECONDPASS_BEST:
+ cpi->pass = 2;
+ cpi->compressor_speed = 0;
+ break;
+ }
+
+ if (cpi->pass == 0)
+ cpi->auto_worst_q = 1;
+
+ cpi->oxcf.worst_allowed_q = q_trans[oxcf->worst_allowed_q];
+ cpi->oxcf.best_allowed_q = q_trans[oxcf->best_allowed_q];
+ cpi->oxcf.cq_level = q_trans[cpi->oxcf.cq_level];
+
+ if (oxcf->fixed_q >= 0)
+ {
+ if (oxcf->worst_allowed_q < 0)
+ cpi->oxcf.fixed_q = q_trans[0];
+ else
+ cpi->oxcf.fixed_q = q_trans[oxcf->worst_allowed_q];
+
+ if (oxcf->alt_q < 0)
+ cpi->oxcf.alt_q = q_trans[0];
+ else
+ cpi->oxcf.alt_q = q_trans[oxcf->alt_q];
+
+ if (oxcf->key_q < 0)
+ cpi->oxcf.key_q = q_trans[0];
+ else
+ cpi->oxcf.key_q = q_trans[oxcf->key_q];
+
+ if (oxcf->gold_q < 0)
+ cpi->oxcf.gold_q = q_trans[0];
+ else
+ cpi->oxcf.gold_q = q_trans[oxcf->gold_q];
+
+ }
+
+ cpi->baseline_gf_interval =
+ cpi->oxcf.alt_freq ? cpi->oxcf.alt_freq : DEFAULT_GF_INTERVAL;
+
+ cpi->ref_frame_flags = VP8_ALTR_FRAME | VP8_GOLD_FRAME | VP8_LAST_FRAME;
+
+ cm->refresh_golden_frame = 0;
+ cm->refresh_last_frame = 1;
+ cm->refresh_entropy_probs = 1;
+
+#if (CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ cpi->oxcf.token_partitions = 3;
+#endif
+
+ if (cpi->oxcf.token_partitions >= 0 && cpi->oxcf.token_partitions <= 3)
+ cm->multi_token_partition =
+ (TOKEN_PARTITION) cpi->oxcf.token_partitions;
+
+ setup_features(cpi);
+
+ {
+ int i;
+
+ for (i = 0; i < MAX_MB_SEGMENTS; i++)
+ cpi->segment_encode_breakout[i] = cpi->oxcf.encode_breakout;
+ }
+
+ /* At the moment the first order values may not be > MAXQ */
+ if (cpi->oxcf.fixed_q > MAXQ)
+ cpi->oxcf.fixed_q = MAXQ;
+
+ /* local file playback mode == really big buffer */
+ if (cpi->oxcf.end_usage == USAGE_LOCAL_FILE_PLAYBACK)
+ {
+ cpi->oxcf.starting_buffer_level = 60000;
+ cpi->oxcf.optimal_buffer_level = 60000;
+ cpi->oxcf.maximum_buffer_size = 240000;
+ cpi->oxcf.starting_buffer_level_in_ms = 60000;
+ cpi->oxcf.optimal_buffer_level_in_ms = 60000;
+ cpi->oxcf.maximum_buffer_size_in_ms = 240000;
+ }
+
+ /* Convert target bandwidth from Kbit/s to Bit/s */
+ cpi->oxcf.target_bandwidth *= 1000;
+
+ cpi->oxcf.starting_buffer_level =
+ rescale((int)cpi->oxcf.starting_buffer_level,
+ cpi->oxcf.target_bandwidth, 1000);
+
+ /* Set or reset optimal and maximum buffer levels. */
+ if (cpi->oxcf.optimal_buffer_level == 0)
+ cpi->oxcf.optimal_buffer_level = cpi->oxcf.target_bandwidth / 8;
+ else
+ cpi->oxcf.optimal_buffer_level =
+ rescale((int)cpi->oxcf.optimal_buffer_level,
+ cpi->oxcf.target_bandwidth, 1000);
+
+ if (cpi->oxcf.maximum_buffer_size == 0)
+ cpi->oxcf.maximum_buffer_size = cpi->oxcf.target_bandwidth / 8;
+ else
+ cpi->oxcf.maximum_buffer_size =
+ rescale((int)cpi->oxcf.maximum_buffer_size,
+ cpi->oxcf.target_bandwidth, 1000);
+
+ /* Set up frame rate and related parameters rate control values. */
+ vp8_new_frame_rate(cpi, cpi->frame_rate);
+
+ /* Set absolute upper and lower quality limits */
+ cpi->worst_quality = cpi->oxcf.worst_allowed_q;
+ cpi->best_quality = cpi->oxcf.best_allowed_q;
+
+ /* active values should only be modified if out of new range */
+ if (cpi->active_worst_quality > cpi->oxcf.worst_allowed_q)
+ {
+ cpi->active_worst_quality = cpi->oxcf.worst_allowed_q;
+ }
+ /* less likely */
+ else if (cpi->active_worst_quality < cpi->oxcf.best_allowed_q)
+ {
+ cpi->active_worst_quality = cpi->oxcf.best_allowed_q;
+ }
+ if (cpi->active_best_quality < cpi->oxcf.best_allowed_q)
+ {
+ cpi->active_best_quality = cpi->oxcf.best_allowed_q;
+ }
+ /* less likely */
+ else if (cpi->active_best_quality > cpi->oxcf.worst_allowed_q)
+ {
+ cpi->active_best_quality = cpi->oxcf.worst_allowed_q;
+ }
+
+ cpi->buffered_mode = cpi->oxcf.optimal_buffer_level > 0;
+
+ cpi->cq_target_quality = cpi->oxcf.cq_level;
+
+ /* Only allow dropped frames in buffered mode */
+ cpi->drop_frames_allowed = cpi->oxcf.allow_df && cpi->buffered_mode;
+
+ cpi->target_bandwidth = cpi->oxcf.target_bandwidth;
+
+
+ cm->Width = cpi->oxcf.Width;
+ cm->Height = cpi->oxcf.Height;
+
+ /* TODO(jkoleszar): if an internal spatial resampling is active,
+ * and we downsize the input image, maybe we should clear the
+ * internal scale immediately rather than waiting for it to
+ * correct.
+ */
+
+ /* VP8 sharpness level mapping 0-7 (vs 0-10 in general VPx dialogs) */
+ if (cpi->oxcf.Sharpness > 7)
+ cpi->oxcf.Sharpness = 7;
+
+ cm->sharpness_level = cpi->oxcf.Sharpness;
+
+ if (cm->horiz_scale != NORMAL || cm->vert_scale != NORMAL)
+ {
+ int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
+ int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
+
+ Scale2Ratio(cm->horiz_scale, &hr, &hs);
+ Scale2Ratio(cm->vert_scale, &vr, &vs);
+
+ /* always go to the next whole number */
+ cm->Width = (hs - 1 + cpi->oxcf.Width * hr) / hs;
+ cm->Height = (vs - 1 + cpi->oxcf.Height * vr) / vs;
+ }
+
+ if (last_w != cpi->oxcf.Width || last_h != cpi->oxcf.Height)
+ cpi->force_next_frame_intra = 1;
+
+ if (((cm->Width + 15) & 0xfffffff0) !=
+ cm->yv12_fb[cm->lst_fb_idx].y_width ||
+ ((cm->Height + 15) & 0xfffffff0) !=
+ cm->yv12_fb[cm->lst_fb_idx].y_height ||
+ cm->yv12_fb[cm->lst_fb_idx].y_width == 0)
+ {
+ dealloc_raw_frame_buffers(cpi);
+ alloc_raw_frame_buffers(cpi);
+ vp8_alloc_compressor_data(cpi);
+ }
+
+ if (cpi->oxcf.fixed_q >= 0)
+ {
+ cpi->last_q[0] = cpi->oxcf.fixed_q;
+ cpi->last_q[1] = cpi->oxcf.fixed_q;
+ }
+
+ cpi->Speed = cpi->oxcf.cpu_used;
+
+ /* force to allowlag to 0 if lag_in_frames is 0; */
+ if (cpi->oxcf.lag_in_frames == 0)
+ {
+ cpi->oxcf.allow_lag = 0;
+ }
+ /* Limit on lag buffers as these are not currently dynamically allocated */
+ else if (cpi->oxcf.lag_in_frames > MAX_LAG_BUFFERS)
+ cpi->oxcf.lag_in_frames = MAX_LAG_BUFFERS;
+
+ /* YX Temp */
+ cpi->alt_ref_source = NULL;
+ cpi->is_src_frame_alt_ref = 0;
+
+#if CONFIG_TEMPORAL_DENOISING
+ if (cpi->oxcf.noise_sensitivity)
+ {
+ if (!cpi->denoiser.yv12_mc_running_avg.buffer_alloc)
+ {
+ int width = (cpi->oxcf.Width + 15) & ~15;
+ int height = (cpi->oxcf.Height + 15) & ~15;
+ vp8_denoiser_allocate(&cpi->denoiser, width, height);
+ }
+ }
+#endif
+
+#if 0
+ /* Experimental RD Code */
+ cpi->frame_distortion = 0;
+ cpi->last_frame_distortion = 0;
+#endif
+
+}
+
+#define M_LOG2_E 0.693147180559945309417
+#define log2f(x) (log (x) / (float) M_LOG2_E)
+static void cal_mvsadcosts(int *mvsadcost[2])
+{
+ int i = 1;
+
+ mvsadcost [0] [0] = 300;
+ mvsadcost [1] [0] = 300;
+
+ do
+ {
+ double z = 256 * (2 * (log2f(8 * i) + .6));
+ mvsadcost [0][i] = (int) z;
+ mvsadcost [1][i] = (int) z;
+ mvsadcost [0][-i] = (int) z;
+ mvsadcost [1][-i] = (int) z;
+ }
+ while (++i <= mvfp_max);
+}
+
+struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf)
+{
+ int i;
+
+ VP8_COMP *cpi;
+ VP8_COMMON *cm;
+
+ cpi = vpx_memalign(32, sizeof(VP8_COMP));
+ /* Check that the CPI instance is valid */
+ if (!cpi)
+ return 0;
+
+ cm = &cpi->common;
+
+ vpx_memset(cpi, 0, sizeof(VP8_COMP));
+
+ if (setjmp(cm->error.jmp))
+ {
+ cpi->common.error.setjmp = 0;
+ vp8_remove_compressor(&cpi);
+ return 0;
+ }
+
+ cpi->common.error.setjmp = 1;
+
+ CHECK_MEM_ERROR(cpi->mb.ss, vpx_calloc(sizeof(search_site), (MAX_MVSEARCH_STEPS * 8) + 1));
+
+ vp8_create_common(&cpi->common);
+
+ init_config(cpi, oxcf);
+
+ memcpy(cpi->base_skip_false_prob, vp8cx_base_skip_false_prob, sizeof(vp8cx_base_skip_false_prob));
+ cpi->common.current_video_frame = 0;
+ cpi->kf_overspend_bits = 0;
+ cpi->kf_bitrate_adjustment = 0;
+ cpi->frames_till_gf_update_due = 0;
+ cpi->gf_overspend_bits = 0;
+ cpi->non_gf_bitrate_adjustment = 0;
+ cpi->prob_last_coded = 128;
+ cpi->prob_gf_coded = 128;
+ cpi->prob_intra_coded = 63;
+
+ /* Prime the recent reference frame usage counters.
+ * Hereafter they will be maintained as a sort of moving average
+ */
+ cpi->recent_ref_frame_usage[INTRA_FRAME] = 1;
+ cpi->recent_ref_frame_usage[LAST_FRAME] = 1;
+ cpi->recent_ref_frame_usage[GOLDEN_FRAME] = 1;
+ cpi->recent_ref_frame_usage[ALTREF_FRAME] = 1;
+
+ /* Set reference frame sign bias for ALTREF frame to 1 (for now) */
+ cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 1;
+
+ cpi->twopass.gf_decay_rate = 0;
+ cpi->baseline_gf_interval = DEFAULT_GF_INTERVAL;
+
+ cpi->gold_is_last = 0 ;
+ cpi->alt_is_last = 0 ;
+ cpi->gold_is_alt = 0 ;
+
+ cpi->active_map_enabled = 0;
+
+#if 0
+ /* Experimental code for lagged and one pass */
+ /* Initialise one_pass GF frames stats */
+ /* Update stats used for GF selection */
+ if (cpi->pass == 0)
+ {
+ cpi->one_pass_frame_index = 0;
+
+ for (i = 0; i < MAX_LAG_BUFFERS; i++)
+ {
+ cpi->one_pass_frame_stats[i].frames_so_far = 0;
+ cpi->one_pass_frame_stats[i].frame_intra_error = 0.0;
+ cpi->one_pass_frame_stats[i].frame_coded_error = 0.0;
+ cpi->one_pass_frame_stats[i].frame_pcnt_inter = 0.0;
+ cpi->one_pass_frame_stats[i].frame_pcnt_motion = 0.0;
+ cpi->one_pass_frame_stats[i].frame_mvr = 0.0;
+ cpi->one_pass_frame_stats[i].frame_mvr_abs = 0.0;
+ cpi->one_pass_frame_stats[i].frame_mvc = 0.0;
+ cpi->one_pass_frame_stats[i].frame_mvc_abs = 0.0;
+ }
+ }
+#endif
+
+ /* Should we use the cyclic refresh method.
+ * Currently this is tied to error resilliant mode
+ */
+ cpi->cyclic_refresh_mode_enabled = cpi->oxcf.error_resilient_mode;
+ cpi->cyclic_refresh_mode_max_mbs_perframe = (cpi->common.mb_rows * cpi->common.mb_cols) / 5;
+ cpi->cyclic_refresh_mode_index = 0;
+ cpi->cyclic_refresh_q = 32;
+
+ if (cpi->cyclic_refresh_mode_enabled)
+ {
+ CHECK_MEM_ERROR(cpi->cyclic_refresh_map, vpx_calloc((cpi->common.mb_rows * cpi->common.mb_cols), 1));
+ }
+ else
+ cpi->cyclic_refresh_map = (signed char *) NULL;
+
+#ifdef ENTROPY_STATS
+ init_context_counters();
+#endif
+
+ /*Initialize the feed-forward activity masking.*/
+ cpi->activity_avg = 90<<12;
+
+ /* Give a sensible default for the first frame. */
+ cpi->frames_since_key = 8;
+ cpi->key_frame_frequency = cpi->oxcf.key_freq;
+ cpi->this_key_frame_forced = 0;
+ cpi->next_key_frame_forced = 0;
+
+ cpi->source_alt_ref_pending = 0;
+ cpi->source_alt_ref_active = 0;
+ cpi->common.refresh_alt_ref_frame = 0;
+
+ cpi->b_calculate_psnr = CONFIG_INTERNAL_STATS;
+#if CONFIG_INTERNAL_STATS
+ cpi->b_calculate_ssimg = 0;
+
+ cpi->count = 0;
+ cpi->bytes = 0;
+
+ if (cpi->b_calculate_psnr)
+ {
+ cpi->total_sq_error = 0.0;
+ cpi->total_sq_error2 = 0.0;
+ cpi->total_y = 0.0;
+ cpi->total_u = 0.0;
+ cpi->total_v = 0.0;
+ cpi->total = 0.0;
+ cpi->totalp_y = 0.0;
+ cpi->totalp_u = 0.0;
+ cpi->totalp_v = 0.0;
+ cpi->totalp = 0.0;
+ cpi->tot_recode_hits = 0;
+ cpi->summed_quality = 0;
+ cpi->summed_weights = 0;
+ }
+
+ if (cpi->b_calculate_ssimg)
+ {
+ cpi->total_ssimg_y = 0;
+ cpi->total_ssimg_u = 0;
+ cpi->total_ssimg_v = 0;
+ cpi->total_ssimg_all = 0;
+ }
+
+#endif
+
+ cpi->first_time_stamp_ever = 0x7FFFFFFF;
+
+ cpi->frames_till_gf_update_due = 0;
+ cpi->key_frame_count = 1;
+
+ cpi->ni_av_qi = cpi->oxcf.worst_allowed_q;
+ cpi->ni_tot_qi = 0;
+ cpi->ni_frames = 0;
+ cpi->total_byte_count = 0;
+
+ cpi->drop_frame = 0;
+
+ cpi->rate_correction_factor = 1.0;
+ cpi->key_frame_rate_correction_factor = 1.0;
+ cpi->gf_rate_correction_factor = 1.0;
+ cpi->twopass.est_max_qcorrection_factor = 1.0;
+
+ for (i = 0; i < KEY_FRAME_CONTEXT; i++)
+ {
+ cpi->prior_key_frame_distance[i] = (int)cpi->output_frame_rate;
+ }
+
+#ifdef OUTPUT_YUV_SRC
+ yuv_file = fopen("bd.yuv", "ab");
+#endif
+
+#if 0
+ framepsnr = fopen("framepsnr.stt", "a");
+ kf_list = fopen("kf_list.stt", "w");
+#endif
+
+ cpi->output_pkt_list = oxcf->output_pkt_list;
+
+#if !(CONFIG_REALTIME_ONLY)
+
+ if (cpi->pass == 1)
+ {
+ vp8_init_first_pass(cpi);
+ }
+ else if (cpi->pass == 2)
+ {
+ size_t packet_sz = sizeof(FIRSTPASS_STATS);
+ int packets = (int)(oxcf->two_pass_stats_in.sz / packet_sz);
+
+ cpi->twopass.stats_in_start = oxcf->two_pass_stats_in.buf;
+ cpi->twopass.stats_in = cpi->twopass.stats_in_start;
+ cpi->twopass.stats_in_end = (void*)((char *)cpi->twopass.stats_in
+ + (packets - 1) * packet_sz);
+ vp8_init_second_pass(cpi);
+ }
+
+#endif
+
+ if (cpi->compressor_speed == 2)
+ {
+ cpi->avg_encode_time = 0;
+ cpi->avg_pick_mode_time = 0;
+ }
+
+ vp8_set_speed_features(cpi);
+
+ /* Set starting values of RD threshold multipliers (128 = *1) */
+ for (i = 0; i < MAX_MODES; i++)
+ {
+ cpi->rd_thresh_mult[i] = 128;
+ }
+
+#ifdef ENTROPY_STATS
+ init_mv_ref_counts();
+#endif
+
+#if CONFIG_MULTITHREAD
+ if(vp8cx_create_encoder_threads(cpi))
+ {
+ vp8_remove_compressor(&cpi);
+ return 0;
+ }
+#endif
+
+ cpi->fn_ptr[BLOCK_16X16].sdf = vp8_sad16x16;
+ cpi->fn_ptr[BLOCK_16X16].vf = vp8_variance16x16;
+ cpi->fn_ptr[BLOCK_16X16].svf = vp8_sub_pixel_variance16x16;
+ cpi->fn_ptr[BLOCK_16X16].svf_halfpix_h = vp8_variance_halfpixvar16x16_h;
+ cpi->fn_ptr[BLOCK_16X16].svf_halfpix_v = vp8_variance_halfpixvar16x16_v;
+ cpi->fn_ptr[BLOCK_16X16].svf_halfpix_hv = vp8_variance_halfpixvar16x16_hv;
+ cpi->fn_ptr[BLOCK_16X16].sdx3f = vp8_sad16x16x3;
+ cpi->fn_ptr[BLOCK_16X16].sdx8f = vp8_sad16x16x8;
+ cpi->fn_ptr[BLOCK_16X16].sdx4df = vp8_sad16x16x4d;
+
+ cpi->fn_ptr[BLOCK_16X8].sdf = vp8_sad16x8;
+ cpi->fn_ptr[BLOCK_16X8].vf = vp8_variance16x8;
+ cpi->fn_ptr[BLOCK_16X8].svf = vp8_sub_pixel_variance16x8;
+ cpi->fn_ptr[BLOCK_16X8].svf_halfpix_h = NULL;
+ cpi->fn_ptr[BLOCK_16X8].svf_halfpix_v = NULL;
+ cpi->fn_ptr[BLOCK_16X8].svf_halfpix_hv = NULL;
+ cpi->fn_ptr[BLOCK_16X8].sdx3f = vp8_sad16x8x3;
+ cpi->fn_ptr[BLOCK_16X8].sdx8f = vp8_sad16x8x8;
+ cpi->fn_ptr[BLOCK_16X8].sdx4df = vp8_sad16x8x4d;
+
+ cpi->fn_ptr[BLOCK_8X16].sdf = vp8_sad8x16;
+ cpi->fn_ptr[BLOCK_8X16].vf = vp8_variance8x16;
+ cpi->fn_ptr[BLOCK_8X16].svf = vp8_sub_pixel_variance8x16;
+ cpi->fn_ptr[BLOCK_8X16].svf_halfpix_h = NULL;
+ cpi->fn_ptr[BLOCK_8X16].svf_halfpix_v = NULL;
+ cpi->fn_ptr[BLOCK_8X16].svf_halfpix_hv = NULL;
+ cpi->fn_ptr[BLOCK_8X16].sdx3f = vp8_sad8x16x3;
+ cpi->fn_ptr[BLOCK_8X16].sdx8f = vp8_sad8x16x8;
+ cpi->fn_ptr[BLOCK_8X16].sdx4df = vp8_sad8x16x4d;
+
+ cpi->fn_ptr[BLOCK_8X8].sdf = vp8_sad8x8;
+ cpi->fn_ptr[BLOCK_8X8].vf = vp8_variance8x8;
+ cpi->fn_ptr[BLOCK_8X8].svf = vp8_sub_pixel_variance8x8;
+ cpi->fn_ptr[BLOCK_8X8].svf_halfpix_h = NULL;
+ cpi->fn_ptr[BLOCK_8X8].svf_halfpix_v = NULL;
+ cpi->fn_ptr[BLOCK_8X8].svf_halfpix_hv = NULL;
+ cpi->fn_ptr[BLOCK_8X8].sdx3f = vp8_sad8x8x3;
+ cpi->fn_ptr[BLOCK_8X8].sdx8f = vp8_sad8x8x8;
+ cpi->fn_ptr[BLOCK_8X8].sdx4df = vp8_sad8x8x4d;
+
+ cpi->fn_ptr[BLOCK_4X4].sdf = vp8_sad4x4;
+ cpi->fn_ptr[BLOCK_4X4].vf = vp8_variance4x4;
+ cpi->fn_ptr[BLOCK_4X4].svf = vp8_sub_pixel_variance4x4;
+ cpi->fn_ptr[BLOCK_4X4].svf_halfpix_h = NULL;
+ cpi->fn_ptr[BLOCK_4X4].svf_halfpix_v = NULL;
+ cpi->fn_ptr[BLOCK_4X4].svf_halfpix_hv = NULL;
+ cpi->fn_ptr[BLOCK_4X4].sdx3f = vp8_sad4x4x3;
+ cpi->fn_ptr[BLOCK_4X4].sdx8f = vp8_sad4x4x8;
+ cpi->fn_ptr[BLOCK_4X4].sdx4df = vp8_sad4x4x4d;
+
+#if ARCH_X86 || ARCH_X86_64
+ cpi->fn_ptr[BLOCK_16X16].copymem = vp8_copy32xn;
+ cpi->fn_ptr[BLOCK_16X8].copymem = vp8_copy32xn;
+ cpi->fn_ptr[BLOCK_8X16].copymem = vp8_copy32xn;
+ cpi->fn_ptr[BLOCK_8X8].copymem = vp8_copy32xn;
+ cpi->fn_ptr[BLOCK_4X4].copymem = vp8_copy32xn;
+#endif
+
+ cpi->full_search_sad = vp8_full_search_sad;
+ cpi->diamond_search_sad = vp8_diamond_search_sad;
+ cpi->refining_search_sad = vp8_refining_search_sad;
+
+ /* make sure frame 1 is okay */
+ cpi->error_bins[0] = cpi->common.MBs;
+
+ /* vp8cx_init_quantizer() is first called here. Add check in
+ * vp8cx_frame_init_quantizer() so that vp8cx_init_quantizer is only
+ * called later when needed. This will avoid unnecessary calls of
+ * vp8cx_init_quantizer() for every frame.
+ */
+ vp8cx_init_quantizer(cpi);
+
+ vp8_loop_filter_init(cm);
+
+ cpi->common.error.setjmp = 0;
+
+#if CONFIG_MULTI_RES_ENCODING
+
+ /* Calculate # of MBs in a row in lower-resolution level image. */
+ if (cpi->oxcf.mr_encoder_id > 0)
+ vp8_cal_low_res_mb_cols(cpi);
+
+#endif
+
+ /* setup RD costs to MACROBLOCK struct */
+
+ cpi->mb.mvcost[0] = &cpi->rd_costs.mvcosts[0][mv_max+1];
+ cpi->mb.mvcost[1] = &cpi->rd_costs.mvcosts[1][mv_max+1];
+ cpi->mb.mvsadcost[0] = &cpi->rd_costs.mvsadcosts[0][mvfp_max+1];
+ cpi->mb.mvsadcost[1] = &cpi->rd_costs.mvsadcosts[1][mvfp_max+1];
+
+ cal_mvsadcosts(cpi->mb.mvsadcost);
+
+ cpi->mb.mbmode_cost = cpi->rd_costs.mbmode_cost;
+ cpi->mb.intra_uv_mode_cost = cpi->rd_costs.intra_uv_mode_cost;
+ cpi->mb.bmode_costs = cpi->rd_costs.bmode_costs;
+ cpi->mb.inter_bmode_costs = cpi->rd_costs.inter_bmode_costs;
+ cpi->mb.token_costs = cpi->rd_costs.token_costs;
+
+ /* setup block ptrs & offsets */
+ vp8_setup_block_ptrs(&cpi->mb);
+ vp8_setup_block_dptrs(&cpi->mb.e_mbd);
+
+ return cpi;
+}
+
+
+void vp8_remove_compressor(VP8_COMP **ptr)
+{
+ VP8_COMP *cpi = *ptr;
+
+ if (!cpi)
+ return;
+
+ if (cpi && (cpi->common.current_video_frame > 0))
+ {
+#if !(CONFIG_REALTIME_ONLY)
+
+ if (cpi->pass == 2)
+ {
+ vp8_end_second_pass(cpi);
+ }
+
+#endif
+
+#ifdef ENTROPY_STATS
+ print_context_counters();
+ print_tree_update_probs();
+ print_mode_context();
+#endif
+
+#if CONFIG_INTERNAL_STATS
+
+ if (cpi->pass != 1)
+ {
+ FILE *f = fopen("opsnr.stt", "a");
+ double time_encoded = (cpi->last_end_time_stamp_seen
+ - cpi->first_time_stamp_ever) / 10000000.000;
+ double total_encode_time = (cpi->time_receive_data +
+ cpi->time_compress_data) / 1000.000;
+ double dr = (double)cpi->bytes * 8.0 / 1000.0 / time_encoded;
+
+ if (cpi->b_calculate_psnr)
+ {
+ YV12_BUFFER_CONFIG *lst_yv12 =
+ &cpi->common.yv12_fb[cpi->common.lst_fb_idx];
+
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ int i;
+
+ fprintf(f, "Layer\tBitrate\tAVGPsnr\tGLBPsnr\tAVPsnrP\t"
+ "GLPsnrP\tVPXSSIM\t\n");
+ for (i=0; i<(int)cpi->oxcf.number_of_layers; i++)
+ {
+ double dr = (double)cpi->bytes_in_layer[i] *
+ 8.0 / 1000.0 / time_encoded;
+ double samples = 3.0 / 2 * cpi->frames_in_layer[i] *
+ lst_yv12->y_width * lst_yv12->y_height;
+ double total_psnr = vp8_mse2psnr(samples, 255.0,
+ cpi->total_error2[i]);
+ double total_psnr2 = vp8_mse2psnr(samples, 255.0,
+ cpi->total_error2_p[i]);
+ double total_ssim = 100 * pow(cpi->sum_ssim[i] /
+ cpi->sum_weights[i], 8.0);
+
+ fprintf(f, "%5d\t%7.3f\t%7.3f\t%7.3f\t%7.3f\t"
+ "%7.3f\t%7.3f\n",
+ i, dr,
+ cpi->sum_psnr[i] / cpi->frames_in_layer[i],
+ total_psnr,
+ cpi->sum_psnr_p[i] / cpi->frames_in_layer[i],
+ total_psnr2, total_ssim);
+ }
+ }
+ else
+ {
+ double samples = 3.0 / 2 * cpi->count *
+ lst_yv12->y_width * lst_yv12->y_height;
+ double total_psnr = vp8_mse2psnr(samples, 255.0,
+ cpi->total_sq_error);
+ double total_psnr2 = vp8_mse2psnr(samples, 255.0,
+ cpi->total_sq_error2);
+ double total_ssim = 100 * pow(cpi->summed_quality /
+ cpi->summed_weights, 8.0);
+
+ fprintf(f, "Bitrate\tAVGPsnr\tGLBPsnr\tAVPsnrP\t"
+ "GLPsnrP\tVPXSSIM\t Time(us)\n");
+ fprintf(f, "%7.3f\t%7.3f\t%7.3f\t%7.3f\t%7.3f\t"
+ "%7.3f\t%8.0f\n",
+ dr, cpi->total / cpi->count, total_psnr,
+ cpi->totalp / cpi->count, total_psnr2,
+ total_ssim, total_encode_time);
+ }
+ }
+
+ if (cpi->b_calculate_ssimg)
+ {
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ int i;
+
+ fprintf(f, "Layer\tBitRate\tSSIM_Y\tSSIM_U\tSSIM_V\tSSIM_A\t"
+ "Time(us)\n");
+ for (i=0; i<(int)cpi->oxcf.number_of_layers; i++)
+ {
+ double dr = (double)cpi->bytes_in_layer[i] *
+ 8.0 / 1000.0 / time_encoded;
+ fprintf(f, "%5d\t%7.3f\t%6.4f\t"
+ "%6.4f\t%6.4f\t%6.4f\t%8.0f\n",
+ i, dr,
+ cpi->total_ssimg_y_in_layer[i] /
+ cpi->frames_in_layer[i],
+ cpi->total_ssimg_u_in_layer[i] /
+ cpi->frames_in_layer[i],
+ cpi->total_ssimg_v_in_layer[i] /
+ cpi->frames_in_layer[i],
+ cpi->total_ssimg_all_in_layer[i] /
+ cpi->frames_in_layer[i],
+ total_encode_time);
+ }
+ }
+ else
+ {
+ fprintf(f, "BitRate\tSSIM_Y\tSSIM_U\tSSIM_V\tSSIM_A\t"
+ "Time(us)\n");
+ fprintf(f, "%7.3f\t%6.4f\t%6.4f\t%6.4f\t%6.4f\t%8.0f\n", dr,
+ cpi->total_ssimg_y / cpi->count,
+ cpi->total_ssimg_u / cpi->count,
+ cpi->total_ssimg_v / cpi->count,
+ cpi->total_ssimg_all / cpi->count, total_encode_time);
+ }
+ }
+
+ fclose(f);
+#if 0
+ f = fopen("qskip.stt", "a");
+ fprintf(f, "minq:%d -maxq:%d skiptrue:skipfalse = %d:%d\n", cpi->oxcf.best_allowed_q, cpi->oxcf.worst_allowed_q, skiptruecount, skipfalsecount);
+ fclose(f);
+#endif
+
+ }
+
+#endif
+
+
+#ifdef SPEEDSTATS
+
+ if (cpi->compressor_speed == 2)
+ {
+ int i;
+ FILE *f = fopen("cxspeed.stt", "a");
+ cnt_pm /= cpi->common.MBs;
+
+ for (i = 0; i < 16; i++)
+ fprintf(f, "%5d", frames_at_speed[i]);
+
+ fprintf(f, "\n");
+ fclose(f);
+ }
+
+#endif
+
+
+#ifdef MODE_STATS
+ {
+ extern int count_mb_seg[4];
+ FILE *f = fopen("modes.stt", "a");
+ double dr = (double)cpi->frame_rate * (double)bytes * (double)8 / (double)count / (double)1000 ;
+ fprintf(f, "intra_mode in Intra Frames:\n");
+ fprintf(f, "Y: %8d, %8d, %8d, %8d, %8d\n", y_modes[0], y_modes[1], y_modes[2], y_modes[3], y_modes[4]);
+ fprintf(f, "UV:%8d, %8d, %8d, %8d\n", uv_modes[0], uv_modes[1], uv_modes[2], uv_modes[3]);
+ fprintf(f, "B: ");
+ {
+ int i;
+
+ for (i = 0; i < 10; i++)
+ fprintf(f, "%8d, ", b_modes[i]);
+
+ fprintf(f, "\n");
+
+ }
+
+ fprintf(f, "Modes in Inter Frames:\n");
+ fprintf(f, "Y: %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d\n",
+ inter_y_modes[0], inter_y_modes[1], inter_y_modes[2], inter_y_modes[3], inter_y_modes[4],
+ inter_y_modes[5], inter_y_modes[6], inter_y_modes[7], inter_y_modes[8], inter_y_modes[9]);
+ fprintf(f, "UV:%8d, %8d, %8d, %8d\n", inter_uv_modes[0], inter_uv_modes[1], inter_uv_modes[2], inter_uv_modes[3]);
+ fprintf(f, "B: ");
+ {
+ int i;
+
+ for (i = 0; i < 15; i++)
+ fprintf(f, "%8d, ", inter_b_modes[i]);
+
+ fprintf(f, "\n");
+
+ }
+ fprintf(f, "P:%8d, %8d, %8d, %8d\n", count_mb_seg[0], count_mb_seg[1], count_mb_seg[2], count_mb_seg[3]);
+ fprintf(f, "PB:%8d, %8d, %8d, %8d\n", inter_b_modes[LEFT4X4], inter_b_modes[ABOVE4X4], inter_b_modes[ZERO4X4], inter_b_modes[NEW4X4]);
+
+
+
+ fclose(f);
+ }
+#endif
+
+#ifdef ENTROPY_STATS
+ {
+ int i, j, k;
+ FILE *fmode = fopen("modecontext.c", "w");
+
+ fprintf(fmode, "\n#include \"entropymode.h\"\n\n");
+ fprintf(fmode, "const unsigned int vp8_kf_default_bmode_counts ");
+ fprintf(fmode, "[VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES] =\n{\n");
+
+ for (i = 0; i < 10; i++)
+ {
+
+ fprintf(fmode, " { /* Above Mode : %d */\n", i);
+
+ for (j = 0; j < 10; j++)
+ {
+
+ fprintf(fmode, " {");
+
+ for (k = 0; k < 10; k++)
+ {
+ if (!intra_mode_stats[i][j][k])
+ fprintf(fmode, " %5d, ", 1);
+ else
+ fprintf(fmode, " %5d, ", intra_mode_stats[i][j][k]);
+ }
+
+ fprintf(fmode, "}, /* left_mode %d */\n", j);
+
+ }
+
+ fprintf(fmode, " },\n");
+
+ }
+
+ fprintf(fmode, "};\n");
+ fclose(fmode);
+ }
+#endif
+
+
+#if defined(SECTIONBITS_OUTPUT)
+
+ if (0)
+ {
+ int i;
+ FILE *f = fopen("tokenbits.stt", "a");
+
+ for (i = 0; i < 28; i++)
+ fprintf(f, "%8d", (int)(Sectionbits[i] / 256));
+
+ fprintf(f, "\n");
+ fclose(f);
+ }
+
+#endif
+
+#if 0
+ {
+ printf("\n_pick_loop_filter_level:%d\n", cpi->time_pick_lpf / 1000);
+ printf("\n_frames recive_data encod_mb_row compress_frame Total\n");
+ printf("%6d %10ld %10ld %10ld %10ld\n", cpi->common.current_video_frame, cpi->time_receive_data / 1000, cpi->time_encode_mb_row / 1000, cpi->time_compress_data / 1000, (cpi->time_receive_data + cpi->time_compress_data) / 1000);
+ }
+#endif
+
+ }
+
+#if CONFIG_MULTITHREAD
+ vp8cx_remove_encoder_threads(cpi);
+#endif
+
+#if CONFIG_TEMPORAL_DENOISING
+ vp8_denoiser_free(&cpi->denoiser);
+#endif
+ dealloc_compressor_data(cpi);
+ vpx_free(cpi->mb.ss);
+ vpx_free(cpi->tok);
+ vpx_free(cpi->cyclic_refresh_map);
+
+ vp8_remove_common(&cpi->common);
+ vpx_free(cpi);
+ *ptr = 0;
+
+#ifdef OUTPUT_YUV_SRC
+ fclose(yuv_file);
+#endif
+
+#if 0
+
+ if (keyfile)
+ fclose(keyfile);
+
+ if (framepsnr)
+ fclose(framepsnr);
+
+ if (kf_list)
+ fclose(kf_list);
+
+#endif
+
+}
+
+
+static uint64_t calc_plane_error(unsigned char *orig, int orig_stride,
+ unsigned char *recon, int recon_stride,
+ unsigned int cols, unsigned int rows)
+{
+ unsigned int row, col;
+ uint64_t total_sse = 0;
+ int diff;
+
+ for (row = 0; row + 16 <= rows; row += 16)
+ {
+ for (col = 0; col + 16 <= cols; col += 16)
+ {
+ unsigned int sse;
+
+ vp8_mse16x16(orig + col, orig_stride,
+ recon + col, recon_stride,
+ &sse);
+ total_sse += sse;
+ }
+
+ /* Handle odd-sized width */
+ if (col < cols)
+ {
+ unsigned int border_row, border_col;
+ unsigned char *border_orig = orig;
+ unsigned char *border_recon = recon;
+
+ for (border_row = 0; border_row < 16; border_row++)
+ {
+ for (border_col = col; border_col < cols; border_col++)
+ {
+ diff = border_orig[border_col] - border_recon[border_col];
+ total_sse += diff * diff;
+ }
+
+ border_orig += orig_stride;
+ border_recon += recon_stride;
+ }
+ }
+
+ orig += orig_stride * 16;
+ recon += recon_stride * 16;
+ }
+
+ /* Handle odd-sized height */
+ for (; row < rows; row++)
+ {
+ for (col = 0; col < cols; col++)
+ {
+ diff = orig[col] - recon[col];
+ total_sse += diff * diff;
+ }
+
+ orig += orig_stride;
+ recon += recon_stride;
+ }
+
+ vp8_clear_system_state();
+ return total_sse;
+}
+
+
+static void generate_psnr_packet(VP8_COMP *cpi)
+{
+ YV12_BUFFER_CONFIG *orig = cpi->Source;
+ YV12_BUFFER_CONFIG *recon = cpi->common.frame_to_show;
+ struct vpx_codec_cx_pkt pkt;
+ uint64_t sse;
+ int i;
+ unsigned int width = cpi->common.Width;
+ unsigned int height = cpi->common.Height;
+
+ pkt.kind = VPX_CODEC_PSNR_PKT;
+ sse = calc_plane_error(orig->y_buffer, orig->y_stride,
+ recon->y_buffer, recon->y_stride,
+ width, height);
+ pkt.data.psnr.sse[0] = sse;
+ pkt.data.psnr.sse[1] = sse;
+ pkt.data.psnr.samples[0] = width * height;
+ pkt.data.psnr.samples[1] = width * height;
+
+ width = (width + 1) / 2;
+ height = (height + 1) / 2;
+
+ sse = calc_plane_error(orig->u_buffer, orig->uv_stride,
+ recon->u_buffer, recon->uv_stride,
+ width, height);
+ pkt.data.psnr.sse[0] += sse;
+ pkt.data.psnr.sse[2] = sse;
+ pkt.data.psnr.samples[0] += width * height;
+ pkt.data.psnr.samples[2] = width * height;
+
+ sse = calc_plane_error(orig->v_buffer, orig->uv_stride,
+ recon->v_buffer, recon->uv_stride,
+ width, height);
+ pkt.data.psnr.sse[0] += sse;
+ pkt.data.psnr.sse[3] = sse;
+ pkt.data.psnr.samples[0] += width * height;
+ pkt.data.psnr.samples[3] = width * height;
+
+ for (i = 0; i < 4; i++)
+ pkt.data.psnr.psnr[i] = vp8_mse2psnr(pkt.data.psnr.samples[i], 255.0,
+ (double)(pkt.data.psnr.sse[i]));
+
+ vpx_codec_pkt_list_add(cpi->output_pkt_list, &pkt);
+}
+
+
+int vp8_use_as_reference(VP8_COMP *cpi, int ref_frame_flags)
+{
+ if (ref_frame_flags > 7)
+ return -1 ;
+
+ cpi->ref_frame_flags = ref_frame_flags;
+ return 0;
+}
+int vp8_update_reference(VP8_COMP *cpi, int ref_frame_flags)
+{
+ if (ref_frame_flags > 7)
+ return -1 ;
+
+ cpi->common.refresh_golden_frame = 0;
+ cpi->common.refresh_alt_ref_frame = 0;
+ cpi->common.refresh_last_frame = 0;
+
+ if (ref_frame_flags & VP8_LAST_FRAME)
+ cpi->common.refresh_last_frame = 1;
+
+ if (ref_frame_flags & VP8_GOLD_FRAME)
+ cpi->common.refresh_golden_frame = 1;
+
+ if (ref_frame_flags & VP8_ALTR_FRAME)
+ cpi->common.refresh_alt_ref_frame = 1;
+
+ return 0;
+}
+
+int vp8_get_reference(VP8_COMP *cpi, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd)
+{
+ VP8_COMMON *cm = &cpi->common;
+ int ref_fb_idx;
+
+ if (ref_frame_flag == VP8_LAST_FRAME)
+ ref_fb_idx = cm->lst_fb_idx;
+ else if (ref_frame_flag == VP8_GOLD_FRAME)
+ ref_fb_idx = cm->gld_fb_idx;
+ else if (ref_frame_flag == VP8_ALTR_FRAME)
+ ref_fb_idx = cm->alt_fb_idx;
+ else
+ return -1;
+
+ vp8_yv12_copy_frame(&cm->yv12_fb[ref_fb_idx], sd);
+
+ return 0;
+}
+int vp8_set_reference(VP8_COMP *cpi, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ int ref_fb_idx;
+
+ if (ref_frame_flag == VP8_LAST_FRAME)
+ ref_fb_idx = cm->lst_fb_idx;
+ else if (ref_frame_flag == VP8_GOLD_FRAME)
+ ref_fb_idx = cm->gld_fb_idx;
+ else if (ref_frame_flag == VP8_ALTR_FRAME)
+ ref_fb_idx = cm->alt_fb_idx;
+ else
+ return -1;
+
+ vp8_yv12_copy_frame(sd, &cm->yv12_fb[ref_fb_idx]);
+
+ return 0;
+}
+int vp8_update_entropy(VP8_COMP *cpi, int update)
+{
+ VP8_COMMON *cm = &cpi->common;
+ cm->refresh_entropy_probs = update;
+
+ return 0;
+}
+
+
+#if OUTPUT_YUV_SRC
+void vp8_write_yuv_frame(const char *name, YV12_BUFFER_CONFIG *s)
+{
+ FILE *yuv_file = fopen(name, "ab");
+ unsigned char *src = s->y_buffer;
+ int h = s->y_height;
+
+ do
+ {
+ fwrite(src, s->y_width, 1, yuv_file);
+ src += s->y_stride;
+ }
+ while (--h);
+
+ src = s->u_buffer;
+ h = s->uv_height;
+
+ do
+ {
+ fwrite(src, s->uv_width, 1, yuv_file);
+ src += s->uv_stride;
+ }
+ while (--h);
+
+ src = s->v_buffer;
+ h = s->uv_height;
+
+ do
+ {
+ fwrite(src, s->uv_width, 1, yuv_file);
+ src += s->uv_stride;
+ }
+ while (--h);
+
+ fclose(yuv_file);
+}
+#endif
+
+
+static void scale_and_extend_source(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ /* are we resizing the image */
+ if (cm->horiz_scale != 0 || cm->vert_scale != 0)
+ {
+#if CONFIG_SPATIAL_RESAMPLING
+ int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
+ int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
+ int tmp_height;
+
+ if (cm->vert_scale == 3)
+ tmp_height = 9;
+ else
+ tmp_height = 11;
+
+ Scale2Ratio(cm->horiz_scale, &hr, &hs);
+ Scale2Ratio(cm->vert_scale, &vr, &vs);
+
+ vp8_scale_frame(sd, &cpi->scaled_source, cm->temp_scale_frame.y_buffer,
+ tmp_height, hs, hr, vs, vr, 0);
+
+ vp8_yv12_extend_frame_borders(&cpi->scaled_source);
+ cpi->Source = &cpi->scaled_source;
+#endif
+ }
+ else
+ cpi->Source = sd;
+}
+
+
+static int resize_key_frame(VP8_COMP *cpi)
+{
+#if CONFIG_SPATIAL_RESAMPLING
+ VP8_COMMON *cm = &cpi->common;
+
+ /* Do we need to apply resampling for one pass cbr.
+ * In one pass this is more limited than in two pass cbr
+ * The test and any change is only made one per key frame sequence
+ */
+ if (cpi->oxcf.allow_spatial_resampling && (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER))
+ {
+ int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
+ int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
+ int new_width, new_height;
+
+ /* If we are below the resample DOWN watermark then scale down a
+ * notch.
+ */
+ if (cpi->buffer_level < (cpi->oxcf.resample_down_water_mark * cpi->oxcf.optimal_buffer_level / 100))
+ {
+ cm->horiz_scale = (cm->horiz_scale < ONETWO) ? cm->horiz_scale + 1 : ONETWO;
+ cm->vert_scale = (cm->vert_scale < ONETWO) ? cm->vert_scale + 1 : ONETWO;
+ }
+ /* Should we now start scaling back up */
+ else if (cpi->buffer_level > (cpi->oxcf.resample_up_water_mark * cpi->oxcf.optimal_buffer_level / 100))
+ {
+ cm->horiz_scale = (cm->horiz_scale > NORMAL) ? cm->horiz_scale - 1 : NORMAL;
+ cm->vert_scale = (cm->vert_scale > NORMAL) ? cm->vert_scale - 1 : NORMAL;
+ }
+
+ /* Get the new hieght and width */
+ Scale2Ratio(cm->horiz_scale, &hr, &hs);
+ Scale2Ratio(cm->vert_scale, &vr, &vs);
+ new_width = ((hs - 1) + (cpi->oxcf.Width * hr)) / hs;
+ new_height = ((vs - 1) + (cpi->oxcf.Height * vr)) / vs;
+
+ /* If the image size has changed we need to reallocate the buffers
+ * and resample the source image
+ */
+ if ((cm->Width != new_width) || (cm->Height != new_height))
+ {
+ cm->Width = new_width;
+ cm->Height = new_height;
+ vp8_alloc_compressor_data(cpi);
+ scale_and_extend_source(cpi->un_scaled_source, cpi);
+ return 1;
+ }
+ }
+
+#endif
+ return 0;
+}
+
+
+static void update_alt_ref_frame_stats(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ /* Select an interval before next GF or altref */
+ if (!cpi->auto_gold)
+ cpi->frames_till_gf_update_due = DEFAULT_GF_INTERVAL;
+
+ if ((cpi->pass != 2) && cpi->frames_till_gf_update_due)
+ {
+ cpi->current_gf_interval = cpi->frames_till_gf_update_due;
+
+ /* Set the bits per frame that we should try and recover in
+ * subsequent inter frames to account for the extra GF spend...
+ * note that his does not apply for GF updates that occur
+ * coincident with a key frame as the extra cost of key frames is
+ * dealt with elsewhere.
+ */
+ cpi->gf_overspend_bits += cpi->projected_frame_size;
+ cpi->non_gf_bitrate_adjustment = cpi->gf_overspend_bits / cpi->frames_till_gf_update_due;
+ }
+
+ /* Update data structure that monitors level of reference to last GF */
+ vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols));
+ cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
+
+ /* this frame refreshes means next frames don't unless specified by user */
+ cpi->common.frames_since_golden = 0;
+
+ /* Clear the alternate reference update pending flag. */
+ cpi->source_alt_ref_pending = 0;
+
+ /* Set the alternate refernce frame active flag */
+ cpi->source_alt_ref_active = 1;
+
+
+}
+static void update_golden_frame_stats(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ /* Update the Golden frame usage counts. */
+ if (cm->refresh_golden_frame)
+ {
+ /* Select an interval before next GF */
+ if (!cpi->auto_gold)
+ cpi->frames_till_gf_update_due = DEFAULT_GF_INTERVAL;
+
+ if ((cpi->pass != 2) && (cpi->frames_till_gf_update_due > 0))
+ {
+ cpi->current_gf_interval = cpi->frames_till_gf_update_due;
+
+ /* Set the bits per frame that we should try and recover in
+ * subsequent inter frames to account for the extra GF spend...
+ * note that his does not apply for GF updates that occur
+ * coincident with a key frame as the extra cost of key frames
+ * is dealt with elsewhere.
+ */
+ if ((cm->frame_type != KEY_FRAME) && !cpi->source_alt_ref_active)
+ {
+ /* Calcluate GF bits to be recovered
+ * Projected size - av frame bits available for inter
+ * frames for clip as a whole
+ */
+ cpi->gf_overspend_bits += (cpi->projected_frame_size - cpi->inter_frame_target);
+ }
+
+ cpi->non_gf_bitrate_adjustment = cpi->gf_overspend_bits / cpi->frames_till_gf_update_due;
+
+ }
+
+ /* Update data structure that monitors level of reference to last GF */
+ vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols));
+ cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
+
+ /* this frame refreshes means next frames don't unless specified by
+ * user
+ */
+ cm->refresh_golden_frame = 0;
+ cpi->common.frames_since_golden = 0;
+
+ cpi->recent_ref_frame_usage[INTRA_FRAME] = 1;
+ cpi->recent_ref_frame_usage[LAST_FRAME] = 1;
+ cpi->recent_ref_frame_usage[GOLDEN_FRAME] = 1;
+ cpi->recent_ref_frame_usage[ALTREF_FRAME] = 1;
+
+ /* ******** Fixed Q test code only ************ */
+ /* If we are going to use the ALT reference for the next group of
+ * frames set a flag to say so.
+ */
+ if (cpi->oxcf.fixed_q >= 0 &&
+ cpi->oxcf.play_alternate && !cpi->common.refresh_alt_ref_frame)
+ {
+ cpi->source_alt_ref_pending = 1;
+ cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
+ }
+
+ if (!cpi->source_alt_ref_pending)
+ cpi->source_alt_ref_active = 0;
+
+ /* Decrement count down till next gf */
+ if (cpi->frames_till_gf_update_due > 0)
+ cpi->frames_till_gf_update_due--;
+
+ }
+ else if (!cpi->common.refresh_alt_ref_frame)
+ {
+ /* Decrement count down till next gf */
+ if (cpi->frames_till_gf_update_due > 0)
+ cpi->frames_till_gf_update_due--;
+
+ if (cpi->common.frames_till_alt_ref_frame)
+ cpi->common.frames_till_alt_ref_frame --;
+
+ cpi->common.frames_since_golden ++;
+
+ if (cpi->common.frames_since_golden > 1)
+ {
+ cpi->recent_ref_frame_usage[INTRA_FRAME] += cpi->count_mb_ref_frame_usage[INTRA_FRAME];
+ cpi->recent_ref_frame_usage[LAST_FRAME] += cpi->count_mb_ref_frame_usage[LAST_FRAME];
+ cpi->recent_ref_frame_usage[GOLDEN_FRAME] += cpi->count_mb_ref_frame_usage[GOLDEN_FRAME];
+ cpi->recent_ref_frame_usage[ALTREF_FRAME] += cpi->count_mb_ref_frame_usage[ALTREF_FRAME];
+ }
+ }
+}
+
+/* This function updates the reference frame probability estimates that
+ * will be used during mode selection
+ */
+static void update_rd_ref_frame_probs(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ const int *const rfct = cpi->count_mb_ref_frame_usage;
+ const int rf_intra = rfct[INTRA_FRAME];
+ const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
+
+ if (cm->frame_type == KEY_FRAME)
+ {
+ cpi->prob_intra_coded = 255;
+ cpi->prob_last_coded = 128;
+ cpi->prob_gf_coded = 128;
+ }
+ else if (!(rf_intra + rf_inter))
+ {
+ cpi->prob_intra_coded = 63;
+ cpi->prob_last_coded = 128;
+ cpi->prob_gf_coded = 128;
+ }
+
+ /* update reference frame costs since we can do better than what we got
+ * last frame.
+ */
+ if (cpi->oxcf.number_of_layers == 1)
+ {
+ if (cpi->common.refresh_alt_ref_frame)
+ {
+ cpi->prob_intra_coded += 40;
+ cpi->prob_last_coded = 200;
+ cpi->prob_gf_coded = 1;
+ }
+ else if (cpi->common.frames_since_golden == 0)
+ {
+ cpi->prob_last_coded = 214;
+ }
+ else if (cpi->common.frames_since_golden == 1)
+ {
+ cpi->prob_last_coded = 192;
+ cpi->prob_gf_coded = 220;
+ }
+ else if (cpi->source_alt_ref_active)
+ {
+ cpi->prob_gf_coded -= 20;
+
+ if (cpi->prob_gf_coded < 10)
+ cpi->prob_gf_coded = 10;
+ }
+ if (!cpi->source_alt_ref_active)
+ cpi->prob_gf_coded = 255;
+ }
+}
+
+
+/* 1 = key, 0 = inter */
+static int decide_key_frame(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ int code_key_frame = 0;
+
+ cpi->kf_boost = 0;
+
+ if (cpi->Speed > 11)
+ return 0;
+
+ /* Clear down mmx registers */
+ vp8_clear_system_state();
+
+ if ((cpi->compressor_speed == 2) && (cpi->Speed >= 5) && (cpi->sf.RD == 0))
+ {
+ double change = 1.0 * abs((int)(cpi->mb.intra_error -
+ cpi->last_intra_error)) / (1 + cpi->last_intra_error);
+ double change2 = 1.0 * abs((int)(cpi->mb.prediction_error -
+ cpi->last_prediction_error)) / (1 + cpi->last_prediction_error);
+ double minerror = cm->MBs * 256;
+
+ cpi->last_intra_error = cpi->mb.intra_error;
+ cpi->last_prediction_error = cpi->mb.prediction_error;
+
+ if (10 * cpi->mb.intra_error / (1 + cpi->mb.prediction_error) < 15
+ && cpi->mb.prediction_error > minerror
+ && (change > .25 || change2 > .25))
+ {
+ /*(change > 1.4 || change < .75)&& cpi->this_frame_percent_intra > cpi->last_frame_percent_intra + 3*/
+ return 1;
+ }
+
+ return 0;
+
+ }
+
+ /* If the following are true we might as well code a key frame */
+ if (((cpi->this_frame_percent_intra == 100) &&
+ (cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra + 2))) ||
+ ((cpi->this_frame_percent_intra > 95) &&
+ (cpi->this_frame_percent_intra >= (cpi->last_frame_percent_intra + 5))))
+ {
+ code_key_frame = 1;
+ }
+ /* in addition if the following are true and this is not a golden frame
+ * then code a key frame Note that on golden frames there often seems
+ * to be a pop in intra useage anyway hence this restriction is
+ * designed to prevent spurious key frames. The Intra pop needs to be
+ * investigated.
+ */
+ else if (((cpi->this_frame_percent_intra > 60) &&
+ (cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra * 2))) ||
+ ((cpi->this_frame_percent_intra > 75) &&
+ (cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra * 3 / 2))) ||
+ ((cpi->this_frame_percent_intra > 90) &&
+ (cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra + 10))))
+ {
+ if (!cm->refresh_golden_frame)
+ code_key_frame = 1;
+ }
+
+ return code_key_frame;
+
+}
+
+#if !(CONFIG_REALTIME_ONLY)
+static void Pass1Encode(VP8_COMP *cpi, unsigned long *size, unsigned char *dest, unsigned int *frame_flags)
+{
+ (void) size;
+ (void) dest;
+ (void) frame_flags;
+ vp8_set_quantizer(cpi, 26);
+
+ vp8_first_pass(cpi);
+}
+#endif
+
+#if 0
+void write_cx_frame_to_file(YV12_BUFFER_CONFIG *frame, int this_frame)
+{
+
+ /* write the frame */
+ FILE *yframe;
+ int i;
+ char filename[255];
+
+ sprintf(filename, "cx\\y%04d.raw", this_frame);
+ yframe = fopen(filename, "wb");
+
+ for (i = 0; i < frame->y_height; i++)
+ fwrite(frame->y_buffer + i * frame->y_stride, frame->y_width, 1, yframe);
+
+ fclose(yframe);
+ sprintf(filename, "cx\\u%04d.raw", this_frame);
+ yframe = fopen(filename, "wb");
+
+ for (i = 0; i < frame->uv_height; i++)
+ fwrite(frame->u_buffer + i * frame->uv_stride, frame->uv_width, 1, yframe);
+
+ fclose(yframe);
+ sprintf(filename, "cx\\v%04d.raw", this_frame);
+ yframe = fopen(filename, "wb");
+
+ for (i = 0; i < frame->uv_height; i++)
+ fwrite(frame->v_buffer + i * frame->uv_stride, frame->uv_width, 1, yframe);
+
+ fclose(yframe);
+}
+#endif
+/* return of 0 means drop frame */
+
+/* Function to test for conditions that indeicate we should loop
+ * back and recode a frame.
+ */
+static int recode_loop_test( VP8_COMP *cpi,
+ int high_limit, int low_limit,
+ int q, int maxq, int minq )
+{
+ int force_recode = 0;
+ VP8_COMMON *cm = &cpi->common;
+
+ /* Is frame recode allowed at all
+ * Yes if either recode mode 1 is selected or mode two is selcted
+ * and the frame is a key frame. golden frame or alt_ref_frame
+ */
+ if ( (cpi->sf.recode_loop == 1) ||
+ ( (cpi->sf.recode_loop == 2) &&
+ ( (cm->frame_type == KEY_FRAME) ||
+ cm->refresh_golden_frame ||
+ cm->refresh_alt_ref_frame ) ) )
+ {
+ /* General over and under shoot tests */
+ if ( ((cpi->projected_frame_size > high_limit) && (q < maxq)) ||
+ ((cpi->projected_frame_size < low_limit) && (q > minq)) )
+ {
+ force_recode = 1;
+ }
+ /* Special Constrained quality tests */
+ else if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
+ {
+ /* Undershoot and below auto cq level */
+ if ( (q > cpi->cq_target_quality) &&
+ (cpi->projected_frame_size <
+ ((cpi->this_frame_target * 7) >> 3)))
+ {
+ force_recode = 1;
+ }
+ /* Severe undershoot and between auto and user cq level */
+ else if ( (q > cpi->oxcf.cq_level) &&
+ (cpi->projected_frame_size < cpi->min_frame_bandwidth) &&
+ (cpi->active_best_quality > cpi->oxcf.cq_level))
+ {
+ force_recode = 1;
+ cpi->active_best_quality = cpi->oxcf.cq_level;
+ }
+ }
+ }
+
+ return force_recode;
+}
+
+static void update_reference_frames(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+ YV12_BUFFER_CONFIG *yv12_fb = cm->yv12_fb;
+
+ /* At this point the new frame has been encoded.
+ * If any buffer copy / swapping is signaled it should be done here.
+ */
+
+ if (cm->frame_type == KEY_FRAME)
+ {
+ yv12_fb[cm->new_fb_idx].flags |= VP8_GOLD_FRAME | VP8_ALTR_FRAME ;
+
+ yv12_fb[cm->gld_fb_idx].flags &= ~VP8_GOLD_FRAME;
+ yv12_fb[cm->alt_fb_idx].flags &= ~VP8_ALTR_FRAME;
+
+ cm->alt_fb_idx = cm->gld_fb_idx = cm->new_fb_idx;
+
+#if CONFIG_MULTI_RES_ENCODING
+ cpi->current_ref_frames[GOLDEN_FRAME] = cm->current_video_frame;
+ cpi->current_ref_frames[ALTREF_FRAME] = cm->current_video_frame;
+#endif
+ }
+ else /* For non key frames */
+ {
+ if (cm->refresh_alt_ref_frame)
+ {
+ assert(!cm->copy_buffer_to_arf);
+
+ cm->yv12_fb[cm->new_fb_idx].flags |= VP8_ALTR_FRAME;
+ cm->yv12_fb[cm->alt_fb_idx].flags &= ~VP8_ALTR_FRAME;
+ cm->alt_fb_idx = cm->new_fb_idx;
+
+#if CONFIG_MULTI_RES_ENCODING
+ cpi->current_ref_frames[ALTREF_FRAME] = cm->current_video_frame;
+#endif
+ }
+ else if (cm->copy_buffer_to_arf)
+ {
+ assert(!(cm->copy_buffer_to_arf & ~0x3));
+
+ if (cm->copy_buffer_to_arf == 1)
+ {
+ if(cm->alt_fb_idx != cm->lst_fb_idx)
+ {
+ yv12_fb[cm->lst_fb_idx].flags |= VP8_ALTR_FRAME;
+ yv12_fb[cm->alt_fb_idx].flags &= ~VP8_ALTR_FRAME;
+ cm->alt_fb_idx = cm->lst_fb_idx;
+
+#if CONFIG_MULTI_RES_ENCODING
+ cpi->current_ref_frames[ALTREF_FRAME] =
+ cpi->current_ref_frames[LAST_FRAME];
+#endif
+ }
+ }
+ else /* if (cm->copy_buffer_to_arf == 2) */
+ {
+ if(cm->alt_fb_idx != cm->gld_fb_idx)
+ {
+ yv12_fb[cm->gld_fb_idx].flags |= VP8_ALTR_FRAME;
+ yv12_fb[cm->alt_fb_idx].flags &= ~VP8_ALTR_FRAME;
+ cm->alt_fb_idx = cm->gld_fb_idx;
+
+#if CONFIG_MULTI_RES_ENCODING
+ cpi->current_ref_frames[ALTREF_FRAME] =
+ cpi->current_ref_frames[GOLDEN_FRAME];
+#endif
+ }
+ }
+ }
+
+ if (cm->refresh_golden_frame)
+ {
+ assert(!cm->copy_buffer_to_gf);
+
+ cm->yv12_fb[cm->new_fb_idx].flags |= VP8_GOLD_FRAME;
+ cm->yv12_fb[cm->gld_fb_idx].flags &= ~VP8_GOLD_FRAME;
+ cm->gld_fb_idx = cm->new_fb_idx;
+
+#if CONFIG_MULTI_RES_ENCODING
+ cpi->current_ref_frames[GOLDEN_FRAME] = cm->current_video_frame;
+#endif
+ }
+ else if (cm->copy_buffer_to_gf)
+ {
+ assert(!(cm->copy_buffer_to_arf & ~0x3));
+
+ if (cm->copy_buffer_to_gf == 1)
+ {
+ if(cm->gld_fb_idx != cm->lst_fb_idx)
+ {
+ yv12_fb[cm->lst_fb_idx].flags |= VP8_GOLD_FRAME;
+ yv12_fb[cm->gld_fb_idx].flags &= ~VP8_GOLD_FRAME;
+ cm->gld_fb_idx = cm->lst_fb_idx;
+
+#if CONFIG_MULTI_RES_ENCODING
+ cpi->current_ref_frames[GOLDEN_FRAME] =
+ cpi->current_ref_frames[LAST_FRAME];
+#endif
+ }
+ }
+ else /* if (cm->copy_buffer_to_gf == 2) */
+ {
+ if(cm->alt_fb_idx != cm->gld_fb_idx)
+ {
+ yv12_fb[cm->alt_fb_idx].flags |= VP8_GOLD_FRAME;
+ yv12_fb[cm->gld_fb_idx].flags &= ~VP8_GOLD_FRAME;
+ cm->gld_fb_idx = cm->alt_fb_idx;
+
+#if CONFIG_MULTI_RES_ENCODING
+ cpi->current_ref_frames[GOLDEN_FRAME] =
+ cpi->current_ref_frames[ALTREF_FRAME];
+#endif
+ }
+ }
+ }
+ }
+
+ if (cm->refresh_last_frame)
+ {
+ cm->yv12_fb[cm->new_fb_idx].flags |= VP8_LAST_FRAME;
+ cm->yv12_fb[cm->lst_fb_idx].flags &= ~VP8_LAST_FRAME;
+ cm->lst_fb_idx = cm->new_fb_idx;
+
+#if CONFIG_MULTI_RES_ENCODING
+ cpi->current_ref_frames[LAST_FRAME] = cm->current_video_frame;
+#endif
+ }
+}
+
+void vp8_loopfilter_frame(VP8_COMP *cpi, VP8_COMMON *cm)
+{
+ const FRAME_TYPE frame_type = cm->frame_type;
+
+ if (cm->no_lpf)
+ {
+ cm->filter_level = 0;
+ }
+ else
+ {
+ struct vpx_usec_timer timer;
+
+ vp8_clear_system_state();
+
+ vpx_usec_timer_start(&timer);
+ if (cpi->sf.auto_filter == 0)
+ vp8cx_pick_filter_level_fast(cpi->Source, cpi);
+
+ else
+ vp8cx_pick_filter_level(cpi->Source, cpi);
+
+ if (cm->filter_level > 0)
+ {
+ vp8cx_set_alt_lf_level(cpi, cm->filter_level);
+ }
+
+ vpx_usec_timer_mark(&timer);
+ cpi->time_pick_lpf += vpx_usec_timer_elapsed(&timer);
+ }
+
+#if CONFIG_MULTITHREAD
+ if (cpi->b_multi_threaded)
+ sem_post(&cpi->h_event_end_lpf); /* signal that we have set filter_level */
+#endif
+
+ if (cm->filter_level > 0)
+ {
+ vp8_loop_filter_frame(cm, &cpi->mb.e_mbd, frame_type);
+ }
+
+ vp8_yv12_extend_frame_borders(cm->frame_to_show);
+#if CONFIG_TEMPORAL_DENOISING
+ if (cpi->oxcf.noise_sensitivity)
+ {
+
+
+ /* we shouldn't have to keep multiple copies as we know in advance which
+ * buffer we should start - for now to get something up and running
+ * I've chosen to copy the buffers
+ */
+ if (cm->frame_type == KEY_FRAME)
+ {
+ int i;
+ vp8_yv12_copy_frame(
+ cpi->Source,
+ &cpi->denoiser.yv12_running_avg[LAST_FRAME]);
+
+ vp8_yv12_extend_frame_borders(
+ &cpi->denoiser.yv12_running_avg[LAST_FRAME]);
+
+ for (i = 2; i < MAX_REF_FRAMES - 1; i++)
+ vp8_yv12_copy_frame(
+ cpi->Source,
+ &cpi->denoiser.yv12_running_avg[i]);
+ }
+ else /* For non key frames */
+ {
+ vp8_yv12_extend_frame_borders(
+ &cpi->denoiser.yv12_running_avg[LAST_FRAME]);
+
+ if (cm->refresh_alt_ref_frame || cm->copy_buffer_to_arf)
+ {
+ vp8_yv12_copy_frame(
+ &cpi->denoiser.yv12_running_avg[LAST_FRAME],
+ &cpi->denoiser.yv12_running_avg[ALTREF_FRAME]);
+ }
+ if (cm->refresh_golden_frame || cm->copy_buffer_to_gf)
+ {
+ vp8_yv12_copy_frame(
+ &cpi->denoiser.yv12_running_avg[LAST_FRAME],
+ &cpi->denoiser.yv12_running_avg[GOLDEN_FRAME]);
+ }
+ }
+
+ }
+#endif
+
+}
+
+static void encode_frame_to_data_rate
+(
+ VP8_COMP *cpi,
+ unsigned long *size,
+ unsigned char *dest,
+ unsigned char* dest_end,
+ unsigned int *frame_flags
+)
+{
+ int Q;
+ int frame_over_shoot_limit;
+ int frame_under_shoot_limit;
+
+ int Loop = 0;
+ int loop_count;
+
+ VP8_COMMON *cm = &cpi->common;
+ int active_worst_qchanged = 0;
+
+#if !(CONFIG_REALTIME_ONLY)
+ int q_low;
+ int q_high;
+ int zbin_oq_high;
+ int zbin_oq_low = 0;
+ int top_index;
+ int bottom_index;
+ int overshoot_seen = 0;
+ int undershoot_seen = 0;
+#endif
+
+ int drop_mark = (int)(cpi->oxcf.drop_frames_water_mark *
+ cpi->oxcf.optimal_buffer_level / 100);
+ int drop_mark75 = drop_mark * 2 / 3;
+ int drop_mark50 = drop_mark / 4;
+ int drop_mark25 = drop_mark / 8;
+
+
+ /* Clear down mmx registers to allow floating point in what follows */
+ vp8_clear_system_state();
+
+#if CONFIG_MULTITHREAD
+ /* wait for the last picture loopfilter thread done */
+ if (cpi->b_lpf_running)
+ {
+ sem_wait(&cpi->h_event_end_lpf);
+ cpi->b_lpf_running = 0;
+ }
+#endif
+
+ if(cpi->force_next_frame_intra)
+ {
+ cm->frame_type = KEY_FRAME; /* delayed intra frame */
+ cpi->force_next_frame_intra = 0;
+ }
+
+ /* For an alt ref frame in 2 pass we skip the call to the second pass
+ * function that sets the target bandwidth
+ */
+#if !(CONFIG_REALTIME_ONLY)
+
+ if (cpi->pass == 2)
+ {
+ if (cpi->common.refresh_alt_ref_frame)
+ {
+ /* Per frame bit target for the alt ref frame */
+ cpi->per_frame_bandwidth = cpi->twopass.gf_bits;
+ /* per second target bitrate */
+ cpi->target_bandwidth = (int)(cpi->twopass.gf_bits *
+ cpi->output_frame_rate);
+ }
+ }
+ else
+#endif
+ cpi->per_frame_bandwidth = (int)(cpi->target_bandwidth / cpi->output_frame_rate);
+
+ /* Default turn off buffer to buffer copying */
+ cm->copy_buffer_to_gf = 0;
+ cm->copy_buffer_to_arf = 0;
+
+ /* Clear zbin over-quant value and mode boost values. */
+ cpi->zbin_over_quant = 0;
+ cpi->zbin_mode_boost = 0;
+
+ /* Enable or disable mode based tweaking of the zbin
+ * For 2 Pass Only used where GF/ARF prediction quality
+ * is above a threshold
+ */
+ cpi->zbin_mode_boost_enabled = 1;
+ if (cpi->pass == 2)
+ {
+ if ( cpi->gfu_boost <= 400 )
+ {
+ cpi->zbin_mode_boost_enabled = 0;
+ }
+ }
+
+ /* Current default encoder behaviour for the altref sign bias */
+ if (cpi->source_alt_ref_active)
+ cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 1;
+ else
+ cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 0;
+
+ /* Check to see if a key frame is signalled
+ * For two pass with auto key frame enabled cm->frame_type may already
+ * be set, but not for one pass.
+ */
+ if ((cm->current_video_frame == 0) ||
+ (cm->frame_flags & FRAMEFLAGS_KEY) ||
+ (cpi->oxcf.auto_key && (cpi->frames_since_key % cpi->key_frame_frequency == 0)))
+ {
+ /* Key frame from VFW/auto-keyframe/first frame */
+ cm->frame_type = KEY_FRAME;
+ }
+
+#if CONFIG_MULTI_RES_ENCODING
+ /* In multi-resolution encoding, frame_type is decided by lowest-resolution
+ * encoder. Same frame_type is adopted while encoding at other resolution.
+ */
+ if (cpi->oxcf.mr_encoder_id)
+ {
+ LOWER_RES_FRAME_INFO* low_res_frame_info
+ = (LOWER_RES_FRAME_INFO*)cpi->oxcf.mr_low_res_mode_info;
+
+ cm->frame_type = low_res_frame_info->frame_type;
+
+ if(cm->frame_type != KEY_FRAME)
+ {
+ cpi->mr_low_res_mv_avail = 1;
+ cpi->mr_low_res_mv_avail &= !(low_res_frame_info->is_frame_dropped);
+
+ if (cpi->ref_frame_flags & VP8_LAST_FRAME)
+ cpi->mr_low_res_mv_avail &= (cpi->current_ref_frames[LAST_FRAME]
+ == low_res_frame_info->low_res_ref_frames[LAST_FRAME]);
+
+ if (cpi->ref_frame_flags & VP8_GOLD_FRAME)
+ cpi->mr_low_res_mv_avail &= (cpi->current_ref_frames[GOLDEN_FRAME]
+ == low_res_frame_info->low_res_ref_frames[GOLDEN_FRAME]);
+
+ if (cpi->ref_frame_flags & VP8_ALTR_FRAME)
+ cpi->mr_low_res_mv_avail &= (cpi->current_ref_frames[ALTREF_FRAME]
+ == low_res_frame_info->low_res_ref_frames[ALTREF_FRAME]);
+ }
+ }
+#endif
+
+ /* Set various flags etc to special state if it is a key frame */
+ if (cm->frame_type == KEY_FRAME)
+ {
+ int i;
+
+ // Set the loop filter deltas and segmentation map update
+ setup_features(cpi);
+
+ /* The alternate reference frame cannot be active for a key frame */
+ cpi->source_alt_ref_active = 0;
+
+ /* Reset the RD threshold multipliers to default of * 1 (128) */
+ for (i = 0; i < MAX_MODES; i++)
+ {
+ cpi->rd_thresh_mult[i] = 128;
+ }
+ }
+
+#if 0
+ /* Experimental code for lagged compress and one pass
+ * Initialise one_pass GF frames stats
+ * Update stats used for GF selection
+ */
+ {
+ cpi->one_pass_frame_index = cm->current_video_frame % MAX_LAG_BUFFERS;
+
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frames_so_far = 0;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_intra_error = 0.0;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_coded_error = 0.0;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_pcnt_inter = 0.0;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_pcnt_motion = 0.0;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvr = 0.0;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvr_abs = 0.0;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvc = 0.0;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvc_abs = 0.0;
+ }
+#endif
+
+ update_rd_ref_frame_probs(cpi);
+
+ if (cpi->drop_frames_allowed)
+ {
+ /* The reset to decimation 0 is only done here for one pass.
+ * Once it is set two pass leaves decimation on till the next kf.
+ */
+ if ((cpi->buffer_level > drop_mark) && (cpi->decimation_factor > 0))
+ cpi->decimation_factor --;
+
+ if (cpi->buffer_level > drop_mark75 && cpi->decimation_factor > 0)
+ cpi->decimation_factor = 1;
+
+ else if (cpi->buffer_level < drop_mark25 && (cpi->decimation_factor == 2 || cpi->decimation_factor == 3))
+ {
+ cpi->decimation_factor = 3;
+ }
+ else if (cpi->buffer_level < drop_mark50 && (cpi->decimation_factor == 1 || cpi->decimation_factor == 2))
+ {
+ cpi->decimation_factor = 2;
+ }
+ else if (cpi->buffer_level < drop_mark75 && (cpi->decimation_factor == 0 || cpi->decimation_factor == 1))
+ {
+ cpi->decimation_factor = 1;
+ }
+ }
+
+ /* The following decimates the frame rate according to a regular
+ * pattern (i.e. to 1/2 or 2/3 frame rate) This can be used to help
+ * prevent buffer under-run in CBR mode. Alternatively it might be
+ * desirable in some situations to drop frame rate but throw more bits
+ * at each frame.
+ *
+ * Note that dropping a key frame can be problematic if spatial
+ * resampling is also active
+ */
+ if (cpi->decimation_factor > 0)
+ {
+ switch (cpi->decimation_factor)
+ {
+ case 1:
+ cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 3 / 2;
+ break;
+ case 2:
+ cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 5 / 4;
+ break;
+ case 3:
+ cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 5 / 4;
+ break;
+ }
+
+ /* Note that we should not throw out a key frame (especially when
+ * spatial resampling is enabled).
+ */
+ if ((cm->frame_type == KEY_FRAME))
+ {
+ cpi->decimation_count = cpi->decimation_factor;
+ }
+ else if (cpi->decimation_count > 0)
+ {
+ cpi->decimation_count --;
+
+ cpi->bits_off_target += cpi->av_per_frame_bandwidth;
+ if (cpi->bits_off_target > cpi->oxcf.maximum_buffer_size)
+ cpi->bits_off_target = cpi->oxcf.maximum_buffer_size;
+
+#if CONFIG_MULTI_RES_ENCODING
+ vp8_store_drop_frame_info(cpi);
+#endif
+
+ cm->current_video_frame++;
+ cpi->frames_since_key++;
+
+#if CONFIG_INTERNAL_STATS
+ cpi->count ++;
+#endif
+
+ cpi->buffer_level = cpi->bits_off_target;
+
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ unsigned int i;
+
+ /* Propagate bits saved by dropping the frame to higher
+ * layers
+ */
+ for (i=cpi->current_layer+1; i<cpi->oxcf.number_of_layers; i++)
+ {
+ LAYER_CONTEXT *lc = &cpi->layer_context[i];
+ lc->bits_off_target += cpi->av_per_frame_bandwidth;
+ if (lc->bits_off_target > lc->maximum_buffer_size)
+ lc->bits_off_target = lc->maximum_buffer_size;
+ lc->buffer_level = lc->bits_off_target;
+ }
+ }
+
+ return;
+ }
+ else
+ cpi->decimation_count = cpi->decimation_factor;
+ }
+ else
+ cpi->decimation_count = 0;
+
+ /* Decide how big to make the frame */
+ if (!vp8_pick_frame_size(cpi))
+ {
+ /*TODO: 2 drop_frame and return code could be put together. */
+#if CONFIG_MULTI_RES_ENCODING
+ vp8_store_drop_frame_info(cpi);
+#endif
+ cm->current_video_frame++;
+ cpi->frames_since_key++;
+ return;
+ }
+
+ /* Reduce active_worst_allowed_q for CBR if our buffer is getting too full.
+ * This has a knock on effect on active best quality as well.
+ * For CBR if the buffer reaches its maximum level then we can no longer
+ * save up bits for later frames so we might as well use them up
+ * on the current frame.
+ */
+ if ((cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
+ (cpi->buffer_level >= cpi->oxcf.optimal_buffer_level) && cpi->buffered_mode)
+ {
+ /* Max adjustment is 1/4 */
+ int Adjustment = cpi->active_worst_quality / 4;
+
+ if (Adjustment)
+ {
+ int buff_lvl_step;
+
+ if (cpi->buffer_level < cpi->oxcf.maximum_buffer_size)
+ {
+ buff_lvl_step = (int)
+ ((cpi->oxcf.maximum_buffer_size -
+ cpi->oxcf.optimal_buffer_level) /
+ Adjustment);
+
+ if (buff_lvl_step)
+ Adjustment = (int)
+ ((cpi->buffer_level -
+ cpi->oxcf.optimal_buffer_level) /
+ buff_lvl_step);
+ else
+ Adjustment = 0;
+ }
+
+ cpi->active_worst_quality -= Adjustment;
+
+ if(cpi->active_worst_quality < cpi->active_best_quality)
+ cpi->active_worst_quality = cpi->active_best_quality;
+ }
+ }
+
+ /* Set an active best quality and if necessary active worst quality
+ * There is some odd behavior for one pass here that needs attention.
+ */
+ if ( (cpi->pass == 2) || (cpi->ni_frames > 150))
+ {
+ vp8_clear_system_state();
+
+ Q = cpi->active_worst_quality;
+
+ if ( cm->frame_type == KEY_FRAME )
+ {
+ if ( cpi->pass == 2 )
+ {
+ if (cpi->gfu_boost > 600)
+ cpi->active_best_quality = kf_low_motion_minq[Q];
+ else
+ cpi->active_best_quality = kf_high_motion_minq[Q];
+
+ /* Special case for key frames forced because we have reached
+ * the maximum key frame interval. Here force the Q to a range
+ * based on the ambient Q to reduce the risk of popping
+ */
+ if ( cpi->this_key_frame_forced )
+ {
+ if ( cpi->active_best_quality > cpi->avg_frame_qindex * 7/8)
+ cpi->active_best_quality = cpi->avg_frame_qindex * 7/8;
+ else if ( cpi->active_best_quality < cpi->avg_frame_qindex >> 2 )
+ cpi->active_best_quality = cpi->avg_frame_qindex >> 2;
+ }
+ }
+ /* One pass more conservative */
+ else
+ cpi->active_best_quality = kf_high_motion_minq[Q];
+ }
+
+ else if (cpi->oxcf.number_of_layers==1 &&
+ (cm->refresh_golden_frame || cpi->common.refresh_alt_ref_frame))
+ {
+ /* Use the lower of cpi->active_worst_quality and recent
+ * average Q as basis for GF/ARF Q limit unless last frame was
+ * a key frame.
+ */
+ if ( (cpi->frames_since_key > 1) &&
+ (cpi->avg_frame_qindex < cpi->active_worst_quality) )
+ {
+ Q = cpi->avg_frame_qindex;
+ }
+
+ /* For constrained quality dont allow Q less than the cq level */
+ if ( (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
+ (Q < cpi->cq_target_quality) )
+ {
+ Q = cpi->cq_target_quality;
+ }
+
+ if ( cpi->pass == 2 )
+ {
+ if ( cpi->gfu_boost > 1000 )
+ cpi->active_best_quality = gf_low_motion_minq[Q];
+ else if ( cpi->gfu_boost < 400 )
+ cpi->active_best_quality = gf_high_motion_minq[Q];
+ else
+ cpi->active_best_quality = gf_mid_motion_minq[Q];
+
+ /* Constrained quality use slightly lower active best. */
+ if ( cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY )
+ {
+ cpi->active_best_quality =
+ cpi->active_best_quality * 15/16;
+ }
+ }
+ /* One pass more conservative */
+ else
+ cpi->active_best_quality = gf_high_motion_minq[Q];
+ }
+ else
+ {
+ cpi->active_best_quality = inter_minq[Q];
+
+ /* For the constant/constrained quality mode we dont want
+ * q to fall below the cq level.
+ */
+ if ((cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
+ (cpi->active_best_quality < cpi->cq_target_quality) )
+ {
+ /* If we are strongly undershooting the target rate in the last
+ * frames then use the user passed in cq value not the auto
+ * cq value.
+ */
+ if ( cpi->rolling_actual_bits < cpi->min_frame_bandwidth )
+ cpi->active_best_quality = cpi->oxcf.cq_level;
+ else
+ cpi->active_best_quality = cpi->cq_target_quality;
+ }
+ }
+
+ /* If CBR and the buffer is as full then it is reasonable to allow
+ * higher quality on the frames to prevent bits just going to waste.
+ */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ /* Note that the use of >= here elliminates the risk of a devide
+ * by 0 error in the else if clause
+ */
+ if (cpi->buffer_level >= cpi->oxcf.maximum_buffer_size)
+ cpi->active_best_quality = cpi->best_quality;
+
+ else if (cpi->buffer_level > cpi->oxcf.optimal_buffer_level)
+ {
+ int Fraction = (int)
+ (((cpi->buffer_level - cpi->oxcf.optimal_buffer_level) * 128)
+ / (cpi->oxcf.maximum_buffer_size -
+ cpi->oxcf.optimal_buffer_level));
+ int min_qadjustment = ((cpi->active_best_quality -
+ cpi->best_quality) * Fraction) / 128;
+
+ cpi->active_best_quality -= min_qadjustment;
+ }
+ }
+ }
+ /* Make sure constrained quality mode limits are adhered to for the first
+ * few frames of one pass encodes
+ */
+ else if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
+ {
+ if ( (cm->frame_type == KEY_FRAME) ||
+ cm->refresh_golden_frame || cpi->common.refresh_alt_ref_frame )
+ {
+ cpi->active_best_quality = cpi->best_quality;
+ }
+ else if (cpi->active_best_quality < cpi->cq_target_quality)
+ {
+ cpi->active_best_quality = cpi->cq_target_quality;
+ }
+ }
+
+ /* Clip the active best and worst quality values to limits */
+ if (cpi->active_worst_quality > cpi->worst_quality)
+ cpi->active_worst_quality = cpi->worst_quality;
+
+ if (cpi->active_best_quality < cpi->best_quality)
+ cpi->active_best_quality = cpi->best_quality;
+
+ if ( cpi->active_worst_quality < cpi->active_best_quality )
+ cpi->active_worst_quality = cpi->active_best_quality;
+
+ /* Determine initial Q to try */
+ Q = vp8_regulate_q(cpi, cpi->this_frame_target);
+
+#if !(CONFIG_REALTIME_ONLY)
+
+ /* Set highest allowed value for Zbin over quant */
+ if (cm->frame_type == KEY_FRAME)
+ zbin_oq_high = 0;
+ else if ((cpi->oxcf.number_of_layers == 1) && ((cm->refresh_alt_ref_frame ||
+ (cm->refresh_golden_frame && !cpi->source_alt_ref_active))))
+ {
+ zbin_oq_high = 16;
+ }
+ else
+ zbin_oq_high = ZBIN_OQ_MAX;
+#endif
+
+ /* Setup background Q adjustment for error resilient mode.
+ * For multi-layer encodes only enable this for the base layer.
+ */
+ if (cpi->cyclic_refresh_mode_enabled)
+ {
+ if (cpi->current_layer==0)
+ cyclic_background_refresh(cpi, Q, 0);
+ else
+ disable_segmentation(cpi);
+ }
+
+ vp8_compute_frame_size_bounds(cpi, &frame_under_shoot_limit, &frame_over_shoot_limit);
+
+#if !(CONFIG_REALTIME_ONLY)
+ /* Limit Q range for the adaptive loop. */
+ bottom_index = cpi->active_best_quality;
+ top_index = cpi->active_worst_quality;
+ q_low = cpi->active_best_quality;
+ q_high = cpi->active_worst_quality;
+#endif
+
+ vp8_save_coding_context(cpi);
+
+ loop_count = 0;
+
+ scale_and_extend_source(cpi->un_scaled_source, cpi);
+
+#if !(CONFIG_REALTIME_ONLY) && CONFIG_POSTPROC && !(CONFIG_TEMPORAL_DENOISING)
+
+ if (cpi->oxcf.noise_sensitivity > 0)
+ {
+ unsigned char *src;
+ int l = 0;
+
+ switch (cpi->oxcf.noise_sensitivity)
+ {
+ case 1:
+ l = 20;
+ break;
+ case 2:
+ l = 40;
+ break;
+ case 3:
+ l = 60;
+ break;
+ case 4:
+ l = 80;
+ break;
+ case 5:
+ l = 100;
+ break;
+ case 6:
+ l = 150;
+ break;
+ }
+
+
+ if (cm->frame_type == KEY_FRAME)
+ {
+ vp8_de_noise(cm, cpi->Source, cpi->Source, l , 1, 0);
+ }
+ else
+ {
+ vp8_de_noise(cm, cpi->Source, cpi->Source, l , 1, 0);
+
+ src = cpi->Source->y_buffer;
+
+ if (cpi->Source->y_stride < 0)
+ {
+ src += cpi->Source->y_stride * (cpi->Source->y_height - 1);
+ }
+ }
+ }
+
+#endif
+
+#ifdef OUTPUT_YUV_SRC
+ vp8_write_yuv_frame(cpi->Source);
+#endif
+
+ do
+ {
+ vp8_clear_system_state();
+
+ vp8_set_quantizer(cpi, Q);
+
+ /* setup skip prob for costing in mode/mv decision */
+ if (cpi->common.mb_no_coeff_skip)
+ {
+ cpi->prob_skip_false = cpi->base_skip_false_prob[Q];
+
+ if (cm->frame_type != KEY_FRAME)
+ {
+ if (cpi->common.refresh_alt_ref_frame)
+ {
+ if (cpi->last_skip_false_probs[2] != 0)
+ cpi->prob_skip_false = cpi->last_skip_false_probs[2];
+
+ /*
+ if(cpi->last_skip_false_probs[2]!=0 && abs(Q- cpi->last_skip_probs_q[2])<=16 )
+ cpi->prob_skip_false = cpi->last_skip_false_probs[2];
+ else if (cpi->last_skip_false_probs[2]!=0)
+ cpi->prob_skip_false = (cpi->last_skip_false_probs[2] + cpi->prob_skip_false ) / 2;
+ */
+ }
+ else if (cpi->common.refresh_golden_frame)
+ {
+ if (cpi->last_skip_false_probs[1] != 0)
+ cpi->prob_skip_false = cpi->last_skip_false_probs[1];
+
+ /*
+ if(cpi->last_skip_false_probs[1]!=0 && abs(Q- cpi->last_skip_probs_q[1])<=16 )
+ cpi->prob_skip_false = cpi->last_skip_false_probs[1];
+ else if (cpi->last_skip_false_probs[1]!=0)
+ cpi->prob_skip_false = (cpi->last_skip_false_probs[1] + cpi->prob_skip_false ) / 2;
+ */
+ }
+ else
+ {
+ if (cpi->last_skip_false_probs[0] != 0)
+ cpi->prob_skip_false = cpi->last_skip_false_probs[0];
+
+ /*
+ if(cpi->last_skip_false_probs[0]!=0 && abs(Q- cpi->last_skip_probs_q[0])<=16 )
+ cpi->prob_skip_false = cpi->last_skip_false_probs[0];
+ else if(cpi->last_skip_false_probs[0]!=0)
+ cpi->prob_skip_false = (cpi->last_skip_false_probs[0] + cpi->prob_skip_false ) / 2;
+ */
+ }
+
+ /* as this is for cost estimate, let's make sure it does not
+ * go extreme eitehr way
+ */
+ if (cpi->prob_skip_false < 5)
+ cpi->prob_skip_false = 5;
+
+ if (cpi->prob_skip_false > 250)
+ cpi->prob_skip_false = 250;
+
+ if (cpi->oxcf.number_of_layers == 1 && cpi->is_src_frame_alt_ref)
+ cpi->prob_skip_false = 1;
+ }
+
+#if 0
+
+ if (cpi->pass != 1)
+ {
+ FILE *f = fopen("skip.stt", "a");
+ fprintf(f, "%d, %d, %4d ", cpi->common.refresh_golden_frame, cpi->common.refresh_alt_ref_frame, cpi->prob_skip_false);
+ fclose(f);
+ }
+
+#endif
+
+ }
+
+ if (cm->frame_type == KEY_FRAME)
+ {
+ if(resize_key_frame(cpi))
+ {
+ /* If the frame size has changed, need to reset Q, quantizer,
+ * and background refresh.
+ */
+ Q = vp8_regulate_q(cpi, cpi->this_frame_target);
+ if (cpi->cyclic_refresh_mode_enabled)
+ {
+ if (cpi->current_layer==0)
+ cyclic_background_refresh(cpi, Q, 0);
+ else
+ disable_segmentation(cpi);
+ }
+ vp8_set_quantizer(cpi, Q);
+ }
+
+ vp8_setup_key_frame(cpi);
+ }
+
+
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ {
+ if(cpi->oxcf.error_resilient_mode)
+ cm->refresh_entropy_probs = 0;
+
+ if (cpi->oxcf.error_resilient_mode & VPX_ERROR_RESILIENT_PARTITIONS)
+ {
+ if (cm->frame_type == KEY_FRAME)
+ cm->refresh_entropy_probs = 1;
+ }
+
+ if (cm->refresh_entropy_probs == 0)
+ {
+ /* save a copy for later refresh */
+ vpx_memcpy(&cm->lfc, &cm->fc, sizeof(cm->fc));
+ }
+
+ vp8_update_coef_context(cpi);
+
+ vp8_update_coef_probs(cpi);
+
+ /* transform / motion compensation build reconstruction frame
+ * +pack coef partitions
+ */
+ vp8_encode_frame(cpi);
+
+ /* cpi->projected_frame_size is not needed for RT mode */
+ }
+#else
+ /* transform / motion compensation build reconstruction frame */
+ vp8_encode_frame(cpi);
+
+ cpi->projected_frame_size -= vp8_estimate_entropy_savings(cpi);
+ cpi->projected_frame_size = (cpi->projected_frame_size > 0) ? cpi->projected_frame_size : 0;
+#endif
+ vp8_clear_system_state();
+
+ /* Test to see if the stats generated for this frame indicate that
+ * we should have coded a key frame (assuming that we didn't)!
+ */
+
+ if (cpi->pass != 2 && cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME
+ && cpi->compressor_speed != 2)
+ {
+#if !(CONFIG_REALTIME_ONLY)
+ if (decide_key_frame(cpi))
+ {
+ /* Reset all our sizing numbers and recode */
+ cm->frame_type = KEY_FRAME;
+
+ vp8_pick_frame_size(cpi);
+
+ /* Clear the Alt reference frame active flag when we have
+ * a key frame
+ */
+ cpi->source_alt_ref_active = 0;
+
+ // Set the loop filter deltas and segmentation map update
+ setup_features(cpi);
+
+ vp8_restore_coding_context(cpi);
+
+ Q = vp8_regulate_q(cpi, cpi->this_frame_target);
+
+ vp8_compute_frame_size_bounds(cpi, &frame_under_shoot_limit, &frame_over_shoot_limit);
+
+ /* Limit Q range for the adaptive loop. */
+ bottom_index = cpi->active_best_quality;
+ top_index = cpi->active_worst_quality;
+ q_low = cpi->active_best_quality;
+ q_high = cpi->active_worst_quality;
+
+ loop_count++;
+ Loop = 1;
+
+ continue;
+ }
+#endif
+ }
+
+ vp8_clear_system_state();
+
+ if (frame_over_shoot_limit == 0)
+ frame_over_shoot_limit = 1;
+
+ /* Are we are overshooting and up against the limit of active max Q. */
+ if (((cpi->pass != 2) || (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)) &&
+ (Q == cpi->active_worst_quality) &&
+ (cpi->active_worst_quality < cpi->worst_quality) &&
+ (cpi->projected_frame_size > frame_over_shoot_limit))
+ {
+ int over_size_percent = ((cpi->projected_frame_size - frame_over_shoot_limit) * 100) / frame_over_shoot_limit;
+
+ /* If so is there any scope for relaxing it */
+ while ((cpi->active_worst_quality < cpi->worst_quality) && (over_size_percent > 0))
+ {
+ cpi->active_worst_quality++;
+ /* Assume 1 qstep = about 4% on frame size. */
+ over_size_percent = (int)(over_size_percent * 0.96);
+ }
+#if !(CONFIG_REALTIME_ONLY)
+ top_index = cpi->active_worst_quality;
+#endif
+ /* If we have updated the active max Q do not call
+ * vp8_update_rate_correction_factors() this loop.
+ */
+ active_worst_qchanged = 1;
+ }
+ else
+ active_worst_qchanged = 0;
+
+#if !(CONFIG_REALTIME_ONLY)
+ /* Special case handling for forced key frames */
+ if ( (cm->frame_type == KEY_FRAME) && cpi->this_key_frame_forced )
+ {
+ int last_q = Q;
+ int kf_err = vp8_calc_ss_err(cpi->Source,
+ &cm->yv12_fb[cm->new_fb_idx]);
+
+ /* The key frame is not good enough */
+ if ( kf_err > ((cpi->ambient_err * 7) >> 3) )
+ {
+ /* Lower q_high */
+ q_high = (Q > q_low) ? (Q - 1) : q_low;
+
+ /* Adjust Q */
+ Q = (q_high + q_low) >> 1;
+ }
+ /* The key frame is much better than the previous frame */
+ else if ( kf_err < (cpi->ambient_err >> 1) )
+ {
+ /* Raise q_low */
+ q_low = (Q < q_high) ? (Q + 1) : q_high;
+
+ /* Adjust Q */
+ Q = (q_high + q_low + 1) >> 1;
+ }
+
+ /* Clamp Q to upper and lower limits: */
+ if (Q > q_high)
+ Q = q_high;
+ else if (Q < q_low)
+ Q = q_low;
+
+ Loop = Q != last_q;
+ }
+
+ /* Is the projected frame size out of range and are we allowed
+ * to attempt to recode.
+ */
+ else if ( recode_loop_test( cpi,
+ frame_over_shoot_limit, frame_under_shoot_limit,
+ Q, top_index, bottom_index ) )
+ {
+ int last_q = Q;
+ int Retries = 0;
+
+ /* Frame size out of permitted range. Update correction factor
+ * & compute new Q to try...
+ */
+
+ /* Frame is too large */
+ if (cpi->projected_frame_size > cpi->this_frame_target)
+ {
+ /* Raise Qlow as to at least the current value */
+ q_low = (Q < q_high) ? (Q + 1) : q_high;
+
+ /* If we are using over quant do the same for zbin_oq_low */
+ if (cpi->zbin_over_quant > 0)
+ zbin_oq_low = (cpi->zbin_over_quant < zbin_oq_high) ? (cpi->zbin_over_quant + 1) : zbin_oq_high;
+
+ if (undershoot_seen)
+ {
+ /* Update rate_correction_factor unless
+ * cpi->active_worst_quality has changed.
+ */
+ if (!active_worst_qchanged)
+ vp8_update_rate_correction_factors(cpi, 1);
+
+ Q = (q_high + q_low + 1) / 2;
+
+ /* Adjust cpi->zbin_over_quant (only allowed when Q
+ * is max)
+ */
+ if (Q < MAXQ)
+ cpi->zbin_over_quant = 0;
+ else
+ {
+ zbin_oq_low = (cpi->zbin_over_quant < zbin_oq_high) ? (cpi->zbin_over_quant + 1) : zbin_oq_high;
+ cpi->zbin_over_quant = (zbin_oq_high + zbin_oq_low) / 2;
+ }
+ }
+ else
+ {
+ /* Update rate_correction_factor unless
+ * cpi->active_worst_quality has changed.
+ */
+ if (!active_worst_qchanged)
+ vp8_update_rate_correction_factors(cpi, 0);
+
+ Q = vp8_regulate_q(cpi, cpi->this_frame_target);
+
+ while (((Q < q_low) || (cpi->zbin_over_quant < zbin_oq_low)) && (Retries < 10))
+ {
+ vp8_update_rate_correction_factors(cpi, 0);
+ Q = vp8_regulate_q(cpi, cpi->this_frame_target);
+ Retries ++;
+ }
+ }
+
+ overshoot_seen = 1;
+ }
+ /* Frame is too small */
+ else
+ {
+ if (cpi->zbin_over_quant == 0)
+ /* Lower q_high if not using over quant */
+ q_high = (Q > q_low) ? (Q - 1) : q_low;
+ else
+ /* else lower zbin_oq_high */
+ zbin_oq_high = (cpi->zbin_over_quant > zbin_oq_low) ? (cpi->zbin_over_quant - 1) : zbin_oq_low;
+
+ if (overshoot_seen)
+ {
+ /* Update rate_correction_factor unless
+ * cpi->active_worst_quality has changed.
+ */
+ if (!active_worst_qchanged)
+ vp8_update_rate_correction_factors(cpi, 1);
+
+ Q = (q_high + q_low) / 2;
+
+ /* Adjust cpi->zbin_over_quant (only allowed when Q
+ * is max)
+ */
+ if (Q < MAXQ)
+ cpi->zbin_over_quant = 0;
+ else
+ cpi->zbin_over_quant = (zbin_oq_high + zbin_oq_low) / 2;
+ }
+ else
+ {
+ /* Update rate_correction_factor unless
+ * cpi->active_worst_quality has changed.
+ */
+ if (!active_worst_qchanged)
+ vp8_update_rate_correction_factors(cpi, 0);
+
+ Q = vp8_regulate_q(cpi, cpi->this_frame_target);
+
+ /* Special case reset for qlow for constrained quality.
+ * This should only trigger where there is very substantial
+ * undershoot on a frame and the auto cq level is above
+ * the user passsed in value.
+ */
+ if ( (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
+ (Q < q_low) )
+ {
+ q_low = Q;
+ }
+
+ while (((Q > q_high) || (cpi->zbin_over_quant > zbin_oq_high)) && (Retries < 10))
+ {
+ vp8_update_rate_correction_factors(cpi, 0);
+ Q = vp8_regulate_q(cpi, cpi->this_frame_target);
+ Retries ++;
+ }
+ }
+
+ undershoot_seen = 1;
+ }
+
+ /* Clamp Q to upper and lower limits: */
+ if (Q > q_high)
+ Q = q_high;
+ else if (Q < q_low)
+ Q = q_low;
+
+ /* Clamp cpi->zbin_over_quant */
+ cpi->zbin_over_quant = (cpi->zbin_over_quant < zbin_oq_low) ? zbin_oq_low : (cpi->zbin_over_quant > zbin_oq_high) ? zbin_oq_high : cpi->zbin_over_quant;
+
+ Loop = Q != last_q;
+ }
+ else
+#endif
+ Loop = 0;
+
+ if (cpi->is_src_frame_alt_ref)
+ Loop = 0;
+
+ if (Loop == 1)
+ {
+ vp8_restore_coding_context(cpi);
+ loop_count++;
+#if CONFIG_INTERNAL_STATS
+ cpi->tot_recode_hits++;
+#endif
+ }
+ }
+ while (Loop == 1);
+
+#if 0
+ /* Experimental code for lagged and one pass
+ * Update stats used for one pass GF selection
+ */
+ {
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_coded_error = (double)cpi->prediction_error;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_intra_error = (double)cpi->intra_error;
+ cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_pcnt_inter = (double)(100 - cpi->this_frame_percent_intra) / 100.0;
+ }
+#endif
+
+ /* Special case code to reduce pulsing when key frames are forced at a
+ * fixed interval. Note the reconstruction error if it is the frame before
+ * the force key frame
+ */
+ if ( cpi->next_key_frame_forced && (cpi->twopass.frames_to_key == 0) )
+ {
+ cpi->ambient_err = vp8_calc_ss_err(cpi->Source,
+ &cm->yv12_fb[cm->new_fb_idx]);
+ }
+
+ /* This frame's MVs are saved and will be used in next frame's MV predictor.
+ * Last frame has one more line(add to bottom) and one more column(add to
+ * right) than cm->mip. The edge elements are initialized to 0.
+ */
+#if CONFIG_MULTI_RES_ENCODING
+ if(!cpi->oxcf.mr_encoder_id && cm->show_frame)
+#else
+ if(cm->show_frame) /* do not save for altref frame */
+#endif
+ {
+ int mb_row;
+ int mb_col;
+ /* Point to beginning of allocated MODE_INFO arrays. */
+ MODE_INFO *tmp = cm->mip;
+
+ if(cm->frame_type != KEY_FRAME)
+ {
+ for (mb_row = 0; mb_row < cm->mb_rows+1; mb_row ++)
+ {
+ for (mb_col = 0; mb_col < cm->mb_cols+1; mb_col ++)
+ {
+ if(tmp->mbmi.ref_frame != INTRA_FRAME)
+ cpi->lfmv[mb_col + mb_row*(cm->mode_info_stride+1)].as_int = tmp->mbmi.mv.as_int;
+
+ cpi->lf_ref_frame_sign_bias[mb_col + mb_row*(cm->mode_info_stride+1)] = cm->ref_frame_sign_bias[tmp->mbmi.ref_frame];
+ cpi->lf_ref_frame[mb_col + mb_row*(cm->mode_info_stride+1)] = tmp->mbmi.ref_frame;
+ tmp++;
+ }
+ }
+ }
+ }
+
+ /* Count last ref frame 0,0 usage on current encoded frame. */
+ {
+ int mb_row;
+ int mb_col;
+ /* Point to beginning of MODE_INFO arrays. */
+ MODE_INFO *tmp = cm->mi;
+
+ cpi->inter_zz_count = 0;
+ cpi->zeromv_count = 0;
+
+ if(cm->frame_type != KEY_FRAME)
+ {
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row ++)
+ {
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col ++)
+ {
+ if(tmp->mbmi.mode == ZEROMV && tmp->mbmi.ref_frame == LAST_FRAME)
+ cpi->inter_zz_count++;
+ if(tmp->mbmi.mode == ZEROMV)
+ cpi->zeromv_count++;
+ tmp++;
+ }
+ tmp++;
+ }
+ }
+ }
+
+#if CONFIG_MULTI_RES_ENCODING
+ vp8_cal_dissimilarity(cpi);
+#endif
+
+ /* Update the GF useage maps.
+ * This is done after completing the compression of a frame when all
+ * modes etc. are finalized but before loop filter
+ */
+ if (cpi->oxcf.number_of_layers == 1)
+ vp8_update_gf_useage_maps(cpi, cm, &cpi->mb);
+
+ if (cm->frame_type == KEY_FRAME)
+ cm->refresh_last_frame = 1;
+
+#if 0
+ {
+ FILE *f = fopen("gfactive.stt", "a");
+ fprintf(f, "%8d %8d %8d %8d %8d\n", cm->current_video_frame, (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols), cpi->this_iiratio, cpi->next_iiratio, cm->refresh_golden_frame);
+ fclose(f);
+ }
+#endif
+
+ /* For inter frames the current default behavior is that when
+ * cm->refresh_golden_frame is set we copy the old GF over to the ARF buffer
+ * This is purely an encoder decision at present.
+ */
+ if (!cpi->oxcf.error_resilient_mode && cm->refresh_golden_frame)
+ cm->copy_buffer_to_arf = 2;
+ else
+ cm->copy_buffer_to_arf = 0;
+
+ cm->frame_to_show = &cm->yv12_fb[cm->new_fb_idx];
+
+#if CONFIG_MULTITHREAD
+ if (cpi->b_multi_threaded)
+ {
+ /* start loopfilter in separate thread */
+ sem_post(&cpi->h_event_start_lpf);
+ cpi->b_lpf_running = 1;
+ }
+ else
+#endif
+ {
+ vp8_loopfilter_frame(cpi, cm);
+ }
+
+ update_reference_frames(cpi);
+
+#if !(CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ if (cpi->oxcf.error_resilient_mode)
+ {
+ cm->refresh_entropy_probs = 0;
+ }
+#endif
+
+#if CONFIG_MULTITHREAD
+ /* wait that filter_level is picked so that we can continue with stream packing */
+ if (cpi->b_multi_threaded)
+ sem_wait(&cpi->h_event_end_lpf);
+#endif
+
+ /* build the bitstream */
+ vp8_pack_bitstream(cpi, dest, dest_end, size);
+
+#if CONFIG_MULTITHREAD
+ /* if PSNR packets are generated we have to wait for the lpf */
+ if (cpi->b_lpf_running && cpi->b_calculate_psnr)
+ {
+ sem_wait(&cpi->h_event_end_lpf);
+ cpi->b_lpf_running = 0;
+ }
+#endif
+
+ /* Move storing frame_type out of the above loop since it is also
+ * needed in motion search besides loopfilter */
+ cm->last_frame_type = cm->frame_type;
+
+ /* Update rate control heuristics */
+ cpi->total_byte_count += (*size);
+ cpi->projected_frame_size = (*size) << 3;
+
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ unsigned int i;
+ for (i=cpi->current_layer+1; i<cpi->oxcf.number_of_layers; i++)
+ cpi->layer_context[i].total_byte_count += (*size);
+ }
+
+ if (!active_worst_qchanged)
+ vp8_update_rate_correction_factors(cpi, 2);
+
+ cpi->last_q[cm->frame_type] = cm->base_qindex;
+
+ if (cm->frame_type == KEY_FRAME)
+ {
+ vp8_adjust_key_frame_context(cpi);
+ }
+
+ /* Keep a record of ambient average Q. */
+ if (cm->frame_type != KEY_FRAME)
+ cpi->avg_frame_qindex = (2 + 3 * cpi->avg_frame_qindex + cm->base_qindex) >> 2;
+
+ /* Keep a record from which we can calculate the average Q excluding
+ * GF updates and key frames
+ */
+ if ((cm->frame_type != KEY_FRAME) && ((cpi->oxcf.number_of_layers > 1) ||
+ (!cm->refresh_golden_frame && !cm->refresh_alt_ref_frame)))
+ {
+ cpi->ni_frames++;
+
+ /* Calculate the average Q for normal inter frames (not key or GFU
+ * frames).
+ */
+ if ( cpi->pass == 2 )
+ {
+ cpi->ni_tot_qi += Q;
+ cpi->ni_av_qi = (cpi->ni_tot_qi / cpi->ni_frames);
+ }
+ else
+ {
+ /* Damp value for first few frames */
+ if (cpi->ni_frames > 150 )
+ {
+ cpi->ni_tot_qi += Q;
+ cpi->ni_av_qi = (cpi->ni_tot_qi / cpi->ni_frames);
+ }
+ /* For one pass, early in the clip ... average the current frame Q
+ * value with the worstq entered by the user as a dampening measure
+ */
+ else
+ {
+ cpi->ni_tot_qi += Q;
+ cpi->ni_av_qi = ((cpi->ni_tot_qi / cpi->ni_frames) + cpi->worst_quality + 1) / 2;
+ }
+
+ /* If the average Q is higher than what was used in the last
+ * frame (after going through the recode loop to keep the frame
+ * size within range) then use the last frame value - 1. The -1
+ * is designed to stop Q and hence the data rate, from
+ * progressively falling away during difficult sections, but at
+ * the same time reduce the number of itterations around the
+ * recode loop.
+ */
+ if (Q > cpi->ni_av_qi)
+ cpi->ni_av_qi = Q - 1;
+ }
+ }
+
+ /* Update the buffer level variable. */
+ /* Non-viewable frames are a special case and are treated as pure overhead. */
+ if ( !cm->show_frame )
+ cpi->bits_off_target -= cpi->projected_frame_size;
+ else
+ cpi->bits_off_target += cpi->av_per_frame_bandwidth - cpi->projected_frame_size;
+
+ /* Clip the buffer level to the maximum specified buffer size */
+ if (cpi->bits_off_target > cpi->oxcf.maximum_buffer_size)
+ cpi->bits_off_target = cpi->oxcf.maximum_buffer_size;
+
+ /* Rolling monitors of whether we are over or underspending used to
+ * help regulate min and Max Q in two pass.
+ */
+ cpi->rolling_target_bits = ((cpi->rolling_target_bits * 3) + cpi->this_frame_target + 2) / 4;
+ cpi->rolling_actual_bits = ((cpi->rolling_actual_bits * 3) + cpi->projected_frame_size + 2) / 4;
+ cpi->long_rolling_target_bits = ((cpi->long_rolling_target_bits * 31) + cpi->this_frame_target + 16) / 32;
+ cpi->long_rolling_actual_bits = ((cpi->long_rolling_actual_bits * 31) + cpi->projected_frame_size + 16) / 32;
+
+ /* Actual bits spent */
+ cpi->total_actual_bits += cpi->projected_frame_size;
+
+ /* Debug stats */
+ cpi->total_target_vs_actual += (cpi->this_frame_target - cpi->projected_frame_size);
+
+ cpi->buffer_level = cpi->bits_off_target;
+
+ /* Propagate values to higher temporal layers */
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ unsigned int i;
+
+ for (i=cpi->current_layer+1; i<cpi->oxcf.number_of_layers; i++)
+ {
+ LAYER_CONTEXT *lc = &cpi->layer_context[i];
+ int bits_off_for_this_layer =
+ (int)(lc->target_bandwidth / lc->frame_rate -
+ cpi->projected_frame_size);
+
+ lc->bits_off_target += bits_off_for_this_layer;
+
+ /* Clip buffer level to maximum buffer size for the layer */
+ if (lc->bits_off_target > lc->maximum_buffer_size)
+ lc->bits_off_target = lc->maximum_buffer_size;
+
+ lc->total_actual_bits += cpi->projected_frame_size;
+ lc->total_target_vs_actual += bits_off_for_this_layer;
+ lc->buffer_level = lc->bits_off_target;
+ }
+ }
+
+ /* Update bits left to the kf and gf groups to account for overshoot
+ * or undershoot on these frames
+ */
+ if (cm->frame_type == KEY_FRAME)
+ {
+ cpi->twopass.kf_group_bits += cpi->this_frame_target - cpi->projected_frame_size;
+
+ if (cpi->twopass.kf_group_bits < 0)
+ cpi->twopass.kf_group_bits = 0 ;
+ }
+ else if (cm->refresh_golden_frame || cm->refresh_alt_ref_frame)
+ {
+ cpi->twopass.gf_group_bits += cpi->this_frame_target - cpi->projected_frame_size;
+
+ if (cpi->twopass.gf_group_bits < 0)
+ cpi->twopass.gf_group_bits = 0 ;
+ }
+
+ if (cm->frame_type != KEY_FRAME)
+ {
+ if (cpi->common.refresh_alt_ref_frame)
+ {
+ cpi->last_skip_false_probs[2] = cpi->prob_skip_false;
+ cpi->last_skip_probs_q[2] = cm->base_qindex;
+ }
+ else if (cpi->common.refresh_golden_frame)
+ {
+ cpi->last_skip_false_probs[1] = cpi->prob_skip_false;
+ cpi->last_skip_probs_q[1] = cm->base_qindex;
+ }
+ else
+ {
+ cpi->last_skip_false_probs[0] = cpi->prob_skip_false;
+ cpi->last_skip_probs_q[0] = cm->base_qindex;
+
+ /* update the baseline */
+ cpi->base_skip_false_prob[cm->base_qindex] = cpi->prob_skip_false;
+
+ }
+ }
+
+#if 0 && CONFIG_INTERNAL_STATS
+ {
+ FILE *f = fopen("tmp.stt", "a");
+
+ vp8_clear_system_state();
+
+ if (cpi->twopass.total_left_stats.coded_error != 0.0)
+ fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %10d %6d %6d"
+ "%6d %6d %6d %5d %5d %5d %8d %8.2f %10d %10.3f"
+ "%10.3f %8d\n",
+ cpi->common.current_video_frame, cpi->this_frame_target,
+ cpi->projected_frame_size,
+ (cpi->projected_frame_size - cpi->this_frame_target),
+ (int)cpi->total_target_vs_actual,
+ cpi->buffer_level,
+ (cpi->oxcf.starting_buffer_level-cpi->bits_off_target),
+ (int)cpi->total_actual_bits, cm->base_qindex,
+ cpi->active_best_quality, cpi->active_worst_quality,
+ cpi->ni_av_qi, cpi->cq_target_quality,
+ cpi->zbin_over_quant,
+ cm->refresh_golden_frame, cm->refresh_alt_ref_frame,
+ cm->frame_type, cpi->gfu_boost,
+ cpi->twopass.est_max_qcorrection_factor,
+ (int)cpi->twopass.bits_left,
+ cpi->twopass.total_left_stats.coded_error,
+ (double)cpi->twopass.bits_left /
+ cpi->twopass.total_left_stats.coded_error,
+ cpi->tot_recode_hits);
+ else
+ fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %10d %6d %6d"
+ "%6d %6d %6d %5d %5d %5d %8d %8.2f %10d %10.3f"
+ "%8d\n",
+ cpi->common.current_video_frame,
+ cpi->this_frame_target, cpi->projected_frame_size,
+ (cpi->projected_frame_size - cpi->this_frame_target),
+ (int)cpi->total_target_vs_actual,
+ cpi->buffer_level,
+ (cpi->oxcf.starting_buffer_level-cpi->bits_off_target),
+ (int)cpi->total_actual_bits, cm->base_qindex,
+ cpi->active_best_quality, cpi->active_worst_quality,
+ cpi->ni_av_qi, cpi->cq_target_quality,
+ cpi->zbin_over_quant,
+ cm->refresh_golden_frame, cm->refresh_alt_ref_frame,
+ cm->frame_type, cpi->gfu_boost,
+ cpi->twopass.est_max_qcorrection_factor,
+ (int)cpi->twopass.bits_left,
+ cpi->twopass.total_left_stats.coded_error,
+ cpi->tot_recode_hits);
+
+ fclose(f);
+
+ {
+ FILE *fmodes = fopen("Modes.stt", "a");
+ int i;
+
+ fprintf(fmodes, "%6d:%1d:%1d:%1d ",
+ cpi->common.current_video_frame,
+ cm->frame_type, cm->refresh_golden_frame,
+ cm->refresh_alt_ref_frame);
+
+ for (i = 0; i < MAX_MODES; i++)
+ fprintf(fmodes, "%5d ", cpi->mode_chosen_counts[i]);
+
+ fprintf(fmodes, "\n");
+
+ fclose(fmodes);
+ }
+ }
+
+#endif
+
+ if (cm->refresh_golden_frame == 1)
+ cm->frame_flags = cm->frame_flags | FRAMEFLAGS_GOLDEN;
+ else
+ cm->frame_flags = cm->frame_flags&~FRAMEFLAGS_GOLDEN;
+
+ if (cm->refresh_alt_ref_frame == 1)
+ cm->frame_flags = cm->frame_flags | FRAMEFLAGS_ALTREF;
+ else
+ cm->frame_flags = cm->frame_flags&~FRAMEFLAGS_ALTREF;
+
+
+ if (cm->refresh_last_frame & cm->refresh_golden_frame)
+ /* both refreshed */
+ cpi->gold_is_last = 1;
+ else if (cm->refresh_last_frame ^ cm->refresh_golden_frame)
+ /* 1 refreshed but not the other */
+ cpi->gold_is_last = 0;
+
+ if (cm->refresh_last_frame & cm->refresh_alt_ref_frame)
+ /* both refreshed */
+ cpi->alt_is_last = 1;
+ else if (cm->refresh_last_frame ^ cm->refresh_alt_ref_frame)
+ /* 1 refreshed but not the other */
+ cpi->alt_is_last = 0;
+
+ if (cm->refresh_alt_ref_frame & cm->refresh_golden_frame)
+ /* both refreshed */
+ cpi->gold_is_alt = 1;
+ else if (cm->refresh_alt_ref_frame ^ cm->refresh_golden_frame)
+ /* 1 refreshed but not the other */
+ cpi->gold_is_alt = 0;
+
+ cpi->ref_frame_flags = VP8_ALTR_FRAME | VP8_GOLD_FRAME | VP8_LAST_FRAME;
+
+ if (cpi->gold_is_last)
+ cpi->ref_frame_flags &= ~VP8_GOLD_FRAME;
+
+ if (cpi->alt_is_last)
+ cpi->ref_frame_flags &= ~VP8_ALTR_FRAME;
+
+ if (cpi->gold_is_alt)
+ cpi->ref_frame_flags &= ~VP8_ALTR_FRAME;
+
+
+ if (!cpi->oxcf.error_resilient_mode)
+ {
+ if (cpi->oxcf.play_alternate && cm->refresh_alt_ref_frame && (cm->frame_type != KEY_FRAME))
+ /* Update the alternate reference frame stats as appropriate. */
+ update_alt_ref_frame_stats(cpi);
+ else
+ /* Update the Golden frame stats as appropriate. */
+ update_golden_frame_stats(cpi);
+ }
+
+ if (cm->frame_type == KEY_FRAME)
+ {
+ /* Tell the caller that the frame was coded as a key frame */
+ *frame_flags = cm->frame_flags | FRAMEFLAGS_KEY;
+
+ /* As this frame is a key frame the next defaults to an inter frame. */
+ cm->frame_type = INTER_FRAME;
+
+ cpi->last_frame_percent_intra = 100;
+ }
+ else
+ {
+ *frame_flags = cm->frame_flags&~FRAMEFLAGS_KEY;
+
+ cpi->last_frame_percent_intra = cpi->this_frame_percent_intra;
+ }
+
+ /* Clear the one shot update flags for segmentation map and mode/ref
+ * loop filter deltas.
+ */
+ cpi->mb.e_mbd.update_mb_segmentation_map = 0;
+ cpi->mb.e_mbd.update_mb_segmentation_data = 0;
+ cpi->mb.e_mbd.mode_ref_lf_delta_update = 0;
+
+
+ /* Dont increment frame counters if this was an altref buffer update
+ * not a real frame
+ */
+ if (cm->show_frame)
+ {
+ cm->current_video_frame++;
+ cpi->frames_since_key++;
+ }
+
+ /* reset to normal state now that we are done. */
+
+
+
+#if 0
+ {
+ char filename[512];
+ FILE *recon_file;
+ sprintf(filename, "enc%04d.yuv", (int) cm->current_video_frame);
+ recon_file = fopen(filename, "wb");
+ fwrite(cm->yv12_fb[cm->lst_fb_idx].buffer_alloc,
+ cm->yv12_fb[cm->lst_fb_idx].frame_size, 1, recon_file);
+ fclose(recon_file);
+ }
+#endif
+
+ /* DEBUG */
+ /* vp8_write_yuv_frame("encoder_recon.yuv", cm->frame_to_show); */
+
+
+}
+
+
+static void check_gf_quality(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+ int gf_active_pct = (100 * cpi->gf_active_count) / (cm->mb_rows * cm->mb_cols);
+ int gf_ref_usage_pct = (cpi->count_mb_ref_frame_usage[GOLDEN_FRAME] * 100) / (cm->mb_rows * cm->mb_cols);
+ int last_ref_zz_useage = (cpi->inter_zz_count * 100) / (cm->mb_rows * cm->mb_cols);
+
+ /* Gf refresh is not currently being signalled */
+ if (cpi->gf_update_recommended == 0)
+ {
+ if (cpi->common.frames_since_golden > 7)
+ {
+ /* Low use of gf */
+ if ((gf_active_pct < 10) || ((gf_active_pct + gf_ref_usage_pct) < 15))
+ {
+ /* ...but last frame zero zero usage is reasonbable so a
+ * new gf might be appropriate
+ */
+ if (last_ref_zz_useage >= 25)
+ {
+ cpi->gf_bad_count ++;
+
+ /* Check that the condition is stable */
+ if (cpi->gf_bad_count >= 8)
+ {
+ cpi->gf_update_recommended = 1;
+ cpi->gf_bad_count = 0;
+ }
+ }
+ else
+ /* Restart count as the background is not stable enough */
+ cpi->gf_bad_count = 0;
+ }
+ else
+ /* Gf useage has picked up so reset count */
+ cpi->gf_bad_count = 0;
+ }
+ }
+ /* If the signal is set but has not been read should we cancel it. */
+ else if (last_ref_zz_useage < 15)
+ {
+ cpi->gf_update_recommended = 0;
+ cpi->gf_bad_count = 0;
+ }
+
+#if 0
+ {
+ FILE *f = fopen("gfneeded.stt", "a");
+ fprintf(f, "%10d %10d %10d %10d %10ld \n",
+ cm->current_video_frame,
+ cpi->common.frames_since_golden,
+ gf_active_pct, gf_ref_usage_pct,
+ cpi->gf_update_recommended);
+ fclose(f);
+ }
+
+#endif
+}
+
+#if !(CONFIG_REALTIME_ONLY)
+static void Pass2Encode(VP8_COMP *cpi, unsigned long *size, unsigned char *dest, unsigned char * dest_end, unsigned int *frame_flags)
+{
+
+ if (!cpi->common.refresh_alt_ref_frame)
+ vp8_second_pass(cpi);
+
+ encode_frame_to_data_rate(cpi, size, dest, dest_end, frame_flags);
+ cpi->twopass.bits_left -= 8 * *size;
+
+ if (!cpi->common.refresh_alt_ref_frame)
+ {
+ double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth
+ *cpi->oxcf.two_pass_vbrmin_section / 100);
+ cpi->twopass.bits_left += (int64_t)(two_pass_min_rate / cpi->frame_rate);
+ }
+}
+#endif
+
+/* For ARM NEON, d8-d15 are callee-saved registers, and need to be saved. */
+#if HAVE_NEON
+extern void vp8_push_neon(int64_t *store);
+extern void vp8_pop_neon(int64_t *store);
+#endif
+
+
+int vp8_receive_raw_frame(VP8_COMP *cpi, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, int64_t time_stamp, int64_t end_time)
+{
+#if HAVE_NEON
+ int64_t store_reg[8];
+#endif
+ VP8_COMMON *cm = &cpi->common;
+ struct vpx_usec_timer timer;
+ int res = 0;
+
+#if HAVE_NEON
+#if CONFIG_RUNTIME_CPU_DETECT
+ if (cm->cpu_caps & HAS_NEON)
+#endif
+ {
+ vp8_push_neon(store_reg);
+ }
+#endif
+
+ vpx_usec_timer_start(&timer);
+
+ /* Reinit the lookahead buffer if the frame size changes */
+ if (sd->y_width != cpi->oxcf.Width || sd->y_height != cpi->oxcf.Height)
+ {
+ assert(cpi->oxcf.lag_in_frames < 2);
+ dealloc_raw_frame_buffers(cpi);
+ alloc_raw_frame_buffers(cpi);
+ }
+
+ if(vp8_lookahead_push(cpi->lookahead, sd, time_stamp, end_time,
+ frame_flags, cpi->active_map_enabled ? cpi->active_map : NULL))
+ res = -1;
+ cm->clr_type = sd->clrtype;
+ vpx_usec_timer_mark(&timer);
+ cpi->time_receive_data += vpx_usec_timer_elapsed(&timer);
+
+#if HAVE_NEON
+#if CONFIG_RUNTIME_CPU_DETECT
+ if (cm->cpu_caps & HAS_NEON)
+#endif
+ {
+ vp8_pop_neon(store_reg);
+ }
+#endif
+
+ return res;
+}
+
+
+static int frame_is_reference(const VP8_COMP *cpi)
+{
+ const VP8_COMMON *cm = &cpi->common;
+ const MACROBLOCKD *xd = &cpi->mb.e_mbd;
+
+ return cm->frame_type == KEY_FRAME || cm->refresh_last_frame
+ || cm->refresh_golden_frame || cm->refresh_alt_ref_frame
+ || cm->copy_buffer_to_gf || cm->copy_buffer_to_arf
+ || cm->refresh_entropy_probs
+ || xd->mode_ref_lf_delta_update
+ || xd->update_mb_segmentation_map || xd->update_mb_segmentation_data;
+}
+
+
+int vp8_get_compressed_data(VP8_COMP *cpi, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, unsigned char *dest_end, int64_t *time_stamp, int64_t *time_end, int flush)
+{
+#if HAVE_NEON
+ int64_t store_reg[8];
+#endif
+ VP8_COMMON *cm;
+ struct vpx_usec_timer tsctimer;
+ struct vpx_usec_timer ticktimer;
+ struct vpx_usec_timer cmptimer;
+ YV12_BUFFER_CONFIG *force_src_buffer = NULL;
+
+ if (!cpi)
+ return -1;
+
+ cm = &cpi->common;
+
+ if (setjmp(cpi->common.error.jmp))
+ {
+ cpi->common.error.setjmp = 0;
+ return VPX_CODEC_CORRUPT_FRAME;
+ }
+
+ cpi->common.error.setjmp = 1;
+
+#if HAVE_NEON
+#if CONFIG_RUNTIME_CPU_DETECT
+ if (cm->cpu_caps & HAS_NEON)
+#endif
+ {
+ vp8_push_neon(store_reg);
+ }
+#endif
+
+ vpx_usec_timer_start(&cmptimer);
+
+ cpi->source = NULL;
+
+#if !(CONFIG_REALTIME_ONLY)
+ /* Should we code an alternate reference frame */
+ if (cpi->oxcf.error_resilient_mode == 0 &&
+ cpi->oxcf.play_alternate &&
+ cpi->source_alt_ref_pending)
+ {
+ if ((cpi->source = vp8_lookahead_peek(cpi->lookahead,
+ cpi->frames_till_gf_update_due,
+ PEEK_FORWARD)))
+ {
+ cpi->alt_ref_source = cpi->source;
+ if (cpi->oxcf.arnr_max_frames > 0)
+ {
+ vp8_temporal_filter_prepare_c(cpi,
+ cpi->frames_till_gf_update_due);
+ force_src_buffer = &cpi->alt_ref_buffer;
+ }
+ cm->frames_till_alt_ref_frame = cpi->frames_till_gf_update_due;
+ cm->refresh_alt_ref_frame = 1;
+ cm->refresh_golden_frame = 0;
+ cm->refresh_last_frame = 0;
+ cm->show_frame = 0;
+ /* Clear Pending alt Ref flag. */
+ cpi->source_alt_ref_pending = 0;
+ cpi->is_src_frame_alt_ref = 0;
+ }
+ }
+#endif
+
+ if (!cpi->source)
+ {
+ /* Read last frame source if we are encoding first pass. */
+ if (cpi->pass == 1 && cm->current_video_frame > 0)
+ {
+ if((cpi->last_source = vp8_lookahead_peek(cpi->lookahead, 1,
+ PEEK_BACKWARD)) == NULL)
+ return -1;
+ }
+
+
+ if ((cpi->source = vp8_lookahead_pop(cpi->lookahead, flush)))
+ {
+ cm->show_frame = 1;
+
+ cpi->is_src_frame_alt_ref = cpi->alt_ref_source
+ && (cpi->source == cpi->alt_ref_source);
+
+ if(cpi->is_src_frame_alt_ref)
+ cpi->alt_ref_source = NULL;
+ }
+ }
+
+ if (cpi->source)
+ {
+ cpi->Source = force_src_buffer ? force_src_buffer : &cpi->source->img;
+ cpi->un_scaled_source = cpi->Source;
+ *time_stamp = cpi->source->ts_start;
+ *time_end = cpi->source->ts_end;
+ *frame_flags = cpi->source->flags;
+
+ if (cpi->pass == 1 && cm->current_video_frame > 0)
+ {
+ cpi->last_frame_unscaled_source = &cpi->last_source->img;
+ }
+ }
+ else
+ {
+ *size = 0;
+#if !(CONFIG_REALTIME_ONLY)
+
+ if (flush && cpi->pass == 1 && !cpi->twopass.first_pass_done)
+ {
+ vp8_end_first_pass(cpi); /* get last stats packet */
+ cpi->twopass.first_pass_done = 1;
+ }
+
+#endif
+
+#if HAVE_NEON
+#if CONFIG_RUNTIME_CPU_DETECT
+ if (cm->cpu_caps & HAS_NEON)
+#endif
+ {
+ vp8_pop_neon(store_reg);
+ }
+#endif
+ return -1;
+ }
+
+ if (cpi->source->ts_start < cpi->first_time_stamp_ever)
+ {
+ cpi->first_time_stamp_ever = cpi->source->ts_start;
+ cpi->last_end_time_stamp_seen = cpi->source->ts_start;
+ }
+
+ /* adjust frame rates based on timestamps given */
+ if (cm->show_frame)
+ {
+ int64_t this_duration;
+ int step = 0;
+
+ if (cpi->source->ts_start == cpi->first_time_stamp_ever)
+ {
+ this_duration = cpi->source->ts_end - cpi->source->ts_start;
+ step = 1;
+ }
+ else
+ {
+ int64_t last_duration;
+
+ this_duration = cpi->source->ts_end - cpi->last_end_time_stamp_seen;
+ last_duration = cpi->last_end_time_stamp_seen
+ - cpi->last_time_stamp_seen;
+ /* do a step update if the duration changes by 10% */
+ if (last_duration)
+ step = (int)(((this_duration - last_duration) *
+ 10 / last_duration));
+ }
+
+ if (this_duration)
+ {
+ if (step)
+ cpi->ref_frame_rate = 10000000.0 / this_duration;
+ else
+ {
+ double avg_duration, interval;
+
+ /* Average this frame's rate into the last second's average
+ * frame rate. If we haven't seen 1 second yet, then average
+ * over the whole interval seen.
+ */
+ interval = (double)(cpi->source->ts_end -
+ cpi->first_time_stamp_ever);
+ if(interval > 10000000.0)
+ interval = 10000000;
+
+ avg_duration = 10000000.0 / cpi->ref_frame_rate;
+ avg_duration *= (interval - avg_duration + this_duration);
+ avg_duration /= interval;
+
+ cpi->ref_frame_rate = 10000000.0 / avg_duration;
+ }
+
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ unsigned int i;
+
+ /* Update frame rates for each layer */
+ for (i=0; i<cpi->oxcf.number_of_layers; i++)
+ {
+ LAYER_CONTEXT *lc = &cpi->layer_context[i];
+ lc->frame_rate = cpi->ref_frame_rate /
+ cpi->oxcf.rate_decimator[i];
+ }
+ }
+ else
+ vp8_new_frame_rate(cpi, cpi->ref_frame_rate);
+ }
+
+ cpi->last_time_stamp_seen = cpi->source->ts_start;
+ cpi->last_end_time_stamp_seen = cpi->source->ts_end;
+ }
+
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ int layer;
+
+ update_layer_contexts (cpi);
+
+ /* Restore layer specific context & set frame rate */
+ layer = cpi->oxcf.layer_id[
+ cm->current_video_frame % cpi->oxcf.periodicity];
+ restore_layer_context (cpi, layer);
+ vp8_new_frame_rate (cpi, cpi->layer_context[layer].frame_rate);
+ }
+
+ if (cpi->compressor_speed == 2)
+ {
+ if (cpi->oxcf.number_of_layers == 1)
+ check_gf_quality(cpi);
+ vpx_usec_timer_start(&tsctimer);
+ vpx_usec_timer_start(&ticktimer);
+ }
+
+ cpi->lf_zeromv_pct = (cpi->zeromv_count * 100)/cm->MBs;
+
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ {
+ int i;
+ const int num_part = (1 << cm->multi_token_partition);
+ /* the available bytes in dest */
+ const unsigned long dest_size = dest_end - dest;
+ const int tok_part_buff_size = (dest_size * 9) / (10 * num_part);
+
+ unsigned char *dp = dest;
+
+ cpi->partition_d[0] = dp;
+ dp += dest_size/10; /* reserve 1/10 for control partition */
+ cpi->partition_d_end[0] = dp;
+
+ for(i = 0; i < num_part; i++)
+ {
+ cpi->partition_d[i + 1] = dp;
+ dp += tok_part_buff_size;
+ cpi->partition_d_end[i + 1] = dp;
+ }
+ }
+#endif
+
+ /* start with a 0 size frame */
+ *size = 0;
+
+ /* Clear down mmx registers */
+ vp8_clear_system_state();
+
+ cm->frame_type = INTER_FRAME;
+ cm->frame_flags = *frame_flags;
+
+#if 0
+
+ if (cm->refresh_alt_ref_frame)
+ {
+ cm->refresh_golden_frame = 0;
+ cm->refresh_last_frame = 0;
+ }
+ else
+ {
+ cm->refresh_golden_frame = 0;
+ cm->refresh_last_frame = 1;
+ }
+
+#endif
+ /* find a free buffer for the new frame */
+ {
+ int i = 0;
+ for(; i < NUM_YV12_BUFFERS; i++)
+ {
+ if(!cm->yv12_fb[i].flags)
+ {
+ cm->new_fb_idx = i;
+ break;
+ }
+ }
+
+ assert(i < NUM_YV12_BUFFERS );
+ }
+#if !(CONFIG_REALTIME_ONLY)
+
+ if (cpi->pass == 1)
+ {
+ Pass1Encode(cpi, size, dest, frame_flags);
+ }
+ else if (cpi->pass == 2)
+ {
+ Pass2Encode(cpi, size, dest, dest_end, frame_flags);
+ }
+ else
+#endif
+ encode_frame_to_data_rate(cpi, size, dest, dest_end, frame_flags);
+
+ if (cpi->compressor_speed == 2)
+ {
+ unsigned int duration, duration2;
+ vpx_usec_timer_mark(&tsctimer);
+ vpx_usec_timer_mark(&ticktimer);
+
+ duration = (int)(vpx_usec_timer_elapsed(&ticktimer));
+ duration2 = (unsigned int)((double)duration / 2);
+
+ if (cm->frame_type != KEY_FRAME)
+ {
+ if (cpi->avg_encode_time == 0)
+ cpi->avg_encode_time = duration;
+ else
+ cpi->avg_encode_time = (7 * cpi->avg_encode_time + duration) >> 3;
+ }
+
+ if (duration2)
+ {
+ {
+
+ if (cpi->avg_pick_mode_time == 0)
+ cpi->avg_pick_mode_time = duration2;
+ else
+ cpi->avg_pick_mode_time = (7 * cpi->avg_pick_mode_time + duration2) >> 3;
+ }
+ }
+
+ }
+
+ if (cm->refresh_entropy_probs == 0)
+ {
+ vpx_memcpy(&cm->fc, &cm->lfc, sizeof(cm->fc));
+ }
+
+ /* Save the contexts separately for alt ref, gold and last. */
+ /* (TODO jbb -> Optimize this with pointers to avoid extra copies. ) */
+ if(cm->refresh_alt_ref_frame)
+ vpx_memcpy(&cpi->lfc_a, &cm->fc, sizeof(cm->fc));
+
+ if(cm->refresh_golden_frame)
+ vpx_memcpy(&cpi->lfc_g, &cm->fc, sizeof(cm->fc));
+
+ if(cm->refresh_last_frame)
+ vpx_memcpy(&cpi->lfc_n, &cm->fc, sizeof(cm->fc));
+
+ /* if its a dropped frame honor the requests on subsequent frames */
+ if (*size > 0)
+ {
+ cpi->droppable = !frame_is_reference(cpi);
+
+ /* return to normal state */
+ cm->refresh_entropy_probs = 1;
+ cm->refresh_alt_ref_frame = 0;
+ cm->refresh_golden_frame = 0;
+ cm->refresh_last_frame = 1;
+ cm->frame_type = INTER_FRAME;
+
+ }
+
+ /* Save layer specific state */
+ if (cpi->oxcf.number_of_layers > 1)
+ save_layer_context (cpi);
+
+ vpx_usec_timer_mark(&cmptimer);
+ cpi->time_compress_data += vpx_usec_timer_elapsed(&cmptimer);
+
+ if (cpi->b_calculate_psnr && cpi->pass != 1 && cm->show_frame)
+ {
+ generate_psnr_packet(cpi);
+ }
+
+#if CONFIG_INTERNAL_STATS
+
+ if (cpi->pass != 1)
+ {
+ cpi->bytes += *size;
+
+ if (cm->show_frame)
+ {
+
+ cpi->count ++;
+
+ if (cpi->b_calculate_psnr)
+ {
+ uint64_t ye,ue,ve;
+ double frame_psnr;
+ YV12_BUFFER_CONFIG *orig = cpi->Source;
+ YV12_BUFFER_CONFIG *recon = cpi->common.frame_to_show;
+ int y_samples = orig->y_height * orig->y_width ;
+ int uv_samples = orig->uv_height * orig->uv_width ;
+ int t_samples = y_samples + 2 * uv_samples;
+ double sq_error, sq_error2;
+
+ ye = calc_plane_error(orig->y_buffer, orig->y_stride,
+ recon->y_buffer, recon->y_stride, orig->y_width, orig->y_height);
+
+ ue = calc_plane_error(orig->u_buffer, orig->uv_stride,
+ recon->u_buffer, recon->uv_stride, orig->uv_width, orig->uv_height);
+
+ ve = calc_plane_error(orig->v_buffer, orig->uv_stride,
+ recon->v_buffer, recon->uv_stride, orig->uv_width, orig->uv_height);
+
+ sq_error = (double)(ye + ue + ve);
+
+ frame_psnr = vp8_mse2psnr(t_samples, 255.0, sq_error);
+
+ cpi->total_y += vp8_mse2psnr(y_samples, 255.0, (double)ye);
+ cpi->total_u += vp8_mse2psnr(uv_samples, 255.0, (double)ue);
+ cpi->total_v += vp8_mse2psnr(uv_samples, 255.0, (double)ve);
+ cpi->total_sq_error += sq_error;
+ cpi->total += frame_psnr;
+#if CONFIG_POSTPROC
+ {
+ YV12_BUFFER_CONFIG *pp = &cm->post_proc_buffer;
+ double frame_psnr2, frame_ssim2 = 0;
+ double weight = 0;
+
+ vp8_deblock(cm, cm->frame_to_show, &cm->post_proc_buffer, cm->filter_level * 10 / 6, 1, 0);
+ vp8_clear_system_state();
+
+ ye = calc_plane_error(orig->y_buffer, orig->y_stride,
+ pp->y_buffer, pp->y_stride, orig->y_width, orig->y_height);
+
+ ue = calc_plane_error(orig->u_buffer, orig->uv_stride,
+ pp->u_buffer, pp->uv_stride, orig->uv_width, orig->uv_height);
+
+ ve = calc_plane_error(orig->v_buffer, orig->uv_stride,
+ pp->v_buffer, pp->uv_stride, orig->uv_width, orig->uv_height);
+
+ sq_error2 = (double)(ye + ue + ve);
+
+ frame_psnr2 = vp8_mse2psnr(t_samples, 255.0, sq_error2);
+
+ cpi->totalp_y += vp8_mse2psnr(y_samples,
+ 255.0, (double)ye);
+ cpi->totalp_u += vp8_mse2psnr(uv_samples,
+ 255.0, (double)ue);
+ cpi->totalp_v += vp8_mse2psnr(uv_samples,
+ 255.0, (double)ve);
+ cpi->total_sq_error2 += sq_error2;
+ cpi->totalp += frame_psnr2;
+
+ frame_ssim2 = vp8_calc_ssim(cpi->Source,
+ &cm->post_proc_buffer, 1, &weight);
+
+ cpi->summed_quality += frame_ssim2 * weight;
+ cpi->summed_weights += weight;
+
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ unsigned int i;
+
+ for (i=cpi->current_layer;
+ i<cpi->oxcf.number_of_layers; i++)
+ {
+ cpi->frames_in_layer[i]++;
+
+ cpi->bytes_in_layer[i] += *size;
+ cpi->sum_psnr[i] += frame_psnr;
+ cpi->sum_psnr_p[i] += frame_psnr2;
+ cpi->total_error2[i] += sq_error;
+ cpi->total_error2_p[i] += sq_error2;
+ cpi->sum_ssim[i] += frame_ssim2 * weight;
+ cpi->sum_weights[i] += weight;
+ }
+ }
+ }
+#endif
+ }
+
+ if (cpi->b_calculate_ssimg)
+ {
+ double y, u, v, frame_all;
+ frame_all = vp8_calc_ssimg(cpi->Source, cm->frame_to_show,
+ &y, &u, &v);
+
+ if (cpi->oxcf.number_of_layers > 1)
+ {
+ unsigned int i;
+
+ for (i=cpi->current_layer;
+ i<cpi->oxcf.number_of_layers; i++)
+ {
+ if (!cpi->b_calculate_psnr)
+ cpi->frames_in_layer[i]++;
+
+ cpi->total_ssimg_y_in_layer[i] += y;
+ cpi->total_ssimg_u_in_layer[i] += u;
+ cpi->total_ssimg_v_in_layer[i] += v;
+ cpi->total_ssimg_all_in_layer[i] += frame_all;
+ }
+ }
+ else
+ {
+ cpi->total_ssimg_y += y;
+ cpi->total_ssimg_u += u;
+ cpi->total_ssimg_v += v;
+ cpi->total_ssimg_all += frame_all;
+ }
+ }
+
+ }
+ }
+
+#if 0
+
+ if (cpi->common.frame_type != 0 && cpi->common.base_qindex == cpi->oxcf.worst_allowed_q)
+ {
+ skiptruecount += cpi->skip_true_count;
+ skipfalsecount += cpi->skip_false_count;
+ }
+
+#endif
+#if 0
+
+ if (cpi->pass != 1)
+ {
+ FILE *f = fopen("skip.stt", "a");
+ fprintf(f, "frame:%4d flags:%4x Q:%4d P:%4d Size:%5d\n", cpi->common.current_video_frame, *frame_flags, cpi->common.base_qindex, cpi->prob_skip_false, *size);
+
+ if (cpi->is_src_frame_alt_ref == 1)
+ fprintf(f, "skipcount: %4d framesize: %d\n", cpi->skip_true_count , *size);
+
+ fclose(f);
+ }
+
+#endif
+#endif
+
+#if HAVE_NEON
+#if CONFIG_RUNTIME_CPU_DETECT
+ if (cm->cpu_caps & HAS_NEON)
+#endif
+ {
+ vp8_pop_neon(store_reg);
+ }
+#endif
+
+ cpi->common.error.setjmp = 0;
+
+ return 0;
+}
+
+int vp8_get_preview_raw_frame(VP8_COMP *cpi, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags)
+{
+ if (cpi->common.refresh_alt_ref_frame)
+ return -1;
+ else
+ {
+ int ret;
+
+#if CONFIG_MULTITHREAD
+ if(cpi->b_lpf_running)
+ {
+ sem_wait(&cpi->h_event_end_lpf);
+ cpi->b_lpf_running = 0;
+ }
+#endif
+
+#if CONFIG_POSTPROC
+ ret = vp8_post_proc_frame(&cpi->common, dest, flags);
+#else
+
+ if (cpi->common.frame_to_show)
+ {
+ *dest = *cpi->common.frame_to_show;
+ dest->y_width = cpi->common.Width;
+ dest->y_height = cpi->common.Height;
+ dest->uv_height = cpi->common.Height / 2;
+ ret = 0;
+ }
+ else
+ {
+ ret = -1;
+ }
+
+#endif
+ vp8_clear_system_state();
+ return ret;
+ }
+}
+
+int vp8_set_roimap(VP8_COMP *cpi, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4])
+{
+ signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
+ int internal_delta_q[MAX_MB_SEGMENTS];
+ const int range = 63;
+ int i;
+
+ // This method is currently incompatible with the cyclic refresh method
+ if ( cpi->cyclic_refresh_mode_enabled )
+ return -1;
+
+ // Check number of rows and columns match
+ if (cpi->common.mb_rows != rows || cpi->common.mb_cols != cols)
+ return -1;
+
+ // Range check the delta Q values and convert the external Q range values
+ // to internal ones.
+ if ( (abs(delta_q[0]) > range) || (abs(delta_q[1]) > range) ||
+ (abs(delta_q[2]) > range) || (abs(delta_q[3]) > range) )
+ return -1;
+
+ // Range check the delta lf values
+ if ( (abs(delta_lf[0]) > range) || (abs(delta_lf[1]) > range) ||
+ (abs(delta_lf[2]) > range) || (abs(delta_lf[3]) > range) )
+ return -1;
+
+ if (!map)
+ {
+ disable_segmentation(cpi);
+ return 0;
+ }
+
+ // Translate the external delta q values to internal values.
+ for ( i = 0; i < MAX_MB_SEGMENTS; i++ )
+ internal_delta_q[i] =
+ ( delta_q[i] >= 0 ) ? q_trans[delta_q[i]] : -q_trans[-delta_q[i]];
+
+ /* Set the segmentation Map */
+ set_segmentation_map(cpi, map);
+
+ /* Activate segmentation. */
+ enable_segmentation(cpi);
+
+ /* Set up the quant segment data */
+ feature_data[MB_LVL_ALT_Q][0] = internal_delta_q[0];
+ feature_data[MB_LVL_ALT_Q][1] = internal_delta_q[1];
+ feature_data[MB_LVL_ALT_Q][2] = internal_delta_q[2];
+ feature_data[MB_LVL_ALT_Q][3] = internal_delta_q[3];
+
+ /* Set up the loop segment data s */
+ feature_data[MB_LVL_ALT_LF][0] = delta_lf[0];
+ feature_data[MB_LVL_ALT_LF][1] = delta_lf[1];
+ feature_data[MB_LVL_ALT_LF][2] = delta_lf[2];
+ feature_data[MB_LVL_ALT_LF][3] = delta_lf[3];
+
+ cpi->segment_encode_breakout[0] = threshold[0];
+ cpi->segment_encode_breakout[1] = threshold[1];
+ cpi->segment_encode_breakout[2] = threshold[2];
+ cpi->segment_encode_breakout[3] = threshold[3];
+
+ /* Initialise the feature data structure */
+ set_segment_data(cpi, &feature_data[0][0], SEGMENT_DELTADATA);
+
+ return 0;
+}
+
+int vp8_set_active_map(VP8_COMP *cpi, unsigned char *map, unsigned int rows, unsigned int cols)
+{
+ if (rows == cpi->common.mb_rows && cols == cpi->common.mb_cols)
+ {
+ if (map)
+ {
+ vpx_memcpy(cpi->active_map, map, rows * cols);
+ cpi->active_map_enabled = 1;
+ }
+ else
+ cpi->active_map_enabled = 0;
+
+ return 0;
+ }
+ else
+ {
+ return -1 ;
+ }
+}
+
+int vp8_set_internal_size(VP8_COMP *cpi, VPX_SCALING horiz_mode, VPX_SCALING vert_mode)
+{
+ if (horiz_mode <= ONETWO)
+ cpi->common.horiz_scale = horiz_mode;
+ else
+ return -1;
+
+ if (vert_mode <= ONETWO)
+ cpi->common.vert_scale = vert_mode;
+ else
+ return -1;
+
+ return 0;
+}
+
+
+
+int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest)
+{
+ int i, j;
+ int Total = 0;
+
+ unsigned char *src = source->y_buffer;
+ unsigned char *dst = dest->y_buffer;
+
+ /* Loop through the Y plane raw and reconstruction data summing
+ * (square differences)
+ */
+ for (i = 0; i < source->y_height; i += 16)
+ {
+ for (j = 0; j < source->y_width; j += 16)
+ {
+ unsigned int sse;
+ Total += vp8_mse16x16(src + j, source->y_stride, dst + j, dest->y_stride, &sse);
+ }
+
+ src += 16 * source->y_stride;
+ dst += 16 * dest->y_stride;
+ }
+
+ return Total;
+}
+
+
+int vp8_get_quantizer(VP8_COMP *cpi)
+{
+ return cpi->common.base_qindex;
+}
diff --git a/libvpx/vp8/encoder/onyx_int.h b/libvpx/vp8/encoder/onyx_int.h
new file mode 100644
index 0000000..ed9c762
--- /dev/null
+++ b/libvpx/vp8/encoder/onyx_int.h
@@ -0,0 +1,741 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_VP8_INT_H
+#define __INC_VP8_INT_H
+
+#include <stdio.h>
+#include "vpx_config.h"
+#include "vp8/common/onyx.h"
+#include "treewriter.h"
+#include "tokenize.h"
+#include "vp8/common/onyxc_int.h"
+#include "vp8/common/variance.h"
+#include "encodemb.h"
+#include "quantize.h"
+#include "vp8/common/entropy.h"
+#include "vp8/common/threading.h"
+#include "vpx_ports/mem.h"
+#include "vpx/internal/vpx_codec_internal.h"
+#include "vpx/vp8.h"
+#include "mcomp.h"
+#include "vp8/common/findnearmv.h"
+#include "lookahead.h"
+#if CONFIG_TEMPORAL_DENOISING
+#include "vp8/encoder/denoising.h"
+#endif
+
+#define MIN_GF_INTERVAL 4
+#define DEFAULT_GF_INTERVAL 7
+
+#define KEY_FRAME_CONTEXT 5
+
+#define MAX_LAG_BUFFERS (CONFIG_REALTIME_ONLY? 1 : 25)
+
+#define AF_THRESH 25
+#define AF_THRESH2 100
+#define ARF_DECAY_THRESH 12
+#define MAX_MODES 20
+
+#define MIN_THRESHMULT 32
+#define MAX_THRESHMULT 512
+
+#define GF_ZEROMV_ZBIN_BOOST 12
+#define LF_ZEROMV_ZBIN_BOOST 6
+#define MV_ZBIN_BOOST 4
+#define ZBIN_OQ_MAX 192
+
+#if !(CONFIG_REALTIME_ONLY)
+#define VP8_TEMPORAL_ALT_REF 1
+#endif
+
+#define MAX(x,y) (((x)>(y))?(x):(y))
+#define MIN(x,y) (((x)<(y))?(x):(y))
+
+typedef struct
+{
+ int kf_indicated;
+ unsigned int frames_since_key;
+ unsigned int frames_since_golden;
+ int filter_level;
+ int frames_till_gf_update_due;
+ int recent_ref_frame_usage[MAX_REF_FRAMES];
+
+ MV_CONTEXT mvc[2];
+ int mvcosts[2][MVvals+1];
+
+#ifdef MODE_STATS
+ int y_modes[5];
+ int uv_modes[4];
+ int b_modes[10];
+ int inter_y_modes[10];
+ int inter_uv_modes[4];
+ int inter_b_modes[10];
+#endif
+
+ vp8_prob ymode_prob[4], uv_mode_prob[3]; /* interframe intra mode probs */
+ vp8_prob kf_ymode_prob[4], kf_uv_mode_prob[3]; /* keyframe "" */
+
+ int ymode_count[5], uv_mode_count[4]; /* intra MB type cts this frame */
+
+ int count_mb_ref_frame_usage[MAX_REF_FRAMES];
+
+ int this_frame_percent_intra;
+ int last_frame_percent_intra;
+
+
+} CODING_CONTEXT;
+
+typedef struct
+{
+ double frame;
+ double intra_error;
+ double coded_error;
+ double ssim_weighted_pred_err;
+ double pcnt_inter;
+ double pcnt_motion;
+ double pcnt_second_ref;
+ double pcnt_neutral;
+ double MVr;
+ double mvr_abs;
+ double MVc;
+ double mvc_abs;
+ double MVrv;
+ double MVcv;
+ double mv_in_out_count;
+ double new_mv_count;
+ double duration;
+ double count;
+}
+FIRSTPASS_STATS;
+
+typedef struct
+{
+ int frames_so_far;
+ double frame_intra_error;
+ double frame_coded_error;
+ double frame_pcnt_inter;
+ double frame_pcnt_motion;
+ double frame_mvr;
+ double frame_mvr_abs;
+ double frame_mvc;
+ double frame_mvc_abs;
+
+} ONEPASS_FRAMESTATS;
+
+
+typedef enum
+{
+ THR_ZERO1 = 0,
+ THR_DC = 1,
+
+ THR_NEAREST1 = 2,
+ THR_NEAR1 = 3,
+
+ THR_ZERO2 = 4,
+ THR_NEAREST2 = 5,
+
+ THR_ZERO3 = 6,
+ THR_NEAREST3 = 7,
+
+ THR_NEAR2 = 8,
+ THR_NEAR3 = 9,
+
+ THR_V_PRED = 10,
+ THR_H_PRED = 11,
+ THR_TM = 12,
+
+ THR_NEW1 = 13,
+ THR_NEW2 = 14,
+ THR_NEW3 = 15,
+
+ THR_SPLIT1 = 16,
+ THR_SPLIT2 = 17,
+ THR_SPLIT3 = 18,
+
+ THR_B_PRED = 19
+}
+THR_MODES;
+
+typedef enum
+{
+ DIAMOND = 0,
+ NSTEP = 1,
+ HEX = 2
+} SEARCH_METHODS;
+
+typedef struct
+{
+ int RD;
+ SEARCH_METHODS search_method;
+ int improved_quant;
+ int improved_dct;
+ int auto_filter;
+ int recode_loop;
+ int iterative_sub_pixel;
+ int half_pixel_search;
+ int quarter_pixel_search;
+ int thresh_mult[MAX_MODES];
+ int max_step_search_steps;
+ int first_step;
+ int optimize_coefficients;
+
+ int use_fastquant_for_pick;
+ int no_skip_block4x4_search;
+ int improved_mv_pred;
+
+} SPEED_FEATURES;
+
+typedef struct
+{
+ MACROBLOCK mb;
+ int segment_counts[MAX_MB_SEGMENTS];
+ int totalrate;
+} MB_ROW_COMP;
+
+typedef struct
+{
+ TOKENEXTRA *start;
+ TOKENEXTRA *stop;
+} TOKENLIST;
+
+typedef struct
+{
+ int ithread;
+ void *ptr1;
+ void *ptr2;
+} ENCODETHREAD_DATA;
+typedef struct
+{
+ int ithread;
+ void *ptr1;
+} LPFTHREAD_DATA;
+
+enum
+{
+ BLOCK_16X8,
+ BLOCK_8X16,
+ BLOCK_8X8,
+ BLOCK_4X4,
+ BLOCK_16X16,
+ BLOCK_MAX_SEGMENTS
+};
+
+typedef struct
+{
+ /* Layer configuration */
+ double frame_rate;
+ int target_bandwidth;
+
+ /* Layer specific coding parameters */
+ int64_t starting_buffer_level;
+ int64_t optimal_buffer_level;
+ int64_t maximum_buffer_size;
+ int64_t starting_buffer_level_in_ms;
+ int64_t optimal_buffer_level_in_ms;
+ int64_t maximum_buffer_size_in_ms;
+
+ int avg_frame_size_for_layer;
+
+ int64_t buffer_level;
+ int64_t bits_off_target;
+
+ int64_t total_actual_bits;
+ int total_target_vs_actual;
+
+ int worst_quality;
+ int active_worst_quality;
+ int best_quality;
+ int active_best_quality;
+
+ int ni_av_qi;
+ int ni_tot_qi;
+ int ni_frames;
+ int avg_frame_qindex;
+
+ double rate_correction_factor;
+ double key_frame_rate_correction_factor;
+ double gf_rate_correction_factor;
+
+ int zbin_over_quant;
+
+ int inter_frame_target;
+ int64_t total_byte_count;
+
+ int filter_level;
+
+ int last_frame_percent_intra;
+
+ int count_mb_ref_frame_usage[MAX_REF_FRAMES];
+
+} LAYER_CONTEXT;
+
+typedef struct VP8_COMP
+{
+
+ DECLARE_ALIGNED(16, short, Y1quant[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, unsigned char, Y1quant_shift[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, Y1zbin[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, Y1round[QINDEX_RANGE][16]);
+
+ DECLARE_ALIGNED(16, short, Y2quant[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, unsigned char, Y2quant_shift[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, Y2zbin[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, Y2round[QINDEX_RANGE][16]);
+
+ DECLARE_ALIGNED(16, short, UVquant[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, unsigned char, UVquant_shift[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, UVzbin[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, UVround[QINDEX_RANGE][16]);
+
+ DECLARE_ALIGNED(16, short, zrun_zbin_boost_y1[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, zrun_zbin_boost_y2[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, zrun_zbin_boost_uv[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, Y1quant_fast[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, Y2quant_fast[QINDEX_RANGE][16]);
+ DECLARE_ALIGNED(16, short, UVquant_fast[QINDEX_RANGE][16]);
+
+
+ MACROBLOCK mb;
+ VP8_COMMON common;
+ vp8_writer bc[9]; /* one boolcoder for each partition */
+
+ VP8_CONFIG oxcf;
+
+ struct lookahead_ctx *lookahead;
+ struct lookahead_entry *source;
+ struct lookahead_entry *alt_ref_source;
+ struct lookahead_entry *last_source;
+
+ YV12_BUFFER_CONFIG *Source;
+ YV12_BUFFER_CONFIG *un_scaled_source;
+ YV12_BUFFER_CONFIG scaled_source;
+ YV12_BUFFER_CONFIG *last_frame_unscaled_source;
+
+ /* frame in src_buffers has been identified to be encoded as an alt ref */
+ int source_alt_ref_pending;
+ /* an alt ref frame has been encoded and is usable */
+ int source_alt_ref_active;
+ /* source of frame to encode is an exact copy of an alt ref frame */
+ int is_src_frame_alt_ref;
+
+ /* golden frame same as last frame ( short circuit gold searches) */
+ int gold_is_last;
+ /* Alt reference frame same as last ( short circuit altref search) */
+ int alt_is_last;
+ /* don't do both alt and gold search ( just do gold). */
+ int gold_is_alt;
+
+ YV12_BUFFER_CONFIG pick_lf_lvl_frame;
+
+ TOKENEXTRA *tok;
+ unsigned int tok_count;
+
+
+ unsigned int frames_since_key;
+ unsigned int key_frame_frequency;
+ unsigned int this_key_frame_forced;
+ unsigned int next_key_frame_forced;
+
+ /* Ambient reconstruction err target for force key frames */
+ int ambient_err;
+
+ unsigned int mode_check_freq[MAX_MODES];
+ unsigned int mode_test_hit_counts[MAX_MODES];
+ unsigned int mode_chosen_counts[MAX_MODES];
+ unsigned int mbs_tested_so_far;
+
+ int rd_thresh_mult[MAX_MODES];
+ int rd_baseline_thresh[MAX_MODES];
+ int rd_threshes[MAX_MODES];
+
+ int RDMULT;
+ int RDDIV ;
+
+ CODING_CONTEXT coding_context;
+
+ /* Rate targetting variables */
+ int64_t last_prediction_error;
+ int64_t last_intra_error;
+
+ int this_frame_target;
+ int projected_frame_size;
+ int last_q[2]; /* Separate values for Intra/Inter */
+
+ double rate_correction_factor;
+ double key_frame_rate_correction_factor;
+ double gf_rate_correction_factor;
+
+ /* Count down till next GF */
+ int frames_till_gf_update_due;
+
+ /* GF interval chosen when we coded the last GF */
+ int current_gf_interval;
+
+ /* Total bits overspent becasue of GF boost (cumulative) */
+ int gf_overspend_bits;
+
+ /* Used in the few frames following a GF to recover the extra bits
+ * spent in that GF
+ */
+ int non_gf_bitrate_adjustment;
+
+ /* Extra bits spent on key frames that need to be recovered */
+ int kf_overspend_bits;
+
+ /* Current number of bit s to try and recover on each inter frame. */
+ int kf_bitrate_adjustment;
+ int max_gf_interval;
+ int baseline_gf_interval;
+ int active_arnr_frames;
+
+ int64_t key_frame_count;
+ int prior_key_frame_distance[KEY_FRAME_CONTEXT];
+ /* Current section per frame bandwidth target */
+ int per_frame_bandwidth;
+ /* Average frame size target for clip */
+ int av_per_frame_bandwidth;
+ /* Minimum allocation that should be used for any frame */
+ int min_frame_bandwidth;
+ int inter_frame_target;
+ double output_frame_rate;
+ int64_t last_time_stamp_seen;
+ int64_t last_end_time_stamp_seen;
+ int64_t first_time_stamp_ever;
+
+ int ni_av_qi;
+ int ni_tot_qi;
+ int ni_frames;
+ int avg_frame_qindex;
+
+ int zbin_over_quant;
+ int zbin_mode_boost;
+ int zbin_mode_boost_enabled;
+ int last_zbin_over_quant;
+ int last_zbin_mode_boost;
+
+ int64_t total_byte_count;
+
+ int buffered_mode;
+
+ double frame_rate;
+ double ref_frame_rate;
+ int64_t buffer_level;
+ int64_t bits_off_target;
+
+ int rolling_target_bits;
+ int rolling_actual_bits;
+
+ int long_rolling_target_bits;
+ int long_rolling_actual_bits;
+
+ int64_t total_actual_bits;
+ int total_target_vs_actual; /* debug stats */
+
+ int worst_quality;
+ int active_worst_quality;
+ int best_quality;
+ int active_best_quality;
+
+ int cq_target_quality;
+
+ int drop_frames_allowed; /* Are we permitted to drop frames? */
+ int drop_frame; /* Drop this frame? */
+
+ vp8_prob frame_coef_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES];
+ char update_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES];
+
+ unsigned int frame_branch_ct [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES][2];
+
+ int gfu_boost;
+ int kf_boost;
+ int last_boost;
+
+ int target_bandwidth;
+ struct vpx_codec_pkt_list *output_pkt_list;
+
+#if 0
+ /* Experimental code for lagged and one pass */
+ ONEPASS_FRAMESTATS one_pass_frame_stats[MAX_LAG_BUFFERS];
+ int one_pass_frame_index;
+#endif
+
+ int decimation_factor;
+ int decimation_count;
+
+ /* for real time encoding */
+ int avg_encode_time; /* microsecond */
+ int avg_pick_mode_time; /* microsecond */
+ int Speed;
+ int compressor_speed;
+
+ int interquantizer;
+ int auto_gold;
+ int auto_adjust_gold_quantizer;
+ int auto_worst_q;
+ int cpu_used;
+ int pass;
+
+
+ int prob_intra_coded;
+ int prob_last_coded;
+ int prob_gf_coded;
+ int prob_skip_false;
+ int last_skip_false_probs[3];
+ int last_skip_probs_q[3];
+ int recent_ref_frame_usage[MAX_REF_FRAMES];
+
+ int count_mb_ref_frame_usage[MAX_REF_FRAMES];
+ int this_frame_percent_intra;
+ int last_frame_percent_intra;
+
+ int ref_frame_flags;
+
+ SPEED_FEATURES sf;
+ int error_bins[1024];
+
+ /* Data used for real time conferencing mode to help determine if it
+ * would be good to update the gf
+ */
+ int inter_zz_count;
+ /* Count ZEROMV on all reference frames. */
+ int zeromv_count;
+ int lf_zeromv_pct;
+ int gf_bad_count;
+ int gf_update_recommended;
+
+ unsigned char *segmentation_map;
+ signed char segment_feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
+ int segment_encode_breakout[MAX_MB_SEGMENTS];
+
+ unsigned char *active_map;
+ unsigned int active_map_enabled;
+
+ /* Video conferencing cyclic refresh mode flags. This is a mode
+ * designed to clean up the background over time in live encoding
+ * scenarious. It uses segmentation.
+ */
+ int cyclic_refresh_mode_enabled;
+ int cyclic_refresh_mode_max_mbs_perframe;
+ int cyclic_refresh_mode_index;
+ int cyclic_refresh_q;
+ signed char *cyclic_refresh_map;
+
+#if CONFIG_MULTITHREAD
+ /* multithread data */
+ int * mt_current_mb_col;
+ int mt_sync_range;
+ int b_multi_threaded;
+ int encoding_thread_count;
+ int b_lpf_running;
+
+ pthread_t *h_encoding_thread;
+ pthread_t h_filter_thread;
+
+ MB_ROW_COMP *mb_row_ei;
+ ENCODETHREAD_DATA *en_thread_data;
+ LPFTHREAD_DATA lpf_thread_data;
+
+ /* events */
+ sem_t *h_event_start_encoding;
+ sem_t h_event_end_encoding;
+ sem_t h_event_start_lpf;
+ sem_t h_event_end_lpf;
+#endif
+
+ TOKENLIST *tplist;
+ unsigned int partition_sz[MAX_PARTITIONS];
+ unsigned char *partition_d[MAX_PARTITIONS];
+ unsigned char *partition_d_end[MAX_PARTITIONS];
+
+
+ fractional_mv_step_fp *find_fractional_mv_step;
+ vp8_full_search_fn_t full_search_sad;
+ vp8_refining_search_fn_t refining_search_sad;
+ vp8_diamond_search_fn_t diamond_search_sad;
+ vp8_variance_fn_ptr_t fn_ptr[BLOCK_MAX_SEGMENTS];
+ uint64_t time_receive_data;
+ uint64_t time_compress_data;
+ uint64_t time_pick_lpf;
+ uint64_t time_encode_mb_row;
+
+ int base_skip_false_prob[128];
+
+ FRAME_CONTEXT lfc_n; /* last frame entropy */
+ FRAME_CONTEXT lfc_a; /* last alt ref entropy */
+ FRAME_CONTEXT lfc_g; /* last gold ref entropy */
+
+
+ struct twopass_rc
+ {
+ unsigned int section_intra_rating;
+ double section_max_qfactor;
+ unsigned int next_iiratio;
+ unsigned int this_iiratio;
+ FIRSTPASS_STATS total_stats;
+ FIRSTPASS_STATS this_frame_stats;
+ FIRSTPASS_STATS *stats_in, *stats_in_end, *stats_in_start;
+ FIRSTPASS_STATS total_left_stats;
+ int first_pass_done;
+ int64_t bits_left;
+ int64_t clip_bits_total;
+ double avg_iiratio;
+ double modified_error_total;
+ double modified_error_used;
+ double modified_error_left;
+ double kf_intra_err_min;
+ double gf_intra_err_min;
+ int frames_to_key;
+ int maxq_max_limit;
+ int maxq_min_limit;
+ int gf_decay_rate;
+ int static_scene_max_gf_interval;
+ int kf_bits;
+ /* Remaining error from uncoded frames in a gf group. */
+ int gf_group_error_left;
+ /* Projected total bits available for a key frame group of frames */
+ int64_t kf_group_bits;
+ /* Error score of frames still to be coded in kf group */
+ int64_t kf_group_error_left;
+ /* Projected Bits available for a group including 1 GF or ARF */
+ int gf_group_bits;
+ /* Bits for the golden frame or ARF */
+ int gf_bits;
+ int alt_extra_bits;
+ double est_max_qcorrection_factor;
+ } twopass;
+
+#if VP8_TEMPORAL_ALT_REF
+ YV12_BUFFER_CONFIG alt_ref_buffer;
+ YV12_BUFFER_CONFIG *frames[MAX_LAG_BUFFERS];
+ int fixed_divide[512];
+#endif
+
+#if CONFIG_INTERNAL_STATS
+ int count;
+ double total_y;
+ double total_u;
+ double total_v;
+ double total ;
+ double total_sq_error;
+ double totalp_y;
+ double totalp_u;
+ double totalp_v;
+ double totalp;
+ double total_sq_error2;
+ int bytes;
+ double summed_quality;
+ double summed_weights;
+ unsigned int tot_recode_hits;
+
+
+ double total_ssimg_y;
+ double total_ssimg_u;
+ double total_ssimg_v;
+ double total_ssimg_all;
+
+ int b_calculate_ssimg;
+#endif
+ int b_calculate_psnr;
+
+ /* Per MB activity measurement */
+ unsigned int activity_avg;
+ unsigned int * mb_activity_map;
+
+ /* Record of which MBs still refer to last golden frame either
+ * directly or through 0,0
+ */
+ unsigned char *gf_active_flags;
+ int gf_active_count;
+
+ int output_partition;
+
+ /* Store last frame's MV info for next frame MV prediction */
+ int_mv *lfmv;
+ int *lf_ref_frame_sign_bias;
+ int *lf_ref_frame;
+
+ /* force next frame to intra when kf_auto says so */
+ int force_next_frame_intra;
+
+ int droppable;
+
+#if CONFIG_TEMPORAL_DENOISING
+ VP8_DENOISER denoiser;
+#endif
+
+ /* Coding layer state variables */
+ unsigned int current_layer;
+ LAYER_CONTEXT layer_context[VPX_TS_MAX_LAYERS];
+
+ int64_t frames_in_layer[VPX_TS_MAX_LAYERS];
+ int64_t bytes_in_layer[VPX_TS_MAX_LAYERS];
+ double sum_psnr[VPX_TS_MAX_LAYERS];
+ double sum_psnr_p[VPX_TS_MAX_LAYERS];
+ double total_error2[VPX_TS_MAX_LAYERS];
+ double total_error2_p[VPX_TS_MAX_LAYERS];
+ double sum_ssim[VPX_TS_MAX_LAYERS];
+ double sum_weights[VPX_TS_MAX_LAYERS];
+
+ double total_ssimg_y_in_layer[VPX_TS_MAX_LAYERS];
+ double total_ssimg_u_in_layer[VPX_TS_MAX_LAYERS];
+ double total_ssimg_v_in_layer[VPX_TS_MAX_LAYERS];
+ double total_ssimg_all_in_layer[VPX_TS_MAX_LAYERS];
+
+#if CONFIG_MULTI_RES_ENCODING
+ /* Number of MBs per row at lower-resolution level */
+ int mr_low_res_mb_cols;
+ /* Indicate if lower-res mv info is available */
+ unsigned char mr_low_res_mv_avail;
+ /* The frame number of each reference frames */
+ unsigned int current_ref_frames[MAX_REF_FRAMES];
+#endif
+
+ struct rd_costs_struct
+ {
+ int mvcosts[2][MVvals+1];
+ int mvsadcosts[2][MVfpvals+1];
+ int mbmode_cost[2][MB_MODE_COUNT];
+ int intra_uv_mode_cost[2][MB_MODE_COUNT];
+ int bmode_costs[10][10][10];
+ int inter_bmode_costs[B_MODE_COUNT];
+ int token_costs[BLOCK_TYPES][COEF_BANDS]
+ [PREV_COEF_CONTEXTS][MAX_ENTROPY_TOKENS];
+ } rd_costs;
+} VP8_COMP;
+
+void control_data_rate(VP8_COMP *cpi);
+
+void vp8_pack_bitstream(VP8_COMP *cpi, unsigned char *dest, unsigned char *dest_end, unsigned long *size);
+
+int rd_cost_intra_mb(MACROBLOCKD *x);
+
+void vp8_tokenize_mb(VP8_COMP *, MACROBLOCK *, TOKENEXTRA **);
+
+void vp8_set_speed_features(VP8_COMP *cpi);
+
+#if CONFIG_DEBUG
+#define CHECK_MEM_ERROR(lval,expr) do {\
+ lval = (expr); \
+ if(!lval) \
+ vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,\
+ "Failed to allocate "#lval" at %s:%d", \
+ __FILE__,__LINE__);\
+ } while(0)
+#else
+#define CHECK_MEM_ERROR(lval,expr) do {\
+ lval = (expr); \
+ if(!lval) \
+ vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,\
+ "Failed to allocate "#lval);\
+ } while(0)
+#endif
+#endif
diff --git a/libvpx/vp8/encoder/pickinter.c b/libvpx/vp8/encoder/pickinter.c
new file mode 100644
index 0000000..3f09a9f
--- /dev/null
+++ b/libvpx/vp8/encoder/pickinter.c
@@ -0,0 +1,1295 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <limits.h>
+#include "vpx_config.h"
+#include "onyx_int.h"
+#include "modecosts.h"
+#include "encodeintra.h"
+#include "vp8/common/entropymode.h"
+#include "pickinter.h"
+#include "vp8/common/findnearmv.h"
+#include "encodemb.h"
+#include "vp8/common/reconinter.h"
+#include "vp8/common/reconintra4x4.h"
+#include "vp8/common/variance.h"
+#include "mcomp.h"
+#include "rdopt.h"
+#include "vpx_mem/vpx_mem.h"
+#if CONFIG_TEMPORAL_DENOISING
+#include "denoising.h"
+#endif
+
+extern int VP8_UVSSE(MACROBLOCK *x);
+
+#ifdef SPEEDSTATS
+extern unsigned int cnt_pm;
+#endif
+
+extern const int vp8_ref_frame_order[MAX_MODES];
+extern const MB_PREDICTION_MODE vp8_mode_order[MAX_MODES];
+
+extern int vp8_cost_mv_ref(MB_PREDICTION_MODE m, const int near_mv_ref_ct[4]);
+
+
+int vp8_skip_fractional_mv_step(MACROBLOCK *mb, BLOCK *b, BLOCKD *d,
+ int_mv *bestmv, int_mv *ref_mv,
+ int error_per_bit,
+ const vp8_variance_fn_ptr_t *vfp,
+ int *mvcost[2], int *distortion,
+ unsigned int *sse)
+{
+ (void) b;
+ (void) d;
+ (void) ref_mv;
+ (void) error_per_bit;
+ (void) vfp;
+ (void) mvcost;
+ (void) distortion;
+ (void) sse;
+ bestmv->as_mv.row <<= 3;
+ bestmv->as_mv.col <<= 3;
+ return 0;
+}
+
+
+int vp8_get_inter_mbpred_error(MACROBLOCK *mb,
+ const vp8_variance_fn_ptr_t *vfp,
+ unsigned int *sse,
+ int_mv this_mv)
+{
+
+ BLOCK *b = &mb->block[0];
+ BLOCKD *d = &mb->e_mbd.block[0];
+ unsigned char *what = (*(b->base_src) + b->src);
+ int what_stride = b->src_stride;
+ int pre_stride = mb->e_mbd.pre.y_stride;
+ unsigned char *in_what = mb->e_mbd.pre.y_buffer + d->offset ;
+ int in_what_stride = pre_stride;
+ int xoffset = this_mv.as_mv.col & 7;
+ int yoffset = this_mv.as_mv.row & 7;
+
+ in_what += (this_mv.as_mv.row >> 3) * pre_stride + (this_mv.as_mv.col >> 3);
+
+ if (xoffset | yoffset)
+ {
+ return vfp->svf(in_what, in_what_stride, xoffset, yoffset, what, what_stride, sse);
+ }
+ else
+ {
+ return vfp->vf(what, what_stride, in_what, in_what_stride, sse);
+ }
+
+}
+
+
+unsigned int vp8_get4x4sse_cs_c
+(
+ const unsigned char *src_ptr,
+ int source_stride,
+ const unsigned char *ref_ptr,
+ int recon_stride
+)
+{
+ int distortion = 0;
+ int r, c;
+
+ for (r = 0; r < 4; r++)
+ {
+ for (c = 0; c < 4; c++)
+ {
+ int diff = src_ptr[c] - ref_ptr[c];
+ distortion += diff * diff;
+ }
+
+ src_ptr += source_stride;
+ ref_ptr += recon_stride;
+ }
+
+ return distortion;
+}
+
+static int get_prediction_error(BLOCK *be, BLOCKD *b)
+{
+ unsigned char *sptr;
+ unsigned char *dptr;
+ sptr = (*(be->base_src) + be->src);
+ dptr = b->predictor;
+
+ return vp8_get4x4sse_cs(sptr, be->src_stride, dptr, 16);
+
+}
+
+static int pick_intra4x4block(
+ MACROBLOCK *x,
+ int ib,
+ B_PREDICTION_MODE *best_mode,
+ const int *mode_costs,
+
+ int *bestrate,
+ int *bestdistortion)
+{
+
+ BLOCKD *b = &x->e_mbd.block[ib];
+ BLOCK *be = &x->block[ib];
+ int dst_stride = x->e_mbd.dst.y_stride;
+ unsigned char *dst = x->e_mbd.dst.y_buffer + b->offset;
+ B_PREDICTION_MODE mode;
+ int best_rd = INT_MAX;
+ int rate;
+ int distortion;
+
+ unsigned char *Above = dst - dst_stride;
+ unsigned char *yleft = dst - 1;
+ unsigned char top_left = Above[-1];
+
+ for (mode = B_DC_PRED; mode <= B_HE_PRED; mode++)
+ {
+ int this_rd;
+
+ rate = mode_costs[mode];
+
+ vp8_intra4x4_predict(Above, yleft, dst_stride, mode,
+ b->predictor, 16, top_left);
+ distortion = get_prediction_error(be, b);
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
+
+ if (this_rd < best_rd)
+ {
+ *bestrate = rate;
+ *bestdistortion = distortion;
+ best_rd = this_rd;
+ *best_mode = mode;
+ }
+ }
+
+ b->bmi.as_mode = *best_mode;
+ vp8_encode_intra4x4block(x, ib);
+ return best_rd;
+}
+
+
+static int pick_intra4x4mby_modes
+(
+ MACROBLOCK *mb,
+ int *Rate,
+ int *best_dist
+)
+{
+ MACROBLOCKD *const xd = &mb->e_mbd;
+ int i;
+ int cost = mb->mbmode_cost [xd->frame_type] [B_PRED];
+ int error;
+ int distortion = 0;
+ const int *bmode_costs;
+
+ intra_prediction_down_copy(xd, xd->dst.y_buffer - xd->dst.y_stride + 16);
+
+ bmode_costs = mb->inter_bmode_costs;
+
+ for (i = 0; i < 16; i++)
+ {
+ MODE_INFO *const mic = xd->mode_info_context;
+ const int mis = xd->mode_info_stride;
+
+ B_PREDICTION_MODE UNINITIALIZED_IS_SAFE(best_mode);
+ int UNINITIALIZED_IS_SAFE(r), UNINITIALIZED_IS_SAFE(d);
+
+ if (mb->e_mbd.frame_type == KEY_FRAME)
+ {
+ const B_PREDICTION_MODE A = above_block_mode(mic, i, mis);
+ const B_PREDICTION_MODE L = left_block_mode(mic, i);
+
+ bmode_costs = mb->bmode_costs[A][L];
+ }
+
+
+ pick_intra4x4block(mb, i, &best_mode, bmode_costs, &r, &d);
+
+ cost += r;
+ distortion += d;
+ mic->bmi[i].as_mode = best_mode;
+
+ /* Break out case where we have already exceeded best so far value
+ * that was passed in
+ */
+ if (distortion > *best_dist)
+ break;
+ }
+
+ *Rate = cost;
+
+ if (i == 16)
+ {
+ *best_dist = distortion;
+ error = RDCOST(mb->rdmult, mb->rddiv, cost, distortion);
+ }
+ else
+ {
+ *best_dist = INT_MAX;
+ error = INT_MAX;
+ }
+
+ return error;
+}
+
+static void pick_intra_mbuv_mode(MACROBLOCK *mb)
+{
+
+ MACROBLOCKD *x = &mb->e_mbd;
+ unsigned char *uabove_row = x->dst.u_buffer - x->dst.uv_stride;
+ unsigned char *vabove_row = x->dst.v_buffer - x->dst.uv_stride;
+ unsigned char *usrc_ptr = (mb->block[16].src + *mb->block[16].base_src);
+ unsigned char *vsrc_ptr = (mb->block[20].src + *mb->block[20].base_src);
+ int uvsrc_stride = mb->block[16].src_stride;
+ unsigned char uleft_col[8];
+ unsigned char vleft_col[8];
+ unsigned char utop_left = uabove_row[-1];
+ unsigned char vtop_left = vabove_row[-1];
+ int i, j;
+ int expected_udc;
+ int expected_vdc;
+ int shift;
+ int Uaverage = 0;
+ int Vaverage = 0;
+ int diff;
+ int pred_error[4] = {0, 0, 0, 0}, best_error = INT_MAX;
+ MB_PREDICTION_MODE UNINITIALIZED_IS_SAFE(best_mode);
+
+
+ for (i = 0; i < 8; i++)
+ {
+ uleft_col[i] = x->dst.u_buffer [i* x->dst.uv_stride -1];
+ vleft_col[i] = x->dst.v_buffer [i* x->dst.uv_stride -1];
+ }
+
+ if (!x->up_available && !x->left_available)
+ {
+ expected_udc = 128;
+ expected_vdc = 128;
+ }
+ else
+ {
+ shift = 2;
+
+ if (x->up_available)
+ {
+
+ for (i = 0; i < 8; i++)
+ {
+ Uaverage += uabove_row[i];
+ Vaverage += vabove_row[i];
+ }
+
+ shift ++;
+
+ }
+
+ if (x->left_available)
+ {
+ for (i = 0; i < 8; i++)
+ {
+ Uaverage += uleft_col[i];
+ Vaverage += vleft_col[i];
+ }
+
+ shift ++;
+
+ }
+
+ expected_udc = (Uaverage + (1 << (shift - 1))) >> shift;
+ expected_vdc = (Vaverage + (1 << (shift - 1))) >> shift;
+ }
+
+
+ for (i = 0; i < 8; i++)
+ {
+ for (j = 0; j < 8; j++)
+ {
+
+ int predu = uleft_col[i] + uabove_row[j] - utop_left;
+ int predv = vleft_col[i] + vabove_row[j] - vtop_left;
+ int u_p, v_p;
+
+ u_p = usrc_ptr[j];
+ v_p = vsrc_ptr[j];
+
+ if (predu < 0)
+ predu = 0;
+
+ if (predu > 255)
+ predu = 255;
+
+ if (predv < 0)
+ predv = 0;
+
+ if (predv > 255)
+ predv = 255;
+
+
+ diff = u_p - expected_udc;
+ pred_error[DC_PRED] += diff * diff;
+ diff = v_p - expected_vdc;
+ pred_error[DC_PRED] += diff * diff;
+
+
+ diff = u_p - uabove_row[j];
+ pred_error[V_PRED] += diff * diff;
+ diff = v_p - vabove_row[j];
+ pred_error[V_PRED] += diff * diff;
+
+
+ diff = u_p - uleft_col[i];
+ pred_error[H_PRED] += diff * diff;
+ diff = v_p - vleft_col[i];
+ pred_error[H_PRED] += diff * diff;
+
+
+ diff = u_p - predu;
+ pred_error[TM_PRED] += diff * diff;
+ diff = v_p - predv;
+ pred_error[TM_PRED] += diff * diff;
+
+
+ }
+
+ usrc_ptr += uvsrc_stride;
+ vsrc_ptr += uvsrc_stride;
+
+ if (i == 3)
+ {
+ usrc_ptr = (mb->block[18].src + *mb->block[18].base_src);
+ vsrc_ptr = (mb->block[22].src + *mb->block[22].base_src);
+ }
+
+
+
+ }
+
+
+ for (i = DC_PRED; i <= TM_PRED; i++)
+ {
+ if (best_error > pred_error[i])
+ {
+ best_error = pred_error[i];
+ best_mode = (MB_PREDICTION_MODE)i;
+ }
+ }
+
+
+ mb->e_mbd.mode_info_context->mbmi.uv_mode = best_mode;
+
+}
+
+static void update_mvcount(VP8_COMP *cpi, MACROBLOCK *x, int_mv *best_ref_mv)
+{
+ MACROBLOCKD *xd = &x->e_mbd;
+ /* Split MV modes currently not supported when RD is nopt enabled,
+ * therefore, only need to modify MVcount in NEWMV mode. */
+ if (xd->mode_info_context->mbmi.mode == NEWMV)
+ {
+ x->MVcount[0][mv_max+((xd->mode_info_context->mbmi.mv.as_mv.row -
+ best_ref_mv->as_mv.row) >> 1)]++;
+ x->MVcount[1][mv_max+((xd->mode_info_context->mbmi.mv.as_mv.col -
+ best_ref_mv->as_mv.col) >> 1)]++;
+ }
+}
+
+
+#if CONFIG_MULTI_RES_ENCODING
+static
+void get_lower_res_motion_info(VP8_COMP *cpi, MACROBLOCKD *xd, int *dissim,
+ int *parent_ref_frame,
+ MB_PREDICTION_MODE *parent_mode,
+ int_mv *parent_ref_mv, int mb_row, int mb_col)
+{
+ LOWER_RES_MB_INFO* store_mode_info
+ = ((LOWER_RES_FRAME_INFO*)cpi->oxcf.mr_low_res_mode_info)->mb_info;
+ unsigned int parent_mb_index;
+
+ /* Consider different down_sampling_factor. */
+ {
+ /* TODO: Removed the loop that supports special down_sampling_factor
+ * such as 2, 4, 8. Will revisit it if needed.
+ * Should also try using a look-up table to see if it helps
+ * performance. */
+ int parent_mb_row, parent_mb_col;
+
+ parent_mb_row = mb_row*cpi->oxcf.mr_down_sampling_factor.den
+ /cpi->oxcf.mr_down_sampling_factor.num;
+ parent_mb_col = mb_col*cpi->oxcf.mr_down_sampling_factor.den
+ /cpi->oxcf.mr_down_sampling_factor.num;
+ parent_mb_index = parent_mb_row*cpi->mr_low_res_mb_cols + parent_mb_col;
+ }
+
+ /* Read lower-resolution mode & motion result from memory.*/
+ *parent_ref_frame = store_mode_info[parent_mb_index].ref_frame;
+ *parent_mode = store_mode_info[parent_mb_index].mode;
+ *dissim = store_mode_info[parent_mb_index].dissim;
+
+ /* For highest-resolution encoder, adjust dissim value. Lower its quality
+ * for good performance. */
+ if (cpi->oxcf.mr_encoder_id == (cpi->oxcf.mr_total_resolutions - 1))
+ *dissim>>=1;
+
+ if(*parent_ref_frame != INTRA_FRAME)
+ {
+ /* Consider different down_sampling_factor.
+ * The result can be rounded to be more precise, but it takes more time.
+ */
+ (*parent_ref_mv).as_mv.row = store_mode_info[parent_mb_index].mv.as_mv.row
+ *cpi->oxcf.mr_down_sampling_factor.num
+ /cpi->oxcf.mr_down_sampling_factor.den;
+ (*parent_ref_mv).as_mv.col = store_mode_info[parent_mb_index].mv.as_mv.col
+ *cpi->oxcf.mr_down_sampling_factor.num
+ /cpi->oxcf.mr_down_sampling_factor.den;
+
+ vp8_clamp_mv2(parent_ref_mv, xd);
+ }
+}
+#endif
+
+static void check_for_encode_breakout(unsigned int sse, MACROBLOCK* x)
+{
+ MACROBLOCKD *xd = &x->e_mbd;
+
+ unsigned int threshold = (xd->block[0].dequant[1]
+ * xd->block[0].dequant[1] >>4);
+
+ if(threshold < x->encode_breakout)
+ threshold = x->encode_breakout;
+
+ if (sse < threshold )
+ {
+ /* Check u and v to make sure skip is ok */
+ unsigned int sse2 = 0;
+
+ sse2 = VP8_UVSSE(x);
+
+ if (sse2 * 2 < x->encode_breakout)
+ x->skip = 1;
+ else
+ x->skip = 0;
+ }
+}
+
+static int evaluate_inter_mode(unsigned int* sse, int rate2, int* distortion2,
+ VP8_COMP *cpi, MACROBLOCK *x, int rd_adj)
+{
+ MB_PREDICTION_MODE this_mode = x->e_mbd.mode_info_context->mbmi.mode;
+ int_mv mv = x->e_mbd.mode_info_context->mbmi.mv;
+ int this_rd;
+ /* Exit early and don't compute the distortion if this macroblock
+ * is marked inactive. */
+ if (cpi->active_map_enabled && x->active_ptr[0] == 0)
+ {
+ *sse = 0;
+ *distortion2 = 0;
+ x->skip = 1;
+ return INT_MAX;
+ }
+
+ if((this_mode != NEWMV) ||
+ !(cpi->sf.half_pixel_search) || cpi->common.full_pixel==1)
+ *distortion2 = vp8_get_inter_mbpred_error(x,
+ &cpi->fn_ptr[BLOCK_16X16],
+ sse, mv);
+
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate2, *distortion2);
+
+ /* Adjust rd to bias to ZEROMV */
+ if(this_mode == ZEROMV)
+ {
+ /* Bias to ZEROMV on LAST_FRAME reference when it is available. */
+ if ((cpi->ref_frame_flags & VP8_LAST_FRAME &
+ cpi->common.refresh_last_frame)
+ && x->e_mbd.mode_info_context->mbmi.ref_frame != LAST_FRAME)
+ rd_adj = 100;
+
+ // rd_adj <= 100
+ this_rd = ((int64_t)this_rd) * rd_adj / 100;
+ }
+
+ check_for_encode_breakout(*sse, x);
+ return this_rd;
+}
+
+static void calculate_zeromv_rd_adjustment(VP8_COMP *cpi, MACROBLOCK *x,
+ int *rd_adjustment)
+{
+ MODE_INFO *mic = x->e_mbd.mode_info_context;
+ int_mv mv_l, mv_a, mv_al;
+ int local_motion_check = 0;
+
+ if (cpi->lf_zeromv_pct > 40)
+ {
+ /* left mb */
+ mic -= 1;
+ mv_l = mic->mbmi.mv;
+
+ if (mic->mbmi.ref_frame != INTRA_FRAME)
+ if( abs(mv_l.as_mv.row) < 8 && abs(mv_l.as_mv.col) < 8)
+ local_motion_check++;
+
+ /* above-left mb */
+ mic -= x->e_mbd.mode_info_stride;
+ mv_al = mic->mbmi.mv;
+
+ if (mic->mbmi.ref_frame != INTRA_FRAME)
+ if( abs(mv_al.as_mv.row) < 8 && abs(mv_al.as_mv.col) < 8)
+ local_motion_check++;
+
+ /* above mb */
+ mic += 1;
+ mv_a = mic->mbmi.mv;
+
+ if (mic->mbmi.ref_frame != INTRA_FRAME)
+ if( abs(mv_a.as_mv.row) < 8 && abs(mv_a.as_mv.col) < 8)
+ local_motion_check++;
+
+ if (((!x->e_mbd.mb_to_top_edge || !x->e_mbd.mb_to_left_edge)
+ && local_motion_check >0) || local_motion_check >2 )
+ *rd_adjustment = 80;
+ else if (local_motion_check > 0)
+ *rd_adjustment = 90;
+ }
+}
+
+void vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset,
+ int recon_uvoffset, int *returnrate,
+ int *returndistortion, int *returnintra, int mb_row,
+ int mb_col)
+{
+ BLOCK *b = &x->block[0];
+ BLOCKD *d = &x->e_mbd.block[0];
+ MACROBLOCKD *xd = &x->e_mbd;
+ MB_MODE_INFO best_mbmode;
+
+ int_mv best_ref_mv_sb[2];
+ int_mv mode_mv_sb[2][MB_MODE_COUNT];
+ int_mv best_ref_mv;
+ int_mv *mode_mv;
+ MB_PREDICTION_MODE this_mode;
+ int num00;
+ int mdcounts[4];
+ int best_rd = INT_MAX;
+ int rd_adjustment = 100;
+ int best_intra_rd = INT_MAX;
+ int mode_index;
+ int rate;
+ int rate2;
+ int distortion2;
+ int bestsme = INT_MAX;
+ int best_mode_index = 0;
+ unsigned int sse = INT_MAX, best_rd_sse = INT_MAX;
+#if CONFIG_TEMPORAL_DENOISING
+ unsigned int zero_mv_sse = INT_MAX, best_sse = INT_MAX;
+#endif
+
+ int_mv mvp;
+
+ int near_sadidx[8] = {0, 1, 2, 3, 4, 5, 6, 7};
+ int saddone=0;
+ /* search range got from mv_pred(). It uses step_param levels. (0-7) */
+ int sr=0;
+
+ unsigned char *plane[4][3];
+ int ref_frame_map[4];
+ int sign_bias = 0;
+
+#if CONFIG_MULTI_RES_ENCODING
+ int dissim = INT_MAX;
+ int parent_ref_frame = 0;
+ int parent_ref_valid = cpi->oxcf.mr_encoder_id && cpi->mr_low_res_mv_avail;
+ int_mv parent_ref_mv;
+ MB_PREDICTION_MODE parent_mode = 0;
+
+ if (parent_ref_valid)
+ {
+ int parent_ref_flag;
+
+ get_lower_res_motion_info(cpi, xd, &dissim, &parent_ref_frame,
+ &parent_mode, &parent_ref_mv, mb_row, mb_col);
+
+ /* TODO(jkoleszar): The references available (ref_frame_flags) to the
+ * lower res encoder should match those available to this encoder, but
+ * there seems to be a situation where this mismatch can happen in the
+ * case of frame dropping and temporal layers. For example,
+ * GOLD being disallowed in ref_frame_flags, but being returned as
+ * parent_ref_frame.
+ *
+ * In this event, take the conservative approach of disabling the
+ * lower res info for this MB.
+ */
+ parent_ref_flag = 0;
+ if (parent_ref_frame == LAST_FRAME)
+ parent_ref_flag = (cpi->ref_frame_flags & VP8_LAST_FRAME);
+ else if (parent_ref_frame == GOLDEN_FRAME)
+ parent_ref_flag = (cpi->ref_frame_flags & VP8_GOLD_FRAME);
+ else if (parent_ref_frame == ALTREF_FRAME)
+ parent_ref_flag = (cpi->ref_frame_flags & VP8_ALTR_FRAME);
+
+ //assert(!parent_ref_frame || parent_ref_flag);
+ if (parent_ref_frame && !parent_ref_flag)
+ parent_ref_valid = 0;
+ }
+#endif
+
+ mode_mv = mode_mv_sb[sign_bias];
+ best_ref_mv.as_int = 0;
+ vpx_memset(mode_mv_sb, 0, sizeof(mode_mv_sb));
+ vpx_memset(&best_mbmode, 0, sizeof(best_mbmode));
+
+ /* Setup search priorities */
+#if CONFIG_MULTI_RES_ENCODING
+ if (parent_ref_valid && parent_ref_frame && dissim < 8)
+ {
+ ref_frame_map[0] = -1;
+ ref_frame_map[1] = parent_ref_frame;
+ ref_frame_map[2] = -1;
+ ref_frame_map[3] = -1;
+ } else
+#endif
+ get_reference_search_order(cpi, ref_frame_map);
+
+ /* Check to see if there is at least 1 valid reference frame that we need
+ * to calculate near_mvs.
+ */
+ if (ref_frame_map[1] > 0)
+ {
+ sign_bias = vp8_find_near_mvs_bias(&x->e_mbd,
+ x->e_mbd.mode_info_context,
+ mode_mv_sb,
+ best_ref_mv_sb,
+ mdcounts,
+ ref_frame_map[1],
+ cpi->common.ref_frame_sign_bias);
+
+ mode_mv = mode_mv_sb[sign_bias];
+ best_ref_mv.as_int = best_ref_mv_sb[sign_bias].as_int;
+ }
+
+ get_predictor_pointers(cpi, plane, recon_yoffset, recon_uvoffset);
+
+ /* Count of the number of MBs tested so far this frame */
+ cpi->mbs_tested_so_far++;
+
+ *returnintra = INT_MAX;
+ x->skip = 0;
+
+ x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
+
+ /* If the frame has big static background and current MB is in low
+ * motion area, its mode decision is biased to ZEROMV mode.
+ */
+ calculate_zeromv_rd_adjustment(cpi, x, &rd_adjustment);
+
+ /* if we encode a new mv this is important
+ * find the best new motion vector
+ */
+ for (mode_index = 0; mode_index < MAX_MODES; mode_index++)
+ {
+ int frame_cost;
+ int this_rd = INT_MAX;
+ int this_ref_frame = ref_frame_map[vp8_ref_frame_order[mode_index]];
+
+ if (best_rd <= cpi->rd_threshes[mode_index])
+ continue;
+
+ if (this_ref_frame < 0)
+ continue;
+
+ x->e_mbd.mode_info_context->mbmi.ref_frame = this_ref_frame;
+
+ /* everything but intra */
+ if (x->e_mbd.mode_info_context->mbmi.ref_frame)
+ {
+ x->e_mbd.pre.y_buffer = plane[this_ref_frame][0];
+ x->e_mbd.pre.u_buffer = plane[this_ref_frame][1];
+ x->e_mbd.pre.v_buffer = plane[this_ref_frame][2];
+
+ if (sign_bias != cpi->common.ref_frame_sign_bias[this_ref_frame])
+ {
+ sign_bias = cpi->common.ref_frame_sign_bias[this_ref_frame];
+ mode_mv = mode_mv_sb[sign_bias];
+ best_ref_mv.as_int = best_ref_mv_sb[sign_bias].as_int;
+ }
+
+#if CONFIG_MULTI_RES_ENCODING
+ if (parent_ref_valid)
+ {
+ if (vp8_mode_order[mode_index] == NEARESTMV &&
+ mode_mv[NEARESTMV].as_int ==0)
+ continue;
+ if (vp8_mode_order[mode_index] == NEARMV &&
+ mode_mv[NEARMV].as_int ==0)
+ continue;
+
+ if (vp8_mode_order[mode_index] == NEWMV && parent_mode == ZEROMV
+ && best_ref_mv.as_int==0)
+ continue;
+ else if(vp8_mode_order[mode_index] == NEWMV && dissim==0
+ && best_ref_mv.as_int==parent_ref_mv.as_int)
+ continue;
+ }
+#endif
+ }
+
+ /* Check to see if the testing frequency for this mode is at its max
+ * If so then prevent it from being tested and increase the threshold
+ * for its testing */
+ if (cpi->mode_test_hit_counts[mode_index] &&
+ (cpi->mode_check_freq[mode_index] > 1))
+ {
+ if (cpi->mbs_tested_so_far <= (cpi->mode_check_freq[mode_index] *
+ cpi->mode_test_hit_counts[mode_index]))
+ {
+ /* Increase the threshold for coding this mode to make it less
+ * likely to be chosen */
+ cpi->rd_thresh_mult[mode_index] += 4;
+
+ if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
+ cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
+
+ cpi->rd_threshes[mode_index] =
+ (cpi->rd_baseline_thresh[mode_index] >> 7) *
+ cpi->rd_thresh_mult[mode_index];
+ continue;
+ }
+ }
+
+ /* We have now reached the point where we are going to test the current
+ * mode so increment the counter for the number of times it has been
+ * tested */
+ cpi->mode_test_hit_counts[mode_index] ++;
+
+ rate2 = 0;
+ distortion2 = 0;
+
+ this_mode = vp8_mode_order[mode_index];
+
+ x->e_mbd.mode_info_context->mbmi.mode = this_mode;
+ x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
+
+ /* Work out the cost assosciated with selecting the reference frame */
+ frame_cost =
+ x->ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
+ rate2 += frame_cost;
+
+ /* Only consider ZEROMV/ALTREF_FRAME for alt ref frame,
+ * unless ARNR filtering is enabled in which case we want
+ * an unfiltered alternative */
+ if (cpi->is_src_frame_alt_ref && (cpi->oxcf.arnr_max_frames == 0))
+ {
+ if (this_mode != ZEROMV ||
+ x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME)
+ continue;
+ }
+
+ switch (this_mode)
+ {
+ case B_PRED:
+ /* Pass best so far to pick_intra4x4mby_modes to use as breakout */
+ distortion2 = best_rd_sse;
+ pick_intra4x4mby_modes(x, &rate, &distortion2);
+
+ if (distortion2 == INT_MAX)
+ {
+ this_rd = INT_MAX;
+ }
+ else
+ {
+ rate2 += rate;
+ distortion2 = vp8_variance16x16(
+ *(b->base_src), b->src_stride,
+ x->e_mbd.predictor, 16, &sse);
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
+
+ if (this_rd < best_intra_rd)
+ {
+ best_intra_rd = this_rd;
+ *returnintra = distortion2;
+ }
+ }
+
+ break;
+
+ case SPLITMV:
+
+ /* Split MV modes currently not supported when RD is not enabled. */
+ break;
+
+ case DC_PRED:
+ case V_PRED:
+ case H_PRED:
+ case TM_PRED:
+ vp8_build_intra_predictors_mby_s(xd,
+ xd->dst.y_buffer - xd->dst.y_stride,
+ xd->dst.y_buffer - 1,
+ xd->dst.y_stride,
+ xd->predictor,
+ 16);
+ distortion2 = vp8_variance16x16
+ (*(b->base_src), b->src_stride,
+ x->e_mbd.predictor, 16, &sse);
+ rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode];
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
+
+ if (this_rd < best_intra_rd)
+ {
+ best_intra_rd = this_rd;
+ *returnintra = distortion2;
+ }
+ break;
+
+ case NEWMV:
+ {
+ int thissme;
+ int step_param;
+ int further_steps;
+ int n = 0;
+ int sadpb = x->sadperbit16;
+ int_mv mvp_full;
+
+ int col_min = ((best_ref_mv.as_mv.col+7)>>3) - MAX_FULL_PEL_VAL;
+ int row_min = ((best_ref_mv.as_mv.row+7)>>3) - MAX_FULL_PEL_VAL;
+ int col_max = (best_ref_mv.as_mv.col>>3)
+ + MAX_FULL_PEL_VAL;
+ int row_max = (best_ref_mv.as_mv.row>>3)
+ + MAX_FULL_PEL_VAL;
+
+ int tmp_col_min = x->mv_col_min;
+ int tmp_col_max = x->mv_col_max;
+ int tmp_row_min = x->mv_row_min;
+ int tmp_row_max = x->mv_row_max;
+
+ int speed_adjust = (cpi->Speed > 5) ? ((cpi->Speed >= 8)? 3 : 2) : 1;
+
+ /* Further step/diamond searches as necessary */
+ step_param = cpi->sf.first_step + speed_adjust;
+
+#if CONFIG_MULTI_RES_ENCODING
+ /* If lower-res drops this frame, then higher-res encoder does
+ motion search without any previous knowledge. Also, since
+ last frame motion info is not stored, then we can not
+ use improved_mv_pred. */
+ if (cpi->oxcf.mr_encoder_id && !parent_ref_valid)
+ cpi->sf.improved_mv_pred = 0;
+
+ if (parent_ref_valid && parent_ref_frame)
+ {
+ /* Use parent MV as predictor. Adjust search range
+ * accordingly.
+ */
+ mvp.as_int = parent_ref_mv.as_int;
+ mvp_full.as_mv.col = parent_ref_mv.as_mv.col>>3;
+ mvp_full.as_mv.row = parent_ref_mv.as_mv.row>>3;
+
+ if(dissim <=32) step_param += 3;
+ else if(dissim <=128) step_param += 2;
+ else step_param += 1;
+ }else
+#endif
+ {
+ if(cpi->sf.improved_mv_pred)
+ {
+ if(!saddone)
+ {
+ vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] );
+ saddone = 1;
+ }
+
+ vp8_mv_pred(cpi, &x->e_mbd, x->e_mbd.mode_info_context,
+ &mvp,x->e_mbd.mode_info_context->mbmi.ref_frame,
+ cpi->common.ref_frame_sign_bias, &sr,
+ &near_sadidx[0]);
+
+ sr += speed_adjust;
+ /* adjust search range according to sr from mv prediction */
+ if(sr > step_param)
+ step_param = sr;
+
+ mvp_full.as_mv.col = mvp.as_mv.col>>3;
+ mvp_full.as_mv.row = mvp.as_mv.row>>3;
+ }else
+ {
+ mvp.as_int = best_ref_mv.as_int;
+ mvp_full.as_mv.col = best_ref_mv.as_mv.col>>3;
+ mvp_full.as_mv.row = best_ref_mv.as_mv.row>>3;
+ }
+ }
+
+#if CONFIG_MULTI_RES_ENCODING
+ if (parent_ref_valid && parent_ref_frame && dissim <= 2 &&
+ MAX(abs(best_ref_mv.as_mv.row - parent_ref_mv.as_mv.row),
+ abs(best_ref_mv.as_mv.col - parent_ref_mv.as_mv.col)) <= 4)
+ {
+ d->bmi.mv.as_int = mvp_full.as_int;
+ mode_mv[NEWMV].as_int = mvp_full.as_int;
+
+ cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv, &best_ref_mv,
+ x->errorperbit,
+ &cpi->fn_ptr[BLOCK_16X16],
+ cpi->mb.mvcost,
+ &distortion2,&sse);
+ }else
+#endif
+ {
+ /* Get intersection of UMV window and valid MV window to
+ * reduce # of checks in diamond search. */
+ if (x->mv_col_min < col_min )
+ x->mv_col_min = col_min;
+ if (x->mv_col_max > col_max )
+ x->mv_col_max = col_max;
+ if (x->mv_row_min < row_min )
+ x->mv_row_min = row_min;
+ if (x->mv_row_max > row_max )
+ x->mv_row_max = row_max;
+
+ further_steps = (cpi->Speed >= 8)?
+ 0: (cpi->sf.max_step_search_steps - 1 - step_param);
+
+ if (cpi->sf.search_method == HEX)
+ {
+#if CONFIG_MULTI_RES_ENCODING
+ /* TODO: In higher-res pick_inter_mode, step_param is used to
+ * modify hex search range. Here, set step_param to 0 not to
+ * change the behavior in lowest-resolution encoder.
+ * Will improve it later.
+ */
+ /* Set step_param to 0 to ensure large-range motion search
+ when encoder drops this frame at lower-resolution.
+ */
+ if (!parent_ref_valid)
+ step_param = 0;
+#endif
+ bestsme = vp8_hex_search(x, b, d, &mvp_full, &d->bmi.mv,
+ step_param, sadpb,
+ &cpi->fn_ptr[BLOCK_16X16],
+ x->mvsadcost, x->mvcost, &best_ref_mv);
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+ }
+ else
+ {
+ bestsme = cpi->diamond_search_sad(x, b, d, &mvp_full,
+ &d->bmi.mv, step_param, sadpb, &num00,
+ &cpi->fn_ptr[BLOCK_16X16],
+ x->mvcost, &best_ref_mv);
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+
+ /* Further step/diamond searches as necessary */
+ n = num00;
+ num00 = 0;
+
+ while (n < further_steps)
+ {
+ n++;
+
+ if (num00)
+ num00--;
+ else
+ {
+ thissme =
+ cpi->diamond_search_sad(x, b, d, &mvp_full,
+ &d->bmi.mv,
+ step_param + n,
+ sadpb, &num00,
+ &cpi->fn_ptr[BLOCK_16X16],
+ x->mvcost, &best_ref_mv);
+ if (thissme < bestsme)
+ {
+ bestsme = thissme;
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+ }
+ else
+ {
+ d->bmi.mv.as_int = mode_mv[NEWMV].as_int;
+ }
+ }
+ }
+ }
+
+ x->mv_col_min = tmp_col_min;
+ x->mv_col_max = tmp_col_max;
+ x->mv_row_min = tmp_row_min;
+ x->mv_row_max = tmp_row_max;
+
+ if (bestsme < INT_MAX)
+ cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv,
+ &best_ref_mv, x->errorperbit,
+ &cpi->fn_ptr[BLOCK_16X16],
+ cpi->mb.mvcost,
+ &distortion2,&sse);
+ }
+
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+
+ /* mv cost; */
+ rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv,
+ cpi->mb.mvcost, 128);
+ }
+
+ case NEARESTMV:
+ case NEARMV:
+
+ if (mode_mv[this_mode].as_int == 0)
+ continue;
+
+ case ZEROMV:
+
+ /* Trap vectors that reach beyond the UMV borders
+ * Note that ALL New MV, Nearest MV Near MV and Zero MV code drops
+ * through to this point because of the lack of break statements
+ * in the previous two cases.
+ */
+ if (((mode_mv[this_mode].as_mv.row >> 3) < x->mv_row_min) ||
+ ((mode_mv[this_mode].as_mv.row >> 3) > x->mv_row_max) ||
+ ((mode_mv[this_mode].as_mv.col >> 3) < x->mv_col_min) ||
+ ((mode_mv[this_mode].as_mv.col >> 3) > x->mv_col_max))
+ continue;
+
+ rate2 += vp8_cost_mv_ref(this_mode, mdcounts);
+ x->e_mbd.mode_info_context->mbmi.mv.as_int =
+ mode_mv[this_mode].as_int;
+ this_rd = evaluate_inter_mode(&sse, rate2, &distortion2, cpi, x,
+ rd_adjustment);
+
+ break;
+ default:
+ break;
+ }
+
+#if CONFIG_TEMPORAL_DENOISING
+ if (cpi->oxcf.noise_sensitivity)
+ {
+
+ /* Store for later use by denoiser. */
+ if (this_mode == ZEROMV && sse < zero_mv_sse )
+ {
+ zero_mv_sse = sse;
+ x->best_zeromv_reference_frame =
+ x->e_mbd.mode_info_context->mbmi.ref_frame;
+ }
+
+ /* Store the best NEWMV in x for later use in the denoiser. */
+ if (x->e_mbd.mode_info_context->mbmi.mode == NEWMV &&
+ sse < best_sse)
+ {
+ best_sse = sse;
+ x->best_sse_inter_mode = NEWMV;
+ x->best_sse_mv = x->e_mbd.mode_info_context->mbmi.mv;
+ x->need_to_clamp_best_mvs =
+ x->e_mbd.mode_info_context->mbmi.need_to_clamp_mvs;
+ x->best_reference_frame =
+ x->e_mbd.mode_info_context->mbmi.ref_frame;
+ }
+ }
+#endif
+
+ if (this_rd < best_rd || x->skip)
+ {
+ /* Note index of best mode */
+ best_mode_index = mode_index;
+
+ *returnrate = rate2;
+ *returndistortion = distortion2;
+ best_rd_sse = sse;
+ best_rd = this_rd;
+ vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi,
+ sizeof(MB_MODE_INFO));
+
+ /* Testing this mode gave rise to an improvement in best error
+ * score. Lower threshold a bit for next time
+ */
+ cpi->rd_thresh_mult[mode_index] =
+ (cpi->rd_thresh_mult[mode_index] >= (MIN_THRESHMULT + 2)) ?
+ cpi->rd_thresh_mult[mode_index] - 2 : MIN_THRESHMULT;
+ cpi->rd_threshes[mode_index] =
+ (cpi->rd_baseline_thresh[mode_index] >> 7) *
+ cpi->rd_thresh_mult[mode_index];
+ }
+
+ /* If the mode did not help improve the best error case then raise the
+ * threshold for testing that mode next time around.
+ */
+ else
+ {
+ cpi->rd_thresh_mult[mode_index] += 4;
+
+ if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
+ cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
+
+ cpi->rd_threshes[mode_index] =
+ (cpi->rd_baseline_thresh[mode_index] >> 7) *
+ cpi->rd_thresh_mult[mode_index];
+ }
+
+ if (x->skip)
+ break;
+ }
+
+ /* Reduce the activation RD thresholds for the best choice mode */
+ if ((cpi->rd_baseline_thresh[best_mode_index] > 0) && (cpi->rd_baseline_thresh[best_mode_index] < (INT_MAX >> 2)))
+ {
+ int best_adjustment = (cpi->rd_thresh_mult[best_mode_index] >> 3);
+
+ cpi->rd_thresh_mult[best_mode_index] =
+ (cpi->rd_thresh_mult[best_mode_index]
+ >= (MIN_THRESHMULT + best_adjustment)) ?
+ cpi->rd_thresh_mult[best_mode_index] - best_adjustment :
+ MIN_THRESHMULT;
+ cpi->rd_threshes[best_mode_index] =
+ (cpi->rd_baseline_thresh[best_mode_index] >> 7) *
+ cpi->rd_thresh_mult[best_mode_index];
+ }
+
+
+ {
+ int this_rdbin = (*returndistortion >> 7);
+
+ if (this_rdbin >= 1024)
+ {
+ this_rdbin = 1023;
+ }
+
+ cpi->error_bins[this_rdbin] ++;
+ }
+
+#if CONFIG_TEMPORAL_DENOISING
+ if (cpi->oxcf.noise_sensitivity)
+ {
+ if (x->best_sse_inter_mode == DC_PRED)
+ {
+ /* No best MV found. */
+ x->best_sse_inter_mode = best_mbmode.mode;
+ x->best_sse_mv = best_mbmode.mv;
+ x->need_to_clamp_best_mvs = best_mbmode.need_to_clamp_mvs;
+ x->best_reference_frame = best_mbmode.ref_frame;
+ best_sse = best_rd_sse;
+ }
+ vp8_denoiser_denoise_mb(&cpi->denoiser, x, best_sse, zero_mv_sse,
+ recon_yoffset, recon_uvoffset);
+
+
+ /* Reevaluate ZEROMV after denoising. */
+ if (best_mbmode.ref_frame == INTRA_FRAME &&
+ x->best_zeromv_reference_frame != INTRA_FRAME)
+ {
+ int this_rd = 0;
+ int this_ref_frame = x->best_zeromv_reference_frame;
+ rate2 = x->ref_frame_cost[this_ref_frame] +
+ vp8_cost_mv_ref(ZEROMV, mdcounts);
+ distortion2 = 0;
+
+ /* set up the proper prediction buffers for the frame */
+ x->e_mbd.mode_info_context->mbmi.ref_frame = this_ref_frame;
+ x->e_mbd.pre.y_buffer = plane[this_ref_frame][0];
+ x->e_mbd.pre.u_buffer = plane[this_ref_frame][1];
+ x->e_mbd.pre.v_buffer = plane[this_ref_frame][2];
+
+ x->e_mbd.mode_info_context->mbmi.mode = ZEROMV;
+ x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
+ x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
+ this_rd = evaluate_inter_mode(&sse, rate2, &distortion2, cpi, x,
+ rd_adjustment);
+
+ if (this_rd < best_rd)
+ {
+ vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi,
+ sizeof(MB_MODE_INFO));
+ }
+ }
+
+ }
+#endif
+
+ if (cpi->is_src_frame_alt_ref &&
+ (best_mbmode.mode != ZEROMV || best_mbmode.ref_frame != ALTREF_FRAME))
+ {
+ x->e_mbd.mode_info_context->mbmi.mode = ZEROMV;
+ x->e_mbd.mode_info_context->mbmi.ref_frame = ALTREF_FRAME;
+ x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
+ x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
+ x->e_mbd.mode_info_context->mbmi.mb_skip_coeff =
+ (cpi->common.mb_no_coeff_skip);
+ x->e_mbd.mode_info_context->mbmi.partitioning = 0;
+
+ return;
+ }
+
+ /* set to the best mb mode, this copy can be skip if x->skip since it
+ * already has the right content */
+ if (!x->skip)
+ vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode,
+ sizeof(MB_MODE_INFO));
+
+ if (best_mbmode.mode <= B_PRED)
+ {
+ /* set mode_info_context->mbmi.uv_mode */
+ pick_intra_mbuv_mode(x);
+ }
+
+ if (sign_bias
+ != cpi->common.ref_frame_sign_bias[xd->mode_info_context->mbmi.ref_frame])
+ best_ref_mv.as_int = best_ref_mv_sb[!sign_bias].as_int;
+
+ update_mvcount(cpi, x, &best_ref_mv);
+}
+
+
+void vp8_pick_intra_mode(MACROBLOCK *x, int *rate_)
+{
+ int error4x4, error16x16 = INT_MAX;
+ int rate, best_rate = 0, distortion, best_sse;
+ MB_PREDICTION_MODE mode, best_mode = DC_PRED;
+ int this_rd;
+ unsigned int sse;
+ BLOCK *b = &x->block[0];
+ MACROBLOCKD *xd = &x->e_mbd;
+
+ xd->mode_info_context->mbmi.ref_frame = INTRA_FRAME;
+
+ pick_intra_mbuv_mode(x);
+
+ for (mode = DC_PRED; mode <= TM_PRED; mode ++)
+ {
+ xd->mode_info_context->mbmi.mode = mode;
+ vp8_build_intra_predictors_mby_s(xd,
+ xd->dst.y_buffer - xd->dst.y_stride,
+ xd->dst.y_buffer - 1,
+ xd->dst.y_stride,
+ xd->predictor,
+ 16);
+ distortion = vp8_variance16x16
+ (*(b->base_src), b->src_stride, xd->predictor, 16, &sse);
+ rate = x->mbmode_cost[xd->frame_type][mode];
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
+
+ if (error16x16 > this_rd)
+ {
+ error16x16 = this_rd;
+ best_mode = mode;
+ best_sse = sse;
+ best_rate = rate;
+ }
+ }
+ xd->mode_info_context->mbmi.mode = best_mode;
+
+ error4x4 = pick_intra4x4mby_modes(x, &rate,
+ &best_sse);
+ if (error4x4 < error16x16)
+ {
+ xd->mode_info_context->mbmi.mode = B_PRED;
+ best_rate = rate;
+ }
+
+ *rate_ = best_rate;
+}
diff --git a/libvpx/vp8/encoder/pickinter.h b/libvpx/vp8/encoder/pickinter.h
new file mode 100644
index 0000000..35011ca
--- /dev/null
+++ b/libvpx/vp8/encoder/pickinter.h
@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_PICKINTER_H
+#define __INC_PICKINTER_H
+#include "vpx_config.h"
+#include "vp8/common/onyxc_int.h"
+
+extern void vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset,
+ int recon_uvoffset, int *returnrate,
+ int *returndistortion, int *returnintra,
+ int mb_row, int mb_col);
+extern void vp8_pick_intra_mode(MACROBLOCK *x, int *rate);
+
+extern int vp8_get_inter_mbpred_error(MACROBLOCK *mb,
+ const vp8_variance_fn_ptr_t *vfp,
+ unsigned int *sse,
+ int_mv this_mv);
+#endif
diff --git a/libvpx/vp8/encoder/picklpf.c b/libvpx/vp8/encoder/picklpf.c
new file mode 100644
index 0000000..4121349
--- /dev/null
+++ b/libvpx/vp8/encoder/picklpf.c
@@ -0,0 +1,406 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/common/onyxc_int.h"
+#include "onyx_int.h"
+#include "quantize.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vpx_scale/vpxscale.h"
+#include "vp8/common/alloccommon.h"
+#include "vp8/common/loopfilter.h"
+#if ARCH_ARM
+#include "vpx_ports/arm.h"
+#endif
+
+extern int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest);
+
+void vp8_yv12_copy_partial_frame_c(YV12_BUFFER_CONFIG *src_ybc,
+ YV12_BUFFER_CONFIG *dst_ybc)
+{
+ unsigned char *src_y, *dst_y;
+ int yheight;
+ int ystride;
+ int yoffset;
+ int linestocopy;
+
+ yheight = src_ybc->y_height;
+ ystride = src_ybc->y_stride;
+
+ /* number of MB rows to use in partial filtering */
+ linestocopy = (yheight >> 4) / PARTIAL_FRAME_FRACTION;
+ linestocopy = linestocopy ? linestocopy << 4 : 16; /* 16 lines per MB */
+
+ /* Copy extra 4 so that full filter context is available if filtering done
+ * on the copied partial frame and not original. Partial filter does mb
+ * filtering for top row also, which can modify3 pixels above.
+ */
+ linestocopy += 4;
+ /* partial image starts at ~middle of frame (macroblock border)*/
+ yoffset = ystride * (((yheight >> 5) * 16) - 4);
+ src_y = src_ybc->y_buffer + yoffset;
+ dst_y = dst_ybc->y_buffer + yoffset;
+
+ vpx_memcpy(dst_y, src_y, ystride * linestocopy);
+}
+
+static int calc_partial_ssl_err(YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *dest)
+{
+ int i, j;
+ int Total = 0;
+ int srcoffset, dstoffset;
+ unsigned char *src = source->y_buffer;
+ unsigned char *dst = dest->y_buffer;
+
+ int linestocopy;
+
+ /* number of MB rows to use in partial filtering */
+ linestocopy = (source->y_height >> 4) / PARTIAL_FRAME_FRACTION;
+ linestocopy = linestocopy ? linestocopy << 4 : 16; /* 16 lines per MB */
+
+
+ /* partial image starts at ~middle of frame (macroblock border)*/
+ srcoffset = source->y_stride * ((dest->y_height >> 5) * 16);
+ dstoffset = dest->y_stride * ((dest->y_height >> 5) * 16);
+
+ src += srcoffset;
+ dst += dstoffset;
+
+ /* Loop through the Y plane raw and reconstruction data summing
+ * (square differences)
+ */
+ for (i = 0; i < linestocopy; i += 16)
+ {
+ for (j = 0; j < source->y_width; j += 16)
+ {
+ unsigned int sse;
+ Total += vp8_mse16x16(src + j, source->y_stride,
+ dst + j, dest->y_stride,
+ &sse);
+ }
+
+ src += 16 * source->y_stride;
+ dst += 16 * dest->y_stride;
+ }
+
+ return Total;
+}
+
+/* Enforce a minimum filter level based upon baseline Q */
+static int get_min_filter_level(VP8_COMP *cpi, int base_qindex)
+{
+ int min_filter_level;
+
+ if (cpi->source_alt_ref_active && cpi->common.refresh_golden_frame &&
+ !cpi->common.refresh_alt_ref_frame)
+ min_filter_level = 0;
+ else
+ {
+ if (base_qindex <= 6)
+ min_filter_level = 0;
+ else if (base_qindex <= 16)
+ min_filter_level = 1;
+ else
+ min_filter_level = (base_qindex / 8);
+ }
+
+ return min_filter_level;
+}
+
+/* Enforce a maximum filter level based upon baseline Q */
+static int get_max_filter_level(VP8_COMP *cpi, int base_qindex)
+{
+ /* PGW August 2006: Highest filter values almost always a bad idea */
+
+ /* jbb chg: 20100118 - not so any more with this overquant stuff allow
+ * high values with lots of intra coming in.
+ */
+ int max_filter_level = MAX_LOOP_FILTER;
+ (void)base_qindex;
+
+ if (cpi->twopass.section_intra_rating > 8)
+ max_filter_level = MAX_LOOP_FILTER * 3 / 4;
+
+ return max_filter_level;
+}
+
+void vp8cx_pick_filter_level_fast(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ int best_err = 0;
+ int filt_err = 0;
+ int min_filter_level = get_min_filter_level(cpi, cm->base_qindex);
+ int max_filter_level = get_max_filter_level(cpi, cm->base_qindex);
+ int filt_val;
+ int best_filt_val = cm->filter_level;
+ YV12_BUFFER_CONFIG * saved_frame = cm->frame_to_show;
+
+ /* Replace unfiltered frame buffer with a new one */
+ cm->frame_to_show = &cpi->pick_lf_lvl_frame;
+
+ if (cm->frame_type == KEY_FRAME)
+ cm->sharpness_level = 0;
+ else
+ cm->sharpness_level = cpi->oxcf.Sharpness;
+
+ if (cm->sharpness_level != cm->last_sharpness_level)
+ {
+ vp8_loop_filter_update_sharpness(&cm->lf_info, cm->sharpness_level);
+ cm->last_sharpness_level = cm->sharpness_level;
+ }
+
+ /* Start the search at the previous frame filter level unless it is
+ * now out of range.
+ */
+ if (cm->filter_level < min_filter_level)
+ cm->filter_level = min_filter_level;
+ else if (cm->filter_level > max_filter_level)
+ cm->filter_level = max_filter_level;
+
+ filt_val = cm->filter_level;
+ best_filt_val = filt_val;
+
+ /* Get the err using the previous frame's filter value. */
+
+ /* Copy the unfiltered / processed recon buffer to the new buffer */
+ vp8_yv12_copy_partial_frame(saved_frame, cm->frame_to_show);
+ vp8_loop_filter_partial_frame(cm, &cpi->mb.e_mbd, filt_val);
+
+ best_err = calc_partial_ssl_err(sd, cm->frame_to_show);
+
+ filt_val -= 1 + (filt_val > 10);
+
+ /* Search lower filter levels */
+ while (filt_val >= min_filter_level)
+ {
+ /* Apply the loop filter */
+ vp8_yv12_copy_partial_frame(saved_frame, cm->frame_to_show);
+ vp8_loop_filter_partial_frame(cm, &cpi->mb.e_mbd, filt_val);
+
+ /* Get the err for filtered frame */
+ filt_err = calc_partial_ssl_err(sd, cm->frame_to_show);
+
+ /* Update the best case record or exit loop. */
+ if (filt_err < best_err)
+ {
+ best_err = filt_err;
+ best_filt_val = filt_val;
+ }
+ else
+ break;
+
+ /* Adjust filter level */
+ filt_val -= 1 + (filt_val > 10);
+ }
+
+ /* Search up (note that we have already done filt_val = cm->filter_level) */
+ filt_val = cm->filter_level + 1 + (filt_val > 10);
+
+ if (best_filt_val == cm->filter_level)
+ {
+ /* Resist raising filter level for very small gains */
+ best_err -= (best_err >> 10);
+
+ while (filt_val < max_filter_level)
+ {
+ /* Apply the loop filter */
+ vp8_yv12_copy_partial_frame(saved_frame, cm->frame_to_show);
+
+ vp8_loop_filter_partial_frame(cm, &cpi->mb.e_mbd, filt_val);
+
+ /* Get the err for filtered frame */
+ filt_err = calc_partial_ssl_err(sd, cm->frame_to_show);
+
+ /* Update the best case record or exit loop. */
+ if (filt_err < best_err)
+ {
+ /* Do not raise filter level if improvement is < 1 part
+ * in 4096
+ */
+ best_err = filt_err - (filt_err >> 10);
+
+ best_filt_val = filt_val;
+ }
+ else
+ break;
+
+ /* Adjust filter level */
+ filt_val += 1 + (filt_val > 10);
+ }
+ }
+
+ cm->filter_level = best_filt_val;
+
+ if (cm->filter_level < min_filter_level)
+ cm->filter_level = min_filter_level;
+
+ if (cm->filter_level > max_filter_level)
+ cm->filter_level = max_filter_level;
+
+ /* restore unfiltered frame pointer */
+ cm->frame_to_show = saved_frame;
+}
+
+/* Stub function for now Alt LF not used */
+void vp8cx_set_alt_lf_level(VP8_COMP *cpi, int filt_val)
+{
+ MACROBLOCKD *mbd = &cpi->mb.e_mbd;
+ (void) filt_val;
+
+ mbd->segment_feature_data[MB_LVL_ALT_LF][0] = cpi->segment_feature_data[MB_LVL_ALT_LF][0];
+ mbd->segment_feature_data[MB_LVL_ALT_LF][1] = cpi->segment_feature_data[MB_LVL_ALT_LF][1];
+ mbd->segment_feature_data[MB_LVL_ALT_LF][2] = cpi->segment_feature_data[MB_LVL_ALT_LF][2];
+ mbd->segment_feature_data[MB_LVL_ALT_LF][3] = cpi->segment_feature_data[MB_LVL_ALT_LF][3];
+}
+
+void vp8cx_pick_filter_level(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ int best_err = 0;
+ int filt_err = 0;
+ int min_filter_level = get_min_filter_level(cpi, cm->base_qindex);
+ int max_filter_level = get_max_filter_level(cpi, cm->base_qindex);
+
+ int filter_step;
+ int filt_high = 0;
+ /* Start search at previous frame filter level */
+ int filt_mid = cm->filter_level;
+ int filt_low = 0;
+ int filt_best;
+ int filt_direction = 0;
+
+ /* Bias against raising loop filter and in favor of lowering it */
+ int Bias = 0;
+
+ int ss_err[MAX_LOOP_FILTER + 1];
+
+ YV12_BUFFER_CONFIG * saved_frame = cm->frame_to_show;
+
+ vpx_memset(ss_err, 0, sizeof(ss_err));
+
+ /* Replace unfiltered frame buffer with a new one */
+ cm->frame_to_show = &cpi->pick_lf_lvl_frame;
+
+ if (cm->frame_type == KEY_FRAME)
+ cm->sharpness_level = 0;
+ else
+ cm->sharpness_level = cpi->oxcf.Sharpness;
+
+ /* Start the search at the previous frame filter level unless it is
+ * now out of range.
+ */
+ filt_mid = cm->filter_level;
+
+ if (filt_mid < min_filter_level)
+ filt_mid = min_filter_level;
+ else if (filt_mid > max_filter_level)
+ filt_mid = max_filter_level;
+
+ /* Define the initial step size */
+ filter_step = (filt_mid < 16) ? 4 : filt_mid / 4;
+
+ /* Get baseline error score */
+
+ /* Copy the unfiltered / processed recon buffer to the new buffer */
+ vp8_yv12_copy_y(saved_frame, cm->frame_to_show);
+
+ vp8cx_set_alt_lf_level(cpi, filt_mid);
+ vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_mid);
+
+ best_err = vp8_calc_ss_err(sd, cm->frame_to_show);
+
+ ss_err[filt_mid] = best_err;
+
+ filt_best = filt_mid;
+
+ while (filter_step > 0)
+ {
+ Bias = (best_err >> (15 - (filt_mid / 8))) * filter_step;
+
+ if (cpi->twopass.section_intra_rating < 20)
+ Bias = Bias * cpi->twopass.section_intra_rating / 20;
+
+ filt_high = ((filt_mid + filter_step) > max_filter_level) ? max_filter_level : (filt_mid + filter_step);
+ filt_low = ((filt_mid - filter_step) < min_filter_level) ? min_filter_level : (filt_mid - filter_step);
+
+ if ((filt_direction <= 0) && (filt_low != filt_mid))
+ {
+ if(ss_err[filt_low] == 0)
+ {
+ /* Get Low filter error score */
+ vp8_yv12_copy_y(saved_frame, cm->frame_to_show);
+ vp8cx_set_alt_lf_level(cpi, filt_low);
+ vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_low);
+
+ filt_err = vp8_calc_ss_err(sd, cm->frame_to_show);
+ ss_err[filt_low] = filt_err;
+ }
+ else
+ filt_err = ss_err[filt_low];
+
+ /* If value is close to the best so far then bias towards a
+ * lower loop filter value.
+ */
+ if ((filt_err - Bias) < best_err)
+ {
+ /* Was it actually better than the previous best? */
+ if (filt_err < best_err)
+ best_err = filt_err;
+
+ filt_best = filt_low;
+ }
+ }
+
+ /* Now look at filt_high */
+ if ((filt_direction >= 0) && (filt_high != filt_mid))
+ {
+ if(ss_err[filt_high] == 0)
+ {
+ vp8_yv12_copy_y(saved_frame, cm->frame_to_show);
+ vp8cx_set_alt_lf_level(cpi, filt_high);
+ vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_high);
+
+ filt_err = vp8_calc_ss_err(sd, cm->frame_to_show);
+ ss_err[filt_high] = filt_err;
+ }
+ else
+ filt_err = ss_err[filt_high];
+
+ /* Was it better than the previous best? */
+ if (filt_err < (best_err - Bias))
+ {
+ best_err = filt_err;
+ filt_best = filt_high;
+ }
+ }
+
+ /* Half the step distance if the best filter value was the same
+ * as last time
+ */
+ if (filt_best == filt_mid)
+ {
+ filter_step = filter_step / 2;
+ filt_direction = 0;
+ }
+ else
+ {
+ filt_direction = (filt_best < filt_mid) ? -1 : 1;
+ filt_mid = filt_best;
+ }
+ }
+
+ cm->filter_level = filt_best;
+
+ /* restore unfiltered frame pointer */
+ cm->frame_to_show = saved_frame;
+}
diff --git a/libvpx/vp8/encoder/ppc/csystemdependent.c b/libvpx/vp8/encoder/ppc/csystemdependent.c
new file mode 100644
index 0000000..63f2357
--- /dev/null
+++ b/libvpx/vp8/encoder/ppc/csystemdependent.c
@@ -0,0 +1,160 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/encoder/variance.h"
+#include "vp8/encoder/onyx_int.h"
+
+SADFunction *vp8_sad16x16;
+SADFunction *vp8_sad16x8;
+SADFunction *vp8_sad8x16;
+SADFunction *vp8_sad8x8;
+SADFunction *vp8_sad4x4;
+
+variance_function *vp8_variance4x4;
+variance_function *vp8_variance8x8;
+variance_function *vp8_variance8x16;
+variance_function *vp8_variance16x8;
+variance_function *vp8_variance16x16;
+
+variance_function *vp8_mse16x16;
+
+sub_pixel_variance_function *vp8_sub_pixel_variance4x4;
+sub_pixel_variance_function *vp8_sub_pixel_variance8x8;
+sub_pixel_variance_function *vp8_sub_pixel_variance8x16;
+sub_pixel_variance_function *vp8_sub_pixel_variance16x8;
+sub_pixel_variance_function *vp8_sub_pixel_variance16x16;
+
+int (*vp8_block_error)(short *coeff, short *dqcoeff);
+int (*vp8_mbblock_error)(MACROBLOCK *mb, int dc);
+
+int (*vp8_mbuverror)(MACROBLOCK *mb);
+unsigned int (*vp8_get_mb_ss)(short *);
+void (*vp8_short_fdct4x4)(short *input, short *output, int pitch);
+void (*vp8_short_fdct8x4)(short *input, short *output, int pitch);
+void (*vp8_fast_fdct4x4)(short *input, short *output, int pitch);
+void (*vp8_fast_fdct8x4)(short *input, short *output, int pitch);
+void (*short_walsh4x4)(short *input, short *output, int pitch);
+
+void (*vp8_subtract_b)(BLOCK *be, BLOCKD *bd, int pitch);
+void (*vp8_subtract_mby)(short *diff, unsigned char *src, unsigned char *pred, int stride);
+void (*vp8_subtract_mbuv)(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride);
+void (*vp8_fast_quantize_b)(BLOCK *b, BLOCKD *d);
+
+unsigned int (*vp8_get4x4sse_cs)(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride);
+
+// c imports
+extern int block_error_c(short *coeff, short *dqcoeff);
+extern int vp8_mbblock_error_c(MACROBLOCK *mb, int dc);
+
+extern int vp8_mbuverror_c(MACROBLOCK *mb);
+extern unsigned int vp8_get8x8var_c(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
+extern void short_fdct4x4_c(short *input, short *output, int pitch);
+extern void short_fdct8x4_c(short *input, short *output, int pitch);
+extern void vp8_short_walsh4x4_c(short *input, short *output, int pitch);
+
+extern void vp8_subtract_b_c(BLOCK *be, BLOCKD *bd, int pitch);
+extern void subtract_mby_c(short *diff, unsigned char *src, unsigned char *pred, int stride);
+extern void subtract_mbuv_c(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride);
+extern void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d);
+
+extern SADFunction sad16x16_c;
+extern SADFunction sad16x8_c;
+extern SADFunction sad8x16_c;
+extern SADFunction sad8x8_c;
+extern SADFunction sad4x4_c;
+
+extern variance_function variance16x16_c;
+extern variance_function variance8x16_c;
+extern variance_function variance16x8_c;
+extern variance_function variance8x8_c;
+extern variance_function variance4x4_c;
+extern variance_function mse16x16_c;
+
+extern sub_pixel_variance_function sub_pixel_variance4x4_c;
+extern sub_pixel_variance_function sub_pixel_variance8x8_c;
+extern sub_pixel_variance_function sub_pixel_variance8x16_c;
+extern sub_pixel_variance_function sub_pixel_variance16x8_c;
+extern sub_pixel_variance_function sub_pixel_variance16x16_c;
+
+extern unsigned int vp8_get_mb_ss_c(short *);
+extern unsigned int vp8_get4x4sse_cs_c(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride);
+
+// ppc
+extern int vp8_block_error_ppc(short *coeff, short *dqcoeff);
+
+extern void vp8_short_fdct4x4_ppc(short *input, short *output, int pitch);
+extern void vp8_short_fdct8x4_ppc(short *input, short *output, int pitch);
+
+extern void vp8_subtract_mby_ppc(short *diff, unsigned char *src, unsigned char *pred, int stride);
+extern void vp8_subtract_mbuv_ppc(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride);
+
+extern SADFunction vp8_sad16x16_ppc;
+extern SADFunction vp8_sad16x8_ppc;
+extern SADFunction vp8_sad8x16_ppc;
+extern SADFunction vp8_sad8x8_ppc;
+extern SADFunction vp8_sad4x4_ppc;
+
+extern variance_function vp8_variance16x16_ppc;
+extern variance_function vp8_variance8x16_ppc;
+extern variance_function vp8_variance16x8_ppc;
+extern variance_function vp8_variance8x8_ppc;
+extern variance_function vp8_variance4x4_ppc;
+extern variance_function vp8_mse16x16_ppc;
+
+extern sub_pixel_variance_function vp8_sub_pixel_variance4x4_ppc;
+extern sub_pixel_variance_function vp8_sub_pixel_variance8x8_ppc;
+extern sub_pixel_variance_function vp8_sub_pixel_variance8x16_ppc;
+extern sub_pixel_variance_function vp8_sub_pixel_variance16x8_ppc;
+extern sub_pixel_variance_function vp8_sub_pixel_variance16x16_ppc;
+
+extern unsigned int vp8_get8x8var_ppc(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
+extern unsigned int vp8_get16x16var_ppc(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
+
+void vp8_cmachine_specific_config(void)
+{
+ // Pure C:
+ vp8_mbuverror = vp8_mbuverror_c;
+ vp8_fast_quantize_b = vp8_fast_quantize_b_c;
+ vp8_short_fdct4x4 = vp8_short_fdct4x4_ppc;
+ vp8_short_fdct8x4 = vp8_short_fdct8x4_ppc;
+ vp8_fast_fdct4x4 = vp8_short_fdct4x4_ppc;
+ vp8_fast_fdct8x4 = vp8_short_fdct8x4_ppc;
+ short_walsh4x4 = vp8_short_walsh4x4_c;
+
+ vp8_variance4x4 = vp8_variance4x4_ppc;
+ vp8_variance8x8 = vp8_variance8x8_ppc;
+ vp8_variance8x16 = vp8_variance8x16_ppc;
+ vp8_variance16x8 = vp8_variance16x8_ppc;
+ vp8_variance16x16 = vp8_variance16x16_ppc;
+ vp8_mse16x16 = vp8_mse16x16_ppc;
+
+ vp8_sub_pixel_variance4x4 = vp8_sub_pixel_variance4x4_ppc;
+ vp8_sub_pixel_variance8x8 = vp8_sub_pixel_variance8x8_ppc;
+ vp8_sub_pixel_variance8x16 = vp8_sub_pixel_variance8x16_ppc;
+ vp8_sub_pixel_variance16x8 = vp8_sub_pixel_variance16x8_ppc;
+ vp8_sub_pixel_variance16x16 = vp8_sub_pixel_variance16x16_ppc;
+
+ vp8_get_mb_ss = vp8_get_mb_ss_c;
+ vp8_get4x4sse_cs = vp8_get4x4sse_cs_c;
+
+ vp8_sad16x16 = vp8_sad16x16_ppc;
+ vp8_sad16x8 = vp8_sad16x8_ppc;
+ vp8_sad8x16 = vp8_sad8x16_ppc;
+ vp8_sad8x8 = vp8_sad8x8_ppc;
+ vp8_sad4x4 = vp8_sad4x4_ppc;
+
+ vp8_block_error = vp8_block_error_ppc;
+ vp8_mbblock_error = vp8_mbblock_error_c;
+
+ vp8_subtract_b = vp8_subtract_b_c;
+ vp8_subtract_mby = vp8_subtract_mby_ppc;
+ vp8_subtract_mbuv = vp8_subtract_mbuv_ppc;
+}
diff --git a/vp8/encoder/ppc/encodemb_altivec.asm b/libvpx/vp8/encoder/ppc/encodemb_altivec.asm
similarity index 100%
rename from vp8/encoder/ppc/encodemb_altivec.asm
rename to libvpx/vp8/encoder/ppc/encodemb_altivec.asm
diff --git a/vp8/encoder/ppc/fdct_altivec.asm b/libvpx/vp8/encoder/ppc/fdct_altivec.asm
similarity index 100%
rename from vp8/encoder/ppc/fdct_altivec.asm
rename to libvpx/vp8/encoder/ppc/fdct_altivec.asm
diff --git a/vp8/encoder/ppc/rdopt_altivec.asm b/libvpx/vp8/encoder/ppc/rdopt_altivec.asm
similarity index 100%
rename from vp8/encoder/ppc/rdopt_altivec.asm
rename to libvpx/vp8/encoder/ppc/rdopt_altivec.asm
diff --git a/libvpx/vp8/encoder/psnr.c b/libvpx/vp8/encoder/psnr.c
new file mode 100644
index 0000000..5bb49ad
--- /dev/null
+++ b/libvpx/vp8/encoder/psnr.c
@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_scale/yv12config.h"
+#include "math.h"
+#include "vp8/common/systemdependent.h" /* for vp8_clear_system_state() */
+
+#define MAX_PSNR 60
+
+double vp8_mse2psnr(double Samples, double Peak, double Mse)
+{
+ double psnr;
+
+ if ((double)Mse > 0.0)
+ psnr = 10.0 * log10(Peak * Peak * Samples / Mse);
+ else
+ psnr = MAX_PSNR; /* Limit to prevent / 0 */
+
+ if (psnr > MAX_PSNR)
+ psnr = MAX_PSNR;
+
+ return psnr;
+}
diff --git a/vp8/encoder/psnr.h b/libvpx/vp8/encoder/psnr.h
similarity index 100%
rename from vp8/encoder/psnr.h
rename to libvpx/vp8/encoder/psnr.h
diff --git a/libvpx/vp8/encoder/quantize.c b/libvpx/vp8/encoder/quantize.c
new file mode 100644
index 0000000..88fea11
--- /dev/null
+++ b/libvpx/vp8/encoder/quantize.c
@@ -0,0 +1,811 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <math.h>
+#include "vpx_mem/vpx_mem.h"
+
+#include "onyx_int.h"
+#include "quantize.h"
+#include "vp8/common/quant_common.h"
+
+#define EXACT_QUANT
+
+#ifdef EXACT_FASTQUANT
+void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d)
+{
+ int i, rc, eob;
+ int zbin;
+ int x, y, z, sz;
+ short *coeff_ptr = b->coeff;
+ short *zbin_ptr = b->zbin;
+ short *round_ptr = b->round;
+ short *quant_ptr = b->quant_fast;
+ unsigned char *quant_shift_ptr = b->quant_shift;
+ short *qcoeff_ptr = d->qcoeff;
+ short *dqcoeff_ptr = d->dqcoeff;
+ short *dequant_ptr = d->dequant;
+
+ vpx_memset(qcoeff_ptr, 0, 32);
+ vpx_memset(dqcoeff_ptr, 0, 32);
+
+ eob = -1;
+
+ for (i = 0; i < 16; i++)
+ {
+ rc = vp8_default_zig_zag1d[i];
+ z = coeff_ptr[rc];
+ zbin = zbin_ptr[rc] ;
+
+ sz = (z >> 31); /* sign of z */
+ x = (z ^ sz) - sz; /* x = abs(z) */
+
+ if (x >= zbin)
+ {
+ x += round_ptr[rc];
+ y = (((x * quant_ptr[rc]) >> 16) + x)
+ >> quant_shift_ptr[rc]; /* quantize (x) */
+ x = (y ^ sz) - sz; /* get the sign back */
+ qcoeff_ptr[rc] = x; /* write to destination */
+ dqcoeff_ptr[rc] = x * dequant_ptr[rc]; /* dequantized value */
+
+ if (y)
+ {
+ eob = i; /* last nonzero coeffs */
+ }
+ }
+ }
+ *d->eob = (char)(eob + 1);
+}
+
+#else
+
+void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d)
+{
+ int i, rc, eob;
+ int x, y, z, sz;
+ short *coeff_ptr = b->coeff;
+ short *round_ptr = b->round;
+ short *quant_ptr = b->quant_fast;
+ short *qcoeff_ptr = d->qcoeff;
+ short *dqcoeff_ptr = d->dqcoeff;
+ short *dequant_ptr = d->dequant;
+
+ eob = -1;
+ for (i = 0; i < 16; i++)
+ {
+ rc = vp8_default_zig_zag1d[i];
+ z = coeff_ptr[rc];
+
+ sz = (z >> 31); /* sign of z */
+ x = (z ^ sz) - sz; /* x = abs(z) */
+
+ y = ((x + round_ptr[rc]) * quant_ptr[rc]) >> 16; /* quantize (x) */
+ x = (y ^ sz) - sz; /* get the sign back */
+ qcoeff_ptr[rc] = x; /* write to destination */
+ dqcoeff_ptr[rc] = x * dequant_ptr[rc]; /* dequantized value */
+
+ if (y)
+ {
+ eob = i; /* last nonzero coeffs */
+ }
+ }
+ *d->eob = (char)(eob + 1);
+}
+
+#endif
+
+#ifdef EXACT_QUANT
+void vp8_regular_quantize_b_c(BLOCK *b, BLOCKD *d)
+{
+ int i, rc, eob;
+ int zbin;
+ int x, y, z, sz;
+ short *zbin_boost_ptr = b->zrun_zbin_boost;
+ short *coeff_ptr = b->coeff;
+ short *zbin_ptr = b->zbin;
+ short *round_ptr = b->round;
+ short *quant_ptr = b->quant;
+ unsigned char *quant_shift_ptr = b->quant_shift;
+ short *qcoeff_ptr = d->qcoeff;
+ short *dqcoeff_ptr = d->dqcoeff;
+ short *dequant_ptr = d->dequant;
+ short zbin_oq_value = b->zbin_extra;
+
+ vpx_memset(qcoeff_ptr, 0, 32);
+ vpx_memset(dqcoeff_ptr, 0, 32);
+
+ eob = -1;
+
+ for (i = 0; i < 16; i++)
+ {
+ rc = vp8_default_zig_zag1d[i];
+ z = coeff_ptr[rc];
+
+ zbin = zbin_ptr[rc] + *zbin_boost_ptr + zbin_oq_value;
+
+ zbin_boost_ptr ++;
+ sz = (z >> 31); /* sign of z */
+ x = (z ^ sz) - sz; /* x = abs(z) */
+
+ if (x >= zbin)
+ {
+ x += round_ptr[rc];
+ y = (((x * quant_ptr[rc]) >> 16) + x)
+ >> quant_shift_ptr[rc]; /* quantize (x) */
+ x = (y ^ sz) - sz; /* get the sign back */
+ qcoeff_ptr[rc] = x; /* write to destination */
+ dqcoeff_ptr[rc] = x * dequant_ptr[rc]; /* dequantized value */
+
+ if (y)
+ {
+ eob = i; /* last nonzero coeffs */
+ zbin_boost_ptr = b->zrun_zbin_boost; /* reset zero runlength */
+ }
+ }
+ }
+
+ *d->eob = (char)(eob + 1);
+}
+
+/* Perform regular quantization, with unbiased rounding and no zero bin. */
+void vp8_strict_quantize_b_c(BLOCK *b, BLOCKD *d)
+{
+ int i;
+ int rc;
+ int eob;
+ int x;
+ int y;
+ int z;
+ int sz;
+ short *coeff_ptr;
+ short *quant_ptr;
+ unsigned char *quant_shift_ptr;
+ short *qcoeff_ptr;
+ short *dqcoeff_ptr;
+ short *dequant_ptr;
+
+ coeff_ptr = b->coeff;
+ quant_ptr = b->quant;
+ quant_shift_ptr = b->quant_shift;
+ qcoeff_ptr = d->qcoeff;
+ dqcoeff_ptr = d->dqcoeff;
+ dequant_ptr = d->dequant;
+ eob = - 1;
+ vpx_memset(qcoeff_ptr, 0, 32);
+ vpx_memset(dqcoeff_ptr, 0, 32);
+ for (i = 0; i < 16; i++)
+ {
+ int dq;
+ int round;
+
+ /*TODO: These arrays should be stored in zig-zag order.*/
+ rc = vp8_default_zig_zag1d[i];
+ z = coeff_ptr[rc];
+ dq = dequant_ptr[rc];
+ round = dq >> 1;
+ /* Sign of z. */
+ sz = -(z < 0);
+ x = (z + sz) ^ sz;
+ x += round;
+ if (x >= dq)
+ {
+ /* Quantize x. */
+ y = (((x * quant_ptr[rc]) >> 16) + x) >> quant_shift_ptr[rc];
+ /* Put the sign back. */
+ x = (y + sz) ^ sz;
+ /* Save the coefficient and its dequantized value. */
+ qcoeff_ptr[rc] = x;
+ dqcoeff_ptr[rc] = x * dq;
+ /* Remember the last non-zero coefficient. */
+ if (y)
+ eob = i;
+ }
+ }
+
+ *d->eob = (char)(eob + 1);
+}
+
+#else
+
+void vp8_regular_quantize_b_c(BLOCK *b, BLOCKD *d)
+{
+ int i, rc, eob;
+ int zbin;
+ int x, y, z, sz;
+ short *zbin_boost_ptr = b->zrun_zbin_boost;
+ short *coeff_ptr = b->coeff;
+ short *zbin_ptr = b->zbin;
+ short *round_ptr = b->round;
+ short *quant_ptr = b->quant;
+ short *qcoeff_ptr = d->qcoeff;
+ short *dqcoeff_ptr = d->dqcoeff;
+ short *dequant_ptr = d->dequant;
+ short zbin_oq_value = b->zbin_extra;
+
+ vpx_memset(qcoeff_ptr, 0, 32);
+ vpx_memset(dqcoeff_ptr, 0, 32);
+
+ eob = -1;
+
+ for (i = 0; i < 16; i++)
+ {
+ rc = vp8_default_zig_zag1d[i];
+ z = coeff_ptr[rc];
+
+ zbin = zbin_ptr[rc] + *zbin_boost_ptr + zbin_oq_value;
+
+ zbin_boost_ptr ++;
+ sz = (z >> 31); /* sign of z */
+ x = (z ^ sz) - sz; /* x = abs(z) */
+
+ if (x >= zbin)
+ {
+ y = ((x + round_ptr[rc]) * quant_ptr[rc]) >> 16; /* quantize (x) */
+ x = (y ^ sz) - sz; /* get the sign back */
+ qcoeff_ptr[rc] = x; /* write to destination */
+ dqcoeff_ptr[rc] = x * dequant_ptr[rc]; /* dequantized value */
+
+ if (y)
+ {
+ eob = i; /* last nonzero coeffs */
+ zbin_boost_ptr = &b->zrun_zbin_boost[0]; /* reset zrl */
+ }
+ }
+ }
+
+ *d->eob = (char)(eob + 1);
+}
+
+#endif
+
+void vp8_quantize_mby_c(MACROBLOCK *x)
+{
+ int i;
+ int has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED
+ && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
+
+ for (i = 0; i < 16; i++)
+ x->quantize_b(&x->block[i], &x->e_mbd.block[i]);
+
+ if(has_2nd_order)
+ x->quantize_b(&x->block[24], &x->e_mbd.block[24]);
+}
+
+void vp8_quantize_mb_c(MACROBLOCK *x)
+{
+ int i;
+ int has_2nd_order=(x->e_mbd.mode_info_context->mbmi.mode != B_PRED
+ && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
+
+ for (i = 0; i < 24+has_2nd_order; i++)
+ x->quantize_b(&x->block[i], &x->e_mbd.block[i]);
+}
+
+
+void vp8_quantize_mbuv_c(MACROBLOCK *x)
+{
+ int i;
+
+ for (i = 16; i < 24; i++)
+ x->quantize_b(&x->block[i], &x->e_mbd.block[i]);
+}
+
+/* quantize_b_pair function pointer in MACROBLOCK structure is set to one of
+ * these two C functions if corresponding optimized routine is not available.
+ * NEON optimized version implements currently the fast quantization for pair
+ * of blocks. */
+void vp8_regular_quantize_b_pair(BLOCK *b1, BLOCK *b2, BLOCKD *d1, BLOCKD *d2)
+{
+ vp8_regular_quantize_b(b1, d1);
+ vp8_regular_quantize_b(b2, d2);
+}
+
+void vp8_fast_quantize_b_pair_c(BLOCK *b1, BLOCK *b2, BLOCKD *d1, BLOCKD *d2)
+{
+ vp8_fast_quantize_b_c(b1, d1);
+ vp8_fast_quantize_b_c(b2, d2);
+}
+
+
+static const int qrounding_factors[129] =
+{
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48
+};
+
+
+static const int qzbin_factors[129] =
+{
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80
+};
+
+
+static const int qrounding_factors_y2[129] =
+{
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48, 48, 48, 48, 48, 48, 48, 48,
+ 48
+};
+
+
+static const int qzbin_factors_y2[129] =
+{
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 84, 84, 84, 84, 84, 84, 84, 84,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80, 80, 80, 80, 80, 80, 80, 80,
+ 80
+};
+
+
+#define EXACT_QUANT
+#ifdef EXACT_QUANT
+static void invert_quant(int improved_quant, short *quant,
+ unsigned char *shift, short d)
+{
+ if(improved_quant)
+ {
+ unsigned t;
+ int l;
+ t = d;
+ for(l = 0; t > 1; l++)
+ t>>=1;
+ t = 1 + (1<<(16+l))/d;
+ *quant = (short)(t - (1<<16));
+ *shift = l;
+ }
+ else
+ {
+ *quant = (1 << 16) / d;
+ *shift = 0;
+ }
+}
+
+
+void vp8cx_init_quantizer(VP8_COMP *cpi)
+{
+ int i;
+ int quant_val;
+ int Q;
+
+ int zbin_boost[16] = {0, 0, 8, 10, 12, 14, 16, 20, 24, 28, 32, 36, 40, 44,
+ 44, 44};
+
+ for (Q = 0; Q < QINDEX_RANGE; Q++)
+ {
+ /* dc values */
+ quant_val = vp8_dc_quant(Q, cpi->common.y1dc_delta_q);
+ cpi->Y1quant_fast[Q][0] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->Y1quant[Q] + 0,
+ cpi->Y1quant_shift[Q] + 0, quant_val);
+ cpi->Y1zbin[Q][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
+ cpi->Y1round[Q][0] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.Y1dequant[Q][0] = quant_val;
+ cpi->zrun_zbin_boost_y1[Q][0] = (quant_val * zbin_boost[0]) >> 7;
+
+ quant_val = vp8_dc2quant(Q, cpi->common.y2dc_delta_q);
+ cpi->Y2quant_fast[Q][0] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->Y2quant[Q] + 0,
+ cpi->Y2quant_shift[Q] + 0, quant_val);
+ cpi->Y2zbin[Q][0] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
+ cpi->Y2round[Q][0] = (qrounding_factors_y2[Q] * quant_val) >> 7;
+ cpi->common.Y2dequant[Q][0] = quant_val;
+ cpi->zrun_zbin_boost_y2[Q][0] = (quant_val * zbin_boost[0]) >> 7;
+
+ quant_val = vp8_dc_uv_quant(Q, cpi->common.uvdc_delta_q);
+ cpi->UVquant_fast[Q][0] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->UVquant[Q] + 0,
+ cpi->UVquant_shift[Q] + 0, quant_val);
+ cpi->UVzbin[Q][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;;
+ cpi->UVround[Q][0] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.UVdequant[Q][0] = quant_val;
+ cpi->zrun_zbin_boost_uv[Q][0] = (quant_val * zbin_boost[0]) >> 7;
+
+ /* all the ac values = ; */
+ quant_val = vp8_ac_yquant(Q);
+ cpi->Y1quant_fast[Q][1] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->Y1quant[Q] + 1,
+ cpi->Y1quant_shift[Q] + 1, quant_val);
+ cpi->Y1zbin[Q][1] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
+ cpi->Y1round[Q][1] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.Y1dequant[Q][1] = quant_val;
+ cpi->zrun_zbin_boost_y1[Q][1] = (quant_val * zbin_boost[1]) >> 7;
+
+ quant_val = vp8_ac2quant(Q, cpi->common.y2ac_delta_q);
+ cpi->Y2quant_fast[Q][1] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->Y2quant[Q] + 1,
+ cpi->Y2quant_shift[Q] + 1, quant_val);
+ cpi->Y2zbin[Q][1] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
+ cpi->Y2round[Q][1] = (qrounding_factors_y2[Q] * quant_val) >> 7;
+ cpi->common.Y2dequant[Q][1] = quant_val;
+ cpi->zrun_zbin_boost_y2[Q][1] = (quant_val * zbin_boost[1]) >> 7;
+
+ quant_val = vp8_ac_uv_quant(Q, cpi->common.uvac_delta_q);
+ cpi->UVquant_fast[Q][1] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->UVquant[Q] + 1,
+ cpi->UVquant_shift[Q] + 1, quant_val);
+ cpi->UVzbin[Q][1] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
+ cpi->UVround[Q][1] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.UVdequant[Q][1] = quant_val;
+ cpi->zrun_zbin_boost_uv[Q][1] = (quant_val * zbin_boost[1]) >> 7;
+
+ for (i = 2; i < 16; i++)
+ {
+ cpi->Y1quant_fast[Q][i] = cpi->Y1quant_fast[Q][1];
+ cpi->Y1quant[Q][i] = cpi->Y1quant[Q][1];
+ cpi->Y1quant_shift[Q][i] = cpi->Y1quant_shift[Q][1];
+ cpi->Y1zbin[Q][i] = cpi->Y1zbin[Q][1];
+ cpi->Y1round[Q][i] = cpi->Y1round[Q][1];
+ cpi->zrun_zbin_boost_y1[Q][i] = (cpi->common.Y1dequant[Q][1] *
+ zbin_boost[i]) >> 7;
+
+ cpi->Y2quant_fast[Q][i] = cpi->Y2quant_fast[Q][1];
+ cpi->Y2quant[Q][i] = cpi->Y2quant[Q][1];
+ cpi->Y2quant_shift[Q][i] = cpi->Y2quant_shift[Q][1];
+ cpi->Y2zbin[Q][i] = cpi->Y2zbin[Q][1];
+ cpi->Y2round[Q][i] = cpi->Y2round[Q][1];
+ cpi->zrun_zbin_boost_y2[Q][i] = (cpi->common.Y2dequant[Q][1] *
+ zbin_boost[i]) >> 7;
+
+ cpi->UVquant_fast[Q][i] = cpi->UVquant_fast[Q][1];
+ cpi->UVquant[Q][i] = cpi->UVquant[Q][1];
+ cpi->UVquant_shift[Q][i] = cpi->UVquant_shift[Q][1];
+ cpi->UVzbin[Q][i] = cpi->UVzbin[Q][1];
+ cpi->UVround[Q][i] = cpi->UVround[Q][1];
+ cpi->zrun_zbin_boost_uv[Q][i] = (cpi->common.UVdequant[Q][1] *
+ zbin_boost[i]) >> 7;
+ }
+ }
+}
+#else
+void vp8cx_init_quantizer(VP8_COMP *cpi)
+{
+ int i;
+ int quant_val;
+ int Q;
+
+ int zbin_boost[16] = {0, 0, 8, 10, 12, 14, 16, 20, 24, 28, 32, 36, 40, 44, 44, 44};
+
+ for (Q = 0; Q < QINDEX_RANGE; Q++)
+ {
+ /* dc values */
+ quant_val = vp8_dc_quant(Q, cpi->common.y1dc_delta_q);
+ cpi->Y1quant[Q][0] = (1 << 16) / quant_val;
+ cpi->Y1zbin[Q][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
+ cpi->Y1round[Q][0] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.Y1dequant[Q][0] = quant_val;
+ cpi->zrun_zbin_boost_y1[Q][0] = (quant_val * zbin_boost[0]) >> 7;
+
+ quant_val = vp8_dc2quant(Q, cpi->common.y2dc_delta_q);
+ cpi->Y2quant[Q][0] = (1 << 16) / quant_val;
+ cpi->Y2zbin[Q][0] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
+ cpi->Y2round[Q][0] = (qrounding_factors_y2[Q] * quant_val) >> 7;
+ cpi->common.Y2dequant[Q][0] = quant_val;
+ cpi->zrun_zbin_boost_y2[Q][0] = (quant_val * zbin_boost[0]) >> 7;
+
+ quant_val = vp8_dc_uv_quant(Q, cpi->common.uvdc_delta_q);
+ cpi->UVquant[Q][0] = (1 << 16) / quant_val;
+ cpi->UVzbin[Q][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;;
+ cpi->UVround[Q][0] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.UVdequant[Q][0] = quant_val;
+ cpi->zrun_zbin_boost_uv[Q][0] = (quant_val * zbin_boost[0]) >> 7;
+
+ /* all the ac values = ; */
+ for (i = 1; i < 16; i++)
+ {
+ int rc = vp8_default_zig_zag1d[i];
+
+ quant_val = vp8_ac_yquant(Q);
+ cpi->Y1quant[Q][rc] = (1 << 16) / quant_val;
+ cpi->Y1zbin[Q][rc] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
+ cpi->Y1round[Q][rc] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.Y1dequant[Q][rc] = quant_val;
+ cpi->zrun_zbin_boost_y1[Q][i] = (quant_val * zbin_boost[i]) >> 7;
+
+ quant_val = vp8_ac2quant(Q, cpi->common.y2ac_delta_q);
+ cpi->Y2quant[Q][rc] = (1 << 16) / quant_val;
+ cpi->Y2zbin[Q][rc] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
+ cpi->Y2round[Q][rc] = (qrounding_factors_y2[Q] * quant_val) >> 7;
+ cpi->common.Y2dequant[Q][rc] = quant_val;
+ cpi->zrun_zbin_boost_y2[Q][i] = (quant_val * zbin_boost[i]) >> 7;
+
+ quant_val = vp8_ac_uv_quant(Q, cpi->common.uvac_delta_q);
+ cpi->UVquant[Q][rc] = (1 << 16) / quant_val;
+ cpi->UVzbin[Q][rc] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
+ cpi->UVround[Q][rc] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.UVdequant[Q][rc] = quant_val;
+ cpi->zrun_zbin_boost_uv[Q][i] = (quant_val * zbin_boost[i]) >> 7;
+ }
+ }
+}
+#endif
+
+#define ZBIN_EXTRA_Y \
+ (( cpi->common.Y1dequant[QIndex][1] * \
+ ( cpi->zbin_over_quant + \
+ cpi->zbin_mode_boost + \
+ x->act_zbin_adj ) ) >> 7)
+
+#define ZBIN_EXTRA_UV \
+ (( cpi->common.UVdequant[QIndex][1] * \
+ ( cpi->zbin_over_quant + \
+ cpi->zbin_mode_boost + \
+ x->act_zbin_adj ) ) >> 7)
+
+#define ZBIN_EXTRA_Y2 \
+ (( cpi->common.Y2dequant[QIndex][1] * \
+ ( (cpi->zbin_over_quant / 2) + \
+ cpi->zbin_mode_boost + \
+ x->act_zbin_adj ) ) >> 7)
+
+void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x, int ok_to_skip)
+{
+ int i;
+ int QIndex;
+ MACROBLOCKD *xd = &x->e_mbd;
+ int zbin_extra;
+
+ /* Select the baseline MB Q index. */
+ if (xd->segmentation_enabled)
+ {
+ /* Abs Value */
+ if (xd->mb_segement_abs_delta == SEGMENT_ABSDATA)
+ QIndex = xd->segment_feature_data[MB_LVL_ALT_Q][xd->mode_info_context->mbmi.segment_id];
+ /* Delta Value */
+ else
+ {
+ QIndex = cpi->common.base_qindex + xd->segment_feature_data[MB_LVL_ALT_Q][xd->mode_info_context->mbmi.segment_id];
+ /* Clamp to valid range */
+ QIndex = (QIndex >= 0) ? ((QIndex <= MAXQ) ? QIndex : MAXQ) : 0;
+ }
+ }
+ else
+ QIndex = cpi->common.base_qindex;
+
+ /* This initialization should be called at least once. Use ok_to_skip to
+ * decide if it is ok to skip.
+ * Before encoding a frame, this function is always called with ok_to_skip
+ * =0, which means no skiping of calculations. The "last" values are
+ * initialized at that time.
+ */
+ if (!ok_to_skip || QIndex != x->q_index)
+ {
+
+ xd->dequant_y1_dc[0] = 1;
+ xd->dequant_y1[0] = cpi->common.Y1dequant[QIndex][0];
+ xd->dequant_y2[0] = cpi->common.Y2dequant[QIndex][0];
+ xd->dequant_uv[0] = cpi->common.UVdequant[QIndex][0];
+
+ for (i = 1; i < 16; i++)
+ {
+ xd->dequant_y1_dc[i] =
+ xd->dequant_y1[i] = cpi->common.Y1dequant[QIndex][1];
+ xd->dequant_y2[i] = cpi->common.Y2dequant[QIndex][1];
+ xd->dequant_uv[i] = cpi->common.UVdequant[QIndex][1];
+ }
+#if 1
+ /*TODO: Remove dequant from BLOCKD. This is a temporary solution until
+ * the quantizer code uses a passed in pointer to the dequant constants.
+ * This will also require modifications to the x86 and neon assembly.
+ * */
+ for (i = 0; i < 16; i++)
+ x->e_mbd.block[i].dequant = xd->dequant_y1;
+ for (i = 16; i < 24; i++)
+ x->e_mbd.block[i].dequant = xd->dequant_uv;
+ x->e_mbd.block[24].dequant = xd->dequant_y2;
+#endif
+
+ /* Y */
+ zbin_extra = ZBIN_EXTRA_Y;
+
+ for (i = 0; i < 16; i++)
+ {
+ x->block[i].quant = cpi->Y1quant[QIndex];
+ x->block[i].quant_fast = cpi->Y1quant_fast[QIndex];
+ x->block[i].quant_shift = cpi->Y1quant_shift[QIndex];
+ x->block[i].zbin = cpi->Y1zbin[QIndex];
+ x->block[i].round = cpi->Y1round[QIndex];
+ x->block[i].zrun_zbin_boost = cpi->zrun_zbin_boost_y1[QIndex];
+ x->block[i].zbin_extra = (short)zbin_extra;
+ }
+
+ /* UV */
+ zbin_extra = ZBIN_EXTRA_UV;
+
+ for (i = 16; i < 24; i++)
+ {
+ x->block[i].quant = cpi->UVquant[QIndex];
+ x->block[i].quant_fast = cpi->UVquant_fast[QIndex];
+ x->block[i].quant_shift = cpi->UVquant_shift[QIndex];
+ x->block[i].zbin = cpi->UVzbin[QIndex];
+ x->block[i].round = cpi->UVround[QIndex];
+ x->block[i].zrun_zbin_boost = cpi->zrun_zbin_boost_uv[QIndex];
+ x->block[i].zbin_extra = (short)zbin_extra;
+ }
+
+ /* Y2 */
+ zbin_extra = ZBIN_EXTRA_Y2;
+
+ x->block[24].quant_fast = cpi->Y2quant_fast[QIndex];
+ x->block[24].quant = cpi->Y2quant[QIndex];
+ x->block[24].quant_shift = cpi->Y2quant_shift[QIndex];
+ x->block[24].zbin = cpi->Y2zbin[QIndex];
+ x->block[24].round = cpi->Y2round[QIndex];
+ x->block[24].zrun_zbin_boost = cpi->zrun_zbin_boost_y2[QIndex];
+ x->block[24].zbin_extra = (short)zbin_extra;
+
+ /* save this macroblock QIndex for vp8_update_zbin_extra() */
+ x->q_index = QIndex;
+
+ cpi->last_zbin_over_quant = cpi->zbin_over_quant;
+ cpi->last_zbin_mode_boost = cpi->zbin_mode_boost;
+ x->last_act_zbin_adj = x->act_zbin_adj;
+
+
+
+ }
+ else if(cpi->last_zbin_over_quant != cpi->zbin_over_quant
+ || cpi->last_zbin_mode_boost != cpi->zbin_mode_boost
+ || x->last_act_zbin_adj != x->act_zbin_adj)
+ {
+ /* Y */
+ zbin_extra = ZBIN_EXTRA_Y;
+
+ for (i = 0; i < 16; i++)
+ x->block[i].zbin_extra = (short)zbin_extra;
+
+ /* UV */
+ zbin_extra = ZBIN_EXTRA_UV;
+
+ for (i = 16; i < 24; i++)
+ x->block[i].zbin_extra = (short)zbin_extra;
+
+ /* Y2 */
+ zbin_extra = ZBIN_EXTRA_Y2;
+ x->block[24].zbin_extra = (short)zbin_extra;
+
+ cpi->last_zbin_over_quant = cpi->zbin_over_quant;
+ cpi->last_zbin_mode_boost = cpi->zbin_mode_boost;
+ x->last_act_zbin_adj = x->act_zbin_adj;
+ }
+}
+
+void vp8_update_zbin_extra(VP8_COMP *cpi, MACROBLOCK *x)
+{
+ int i;
+ int QIndex = x->q_index;
+ int zbin_extra;
+
+ /* Y */
+ zbin_extra = ZBIN_EXTRA_Y;
+
+ for (i = 0; i < 16; i++)
+ x->block[i].zbin_extra = (short)zbin_extra;
+
+ /* UV */
+ zbin_extra = ZBIN_EXTRA_UV;
+
+ for (i = 16; i < 24; i++)
+ x->block[i].zbin_extra = (short)zbin_extra;
+
+ /* Y2 */
+ zbin_extra = ZBIN_EXTRA_Y2;
+ x->block[24].zbin_extra = (short)zbin_extra;
+}
+#undef ZBIN_EXTRA_Y
+#undef ZBIN_EXTRA_UV
+#undef ZBIN_EXTRA_Y2
+
+void vp8cx_frame_init_quantizer(VP8_COMP *cpi)
+{
+ /* Clear Zbin mode boost for default case */
+ cpi->zbin_mode_boost = 0;
+
+ /* MB level quantizer setup */
+ vp8cx_mb_init_quantizer(cpi, &cpi->mb, 0);
+}
+
+
+void vp8_set_quantizer(struct VP8_COMP *cpi, int Q)
+{
+ VP8_COMMON *cm = &cpi->common;
+ MACROBLOCKD *mbd = &cpi->mb.e_mbd;
+ int update = 0;
+ int new_delta_q;
+ cm->base_qindex = Q;
+
+ /* if any of the delta_q values are changing update flag has to be set */
+ /* currently only y2dc_delta_q may change */
+
+ cm->y1dc_delta_q = 0;
+ cm->y2ac_delta_q = 0;
+ cm->uvdc_delta_q = 0;
+ cm->uvac_delta_q = 0;
+
+ if (Q < 4)
+ {
+ new_delta_q = 4-Q;
+ }
+ else
+ new_delta_q = 0;
+
+ update |= cm->y2dc_delta_q != new_delta_q;
+ cm->y2dc_delta_q = new_delta_q;
+
+
+ /* Set Segment specific quatizers */
+ mbd->segment_feature_data[MB_LVL_ALT_Q][0] = cpi->segment_feature_data[MB_LVL_ALT_Q][0];
+ mbd->segment_feature_data[MB_LVL_ALT_Q][1] = cpi->segment_feature_data[MB_LVL_ALT_Q][1];
+ mbd->segment_feature_data[MB_LVL_ALT_Q][2] = cpi->segment_feature_data[MB_LVL_ALT_Q][2];
+ mbd->segment_feature_data[MB_LVL_ALT_Q][3] = cpi->segment_feature_data[MB_LVL_ALT_Q][3];
+
+ /* quantizer has to be reinitialized for any delta_q changes */
+ if(update)
+ vp8cx_init_quantizer(cpi);
+
+}
diff --git a/libvpx/vp8/encoder/quantize.h b/libvpx/vp8/encoder/quantize.h
new file mode 100644
index 0000000..d55496c
--- /dev/null
+++ b/libvpx/vp8/encoder/quantize.h
@@ -0,0 +1,23 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_QUANTIZE_H
+#define __INC_QUANTIZE_H
+
+struct VP8_COMP;
+struct macroblock;
+extern void vp8_set_quantizer(struct VP8_COMP *cpi, int Q);
+extern void vp8cx_frame_init_quantizer(struct VP8_COMP *cpi);
+extern void vp8_update_zbin_extra(struct VP8_COMP *cpi, struct macroblock *x);
+extern void vp8cx_mb_init_quantizer(struct VP8_COMP *cpi, struct macroblock *x, int ok_to_skip);
+extern void vp8cx_init_quantizer(struct VP8_COMP *cpi);
+
+#endif
diff --git a/libvpx/vp8/encoder/ratectrl.c b/libvpx/vp8/encoder/ratectrl.c
new file mode 100644
index 0000000..77c1c5a
--- /dev/null
+++ b/libvpx/vp8/encoder/ratectrl.c
@@ -0,0 +1,1534 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <limits.h>
+#include <assert.h>
+
+#include "math.h"
+#include "vp8/common/common.h"
+#include "ratectrl.h"
+#include "vp8/common/entropymode.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/systemdependent.h"
+#include "encodemv.h"
+
+
+#define MIN_BPB_FACTOR 0.01
+#define MAX_BPB_FACTOR 50
+
+extern const MB_PREDICTION_MODE vp8_mode_order[MAX_MODES];
+
+
+
+#ifdef MODE_STATS
+extern int y_modes[5];
+extern int uv_modes[4];
+extern int b_modes[10];
+
+extern int inter_y_modes[10];
+extern int inter_uv_modes[4];
+extern int inter_b_modes[10];
+#endif
+
+/* Bits Per MB at different Q (Multiplied by 512) */
+#define BPER_MB_NORMBITS 9
+
+/* Work in progress recalibration of baseline rate tables based on
+ * the assumption that bits per mb is inversely proportional to the
+ * quantizer value.
+ */
+const int vp8_bits_per_mb[2][QINDEX_RANGE] =
+{
+ /* Intra case 450000/Qintra */
+ {
+ 1125000,900000, 750000, 642857, 562500, 500000, 450000, 450000,
+ 409090, 375000, 346153, 321428, 300000, 281250, 264705, 264705,
+ 250000, 236842, 225000, 225000, 214285, 214285, 204545, 204545,
+ 195652, 195652, 187500, 180000, 180000, 173076, 166666, 160714,
+ 155172, 150000, 145161, 140625, 136363, 132352, 128571, 125000,
+ 121621, 121621, 118421, 115384, 112500, 109756, 107142, 104651,
+ 102272, 100000, 97826, 97826, 95744, 93750, 91836, 90000,
+ 88235, 86538, 84905, 83333, 81818, 80357, 78947, 77586,
+ 76271, 75000, 73770, 72580, 71428, 70312, 69230, 68181,
+ 67164, 66176, 65217, 64285, 63380, 62500, 61643, 60810,
+ 60000, 59210, 59210, 58441, 57692, 56962, 56250, 55555,
+ 54878, 54216, 53571, 52941, 52325, 51724, 51136, 50561,
+ 49450, 48387, 47368, 46875, 45918, 45000, 44554, 44117,
+ 43269, 42452, 41666, 40909, 40178, 39473, 38793, 38135,
+ 36885, 36290, 35714, 35156, 34615, 34090, 33582, 33088,
+ 32608, 32142, 31468, 31034, 30405, 29801, 29220, 28662,
+ },
+ /* Inter case 285000/Qinter */
+ {
+ 712500, 570000, 475000, 407142, 356250, 316666, 285000, 259090,
+ 237500, 219230, 203571, 190000, 178125, 167647, 158333, 150000,
+ 142500, 135714, 129545, 123913, 118750, 114000, 109615, 105555,
+ 101785, 98275, 95000, 91935, 89062, 86363, 83823, 81428,
+ 79166, 77027, 75000, 73076, 71250, 69512, 67857, 66279,
+ 64772, 63333, 61956, 60638, 59375, 58163, 57000, 55882,
+ 54807, 53773, 52777, 51818, 50892, 50000, 49137, 47500,
+ 45967, 44531, 43181, 41911, 40714, 39583, 38513, 37500,
+ 36538, 35625, 34756, 33928, 33139, 32386, 31666, 30978,
+ 30319, 29687, 29081, 28500, 27941, 27403, 26886, 26388,
+ 25909, 25446, 25000, 24568, 23949, 23360, 22800, 22265,
+ 21755, 21268, 20802, 20357, 19930, 19520, 19127, 18750,
+ 18387, 18037, 17701, 17378, 17065, 16764, 16473, 16101,
+ 15745, 15405, 15079, 14766, 14467, 14179, 13902, 13636,
+ 13380, 13133, 12895, 12666, 12445, 12179, 11924, 11632,
+ 11445, 11220, 11003, 10795, 10594, 10401, 10215, 10035,
+ }
+};
+
+static const int kf_boost_qadjustment[QINDEX_RANGE] =
+{
+ 128, 129, 130, 131, 132, 133, 134, 135,
+ 136, 137, 138, 139, 140, 141, 142, 143,
+ 144, 145, 146, 147, 148, 149, 150, 151,
+ 152, 153, 154, 155, 156, 157, 158, 159,
+ 160, 161, 162, 163, 164, 165, 166, 167,
+ 168, 169, 170, 171, 172, 173, 174, 175,
+ 176, 177, 178, 179, 180, 181, 182, 183,
+ 184, 185, 186, 187, 188, 189, 190, 191,
+ 192, 193, 194, 195, 196, 197, 198, 199,
+ 200, 200, 201, 201, 202, 203, 203, 203,
+ 204, 204, 205, 205, 206, 206, 207, 207,
+ 208, 208, 209, 209, 210, 210, 211, 211,
+ 212, 212, 213, 213, 214, 214, 215, 215,
+ 216, 216, 217, 217, 218, 218, 219, 219,
+ 220, 220, 220, 220, 220, 220, 220, 220,
+ 220, 220, 220, 220, 220, 220, 220, 220,
+};
+
+/* #define GFQ_ADJUSTMENT (Q+100) */
+#define GFQ_ADJUSTMENT vp8_gf_boost_qadjustment[Q]
+const int vp8_gf_boost_qadjustment[QINDEX_RANGE] =
+{
+ 80, 82, 84, 86, 88, 90, 92, 94,
+ 96, 97, 98, 99, 100, 101, 102, 103,
+ 104, 105, 106, 107, 108, 109, 110, 111,
+ 112, 113, 114, 115, 116, 117, 118, 119,
+ 120, 121, 122, 123, 124, 125, 126, 127,
+ 128, 129, 130, 131, 132, 133, 134, 135,
+ 136, 137, 138, 139, 140, 141, 142, 143,
+ 144, 145, 146, 147, 148, 149, 150, 151,
+ 152, 153, 154, 155, 156, 157, 158, 159,
+ 160, 161, 162, 163, 164, 165, 166, 167,
+ 168, 169, 170, 171, 172, 173, 174, 175,
+ 176, 177, 178, 179, 180, 181, 182, 183,
+ 184, 184, 185, 185, 186, 186, 187, 187,
+ 188, 188, 189, 189, 190, 190, 191, 191,
+ 192, 192, 193, 193, 194, 194, 194, 194,
+ 195, 195, 196, 196, 197, 197, 198, 198
+};
+
+/*
+const int vp8_gf_boost_qadjustment[QINDEX_RANGE] =
+{
+ 100,101,102,103,104,105,105,106,
+ 106,107,107,108,109,109,110,111,
+ 112,113,114,115,116,117,118,119,
+ 120,121,122,123,124,125,126,127,
+ 128,129,130,131,132,133,134,135,
+ 136,137,138,139,140,141,142,143,
+ 144,145,146,147,148,149,150,151,
+ 152,153,154,155,156,157,158,159,
+ 160,161,162,163,164,165,166,167,
+ 168,169,170,170,171,171,172,172,
+ 173,173,173,174,174,174,175,175,
+ 175,176,176,176,177,177,177,177,
+ 178,178,179,179,180,180,181,181,
+ 182,182,183,183,184,184,185,185,
+ 186,186,187,187,188,188,189,189,
+ 190,190,191,191,192,192,193,193,
+};
+*/
+
+static const int kf_gf_boost_qlimits[QINDEX_RANGE] =
+{
+ 150, 155, 160, 165, 170, 175, 180, 185,
+ 190, 195, 200, 205, 210, 215, 220, 225,
+ 230, 235, 240, 245, 250, 255, 260, 265,
+ 270, 275, 280, 285, 290, 295, 300, 305,
+ 310, 320, 330, 340, 350, 360, 370, 380,
+ 390, 400, 410, 420, 430, 440, 450, 460,
+ 470, 480, 490, 500, 510, 520, 530, 540,
+ 550, 560, 570, 580, 590, 600, 600, 600,
+ 600, 600, 600, 600, 600, 600, 600, 600,
+ 600, 600, 600, 600, 600, 600, 600, 600,
+ 600, 600, 600, 600, 600, 600, 600, 600,
+ 600, 600, 600, 600, 600, 600, 600, 600,
+ 600, 600, 600, 600, 600, 600, 600, 600,
+ 600, 600, 600, 600, 600, 600, 600, 600,
+ 600, 600, 600, 600, 600, 600, 600, 600,
+ 600, 600, 600, 600, 600, 600, 600, 600,
+};
+
+/* % adjustment to target kf size based on seperation from previous frame */
+static const int kf_boost_seperation_adjustment[16] =
+{
+ 30, 40, 50, 55, 60, 65, 70, 75,
+ 80, 85, 90, 95, 100, 100, 100, 100,
+};
+
+
+static const int gf_adjust_table[101] =
+{
+ 100,
+ 115, 130, 145, 160, 175, 190, 200, 210, 220, 230,
+ 240, 260, 270, 280, 290, 300, 310, 320, 330, 340,
+ 350, 360, 370, 380, 390, 400, 400, 400, 400, 400,
+ 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
+ 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
+ 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
+ 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
+ 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
+ 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
+ 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
+};
+
+static const int gf_intra_usage_adjustment[20] =
+{
+ 125, 120, 115, 110, 105, 100, 95, 85, 80, 75,
+ 70, 65, 60, 55, 50, 50, 50, 50, 50, 50,
+};
+
+static const int gf_interval_table[101] =
+{
+ 7,
+ 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+ 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+ 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+ 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
+ 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
+ 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
+ 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
+ 10, 10, 10, 10, 10, 10, 10, 10, 10, 10,
+ 10, 10, 10, 10, 10, 10, 10, 10, 10, 10,
+ 11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+};
+
+static const unsigned int prior_key_frame_weight[KEY_FRAME_CONTEXT] = { 1, 2, 3, 4, 5 };
+
+
+void vp8_save_coding_context(VP8_COMP *cpi)
+{
+ CODING_CONTEXT *const cc = & cpi->coding_context;
+
+ /* Stores a snapshot of key state variables which can subsequently be
+ * restored with a call to vp8_restore_coding_context. These functions are
+ * intended for use in a re-code loop in vp8_compress_frame where the
+ * quantizer value is adjusted between loop iterations.
+ */
+
+ cc->frames_since_key = cpi->frames_since_key;
+ cc->filter_level = cpi->common.filter_level;
+ cc->frames_till_gf_update_due = cpi->frames_till_gf_update_due;
+ cc->frames_since_golden = cpi->common.frames_since_golden;
+
+ vp8_copy(cc->mvc, cpi->common.fc.mvc);
+ vp8_copy(cc->mvcosts, cpi->rd_costs.mvcosts);
+
+ vp8_copy(cc->ymode_prob, cpi->common.fc.ymode_prob);
+ vp8_copy(cc->uv_mode_prob, cpi->common.fc.uv_mode_prob);
+
+ vp8_copy(cc->ymode_count, cpi->mb.ymode_count);
+ vp8_copy(cc->uv_mode_count, cpi->mb.uv_mode_count);
+
+
+ /* Stats */
+#ifdef MODE_STATS
+ vp8_copy(cc->y_modes, y_modes);
+ vp8_copy(cc->uv_modes, uv_modes);
+ vp8_copy(cc->b_modes, b_modes);
+ vp8_copy(cc->inter_y_modes, inter_y_modes);
+ vp8_copy(cc->inter_uv_modes, inter_uv_modes);
+ vp8_copy(cc->inter_b_modes, inter_b_modes);
+#endif
+
+ cc->this_frame_percent_intra = cpi->this_frame_percent_intra;
+}
+
+
+void vp8_restore_coding_context(VP8_COMP *cpi)
+{
+ CODING_CONTEXT *const cc = & cpi->coding_context;
+
+ /* Restore key state variables to the snapshot state stored in the
+ * previous call to vp8_save_coding_context.
+ */
+
+ cpi->frames_since_key = cc->frames_since_key;
+ cpi->common.filter_level = cc->filter_level;
+ cpi->frames_till_gf_update_due = cc->frames_till_gf_update_due;
+ cpi->common.frames_since_golden = cc->frames_since_golden;
+
+ vp8_copy(cpi->common.fc.mvc, cc->mvc);
+
+ vp8_copy(cpi->rd_costs.mvcosts, cc->mvcosts);
+
+ vp8_copy(cpi->common.fc.ymode_prob, cc->ymode_prob);
+ vp8_copy(cpi->common.fc.uv_mode_prob, cc->uv_mode_prob);
+
+ vp8_copy(cpi->mb.ymode_count, cc->ymode_count);
+ vp8_copy(cpi->mb.uv_mode_count, cc->uv_mode_count);
+
+ /* Stats */
+#ifdef MODE_STATS
+ vp8_copy(y_modes, cc->y_modes);
+ vp8_copy(uv_modes, cc->uv_modes);
+ vp8_copy(b_modes, cc->b_modes);
+ vp8_copy(inter_y_modes, cc->inter_y_modes);
+ vp8_copy(inter_uv_modes, cc->inter_uv_modes);
+ vp8_copy(inter_b_modes, cc->inter_b_modes);
+#endif
+
+
+ cpi->this_frame_percent_intra = cc->this_frame_percent_intra;
+}
+
+
+void vp8_setup_key_frame(VP8_COMP *cpi)
+{
+ /* Setup for Key frame: */
+
+ vp8_default_coef_probs(& cpi->common);
+
+ vpx_memcpy(cpi->common.fc.mvc, vp8_default_mv_context, sizeof(vp8_default_mv_context));
+ {
+ int flag[2] = {1, 1};
+ vp8_build_component_cost_table(cpi->mb.mvcost, (const MV_CONTEXT *) cpi->common.fc.mvc, flag);
+ }
+
+ /* Make sure we initialize separate contexts for altref,gold, and normal.
+ * TODO shouldn't need 3 different copies of structure to do this!
+ */
+ vpx_memcpy(&cpi->lfc_a, &cpi->common.fc, sizeof(cpi->common.fc));
+ vpx_memcpy(&cpi->lfc_g, &cpi->common.fc, sizeof(cpi->common.fc));
+ vpx_memcpy(&cpi->lfc_n, &cpi->common.fc, sizeof(cpi->common.fc));
+
+ cpi->common.filter_level = cpi->common.base_qindex * 3 / 8 ;
+
+ /* Provisional interval before next GF */
+ if (cpi->auto_gold)
+ cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
+ else
+ cpi->frames_till_gf_update_due = DEFAULT_GF_INTERVAL;
+
+ cpi->common.refresh_golden_frame = 1;
+ cpi->common.refresh_alt_ref_frame = 1;
+}
+
+
+static int estimate_bits_at_q(int frame_kind, int Q, int MBs,
+ double correction_factor)
+{
+ int Bpm = (int)(.5 + correction_factor * vp8_bits_per_mb[frame_kind][Q]);
+
+ /* Attempt to retain reasonable accuracy without overflow. The cutoff is
+ * chosen such that the maximum product of Bpm and MBs fits 31 bits. The
+ * largest Bpm takes 20 bits.
+ */
+ if (MBs > (1 << 11))
+ return (Bpm >> BPER_MB_NORMBITS) * MBs;
+ else
+ return (Bpm * MBs) >> BPER_MB_NORMBITS;
+}
+
+
+static void calc_iframe_target_size(VP8_COMP *cpi)
+{
+ /* boost defaults to half second */
+ int kf_boost;
+ uint64_t target;
+
+ /* Clear down mmx registers to allow floating point in what follows */
+ vp8_clear_system_state();
+
+ if (cpi->oxcf.fixed_q >= 0)
+ {
+ int Q = cpi->oxcf.key_q;
+
+ target = estimate_bits_at_q(INTRA_FRAME, Q, cpi->common.MBs,
+ cpi->key_frame_rate_correction_factor);
+ }
+ else if (cpi->pass == 2)
+ {
+ /* New Two pass RC */
+ target = cpi->per_frame_bandwidth;
+ }
+ /* First Frame is a special case */
+ else if (cpi->common.current_video_frame == 0)
+ {
+ /* 1 Pass there is no information on which to base size so use
+ * bandwidth per second * fraction of the initial buffer
+ * level
+ */
+ target = cpi->oxcf.starting_buffer_level / 2;
+
+ if(target > cpi->oxcf.target_bandwidth * 3 / 2)
+ target = cpi->oxcf.target_bandwidth * 3 / 2;
+ }
+ else
+ {
+ /* if this keyframe was forced, use a more recent Q estimate */
+ int Q = (cpi->common.frame_flags & FRAMEFLAGS_KEY)
+ ? cpi->avg_frame_qindex : cpi->ni_av_qi;
+
+ int initial_boost = 32; /* |3.0 * per_frame_bandwidth| */
+ /* Boost depends somewhat on frame rate: only used for 1 layer case. */
+ if (cpi->oxcf.number_of_layers == 1) {
+ kf_boost = MAX(initial_boost, (int)(2 * cpi->output_frame_rate - 16));
+ }
+ else {
+ /* Initial factor: set target size to: |3.0 * per_frame_bandwidth|. */
+ kf_boost = initial_boost;
+ }
+
+ /* adjustment up based on q: this factor ranges from ~1.2 to 2.2. */
+ kf_boost = kf_boost * kf_boost_qadjustment[Q] / 100;
+
+ /* frame separation adjustment ( down) */
+ if (cpi->frames_since_key < cpi->output_frame_rate / 2)
+ kf_boost = (int)(kf_boost
+ * cpi->frames_since_key / (cpi->output_frame_rate / 2));
+
+ /* Minimal target size is |2* per_frame_bandwidth|. */
+ if (kf_boost < 16)
+ kf_boost = 16;
+
+ target = ((16 + kf_boost) * cpi->per_frame_bandwidth) >> 4;
+ }
+
+
+ if (cpi->oxcf.rc_max_intra_bitrate_pct)
+ {
+ unsigned int max_rate = cpi->per_frame_bandwidth
+ * cpi->oxcf.rc_max_intra_bitrate_pct / 100;
+
+ if (target > max_rate)
+ target = max_rate;
+ }
+
+ cpi->this_frame_target = (int)target;
+
+ /* TODO: if we separate rate targeting from Q targetting, move this.
+ * Reset the active worst quality to the baseline value for key frames.
+ */
+ if (cpi->pass != 2)
+ cpi->active_worst_quality = cpi->worst_quality;
+
+#if 0
+ {
+ FILE *f;
+
+ f = fopen("kf_boost.stt", "a");
+ fprintf(f, " %8u %10d %10d %10d\n",
+ cpi->common.current_video_frame, cpi->gfu_boost, cpi->baseline_gf_interval, cpi->source_alt_ref_pending);
+
+ fclose(f);
+ }
+#endif
+}
+
+
+/* Do the best we can to define the parameters for the next GF based on what
+ * information we have available.
+ */
+static void calc_gf_params(VP8_COMP *cpi)
+{
+ int Q = (cpi->oxcf.fixed_q < 0) ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
+ int Boost = 0;
+
+ int gf_frame_useage = 0; /* Golden frame useage since last GF */
+ int tot_mbs = cpi->recent_ref_frame_usage[INTRA_FRAME] +
+ cpi->recent_ref_frame_usage[LAST_FRAME] +
+ cpi->recent_ref_frame_usage[GOLDEN_FRAME] +
+ cpi->recent_ref_frame_usage[ALTREF_FRAME];
+
+ int pct_gf_active = (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols);
+
+ if (tot_mbs)
+ gf_frame_useage = (cpi->recent_ref_frame_usage[GOLDEN_FRAME] + cpi->recent_ref_frame_usage[ALTREF_FRAME]) * 100 / tot_mbs;
+
+ if (pct_gf_active > gf_frame_useage)
+ gf_frame_useage = pct_gf_active;
+
+ /* Not two pass */
+ if (cpi->pass != 2)
+ {
+ /* Single Pass lagged mode: TBD */
+ if (0)
+ {
+ }
+
+ /* Single Pass compression: Has to use current and historical data */
+ else
+ {
+#if 0
+ /* Experimental code */
+ int index = cpi->one_pass_frame_index;
+ int frames_to_scan = (cpi->max_gf_interval <= MAX_LAG_BUFFERS) ? cpi->max_gf_interval : MAX_LAG_BUFFERS;
+
+ /* ************** Experimental code - incomplete */
+ /*
+ double decay_val = 1.0;
+ double IIAccumulator = 0.0;
+ double last_iiaccumulator = 0.0;
+ double IIRatio;
+
+ cpi->one_pass_frame_index = cpi->common.current_video_frame%MAX_LAG_BUFFERS;
+
+ for ( i = 0; i < (frames_to_scan - 1); i++ )
+ {
+ if ( index < 0 )
+ index = MAX_LAG_BUFFERS;
+ index --;
+
+ if ( cpi->one_pass_frame_stats[index].frame_coded_error > 0.0 )
+ {
+ IIRatio = cpi->one_pass_frame_stats[index].frame_intra_error / cpi->one_pass_frame_stats[index].frame_coded_error;
+
+ if ( IIRatio > 30.0 )
+ IIRatio = 30.0;
+ }
+ else
+ IIRatio = 30.0;
+
+ IIAccumulator += IIRatio * decay_val;
+
+ decay_val = decay_val * cpi->one_pass_frame_stats[index].frame_pcnt_inter;
+
+ if ( (i > MIN_GF_INTERVAL) &&
+ ((IIAccumulator - last_iiaccumulator) < 2.0) )
+ {
+ break;
+ }
+ last_iiaccumulator = IIAccumulator;
+ }
+
+ Boost = IIAccumulator*100.0/16.0;
+ cpi->baseline_gf_interval = i;
+
+ */
+#else
+
+ /*************************************************************/
+ /* OLD code */
+
+ /* Adjust boost based upon ambient Q */
+ Boost = GFQ_ADJUSTMENT;
+
+ /* Adjust based upon most recently measure intra useage */
+ Boost = Boost * gf_intra_usage_adjustment[(cpi->this_frame_percent_intra < 15) ? cpi->this_frame_percent_intra : 14] / 100;
+
+ /* Adjust gf boost based upon GF usage since last GF */
+ Boost = Boost * gf_adjust_table[gf_frame_useage] / 100;
+#endif
+ }
+
+ /* golden frame boost without recode loop often goes awry. be
+ * safe by keeping numbers down.
+ */
+ if (!cpi->sf.recode_loop)
+ {
+ if (cpi->compressor_speed == 2)
+ Boost = Boost / 2;
+ }
+
+ /* Apply an upper limit based on Q for 1 pass encodes */
+ if (Boost > kf_gf_boost_qlimits[Q] && (cpi->pass == 0))
+ Boost = kf_gf_boost_qlimits[Q];
+
+ /* Apply lower limits to boost. */
+ else if (Boost < 110)
+ Boost = 110;
+
+ /* Note the boost used */
+ cpi->last_boost = Boost;
+
+ }
+
+ /* Estimate next interval
+ * This is updated once the real frame size/boost is known.
+ */
+ if (cpi->oxcf.fixed_q == -1)
+ {
+ if (cpi->pass == 2) /* 2 Pass */
+ {
+ cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
+ }
+ else /* 1 Pass */
+ {
+ cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
+
+ if (cpi->last_boost > 750)
+ cpi->frames_till_gf_update_due++;
+
+ if (cpi->last_boost > 1000)
+ cpi->frames_till_gf_update_due++;
+
+ if (cpi->last_boost > 1250)
+ cpi->frames_till_gf_update_due++;
+
+ if (cpi->last_boost >= 1500)
+ cpi->frames_till_gf_update_due ++;
+
+ if (gf_interval_table[gf_frame_useage] > cpi->frames_till_gf_update_due)
+ cpi->frames_till_gf_update_due = gf_interval_table[gf_frame_useage];
+
+ if (cpi->frames_till_gf_update_due > cpi->max_gf_interval)
+ cpi->frames_till_gf_update_due = cpi->max_gf_interval;
+ }
+ }
+ else
+ cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
+
+ /* ARF on or off */
+ if (cpi->pass != 2)
+ {
+ /* For now Alt ref is not allowed except in 2 pass modes. */
+ cpi->source_alt_ref_pending = 0;
+
+ /*if ( cpi->oxcf.fixed_q == -1)
+ {
+ if ( cpi->oxcf.play_alternate && (cpi->last_boost > (100 + (AF_THRESH*cpi->frames_till_gf_update_due)) ) )
+ cpi->source_alt_ref_pending = 1;
+ else
+ cpi->source_alt_ref_pending = 0;
+ }*/
+ }
+}
+
+
+static void calc_pframe_target_size(VP8_COMP *cpi)
+{
+ int min_frame_target;
+ int Adjustment;
+ int old_per_frame_bandwidth = cpi->per_frame_bandwidth;
+
+ if ( cpi->current_layer > 0)
+ cpi->per_frame_bandwidth =
+ cpi->layer_context[cpi->current_layer].avg_frame_size_for_layer;
+
+ min_frame_target = 0;
+
+ if (cpi->pass == 2)
+ {
+ min_frame_target = cpi->min_frame_bandwidth;
+
+ if (min_frame_target < (cpi->av_per_frame_bandwidth >> 5))
+ min_frame_target = cpi->av_per_frame_bandwidth >> 5;
+ }
+ else if (min_frame_target < cpi->per_frame_bandwidth / 4)
+ min_frame_target = cpi->per_frame_bandwidth / 4;
+
+
+ /* Special alt reference frame case */
+ if((cpi->common.refresh_alt_ref_frame) && (cpi->oxcf.number_of_layers == 1))
+ {
+ if (cpi->pass == 2)
+ {
+ /* Per frame bit target for the alt ref frame */
+ cpi->per_frame_bandwidth = cpi->twopass.gf_bits;
+ cpi->this_frame_target = cpi->per_frame_bandwidth;
+ }
+
+ /* One Pass ??? TBD */
+ }
+
+ /* Normal frames (gf,and inter) */
+ else
+ {
+ /* 2 pass */
+ if (cpi->pass == 2)
+ {
+ cpi->this_frame_target = cpi->per_frame_bandwidth;
+ }
+ /* 1 pass */
+ else
+ {
+ /* Make rate adjustment to recover bits spent in key frame
+ * Test to see if the key frame inter data rate correction
+ * should still be in force
+ */
+ if (cpi->kf_overspend_bits > 0)
+ {
+ Adjustment = (cpi->kf_bitrate_adjustment <= cpi->kf_overspend_bits) ? cpi->kf_bitrate_adjustment : cpi->kf_overspend_bits;
+
+ if (Adjustment > (cpi->per_frame_bandwidth - min_frame_target))
+ Adjustment = (cpi->per_frame_bandwidth - min_frame_target);
+
+ cpi->kf_overspend_bits -= Adjustment;
+
+ /* Calculate an inter frame bandwidth target for the next
+ * few frames designed to recover any extra bits spent on
+ * the key frame.
+ */
+ cpi->this_frame_target = cpi->per_frame_bandwidth - Adjustment;
+
+ if (cpi->this_frame_target < min_frame_target)
+ cpi->this_frame_target = min_frame_target;
+ }
+ else
+ cpi->this_frame_target = cpi->per_frame_bandwidth;
+
+ /* If appropriate make an adjustment to recover bits spent on a
+ * recent GF
+ */
+ if ((cpi->gf_overspend_bits > 0) && (cpi->this_frame_target > min_frame_target))
+ {
+ int Adjustment = (cpi->non_gf_bitrate_adjustment <= cpi->gf_overspend_bits) ? cpi->non_gf_bitrate_adjustment : cpi->gf_overspend_bits;
+
+ if (Adjustment > (cpi->this_frame_target - min_frame_target))
+ Adjustment = (cpi->this_frame_target - min_frame_target);
+
+ cpi->gf_overspend_bits -= Adjustment;
+ cpi->this_frame_target -= Adjustment;
+ }
+
+ /* Apply small + and - boosts for non gf frames */
+ if ((cpi->last_boost > 150) && (cpi->frames_till_gf_update_due > 0) &&
+ (cpi->current_gf_interval >= (MIN_GF_INTERVAL << 1)))
+ {
+ /* % Adjustment limited to the range 1% to 10% */
+ Adjustment = (cpi->last_boost - 100) >> 5;
+
+ if (Adjustment < 1)
+ Adjustment = 1;
+ else if (Adjustment > 10)
+ Adjustment = 10;
+
+ /* Convert to bits */
+ Adjustment = (cpi->this_frame_target * Adjustment) / 100;
+
+ if (Adjustment > (cpi->this_frame_target - min_frame_target))
+ Adjustment = (cpi->this_frame_target - min_frame_target);
+
+ if (cpi->common.frames_since_golden == (cpi->current_gf_interval >> 1))
+ cpi->this_frame_target += ((cpi->current_gf_interval - 1) * Adjustment);
+ else
+ cpi->this_frame_target -= Adjustment;
+ }
+ }
+ }
+
+ /* Sanity check that the total sum of adjustments is not above the
+ * maximum allowed That is that having allowed for KF and GF penalties
+ * we have not pushed the current interframe target to low. If the
+ * adjustment we apply here is not capable of recovering all the extra
+ * bits we have spent in the KF or GF then the remainder will have to
+ * be recovered over a longer time span via other buffer / rate control
+ * mechanisms.
+ */
+ if (cpi->this_frame_target < min_frame_target)
+ cpi->this_frame_target = min_frame_target;
+
+ if (!cpi->common.refresh_alt_ref_frame)
+ /* Note the baseline target data rate for this inter frame. */
+ cpi->inter_frame_target = cpi->this_frame_target;
+
+ /* One Pass specific code */
+ if (cpi->pass == 0)
+ {
+ /* Adapt target frame size with respect to any buffering constraints: */
+ if (cpi->buffered_mode)
+ {
+ int one_percent_bits = (int)
+ (1 + cpi->oxcf.optimal_buffer_level / 100);
+
+ if ((cpi->buffer_level < cpi->oxcf.optimal_buffer_level) ||
+ (cpi->bits_off_target < cpi->oxcf.optimal_buffer_level))
+ {
+ int percent_low = 0;
+
+ /* Decide whether or not we need to adjust the frame data
+ * rate target.
+ *
+ * If we are are below the optimal buffer fullness level
+ * and adherence to buffering constraints is important to
+ * the end usage then adjust the per frame target.
+ */
+ if ((cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
+ (cpi->buffer_level < cpi->oxcf.optimal_buffer_level))
+ {
+ percent_low = (int)
+ ((cpi->oxcf.optimal_buffer_level - cpi->buffer_level) /
+ one_percent_bits);
+ }
+ /* Are we overshooting the long term clip data rate... */
+ else if (cpi->bits_off_target < 0)
+ {
+ /* Adjust per frame data target downwards to compensate. */
+ percent_low = (int)(100 * -cpi->bits_off_target /
+ (cpi->total_byte_count * 8));
+ }
+
+ if (percent_low > cpi->oxcf.under_shoot_pct)
+ percent_low = cpi->oxcf.under_shoot_pct;
+ else if (percent_low < 0)
+ percent_low = 0;
+
+ /* lower the target bandwidth for this frame. */
+ cpi->this_frame_target -=
+ (cpi->this_frame_target * percent_low) / 200;
+
+ /* Are we using allowing control of active_worst_allowed_q
+ * according to buffer level.
+ */
+ if (cpi->auto_worst_q && cpi->ni_frames > 150)
+ {
+ int64_t critical_buffer_level;
+
+ /* For streaming applications the most important factor is
+ * cpi->buffer_level as this takes into account the
+ * specified short term buffering constraints. However,
+ * hitting the long term clip data rate target is also
+ * important.
+ */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ /* Take the smaller of cpi->buffer_level and
+ * cpi->bits_off_target
+ */
+ critical_buffer_level =
+ (cpi->buffer_level < cpi->bits_off_target)
+ ? cpi->buffer_level : cpi->bits_off_target;
+ }
+ /* For local file playback short term buffering constraints
+ * are less of an issue
+ */
+ else
+ {
+ /* Consider only how we are doing for the clip as a
+ * whole
+ */
+ critical_buffer_level = cpi->bits_off_target;
+ }
+
+ /* Set the active worst quality based upon the selected
+ * buffer fullness number.
+ */
+ if (critical_buffer_level < cpi->oxcf.optimal_buffer_level)
+ {
+ if ( critical_buffer_level >
+ (cpi->oxcf.optimal_buffer_level >> 2) )
+ {
+ int64_t qadjustment_range =
+ cpi->worst_quality - cpi->ni_av_qi;
+ int64_t above_base =
+ (critical_buffer_level -
+ (cpi->oxcf.optimal_buffer_level >> 2));
+
+ /* Step active worst quality down from
+ * cpi->ni_av_qi when (critical_buffer_level ==
+ * cpi->optimal_buffer_level) to
+ * cpi->worst_quality when
+ * (critical_buffer_level ==
+ * cpi->optimal_buffer_level >> 2)
+ */
+ cpi->active_worst_quality =
+ cpi->worst_quality -
+ (int)((qadjustment_range * above_base) /
+ (cpi->oxcf.optimal_buffer_level*3>>2));
+ }
+ else
+ {
+ cpi->active_worst_quality = cpi->worst_quality;
+ }
+ }
+ else
+ {
+ cpi->active_worst_quality = cpi->ni_av_qi;
+ }
+ }
+ else
+ {
+ cpi->active_worst_quality = cpi->worst_quality;
+ }
+ }
+ else
+ {
+ int percent_high = 0;
+
+ if ((cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ && (cpi->buffer_level > cpi->oxcf.optimal_buffer_level))
+ {
+ percent_high = (int)((cpi->buffer_level
+ - cpi->oxcf.optimal_buffer_level)
+ / one_percent_bits);
+ }
+ else if (cpi->bits_off_target > cpi->oxcf.optimal_buffer_level)
+ {
+ percent_high = (int)((100 * cpi->bits_off_target)
+ / (cpi->total_byte_count * 8));
+ }
+
+ if (percent_high > cpi->oxcf.over_shoot_pct)
+ percent_high = cpi->oxcf.over_shoot_pct;
+ else if (percent_high < 0)
+ percent_high = 0;
+
+ cpi->this_frame_target += (cpi->this_frame_target *
+ percent_high) / 200;
+
+ /* Are we allowing control of active_worst_allowed_q according
+ * to buffer level.
+ */
+ if (cpi->auto_worst_q && cpi->ni_frames > 150)
+ {
+ /* When using the relaxed buffer model stick to the
+ * user specified value
+ */
+ cpi->active_worst_quality = cpi->ni_av_qi;
+ }
+ else
+ {
+ cpi->active_worst_quality = cpi->worst_quality;
+ }
+ }
+
+ /* Set active_best_quality to prevent quality rising too high */
+ cpi->active_best_quality = cpi->best_quality;
+
+ /* Worst quality obviously must not be better than best quality */
+ if (cpi->active_worst_quality <= cpi->active_best_quality)
+ cpi->active_worst_quality = cpi->active_best_quality + 1;
+
+ if(cpi->active_worst_quality > 127)
+ cpi->active_worst_quality = 127;
+ }
+ /* Unbuffered mode (eg. video conferencing) */
+ else
+ {
+ /* Set the active worst quality */
+ cpi->active_worst_quality = cpi->worst_quality;
+ }
+
+ /* Special trap for constrained quality mode
+ * "active_worst_quality" may never drop below cq level
+ * for any frame type.
+ */
+ if ( cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY &&
+ cpi->active_worst_quality < cpi->cq_target_quality)
+ {
+ cpi->active_worst_quality = cpi->cq_target_quality;
+ }
+ }
+
+ /* Test to see if we have to drop a frame
+ * The auto-drop frame code is only used in buffered mode.
+ * In unbufferd mode (eg vide conferencing) the descision to
+ * code or drop a frame is made outside the codec in response to real
+ * world comms or buffer considerations.
+ */
+ if (cpi->drop_frames_allowed &&
+ (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
+ ((cpi->common.frame_type != KEY_FRAME)))
+ {
+ /* Check for a buffer underun-crisis in which case we have to drop
+ * a frame
+ */
+ if ((cpi->buffer_level < 0))
+ {
+#if 0
+ FILE *f = fopen("dec.stt", "a");
+ fprintf(f, "%10d %10d %10d %10d ***** BUFFER EMPTY\n",
+ (int) cpi->common.current_video_frame,
+ cpi->decimation_factor, cpi->common.horiz_scale,
+ (cpi->buffer_level * 100) / cpi->oxcf.optimal_buffer_level);
+ fclose(f);
+#endif
+ cpi->drop_frame = 1;
+
+ /* Update the buffer level variable. */
+ cpi->bits_off_target += cpi->av_per_frame_bandwidth;
+ if (cpi->bits_off_target > cpi->oxcf.maximum_buffer_size)
+ cpi->bits_off_target = (int)cpi->oxcf.maximum_buffer_size;
+ cpi->buffer_level = cpi->bits_off_target;
+ }
+ }
+
+ /* Adjust target frame size for Golden Frames: */
+ if (cpi->oxcf.error_resilient_mode == 0 &&
+ (cpi->frames_till_gf_update_due == 0) && !cpi->drop_frame)
+ {
+ int Q = (cpi->oxcf.fixed_q < 0) ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
+
+ int gf_frame_useage = 0; /* Golden frame useage since last GF */
+ int tot_mbs = cpi->recent_ref_frame_usage[INTRA_FRAME] +
+ cpi->recent_ref_frame_usage[LAST_FRAME] +
+ cpi->recent_ref_frame_usage[GOLDEN_FRAME] +
+ cpi->recent_ref_frame_usage[ALTREF_FRAME];
+
+ int pct_gf_active = (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols);
+
+ if (tot_mbs)
+ gf_frame_useage = (cpi->recent_ref_frame_usage[GOLDEN_FRAME] + cpi->recent_ref_frame_usage[ALTREF_FRAME]) * 100 / tot_mbs;
+
+ if (pct_gf_active > gf_frame_useage)
+ gf_frame_useage = pct_gf_active;
+
+ /* Is a fixed manual GF frequency being used */
+ if (cpi->auto_gold)
+ {
+ /* For one pass throw a GF if recent frame intra useage is
+ * low or the GF useage is high
+ */
+ if ((cpi->pass == 0) && (cpi->this_frame_percent_intra < 15 || gf_frame_useage >= 5))
+ cpi->common.refresh_golden_frame = 1;
+
+ /* Two pass GF descision */
+ else if (cpi->pass == 2)
+ cpi->common.refresh_golden_frame = 1;
+ }
+
+#if 0
+
+ /* Debug stats */
+ if (0)
+ {
+ FILE *f;
+
+ f = fopen("gf_useaget.stt", "a");
+ fprintf(f, " %8ld %10ld %10ld %10ld %10ld\n",
+ cpi->common.current_video_frame, cpi->gfu_boost, GFQ_ADJUSTMENT, cpi->gfu_boost, gf_frame_useage);
+ fclose(f);
+ }
+
+#endif
+
+ if (cpi->common.refresh_golden_frame == 1)
+ {
+#if 0
+
+ if (0)
+ {
+ FILE *f;
+
+ f = fopen("GFexit.stt", "a");
+ fprintf(f, "%8ld GF coded\n", cpi->common.current_video_frame);
+ fclose(f);
+ }
+
+#endif
+
+ if (cpi->auto_adjust_gold_quantizer)
+ {
+ calc_gf_params(cpi);
+ }
+
+ /* If we are using alternate ref instead of gf then do not apply the
+ * boost It will instead be applied to the altref update Jims
+ * modified boost
+ */
+ if (!cpi->source_alt_ref_active)
+ {
+ if (cpi->oxcf.fixed_q < 0)
+ {
+ if (cpi->pass == 2)
+ {
+ /* The spend on the GF is defined in the two pass
+ * code for two pass encodes
+ */
+ cpi->this_frame_target = cpi->per_frame_bandwidth;
+ }
+ else
+ {
+ int Boost = cpi->last_boost;
+ int frames_in_section = cpi->frames_till_gf_update_due + 1;
+ int allocation_chunks = (frames_in_section * 100) + (Boost - 100);
+ int bits_in_section = cpi->inter_frame_target * frames_in_section;
+
+ /* Normalize Altboost and allocations chunck down to
+ * prevent overflow
+ */
+ while (Boost > 1000)
+ {
+ Boost /= 2;
+ allocation_chunks /= 2;
+ }
+
+ /* Avoid loss of precision but avoid overflow */
+ if ((bits_in_section >> 7) > allocation_chunks)
+ cpi->this_frame_target = Boost * (bits_in_section / allocation_chunks);
+ else
+ cpi->this_frame_target = (Boost * bits_in_section) / allocation_chunks;
+ }
+ }
+ else
+ cpi->this_frame_target =
+ (estimate_bits_at_q(1, Q, cpi->common.MBs, 1.0)
+ * cpi->last_boost) / 100;
+
+ }
+ /* If there is an active ARF at this location use the minimum
+ * bits on this frame even if it is a contructed arf.
+ * The active maximum quantizer insures that an appropriate
+ * number of bits will be spent if needed for contstructed ARFs.
+ */
+ else
+ {
+ cpi->this_frame_target = 0;
+ }
+
+ cpi->current_gf_interval = cpi->frames_till_gf_update_due;
+
+ }
+ }
+
+ cpi->per_frame_bandwidth = old_per_frame_bandwidth;
+}
+
+
+void vp8_update_rate_correction_factors(VP8_COMP *cpi, int damp_var)
+{
+ int Q = cpi->common.base_qindex;
+ int correction_factor = 100;
+ double rate_correction_factor;
+ double adjustment_limit;
+
+ int projected_size_based_on_q = 0;
+
+ /* Clear down mmx registers to allow floating point in what follows */
+ vp8_clear_system_state();
+
+ if (cpi->common.frame_type == KEY_FRAME)
+ {
+ rate_correction_factor = cpi->key_frame_rate_correction_factor;
+ }
+ else
+ {
+ if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
+ rate_correction_factor = cpi->gf_rate_correction_factor;
+ else
+ rate_correction_factor = cpi->rate_correction_factor;
+ }
+
+ /* Work out how big we would have expected the frame to be at this Q
+ * given the current correction factor. Stay in double to avoid int
+ * overflow when values are large
+ */
+ projected_size_based_on_q = (int)(((.5 + rate_correction_factor * vp8_bits_per_mb[cpi->common.frame_type][Q]) * cpi->common.MBs) / (1 << BPER_MB_NORMBITS));
+
+ /* Make some allowance for cpi->zbin_over_quant */
+ if (cpi->zbin_over_quant > 0)
+ {
+ int Z = cpi->zbin_over_quant;
+ double Factor = 0.99;
+ double factor_adjustment = 0.01 / 256.0;
+
+ while (Z > 0)
+ {
+ Z --;
+ projected_size_based_on_q =
+ (int)(Factor * projected_size_based_on_q);
+ Factor += factor_adjustment;
+
+ if (Factor >= 0.999)
+ Factor = 0.999;
+ }
+ }
+
+ /* Work out a size correction factor. */
+ if (projected_size_based_on_q > 0)
+ correction_factor = (100 * cpi->projected_frame_size) / projected_size_based_on_q;
+
+ /* More heavily damped adjustment used if we have been oscillating
+ * either side of target
+ */
+ switch (damp_var)
+ {
+ case 0:
+ adjustment_limit = 0.75;
+ break;
+ case 1:
+ adjustment_limit = 0.375;
+ break;
+ case 2:
+ default:
+ adjustment_limit = 0.25;
+ break;
+ }
+
+ if (correction_factor > 102)
+ {
+ /* We are not already at the worst allowable quality */
+ correction_factor = (int)(100.5 + ((correction_factor - 100) * adjustment_limit));
+ rate_correction_factor = ((rate_correction_factor * correction_factor) / 100);
+
+ /* Keep rate_correction_factor within limits */
+ if (rate_correction_factor > MAX_BPB_FACTOR)
+ rate_correction_factor = MAX_BPB_FACTOR;
+ }
+ else if (correction_factor < 99)
+ {
+ /* We are not already at the best allowable quality */
+ correction_factor = (int)(100.5 - ((100 - correction_factor) * adjustment_limit));
+ rate_correction_factor = ((rate_correction_factor * correction_factor) / 100);
+
+ /* Keep rate_correction_factor within limits */
+ if (rate_correction_factor < MIN_BPB_FACTOR)
+ rate_correction_factor = MIN_BPB_FACTOR;
+ }
+
+ if (cpi->common.frame_type == KEY_FRAME)
+ cpi->key_frame_rate_correction_factor = rate_correction_factor;
+ else
+ {
+ if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
+ cpi->gf_rate_correction_factor = rate_correction_factor;
+ else
+ cpi->rate_correction_factor = rate_correction_factor;
+ }
+}
+
+
+int vp8_regulate_q(VP8_COMP *cpi, int target_bits_per_frame)
+{
+ int Q = cpi->active_worst_quality;
+
+ /* Reset Zbin OQ value */
+ cpi->zbin_over_quant = 0;
+
+ if (cpi->oxcf.fixed_q >= 0)
+ {
+ Q = cpi->oxcf.fixed_q;
+
+ if (cpi->common.frame_type == KEY_FRAME)
+ {
+ Q = cpi->oxcf.key_q;
+ }
+ else if (cpi->common.refresh_alt_ref_frame)
+ {
+ Q = cpi->oxcf.alt_q;
+ }
+ else if (cpi->common.refresh_golden_frame)
+ {
+ Q = cpi->oxcf.gold_q;
+ }
+
+ }
+ else
+ {
+ int i;
+ int last_error = INT_MAX;
+ int target_bits_per_mb;
+ int bits_per_mb_at_this_q;
+ double correction_factor;
+
+ /* Select the appropriate correction factor based upon type of frame. */
+ if (cpi->common.frame_type == KEY_FRAME)
+ correction_factor = cpi->key_frame_rate_correction_factor;
+ else
+ {
+ if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
+ correction_factor = cpi->gf_rate_correction_factor;
+ else
+ correction_factor = cpi->rate_correction_factor;
+ }
+
+ /* Calculate required scaling factor based on target frame size and
+ * size of frame produced using previous Q
+ */
+ if (target_bits_per_frame >= (INT_MAX >> BPER_MB_NORMBITS))
+ /* Case where we would overflow int */
+ target_bits_per_mb = (target_bits_per_frame / cpi->common.MBs) << BPER_MB_NORMBITS;
+ else
+ target_bits_per_mb = (target_bits_per_frame << BPER_MB_NORMBITS) / cpi->common.MBs;
+
+ i = cpi->active_best_quality;
+
+ do
+ {
+ bits_per_mb_at_this_q = (int)(.5 + correction_factor * vp8_bits_per_mb[cpi->common.frame_type][i]);
+
+ if (bits_per_mb_at_this_q <= target_bits_per_mb)
+ {
+ if ((target_bits_per_mb - bits_per_mb_at_this_q) <= last_error)
+ Q = i;
+ else
+ Q = i - 1;
+
+ break;
+ }
+ else
+ last_error = bits_per_mb_at_this_q - target_bits_per_mb;
+ }
+ while (++i <= cpi->active_worst_quality);
+
+
+ /* If we are at MAXQ then enable Q over-run which seeks to claw
+ * back additional bits through things like the RD multiplier
+ * and zero bin size.
+ */
+ if (Q >= MAXQ)
+ {
+ int zbin_oqmax;
+
+ double Factor = 0.99;
+ double factor_adjustment = 0.01 / 256.0;
+
+ if (cpi->common.frame_type == KEY_FRAME)
+ zbin_oqmax = 0;
+ else if (cpi->common.refresh_alt_ref_frame || (cpi->common.refresh_golden_frame && !cpi->source_alt_ref_active))
+ zbin_oqmax = 16;
+ else
+ zbin_oqmax = ZBIN_OQ_MAX;
+
+ /*{
+ double Factor = (double)target_bits_per_mb/(double)bits_per_mb_at_this_q;
+ double Oq;
+
+ Factor = Factor/1.2683;
+
+ Oq = pow( Factor, (1.0/-0.165) );
+
+ if ( Oq > zbin_oqmax )
+ Oq = zbin_oqmax;
+
+ cpi->zbin_over_quant = (int)Oq;
+ }*/
+
+ /* Each incrment in the zbin is assumed to have a fixed effect
+ * on bitrate. This is not of course true. The effect will be
+ * highly clip dependent and may well have sudden steps. The
+ * idea here is to acheive higher effective quantizers than the
+ * normal maximum by expanding the zero bin and hence
+ * decreasing the number of low magnitude non zero coefficients.
+ */
+ while (cpi->zbin_over_quant < zbin_oqmax)
+ {
+ cpi->zbin_over_quant ++;
+
+ if (cpi->zbin_over_quant > zbin_oqmax)
+ cpi->zbin_over_quant = zbin_oqmax;
+
+ /* Adjust bits_per_mb_at_this_q estimate */
+ bits_per_mb_at_this_q = (int)(Factor * bits_per_mb_at_this_q);
+ Factor += factor_adjustment;
+
+ if (Factor >= 0.999)
+ Factor = 0.999;
+
+ /* Break out if we get down to the target rate */
+ if (bits_per_mb_at_this_q <= target_bits_per_mb)
+ break;
+ }
+
+ }
+ }
+
+ return Q;
+}
+
+
+static int estimate_keyframe_frequency(VP8_COMP *cpi)
+{
+ int i;
+
+ /* Average key frame frequency */
+ int av_key_frame_frequency = 0;
+
+ /* First key frame at start of sequence is a special case. We have no
+ * frequency data.
+ */
+ if (cpi->key_frame_count == 1)
+ {
+ /* Assume a default of 1 kf every 2 seconds, or the max kf interval,
+ * whichever is smaller.
+ */
+ int key_freq = cpi->oxcf.key_freq>0 ? cpi->oxcf.key_freq : 1;
+ av_key_frame_frequency = (int)cpi->output_frame_rate * 2;
+
+ if (cpi->oxcf.auto_key && av_key_frame_frequency > key_freq)
+ av_key_frame_frequency = cpi->oxcf.key_freq;
+
+ cpi->prior_key_frame_distance[KEY_FRAME_CONTEXT - 1]
+ = av_key_frame_frequency;
+ }
+ else
+ {
+ unsigned int total_weight = 0;
+ int last_kf_interval =
+ (cpi->frames_since_key > 0) ? cpi->frames_since_key : 1;
+
+ /* reset keyframe context and calculate weighted average of last
+ * KEY_FRAME_CONTEXT keyframes
+ */
+ for (i = 0; i < KEY_FRAME_CONTEXT; i++)
+ {
+ if (i < KEY_FRAME_CONTEXT - 1)
+ cpi->prior_key_frame_distance[i]
+ = cpi->prior_key_frame_distance[i+1];
+ else
+ cpi->prior_key_frame_distance[i] = last_kf_interval;
+
+ av_key_frame_frequency += prior_key_frame_weight[i]
+ * cpi->prior_key_frame_distance[i];
+ total_weight += prior_key_frame_weight[i];
+ }
+
+ av_key_frame_frequency /= total_weight;
+
+ }
+ return av_key_frame_frequency;
+}
+
+
+void vp8_adjust_key_frame_context(VP8_COMP *cpi)
+{
+ /* Clear down mmx registers to allow floating point in what follows */
+ vp8_clear_system_state();
+
+ /* Do we have any key frame overspend to recover? */
+ /* Two-pass overspend handled elsewhere. */
+ if ((cpi->pass != 2)
+ && (cpi->projected_frame_size > cpi->per_frame_bandwidth))
+ {
+ int overspend;
+
+ /* Update the count of key frame overspend to be recovered in
+ * subsequent frames. A portion of the KF overspend is treated as gf
+ * overspend (and hence recovered more quickly) as the kf is also a
+ * gf. Otherwise the few frames following each kf tend to get more
+ * bits allocated than those following other gfs.
+ */
+ overspend = (cpi->projected_frame_size - cpi->per_frame_bandwidth);
+
+ if (cpi->oxcf.number_of_layers > 1)
+ cpi->kf_overspend_bits += overspend;
+ else
+ {
+ cpi->kf_overspend_bits += overspend * 7 / 8;
+ cpi->gf_overspend_bits += overspend * 1 / 8;
+ }
+
+ /* Work out how much to try and recover per frame. */
+ cpi->kf_bitrate_adjustment = cpi->kf_overspend_bits
+ / estimate_keyframe_frequency(cpi);
+ }
+
+ cpi->frames_since_key = 0;
+ cpi->key_frame_count++;
+}
+
+
+void vp8_compute_frame_size_bounds(VP8_COMP *cpi, int *frame_under_shoot_limit, int *frame_over_shoot_limit)
+{
+ /* Set-up bounds on acceptable frame size: */
+ if (cpi->oxcf.fixed_q >= 0)
+ {
+ /* Fixed Q scenario: frame size never outranges target
+ * (there is no target!)
+ */
+ *frame_under_shoot_limit = 0;
+ *frame_over_shoot_limit = INT_MAX;
+ }
+ else
+ {
+ if (cpi->common.frame_type == KEY_FRAME)
+ {
+ *frame_over_shoot_limit = cpi->this_frame_target * 9 / 8;
+ *frame_under_shoot_limit = cpi->this_frame_target * 7 / 8;
+ }
+ else
+ {
+ if (cpi->oxcf.number_of_layers > 1 ||
+ cpi->common.refresh_alt_ref_frame ||
+ cpi->common.refresh_golden_frame)
+ {
+ *frame_over_shoot_limit = cpi->this_frame_target * 9 / 8;
+ *frame_under_shoot_limit = cpi->this_frame_target * 7 / 8;
+ }
+ else
+ {
+ /* For CBR take buffer fullness into account */
+ if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ {
+ if (cpi->buffer_level >= ((cpi->oxcf.optimal_buffer_level + cpi->oxcf.maximum_buffer_size) >> 1))
+ {
+ /* Buffer is too full so relax overshoot and tighten
+ * undershoot
+ */
+ *frame_over_shoot_limit = cpi->this_frame_target * 12 / 8;
+ *frame_under_shoot_limit = cpi->this_frame_target * 6 / 8;
+ }
+ else if (cpi->buffer_level <= (cpi->oxcf.optimal_buffer_level >> 1))
+ {
+ /* Buffer is too low so relax undershoot and tighten
+ * overshoot
+ */
+ *frame_over_shoot_limit = cpi->this_frame_target * 10 / 8;
+ *frame_under_shoot_limit = cpi->this_frame_target * 4 / 8;
+ }
+ else
+ {
+ *frame_over_shoot_limit = cpi->this_frame_target * 11 / 8;
+ *frame_under_shoot_limit = cpi->this_frame_target * 5 / 8;
+ }
+ }
+ /* VBR and CQ mode */
+ /* Note that tighter restrictions here can help quality
+ * but hurt encode speed
+ */
+ else
+ {
+ /* Stron overshoot limit for constrained quality */
+ if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
+ {
+ *frame_over_shoot_limit = cpi->this_frame_target * 11 / 8;
+ *frame_under_shoot_limit = cpi->this_frame_target * 2 / 8;
+ }
+ else
+ {
+ *frame_over_shoot_limit = cpi->this_frame_target * 11 / 8;
+ *frame_under_shoot_limit = cpi->this_frame_target * 5 / 8;
+ }
+ }
+ }
+ }
+
+ /* For very small rate targets where the fractional adjustment
+ * (eg * 7/8) may be tiny make sure there is at least a minimum
+ * range.
+ */
+ *frame_over_shoot_limit += 200;
+ *frame_under_shoot_limit -= 200;
+ if ( *frame_under_shoot_limit < 0 )
+ *frame_under_shoot_limit = 0;
+
+ }
+}
+
+
+/* return of 0 means drop frame */
+int vp8_pick_frame_size(VP8_COMP *cpi)
+{
+ VP8_COMMON *cm = &cpi->common;
+
+ if (cm->frame_type == KEY_FRAME)
+ calc_iframe_target_size(cpi);
+ else
+ {
+ calc_pframe_target_size(cpi);
+
+ /* Check if we're dropping the frame: */
+ if (cpi->drop_frame)
+ {
+ cpi->drop_frame = 0;
+ return 0;
+ }
+ }
+ return 1;
+}
diff --git a/libvpx/vp8/encoder/ratectrl.h b/libvpx/vp8/encoder/ratectrl.h
new file mode 100644
index 0000000..c43f08d
--- /dev/null
+++ b/libvpx/vp8/encoder/ratectrl.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#if !defined __INC_RATECTRL_H
+
+#include "onyx_int.h"
+
+extern void vp8_save_coding_context(VP8_COMP *cpi);
+extern void vp8_restore_coding_context(VP8_COMP *cpi);
+
+extern void vp8_setup_key_frame(VP8_COMP *cpi);
+extern void vp8_update_rate_correction_factors(VP8_COMP *cpi, int damp_var);
+extern int vp8_regulate_q(VP8_COMP *cpi, int target_bits_per_frame);
+extern void vp8_adjust_key_frame_context(VP8_COMP *cpi);
+extern void vp8_compute_frame_size_bounds(VP8_COMP *cpi, int *frame_under_shoot_limit, int *frame_over_shoot_limit);
+
+/* return of 0 means drop frame */
+extern int vp8_pick_frame_size(VP8_COMP *cpi);
+
+#endif
diff --git a/libvpx/vp8/encoder/rdopt.c b/libvpx/vp8/encoder/rdopt.c
new file mode 100644
index 0000000..7d80606
--- /dev/null
+++ b/libvpx/vp8/encoder/rdopt.c
@@ -0,0 +1,2629 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdio.h>
+#include <math.h>
+#include <limits.h>
+#include <assert.h>
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/common/pragmas.h"
+#include "tokenize.h"
+#include "treewriter.h"
+#include "onyx_int.h"
+#include "modecosts.h"
+#include "encodeintra.h"
+#include "pickinter.h"
+#include "vp8/common/entropymode.h"
+#include "vp8/common/reconinter.h"
+#include "vp8/common/reconintra4x4.h"
+#include "vp8/common/findnearmv.h"
+#include "vp8/common/quant_common.h"
+#include "encodemb.h"
+#include "quantize.h"
+#include "vp8/common/variance.h"
+#include "mcomp.h"
+#include "rdopt.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/systemdependent.h"
+#if CONFIG_TEMPORAL_DENOISING
+#include "denoising.h"
+#endif
+extern void vp8_update_zbin_extra(VP8_COMP *cpi, MACROBLOCK *x);
+
+#define MAXF(a,b) (((a) > (b)) ? (a) : (b))
+
+typedef struct rate_distortion_struct
+{
+ int rate2;
+ int rate_y;
+ int rate_uv;
+ int distortion2;
+ int distortion_uv;
+} RATE_DISTORTION;
+
+typedef struct best_mode_struct
+{
+ int yrd;
+ int rd;
+ int intra_rd;
+ MB_MODE_INFO mbmode;
+ union b_mode_info bmodes[16];
+ PARTITION_INFO partition;
+} BEST_MODE;
+
+static const int auto_speed_thresh[17] =
+{
+ 1000,
+ 200,
+ 150,
+ 130,
+ 150,
+ 125,
+ 120,
+ 115,
+ 115,
+ 115,
+ 115,
+ 115,
+ 115,
+ 115,
+ 115,
+ 115,
+ 105
+};
+
+const MB_PREDICTION_MODE vp8_mode_order[MAX_MODES] =
+{
+ ZEROMV,
+ DC_PRED,
+
+ NEARESTMV,
+ NEARMV,
+
+ ZEROMV,
+ NEARESTMV,
+
+ ZEROMV,
+ NEARESTMV,
+
+ NEARMV,
+ NEARMV,
+
+ V_PRED,
+ H_PRED,
+ TM_PRED,
+
+ NEWMV,
+ NEWMV,
+ NEWMV,
+
+ SPLITMV,
+ SPLITMV,
+ SPLITMV,
+
+ B_PRED,
+};
+
+/* This table determines the search order in reference frame priority order,
+ * which may not necessarily match INTRA,LAST,GOLDEN,ARF
+ */
+const int vp8_ref_frame_order[MAX_MODES] =
+{
+ 1,
+ 0,
+
+ 1,
+ 1,
+
+ 2,
+ 2,
+
+ 3,
+ 3,
+
+ 2,
+ 3,
+
+ 0,
+ 0,
+ 0,
+
+ 1,
+ 2,
+ 3,
+
+ 1,
+ 2,
+ 3,
+
+ 0,
+};
+
+static void fill_token_costs(
+ int c[BLOCK_TYPES][COEF_BANDS][PREV_COEF_CONTEXTS][MAX_ENTROPY_TOKENS],
+ const vp8_prob p[BLOCK_TYPES][COEF_BANDS][PREV_COEF_CONTEXTS][ENTROPY_NODES]
+)
+{
+ int i, j, k;
+
+
+ for (i = 0; i < BLOCK_TYPES; i++)
+ for (j = 0; j < COEF_BANDS; j++)
+ for (k = 0; k < PREV_COEF_CONTEXTS; k++)
+
+ /* check for pt=0 and band > 1 if block type 0
+ * and 0 if blocktype 1
+ */
+ if (k == 0 && j > (i == 0))
+ vp8_cost_tokens2(c[i][j][k], p [i][j][k], vp8_coef_tree, 2);
+ else
+ vp8_cost_tokens(c[i][j][k], p [i][j][k], vp8_coef_tree);
+}
+
+static const int rd_iifactor[32] =
+{
+ 4, 4, 3, 2, 1, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0
+};
+
+/* values are now correlated to quantizer */
+static const int sad_per_bit16lut[QINDEX_RANGE] =
+{
+ 2, 2, 2, 2, 2, 2, 2, 2,
+ 2, 2, 2, 2, 2, 2, 2, 2,
+ 3, 3, 3, 3, 3, 3, 3, 3,
+ 3, 3, 3, 3, 3, 3, 4, 4,
+ 4, 4, 4, 4, 4, 4, 4, 4,
+ 4, 4, 5, 5, 5, 5, 5, 5,
+ 5, 5, 5, 5, 5, 5, 6, 6,
+ 6, 6, 6, 6, 6, 6, 6, 6,
+ 6, 6, 7, 7, 7, 7, 7, 7,
+ 7, 7, 7, 7, 7, 7, 8, 8,
+ 8, 8, 8, 8, 8, 8, 8, 8,
+ 8, 8, 9, 9, 9, 9, 9, 9,
+ 9, 9, 9, 9, 9, 9, 10, 10,
+ 10, 10, 10, 10, 10, 10, 11, 11,
+ 11, 11, 11, 11, 12, 12, 12, 12,
+ 12, 12, 13, 13, 13, 13, 14, 14
+};
+static const int sad_per_bit4lut[QINDEX_RANGE] =
+{
+ 2, 2, 2, 2, 2, 2, 3, 3,
+ 3, 3, 3, 3, 3, 3, 3, 3,
+ 3, 3, 3, 3, 4, 4, 4, 4,
+ 4, 4, 4, 4, 4, 4, 5, 5,
+ 5, 5, 5, 5, 6, 6, 6, 6,
+ 6, 6, 6, 6, 6, 6, 6, 6,
+ 7, 7, 7, 7, 7, 7, 7, 7,
+ 7, 7, 7, 7, 7, 8, 8, 8,
+ 8, 8, 9, 9, 9, 9, 9, 9,
+ 10, 10, 10, 10, 10, 10, 10, 10,
+ 11, 11, 11, 11, 11, 11, 11, 11,
+ 12, 12, 12, 12, 12, 12, 12, 12,
+ 13, 13, 13, 13, 13, 13, 13, 14,
+ 14, 14, 14, 14, 15, 15, 15, 15,
+ 16, 16, 16, 16, 17, 17, 17, 18,
+ 18, 18, 19, 19, 19, 20, 20, 20,
+};
+
+void vp8cx_initialize_me_consts(VP8_COMP *cpi, int QIndex)
+{
+ cpi->mb.sadperbit16 = sad_per_bit16lut[QIndex];
+ cpi->mb.sadperbit4 = sad_per_bit4lut[QIndex];
+}
+
+void vp8_initialize_rd_consts(VP8_COMP *cpi, int Qvalue)
+{
+ int q;
+ int i;
+ double capped_q = (Qvalue < 160) ? (double)Qvalue : 160.0;
+ double rdconst = 2.80;
+
+ vp8_clear_system_state();
+
+ /* Further tests required to see if optimum is different
+ * for key frames, golden frames and arf frames.
+ */
+ cpi->RDMULT = (int)(rdconst * (capped_q * capped_q));
+
+ /* Extend rate multiplier along side quantizer zbin increases */
+ if (cpi->zbin_over_quant > 0)
+ {
+ double oq_factor;
+ double modq;
+
+ /* Experimental code using the same basic equation as used for Q above
+ * The units of cpi->zbin_over_quant are 1/128 of Q bin size
+ */
+ oq_factor = 1.0 + ((double)0.0015625 * cpi->zbin_over_quant);
+ modq = (int)((double)capped_q * oq_factor);
+ cpi->RDMULT = (int)(rdconst * (modq * modq));
+ }
+
+ if (cpi->pass == 2 && (cpi->common.frame_type != KEY_FRAME))
+ {
+ if (cpi->twopass.next_iiratio > 31)
+ cpi->RDMULT += (cpi->RDMULT * rd_iifactor[31]) >> 4;
+ else
+ cpi->RDMULT +=
+ (cpi->RDMULT * rd_iifactor[cpi->twopass.next_iiratio]) >> 4;
+ }
+
+ cpi->mb.errorperbit = (cpi->RDMULT / 110);
+ cpi->mb.errorperbit += (cpi->mb.errorperbit==0);
+
+ vp8_set_speed_features(cpi);
+
+ q = (int)pow(Qvalue, 1.25);
+
+ if (q < 8)
+ q = 8;
+
+ if (cpi->RDMULT > 1000)
+ {
+ cpi->RDDIV = 1;
+ cpi->RDMULT /= 100;
+
+ for (i = 0; i < MAX_MODES; i++)
+ {
+ if (cpi->sf.thresh_mult[i] < INT_MAX)
+ {
+ cpi->rd_threshes[i] = cpi->sf.thresh_mult[i] * q / 100;
+ }
+ else
+ {
+ cpi->rd_threshes[i] = INT_MAX;
+ }
+
+ cpi->rd_baseline_thresh[i] = cpi->rd_threshes[i];
+ }
+ }
+ else
+ {
+ cpi->RDDIV = 100;
+
+ for (i = 0; i < MAX_MODES; i++)
+ {
+ if (cpi->sf.thresh_mult[i] < (INT_MAX / q))
+ {
+ cpi->rd_threshes[i] = cpi->sf.thresh_mult[i] * q;
+ }
+ else
+ {
+ cpi->rd_threshes[i] = INT_MAX;
+ }
+
+ cpi->rd_baseline_thresh[i] = cpi->rd_threshes[i];
+ }
+ }
+
+ {
+ /* build token cost array for the type of frame we have now */
+ FRAME_CONTEXT *l = &cpi->lfc_n;
+
+ if(cpi->common.refresh_alt_ref_frame)
+ l = &cpi->lfc_a;
+ else if(cpi->common.refresh_golden_frame)
+ l = &cpi->lfc_g;
+
+ fill_token_costs(
+ cpi->mb.token_costs,
+ (const vp8_prob( *)[8][3][11]) l->coef_probs
+ );
+ /*
+ fill_token_costs(
+ cpi->mb.token_costs,
+ (const vp8_prob( *)[8][3][11]) cpi->common.fc.coef_probs);
+ */
+
+
+ /* TODO make these mode costs depend on last,alt or gold too. (jbb) */
+ vp8_init_mode_costs(cpi);
+ }
+
+}
+
+void vp8_auto_select_speed(VP8_COMP *cpi)
+{
+ int milliseconds_for_compress = (int)(1000000 / cpi->frame_rate);
+
+ milliseconds_for_compress = milliseconds_for_compress * (16 - cpi->oxcf.cpu_used) / 16;
+
+#if 0
+
+ if (0)
+ {
+ FILE *f;
+
+ f = fopen("speed.stt", "a");
+ fprintf(f, " %8ld %10ld %10ld %10ld\n",
+ cpi->common.current_video_frame, cpi->Speed, milliseconds_for_compress, cpi->avg_pick_mode_time);
+ fclose(f);
+ }
+
+#endif
+
+ if (cpi->avg_pick_mode_time < milliseconds_for_compress && (cpi->avg_encode_time - cpi->avg_pick_mode_time) < milliseconds_for_compress)
+ {
+ if (cpi->avg_pick_mode_time == 0)
+ {
+ cpi->Speed = 4;
+ }
+ else
+ {
+ if (milliseconds_for_compress * 100 < cpi->avg_encode_time * 95)
+ {
+ cpi->Speed += 2;
+ cpi->avg_pick_mode_time = 0;
+ cpi->avg_encode_time = 0;
+
+ if (cpi->Speed > 16)
+ {
+ cpi->Speed = 16;
+ }
+ }
+
+ if (milliseconds_for_compress * 100 > cpi->avg_encode_time * auto_speed_thresh[cpi->Speed])
+ {
+ cpi->Speed -= 1;
+ cpi->avg_pick_mode_time = 0;
+ cpi->avg_encode_time = 0;
+
+ /* In real-time mode, cpi->speed is in [4, 16]. */
+ if (cpi->Speed < 4)
+ {
+ cpi->Speed = 4;
+ }
+ }
+ }
+ }
+ else
+ {
+ cpi->Speed += 4;
+
+ if (cpi->Speed > 16)
+ cpi->Speed = 16;
+
+
+ cpi->avg_pick_mode_time = 0;
+ cpi->avg_encode_time = 0;
+ }
+}
+
+int vp8_block_error_c(short *coeff, short *dqcoeff)
+{
+ int i;
+ int error = 0;
+
+ for (i = 0; i < 16; i++)
+ {
+ int this_diff = coeff[i] - dqcoeff[i];
+ error += this_diff * this_diff;
+ }
+
+ return error;
+}
+
+int vp8_mbblock_error_c(MACROBLOCK *mb, int dc)
+{
+ BLOCK *be;
+ BLOCKD *bd;
+ int i, j;
+ int berror, error = 0;
+
+ for (i = 0; i < 16; i++)
+ {
+ be = &mb->block[i];
+ bd = &mb->e_mbd.block[i];
+
+ berror = 0;
+
+ for (j = dc; j < 16; j++)
+ {
+ int this_diff = be->coeff[j] - bd->dqcoeff[j];
+ berror += this_diff * this_diff;
+ }
+
+ error += berror;
+ }
+
+ return error;
+}
+
+int vp8_mbuverror_c(MACROBLOCK *mb)
+{
+
+ BLOCK *be;
+ BLOCKD *bd;
+
+
+ int i;
+ int error = 0;
+
+ for (i = 16; i < 24; i++)
+ {
+ be = &mb->block[i];
+ bd = &mb->e_mbd.block[i];
+
+ error += vp8_block_error_c(be->coeff, bd->dqcoeff);
+ }
+
+ return error;
+}
+
+int VP8_UVSSE(MACROBLOCK *x)
+{
+ unsigned char *uptr, *vptr;
+ unsigned char *upred_ptr = (*(x->block[16].base_src) + x->block[16].src);
+ unsigned char *vpred_ptr = (*(x->block[20].base_src) + x->block[20].src);
+ int uv_stride = x->block[16].src_stride;
+
+ unsigned int sse1 = 0;
+ unsigned int sse2 = 0;
+ int mv_row = x->e_mbd.mode_info_context->mbmi.mv.as_mv.row;
+ int mv_col = x->e_mbd.mode_info_context->mbmi.mv.as_mv.col;
+ int offset;
+ int pre_stride = x->e_mbd.pre.uv_stride;
+
+ if (mv_row < 0)
+ mv_row -= 1;
+ else
+ mv_row += 1;
+
+ if (mv_col < 0)
+ mv_col -= 1;
+ else
+ mv_col += 1;
+
+ mv_row /= 2;
+ mv_col /= 2;
+
+ offset = (mv_row >> 3) * pre_stride + (mv_col >> 3);
+ uptr = x->e_mbd.pre.u_buffer + offset;
+ vptr = x->e_mbd.pre.v_buffer + offset;
+
+ if ((mv_row | mv_col) & 7)
+ {
+ vp8_sub_pixel_variance8x8(uptr, pre_stride,
+ mv_col & 7, mv_row & 7, upred_ptr, uv_stride, &sse2);
+ vp8_sub_pixel_variance8x8(vptr, pre_stride,
+ mv_col & 7, mv_row & 7, vpred_ptr, uv_stride, &sse1);
+ sse2 += sse1;
+ }
+ else
+ {
+ vp8_variance8x8(uptr, pre_stride,
+ upred_ptr, uv_stride, &sse2);
+ vp8_variance8x8(vptr, pre_stride,
+ vpred_ptr, uv_stride, &sse1);
+ sse2 += sse1;
+ }
+ return sse2;
+
+}
+
+static int cost_coeffs(MACROBLOCK *mb, BLOCKD *b, int type, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l)
+{
+ int c = !type; /* start at coef 0, unless Y with Y2 */
+ int eob = (int)(*b->eob);
+ int pt ; /* surrounding block/prev coef predictor */
+ int cost = 0;
+ short *qcoeff_ptr = b->qcoeff;
+
+ VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
+
+# define QC( I) ( qcoeff_ptr [vp8_default_zig_zag1d[I]] )
+
+ for (; c < eob; c++)
+ {
+ int v = QC(c);
+ int t = vp8_dct_value_tokens_ptr[v].Token;
+ cost += mb->token_costs [type] [vp8_coef_bands[c]] [pt] [t];
+ cost += vp8_dct_value_cost_ptr[v];
+ pt = vp8_prev_token_class[t];
+ }
+
+# undef QC
+
+ if (c < 16)
+ cost += mb->token_costs [type] [vp8_coef_bands[c]] [pt] [DCT_EOB_TOKEN];
+
+ pt = (c != !type); /* is eob first coefficient; */
+ *a = *l = pt;
+
+ return cost;
+}
+
+static int vp8_rdcost_mby(MACROBLOCK *mb)
+{
+ int cost = 0;
+ int b;
+ MACROBLOCKD *x = &mb->e_mbd;
+ ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT *ta;
+ ENTROPY_CONTEXT *tl;
+
+ vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta = (ENTROPY_CONTEXT *)&t_above;
+ tl = (ENTROPY_CONTEXT *)&t_left;
+
+ for (b = 0; b < 16; b++)
+ cost += cost_coeffs(mb, x->block + b, PLANE_TYPE_Y_NO_DC,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+
+ cost += cost_coeffs(mb, x->block + 24, PLANE_TYPE_Y2,
+ ta + vp8_block2above[24], tl + vp8_block2left[24]);
+
+ return cost;
+}
+
+static void macro_block_yrd( MACROBLOCK *mb,
+ int *Rate,
+ int *Distortion)
+{
+ int b;
+ MACROBLOCKD *const x = &mb->e_mbd;
+ BLOCK *const mb_y2 = mb->block + 24;
+ BLOCKD *const x_y2 = x->block + 24;
+ short *Y2DCPtr = mb_y2->src_diff;
+ BLOCK *beptr;
+ int d;
+
+ vp8_subtract_mby( mb->src_diff, *(mb->block[0].base_src),
+ mb->block[0].src_stride, mb->e_mbd.predictor, 16);
+
+ /* Fdct and building the 2nd order block */
+ for (beptr = mb->block; beptr < mb->block + 16; beptr += 2)
+ {
+ mb->short_fdct8x4(beptr->src_diff, beptr->coeff, 32);
+ *Y2DCPtr++ = beptr->coeff[0];
+ *Y2DCPtr++ = beptr->coeff[16];
+ }
+
+ /* 2nd order fdct */
+ mb->short_walsh4x4(mb_y2->src_diff, mb_y2->coeff, 8);
+
+ /* Quantization */
+ for (b = 0; b < 16; b++)
+ {
+ mb->quantize_b(&mb->block[b], &mb->e_mbd.block[b]);
+ }
+
+ /* DC predication and Quantization of 2nd Order block */
+ mb->quantize_b(mb_y2, x_y2);
+
+ /* Distortion */
+ d = vp8_mbblock_error(mb, 1) << 2;
+ d += vp8_block_error(mb_y2->coeff, x_y2->dqcoeff);
+
+ *Distortion = (d >> 4);
+
+ /* rate */
+ *Rate = vp8_rdcost_mby(mb);
+}
+
+static void copy_predictor(unsigned char *dst, const unsigned char *predictor)
+{
+ const unsigned int *p = (const unsigned int *)predictor;
+ unsigned int *d = (unsigned int *)dst;
+ d[0] = p[0];
+ d[4] = p[4];
+ d[8] = p[8];
+ d[12] = p[12];
+}
+static int rd_pick_intra4x4block(
+ MACROBLOCK *x,
+ BLOCK *be,
+ BLOCKD *b,
+ B_PREDICTION_MODE *best_mode,
+ const int *bmode_costs,
+ ENTROPY_CONTEXT *a,
+ ENTROPY_CONTEXT *l,
+
+ int *bestrate,
+ int *bestratey,
+ int *bestdistortion)
+{
+ B_PREDICTION_MODE mode;
+ int best_rd = INT_MAX;
+ int rate = 0;
+ int distortion;
+
+ ENTROPY_CONTEXT ta = *a, tempa = *a;
+ ENTROPY_CONTEXT tl = *l, templ = *l;
+ /*
+ * The predictor buffer is a 2d buffer with a stride of 16. Create
+ * a temp buffer that meets the stride requirements, but we are only
+ * interested in the left 4x4 block
+ * */
+ DECLARE_ALIGNED_ARRAY(16, unsigned char, best_predictor, 16*4);
+ DECLARE_ALIGNED_ARRAY(16, short, best_dqcoeff, 16);
+ int dst_stride = x->e_mbd.dst.y_stride;
+ unsigned char *dst = x->e_mbd.dst.y_buffer + b->offset;
+
+ unsigned char *Above = dst - dst_stride;
+ unsigned char *yleft = dst - 1;
+ unsigned char top_left = Above[-1];
+
+ for (mode = B_DC_PRED; mode <= B_HU_PRED; mode++)
+ {
+ int this_rd;
+ int ratey;
+
+ rate = bmode_costs[mode];
+
+ vp8_intra4x4_predict(Above, yleft, dst_stride, mode,
+ b->predictor, 16, top_left);
+ vp8_subtract_b(be, b, 16);
+ x->short_fdct4x4(be->src_diff, be->coeff, 32);
+ x->quantize_b(be, b);
+
+ tempa = ta;
+ templ = tl;
+
+ ratey = cost_coeffs(x, b, PLANE_TYPE_Y_WITH_DC, &tempa, &templ);
+ rate += ratey;
+ distortion = vp8_block_error(be->coeff, b->dqcoeff) >> 2;
+
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
+
+ if (this_rd < best_rd)
+ {
+ *bestrate = rate;
+ *bestratey = ratey;
+ *bestdistortion = distortion;
+ best_rd = this_rd;
+ *best_mode = mode;
+ *a = tempa;
+ *l = templ;
+ copy_predictor(best_predictor, b->predictor);
+ vpx_memcpy(best_dqcoeff, b->dqcoeff, 32);
+ }
+ }
+ b->bmi.as_mode = *best_mode;
+
+ vp8_short_idct4x4llm(best_dqcoeff, best_predictor, 16, dst, dst_stride);
+
+ return best_rd;
+}
+
+static int rd_pick_intra4x4mby_modes(MACROBLOCK *mb, int *Rate,
+ int *rate_y, int *Distortion, int best_rd)
+{
+ MACROBLOCKD *const xd = &mb->e_mbd;
+ int i;
+ int cost = mb->mbmode_cost [xd->frame_type] [B_PRED];
+ int distortion = 0;
+ int tot_rate_y = 0;
+ int64_t total_rd = 0;
+ ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT *ta;
+ ENTROPY_CONTEXT *tl;
+ const int *bmode_costs;
+
+ vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta = (ENTROPY_CONTEXT *)&t_above;
+ tl = (ENTROPY_CONTEXT *)&t_left;
+
+ intra_prediction_down_copy(xd, xd->dst.y_buffer - xd->dst.y_stride + 16);
+
+ bmode_costs = mb->inter_bmode_costs;
+
+ for (i = 0; i < 16; i++)
+ {
+ MODE_INFO *const mic = xd->mode_info_context;
+ const int mis = xd->mode_info_stride;
+ B_PREDICTION_MODE UNINITIALIZED_IS_SAFE(best_mode);
+ int UNINITIALIZED_IS_SAFE(r), UNINITIALIZED_IS_SAFE(ry), UNINITIALIZED_IS_SAFE(d);
+
+ if (mb->e_mbd.frame_type == KEY_FRAME)
+ {
+ const B_PREDICTION_MODE A = above_block_mode(mic, i, mis);
+ const B_PREDICTION_MODE L = left_block_mode(mic, i);
+
+ bmode_costs = mb->bmode_costs[A][L];
+ }
+
+ total_rd += rd_pick_intra4x4block(
+ mb, mb->block + i, xd->block + i, &best_mode, bmode_costs,
+ ta + vp8_block2above[i],
+ tl + vp8_block2left[i], &r, &ry, &d);
+
+ cost += r;
+ distortion += d;
+ tot_rate_y += ry;
+
+ mic->bmi[i].as_mode = best_mode;
+
+ if(total_rd >= (int64_t)best_rd)
+ break;
+ }
+
+ if(total_rd >= (int64_t)best_rd)
+ return INT_MAX;
+
+ *Rate = cost;
+ *rate_y = tot_rate_y;
+ *Distortion = distortion;
+
+ return RDCOST(mb->rdmult, mb->rddiv, cost, distortion);
+}
+
+
+static int rd_pick_intra16x16mby_mode(MACROBLOCK *x,
+ int *Rate,
+ int *rate_y,
+ int *Distortion)
+{
+ MB_PREDICTION_MODE mode;
+ MB_PREDICTION_MODE UNINITIALIZED_IS_SAFE(mode_selected);
+ int rate, ratey;
+ int distortion;
+ int best_rd = INT_MAX;
+ int this_rd;
+ MACROBLOCKD *xd = &x->e_mbd;
+
+ /* Y Search for 16x16 intra prediction mode */
+ for (mode = DC_PRED; mode <= TM_PRED; mode++)
+ {
+ xd->mode_info_context->mbmi.mode = mode;
+
+ vp8_build_intra_predictors_mby_s(xd,
+ xd->dst.y_buffer - xd->dst.y_stride,
+ xd->dst.y_buffer - 1,
+ xd->dst.y_stride,
+ xd->predictor,
+ 16);
+
+ macro_block_yrd(x, &ratey, &distortion);
+ rate = ratey + x->mbmode_cost[xd->frame_type]
+ [xd->mode_info_context->mbmi.mode];
+
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
+
+ if (this_rd < best_rd)
+ {
+ mode_selected = mode;
+ best_rd = this_rd;
+ *Rate = rate;
+ *rate_y = ratey;
+ *Distortion = distortion;
+ }
+ }
+
+ xd->mode_info_context->mbmi.mode = mode_selected;
+ return best_rd;
+}
+
+static int rd_cost_mbuv(MACROBLOCK *mb)
+{
+ int b;
+ int cost = 0;
+ MACROBLOCKD *x = &mb->e_mbd;
+ ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT *ta;
+ ENTROPY_CONTEXT *tl;
+
+ vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta = (ENTROPY_CONTEXT *)&t_above;
+ tl = (ENTROPY_CONTEXT *)&t_left;
+
+ for (b = 16; b < 24; b++)
+ cost += cost_coeffs(mb, x->block + b, PLANE_TYPE_UV,
+ ta + vp8_block2above[b], tl + vp8_block2left[b]);
+
+ return cost;
+}
+
+
+static int rd_inter16x16_uv(VP8_COMP *cpi, MACROBLOCK *x, int *rate,
+ int *distortion, int fullpixel)
+{
+ vp8_build_inter16x16_predictors_mbuv(&x->e_mbd);
+ vp8_subtract_mbuv(x->src_diff,
+ x->src.u_buffer, x->src.v_buffer, x->src.uv_stride,
+ &x->e_mbd.predictor[256], &x->e_mbd.predictor[320], 8);
+
+ vp8_transform_mbuv(x);
+ vp8_quantize_mbuv(x);
+
+ *rate = rd_cost_mbuv(x);
+ *distortion = vp8_mbuverror(x) / 4;
+
+ return RDCOST(x->rdmult, x->rddiv, *rate, *distortion);
+}
+
+static int rd_inter4x4_uv(VP8_COMP *cpi, MACROBLOCK *x, int *rate,
+ int *distortion, int fullpixel)
+{
+ vp8_build_inter4x4_predictors_mbuv(&x->e_mbd);
+ vp8_subtract_mbuv(x->src_diff,
+ x->src.u_buffer, x->src.v_buffer, x->src.uv_stride,
+ &x->e_mbd.predictor[256], &x->e_mbd.predictor[320], 8);
+
+ vp8_transform_mbuv(x);
+ vp8_quantize_mbuv(x);
+
+ *rate = rd_cost_mbuv(x);
+ *distortion = vp8_mbuverror(x) / 4;
+
+ return RDCOST(x->rdmult, x->rddiv, *rate, *distortion);
+}
+
+static void rd_pick_intra_mbuv_mode(MACROBLOCK *x, int *rate,
+ int *rate_tokenonly, int *distortion)
+{
+ MB_PREDICTION_MODE mode;
+ MB_PREDICTION_MODE UNINITIALIZED_IS_SAFE(mode_selected);
+ int best_rd = INT_MAX;
+ int UNINITIALIZED_IS_SAFE(d), UNINITIALIZED_IS_SAFE(r);
+ int rate_to;
+ MACROBLOCKD *xd = &x->e_mbd;
+
+ for (mode = DC_PRED; mode <= TM_PRED; mode++)
+ {
+ int rate;
+ int distortion;
+ int this_rd;
+
+ xd->mode_info_context->mbmi.uv_mode = mode;
+
+ vp8_build_intra_predictors_mbuv_s(xd,
+ xd->dst.u_buffer - xd->dst.uv_stride,
+ xd->dst.v_buffer - xd->dst.uv_stride,
+ xd->dst.u_buffer - 1,
+ xd->dst.v_buffer - 1,
+ xd->dst.uv_stride,
+ &xd->predictor[256], &xd->predictor[320],
+ 8);
+
+
+ vp8_subtract_mbuv(x->src_diff,
+ x->src.u_buffer, x->src.v_buffer, x->src.uv_stride,
+ &xd->predictor[256], &xd->predictor[320], 8);
+ vp8_transform_mbuv(x);
+ vp8_quantize_mbuv(x);
+
+ rate_to = rd_cost_mbuv(x);
+ rate = rate_to + x->intra_uv_mode_cost[xd->frame_type][xd->mode_info_context->mbmi.uv_mode];
+
+ distortion = vp8_mbuverror(x) / 4;
+
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
+
+ if (this_rd < best_rd)
+ {
+ best_rd = this_rd;
+ d = distortion;
+ r = rate;
+ *rate_tokenonly = rate_to;
+ mode_selected = mode;
+ }
+ }
+
+ *rate = r;
+ *distortion = d;
+
+ xd->mode_info_context->mbmi.uv_mode = mode_selected;
+}
+
+int vp8_cost_mv_ref(MB_PREDICTION_MODE m, const int near_mv_ref_ct[4])
+{
+ vp8_prob p [VP8_MVREFS-1];
+ assert(NEARESTMV <= m && m <= SPLITMV);
+ vp8_mv_ref_probs(p, near_mv_ref_ct);
+ return vp8_cost_token(vp8_mv_ref_tree, p,
+ vp8_mv_ref_encoding_array - NEARESTMV + m);
+}
+
+void vp8_set_mbmode_and_mvs(MACROBLOCK *x, MB_PREDICTION_MODE mb, int_mv *mv)
+{
+ x->e_mbd.mode_info_context->mbmi.mode = mb;
+ x->e_mbd.mode_info_context->mbmi.mv.as_int = mv->as_int;
+}
+
+static int labels2mode(
+ MACROBLOCK *x,
+ int const *labelings, int which_label,
+ B_PREDICTION_MODE this_mode,
+ int_mv *this_mv, int_mv *best_ref_mv,
+ int *mvcost[2]
+)
+{
+ MACROBLOCKD *const xd = & x->e_mbd;
+ MODE_INFO *const mic = xd->mode_info_context;
+ const int mis = xd->mode_info_stride;
+
+ int cost = 0;
+ int thismvcost = 0;
+
+ /* We have to be careful retrieving previously-encoded motion vectors.
+ Ones from this macroblock have to be pulled from the BLOCKD array
+ as they have not yet made it to the bmi array in our MB_MODE_INFO. */
+
+ int i = 0;
+
+ do
+ {
+ BLOCKD *const d = xd->block + i;
+ const int row = i >> 2, col = i & 3;
+
+ B_PREDICTION_MODE m;
+
+ if (labelings[i] != which_label)
+ continue;
+
+ if (col && labelings[i] == labelings[i-1])
+ m = LEFT4X4;
+ else if (row && labelings[i] == labelings[i-4])
+ m = ABOVE4X4;
+ else
+ {
+ /* the only time we should do costing for new motion vector
+ * or mode is when we are on a new label (jbb May 08, 2007)
+ */
+ switch (m = this_mode)
+ {
+ case NEW4X4 :
+ thismvcost = vp8_mv_bit_cost(this_mv, best_ref_mv, mvcost, 102);
+ break;
+ case LEFT4X4:
+ this_mv->as_int = col ? d[-1].bmi.mv.as_int : left_block_mv(mic, i);
+ break;
+ case ABOVE4X4:
+ this_mv->as_int = row ? d[-4].bmi.mv.as_int : above_block_mv(mic, i, mis);
+ break;
+ case ZERO4X4:
+ this_mv->as_int = 0;
+ break;
+ default:
+ break;
+ }
+
+ if (m == ABOVE4X4) /* replace above with left if same */
+ {
+ int_mv left_mv;
+
+ left_mv.as_int = col ? d[-1].bmi.mv.as_int :
+ left_block_mv(mic, i);
+
+ if (left_mv.as_int == this_mv->as_int)
+ m = LEFT4X4;
+ }
+
+ cost = x->inter_bmode_costs[ m];
+ }
+
+ d->bmi.mv.as_int = this_mv->as_int;
+
+ x->partition_info->bmi[i].mode = m;
+ x->partition_info->bmi[i].mv.as_int = this_mv->as_int;
+
+ }
+ while (++i < 16);
+
+ cost += thismvcost ;
+ return cost;
+}
+
+static int rdcost_mbsegment_y(MACROBLOCK *mb, const int *labels,
+ int which_label, ENTROPY_CONTEXT *ta,
+ ENTROPY_CONTEXT *tl)
+{
+ int cost = 0;
+ int b;
+ MACROBLOCKD *x = &mb->e_mbd;
+
+ for (b = 0; b < 16; b++)
+ if (labels[ b] == which_label)
+ cost += cost_coeffs(mb, x->block + b, PLANE_TYPE_Y_WITH_DC,
+ ta + vp8_block2above[b],
+ tl + vp8_block2left[b]);
+
+ return cost;
+
+}
+static unsigned int vp8_encode_inter_mb_segment(MACROBLOCK *x, int const *labels, int which_label)
+{
+ int i;
+ unsigned int distortion = 0;
+ int pre_stride = x->e_mbd.pre.y_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+
+
+ for (i = 0; i < 16; i++)
+ {
+ if (labels[i] == which_label)
+ {
+ BLOCKD *bd = &x->e_mbd.block[i];
+ BLOCK *be = &x->block[i];
+
+ vp8_build_inter_predictors_b(bd, 16, base_pre, pre_stride, x->e_mbd.subpixel_predict);
+ vp8_subtract_b(be, bd, 16);
+ x->short_fdct4x4(be->src_diff, be->coeff, 32);
+ x->quantize_b(be, bd);
+
+ distortion += vp8_block_error(be->coeff, bd->dqcoeff);
+ }
+ }
+
+ return distortion;
+}
+
+
+static const unsigned int segmentation_to_sseshift[4] = {3, 3, 2, 0};
+
+
+typedef struct
+{
+ int_mv *ref_mv;
+ int_mv mvp;
+
+ int segment_rd;
+ int segment_num;
+ int r;
+ int d;
+ int segment_yrate;
+ B_PREDICTION_MODE modes[16];
+ int_mv mvs[16];
+ unsigned char eobs[16];
+
+ int mvthresh;
+ int *mdcounts;
+
+ int_mv sv_mvp[4]; /* save 4 mvp from 8x8 */
+ int sv_istep[2]; /* save 2 initial step_param for 16x8/8x16 */
+
+} BEST_SEG_INFO;
+
+
+static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x,
+ BEST_SEG_INFO *bsi, unsigned int segmentation)
+{
+ int i;
+ int const *labels;
+ int br = 0;
+ int bd = 0;
+ B_PREDICTION_MODE this_mode;
+
+
+ int label_count;
+ int this_segment_rd = 0;
+ int label_mv_thresh;
+ int rate = 0;
+ int sbr = 0;
+ int sbd = 0;
+ int segmentyrate = 0;
+
+ vp8_variance_fn_ptr_t *v_fn_ptr;
+
+ ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT *ta;
+ ENTROPY_CONTEXT *tl;
+ ENTROPY_CONTEXT_PLANES t_above_b, t_left_b;
+ ENTROPY_CONTEXT *ta_b;
+ ENTROPY_CONTEXT *tl_b;
+
+ vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta = (ENTROPY_CONTEXT *)&t_above;
+ tl = (ENTROPY_CONTEXT *)&t_left;
+ ta_b = (ENTROPY_CONTEXT *)&t_above_b;
+ tl_b = (ENTROPY_CONTEXT *)&t_left_b;
+
+ br = 0;
+ bd = 0;
+
+ v_fn_ptr = &cpi->fn_ptr[segmentation];
+ labels = vp8_mbsplits[segmentation];
+ label_count = vp8_mbsplit_count[segmentation];
+
+ /* 64 makes this threshold really big effectively making it so that we
+ * very rarely check mvs on segments. setting this to 1 would make mv
+ * thresh roughly equal to what it is for macroblocks
+ */
+ label_mv_thresh = 1 * bsi->mvthresh / label_count ;
+
+ /* Segmentation method overheads */
+ rate = vp8_cost_token(vp8_mbsplit_tree, vp8_mbsplit_probs, vp8_mbsplit_encodings + segmentation);
+ rate += vp8_cost_mv_ref(SPLITMV, bsi->mdcounts);
+ this_segment_rd += RDCOST(x->rdmult, x->rddiv, rate, 0);
+ br += rate;
+
+ for (i = 0; i < label_count; i++)
+ {
+ int_mv mode_mv[B_MODE_COUNT];
+ int best_label_rd = INT_MAX;
+ B_PREDICTION_MODE mode_selected = ZERO4X4;
+ int bestlabelyrate = 0;
+
+ /* search for the best motion vector on this segment */
+ for (this_mode = LEFT4X4; this_mode <= NEW4X4 ; this_mode ++)
+ {
+ int this_rd;
+ int distortion;
+ int labelyrate;
+ ENTROPY_CONTEXT_PLANES t_above_s, t_left_s;
+ ENTROPY_CONTEXT *ta_s;
+ ENTROPY_CONTEXT *tl_s;
+
+ vpx_memcpy(&t_above_s, &t_above, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_left_s, &t_left, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta_s = (ENTROPY_CONTEXT *)&t_above_s;
+ tl_s = (ENTROPY_CONTEXT *)&t_left_s;
+
+ if (this_mode == NEW4X4)
+ {
+ int sseshift;
+ int num00;
+ int step_param = 0;
+ int further_steps;
+ int n;
+ int thissme;
+ int bestsme = INT_MAX;
+ int_mv temp_mv;
+ BLOCK *c;
+ BLOCKD *e;
+
+ /* Is the best so far sufficiently good that we cant justify
+ * doing a new motion search.
+ */
+ if (best_label_rd < label_mv_thresh)
+ break;
+
+ if(cpi->compressor_speed)
+ {
+ if (segmentation == BLOCK_8X16 || segmentation == BLOCK_16X8)
+ {
+ bsi->mvp.as_int = bsi->sv_mvp[i].as_int;
+ if (i==1 && segmentation == BLOCK_16X8)
+ bsi->mvp.as_int = bsi->sv_mvp[2].as_int;
+
+ step_param = bsi->sv_istep[i];
+ }
+
+ /* use previous block's result as next block's MV
+ * predictor.
+ */
+ if (segmentation == BLOCK_4X4 && i>0)
+ {
+ bsi->mvp.as_int = x->e_mbd.block[i-1].bmi.mv.as_int;
+ if (i==4 || i==8 || i==12)
+ bsi->mvp.as_int = x->e_mbd.block[i-4].bmi.mv.as_int;
+ step_param = 2;
+ }
+ }
+
+ further_steps = (MAX_MVSEARCH_STEPS - 1) - step_param;
+
+ {
+ int sadpb = x->sadperbit4;
+ int_mv mvp_full;
+
+ mvp_full.as_mv.row = bsi->mvp.as_mv.row >>3;
+ mvp_full.as_mv.col = bsi->mvp.as_mv.col >>3;
+
+ /* find first label */
+ n = vp8_mbsplit_offset[segmentation][i];
+
+ c = &x->block[n];
+ e = &x->e_mbd.block[n];
+
+ {
+ bestsme = cpi->diamond_search_sad(x, c, e, &mvp_full,
+ &mode_mv[NEW4X4], step_param,
+ sadpb, &num00, v_fn_ptr,
+ x->mvcost, bsi->ref_mv);
+
+ n = num00;
+ num00 = 0;
+
+ while (n < further_steps)
+ {
+ n++;
+
+ if (num00)
+ num00--;
+ else
+ {
+ thissme = cpi->diamond_search_sad(x, c, e,
+ &mvp_full, &temp_mv,
+ step_param + n, sadpb,
+ &num00, v_fn_ptr,
+ x->mvcost, bsi->ref_mv);
+
+ if (thissme < bestsme)
+ {
+ bestsme = thissme;
+ mode_mv[NEW4X4].as_int = temp_mv.as_int;
+ }
+ }
+ }
+ }
+
+ sseshift = segmentation_to_sseshift[segmentation];
+
+ /* Should we do a full search (best quality only) */
+ if ((cpi->compressor_speed == 0) && (bestsme >> sseshift) > 4000)
+ {
+ /* Check if mvp_full is within the range. */
+ vp8_clamp_mv(&mvp_full, x->mv_col_min, x->mv_col_max, x->mv_row_min, x->mv_row_max);
+
+ thissme = cpi->full_search_sad(x, c, e, &mvp_full,
+ sadpb, 16, v_fn_ptr,
+ x->mvcost, bsi->ref_mv);
+
+ if (thissme < bestsme)
+ {
+ bestsme = thissme;
+ mode_mv[NEW4X4].as_int = e->bmi.mv.as_int;
+ }
+ else
+ {
+ /* The full search result is actually worse so
+ * re-instate the previous best vector
+ */
+ e->bmi.mv.as_int = mode_mv[NEW4X4].as_int;
+ }
+ }
+ }
+
+ if (bestsme < INT_MAX)
+ {
+ int distortion;
+ unsigned int sse;
+ cpi->find_fractional_mv_step(x, c, e, &mode_mv[NEW4X4],
+ bsi->ref_mv, x->errorperbit, v_fn_ptr, x->mvcost,
+ &distortion, &sse);
+
+ }
+ } /* NEW4X4 */
+
+ rate = labels2mode(x, labels, i, this_mode, &mode_mv[this_mode],
+ bsi->ref_mv, x->mvcost);
+
+ /* Trap vectors that reach beyond the UMV borders */
+ if (((mode_mv[this_mode].as_mv.row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].as_mv.row >> 3) > x->mv_row_max) ||
+ ((mode_mv[this_mode].as_mv.col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].as_mv.col >> 3) > x->mv_col_max))
+ {
+ continue;
+ }
+
+ distortion = vp8_encode_inter_mb_segment(x, labels, i) / 4;
+
+ labelyrate = rdcost_mbsegment_y(x, labels, i, ta_s, tl_s);
+ rate += labelyrate;
+
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
+
+ if (this_rd < best_label_rd)
+ {
+ sbr = rate;
+ sbd = distortion;
+ bestlabelyrate = labelyrate;
+ mode_selected = this_mode;
+ best_label_rd = this_rd;
+
+ vpx_memcpy(ta_b, ta_s, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(tl_b, tl_s, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ }
+ } /*for each 4x4 mode*/
+
+ vpx_memcpy(ta, ta_b, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(tl, tl_b, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ labels2mode(x, labels, i, mode_selected, &mode_mv[mode_selected],
+ bsi->ref_mv, x->mvcost);
+
+ br += sbr;
+ bd += sbd;
+ segmentyrate += bestlabelyrate;
+ this_segment_rd += best_label_rd;
+
+ if (this_segment_rd >= bsi->segment_rd)
+ break;
+
+ } /* for each label */
+
+ if (this_segment_rd < bsi->segment_rd)
+ {
+ bsi->r = br;
+ bsi->d = bd;
+ bsi->segment_yrate = segmentyrate;
+ bsi->segment_rd = this_segment_rd;
+ bsi->segment_num = segmentation;
+
+ /* store everything needed to come back to this!! */
+ for (i = 0; i < 16; i++)
+ {
+ bsi->mvs[i].as_mv = x->partition_info->bmi[i].mv.as_mv;
+ bsi->modes[i] = x->partition_info->bmi[i].mode;
+ bsi->eobs[i] = x->e_mbd.eobs[i];
+ }
+ }
+}
+
+static
+void vp8_cal_step_param(int sr, int *sp)
+{
+ int step = 0;
+
+ if (sr > MAX_FIRST_STEP) sr = MAX_FIRST_STEP;
+ else if (sr < 1) sr = 1;
+
+ while (sr>>=1)
+ step++;
+
+ *sp = MAX_MVSEARCH_STEPS - 1 - step;
+}
+
+static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x,
+ int_mv *best_ref_mv, int best_rd,
+ int *mdcounts, int *returntotrate,
+ int *returnyrate, int *returndistortion,
+ int mvthresh)
+{
+ int i;
+ BEST_SEG_INFO bsi;
+
+ vpx_memset(&bsi, 0, sizeof(bsi));
+
+ bsi.segment_rd = best_rd;
+ bsi.ref_mv = best_ref_mv;
+ bsi.mvp.as_int = best_ref_mv->as_int;
+ bsi.mvthresh = mvthresh;
+ bsi.mdcounts = mdcounts;
+
+ for(i = 0; i < 16; i++)
+ {
+ bsi.modes[i] = ZERO4X4;
+ }
+
+ if(cpi->compressor_speed == 0)
+ {
+ /* for now, we will keep the original segmentation order
+ when in best quality mode */
+ rd_check_segment(cpi, x, &bsi, BLOCK_16X8);
+ rd_check_segment(cpi, x, &bsi, BLOCK_8X16);
+ rd_check_segment(cpi, x, &bsi, BLOCK_8X8);
+ rd_check_segment(cpi, x, &bsi, BLOCK_4X4);
+ }
+ else
+ {
+ int sr;
+
+ rd_check_segment(cpi, x, &bsi, BLOCK_8X8);
+
+ if (bsi.segment_rd < best_rd)
+ {
+ int col_min = ((best_ref_mv->as_mv.col+7)>>3) - MAX_FULL_PEL_VAL;
+ int row_min = ((best_ref_mv->as_mv.row+7)>>3) - MAX_FULL_PEL_VAL;
+ int col_max = (best_ref_mv->as_mv.col>>3) + MAX_FULL_PEL_VAL;
+ int row_max = (best_ref_mv->as_mv.row>>3) + MAX_FULL_PEL_VAL;
+
+ int tmp_col_min = x->mv_col_min;
+ int tmp_col_max = x->mv_col_max;
+ int tmp_row_min = x->mv_row_min;
+ int tmp_row_max = x->mv_row_max;
+
+ /* Get intersection of UMV window and valid MV window to reduce # of checks in diamond search. */
+ if (x->mv_col_min < col_min )
+ x->mv_col_min = col_min;
+ if (x->mv_col_max > col_max )
+ x->mv_col_max = col_max;
+ if (x->mv_row_min < row_min )
+ x->mv_row_min = row_min;
+ if (x->mv_row_max > row_max )
+ x->mv_row_max = row_max;
+
+ /* Get 8x8 result */
+ bsi.sv_mvp[0].as_int = bsi.mvs[0].as_int;
+ bsi.sv_mvp[1].as_int = bsi.mvs[2].as_int;
+ bsi.sv_mvp[2].as_int = bsi.mvs[8].as_int;
+ bsi.sv_mvp[3].as_int = bsi.mvs[10].as_int;
+
+ /* Use 8x8 result as 16x8/8x16's predictor MV. Adjust search range according to the closeness of 2 MV. */
+ /* block 8X16 */
+ {
+ sr = MAXF((abs(bsi.sv_mvp[0].as_mv.row - bsi.sv_mvp[2].as_mv.row))>>3, (abs(bsi.sv_mvp[0].as_mv.col - bsi.sv_mvp[2].as_mv.col))>>3);
+ vp8_cal_step_param(sr, &bsi.sv_istep[0]);
+
+ sr = MAXF((abs(bsi.sv_mvp[1].as_mv.row - bsi.sv_mvp[3].as_mv.row))>>3, (abs(bsi.sv_mvp[1].as_mv.col - bsi.sv_mvp[3].as_mv.col))>>3);
+ vp8_cal_step_param(sr, &bsi.sv_istep[1]);
+
+ rd_check_segment(cpi, x, &bsi, BLOCK_8X16);
+ }
+
+ /* block 16X8 */
+ {
+ sr = MAXF((abs(bsi.sv_mvp[0].as_mv.row - bsi.sv_mvp[1].as_mv.row))>>3, (abs(bsi.sv_mvp[0].as_mv.col - bsi.sv_mvp[1].as_mv.col))>>3);
+ vp8_cal_step_param(sr, &bsi.sv_istep[0]);
+
+ sr = MAXF((abs(bsi.sv_mvp[2].as_mv.row - bsi.sv_mvp[3].as_mv.row))>>3, (abs(bsi.sv_mvp[2].as_mv.col - bsi.sv_mvp[3].as_mv.col))>>3);
+ vp8_cal_step_param(sr, &bsi.sv_istep[1]);
+
+ rd_check_segment(cpi, x, &bsi, BLOCK_16X8);
+ }
+
+ /* If 8x8 is better than 16x8/8x16, then do 4x4 search */
+ /* Not skip 4x4 if speed=0 (good quality) */
+ if (cpi->sf.no_skip_block4x4_search || bsi.segment_num == BLOCK_8X8) /* || (sv_segment_rd8x8-bsi.segment_rd) < sv_segment_rd8x8>>5) */
+ {
+ bsi.mvp.as_int = bsi.sv_mvp[0].as_int;
+ rd_check_segment(cpi, x, &bsi, BLOCK_4X4);
+ }
+
+ /* restore UMV window */
+ x->mv_col_min = tmp_col_min;
+ x->mv_col_max = tmp_col_max;
+ x->mv_row_min = tmp_row_min;
+ x->mv_row_max = tmp_row_max;
+ }
+ }
+
+ /* set it to the best */
+ for (i = 0; i < 16; i++)
+ {
+ BLOCKD *bd = &x->e_mbd.block[i];
+
+ bd->bmi.mv.as_int = bsi.mvs[i].as_int;
+ *bd->eob = bsi.eobs[i];
+ }
+
+ *returntotrate = bsi.r;
+ *returndistortion = bsi.d;
+ *returnyrate = bsi.segment_yrate;
+
+ /* save partitions */
+ x->e_mbd.mode_info_context->mbmi.partitioning = bsi.segment_num;
+ x->partition_info->count = vp8_mbsplit_count[bsi.segment_num];
+
+ for (i = 0; i < x->partition_info->count; i++)
+ {
+ int j;
+
+ j = vp8_mbsplit_offset[bsi.segment_num][i];
+
+ x->partition_info->bmi[i].mode = bsi.modes[j];
+ x->partition_info->bmi[i].mv.as_mv = bsi.mvs[j].as_mv;
+ }
+ /*
+ * used to set x->e_mbd.mode_info_context->mbmi.mv.as_int
+ */
+ x->partition_info->bmi[15].mv.as_int = bsi.mvs[15].as_int;
+
+ return bsi.segment_rd;
+}
+
+/* The improved MV prediction */
+void vp8_mv_pred
+(
+ VP8_COMP *cpi,
+ MACROBLOCKD *xd,
+ const MODE_INFO *here,
+ int_mv *mvp,
+ int refframe,
+ int *ref_frame_sign_bias,
+ int *sr,
+ int near_sadidx[]
+)
+{
+ const MODE_INFO *above = here - xd->mode_info_stride;
+ const MODE_INFO *left = here - 1;
+ const MODE_INFO *aboveleft = above - 1;
+ int_mv near_mvs[8];
+ int near_ref[8];
+ int_mv mv;
+ int vcnt=0;
+ int find=0;
+ int mb_offset;
+
+ int mvx[8];
+ int mvy[8];
+ int i;
+
+ mv.as_int = 0;
+
+ if(here->mbmi.ref_frame != INTRA_FRAME)
+ {
+ near_mvs[0].as_int = near_mvs[1].as_int = near_mvs[2].as_int = near_mvs[3].as_int = near_mvs[4].as_int = near_mvs[5].as_int = near_mvs[6].as_int = near_mvs[7].as_int = 0;
+ near_ref[0] = near_ref[1] = near_ref[2] = near_ref[3] = near_ref[4] = near_ref[5] = near_ref[6] = near_ref[7] = 0;
+
+ /* read in 3 nearby block's MVs from current frame as prediction
+ * candidates.
+ */
+ if (above->mbmi.ref_frame != INTRA_FRAME)
+ {
+ near_mvs[vcnt].as_int = above->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[above->mbmi.ref_frame], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
+ near_ref[vcnt] = above->mbmi.ref_frame;
+ }
+ vcnt++;
+ if (left->mbmi.ref_frame != INTRA_FRAME)
+ {
+ near_mvs[vcnt].as_int = left->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[left->mbmi.ref_frame], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
+ near_ref[vcnt] = left->mbmi.ref_frame;
+ }
+ vcnt++;
+ if (aboveleft->mbmi.ref_frame != INTRA_FRAME)
+ {
+ near_mvs[vcnt].as_int = aboveleft->mbmi.mv.as_int;
+ mv_bias(ref_frame_sign_bias[aboveleft->mbmi.ref_frame], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
+ near_ref[vcnt] = aboveleft->mbmi.ref_frame;
+ }
+ vcnt++;
+
+ /* read in 5 nearby block's MVs from last frame. */
+ if(cpi->common.last_frame_type != KEY_FRAME)
+ {
+ mb_offset = (-xd->mb_to_top_edge/128 + 1) * (xd->mode_info_stride +1) + (-xd->mb_to_left_edge/128 +1) ;
+
+ /* current in last frame */
+ if (cpi->lf_ref_frame[mb_offset] != INTRA_FRAME)
+ {
+ near_mvs[vcnt].as_int = cpi->lfmv[mb_offset].as_int;
+ mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
+ near_ref[vcnt] = cpi->lf_ref_frame[mb_offset];
+ }
+ vcnt++;
+
+ /* above in last frame */
+ if (cpi->lf_ref_frame[mb_offset - xd->mode_info_stride-1] != INTRA_FRAME)
+ {
+ near_mvs[vcnt].as_int = cpi->lfmv[mb_offset - xd->mode_info_stride-1].as_int;
+ mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset - xd->mode_info_stride-1], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
+ near_ref[vcnt] = cpi->lf_ref_frame[mb_offset - xd->mode_info_stride-1];
+ }
+ vcnt++;
+
+ /* left in last frame */
+ if (cpi->lf_ref_frame[mb_offset-1] != INTRA_FRAME)
+ {
+ near_mvs[vcnt].as_int = cpi->lfmv[mb_offset -1].as_int;
+ mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset -1], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
+ near_ref[vcnt] = cpi->lf_ref_frame[mb_offset - 1];
+ }
+ vcnt++;
+
+ /* right in last frame */
+ if (cpi->lf_ref_frame[mb_offset +1] != INTRA_FRAME)
+ {
+ near_mvs[vcnt].as_int = cpi->lfmv[mb_offset +1].as_int;
+ mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset +1], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
+ near_ref[vcnt] = cpi->lf_ref_frame[mb_offset +1];
+ }
+ vcnt++;
+
+ /* below in last frame */
+ if (cpi->lf_ref_frame[mb_offset + xd->mode_info_stride +1] != INTRA_FRAME)
+ {
+ near_mvs[vcnt].as_int = cpi->lfmv[mb_offset + xd->mode_info_stride +1].as_int;
+ mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset + xd->mode_info_stride +1], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
+ near_ref[vcnt] = cpi->lf_ref_frame[mb_offset + xd->mode_info_stride +1];
+ }
+ vcnt++;
+ }
+
+ for(i=0; i< vcnt; i++)
+ {
+ if(near_ref[near_sadidx[i]] != INTRA_FRAME)
+ {
+ if(here->mbmi.ref_frame == near_ref[near_sadidx[i]])
+ {
+ mv.as_int = near_mvs[near_sadidx[i]].as_int;
+ find = 1;
+ if (i < 3)
+ *sr = 3;
+ else
+ *sr = 2;
+ break;
+ }
+ }
+ }
+
+ if(!find)
+ {
+ for(i=0; i<vcnt; i++)
+ {
+ mvx[i] = near_mvs[i].as_mv.row;
+ mvy[i] = near_mvs[i].as_mv.col;
+ }
+
+ insertsortmv(mvx, vcnt);
+ insertsortmv(mvy, vcnt);
+ mv.as_mv.row = mvx[vcnt/2];
+ mv.as_mv.col = mvy[vcnt/2];
+
+ find = 1;
+ /* sr is set to 0 to allow calling function to decide the search
+ * range.
+ */
+ *sr = 0;
+ }
+ }
+
+ /* Set up return values */
+ mvp->as_int = mv.as_int;
+ vp8_clamp_mv2(mvp, xd);
+}
+
+void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffset, int near_sadidx[])
+{
+ /* near_sad indexes:
+ * 0-cf above, 1-cf left, 2-cf aboveleft,
+ * 3-lf current, 4-lf above, 5-lf left, 6-lf right, 7-lf below
+ */
+ int near_sad[8] = {0};
+ BLOCK *b = &x->block[0];
+ unsigned char *src_y_ptr = *(b->base_src);
+
+ /* calculate sad for current frame 3 nearby MBs. */
+ if( xd->mb_to_top_edge==0 && xd->mb_to_left_edge ==0)
+ {
+ near_sad[0] = near_sad[1] = near_sad[2] = INT_MAX;
+ }else if(xd->mb_to_top_edge==0)
+ { /* only has left MB for sad calculation. */
+ near_sad[0] = near_sad[2] = INT_MAX;
+ near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, UINT_MAX);
+ }else if(xd->mb_to_left_edge ==0)
+ { /* only has left MB for sad calculation. */
+ near_sad[1] = near_sad[2] = INT_MAX;
+ near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, UINT_MAX);
+ }else
+ {
+ near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, UINT_MAX);
+ near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, UINT_MAX);
+ near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, UINT_MAX);
+ }
+
+ if(cpi->common.last_frame_type != KEY_FRAME)
+ {
+ /* calculate sad for last frame 5 nearby MBs. */
+ unsigned char *pre_y_buffer = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_buffer + recon_yoffset;
+ int pre_y_stride = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_stride;
+
+ if(xd->mb_to_top_edge==0) near_sad[4] = INT_MAX;
+ if(xd->mb_to_left_edge ==0) near_sad[5] = INT_MAX;
+ if(xd->mb_to_right_edge ==0) near_sad[6] = INT_MAX;
+ if(xd->mb_to_bottom_edge==0) near_sad[7] = INT_MAX;
+
+ if(near_sad[4] != INT_MAX)
+ near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, UINT_MAX);
+ if(near_sad[5] != INT_MAX)
+ near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer - 16, pre_y_stride, UINT_MAX);
+ near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer, pre_y_stride, UINT_MAX);
+ if(near_sad[6] != INT_MAX)
+ near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer + 16, pre_y_stride, UINT_MAX);
+ if(near_sad[7] != INT_MAX)
+ near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, UINT_MAX);
+ }
+
+ if(cpi->common.last_frame_type != KEY_FRAME)
+ {
+ insertsortsad(near_sad, near_sadidx, 8);
+ }else
+ {
+ insertsortsad(near_sad, near_sadidx, 3);
+ }
+}
+
+static void rd_update_mvcount(VP8_COMP *cpi, MACROBLOCK *x, int_mv *best_ref_mv)
+{
+ if (x->e_mbd.mode_info_context->mbmi.mode == SPLITMV)
+ {
+ int i;
+
+ for (i = 0; i < x->partition_info->count; i++)
+ {
+ if (x->partition_info->bmi[i].mode == NEW4X4)
+ {
+ x->MVcount[0][mv_max+((x->partition_info->bmi[i].mv.as_mv.row
+ - best_ref_mv->as_mv.row) >> 1)]++;
+ x->MVcount[1][mv_max+((x->partition_info->bmi[i].mv.as_mv.col
+ - best_ref_mv->as_mv.col) >> 1)]++;
+ }
+ }
+ }
+ else if (x->e_mbd.mode_info_context->mbmi.mode == NEWMV)
+ {
+ x->MVcount[0][mv_max+((x->e_mbd.mode_info_context->mbmi.mv.as_mv.row
+ - best_ref_mv->as_mv.row) >> 1)]++;
+ x->MVcount[1][mv_max+((x->e_mbd.mode_info_context->mbmi.mv.as_mv.col
+ - best_ref_mv->as_mv.col) >> 1)]++;
+ }
+}
+
+static int evaluate_inter_mode_rd(int mdcounts[4],
+ RATE_DISTORTION* rd,
+ int* disable_skip,
+ VP8_COMP *cpi, MACROBLOCK *x)
+{
+ MB_PREDICTION_MODE this_mode = x->e_mbd.mode_info_context->mbmi.mode;
+ BLOCK *b = &x->block[0];
+ MACROBLOCKD *xd = &x->e_mbd;
+ int distortion;
+ vp8_build_inter16x16_predictors_mby(&x->e_mbd, x->e_mbd.predictor, 16);
+
+ if (cpi->active_map_enabled && x->active_ptr[0] == 0) {
+ x->skip = 1;
+ }
+ else if (x->encode_breakout)
+ {
+ unsigned int sse;
+ unsigned int var;
+ unsigned int threshold = (xd->block[0].dequant[1]
+ * xd->block[0].dequant[1] >>4);
+
+ if(threshold < x->encode_breakout)
+ threshold = x->encode_breakout;
+
+ var = vp8_variance16x16
+ (*(b->base_src), b->src_stride,
+ x->e_mbd.predictor, 16, &sse);
+
+ if (sse < threshold)
+ {
+ unsigned int q2dc = xd->block[24].dequant[0];
+ /* If theres is no codeable 2nd order dc
+ or a very small uniform pixel change change */
+ if ((sse - var < q2dc * q2dc >>4) ||
+ (sse /2 > var && sse-var < 64))
+ {
+ /* Check u and v to make sure skip is ok */
+ unsigned int sse2 = VP8_UVSSE(x);
+ if (sse2 * 2 < threshold)
+ {
+ x->skip = 1;
+ rd->distortion2 = sse + sse2;
+ rd->rate2 = 500;
+
+ /* for best_yrd calculation */
+ rd->rate_uv = 0;
+ rd->distortion_uv = sse2;
+
+ *disable_skip = 1;
+ return RDCOST(x->rdmult, x->rddiv, rd->rate2,
+ rd->distortion2);
+ }
+ }
+ }
+ }
+
+
+ /* Add in the Mv/mode cost */
+ rd->rate2 += vp8_cost_mv_ref(this_mode, mdcounts);
+
+ /* Y cost and distortion */
+ macro_block_yrd(x, &rd->rate_y, &distortion);
+ rd->rate2 += rd->rate_y;
+ rd->distortion2 += distortion;
+
+ /* UV cost and distortion */
+ rd_inter16x16_uv(cpi, x, &rd->rate_uv, &rd->distortion_uv,
+ cpi->common.full_pixel);
+ rd->rate2 += rd->rate_uv;
+ rd->distortion2 += rd->distortion_uv;
+ return INT_MAX;
+}
+
+static int calculate_final_rd_costs(int this_rd,
+ RATE_DISTORTION* rd,
+ int* other_cost,
+ int disable_skip,
+ int uv_intra_tteob,
+ int intra_rd_penalty,
+ VP8_COMP *cpi, MACROBLOCK *x)
+{
+ MB_PREDICTION_MODE this_mode = x->e_mbd.mode_info_context->mbmi.mode;
+
+ /* Where skip is allowable add in the default per mb cost for the no
+ * skip case. where we then decide to skip we have to delete this and
+ * replace it with the cost of signalling a skip
+ */
+ if (cpi->common.mb_no_coeff_skip)
+ {
+ *other_cost += vp8_cost_bit(cpi->prob_skip_false, 0);
+ rd->rate2 += *other_cost;
+ }
+
+ /* Estimate the reference frame signaling cost and add it
+ * to the rolling cost variable.
+ */
+ rd->rate2 +=
+ x->ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
+
+ if (!disable_skip)
+ {
+ /* Test for the condition where skip block will be activated
+ * because there are no non zero coefficients and make any
+ * necessary adjustment for rate
+ */
+ if (cpi->common.mb_no_coeff_skip)
+ {
+ int i;
+ int tteob;
+ int has_y2_block = (this_mode!=SPLITMV && this_mode!=B_PRED);
+
+ tteob = 0;
+ if(has_y2_block)
+ tteob += x->e_mbd.eobs[24];
+
+ for (i = 0; i < 16; i++)
+ tteob += (x->e_mbd.eobs[i] > has_y2_block);
+
+ if (x->e_mbd.mode_info_context->mbmi.ref_frame)
+ {
+ for (i = 16; i < 24; i++)
+ tteob += x->e_mbd.eobs[i];
+ }
+ else
+ tteob += uv_intra_tteob;
+
+ if (tteob == 0)
+ {
+ rd->rate2 -= (rd->rate_y + rd->rate_uv);
+ /* for best_yrd calculation */
+ rd->rate_uv = 0;
+
+ /* Back out no skip flag costing and add in skip flag costing */
+ if (cpi->prob_skip_false)
+ {
+ int prob_skip_cost;
+
+ prob_skip_cost = vp8_cost_bit(cpi->prob_skip_false, 1);
+ prob_skip_cost -= vp8_cost_bit(cpi->prob_skip_false, 0);
+ rd->rate2 += prob_skip_cost;
+ *other_cost += prob_skip_cost;
+ }
+ }
+ }
+ /* Calculate the final RD estimate for this mode */
+ this_rd = RDCOST(x->rdmult, x->rddiv, rd->rate2, rd->distortion2);
+ if (this_rd < INT_MAX && x->e_mbd.mode_info_context->mbmi.ref_frame
+ == INTRA_FRAME)
+ this_rd += intra_rd_penalty;
+ }
+ return this_rd;
+}
+
+static void update_best_mode(BEST_MODE* best_mode, int this_rd,
+ RATE_DISTORTION* rd, int other_cost, MACROBLOCK *x)
+{
+ MB_PREDICTION_MODE this_mode = x->e_mbd.mode_info_context->mbmi.mode;
+
+ other_cost +=
+ x->ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
+
+ /* Calculate the final y RD estimate for this mode */
+ best_mode->yrd = RDCOST(x->rdmult, x->rddiv, (rd->rate2-rd->rate_uv-other_cost),
+ (rd->distortion2-rd->distortion_uv));
+
+ best_mode->rd = this_rd;
+ vpx_memcpy(&best_mode->mbmode, &x->e_mbd.mode_info_context->mbmi, sizeof(MB_MODE_INFO));
+ vpx_memcpy(&best_mode->partition, x->partition_info, sizeof(PARTITION_INFO));
+
+ if ((this_mode == B_PRED) || (this_mode == SPLITMV))
+ {
+ int i;
+ for (i = 0; i < 16; i++)
+ {
+ best_mode->bmodes[i] = x->e_mbd.block[i].bmi;
+ }
+ }
+}
+
+void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset,
+ int recon_uvoffset, int *returnrate,
+ int *returndistortion, int *returnintra)
+{
+ BLOCK *b = &x->block[0];
+ BLOCKD *d = &x->e_mbd.block[0];
+ MACROBLOCKD *xd = &x->e_mbd;
+ int_mv best_ref_mv_sb[2];
+ int_mv mode_mv_sb[2][MB_MODE_COUNT];
+ int_mv best_ref_mv;
+ int_mv *mode_mv;
+ MB_PREDICTION_MODE this_mode;
+ int num00;
+ int best_mode_index = 0;
+ BEST_MODE best_mode;
+
+ int i;
+ int mode_index;
+ int mdcounts[4];
+ int rate;
+ RATE_DISTORTION rd;
+ int uv_intra_rate, uv_intra_distortion, uv_intra_rate_tokenonly;
+ int uv_intra_tteob = 0;
+ int uv_intra_done = 0;
+
+ MB_PREDICTION_MODE uv_intra_mode = 0;
+ int_mv mvp;
+ int near_sadidx[8] = {0, 1, 2, 3, 4, 5, 6, 7};
+ int saddone=0;
+ /* search range got from mv_pred(). It uses step_param levels. (0-7) */
+ int sr=0;
+
+ unsigned char *plane[4][3];
+ int ref_frame_map[4];
+ int sign_bias = 0;
+
+ int intra_rd_penalty = 10* vp8_dc_quant(cpi->common.base_qindex,
+ cpi->common.y1dc_delta_q);
+
+#if CONFIG_TEMPORAL_DENOISING
+ unsigned int zero_mv_sse = INT_MAX, best_sse = INT_MAX,
+ best_rd_sse = INT_MAX;
+#endif
+
+ mode_mv = mode_mv_sb[sign_bias];
+ best_ref_mv.as_int = 0;
+ best_mode.rd = INT_MAX;
+ best_mode.yrd = INT_MAX;
+ best_mode.intra_rd = INT_MAX;
+ vpx_memset(mode_mv_sb, 0, sizeof(mode_mv_sb));
+ vpx_memset(&best_mode.mbmode, 0, sizeof(best_mode.mbmode));
+ vpx_memset(&best_mode.bmodes, 0, sizeof(best_mode.bmodes));
+
+ /* Setup search priorities */
+ get_reference_search_order(cpi, ref_frame_map);
+
+ /* Check to see if there is at least 1 valid reference frame that we need
+ * to calculate near_mvs.
+ */
+ if (ref_frame_map[1] > 0)
+ {
+ sign_bias = vp8_find_near_mvs_bias(&x->e_mbd,
+ x->e_mbd.mode_info_context,
+ mode_mv_sb,
+ best_ref_mv_sb,
+ mdcounts,
+ ref_frame_map[1],
+ cpi->common.ref_frame_sign_bias);
+
+ mode_mv = mode_mv_sb[sign_bias];
+ best_ref_mv.as_int = best_ref_mv_sb[sign_bias].as_int;
+ }
+
+ get_predictor_pointers(cpi, plane, recon_yoffset, recon_uvoffset);
+
+ *returnintra = INT_MAX;
+ /* Count of the number of MBs tested so far this frame */
+ cpi->mbs_tested_so_far++;
+
+ x->skip = 0;
+
+ for (mode_index = 0; mode_index < MAX_MODES; mode_index++)
+ {
+ int this_rd = INT_MAX;
+ int disable_skip = 0;
+ int other_cost = 0;
+ int this_ref_frame = ref_frame_map[vp8_ref_frame_order[mode_index]];
+
+ /* Test best rd so far against threshold for trying this mode. */
+ if (best_mode.rd <= cpi->rd_threshes[mode_index])
+ continue;
+
+ if (this_ref_frame < 0)
+ continue;
+
+ /* These variables hold are rolling total cost and distortion for
+ * this mode
+ */
+ rd.rate2 = 0;
+ rd.distortion2 = 0;
+
+ this_mode = vp8_mode_order[mode_index];
+
+ x->e_mbd.mode_info_context->mbmi.mode = this_mode;
+ x->e_mbd.mode_info_context->mbmi.ref_frame = this_ref_frame;
+
+ /* Only consider ZEROMV/ALTREF_FRAME for alt ref frame,
+ * unless ARNR filtering is enabled in which case we want
+ * an unfiltered alternative
+ */
+ if (cpi->is_src_frame_alt_ref && (cpi->oxcf.arnr_max_frames == 0))
+ {
+ if (this_mode != ZEROMV || x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME)
+ continue;
+ }
+
+ /* everything but intra */
+ if (x->e_mbd.mode_info_context->mbmi.ref_frame)
+ {
+ x->e_mbd.pre.y_buffer = plane[this_ref_frame][0];
+ x->e_mbd.pre.u_buffer = plane[this_ref_frame][1];
+ x->e_mbd.pre.v_buffer = plane[this_ref_frame][2];
+
+ if (sign_bias != cpi->common.ref_frame_sign_bias[this_ref_frame])
+ {
+ sign_bias = cpi->common.ref_frame_sign_bias[this_ref_frame];
+ mode_mv = mode_mv_sb[sign_bias];
+ best_ref_mv.as_int = best_ref_mv_sb[sign_bias].as_int;
+ }
+ }
+
+ /* Check to see if the testing frequency for this mode is at its
+ * max If so then prevent it from being tested and increase the
+ * threshold for its testing
+ */
+ if (cpi->mode_test_hit_counts[mode_index] && (cpi->mode_check_freq[mode_index] > 1))
+ {
+ if (cpi->mbs_tested_so_far <= cpi->mode_check_freq[mode_index] * cpi->mode_test_hit_counts[mode_index])
+ {
+ /* Increase the threshold for coding this mode to make it
+ * less likely to be chosen
+ */
+ cpi->rd_thresh_mult[mode_index] += 4;
+
+ if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
+ cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
+
+ cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
+
+ continue;
+ }
+ }
+
+ /* We have now reached the point where we are going to test the
+ * current mode so increment the counter for the number of times
+ * it has been tested
+ */
+ cpi->mode_test_hit_counts[mode_index] ++;
+
+ /* Experimental code. Special case for gf and arf zeromv modes.
+ * Increase zbin size to supress noise
+ */
+ if (cpi->zbin_mode_boost_enabled)
+ {
+ if ( this_ref_frame == INTRA_FRAME )
+ cpi->zbin_mode_boost = 0;
+ else
+ {
+ if (vp8_mode_order[mode_index] == ZEROMV)
+ {
+ if (this_ref_frame != LAST_FRAME)
+ cpi->zbin_mode_boost = GF_ZEROMV_ZBIN_BOOST;
+ else
+ cpi->zbin_mode_boost = LF_ZEROMV_ZBIN_BOOST;
+ }
+ else if (vp8_mode_order[mode_index] == SPLITMV)
+ cpi->zbin_mode_boost = 0;
+ else
+ cpi->zbin_mode_boost = MV_ZBIN_BOOST;
+ }
+
+ vp8_update_zbin_extra(cpi, x);
+ }
+
+ if(!uv_intra_done && this_ref_frame == INTRA_FRAME)
+ {
+ rd_pick_intra_mbuv_mode(x, &uv_intra_rate,
+ &uv_intra_rate_tokenonly,
+ &uv_intra_distortion);
+ uv_intra_mode = x->e_mbd.mode_info_context->mbmi.uv_mode;
+
+ /*
+ * Total of the eobs is used later to further adjust rate2. Since uv
+ * block's intra eobs will be overwritten when we check inter modes,
+ * we need to save uv_intra_tteob here.
+ */
+ for (i = 16; i < 24; i++)
+ uv_intra_tteob += x->e_mbd.eobs[i];
+
+ uv_intra_done = 1;
+ }
+
+ switch (this_mode)
+ {
+ case B_PRED:
+ {
+ int tmp_rd;
+
+ /* Note the rate value returned here includes the cost of
+ * coding the BPRED mode: x->mbmode_cost[x->e_mbd.frame_type][BPRED]
+ */
+ int distortion;
+ tmp_rd = rd_pick_intra4x4mby_modes(x, &rate, &rd.rate_y, &distortion, best_mode.yrd);
+ rd.rate2 += rate;
+ rd.distortion2 += distortion;
+
+ if(tmp_rd < best_mode.yrd)
+ {
+ rd.rate2 += uv_intra_rate;
+ rd.rate_uv = uv_intra_rate_tokenonly;
+ rd.distortion2 += uv_intra_distortion;
+ rd.distortion_uv = uv_intra_distortion;
+ }
+ else
+ {
+ this_rd = INT_MAX;
+ disable_skip = 1;
+ }
+ }
+ break;
+
+ case SPLITMV:
+ {
+ int tmp_rd;
+ int this_rd_thresh;
+ int distortion;
+
+ this_rd_thresh = (vp8_ref_frame_order[mode_index] == 1) ? cpi->rd_threshes[THR_NEW1] : cpi->rd_threshes[THR_NEW3];
+ this_rd_thresh = (vp8_ref_frame_order[mode_index] == 2) ? cpi->rd_threshes[THR_NEW2] : this_rd_thresh;
+
+ tmp_rd = vp8_rd_pick_best_mbsegmentation(cpi, x, &best_ref_mv,
+ best_mode.yrd, mdcounts,
+ &rate, &rd.rate_y, &distortion, this_rd_thresh) ;
+
+ rd.rate2 += rate;
+ rd.distortion2 += distortion;
+
+ /* If even the 'Y' rd value of split is higher than best so far
+ * then dont bother looking at UV
+ */
+ if (tmp_rd < best_mode.yrd)
+ {
+ /* Now work out UV cost and add it in */
+ rd_inter4x4_uv(cpi, x, &rd.rate_uv, &rd.distortion_uv, cpi->common.full_pixel);
+ rd.rate2 += rd.rate_uv;
+ rd.distortion2 += rd.distortion_uv;
+ }
+ else
+ {
+ this_rd = INT_MAX;
+ disable_skip = 1;
+ }
+ }
+ break;
+ case DC_PRED:
+ case V_PRED:
+ case H_PRED:
+ case TM_PRED:
+ {
+ int distortion;
+ x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
+
+ vp8_build_intra_predictors_mby_s(xd,
+ xd->dst.y_buffer - xd->dst.y_stride,
+ xd->dst.y_buffer - 1,
+ xd->dst.y_stride,
+ xd->predictor,
+ 16);
+ macro_block_yrd(x, &rd.rate_y, &distortion) ;
+ rd.rate2 += rd.rate_y;
+ rd.distortion2 += distortion;
+ rd.rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode];
+ rd.rate2 += uv_intra_rate;
+ rd.rate_uv = uv_intra_rate_tokenonly;
+ rd.distortion2 += uv_intra_distortion;
+ rd.distortion_uv = uv_intra_distortion;
+ }
+ break;
+
+ case NEWMV:
+ {
+ int thissme;
+ int bestsme = INT_MAX;
+ int step_param = cpi->sf.first_step;
+ int further_steps;
+ int n;
+ int do_refine=1; /* If last step (1-away) of n-step search doesn't pick the center point as the best match,
+ we will do a final 1-away diamond refining search */
+
+ int sadpb = x->sadperbit16;
+ int_mv mvp_full;
+
+ int col_min = ((best_ref_mv.as_mv.col+7)>>3) - MAX_FULL_PEL_VAL;
+ int row_min = ((best_ref_mv.as_mv.row+7)>>3) - MAX_FULL_PEL_VAL;
+ int col_max = (best_ref_mv.as_mv.col>>3) + MAX_FULL_PEL_VAL;
+ int row_max = (best_ref_mv.as_mv.row>>3) + MAX_FULL_PEL_VAL;
+
+ int tmp_col_min = x->mv_col_min;
+ int tmp_col_max = x->mv_col_max;
+ int tmp_row_min = x->mv_row_min;
+ int tmp_row_max = x->mv_row_max;
+
+ if(!saddone)
+ {
+ vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] );
+ saddone = 1;
+ }
+
+ vp8_mv_pred(cpi, &x->e_mbd, x->e_mbd.mode_info_context, &mvp,
+ x->e_mbd.mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias, &sr, &near_sadidx[0]);
+
+ mvp_full.as_mv.col = mvp.as_mv.col>>3;
+ mvp_full.as_mv.row = mvp.as_mv.row>>3;
+
+ /* Get intersection of UMV window and valid MV window to
+ * reduce # of checks in diamond search.
+ */
+ if (x->mv_col_min < col_min )
+ x->mv_col_min = col_min;
+ if (x->mv_col_max > col_max )
+ x->mv_col_max = col_max;
+ if (x->mv_row_min < row_min )
+ x->mv_row_min = row_min;
+ if (x->mv_row_max > row_max )
+ x->mv_row_max = row_max;
+
+ /* adjust search range according to sr from mv prediction */
+ if(sr > step_param)
+ step_param = sr;
+
+ /* Initial step/diamond search */
+ {
+ bestsme = cpi->diamond_search_sad(x, b, d, &mvp_full, &d->bmi.mv,
+ step_param, sadpb, &num00,
+ &cpi->fn_ptr[BLOCK_16X16],
+ x->mvcost, &best_ref_mv);
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+
+ /* Further step/diamond searches as necessary */
+ n = 0;
+ further_steps = (cpi->sf.max_step_search_steps - 1) - step_param;
+
+ n = num00;
+ num00 = 0;
+
+ /* If there won't be more n-step search, check to see if refining search is needed. */
+ if (n > further_steps)
+ do_refine = 0;
+
+ while (n < further_steps)
+ {
+ n++;
+
+ if (num00)
+ num00--;
+ else
+ {
+ thissme = cpi->diamond_search_sad(x, b, d, &mvp_full,
+ &d->bmi.mv, step_param + n, sadpb, &num00,
+ &cpi->fn_ptr[BLOCK_16X16], x->mvcost,
+ &best_ref_mv);
+
+ /* check to see if refining search is needed. */
+ if (num00 > (further_steps-n))
+ do_refine = 0;
+
+ if (thissme < bestsme)
+ {
+ bestsme = thissme;
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+ }
+ else
+ {
+ d->bmi.mv.as_int = mode_mv[NEWMV].as_int;
+ }
+ }
+ }
+ }
+
+ /* final 1-away diamond refining search */
+ if (do_refine == 1)
+ {
+ int search_range;
+
+ search_range = 8;
+
+ thissme = cpi->refining_search_sad(x, b, d, &d->bmi.mv, sadpb,
+ search_range, &cpi->fn_ptr[BLOCK_16X16],
+ x->mvcost, &best_ref_mv);
+
+ if (thissme < bestsme)
+ {
+ bestsme = thissme;
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+ }
+ else
+ {
+ d->bmi.mv.as_int = mode_mv[NEWMV].as_int;
+ }
+ }
+
+ x->mv_col_min = tmp_col_min;
+ x->mv_col_max = tmp_col_max;
+ x->mv_row_min = tmp_row_min;
+ x->mv_row_max = tmp_row_max;
+
+ if (bestsme < INT_MAX)
+ {
+ int dis; /* TODO: use dis in distortion calculation later. */
+ unsigned int sse;
+ cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv, &best_ref_mv,
+ x->errorperbit,
+ &cpi->fn_ptr[BLOCK_16X16],
+ x->mvcost, &dis, &sse);
+ }
+
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+
+ /* Add the new motion vector cost to our rolling cost variable */
+ rd.rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv, x->mvcost, 96);
+ }
+
+ case NEARESTMV:
+ case NEARMV:
+ /* Clip "next_nearest" so that it does not extend to far out
+ * of image
+ */
+ vp8_clamp_mv2(&mode_mv[this_mode], xd);
+
+ /* Do not bother proceeding if the vector (from newmv, nearest
+ * or near) is 0,0 as this should then be coded using the zeromv
+ * mode.
+ */
+ if (((this_mode == NEARMV) || (this_mode == NEARESTMV)) && (mode_mv[this_mode].as_int == 0))
+ continue;
+
+ case ZEROMV:
+
+ /* Trap vectors that reach beyond the UMV borders
+ * Note that ALL New MV, Nearest MV Near MV and Zero MV code
+ * drops through to this point because of the lack of break
+ * statements in the previous two cases.
+ */
+ if (((mode_mv[this_mode].as_mv.row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].as_mv.row >> 3) > x->mv_row_max) ||
+ ((mode_mv[this_mode].as_mv.col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].as_mv.col >> 3) > x->mv_col_max))
+ continue;
+
+ vp8_set_mbmode_and_mvs(x, this_mode, &mode_mv[this_mode]);
+ this_rd = evaluate_inter_mode_rd(mdcounts, &rd,
+ &disable_skip, cpi, x);
+ break;
+
+ default:
+ break;
+ }
+
+ this_rd = calculate_final_rd_costs(this_rd, &rd, &other_cost,
+ disable_skip, uv_intra_tteob,
+ intra_rd_penalty, cpi, x);
+
+ /* Keep record of best intra distortion */
+ if ((x->e_mbd.mode_info_context->mbmi.ref_frame == INTRA_FRAME) &&
+ (this_rd < best_mode.intra_rd) )
+ {
+ best_mode.intra_rd = this_rd;
+ *returnintra = rd.distortion2 ;
+ }
+#if CONFIG_TEMPORAL_DENOISING
+ if (cpi->oxcf.noise_sensitivity)
+ {
+ unsigned int sse;
+ vp8_get_inter_mbpred_error(x,&cpi->fn_ptr[BLOCK_16X16],&sse,
+ mode_mv[this_mode]);
+
+ if (sse < best_rd_sse)
+ best_rd_sse = sse;
+
+ /* Store for later use by denoiser. */
+ if (this_mode == ZEROMV && sse < zero_mv_sse )
+ {
+ zero_mv_sse = sse;
+ x->best_zeromv_reference_frame =
+ x->e_mbd.mode_info_context->mbmi.ref_frame;
+ }
+
+ /* Store the best NEWMV in x for later use in the denoiser. */
+ if (x->e_mbd.mode_info_context->mbmi.mode == NEWMV &&
+ sse < best_sse)
+ {
+ best_sse = sse;
+ vp8_get_inter_mbpred_error(x,&cpi->fn_ptr[BLOCK_16X16],&best_sse,
+ mode_mv[this_mode]);
+ x->best_sse_inter_mode = NEWMV;
+ x->best_sse_mv = x->e_mbd.mode_info_context->mbmi.mv;
+ x->need_to_clamp_best_mvs =
+ x->e_mbd.mode_info_context->mbmi.need_to_clamp_mvs;
+ x->best_reference_frame =
+ x->e_mbd.mode_info_context->mbmi.ref_frame;
+ }
+ }
+#endif
+
+ /* Did this mode help.. i.i is it the new best mode */
+ if (this_rd < best_mode.rd || x->skip)
+ {
+ /* Note index of best mode so far */
+ best_mode_index = mode_index;
+ *returnrate = rd.rate2;
+ *returndistortion = rd.distortion2;
+ if (this_mode <= B_PRED)
+ {
+ x->e_mbd.mode_info_context->mbmi.uv_mode = uv_intra_mode;
+ /* required for left and above block mv */
+ x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
+ }
+ update_best_mode(&best_mode, this_rd, &rd, other_cost, x);
+
+
+ /* Testing this mode gave rise to an improvement in best error
+ * score. Lower threshold a bit for next time
+ */
+ cpi->rd_thresh_mult[mode_index] = (cpi->rd_thresh_mult[mode_index] >= (MIN_THRESHMULT + 2)) ? cpi->rd_thresh_mult[mode_index] - 2 : MIN_THRESHMULT;
+ cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
+ }
+
+ /* If the mode did not help improve the best error case then raise
+ * the threshold for testing that mode next time around.
+ */
+ else
+ {
+ cpi->rd_thresh_mult[mode_index] += 4;
+
+ if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
+ cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
+
+ cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
+ }
+
+ if (x->skip)
+ break;
+
+ }
+
+ /* Reduce the activation RD thresholds for the best choice mode */
+ if ((cpi->rd_baseline_thresh[best_mode_index] > 0) && (cpi->rd_baseline_thresh[best_mode_index] < (INT_MAX >> 2)))
+ {
+ int best_adjustment = (cpi->rd_thresh_mult[best_mode_index] >> 2);
+
+ cpi->rd_thresh_mult[best_mode_index] = (cpi->rd_thresh_mult[best_mode_index] >= (MIN_THRESHMULT + best_adjustment)) ? cpi->rd_thresh_mult[best_mode_index] - best_adjustment : MIN_THRESHMULT;
+ cpi->rd_threshes[best_mode_index] = (cpi->rd_baseline_thresh[best_mode_index] >> 7) * cpi->rd_thresh_mult[best_mode_index];
+ }
+
+ /* Note how often each mode chosen as best */
+ cpi->mode_chosen_counts[best_mode_index] ++;
+
+#if CONFIG_TEMPORAL_DENOISING
+ if (cpi->oxcf.noise_sensitivity)
+ {
+ if (x->best_sse_inter_mode == DC_PRED)
+ {
+ /* No best MV found. */
+ x->best_sse_inter_mode = best_mode.mbmode.mode;
+ x->best_sse_mv = best_mode.mbmode.mv;
+ x->need_to_clamp_best_mvs = best_mode.mbmode.need_to_clamp_mvs;
+ x->best_reference_frame = best_mode.mbmode.ref_frame;
+ best_sse = best_rd_sse;
+ }
+ vp8_denoiser_denoise_mb(&cpi->denoiser, x, best_sse, zero_mv_sse,
+ recon_yoffset, recon_uvoffset);
+
+
+ /* Reevaluate ZEROMV after denoising. */
+ if (best_mode.mbmode.ref_frame == INTRA_FRAME &&
+ x->best_zeromv_reference_frame != INTRA_FRAME)
+ {
+ int this_rd = INT_MAX;
+ int disable_skip = 0;
+ int other_cost = 0;
+ int this_ref_frame = x->best_zeromv_reference_frame;
+ rd.rate2 = x->ref_frame_cost[this_ref_frame] +
+ vp8_cost_mv_ref(ZEROMV, mdcounts);
+ rd.distortion2 = 0;
+
+ /* set up the proper prediction buffers for the frame */
+ x->e_mbd.mode_info_context->mbmi.ref_frame = this_ref_frame;
+ x->e_mbd.pre.y_buffer = plane[this_ref_frame][0];
+ x->e_mbd.pre.u_buffer = plane[this_ref_frame][1];
+ x->e_mbd.pre.v_buffer = plane[this_ref_frame][2];
+
+ x->e_mbd.mode_info_context->mbmi.mode = ZEROMV;
+ x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
+ x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
+
+ this_rd = evaluate_inter_mode_rd(mdcounts, &rd, &disable_skip, cpi, x);
+ this_rd = calculate_final_rd_costs(this_rd, &rd, &other_cost,
+ disable_skip, uv_intra_tteob,
+ intra_rd_penalty, cpi, x);
+ if (this_rd < best_mode.rd || x->skip)
+ {
+ /* Note index of best mode so far */
+ best_mode_index = mode_index;
+ *returnrate = rd.rate2;
+ *returndistortion = rd.distortion2;
+ update_best_mode(&best_mode, this_rd, &rd, other_cost, x);
+ }
+ }
+
+ }
+#endif
+
+ if (cpi->is_src_frame_alt_ref &&
+ (best_mode.mbmode.mode != ZEROMV || best_mode.mbmode.ref_frame != ALTREF_FRAME))
+ {
+ x->e_mbd.mode_info_context->mbmi.mode = ZEROMV;
+ x->e_mbd.mode_info_context->mbmi.ref_frame = ALTREF_FRAME;
+ x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
+ x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
+ x->e_mbd.mode_info_context->mbmi.mb_skip_coeff =
+ (cpi->common.mb_no_coeff_skip);
+ x->e_mbd.mode_info_context->mbmi.partitioning = 0;
+ return;
+ }
+
+
+ /* macroblock modes */
+ vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mode.mbmode, sizeof(MB_MODE_INFO));
+
+ if (best_mode.mbmode.mode == B_PRED)
+ {
+ for (i = 0; i < 16; i++)
+ xd->mode_info_context->bmi[i].as_mode = best_mode.bmodes[i].as_mode;
+ }
+
+ if (best_mode.mbmode.mode == SPLITMV)
+ {
+ for (i = 0; i < 16; i++)
+ xd->mode_info_context->bmi[i].mv.as_int = best_mode.bmodes[i].mv.as_int;
+
+ vpx_memcpy(x->partition_info, &best_mode.partition, sizeof(PARTITION_INFO));
+
+ x->e_mbd.mode_info_context->mbmi.mv.as_int =
+ x->partition_info->bmi[15].mv.as_int;
+ }
+
+ if (sign_bias
+ != cpi->common.ref_frame_sign_bias[xd->mode_info_context->mbmi.ref_frame])
+ best_ref_mv.as_int = best_ref_mv_sb[!sign_bias].as_int;
+
+ rd_update_mvcount(cpi, x, &best_ref_mv);
+}
+
+void vp8_rd_pick_intra_mode(MACROBLOCK *x, int *rate_)
+{
+ int error4x4, error16x16;
+ int rate4x4, rate16x16 = 0, rateuv;
+ int dist4x4, dist16x16, distuv;
+ int rate;
+ int rate4x4_tokenonly = 0;
+ int rate16x16_tokenonly = 0;
+ int rateuv_tokenonly = 0;
+
+ x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
+
+ rd_pick_intra_mbuv_mode(x, &rateuv, &rateuv_tokenonly, &distuv);
+ rate = rateuv;
+
+ error16x16 = rd_pick_intra16x16mby_mode(x, &rate16x16, &rate16x16_tokenonly,
+ &dist16x16);
+
+ error4x4 = rd_pick_intra4x4mby_modes(x, &rate4x4, &rate4x4_tokenonly,
+ &dist4x4, error16x16);
+
+ if (error4x4 < error16x16)
+ {
+ x->e_mbd.mode_info_context->mbmi.mode = B_PRED;
+ rate += rate4x4;
+ }
+ else
+ {
+ rate += rate16x16;
+ }
+
+ *rate_ = rate;
+}
diff --git a/libvpx/vp8/encoder/rdopt.h b/libvpx/vp8/encoder/rdopt.h
new file mode 100644
index 0000000..d7b0442
--- /dev/null
+++ b/libvpx/vp8/encoder/rdopt.h
@@ -0,0 +1,133 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_RDOPT_H
+#define __INC_RDOPT_H
+
+#define RDCOST(RM,DM,R,D) ( ((128+(R)*(RM)) >> 8) + (DM)*(D) )
+
+static void insertsortmv(int arr[], int len)
+{
+ int i, j, k;
+
+ for ( i = 1 ; i <= len-1 ; i++ )
+ {
+ for ( j = 0 ; j < i ; j++ )
+ {
+ if ( arr[j] > arr[i] )
+ {
+ int temp;
+
+ temp = arr[i];
+
+ for ( k = i; k >j; k--)
+ arr[k] = arr[k - 1] ;
+
+ arr[j] = temp ;
+ }
+ }
+ }
+}
+
+static void insertsortsad(int arr[],int idx[], int len)
+{
+ int i, j, k;
+
+ for ( i = 1 ; i <= len-1 ; i++ )
+ {
+ for ( j = 0 ; j < i ; j++ )
+ {
+ if ( arr[j] > arr[i] )
+ {
+ int temp, tempi;
+
+ temp = arr[i];
+ tempi = idx[i];
+
+ for ( k = i; k >j; k--)
+ {
+ arr[k] = arr[k - 1] ;
+ idx[k] = idx[k - 1];
+ }
+
+ arr[j] = temp ;
+ idx[j] = tempi;
+ }
+ }
+ }
+}
+
+extern void vp8_initialize_rd_consts(VP8_COMP *cpi, int Qvalue);
+extern void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra);
+extern void vp8_rd_pick_intra_mode(MACROBLOCK *x, int *rate);
+
+
+static void get_plane_pointers(const YV12_BUFFER_CONFIG *fb,
+ unsigned char *plane[3],
+ unsigned int recon_yoffset,
+ unsigned int recon_uvoffset)
+{
+ plane[0] = fb->y_buffer + recon_yoffset;
+ plane[1] = fb->u_buffer + recon_uvoffset;
+ plane[2] = fb->v_buffer + recon_uvoffset;
+}
+
+
+static void get_predictor_pointers(const VP8_COMP *cpi,
+ unsigned char *plane[4][3],
+ unsigned int recon_yoffset,
+ unsigned int recon_uvoffset)
+{
+ if (cpi->ref_frame_flags & VP8_LAST_FRAME)
+ get_plane_pointers(&cpi->common.yv12_fb[cpi->common.lst_fb_idx],
+ plane[LAST_FRAME], recon_yoffset, recon_uvoffset);
+
+ if (cpi->ref_frame_flags & VP8_GOLD_FRAME)
+ get_plane_pointers(&cpi->common.yv12_fb[cpi->common.gld_fb_idx],
+ plane[GOLDEN_FRAME], recon_yoffset, recon_uvoffset);
+
+ if (cpi->ref_frame_flags & VP8_ALTR_FRAME)
+ get_plane_pointers(&cpi->common.yv12_fb[cpi->common.alt_fb_idx],
+ plane[ALTREF_FRAME], recon_yoffset, recon_uvoffset);
+}
+
+
+static void get_reference_search_order(const VP8_COMP *cpi,
+ int ref_frame_map[4])
+{
+ int i=0;
+
+ ref_frame_map[i++] = INTRA_FRAME;
+ if (cpi->ref_frame_flags & VP8_LAST_FRAME)
+ ref_frame_map[i++] = LAST_FRAME;
+ if (cpi->ref_frame_flags & VP8_GOLD_FRAME)
+ ref_frame_map[i++] = GOLDEN_FRAME;
+ if (cpi->ref_frame_flags & VP8_ALTR_FRAME)
+ ref_frame_map[i++] = ALTREF_FRAME;
+ for(; i<4; i++)
+ ref_frame_map[i] = -1;
+}
+
+
+extern void vp8_mv_pred
+(
+ VP8_COMP *cpi,
+ MACROBLOCKD *xd,
+ const MODE_INFO *here,
+ int_mv *mvp,
+ int refframe,
+ int *ref_frame_sign_bias,
+ int *sr,
+ int near_sadidx[]
+);
+void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffset, int near_sadidx[]);
+
+#endif
diff --git a/libvpx/vp8/encoder/segmentation.c b/libvpx/vp8/encoder/segmentation.c
new file mode 100644
index 0000000..37972e2
--- /dev/null
+++ b/libvpx/vp8/encoder/segmentation.c
@@ -0,0 +1,66 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "segmentation.h"
+#include "vpx_mem/vpx_mem.h"
+
+void vp8_update_gf_useage_maps(VP8_COMP *cpi, VP8_COMMON *cm, MACROBLOCK *x)
+{
+ int mb_row, mb_col;
+
+ MODE_INFO *this_mb_mode_info = cm->mi;
+
+ x->gf_active_ptr = (signed char *)cpi->gf_active_flags;
+
+ if ((cm->frame_type == KEY_FRAME) || (cm->refresh_golden_frame))
+ {
+ /* Reset Gf useage monitors */
+ vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols));
+ cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
+ }
+ else
+ {
+ /* for each macroblock row in image */
+ for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
+ {
+ /* for each macroblock col in image */
+ for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
+ {
+
+ /* If using golden then set GF active flag if not already set.
+ * If using last frame 0,0 mode then leave flag as it is
+ * else if using non 0,0 motion or intra modes then clear
+ * flag if it is currently set
+ */
+ if ((this_mb_mode_info->mbmi.ref_frame == GOLDEN_FRAME) || (this_mb_mode_info->mbmi.ref_frame == ALTREF_FRAME))
+ {
+ if (*(x->gf_active_ptr) == 0)
+ {
+ *(x->gf_active_ptr) = 1;
+ cpi->gf_active_count ++;
+ }
+ }
+ else if ((this_mb_mode_info->mbmi.mode != ZEROMV) && *(x->gf_active_ptr))
+ {
+ *(x->gf_active_ptr) = 0;
+ cpi->gf_active_count--;
+ }
+
+ x->gf_active_ptr++; /* Step onto next entry */
+ this_mb_mode_info++; /* skip to next mb */
+
+ }
+
+ /* this is to account for the border */
+ this_mb_mode_info++;
+ }
+ }
+}
diff --git a/vp8/encoder/segmentation.h b/libvpx/vp8/encoder/segmentation.h
similarity index 100%
rename from vp8/encoder/segmentation.h
rename to libvpx/vp8/encoder/segmentation.h
diff --git a/libvpx/vp8/encoder/ssim.c b/libvpx/vp8/encoder/ssim.c
new file mode 100644
index 0000000..e751608
--- /dev/null
+++ b/libvpx/vp8/encoder/ssim.c
@@ -0,0 +1,233 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "onyx_int.h"
+
+void vp8_ssim_parms_16x16_c
+(
+ unsigned char *s,
+ int sp,
+ unsigned char *r,
+ int rp,
+ unsigned long *sum_s,
+ unsigned long *sum_r,
+ unsigned long *sum_sq_s,
+ unsigned long *sum_sq_r,
+ unsigned long *sum_sxr
+)
+{
+ int i,j;
+ for(i=0;i<16;i++,s+=sp,r+=rp)
+ {
+ for(j=0;j<16;j++)
+ {
+ *sum_s += s[j];
+ *sum_r += r[j];
+ *sum_sq_s += s[j] * s[j];
+ *sum_sq_r += r[j] * r[j];
+ *sum_sxr += s[j] * r[j];
+ }
+ }
+}
+void vp8_ssim_parms_8x8_c
+(
+ unsigned char *s,
+ int sp,
+ unsigned char *r,
+ int rp,
+ unsigned long *sum_s,
+ unsigned long *sum_r,
+ unsigned long *sum_sq_s,
+ unsigned long *sum_sq_r,
+ unsigned long *sum_sxr
+)
+{
+ int i,j;
+ for(i=0;i<8;i++,s+=sp,r+=rp)
+ {
+ for(j=0;j<8;j++)
+ {
+ *sum_s += s[j];
+ *sum_r += r[j];
+ *sum_sq_s += s[j] * s[j];
+ *sum_sq_r += r[j] * r[j];
+ *sum_sxr += s[j] * r[j];
+ }
+ }
+}
+
+const static int64_t cc1 = 26634; // (64^2*(.01*255)^2
+const static int64_t cc2 = 239708; // (64^2*(.03*255)^2
+
+static double similarity
+(
+ unsigned long sum_s,
+ unsigned long sum_r,
+ unsigned long sum_sq_s,
+ unsigned long sum_sq_r,
+ unsigned long sum_sxr,
+ int count
+)
+{
+ int64_t ssim_n, ssim_d;
+ int64_t c1, c2;
+
+ //scale the constants by number of pixels
+ c1 = (cc1*count*count)>>12;
+ c2 = (cc2*count*count)>>12;
+
+ ssim_n = (2*sum_s*sum_r+ c1)*((int64_t) 2*count*sum_sxr-
+ (int64_t) 2*sum_s*sum_r+c2);
+
+ ssim_d = (sum_s*sum_s +sum_r*sum_r+c1)*
+ ((int64_t)count*sum_sq_s-(int64_t)sum_s*sum_s +
+ (int64_t)count*sum_sq_r-(int64_t) sum_r*sum_r +c2) ;
+
+ return ssim_n * 1.0 / ssim_d;
+}
+
+static double ssim_16x16(unsigned char *s,int sp, unsigned char *r,int rp)
+{
+ unsigned long sum_s=0,sum_r=0,sum_sq_s=0,sum_sq_r=0,sum_sxr=0;
+ vp8_ssim_parms_16x16(s, sp, r, rp, &sum_s, &sum_r, &sum_sq_s, &sum_sq_r, &sum_sxr);
+ return similarity(sum_s, sum_r, sum_sq_s, sum_sq_r, sum_sxr, 256);
+}
+static double ssim_8x8(unsigned char *s,int sp, unsigned char *r,int rp)
+{
+ unsigned long sum_s=0,sum_r=0,sum_sq_s=0,sum_sq_r=0,sum_sxr=0;
+ vp8_ssim_parms_8x8(s, sp, r, rp, &sum_s, &sum_r, &sum_sq_s, &sum_sq_r, &sum_sxr);
+ return similarity(sum_s, sum_r, sum_sq_s, sum_sq_r, sum_sxr, 64);
+}
+
+// TODO: (jbb) tried to scale this function such that we may be able to use it
+// for distortion metric in mode selection code ( provided we do a reconstruction)
+long dssim(unsigned char *s,int sp, unsigned char *r,int rp)
+{
+ unsigned long sum_s=0,sum_r=0,sum_sq_s=0,sum_sq_r=0,sum_sxr=0;
+ int64_t ssim3;
+ int64_t ssim_n1,ssim_n2;
+ int64_t ssim_d1,ssim_d2;
+ int64_t ssim_t1,ssim_t2;
+ int64_t c1, c2;
+
+ // normalize by 256/64
+ c1 = cc1*16;
+ c2 = cc2*16;
+
+ vp8_ssim_parms_16x16(s, sp, r, rp, &sum_s, &sum_r, &sum_sq_s, &sum_sq_r, &sum_sxr);
+ ssim_n1 = (2*sum_s*sum_r+ c1);
+
+ ssim_n2 =((int64_t) 2*256*sum_sxr-(int64_t) 2*sum_s*sum_r+c2);
+
+ ssim_d1 =((int64_t)sum_s*sum_s +(int64_t)sum_r*sum_r+c1);
+
+ ssim_d2 = (256 * (int64_t) sum_sq_s-(int64_t) sum_s*sum_s +
+ (int64_t) 256*sum_sq_r-(int64_t) sum_r*sum_r +c2) ;
+
+ ssim_t1 = 256 - 256 * ssim_n1 / ssim_d1;
+ ssim_t2 = 256 - 256 * ssim_n2 / ssim_d2;
+
+ ssim3 = 256 *ssim_t1 * ssim_t2;
+ if(ssim3 <0 )
+ ssim3=0;
+ return (long)( ssim3 );
+}
+
+// We are using a 8x8 moving window with starting location of each 8x8 window
+// on the 4x4 pixel grid. Such arrangement allows the windows to overlap
+// block boundaries to penalize blocking artifacts.
+double vp8_ssim2
+(
+ unsigned char *img1,
+ unsigned char *img2,
+ int stride_img1,
+ int stride_img2,
+ int width,
+ int height
+)
+{
+ int i,j;
+ int samples =0;
+ double ssim_total=0;
+
+ // sample point start with each 4x4 location
+ for(i=0; i < height-8; i+=4, img1 += stride_img1*4, img2 += stride_img2*4)
+ {
+ for(j=0; j < width-8; j+=4 )
+ {
+ double v = ssim_8x8(img1+j, stride_img1, img2+j, stride_img2);
+ ssim_total += v;
+ samples++;
+ }
+ }
+ ssim_total /= samples;
+ return ssim_total;
+}
+double vp8_calc_ssim
+(
+ YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *dest,
+ int lumamask,
+ double *weight
+)
+{
+ double a, b, c;
+ double ssimv;
+
+ a = vp8_ssim2(source->y_buffer, dest->y_buffer,
+ source->y_stride, dest->y_stride, source->y_width,
+ source->y_height);
+
+ b = vp8_ssim2(source->u_buffer, dest->u_buffer,
+ source->uv_stride, dest->uv_stride, source->uv_width,
+ source->uv_height);
+
+ c = vp8_ssim2(source->v_buffer, dest->v_buffer,
+ source->uv_stride, dest->uv_stride, source->uv_width,
+ source->uv_height);
+
+ ssimv = a * .8 + .1 * (b + c);
+
+ *weight = 1;
+
+ return ssimv;
+}
+
+double vp8_calc_ssimg
+(
+ YV12_BUFFER_CONFIG *source,
+ YV12_BUFFER_CONFIG *dest,
+ double *ssim_y,
+ double *ssim_u,
+ double *ssim_v
+)
+{
+ double ssim_all = 0;
+ double a, b, c;
+
+ a = vp8_ssim2(source->y_buffer, dest->y_buffer,
+ source->y_stride, dest->y_stride, source->y_width,
+ source->y_height);
+
+ b = vp8_ssim2(source->u_buffer, dest->u_buffer,
+ source->uv_stride, dest->uv_stride, source->uv_width,
+ source->uv_height);
+
+ c = vp8_ssim2(source->v_buffer, dest->v_buffer,
+ source->uv_stride, dest->uv_stride, source->uv_width,
+ source->uv_height);
+ *ssim_y = a;
+ *ssim_u = b;
+ *ssim_v = c;
+ ssim_all = (a * 4 + b + c) /6;
+
+ return ssim_all;
+}
diff --git a/libvpx/vp8/encoder/temporal_filter.c b/libvpx/vp8/encoder/temporal_filter.c
new file mode 100644
index 0000000..b83ae89
--- /dev/null
+++ b/libvpx/vp8/encoder/temporal_filter.c
@@ -0,0 +1,519 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/common/onyxc_int.h"
+#include "onyx_int.h"
+#include "vp8/common/systemdependent.h"
+#include "quantize.h"
+#include "vp8/common/alloccommon.h"
+#include "mcomp.h"
+#include "firstpass.h"
+#include "psnr.h"
+#include "vpx_scale/vpxscale.h"
+#include "vp8/common/extend.h"
+#include "ratectrl.h"
+#include "vp8/common/quant_common.h"
+#include "segmentation.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vp8/common/swapyv12buffer.h"
+#include "vp8/common/threading.h"
+#include "vpx_ports/vpx_timer.h"
+
+#include <math.h>
+#include <limits.h>
+
+#define ALT_REF_MC_ENABLED 1 /* dis/enable MC in AltRef filtering */
+#define ALT_REF_SUBPEL_ENABLED 1 /* dis/enable subpel in MC AltRef filtering */
+
+#if VP8_TEMPORAL_ALT_REF
+
+static void vp8_temporal_filter_predictors_mb_c
+(
+ MACROBLOCKD *x,
+ unsigned char *y_mb_ptr,
+ unsigned char *u_mb_ptr,
+ unsigned char *v_mb_ptr,
+ int stride,
+ int mv_row,
+ int mv_col,
+ unsigned char *pred
+)
+{
+ int offset;
+ unsigned char *yptr, *uptr, *vptr;
+
+ /* Y */
+ yptr = y_mb_ptr + (mv_row >> 3) * stride + (mv_col >> 3);
+
+ if ((mv_row | mv_col) & 7)
+ {
+ x->subpixel_predict16x16(yptr, stride,
+ mv_col & 7, mv_row & 7, &pred[0], 16);
+ }
+ else
+ {
+ vp8_copy_mem16x16(yptr, stride, &pred[0], 16);
+ }
+
+ /* U & V */
+ mv_row >>= 1;
+ mv_col >>= 1;
+ stride = (stride + 1) >> 1;
+ offset = (mv_row >> 3) * stride + (mv_col >> 3);
+ uptr = u_mb_ptr + offset;
+ vptr = v_mb_ptr + offset;
+
+ if ((mv_row | mv_col) & 7)
+ {
+ x->subpixel_predict8x8(uptr, stride,
+ mv_col & 7, mv_row & 7, &pred[256], 8);
+ x->subpixel_predict8x8(vptr, stride,
+ mv_col & 7, mv_row & 7, &pred[320], 8);
+ }
+ else
+ {
+ vp8_copy_mem8x8(uptr, stride, &pred[256], 8);
+ vp8_copy_mem8x8(vptr, stride, &pred[320], 8);
+ }
+}
+void vp8_temporal_filter_apply_c
+(
+ unsigned char *frame1,
+ unsigned int stride,
+ unsigned char *frame2,
+ unsigned int block_size,
+ int strength,
+ int filter_weight,
+ unsigned int *accumulator,
+ unsigned short *count
+)
+{
+ unsigned int i, j, k;
+ int modifier;
+ int byte = 0;
+
+ for (i = 0,k = 0; i < block_size; i++)
+ {
+ for (j = 0; j < block_size; j++, k++)
+ {
+
+ int src_byte = frame1[byte];
+ int pixel_value = *frame2++;
+
+ modifier = src_byte - pixel_value;
+ /* This is an integer approximation of:
+ * float coeff = (3.0 * modifer * modifier) / pow(2, strength);
+ * modifier = (int)roundf(coeff > 16 ? 0 : 16-coeff);
+ */
+ modifier *= modifier;
+ modifier *= 3;
+ modifier += 1 << (strength - 1);
+ modifier >>= strength;
+
+ if (modifier > 16)
+ modifier = 16;
+
+ modifier = 16 - modifier;
+ modifier *= filter_weight;
+
+ count[k] += modifier;
+ accumulator[k] += modifier * pixel_value;
+
+ byte++;
+ }
+
+ byte += stride - block_size;
+ }
+}
+
+#if ALT_REF_MC_ENABLED
+
+static int vp8_temporal_filter_find_matching_mb_c
+(
+ VP8_COMP *cpi,
+ YV12_BUFFER_CONFIG *arf_frame,
+ YV12_BUFFER_CONFIG *frame_ptr,
+ int mb_offset,
+ int error_thresh
+)
+{
+ MACROBLOCK *x = &cpi->mb;
+ int step_param;
+ int sadpb = x->sadperbit16;
+ int bestsme = INT_MAX;
+
+ BLOCK *b = &x->block[0];
+ BLOCKD *d = &x->e_mbd.block[0];
+ int_mv best_ref_mv1;
+ int_mv best_ref_mv1_full; /* full-pixel value of best_ref_mv1 */
+
+ /* Save input state */
+ unsigned char **base_src = b->base_src;
+ int src = b->src;
+ int src_stride = b->src_stride;
+ unsigned char *base_pre = x->e_mbd.pre.y_buffer;
+ int pre = d->offset;
+ int pre_stride = x->e_mbd.pre.y_stride;
+
+ best_ref_mv1.as_int = 0;
+ best_ref_mv1_full.as_mv.col = best_ref_mv1.as_mv.col >>3;
+ best_ref_mv1_full.as_mv.row = best_ref_mv1.as_mv.row >>3;
+
+ /* Setup frame pointers */
+ b->base_src = &arf_frame->y_buffer;
+ b->src_stride = arf_frame->y_stride;
+ b->src = mb_offset;
+
+ x->e_mbd.pre.y_buffer = frame_ptr->y_buffer;
+ x->e_mbd.pre.y_stride = frame_ptr->y_stride;
+ d->offset = mb_offset;
+
+ /* Further step/diamond searches as necessary */
+ if (cpi->Speed < 8)
+ {
+ step_param = cpi->sf.first_step + (cpi->Speed > 5);
+ }
+ else
+ {
+ step_param = cpi->sf.first_step + 2;
+ }
+
+ /* TODO Check that the 16x16 vf & sdf are selected here */
+ /* Ignore mv costing by sending NULL cost arrays */
+ bestsme = vp8_hex_search(x, b, d, &best_ref_mv1_full, &d->bmi.mv,
+ step_param, sadpb,
+ &cpi->fn_ptr[BLOCK_16X16],
+ NULL, NULL, &best_ref_mv1);
+
+#if ALT_REF_SUBPEL_ENABLED
+ /* Try sub-pixel MC? */
+ {
+ int distortion;
+ unsigned int sse;
+ /* Ignore mv costing by sending NULL cost array */
+ bestsme = cpi->find_fractional_mv_step(x, b, d,
+ &d->bmi.mv,
+ &best_ref_mv1,
+ x->errorperbit,
+ &cpi->fn_ptr[BLOCK_16X16],
+ NULL, &distortion, &sse);
+ }
+#endif
+
+ /* Save input state */
+ b->base_src = base_src;
+ b->src = src;
+ b->src_stride = src_stride;
+ x->e_mbd.pre.y_buffer = base_pre;
+ d->offset = pre;
+ x->e_mbd.pre.y_stride = pre_stride;
+
+ return bestsme;
+}
+#endif
+
+static void vp8_temporal_filter_iterate_c
+(
+ VP8_COMP *cpi,
+ int frame_count,
+ int alt_ref_index,
+ int strength
+)
+{
+ int byte;
+ int frame;
+ int mb_col, mb_row;
+ unsigned int filter_weight;
+ int mb_cols = cpi->common.mb_cols;
+ int mb_rows = cpi->common.mb_rows;
+ int mb_y_offset = 0;
+ int mb_uv_offset = 0;
+ DECLARE_ALIGNED_ARRAY(16, unsigned int, accumulator, 16*16 + 8*8 + 8*8);
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, count, 16*16 + 8*8 + 8*8);
+ MACROBLOCKD *mbd = &cpi->mb.e_mbd;
+ YV12_BUFFER_CONFIG *f = cpi->frames[alt_ref_index];
+ unsigned char *dst1, *dst2;
+ DECLARE_ALIGNED_ARRAY(16, unsigned char, predictor, 16*16 + 8*8 + 8*8);
+
+ /* Save input state */
+ unsigned char *y_buffer = mbd->pre.y_buffer;
+ unsigned char *u_buffer = mbd->pre.u_buffer;
+ unsigned char *v_buffer = mbd->pre.v_buffer;
+
+ for (mb_row = 0; mb_row < mb_rows; mb_row++)
+ {
+#if ALT_REF_MC_ENABLED
+ /* Source frames are extended to 16 pixels. This is different than
+ * L/A/G reference frames that have a border of 32 (VP8BORDERINPIXELS)
+ * A 6 tap filter is used for motion search. This requires 2 pixels
+ * before and 3 pixels after. So the largest Y mv on a border would
+ * then be 16 - 3. The UV blocks are half the size of the Y and
+ * therefore only extended by 8. The largest mv that a UV block
+ * can support is 8 - 3. A UV mv is half of a Y mv.
+ * (16 - 3) >> 1 == 6 which is greater than 8 - 3.
+ * To keep the mv in play for both Y and UV planes the max that it
+ * can be on a border is therefore 16 - 5.
+ */
+ cpi->mb.mv_row_min = -((mb_row * 16) + (16 - 5));
+ cpi->mb.mv_row_max = ((cpi->common.mb_rows - 1 - mb_row) * 16)
+ + (16 - 5);
+#endif
+
+ for (mb_col = 0; mb_col < mb_cols; mb_col++)
+ {
+ int i, j, k;
+ int stride;
+
+ vpx_memset(accumulator, 0, 384*sizeof(unsigned int));
+ vpx_memset(count, 0, 384*sizeof(unsigned short));
+
+#if ALT_REF_MC_ENABLED
+ cpi->mb.mv_col_min = -((mb_col * 16) + (16 - 5));
+ cpi->mb.mv_col_max = ((cpi->common.mb_cols - 1 - mb_col) * 16)
+ + (16 - 5);
+#endif
+
+ for (frame = 0; frame < frame_count; frame++)
+ {
+ if (cpi->frames[frame] == NULL)
+ continue;
+
+ mbd->block[0].bmi.mv.as_mv.row = 0;
+ mbd->block[0].bmi.mv.as_mv.col = 0;
+
+ if (frame == alt_ref_index)
+ {
+ filter_weight = 2;
+ }
+ else
+ {
+ int err = 0;
+#if ALT_REF_MC_ENABLED
+#define THRESH_LOW 10000
+#define THRESH_HIGH 20000
+ /* Find best match in this frame by MC */
+ err = vp8_temporal_filter_find_matching_mb_c
+ (cpi,
+ cpi->frames[alt_ref_index],
+ cpi->frames[frame],
+ mb_y_offset,
+ THRESH_LOW);
+#endif
+ /* Assign higher weight to matching MB if it's error
+ * score is lower. If not applying MC default behavior
+ * is to weight all MBs equal.
+ */
+ filter_weight = err<THRESH_LOW
+ ? 2 : err<THRESH_HIGH ? 1 : 0;
+ }
+
+ if (filter_weight != 0)
+ {
+ /* Construct the predictors */
+ vp8_temporal_filter_predictors_mb_c
+ (mbd,
+ cpi->frames[frame]->y_buffer + mb_y_offset,
+ cpi->frames[frame]->u_buffer + mb_uv_offset,
+ cpi->frames[frame]->v_buffer + mb_uv_offset,
+ cpi->frames[frame]->y_stride,
+ mbd->block[0].bmi.mv.as_mv.row,
+ mbd->block[0].bmi.mv.as_mv.col,
+ predictor);
+
+ /* Apply the filter (YUV) */
+ vp8_temporal_filter_apply
+ (f->y_buffer + mb_y_offset,
+ f->y_stride,
+ predictor,
+ 16,
+ strength,
+ filter_weight,
+ accumulator,
+ count);
+
+ vp8_temporal_filter_apply
+ (f->u_buffer + mb_uv_offset,
+ f->uv_stride,
+ predictor + 256,
+ 8,
+ strength,
+ filter_weight,
+ accumulator + 256,
+ count + 256);
+
+ vp8_temporal_filter_apply
+ (f->v_buffer + mb_uv_offset,
+ f->uv_stride,
+ predictor + 320,
+ 8,
+ strength,
+ filter_weight,
+ accumulator + 320,
+ count + 320);
+ }
+ }
+
+ /* Normalize filter output to produce AltRef frame */
+ dst1 = cpi->alt_ref_buffer.y_buffer;
+ stride = cpi->alt_ref_buffer.y_stride;
+ byte = mb_y_offset;
+ for (i = 0,k = 0; i < 16; i++)
+ {
+ for (j = 0; j < 16; j++, k++)
+ {
+ unsigned int pval = accumulator[k] + (count[k] >> 1);
+ pval *= cpi->fixed_divide[count[k]];
+ pval >>= 19;
+
+ dst1[byte] = (unsigned char)pval;
+
+ /* move to next pixel */
+ byte++;
+ }
+
+ byte += stride - 16;
+ }
+
+ dst1 = cpi->alt_ref_buffer.u_buffer;
+ dst2 = cpi->alt_ref_buffer.v_buffer;
+ stride = cpi->alt_ref_buffer.uv_stride;
+ byte = mb_uv_offset;
+ for (i = 0,k = 256; i < 8; i++)
+ {
+ for (j = 0; j < 8; j++, k++)
+ {
+ int m=k+64;
+
+ /* U */
+ unsigned int pval = accumulator[k] + (count[k] >> 1);
+ pval *= cpi->fixed_divide[count[k]];
+ pval >>= 19;
+ dst1[byte] = (unsigned char)pval;
+
+ /* V */
+ pval = accumulator[m] + (count[m] >> 1);
+ pval *= cpi->fixed_divide[count[m]];
+ pval >>= 19;
+ dst2[byte] = (unsigned char)pval;
+
+ /* move to next pixel */
+ byte++;
+ }
+
+ byte += stride - 8;
+ }
+
+ mb_y_offset += 16;
+ mb_uv_offset += 8;
+ }
+
+ mb_y_offset += 16*(f->y_stride-mb_cols);
+ mb_uv_offset += 8*(f->uv_stride-mb_cols);
+ }
+
+ /* Restore input state */
+ mbd->pre.y_buffer = y_buffer;
+ mbd->pre.u_buffer = u_buffer;
+ mbd->pre.v_buffer = v_buffer;
+}
+
+void vp8_temporal_filter_prepare_c
+(
+ VP8_COMP *cpi,
+ int distance
+)
+{
+ int frame = 0;
+
+ int num_frames_backward = 0;
+ int num_frames_forward = 0;
+ int frames_to_blur_backward = 0;
+ int frames_to_blur_forward = 0;
+ int frames_to_blur = 0;
+ int start_frame = 0;
+
+ int strength = cpi->oxcf.arnr_strength;
+
+ int blur_type = cpi->oxcf.arnr_type;
+
+ int max_frames = cpi->active_arnr_frames;
+
+ num_frames_backward = distance;
+ num_frames_forward = vp8_lookahead_depth(cpi->lookahead)
+ - (num_frames_backward + 1);
+
+ switch (blur_type)
+ {
+ case 1:
+ /* Backward Blur */
+
+ frames_to_blur_backward = num_frames_backward;
+
+ if (frames_to_blur_backward >= max_frames)
+ frames_to_blur_backward = max_frames - 1;
+
+ frames_to_blur = frames_to_blur_backward + 1;
+ break;
+
+ case 2:
+ /* Forward Blur */
+
+ frames_to_blur_forward = num_frames_forward;
+
+ if (frames_to_blur_forward >= max_frames)
+ frames_to_blur_forward = max_frames - 1;
+
+ frames_to_blur = frames_to_blur_forward + 1;
+ break;
+
+ case 3:
+ default:
+ /* Center Blur */
+ frames_to_blur_forward = num_frames_forward;
+ frames_to_blur_backward = num_frames_backward;
+
+ if (frames_to_blur_forward > frames_to_blur_backward)
+ frames_to_blur_forward = frames_to_blur_backward;
+
+ if (frames_to_blur_backward > frames_to_blur_forward)
+ frames_to_blur_backward = frames_to_blur_forward;
+
+ /* When max_frames is even we have 1 more frame backward than forward */
+ if (frames_to_blur_forward > (max_frames - 1) / 2)
+ frames_to_blur_forward = ((max_frames - 1) / 2);
+
+ if (frames_to_blur_backward > (max_frames / 2))
+ frames_to_blur_backward = (max_frames / 2);
+
+ frames_to_blur = frames_to_blur_backward + frames_to_blur_forward + 1;
+ break;
+ }
+
+ start_frame = distance + frames_to_blur_forward;
+
+ /* Setup frame pointers, NULL indicates frame not included in filter */
+ vpx_memset(cpi->frames, 0, max_frames*sizeof(YV12_BUFFER_CONFIG *));
+ for (frame = 0; frame < frames_to_blur; frame++)
+ {
+ int which_buffer = start_frame - frame;
+ struct lookahead_entry* buf = vp8_lookahead_peek(cpi->lookahead,
+ which_buffer,
+ PEEK_FORWARD);
+ cpi->frames[frames_to_blur-1-frame] = &buf->img;
+ }
+
+ vp8_temporal_filter_iterate_c (
+ cpi,
+ frames_to_blur,
+ frames_to_blur_backward,
+ strength );
+}
+#endif
diff --git a/libvpx/vp8/encoder/tokenize.c b/libvpx/vp8/encoder/tokenize.c
new file mode 100644
index 0000000..3b5268b
--- /dev/null
+++ b/libvpx/vp8/encoder/tokenize.c
@@ -0,0 +1,604 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <math.h>
+#include <stdio.h>
+#include <string.h>
+#include <assert.h>
+#include "onyx_int.h"
+#include "tokenize.h"
+#include "vpx_mem/vpx_mem.h"
+
+/* Global event counters used for accumulating statistics across several
+ compressions, then generating context.c = initial stats. */
+
+#ifdef ENTROPY_STATS
+_int64 context_counters[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [MAX_ENTROPY_TOKENS];
+#endif
+void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) ;
+void vp8_fix_contexts(MACROBLOCKD *x);
+
+#include "dct_value_tokens.h"
+#include "dct_value_cost.h"
+
+const TOKENVALUE *const vp8_dct_value_tokens_ptr = dct_value_tokens +
+ DCT_MAX_VALUE;
+const short *const vp8_dct_value_cost_ptr = dct_value_cost + DCT_MAX_VALUE;
+
+#if 0
+int skip_true_count = 0;
+int skip_false_count = 0;
+#endif
+
+/* function used to generate dct_value_tokens and dct_value_cost tables */
+/*
+static void fill_value_tokens()
+{
+
+ TOKENVALUE *t = dct_value_tokens + DCT_MAX_VALUE;
+ const vp8_extra_bit_struct *e = vp8_extra_bits;
+
+ int i = -DCT_MAX_VALUE;
+ int sign = 1;
+
+ do
+ {
+ if (!i)
+ sign = 0;
+
+ {
+ const int a = sign ? -i : i;
+ int eb = sign;
+
+ if (a > 4)
+ {
+ int j = 4;
+
+ while (++j < 11 && e[j].base_val <= a) {}
+
+ t[i].Token = --j;
+ eb |= (a - e[j].base_val) << 1;
+ }
+ else
+ t[i].Token = a;
+
+ t[i].Extra = eb;
+ }
+
+ // initialize the cost for extra bits for all possible coefficient value.
+ {
+ int cost = 0;
+ const vp8_extra_bit_struct *p = vp8_extra_bits + t[i].Token;
+
+ if (p->base_val)
+ {
+ const int extra = t[i].Extra;
+ const int Length = p->Len;
+
+ if (Length)
+ cost += vp8_treed_cost(p->tree, p->prob, extra >> 1, Length);
+
+ cost += vp8_cost_bit(vp8_prob_half, extra & 1); // sign
+ dct_value_cost[i + DCT_MAX_VALUE] = cost;
+ }
+
+ }
+
+ }
+ while (++i < DCT_MAX_VALUE);
+
+ vp8_dct_value_tokens_ptr = dct_value_tokens + DCT_MAX_VALUE;
+ vp8_dct_value_cost_ptr = dct_value_cost + DCT_MAX_VALUE;
+}
+*/
+
+static void tokenize2nd_order_b
+(
+ MACROBLOCK *x,
+ TOKENEXTRA **tp,
+ VP8_COMP *cpi
+)
+{
+ MACROBLOCKD *xd = &x->e_mbd;
+ int pt; /* near block/prev token context index */
+ int c; /* start at DC */
+ TOKENEXTRA *t = *tp;/* store tokens starting here */
+ const BLOCKD *b;
+ const short *qcoeff_ptr;
+ ENTROPY_CONTEXT * a;
+ ENTROPY_CONTEXT * l;
+ int band, rc, v, token;
+ int eob;
+
+ b = xd->block + 24;
+ qcoeff_ptr = b->qcoeff;
+ a = (ENTROPY_CONTEXT *)xd->above_context + 8;
+ l = (ENTROPY_CONTEXT *)xd->left_context + 8;
+ eob = xd->eobs[24];
+ VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
+
+ if(!eob)
+ {
+ /* c = band for this case */
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [1] [0] [pt];
+ t->skip_eob_node = 0;
+
+ ++x->coef_counts [1] [0] [pt] [DCT_EOB_TOKEN];
+ t++;
+ *tp = t;
+ *a = *l = 0;
+ return;
+ }
+
+ v = qcoeff_ptr[0];
+ t->Extra = vp8_dct_value_tokens_ptr[v].Extra;
+ token = vp8_dct_value_tokens_ptr[v].Token;
+ t->Token = token;
+
+ t->context_tree = cpi->common.fc.coef_probs [1] [0] [pt];
+ t->skip_eob_node = 0;
+ ++x->coef_counts [1] [0] [pt] [token];
+ pt = vp8_prev_token_class[token];
+ t++;
+ c = 1;
+
+ for (; c < eob; c++)
+ {
+ rc = vp8_default_zig_zag1d[c];
+ band = vp8_coef_bands[c];
+ v = qcoeff_ptr[rc];
+
+ t->Extra = vp8_dct_value_tokens_ptr[v].Extra;
+ token = vp8_dct_value_tokens_ptr[v].Token;
+
+ t->Token = token;
+ t->context_tree = cpi->common.fc.coef_probs [1] [band] [pt];
+
+ t->skip_eob_node = ((pt == 0));
+
+ ++x->coef_counts [1] [band] [pt] [token];
+
+ pt = vp8_prev_token_class[token];
+ t++;
+ }
+ if (c < 16)
+ {
+ band = vp8_coef_bands[c];
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [1] [band] [pt];
+
+ t->skip_eob_node = 0;
+
+ ++x->coef_counts [1] [band] [pt] [DCT_EOB_TOKEN];
+
+ t++;
+ }
+
+ *tp = t;
+ *a = *l = 1;
+
+}
+
+static void tokenize1st_order_b
+(
+ MACROBLOCK *x,
+ TOKENEXTRA **tp,
+ int type, /* which plane: 0=Y no DC, 1=Y2, 2=UV, 3=Y with DC */
+ VP8_COMP *cpi
+)
+{
+ MACROBLOCKD *xd = &x->e_mbd;
+ unsigned int block;
+ const BLOCKD *b;
+ int pt; /* near block/prev token context index */
+ int c;
+ int token;
+ TOKENEXTRA *t = *tp;/* store tokens starting here */
+ const short *qcoeff_ptr;
+ ENTROPY_CONTEXT * a;
+ ENTROPY_CONTEXT * l;
+ int band, rc, v;
+ int tmp1, tmp2;
+
+ b = xd->block;
+ /* Luma */
+ for (block = 0; block < 16; block++, b++)
+ {
+ tmp1 = vp8_block2above[block];
+ tmp2 = vp8_block2left[block];
+ qcoeff_ptr = b->qcoeff;
+ a = (ENTROPY_CONTEXT *)xd->above_context + tmp1;
+ l = (ENTROPY_CONTEXT *)xd->left_context + tmp2;
+
+ VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
+
+ c = type ? 0 : 1;
+
+ if(c >= *b->eob)
+ {
+ /* c = band for this case */
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [type] [c] [pt];
+ t->skip_eob_node = 0;
+
+ ++x->coef_counts [type] [c] [pt] [DCT_EOB_TOKEN];
+ t++;
+ *tp = t;
+ *a = *l = 0;
+ continue;
+ }
+
+ v = qcoeff_ptr[c];
+
+ t->Extra = vp8_dct_value_tokens_ptr[v].Extra;
+ token = vp8_dct_value_tokens_ptr[v].Token;
+ t->Token = token;
+
+ t->context_tree = cpi->common.fc.coef_probs [type] [c] [pt];
+ t->skip_eob_node = 0;
+ ++x->coef_counts [type] [c] [pt] [token];
+ pt = vp8_prev_token_class[token];
+ t++;
+ c++;
+
+ for (; c < *b->eob; c++)
+ {
+ rc = vp8_default_zig_zag1d[c];
+ band = vp8_coef_bands[c];
+ v = qcoeff_ptr[rc];
+
+ t->Extra = vp8_dct_value_tokens_ptr[v].Extra;
+ token = vp8_dct_value_tokens_ptr[v].Token;
+
+ t->Token = token;
+ t->context_tree = cpi->common.fc.coef_probs [type] [band] [pt];
+
+ t->skip_eob_node = (pt == 0);
+ ++x->coef_counts [type] [band] [pt] [token];
+
+ pt = vp8_prev_token_class[token];
+ t++;
+ }
+ if (c < 16)
+ {
+ band = vp8_coef_bands[c];
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [type] [band] [pt];
+
+ t->skip_eob_node = 0;
+ ++x->coef_counts [type] [band] [pt] [DCT_EOB_TOKEN];
+
+ t++;
+ }
+ *tp = t;
+ *a = *l = 1;
+ }
+
+ /* Chroma */
+ for (block = 16; block < 24; block++, b++)
+ {
+ tmp1 = vp8_block2above[block];
+ tmp2 = vp8_block2left[block];
+ qcoeff_ptr = b->qcoeff;
+ a = (ENTROPY_CONTEXT *)xd->above_context + tmp1;
+ l = (ENTROPY_CONTEXT *)xd->left_context + tmp2;
+
+ VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
+
+ if(!(*b->eob))
+ {
+ /* c = band for this case */
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [2] [0] [pt];
+ t->skip_eob_node = 0;
+
+ ++x->coef_counts [2] [0] [pt] [DCT_EOB_TOKEN];
+ t++;
+ *tp = t;
+ *a = *l = 0;
+ continue;
+ }
+
+ v = qcoeff_ptr[0];
+
+ t->Extra = vp8_dct_value_tokens_ptr[v].Extra;
+ token = vp8_dct_value_tokens_ptr[v].Token;
+ t->Token = token;
+
+ t->context_tree = cpi->common.fc.coef_probs [2] [0] [pt];
+ t->skip_eob_node = 0;
+ ++x->coef_counts [2] [0] [pt] [token];
+ pt = vp8_prev_token_class[token];
+ t++;
+ c = 1;
+
+ for (; c < *b->eob; c++)
+ {
+ rc = vp8_default_zig_zag1d[c];
+ band = vp8_coef_bands[c];
+ v = qcoeff_ptr[rc];
+
+ t->Extra = vp8_dct_value_tokens_ptr[v].Extra;
+ token = vp8_dct_value_tokens_ptr[v].Token;
+
+ t->Token = token;
+ t->context_tree = cpi->common.fc.coef_probs [2] [band] [pt];
+
+ t->skip_eob_node = (pt == 0);
+
+ ++x->coef_counts [2] [band] [pt] [token];
+
+ pt = vp8_prev_token_class[token];
+ t++;
+ }
+ if (c < 16)
+ {
+ band = vp8_coef_bands[c];
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [2] [band] [pt];
+
+ t->skip_eob_node = 0;
+
+ ++x->coef_counts [2] [band] [pt] [DCT_EOB_TOKEN];
+
+ t++;
+ }
+ *tp = t;
+ *a = *l = 1;
+ }
+}
+
+
+static int mb_is_skippable(MACROBLOCKD *x, int has_y2_block)
+{
+ int skip = 1;
+ int i = 0;
+
+ if (has_y2_block)
+ {
+ for (i = 0; i < 16; i++)
+ skip &= (x->eobs[i] < 2);
+ }
+
+ for (; i < 24 + has_y2_block; i++)
+ skip &= (!x->eobs[i]);
+
+ return skip;
+}
+
+
+void vp8_tokenize_mb(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t)
+{
+ MACROBLOCKD *xd = &x->e_mbd;
+ int plane_type;
+ int has_y2_block;
+
+ has_y2_block = (xd->mode_info_context->mbmi.mode != B_PRED
+ && xd->mode_info_context->mbmi.mode != SPLITMV);
+
+ xd->mode_info_context->mbmi.mb_skip_coeff =
+ mb_is_skippable(xd, has_y2_block);
+ if (xd->mode_info_context->mbmi.mb_skip_coeff)
+ {
+ if (!cpi->common.mb_no_coeff_skip)
+ {
+ vp8_stuff_mb(cpi, x, t);
+ }
+ else
+ {
+ vp8_fix_contexts(xd);
+ x->skip_true_count++;
+ }
+
+ return;
+ }
+
+ plane_type = 3;
+ if(has_y2_block)
+ {
+ tokenize2nd_order_b(x, t, cpi);
+ plane_type = 0;
+ }
+
+ tokenize1st_order_b(x, t, plane_type, cpi);
+}
+
+
+#ifdef ENTROPY_STATS
+
+void init_context_counters(void)
+{
+ vpx_memset(context_counters, 0, sizeof(context_counters));
+}
+
+void print_context_counters()
+{
+
+ int type, band, pt, t;
+
+ FILE *const f = fopen("context.c", "w");
+
+ fprintf(f, "#include \"entropy.h\"\n");
+
+ fprintf(f, "\n/* *** GENERATED FILE: DO NOT EDIT *** */\n\n");
+
+ fprintf(f, "int Contexts[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [MAX_ENTROPY_TOKENS];\n\n");
+
+ fprintf(f, "const int default_contexts[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [MAX_ENTROPY_TOKENS] = {");
+
+# define Comma( X) (X? ",":"")
+
+ type = 0;
+
+ do
+ {
+ fprintf(f, "%s\n { /* block Type %d */", Comma(type), type);
+
+ band = 0;
+
+ do
+ {
+ fprintf(f, "%s\n { /* Coeff Band %d */", Comma(band), band);
+
+ pt = 0;
+
+ do
+ {
+ fprintf(f, "%s\n {", Comma(pt));
+
+ t = 0;
+
+ do
+ {
+ const _int64 x = context_counters [type] [band] [pt] [t];
+ const int y = (int) x;
+
+ assert(x == (_int64) y); /* no overflow handling yet */
+ fprintf(f, "%s %d", Comma(t), y);
+
+ }
+ while (++t < MAX_ENTROPY_TOKENS);
+
+ fprintf(f, "}");
+ }
+ while (++pt < PREV_COEF_CONTEXTS);
+
+ fprintf(f, "\n }");
+
+ }
+ while (++band < COEF_BANDS);
+
+ fprintf(f, "\n }");
+ }
+ while (++type < BLOCK_TYPES);
+
+ fprintf(f, "\n};\n");
+ fclose(f);
+}
+#endif
+
+
+static void stuff2nd_order_b
+(
+ TOKENEXTRA **tp,
+ ENTROPY_CONTEXT *a,
+ ENTROPY_CONTEXT *l,
+ VP8_COMP *cpi,
+ MACROBLOCK *x
+)
+{
+ int pt; /* near block/prev token context index */
+ TOKENEXTRA *t = *tp; /* store tokens starting here */
+ VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
+
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [1] [0] [pt];
+ t->skip_eob_node = 0;
+ ++x->coef_counts [1] [0] [pt] [DCT_EOB_TOKEN];
+ ++t;
+
+ *tp = t;
+ pt = 0;
+ *a = *l = pt;
+}
+
+static void stuff1st_order_b
+(
+ TOKENEXTRA **tp,
+ ENTROPY_CONTEXT *a,
+ ENTROPY_CONTEXT *l,
+ int type,
+ VP8_COMP *cpi,
+ MACROBLOCK *x
+)
+{
+ int pt; /* near block/prev token context index */
+ int band;
+ TOKENEXTRA *t = *tp; /* store tokens starting here */
+ VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
+ band = type ? 0 : 1;
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [type] [band] [pt];
+ t->skip_eob_node = 0;
+ ++x->coef_counts [type] [band] [pt] [DCT_EOB_TOKEN];
+ ++t;
+ *tp = t;
+ pt = 0; /* 0 <-> all coeff data is zero */
+ *a = *l = pt;
+}
+
+static
+void stuff1st_order_buv
+(
+ TOKENEXTRA **tp,
+ ENTROPY_CONTEXT *a,
+ ENTROPY_CONTEXT *l,
+ VP8_COMP *cpi,
+ MACROBLOCK *x
+)
+{
+ int pt; /* near block/prev token context index */
+ TOKENEXTRA *t = *tp; /* store tokens starting here */
+ VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
+
+ t->Token = DCT_EOB_TOKEN;
+ t->context_tree = cpi->common.fc.coef_probs [2] [0] [pt];
+ t->skip_eob_node = 0;
+ ++x->coef_counts[2] [0] [pt] [DCT_EOB_TOKEN];
+ ++t;
+ *tp = t;
+ pt = 0; /* 0 <-> all coeff data is zero */
+ *a = *l = pt;
+}
+
+void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t)
+{
+ MACROBLOCKD *xd = &x->e_mbd;
+ ENTROPY_CONTEXT * A = (ENTROPY_CONTEXT *)xd->above_context;
+ ENTROPY_CONTEXT * L = (ENTROPY_CONTEXT *)xd->left_context;
+ int plane_type;
+ int b;
+ plane_type = 3;
+ if((xd->mode_info_context->mbmi.mode != B_PRED
+ && xd->mode_info_context->mbmi.mode != SPLITMV))
+ {
+ stuff2nd_order_b(t,
+ A + vp8_block2above[24], L + vp8_block2left[24], cpi, x);
+ plane_type = 0;
+ }
+
+ for (b = 0; b < 16; b++)
+ stuff1st_order_b(t,
+ A + vp8_block2above[b],
+ L + vp8_block2left[b], plane_type, cpi, x);
+
+ for (b = 16; b < 24; b++)
+ stuff1st_order_buv(t,
+ A + vp8_block2above[b],
+ L + vp8_block2left[b], cpi, x);
+
+}
+void vp8_fix_contexts(MACROBLOCKD *x)
+{
+ /* Clear entropy contexts for Y2 blocks */
+ if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV)
+ {
+ vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES));
+ }
+ else
+ {
+ vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1);
+ vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1);
+ }
+
+}
diff --git a/libvpx/vp8/encoder/tokenize.h b/libvpx/vp8/encoder/tokenize.h
new file mode 100644
index 0000000..c2d1438
--- /dev/null
+++ b/libvpx/vp8/encoder/tokenize.h
@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef tokenize_h
+#define tokenize_h
+
+#include "vp8/common/entropy.h"
+#include "block.h"
+
+void vp8_tokenize_initialize();
+
+typedef struct
+{
+ short Token;
+ short Extra;
+} TOKENVALUE;
+
+typedef struct
+{
+ const vp8_prob *context_tree;
+ short Extra;
+ unsigned char Token;
+ unsigned char skip_eob_node;
+} TOKENEXTRA;
+
+int rd_cost_mby(MACROBLOCKD *);
+
+#ifdef ENTROPY_STATS
+void init_context_counters();
+void print_context_counters();
+
+extern _int64 context_counters[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [MAX_ENTROPY_TOKENS];
+#endif
+
+extern const short *const vp8_dct_value_cost_ptr;
+/* TODO: The Token field should be broken out into a separate char array to
+ * improve cache locality, since it's needed for costing when the rest of the
+ * fields are not.
+ */
+extern const TOKENVALUE *const vp8_dct_value_tokens_ptr;
+
+#endif /* tokenize_h */
diff --git a/libvpx/vp8/encoder/treewriter.c b/libvpx/vp8/encoder/treewriter.c
new file mode 100644
index 0000000..ef25f67
--- /dev/null
+++ b/libvpx/vp8/encoder/treewriter.c
@@ -0,0 +1,43 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "treewriter.h"
+
+static void cost(
+ int *const C,
+ vp8_tree T,
+ const vp8_prob *const P,
+ int i,
+ int c
+)
+{
+ const vp8_prob p = P [i>>1];
+
+ do
+ {
+ const vp8_tree_index j = T[i];
+ const int d = c + vp8_cost_bit(p, i & 1);
+
+ if (j <= 0)
+ C[-j] = d;
+ else
+ cost(C, T, P, j, d);
+ }
+ while (++i & 1);
+}
+void vp8_cost_tokens(int *c, const vp8_prob *p, vp8_tree t)
+{
+ cost(c, t, p, 0, 0);
+}
+void vp8_cost_tokens2(int *c, const vp8_prob *p, vp8_tree t,int start)
+{
+ cost(c, t, p, start, 0);
+}
diff --git a/libvpx/vp8/encoder/treewriter.h b/libvpx/vp8/encoder/treewriter.h
new file mode 100644
index 0000000..48574f3
--- /dev/null
+++ b/libvpx/vp8/encoder/treewriter.h
@@ -0,0 +1,126 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __INC_TREEWRITER_H
+#define __INC_TREEWRITER_H
+
+/* Trees map alphabets into huffman-like codes suitable for an arithmetic
+ bit coder. Timothy S Murphy 11 October 2004 */
+
+#include "vp8/common/treecoder.h"
+
+#include "boolhuff.h" /* for now */
+
+typedef BOOL_CODER vp8_writer;
+
+#define vp8_write vp8_encode_bool
+#define vp8_write_literal vp8_encode_value
+#define vp8_write_bit( W, V) vp8_write( W, V, vp8_prob_half)
+
+#define vp8bc_write vp8bc_write_bool
+#define vp8bc_write_literal vp8bc_write_bits
+#define vp8bc_write_bit( W, V) vp8bc_write_bits( W, V, 1)
+
+
+/* Approximate length of an encoded bool in 256ths of a bit at given prob */
+
+#define vp8_cost_zero( x) ( vp8_prob_cost[x])
+#define vp8_cost_one( x) vp8_cost_zero( vp8_complement(x))
+
+#define vp8_cost_bit( x, b) vp8_cost_zero( (b)? vp8_complement(x) : (x) )
+
+/* VP8BC version is scaled by 2^20 rather than 2^8; see bool_coder.h */
+
+
+/* Both of these return bits, not scaled bits. */
+
+static unsigned int vp8_cost_branch(const unsigned int ct[2], vp8_prob p)
+{
+ /* Imitate existing calculation */
+
+ return ((ct[0] * vp8_cost_zero(p))
+ + (ct[1] * vp8_cost_one(p))) >> 8;
+}
+
+/* Small functions to write explicit values and tokens, as well as
+ estimate their lengths. */
+
+static void vp8_treed_write
+(
+ vp8_writer *const w,
+ vp8_tree t,
+ const vp8_prob *const p,
+ int v,
+ int n /* number of bits in v, assumed nonzero */
+)
+{
+ vp8_tree_index i = 0;
+
+ do
+ {
+ const int b = (v >> --n) & 1;
+ vp8_write(w, b, p[i>>1]);
+ i = t[i+b];
+ }
+ while (n);
+}
+static void vp8_write_token
+(
+ vp8_writer *const w,
+ vp8_tree t,
+ const vp8_prob *const p,
+ vp8_token *const x
+)
+{
+ vp8_treed_write(w, t, p, x->value, x->Len);
+}
+
+static int vp8_treed_cost(
+ vp8_tree t,
+ const vp8_prob *const p,
+ int v,
+ int n /* number of bits in v, assumed nonzero */
+)
+{
+ int c = 0;
+ vp8_tree_index i = 0;
+
+ do
+ {
+ const int b = (v >> --n) & 1;
+ c += vp8_cost_bit(p[i>>1], b);
+ i = t[i+b];
+ }
+ while (n);
+
+ return c;
+}
+static int vp8_cost_token
+(
+ vp8_tree t,
+ const vp8_prob *const p,
+ vp8_token *const x
+)
+{
+ return vp8_treed_cost(t, p, x->value, x->Len);
+}
+
+/* Fill array of costs for all possible token values. */
+
+void vp8_cost_tokens(
+ int *Costs, const vp8_prob *, vp8_tree
+);
+
+void vp8_cost_tokens2(
+ int *Costs, const vp8_prob *, vp8_tree, int
+);
+
+#endif
diff --git a/libvpx/vp8/encoder/x86/dct_mmx.asm b/libvpx/vp8/encoder/x86/dct_mmx.asm
new file mode 100644
index 0000000..6f188cb
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/dct_mmx.asm
@@ -0,0 +1,241 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void vp8_short_fdct4x4_mmx(short *input, short *output, int pitch)
+global sym(vp8_short_fdct4x4_mmx) PRIVATE
+sym(vp8_short_fdct4x4_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 3
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ; input
+ mov rdi, arg(1) ; output
+
+ movsxd rax, dword ptr arg(2) ;pitch
+
+ lea rcx, [rsi + rax*2]
+ ; read the input data
+ movq mm0, [rsi]
+ movq mm1, [rsi + rax]
+
+ movq mm2, [rcx]
+ movq mm4, [rcx + rax]
+
+ ; transpose for the first stage
+ movq mm3, mm0 ; 00 01 02 03
+ movq mm5, mm2 ; 20 21 22 23
+
+ punpcklwd mm0, mm1 ; 00 10 01 11
+ punpckhwd mm3, mm1 ; 02 12 03 13
+
+ punpcklwd mm2, mm4 ; 20 30 21 31
+ punpckhwd mm5, mm4 ; 22 32 23 33
+
+ movq mm1, mm0 ; 00 10 01 11
+ punpckldq mm0, mm2 ; 00 10 20 30
+
+ punpckhdq mm1, mm2 ; 01 11 21 31
+
+ movq mm2, mm3 ; 02 12 03 13
+ punpckldq mm2, mm5 ; 02 12 22 32
+
+ punpckhdq mm3, mm5 ; 03 13 23 33
+
+ ; mm0 0
+ ; mm1 1
+ ; mm2 2
+ ; mm3 3
+
+ ; first stage
+ movq mm5, mm0
+ movq mm4, mm1
+
+ paddw mm0, mm3 ; a1 = 0 + 3
+ paddw mm1, mm2 ; b1 = 1 + 2
+
+ psubw mm4, mm2 ; c1 = 1 - 2
+ psubw mm5, mm3 ; d1 = 0 - 3
+
+ psllw mm5, 3
+ psllw mm4, 3
+
+ psllw mm0, 3
+ psllw mm1, 3
+
+ ; output 0 and 2
+ movq mm2, mm0 ; a1
+
+ paddw mm0, mm1 ; op[0] = a1 + b1
+ psubw mm2, mm1 ; op[2] = a1 - b1
+
+ ; output 1 and 3
+ ; interleave c1, d1
+ movq mm1, mm5 ; d1
+ punpcklwd mm1, mm4 ; c1 d1
+ punpckhwd mm5, mm4 ; c1 d1
+
+ movq mm3, mm1
+ movq mm4, mm5
+
+ pmaddwd mm1, MMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
+ pmaddwd mm4, MMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
+
+ pmaddwd mm3, MMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
+ pmaddwd mm5, MMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
+
+ paddd mm1, MMWORD PTR[GLOBAL(_14500)]
+ paddd mm4, MMWORD PTR[GLOBAL(_14500)]
+ paddd mm3, MMWORD PTR[GLOBAL(_7500)]
+ paddd mm5, MMWORD PTR[GLOBAL(_7500)]
+
+ psrad mm1, 12 ; (c1 * 2217 + d1 * 5352 + 14500)>>12
+ psrad mm4, 12 ; (c1 * 2217 + d1 * 5352 + 14500)>>12
+ psrad mm3, 12 ; (d1 * 2217 - c1 * 5352 + 7500)>>12
+ psrad mm5, 12 ; (d1 * 2217 - c1 * 5352 + 7500)>>12
+
+ packssdw mm1, mm4 ; op[1]
+ packssdw mm3, mm5 ; op[3]
+
+ ; done with vertical
+ ; transpose for the second stage
+ movq mm4, mm0 ; 00 10 20 30
+ movq mm5, mm2 ; 02 12 22 32
+
+ punpcklwd mm0, mm1 ; 00 01 10 11
+ punpckhwd mm4, mm1 ; 20 21 30 31
+
+ punpcklwd mm2, mm3 ; 02 03 12 13
+ punpckhwd mm5, mm3 ; 22 23 32 33
+
+ movq mm1, mm0 ; 00 01 10 11
+ punpckldq mm0, mm2 ; 00 01 02 03
+
+ punpckhdq mm1, mm2 ; 01 22 12 13
+
+ movq mm2, mm4 ; 20 31 30 31
+ punpckldq mm2, mm5 ; 20 21 22 23
+
+ punpckhdq mm4, mm5 ; 30 31 32 33
+
+ ; mm0 0
+ ; mm1 1
+ ; mm2 2
+ ; mm3 4
+
+ movq mm5, mm0
+ movq mm3, mm1
+
+ paddw mm0, mm4 ; a1 = 0 + 3
+ paddw mm1, mm2 ; b1 = 1 + 2
+
+ psubw mm3, mm2 ; c1 = 1 - 2
+ psubw mm5, mm4 ; d1 = 0 - 3
+
+ pxor mm6, mm6 ; zero out for compare
+
+ pcmpeqw mm6, mm5 ; d1 != 0
+
+ pandn mm6, MMWORD PTR[GLOBAL(_cmp_mask)] ; clear upper,
+ ; and keep bit 0 of lower
+
+ ; output 0 and 2
+ movq mm2, mm0 ; a1
+
+ paddw mm0, mm1 ; a1 + b1
+ psubw mm2, mm1 ; a1 - b1
+
+ paddw mm0, MMWORD PTR[GLOBAL(_7w)]
+ paddw mm2, MMWORD PTR[GLOBAL(_7w)]
+
+ psraw mm0, 4 ; op[0] = (a1 + b1 + 7)>>4
+ psraw mm2, 4 ; op[8] = (a1 - b1 + 7)>>4
+
+ movq MMWORD PTR[rdi + 0 ], mm0
+ movq MMWORD PTR[rdi + 16], mm2
+
+ ; output 1 and 3
+ ; interleave c1, d1
+ movq mm1, mm5 ; d1
+ punpcklwd mm1, mm3 ; c1 d1
+ punpckhwd mm5, mm3 ; c1 d1
+
+ movq mm3, mm1
+ movq mm4, mm5
+
+ pmaddwd mm1, MMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
+ pmaddwd mm4, MMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
+
+ pmaddwd mm3, MMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
+ pmaddwd mm5, MMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
+
+ paddd mm1, MMWORD PTR[GLOBAL(_12000)]
+ paddd mm4, MMWORD PTR[GLOBAL(_12000)]
+ paddd mm3, MMWORD PTR[GLOBAL(_51000)]
+ paddd mm5, MMWORD PTR[GLOBAL(_51000)]
+
+ psrad mm1, 16 ; (c1 * 2217 + d1 * 5352 + 14500)>>16
+ psrad mm4, 16 ; (c1 * 2217 + d1 * 5352 + 14500)>>16
+ psrad mm3, 16 ; (d1 * 2217 - c1 * 5352 + 7500)>>16
+ psrad mm5, 16 ; (d1 * 2217 - c1 * 5352 + 7500)>>16
+
+ packssdw mm1, mm4 ; op[4]
+ packssdw mm3, mm5 ; op[12]
+
+ paddw mm1, mm6 ; op[4] += (d1!=0)
+
+ movq MMWORD PTR[rdi + 8 ], mm1
+ movq MMWORD PTR[rdi + 24], mm3
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 8
+_5352_2217:
+ dw 5352
+ dw 2217
+ dw 5352
+ dw 2217
+align 8
+_2217_neg5352:
+ dw 2217
+ dw -5352
+ dw 2217
+ dw -5352
+align 8
+_cmp_mask:
+ times 4 dw 1
+align 8
+_7w:
+ times 4 dw 7
+align 8
+_14500:
+ times 2 dd 14500
+align 8
+_7500:
+ times 2 dd 7500
+align 8
+_12000:
+ times 2 dd 12000
+align 8
+_51000:
+ times 2 dd 51000
diff --git a/libvpx/vp8/encoder/x86/dct_sse2.asm b/libvpx/vp8/encoder/x86/dct_sse2.asm
new file mode 100644
index 0000000..d880ce0
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/dct_sse2.asm
@@ -0,0 +1,432 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%macro STACK_FRAME_CREATE 0
+%if ABI_IS_32BIT
+ %define input rsi
+ %define output rdi
+ %define pitch rax
+ push rbp
+ mov rbp, rsp
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0)
+ mov rdi, arg(1)
+
+ movsxd rax, dword ptr arg(2)
+ lea rcx, [rsi + rax*2]
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ %define input rcx
+ %define output rdx
+ %define pitch r8
+ SAVE_XMM 7, u
+ %else
+ %define input rdi
+ %define output rsi
+ %define pitch rdx
+ %endif
+%endif
+%endmacro
+
+%macro STACK_FRAME_DESTROY 0
+ %define input
+ %define output
+ %define pitch
+
+%if ABI_IS_32BIT
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ pop rbp
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ RESTORE_XMM
+ %endif
+%endif
+ ret
+%endmacro
+
+;void vp8_short_fdct4x4_sse2(short *input, short *output, int pitch)
+global sym(vp8_short_fdct4x4_sse2) PRIVATE
+sym(vp8_short_fdct4x4_sse2):
+
+ STACK_FRAME_CREATE
+
+ movq xmm0, MMWORD PTR[input ] ;03 02 01 00
+ movq xmm2, MMWORD PTR[input+ pitch] ;13 12 11 10
+ lea input, [input+2*pitch]
+ movq xmm1, MMWORD PTR[input ] ;23 22 21 20
+ movq xmm3, MMWORD PTR[input+ pitch] ;33 32 31 30
+
+ punpcklqdq xmm0, xmm2 ;13 12 11 10 03 02 01 00
+ punpcklqdq xmm1, xmm3 ;33 32 31 30 23 22 21 20
+
+ movdqa xmm2, xmm0
+ punpckldq xmm0, xmm1 ;23 22 03 02 21 20 01 00
+ punpckhdq xmm2, xmm1 ;33 32 13 12 31 30 11 10
+ movdqa xmm1, xmm0
+ punpckldq xmm0, xmm2 ;31 21 30 20 11 10 01 00
+ pshufhw xmm1, xmm1, 0b1h ;22 23 02 03 xx xx xx xx
+ pshufhw xmm2, xmm2, 0b1h ;32 33 12 13 xx xx xx xx
+
+ punpckhdq xmm1, xmm2 ;32 33 22 23 12 13 02 03
+ movdqa xmm3, xmm0
+ paddw xmm0, xmm1 ;b1 a1 b1 a1 b1 a1 b1 a1
+ psubw xmm3, xmm1 ;c1 d1 c1 d1 c1 d1 c1 d1
+ psllw xmm0, 3 ;b1 <<= 3 a1 <<= 3
+ psllw xmm3, 3 ;c1 <<= 3 d1 <<= 3
+
+ movdqa xmm1, xmm0
+ pmaddwd xmm0, XMMWORD PTR[GLOBAL(_mult_add)] ;a1 + b1
+ pmaddwd xmm1, XMMWORD PTR[GLOBAL(_mult_sub)] ;a1 - b1
+ movdqa xmm4, xmm3
+ pmaddwd xmm3, XMMWORD PTR[GLOBAL(_5352_2217)] ;c1*2217 + d1*5352
+ pmaddwd xmm4, XMMWORD PTR[GLOBAL(_2217_neg5352)];d1*2217 - c1*5352
+
+ paddd xmm3, XMMWORD PTR[GLOBAL(_14500)]
+ paddd xmm4, XMMWORD PTR[GLOBAL(_7500)]
+ psrad xmm3, 12 ;(c1 * 2217 + d1 * 5352 + 14500)>>12
+ psrad xmm4, 12 ;(d1 * 2217 - c1 * 5352 + 7500)>>12
+
+ packssdw xmm0, xmm1 ;op[2] op[0]
+ packssdw xmm3, xmm4 ;op[3] op[1]
+ ; 23 22 21 20 03 02 01 00
+ ;
+ ; 33 32 31 30 13 12 11 10
+ ;
+ movdqa xmm2, xmm0
+ punpcklqdq xmm0, xmm3 ;13 12 11 10 03 02 01 00
+ punpckhqdq xmm2, xmm3 ;23 22 21 20 33 32 31 30
+
+ movdqa xmm3, xmm0
+ punpcklwd xmm0, xmm2 ;32 30 22 20 12 10 02 00
+ punpckhwd xmm3, xmm2 ;33 31 23 21 13 11 03 01
+ movdqa xmm2, xmm0
+ punpcklwd xmm0, xmm3 ;13 12 11 10 03 02 01 00
+ punpckhwd xmm2, xmm3 ;33 32 31 30 23 22 21 20
+
+ movdqa xmm5, XMMWORD PTR[GLOBAL(_7)]
+ pshufd xmm2, xmm2, 04eh
+ movdqa xmm3, xmm0
+ paddw xmm0, xmm2 ;b1 b1 b1 b1 a1 a1 a1 a1
+ psubw xmm3, xmm2 ;c1 c1 c1 c1 d1 d1 d1 d1
+
+ pshufd xmm0, xmm0, 0d8h ;b1 b1 a1 a1 b1 b1 a1 a1
+ movdqa xmm2, xmm3 ;save d1 for compare
+ pshufd xmm3, xmm3, 0d8h ;c1 c1 d1 d1 c1 c1 d1 d1
+ pshuflw xmm0, xmm0, 0d8h ;b1 b1 a1 a1 b1 a1 b1 a1
+ pshuflw xmm3, xmm3, 0d8h ;c1 c1 d1 d1 c1 d1 c1 d1
+ pshufhw xmm0, xmm0, 0d8h ;b1 a1 b1 a1 b1 a1 b1 a1
+ pshufhw xmm3, xmm3, 0d8h ;c1 d1 c1 d1 c1 d1 c1 d1
+ movdqa xmm1, xmm0
+ pmaddwd xmm0, XMMWORD PTR[GLOBAL(_mult_add)] ;a1 + b1
+ pmaddwd xmm1, XMMWORD PTR[GLOBAL(_mult_sub)] ;a1 - b1
+
+ pxor xmm4, xmm4 ;zero out for compare
+ paddd xmm0, xmm5
+ paddd xmm1, xmm5
+ pcmpeqw xmm2, xmm4
+ psrad xmm0, 4 ;(a1 + b1 + 7)>>4
+ psrad xmm1, 4 ;(a1 - b1 + 7)>>4
+ pandn xmm2, XMMWORD PTR[GLOBAL(_cmp_mask)] ;clear upper,
+ ;and keep bit 0 of lower
+
+ movdqa xmm4, xmm3
+ pmaddwd xmm3, XMMWORD PTR[GLOBAL(_5352_2217)] ;c1*2217 + d1*5352
+ pmaddwd xmm4, XMMWORD PTR[GLOBAL(_2217_neg5352)] ;d1*2217 - c1*5352
+ paddd xmm3, XMMWORD PTR[GLOBAL(_12000)]
+ paddd xmm4, XMMWORD PTR[GLOBAL(_51000)]
+ packssdw xmm0, xmm1 ;op[8] op[0]
+ psrad xmm3, 16 ;(c1 * 2217 + d1 * 5352 + 12000)>>16
+ psrad xmm4, 16 ;(d1 * 2217 - c1 * 5352 + 51000)>>16
+
+ packssdw xmm3, xmm4 ;op[12] op[4]
+ movdqa xmm1, xmm0
+ paddw xmm3, xmm2 ;op[4] += (d1!=0)
+ punpcklqdq xmm0, xmm3 ;op[4] op[0]
+ punpckhqdq xmm1, xmm3 ;op[12] op[8]
+
+ movdqa XMMWORD PTR[output + 0], xmm0
+ movdqa XMMWORD PTR[output + 16], xmm1
+
+ STACK_FRAME_DESTROY
+
+;void vp8_short_fdct8x4_sse2(short *input, short *output, int pitch)
+global sym(vp8_short_fdct8x4_sse2) PRIVATE
+sym(vp8_short_fdct8x4_sse2):
+
+ STACK_FRAME_CREATE
+
+ ; read the input data
+ movdqa xmm0, [input ]
+ movdqa xmm2, [input+ pitch]
+ lea input, [input+2*pitch]
+ movdqa xmm4, [input ]
+ movdqa xmm3, [input+ pitch]
+
+ ; transpose for the first stage
+ movdqa xmm1, xmm0 ; 00 01 02 03 04 05 06 07
+ movdqa xmm5, xmm4 ; 20 21 22 23 24 25 26 27
+
+ punpcklwd xmm0, xmm2 ; 00 10 01 11 02 12 03 13
+ punpckhwd xmm1, xmm2 ; 04 14 05 15 06 16 07 17
+
+ punpcklwd xmm4, xmm3 ; 20 30 21 31 22 32 23 33
+ punpckhwd xmm5, xmm3 ; 24 34 25 35 26 36 27 37
+
+ movdqa xmm2, xmm0 ; 00 10 01 11 02 12 03 13
+ punpckldq xmm0, xmm4 ; 00 10 20 30 01 11 21 31
+
+ punpckhdq xmm2, xmm4 ; 02 12 22 32 03 13 23 33
+
+ movdqa xmm4, xmm1 ; 04 14 05 15 06 16 07 17
+ punpckldq xmm4, xmm5 ; 04 14 24 34 05 15 25 35
+
+ punpckhdq xmm1, xmm5 ; 06 16 26 36 07 17 27 37
+ movdqa xmm3, xmm2 ; 02 12 22 32 03 13 23 33
+
+ punpckhqdq xmm3, xmm1 ; 03 13 23 33 07 17 27 37
+ punpcklqdq xmm2, xmm1 ; 02 12 22 32 06 16 26 36
+
+ movdqa xmm1, xmm0 ; 00 10 20 30 01 11 21 31
+ punpcklqdq xmm0, xmm4 ; 00 10 20 30 04 14 24 34
+
+ punpckhqdq xmm1, xmm4 ; 01 11 21 32 05 15 25 35
+
+ ; xmm0 0
+ ; xmm1 1
+ ; xmm2 2
+ ; xmm3 3
+
+ ; first stage
+ movdqa xmm5, xmm0
+ movdqa xmm4, xmm1
+
+ paddw xmm0, xmm3 ; a1 = 0 + 3
+ paddw xmm1, xmm2 ; b1 = 1 + 2
+
+ psubw xmm4, xmm2 ; c1 = 1 - 2
+ psubw xmm5, xmm3 ; d1 = 0 - 3
+
+ psllw xmm5, 3
+ psllw xmm4, 3
+
+ psllw xmm0, 3
+ psllw xmm1, 3
+
+ ; output 0 and 2
+ movdqa xmm2, xmm0 ; a1
+
+ paddw xmm0, xmm1 ; op[0] = a1 + b1
+ psubw xmm2, xmm1 ; op[2] = a1 - b1
+
+ ; output 1 and 3
+ ; interleave c1, d1
+ movdqa xmm1, xmm5 ; d1
+ punpcklwd xmm1, xmm4 ; c1 d1
+ punpckhwd xmm5, xmm4 ; c1 d1
+
+ movdqa xmm3, xmm1
+ movdqa xmm4, xmm5
+
+ pmaddwd xmm1, XMMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
+ pmaddwd xmm4, XMMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
+
+ pmaddwd xmm3, XMMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
+ pmaddwd xmm5, XMMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
+
+ paddd xmm1, XMMWORD PTR[GLOBAL(_14500)]
+ paddd xmm4, XMMWORD PTR[GLOBAL(_14500)]
+ paddd xmm3, XMMWORD PTR[GLOBAL(_7500)]
+ paddd xmm5, XMMWORD PTR[GLOBAL(_7500)]
+
+ psrad xmm1, 12 ; (c1 * 2217 + d1 * 5352 + 14500)>>12
+ psrad xmm4, 12 ; (c1 * 2217 + d1 * 5352 + 14500)>>12
+ psrad xmm3, 12 ; (d1 * 2217 - c1 * 5352 + 7500)>>12
+ psrad xmm5, 12 ; (d1 * 2217 - c1 * 5352 + 7500)>>12
+
+ packssdw xmm1, xmm4 ; op[1]
+ packssdw xmm3, xmm5 ; op[3]
+
+ ; done with vertical
+ ; transpose for the second stage
+ movdqa xmm4, xmm0 ; 00 10 20 30 04 14 24 34
+ movdqa xmm5, xmm2 ; 02 12 22 32 06 16 26 36
+
+ punpcklwd xmm0, xmm1 ; 00 01 10 11 20 21 30 31
+ punpckhwd xmm4, xmm1 ; 04 05 14 15 24 25 34 35
+
+ punpcklwd xmm2, xmm3 ; 02 03 12 13 22 23 32 33
+ punpckhwd xmm5, xmm3 ; 06 07 16 17 26 27 36 37
+
+ movdqa xmm1, xmm0 ; 00 01 10 11 20 21 30 31
+ punpckldq xmm0, xmm2 ; 00 01 02 03 10 11 12 13
+
+ punpckhdq xmm1, xmm2 ; 20 21 22 23 30 31 32 33
+
+ movdqa xmm2, xmm4 ; 04 05 14 15 24 25 34 35
+ punpckldq xmm2, xmm5 ; 04 05 06 07 14 15 16 17
+
+ punpckhdq xmm4, xmm5 ; 24 25 26 27 34 35 36 37
+ movdqa xmm3, xmm1 ; 20 21 22 23 30 31 32 33
+
+ punpckhqdq xmm3, xmm4 ; 30 31 32 33 34 35 36 37
+ punpcklqdq xmm1, xmm4 ; 20 21 22 23 24 25 26 27
+
+ movdqa xmm4, xmm0 ; 00 01 02 03 10 11 12 13
+ punpcklqdq xmm0, xmm2 ; 00 01 02 03 04 05 06 07
+
+ punpckhqdq xmm4, xmm2 ; 10 11 12 13 14 15 16 17
+
+ ; xmm0 0
+ ; xmm1 4
+ ; xmm2 1
+ ; xmm3 3
+
+ movdqa xmm5, xmm0
+ movdqa xmm2, xmm1
+
+ paddw xmm0, xmm3 ; a1 = 0 + 3
+ paddw xmm1, xmm4 ; b1 = 1 + 2
+
+ psubw xmm4, xmm2 ; c1 = 1 - 2
+ psubw xmm5, xmm3 ; d1 = 0 - 3
+
+ pxor xmm6, xmm6 ; zero out for compare
+
+ pcmpeqw xmm6, xmm5 ; d1 != 0
+
+ pandn xmm6, XMMWORD PTR[GLOBAL(_cmp_mask8x4)] ; clear upper,
+ ; and keep bit 0 of lower
+
+ ; output 0 and 2
+ movdqa xmm2, xmm0 ; a1
+
+ paddw xmm0, xmm1 ; a1 + b1
+ psubw xmm2, xmm1 ; a1 - b1
+
+ paddw xmm0, XMMWORD PTR[GLOBAL(_7w)]
+ paddw xmm2, XMMWORD PTR[GLOBAL(_7w)]
+
+ psraw xmm0, 4 ; op[0] = (a1 + b1 + 7)>>4
+ psraw xmm2, 4 ; op[8] = (a1 - b1 + 7)>>4
+
+ ; output 1 and 3
+ ; interleave c1, d1
+ movdqa xmm1, xmm5 ; d1
+ punpcklwd xmm1, xmm4 ; c1 d1
+ punpckhwd xmm5, xmm4 ; c1 d1
+
+ movdqa xmm3, xmm1
+ movdqa xmm4, xmm5
+
+ pmaddwd xmm1, XMMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
+ pmaddwd xmm4, XMMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
+
+ pmaddwd xmm3, XMMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
+ pmaddwd xmm5, XMMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
+
+ paddd xmm1, XMMWORD PTR[GLOBAL(_12000)]
+ paddd xmm4, XMMWORD PTR[GLOBAL(_12000)]
+ paddd xmm3, XMMWORD PTR[GLOBAL(_51000)]
+ paddd xmm5, XMMWORD PTR[GLOBAL(_51000)]
+
+ psrad xmm1, 16 ; (c1 * 2217 + d1 * 5352 + 14500)>>16
+ psrad xmm4, 16 ; (c1 * 2217 + d1 * 5352 + 14500)>>16
+ psrad xmm3, 16 ; (d1 * 2217 - c1 * 5352 + 7500)>>16
+ psrad xmm5, 16 ; (d1 * 2217 - c1 * 5352 + 7500)>>16
+
+ packssdw xmm1, xmm4 ; op[4]
+ packssdw xmm3, xmm5 ; op[12]
+
+ paddw xmm1, xmm6 ; op[4] += (d1!=0)
+
+ movdqa xmm4, xmm0
+ movdqa xmm5, xmm2
+
+ punpcklqdq xmm0, xmm1
+ punpckhqdq xmm4, xmm1
+
+ punpcklqdq xmm2, xmm3
+ punpckhqdq xmm5, xmm3
+
+ movdqa XMMWORD PTR[output + 0 ], xmm0
+ movdqa XMMWORD PTR[output + 16], xmm2
+ movdqa XMMWORD PTR[output + 32], xmm4
+ movdqa XMMWORD PTR[output + 48], xmm5
+
+ STACK_FRAME_DESTROY
+
+SECTION_RODATA
+align 16
+_5352_2217:
+ dw 5352
+ dw 2217
+ dw 5352
+ dw 2217
+ dw 5352
+ dw 2217
+ dw 5352
+ dw 2217
+align 16
+_2217_neg5352:
+ dw 2217
+ dw -5352
+ dw 2217
+ dw -5352
+ dw 2217
+ dw -5352
+ dw 2217
+ dw -5352
+align 16
+_mult_add:
+ times 8 dw 1
+align 16
+_cmp_mask:
+ times 4 dw 1
+ times 4 dw 0
+align 16
+_cmp_mask8x4:
+ times 8 dw 1
+align 16
+_mult_sub:
+ dw 1
+ dw -1
+ dw 1
+ dw -1
+ dw 1
+ dw -1
+ dw 1
+ dw -1
+align 16
+_7:
+ times 4 dd 7
+align 16
+_7w:
+ times 8 dw 7
+align 16
+_14500:
+ times 4 dd 14500
+align 16
+_7500:
+ times 4 dd 7500
+align 16
+_12000:
+ times 4 dd 12000
+align 16
+_51000:
+ times 4 dd 51000
diff --git a/libvpx/vp8/encoder/x86/denoising_sse2.c b/libvpx/vp8/encoder/x86/denoising_sse2.c
new file mode 100644
index 0000000..c1ac6c1
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/denoising_sse2.c
@@ -0,0 +1,119 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "vp8/encoder/denoising.h"
+#include "vp8/common/reconinter.h"
+#include "vpx/vpx_integer.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vpx_rtcd.h"
+
+#include <emmintrin.h>
+
+union sum_union {
+ __m128i v;
+ signed char e[16];
+};
+
+int vp8_denoiser_filter_sse2(YV12_BUFFER_CONFIG *mc_running_avg,
+ YV12_BUFFER_CONFIG *running_avg,
+ MACROBLOCK *signal, unsigned int motion_magnitude,
+ int y_offset, int uv_offset)
+{
+ unsigned char *sig = signal->thismb;
+ int sig_stride = 16;
+ unsigned char *mc_running_avg_y = mc_running_avg->y_buffer + y_offset;
+ int mc_avg_y_stride = mc_running_avg->y_stride;
+ unsigned char *running_avg_y = running_avg->y_buffer + y_offset;
+ int avg_y_stride = running_avg->y_stride;
+ int r;
+ __m128i acc_diff = _mm_setzero_si128();
+ const __m128i k_0 = _mm_setzero_si128();
+ const __m128i k_4 = _mm_set1_epi8(4);
+ const __m128i k_8 = _mm_set1_epi8(8);
+ const __m128i k_16 = _mm_set1_epi8(16);
+ /* Modify each level's adjustment according to motion_magnitude. */
+ const __m128i l3 = _mm_set1_epi8(
+ (motion_magnitude <= MOTION_MAGNITUDE_THRESHOLD) ? 7 : 6);
+ /* Difference between level 3 and level 2 is 2. */
+ const __m128i l32 = _mm_set1_epi8(2);
+ /* Difference between level 2 and level 1 is 1. */
+ const __m128i l21 = _mm_set1_epi8(1);
+
+ for (r = 0; r < 16; ++r)
+ {
+ /* Calculate differences */
+ const __m128i v_sig = _mm_loadu_si128((__m128i *)(&sig[0]));
+ const __m128i v_mc_running_avg_y = _mm_loadu_si128(
+ (__m128i *)(&mc_running_avg_y[0]));
+ __m128i v_running_avg_y;
+ const __m128i pdiff = _mm_subs_epu8(v_mc_running_avg_y, v_sig);
+ const __m128i ndiff = _mm_subs_epu8(v_sig, v_mc_running_avg_y);
+ /* Obtain the sign. FF if diff is negative. */
+ const __m128i diff_sign = _mm_cmpeq_epi8(pdiff, k_0);
+ /* Clamp absolute difference to 16 to be used to get mask. Doing this
+ * allows us to use _mm_cmpgt_epi8, which operates on signed byte. */
+ const __m128i clamped_absdiff = _mm_min_epu8(
+ _mm_or_si128(pdiff, ndiff), k_16);
+ /* Get masks for l2 l1 and l0 adjustments */
+ const __m128i mask2 = _mm_cmpgt_epi8(k_16, clamped_absdiff);
+ const __m128i mask1 = _mm_cmpgt_epi8(k_8, clamped_absdiff);
+ const __m128i mask0 = _mm_cmpgt_epi8(k_4, clamped_absdiff);
+ /* Get adjustments for l2, l1, and l0 */
+ __m128i adj2 = _mm_and_si128(mask2, l32);
+ const __m128i adj1 = _mm_and_si128(mask1, l21);
+ const __m128i adj0 = _mm_and_si128(mask0, clamped_absdiff);
+ __m128i adj, padj, nadj;
+
+ /* Combine the adjustments and get absolute adjustments. */
+ adj2 = _mm_add_epi8(adj2, adj1);
+ adj = _mm_sub_epi8(l3, adj2);
+ adj = _mm_andnot_si128(mask0, adj);
+ adj = _mm_or_si128(adj, adj0);
+
+ /* Restore the sign and get positive and negative adjustments. */
+ padj = _mm_andnot_si128(diff_sign, adj);
+ nadj = _mm_and_si128(diff_sign, adj);
+
+ /* Calculate filtered value. */
+ v_running_avg_y = _mm_adds_epu8(v_sig, padj);
+ v_running_avg_y = _mm_subs_epu8(v_running_avg_y, nadj);
+ _mm_storeu_si128((__m128i *)running_avg_y, v_running_avg_y);
+
+ /* Adjustments <=7, and each element in acc_diff can fit in signed
+ * char.
+ */
+ acc_diff = _mm_adds_epi8(acc_diff, padj);
+ acc_diff = _mm_subs_epi8(acc_diff, nadj);
+
+ /* Update pointers for next iteration. */
+ sig += sig_stride;
+ mc_running_avg_y += mc_avg_y_stride;
+ running_avg_y += avg_y_stride;
+ }
+
+ {
+ /* Compute the sum of all pixel differences of this MB. */
+ union sum_union s;
+ int sum_diff = 0;
+ s.v = acc_diff;
+ sum_diff = s.e[0] + s.e[1] + s.e[2] + s.e[3] + s.e[4] + s.e[5]
+ + s.e[6] + s.e[7] + s.e[8] + s.e[9] + s.e[10] + s.e[11]
+ + s.e[12] + s.e[13] + s.e[14] + s.e[15];
+
+ if (abs(sum_diff) > SUM_DIFF_THRESHOLD)
+ {
+ return COPY_BLOCK;
+ }
+ }
+
+ vp8_copy_mem16x16(running_avg->y_buffer + y_offset, avg_y_stride,
+ signal->thismb, sig_stride);
+ return FILTER_BLOCK;
+}
diff --git a/libvpx/vp8/encoder/x86/encodeopt.asm b/libvpx/vp8/encoder/x86/encodeopt.asm
new file mode 100644
index 0000000..fe26b18
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/encodeopt.asm
@@ -0,0 +1,386 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;int vp8_block_error_xmm(short *coeff_ptr, short *dcoef_ptr)
+global sym(vp8_block_error_xmm) PRIVATE
+sym(vp8_block_error_xmm):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 2
+ push rsi
+ push rdi
+ ; end prologue
+
+ mov rsi, arg(0) ;coeff_ptr
+ mov rdi, arg(1) ;dcoef_ptr
+
+ movdqa xmm0, [rsi]
+ movdqa xmm1, [rdi]
+
+ movdqa xmm2, [rsi+16]
+ movdqa xmm3, [rdi+16]
+
+ psubw xmm0, xmm1
+ psubw xmm2, xmm3
+
+ pmaddwd xmm0, xmm0
+ pmaddwd xmm2, xmm2
+
+ paddd xmm0, xmm2
+
+ pxor xmm5, xmm5
+ movdqa xmm1, xmm0
+
+ punpckldq xmm0, xmm5
+ punpckhdq xmm1, xmm5
+
+ paddd xmm0, xmm1
+ movdqa xmm1, xmm0
+
+ psrldq xmm0, 8
+ paddd xmm0, xmm1
+
+ movq rax, xmm0
+
+ pop rdi
+ pop rsi
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;int vp8_block_error_mmx(short *coeff_ptr, short *dcoef_ptr)
+global sym(vp8_block_error_mmx) PRIVATE
+sym(vp8_block_error_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 2
+ push rsi
+ push rdi
+ ; end prolog
+
+
+ mov rsi, arg(0) ;coeff_ptr
+ pxor mm7, mm7
+
+ mov rdi, arg(1) ;dcoef_ptr
+ movq mm3, [rsi]
+
+ movq mm4, [rdi]
+ movq mm5, [rsi+8]
+
+ movq mm6, [rdi+8]
+ pxor mm1, mm1 ; from movd mm1, dc ; dc =0
+
+ movq mm2, mm7
+ psubw mm5, mm6
+
+ por mm1, mm2
+ pmaddwd mm5, mm5
+
+ pcmpeqw mm1, mm7
+ psubw mm3, mm4
+
+ pand mm1, mm3
+ pmaddwd mm1, mm1
+
+ paddd mm1, mm5
+ movq mm3, [rsi+16]
+
+ movq mm4, [rdi+16]
+ movq mm5, [rsi+24]
+
+ movq mm6, [rdi+24]
+ psubw mm5, mm6
+
+ pmaddwd mm5, mm5
+ psubw mm3, mm4
+
+ pmaddwd mm3, mm3
+ paddd mm3, mm5
+
+ paddd mm1, mm3
+ movq mm0, mm1
+
+ psrlq mm1, 32
+ paddd mm0, mm1
+
+ movq rax, mm0
+
+ pop rdi
+ pop rsi
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;int vp8_mbblock_error_mmx_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
+global sym(vp8_mbblock_error_mmx_impl) PRIVATE
+sym(vp8_mbblock_error_mmx_impl):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 3
+ push rsi
+ push rdi
+ ; end prolog
+
+
+ mov rsi, arg(0) ;coeff_ptr
+ pxor mm7, mm7
+
+ mov rdi, arg(1) ;dcoef_ptr
+ pxor mm2, mm2
+
+ movd mm1, dword ptr arg(2) ;dc
+ por mm1, mm2
+
+ pcmpeqw mm1, mm7
+ mov rcx, 16
+
+.mberror_loop_mmx:
+ movq mm3, [rsi]
+ movq mm4, [rdi]
+
+ movq mm5, [rsi+8]
+ movq mm6, [rdi+8]
+
+
+ psubw mm5, mm6
+ pmaddwd mm5, mm5
+
+ psubw mm3, mm4
+ pand mm3, mm1
+
+ pmaddwd mm3, mm3
+ paddd mm2, mm5
+
+ paddd mm2, mm3
+ movq mm3, [rsi+16]
+
+ movq mm4, [rdi+16]
+ movq mm5, [rsi+24]
+
+ movq mm6, [rdi+24]
+ psubw mm5, mm6
+
+ pmaddwd mm5, mm5
+ psubw mm3, mm4
+
+ pmaddwd mm3, mm3
+ paddd mm2, mm5
+
+ paddd mm2, mm3
+ add rsi, 32
+
+ add rdi, 32
+ sub rcx, 1
+
+ jnz .mberror_loop_mmx
+
+ movq mm0, mm2
+ psrlq mm2, 32
+
+ paddd mm0, mm2
+ movq rax, mm0
+
+ pop rdi
+ pop rsi
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;int vp8_mbblock_error_xmm_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
+global sym(vp8_mbblock_error_xmm_impl) PRIVATE
+sym(vp8_mbblock_error_xmm_impl):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 3
+ SAVE_XMM 6
+ push rsi
+ push rdi
+ ; end prolog
+
+
+ mov rsi, arg(0) ;coeff_ptr
+ pxor xmm6, xmm6
+
+ mov rdi, arg(1) ;dcoef_ptr
+ pxor xmm4, xmm4
+
+ movd xmm5, dword ptr arg(2) ;dc
+ por xmm5, xmm4
+
+ pcmpeqw xmm5, xmm6
+ mov rcx, 16
+
+.mberror_loop:
+ movdqa xmm0, [rsi]
+ movdqa xmm1, [rdi]
+
+ movdqa xmm2, [rsi+16]
+ movdqa xmm3, [rdi+16]
+
+
+ psubw xmm2, xmm3
+ pmaddwd xmm2, xmm2
+
+ psubw xmm0, xmm1
+ pand xmm0, xmm5
+
+ pmaddwd xmm0, xmm0
+ add rsi, 32
+
+ add rdi, 32
+
+ sub rcx, 1
+ paddd xmm4, xmm2
+
+ paddd xmm4, xmm0
+ jnz .mberror_loop
+
+ movdqa xmm0, xmm4
+ punpckldq xmm0, xmm6
+
+ punpckhdq xmm4, xmm6
+ paddd xmm0, xmm4
+
+ movdqa xmm1, xmm0
+ psrldq xmm0, 8
+
+ paddd xmm0, xmm1
+ movq rax, xmm0
+
+ pop rdi
+ pop rsi
+ ; begin epilog
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;int vp8_mbuverror_mmx_impl(short *s_ptr, short *d_ptr);
+global sym(vp8_mbuverror_mmx_impl) PRIVATE
+sym(vp8_mbuverror_mmx_impl):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 2
+ push rsi
+ push rdi
+ ; end prolog
+
+
+ mov rsi, arg(0) ;s_ptr
+ mov rdi, arg(1) ;d_ptr
+
+ mov rcx, 16
+ pxor mm7, mm7
+
+.mbuverror_loop_mmx:
+
+ movq mm1, [rsi]
+ movq mm2, [rdi]
+
+ psubw mm1, mm2
+ pmaddwd mm1, mm1
+
+
+ movq mm3, [rsi+8]
+ movq mm4, [rdi+8]
+
+ psubw mm3, mm4
+ pmaddwd mm3, mm3
+
+
+ paddd mm7, mm1
+ paddd mm7, mm3
+
+
+ add rsi, 16
+ add rdi, 16
+
+ dec rcx
+ jnz .mbuverror_loop_mmx
+
+ movq mm0, mm7
+ psrlq mm7, 32
+
+ paddd mm0, mm7
+ movq rax, mm0
+
+ pop rdi
+ pop rsi
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;int vp8_mbuverror_xmm_impl(short *s_ptr, short *d_ptr);
+global sym(vp8_mbuverror_xmm_impl) PRIVATE
+sym(vp8_mbuverror_xmm_impl):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 2
+ push rsi
+ push rdi
+ ; end prolog
+
+
+ mov rsi, arg(0) ;s_ptr
+ mov rdi, arg(1) ;d_ptr
+
+ mov rcx, 16
+ pxor xmm3, xmm3
+
+.mbuverror_loop:
+
+ movdqa xmm1, [rsi]
+ movdqa xmm2, [rdi]
+
+ psubw xmm1, xmm2
+ pmaddwd xmm1, xmm1
+
+ paddd xmm3, xmm1
+
+ add rsi, 16
+ add rdi, 16
+
+ dec rcx
+ jnz .mbuverror_loop
+
+ pxor xmm0, xmm0
+ movdqa xmm1, xmm3
+
+ movdqa xmm2, xmm1
+ punpckldq xmm1, xmm0
+
+ punpckhdq xmm2, xmm0
+ paddd xmm1, xmm2
+
+ movdqa xmm2, xmm1
+
+ psrldq xmm1, 8
+ paddd xmm1, xmm2
+
+ movq rax, xmm1
+
+ pop rdi
+ pop rsi
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/encoder/x86/fwalsh_sse2.asm b/libvpx/vp8/encoder/x86/fwalsh_sse2.asm
new file mode 100644
index 0000000..f498927
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/fwalsh_sse2.asm
@@ -0,0 +1,164 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void vp8_short_walsh4x4_sse2(short *input, short *output, int pitch)
+global sym(vp8_short_walsh4x4_sse2) PRIVATE
+sym(vp8_short_walsh4x4_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 3
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ; input
+ mov rdi, arg(1) ; output
+ movsxd rdx, dword ptr arg(2) ; pitch
+
+ ; first for loop
+ movq xmm0, MMWORD PTR [rsi] ; load input
+ movq xmm1, MMWORD PTR [rsi + rdx]
+ lea rsi, [rsi + rdx*2]
+ movq xmm2, MMWORD PTR [rsi]
+ movq xmm3, MMWORD PTR [rsi + rdx]
+
+ punpcklwd xmm0, xmm1
+ punpcklwd xmm2, xmm3
+
+ movdqa xmm1, xmm0
+ punpckldq xmm0, xmm2 ; ip[1] ip[0]
+ punpckhdq xmm1, xmm2 ; ip[3] ip[2]
+
+ movdqa xmm2, xmm0
+ paddw xmm0, xmm1
+ psubw xmm2, xmm1
+
+ psllw xmm0, 2 ; d1 a1
+ psllw xmm2, 2 ; c1 b1
+
+ movdqa xmm1, xmm0
+ punpcklqdq xmm0, xmm2 ; b1 a1
+ punpckhqdq xmm1, xmm2 ; c1 d1
+
+ pxor xmm6, xmm6
+ movq xmm6, xmm0
+ pxor xmm7, xmm7
+ pcmpeqw xmm7, xmm6
+ paddw xmm7, [GLOBAL(c1)]
+
+ movdqa xmm2, xmm0
+ paddw xmm0, xmm1 ; b1+c1 a1+d1
+ psubw xmm2, xmm1 ; b1-c1 a1-d1
+ paddw xmm0, xmm7 ; b1+c1 a1+d1+(a1!=0)
+
+ ; second for loop
+ ; input: 13 9 5 1 12 8 4 0 (xmm0)
+ ; 14 10 6 2 15 11 7 3 (xmm2)
+ ; after shuffle:
+ ; 13 5 9 1 12 4 8 0 (xmm0)
+ ; 14 6 10 2 15 7 11 3 (xmm1)
+ pshuflw xmm3, xmm0, 0xd8
+ pshufhw xmm0, xmm3, 0xd8
+ pshuflw xmm3, xmm2, 0xd8
+ pshufhw xmm1, xmm3, 0xd8
+
+ movdqa xmm2, xmm0
+ pmaddwd xmm0, [GLOBAL(c1)] ; d11 a11 d10 a10
+ pmaddwd xmm2, [GLOBAL(cn1)] ; c11 b11 c10 b10
+ movdqa xmm3, xmm1
+ pmaddwd xmm1, [GLOBAL(c1)] ; d12 a12 d13 a13
+ pmaddwd xmm3, [GLOBAL(cn1)] ; c12 b12 c13 b13
+
+ pshufd xmm4, xmm0, 0xd8 ; d11 d10 a11 a10
+ pshufd xmm5, xmm2, 0xd8 ; c11 c10 b11 b10
+ pshufd xmm6, xmm1, 0x72 ; d13 d12 a13 a12
+ pshufd xmm7, xmm3, 0x72 ; c13 c12 b13 b12
+
+ movdqa xmm0, xmm4
+ punpcklqdq xmm0, xmm5 ; b11 b10 a11 a10
+ punpckhqdq xmm4, xmm5 ; c11 c10 d11 d10
+ movdqa xmm1, xmm6
+ punpcklqdq xmm1, xmm7 ; b13 b12 a13 a12
+ punpckhqdq xmm6, xmm7 ; c13 c12 d13 d12
+
+ movdqa xmm2, xmm0
+ paddd xmm0, xmm4 ; b21 b20 a21 a20
+ psubd xmm2, xmm4 ; c21 c20 d21 d20
+ movdqa xmm3, xmm1
+ paddd xmm1, xmm6 ; b23 b22 a23 a22
+ psubd xmm3, xmm6 ; c23 c22 d23 d22
+
+ pxor xmm4, xmm4
+ movdqa xmm5, xmm4
+ pcmpgtd xmm4, xmm0
+ pcmpgtd xmm5, xmm2
+ pand xmm4, [GLOBAL(cd1)]
+ pand xmm5, [GLOBAL(cd1)]
+
+ pxor xmm6, xmm6
+ movdqa xmm7, xmm6
+ pcmpgtd xmm6, xmm1
+ pcmpgtd xmm7, xmm3
+ pand xmm6, [GLOBAL(cd1)]
+ pand xmm7, [GLOBAL(cd1)]
+
+ paddd xmm0, xmm4
+ paddd xmm2, xmm5
+ paddd xmm0, [GLOBAL(cd3)]
+ paddd xmm2, [GLOBAL(cd3)]
+ paddd xmm1, xmm6
+ paddd xmm3, xmm7
+ paddd xmm1, [GLOBAL(cd3)]
+ paddd xmm3, [GLOBAL(cd3)]
+
+ psrad xmm0, 3
+ psrad xmm1, 3
+ psrad xmm2, 3
+ psrad xmm3, 3
+ movdqa xmm4, xmm0
+ punpcklqdq xmm0, xmm1 ; a23 a22 a21 a20
+ punpckhqdq xmm4, xmm1 ; b23 b22 b21 b20
+ movdqa xmm5, xmm2
+ punpckhqdq xmm2, xmm3 ; c23 c22 c21 c20
+ punpcklqdq xmm5, xmm3 ; d23 d22 d21 d20
+
+ packssdw xmm0, xmm4 ; b23 b22 b21 b20 a23 a22 a21 a20
+ packssdw xmm2, xmm5 ; d23 d22 d21 d20 c23 c22 c21 c20
+
+ movdqa XMMWORD PTR [rdi], xmm0
+ movdqa XMMWORD PTR [rdi + 16], xmm2
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+c1:
+ dw 0x0001, 0x0001, 0x0001, 0x0001, 0x0001, 0x0001, 0x0001, 0x0001
+align 16
+cn1:
+ dw 0x0001, 0xffff, 0x0001, 0xffff, 0x0001, 0xffff, 0x0001, 0xffff
+align 16
+cd1:
+ dd 0x00000001, 0x00000001, 0x00000001, 0x00000001
+align 16
+cd3:
+ dd 0x00000003, 0x00000003, 0x00000003, 0x00000003
diff --git a/libvpx/vp8/encoder/x86/quantize_mmx.asm b/libvpx/vp8/encoder/x86/quantize_mmx.asm
new file mode 100644
index 0000000..2864ce1
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/quantize_mmx.asm
@@ -0,0 +1,286 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;int vp8_fast_quantize_b_impl_mmx(short *coeff_ptr, short *zbin_ptr,
+; short *qcoeff_ptr,short *dequant_ptr,
+; short *scan_mask, short *round_ptr,
+; short *quant_ptr, short *dqcoeff_ptr);
+global sym(vp8_fast_quantize_b_impl_mmx) PRIVATE
+sym(vp8_fast_quantize_b_impl_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 8
+ push rsi
+ push rdi
+ ; end prolog
+
+
+ mov rsi, arg(0) ;coeff_ptr
+ movq mm0, [rsi]
+
+ mov rax, arg(1) ;zbin_ptr
+ movq mm1, [rax]
+
+ movq mm3, mm0
+ psraw mm0, 15
+
+ pxor mm3, mm0
+ psubw mm3, mm0 ; abs
+
+ movq mm2, mm3
+ pcmpgtw mm1, mm2
+
+ pandn mm1, mm2
+ movq mm3, mm1
+
+ mov rdx, arg(6) ;quant_ptr
+ movq mm1, [rdx]
+
+ mov rcx, arg(5) ;round_ptr
+ movq mm2, [rcx]
+
+ paddw mm3, mm2
+ pmulhuw mm3, mm1
+
+ pxor mm3, mm0
+ psubw mm3, mm0 ;gain the sign back
+
+ mov rdi, arg(2) ;qcoeff_ptr
+ movq mm0, mm3
+
+ movq [rdi], mm3
+
+ mov rax, arg(3) ;dequant_ptr
+ movq mm2, [rax]
+
+ pmullw mm3, mm2
+ mov rax, arg(7) ;dqcoeff_ptr
+
+ movq [rax], mm3
+
+ ; next 8
+ movq mm4, [rsi+8]
+
+ mov rax, arg(1) ;zbin_ptr
+ movq mm5, [rax+8]
+
+ movq mm7, mm4
+ psraw mm4, 15
+
+ pxor mm7, mm4
+ psubw mm7, mm4 ; abs
+
+ movq mm6, mm7
+ pcmpgtw mm5, mm6
+
+ pandn mm5, mm6
+ movq mm7, mm5
+
+ movq mm5, [rdx+8]
+ movq mm6, [rcx+8]
+
+ paddw mm7, mm6
+ pmulhuw mm7, mm5
+
+ pxor mm7, mm4
+ psubw mm7, mm4;gain the sign back
+
+ mov rdi, arg(2) ;qcoeff_ptr
+
+ movq mm1, mm7
+ movq [rdi+8], mm7
+
+ mov rax, arg(3) ;dequant_ptr
+ movq mm6, [rax+8]
+
+ pmullw mm7, mm6
+ mov rax, arg(7) ;dqcoeff_ptr
+
+ movq [rax+8], mm7
+
+
+ ; next 8
+ movq mm4, [rsi+16]
+
+ mov rax, arg(1) ;zbin_ptr
+ movq mm5, [rax+16]
+
+ movq mm7, mm4
+ psraw mm4, 15
+
+ pxor mm7, mm4
+ psubw mm7, mm4 ; abs
+
+ movq mm6, mm7
+ pcmpgtw mm5, mm6
+
+ pandn mm5, mm6
+ movq mm7, mm5
+
+ movq mm5, [rdx+16]
+ movq mm6, [rcx+16]
+
+ paddw mm7, mm6
+ pmulhuw mm7, mm5
+
+ pxor mm7, mm4
+ psubw mm7, mm4;gain the sign back
+
+ mov rdi, arg(2) ;qcoeff_ptr
+
+ movq mm1, mm7
+ movq [rdi+16], mm7
+
+ mov rax, arg(3) ;dequant_ptr
+ movq mm6, [rax+16]
+
+ pmullw mm7, mm6
+ mov rax, arg(7) ;dqcoeff_ptr
+
+ movq [rax+16], mm7
+
+
+ ; next 8
+ movq mm4, [rsi+24]
+
+ mov rax, arg(1) ;zbin_ptr
+ movq mm5, [rax+24]
+
+ movq mm7, mm4
+ psraw mm4, 15
+
+ pxor mm7, mm4
+ psubw mm7, mm4 ; abs
+
+ movq mm6, mm7
+ pcmpgtw mm5, mm6
+
+ pandn mm5, mm6
+ movq mm7, mm5
+
+ movq mm5, [rdx+24]
+ movq mm6, [rcx+24]
+
+ paddw mm7, mm6
+ pmulhuw mm7, mm5
+
+ pxor mm7, mm4
+ psubw mm7, mm4;gain the sign back
+
+ mov rdi, arg(2) ;qcoeff_ptr
+
+ movq mm1, mm7
+ movq [rdi+24], mm7
+
+ mov rax, arg(3) ;dequant_ptr
+ movq mm6, [rax+24]
+
+ pmullw mm7, mm6
+ mov rax, arg(7) ;dqcoeff_ptr
+
+ movq [rax+24], mm7
+
+
+
+ mov rdi, arg(4) ;scan_mask
+ mov rsi, arg(2) ;qcoeff_ptr
+
+ pxor mm5, mm5
+ pxor mm7, mm7
+
+ movq mm0, [rsi]
+ movq mm1, [rsi+8]
+
+ movq mm2, [rdi]
+ movq mm3, [rdi+8];
+
+ pcmpeqw mm0, mm7
+ pcmpeqw mm1, mm7
+
+ pcmpeqw mm6, mm6
+ pxor mm0, mm6
+
+ pxor mm1, mm6
+ psrlw mm0, 15
+
+ psrlw mm1, 15
+ pmaddwd mm0, mm2
+
+ pmaddwd mm1, mm3
+ movq mm5, mm0
+
+ paddd mm5, mm1
+
+ movq mm0, [rsi+16]
+ movq mm1, [rsi+24]
+
+ movq mm2, [rdi+16]
+ movq mm3, [rdi+24];
+
+ pcmpeqw mm0, mm7
+ pcmpeqw mm1, mm7
+
+ pcmpeqw mm6, mm6
+ pxor mm0, mm6
+
+ pxor mm1, mm6
+ psrlw mm0, 15
+
+ psrlw mm1, 15
+ pmaddwd mm0, mm2
+
+ pmaddwd mm1, mm3
+ paddd mm5, mm0
+
+ paddd mm5, mm1
+ movq mm0, mm5
+
+ psrlq mm5, 32
+ paddd mm0, mm5
+
+ ; eob adjustment begins here
+ movq rcx, mm0
+ and rcx, 0xffff
+
+ xor rdx, rdx
+ sub rdx, rcx ; rdx=-rcx
+
+ bsr rax, rcx
+ inc rax
+
+ sar rdx, 31
+ and rax, rdx
+ ; Substitute the sse assembly for the old mmx mixed assembly/C. The
+ ; following is kept as reference
+ ; movq rcx, mm0
+ ; bsr rax, rcx
+ ;
+ ; mov eob, rax
+ ; mov eee, rcx
+ ;
+ ;if(eee==0)
+ ;{
+ ; eob=-1;
+ ;}
+ ;else if(eee<0)
+ ;{
+ ; eob=15;
+ ;}
+ ;d->eob = eob+1;
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/encoder/x86/quantize_sse2.asm b/libvpx/vp8/encoder/x86/quantize_sse2.asm
new file mode 100644
index 0000000..724e54c
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/quantize_sse2.asm
@@ -0,0 +1,386 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license and patent
+; grant that can be found in the LICENSE file in the root of the source
+; tree. All contributing project authors may be found in the AUTHORS
+; file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+%include "asm_enc_offsets.asm"
+
+
+; void vp8_regular_quantize_b_sse2 | arg
+; (BLOCK *b, | 0
+; BLOCKD *d) | 1
+
+global sym(vp8_regular_quantize_b_sse2) PRIVATE
+sym(vp8_regular_quantize_b_sse2):
+ push rbp
+ mov rbp, rsp
+ SAVE_XMM 7
+ GET_GOT rbx
+
+%if ABI_IS_32BIT
+ push rdi
+ push rsi
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ push rdi
+ push rsi
+ %endif
+%endif
+
+ ALIGN_STACK 16, rax
+ %define zrun_zbin_boost 0 ; 8
+ %define abs_minus_zbin 8 ; 32
+ %define temp_qcoeff 40 ; 32
+ %define qcoeff 72 ; 32
+ %define stack_size 104
+ sub rsp, stack_size
+ ; end prolog
+
+%if ABI_IS_32BIT
+ mov rdi, arg(0) ; BLOCK *b
+ mov rsi, arg(1) ; BLOCKD *d
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ mov rdi, rcx ; BLOCK *b
+ mov rsi, rdx ; BLOCKD *d
+ %else
+ ;mov rdi, rdi ; BLOCK *b
+ ;mov rsi, rsi ; BLOCKD *d
+ %endif
+%endif
+
+ mov rdx, [rdi + vp8_block_coeff] ; coeff_ptr
+ mov rcx, [rdi + vp8_block_zbin] ; zbin_ptr
+ movd xmm7, [rdi + vp8_block_zbin_extra] ; zbin_oq_value
+
+ ; z
+ movdqa xmm0, [rdx]
+ movdqa xmm4, [rdx + 16]
+ mov rdx, [rdi + vp8_block_round] ; round_ptr
+
+ pshuflw xmm7, xmm7, 0
+ punpcklwd xmm7, xmm7 ; duplicated zbin_oq_value
+
+ movdqa xmm1, xmm0
+ movdqa xmm5, xmm4
+
+ ; sz
+ psraw xmm0, 15
+ psraw xmm4, 15
+
+ ; (z ^ sz)
+ pxor xmm1, xmm0
+ pxor xmm5, xmm4
+
+ ; x = abs(z)
+ psubw xmm1, xmm0
+ psubw xmm5, xmm4
+
+ movdqa xmm2, [rcx]
+ movdqa xmm3, [rcx + 16]
+ mov rcx, [rdi + vp8_block_quant] ; quant_ptr
+
+ ; *zbin_ptr + zbin_oq_value
+ paddw xmm2, xmm7
+ paddw xmm3, xmm7
+
+ ; x - (*zbin_ptr + zbin_oq_value)
+ psubw xmm1, xmm2
+ psubw xmm5, xmm3
+ movdqa [rsp + abs_minus_zbin], xmm1
+ movdqa [rsp + abs_minus_zbin + 16], xmm5
+
+ ; add (zbin_ptr + zbin_oq_value) back
+ paddw xmm1, xmm2
+ paddw xmm5, xmm3
+
+ movdqa xmm2, [rdx]
+ movdqa xmm6, [rdx + 16]
+
+ movdqa xmm3, [rcx]
+ movdqa xmm7, [rcx + 16]
+
+ ; x + round
+ paddw xmm1, xmm2
+ paddw xmm5, xmm6
+
+ ; y = x * quant_ptr >> 16
+ pmulhw xmm3, xmm1
+ pmulhw xmm7, xmm5
+
+ ; y += x
+ paddw xmm1, xmm3
+ paddw xmm5, xmm7
+
+ movdqa [rsp + temp_qcoeff], xmm1
+ movdqa [rsp + temp_qcoeff + 16], xmm5
+
+ pxor xmm6, xmm6
+ ; zero qcoeff
+ movdqa [rsp + qcoeff], xmm6
+ movdqa [rsp + qcoeff + 16], xmm6
+
+ mov rdx, [rdi + vp8_block_zrun_zbin_boost] ; zbin_boost_ptr
+ mov rax, [rdi + vp8_block_quant_shift] ; quant_shift_ptr
+ mov [rsp + zrun_zbin_boost], rdx
+
+%macro ZIGZAG_LOOP 1
+ ; x
+ movsx ecx, WORD PTR[rsp + abs_minus_zbin + %1 * 2]
+
+ ; if (x >= zbin)
+ sub cx, WORD PTR[rdx] ; x - zbin
+ lea rdx, [rdx + 2] ; zbin_boost_ptr++
+ jl .rq_zigzag_loop_%1 ; x < zbin
+
+ movsx edi, WORD PTR[rsp + temp_qcoeff + %1 * 2]
+
+ ; downshift by quant_shift[rc]
+ movsx cx, BYTE PTR[rax + %1] ; quant_shift_ptr[rc]
+ sar edi, cl ; also sets Z bit
+ je .rq_zigzag_loop_%1 ; !y
+ mov WORD PTR[rsp + qcoeff + %1 * 2], di ;qcoeff_ptr[rc] = temp_qcoeff[rc]
+ mov rdx, [rsp + zrun_zbin_boost] ; reset to b->zrun_zbin_boost
+.rq_zigzag_loop_%1:
+%endmacro
+; in vp8_default_zig_zag1d order: see vp8/common/entropy.c
+ZIGZAG_LOOP 0
+ZIGZAG_LOOP 1
+ZIGZAG_LOOP 4
+ZIGZAG_LOOP 8
+ZIGZAG_LOOP 5
+ZIGZAG_LOOP 2
+ZIGZAG_LOOP 3
+ZIGZAG_LOOP 6
+ZIGZAG_LOOP 9
+ZIGZAG_LOOP 12
+ZIGZAG_LOOP 13
+ZIGZAG_LOOP 10
+ZIGZAG_LOOP 7
+ZIGZAG_LOOP 11
+ZIGZAG_LOOP 14
+ZIGZAG_LOOP 15
+
+ movdqa xmm2, [rsp + qcoeff]
+ movdqa xmm3, [rsp + qcoeff + 16]
+
+ mov rcx, [rsi + vp8_blockd_dequant] ; dequant_ptr
+ mov rdi, [rsi + vp8_blockd_dqcoeff] ; dqcoeff_ptr
+
+ ; y ^ sz
+ pxor xmm2, xmm0
+ pxor xmm3, xmm4
+ ; x = (y ^ sz) - sz
+ psubw xmm2, xmm0
+ psubw xmm3, xmm4
+
+ ; dequant
+ movdqa xmm0, [rcx]
+ movdqa xmm1, [rcx + 16]
+
+ mov rcx, [rsi + vp8_blockd_qcoeff] ; qcoeff_ptr
+
+ pmullw xmm0, xmm2
+ pmullw xmm1, xmm3
+
+ movdqa [rcx], xmm2 ; store qcoeff
+ movdqa [rcx + 16], xmm3
+ movdqa [rdi], xmm0 ; store dqcoeff
+ movdqa [rdi + 16], xmm1
+
+ mov rcx, [rsi + vp8_blockd_eob]
+
+ ; select the last value (in zig_zag order) for EOB
+ pcmpeqw xmm2, xmm6
+ pcmpeqw xmm3, xmm6
+ ; !
+ pcmpeqw xmm6, xmm6
+ pxor xmm2, xmm6
+ pxor xmm3, xmm6
+ ; mask inv_zig_zag
+ pand xmm2, [GLOBAL(inv_zig_zag)]
+ pand xmm3, [GLOBAL(inv_zig_zag + 16)]
+ ; select the max value
+ pmaxsw xmm2, xmm3
+ pshufd xmm3, xmm2, 00001110b
+ pmaxsw xmm2, xmm3
+ pshuflw xmm3, xmm2, 00001110b
+ pmaxsw xmm2, xmm3
+ pshuflw xmm3, xmm2, 00000001b
+ pmaxsw xmm2, xmm3
+ movd eax, xmm2
+ and eax, 0xff
+
+ mov BYTE PTR [rcx], al ; store eob
+
+ ; begin epilog
+ add rsp, stack_size
+ pop rsp
+%if ABI_IS_32BIT
+ pop rsi
+ pop rdi
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ pop rsi
+ pop rdi
+ %endif
+%endif
+ RESTORE_GOT
+ RESTORE_XMM
+ pop rbp
+ ret
+
+; void vp8_fast_quantize_b_sse2 | arg
+; (BLOCK *b, | 0
+; BLOCKD *d) | 1
+
+global sym(vp8_fast_quantize_b_sse2) PRIVATE
+sym(vp8_fast_quantize_b_sse2):
+ push rbp
+ mov rbp, rsp
+ GET_GOT rbx
+
+%if ABI_IS_32BIT
+ push rdi
+ push rsi
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ push rdi
+ push rsi
+ %else
+ ; these registers are used for passing arguments
+ %endif
+%endif
+
+ ; end prolog
+
+%if ABI_IS_32BIT
+ mov rdi, arg(0) ; BLOCK *b
+ mov rsi, arg(1) ; BLOCKD *d
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ mov rdi, rcx ; BLOCK *b
+ mov rsi, rdx ; BLOCKD *d
+ %else
+ ;mov rdi, rdi ; BLOCK *b
+ ;mov rsi, rsi ; BLOCKD *d
+ %endif
+%endif
+
+ mov rax, [rdi + vp8_block_coeff]
+ mov rcx, [rdi + vp8_block_round]
+ mov rdx, [rdi + vp8_block_quant_fast]
+
+ ; z = coeff
+ movdqa xmm0, [rax]
+ movdqa xmm4, [rax + 16]
+
+ ; dup z so we can save sz
+ movdqa xmm1, xmm0
+ movdqa xmm5, xmm4
+
+ ; sz = z >> 15
+ psraw xmm0, 15
+ psraw xmm4, 15
+
+ ; x = abs(z) = (z ^ sz) - sz
+ pxor xmm1, xmm0
+ pxor xmm5, xmm4
+ psubw xmm1, xmm0
+ psubw xmm5, xmm4
+
+ ; x += round
+ paddw xmm1, [rcx]
+ paddw xmm5, [rcx + 16]
+
+ mov rax, [rsi + vp8_blockd_qcoeff]
+ mov rcx, [rsi + vp8_blockd_dequant]
+ mov rdi, [rsi + vp8_blockd_dqcoeff]
+
+ ; y = x * quant >> 16
+ pmulhw xmm1, [rdx]
+ pmulhw xmm5, [rdx + 16]
+
+ ; x = (y ^ sz) - sz
+ pxor xmm1, xmm0
+ pxor xmm5, xmm4
+ psubw xmm1, xmm0
+ psubw xmm5, xmm4
+
+ ; qcoeff = x
+ movdqa [rax], xmm1
+ movdqa [rax + 16], xmm5
+
+ ; x * dequant
+ movdqa xmm2, xmm1
+ movdqa xmm3, xmm5
+ pmullw xmm2, [rcx]
+ pmullw xmm3, [rcx + 16]
+
+ ; dqcoeff = x * dequant
+ movdqa [rdi], xmm2
+ movdqa [rdi + 16], xmm3
+
+ pxor xmm4, xmm4 ;clear all bits
+ pcmpeqw xmm1, xmm4
+ pcmpeqw xmm5, xmm4
+
+ pcmpeqw xmm4, xmm4 ;set all bits
+ pxor xmm1, xmm4
+ pxor xmm5, xmm4
+
+ pand xmm1, [GLOBAL(inv_zig_zag)]
+ pand xmm5, [GLOBAL(inv_zig_zag + 16)]
+
+ pmaxsw xmm1, xmm5
+
+ mov rcx, [rsi + vp8_blockd_eob]
+
+ ; now down to 8
+ pshufd xmm5, xmm1, 00001110b
+
+ pmaxsw xmm1, xmm5
+
+ ; only 4 left
+ pshuflw xmm5, xmm1, 00001110b
+
+ pmaxsw xmm1, xmm5
+
+ ; okay, just 2!
+ pshuflw xmm5, xmm1, 00000001b
+
+ pmaxsw xmm1, xmm5
+
+ movd eax, xmm1
+ and eax, 0xff
+
+ mov BYTE PTR [rcx], al ; store eob
+
+ ; begin epilog
+%if ABI_IS_32BIT
+ pop rsi
+ pop rdi
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ pop rsi
+ pop rdi
+ %endif
+%endif
+
+ RESTORE_GOT
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+inv_zig_zag:
+ dw 0x0001, 0x0002, 0x0006, 0x0007
+ dw 0x0003, 0x0005, 0x0008, 0x000d
+ dw 0x0004, 0x0009, 0x000c, 0x000e
+ dw 0x000a, 0x000b, 0x000f, 0x0010
diff --git a/libvpx/vp8/encoder/x86/quantize_sse4.asm b/libvpx/vp8/encoder/x86/quantize_sse4.asm
new file mode 100644
index 0000000..f0e5d40
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/quantize_sse4.asm
@@ -0,0 +1,256 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license and patent
+; grant that can be found in the LICENSE file in the root of the source
+; tree. All contributing project authors may be found in the AUTHORS
+; file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+%include "asm_enc_offsets.asm"
+
+
+; void vp8_regular_quantize_b_sse4 | arg
+; (BLOCK *b, | 0
+; BLOCKD *d) | 1
+
+global sym(vp8_regular_quantize_b_sse4) PRIVATE
+sym(vp8_regular_quantize_b_sse4):
+
+%if ABI_IS_32BIT
+ push rbp
+ mov rbp, rsp
+ GET_GOT rbx
+ push rdi
+ push rsi
+
+ ALIGN_STACK 16, rax
+ %define qcoeff 0 ; 32
+ %define stack_size 32
+ sub rsp, stack_size
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ SAVE_XMM 8, u
+ push rdi
+ push rsi
+ %endif
+%endif
+ ; end prolog
+
+%if ABI_IS_32BIT
+ mov rdi, arg(0) ; BLOCK *b
+ mov rsi, arg(1) ; BLOCKD *d
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ mov rdi, rcx ; BLOCK *b
+ mov rsi, rdx ; BLOCKD *d
+ %else
+ ;mov rdi, rdi ; BLOCK *b
+ ;mov rsi, rsi ; BLOCKD *d
+ %endif
+%endif
+
+ mov rax, [rdi + vp8_block_coeff]
+ mov rcx, [rdi + vp8_block_zbin]
+ mov rdx, [rdi + vp8_block_round]
+ movd xmm7, [rdi + vp8_block_zbin_extra]
+
+ ; z
+ movdqa xmm0, [rax]
+ movdqa xmm1, [rax + 16]
+
+ ; duplicate zbin_oq_value
+ pshuflw xmm7, xmm7, 0
+ punpcklwd xmm7, xmm7
+
+ movdqa xmm2, xmm0
+ movdqa xmm3, xmm1
+
+ ; sz
+ psraw xmm0, 15
+ psraw xmm1, 15
+
+ ; (z ^ sz)
+ pxor xmm2, xmm0
+ pxor xmm3, xmm1
+
+ ; x = abs(z)
+ psubw xmm2, xmm0
+ psubw xmm3, xmm1
+
+ ; zbin
+ movdqa xmm4, [rcx]
+ movdqa xmm5, [rcx + 16]
+
+ ; *zbin_ptr + zbin_oq_value
+ paddw xmm4, xmm7
+ paddw xmm5, xmm7
+
+ movdqa xmm6, xmm2
+ movdqa xmm7, xmm3
+
+ ; x - (*zbin_ptr + zbin_oq_value)
+ psubw xmm6, xmm4
+ psubw xmm7, xmm5
+
+ ; round
+ movdqa xmm4, [rdx]
+ movdqa xmm5, [rdx + 16]
+
+ mov rax, [rdi + vp8_block_quant_shift]
+ mov rcx, [rdi + vp8_block_quant]
+ mov rdx, [rdi + vp8_block_zrun_zbin_boost]
+
+ ; x + round
+ paddw xmm2, xmm4
+ paddw xmm3, xmm5
+
+ ; quant
+ movdqa xmm4, [rcx]
+ movdqa xmm5, [rcx + 16]
+
+ ; y = x * quant_ptr >> 16
+ pmulhw xmm4, xmm2
+ pmulhw xmm5, xmm3
+
+ ; y += x
+ paddw xmm2, xmm4
+ paddw xmm3, xmm5
+
+ pxor xmm4, xmm4
+%if ABI_IS_32BIT
+ movdqa [rsp + qcoeff], xmm4
+ movdqa [rsp + qcoeff + 16], xmm4
+%else
+ pxor xmm8, xmm8
+%endif
+
+ ; quant_shift
+ movdqa xmm5, [rax]
+
+ ; zrun_zbin_boost
+ mov rax, rdx
+
+%macro ZIGZAG_LOOP 5
+ ; x
+ pextrw ecx, %4, %2
+
+ ; if (x >= zbin)
+ sub cx, WORD PTR[rdx] ; x - zbin
+ lea rdx, [rdx + 2] ; zbin_boost_ptr++
+ jl .rq_zigzag_loop_%1 ; x < zbin
+
+ pextrw edi, %3, %2 ; y
+
+ ; downshift by quant_shift[rc]
+ pextrb ecx, xmm5, %1 ; quant_shift[rc]
+ sar edi, cl ; also sets Z bit
+ je .rq_zigzag_loop_%1 ; !y
+%if ABI_IS_32BIT
+ mov WORD PTR[rsp + qcoeff + %1 *2], di
+%else
+ pinsrw %5, edi, %2 ; qcoeff[rc]
+%endif
+ mov rdx, rax ; reset to b->zrun_zbin_boost
+.rq_zigzag_loop_%1:
+%endmacro
+; in vp8_default_zig_zag1d order: see vp8/common/entropy.c
+ZIGZAG_LOOP 0, 0, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 1, 1, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 4, 4, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 8, 0, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 5, 5, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 2, 2, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 3, 3, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 6, 6, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 9, 1, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 12, 4, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 13, 5, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 10, 2, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 7, 7, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 11, 3, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 14, 6, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 15, 7, xmm3, xmm7, xmm8
+
+ mov rcx, [rsi + vp8_blockd_dequant]
+ mov rdi, [rsi + vp8_blockd_dqcoeff]
+
+%if ABI_IS_32BIT
+ movdqa xmm4, [rsp + qcoeff]
+ movdqa xmm5, [rsp + qcoeff + 16]
+%else
+ %define xmm5 xmm8
+%endif
+
+ ; y ^ sz
+ pxor xmm4, xmm0
+ pxor xmm5, xmm1
+ ; x = (y ^ sz) - sz
+ psubw xmm4, xmm0
+ psubw xmm5, xmm1
+
+ ; dequant
+ movdqa xmm0, [rcx]
+ movdqa xmm1, [rcx + 16]
+
+ mov rcx, [rsi + vp8_blockd_qcoeff]
+
+ pmullw xmm0, xmm4
+ pmullw xmm1, xmm5
+
+ ; store qcoeff
+ movdqa [rcx], xmm4
+ movdqa [rcx + 16], xmm5
+
+ ; store dqcoeff
+ movdqa [rdi], xmm0
+ movdqa [rdi + 16], xmm1
+
+ mov rcx, [rsi + vp8_blockd_eob]
+
+ ; select the last value (in zig_zag order) for EOB
+ pxor xmm6, xmm6
+ pcmpeqw xmm4, xmm6
+ pcmpeqw xmm5, xmm6
+
+ packsswb xmm4, xmm5
+ pshufb xmm4, [GLOBAL(zig_zag1d)]
+ pmovmskb edx, xmm4
+ xor rdi, rdi
+ mov eax, -1
+ xor dx, ax
+ bsr eax, edx
+ sub edi, edx
+ sar edi, 31
+ add eax, 1
+ and eax, edi
+
+ mov BYTE PTR [rcx], al ; store eob
+
+ ; begin epilog
+%if ABI_IS_32BIT
+ add rsp, stack_size
+ pop rsp
+
+ pop rsi
+ pop rdi
+ RESTORE_GOT
+ pop rbp
+%else
+ %undef xmm5
+ %ifidn __OUTPUT_FORMAT__,x64
+ pop rsi
+ pop rdi
+ RESTORE_XMM
+ %endif
+%endif
+
+ ret
+
+SECTION_RODATA
+align 16
+; vp8/common/entropy.c: vp8_default_zig_zag1d
+zig_zag1d:
+ db 0, 1, 4, 8, 5, 2, 3, 6, 9, 12, 13, 10, 7, 11, 14, 15
diff --git a/libvpx/vp8/encoder/x86/quantize_ssse3.asm b/libvpx/vp8/encoder/x86/quantize_ssse3.asm
new file mode 100644
index 0000000..dd526f4
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/quantize_ssse3.asm
@@ -0,0 +1,138 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license and patent
+; grant that can be found in the LICENSE file in the root of the source
+; tree. All contributing project authors may be found in the AUTHORS
+; file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+%include "asm_enc_offsets.asm"
+
+
+; void vp8_fast_quantize_b_ssse3 | arg
+; (BLOCK *b, | 0
+; BLOCKD *d) | 1
+;
+
+global sym(vp8_fast_quantize_b_ssse3) PRIVATE
+sym(vp8_fast_quantize_b_ssse3):
+ push rbp
+ mov rbp, rsp
+ GET_GOT rbx
+
+%if ABI_IS_32BIT
+ push rdi
+ push rsi
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ push rdi
+ push rsi
+ %endif
+%endif
+ ; end prolog
+
+%if ABI_IS_32BIT
+ mov rdi, arg(0) ; BLOCK *b
+ mov rsi, arg(1) ; BLOCKD *d
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ mov rdi, rcx ; BLOCK *b
+ mov rsi, rdx ; BLOCKD *d
+ %else
+ ;mov rdi, rdi ; BLOCK *b
+ ;mov rsi, rsi ; BLOCKD *d
+ %endif
+%endif
+
+ mov rax, [rdi + vp8_block_coeff]
+ mov rcx, [rdi + vp8_block_round]
+ mov rdx, [rdi + vp8_block_quant_fast]
+
+ ; coeff
+ movdqa xmm0, [rax]
+ movdqa xmm4, [rax + 16]
+
+ ; round
+ movdqa xmm2, [rcx]
+ movdqa xmm3, [rcx + 16]
+
+ movdqa xmm1, xmm0
+ movdqa xmm5, xmm4
+
+ ; sz = z >> 15
+ psraw xmm0, 15
+ psraw xmm4, 15
+
+ pabsw xmm1, xmm1
+ pabsw xmm5, xmm5
+
+ paddw xmm1, xmm2
+ paddw xmm5, xmm3
+
+ ; quant_fast
+ pmulhw xmm1, [rdx]
+ pmulhw xmm5, [rdx + 16]
+
+ mov rax, [rsi + vp8_blockd_qcoeff]
+ mov rdi, [rsi + vp8_blockd_dequant]
+ mov rcx, [rsi + vp8_blockd_dqcoeff]
+
+ movdqa xmm2, xmm1 ;store y for getting eob
+ movdqa xmm3, xmm5
+
+ pxor xmm1, xmm0
+ pxor xmm5, xmm4
+ psubw xmm1, xmm0
+ psubw xmm5, xmm4
+
+ movdqa [rax], xmm1
+ movdqa [rax + 16], xmm5
+
+ movdqa xmm0, [rdi]
+ movdqa xmm4, [rdi + 16]
+
+ pmullw xmm0, xmm1
+ pmullw xmm4, xmm5
+ pxor xmm1, xmm1
+
+ pcmpgtw xmm2, xmm1 ;calculate eob
+ pcmpgtw xmm3, xmm1
+ packsswb xmm2, xmm3
+ pshufb xmm2, [GLOBAL(zz_shuf)]
+
+ pmovmskb edx, xmm2
+
+ movdqa [rcx], xmm0 ;store dqcoeff
+ movdqa [rcx + 16], xmm4 ;store dqcoeff
+ mov rcx, [rsi + vp8_blockd_eob]
+
+ bsr eax, edx ;count 0
+ add eax, 1
+
+ cmp edx, 0 ;if all 0, eob=0
+ cmove eax, edx
+
+ mov BYTE PTR [rcx], al ;store eob
+
+ ; begin epilog
+%if ABI_IS_32BIT
+ pop rsi
+ pop rdi
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ pop rsi
+ pop rdi
+ %endif
+%endif
+
+ RESTORE_GOT
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+zz_shuf:
+ db 0, 1, 4, 8, 5, 2, 3, 6, 9, 12, 13, 10, 7, 11, 14, 15
diff --git a/libvpx/vp8/encoder/x86/ssim_opt.asm b/libvpx/vp8/encoder/x86/ssim_opt.asm
new file mode 100644
index 0000000..5964a85
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/ssim_opt.asm
@@ -0,0 +1,216 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+%include "vpx_ports/x86_abi_support.asm"
+
+; tabulate_ssim - sums sum_s,sum_r,sum_sq_s,sum_sq_r, sum_sxr
+%macro TABULATE_SSIM 0
+ paddusw xmm15, xmm3 ; sum_s
+ paddusw xmm14, xmm4 ; sum_r
+ movdqa xmm1, xmm3
+ pmaddwd xmm1, xmm1
+ paddd xmm13, xmm1 ; sum_sq_s
+ movdqa xmm2, xmm4
+ pmaddwd xmm2, xmm2
+ paddd xmm12, xmm2 ; sum_sq_r
+ pmaddwd xmm3, xmm4
+ paddd xmm11, xmm3 ; sum_sxr
+%endmacro
+
+; Sum across the register %1 starting with q words
+%macro SUM_ACROSS_Q 1
+ movdqa xmm2,%1
+ punpckldq %1,xmm0
+ punpckhdq xmm2,xmm0
+ paddq %1,xmm2
+ movdqa xmm2,%1
+ punpcklqdq %1,xmm0
+ punpckhqdq xmm2,xmm0
+ paddq %1,xmm2
+%endmacro
+
+; Sum across the register %1 starting with q words
+%macro SUM_ACROSS_W 1
+ movdqa xmm1, %1
+ punpcklwd %1,xmm0
+ punpckhwd xmm1,xmm0
+ paddd %1, xmm1
+ SUM_ACROSS_Q %1
+%endmacro
+;void ssim_parms_sse2(
+; unsigned char *s,
+; int sp,
+; unsigned char *r,
+; int rp
+; unsigned long *sum_s,
+; unsigned long *sum_r,
+; unsigned long *sum_sq_s,
+; unsigned long *sum_sq_r,
+; unsigned long *sum_sxr);
+;
+; TODO: Use parm passing through structure, probably don't need the pxors
+; ( calling app will initialize to 0 ) could easily fit everything in sse2
+; without too much hastle, and can probably do better estimates with psadw
+; or pavgb At this point this is just meant to be first pass for calculating
+; all the parms needed for 16x16 ssim so we can play with dssim as distortion
+; in mode selection code.
+global sym(vp8_ssim_parms_16x16_sse2) PRIVATE
+sym(vp8_ssim_parms_16x16_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 9
+ SAVE_XMM 15
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;s
+ mov rcx, arg(1) ;sp
+ mov rdi, arg(2) ;r
+ mov rax, arg(3) ;rp
+
+ pxor xmm0, xmm0
+ pxor xmm15,xmm15 ;sum_s
+ pxor xmm14,xmm14 ;sum_r
+ pxor xmm13,xmm13 ;sum_sq_s
+ pxor xmm12,xmm12 ;sum_sq_r
+ pxor xmm11,xmm11 ;sum_sxr
+
+ mov rdx, 16 ;row counter
+.NextRow:
+
+ ;grab source and reference pixels
+ movdqu xmm5, [rsi]
+ movdqu xmm6, [rdi]
+ movdqa xmm3, xmm5
+ movdqa xmm4, xmm6
+ punpckhbw xmm3, xmm0 ; high_s
+ punpckhbw xmm4, xmm0 ; high_r
+
+ TABULATE_SSIM
+
+ movdqa xmm3, xmm5
+ movdqa xmm4, xmm6
+ punpcklbw xmm3, xmm0 ; low_s
+ punpcklbw xmm4, xmm0 ; low_r
+
+ TABULATE_SSIM
+
+ add rsi, rcx ; next s row
+ add rdi, rax ; next r row
+
+ dec rdx ; counter
+ jnz .NextRow
+
+ SUM_ACROSS_W xmm15
+ SUM_ACROSS_W xmm14
+ SUM_ACROSS_Q xmm13
+ SUM_ACROSS_Q xmm12
+ SUM_ACROSS_Q xmm11
+
+ mov rdi,arg(4)
+ movd [rdi], xmm15;
+ mov rdi,arg(5)
+ movd [rdi], xmm14;
+ mov rdi,arg(6)
+ movd [rdi], xmm13;
+ mov rdi,arg(7)
+ movd [rdi], xmm12;
+ mov rdi,arg(8)
+ movd [rdi], xmm11;
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void ssim_parms_sse2(
+; unsigned char *s,
+; int sp,
+; unsigned char *r,
+; int rp
+; unsigned long *sum_s,
+; unsigned long *sum_r,
+; unsigned long *sum_sq_s,
+; unsigned long *sum_sq_r,
+; unsigned long *sum_sxr);
+;
+; TODO: Use parm passing through structure, probably don't need the pxors
+; ( calling app will initialize to 0 ) could easily fit everything in sse2
+; without too much hastle, and can probably do better estimates with psadw
+; or pavgb At this point this is just meant to be first pass for calculating
+; all the parms needed for 16x16 ssim so we can play with dssim as distortion
+; in mode selection code.
+global sym(vp8_ssim_parms_8x8_sse2) PRIVATE
+sym(vp8_ssim_parms_8x8_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 9
+ SAVE_XMM 15
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rsi, arg(0) ;s
+ mov rcx, arg(1) ;sp
+ mov rdi, arg(2) ;r
+ mov rax, arg(3) ;rp
+
+ pxor xmm0, xmm0
+ pxor xmm15,xmm15 ;sum_s
+ pxor xmm14,xmm14 ;sum_r
+ pxor xmm13,xmm13 ;sum_sq_s
+ pxor xmm12,xmm12 ;sum_sq_r
+ pxor xmm11,xmm11 ;sum_sxr
+
+ mov rdx, 8 ;row counter
+.NextRow:
+
+ ;grab source and reference pixels
+ movq xmm3, [rsi]
+ movq xmm4, [rdi]
+ punpcklbw xmm3, xmm0 ; low_s
+ punpcklbw xmm4, xmm0 ; low_r
+
+ TABULATE_SSIM
+
+ add rsi, rcx ; next s row
+ add rdi, rax ; next r row
+
+ dec rdx ; counter
+ jnz .NextRow
+
+ SUM_ACROSS_W xmm15
+ SUM_ACROSS_W xmm14
+ SUM_ACROSS_Q xmm13
+ SUM_ACROSS_Q xmm12
+ SUM_ACROSS_Q xmm11
+
+ mov rdi,arg(4)
+ movd [rdi], xmm15;
+ mov rdi,arg(5)
+ movd [rdi], xmm14;
+ mov rdi,arg(6)
+ movd [rdi], xmm13;
+ mov rdi,arg(7)
+ movd [rdi], xmm12;
+ mov rdi,arg(8)
+ movd [rdi], xmm11;
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/encoder/x86/subtract_mmx.asm b/libvpx/vp8/encoder/x86/subtract_mmx.asm
new file mode 100644
index 0000000..794dd22
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/subtract_mmx.asm
@@ -0,0 +1,223 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void vp8_subtract_b_mmx_impl(unsigned char *z, int src_stride,
+; short *diff, unsigned char *Predictor,
+; int pitch);
+global sym(vp8_subtract_b_mmx_impl) PRIVATE
+sym(vp8_subtract_b_mmx_impl):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+
+ mov rdi, arg(2) ;diff
+ mov rax, arg(3) ;Predictor
+ mov rsi, arg(0) ;z
+ movsxd rdx, dword ptr arg(1);src_stride;
+ movsxd rcx, dword ptr arg(4);pitch
+ pxor mm7, mm7
+
+ movd mm0, [rsi]
+ movd mm1, [rax]
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ psubw mm0, mm1
+ movq [rdi], mm0
+
+
+ movd mm0, [rsi+rdx]
+ movd mm1, [rax+rcx]
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ psubw mm0, mm1
+ movq [rdi+rcx*2],mm0
+
+
+ movd mm0, [rsi+rdx*2]
+ movd mm1, [rax+rcx*2]
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ psubw mm0, mm1
+ movq [rdi+rcx*4], mm0
+
+ lea rsi, [rsi+rdx*2]
+ lea rcx, [rcx+rcx*2]
+
+
+
+ movd mm0, [rsi+rdx]
+ movd mm1, [rax+rcx]
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ psubw mm0, mm1
+ movq [rdi+rcx*2], mm0
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_subtract_mby_mmx(short *diff, unsigned char *src, int src_stride,
+;unsigned char *pred, int pred_stride)
+global sym(vp8_subtract_mby_mmx) PRIVATE
+sym(vp8_subtract_mby_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdi, arg(0) ;diff
+ mov rsi, arg(1) ;src
+ movsxd rdx, dword ptr arg(2);src_stride
+ mov rax, arg(3) ;pred
+ push rbx
+ movsxd rbx, dword ptr arg(4);pred_stride
+
+ pxor mm0, mm0
+ mov rcx, 16
+
+
+.submby_loop:
+ movq mm1, [rsi]
+ movq mm3, [rax]
+
+ movq mm2, mm1
+ movq mm4, mm3
+
+ punpcklbw mm1, mm0
+ punpcklbw mm3, mm0
+
+ punpckhbw mm2, mm0
+ punpckhbw mm4, mm0
+
+ psubw mm1, mm3
+ psubw mm2, mm4
+
+ movq [rdi], mm1
+ movq [rdi+8], mm2
+
+ movq mm1, [rsi+8]
+ movq mm3, [rax+8]
+
+ movq mm2, mm1
+ movq mm4, mm3
+
+ punpcklbw mm1, mm0
+ punpcklbw mm3, mm0
+
+ punpckhbw mm2, mm0
+ punpckhbw mm4, mm0
+
+ psubw mm1, mm3
+ psubw mm2, mm4
+
+ movq [rdi+16], mm1
+ movq [rdi+24], mm2
+ add rdi, 32
+ lea rax, [rax+rbx]
+ lea rsi, [rsi+rdx]
+ dec rcx
+ jnz .submby_loop
+
+ pop rbx
+ pop rdi
+ pop rsi
+ ; begin epilog
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;vp8_subtract_mbuv_mmx(short *diff, unsigned char *usrc, unsigned char *vsrc,
+; int src_stride, unsigned char *upred,
+; unsigned char *vpred, int pred_stride)
+
+global sym(vp8_subtract_mbuv_mmx) PRIVATE
+sym(vp8_subtract_mbuv_mmx):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdi, arg(0) ;diff
+ mov rsi, arg(1) ;usrc
+ movsxd rdx, dword ptr arg(3);src_stride;
+ mov rax, arg(4) ;upred
+ add rdi, 256*2 ;diff = diff + 256 (shorts)
+ mov rcx, 8
+ push rbx
+ movsxd rbx, dword ptr arg(6);pred_stride
+
+ pxor mm7, mm7
+
+.submbu_loop:
+ movq mm0, [rsi]
+ movq mm1, [rax]
+ movq mm3, mm0
+ movq mm4, mm1
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ punpckhbw mm3, mm7
+ punpckhbw mm4, mm7
+ psubw mm0, mm1
+ psubw mm3, mm4
+ movq [rdi], mm0
+ movq [rdi+8], mm3
+ add rdi, 16
+ add rsi, rdx
+ add rax, rbx
+
+ dec rcx
+ jnz .submbu_loop
+
+ mov rsi, arg(2) ;vsrc
+ mov rax, arg(5) ;vpred
+ mov rcx, 8
+
+.submbv_loop:
+ movq mm0, [rsi]
+ movq mm1, [rax]
+ movq mm3, mm0
+ movq mm4, mm1
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ punpckhbw mm3, mm7
+ punpckhbw mm4, mm7
+ psubw mm0, mm1
+ psubw mm3, mm4
+ movq [rdi], mm0
+ movq [rdi+8], mm3
+ add rdi, 16
+ add rsi, rdx
+ add rax, rbx
+
+ dec rcx
+ jnz .submbv_loop
+
+ pop rbx
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
diff --git a/libvpx/vp8/encoder/x86/subtract_sse2.asm b/libvpx/vp8/encoder/x86/subtract_sse2.asm
new file mode 100644
index 0000000..a5d17f5
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/subtract_sse2.asm
@@ -0,0 +1,245 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+;void vp8_subtract_b_sse2_impl(unsigned char *z, int src_stride,
+; short *diff, unsigned char *Predictor,
+; int pitch);
+global sym(vp8_subtract_b_sse2_impl) PRIVATE
+sym(vp8_subtract_b_sse2_impl):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdi, arg(2) ;diff
+ mov rax, arg(3) ;Predictor
+ mov rsi, arg(0) ;z
+ movsxd rdx, dword ptr arg(1);src_stride;
+ movsxd rcx, dword ptr arg(4);pitch
+ pxor mm7, mm7
+
+ movd mm0, [rsi]
+ movd mm1, [rax]
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ psubw mm0, mm1
+ movq MMWORD PTR [rdi], mm0
+
+ movd mm0, [rsi+rdx]
+ movd mm1, [rax+rcx]
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ psubw mm0, mm1
+ movq MMWORD PTR [rdi+rcx*2], mm0
+
+ movd mm0, [rsi+rdx*2]
+ movd mm1, [rax+rcx*2]
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ psubw mm0, mm1
+ movq MMWORD PTR [rdi+rcx*4], mm0
+
+ lea rsi, [rsi+rdx*2]
+ lea rcx, [rcx+rcx*2]
+
+ movd mm0, [rsi+rdx]
+ movd mm1, [rax+rcx]
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+ psubw mm0, mm1
+ movq MMWORD PTR [rdi+rcx*2], mm0
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+
+;void vp8_subtract_mby_sse2(short *diff, unsigned char *src, int src_stride,
+;unsigned char *pred, int pred_stride)
+global sym(vp8_subtract_mby_sse2) PRIVATE
+sym(vp8_subtract_mby_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 5
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ mov rdi, arg(0) ;diff
+ mov rsi, arg(1) ;src
+ movsxd rdx, dword ptr arg(2);src_stride
+ mov rax, arg(3) ;pred
+ movdqa xmm4, [GLOBAL(t80)]
+ push rbx
+ mov rcx, 8 ; do two lines at one time
+ movsxd rbx, dword ptr arg(4);pred_stride
+
+.submby_loop:
+ movdqa xmm0, [rsi] ; src
+ movdqa xmm1, [rax] ; pred
+
+ movdqa xmm2, xmm0
+ psubb xmm0, xmm1
+
+ pxor xmm1, xmm4 ;convert to signed values
+ pxor xmm2, xmm4
+ pcmpgtb xmm1, xmm2 ; obtain sign information
+
+ movdqa xmm2, xmm0
+ punpcklbw xmm0, xmm1 ; put sign back to subtraction
+ punpckhbw xmm2, xmm1 ; put sign back to subtraction
+
+ movdqa xmm3, [rsi + rdx]
+ movdqa xmm5, [rax + rbx]
+
+ lea rsi, [rsi+rdx*2]
+ lea rax, [rax+rbx*2]
+
+ movdqa [rdi], xmm0
+ movdqa [rdi +16], xmm2
+
+ movdqa xmm1, xmm3
+ psubb xmm3, xmm5
+
+ pxor xmm5, xmm4 ;convert to signed values
+ pxor xmm1, xmm4
+ pcmpgtb xmm5, xmm1 ; obtain sign information
+
+ movdqa xmm1, xmm3
+ punpcklbw xmm3, xmm5 ; put sign back to subtraction
+ punpckhbw xmm1, xmm5 ; put sign back to subtraction
+
+ movdqa [rdi +32], xmm3
+ movdqa [rdi +48], xmm1
+
+ add rdi, 64
+ dec rcx
+ jnz .submby_loop
+
+ pop rbx
+ pop rdi
+ pop rsi
+ ; begin epilog
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;vp8_subtract_mbuv_sse2(short *diff, unsigned char *usrc, unsigned char *vsrc,
+; int src_stride, unsigned char *upred,
+; unsigned char *vpred, int pred_stride)
+global sym(vp8_subtract_mbuv_sse2) PRIVATE
+sym(vp8_subtract_mbuv_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ; end prolog
+
+ movdqa xmm4, [GLOBAL(t80)]
+ mov rdi, arg(0) ;diff
+ mov rsi, arg(1) ;usrc
+ movsxd rdx, dword ptr arg(3);src_stride;
+ mov rax, arg(4) ;upred
+ add rdi, 256*2 ;diff = diff + 256 (shorts)
+ mov rcx, 4
+ push rbx
+ movsxd rbx, dword ptr arg(6);pred_stride
+
+ ;u
+.submbu_loop:
+ movq xmm0, [rsi] ; src
+ movq xmm2, [rsi+rdx] ; src -- next line
+ movq xmm1, [rax] ; pred
+ movq xmm3, [rax+rbx] ; pred -- next line
+ lea rsi, [rsi + rdx*2]
+ lea rax, [rax + rbx*2]
+
+ punpcklqdq xmm0, xmm2
+ punpcklqdq xmm1, xmm3
+
+ movdqa xmm2, xmm0
+ psubb xmm0, xmm1 ; subtraction with sign missed
+
+ pxor xmm1, xmm4 ;convert to signed values
+ pxor xmm2, xmm4
+ pcmpgtb xmm1, xmm2 ; obtain sign information
+
+ movdqa xmm2, xmm0
+ movdqa xmm3, xmm1
+ punpcklbw xmm0, xmm1 ; put sign back to subtraction
+ punpckhbw xmm2, xmm3 ; put sign back to subtraction
+
+ movdqa [rdi], xmm0 ; store difference
+ movdqa [rdi +16], xmm2 ; store difference
+ add rdi, 32
+ sub rcx, 1
+ jnz .submbu_loop
+
+ mov rsi, arg(2) ;vsrc
+ mov rax, arg(5) ;vpred
+ mov rcx, 4
+
+ ;v
+.submbv_loop:
+ movq xmm0, [rsi] ; src
+ movq xmm2, [rsi+rdx] ; src -- next line
+ movq xmm1, [rax] ; pred
+ movq xmm3, [rax+rbx] ; pred -- next line
+ lea rsi, [rsi + rdx*2]
+ lea rax, [rax + rbx*2]
+
+ punpcklqdq xmm0, xmm2
+ punpcklqdq xmm1, xmm3
+
+ movdqa xmm2, xmm0
+ psubb xmm0, xmm1 ; subtraction with sign missed
+
+ pxor xmm1, xmm4 ;convert to signed values
+ pxor xmm2, xmm4
+ pcmpgtb xmm1, xmm2 ; obtain sign information
+
+ movdqa xmm2, xmm0
+ movdqa xmm3, xmm1
+ punpcklbw xmm0, xmm1 ; put sign back to subtraction
+ punpckhbw xmm2, xmm3 ; put sign back to subtraction
+
+ movdqa [rdi], xmm0 ; store difference
+ movdqa [rdi +16], xmm2 ; store difference
+ add rdi, 32
+ sub rcx, 1
+ jnz .submbv_loop
+
+ pop rbx
+ ; begin epilog
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+t80:
+ times 16 db 0x80
diff --git a/libvpx/vp8/encoder/x86/temporal_filter_apply_sse2.asm b/libvpx/vp8/encoder/x86/temporal_filter_apply_sse2.asm
new file mode 100644
index 0000000..ce9d983
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/temporal_filter_apply_sse2.asm
@@ -0,0 +1,207 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+; void vp8_temporal_filter_apply_sse2 | arg
+; (unsigned char *frame1, | 0
+; unsigned int stride, | 1
+; unsigned char *frame2, | 2
+; unsigned int block_size, | 3
+; int strength, | 4
+; int filter_weight, | 5
+; unsigned int *accumulator, | 6
+; unsigned short *count) | 7
+global sym(vp8_temporal_filter_apply_sse2) PRIVATE
+sym(vp8_temporal_filter_apply_sse2):
+
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 8
+ SAVE_XMM 7
+ GET_GOT rbx
+ push rsi
+ push rdi
+ ALIGN_STACK 16, rax
+ %define block_size 0
+ %define strength 16
+ %define filter_weight 32
+ %define rounding_bit 48
+ %define rbp_backup 64
+ %define stack_size 80
+ sub rsp, stack_size
+ mov [rsp + rbp_backup], rbp
+ ; end prolog
+
+ mov rdx, arg(3)
+ mov [rsp + block_size], rdx
+ movd xmm6, arg(4)
+ movdqa [rsp + strength], xmm6 ; where strength is used, all 16 bytes are read
+
+ ; calculate the rounding bit outside the loop
+ ; 0x8000 >> (16 - strength)
+ mov rdx, 16
+ sub rdx, arg(4) ; 16 - strength
+ movd xmm4, rdx ; can't use rdx w/ shift
+ movdqa xmm5, [GLOBAL(_const_top_bit)]
+ psrlw xmm5, xmm4
+ movdqa [rsp + rounding_bit], xmm5
+
+ mov rsi, arg(0) ; src/frame1
+ mov rdx, arg(2) ; predictor frame
+ mov rdi, arg(6) ; accumulator
+ mov rax, arg(7) ; count
+
+ ; dup the filter weight and store for later
+ movd xmm0, arg(5) ; filter_weight
+ pshuflw xmm0, xmm0, 0
+ punpcklwd xmm0, xmm0
+ movdqa [rsp + filter_weight], xmm0
+
+ mov rbp, arg(1) ; stride
+ pxor xmm7, xmm7 ; zero for extraction
+
+ lea rcx, [rdx + 16*16*1]
+ cmp dword ptr [rsp + block_size], 8
+ jne .temporal_filter_apply_load_16
+ lea rcx, [rdx + 8*8*1]
+
+.temporal_filter_apply_load_8:
+ movq xmm0, [rsi] ; first row
+ lea rsi, [rsi + rbp] ; += stride
+ punpcklbw xmm0, xmm7 ; src[ 0- 7]
+ movq xmm1, [rsi] ; second row
+ lea rsi, [rsi + rbp] ; += stride
+ punpcklbw xmm1, xmm7 ; src[ 8-15]
+ jmp .temporal_filter_apply_load_finished
+
+.temporal_filter_apply_load_16:
+ movdqa xmm0, [rsi] ; src (frame1)
+ lea rsi, [rsi + rbp] ; += stride
+ movdqa xmm1, xmm0
+ punpcklbw xmm0, xmm7 ; src[ 0- 7]
+ punpckhbw xmm1, xmm7 ; src[ 8-15]
+
+.temporal_filter_apply_load_finished:
+ movdqa xmm2, [rdx] ; predictor (frame2)
+ movdqa xmm3, xmm2
+ punpcklbw xmm2, xmm7 ; pred[ 0- 7]
+ punpckhbw xmm3, xmm7 ; pred[ 8-15]
+
+ ; modifier = src_byte - pixel_value
+ psubw xmm0, xmm2 ; src - pred[ 0- 7]
+ psubw xmm1, xmm3 ; src - pred[ 8-15]
+
+ ; modifier *= modifier
+ pmullw xmm0, xmm0 ; modifer[ 0- 7]^2
+ pmullw xmm1, xmm1 ; modifer[ 8-15]^2
+
+ ; modifier *= 3
+ pmullw xmm0, [GLOBAL(_const_3w)]
+ pmullw xmm1, [GLOBAL(_const_3w)]
+
+ ; modifer += 0x8000 >> (16 - strength)
+ paddw xmm0, [rsp + rounding_bit]
+ paddw xmm1, [rsp + rounding_bit]
+
+ ; modifier >>= strength
+ psrlw xmm0, [rsp + strength]
+ psrlw xmm1, [rsp + strength]
+
+ ; modifier = 16 - modifier
+ ; saturation takes care of modifier > 16
+ movdqa xmm3, [GLOBAL(_const_16w)]
+ movdqa xmm2, [GLOBAL(_const_16w)]
+ psubusw xmm3, xmm1
+ psubusw xmm2, xmm0
+
+ ; modifier *= filter_weight
+ pmullw xmm2, [rsp + filter_weight]
+ pmullw xmm3, [rsp + filter_weight]
+
+ ; count
+ movdqa xmm4, [rax]
+ movdqa xmm5, [rax+16]
+ ; += modifier
+ paddw xmm4, xmm2
+ paddw xmm5, xmm3
+ ; write back
+ movdqa [rax], xmm4
+ movdqa [rax+16], xmm5
+ lea rax, [rax + 16*2] ; count += 16*(sizeof(short))
+
+ ; load and extract the predictor up to shorts
+ pxor xmm7, xmm7
+ movdqa xmm0, [rdx]
+ lea rdx, [rdx + 16*1] ; pred += 16*(sizeof(char))
+ movdqa xmm1, xmm0
+ punpcklbw xmm0, xmm7 ; pred[ 0- 7]
+ punpckhbw xmm1, xmm7 ; pred[ 8-15]
+
+ ; modifier *= pixel_value
+ pmullw xmm0, xmm2
+ pmullw xmm1, xmm3
+
+ ; expand to double words
+ movdqa xmm2, xmm0
+ punpcklwd xmm0, xmm7 ; [ 0- 3]
+ punpckhwd xmm2, xmm7 ; [ 4- 7]
+ movdqa xmm3, xmm1
+ punpcklwd xmm1, xmm7 ; [ 8-11]
+ punpckhwd xmm3, xmm7 ; [12-15]
+
+ ; accumulator
+ movdqa xmm4, [rdi]
+ movdqa xmm5, [rdi+16]
+ movdqa xmm6, [rdi+32]
+ movdqa xmm7, [rdi+48]
+ ; += modifier
+ paddd xmm4, xmm0
+ paddd xmm5, xmm2
+ paddd xmm6, xmm1
+ paddd xmm7, xmm3
+ ; write back
+ movdqa [rdi], xmm4
+ movdqa [rdi+16], xmm5
+ movdqa [rdi+32], xmm6
+ movdqa [rdi+48], xmm7
+ lea rdi, [rdi + 16*4] ; accumulator += 16*(sizeof(int))
+
+ cmp rdx, rcx
+ je .temporal_filter_apply_epilog
+ pxor xmm7, xmm7 ; zero for extraction
+ cmp dword ptr [rsp + block_size], 16
+ je .temporal_filter_apply_load_16
+ jmp .temporal_filter_apply_load_8
+
+.temporal_filter_apply_epilog:
+ ; begin epilog
+ mov rbp, [rsp + rbp_backup]
+ add rsp, stack_size
+ pop rsp
+ pop rdi
+ pop rsi
+ RESTORE_GOT
+ RESTORE_XMM
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+SECTION_RODATA
+align 16
+_const_3w:
+ times 8 dw 3
+align 16
+_const_top_bit:
+ times 8 dw 1<<15
+align 16
+_const_16w
+ times 8 dw 16
diff --git a/libvpx/vp8/encoder/x86/vp8_enc_stubs_mmx.c b/libvpx/vp8/encoder/x86/vp8_enc_stubs_mmx.c
new file mode 100644
index 0000000..da25f52
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/vp8_enc_stubs_mmx.c
@@ -0,0 +1,78 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx_ports/x86.h"
+#include "vp8/encoder/block.h"
+
+void vp8_short_fdct4x4_mmx(short *input, short *output, int pitch);
+void vp8_short_fdct8x4_mmx(short *input, short *output, int pitch)
+{
+ vp8_short_fdct4x4_mmx(input, output, pitch);
+ vp8_short_fdct4x4_mmx(input + 4, output + 16, pitch);
+}
+
+int vp8_fast_quantize_b_impl_mmx(short *coeff_ptr, short *zbin_ptr,
+ short *qcoeff_ptr, short *dequant_ptr,
+ const short *scan_mask, short *round_ptr,
+ short *quant_ptr, short *dqcoeff_ptr);
+void vp8_fast_quantize_b_mmx(BLOCK *b, BLOCKD *d)
+{
+ const short *scan_mask = vp8_default_zig_zag_mask;
+ short *coeff_ptr = b->coeff;
+ short *zbin_ptr = b->zbin;
+ short *round_ptr = b->round;
+ short *quant_ptr = b->quant_fast;
+ short *qcoeff_ptr = d->qcoeff;
+ short *dqcoeff_ptr = d->dqcoeff;
+ short *dequant_ptr = d->dequant;
+
+ *d->eob = (char)vp8_fast_quantize_b_impl_mmx(
+ coeff_ptr,
+ zbin_ptr,
+ qcoeff_ptr,
+ dequant_ptr,
+ scan_mask,
+
+ round_ptr,
+ quant_ptr,
+ dqcoeff_ptr
+ );
+}
+
+int vp8_mbblock_error_mmx_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
+int vp8_mbblock_error_mmx(MACROBLOCK *mb, int dc)
+{
+ short *coeff_ptr = mb->block[0].coeff;
+ short *dcoef_ptr = mb->e_mbd.block[0].dqcoeff;
+ return vp8_mbblock_error_mmx_impl(coeff_ptr, dcoef_ptr, dc);
+}
+
+int vp8_mbuverror_mmx_impl(short *s_ptr, short *d_ptr);
+int vp8_mbuverror_mmx(MACROBLOCK *mb)
+{
+ short *s_ptr = &mb->coeff[256];
+ short *d_ptr = &mb->e_mbd.dqcoeff[256];
+ return vp8_mbuverror_mmx_impl(s_ptr, d_ptr);
+}
+
+void vp8_subtract_b_mmx_impl(unsigned char *z, int src_stride,
+ short *diff, unsigned char *predictor,
+ int pitch);
+void vp8_subtract_b_mmx(BLOCK *be, BLOCKD *bd, int pitch)
+{
+ unsigned char *z = *(be->base_src) + be->src;
+ unsigned int src_stride = be->src_stride;
+ short *diff = &be->src_diff[0];
+ unsigned char *predictor = &bd->predictor[0];
+ vp8_subtract_b_mmx_impl(z, src_stride, diff, predictor, pitch);
+}
diff --git a/libvpx/vp8/encoder/x86/vp8_enc_stubs_sse2.c b/libvpx/vp8/encoder/x86/vp8_enc_stubs_sse2.c
new file mode 100644
index 0000000..68db815
--- /dev/null
+++ b/libvpx/vp8/encoder/x86/vp8_enc_stubs_sse2.c
@@ -0,0 +1,43 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vpx_ports/x86.h"
+#include "vp8/encoder/block.h"
+
+int vp8_mbblock_error_xmm_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
+int vp8_mbblock_error_xmm(MACROBLOCK *mb, int dc)
+{
+ short *coeff_ptr = mb->block[0].coeff;
+ short *dcoef_ptr = mb->e_mbd.block[0].dqcoeff;
+ return vp8_mbblock_error_xmm_impl(coeff_ptr, dcoef_ptr, dc);
+}
+
+int vp8_mbuverror_xmm_impl(short *s_ptr, short *d_ptr);
+int vp8_mbuverror_xmm(MACROBLOCK *mb)
+{
+ short *s_ptr = &mb->coeff[256];
+ short *d_ptr = &mb->e_mbd.dqcoeff[256];
+ return vp8_mbuverror_xmm_impl(s_ptr, d_ptr);
+}
+
+void vp8_subtract_b_sse2_impl(unsigned char *z, int src_stride,
+ short *diff, unsigned char *predictor,
+ int pitch);
+void vp8_subtract_b_sse2(BLOCK *be, BLOCKD *bd, int pitch)
+{
+ unsigned char *z = *(be->base_src) + be->src;
+ unsigned int src_stride = be->src_stride;
+ short *diff = &be->src_diff[0];
+ unsigned char *predictor = &bd->predictor[0];
+ vp8_subtract_b_sse2_impl(z, src_stride, diff, predictor, pitch);
+}
diff --git a/vp8/exports_dec b/libvpx/vp8/exports_dec
similarity index 100%
rename from vp8/exports_dec
rename to libvpx/vp8/exports_dec
diff --git a/vp8/exports_enc b/libvpx/vp8/exports_enc
similarity index 100%
rename from vp8/exports_enc
rename to libvpx/vp8/exports_enc
diff --git a/libvpx/vp8/vp8_common.mk b/libvpx/vp8/vp8_common.mk
new file mode 100644
index 0000000..a328f46
--- /dev/null
+++ b/libvpx/vp8/vp8_common.mk
@@ -0,0 +1,193 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+VP8_COMMON_SRCS-yes += vp8_common.mk
+VP8_COMMON_SRCS-yes += common/pragmas.h
+VP8_COMMON_SRCS-yes += common/ppflags.h
+VP8_COMMON_SRCS-yes += common/onyx.h
+VP8_COMMON_SRCS-yes += common/onyxd.h
+VP8_COMMON_SRCS-yes += common/alloccommon.c
+VP8_COMMON_SRCS-yes += common/asm_com_offsets.c
+VP8_COMMON_SRCS-yes += common/blockd.c
+VP8_COMMON_SRCS-yes += common/coefupdateprobs.h
+VP8_COMMON_SRCS-yes += common/debugmodes.c
+VP8_COMMON_SRCS-yes += common/default_coef_probs.h
+VP8_COMMON_SRCS-yes += common/dequantize.c
+VP8_COMMON_SRCS-yes += common/entropy.c
+VP8_COMMON_SRCS-yes += common/entropymode.c
+VP8_COMMON_SRCS-yes += common/entropymv.c
+VP8_COMMON_SRCS-yes += common/extend.c
+VP8_COMMON_SRCS-yes += common/filter.c
+VP8_COMMON_SRCS-yes += common/filter.h
+VP8_COMMON_SRCS-yes += common/findnearmv.c
+VP8_COMMON_SRCS-yes += common/generic/systemdependent.c
+VP8_COMMON_SRCS-yes += common/idct_blk.c
+VP8_COMMON_SRCS-yes += common/idctllm.c
+VP8_COMMON_SRCS-yes += common/alloccommon.h
+VP8_COMMON_SRCS-yes += common/blockd.h
+VP8_COMMON_SRCS-yes += common/common.h
+VP8_COMMON_SRCS-yes += common/entropy.h
+VP8_COMMON_SRCS-yes += common/entropymode.h
+VP8_COMMON_SRCS-yes += common/entropymv.h
+VP8_COMMON_SRCS-yes += common/extend.h
+VP8_COMMON_SRCS-yes += common/findnearmv.h
+VP8_COMMON_SRCS-yes += common/header.h
+VP8_COMMON_SRCS-yes += common/invtrans.h
+VP8_COMMON_SRCS-yes += common/loopfilter.h
+VP8_COMMON_SRCS-yes += common/modecont.h
+VP8_COMMON_SRCS-yes += common/mv.h
+VP8_COMMON_SRCS-yes += common/onyxc_int.h
+VP8_COMMON_SRCS-yes += common/quant_common.h
+VP8_COMMON_SRCS-yes += common/reconinter.h
+VP8_COMMON_SRCS-yes += common/reconintra4x4.h
+VP8_COMMON_SRCS-yes += common/rtcd.c
+VP8_COMMON_SRCS-yes += common/rtcd_defs.sh
+VP8_COMMON_SRCS-yes += common/setupintrarecon.h
+VP8_COMMON_SRCS-yes += common/swapyv12buffer.h
+VP8_COMMON_SRCS-yes += common/systemdependent.h
+VP8_COMMON_SRCS-yes += common/threading.h
+VP8_COMMON_SRCS-yes += common/treecoder.h
+VP8_COMMON_SRCS-yes += common/loopfilter.c
+VP8_COMMON_SRCS-yes += common/loopfilter_filters.c
+VP8_COMMON_SRCS-yes += common/mbpitch.c
+VP8_COMMON_SRCS-yes += common/modecont.c
+VP8_COMMON_SRCS-yes += common/quant_common.c
+VP8_COMMON_SRCS-yes += common/reconinter.c
+VP8_COMMON_SRCS-yes += common/reconintra.c
+VP8_COMMON_SRCS-yes += common/reconintra4x4.c
+VP8_COMMON_SRCS-yes += common/sad_c.c
+VP8_COMMON_SRCS-yes += common/setupintrarecon.c
+VP8_COMMON_SRCS-yes += common/swapyv12buffer.c
+VP8_COMMON_SRCS-yes += common/variance_c.c
+VP8_COMMON_SRCS-yes += common/variance.h
+VP8_COMMON_SRCS-yes += common/vp8_entropymodedata.h
+
+
+
+VP8_COMMON_SRCS-$(CONFIG_POSTPROC_VISUALIZER) += common/textblit.c
+VP8_COMMON_SRCS-yes += common/treecoder.c
+
+VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/filter_x86.c
+VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/filter_x86.h
+VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/vp8_asm_stubs.c
+VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/loopfilter_x86.c
+VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/mfqe.c
+VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/postproc.h
+VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/postproc.c
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/dequantize_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/idct_blk_mmx.c
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/idctllm_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/iwalsh_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/loopfilter_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/recon_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/sad_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/subpixel_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/variance_mmx.c
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/variance_impl_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/idct_blk_sse2.c
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/idctllm_sse2.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/recon_sse2.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/recon_wrapper_sse2.c
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/sad_sse2.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/subpixel_sse2.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/loopfilter_sse2.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/iwalsh_sse2.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/variance_sse2.c
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/variance_impl_sse2.asm
+VP8_COMMON_SRCS-$(HAVE_SSE3) += common/x86/sad_sse3.asm
+VP8_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/sad_ssse3.asm
+VP8_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/subpixel_ssse3.asm
+VP8_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/variance_ssse3.c
+VP8_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/variance_impl_ssse3.asm
+VP8_COMMON_SRCS-$(HAVE_SSE4_1) += common/x86/sad_sse4.asm
+
+ifeq ($(CONFIG_POSTPROC),yes)
+VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/postproc_x86.c
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/postproc_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/mfqe_sse2.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/postproc_sse2.asm
+endif
+
+ifeq ($(ARCH_X86_64),yes)
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/loopfilter_block_sse2.asm
+endif
+
+# common (c)
+VP8_COMMON_SRCS-$(HAVE_DSPR2) += common/mips/dspr2/idctllm_dspr2.c
+VP8_COMMON_SRCS-$(HAVE_DSPR2) += common/mips/dspr2/filter_dspr2.c
+VP8_COMMON_SRCS-$(HAVE_DSPR2) += common/mips/dspr2/loopfilter_filters_dspr2.c
+VP8_COMMON_SRCS-$(HAVE_DSPR2) += common/mips/dspr2/reconinter_dspr2.c
+VP8_COMMON_SRCS-$(HAVE_DSPR2) += common/mips/dspr2/idct_blk_dspr2.c
+VP8_COMMON_SRCS-$(HAVE_DSPR2) += common/mips/dspr2/dequantize_dspr2.c
+
+# common (c)
+VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/filter_arm.c
+VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/loopfilter_arm.c
+VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/reconintra_arm.c
+VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/dequantize_arm.c
+VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/variance_arm.c
+
+# common (media)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/bilinearfilter_arm.c
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/bilinearfilter_arm.h
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/bilinearfilter_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/copymem8x4_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/copymem8x8_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/copymem16x16_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/dc_only_idct_add_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/iwalsh_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/filter_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/idct_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/loopfilter_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/simpleloopfilter_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/sixtappredict8x4_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/intra4x4_predict_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/dequant_idct_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/dequantize_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/idct_blk_v6.c
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/vp8_sad16x16_armv6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/vp8_variance8x8_armv6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/vp8_variance16x16_armv6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_MEDIA) += common/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6$(ASM)
+
+# common (neon)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/bilinearpredict4x4_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/bilinearpredict8x4_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/bilinearpredict8x8_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/bilinearpredict16x16_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/copymem8x4_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/copymem8x8_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/copymem16x16_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/dc_only_idct_add_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/iwalsh_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/loopfilter_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/loopfiltersimplehorizontaledge_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/loopfiltersimpleverticaledge_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/mbloopfilter_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/shortidct4x4llm_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/sad8_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/sad16_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/sixtappredict4x4_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/sixtappredict8x4_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/sixtappredict8x8_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/sixtappredict16x16_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/buildintrapredictorsmby_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/save_reg_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/dequant_idct_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/idct_dequant_full_2x_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/idct_dequant_0_2x_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/dequantizeb_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/idct_blk_neon.c
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/variance_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/vp8_subpixelvariance8x8_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/vp8_subpixelvariance16x16_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_NEON) += common/arm/neon/vp8_subpixelvariance16x16s_neon$(ASM)
diff --git a/libvpx/vp8/vp8_cx_iface.c b/libvpx/vp8/vp8_cx_iface.c
new file mode 100644
index 0000000..eeac3a8
--- /dev/null
+++ b/libvpx/vp8/vp8_cx_iface.c
@@ -0,0 +1,1317 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_rtcd.h"
+#include "vpx/vpx_codec.h"
+#include "vpx/internal/vpx_codec_internal.h"
+#include "vpx_version.h"
+#include "vp8/encoder/onyx_int.h"
+#include "vpx/vp8cx.h"
+#include "vp8/encoder/firstpass.h"
+#include "vp8/common/onyx.h"
+#include <stdlib.h>
+#include <string.h>
+
+struct vp8_extracfg
+{
+ struct vpx_codec_pkt_list *pkt_list;
+ int cpu_used; /** available cpu percentage in 1/16*/
+ unsigned int enable_auto_alt_ref; /** if encoder decides to uses alternate reference frame */
+ unsigned int noise_sensitivity;
+ unsigned int Sharpness;
+ unsigned int static_thresh;
+ unsigned int token_partitions;
+ unsigned int arnr_max_frames; /* alt_ref Noise Reduction Max Frame Count */
+ unsigned int arnr_strength; /* alt_ref Noise Reduction Strength */
+ unsigned int arnr_type; /* alt_ref filter type */
+ vp8e_tuning tuning;
+ unsigned int cq_level; /* constrained quality level */
+ unsigned int rc_max_intra_bitrate_pct;
+
+};
+
+struct extraconfig_map
+{
+ int usage;
+ struct vp8_extracfg cfg;
+};
+
+static const struct extraconfig_map extracfg_map[] =
+{
+ {
+ 0,
+ {
+ NULL,
+#if !(CONFIG_REALTIME_ONLY)
+ 0, /* cpu_used */
+#else
+ 4, /* cpu_used */
+#endif
+ 0, /* enable_auto_alt_ref */
+ 0, /* noise_sensitivity */
+ 0, /* Sharpness */
+ 0, /* static_thresh */
+#if (CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ VP8_EIGHT_TOKENPARTITION,
+#else
+ VP8_ONE_TOKENPARTITION, /* token_partitions */
+#endif
+ 0, /* arnr_max_frames */
+ 3, /* arnr_strength */
+ 3, /* arnr_type*/
+ 0, /* tuning*/
+ 10, /* cq_level */
+ 0, /* rc_max_intra_bitrate_pct */
+ }
+ }
+};
+
+struct vpx_codec_alg_priv
+{
+ vpx_codec_priv_t base;
+ vpx_codec_enc_cfg_t cfg;
+ struct vp8_extracfg vp8_cfg;
+ VP8_CONFIG oxcf;
+ struct VP8_COMP *cpi;
+ unsigned char *cx_data;
+ unsigned int cx_data_sz;
+ vpx_image_t preview_img;
+ unsigned int next_frame_flag;
+ vp8_postproc_cfg_t preview_ppcfg;
+ /* pkt_list size depends on the maximum number of lagged frames allowed. */
+ vpx_codec_pkt_list_decl(64) pkt_list;
+ unsigned int fixed_kf_cntr;
+};
+
+
+static vpx_codec_err_t
+update_error_state(vpx_codec_alg_priv_t *ctx,
+ const struct vpx_internal_error_info *error)
+{
+ vpx_codec_err_t res;
+
+ if ((res = error->error_code))
+ ctx->base.err_detail = error->has_detail
+ ? error->detail
+ : NULL;
+
+ return res;
+}
+
+
+#undef ERROR
+#define ERROR(str) do {\
+ ctx->base.err_detail = str;\
+ return VPX_CODEC_INVALID_PARAM;\
+ } while(0)
+
+#define RANGE_CHECK(p,memb,lo,hi) do {\
+ if(!(((p)->memb == lo || (p)->memb > (lo)) && (p)->memb <= hi)) \
+ ERROR(#memb " out of range ["#lo".."#hi"]");\
+ } while(0)
+
+#define RANGE_CHECK_HI(p,memb,hi) do {\
+ if(!((p)->memb <= (hi))) \
+ ERROR(#memb " out of range [.."#hi"]");\
+ } while(0)
+
+#define RANGE_CHECK_LO(p,memb,lo) do {\
+ if(!((p)->memb >= (lo))) \
+ ERROR(#memb " out of range ["#lo"..]");\
+ } while(0)
+
+#define RANGE_CHECK_BOOL(p,memb) do {\
+ if(!!((p)->memb) != (p)->memb) ERROR(#memb " expected boolean");\
+ } while(0)
+
+static vpx_codec_err_t validate_config(vpx_codec_alg_priv_t *ctx,
+ const vpx_codec_enc_cfg_t *cfg,
+ const struct vp8_extracfg *vp8_cfg,
+ int finalize)
+{
+ RANGE_CHECK(cfg, g_w, 1, 16383); /* 14 bits available */
+ RANGE_CHECK(cfg, g_h, 1, 16383); /* 14 bits available */
+ RANGE_CHECK(cfg, g_timebase.den, 1, 1000000000);
+ RANGE_CHECK(cfg, g_timebase.num, 1, cfg->g_timebase.den);
+ RANGE_CHECK_HI(cfg, g_profile, 3);
+ RANGE_CHECK_HI(cfg, rc_max_quantizer, 63);
+ RANGE_CHECK_HI(cfg, rc_min_quantizer, cfg->rc_max_quantizer);
+ RANGE_CHECK_HI(cfg, g_threads, 64);
+#if CONFIG_REALTIME_ONLY
+ RANGE_CHECK_HI(cfg, g_lag_in_frames, 0);
+#elif CONFIG_MULTI_RES_ENCODING
+ if (ctx->base.enc.total_encoders > 1)
+ RANGE_CHECK_HI(cfg, g_lag_in_frames, 0);
+#else
+ RANGE_CHECK_HI(cfg, g_lag_in_frames, 25);
+#endif
+ RANGE_CHECK(cfg, rc_end_usage, VPX_VBR, VPX_CQ);
+ RANGE_CHECK_HI(cfg, rc_undershoot_pct, 1000);
+ RANGE_CHECK_HI(cfg, rc_overshoot_pct, 1000);
+ RANGE_CHECK_HI(cfg, rc_2pass_vbr_bias_pct, 100);
+ RANGE_CHECK(cfg, kf_mode, VPX_KF_DISABLED, VPX_KF_AUTO);
+
+/* TODO: add spatial re-sampling support and frame dropping in
+ * multi-res-encoder.*/
+#if CONFIG_MULTI_RES_ENCODING
+ if (ctx->base.enc.total_encoders > 1)
+ RANGE_CHECK_HI(cfg, rc_resize_allowed, 0);
+#else
+ RANGE_CHECK_BOOL(cfg, rc_resize_allowed);
+#endif
+ RANGE_CHECK_HI(cfg, rc_dropframe_thresh, 100);
+ RANGE_CHECK_HI(cfg, rc_resize_up_thresh, 100);
+ RANGE_CHECK_HI(cfg, rc_resize_down_thresh, 100);
+
+#if CONFIG_REALTIME_ONLY
+ RANGE_CHECK(cfg, g_pass, VPX_RC_ONE_PASS, VPX_RC_ONE_PASS);
+#elif CONFIG_MULTI_RES_ENCODING
+ if (ctx->base.enc.total_encoders > 1)
+ RANGE_CHECK(cfg, g_pass, VPX_RC_ONE_PASS, VPX_RC_ONE_PASS);
+#else
+ RANGE_CHECK(cfg, g_pass, VPX_RC_ONE_PASS, VPX_RC_LAST_PASS);
+#endif
+
+ /* VP8 does not support a lower bound on the keyframe interval in
+ * automatic keyframe placement mode.
+ */
+ if (cfg->kf_mode != VPX_KF_DISABLED && cfg->kf_min_dist != cfg->kf_max_dist
+ && cfg->kf_min_dist > 0)
+ ERROR("kf_min_dist not supported in auto mode, use 0 "
+ "or kf_max_dist instead.");
+
+ RANGE_CHECK_BOOL(vp8_cfg, enable_auto_alt_ref);
+ RANGE_CHECK(vp8_cfg, cpu_used, -16, 16);
+
+#if CONFIG_REALTIME_ONLY && !CONFIG_TEMPORAL_DENOISING
+ RANGE_CHECK(vp8_cfg, noise_sensitivity, 0, 0);
+#else
+ RANGE_CHECK_HI(vp8_cfg, noise_sensitivity, 6);
+#endif
+
+ RANGE_CHECK(vp8_cfg, token_partitions, VP8_ONE_TOKENPARTITION,
+ VP8_EIGHT_TOKENPARTITION);
+ RANGE_CHECK_HI(vp8_cfg, Sharpness, 7);
+ RANGE_CHECK(vp8_cfg, arnr_max_frames, 0, 15);
+ RANGE_CHECK_HI(vp8_cfg, arnr_strength, 6);
+ RANGE_CHECK(vp8_cfg, arnr_type, 1, 3);
+ RANGE_CHECK(vp8_cfg, cq_level, 0, 63);
+ if(finalize && cfg->rc_end_usage == VPX_CQ)
+ RANGE_CHECK(vp8_cfg, cq_level,
+ cfg->rc_min_quantizer, cfg->rc_max_quantizer);
+
+#if !(CONFIG_REALTIME_ONLY)
+ if (cfg->g_pass == VPX_RC_LAST_PASS)
+ {
+ size_t packet_sz = sizeof(FIRSTPASS_STATS);
+ int n_packets = (int)(cfg->rc_twopass_stats_in.sz /
+ packet_sz);
+ FIRSTPASS_STATS *stats;
+
+ if (!cfg->rc_twopass_stats_in.buf)
+ ERROR("rc_twopass_stats_in.buf not set.");
+
+ if (cfg->rc_twopass_stats_in.sz % packet_sz)
+ ERROR("rc_twopass_stats_in.sz indicates truncated packet.");
+
+ if (cfg->rc_twopass_stats_in.sz < 2 * packet_sz)
+ ERROR("rc_twopass_stats_in requires at least two packets.");
+
+ stats = (void*)((char *)cfg->rc_twopass_stats_in.buf
+ + (n_packets - 1) * packet_sz);
+
+ if ((int)(stats->count + 0.5) != n_packets - 1)
+ ERROR("rc_twopass_stats_in missing EOS stats packet");
+ }
+#endif
+
+ RANGE_CHECK(cfg, ts_number_layers, 1, 5);
+
+ if (cfg->ts_number_layers > 1)
+ {
+ unsigned int i;
+ RANGE_CHECK_HI(cfg, ts_periodicity, 16);
+
+ for (i=1; i<cfg->ts_number_layers; i++)
+ if (cfg->ts_target_bitrate[i] <= cfg->ts_target_bitrate[i-1])
+ ERROR("ts_target_bitrate entries are not strictly increasing");
+
+ RANGE_CHECK(cfg, ts_rate_decimator[cfg->ts_number_layers-1], 1, 1);
+ for (i=cfg->ts_number_layers-2; i>0; i--)
+ if (cfg->ts_rate_decimator[i-1] != 2*cfg->ts_rate_decimator[i])
+ ERROR("ts_rate_decimator factors are not powers of 2");
+
+ RANGE_CHECK_HI(cfg, ts_layer_id[i], cfg->ts_number_layers-1);
+ }
+
+#if (CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING)
+ if(cfg->g_threads > (1 << vp8_cfg->token_partitions))
+ ERROR("g_threads cannot be bigger than number of token partitions");
+#endif
+
+ return VPX_CODEC_OK;
+}
+
+
+static vpx_codec_err_t validate_img(vpx_codec_alg_priv_t *ctx,
+ const vpx_image_t *img)
+{
+ switch (img->fmt)
+ {
+ case VPX_IMG_FMT_YV12:
+ case VPX_IMG_FMT_I420:
+ case VPX_IMG_FMT_VPXI420:
+ case VPX_IMG_FMT_VPXYV12:
+ break;
+ default:
+ ERROR("Invalid image format. Only YV12 and I420 images are supported");
+ }
+
+ if ((img->d_w != ctx->cfg.g_w) || (img->d_h != ctx->cfg.g_h))
+ ERROR("Image size must match encoder init configuration size");
+
+ return VPX_CODEC_OK;
+}
+
+
+static vpx_codec_err_t set_vp8e_config(VP8_CONFIG *oxcf,
+ vpx_codec_enc_cfg_t cfg,
+ struct vp8_extracfg vp8_cfg,
+ vpx_codec_priv_enc_mr_cfg_t *mr_cfg)
+{
+ oxcf->multi_threaded = cfg.g_threads;
+ oxcf->Version = cfg.g_profile;
+
+ oxcf->Width = cfg.g_w;
+ oxcf->Height = cfg.g_h;
+ oxcf->timebase = cfg.g_timebase;
+
+ oxcf->error_resilient_mode = cfg.g_error_resilient;
+
+ switch (cfg.g_pass)
+ {
+ case VPX_RC_ONE_PASS:
+ oxcf->Mode = MODE_BESTQUALITY;
+ break;
+ case VPX_RC_FIRST_PASS:
+ oxcf->Mode = MODE_FIRSTPASS;
+ break;
+ case VPX_RC_LAST_PASS:
+ oxcf->Mode = MODE_SECONDPASS_BEST;
+ break;
+ }
+
+ if (cfg.g_pass == VPX_RC_FIRST_PASS || cfg.g_pass == VPX_RC_ONE_PASS)
+ {
+ oxcf->allow_lag = 0;
+ oxcf->lag_in_frames = 0;
+ }
+ else
+ {
+ oxcf->allow_lag = (cfg.g_lag_in_frames) > 0;
+ oxcf->lag_in_frames = cfg.g_lag_in_frames;
+ }
+
+ oxcf->allow_df = (cfg.rc_dropframe_thresh > 0);
+ oxcf->drop_frames_water_mark = cfg.rc_dropframe_thresh;
+
+ oxcf->allow_spatial_resampling = cfg.rc_resize_allowed;
+ oxcf->resample_up_water_mark = cfg.rc_resize_up_thresh;
+ oxcf->resample_down_water_mark = cfg.rc_resize_down_thresh;
+
+ if (cfg.rc_end_usage == VPX_VBR)
+ {
+ oxcf->end_usage = USAGE_LOCAL_FILE_PLAYBACK;
+ }
+ else if (cfg.rc_end_usage == VPX_CBR)
+ {
+ oxcf->end_usage = USAGE_STREAM_FROM_SERVER;
+ }
+ else if (cfg.rc_end_usage == VPX_CQ)
+ {
+ oxcf->end_usage = USAGE_CONSTRAINED_QUALITY;
+ }
+
+ oxcf->target_bandwidth = cfg.rc_target_bitrate;
+ oxcf->rc_max_intra_bitrate_pct = vp8_cfg.rc_max_intra_bitrate_pct;
+
+ oxcf->best_allowed_q = cfg.rc_min_quantizer;
+ oxcf->worst_allowed_q = cfg.rc_max_quantizer;
+ oxcf->cq_level = vp8_cfg.cq_level;
+ oxcf->fixed_q = -1;
+
+ oxcf->under_shoot_pct = cfg.rc_undershoot_pct;
+ oxcf->over_shoot_pct = cfg.rc_overshoot_pct;
+
+ oxcf->maximum_buffer_size_in_ms = cfg.rc_buf_sz;
+ oxcf->starting_buffer_level_in_ms = cfg.rc_buf_initial_sz;
+ oxcf->optimal_buffer_level_in_ms = cfg.rc_buf_optimal_sz;
+
+ oxcf->maximum_buffer_size = cfg.rc_buf_sz;
+ oxcf->starting_buffer_level = cfg.rc_buf_initial_sz;
+ oxcf->optimal_buffer_level = cfg.rc_buf_optimal_sz;
+
+ oxcf->two_pass_vbrbias = cfg.rc_2pass_vbr_bias_pct;
+ oxcf->two_pass_vbrmin_section = cfg.rc_2pass_vbr_minsection_pct;
+ oxcf->two_pass_vbrmax_section = cfg.rc_2pass_vbr_maxsection_pct;
+
+ oxcf->auto_key = cfg.kf_mode == VPX_KF_AUTO
+ && cfg.kf_min_dist != cfg.kf_max_dist;
+ oxcf->key_freq = cfg.kf_max_dist;
+
+ oxcf->number_of_layers = cfg.ts_number_layers;
+ oxcf->periodicity = cfg.ts_periodicity;
+
+ if (oxcf->number_of_layers > 1)
+ {
+ memcpy (oxcf->target_bitrate, cfg.ts_target_bitrate,
+ sizeof(cfg.ts_target_bitrate));
+ memcpy (oxcf->rate_decimator, cfg.ts_rate_decimator,
+ sizeof(cfg.ts_rate_decimator));
+ memcpy (oxcf->layer_id, cfg.ts_layer_id, sizeof(cfg.ts_layer_id));
+ }
+
+#if CONFIG_MULTI_RES_ENCODING
+ /* When mr_cfg is NULL, oxcf->mr_total_resolutions and oxcf->mr_encoder_id
+ * are both memset to 0, which ensures the correct logic under this
+ * situation.
+ */
+ if(mr_cfg)
+ {
+ oxcf->mr_total_resolutions = mr_cfg->mr_total_resolutions;
+ oxcf->mr_encoder_id = mr_cfg->mr_encoder_id;
+ oxcf->mr_down_sampling_factor.num = mr_cfg->mr_down_sampling_factor.num;
+ oxcf->mr_down_sampling_factor.den = mr_cfg->mr_down_sampling_factor.den;
+ oxcf->mr_low_res_mode_info = mr_cfg->mr_low_res_mode_info;
+ }
+#endif
+
+ oxcf->cpu_used = vp8_cfg.cpu_used;
+ oxcf->encode_breakout = vp8_cfg.static_thresh;
+ oxcf->play_alternate = vp8_cfg.enable_auto_alt_ref;
+ oxcf->noise_sensitivity = vp8_cfg.noise_sensitivity;
+ oxcf->Sharpness = vp8_cfg.Sharpness;
+ oxcf->token_partitions = vp8_cfg.token_partitions;
+
+ oxcf->two_pass_stats_in = cfg.rc_twopass_stats_in;
+ oxcf->output_pkt_list = vp8_cfg.pkt_list;
+
+ oxcf->arnr_max_frames = vp8_cfg.arnr_max_frames;
+ oxcf->arnr_strength = vp8_cfg.arnr_strength;
+ oxcf->arnr_type = vp8_cfg.arnr_type;
+
+ oxcf->tuning = vp8_cfg.tuning;
+
+ /*
+ printf("Current VP8 Settings: \n");
+ printf("target_bandwidth: %d\n", oxcf->target_bandwidth);
+ printf("noise_sensitivity: %d\n", oxcf->noise_sensitivity);
+ printf("Sharpness: %d\n", oxcf->Sharpness);
+ printf("cpu_used: %d\n", oxcf->cpu_used);
+ printf("Mode: %d\n", oxcf->Mode);
+ printf("delete_first_pass_file: %d\n", oxcf->delete_first_pass_file);
+ printf("auto_key: %d\n", oxcf->auto_key);
+ printf("key_freq: %d\n", oxcf->key_freq);
+ printf("end_usage: %d\n", oxcf->end_usage);
+ printf("under_shoot_pct: %d\n", oxcf->under_shoot_pct);
+ printf("over_shoot_pct: %d\n", oxcf->over_shoot_pct);
+ printf("starting_buffer_level: %d\n", oxcf->starting_buffer_level);
+ printf("optimal_buffer_level: %d\n", oxcf->optimal_buffer_level);
+ printf("maximum_buffer_size: %d\n", oxcf->maximum_buffer_size);
+ printf("fixed_q: %d\n", oxcf->fixed_q);
+ printf("worst_allowed_q: %d\n", oxcf->worst_allowed_q);
+ printf("best_allowed_q: %d\n", oxcf->best_allowed_q);
+ printf("allow_spatial_resampling: %d\n", oxcf->allow_spatial_resampling);
+ printf("resample_down_water_mark: %d\n", oxcf->resample_down_water_mark);
+ printf("resample_up_water_mark: %d\n", oxcf->resample_up_water_mark);
+ printf("allow_df: %d\n", oxcf->allow_df);
+ printf("drop_frames_water_mark: %d\n", oxcf->drop_frames_water_mark);
+ printf("two_pass_vbrbias: %d\n", oxcf->two_pass_vbrbias);
+ printf("two_pass_vbrmin_section: %d\n", oxcf->two_pass_vbrmin_section);
+ printf("two_pass_vbrmax_section: %d\n", oxcf->two_pass_vbrmax_section);
+ printf("allow_lag: %d\n", oxcf->allow_lag);
+ printf("lag_in_frames: %d\n", oxcf->lag_in_frames);
+ printf("play_alternate: %d\n", oxcf->play_alternate);
+ printf("Version: %d\n", oxcf->Version);
+ printf("multi_threaded: %d\n", oxcf->multi_threaded);
+ printf("encode_breakout: %d\n", oxcf->encode_breakout);
+ */
+ return VPX_CODEC_OK;
+}
+
+static vpx_codec_err_t vp8e_set_config(vpx_codec_alg_priv_t *ctx,
+ const vpx_codec_enc_cfg_t *cfg)
+{
+ vpx_codec_err_t res;
+
+ if (((cfg->g_w != ctx->cfg.g_w) || (cfg->g_h != ctx->cfg.g_h))
+ && (cfg->g_lag_in_frames > 1 || cfg->g_pass != VPX_RC_ONE_PASS))
+ ERROR("Cannot change width or height after initialization");
+
+ /* Prevent increasing lag_in_frames. This check is stricter than it needs
+ * to be -- the limit is not increasing past the first lag_in_frames
+ * value, but we don't track the initial config, only the last successful
+ * config.
+ */
+ if ((cfg->g_lag_in_frames > ctx->cfg.g_lag_in_frames))
+ ERROR("Cannot increase lag_in_frames");
+
+ res = validate_config(ctx, cfg, &ctx->vp8_cfg, 0);
+
+ if (!res)
+ {
+ ctx->cfg = *cfg;
+ set_vp8e_config(&ctx->oxcf, ctx->cfg, ctx->vp8_cfg, NULL);
+ vp8_change_config(ctx->cpi, &ctx->oxcf);
+ }
+
+ return res;
+}
+
+
+int vp8_reverse_trans(int);
+
+
+static vpx_codec_err_t get_param(vpx_codec_alg_priv_t *ctx,
+ int ctrl_id,
+ va_list args)
+{
+ void *arg = va_arg(args, void *);
+
+#define MAP(id, var) case id: *(RECAST(id, arg)) = var; break
+
+ if (!arg)
+ return VPX_CODEC_INVALID_PARAM;
+
+ switch (ctrl_id)
+ {
+ MAP(VP8E_GET_LAST_QUANTIZER, vp8_get_quantizer(ctx->cpi));
+ MAP(VP8E_GET_LAST_QUANTIZER_64, vp8_reverse_trans(vp8_get_quantizer(ctx->cpi)));
+ }
+
+ return VPX_CODEC_OK;
+#undef MAP
+}
+
+
+static vpx_codec_err_t set_param(vpx_codec_alg_priv_t *ctx,
+ int ctrl_id,
+ va_list args)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+ struct vp8_extracfg xcfg = ctx->vp8_cfg;
+
+#define MAP(id, var) case id: var = CAST(id, args); break;
+
+ switch (ctrl_id)
+ {
+ MAP(VP8E_SET_CPUUSED, xcfg.cpu_used);
+ MAP(VP8E_SET_ENABLEAUTOALTREF, xcfg.enable_auto_alt_ref);
+ MAP(VP8E_SET_NOISE_SENSITIVITY, xcfg.noise_sensitivity);
+ MAP(VP8E_SET_SHARPNESS, xcfg.Sharpness);
+ MAP(VP8E_SET_STATIC_THRESHOLD, xcfg.static_thresh);
+ MAP(VP8E_SET_TOKEN_PARTITIONS, xcfg.token_partitions);
+
+ MAP(VP8E_SET_ARNR_MAXFRAMES, xcfg.arnr_max_frames);
+ MAP(VP8E_SET_ARNR_STRENGTH , xcfg.arnr_strength);
+ MAP(VP8E_SET_ARNR_TYPE , xcfg.arnr_type);
+ MAP(VP8E_SET_TUNING, xcfg.tuning);
+ MAP(VP8E_SET_CQ_LEVEL, xcfg.cq_level);
+ MAP(VP8E_SET_MAX_INTRA_BITRATE_PCT, xcfg.rc_max_intra_bitrate_pct);
+
+ }
+
+ res = validate_config(ctx, &ctx->cfg, &xcfg, 0);
+
+ if (!res)
+ {
+ ctx->vp8_cfg = xcfg;
+ set_vp8e_config(&ctx->oxcf, ctx->cfg, ctx->vp8_cfg, NULL);
+ vp8_change_config(ctx->cpi, &ctx->oxcf);
+ }
+
+ return res;
+#undef MAP
+}
+
+static vpx_codec_err_t vp8e_mr_alloc_mem(const vpx_codec_enc_cfg_t *cfg,
+ void **mem_loc)
+{
+ vpx_codec_err_t res = 0;
+
+#if CONFIG_MULTI_RES_ENCODING
+ LOWER_RES_FRAME_INFO *shared_mem_loc;
+ int mb_rows = ((cfg->g_w + 15) >>4);
+ int mb_cols = ((cfg->g_h + 15) >>4);
+
+ shared_mem_loc = calloc(1, sizeof(LOWER_RES_FRAME_INFO));
+ if(!shared_mem_loc)
+ {
+ res = VPX_CODEC_MEM_ERROR;
+ }
+
+ shared_mem_loc->mb_info = calloc(mb_rows*mb_cols, sizeof(LOWER_RES_MB_INFO));
+ if(!(shared_mem_loc->mb_info))
+ {
+ res = VPX_CODEC_MEM_ERROR;
+ }
+ else
+ {
+ *mem_loc = (void *)shared_mem_loc;
+ res = VPX_CODEC_OK;
+ }
+#endif
+ return res;
+}
+
+static vpx_codec_err_t vp8e_init(vpx_codec_ctx_t *ctx,
+ vpx_codec_priv_enc_mr_cfg_t *mr_cfg)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+ struct vpx_codec_alg_priv *priv;
+ vpx_codec_enc_cfg_t *cfg;
+ unsigned int i;
+
+ struct VP8_COMP *optr;
+
+ vpx_rtcd();
+
+ if (!ctx->priv)
+ {
+ priv = calloc(1, sizeof(struct vpx_codec_alg_priv));
+
+ if (!priv)
+ {
+ return VPX_CODEC_MEM_ERROR;
+ }
+
+ ctx->priv = &priv->base;
+ ctx->priv->sz = sizeof(*ctx->priv);
+ ctx->priv->iface = ctx->iface;
+ ctx->priv->alg_priv = priv;
+ ctx->priv->init_flags = ctx->init_flags;
+
+ if (ctx->config.enc)
+ {
+ /* Update the reference to the config structure to an
+ * internal copy.
+ */
+ ctx->priv->alg_priv->cfg = *ctx->config.enc;
+ ctx->config.enc = &ctx->priv->alg_priv->cfg;
+ }
+
+ cfg = &ctx->priv->alg_priv->cfg;
+
+ /* Select the extra vp8 configuration table based on the current
+ * usage value. If the current usage value isn't found, use the
+ * values for usage case 0.
+ */
+ for (i = 0;
+ extracfg_map[i].usage && extracfg_map[i].usage != cfg->g_usage;
+ i++);
+
+ priv->vp8_cfg = extracfg_map[i].cfg;
+ priv->vp8_cfg.pkt_list = &priv->pkt_list.head;
+
+ priv->cx_data_sz = priv->cfg.g_w * priv->cfg.g_h * 3 / 2 * 2;
+
+ if (priv->cx_data_sz < 32768) priv->cx_data_sz = 32768;
+
+ priv->cx_data = malloc(priv->cx_data_sz);
+
+ if (!priv->cx_data)
+ {
+ return VPX_CODEC_MEM_ERROR;
+ }
+
+ if(mr_cfg)
+ ctx->priv->enc.total_encoders = mr_cfg->mr_total_resolutions;
+ else
+ ctx->priv->enc.total_encoders = 1;
+
+ res = validate_config(priv, &priv->cfg, &priv->vp8_cfg, 0);
+
+ if (!res)
+ {
+ set_vp8e_config(&ctx->priv->alg_priv->oxcf,
+ ctx->priv->alg_priv->cfg,
+ ctx->priv->alg_priv->vp8_cfg,
+ mr_cfg);
+
+ optr = vp8_create_compressor(&ctx->priv->alg_priv->oxcf);
+
+ if (!optr)
+ res = VPX_CODEC_MEM_ERROR;
+ else
+ ctx->priv->alg_priv->cpi = optr;
+ }
+ }
+
+ return res;
+}
+
+static vpx_codec_err_t vp8e_destroy(vpx_codec_alg_priv_t *ctx)
+{
+#if CONFIG_MULTI_RES_ENCODING
+ /* Free multi-encoder shared memory */
+ if (ctx->oxcf.mr_total_resolutions > 0 && (ctx->oxcf.mr_encoder_id == ctx->oxcf.mr_total_resolutions-1))
+ {
+ LOWER_RES_FRAME_INFO *shared_mem_loc = (LOWER_RES_FRAME_INFO *)ctx->oxcf.mr_low_res_mode_info;
+ free(shared_mem_loc->mb_info);
+ free(ctx->oxcf.mr_low_res_mode_info);
+ }
+#endif
+
+ free(ctx->cx_data);
+ vp8_remove_compressor(&ctx->cpi);
+ free(ctx);
+ return VPX_CODEC_OK;
+}
+
+static vpx_codec_err_t image2yuvconfig(const vpx_image_t *img,
+ YV12_BUFFER_CONFIG *yv12)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+ yv12->y_buffer = img->planes[VPX_PLANE_Y];
+ yv12->u_buffer = img->planes[VPX_PLANE_U];
+ yv12->v_buffer = img->planes[VPX_PLANE_V];
+
+ yv12->y_width = img->d_w;
+ yv12->y_height = img->d_h;
+ yv12->uv_width = (1 + yv12->y_width) / 2;
+ yv12->uv_height = (1 + yv12->y_height) / 2;
+
+ yv12->y_stride = img->stride[VPX_PLANE_Y];
+ yv12->uv_stride = img->stride[VPX_PLANE_U];
+
+ yv12->border = (img->stride[VPX_PLANE_Y] - img->w) / 2;
+ yv12->clrtype = (img->fmt == VPX_IMG_FMT_VPXI420 || img->fmt == VPX_IMG_FMT_VPXYV12);
+ return res;
+}
+
+static void pick_quickcompress_mode(vpx_codec_alg_priv_t *ctx,
+ unsigned long duration,
+ unsigned long deadline)
+{
+ unsigned int new_qc;
+
+#if !(CONFIG_REALTIME_ONLY)
+ /* Use best quality mode if no deadline is given. */
+ new_qc = MODE_BESTQUALITY;
+
+ if (deadline)
+ {
+ uint64_t duration_us;
+
+ /* Convert duration parameter from stream timebase to microseconds */
+ duration_us = (uint64_t)duration * 1000000
+ * (uint64_t)ctx->cfg.g_timebase.num
+ / (uint64_t)ctx->cfg.g_timebase.den;
+
+ /* If the deadline is more that the duration this frame is to be shown,
+ * use good quality mode. Otherwise use realtime mode.
+ */
+ new_qc = (deadline > duration_us) ? MODE_GOODQUALITY : MODE_REALTIME;
+ }
+
+#else
+ new_qc = MODE_REALTIME;
+#endif
+
+ if (ctx->cfg.g_pass == VPX_RC_FIRST_PASS)
+ new_qc = MODE_FIRSTPASS;
+ else if (ctx->cfg.g_pass == VPX_RC_LAST_PASS)
+ new_qc = (new_qc == MODE_BESTQUALITY)
+ ? MODE_SECONDPASS_BEST
+ : MODE_SECONDPASS;
+
+ if (ctx->oxcf.Mode != new_qc)
+ {
+ ctx->oxcf.Mode = new_qc;
+ vp8_change_config(ctx->cpi, &ctx->oxcf);
+ }
+}
+
+
+static vpx_codec_err_t vp8e_encode(vpx_codec_alg_priv_t *ctx,
+ const vpx_image_t *img,
+ vpx_codec_pts_t pts,
+ unsigned long duration,
+ vpx_enc_frame_flags_t flags,
+ unsigned long deadline)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+
+ if (!ctx->cfg.rc_target_bitrate)
+ return res;
+
+ if (!ctx->cfg.rc_target_bitrate)
+ return res;
+
+ if (img)
+ res = validate_img(ctx, img);
+
+ if (!res)
+ res = validate_config(ctx, &ctx->cfg, &ctx->vp8_cfg, 1);
+
+ pick_quickcompress_mode(ctx, duration, deadline);
+ vpx_codec_pkt_list_init(&ctx->pkt_list);
+
+ /* Handle Flags */
+ if (((flags & VP8_EFLAG_NO_UPD_GF) && (flags & VP8_EFLAG_FORCE_GF))
+ || ((flags & VP8_EFLAG_NO_UPD_ARF) && (flags & VP8_EFLAG_FORCE_ARF)))
+ {
+ ctx->base.err_detail = "Conflicting flags.";
+ return VPX_CODEC_INVALID_PARAM;
+ }
+
+ if (flags & (VP8_EFLAG_NO_REF_LAST | VP8_EFLAG_NO_REF_GF
+ | VP8_EFLAG_NO_REF_ARF))
+ {
+ int ref = 7;
+
+ if (flags & VP8_EFLAG_NO_REF_LAST)
+ ref ^= VP8_LAST_FRAME;
+
+ if (flags & VP8_EFLAG_NO_REF_GF)
+ ref ^= VP8_GOLD_FRAME;
+
+ if (flags & VP8_EFLAG_NO_REF_ARF)
+ ref ^= VP8_ALTR_FRAME;
+
+ vp8_use_as_reference(ctx->cpi, ref);
+ }
+
+ if (flags & (VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF
+ | VP8_EFLAG_NO_UPD_ARF | VP8_EFLAG_FORCE_GF
+ | VP8_EFLAG_FORCE_ARF))
+ {
+ int upd = 7;
+
+ if (flags & VP8_EFLAG_NO_UPD_LAST)
+ upd ^= VP8_LAST_FRAME;
+
+ if (flags & VP8_EFLAG_NO_UPD_GF)
+ upd ^= VP8_GOLD_FRAME;
+
+ if (flags & VP8_EFLAG_NO_UPD_ARF)
+ upd ^= VP8_ALTR_FRAME;
+
+ vp8_update_reference(ctx->cpi, upd);
+ }
+
+ if (flags & VP8_EFLAG_NO_UPD_ENTROPY)
+ {
+ vp8_update_entropy(ctx->cpi, 0);
+ }
+
+ /* Handle fixed keyframe intervals */
+ if (ctx->cfg.kf_mode == VPX_KF_AUTO
+ && ctx->cfg.kf_min_dist == ctx->cfg.kf_max_dist)
+ {
+ if (++ctx->fixed_kf_cntr > ctx->cfg.kf_min_dist)
+ {
+ flags |= VPX_EFLAG_FORCE_KF;
+ ctx->fixed_kf_cntr = 1;
+ }
+ }
+
+ /* Initialize the encoder instance on the first frame*/
+ if (!res && ctx->cpi)
+ {
+ unsigned int lib_flags;
+ YV12_BUFFER_CONFIG sd;
+ int64_t dst_time_stamp, dst_end_time_stamp;
+ unsigned long size, cx_data_sz;
+ unsigned char *cx_data;
+ unsigned char *cx_data_end;
+ int comp_data_state = 0;
+
+ /* Set up internal flags */
+ if (ctx->base.init_flags & VPX_CODEC_USE_PSNR)
+ ((VP8_COMP *)ctx->cpi)->b_calculate_psnr = 1;
+
+ if (ctx->base.init_flags & VPX_CODEC_USE_OUTPUT_PARTITION)
+ ((VP8_COMP *)ctx->cpi)->output_partition = 1;
+
+ /* Convert API flags to internal codec lib flags */
+ lib_flags = (flags & VPX_EFLAG_FORCE_KF) ? FRAMEFLAGS_KEY : 0;
+
+ /* vp8 use 10,000,000 ticks/second as time stamp */
+ dst_time_stamp = pts * 10000000 * ctx->cfg.g_timebase.num / ctx->cfg.g_timebase.den;
+ dst_end_time_stamp = (pts + duration) * 10000000 * ctx->cfg.g_timebase.num / ctx->cfg.g_timebase.den;
+
+ if (img != NULL)
+ {
+ res = image2yuvconfig(img, &sd);
+
+ if (vp8_receive_raw_frame(ctx->cpi, ctx->next_frame_flag | lib_flags,
+ &sd, dst_time_stamp, dst_end_time_stamp))
+ {
+ VP8_COMP *cpi = (VP8_COMP *)ctx->cpi;
+ res = update_error_state(ctx, &cpi->common.error);
+ }
+
+ /* reset for next frame */
+ ctx->next_frame_flag = 0;
+ }
+
+ cx_data = ctx->cx_data;
+ cx_data_sz = ctx->cx_data_sz;
+ cx_data_end = ctx->cx_data + cx_data_sz;
+ lib_flags = 0;
+
+ while (cx_data_sz >= ctx->cx_data_sz / 2)
+ {
+ comp_data_state = vp8_get_compressed_data(ctx->cpi,
+ &lib_flags,
+ &size,
+ cx_data,
+ cx_data_end,
+ &dst_time_stamp,
+ &dst_end_time_stamp,
+ !img);
+
+ if(comp_data_state == VPX_CODEC_CORRUPT_FRAME)
+ return VPX_CODEC_CORRUPT_FRAME;
+ else if(comp_data_state == -1)
+ break;
+
+ if (size)
+ {
+ vpx_codec_pts_t round, delta;
+ vpx_codec_cx_pkt_t pkt;
+ VP8_COMP *cpi = (VP8_COMP *)ctx->cpi;
+
+ /* Add the frame packet to the list of returned packets. */
+ round = (vpx_codec_pts_t)1000000
+ * ctx->cfg.g_timebase.num / 2 - 1;
+ delta = (dst_end_time_stamp - dst_time_stamp);
+ pkt.kind = VPX_CODEC_CX_FRAME_PKT;
+ pkt.data.frame.pts =
+ (dst_time_stamp * ctx->cfg.g_timebase.den + round)
+ / ctx->cfg.g_timebase.num / 10000000;
+ pkt.data.frame.duration = (unsigned long)
+ ((delta * ctx->cfg.g_timebase.den + round)
+ / ctx->cfg.g_timebase.num / 10000000);
+ pkt.data.frame.flags = lib_flags << 16;
+
+ if (lib_flags & FRAMEFLAGS_KEY)
+ pkt.data.frame.flags |= VPX_FRAME_IS_KEY;
+
+ if (!cpi->common.show_frame)
+ {
+ pkt.data.frame.flags |= VPX_FRAME_IS_INVISIBLE;
+
+ /* This timestamp should be as close as possible to the
+ * prior PTS so that if a decoder uses pts to schedule when
+ * to do this, we start right after last frame was decoded.
+ * Invisible frames have no duration.
+ */
+ pkt.data.frame.pts = ((cpi->last_time_stamp_seen
+ * ctx->cfg.g_timebase.den + round)
+ / ctx->cfg.g_timebase.num / 10000000) + 1;
+ pkt.data.frame.duration = 0;
+ }
+
+ if (cpi->droppable)
+ pkt.data.frame.flags |= VPX_FRAME_IS_DROPPABLE;
+
+ if (cpi->output_partition)
+ {
+ int i;
+ const int num_partitions =
+ (1 << cpi->common.multi_token_partition) + 1;
+
+ pkt.data.frame.flags |= VPX_FRAME_IS_FRAGMENT;
+
+ for (i = 0; i < num_partitions; ++i)
+ {
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ pkt.data.frame.buf = cpi->partition_d[i];
+#else
+ pkt.data.frame.buf = cx_data;
+ cx_data += cpi->partition_sz[i];
+ cx_data_sz -= cpi->partition_sz[i];
+#endif
+ pkt.data.frame.sz = cpi->partition_sz[i];
+ pkt.data.frame.partition_id = i;
+ /* don't set the fragment bit for the last partition */
+ if (i == (num_partitions - 1))
+ pkt.data.frame.flags &= ~VPX_FRAME_IS_FRAGMENT;
+ vpx_codec_pkt_list_add(&ctx->pkt_list.head, &pkt);
+ }
+#if CONFIG_REALTIME_ONLY & CONFIG_ONTHEFLY_BITPACKING
+ /* In lagged mode the encoder can buffer multiple frames.
+ * We don't want this in partitioned output because
+ * partitions are spread all over the output buffer.
+ * So, force an exit!
+ */
+ cx_data_sz -= ctx->cx_data_sz / 2;
+#endif
+ }
+ else
+ {
+ pkt.data.frame.buf = cx_data;
+ pkt.data.frame.sz = size;
+ pkt.data.frame.partition_id = -1;
+ vpx_codec_pkt_list_add(&ctx->pkt_list.head, &pkt);
+ cx_data += size;
+ cx_data_sz -= size;
+ }
+ }
+ }
+ }
+
+ return res;
+}
+
+
+static const vpx_codec_cx_pkt_t *vp8e_get_cxdata(vpx_codec_alg_priv_t *ctx,
+ vpx_codec_iter_t *iter)
+{
+ return vpx_codec_pkt_list_get(&ctx->pkt_list.head, iter);
+}
+
+static vpx_codec_err_t vp8e_set_reference(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+ vpx_ref_frame_t *data = va_arg(args, vpx_ref_frame_t *);
+
+ if (data)
+ {
+ vpx_ref_frame_t *frame = (vpx_ref_frame_t *)data;
+ YV12_BUFFER_CONFIG sd;
+
+ image2yuvconfig(&frame->img, &sd);
+ vp8_set_reference(ctx->cpi, frame->frame_type, &sd);
+ return VPX_CODEC_OK;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+
+}
+
+static vpx_codec_err_t vp8e_get_reference(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+
+ vpx_ref_frame_t *data = va_arg(args, vpx_ref_frame_t *);
+
+ if (data)
+ {
+ vpx_ref_frame_t *frame = (vpx_ref_frame_t *)data;
+ YV12_BUFFER_CONFIG sd;
+
+ image2yuvconfig(&frame->img, &sd);
+ vp8_get_reference(ctx->cpi, frame->frame_type, &sd);
+ return VPX_CODEC_OK;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+}
+
+static vpx_codec_err_t vp8e_set_previewpp(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+#if CONFIG_POSTPROC
+ vp8_postproc_cfg_t *data = va_arg(args, vp8_postproc_cfg_t *);
+ (void)ctr_id;
+
+ if (data)
+ {
+ ctx->preview_ppcfg = *((vp8_postproc_cfg_t *)data);
+ return VPX_CODEC_OK;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+#else
+ (void)ctx;
+ (void)ctr_id;
+ (void)args;
+ return VPX_CODEC_INCAPABLE;
+#endif
+}
+
+
+static vpx_image_t *vp8e_get_preview(vpx_codec_alg_priv_t *ctx)
+{
+
+ YV12_BUFFER_CONFIG sd;
+ vp8_ppflags_t flags = {0};
+
+ if (ctx->preview_ppcfg.post_proc_flag)
+ {
+ flags.post_proc_flag = ctx->preview_ppcfg.post_proc_flag;
+ flags.deblocking_level = ctx->preview_ppcfg.deblocking_level;
+ flags.noise_level = ctx->preview_ppcfg.noise_level;
+ }
+
+ if (0 == vp8_get_preview_raw_frame(ctx->cpi, &sd, &flags))
+ {
+
+ /*
+ vpx_img_wrap(&ctx->preview_img, VPX_IMG_FMT_YV12,
+ sd.y_width + 2*VP8BORDERINPIXELS,
+ sd.y_height + 2*VP8BORDERINPIXELS,
+ 1,
+ sd.buffer_alloc);
+ vpx_img_set_rect(&ctx->preview_img,
+ VP8BORDERINPIXELS, VP8BORDERINPIXELS,
+ sd.y_width, sd.y_height);
+ */
+
+ ctx->preview_img.bps = 12;
+ ctx->preview_img.planes[VPX_PLANE_Y] = sd.y_buffer;
+ ctx->preview_img.planes[VPX_PLANE_U] = sd.u_buffer;
+ ctx->preview_img.planes[VPX_PLANE_V] = sd.v_buffer;
+
+ if (sd.clrtype == REG_YUV)
+ ctx->preview_img.fmt = VPX_IMG_FMT_I420;
+ else
+ ctx->preview_img.fmt = VPX_IMG_FMT_VPXI420;
+
+ ctx->preview_img.x_chroma_shift = 1;
+ ctx->preview_img.y_chroma_shift = 1;
+
+ ctx->preview_img.d_w = sd.y_width;
+ ctx->preview_img.d_h = sd.y_height;
+ ctx->preview_img.stride[VPX_PLANE_Y] = sd.y_stride;
+ ctx->preview_img.stride[VPX_PLANE_U] = sd.uv_stride;
+ ctx->preview_img.stride[VPX_PLANE_V] = sd.uv_stride;
+ ctx->preview_img.w = sd.y_width;
+ ctx->preview_img.h = sd.y_height;
+
+ return &ctx->preview_img;
+ }
+ else
+ return NULL;
+}
+
+static vpx_codec_err_t vp8e_update_entropy(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+ int update = va_arg(args, int);
+ vp8_update_entropy(ctx->cpi, update);
+ return VPX_CODEC_OK;
+
+}
+
+static vpx_codec_err_t vp8e_update_reference(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+ int update = va_arg(args, int);
+ vp8_update_reference(ctx->cpi, update);
+ return VPX_CODEC_OK;
+}
+
+static vpx_codec_err_t vp8e_use_reference(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+ int reference_flag = va_arg(args, int);
+ vp8_use_as_reference(ctx->cpi, reference_flag);
+ return VPX_CODEC_OK;
+}
+
+static vpx_codec_err_t vp8e_set_roi_map(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+ vpx_roi_map_t *data = va_arg(args, vpx_roi_map_t *);
+
+ if (data)
+ {
+ vpx_roi_map_t *roi = (vpx_roi_map_t *)data;
+
+ if (!vp8_set_roimap(ctx->cpi, roi->roi_map, roi->rows, roi->cols, roi->delta_q, roi->delta_lf, roi->static_threshold))
+ return VPX_CODEC_OK;
+ else
+ return VPX_CODEC_INVALID_PARAM;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+}
+
+
+static vpx_codec_err_t vp8e_set_activemap(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+ vpx_active_map_t *data = va_arg(args, vpx_active_map_t *);
+
+ if (data)
+ {
+
+ vpx_active_map_t *map = (vpx_active_map_t *)data;
+
+ if (!vp8_set_active_map(ctx->cpi, map->active_map, map->rows, map->cols))
+ return VPX_CODEC_OK;
+ else
+ return VPX_CODEC_INVALID_PARAM;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+}
+
+static vpx_codec_err_t vp8e_set_scalemode(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+
+ vpx_scaling_mode_t *data = va_arg(args, vpx_scaling_mode_t *);
+
+ if (data)
+ {
+ int res;
+ vpx_scaling_mode_t scalemode = *(vpx_scaling_mode_t *)data ;
+ res = vp8_set_internal_size(ctx->cpi, scalemode.h_scaling_mode, scalemode.v_scaling_mode);
+
+ if (!res)
+ {
+ /*force next frame a key frame to effect scaling mode */
+ ctx->next_frame_flag |= FRAMEFLAGS_KEY;
+ return VPX_CODEC_OK;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+}
+
+
+static vpx_codec_ctrl_fn_map_t vp8e_ctf_maps[] =
+{
+ {VP8_SET_REFERENCE, vp8e_set_reference},
+ {VP8_COPY_REFERENCE, vp8e_get_reference},
+ {VP8_SET_POSTPROC, vp8e_set_previewpp},
+ {VP8E_UPD_ENTROPY, vp8e_update_entropy},
+ {VP8E_UPD_REFERENCE, vp8e_update_reference},
+ {VP8E_USE_REFERENCE, vp8e_use_reference},
+ {VP8E_SET_ROI_MAP, vp8e_set_roi_map},
+ {VP8E_SET_ACTIVEMAP, vp8e_set_activemap},
+ {VP8E_SET_SCALEMODE, vp8e_set_scalemode},
+ {VP8E_SET_CPUUSED, set_param},
+ {VP8E_SET_NOISE_SENSITIVITY, set_param},
+ {VP8E_SET_ENABLEAUTOALTREF, set_param},
+ {VP8E_SET_SHARPNESS, set_param},
+ {VP8E_SET_STATIC_THRESHOLD, set_param},
+ {VP8E_SET_TOKEN_PARTITIONS, set_param},
+ {VP8E_GET_LAST_QUANTIZER, get_param},
+ {VP8E_GET_LAST_QUANTIZER_64, get_param},
+ {VP8E_SET_ARNR_MAXFRAMES, set_param},
+ {VP8E_SET_ARNR_STRENGTH , set_param},
+ {VP8E_SET_ARNR_TYPE , set_param},
+ {VP8E_SET_TUNING, set_param},
+ {VP8E_SET_CQ_LEVEL, set_param},
+ {VP8E_SET_MAX_INTRA_BITRATE_PCT, set_param},
+ { -1, NULL},
+};
+
+static vpx_codec_enc_cfg_map_t vp8e_usage_cfg_map[] =
+{
+ {
+ 0,
+ {
+ 0, /* g_usage */
+ 0, /* g_threads */
+ 0, /* g_profile */
+
+ 320, /* g_width */
+ 240, /* g_height */
+ {1, 30}, /* g_timebase */
+
+ 0, /* g_error_resilient */
+
+ VPX_RC_ONE_PASS, /* g_pass */
+
+ 0, /* g_lag_in_frames */
+
+ 0, /* rc_dropframe_thresh */
+ 0, /* rc_resize_allowed */
+ 60, /* rc_resize_down_thresold */
+ 30, /* rc_resize_up_thresold */
+
+ VPX_VBR, /* rc_end_usage */
+#if VPX_ENCODER_ABI_VERSION > (1 + VPX_CODEC_ABI_VERSION)
+ {0}, /* rc_twopass_stats_in */
+#endif
+ 256, /* rc_target_bandwidth */
+ 4, /* rc_min_quantizer */
+ 63, /* rc_max_quantizer */
+ 100, /* rc_undershoot_pct */
+ 100, /* rc_overshoot_pct */
+
+ 6000, /* rc_max_buffer_size */
+ 4000, /* rc_buffer_initial_size; */
+ 5000, /* rc_buffer_optimal_size; */
+
+ 50, /* rc_two_pass_vbrbias */
+ 0, /* rc_two_pass_vbrmin_section */
+ 400, /* rc_two_pass_vbrmax_section */
+
+ /* keyframing settings (kf) */
+ VPX_KF_AUTO, /* g_kfmode*/
+ 0, /* kf_min_dist */
+ 128, /* kf_max_dist */
+
+#if VPX_ENCODER_ABI_VERSION == (1 + VPX_CODEC_ABI_VERSION)
+ 1, /* g_delete_first_pass_file */
+ "vp8.fpf" /* first pass filename */
+#endif
+
+ 1, /* ts_number_layers */
+ {0}, /* ts_target_bitrate */
+ {0}, /* ts_rate_decimator */
+ 0, /* ts_periodicity */
+ {0}, /* ts_layer_id */
+ }},
+ { -1, {NOT_IMPLEMENTED}}
+};
+
+
+#ifndef VERSION_STRING
+#define VERSION_STRING
+#endif
+CODEC_INTERFACE(vpx_codec_vp8_cx) =
+{
+ "WebM Project VP8 Encoder" VERSION_STRING,
+ VPX_CODEC_INTERNAL_ABI_VERSION,
+ VPX_CODEC_CAP_ENCODER | VPX_CODEC_CAP_PSNR |
+ VPX_CODEC_CAP_OUTPUT_PARTITION,
+ /* vpx_codec_caps_t caps; */
+ vp8e_init, /* vpx_codec_init_fn_t init; */
+ vp8e_destroy, /* vpx_codec_destroy_fn_t destroy; */
+ vp8e_ctf_maps, /* vpx_codec_ctrl_fn_map_t *ctrl_maps; */
+ NOT_IMPLEMENTED, /* vpx_codec_get_mmap_fn_t get_mmap; */
+ NOT_IMPLEMENTED, /* vpx_codec_set_mmap_fn_t set_mmap; */
+ {
+ NOT_IMPLEMENTED, /* vpx_codec_peek_si_fn_t peek_si; */
+ NOT_IMPLEMENTED, /* vpx_codec_get_si_fn_t get_si; */
+ NOT_IMPLEMENTED, /* vpx_codec_decode_fn_t decode; */
+ NOT_IMPLEMENTED, /* vpx_codec_frame_get_fn_t frame_get; */
+ },
+ {
+ vp8e_usage_cfg_map, /* vpx_codec_enc_cfg_map_t peek_si; */
+ vp8e_encode, /* vpx_codec_encode_fn_t encode; */
+ vp8e_get_cxdata, /* vpx_codec_get_cx_data_fn_t frame_get; */
+ vp8e_set_config,
+ NOT_IMPLEMENTED,
+ vp8e_get_preview,
+ vp8e_mr_alloc_mem,
+ } /* encoder functions */
+};
diff --git a/libvpx/vp8/vp8_dx_iface.c b/libvpx/vp8/vp8_dx_iface.c
new file mode 100644
index 0000000..c13d697
--- /dev/null
+++ b/libvpx/vp8/vp8_dx_iface.c
@@ -0,0 +1,902 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdlib.h>
+#include <string.h>
+#include "vpx_rtcd.h"
+#include "vpx/vpx_decoder.h"
+#include "vpx/vp8dx.h"
+#include "vpx/internal/vpx_codec_internal.h"
+#include "vpx_version.h"
+#include "common/onyxd.h"
+#include "decoder/onyxd_int.h"
+#include "common/alloccommon.h"
+#include "vpx_mem/vpx_mem.h"
+#if CONFIG_ERROR_CONCEALMENT
+#include "decoder/error_concealment.h"
+#endif
+#include "decoder/decoderthreading.h"
+
+#define VP8_CAP_POSTPROC (CONFIG_POSTPROC ? VPX_CODEC_CAP_POSTPROC : 0)
+#define VP8_CAP_ERROR_CONCEALMENT (CONFIG_ERROR_CONCEALMENT ? \
+ VPX_CODEC_CAP_ERROR_CONCEALMENT : 0)
+
+typedef vpx_codec_stream_info_t vp8_stream_info_t;
+
+/* Structures for handling memory allocations */
+typedef enum
+{
+ VP8_SEG_ALG_PRIV = 256,
+ VP8_SEG_MAX
+} mem_seg_id_t;
+#define NELEMENTS(x) ((int)(sizeof(x)/sizeof(x[0])))
+
+static unsigned long vp8_priv_sz(const vpx_codec_dec_cfg_t *si, vpx_codec_flags_t);
+
+typedef struct
+{
+ unsigned int id;
+ unsigned long sz;
+ unsigned int align;
+ unsigned int flags;
+ unsigned long(*calc_sz)(const vpx_codec_dec_cfg_t *, vpx_codec_flags_t);
+} mem_req_t;
+
+static const mem_req_t vp8_mem_req_segs[] =
+{
+ {VP8_SEG_ALG_PRIV, 0, 8, VPX_CODEC_MEM_ZERO, vp8_priv_sz},
+ {VP8_SEG_MAX, 0, 0, 0, NULL}
+};
+
+struct vpx_codec_alg_priv
+{
+ vpx_codec_priv_t base;
+ vpx_codec_mmap_t mmaps[NELEMENTS(vp8_mem_req_segs)-1];
+ vpx_codec_dec_cfg_t cfg;
+ vp8_stream_info_t si;
+ int defer_alloc;
+ int decoder_init;
+ struct VP8D_COMP *pbi;
+ int postproc_cfg_set;
+ vp8_postproc_cfg_t postproc_cfg;
+#if CONFIG_POSTPROC_VISUALIZER
+ unsigned int dbg_postproc_flag;
+ int dbg_color_ref_frame_flag;
+ int dbg_color_mb_modes_flag;
+ int dbg_color_b_modes_flag;
+ int dbg_display_mv_flag;
+#endif
+ vpx_image_t img;
+ int img_setup;
+ void *user_priv;
+};
+
+static unsigned long vp8_priv_sz(const vpx_codec_dec_cfg_t *si, vpx_codec_flags_t flags)
+{
+ /* Although this declaration is constant, we can't use it in the requested
+ * segments list because we want to define the requested segments list
+ * before defining the private type (so that the number of memory maps is
+ * known)
+ */
+ (void)si;
+ return sizeof(vpx_codec_alg_priv_t);
+}
+
+
+static void vp8_mmap_dtor(vpx_codec_mmap_t *mmap)
+{
+ free(mmap->priv);
+}
+
+static vpx_codec_err_t vp8_mmap_alloc(vpx_codec_mmap_t *mmap)
+{
+ vpx_codec_err_t res;
+ unsigned int align;
+
+ align = mmap->align ? mmap->align - 1 : 0;
+
+ if (mmap->flags & VPX_CODEC_MEM_ZERO)
+ mmap->priv = calloc(1, mmap->sz + align);
+ else
+ mmap->priv = malloc(mmap->sz + align);
+
+ res = (mmap->priv) ? VPX_CODEC_OK : VPX_CODEC_MEM_ERROR;
+ mmap->base = (void *)((((uintptr_t)mmap->priv) + align) & ~(uintptr_t)align);
+ mmap->dtor = vp8_mmap_dtor;
+ return res;
+}
+
+static vpx_codec_err_t vp8_validate_mmaps(const vp8_stream_info_t *si,
+ const vpx_codec_mmap_t *mmaps,
+ vpx_codec_flags_t init_flags)
+{
+ int i;
+ vpx_codec_err_t res = VPX_CODEC_OK;
+
+ for (i = 0; i < NELEMENTS(vp8_mem_req_segs) - 1; i++)
+ {
+ /* Ensure the segment has been allocated */
+ if (!mmaps[i].base)
+ {
+ res = VPX_CODEC_MEM_ERROR;
+ break;
+ }
+
+ /* Verify variable size segment is big enough for the current si. */
+ if (vp8_mem_req_segs[i].calc_sz)
+ {
+ vpx_codec_dec_cfg_t cfg;
+
+ cfg.w = si->w;
+ cfg.h = si->h;
+
+ if (mmaps[i].sz < vp8_mem_req_segs[i].calc_sz(&cfg, init_flags))
+ {
+ res = VPX_CODEC_MEM_ERROR;
+ break;
+ }
+ }
+ }
+
+ return res;
+}
+
+static void vp8_init_ctx(vpx_codec_ctx_t *ctx, const vpx_codec_mmap_t *mmap)
+{
+ int i;
+
+ ctx->priv = mmap->base;
+ ctx->priv->sz = sizeof(*ctx->priv);
+ ctx->priv->iface = ctx->iface;
+ ctx->priv->alg_priv = mmap->base;
+
+ for (i = 0; i < NELEMENTS(ctx->priv->alg_priv->mmaps); i++)
+ ctx->priv->alg_priv->mmaps[i].id = vp8_mem_req_segs[i].id;
+
+ ctx->priv->alg_priv->mmaps[0] = *mmap;
+ ctx->priv->alg_priv->si.sz = sizeof(ctx->priv->alg_priv->si);
+ ctx->priv->init_flags = ctx->init_flags;
+
+ if (ctx->config.dec)
+ {
+ /* Update the reference to the config structure to an internal copy. */
+ ctx->priv->alg_priv->cfg = *ctx->config.dec;
+ ctx->config.dec = &ctx->priv->alg_priv->cfg;
+ }
+}
+
+static void *mmap_lkup(vpx_codec_alg_priv_t *ctx, unsigned int id)
+{
+ int i;
+
+ for (i = 0; i < NELEMENTS(ctx->mmaps); i++)
+ if (ctx->mmaps[i].id == id)
+ return ctx->mmaps[i].base;
+
+ return NULL;
+}
+static void vp8_finalize_mmaps(vpx_codec_alg_priv_t *ctx)
+{
+ /* nothing to clean up */
+}
+
+static vpx_codec_err_t vp8_init(vpx_codec_ctx_t *ctx,
+ vpx_codec_priv_enc_mr_cfg_t *data)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+ (void) data;
+
+ vpx_rtcd();
+
+ /* This function only allocates space for the vpx_codec_alg_priv_t
+ * structure. More memory may be required at the time the stream
+ * information becomes known.
+ */
+ if (!ctx->priv)
+ {
+ vpx_codec_mmap_t mmap;
+
+ mmap.id = vp8_mem_req_segs[0].id;
+ mmap.sz = sizeof(vpx_codec_alg_priv_t);
+ mmap.align = vp8_mem_req_segs[0].align;
+ mmap.flags = vp8_mem_req_segs[0].flags;
+
+ res = vp8_mmap_alloc(&mmap);
+
+ if (!res)
+ {
+ vp8_init_ctx(ctx, &mmap);
+
+ ctx->priv->alg_priv->defer_alloc = 1;
+ /*post processing level initialized to do nothing */
+ }
+ }
+
+ return res;
+}
+
+static vpx_codec_err_t vp8_destroy(vpx_codec_alg_priv_t *ctx)
+{
+ int i;
+
+ vp8dx_remove_decompressor(ctx->pbi);
+
+ for (i = NELEMENTS(ctx->mmaps) - 1; i >= 0; i--)
+ {
+ if (ctx->mmaps[i].dtor)
+ ctx->mmaps[i].dtor(&ctx->mmaps[i]);
+ }
+
+ return VPX_CODEC_OK;
+}
+
+static vpx_codec_err_t vp8_peek_si(const uint8_t *data,
+ unsigned int data_sz,
+ vpx_codec_stream_info_t *si)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+
+ if(data + data_sz <= data)
+ res = VPX_CODEC_INVALID_PARAM;
+ else
+ {
+ /* Parse uncompresssed part of key frame header.
+ * 3 bytes:- including version, frame type and an offset
+ * 3 bytes:- sync code (0x9d, 0x01, 0x2a)
+ * 4 bytes:- including image width and height in the lowest 14 bits
+ * of each 2-byte value.
+ */
+ si->is_kf = 0;
+
+ if (data_sz >= 10 && !(data[0] & 0x01)) /* I-Frame */
+ {
+ const uint8_t *c = data + 3;
+ si->is_kf = 1;
+
+ /* vet via sync code */
+ if (c[0] != 0x9d || c[1] != 0x01 || c[2] != 0x2a)
+ res = VPX_CODEC_UNSUP_BITSTREAM;
+
+ si->w = (c[3] | (c[4] << 8)) & 0x3fff;
+ si->h = (c[5] | (c[6] << 8)) & 0x3fff;
+
+ /*printf("w=%d, h=%d\n", si->w, si->h);*/
+ if (!(si->h | si->w))
+ res = VPX_CODEC_UNSUP_BITSTREAM;
+ }
+ else
+ res = VPX_CODEC_UNSUP_BITSTREAM;
+ }
+
+ return res;
+
+}
+
+static vpx_codec_err_t vp8_get_si(vpx_codec_alg_priv_t *ctx,
+ vpx_codec_stream_info_t *si)
+{
+
+ unsigned int sz;
+
+ if (si->sz >= sizeof(vp8_stream_info_t))
+ sz = sizeof(vp8_stream_info_t);
+ else
+ sz = sizeof(vpx_codec_stream_info_t);
+
+ memcpy(si, &ctx->si, sz);
+ si->sz = sz;
+
+ return VPX_CODEC_OK;
+}
+
+
+static vpx_codec_err_t
+update_error_state(vpx_codec_alg_priv_t *ctx,
+ const struct vpx_internal_error_info *error)
+{
+ vpx_codec_err_t res;
+
+ if ((res = error->error_code))
+ ctx->base.err_detail = error->has_detail
+ ? error->detail
+ : NULL;
+
+ return res;
+}
+
+static void yuvconfig2image(vpx_image_t *img,
+ const YV12_BUFFER_CONFIG *yv12,
+ void *user_priv)
+{
+ /** vpx_img_wrap() doesn't allow specifying independent strides for
+ * the Y, U, and V planes, nor other alignment adjustments that
+ * might be representable by a YV12_BUFFER_CONFIG, so we just
+ * initialize all the fields.*/
+ img->fmt = yv12->clrtype == REG_YUV ?
+ VPX_IMG_FMT_I420 : VPX_IMG_FMT_VPXI420;
+ img->w = yv12->y_stride;
+ img->h = (yv12->y_height + 2 * VP8BORDERINPIXELS + 15) & ~15;
+ img->d_w = yv12->y_width;
+ img->d_h = yv12->y_height;
+ img->x_chroma_shift = 1;
+ img->y_chroma_shift = 1;
+ img->planes[VPX_PLANE_Y] = yv12->y_buffer;
+ img->planes[VPX_PLANE_U] = yv12->u_buffer;
+ img->planes[VPX_PLANE_V] = yv12->v_buffer;
+ img->planes[VPX_PLANE_ALPHA] = NULL;
+ img->stride[VPX_PLANE_Y] = yv12->y_stride;
+ img->stride[VPX_PLANE_U] = yv12->uv_stride;
+ img->stride[VPX_PLANE_V] = yv12->uv_stride;
+ img->stride[VPX_PLANE_ALPHA] = yv12->y_stride;
+ img->bps = 12;
+ img->user_priv = user_priv;
+ img->img_data = yv12->buffer_alloc;
+ img->img_data_owner = 0;
+ img->self_allocd = 0;
+}
+
+static vpx_codec_err_t vp8_decode(vpx_codec_alg_priv_t *ctx,
+ const uint8_t *data,
+ unsigned int data_sz,
+ void *user_priv,
+ long deadline)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+ unsigned int resolution_change = 0;
+ unsigned int w, h;
+
+ /* Determine the stream parameters. Note that we rely on peek_si to
+ * validate that we have a buffer that does not wrap around the top
+ * of the heap.
+ */
+ w = ctx->si.w;
+ h = ctx->si.h;
+
+ res = ctx->base.iface->dec.peek_si(data, data_sz, &ctx->si);
+
+ if((res == VPX_CODEC_UNSUP_BITSTREAM) && !ctx->si.is_kf)
+ {
+ /* the peek function returns an error for non keyframes, however for
+ * this case, it is not an error */
+ res = VPX_CODEC_OK;
+ }
+
+ if(!ctx->decoder_init && !ctx->si.is_kf)
+ res = VPX_CODEC_UNSUP_BITSTREAM;
+
+ if ((ctx->si.h != h) || (ctx->si.w != w))
+ resolution_change = 1;
+
+ /* Perform deferred allocations, if required */
+ if (!res && ctx->defer_alloc)
+ {
+ int i;
+
+ for (i = 1; !res && i < NELEMENTS(ctx->mmaps); i++)
+ {
+ vpx_codec_dec_cfg_t cfg;
+
+ cfg.w = ctx->si.w;
+ cfg.h = ctx->si.h;
+ ctx->mmaps[i].id = vp8_mem_req_segs[i].id;
+ ctx->mmaps[i].sz = vp8_mem_req_segs[i].sz;
+ ctx->mmaps[i].align = vp8_mem_req_segs[i].align;
+ ctx->mmaps[i].flags = vp8_mem_req_segs[i].flags;
+
+ if (!ctx->mmaps[i].sz)
+ ctx->mmaps[i].sz = vp8_mem_req_segs[i].calc_sz(&cfg,
+ ctx->base.init_flags);
+
+ res = vp8_mmap_alloc(&ctx->mmaps[i]);
+ }
+
+ if (!res)
+ vp8_finalize_mmaps(ctx);
+
+ ctx->defer_alloc = 0;
+ }
+
+ /* Initialize the decoder instance on the first frame*/
+ if (!res && !ctx->decoder_init)
+ {
+ res = vp8_validate_mmaps(&ctx->si, ctx->mmaps, ctx->base.init_flags);
+
+ if (!res)
+ {
+ VP8D_CONFIG oxcf;
+ struct VP8D_COMP* optr;
+
+ oxcf.Width = ctx->si.w;
+ oxcf.Height = ctx->si.h;
+ oxcf.Version = 9;
+ oxcf.postprocess = 0;
+ oxcf.max_threads = ctx->cfg.threads;
+ oxcf.error_concealment =
+ (ctx->base.init_flags & VPX_CODEC_USE_ERROR_CONCEALMENT);
+ oxcf.input_fragments =
+ (ctx->base.init_flags & VPX_CODEC_USE_INPUT_FRAGMENTS);
+
+ optr = vp8dx_create_decompressor(&oxcf);
+
+ /* If postprocessing was enabled by the application and a
+ * configuration has not been provided, default it.
+ */
+ if (!ctx->postproc_cfg_set
+ && (ctx->base.init_flags & VPX_CODEC_USE_POSTPROC))
+ {
+ ctx->postproc_cfg.post_proc_flag =
+ VP8_DEBLOCK | VP8_DEMACROBLOCK | VP8_MFQE;
+ ctx->postproc_cfg.deblocking_level = 4;
+ ctx->postproc_cfg.noise_level = 0;
+ }
+
+ if (!optr)
+ res = VPX_CODEC_ERROR;
+ else
+ ctx->pbi = optr;
+ }
+
+ ctx->decoder_init = 1;
+ }
+
+ if (!res && ctx->pbi)
+ {
+ if(resolution_change)
+ {
+ VP8D_COMP *pbi = ctx->pbi;
+ VP8_COMMON *const pc = & pbi->common;
+ MACROBLOCKD *const xd = & pbi->mb;
+#if CONFIG_MULTITHREAD
+ int i;
+#endif
+ pc->Width = ctx->si.w;
+ pc->Height = ctx->si.h;
+ {
+ int prev_mb_rows = pc->mb_rows;
+
+ if (setjmp(pbi->common.error.jmp))
+ {
+ pbi->common.error.setjmp = 0;
+ /* same return value as used in vp8dx_receive_compressed_data */
+ return -1;
+ }
+
+ pbi->common.error.setjmp = 1;
+
+ if (pc->Width <= 0)
+ {
+ pc->Width = w;
+ vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
+ "Invalid frame width");
+ }
+
+ if (pc->Height <= 0)
+ {
+ pc->Height = h;
+ vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
+ "Invalid frame height");
+ }
+
+ if (vp8_alloc_frame_buffers(pc, pc->Width, pc->Height))
+ vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate frame buffers");
+
+ xd->pre = pc->yv12_fb[pc->lst_fb_idx];
+ xd->dst = pc->yv12_fb[pc->new_fb_idx];
+
+#if CONFIG_MULTITHREAD
+ for (i = 0; i < pbi->allocated_decoding_thread_count; i++)
+ {
+ pbi->mb_row_di[i].mbd.dst = pc->yv12_fb[pc->new_fb_idx];
+ vp8_build_block_doffsets(&pbi->mb_row_di[i].mbd);
+ }
+#endif
+ vp8_build_block_doffsets(&pbi->mb);
+
+ /* allocate memory for last frame MODE_INFO array */
+#if CONFIG_ERROR_CONCEALMENT
+
+ if (pbi->ec_enabled)
+ {
+ /* old prev_mip was released by vp8_de_alloc_frame_buffers()
+ * called in vp8_alloc_frame_buffers() */
+ pc->prev_mip = vpx_calloc(
+ (pc->mb_cols + 1) * (pc->mb_rows + 1),
+ sizeof(MODE_INFO));
+
+ if (!pc->prev_mip)
+ {
+ vp8_de_alloc_frame_buffers(pc);
+ vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate"
+ "last frame MODE_INFO array");
+ }
+
+ pc->prev_mi = pc->prev_mip + pc->mode_info_stride + 1;
+
+ if (vp8_alloc_overlap_lists(pbi))
+ vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
+ "Failed to allocate overlap lists "
+ "for error concealment");
+ }
+
+#endif
+
+#if CONFIG_MULTITHREAD
+ if (pbi->b_multithreaded_rd)
+ vp8mt_alloc_temp_buffers(pbi, pc->Width, prev_mb_rows);
+#else
+ (void)prev_mb_rows;
+#endif
+ }
+
+ pbi->common.error.setjmp = 0;
+
+ /* required to get past the first get_free_fb() call */
+ ctx->pbi->common.fb_idx_ref_cnt[0] = 0;
+ }
+
+ ctx->user_priv = user_priv;
+ if (vp8dx_receive_compressed_data(ctx->pbi, data_sz, data, deadline))
+ {
+ VP8D_COMP *pbi = (VP8D_COMP *)ctx->pbi;
+ res = update_error_state(ctx, &pbi->common.error);
+ }
+ }
+
+ return res;
+}
+
+static vpx_image_t *vp8_get_frame(vpx_codec_alg_priv_t *ctx,
+ vpx_codec_iter_t *iter)
+{
+ vpx_image_t *img = NULL;
+
+ /* iter acts as a flip flop, so an image is only returned on the first
+ * call to get_frame.
+ */
+ if (!(*iter))
+ {
+ YV12_BUFFER_CONFIG sd;
+ int64_t time_stamp = 0, time_end_stamp = 0;
+ vp8_ppflags_t flags = {0};
+
+ if (ctx->base.init_flags & VPX_CODEC_USE_POSTPROC)
+ {
+ flags.post_proc_flag= ctx->postproc_cfg.post_proc_flag
+#if CONFIG_POSTPROC_VISUALIZER
+
+ | ((ctx->dbg_color_ref_frame_flag != 0) ? VP8D_DEBUG_CLR_FRM_REF_BLKS : 0)
+ | ((ctx->dbg_color_mb_modes_flag != 0) ? VP8D_DEBUG_CLR_BLK_MODES : 0)
+ | ((ctx->dbg_color_b_modes_flag != 0) ? VP8D_DEBUG_CLR_BLK_MODES : 0)
+ | ((ctx->dbg_display_mv_flag != 0) ? VP8D_DEBUG_DRAW_MV : 0)
+#endif
+ ;
+ flags.deblocking_level = ctx->postproc_cfg.deblocking_level;
+ flags.noise_level = ctx->postproc_cfg.noise_level;
+#if CONFIG_POSTPROC_VISUALIZER
+ flags.display_ref_frame_flag= ctx->dbg_color_ref_frame_flag;
+ flags.display_mb_modes_flag = ctx->dbg_color_mb_modes_flag;
+ flags.display_b_modes_flag = ctx->dbg_color_b_modes_flag;
+ flags.display_mv_flag = ctx->dbg_display_mv_flag;
+#endif
+ }
+
+ if (0 == vp8dx_get_raw_frame(ctx->pbi, &sd, &time_stamp, &time_end_stamp, &flags))
+ {
+ yuvconfig2image(&ctx->img, &sd, ctx->user_priv);
+
+ img = &ctx->img;
+ *iter = img;
+ }
+ }
+
+ return img;
+}
+
+
+static
+vpx_codec_err_t vp8_xma_get_mmap(const vpx_codec_ctx_t *ctx,
+ vpx_codec_mmap_t *mmap,
+ vpx_codec_iter_t *iter)
+{
+ vpx_codec_err_t res;
+ const mem_req_t *seg_iter = *iter;
+
+ /* Get address of next segment request */
+ do
+ {
+ if (!seg_iter)
+ seg_iter = vp8_mem_req_segs;
+ else if (seg_iter->id != VP8_SEG_MAX)
+ seg_iter++;
+
+ *iter = (vpx_codec_iter_t)seg_iter;
+
+ if (seg_iter->id != VP8_SEG_MAX)
+ {
+ mmap->id = seg_iter->id;
+ mmap->sz = seg_iter->sz;
+ mmap->align = seg_iter->align;
+ mmap->flags = seg_iter->flags;
+
+ if (!seg_iter->sz)
+ mmap->sz = seg_iter->calc_sz(ctx->config.dec, ctx->init_flags);
+
+ res = VPX_CODEC_OK;
+ }
+ else
+ res = VPX_CODEC_LIST_END;
+ }
+ while (!mmap->sz && res != VPX_CODEC_LIST_END);
+
+ return res;
+}
+
+static vpx_codec_err_t vp8_xma_set_mmap(vpx_codec_ctx_t *ctx,
+ const vpx_codec_mmap_t *mmap)
+{
+ vpx_codec_err_t res = VPX_CODEC_MEM_ERROR;
+ int i, done;
+
+ if (!ctx->priv)
+ {
+ if (mmap->id == VP8_SEG_ALG_PRIV)
+ {
+ if (!ctx->priv)
+ {
+ vp8_init_ctx(ctx, mmap);
+ res = VPX_CODEC_OK;
+ }
+ }
+ }
+
+ done = 1;
+
+ if (!res && ctx->priv->alg_priv)
+ {
+ for (i = 0; i < NELEMENTS(ctx->priv->alg_priv->mmaps); i++)
+ {
+ if (ctx->priv->alg_priv->mmaps[i].id == mmap->id)
+ if (!ctx->priv->alg_priv->mmaps[i].base)
+ {
+ ctx->priv->alg_priv->mmaps[i] = *mmap;
+ res = VPX_CODEC_OK;
+ }
+
+ done &= (ctx->priv->alg_priv->mmaps[i].base != NULL);
+ }
+ }
+
+ if (done && !res)
+ {
+ vp8_finalize_mmaps(ctx->priv->alg_priv);
+ res = ctx->iface->init(ctx, NULL);
+ }
+
+ return res;
+}
+
+static vpx_codec_err_t image2yuvconfig(const vpx_image_t *img,
+ YV12_BUFFER_CONFIG *yv12)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+ yv12->y_buffer = img->planes[VPX_PLANE_Y];
+ yv12->u_buffer = img->planes[VPX_PLANE_U];
+ yv12->v_buffer = img->planes[VPX_PLANE_V];
+
+ yv12->y_width = img->d_w;
+ yv12->y_height = img->d_h;
+ yv12->uv_width = yv12->y_width / 2;
+ yv12->uv_height = yv12->y_height / 2;
+
+ yv12->y_stride = img->stride[VPX_PLANE_Y];
+ yv12->uv_stride = img->stride[VPX_PLANE_U];
+
+ yv12->border = (img->stride[VPX_PLANE_Y] - img->d_w) / 2;
+ yv12->clrtype = (img->fmt == VPX_IMG_FMT_VPXI420 || img->fmt == VPX_IMG_FMT_VPXYV12);
+
+ return res;
+}
+
+
+static vpx_codec_err_t vp8_set_reference(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+
+ vpx_ref_frame_t *data = va_arg(args, vpx_ref_frame_t *);
+
+ if (data)
+ {
+ vpx_ref_frame_t *frame = (vpx_ref_frame_t *)data;
+ YV12_BUFFER_CONFIG sd;
+
+ image2yuvconfig(&frame->img, &sd);
+
+ return vp8dx_set_reference(ctx->pbi, frame->frame_type, &sd);
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+
+}
+
+static vpx_codec_err_t vp8_get_reference(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+
+ vpx_ref_frame_t *data = va_arg(args, vpx_ref_frame_t *);
+
+ if (data)
+ {
+ vpx_ref_frame_t *frame = (vpx_ref_frame_t *)data;
+ YV12_BUFFER_CONFIG sd;
+
+ image2yuvconfig(&frame->img, &sd);
+
+ return vp8dx_get_reference(ctx->pbi, frame->frame_type, &sd);
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+
+}
+
+static vpx_codec_err_t vp8_set_postproc(vpx_codec_alg_priv_t *ctx,
+ int ctr_id,
+ va_list args)
+{
+#if CONFIG_POSTPROC
+ vp8_postproc_cfg_t *data = va_arg(args, vp8_postproc_cfg_t *);
+
+ if (data)
+ {
+ ctx->postproc_cfg_set = 1;
+ ctx->postproc_cfg = *((vp8_postproc_cfg_t *)data);
+ return VPX_CODEC_OK;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+
+#else
+ return VPX_CODEC_INCAPABLE;
+#endif
+}
+
+static vpx_codec_err_t vp8_set_dbg_options(vpx_codec_alg_priv_t *ctx,
+ int ctrl_id,
+ va_list args)
+{
+#if CONFIG_POSTPROC_VISUALIZER && CONFIG_POSTPROC
+ int data = va_arg(args, int);
+
+#define MAP(id, var) case id: var = data; break;
+
+ switch (ctrl_id)
+ {
+ MAP (VP8_SET_DBG_COLOR_REF_FRAME, ctx->dbg_color_ref_frame_flag);
+ MAP (VP8_SET_DBG_COLOR_MB_MODES, ctx->dbg_color_mb_modes_flag);
+ MAP (VP8_SET_DBG_COLOR_B_MODES, ctx->dbg_color_b_modes_flag);
+ MAP (VP8_SET_DBG_DISPLAY_MV, ctx->dbg_display_mv_flag);
+ }
+
+ return VPX_CODEC_OK;
+#else
+ return VPX_CODEC_INCAPABLE;
+#endif
+}
+
+static vpx_codec_err_t vp8_get_last_ref_updates(vpx_codec_alg_priv_t *ctx,
+ int ctrl_id,
+ va_list args)
+{
+ int *update_info = va_arg(args, int *);
+ VP8D_COMP *pbi = (VP8D_COMP *)ctx->pbi;
+
+ if (update_info)
+ {
+ *update_info = pbi->common.refresh_alt_ref_frame * (int) VP8_ALTR_FRAME
+ + pbi->common.refresh_golden_frame * (int) VP8_GOLD_FRAME
+ + pbi->common.refresh_last_frame * (int) VP8_LAST_FRAME;
+
+ return VPX_CODEC_OK;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+}
+
+extern int vp8dx_references_buffer( VP8_COMMON *oci, int ref_frame );
+static vpx_codec_err_t vp8_get_last_ref_frame(vpx_codec_alg_priv_t *ctx,
+ int ctrl_id,
+ va_list args)
+{
+ int *ref_info = va_arg(args, int *);
+ VP8D_COMP *pbi = (VP8D_COMP *)ctx->pbi;
+ VP8_COMMON *oci = &pbi->common;
+
+ if (ref_info)
+ {
+ *ref_info =
+ (vp8dx_references_buffer( oci, ALTREF_FRAME )?VP8_ALTR_FRAME:0) |
+ (vp8dx_references_buffer( oci, GOLDEN_FRAME )?VP8_GOLD_FRAME:0) |
+ (vp8dx_references_buffer( oci, LAST_FRAME )?VP8_LAST_FRAME:0);
+
+ return VPX_CODEC_OK;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+}
+
+static vpx_codec_err_t vp8_get_frame_corrupted(vpx_codec_alg_priv_t *ctx,
+ int ctrl_id,
+ va_list args)
+{
+
+ int *corrupted = va_arg(args, int *);
+
+ if (corrupted)
+ {
+ VP8D_COMP *pbi = (VP8D_COMP *)ctx->pbi;
+ *corrupted = pbi->common.frame_to_show->corrupted;
+
+ return VPX_CODEC_OK;
+ }
+ else
+ return VPX_CODEC_INVALID_PARAM;
+
+}
+
+vpx_codec_ctrl_fn_map_t vp8_ctf_maps[] =
+{
+ {VP8_SET_REFERENCE, vp8_set_reference},
+ {VP8_COPY_REFERENCE, vp8_get_reference},
+ {VP8_SET_POSTPROC, vp8_set_postproc},
+ {VP8_SET_DBG_COLOR_REF_FRAME, vp8_set_dbg_options},
+ {VP8_SET_DBG_COLOR_MB_MODES, vp8_set_dbg_options},
+ {VP8_SET_DBG_COLOR_B_MODES, vp8_set_dbg_options},
+ {VP8_SET_DBG_DISPLAY_MV, vp8_set_dbg_options},
+ {VP8D_GET_LAST_REF_UPDATES, vp8_get_last_ref_updates},
+ {VP8D_GET_FRAME_CORRUPTED, vp8_get_frame_corrupted},
+ {VP8D_GET_LAST_REF_USED, vp8_get_last_ref_frame},
+ { -1, NULL},
+};
+
+
+#ifndef VERSION_STRING
+#define VERSION_STRING
+#endif
+CODEC_INTERFACE(vpx_codec_vp8_dx) =
+{
+ "WebM Project VP8 Decoder" VERSION_STRING,
+ VPX_CODEC_INTERNAL_ABI_VERSION,
+ VPX_CODEC_CAP_DECODER | VP8_CAP_POSTPROC | VP8_CAP_ERROR_CONCEALMENT |
+ VPX_CODEC_CAP_INPUT_FRAGMENTS,
+ /* vpx_codec_caps_t caps; */
+ vp8_init, /* vpx_codec_init_fn_t init; */
+ vp8_destroy, /* vpx_codec_destroy_fn_t destroy; */
+ vp8_ctf_maps, /* vpx_codec_ctrl_fn_map_t *ctrl_maps; */
+ vp8_xma_get_mmap, /* vpx_codec_get_mmap_fn_t get_mmap; */
+ vp8_xma_set_mmap, /* vpx_codec_set_mmap_fn_t set_mmap; */
+ {
+ vp8_peek_si, /* vpx_codec_peek_si_fn_t peek_si; */
+ vp8_get_si, /* vpx_codec_get_si_fn_t get_si; */
+ vp8_decode, /* vpx_codec_decode_fn_t decode; */
+ vp8_get_frame, /* vpx_codec_frame_get_fn_t frame_get; */
+ },
+ { /* encoder functions */
+ NOT_IMPLEMENTED,
+ NOT_IMPLEMENTED,
+ NOT_IMPLEMENTED,
+ NOT_IMPLEMENTED,
+ NOT_IMPLEMENTED,
+ NOT_IMPLEMENTED
+ }
+};
diff --git a/libvpx/vp8/vp8cx.mk b/libvpx/vp8/vp8cx.mk
new file mode 100644
index 0000000..0ae2f10
--- /dev/null
+++ b/libvpx/vp8/vp8cx.mk
@@ -0,0 +1,117 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8_common.mk
+
+VP8_CX_EXPORTS += exports_enc
+
+VP8_CX_SRCS-yes += $(VP8_COMMON_SRCS-yes)
+VP8_CX_SRCS-no += $(VP8_COMMON_SRCS-no)
+VP8_CX_SRCS_REMOVE-yes += $(VP8_COMMON_SRCS_REMOVE-yes)
+VP8_CX_SRCS_REMOVE-no += $(VP8_COMMON_SRCS_REMOVE-no)
+
+ifeq ($(ARCH_ARM),yes)
+ include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8cx_arm.mk
+endif
+
+VP8_CX_SRCS-yes += vp8cx.mk
+
+VP8_CX_SRCS-yes += vp8_cx_iface.c
+
+VP8_CX_SRCS-yes += encoder/asm_enc_offsets.c
+VP8_CX_SRCS-yes += encoder/defaultcoefcounts.h
+VP8_CX_SRCS-yes += encoder/bitstream.c
+VP8_CX_SRCS-yes += encoder/boolhuff.c
+VP8_CX_SRCS-yes += encoder/dct.c
+VP8_CX_SRCS-yes += encoder/encodeframe.c
+VP8_CX_SRCS-yes += encoder/encodeframe.h
+VP8_CX_SRCS-yes += encoder/encodeintra.c
+VP8_CX_SRCS-yes += encoder/encodemb.c
+VP8_CX_SRCS-yes += encoder/encodemv.c
+VP8_CX_SRCS-$(CONFIG_MULTITHREAD) += encoder/ethreading.c
+VP8_CX_SRCS-yes += encoder/firstpass.c
+VP8_CX_SRCS-yes += encoder/block.h
+VP8_CX_SRCS-yes += encoder/boolhuff.h
+VP8_CX_SRCS-yes += encoder/bitstream.h
+VP8_CX_SRCS-$(CONFIG_TEMPORAL_DENOISING) += encoder/denoising.h
+VP8_CX_SRCS-$(CONFIG_TEMPORAL_DENOISING) += encoder/denoising.c
+VP8_CX_SRCS-yes += encoder/encodeintra.h
+VP8_CX_SRCS-yes += encoder/encodemb.h
+VP8_CX_SRCS-yes += encoder/encodemv.h
+VP8_CX_SRCS-yes += encoder/firstpass.h
+VP8_CX_SRCS-yes += encoder/lookahead.c
+VP8_CX_SRCS-yes += encoder/lookahead.h
+VP8_CX_SRCS-yes += encoder/mcomp.h
+VP8_CX_SRCS-yes += encoder/modecosts.h
+VP8_CX_SRCS-yes += encoder/onyx_int.h
+VP8_CX_SRCS-yes += encoder/pickinter.h
+VP8_CX_SRCS-yes += encoder/psnr.h
+VP8_CX_SRCS-yes += encoder/quantize.h
+VP8_CX_SRCS-yes += encoder/ratectrl.h
+VP8_CX_SRCS-yes += encoder/rdopt.h
+VP8_CX_SRCS-yes += encoder/tokenize.h
+VP8_CX_SRCS-yes += encoder/treewriter.h
+VP8_CX_SRCS-yes += encoder/mcomp.c
+VP8_CX_SRCS-yes += encoder/modecosts.c
+VP8_CX_SRCS-yes += encoder/onyx_if.c
+VP8_CX_SRCS-yes += encoder/pickinter.c
+VP8_CX_SRCS-yes += encoder/picklpf.c
+VP8_CX_SRCS-yes += encoder/psnr.c
+VP8_CX_SRCS-yes += encoder/quantize.c
+VP8_CX_SRCS-yes += encoder/ratectrl.c
+VP8_CX_SRCS-yes += encoder/rdopt.c
+VP8_CX_SRCS-yes += encoder/segmentation.c
+VP8_CX_SRCS-yes += encoder/segmentation.h
+VP8_CX_SRCS-$(CONFIG_INTERNAL_STATS) += encoder/ssim.c
+VP8_CX_SRCS-yes += encoder/tokenize.c
+VP8_CX_SRCS-yes += encoder/dct_value_cost.h
+VP8_CX_SRCS-yes += encoder/dct_value_tokens.h
+VP8_CX_SRCS-yes += encoder/treewriter.c
+VP8_CX_SRCS-$(CONFIG_INTERNAL_STATS) += common/postproc.h
+VP8_CX_SRCS-$(CONFIG_INTERNAL_STATS) += common/postproc.c
+VP8_CX_SRCS-yes += encoder/temporal_filter.c
+VP8_CX_SRCS-$(CONFIG_MULTI_RES_ENCODING) += encoder/mr_dissim.c
+VP8_CX_SRCS-$(CONFIG_MULTI_RES_ENCODING) += encoder/mr_dissim.h
+
+ifeq ($(CONFIG_REALTIME_ONLY),yes)
+VP8_CX_SRCS_REMOVE-yes += encoder/firstpass.c
+VP8_CX_SRCS_REMOVE-yes += encoder/temporal_filter.c
+endif
+
+VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/dct_mmx.asm
+VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/subtract_mmx.asm
+VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/vp8_enc_stubs_mmx.c
+VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/dct_sse2.asm
+VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/fwalsh_sse2.asm
+VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/quantize_sse2.asm
+
+ifeq ($(CONFIG_TEMPORAL_DENOISING),yes)
+VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/denoising_sse2.c
+ifeq ($(HAVE_SSE2),yes)
+vp8/encoder/x86/denoising_sse2.c.o: CFLAGS += -msse2
+endif
+endif
+
+VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/subtract_sse2.asm
+VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/temporal_filter_apply_sse2.asm
+VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/vp8_enc_stubs_sse2.c
+VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/quantize_ssse3.asm
+VP8_CX_SRCS-$(HAVE_SSE4_1) += encoder/x86/quantize_sse4.asm
+VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/quantize_mmx.asm
+VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/encodeopt.asm
+VP8_CX_SRCS-$(ARCH_X86_64) += encoder/x86/ssim_opt.asm
+
+ifeq ($(CONFIG_REALTIME_ONLY),yes)
+VP8_CX_SRCS_REMOVE-$(HAVE_SSE2) += encoder/x86/temporal_filter_apply_sse2.asm
+endif
+
+
+VP8_CX_SRCS-yes := $(filter-out $(VP8_CX_SRCS_REMOVE-yes),$(VP8_CX_SRCS-yes))
diff --git a/libvpx/vp8/vp8cx_arm.mk b/libvpx/vp8/vp8cx_arm.mk
new file mode 100644
index 0000000..b030ee5
--- /dev/null
+++ b/libvpx/vp8/vp8cx_arm.mk
@@ -0,0 +1,44 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+VP8_CX_SRCS-$(ARCH_ARM) += vp8cx_arm.mk
+
+#File list for arm
+# encoder
+VP8_CX_SRCS-$(ARCH_ARM) += encoder/arm/dct_arm.c
+VP8_CX_SRCS-$(ARCH_ARM) += encoder/arm/quantize_arm.c
+
+#File list for edsp
+# encoder
+VP8_CX_SRCS-$(HAVE_EDSP) += encoder/arm/boolhuff_arm.c
+VP8_CX_SRCS_REMOVE-$(HAVE_EDSP) += encoder/boolhuff.c
+VP8_CX_SRCS-$(HAVE_EDSP) += encoder/arm/armv5te/boolhuff_armv5te$(ASM)
+VP8_CX_SRCS-$(HAVE_EDSP) += encoder/arm/armv5te/vp8_packtokens_armv5$(ASM)
+VP8_CX_SRCS-$(HAVE_EDSP) += encoder/arm/armv5te/vp8_packtokens_mbrow_armv5$(ASM)
+VP8_CX_SRCS-$(HAVE_EDSP) += encoder/arm/armv5te/vp8_packtokens_partitions_armv5$(ASM)
+
+#File list for media
+# encoder
+VP8_CX_SRCS-$(HAVE_MEDIA) += encoder/arm/armv6/vp8_subtract_armv6$(ASM)
+VP8_CX_SRCS-$(HAVE_MEDIA) += encoder/arm/armv6/vp8_short_fdct4x4_armv6$(ASM)
+VP8_CX_SRCS-$(HAVE_MEDIA) += encoder/arm/armv6/vp8_fast_quantize_b_armv6$(ASM)
+VP8_CX_SRCS-$(HAVE_MEDIA) += encoder/arm/armv6/vp8_mse16x16_armv6$(ASM)
+VP8_CX_SRCS-$(HAVE_MEDIA) += encoder/arm/armv6/walsh_v6$(ASM)
+
+#File list for neon
+# encoder
+VP8_CX_SRCS-$(HAVE_NEON) += encoder/arm/neon/fastquantizeb_neon$(ASM)
+VP8_CX_SRCS-$(HAVE_NEON) += encoder/arm/neon/picklpf_arm.c
+VP8_CX_SRCS-$(HAVE_NEON) += encoder/arm/neon/shortfdct_neon$(ASM)
+VP8_CX_SRCS-$(HAVE_NEON) += encoder/arm/neon/subtract_neon$(ASM)
+VP8_CX_SRCS-$(HAVE_NEON) += encoder/arm/neon/vp8_mse16x16_neon$(ASM)
+VP8_CX_SRCS-$(HAVE_NEON) += encoder/arm/neon/vp8_memcpy_neon$(ASM)
+VP8_CX_SRCS-$(HAVE_NEON) += encoder/arm/neon/vp8_shortwalsh4x4_neon$(ASM)
diff --git a/libvpx/vp8/vp8dx.mk b/libvpx/vp8/vp8dx.mk
new file mode 100644
index 0000000..dd39190
--- /dev/null
+++ b/libvpx/vp8/vp8dx.mk
@@ -0,0 +1,66 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8_common.mk
+
+VP8_DX_EXPORTS += exports_dec
+
+VP8_DX_SRCS-yes += $(VP8_COMMON_SRCS-yes)
+VP8_DX_SRCS-no += $(VP8_COMMON_SRCS-no)
+VP8_DX_SRCS_REMOVE-yes += $(VP8_COMMON_SRCS_REMOVE-yes)
+VP8_DX_SRCS_REMOVE-no += $(VP8_COMMON_SRCS_REMOVE-no)
+
+VP8_DX_SRCS-yes += vp8dx.mk
+
+VP8_DX_SRCS-yes += vp8_dx_iface.c
+
+# common
+#define ARM
+#define DISABLE_THREAD
+
+#INCLUDES += algo/vpx_common/vpx_mem/include
+#INCLUDES += common
+#INCLUDES += common
+#INCLUDES += common
+#INCLUDES += common
+#INCLUDES += decoder
+
+
+
+# decoder
+#define ARM
+#define DISABLE_THREAD
+
+#INCLUDES += algo/vpx_common/vpx_mem/include
+#INCLUDES += common
+#INCLUDES += common
+#INCLUDES += common
+#INCLUDES += common
+#INCLUDES += decoder
+
+VP8_DX_SRCS-yes += decoder/asm_dec_offsets.c
+VP8_DX_SRCS-yes += decoder/dboolhuff.c
+VP8_DX_SRCS-yes += decoder/decodemv.c
+VP8_DX_SRCS-yes += decoder/decodframe.c
+VP8_DX_SRCS-yes += decoder/detokenize.c
+VP8_DX_SRCS-$(CONFIG_ERROR_CONCEALMENT) += decoder/ec_types.h
+VP8_DX_SRCS-$(CONFIG_ERROR_CONCEALMENT) += decoder/error_concealment.h
+VP8_DX_SRCS-$(CONFIG_ERROR_CONCEALMENT) += decoder/error_concealment.c
+VP8_DX_SRCS-yes += decoder/dboolhuff.h
+VP8_DX_SRCS-yes += decoder/decodemv.h
+VP8_DX_SRCS-yes += decoder/decoderthreading.h
+VP8_DX_SRCS-yes += decoder/detokenize.h
+VP8_DX_SRCS-yes += decoder/onyxd_int.h
+VP8_DX_SRCS-yes += decoder/treereader.h
+VP8_DX_SRCS-yes += decoder/onyxd_if.c
+VP8_DX_SRCS-$(CONFIG_MULTITHREAD) += decoder/threading.c
+
+VP8_DX_SRCS-yes := $(filter-out $(VP8_DX_SRCS_REMOVE-yes),$(VP8_DX_SRCS-yes))
diff --git a/libvpx/vp8_multi_resolution_encoder.c b/libvpx/vp8_multi_resolution_encoder.c
new file mode 100644
index 0000000..eae36a4
--- /dev/null
+++ b/libvpx/vp8_multi_resolution_encoder.c
@@ -0,0 +1,486 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+/*
+ * This is an example demonstrating multi-resolution encoding in VP8.
+ * High-resolution input video is down-sampled to lower-resolutions. The
+ * encoder then encodes the video and outputs multiple bitstreams with
+ * different resolutions.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#include "math.h"
+#define VPX_CODEC_DISABLE_COMPAT 1
+#include "vpx/vpx_encoder.h"
+#include "vpx/vp8cx.h"
+#include "vpx_ports/mem_ops.h"
+#define interface (vpx_codec_vp8_cx())
+#define fourcc 0x30385056
+
+#define IVF_FILE_HDR_SZ (32)
+#define IVF_FRAME_HDR_SZ (12)
+
+/*
+ * The input video frame is downsampled several times to generate a multi-level
+ * hierarchical structure. NUM_ENCODERS is defined as the number of encoding
+ * levels required. For example, if the size of input video is 1280x720,
+ * NUM_ENCODERS is 3, and down-sampling factor is 2, the encoder outputs 3
+ * bitstreams with resolution of 1280x720(level 0), 640x360(level 1), and
+ * 320x180(level 2) respectively.
+ */
+#define NUM_ENCODERS 3
+
+/* This example uses the scaler function in libyuv. */
+#include "third_party/libyuv/include/libyuv/basic_types.h"
+#include "third_party/libyuv/include/libyuv/scale.h"
+#include "third_party/libyuv/include/libyuv/cpu_id.h"
+
+static double vp8_mse2psnr(double Samples, double Peak, double Mse)
+{
+ double psnr;
+
+ if ((double)Mse > 0.0)
+ psnr = 10.0 * log10(Peak * Peak * Samples / Mse);
+ else
+ psnr = 60; // Limit to prevent / 0
+
+ if (psnr > 60)
+ psnr = 60;
+
+ return psnr;
+}
+
+static void die(const char *fmt, ...) {
+ va_list ap;
+
+ va_start(ap, fmt);
+ vprintf(fmt, ap);
+ if(fmt[strlen(fmt)-1] != '\n')
+ printf("\n");
+ exit(EXIT_FAILURE);
+}
+
+static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
+ const char *detail = vpx_codec_error_detail(ctx);
+
+ printf("%s: %s\n", s, vpx_codec_error(ctx));
+ if(detail)
+ printf(" %s\n",detail);
+ exit(EXIT_FAILURE);
+}
+
+int (*read_frame_p)(FILE *f, vpx_image_t *img);
+
+static int read_frame(FILE *f, vpx_image_t *img) {
+ size_t nbytes, to_read;
+ int res = 1;
+
+ to_read = img->w*img->h*3/2;
+ nbytes = fread(img->planes[0], 1, to_read, f);
+ if(nbytes != to_read) {
+ res = 0;
+ if(nbytes > 0)
+ printf("Warning: Read partial frame. Check your width & height!\n");
+ }
+ return res;
+}
+
+static int read_frame_by_row(FILE *f, vpx_image_t *img) {
+ size_t nbytes, to_read;
+ int res = 1;
+ int plane;
+
+ for (plane = 0; plane < 3; plane++)
+ {
+ unsigned char *ptr;
+ int w = (plane ? (1 + img->d_w) / 2 : img->d_w);
+ int h = (plane ? (1 + img->d_h) / 2 : img->d_h);
+ int r;
+
+ /* Determine the correct plane based on the image format. The for-loop
+ * always counts in Y,U,V order, but this may not match the order of
+ * the data on disk.
+ */
+ switch (plane)
+ {
+ case 1:
+ ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12? VPX_PLANE_V : VPX_PLANE_U];
+ break;
+ case 2:
+ ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12?VPX_PLANE_U : VPX_PLANE_V];
+ break;
+ default:
+ ptr = img->planes[plane];
+ }
+
+ for (r = 0; r < h; r++)
+ {
+ to_read = w;
+
+ nbytes = fread(ptr, 1, to_read, f);
+ if(nbytes != to_read) {
+ res = 0;
+ if(nbytes > 0)
+ printf("Warning: Read partial frame. Check your width & height!\n");
+ break;
+ }
+
+ ptr += img->stride[plane];
+ }
+ if (!res)
+ break;
+ }
+
+ return res;
+}
+
+static void write_ivf_file_header(FILE *outfile,
+ const vpx_codec_enc_cfg_t *cfg,
+ int frame_cnt) {
+ char header[32];
+
+ if(cfg->g_pass != VPX_RC_ONE_PASS && cfg->g_pass != VPX_RC_LAST_PASS)
+ return;
+ header[0] = 'D';
+ header[1] = 'K';
+ header[2] = 'I';
+ header[3] = 'F';
+ mem_put_le16(header+4, 0); /* version */
+ mem_put_le16(header+6, 32); /* headersize */
+ mem_put_le32(header+8, fourcc); /* headersize */
+ mem_put_le16(header+12, cfg->g_w); /* width */
+ mem_put_le16(header+14, cfg->g_h); /* height */
+ mem_put_le32(header+16, cfg->g_timebase.den); /* rate */
+ mem_put_le32(header+20, cfg->g_timebase.num); /* scale */
+ mem_put_le32(header+24, frame_cnt); /* length */
+ mem_put_le32(header+28, 0); /* unused */
+
+ (void) fwrite(header, 1, 32, outfile);
+}
+
+static void write_ivf_frame_header(FILE *outfile,
+ const vpx_codec_cx_pkt_t *pkt)
+{
+ char header[12];
+ vpx_codec_pts_t pts;
+
+ if(pkt->kind != VPX_CODEC_CX_FRAME_PKT)
+ return;
+
+ pts = pkt->data.frame.pts;
+ mem_put_le32(header, pkt->data.frame.sz);
+ mem_put_le32(header+4, pts&0xFFFFFFFF);
+ mem_put_le32(header+8, pts >> 32);
+
+ (void) fwrite(header, 1, 12, outfile);
+}
+
+int main(int argc, char **argv)
+{
+ FILE *infile, *outfile[NUM_ENCODERS];
+ vpx_codec_ctx_t codec[NUM_ENCODERS];
+ vpx_codec_enc_cfg_t cfg[NUM_ENCODERS];
+ vpx_codec_pts_t frame_cnt = 0;
+ vpx_image_t raw[NUM_ENCODERS];
+ vpx_codec_err_t res[NUM_ENCODERS];
+
+ int i;
+ long width;
+ long height;
+ int frame_avail;
+ int got_data;
+ int flags = 0;
+
+ /*Currently, only realtime mode is supported in multi-resolution encoding.*/
+ int arg_deadline = VPX_DL_REALTIME;
+
+ /* Set show_psnr to 1/0 to show/not show PSNR. Choose show_psnr=0 if you
+ don't need to know PSNR, which will skip PSNR calculation and save
+ encoding time. */
+ int show_psnr = 0;
+ uint64_t psnr_sse_total[NUM_ENCODERS] = {0};
+ uint64_t psnr_samples_total[NUM_ENCODERS] = {0};
+ double psnr_totals[NUM_ENCODERS][4] = {{0,0}};
+ int psnr_count[NUM_ENCODERS] = {0};
+
+ /* Set the required target bitrates for each resolution level.
+ * If target bitrate for highest-resolution level is set to 0,
+ * (i.e. target_bitrate[0]=0), we skip encoding at that level.
+ */
+ unsigned int target_bitrate[NUM_ENCODERS]={1400, 500, 100};
+ /* Enter the frame rate of the input video */
+ int framerate = 30;
+ /* Set down-sampling factor for each resolution level.
+ dsf[0] controls down sampling from level 0 to level 1;
+ dsf[1] controls down sampling from level 1 to level 2;
+ dsf[2] is not used. */
+ vpx_rational_t dsf[NUM_ENCODERS] = {{2, 1}, {2, 1}, {1, 1}};
+
+ if(argc!= (5+NUM_ENCODERS))
+ die("Usage: %s <width> <height> <infile> <outfile(s)> <output psnr?>\n",
+ argv[0]);
+
+ printf("Using %s\n",vpx_codec_iface_name(interface));
+
+ width = strtol(argv[1], NULL, 0);
+ height = strtol(argv[2], NULL, 0);
+
+ if(width < 16 || width%2 || height <16 || height%2)
+ die("Invalid resolution: %ldx%ld", width, height);
+
+ /* Open input video file for encoding */
+ if(!(infile = fopen(argv[3], "rb")))
+ die("Failed to open %s for reading", argv[3]);
+
+ /* Open output file for each encoder to output bitstreams */
+ for (i=0; i< NUM_ENCODERS; i++)
+ {
+ if(!target_bitrate[i])
+ {
+ outfile[i] = NULL;
+ continue;
+ }
+
+ if(!(outfile[i] = fopen(argv[i+4], "wb")))
+ die("Failed to open %s for writing", argv[i+4]);
+ }
+
+ show_psnr = strtol(argv[NUM_ENCODERS + 4], NULL, 0);
+
+ /* Populate default encoder configuration */
+ for (i=0; i< NUM_ENCODERS; i++)
+ {
+ res[i] = vpx_codec_enc_config_default(interface, &cfg[i], 0);
+ if(res[i]) {
+ printf("Failed to get config: %s\n", vpx_codec_err_to_string(res[i]));
+ return EXIT_FAILURE;
+ }
+ }
+
+ /*
+ * Update the default configuration according to needs of the application.
+ */
+ /* Highest-resolution encoder settings */
+ cfg[0].g_w = width;
+ cfg[0].g_h = height;
+ cfg[0].g_threads = 1; /* number of threads used */
+ cfg[0].rc_dropframe_thresh = 30;
+ cfg[0].rc_end_usage = VPX_CBR;
+ cfg[0].rc_resize_allowed = 0;
+ cfg[0].rc_min_quantizer = 4;
+ cfg[0].rc_max_quantizer = 56;
+ cfg[0].rc_undershoot_pct = 98;
+ cfg[0].rc_overshoot_pct = 100;
+ cfg[0].rc_buf_initial_sz = 500;
+ cfg[0].rc_buf_optimal_sz = 600;
+ cfg[0].rc_buf_sz = 1000;
+ cfg[0].g_error_resilient = 1; /* Enable error resilient mode */
+ cfg[0].g_lag_in_frames = 0;
+
+ /* Disable automatic keyframe placement */
+ /* Note: These 3 settings are copied to all levels. But, except the lowest
+ * resolution level, all other levels are set to VPX_KF_DISABLED internally.
+ */
+ //cfg[0].kf_mode = VPX_KF_DISABLED;
+ cfg[0].kf_mode = VPX_KF_AUTO;
+ cfg[0].kf_min_dist = 3000;
+ cfg[0].kf_max_dist = 3000;
+
+ cfg[0].rc_target_bitrate = target_bitrate[0]; /* Set target bitrate */
+ cfg[0].g_timebase.num = 1; /* Set fps */
+ cfg[0].g_timebase.den = framerate;
+
+ /* Other-resolution encoder settings */
+ for (i=1; i< NUM_ENCODERS; i++)
+ {
+ memcpy(&cfg[i], &cfg[0], sizeof(vpx_codec_enc_cfg_t));
+
+ cfg[i].g_threads = 1; /* number of threads used */
+ cfg[i].rc_target_bitrate = target_bitrate[i];
+
+ /* Note: Width & height of other-resolution encoders are calculated
+ * from the highest-resolution encoder's size and the corresponding
+ * down_sampling_factor.
+ */
+ {
+ unsigned int iw = cfg[i-1].g_w*dsf[i-1].den + dsf[i-1].num - 1;
+ unsigned int ih = cfg[i-1].g_h*dsf[i-1].den + dsf[i-1].num - 1;
+ cfg[i].g_w = iw/dsf[i-1].num;
+ cfg[i].g_h = ih/dsf[i-1].num;
+ }
+
+ /* Make width & height to be multiplier of 2. */
+ // Should support odd size ???
+ if((cfg[i].g_w)%2)cfg[i].g_w++;
+ if((cfg[i].g_h)%2)cfg[i].g_h++;
+ }
+
+ /* Allocate image for each encoder */
+ for (i=0; i< NUM_ENCODERS; i++)
+ if(!vpx_img_alloc(&raw[i], VPX_IMG_FMT_I420, cfg[i].g_w, cfg[i].g_h, 32))
+ die("Failed to allocate image", cfg[i].g_w, cfg[i].g_h);
+
+ if (raw[0].stride[VPX_PLANE_Y] == raw[0].d_w)
+ read_frame_p = read_frame;
+ else
+ read_frame_p = read_frame_by_row;
+
+ for (i=0; i< NUM_ENCODERS; i++)
+ if(outfile[i])
+ write_ivf_file_header(outfile[i], &cfg[i], 0);
+
+ /* Initialize multi-encoder */
+ if(vpx_codec_enc_init_multi(&codec[0], interface, &cfg[0], NUM_ENCODERS,
+ (show_psnr ? VPX_CODEC_USE_PSNR : 0), &dsf[0]))
+ die_codec(&codec[0], "Failed to initialize encoder");
+
+ /* The extra encoding configuration parameters can be set as follows. */
+ /* Set encoding speed */
+ for ( i=0; i<NUM_ENCODERS; i++)
+ {
+ int speed = -6;
+ if(vpx_codec_control(&codec[i], VP8E_SET_CPUUSED, speed))
+ die_codec(&codec[i], "Failed to set cpu_used");
+ }
+ /* Set static thresh for highest-resolution encoder. Set it to 1000 for
+ * better performance. */
+ {
+ unsigned int static_thresh = 1000;
+ if(vpx_codec_control(&codec[0], VP8E_SET_STATIC_THRESHOLD, static_thresh))
+ die_codec(&codec[0], "Failed to set static threshold");
+ }
+ /* Set static thresh = 0 for other encoders for better quality */
+ for ( i=1; i<NUM_ENCODERS; i++)
+ {
+ unsigned int static_thresh = 0;
+ if(vpx_codec_control(&codec[i], VP8E_SET_STATIC_THRESHOLD, static_thresh))
+ die_codec(&codec[i], "Failed to set static threshold");
+ }
+ /* Set NOISE_SENSITIVITY to do TEMPORAL_DENOISING */
+ for ( i=0; i< NUM_ENCODERS; i++)
+ {
+ if(vpx_codec_control(&codec[i], VP8E_SET_NOISE_SENSITIVITY, 0))
+ die_codec(&codec[i], "Failed to set noise_sensitivity");
+ }
+
+ frame_avail = 1;
+ got_data = 0;
+
+ while(frame_avail || got_data)
+ {
+ vpx_codec_iter_t iter[NUM_ENCODERS]={NULL};
+ const vpx_codec_cx_pkt_t *pkt[NUM_ENCODERS];
+
+ flags = 0;
+ frame_avail = read_frame_p(infile, &raw[0]);
+
+ if(frame_avail)
+ {
+ for ( i=1; i<NUM_ENCODERS; i++)
+ {
+ /*Scale the image down a number of times by downsampling factor*/
+ /* FilterMode 1 or 2 give better psnr than FilterMode 0. */
+ I420Scale(raw[i-1].planes[VPX_PLANE_Y], raw[i-1].stride[VPX_PLANE_Y],
+ raw[i-1].planes[VPX_PLANE_U], raw[i-1].stride[VPX_PLANE_U],
+ raw[i-1].planes[VPX_PLANE_V], raw[i-1].stride[VPX_PLANE_V],
+ raw[i-1].d_w, raw[i-1].d_h,
+ raw[i].planes[VPX_PLANE_Y], raw[i].stride[VPX_PLANE_Y],
+ raw[i].planes[VPX_PLANE_U], raw[i].stride[VPX_PLANE_U],
+ raw[i].planes[VPX_PLANE_V], raw[i].stride[VPX_PLANE_V],
+ raw[i].d_w, raw[i].d_h, 1);
+ }
+ }
+
+ /* Encode each frame at multi-levels */
+ if(vpx_codec_encode(&codec[0], frame_avail? &raw[0] : NULL,
+ frame_cnt, 1, flags, arg_deadline))
+ die_codec(&codec[0], "Failed to encode frame");
+
+ for (i=NUM_ENCODERS-1; i>=0 ; i--)
+ {
+ got_data = 0;
+
+ while( (pkt[i] = vpx_codec_get_cx_data(&codec[i], &iter[i])) )
+ {
+ got_data = 1;
+ switch(pkt[i]->kind) {
+ case VPX_CODEC_CX_FRAME_PKT:
+ write_ivf_frame_header(outfile[i], pkt[i]);
+ (void) fwrite(pkt[i]->data.frame.buf, 1,
+ pkt[i]->data.frame.sz, outfile[i]);
+ break;
+ case VPX_CODEC_PSNR_PKT:
+ if (show_psnr)
+ {
+ int j;
+
+ psnr_sse_total[i] += pkt[i]->data.psnr.sse[0];
+ psnr_samples_total[i] += pkt[i]->data.psnr.samples[0];
+ for (j = 0; j < 4; j++)
+ {
+ //fprintf(stderr, "%.3lf ", pkt[i]->data.psnr.psnr[j]);
+ psnr_totals[i][j] += pkt[i]->data.psnr.psnr[j];
+ }
+ psnr_count[i]++;
+ }
+
+ break;
+ default:
+ break;
+ }
+ printf(pkt[i]->kind == VPX_CODEC_CX_FRAME_PKT
+ && (pkt[i]->data.frame.flags & VPX_FRAME_IS_KEY)? "K":".");
+ fflush(stdout);
+ }
+ }
+ frame_cnt++;
+ }
+ printf("\n");
+
+ fclose(infile);
+
+ printf("Processed %ld frames.\n",(long int)frame_cnt-1);
+ for (i=0; i< NUM_ENCODERS; i++)
+ {
+ /* Calculate PSNR and print it out */
+ if ( (show_psnr) && (psnr_count[i]>0) )
+ {
+ int j;
+ double ovpsnr = vp8_mse2psnr(psnr_samples_total[i], 255.0,
+ psnr_sse_total[i]);
+
+ fprintf(stderr, "\n ENC%d PSNR (Overall/Avg/Y/U/V)", i);
+
+ fprintf(stderr, " %.3lf", ovpsnr);
+ for (j = 0; j < 4; j++)
+ {
+ fprintf(stderr, " %.3lf", psnr_totals[i][j]/psnr_count[i]);
+ }
+ }
+
+ if(vpx_codec_destroy(&codec[i]))
+ die_codec(&codec[i], "Failed to destroy codec");
+
+ vpx_img_free(&raw[i]);
+
+ if(!outfile[i])
+ continue;
+
+ /* Try to rewrite the file header with the actual frame count */
+ if(!fseek(outfile[i], 0, SEEK_SET))
+ write_ivf_file_header(outfile[i], &cfg[i], frame_cnt-1);
+ fclose(outfile[i]);
+ }
+ printf("\n");
+
+ return EXIT_SUCCESS;
+}
diff --git a/libvpx/vp8_scalable_patterns.c b/libvpx/vp8_scalable_patterns.c
new file mode 100644
index 0000000..06270fe
--- /dev/null
+++ b/libvpx/vp8_scalable_patterns.c
@@ -0,0 +1,694 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*
+ * This is an example demonstrating how to implement a multi-layer VP8
+ * encoding scheme based on temporal scalability for video applications
+ * that benefit from a scalable bitstream.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#define VPX_CODEC_DISABLE_COMPAT 1
+#include "vpx/vpx_encoder.h"
+#include "vpx/vp8cx.h"
+#define interface (vpx_codec_vp8_cx())
+#define fourcc 0x30385056
+
+#define IVF_FILE_HDR_SZ (32)
+#define IVF_FRAME_HDR_SZ (12)
+
+static void mem_put_le16(char *mem, unsigned int val) {
+ mem[0] = val;
+ mem[1] = val>>8;
+}
+
+static void mem_put_le32(char *mem, unsigned int val) {
+ mem[0] = val;
+ mem[1] = val>>8;
+ mem[2] = val>>16;
+ mem[3] = val>>24;
+}
+
+static void die(const char *fmt, ...) {
+ va_list ap;
+
+ va_start(ap, fmt);
+ vprintf(fmt, ap);
+ if(fmt[strlen(fmt)-1] != '\n')
+ printf("\n");
+ exit(EXIT_FAILURE);
+}
+
+static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
+ const char *detail = vpx_codec_error_detail(ctx);
+
+ printf("%s: %s\n", s, vpx_codec_error(ctx));
+ if(detail)
+ printf(" %s\n",detail);
+ exit(EXIT_FAILURE);
+}
+
+static int read_frame(FILE *f, vpx_image_t *img) {
+ size_t nbytes, to_read;
+ int res = 1;
+
+ to_read = img->w*img->h*3/2;
+ nbytes = fread(img->planes[0], 1, to_read, f);
+ if(nbytes != to_read) {
+ res = 0;
+ if(nbytes > 0)
+ printf("Warning: Read partial frame. Check your width & height!\n");
+ }
+ return res;
+}
+
+static void write_ivf_file_header(FILE *outfile,
+ const vpx_codec_enc_cfg_t *cfg,
+ int frame_cnt) {
+ char header[32];
+
+ if(cfg->g_pass != VPX_RC_ONE_PASS && cfg->g_pass != VPX_RC_LAST_PASS)
+ return;
+ header[0] = 'D';
+ header[1] = 'K';
+ header[2] = 'I';
+ header[3] = 'F';
+ mem_put_le16(header+4, 0); /* version */
+ mem_put_le16(header+6, 32); /* headersize */
+ mem_put_le32(header+8, fourcc); /* headersize */
+ mem_put_le16(header+12, cfg->g_w); /* width */
+ mem_put_le16(header+14, cfg->g_h); /* height */
+ mem_put_le32(header+16, cfg->g_timebase.den); /* rate */
+ mem_put_le32(header+20, cfg->g_timebase.num); /* scale */
+ mem_put_le32(header+24, frame_cnt); /* length */
+ mem_put_le32(header+28, 0); /* unused */
+
+ (void) fwrite(header, 1, 32, outfile);
+}
+
+
+static void write_ivf_frame_header(FILE *outfile,
+ const vpx_codec_cx_pkt_t *pkt)
+{
+ char header[12];
+ vpx_codec_pts_t pts;
+
+ if(pkt->kind != VPX_CODEC_CX_FRAME_PKT)
+ return;
+
+ pts = pkt->data.frame.pts;
+ mem_put_le32(header, pkt->data.frame.sz);
+ mem_put_le32(header+4, pts&0xFFFFFFFF);
+ mem_put_le32(header+8, pts >> 32);
+
+ (void) fwrite(header, 1, 12, outfile);
+}
+
+static int mode_to_num_layers[12] = {1, 2, 2, 3, 3, 3, 3, 5, 2, 3, 3, 3};
+
+int main(int argc, char **argv) {
+ FILE *infile, *outfile[VPX_TS_MAX_LAYERS];
+ vpx_codec_ctx_t codec;
+ vpx_codec_enc_cfg_t cfg;
+ int frame_cnt = 0;
+ vpx_image_t raw;
+ vpx_codec_err_t res;
+ unsigned int width;
+ unsigned int height;
+ int frame_avail;
+ int got_data;
+ int flags = 0;
+ int i;
+ int pts = 0; /* PTS starts at 0 */
+ int frame_duration = 1; /* 1 timebase tick per frame */
+
+ int layering_mode = 0;
+ int frames_in_layer[VPX_TS_MAX_LAYERS] = {0};
+ int layer_flags[VPX_TS_MAX_PERIODICITY] = {0};
+ int flag_periodicity;
+ int max_intra_size_pct;
+
+ /* Check usage and arguments */
+ if (argc < 9)
+ die("Usage: %s <infile> <outfile> <width> <height> <rate_num> "
+ " <rate_den> <mode> <Rate_0> ... <Rate_nlayers-1>\n", argv[0]);
+
+ width = strtol (argv[3], NULL, 0);
+ height = strtol (argv[4], NULL, 0);
+ if (width < 16 || width%2 || height <16 || height%2)
+ die ("Invalid resolution: %d x %d", width, height);
+
+ if (!sscanf(argv[7], "%d", &layering_mode))
+ die ("Invalid mode %s", argv[7]);
+ if (layering_mode<0 || layering_mode>11)
+ die ("Invalid mode (0..11) %s", argv[7]);
+
+ if (argc != 8+mode_to_num_layers[layering_mode])
+ die ("Invalid number of arguments");
+
+ if (!vpx_img_alloc (&raw, VPX_IMG_FMT_I420, width, height, 32))
+ die ("Failed to allocate image", width, height);
+
+ printf("Using %s\n",vpx_codec_iface_name(interface));
+
+ /* Populate encoder configuration */
+ res = vpx_codec_enc_config_default(interface, &cfg, 0);
+ if(res) {
+ printf("Failed to get config: %s\n", vpx_codec_err_to_string(res));
+ return EXIT_FAILURE;
+ }
+
+ /* Update the default configuration with our settings */
+ cfg.g_w = width;
+ cfg.g_h = height;
+
+ /* Timebase format e.g. 30fps: numerator=1, demoninator=30 */
+ if (!sscanf (argv[5], "%d", &cfg.g_timebase.num ))
+ die ("Invalid timebase numerator %s", argv[5]);
+ if (!sscanf (argv[6], "%d", &cfg.g_timebase.den ))
+ die ("Invalid timebase denominator %s", argv[6]);
+
+ for (i=8; i<8+mode_to_num_layers[layering_mode]; i++)
+ if (!sscanf(argv[i], "%ud", &cfg.ts_target_bitrate[i-8]))
+ die ("Invalid data rate %s", argv[i]);
+
+ /* Real time parameters */
+ cfg.rc_dropframe_thresh = 0;
+ cfg.rc_end_usage = VPX_CBR;
+ cfg.rc_resize_allowed = 0;
+ cfg.rc_min_quantizer = 2;
+ cfg.rc_max_quantizer = 56;
+ cfg.rc_undershoot_pct = 100;
+ cfg.rc_overshoot_pct = 15;
+ cfg.rc_buf_initial_sz = 500;
+ cfg.rc_buf_optimal_sz = 600;
+ cfg.rc_buf_sz = 1000;
+
+ /* Enable error resilient mode */
+ cfg.g_error_resilient = 1;
+ cfg.g_lag_in_frames = 0;
+ cfg.kf_mode = VPX_KF_DISABLED;
+
+ /* Disable automatic keyframe placement */
+ cfg.kf_min_dist = cfg.kf_max_dist = 3000;
+
+ /* Default setting for bitrate: used in special case of 1 layer (case 0). */
+ cfg.rc_target_bitrate = cfg.ts_target_bitrate[0];
+
+ /* Temporal scaling parameters: */
+ /* NOTE: The 3 prediction frames cannot be used interchangeably due to
+ * differences in the way they are handled throughout the code. The
+ * frames should be allocated to layers in the order LAST, GF, ARF.
+ * Other combinations work, but may produce slightly inferior results.
+ */
+ switch (layering_mode)
+ {
+ case 0:
+ {
+ /* 1-layer */
+ int ids[1] = {0};
+ cfg.ts_number_layers = 1;
+ cfg.ts_periodicity = 1;
+ cfg.ts_rate_decimator[0] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = cfg.ts_periodicity;
+
+ // Update L only.
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+ break;
+ }
+ case 1:
+ {
+ /* 2-layers, 2-frame period */
+ int ids[2] = {0,1};
+ cfg.ts_number_layers = 2;
+ cfg.ts_periodicity = 2;
+ cfg.ts_rate_decimator[0] = 2;
+ cfg.ts_rate_decimator[1] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = cfg.ts_periodicity;
+#if 1
+ /* 0=L, 1=GF, Intra-layer prediction enabled */
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF;
+ layer_flags[1] = VP8_EFLAG_NO_UPD_ARF | VP8_EFLAG_NO_UPD_LAST |
+ VP8_EFLAG_NO_REF_ARF;
+#else
+ /* 0=L, 1=GF, Intra-layer prediction disabled */
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF;
+ layer_flags[1] = VP8_EFLAG_NO_UPD_ARF | VP8_EFLAG_NO_UPD_LAST |
+ VP8_EFLAG_NO_REF_ARF | VP8_EFLAG_NO_REF_LAST;
+#endif
+ break;
+ }
+
+ case 2:
+ {
+ /* 2-layers, 3-frame period */
+ int ids[3] = {0,1,1};
+ cfg.ts_number_layers = 2;
+ cfg.ts_periodicity = 3;
+ cfg.ts_rate_decimator[0] = 3;
+ cfg.ts_rate_decimator[1] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = cfg.ts_periodicity;
+
+ /* 0=L, 1=GF, Intra-layer prediction enabled */
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[1] =
+ layer_flags[2] = VP8_EFLAG_NO_REF_GF |
+ VP8_EFLAG_NO_REF_ARF | VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST;
+ break;
+ }
+
+ case 3:
+ {
+ /* 3-layers, 6-frame period */
+ int ids[6] = {0,2,2,1,2,2};
+ cfg.ts_number_layers = 3;
+ cfg.ts_periodicity = 6;
+ cfg.ts_rate_decimator[0] = 6;
+ cfg.ts_rate_decimator[1] = 3;
+ cfg.ts_rate_decimator[2] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = cfg.ts_periodicity;
+
+ /* 0=L, 1=GF, 2=ARF, Intra-layer prediction enabled */
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[3] = VP8_EFLAG_NO_REF_ARF | VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST;
+ layer_flags[1] =
+ layer_flags[2] =
+ layer_flags[4] =
+ layer_flags[5] = VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_LAST;
+ break;
+ }
+
+ case 4:
+ {
+ /* 3-layers, 4-frame period */
+ int ids[4] = {0,2,1,2};
+ cfg.ts_number_layers = 3;
+ cfg.ts_periodicity = 4;
+ cfg.ts_rate_decimator[0] = 4;
+ cfg.ts_rate_decimator[1] = 2;
+ cfg.ts_rate_decimator[2] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = cfg.ts_periodicity;
+
+ /* 0=L, 1=GF, 2=ARF, Intra-layer prediction disabled */
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[2] = VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST;
+ layer_flags[1] =
+ layer_flags[3] = VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF;
+ break;
+ }
+
+ case 5:
+ {
+ /* 3-layers, 4-frame period */
+ int ids[4] = {0,2,1,2};
+ cfg.ts_number_layers = 3;
+ cfg.ts_periodicity = 4;
+ cfg.ts_rate_decimator[0] = 4;
+ cfg.ts_rate_decimator[1] = 2;
+ cfg.ts_rate_decimator[2] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = cfg.ts_periodicity;
+
+ /* 0=L, 1=GF, 2=ARF, Intra-layer prediction enabled in layer 1,
+ * disabled in layer 2
+ */
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[2] = VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[1] =
+ layer_flags[3] = VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF;
+ break;
+ }
+
+ case 6:
+ {
+ /* 3-layers, 4-frame period */
+ int ids[4] = {0,2,1,2};
+ cfg.ts_number_layers = 3;
+ cfg.ts_periodicity = 4;
+ cfg.ts_rate_decimator[0] = 4;
+ cfg.ts_rate_decimator[1] = 2;
+ cfg.ts_rate_decimator[2] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = cfg.ts_periodicity;
+
+ /* 0=L, 1=GF, 2=ARF, Intra-layer prediction enabled */
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[2] = VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[1] =
+ layer_flags[3] = VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF;
+ break;
+ }
+
+ case 7:
+ {
+ /* NOTE: Probably of academic interest only */
+
+ /* 5-layers, 16-frame period */
+ int ids[16] = {0,4,3,4,2,4,3,4,1,4,3,4,2,4,3,4};
+ cfg.ts_number_layers = 5;
+ cfg.ts_periodicity = 16;
+ cfg.ts_rate_decimator[0] = 16;
+ cfg.ts_rate_decimator[1] = 8;
+ cfg.ts_rate_decimator[2] = 4;
+ cfg.ts_rate_decimator[3] = 2;
+ cfg.ts_rate_decimator[4] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = cfg.ts_periodicity;
+
+ layer_flags[0] = VPX_EFLAG_FORCE_KF;
+ layer_flags[1] =
+ layer_flags[3] =
+ layer_flags[5] =
+ layer_flags[7] =
+ layer_flags[9] =
+ layer_flags[11] =
+ layer_flags[13] =
+ layer_flags[15] = VP8_EFLAG_NO_UPD_LAST |
+ VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[2] =
+ layer_flags[6] =
+ layer_flags[10] =
+ layer_flags[14] = VP8_EFLAG_NO_UPD_ARF | VP8_EFLAG_NO_UPD_GF;
+ layer_flags[4] =
+ layer_flags[12] = VP8_EFLAG_NO_REF_LAST |
+ VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[8] = VP8_EFLAG_NO_REF_LAST | VP8_EFLAG_NO_REF_GF;
+ break;
+ }
+
+ case 8:
+ {
+ /* 2-layers, with sync point at first frame of layer 1. */
+ int ids[2] = {0,1};
+ cfg.ts_number_layers = 2;
+ cfg.ts_periodicity = 2;
+ cfg.ts_rate_decimator[0] = 2;
+ cfg.ts_rate_decimator[1] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = 8;
+
+ /* 0=L, 1=GF */
+ // ARF is used as predictor for all frames, and is only updated on
+ // key frame. Sync point every 8 frames.
+
+ // Layer 0: predict from L and ARF, update L and G.
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_REF_GF |
+ VP8_EFLAG_NO_UPD_ARF;
+
+ // Layer 1: sync point: predict from L and ARF, and update G.
+ layer_flags[1] = VP8_EFLAG_NO_REF_GF |
+ VP8_EFLAG_NO_UPD_LAST |
+ VP8_EFLAG_NO_UPD_ARF;
+
+ // Layer 0, predict from L and ARF, update L.
+ layer_flags[2] = VP8_EFLAG_NO_REF_GF |
+ VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF;
+
+ // Layer 1: predict from L, G and ARF, and update G.
+ layer_flags[3] = VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST |
+ VP8_EFLAG_NO_UPD_ENTROPY;
+
+ // Layer 0
+ layer_flags[4] = layer_flags[2];
+
+ // Layer 1
+ layer_flags[5] = layer_flags[3];
+
+ // Layer 0
+ layer_flags[6] = layer_flags[4];
+
+ // Layer 1
+ layer_flags[7] = layer_flags[5];
+ break;
+ }
+
+ case 9:
+ {
+ /* 3-layers */
+ // Sync points for layer 1 and 2 every 8 frames.
+
+ int ids[4] = {0,2,1,2};
+ cfg.ts_number_layers = 3;
+ cfg.ts_periodicity = 4;
+ cfg.ts_rate_decimator[0] = 4;
+ cfg.ts_rate_decimator[1] = 2;
+ cfg.ts_rate_decimator[2] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = 8;
+
+ /* 0=L, 1=GF, 2=ARF */
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[1] = VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF;
+ layer_flags[2] = VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[3] =
+ layer_flags[5] = VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF;
+ layer_flags[4] = VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[6] = VP8_EFLAG_NO_REF_ARF |
+ VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_ARF;
+ layer_flags[7] = VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_ENTROPY;
+ break;
+ }
+ case 10:
+ {
+ // 3-layers structure where ARF is used as predictor for all frames,
+ // and is only updated on key frame.
+ // Sync points for layer 1 and 2 every 8 frames.
+
+ int ids[4] = {0,2,1,2};
+ cfg.ts_number_layers = 3;
+ cfg.ts_periodicity = 4;
+ cfg.ts_rate_decimator[0] = 4;
+ cfg.ts_rate_decimator[1] = 2;
+ cfg.ts_rate_decimator[2] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = 8;
+
+ /* 0=L, 1=GF, 2=ARF */
+
+ // Layer 0: predict from L and ARF; update L and G.
+ layer_flags[0] = VPX_EFLAG_FORCE_KF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_REF_GF;
+
+ // Layer 2: sync point: predict from L and ARF; update none.
+ layer_flags[1] = VP8_EFLAG_NO_REF_GF |
+ VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST |
+ VP8_EFLAG_NO_UPD_ENTROPY;
+
+ // Layer 1: sync point: predict from L and ARF; update G.
+ layer_flags[2] = VP8_EFLAG_NO_REF_GF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST;
+
+ // Layer 2: predict from L, G, ARF; update none.
+ layer_flags[3] = VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST |
+ VP8_EFLAG_NO_UPD_ENTROPY;
+
+ // Layer 0: predict from L and ARF; update L.
+ layer_flags[4] = VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_REF_GF;
+
+ // Layer 2: predict from L, G, ARF; update none.
+ layer_flags[5] = layer_flags[3];
+
+ // Layer 1: predict from L, G, ARF; update G.
+ layer_flags[6] = VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST;
+
+ // Layer 2: predict from L, G, ARF; update none.
+ layer_flags[7] = layer_flags[3];
+ break;
+ }
+ case 11:
+ default:
+ {
+ // 3-layers structure as in case 10, but no sync/refresh points for
+ // layer 1 and 2.
+
+ int ids[4] = {0,2,1,2};
+ cfg.ts_number_layers = 3;
+ cfg.ts_periodicity = 4;
+ cfg.ts_rate_decimator[0] = 4;
+ cfg.ts_rate_decimator[1] = 2;
+ cfg.ts_rate_decimator[2] = 1;
+ memcpy(cfg.ts_layer_id, ids, sizeof(ids));
+
+ flag_periodicity = 8;
+
+ /* 0=L, 1=GF, 2=ARF */
+
+ // Layer 0: predict from L and ARF; update L.
+ layer_flags[0] = VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_REF_GF;
+ layer_flags[4] = layer_flags[0];
+
+ // Layer 1: predict from L, G, ARF; update G.
+ layer_flags[2] = VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST;
+ layer_flags[6] = layer_flags[2];
+
+ // Layer 2: predict from L, G, ARF; update none.
+ layer_flags[1] = VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF |
+ VP8_EFLAG_NO_UPD_LAST |
+ VP8_EFLAG_NO_UPD_ENTROPY;
+ layer_flags[3] = layer_flags[1];
+ layer_flags[5] = layer_flags[1];
+ layer_flags[7] = layer_flags[1];
+ break;
+ }
+ }
+
+ /* Open input file */
+ if(!(infile = fopen(argv[1], "rb")))
+ die("Failed to open %s for reading", argv[1]);
+
+ /* Open an output file for each stream */
+ for (i=0; i<cfg.ts_number_layers; i++)
+ {
+ char file_name[512];
+ sprintf (file_name, "%s_%d.ivf", argv[2], i);
+ if (!(outfile[i] = fopen(file_name, "wb")))
+ die("Failed to open %s for writing", file_name);
+ write_ivf_file_header(outfile[i], &cfg, 0);
+ }
+
+ /* Initialize codec */
+ if (vpx_codec_enc_init (&codec, interface, &cfg, 0))
+ die_codec (&codec, "Failed to initialize encoder");
+
+ /* Cap CPU & first I-frame size */
+ vpx_codec_control (&codec, VP8E_SET_CPUUSED, -6);
+ vpx_codec_control (&codec, VP8E_SET_STATIC_THRESHOLD, 1);
+ vpx_codec_control (&codec, VP8E_SET_NOISE_SENSITIVITY, 1);
+ vpx_codec_control(&codec, VP8E_SET_TOKEN_PARTITIONS, 1);
+
+ max_intra_size_pct = (int) (((double)cfg.rc_buf_optimal_sz * 0.5)
+ * ((double) cfg.g_timebase.den / cfg.g_timebase.num)
+ / 10.0);
+ /* printf ("max_intra_size_pct=%d\n", max_intra_size_pct); */
+
+ vpx_codec_control(&codec, VP8E_SET_MAX_INTRA_BITRATE_PCT,
+ max_intra_size_pct);
+
+ frame_avail = 1;
+ while (frame_avail || got_data) {
+ vpx_codec_iter_t iter = NULL;
+ const vpx_codec_cx_pkt_t *pkt;
+
+ flags = layer_flags[frame_cnt % flag_periodicity];
+
+ frame_avail = read_frame(infile, &raw);
+ if (vpx_codec_encode(&codec, frame_avail? &raw : NULL, pts,
+ 1, flags, VPX_DL_REALTIME))
+ die_codec(&codec, "Failed to encode frame");
+
+ /* Reset KF flag */
+ if (layering_mode != 7)
+ layer_flags[0] &= ~VPX_EFLAG_FORCE_KF;
+
+ got_data = 0;
+ while ( (pkt = vpx_codec_get_cx_data(&codec, &iter)) ) {
+ got_data = 1;
+ switch (pkt->kind) {
+ case VPX_CODEC_CX_FRAME_PKT:
+ for (i=cfg.ts_layer_id[frame_cnt % cfg.ts_periodicity];
+ i<cfg.ts_number_layers; i++)
+ {
+ write_ivf_frame_header(outfile[i], pkt);
+ (void) fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
+ outfile[i]);
+ frames_in_layer[i]++;
+ }
+ break;
+ default:
+ break;
+ }
+ }
+ frame_cnt++;
+ pts += frame_duration;
+ }
+ fclose (infile);
+
+ printf ("Processed %d frames.\n",frame_cnt-1);
+ if (vpx_codec_destroy(&codec))
+ die_codec (&codec, "Failed to destroy codec");
+
+ /* Try to rewrite the output file headers with the actual frame count */
+ for (i=0; i<cfg.ts_number_layers; i++)
+ {
+ if (!fseek(outfile[i], 0, SEEK_SET))
+ write_ivf_file_header (outfile[i], &cfg, frames_in_layer[i]);
+ fclose (outfile[i]);
+ }
+
+ return EXIT_SUCCESS;
+}
diff --git a/vpx/exports_com b/libvpx/vpx/exports_com
similarity index 100%
rename from vpx/exports_com
rename to libvpx/vpx/exports_com
diff --git a/vpx/exports_dec b/libvpx/vpx/exports_dec
similarity index 100%
rename from vpx/exports_dec
rename to libvpx/vpx/exports_dec
diff --git a/vpx/exports_enc b/libvpx/vpx/exports_enc
similarity index 100%
rename from vpx/exports_enc
rename to libvpx/vpx/exports_enc
diff --git a/libvpx/vpx/internal/vpx_codec_internal.h b/libvpx/vpx/internal/vpx_codec_internal.h
new file mode 100644
index 0000000..4474331
--- /dev/null
+++ b/libvpx/vpx/internal/vpx_codec_internal.h
@@ -0,0 +1,490 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*!\file
+ * \brief Describes the decoder algorithm interface for algorithm
+ * implementations.
+ *
+ * This file defines the private structures and data types that are only
+ * relevant to implementing an algorithm, as opposed to using it.
+ *
+ * To create a decoder algorithm class, an interface structure is put
+ * into the global namespace:
+ * <pre>
+ * my_codec.c:
+ * vpx_codec_iface_t my_codec = {
+ * "My Codec v1.0",
+ * VPX_CODEC_ALG_ABI_VERSION,
+ * ...
+ * };
+ * </pre>
+ *
+ * An application instantiates a specific decoder instance by using
+ * vpx_codec_init() and a pointer to the algorithm's interface structure:
+ * <pre>
+ * my_app.c:
+ * extern vpx_codec_iface_t my_codec;
+ * {
+ * vpx_codec_ctx_t algo;
+ * res = vpx_codec_init(&algo, &my_codec);
+ * }
+ * </pre>
+ *
+ * Once initialized, the instance is manged using other functions from
+ * the vpx_codec_* family.
+ */
+#ifndef VPX_CODEC_INTERNAL_H
+#define VPX_CODEC_INTERNAL_H
+#include "../vpx_decoder.h"
+#include "../vpx_encoder.h"
+#include <stdarg.h>
+
+
+/*!\brief Current ABI version number
+ *
+ * \internal
+ * If this file is altered in any way that changes the ABI, this value
+ * must be bumped. Examples include, but are not limited to, changing
+ * types, removing or reassigning enums, adding/removing/rearranging
+ * fields to structures
+ */
+#define VPX_CODEC_INTERNAL_ABI_VERSION (4) /**<\hideinitializer*/
+
+typedef struct vpx_codec_alg_priv vpx_codec_alg_priv_t;
+typedef struct vpx_codec_priv_enc_mr_cfg vpx_codec_priv_enc_mr_cfg_t;
+
+/*!\brief init function pointer prototype
+ *
+ * Performs algorithm-specific initialization of the decoder context. This
+ * function is called by the generic vpx_codec_init() wrapper function, so
+ * plugins implementing this interface may trust the input parameters to be
+ * properly initialized.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \retval #VPX_CODEC_OK
+ * The input stream was recognized and decoder initialized.
+ * \retval #VPX_CODEC_MEM_ERROR
+ * Memory operation failed.
+ */
+typedef vpx_codec_err_t (*vpx_codec_init_fn_t)(vpx_codec_ctx_t *ctx,
+ vpx_codec_priv_enc_mr_cfg_t *data);
+
+/*!\brief destroy function pointer prototype
+ *
+ * Performs algorithm-specific destruction of the decoder context. This
+ * function is called by the generic vpx_codec_destroy() wrapper function,
+ * so plugins implementing this interface may trust the input parameters
+ * to be properly initialized.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \retval #VPX_CODEC_OK
+ * The input stream was recognized and decoder initialized.
+ * \retval #VPX_CODEC_MEM_ERROR
+ * Memory operation failed.
+ */
+typedef vpx_codec_err_t (*vpx_codec_destroy_fn_t)(vpx_codec_alg_priv_t *ctx);
+
+/*!\brief parse stream info function pointer prototype
+ *
+ * Performs high level parsing of the bitstream. This function is called by
+ * the generic vpx_codec_parse_stream() wrapper function, so plugins implementing
+ * this interface may trust the input parameters to be properly initialized.
+ *
+ * \param[in] data Pointer to a block of data to parse
+ * \param[in] data_sz Size of the data buffer
+ * \param[in,out] si Pointer to stream info to update. The size member
+ * \ref MUST be properly initialized, but \ref MAY be
+ * clobbered by the algorithm. This parameter \ref MAY
+ * be NULL.
+ *
+ * \retval #VPX_CODEC_OK
+ * Bitstream is parsable and stream information updated
+ */
+typedef vpx_codec_err_t (*vpx_codec_peek_si_fn_t)(const uint8_t *data,
+ unsigned int data_sz,
+ vpx_codec_stream_info_t *si);
+
+/*!\brief Return information about the current stream.
+ *
+ * Returns information about the stream that has been parsed during decoding.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in,out] si Pointer to stream info to update. The size member
+ * \ref MUST be properly initialized, but \ref MAY be
+ * clobbered by the algorithm. This parameter \ref MAY
+ * be NULL.
+ *
+ * \retval #VPX_CODEC_OK
+ * Bitstream is parsable and stream information updated
+ */
+typedef vpx_codec_err_t (*vpx_codec_get_si_fn_t)(vpx_codec_alg_priv_t *ctx,
+ vpx_codec_stream_info_t *si);
+
+/*!\brief control function pointer prototype
+ *
+ * This function is used to exchange algorithm specific data with the decoder
+ * instance. This can be used to implement features specific to a particular
+ * algorithm.
+ *
+ * This function is called by the generic vpx_codec_control() wrapper
+ * function, so plugins implementing this interface may trust the input
+ * parameters to be properly initialized. However, this interface does not
+ * provide type safety for the exchanged data or assign meanings to the
+ * control codes. Those details should be specified in the algorithm's
+ * header file. In particular, the ctrl_id parameter is guaranteed to exist
+ * in the algorithm's control mapping table, and the data parameter may be NULL.
+ *
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] ctrl_id Algorithm specific control identifier
+ * \param[in,out] data Data to exchange with algorithm instance.
+ *
+ * \retval #VPX_CODEC_OK
+ * The internal state data was deserialized.
+ */
+typedef vpx_codec_err_t (*vpx_codec_control_fn_t)(vpx_codec_alg_priv_t *ctx,
+ int ctrl_id,
+ va_list ap);
+
+/*!\brief control function pointer mapping
+ *
+ * This structure stores the mapping between control identifiers and
+ * implementing functions. Each algorithm provides a list of these
+ * mappings. This list is searched by the vpx_codec_control() wrapper
+ * function to determine which function to invoke. The special
+ * value {0, NULL} is used to indicate end-of-list, and must be
+ * present. The special value {0, <non-null>} can be used as a catch-all
+ * mapping. This implies that ctrl_id values chosen by the algorithm
+ * \ref MUST be non-zero.
+ */
+typedef const struct vpx_codec_ctrl_fn_map
+{
+ int ctrl_id;
+ vpx_codec_control_fn_t fn;
+} vpx_codec_ctrl_fn_map_t;
+
+/*!\brief decode data function pointer prototype
+ *
+ * Processes a buffer of coded data. If the processing results in a new
+ * decoded frame becoming available, #VPX_CODEC_CB_PUT_SLICE and
+ * #VPX_CODEC_CB_PUT_FRAME events are generated as appropriate. This
+ * function is called by the generic vpx_codec_decode() wrapper function,
+ * so plugins implementing this interface may trust the input parameters
+ * to be properly initialized.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] data Pointer to this block of new coded data. If
+ * NULL, a #VPX_CODEC_CB_PUT_FRAME event is posted
+ * for the previously decoded frame.
+ * \param[in] data_sz Size of the coded data, in bytes.
+ *
+ * \return Returns #VPX_CODEC_OK if the coded data was processed completely
+ * and future pictures can be decoded without error. Otherwise,
+ * see the descriptions of the other error codes in ::vpx_codec_err_t
+ * for recoverability capabilities.
+ */
+typedef vpx_codec_err_t (*vpx_codec_decode_fn_t)(vpx_codec_alg_priv_t *ctx,
+ const uint8_t *data,
+ unsigned int data_sz,
+ void *user_priv,
+ long deadline);
+
+/*!\brief Decoded frames iterator
+ *
+ * Iterates over a list of the frames available for display. The iterator
+ * storage should be initialized to NULL to start the iteration. Iteration is
+ * complete when this function returns NULL.
+ *
+ * The list of available frames becomes valid upon completion of the
+ * vpx_codec_decode call, and remains valid until the next call to vpx_codec_decode.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in out] iter Iterator storage, initialized to NULL
+ *
+ * \return Returns a pointer to an image, if one is ready for display. Frames
+ * produced will always be in PTS (presentation time stamp) order.
+ */
+typedef vpx_image_t*(*vpx_codec_get_frame_fn_t)(vpx_codec_alg_priv_t *ctx,
+ vpx_codec_iter_t *iter);
+
+
+/*\brief eXternal Memory Allocation memory map get iterator
+ *
+ * Iterates over a list of the memory maps requested by the decoder. The
+ * iterator storage should be initialized to NULL to start the iteration.
+ * Iteration is complete when this function returns NULL.
+ *
+ * \param[in out] iter Iterator storage, initialized to NULL
+ *
+ * \return Returns a pointer to an memory segment descriptor, or NULL to
+ * indicate end-of-list.
+ */
+typedef vpx_codec_err_t (*vpx_codec_get_mmap_fn_t)(const vpx_codec_ctx_t *ctx,
+ vpx_codec_mmap_t *mmap,
+ vpx_codec_iter_t *iter);
+
+
+/*\brief eXternal Memory Allocation memory map set iterator
+ *
+ * Sets a memory descriptor inside the decoder instance.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] mmap Memory map to store.
+ *
+ * \retval #VPX_CODEC_OK
+ * The memory map was accepted and stored.
+ * \retval #VPX_CODEC_MEM_ERROR
+ * The memory map was rejected.
+ */
+typedef vpx_codec_err_t (*vpx_codec_set_mmap_fn_t)(vpx_codec_ctx_t *ctx,
+ const vpx_codec_mmap_t *mmap);
+
+
+typedef vpx_codec_err_t (*vpx_codec_encode_fn_t)(vpx_codec_alg_priv_t *ctx,
+ const vpx_image_t *img,
+ vpx_codec_pts_t pts,
+ unsigned long duration,
+ vpx_enc_frame_flags_t flags,
+ unsigned long deadline);
+typedef const vpx_codec_cx_pkt_t*(*vpx_codec_get_cx_data_fn_t)(vpx_codec_alg_priv_t *ctx,
+ vpx_codec_iter_t *iter);
+
+typedef vpx_codec_err_t
+(*vpx_codec_enc_config_set_fn_t)(vpx_codec_alg_priv_t *ctx,
+ const vpx_codec_enc_cfg_t *cfg);
+typedef vpx_fixed_buf_t *
+(*vpx_codec_get_global_headers_fn_t)(vpx_codec_alg_priv_t *ctx);
+
+typedef vpx_image_t *
+(*vpx_codec_get_preview_frame_fn_t)(vpx_codec_alg_priv_t *ctx);
+
+typedef vpx_codec_err_t
+(*vpx_codec_enc_mr_get_mem_loc_fn_t)(const vpx_codec_enc_cfg_t *cfg,
+ void **mem_loc);
+
+/*!\brief usage configuration mapping
+ *
+ * This structure stores the mapping between usage identifiers and
+ * configuration structures. Each algorithm provides a list of these
+ * mappings. This list is searched by the vpx_codec_enc_config_default()
+ * wrapper function to determine which config to return. The special value
+ * {-1, {0}} is used to indicate end-of-list, and must be present. At least
+ * one mapping must be present, in addition to the end-of-list.
+ *
+ */
+typedef const struct vpx_codec_enc_cfg_map
+{
+ int usage;
+ vpx_codec_enc_cfg_t cfg;
+} vpx_codec_enc_cfg_map_t;
+
+#define NOT_IMPLEMENTED 0
+
+/*!\brief Decoder algorithm interface interface
+ *
+ * All decoders \ref MUST expose a variable of this type.
+ */
+struct vpx_codec_iface
+{
+ const char *name; /**< Identification String */
+ int abi_version; /**< Implemented ABI version */
+ vpx_codec_caps_t caps; /**< Decoder capabilities */
+ vpx_codec_init_fn_t init; /**< \copydoc ::vpx_codec_init_fn_t */
+ vpx_codec_destroy_fn_t destroy; /**< \copydoc ::vpx_codec_destroy_fn_t */
+ vpx_codec_ctrl_fn_map_t *ctrl_maps; /**< \copydoc ::vpx_codec_ctrl_fn_map_t */
+ vpx_codec_get_mmap_fn_t get_mmap; /**< \copydoc ::vpx_codec_get_mmap_fn_t */
+ vpx_codec_set_mmap_fn_t set_mmap; /**< \copydoc ::vpx_codec_set_mmap_fn_t */
+ struct vpx_codec_dec_iface
+ {
+ vpx_codec_peek_si_fn_t peek_si; /**< \copydoc ::vpx_codec_peek_si_fn_t */
+ vpx_codec_get_si_fn_t get_si; /**< \copydoc ::vpx_codec_peek_si_fn_t */
+ vpx_codec_decode_fn_t decode; /**< \copydoc ::vpx_codec_decode_fn_t */
+ vpx_codec_get_frame_fn_t get_frame; /**< \copydoc ::vpx_codec_get_frame_fn_t */
+ } dec;
+ struct vpx_codec_enc_iface
+ {
+ vpx_codec_enc_cfg_map_t *cfg_maps; /**< \copydoc ::vpx_codec_enc_cfg_map_t */
+ vpx_codec_encode_fn_t encode; /**< \copydoc ::vpx_codec_encode_fn_t */
+ vpx_codec_get_cx_data_fn_t get_cx_data; /**< \copydoc ::vpx_codec_get_cx_data_fn_t */
+ vpx_codec_enc_config_set_fn_t cfg_set; /**< \copydoc ::vpx_codec_enc_config_set_fn_t */
+ vpx_codec_get_global_headers_fn_t get_glob_hdrs; /**< \copydoc ::vpx_codec_get_global_headers_fn_t */
+ vpx_codec_get_preview_frame_fn_t get_preview; /**< \copydoc ::vpx_codec_get_preview_frame_fn_t */
+ vpx_codec_enc_mr_get_mem_loc_fn_t mr_get_mem_loc; /**< \copydoc ::vpx_codec_enc_mr_get_mem_loc_fn_t */
+ } enc;
+};
+
+/*!\brief Callback function pointer / user data pair storage */
+typedef struct vpx_codec_priv_cb_pair
+{
+ union
+ {
+ vpx_codec_put_frame_cb_fn_t put_frame;
+ vpx_codec_put_slice_cb_fn_t put_slice;
+ } u;
+ void *user_priv;
+} vpx_codec_priv_cb_pair_t;
+
+
+/*!\brief Instance private storage
+ *
+ * This structure is allocated by the algorithm's init function. It can be
+ * extended in one of two ways. First, a second, algorithm specific structure
+ * can be allocated and the priv member pointed to it. Alternatively, this
+ * structure can be made the first member of the algorithm specific structure,
+ * and the pointer cast to the proper type.
+ */
+struct vpx_codec_priv
+{
+ unsigned int sz;
+ vpx_codec_iface_t *iface;
+ struct vpx_codec_alg_priv *alg_priv;
+ const char *err_detail;
+ vpx_codec_flags_t init_flags;
+ struct
+ {
+ vpx_codec_priv_cb_pair_t put_frame_cb;
+ vpx_codec_priv_cb_pair_t put_slice_cb;
+ } dec;
+ struct
+ {
+ int tbd;
+ struct vpx_fixed_buf cx_data_dst_buf;
+ unsigned int cx_data_pad_before;
+ unsigned int cx_data_pad_after;
+ vpx_codec_cx_pkt_t cx_data_pkt;
+ unsigned int total_encoders;
+ } enc;
+};
+
+/*
+ * Multi-resolution encoding internal configuration
+ */
+struct vpx_codec_priv_enc_mr_cfg
+{
+ unsigned int mr_total_resolutions;
+ unsigned int mr_encoder_id;
+ struct vpx_rational mr_down_sampling_factor;
+ void* mr_low_res_mode_info;
+};
+
+#undef VPX_CTRL_USE_TYPE
+#define VPX_CTRL_USE_TYPE(id, typ) \
+ static typ id##__value(va_list args) {return va_arg(args, typ);} \
+ static typ id##__convert(void *x)\
+ {\
+ union\
+ {\
+ void *x;\
+ typ d;\
+ } u;\
+ u.x = x;\
+ return u.d;\
+ }
+
+
+#undef VPX_CTRL_USE_TYPE_DEPRECATED
+#define VPX_CTRL_USE_TYPE_DEPRECATED(id, typ) \
+ static typ id##__value(va_list args) {return va_arg(args, typ);} \
+ static typ id##__convert(void *x)\
+ {\
+ union\
+ {\
+ void *x;\
+ typ d;\
+ } u;\
+ u.x = x;\
+ return u.d;\
+ }
+
+#define CAST(id, arg) id##__value(arg)
+#define RECAST(id, x) id##__convert(x)
+
+
+/* CODEC_INTERFACE convenience macro
+ *
+ * By convention, each codec interface is a struct with extern linkage, where
+ * the symbol is suffixed with _algo. A getter function is also defined to
+ * return a pointer to the struct, since in some cases it's easier to work
+ * with text symbols than data symbols (see issue #169). This function has
+ * the same name as the struct, less the _algo suffix. The CODEC_INTERFACE
+ * macro is provided to define this getter function automatically.
+ */
+#define CODEC_INTERFACE(id)\
+vpx_codec_iface_t* id(void) { return &id##_algo; }\
+vpx_codec_iface_t id##_algo
+
+
+/* Internal Utility Functions
+ *
+ * The following functions are intended to be used inside algorithms as
+ * utilities for manipulating vpx_codec_* data structures.
+ */
+struct vpx_codec_pkt_list
+{
+ unsigned int cnt;
+ unsigned int max;
+ struct vpx_codec_cx_pkt pkts[1];
+};
+
+#define vpx_codec_pkt_list_decl(n)\
+ union {struct vpx_codec_pkt_list head;\
+ struct {struct vpx_codec_pkt_list head;\
+ struct vpx_codec_cx_pkt pkts[n];} alloc;}
+
+#define vpx_codec_pkt_list_init(m)\
+ (m)->alloc.head.cnt = 0,\
+ (m)->alloc.head.max = sizeof((m)->alloc.pkts) / sizeof((m)->alloc.pkts[0])
+
+int
+vpx_codec_pkt_list_add(struct vpx_codec_pkt_list *,
+ const struct vpx_codec_cx_pkt *);
+
+const vpx_codec_cx_pkt_t*
+vpx_codec_pkt_list_get(struct vpx_codec_pkt_list *list,
+ vpx_codec_iter_t *iter);
+
+
+#include <stdio.h>
+#include <setjmp.h>
+struct vpx_internal_error_info
+{
+ vpx_codec_err_t error_code;
+ int has_detail;
+ char detail[80];
+ int setjmp;
+ jmp_buf jmp;
+};
+
+static void vpx_internal_error(struct vpx_internal_error_info *info,
+ vpx_codec_err_t error,
+ const char *fmt,
+ ...)
+{
+ va_list ap;
+
+ info->error_code = error;
+ info->has_detail = 0;
+
+ if (fmt)
+ {
+ size_t sz = sizeof(info->detail);
+
+ info->has_detail = 1;
+ va_start(ap, fmt);
+ vsnprintf(info->detail, sz - 1, fmt, ap);
+ va_end(ap);
+ info->detail[sz-1] = '\0';
+ }
+
+ if (info->setjmp)
+ longjmp(info->jmp, info->error_code);
+}
+#endif
diff --git a/vpx/src/vpx_codec.c b/libvpx/vpx/src/vpx_codec.c
similarity index 100%
rename from vpx/src/vpx_codec.c
rename to libvpx/vpx/src/vpx_codec.c
diff --git a/libvpx/vpx/src/vpx_decoder.c b/libvpx/vpx/src/vpx_decoder.c
new file mode 100644
index 0000000..59a783d
--- /dev/null
+++ b/libvpx/vpx/src/vpx_decoder.c
@@ -0,0 +1,247 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*!\file
+ * \brief Provides the high level interface to wrap decoder algorithms.
+ *
+ */
+#include <string.h>
+#include "vpx/internal/vpx_codec_internal.h"
+
+#define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var)
+
+vpx_codec_err_t vpx_codec_dec_init_ver(vpx_codec_ctx_t *ctx,
+ vpx_codec_iface_t *iface,
+ vpx_codec_dec_cfg_t *cfg,
+ vpx_codec_flags_t flags,
+ int ver)
+{
+ vpx_codec_err_t res;
+
+ if (ver != VPX_DECODER_ABI_VERSION)
+ res = VPX_CODEC_ABI_MISMATCH;
+ else if (!ctx || !iface)
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (iface->abi_version != VPX_CODEC_INTERNAL_ABI_VERSION)
+ res = VPX_CODEC_ABI_MISMATCH;
+ else if ((flags & VPX_CODEC_USE_XMA) && !(iface->caps & VPX_CODEC_CAP_XMA))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_POSTPROC) && !(iface->caps & VPX_CODEC_CAP_POSTPROC))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_ERROR_CONCEALMENT) &&
+ !(iface->caps & VPX_CODEC_CAP_ERROR_CONCEALMENT))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_INPUT_FRAGMENTS) &&
+ !(iface->caps & VPX_CODEC_CAP_INPUT_FRAGMENTS))
+ res = VPX_CODEC_INCAPABLE;
+ else if (!(iface->caps & VPX_CODEC_CAP_DECODER))
+ res = VPX_CODEC_INCAPABLE;
+ else
+ {
+ memset(ctx, 0, sizeof(*ctx));
+ ctx->iface = iface;
+ ctx->name = iface->name;
+ ctx->priv = NULL;
+ ctx->init_flags = flags;
+ ctx->config.dec = cfg;
+ res = VPX_CODEC_OK;
+
+ if (!(flags & VPX_CODEC_USE_XMA))
+ {
+ res = ctx->iface->init(ctx, NULL);
+
+ if (res)
+ {
+ ctx->err_detail = ctx->priv ? ctx->priv->err_detail : NULL;
+ vpx_codec_destroy(ctx);
+ }
+
+ if (ctx->priv)
+ ctx->priv->iface = ctx->iface;
+ }
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
+
+
+vpx_codec_err_t vpx_codec_peek_stream_info(vpx_codec_iface_t *iface,
+ const uint8_t *data,
+ unsigned int data_sz,
+ vpx_codec_stream_info_t *si)
+{
+ vpx_codec_err_t res;
+
+ if (!iface || !data || !data_sz || !si
+ || si->sz < sizeof(vpx_codec_stream_info_t))
+ res = VPX_CODEC_INVALID_PARAM;
+ else
+ {
+ /* Set default/unknown values */
+ si->w = 0;
+ si->h = 0;
+
+ res = iface->dec.peek_si(data, data_sz, si);
+ }
+
+ return res;
+}
+
+
+vpx_codec_err_t vpx_codec_get_stream_info(vpx_codec_ctx_t *ctx,
+ vpx_codec_stream_info_t *si)
+{
+ vpx_codec_err_t res;
+
+ if (!ctx || !si || si->sz < sizeof(vpx_codec_stream_info_t))
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!ctx->iface || !ctx->priv)
+ res = VPX_CODEC_ERROR;
+ else
+ {
+ /* Set default/unknown values */
+ si->w = 0;
+ si->h = 0;
+
+ res = ctx->iface->dec.get_si(ctx->priv->alg_priv, si);
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
+
+
+vpx_codec_err_t vpx_codec_decode(vpx_codec_ctx_t *ctx,
+ const uint8_t *data,
+ unsigned int data_sz,
+ void *user_priv,
+ long deadline)
+{
+ vpx_codec_err_t res;
+
+ /* Sanity checks */
+ /* NULL data ptr allowed if data_sz is 0 too */
+ if (!ctx || (!data && data_sz))
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!ctx->iface || !ctx->priv)
+ res = VPX_CODEC_ERROR;
+ else
+ {
+ res = ctx->iface->dec.decode(ctx->priv->alg_priv, data, data_sz,
+ user_priv, deadline);
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
+
+vpx_image_t *vpx_codec_get_frame(vpx_codec_ctx_t *ctx,
+ vpx_codec_iter_t *iter)
+{
+ vpx_image_t *img;
+
+ if (!ctx || !iter || !ctx->iface || !ctx->priv)
+ img = NULL;
+ else
+ img = ctx->iface->dec.get_frame(ctx->priv->alg_priv, iter);
+
+ return img;
+}
+
+
+vpx_codec_err_t vpx_codec_register_put_frame_cb(vpx_codec_ctx_t *ctx,
+ vpx_codec_put_frame_cb_fn_t cb,
+ void *user_priv)
+{
+ vpx_codec_err_t res;
+
+ if (!ctx || !cb)
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!ctx->iface || !ctx->priv
+ || !(ctx->iface->caps & VPX_CODEC_CAP_PUT_FRAME))
+ res = VPX_CODEC_ERROR;
+ else
+ {
+ ctx->priv->dec.put_frame_cb.u.put_frame = cb;
+ ctx->priv->dec.put_frame_cb.user_priv = user_priv;
+ res = VPX_CODEC_OK;
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
+
+
+vpx_codec_err_t vpx_codec_register_put_slice_cb(vpx_codec_ctx_t *ctx,
+ vpx_codec_put_slice_cb_fn_t cb,
+ void *user_priv)
+{
+ vpx_codec_err_t res;
+
+ if (!ctx || !cb)
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!ctx->iface || !ctx->priv
+ || !(ctx->iface->caps & VPX_CODEC_CAP_PUT_FRAME))
+ res = VPX_CODEC_ERROR;
+ else
+ {
+ ctx->priv->dec.put_slice_cb.u.put_slice = cb;
+ ctx->priv->dec.put_slice_cb.user_priv = user_priv;
+ res = VPX_CODEC_OK;
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
+
+
+vpx_codec_err_t vpx_codec_get_mem_map(vpx_codec_ctx_t *ctx,
+ vpx_codec_mmap_t *mmap,
+ vpx_codec_iter_t *iter)
+{
+ vpx_codec_err_t res = VPX_CODEC_OK;
+
+ if (!ctx || !mmap || !iter || !ctx->iface)
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!(ctx->iface->caps & VPX_CODEC_CAP_XMA))
+ res = VPX_CODEC_ERROR;
+ else
+ res = ctx->iface->get_mmap(ctx, mmap, iter);
+
+ return SAVE_STATUS(ctx, res);
+}
+
+
+vpx_codec_err_t vpx_codec_set_mem_map(vpx_codec_ctx_t *ctx,
+ vpx_codec_mmap_t *mmap,
+ unsigned int num_maps)
+{
+ vpx_codec_err_t res = VPX_CODEC_MEM_ERROR;
+
+ if (!ctx || !mmap || !ctx->iface)
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!(ctx->iface->caps & VPX_CODEC_CAP_XMA))
+ res = VPX_CODEC_ERROR;
+ else
+ {
+ unsigned int i;
+
+ for (i = 0; i < num_maps; i++, mmap++)
+ {
+ if (!mmap->base)
+ break;
+
+ /* Everything look ok, set the mmap in the decoder */
+ res = ctx->iface->set_mmap(ctx, mmap);
+
+ if (res)
+ break;
+ }
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
diff --git a/libvpx/vpx/src/vpx_encoder.c b/libvpx/vpx/src/vpx_encoder.c
new file mode 100644
index 0000000..73c1c66
--- /dev/null
+++ b/libvpx/vpx/src/vpx_encoder.c
@@ -0,0 +1,442 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*!\file
+ * \brief Provides the high level interface to wrap encoder algorithms.
+ *
+ */
+#include <limits.h>
+#include <string.h>
+#include "vpx/internal/vpx_codec_internal.h"
+#include "vpx_config.h"
+
+#define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var)
+
+vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx,
+ vpx_codec_iface_t *iface,
+ vpx_codec_enc_cfg_t *cfg,
+ vpx_codec_flags_t flags,
+ int ver)
+{
+ vpx_codec_err_t res;
+
+ if (ver != VPX_ENCODER_ABI_VERSION)
+ res = VPX_CODEC_ABI_MISMATCH;
+ else if (!ctx || !iface || !cfg)
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (iface->abi_version != VPX_CODEC_INTERNAL_ABI_VERSION)
+ res = VPX_CODEC_ABI_MISMATCH;
+ else if (!(iface->caps & VPX_CODEC_CAP_ENCODER))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_XMA) && !(iface->caps & VPX_CODEC_CAP_XMA))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_PSNR)
+ && !(iface->caps & VPX_CODEC_CAP_PSNR))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_OUTPUT_PARTITION)
+ && !(iface->caps & VPX_CODEC_CAP_OUTPUT_PARTITION))
+ res = VPX_CODEC_INCAPABLE;
+ else
+ {
+ ctx->iface = iface;
+ ctx->name = iface->name;
+ ctx->priv = NULL;
+ ctx->init_flags = flags;
+ ctx->config.enc = cfg;
+ res = ctx->iface->init(ctx, NULL);
+
+ if (res)
+ {
+ ctx->err_detail = ctx->priv ? ctx->priv->err_detail : NULL;
+ vpx_codec_destroy(ctx);
+ }
+
+ if (ctx->priv)
+ ctx->priv->iface = ctx->iface;
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
+
+vpx_codec_err_t vpx_codec_enc_init_multi_ver(vpx_codec_ctx_t *ctx,
+ vpx_codec_iface_t *iface,
+ vpx_codec_enc_cfg_t *cfg,
+ int num_enc,
+ vpx_codec_flags_t flags,
+ vpx_rational_t *dsf,
+ int ver)
+{
+ vpx_codec_err_t res = 0;
+
+ if (ver != VPX_ENCODER_ABI_VERSION)
+ res = VPX_CODEC_ABI_MISMATCH;
+ else if (!ctx || !iface || !cfg || (num_enc > 16 || num_enc < 1))
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (iface->abi_version != VPX_CODEC_INTERNAL_ABI_VERSION)
+ res = VPX_CODEC_ABI_MISMATCH;
+ else if (!(iface->caps & VPX_CODEC_CAP_ENCODER))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_XMA) && !(iface->caps & VPX_CODEC_CAP_XMA))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_PSNR)
+ && !(iface->caps & VPX_CODEC_CAP_PSNR))
+ res = VPX_CODEC_INCAPABLE;
+ else if ((flags & VPX_CODEC_USE_OUTPUT_PARTITION)
+ && !(iface->caps & VPX_CODEC_CAP_OUTPUT_PARTITION))
+ res = VPX_CODEC_INCAPABLE;
+ else
+ {
+ int i;
+ void *mem_loc = NULL;
+
+ if(!(res = iface->enc.mr_get_mem_loc(cfg, &mem_loc)))
+ {
+ for (i = 0; i < num_enc; i++)
+ {
+ vpx_codec_priv_enc_mr_cfg_t mr_cfg;
+
+ /* Validate down-sampling factor. */
+ if(dsf->num < 1 || dsf->num > 4096 || dsf->den < 1 ||
+ dsf->den > dsf->num)
+ {
+ res = VPX_CODEC_INVALID_PARAM;
+ break;
+ }
+
+ mr_cfg.mr_low_res_mode_info = mem_loc;
+ mr_cfg.mr_total_resolutions = num_enc;
+ mr_cfg.mr_encoder_id = num_enc-1-i;
+ mr_cfg.mr_down_sampling_factor.num = dsf->num;
+ mr_cfg.mr_down_sampling_factor.den = dsf->den;
+
+ /* Force Key-frame synchronization. Namely, encoder at higher
+ * resolution always use the same frame_type chosen by the
+ * lowest-resolution encoder.
+ */
+ if(mr_cfg.mr_encoder_id)
+ cfg->kf_mode = VPX_KF_DISABLED;
+
+ ctx->iface = iface;
+ ctx->name = iface->name;
+ ctx->priv = NULL;
+ ctx->init_flags = flags;
+ ctx->config.enc = cfg;
+ res = ctx->iface->init(ctx, &mr_cfg);
+
+ if (res)
+ {
+ const char *error_detail =
+ ctx->priv ? ctx->priv->err_detail : NULL;
+ /* Destroy current ctx */
+ ctx->err_detail = error_detail;
+ vpx_codec_destroy(ctx);
+
+ /* Destroy already allocated high-level ctx */
+ while (i)
+ {
+ ctx--;
+ ctx->err_detail = error_detail;
+ vpx_codec_destroy(ctx);
+ i--;
+ }
+ }
+
+ if (ctx->priv)
+ ctx->priv->iface = ctx->iface;
+
+ if (res)
+ break;
+
+ ctx++;
+ cfg++;
+ dsf++;
+ }
+ }
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
+
+
+vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface,
+ vpx_codec_enc_cfg_t *cfg,
+ unsigned int usage)
+{
+ vpx_codec_err_t res;
+ vpx_codec_enc_cfg_map_t *map;
+
+ if (!iface || !cfg || usage > INT_MAX)
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!(iface->caps & VPX_CODEC_CAP_ENCODER))
+ res = VPX_CODEC_INCAPABLE;
+ else
+ {
+ res = VPX_CODEC_INVALID_PARAM;
+
+ for (map = iface->enc.cfg_maps; map->usage >= 0; map++)
+ {
+ if (map->usage == (int)usage)
+ {
+ *cfg = map->cfg;
+ cfg->g_usage = usage;
+ res = VPX_CODEC_OK;
+ break;
+ }
+ }
+ }
+
+ return res;
+}
+
+
+#if ARCH_X86 || ARCH_X86_64
+/* On X86, disable the x87 unit's internal 80 bit precision for better
+ * consistency with the SSE unit's 64 bit precision.
+ */
+#include "vpx_ports/x86.h"
+#define FLOATING_POINT_INIT() do {\
+ unsigned short x87_orig_mode = x87_set_double_precision();
+#define FLOATING_POINT_RESTORE() \
+ x87_set_control_word(x87_orig_mode); }while(0)
+
+
+#else
+static void FLOATING_POINT_INIT() {}
+static void FLOATING_POINT_RESTORE() {}
+#endif
+
+
+vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx,
+ const vpx_image_t *img,
+ vpx_codec_pts_t pts,
+ unsigned long duration,
+ vpx_enc_frame_flags_t flags,
+ unsigned long deadline)
+{
+ vpx_codec_err_t res = 0;
+
+ if (!ctx || (img && !duration))
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!ctx->iface || !ctx->priv)
+ res = VPX_CODEC_ERROR;
+ else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
+ res = VPX_CODEC_INCAPABLE;
+ else
+ {
+ /* Execute in a normalized floating point environment, if the platform
+ * requires it.
+ */
+ unsigned int num_enc =ctx->priv->enc.total_encoders;
+
+ FLOATING_POINT_INIT();
+
+ if (num_enc == 1)
+ res = ctx->iface->enc.encode(ctx->priv->alg_priv, img, pts,
+ duration, flags, deadline);
+ else
+ {
+ /* Multi-resolution encoding:
+ * Encode multi-levels in reverse order. For example,
+ * if mr_total_resolutions = 3, first encode level 2,
+ * then encode level 1, and finally encode level 0.
+ */
+ int i;
+
+ ctx += num_enc - 1;
+ if (img) img += num_enc - 1;
+
+ for (i = num_enc-1; i >= 0; i--)
+ {
+ if ((res = ctx->iface->enc.encode(ctx->priv->alg_priv, img, pts,
+ duration, flags, deadline)))
+ break;
+
+ ctx--;
+ if (img) img--;
+ }
+ ctx++;
+ }
+
+ FLOATING_POINT_RESTORE();
+ }
+
+ return SAVE_STATUS(ctx, res);
+}
+
+
+const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx,
+ vpx_codec_iter_t *iter)
+{
+ const vpx_codec_cx_pkt_t *pkt = NULL;
+
+ if (ctx)
+ {
+ if (!iter)
+ ctx->err = VPX_CODEC_INVALID_PARAM;
+ else if (!ctx->iface || !ctx->priv)
+ ctx->err = VPX_CODEC_ERROR;
+ else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
+ ctx->err = VPX_CODEC_INCAPABLE;
+ else
+ pkt = ctx->iface->enc.get_cx_data(ctx->priv->alg_priv, iter);
+ }
+
+ if (pkt && pkt->kind == VPX_CODEC_CX_FRAME_PKT)
+ {
+ /* If the application has specified a destination area for the
+ * compressed data, and the codec has not placed the data there,
+ * and it fits, copy it.
+ */
+ char *dst_buf = ctx->priv->enc.cx_data_dst_buf.buf;
+
+ if (dst_buf
+ && pkt->data.raw.buf != dst_buf
+ && pkt->data.raw.sz
+ + ctx->priv->enc.cx_data_pad_before
+ + ctx->priv->enc.cx_data_pad_after
+ <= ctx->priv->enc.cx_data_dst_buf.sz)
+ {
+ vpx_codec_cx_pkt_t *modified_pkt = &ctx->priv->enc.cx_data_pkt;
+
+ memcpy(dst_buf + ctx->priv->enc.cx_data_pad_before,
+ pkt->data.raw.buf, pkt->data.raw.sz);
+ *modified_pkt = *pkt;
+ modified_pkt->data.raw.buf = dst_buf;
+ modified_pkt->data.raw.sz += ctx->priv->enc.cx_data_pad_before
+ + ctx->priv->enc.cx_data_pad_after;
+ pkt = modified_pkt;
+ }
+
+ if (dst_buf == pkt->data.raw.buf)
+ {
+ ctx->priv->enc.cx_data_dst_buf.buf = dst_buf + pkt->data.raw.sz;
+ ctx->priv->enc.cx_data_dst_buf.sz -= pkt->data.raw.sz;
+ }
+ }
+
+ return pkt;
+}
+
+
+vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx,
+ const vpx_fixed_buf_t *buf,
+ unsigned int pad_before,
+ unsigned int pad_after)
+{
+ if (!ctx || !ctx->priv)
+ return VPX_CODEC_INVALID_PARAM;
+
+ if (buf)
+ {
+ ctx->priv->enc.cx_data_dst_buf = *buf;
+ ctx->priv->enc.cx_data_pad_before = pad_before;
+ ctx->priv->enc.cx_data_pad_after = pad_after;
+ }
+ else
+ {
+ ctx->priv->enc.cx_data_dst_buf.buf = NULL;
+ ctx->priv->enc.cx_data_dst_buf.sz = 0;
+ ctx->priv->enc.cx_data_pad_before = 0;
+ ctx->priv->enc.cx_data_pad_after = 0;
+ }
+
+ return VPX_CODEC_OK;
+}
+
+
+const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx)
+{
+ vpx_image_t *img = NULL;
+
+ if (ctx)
+ {
+ if (!ctx->iface || !ctx->priv)
+ ctx->err = VPX_CODEC_ERROR;
+ else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
+ ctx->err = VPX_CODEC_INCAPABLE;
+ else if (!ctx->iface->enc.get_preview)
+ ctx->err = VPX_CODEC_INCAPABLE;
+ else
+ img = ctx->iface->enc.get_preview(ctx->priv->alg_priv);
+ }
+
+ return img;
+}
+
+
+vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx)
+{
+ vpx_fixed_buf_t *buf = NULL;
+
+ if (ctx)
+ {
+ if (!ctx->iface || !ctx->priv)
+ ctx->err = VPX_CODEC_ERROR;
+ else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
+ ctx->err = VPX_CODEC_INCAPABLE;
+ else if (!ctx->iface->enc.get_glob_hdrs)
+ ctx->err = VPX_CODEC_INCAPABLE;
+ else
+ buf = ctx->iface->enc.get_glob_hdrs(ctx->priv->alg_priv);
+ }
+
+ return buf;
+}
+
+
+vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx,
+ const vpx_codec_enc_cfg_t *cfg)
+{
+ vpx_codec_err_t res;
+
+ if (!ctx || !ctx->iface || !ctx->priv || !cfg)
+ res = VPX_CODEC_INVALID_PARAM;
+ else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
+ res = VPX_CODEC_INCAPABLE;
+ else
+ res = ctx->iface->enc.cfg_set(ctx->priv->alg_priv, cfg);
+
+ return SAVE_STATUS(ctx, res);
+}
+
+
+int vpx_codec_pkt_list_add(struct vpx_codec_pkt_list *list,
+ const struct vpx_codec_cx_pkt *pkt)
+{
+ if (list->cnt < list->max)
+ {
+ list->pkts[list->cnt++] = *pkt;
+ return 0;
+ }
+
+ return 1;
+}
+
+
+const vpx_codec_cx_pkt_t *vpx_codec_pkt_list_get(struct vpx_codec_pkt_list *list,
+ vpx_codec_iter_t *iter)
+{
+ const vpx_codec_cx_pkt_t *pkt;
+
+ if (!(*iter))
+ {
+ *iter = list->pkts;
+ }
+
+ pkt = (const void *) * iter;
+
+ if ((size_t)(pkt - list->pkts) < list->cnt)
+ *iter = pkt + 1;
+ else
+ pkt = NULL;
+
+ return pkt;
+}
diff --git a/libvpx/vpx/src/vpx_image.c b/libvpx/vpx/src/vpx_image.c
new file mode 100644
index 0000000..336b6e2
--- /dev/null
+++ b/libvpx/vpx/src/vpx_image.c
@@ -0,0 +1,305 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <stdlib.h>
+#include <string.h>
+#include "vpx/vpx_image.h"
+
+#define ADDRESS_STORAGE_SIZE sizeof(size_t)
+/*returns an addr aligned to the byte boundary specified by align*/
+#define align_addr(addr,align) (void*)(((size_t)(addr) + ((align) - 1)) & (size_t)-(align))
+
+/* Memalign code is copied from vpx_mem.c */
+static void *img_buf_memalign(size_t align, size_t size)
+{
+ void *addr,
+ * x = NULL;
+
+ addr = malloc(size + align - 1 + ADDRESS_STORAGE_SIZE);
+
+ if (addr)
+ {
+ x = align_addr((unsigned char *)addr + ADDRESS_STORAGE_SIZE, (int)align);
+ /* save the actual malloc address */
+ ((size_t *)x)[-1] = (size_t)addr;
+ }
+
+ return x;
+}
+
+static void img_buf_free(void *memblk)
+{
+ if (memblk)
+ {
+ void *addr = (void *)(((size_t *)memblk)[-1]);
+ free(addr);
+ }
+}
+
+static vpx_image_t *img_alloc_helper(vpx_image_t *img,
+ vpx_img_fmt_t fmt,
+ unsigned int d_w,
+ unsigned int d_h,
+ unsigned int buf_align,
+ unsigned int stride_align,
+ unsigned char *img_data)
+{
+
+ unsigned int h, w, s, xcs, ycs, bps;
+ int align;
+
+ /* Treat align==0 like align==1 */
+ if (!buf_align)
+ buf_align = 1;
+
+ /* Validate alignment (must be power of 2) */
+ if (buf_align & (buf_align - 1))
+ goto fail;
+
+ /* Treat align==0 like align==1 */
+ if (!stride_align)
+ stride_align = 1;
+
+ /* Validate alignment (must be power of 2) */
+ if (stride_align & (stride_align - 1))
+ goto fail;
+
+ /* Get sample size for this format */
+ switch (fmt)
+ {
+ case VPX_IMG_FMT_RGB32:
+ case VPX_IMG_FMT_RGB32_LE:
+ case VPX_IMG_FMT_ARGB:
+ case VPX_IMG_FMT_ARGB_LE:
+ bps = 32;
+ break;
+ case VPX_IMG_FMT_RGB24:
+ case VPX_IMG_FMT_BGR24:
+ bps = 24;
+ break;
+ case VPX_IMG_FMT_RGB565:
+ case VPX_IMG_FMT_RGB565_LE:
+ case VPX_IMG_FMT_RGB555:
+ case VPX_IMG_FMT_RGB555_LE:
+ case VPX_IMG_FMT_UYVY:
+ case VPX_IMG_FMT_YUY2:
+ case VPX_IMG_FMT_YVYU:
+ bps = 16;
+ break;
+ case VPX_IMG_FMT_I420:
+ case VPX_IMG_FMT_YV12:
+ case VPX_IMG_FMT_VPXI420:
+ case VPX_IMG_FMT_VPXYV12:
+ bps = 12;
+ break;
+ default:
+ bps = 16;
+ break;
+ }
+
+ /* Get chroma shift values for this format */
+ switch (fmt)
+ {
+ case VPX_IMG_FMT_I420:
+ case VPX_IMG_FMT_YV12:
+ case VPX_IMG_FMT_VPXI420:
+ case VPX_IMG_FMT_VPXYV12:
+ xcs = 1;
+ break;
+ default:
+ xcs = 0;
+ break;
+ }
+
+ switch (fmt)
+ {
+ case VPX_IMG_FMT_I420:
+ case VPX_IMG_FMT_YV12:
+ case VPX_IMG_FMT_VPXI420:
+ case VPX_IMG_FMT_VPXYV12:
+ ycs = 1;
+ break;
+ default:
+ ycs = 0;
+ break;
+ }
+
+ /* Calculate storage sizes given the chroma subsampling */
+ align = (1 << xcs) - 1;
+ w = (d_w + align) & ~align;
+ align = (1 << ycs) - 1;
+ h = (d_h + align) & ~align;
+ s = (fmt & VPX_IMG_FMT_PLANAR) ? w : bps * w / 8;
+ s = (s + stride_align - 1) & ~(stride_align - 1);
+
+ /* Allocate the new image */
+ if (!img)
+ {
+ img = (vpx_image_t *)calloc(1, sizeof(vpx_image_t));
+
+ if (!img)
+ goto fail;
+
+ img->self_allocd = 1;
+ }
+ else
+ {
+ memset(img, 0, sizeof(vpx_image_t));
+ }
+
+ img->img_data = img_data;
+
+ if (!img_data)
+ {
+ img->img_data = img_buf_memalign(buf_align, ((fmt & VPX_IMG_FMT_PLANAR)?
+ h * s * bps / 8 : h * s));
+ img->img_data_owner = 1;
+ }
+
+ if (!img->img_data)
+ goto fail;
+
+ img->fmt = fmt;
+ img->w = w;
+ img->h = h;
+ img->x_chroma_shift = xcs;
+ img->y_chroma_shift = ycs;
+ img->bps = bps;
+
+ /* Calculate strides */
+ img->stride[VPX_PLANE_Y] = img->stride[VPX_PLANE_ALPHA] = s;
+ img->stride[VPX_PLANE_U] = img->stride[VPX_PLANE_V] = s >> xcs;
+
+ /* Default viewport to entire image */
+ if (!vpx_img_set_rect(img, 0, 0, d_w, d_h))
+ return img;
+
+fail:
+ vpx_img_free(img);
+ return NULL;
+}
+
+vpx_image_t *vpx_img_alloc(vpx_image_t *img,
+ vpx_img_fmt_t fmt,
+ unsigned int d_w,
+ unsigned int d_h,
+ unsigned int align)
+{
+ return img_alloc_helper(img, fmt, d_w, d_h, align, align, NULL);
+}
+
+vpx_image_t *vpx_img_wrap(vpx_image_t *img,
+ vpx_img_fmt_t fmt,
+ unsigned int d_w,
+ unsigned int d_h,
+ unsigned int stride_align,
+ unsigned char *img_data)
+{
+ /* By setting buf_align = 1, we don't change buffer alignment in this
+ * function. */
+ return img_alloc_helper(img, fmt, d_w, d_h, 1, stride_align, img_data);
+}
+
+int vpx_img_set_rect(vpx_image_t *img,
+ unsigned int x,
+ unsigned int y,
+ unsigned int w,
+ unsigned int h)
+{
+ unsigned char *data;
+
+ if (x + w <= img->w && y + h <= img->h)
+ {
+ img->d_w = w;
+ img->d_h = h;
+
+ /* Calculate plane pointers */
+ if (!(img->fmt & VPX_IMG_FMT_PLANAR))
+ {
+ img->planes[VPX_PLANE_PACKED] =
+ img->img_data + x * img->bps / 8 + y * img->stride[VPX_PLANE_PACKED];
+ }
+ else
+ {
+ data = img->img_data;
+
+ if (img->fmt & VPX_IMG_FMT_HAS_ALPHA)
+ {
+ img->planes[VPX_PLANE_ALPHA] =
+ data + x + y * img->stride[VPX_PLANE_ALPHA];
+ data += img->h * img->stride[VPX_PLANE_ALPHA];
+ }
+
+ img->planes[VPX_PLANE_Y] = data + x + y * img->stride[VPX_PLANE_Y];
+ data += img->h * img->stride[VPX_PLANE_Y];
+
+ if (!(img->fmt & VPX_IMG_FMT_UV_FLIP))
+ {
+ img->planes[VPX_PLANE_U] = data
+ + (x >> img->x_chroma_shift)
+ + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U];
+ data += (img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_U];
+ img->planes[VPX_PLANE_V] = data
+ + (x >> img->x_chroma_shift)
+ + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_V];
+ }
+ else
+ {
+ img->planes[VPX_PLANE_V] = data
+ + (x >> img->x_chroma_shift)
+ + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_V];
+ data += (img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_V];
+ img->planes[VPX_PLANE_U] = data
+ + (x >> img->x_chroma_shift)
+ + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U];
+ }
+ }
+
+ return 0;
+ }
+
+ return -1;
+}
+
+void vpx_img_flip(vpx_image_t *img)
+{
+ /* Note: In the calculation pointer adjustment calculation, we want the
+ * rhs to be promoted to a signed type. Section 6.3.1.8 of the ISO C99
+ * standard indicates that if the adjustment parameter is unsigned, the
+ * stride parameter will be promoted to unsigned, causing errors when
+ * the lhs is a larger type than the rhs.
+ */
+ img->planes[VPX_PLANE_Y] += (signed)(img->d_h - 1) * img->stride[VPX_PLANE_Y];
+ img->stride[VPX_PLANE_Y] = -img->stride[VPX_PLANE_Y];
+
+ img->planes[VPX_PLANE_U] += (signed)((img->d_h >> img->y_chroma_shift) - 1)
+ * img->stride[VPX_PLANE_U];
+ img->stride[VPX_PLANE_U] = -img->stride[VPX_PLANE_U];
+
+ img->planes[VPX_PLANE_V] += (signed)((img->d_h >> img->y_chroma_shift) - 1)
+ * img->stride[VPX_PLANE_V];
+ img->stride[VPX_PLANE_V] = -img->stride[VPX_PLANE_V];
+
+ img->planes[VPX_PLANE_ALPHA] += (signed)(img->d_h - 1) * img->stride[VPX_PLANE_ALPHA];
+ img->stride[VPX_PLANE_ALPHA] = -img->stride[VPX_PLANE_ALPHA];
+}
+
+void vpx_img_free(vpx_image_t *img)
+{
+ if (img)
+ {
+ if (img->img_data && img->img_data_owner)
+ img_buf_free(img->img_data);
+
+ if (img->self_allocd)
+ free(img);
+ }
+}
diff --git a/libvpx/vpx/vp8.h b/libvpx/vpx/vp8.h
new file mode 100644
index 0000000..2952203
--- /dev/null
+++ b/libvpx/vpx/vp8.h
@@ -0,0 +1,123 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*!\defgroup vp8 VP8
+ * \ingroup codecs
+ * VP8 is vpx's newest video compression algorithm that uses motion
+ * compensated prediction, Discrete Cosine Transform (DCT) coding of the
+ * prediction error signal and context dependent entropy coding techniques
+ * based on arithmetic principles. It features:
+ * - YUV 4:2:0 image format
+ * - Macro-block based coding (16x16 luma plus two 8x8 chroma)
+ * - 1/4 (1/8) pixel accuracy motion compensated prediction
+ * - 4x4 DCT transform
+ * - 128 level linear quantizer
+ * - In loop deblocking filter
+ * - Context-based entropy coding
+ *
+ * @{
+ */
+/*!\file
+ * \brief Provides controls common to both the VP8 encoder and decoder.
+ */
+#ifndef VP8_H
+#define VP8_H
+#include "vpx_codec_impl_top.h"
+
+/*!\brief Control functions
+ *
+ * The set of macros define the control functions of VP8 interface
+ */
+enum vp8_com_control_id
+{
+ VP8_SET_REFERENCE = 1, /**< pass in an external frame into decoder to be used as reference frame */
+ VP8_COPY_REFERENCE = 2, /**< get a copy of reference frame from the decoder */
+ VP8_SET_POSTPROC = 3, /**< set the decoder's post processing settings */
+ VP8_SET_DBG_COLOR_REF_FRAME = 4, /**< set the reference frames to color for each macroblock */
+ VP8_SET_DBG_COLOR_MB_MODES = 5, /**< set which macro block modes to color */
+ VP8_SET_DBG_COLOR_B_MODES = 6, /**< set which blocks modes to color */
+ VP8_SET_DBG_DISPLAY_MV = 7, /**< set which motion vector modes to draw */
+ VP8_COMMON_CTRL_ID_MAX,
+ VP8_DECODER_CTRL_ID_START = 256
+};
+
+/*!\brief post process flags
+ *
+ * The set of macros define VP8 decoder post processing flags
+ */
+enum vp8_postproc_level
+{
+ VP8_NOFILTERING = 0,
+ VP8_DEBLOCK = 1<<0,
+ VP8_DEMACROBLOCK = 1<<1,
+ VP8_ADDNOISE = 1<<2,
+ VP8_DEBUG_TXT_FRAME_INFO = 1<<3, /**< print frame information */
+ VP8_DEBUG_TXT_MBLK_MODES = 1<<4, /**< print macro block modes over each macro block */
+ VP8_DEBUG_TXT_DC_DIFF = 1<<5, /**< print dc diff for each macro block */
+ VP8_DEBUG_TXT_RATE_INFO = 1<<6, /**< print video rate info (encoder only) */
+ VP8_MFQE = 1<<10
+};
+
+/*!\brief post process flags
+ *
+ * This define a structure that describe the post processing settings. For
+ * the best objective measure (using the PSNR metric) set post_proc_flag
+ * to VP8_DEBLOCK and deblocking_level to 1.
+ */
+
+typedef struct vp8_postproc_cfg
+{
+ int post_proc_flag; /**< the types of post processing to be done, should be combination of "vp8_postproc_level" */
+ int deblocking_level; /**< the strength of deblocking, valid range [0, 16] */
+ int noise_level; /**< the strength of additive noise, valid range [0, 16] */
+} vp8_postproc_cfg_t;
+
+/*!\brief reference frame type
+ *
+ * The set of macros define the type of VP8 reference frames
+ */
+typedef enum vpx_ref_frame_type
+{
+ VP8_LAST_FRAME = 1,
+ VP8_GOLD_FRAME = 2,
+ VP8_ALTR_FRAME = 4
+} vpx_ref_frame_type_t;
+
+/*!\brief reference frame data struct
+ *
+ * define the data struct to access vp8 reference frames
+ */
+
+typedef struct vpx_ref_frame
+{
+ vpx_ref_frame_type_t frame_type; /**< which reference frame */
+ vpx_image_t img; /**< reference frame data in image format */
+} vpx_ref_frame_t;
+
+
+/*!\brief vp8 decoder control function parameter type
+ *
+ * defines the data type for each of VP8 decoder control function requires
+ */
+
+VPX_CTRL_USE_TYPE(VP8_SET_REFERENCE, vpx_ref_frame_t *)
+VPX_CTRL_USE_TYPE(VP8_COPY_REFERENCE, vpx_ref_frame_t *)
+VPX_CTRL_USE_TYPE(VP8_SET_POSTPROC, vp8_postproc_cfg_t *)
+VPX_CTRL_USE_TYPE(VP8_SET_DBG_COLOR_REF_FRAME, int)
+VPX_CTRL_USE_TYPE(VP8_SET_DBG_COLOR_MB_MODES, int)
+VPX_CTRL_USE_TYPE(VP8_SET_DBG_COLOR_B_MODES, int)
+VPX_CTRL_USE_TYPE(VP8_SET_DBG_DISPLAY_MV, int)
+
+
+/*! @} - end defgroup vp8 */
+
+#include "vpx_codec_impl_bottom.h"
+#endif
diff --git a/libvpx/vpx/vp8cx.h b/libvpx/vpx/vp8cx.h
new file mode 100644
index 0000000..a3c95d2
--- /dev/null
+++ b/libvpx/vpx/vp8cx.h
@@ -0,0 +1,305 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*!\defgroup vp8_encoder WebM VP8 Encoder
+ * \ingroup vp8
+ *
+ * @{
+ */
+#include "vp8.h"
+
+/*!\file
+ * \brief Provides definitions for using the VP8 encoder algorithm within the
+ * vpx Codec Interface.
+ */
+#ifndef VP8CX_H
+#define VP8CX_H
+#include "vpx_codec_impl_top.h"
+
+/*!\name Algorithm interface for VP8
+ *
+ * This interface provides the capability to encode raw VP8 streams, as would
+ * be found in AVI files.
+ * @{
+ */
+extern vpx_codec_iface_t vpx_codec_vp8_cx_algo;
+extern vpx_codec_iface_t* vpx_codec_vp8_cx(void);
+/*!@} - end algorithm interface member group*/
+
+
+/*
+ * Algorithm Flags
+ */
+
+/*!\brief Don't reference the last frame
+ *
+ * When this flag is set, the encoder will not use the last frame as a
+ * predictor. When not set, the encoder will choose whether to use the
+ * last frame or not automatically.
+ */
+#define VP8_EFLAG_NO_REF_LAST (1<<16)
+
+
+/*!\brief Don't reference the golden frame
+ *
+ * When this flag is set, the encoder will not use the golden frame as a
+ * predictor. When not set, the encoder will choose whether to use the
+ * golden frame or not automatically.
+ */
+#define VP8_EFLAG_NO_REF_GF (1<<17)
+
+
+/*!\brief Don't reference the alternate reference frame
+ *
+ * When this flag is set, the encoder will not use the alt ref frame as a
+ * predictor. When not set, the encoder will choose whether to use the
+ * alt ref frame or not automatically.
+ */
+#define VP8_EFLAG_NO_REF_ARF (1<<21)
+
+
+/*!\brief Don't update the last frame
+ *
+ * When this flag is set, the encoder will not update the last frame with
+ * the contents of the current frame.
+ */
+#define VP8_EFLAG_NO_UPD_LAST (1<<18)
+
+
+/*!\brief Don't update the golden frame
+ *
+ * When this flag is set, the encoder will not update the golden frame with
+ * the contents of the current frame.
+ */
+#define VP8_EFLAG_NO_UPD_GF (1<<22)
+
+
+/*!\brief Don't update the alternate reference frame
+ *
+ * When this flag is set, the encoder will not update the alt ref frame with
+ * the contents of the current frame.
+ */
+#define VP8_EFLAG_NO_UPD_ARF (1<<23)
+
+
+/*!\brief Force golden frame update
+ *
+ * When this flag is set, the encoder copy the contents of the current frame
+ * to the golden frame buffer.
+ */
+#define VP8_EFLAG_FORCE_GF (1<<19)
+
+
+/*!\brief Force alternate reference frame update
+ *
+ * When this flag is set, the encoder copy the contents of the current frame
+ * to the alternate reference frame buffer.
+ */
+#define VP8_EFLAG_FORCE_ARF (1<<24)
+
+
+/*!\brief Disable entropy update
+ *
+ * When this flag is set, the encoder will not update its internal entropy
+ * model based on the entropy of this frame.
+ */
+#define VP8_EFLAG_NO_UPD_ENTROPY (1<<20)
+
+
+/*!\brief VP8 encoder control functions
+ *
+ * This set of macros define the control functions available for the VP8
+ * encoder interface.
+ *
+ * \sa #vpx_codec_control
+ */
+enum vp8e_enc_control_id
+{
+ VP8E_UPD_ENTROPY = 5, /**< control function to set mode of entropy update in encoder */
+ VP8E_UPD_REFERENCE, /**< control function to set reference update mode in encoder */
+ VP8E_USE_REFERENCE, /**< control function to set which reference frame encoder can use */
+ VP8E_SET_ROI_MAP, /**< control function to pass an ROI map to encoder */
+ VP8E_SET_ACTIVEMAP, /**< control function to pass an Active map to encoder */
+ VP8E_SET_SCALEMODE = 11, /**< control function to set encoder scaling mode */
+ /*!\brief control function to set vp8 encoder cpuused
+ *
+ * Changes in this value influences, among others, the encoder's selection
+ * of motion estimation methods. Values greater than 0 will increase encoder
+ * speed at the expense of quality.
+ * The full set of adjustments can be found in
+ * onyx_if.c:vp8_set_speed_features().
+ * \todo List highlights of the changes at various levels.
+ *
+ * \note Valid range: -16..16
+ */
+ VP8E_SET_CPUUSED = 13,
+ VP8E_SET_ENABLEAUTOALTREF, /**< control function to enable vp8 to automatic set and use altref frame */
+ VP8E_SET_NOISE_SENSITIVITY, /**< control function to set noise sensitivity */
+ VP8E_SET_SHARPNESS, /**< control function to set sharpness */
+ VP8E_SET_STATIC_THRESHOLD, /**< control function to set the threshold for macroblocks treated static */
+ VP8E_SET_TOKEN_PARTITIONS, /**< control function to set the number of token partitions */
+ VP8E_GET_LAST_QUANTIZER, /**< return the quantizer chosen by the
+ encoder for the last frame using the internal
+ scale */
+ VP8E_GET_LAST_QUANTIZER_64, /**< return the quantizer chosen by the
+ encoder for the last frame, using the 0..63
+ scale as used by the rc_*_quantizer config
+ parameters */
+ VP8E_SET_ARNR_MAXFRAMES, /**< control function to set the max number of frames blurred creating arf*/
+ VP8E_SET_ARNR_STRENGTH , /**< control function to set the filter strength for the arf */
+ VP8E_SET_ARNR_TYPE , /**< control function to set the type of filter to use for the arf*/
+ VP8E_SET_TUNING, /**< control function to set visual tuning */
+ /*!\brief control function to set constrained quality level
+ *
+ * \attention For this value to be used vpx_codec_enc_cfg_t::g_usage must be
+ * set to #VPX_CQ.
+ * \note Valid range: 0..63
+ */
+ VP8E_SET_CQ_LEVEL,
+
+ /*!\brief Max data rate for Intra frames
+ *
+ * This value controls additional clamping on the maximum size of a
+ * keyframe. It is expressed as a percentage of the average
+ * per-frame bitrate, with the special (and default) value 0 meaning
+ * unlimited, or no additional clamping beyond the codec's built-in
+ * algorithm.
+ *
+ * For example, to allocate no more than 4.5 frames worth of bitrate
+ * to a keyframe, set this to 450.
+ *
+ */
+ VP8E_SET_MAX_INTRA_BITRATE_PCT
+};
+
+/*!\brief vpx 1-D scaling mode
+ *
+ * This set of constants define 1-D vpx scaling modes
+ */
+typedef enum vpx_scaling_mode_1d
+{
+ VP8E_NORMAL = 0,
+ VP8E_FOURFIVE = 1,
+ VP8E_THREEFIVE = 2,
+ VP8E_ONETWO = 3
+} VPX_SCALING_MODE;
+
+
+/*!\brief vpx region of interest map
+ *
+ * These defines the data structures for the region of interest map
+ *
+ */
+
+typedef struct vpx_roi_map
+{
+ unsigned char *roi_map; /**< specify an id between 0 and 3 for each 16x16 region within a frame */
+ unsigned int rows; /**< number of rows */
+ unsigned int cols; /**< number of cols */
+ int delta_q[4]; /**< quantizer delta [-63, 63] off baseline for regions with id between 0 and 3*/
+ int delta_lf[4]; /**< loop filter strength delta [-63, 63] for regions with id between 0 and 3 */
+ unsigned int static_threshold[4];/**< threshold for region to be treated as static */
+} vpx_roi_map_t;
+
+/*!\brief vpx active region map
+ *
+ * These defines the data structures for active region map
+ *
+ */
+
+
+typedef struct vpx_active_map
+{
+ unsigned char *active_map; /**< specify an on (1) or off (0) each 16x16 region within a frame */
+ unsigned int rows; /**< number of rows */
+ unsigned int cols; /**< number of cols */
+} vpx_active_map_t;
+
+/*!\brief vpx image scaling mode
+ *
+ * This defines the data structure for image scaling mode
+ *
+ */
+typedef struct vpx_scaling_mode
+{
+ VPX_SCALING_MODE h_scaling_mode; /**< horizontal scaling mode */
+ VPX_SCALING_MODE v_scaling_mode; /**< vertical scaling mode */
+} vpx_scaling_mode_t;
+
+/*!\brief VP8 token partition mode
+ *
+ * This defines VP8 partitioning mode for compressed data, i.e., the number of
+ * sub-streams in the bitstream. Used for parallelized decoding.
+ *
+ */
+
+typedef enum
+{
+ VP8_ONE_TOKENPARTITION = 0,
+ VP8_TWO_TOKENPARTITION = 1,
+ VP8_FOUR_TOKENPARTITION = 2,
+ VP8_EIGHT_TOKENPARTITION = 3
+} vp8e_token_partitions;
+
+
+/*!\brief VP8 model tuning parameters
+ *
+ * Changes the encoder to tune for certain types of input material.
+ *
+ */
+typedef enum
+{
+ VP8_TUNE_PSNR,
+ VP8_TUNE_SSIM
+} vp8e_tuning;
+
+
+/*!\brief VP8 encoder control function parameter type
+ *
+ * Defines the data types that VP8E control functions take. Note that
+ * additional common controls are defined in vp8.h
+ *
+ */
+
+
+/* These controls have been deprecated in favor of the flags parameter to
+ * vpx_codec_encode(). See the definition of VP8_EFLAG_* above.
+ */
+VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_UPD_ENTROPY, int)
+VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_UPD_REFERENCE, int)
+VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_USE_REFERENCE, int)
+
+VPX_CTRL_USE_TYPE(VP8E_SET_ROI_MAP, vpx_roi_map_t *)
+VPX_CTRL_USE_TYPE(VP8E_SET_ACTIVEMAP, vpx_active_map_t *)
+VPX_CTRL_USE_TYPE(VP8E_SET_SCALEMODE, vpx_scaling_mode_t *)
+
+VPX_CTRL_USE_TYPE(VP8E_SET_CPUUSED, int)
+VPX_CTRL_USE_TYPE(VP8E_SET_ENABLEAUTOALTREF, unsigned int)
+VPX_CTRL_USE_TYPE(VP8E_SET_NOISE_SENSITIVITY, unsigned int)
+VPX_CTRL_USE_TYPE(VP8E_SET_SHARPNESS, unsigned int)
+VPX_CTRL_USE_TYPE(VP8E_SET_STATIC_THRESHOLD, unsigned int)
+VPX_CTRL_USE_TYPE(VP8E_SET_TOKEN_PARTITIONS, int) /* vp8e_token_partitions */
+
+VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_MAXFRAMES, unsigned int)
+VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_STRENGTH , unsigned int)
+VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_TYPE , unsigned int)
+VPX_CTRL_USE_TYPE(VP8E_SET_TUNING, int) /* vp8e_tuning */
+VPX_CTRL_USE_TYPE(VP8E_SET_CQ_LEVEL , unsigned int)
+
+VPX_CTRL_USE_TYPE(VP8E_GET_LAST_QUANTIZER, int *)
+VPX_CTRL_USE_TYPE(VP8E_GET_LAST_QUANTIZER_64, int *)
+
+VPX_CTRL_USE_TYPE(VP8E_SET_MAX_INTRA_BITRATE_PCT, unsigned int)
+
+
+/*! @} - end defgroup vp8_encoder */
+#include "vpx_codec_impl_bottom.h"
+#endif
diff --git a/libvpx/vpx/vp8dx.h b/libvpx/vpx/vp8dx.h
new file mode 100644
index 0000000..8661035
--- /dev/null
+++ b/libvpx/vpx/vp8dx.h
@@ -0,0 +1,83 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8.h"
+
+/*!\defgroup vp8_decoder WebM VP8 Decoder
+ * \ingroup vp8
+ *
+ * @{
+ */
+/*!\file
+ * \brief Provides definitions for using the VP8 algorithm within the vpx Decoder
+ * interface.
+ */
+#ifndef VP8DX_H
+#define VP8DX_H
+#include "vpx_codec_impl_top.h"
+
+/*!\name Algorithm interface for VP8
+ *
+ * This interface provides the capability to decode raw VP8 streams, as would
+ * be found in AVI files and other non-Flash uses.
+ * @{
+ */
+extern vpx_codec_iface_t vpx_codec_vp8_dx_algo;
+extern vpx_codec_iface_t* vpx_codec_vp8_dx(void);
+/*!@} - end algorithm interface member group*/
+
+/* Include controls common to both the encoder and decoder */
+#include "vp8.h"
+
+
+/*!\brief VP8 decoder control functions
+ *
+ * This set of macros define the control functions available for the VP8
+ * decoder interface.
+ *
+ * \sa #vpx_codec_control
+ */
+enum vp8_dec_control_id
+{
+ /** control function to get info on which reference frames were updated
+ * by the last decode
+ */
+ VP8D_GET_LAST_REF_UPDATES = VP8_DECODER_CTRL_ID_START,
+
+ /** check if the indicated frame is corrupted */
+ VP8D_GET_FRAME_CORRUPTED,
+
+ /** control function to get info on which reference frames were used
+ * by the last decode
+ */
+ VP8D_GET_LAST_REF_USED,
+
+ VP8_DECODER_CTRL_ID_MAX
+} ;
+
+
+/*!\brief VP8 decoder control function parameter type
+ *
+ * Defines the data types that VP8D control functions take. Note that
+ * additional common controls are defined in vp8.h
+ *
+ */
+
+
+VPX_CTRL_USE_TYPE(VP8D_GET_LAST_REF_UPDATES, int *)
+VPX_CTRL_USE_TYPE(VP8D_GET_FRAME_CORRUPTED, int *)
+VPX_CTRL_USE_TYPE(VP8D_GET_LAST_REF_USED, int *)
+
+/*! @} - end defgroup vp8_decoder */
+
+
+#include "vpx_codec_impl_bottom.h"
+#endif
diff --git a/vpx/vpx_codec.h b/libvpx/vpx/vpx_codec.h
similarity index 100%
rename from vpx/vpx_codec.h
rename to libvpx/vpx/vpx_codec.h
diff --git a/libvpx/vpx/vpx_codec.mk b/libvpx/vpx/vpx_codec.mk
new file mode 100644
index 0000000..ffa123f
--- /dev/null
+++ b/libvpx/vpx/vpx_codec.mk
@@ -0,0 +1,41 @@
+##
+## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+API_EXPORTS += exports
+
+API_SRCS-$(CONFIG_VP8_ENCODER) += vp8.h
+API_SRCS-$(CONFIG_VP8_ENCODER) += vp8cx.h
+API_DOC_SRCS-$(CONFIG_VP8_ENCODER) += vp8.h
+API_DOC_SRCS-$(CONFIG_VP8_ENCODER) += vp8cx.h
+
+API_SRCS-$(CONFIG_VP8_DECODER) += vp8.h
+API_SRCS-$(CONFIG_VP8_DECODER) += vp8dx.h
+API_DOC_SRCS-$(CONFIG_VP8_DECODER) += vp8.h
+API_DOC_SRCS-$(CONFIG_VP8_DECODER) += vp8dx.h
+
+API_DOC_SRCS-yes += vpx_codec.h
+API_DOC_SRCS-yes += vpx_decoder.h
+API_DOC_SRCS-yes += vpx_encoder.h
+API_DOC_SRCS-yes += vpx_image.h
+
+API_SRCS-yes += src/vpx_decoder.c
+API_SRCS-yes += vpx_decoder.h
+API_SRCS-yes += src/vpx_encoder.c
+API_SRCS-yes += vpx_encoder.h
+API_SRCS-yes += internal/vpx_codec_internal.h
+API_SRCS-yes += src/vpx_codec.c
+API_SRCS-yes += src/vpx_image.c
+API_SRCS-yes += vpx_codec.h
+API_SRCS-yes += vpx_codec.mk
+API_SRCS-yes += vpx_codec_impl_bottom.h
+API_SRCS-yes += vpx_codec_impl_top.h
+API_SRCS-yes += vpx_image.h
+API_SRCS-$(BUILD_LIBVPX) += vpx_integer.h
diff --git a/vpx/vpx_codec_impl_bottom.h b/libvpx/vpx/vpx_codec_impl_bottom.h
similarity index 100%
rename from vpx/vpx_codec_impl_bottom.h
rename to libvpx/vpx/vpx_codec_impl_bottom.h
diff --git a/vpx/vpx_codec_impl_top.h b/libvpx/vpx/vpx_codec_impl_top.h
similarity index 100%
rename from vpx/vpx_codec_impl_top.h
rename to libvpx/vpx/vpx_codec_impl_top.h
diff --git a/libvpx/vpx/vpx_decoder.h b/libvpx/vpx/vpx_decoder.h
new file mode 100644
index 0000000..1ccf1c5
--- /dev/null
+++ b/libvpx/vpx/vpx_decoder.h
@@ -0,0 +1,333 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*!\defgroup decoder Decoder Algorithm Interface
+ * \ingroup codec
+ * This abstraction allows applications using this decoder to easily support
+ * multiple video formats with minimal code duplication. This section describes
+ * the interface common to all decoders.
+ * @{
+ */
+
+/*!\file
+ * \brief Describes the decoder algorithm interface to applications.
+ *
+ * This file describes the interface between an application and a
+ * video decoder algorithm.
+ *
+ */
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef VPX_DECODER_H
+#define VPX_DECODER_H
+#include "vpx_codec.h"
+
+ /*!\brief Current ABI version number
+ *
+ * \internal
+ * If this file is altered in any way that changes the ABI, this value
+ * must be bumped. Examples include, but are not limited to, changing
+ * types, removing or reassigning enums, adding/removing/rearranging
+ * fields to structures
+ */
+#define VPX_DECODER_ABI_VERSION (2 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/
+
+ /*! \brief Decoder capabilities bitfield
+ *
+ * Each decoder advertises the capabilities it supports as part of its
+ * ::vpx_codec_iface_t interface structure. Capabilities are extra interfaces
+ * or functionality, and are not required to be supported by a decoder.
+ *
+ * The available flags are specified by VPX_CODEC_CAP_* defines.
+ */
+#define VPX_CODEC_CAP_PUT_SLICE 0x10000 /**< Will issue put_slice callbacks */
+#define VPX_CODEC_CAP_PUT_FRAME 0x20000 /**< Will issue put_frame callbacks */
+#define VPX_CODEC_CAP_POSTPROC 0x40000 /**< Can postprocess decoded frame */
+#define VPX_CODEC_CAP_ERROR_CONCEALMENT 0x80000 /**< Can conceal errors due to
+ packet loss */
+#define VPX_CODEC_CAP_INPUT_FRAGMENTS 0x100000 /**< Can receive encoded frames
+ one fragment at a time */
+
+ /*! \brief Initialization-time Feature Enabling
+ *
+ * Certain codec features must be known at initialization time, to allow for
+ * proper memory allocation.
+ *
+ * The available flags are specified by VPX_CODEC_USE_* defines.
+ */
+#define VPX_CODEC_USE_POSTPROC 0x10000 /**< Postprocess decoded frame */
+#define VPX_CODEC_USE_ERROR_CONCEALMENT 0x20000 /**< Conceal errors in decoded
+ frames */
+#define VPX_CODEC_USE_INPUT_FRAGMENTS 0x40000 /**< The input frame should be
+ passed to the decoder one
+ fragment at a time */
+
+ /*!\brief Stream properties
+ *
+ * This structure is used to query or set properties of the decoded
+ * stream. Algorithms may extend this structure with data specific
+ * to their bitstream by setting the sz member appropriately.
+ */
+ typedef struct vpx_codec_stream_info
+ {
+ unsigned int sz; /**< Size of this structure */
+ unsigned int w; /**< Width (or 0 for unknown/default) */
+ unsigned int h; /**< Height (or 0 for unknown/default) */
+ unsigned int is_kf; /**< Current frame is a keyframe */
+ } vpx_codec_stream_info_t;
+
+ /* REQUIRED FUNCTIONS
+ *
+ * The following functions are required to be implemented for all decoders.
+ * They represent the base case functionality expected of all decoders.
+ */
+
+
+ /*!\brief Initialization Configurations
+ *
+ * This structure is used to pass init time configuration options to the
+ * decoder.
+ */
+ typedef struct vpx_codec_dec_cfg
+ {
+ unsigned int threads; /**< Maximum number of threads to use, default 1 */
+ unsigned int w; /**< Width */
+ unsigned int h; /**< Height */
+ } vpx_codec_dec_cfg_t; /**< alias for struct vpx_codec_dec_cfg */
+
+
+ /*!\brief Initialize a decoder instance
+ *
+ * Initializes a decoder context using the given interface. Applications
+ * should call the vpx_codec_dec_init convenience macro instead of this
+ * function directly, to ensure that the ABI version number parameter
+ * is properly initialized.
+ *
+ * If the library was configured with --disable-multithread, this call
+ * is not thread safe and should be guarded with a lock if being used
+ * in a multithreaded context.
+ *
+ * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
+ * parameter), the storage pointed to by the cfg parameter must be
+ * kept readable and stable until all memory maps have been set.
+ *
+ * \param[in] ctx Pointer to this instance's context.
+ * \param[in] iface Pointer to the algorithm interface to use.
+ * \param[in] cfg Configuration to use, if known. May be NULL.
+ * \param[in] flags Bitfield of VPX_CODEC_USE_* flags
+ * \param[in] ver ABI version number. Must be set to
+ * VPX_DECODER_ABI_VERSION
+ * \retval #VPX_CODEC_OK
+ * The decoder algorithm initialized.
+ * \retval #VPX_CODEC_MEM_ERROR
+ * Memory allocation failed.
+ */
+ vpx_codec_err_t vpx_codec_dec_init_ver(vpx_codec_ctx_t *ctx,
+ vpx_codec_iface_t *iface,
+ vpx_codec_dec_cfg_t *cfg,
+ vpx_codec_flags_t flags,
+ int ver);
+
+ /*!\brief Convenience macro for vpx_codec_dec_init_ver()
+ *
+ * Ensures the ABI version parameter is properly set.
+ */
+#define vpx_codec_dec_init(ctx, iface, cfg, flags) \
+ vpx_codec_dec_init_ver(ctx, iface, cfg, flags, VPX_DECODER_ABI_VERSION)
+
+
+ /*!\brief Parse stream info from a buffer
+ *
+ * Performs high level parsing of the bitstream. Construction of a decoder
+ * context is not necessary. Can be used to determine if the bitstream is
+ * of the proper format, and to extract information from the stream.
+ *
+ * \param[in] iface Pointer to the algorithm interface
+ * \param[in] data Pointer to a block of data to parse
+ * \param[in] data_sz Size of the data buffer
+ * \param[in,out] si Pointer to stream info to update. The size member
+ * \ref MUST be properly initialized, but \ref MAY be
+ * clobbered by the algorithm. This parameter \ref MAY
+ * be NULL.
+ *
+ * \retval #VPX_CODEC_OK
+ * Bitstream is parsable and stream information updated
+ */
+ vpx_codec_err_t vpx_codec_peek_stream_info(vpx_codec_iface_t *iface,
+ const uint8_t *data,
+ unsigned int data_sz,
+ vpx_codec_stream_info_t *si);
+
+
+ /*!\brief Return information about the current stream.
+ *
+ * Returns information about the stream that has been parsed during decoding.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in,out] si Pointer to stream info to update. The size member
+ * \ref MUST be properly initialized, but \ref MAY be
+ * clobbered by the algorithm. This parameter \ref MAY
+ * be NULL.
+ *
+ * \retval #VPX_CODEC_OK
+ * Bitstream is parsable and stream information updated
+ */
+ vpx_codec_err_t vpx_codec_get_stream_info(vpx_codec_ctx_t *ctx,
+ vpx_codec_stream_info_t *si);
+
+
+ /*!\brief Decode data
+ *
+ * Processes a buffer of coded data. If the processing results in a new
+ * decoded frame becoming available, PUT_SLICE and PUT_FRAME events may be
+ * generated, as appropriate. Encoded data \ref MUST be passed in DTS (decode
+ * time stamp) order. Frames produced will always be in PTS (presentation
+ * time stamp) order.
+ * If the decoder is configured with VPX_CODEC_USE_INPUT_FRAGMENTS enabled,
+ * data and data_sz can contain a fragment of the encoded frame. Fragment
+ * \#n must contain at least partition \#n, but can also contain subsequent
+ * partitions (\#n+1 - \#n+i), and if so, fragments \#n+1, .., \#n+i must
+ * be empty. When no more data is available, this function should be called
+ * with NULL as data and 0 as data_sz. The memory passed to this function
+ * must be available until the frame has been decoded.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] data Pointer to this block of new coded data. If
+ * NULL, a VPX_CODEC_CB_PUT_FRAME event is posted
+ * for the previously decoded frame.
+ * \param[in] data_sz Size of the coded data, in bytes.
+ * \param[in] user_priv Application specific data to associate with
+ * this frame.
+ * \param[in] deadline Soft deadline the decoder should attempt to meet,
+ * in us. Set to zero for unlimited.
+ *
+ * \return Returns #VPX_CODEC_OK if the coded data was processed completely
+ * and future pictures can be decoded without error. Otherwise,
+ * see the descriptions of the other error codes in ::vpx_codec_err_t
+ * for recoverability capabilities.
+ */
+ vpx_codec_err_t vpx_codec_decode(vpx_codec_ctx_t *ctx,
+ const uint8_t *data,
+ unsigned int data_sz,
+ void *user_priv,
+ long deadline);
+
+
+ /*!\brief Decoded frames iterator
+ *
+ * Iterates over a list of the frames available for display. The iterator
+ * storage should be initialized to NULL to start the iteration. Iteration is
+ * complete when this function returns NULL.
+ *
+ * The list of available frames becomes valid upon completion of the
+ * vpx_codec_decode call, and remains valid until the next call to vpx_codec_decode.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in,out] iter Iterator storage, initialized to NULL
+ *
+ * \return Returns a pointer to an image, if one is ready for display. Frames
+ * produced will always be in PTS (presentation time stamp) order.
+ */
+ vpx_image_t *vpx_codec_get_frame(vpx_codec_ctx_t *ctx,
+ vpx_codec_iter_t *iter);
+
+
+ /*!\defgroup cap_put_frame Frame-Based Decoding Functions
+ *
+ * The following functions are required to be implemented for all decoders
+ * that advertise the VPX_CODEC_CAP_PUT_FRAME capability. Calling these functions
+ * for codecs that don't advertise this capability will result in an error
+ * code being returned, usually VPX_CODEC_ERROR
+ * @{
+ */
+
+ /*!\brief put frame callback prototype
+ *
+ * This callback is invoked by the decoder to notify the application of
+ * the availability of decoded image data.
+ */
+ typedef void (*vpx_codec_put_frame_cb_fn_t)(void *user_priv,
+ const vpx_image_t *img);
+
+
+ /*!\brief Register for notification of frame completion.
+ *
+ * Registers a given function to be called when a decoded frame is
+ * available.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] cb Pointer to the callback function
+ * \param[in] user_priv User's private data
+ *
+ * \retval #VPX_CODEC_OK
+ * Callback successfully registered.
+ * \retval #VPX_CODEC_ERROR
+ * Decoder context not initialized, or algorithm not capable of
+ * posting slice completion.
+ */
+ vpx_codec_err_t vpx_codec_register_put_frame_cb(vpx_codec_ctx_t *ctx,
+ vpx_codec_put_frame_cb_fn_t cb,
+ void *user_priv);
+
+
+ /*!@} - end defgroup cap_put_frame */
+
+ /*!\defgroup cap_put_slice Slice-Based Decoding Functions
+ *
+ * The following functions are required to be implemented for all decoders
+ * that advertise the VPX_CODEC_CAP_PUT_SLICE capability. Calling these functions
+ * for codecs that don't advertise this capability will result in an error
+ * code being returned, usually VPX_CODEC_ERROR
+ * @{
+ */
+
+ /*!\brief put slice callback prototype
+ *
+ * This callback is invoked by the decoder to notify the application of
+ * the availability of partially decoded image data. The
+ */
+ typedef void (*vpx_codec_put_slice_cb_fn_t)(void *user_priv,
+ const vpx_image_t *img,
+ const vpx_image_rect_t *valid,
+ const vpx_image_rect_t *update);
+
+
+ /*!\brief Register for notification of slice completion.
+ *
+ * Registers a given function to be called when a decoded slice is
+ * available.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] cb Pointer to the callback function
+ * \param[in] user_priv User's private data
+ *
+ * \retval #VPX_CODEC_OK
+ * Callback successfully registered.
+ * \retval #VPX_CODEC_ERROR
+ * Decoder context not initialized, or algorithm not capable of
+ * posting slice completion.
+ */
+ vpx_codec_err_t vpx_codec_register_put_slice_cb(vpx_codec_ctx_t *ctx,
+ vpx_codec_put_slice_cb_fn_t cb,
+ void *user_priv);
+
+
+ /*!@} - end defgroup cap_put_slice*/
+
+ /*!@} - end defgroup decoder*/
+
+#endif
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/libvpx/vpx/vpx_encoder.h b/libvpx/vpx/vpx_encoder.h
new file mode 100644
index 0000000..67d9033
--- /dev/null
+++ b/libvpx/vpx/vpx_encoder.h
@@ -0,0 +1,935 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*!\defgroup encoder Encoder Algorithm Interface
+ * \ingroup codec
+ * This abstraction allows applications using this encoder to easily support
+ * multiple video formats with minimal code duplication. This section describes
+ * the interface common to all encoders.
+ * @{
+ */
+
+/*!\file
+ * \brief Describes the encoder algorithm interface to applications.
+ *
+ * This file describes the interface between an application and a
+ * video encoder algorithm.
+ *
+ */
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef VPX_ENCODER_H
+#define VPX_ENCODER_H
+#include "vpx_codec.h"
+
+/*! Temporal Scalability: Maximum length of the sequence defining frame
+ * layer membership
+ */
+#define VPX_TS_MAX_PERIODICITY 16
+
+/*! Temporal Scalability: Maximum number of coding layers */
+#define VPX_TS_MAX_LAYERS 5
+
+/*!\deprecated Use #VPX_TS_MAX_PERIODICITY instead. */
+#define MAX_PERIODICITY VPX_TS_MAX_PERIODICITY
+
+/*!\deprecated Use #VPX_TS_MAX_LAYERS instead. */
+#define MAX_LAYERS VPX_TS_MAX_LAYERS
+
+ /*!\brief Current ABI version number
+ *
+ * \internal
+ * If this file is altered in any way that changes the ABI, this value
+ * must be bumped. Examples include, but are not limited to, changing
+ * types, removing or reassigning enums, adding/removing/rearranging
+ * fields to structures
+ */
+#define VPX_ENCODER_ABI_VERSION (3 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/
+
+
+ /*! \brief Encoder capabilities bitfield
+ *
+ * Each encoder advertises the capabilities it supports as part of its
+ * ::vpx_codec_iface_t interface structure. Capabilities are extra
+ * interfaces or functionality, and are not required to be supported
+ * by an encoder.
+ *
+ * The available flags are specified by VPX_CODEC_CAP_* defines.
+ */
+#define VPX_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */
+
+ /*! Can output one partition at a time. Each partition is returned in its
+ * own VPX_CODEC_CX_FRAME_PKT, with the FRAME_IS_FRAGMENT flag set for
+ * every partition but the last. In this mode all frames are always
+ * returned partition by partition.
+ */
+#define VPX_CODEC_CAP_OUTPUT_PARTITION 0x20000
+
+
+ /*! \brief Initialization-time Feature Enabling
+ *
+ * Certain codec features must be known at initialization time, to allow
+ * for proper memory allocation.
+ *
+ * The available flags are specified by VPX_CODEC_USE_* defines.
+ */
+#define VPX_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */
+#define VPX_CODEC_USE_OUTPUT_PARTITION 0x20000 /**< Make the encoder output one
+ partition at a time. */
+
+
+ /*!\brief Generic fixed size buffer structure
+ *
+ * This structure is able to hold a reference to any fixed size buffer.
+ */
+ typedef struct vpx_fixed_buf
+ {
+ void *buf; /**< Pointer to the data */
+ size_t sz; /**< Length of the buffer, in chars */
+ } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */
+
+
+ /*!\brief Time Stamp Type
+ *
+ * An integer, which when multiplied by the stream's time base, provides
+ * the absolute time of a sample.
+ */
+ typedef int64_t vpx_codec_pts_t;
+
+
+ /*!\brief Compressed Frame Flags
+ *
+ * This type represents a bitfield containing information about a compressed
+ * frame that may be useful to an application. The most significant 16 bits
+ * can be used by an algorithm to provide additional detail, for example to
+ * support frame types that are codec specific (MPEG-1 D-frames for example)
+ */
+ typedef uint32_t vpx_codec_frame_flags_t;
+#define VPX_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */
+#define VPX_FRAME_IS_DROPPABLE 0x2 /**< frame can be dropped without affecting
+ the stream (no future frame depends on
+ this one) */
+#define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not
+ be shown */
+#define VPX_FRAME_IS_FRAGMENT 0x8 /**< this is a fragment of the encoded
+ frame */
+
+ /*!\brief Error Resilient flags
+ *
+ * These flags define which error resilient features to enable in the
+ * encoder. The flags are specified through the
+ * vpx_codec_enc_cfg::g_error_resilient variable.
+ */
+ typedef uint32_t vpx_codec_er_flags_t;
+#define VPX_ERROR_RESILIENT_DEFAULT 0x1 /**< Improve resiliency against
+ losses of whole frames */
+#define VPX_ERROR_RESILIENT_PARTITIONS 0x2 /**< The frame partitions are
+ independently decodable by the
+ bool decoder, meaning that
+ partitions can be decoded even
+ though earlier partitions have
+ been lost. Note that intra
+ predicition is still done over
+ the partition boundary. */
+
+ /*!\brief Encoder output packet variants
+ *
+ * This enumeration lists the different kinds of data packets that can be
+ * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY
+ * extend this list to provide additional functionality.
+ */
+ enum vpx_codec_cx_pkt_kind
+ {
+ VPX_CODEC_CX_FRAME_PKT, /**< Compressed video frame */
+ VPX_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */
+ VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */
+ VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */
+ };
+
+
+ /*!\brief Encoder output packet
+ *
+ * This structure contains the different kinds of output data the encoder
+ * may produce while compressing a frame.
+ */
+ typedef struct vpx_codec_cx_pkt
+ {
+ enum vpx_codec_cx_pkt_kind kind; /**< packet variant */
+ union
+ {
+ struct
+ {
+ void *buf; /**< compressed data buffer */
+ size_t sz; /**< length of compressed data */
+ vpx_codec_pts_t pts; /**< time stamp to show frame
+ (in timebase units) */
+ unsigned long duration; /**< duration to show frame
+ (in timebase units) */
+ vpx_codec_frame_flags_t flags; /**< flags for this frame */
+ int partition_id; /**< the partition id
+ defines the decoding order
+ of the partitions. Only
+ applicable when "output partition"
+ mode is enabled. First partition
+ has id 0.*/
+
+ } frame; /**< data for compressed frame packet */
+ struct vpx_fixed_buf twopass_stats; /**< data for two-pass packet */
+ struct vpx_psnr_pkt
+ {
+ unsigned int samples[4]; /**< Number of samples, total/y/u/v */
+ uint64_t sse[4]; /**< sum squared error, total/y/u/v */
+ double psnr[4]; /**< PSNR, total/y/u/v */
+ } psnr; /**< data for PSNR packet */
+ struct vpx_fixed_buf raw; /**< data for arbitrary packets */
+
+ /* This packet size is fixed to allow codecs to extend this
+ * interface without having to manage storage for raw packets,
+ * i.e., if it's smaller than 128 bytes, you can store in the
+ * packet list directly.
+ */
+ char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */
+ } data; /**< packet data */
+ } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */
+
+
+ /*!\brief Rational Number
+ *
+ * This structure holds a fractional value.
+ */
+ typedef struct vpx_rational
+ {
+ int num; /**< fraction numerator */
+ int den; /**< fraction denominator */
+ } vpx_rational_t; /**< alias for struct vpx_rational */
+
+
+ /*!\brief Multi-pass Encoding Pass */
+ enum vpx_enc_pass
+ {
+ VPX_RC_ONE_PASS, /**< Single pass mode */
+ VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */
+ VPX_RC_LAST_PASS /**< Final pass of multi-pass mode */
+ };
+
+
+ /*!\brief Rate control mode */
+ enum vpx_rc_mode
+ {
+ VPX_VBR, /**< Variable Bit Rate (VBR) mode */
+ VPX_CBR, /**< Constant Bit Rate (CBR) mode */
+ VPX_CQ /**< Constant Quality (CQ) mode */
+ };
+
+
+ /*!\brief Keyframe placement mode.
+ *
+ * This enumeration determines whether keyframes are placed automatically by
+ * the encoder or whether this behavior is disabled. Older releases of this
+ * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled.
+ * This name is confusing for this behavior, so the new symbols to be used
+ * are VPX_KF_AUTO and VPX_KF_DISABLED.
+ */
+ enum vpx_kf_mode
+ {
+ VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */
+ VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */
+ VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */
+ };
+
+
+ /*!\brief Encoded Frame Flags
+ *
+ * This type indicates a bitfield to be passed to vpx_codec_encode(), defining
+ * per-frame boolean values. By convention, bits common to all codecs will be
+ * named VPX_EFLAG_*, and bits specific to an algorithm will be named
+ * /algo/_eflag_*. The lower order 16 bits are reserved for common use.
+ */
+ typedef long vpx_enc_frame_flags_t;
+#define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */
+
+
+ /*!\brief Encoder configuration structure
+ *
+ * This structure contains the encoder settings that have common representations
+ * across all codecs. This doesn't imply that all codecs support all features,
+ * however.
+ */
+ typedef struct vpx_codec_enc_cfg
+ {
+ /*
+ * generic settings (g)
+ */
+
+ /*!\brief Algorithm specific "usage" value
+ *
+ * Algorithms may define multiple values for usage, which may convey the
+ * intent of how the application intends to use the stream. If this value
+ * is non-zero, consult the documentation for the codec to determine its
+ * meaning.
+ */
+ unsigned int g_usage;
+
+
+ /*!\brief Maximum number of threads to use
+ *
+ * For multi-threaded implementations, use no more than this number of
+ * threads. The codec may use fewer threads than allowed. The value
+ * 0 is equivalent to the value 1.
+ */
+ unsigned int g_threads;
+
+
+ /*!\brief Bitstream profile to use
+ *
+ * Some codecs support a notion of multiple bitstream profiles. Typically
+ * this maps to a set of features that are turned on or off. Often the
+ * profile to use is determined by the features of the intended decoder.
+ * Consult the documentation for the codec to determine the valid values
+ * for this parameter, or set to zero for a sane default.
+ */
+ unsigned int g_profile; /**< profile of bitstream to use */
+
+
+
+ /*!\brief Width of the frame
+ *
+ * This value identifies the presentation resolution of the frame,
+ * in pixels. Note that the frames passed as input to the encoder must
+ * have this resolution. Frames will be presented by the decoder in this
+ * resolution, independent of any spatial resampling the encoder may do.
+ */
+ unsigned int g_w;
+
+
+ /*!\brief Height of the frame
+ *
+ * This value identifies the presentation resolution of the frame,
+ * in pixels. Note that the frames passed as input to the encoder must
+ * have this resolution. Frames will be presented by the decoder in this
+ * resolution, independent of any spatial resampling the encoder may do.
+ */
+ unsigned int g_h;
+
+
+ /*!\brief Stream timebase units
+ *
+ * Indicates the smallest interval of time, in seconds, used by the stream.
+ * For fixed frame rate material, or variable frame rate material where
+ * frames are timed at a multiple of a given clock (ex: video capture),
+ * the \ref RECOMMENDED method is to set the timebase to the reciprocal
+ * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the
+ * pts to correspond to the frame number, which can be handy. For
+ * re-encoding video from containers with absolute time timestamps, the
+ * \ref RECOMMENDED method is to set the timebase to that of the parent
+ * container or multimedia framework (ex: 1/1000 for ms, as in FLV).
+ */
+ struct vpx_rational g_timebase;
+
+
+ /*!\brief Enable error resilient modes.
+ *
+ * The error resilient bitfield indicates to the encoder which features
+ * it should enable to take measures for streaming over lossy or noisy
+ * links.
+ */
+ vpx_codec_er_flags_t g_error_resilient;
+
+
+ /*!\brief Multi-pass Encoding Mode
+ *
+ * This value should be set to the current phase for multi-pass encoding.
+ * For single pass, set to #VPX_RC_ONE_PASS.
+ */
+ enum vpx_enc_pass g_pass;
+
+
+ /*!\brief Allow lagged encoding
+ *
+ * If set, this value allows the encoder to consume a number of input
+ * frames before producing output frames. This allows the encoder to
+ * base decisions for the current frame on future frames. This does
+ * increase the latency of the encoding pipeline, so it is not appropriate
+ * in all situations (ex: realtime encoding).
+ *
+ * Note that this is a maximum value -- the encoder may produce frames
+ * sooner than the given limit. Set this value to 0 to disable this
+ * feature.
+ */
+ unsigned int g_lag_in_frames;
+
+
+ /*
+ * rate control settings (rc)
+ */
+
+ /*!\brief Temporal resampling configuration, if supported by the codec.
+ *
+ * Temporal resampling allows the codec to "drop" frames as a strategy to
+ * meet its target data rate. This can cause temporal discontinuities in
+ * the encoded video, which may appear as stuttering during playback. This
+ * trade-off is often acceptable, but for many applications is not. It can
+ * be disabled in these cases.
+ *
+ * Note that not all codecs support this feature. All vpx VPx codecs do.
+ * For other codecs, consult the documentation for that algorithm.
+ *
+ * This threshold is described as a percentage of the target data buffer.
+ * When the data buffer falls below this percentage of fullness, a
+ * dropped frame is indicated. Set the threshold to zero (0) to disable
+ * this feature.
+ */
+ unsigned int rc_dropframe_thresh;
+
+
+ /*!\brief Enable/disable spatial resampling, if supported by the codec.
+ *
+ * Spatial resampling allows the codec to compress a lower resolution
+ * version of the frame, which is then upscaled by the encoder to the
+ * correct presentation resolution. This increases visual quality at
+ * low data rates, at the expense of CPU time on the encoder/decoder.
+ */
+ unsigned int rc_resize_allowed;
+
+
+ /*!\brief Spatial resampling up watermark.
+ *
+ * This threshold is described as a percentage of the target data buffer.
+ * When the data buffer rises above this percentage of fullness, the
+ * encoder will step up to a higher resolution version of the frame.
+ */
+ unsigned int rc_resize_up_thresh;
+
+
+ /*!\brief Spatial resampling down watermark.
+ *
+ * This threshold is described as a percentage of the target data buffer.
+ * When the data buffer falls below this percentage of fullness, the
+ * encoder will step down to a lower resolution version of the frame.
+ */
+ unsigned int rc_resize_down_thresh;
+
+
+ /*!\brief Rate control algorithm to use.
+ *
+ * Indicates whether the end usage of this stream is to be streamed over
+ * a bandwidth constrained link, indicating that Constant Bit Rate (CBR)
+ * mode should be used, or whether it will be played back on a high
+ * bandwidth link, as from a local disk, where higher variations in
+ * bitrate are acceptable.
+ */
+ enum vpx_rc_mode rc_end_usage;
+
+
+ /*!\brief Two-pass stats buffer.
+ *
+ * A buffer containing all of the stats packets produced in the first
+ * pass, concatenated.
+ */
+ struct vpx_fixed_buf rc_twopass_stats_in;
+
+
+ /*!\brief Target data rate
+ *
+ * Target bandwidth to use for this stream, in kilobits per second.
+ */
+ unsigned int rc_target_bitrate;
+
+
+ /*
+ * quantizer settings
+ */
+
+
+ /*!\brief Minimum (Best Quality) Quantizer
+ *
+ * The quantizer is the most direct control over the quality of the
+ * encoded image. The range of valid values for the quantizer is codec
+ * specific. Consult the documentation for the codec to determine the
+ * values to use. To determine the range programmatically, call
+ * vpx_codec_enc_config_default() with a usage value of 0.
+ */
+ unsigned int rc_min_quantizer;
+
+
+ /*!\brief Maximum (Worst Quality) Quantizer
+ *
+ * The quantizer is the most direct control over the quality of the
+ * encoded image. The range of valid values for the quantizer is codec
+ * specific. Consult the documentation for the codec to determine the
+ * values to use. To determine the range programmatically, call
+ * vpx_codec_enc_config_default() with a usage value of 0.
+ */
+ unsigned int rc_max_quantizer;
+
+
+ /*
+ * bitrate tolerance
+ */
+
+
+ /*!\brief Rate control adaptation undershoot control
+ *
+ * This value, expressed as a percentage of the target bitrate,
+ * controls the maximum allowed adaptation speed of the codec.
+ * This factor controls the maximum amount of bits that can
+ * be subtracted from the target bitrate in order to compensate
+ * for prior overshoot.
+ *
+ * Valid values in the range 0-1000.
+ */
+ unsigned int rc_undershoot_pct;
+
+
+ /*!\brief Rate control adaptation overshoot control
+ *
+ * This value, expressed as a percentage of the target bitrate,
+ * controls the maximum allowed adaptation speed of the codec.
+ * This factor controls the maximum amount of bits that can
+ * be added to the target bitrate in order to compensate for
+ * prior undershoot.
+ *
+ * Valid values in the range 0-1000.
+ */
+ unsigned int rc_overshoot_pct;
+
+
+ /*
+ * decoder buffer model parameters
+ */
+
+
+ /*!\brief Decoder Buffer Size
+ *
+ * This value indicates the amount of data that may be buffered by the
+ * decoding application. Note that this value is expressed in units of
+ * time (milliseconds). For example, a value of 5000 indicates that the
+ * client will buffer (at least) 5000ms worth of encoded data. Use the
+ * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if
+ * necessary.
+ */
+ unsigned int rc_buf_sz;
+
+
+ /*!\brief Decoder Buffer Initial Size
+ *
+ * This value indicates the amount of data that will be buffered by the
+ * decoding application prior to beginning playback. This value is
+ * expressed in units of time (milliseconds). Use the target bitrate
+ * (#rc_target_bitrate) to convert to bits/bytes, if necessary.
+ */
+ unsigned int rc_buf_initial_sz;
+
+
+ /*!\brief Decoder Buffer Optimal Size
+ *
+ * This value indicates the amount of data that the encoder should try
+ * to maintain in the decoder's buffer. This value is expressed in units
+ * of time (milliseconds). Use the target bitrate (#rc_target_bitrate)
+ * to convert to bits/bytes, if necessary.
+ */
+ unsigned int rc_buf_optimal_sz;
+
+
+ /*
+ * 2 pass rate control parameters
+ */
+
+
+ /*!\brief Two-pass mode CBR/VBR bias
+ *
+ * Bias, expressed on a scale of 0 to 100, for determining target size
+ * for the current frame. The value 0 indicates the optimal CBR mode
+ * value should be used. The value 100 indicates the optimal VBR mode
+ * value should be used. Values in between indicate which way the
+ * encoder should "lean."
+ */
+ unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */
+
+
+ /*!\brief Two-pass mode per-GOP minimum bitrate
+ *
+ * This value, expressed as a percentage of the target bitrate, indicates
+ * the minimum bitrate to be used for a single GOP (aka "section")
+ */
+ unsigned int rc_2pass_vbr_minsection_pct;
+
+
+ /*!\brief Two-pass mode per-GOP maximum bitrate
+ *
+ * This value, expressed as a percentage of the target bitrate, indicates
+ * the maximum bitrate to be used for a single GOP (aka "section")
+ */
+ unsigned int rc_2pass_vbr_maxsection_pct;
+
+
+ /*
+ * keyframing settings (kf)
+ */
+
+ /*!\brief Keyframe placement mode
+ *
+ * This value indicates whether the encoder should place keyframes at a
+ * fixed interval, or determine the optimal placement automatically
+ * (as governed by the #kf_min_dist and #kf_max_dist parameters)
+ */
+ enum vpx_kf_mode kf_mode;
+
+
+ /*!\brief Keyframe minimum interval
+ *
+ * This value, expressed as a number of frames, prevents the encoder from
+ * placing a keyframe nearer than kf_min_dist to the previous keyframe. At
+ * least kf_min_dist frames non-keyframes will be coded before the next
+ * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval.
+ */
+ unsigned int kf_min_dist;
+
+
+ /*!\brief Keyframe maximum interval
+ *
+ * This value, expressed as a number of frames, forces the encoder to code
+ * a keyframe if one has not been coded in the last kf_max_dist frames.
+ * A value of 0 implies all frames will be keyframes. Set kf_min_dist
+ * equal to kf_max_dist for a fixed interval.
+ */
+ unsigned int kf_max_dist;
+
+ /*
+ * Temporal scalability settings (ts)
+ */
+
+ /*!\brief Number of coding layers
+ *
+ * This value specifies the number of coding layers to be used.
+ */
+ unsigned int ts_number_layers;
+
+ /*!\brief Target bitrate for each layer
+ *
+ * These values specify the target coding bitrate for each coding layer.
+ */
+ unsigned int ts_target_bitrate[VPX_TS_MAX_LAYERS];
+
+ /*!\brief Frame rate decimation factor for each layer
+ *
+ * These values specify the frame rate decimation factors to apply
+ * to each layer.
+ */
+ unsigned int ts_rate_decimator[VPX_TS_MAX_LAYERS];
+
+ /*!\brief Length of the sequence defining frame layer membership
+ *
+ * This value specifies the length of the sequence that defines the
+ * membership of frames to layers. For example, if ts_periodicity=8 then
+ * frames are assigned to coding layers with a repeated sequence of
+ * length 8.
+ */
+ unsigned int ts_periodicity;
+
+ /*!\brief Template defining the membership of frames to coding layers
+ *
+ * This array defines the membership of frames to coding layers. For a
+ * 2-layer encoding that assigns even numbered frames to one layer (0)
+ * and odd numbered frames to a second layer (1) with ts_periodicity=8,
+ * then ts_layer_id = (0,1,0,1,0,1,0,1).
+ */
+ unsigned int ts_layer_id[VPX_TS_MAX_PERIODICITY];
+ } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */
+
+
+ /*!\brief Initialize an encoder instance
+ *
+ * Initializes a encoder context using the given interface. Applications
+ * should call the vpx_codec_enc_init convenience macro instead of this
+ * function directly, to ensure that the ABI version number parameter
+ * is properly initialized.
+ *
+ * If the library was configured with --disable-multithread, this call
+ * is not thread safe and should be guarded with a lock if being used
+ * in a multithreaded context.
+ *
+ * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
+ * parameter), the storage pointed to by the cfg parameter must be
+ * kept readable and stable until all memory maps have been set.
+ *
+ * \param[in] ctx Pointer to this instance's context.
+ * \param[in] iface Pointer to the algorithm interface to use.
+ * \param[in] cfg Configuration to use, if known. May be NULL.
+ * \param[in] flags Bitfield of VPX_CODEC_USE_* flags
+ * \param[in] ver ABI version number. Must be set to
+ * VPX_ENCODER_ABI_VERSION
+ * \retval #VPX_CODEC_OK
+ * The decoder algorithm initialized.
+ * \retval #VPX_CODEC_MEM_ERROR
+ * Memory allocation failed.
+ */
+ vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx,
+ vpx_codec_iface_t *iface,
+ vpx_codec_enc_cfg_t *cfg,
+ vpx_codec_flags_t flags,
+ int ver);
+
+
+ /*!\brief Convenience macro for vpx_codec_enc_init_ver()
+ *
+ * Ensures the ABI version parameter is properly set.
+ */
+#define vpx_codec_enc_init(ctx, iface, cfg, flags) \
+ vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION)
+
+
+ /*!\brief Initialize multi-encoder instance
+ *
+ * Initializes multi-encoder context using the given interface.
+ * Applications should call the vpx_codec_enc_init_multi convenience macro
+ * instead of this function directly, to ensure that the ABI version number
+ * parameter is properly initialized.
+ *
+ * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
+ * parameter), the storage pointed to by the cfg parameter must be
+ * kept readable and stable until all memory maps have been set.
+ *
+ * \param[in] ctx Pointer to this instance's context.
+ * \param[in] iface Pointer to the algorithm interface to use.
+ * \param[in] cfg Configuration to use, if known. May be NULL.
+ * \param[in] num_enc Total number of encoders.
+ * \param[in] flags Bitfield of VPX_CODEC_USE_* flags
+ * \param[in] dsf Pointer to down-sampling factors.
+ * \param[in] ver ABI version number. Must be set to
+ * VPX_ENCODER_ABI_VERSION
+ * \retval #VPX_CODEC_OK
+ * The decoder algorithm initialized.
+ * \retval #VPX_CODEC_MEM_ERROR
+ * Memory allocation failed.
+ */
+ vpx_codec_err_t vpx_codec_enc_init_multi_ver(vpx_codec_ctx_t *ctx,
+ vpx_codec_iface_t *iface,
+ vpx_codec_enc_cfg_t *cfg,
+ int num_enc,
+ vpx_codec_flags_t flags,
+ vpx_rational_t *dsf,
+ int ver);
+
+
+ /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver()
+ *
+ * Ensures the ABI version parameter is properly set.
+ */
+#define vpx_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \
+ vpx_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \
+ VPX_ENCODER_ABI_VERSION)
+
+
+ /*!\brief Get a default configuration
+ *
+ * Initializes a encoder configuration structure with default values. Supports
+ * the notion of "usages" so that an algorithm may offer different default
+ * settings depending on the user's intended goal. This function \ref SHOULD
+ * be called by all applications to initialize the configuration structure
+ * before specializing the configuration with application specific values.
+ *
+ * \param[in] iface Pointer to the algorithm interface to use.
+ * \param[out] cfg Configuration buffer to populate
+ * \param[in] usage End usage. Set to 0 or use codec specific values.
+ *
+ * \retval #VPX_CODEC_OK
+ * The configuration was populated.
+ * \retval #VPX_CODEC_INCAPABLE
+ * Interface is not an encoder interface.
+ * \retval #VPX_CODEC_INVALID_PARAM
+ * A parameter was NULL, or the usage value was not recognized.
+ */
+ vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface,
+ vpx_codec_enc_cfg_t *cfg,
+ unsigned int usage);
+
+
+ /*!\brief Set or change configuration
+ *
+ * Reconfigures an encoder instance according to the given configuration.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] cfg Configuration buffer to use
+ *
+ * \retval #VPX_CODEC_OK
+ * The configuration was populated.
+ * \retval #VPX_CODEC_INCAPABLE
+ * Interface is not an encoder interface.
+ * \retval #VPX_CODEC_INVALID_PARAM
+ * A parameter was NULL, or the usage value was not recognized.
+ */
+ vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx,
+ const vpx_codec_enc_cfg_t *cfg);
+
+
+ /*!\brief Get global stream headers
+ *
+ * Retrieves a stream level global header packet, if supported by the codec.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ *
+ * \retval NULL
+ * Encoder does not support global header
+ * \retval Non-NULL
+ * Pointer to buffer containing global header packet
+ */
+ vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx);
+
+
+#define VPX_DL_REALTIME (1) /**< deadline parameter analogous to
+ * VPx REALTIME mode. */
+#define VPX_DL_GOOD_QUALITY (1000000) /**< deadline parameter analogous to
+ * VPx GOOD QUALITY mode. */
+#define VPX_DL_BEST_QUALITY (0) /**< deadline parameter analogous to
+ * VPx BEST QUALITY mode. */
+ /*!\brief Encode a frame
+ *
+ * Encodes a video frame at the given "presentation time." The presentation
+ * time stamp (PTS) \ref MUST be strictly increasing.
+ *
+ * The encoder supports the notion of a soft real-time deadline. Given a
+ * non-zero value to the deadline parameter, the encoder will make a "best
+ * effort" guarantee to return before the given time slice expires. It is
+ * implicit that limiting the available time to encode will degrade the
+ * output quality. The encoder can be given an unlimited time to produce the
+ * best possible frame by specifying a deadline of '0'. This deadline
+ * supercedes the VPx notion of "best quality, good quality, realtime".
+ * Applications that wish to map these former settings to the new deadline
+ * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY,
+ * and #VPX_DL_BEST_QUALITY.
+ *
+ * When the last frame has been passed to the encoder, this function should
+ * continue to be called, with the img parameter set to NULL. This will
+ * signal the end-of-stream condition to the encoder and allow it to encode
+ * any held buffers. Encoding is complete when vpx_codec_encode() is called
+ * and vpx_codec_get_cx_data() returns no data.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] img Image data to encode, NULL to flush.
+ * \param[in] pts Presentation time stamp, in timebase units.
+ * \param[in] duration Duration to show frame, in timebase units.
+ * \param[in] flags Flags to use for encoding this frame.
+ * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite)
+ *
+ * \retval #VPX_CODEC_OK
+ * The configuration was populated.
+ * \retval #VPX_CODEC_INCAPABLE
+ * Interface is not an encoder interface.
+ * \retval #VPX_CODEC_INVALID_PARAM
+ * A parameter was NULL, the image format is unsupported, etc.
+ */
+ vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx,
+ const vpx_image_t *img,
+ vpx_codec_pts_t pts,
+ unsigned long duration,
+ vpx_enc_frame_flags_t flags,
+ unsigned long deadline);
+
+ /*!\brief Set compressed data output buffer
+ *
+ * Sets the buffer that the codec should output the compressed data
+ * into. This call effectively sets the buffer pointer returned in the
+ * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be
+ * appended into this buffer. The buffer is preserved across frames,
+ * so applications must periodically call this function after flushing
+ * the accumulated compressed data to disk or to the network to reset
+ * the pointer to the buffer's head.
+ *
+ * `pad_before` bytes will be skipped before writing the compressed
+ * data, and `pad_after` bytes will be appended to the packet. The size
+ * of the packet will be the sum of the size of the actual compressed
+ * data, pad_before, and pad_after. The padding bytes will be preserved
+ * (not overwritten).
+ *
+ * Note that calling this function does not guarantee that the returned
+ * compressed data will be placed into the specified buffer. In the
+ * event that the encoded data will not fit into the buffer provided,
+ * the returned packet \ref MAY point to an internal buffer, as it would
+ * if this call were never used. In this event, the output packet will
+ * NOT have any padding, and the application must free space and copy it
+ * to the proper place. This is of particular note in configurations
+ * that may output multiple packets for a single encoded frame (e.g., lagged
+ * encoding) or if the application does not reset the buffer periodically.
+ *
+ * Applications may restore the default behavior of the codec providing
+ * the compressed data buffer by calling this function with a NULL
+ * buffer.
+ *
+ * Applications \ref MUSTNOT call this function during iteration of
+ * vpx_codec_get_cx_data().
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in] buf Buffer to store compressed data into
+ * \param[in] pad_before Bytes to skip before writing compressed data
+ * \param[in] pad_after Bytes to skip after writing compressed data
+ *
+ * \retval #VPX_CODEC_OK
+ * The buffer was set successfully.
+ * \retval #VPX_CODEC_INVALID_PARAM
+ * A parameter was NULL, the image format is unsupported, etc.
+ */
+ vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx,
+ const vpx_fixed_buf_t *buf,
+ unsigned int pad_before,
+ unsigned int pad_after);
+
+
+ /*!\brief Encoded data iterator
+ *
+ * Iterates over a list of data packets to be passed from the encoder to the
+ * application. The different kinds of packets available are enumerated in
+ * #vpx_codec_cx_pkt_kind.
+ *
+ * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's
+ * muxer. Multiple compressed frames may be in the list.
+ * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer.
+ *
+ * The application \ref MUST silently ignore any packet kinds that it does
+ * not recognize or support.
+ *
+ * The data buffers returned from this function are only guaranteed to be
+ * valid until the application makes another call to any vpx_codec_* function.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ * \param[in,out] iter Iterator storage, initialized to NULL
+ *
+ * \return Returns a pointer to an output data packet (compressed frame data,
+ * two-pass statistics, etc.) or NULL to signal end-of-list.
+ *
+ */
+ const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx,
+ vpx_codec_iter_t *iter);
+
+
+ /*!\brief Get Preview Frame
+ *
+ * Returns an image that can be used as a preview. Shows the image as it would
+ * exist at the decompressor. The application \ref MUST NOT write into this
+ * image buffer.
+ *
+ * \param[in] ctx Pointer to this instance's context
+ *
+ * \return Returns a pointer to a preview image, or NULL if no image is
+ * available.
+ *
+ */
+ const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx);
+
+
+ /*!@} - end defgroup encoder*/
+
+#endif
+#ifdef __cplusplus
+}
+#endif
diff --git a/libvpx/vpx/vpx_image.h b/libvpx/vpx/vpx_image.h
new file mode 100644
index 0000000..3e42447
--- /dev/null
+++ b/libvpx/vpx/vpx_image.h
@@ -0,0 +1,243 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*!\file
+ * \brief Describes the vpx image descriptor and associated operations
+ *
+ */
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef VPX_IMAGE_H
+#define VPX_IMAGE_H
+
+ /*!\brief Current ABI version number
+ *
+ * \internal
+ * If this file is altered in any way that changes the ABI, this value
+ * must be bumped. Examples include, but are not limited to, changing
+ * types, removing or reassigning enums, adding/removing/rearranging
+ * fields to structures
+ */
+#define VPX_IMAGE_ABI_VERSION (1) /**<\hideinitializer*/
+
+
+#define VPX_IMG_FMT_PLANAR 0x100 /**< Image is a planar format */
+#define VPX_IMG_FMT_UV_FLIP 0x200 /**< V plane precedes U plane in memory */
+#define VPX_IMG_FMT_HAS_ALPHA 0x400 /**< Image has an alpha channel component */
+
+
+ /*!\brief List of supported image formats */
+ typedef enum vpx_img_fmt {
+ VPX_IMG_FMT_NONE,
+ VPX_IMG_FMT_RGB24, /**< 24 bit per pixel packed RGB */
+ VPX_IMG_FMT_RGB32, /**< 32 bit per pixel packed 0RGB */
+ VPX_IMG_FMT_RGB565, /**< 16 bit per pixel, 565 */
+ VPX_IMG_FMT_RGB555, /**< 16 bit per pixel, 555 */
+ VPX_IMG_FMT_UYVY, /**< UYVY packed YUV */
+ VPX_IMG_FMT_YUY2, /**< YUYV packed YUV */
+ VPX_IMG_FMT_YVYU, /**< YVYU packed YUV */
+ VPX_IMG_FMT_BGR24, /**< 24 bit per pixel packed BGR */
+ VPX_IMG_FMT_RGB32_LE, /**< 32 bit packed BGR0 */
+ VPX_IMG_FMT_ARGB, /**< 32 bit packed ARGB, alpha=255 */
+ VPX_IMG_FMT_ARGB_LE, /**< 32 bit packed BGRA, alpha=255 */
+ VPX_IMG_FMT_RGB565_LE, /**< 16 bit per pixel, gggbbbbb rrrrrggg */
+ VPX_IMG_FMT_RGB555_LE, /**< 16 bit per pixel, gggbbbbb 0rrrrrgg */
+ VPX_IMG_FMT_YV12 = VPX_IMG_FMT_PLANAR | VPX_IMG_FMT_UV_FLIP | 1, /**< planar YVU */
+ VPX_IMG_FMT_I420 = VPX_IMG_FMT_PLANAR | 2,
+ VPX_IMG_FMT_VPXYV12 = VPX_IMG_FMT_PLANAR | VPX_IMG_FMT_UV_FLIP | 3, /** < planar 4:2:0 format with vpx color space */
+ VPX_IMG_FMT_VPXI420 = VPX_IMG_FMT_PLANAR | 4 /** < planar 4:2:0 format with vpx color space */
+ }
+ vpx_img_fmt_t; /**< alias for enum vpx_img_fmt */
+
+#if !defined(VPX_CODEC_DISABLE_COMPAT) || !VPX_CODEC_DISABLE_COMPAT
+#define IMG_FMT_PLANAR VPX_IMG_FMT_PLANAR /**< \deprecated Use #VPX_IMG_FMT_PLANAR */
+#define IMG_FMT_UV_FLIP VPX_IMG_FMT_UV_FLIP /**< \deprecated Use #VPX_IMG_FMT_UV_FLIP */
+#define IMG_FMT_HAS_ALPHA VPX_IMG_FMT_HAS_ALPHA /**< \deprecated Use #VPX_IMG_FMT_HAS_ALPHA */
+
+ /*!\brief Deprecated list of supported image formats
+ * \deprecated New code should use #vpx_img_fmt
+ */
+#define img_fmt vpx_img_fmt
+ /*!\brief alias for enum img_fmt.
+ * \deprecated New code should use #vpx_img_fmt_t
+ */
+#define img_fmt_t vpx_img_fmt_t
+
+#define IMG_FMT_NONE VPX_IMG_FMT_NONE /**< \deprecated Use #VPX_IMG_FMT_NONE */
+#define IMG_FMT_RGB24 VPX_IMG_FMT_RGB24 /**< \deprecated Use #VPX_IMG_FMT_RGB24 */
+#define IMG_FMT_RGB32 VPX_IMG_FMT_RGB32 /**< \deprecated Use #VPX_IMG_FMT_RGB32 */
+#define IMG_FMT_RGB565 VPX_IMG_FMT_RGB565 /**< \deprecated Use #VPX_IMG_FMT_RGB565 */
+#define IMG_FMT_RGB555 VPX_IMG_FMT_RGB555 /**< \deprecated Use #VPX_IMG_FMT_RGB555 */
+#define IMG_FMT_UYVY VPX_IMG_FMT_UYVY /**< \deprecated Use #VPX_IMG_FMT_UYVY */
+#define IMG_FMT_YUY2 VPX_IMG_FMT_YUY2 /**< \deprecated Use #VPX_IMG_FMT_YUY2 */
+#define IMG_FMT_YVYU VPX_IMG_FMT_YVYU /**< \deprecated Use #VPX_IMG_FMT_YVYU */
+#define IMG_FMT_BGR24 VPX_IMG_FMT_BGR24 /**< \deprecated Use #VPX_IMG_FMT_BGR24 */
+#define IMG_FMT_RGB32_LE VPX_IMG_FMT_RGB32_LE /**< \deprecated Use #VPX_IMG_FMT_RGB32_LE */
+#define IMG_FMT_ARGB VPX_IMG_FMT_ARGB /**< \deprecated Use #VPX_IMG_FMT_ARGB */
+#define IMG_FMT_ARGB_LE VPX_IMG_FMT_ARGB_LE /**< \deprecated Use #VPX_IMG_FMT_ARGB_LE */
+#define IMG_FMT_RGB565_LE VPX_IMG_FMT_RGB565_LE /**< \deprecated Use #VPX_IMG_FMT_RGB565_LE */
+#define IMG_FMT_RGB555_LE VPX_IMG_FMT_RGB555_LE /**< \deprecated Use #VPX_IMG_FMT_RGB555_LE */
+#define IMG_FMT_YV12 VPX_IMG_FMT_YV12 /**< \deprecated Use #VPX_IMG_FMT_YV12 */
+#define IMG_FMT_I420 VPX_IMG_FMT_I420 /**< \deprecated Use #VPX_IMG_FMT_I420 */
+#define IMG_FMT_VPXYV12 VPX_IMG_FMT_VPXYV12 /**< \deprecated Use #VPX_IMG_FMT_VPXYV12 */
+#define IMG_FMT_VPXI420 VPX_IMG_FMT_VPXI420 /**< \deprecated Use #VPX_IMG_FMT_VPXI420 */
+#endif /* VPX_CODEC_DISABLE_COMPAT */
+
+ /**\brief Image Descriptor */
+ typedef struct vpx_image
+ {
+ vpx_img_fmt_t fmt; /**< Image Format */
+
+ /* Image storage dimensions */
+ unsigned int w; /**< Stored image width */
+ unsigned int h; /**< Stored image height */
+
+ /* Image display dimensions */
+ unsigned int d_w; /**< Displayed image width */
+ unsigned int d_h; /**< Displayed image height */
+
+ /* Chroma subsampling info */
+ unsigned int x_chroma_shift; /**< subsampling order, X */
+ unsigned int y_chroma_shift; /**< subsampling order, Y */
+
+ /* Image data pointers. */
+#define VPX_PLANE_PACKED 0 /**< To be used for all packed formats */
+#define VPX_PLANE_Y 0 /**< Y (Luminance) plane */
+#define VPX_PLANE_U 1 /**< U (Chroma) plane */
+#define VPX_PLANE_V 2 /**< V (Chroma) plane */
+#define VPX_PLANE_ALPHA 3 /**< A (Transparency) plane */
+#if !defined(VPX_CODEC_DISABLE_COMPAT) || !VPX_CODEC_DISABLE_COMPAT
+#define PLANE_PACKED VPX_PLANE_PACKED
+#define PLANE_Y VPX_PLANE_Y
+#define PLANE_U VPX_PLANE_U
+#define PLANE_V VPX_PLANE_V
+#define PLANE_ALPHA VPX_PLANE_ALPHA
+#endif
+ unsigned char *planes[4]; /**< pointer to the top left pixel for each plane */
+ int stride[4]; /**< stride between rows for each plane */
+
+ int bps; /**< bits per sample (for packed formats) */
+
+ /* The following member may be set by the application to associate data
+ * with this image.
+ */
+ void *user_priv; /**< may be set by the application to associate data
+ * with this image. */
+
+ /* The following members should be treated as private. */
+ unsigned char *img_data; /**< private */
+ int img_data_owner; /**< private */
+ int self_allocd; /**< private */
+ } vpx_image_t; /**< alias for struct vpx_image */
+
+ /**\brief Representation of a rectangle on a surface */
+ typedef struct vpx_image_rect
+ {
+ unsigned int x; /**< leftmost column */
+ unsigned int y; /**< topmost row */
+ unsigned int w; /**< width */
+ unsigned int h; /**< height */
+ } vpx_image_rect_t; /**< alias for struct vpx_image_rect */
+
+ /*!\brief Open a descriptor, allocating storage for the underlying image
+ *
+ * Returns a descriptor for storing an image of the given format. The
+ * storage for the descriptor is allocated on the heap.
+ *
+ * \param[in] img Pointer to storage for descriptor. If this parameter
+ * is NULL, the storage for the descriptor will be
+ * allocated on the heap.
+ * \param[in] fmt Format for the image
+ * \param[in] d_w Width of the image
+ * \param[in] d_h Height of the image
+ * \param[in] align Alignment, in bytes, of the image buffer and
+ * each row in the image(stride).
+ *
+ * \return Returns a pointer to the initialized image descriptor. If the img
+ * parameter is non-null, the value of the img parameter will be
+ * returned.
+ */
+ vpx_image_t *vpx_img_alloc(vpx_image_t *img,
+ vpx_img_fmt_t fmt,
+ unsigned int d_w,
+ unsigned int d_h,
+ unsigned int align);
+
+ /*!\brief Open a descriptor, using existing storage for the underlying image
+ *
+ * Returns a descriptor for storing an image of the given format. The
+ * storage for descriptor has been allocated elsewhere, and a descriptor is
+ * desired to "wrap" that storage.
+ *
+ * \param[in] img Pointer to storage for descriptor. If this parameter
+ * is NULL, the storage for the descriptor will be
+ * allocated on the heap.
+ * \param[in] fmt Format for the image
+ * \param[in] d_w Width of the image
+ * \param[in] d_h Height of the image
+ * \param[in] align Alignment, in bytes, of each row in the image.
+ * \param[in] img_data Storage to use for the image
+ *
+ * \return Returns a pointer to the initialized image descriptor. If the img
+ * parameter is non-null, the value of the img parameter will be
+ * returned.
+ */
+ vpx_image_t *vpx_img_wrap(vpx_image_t *img,
+ vpx_img_fmt_t fmt,
+ unsigned int d_w,
+ unsigned int d_h,
+ unsigned int align,
+ unsigned char *img_data);
+
+
+ /*!\brief Set the rectangle identifying the displayed portion of the image
+ *
+ * Updates the displayed rectangle (aka viewport) on the image surface to
+ * match the specified coordinates and size.
+ *
+ * \param[in] img Image descriptor
+ * \param[in] x leftmost column
+ * \param[in] y topmost row
+ * \param[in] w width
+ * \param[in] h height
+ *
+ * \return 0 if the requested rectangle is valid, nonzero otherwise.
+ */
+ int vpx_img_set_rect(vpx_image_t *img,
+ unsigned int x,
+ unsigned int y,
+ unsigned int w,
+ unsigned int h);
+
+
+ /*!\brief Flip the image vertically (top for bottom)
+ *
+ * Adjusts the image descriptor's pointers and strides to make the image
+ * be referenced upside-down.
+ *
+ * \param[in] img Image descriptor
+ */
+ void vpx_img_flip(vpx_image_t *img);
+
+ /*!\brief Close an image descriptor
+ *
+ * Frees all allocated storage associated with an image descriptor.
+ *
+ * \param[in] img Image descriptor
+ */
+ void vpx_img_free(vpx_image_t *img);
+
+#endif
+#ifdef __cplusplus
+}
+#endif
diff --git a/libvpx/vpx/vpx_integer.h b/libvpx/vpx/vpx_integer.h
new file mode 100644
index 0000000..218bca7
--- /dev/null
+++ b/libvpx/vpx/vpx_integer.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef VPX_INTEGER_H
+#define VPX_INTEGER_H
+
+/* get ptrdiff_t, size_t, wchar_t, NULL */
+#include <stddef.h>
+
+#if (defined(_MSC_VER) && (_MSC_VER < 1600)) || defined(VPX_EMULATE_INTTYPES)
+typedef signed char int8_t;
+typedef signed short int16_t;
+typedef signed int int32_t;
+
+typedef unsigned char uint8_t;
+typedef unsigned short uint16_t;
+typedef unsigned int uint32_t;
+
+#if (defined(_MSC_VER) && (_MSC_VER < 1600))
+typedef signed __int64 int64_t;
+typedef unsigned __int64 uint64_t;
+#endif
+
+#ifndef _UINTPTR_T_DEFINED
+typedef size_t uintptr_t;
+#endif
+
+#else
+
+/* Most platforms have the C99 standard integer types. */
+
+#if defined(__cplusplus) && !defined(__STDC_FORMAT_MACROS)
+#define __STDC_FORMAT_MACROS
+#endif
+#include <stdint.h>
+
+#endif
+
+/* VS2010 defines stdint.h, but not inttypes.h */
+#if defined(_MSC_VER)
+#define PRId64 "I64d"
+#else
+#include <inttypes.h>
+#endif
+
+#endif
diff --git a/libvpx/vpx_mem/include/vpx_mem_intrnl.h b/libvpx/vpx_mem/include/vpx_mem_intrnl.h
new file mode 100644
index 0000000..63c6b77
--- /dev/null
+++ b/libvpx/vpx_mem/include/vpx_mem_intrnl.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __VPX_MEM_INTRNL_H__
+#define __VPX_MEM_INTRNL_H__
+#include "vpx_config.h"
+
+#ifndef CONFIG_MEM_MANAGER
+# if defined(VXWORKS)
+# define CONFIG_MEM_MANAGER 1 /*include heap manager functionality,*/
+/*default: enabled on vxworks*/
+# else
+# define CONFIG_MEM_MANAGER 0 /*include heap manager functionality*/
+# endif
+#endif /*CONFIG_MEM_MANAGER*/
+
+#ifndef CONFIG_MEM_TRACKER
+# define CONFIG_MEM_TRACKER 1 /*include xvpx_* calls in the lib*/
+#endif
+
+#ifndef CONFIG_MEM_CHECKS
+# define CONFIG_MEM_CHECKS 0 /*include some basic safety checks in
+vpx_memcpy, _memset, and _memmove*/
+#endif
+
+#ifndef USE_GLOBAL_FUNCTION_POINTERS
+# define USE_GLOBAL_FUNCTION_POINTERS 0 /*use function pointers instead of compiled functions.*/
+#endif
+
+#if CONFIG_MEM_TRACKER
+# include "vpx_mem_tracker.h"
+# if VPX_MEM_TRACKER_VERSION_CHIEF != 2 || VPX_MEM_TRACKER_VERSION_MAJOR != 5
+# error "vpx_mem requires memory tracker version 2.5 to track memory usage"
+# endif
+#endif
+
+#define ADDRESS_STORAGE_SIZE sizeof(size_t)
+
+#ifndef DEFAULT_ALIGNMENT
+# if defined(VXWORKS)
+# define DEFAULT_ALIGNMENT 32 /*default addr alignment to use in
+ calls to vpx_* functions other
+ than vpx_memalign*/
+# else
+# define DEFAULT_ALIGNMENT 1
+# endif
+#endif
+
+#if DEFAULT_ALIGNMENT < 1
+# error "DEFAULT_ALIGNMENT must be >= 1!"
+#endif
+
+#if CONFIG_MEM_TRACKER
+# define TRY_BOUNDS_CHECK 1 /*when set to 1 pads each allocation,
+ integrity can be checked using
+ vpx_memory_tracker_check_integrity
+ or on free by defining*/
+/*TRY_BOUNDS_CHECK_ON_FREE*/
+#else
+# define TRY_BOUNDS_CHECK 0
+#endif /*CONFIG_MEM_TRACKER*/
+
+#if TRY_BOUNDS_CHECK
+# define TRY_BOUNDS_CHECK_ON_FREE 0 /*checks mem integrity on every
+ free, very expensive*/
+# define BOUNDS_CHECK_VALUE 0xdeadbeef /*value stored before/after ea.
+ mem addr for bounds checking*/
+# define BOUNDS_CHECK_PAD_SIZE 32 /*size of the padding before and
+ after ea allocation to be filled
+ with BOUNDS_CHECK_VALUE.
+ this should be a multiple of 4*/
+#else
+# define BOUNDS_CHECK_VALUE 0
+# define BOUNDS_CHECK_PAD_SIZE 0
+#endif /*TRY_BOUNDS_CHECK*/
+
+#ifndef REMOVE_PRINTFS
+# define REMOVE_PRINTFS 0
+#endif
+
+/* Should probably use a vpx_mem logger function. */
+#if REMOVE_PRINTFS
+# define _P(x)
+#else
+# define _P(x) x
+#endif
+
+/*returns an addr aligned to the byte boundary specified by align*/
+#define align_addr(addr,align) (void*)(((size_t)(addr) + ((align) - 1)) & (size_t)-(align))
+
+#endif /*__VPX_MEM_INTRNL_H__*/
diff --git a/vpx_mem/include/vpx_mem_tracker.h b/libvpx/vpx_mem/include/vpx_mem_tracker.h
similarity index 100%
rename from vpx_mem/include/vpx_mem_tracker.h
rename to libvpx/vpx_mem/include/vpx_mem_tracker.h
diff --git a/vpx_mem/memory_manager/hmm_alloc.c b/libvpx/vpx_mem/memory_manager/hmm_alloc.c
similarity index 100%
rename from vpx_mem/memory_manager/hmm_alloc.c
rename to libvpx/vpx_mem/memory_manager/hmm_alloc.c
diff --git a/vpx_mem/memory_manager/hmm_base.c b/libvpx/vpx_mem/memory_manager/hmm_base.c
similarity index 100%
rename from vpx_mem/memory_manager/hmm_base.c
rename to libvpx/vpx_mem/memory_manager/hmm_base.c
diff --git a/vpx_mem/memory_manager/hmm_dflt_abort.c b/libvpx/vpx_mem/memory_manager/hmm_dflt_abort.c
similarity index 100%
rename from vpx_mem/memory_manager/hmm_dflt_abort.c
rename to libvpx/vpx_mem/memory_manager/hmm_dflt_abort.c
diff --git a/vpx_mem/memory_manager/hmm_grow.c b/libvpx/vpx_mem/memory_manager/hmm_grow.c
similarity index 100%
rename from vpx_mem/memory_manager/hmm_grow.c
rename to libvpx/vpx_mem/memory_manager/hmm_grow.c
diff --git a/vpx_mem/memory_manager/hmm_largest.c b/libvpx/vpx_mem/memory_manager/hmm_largest.c
similarity index 100%
rename from vpx_mem/memory_manager/hmm_largest.c
rename to libvpx/vpx_mem/memory_manager/hmm_largest.c
diff --git a/vpx_mem/memory_manager/hmm_resize.c b/libvpx/vpx_mem/memory_manager/hmm_resize.c
similarity index 100%
rename from vpx_mem/memory_manager/hmm_resize.c
rename to libvpx/vpx_mem/memory_manager/hmm_resize.c
diff --git a/vpx_mem/memory_manager/hmm_shrink.c b/libvpx/vpx_mem/memory_manager/hmm_shrink.c
similarity index 100%
rename from vpx_mem/memory_manager/hmm_shrink.c
rename to libvpx/vpx_mem/memory_manager/hmm_shrink.c
diff --git a/vpx_mem/memory_manager/hmm_true.c b/libvpx/vpx_mem/memory_manager/hmm_true.c
similarity index 100%
rename from vpx_mem/memory_manager/hmm_true.c
rename to libvpx/vpx_mem/memory_manager/hmm_true.c
diff --git a/vpx_mem/memory_manager/include/cavl_if.h b/libvpx/vpx_mem/memory_manager/include/cavl_if.h
similarity index 100%
rename from vpx_mem/memory_manager/include/cavl_if.h
rename to libvpx/vpx_mem/memory_manager/include/cavl_if.h
diff --git a/vpx_mem/memory_manager/include/cavl_impl.h b/libvpx/vpx_mem/memory_manager/include/cavl_impl.h
similarity index 100%
rename from vpx_mem/memory_manager/include/cavl_impl.h
rename to libvpx/vpx_mem/memory_manager/include/cavl_impl.h
diff --git a/vpx_mem/memory_manager/include/heapmm.h b/libvpx/vpx_mem/memory_manager/include/heapmm.h
similarity index 100%
rename from vpx_mem/memory_manager/include/heapmm.h
rename to libvpx/vpx_mem/memory_manager/include/heapmm.h
diff --git a/vpx_mem/memory_manager/include/hmm_cnfg.h b/libvpx/vpx_mem/memory_manager/include/hmm_cnfg.h
similarity index 100%
rename from vpx_mem/memory_manager/include/hmm_cnfg.h
rename to libvpx/vpx_mem/memory_manager/include/hmm_cnfg.h
diff --git a/vpx_mem/memory_manager/include/hmm_intrnl.h b/libvpx/vpx_mem/memory_manager/include/hmm_intrnl.h
similarity index 100%
rename from vpx_mem/memory_manager/include/hmm_intrnl.h
rename to libvpx/vpx_mem/memory_manager/include/hmm_intrnl.h
diff --git a/vpx_mem/vpx_mem.c b/libvpx/vpx_mem/vpx_mem.c
similarity index 100%
rename from vpx_mem/vpx_mem.c
rename to libvpx/vpx_mem/vpx_mem.c
diff --git a/vpx_mem/vpx_mem.h b/libvpx/vpx_mem/vpx_mem.h
similarity index 100%
rename from vpx_mem/vpx_mem.h
rename to libvpx/vpx_mem/vpx_mem.h
diff --git a/vpx_mem/vpx_mem.mk b/libvpx/vpx_mem/vpx_mem.mk
similarity index 100%
rename from vpx_mem/vpx_mem.mk
rename to libvpx/vpx_mem/vpx_mem.mk
diff --git a/libvpx/vpx_mem/vpx_mem_tracker.c b/libvpx/vpx_mem/vpx_mem_tracker.c
new file mode 100644
index 0000000..b37076e
--- /dev/null
+++ b/libvpx/vpx_mem/vpx_mem_tracker.c
@@ -0,0 +1,798 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/*
+ vpx_mem_tracker.c
+
+ jwz 2003-09-30:
+ Stores a list of addreses, their size, and file and line they came from.
+ All exposed lib functions are prefaced by vpx_ and allow the global list
+ to be thread safe.
+ Current supported platforms are:
+ Linux, Win32, win_ce and vx_works
+ Further support can be added by defining the platform specific mutex
+ in the memory_tracker struct as well as calls to create/destroy/lock/unlock
+ the mutex in vpx_memory_tracker_init/Destroy and memory_tracker_lock_mutex/unlock_mutex
+*/
+#include "vpx_config.h"
+
+#if defined(__uClinux__)
+# include <lddk.h>
+#endif
+
+#if HAVE_PTHREAD_H
+# include <pthread.h>
+#elif defined(WIN32) || defined(_WIN32_WCE)
+# define WIN32_LEAN_AND_MEAN
+# include <windows.h>
+# include <winbase.h>
+#elif defined(VXWORKS)
+# include <sem_lib.h>
+#endif
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h> //VXWORKS doesn't have a malloc/memory.h file,
+//this should pull in malloc,free,etc.
+#include <stdarg.h>
+
+#include "include/vpx_mem_tracker.h"
+
+#undef vpx_malloc //undefine any vpx_mem macros that may affect calls to
+#undef vpx_free //memory functions in this file
+#undef vpx_memcpy
+#undef vpx_memset
+
+
+#ifndef USE_GLOBAL_FUNCTION_POINTERS
+# define USE_GLOBAL_FUNCTION_POINTERS 0 //use function pointers instead of compiled functions.
+#endif
+
+#if USE_GLOBAL_FUNCTION_POINTERS
+static mem_track_malloc_func g_malloc = malloc;
+static mem_track_calloc_func g_calloc = calloc;
+static mem_track_realloc_func g_realloc = realloc;
+static mem_track_free_func g_free = free;
+static mem_track_memcpy_func g_memcpy = memcpy;
+static mem_track_memset_func g_memset = memset;
+static mem_track_memmove_func g_memmove = memmove;
+# define MEM_TRACK_MALLOC g_malloc
+# define MEM_TRACK_FREE g_free
+# define MEM_TRACK_MEMCPY g_memcpy
+# define MEM_TRACK_MEMSET g_memset
+#else
+# define MEM_TRACK_MALLOC vpx_malloc
+# define MEM_TRACK_FREE vpx_free
+# define MEM_TRACK_MEMCPY vpx_memcpy
+# define MEM_TRACK_MEMSET vpx_memset
+#endif // USE_GLOBAL_FUNCTION_POINTERS
+
+/* prototypes for internal library functions */
+static void memtrack_log(const char *fmt, ...);
+static void memory_tracker_dump();
+static void memory_tracker_check_integrity(char *file, unsigned int line);
+static void memory_tracker_add(size_t addr, unsigned int size,
+ char *file, unsigned int line,
+ int padded);
+static int memory_tracker_remove(size_t addr);
+static struct mem_block *memory_tracker_find(size_t addr);
+
+#if defined(NO_MUTEX)
+# define memory_tracker_lock_mutex() (!g_b_mem_tracker_inited)
+# define memory_tracker_unlock_mutex()
+#else
+static int memory_tracker_lock_mutex();
+static int memory_tracker_unlock_mutex();
+#endif
+
+#ifndef VPX_NO_GLOBALS
+struct memory_tracker
+{
+ struct mem_block *head,
+ * tail;
+ int len,
+ totalsize;
+ unsigned int current_allocated,
+ max_allocated;
+
+#if HAVE_PTHREAD_H
+ pthread_mutex_t mutex;
+#elif defined(WIN32) || defined(_WIN32_WCE)
+ HANDLE mutex;
+#elif defined(VXWORKS)
+ SEM_ID mutex;
+#elif defined(NO_MUTEX)
+#else
+#error "No mutex type defined for this platform!"
+#endif
+
+ int padding_size,
+ pad_value;
+};
+
+static struct memory_tracker memtrack; //our global memory allocation list
+static int g_b_mem_tracker_inited = 0; //indicates whether the global list has
+//been initialized (1:yes/0:no)
+static struct
+{
+ FILE *file;
+ int type;
+ void (*func)(void *userdata, const char *fmt, va_list args);
+ void *userdata;
+} g_logging = {NULL, 0, NULL, NULL};
+#else
+# include "vpx_global_handling.h"
+#define g_b_mem_tracker_inited vpxglobalm(vpxmem,g_b_mem_tracker_inited)
+#define g_logging vpxglobalm(vpxmem,g_logging)
+#define memtrack vpxglobalm(vpxmem,memtrack)
+#endif // #ifndef VPX_NO_GLOBALS
+
+extern void *vpx_malloc(size_t size);
+extern void vpx_free(void *memblk);
+extern void *vpx_memcpy(void *dest, const void *src, size_t length);
+extern void *vpx_memset(void *dest, int val, size_t length);
+
+/*
+ *
+ * Exposed library functions
+ *
+*/
+
+/*
+ vpx_memory_tracker_init(int padding_size, int pad_value)
+ padding_size - the size of the padding before and after each mem addr.
+ Values > 0 indicate that integrity checks can be performed
+ by inspecting these areas.
+ pad_value - the initial value within the padding area before and after
+ each mem addr.
+
+ Initializes global memory tracker structure
+ Allocates the head of the list
+*/
+int vpx_memory_tracker_init(int padding_size, int pad_value)
+{
+ if (!g_b_mem_tracker_inited)
+ {
+ if ((memtrack.head = (struct mem_block *)
+ MEM_TRACK_MALLOC(sizeof(struct mem_block))))
+ {
+ int ret;
+
+ MEM_TRACK_MEMSET(memtrack.head, 0, sizeof(struct mem_block));
+
+ memtrack.tail = memtrack.head;
+
+ memtrack.current_allocated = 0;
+ memtrack.max_allocated = 0;
+
+ memtrack.padding_size = padding_size;
+ memtrack.pad_value = pad_value;
+
+#if HAVE_PTHREAD_H
+ ret = pthread_mutex_init(&memtrack.mutex,
+ NULL); /*mutex attributes (NULL=default)*/
+#elif defined(WIN32) || defined(_WIN32_WCE)
+ memtrack.mutex = CreateMutex(NULL, /*security attributes*/
+ FALSE, /*we don't want initial ownership*/
+ NULL); /*mutex name*/
+ ret = !memtrack.mutex;
+#elif defined(VXWORKS)
+ memtrack.mutex = sem_bcreate(SEM_Q_FIFO, /*SEM_Q_FIFO non-priority based mutex*/
+ SEM_FULL); /*SEM_FULL initial state is unlocked*/
+ ret = !memtrack.mutex;
+#elif defined(NO_MUTEX)
+ ret = 0;
+#endif
+
+ if (ret)
+ {
+ memtrack_log("vpx_memory_tracker_init: Error creating mutex!\n");
+
+ MEM_TRACK_FREE(memtrack.head);
+ memtrack.head = NULL;
+ }
+ else
+ {
+ memtrack_log("Memory Tracker init'd, v."vpx_mem_tracker_version" pad_size:%d pad_val:0x%x %d\n"
+ , padding_size
+ , pad_value
+ , pad_value);
+ g_b_mem_tracker_inited = 1;
+ }
+ }
+ }
+
+ return g_b_mem_tracker_inited;
+}
+
+/*
+ vpx_memory_tracker_destroy()
+ If our global struct was initialized zeros out all its members,
+ frees memory and destroys it's mutex
+*/
+void vpx_memory_tracker_destroy()
+{
+ if (!memory_tracker_lock_mutex())
+ {
+ struct mem_block *p = memtrack.head,
+ * p2 = memtrack.head;
+
+ memory_tracker_dump();
+
+ while (p)
+ {
+ p2 = p;
+ p = p->next;
+
+ MEM_TRACK_FREE(p2);
+ }
+
+ memtrack.head = NULL;
+ memtrack.tail = NULL;
+ memtrack.len = 0;
+ memtrack.current_allocated = 0;
+ memtrack.max_allocated = 0;
+
+ if (!g_logging.type && g_logging.file && g_logging.file != stderr)
+ {
+ fclose(g_logging.file);
+ g_logging.file = NULL;
+ }
+
+ memory_tracker_unlock_mutex();
+
+ g_b_mem_tracker_inited = 0;
+ }
+}
+
+/*
+ vpx_memory_tracker_add(size_t addr, unsigned int size,
+ char * file, unsigned int line)
+ addr - memory address to be added to list
+ size - size of addr
+ file - the file addr was referenced from
+ line - the line in file addr was referenced from
+ Adds memory address addr, it's size, file and line it came from
+ to the global list via the thread safe internal library function
+*/
+void vpx_memory_tracker_add(size_t addr, unsigned int size,
+ char *file, unsigned int line,
+ int padded)
+{
+ memory_tracker_add(addr, size, file, line, padded);
+}
+
+/*
+ vpx_memory_tracker_remove(size_t addr)
+ addr - memory address to be removed from list
+ Removes addr from the global list via the thread safe
+ internal remove function
+ Return:
+ Same as described for memory_tracker_remove
+*/
+int vpx_memory_tracker_remove(size_t addr)
+{
+ return memory_tracker_remove(addr);
+}
+
+/*
+ vpx_memory_tracker_find(size_t addr)
+ addr - address to be found in list
+ Return:
+ If found, pointer to the memory block that matches addr
+ NULL otherwise
+*/
+struct mem_block *vpx_memory_tracker_find(size_t addr)
+{
+ struct mem_block *p = NULL;
+
+ if (!memory_tracker_lock_mutex())
+ {
+ p = memory_tracker_find(addr);
+ memory_tracker_unlock_mutex();
+ }
+
+ return p;
+}
+
+/*
+ vpx_memory_tracker_dump()
+ Locks the memory tracker's mutex and calls the internal
+ library function to dump the current contents of the
+ global memory allocation list
+*/
+void vpx_memory_tracker_dump()
+{
+ if (!memory_tracker_lock_mutex())
+ {
+ memory_tracker_dump();
+ memory_tracker_unlock_mutex();
+ }
+}
+
+/*
+ vpx_memory_tracker_check_integrity(char* file, unsigned int line)
+ file - The file name where the check was placed
+ line - The line in file where the check was placed
+ Locks the memory tracker's mutex and calls the internal
+ integrity check function to inspect every address in the global
+ memory allocation list
+*/
+void vpx_memory_tracker_check_integrity(char *file, unsigned int line)
+{
+ if (!memory_tracker_lock_mutex())
+ {
+ memory_tracker_check_integrity(file, line);
+ memory_tracker_unlock_mutex();
+ }
+}
+
+/*
+ vpx_memory_tracker_set_log_type
+ Sets the logging type for the memory tracker. Based on the value it will
+ direct its output to the appropriate place.
+ Return:
+ 0: on success
+ -1: if the logging type could not be set, because the value was invalid
+ or because a file could not be opened
+*/
+int vpx_memory_tracker_set_log_type(int type, char *option)
+{
+ int ret = -1;
+
+ switch (type)
+ {
+ case 0:
+ g_logging.type = 0;
+
+ if (!option)
+ {
+ g_logging.file = stderr;
+ ret = 0;
+ }
+ else
+ {
+ if ((g_logging.file = fopen((char *)option, "w")))
+ ret = 0;
+ }
+
+ break;
+#if defined(WIN32) && !defined(_WIN32_WCE)
+ case 1:
+ g_logging.type = type;
+ ret = 0;
+ break;
+#endif
+ default:
+ break;
+ }
+
+ //output the version to the new logging destination
+ if (!ret)
+ memtrack_log("Memory Tracker logging initialized, "
+ "Memory Tracker v."vpx_mem_tracker_version"\n");
+
+ return ret;
+}
+
+/*
+ vpx_memory_tracker_set_log_func
+ Sets a logging function to be used by the memory tracker.
+ Return:
+ 0: on success
+ -1: if the logging type could not be set because logfunc was NULL
+*/
+int vpx_memory_tracker_set_log_func(void *userdata,
+ void(*logfunc)(void *userdata,
+ const char *fmt, va_list args))
+{
+ int ret = -1;
+
+ if (logfunc)
+ {
+ g_logging.type = -1;
+ g_logging.userdata = userdata;
+ g_logging.func = logfunc;
+ ret = 0;
+ }
+
+ //output the version to the new logging destination
+ if (!ret)
+ memtrack_log("Memory Tracker logging initialized, "
+ "Memory Tracker v."vpx_mem_tracker_version"\n");
+
+ return ret;
+}
+
+/*
+ *
+ * END - Exposed library functions
+ *
+*/
+
+
+/*
+ *
+ * Internal library functions
+ *
+*/
+
+static void memtrack_log(const char *fmt, ...)
+{
+ va_list list;
+
+ va_start(list, fmt);
+
+ switch (g_logging.type)
+ {
+ case -1:
+
+ if (g_logging.func)
+ g_logging.func(g_logging.userdata, fmt, list);
+
+ break;
+ case 0:
+
+ if (g_logging.file)
+ {
+ vfprintf(g_logging.file, fmt, list);
+ fflush(g_logging.file);
+ }
+
+ break;
+#if defined(WIN32) && !defined(_WIN32_WCE)
+ case 1:
+ {
+ char temp[1024];
+ _vsnprintf(temp, sizeof(temp) / sizeof(char) - 1, fmt, list);
+ OutputDebugString(temp);
+ }
+ break;
+#endif
+ default:
+ break;
+ }
+
+ va_end(list);
+}
+
+/*
+ memory_tracker_dump()
+ Dumps the current contents of the global memory allocation list
+*/
+static void memory_tracker_dump()
+{
+ int i = 0;
+ struct mem_block *p = (memtrack.head ? memtrack.head->next : NULL);
+
+ memtrack_log("\n_currently Allocated= %d; Max allocated= %d\n",
+ memtrack.current_allocated, memtrack.max_allocated);
+
+ while (p)
+ {
+#if defined(WIN32) && !defined(_WIN32_WCE)
+
+ /*when using outputdebugstring, output filenames so they
+ can be clicked to be opened in visual studio*/
+ if (g_logging.type == 1)
+ memtrack_log("memblocks[%d].addr= 0x%.8x, memblocks[%d].size= %d, file:\n"
+ " %s(%d):\n", i,
+ p->addr, i, p->size,
+ p->file, p->line);
+ else
+#endif
+ memtrack_log("memblocks[%d].addr= 0x%.8x, memblocks[%d].size= %d, file: %s, line: %d\n", i,
+ p->addr, i, p->size,
+ p->file, p->line);
+
+ p = p->next;
+ ++i;
+ }
+
+ memtrack_log("\n");
+}
+
+/*
+ memory_tracker_check_integrity(char* file, unsigned int file)
+ file - the file name where the check was placed
+ line - the line in file where the check was placed
+ If a padding_size was supplied to vpx_memory_tracker_init()
+ this function will check ea. addr in the list verifying that
+ addr-padding_size and addr+padding_size is filled with pad_value
+*/
+static void memory_tracker_check_integrity(char *file, unsigned int line)
+{
+ if (memtrack.padding_size)
+ {
+ int i,
+ index = 0;
+ unsigned char *p_show_me,
+ * p_show_me2;
+ unsigned int tempme = memtrack.pad_value,
+ dead1,
+ dead2;
+ unsigned char *x_bounds;
+ struct mem_block *p = memtrack.head->next;
+
+ while (p)
+ {
+ //x_bounds = (unsigned char*)p->addr;
+ //back up VPX_BYTE_ALIGNMENT
+ //x_bounds -= memtrack.padding_size;
+
+ if (p->padded) // can the bounds be checked?
+ {
+ /*yes, move to the address that was actually allocated
+ by the vpx_* calls*/
+ x_bounds = (unsigned char *)(((size_t *)p->addr)[-1]);
+
+ for (i = 0; i < memtrack.padding_size; i += sizeof(unsigned int))
+ {
+ p_show_me = (x_bounds + i);
+ p_show_me2 = (unsigned char *)(p->addr + p->size + i);
+
+ MEM_TRACK_MEMCPY(&dead1, p_show_me, sizeof(unsigned int));
+ MEM_TRACK_MEMCPY(&dead2, p_show_me2, sizeof(unsigned int));
+
+ if ((dead1 != tempme) || (dead2 != tempme))
+ {
+ memtrack_log("\n[vpx_mem integrity check failed]:\n"
+ " index[%d,%d] {%s:%d} addr=0x%x, size=%d,"
+ " file: %s, line: %d c0:0x%x c1:0x%x\n",
+ index, i, file, line, p->addr, p->size, p->file,
+ p->line, dead1, dead2);
+ }
+ }
+ }
+
+ ++index;
+ p = p->next;
+ }
+ }
+}
+
+/*
+ memory_tracker_add(size_t addr, unsigned int size,
+ char * file, unsigned int line)
+ Adds an address (addr), it's size, file and line number to our list.
+ Adjusts the total bytes allocated and max bytes allocated if necessary.
+ If memory cannot be allocated the list will be destroyed.
+*/
+void memory_tracker_add(size_t addr, unsigned int size,
+ char *file, unsigned int line,
+ int padded)
+{
+ if (!memory_tracker_lock_mutex())
+ {
+ struct mem_block *p;
+
+ p = MEM_TRACK_MALLOC(sizeof(struct mem_block));
+
+ if (p)
+ {
+ p->prev = memtrack.tail;
+ p->prev->next = p;
+ p->addr = addr;
+ p->size = size;
+ p->line = line;
+ p->file = file;
+ p->padded = padded;
+ p->next = NULL;
+
+ memtrack.tail = p;
+
+ memtrack.current_allocated += size;
+
+ if (memtrack.current_allocated > memtrack.max_allocated)
+ memtrack.max_allocated = memtrack.current_allocated;
+
+ //memtrack_log("memory_tracker_add: added addr=0x%.8x\n", addr);
+
+ memory_tracker_unlock_mutex();
+ }
+ else
+ {
+ memtrack_log("memory_tracker_add: error allocating memory!\n");
+ memory_tracker_unlock_mutex();
+ vpx_memory_tracker_destroy();
+ }
+ }
+}
+
+/*
+ memory_tracker_remove(size_t addr)
+ Removes an address and its corresponding size (if they exist)
+ from the memory tracker list and adjusts the current number
+ of bytes allocated.
+ Return:
+ 0: on success
+ -1: if the mutex could not be locked
+ -2: if the addr was not found in the list
+*/
+int memory_tracker_remove(size_t addr)
+{
+ int ret = -1;
+
+ if (!memory_tracker_lock_mutex())
+ {
+ struct mem_block *p;
+
+ if ((p = memory_tracker_find(addr)))
+ {
+ memtrack.current_allocated -= p->size;
+
+ p->prev->next = p->next;
+
+ if (p->next)
+ p->next->prev = p->prev;
+ else
+ memtrack.tail = p->prev;
+
+ ret = 0;
+ MEM_TRACK_FREE(p);
+ }
+ else
+ {
+ if (addr)
+ memtrack_log("memory_tracker_remove(): addr not found in list,"
+ " 0x%.8x\n", addr);
+
+ ret = -2;
+ }
+
+ memory_tracker_unlock_mutex();
+ }
+
+ return ret;
+}
+
+/*
+ memory_tracker_find(size_t addr)
+ Finds an address in our addrs list
+ NOTE: the mutex MUST be locked in the other internal
+ functions before calling this one. This avoids
+ the need for repeated locking and unlocking as in Remove
+ Returns: pointer to the mem block if found, NULL otherwise
+*/
+static struct mem_block *memory_tracker_find(size_t addr)
+{
+ struct mem_block *p = NULL;
+
+ if (memtrack.head)
+ {
+ p = memtrack.head->next;
+
+ while (p && (p->addr != addr))
+ p = p->next;
+ }
+
+ return p;
+}
+
+
+#if !defined(NO_MUTEX)
+/*
+ memory_tracker_lock_mutex()
+ Locks the memory tracker mutex with a platform specific call
+ Returns:
+ 0: Success
+ <0: Failure, either the mutex was not initialized
+ or the call to lock the mutex failed
+*/
+static int memory_tracker_lock_mutex()
+{
+ int ret = -1;
+
+ if (g_b_mem_tracker_inited)
+ {
+
+#if HAVE_PTHREAD_H
+ ret = pthread_mutex_lock(&memtrack.mutex);
+#elif defined(WIN32) || defined(_WIN32_WCE)
+ ret = WaitForSingleObject(memtrack.mutex, INFINITE);
+#elif defined(VXWORKS)
+ ret = sem_take(memtrack.mutex, WAIT_FOREVER);
+#endif
+
+ if (ret)
+ {
+ memtrack_log("memory_tracker_lock_mutex: mutex lock failed\n");
+ }
+ }
+
+ return ret;
+}
+
+/*
+ memory_tracker_unlock_mutex()
+ Unlocks the memory tracker mutex with a platform specific call
+ Returns:
+ 0: Success
+ <0: Failure, either the mutex was not initialized
+ or the call to unlock the mutex failed
+*/
+static int memory_tracker_unlock_mutex()
+{
+ int ret = -1;
+
+ if (g_b_mem_tracker_inited)
+ {
+
+#if HAVE_PTHREAD_H
+ ret = pthread_mutex_unlock(&memtrack.mutex);
+#elif defined(WIN32) || defined(_WIN32_WCE)
+ ret = !ReleaseMutex(memtrack.mutex);
+#elif defined(VXWORKS)
+ ret = sem_give(memtrack.mutex);
+#endif
+
+ if (ret)
+ {
+ memtrack_log("memory_tracker_unlock_mutex: mutex unlock failed\n");
+ }
+ }
+
+ return ret;
+}
+#endif
+
+/*
+ vpx_memory_tracker_set_functions
+
+ Sets the function pointers for the standard library functions.
+
+ Return:
+ 0: on success
+ -1: if the use global function pointers is not set.
+*/
+int vpx_memory_tracker_set_functions(mem_track_malloc_func g_malloc_l
+ , mem_track_calloc_func g_calloc_l
+ , mem_track_realloc_func g_realloc_l
+ , mem_track_free_func g_free_l
+ , mem_track_memcpy_func g_memcpy_l
+ , mem_track_memset_func g_memset_l
+ , mem_track_memmove_func g_memmove_l)
+{
+#if USE_GLOBAL_FUNCTION_POINTERS
+
+ if (g_malloc_l)
+ g_malloc = g_malloc_l;
+
+ if (g_calloc_l)
+ g_calloc = g_calloc_l;
+
+ if (g_realloc_l)
+ g_realloc = g_realloc_l;
+
+ if (g_free_l)
+ g_free = g_free_l;
+
+ if (g_memcpy_l)
+ g_memcpy = g_memcpy_l;
+
+ if (g_memset_l)
+ g_memset = g_memset_l;
+
+ if (g_memmove_l)
+ g_memmove = g_memmove_l;
+
+ return 0;
+#else
+ (void)g_malloc_l;
+ (void)g_calloc_l;
+ (void)g_realloc_l;
+ (void)g_free_l;
+ (void)g_memcpy_l;
+ (void)g_memset_l;
+ (void)g_memmove_l;
+ return -1;
+#endif
+}
diff --git a/libvpx/vpx_ports/arm.h b/libvpx/vpx_ports/arm.h
new file mode 100644
index 0000000..525a764
--- /dev/null
+++ b/libvpx/vpx_ports/arm.h
@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef VPX_PORTS_ARM_H
+#define VPX_PORTS_ARM_H
+#include <stdlib.h>
+#include "vpx_config.h"
+
+/*ARMv5TE "Enhanced DSP" instructions.*/
+#define HAS_EDSP 0x01
+/*ARMv6 "Parallel" or "Media" instructions.*/
+#define HAS_MEDIA 0x02
+/*ARMv7 optional NEON instructions.*/
+#define HAS_NEON 0x04
+
+int arm_cpu_caps(void);
+
+#endif
+
diff --git a/libvpx/vpx_ports/arm_cpudetect.c b/libvpx/vpx_ports/arm_cpudetect.c
new file mode 100644
index 0000000..8ff95a1
--- /dev/null
+++ b/libvpx/vpx_ports/arm_cpudetect.c
@@ -0,0 +1,219 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include <stdlib.h>
+#include <string.h>
+#include "arm.h"
+
+static int arm_cpu_env_flags(int *flags)
+{
+ char *env;
+ env = getenv("VPX_SIMD_CAPS");
+ if (env && *env)
+ {
+ *flags = (int)strtol(env, NULL, 0);
+ return 0;
+ }
+ *flags = 0;
+ return -1;
+}
+
+static int arm_cpu_env_mask(void)
+{
+ char *env;
+ env = getenv("VPX_SIMD_CAPS_MASK");
+ return env && *env ? (int)strtol(env, NULL, 0) : ~0;
+}
+
+#if !CONFIG_RUNTIME_CPU_DETECT
+
+int arm_cpu_caps(void)
+{
+ /* This function should actually be a no-op. There is no way to adjust any of
+ * these because the RTCD tables do not exist: the functions are called
+ * statically */
+ int flags;
+ int mask;
+ if (!arm_cpu_env_flags(&flags))
+ {
+ return flags;
+ }
+ mask = arm_cpu_env_mask();
+#if HAVE_EDSP
+ flags |= HAS_EDSP;
+#endif /* HAVE_EDSP */
+#if HAVE_MEDIA
+ flags |= HAS_MEDIA;
+#endif /* HAVE_MEDIA */
+#if HAVE_NEON
+ flags |= HAS_NEON;
+#endif /* HAVE_NEON */
+ return flags & mask;
+}
+
+#elif defined(_MSC_VER) /* end !CONFIG_RUNTIME_CPU_DETECT */
+/*For GetExceptionCode() and EXCEPTION_ILLEGAL_INSTRUCTION.*/
+#define WIN32_LEAN_AND_MEAN
+#define WIN32_EXTRA_LEAN
+#include <windows.h>
+
+int arm_cpu_caps(void)
+{
+ int flags;
+ int mask;
+ if (!arm_cpu_env_flags(&flags))
+ {
+ return flags;
+ }
+ mask = arm_cpu_env_mask();
+ /* MSVC has no inline __asm support for ARM, but it does let you __emit
+ * instructions via their assembled hex code.
+ * All of these instructions should be essentially nops.
+ */
+#if HAVE_EDSP
+ if (mask & HAS_EDSP)
+ {
+ __try
+ {
+ /*PLD [r13]*/
+ __emit(0xF5DDF000);
+ flags |= HAS_EDSP;
+ }
+ __except(GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION)
+ {
+ /*Ignore exception.*/
+ }
+ }
+#if HAVE_MEDIA
+ if (mask & HAS_MEDIA)
+ __try
+ {
+ /*SHADD8 r3,r3,r3*/
+ __emit(0xE6333F93);
+ flags |= HAS_MEDIA;
+ }
+ __except(GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION)
+ {
+ /*Ignore exception.*/
+ }
+ }
+#if HAVE_NEON
+ if (mask & HAS_NEON)
+ {
+ __try
+ {
+ /*VORR q0,q0,q0*/
+ __emit(0xF2200150);
+ flags |= HAS_NEON;
+ }
+ __except(GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION)
+ {
+ /*Ignore exception.*/
+ }
+ }
+#endif /* HAVE_NEON */
+#endif /* HAVE_MEDIA */
+#endif /* HAVE_EDSP */
+ return flags & mask;
+}
+
+#elif defined(__ANDROID__) /* end _MSC_VER */
+#include <cpu-features.h>
+
+int arm_cpu_caps(void)
+{
+ int flags;
+ int mask;
+ uint64_t features;
+ if (!arm_cpu_env_flags(&flags))
+ {
+ return flags;
+ }
+ mask = arm_cpu_env_mask();
+ features = android_getCpuFeatures();
+
+#if HAVE_EDSP
+ flags |= HAS_EDSP;
+#endif /* HAVE_EDSP */
+#if HAVE_MEDIA
+ flags |= HAS_MEDIA;
+#endif /* HAVE_MEDIA */
+#if HAVE_NEON
+ if (features & ANDROID_CPU_ARM_FEATURE_NEON)
+ flags |= HAS_NEON;
+#endif /* HAVE_NEON */
+ return flags & mask;
+}
+
+#elif defined(__linux__) /* end __ANDROID__ */
+#include <stdio.h>
+
+int arm_cpu_caps(void)
+{
+ FILE *fin;
+ int flags;
+ int mask;
+ if (!arm_cpu_env_flags(&flags))
+ {
+ return flags;
+ }
+ mask = arm_cpu_env_mask();
+ /* Reading /proc/self/auxv would be easier, but that doesn't work reliably
+ * on Android.
+ * This also means that detection will fail in Scratchbox.
+ */
+ fin = fopen("/proc/cpuinfo","r");
+ if(fin != NULL)
+ {
+ /* 512 should be enough for anybody (it's even enough for all the flags
+ * that x86 has accumulated... so far).
+ */
+ char buf[512];
+ while (fgets(buf, 511, fin) != NULL)
+ {
+#if HAVE_EDSP || HAVE_NEON
+ if (memcmp(buf, "Features", 8) == 0)
+ {
+ char *p;
+#if HAVE_EDSP
+ p=strstr(buf, " edsp");
+ if (p != NULL && (p[5] == ' ' || p[5] == '\n'))
+ {
+ flags |= HAS_EDSP;
+ }
+#if HAVE_NEON
+ p = strstr(buf, " neon");
+ if (p != NULL && (p[5] == ' ' || p[5] == '\n'))
+ {
+ flags |= HAS_NEON;
+ }
+#endif /* HAVE_NEON */
+#endif /* HAVE_EDSP */
+ }
+#endif /* HAVE_EDSP || HAVE_NEON */
+#if HAVE_MEDIA
+ if (memcmp(buf, "CPU architecture:",17) == 0){
+ int version;
+ version = atoi(buf+17);
+ if (version >= 6)
+ {
+ flags |= HAS_MEDIA;
+ }
+ }
+#endif /* HAVE_MEDIA */
+ }
+ fclose(fin);
+ }
+ return flags & mask;
+}
+#else /* end __linux__ */
+#error "--enable-runtime-cpu-detect selected, but no CPU detection method " \
+ "available for your platform. Reconfigure with --disable-runtime-cpu-detect."
+#endif
diff --git a/libvpx/vpx_ports/asm_offsets.h b/libvpx/vpx_ports/asm_offsets.h
new file mode 100644
index 0000000..7b6ae4a
--- /dev/null
+++ b/libvpx/vpx_ports/asm_offsets.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef VPX_PORTS_ASM_OFFSETS_H
+#define VPX_PORTS_ASM_OFFSETS_H
+
+#include <stddef.h>
+
+#define ct_assert(name,cond) \
+ static void assert_##name(void) UNUSED;\
+ static void assert_##name(void) {switch(0){case 0:case !!(cond):;}}
+
+#if INLINE_ASM
+#define DEFINE(sym, val) asm("\n" #sym " EQU %0" : : "i" (val))
+#define BEGIN int main(void) {
+#define END return 0; }
+#else
+#define DEFINE(sym, val) const int sym = val
+#define BEGIN
+#define END
+#endif
+
+#endif /* VPX_PORTS_ASM_OFFSETS_H */
diff --git a/libvpx/vpx_ports/emms.asm b/libvpx/vpx_ports/emms.asm
new file mode 100644
index 0000000..efad1a5
--- /dev/null
+++ b/libvpx/vpx_ports/emms.asm
@@ -0,0 +1,38 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+section .text
+global sym(vpx_reset_mmx_state) PRIVATE
+sym(vpx_reset_mmx_state):
+ emms
+ ret
+
+
+%ifidn __OUTPUT_FORMAT__,x64
+global sym(vpx_winx64_fldcw) PRIVATE
+sym(vpx_winx64_fldcw):
+ sub rsp, 8
+ mov [rsp], rcx ; win x64 specific
+ fldcw [rsp]
+ add rsp, 8
+ ret
+
+
+global sym(vpx_winx64_fstcw) PRIVATE
+sym(vpx_winx64_fstcw):
+ sub rsp, 8
+ fstcw [rsp]
+ mov rax, [rsp]
+ add rsp, 8
+ ret
+%endif
diff --git a/libvpx/vpx_ports/mem.h b/libvpx/vpx_ports/mem.h
new file mode 100644
index 0000000..29e507f
--- /dev/null
+++ b/libvpx/vpx_ports/mem.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef VPX_PORTS_MEM_H
+#define VPX_PORTS_MEM_H
+#include "vpx_config.h"
+#include "vpx/vpx_integer.h"
+
+#if (defined(__GNUC__) && __GNUC__) || defined(__SUNPRO_C)
+#define DECLARE_ALIGNED(n,typ,val) typ val __attribute__ ((aligned (n)))
+#elif defined(_MSC_VER)
+#define DECLARE_ALIGNED(n,typ,val) __declspec(align(n)) typ val
+#else
+#warning No alignment directives known for this compiler.
+#define DECLARE_ALIGNED(n,typ,val) typ val
+#endif
+#endif
+
+
+/* Declare an aligned array on the stack, for situations where the stack
+ * pointer may not have the alignment we expect. Creates an array with a
+ * modified name, then defines val to be a pointer, and aligns that pointer
+ * within the array.
+ */
+#define DECLARE_ALIGNED_ARRAY(a,typ,val,n)\
+typ val##_[(n)+(a)/sizeof(typ)+1];\
+typ *val = (typ*)((((intptr_t)val##_)+(a)-1)&((intptr_t)-(a)))
+
+
+/* Indicates that the usage of the specified variable has been audited to assure
+ * that it's safe to use uninitialized. Silences 'may be used uninitialized'
+ * warnings on gcc.
+ */
+#if defined(__GNUC__) && __GNUC__
+#define UNINITIALIZED_IS_SAFE(x) x=x
+#else
+#define UNINITIALIZED_IS_SAFE(x) x
+#endif
diff --git a/libvpx/vpx_ports/mem_ops.h b/libvpx/vpx_ports/mem_ops.h
new file mode 100644
index 0000000..dec28d5
--- /dev/null
+++ b/libvpx/vpx_ports/mem_ops.h
@@ -0,0 +1,234 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/* \file
+ * \brief Provides portable memory access primitives
+ *
+ * This function provides portable primitives for getting and setting of
+ * signed and unsigned integers in 16, 24, and 32 bit sizes. The operations
+ * can be performed on unaligned data regardless of hardware support for
+ * unaligned accesses.
+ *
+ * The type used to pass the integral values may be changed by defining
+ * MEM_VALUE_T with the appropriate type. The type given must be an integral
+ * numeric type.
+ *
+ * The actual functions instantiated have the MEM_VALUE_T type name pasted
+ * on to the symbol name. This allows the developer to instantiate these
+ * operations for multiple types within the same translation unit. This is
+ * of somewhat questionable utility, but the capability exists nonetheless.
+ * Users not making use of this functionality should call the functions
+ * without the type name appended, and the preprocessor will take care of
+ * it.
+ *
+ * NOTE: This code is not supported on platforms where char > 1 octet ATM.
+ */
+
+#ifndef MAU_T
+/* Minimum Access Unit for this target */
+#define MAU_T unsigned char
+#endif
+
+#ifndef MEM_VALUE_T
+#define MEM_VALUE_T int
+#endif
+
+#undef MEM_VALUE_T_SZ_BITS
+#define MEM_VALUE_T_SZ_BITS (sizeof(MEM_VALUE_T) << 3)
+
+#undef mem_ops_wrap_symbol
+#define mem_ops_wrap_symbol(fn) mem_ops_wrap_symbol2(fn, MEM_VALUE_T)
+#undef mem_ops_wrap_symbol2
+#define mem_ops_wrap_symbol2(fn,typ) mem_ops_wrap_symbol3(fn,typ)
+#undef mem_ops_wrap_symbol3
+#define mem_ops_wrap_symbol3(fn,typ) fn##_as_##typ
+
+/*
+ * Include aligned access routines
+ */
+#define INCLUDED_BY_MEM_OPS_H
+#include "mem_ops_aligned.h"
+#undef INCLUDED_BY_MEM_OPS_H
+
+#undef mem_get_be16
+#define mem_get_be16 mem_ops_wrap_symbol(mem_get_be16)
+static unsigned MEM_VALUE_T mem_get_be16(const void *vmem)
+{
+ unsigned MEM_VALUE_T val;
+ const MAU_T *mem = (const MAU_T *)vmem;
+
+ val = mem[0] << 8;
+ val |= mem[1];
+ return val;
+}
+
+#undef mem_get_be24
+#define mem_get_be24 mem_ops_wrap_symbol(mem_get_be24)
+static unsigned MEM_VALUE_T mem_get_be24(const void *vmem)
+{
+ unsigned MEM_VALUE_T val;
+ const MAU_T *mem = (const MAU_T *)vmem;
+
+ val = mem[0] << 16;
+ val |= mem[1] << 8;
+ val |= mem[2];
+ return val;
+}
+
+#undef mem_get_be32
+#define mem_get_be32 mem_ops_wrap_symbol(mem_get_be32)
+static unsigned MEM_VALUE_T mem_get_be32(const void *vmem)
+{
+ unsigned MEM_VALUE_T val;
+ const MAU_T *mem = (const MAU_T *)vmem;
+
+ val = mem[0] << 24;
+ val |= mem[1] << 16;
+ val |= mem[2] << 8;
+ val |= mem[3];
+ return val;
+}
+
+#undef mem_get_le16
+#define mem_get_le16 mem_ops_wrap_symbol(mem_get_le16)
+static unsigned MEM_VALUE_T mem_get_le16(const void *vmem)
+{
+ unsigned MEM_VALUE_T val;
+ const MAU_T *mem = (const MAU_T *)vmem;
+
+ val = mem[1] << 8;
+ val |= mem[0];
+ return val;
+}
+
+#undef mem_get_le24
+#define mem_get_le24 mem_ops_wrap_symbol(mem_get_le24)
+static unsigned MEM_VALUE_T mem_get_le24(const void *vmem)
+{
+ unsigned MEM_VALUE_T val;
+ const MAU_T *mem = (const MAU_T *)vmem;
+
+ val = mem[2] << 16;
+ val |= mem[1] << 8;
+ val |= mem[0];
+ return val;
+}
+
+#undef mem_get_le32
+#define mem_get_le32 mem_ops_wrap_symbol(mem_get_le32)
+static unsigned MEM_VALUE_T mem_get_le32(const void *vmem)
+{
+ unsigned MEM_VALUE_T val;
+ const MAU_T *mem = (const MAU_T *)vmem;
+
+ val = mem[3] << 24;
+ val |= mem[2] << 16;
+ val |= mem[1] << 8;
+ val |= mem[0];
+ return val;
+}
+
+#define mem_get_s_generic(end,sz) \
+ static signed MEM_VALUE_T mem_get_s##end##sz(const void *vmem) {\
+ const MAU_T *mem = (const MAU_T*)vmem;\
+ signed MEM_VALUE_T val = mem_get_##end##sz(mem);\
+ return (val << (MEM_VALUE_T_SZ_BITS - sz)) >> (MEM_VALUE_T_SZ_BITS - sz);\
+ }
+
+#undef mem_get_sbe16
+#define mem_get_sbe16 mem_ops_wrap_symbol(mem_get_sbe16)
+mem_get_s_generic(be, 16)
+
+#undef mem_get_sbe24
+#define mem_get_sbe24 mem_ops_wrap_symbol(mem_get_sbe24)
+mem_get_s_generic(be, 24)
+
+#undef mem_get_sbe32
+#define mem_get_sbe32 mem_ops_wrap_symbol(mem_get_sbe32)
+mem_get_s_generic(be, 32)
+
+#undef mem_get_sle16
+#define mem_get_sle16 mem_ops_wrap_symbol(mem_get_sle16)
+mem_get_s_generic(le, 16)
+
+#undef mem_get_sle24
+#define mem_get_sle24 mem_ops_wrap_symbol(mem_get_sle24)
+mem_get_s_generic(le, 24)
+
+#undef mem_get_sle32
+#define mem_get_sle32 mem_ops_wrap_symbol(mem_get_sle32)
+mem_get_s_generic(le, 32)
+
+#undef mem_put_be16
+#define mem_put_be16 mem_ops_wrap_symbol(mem_put_be16)
+static void mem_put_be16(void *vmem, MEM_VALUE_T val)
+{
+ MAU_T *mem = (MAU_T *)vmem;
+
+ mem[0] = (val >> 8) & 0xff;
+ mem[1] = (val >> 0) & 0xff;
+}
+
+#undef mem_put_be24
+#define mem_put_be24 mem_ops_wrap_symbol(mem_put_be24)
+static void mem_put_be24(void *vmem, MEM_VALUE_T val)
+{
+ MAU_T *mem = (MAU_T *)vmem;
+
+ mem[0] = (val >> 16) & 0xff;
+ mem[1] = (val >> 8) & 0xff;
+ mem[2] = (val >> 0) & 0xff;
+}
+
+#undef mem_put_be32
+#define mem_put_be32 mem_ops_wrap_symbol(mem_put_be32)
+static void mem_put_be32(void *vmem, MEM_VALUE_T val)
+{
+ MAU_T *mem = (MAU_T *)vmem;
+
+ mem[0] = (val >> 24) & 0xff;
+ mem[1] = (val >> 16) & 0xff;
+ mem[2] = (val >> 8) & 0xff;
+ mem[3] = (val >> 0) & 0xff;
+}
+
+#undef mem_put_le16
+#define mem_put_le16 mem_ops_wrap_symbol(mem_put_le16)
+static void mem_put_le16(void *vmem, MEM_VALUE_T val)
+{
+ MAU_T *mem = (MAU_T *)vmem;
+
+ mem[0] = (val >> 0) & 0xff;
+ mem[1] = (val >> 8) & 0xff;
+}
+
+#undef mem_put_le24
+#define mem_put_le24 mem_ops_wrap_symbol(mem_put_le24)
+static void mem_put_le24(void *vmem, MEM_VALUE_T val)
+{
+ MAU_T *mem = (MAU_T *)vmem;
+
+ mem[0] = (val >> 0) & 0xff;
+ mem[1] = (val >> 8) & 0xff;
+ mem[2] = (val >> 16) & 0xff;
+}
+
+#undef mem_put_le32
+#define mem_put_le32 mem_ops_wrap_symbol(mem_put_le32)
+static void mem_put_le32(void *vmem, MEM_VALUE_T val)
+{
+ MAU_T *mem = (MAU_T *)vmem;
+
+ mem[0] = (val >> 0) & 0xff;
+ mem[1] = (val >> 8) & 0xff;
+ mem[2] = (val >> 16) & 0xff;
+ mem[3] = (val >> 24) & 0xff;
+}
diff --git a/libvpx/vpx_ports/mem_ops_aligned.h b/libvpx/vpx_ports/mem_ops_aligned.h
new file mode 100644
index 0000000..fca653a
--- /dev/null
+++ b/libvpx/vpx_ports/mem_ops_aligned.h
@@ -0,0 +1,157 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/* \file
+ * \brief Provides portable memory access primitives for operating on aligned
+ * data
+ *
+ * This file is split from mem_ops.h for easier maintenance. See mem_ops.h
+ * for a more detailed description of these primitives.
+ */
+#ifndef INCLUDED_BY_MEM_OPS_H
+#error Include mem_ops.h, not mem_ops_aligned.h directly.
+#endif
+
+/* Architectures that provide instructions for doing this byte swapping
+ * could redefine these macros.
+ */
+#define swap_endian_16(val,raw) do {\
+ val = ((raw>>8) & 0x00ff) \
+ | ((raw<<8) & 0xff00);\
+ } while(0)
+#define swap_endian_32(val,raw) do {\
+ val = ((raw>>24) & 0x000000ff) \
+ | ((raw>>8) & 0x0000ff00) \
+ | ((raw<<8) & 0x00ff0000) \
+ | ((raw<<24) & 0xff000000); \
+ } while(0)
+#define swap_endian_16_se(val,raw) do {\
+ swap_endian_16(val,raw);\
+ val = ((val << 16) >> 16);\
+ } while(0)
+#define swap_endian_32_se(val,raw) swap_endian_32(val,raw)
+
+#define mem_get_ne_aligned_generic(end,sz) \
+ static unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\
+ const uint##sz##_t *mem = (const uint##sz##_t *)vmem;\
+ return *mem;\
+ }
+
+#define mem_get_sne_aligned_generic(end,sz) \
+ static signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\
+ const int##sz##_t *mem = (const int##sz##_t *)vmem;\
+ return *mem;\
+ }
+
+#define mem_get_se_aligned_generic(end,sz) \
+ static unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\
+ const uint##sz##_t *mem = (const uint##sz##_t *)vmem;\
+ unsigned MEM_VALUE_T val, raw = *mem;\
+ swap_endian_##sz(val,raw);\
+ return val;\
+ }
+
+#define mem_get_sse_aligned_generic(end,sz) \
+ static signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\
+ const int##sz##_t *mem = (const int##sz##_t *)vmem;\
+ unsigned MEM_VALUE_T val, raw = *mem;\
+ swap_endian_##sz##_se(val,raw);\
+ return val;\
+ }
+
+#define mem_put_ne_aligned_generic(end,sz) \
+ static void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\
+ uint##sz##_t *mem = (uint##sz##_t *)vmem;\
+ *mem = (uint##sz##_t)val;\
+ }
+
+#define mem_put_se_aligned_generic(end,sz) \
+ static void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\
+ uint##sz##_t *mem = (uint##sz##_t *)vmem, raw;\
+ swap_endian_##sz(raw,val);\
+ *mem = (uint##sz##_t)raw;\
+ }
+
+#include "vpx_config.h"
+#if CONFIG_BIG_ENDIAN
+#define mem_get_be_aligned_generic(sz) mem_get_ne_aligned_generic(be,sz)
+#define mem_get_sbe_aligned_generic(sz) mem_get_sne_aligned_generic(be,sz)
+#define mem_get_le_aligned_generic(sz) mem_get_se_aligned_generic(le,sz)
+#define mem_get_sle_aligned_generic(sz) mem_get_sse_aligned_generic(le,sz)
+#define mem_put_be_aligned_generic(sz) mem_put_ne_aligned_generic(be,sz)
+#define mem_put_le_aligned_generic(sz) mem_put_se_aligned_generic(le,sz)
+#else
+#define mem_get_be_aligned_generic(sz) mem_get_se_aligned_generic(be,sz)
+#define mem_get_sbe_aligned_generic(sz) mem_get_sse_aligned_generic(be,sz)
+#define mem_get_le_aligned_generic(sz) mem_get_ne_aligned_generic(le,sz)
+#define mem_get_sle_aligned_generic(sz) mem_get_sne_aligned_generic(le,sz)
+#define mem_put_be_aligned_generic(sz) mem_put_se_aligned_generic(be,sz)
+#define mem_put_le_aligned_generic(sz) mem_put_ne_aligned_generic(le,sz)
+#endif
+
+#undef mem_get_be16_aligned
+#define mem_get_be16_aligned mem_ops_wrap_symbol(mem_get_be16_aligned)
+mem_get_be_aligned_generic(16)
+
+#undef mem_get_be32_aligned
+#define mem_get_be32_aligned mem_ops_wrap_symbol(mem_get_be32_aligned)
+mem_get_be_aligned_generic(32)
+
+#undef mem_get_le16_aligned
+#define mem_get_le16_aligned mem_ops_wrap_symbol(mem_get_le16_aligned)
+mem_get_le_aligned_generic(16)
+
+#undef mem_get_le32_aligned
+#define mem_get_le32_aligned mem_ops_wrap_symbol(mem_get_le32_aligned)
+mem_get_le_aligned_generic(32)
+
+#undef mem_get_sbe16_aligned
+#define mem_get_sbe16_aligned mem_ops_wrap_symbol(mem_get_sbe16_aligned)
+mem_get_sbe_aligned_generic(16)
+
+#undef mem_get_sbe32_aligned
+#define mem_get_sbe32_aligned mem_ops_wrap_symbol(mem_get_sbe32_aligned)
+mem_get_sbe_aligned_generic(32)
+
+#undef mem_get_sle16_aligned
+#define mem_get_sle16_aligned mem_ops_wrap_symbol(mem_get_sle16_aligned)
+mem_get_sle_aligned_generic(16)
+
+#undef mem_get_sle32_aligned
+#define mem_get_sle32_aligned mem_ops_wrap_symbol(mem_get_sle32_aligned)
+mem_get_sle_aligned_generic(32)
+
+#undef mem_put_be16_aligned
+#define mem_put_be16_aligned mem_ops_wrap_symbol(mem_put_be16_aligned)
+mem_put_be_aligned_generic(16)
+
+#undef mem_put_be32_aligned
+#define mem_put_be32_aligned mem_ops_wrap_symbol(mem_put_be32_aligned)
+mem_put_be_aligned_generic(32)
+
+#undef mem_put_le16_aligned
+#define mem_put_le16_aligned mem_ops_wrap_symbol(mem_put_le16_aligned)
+mem_put_le_aligned_generic(16)
+
+#undef mem_put_le32_aligned
+#define mem_put_le32_aligned mem_ops_wrap_symbol(mem_put_le32_aligned)
+mem_put_le_aligned_generic(32)
+
+#undef mem_get_ne_aligned_generic
+#undef mem_get_se_aligned_generic
+#undef mem_get_sne_aligned_generic
+#undef mem_get_sse_aligned_generic
+#undef mem_put_ne_aligned_generic
+#undef mem_put_se_aligned_generic
+#undef swap_endian_16
+#undef swap_endian_32
+#undef swap_endian_16_se
+#undef swap_endian_32_se
diff --git a/libvpx/vpx_ports/vpx_ports.mk b/libvpx/vpx_ports/vpx_ports.mk
new file mode 100644
index 0000000..e6cb52f
--- /dev/null
+++ b/libvpx/vpx_ports/vpx_ports.mk
@@ -0,0 +1,26 @@
+##
+## Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+PORTS_SRCS-yes += vpx_ports.mk
+
+PORTS_SRCS-$(BUILD_LIBVPX) += asm_offsets.h
+PORTS_SRCS-$(BUILD_LIBVPX) += mem.h
+PORTS_SRCS-$(BUILD_LIBVPX) += vpx_timer.h
+
+ifeq ($(ARCH_X86)$(ARCH_X86_64),yes)
+PORTS_SRCS-$(BUILD_LIBVPX) += emms.asm
+PORTS_SRCS-$(BUILD_LIBVPX) += x86.h
+PORTS_SRCS-$(BUILD_LIBVPX) += x86_abi_support.asm
+PORTS_SRCS-$(BUILD_LIBVPX) += x86_cpuid.c
+endif
+
+PORTS_SRCS-$(ARCH_ARM) += arm_cpudetect.c
+PORTS_SRCS-$(ARCH_ARM) += arm.h
diff --git a/libvpx/vpx_ports/vpx_timer.h b/libvpx/vpx_ports/vpx_timer.h
new file mode 100644
index 0000000..d07e086
--- /dev/null
+++ b/libvpx/vpx_ports/vpx_timer.h
@@ -0,0 +1,120 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef VPX_TIMER_H
+#define VPX_TIMER_H
+#include "vpx/vpx_integer.h"
+
+#if CONFIG_OS_SUPPORT
+
+#if defined(_WIN32)
+/*
+ * Win32 specific includes
+ */
+#ifndef WIN32_LEAN_AND_MEAN
+#define WIN32_LEAN_AND_MEAN
+#endif
+#include <windows.h>
+#else
+/*
+ * POSIX specific includes
+ */
+#include <sys/time.h>
+
+/* timersub is not provided by msys at this time. */
+#ifndef timersub
+#define timersub(a, b, result) \
+ do { \
+ (result)->tv_sec = (a)->tv_sec - (b)->tv_sec; \
+ (result)->tv_usec = (a)->tv_usec - (b)->tv_usec; \
+ if ((result)->tv_usec < 0) { \
+ --(result)->tv_sec; \
+ (result)->tv_usec += 1000000; \
+ } \
+ } while (0)
+#endif
+#endif
+
+
+struct vpx_usec_timer
+{
+#if defined(_WIN32)
+ LARGE_INTEGER begin, end;
+#else
+ struct timeval begin, end;
+#endif
+};
+
+
+static void
+vpx_usec_timer_start(struct vpx_usec_timer *t)
+{
+#if defined(_WIN32)
+ QueryPerformanceCounter(&t->begin);
+#else
+ gettimeofday(&t->begin, NULL);
+#endif
+}
+
+
+static void
+vpx_usec_timer_mark(struct vpx_usec_timer *t)
+{
+#if defined(_WIN32)
+ QueryPerformanceCounter(&t->end);
+#else
+ gettimeofday(&t->end, NULL);
+#endif
+}
+
+
+static int64_t
+vpx_usec_timer_elapsed(struct vpx_usec_timer *t)
+{
+#if defined(_WIN32)
+ LARGE_INTEGER freq, diff;
+
+ diff.QuadPart = t->end.QuadPart - t->begin.QuadPart;
+
+ QueryPerformanceFrequency(&freq);
+ return diff.QuadPart * 1000000 / freq.QuadPart;
+#else
+ struct timeval diff;
+
+ timersub(&t->end, &t->begin, &diff);
+ return diff.tv_sec * 1000000 + diff.tv_usec;
+#endif
+}
+
+#else /* CONFIG_OS_SUPPORT = 0*/
+
+/* Empty timer functions if CONFIG_OS_SUPPORT = 0 */
+#ifndef timersub
+#define timersub(a, b, result)
+#endif
+
+struct vpx_usec_timer
+{
+ void *dummy;
+};
+
+static void
+vpx_usec_timer_start(struct vpx_usec_timer *t) { }
+
+static void
+vpx_usec_timer_mark(struct vpx_usec_timer *t) { }
+
+static long
+vpx_usec_timer_elapsed(struct vpx_usec_timer *t) { return 0; }
+
+#endif /* CONFIG_OS_SUPPORT */
+
+#endif
diff --git a/libvpx/vpx_ports/vpxtypes.h b/libvpx/vpx_ports/vpxtypes.h
new file mode 100644
index 0000000..f2fb089
--- /dev/null
+++ b/libvpx/vpx_ports/vpxtypes.h
@@ -0,0 +1,167 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __VPXTYPES_H__
+#define __VPXTYPES_H__
+
+#include "vpx_config.h"
+
+//#include <sys/types.h>
+#ifdef _MSC_VER
+# include <basetsd.h>
+typedef SSIZE_T ssize_t;
+#endif
+
+#if defined(HAVE_STDINT_H) && HAVE_STDINT_H
+/* C99 types are preferred to vpx integer types */
+# include <stdint.h>
+#endif
+
+/*!\defgroup basetypes Base Types
+ @{*/
+#if !defined(HAVE_STDINT_H) && !defined(INT_T_DEFINED)
+# ifdef STRICTTYPES
+typedef signed char int8_t;
+typedef signed short int16_t;
+typedef signed int int32_t;
+# else
+typedef char int8_t;
+typedef short int16_t;
+typedef int int32_t;
+# endif
+typedef unsigned char uint8_t;
+typedef unsigned short uint16_t;
+typedef unsigned int uint32_t;
+#endif
+
+typedef int8_t vpxs8;
+typedef uint8_t vpxu8;
+typedef int16_t vpxs16;
+typedef uint16_t vpxu16;
+typedef int32_t vpxs32;
+typedef uint32_t vpxu32;
+typedef int32_t vpxbool;
+
+enum {vpxfalse, vpxtrue};
+
+/*!\def OTC
+ \brief a macro suitable for declaring a constant #vpxtc*/
+/*!\def VPXTC
+ \brief printf format string suitable for printing an #vpxtc*/
+#ifdef UNICODE
+# ifdef NO_WCHAR
+# error "no non-wchar support added yet"
+# else
+# include <wchar.h>
+typedef wchar_t vpxtc;
+# define OTC(str) L ## str
+# define VPXTC "ls"
+# endif /*NO_WCHAR*/
+#else
+typedef char vpxtc;
+# define OTC(str) (vpxtc*)str
+# define VPXTC "s"
+#endif /*UNICODE*/
+/*@} end - base types*/
+
+/*!\addtogroup basetypes
+ @{*/
+/*!\def VPX64
+ \brief printf format string suitable for printing an #vpxs64*/
+#if defined(HAVE_STDINT_H)
+# define VPX64 PRId64
+typedef int64_t vpxs64;
+#elif defined(HASLONGLONG)
+# undef PRId64
+# define PRId64 "lld"
+# define VPX64 PRId64
+typedef long long vpxs64;
+#elif defined(WIN32) || defined(_WIN32_WCE)
+# undef PRId64
+# define PRId64 "I64d"
+# define VPX64 PRId64
+typedef __int64 vpxs64;
+typedef unsigned __int64 vpxu64;
+#elif defined(__uClinux__) && defined(CHIP_DM642)
+# include <lddk.h>
+# undef PRId64
+# define PRId64 "lld"
+# define VPX64 PRId64
+typedef long vpxs64;
+#else
+# error "64 bit integer type undefined for this platform!"
+#endif
+#if !defined(HAVE_STDINT_H) && !defined(INT_T_DEFINED)
+typedef vpxs64 int64_t;
+typedef vpxu64 uint64_t;
+#endif
+/*!@} end - base types*/
+
+/*!\ingroup basetypes
+ \brief Common return type*/
+typedef enum
+{
+ VPX_NOT_FOUND = -404,
+ VPX_BUFFER_EMPTY = -202,
+ VPX_BUFFER_FULL = -201,
+
+ VPX_CONNREFUSED = -102,
+ VPX_TIMEDOUT = -101,
+ VPX_WOULDBLOCK = -100,
+
+ VPX_NET_ERROR = -9,
+ VPX_INVALID_VERSION = -8,
+ VPX_INPROGRESS = -7,
+ VPX_NOT_SUPP = -6,
+ VPX_NO_MEM = -3,
+ VPX_INVALID_PARAMS = -2,
+ VPX_ERROR = -1,
+ VPX_OK = 0,
+ VPX_DONE = 1
+} vpxsc;
+
+#if defined(WIN32) || defined(_WIN32_WCE)
+# define DLLIMPORT __declspec(dllimport)
+# define DLLEXPORT __declspec(dllexport)
+# define DLLLOCAL
+#elif defined(LINUX)
+# define DLLIMPORT
+/*visibility attribute support is available in 3.4 and later.
+ see: http://gcc.gnu.org/wiki/Visibility for more info*/
+# if defined(__GNUC__) && ((__GNUC__<<16|(__GNUC_MINOR__&0xff)) >= (3<<16|4))
+# define GCC_HASCLASSVISIBILITY
+# endif /*defined(__GNUC__) && __GNUC_PREREQ(3,4)*/
+# ifdef GCC_HASCLASSVISIBILITY
+# define DLLEXPORT __attribute__ ((visibility("default")))
+# define DLLLOCAL __attribute__ ((visibility("hidden")))
+# else
+# define DLLEXPORT
+# define DLLLOCAL
+# endif /*GCC_HASCLASSVISIBILITY*/
+#endif /*platform ifdefs*/
+
+#endif /*__VPXTYPES_H__*/
+
+#undef VPXAPI
+/*!\def VPXAPI
+ \brief library calling convention/storage class attributes.
+
+ Specifies whether the function is imported through a dll
+ or is from a static library.*/
+#ifdef VPXDLL
+# ifdef VPXDLLEXPORT
+# define VPXAPI DLLEXPORT
+# else
+# define VPXAPI DLLIMPORT
+# endif /*VPXDLLEXPORT*/
+#else
+# define VPXAPI
+#endif /*VPXDLL*/
diff --git a/libvpx/vpx_ports/x86.h b/libvpx/vpx_ports/x86.h
new file mode 100644
index 0000000..9dd8c4b
--- /dev/null
+++ b/libvpx/vpx_ports/x86.h
@@ -0,0 +1,247 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef VPX_PORTS_X86_H
+#define VPX_PORTS_X86_H
+#include <stdlib.h>
+#include "vpx_config.h"
+
+typedef enum
+{
+ VPX_CPU_UNKNOWN = -1,
+ VPX_CPU_AMD,
+ VPX_CPU_AMD_OLD,
+ VPX_CPU_CENTAUR,
+ VPX_CPU_CYRIX,
+ VPX_CPU_INTEL,
+ VPX_CPU_NEXGEN,
+ VPX_CPU_NSC,
+ VPX_CPU_RISE,
+ VPX_CPU_SIS,
+ VPX_CPU_TRANSMETA,
+ VPX_CPU_TRANSMETA_OLD,
+ VPX_CPU_UMC,
+ VPX_CPU_VIA,
+
+ VPX_CPU_LAST
+} vpx_cpu_t;
+
+#if defined(__GNUC__) && __GNUC__
+#if ARCH_X86_64
+#define cpuid(func,ax,bx,cx,dx)\
+ __asm__ __volatile__ (\
+ "cpuid \n\t" \
+ : "=a" (ax), "=b" (bx), "=c" (cx), "=d" (dx) \
+ : "a" (func));
+#else
+#define cpuid(func,ax,bx,cx,dx)\
+ __asm__ __volatile__ (\
+ "mov %%ebx, %%edi \n\t" \
+ "cpuid \n\t" \
+ "xchg %%edi, %%ebx \n\t" \
+ : "=a" (ax), "=D" (bx), "=c" (cx), "=d" (dx) \
+ : "a" (func));
+#endif
+#elif defined(__SUNPRO_C) || defined(__SUNPRO_CC)
+#if ARCH_X86_64
+#define cpuid(func,ax,bx,cx,dx)\
+ asm volatile (\
+ "xchg %rsi, %rbx \n\t" \
+ "cpuid \n\t" \
+ "movl %ebx, %edi \n\t" \
+ "xchg %rsi, %rbx \n\t" \
+ : "=a" (ax), "=D" (bx), "=c" (cx), "=d" (dx) \
+ : "a" (func));
+#else
+#define cpuid(func,ax,bx,cx,dx)\
+ asm volatile (\
+ "pushl %ebx \n\t" \
+ "cpuid \n\t" \
+ "movl %ebx, %edi \n\t" \
+ "popl %ebx \n\t" \
+ : "=a" (ax), "=D" (bx), "=c" (cx), "=d" (dx) \
+ : "a" (func));
+#endif
+#else
+#if ARCH_X86_64
+void __cpuid(int CPUInfo[4], int info_type);
+#pragma intrinsic(__cpuid)
+#define cpuid(func,a,b,c,d) do{\
+ int regs[4];\
+ __cpuid(regs,func); a=regs[0]; b=regs[1]; c=regs[2]; d=regs[3];\
+ } while(0)
+#else
+#define cpuid(func,a,b,c,d)\
+ __asm mov eax, func\
+ __asm cpuid\
+ __asm mov a, eax\
+ __asm mov b, ebx\
+ __asm mov c, ecx\
+ __asm mov d, edx
+#endif
+#endif
+
+#define HAS_MMX 0x01
+#define HAS_SSE 0x02
+#define HAS_SSE2 0x04
+#define HAS_SSE3 0x08
+#define HAS_SSSE3 0x10
+#define HAS_SSE4_1 0x20
+#ifndef BIT
+#define BIT(n) (1<<n)
+#endif
+
+static int
+x86_simd_caps(void)
+{
+ unsigned int flags = 0;
+ unsigned int mask = ~0;
+ unsigned int reg_eax, reg_ebx, reg_ecx, reg_edx;
+ char *env;
+ (void)reg_ebx;
+
+ /* See if the CPU capabilities are being overridden by the environment */
+ env = getenv("VPX_SIMD_CAPS");
+
+ if (env && *env)
+ return (int)strtol(env, NULL, 0);
+
+ env = getenv("VPX_SIMD_CAPS_MASK");
+
+ if (env && *env)
+ mask = strtol(env, NULL, 0);
+
+ /* Ensure that the CPUID instruction supports extended features */
+ cpuid(0, reg_eax, reg_ebx, reg_ecx, reg_edx);
+
+ if (reg_eax < 1)
+ return 0;
+
+ /* Get the standard feature flags */
+ cpuid(1, reg_eax, reg_ebx, reg_ecx, reg_edx);
+
+ if (reg_edx & BIT(23)) flags |= HAS_MMX;
+
+ if (reg_edx & BIT(25)) flags |= HAS_SSE; /* aka xmm */
+
+ if (reg_edx & BIT(26)) flags |= HAS_SSE2; /* aka wmt */
+
+ if (reg_ecx & BIT(0)) flags |= HAS_SSE3;
+
+ if (reg_ecx & BIT(9)) flags |= HAS_SSSE3;
+
+ if (reg_ecx & BIT(19)) flags |= HAS_SSE4_1;
+
+ return flags & mask;
+}
+
+vpx_cpu_t vpx_x86_vendor(void);
+
+#if ARCH_X86_64 && defined(_MSC_VER)
+unsigned __int64 __rdtsc(void);
+#pragma intrinsic(__rdtsc)
+#endif
+static unsigned int
+x86_readtsc(void)
+{
+#if defined(__GNUC__) && __GNUC__
+ unsigned int tsc;
+ __asm__ __volatile__("rdtsc\n\t":"=a"(tsc):);
+ return tsc;
+#elif defined(__SUNPRO_C) || defined(__SUNPRO_CC)
+ unsigned int tsc;
+ asm volatile("rdtsc\n\t":"=a"(tsc):);
+ return tsc;
+#else
+#if ARCH_X86_64
+ return (unsigned int)__rdtsc();
+#else
+ __asm rdtsc;
+#endif
+#endif
+}
+
+
+#if defined(__GNUC__) && __GNUC__
+#define x86_pause_hint()\
+ __asm__ __volatile__ ("pause \n\t")
+#elif defined(__SUNPRO_C) || defined(__SUNPRO_CC)
+#define x86_pause_hint()\
+ asm volatile ("pause \n\t")
+#else
+#if ARCH_X86_64
+#define x86_pause_hint()\
+ _mm_pause();
+#else
+#define x86_pause_hint()\
+ __asm pause
+#endif
+#endif
+
+#if defined(__GNUC__) && __GNUC__
+static void
+x87_set_control_word(unsigned short mode)
+{
+ __asm__ __volatile__("fldcw %0" : : "m"(*&mode));
+}
+static unsigned short
+x87_get_control_word(void)
+{
+ unsigned short mode;
+ __asm__ __volatile__("fstcw %0\n\t":"=m"(*&mode):);
+ return mode;
+}
+#elif defined(__SUNPRO_C) || defined(__SUNPRO_CC)
+static void
+x87_set_control_word(unsigned short mode)
+{
+ asm volatile("fldcw %0" : : "m"(*&mode));
+}
+static unsigned short
+x87_get_control_word(void)
+{
+ unsigned short mode;
+ asm volatile("fstcw %0\n\t":"=m"(*&mode):);
+ return mode;
+}
+#elif ARCH_X86_64
+/* No fldcw intrinsics on Windows x64, punt to external asm */
+extern void vpx_winx64_fldcw(unsigned short mode);
+extern unsigned short vpx_winx64_fstcw(void);
+#define x87_set_control_word vpx_winx64_fldcw
+#define x87_get_control_word vpx_winx64_fstcw
+#else
+static void
+x87_set_control_word(unsigned short mode)
+{
+ __asm { fldcw mode }
+}
+static unsigned short
+x87_get_control_word(void)
+{
+ unsigned short mode;
+ __asm { fstcw mode }
+ return mode;
+}
+#endif
+
+static unsigned short
+x87_set_double_precision(void)
+{
+ unsigned short mode = x87_get_control_word();
+ x87_set_control_word((mode&~0x300) | 0x200);
+ return mode;
+}
+
+
+extern void vpx_reset_mmx_state(void);
+#endif
+
diff --git a/libvpx/vpx_ports/x86_abi_support.asm b/libvpx/vpx_ports/x86_abi_support.asm
new file mode 100644
index 0000000..0c9fe37
--- /dev/null
+++ b/libvpx/vpx_ports/x86_abi_support.asm
@@ -0,0 +1,378 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_config.asm"
+
+; 32/64 bit compatibility macros
+;
+; In general, we make the source use 64 bit syntax, then twiddle with it using
+; the preprocessor to get the 32 bit syntax on 32 bit platforms.
+;
+%ifidn __OUTPUT_FORMAT__,elf32
+%define ABI_IS_32BIT 1
+%elifidn __OUTPUT_FORMAT__,macho32
+%define ABI_IS_32BIT 1
+%elifidn __OUTPUT_FORMAT__,win32
+%define ABI_IS_32BIT 1
+%elifidn __OUTPUT_FORMAT__,aout
+%define ABI_IS_32BIT 1
+%else
+%define ABI_IS_32BIT 0
+%endif
+
+%if ABI_IS_32BIT
+%define rax eax
+%define rbx ebx
+%define rcx ecx
+%define rdx edx
+%define rsi esi
+%define rdi edi
+%define rsp esp
+%define rbp ebp
+%define movsxd mov
+%macro movq 2
+ %ifidn %1,eax
+ movd %1,%2
+ %elifidn %2,eax
+ movd %1,%2
+ %elifidn %1,ebx
+ movd %1,%2
+ %elifidn %2,ebx
+ movd %1,%2
+ %elifidn %1,ecx
+ movd %1,%2
+ %elifidn %2,ecx
+ movd %1,%2
+ %elifidn %1,edx
+ movd %1,%2
+ %elifidn %2,edx
+ movd %1,%2
+ %elifidn %1,esi
+ movd %1,%2
+ %elifidn %2,esi
+ movd %1,%2
+ %elifidn %1,edi
+ movd %1,%2
+ %elifidn %2,edi
+ movd %1,%2
+ %elifidn %1,esp
+ movd %1,%2
+ %elifidn %2,esp
+ movd %1,%2
+ %elifidn %1,ebp
+ movd %1,%2
+ %elifidn %2,ebp
+ movd %1,%2
+ %else
+ movq %1,%2
+ %endif
+%endmacro
+%endif
+
+
+; sym()
+; Return the proper symbol name for the target ABI.
+;
+; Certain ABIs, notably MS COFF and Darwin MACH-O, require that symbols
+; with C linkage be prefixed with an underscore.
+;
+%ifidn __OUTPUT_FORMAT__,elf32
+%define sym(x) x
+%elifidn __OUTPUT_FORMAT__,elf64
+%define sym(x) x
+%elifidn __OUTPUT_FORMAT__,elfx32
+%define sym(x) x
+%elifidn __OUTPUT_FORMAT__,x64
+%define sym(x) x
+%else
+%define sym(x) _ %+ x
+%endif
+
+; PRIVATE
+; Macro for the attribute to hide a global symbol for the target ABI.
+; This is only active if CHROMIUM is defined.
+;
+; Chromium doesn't like exported global symbols due to symbol clashing with
+; plugins among other things.
+;
+; Requires Chromium's patched copy of yasm:
+; http://src.chromium.org/viewvc/chrome?view=rev&revision=73761
+; http://www.tortall.net/projects/yasm/ticket/236
+;
+%ifdef CHROMIUM
+ %ifidn __OUTPUT_FORMAT__,elf32
+ %define PRIVATE :hidden
+ %elifidn __OUTPUT_FORMAT__,elf64
+ %define PRIVATE :hidden
+ %elifidn __OUTPUT_FORMAT__,elfx32
+ %define PRIVATE :hidden
+ %elifidn __OUTPUT_FORMAT__,x64
+ %define PRIVATE
+ %else
+ %define PRIVATE :private_extern
+ %endif
+%else
+ %define PRIVATE
+%endif
+
+; arg()
+; Return the address specification of the given argument
+;
+%if ABI_IS_32BIT
+ %define arg(x) [ebp+8+4*x]
+%else
+ ; 64 bit ABI passes arguments in registers. This is a workaround to get up
+ ; and running quickly. Relies on SHADOW_ARGS_TO_STACK
+ %ifidn __OUTPUT_FORMAT__,x64
+ %define arg(x) [rbp+16+8*x]
+ %else
+ %define arg(x) [rbp-8-8*x]
+ %endif
+%endif
+
+; REG_SZ_BYTES, REG_SZ_BITS
+; Size of a register
+%if ABI_IS_32BIT
+%define REG_SZ_BYTES 4
+%define REG_SZ_BITS 32
+%else
+%define REG_SZ_BYTES 8
+%define REG_SZ_BITS 64
+%endif
+
+
+; ALIGN_STACK <alignment> <register>
+; This macro aligns the stack to the given alignment (in bytes). The stack
+; is left such that the previous value of the stack pointer is the first
+; argument on the stack (ie, the inverse of this macro is 'pop rsp.')
+; This macro uses one temporary register, which is not preserved, and thus
+; must be specified as an argument.
+%macro ALIGN_STACK 2
+ mov %2, rsp
+ and rsp, -%1
+ lea rsp, [rsp - (%1 - REG_SZ_BYTES)]
+ push %2
+%endmacro
+
+
+;
+; The Microsoft assembler tries to impose a certain amount of type safety in
+; its register usage. YASM doesn't recognize these directives, so we just
+; %define them away to maintain as much compatibility as possible with the
+; original inline assembler we're porting from.
+;
+%idefine PTR
+%idefine XMMWORD
+%idefine MMWORD
+
+; PIC macros
+;
+%if ABI_IS_32BIT
+ %if CONFIG_PIC=1
+ %ifidn __OUTPUT_FORMAT__,elf32
+ %define GET_GOT_SAVE_ARG 1
+ %define WRT_PLT wrt ..plt
+ %macro GET_GOT 1
+ extern _GLOBAL_OFFSET_TABLE_
+ push %1
+ call %%get_got
+ %%sub_offset:
+ jmp %%exitGG
+ %%get_got:
+ mov %1, [esp]
+ add %1, _GLOBAL_OFFSET_TABLE_ + $$ - %%sub_offset wrt ..gotpc
+ ret
+ %%exitGG:
+ %undef GLOBAL
+ %define GLOBAL(x) x + %1 wrt ..gotoff
+ %undef RESTORE_GOT
+ %define RESTORE_GOT pop %1
+ %endmacro
+ %elifidn __OUTPUT_FORMAT__,macho32
+ %define GET_GOT_SAVE_ARG 1
+ %macro GET_GOT 1
+ push %1
+ call %%get_got
+ %%get_got:
+ pop %1
+ %undef GLOBAL
+ %define GLOBAL(x) x + %1 - %%get_got
+ %undef RESTORE_GOT
+ %define RESTORE_GOT pop %1
+ %endmacro
+ %endif
+ %endif
+
+ %ifdef CHROMIUM
+ %ifidn __OUTPUT_FORMAT__,macho32
+ %define HIDDEN_DATA(x) x:private_extern
+ %else
+ %define HIDDEN_DATA(x) x
+ %endif
+ %else
+ %define HIDDEN_DATA(x) x
+ %endif
+%else
+ %macro GET_GOT 1
+ %endmacro
+ %define GLOBAL(x) rel x
+ %ifidn __OUTPUT_FORMAT__,elf64
+ %define WRT_PLT wrt ..plt
+ %define HIDDEN_DATA(x) x:data hidden
+ %elifidn __OUTPUT_FORMAT__,elfx32
+ %define WRT_PLT wrt ..plt
+ %define HIDDEN_DATA(x) x:data hidden
+ %else
+ %define HIDDEN_DATA(x) x
+ %endif
+%endif
+%ifnmacro GET_GOT
+ %macro GET_GOT 1
+ %endmacro
+ %define GLOBAL(x) x
+%endif
+%ifndef RESTORE_GOT
+%define RESTORE_GOT
+%endif
+%ifndef WRT_PLT
+%define WRT_PLT
+%endif
+
+%if ABI_IS_32BIT
+ %macro SHADOW_ARGS_TO_STACK 1
+ %endm
+ %define UNSHADOW_ARGS
+%else
+%ifidn __OUTPUT_FORMAT__,x64
+ %macro SHADOW_ARGS_TO_STACK 1 ; argc
+ %if %1 > 0
+ mov arg(0),rcx
+ %endif
+ %if %1 > 1
+ mov arg(1),rdx
+ %endif
+ %if %1 > 2
+ mov arg(2),r8
+ %endif
+ %if %1 > 3
+ mov arg(3),r9
+ %endif
+ %endm
+%else
+ %macro SHADOW_ARGS_TO_STACK 1 ; argc
+ %if %1 > 0
+ push rdi
+ %endif
+ %if %1 > 1
+ push rsi
+ %endif
+ %if %1 > 2
+ push rdx
+ %endif
+ %if %1 > 3
+ push rcx
+ %endif
+ %if %1 > 4
+ push r8
+ %endif
+ %if %1 > 5
+ push r9
+ %endif
+ %if %1 > 6
+ %assign i %1-6
+ %assign off 16
+ %rep i
+ mov rax,[rbp+off]
+ push rax
+ %assign off off+8
+ %endrep
+ %endif
+ %endm
+%endif
+ %define UNSHADOW_ARGS mov rsp, rbp
+%endif
+
+; Win64 ABI requires that XMM6:XMM15 are callee saved
+; SAVE_XMM n, [u]
+; store registers 6-n on the stack
+; if u is specified, use unaligned movs.
+; Win64 ABI requires 16 byte stack alignment, but then pushes an 8 byte return
+; value. Typically we follow this up with 'push rbp' - re-aligning the stack -
+; but in some cases this is not done and unaligned movs must be used.
+%ifidn __OUTPUT_FORMAT__,x64
+%macro SAVE_XMM 1-2 a
+ %if %1 < 6
+ %error Only xmm registers 6-15 must be preserved
+ %else
+ %assign last_xmm %1
+ %define movxmm movdq %+ %2
+ %assign xmm_stack_space ((last_xmm - 5) * 16)
+ sub rsp, xmm_stack_space
+ %assign i 6
+ %rep (last_xmm - 5)
+ movxmm [rsp + ((i - 6) * 16)], xmm %+ i
+ %assign i i+1
+ %endrep
+ %endif
+%endmacro
+%macro RESTORE_XMM 0
+ %ifndef last_xmm
+ %error RESTORE_XMM must be paired with SAVE_XMM n
+ %else
+ %assign i last_xmm
+ %rep (last_xmm - 5)
+ movxmm xmm %+ i, [rsp +((i - 6) * 16)]
+ %assign i i-1
+ %endrep
+ add rsp, xmm_stack_space
+ ; there are a couple functions which return from multiple places.
+ ; otherwise, we could uncomment these:
+ ; %undef last_xmm
+ ; %undef xmm_stack_space
+ ; %undef movxmm
+ %endif
+%endmacro
+%else
+%macro SAVE_XMM 1-2
+%endmacro
+%macro RESTORE_XMM 0
+%endmacro
+%endif
+
+; Name of the rodata section
+;
+; .rodata seems to be an elf-ism, as it doesn't work on OSX.
+;
+%ifidn __OUTPUT_FORMAT__,macho64
+%define SECTION_RODATA section .text
+%elifidn __OUTPUT_FORMAT__,macho32
+%macro SECTION_RODATA 0
+section .text
+%endmacro
+%elifidn __OUTPUT_FORMAT__,aout
+%define SECTION_RODATA section .data
+%else
+%define SECTION_RODATA section .rodata
+%endif
+
+
+; Tell GNU ld that we don't require an executable stack.
+%ifidn __OUTPUT_FORMAT__,elf32
+section .note.GNU-stack noalloc noexec nowrite progbits
+section .text
+%elifidn __OUTPUT_FORMAT__,elf64
+section .note.GNU-stack noalloc noexec nowrite progbits
+section .text
+%elifidn __OUTPUT_FORMAT__,elfx32
+section .note.GNU-stack noalloc noexec nowrite progbits
+section .text
+%endif
+
diff --git a/vpx_ports/x86_cpuid.c b/libvpx/vpx_ports/x86_cpuid.c
similarity index 100%
rename from vpx_ports/x86_cpuid.c
rename to libvpx/vpx_ports/x86_cpuid.c
diff --git a/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_copy_y_neon.asm b/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_copy_y_neon.asm
new file mode 100644
index 0000000..9189641
--- /dev/null
+++ b/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_copy_y_neon.asm
@@ -0,0 +1,122 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_yv12_copy_y_neon|
+
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ INCLUDE asm_com_offsets.asm
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+;void vpxyv12_copy_y_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc)
+|vp8_yv12_copy_y_neon| PROC
+ push {r4 - r11, lr}
+ vpush {d8-d15}
+
+ ldr r4, [r0, #yv12_buffer_config_y_height]
+ ldr r5, [r0, #yv12_buffer_config_y_width]
+ ldr r6, [r0, #yv12_buffer_config_y_stride]
+ ldr r7, [r1, #yv12_buffer_config_y_stride]
+ ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
+ ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
+
+ ; copy two rows at one time
+ mov lr, r4, lsr #1
+
+cp_src_to_dst_height_loop1
+ mov r8, r2
+ mov r9, r3
+ add r10, r2, r6
+ add r11, r3, r7
+ movs r12, r5, lsr #7
+ ble extra_copy_needed ; y_width < 128
+
+cp_src_to_dst_width_loop1
+ vld1.8 {q0, q1}, [r8]!
+ vld1.8 {q8, q9}, [r10]!
+ vld1.8 {q2, q3}, [r8]!
+ vld1.8 {q10, q11}, [r10]!
+ vld1.8 {q4, q5}, [r8]!
+ vld1.8 {q12, q13}, [r10]!
+ vld1.8 {q6, q7}, [r8]!
+ vld1.8 {q14, q15}, [r10]!
+
+ subs r12, r12, #1
+
+ vst1.8 {q0, q1}, [r9]!
+ vst1.8 {q8, q9}, [r11]!
+ vst1.8 {q2, q3}, [r9]!
+ vst1.8 {q10, q11}, [r11]!
+ vst1.8 {q4, q5}, [r9]!
+ vst1.8 {q12, q13}, [r11]!
+ vst1.8 {q6, q7}, [r9]!
+ vst1.8 {q14, q15}, [r11]!
+
+ bne cp_src_to_dst_width_loop1
+
+ subs lr, lr, #1
+ add r2, r2, r6, lsl #1
+ add r3, r3, r7, lsl #1
+
+ bne cp_src_to_dst_height_loop1
+
+extra_copy_needed
+ ands r10, r5, #0x7f ;check to see if extra copy is needed
+ sub r11, r5, r10
+ ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
+ ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
+ bne extra_cp_src_to_dst_width1
+end_of_cp_src_to_dst1
+
+ vpop {d8 - d15}
+ pop {r4-r11, pc}
+
+;=============================
+extra_cp_src_to_dst_width1
+ add r2, r2, r11
+ add r3, r3, r11
+ add r0, r8, r6
+ add r11, r9, r7
+
+ mov lr, r4, lsr #1
+extra_cp_src_to_dst_height_loop1
+ mov r8, r2
+ mov r9, r3
+ add r0, r8, r6
+ add r11, r9, r7
+
+ mov r12, r10
+
+extra_cp_src_to_dst_width_loop1
+ vld1.8 {q0}, [r8]!
+ vld1.8 {q1}, [r0]!
+
+ subs r12, r12, #16
+
+ vst1.8 {q0}, [r9]!
+ vst1.8 {q1}, [r11]!
+ bne extra_cp_src_to_dst_width_loop1
+
+ subs lr, lr, #1
+
+ add r2, r2, r6, lsl #1
+ add r3, r3, r7, lsl #1
+
+ bne extra_cp_src_to_dst_height_loop1
+
+ b end_of_cp_src_to_dst1
+
+ ENDP
+
+ END
diff --git a/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm b/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm
new file mode 100644
index 0000000..e55d076
--- /dev/null
+++ b/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm
@@ -0,0 +1,233 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_yv12_copy_frame_func_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ INCLUDE asm_com_offsets.asm
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+
+;void vp8_yv12_copy_frame_func_neon(YV12_BUFFER_CONFIG *src_ybc,
+; YV12_BUFFER_CONFIG *dst_ybc);
+
+|vp8_yv12_copy_frame_func_neon| PROC
+ push {r4 - r11, lr}
+ vpush {d8 - d15}
+
+ sub sp, sp, #16
+
+ ;Copy Y plane
+ ldr r8, [r0, #yv12_buffer_config_u_buffer] ;srcptr1
+ ldr r9, [r1, #yv12_buffer_config_u_buffer] ;srcptr1
+ ldr r10, [r0, #yv12_buffer_config_v_buffer] ;srcptr1
+ ldr r11, [r1, #yv12_buffer_config_v_buffer] ;srcptr1
+
+ ldr r4, [r0, #yv12_buffer_config_y_height]
+ ldr r5, [r0, #yv12_buffer_config_y_width]
+ ldr r6, [r0, #yv12_buffer_config_y_stride]
+ ldr r7, [r1, #yv12_buffer_config_y_stride]
+ ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
+ ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
+
+ str r8, [sp]
+ str r9, [sp, #4]
+ str r10, [sp, #8]
+ str r11, [sp, #12]
+
+ ; copy two rows at one time
+ mov lr, r4, lsr #1
+
+cp_src_to_dst_height_loop
+ mov r8, r2
+ mov r9, r3
+ add r10, r2, r6
+ add r11, r3, r7
+ movs r12, r5, lsr #7
+ ble extra_cp_needed ; y_width < 128
+
+cp_src_to_dst_width_loop
+ vld1.8 {q0, q1}, [r8]!
+ vld1.8 {q8, q9}, [r10]!
+ vld1.8 {q2, q3}, [r8]!
+ vld1.8 {q10, q11}, [r10]!
+ vld1.8 {q4, q5}, [r8]!
+ vld1.8 {q12, q13}, [r10]!
+ vld1.8 {q6, q7}, [r8]!
+ vld1.8 {q14, q15}, [r10]!
+
+ subs r12, r12, #1
+
+ vst1.8 {q0, q1}, [r9]!
+ vst1.8 {q8, q9}, [r11]!
+ vst1.8 {q2, q3}, [r9]!
+ vst1.8 {q10, q11}, [r11]!
+ vst1.8 {q4, q5}, [r9]!
+ vst1.8 {q12, q13}, [r11]!
+ vst1.8 {q6, q7}, [r9]!
+ vst1.8 {q14, q15}, [r11]!
+
+ bne cp_src_to_dst_width_loop
+
+ subs lr, lr, #1
+ add r2, r2, r6, lsl #1
+ add r3, r3, r7, lsl #1
+
+ bne cp_src_to_dst_height_loop
+
+extra_cp_needed
+ ands r10, r5, #0x7f ;check to see if extra copy is needed
+ sub r11, r5, r10
+ ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
+ ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
+ bne extra_cp_src_to_dst_width
+end_of_cp_src_to_dst
+
+;Copy U & V planes
+ ldr r2, [sp] ;srcptr1
+ ldr r3, [sp, #4] ;dstptr1
+ mov r4, r4, lsr #1 ;src uv_height
+ mov r5, r5, lsr #1 ;src uv_width
+ mov r6, r6, lsr #1 ;src uv_stride
+ mov r7, r7, lsr #1 ;dst uv_stride
+
+ mov r1, #2
+
+cp_uv_loop
+
+ ;copy two rows at one time
+ mov lr, r4, lsr #1
+
+cp_src_to_dst_height_uv_loop
+ mov r8, r2
+ mov r9, r3
+ add r10, r2, r6
+ add r11, r3, r7
+ movs r12, r5, lsr #6
+ ble extra_uv_cp_needed
+
+cp_src_to_dst_width_uv_loop
+ vld1.8 {q0, q1}, [r8]!
+ vld1.8 {q8, q9}, [r10]!
+ vld1.8 {q2, q3}, [r8]!
+ vld1.8 {q10, q11}, [r10]!
+
+ subs r12, r12, #1
+
+ vst1.8 {q0, q1}, [r9]!
+ vst1.8 {q8, q9}, [r11]!
+ vst1.8 {q2, q3}, [r9]!
+ vst1.8 {q10, q11}, [r11]!
+
+ bne cp_src_to_dst_width_uv_loop
+
+ subs lr, lr, #1
+ add r2, r2, r6, lsl #1
+ add r3, r3, r7, lsl #1
+
+ bne cp_src_to_dst_height_uv_loop
+
+extra_uv_cp_needed
+ ands r10, r5, #0x3f ;check to see if extra copy is needed
+ sub r11, r5, r10
+ ldr r2, [sp] ;srcptr1
+ ldr r3, [sp, #4] ;dstptr1
+ bne extra_cp_src_to_dst_uv_width
+end_of_cp_src_to_dst_uv
+
+ subs r1, r1, #1
+
+ addne sp, sp, #8
+
+ ldrne r2, [sp] ;srcptr1
+ ldrne r3, [sp, #4] ;dstptr1
+
+ bne cp_uv_loop
+
+ add sp, sp, #8
+
+ vpop {d8 - d15}
+ pop {r4 - r11, pc}
+
+;=============================
+extra_cp_src_to_dst_width
+ add r2, r2, r11
+ add r3, r3, r11
+ add r0, r8, r6
+ add r11, r9, r7
+
+ mov lr, r4, lsr #1
+extra_cp_src_to_dst_height_loop
+ mov r8, r2
+ mov r9, r3
+ add r0, r8, r6
+ add r11, r9, r7
+
+ mov r12, r10
+
+extra_cp_src_to_dst_width_loop
+ vld1.8 {q0}, [r8]!
+ vld1.8 {q1}, [r0]!
+
+ subs r12, r12, #16
+
+ vst1.8 {q0}, [r9]!
+ vst1.8 {q1}, [r11]!
+ bne extra_cp_src_to_dst_width_loop
+
+ subs lr, lr, #1
+
+ add r2, r2, r6, lsl #1
+ add r3, r3, r7, lsl #1
+
+ bne extra_cp_src_to_dst_height_loop
+
+ b end_of_cp_src_to_dst
+
+;=================================
+extra_cp_src_to_dst_uv_width
+ add r2, r2, r11
+ add r3, r3, r11
+ add r0, r8, r6
+ add r11, r9, r7
+
+ mov lr, r4, lsr #1
+extra_cp_src_to_dst_height_uv_loop
+ mov r8, r2
+ mov r9, r3
+ add r0, r8, r6
+ add r11, r9, r7
+
+ mov r12, r10
+
+extra_cp_src_to_dst_width_uv_loop
+ vld1.8 {d0}, [r8]!
+ vld1.8 {d1}, [r0]!
+
+ subs r12, r12, #8
+
+ vst1.8 {d0}, [r9]!
+ vst1.8 {d1}, [r11]!
+ bne extra_cp_src_to_dst_width_uv_loop
+
+ subs lr, lr, #1
+
+ add r2, r2, r6, lsl #1
+ add r3, r3, r7, lsl #1
+
+ bne extra_cp_src_to_dst_height_uv_loop
+
+ b end_of_cp_src_to_dst_uv
+
+ ENDP
+ END
diff --git a/vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm b/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm
similarity index 100%
rename from vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm
rename to libvpx/vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm
diff --git a/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm b/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm
new file mode 100644
index 0000000..ebc4242
--- /dev/null
+++ b/libvpx/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm
@@ -0,0 +1,308 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license
+; that can be found in the LICENSE file in the root of the source
+; tree. An additional intellectual property rights grant can be found
+; in the file PATENTS. All contributing project authors may
+; be found in the AUTHORS file in the root of the source tree.
+;
+
+
+ EXPORT |vp8_yv12_extend_frame_borders_neon|
+ ARM
+ REQUIRE8
+ PRESERVE8
+
+ INCLUDE asm_com_offsets.asm
+
+ AREA ||.text||, CODE, READONLY, ALIGN=2
+;void vp8_yv12_extend_frame_borders_neon (YV12_BUFFER_CONFIG *ybf);
+; we depend on VP8BORDERINPIXELS being 32
+
+|vp8_yv12_extend_frame_borders_neon| PROC
+ push {r4 - r10, lr}
+ vpush {d8 - d15}
+
+ ; Border = 32
+ ldr r3, [r0, #yv12_buffer_config_y_width] ; plane_width
+ ldr r1, [r0, #yv12_buffer_config_y_buffer] ; src_ptr1
+ ldr r4, [r0, #yv12_buffer_config_y_height] ; plane_height
+ ldr lr, [r0, #yv12_buffer_config_y_stride] ; plane_stride
+
+; Border copy for Y plane
+; copy the left and right most columns out
+ add r6, r1, r3 ; dest_ptr2 = src_ptr2 + 1 (src_ptr1 + plane_width)
+ sub r2, r6, #1 ; src_ptr2 = src_ptr1 + plane_width - 1
+ sub r5, r1, #32 ; dest_ptr1 = src_ptr1 - Border
+
+ mov r12, r4, lsr #2 ; plane_height / 4
+
+copy_left_right_y
+ vld1.8 {d0[], d1[]}, [r1], lr
+ vld1.8 {d4[], d5[]}, [r2], lr
+ vld1.8 {d8[], d9[]}, [r1], lr
+ vld1.8 {d12[], d13[]}, [r2], lr
+ vld1.8 {d16[], d17[]}, [r1], lr
+ vld1.8 {d20[], d21[]}, [r2], lr
+ vld1.8 {d24[], d25[]}, [r1], lr
+ vld1.8 {d28[], d29[]}, [r2], lr
+
+ vmov q1, q0
+ vmov q3, q2
+ vmov q5, q4
+ vmov q7, q6
+ vmov q9, q8
+ vmov q11, q10
+ vmov q13, q12
+ vmov q15, q14
+
+ subs r12, r12, #1
+
+ vst1.8 {q0, q1}, [r5], lr
+ vst1.8 {q2, q3}, [r6], lr
+ vst1.8 {q4, q5}, [r5], lr
+ vst1.8 {q6, q7}, [r6], lr
+ vst1.8 {q8, q9}, [r5], lr
+ vst1.8 {q10, q11}, [r6], lr
+ vst1.8 {q12, q13}, [r5], lr
+ vst1.8 {q14, q15}, [r6], lr
+
+ bne copy_left_right_y
+
+;Now copy the top and bottom source lines into each line of the respective borders
+ ldr r1, [r0, #yv12_buffer_config_y_buffer] ; y_buffer
+ mul r8, r4, lr ; plane_height * plane_stride
+
+ ; copy width is plane_stride
+ movs r12, lr, lsr #7 ; plane_stride / 128
+
+ sub r1, r1, #32 ; src_ptr1 = y_buffer - Border
+ add r6, r1, r8 ; dest_ptr2 = src_ptr2 - plane_stride (src_ptr1 + (plane_height * plane_stride))
+ sub r2, r6, lr ; src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride
+ sub r5, r1, lr, asl #5 ; dest_ptr1 = src_ptr1 - (Border * plane_stride)
+ ble extra_y_copy_needed ; plane stride < 128
+
+copy_top_bottom_y
+ vld1.8 {q0, q1}, [r1]!
+ vld1.8 {q8, q9}, [r2]!
+ vld1.8 {q2, q3}, [r1]!
+ vld1.8 {q10, q11}, [r2]!
+ vld1.8 {q4, q5}, [r1]!
+ vld1.8 {q12, q13}, [r2]!
+ vld1.8 {q6, q7}, [r1]!
+ vld1.8 {q14, q15}, [r2]!
+
+ mov r7, #32 ; Border
+
+top_bottom_32
+ subs r7, r7, #1
+
+ vst1.8 {q0, q1}, [r5]!
+ vst1.8 {q8, q9}, [r6]!
+ vst1.8 {q2, q3}, [r5]!
+ vst1.8 {q10, q11}, [r6]!
+ vst1.8 {q4, q5}, [r5]!
+ vst1.8 {q12, q13}, [r6]!
+ vst1.8 {q6, q7}, [r5]!
+ vst1.8 {q14, q15}, [r6]!
+
+ add r5, r5, lr ; dest_ptr1 += plane_stride
+ sub r5, r5, #128 ; dest_ptr1 -= 128
+ add r6, r6, lr ; dest_ptr2 += plane_stride
+ sub r6, r6, #128 ; dest_ptr2 -= 128
+
+ bne top_bottom_32
+
+ sub r5, r1, lr, asl #5 ; src_ptr1 - (Border* plane_stride)
+ add r6, r2, lr ; src_ptr2 + plane_stride
+
+ subs r12, r12, #1
+ bne copy_top_bottom_y
+
+extra_y_copy_needed
+ mov r7, lr, lsr #4 ; check to see if extra copy is needed
+ ands r7, r7, #0x7
+ bne extra_top_bottom_y
+end_of_border_copy_y
+
+;Border copy for U, V planes
+; Border = 16
+ ldr r7, [r0, #yv12_buffer_config_u_buffer] ; src_ptr1
+ ldr lr, [r0, #yv12_buffer_config_uv_stride] ; plane_stride
+ ldr r3, [r0, #yv12_buffer_config_uv_width] ; plane_width
+ ldr r4, [r0, #yv12_buffer_config_uv_height] ; plane_height
+
+ mov r10, #2
+
+;copy the left and right most columns out
+border_copy_uv
+ mov r1, r7 ; src_ptr1 needs to be saved for second half of loop
+ sub r5, r1, #16 ; dest_ptr1 = src_ptr1 - Border
+ add r6, r1, r3 ; dest_ptr2 = src_ptr2 + 1 (src_ptr1 + plane_width)
+ sub r2, r6, #1 ; src_ptr2 = src_ptr1 + plane_width - 1
+
+ mov r12, r4, lsr #3 ; plane_height / 8
+
+copy_left_right_uv
+ vld1.8 {d0[], d1[]}, [r1], lr
+ vld1.8 {d2[], d3[]}, [r2], lr
+ vld1.8 {d4[], d5[]}, [r1], lr
+ vld1.8 {d6[], d7[]}, [r2], lr
+ vld1.8 {d8[], d9[]}, [r1], lr
+ vld1.8 {d10[], d11[]}, [r2], lr
+ vld1.8 {d12[], d13[]}, [r1], lr
+ vld1.8 {d14[], d15[]}, [r2], lr
+ vld1.8 {d16[], d17[]}, [r1], lr
+ vld1.8 {d18[], d19[]}, [r2], lr
+ vld1.8 {d20[], d21[]}, [r1], lr
+ vld1.8 {d22[], d23[]}, [r2], lr
+ vld1.8 {d24[], d25[]}, [r1], lr
+ vld1.8 {d26[], d27[]}, [r2], lr
+ vld1.8 {d28[], d29[]}, [r1], lr
+ vld1.8 {d30[], d31[]}, [r2], lr
+
+ subs r12, r12, #1
+
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q1}, [r6], lr
+ vst1.8 {q2}, [r5], lr
+ vst1.8 {q3}, [r6], lr
+ vst1.8 {q4}, [r5], lr
+ vst1.8 {q5}, [r6], lr
+ vst1.8 {q6}, [r5], lr
+ vst1.8 {q7}, [r6], lr
+ vst1.8 {q8}, [r5], lr
+ vst1.8 {q9}, [r6], lr
+ vst1.8 {q10}, [r5], lr
+ vst1.8 {q11}, [r6], lr
+ vst1.8 {q12}, [r5], lr
+ vst1.8 {q13}, [r6], lr
+ vst1.8 {q14}, [r5], lr
+ vst1.8 {q15}, [r6], lr
+
+ bne copy_left_right_uv
+
+;Now copy the top and bottom source lines into each line of the respective borders
+ mov r1, r7
+ mul r8, r4, lr ; plane_height * plane_stride
+ movs r12, lr, lsr #6 ; plane_stride / 64
+
+ sub r1, r1, #16 ; src_ptr1 = u_buffer - Border
+ add r6, r1, r8 ; dest_ptr2 = src_ptr2 + plane_stride (src_ptr1 + (plane_height * plane_stride)
+ sub r2, r6, lr ; src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride
+ sub r5, r1, lr, asl #4 ; dest_ptr1 = src_ptr1 - (Border * plane_stride)
+ ble extra_uv_copy_needed ; plane_stride < 64
+
+copy_top_bottom_uv
+ vld1.8 {q0, q1}, [r1]!
+ vld1.8 {q8, q9}, [r2]!
+ vld1.8 {q2, q3}, [r1]!
+ vld1.8 {q10, q11}, [r2]!
+
+ mov r7, #16 ; Border
+
+top_bottom_16
+ subs r7, r7, #1
+
+ vst1.8 {q0, q1}, [r5]!
+ vst1.8 {q8, q9}, [r6]!
+ vst1.8 {q2, q3}, [r5]!
+ vst1.8 {q10, q11}, [r6]!
+
+ add r5, r5, lr ; dest_ptr1 += plane_stride
+ sub r5, r5, #64
+ add r6, r6, lr ; dest_ptr2 += plane_stride
+ sub r6, r6, #64
+
+ bne top_bottom_16
+
+ sub r5, r1, lr, asl #4 ; dest_ptr1 = src_ptr1 - (Border * plane_stride)
+ add r6, r2, lr ; dest_ptr2 = src_ptr2 + plane_stride
+
+ subs r12, r12, #1
+ bne copy_top_bottom_uv
+extra_uv_copy_needed
+ mov r7, lr, lsr #3 ; check to see if extra copy is needed
+ ands r7, r7, #0x7
+ bne extra_top_bottom_uv
+
+end_of_border_copy_uv
+ subs r10, r10, #1
+ ldrne r7, [r0, #yv12_buffer_config_v_buffer] ; src_ptr1
+ bne border_copy_uv
+
+ vpop {d8 - d15}
+ pop {r4 - r10, pc}
+
+;;;;;;;;;;;;;;;;;;;;;;
+extra_top_bottom_y
+ vld1.8 {q0}, [r1]!
+ vld1.8 {q2}, [r2]!
+
+ mov r9, #4 ; 32 >> 3
+
+extra_top_bottom_32
+ subs r9, r9, #1
+
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q2}, [r6], lr
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q2}, [r6], lr
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q2}, [r6], lr
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q2}, [r6], lr
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q2}, [r6], lr
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q2}, [r6], lr
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q2}, [r6], lr
+ vst1.8 {q0}, [r5], lr
+ vst1.8 {q2}, [r6], lr
+ bne extra_top_bottom_32
+
+ sub r5, r1, lr, asl #5 ; src_ptr1 - (Border * plane_stride)
+ add r6, r2, lr ; src_ptr2 + plane_stride
+ subs r7, r7, #1
+ bne extra_top_bottom_y
+
+ b end_of_border_copy_y
+
+extra_top_bottom_uv
+ vld1.8 {d0}, [r1]!
+ vld1.8 {d8}, [r2]!
+
+ mov r9, #2 ; 16 >> 3
+
+extra_top_bottom_16
+ subs r9, r9, #1
+
+ vst1.8 {d0}, [r5], lr
+ vst1.8 {d8}, [r6], lr
+ vst1.8 {d0}, [r5], lr
+ vst1.8 {d8}, [r6], lr
+ vst1.8 {d0}, [r5], lr
+ vst1.8 {d8}, [r6], lr
+ vst1.8 {d0}, [r5], lr
+ vst1.8 {d8}, [r6], lr
+ vst1.8 {d0}, [r5], lr
+ vst1.8 {d8}, [r6], lr
+ vst1.8 {d0}, [r5], lr
+ vst1.8 {d8}, [r6], lr
+ vst1.8 {d0}, [r5], lr
+ vst1.8 {d8}, [r6], lr
+ vst1.8 {d0}, [r5], lr
+ vst1.8 {d8}, [r6], lr
+ bne extra_top_bottom_16
+
+ sub r5, r1, lr, asl #4 ; src_ptr1 - (Border * plane_stride)
+ add r6, r2, lr ; src_ptr2 + plane_stride
+ subs r7, r7, #1
+ bne extra_top_bottom_uv
+
+ b end_of_border_copy_uv
+
+ ENDP
+ END
diff --git a/libvpx/vpx_scale/arm/neon/yv12extend_arm.c b/libvpx/vpx_scale/arm/neon/yv12extend_arm.c
new file mode 100644
index 0000000..eabd495
--- /dev/null
+++ b/libvpx/vpx_scale/arm/neon/yv12extend_arm.c
@@ -0,0 +1,21 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "./vpx_rtcd.h"
+
+extern void vp8_yv12_copy_frame_func_neon(struct yv12_buffer_config *src_ybc,
+ struct yv12_buffer_config *dst_ybc);
+
+void vp8_yv12_copy_frame_neon(struct yv12_buffer_config *src_ybc,
+ struct yv12_buffer_config *dst_ybc) {
+ vp8_yv12_copy_frame_func_neon(src_ybc, dst_ybc);
+
+ vp8_yv12_extend_frame_borders_neon(dst_ybc);
+}
diff --git a/libvpx/vpx_scale/generic/bicubic_scaler.c b/libvpx/vpx_scale/generic/bicubic_scaler.c
new file mode 100644
index 0000000..c116740
--- /dev/null
+++ b/libvpx/vpx_scale/generic/bicubic_scaler.c
@@ -0,0 +1,569 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include <float.h>
+#include <math.h>
+#include <stdio.h>
+#include "vpx_mem/vpx_mem.h"
+#include "vpxscale_arbitrary.h"
+
+#define FIXED_POINT
+
+#define MAX_IN_WIDTH 800
+#define MAX_IN_HEIGHT 600
+#define MAX_OUT_WIDTH 800
+#define MAX_OUT_HEIGHT 600
+#define MAX_OUT_DIMENSION ((MAX_OUT_WIDTH > MAX_OUT_HEIGHT) ? \
+ MAX_OUT_WIDTH : MAX_OUT_HEIGHT)
+
+BICUBIC_SCALER_STRUCT g_b_scaler;
+static int g_first_time = 1;
+
+#pragma DATA_SECTION(g_hbuf, "VP6_HEAP")
+#pragma DATA_ALIGN (g_hbuf, 32);
+unsigned char g_hbuf[MAX_OUT_DIMENSION];
+
+#pragma DATA_SECTION(g_hbuf_uv, "VP6_HEAP")
+#pragma DATA_ALIGN (g_hbuf_uv, 32);
+unsigned char g_hbuf_uv[MAX_OUT_DIMENSION];
+
+
+#ifdef FIXED_POINT
+static int a_i = 0.6 * 65536;
+#else
+static float a = -0.6;
+#endif
+
+#ifdef FIXED_POINT
+// 3 2
+// C0 = a*t - a*t
+//
+static short c0_fixed(unsigned int t) {
+ // put t in Q16 notation
+ unsigned short v1, v2;
+
+ // Q16
+ v1 = (a_i * t) >> 16;
+ v1 = (v1 * t) >> 16;
+
+ // Q16
+ v2 = (a_i * t) >> 16;
+ v2 = (v2 * t) >> 16;
+ v2 = (v2 * t) >> 16;
+
+ // Q12
+ return -((v1 - v2) >> 4);
+}
+
+// 2 3
+// C1 = a*t + (3-2*a)*t - (2-a)*t
+//
+static short c1_fixed(unsigned int t) {
+ unsigned short v1, v2, v3;
+ unsigned short two, three;
+
+ // Q16
+ v1 = (a_i * t) >> 16;
+
+ // Q13
+ two = 2 << 13;
+ v2 = two - (a_i >> 3);
+ v2 = (v2 * t) >> 16;
+ v2 = (v2 * t) >> 16;
+ v2 = (v2 * t) >> 16;
+
+ // Q13
+ three = 3 << 13;
+ v3 = three - (2 * (a_i >> 3));
+ v3 = (v3 * t) >> 16;
+ v3 = (v3 * t) >> 16;
+
+ // Q12
+ return (((v1 >> 3) - v2 + v3) >> 1);
+
+}
+
+// 2 3
+// C2 = 1 - (3-a)*t + (2-a)*t
+//
+static short c2_fixed(unsigned int t) {
+ unsigned short v1, v2, v3;
+ unsigned short two, three;
+
+ // Q13
+ v1 = 1 << 13;
+
+ // Q13
+ three = 3 << 13;
+ v2 = three - (a_i >> 3);
+ v2 = (v2 * t) >> 16;
+ v2 = (v2 * t) >> 16;
+
+ // Q13
+ two = 2 << 13;
+ v3 = two - (a_i >> 3);
+ v3 = (v3 * t) >> 16;
+ v3 = (v3 * t) >> 16;
+ v3 = (v3 * t) >> 16;
+
+ // Q12
+ return (v1 - v2 + v3) >> 1;
+}
+
+// 2 3
+// C3 = a*t - 2*a*t + a*t
+//
+static short c3_fixed(unsigned int t) {
+ int v1, v2, v3;
+
+ // Q16
+ v1 = (a_i * t) >> 16;
+
+ // Q15
+ v2 = 2 * (a_i >> 1);
+ v2 = (v2 * t) >> 16;
+ v2 = (v2 * t) >> 16;
+
+ // Q16
+ v3 = (a_i * t) >> 16;
+ v3 = (v3 * t) >> 16;
+ v3 = (v3 * t) >> 16;
+
+ // Q12
+ return ((v2 - (v1 >> 1) - (v3 >> 1)) >> 3);
+}
+#else
+// 3 2
+// C0 = -a*t + a*t
+//
+float C0(float t) {
+ return -a * t * t * t + a * t * t;
+}
+
+// 2 3
+// C1 = -a*t + (2*a+3)*t - (a+2)*t
+//
+float C1(float t) {
+ return -(a + 2.0f) * t * t * t + (2.0f * a + 3.0f) * t * t - a * t;
+}
+
+// 2 3
+// C2 = 1 - (a+3)*t + (a+2)*t
+//
+float C2(float t) {
+ return (a + 2.0f) * t * t * t - (a + 3.0f) * t * t + 1.0f;
+}
+
+// 2 3
+// C3 = a*t - 2*a*t + a*t
+//
+float C3(float t) {
+ return a * t * t * t - 2.0f * a * t * t + a * t;
+}
+#endif
+
+#if 0
+int compare_real_fixed() {
+ int i, errors = 0;
+ float mult = 1.0 / 10000.0;
+ unsigned int fixed_mult = mult * 4294967296;// 65536;
+ unsigned int phase_offset_int;
+ float phase_offset_real;
+
+ for (i = 0; i < 10000; i++) {
+ int fixed0, fixed1, fixed2, fixed3, fixed_total;
+ int real0, real1, real2, real3, real_total;
+
+ phase_offset_real = (float)i * mult;
+ phase_offset_int = (fixed_mult * i) >> 16;
+// phase_offset_int = phase_offset_real * 65536;
+
+ fixed0 = c0_fixed(phase_offset_int);
+ real0 = C0(phase_offset_real) * 4096.0;
+
+ if ((abs(fixed0) > (abs(real0) + 1)) || (abs(fixed0) < (abs(real0) - 1)))
+ errors++;
+
+ fixed1 = c1_fixed(phase_offset_int);
+ real1 = C1(phase_offset_real) * 4096.0;
+
+ if ((abs(fixed1) > (abs(real1) + 1)) || (abs(fixed1) < (abs(real1) - 1)))
+ errors++;
+
+ fixed2 = c2_fixed(phase_offset_int);
+ real2 = C2(phase_offset_real) * 4096.0;
+
+ if ((abs(fixed2) > (abs(real2) + 1)) || (abs(fixed2) < (abs(real2) - 1)))
+ errors++;
+
+ fixed3 = c3_fixed(phase_offset_int);
+ real3 = C3(phase_offset_real) * 4096.0;
+
+ if ((abs(fixed3) > (abs(real3) + 1)) || (abs(fixed3) < (abs(real3) - 1)))
+ errors++;
+
+ fixed_total = fixed0 + fixed1 + fixed2 + fixed3;
+ real_total = real0 + real1 + real2 + real3;
+
+ if ((fixed_total > 4097) || (fixed_total < 4094))
+ errors++;
+
+ if ((real_total > 4097) || (real_total < 4095))
+ errors++;
+ }
+
+ return errors;
+}
+#endif
+
+// Find greatest common denominator between two integers. Method used here is
+// slow compared to Euclid's algorithm, but does not require any division.
+int gcd(int a, int b) {
+ // Problem with this algorithm is that if a or b = 0 this function
+ // will never exit. Don't want to return 0 because any computation
+ // that was based on a common denoninator and tried to reduce by
+ // dividing by 0 would fail. Best solution that could be thought of
+ // would to be fail by returing a 1;
+ if (a <= 0 || b <= 0)
+ return 1;
+
+ while (a != b) {
+ if (b > a)
+ b = b - a;
+ else {
+ int tmp = a;// swap large and
+ a = b; // small
+ b = tmp;
+ }
+ }
+
+ return b;
+}
+
+void bicubic_coefficient_init() {
+ vpx_memset(&g_b_scaler, 0, sizeof(BICUBIC_SCALER_STRUCT));
+ g_first_time = 0;
+}
+
+void bicubic_coefficient_destroy() {
+ if (!g_first_time) {
+ vpx_free(g_b_scaler.l_w);
+
+ vpx_free(g_b_scaler.l_h);
+
+ vpx_free(g_b_scaler.l_h_uv);
+
+ vpx_free(g_b_scaler.c_w);
+
+ vpx_free(g_b_scaler.c_h);
+
+ vpx_free(g_b_scaler.c_h_uv);
+
+ vpx_memset(&g_b_scaler, 0, sizeof(BICUBIC_SCALER_STRUCT));
+ }
+}
+
+// Create the coeffients that will be used for the cubic interpolation.
+// Because scaling does not have to be equal in the vertical and horizontal
+// regimes the phase offsets will be different. There are 4 coefficents
+// for each point, two on each side. The layout is that there are the
+// 4 coefficents for each phase in the array and then the next phase.
+int bicubic_coefficient_setup(int in_width, int in_height, int out_width, int out_height) {
+ int i;
+#ifdef FIXED_POINT
+ int phase_offset_int;
+ unsigned int fixed_mult;
+ int product_val = 0;
+#else
+ float phase_offset;
+#endif
+ int gcd_w, gcd_h, gcd_h_uv, d_w, d_h, d_h_uv;
+
+ if (g_first_time)
+ bicubic_coefficient_init();
+
+
+ // check to see if the coefficents have already been set up correctly
+ if ((in_width == g_b_scaler.in_width) && (in_height == g_b_scaler.in_height)
+ && (out_width == g_b_scaler.out_width) && (out_height == g_b_scaler.out_height))
+ return 0;
+
+ g_b_scaler.in_width = in_width;
+ g_b_scaler.in_height = in_height;
+ g_b_scaler.out_width = out_width;
+ g_b_scaler.out_height = out_height;
+
+ // Don't want to allow crazy scaling, just try and prevent a catastrophic
+ // failure here. Want to fail after setting the member functions so if
+ // if the scaler is called the member functions will not scale.
+ if (out_width <= 0 || out_height <= 0)
+ return -1;
+
+ // reduce in/out width and height ratios using the gcd
+ gcd_w = gcd(out_width, in_width);
+ gcd_h = gcd(out_height, in_height);
+ gcd_h_uv = gcd(out_height, in_height / 2);
+
+ // the numerator width and height are to be saved in
+ // globals so they can be used during the scaling process
+ // without having to be recalculated.
+ g_b_scaler.nw = out_width / gcd_w;
+ d_w = in_width / gcd_w;
+
+ g_b_scaler.nh = out_height / gcd_h;
+ d_h = in_height / gcd_h;
+
+ g_b_scaler.nh_uv = out_height / gcd_h_uv;
+ d_h_uv = (in_height / 2) / gcd_h_uv;
+
+ // allocate memory for the coefficents
+ vpx_free(g_b_scaler.l_w);
+
+ vpx_free(g_b_scaler.l_h);
+
+ vpx_free(g_b_scaler.l_h_uv);
+
+ g_b_scaler.l_w = (short *)vpx_memalign(32, out_width * 2);
+ g_b_scaler.l_h = (short *)vpx_memalign(32, out_height * 2);
+ g_b_scaler.l_h_uv = (short *)vpx_memalign(32, out_height * 2);
+
+ vpx_free(g_b_scaler.c_w);
+
+ vpx_free(g_b_scaler.c_h);
+
+ vpx_free(g_b_scaler.c_h_uv);
+
+ g_b_scaler.c_w = (short *)vpx_memalign(32, g_b_scaler.nw * 4 * 2);
+ g_b_scaler.c_h = (short *)vpx_memalign(32, g_b_scaler.nh * 4 * 2);
+ g_b_scaler.c_h_uv = (short *)vpx_memalign(32, g_b_scaler.nh_uv * 4 * 2);
+
+ g_b_scaler.hbuf = g_hbuf;
+ g_b_scaler.hbuf_uv = g_hbuf_uv;
+
+ // Set up polyphase filter taps. This needs to be done before
+ // the scaling because of the floating point math required. The
+ // coefficients are multiplied by 2^12 so that fixed point math
+ // can be used in the main scaling loop.
+#ifdef FIXED_POINT
+ fixed_mult = (1.0 / (float)g_b_scaler.nw) * 4294967296;
+
+ product_val = 0;
+
+ for (i = 0; i < g_b_scaler.nw; i++) {
+ if (product_val > g_b_scaler.nw)
+ product_val -= g_b_scaler.nw;
+
+ phase_offset_int = (fixed_mult * product_val) >> 16;
+
+ g_b_scaler.c_w[i * 4] = c3_fixed(phase_offset_int);
+ g_b_scaler.c_w[i * 4 + 1] = c2_fixed(phase_offset_int);
+ g_b_scaler.c_w[i * 4 + 2] = c1_fixed(phase_offset_int);
+ g_b_scaler.c_w[i * 4 + 3] = c0_fixed(phase_offset_int);
+
+ product_val += d_w;
+ }
+
+
+ fixed_mult = (1.0 / (float)g_b_scaler.nh) * 4294967296;
+
+ product_val = 0;
+
+ for (i = 0; i < g_b_scaler.nh; i++) {
+ if (product_val > g_b_scaler.nh)
+ product_val -= g_b_scaler.nh;
+
+ phase_offset_int = (fixed_mult * product_val) >> 16;
+
+ g_b_scaler.c_h[i * 4] = c0_fixed(phase_offset_int);
+ g_b_scaler.c_h[i * 4 + 1] = c1_fixed(phase_offset_int);
+ g_b_scaler.c_h[i * 4 + 2] = c2_fixed(phase_offset_int);
+ g_b_scaler.c_h[i * 4 + 3] = c3_fixed(phase_offset_int);
+
+ product_val += d_h;
+ }
+
+ fixed_mult = (1.0 / (float)g_b_scaler.nh_uv) * 4294967296;
+
+ product_val = 0;
+
+ for (i = 0; i < g_b_scaler.nh_uv; i++) {
+ if (product_val > g_b_scaler.nh_uv)
+ product_val -= g_b_scaler.nh_uv;
+
+ phase_offset_int = (fixed_mult * product_val) >> 16;
+
+ g_b_scaler.c_h_uv[i * 4] = c0_fixed(phase_offset_int);
+ g_b_scaler.c_h_uv[i * 4 + 1] = c1_fixed(phase_offset_int);
+ g_b_scaler.c_h_uv[i * 4 + 2] = c2_fixed(phase_offset_int);
+ g_b_scaler.c_h_uv[i * 4 + 3] = c3_fixed(phase_offset_int);
+
+ product_val += d_h_uv;
+ }
+
+#else
+
+ for (i = 0; i < g_nw; i++) {
+ phase_offset = (float)((i * d_w) % g_nw) / (float)g_nw;
+ g_c_w[i * 4] = (C3(phase_offset) * 4096.0);
+ g_c_w[i * 4 + 1] = (C2(phase_offset) * 4096.0);
+ g_c_w[i * 4 + 2] = (C1(phase_offset) * 4096.0);
+ g_c_w[i * 4 + 3] = (C0(phase_offset) * 4096.0);
+ }
+
+ for (i = 0; i < g_nh; i++) {
+ phase_offset = (float)((i * d_h) % g_nh) / (float)g_nh;
+ g_c_h[i * 4] = (C0(phase_offset) * 4096.0);
+ g_c_h[i * 4 + 1] = (C1(phase_offset) * 4096.0);
+ g_c_h[i * 4 + 2] = (C2(phase_offset) * 4096.0);
+ g_c_h[i * 4 + 3] = (C3(phase_offset) * 4096.0);
+ }
+
+ for (i = 0; i < g_nh_uv; i++) {
+ phase_offset = (float)((i * d_h_uv) % g_nh_uv) / (float)g_nh_uv;
+ g_c_h_uv[i * 4] = (C0(phase_offset) * 4096.0);
+ g_c_h_uv[i * 4 + 1] = (C1(phase_offset) * 4096.0);
+ g_c_h_uv[i * 4 + 2] = (C2(phase_offset) * 4096.0);
+ g_c_h_uv[i * 4 + 3] = (C3(phase_offset) * 4096.0);
+ }
+
+#endif
+
+ // Create an array that corresponds input lines to output lines.
+ // This doesn't require floating point math, but it does require
+ // a division and because hardware division is not present that
+ // is a call.
+ for (i = 0; i < out_width; i++) {
+ g_b_scaler.l_w[i] = (i * d_w) / g_b_scaler.nw;
+
+ if ((g_b_scaler.l_w[i] + 2) <= in_width)
+ g_b_scaler.max_usable_out_width = i;
+
+ }
+
+ for (i = 0; i < out_height + 1; i++) {
+ g_b_scaler.l_h[i] = (i * d_h) / g_b_scaler.nh;
+ g_b_scaler.l_h_uv[i] = (i * d_h_uv) / g_b_scaler.nh_uv;
+ }
+
+ return 0;
+}
+
+int bicubic_scale(int in_width, int in_height, int in_stride,
+ int out_width, int out_height, int out_stride,
+ unsigned char *input_image, unsigned char *output_image) {
+ short *RESTRICT l_w, * RESTRICT l_h;
+ short *RESTRICT c_w, * RESTRICT c_h;
+ unsigned char *RESTRICT ip, * RESTRICT op;
+ unsigned char *RESTRICT hbuf;
+ int h, w, lw, lh;
+ int temp_sum;
+ int phase_offset_w, phase_offset_h;
+
+ c_w = g_b_scaler.c_w;
+ c_h = g_b_scaler.c_h;
+
+ op = output_image;
+
+ l_w = g_b_scaler.l_w;
+ l_h = g_b_scaler.l_h;
+
+ phase_offset_h = 0;
+
+ for (h = 0; h < out_height; h++) {
+ // select the row to work on
+ lh = l_h[h];
+ ip = input_image + (in_stride * lh);
+
+ // vp8_filter the row vertically into an temporary buffer.
+ // If the phase offset == 0 then all the multiplication
+ // is going to result in the output equalling the input.
+ // So instead point the temporary buffer to the input.
+ // Also handle the boundry condition of not being able to
+ // filter that last lines.
+ if (phase_offset_h && (lh < in_height - 2)) {
+ hbuf = g_b_scaler.hbuf;
+
+ for (w = 0; w < in_width; w++) {
+ temp_sum = c_h[phase_offset_h * 4 + 3] * ip[w - in_stride];
+ temp_sum += c_h[phase_offset_h * 4 + 2] * ip[w];
+ temp_sum += c_h[phase_offset_h * 4 + 1] * ip[w + in_stride];
+ temp_sum += c_h[phase_offset_h * 4] * ip[w + 2 * in_stride];
+
+ hbuf[w] = temp_sum >> 12;
+ }
+ } else
+ hbuf = ip;
+
+ // increase the phase offset for the next time around.
+ if (++phase_offset_h >= g_b_scaler.nh)
+ phase_offset_h = 0;
+
+ // now filter and expand it horizontally into the final
+ // output buffer
+ phase_offset_w = 0;
+
+ for (w = 0; w < out_width; w++) {
+ // get the index to use to expand the image
+ lw = l_w[w];
+
+ temp_sum = c_w[phase_offset_w * 4] * hbuf[lw - 1];
+ temp_sum += c_w[phase_offset_w * 4 + 1] * hbuf[lw];
+ temp_sum += c_w[phase_offset_w * 4 + 2] * hbuf[lw + 1];
+ temp_sum += c_w[phase_offset_w * 4 + 3] * hbuf[lw + 2];
+ temp_sum = temp_sum >> 12;
+
+ if (++phase_offset_w >= g_b_scaler.nw)
+ phase_offset_w = 0;
+
+ // boundry conditions
+ if ((lw + 2) >= in_width)
+ temp_sum = hbuf[lw];
+
+ if (lw == 0)
+ temp_sum = hbuf[0];
+
+ op[w] = temp_sum;
+ }
+
+ op += out_stride;
+ }
+
+ return 0;
+}
+
+void bicubic_scale_frame_reset() {
+ g_b_scaler.out_width = 0;
+ g_b_scaler.out_height = 0;
+}
+
+void bicubic_scale_frame(YV12_BUFFER_CONFIG *src, YV12_BUFFER_CONFIG *dst,
+ int new_width, int new_height) {
+
+ dst->y_width = new_width;
+ dst->y_height = new_height;
+ dst->uv_width = new_width / 2;
+ dst->uv_height = new_height / 2;
+
+ dst->y_stride = dst->y_width;
+ dst->uv_stride = dst->uv_width;
+
+ bicubic_scale(src->y_width, src->y_height, src->y_stride,
+ new_width, new_height, dst->y_stride,
+ src->y_buffer, dst->y_buffer);
+
+ bicubic_scale(src->uv_width, src->uv_height, src->uv_stride,
+ new_width / 2, new_height / 2, dst->uv_stride,
+ src->u_buffer, dst->u_buffer);
+
+ bicubic_scale(src->uv_width, src->uv_height, src->uv_stride,
+ new_width / 2, new_height / 2, dst->uv_stride,
+ src->v_buffer, dst->v_buffer);
+}
diff --git a/libvpx/vpx_scale/generic/gen_scalers.c b/libvpx/vpx_scale/generic/gen_scalers.c
new file mode 100644
index 0000000..60c21fb
--- /dev/null
+++ b/libvpx/vpx_scale/generic/gen_scalers.c
@@ -0,0 +1,922 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_scale/vpxscale.h"
+#include "vpx_mem/vpx_mem.h"
+/****************************************************************************
+* Imports
+****************************************************************************/
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_horizontal_line_4_5_scale_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 4 to 5.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+void vp8_horizontal_line_4_5_scale_c(const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width) {
+ unsigned i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for (i = 0; i < source_width - 4; i += 4) {
+ a = src[0];
+ b = src[1];
+ des [0] = (unsigned char) a;
+ des [1] = (unsigned char)((a * 51 + 205 * b + 128) >> 8);
+ c = src[2] * 154;
+ a = src[3];
+ des [2] = (unsigned char)((b * 102 + c + 128) >> 8);
+ des [3] = (unsigned char)((c + 102 * a + 128) >> 8);
+ b = src[4];
+ des [4] = (unsigned char)((a * 205 + 51 * b + 128) >> 8);
+
+ src += 4;
+ des += 5;
+ }
+
+ a = src[0];
+ b = src[1];
+ des [0] = (unsigned char)(a);
+ des [1] = (unsigned char)((a * 51 + 205 * b + 128) >> 8);
+ c = src[2] * 154;
+ a = src[3];
+ des [2] = (unsigned char)((b * 102 + c + 128) >> 8);
+ des [3] = (unsigned char)((c + 102 * a + 128) >> 8);
+ des [4] = (unsigned char)(a);
+
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_vertical_band_4_5_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales vertical band of pixels by scale 4 to 5. The
+ * height of the band scaled is 4-pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band.
+ *
+ ****************************************************************************/
+void vp8_vertical_band_4_5_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c, d;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; i++) {
+ a = des [0];
+ b = des [dest_pitch];
+
+ des[dest_pitch] = (unsigned char)((a * 51 + 205 * b + 128) >> 8);
+
+ c = des[dest_pitch * 2] * 154;
+ d = des[dest_pitch * 3];
+
+ des [dest_pitch * 2] = (unsigned char)((b * 102 + c + 128) >> 8);
+ des [dest_pitch * 3] = (unsigned char)((c + 102 * d + 128) >> 8);
+
+ /* First line in next band */
+ a = des [dest_pitch * 5];
+ des [dest_pitch * 4] = (unsigned char)((d * 205 + 51 * a + 128) >> 8);
+
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_last_vertical_band_4_5_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales last vertical band of pixels by scale 4 to 5. The
+ * height of the band scaled is 4-pixels.
+ *
+ * SPECIAL NOTES : The routine does not have available the first line of
+ * the band below the current band, since this is the
+ * last band.
+ *
+ ****************************************************************************/
+void vp8_last_vertical_band_4_5_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c, d;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; ++i) {
+ a = des[0];
+ b = des[dest_pitch];
+
+ des[dest_pitch] = (unsigned char)((a * 51 + 205 * b + 128) >> 8);
+
+ c = des[dest_pitch * 2] * 154;
+ d = des[dest_pitch * 3];
+
+ des [dest_pitch * 2] = (unsigned char)((b * 102 + c + 128) >> 8);
+ des [dest_pitch * 3] = (unsigned char)((c + 102 * d + 128) >> 8);
+
+ /* No other line for interplation of this line, so .. */
+ des[dest_pitch * 4] = (unsigned char) d;
+
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_horizontal_line_2_3_scale_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 2 to 3.
+ *
+ * SPECIAL NOTES : None.
+ *
+ *
+ ****************************************************************************/
+void vp8_horizontal_line_2_3_scale_c(const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for (i = 0; i < source_width - 2; i += 2) {
+ a = src[0];
+ b = src[1];
+ c = src[2];
+
+ des [0] = (unsigned char)(a);
+ des [1] = (unsigned char)((a * 85 + 171 * b + 128) >> 8);
+ des [2] = (unsigned char)((b * 171 + 85 * c + 128) >> 8);
+
+ src += 2;
+ des += 3;
+ }
+
+ a = src[0];
+ b = src[1];
+ des [0] = (unsigned char)(a);
+ des [1] = (unsigned char)((a * 85 + 171 * b + 128) >> 8);
+ des [2] = (unsigned char)(b);
+}
+
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_vertical_band_2_3_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales vertical band of pixels by scale 2 to 3. The
+ * height of the band scaled is 2-pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band.
+ *
+ ****************************************************************************/
+void vp8_vertical_band_2_3_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; i++) {
+ a = des [0];
+ b = des [dest_pitch];
+ c = des[dest_pitch * 3];
+ des [dest_pitch ] = (unsigned char)((a * 85 + 171 * b + 128) >> 8);
+ des [dest_pitch * 2] = (unsigned char)((b * 171 + 85 * c + 128) >> 8);
+
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_last_vertical_band_2_3_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales last vertical band of pixels by scale 2 to 3. The
+ * height of the band scaled is 2-pixels.
+ *
+ * SPECIAL NOTES : The routine does not have available the first line of
+ * the band below the current band, since this is the
+ * last band.
+ *
+ ****************************************************************************/
+void vp8_last_vertical_band_2_3_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; ++i) {
+ a = des [0];
+ b = des [dest_pitch];
+
+ des [dest_pitch ] = (unsigned char)((a * 85 + 171 * b + 128) >> 8);
+ des [dest_pitch * 2] = (unsigned char)(b);
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_horizontal_line_3_5_scale_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 3 to 5.
+ *
+ * SPECIAL NOTES : None.
+ *
+ *
+ ****************************************************************************/
+void vp8_horizontal_line_3_5_scale_c(const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for (i = 0; i < source_width - 3; i += 3) {
+ a = src[0];
+ b = src[1];
+ des [0] = (unsigned char)(a);
+ des [1] = (unsigned char)((a * 102 + 154 * b + 128) >> 8);
+
+ c = src[2];
+ des [2] = (unsigned char)((b * 205 + c * 51 + 128) >> 8);
+ des [3] = (unsigned char)((b * 51 + c * 205 + 128) >> 8);
+
+ a = src[3];
+ des [4] = (unsigned char)((c * 154 + a * 102 + 128) >> 8);
+
+ src += 3;
+ des += 5;
+ }
+
+ a = src[0];
+ b = src[1];
+ des [0] = (unsigned char)(a);
+
+ des [1] = (unsigned char)((a * 102 + 154 * b + 128) >> 8);
+ c = src[2];
+ des [2] = (unsigned char)((b * 205 + c * 51 + 128) >> 8);
+ des [3] = (unsigned char)((b * 51 + c * 205 + 128) >> 8);
+
+ des [4] = (unsigned char)(c);
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_vertical_band_3_5_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales vertical band of pixels by scale 3 to 5. The
+ * height of the band scaled is 3-pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band.
+ *
+ ****************************************************************************/
+void vp8_vertical_band_3_5_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; i++) {
+ a = des [0];
+ b = des [dest_pitch];
+ des [dest_pitch] = (unsigned char)((a * 102 + 154 * b + 128) >> 8);
+
+ c = des[dest_pitch * 2];
+ des [dest_pitch * 2] = (unsigned char)((b * 205 + c * 51 + 128) >> 8);
+ des [dest_pitch * 3] = (unsigned char)((b * 51 + c * 205 + 128) >> 8);
+
+ /* First line in next band... */
+ a = des [dest_pitch * 5];
+ des [dest_pitch * 4] = (unsigned char)((c * 154 + a * 102 + 128) >> 8);
+
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_last_vertical_band_3_5_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales last vertical band of pixels by scale 3 to 5. The
+ * height of the band scaled is 3-pixels.
+ *
+ * SPECIAL NOTES : The routine does not have available the first line of
+ * the band below the current band, since this is the
+ * last band.
+ *
+ ****************************************************************************/
+void vp8_last_vertical_band_3_5_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; ++i) {
+ a = des [0];
+ b = des [dest_pitch];
+
+ des [ dest_pitch ] = (unsigned char)((a * 102 + 154 * b + 128) >> 8);
+
+ c = des[dest_pitch * 2];
+ des [dest_pitch * 2] = (unsigned char)((b * 205 + c * 51 + 128) >> 8);
+ des [dest_pitch * 3] = (unsigned char)((b * 51 + c * 205 + 128) >> 8);
+
+ /* No other line for interplation of this line, so .. */
+ des [ dest_pitch * 4 ] = (unsigned char)(c);
+
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_horizontal_line_3_4_scale_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 3 to 4.
+ *
+ * SPECIAL NOTES : None.
+ *
+ *
+ ****************************************************************************/
+void vp8_horizontal_line_3_4_scale_c(const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for (i = 0; i < source_width - 3; i += 3) {
+ a = src[0];
+ b = src[1];
+ des [0] = (unsigned char)(a);
+ des [1] = (unsigned char)((a * 64 + b * 192 + 128) >> 8);
+
+ c = src[2];
+ des [2] = (unsigned char)((b + c + 1) >> 1);
+
+ a = src[3];
+ des [3] = (unsigned char)((c * 192 + a * 64 + 128) >> 8);
+
+ src += 3;
+ des += 4;
+ }
+
+ a = src[0];
+ b = src[1];
+ des [0] = (unsigned char)(a);
+ des [1] = (unsigned char)((a * 64 + b * 192 + 128) >> 8);
+
+ c = src[2];
+ des [2] = (unsigned char)((b + c + 1) >> 1);
+ des [3] = (unsigned char)(c);
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_vertical_band_3_4_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales vertical band of pixels by scale 3 to 4. The
+ * height of the band scaled is 3-pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band.
+ *
+ ****************************************************************************/
+void vp8_vertical_band_3_4_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; i++) {
+ a = des [0];
+ b = des [dest_pitch];
+ des [dest_pitch] = (unsigned char)((a * 64 + b * 192 + 128) >> 8);
+
+ c = des[dest_pitch * 2];
+ des [dest_pitch * 2] = (unsigned char)((b + c + 1) >> 1);
+
+ /* First line in next band... */
+ a = des [dest_pitch * 4];
+ des [dest_pitch * 3] = (unsigned char)((c * 192 + a * 64 + 128) >> 8);
+
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_last_vertical_band_3_4_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales last vertical band of pixels by scale 3 to 4. The
+ * height of the band scaled is 3-pixels.
+ *
+ * SPECIAL NOTES : The routine does not have available the first line of
+ * the band below the current band, since this is the
+ * last band.
+ *
+ ****************************************************************************/
+void vp8_last_vertical_band_3_4_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; ++i) {
+ a = des [0];
+ b = des [dest_pitch];
+
+ des [dest_pitch] = (unsigned char)((a * 64 + b * 192 + 128) >> 8);
+
+ c = des[dest_pitch * 2];
+ des [dest_pitch * 2] = (unsigned char)((b + c + 1) >> 1);
+
+ /* No other line for interplation of this line, so .. */
+ des [dest_pitch * 3] = (unsigned char)(c);
+
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_horizontal_line_1_2_scale_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 1 to 2.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+void vp8_horizontal_line_1_2_scale_c(const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for (i = 0; i < source_width - 1; i += 1) {
+ a = src[0];
+ b = src[1];
+ des [0] = (unsigned char)(a);
+ des [1] = (unsigned char)((a + b + 1) >> 1);
+ src += 1;
+ des += 2;
+ }
+
+ a = src[0];
+ des [0] = (unsigned char)(a);
+ des [1] = (unsigned char)(a);
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_vertical_band_1_2_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales vertical band of pixels by scale 1 to 2. The
+ * height of the band scaled is 1-pixel.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band.
+ *
+ ****************************************************************************/
+void vp8_vertical_band_1_2_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; i++) {
+ a = des [0];
+ b = des [dest_pitch * 2];
+
+ des[dest_pitch] = (unsigned char)((a + b + 1) >> 1);
+
+ des++;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_last_vertical_band_1_2_scale_c
+ *
+ * INPUTS : unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_pitch : Stride of destination data.
+ * unsigned int dest_width : Width of destination data.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Scales last vertical band of pixels by scale 1 to 2. The
+ * height of the band scaled is 1-pixel.
+ *
+ * SPECIAL NOTES : The routine does not have available the first line of
+ * the band below the current band, since this is the
+ * last band.
+ *
+ ****************************************************************************/
+void vp8_last_vertical_band_1_2_scale_c(unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned char *des = dest;
+
+ for (i = 0; i < dest_width; ++i) {
+ des[dest_pitch] = des[0];
+ des++;
+ }
+}
+
+
+
+
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_horizontal_line_4_5_scale_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 4 to 5.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+void vp8_horizontal_line_5_4_scale_c(const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width) {
+ unsigned i;
+ unsigned int a, b, c, d, e;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for (i = 0; i < source_width; i += 5) {
+ a = src[0];
+ b = src[1];
+ c = src[2];
+ d = src[3];
+ e = src[4];
+
+ des[0] = (unsigned char) a;
+ des[1] = (unsigned char)((b * 192 + c * 64 + 128) >> 8);
+ des[2] = (unsigned char)((c * 128 + d * 128 + 128) >> 8);
+ des[3] = (unsigned char)((d * 64 + e * 192 + 128) >> 8);
+
+ src += 5;
+ des += 4;
+ }
+}
+
+
+
+
+void vp8_vertical_band_5_4_scale_c(unsigned char *source,
+ unsigned int src_pitch,
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c, d, e;
+ unsigned char *des = dest;
+ unsigned char *src = source;
+
+ for (i = 0; i < dest_width; i++) {
+
+ a = src[0 * src_pitch];
+ b = src[1 * src_pitch];
+ c = src[2 * src_pitch];
+ d = src[3 * src_pitch];
+ e = src[4 * src_pitch];
+
+ des[0 * dest_pitch] = (unsigned char) a;
+ des[1 * dest_pitch] = (unsigned char)((b * 192 + c * 64 + 128) >> 8);
+ des[2 * dest_pitch] = (unsigned char)((c * 128 + d * 128 + 128) >> 8);
+ des[3 * dest_pitch] = (unsigned char)((d * 64 + e * 192 + 128) >> 8);
+
+ src++;
+ des++;
+
+ }
+}
+
+
+/*7***************************************************************************
+ *
+ * ROUTINE : vp8_horizontal_line_3_5_scale_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 3 to 5.
+ *
+ * SPECIAL NOTES : None.
+ *
+ *
+ ****************************************************************************/
+void vp8_horizontal_line_5_3_scale_c(const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c, d, e;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for (i = 0; i < source_width; i += 5) {
+ a = src[0];
+ b = src[1];
+ c = src[2];
+ d = src[3];
+ e = src[4];
+
+ des[0] = (unsigned char) a;
+ des[1] = (unsigned char)((b * 85 + c * 171 + 128) >> 8);
+ des[2] = (unsigned char)((d * 171 + e * 85 + 128) >> 8);
+
+ src += 5;
+ des += 3;
+ }
+
+}
+
+void vp8_vertical_band_5_3_scale_c(unsigned char *source,
+ unsigned int src_pitch,
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a, b, c, d, e;
+ unsigned char *des = dest;
+ unsigned char *src = source;
+
+ for (i = 0; i < dest_width; i++) {
+
+ a = src[0 * src_pitch];
+ b = src[1 * src_pitch];
+ c = src[2 * src_pitch];
+ d = src[3 * src_pitch];
+ e = src[4 * src_pitch];
+
+ des[0 * dest_pitch] = (unsigned char) a;
+ des[1 * dest_pitch] = (unsigned char)((b * 85 + c * 171 + 128) >> 8);
+ des[2 * dest_pitch] = (unsigned char)((d * 171 + e * 85 + 128) >> 8);
+
+ src++;
+ des++;
+
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_horizontal_line_1_2_scale_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 1 to 2.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+void vp8_horizontal_line_2_1_scale_c(const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width) {
+ unsigned int i;
+ unsigned int a;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for (i = 0; i < source_width; i += 2) {
+ a = src[0];
+ des [0] = (unsigned char)(a);
+ src += 2;
+ des += 1;
+ }
+}
+
+void vp8_vertical_band_2_1_scale_c(unsigned char *source,
+ unsigned int src_pitch,
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ (void) dest_pitch;
+ (void) src_pitch;
+ vpx_memcpy(dest, source, dest_width);
+}
+
+void vp8_vertical_band_2_1_scale_i_c(unsigned char *source,
+ unsigned int src_pitch,
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width) {
+ int i;
+ int temp;
+ int width = dest_width;
+
+ (void) dest_pitch;
+
+ for (i = 0; i < width; i++) {
+ temp = 8;
+ temp += source[i - (int)src_pitch] * 3;
+ temp += source[i] * 10;
+ temp += source[i + src_pitch] * 3;
+ temp >>= 4;
+ dest[i] = (unsigned char)(temp);
+ }
+}
diff --git a/libvpx/vpx_scale/generic/vpxscale.c b/libvpx/vpx_scale/generic/vpxscale.c
new file mode 100644
index 0000000..7de85ca
--- /dev/null
+++ b/libvpx/vpx_scale/generic/vpxscale.c
@@ -0,0 +1,1010 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/****************************************************************************
+ *
+ * Module Title : scale.c
+ *
+ * Description : Image scaling functions.
+ *
+ ***************************************************************************/
+
+/****************************************************************************
+* Header Files
+****************************************************************************/
+#include "./vpx_rtcd.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vpx_scale/yv12config.h"
+#include "vpx_scale/scale_mode.h"
+
+typedef struct {
+ int expanded_frame_width;
+ int expanded_frame_height;
+
+ int HScale;
+ int HRatio;
+ int VScale;
+ int VRatio;
+
+ YV12_BUFFER_CONFIG *src_yuv_config;
+ YV12_BUFFER_CONFIG *dst_yuv_config;
+
+} SCALE_VARS;
+
+/****************************************************************************
+ *
+ * ROUTINE : horizontal_line_copy
+ *
+ * INPUTS : None
+ *
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : None
+ *
+ * FUNCTION : 1 to 1 scaling up for a horizontal line of pixles
+ *
+ * SPECIAL NOTES : None.
+ *
+ * ERRORS : None.
+ *
+ ****************************************************************************/
+static
+void horizontal_line_copy(
+ const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width
+) {
+ (void) dest_width;
+
+ duck_memcpy(dest, source, source_width);
+}
+/****************************************************************************
+ *
+ * ROUTINE : null_scale
+ *
+ * INPUTS : None
+ *
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : None
+ *
+ * FUNCTION : 1 to 1 scaling up for a vertical band
+ *
+ * SPECIAL NOTES : None.
+ *
+ * ERRORS : None.
+ *
+ ****************************************************************************/
+static
+void null_scale(
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width
+) {
+ (void) dest;
+ (void) dest_pitch;
+ (void) dest_width;
+
+ return;
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : scale1d_2t1_i
+ *
+ * INPUTS : const unsigned char *source : Pointer to data to be scaled.
+ * int source_step : Number of pixels to step on in source.
+ * unsigned int source_scale : Scale for source (UNUSED).
+ * unsigned int source_length : Length of source (UNUSED).
+ * unsigned char *dest : Pointer to output data array.
+ * int dest_step : Number of pixels to step on in destination.
+ * unsigned int dest_scale : Scale for destination (UNUSED).
+ * unsigned int dest_length : Length of destination.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Performs 2-to-1 interpolated scaling.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+void scale1d_2t1_i
+(
+ const unsigned char *source,
+ int source_step,
+ unsigned int source_scale,
+ unsigned int source_length,
+ unsigned char *dest,
+ int dest_step,
+ unsigned int dest_scale,
+ unsigned int dest_length
+) {
+ unsigned int i, j;
+ unsigned int temp;
+ int source_pitch = source_step;
+ (void) source_length;
+ (void) source_scale;
+ (void) dest_scale;
+
+ source_step *= 2;
+ dest[0] = source[0];
+
+ for (i = dest_step, j = source_step; i < dest_length * dest_step; i += dest_step, j += source_step) {
+ temp = 8;
+ temp += 3 * source[j - source_pitch];
+ temp += 10 * source[j];
+ temp += 3 * source[j + source_pitch];
+ temp >>= 4;
+ dest[i] = (char)(temp);
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : scale1d_2t1_ps
+ *
+ * INPUTS : const unsigned char *source : Pointer to data to be scaled.
+ * int source_step : Number of pixels to step on in source.
+ * unsigned int source_scale : Scale for source (UNUSED).
+ * unsigned int source_length : Length of source (UNUSED).
+ * unsigned char *dest : Pointer to output data array.
+ * int dest_step : Number of pixels to step on in destination.
+ * unsigned int dest_scale : Scale for destination (UNUSED).
+ * unsigned int dest_length : Length of destination.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Performs 2-to-1 point subsampled scaling.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+void scale1d_2t1_ps
+(
+ const unsigned char *source,
+ int source_step,
+ unsigned int source_scale,
+ unsigned int source_length,
+ unsigned char *dest,
+ int dest_step,
+ unsigned int dest_scale,
+ unsigned int dest_length
+) {
+ unsigned int i, j;
+
+ (void) source_length;
+ (void) source_scale;
+ (void) dest_scale;
+
+ source_step *= 2;
+ j = 0;
+
+ for (i = 0; i < dest_length * dest_step; i += dest_step, j += source_step)
+ dest[i] = source[j];
+}
+/****************************************************************************
+ *
+ * ROUTINE : scale1d_c
+ *
+ * INPUTS : const unsigned char *source : Pointer to data to be scaled.
+ * int source_step : Number of pixels to step on in source.
+ * unsigned int source_scale : Scale for source.
+ * unsigned int source_length : Length of source (UNUSED).
+ * unsigned char *dest : Pointer to output data array.
+ * int dest_step : Number of pixels to step on in destination.
+ * unsigned int dest_scale : Scale for destination.
+ * unsigned int dest_length : Length of destination.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Performs linear interpolation in one dimension.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+void scale1d_c
+(
+ const unsigned char *source,
+ int source_step,
+ unsigned int source_scale,
+ unsigned int source_length,
+ unsigned char *dest,
+ int dest_step,
+ unsigned int dest_scale,
+ unsigned int dest_length
+) {
+ unsigned int i;
+ unsigned int round_value = dest_scale / 2;
+ unsigned int left_modifier = dest_scale;
+ unsigned int right_modifier = 0;
+ unsigned char left_pixel = *source;
+ unsigned char right_pixel = *(source + source_step);
+
+ (void) source_length;
+
+ /* These asserts are needed if there are boundary issues... */
+ /*assert ( dest_scale > source_scale );*/
+ /*assert ( (source_length-1) * dest_scale >= (dest_length-1) * source_scale );*/
+
+ for (i = 0; i < dest_length * dest_step; i += dest_step) {
+ dest[i] = (char)((left_modifier * left_pixel + right_modifier * right_pixel + round_value) / dest_scale);
+
+ right_modifier += source_scale;
+
+ while (right_modifier > dest_scale) {
+ right_modifier -= dest_scale;
+ source += source_step;
+ left_pixel = *source;
+ right_pixel = *(source + source_step);
+ }
+
+ left_modifier = dest_scale - right_modifier;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : Scale2D
+ *
+ * INPUTS : const unsigned char *source : Pointer to data to be scaled.
+ * int source_pitch : Stride of source image.
+ * unsigned int source_width : Width of input image.
+ * unsigned int source_height : Height of input image.
+ * unsigned char *dest : Pointer to output data array.
+ * int dest_pitch : Stride of destination image.
+ * unsigned int dest_width : Width of destination image.
+ * unsigned int dest_height : Height of destination image.
+ * unsigned char *temp_area : Pointer to temp work area.
+ * unsigned char temp_area_height : Height of temp work area.
+ * unsigned int hscale : Horizontal scale factor numerator.
+ * unsigned int hratio : Horizontal scale factor denominator.
+ * unsigned int vscale : Vertical scale factor numerator.
+ * unsigned int vratio : Vertical scale factor denominator.
+ * unsigned int interlaced : Interlace flag.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Performs 2-tap linear interpolation in two dimensions.
+ *
+ * SPECIAL NOTES : Expansion is performed one band at a time to help with
+ * caching.
+ *
+ ****************************************************************************/
+static
+void Scale2D
+(
+ /*const*/
+ unsigned char *source,
+ int source_pitch,
+ unsigned int source_width,
+ unsigned int source_height,
+ unsigned char *dest,
+ int dest_pitch,
+ unsigned int dest_width,
+ unsigned int dest_height,
+ unsigned char *temp_area,
+ unsigned char temp_area_height,
+ unsigned int hscale,
+ unsigned int hratio,
+ unsigned int vscale,
+ unsigned int vratio,
+ unsigned int interlaced
+) {
+ /*unsigned*/
+ int i, j, k;
+ int bands;
+ int dest_band_height;
+ int source_band_height;
+
+ typedef void (*Scale1D)(const unsigned char * source, int source_step, unsigned int source_scale, unsigned int source_length,
+ unsigned char * dest, int dest_step, unsigned int dest_scale, unsigned int dest_length);
+
+ Scale1D Scale1Dv = scale1d_c;
+ Scale1D Scale1Dh = scale1d_c;
+
+ void (*horiz_line_scale)(const unsigned char *, unsigned int, unsigned char *, unsigned int) = NULL;
+ void (*vert_band_scale)(unsigned char *, unsigned int, unsigned char *, unsigned int, unsigned int) = NULL;
+
+ int ratio_scalable = 1;
+ int interpolation = 0;
+
+ unsigned char *source_base; /* = (unsigned char *) ((source_pitch >= 0) ? source : (source + ((source_height-1) * source_pitch))); */
+ unsigned char *line_src;
+
+
+ source_base = (unsigned char *)source;
+
+ if (source_pitch < 0) {
+ int offset;
+
+ offset = (source_height - 1);
+ offset *= source_pitch;
+
+ source_base += offset;
+ }
+
+ /* find out the ratio for each direction */
+ switch (hratio * 10 / hscale) {
+ case 8:
+ /* 4-5 Scale in Width direction */
+ horiz_line_scale = vp8_horizontal_line_5_4_scale;
+ break;
+ case 6:
+ /* 3-5 Scale in Width direction */
+ horiz_line_scale = vp8_horizontal_line_5_3_scale;
+ break;
+ case 5:
+ /* 1-2 Scale in Width direction */
+ horiz_line_scale = vp8_horizontal_line_2_1_scale;
+ break;
+ default:
+ /* The ratio is not acceptable now */
+ /* throw("The ratio is not acceptable for now!"); */
+ ratio_scalable = 0;
+ break;
+ }
+
+ switch (vratio * 10 / vscale) {
+ case 8:
+ /* 4-5 Scale in vertical direction */
+ vert_band_scale = vp8_vertical_band_5_4_scale;
+ source_band_height = 5;
+ dest_band_height = 4;
+ break;
+ case 6:
+ /* 3-5 Scale in vertical direction */
+ vert_band_scale = vp8_vertical_band_5_3_scale;
+ source_band_height = 5;
+ dest_band_height = 3;
+ break;
+ case 5:
+ /* 1-2 Scale in vertical direction */
+
+ if (interlaced) {
+ /* if the content is interlaced, point sampling is used */
+ vert_band_scale = vp8_vertical_band_2_1_scale;
+ } else {
+
+ interpolation = 1;
+ /* if the content is progressive, interplo */
+ vert_band_scale = vp8_vertical_band_2_1_scale_i;
+
+ }
+
+ source_band_height = 2;
+ dest_band_height = 1;
+ break;
+ default:
+ /* The ratio is not acceptable now */
+ /* throw("The ratio is not acceptable for now!"); */
+ ratio_scalable = 0;
+ break;
+ }
+
+ if (ratio_scalable) {
+ if (source_height == dest_height) {
+ /* for each band of the image */
+ for (k = 0; k < (int)dest_height; k++) {
+ horiz_line_scale(source, source_width, dest, dest_width);
+ source += source_pitch;
+ dest += dest_pitch;
+ }
+
+ return;
+ }
+
+ if (interpolation) {
+ if (source < source_base)
+ source = source_base;
+
+ horiz_line_scale(source, source_width, temp_area, dest_width);
+ }
+
+ for (k = 0; k < (int)(dest_height + dest_band_height - 1) / dest_band_height; k++) {
+ /* scale one band horizontally */
+ for (i = 0; i < source_band_height; i++) {
+ /* Trap case where we could read off the base of the source buffer */
+
+ line_src = (unsigned char *)source + i * source_pitch;
+
+ if (line_src < source_base)
+ line_src = source_base;
+
+ horiz_line_scale(line_src, source_width,
+ temp_area + (i + 1)*dest_pitch, dest_width);
+ }
+
+ /* Vertical scaling is in place */
+ vert_band_scale(temp_area + dest_pitch, dest_pitch, dest, dest_pitch, dest_width);
+
+ if (interpolation)
+ vpx_memcpy(temp_area, temp_area + source_band_height * dest_pitch, dest_width);
+
+ /* Next band... */
+ source += (unsigned long) source_band_height * source_pitch;
+ dest += (unsigned long) dest_band_height * dest_pitch;
+ }
+
+ return;
+ }
+
+ if (hscale == 2 && hratio == 1)
+ Scale1Dh = scale1d_2t1_ps;
+
+ if (vscale == 2 && vratio == 1) {
+ if (interlaced)
+ Scale1Dv = scale1d_2t1_ps;
+ else
+ Scale1Dv = scale1d_2t1_i;
+ }
+
+ if (source_height == dest_height) {
+ /* for each band of the image */
+ for (k = 0; k < (int)dest_height; k++) {
+ Scale1Dh(source, 1, hscale, source_width + 1, dest, 1, hratio, dest_width);
+ source += source_pitch;
+ dest += dest_pitch;
+ }
+
+ return;
+ }
+
+ if (dest_height > source_height) {
+ dest_band_height = temp_area_height - 1;
+ source_band_height = dest_band_height * source_height / dest_height;
+ } else {
+ source_band_height = temp_area_height - 1;
+ dest_band_height = source_band_height * vratio / vscale;
+ }
+
+ /* first row needs to be done so that we can stay one row ahead for vertical zoom */
+ Scale1Dh(source, 1, hscale, source_width + 1, temp_area, 1, hratio, dest_width);
+
+ /* for each band of the image */
+ bands = (dest_height + dest_band_height - 1) / dest_band_height;
+
+ for (k = 0; k < bands; k++) {
+ /* scale one band horizontally */
+ for (i = 1; i < source_band_height + 1; i++) {
+ if (k * source_band_height + i < (int) source_height) {
+ Scale1Dh(source + i * source_pitch, 1, hscale, source_width + 1,
+ temp_area + i * dest_pitch, 1, hratio, dest_width);
+ } else { /* Duplicate the last row */
+ /* copy temp_area row 0 over from last row in the past */
+ duck_memcpy(temp_area + i * dest_pitch, temp_area + (i - 1)*dest_pitch, dest_pitch);
+ }
+ }
+
+ /* scale one band vertically */
+ for (j = 0; j < (int)dest_width; j++) {
+ Scale1Dv(&temp_area[j], dest_pitch, vscale, source_band_height + 1,
+ &dest[j], dest_pitch, vratio, dest_band_height);
+ }
+
+ /* copy temp_area row 0 over from last row in the past */
+ duck_memcpy(temp_area, temp_area + source_band_height * dest_pitch, dest_pitch);
+
+ /* move to the next band */
+ source += source_band_height * source_pitch;
+ dest += dest_band_height * dest_pitch;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE :
+ *
+ * INPUTS : YV12_BUFFER_CONFIG *src : Pointer to frame to be scaled.
+ * YV12_BUFFER_CONFIG *dst : Pointer to buffer to hold scaled frame.
+ * unsigned char *temp_area : Pointer to temp work area.
+ * unsigned char temp_area_height : Height of temp work area.
+ * unsigned int hscale : Horizontal scale factor numerator.
+ * unsigned int hratio : Horizontal scale factor denominator.
+ * unsigned int vscale : Vertical scale factor numerator.
+ * unsigned int vratio : Vertical scale factor denominator.
+ * unsigned int interlaced : Interlace flag.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Performs 2-tap linear interpolation in two dimensions.
+ *
+ * SPECIAL NOTES : Expansion is performed one band at a time to help with
+ * caching.
+ *
+ ****************************************************************************/
+void vp8_scale_frame
+(
+ YV12_BUFFER_CONFIG *src,
+ YV12_BUFFER_CONFIG *dst,
+ unsigned char *temp_area,
+ unsigned char temp_height,
+ unsigned int hscale,
+ unsigned int hratio,
+ unsigned int vscale,
+ unsigned int vratio,
+ unsigned int interlaced
+) {
+ int i;
+ int dw = (hscale - 1 + src->y_width * hratio) / hscale;
+ int dh = (vscale - 1 + src->y_height * vratio) / vscale;
+
+ /* call our internal scaling routines!! */
+ Scale2D((unsigned char *) src->y_buffer, src->y_stride, src->y_width, src->y_height,
+ (unsigned char *) dst->y_buffer, dst->y_stride, dw, dh,
+ temp_area, temp_height, hscale, hratio, vscale, vratio, interlaced);
+
+ if (dw < (int)dst->y_width)
+ for (i = 0; i < dh; i++)
+ duck_memset(dst->y_buffer + i * dst->y_stride + dw - 1, dst->y_buffer[i * dst->y_stride + dw - 2], dst->y_width - dw + 1);
+
+ if (dh < (int)dst->y_height)
+ for (i = dh - 1; i < (int)dst->y_height; i++)
+ duck_memcpy(dst->y_buffer + i * dst->y_stride, dst->y_buffer + (dh - 2) * dst->y_stride, dst->y_width + 1);
+
+ Scale2D((unsigned char *) src->u_buffer, src->uv_stride, src->uv_width, src->uv_height,
+ (unsigned char *) dst->u_buffer, dst->uv_stride, dw / 2, dh / 2,
+ temp_area, temp_height, hscale, hratio, vscale, vratio, interlaced);
+
+ if (dw / 2 < (int)dst->uv_width)
+ for (i = 0; i < dst->uv_height; i++)
+ duck_memset(dst->u_buffer + i * dst->uv_stride + dw / 2 - 1, dst->u_buffer[i * dst->uv_stride + dw / 2 - 2], dst->uv_width - dw / 2 + 1);
+
+ if (dh / 2 < (int)dst->uv_height)
+ for (i = dh / 2 - 1; i < (int)dst->y_height / 2; i++)
+ duck_memcpy(dst->u_buffer + i * dst->uv_stride, dst->u_buffer + (dh / 2 - 2)*dst->uv_stride, dst->uv_width);
+
+ Scale2D((unsigned char *) src->v_buffer, src->uv_stride, src->uv_width, src->uv_height,
+ (unsigned char *) dst->v_buffer, dst->uv_stride, dw / 2, dh / 2,
+ temp_area, temp_height, hscale, hratio, vscale, vratio, interlaced);
+
+ if (dw / 2 < (int)dst->uv_width)
+ for (i = 0; i < dst->uv_height; i++)
+ duck_memset(dst->v_buffer + i * dst->uv_stride + dw / 2 - 1, dst->v_buffer[i * dst->uv_stride + dw / 2 - 2], dst->uv_width - dw / 2 + 1);
+
+ if (dh / 2 < (int) dst->uv_height)
+ for (i = dh / 2 - 1; i < (int)dst->y_height / 2; i++)
+ duck_memcpy(dst->v_buffer + i * dst->uv_stride, dst->v_buffer + (dh / 2 - 2)*dst->uv_stride, dst->uv_width);
+}
+/****************************************************************************
+ *
+ * ROUTINE : any_ratio_2d_scale
+ *
+ * INPUTS : SCALE_INSTANCE *si : Pointer to post-processor instance (NOT USED).
+ * const unsigned char *source : Pointer to source image.
+ * unsigned int source_pitch : Stride of source image.
+ * unsigned int source_width : Width of source image.
+ * unsigned int source_height : Height of source image (NOT USED).
+ * unsigned char *dest : Pointer to destination image.
+ * unsigned int dest_pitch : Stride of destination image.
+ * unsigned int dest_width : Width of destination image.
+ * unsigned int dest_height : Height of destination image.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : int: 1 if image scaled, 0 if image could not be scaled.
+ *
+ * FUNCTION : Scale the image with changing apect ratio.
+ *
+ * SPECIAL NOTES : This scaling is a bi-linear scaling. Need to re-work the
+ * whole function for new scaling algorithm.
+ *
+ ****************************************************************************/
+static
+int any_ratio_2d_scale
+(
+ SCALE_VARS *si,
+ const unsigned char *source,
+ int source_pitch,
+ unsigned int source_width,
+ unsigned int source_height,
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width,
+ unsigned int dest_height
+) {
+ unsigned int i, k;
+ unsigned int src_band_height = 0;
+ unsigned int dest_band_height = 0;
+
+ /* suggested scale factors */
+ int hs = si->HScale;
+ int hr = si->HRatio;
+ int vs = si->VScale;
+ int vr = si->VRatio;
+
+ /* assume the ratios are scalable instead of should be centered */
+ int ratio_scalable = 1;
+
+ const unsigned char *source_base = ((source_pitch >= 0) ? source : (source + ((source_height - 1) * source_pitch)));
+ const unsigned char *line_src;
+
+ void (*horiz_line_scale)(const unsigned char *, unsigned int, unsigned char *, unsigned int) = NULL;
+ void (*vert_band_scale)(unsigned char *, unsigned int, unsigned int) = NULL;
+ void (*last_vert_band_scale)(unsigned char *, unsigned int, unsigned int) = NULL;
+
+ (void) si;
+
+ /* find out the ratio for each direction */
+ switch (hr * 30 / hs) {
+ case 24:
+ /* 4-5 Scale in Width direction */
+ horiz_line_scale = vp8_horizontal_line_4_5_scale;
+ break;
+ case 22:
+ /* 3-4 Scale in Width direction */
+ horiz_line_scale = vp8_horizontal_line_3_4_scale;
+ break;
+
+ case 20:
+ /* 4-5 Scale in Width direction */
+ horiz_line_scale = vp8_horizontal_line_2_3_scale;
+ break;
+ case 18:
+ /* 3-5 Scale in Width direction */
+ horiz_line_scale = vp8_horizontal_line_3_5_scale;
+ break;
+ case 15:
+ /* 1-2 Scale in Width direction */
+ horiz_line_scale = vp8_horizontal_line_1_2_scale;
+ break;
+ case 30:
+ /* no scale in Width direction */
+ horiz_line_scale = horizontal_line_copy;
+ break;
+ default:
+ /* The ratio is not acceptable now */
+ /* throw("The ratio is not acceptable for now!"); */
+ ratio_scalable = 0;
+ break;
+ }
+
+ switch (vr * 30 / vs) {
+ case 24:
+ /* 4-5 Scale in vertical direction */
+ vert_band_scale = vp8_vertical_band_4_5_scale;
+ last_vert_band_scale = vp8_last_vertical_band_4_5_scale;
+ src_band_height = 4;
+ dest_band_height = 5;
+ break;
+ case 22:
+ /* 3-4 Scale in vertical direction */
+ vert_band_scale = vp8_vertical_band_3_4_scale;
+ last_vert_band_scale = vp8_last_vertical_band_3_4_scale;
+ src_band_height = 3;
+ dest_band_height = 4;
+ break;
+ case 20:
+ /* 2-3 Scale in vertical direction */
+ vert_band_scale = vp8_vertical_band_2_3_scale;
+ last_vert_band_scale = vp8_last_vertical_band_2_3_scale;
+ src_band_height = 2;
+ dest_band_height = 3;
+ break;
+ case 18:
+ /* 3-5 Scale in vertical direction */
+ vert_band_scale = vp8_vertical_band_3_5_scale;
+ last_vert_band_scale = vp8_last_vertical_band_3_5_scale;
+ src_band_height = 3;
+ dest_band_height = 5;
+ break;
+ case 15:
+ /* 1-2 Scale in vertical direction */
+ vert_band_scale = vp8_vertical_band_1_2_scale;
+ last_vert_band_scale = vp8_last_vertical_band_1_2_scale;
+ src_band_height = 1;
+ dest_band_height = 2;
+ break;
+ case 30:
+ /* no scale in Width direction */
+ vert_band_scale = null_scale;
+ last_vert_band_scale = null_scale;
+ src_band_height = 4;
+ dest_band_height = 4;
+ break;
+ default:
+ /* The ratio is not acceptable now */
+ /* throw("The ratio is not acceptable for now!"); */
+ ratio_scalable = 0;
+ break;
+ }
+
+ if (ratio_scalable == 0)
+ return ratio_scalable;
+
+ horiz_line_scale(source, source_width, dest, dest_width);
+
+ /* except last band */
+ for (k = 0; k < (dest_height + dest_band_height - 1) / dest_band_height - 1; k++) {
+ /* scale one band horizontally */
+ for (i = 1; i < src_band_height; i++) {
+ /* Trap case where we could read off the base of the source buffer */
+ line_src = source + i * source_pitch;
+
+ if (line_src < source_base)
+ line_src = source_base;
+
+ horiz_line_scale(line_src, source_width,
+ dest + i * dest_pitch, dest_width);
+ }
+
+ /* first line of next band */
+ /* Trap case where we could read off the base of the source buffer */
+ line_src = source + src_band_height * source_pitch;
+
+ if (line_src < source_base)
+ line_src = source_base;
+
+ horiz_line_scale(line_src, source_width,
+ dest + dest_band_height * dest_pitch,
+ dest_width);
+
+ /* Vertical scaling is in place */
+ vert_band_scale(dest, dest_pitch, dest_width);
+
+ /* Next band... */
+ source += src_band_height * source_pitch;
+ dest += dest_band_height * dest_pitch;
+ }
+
+ /* scale one band horizontally */
+ for (i = 1; i < src_band_height; i++) {
+ /* Trap case where we could read off the base of the source buffer */
+ line_src = source + i * source_pitch;
+
+ if (line_src < source_base)
+ line_src = source_base;
+
+ horiz_line_scale(line_src, source_width,
+ dest + i * dest_pitch,
+ dest_width);
+ }
+
+ /* Vertical scaling is in place */
+ last_vert_band_scale(dest, dest_pitch, dest_width);
+
+ return ratio_scalable;
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : any_ratio_frame_scale
+ *
+ * INPUTS : SCALE_INSTANCE *si : Pointer to post-processor instance (NOT USED).
+ * unsigned char *frame_buffer : Pointer to source image.
+ * int YOffset : Offset from start of buffer to Y samples.
+ * int UVOffset : Offset from start of buffer to UV samples.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : int: 1 if image scaled, 0 if image could not be scaled.
+ *
+ * FUNCTION : Scale the image with changing apect ratio.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+int any_ratio_frame_scale(SCALE_VARS *scale_vars, int YOffset, int UVOffset) {
+ int i;
+ int ew;
+ int eh;
+
+ /* suggested scale factors */
+ int hs = scale_vars->HScale;
+ int hr = scale_vars->HRatio;
+ int vs = scale_vars->VScale;
+ int vr = scale_vars->VRatio;
+
+ int ratio_scalable = 1;
+
+ int sw = (scale_vars->expanded_frame_width * hr + hs - 1) / hs;
+ int sh = (scale_vars->expanded_frame_height * vr + vs - 1) / vs;
+ int dw = scale_vars->expanded_frame_width;
+ int dh = scale_vars->expanded_frame_height;
+ YV12_BUFFER_CONFIG *src_yuv_config = scale_vars->src_yuv_config;
+ YV12_BUFFER_CONFIG *dst_yuv_config = scale_vars->dst_yuv_config;
+
+ if (hr == 3)
+ ew = (sw + 2) / 3 * 3 * hs / hr;
+ else
+ ew = (sw + 7) / 8 * 8 * hs / hr;
+
+ if (vr == 3)
+ eh = (sh + 2) / 3 * 3 * vs / vr;
+ else
+ eh = (sh + 7) / 8 * 8 * vs / vr;
+
+ ratio_scalable = any_ratio_2d_scale(scale_vars,
+ (const unsigned char *)src_yuv_config->y_buffer,
+ src_yuv_config->y_stride, sw, sh,
+ (unsigned char *) dst_yuv_config->y_buffer + YOffset,
+ dst_yuv_config->y_stride, dw, dh);
+
+ for (i = 0; i < eh; i++)
+ duck_memset(dst_yuv_config->y_buffer + YOffset + i * dst_yuv_config->y_stride + dw, 0, ew - dw);
+
+ for (i = dh; i < eh; i++)
+ duck_memset(dst_yuv_config->y_buffer + YOffset + i * dst_yuv_config->y_stride, 0, ew);
+
+ if (ratio_scalable == 0)
+ return ratio_scalable;
+
+ sw = (sw + 1) >> 1;
+ sh = (sh + 1) >> 1;
+ dw = (dw + 1) >> 1;
+ dh = (dh + 1) >> 1;
+
+ any_ratio_2d_scale(scale_vars,
+ (const unsigned char *)src_yuv_config->u_buffer,
+ src_yuv_config->y_stride / 2, sw, sh,
+ (unsigned char *)dst_yuv_config->u_buffer + UVOffset,
+ dst_yuv_config->uv_stride, dw, dh);
+
+ any_ratio_2d_scale(scale_vars,
+ (const unsigned char *)src_yuv_config->v_buffer,
+ src_yuv_config->y_stride / 2, sw, sh,
+ (unsigned char *)dst_yuv_config->v_buffer + UVOffset,
+ dst_yuv_config->uv_stride, dw, dh);
+
+ return ratio_scalable;
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : center_image
+ *
+ * INPUTS : SCALE_INSTANCE *si : Pointer to post-processor instance.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Centers the image without scaling in the output buffer.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static void
+center_image(YV12_BUFFER_CONFIG *src_yuv_config, YV12_BUFFER_CONFIG *dst_yuv_config) {
+ int i;
+ int row_offset, col_offset;
+ unsigned char *src_data_pointer;
+ unsigned char *dst_data_pointer;
+
+ /* center values */
+ row_offset = (dst_yuv_config->y_height - src_yuv_config->y_height) / 2;
+ col_offset = (dst_yuv_config->y_width - src_yuv_config->y_width) / 2;
+
+ /* Y's */
+ src_data_pointer = src_yuv_config->y_buffer;
+ dst_data_pointer = (unsigned char *)dst_yuv_config->y_buffer + (row_offset * dst_yuv_config->y_stride) + col_offset;
+
+ for (i = 0; i < src_yuv_config->y_height; i++) {
+ duck_memcpy(dst_data_pointer, src_data_pointer, src_yuv_config->y_width);
+ dst_data_pointer += dst_yuv_config->y_stride;
+ src_data_pointer += src_yuv_config->y_stride;
+ }
+
+ row_offset /= 2;
+ col_offset /= 2;
+
+ /* U's */
+ src_data_pointer = src_yuv_config->u_buffer;
+ dst_data_pointer = (unsigned char *)dst_yuv_config->u_buffer + (row_offset * dst_yuv_config->uv_stride) + col_offset;
+
+ for (i = 0; i < src_yuv_config->uv_height; i++) {
+ duck_memcpy(dst_data_pointer, src_data_pointer, src_yuv_config->uv_width);
+ dst_data_pointer += dst_yuv_config->uv_stride;
+ src_data_pointer += src_yuv_config->uv_stride;
+ }
+
+ /* V's */
+ src_data_pointer = src_yuv_config->v_buffer;
+ dst_data_pointer = (unsigned char *)dst_yuv_config->v_buffer + (row_offset * dst_yuv_config->uv_stride) + col_offset;
+
+ for (i = 0; i < src_yuv_config->uv_height; i++) {
+ duck_memcpy(dst_data_pointer, src_data_pointer, src_yuv_config->uv_width);
+ dst_data_pointer += dst_yuv_config->uv_stride;
+ src_data_pointer += src_yuv_config->uv_stride;
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : scale_or_center
+ *
+ * INPUTS : SCALE_INSTANCE *si : Pointer to post-processor instance.
+ *
+ *
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Decides to scale or center image in scale buffer for blit
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+void
+vp8_yv12_scale_or_center
+(
+ YV12_BUFFER_CONFIG *src_yuv_config,
+ YV12_BUFFER_CONFIG *dst_yuv_config,
+ int expanded_frame_width,
+ int expanded_frame_height,
+ int scaling_mode,
+ int HScale,
+ int HRatio,
+ int VScale,
+ int VRatio
+) {
+ /*if ( ppi->post_processing_level )
+ update_umvborder ( ppi, frame_buffer );*/
+
+
+ switch (scaling_mode) {
+ case SCALE_TO_FIT:
+ case MAINTAIN_ASPECT_RATIO: {
+ SCALE_VARS scale_vars;
+ /* center values */
+#if 1
+ int row = (dst_yuv_config->y_height - expanded_frame_height) / 2;
+ int col = (dst_yuv_config->y_width - expanded_frame_width) / 2;
+ /*int YOffset = row * dst_yuv_config->y_width + col;
+ int UVOffset = (row>>1) * dst_yuv_config->uv_width + (col>>1);*/
+ int YOffset = row * dst_yuv_config->y_stride + col;
+ int UVOffset = (row >> 1) * dst_yuv_config->uv_stride + (col >> 1);
+#else
+ int row = (src_yuv_config->y_height - expanded_frame_height) / 2;
+ int col = (src_yuv_config->y_width - expanded_frame_width) / 2;
+ int YOffset = row * src_yuv_config->y_width + col;
+ int UVOffset = (row >> 1) * src_yuv_config->uv_width + (col >> 1);
+#endif
+
+ scale_vars.dst_yuv_config = dst_yuv_config;
+ scale_vars.src_yuv_config = src_yuv_config;
+ scale_vars.HScale = HScale;
+ scale_vars.HRatio = HRatio;
+ scale_vars.VScale = VScale;
+ scale_vars.VRatio = VRatio;
+ scale_vars.expanded_frame_width = expanded_frame_width;
+ scale_vars.expanded_frame_height = expanded_frame_height;
+
+ /* perform center and scale */
+ any_ratio_frame_scale(&scale_vars, YOffset, UVOffset);
+
+ break;
+ }
+ case CENTER:
+ center_image(src_yuv_config, dst_yuv_config);
+ break;
+
+ default:
+ break;
+ }
+}
diff --git a/libvpx/vpx_scale/generic/yv12config.c b/libvpx/vpx_scale/generic/yv12config.c
new file mode 100644
index 0000000..4cb2a41
--- /dev/null
+++ b/libvpx/vpx_scale/generic/yv12config.c
@@ -0,0 +1,92 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_scale/yv12config.h"
+#include "vpx_mem/vpx_mem.h"
+
+/****************************************************************************
+* Exports
+****************************************************************************/
+
+/****************************************************************************
+ *
+ ****************************************************************************/
+int
+vp8_yv12_de_alloc_frame_buffer(YV12_BUFFER_CONFIG *ybf) {
+ if (ybf) {
+ vpx_free(ybf->buffer_alloc);
+
+ /* buffer_alloc isn't accessed by most functions. Rather y_buffer,
+ u_buffer and v_buffer point to buffer_alloc and are used. Clear out
+ all of this so that a freed pointer isn't inadvertently used */
+ vpx_memset(ybf, 0, sizeof(YV12_BUFFER_CONFIG));
+ } else {
+ return -1;
+ }
+
+ return 0;
+}
+
+/****************************************************************************
+ *
+ ****************************************************************************/
+int
+vp8_yv12_alloc_frame_buffer(YV12_BUFFER_CONFIG *ybf, int width, int height, int border) {
+ /*NOTE:*/
+
+ if (ybf) {
+ int y_stride = ((width + 2 * border) + 31) & ~31;
+ int yplane_size = (height + 2 * border) * y_stride;
+ int uv_width = width >> 1;
+ int uv_height = height >> 1;
+ /** There is currently a bunch of code which assumes
+ * uv_stride == y_stride/2, so enforce this here. */
+ int uv_stride = y_stride >> 1;
+ int uvplane_size = (uv_height + border) * uv_stride;
+
+ vp8_yv12_de_alloc_frame_buffer(ybf);
+
+ /** Only support allocating buffers that have a height and width that
+ * are multiples of 16, and a border that's a multiple of 32.
+ * The border restriction is required to get 16-byte alignment of the
+ * start of the chroma rows without intoducing an arbitrary gap
+ * between planes, which would break the semantics of things like
+ * vpx_img_set_rect(). */
+ if ((width & 0xf) | (height & 0xf) | (border & 0x1f))
+ return -3;
+
+ ybf->y_width = width;
+ ybf->y_height = height;
+ ybf->y_stride = y_stride;
+
+ ybf->uv_width = uv_width;
+ ybf->uv_height = uv_height;
+ ybf->uv_stride = uv_stride;
+
+ ybf->border = border;
+ ybf->frame_size = yplane_size + 2 * uvplane_size;
+
+ ybf->buffer_alloc = (unsigned char *) vpx_memalign(32, ybf->frame_size);
+
+ if (ybf->buffer_alloc == NULL)
+ return -1;
+
+ ybf->y_buffer = ybf->buffer_alloc + (border * y_stride) + border;
+ ybf->u_buffer = ybf->buffer_alloc + yplane_size + (border / 2 * uv_stride) + border / 2;
+ ybf->v_buffer = ybf->buffer_alloc + yplane_size + uvplane_size + (border / 2 * uv_stride) + border / 2;
+
+ ybf->corrupted = 0; /* assume not currupted by errors */
+ } else {
+ return -2;
+ }
+
+ return 0;
+}
diff --git a/libvpx/vpx_scale/generic/yv12extend.c b/libvpx/vpx_scale/generic/yv12extend.c
new file mode 100644
index 0000000..247078c
--- /dev/null
+++ b/libvpx/vpx_scale/generic/yv12extend.c
@@ -0,0 +1,263 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vpx_scale/yv12config.h"
+#include "vpx_mem/vpx_mem.h"
+#include "vpx_scale/vpxscale.h"
+
+/****************************************************************************
+* Exports
+****************************************************************************/
+
+/****************************************************************************
+ *
+ ****************************************************************************/
+void
+vp8_yv12_extend_frame_borders_c(YV12_BUFFER_CONFIG *ybf) {
+ int i;
+ unsigned char *src_ptr1, *src_ptr2;
+ unsigned char *dest_ptr1, *dest_ptr2;
+
+ unsigned int Border;
+ int plane_stride;
+ int plane_height;
+ int plane_width;
+
+ /***********/
+ /* Y Plane */
+ /***********/
+ Border = ybf->border;
+ plane_stride = ybf->y_stride;
+ plane_height = ybf->y_height;
+ plane_width = ybf->y_width;
+
+ /* copy the left and right most columns out */
+ src_ptr1 = ybf->y_buffer;
+ src_ptr2 = src_ptr1 + plane_width - 1;
+ dest_ptr1 = src_ptr1 - Border;
+ dest_ptr2 = src_ptr2 + 1;
+
+ for (i = 0; i < plane_height; i++) {
+ vpx_memset(dest_ptr1, src_ptr1[0], Border);
+ vpx_memset(dest_ptr2, src_ptr2[0], Border);
+ src_ptr1 += plane_stride;
+ src_ptr2 += plane_stride;
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+ /* Now copy the top and bottom source lines into each line of the respective borders */
+ src_ptr1 = ybf->y_buffer - Border;
+ src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
+ dest_ptr1 = src_ptr1 - (Border * plane_stride);
+ dest_ptr2 = src_ptr2 + plane_stride;
+
+ for (i = 0; i < (int)Border; i++) {
+ vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
+ vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+
+ /***********/
+ /* U Plane */
+ /***********/
+ plane_stride = ybf->uv_stride;
+ plane_height = ybf->uv_height;
+ plane_width = ybf->uv_width;
+ Border /= 2;
+
+ /* copy the left and right most columns out */
+ src_ptr1 = ybf->u_buffer;
+ src_ptr2 = src_ptr1 + plane_width - 1;
+ dest_ptr1 = src_ptr1 - Border;
+ dest_ptr2 = src_ptr2 + 1;
+
+ for (i = 0; i < plane_height; i++) {
+ vpx_memset(dest_ptr1, src_ptr1[0], Border);
+ vpx_memset(dest_ptr2, src_ptr2[0], Border);
+ src_ptr1 += plane_stride;
+ src_ptr2 += plane_stride;
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+ /* Now copy the top and bottom source lines into each line of the respective borders */
+ src_ptr1 = ybf->u_buffer - Border;
+ src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
+ dest_ptr1 = src_ptr1 - (Border * plane_stride);
+ dest_ptr2 = src_ptr2 + plane_stride;
+
+ for (i = 0; i < (int)(Border); i++) {
+ vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
+ vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+ /***********/
+ /* V Plane */
+ /***********/
+
+ /* copy the left and right most columns out */
+ src_ptr1 = ybf->v_buffer;
+ src_ptr2 = src_ptr1 + plane_width - 1;
+ dest_ptr1 = src_ptr1 - Border;
+ dest_ptr2 = src_ptr2 + 1;
+
+ for (i = 0; i < plane_height; i++) {
+ vpx_memset(dest_ptr1, src_ptr1[0], Border);
+ vpx_memset(dest_ptr2, src_ptr2[0], Border);
+ src_ptr1 += plane_stride;
+ src_ptr2 += plane_stride;
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+ /* Now copy the top and bottom source lines into each line of the respective borders */
+ src_ptr1 = ybf->v_buffer - Border;
+ src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
+ dest_ptr1 = src_ptr1 - (Border * plane_stride);
+ dest_ptr2 = src_ptr2 + plane_stride;
+
+ for (i = 0; i < (int)(Border); i++) {
+ vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
+ vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+}
+
+
+static void
+extend_frame_borders_yonly_c(YV12_BUFFER_CONFIG *ybf) {
+ int i;
+ unsigned char *src_ptr1, *src_ptr2;
+ unsigned char *dest_ptr1, *dest_ptr2;
+
+ unsigned int Border;
+ int plane_stride;
+ int plane_height;
+ int plane_width;
+
+ /***********/
+ /* Y Plane */
+ /***********/
+ Border = ybf->border;
+ plane_stride = ybf->y_stride;
+ plane_height = ybf->y_height;
+ plane_width = ybf->y_width;
+
+ /* copy the left and right most columns out */
+ src_ptr1 = ybf->y_buffer;
+ src_ptr2 = src_ptr1 + plane_width - 1;
+ dest_ptr1 = src_ptr1 - Border;
+ dest_ptr2 = src_ptr2 + 1;
+
+ for (i = 0; i < plane_height; i++) {
+ vpx_memset(dest_ptr1, src_ptr1[0], Border);
+ vpx_memset(dest_ptr2, src_ptr2[0], Border);
+ src_ptr1 += plane_stride;
+ src_ptr2 += plane_stride;
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+ /* Now copy the top and bottom source lines into each line of the respective borders */
+ src_ptr1 = ybf->y_buffer - Border;
+ src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
+ dest_ptr1 = src_ptr1 - (Border * plane_stride);
+ dest_ptr2 = src_ptr2 + plane_stride;
+
+ for (i = 0; i < (int)Border; i++) {
+ vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
+ vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
+ dest_ptr1 += plane_stride;
+ dest_ptr2 += plane_stride;
+ }
+
+ plane_stride /= 2;
+ plane_height /= 2;
+ plane_width /= 2;
+ Border /= 2;
+
+}
+
+
+
+/****************************************************************************
+ *
+ * ROUTINE : vp8_yv12_copy_frame
+ *
+ * INPUTS :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies the source image into the destination image and
+ * updates the destination's UMV borders.
+ *
+ * SPECIAL NOTES : The frames are assumed to be identical in size.
+ *
+ ****************************************************************************/
+void
+vp8_yv12_copy_frame_c(YV12_BUFFER_CONFIG *src_ybc,
+ YV12_BUFFER_CONFIG *dst_ybc) {
+ int row;
+ unsigned char *source, *dest;
+
+ source = src_ybc->y_buffer;
+ dest = dst_ybc->y_buffer;
+
+ for (row = 0; row < src_ybc->y_height; row++) {
+ vpx_memcpy(dest, source, src_ybc->y_width);
+ source += src_ybc->y_stride;
+ dest += dst_ybc->y_stride;
+ }
+
+ source = src_ybc->u_buffer;
+ dest = dst_ybc->u_buffer;
+
+ for (row = 0; row < src_ybc->uv_height; row++) {
+ vpx_memcpy(dest, source, src_ybc->uv_width);
+ source += src_ybc->uv_stride;
+ dest += dst_ybc->uv_stride;
+ }
+
+ source = src_ybc->v_buffer;
+ dest = dst_ybc->v_buffer;
+
+ for (row = 0; row < src_ybc->uv_height; row++) {
+ vpx_memcpy(dest, source, src_ybc->uv_width);
+ source += src_ybc->uv_stride;
+ dest += dst_ybc->uv_stride;
+ }
+
+ vp8_yv12_extend_frame_borders_c(dst_ybc);
+}
+
+void vp8_yv12_copy_y_c(YV12_BUFFER_CONFIG *src_ybc,
+ YV12_BUFFER_CONFIG *dst_ybc) {
+ int row;
+ unsigned char *source, *dest;
+
+
+ source = src_ybc->y_buffer;
+ dest = dst_ybc->y_buffer;
+
+ for (row = 0; row < src_ybc->y_height; row++) {
+ vpx_memcpy(dest, source, src_ybc->y_width);
+ source += src_ybc->y_stride;
+ dest += dst_ybc->y_stride;
+ }
+}
diff --git a/libvpx/vpx_scale/generic/yv12extend_generic.h b/libvpx/vpx_scale/generic/yv12extend_generic.h
new file mode 100644
index 0000000..cc2a554
--- /dev/null
+++ b/libvpx/vpx_scale/generic/yv12extend_generic.h
@@ -0,0 +1,25 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef YV12_EXTEND_GENERIC_H
+#define YV12_EXTEND_GENERIC_H
+
+#include "vpx_scale/yv12config.h"
+
+ void vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf);
+
+ /* Copy Y,U,V buffer data from src to dst, filling border of dst as well. */
+ void vp8_yv12_copy_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
+
+ /* Copy Y buffer data from src_ybc to dst_ybc without filling border data */
+ void vp8_yv12_copy_y_c(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
+
+#endif /* YV12_EXTEND_GENERIC_H */
diff --git a/libvpx/vpx_scale/include/generic/vpxscale_arbitrary.h b/libvpx/vpx_scale/include/generic/vpxscale_arbitrary.h
new file mode 100644
index 0000000..c535252
--- /dev/null
+++ b/libvpx/vpx_scale/include/generic/vpxscale_arbitrary.h
@@ -0,0 +1,55 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef __VPX_SCALE_ARBITRARY_H__
+#define __VPX_SCALE_ARBITRARY_H__
+
+#include "vpx_scale/yv12config.h"
+
+typedef struct {
+ int in_width;
+ int in_height;
+
+ int out_width;
+ int out_height;
+ int max_usable_out_width;
+
+ // numerator for the width and height
+ int nw;
+ int nh;
+ int nh_uv;
+
+ // output to input correspondance array
+ short *l_w;
+ short *l_h;
+ short *l_h_uv;
+
+ // polyphase coefficients
+ short *c_w;
+ short *c_h;
+ short *c_h_uv;
+
+ // buffer for horizontal filtering.
+ unsigned char *hbuf;
+ unsigned char *hbuf_uv;
+} BICUBIC_SCALER_STRUCT;
+
+int bicubic_coefficient_setup(int in_width, int in_height, int out_width, int out_height);
+int bicubic_scale(int in_width, int in_height, int in_stride,
+ int out_width, int out_height, int out_stride,
+ unsigned char *input_image, unsigned char *output_image);
+void bicubic_scale_frame_reset();
+void bicubic_scale_frame(YV12_BUFFER_CONFIG *src, YV12_BUFFER_CONFIG *dst,
+ int new_width, int new_height);
+void bicubic_coefficient_init();
+void bicubic_coefficient_destroy();
+
+#endif /* __VPX_SCALE_ARBITRARY_H__ */
diff --git a/vpx_scale/include/generic/vpxscale_depricated.h b/libvpx/vpx_scale/include/generic/vpxscale_depricated.h
similarity index 100%
rename from vpx_scale/include/generic/vpxscale_depricated.h
rename to libvpx/vpx_scale/include/generic/vpxscale_depricated.h
diff --git a/libvpx/vpx_scale/scale_mode.h b/libvpx/vpx_scale/scale_mode.h
new file mode 100644
index 0000000..5581385
--- /dev/null
+++ b/libvpx/vpx_scale/scale_mode.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/****************************************************************************
+*
+*****************************************************************************
+*/
+
+#ifndef SCALE_MODE_H
+#define SCALE_MODE_H
+
+typedef enum {
+ MAINTAIN_ASPECT_RATIO = 0x0,
+ SCALE_TO_FIT = 0x1,
+ CENTER = 0x2,
+ OTHER = 0x3
+} SCALE_MODE;
+
+
+#endif
diff --git a/libvpx/vpx_scale/vpx_scale.mk b/libvpx/vpx_scale/vpx_scale.mk
new file mode 100644
index 0000000..dc89478
--- /dev/null
+++ b/libvpx/vpx_scale/vpx_scale.mk
@@ -0,0 +1,18 @@
+SCALE_SRCS-yes += vpx_scale.mk
+SCALE_SRCS-yes += scale_mode.h
+SCALE_SRCS-yes += yv12config.h
+SCALE_SRCS-yes += vpxscale.h
+SCALE_SRCS-yes += generic/vpxscale.c
+SCALE_SRCS-yes += generic/yv12config.c
+SCALE_SRCS-yes += generic/yv12extend.c
+SCALE_SRCS-yes += generic/yv12extend_generic.h
+SCALE_SRCS-$(CONFIG_SPATIAL_RESAMPLING) += generic/gen_scalers.c
+
+#neon
+SCALE_SRCS-$(HAVE_NEON) += arm/neon/vp8_vpxyv12_copyframe_func_neon$(ASM)
+SCALE_SRCS-$(HAVE_NEON) += arm/neon/vp8_vpxyv12_copy_y_neon$(ASM)
+SCALE_SRCS-$(HAVE_NEON) += arm/neon/vp8_vpxyv12_copysrcframe_func_neon$(ASM)
+SCALE_SRCS-$(HAVE_NEON) += arm/neon/vp8_vpxyv12_extendframeborders_neon$(ASM)
+SCALE_SRCS-$(HAVE_NEON) += arm/neon/yv12extend_arm.c
+
+SCALE_SRCS-no += $(SCALE_SRCS_REMOVE-yes)
diff --git a/libvpx/vpx_scale/vpxscale.h b/libvpx/vpx_scale/vpxscale.h
new file mode 100644
index 0000000..3c2194d
--- /dev/null
+++ b/libvpx/vpx_scale/vpxscale.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef VPXSCALE_H
+#define VPXSCALE_H
+
+#include "vpx_scale/yv12config.h"
+
+extern void vp8_yv12_scale_or_center(YV12_BUFFER_CONFIG *src_yuv_config,
+ YV12_BUFFER_CONFIG *dst_yuv_config,
+ int expanded_frame_width,
+ int expanded_frame_height,
+ int scaling_mode,
+ int HScale,
+ int HRatio,
+ int VScale,
+ int VRatio);
+
+extern void vp8_scale_frame(YV12_BUFFER_CONFIG *src,
+ YV12_BUFFER_CONFIG *dst,
+ unsigned char *temp_area,
+ unsigned char temp_height,
+ unsigned int hscale,
+ unsigned int hratio,
+ unsigned int vscale,
+ unsigned int vratio,
+ unsigned int interlaced);
+
+#endif
diff --git a/libvpx/vpx_scale/win32/scaleopt.c b/libvpx/vpx_scale/win32/scaleopt.c
new file mode 100644
index 0000000..2d96cc7
--- /dev/null
+++ b/libvpx/vpx_scale/win32/scaleopt.c
@@ -0,0 +1,1718 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/****************************************************************************
+*
+* Module Title : scaleopt.cpp
+*
+* Description : Optimized scaling functions
+*
+****************************************************************************/
+#include "pragmas.h"
+
+
+
+/****************************************************************************
+* Module Statics
+****************************************************************************/
+__declspec(align(16)) const static unsigned short one_fifth[] = { 51, 51, 51, 51 };
+__declspec(align(16)) const static unsigned short two_fifths[] = { 102, 102, 102, 102 };
+__declspec(align(16)) const static unsigned short three_fifths[] = { 154, 154, 154, 154 };
+__declspec(align(16)) const static unsigned short four_fifths[] = { 205, 205, 205, 205 };
+__declspec(align(16)) const static unsigned short round_values[] = { 128, 128, 128, 128 };
+__declspec(align(16)) const static unsigned short four_ones[] = { 1, 1, 1, 1};
+__declspec(align(16)) const static unsigned short const45_2[] = {205, 154, 102, 51 };
+__declspec(align(16)) const static unsigned short const45_1[] = { 51, 102, 154, 205 };
+__declspec(align(16)) const static unsigned char mask45[] = { 0, 0, 0, 0, 0, 0, 255, 0};
+__declspec(align(16)) const static unsigned short const35_2[] = { 154, 51, 205, 102 };
+__declspec(align(16)) const static unsigned short const35_1[] = { 102, 205, 51, 154 };
+
+
+
+#include "vpx_scale/vpxscale.h"
+#include "vpx_mem/vpx_mem.h"
+
+/****************************************************************************
+ *
+ * ROUTINE : horizontal_line_3_5_scale_mmx
+ *
+ * INPUTS : const unsigned char *source :
+ * unsigned int source_width :
+ * unsigned char *dest :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 3 to 5 up-scaling of a horizontal line of pixels.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+void horizontal_line_3_5_scale_mmx
+(
+ const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width
+) {
+ (void) dest_width;
+
+ __asm {
+
+ push ebx
+
+ mov esi, source
+ mov edi, dest
+
+ mov ecx, source_width
+ lea edx, [esi+ecx-3];
+
+ movq mm5, const35_1 // mm5 = 66 xx cd xx 33 xx 9a xx
+ movq mm6, const35_2 // mm6 = 9a xx 33 xx cd xx 66 xx
+
+ movq mm4, round_values // mm4 = 80 xx 80 xx 80 xx 80 xx
+ pxor mm7, mm7 // clear mm7
+
+ horiz_line_3_5_loop:
+
+ mov eax, DWORD PTR [esi] // eax = 00 01 02 03
+ mov ebx, eax
+
+ and ebx, 0xffff00 // ebx = xx 01 02 xx
+ mov ecx, eax // ecx = 00 01 02 03
+
+ and eax, 0xffff0000 // eax = xx xx 02 03
+ xor ecx, eax // ecx = 00 01 xx xx
+
+ shr ebx, 8 // ebx = 01 02 xx xx
+ or eax, ebx // eax = 01 02 02 03
+
+ shl ebx, 16 // ebx = xx xx 01 02
+ movd mm1, eax // mm1 = 01 02 02 03 xx xx xx xx
+
+ or ebx, ecx // ebx = 00 01 01 02
+ punpcklbw mm1, mm7 // mm1 = 01 xx 02 xx 02 xx 03 xx
+
+ movd mm0, ebx // mm0 = 00 01 01 02
+ pmullw mm1, mm6 //
+
+ punpcklbw mm0, mm7 // mm0 = 00 xx 01 xx 01 xx 02 xx
+ pmullw mm0, mm5 //
+
+ mov [edi], ebx // writeoutput 00 xx xx xx
+ add esi, 3
+
+ add edi, 5
+ paddw mm0, mm1
+
+ paddw mm0, mm4
+ psrlw mm0, 8
+
+ cmp esi, edx
+ packuswb mm0, mm7
+
+ movd DWORD Ptr [edi-4], mm0
+ jl horiz_line_3_5_loop
+
+// Exit:
+ mov eax, DWORD PTR [esi] // eax = 00 01 02 03
+ mov ebx, eax
+
+ and ebx, 0xffff00 // ebx = xx 01 02 xx
+ mov ecx, eax // ecx = 00 01 02 03
+
+ and eax, 0xffff0000 // eax = xx xx 02 03
+ xor ecx, eax // ecx = 00 01 xx xx
+
+ shr ebx, 8 // ebx = 01 02 xx xx
+ or eax, ebx // eax = 01 02 02 03
+
+ shl eax, 8 // eax = xx 01 02 02
+ and eax, 0xffff0000 // eax = xx xx 02 02
+
+ or eax, ebx // eax = 01 02 02 02
+
+ shl ebx, 16 // ebx = xx xx 01 02
+ movd mm1, eax // mm1 = 01 02 02 02 xx xx xx xx
+
+ or ebx, ecx // ebx = 00 01 01 02
+ punpcklbw mm1, mm7 // mm1 = 01 xx 02 xx 02 xx 02 xx
+
+ movd mm0, ebx // mm0 = 00 01 01 02
+ pmullw mm1, mm6 //
+
+ punpcklbw mm0, mm7 // mm0 = 00 xx 01 xx 01 xx 02 xx
+ pmullw mm0, mm5 //
+
+ mov [edi], ebx // writeoutput 00 xx xx xx
+ paddw mm0, mm1
+
+ paddw mm0, mm4
+ psrlw mm0, 8
+
+ packuswb mm0, mm7
+ movd DWORD Ptr [edi+1], mm0
+
+ pop ebx
+
+ }
+
+}
+
+
+/****************************************************************************
+ *
+ * ROUTINE : horizontal_line_4_5_scale_mmx
+ *
+ * INPUTS : const unsigned char *source :
+ * unsigned int source_width :
+ * unsigned char *dest :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 4 to 5 up-scaling of a horizontal line of pixels.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+void horizontal_line_4_5_scale_mmx
+(
+ const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width
+) {
+ (void)dest_width;
+
+ __asm {
+
+ mov esi, source
+ mov edi, dest
+
+ mov ecx, source_width
+ lea edx, [esi+ecx-8];
+
+ movq mm5, const45_1 // mm5 = 33 xx 66 xx 9a xx cd xx
+ movq mm6, const45_2 // mm6 = cd xx 9a xx 66 xx 33 xx
+
+ movq mm4, round_values // mm4 = 80 xx 80 xx 80 xx 80 xx
+ pxor mm7, mm7 // clear mm7
+
+ horiz_line_4_5_loop:
+
+ movq mm0, QWORD PTR [esi] // mm0 = 00 01 02 03 04 05 06 07
+ movq mm1, QWORD PTR [esi+1]; // mm1 = 01 02 03 04 05 06 07 08
+
+ movq mm2, mm0 // mm2 = 00 01 02 03 04 05 06 07
+ movq mm3, mm1 // mm3 = 01 02 03 04 05 06 07 08
+
+ movd DWORD PTR [edi], mm0 // write output 00 xx xx xx
+ punpcklbw mm0, mm7 // mm0 = 00 xx 01 xx 02 xx 03 xx
+
+ punpcklbw mm1, mm7 // mm1 = 01 xx 02 xx 03 xx 04 xx
+ pmullw mm0, mm5 // 00* 51 01*102 02*154 03*205
+
+ pmullw mm1, mm6 // 01*205 02*154 03*102 04* 51
+ punpckhbw mm2, mm7 // mm2 = 04 xx 05 xx 06 xx 07 xx
+
+ movd DWORD PTR [edi+5], mm2 // write ouput 05 xx xx xx
+ pmullw mm2, mm5 // 04* 51 05*102 06*154 07*205
+
+ punpckhbw mm3, mm7 // mm3 = 05 xx 06 xx 07 xx 08 xx
+ pmullw mm3, mm6 // 05*205 06*154 07*102 08* 51
+
+ paddw mm0, mm1 // added round values
+ paddw mm0, mm4
+
+ psrlw mm0, 8 // output: 01 xx 02 xx 03 xx 04 xx
+ packuswb mm0, mm7
+
+ movd DWORD PTR [edi+1], mm0 // write output 01 02 03 04
+ add edi, 10
+
+ add esi, 8
+ paddw mm2, mm3 //
+
+ paddw mm2, mm4 // added round values
+ cmp esi, edx
+
+ psrlw mm2, 8
+ packuswb mm2, mm7
+
+ movd DWORD PTR [edi-4], mm2 // writeoutput 06 07 08 09
+ jl horiz_line_4_5_loop
+
+// Exit:
+ movq mm0, [esi] // mm0 = 00 01 02 03 04 05 06 07
+ movq mm1, mm0 // mm1 = 00 01 02 03 04 05 06 07
+
+ movq mm2, mm0 // mm2 = 00 01 02 03 04 05 06 07
+ psrlq mm1, 8 // mm1 = 01 02 03 04 05 06 07 00
+
+ movq mm3, mask45 // mm3 = 00 00 00 00 00 00 ff 00
+ pand mm3, mm1 // mm3 = 00 00 00 00 00 00 07 00
+
+ psllq mm3, 8 // mm3 = 00 00 00 00 00 00 00 07
+ por mm1, mm3 // mm1 = 01 02 03 04 05 06 07 07
+
+ movq mm3, mm1
+
+ movd DWORD PTR [edi], mm0 // write output 00 xx xx xx
+ punpcklbw mm0, mm7 // mm0 = 00 xx 01 xx 02 xx 03 xx
+
+ punpcklbw mm1, mm7 // mm1 = 01 xx 02 xx 03 xx 04 xx
+ pmullw mm0, mm5 // 00* 51 01*102 02*154 03*205
+
+ pmullw mm1, mm6 // 01*205 02*154 03*102 04* 51
+ punpckhbw mm2, mm7 // mm2 = 04 xx 05 xx 06 xx 07 xx
+
+ movd DWORD PTR [edi+5], mm2 // write ouput 05 xx xx xx
+ pmullw mm2, mm5 // 04* 51 05*102 06*154 07*205
+
+ punpckhbw mm3, mm7 // mm3 = 05 xx 06 xx 07 xx 08 xx
+ pmullw mm3, mm6 // 05*205 06*154 07*102 07* 51
+
+ paddw mm0, mm1 // added round values
+ paddw mm0, mm4
+
+ psrlw mm0, 8 // output: 01 xx 02 xx 03 xx 04 xx
+ packuswb mm0, mm7 // 01 02 03 04 xx xx xx xx
+
+ movd DWORD PTR [edi+1], mm0 // write output 01 02 03 04
+ paddw mm2, mm3 //
+
+ paddw mm2, mm4 // added round values
+ psrlw mm2, 8
+
+ packuswb mm2, mm7
+ movd DWORD PTR [edi+6], mm2 // writeoutput 06 07 08 09
+
+
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vertical_band_4_5_scale_mmx
+ *
+ * INPUTS : unsigned char *dest :
+ * unsigned int dest_pitch :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 4 to 5 up-scaling of a 4 pixel high band of pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band. The function also has a "C" only
+ * version.
+ *
+ ****************************************************************************/
+static
+void vertical_band_4_5_scale_mmx
+(
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width
+) {
+ __asm {
+
+ mov esi, dest // Get the source and destination pointer
+ mov ecx, dest_pitch // Get the pitch size
+
+ lea edi, [esi+ecx*2] // tow lines below
+ add edi, ecx // three lines below
+
+ pxor mm7, mm7 // clear out mm7
+ mov edx, dest_width // Loop counter
+
+ vs_4_5_loop:
+
+ movq mm0, QWORD ptr [esi] // src[0];
+ movq mm1, QWORD ptr [esi+ecx] // src[1];
+
+ movq mm2, mm0 // Make a copy
+ punpcklbw mm0, mm7 // unpack low to word
+
+ movq mm5, one_fifth
+ punpckhbw mm2, mm7 // unpack high to word
+
+ pmullw mm0, mm5 // a * 1/5
+
+ movq mm3, mm1 // make a copy
+ punpcklbw mm1, mm7 // unpack low to word
+
+ pmullw mm2, mm5 // a * 1/5
+ movq mm6, four_fifths // constan
+
+ movq mm4, mm1 // copy of low b
+ pmullw mm4, mm6 // b * 4/5
+
+ punpckhbw mm3, mm7 // unpack high to word
+ movq mm5, mm3 // copy of high b
+
+ pmullw mm5, mm6 // b * 4/5
+ paddw mm0, mm4 // a * 1/5 + b * 4/5
+
+ paddw mm2, mm5 // a * 1/5 + b * 4/5
+ paddw mm0, round_values // + 128
+
+ paddw mm2, round_values // + 128
+ psrlw mm0, 8
+
+ psrlw mm2, 8
+ packuswb mm0, mm2 // des [1]
+
+ movq QWORD ptr [esi+ecx], mm0 // write des[1]
+ movq mm0, [esi+ecx*2] // mm0 = src[2]
+
+ // mm1, mm3 --- Src[1]
+ // mm0 --- Src[2]
+ // mm7 for unpacking
+
+ movq mm5, two_fifths
+ movq mm2, mm0 // make a copy
+
+ pmullw mm1, mm5 // b * 2/5
+ movq mm6, three_fifths
+
+
+ punpcklbw mm0, mm7 // unpack low to word
+ pmullw mm3, mm5 // b * 2/5
+
+ movq mm4, mm0 // make copy of c
+ punpckhbw mm2, mm7 // unpack high to word
+
+ pmullw mm4, mm6 // c * 3/5
+ movq mm5, mm2
+
+ pmullw mm5, mm6 // c * 3/5
+ paddw mm1, mm4 // b * 2/5 + c * 3/5
+
+ paddw mm3, mm5 // b * 2/5 + c * 3/5
+ paddw mm1, round_values // + 128
+
+ paddw mm3, round_values // + 128
+ psrlw mm1, 8
+
+ psrlw mm3, 8
+ packuswb mm1, mm3 // des[2]
+
+ movq QWORD ptr [esi+ecx*2], mm1 // write des[2]
+ movq mm1, [edi] // mm1=Src[3];
+
+ // mm0, mm2 --- Src[2]
+ // mm1 --- Src[3]
+ // mm6 --- 3/5
+ // mm7 for unpacking
+
+ pmullw mm0, mm6 // c * 3/5
+ movq mm5, two_fifths // mm5 = 2/5
+
+ movq mm3, mm1 // make a copy
+ pmullw mm2, mm6 // c * 3/5
+
+ punpcklbw mm1, mm7 // unpack low
+ movq mm4, mm1 // make a copy
+
+ punpckhbw mm3, mm7 // unpack high
+ pmullw mm4, mm5 // d * 2/5
+
+ movq mm6, mm3 // make a copy
+ pmullw mm6, mm5 // d * 2/5
+
+ paddw mm0, mm4 // c * 3/5 + d * 2/5
+ paddw mm2, mm6 // c * 3/5 + d * 2/5
+
+ paddw mm0, round_values // + 128
+ paddw mm2, round_values // + 128
+
+ psrlw mm0, 8
+ psrlw mm2, 8
+
+ packuswb mm0, mm2 // des[3]
+ movq QWORD ptr [edi], mm0 // write des[3]
+
+ // mm1, mm3 --- Src[3]
+ // mm7 -- cleared for unpacking
+
+ movq mm0, [edi+ecx*2] // mm0, Src[0] of the next group
+
+ movq mm5, four_fifths // mm5 = 4/5
+ pmullw mm1, mm5 // d * 4/5
+
+ movq mm6, one_fifth // mm6 = 1/5
+ movq mm2, mm0 // make a copy
+
+ pmullw mm3, mm5 // d * 4/5
+ punpcklbw mm0, mm7 // unpack low
+
+ pmullw mm0, mm6 // an * 1/5
+ punpckhbw mm2, mm7 // unpack high
+
+ paddw mm1, mm0 // d * 4/5 + an * 1/5
+ pmullw mm2, mm6 // an * 1/5
+
+ paddw mm3, mm2 // d * 4/5 + an * 1/5
+ paddw mm1, round_values // + 128
+
+ paddw mm3, round_values // + 128
+ psrlw mm1, 8
+
+ psrlw mm3, 8
+ packuswb mm1, mm3 // des[4]
+
+ movq QWORD ptr [edi+ecx], mm1 // write des[4]
+
+ add edi, 8
+ add esi, 8
+
+ sub edx, 8
+ jg vs_4_5_loop
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : last_vertical_band_4_5_scale_mmx
+ *
+ * INPUTS : unsigned char *dest :
+ * unsigned int dest_pitch :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : None
+ *
+ * FUNCTION : 4 to 5 up-scaling of the last 4-pixel high band in an image.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band. The function also has an "C" only
+ * version.
+ *
+ ****************************************************************************/
+static
+void last_vertical_band_4_5_scale_mmx
+(
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width
+) {
+ __asm {
+ mov esi, dest // Get the source and destination pointer
+ mov ecx, dest_pitch // Get the pitch size
+
+ lea edi, [esi+ecx*2] // tow lines below
+ add edi, ecx // three lines below
+
+ pxor mm7, mm7 // clear out mm7
+ mov edx, dest_width // Loop counter
+
+ last_vs_4_5_loop:
+
+ movq mm0, QWORD ptr [esi] // src[0];
+ movq mm1, QWORD ptr [esi+ecx] // src[1];
+
+ movq mm2, mm0 // Make a copy
+ punpcklbw mm0, mm7 // unpack low to word
+
+ movq mm5, one_fifth
+ punpckhbw mm2, mm7 // unpack high to word
+
+ pmullw mm0, mm5 // a * 1/5
+
+ movq mm3, mm1 // make a copy
+ punpcklbw mm1, mm7 // unpack low to word
+
+ pmullw mm2, mm5 // a * 1/5
+ movq mm6, four_fifths // constan
+
+ movq mm4, mm1 // copy of low b
+ pmullw mm4, mm6 // b * 4/5
+
+ punpckhbw mm3, mm7 // unpack high to word
+ movq mm5, mm3 // copy of high b
+
+ pmullw mm5, mm6 // b * 4/5
+ paddw mm0, mm4 // a * 1/5 + b * 4/5
+
+ paddw mm2, mm5 // a * 1/5 + b * 4/5
+ paddw mm0, round_values // + 128
+
+ paddw mm2, round_values // + 128
+ psrlw mm0, 8
+
+ psrlw mm2, 8
+ packuswb mm0, mm2 // des [1]
+
+ movq QWORD ptr [esi+ecx], mm0 // write des[1]
+ movq mm0, [esi+ecx*2] // mm0 = src[2]
+
+ // mm1, mm3 --- Src[1]
+ // mm0 --- Src[2]
+ // mm7 for unpacking
+
+ movq mm5, two_fifths
+ movq mm2, mm0 // make a copy
+
+ pmullw mm1, mm5 // b * 2/5
+ movq mm6, three_fifths
+
+
+ punpcklbw mm0, mm7 // unpack low to word
+ pmullw mm3, mm5 // b * 2/5
+
+ movq mm4, mm0 // make copy of c
+ punpckhbw mm2, mm7 // unpack high to word
+
+ pmullw mm4, mm6 // c * 3/5
+ movq mm5, mm2
+
+ pmullw mm5, mm6 // c * 3/5
+ paddw mm1, mm4 // b * 2/5 + c * 3/5
+
+ paddw mm3, mm5 // b * 2/5 + c * 3/5
+ paddw mm1, round_values // + 128
+
+ paddw mm3, round_values // + 128
+ psrlw mm1, 8
+
+ psrlw mm3, 8
+ packuswb mm1, mm3 // des[2]
+
+ movq QWORD ptr [esi+ecx*2], mm1 // write des[2]
+ movq mm1, [edi] // mm1=Src[3];
+
+ movq QWORD ptr [edi+ecx], mm1 // write des[4];
+
+ // mm0, mm2 --- Src[2]
+ // mm1 --- Src[3]
+ // mm6 --- 3/5
+ // mm7 for unpacking
+
+ pmullw mm0, mm6 // c * 3/5
+ movq mm5, two_fifths // mm5 = 2/5
+
+ movq mm3, mm1 // make a copy
+ pmullw mm2, mm6 // c * 3/5
+
+ punpcklbw mm1, mm7 // unpack low
+ movq mm4, mm1 // make a copy
+
+ punpckhbw mm3, mm7 // unpack high
+ pmullw mm4, mm5 // d * 2/5
+
+ movq mm6, mm3 // make a copy
+ pmullw mm6, mm5 // d * 2/5
+
+ paddw mm0, mm4 // c * 3/5 + d * 2/5
+ paddw mm2, mm6 // c * 3/5 + d * 2/5
+
+ paddw mm0, round_values // + 128
+ paddw mm2, round_values // + 128
+
+ psrlw mm0, 8
+ psrlw mm2, 8
+
+ packuswb mm0, mm2 // des[3]
+ movq QWORD ptr [edi], mm0 // write des[3]
+
+ // mm1, mm3 --- Src[3]
+ // mm7 -- cleared for unpacking
+ add edi, 8
+ add esi, 8
+
+ sub edx, 8
+ jg last_vs_4_5_loop
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vertical_band_3_5_scale_mmx
+ *
+ * INPUTS : unsigned char *dest :
+ * unsigned int dest_pitch :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 3 to 5 up-scaling of a 3-pixel high band of pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band. The function also has an "C" only
+ * version.
+ *
+ ****************************************************************************/
+static
+void vertical_band_3_5_scale_mmx
+(
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width
+) {
+ __asm {
+ mov esi, dest // Get the source and destination pointer
+ mov ecx, dest_pitch // Get the pitch size
+
+ lea edi, [esi+ecx*2] // tow lines below
+ add edi, ecx // three lines below
+
+ pxor mm7, mm7 // clear out mm7
+ mov edx, dest_width // Loop counter
+
+ vs_3_5_loop:
+
+ movq mm0, QWORD ptr [esi] // src[0];
+ movq mm1, QWORD ptr [esi+ecx] // src[1];
+
+ movq mm2, mm0 // Make a copy
+ punpcklbw mm0, mm7 // unpack low to word
+
+ movq mm5, two_fifths // mm5 = 2/5
+ punpckhbw mm2, mm7 // unpack high to word
+
+ pmullw mm0, mm5 // a * 2/5
+
+ movq mm3, mm1 // make a copy
+ punpcklbw mm1, mm7 // unpack low to word
+
+ pmullw mm2, mm5 // a * 2/5
+ movq mm6, three_fifths // mm6 = 3/5
+
+ movq mm4, mm1 // copy of low b
+ pmullw mm4, mm6 // b * 3/5
+
+ punpckhbw mm3, mm7 // unpack high to word
+ movq mm5, mm3 // copy of high b
+
+ pmullw mm5, mm6 // b * 3/5
+ paddw mm0, mm4 // a * 2/5 + b * 3/5
+
+ paddw mm2, mm5 // a * 2/5 + b * 3/5
+ paddw mm0, round_values // + 128
+
+ paddw mm2, round_values // + 128
+ psrlw mm0, 8
+
+ psrlw mm2, 8
+ packuswb mm0, mm2 // des [1]
+
+ movq QWORD ptr [esi+ecx], mm0 // write des[1]
+ movq mm0, [esi+ecx*2] // mm0 = src[2]
+
+ // mm1, mm3 --- Src[1]
+ // mm0 --- Src[2]
+ // mm7 for unpacking
+
+ movq mm4, mm1 // b low
+ pmullw mm1, four_fifths // b * 4/5 low
+
+ movq mm5, mm3 // b high
+ pmullw mm3, four_fifths // b * 4/5 high
+
+ movq mm2, mm0 // c
+ pmullw mm4, one_fifth // b * 1/5
+
+ punpcklbw mm0, mm7 // c low
+ pmullw mm5, one_fifth // b * 1/5
+
+ movq mm6, mm0 // make copy of c low
+ punpckhbw mm2, mm7 // c high
+
+ pmullw mm6, one_fifth // c * 1/5 low
+ movq mm7, mm2 // make copy of c high
+
+ pmullw mm7, one_fifth // c * 1/5 high
+ paddw mm1, mm6 // b * 4/5 + c * 1/5 low
+
+ paddw mm3, mm7 // b * 4/5 + c * 1/5 high
+ movq mm6, mm0 // make copy of c low
+
+ pmullw mm6, four_fifths // c * 4/5 low
+ movq mm7, mm2 // make copy of c high
+
+ pmullw mm7, four_fifths // c * 4/5 high
+
+ paddw mm4, mm6 // b * 1/5 + c * 4/5 low
+ paddw mm5, mm7 // b * 1/5 + c * 4/5 high
+
+ paddw mm1, round_values // + 128
+ paddw mm3, round_values // + 128
+
+ psrlw mm1, 8
+ psrlw mm3, 8
+
+ packuswb mm1, mm3 // des[2]
+ movq QWORD ptr [esi+ecx*2], mm1 // write des[2]
+
+ paddw mm4, round_values // + 128
+ paddw mm5, round_values // + 128
+
+ psrlw mm4, 8
+ psrlw mm5, 8
+
+ packuswb mm4, mm5 // des[3]
+ movq QWORD ptr [edi], mm4 // write des[3]
+
+ // mm0, mm2 --- Src[3]
+
+ pxor mm7, mm7 // clear mm7 for unpacking
+ movq mm1, [edi+ecx*2] // mm1 = Src[0] of the next group
+
+ movq mm5, three_fifths // mm5 = 3/5
+ pmullw mm0, mm5 // d * 3/5
+
+ movq mm6, two_fifths // mm6 = 2/5
+ movq mm3, mm1 // make a copy
+
+ pmullw mm2, mm5 // d * 3/5
+ punpcklbw mm1, mm7 // unpack low
+
+ pmullw mm1, mm6 // an * 2/5
+ punpckhbw mm3, mm7 // unpack high
+
+ paddw mm0, mm1 // d * 3/5 + an * 2/5
+ pmullw mm3, mm6 // an * 2/5
+
+ paddw mm2, mm3 // d * 3/5 + an * 2/5
+ paddw mm0, round_values // + 128
+
+ paddw mm2, round_values // + 128
+ psrlw mm0, 8
+
+ psrlw mm2, 8
+ packuswb mm0, mm2 // des[4]
+
+ movq QWORD ptr [edi+ecx], mm0 // write des[4]
+
+ add edi, 8
+ add esi, 8
+
+ sub edx, 8
+ jg vs_3_5_loop
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : last_vertical_band_3_5_scale_mmx
+ *
+ * INPUTS : unsigned char *dest :
+ * unsigned int dest_pitch :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 3 to 5 up-scaling of a 3-pixel high band of pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band. The function also has an "C" only
+ * version.
+ *
+ ****************************************************************************/
+static
+void last_vertical_band_3_5_scale_mmx
+(
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width
+) {
+ __asm {
+ mov esi, dest // Get the source and destination pointer
+ mov ecx, dest_pitch // Get the pitch size
+
+ lea edi, [esi+ecx*2] // tow lines below
+ add edi, ecx // three lines below
+
+ pxor mm7, mm7 // clear out mm7
+ mov edx, dest_width // Loop counter
+
+
+ last_vs_3_5_loop:
+
+ movq mm0, QWORD ptr [esi] // src[0];
+ movq mm1, QWORD ptr [esi+ecx] // src[1];
+
+ movq mm2, mm0 // Make a copy
+ punpcklbw mm0, mm7 // unpack low to word
+
+ movq mm5, two_fifths // mm5 = 2/5
+ punpckhbw mm2, mm7 // unpack high to word
+
+ pmullw mm0, mm5 // a * 2/5
+
+ movq mm3, mm1 // make a copy
+ punpcklbw mm1, mm7 // unpack low to word
+
+ pmullw mm2, mm5 // a * 2/5
+ movq mm6, three_fifths // mm6 = 3/5
+
+ movq mm4, mm1 // copy of low b
+ pmullw mm4, mm6 // b * 3/5
+
+ punpckhbw mm3, mm7 // unpack high to word
+ movq mm5, mm3 // copy of high b
+
+ pmullw mm5, mm6 // b * 3/5
+ paddw mm0, mm4 // a * 2/5 + b * 3/5
+
+ paddw mm2, mm5 // a * 2/5 + b * 3/5
+ paddw mm0, round_values // + 128
+
+ paddw mm2, round_values // + 128
+ psrlw mm0, 8
+
+ psrlw mm2, 8
+ packuswb mm0, mm2 // des [1]
+
+ movq QWORD ptr [esi+ecx], mm0 // write des[1]
+ movq mm0, [esi+ecx*2] // mm0 = src[2]
+
+
+
+ // mm1, mm3 --- Src[1]
+ // mm0 --- Src[2]
+ // mm7 for unpacking
+
+ movq mm4, mm1 // b low
+ pmullw mm1, four_fifths // b * 4/5 low
+
+ movq QWORD ptr [edi+ecx], mm0 // write des[4]
+
+ movq mm5, mm3 // b high
+ pmullw mm3, four_fifths // b * 4/5 high
+
+ movq mm2, mm0 // c
+ pmullw mm4, one_fifth // b * 1/5
+
+ punpcklbw mm0, mm7 // c low
+ pmullw mm5, one_fifth // b * 1/5
+
+ movq mm6, mm0 // make copy of c low
+ punpckhbw mm2, mm7 // c high
+
+ pmullw mm6, one_fifth // c * 1/5 low
+ movq mm7, mm2 // make copy of c high
+
+ pmullw mm7, one_fifth // c * 1/5 high
+ paddw mm1, mm6 // b * 4/5 + c * 1/5 low
+
+ paddw mm3, mm7 // b * 4/5 + c * 1/5 high
+ movq mm6, mm0 // make copy of c low
+
+ pmullw mm6, four_fifths // c * 4/5 low
+ movq mm7, mm2 // make copy of c high
+
+ pmullw mm7, four_fifths // c * 4/5 high
+
+ paddw mm4, mm6 // b * 1/5 + c * 4/5 low
+ paddw mm5, mm7 // b * 1/5 + c * 4/5 high
+
+ paddw mm1, round_values // + 128
+ paddw mm3, round_values // + 128
+
+ psrlw mm1, 8
+ psrlw mm3, 8
+
+ packuswb mm1, mm3 // des[2]
+ movq QWORD ptr [esi+ecx*2], mm1 // write des[2]
+
+ paddw mm4, round_values // + 128
+ paddw mm5, round_values // + 128
+
+ psrlw mm4, 8
+ psrlw mm5, 8
+
+ packuswb mm4, mm5 // des[3]
+ movq QWORD ptr [edi], mm4 // write des[3]
+
+ // mm0, mm2 --- Src[3]
+
+ add edi, 8
+ add esi, 8
+
+ sub edx, 8
+ jg last_vs_3_5_loop
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : vertical_band_1_2_scale_mmx
+ *
+ * INPUTS : unsigned char *dest :
+ * unsigned int dest_pitch :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 1 to 2 up-scaling of a band of pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band. The function also has an "C" only
+ * version.
+ *
+ ****************************************************************************/
+static
+void vertical_band_1_2_scale_mmx
+(
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width
+) {
+ __asm {
+
+ mov esi, dest // Get the source and destination pointer
+ mov ecx, dest_pitch // Get the pitch size
+
+ pxor mm7, mm7 // clear out mm7
+ mov edx, dest_width // Loop counter
+
+ vs_1_2_loop:
+
+ movq mm0, [esi] // get Src[0]
+ movq mm1, [esi + ecx * 2] // get Src[1]
+
+ movq mm2, mm0 // make copy before unpack
+ movq mm3, mm1 // make copy before unpack
+
+ punpcklbw mm0, mm7 // low Src[0]
+ movq mm6, four_ones // mm6= 1, 1, 1, 1
+
+ punpcklbw mm1, mm7 // low Src[1]
+ paddw mm0, mm1 // low (a + b)
+
+ punpckhbw mm2, mm7 // high Src[0]
+ paddw mm0, mm6 // low (a + b + 1)
+
+ punpckhbw mm3, mm7
+ paddw mm2, mm3 // high (a + b )
+
+ psraw mm0, 1 // low (a + b +1 )/2
+ paddw mm2, mm6 // high (a + b + 1)
+
+ psraw mm2, 1 // high (a + b + 1)/2
+ packuswb mm0, mm2 // pack results
+
+ movq [esi+ecx], mm0 // write out eight bytes
+ add esi, 8
+
+ sub edx, 8
+ jg vs_1_2_loop
+ }
+
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : last_vertical_band_1_2_scale_mmx
+ *
+ * INPUTS : unsigned char *dest :
+ * unsigned int dest_pitch :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 1 to 2 up-scaling of band of pixels.
+ *
+ * SPECIAL NOTES : The routine uses the first line of the band below
+ * the current band. The function also has an "C" only
+ * version.
+ *
+ ****************************************************************************/
+static
+void last_vertical_band_1_2_scale_mmx
+(
+ unsigned char *dest,
+ unsigned int dest_pitch,
+ unsigned int dest_width
+) {
+ __asm {
+ mov esi, dest // Get the source and destination pointer
+ mov ecx, dest_pitch // Get the pitch size
+
+ mov edx, dest_width // Loop counter
+
+ last_vs_1_2_loop:
+
+ movq mm0, [esi] // get Src[0]
+ movq [esi+ecx], mm0 // write out eight bytes
+
+ add esi, 8
+ sub edx, 8
+
+ jg last_vs_1_2_loop
+ }
+}
+
+/****************************************************************************
+ *
+ * ROUTINE : horizontal_line_1_2_scale
+ *
+ * INPUTS : const unsigned char *source :
+ * unsigned int source_width :
+ * unsigned char *dest :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 1 to 2 up-scaling of a horizontal line of pixels.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+void horizontal_line_1_2_scale_mmx
+(
+ const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width
+) {
+ (void) dest_width;
+
+ __asm {
+ mov esi, source
+ mov edi, dest
+
+ pxor mm7, mm7
+ movq mm6, four_ones
+
+ mov ecx, source_width
+
+ hs_1_2_loop:
+
+ movq mm0, [esi]
+ movq mm1, [esi+1]
+
+ movq mm2, mm0
+ movq mm3, mm1
+
+ movq mm4, mm0
+ punpcklbw mm0, mm7
+
+ punpcklbw mm1, mm7
+ paddw mm0, mm1
+
+ paddw mm0, mm6
+ punpckhbw mm2, mm7
+
+ punpckhbw mm3, mm7
+ paddw mm2, mm3
+
+ paddw mm2, mm6
+ psraw mm0, 1
+
+ psraw mm2, 1
+ packuswb mm0, mm2
+
+ movq mm2, mm4
+ punpcklbw mm2, mm0
+
+ movq [edi], mm2
+ punpckhbw mm4, mm0
+
+ movq [edi+8], mm4
+ add esi, 8
+
+ add edi, 16
+ sub ecx, 8
+
+ cmp ecx, 8
+ jg hs_1_2_loop
+
+// last eight pixel
+
+ movq mm0, [esi]
+ movq mm1, mm0
+
+ movq mm2, mm0
+ movq mm3, mm1
+
+ psrlq mm1, 8
+ psrlq mm3, 56
+
+ psllq mm3, 56
+ por mm1, mm3
+
+ movq mm3, mm1
+ movq mm4, mm0
+
+ punpcklbw mm0, mm7
+ punpcklbw mm1, mm7
+
+ paddw mm0, mm1
+ paddw mm0, mm6
+
+ punpckhbw mm2, mm7
+ punpckhbw mm3, mm7
+
+ paddw mm2, mm3
+ paddw mm2, mm6
+
+ psraw mm0, 1
+ psraw mm2, 1
+
+ packuswb mm0, mm2
+ movq mm2, mm4
+
+ punpcklbw mm2, mm0
+ movq [edi], mm2
+
+ punpckhbw mm4, mm0
+ movq [edi+8], mm4
+ }
+}
+
+
+
+
+
+__declspec(align(16)) const static unsigned short const54_2[] = { 0, 64, 128, 192 };
+__declspec(align(16)) const static unsigned short const54_1[] = {256, 192, 128, 64 };
+
+
+/****************************************************************************
+ *
+ * ROUTINE : horizontal_line_5_4_scale_mmx
+ *
+ * INPUTS : const unsigned char *source : Pointer to source data.
+ * unsigned int source_width : Stride of source.
+ * unsigned char *dest : Pointer to destination data.
+ * unsigned int dest_width : Stride of destination (NOT USED).
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Copies horizontal line of pixels from source to
+ * destination scaling up by 4 to 5.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+void horizontal_line_5_4_scale_mmx
+(
+ const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width
+) {
+ /*
+ unsigned i;
+ unsigned int a, b, c, d, e;
+ unsigned char *des = dest;
+ const unsigned char *src = source;
+
+ (void) dest_width;
+
+ for ( i=0; i<source_width; i+=5 )
+ {
+ a = src[0];
+ b = src[1];
+ c = src[2];
+ d = src[3];
+ e = src[4];
+
+ des[0] = a;
+ des[1] = ((b*192 + c* 64 + 128)>>8);
+ des[2] = ((c*128 + d*128 + 128)>>8);
+ des[3] = ((d* 64 + e*192 + 128)>>8);
+
+ src += 5;
+ des += 4;
+ }
+ */
+ (void) dest_width;
+
+ __asm {
+
+ mov esi, source;
+ mov edi, dest;
+
+ mov ecx, source_width;
+ movq mm5, const54_1;
+
+ pxor mm7, mm7;
+ movq mm6, const54_2;
+
+ movq mm4, round_values;
+ lea edx, [esi+ecx];
+ horizontal_line_5_4_loop:
+
+ movq mm0, QWORD PTR [esi];
+ 00 01 02 03 04 05 06 07
+ movq mm1, mm0;
+ 00 01 02 03 04 05 06 07
+
+ psrlq mm0, 8;
+ 01 02 03 04 05 06 07 xx
+ punpcklbw mm1, mm7;
+ xx 00 xx 01 xx 02 xx 03
+
+ punpcklbw mm0, mm7;
+ xx 01 xx 02 xx 03 xx 04
+ pmullw mm1, mm5
+
+ pmullw mm0, mm6
+ add esi, 5
+
+ add edi, 4
+ paddw mm1, mm0
+
+ paddw mm1, mm4
+ psrlw mm1, 8
+
+ cmp esi, edx
+ packuswb mm1, mm7
+
+ movd DWORD PTR [edi-4], mm1
+
+ jl horizontal_line_5_4_loop
+
+ }
+
+}
+__declspec(align(16)) const static unsigned short one_fourths[] = { 64, 64, 64, 64 };
+__declspec(align(16)) const static unsigned short two_fourths[] = { 128, 128, 128, 128 };
+__declspec(align(16)) const static unsigned short three_fourths[] = { 192, 192, 192, 192 };
+
+static
+void vertical_band_5_4_scale_mmx(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) {
+
+ __asm {
+ push ebx
+
+ mov esi, source // Get the source and destination pointer
+ mov ecx, src_pitch // Get the pitch size
+
+ mov edi, dest // tow lines below
+ pxor mm7, mm7 // clear out mm7
+
+ mov edx, dest_pitch // Loop counter
+ mov ebx, dest_width
+
+ vs_5_4_loop:
+
+ movd mm0, DWORD ptr [esi] // src[0];
+ movd mm1, DWORD ptr [esi+ecx] // src[1];
+
+ movd mm2, DWORD ptr [esi+ecx*2]
+ lea eax, [esi+ecx*2] //
+
+ punpcklbw mm1, mm7
+ punpcklbw mm2, mm7
+
+ movq mm3, mm2
+ pmullw mm1, three_fourths
+
+ pmullw mm2, one_fourths
+ movd mm4, [eax+ecx]
+
+ pmullw mm3, two_fourths
+ punpcklbw mm4, mm7
+
+ movq mm5, mm4
+ pmullw mm4, two_fourths
+
+ paddw mm1, mm2
+ movd mm6, [eax+ecx*2]
+
+ pmullw mm5, one_fourths
+ paddw mm1, round_values;
+
+ paddw mm3, mm4
+ psrlw mm1, 8
+
+ punpcklbw mm6, mm7
+ paddw mm3, round_values
+
+ pmullw mm6, three_fourths
+ psrlw mm3, 8
+
+ packuswb mm1, mm7
+ packuswb mm3, mm7
+
+ movd DWORD PTR [edi], mm0
+ movd DWORD PTR [edi+edx], mm1
+
+
+ paddw mm5, mm6
+ movd DWORD PTR [edi+edx*2], mm3
+
+ lea eax, [edi+edx*2]
+ paddw mm5, round_values
+
+ psrlw mm5, 8
+ add edi, 4
+
+ packuswb mm5, mm7
+ movd DWORD PTR [eax+edx], mm5
+
+ add esi, 4
+ sub ebx, 4
+
+ jg vs_5_4_loop
+
+ pop ebx
+ }
+}
+
+
+__declspec(align(16)) const static unsigned short const53_1[] = { 0, 85, 171, 0 };
+__declspec(align(16)) const static unsigned short const53_2[] = {256, 171, 85, 0 };
+
+
+static
+void horizontal_line_5_3_scale_mmx
+(
+ const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width
+) {
+
+ (void) dest_width;
+ __asm {
+
+ mov esi, source;
+ mov edi, dest;
+
+ mov ecx, source_width;
+ movq mm5, const53_1;
+
+ pxor mm7, mm7;
+ movq mm6, const53_2;
+
+ movq mm4, round_values;
+ lea edx, [esi+ecx-5];
+ horizontal_line_5_3_loop:
+
+ movq mm0, QWORD PTR [esi];
+ 00 01 02 03 04 05 06 07
+ movq mm1, mm0;
+ 00 01 02 03 04 05 06 07
+
+ psllw mm0, 8;
+ xx 00 xx 02 xx 04 xx 06
+ psrlw mm1, 8;
+ 01 xx 03 xx 05 xx 07 xx
+
+ psrlw mm0, 8;
+ 00 xx 02 xx 04 xx 06 xx
+ psllq mm1, 16;
+ xx xx 01 xx 03 xx 05 xx
+
+ pmullw mm0, mm6
+
+ pmullw mm1, mm5
+ add esi, 5
+
+ add edi, 3
+ paddw mm1, mm0
+
+ paddw mm1, mm4
+ psrlw mm1, 8
+
+ cmp esi, edx
+ packuswb mm1, mm7
+
+ movd DWORD PTR [edi-3], mm1
+ jl horizontal_line_5_3_loop
+
+// exit condition
+ movq mm0, QWORD PTR [esi];
+ 00 01 02 03 04 05 06 07
+ movq mm1, mm0;
+ 00 01 02 03 04 05 06 07
+
+ psllw mm0, 8;
+ xx 00 xx 02 xx 04 xx 06
+ psrlw mm1, 8;
+ 01 xx 03 xx 05 xx 07 xx
+
+ psrlw mm0, 8;
+ 00 xx 02 xx 04 xx 06 xx
+ psllq mm1, 16;
+ xx xx 01 xx 03 xx 05 xx
+
+ pmullw mm0, mm6
+
+ pmullw mm1, mm5
+ paddw mm1, mm0
+
+ paddw mm1, mm4
+ psrlw mm1, 8
+
+ packuswb mm1, mm7
+ movd eax, mm1
+
+ mov edx, eax
+ shr edx, 16
+
+ mov WORD PTR[edi], ax
+ mov BYTE PTR[edi+2], dl
+
+ }
+
+}
+
+__declspec(align(16)) const static unsigned short one_thirds[] = { 85, 85, 85, 85 };
+__declspec(align(16)) const static unsigned short two_thirds[] = { 171, 171, 171, 171 };
+
+static
+void vertical_band_5_3_scale_mmx(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) {
+
+ __asm {
+ push ebx
+
+ mov esi, source // Get the source and destination pointer
+ mov ecx, src_pitch // Get the pitch size
+
+ mov edi, dest // tow lines below
+ pxor mm7, mm7 // clear out mm7
+
+ mov edx, dest_pitch // Loop counter
+ movq mm5, one_thirds
+
+ movq mm6, two_thirds
+ mov ebx, dest_width;
+
+ vs_5_3_loop:
+
+ movd mm0, DWORD ptr [esi] // src[0];
+ movd mm1, DWORD ptr [esi+ecx] // src[1];
+
+ movd mm2, DWORD ptr [esi+ecx*2]
+ lea eax, [esi+ecx*2] //
+
+ punpcklbw mm1, mm7
+ punpcklbw mm2, mm7
+
+ pmullw mm1, mm5
+ pmullw mm2, mm6
+
+ movd mm3, DWORD ptr [eax+ecx]
+ movd mm4, DWORD ptr [eax+ecx*2]
+
+ punpcklbw mm3, mm7
+ punpcklbw mm4, mm7
+
+ pmullw mm3, mm6
+ pmullw mm4, mm5
+
+
+ movd DWORD PTR [edi], mm0
+ paddw mm1, mm2
+
+ paddw mm1, round_values
+ psrlw mm1, 8
+
+ packuswb mm1, mm7
+ paddw mm3, mm4
+
+ paddw mm3, round_values
+ movd DWORD PTR [edi+edx], mm1
+
+ psrlw mm3, 8
+ packuswb mm3, mm7
+
+ movd DWORD PTR [edi+edx*2], mm3
+
+
+ add edi, 4
+ add esi, 4
+
+ sub ebx, 4
+ jg vs_5_3_loop
+
+ pop ebx
+ }
+}
+
+
+
+
+/****************************************************************************
+ *
+ * ROUTINE : horizontal_line_2_1_scale
+ *
+ * INPUTS : const unsigned char *source :
+ * unsigned int source_width :
+ * unsigned char *dest :
+ * unsigned int dest_width :
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : 1 to 2 up-scaling of a horizontal line of pixels.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+static
+void horizontal_line_2_1_scale_mmx
+(
+ const unsigned char *source,
+ unsigned int source_width,
+ unsigned char *dest,
+ unsigned int dest_width
+) {
+ (void) dest_width;
+ (void) source_width;
+ __asm {
+ mov esi, source
+ mov edi, dest
+
+ pxor mm7, mm7
+ mov ecx, dest_width
+
+ xor edx, edx
+ hs_2_1_loop:
+
+ movq mm0, [esi+edx*2]
+ psllw mm0, 8
+
+ psrlw mm0, 8
+ packuswb mm0, mm7
+
+ movd DWORD Ptr [edi+edx], mm0;
+ add edx, 4
+
+ cmp edx, ecx
+ jl hs_2_1_loop
+
+ }
+}
+
+
+
+static
+void vertical_band_2_1_scale_mmx(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) {
+ (void) dest_pitch;
+ (void) src_pitch;
+ vpx_memcpy(dest, source, dest_width);
+}
+
+
+__declspec(align(16)) const static unsigned short three_sixteenths[] = { 48, 48, 48, 48 };
+__declspec(align(16)) const static unsigned short ten_sixteenths[] = { 160, 160, 160, 160 };
+
+static
+void vertical_band_2_1_scale_i_mmx(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) {
+
+ (void) dest_pitch;
+ __asm {
+ mov esi, source
+ mov edi, dest
+
+ mov eax, src_pitch
+ mov edx, dest_width
+
+ pxor mm7, mm7
+ sub esi, eax // back one line
+
+
+ lea ecx, [esi+edx];
+ movq mm6, round_values;
+
+ movq mm5, three_sixteenths;
+ movq mm4, ten_sixteenths;
+
+ vs_2_1_i_loop:
+ movd mm0, [esi] //
+ movd mm1, [esi+eax] //
+
+ movd mm2, [esi+eax*2] //
+ punpcklbw mm0, mm7
+
+ pmullw mm0, mm5
+ punpcklbw mm1, mm7
+
+ pmullw mm1, mm4
+ punpcklbw mm2, mm7
+
+ pmullw mm2, mm5
+ paddw mm0, round_values
+
+ paddw mm1, mm2
+ paddw mm0, mm1
+
+ psrlw mm0, 8
+ packuswb mm0, mm7
+
+ movd DWORD PTR [edi], mm0
+ add esi, 4
+
+ add edi, 4;
+ cmp esi, ecx
+ jl vs_2_1_i_loop
+
+ }
+}
+
+
+
+void
+register_mmxscalers(void) {
+ vp8_horizontal_line_1_2_scale = horizontal_line_1_2_scale_mmx;
+ vp8_vertical_band_1_2_scale = vertical_band_1_2_scale_mmx;
+ vp8_last_vertical_band_1_2_scale = last_vertical_band_1_2_scale_mmx;
+ vp8_horizontal_line_3_5_scale = horizontal_line_3_5_scale_mmx;
+ vp8_vertical_band_3_5_scale = vertical_band_3_5_scale_mmx;
+ vp8_last_vertical_band_3_5_scale = last_vertical_band_3_5_scale_mmx;
+ vp8_horizontal_line_4_5_scale = horizontal_line_4_5_scale_mmx;
+ vp8_vertical_band_4_5_scale = vertical_band_4_5_scale_mmx;
+ vp8_last_vertical_band_4_5_scale = last_vertical_band_4_5_scale_mmx;
+
+ vp8_horizontal_line_3_4_scale = vp8cx_horizontal_line_3_4_scale_c;
+ vp8_vertical_band_3_4_scale = vp8cx_vertical_band_3_4_scale_c;
+ vp8_last_vertical_band_3_4_scale = vp8cx_last_vertical_band_3_4_scale_c;
+ vp8_horizontal_line_2_3_scale = vp8cx_horizontal_line_2_3_scale_c;
+ vp8_vertical_band_2_3_scale = vp8cx_vertical_band_2_3_scale_c;
+ vp8_last_vertical_band_2_3_scale = vp8cx_last_vertical_band_2_3_scale_c;
+
+
+
+ vp8_vertical_band_5_4_scale = vertical_band_5_4_scale_mmx;
+ vp8_vertical_band_5_3_scale = vertical_band_5_3_scale_mmx;
+ vp8_vertical_band_2_1_scale = vertical_band_2_1_scale_mmx;
+ vp8_vertical_band_2_1_scale_i = vertical_band_2_1_scale_i_mmx;
+ vp8_horizontal_line_2_1_scale = horizontal_line_2_1_scale_mmx;
+ vp8_horizontal_line_5_3_scale = horizontal_line_5_3_scale_mmx;
+ vp8_horizontal_line_5_4_scale = horizontal_line_5_4_scale_mmx;
+
+
+
+
+}
diff --git a/libvpx/vpx_scale/win32/scalesystemdependent.c b/libvpx/vpx_scale/win32/scalesystemdependent.c
new file mode 100644
index 0000000..98913d1
--- /dev/null
+++ b/libvpx/vpx_scale/win32/scalesystemdependent.c
@@ -0,0 +1,87 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/****************************************************************************
+*
+* Module Title : system_dependent.c
+*
+* Description : Miscellaneous system dependent functions
+*
+****************************************************************************/
+
+/****************************************************************************
+* Header Files
+****************************************************************************/
+#include "vpx_scale/vpxscale.h"
+#include "cpuidlib.h"
+
+/****************************************************************************
+* Imports
+*****************************************************************************/
+extern void register_generic_scalers(void);
+extern void register_mmxscalers(void);
+
+/****************************************************************************
+ *
+ * ROUTINE : post_proc_machine_specific_config
+ *
+ * INPUTS : UINT32 Version : Codec version number.
+ *
+ * OUTPUTS : None.
+ *
+ * RETURNS : void
+ *
+ * FUNCTION : Checks for machine specifc features such as MMX support
+ * sets appropriate flags and function pointers.
+ *
+ * SPECIAL NOTES : None.
+ *
+ ****************************************************************************/
+void
+vp8_scale_machine_specific_config(void) {
+ // If MMX supported then set to use MMX versions of functions else
+ // use original 'C' versions.
+ int mmx_enabled;
+ int xmm_enabled;
+ int wmt_enabled;
+
+ vpx_get_processor_flags(&mmx_enabled, &xmm_enabled, &wmt_enabled);
+
+ if (mmx_enabled || xmm_enabled || wmt_enabled) {
+ register_mmxscalers();
+ } else {
+ vp8_horizontal_line_1_2_scale = vp8cx_horizontal_line_1_2_scale_c;
+ vp8_vertical_band_1_2_scale = vp8cx_vertical_band_1_2_scale_c;
+ vp8_last_vertical_band_1_2_scale = vp8cx_last_vertical_band_1_2_scale_c;
+ vp8_horizontal_line_3_5_scale = vp8cx_horizontal_line_3_5_scale_c;
+ vp8_vertical_band_3_5_scale = vp8cx_vertical_band_3_5_scale_c;
+ vp8_last_vertical_band_3_5_scale = vp8cx_last_vertical_band_3_5_scale_c;
+ vp8_horizontal_line_3_4_scale = vp8cx_horizontal_line_3_4_scale_c;
+ vp8_vertical_band_3_4_scale = vp8cx_vertical_band_3_4_scale_c;
+ vp8_last_vertical_band_3_4_scale = vp8cx_last_vertical_band_3_4_scale_c;
+ vp8_horizontal_line_2_3_scale = vp8cx_horizontal_line_2_3_scale_c;
+ vp8_vertical_band_2_3_scale = vp8cx_vertical_band_2_3_scale_c;
+ vp8_last_vertical_band_2_3_scale = vp8cx_last_vertical_band_2_3_scale_c;
+ vp8_horizontal_line_4_5_scale = vp8cx_horizontal_line_4_5_scale_c;
+ vp8_vertical_band_4_5_scale = vp8cx_vertical_band_4_5_scale_c;
+ vp8_last_vertical_band_4_5_scale = vp8cx_last_vertical_band_4_5_scale_c;
+
+
+ vp8_vertical_band_5_4_scale = vp8cx_vertical_band_5_4_scale_c;
+ vp8_vertical_band_5_3_scale = vp8cx_vertical_band_5_3_scale_c;
+ vp8_vertical_band_2_1_scale = vp8cx_vertical_band_2_1_scale_c;
+ vp8_vertical_band_2_1_scale_i = vp8cx_vertical_band_2_1_scale_i_c;
+ vp8_horizontal_line_2_1_scale = vp8cx_horizontal_line_2_1_scale_c;
+ vp8_horizontal_line_5_3_scale = vp8cx_horizontal_line_5_3_scale_c;
+ vp8_horizontal_line_5_4_scale = vp8cx_horizontal_line_5_4_scale_c;
+
+ }
+}
diff --git a/libvpx/vpx_scale/yv12config.h b/libvpx/vpx_scale/yv12config.h
new file mode 100644
index 0000000..6a8a1fc
--- /dev/null
+++ b/libvpx/vpx_scale/yv12config.h
@@ -0,0 +1,73 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#ifndef YV12_CONFIG_H
+#define YV12_CONFIG_H
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+
+#define VP8BORDERINPIXELS 32
+#define VP9BORDERINPIXELS 64
+#define VP9_INTERP_EXTEND 4
+
+ /*************************************
+ For INT_YUV:
+
+ Y = (R+G*2+B)/4;
+ U = (R-B)/2;
+ V = (G*2 - R - B)/4;
+ And
+ R = Y+U-V;
+ G = Y+V;
+ B = Y-U-V;
+ ************************************/
+ typedef enum
+ {
+ REG_YUV = 0, /* Regular yuv */
+ INT_YUV = 1 /* The type of yuv that can be tranfer to and from RGB through integer transform */
+ }
+ YUV_TYPE;
+
+ typedef struct yv12_buffer_config {
+ int y_width;
+ int y_height;
+ int y_stride;
+ /* int yinternal_width; */
+
+ int uv_width;
+ int uv_height;
+ int uv_stride;
+ /* int uvinternal_width; */
+
+ unsigned char *y_buffer;
+ unsigned char *u_buffer;
+ unsigned char *v_buffer;
+
+ unsigned char *buffer_alloc;
+ int border;
+ int frame_size;
+ YUV_TYPE clrtype;
+
+ int corrupted;
+ int flags;
+ } YV12_BUFFER_CONFIG;
+
+ int vp8_yv12_alloc_frame_buffer(YV12_BUFFER_CONFIG *ybf, int width, int height, int border);
+ int vp8_yv12_de_alloc_frame_buffer(YV12_BUFFER_CONFIG *ybf);
+
+#ifdef __cplusplus
+}
+#endif
+
+
+#endif /*YV12_CONFIG_H*/
diff --git a/libvpx/vpxdec.c b/libvpx/vpxdec.c
new file mode 100644
index 0000000..9b728bf
--- /dev/null
+++ b/libvpx/vpxdec.c
@@ -0,0 +1,1161 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/* This is a simple program that reads ivf files and decodes them
+ * using the new interface. Decoded frames are output as YV12 raw.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#include <limits.h>
+
+#define VPX_CODEC_DISABLE_COMPAT 1
+#include "vpx_config.h"
+#include "vpx/vpx_decoder.h"
+#include "vpx_ports/vpx_timer.h"
+#if CONFIG_VP8_DECODER
+#include "vpx/vp8dx.h"
+#endif
+#if CONFIG_MD5
+#include "md5_utils.h"
+#endif
+#include "tools_common.h"
+#include "nestegg/include/nestegg/nestegg.h"
+
+#if CONFIG_OS_SUPPORT
+#if defined(_MSC_VER)
+#include <io.h>
+#define snprintf _snprintf
+#define isatty _isatty
+#define fileno _fileno
+#else
+#include <unistd.h>
+#endif
+#endif
+
+#ifndef PATH_MAX
+#define PATH_MAX 256
+#endif
+
+static const char *exec_name;
+
+#define VP8_FOURCC (0x00385056)
+static const struct
+{
+ char const *name;
+ vpx_codec_iface_t *iface;
+ unsigned int fourcc;
+ unsigned int fourcc_mask;
+} ifaces[] =
+{
+#if CONFIG_VP8_DECODER
+ {"vp8", &vpx_codec_vp8_dx_algo, VP8_FOURCC, 0x00FFFFFF},
+#endif
+};
+
+#include "args.h"
+static const arg_def_t codecarg = ARG_DEF(NULL, "codec", 1,
+ "Codec to use");
+static const arg_def_t use_yv12 = ARG_DEF(NULL, "yv12", 0,
+ "Output raw YV12 frames");
+static const arg_def_t use_i420 = ARG_DEF(NULL, "i420", 0,
+ "Output raw I420 frames");
+static const arg_def_t flipuvarg = ARG_DEF(NULL, "flipuv", 0,
+ "Flip the chroma planes in the output");
+static const arg_def_t noblitarg = ARG_DEF(NULL, "noblit", 0,
+ "Don't process the decoded frames");
+static const arg_def_t progressarg = ARG_DEF(NULL, "progress", 0,
+ "Show progress after each frame decodes");
+static const arg_def_t limitarg = ARG_DEF(NULL, "limit", 1,
+ "Stop decoding after n frames");
+static const arg_def_t postprocarg = ARG_DEF(NULL, "postproc", 0,
+ "Postprocess decoded frames");
+static const arg_def_t summaryarg = ARG_DEF(NULL, "summary", 0,
+ "Show timing summary");
+static const arg_def_t outputfile = ARG_DEF("o", "output", 1,
+ "Output file name pattern (see below)");
+static const arg_def_t threadsarg = ARG_DEF("t", "threads", 1,
+ "Max threads to use");
+static const arg_def_t verbosearg = ARG_DEF("v", "verbose", 0,
+ "Show version string");
+static const arg_def_t error_concealment = ARG_DEF(NULL, "error-concealment", 0,
+ "Enable decoder error-concealment");
+
+
+#if CONFIG_MD5
+static const arg_def_t md5arg = ARG_DEF(NULL, "md5", 0,
+ "Compute the MD5 sum of the decoded frame");
+#endif
+static const arg_def_t *all_args[] =
+{
+ &codecarg, &use_yv12, &use_i420, &flipuvarg, &noblitarg,
+ &progressarg, &limitarg, &postprocarg, &summaryarg, &outputfile,
+ &threadsarg, &verbosearg,
+#if CONFIG_MD5
+ &md5arg,
+#endif
+ &error_concealment,
+ NULL
+};
+
+#if CONFIG_VP8_DECODER
+static const arg_def_t addnoise_level = ARG_DEF(NULL, "noise-level", 1,
+ "Enable VP8 postproc add noise");
+static const arg_def_t deblock = ARG_DEF(NULL, "deblock", 0,
+ "Enable VP8 deblocking");
+static const arg_def_t demacroblock_level = ARG_DEF(NULL, "demacroblock-level", 1,
+ "Enable VP8 demacroblocking, w/ level");
+static const arg_def_t pp_debug_info = ARG_DEF(NULL, "pp-debug-info", 1,
+ "Enable VP8 visible debug info");
+static const arg_def_t pp_disp_ref_frame = ARG_DEF(NULL, "pp-dbg-ref-frame", 1,
+ "Display only selected reference frame per macro block");
+static const arg_def_t pp_disp_mb_modes = ARG_DEF(NULL, "pp-dbg-mb-modes", 1,
+ "Display only selected macro block modes");
+static const arg_def_t pp_disp_b_modes = ARG_DEF(NULL, "pp-dbg-b-modes", 1,
+ "Display only selected block modes");
+static const arg_def_t pp_disp_mvs = ARG_DEF(NULL, "pp-dbg-mvs", 1,
+ "Draw only selected motion vectors");
+static const arg_def_t mfqe = ARG_DEF(NULL, "mfqe", 0,
+ "Enable multiframe quality enhancement");
+
+static const arg_def_t *vp8_pp_args[] =
+{
+ &addnoise_level, &deblock, &demacroblock_level, &pp_debug_info,
+ &pp_disp_ref_frame, &pp_disp_mb_modes, &pp_disp_b_modes, &pp_disp_mvs, &mfqe,
+ NULL
+};
+#endif
+
+static void usage_exit()
+{
+ int i;
+
+ fprintf(stderr, "Usage: %s <options> filename\n\n"
+ "Options:\n", exec_name);
+ arg_show_usage(stderr, all_args);
+#if CONFIG_VP8_DECODER
+ fprintf(stderr, "\nVP8 Postprocessing Options:\n");
+ arg_show_usage(stderr, vp8_pp_args);
+#endif
+ fprintf(stderr,
+ "\nOutput File Patterns:\n\n"
+ " The -o argument specifies the name of the file(s) to "
+ "write to. If the\n argument does not include any escape "
+ "characters, the output will be\n written to a single file. "
+ "Otherwise, the filename will be calculated by\n expanding "
+ "the following escape characters:\n");
+ fprintf(stderr,
+ "\n\t%%w - Frame width"
+ "\n\t%%h - Frame height"
+ "\n\t%%<n> - Frame number, zero padded to <n> places (1..9)"
+ "\n\n Pattern arguments are only supported in conjunction "
+ "with the --yv12 and\n --i420 options. If the -o option is "
+ "not specified, the output will be\n directed to stdout.\n"
+ );
+ fprintf(stderr, "\nIncluded decoders:\n\n");
+
+ for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++)
+ fprintf(stderr, " %-6s - %s\n",
+ ifaces[i].name,
+ vpx_codec_iface_name(ifaces[i].iface));
+
+ exit(EXIT_FAILURE);
+}
+
+void die(const char *fmt, ...)
+{
+ va_list ap;
+ va_start(ap, fmt);
+ vfprintf(stderr, fmt, ap);
+ fprintf(stderr, "\n");
+ usage_exit();
+}
+
+static unsigned int mem_get_le16(const void *vmem)
+{
+ unsigned int val;
+ const unsigned char *mem = (const unsigned char *)vmem;
+
+ val = mem[1] << 8;
+ val |= mem[0];
+ return val;
+}
+
+static unsigned int mem_get_le32(const void *vmem)
+{
+ unsigned int val;
+ const unsigned char *mem = (const unsigned char *)vmem;
+
+ val = mem[3] << 24;
+ val |= mem[2] << 16;
+ val |= mem[1] << 8;
+ val |= mem[0];
+ return val;
+}
+
+enum file_kind
+{
+ RAW_FILE,
+ IVF_FILE,
+ WEBM_FILE
+};
+
+struct input_ctx
+{
+ enum file_kind kind;
+ FILE *infile;
+ nestegg *nestegg_ctx;
+ nestegg_packet *pkt;
+ unsigned int chunk;
+ unsigned int chunks;
+ unsigned int video_track;
+};
+
+#define IVF_FRAME_HDR_SZ (sizeof(uint32_t) + sizeof(uint64_t))
+#define RAW_FRAME_HDR_SZ (sizeof(uint32_t))
+static int read_frame(struct input_ctx *input,
+ uint8_t **buf,
+ size_t *buf_sz,
+ size_t *buf_alloc_sz)
+{
+ char raw_hdr[IVF_FRAME_HDR_SZ];
+ size_t new_buf_sz;
+ FILE *infile = input->infile;
+ enum file_kind kind = input->kind;
+ if(kind == WEBM_FILE)
+ {
+ if(input->chunk >= input->chunks)
+ {
+ unsigned int track;
+
+ do
+ {
+ /* End of this packet, get another. */
+ if(input->pkt)
+ nestegg_free_packet(input->pkt);
+
+ if(nestegg_read_packet(input->nestegg_ctx, &input->pkt) <= 0
+ || nestegg_packet_track(input->pkt, &track))
+ return 1;
+
+ } while(track != input->video_track);
+
+ if(nestegg_packet_count(input->pkt, &input->chunks))
+ return 1;
+ input->chunk = 0;
+ }
+
+ if(nestegg_packet_data(input->pkt, input->chunk, buf, buf_sz))
+ return 1;
+ input->chunk++;
+
+ return 0;
+ }
+ /* For both the raw and ivf formats, the frame size is the first 4 bytes
+ * of the frame header. We just need to special case on the header
+ * size.
+ */
+ else if (fread(raw_hdr, kind==IVF_FILE
+ ? IVF_FRAME_HDR_SZ : RAW_FRAME_HDR_SZ, 1, infile) != 1)
+ {
+ if (!feof(infile))
+ fprintf(stderr, "Failed to read frame size\n");
+
+ new_buf_sz = 0;
+ }
+ else
+ {
+ new_buf_sz = mem_get_le32(raw_hdr);
+
+ if (new_buf_sz > 256 * 1024 * 1024)
+ {
+ fprintf(stderr, "Error: Read invalid frame size (%u)\n",
+ (unsigned int)new_buf_sz);
+ new_buf_sz = 0;
+ }
+
+ if (kind == RAW_FILE && new_buf_sz > 256 * 1024)
+ fprintf(stderr, "Warning: Read invalid frame size (%u)"
+ " - not a raw file?\n", (unsigned int)new_buf_sz);
+
+ if (new_buf_sz > *buf_alloc_sz)
+ {
+ uint8_t *new_buf = realloc(*buf, 2 * new_buf_sz);
+
+ if (new_buf)
+ {
+ *buf = new_buf;
+ *buf_alloc_sz = 2 * new_buf_sz;
+ }
+ else
+ {
+ fprintf(stderr, "Failed to allocate compressed data buffer\n");
+ new_buf_sz = 0;
+ }
+ }
+ }
+
+ *buf_sz = new_buf_sz;
+
+ if (!feof(infile))
+ {
+ if (fread(*buf, 1, *buf_sz, infile) != *buf_sz)
+ {
+ fprintf(stderr, "Failed to read full frame\n");
+ return 1;
+ }
+
+ return 0;
+ }
+
+ return 1;
+}
+
+void *out_open(const char *out_fn, int do_md5)
+{
+ void *out = NULL;
+
+ if (do_md5)
+ {
+#if CONFIG_MD5
+ MD5Context *md5_ctx = out = malloc(sizeof(MD5Context));
+ (void)out_fn;
+ MD5Init(md5_ctx);
+#endif
+ }
+ else
+ {
+ FILE *outfile = out = strcmp("-", out_fn) ? fopen(out_fn, "wb")
+ : set_binary_mode(stdout);
+
+ if (!outfile)
+ {
+ fprintf(stderr, "Failed to output file");
+ exit(EXIT_FAILURE);
+ }
+ }
+
+ return out;
+}
+
+void out_put(void *out, const uint8_t *buf, unsigned int len, int do_md5)
+{
+ if (do_md5)
+ {
+#if CONFIG_MD5
+ MD5Update(out, buf, len);
+#endif
+ }
+ else
+ {
+ (void) fwrite(buf, 1, len, out);
+ }
+}
+
+void out_close(void *out, const char *out_fn, int do_md5)
+{
+ if (do_md5)
+ {
+#if CONFIG_MD5
+ uint8_t md5[16];
+ int i;
+
+ MD5Final(md5, out);
+ free(out);
+
+ for (i = 0; i < 16; i++)
+ printf("%02x", md5[i]);
+
+ printf(" %s\n", out_fn);
+#endif
+ }
+ else
+ {
+ fclose(out);
+ }
+}
+
+unsigned int file_is_ivf(FILE *infile,
+ unsigned int *fourcc,
+ unsigned int *width,
+ unsigned int *height,
+ unsigned int *fps_den,
+ unsigned int *fps_num)
+{
+ char raw_hdr[32];
+ int is_ivf = 0;
+
+ if (fread(raw_hdr, 1, 32, infile) == 32)
+ {
+ if (raw_hdr[0] == 'D' && raw_hdr[1] == 'K'
+ && raw_hdr[2] == 'I' && raw_hdr[3] == 'F')
+ {
+ is_ivf = 1;
+
+ if (mem_get_le16(raw_hdr + 4) != 0)
+ fprintf(stderr, "Error: Unrecognized IVF version! This file may not"
+ " decode properly.");
+
+ *fourcc = mem_get_le32(raw_hdr + 8);
+ *width = mem_get_le16(raw_hdr + 12);
+ *height = mem_get_le16(raw_hdr + 14);
+ *fps_num = mem_get_le32(raw_hdr + 16);
+ *fps_den = mem_get_le32(raw_hdr + 20);
+
+ /* Some versions of vpxenc used 1/(2*fps) for the timebase, so
+ * we can guess the framerate using only the timebase in this
+ * case. Other files would require reading ahead to guess the
+ * timebase, like we do for webm.
+ */
+ if(*fps_num < 1000)
+ {
+ /* Correct for the factor of 2 applied to the timebase in the
+ * encoder.
+ */
+ if(*fps_num&1)*fps_den<<=1;
+ else *fps_num>>=1;
+ }
+ else
+ {
+ /* Don't know FPS for sure, and don't have readahead code
+ * (yet?), so just default to 30fps.
+ */
+ *fps_num = 30;
+ *fps_den = 1;
+ }
+ }
+ }
+
+ if (!is_ivf)
+ rewind(infile);
+
+ return is_ivf;
+}
+
+
+unsigned int file_is_raw(FILE *infile,
+ unsigned int *fourcc,
+ unsigned int *width,
+ unsigned int *height,
+ unsigned int *fps_den,
+ unsigned int *fps_num)
+{
+ unsigned char buf[32];
+ int is_raw = 0;
+ vpx_codec_stream_info_t si;
+
+ si.sz = sizeof(si);
+
+ if (fread(buf, 1, 32, infile) == 32)
+ {
+ int i;
+
+ if(mem_get_le32(buf) < 256 * 1024 * 1024)
+ for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++)
+ if(!vpx_codec_peek_stream_info(ifaces[i].iface,
+ buf + 4, 32 - 4, &si))
+ {
+ is_raw = 1;
+ *fourcc = ifaces[i].fourcc;
+ *width = si.w;
+ *height = si.h;
+ *fps_num = 30;
+ *fps_den = 1;
+ break;
+ }
+ }
+
+ rewind(infile);
+ return is_raw;
+}
+
+
+static int
+nestegg_read_cb(void *buffer, size_t length, void *userdata)
+{
+ FILE *f = userdata;
+
+ if(fread(buffer, 1, length, f) < length)
+ {
+ if (ferror(f))
+ return -1;
+ if (feof(f))
+ return 0;
+ }
+ return 1;
+}
+
+
+static int
+nestegg_seek_cb(int64_t offset, int whence, void * userdata)
+{
+ switch(whence) {
+ case NESTEGG_SEEK_SET: whence = SEEK_SET; break;
+ case NESTEGG_SEEK_CUR: whence = SEEK_CUR; break;
+ case NESTEGG_SEEK_END: whence = SEEK_END; break;
+ };
+ return fseek(userdata, (long)offset, whence)? -1 : 0;
+}
+
+
+static int64_t
+nestegg_tell_cb(void * userdata)
+{
+ return ftell(userdata);
+}
+
+
+static void
+nestegg_log_cb(nestegg * context, unsigned int severity, char const * format,
+ ...)
+{
+ va_list ap;
+
+ va_start(ap, format);
+ vfprintf(stderr, format, ap);
+ fprintf(stderr, "\n");
+ va_end(ap);
+}
+
+
+static int
+webm_guess_framerate(struct input_ctx *input,
+ unsigned int *fps_den,
+ unsigned int *fps_num)
+{
+ unsigned int i;
+ uint64_t tstamp=0;
+
+ /* Guess the framerate. Read up to 1 second, or 50 video packets,
+ * whichever comes first.
+ */
+ for(i=0; tstamp < 1000000000 && i < 50;)
+ {
+ nestegg_packet * pkt;
+ unsigned int track;
+
+ if(nestegg_read_packet(input->nestegg_ctx, &pkt) <= 0)
+ break;
+
+ nestegg_packet_track(pkt, &track);
+ if(track == input->video_track)
+ {
+ nestegg_packet_tstamp(pkt, &tstamp);
+ i++;
+ }
+
+ nestegg_free_packet(pkt);
+ }
+
+ if(nestegg_track_seek(input->nestegg_ctx, input->video_track, 0))
+ goto fail;
+
+ *fps_num = (i - 1) * 1000000;
+ *fps_den = (unsigned int)(tstamp / 1000);
+ return 0;
+fail:
+ nestegg_destroy(input->nestegg_ctx);
+ input->nestegg_ctx = NULL;
+ rewind(input->infile);
+ return 1;
+}
+
+
+static int
+file_is_webm(struct input_ctx *input,
+ unsigned int *fourcc,
+ unsigned int *width,
+ unsigned int *height,
+ unsigned int *fps_den,
+ unsigned int *fps_num)
+{
+ unsigned int i, n;
+ int track_type = -1;
+
+ nestegg_io io = {nestegg_read_cb, nestegg_seek_cb, nestegg_tell_cb, 0};
+ nestegg_video_params params;
+
+ io.userdata = input->infile;
+ if(nestegg_init(&input->nestegg_ctx, io, NULL))
+ goto fail;
+
+ if(nestegg_track_count(input->nestegg_ctx, &n))
+ goto fail;
+
+ for(i=0; i<n; i++)
+ {
+ track_type = nestegg_track_type(input->nestegg_ctx, i);
+
+ if(track_type == NESTEGG_TRACK_VIDEO)
+ break;
+ else if(track_type < 0)
+ goto fail;
+ }
+
+ if(nestegg_track_codec_id(input->nestegg_ctx, i) != NESTEGG_CODEC_VP8)
+ {
+ fprintf(stderr, "Not VP8 video, quitting.\n");
+ exit(1);
+ }
+
+ input->video_track = i;
+
+ if(nestegg_track_video_params(input->nestegg_ctx, i, ¶ms))
+ goto fail;
+
+ *fps_den = 0;
+ *fps_num = 0;
+ *fourcc = VP8_FOURCC;
+ *width = params.width;
+ *height = params.height;
+ return 1;
+fail:
+ input->nestegg_ctx = NULL;
+ rewind(input->infile);
+ return 0;
+}
+
+
+void show_progress(int frame_in, int frame_out, unsigned long dx_time)
+{
+ fprintf(stderr, "%d decoded frames/%d showed frames in %lu us (%.2f fps)\r",
+ frame_in, frame_out, dx_time,
+ (float)frame_out * 1000000.0 / (float)dx_time);
+}
+
+
+void generate_filename(const char *pattern, char *out, size_t q_len,
+ unsigned int d_w, unsigned int d_h,
+ unsigned int frame_in)
+{
+ const char *p = pattern;
+ char *q = out;
+
+ do
+ {
+ char *next_pat = strchr(p, '%');
+
+ if(p == next_pat)
+ {
+ size_t pat_len;
+
+ /* parse the pattern */
+ q[q_len - 1] = '\0';
+ switch(p[1])
+ {
+ case 'w': snprintf(q, q_len - 1, "%d", d_w); break;
+ case 'h': snprintf(q, q_len - 1, "%d", d_h); break;
+ case '1': snprintf(q, q_len - 1, "%d", frame_in); break;
+ case '2': snprintf(q, q_len - 1, "%02d", frame_in); break;
+ case '3': snprintf(q, q_len - 1, "%03d", frame_in); break;
+ case '4': snprintf(q, q_len - 1, "%04d", frame_in); break;
+ case '5': snprintf(q, q_len - 1, "%05d", frame_in); break;
+ case '6': snprintf(q, q_len - 1, "%06d", frame_in); break;
+ case '7': snprintf(q, q_len - 1, "%07d", frame_in); break;
+ case '8': snprintf(q, q_len - 1, "%08d", frame_in); break;
+ case '9': snprintf(q, q_len - 1, "%09d", frame_in); break;
+ default:
+ die("Unrecognized pattern %%%c\n", p[1]);
+ }
+
+ pat_len = strlen(q);
+ if(pat_len >= q_len - 1)
+ die("Output filename too long.\n");
+ q += pat_len;
+ p += 2;
+ q_len -= pat_len;
+ }
+ else
+ {
+ size_t copy_len;
+
+ /* copy the next segment */
+ if(!next_pat)
+ copy_len = strlen(p);
+ else
+ copy_len = next_pat - p;
+
+ if(copy_len >= q_len - 1)
+ die("Output filename too long.\n");
+
+ memcpy(q, p, copy_len);
+ q[copy_len] = '\0';
+ q += copy_len;
+ p += copy_len;
+ q_len -= copy_len;
+ }
+ } while(*p);
+}
+
+
+int main(int argc, const char **argv_)
+{
+ vpx_codec_ctx_t decoder;
+ char *fn = NULL;
+ int i;
+ uint8_t *buf = NULL;
+ size_t buf_sz = 0, buf_alloc_sz = 0;
+ FILE *infile;
+ int frame_in = 0, frame_out = 0, flipuv = 0, noblit = 0, do_md5 = 0, progress = 0;
+ int stop_after = 0, postproc = 0, summary = 0, quiet = 1;
+ int ec_enabled = 0;
+ vpx_codec_iface_t *iface = NULL;
+ unsigned int fourcc;
+ unsigned long dx_time = 0;
+ struct arg arg;
+ char **argv, **argi, **argj;
+ const char *outfile_pattern = 0;
+ char outfile[PATH_MAX];
+ int single_file;
+ int use_y4m = 1;
+ unsigned int width;
+ unsigned int height;
+ unsigned int fps_den;
+ unsigned int fps_num;
+ void *out = NULL;
+ vpx_codec_dec_cfg_t cfg = {0};
+#if CONFIG_VP8_DECODER
+ vp8_postproc_cfg_t vp8_pp_cfg = {0};
+ int vp8_dbg_color_ref_frame = 0;
+ int vp8_dbg_color_mb_modes = 0;
+ int vp8_dbg_color_b_modes = 0;
+ int vp8_dbg_display_mv = 0;
+#endif
+ struct input_ctx input = {0};
+ int frames_corrupted = 0;
+ int dec_flags = 0;
+
+ /* Parse command line */
+ exec_name = argv_[0];
+ argv = argv_dup(argc - 1, argv_ + 1);
+
+ for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step)
+ {
+ memset(&arg, 0, sizeof(arg));
+ arg.argv_step = 1;
+
+ if (arg_match(&arg, &codecarg, argi))
+ {
+ int j, k = -1;
+
+ for (j = 0; j < sizeof(ifaces) / sizeof(ifaces[0]); j++)
+ if (!strcmp(ifaces[j].name, arg.val))
+ k = j;
+
+ if (k >= 0)
+ iface = ifaces[k].iface;
+ else
+ die("Error: Unrecognized argument (%s) to --codec\n",
+ arg.val);
+ }
+ else if (arg_match(&arg, &outputfile, argi))
+ outfile_pattern = arg.val;
+ else if (arg_match(&arg, &use_yv12, argi))
+ {
+ use_y4m = 0;
+ flipuv = 1;
+ }
+ else if (arg_match(&arg, &use_i420, argi))
+ {
+ use_y4m = 0;
+ flipuv = 0;
+ }
+ else if (arg_match(&arg, &flipuvarg, argi))
+ flipuv = 1;
+ else if (arg_match(&arg, &noblitarg, argi))
+ noblit = 1;
+ else if (arg_match(&arg, &progressarg, argi))
+ progress = 1;
+ else if (arg_match(&arg, &limitarg, argi))
+ stop_after = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &postprocarg, argi))
+ postproc = 1;
+ else if (arg_match(&arg, &md5arg, argi))
+ do_md5 = 1;
+ else if (arg_match(&arg, &summaryarg, argi))
+ summary = 1;
+ else if (arg_match(&arg, &threadsarg, argi))
+ cfg.threads = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &verbosearg, argi))
+ quiet = 0;
+
+#if CONFIG_VP8_DECODER
+ else if (arg_match(&arg, &addnoise_level, argi))
+ {
+ postproc = 1;
+ vp8_pp_cfg.post_proc_flag |= VP8_ADDNOISE;
+ vp8_pp_cfg.noise_level = arg_parse_uint(&arg);
+ }
+ else if (arg_match(&arg, &demacroblock_level, argi))
+ {
+ postproc = 1;
+ vp8_pp_cfg.post_proc_flag |= VP8_DEMACROBLOCK;
+ vp8_pp_cfg.deblocking_level = arg_parse_uint(&arg);
+ }
+ else if (arg_match(&arg, &deblock, argi))
+ {
+ postproc = 1;
+ vp8_pp_cfg.post_proc_flag |= VP8_DEBLOCK;
+ }
+ else if (arg_match(&arg, &mfqe, argi))
+ {
+ postproc = 1;
+ vp8_pp_cfg.post_proc_flag |= VP8_MFQE;
+ }
+ else if (arg_match(&arg, &pp_debug_info, argi))
+ {
+ unsigned int level = arg_parse_uint(&arg);
+
+ postproc = 1;
+ vp8_pp_cfg.post_proc_flag &= ~0x7;
+
+ if (level)
+ vp8_pp_cfg.post_proc_flag |= level;
+ }
+ else if (arg_match(&arg, &pp_disp_ref_frame, argi))
+ {
+ unsigned int flags = arg_parse_int(&arg);
+ if (flags)
+ {
+ postproc = 1;
+ vp8_dbg_color_ref_frame = flags;
+ }
+ }
+ else if (arg_match(&arg, &pp_disp_mb_modes, argi))
+ {
+ unsigned int flags = arg_parse_int(&arg);
+ if (flags)
+ {
+ postproc = 1;
+ vp8_dbg_color_mb_modes = flags;
+ }
+ }
+ else if (arg_match(&arg, &pp_disp_b_modes, argi))
+ {
+ unsigned int flags = arg_parse_int(&arg);
+ if (flags)
+ {
+ postproc = 1;
+ vp8_dbg_color_b_modes = flags;
+ }
+ }
+ else if (arg_match(&arg, &pp_disp_mvs, argi))
+ {
+ unsigned int flags = arg_parse_int(&arg);
+ if (flags)
+ {
+ postproc = 1;
+ vp8_dbg_display_mv = flags;
+ }
+ }
+ else if (arg_match(&arg, &error_concealment, argi))
+ {
+ ec_enabled = 1;
+ }
+
+#endif
+ else
+ argj++;
+ }
+
+ /* Check for unrecognized options */
+ for (argi = argv; *argi; argi++)
+ if (argi[0][0] == '-' && strlen(argi[0]) > 1)
+ die("Error: Unrecognized option %s\n", *argi);
+
+ /* Handle non-option arguments */
+ fn = argv[0];
+
+ if (!fn)
+ usage_exit();
+
+ /* Open file */
+ infile = strcmp(fn, "-") ? fopen(fn, "rb") : set_binary_mode(stdin);
+
+ if (!infile)
+ {
+ fprintf(stderr, "Failed to open file '%s'",
+ strcmp(fn, "-") ? fn : "stdin");
+ return EXIT_FAILURE;
+ }
+#if CONFIG_OS_SUPPORT
+ /* Make sure we don't dump to the terminal, unless forced to with -o - */
+ if(!outfile_pattern && isatty(fileno(stdout)) && !do_md5 && !noblit)
+ {
+ fprintf(stderr,
+ "Not dumping raw video to your terminal. Use '-o -' to "
+ "override.\n");
+ return EXIT_FAILURE;
+ }
+#endif
+ input.infile = infile;
+ if(file_is_ivf(infile, &fourcc, &width, &height, &fps_den,
+ &fps_num))
+ input.kind = IVF_FILE;
+ else if(file_is_webm(&input, &fourcc, &width, &height, &fps_den, &fps_num))
+ input.kind = WEBM_FILE;
+ else if(file_is_raw(infile, &fourcc, &width, &height, &fps_den, &fps_num))
+ input.kind = RAW_FILE;
+ else
+ {
+ fprintf(stderr, "Unrecognized input file type.\n");
+ return EXIT_FAILURE;
+ }
+
+ /* If the output file is not set or doesn't have a sequence number in
+ * it, then we only open it once.
+ */
+ outfile_pattern = outfile_pattern ? outfile_pattern : "-";
+ single_file = 1;
+ {
+ const char *p = outfile_pattern;
+ do
+ {
+ p = strchr(p, '%');
+ if(p && p[1] >= '1' && p[1] <= '9')
+ {
+ /* pattern contains sequence number, so it's not unique. */
+ single_file = 0;
+ break;
+ }
+ if(p)
+ p++;
+ } while(p);
+ }
+
+ if(single_file && !noblit)
+ {
+ generate_filename(outfile_pattern, outfile, sizeof(outfile)-1,
+ width, height, 0);
+ out = out_open(outfile, do_md5);
+ }
+
+ if (use_y4m && !noblit)
+ {
+ char buffer[128];
+ if (!single_file)
+ {
+ fprintf(stderr, "YUV4MPEG2 not supported with output patterns,"
+ " try --i420 or --yv12.\n");
+ return EXIT_FAILURE;
+ }
+
+ if(input.kind == WEBM_FILE)
+ if(webm_guess_framerate(&input, &fps_den, &fps_num))
+ {
+ fprintf(stderr, "Failed to guess framerate -- error parsing "
+ "webm file?\n");
+ return EXIT_FAILURE;
+ }
+
+
+ /*Note: We can't output an aspect ratio here because IVF doesn't
+ store one, and neither does VP8.
+ That will have to wait until these tools support WebM natively.*/
+ sprintf(buffer, "YUV4MPEG2 C%s W%u H%u F%u:%u I%c\n",
+ "420jpeg", width, height, fps_num, fps_den, 'p');
+ out_put(out, (unsigned char *)buffer,
+ (unsigned int)strlen(buffer), do_md5);
+ }
+
+ /* Try to determine the codec from the fourcc. */
+ for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++)
+ if ((fourcc & ifaces[i].fourcc_mask) == ifaces[i].fourcc)
+ {
+ vpx_codec_iface_t *ivf_iface = ifaces[i].iface;
+
+ if (iface && iface != ivf_iface)
+ fprintf(stderr, "Notice -- IVF header indicates codec: %s\n",
+ ifaces[i].name);
+ else
+ iface = ivf_iface;
+
+ break;
+ }
+
+ dec_flags = (postproc ? VPX_CODEC_USE_POSTPROC : 0) |
+ (ec_enabled ? VPX_CODEC_USE_ERROR_CONCEALMENT : 0);
+ if (vpx_codec_dec_init(&decoder, iface ? iface : ifaces[0].iface, &cfg,
+ dec_flags))
+ {
+ fprintf(stderr, "Failed to initialize decoder: %s\n", vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+
+ if (!quiet)
+ fprintf(stderr, "%s\n", decoder.name);
+
+#if CONFIG_VP8_DECODER
+
+ if (vp8_pp_cfg.post_proc_flag
+ && vpx_codec_control(&decoder, VP8_SET_POSTPROC, &vp8_pp_cfg))
+ {
+ fprintf(stderr, "Failed to configure postproc: %s\n", vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+
+ if (vp8_dbg_color_ref_frame
+ && vpx_codec_control(&decoder, VP8_SET_DBG_COLOR_REF_FRAME, vp8_dbg_color_ref_frame))
+ {
+ fprintf(stderr, "Failed to configure reference block visualizer: %s\n", vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+
+ if (vp8_dbg_color_mb_modes
+ && vpx_codec_control(&decoder, VP8_SET_DBG_COLOR_MB_MODES, vp8_dbg_color_mb_modes))
+ {
+ fprintf(stderr, "Failed to configure macro block visualizer: %s\n", vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+
+ if (vp8_dbg_color_b_modes
+ && vpx_codec_control(&decoder, VP8_SET_DBG_COLOR_B_MODES, vp8_dbg_color_b_modes))
+ {
+ fprintf(stderr, "Failed to configure block visualizer: %s\n", vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+
+ if (vp8_dbg_display_mv
+ && vpx_codec_control(&decoder, VP8_SET_DBG_DISPLAY_MV, vp8_dbg_display_mv))
+ {
+ fprintf(stderr, "Failed to configure motion vector visualizer: %s\n", vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+#endif
+
+ /* Decode file */
+ while (!read_frame(&input, &buf, &buf_sz, &buf_alloc_sz))
+ {
+ vpx_codec_iter_t iter = NULL;
+ vpx_image_t *img;
+ struct vpx_usec_timer timer;
+ int corrupted;
+
+ vpx_usec_timer_start(&timer);
+
+ if (vpx_codec_decode(&decoder, buf, (unsigned int)buf_sz, NULL, 0))
+ {
+ const char *detail = vpx_codec_error_detail(&decoder);
+ fprintf(stderr, "Failed to decode frame: %s\n", vpx_codec_error(&decoder));
+
+ if (detail)
+ fprintf(stderr, " Additional information: %s\n", detail);
+
+ goto fail;
+ }
+
+ vpx_usec_timer_mark(&timer);
+ dx_time += (unsigned int)vpx_usec_timer_elapsed(&timer);
+
+ ++frame_in;
+
+ if (vpx_codec_control(&decoder, VP8D_GET_FRAME_CORRUPTED, &corrupted))
+ {
+ fprintf(stderr, "Failed VP8_GET_FRAME_CORRUPTED: %s\n",
+ vpx_codec_error(&decoder));
+ goto fail;
+ }
+ frames_corrupted += corrupted;
+
+ vpx_usec_timer_start(&timer);
+
+ if ((img = vpx_codec_get_frame(&decoder, &iter)))
+ ++frame_out;
+
+ vpx_usec_timer_mark(&timer);
+ dx_time += (unsigned int)vpx_usec_timer_elapsed(&timer);
+
+ if (progress)
+ show_progress(frame_in, frame_out, dx_time);
+
+ if (!noblit)
+ {
+ if (img)
+ {
+ unsigned int y;
+ char out_fn[PATH_MAX];
+ uint8_t *buf;
+
+ if (!single_file)
+ {
+ size_t len = sizeof(out_fn)-1;
+
+ out_fn[len] = '\0';
+ generate_filename(outfile_pattern, out_fn, len-1,
+ img->d_w, img->d_h, frame_in);
+ out = out_open(out_fn, do_md5);
+ }
+ else if(use_y4m)
+ out_put(out, (unsigned char *)"FRAME\n", 6, do_md5);
+
+ buf = img->planes[VPX_PLANE_Y];
+
+ for (y = 0; y < img->d_h; y++)
+ {
+ out_put(out, buf, img->d_w, do_md5);
+ buf += img->stride[VPX_PLANE_Y];
+ }
+
+ buf = img->planes[flipuv?VPX_PLANE_V:VPX_PLANE_U];
+
+ for (y = 0; y < (1 + img->d_h) / 2; y++)
+ {
+ out_put(out, buf, (1 + img->d_w) / 2, do_md5);
+ buf += img->stride[VPX_PLANE_U];
+ }
+
+ buf = img->planes[flipuv?VPX_PLANE_U:VPX_PLANE_V];
+
+ for (y = 0; y < (1 + img->d_h) / 2; y++)
+ {
+ out_put(out, buf, (1 + img->d_w) / 2, do_md5);
+ buf += img->stride[VPX_PLANE_V];
+ }
+
+ if (!single_file)
+ out_close(out, out_fn, do_md5);
+ }
+ }
+
+ if (stop_after && frame_in >= stop_after)
+ break;
+ }
+
+ if (summary || progress)
+ {
+ show_progress(frame_in, frame_out, dx_time);
+ fprintf(stderr, "\n");
+ }
+
+ if (frames_corrupted)
+ fprintf(stderr, "WARNING: %d frames corrupted.\n",frames_corrupted);
+
+fail:
+
+ if (vpx_codec_destroy(&decoder))
+ {
+ fprintf(stderr, "Failed to destroy decoder: %s\n", vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+
+ if (single_file && !noblit)
+ out_close(out, outfile, do_md5);
+
+ if(input.nestegg_ctx)
+ nestegg_destroy(input.nestegg_ctx);
+ if(input.kind != WEBM_FILE)
+ free(buf);
+ fclose(infile);
+ free(argv);
+
+ return frames_corrupted ? EXIT_FAILURE : EXIT_SUCCESS;
+}
diff --git a/libvpx/vpxenc.c b/libvpx/vpxenc.c
new file mode 100644
index 0000000..7449e6c
--- /dev/null
+++ b/libvpx/vpxenc.c
@@ -0,0 +1,2567 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+/* This is a simple program that encodes YV12 files and generates ivf
+ * files using the new interface.
+ */
+#if defined(_WIN32) || !CONFIG_OS_SUPPORT
+#define USE_POSIX_MMAP 0
+#else
+#define USE_POSIX_MMAP 1
+#endif
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#include <limits.h>
+#include <assert.h>
+#include "vpx/vpx_encoder.h"
+#if USE_POSIX_MMAP
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/mman.h>
+#include <fcntl.h>
+#include <unistd.h>
+#endif
+#include "vpx/vp8cx.h"
+#include "vpx_ports/mem_ops.h"
+#include "vpx_ports/vpx_timer.h"
+#include "tools_common.h"
+#include "y4minput.h"
+#include "libmkv/EbmlWriter.h"
+#include "libmkv/EbmlIDs.h"
+
+/* Need special handling of these functions on Windows */
+#if defined(_MSC_VER)
+/* MSVS doesn't define off_t, and uses _f{seek,tell}i64 */
+typedef __int64 off_t;
+#define fseeko _fseeki64
+#define ftello _ftelli64
+#elif defined(_WIN32)
+/* MinGW defines off_t as long
+ and uses f{seek,tell}o64/off64_t for large files */
+#define fseeko fseeko64
+#define ftello ftello64
+#define off_t off64_t
+#endif
+
+#define LITERALU64(hi,lo) ((((uint64_t)hi)<<32)|lo)
+
+/* We should use 32-bit file operations in WebM file format
+ * when building ARM executable file (.axf) with RVCT */
+#if !CONFIG_OS_SUPPORT
+typedef long off_t;
+#define fseeko fseek
+#define ftello ftell
+#endif
+
+/* Swallow warnings about unused results of fread/fwrite */
+static size_t wrap_fread(void *ptr, size_t size, size_t nmemb,
+ FILE *stream)
+{
+ return fread(ptr, size, nmemb, stream);
+}
+#define fread wrap_fread
+
+static size_t wrap_fwrite(const void *ptr, size_t size, size_t nmemb,
+ FILE *stream)
+{
+ return fwrite(ptr, size, nmemb, stream);
+}
+#define fwrite wrap_fwrite
+
+
+static const char *exec_name;
+
+static const struct codec_item
+{
+ char const *name;
+ vpx_codec_iface_t *iface;
+ unsigned int fourcc;
+} codecs[] =
+{
+#if CONFIG_VP8_ENCODER
+ {"vp8", &vpx_codec_vp8_cx_algo, 0x30385056},
+#endif
+};
+
+static void usage_exit();
+
+#define LOG_ERROR(label) do \
+{\
+ const char *l=label;\
+ va_list ap;\
+ va_start(ap, fmt);\
+ if(l)\
+ fprintf(stderr, "%s: ", l);\
+ vfprintf(stderr, fmt, ap);\
+ fprintf(stderr, "\n");\
+ va_end(ap);\
+} while(0)
+
+void die(const char *fmt, ...)
+{
+ LOG_ERROR(NULL);
+ usage_exit();
+}
+
+
+void fatal(const char *fmt, ...)
+{
+ LOG_ERROR("Fatal");
+ exit(EXIT_FAILURE);
+}
+
+
+void warn(const char *fmt, ...)
+{
+ LOG_ERROR("Warning");
+}
+
+
+static void ctx_exit_on_error(vpx_codec_ctx_t *ctx, const char *s, ...)
+{
+ va_list ap;
+
+ va_start(ap, s);
+ if (ctx->err)
+ {
+ const char *detail = vpx_codec_error_detail(ctx);
+
+ vfprintf(stderr, s, ap);
+ fprintf(stderr, ": %s\n", vpx_codec_error(ctx));
+
+ if (detail)
+ fprintf(stderr, " %s\n", detail);
+
+ exit(EXIT_FAILURE);
+ }
+}
+
+/* This structure is used to abstract the different ways of handling
+ * first pass statistics.
+ */
+typedef struct
+{
+ vpx_fixed_buf_t buf;
+ int pass;
+ FILE *file;
+ char *buf_ptr;
+ size_t buf_alloc_sz;
+} stats_io_t;
+
+int stats_open_file(stats_io_t *stats, const char *fpf, int pass)
+{
+ int res;
+
+ stats->pass = pass;
+
+ if (pass == 0)
+ {
+ stats->file = fopen(fpf, "wb");
+ stats->buf.sz = 0;
+ stats->buf.buf = NULL,
+ res = (stats->file != NULL);
+ }
+ else
+ {
+#if 0
+#elif USE_POSIX_MMAP
+ struct stat stat_buf;
+ int fd;
+
+ fd = open(fpf, O_RDONLY);
+ stats->file = fdopen(fd, "rb");
+ fstat(fd, &stat_buf);
+ stats->buf.sz = stat_buf.st_size;
+ stats->buf.buf = mmap(NULL, stats->buf.sz, PROT_READ, MAP_PRIVATE,
+ fd, 0);
+ res = (stats->buf.buf != NULL);
+#else
+ size_t nbytes;
+
+ stats->file = fopen(fpf, "rb");
+
+ if (fseek(stats->file, 0, SEEK_END))
+ fatal("First-pass stats file must be seekable!");
+
+ stats->buf.sz = stats->buf_alloc_sz = ftell(stats->file);
+ rewind(stats->file);
+
+ stats->buf.buf = malloc(stats->buf_alloc_sz);
+
+ if (!stats->buf.buf)
+ fatal("Failed to allocate first-pass stats buffer (%lu bytes)",
+ (unsigned long)stats->buf_alloc_sz);
+
+ nbytes = fread(stats->buf.buf, 1, stats->buf.sz, stats->file);
+ res = (nbytes == stats->buf.sz);
+#endif
+ }
+
+ return res;
+}
+
+int stats_open_mem(stats_io_t *stats, int pass)
+{
+ int res;
+ stats->pass = pass;
+
+ if (!pass)
+ {
+ stats->buf.sz = 0;
+ stats->buf_alloc_sz = 64 * 1024;
+ stats->buf.buf = malloc(stats->buf_alloc_sz);
+ }
+
+ stats->buf_ptr = stats->buf.buf;
+ res = (stats->buf.buf != NULL);
+ return res;
+}
+
+
+void stats_close(stats_io_t *stats, int last_pass)
+{
+ if (stats->file)
+ {
+ if (stats->pass == last_pass)
+ {
+#if 0
+#elif USE_POSIX_MMAP
+ munmap(stats->buf.buf, stats->buf.sz);
+#else
+ free(stats->buf.buf);
+#endif
+ }
+
+ fclose(stats->file);
+ stats->file = NULL;
+ }
+ else
+ {
+ if (stats->pass == last_pass)
+ free(stats->buf.buf);
+ }
+}
+
+void stats_write(stats_io_t *stats, const void *pkt, size_t len)
+{
+ if (stats->file)
+ {
+ (void) fwrite(pkt, 1, len, stats->file);
+ }
+ else
+ {
+ if (stats->buf.sz + len > stats->buf_alloc_sz)
+ {
+ size_t new_sz = stats->buf_alloc_sz + 64 * 1024;
+ char *new_ptr = realloc(stats->buf.buf, new_sz);
+
+ if (new_ptr)
+ {
+ stats->buf_ptr = new_ptr + (stats->buf_ptr - (char *)stats->buf.buf);
+ stats->buf.buf = new_ptr;
+ stats->buf_alloc_sz = new_sz;
+ }
+ else
+ fatal("Failed to realloc firstpass stats buffer.");
+ }
+
+ memcpy(stats->buf_ptr, pkt, len);
+ stats->buf.sz += len;
+ stats->buf_ptr += len;
+ }
+}
+
+vpx_fixed_buf_t stats_get(stats_io_t *stats)
+{
+ return stats->buf;
+}
+
+/* Stereo 3D packed frame format */
+typedef enum stereo_format
+{
+ STEREO_FORMAT_MONO = 0,
+ STEREO_FORMAT_LEFT_RIGHT = 1,
+ STEREO_FORMAT_BOTTOM_TOP = 2,
+ STEREO_FORMAT_TOP_BOTTOM = 3,
+ STEREO_FORMAT_RIGHT_LEFT = 11
+} stereo_format_t;
+
+enum video_file_type
+{
+ FILE_TYPE_RAW,
+ FILE_TYPE_IVF,
+ FILE_TYPE_Y4M
+};
+
+struct detect_buffer {
+ char buf[4];
+ size_t buf_read;
+ size_t position;
+};
+
+
+struct input_state
+{
+ char *fn;
+ FILE *file;
+ y4m_input y4m;
+ struct detect_buffer detect;
+ enum video_file_type file_type;
+ unsigned int w;
+ unsigned int h;
+ struct vpx_rational framerate;
+ int use_i420;
+};
+
+
+#define IVF_FRAME_HDR_SZ (4+8) /* 4 byte size + 8 byte timestamp */
+static int read_frame(struct input_state *input, vpx_image_t *img)
+{
+ FILE *f = input->file;
+ enum video_file_type file_type = input->file_type;
+ y4m_input *y4m = &input->y4m;
+ struct detect_buffer *detect = &input->detect;
+ int plane = 0;
+ int shortread = 0;
+
+ if (file_type == FILE_TYPE_Y4M)
+ {
+ if (y4m_input_fetch_frame(y4m, f, img) < 1)
+ return 0;
+ }
+ else
+ {
+ if (file_type == FILE_TYPE_IVF)
+ {
+ char junk[IVF_FRAME_HDR_SZ];
+
+ /* Skip the frame header. We know how big the frame should be. See
+ * write_ivf_frame_header() for documentation on the frame header
+ * layout.
+ */
+ (void) fread(junk, 1, IVF_FRAME_HDR_SZ, f);
+ }
+
+ for (plane = 0; plane < 3; plane++)
+ {
+ unsigned char *ptr;
+ int w = (plane ? (1 + img->d_w) / 2 : img->d_w);
+ int h = (plane ? (1 + img->d_h) / 2 : img->d_h);
+ int r;
+
+ /* Determine the correct plane based on the image format. The for-loop
+ * always counts in Y,U,V order, but this may not match the order of
+ * the data on disk.
+ */
+ switch (plane)
+ {
+ case 1:
+ ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12? VPX_PLANE_V : VPX_PLANE_U];
+ break;
+ case 2:
+ ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12?VPX_PLANE_U : VPX_PLANE_V];
+ break;
+ default:
+ ptr = img->planes[plane];
+ }
+
+ for (r = 0; r < h; r++)
+ {
+ size_t needed = w;
+ size_t buf_position = 0;
+ const size_t left = detect->buf_read - detect->position;
+ if (left > 0)
+ {
+ const size_t more = (left < needed) ? left : needed;
+ memcpy(ptr, detect->buf + detect->position, more);
+ buf_position = more;
+ needed -= more;
+ detect->position += more;
+ }
+ if (needed > 0)
+ {
+ shortread |= (fread(ptr + buf_position, 1, needed, f) < needed);
+ }
+
+ ptr += img->stride[plane];
+ }
+ }
+ }
+
+ return !shortread;
+}
+
+
+unsigned int file_is_y4m(FILE *infile,
+ y4m_input *y4m,
+ char detect[4])
+{
+ if(memcmp(detect, "YUV4", 4) == 0)
+ {
+ return 1;
+ }
+ return 0;
+}
+
+#define IVF_FILE_HDR_SZ (32)
+unsigned int file_is_ivf(struct input_state *input,
+ unsigned int *fourcc)
+{
+ char raw_hdr[IVF_FILE_HDR_SZ];
+ int is_ivf = 0;
+ FILE *infile = input->file;
+ unsigned int *width = &input->w;
+ unsigned int *height = &input->h;
+ struct detect_buffer *detect = &input->detect;
+
+ if(memcmp(detect->buf, "DKIF", 4) != 0)
+ return 0;
+
+ /* See write_ivf_file_header() for more documentation on the file header
+ * layout.
+ */
+ if (fread(raw_hdr + 4, 1, IVF_FILE_HDR_SZ - 4, infile)
+ == IVF_FILE_HDR_SZ - 4)
+ {
+ {
+ is_ivf = 1;
+
+ if (mem_get_le16(raw_hdr + 4) != 0)
+ warn("Unrecognized IVF version! This file may not decode "
+ "properly.");
+
+ *fourcc = mem_get_le32(raw_hdr + 8);
+ }
+ }
+
+ if (is_ivf)
+ {
+ *width = mem_get_le16(raw_hdr + 12);
+ *height = mem_get_le16(raw_hdr + 14);
+ detect->position = 4;
+ }
+
+ return is_ivf;
+}
+
+
+static void write_ivf_file_header(FILE *outfile,
+ const vpx_codec_enc_cfg_t *cfg,
+ unsigned int fourcc,
+ int frame_cnt)
+{
+ char header[32];
+
+ if (cfg->g_pass != VPX_RC_ONE_PASS && cfg->g_pass != VPX_RC_LAST_PASS)
+ return;
+
+ header[0] = 'D';
+ header[1] = 'K';
+ header[2] = 'I';
+ header[3] = 'F';
+ mem_put_le16(header + 4, 0); /* version */
+ mem_put_le16(header + 6, 32); /* headersize */
+ mem_put_le32(header + 8, fourcc); /* headersize */
+ mem_put_le16(header + 12, cfg->g_w); /* width */
+ mem_put_le16(header + 14, cfg->g_h); /* height */
+ mem_put_le32(header + 16, cfg->g_timebase.den); /* rate */
+ mem_put_le32(header + 20, cfg->g_timebase.num); /* scale */
+ mem_put_le32(header + 24, frame_cnt); /* length */
+ mem_put_le32(header + 28, 0); /* unused */
+
+ (void) fwrite(header, 1, 32, outfile);
+}
+
+
+static void write_ivf_frame_header(FILE *outfile,
+ const vpx_codec_cx_pkt_t *pkt)
+{
+ char header[12];
+ vpx_codec_pts_t pts;
+
+ if (pkt->kind != VPX_CODEC_CX_FRAME_PKT)
+ return;
+
+ pts = pkt->data.frame.pts;
+ mem_put_le32(header, (int)pkt->data.frame.sz);
+ mem_put_le32(header + 4, pts & 0xFFFFFFFF);
+ mem_put_le32(header + 8, pts >> 32);
+
+ (void) fwrite(header, 1, 12, outfile);
+}
+
+static void write_ivf_frame_size(FILE *outfile, size_t size)
+{
+ char header[4];
+ mem_put_le32(header, (int)size);
+ (void) fwrite(header, 1, 4, outfile);
+}
+
+
+typedef off_t EbmlLoc;
+
+
+struct cue_entry
+{
+ unsigned int time;
+ uint64_t loc;
+};
+
+
+struct EbmlGlobal
+{
+ int debug;
+
+ FILE *stream;
+ int64_t last_pts_ms;
+ vpx_rational_t framerate;
+
+ /* These pointers are to the start of an element */
+ off_t position_reference;
+ off_t seek_info_pos;
+ off_t segment_info_pos;
+ off_t track_pos;
+ off_t cue_pos;
+ off_t cluster_pos;
+
+ /* This pointer is to a specific element to be serialized */
+ off_t track_id_pos;
+
+ /* These pointers are to the size field of the element */
+ EbmlLoc startSegment;
+ EbmlLoc startCluster;
+
+ uint32_t cluster_timecode;
+ int cluster_open;
+
+ struct cue_entry *cue_list;
+ unsigned int cues;
+
+};
+
+
+void Ebml_Write(EbmlGlobal *glob, const void *buffer_in, unsigned long len)
+{
+ (void) fwrite(buffer_in, 1, len, glob->stream);
+}
+
+#define WRITE_BUFFER(s) \
+for(i = len-1; i>=0; i--)\
+{ \
+ x = (char)(*(const s *)buffer_in >> (i * CHAR_BIT)); \
+ Ebml_Write(glob, &x, 1); \
+}
+void Ebml_Serialize(EbmlGlobal *glob, const void *buffer_in, int buffer_size, unsigned long len)
+{
+ char x;
+ int i;
+
+ /* buffer_size:
+ * 1 - int8_t;
+ * 2 - int16_t;
+ * 3 - int32_t;
+ * 4 - int64_t;
+ */
+ switch (buffer_size)
+ {
+ case 1:
+ WRITE_BUFFER(int8_t)
+ break;
+ case 2:
+ WRITE_BUFFER(int16_t)
+ break;
+ case 4:
+ WRITE_BUFFER(int32_t)
+ break;
+ case 8:
+ WRITE_BUFFER(int64_t)
+ break;
+ default:
+ break;
+ }
+}
+#undef WRITE_BUFFER
+
+/* Need a fixed size serializer for the track ID. libmkv provides a 64 bit
+ * one, but not a 32 bit one.
+ */
+static void Ebml_SerializeUnsigned32(EbmlGlobal *glob, unsigned long class_id, uint64_t ui)
+{
+ unsigned char sizeSerialized = 4 | 0x80;
+ Ebml_WriteID(glob, class_id);
+ Ebml_Serialize(glob, &sizeSerialized, sizeof(sizeSerialized), 1);
+ Ebml_Serialize(glob, &ui, sizeof(ui), 4);
+}
+
+
+static void
+Ebml_StartSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc,
+ unsigned long class_id)
+{
+ /* todo this is always taking 8 bytes, this may need later optimization */
+ /* this is a key that says length unknown */
+ uint64_t unknownLen = LITERALU64(0x01FFFFFF, 0xFFFFFFFF);
+
+ Ebml_WriteID(glob, class_id);
+ *ebmlLoc = ftello(glob->stream);
+ Ebml_Serialize(glob, &unknownLen, sizeof(unknownLen), 8);
+}
+
+static void
+Ebml_EndSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc)
+{
+ off_t pos;
+ uint64_t size;
+
+ /* Save the current stream pointer */
+ pos = ftello(glob->stream);
+
+ /* Calculate the size of this element */
+ size = pos - *ebmlLoc - 8;
+ size |= LITERALU64(0x01000000,0x00000000);
+
+ /* Seek back to the beginning of the element and write the new size */
+ fseeko(glob->stream, *ebmlLoc, SEEK_SET);
+ Ebml_Serialize(glob, &size, sizeof(size), 8);
+
+ /* Reset the stream pointer */
+ fseeko(glob->stream, pos, SEEK_SET);
+}
+
+
+static void
+write_webm_seek_element(EbmlGlobal *ebml, unsigned long id, off_t pos)
+{
+ uint64_t offset = pos - ebml->position_reference;
+ EbmlLoc start;
+ Ebml_StartSubElement(ebml, &start, Seek);
+ Ebml_SerializeBinary(ebml, SeekID, id);
+ Ebml_SerializeUnsigned64(ebml, SeekPosition, offset);
+ Ebml_EndSubElement(ebml, &start);
+}
+
+
+static void
+write_webm_seek_info(EbmlGlobal *ebml)
+{
+
+ off_t pos;
+
+ /* Save the current stream pointer */
+ pos = ftello(ebml->stream);
+
+ if(ebml->seek_info_pos)
+ fseeko(ebml->stream, ebml->seek_info_pos, SEEK_SET);
+ else
+ ebml->seek_info_pos = pos;
+
+ {
+ EbmlLoc start;
+
+ Ebml_StartSubElement(ebml, &start, SeekHead);
+ write_webm_seek_element(ebml, Tracks, ebml->track_pos);
+ write_webm_seek_element(ebml, Cues, ebml->cue_pos);
+ write_webm_seek_element(ebml, Info, ebml->segment_info_pos);
+ Ebml_EndSubElement(ebml, &start);
+ }
+ {
+ /* segment info */
+ EbmlLoc startInfo;
+ uint64_t frame_time;
+ char version_string[64];
+
+ /* Assemble version string */
+ if(ebml->debug)
+ strcpy(version_string, "vpxenc");
+ else
+ {
+ strcpy(version_string, "vpxenc ");
+ strncat(version_string,
+ vpx_codec_version_str(),
+ sizeof(version_string) - 1 - strlen(version_string));
+ }
+
+ frame_time = (uint64_t)1000 * ebml->framerate.den
+ / ebml->framerate.num;
+ ebml->segment_info_pos = ftello(ebml->stream);
+ Ebml_StartSubElement(ebml, &startInfo, Info);
+ Ebml_SerializeUnsigned(ebml, TimecodeScale, 1000000);
+ Ebml_SerializeFloat(ebml, Segment_Duration,
+ (double)(ebml->last_pts_ms + frame_time));
+ Ebml_SerializeString(ebml, 0x4D80, version_string);
+ Ebml_SerializeString(ebml, 0x5741, version_string);
+ Ebml_EndSubElement(ebml, &startInfo);
+ }
+}
+
+
+static void
+write_webm_file_header(EbmlGlobal *glob,
+ const vpx_codec_enc_cfg_t *cfg,
+ const struct vpx_rational *fps,
+ stereo_format_t stereo_fmt)
+{
+ {
+ EbmlLoc start;
+ Ebml_StartSubElement(glob, &start, EBML);
+ Ebml_SerializeUnsigned(glob, EBMLVersion, 1);
+ Ebml_SerializeUnsigned(glob, EBMLReadVersion, 1);
+ Ebml_SerializeUnsigned(glob, EBMLMaxIDLength, 4);
+ Ebml_SerializeUnsigned(glob, EBMLMaxSizeLength, 8);
+ Ebml_SerializeString(glob, DocType, "webm");
+ Ebml_SerializeUnsigned(glob, DocTypeVersion, 2);
+ Ebml_SerializeUnsigned(glob, DocTypeReadVersion, 2);
+ Ebml_EndSubElement(glob, &start);
+ }
+ {
+ Ebml_StartSubElement(glob, &glob->startSegment, Segment);
+ glob->position_reference = ftello(glob->stream);
+ glob->framerate = *fps;
+ write_webm_seek_info(glob);
+
+ {
+ EbmlLoc trackStart;
+ glob->track_pos = ftello(glob->stream);
+ Ebml_StartSubElement(glob, &trackStart, Tracks);
+ {
+ unsigned int trackNumber = 1;
+ uint64_t trackID = 0;
+
+ EbmlLoc start;
+ Ebml_StartSubElement(glob, &start, TrackEntry);
+ Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber);
+ glob->track_id_pos = ftello(glob->stream);
+ Ebml_SerializeUnsigned32(glob, TrackUID, trackID);
+ Ebml_SerializeUnsigned(glob, TrackType, 1);
+ Ebml_SerializeString(glob, CodecID, "V_VP8");
+ {
+ unsigned int pixelWidth = cfg->g_w;
+ unsigned int pixelHeight = cfg->g_h;
+ float frameRate = (float)fps->num/(float)fps->den;
+
+ EbmlLoc videoStart;
+ Ebml_StartSubElement(glob, &videoStart, Video);
+ Ebml_SerializeUnsigned(glob, PixelWidth, pixelWidth);
+ Ebml_SerializeUnsigned(glob, PixelHeight, pixelHeight);
+ Ebml_SerializeUnsigned(glob, StereoMode, stereo_fmt);
+ Ebml_SerializeFloat(glob, FrameRate, frameRate);
+ Ebml_EndSubElement(glob, &videoStart);
+ }
+ Ebml_EndSubElement(glob, &start); /* Track Entry */
+ }
+ Ebml_EndSubElement(glob, &trackStart);
+ }
+ /* segment element is open */
+ }
+}
+
+
+static void
+write_webm_block(EbmlGlobal *glob,
+ const vpx_codec_enc_cfg_t *cfg,
+ const vpx_codec_cx_pkt_t *pkt)
+{
+ unsigned long block_length;
+ unsigned char track_number;
+ unsigned short block_timecode = 0;
+ unsigned char flags;
+ int64_t pts_ms;
+ int start_cluster = 0, is_keyframe;
+
+ /* Calculate the PTS of this frame in milliseconds */
+ pts_ms = pkt->data.frame.pts * 1000
+ * (uint64_t)cfg->g_timebase.num / (uint64_t)cfg->g_timebase.den;
+ if(pts_ms <= glob->last_pts_ms)
+ pts_ms = glob->last_pts_ms + 1;
+ glob->last_pts_ms = pts_ms;
+
+ /* Calculate the relative time of this block */
+ if(pts_ms - glob->cluster_timecode > SHRT_MAX)
+ start_cluster = 1;
+ else
+ block_timecode = (unsigned short)pts_ms - glob->cluster_timecode;
+
+ is_keyframe = (pkt->data.frame.flags & VPX_FRAME_IS_KEY);
+ if(start_cluster || is_keyframe)
+ {
+ if(glob->cluster_open)
+ Ebml_EndSubElement(glob, &glob->startCluster);
+
+ /* Open the new cluster */
+ block_timecode = 0;
+ glob->cluster_open = 1;
+ glob->cluster_timecode = (uint32_t)pts_ms;
+ glob->cluster_pos = ftello(glob->stream);
+ Ebml_StartSubElement(glob, &glob->startCluster, Cluster); /* cluster */
+ Ebml_SerializeUnsigned(glob, Timecode, glob->cluster_timecode);
+
+ /* Save a cue point if this is a keyframe. */
+ if(is_keyframe)
+ {
+ struct cue_entry *cue, *new_cue_list;
+
+ new_cue_list = realloc(glob->cue_list,
+ (glob->cues+1) * sizeof(struct cue_entry));
+ if(new_cue_list)
+ glob->cue_list = new_cue_list;
+ else
+ fatal("Failed to realloc cue list.");
+
+ cue = &glob->cue_list[glob->cues];
+ cue->time = glob->cluster_timecode;
+ cue->loc = glob->cluster_pos;
+ glob->cues++;
+ }
+ }
+
+ /* Write the Simple Block */
+ Ebml_WriteID(glob, SimpleBlock);
+
+ block_length = (unsigned long)pkt->data.frame.sz + 4;
+ block_length |= 0x10000000;
+ Ebml_Serialize(glob, &block_length, sizeof(block_length), 4);
+
+ track_number = 1;
+ track_number |= 0x80;
+ Ebml_Write(glob, &track_number, 1);
+
+ Ebml_Serialize(glob, &block_timecode, sizeof(block_timecode), 2);
+
+ flags = 0;
+ if(is_keyframe)
+ flags |= 0x80;
+ if(pkt->data.frame.flags & VPX_FRAME_IS_INVISIBLE)
+ flags |= 0x08;
+ Ebml_Write(glob, &flags, 1);
+
+ Ebml_Write(glob, pkt->data.frame.buf, (unsigned long)pkt->data.frame.sz);
+}
+
+
+static void
+write_webm_file_footer(EbmlGlobal *glob, long hash)
+{
+
+ if(glob->cluster_open)
+ Ebml_EndSubElement(glob, &glob->startCluster);
+
+ {
+ EbmlLoc start;
+ unsigned int i;
+
+ glob->cue_pos = ftello(glob->stream);
+ Ebml_StartSubElement(glob, &start, Cues);
+ for(i=0; i<glob->cues; i++)
+ {
+ struct cue_entry *cue = &glob->cue_list[i];
+ EbmlLoc start;
+
+ Ebml_StartSubElement(glob, &start, CuePoint);
+ {
+ EbmlLoc start;
+
+ Ebml_SerializeUnsigned(glob, CueTime, cue->time);
+
+ Ebml_StartSubElement(glob, &start, CueTrackPositions);
+ Ebml_SerializeUnsigned(glob, CueTrack, 1);
+ Ebml_SerializeUnsigned64(glob, CueClusterPosition,
+ cue->loc - glob->position_reference);
+ Ebml_EndSubElement(glob, &start);
+ }
+ Ebml_EndSubElement(glob, &start);
+ }
+ Ebml_EndSubElement(glob, &start);
+ }
+
+ Ebml_EndSubElement(glob, &glob->startSegment);
+
+ /* Patch up the seek info block */
+ write_webm_seek_info(glob);
+
+ /* Patch up the track id */
+ fseeko(glob->stream, glob->track_id_pos, SEEK_SET);
+ Ebml_SerializeUnsigned32(glob, TrackUID, glob->debug ? 0xDEADBEEF : hash);
+
+ fseeko(glob->stream, 0, SEEK_END);
+}
+
+
+/* Murmur hash derived from public domain reference implementation at
+ * http://sites.google.com/site/murmurhash/
+ */
+static unsigned int murmur ( const void * key, int len, unsigned int seed )
+{
+ const unsigned int m = 0x5bd1e995;
+ const int r = 24;
+
+ unsigned int h = seed ^ len;
+
+ const unsigned char * data = (const unsigned char *)key;
+
+ while(len >= 4)
+ {
+ unsigned int k;
+
+ k = data[0];
+ k |= data[1] << 8;
+ k |= data[2] << 16;
+ k |= data[3] << 24;
+
+ k *= m;
+ k ^= k >> r;
+ k *= m;
+
+ h *= m;
+ h ^= k;
+
+ data += 4;
+ len -= 4;
+ }
+
+ switch(len)
+ {
+ case 3: h ^= data[2] << 16;
+ case 2: h ^= data[1] << 8;
+ case 1: h ^= data[0];
+ h *= m;
+ };
+
+ h ^= h >> 13;
+ h *= m;
+ h ^= h >> 15;
+
+ return h;
+}
+
+#include "math.h"
+
+static double vp8_mse2psnr(double Samples, double Peak, double Mse)
+{
+ double psnr;
+
+ if ((double)Mse > 0.0)
+ psnr = 10.0 * log10(Peak * Peak * Samples / Mse);
+ else
+ psnr = 60; /* Limit to prevent / 0 */
+
+ if (psnr > 60)
+ psnr = 60;
+
+ return psnr;
+}
+
+
+#include "args.h"
+static const arg_def_t debugmode = ARG_DEF("D", "debug", 0,
+ "Debug mode (makes output deterministic)");
+static const arg_def_t outputfile = ARG_DEF("o", "output", 1,
+ "Output filename");
+static const arg_def_t use_yv12 = ARG_DEF(NULL, "yv12", 0,
+ "Input file is YV12 ");
+static const arg_def_t use_i420 = ARG_DEF(NULL, "i420", 0,
+ "Input file is I420 (default)");
+static const arg_def_t codecarg = ARG_DEF(NULL, "codec", 1,
+ "Codec to use");
+static const arg_def_t passes = ARG_DEF("p", "passes", 1,
+ "Number of passes (1/2)");
+static const arg_def_t pass_arg = ARG_DEF(NULL, "pass", 1,
+ "Pass to execute (1/2)");
+static const arg_def_t fpf_name = ARG_DEF(NULL, "fpf", 1,
+ "First pass statistics file name");
+static const arg_def_t limit = ARG_DEF(NULL, "limit", 1,
+ "Stop encoding after n input frames");
+static const arg_def_t deadline = ARG_DEF("d", "deadline", 1,
+ "Deadline per frame (usec)");
+static const arg_def_t best_dl = ARG_DEF(NULL, "best", 0,
+ "Use Best Quality Deadline");
+static const arg_def_t good_dl = ARG_DEF(NULL, "good", 0,
+ "Use Good Quality Deadline");
+static const arg_def_t rt_dl = ARG_DEF(NULL, "rt", 0,
+ "Use Realtime Quality Deadline");
+static const arg_def_t quietarg = ARG_DEF("q", "quiet", 0,
+ "Do not print encode progress");
+static const arg_def_t verbosearg = ARG_DEF("v", "verbose", 0,
+ "Show encoder parameters");
+static const arg_def_t psnrarg = ARG_DEF(NULL, "psnr", 0,
+ "Show PSNR in status line");
+static const arg_def_t framerate = ARG_DEF(NULL, "fps", 1,
+ "Stream frame rate (rate/scale)");
+static const arg_def_t use_ivf = ARG_DEF(NULL, "ivf", 0,
+ "Output IVF (default is WebM)");
+static const arg_def_t out_part = ARG_DEF("P", "output-partitions", 0,
+ "Makes encoder output partitions. Requires IVF output!");
+static const arg_def_t q_hist_n = ARG_DEF(NULL, "q-hist", 1,
+ "Show quantizer histogram (n-buckets)");
+static const arg_def_t rate_hist_n = ARG_DEF(NULL, "rate-hist", 1,
+ "Show rate histogram (n-buckets)");
+static const arg_def_t *main_args[] =
+{
+ &debugmode,
+ &outputfile, &codecarg, &passes, &pass_arg, &fpf_name, &limit, &deadline,
+ &best_dl, &good_dl, &rt_dl,
+ &quietarg, &verbosearg, &psnrarg, &use_ivf, &out_part, &q_hist_n, &rate_hist_n,
+ NULL
+};
+
+static const arg_def_t usage = ARG_DEF("u", "usage", 1,
+ "Usage profile number to use");
+static const arg_def_t threads = ARG_DEF("t", "threads", 1,
+ "Max number of threads to use");
+static const arg_def_t profile = ARG_DEF(NULL, "profile", 1,
+ "Bitstream profile number to use");
+static const arg_def_t width = ARG_DEF("w", "width", 1,
+ "Frame width");
+static const arg_def_t height = ARG_DEF("h", "height", 1,
+ "Frame height");
+static const struct arg_enum_list stereo_mode_enum[] = {
+ {"mono" , STEREO_FORMAT_MONO},
+ {"left-right", STEREO_FORMAT_LEFT_RIGHT},
+ {"bottom-top", STEREO_FORMAT_BOTTOM_TOP},
+ {"top-bottom", STEREO_FORMAT_TOP_BOTTOM},
+ {"right-left", STEREO_FORMAT_RIGHT_LEFT},
+ {NULL, 0}
+};
+static const arg_def_t stereo_mode = ARG_DEF_ENUM(NULL, "stereo-mode", 1,
+ "Stereo 3D video format", stereo_mode_enum);
+static const arg_def_t timebase = ARG_DEF(NULL, "timebase", 1,
+ "Output timestamp precision (fractional seconds)");
+static const arg_def_t error_resilient = ARG_DEF(NULL, "error-resilient", 1,
+ "Enable error resiliency features");
+static const arg_def_t lag_in_frames = ARG_DEF(NULL, "lag-in-frames", 1,
+ "Max number of frames to lag");
+
+static const arg_def_t *global_args[] =
+{
+ &use_yv12, &use_i420, &usage, &threads, &profile,
+ &width, &height, &stereo_mode, &timebase, &framerate, &error_resilient,
+ &lag_in_frames, NULL
+};
+
+static const arg_def_t dropframe_thresh = ARG_DEF(NULL, "drop-frame", 1,
+ "Temporal resampling threshold (buf %)");
+static const arg_def_t resize_allowed = ARG_DEF(NULL, "resize-allowed", 1,
+ "Spatial resampling enabled (bool)");
+static const arg_def_t resize_up_thresh = ARG_DEF(NULL, "resize-up", 1,
+ "Upscale threshold (buf %)");
+static const arg_def_t resize_down_thresh = ARG_DEF(NULL, "resize-down", 1,
+ "Downscale threshold (buf %)");
+static const struct arg_enum_list end_usage_enum[] = {
+ {"vbr", VPX_VBR},
+ {"cbr", VPX_CBR},
+ {"cq", VPX_CQ},
+ {NULL, 0}
+};
+static const arg_def_t end_usage = ARG_DEF_ENUM(NULL, "end-usage", 1,
+ "Rate control mode", end_usage_enum);
+static const arg_def_t target_bitrate = ARG_DEF(NULL, "target-bitrate", 1,
+ "Bitrate (kbps)");
+static const arg_def_t min_quantizer = ARG_DEF(NULL, "min-q", 1,
+ "Minimum (best) quantizer");
+static const arg_def_t max_quantizer = ARG_DEF(NULL, "max-q", 1,
+ "Maximum (worst) quantizer");
+static const arg_def_t undershoot_pct = ARG_DEF(NULL, "undershoot-pct", 1,
+ "Datarate undershoot (min) target (%)");
+static const arg_def_t overshoot_pct = ARG_DEF(NULL, "overshoot-pct", 1,
+ "Datarate overshoot (max) target (%)");
+static const arg_def_t buf_sz = ARG_DEF(NULL, "buf-sz", 1,
+ "Client buffer size (ms)");
+static const arg_def_t buf_initial_sz = ARG_DEF(NULL, "buf-initial-sz", 1,
+ "Client initial buffer size (ms)");
+static const arg_def_t buf_optimal_sz = ARG_DEF(NULL, "buf-optimal-sz", 1,
+ "Client optimal buffer size (ms)");
+static const arg_def_t *rc_args[] =
+{
+ &dropframe_thresh, &resize_allowed, &resize_up_thresh, &resize_down_thresh,
+ &end_usage, &target_bitrate, &min_quantizer, &max_quantizer,
+ &undershoot_pct, &overshoot_pct, &buf_sz, &buf_initial_sz, &buf_optimal_sz,
+ NULL
+};
+
+
+static const arg_def_t bias_pct = ARG_DEF(NULL, "bias-pct", 1,
+ "CBR/VBR bias (0=CBR, 100=VBR)");
+static const arg_def_t minsection_pct = ARG_DEF(NULL, "minsection-pct", 1,
+ "GOP min bitrate (% of target)");
+static const arg_def_t maxsection_pct = ARG_DEF(NULL, "maxsection-pct", 1,
+ "GOP max bitrate (% of target)");
+static const arg_def_t *rc_twopass_args[] =
+{
+ &bias_pct, &minsection_pct, &maxsection_pct, NULL
+};
+
+
+static const arg_def_t kf_min_dist = ARG_DEF(NULL, "kf-min-dist", 1,
+ "Minimum keyframe interval (frames)");
+static const arg_def_t kf_max_dist = ARG_DEF(NULL, "kf-max-dist", 1,
+ "Maximum keyframe interval (frames)");
+static const arg_def_t kf_disabled = ARG_DEF(NULL, "disable-kf", 0,
+ "Disable keyframe placement");
+static const arg_def_t *kf_args[] =
+{
+ &kf_min_dist, &kf_max_dist, &kf_disabled, NULL
+};
+
+
+#if CONFIG_VP8_ENCODER
+static const arg_def_t noise_sens = ARG_DEF(NULL, "noise-sensitivity", 1,
+ "Noise sensitivity (frames to blur)");
+static const arg_def_t sharpness = ARG_DEF(NULL, "sharpness", 1,
+ "Filter sharpness (0-7)");
+static const arg_def_t static_thresh = ARG_DEF(NULL, "static-thresh", 1,
+ "Motion detection threshold");
+#endif
+
+#if CONFIG_VP8_ENCODER
+static const arg_def_t cpu_used = ARG_DEF(NULL, "cpu-used", 1,
+ "CPU Used (-16..16)");
+#endif
+
+
+#if CONFIG_VP8_ENCODER
+static const arg_def_t token_parts = ARG_DEF(NULL, "token-parts", 1,
+ "Number of token partitions to use, log2");
+static const arg_def_t auto_altref = ARG_DEF(NULL, "auto-alt-ref", 1,
+ "Enable automatic alt reference frames");
+static const arg_def_t arnr_maxframes = ARG_DEF(NULL, "arnr-maxframes", 1,
+ "AltRef Max Frames");
+static const arg_def_t arnr_strength = ARG_DEF(NULL, "arnr-strength", 1,
+ "AltRef Strength");
+static const arg_def_t arnr_type = ARG_DEF(NULL, "arnr-type", 1,
+ "AltRef Type");
+static const struct arg_enum_list tuning_enum[] = {
+ {"psnr", VP8_TUNE_PSNR},
+ {"ssim", VP8_TUNE_SSIM},
+ {NULL, 0}
+};
+static const arg_def_t tune_ssim = ARG_DEF_ENUM(NULL, "tune", 1,
+ "Material to favor", tuning_enum);
+static const arg_def_t cq_level = ARG_DEF(NULL, "cq-level", 1,
+ "Constrained Quality Level");
+static const arg_def_t max_intra_rate_pct = ARG_DEF(NULL, "max-intra-rate", 1,
+ "Max I-frame bitrate (pct)");
+
+static const arg_def_t *vp8_args[] =
+{
+ &cpu_used, &auto_altref, &noise_sens, &sharpness, &static_thresh,
+ &token_parts, &arnr_maxframes, &arnr_strength, &arnr_type,
+ &tune_ssim, &cq_level, &max_intra_rate_pct, NULL
+};
+static const int vp8_arg_ctrl_map[] =
+{
+ VP8E_SET_CPUUSED, VP8E_SET_ENABLEAUTOALTREF,
+ VP8E_SET_NOISE_SENSITIVITY, VP8E_SET_SHARPNESS, VP8E_SET_STATIC_THRESHOLD,
+ VP8E_SET_TOKEN_PARTITIONS,
+ VP8E_SET_ARNR_MAXFRAMES, VP8E_SET_ARNR_STRENGTH , VP8E_SET_ARNR_TYPE,
+ VP8E_SET_TUNING, VP8E_SET_CQ_LEVEL, VP8E_SET_MAX_INTRA_BITRATE_PCT, 0
+};
+#endif
+
+static const arg_def_t *no_args[] = { NULL };
+
+static void usage_exit()
+{
+ int i;
+
+ fprintf(stderr, "Usage: %s <options> -o dst_filename src_filename \n",
+ exec_name);
+
+ fprintf(stderr, "\nOptions:\n");
+ arg_show_usage(stdout, main_args);
+ fprintf(stderr, "\nEncoder Global Options:\n");
+ arg_show_usage(stdout, global_args);
+ fprintf(stderr, "\nRate Control Options:\n");
+ arg_show_usage(stdout, rc_args);
+ fprintf(stderr, "\nTwopass Rate Control Options:\n");
+ arg_show_usage(stdout, rc_twopass_args);
+ fprintf(stderr, "\nKeyframe Placement Options:\n");
+ arg_show_usage(stdout, kf_args);
+#if CONFIG_VP8_ENCODER
+ fprintf(stderr, "\nVP8 Specific Options:\n");
+ arg_show_usage(stdout, vp8_args);
+#endif
+ fprintf(stderr, "\nStream timebase (--timebase):\n"
+ " The desired precision of timestamps in the output, expressed\n"
+ " in fractional seconds. Default is 1/1000.\n");
+ fprintf(stderr, "\n"
+ "Included encoders:\n"
+ "\n");
+
+ for (i = 0; i < sizeof(codecs) / sizeof(codecs[0]); i++)
+ fprintf(stderr, " %-6s - %s\n",
+ codecs[i].name,
+ vpx_codec_iface_name(codecs[i].iface));
+
+ exit(EXIT_FAILURE);
+}
+
+
+#define HIST_BAR_MAX 40
+struct hist_bucket
+{
+ int low, high, count;
+};
+
+
+static int merge_hist_buckets(struct hist_bucket *bucket,
+ int *buckets_,
+ int max_buckets)
+{
+ int small_bucket = 0, merge_bucket = INT_MAX, big_bucket=0;
+ int buckets = *buckets_;
+ int i;
+
+ /* Find the extrema for this list of buckets */
+ big_bucket = small_bucket = 0;
+ for(i=0; i < buckets; i++)
+ {
+ if(bucket[i].count < bucket[small_bucket].count)
+ small_bucket = i;
+ if(bucket[i].count > bucket[big_bucket].count)
+ big_bucket = i;
+ }
+
+ /* If we have too many buckets, merge the smallest with an adjacent
+ * bucket.
+ */
+ while(buckets > max_buckets)
+ {
+ int last_bucket = buckets - 1;
+
+ /* merge the small bucket with an adjacent one. */
+ if(small_bucket == 0)
+ merge_bucket = 1;
+ else if(small_bucket == last_bucket)
+ merge_bucket = last_bucket - 1;
+ else if(bucket[small_bucket - 1].count < bucket[small_bucket + 1].count)
+ merge_bucket = small_bucket - 1;
+ else
+ merge_bucket = small_bucket + 1;
+
+ assert(abs(merge_bucket - small_bucket) <= 1);
+ assert(small_bucket < buckets);
+ assert(big_bucket < buckets);
+ assert(merge_bucket < buckets);
+
+ if(merge_bucket < small_bucket)
+ {
+ bucket[merge_bucket].high = bucket[small_bucket].high;
+ bucket[merge_bucket].count += bucket[small_bucket].count;
+ }
+ else
+ {
+ bucket[small_bucket].high = bucket[merge_bucket].high;
+ bucket[small_bucket].count += bucket[merge_bucket].count;
+ merge_bucket = small_bucket;
+ }
+
+ assert(bucket[merge_bucket].low != bucket[merge_bucket].high);
+
+ buckets--;
+
+ /* Remove the merge_bucket from the list, and find the new small
+ * and big buckets while we're at it
+ */
+ big_bucket = small_bucket = 0;
+ for(i=0; i < buckets; i++)
+ {
+ if(i > merge_bucket)
+ bucket[i] = bucket[i+1];
+
+ if(bucket[i].count < bucket[small_bucket].count)
+ small_bucket = i;
+ if(bucket[i].count > bucket[big_bucket].count)
+ big_bucket = i;
+ }
+
+ }
+
+ *buckets_ = buckets;
+ return bucket[big_bucket].count;
+}
+
+
+static void show_histogram(const struct hist_bucket *bucket,
+ int buckets,
+ int total,
+ int scale)
+{
+ const char *pat1, *pat2;
+ int i;
+
+ switch((int)(log(bucket[buckets-1].high)/log(10))+1)
+ {
+ case 1:
+ case 2:
+ pat1 = "%4d %2s: ";
+ pat2 = "%4d-%2d: ";
+ break;
+ case 3:
+ pat1 = "%5d %3s: ";
+ pat2 = "%5d-%3d: ";
+ break;
+ case 4:
+ pat1 = "%6d %4s: ";
+ pat2 = "%6d-%4d: ";
+ break;
+ case 5:
+ pat1 = "%7d %5s: ";
+ pat2 = "%7d-%5d: ";
+ break;
+ case 6:
+ pat1 = "%8d %6s: ";
+ pat2 = "%8d-%6d: ";
+ break;
+ case 7:
+ pat1 = "%9d %7s: ";
+ pat2 = "%9d-%7d: ";
+ break;
+ default:
+ pat1 = "%12d %10s: ";
+ pat2 = "%12d-%10d: ";
+ break;
+ }
+
+ for(i=0; i<buckets; i++)
+ {
+ int len;
+ int j;
+ float pct;
+
+ pct = (float)(100.0 * bucket[i].count / total);
+ len = HIST_BAR_MAX * bucket[i].count / scale;
+ if(len < 1)
+ len = 1;
+ assert(len <= HIST_BAR_MAX);
+
+ if(bucket[i].low == bucket[i].high)
+ fprintf(stderr, pat1, bucket[i].low, "");
+ else
+ fprintf(stderr, pat2, bucket[i].low, bucket[i].high);
+
+ for(j=0; j<HIST_BAR_MAX; j++)
+ fprintf(stderr, j<len?"=":" ");
+ fprintf(stderr, "\t%5d (%6.2f%%)\n",bucket[i].count,pct);
+ }
+}
+
+
+static void show_q_histogram(const int counts[64], int max_buckets)
+{
+ struct hist_bucket bucket[64];
+ int buckets = 0;
+ int total = 0;
+ int scale;
+ int i;
+
+
+ for(i=0; i<64; i++)
+ {
+ if(counts[i])
+ {
+ bucket[buckets].low = bucket[buckets].high = i;
+ bucket[buckets].count = counts[i];
+ buckets++;
+ total += counts[i];
+ }
+ }
+
+ fprintf(stderr, "\nQuantizer Selection:\n");
+ scale = merge_hist_buckets(bucket, &buckets, max_buckets);
+ show_histogram(bucket, buckets, total, scale);
+}
+
+
+#define RATE_BINS (100)
+struct rate_hist
+{
+ int64_t *pts;
+ int *sz;
+ int samples;
+ int frames;
+ struct hist_bucket bucket[RATE_BINS];
+ int total;
+};
+
+
+static void init_rate_histogram(struct rate_hist *hist,
+ const vpx_codec_enc_cfg_t *cfg,
+ const vpx_rational_t *fps)
+{
+ int i;
+
+ /* Determine the number of samples in the buffer. Use the file's framerate
+ * to determine the number of frames in rc_buf_sz milliseconds, with an
+ * adjustment (5/4) to account for alt-refs
+ */
+ hist->samples = cfg->rc_buf_sz * 5 / 4 * fps->num / fps->den / 1000;
+
+ /* prevent division by zero */
+ if (hist->samples == 0)
+ hist->samples=1;
+
+ hist->pts = calloc(hist->samples, sizeof(*hist->pts));
+ hist->sz = calloc(hist->samples, sizeof(*hist->sz));
+ for(i=0; i<RATE_BINS; i++)
+ {
+ hist->bucket[i].low = INT_MAX;
+ hist->bucket[i].high = 0;
+ hist->bucket[i].count = 0;
+ }
+}
+
+
+static void destroy_rate_histogram(struct rate_hist *hist)
+{
+ free(hist->pts);
+ free(hist->sz);
+}
+
+
+static void update_rate_histogram(struct rate_hist *hist,
+ const vpx_codec_enc_cfg_t *cfg,
+ const vpx_codec_cx_pkt_t *pkt)
+{
+ int i, idx;
+ int64_t now, then, sum_sz = 0, avg_bitrate;
+
+ now = pkt->data.frame.pts * 1000
+ * (uint64_t)cfg->g_timebase.num / (uint64_t)cfg->g_timebase.den;
+
+ idx = hist->frames++ % hist->samples;
+ hist->pts[idx] = now;
+ hist->sz[idx] = (int)pkt->data.frame.sz;
+
+ if(now < cfg->rc_buf_initial_sz)
+ return;
+
+ then = now;
+
+ /* Sum the size over the past rc_buf_sz ms */
+ for(i = hist->frames; i > 0 && hist->frames - i < hist->samples; i--)
+ {
+ int i_idx = (i-1) % hist->samples;
+
+ then = hist->pts[i_idx];
+ if(now - then > cfg->rc_buf_sz)
+ break;
+ sum_sz += hist->sz[i_idx];
+ }
+
+ if (now == then)
+ return;
+
+ avg_bitrate = sum_sz * 8 * 1000 / (now - then);
+ idx = (int)(avg_bitrate * (RATE_BINS/2) / (cfg->rc_target_bitrate * 1000));
+ if(idx < 0)
+ idx = 0;
+ if(idx > RATE_BINS-1)
+ idx = RATE_BINS-1;
+ if(hist->bucket[idx].low > avg_bitrate)
+ hist->bucket[idx].low = (int)avg_bitrate;
+ if(hist->bucket[idx].high < avg_bitrate)
+ hist->bucket[idx].high = (int)avg_bitrate;
+ hist->bucket[idx].count++;
+ hist->total++;
+}
+
+
+static void show_rate_histogram(struct rate_hist *hist,
+ const vpx_codec_enc_cfg_t *cfg,
+ int max_buckets)
+{
+ int i, scale;
+ int buckets = 0;
+
+ for(i = 0; i < RATE_BINS; i++)
+ {
+ if(hist->bucket[i].low == INT_MAX)
+ continue;
+ hist->bucket[buckets++] = hist->bucket[i];
+ }
+
+ fprintf(stderr, "\nRate (over %dms window):\n", cfg->rc_buf_sz);
+ scale = merge_hist_buckets(hist->bucket, &buckets, max_buckets);
+ show_histogram(hist->bucket, buckets, hist->total, scale);
+}
+
+#define NELEMENTS(x) (sizeof(x)/sizeof(x[0]))
+#define ARG_CTRL_CNT_MAX NELEMENTS(vp8_arg_ctrl_map)
+
+
+/* Configuration elements common to all streams */
+struct global_config
+{
+ const struct codec_item *codec;
+ int passes;
+ int pass;
+ int usage;
+ int deadline;
+ int use_i420;
+ int quiet;
+ int verbose;
+ int limit;
+ int show_psnr;
+ int have_framerate;
+ struct vpx_rational framerate;
+ int out_part;
+ int debug;
+ int show_q_hist_buckets;
+ int show_rate_hist_buckets;
+};
+
+
+/* Per-stream configuration */
+struct stream_config
+{
+ struct vpx_codec_enc_cfg cfg;
+ const char *out_fn;
+ const char *stats_fn;
+ stereo_format_t stereo_fmt;
+ int arg_ctrls[ARG_CTRL_CNT_MAX][2];
+ int arg_ctrl_cnt;
+ int write_webm;
+ int have_kf_max_dist;
+};
+
+
+struct stream_state
+{
+ int index;
+ struct stream_state *next;
+ struct stream_config config;
+ FILE *file;
+ struct rate_hist rate_hist;
+ EbmlGlobal ebml;
+ uint32_t hash;
+ uint64_t psnr_sse_total;
+ uint64_t psnr_samples_total;
+ double psnr_totals[4];
+ int psnr_count;
+ int counts[64];
+ vpx_codec_ctx_t encoder;
+ unsigned int frames_out;
+ uint64_t cx_time;
+ size_t nbytes;
+ stats_io_t stats;
+};
+
+
+void validate_positive_rational(const char *msg,
+ struct vpx_rational *rat)
+{
+ if (rat->den < 0)
+ {
+ rat->num *= -1;
+ rat->den *= -1;
+ }
+
+ if (rat->num < 0)
+ die("Error: %s must be positive\n", msg);
+
+ if (!rat->den)
+ die("Error: %s has zero denominator\n", msg);
+}
+
+
+static void parse_global_config(struct global_config *global, char **argv)
+{
+ char **argi, **argj;
+ struct arg arg;
+
+ /* Initialize default parameters */
+ memset(global, 0, sizeof(*global));
+ global->codec = codecs;
+ global->passes = 1;
+ global->use_i420 = 1;
+
+ for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step)
+ {
+ arg.argv_step = 1;
+
+ if (arg_match(&arg, &codecarg, argi))
+ {
+ int j, k = -1;
+
+ for (j = 0; j < sizeof(codecs) / sizeof(codecs[0]); j++)
+ if (!strcmp(codecs[j].name, arg.val))
+ k = j;
+
+ if (k >= 0)
+ global->codec = codecs + k;
+ else
+ die("Error: Unrecognized argument (%s) to --codec\n",
+ arg.val);
+
+ }
+ else if (arg_match(&arg, &passes, argi))
+ {
+ global->passes = arg_parse_uint(&arg);
+
+ if (global->passes < 1 || global->passes > 2)
+ die("Error: Invalid number of passes (%d)\n", global->passes);
+ }
+ else if (arg_match(&arg, &pass_arg, argi))
+ {
+ global->pass = arg_parse_uint(&arg);
+
+ if (global->pass < 1 || global->pass > 2)
+ die("Error: Invalid pass selected (%d)\n",
+ global->pass);
+ }
+ else if (arg_match(&arg, &usage, argi))
+ global->usage = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &deadline, argi))
+ global->deadline = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &best_dl, argi))
+ global->deadline = VPX_DL_BEST_QUALITY;
+ else if (arg_match(&arg, &good_dl, argi))
+ global->deadline = VPX_DL_GOOD_QUALITY;
+ else if (arg_match(&arg, &rt_dl, argi))
+ global->deadline = VPX_DL_REALTIME;
+ else if (arg_match(&arg, &use_yv12, argi))
+ global->use_i420 = 0;
+ else if (arg_match(&arg, &use_i420, argi))
+ global->use_i420 = 1;
+ else if (arg_match(&arg, &quietarg, argi))
+ global->quiet = 1;
+ else if (arg_match(&arg, &verbosearg, argi))
+ global->verbose = 1;
+ else if (arg_match(&arg, &limit, argi))
+ global->limit = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &psnrarg, argi))
+ global->show_psnr = 1;
+ else if (arg_match(&arg, &framerate, argi))
+ {
+ global->framerate = arg_parse_rational(&arg);
+ validate_positive_rational(arg.name, &global->framerate);
+ global->have_framerate = 1;
+ }
+ else if (arg_match(&arg,&out_part, argi))
+ global->out_part = 1;
+ else if (arg_match(&arg, &debugmode, argi))
+ global->debug = 1;
+ else if (arg_match(&arg, &q_hist_n, argi))
+ global->show_q_hist_buckets = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &rate_hist_n, argi))
+ global->show_rate_hist_buckets = arg_parse_uint(&arg);
+ else
+ argj++;
+ }
+
+ /* Validate global config */
+
+ if (global->pass)
+ {
+ /* DWIM: Assume the user meant passes=2 if pass=2 is specified */
+ if (global->pass > global->passes)
+ {
+ warn("Assuming --pass=%d implies --passes=%d\n",
+ global->pass, global->pass);
+ global->passes = global->pass;
+ }
+ }
+}
+
+
+void open_input_file(struct input_state *input)
+{
+ unsigned int fourcc;
+
+ /* Parse certain options from the input file, if possible */
+ input->file = strcmp(input->fn, "-") ? fopen(input->fn, "rb")
+ : set_binary_mode(stdin);
+
+ if (!input->file)
+ fatal("Failed to open input file");
+
+ /* For RAW input sources, these bytes will applied on the first frame
+ * in read_frame().
+ */
+ input->detect.buf_read = fread(input->detect.buf, 1, 4, input->file);
+ input->detect.position = 0;
+
+ if (input->detect.buf_read == 4
+ && file_is_y4m(input->file, &input->y4m, input->detect.buf))
+ {
+ if (y4m_input_open(&input->y4m, input->file, input->detect.buf, 4) >= 0)
+ {
+ input->file_type = FILE_TYPE_Y4M;
+ input->w = input->y4m.pic_w;
+ input->h = input->y4m.pic_h;
+ input->framerate.num = input->y4m.fps_n;
+ input->framerate.den = input->y4m.fps_d;
+ input->use_i420 = 0;
+ }
+ else
+ fatal("Unsupported Y4M stream.");
+ }
+ else if (input->detect.buf_read == 4 && file_is_ivf(input, &fourcc))
+ {
+ input->file_type = FILE_TYPE_IVF;
+ switch (fourcc)
+ {
+ case 0x32315659:
+ input->use_i420 = 0;
+ break;
+ case 0x30323449:
+ input->use_i420 = 1;
+ break;
+ default:
+ fatal("Unsupported fourcc (%08x) in IVF", fourcc);
+ }
+ }
+ else
+ {
+ input->file_type = FILE_TYPE_RAW;
+ }
+}
+
+
+static void close_input_file(struct input_state *input)
+{
+ fclose(input->file);
+ if (input->file_type == FILE_TYPE_Y4M)
+ y4m_input_close(&input->y4m);
+}
+
+static struct stream_state *new_stream(struct global_config *global,
+ struct stream_state *prev)
+{
+ struct stream_state *stream;
+
+ stream = calloc(1, sizeof(*stream));
+ if(!stream)
+ fatal("Failed to allocate new stream.");
+ if(prev)
+ {
+ memcpy(stream, prev, sizeof(*stream));
+ stream->index++;
+ prev->next = stream;
+ }
+ else
+ {
+ vpx_codec_err_t res;
+
+ /* Populate encoder configuration */
+ res = vpx_codec_enc_config_default(global->codec->iface,
+ &stream->config.cfg,
+ global->usage);
+ if (res)
+ fatal("Failed to get config: %s\n", vpx_codec_err_to_string(res));
+
+ /* Change the default timebase to a high enough value so that the
+ * encoder will always create strictly increasing timestamps.
+ */
+ stream->config.cfg.g_timebase.den = 1000;
+
+ /* Never use the library's default resolution, require it be parsed
+ * from the file or set on the command line.
+ */
+ stream->config.cfg.g_w = 0;
+ stream->config.cfg.g_h = 0;
+
+ /* Initialize remaining stream parameters */
+ stream->config.stereo_fmt = STEREO_FORMAT_MONO;
+ stream->config.write_webm = 1;
+ stream->ebml.last_pts_ms = -1;
+
+ /* Allows removal of the application version from the EBML tags */
+ stream->ebml.debug = global->debug;
+ }
+
+ /* Output files must be specified for each stream */
+ stream->config.out_fn = NULL;
+
+ stream->next = NULL;
+ return stream;
+}
+
+
+static int parse_stream_params(struct global_config *global,
+ struct stream_state *stream,
+ char **argv)
+{
+ char **argi, **argj;
+ struct arg arg;
+ static const arg_def_t **ctrl_args = no_args;
+ static const int *ctrl_args_map = NULL;
+ struct stream_config *config = &stream->config;
+ int eos_mark_found = 0;
+
+ /* Handle codec specific options */
+ if (global->codec->iface == &vpx_codec_vp8_cx_algo)
+ {
+ ctrl_args = vp8_args;
+ ctrl_args_map = vp8_arg_ctrl_map;
+ }
+
+ for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step)
+ {
+ arg.argv_step = 1;
+
+ /* Once we've found an end-of-stream marker (--) we want to continue
+ * shifting arguments but not consuming them.
+ */
+ if (eos_mark_found)
+ {
+ argj++;
+ continue;
+ }
+ else if (!strcmp(*argj, "--"))
+ {
+ eos_mark_found = 1;
+ continue;
+ }
+
+ if (0);
+ else if (arg_match(&arg, &outputfile, argi))
+ config->out_fn = arg.val;
+ else if (arg_match(&arg, &fpf_name, argi))
+ config->stats_fn = arg.val;
+ else if (arg_match(&arg, &use_ivf, argi))
+ config->write_webm = 0;
+ else if (arg_match(&arg, &threads, argi))
+ config->cfg.g_threads = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &profile, argi))
+ config->cfg.g_profile = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &width, argi))
+ config->cfg.g_w = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &height, argi))
+ config->cfg.g_h = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &stereo_mode, argi))
+ config->stereo_fmt = arg_parse_enum_or_int(&arg);
+ else if (arg_match(&arg, &timebase, argi))
+ {
+ config->cfg.g_timebase = arg_parse_rational(&arg);
+ validate_positive_rational(arg.name, &config->cfg.g_timebase);
+ }
+ else if (arg_match(&arg, &error_resilient, argi))
+ config->cfg.g_error_resilient = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &lag_in_frames, argi))
+ config->cfg.g_lag_in_frames = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &dropframe_thresh, argi))
+ config->cfg.rc_dropframe_thresh = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &resize_allowed, argi))
+ config->cfg.rc_resize_allowed = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &resize_up_thresh, argi))
+ config->cfg.rc_resize_up_thresh = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &resize_down_thresh, argi))
+ config->cfg.rc_resize_down_thresh = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &end_usage, argi))
+ config->cfg.rc_end_usage = arg_parse_enum_or_int(&arg);
+ else if (arg_match(&arg, &target_bitrate, argi))
+ config->cfg.rc_target_bitrate = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &min_quantizer, argi))
+ config->cfg.rc_min_quantizer = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &max_quantizer, argi))
+ config->cfg.rc_max_quantizer = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &undershoot_pct, argi))
+ config->cfg.rc_undershoot_pct = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &overshoot_pct, argi))
+ config->cfg.rc_overshoot_pct = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &buf_sz, argi))
+ config->cfg.rc_buf_sz = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &buf_initial_sz, argi))
+ config->cfg.rc_buf_initial_sz = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &buf_optimal_sz, argi))
+ config->cfg.rc_buf_optimal_sz = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &bias_pct, argi))
+ {
+ config->cfg.rc_2pass_vbr_bias_pct = arg_parse_uint(&arg);
+
+ if (global->passes < 2)
+ warn("option %s ignored in one-pass mode.\n", arg.name);
+ }
+ else if (arg_match(&arg, &minsection_pct, argi))
+ {
+ config->cfg.rc_2pass_vbr_minsection_pct = arg_parse_uint(&arg);
+
+ if (global->passes < 2)
+ warn("option %s ignored in one-pass mode.\n", arg.name);
+ }
+ else if (arg_match(&arg, &maxsection_pct, argi))
+ {
+ config->cfg.rc_2pass_vbr_maxsection_pct = arg_parse_uint(&arg);
+
+ if (global->passes < 2)
+ warn("option %s ignored in one-pass mode.\n", arg.name);
+ }
+ else if (arg_match(&arg, &kf_min_dist, argi))
+ config->cfg.kf_min_dist = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &kf_max_dist, argi))
+ {
+ config->cfg.kf_max_dist = arg_parse_uint(&arg);
+ config->have_kf_max_dist = 1;
+ }
+ else if (arg_match(&arg, &kf_disabled, argi))
+ config->cfg.kf_mode = VPX_KF_DISABLED;
+ else
+ {
+ int i, match = 0;
+
+ for (i = 0; ctrl_args[i]; i++)
+ {
+ if (arg_match(&arg, ctrl_args[i], argi))
+ {
+ int j;
+ match = 1;
+
+ /* Point either to the next free element or the first
+ * instance of this control.
+ */
+ for(j=0; j<config->arg_ctrl_cnt; j++)
+ if(config->arg_ctrls[j][0] == ctrl_args_map[i])
+ break;
+
+ /* Update/insert */
+ assert(j < ARG_CTRL_CNT_MAX);
+ if (j < ARG_CTRL_CNT_MAX)
+ {
+ config->arg_ctrls[j][0] = ctrl_args_map[i];
+ config->arg_ctrls[j][1] = arg_parse_enum_or_int(&arg);
+ if(j == config->arg_ctrl_cnt)
+ config->arg_ctrl_cnt++;
+ }
+
+ }
+ }
+
+ if (!match)
+ argj++;
+ }
+ }
+
+ return eos_mark_found;
+}
+
+
+#define FOREACH_STREAM(func)\
+do\
+{\
+ struct stream_state *stream;\
+\
+ for(stream = streams; stream; stream = stream->next)\
+ func;\
+}while(0)
+
+
+static void validate_stream_config(struct stream_state *stream)
+{
+ struct stream_state *streami;
+
+ if(!stream->config.cfg.g_w || !stream->config.cfg.g_h)
+ fatal("Stream %d: Specify stream dimensions with --width (-w) "
+ " and --height (-h)", stream->index);
+
+ for(streami = stream; streami; streami = streami->next)
+ {
+ /* All streams require output files */
+ if(!streami->config.out_fn)
+ fatal("Stream %d: Output file is required (specify with -o)",
+ streami->index);
+
+ /* Check for two streams outputting to the same file */
+ if(streami != stream)
+ {
+ const char *a = stream->config.out_fn;
+ const char *b = streami->config.out_fn;
+ if(!strcmp(a,b) && strcmp(a, "/dev/null") && strcmp(a, ":nul"))
+ fatal("Stream %d: duplicate output file (from stream %d)",
+ streami->index, stream->index);
+ }
+
+ /* Check for two streams sharing a stats file. */
+ if(streami != stream)
+ {
+ const char *a = stream->config.stats_fn;
+ const char *b = streami->config.stats_fn;
+ if(a && b && !strcmp(a,b))
+ fatal("Stream %d: duplicate stats file (from stream %d)",
+ streami->index, stream->index);
+ }
+ }
+}
+
+
+static void set_stream_dimensions(struct stream_state *stream,
+ unsigned int w,
+ unsigned int h)
+{
+ if ((stream->config.cfg.g_w && stream->config.cfg.g_w != w)
+ ||(stream->config.cfg.g_h && stream->config.cfg.g_h != h))
+ fatal("Stream %d: Resizing not yet supported", stream->index);
+ stream->config.cfg.g_w = w;
+ stream->config.cfg.g_h = h;
+}
+
+
+static void set_default_kf_interval(struct stream_state *stream,
+ struct global_config *global)
+{
+ /* Use a max keyframe interval of 5 seconds, if none was
+ * specified on the command line.
+ */
+ if (!stream->config.have_kf_max_dist)
+ {
+ double framerate = (double)global->framerate.num/global->framerate.den;
+ if (framerate > 0.0)
+ stream->config.cfg.kf_max_dist = (unsigned int)(5.0*framerate);
+ }
+}
+
+
+static void show_stream_config(struct stream_state *stream,
+ struct global_config *global,
+ struct input_state *input)
+{
+
+#define SHOW(field) \
+ fprintf(stderr, " %-28s = %d\n", #field, stream->config.cfg.field)
+
+ if(stream->index == 0)
+ {
+ fprintf(stderr, "Codec: %s\n",
+ vpx_codec_iface_name(global->codec->iface));
+ fprintf(stderr, "Source file: %s Format: %s\n", input->fn,
+ input->use_i420 ? "I420" : "YV12");
+ }
+ if(stream->next || stream->index)
+ fprintf(stderr, "\nStream Index: %d\n", stream->index);
+ fprintf(stderr, "Destination file: %s\n", stream->config.out_fn);
+ fprintf(stderr, "Encoder parameters:\n");
+
+ SHOW(g_usage);
+ SHOW(g_threads);
+ SHOW(g_profile);
+ SHOW(g_w);
+ SHOW(g_h);
+ SHOW(g_timebase.num);
+ SHOW(g_timebase.den);
+ SHOW(g_error_resilient);
+ SHOW(g_pass);
+ SHOW(g_lag_in_frames);
+ SHOW(rc_dropframe_thresh);
+ SHOW(rc_resize_allowed);
+ SHOW(rc_resize_up_thresh);
+ SHOW(rc_resize_down_thresh);
+ SHOW(rc_end_usage);
+ SHOW(rc_target_bitrate);
+ SHOW(rc_min_quantizer);
+ SHOW(rc_max_quantizer);
+ SHOW(rc_undershoot_pct);
+ SHOW(rc_overshoot_pct);
+ SHOW(rc_buf_sz);
+ SHOW(rc_buf_initial_sz);
+ SHOW(rc_buf_optimal_sz);
+ SHOW(rc_2pass_vbr_bias_pct);
+ SHOW(rc_2pass_vbr_minsection_pct);
+ SHOW(rc_2pass_vbr_maxsection_pct);
+ SHOW(kf_mode);
+ SHOW(kf_min_dist);
+ SHOW(kf_max_dist);
+}
+
+
+static void open_output_file(struct stream_state *stream,
+ struct global_config *global)
+{
+ const char *fn = stream->config.out_fn;
+
+ stream->file = strcmp(fn, "-") ? fopen(fn, "wb") : set_binary_mode(stdout);
+
+ if (!stream->file)
+ fatal("Failed to open output file");
+
+ if(stream->config.write_webm && fseek(stream->file, 0, SEEK_CUR))
+ fatal("WebM output to pipes not supported.");
+
+ if(stream->config.write_webm)
+ {
+ stream->ebml.stream = stream->file;
+ write_webm_file_header(&stream->ebml, &stream->config.cfg,
+ &global->framerate,
+ stream->config.stereo_fmt);
+ }
+ else
+ write_ivf_file_header(stream->file, &stream->config.cfg,
+ global->codec->fourcc, 0);
+}
+
+
+static void close_output_file(struct stream_state *stream,
+ unsigned int fourcc)
+{
+ if(stream->config.write_webm)
+ {
+ write_webm_file_footer(&stream->ebml, stream->hash);
+ free(stream->ebml.cue_list);
+ stream->ebml.cue_list = NULL;
+ }
+ else
+ {
+ if (!fseek(stream->file, 0, SEEK_SET))
+ write_ivf_file_header(stream->file, &stream->config.cfg,
+ fourcc,
+ stream->frames_out);
+ }
+
+ fclose(stream->file);
+}
+
+
+static void setup_pass(struct stream_state *stream,
+ struct global_config *global,
+ int pass)
+{
+ if (stream->config.stats_fn)
+ {
+ if (!stats_open_file(&stream->stats, stream->config.stats_fn,
+ pass))
+ fatal("Failed to open statistics store");
+ }
+ else
+ {
+ if (!stats_open_mem(&stream->stats, pass))
+ fatal("Failed to open statistics store");
+ }
+
+ stream->config.cfg.g_pass = global->passes == 2
+ ? pass ? VPX_RC_LAST_PASS : VPX_RC_FIRST_PASS
+ : VPX_RC_ONE_PASS;
+ if (pass)
+ stream->config.cfg.rc_twopass_stats_in = stats_get(&stream->stats);
+
+ stream->cx_time = 0;
+ stream->nbytes = 0;
+ stream->frames_out = 0;
+}
+
+
+static void initialize_encoder(struct stream_state *stream,
+ struct global_config *global)
+{
+ int i;
+ int flags = 0;
+
+ flags |= global->show_psnr ? VPX_CODEC_USE_PSNR : 0;
+ flags |= global->out_part ? VPX_CODEC_USE_OUTPUT_PARTITION : 0;
+
+ /* Construct Encoder Context */
+ vpx_codec_enc_init(&stream->encoder, global->codec->iface,
+ &stream->config.cfg, flags);
+ ctx_exit_on_error(&stream->encoder, "Failed to initialize encoder");
+
+ /* Note that we bypass the vpx_codec_control wrapper macro because
+ * we're being clever to store the control IDs in an array. Real
+ * applications will want to make use of the enumerations directly
+ */
+ for (i = 0; i < stream->config.arg_ctrl_cnt; i++)
+ {
+ int ctrl = stream->config.arg_ctrls[i][0];
+ int value = stream->config.arg_ctrls[i][1];
+ if (vpx_codec_control_(&stream->encoder, ctrl, value))
+ fprintf(stderr, "Error: Tried to set control %d = %d\n",
+ ctrl, value);
+
+ ctx_exit_on_error(&stream->encoder, "Failed to control codec");
+ }
+}
+
+
+static void encode_frame(struct stream_state *stream,
+ struct global_config *global,
+ struct vpx_image *img,
+ unsigned int frames_in)
+{
+ vpx_codec_pts_t frame_start, next_frame_start;
+ struct vpx_codec_enc_cfg *cfg = &stream->config.cfg;
+ struct vpx_usec_timer timer;
+
+ frame_start = (cfg->g_timebase.den * (int64_t)(frames_in - 1)
+ * global->framerate.den)
+ / cfg->g_timebase.num / global->framerate.num;
+ next_frame_start = (cfg->g_timebase.den * (int64_t)(frames_in)
+ * global->framerate.den)
+ / cfg->g_timebase.num / global->framerate.num;
+ vpx_usec_timer_start(&timer);
+ vpx_codec_encode(&stream->encoder, img, frame_start,
+ (unsigned long)(next_frame_start - frame_start),
+ 0, global->deadline);
+ vpx_usec_timer_mark(&timer);
+ stream->cx_time += vpx_usec_timer_elapsed(&timer);
+ ctx_exit_on_error(&stream->encoder, "Stream %d: Failed to encode frame",
+ stream->index);
+}
+
+
+static void update_quantizer_histogram(struct stream_state *stream)
+{
+ if(stream->config.cfg.g_pass != VPX_RC_FIRST_PASS)
+ {
+ int q;
+
+ vpx_codec_control(&stream->encoder, VP8E_GET_LAST_QUANTIZER_64, &q);
+ ctx_exit_on_error(&stream->encoder, "Failed to read quantizer");
+ stream->counts[q]++;
+ }
+}
+
+
+static void get_cx_data(struct stream_state *stream,
+ struct global_config *global,
+ int *got_data)
+{
+ const vpx_codec_cx_pkt_t *pkt;
+ const struct vpx_codec_enc_cfg *cfg = &stream->config.cfg;
+ vpx_codec_iter_t iter = NULL;
+
+ while ((pkt = vpx_codec_get_cx_data(&stream->encoder, &iter)))
+ {
+ static size_t fsize = 0;
+ static off_t ivf_header_pos = 0;
+
+ *got_data = 1;
+
+ switch (pkt->kind)
+ {
+ case VPX_CODEC_CX_FRAME_PKT:
+ if (!(pkt->data.frame.flags & VPX_FRAME_IS_FRAGMENT))
+ {
+ stream->frames_out++;
+ }
+ if (!global->quiet)
+ fprintf(stderr, " %6luF",
+ (unsigned long)pkt->data.frame.sz);
+
+ update_rate_histogram(&stream->rate_hist, cfg, pkt);
+ if(stream->config.write_webm)
+ {
+ /* Update the hash */
+ if(!stream->ebml.debug)
+ stream->hash = murmur(pkt->data.frame.buf,
+ (int)pkt->data.frame.sz,
+ stream->hash);
+
+ write_webm_block(&stream->ebml, cfg, pkt);
+ }
+ else
+ {
+ if (pkt->data.frame.partition_id <= 0)
+ {
+ ivf_header_pos = ftello(stream->file);
+ fsize = pkt->data.frame.sz;
+
+ write_ivf_frame_header(stream->file, pkt);
+ }
+ else
+ {
+ fsize += pkt->data.frame.sz;
+
+ if (!(pkt->data.frame.flags & VPX_FRAME_IS_FRAGMENT))
+ {
+ off_t currpos = ftello(stream->file);
+ fseeko(stream->file, ivf_header_pos, SEEK_SET);
+ write_ivf_frame_size(stream->file, fsize);
+ fseeko(stream->file, currpos, SEEK_SET);
+ }
+ }
+
+ (void) fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
+ stream->file);
+ }
+ stream->nbytes += pkt->data.raw.sz;
+ break;
+ case VPX_CODEC_STATS_PKT:
+ stream->frames_out++;
+ fprintf(stderr, " %6luS",
+ (unsigned long)pkt->data.twopass_stats.sz);
+ stats_write(&stream->stats,
+ pkt->data.twopass_stats.buf,
+ pkt->data.twopass_stats.sz);
+ stream->nbytes += pkt->data.raw.sz;
+ break;
+ case VPX_CODEC_PSNR_PKT:
+
+ if (global->show_psnr)
+ {
+ int i;
+
+ stream->psnr_sse_total += pkt->data.psnr.sse[0];
+ stream->psnr_samples_total += pkt->data.psnr.samples[0];
+ for (i = 0; i < 4; i++)
+ {
+ if (!global->quiet)
+ fprintf(stderr, "%.3f ", pkt->data.psnr.psnr[i]);
+ stream->psnr_totals[i] += pkt->data.psnr.psnr[i];
+ }
+ stream->psnr_count++;
+ }
+
+ break;
+ default:
+ break;
+ }
+ }
+}
+
+
+static void show_psnr(struct stream_state *stream)
+{
+ int i;
+ double ovpsnr;
+
+ if (!stream->psnr_count)
+ return;
+
+ fprintf(stderr, "Stream %d PSNR (Overall/Avg/Y/U/V)", stream->index);
+ ovpsnr = vp8_mse2psnr((double)stream->psnr_samples_total, 255.0,
+ (double)stream->psnr_sse_total);
+ fprintf(stderr, " %.3f", ovpsnr);
+
+ for (i = 0; i < 4; i++)
+ {
+ fprintf(stderr, " %.3f", stream->psnr_totals[i]/stream->psnr_count);
+ }
+ fprintf(stderr, "\n");
+}
+
+
+float usec_to_fps(uint64_t usec, unsigned int frames)
+{
+ return (float)(usec > 0 ? frames * 1000000.0 / (float)usec : 0);
+}
+
+
+int main(int argc, const char **argv_)
+{
+ int pass;
+ vpx_image_t raw;
+ int frame_avail, got_data;
+
+ struct input_state input = {0};
+ struct global_config global;
+ struct stream_state *streams = NULL;
+ char **argv, **argi;
+ unsigned long cx_time = 0;
+ int stream_cnt = 0;
+
+ exec_name = argv_[0];
+
+ if (argc < 3)
+ usage_exit();
+
+ /* Setup default input stream settings */
+ input.framerate.num = 30;
+ input.framerate.den = 1;
+ input.use_i420 = 1;
+
+ /* First parse the global configuration values, because we want to apply
+ * other parameters on top of the default configuration provided by the
+ * codec.
+ */
+ argv = argv_dup(argc - 1, argv_ + 1);
+ parse_global_config(&global, argv);
+
+ {
+ /* Now parse each stream's parameters. Using a local scope here
+ * due to the use of 'stream' as loop variable in FOREACH_STREAM
+ * loops
+ */
+ struct stream_state *stream = NULL;
+
+ do
+ {
+ stream = new_stream(&global, stream);
+ stream_cnt++;
+ if(!streams)
+ streams = stream;
+ } while(parse_stream_params(&global, stream, argv));
+ }
+
+ /* Check for unrecognized options */
+ for (argi = argv; *argi; argi++)
+ if (argi[0][0] == '-' && argi[0][1])
+ die("Error: Unrecognized option %s\n", *argi);
+
+ /* Handle non-option arguments */
+ input.fn = argv[0];
+
+ if (!input.fn)
+ usage_exit();
+
+ for (pass = global.pass ? global.pass - 1 : 0; pass < global.passes; pass++)
+ {
+ int frames_in = 0;
+
+ open_input_file(&input);
+
+ /* If the input file doesn't specify its w/h (raw files), try to get
+ * the data from the first stream's configuration.
+ */
+ if(!input.w || !input.h)
+ FOREACH_STREAM({
+ if(stream->config.cfg.g_w && stream->config.cfg.g_h)
+ {
+ input.w = stream->config.cfg.g_w;
+ input.h = stream->config.cfg.g_h;
+ break;
+ }
+ });
+
+ /* Update stream configurations from the input file's parameters */
+ FOREACH_STREAM(set_stream_dimensions(stream, input.w, input.h));
+ FOREACH_STREAM(validate_stream_config(stream));
+
+ /* Ensure that --passes and --pass are consistent. If --pass is set and
+ * --passes=2, ensure --fpf was set.
+ */
+ if (global.pass && global.passes == 2)
+ FOREACH_STREAM({
+ if(!stream->config.stats_fn)
+ die("Stream %d: Must specify --fpf when --pass=%d"
+ " and --passes=2\n", stream->index, global.pass);
+ });
+
+
+ /* Use the frame rate from the file only if none was specified
+ * on the command-line.
+ */
+ if (!global.have_framerate)
+ global.framerate = input.framerate;
+
+ FOREACH_STREAM(set_default_kf_interval(stream, &global));
+
+ /* Show configuration */
+ if (global.verbose && pass == 0)
+ FOREACH_STREAM(show_stream_config(stream, &global, &input));
+
+ if(pass == (global.pass ? global.pass - 1 : 0)) {
+ if (input.file_type == FILE_TYPE_Y4M)
+ /*The Y4M reader does its own allocation.
+ Just initialize this here to avoid problems if we never read any
+ frames.*/
+ memset(&raw, 0, sizeof(raw));
+ else
+ vpx_img_alloc(&raw,
+ input.use_i420 ? VPX_IMG_FMT_I420
+ : VPX_IMG_FMT_YV12,
+ input.w, input.h, 32);
+
+ FOREACH_STREAM(init_rate_histogram(&stream->rate_hist,
+ &stream->config.cfg,
+ &global.framerate));
+ }
+
+ FOREACH_STREAM(open_output_file(stream, &global));
+ FOREACH_STREAM(setup_pass(stream, &global, pass));
+ FOREACH_STREAM(initialize_encoder(stream, &global));
+
+ frame_avail = 1;
+ got_data = 0;
+
+ while (frame_avail || got_data)
+ {
+ struct vpx_usec_timer timer;
+
+ if (!global.limit || frames_in < global.limit)
+ {
+ frame_avail = read_frame(&input, &raw);
+
+ if (frame_avail)
+ frames_in++;
+
+ if (!global.quiet)
+ {
+ if(stream_cnt == 1)
+ fprintf(stderr,
+ "\rPass %d/%d frame %4d/%-4d %7"PRId64"B \033[K",
+ pass + 1, global.passes, frames_in,
+ streams->frames_out, (int64_t)streams->nbytes);
+ else
+ fprintf(stderr,
+ "\rPass %d/%d frame %4d %7lu %s (%.2f fps)\033[K",
+ pass + 1, global.passes, frames_in,
+ cx_time > 9999999 ? cx_time / 1000 : cx_time,
+ cx_time > 9999999 ? "ms" : "us",
+ usec_to_fps(cx_time, frames_in));
+ }
+
+ }
+ else
+ frame_avail = 0;
+
+ vpx_usec_timer_start(&timer);
+ FOREACH_STREAM(encode_frame(stream, &global,
+ frame_avail ? &raw : NULL,
+ frames_in));
+ vpx_usec_timer_mark(&timer);
+ cx_time += (unsigned long)vpx_usec_timer_elapsed(&timer);
+
+ FOREACH_STREAM(update_quantizer_histogram(stream));
+
+ got_data = 0;
+ FOREACH_STREAM(get_cx_data(stream, &global, &got_data));
+
+ fflush(stdout);
+ }
+
+ if(stream_cnt > 1)
+ fprintf(stderr, "\n");
+
+ if (!global.quiet)
+ FOREACH_STREAM(fprintf(
+ stderr,
+ "\rPass %d/%d frame %4d/%-4d %7"PRId64"B %7lub/f %7"PRId64"b/s"
+ " %7"PRId64" %s (%.2f fps)\033[K\n", pass + 1,
+ global.passes, frames_in, stream->frames_out, (int64_t)stream->nbytes,
+ frames_in ? (unsigned long)(stream->nbytes * 8 / frames_in) : 0,
+ frames_in ? (int64_t)stream->nbytes * 8
+ * (int64_t)global.framerate.num / global.framerate.den
+ / frames_in
+ : 0,
+ stream->cx_time > 9999999 ? stream->cx_time / 1000 : stream->cx_time,
+ stream->cx_time > 9999999 ? "ms" : "us",
+ usec_to_fps(stream->cx_time, frames_in));
+ );
+
+ if (global.show_psnr)
+ FOREACH_STREAM(show_psnr(stream));
+
+ FOREACH_STREAM(vpx_codec_destroy(&stream->encoder));
+
+ close_input_file(&input);
+
+ FOREACH_STREAM(close_output_file(stream, global.codec->fourcc));
+
+ FOREACH_STREAM(stats_close(&stream->stats, global.passes-1));
+
+ if (global.pass)
+ break;
+ }
+
+ if (global.show_q_hist_buckets)
+ FOREACH_STREAM(show_q_histogram(stream->counts,
+ global.show_q_hist_buckets));
+
+ if (global.show_rate_hist_buckets)
+ FOREACH_STREAM(show_rate_histogram(&stream->rate_hist,
+ &stream->config.cfg,
+ global.show_rate_hist_buckets));
+ FOREACH_STREAM(destroy_rate_histogram(&stream->rate_hist));
+
+ vpx_img_free(&raw);
+ free(argv);
+ free(streams);
+ return EXIT_SUCCESS;
+}
diff --git a/libvpx/y4minput.c b/libvpx/y4minput.c
new file mode 100644
index 0000000..ff9ffbc
--- /dev/null
+++ b/libvpx/y4minput.c
@@ -0,0 +1,871 @@
+/*
+ * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ *
+ * Based on code from the OggTheora software codec source code,
+ * Copyright (C) 2002-2010 The Xiph.Org Foundation and contributors.
+ */
+#include <stdlib.h>
+#include <string.h>
+#include "y4minput.h"
+
+static int y4m_parse_tags(y4m_input *_y4m,char *_tags){
+ int got_w;
+ int got_h;
+ int got_fps;
+ int got_interlace;
+ int got_par;
+ int got_chroma;
+ char *p;
+ char *q;
+ got_w=got_h=got_fps=got_interlace=got_par=got_chroma=0;
+ for(p=_tags;;p=q){
+ /*Skip any leading spaces.*/
+ while(*p==' ')p++;
+ /*If that's all we have, stop.*/
+ if(p[0]=='\0')break;
+ /*Find the end of this tag.*/
+ for(q=p+1;*q!='\0'&&*q!=' ';q++);
+ /*Process the tag.*/
+ switch(p[0]){
+ case 'W':{
+ if(sscanf(p+1,"%d",&_y4m->pic_w)!=1)return -1;
+ got_w=1;
+ }break;
+ case 'H':{
+ if(sscanf(p+1,"%d",&_y4m->pic_h)!=1)return -1;
+ got_h=1;
+ }break;
+ case 'F':{
+ if(sscanf(p+1,"%d:%d",&_y4m->fps_n,&_y4m->fps_d)!=2){
+ return -1;
+ }
+ got_fps=1;
+ }break;
+ case 'I':{
+ _y4m->interlace=p[1];
+ got_interlace=1;
+ }break;
+ case 'A':{
+ if(sscanf(p+1,"%d:%d",&_y4m->par_n,&_y4m->par_d)!=2){
+ return -1;
+ }
+ got_par=1;
+ }break;
+ case 'C':{
+ if(q-p>16)return -1;
+ memcpy(_y4m->chroma_type,p+1,q-p-1);
+ _y4m->chroma_type[q-p-1]='\0';
+ got_chroma=1;
+ }break;
+ /*Ignore unknown tags.*/
+ }
+ }
+ if(!got_w||!got_h||!got_fps)return -1;
+ if(!got_interlace)_y4m->interlace='?';
+ if(!got_par)_y4m->par_n=_y4m->par_d=0;
+ /*Chroma-type is not specified in older files, e.g., those generated by
+ mplayer.*/
+ if(!got_chroma)strcpy(_y4m->chroma_type,"420");
+ return 0;
+}
+
+
+
+/*All anti-aliasing filters in the following conversion functions are based on
+ one of two window functions:
+ The 6-tap Lanczos window (for down-sampling and shifts):
+ sinc(\pi*t)*sinc(\pi*t/3), |t|<3 (sinc(t)==sin(t)/t)
+ 0, |t|>=3
+ The 4-tap Mitchell window (for up-sampling):
+ 7|t|^3-12|t|^2+16/3, |t|<1
+ -(7/3)|x|^3+12|x|^2-20|x|+32/3, |t|<2
+ 0, |t|>=2
+ The number of taps is intentionally kept small to reduce computational
+ overhead and limit ringing.
+
+ The taps from these filters are scaled so that their sum is 1, and the result
+ is scaled by 128 and rounded to integers to create a filter whose
+ intermediate values fit inside 16 bits.
+ Coefficients are rounded in such a way as to ensure their sum is still 128,
+ which is usually equivalent to normal rounding.
+
+ Conversions which require both horizontal and vertical filtering could
+ have these steps pipelined, for less memory consumption and better cache
+ performance, but we do them separately for simplicity.*/
+
+#define OC_MINI(_a,_b) ((_a)>(_b)?(_b):(_a))
+#define OC_MAXI(_a,_b) ((_a)<(_b)?(_b):(_a))
+#define OC_CLAMPI(_a,_b,_c) (OC_MAXI(_a,OC_MINI(_b,_c)))
+
+/*420jpeg chroma samples are sited like:
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ 420mpeg2 chroma samples are sited like:
+ Y-------Y-------Y-------Y-------
+ | | | |
+ BR | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ BR | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ We use a resampling filter to shift the site locations one quarter pixel (at
+ the chroma plane's resolution) to the right.
+ The 4:2:2 modes look exactly the same, except there are twice as many chroma
+ lines, and they are vertically co-sited with the luma samples in both the
+ mpeg2 and jpeg cases (thus requiring no vertical resampling).*/
+static void y4m_42xmpeg2_42xjpeg_helper(unsigned char *_dst,
+ const unsigned char *_src,int _c_w,int _c_h){
+ int y;
+ int x;
+ for(y=0;y<_c_h;y++){
+ /*Filter: [4 -17 114 35 -9 1]/128, derived from a 6-tap Lanczos
+ window.*/
+ for(x=0;x<OC_MINI(_c_w,2);x++){
+ _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[0]-17*_src[OC_MAXI(x-1,0)]+
+ 114*_src[x]+35*_src[OC_MINI(x+1,_c_w-1)]-9*_src[OC_MINI(x+2,_c_w-1)]+
+ _src[OC_MINI(x+3,_c_w-1)]+64)>>7,255);
+ }
+ for(;x<_c_w-3;x++){
+ _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[x-2]-17*_src[x-1]+
+ 114*_src[x]+35*_src[x+1]-9*_src[x+2]+_src[x+3]+64)>>7,255);
+ }
+ for(;x<_c_w;x++){
+ _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[x-2]-17*_src[x-1]+
+ 114*_src[x]+35*_src[OC_MINI(x+1,_c_w-1)]-9*_src[OC_MINI(x+2,_c_w-1)]+
+ _src[_c_w-1]+64)>>7,255);
+ }
+ _dst+=_c_w;
+ _src+=_c_w;
+ }
+}
+
+/*Handles both 422 and 420mpeg2 to 422jpeg and 420jpeg, respectively.*/
+static void y4m_convert_42xmpeg2_42xjpeg(y4m_input *_y4m,unsigned char *_dst,
+ unsigned char *_aux){
+ int c_w;
+ int c_h;
+ int c_sz;
+ int pli;
+ /*Skip past the luma data.*/
+ _dst+=_y4m->pic_w*_y4m->pic_h;
+ /*Compute the size of each chroma plane.*/
+ c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
+ c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
+ c_sz=c_w*c_h;
+ for(pli=1;pli<3;pli++){
+ y4m_42xmpeg2_42xjpeg_helper(_dst,_aux,c_w,c_h);
+ _dst+=c_sz;
+ _aux+=c_sz;
+ }
+}
+
+/*This format is only used for interlaced content, but is included for
+ completeness.
+
+ 420jpeg chroma samples are sited like:
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ 420paldv chroma samples are sited like:
+ YR------Y-------YR------Y-------
+ | | | |
+ | | | |
+ | | | |
+ YB------Y-------YB------Y-------
+ | | | |
+ | | | |
+ | | | |
+ YR------Y-------YR------Y-------
+ | | | |
+ | | | |
+ | | | |
+ YB------Y-------YB------Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ We use a resampling filter to shift the site locations one quarter pixel (at
+ the chroma plane's resolution) to the right.
+ Then we use another filter to move the C_r location down one quarter pixel,
+ and the C_b location up one quarter pixel.*/
+static void y4m_convert_42xpaldv_42xjpeg(y4m_input *_y4m,unsigned char *_dst,
+ unsigned char *_aux){
+ unsigned char *tmp;
+ int c_w;
+ int c_h;
+ int c_sz;
+ int pli;
+ int y;
+ int x;
+ /*Skip past the luma data.*/
+ _dst+=_y4m->pic_w*_y4m->pic_h;
+ /*Compute the size of each chroma plane.*/
+ c_w=(_y4m->pic_w+1)/2;
+ c_h=(_y4m->pic_h+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
+ c_sz=c_w*c_h;
+ tmp=_aux+2*c_sz;
+ for(pli=1;pli<3;pli++){
+ /*First do the horizontal re-sampling.
+ This is the same as the mpeg2 case, except that after the horizontal
+ case, we need to apply a second vertical filter.*/
+ y4m_42xmpeg2_42xjpeg_helper(tmp,_aux,c_w,c_h);
+ _aux+=c_sz;
+ switch(pli){
+ case 1:{
+ /*Slide C_b up a quarter-pel.
+ This is the same filter used above, but in the other order.*/
+ for(x=0;x<c_w;x++){
+ for(y=0;y<OC_MINI(c_h,3);y++){
+ _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[0]
+ -9*tmp[OC_MAXI(y-2,0)*c_w]+35*tmp[OC_MAXI(y-1,0)*c_w]
+ +114*tmp[y*c_w]-17*tmp[OC_MINI(y+1,c_h-1)*c_w]
+ +4*tmp[OC_MINI(y+2,c_h-1)*c_w]+64)>>7,255);
+ }
+ for(;y<c_h-2;y++){
+ _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[(y-3)*c_w]
+ -9*tmp[(y-2)*c_w]+35*tmp[(y-1)*c_w]+114*tmp[y*c_w]
+ -17*tmp[(y+1)*c_w]+4*tmp[(y+2)*c_w]+64)>>7,255);
+ }
+ for(;y<c_h;y++){
+ _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[(y-3)*c_w]
+ -9*tmp[(y-2)*c_w]+35*tmp[(y-1)*c_w]+114*tmp[y*c_w]
+ -17*tmp[OC_MINI(y+1,c_h-1)*c_w]+4*tmp[(c_h-1)*c_w]+64)>>7,255);
+ }
+ _dst++;
+ tmp++;
+ }
+ _dst+=c_sz-c_w;
+ tmp-=c_w;
+ }break;
+ case 2:{
+ /*Slide C_r down a quarter-pel.
+ This is the same as the horizontal filter.*/
+ for(x=0;x<c_w;x++){
+ for(y=0;y<OC_MINI(c_h,2);y++){
+ _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[0]
+ -17*tmp[OC_MAXI(y-1,0)*c_w]+114*tmp[y*c_w]
+ +35*tmp[OC_MINI(y+1,c_h-1)*c_w]-9*tmp[OC_MINI(y+2,c_h-1)*c_w]
+ +tmp[OC_MINI(y+3,c_h-1)*c_w]+64)>>7,255);
+ }
+ for(;y<c_h-3;y++){
+ _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[(y-2)*c_w]
+ -17*tmp[(y-1)*c_w]+114*tmp[y*c_w]+35*tmp[(y+1)*c_w]
+ -9*tmp[(y+2)*c_w]+tmp[(y+3)*c_w]+64)>>7,255);
+ }
+ for(;y<c_h;y++){
+ _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[(y-2)*c_w]
+ -17*tmp[(y-1)*c_w]+114*tmp[y*c_w]+35*tmp[OC_MINI(y+1,c_h-1)*c_w]
+ -9*tmp[OC_MINI(y+2,c_h-1)*c_w]+tmp[(c_h-1)*c_w]+64)>>7,255);
+ }
+ _dst++;
+ tmp++;
+ }
+ }break;
+ }
+ /*For actual interlaced material, this would have to be done separately on
+ each field, and the shift amounts would be different.
+ C_r moves down 1/8, C_b up 3/8 in the top field, and C_r moves down 3/8,
+ C_b up 1/8 in the bottom field.
+ The corresponding filters would be:
+ Down 1/8 (reverse order for up): [3 -11 125 15 -4 0]/128
+ Down 3/8 (reverse order for up): [4 -19 98 56 -13 2]/128*/
+ }
+}
+
+/*Perform vertical filtering to reduce a single plane from 4:2:2 to 4:2:0.
+ This is used as a helper by several converation routines.*/
+static void y4m_422jpeg_420jpeg_helper(unsigned char *_dst,
+ const unsigned char *_src,int _c_w,int _c_h){
+ int y;
+ int x;
+ /*Filter: [3 -17 78 78 -17 3]/128, derived from a 6-tap Lanczos window.*/
+ for(x=0;x<_c_w;x++){
+ for(y=0;y<OC_MINI(_c_h,2);y+=2){
+ _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(64*_src[0]
+ +78*_src[OC_MINI(1,_c_h-1)*_c_w]
+ -17*_src[OC_MINI(2,_c_h-1)*_c_w]
+ +3*_src[OC_MINI(3,_c_h-1)*_c_w]+64)>>7,255);
+ }
+ for(;y<_c_h-3;y+=2){
+ _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(3*(_src[(y-2)*_c_w]+_src[(y+3)*_c_w])
+ -17*(_src[(y-1)*_c_w]+_src[(y+2)*_c_w])
+ +78*(_src[y*_c_w]+_src[(y+1)*_c_w])+64)>>7,255);
+ }
+ for(;y<_c_h;y+=2){
+ _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(3*(_src[(y-2)*_c_w]
+ +_src[(_c_h-1)*_c_w])-17*(_src[(y-1)*_c_w]
+ +_src[OC_MINI(y+2,_c_h-1)*_c_w])
+ +78*(_src[y*_c_w]+_src[OC_MINI(y+1,_c_h-1)*_c_w])+64)>>7,255);
+ }
+ _src++;
+ _dst++;
+ }
+}
+
+/*420jpeg chroma samples are sited like:
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ 422jpeg chroma samples are sited like:
+ Y---BR--Y-------Y---BR--Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y---BR--Y-------Y---BR--Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y---BR--Y-------Y---BR--Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y---BR--Y-------Y---BR--Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ We use a resampling filter to decimate the chroma planes by two in the
+ vertical direction.*/
+static void y4m_convert_422jpeg_420jpeg(y4m_input *_y4m,unsigned char *_dst,
+ unsigned char *_aux){
+ int c_w;
+ int c_h;
+ int c_sz;
+ int dst_c_w;
+ int dst_c_h;
+ int dst_c_sz;
+ int pli;
+ /*Skip past the luma data.*/
+ _dst+=_y4m->pic_w*_y4m->pic_h;
+ /*Compute the size of each chroma plane.*/
+ c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h;
+ c_h=_y4m->pic_h;
+ dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
+ dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
+ c_sz=c_w*c_h;
+ dst_c_sz=dst_c_w*dst_c_h;
+ for(pli=1;pli<3;pli++){
+ y4m_422jpeg_420jpeg_helper(_dst,_aux,c_w,c_h);
+ _aux+=c_sz;
+ _dst+=dst_c_sz;
+ }
+}
+
+/*420jpeg chroma samples are sited like:
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ 422 chroma samples are sited like:
+ YBR-----Y-------YBR-----Y-------
+ | | | |
+ | | | |
+ | | | |
+ YBR-----Y-------YBR-----Y-------
+ | | | |
+ | | | |
+ | | | |
+ YBR-----Y-------YBR-----Y-------
+ | | | |
+ | | | |
+ | | | |
+ YBR-----Y-------YBR-----Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ We use a resampling filter to shift the original site locations one quarter
+ pixel (at the original chroma resolution) to the right.
+ Then we use a second resampling filter to decimate the chroma planes by two
+ in the vertical direction.*/
+static void y4m_convert_422_420jpeg(y4m_input *_y4m,unsigned char *_dst,
+ unsigned char *_aux){
+ unsigned char *tmp;
+ int c_w;
+ int c_h;
+ int c_sz;
+ int dst_c_h;
+ int dst_c_sz;
+ int pli;
+ /*Skip past the luma data.*/
+ _dst+=_y4m->pic_w*_y4m->pic_h;
+ /*Compute the size of each chroma plane.*/
+ c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h;
+ c_h=_y4m->pic_h;
+ dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
+ c_sz=c_w*c_h;
+ dst_c_sz=c_w*dst_c_h;
+ tmp=_aux+2*c_sz;
+ for(pli=1;pli<3;pli++){
+ /*In reality, the horizontal and vertical steps could be pipelined, for
+ less memory consumption and better cache performance, but we do them
+ separately for simplicity.*/
+ /*First do horizontal filtering (convert to 422jpeg)*/
+ y4m_42xmpeg2_42xjpeg_helper(tmp,_aux,c_w,c_h);
+ /*Now do the vertical filtering.*/
+ y4m_422jpeg_420jpeg_helper(_dst,tmp,c_w,c_h);
+ _aux+=c_sz;
+ _dst+=dst_c_sz;
+ }
+}
+
+/*420jpeg chroma samples are sited like:
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | BR | | BR |
+ | | | |
+ Y-------Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ 411 chroma samples are sited like:
+ YBR-----Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ YBR-----Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ YBR-----Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+ YBR-----Y-------Y-------Y-------
+ | | | |
+ | | | |
+ | | | |
+
+ We use a filter to resample at site locations one eighth pixel (at the source
+ chroma plane's horizontal resolution) and five eighths of a pixel to the
+ right.
+ Then we use another filter to decimate the planes by 2 in the vertical
+ direction.*/
+static void y4m_convert_411_420jpeg(y4m_input *_y4m,unsigned char *_dst,
+ unsigned char *_aux){
+ unsigned char *tmp;
+ int c_w;
+ int c_h;
+ int c_sz;
+ int dst_c_w;
+ int dst_c_h;
+ int dst_c_sz;
+ int tmp_sz;
+ int pli;
+ int y;
+ int x;
+ /*Skip past the luma data.*/
+ _dst+=_y4m->pic_w*_y4m->pic_h;
+ /*Compute the size of each chroma plane.*/
+ c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h;
+ c_h=_y4m->pic_h;
+ dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
+ dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
+ c_sz=c_w*c_h;
+ dst_c_sz=dst_c_w*dst_c_h;
+ tmp_sz=dst_c_w*c_h;
+ tmp=_aux+2*c_sz;
+ for(pli=1;pli<3;pli++){
+ /*In reality, the horizontal and vertical steps could be pipelined, for
+ less memory consumption and better cache performance, but we do them
+ separately for simplicity.*/
+ /*First do horizontal filtering (convert to 422jpeg)*/
+ for(y=0;y<c_h;y++){
+ /*Filters: [1 110 18 -1]/128 and [-3 50 86 -5]/128, both derived from a
+ 4-tap Mitchell window.*/
+ for(x=0;x<OC_MINI(c_w,1);x++){
+ tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(111*_aux[0]
+ +18*_aux[OC_MINI(1,c_w-1)]-_aux[OC_MINI(2,c_w-1)]+64)>>7,255);
+ tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(47*_aux[0]
+ +86*_aux[OC_MINI(1,c_w-1)]-5*_aux[OC_MINI(2,c_w-1)]+64)>>7,255);
+ }
+ for(;x<c_w-2;x++){
+ tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(_aux[x-1]+110*_aux[x]
+ +18*_aux[x+1]-_aux[x+2]+64)>>7,255);
+ tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(-3*_aux[x-1]+50*_aux[x]
+ +86*_aux[x+1]-5*_aux[x+2]+64)>>7,255);
+ }
+ for(;x<c_w;x++){
+ tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(_aux[x-1]+110*_aux[x]
+ +18*_aux[OC_MINI(x+1,c_w-1)]-_aux[c_w-1]+64)>>7,255);
+ if((x<<1|1)<dst_c_w){
+ tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(-3*_aux[x-1]+50*_aux[x]
+ +86*_aux[OC_MINI(x+1,c_w-1)]-5*_aux[c_w-1]+64)>>7,255);
+ }
+ }
+ tmp+=dst_c_w;
+ _aux+=c_w;
+ }
+ tmp-=tmp_sz;
+ /*Now do the vertical filtering.*/
+ y4m_422jpeg_420jpeg_helper(_dst,tmp,dst_c_w,c_h);
+ _dst+=dst_c_sz;
+ }
+}
+
+/*Convert 444 to 420jpeg.*/
+static void y4m_convert_444_420jpeg(y4m_input *_y4m,unsigned char *_dst,
+ unsigned char *_aux){
+ unsigned char *tmp;
+ int c_w;
+ int c_h;
+ int c_sz;
+ int dst_c_w;
+ int dst_c_h;
+ int dst_c_sz;
+ int tmp_sz;
+ int pli;
+ int y;
+ int x;
+ /*Skip past the luma data.*/
+ _dst+=_y4m->pic_w*_y4m->pic_h;
+ /*Compute the size of each chroma plane.*/
+ c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h;
+ c_h=_y4m->pic_h;
+ dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
+ dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
+ c_sz=c_w*c_h;
+ dst_c_sz=dst_c_w*dst_c_h;
+ tmp_sz=dst_c_w*c_h;
+ tmp=_aux+2*c_sz;
+ for(pli=1;pli<3;pli++){
+ /*Filter: [3 -17 78 78 -17 3]/128, derived from a 6-tap Lanczos window.*/
+ for(y=0;y<c_h;y++){
+ for(x=0;x<OC_MINI(c_w,2);x+=2){
+ tmp[x>>1]=OC_CLAMPI(0,(64*_aux[0]+78*_aux[OC_MINI(1,c_w-1)]
+ -17*_aux[OC_MINI(2,c_w-1)]
+ +3*_aux[OC_MINI(3,c_w-1)]+64)>>7,255);
+ }
+ for(;x<c_w-3;x+=2){
+ tmp[x>>1]=OC_CLAMPI(0,(3*(_aux[x-2]+_aux[x+3])
+ -17*(_aux[x-1]+_aux[x+2])+78*(_aux[x]+_aux[x+1])+64)>>7,255);
+ }
+ for(;x<c_w;x+=2){
+ tmp[x>>1]=OC_CLAMPI(0,(3*(_aux[x-2]+_aux[c_w-1])-
+ 17*(_aux[x-1]+_aux[OC_MINI(x+2,c_w-1)])+
+ 78*(_aux[x]+_aux[OC_MINI(x+1,c_w-1)])+64)>>7,255);
+ }
+ tmp+=dst_c_w;
+ _aux+=c_w;
+ }
+ tmp-=tmp_sz;
+ /*Now do the vertical filtering.*/
+ y4m_422jpeg_420jpeg_helper(_dst,tmp,dst_c_w,c_h);
+ _dst+=dst_c_sz;
+ }
+}
+
+/*The image is padded with empty chroma components at 4:2:0.*/
+static void y4m_convert_mono_420jpeg(y4m_input *_y4m,unsigned char *_dst,
+ unsigned char *_aux){
+ int c_sz;
+ _dst+=_y4m->pic_w*_y4m->pic_h;
+ c_sz=((_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h)*
+ ((_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v);
+ memset(_dst,128,c_sz*2);
+}
+
+/*No conversion function needed.*/
+static void y4m_convert_null(y4m_input *_y4m,unsigned char *_dst,
+ unsigned char *_aux){
+}
+
+int y4m_input_open(y4m_input *_y4m,FILE *_fin,char *_skip,int _nskip){
+ char buffer[80];
+ int ret;
+ int i;
+ /*Read until newline, or 80 cols, whichever happens first.*/
+ for(i=0;i<79;i++){
+ if(_nskip>0){
+ buffer[i]=*_skip++;
+ _nskip--;
+ }
+ else{
+ ret=(int)fread(buffer+i,1,1,_fin);
+ if(ret<1)return -1;
+ }
+ if(buffer[i]=='\n')break;
+ }
+ /*We skipped too much header data.*/
+ if(_nskip>0)return -1;
+ if(i==79){
+ fprintf(stderr,"Error parsing header; not a YUV2MPEG2 file?\n");
+ return -1;
+ }
+ buffer[i]='\0';
+ if(memcmp(buffer,"YUV4MPEG",8)){
+ fprintf(stderr,"Incomplete magic for YUV4MPEG file.\n");
+ return -1;
+ }
+ if(buffer[8]!='2'){
+ fprintf(stderr,"Incorrect YUV input file version; YUV4MPEG2 required.\n");
+ }
+ ret=y4m_parse_tags(_y4m,buffer+5);
+ if(ret<0){
+ fprintf(stderr,"Error parsing YUV4MPEG2 header.\n");
+ return ret;
+ }
+ if(_y4m->interlace=='?'){
+ fprintf(stderr,"Warning: Input video interlacing format unknown; "
+ "assuming progressive scan.\n");
+ }
+ else if(_y4m->interlace!='p'){
+ fprintf(stderr,"Input video is interlaced; "
+ "Only progressive scan handled.\n");
+ return -1;
+ }
+ if(strcmp(_y4m->chroma_type,"420")==0||
+ strcmp(_y4m->chroma_type,"420jpeg")==0){
+ _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h
+ +2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2);
+ /*Natively supported: no conversion required.*/
+ _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=0;
+ _y4m->convert=y4m_convert_null;
+ }
+ else if(strcmp(_y4m->chroma_type,"420mpeg2")==0){
+ _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
+ /*Chroma filter required: read into the aux buf first.*/
+ _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=
+ 2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2);
+ _y4m->convert=y4m_convert_42xmpeg2_42xjpeg;
+ }
+ else if(strcmp(_y4m->chroma_type,"420paldv")==0){
+ _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
+ /*Chroma filter required: read into the aux buf first.
+ We need to make two filter passes, so we need some extra space in the
+ aux buffer.*/
+ _y4m->aux_buf_sz=3*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2);
+ _y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2);
+ _y4m->convert=y4m_convert_42xpaldv_42xjpeg;
+ }
+ else if(strcmp(_y4m->chroma_type,"422jpeg")==0){
+ _y4m->src_c_dec_h=_y4m->dst_c_dec_h=2;
+ _y4m->src_c_dec_v=1;
+ _y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
+ /*Chroma filter required: read into the aux buf first.*/
+ _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*_y4m->pic_h;
+ _y4m->convert=y4m_convert_422jpeg_420jpeg;
+ }
+ else if(strcmp(_y4m->chroma_type,"422")==0){
+ _y4m->src_c_dec_h=_y4m->dst_c_dec_h=2;
+ _y4m->src_c_dec_v=1;
+ _y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
+ /*Chroma filter required: read into the aux buf first.
+ We need to make two filter passes, so we need some extra space in the
+ aux buffer.*/
+ _y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*_y4m->pic_h;
+ _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h;
+ _y4m->convert=y4m_convert_422_420jpeg;
+ }
+ else if(strcmp(_y4m->chroma_type,"411")==0){
+ _y4m->src_c_dec_h=4;
+ _y4m->dst_c_dec_h=2;
+ _y4m->src_c_dec_v=1;
+ _y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
+ /*Chroma filter required: read into the aux buf first.
+ We need to make two filter passes, so we need some extra space in the
+ aux buffer.*/
+ _y4m->aux_buf_read_sz=2*((_y4m->pic_w+3)/4)*_y4m->pic_h;
+ _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h;
+ _y4m->convert=y4m_convert_411_420jpeg;
+ }
+ else if(strcmp(_y4m->chroma_type,"444")==0){
+ _y4m->src_c_dec_h=1;
+ _y4m->dst_c_dec_h=2;
+ _y4m->src_c_dec_v=1;
+ _y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
+ /*Chroma filter required: read into the aux buf first.
+ We need to make two filter passes, so we need some extra space in the
+ aux buffer.*/
+ _y4m->aux_buf_read_sz=2*_y4m->pic_w*_y4m->pic_h;
+ _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h;
+ _y4m->convert=y4m_convert_444_420jpeg;
+ }
+ else if(strcmp(_y4m->chroma_type,"444alpha")==0){
+ _y4m->src_c_dec_h=1;
+ _y4m->dst_c_dec_h=2;
+ _y4m->src_c_dec_v=1;
+ _y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
+ /*Chroma filter required: read into the aux buf first.
+ We need to make two filter passes, so we need some extra space in the
+ aux buffer.
+ The extra plane also gets read into the aux buf.
+ It will be discarded.*/
+ _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=3*_y4m->pic_w*_y4m->pic_h;
+ _y4m->convert=y4m_convert_444_420jpeg;
+ }
+ else if(strcmp(_y4m->chroma_type,"mono")==0){
+ _y4m->src_c_dec_h=_y4m->src_c_dec_v=0;
+ _y4m->dst_c_dec_h=_y4m->dst_c_dec_v=2;
+ _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
+ /*No extra space required, but we need to clear the chroma planes.*/
+ _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=0;
+ _y4m->convert=y4m_convert_mono_420jpeg;
+ }
+ else{
+ fprintf(stderr,"Unknown chroma sampling type: %s\n",_y4m->chroma_type);
+ return -1;
+ }
+ /*The size of the final frame buffers is always computed from the
+ destination chroma decimation type.*/
+ _y4m->dst_buf_sz=_y4m->pic_w*_y4m->pic_h
+ +2*((_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h)*
+ ((_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v);
+ _y4m->dst_buf=(unsigned char *)malloc(_y4m->dst_buf_sz);
+ _y4m->aux_buf=(unsigned char *)malloc(_y4m->aux_buf_sz);
+ return 0;
+}
+
+void y4m_input_close(y4m_input *_y4m){
+ free(_y4m->dst_buf);
+ free(_y4m->aux_buf);
+}
+
+int y4m_input_fetch_frame(y4m_input *_y4m,FILE *_fin,vpx_image_t *_img){
+ char frame[6];
+ int pic_sz;
+ int c_w;
+ int c_h;
+ int c_sz;
+ int ret;
+ /*Read and skip the frame header.*/
+ ret=(int)fread(frame,1,6,_fin);
+ if(ret<6)return 0;
+ if(memcmp(frame,"FRAME",5)){
+ fprintf(stderr,"Loss of framing in Y4M input data\n");
+ return -1;
+ }
+ if(frame[5]!='\n'){
+ char c;
+ int j;
+ for(j=0;j<79&&fread(&c,1,1,_fin)&&c!='\n';j++);
+ if(j==79){
+ fprintf(stderr,"Error parsing Y4M frame header\n");
+ return -1;
+ }
+ }
+ /*Read the frame data that needs no conversion.*/
+ if(fread(_y4m->dst_buf,1,_y4m->dst_buf_read_sz,_fin)!=_y4m->dst_buf_read_sz){
+ fprintf(stderr,"Error reading Y4M frame data.\n");
+ return -1;
+ }
+ /*Read the frame data that does need conversion.*/
+ if(fread(_y4m->aux_buf,1,_y4m->aux_buf_read_sz,_fin)!=_y4m->aux_buf_read_sz){
+ fprintf(stderr,"Error reading Y4M frame data.\n");
+ return -1;
+ }
+ /*Now convert the just read frame.*/
+ (*_y4m->convert)(_y4m,_y4m->dst_buf,_y4m->aux_buf);
+ /*Fill in the frame buffer pointers.
+ We don't use vpx_img_wrap() because it forces padding for odd picture
+ sizes, which would require a separate fread call for every row.*/
+ memset(_img,0,sizeof(*_img));
+ /*Y4M has the planes in Y'CbCr order, which libvpx calls Y, U, and V.*/
+ _img->fmt=IMG_FMT_I420;
+ _img->w=_img->d_w=_y4m->pic_w;
+ _img->h=_img->d_h=_y4m->pic_h;
+ /*This is hard-coded to 4:2:0 for now, as that's all VP8 supports.*/
+ _img->x_chroma_shift=1;
+ _img->y_chroma_shift=1;
+ _img->bps=12;
+ /*Set up the buffer pointers.*/
+ pic_sz=_y4m->pic_w*_y4m->pic_h;
+ c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
+ c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
+ c_sz=c_w*c_h;
+ _img->stride[PLANE_Y]=_y4m->pic_w;
+ _img->stride[PLANE_U]=_img->stride[PLANE_V]=c_w;
+ _img->planes[PLANE_Y]=_y4m->dst_buf;
+ _img->planes[PLANE_U]=_y4m->dst_buf+pic_sz;
+ _img->planes[PLANE_V]=_y4m->dst_buf+pic_sz+c_sz;
+ return 1;
+}
diff --git a/y4minput.h b/libvpx/y4minput.h
similarity index 100%
rename from y4minput.h
rename to libvpx/y4minput.h
diff --git a/libwebm.mk b/libwebm.mk
new file mode 100644
index 0000000..0528cfb
--- /dev/null
+++ b/libwebm.mk
@@ -0,0 +1,9 @@
+# libwebm
+LOCAL_PATH := $(call my-dir)
+include $(CLEAR_VARS)
+
+LOCAL_SRC_FILES := libwebm/mkvparser.cpp
+
+LOCAL_MODULE := libwebm
+
+include $(BUILD_STATIC_LIBRARY)
diff --git a/mkvparser/mkvparser.cpp b/libwebm/mkvparser.cpp
similarity index 100%
rename from mkvparser/mkvparser.cpp
rename to libwebm/mkvparser.cpp
diff --git a/mkvparser/mkvparser.hpp b/libwebm/mkvparser.hpp
similarity index 100%
rename from mkvparser/mkvparser.hpp
rename to libwebm/mkvparser.hpp
diff --git a/mainpage.dox b/mainpage.dox
deleted file mode 100644
index 5613ae7..0000000
--- a/mainpage.dox
+++ /dev/null
@@ -1,45 +0,0 @@
-/*!\mainpage WebM VP8 Codec SDK
-
- \section main_contents Page Contents
- - \ref main_intro
- - \ref main_startpoints
- - \ref main_support
-
- \section main_intro Introduction
- Welcome to the WebM VP8 Codec SDK. This SDK allows you to integrate your
- applications with the VP8 video codec, a high quality, royalty free, open
- source codec deployed on millions of computers and devices worldwide.
-
- This distribution of the WebM VP8 Codec SDK includes the following support:
-
- \if vp8_encoder - \ref vp8_encoder \endif
- \if vp8_decoder - \ref vp8_decoder \endif
-
-
- \section main_startpoints Starting Points
- - Consult the \ref changelog for a complete list of improvements in this
- release.
- - The \ref readme contains instructions on recompiling the sample applications.
- - Read the \ref usage "usage" for a narrative on codec usage.
- - Read the \ref samples "sample code" for examples of how to interact with the
- codec.
- - \ref codec reference
- \if encoder - \ref encoder reference \endif
- \if decoder - \ref decoder reference \endif
-
- \section main_support Support Options & FAQ
- The WebM project is an open source project supported by its community. For
- questions about this SDK, please mail the apps-devel@webmproject.org list.
- To contribute, see http://www.webmproject.org/code/contribute and mail
- codec-devel@webmproject.org.
-*/
-
-/*!\page changelog CHANGELOG
- \verbinclude CHANGELOG
-*/
-
-/*!\page readme README
- \verbinclude README
-*/
-
-/*!\defgroup codecs Supported Codecs */
diff --git a/mips-dspr2/libvpx_srcs.txt b/mips-dspr2/libvpx_srcs.txt
new file mode 100644
index 0000000..fa2d647
--- /dev/null
+++ b/mips-dspr2/libvpx_srcs.txt
@@ -0,0 +1,119 @@
+build/make/rtcd.sh
+build/make/version.sh
+CHANGELOG
+libs.mk
+vp8/common/alloccommon.c
+vp8/common/alloccommon.h
+vp8/common/asm_com_offsets.c
+vp8/common/blockd.c
+vp8/common/blockd.h
+vp8/common/coefupdateprobs.h
+vp8/common/common.h
+vp8/common/debugmodes.c
+vp8/common/default_coef_probs.h
+vp8/common/dequantize.c
+vp8/common/entropy.c
+vp8/common/entropy.h
+vp8/common/entropymode.c
+vp8/common/entropymode.h
+vp8/common/entropymv.c
+vp8/common/entropymv.h
+vp8/common/extend.c
+vp8/common/extend.h
+vp8/common/filter.c
+vp8/common/filter.h
+vp8/common/findnearmv.c
+vp8/common/findnearmv.h
+vp8/common/generic/systemdependent.c
+vp8/common/header.h
+vp8/common/idct_blk.c
+vp8/common/idctllm.c
+vp8/common/invtrans.h
+vp8/common/loopfilter.c
+vp8/common/loopfilter_filters.c
+vp8/common/loopfilter.h
+vp8/common/mbpitch.c
+vp8/common/mips/dspr2/dequantize_dspr2.c
+vp8/common/mips/dspr2/filter_dspr2.c
+vp8/common/mips/dspr2/idct_blk_dspr2.c
+vp8/common/mips/dspr2/idctllm_dspr2.c
+vp8/common/mips/dspr2/loopfilter_filters_dspr2.c
+vp8/common/mips/dspr2/reconinter_dspr2.c
+vp8/common/modecont.c
+vp8/common/modecont.h
+vp8/common/mv.h
+vp8/common/onyxc_int.h
+vp8/common/onyxd.h
+vp8/common/onyx.h
+vp8/common/ppflags.h
+vp8/common/pragmas.h
+vp8/common/quant_common.c
+vp8/common/quant_common.h
+vp8/common/reconinter.c
+vp8/common/reconinter.h
+vp8/common/reconintra4x4.c
+vp8/common/reconintra4x4.h
+vp8/common/reconintra.c
+vp8/common/rtcd.c
+vp8/common/rtcd_defs.sh
+vp8/common/sad_c.c
+vp8/common/setupintrarecon.c
+vp8/common/setupintrarecon.h
+vp8/common/swapyv12buffer.c
+vp8/common/swapyv12buffer.h
+vp8/common/systemdependent.h
+vp8/common/threading.h
+vp8/common/treecoder.c
+vp8/common/treecoder.h
+vp8/common/variance_c.c
+vp8/common/variance.h
+vp8/common/vp8_entropymodedata.h
+vp8/decoder/asm_dec_offsets.c
+vp8/decoder/dboolhuff.c
+vp8/decoder/dboolhuff.h
+vp8/decoder/decodemv.c
+vp8/decoder/decodemv.h
+vp8/decoder/decoderthreading.h
+vp8/decoder/decodframe.c
+vp8/decoder/detokenize.c
+vp8/decoder/detokenize.h
+vp8/decoder/onyxd_if.c
+vp8/decoder/onyxd_int.h
+vp8/decoder/threading.c
+vp8/decoder/treereader.h
+vp8/vp8_common.mk
+vp8/vp8_dx_iface.c
+vp8/vp8dx.mk
+vpx_config.c
+vpx/internal/vpx_codec_internal.h
+vpx_mem/include/vpx_mem_intrnl.h
+vpx_mem/vpx_mem.c
+vpx_mem/vpx_mem.h
+vpx_mem/vpx_mem.mk
+vpx_ports/asm_offsets.h
+vpx_ports/mem.h
+vpx_ports/vpx_ports.mk
+vpx_ports/vpx_timer.h
+vpx_scale/generic/gen_scalers.c
+vpx_scale/generic/vpxscale.c
+vpx_scale/generic/yv12config.c
+vpx_scale/generic/yv12extend.c
+vpx_scale/generic/yv12extend_generic.h
+vpx_scale/scale_mode.h
+vpx_scale/vpxscale.h
+vpx_scale/vpx_scale.mk
+vpx_scale/yv12config.h
+vpx/src/vpx_codec.c
+vpx/src/vpx_decoder.c
+vpx/src/vpx_encoder.c
+vpx/src/vpx_image.c
+vpx/vp8dx.h
+vpx/vp8.h
+vpx/vpx_codec.h
+vpx/vpx_codec_impl_bottom.h
+vpx/vpx_codec_impl_top.h
+vpx/vpx_codec.mk
+vpx/vpx_decoder.h
+vpx/vpx_encoder.h
+vpx/vpx_image.h
+vpx/vpx_integer.h
diff --git a/mips-dspr2/vpx_config.c b/mips-dspr2/vpx_config.c
new file mode 100644
index 0000000..49cf24e
--- /dev/null
+++ b/mips-dspr2/vpx_config.c
@@ -0,0 +1,9 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+static const char* const cfg = "--force-target=mips32-android-gcc --disable-runtime-cpu-detect --sdk-path=/usr/local/google/home/johannkoenig/android-ndk --disable-vp8-encoder --disable-examples --disable-docs --enable-dspr2";
+const char *vpx_codec_build_config(void) {return cfg;}
diff --git a/mips-dspr2/vpx_config.h b/mips-dspr2/vpx_config.h
new file mode 100644
index 0000000..3a9e711
--- /dev/null
+++ b/mips-dspr2/vpx_config.h
@@ -0,0 +1,80 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+/* This file automatically generated by configure. Do not edit! */
+#ifndef VPX_CONFIG_H
+#define VPX_CONFIG_H
+#define RESTRICT
+#define ARCH_ARM 0
+#define ARCH_MIPS 1
+#define ARCH_X86 0
+#define ARCH_X86_64 0
+#define ARCH_PPC32 0
+#define ARCH_PPC64 0
+#define HAVE_EDSP 0
+#define HAVE_MEDIA 0
+#define HAVE_NEON 0
+#define HAVE_MIPS32 1
+#define HAVE_DSPR2 1
+#define HAVE_MMX 0
+#define HAVE_SSE 0
+#define HAVE_SSE2 0
+#define HAVE_SSE3 0
+#define HAVE_SSSE3 0
+#define HAVE_SSE4_1 0
+#define HAVE_ALTIVEC 0
+#define HAVE_VPX_PORTS 1
+#define HAVE_STDINT_H 1
+#define HAVE_ALT_TREE_LAYOUT 0
+#define HAVE_PTHREAD_H 1
+#define HAVE_SYS_MMAN_H 1
+#define HAVE_UNISTD_H 1
+#define CONFIG_EXTERNAL_BUILD 0
+#define CONFIG_INSTALL_DOCS 1
+#define CONFIG_INSTALL_BINS 1
+#define CONFIG_INSTALL_LIBS 1
+#define CONFIG_INSTALL_SRCS 0
+#define CONFIG_DEBUG 0
+#define CONFIG_GPROF 0
+#define CONFIG_GCOV 0
+#define CONFIG_RVCT 0
+#define CONFIG_GCC 1
+#define CONFIG_MSVS 0
+#define CONFIG_PIC 0
+#define CONFIG_BIG_ENDIAN 0
+#define CONFIG_CODEC_SRCS 0
+#define CONFIG_DEBUG_LIBS 0
+#define CONFIG_FAST_UNALIGNED 0
+#define CONFIG_MEM_MANAGER 0
+#define CONFIG_MEM_TRACKER 0
+#define CONFIG_MEM_CHECKS 0
+#define CONFIG_MD5 1
+#define CONFIG_DEQUANT_TOKENS 1
+#define CONFIG_DC_RECON 1
+#define CONFIG_RUNTIME_CPU_DETECT 0
+#define CONFIG_POSTPROC 0
+#define CONFIG_MULTITHREAD 1
+#define CONFIG_INTERNAL_STATS 0
+#define CONFIG_VP8_ENCODER 0
+#define CONFIG_VP8_DECODER 1
+#define CONFIG_VP8 1
+#define CONFIG_ENCODERS 0
+#define CONFIG_DECODERS 1
+#define CONFIG_STATIC_MSVCRT 0
+#define CONFIG_SPATIAL_RESAMPLING 1
+#define CONFIG_REALTIME_ONLY 0
+#define CONFIG_ONTHEFLY_BITPACKING 0
+#define CONFIG_ERROR_CONCEALMENT 0
+#define CONFIG_SHARED 0
+#define CONFIG_STATIC 1
+#define CONFIG_SMALL 0
+#define CONFIG_POSTPROC_VISUALIZER 0
+#define CONFIG_OS_SUPPORT 1
+#define CONFIG_UNIT_TESTS 0
+#define CONFIG_MULTI_RES_ENCODING 0
+#define CONFIG_TEMPORAL_DENOISING 1
+#endif /* VPX_CONFIG_H */
diff --git a/mips-dspr2/vpx_rtcd.h b/mips-dspr2/vpx_rtcd.h
new file mode 100644
index 0000000..880e0f4
--- /dev/null
+++ b/mips-dspr2/vpx_rtcd.h
@@ -0,0 +1,318 @@
+#ifndef VPX_RTCD_
+#define VPX_RTCD_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+#include "vp8/common/blockd.h"
+
+struct blockd;
+struct macroblockd;
+struct loop_filter_info;
+
+/* Encoder forward decls */
+struct block;
+struct macroblock;
+struct variance_vtable;
+union int_mv;
+struct yv12_buffer_config;
+
+void vp8_dequantize_b_c(struct blockd*, short *dqc);
+#define vp8_dequantize_b vp8_dequantize_b_c
+
+void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *output, int stride);
+void vp8_dequant_idct_add_dspr2(short *input, short *dq, unsigned char *output, int stride);
+#define vp8_dequant_idct_add vp8_dequant_idct_add_dspr2
+
+void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+void vp8_dequant_idct_add_y_block_dspr2(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_dspr2
+
+void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+void vp8_dequant_idct_add_uv_block_dspr2(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_dspr2
+
+void vp8_loop_filter_mbv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbv_dspr2(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbv vp8_loop_filter_mbv_dspr2
+
+void vp8_loop_filter_bv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bv_dspr2(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bv vp8_loop_filter_bv_dspr2
+
+void vp8_loop_filter_mbh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbh_dspr2(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbh vp8_loop_filter_mbh_dspr2
+
+void vp8_loop_filter_bh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bh_dspr2(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bh vp8_loop_filter_bh_dspr2
+
+void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbv vp8_loop_filter_simple_vertical_edge_c
+
+void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbh vp8_loop_filter_simple_horizontal_edge_c
+
+void vp8_loop_filter_bvs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bv vp8_loop_filter_bvs_c
+
+void vp8_loop_filter_bhs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bh vp8_loop_filter_bhs_c
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+void vp8_short_idct4x4llm_dspr2(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+#define vp8_short_idct4x4llm vp8_short_idct4x4llm_dspr2
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *output);
+void vp8_short_inv_walsh4x4_1_dspr2(short *input, short *output);
+#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_dspr2
+
+void vp8_short_inv_walsh4x4_c(short *input, short *output);
+void vp8_short_inv_walsh4x4_dspr2(short *input, short *output);
+#define vp8_short_inv_walsh4x4 vp8_short_inv_walsh4x4_dspr2
+
+void vp8_dc_only_idct_add_c(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+void vp8_dc_only_idct_add_dspr2(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+#define vp8_dc_only_idct_add vp8_dc_only_idct_add_dspr2
+
+void vp8_copy_mem16x16_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem16x16_dspr2(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem16x16 vp8_copy_mem16x16_dspr2
+
+void vp8_copy_mem8x8_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x8_dspr2(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x8 vp8_copy_mem8x8_dspr2
+
+void vp8_copy_mem8x4_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x4_dspr2(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x4 vp8_copy_mem8x4_dspr2
+
+void vp8_build_intra_predictors_mby_s_c(struct macroblockd *x, unsigned char * yabove_row, unsigned char * yleft, int left_stride, unsigned char * ypred_ptr, int y_stride);
+#define vp8_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s_c
+
+void vp8_build_intra_predictors_mbuv_s_c(struct macroblockd *x, unsigned char * uabove_row, unsigned char * vabove_row, unsigned char *uleft, unsigned char *vleft, int left_stride, unsigned char * upred_ptr, unsigned char * vpred_ptr, int pred_stride);
+#define vp8_build_intra_predictors_mbuv_s vp8_build_intra_predictors_mbuv_s_c
+
+void vp8_intra4x4_predict_c(unsigned char *Above, unsigned char *yleft, int left_stride, B_PREDICTION_MODE b_mode, unsigned char *dst, int dst_stride, unsigned char top_left);
+#define vp8_intra4x4_predict vp8_intra4x4_predict_c
+
+void vp8_sixtap_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict16x16_dspr2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict16x16 vp8_sixtap_predict16x16_dspr2
+
+void vp8_sixtap_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x8_dspr2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x8 vp8_sixtap_predict8x8_dspr2
+
+void vp8_sixtap_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x4_dspr2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x4 vp8_sixtap_predict8x4_dspr2
+
+void vp8_sixtap_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict4x4_dspr2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict4x4 vp8_sixtap_predict4x4_dspr2
+
+void vp8_bilinear_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict16x16 vp8_bilinear_predict16x16_c
+
+void vp8_bilinear_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x8 vp8_bilinear_predict8x8_c
+
+void vp8_bilinear_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x4 vp8_bilinear_predict8x4_c
+
+void vp8_bilinear_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict4x4 vp8_bilinear_predict4x4_c
+
+unsigned int vp8_variance4x4_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance4x4 vp8_variance4x4_c
+
+unsigned int vp8_variance8x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x8 vp8_variance8x8_c
+
+unsigned int vp8_variance8x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x16 vp8_variance8x16_c
+
+unsigned int vp8_variance16x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x8 vp8_variance16x8_c
+
+unsigned int vp8_variance16x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x16 vp8_variance16x16_c
+
+unsigned int vp8_sub_pixel_variance4x4_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance4x4 vp8_sub_pixel_variance4x4_c
+
+unsigned int vp8_sub_pixel_variance8x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x8 vp8_sub_pixel_variance8x8_c
+
+unsigned int vp8_sub_pixel_variance8x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x16 vp8_sub_pixel_variance8x16_c
+
+unsigned int vp8_sub_pixel_variance16x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x8 vp8_sub_pixel_variance16x8_c
+
+unsigned int vp8_sub_pixel_variance16x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x16 vp8_sub_pixel_variance16x16_c
+
+unsigned int vp8_variance_halfpixvar16x16_h_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_c
+
+unsigned int vp8_variance_halfpixvar16x16_v_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_c
+
+unsigned int vp8_variance_halfpixvar16x16_hv_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_c
+
+unsigned int vp8_sad4x4_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad4x4 vp8_sad4x4_c
+
+unsigned int vp8_sad8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x8 vp8_sad8x8_c
+
+unsigned int vp8_sad8x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x16 vp8_sad8x16_c
+
+unsigned int vp8_sad16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x8 vp8_sad16x8_c
+
+unsigned int vp8_sad16x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x16 vp8_sad16x16_c
+
+void vp8_sad4x4x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x3 vp8_sad4x4x3_c
+
+void vp8_sad8x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x3 vp8_sad8x8x3_c
+
+void vp8_sad8x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x3 vp8_sad8x16x3_c
+
+void vp8_sad16x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x3 vp8_sad16x8x3_c
+
+void vp8_sad16x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x3 vp8_sad16x16x3_c
+
+void vp8_sad4x4x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad4x4x8 vp8_sad4x4x8_c
+
+void vp8_sad8x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x8x8 vp8_sad8x8x8_c
+
+void vp8_sad8x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x16x8 vp8_sad8x16x8_c
+
+void vp8_sad16x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x8x8 vp8_sad16x8x8_c
+
+void vp8_sad16x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x16x8 vp8_sad16x16x8_c
+
+void vp8_sad4x4x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x4d vp8_sad4x4x4d_c
+
+void vp8_sad8x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x4d vp8_sad8x8x4d_c
+
+void vp8_sad8x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x4d vp8_sad8x16x4d_c
+
+void vp8_sad16x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x4d vp8_sad16x8x4d_c
+
+void vp8_sad16x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x4d vp8_sad16x16x4d_c
+
+void vp8_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_4_5_scale vp8_horizontal_line_4_5_scale_c
+
+void vp8_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_4_5_scale vp8_vertical_band_4_5_scale_c
+
+void vp8_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_4_5_scale vp8_last_vertical_band_4_5_scale_c
+
+void vp8_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_3_scale vp8_horizontal_line_2_3_scale_c
+
+void vp8_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_3_scale vp8_vertical_band_2_3_scale_c
+
+void vp8_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_2_3_scale vp8_last_vertical_band_2_3_scale_c
+
+void vp8_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_5_scale vp8_horizontal_line_3_5_scale_c
+
+void vp8_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_5_scale vp8_vertical_band_3_5_scale_c
+
+void vp8_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_5_scale vp8_last_vertical_band_3_5_scale_c
+
+void vp8_horizontal_line_3_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_4_scale vp8_horizontal_line_3_4_scale_c
+
+void vp8_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_4_scale vp8_vertical_band_3_4_scale_c
+
+void vp8_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_4_scale vp8_last_vertical_band_3_4_scale_c
+
+void vp8_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_1_2_scale vp8_horizontal_line_1_2_scale_c
+
+void vp8_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_1_2_scale vp8_vertical_band_1_2_scale_c
+
+void vp8_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_1_2_scale vp8_last_vertical_band_1_2_scale_c
+
+void vp8_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_4_scale vp8_horizontal_line_5_4_scale_c
+
+void vp8_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_4_scale vp8_vertical_band_5_4_scale_c
+
+void vp8_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_3_scale vp8_horizontal_line_5_3_scale_c
+
+void vp8_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_3_scale vp8_vertical_band_5_3_scale_c
+
+void vp8_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_1_scale vp8_horizontal_line_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale vp8_vertical_band_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale_i vp8_vertical_band_2_1_scale_i_c
+
+void vp8_yv12_extend_frame_borders_c(struct yv12_buffer_config *ybf);
+#define vp8_yv12_extend_frame_borders vp8_yv12_extend_frame_borders_c
+
+void vp8_yv12_copy_frame_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_frame vp8_yv12_copy_frame_c
+
+void vp8_yv12_copy_y_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_y vp8_yv12_copy_y_c
+
+void vpx_rtcd(void);
+#include "vpx_config.h"
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+
+#if HAVE_DSPR2
+void dsputil_static_init();
+dsputil_static_init();
+#endif
+}
+#endif
+#endif
diff --git a/mips-dspr2/vpx_version.h b/mips-dspr2/vpx_version.h
new file mode 100644
index 0000000..663dd49
--- /dev/null
+++ b/mips-dspr2/vpx_version.h
@@ -0,0 +1,7 @@
+#define VERSION_MAJOR 1
+#define VERSION_MINOR 1
+#define VERSION_PATCH 0
+#define VERSION_EXTRA ""
+#define VERSION_PACKED ((VERSION_MAJOR<<16)|(VERSION_MINOR<<8)|(VERSION_PATCH))
+#define VERSION_STRING_NOSP "v1.1.0"
+#define VERSION_STRING " v1.1.0"
diff --git a/mips/.bins b/mips/.bins
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/mips/.bins
diff --git a/mips/.docs b/mips/.docs
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/mips/.docs
diff --git a/mips/.libs b/mips/.libs
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/mips/.libs
diff --git a/mips/libvpx_srcs.txt b/mips/libvpx_srcs.txt
new file mode 100644
index 0000000..4ee59ef
--- /dev/null
+++ b/mips/libvpx_srcs.txt
@@ -0,0 +1,113 @@
+build/make/rtcd.sh
+build/make/version.sh
+CHANGELOG
+libs.mk
+vp8/common/alloccommon.c
+vp8/common/alloccommon.h
+vp8/common/asm_com_offsets.c
+vp8/common/blockd.c
+vp8/common/blockd.h
+vp8/common/coefupdateprobs.h
+vp8/common/common.h
+vp8/common/debugmodes.c
+vp8/common/default_coef_probs.h
+vp8/common/dequantize.c
+vp8/common/entropy.c
+vp8/common/entropy.h
+vp8/common/entropymode.c
+vp8/common/entropymode.h
+vp8/common/entropymv.c
+vp8/common/entropymv.h
+vp8/common/extend.c
+vp8/common/extend.h
+vp8/common/filter.c
+vp8/common/filter.h
+vp8/common/findnearmv.c
+vp8/common/findnearmv.h
+vp8/common/generic/systemdependent.c
+vp8/common/header.h
+vp8/common/idct_blk.c
+vp8/common/idctllm.c
+vp8/common/invtrans.h
+vp8/common/loopfilter.c
+vp8/common/loopfilter_filters.c
+vp8/common/loopfilter.h
+vp8/common/mbpitch.c
+vp8/common/modecont.c
+vp8/common/modecont.h
+vp8/common/mv.h
+vp8/common/onyxc_int.h
+vp8/common/onyxd.h
+vp8/common/onyx.h
+vp8/common/ppflags.h
+vp8/common/pragmas.h
+vp8/common/quant_common.c
+vp8/common/quant_common.h
+vp8/common/reconinter.c
+vp8/common/reconinter.h
+vp8/common/reconintra4x4.c
+vp8/common/reconintra4x4.h
+vp8/common/reconintra.c
+vp8/common/rtcd.c
+vp8/common/rtcd_defs.sh
+vp8/common/sad_c.c
+vp8/common/setupintrarecon.c
+vp8/common/setupintrarecon.h
+vp8/common/swapyv12buffer.c
+vp8/common/swapyv12buffer.h
+vp8/common/systemdependent.h
+vp8/common/threading.h
+vp8/common/treecoder.c
+vp8/common/treecoder.h
+vp8/common/variance_c.c
+vp8/common/variance.h
+vp8/common/vp8_entropymodedata.h
+vp8/decoder/asm_dec_offsets.c
+vp8/decoder/dboolhuff.c
+vp8/decoder/dboolhuff.h
+vp8/decoder/decodemv.c
+vp8/decoder/decodemv.h
+vp8/decoder/decoderthreading.h
+vp8/decoder/decodframe.c
+vp8/decoder/detokenize.c
+vp8/decoder/detokenize.h
+vp8/decoder/onyxd_if.c
+vp8/decoder/onyxd_int.h
+vp8/decoder/threading.c
+vp8/decoder/treereader.h
+vp8/vp8_common.mk
+vp8/vp8_dx_iface.c
+vp8/vp8dx.mk
+vpx_config.c
+vpx/internal/vpx_codec_internal.h
+vpx_mem/include/vpx_mem_intrnl.h
+vpx_mem/vpx_mem.c
+vpx_mem/vpx_mem.h
+vpx_mem/vpx_mem.mk
+vpx_ports/asm_offsets.h
+vpx_ports/mem.h
+vpx_ports/vpx_ports.mk
+vpx_ports/vpx_timer.h
+vpx_scale/generic/gen_scalers.c
+vpx_scale/generic/vpxscale.c
+vpx_scale/generic/yv12config.c
+vpx_scale/generic/yv12extend.c
+vpx_scale/generic/yv12extend_generic.h
+vpx_scale/scale_mode.h
+vpx_scale/vpxscale.h
+vpx_scale/vpx_scale.mk
+vpx_scale/yv12config.h
+vpx/src/vpx_codec.c
+vpx/src/vpx_decoder.c
+vpx/src/vpx_encoder.c
+vpx/src/vpx_image.c
+vpx/vp8dx.h
+vpx/vp8.h
+vpx/vpx_codec.h
+vpx/vpx_codec_impl_bottom.h
+vpx/vpx_codec_impl_top.h
+vpx/vpx_codec.mk
+vpx/vpx_decoder.h
+vpx/vpx_encoder.h
+vpx/vpx_image.h
+vpx/vpx_integer.h
diff --git a/mips/vpx_config.c b/mips/vpx_config.c
new file mode 100644
index 0000000..1a07e06
--- /dev/null
+++ b/mips/vpx_config.c
@@ -0,0 +1,9 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+static const char* const cfg = "--force-target=mips32-android-gcc --disable-runtime-cpu-detect --sdk-path=/usr/local/google/home/johannkoenig/android-ndk --disable-vp8-encoder --disable-examples --disable-docs";
+const char *vpx_codec_build_config(void) {return cfg;}
diff --git a/mips/vpx_config.h b/mips/vpx_config.h
new file mode 100644
index 0000000..c567363
--- /dev/null
+++ b/mips/vpx_config.h
@@ -0,0 +1,80 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/* */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS. All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+/* This file automatically generated by configure. Do not edit! */
+#ifndef VPX_CONFIG_H
+#define VPX_CONFIG_H
+#define RESTRICT
+#define ARCH_ARM 0
+#define ARCH_MIPS 1
+#define ARCH_X86 0
+#define ARCH_X86_64 0
+#define ARCH_PPC32 0
+#define ARCH_PPC64 0
+#define HAVE_EDSP 0
+#define HAVE_MEDIA 0
+#define HAVE_NEON 0
+#define HAVE_MIPS32 1
+#define HAVE_DSPR2 0
+#define HAVE_MMX 0
+#define HAVE_SSE 0
+#define HAVE_SSE2 0
+#define HAVE_SSE3 0
+#define HAVE_SSSE3 0
+#define HAVE_SSE4_1 0
+#define HAVE_ALTIVEC 0
+#define HAVE_VPX_PORTS 1
+#define HAVE_STDINT_H 1
+#define HAVE_ALT_TREE_LAYOUT 0
+#define HAVE_PTHREAD_H 1
+#define HAVE_SYS_MMAN_H 1
+#define HAVE_UNISTD_H 1
+#define CONFIG_EXTERNAL_BUILD 0
+#define CONFIG_INSTALL_DOCS 1
+#define CONFIG_INSTALL_BINS 1
+#define CONFIG_INSTALL_LIBS 1
+#define CONFIG_INSTALL_SRCS 0
+#define CONFIG_DEBUG 0
+#define CONFIG_GPROF 0
+#define CONFIG_GCOV 0
+#define CONFIG_RVCT 0
+#define CONFIG_GCC 1
+#define CONFIG_MSVS 0
+#define CONFIG_PIC 0
+#define CONFIG_BIG_ENDIAN 0
+#define CONFIG_CODEC_SRCS 0
+#define CONFIG_DEBUG_LIBS 0
+#define CONFIG_FAST_UNALIGNED 1
+#define CONFIG_MEM_MANAGER 0
+#define CONFIG_MEM_TRACKER 0
+#define CONFIG_MEM_CHECKS 0
+#define CONFIG_MD5 1
+#define CONFIG_DEQUANT_TOKENS 1
+#define CONFIG_DC_RECON 1
+#define CONFIG_RUNTIME_CPU_DETECT 0
+#define CONFIG_POSTPROC 0
+#define CONFIG_MULTITHREAD 1
+#define CONFIG_INTERNAL_STATS 0
+#define CONFIG_VP8_ENCODER 0
+#define CONFIG_VP8_DECODER 1
+#define CONFIG_VP8 1
+#define CONFIG_ENCODERS 0
+#define CONFIG_DECODERS 1
+#define CONFIG_STATIC_MSVCRT 0
+#define CONFIG_SPATIAL_RESAMPLING 1
+#define CONFIG_REALTIME_ONLY 0
+#define CONFIG_ONTHEFLY_BITPACKING 0
+#define CONFIG_ERROR_CONCEALMENT 0
+#define CONFIG_SHARED 0
+#define CONFIG_STATIC 1
+#define CONFIG_SMALL 0
+#define CONFIG_POSTPROC_VISUALIZER 0
+#define CONFIG_OS_SUPPORT 1
+#define CONFIG_UNIT_TESTS 0
+#define CONFIG_MULTI_RES_ENCODING 0
+#define CONFIG_TEMPORAL_DENOISING 1
+#endif /* VPX_CONFIG_H */
diff --git a/mips/vpx_rtcd.h b/mips/vpx_rtcd.h
new file mode 100644
index 0000000..00d86af
--- /dev/null
+++ b/mips/vpx_rtcd.h
@@ -0,0 +1,300 @@
+#ifndef VPX_RTCD_
+#define VPX_RTCD_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+#include "vp8/common/blockd.h"
+
+struct blockd;
+struct macroblockd;
+struct loop_filter_info;
+
+/* Encoder forward decls */
+struct block;
+struct macroblock;
+struct variance_vtable;
+union int_mv;
+struct yv12_buffer_config;
+
+void vp8_dequantize_b_c(struct blockd*, short *dqc);
+#define vp8_dequantize_b vp8_dequantize_b_c
+
+void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *output, int stride);
+#define vp8_dequant_idct_add vp8_dequant_idct_add_c
+
+void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_c
+
+void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_c
+
+void vp8_loop_filter_mbv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbv vp8_loop_filter_mbv_c
+
+void vp8_loop_filter_bv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bv vp8_loop_filter_bv_c
+
+void vp8_loop_filter_mbh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbh vp8_loop_filter_mbh_c
+
+void vp8_loop_filter_bh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bh vp8_loop_filter_bh_c
+
+void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbv vp8_loop_filter_simple_vertical_edge_c
+
+void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbh vp8_loop_filter_simple_horizontal_edge_c
+
+void vp8_loop_filter_bvs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bv vp8_loop_filter_bvs_c
+
+void vp8_loop_filter_bhs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bh vp8_loop_filter_bhs_c
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+#define vp8_short_idct4x4llm vp8_short_idct4x4llm_c
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_c
+
+void vp8_short_inv_walsh4x4_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4 vp8_short_inv_walsh4x4_c
+
+void vp8_dc_only_idct_add_c(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+#define vp8_dc_only_idct_add vp8_dc_only_idct_add_c
+
+void vp8_copy_mem16x16_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem16x16 vp8_copy_mem16x16_c
+
+void vp8_copy_mem8x8_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x8 vp8_copy_mem8x8_c
+
+void vp8_copy_mem8x4_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x4 vp8_copy_mem8x4_c
+
+void vp8_build_intra_predictors_mby_s_c(struct macroblockd *x, unsigned char * yabove_row, unsigned char * yleft, int left_stride, unsigned char * ypred_ptr, int y_stride);
+#define vp8_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s_c
+
+void vp8_build_intra_predictors_mbuv_s_c(struct macroblockd *x, unsigned char * uabove_row, unsigned char * vabove_row, unsigned char *uleft, unsigned char *vleft, int left_stride, unsigned char * upred_ptr, unsigned char * vpred_ptr, int pred_stride);
+#define vp8_build_intra_predictors_mbuv_s vp8_build_intra_predictors_mbuv_s_c
+
+void vp8_intra4x4_predict_c(unsigned char *Above, unsigned char *yleft, int left_stride, B_PREDICTION_MODE b_mode, unsigned char *dst, int dst_stride, unsigned char top_left);
+#define vp8_intra4x4_predict vp8_intra4x4_predict_c
+
+void vp8_sixtap_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict16x16 vp8_sixtap_predict16x16_c
+
+void vp8_sixtap_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x8 vp8_sixtap_predict8x8_c
+
+void vp8_sixtap_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x4 vp8_sixtap_predict8x4_c
+
+void vp8_sixtap_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict4x4 vp8_sixtap_predict4x4_c
+
+void vp8_bilinear_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict16x16 vp8_bilinear_predict16x16_c
+
+void vp8_bilinear_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x8 vp8_bilinear_predict8x8_c
+
+void vp8_bilinear_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x4 vp8_bilinear_predict8x4_c
+
+void vp8_bilinear_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict4x4 vp8_bilinear_predict4x4_c
+
+unsigned int vp8_variance4x4_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance4x4 vp8_variance4x4_c
+
+unsigned int vp8_variance8x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x8 vp8_variance8x8_c
+
+unsigned int vp8_variance8x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance8x16 vp8_variance8x16_c
+
+unsigned int vp8_variance16x8_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x8 vp8_variance16x8_c
+
+unsigned int vp8_variance16x16_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance16x16 vp8_variance16x16_c
+
+unsigned int vp8_sub_pixel_variance4x4_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance4x4 vp8_sub_pixel_variance4x4_c
+
+unsigned int vp8_sub_pixel_variance8x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x8 vp8_sub_pixel_variance8x8_c
+
+unsigned int vp8_sub_pixel_variance8x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance8x16 vp8_sub_pixel_variance8x16_c
+
+unsigned int vp8_sub_pixel_variance16x8_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x8 vp8_sub_pixel_variance16x8_c
+
+unsigned int vp8_sub_pixel_variance16x16_c(const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse);
+#define vp8_sub_pixel_variance16x16 vp8_sub_pixel_variance16x16_c
+
+unsigned int vp8_variance_halfpixvar16x16_h_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_c
+
+unsigned int vp8_variance_halfpixvar16x16_v_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_c
+
+unsigned int vp8_variance_halfpixvar16x16_hv_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse);
+#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_c
+
+unsigned int vp8_sad4x4_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad4x4 vp8_sad4x4_c
+
+unsigned int vp8_sad8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x8 vp8_sad8x8_c
+
+unsigned int vp8_sad8x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad8x16 vp8_sad8x16_c
+
+unsigned int vp8_sad16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x8 vp8_sad16x8_c
+
+unsigned int vp8_sad16x16_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad);
+#define vp8_sad16x16 vp8_sad16x16_c
+
+void vp8_sad4x4x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x3 vp8_sad4x4x3_c
+
+void vp8_sad8x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x3 vp8_sad8x8x3_c
+
+void vp8_sad8x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x3 vp8_sad8x16x3_c
+
+void vp8_sad16x8x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x3 vp8_sad16x8x3_c
+
+void vp8_sad16x16x3_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x3 vp8_sad16x16x3_c
+
+void vp8_sad4x4x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad4x4x8 vp8_sad4x4x8_c
+
+void vp8_sad8x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x8x8 vp8_sad8x8x8_c
+
+void vp8_sad8x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad8x16x8 vp8_sad8x16x8_c
+
+void vp8_sad16x8x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x8x8 vp8_sad16x8x8_c
+
+void vp8_sad16x16x8_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array);
+#define vp8_sad16x16x8 vp8_sad16x16x8_c
+
+void vp8_sad4x4x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad4x4x4d vp8_sad4x4x4d_c
+
+void vp8_sad8x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x8x4d vp8_sad8x8x4d_c
+
+void vp8_sad8x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad8x16x4d vp8_sad8x16x4d_c
+
+void vp8_sad16x8x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x8x4d vp8_sad16x8x4d_c
+
+void vp8_sad16x16x4d_c(const unsigned char *src_ptr, int src_stride, const unsigned char * const ref_ptr[], int ref_stride, unsigned int *sad_array);
+#define vp8_sad16x16x4d vp8_sad16x16x4d_c
+
+void vp8_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_4_5_scale vp8_horizontal_line_4_5_scale_c
+
+void vp8_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_4_5_scale vp8_vertical_band_4_5_scale_c
+
+void vp8_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_4_5_scale vp8_last_vertical_band_4_5_scale_c
+
+void vp8_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_3_scale vp8_horizontal_line_2_3_scale_c
+
+void vp8_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_3_scale vp8_vertical_band_2_3_scale_c
+
+void vp8_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_2_3_scale vp8_last_vertical_band_2_3_scale_c
+
+void vp8_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_5_scale vp8_horizontal_line_3_5_scale_c
+
+void vp8_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_5_scale vp8_vertical_band_3_5_scale_c
+
+void vp8_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_5_scale vp8_last_vertical_band_3_5_scale_c
+
+void vp8_horizontal_line_3_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_3_4_scale vp8_horizontal_line_3_4_scale_c
+
+void vp8_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_3_4_scale vp8_vertical_band_3_4_scale_c
+
+void vp8_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_3_4_scale vp8_last_vertical_band_3_4_scale_c
+
+void vp8_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_1_2_scale vp8_horizontal_line_1_2_scale_c
+
+void vp8_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_1_2_scale vp8_vertical_band_1_2_scale_c
+
+void vp8_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_last_vertical_band_1_2_scale vp8_last_vertical_band_1_2_scale_c
+
+void vp8_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_4_scale vp8_horizontal_line_5_4_scale_c
+
+void vp8_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_4_scale vp8_vertical_band_5_4_scale_c
+
+void vp8_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_5_3_scale vp8_horizontal_line_5_3_scale_c
+
+void vp8_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_5_3_scale vp8_vertical_band_5_3_scale_c
+
+void vp8_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
+#define vp8_horizontal_line_2_1_scale vp8_horizontal_line_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale vp8_vertical_band_2_1_scale_c
+
+void vp8_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
+#define vp8_vertical_band_2_1_scale_i vp8_vertical_band_2_1_scale_i_c
+
+void vp8_yv12_extend_frame_borders_c(struct yv12_buffer_config *ybf);
+#define vp8_yv12_extend_frame_borders vp8_yv12_extend_frame_borders_c
+
+void vp8_yv12_copy_frame_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_frame vp8_yv12_copy_frame_c
+
+void vp8_yv12_copy_y_c(struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_y vp8_yv12_copy_y_c
+
+void vpx_rtcd(void);
+#include "vpx_config.h"
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+
+#if HAVE_DSPR2
+void dsputil_static_init();
+dsputil_static_init();
+#endif
+}
+#endif
+#endif
diff --git a/mips/vpx_version.h b/mips/vpx_version.h
new file mode 100644
index 0000000..663dd49
--- /dev/null
+++ b/mips/vpx_version.h
@@ -0,0 +1,7 @@
+#define VERSION_MAJOR 1
+#define VERSION_MINOR 1
+#define VERSION_PATCH 0
+#define VERSION_EXTRA ""
+#define VERSION_PACKED ((VERSION_MAJOR<<16)|(VERSION_MINOR<<8)|(VERSION_PATCH))
+#define VERSION_STRING_NOSP "v1.1.0"
+#define VERSION_STRING " v1.1.0"
diff --git a/nestegg/src/nestegg.c b/nestegg/src/nestegg.c
deleted file mode 100644
index 63a0e83..0000000
--- a/nestegg/src/nestegg.c
+++ /dev/null
@@ -1,1938 +0,0 @@
-/*
- * Copyright © 2010 Mozilla Foundation
- *
- * This program is made available under an ISC-style license. See the
- * accompanying file LICENSE for details.
- */
-#include <assert.h>
-#include <stdlib.h>
-#include <string.h>
-
-#include "nestegg/halloc/halloc.h"
-#include "nestegg/include/nestegg/nestegg.h"
-
-/* EBML Elements */
-#define ID_EBML 0x1a45dfa3
-#define ID_EBML_VERSION 0x4286
-#define ID_EBML_READ_VERSION 0x42f7
-#define ID_EBML_MAX_ID_LENGTH 0x42f2
-#define ID_EBML_MAX_SIZE_LENGTH 0x42f3
-#define ID_DOCTYPE 0x4282
-#define ID_DOCTYPE_VERSION 0x4287
-#define ID_DOCTYPE_READ_VERSION 0x4285
-
-/* Global Elements */
-#define ID_VOID 0xec
-#define ID_CRC32 0xbf
-
-/* WebMedia Elements */
-#define ID_SEGMENT 0x18538067
-
-/* Seek Head Elements */
-#define ID_SEEK_HEAD 0x114d9b74
-#define ID_SEEK 0x4dbb
-#define ID_SEEK_ID 0x53ab
-#define ID_SEEK_POSITION 0x53ac
-
-/* Info Elements */
-#define ID_INFO 0x1549a966
-#define ID_TIMECODE_SCALE 0x2ad7b1
-#define ID_DURATION 0x4489
-
-/* Cluster Elements */
-#define ID_CLUSTER 0x1f43b675
-#define ID_TIMECODE 0xe7
-#define ID_BLOCK_GROUP 0xa0
-#define ID_SIMPLE_BLOCK 0xa3
-
-/* BlockGroup Elements */
-#define ID_BLOCK 0xa1
-#define ID_BLOCK_DURATION 0x9b
-#define ID_REFERENCE_BLOCK 0xfb
-
-/* Tracks Elements */
-#define ID_TRACKS 0x1654ae6b
-#define ID_TRACK_ENTRY 0xae
-#define ID_TRACK_NUMBER 0xd7
-#define ID_TRACK_UID 0x73c5
-#define ID_TRACK_TYPE 0x83
-#define ID_FLAG_ENABLED 0xb9
-#define ID_FLAG_DEFAULT 0x88
-#define ID_FLAG_LACING 0x9c
-#define ID_TRACK_TIMECODE_SCALE 0x23314f
-#define ID_LANGUAGE 0x22b59c
-#define ID_CODEC_ID 0x86
-#define ID_CODEC_PRIVATE 0x63a2
-
-/* Video Elements */
-#define ID_VIDEO 0xe0
-#define ID_PIXEL_WIDTH 0xb0
-#define ID_PIXEL_HEIGHT 0xba
-#define ID_PIXEL_CROP_BOTTOM 0x54aa
-#define ID_PIXEL_CROP_TOP 0x54bb
-#define ID_PIXEL_CROP_LEFT 0x54cc
-#define ID_PIXEL_CROP_RIGHT 0x54dd
-#define ID_DISPLAY_WIDTH 0x54b0
-#define ID_DISPLAY_HEIGHT 0x54ba
-
-/* Audio Elements */
-#define ID_AUDIO 0xe1
-#define ID_SAMPLING_FREQUENCY 0xb5
-#define ID_CHANNELS 0x9f
-#define ID_BIT_DEPTH 0x6264
-
-/* Cues Elements */
-#define ID_CUES 0x1c53bb6b
-#define ID_CUE_POINT 0xbb
-#define ID_CUE_TIME 0xb3
-#define ID_CUE_TRACK_POSITIONS 0xb7
-#define ID_CUE_TRACK 0xf7
-#define ID_CUE_CLUSTER_POSITION 0xf1
-#define ID_CUE_BLOCK_NUMBER 0x5378
-
-/* EBML Types */
-enum ebml_type_enum {
- TYPE_UNKNOWN,
- TYPE_MASTER,
- TYPE_UINT,
- TYPE_FLOAT,
- TYPE_INT,
- TYPE_STRING,
- TYPE_BINARY
-};
-
-#define LIMIT_STRING (1 << 20)
-#define LIMIT_BINARY (1 << 24)
-#define LIMIT_BLOCK (1 << 30)
-#define LIMIT_FRAME (1 << 28)
-
-/* Field Flags */
-#define DESC_FLAG_NONE 0
-#define DESC_FLAG_MULTI (1 << 0)
-#define DESC_FLAG_SUSPEND (1 << 1)
-#define DESC_FLAG_OFFSET (1 << 2)
-
-/* Block Header Flags */
-#define BLOCK_FLAGS_LACING 6
-
-/* Lacing Constants */
-#define LACING_NONE 0
-#define LACING_XIPH 1
-#define LACING_FIXED 2
-#define LACING_EBML 3
-
-/* Track Types */
-#define TRACK_TYPE_VIDEO 1
-#define TRACK_TYPE_AUDIO 2
-
-/* Track IDs */
-#define TRACK_ID_VP8 "V_VP8"
-#define TRACK_ID_VORBIS "A_VORBIS"
-
-enum vint_mask {
- MASK_NONE,
- MASK_FIRST_BIT
-};
-
-struct ebml_binary {
- unsigned char * data;
- size_t length;
-};
-
-struct ebml_list_node {
- struct ebml_list_node * next;
- uint64_t id;
- void * data;
-};
-
-struct ebml_list {
- struct ebml_list_node * head;
- struct ebml_list_node * tail;
-};
-
-struct ebml_type {
- union ebml_value {
- uint64_t u;
- double f;
- int64_t i;
- char * s;
- struct ebml_binary b;
- } v;
- enum ebml_type_enum type;
- int read;
-};
-
-/* EBML Definitions */
-struct ebml {
- struct ebml_type ebml_version;
- struct ebml_type ebml_read_version;
- struct ebml_type ebml_max_id_length;
- struct ebml_type ebml_max_size_length;
- struct ebml_type doctype;
- struct ebml_type doctype_version;
- struct ebml_type doctype_read_version;
-};
-
-/* Matroksa Definitions */
-struct seek {
- struct ebml_type id;
- struct ebml_type position;
-};
-
-struct seek_head {
- struct ebml_list seek;
-};
-
-struct info {
- struct ebml_type timecode_scale;
- struct ebml_type duration;
-};
-
-struct block_group {
- struct ebml_type duration;
- struct ebml_type reference_block;
-};
-
-struct cluster {
- struct ebml_type timecode;
- struct ebml_list block_group;
-};
-
-struct video {
- struct ebml_type pixel_width;
- struct ebml_type pixel_height;
- struct ebml_type pixel_crop_bottom;
- struct ebml_type pixel_crop_top;
- struct ebml_type pixel_crop_left;
- struct ebml_type pixel_crop_right;
- struct ebml_type display_width;
- struct ebml_type display_height;
-};
-
-struct audio {
- struct ebml_type sampling_frequency;
- struct ebml_type channels;
- struct ebml_type bit_depth;
-};
-
-struct track_entry {
- struct ebml_type number;
- struct ebml_type uid;
- struct ebml_type type;
- struct ebml_type flag_enabled;
- struct ebml_type flag_default;
- struct ebml_type flag_lacing;
- struct ebml_type track_timecode_scale;
- struct ebml_type language;
- struct ebml_type codec_id;
- struct ebml_type codec_private;
- struct video video;
- struct audio audio;
-};
-
-struct tracks {
- struct ebml_list track_entry;
-};
-
-struct cue_track_positions {
- struct ebml_type track;
- struct ebml_type cluster_position;
- struct ebml_type block_number;
-};
-
-struct cue_point {
- struct ebml_type time;
- struct ebml_list cue_track_positions;
-};
-
-struct cues {
- struct ebml_list cue_point;
-};
-
-struct segment {
- struct ebml_list seek_head;
- struct info info;
- struct ebml_list cluster;
- struct tracks tracks;
- struct cues cues;
-};
-
-/* Misc. */
-struct pool_ctx {
- char dummy;
-};
-
-struct list_node {
- struct list_node * previous;
- struct ebml_element_desc * node;
- unsigned char * data;
-};
-
-struct saved_state {
- int64_t stream_offset;
- struct list_node * ancestor;
- uint64_t last_id;
- uint64_t last_size;
-};
-
-struct frame {
- unsigned char * data;
- size_t length;
- struct frame * next;
-};
-
-/* Public (opaque) Structures */
-struct nestegg {
- nestegg_io * io;
- nestegg_log log;
- struct pool_ctx * alloc_pool;
- uint64_t last_id;
- uint64_t last_size;
- struct list_node * ancestor;
- struct ebml ebml;
- struct segment segment;
- int64_t segment_offset;
- unsigned int track_count;
-};
-
-struct nestegg_packet {
- uint64_t track;
- uint64_t timecode;
- struct frame * frame;
-};
-
-/* Element Descriptor */
-struct ebml_element_desc {
- char const * name;
- uint64_t id;
- enum ebml_type_enum type;
- size_t offset;
- unsigned int flags;
- struct ebml_element_desc * children;
- size_t size;
- size_t data_offset;
-};
-
-#define E_FIELD(ID, TYPE, STRUCT, FIELD) \
- { #ID, ID, TYPE, offsetof(STRUCT, FIELD), DESC_FLAG_NONE, NULL, 0, 0 }
-#define E_MASTER(ID, TYPE, STRUCT, FIELD) \
- { #ID, ID, TYPE, offsetof(STRUCT, FIELD), DESC_FLAG_MULTI, ne_ ## FIELD ## _elements, \
- sizeof(struct FIELD), 0 }
-#define E_SINGLE_MASTER_O(ID, TYPE, STRUCT, FIELD) \
- { #ID, ID, TYPE, offsetof(STRUCT, FIELD), DESC_FLAG_OFFSET, ne_ ## FIELD ## _elements, 0, \
- offsetof(STRUCT, FIELD ## _offset) }
-#define E_SINGLE_MASTER(ID, TYPE, STRUCT, FIELD) \
- { #ID, ID, TYPE, offsetof(STRUCT, FIELD), DESC_FLAG_NONE, ne_ ## FIELD ## _elements, 0, 0 }
-#define E_SUSPEND(ID, TYPE) \
- { #ID, ID, TYPE, 0, DESC_FLAG_SUSPEND, NULL, 0, 0 }
-#define E_LAST \
- { NULL, 0, 0, 0, DESC_FLAG_NONE, NULL, 0, 0 }
-
-/* EBML Element Lists */
-static struct ebml_element_desc ne_ebml_elements[] = {
- E_FIELD(ID_EBML_VERSION, TYPE_UINT, struct ebml, ebml_version),
- E_FIELD(ID_EBML_READ_VERSION, TYPE_UINT, struct ebml, ebml_read_version),
- E_FIELD(ID_EBML_MAX_ID_LENGTH, TYPE_UINT, struct ebml, ebml_max_id_length),
- E_FIELD(ID_EBML_MAX_SIZE_LENGTH, TYPE_UINT, struct ebml, ebml_max_size_length),
- E_FIELD(ID_DOCTYPE, TYPE_STRING, struct ebml, doctype),
- E_FIELD(ID_DOCTYPE_VERSION, TYPE_UINT, struct ebml, doctype_version),
- E_FIELD(ID_DOCTYPE_READ_VERSION, TYPE_UINT, struct ebml, doctype_read_version),
- E_LAST
-};
-
-/* WebMedia Element Lists */
-static struct ebml_element_desc ne_seek_elements[] = {
- E_FIELD(ID_SEEK_ID, TYPE_BINARY, struct seek, id),
- E_FIELD(ID_SEEK_POSITION, TYPE_UINT, struct seek, position),
- E_LAST
-};
-
-static struct ebml_element_desc ne_seek_head_elements[] = {
- E_MASTER(ID_SEEK, TYPE_MASTER, struct seek_head, seek),
- E_LAST
-};
-
-static struct ebml_element_desc ne_info_elements[] = {
- E_FIELD(ID_TIMECODE_SCALE, TYPE_UINT, struct info, timecode_scale),
- E_FIELD(ID_DURATION, TYPE_FLOAT, struct info, duration),
- E_LAST
-};
-
-static struct ebml_element_desc ne_block_group_elements[] = {
- E_SUSPEND(ID_BLOCK, TYPE_BINARY),
- E_FIELD(ID_BLOCK_DURATION, TYPE_UINT, struct block_group, duration),
- E_FIELD(ID_REFERENCE_BLOCK, TYPE_INT, struct block_group, reference_block),
- E_LAST
-};
-
-static struct ebml_element_desc ne_cluster_elements[] = {
- E_FIELD(ID_TIMECODE, TYPE_UINT, struct cluster, timecode),
- E_MASTER(ID_BLOCK_GROUP, TYPE_MASTER, struct cluster, block_group),
- E_SUSPEND(ID_SIMPLE_BLOCK, TYPE_BINARY),
- E_LAST
-};
-
-static struct ebml_element_desc ne_video_elements[] = {
- E_FIELD(ID_PIXEL_WIDTH, TYPE_UINT, struct video, pixel_width),
- E_FIELD(ID_PIXEL_HEIGHT, TYPE_UINT, struct video, pixel_height),
- E_FIELD(ID_PIXEL_CROP_BOTTOM, TYPE_UINT, struct video, pixel_crop_bottom),
- E_FIELD(ID_PIXEL_CROP_TOP, TYPE_UINT, struct video, pixel_crop_top),
- E_FIELD(ID_PIXEL_CROP_LEFT, TYPE_UINT, struct video, pixel_crop_left),
- E_FIELD(ID_PIXEL_CROP_RIGHT, TYPE_UINT, struct video, pixel_crop_right),
- E_FIELD(ID_DISPLAY_WIDTH, TYPE_UINT, struct video, display_width),
- E_FIELD(ID_DISPLAY_HEIGHT, TYPE_UINT, struct video, display_height),
- E_LAST
-};
-
-static struct ebml_element_desc ne_audio_elements[] = {
- E_FIELD(ID_SAMPLING_FREQUENCY, TYPE_FLOAT, struct audio, sampling_frequency),
- E_FIELD(ID_CHANNELS, TYPE_UINT, struct audio, channels),
- E_FIELD(ID_BIT_DEPTH, TYPE_UINT, struct audio, bit_depth),
- E_LAST
-};
-
-static struct ebml_element_desc ne_track_entry_elements[] = {
- E_FIELD(ID_TRACK_NUMBER, TYPE_UINT, struct track_entry, number),
- E_FIELD(ID_TRACK_UID, TYPE_UINT, struct track_entry, uid),
- E_FIELD(ID_TRACK_TYPE, TYPE_UINT, struct track_entry, type),
- E_FIELD(ID_FLAG_ENABLED, TYPE_UINT, struct track_entry, flag_enabled),
- E_FIELD(ID_FLAG_DEFAULT, TYPE_UINT, struct track_entry, flag_default),
- E_FIELD(ID_FLAG_LACING, TYPE_UINT, struct track_entry, flag_lacing),
- E_FIELD(ID_TRACK_TIMECODE_SCALE, TYPE_FLOAT, struct track_entry, track_timecode_scale),
- E_FIELD(ID_LANGUAGE, TYPE_STRING, struct track_entry, language),
- E_FIELD(ID_CODEC_ID, TYPE_STRING, struct track_entry, codec_id),
- E_FIELD(ID_CODEC_PRIVATE, TYPE_BINARY, struct track_entry, codec_private),
- E_SINGLE_MASTER(ID_VIDEO, TYPE_MASTER, struct track_entry, video),
- E_SINGLE_MASTER(ID_AUDIO, TYPE_MASTER, struct track_entry, audio),
- E_LAST
-};
-
-static struct ebml_element_desc ne_tracks_elements[] = {
- E_MASTER(ID_TRACK_ENTRY, TYPE_MASTER, struct tracks, track_entry),
- E_LAST
-};
-
-static struct ebml_element_desc ne_cue_track_positions_elements[] = {
- E_FIELD(ID_CUE_TRACK, TYPE_UINT, struct cue_track_positions, track),
- E_FIELD(ID_CUE_CLUSTER_POSITION, TYPE_UINT, struct cue_track_positions, cluster_position),
- E_FIELD(ID_CUE_BLOCK_NUMBER, TYPE_UINT, struct cue_track_positions, block_number),
- E_LAST
-};
-
-static struct ebml_element_desc ne_cue_point_elements[] = {
- E_FIELD(ID_CUE_TIME, TYPE_UINT, struct cue_point, time),
- E_MASTER(ID_CUE_TRACK_POSITIONS, TYPE_MASTER, struct cue_point, cue_track_positions),
- E_LAST
-};
-
-static struct ebml_element_desc ne_cues_elements[] = {
- E_MASTER(ID_CUE_POINT, TYPE_MASTER, struct cues, cue_point),
- E_LAST
-};
-
-static struct ebml_element_desc ne_segment_elements[] = {
- E_MASTER(ID_SEEK_HEAD, TYPE_MASTER, struct segment, seek_head),
- E_SINGLE_MASTER(ID_INFO, TYPE_MASTER, struct segment, info),
- E_MASTER(ID_CLUSTER, TYPE_MASTER, struct segment, cluster),
- E_SINGLE_MASTER(ID_TRACKS, TYPE_MASTER, struct segment, tracks),
- E_SINGLE_MASTER(ID_CUES, TYPE_MASTER, struct segment, cues),
- E_LAST
-};
-
-static struct ebml_element_desc ne_top_level_elements[] = {
- E_SINGLE_MASTER(ID_EBML, TYPE_MASTER, nestegg, ebml),
- E_SINGLE_MASTER_O(ID_SEGMENT, TYPE_MASTER, nestegg, segment),
- E_LAST
-};
-
-#undef E_FIELD
-#undef E_MASTER
-#undef E_SINGLE_MASTER_O
-#undef E_SINGLE_MASTER
-#undef E_SUSPEND
-#undef E_LAST
-
-static struct pool_ctx *
-ne_pool_init(void)
-{
- struct pool_ctx * pool;
-
- pool = h_malloc(sizeof(*pool));
- if (!pool)
- abort();
- return pool;
-}
-
-static void
-ne_pool_destroy(struct pool_ctx * pool)
-{
- h_free(pool);
-}
-
-static void *
-ne_pool_alloc(size_t size, struct pool_ctx * pool)
-{
- void * p;
-
- p = h_malloc(size);
- if (!p)
- abort();
- hattach(p, pool);
- memset(p, 0, size);
- return p;
-}
-
-static void *
-ne_alloc(size_t size)
-{
- void * p;
-
- p = calloc(1, size);
- if (!p)
- abort();
- return p;
-}
-
-static int
-ne_io_read(nestegg_io * io, void * buffer, size_t length)
-{
- return io->read(buffer, length, io->userdata);
-}
-
-static int
-ne_io_seek(nestegg_io * io, int64_t offset, int whence)
-{
- return io->seek(offset, whence, io->userdata);
-}
-
-static int
-ne_io_read_skip(nestegg_io * io, size_t length)
-{
- size_t get;
- unsigned char buf[8192];
- int r = 1;
-
- while (length > 0) {
- get = length < sizeof(buf) ? length : sizeof(buf);
- r = ne_io_read(io, buf, get);
- if (r != 1)
- break;
- length -= get;
- }
-
- return r;
-}
-
-static int64_t
-ne_io_tell(nestegg_io * io)
-{
- return io->tell(io->userdata);
-}
-
-static int
-ne_bare_read_vint(nestegg_io * io, uint64_t * value, uint64_t * length, enum vint_mask maskflag)
-{
- int r;
- unsigned char b;
- size_t maxlen = 8;
- unsigned int count = 1, mask = 1 << 7;
-
- r = ne_io_read(io, &b, 1);
- if (r != 1)
- return r;
-
- while (count < maxlen) {
- if ((b & mask) != 0)
- break;
- mask >>= 1;
- count += 1;
- }
-
- if (length)
- *length = count;
- *value = b;
-
- if (maskflag == MASK_FIRST_BIT)
- *value = b & ~mask;
-
- while (--count) {
- r = ne_io_read(io, &b, 1);
- if (r != 1)
- return r;
- *value <<= 8;
- *value |= b;
- }
-
- return 1;
-}
-
-static int
-ne_read_id(nestegg_io * io, uint64_t * value, uint64_t * length)
-{
- return ne_bare_read_vint(io, value, length, MASK_NONE);
-}
-
-static int
-ne_read_vint(nestegg_io * io, uint64_t * value, uint64_t * length)
-{
- return ne_bare_read_vint(io, value, length, MASK_FIRST_BIT);
-}
-
-static int
-ne_read_svint(nestegg_io * io, int64_t * value, uint64_t * length)
-{
- int r;
- uint64_t uvalue;
- uint64_t ulength;
- int64_t svint_subtr[] = {
- 0x3f, 0x1fff,
- 0xfffff, 0x7ffffff,
- 0x3ffffffffLL, 0x1ffffffffffLL,
- 0xffffffffffffLL, 0x7fffffffffffffLL
- };
-
- r = ne_bare_read_vint(io, &uvalue, &ulength, MASK_FIRST_BIT);
- if (r != 1)
- return r;
- *value = uvalue - svint_subtr[ulength - 1];
- if (length)
- *length = ulength;
- return r;
-}
-
-static int
-ne_read_uint(nestegg_io * io, uint64_t * val, uint64_t length)
-{
- unsigned char b;
- int r;
-
- if (length == 0 || length > 8)
- return -1;
- r = ne_io_read(io, &b, 1);
- if (r != 1)
- return r;
- *val = b;
- while (--length) {
- r = ne_io_read(io, &b, 1);
- if (r != 1)
- return r;
- *val <<= 8;
- *val |= b;
- }
- return 1;
-}
-
-static int
-ne_read_int(nestegg_io * io, int64_t * val, uint64_t length)
-{
- int r;
- uint64_t uval, base;
-
- r = ne_read_uint(io, &uval, length);
- if (r != 1)
- return r;
-
- if (length < sizeof(int64_t)) {
- base = 1;
- base <<= length * 8 - 1;
- if (uval >= base) {
- base = 1;
- base <<= length * 8;
- } else {
- base = 0;
- }
- *val = uval - base;
- } else {
- *val = (int64_t) uval;
- }
-
- return 1;
-}
-
-static int
-ne_read_float(nestegg_io * io, double * val, uint64_t length)
-{
- union {
- uint64_t u;
- float f;
- double d;
- } value;
- int r;
-
- /* length == 10 not implemented */
- if (length != 4 && length != 8)
- return -1;
- r = ne_read_uint(io, &value.u, length);
- if (r != 1)
- return r;
- if (length == 4)
- *val = value.f;
- else
- *val = value.d;
- return 1;
-}
-
-static int
-ne_read_string(nestegg * ctx, char ** val, uint64_t length)
-{
- char * str;
- int r;
-
- if (length == 0 || length > LIMIT_STRING)
- return -1;
- str = ne_pool_alloc(length + 1, ctx->alloc_pool);
- r = ne_io_read(ctx->io, (unsigned char *) str, length);
- if (r != 1)
- return r;
- str[length] = '\0';
- *val = str;
- return 1;
-}
-
-static int
-ne_read_binary(nestegg * ctx, struct ebml_binary * val, uint64_t length)
-{
- if (length == 0 || length > LIMIT_BINARY)
- return -1;
- val->data = ne_pool_alloc(length, ctx->alloc_pool);
- val->length = length;
- return ne_io_read(ctx->io, val->data, length);
-}
-
-static int
-ne_get_uint(struct ebml_type type, uint64_t * value)
-{
- if (!type.read)
- return -1;
-
- assert(type.type == TYPE_UINT);
-
- *value = type.v.u;
-
- return 0;
-}
-
-static int
-ne_get_float(struct ebml_type type, double * value)
-{
- if (!type.read)
- return -1;
-
- assert(type.type == TYPE_FLOAT);
-
- *value = type.v.f;
-
- return 0;
-}
-
-static int
-ne_get_string(struct ebml_type type, char ** value)
-{
- if (!type.read)
- return -1;
-
- assert(type.type == TYPE_STRING);
-
- *value = type.v.s;
-
- return 0;
-}
-
-static int
-ne_get_binary(struct ebml_type type, struct ebml_binary * value)
-{
- if (!type.read)
- return -1;
-
- assert(type.type == TYPE_BINARY);
-
- *value = type.v.b;
-
- return 0;
-}
-
-static int
-ne_is_ancestor_element(uint64_t id, struct list_node * ancestor)
-{
- struct ebml_element_desc * element;
-
- for (; ancestor; ancestor = ancestor->previous)
- for (element = ancestor->node; element->id; ++element)
- if (element->id == id)
- return 1;
-
- return 0;
-}
-
-static struct ebml_element_desc *
-ne_find_element(uint64_t id, struct ebml_element_desc * elements)
-{
- struct ebml_element_desc * element;
-
- for (element = elements; element->id; ++element)
- if (element->id == id)
- return element;
-
- return NULL;
-}
-
-static void
-ne_ctx_push(nestegg * ctx, struct ebml_element_desc * ancestor, void * data)
-{
- struct list_node * item;
-
- item = ne_alloc(sizeof(*item));
- item->previous = ctx->ancestor;
- item->node = ancestor;
- item->data = data;
- ctx->ancestor = item;
-}
-
-static void
-ne_ctx_pop(nestegg * ctx)
-{
- struct list_node * item;
-
- item = ctx->ancestor;
- ctx->ancestor = item->previous;
- free(item);
-}
-
-static int
-ne_ctx_save(nestegg * ctx, struct saved_state * s)
-{
- s->stream_offset = ne_io_tell(ctx->io);
- if (s->stream_offset < 0)
- return -1;
- s->ancestor = ctx->ancestor;
- s->last_id = ctx->last_id;
- s->last_size = ctx->last_size;
- return 0;
-}
-
-static int
-ne_ctx_restore(nestegg * ctx, struct saved_state * s)
-{
- int r;
-
- r = ne_io_seek(ctx->io, s->stream_offset, NESTEGG_SEEK_SET);
- if (r != 0)
- return -1;
- ctx->ancestor = s->ancestor;
- ctx->last_id = s->last_id;
- ctx->last_size = s->last_size;
- return 0;
-}
-
-static int
-ne_peek_element(nestegg * ctx, uint64_t * id, uint64_t * size)
-{
- int r;
-
- if (ctx->last_id && ctx->last_size) {
- if (id)
- *id = ctx->last_id;
- if (size)
- *size = ctx->last_size;
- return 1;
- }
-
- r = ne_read_id(ctx->io, &ctx->last_id, NULL);
- if (r != 1)
- return r;
-
- r = ne_read_vint(ctx->io, &ctx->last_size, NULL);
- if (r != 1)
- return r;
-
- if (id)
- *id = ctx->last_id;
- if (size)
- *size = ctx->last_size;
-
- return 1;
-}
-
-static int
-ne_read_element(nestegg * ctx, uint64_t * id, uint64_t * size)
-{
- int r;
-
- r = ne_peek_element(ctx, id, size);
- if (r != 1)
- return r;
-
- ctx->last_id = 0;
- ctx->last_size = 0;
-
- return 1;
-}
-
-static void
-ne_read_master(nestegg * ctx, struct ebml_element_desc * desc)
-{
- struct ebml_list * list;
- struct ebml_list_node * node, * oldtail;
-
- assert(desc->type == TYPE_MASTER && desc->flags & DESC_FLAG_MULTI);
-
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "multi master element %llx (%s)",
- desc->id, desc->name);
-
- list = (struct ebml_list *) (ctx->ancestor->data + desc->offset);
-
- node = ne_pool_alloc(sizeof(*node), ctx->alloc_pool);
- node->id = desc->id;
- node->data = ne_pool_alloc(desc->size, ctx->alloc_pool);
-
- oldtail = list->tail;
- if (oldtail)
- oldtail->next = node;
- list->tail = node;
- if (!list->head)
- list->head = node;
-
- ctx->log(ctx, NESTEGG_LOG_DEBUG, " -> using data %p", node->data);
-
- ne_ctx_push(ctx, desc->children, node->data);
-}
-
-static void
-ne_read_single_master(nestegg * ctx, struct ebml_element_desc * desc)
-{
- assert(desc->type == TYPE_MASTER && !(desc->flags & DESC_FLAG_MULTI));
-
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "single master element %llx (%s)",
- desc->id, desc->name);
- ctx->log(ctx, NESTEGG_LOG_DEBUG, " -> using data %p (%u)",
- ctx->ancestor->data + desc->offset, desc->offset);
-
- ne_ctx_push(ctx, desc->children, ctx->ancestor->data + desc->offset);
-}
-
-static int
-ne_read_simple(nestegg * ctx, struct ebml_element_desc * desc, size_t length)
-{
- struct ebml_type * storage;
- int r;
-
- storage = (struct ebml_type *) (ctx->ancestor->data + desc->offset);
-
- if (storage->read) {
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "element %llx (%s) already read, skipping",
- desc->id, desc->name);
- return 0;
- }
-
- storage->type = desc->type;
-
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "element %llx (%s) -> %p (%u)",
- desc->id, desc->name, storage, desc->offset);
-
- r = -1;
-
- switch (desc->type) {
- case TYPE_UINT:
- r = ne_read_uint(ctx->io, &storage->v.u, length);
- break;
- case TYPE_FLOAT:
- r = ne_read_float(ctx->io, &storage->v.f, length);
- break;
- case TYPE_INT:
- r = ne_read_int(ctx->io, &storage->v.i, length);
- break;
- case TYPE_STRING:
- r = ne_read_string(ctx, &storage->v.s, length);
- break;
- case TYPE_BINARY:
- r = ne_read_binary(ctx, &storage->v.b, length);
- break;
- case TYPE_MASTER:
- case TYPE_UNKNOWN:
- assert(0);
- break;
- }
-
- if (r == 1)
- storage->read = 1;
-
- return r;
-}
-
-static int
-ne_parse(nestegg * ctx, struct ebml_element_desc * top_level)
-{
- int r;
- int64_t * data_offset;
- uint64_t id, size;
- struct ebml_element_desc * element;
-
- /* loop until we need to return:
- - hit suspend point
- - parse complete
- - error occurred */
-
- /* loop over elements at current level reading them if sublevel found,
- push ctx onto stack and continue if sublevel ended, pop ctx off stack
- and continue */
-
- if (!ctx->ancestor)
- return -1;
-
- for (;;) {
- r = ne_peek_element(ctx, &id, &size);
- if (r != 1)
- break;
-
- element = ne_find_element(id, ctx->ancestor->node);
- if (element) {
- if (element->flags & DESC_FLAG_SUSPEND) {
- assert(element->type == TYPE_BINARY);
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "suspend parse at %llx", id);
- r = 1;
- break;
- }
-
- r = ne_read_element(ctx, &id, &size);
- if (r != 1)
- break;
-
- if (element->flags & DESC_FLAG_OFFSET) {
- data_offset = (int64_t *) (ctx->ancestor->data + element->data_offset);
- *data_offset = ne_io_tell(ctx->io);
- if (*data_offset < 0) {
- r = -1;
- break;
- }
- }
-
- if (element->type == TYPE_MASTER) {
- if (element->flags & DESC_FLAG_MULTI)
- ne_read_master(ctx, element);
- else
- ne_read_single_master(ctx, element);
- continue;
- } else {
- r = ne_read_simple(ctx, element, size);
- if (r < 0)
- break;
- }
- } else if (ne_is_ancestor_element(id, ctx->ancestor->previous)) {
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "parent element %llx", id);
- if (top_level && ctx->ancestor->node == top_level) {
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "*** parse about to back up past top_level");
- r = 1;
- break;
- }
- ne_ctx_pop(ctx);
- } else {
- r = ne_read_element(ctx, &id, &size);
- if (r != 1)
- break;
-
- if (id != ID_VOID && id != ID_CRC32)
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "unknown element %llx", id);
- r = ne_io_read_skip(ctx->io, size);
- if (r != 1)
- break;
- }
- }
-
- if (r != 1)
- while (ctx->ancestor)
- ne_ctx_pop(ctx);
-
- return r;
-}
-
-static uint64_t
-ne_xiph_lace_value(unsigned char ** np)
-{
- uint64_t lace;
- uint64_t value;
- unsigned char * p = *np;
-
- lace = *p++;
- value = lace;
- while (lace == 255) {
- lace = *p++;
- value += lace;
- }
-
- *np = p;
-
- return value;
-}
-
-static int
-ne_read_xiph_lace_value(nestegg_io * io, uint64_t * value, size_t * consumed)
-{
- int r;
- uint64_t lace;
-
- r = ne_read_uint(io, &lace, 1);
- if (r != 1)
- return r;
- *consumed += 1;
-
- *value = lace;
- while (lace == 255) {
- r = ne_read_uint(io, &lace, 1);
- if (r != 1)
- return r;
- *consumed += 1;
- *value += lace;
- }
-
- return 1;
-}
-
-static int
-ne_read_xiph_lacing(nestegg_io * io, size_t block, size_t * read, uint64_t n, uint64_t * sizes)
-{
- int r;
- size_t i = 0;
- uint64_t sum = 0;
-
- while (--n) {
- r = ne_read_xiph_lace_value(io, &sizes[i], read);
- if (r != 1)
- return r;
- sum += sizes[i];
- i += 1;
- }
-
- if (*read + sum > block)
- return -1;
-
- /* last frame is the remainder of the block */
- sizes[i] = block - *read - sum;
- return 1;
-}
-
-static int
-ne_read_ebml_lacing(nestegg_io * io, size_t block, size_t * read, uint64_t n, uint64_t * sizes)
-{
- int r;
- uint64_t lace, sum, length;
- int64_t slace;
- size_t i = 0;
-
- r = ne_read_vint(io, &lace, &length);
- if (r != 1)
- return r;
- *read += length;
-
- sizes[i] = lace;
- sum = sizes[i];
-
- i += 1;
- n -= 1;
-
- while (--n) {
- r = ne_read_svint(io, &slace, &length);
- if (r != 1)
- return r;
- *read += length;
- sizes[i] = sizes[i - 1] + slace;
- sum += sizes[i];
- i += 1;
- }
-
- if (*read + sum > block)
- return -1;
-
- /* last frame is the remainder of the block */
- sizes[i] = block - *read - sum;
- return 1;
-}
-
-static uint64_t
-ne_get_timecode_scale(nestegg * ctx)
-{
- uint64_t scale;
-
- if (ne_get_uint(ctx->segment.info.timecode_scale, &scale) != 0)
- scale = 1000000;
-
- return scale;
-}
-
-static struct track_entry *
-ne_find_track_entry(nestegg * ctx, unsigned int track)
-{
- struct ebml_list_node * node;
- unsigned int tracks = 0;
-
- node = ctx->segment.tracks.track_entry.head;
- while (node) {
- assert(node->id == ID_TRACK_ENTRY);
- if (track == tracks)
- return node->data;
- tracks += 1;
- node = node->next;
- }
-
- return NULL;
-}
-
-static int
-ne_read_block(nestegg * ctx, uint64_t block_id, uint64_t block_size, nestegg_packet ** data)
-{
- int r;
- int64_t timecode, abs_timecode;
- nestegg_packet * pkt;
- struct cluster * cluster;
- struct frame * f, * last;
- struct track_entry * entry;
- double track_scale;
- uint64_t track, length, frame_sizes[256], cluster_tc, flags, frames, tc_scale, total;
- unsigned int i, lacing;
- size_t consumed = 0;
-
- *data = NULL;
-
- if (block_size > LIMIT_BLOCK)
- return -1;
-
- r = ne_read_vint(ctx->io, &track, &length);
- if (r != 1)
- return r;
-
- if (track == 0 || track > ctx->track_count)
- return -1;
-
- consumed += length;
-
- r = ne_read_int(ctx->io, &timecode, 2);
- if (r != 1)
- return r;
-
- consumed += 2;
-
- r = ne_read_uint(ctx->io, &flags, 1);
- if (r != 1)
- return r;
-
- consumed += 1;
-
- frames = 0;
-
- /* flags are different between block and simpleblock, but lacing is
- encoded the same way */
- lacing = (flags & BLOCK_FLAGS_LACING) >> 1;
-
- switch (lacing) {
- case LACING_NONE:
- frames = 1;
- break;
- case LACING_XIPH:
- case LACING_FIXED:
- case LACING_EBML:
- r = ne_read_uint(ctx->io, &frames, 1);
- if (r != 1)
- return r;
- consumed += 1;
- frames += 1;
- }
-
- if (frames > 256)
- return -1;
-
- switch (lacing) {
- case LACING_NONE:
- frame_sizes[0] = block_size - consumed;
- break;
- case LACING_XIPH:
- if (frames == 1)
- return -1;
- r = ne_read_xiph_lacing(ctx->io, block_size, &consumed, frames, frame_sizes);
- if (r != 1)
- return r;
- break;
- case LACING_FIXED:
- if ((block_size - consumed) % frames)
- return -1;
- for (i = 0; i < frames; ++i)
- frame_sizes[i] = (block_size - consumed) / frames;
- break;
- case LACING_EBML:
- if (frames == 1)
- return -1;
- r = ne_read_ebml_lacing(ctx->io, block_size, &consumed, frames, frame_sizes);
- if (r != 1)
- return r;
- break;
- }
-
- /* sanity check unlaced frame sizes against total block size. */
- total = consumed;
- for (i = 0; i < frames; ++i)
- total += frame_sizes[i];
- if (total > block_size)
- return -1;
-
- entry = ne_find_track_entry(ctx, track - 1);
- if (!entry)
- return -1;
-
- track_scale = 1.0;
-
- tc_scale = ne_get_timecode_scale(ctx);
-
- assert(ctx->segment.cluster.tail->id == ID_CLUSTER);
- cluster = ctx->segment.cluster.tail->data;
- if (ne_get_uint(cluster->timecode, &cluster_tc) != 0)
- return -1;
-
- abs_timecode = timecode + cluster_tc;
- if (abs_timecode < 0)
- return -1;
-
- pkt = ne_alloc(sizeof(*pkt));
- pkt->track = track - 1;
- pkt->timecode = abs_timecode * tc_scale * track_scale;
-
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "%sblock t %lld pts %f f %llx frames: %llu",
- block_id == ID_BLOCK ? "" : "simple", pkt->track, pkt->timecode / 1e9, flags, frames);
-
- last = NULL;
- for (i = 0; i < frames; ++i) {
- if (frame_sizes[i] > LIMIT_FRAME) {
- nestegg_free_packet(pkt);
- return -1;
- }
- f = ne_alloc(sizeof(*f));
- f->data = ne_alloc(frame_sizes[i]);
- f->length = frame_sizes[i];
- r = ne_io_read(ctx->io, f->data, frame_sizes[i]);
- if (r != 1) {
- free(f->data);
- free(f);
- nestegg_free_packet(pkt);
- return -1;
- }
-
- if (!last)
- pkt->frame = f;
- else
- last->next = f;
- last = f;
- }
-
- *data = pkt;
-
- return 1;
-}
-
-static uint64_t
-ne_buf_read_id(unsigned char const * p, size_t length)
-{
- uint64_t id = 0;
-
- while (length--) {
- id <<= 8;
- id |= *p++;
- }
-
- return id;
-}
-
-static struct seek *
-ne_find_seek_for_id(struct ebml_list_node * seek_head, uint64_t id)
-{
- struct ebml_list * head;
- struct ebml_list_node * seek;
- struct ebml_binary binary_id;
- struct seek * s;
-
- while (seek_head) {
- assert(seek_head->id == ID_SEEK_HEAD);
- head = seek_head->data;
- seek = head->head;
-
- while (seek) {
- assert(seek->id == ID_SEEK);
- s = seek->data;
-
- if (ne_get_binary(s->id, &binary_id) == 0 &&
- ne_buf_read_id(binary_id.data, binary_id.length) == id)
- return s;
-
- seek = seek->next;
- }
-
- seek_head = seek_head->next;
- }
-
- return NULL;
-}
-
-static struct cue_point *
-ne_find_cue_point_for_tstamp(struct ebml_list_node * cue_point, uint64_t scale, uint64_t tstamp)
-{
- uint64_t time;
- struct cue_point * c, * prev = NULL;
-
- while (cue_point) {
- assert(cue_point->id == ID_CUE_POINT);
- c = cue_point->data;
-
- if (!prev)
- prev = c;
-
- if (ne_get_uint(c->time, &time) == 0 && time * scale > tstamp)
- break;
-
- prev = cue_point->data;
- cue_point = cue_point->next;
- }
-
- return prev;
-}
-
-static int
-ne_is_suspend_element(uint64_t id)
-{
- /* this could search the tree of elements for DESC_FLAG_SUSPEND */
- if (id == ID_SIMPLE_BLOCK || id == ID_BLOCK)
- return 1;
- return 0;
-}
-
-static void
-ne_null_log_callback(nestegg * ctx, unsigned int severity, char const * fmt, ...)
-{
- if (ctx && severity && fmt)
- return;
-}
-
-int
-nestegg_init(nestegg ** context, nestegg_io io, nestegg_log callback)
-{
- int r;
- uint64_t id, version, docversion;
- struct ebml_list_node * track;
- char * doctype;
- nestegg * ctx = NULL;
-
- if (!(io.read && io.seek && io.tell))
- return -1;
-
- ctx = ne_alloc(sizeof(*ctx));
-
- ctx->io = ne_alloc(sizeof(*ctx->io));
- *ctx->io = io;
- ctx->log = callback;
- ctx->alloc_pool = ne_pool_init();
-
- if (!ctx->log)
- ctx->log = ne_null_log_callback;
-
- r = ne_peek_element(ctx, &id, NULL);
- if (r != 1) {
- nestegg_destroy(ctx);
- return -1;
- }
-
- if (id != ID_EBML) {
- nestegg_destroy(ctx);
- return -1;
- }
-
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "ctx %p", ctx);
-
- ne_ctx_push(ctx, ne_top_level_elements, ctx);
-
- r = ne_parse(ctx, NULL);
-
- if (r != 1) {
- nestegg_destroy(ctx);
- return -1;
- }
-
- if (ne_get_uint(ctx->ebml.ebml_read_version, &version) != 0)
- version = 1;
- if (version != 1) {
- nestegg_destroy(ctx);
- return -1;
- }
-
- if (ne_get_string(ctx->ebml.doctype, &doctype) != 0)
- doctype = "matroska";
- if (strcmp(doctype, "webm") != 0) {
- nestegg_destroy(ctx);
- return -1;
- }
-
- if (ne_get_uint(ctx->ebml.doctype_read_version, &docversion) != 0)
- docversion = 1;
- if (docversion < 1 || docversion > 2) {
- nestegg_destroy(ctx);
- return -1;
- }
-
- if (!ctx->segment.tracks.track_entry.head) {
- nestegg_destroy(ctx);
- return -1;
- }
-
- track = ctx->segment.tracks.track_entry.head;
- ctx->track_count = 0;
-
- while (track) {
- ctx->track_count += 1;
- track = track->next;
- }
-
- *context = ctx;
-
- return 0;
-}
-
-void
-nestegg_destroy(nestegg * ctx)
-{
- while (ctx->ancestor)
- ne_ctx_pop(ctx);
- ne_pool_destroy(ctx->alloc_pool);
- free(ctx->io);
- free(ctx);
-}
-
-int
-nestegg_duration(nestegg * ctx, uint64_t * duration)
-{
- uint64_t tc_scale;
- double unscaled_duration;
-
- if (ne_get_float(ctx->segment.info.duration, &unscaled_duration) != 0)
- return -1;
-
- tc_scale = ne_get_timecode_scale(ctx);
-
- *duration = (uint64_t) (unscaled_duration * tc_scale);
- return 0;
-}
-
-int
-nestegg_tstamp_scale(nestegg * ctx, uint64_t * scale)
-{
- *scale = ne_get_timecode_scale(ctx);
- return 0;
-}
-
-int
-nestegg_track_count(nestegg * ctx, unsigned int * tracks)
-{
- *tracks = ctx->track_count;
- return 0;
-}
-
-int
-nestegg_track_seek(nestegg * ctx, unsigned int track, uint64_t tstamp)
-{
- int r;
- struct cue_point * cue_point;
- struct cue_track_positions * pos;
- struct saved_state state;
- struct seek * found;
- uint64_t seek_pos, tc_scale, t, id;
- struct ebml_list_node * node = ctx->segment.cues.cue_point.head;
-
- /* If there are no cues loaded, check for cues element in the seek head
- and load it. */
- if (!node) {
- found = ne_find_seek_for_id(ctx->segment.seek_head.head, ID_CUES);
- if (!found)
- return -1;
-
- if (ne_get_uint(found->position, &seek_pos) != 0)
- return -1;
-
- /* Save old parser state. */
- r = ne_ctx_save(ctx, &state);
- if (r != 0)
- return -1;
-
- /* Seek and set up parser state for segment-level element (Cues). */
- r = ne_io_seek(ctx->io, ctx->segment_offset + seek_pos, NESTEGG_SEEK_SET);
- if (r != 0)
- return -1;
- ctx->last_id = 0;
- ctx->last_size = 0;
-
- r = ne_read_element(ctx, &id, NULL);
- if (r != 1)
- return -1;
-
- if (id != ID_CUES)
- return -1;
-
- ctx->ancestor = NULL;
- ne_ctx_push(ctx, ne_top_level_elements, ctx);
- ne_ctx_push(ctx, ne_segment_elements, &ctx->segment);
- ne_ctx_push(ctx, ne_cues_elements, &ctx->segment.cues);
- /* parser will run until end of cues element. */
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "seek: parsing cue elements");
- r = ne_parse(ctx, ne_cues_elements);
- while (ctx->ancestor)
- ne_ctx_pop(ctx);
-
- /* Reset parser state to original state and seek back to old position. */
- if (ne_ctx_restore(ctx, &state) != 0)
- return -1;
-
- if (r < 0)
- return -1;
- }
-
- tc_scale = ne_get_timecode_scale(ctx);
-
- cue_point = ne_find_cue_point_for_tstamp(ctx->segment.cues.cue_point.head, tc_scale, tstamp);
- if (!cue_point)
- return -1;
-
- node = cue_point->cue_track_positions.head;
-
- seek_pos = 0;
-
- while (node) {
- assert(node->id == ID_CUE_TRACK_POSITIONS);
- pos = node->data;
- if (ne_get_uint(pos->track, &t) == 0 && t - 1 == track) {
- if (ne_get_uint(pos->cluster_position, &seek_pos) != 0)
- return -1;
- break;
- }
- node = node->next;
- }
-
- /* Seek and set up parser state for segment-level element (Cluster). */
- r = ne_io_seek(ctx->io, ctx->segment_offset + seek_pos, NESTEGG_SEEK_SET);
- if (r != 0)
- return -1;
- ctx->last_id = 0;
- ctx->last_size = 0;
-
- while (ctx->ancestor)
- ne_ctx_pop(ctx);
-
- ne_ctx_push(ctx, ne_top_level_elements, ctx);
- ne_ctx_push(ctx, ne_segment_elements, &ctx->segment);
- ctx->log(ctx, NESTEGG_LOG_DEBUG, "seek: parsing cluster elements");
- r = ne_parse(ctx, NULL);
- if (r != 1)
- return -1;
-
- if (!ne_is_suspend_element(ctx->last_id))
- return -1;
-
- return 0;
-}
-
-int
-nestegg_track_type(nestegg * ctx, unsigned int track)
-{
- struct track_entry * entry;
- uint64_t type;
-
- entry = ne_find_track_entry(ctx, track);
- if (!entry)
- return -1;
-
- if (ne_get_uint(entry->type, &type) != 0)
- return -1;
-
- if (type & TRACK_TYPE_VIDEO)
- return NESTEGG_TRACK_VIDEO;
-
- if (type & TRACK_TYPE_AUDIO)
- return NESTEGG_TRACK_AUDIO;
-
- return -1;
-}
-
-int
-nestegg_track_codec_id(nestegg * ctx, unsigned int track)
-{
- char * codec_id;
- struct track_entry * entry;
-
- entry = ne_find_track_entry(ctx, track);
- if (!entry)
- return -1;
-
- if (ne_get_string(entry->codec_id, &codec_id) != 0)
- return -1;
-
- if (strcmp(codec_id, TRACK_ID_VP8) == 0)
- return NESTEGG_CODEC_VP8;
-
- if (strcmp(codec_id, TRACK_ID_VORBIS) == 0)
- return NESTEGG_CODEC_VORBIS;
-
- return -1;
-}
-
-int
-nestegg_track_codec_data_count(nestegg * ctx, unsigned int track,
- unsigned int * count)
-{
- struct track_entry * entry;
- struct ebml_binary codec_private;
- unsigned char * p;
-
- *count = 0;
-
- entry = ne_find_track_entry(ctx, track);
- if (!entry)
- return -1;
-
- if (nestegg_track_codec_id(ctx, track) != NESTEGG_CODEC_VORBIS)
- return -1;
-
- if (ne_get_binary(entry->codec_private, &codec_private) != 0)
- return -1;
-
- if (codec_private.length < 1)
- return -1;
-
- p = codec_private.data;
- *count = *p + 1;
-
- if (*count > 3)
- return -1;
-
- return 0;
-}
-
-int
-nestegg_track_codec_data(nestegg * ctx, unsigned int track, unsigned int item,
- unsigned char ** data, size_t * length)
-{
- struct track_entry * entry;
- struct ebml_binary codec_private;
- uint64_t sizes[3], total;
- unsigned char * p;
- unsigned int count, i;
-
- *data = NULL;
- *length = 0;
-
- entry = ne_find_track_entry(ctx, track);
- if (!entry)
- return -1;
-
- if (nestegg_track_codec_id(ctx, track) != NESTEGG_CODEC_VORBIS)
- return -1;
-
- if (ne_get_binary(entry->codec_private, &codec_private) != 0)
- return -1;
-
- p = codec_private.data;
- count = *p++ + 1;
-
- if (count > 3)
- return -1;
-
- i = 0;
- total = 0;
- while (--count) {
- sizes[i] = ne_xiph_lace_value(&p);
- total += sizes[i];
- i += 1;
- }
- sizes[i] = codec_private.length - total - (p - codec_private.data);
-
- for (i = 0; i < item; ++i) {
- if (sizes[i] > LIMIT_FRAME)
- return -1;
- p += sizes[i];
- }
- *data = p;
- *length = sizes[item];
-
- return 0;
-}
-
-int
-nestegg_track_video_params(nestegg * ctx, unsigned int track,
- nestegg_video_params * params)
-{
- struct track_entry * entry;
- uint64_t value;
-
- memset(params, 0, sizeof(*params));
-
- entry = ne_find_track_entry(ctx, track);
- if (!entry)
- return -1;
-
- if (nestegg_track_type(ctx, track) != NESTEGG_TRACK_VIDEO)
- return -1;
-
- if (ne_get_uint(entry->video.pixel_width, &value) != 0)
- return -1;
- params->width = value;
-
- if (ne_get_uint(entry->video.pixel_height, &value) != 0)
- return -1;
- params->height = value;
-
- value = 0;
- ne_get_uint(entry->video.pixel_crop_bottom, &value);
- params->crop_bottom = value;
-
- value = 0;
- ne_get_uint(entry->video.pixel_crop_top, &value);
- params->crop_top = value;
-
- value = 0;
- ne_get_uint(entry->video.pixel_crop_left, &value);
- params->crop_left = value;
-
- value = 0;
- ne_get_uint(entry->video.pixel_crop_right, &value);
- params->crop_right = value;
-
- value = params->width;
- ne_get_uint(entry->video.display_width, &value);
- params->display_width = value;
-
- value = params->height;
- ne_get_uint(entry->video.display_height, &value);
- params->display_height = value;
-
- return 0;
-}
-
-int
-nestegg_track_audio_params(nestegg * ctx, unsigned int track,
- nestegg_audio_params * params)
-{
- struct track_entry * entry;
- uint64_t value;
-
- memset(params, 0, sizeof(*params));
-
- entry = ne_find_track_entry(ctx, track);
- if (!entry)
- return -1;
-
- if (nestegg_track_type(ctx, track) != NESTEGG_TRACK_AUDIO)
- return -1;
-
- params->rate = 8000;
- ne_get_float(entry->audio.sampling_frequency, ¶ms->rate);
-
- value = 1;
- ne_get_uint(entry->audio.channels, &value);
- params->channels = value;
-
- value = 16;
- ne_get_uint(entry->audio.bit_depth, &value);
- params->depth = value;
-
- return 0;
-}
-
-int
-nestegg_read_packet(nestegg * ctx, nestegg_packet ** pkt)
-{
- int r;
- uint64_t id, size;
-
- *pkt = NULL;
-
- for (;;) {
- r = ne_peek_element(ctx, &id, &size);
- if (r != 1)
- return r;
-
- /* any suspend fields must be handled here */
- if (ne_is_suspend_element(id)) {
- r = ne_read_element(ctx, &id, &size);
- if (r != 1)
- return r;
-
- /* the only suspend fields are blocks and simple blocks, which we
- handle directly. */
- r = ne_read_block(ctx, id, size, pkt);
- return r;
- }
-
- r = ne_parse(ctx, NULL);
- if (r != 1)
- return r;
- }
-
- return 1;
-}
-
-void
-nestegg_free_packet(nestegg_packet * pkt)
-{
- struct frame * frame;
-
- while (pkt->frame) {
- frame = pkt->frame;
- pkt->frame = frame->next;
- free(frame->data);
- free(frame);
- }
-
- free(pkt);
-}
-
-int
-nestegg_packet_track(nestegg_packet * pkt, unsigned int * track)
-{
- *track = pkt->track;
- return 0;
-}
-
-int
-nestegg_packet_tstamp(nestegg_packet * pkt, uint64_t * tstamp)
-{
- *tstamp = pkt->timecode;
- return 0;
-}
-
-int
-nestegg_packet_count(nestegg_packet * pkt, unsigned int * count)
-{
- struct frame * f = pkt->frame;
-
- *count = 0;
-
- while (f) {
- *count += 1;
- f = f->next;
- }
-
- return 0;
-}
-
-int
-nestegg_packet_data(nestegg_packet * pkt, unsigned int item,
- unsigned char ** data, size_t * length)
-{
- struct frame * f = pkt->frame;
- unsigned int count = 0;
-
- *data = NULL;
- *length = 0;
-
- while (f) {
- if (count == item) {
- *data = f->data;
- *length = f->length;
- return 0;
- }
- count += 1;
- f = f->next;
- }
-
- return -1;
-}
diff --git a/solution.mk b/solution.mk
deleted file mode 100644
index 782150f..0000000
--- a/solution.mk
+++ /dev/null
@@ -1,29 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-vpx.sln: $(wildcard *.vcproj)
- @echo " [CREATE] $@"
- $(SRC_PATH_BARE)/build/make/gen_msvs_sln.sh \
- $(if $(filter %vpx.vcproj,$^),\
- $(foreach vcp,$(filter-out %vpx.vcproj %obj_int_extract.vcproj,$^),\
- --dep=$(vcp:.vcproj=):vpx)) \
- --dep=vpx:obj_int_extract \
- --ver=$(CONFIG_VS_VERSION)\
- --out=$@ $^
-vpx.sln.mk: vpx.sln
- @true
-
-PROJECTS-yes += vpx.sln vpx.sln.mk
--include vpx.sln.mk
-
-# Always install this file, as it is an unconditional post-build rule.
-INSTALL_MAPS += src/% $(SRC_PATH_BARE)/%
-INSTALL-SRCS-yes += $(target).mk
diff --git a/tools_common.c b/tools_common.c
deleted file mode 100644
index d188bbe..0000000
--- a/tools_common.c
+++ /dev/null
@@ -1,24 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-#include <stdio.h>
-#include "tools_common.h"
-#ifdef _WIN32
-#include <io.h>
-#include <fcntl.h>
-#endif
-
-FILE* set_binary_mode(FILE *stream)
-{
- (void)stream;
-#ifdef _WIN32
- _setmode(_fileno(stream), _O_BINARY);
-#endif
- return stream;
-}
diff --git a/usage.dox b/usage.dox
deleted file mode 100644
index 0db080b..0000000
--- a/usage.dox
+++ /dev/null
@@ -1,203 +0,0 @@
-/*!\page usage Usage
-
- The vpx Multi-Format codec SDK provides a unified interface amongst its
- supported codecs. This abstraction allows applications using this SDK to
- easily support multiple video formats with minimal code duplication or
- "special casing." This section describes the interface common to all codecs.
- For codec-specific details, see the \ref codecs page.
-
- The following sections are common to all codecs:
- - \ref usage_types
- - \ref usage_features
- - \ref usage_init
- - \ref usage_errors
-
- Fore more information on decoder and encoder specific usage, see the
- following pages:
- \if decoder - \subpage usage_decode \endif
- \if decoder - \subpage usage_encode \endif
-
- \section usage_types Important Data Types
- There are two important data structures to consider in this interface.
-
- \subsection usage_ctxs Contexts
- A context is a storage area allocated by the calling application that the
- codec may write into to store details about a single instance of that codec.
- Most of the context is implementation specific, and thus opaque to the
- application. The context structure as seen by the application is of fixed
- size, and thus can be allocated with automatic storage or dynamically
- on the heap.
-
- Most operations require an initialized codec context. Codec context
- instances are codec specific. That is, the codec to be used for the encoded
- video must be known at initialization time. See #vpx_codec_ctx_t for further
- information.
-
- \subsection usage_ifaces Interfaces
- A codec interface is an opaque structure that controls how function calls
- into the generic interface are dispatched to their codec-specific
- implementations. Applications \ref MUSTNOT attempt to examine or override
- this storage, as it contains internal implementation details likely to
- change from release to release.
-
- Each supported codec will expose an interface structure to the application
- as an <code>extern</code> reference to a structure of the incomplete type
- #vpx_codec_iface_t.
-
- \section usage_features Features
- Several "features" are defined that are optionally implemented by codec
- algorithms. Indeed, the same algorithm may support different features on
- different platforms. The purpose of defining these features is that when
- they are implemented, they conform to a common interface. The features, or
- capabilities, of an algorithm can be queried from it's interface by using
- the vpx_codec_get_caps() method. Attempts to invoke features not supported
- by an algorithm will generally result in #VPX_CODEC_INCAPABLE.
-
- Currently defined features available in both encoders and decoders include:
- - \subpage usage_xma
-
- \if decoder
- Currently defined decoder features include:
- - \ref usage_cb
- - \ref usage_postproc
- \endif
-
- \section usage_init Initialization
- To initialize a codec instance, the address of the codec context
- and interface structures are passed to an initialization function. Depending
- on the \ref usage_features that the codec supports, the codec could be
- initialized in different modes. Most notably, the application may choose to
- use \ref usage_xma mode to gain fine grained control over how and where
- memory is allocated for the codec.
-
- To prevent cases of confusion where the ABI of the library changes,
- the ABI is versioned. The ABI version number must be passed at
- initialization time to ensure the application is using a header file that
- matches the library. The current ABI version number is stored in the
- preprocessor macros #VPX_CODEC_ABI_VERSION, #VPX_ENCODER_ABI_VERSION, and
- #VPX_DECODER_ABI_VERSION. For convenience, each initialization function has
- a wrapper macro that inserts the correct version number. These macros are
- named like the initialization methods, but without the _ver suffix.
-
-
- The available initialization methods are:
- \if encoder - #vpx_codec_enc_init (calls vpx_codec_enc_init_ver()) \endif
- \if decoder - #vpx_codec_dec_init (calls vpx_codec_dec_init_ver()) \endif
-
-
-
- \section usage_errors Error Handling
- Almost all codec functions return an error status of type #vpx_codec_err_t.
- The semantics of how each error condition should be processed is clearly
- defined in the definitions of each enumerated value. Error values can be
- converted into ASCII strings with the vpx_codec_error() and
- vpx_codec_err_to_string() methods. The difference between these two methods is
- that vpx_codec_error() returns the error state from an initialized context,
- whereas vpx_codec_err_to_string() can be used in cases where an error occurs
- outside any context. The enumerated value returned from the last call can be
- retrieved from the <code>err</code> member of the decoder context as well.
- Finally, more detailed error information may be able to be obtained by using
- the vpx_codec_error_detail() method. Not all errors produce detailed error
- information.
-
- In addition to error information, the codec library's build configuration
- is available at runtime on some platforms. This information can be returned
- by calling vpx_codec_build_config(), and is formatted as a base64 coded string
- (comprised of characters in the set [a-z_a-Z0-9+/]). This information is not
- useful to an application at runtime, but may be of use to vpx for support.
-
-
- \section usage_deadline Deadline
- Both the encoding and decoding functions have a <code>deadline</code>
- parameter. This parameter indicates the amount of time, in microseconds
- (us), that the application wants the codec to spend processing before
- returning. This is a soft deadline -- that is, the semantics of the
- requested operation take precedence over meeting the deadline. If, for
- example, an application sets a <code>deadline</code> of 1000us, and the
- frame takes 2000us to decode, the call to vpx_codec_decode() will return
- after 2000us. In this case the deadline is not met, but the semantics of the
- function are preserved. If, for the same frame, an application instead sets
- a <code>deadline</code> of 5000us, the decoder will see that it has 3000us
- remaining in its time slice when decoding completes. It could then choose to
- run a set of \ref usage_postproc filters, and perhaps would return after
- 4000us (instead of the allocated 5000us). In this case the deadline is met,
- and the semantics of the call are preserved, as before.
-
- The special value <code>0</code> is reserved to represent an infinite
- deadline. In this case, the codec will perform as much processing as
- possible to yield the highest quality frame.
-
- By convention, the value <code>1</code> is used to mean "return as fast as
- possible."
-
-*/
-
-
-/*! \page usage_xma External Memory Allocation
- Applications that wish to have fine grained control over how and where
- decoders allocate memory \ref MAY make use of the eXternal Memory Allocation
- (XMA) interface. Not all codecs support the XMA \ref usage_features.
-
- To use a decoder in XMA mode, the decoder \ref MUST be initialized with the
- vpx_codec_xma_init_ver() function. The amount of memory a decoder needs to
- allocate is heavily dependent on the size of the encoded video frames. The
- size of the video must be known before requesting the decoder's memory map.
- This stream information can be obtained with the vpx_codec_peek_stream_info()
- function, which does not require a constructed decoder context. If the exact
- stream is not known, a stream info structure can be created that reflects
- the maximum size that the decoder instance is required to support.
-
- Once the decoder instance has been initialized and the stream information
- determined, the application calls the vpx_codec_get_mem_map() iterator
- repeatedly to get a list of the memory segments requested by the decoder.
- The iterator value should be initialized to NULL to request the first
- element, and the function will return #VPX_CODEC_LIST_END to signal the end of
- the list.
-
- After each segment is identified, it must be passed to the codec through the
- vpx_codec_set_mem_map() function. Segments \ref MUST be passed in the same
- order as they are returned from vpx_codec_get_mem_map(), but there is no
- requirement that vpx_codec_get_mem_map() must finish iterating before
- vpx_codec_set_mem_map() is called. For instance, some applications may choose
- to get a list of all requests, construct an optimal heap, and then set all
- maps at once with one call. Other applications may set one map at a time,
- allocating it immediately after it is returned from vpx_codec_get_mem_map().
-
- After all segments have been set using vpx_codec_set_mem_map(), the codec may
- be used as it would be in normal internal allocation mode.
-
- \section usage_xma_seg_id Segment Identifiers
- Each requested segment is identified by an identifier unique to
- that decoder type. Some of these identifiers are private, while others are
- enumerated for application use. Identifiers not enumerated publicly are
- subject to change. Identifiers are non-consecutive.
-
- \section usage_xma_seg_szalign Segment Size and Alignment
- The sz (size) and align (alignment) parameters describe the required size
- and alignment of the requested segment. Alignment will always be a power of
- two. Applications \ref MUST honor the alignment requested. Failure to do so
- could result in program crashes or may incur a speed penalty.
-
- \section usage_xma_seg_flags Segment Flags
- The flags member of the segment structure indicates any requirements or
- desires of the codec for the particular segment. The #VPX_CODEC_MEM_ZERO flag
- indicates that the segment \ref MUST be zeroed by the application prior to
- passing it to the application. The #VPX_CODEC_MEM_WRONLY flag indicates that
- the segment will only be written into by the decoder, not read. If this flag
- is not set, the application \ref MUST insure that the memory segment is
- readable. On some platforms, framebuffer memory is writable but not
- readable, for example. The #VPX_CODEC_MEM_FAST flag indicates that the segment
- will be frequently accessed, and that it should be placed into fast memory,
- if any is available. The application \ref MAY choose to place other segments
- in fast memory as well, but the most critical segments will be identified by
- this flag.
-
- \section usage_xma_seg_basedtor Segment Base Address and Destructor
- For each requested memory segment, the application must determine the
- address of a memory segment that meets the requirements of the codec. This
- address is set in the <code>base</code> member of the #vpx_codec_mmap
- structure. If the application requires processing when the segment is no
- longer used by the codec (for instance to deallocate it or close an
- associated file descriptor) the <code>dtor</code> and <code>priv</code>
- members can be set.
-*/
diff --git a/usage_cx.dox b/usage_cx.dox
deleted file mode 100644
index 980a034..0000000
--- a/usage_cx.dox
+++ /dev/null
@@ -1,14 +0,0 @@
-/*! \page usage_encode Encode
-
- The vpx_codec_encode() function is at the core of the decode loop. It
- processes raw images passed by the application, producing packets of
- compressed data. The <code>deadline</code> parameter controls the amount
- of time in microseconds the encoder should spend working on the frame. For
- more information on the <code>deadline</code> parameter, see
- \ref usage_deadline.
-
-
- \ref samples
-
-
-*/
diff --git a/vp8/common/alloccommon.c b/vp8/common/alloccommon.c
deleted file mode 100644
index edef360..0000000
--- a/vp8/common/alloccommon.c
+++ /dev/null
@@ -1,200 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "blockd.h"
-#include "vpx_mem/vpx_mem.h"
-#include "onyxc_int.h"
-#include "findnearmv.h"
-#include "entropymode.h"
-#include "systemdependent.h"
-
-
-extern void vp8_init_scan_order_mask();
-
-static void update_mode_info_border(MODE_INFO *mi, int rows, int cols)
-{
- int i;
- vpx_memset(mi - cols - 2, 0, sizeof(MODE_INFO) * (cols + 1));
-
- for (i = 0; i < rows; i++)
- {
- vpx_memset(&mi[i*cols-1], 0, sizeof(MODE_INFO));
- }
-}
-
-void vp8_de_alloc_frame_buffers(VP8_COMMON *oci)
-{
- int i;
-
- for (i = 0; i < NUM_YV12_BUFFERS; i++)
- vp8_yv12_de_alloc_frame_buffer(&oci->yv12_fb[i]);
-
- vp8_yv12_de_alloc_frame_buffer(&oci->temp_scale_frame);
- vp8_yv12_de_alloc_frame_buffer(&oci->post_proc_buffer);
-
- vpx_free(oci->above_context);
- vpx_free(oci->mip);
-
- oci->above_context = 0;
- oci->mip = 0;
-
-}
-
-int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height)
-{
- int i;
-
- vp8_de_alloc_frame_buffers(oci);
-
- /* our internal buffers are always multiples of 16 */
- if ((width & 0xf) != 0)
- width += 16 - (width & 0xf);
-
- if ((height & 0xf) != 0)
- height += 16 - (height & 0xf);
-
-
- for (i = 0; i < NUM_YV12_BUFFERS; i++)
- {
- oci->fb_idx_ref_cnt[0] = 0;
-
- if (vp8_yv12_alloc_frame_buffer(&oci->yv12_fb[i], width, height, VP8BORDERINPIXELS) < 0)
- {
- vp8_de_alloc_frame_buffers(oci);
- return 1;
- }
- }
-
- oci->new_fb_idx = 0;
- oci->lst_fb_idx = 1;
- oci->gld_fb_idx = 2;
- oci->alt_fb_idx = 3;
-
- oci->fb_idx_ref_cnt[0] = 1;
- oci->fb_idx_ref_cnt[1] = 1;
- oci->fb_idx_ref_cnt[2] = 1;
- oci->fb_idx_ref_cnt[3] = 1;
-
- if (vp8_yv12_alloc_frame_buffer(&oci->temp_scale_frame, width, 16, VP8BORDERINPIXELS) < 0)
- {
- vp8_de_alloc_frame_buffers(oci);
- return 1;
- }
-
- if (vp8_yv12_alloc_frame_buffer(&oci->post_proc_buffer, width, height, VP8BORDERINPIXELS) < 0)
- {
- vp8_de_alloc_frame_buffers(oci);
- return 1;
- }
-
- oci->mb_rows = height >> 4;
- oci->mb_cols = width >> 4;
- oci->MBs = oci->mb_rows * oci->mb_cols;
- oci->mode_info_stride = oci->mb_cols + 1;
- oci->mip = vpx_calloc((oci->mb_cols + 1) * (oci->mb_rows + 1), sizeof(MODE_INFO));
-
- if (!oci->mip)
- {
- vp8_de_alloc_frame_buffers(oci);
- return 1;
- }
-
- oci->mi = oci->mip + oci->mode_info_stride + 1;
-
-
- oci->above_context = vpx_calloc(sizeof(ENTROPY_CONTEXT_PLANES) * oci->mb_cols, 1);
-
- if (!oci->above_context)
- {
- vp8_de_alloc_frame_buffers(oci);
- return 1;
- }
-
- update_mode_info_border(oci->mi, oci->mb_rows, oci->mb_cols);
-
- return 0;
-}
-void vp8_setup_version(VP8_COMMON *cm)
-{
- switch (cm->version)
- {
- case 0:
- cm->no_lpf = 0;
- cm->simpler_lpf = 0;
- cm->use_bilinear_mc_filter = 0;
- cm->full_pixel = 0;
- break;
- case 1:
- cm->no_lpf = 0;
- cm->simpler_lpf = 1;
- cm->use_bilinear_mc_filter = 1;
- cm->full_pixel = 0;
- break;
- case 2:
- cm->no_lpf = 1;
- cm->simpler_lpf = 0;
- cm->use_bilinear_mc_filter = 1;
- cm->full_pixel = 0;
- break;
- case 3:
- cm->no_lpf = 1;
- cm->simpler_lpf = 1;
- cm->use_bilinear_mc_filter = 1;
- cm->full_pixel = 1;
- break;
- default:
- /*4,5,6,7 are reserved for future use*/
- cm->no_lpf = 0;
- cm->simpler_lpf = 0;
- cm->use_bilinear_mc_filter = 0;
- cm->full_pixel = 0;
- break;
- }
-}
-void vp8_create_common(VP8_COMMON *oci)
-{
- vp8_machine_specific_config(oci);
- vp8_default_coef_probs(oci);
- vp8_init_mbmode_probs(oci);
- vp8_default_bmode_probs(oci->fc.bmode_prob);
-
- oci->mb_no_coeff_skip = 1;
- oci->no_lpf = 0;
- oci->simpler_lpf = 0;
- oci->use_bilinear_mc_filter = 0;
- oci->full_pixel = 0;
- oci->multi_token_partition = ONE_PARTITION;
- oci->clr_type = REG_YUV;
- oci->clamp_type = RECON_CLAMP_REQUIRED;
-
- /* Initialise reference frame sign bias structure to defaults */
- vpx_memset(oci->ref_frame_sign_bias, 0, sizeof(oci->ref_frame_sign_bias));
-
- /* Default disable buffer to buffer copying */
- oci->copy_buffer_to_gf = 0;
- oci->copy_buffer_to_arf = 0;
-}
-
-void vp8_remove_common(VP8_COMMON *oci)
-{
- vp8_de_alloc_frame_buffers(oci);
-}
-
-void vp8_initialize_common()
-{
- vp8_coef_tree_initialize();
-
- vp8_entropy_mode_init();
-
- vp8_init_scan_order_mask();
-
-}
diff --git a/vp8/common/arm/arm_systemdependent.c b/vp8/common/arm/arm_systemdependent.c
deleted file mode 100644
index bd5c075..0000000
--- a/vp8/common/arm/arm_systemdependent.c
+++ /dev/null
@@ -1,109 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_ports/arm.h"
-#include "vp8/common/g_common.h"
-#include "vp8/common/pragmas.h"
-#include "vp8/common/subpixel.h"
-#include "vp8/common/loopfilter.h"
-#include "vp8/common/recon.h"
-#include "vp8/common/idct.h"
-#include "vp8/common/onyxc_int.h"
-
-void vp8_arch_arm_common_init(VP8_COMMON *ctx)
-{
-#if CONFIG_RUNTIME_CPU_DETECT
- VP8_COMMON_RTCD *rtcd = &ctx->rtcd;
- int flags = arm_cpu_caps();
- int has_edsp = flags & HAS_EDSP;
- int has_media = flags & HAS_MEDIA;
- int has_neon = flags & HAS_NEON;
- rtcd->flags = flags;
-
- /* Override default functions with fastest ones for this CPU. */
-#if HAVE_ARMV6
- if (has_media)
- {
- rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_armv6;
- rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_armv6;
- rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_armv6;
- rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_armv6;
- rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_armv6;
- rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_armv6;
- rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_armv6;
- rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_armv6;
-
- rtcd->idct.idct1 = vp8_short_idct4x4llm_1_v6;
- rtcd->idct.idct16 = vp8_short_idct4x4llm_v6_dual;
- rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_v6;
- rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_v6;
-
- rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_armv6;
- rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_armv6;
- rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_armv6;
- rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_armv6;
- rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_armv6;
- rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_armv6;
- rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_armv6;
- rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_armv6;
-
- rtcd->recon.copy16x16 = vp8_copy_mem16x16_v6;
- rtcd->recon.copy8x8 = vp8_copy_mem8x8_v6;
- rtcd->recon.copy8x4 = vp8_copy_mem8x4_v6;
- rtcd->recon.recon = vp8_recon_b_armv6;
- rtcd->recon.recon2 = vp8_recon2b_armv6;
- rtcd->recon.recon4 = vp8_recon4b_armv6;
- }
-#endif
-
-#if HAVE_ARMV7
- if (has_neon)
- {
- rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_neon;
- rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_neon;
- rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_neon;
- rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_neon;
- rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_neon;
- rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_neon;
- rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_neon;
- rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_neon;
-
- rtcd->idct.idct1 = vp8_short_idct4x4llm_1_neon;
- rtcd->idct.idct16 = vp8_short_idct4x4llm_neon;
- rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_neon;
- rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_neon;
-
- rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_neon;
- rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_neon;
- rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_neon;
- rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_neon;
- rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_neon;
- rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_neon;
- rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_neon;
- rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_neon;
-
- rtcd->recon.copy16x16 = vp8_copy_mem16x16_neon;
- rtcd->recon.copy8x8 = vp8_copy_mem8x8_neon;
- rtcd->recon.copy8x4 = vp8_copy_mem8x4_neon;
- rtcd->recon.recon = vp8_recon_b_neon;
- rtcd->recon.recon2 = vp8_recon2b_neon;
- rtcd->recon.recon4 = vp8_recon4b_neon;
- rtcd->recon.recon_mb = vp8_recon_mb_neon;
- rtcd->recon.build_intra_predictors_mby =
- vp8_build_intra_predictors_mby_neon;
- rtcd->recon.build_intra_predictors_mby_s =
- vp8_build_intra_predictors_mby_s_neon;
- }
-#endif
-
-#endif
-}
diff --git a/vp8/common/arm/armv6/bilinearfilter_v6.asm b/vp8/common/arm/armv6/bilinearfilter_v6.asm
deleted file mode 100644
index a86ed5d..0000000
--- a/vp8/common/arm/armv6/bilinearfilter_v6.asm
+++ /dev/null
@@ -1,238 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_filter_block2d_bil_first_pass_armv6|
- EXPORT |vp8_filter_block2d_bil_second_pass_armv6|
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
-;-------------------------------------
-; r0 unsigned char *src_ptr,
-; r1 unsigned short *dst_ptr,
-; r2 unsigned int src_pitch,
-; r3 unsigned int height,
-; stack unsigned int width,
-; stack const short *vp8_filter
-;-------------------------------------
-; The output is transposed stroed in output array to make it easy for second pass filtering.
-|vp8_filter_block2d_bil_first_pass_armv6| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r11, [sp, #40] ; vp8_filter address
- ldr r4, [sp, #36] ; width
-
- mov r12, r3 ; outer-loop counter
- sub r2, r2, r4 ; src increment for height loop
-
- ;;IF ARCHITECTURE=6
- pld [r0]
- ;;ENDIF
-
- ldr r5, [r11] ; load up filter coefficients
-
- mov r3, r3, lsl #1 ; height*2
- add r3, r3, #2 ; plus 2 to make output buffer 4-bit aligned since height is actually (height+1)
-
- mov r11, r1 ; save dst_ptr for each row
-
- cmp r5, #128 ; if filter coef = 128, then skip the filter
- beq bil_null_1st_filter
-
-|bil_height_loop_1st_v6|
- ldrb r6, [r0] ; load source data
- ldrb r7, [r0, #1]
- ldrb r8, [r0, #2]
- mov lr, r4, lsr #2 ; 4-in-parellel loop counter
-
-|bil_width_loop_1st_v6|
- ldrb r9, [r0, #3]
- ldrb r10, [r0, #4]
-
- pkhbt r6, r6, r7, lsl #16 ; src[1] | src[0]
- pkhbt r7, r7, r8, lsl #16 ; src[2] | src[1]
-
- smuad r6, r6, r5 ; apply the filter
- pkhbt r8, r8, r9, lsl #16 ; src[3] | src[2]
- smuad r7, r7, r5
- pkhbt r9, r9, r10, lsl #16 ; src[4] | src[3]
-
- smuad r8, r8, r5
- smuad r9, r9, r5
-
- add r0, r0, #4
- subs lr, lr, #1
-
- add r6, r6, #0x40 ; round_shift_and_clamp
- add r7, r7, #0x40
- usat r6, #16, r6, asr #7
- usat r7, #16, r7, asr #7
-
- strh r6, [r1], r3 ; result is transposed and stored
-
- add r8, r8, #0x40 ; round_shift_and_clamp
- strh r7, [r1], r3
- add r9, r9, #0x40
- usat r8, #16, r8, asr #7
- usat r9, #16, r9, asr #7
-
- strh r8, [r1], r3 ; result is transposed and stored
-
- ldrneb r6, [r0] ; load source data
- strh r9, [r1], r3
-
- ldrneb r7, [r0, #1]
- ldrneb r8, [r0, #2]
-
- bne bil_width_loop_1st_v6
-
- add r0, r0, r2 ; move to next input row
- subs r12, r12, #1
-
- ;;IF ARCHITECTURE=6
- pld [r0]
- ;;ENDIF
-
- add r11, r11, #2 ; move over to next column
- mov r1, r11
-
- bne bil_height_loop_1st_v6
-
- ldmia sp!, {r4 - r11, pc}
-
-|bil_null_1st_filter|
-|bil_height_loop_null_1st|
- mov lr, r4, lsr #2 ; loop counter
-
-|bil_width_loop_null_1st|
- ldrb r6, [r0] ; load data
- ldrb r7, [r0, #1]
- ldrb r8, [r0, #2]
- ldrb r9, [r0, #3]
-
- strh r6, [r1], r3 ; store it to immediate buffer
- add r0, r0, #4
- strh r7, [r1], r3
- subs lr, lr, #1
- strh r8, [r1], r3
- strh r9, [r1], r3
-
- bne bil_width_loop_null_1st
-
- subs r12, r12, #1
- add r0, r0, r2 ; move to next input line
- add r11, r11, #2 ; move over to next column
- mov r1, r11
-
- bne bil_height_loop_null_1st
-
- ldmia sp!, {r4 - r11, pc}
-
- ENDP ; |vp8_filter_block2d_bil_first_pass_armv6|
-
-
-;---------------------------------
-; r0 unsigned short *src_ptr,
-; r1 unsigned char *dst_ptr,
-; r2 int dst_pitch,
-; r3 unsigned int height,
-; stack unsigned int width,
-; stack const short *vp8_filter
-;---------------------------------
-|vp8_filter_block2d_bil_second_pass_armv6| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r11, [sp, #40] ; vp8_filter address
- ldr r4, [sp, #36] ; width
-
- ldr r5, [r11] ; load up filter coefficients
- mov r12, r4 ; outer-loop counter = width, since we work on transposed data matrix
- mov r11, r1
-
- cmp r5, #128 ; if filter coef = 128, then skip the filter
- beq bil_null_2nd_filter
-
-|bil_height_loop_2nd|
- ldr r6, [r0] ; load the data
- ldr r8, [r0, #4]
- ldrh r10, [r0, #8]
- mov lr, r3, lsr #2 ; loop counter
-
-|bil_width_loop_2nd|
- pkhtb r7, r6, r8 ; src[1] | src[2]
- pkhtb r9, r8, r10 ; src[3] | src[4]
-
- smuad r6, r6, r5 ; apply filter
- smuad r8, r8, r5 ; apply filter
-
- subs lr, lr, #1
-
- smuadx r7, r7, r5 ; apply filter
- smuadx r9, r9, r5 ; apply filter
-
- add r0, r0, #8
-
- add r6, r6, #0x40 ; round_shift_and_clamp
- add r7, r7, #0x40
- usat r6, #8, r6, asr #7
- usat r7, #8, r7, asr #7
- strb r6, [r1], r2 ; the result is transposed back and stored
-
- add r8, r8, #0x40 ; round_shift_and_clamp
- strb r7, [r1], r2
- add r9, r9, #0x40
- usat r8, #8, r8, asr #7
- usat r9, #8, r9, asr #7
- strb r8, [r1], r2 ; the result is transposed back and stored
-
- ldrne r6, [r0] ; load data
- strb r9, [r1], r2
- ldrne r8, [r0, #4]
- ldrneh r10, [r0, #8]
-
- bne bil_width_loop_2nd
-
- subs r12, r12, #1
- add r0, r0, #4 ; update src for next row
- add r11, r11, #1
- mov r1, r11
-
- bne bil_height_loop_2nd
- ldmia sp!, {r4 - r11, pc}
-
-|bil_null_2nd_filter|
-|bil_height_loop_null_2nd|
- mov lr, r3, lsr #2
-
-|bil_width_loop_null_2nd|
- ldr r6, [r0], #4 ; load data
- subs lr, lr, #1
- ldr r8, [r0], #4
-
- strb r6, [r1], r2 ; store data
- mov r7, r6, lsr #16
- strb r7, [r1], r2
- mov r9, r8, lsr #16
- strb r8, [r1], r2
- strb r9, [r1], r2
-
- bne bil_width_loop_null_2nd
-
- subs r12, r12, #1
- add r0, r0, #4
- add r11, r11, #1
- mov r1, r11
-
- bne bil_height_loop_null_2nd
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_filter_block2d_second_pass_armv6|
-
- END
diff --git a/vp8/common/arm/armv6/copymem16x16_v6.asm b/vp8/common/arm/armv6/copymem16x16_v6.asm
deleted file mode 100644
index fca91a0..0000000
--- a/vp8/common/arm/armv6/copymem16x16_v6.asm
+++ /dev/null
@@ -1,182 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_copy_mem16x16_v6|
- ; ARM
- ; REQUIRE8
- ; PRESERVE8
-
- AREA Block, CODE, READONLY ; name this block of code
-;void copy_mem16x16_v6( unsigned char *src, int src_stride, unsigned char *dst, int dst_stride)
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-|vp8_copy_mem16x16_v6| PROC
- stmdb sp!, {r4 - r7}
- ;push {r4-r7}
-
- ;preload
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- ands r4, r0, #15
- beq copy_mem16x16_fast
-
- ands r4, r0, #7
- beq copy_mem16x16_8
-
- ands r4, r0, #3
- beq copy_mem16x16_4
-
- ;copy one byte each time
- ldrb r4, [r0]
- ldrb r5, [r0, #1]
- ldrb r6, [r0, #2]
- ldrb r7, [r0, #3]
-
- mov r12, #16
-
-copy_mem16x16_1_loop
- strb r4, [r2]
- strb r5, [r2, #1]
- strb r6, [r2, #2]
- strb r7, [r2, #3]
-
- ldrb r4, [r0, #4]
- ldrb r5, [r0, #5]
- ldrb r6, [r0, #6]
- ldrb r7, [r0, #7]
-
- subs r12, r12, #1
-
- strb r4, [r2, #4]
- strb r5, [r2, #5]
- strb r6, [r2, #6]
- strb r7, [r2, #7]
-
- ldrb r4, [r0, #8]
- ldrb r5, [r0, #9]
- ldrb r6, [r0, #10]
- ldrb r7, [r0, #11]
-
- strb r4, [r2, #8]
- strb r5, [r2, #9]
- strb r6, [r2, #10]
- strb r7, [r2, #11]
-
- ldrb r4, [r0, #12]
- ldrb r5, [r0, #13]
- ldrb r6, [r0, #14]
- ldrb r7, [r0, #15]
-
- add r0, r0, r1
-
- strb r4, [r2, #12]
- strb r5, [r2, #13]
- strb r6, [r2, #14]
- strb r7, [r2, #15]
-
- add r2, r2, r3
-
- ldrneb r4, [r0]
- ldrneb r5, [r0, #1]
- ldrneb r6, [r0, #2]
- ldrneb r7, [r0, #3]
-
- bne copy_mem16x16_1_loop
-
- ldmia sp!, {r4 - r7}
- ;pop {r4-r7}
- mov pc, lr
-
-;copy 4 bytes each time
-copy_mem16x16_4
- ldr r4, [r0]
- ldr r5, [r0, #4]
- ldr r6, [r0, #8]
- ldr r7, [r0, #12]
-
- mov r12, #16
-
-copy_mem16x16_4_loop
- subs r12, r12, #1
- add r0, r0, r1
-
- str r4, [r2]
- str r5, [r2, #4]
- str r6, [r2, #8]
- str r7, [r2, #12]
-
- add r2, r2, r3
-
- ldrne r4, [r0]
- ldrne r5, [r0, #4]
- ldrne r6, [r0, #8]
- ldrne r7, [r0, #12]
-
- bne copy_mem16x16_4_loop
-
- ldmia sp!, {r4 - r7}
- ;pop {r4-r7}
- mov pc, lr
-
-;copy 8 bytes each time
-copy_mem16x16_8
- sub r1, r1, #16
- sub r3, r3, #16
-
- mov r12, #16
-
-copy_mem16x16_8_loop
- ldmia r0!, {r4-r5}
- ;ldm r0, {r4-r5}
- ldmia r0!, {r6-r7}
-
- add r0, r0, r1
-
- stmia r2!, {r4-r5}
- subs r12, r12, #1
- ;stm r2, {r4-r5}
- stmia r2!, {r6-r7}
-
- add r2, r2, r3
-
- bne copy_mem16x16_8_loop
-
- ldmia sp!, {r4 - r7}
- ;pop {r4-r7}
- mov pc, lr
-
-;copy 16 bytes each time
-copy_mem16x16_fast
- ;sub r1, r1, #16
- ;sub r3, r3, #16
-
- mov r12, #16
-
-copy_mem16x16_fast_loop
- ldmia r0, {r4-r7}
- ;ldm r0, {r4-r7}
- add r0, r0, r1
-
- subs r12, r12, #1
- stmia r2, {r4-r7}
- ;stm r2, {r4-r7}
- add r2, r2, r3
-
- bne copy_mem16x16_fast_loop
-
- ldmia sp!, {r4 - r7}
- ;pop {r4-r7}
- mov pc, lr
-
- ENDP ; |vp8_copy_mem16x16_v6|
-
- END
diff --git a/vp8/common/arm/armv6/dc_only_idct_add_v6.asm b/vp8/common/arm/armv6/dc_only_idct_add_v6.asm
deleted file mode 100644
index e0660e9..0000000
--- a/vp8/common/arm/armv6/dc_only_idct_add_v6.asm
+++ /dev/null
@@ -1,67 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license and patent
-; grant that can be found in the LICENSE file in the root of the source
-; tree. All contributing project authors may be found in the AUTHORS
-; file in the root of the source tree.
-;
-
- EXPORT |vp8_dc_only_idct_add_v6|
-
- AREA |.text|, CODE, READONLY
-
-;void vp8_dc_only_idct_add_v6(short input_dc, unsigned char *pred_ptr,
-; unsigned char *dst_ptr, int pitch, int stride)
-; r0 input_dc
-; r1 pred_ptr
-; r2 dest_ptr
-; r3 pitch
-; sp stride
-
-|vp8_dc_only_idct_add_v6| PROC
- stmdb sp!, {r4 - r7, lr}
-
- add r0, r0, #4 ; input_dc += 4
- ldr r12, c0x0000FFFF
- ldr r4, [r1], r3
- ldr r6, [r1], r3
- and r0, r12, r0, asr #3 ; input_dc >> 3 + mask
- ldr lr, [sp, #20]
- orr r0, r0, r0, lsl #16 ; a1 | a1
-
- uxtab16 r5, r0, r4 ; a1+2 | a1+0
- uxtab16 r4, r0, r4, ror #8 ; a1+3 | a1+1
- uxtab16 r7, r0, r6
- uxtab16 r6, r0, r6, ror #8
- usat16 r5, #8, r5
- usat16 r4, #8, r4
- usat16 r7, #8, r7
- usat16 r6, #8, r6
- orr r5, r5, r4, lsl #8
- orr r7, r7, r6, lsl #8
- ldr r4, [r1], r3
- ldr r6, [r1]
- str r5, [r2], lr
- str r7, [r2], lr
-
- uxtab16 r5, r0, r4
- uxtab16 r4, r0, r4, ror #8
- uxtab16 r7, r0, r6
- uxtab16 r6, r0, r6, ror #8
- usat16 r5, #8, r5
- usat16 r4, #8, r4
- usat16 r7, #8, r7
- usat16 r6, #8, r6
- orr r5, r5, r4, lsl #8
- orr r7, r7, r6, lsl #8
- str r5, [r2], lr
- str r7, [r2]
-
- ldmia sp!, {r4 - r7, pc}
-
- ENDP ; |vp8_dc_only_idct_add_v6|
-
-; Constant Pool
-c0x0000FFFF DCD 0x0000FFFF
- END
diff --git a/vp8/common/arm/armv6/filter_v6.asm b/vp8/common/arm/armv6/filter_v6.asm
deleted file mode 100644
index 03b5bcc..0000000
--- a/vp8/common/arm/armv6/filter_v6.asm
+++ /dev/null
@@ -1,443 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_filter_block2d_first_pass_armv6|
- EXPORT |vp8_filter_block2d_second_pass_armv6|
- EXPORT |vp8_filter4_block2d_second_pass_armv6|
- EXPORT |vp8_filter_block2d_first_pass_only_armv6|
- EXPORT |vp8_filter_block2d_second_pass_only_armv6|
-
- AREA |.text|, CODE, READONLY ; name this block of code
-;-------------------------------------
-; r0 unsigned char *src_ptr
-; r1 short *output_ptr
-; r2 unsigned int src_pixels_per_line
-; r3 unsigned int output_width
-; stack unsigned int output_height
-; stack const short *vp8_filter
-;-------------------------------------
-; vp8_filter the input and put in the output array. Apply the 6 tap FIR filter with
-; the output being a 2 byte value and the intput being a 1 byte value.
-|vp8_filter_block2d_first_pass_armv6| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r11, [sp, #40] ; vp8_filter address
- ldr r7, [sp, #36] ; output height
-
- sub r2, r2, r3 ; inside loop increments input array,
- ; so the height loop only needs to add
- ; r2 - width to the input pointer
-
- mov r3, r3, lsl #1 ; multiply width by 2 because using shorts
- add r12, r3, #16 ; square off the output
- sub sp, sp, #4
-
- ;;IF ARCHITECTURE=6
- ;pld [r0, #-2]
- ;;pld [r0, #30]
- ;;ENDIF
-
- ldr r4, [r11] ; load up packed filter coefficients
- ldr r5, [r11, #4]
- ldr r6, [r11, #8]
-
- str r1, [sp] ; push destination to stack
- mov r7, r7, lsl #16 ; height is top part of counter
-
-; six tap filter
-|height_loop_1st_6|
- ldrb r8, [r0, #-2] ; load source data
- ldrb r9, [r0, #-1]
- ldrb r10, [r0], #2
- orr r7, r7, r3, lsr #2 ; construct loop counter
-
-|width_loop_1st_6|
- ldrb r11, [r0, #-1]
-
- pkhbt lr, r8, r9, lsl #16 ; r9 | r8
- pkhbt r8, r9, r10, lsl #16 ; r10 | r9
-
- ldrb r9, [r0]
-
- smuad lr, lr, r4 ; apply the filter
- pkhbt r10, r10, r11, lsl #16 ; r11 | r10
- smuad r8, r8, r4
- pkhbt r11, r11, r9, lsl #16 ; r9 | r11
-
- smlad lr, r10, r5, lr
- ldrb r10, [r0, #1]
- smlad r8, r11, r5, r8
- ldrb r11, [r0, #2]
-
- sub r7, r7, #1
-
- pkhbt r9, r9, r10, lsl #16 ; r10 | r9
- pkhbt r10, r10, r11, lsl #16 ; r11 | r10
-
- smlad lr, r9, r6, lr
- smlad r11, r10, r6, r8
-
- ands r10, r7, #0xff ; test loop counter
-
- add lr, lr, #0x40 ; round_shift_and_clamp
- ldrneb r8, [r0, #-2] ; load data for next loop
- usat lr, #8, lr, asr #7
- add r11, r11, #0x40
- ldrneb r9, [r0, #-1]
- usat r11, #8, r11, asr #7
-
- strh lr, [r1], r12 ; result is transposed and stored, which
- ; will make second pass filtering easier.
- ldrneb r10, [r0], #2
- strh r11, [r1], r12
-
- bne width_loop_1st_6
-
- ;;add r9, r2, #30 ; attempt to load 2 adjacent cache lines
- ;;IF ARCHITECTURE=6
- ;pld [r0, r2]
- ;;pld [r0, r9]
- ;;ENDIF
-
- ldr r1, [sp] ; load and update dst address
- subs r7, r7, #0x10000
- add r0, r0, r2 ; move to next input line
- add r1, r1, #2 ; move over to next column
- str r1, [sp]
-
- bne height_loop_1st_6
-
- add sp, sp, #4
- ldmia sp!, {r4 - r11, pc}
-
- ENDP
-
-;---------------------------------
-; r0 short *src_ptr,
-; r1 unsigned char *output_ptr,
-; r2 unsigned int output_pitch,
-; r3 unsigned int cnt,
-; stack const short *vp8_filter
-;---------------------------------
-|vp8_filter_block2d_second_pass_armv6| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r11, [sp, #36] ; vp8_filter address
- sub sp, sp, #4
- mov r7, r3, lsl #16 ; height is top part of counter
- str r1, [sp] ; push destination to stack
-
- ldr r4, [r11] ; load up packed filter coefficients
- ldr r5, [r11, #4]
- ldr r6, [r11, #8]
-
- pkhbt r12, r5, r4 ; pack the filter differently
- pkhbt r11, r6, r5
-
- sub r0, r0, #4 ; offset input buffer
-
-|height_loop_2nd|
- ldr r8, [r0] ; load the data
- ldr r9, [r0, #4]
- orr r7, r7, r3, lsr #1 ; loop counter
-
-|width_loop_2nd|
- smuad lr, r4, r8 ; apply filter
- sub r7, r7, #1
- smulbt r8, r4, r8
-
- ldr r10, [r0, #8]
-
- smlad lr, r5, r9, lr
- smladx r8, r12, r9, r8
-
- ldrh r9, [r0, #12]
-
- smlad lr, r6, r10, lr
- smladx r8, r11, r10, r8
-
- add r0, r0, #4
- smlatb r10, r6, r9, r8
-
- add lr, lr, #0x40 ; round_shift_and_clamp
- ands r8, r7, #0xff
- usat lr, #8, lr, asr #7
- add r10, r10, #0x40
- strb lr, [r1], r2 ; the result is transposed back and stored
- usat r10, #8, r10, asr #7
-
- ldrne r8, [r0] ; load data for next loop
- ldrne r9, [r0, #4]
- strb r10, [r1], r2
-
- bne width_loop_2nd
-
- ldr r1, [sp] ; update dst for next loop
- subs r7, r7, #0x10000
- add r0, r0, #16 ; updata src for next loop
- add r1, r1, #1
- str r1, [sp]
-
- bne height_loop_2nd
-
- add sp, sp, #4
- ldmia sp!, {r4 - r11, pc}
-
- ENDP
-
-;---------------------------------
-; r0 short *src_ptr,
-; r1 unsigned char *output_ptr,
-; r2 unsigned int output_pitch,
-; r3 unsigned int cnt,
-; stack const short *vp8_filter
-;---------------------------------
-|vp8_filter4_block2d_second_pass_armv6| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r11, [sp, #36] ; vp8_filter address
- mov r7, r3, lsl #16 ; height is top part of counter
-
- ldr r4, [r11] ; load up packed filter coefficients
- add lr, r1, r3 ; save final destination pointer
- ldr r5, [r11, #4]
- ldr r6, [r11, #8]
-
- pkhbt r12, r5, r4 ; pack the filter differently
- pkhbt r11, r6, r5
- mov r4, #0x40 ; rounding factor (for smlad{x})
-
-|height_loop_2nd_4|
- ldrd r8, [r0, #-4] ; load the data
- orr r7, r7, r3, lsr #1 ; loop counter
-
-|width_loop_2nd_4|
- ldr r10, [r0, #4]!
- smladx r6, r9, r12, r4 ; apply filter
- pkhbt r8, r9, r8
- smlad r5, r8, r12, r4
- pkhbt r8, r10, r9
- smladx r6, r10, r11, r6
- sub r7, r7, #1
- smlad r5, r8, r11, r5
-
- mov r8, r9 ; shift the data for the next loop
- mov r9, r10
-
- usat r6, #8, r6, asr #7 ; shift and clamp
- usat r5, #8, r5, asr #7
-
- strb r5, [r1], r2 ; the result is transposed back and stored
- tst r7, #0xff
- strb r6, [r1], r2
-
- bne width_loop_2nd_4
-
- subs r7, r7, #0x10000
- add r0, r0, #16 ; update src for next loop
- sub r1, lr, r7, lsr #16 ; update dst for next loop
-
- bne height_loop_2nd_4
-
- ldmia sp!, {r4 - r11, pc}
-
- ENDP
-
-;------------------------------------
-; r0 unsigned char *src_ptr
-; r1 unsigned char *output_ptr,
-; r2 unsigned int src_pixels_per_line
-; r3 unsigned int cnt,
-; stack unsigned int output_pitch,
-; stack const short *vp8_filter
-;------------------------------------
-|vp8_filter_block2d_first_pass_only_armv6| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r4, [sp, #36] ; output pitch
- ldr r11, [sp, #40] ; HFilter address
- sub sp, sp, #8
-
- mov r7, r3
- sub r2, r2, r3 ; inside loop increments input array,
- ; so the height loop only needs to add
- ; r2 - width to the input pointer
-
- sub r4, r4, r3
- str r4, [sp] ; save modified output pitch
- str r2, [sp, #4]
-
- mov r2, #0x40
-
- ldr r4, [r11] ; load up packed filter coefficients
- ldr r5, [r11, #4]
- ldr r6, [r11, #8]
-
-; six tap filter
-|height_loop_1st_only_6|
- ldrb r8, [r0, #-2] ; load data
- ldrb r9, [r0, #-1]
- ldrb r10, [r0], #2
-
- mov r12, r3, lsr #1 ; loop counter
-
-|width_loop_1st_only_6|
- ldrb r11, [r0, #-1]
-
- pkhbt lr, r8, r9, lsl #16 ; r9 | r8
- pkhbt r8, r9, r10, lsl #16 ; r10 | r9
-
- ldrb r9, [r0]
-
-;; smuad lr, lr, r4
- smlad lr, lr, r4, r2
- pkhbt r10, r10, r11, lsl #16 ; r11 | r10
-;; smuad r8, r8, r4
- smlad r8, r8, r4, r2
- pkhbt r11, r11, r9, lsl #16 ; r9 | r11
-
- smlad lr, r10, r5, lr
- ldrb r10, [r0, #1]
- smlad r8, r11, r5, r8
- ldrb r11, [r0, #2]
-
- subs r12, r12, #1
-
- pkhbt r9, r9, r10, lsl #16 ; r10 | r9
- pkhbt r10, r10, r11, lsl #16 ; r11 | r10
-
- smlad lr, r9, r6, lr
- smlad r10, r10, r6, r8
-
-;; add lr, lr, #0x40 ; round_shift_and_clamp
- ldrneb r8, [r0, #-2] ; load data for next loop
- usat lr, #8, lr, asr #7
-;; add r10, r10, #0x40
- strb lr, [r1], #1 ; store the result
- usat r10, #8, r10, asr #7
-
- ldrneb r9, [r0, #-1]
- strb r10, [r1], #1
- ldrneb r10, [r0], #2
-
- bne width_loop_1st_only_6
-
- ;;add r9, r2, #30 ; attempt to load 2 adjacent cache lines
- ;;IF ARCHITECTURE=6
- ;pld [r0, r2]
- ;;pld [r0, r9]
- ;;ENDIF
-
- ldr lr, [sp] ; load back output pitch
- ldr r12, [sp, #4] ; load back output pitch
- subs r7, r7, #1
- add r0, r0, r12 ; updata src for next loop
- add r1, r1, lr ; update dst for next loop
-
- bne height_loop_1st_only_6
-
- add sp, sp, #8
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_filter_block2d_first_pass_only_armv6|
-
-
-;------------------------------------
-; r0 unsigned char *src_ptr,
-; r1 unsigned char *output_ptr,
-; r2 unsigned int src_pixels_per_line
-; r3 unsigned int cnt,
-; stack unsigned int output_pitch,
-; stack const short *vp8_filter
-;------------------------------------
-|vp8_filter_block2d_second_pass_only_armv6| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r11, [sp, #40] ; VFilter address
- ldr r12, [sp, #36] ; output pitch
-
- mov r7, r3, lsl #16 ; height is top part of counter
- sub r0, r0, r2, lsl #1 ; need 6 elements for filtering, 2 before, 3 after
-
- sub sp, sp, #8
-
- ldr r4, [r11] ; load up packed filter coefficients
- ldr r5, [r11, #4]
- ldr r6, [r11, #8]
-
- str r0, [sp] ; save r0 to stack
- str r1, [sp, #4] ; save dst to stack
-
-; six tap filter
-|width_loop_2nd_only_6|
- ldrb r8, [r0], r2 ; load data
- orr r7, r7, r3 ; loop counter
- ldrb r9, [r0], r2
- ldrb r10, [r0], r2
-
-|height_loop_2nd_only_6|
- ; filter first column in this inner loop, than, move to next colum.
- ldrb r11, [r0], r2
-
- pkhbt lr, r8, r9, lsl #16 ; r9 | r8
- pkhbt r8, r9, r10, lsl #16 ; r10 | r9
-
- ldrb r9, [r0], r2
-
- smuad lr, lr, r4
- pkhbt r10, r10, r11, lsl #16 ; r11 | r10
- smuad r8, r8, r4
- pkhbt r11, r11, r9, lsl #16 ; r9 | r11
-
- smlad lr, r10, r5, lr
- ldrb r10, [r0], r2
- smlad r8, r11, r5, r8
- ldrb r11, [r0]
-
- sub r7, r7, #2
- sub r0, r0, r2, lsl #2
-
- pkhbt r9, r9, r10, lsl #16 ; r10 | r9
- pkhbt r10, r10, r11, lsl #16 ; r11 | r10
-
- smlad lr, r9, r6, lr
- smlad r10, r10, r6, r8
-
- ands r9, r7, #0xff
-
- add lr, lr, #0x40 ; round_shift_and_clamp
- ldrneb r8, [r0], r2 ; load data for next loop
- usat lr, #8, lr, asr #7
- add r10, r10, #0x40
- strb lr, [r1], r12 ; store the result for the column
- usat r10, #8, r10, asr #7
-
- ldrneb r9, [r0], r2
- strb r10, [r1], r12
- ldrneb r10, [r0], r2
-
- bne height_loop_2nd_only_6
-
- ldr r0, [sp]
- ldr r1, [sp, #4]
- subs r7, r7, #0x10000
- add r0, r0, #1 ; move to filter next column
- str r0, [sp]
- add r1, r1, #1
- str r1, [sp, #4]
-
- bne width_loop_2nd_only_6
-
- add sp, sp, #8
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_filter_block2d_second_pass_only_armv6|
-
- END
diff --git a/vp8/common/arm/armv6/idct_v6.asm b/vp8/common/arm/armv6/idct_v6.asm
deleted file mode 100644
index 27215af..0000000
--- a/vp8/common/arm/armv6/idct_v6.asm
+++ /dev/null
@@ -1,345 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-; r0 r1 r2 r3 r4 r5 r6 r7 r8 r9 r10 r11 r12 r14
- EXPORT |vp8_short_idct4x4llm_1_v6|
- EXPORT |vp8_short_idct4x4llm_v6|
- EXPORT |vp8_short_idct4x4llm_v6_scott|
- EXPORT |vp8_short_idct4x4llm_v6_dual|
-
- AREA |.text|, CODE, READONLY
-
-;********************************************************************************
-;* void short_idct4x4llm_1_v6(INT16 * input, INT16 * output, INT32 pitch)
-;* r0 INT16 * input
-;* r1 INT16 * output
-;* r2 INT32 pitch
-;* bench: 3/5
-;********************************************************************************
-
-|vp8_short_idct4x4llm_1_v6| PROC ; cycles in out pit
- ;
- ldrsh r0, [r0] ; load input[0] 1, r0 un 2
- add r0, r0, #4 ; 1 +4
- stmdb sp!, {r4, r5, lr} ; make room for wide writes 1 backup
- mov r0, r0, asr #3 ; (input[0] + 4) >> 3 1, r0 req`d ^1 >> 3
- pkhbt r4, r0, r0, lsl #16 ; pack r0 into r4 1, r0 req`d ^1 pack
- mov r5, r4 ; expand expand
-
- strd r4, [r1], r2 ; *output = r0, post inc 1
- strd r4, [r1], r2 ; 1
- strd r4, [r1], r2 ; 1
- strd r4, [r1] ; 1
- ;
- ldmia sp!, {r4, r5, pc} ; replace vars, return restore
- ENDP ; |vp8_short_idct4x4llm_1_v6|
-;********************************************************************************
-;********************************************************************************
-;********************************************************************************
-
-;********************************************************************************
-;* void short_idct4x4llm_v6(INT16 * input, INT16 * output, INT32 pitch)
-;* r0 INT16 * input
-;* r1 INT16 * output
-;* r2 INT32 pitch
-;* bench:
-;********************************************************************************
-
-|vp8_short_idct4x4llm_v6| PROC ; cycles in out pit
- ;
- stmdb sp!, {r4-r11, lr} ; backup registers 1 backup
- ;
- mov r4, #0x00004E00 ; 1 cst
- orr r4, r4, #0x0000007B ; cospi8sqrt2minus1
- mov r5, #0x00008A00 ; 1 cst
- orr r5, r5, #0x0000008C ; sinpi8sqrt2
- ;
- mov r6, #4 ; i=4 1 i
-loop1 ;
- ldrsh r12, [r0, #8] ; input[4] 1, r12 unavail 2 [4]
- ldrsh r3, [r0, #24] ; input[12] 1, r3 unavail 2 [12]
- ldrsh r8, [r0, #16] ; input[8] 1, r8 unavail 2 [8]
- ldrsh r7, [r0], #0x2 ; input[0] 1, r7 unavail 2 ++ [0]
- smulwb r10, r5, r12 ; ([4] * sinpi8sqrt2) >> 16 1, r10 un 2, r12/r5 ^1 t1
- smulwb r11, r4, r3 ; ([12] * cospi8sqrt2minus1) >> 16 1, r11 un 2, r3/r4 ^1 t2
- add r9, r7, r8 ; a1 = [0] + [8] 1 a1
- sub r7, r7, r8 ; b1 = [0] - [8] 1 b1
- add r11, r3, r11 ; temp2 1
- rsb r11, r11, r10 ; c1 = temp1 - temp2 1 c1
- smulwb r3, r5, r3 ; ([12] * sinpi8sqrt2) >> 16 1, r3 un 2, r3/r5 ^ 1 t2
- smulwb r10, r4, r12 ; ([4] * cospi8sqrt2minus1) >> 16 1, r10 un 2, r12/r4 ^1 t1
- add r8, r7, r11 ; b1 + c1 1 b+c
- strh r8, [r1, r2] ; out[pitch] = b1+c1 1
- sub r7, r7, r11 ; b1 - c1 1 b-c
- add r10, r12, r10 ; temp1 1
- add r3, r10, r3 ; d1 = temp1 + temp2 1 d1
- add r10, r9, r3 ; a1 + d1 1 a+d
- sub r3, r9, r3 ; a1 - d1 1 a-d
- add r8, r2, r2 ; pitch * 2 1 p*2
- strh r7, [r1, r8] ; out[pitch*2] = b1-c1 1
- add r7, r2, r2, lsl #1 ; pitch * 3 1 p*3
- strh r3, [r1, r7] ; out[pitch*3] = a1-d1 1
- subs r6, r6, #1 ; i-- 1 --
- strh r10, [r1], #0x2 ; out[0] = a1+d1 1 ++
- bne loop1 ; if i>0, continue
- ;
- sub r1, r1, #8 ; set up out for next loop 1 -4
- ; for this iteration, input=prev output
- mov r6, #4 ; i=4 1 i
-; b returnfull
-loop2 ;
- ldrsh r11, [r1, #2] ; input[1] 1, r11 un 2 [1]
- ldrsh r8, [r1, #6] ; input[3] 1, r8 un 2 [3]
- ldrsh r3, [r1, #4] ; input[2] 1, r3 un 2 [2]
- ldrsh r0, [r1] ; input[0] 1, r0 un 2 [0]
- smulwb r9, r5, r11 ; ([1] * sinpi8sqrt2) >> 16 1, r9 un 2, r5/r11 ^1 t1
- smulwb r10, r4, r8 ; ([3] * cospi8sqrt2minus1) >> 16 1, r10 un 2, r4/r8 ^1 t2
- add r7, r0, r3 ; a1 = [0] + [2] 1 a1
- sub r0, r0, r3 ; b1 = [0] - [2] 1 b1
- add r10, r8, r10 ; temp2 1
- rsb r9, r10, r9 ; c1 = temp1 - temp2 1 c1
- smulwb r8, r5, r8 ; ([3] * sinpi8sqrt2) >> 16 1, r8 un 2, r5/r8 ^1 t2
- smulwb r10, r4, r11 ; ([1] * cospi8sqrt2minus1) >> 16 1, r10 un 2, r4/r11 ^1 t1
- add r3, r0, r9 ; b1+c1 1 b+c
- add r3, r3, #4 ; b1+c1+4 1 +4
- add r10, r11, r10 ; temp1 1
- mov r3, r3, asr #3 ; b1+c1+4 >> 3 1, r3 ^1 >>3
- strh r3, [r1, #2] ; out[1] = b1+c1 1
- add r10, r10, r8 ; d1 = temp1 + temp2 1 d1
- add r3, r7, r10 ; a1+d1 1 a+d
- add r3, r3, #4 ; a1+d1+4 1 +4
- sub r7, r7, r10 ; a1-d1 1 a-d
- add r7, r7, #4 ; a1-d1+4 1 +4
- mov r3, r3, asr #3 ; a1+d1+4 >> 3 1, r3 ^1 >>3
- mov r7, r7, asr #3 ; a1-d1+4 >> 3 1, r7 ^1 >>3
- strh r7, [r1, #6] ; out[3] = a1-d1 1
- sub r0, r0, r9 ; b1-c1 1 b-c
- add r0, r0, #4 ; b1-c1+4 1 +4
- subs r6, r6, #1 ; i-- 1 --
- mov r0, r0, asr #3 ; b1-c1+4 >> 3 1, r0 ^1 >>3
- strh r0, [r1, #4] ; out[2] = b1-c1 1
- strh r3, [r1], r2 ; out[0] = a1+d1 1
-; add r1, r1, r2 ; out += pitch 1 ++
- bne loop2 ; if i>0, continue
-returnfull ;
- ldmia sp!, {r4 - r11, pc} ; replace vars, return restore
- ENDP
-
-;********************************************************************************
-;********************************************************************************
-;********************************************************************************
-
-;********************************************************************************
-;* void short_idct4x4llm_v6_scott(INT16 * input, INT16 * output, INT32 pitch)
-;* r0 INT16 * input
-;* r1 INT16 * output
-;* r2 INT32 pitch
-;* bench:
-;********************************************************************************
-
-|vp8_short_idct4x4llm_v6_scott| PROC ; cycles in out pit
-; mov r0, #0 ;
-; ldr r0, [r0] ;
- stmdb sp!, {r4 - r11, lr} ; backup registers 1 backup
- ;
- mov r3, #0x00004E00 ; cos
- orr r3, r3, #0x0000007B ; cospi8sqrt2minus1
- mov r4, #0x00008A00 ; sin
- orr r4, r4, #0x0000008C ; sinpi8sqrt2
- ;
- mov r5, #0x2 ; i i
- ;
-short_idct4x4llm_v6_scott_loop1 ;
- ldr r10, [r0, #(4*2)] ; i5 | i4 5,4
- ldr r11, [r0, #(12*2)] ; i13 | i12 13,12
- ;
- smulwb r6, r4, r10 ; ((ip[4] * sinpi8sqrt2) >> 16) lt1
- smulwb r7, r3, r11 ; ((ip[12] * cospi8sqrt2minus1) >> 16) lt2
- ;
- smulwb r12, r3, r10 ; ((ip[4] * cospi8sqrt2misu1) >> 16) l2t2
- smulwb r14, r4, r11 ; ((ip[12] * sinpi8sqrt2) >> 16) l2t1
- ;
- add r6, r6, r7 ; partial c1 lt1-lt2
- add r12, r12, r14 ; partial d1 l2t2+l2t1
- ;
- smulwt r14, r4, r10 ; ((ip[5] * sinpi8sqrt2) >> 16) ht1
- smulwt r7, r3, r11 ; ((ip[13] * cospi8sqrt2minus1) >> 16) ht2
- ;
- smulwt r8, r3, r10 ; ((ip[5] * cospi8sqrt2minus1) >> 16) h2t1
- smulwt r9, r4, r11 ; ((ip[13] * sinpi8sqrt2) >> 16) h2t2
- ;
- add r7, r14, r7 ; partial c1_2 ht1+ht2
- sub r8, r8, r9 ; partial d1_2 h2t1-h2t2
- ;
- pkhbt r6, r6, r7, lsl #16 ; partial c1_2 | partial c1_1 pack
- pkhbt r12, r12, r8, lsl #16 ; partial d1_2 | partial d1_1 pack
- ;
- usub16 r6, r6, r10 ; c1_2 | c1_1 c
- uadd16 r12, r12, r11 ; d1_2 | d1_1 d
- ;
- ldr r10, [r0, #0] ; i1 | i0 1,0
- ldr r11, [r0, #(8*2)] ; i9 | i10 9,10
- ;
-;;;;;; add r0, r0, #0x4 ; +4
-;;;;;; add r1, r1, #0x4 ; +4
- ;
- uadd16 r8, r10, r11 ; i1 + i9 | i0 + i8 aka a1 a
- usub16 r9, r10, r11 ; i1 - i9 | i0 - i8 aka b1 b
- ;
- uadd16 r7, r8, r12 ; a1 + d1 pair a+d
- usub16 r14, r8, r12 ; a1 - d1 pair a-d
- ;
- str r7, [r1] ; op[0] = a1 + d1
- str r14, [r1, r2] ; op[pitch*3] = a1 - d1
- ;
- add r0, r0, #0x4 ; op[pitch] = b1 + c1 ++
- add r1, r1, #0x4 ; op[pitch*2] = b1 - c1 ++
- ;
- subs r5, r5, #0x1 ; --
- bne short_idct4x4llm_v6_scott_loop1 ;
- ;
- sub r1, r1, #16 ; reset output ptr
- mov r5, #0x4 ;
- mov r0, r1 ; input = output
- ;
-short_idct4x4llm_v6_scott_loop2 ;
- ;
- subs r5, r5, #0x1 ;
- bne short_idct4x4llm_v6_scott_loop2 ;
- ;
- ldmia sp!, {r4 - r11, pc} ;
- ENDP ;
- ;
-;********************************************************************************
-;********************************************************************************
-;********************************************************************************
-
-;********************************************************************************
-;* void short_idct4x4llm_v6_dual(INT16 * input, INT16 * output, INT32 pitch)
-;* r0 INT16 * input
-;* r1 INT16 * output
-;* r2 INT32 pitch
-;* bench:
-;********************************************************************************
-
-|vp8_short_idct4x4llm_v6_dual| PROC ; cycles in out pit
- ;
- stmdb sp!, {r4-r11, lr} ; backup registers 1 backup
- mov r3, #0x00004E00 ; cos
- orr r3, r3, #0x0000007B ; cospi8sqrt2minus1
- mov r4, #0x00008A00 ; sin
- orr r4, r4, #0x0000008C ; sinpi8sqrt2
- mov r5, #0x2 ; i=2 i
-loop1_dual
- ldr r6, [r0, #(4*2)] ; i5 | i4 5|4
- ldr r12, [r0, #(12*2)] ; i13 | i12 13|12
- ldr r14, [r0, #(8*2)] ; i9 | i8 9|8
-
- smulwt r9, r3, r6 ; (ip[5] * cospi8sqrt2minus1) >> 16 5c
- smulwb r7, r3, r6 ; (ip[4] * cospi8sqrt2minus1) >> 16 4c
- smulwt r10, r4, r6 ; (ip[5] * sinpi8sqrt2) >> 16 5s
- smulwb r8, r4, r6 ; (ip[4] * sinpi8sqrt2) >> 16 4s
- pkhbt r7, r7, r9, lsl #16 ; 5c | 4c
- smulwt r11, r3, r12 ; (ip[13] * cospi8sqrt2minus1) >> 16 13c
- pkhbt r8, r8, r10, lsl #16 ; 5s | 4s
- uadd16 r6, r6, r7 ; 5c+5 | 4c+4
- smulwt r7, r4, r12 ; (ip[13] * sinpi8sqrt2) >> 16 13s
- smulwb r9, r3, r12 ; (ip[12] * cospi8sqrt2minus1) >> 16 12c
- smulwb r10, r4, r12 ; (ip[12] * sinpi8sqrt2) >> 16 12s
- subs r5, r5, #0x1 ; i-- --
- pkhbt r9, r9, r11, lsl #16 ; 13c | 12c
- ldr r11, [r0], #0x4 ; i1 | i0 ++ 1|0
- pkhbt r10, r10, r7, lsl #16 ; 13s | 12s
- uadd16 r7, r12, r9 ; 13c+13 | 12c+12
- usub16 r7, r8, r7 ; c c
- uadd16 r6, r6, r10 ; d d
- uadd16 r10, r11, r14 ; a a
- usub16 r8, r11, r14 ; b b
- uadd16 r9, r10, r6 ; a+d a+d
- usub16 r10, r10, r6 ; a-d a-d
- uadd16 r6, r8, r7 ; b+c b+c
- usub16 r7, r8, r7 ; b-c b-c
- str r6, [r1, r2] ; o5 | o4
- add r6, r2, r2 ; pitch * 2 p2
- str r7, [r1, r6] ; o9 | o8
- add r6, r6, r2 ; pitch * 3 p3
- str r10, [r1, r6] ; o13 | o12
- str r9, [r1], #0x4 ; o1 | o0 ++
- bne loop1_dual ;
- mov r5, #0x2 ; i=2 i
- sub r0, r1, #8 ; reset input/output i/o
-loop2_dual
- ldr r6, [r0, r2] ; i5 | i4 5|4
- ldr r1, [r0] ; i1 | i0 1|0
- ldr r12, [r0, #0x4] ; i3 | i2 3|2
- add r14, r2, #0x4 ; pitch + 2 p+2
- ldr r14, [r0, r14] ; i7 | i6 7|6
- smulwt r9, r3, r6 ; (ip[5] * cospi8sqrt2minus1) >> 16 5c
- smulwt r7, r3, r1 ; (ip[1] * cospi8sqrt2minus1) >> 16 1c
- smulwt r10, r4, r6 ; (ip[5] * sinpi8sqrt2) >> 16 5s
- smulwt r8, r4, r1 ; (ip[1] * sinpi8sqrt2) >> 16 1s
- pkhbt r11, r6, r1, lsl #16 ; i0 | i4 0|4
- pkhbt r7, r9, r7, lsl #16 ; 1c | 5c
- pkhbt r8, r10, r8, lsl #16 ; 1s | 5s = temp1 © tc1
- pkhtb r1, r1, r6, asr #16 ; i1 | i5 1|5
- uadd16 r1, r7, r1 ; 1c+1 | 5c+5 = temp2 (d) td2
- pkhbt r9, r14, r12, lsl #16 ; i2 | i6 2|6
- uadd16 r10, r11, r9 ; a a
- usub16 r9, r11, r9 ; b b
- pkhtb r6, r12, r14, asr #16 ; i3 | i7 3|7
- subs r5, r5, #0x1 ; i-- --
- smulwt r7, r3, r6 ; (ip[3] * cospi8sqrt2minus1) >> 16 3c
- smulwt r11, r4, r6 ; (ip[3] * sinpi8sqrt2) >> 16 3s
- smulwb r12, r3, r6 ; (ip[7] * cospi8sqrt2minus1) >> 16 7c
- smulwb r14, r4, r6 ; (ip[7] * sinpi8sqrt2) >> 16 7s
-
- pkhbt r7, r12, r7, lsl #16 ; 3c | 7c
- pkhbt r11, r14, r11, lsl #16 ; 3s | 7s = temp1 (d) td1
- uadd16 r6, r7, r6 ; 3c+3 | 7c+7 = temp2 (c) tc2
- usub16 r12, r8, r6 ; c (o1 | o5) c
- uadd16 r6, r11, r1 ; d (o3 | o7) d
- uadd16 r7, r10, r6 ; a+d a+d
- mov r8, #0x4 ; set up 4's 4
- orr r8, r8, #0x40000 ; 4|4
- usub16 r6, r10, r6 ; a-d a-d
- uadd16 r6, r6, r8 ; a-d+4 3|7
- uadd16 r7, r7, r8 ; a+d+4 0|4
- uadd16 r10, r9, r12 ; b+c b+c
- usub16 r1, r9, r12 ; b-c b-c
- uadd16 r10, r10, r8 ; b+c+4 1|5
- uadd16 r1, r1, r8 ; b-c+4 2|6
- mov r8, r10, asr #19 ; o1 >> 3
- strh r8, [r0, #2] ; o1
- mov r8, r1, asr #19 ; o2 >> 3
- strh r8, [r0, #4] ; o2
- mov r8, r6, asr #19 ; o3 >> 3
- strh r8, [r0, #6] ; o3
- mov r8, r7, asr #19 ; o0 >> 3
- strh r8, [r0], r2 ; o0 +p
- sxth r10, r10 ;
- mov r8, r10, asr #3 ; o5 >> 3
- strh r8, [r0, #2] ; o5
- sxth r1, r1 ;
- mov r8, r1, asr #3 ; o6 >> 3
- strh r8, [r0, #4] ; o6
- sxth r6, r6 ;
- mov r8, r6, asr #3 ; o7 >> 3
- strh r8, [r0, #6] ; o7
- sxth r7, r7 ;
- mov r8, r7, asr #3 ; o4 >> 3
- strh r8, [r0], r2 ; o4 +p
-;;;;; subs r5, r5, #0x1 ; i-- --
- bne loop2_dual ;
- ;
- ldmia sp!, {r4 - r11, pc} ; replace vars, return restore
- ENDP
-
- END
diff --git a/vp8/common/arm/armv6/iwalsh_v6.asm b/vp8/common/arm/armv6/iwalsh_v6.asm
deleted file mode 100644
index 463bff0..0000000
--- a/vp8/common/arm/armv6/iwalsh_v6.asm
+++ /dev/null
@@ -1,152 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
- EXPORT |vp8_short_inv_walsh4x4_v6|
- EXPORT |vp8_short_inv_walsh4x4_1_v6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
-;short vp8_short_inv_walsh4x4_v6(short *input, short *output)
-|vp8_short_inv_walsh4x4_v6| PROC
-
- stmdb sp!, {r4 - r11, lr}
-
- ldr r2, [r0], #4 ; [1 | 0]
- ldr r3, [r0], #4 ; [3 | 2]
- ldr r4, [r0], #4 ; [5 | 4]
- ldr r5, [r0], #4 ; [7 | 6]
- ldr r6, [r0], #4 ; [9 | 8]
- ldr r7, [r0], #4 ; [11 | 10]
- ldr r8, [r0], #4 ; [13 | 12]
- ldr r9, [r0] ; [15 | 14]
-
- qadd16 r10, r2, r8 ; a1 [1+13 | 0+12]
- qadd16 r11, r4, r6 ; b1 [5+9 | 4+8]
- qsub16 r12, r4, r6 ; c1 [5-9 | 4-8]
- qsub16 lr, r2, r8 ; d1 [1-13 | 0-12]
-
- qadd16 r2, r10, r11 ; a1 + b1 [1 | 0]
- qadd16 r4, r12, lr ; c1 + d1 [5 | 4]
- qsub16 r6, r10, r11 ; a1 - b1 [9 | 8]
- qsub16 r8, lr, r12 ; d1 - c1 [13 | 12]
-
- qadd16 r10, r3, r9 ; a1 [3+15 | 2+14]
- qadd16 r11, r5, r7 ; b1 [7+11 | 6+10]
- qsub16 r12, r5, r7 ; c1 [7-11 | 6-10]
- qsub16 lr, r3, r9 ; d1 [3-15 | 2-14]
-
- qadd16 r3, r10, r11 ; a1 + b1 [3 | 2]
- qadd16 r5, r12, lr ; c1 + d1 [7 | 6]
- qsub16 r7, r10, r11 ; a1 - b1 [11 | 10]
- qsub16 r9, lr, r12 ; d1 - c1 [15 | 14]
-
- ; first transform complete
-
- qsubaddx r10, r2, r3 ; [c1|a1] [1-2 | 0+3]
- qaddsubx r11, r2, r3 ; [b1|d1] [1+2 | 0-3]
- qsubaddx r12, r4, r5 ; [c1|a1] [5-6 | 4+7]
- qaddsubx lr, r4, r5 ; [b1|d1] [5+6 | 4-7]
-
- qaddsubx r2, r10, r11 ; [b2|c2] [c1+d1 | a1-b1]
- qaddsubx r3, r11, r10 ; [a2|d2] [b1+a1 | d1-c1]
- ldr r10, c0x00030003
- qaddsubx r4, r12, lr ; [b2|c2] [c1+d1 | a1-b1]
- qaddsubx r5, lr, r12 ; [a2|d2] [b1+a1 | d1-c1]
-
- qadd16 r2, r2, r10 ; [b2+3|c2+3]
- qadd16 r3, r3, r10 ; [a2+3|d2+3]
- qadd16 r4, r4, r10 ; [b2+3|c2+3]
- qadd16 r5, r5, r10 ; [a2+3|d2+3]
-
- asr r12, r2, #3 ; [1 | x]
- pkhtb r12, r12, r3, asr #19; [1 | 0]
- lsl lr, r3, #16 ; [~3 | x]
- lsl r2, r2, #16 ; [~2 | x]
- asr lr, lr, #3 ; [3 | x]
- pkhtb lr, lr, r2, asr #19 ; [3 | 2]
-
- asr r2, r4, #3 ; [5 | x]
- pkhtb r2, r2, r5, asr #19 ; [5 | 4]
- lsl r3, r5, #16 ; [~7 | x]
- lsl r4, r4, #16 ; [~6 | x]
- asr r3, r3, #3 ; [7 | x]
- pkhtb r3, r3, r4, asr #19 ; [7 | 6]
-
- str r12, [r1], #4
- str lr, [r1], #4
- str r2, [r1], #4
- str r3, [r1], #4
-
- qsubaddx r2, r6, r7 ; [c1|a1] [9-10 | 8+11]
- qaddsubx r3, r6, r7 ; [b1|d1] [9+10 | 8-11]
- qsubaddx r4, r8, r9 ; [c1|a1] [13-14 | 12+15]
- qaddsubx r5, r8, r9 ; [b1|d1] [13+14 | 12-15]
-
- qaddsubx r6, r2, r3 ; [b2|c2] [c1+d1 | a1-b1]
- qaddsubx r7, r3, r2 ; [a2|d2] [b1+a1 | d1-c1]
- qaddsubx r8, r4, r5 ; [b2|c2] [c1+d1 | a1-b1]
- qaddsubx r9, r5, r4 ; [a2|d2] [b1+a1 | d1-c1]
-
- qadd16 r6, r6, r10 ; [b2+3|c2+3]
- qadd16 r7, r7, r10 ; [a2+3|d2+3]
- qadd16 r8, r8, r10 ; [b2+3|c2+3]
- qadd16 r9, r9, r10 ; [a2+3|d2+3]
-
- asr r2, r6, #3 ; [9 | x]
- pkhtb r2, r2, r7, asr #19 ; [9 | 8]
- lsl r3, r7, #16 ; [~11| x]
- lsl r4, r6, #16 ; [~10| x]
- asr r3, r3, #3 ; [11 | x]
- pkhtb r3, r3, r4, asr #19 ; [11 | 10]
-
- asr r4, r8, #3 ; [13 | x]
- pkhtb r4, r4, r9, asr #19 ; [13 | 12]
- lsl r5, r9, #16 ; [~15| x]
- lsl r6, r8, #16 ; [~14| x]
- asr r5, r5, #3 ; [15 | x]
- pkhtb r5, r5, r6, asr #19 ; [15 | 14]
-
- str r2, [r1], #4
- str r3, [r1], #4
- str r4, [r1], #4
- str r5, [r1]
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_short_inv_walsh4x4_v6|
-
-
-;short vp8_short_inv_walsh4x4_1_v6(short *input, short *output)
-|vp8_short_inv_walsh4x4_1_v6| PROC
-
- ldrsh r2, [r0] ; [0]
- add r2, r2, #3 ; [0] + 3
- asr r2, r2, #3 ; a1 ([0]+3) >> 3
- lsl r2, r2, #16 ; [a1 | x]
- orr r2, r2, r2, lsr #16 ; [a1 | a1]
-
- str r2, [r1], #4
- str r2, [r1], #4
- str r2, [r1], #4
- str r2, [r1], #4
- str r2, [r1], #4
- str r2, [r1], #4
- str r2, [r1], #4
- str r2, [r1]
-
- bx lr
- ENDP ; |vp8_short_inv_walsh4x4_1_v6|
-
-; Constant Pool
-c0x00030003 DCD 0x00030003
- END
diff --git a/vp8/common/arm/armv6/loopfilter_v6.asm b/vp8/common/arm/armv6/loopfilter_v6.asm
deleted file mode 100644
index b6417de..0000000
--- a/vp8/common/arm/armv6/loopfilter_v6.asm
+++ /dev/null
@@ -1,1264 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_loop_filter_horizontal_edge_armv6|
- EXPORT |vp8_mbloop_filter_horizontal_edge_armv6|
- EXPORT |vp8_loop_filter_vertical_edge_armv6|
- EXPORT |vp8_mbloop_filter_vertical_edge_armv6|
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
- MACRO
- TRANSPOSE_MATRIX $a0, $a1, $a2, $a3, $b0, $b1, $b2, $b3
- ; input: $a0, $a1, $a2, $a3; output: $b0, $b1, $b2, $b3
- ; a0: 03 02 01 00
- ; a1: 13 12 11 10
- ; a2: 23 22 21 20
- ; a3: 33 32 31 30
- ; b3 b2 b1 b0
-
- uxtb16 $b1, $a1 ; xx 12 xx 10
- uxtb16 $b0, $a0 ; xx 02 xx 00
- uxtb16 $b3, $a3 ; xx 32 xx 30
- uxtb16 $b2, $a2 ; xx 22 xx 20
- orr $b1, $b0, $b1, lsl #8 ; 12 02 10 00
- orr $b3, $b2, $b3, lsl #8 ; 32 22 30 20
-
- uxtb16 $a1, $a1, ror #8 ; xx 13 xx 11
- uxtb16 $a3, $a3, ror #8 ; xx 33 xx 31
- uxtb16 $a0, $a0, ror #8 ; xx 03 xx 01
- uxtb16 $a2, $a2, ror #8 ; xx 23 xx 21
- orr $a0, $a0, $a1, lsl #8 ; 13 03 11 01
- orr $a2, $a2, $a3, lsl #8 ; 33 23 31 21
-
- pkhtb $b2, $b3, $b1, asr #16 ; 32 22 12 02 -- p1
- pkhbt $b0, $b1, $b3, lsl #16 ; 30 20 10 00 -- p3
-
- pkhtb $b3, $a2, $a0, asr #16 ; 33 23 13 03 -- p0
- pkhbt $b1, $a0, $a2, lsl #16 ; 31 21 11 01 -- p2
- MEND
-
-
-src RN r0
-pstep RN r1
-count RN r5
-
-;r0 unsigned char *src_ptr,
-;r1 int src_pixel_step,
-;r2 const char *flimit,
-;r3 const char *limit,
-;stack const char *thresh,
-;stack int count
-
-;Note: All 16 elements in flimit are equal. So, in the code, only one load is needed
-;for flimit. Same way applies to limit and thresh.
-
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-|vp8_loop_filter_horizontal_edge_armv6| PROC
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
- stmdb sp!, {r4 - r11, lr}
-
- sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
- ldr count, [sp, #40] ; count for 8-in-parallel
- ldr r6, [sp, #36] ; load thresh address
- sub sp, sp, #16 ; create temp buffer
-
- ldr r9, [src], pstep ; p3
- ldr r4, [r2], #4 ; flimit
- ldr r10, [src], pstep ; p2
- ldr r2, [r3], #4 ; limit
- ldr r11, [src], pstep ; p1
- uadd8 r4, r4, r4 ; flimit * 2
- ldr r3, [r6], #4 ; thresh
- mov count, count, lsl #1 ; 4-in-parallel
- uadd8 r4, r4, r2 ; flimit * 2 + limit
-
-|Hnext8|
- ; vp8_filter_mask() function
- ; calculate breakout conditions
- ldr r12, [src], pstep ; p0
-
- uqsub8 r6, r9, r10 ; p3 - p2
- uqsub8 r7, r10, r9 ; p2 - p3
- uqsub8 r8, r10, r11 ; p2 - p1
- uqsub8 r10, r11, r10 ; p1 - p2
-
- orr r6, r6, r7 ; abs (p3-p2)
- orr r8, r8, r10 ; abs (p2-p1)
- uqsub8 lr, r6, r2 ; compare to limit. lr: vp8_filter_mask
- uqsub8 r8, r8, r2 ; compare to limit
- uqsub8 r6, r11, r12 ; p1 - p0
- orr lr, lr, r8
- uqsub8 r7, r12, r11 ; p0 - p1
- ldr r9, [src], pstep ; q0
- ldr r10, [src], pstep ; q1
- orr r6, r6, r7 ; abs (p1-p0)
- uqsub8 r7, r6, r2 ; compare to limit
- uqsub8 r8, r6, r3 ; compare to thresh -- save r8 for later
- orr lr, lr, r7
-
- uqsub8 r6, r11, r10 ; p1 - q1
- uqsub8 r7, r10, r11 ; q1 - p1
- uqsub8 r11, r12, r9 ; p0 - q0
- uqsub8 r12, r9, r12 ; q0 - p0
- orr r6, r6, r7 ; abs (p1-q1)
- ldr r7, c0x7F7F7F7F
- orr r12, r11, r12 ; abs (p0-q0)
- ldr r11, [src], pstep ; q2
- uqadd8 r12, r12, r12 ; abs (p0-q0) * 2
- and r6, r7, r6, lsr #1 ; abs (p1-q1) / 2
- uqsub8 r7, r9, r10 ; q0 - q1
- uqadd8 r12, r12, r6 ; abs (p0-q0)*2 + abs (p1-q1)/2
- uqsub8 r6, r10, r9 ; q1 - q0
- uqsub8 r12, r12, r4 ; compare to flimit
- uqsub8 r9, r11, r10 ; q2 - q1
-
- orr lr, lr, r12
-
- ldr r12, [src], pstep ; q3
- uqsub8 r10, r10, r11 ; q1 - q2
- orr r6, r7, r6 ; abs (q1-q0)
- orr r10, r9, r10 ; abs (q2-q1)
- uqsub8 r7, r6, r2 ; compare to limit
- uqsub8 r10, r10, r2 ; compare to limit
- uqsub8 r6, r6, r3 ; compare to thresh -- save r6 for later
- orr lr, lr, r7
- orr lr, lr, r10
-
- uqsub8 r10, r12, r11 ; q3 - q2
- uqsub8 r9, r11, r12 ; q2 - q3
-
- mvn r11, #0 ; r11 == -1
-
- orr r10, r10, r9 ; abs (q3-q2)
- uqsub8 r10, r10, r2 ; compare to limit
-
- mov r12, #0
- orr lr, lr, r10
- sub src, src, pstep, lsl #2
-
- usub8 lr, r12, lr ; use usub8 instead of ssub8
- sel lr, r11, r12 ; filter mask: lr
-
- cmp lr, #0
- beq hskip_filter ; skip filtering
-
- sub src, src, pstep, lsl #1 ; move src pointer down by 6 lines
-
- ;vp8_hevmask() function
- ;calculate high edge variance
- orr r10, r6, r8 ; calculate vp8_hevmask
-
- ldr r7, [src], pstep ; p1
-
- usub8 r10, r12, r10 ; use usub8 instead of ssub8
- sel r6, r12, r11 ; obtain vp8_hevmask: r6
-
- ;vp8_filter() function
- ldr r8, [src], pstep ; p0
- ldr r12, c0x80808080
- ldr r9, [src], pstep ; q0
- ldr r10, [src], pstep ; q1
-
- eor r7, r7, r12 ; p1 offset to convert to a signed value
- eor r8, r8, r12 ; p0 offset to convert to a signed value
- eor r9, r9, r12 ; q0 offset to convert to a signed value
- eor r10, r10, r12 ; q1 offset to convert to a signed value
-
- str r9, [sp] ; store qs0 temporarily
- str r8, [sp, #4] ; store ps0 temporarily
- str r10, [sp, #8] ; store qs1 temporarily
- str r7, [sp, #12] ; store ps1 temporarily
-
- qsub8 r7, r7, r10 ; vp8_signed_char_clamp(ps1-qs1)
- qsub8 r8, r9, r8 ; vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
-
- and r7, r7, r6 ; vp8_filter (r7) &= hev
-
- qadd8 r7, r7, r8
- ldr r9, c0x03030303 ; r9 = 3 --modified for vp8
-
- qadd8 r7, r7, r8
- ldr r10, c0x04040404
-
- qadd8 r7, r7, r8
- and r7, r7, lr ; vp8_filter &= mask;
-
- ;modify code for vp8 -- Filter1 = vp8_filter (r7)
- qadd8 r8 , r7 , r9 ; Filter2 (r8) = vp8_signed_char_clamp(vp8_filter+3)
- qadd8 r7 , r7 , r10 ; vp8_filter = vp8_signed_char_clamp(vp8_filter+4)
-
- mov r9, #0
- shadd8 r8 , r8 , r9 ; Filter2 >>= 3
- shadd8 r7 , r7 , r9 ; vp8_filter >>= 3
- shadd8 r8 , r8 , r9
- shadd8 r7 , r7 , r9
- shadd8 lr , r8 , r9 ; lr: Filter2
- shadd8 r7 , r7 , r9 ; r7: filter
-
- ;usub8 lr, r8, r10 ; s = (s==4)*-1
- ;sel lr, r11, r9
- ;usub8 r8, r10, r8
- ;sel r8, r11, r9
- ;and r8, r8, lr ; -1 for each element that equals 4
-
- ;calculate output
- ;qadd8 lr, r8, r7 ; u = vp8_signed_char_clamp(s + vp8_filter)
-
- ldr r8, [sp] ; load qs0
- ldr r9, [sp, #4] ; load ps0
-
- ldr r10, c0x01010101
-
- qsub8 r8 ,r8, r7 ; u = vp8_signed_char_clamp(qs0 - vp8_filter)
- qadd8 r9, r9, lr ; u = vp8_signed_char_clamp(ps0 + Filter2)
-
- ;end of modification for vp8
-
- mov lr, #0
- sadd8 r7, r7 , r10 ; vp8_filter += 1
- shadd8 r7, r7, lr ; vp8_filter >>= 1
-
- ldr r11, [sp, #12] ; load ps1
- ldr r10, [sp, #8] ; load qs1
-
- bic r7, r7, r6 ; vp8_filter &= ~hev
- sub src, src, pstep, lsl #2
-
- qadd8 r11, r11, r7 ; u = vp8_signed_char_clamp(ps1 + vp8_filter)
- qsub8 r10, r10,r7 ; u = vp8_signed_char_clamp(qs1 - vp8_filter)
-
- eor r11, r11, r12 ; *op1 = u^0x80
- str r11, [src], pstep ; store op1
- eor r9, r9, r12 ; *op0 = u^0x80
- str r9, [src], pstep ; store op0 result
- eor r8, r8, r12 ; *oq0 = u^0x80
- str r8, [src], pstep ; store oq0 result
- eor r10, r10, r12 ; *oq1 = u^0x80
- str r10, [src], pstep ; store oq1
-
- sub src, src, pstep, lsl #1
-
-|hskip_filter|
- add src, src, #4
- sub src, src, pstep, lsl #2
-
- subs count, count, #1
-
- ;pld [src]
- ;pld [src, pstep]
- ;pld [src, pstep, lsl #1]
- ;pld [src, pstep, lsl #2]
- ;pld [src, pstep, lsl #3]
-
- ldrne r9, [src], pstep ; p3
- ldrne r10, [src], pstep ; p2
- ldrne r11, [src], pstep ; p1
-
- bne Hnext8
-
- add sp, sp, #16
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_loop_filter_horizontal_edge_armv6|
-
-
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-|vp8_mbloop_filter_horizontal_edge_armv6| PROC
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
- stmdb sp!, {r4 - r11, lr}
-
- sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
- ldr count, [sp, #40] ; count for 8-in-parallel
- ldr r6, [sp, #36] ; load thresh address
- sub sp, sp, #16 ; create temp buffer
-
- ldr r9, [src], pstep ; p3
- ldr r4, [r2], #4 ; flimit
- ldr r10, [src], pstep ; p2
- ldr r2, [r3], #4 ; limit
- ldr r11, [src], pstep ; p1
- uadd8 r4, r4, r4 ; flimit * 2
- ldr r3, [r6], #4 ; thresh
- mov count, count, lsl #1 ; 4-in-parallel
- uadd8 r4, r4, r2 ; flimit * 2 + limit
-
-|MBHnext8|
-
- ; vp8_filter_mask() function
- ; calculate breakout conditions
- ldr r12, [src], pstep ; p0
-
- uqsub8 r6, r9, r10 ; p3 - p2
- uqsub8 r7, r10, r9 ; p2 - p3
- uqsub8 r8, r10, r11 ; p2 - p1
- uqsub8 r10, r11, r10 ; p1 - p2
-
- orr r6, r6, r7 ; abs (p3-p2)
- orr r8, r8, r10 ; abs (p2-p1)
- uqsub8 lr, r6, r2 ; compare to limit. lr: vp8_filter_mask
- uqsub8 r8, r8, r2 ; compare to limit
-
- uqsub8 r6, r11, r12 ; p1 - p0
- orr lr, lr, r8
- uqsub8 r7, r12, r11 ; p0 - p1
- ldr r9, [src], pstep ; q0
- ldr r10, [src], pstep ; q1
- orr r6, r6, r7 ; abs (p1-p0)
- uqsub8 r7, r6, r2 ; compare to limit
- uqsub8 r8, r6, r3 ; compare to thresh -- save r8 for later
- orr lr, lr, r7
-
- uqsub8 r6, r11, r10 ; p1 - q1
- uqsub8 r7, r10, r11 ; q1 - p1
- uqsub8 r11, r12, r9 ; p0 - q0
- uqsub8 r12, r9, r12 ; q0 - p0
- orr r6, r6, r7 ; abs (p1-q1)
- ldr r7, c0x7F7F7F7F
- orr r12, r11, r12 ; abs (p0-q0)
- ldr r11, [src], pstep ; q2
- uqadd8 r12, r12, r12 ; abs (p0-q0) * 2
- and r6, r7, r6, lsr #1 ; abs (p1-q1) / 2
- uqsub8 r7, r9, r10 ; q0 - q1
- uqadd8 r12, r12, r6 ; abs (p0-q0)*2 + abs (p1-q1)/2
- uqsub8 r6, r10, r9 ; q1 - q0
- uqsub8 r12, r12, r4 ; compare to flimit
- uqsub8 r9, r11, r10 ; q2 - q1
-
- orr lr, lr, r12
-
- ldr r12, [src], pstep ; q3
-
- uqsub8 r10, r10, r11 ; q1 - q2
- orr r6, r7, r6 ; abs (q1-q0)
- orr r10, r9, r10 ; abs (q2-q1)
- uqsub8 r7, r6, r2 ; compare to limit
- uqsub8 r10, r10, r2 ; compare to limit
- uqsub8 r6, r6, r3 ; compare to thresh -- save r6 for later
- orr lr, lr, r7
- orr lr, lr, r10
-
- uqsub8 r10, r12, r11 ; q3 - q2
- uqsub8 r9, r11, r12 ; q2 - q3
-
- mvn r11, #0 ; r11 == -1
-
- orr r10, r10, r9 ; abs (q3-q2)
- uqsub8 r10, r10, r2 ; compare to limit
-
- mov r12, #0
-
- orr lr, lr, r10
-
- usub8 lr, r12, lr ; use usub8 instead of ssub8
- sel lr, r11, r12 ; filter mask: lr
-
- cmp lr, #0
- beq mbhskip_filter ; skip filtering
-
- ;vp8_hevmask() function
- ;calculate high edge variance
- sub src, src, pstep, lsl #2 ; move src pointer down by 6 lines
- sub src, src, pstep, lsl #1
-
- orr r10, r6, r8
- ldr r7, [src], pstep ; p1
-
- usub8 r10, r12, r10
- sel r6, r12, r11 ; hev mask: r6
-
- ;vp8_mbfilter() function
- ;p2, q2 are only needed at the end. Don't need to load them in now.
- ldr r8, [src], pstep ; p0
- ldr r12, c0x80808080
- ldr r9, [src], pstep ; q0
- ldr r10, [src] ; q1
-
- eor r7, r7, r12 ; ps1
- eor r8, r8, r12 ; ps0
- eor r9, r9, r12 ; qs0
- eor r10, r10, r12 ; qs1
-
- qsub8 r12, r9, r8 ; vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
- str r7, [sp, #12] ; store ps1 temporarily
- qsub8 r7, r7, r10 ; vp8_signed_char_clamp(ps1-qs1)
- str r10, [sp, #8] ; store qs1 temporarily
- qadd8 r7, r7, r12
- str r9, [sp] ; store qs0 temporarily
- qadd8 r7, r7, r12
- str r8, [sp, #4] ; store ps0 temporarily
- qadd8 r7, r7, r12 ; vp8_filter: r7
-
- ldr r10, c0x03030303 ; r10 = 3 --modified for vp8
- ldr r9, c0x04040404
-
- and r7, r7, lr ; vp8_filter &= mask (lr is free)
-
- mov r12, r7 ; Filter2: r12
- and r12, r12, r6 ; Filter2 &= hev
-
- ;modify code for vp8
- ;save bottom 3 bits so that we round one side +4 and the other +3
- qadd8 r8 , r12 , r9 ; Filter1 (r8) = vp8_signed_char_clamp(Filter2+4)
- qadd8 r12 , r12 , r10 ; Filter2 (r12) = vp8_signed_char_clamp(Filter2+3)
-
- mov r10, #0
- shadd8 r8 , r8 , r10 ; Filter1 >>= 3
- shadd8 r12 , r12 , r10 ; Filter2 >>= 3
- shadd8 r8 , r8 , r10
- shadd8 r12 , r12 , r10
- shadd8 r8 , r8 , r10 ; r8: Filter1
- shadd8 r12 , r12 , r10 ; r12: Filter2
-
- ldr r9, [sp] ; load qs0
- ldr r11, [sp, #4] ; load ps0
-
- qsub8 r9 , r9, r8 ; qs0 = vp8_signed_char_clamp(qs0 - Filter1)
- qadd8 r11, r11, r12 ; ps0 = vp8_signed_char_clamp(ps0 + Filter2)
-
- ;save bottom 3 bits so that we round one side +4 and the other +3
- ;and r8, r12, r10 ; s = Filter2 & 7 (s: r8)
- ;qadd8 r12 , r12 , r9 ; Filter2 = vp8_signed_char_clamp(Filter2+4)
- ;mov r10, #0
- ;shadd8 r12 , r12 , r10 ; Filter2 >>= 3
- ;usub8 lr, r8, r9 ; s = (s==4)*-1
- ;sel lr, r11, r10
- ;shadd8 r12 , r12 , r10
- ;usub8 r8, r9, r8
- ;sel r8, r11, r10
- ;ldr r9, [sp] ; load qs0
- ;ldr r11, [sp, #4] ; load ps0
- ;shadd8 r12 , r12 , r10
- ;and r8, r8, lr ; -1 for each element that equals 4
- ;qadd8 r10, r8, r12 ; u = vp8_signed_char_clamp(s + Filter2)
- ;qsub8 r9 , r9, r12 ; qs0 = vp8_signed_char_clamp(qs0 - Filter2)
- ;qadd8 r11, r11, r10 ; ps0 = vp8_signed_char_clamp(ps0 + u)
-
- ;end of modification for vp8
-
- bic r12, r7, r6 ; vp8_filter &= ~hev ( r6 is free)
- ;mov r12, r7
-
- ;roughly 3/7th difference across boundary
- mov lr, #0x1b ; 27
- mov r7, #0x3f ; 63
-
- sxtb16 r6, r12
- sxtb16 r10, r12, ror #8
- smlabb r8, r6, lr, r7
- smlatb r6, r6, lr, r7
- smlabb r7, r10, lr, r7
- smultb r10, r10, lr
- ssat r8, #8, r8, asr #7
- ssat r6, #8, r6, asr #7
- add r10, r10, #63
- ssat r7, #8, r7, asr #7
- ssat r10, #8, r10, asr #7
-
- ldr lr, c0x80808080
-
- pkhbt r6, r8, r6, lsl #16
- pkhbt r10, r7, r10, lsl #16
- uxtb16 r6, r6
- uxtb16 r10, r10
-
- sub src, src, pstep
-
- orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
-
- qsub8 r8, r9, r10 ; s = vp8_signed_char_clamp(qs0 - u)
- qadd8 r10, r11, r10 ; s = vp8_signed_char_clamp(ps0 + u)
- eor r8, r8, lr ; *oq0 = s^0x80
- str r8, [src] ; store *oq0
- sub src, src, pstep
- eor r10, r10, lr ; *op0 = s^0x80
- str r10, [src] ; store *op0
-
- ;roughly 2/7th difference across boundary
- mov lr, #0x12 ; 18
- mov r7, #0x3f ; 63
-
- sxtb16 r6, r12
- sxtb16 r10, r12, ror #8
- smlabb r8, r6, lr, r7
- smlatb r6, r6, lr, r7
- smlabb r9, r10, lr, r7
- smlatb r10, r10, lr, r7
- ssat r8, #8, r8, asr #7
- ssat r6, #8, r6, asr #7
- ssat r9, #8, r9, asr #7
- ssat r10, #8, r10, asr #7
-
- ldr lr, c0x80808080
-
- pkhbt r6, r8, r6, lsl #16
- pkhbt r10, r9, r10, lsl #16
-
- ldr r9, [sp, #8] ; load qs1
- ldr r11, [sp, #12] ; load ps1
-
- uxtb16 r6, r6
- uxtb16 r10, r10
-
- sub src, src, pstep
-
- orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
-
- qadd8 r11, r11, r10 ; s = vp8_signed_char_clamp(ps1 + u)
- qsub8 r8, r9, r10 ; s = vp8_signed_char_clamp(qs1 - u)
- eor r11, r11, lr ; *op1 = s^0x80
- str r11, [src], pstep ; store *op1
- eor r8, r8, lr ; *oq1 = s^0x80
- add src, src, pstep, lsl #1
-
- mov r7, #0x3f ; 63
-
- str r8, [src], pstep ; store *oq1
-
- ;roughly 1/7th difference across boundary
- mov lr, #0x9 ; 9
- ldr r9, [src] ; load q2
-
- sxtb16 r6, r12
- sxtb16 r10, r12, ror #8
- smlabb r8, r6, lr, r7
- smlatb r6, r6, lr, r7
- smlabb r12, r10, lr, r7
- smlatb r10, r10, lr, r7
- ssat r8, #8, r8, asr #7
- ssat r6, #8, r6, asr #7
- ssat r12, #8, r12, asr #7
- ssat r10, #8, r10, asr #7
-
- sub src, src, pstep, lsl #2
-
- pkhbt r6, r8, r6, lsl #16
- pkhbt r10, r12, r10, lsl #16
-
- sub src, src, pstep
- ldr lr, c0x80808080
-
- ldr r11, [src] ; load p2
-
- uxtb16 r6, r6
- uxtb16 r10, r10
-
- eor r9, r9, lr
- eor r11, r11, lr
-
- orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
-
- qadd8 r8, r11, r10 ; s = vp8_signed_char_clamp(ps2 + u)
- qsub8 r10, r9, r10 ; s = vp8_signed_char_clamp(qs2 - u)
- eor r8, r8, lr ; *op2 = s^0x80
- str r8, [src], pstep, lsl #2 ; store *op2
- add src, src, pstep
- eor r10, r10, lr ; *oq2 = s^0x80
- str r10, [src], pstep, lsl #1 ; store *oq2
-
-|mbhskip_filter|
- add src, src, #4
- sub src, src, pstep, lsl #3
- subs count, count, #1
-
- ldrne r9, [src], pstep ; p3
- ldrne r10, [src], pstep ; p2
- ldrne r11, [src], pstep ; p1
-
- bne MBHnext8
-
- add sp, sp, #16
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_mbloop_filter_horizontal_edge_armv6|
-
-
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-|vp8_loop_filter_vertical_edge_armv6| PROC
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
- stmdb sp!, {r4 - r11, lr}
-
- sub src, src, #4 ; move src pointer down by 4
- ldr count, [sp, #40] ; count for 8-in-parallel
- ldr r12, [sp, #36] ; load thresh address
- sub sp, sp, #16 ; create temp buffer
-
- ldr r6, [src], pstep ; load source data
- ldr r4, [r2], #4 ; flimit
- ldr r7, [src], pstep
- ldr r2, [r3], #4 ; limit
- ldr r8, [src], pstep
- uadd8 r4, r4, r4 ; flimit * 2
- ldr r3, [r12], #4 ; thresh
- ldr lr, [src], pstep
- mov count, count, lsl #1 ; 4-in-parallel
- uadd8 r4, r4, r2 ; flimit * 2 + limit
-
-|Vnext8|
-
- ; vp8_filter_mask() function
- ; calculate breakout conditions
- ; transpose the source data for 4-in-parallel operation
- TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
-
- uqsub8 r7, r9, r10 ; p3 - p2
- uqsub8 r8, r10, r9 ; p2 - p3
- uqsub8 r9, r10, r11 ; p2 - p1
- uqsub8 r10, r11, r10 ; p1 - p2
- orr r7, r7, r8 ; abs (p3-p2)
- orr r10, r9, r10 ; abs (p2-p1)
- uqsub8 lr, r7, r2 ; compare to limit. lr: vp8_filter_mask
- uqsub8 r10, r10, r2 ; compare to limit
-
- sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
-
- orr lr, lr, r10
-
- uqsub8 r6, r11, r12 ; p1 - p0
- uqsub8 r7, r12, r11 ; p0 - p1
- add src, src, #4 ; move src pointer up by 4
- orr r6, r6, r7 ; abs (p1-p0)
- str r11, [sp, #12] ; save p1
- uqsub8 r10, r6, r2 ; compare to limit
- uqsub8 r11, r6, r3 ; compare to thresh
- orr lr, lr, r10
-
- ; transpose uses 8 regs(r6 - r12 and lr). Need to save reg value now
- ; transpose the source data for 4-in-parallel operation
- ldr r6, [src], pstep ; load source data
- str r11, [sp] ; push r11 to stack
- ldr r7, [src], pstep
- str r12, [sp, #4] ; save current reg before load q0 - q3 data
- ldr r8, [src], pstep
- str lr, [sp, #8]
- ldr lr, [src], pstep
-
- TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
-
- ldr lr, [sp, #8] ; load back (f)limit accumulator
-
- uqsub8 r6, r12, r11 ; q3 - q2
- uqsub8 r7, r11, r12 ; q2 - q3
- uqsub8 r12, r11, r10 ; q2 - q1
- uqsub8 r11, r10, r11 ; q1 - q2
- orr r6, r6, r7 ; abs (q3-q2)
- orr r7, r12, r11 ; abs (q2-q1)
- uqsub8 r6, r6, r2 ; compare to limit
- uqsub8 r7, r7, r2 ; compare to limit
- ldr r11, [sp, #4] ; load back p0
- ldr r12, [sp, #12] ; load back p1
- orr lr, lr, r6
- orr lr, lr, r7
-
- uqsub8 r6, r11, r9 ; p0 - q0
- uqsub8 r7, r9, r11 ; q0 - p0
- uqsub8 r8, r12, r10 ; p1 - q1
- uqsub8 r11, r10, r12 ; q1 - p1
- orr r6, r6, r7 ; abs (p0-q0)
- ldr r7, c0x7F7F7F7F
- orr r8, r8, r11 ; abs (p1-q1)
- uqadd8 r6, r6, r6 ; abs (p0-q0) * 2
- and r8, r7, r8, lsr #1 ; abs (p1-q1) / 2
- uqsub8 r11, r10, r9 ; q1 - q0
- uqadd8 r6, r8, r6 ; abs (p0-q0)*2 + abs (p1-q1)/2
- uqsub8 r12, r9, r10 ; q0 - q1
- uqsub8 r6, r6, r4 ; compare to flimit
-
- orr r9, r11, r12 ; abs (q1-q0)
- uqsub8 r8, r9, r2 ; compare to limit
- uqsub8 r10, r9, r3 ; compare to thresh
- orr lr, lr, r6
- orr lr, lr, r8
-
- mvn r11, #0 ; r11 == -1
- mov r12, #0
-
- usub8 lr, r12, lr
- ldr r9, [sp] ; load the compared result
- sel lr, r11, r12 ; filter mask: lr
-
- cmp lr, #0
- beq vskip_filter ; skip filtering
-
- ;vp8_hevmask() function
- ;calculate high edge variance
-
- sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
-
- orr r9, r9, r10
-
- ldrh r7, [src, #-2]
- ldrh r8, [src], pstep
-
- usub8 r9, r12, r9
- sel r6, r12, r11 ; hev mask: r6
-
- ;vp8_filter() function
- ; load soure data to r6, r11, r12, lr
- ldrh r9, [src, #-2]
- ldrh r10, [src], pstep
-
- pkhbt r12, r7, r8, lsl #16
-
- ldrh r7, [src, #-2]
- ldrh r8, [src], pstep
-
- pkhbt r11, r9, r10, lsl #16
-
- ldrh r9, [src, #-2]
- ldrh r10, [src], pstep
-
- ; Transpose needs 8 regs(r6 - r12, and lr). Save r6 and lr first
- str r6, [sp]
- str lr, [sp, #4]
-
- pkhbt r6, r7, r8, lsl #16
- pkhbt lr, r9, r10, lsl #16
-
- ;transpose r12, r11, r6, lr to r7, r8, r9, r10
- TRANSPOSE_MATRIX r12, r11, r6, lr, r7, r8, r9, r10
-
- ;load back hev_mask r6 and filter_mask lr
- ldr r12, c0x80808080
- ldr r6, [sp]
- ldr lr, [sp, #4]
-
- eor r7, r7, r12 ; p1 offset to convert to a signed value
- eor r8, r8, r12 ; p0 offset to convert to a signed value
- eor r9, r9, r12 ; q0 offset to convert to a signed value
- eor r10, r10, r12 ; q1 offset to convert to a signed value
-
- str r9, [sp] ; store qs0 temporarily
- str r8, [sp, #4] ; store ps0 temporarily
- str r10, [sp, #8] ; store qs1 temporarily
- str r7, [sp, #12] ; store ps1 temporarily
-
- qsub8 r7, r7, r10 ; vp8_signed_char_clamp(ps1-qs1)
- qsub8 r8, r9, r8 ; vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
-
- and r7, r7, r6 ; vp8_filter (r7) &= hev (r7 : filter)
-
- qadd8 r7, r7, r8
- ldr r9, c0x03030303 ; r9 = 3 --modified for vp8
-
- qadd8 r7, r7, r8
- ldr r10, c0x04040404
-
- qadd8 r7, r7, r8
- ;mvn r11, #0 ; r11 == -1
-
- and r7, r7, lr ; vp8_filter &= mask
-
- ;modify code for vp8 -- Filter1 = vp8_filter (r7)
- qadd8 r8 , r7 , r9 ; Filter2 (r8) = vp8_signed_char_clamp(vp8_filter+3)
- qadd8 r7 , r7 , r10 ; vp8_filter = vp8_signed_char_clamp(vp8_filter+4)
-
- mov r9, #0
- shadd8 r8 , r8 , r9 ; Filter2 >>= 3
- shadd8 r7 , r7 , r9 ; vp8_filter >>= 3
- shadd8 r8 , r8 , r9
- shadd8 r7 , r7 , r9
- shadd8 lr , r8 , r9 ; lr: filter2
- shadd8 r7 , r7 , r9 ; r7: filter
-
- ;usub8 lr, r8, r10 ; s = (s==4)*-1
- ;sel lr, r11, r9
- ;usub8 r8, r10, r8
- ;sel r8, r11, r9
- ;and r8, r8, lr ; -1 for each element that equals 4 -- r8: s
-
- ;calculate output
- ;qadd8 lr, r8, r7 ; u = vp8_signed_char_clamp(s + vp8_filter)
-
- ldr r8, [sp] ; load qs0
- ldr r9, [sp, #4] ; load ps0
-
- ldr r10, c0x01010101
-
- qsub8 r8, r8, r7 ; u = vp8_signed_char_clamp(qs0 - vp8_filter)
- qadd8 r9, r9, lr ; u = vp8_signed_char_clamp(ps0 + Filter2)
- ;end of modification for vp8
-
- eor r8, r8, r12
- eor r9, r9, r12
-
- mov lr, #0
-
- sadd8 r7, r7, r10
- shadd8 r7, r7, lr
-
- ldr r10, [sp, #8] ; load qs1
- ldr r11, [sp, #12] ; load ps1
-
- bic r7, r7, r6 ; r7: vp8_filter
-
- qsub8 r10 , r10, r7 ; u = vp8_signed_char_clamp(qs1 - vp8_filter)
- qadd8 r11, r11, r7 ; u = vp8_signed_char_clamp(ps1 + vp8_filter)
- eor r10, r10, r12
- eor r11, r11, r12
-
- sub src, src, pstep, lsl #2
-
- ;we can use TRANSPOSE_MATRIX macro to transpose output - input: q1, q0, p0, p1
- ;output is b0, b1, b2, b3
- ;b0: 03 02 01 00
- ;b1: 13 12 11 10
- ;b2: 23 22 21 20
- ;b3: 33 32 31 30
- ; p1 p0 q0 q1
- ; (a3 a2 a1 a0)
- TRANSPOSE_MATRIX r11, r9, r8, r10, r6, r7, r12, lr
-
- strh r6, [src, #-2] ; store the result
- mov r6, r6, lsr #16
- strh r6, [src], pstep
-
- strh r7, [src, #-2]
- mov r7, r7, lsr #16
- strh r7, [src], pstep
-
- strh r12, [src, #-2]
- mov r12, r12, lsr #16
- strh r12, [src], pstep
-
- strh lr, [src, #-2]
- mov lr, lr, lsr #16
- strh lr, [src], pstep
-
-|vskip_filter|
- sub src, src, #4
- subs count, count, #1
-
- ldrne r6, [src], pstep ; load source data
- ldrne r7, [src], pstep
- ldrne r8, [src], pstep
- ldrne lr, [src], pstep
-
- bne Vnext8
-
- add sp, sp, #16
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_loop_filter_vertical_edge_armv6|
-
-
-
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-|vp8_mbloop_filter_vertical_edge_armv6| PROC
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
- stmdb sp!, {r4 - r11, lr}
-
- sub src, src, #4 ; move src pointer down by 4
- ldr count, [sp, #40] ; count for 8-in-parallel
- ldr r12, [sp, #36] ; load thresh address
- sub sp, sp, #16 ; create temp buffer
-
- ldr r6, [src], pstep ; load source data
- ldr r4, [r2], #4 ; flimit
- ldr r7, [src], pstep
- ldr r2, [r3], #4 ; limit
- ldr r8, [src], pstep
- uadd8 r4, r4, r4 ; flimit * 2
- ldr r3, [r12], #4 ; thresh
- ldr lr, [src], pstep
- mov count, count, lsl #1 ; 4-in-parallel
- uadd8 r4, r4, r2 ; flimit * 2 + limit
-
-|MBVnext8|
- ; vp8_filter_mask() function
- ; calculate breakout conditions
- ; transpose the source data for 4-in-parallel operation
- TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
-
- uqsub8 r7, r9, r10 ; p3 - p2
- uqsub8 r8, r10, r9 ; p2 - p3
- uqsub8 r9, r10, r11 ; p2 - p1
- uqsub8 r10, r11, r10 ; p1 - p2
- orr r7, r7, r8 ; abs (p3-p2)
- orr r10, r9, r10 ; abs (p2-p1)
- uqsub8 lr, r7, r2 ; compare to limit. lr: vp8_filter_mask
- uqsub8 r10, r10, r2 ; compare to limit
-
- sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
-
- orr lr, lr, r10
-
- uqsub8 r6, r11, r12 ; p1 - p0
- uqsub8 r7, r12, r11 ; p0 - p1
- add src, src, #4 ; move src pointer up by 4
- orr r6, r6, r7 ; abs (p1-p0)
- str r11, [sp, #12] ; save p1
- uqsub8 r10, r6, r2 ; compare to limit
- uqsub8 r11, r6, r3 ; compare to thresh
- orr lr, lr, r10
-
- ; transpose uses 8 regs(r6 - r12 and lr). Need to save reg value now
- ; transpose the source data for 4-in-parallel operation
- ldr r6, [src], pstep ; load source data
- str r11, [sp] ; push r11 to stack
- ldr r7, [src], pstep
- str r12, [sp, #4] ; save current reg before load q0 - q3 data
- ldr r8, [src], pstep
- str lr, [sp, #8]
- ldr lr, [src], pstep
-
- TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
-
- ldr lr, [sp, #8] ; load back (f)limit accumulator
-
- uqsub8 r6, r12, r11 ; q3 - q2
- uqsub8 r7, r11, r12 ; q2 - q3
- uqsub8 r12, r11, r10 ; q2 - q1
- uqsub8 r11, r10, r11 ; q1 - q2
- orr r6, r6, r7 ; abs (q3-q2)
- orr r7, r12, r11 ; abs (q2-q1)
- uqsub8 r6, r6, r2 ; compare to limit
- uqsub8 r7, r7, r2 ; compare to limit
- ldr r11, [sp, #4] ; load back p0
- ldr r12, [sp, #12] ; load back p1
- orr lr, lr, r6
- orr lr, lr, r7
-
- uqsub8 r6, r11, r9 ; p0 - q0
- uqsub8 r7, r9, r11 ; q0 - p0
- uqsub8 r8, r12, r10 ; p1 - q1
- uqsub8 r11, r10, r12 ; q1 - p1
- orr r6, r6, r7 ; abs (p0-q0)
- ldr r7, c0x7F7F7F7F
- orr r8, r8, r11 ; abs (p1-q1)
- uqadd8 r6, r6, r6 ; abs (p0-q0) * 2
- and r8, r7, r8, lsr #1 ; abs (p1-q1) / 2
- uqsub8 r11, r10, r9 ; q1 - q0
- uqadd8 r6, r8, r6 ; abs (p0-q0)*2 + abs (p1-q1)/2
- uqsub8 r12, r9, r10 ; q0 - q1
- uqsub8 r6, r6, r4 ; compare to flimit
-
- orr r9, r11, r12 ; abs (q1-q0)
- uqsub8 r8, r9, r2 ; compare to limit
- uqsub8 r10, r9, r3 ; compare to thresh
- orr lr, lr, r6
- orr lr, lr, r8
-
- mvn r11, #0 ; r11 == -1
- mov r12, #0
-
- usub8 lr, r12, lr
- ldr r9, [sp] ; load the compared result
- sel lr, r11, r12 ; filter mask: lr
-
- cmp lr, #0
- beq mbvskip_filter ; skip filtering
-
-
- ;vp8_hevmask() function
- ;calculate high edge variance
-
- sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
-
- orr r9, r9, r10
-
- ldrh r7, [src, #-2]
- ldrh r8, [src], pstep
-
- usub8 r9, r12, r9
- sel r6, r12, r11 ; hev mask: r6
-
-
- ; vp8_mbfilter() function
- ; p2, q2 are only needed at the end. Don't need to load them in now.
- ; Transpose needs 8 regs(r6 - r12, and lr). Save r6 and lr first
- ; load soure data to r6, r11, r12, lr
- ldrh r9, [src, #-2]
- ldrh r10, [src], pstep
-
- pkhbt r12, r7, r8, lsl #16
-
- ldrh r7, [src, #-2]
- ldrh r8, [src], pstep
-
- pkhbt r11, r9, r10, lsl #16
-
- ldrh r9, [src, #-2]
- ldrh r10, [src], pstep
-
- str r6, [sp] ; save r6
- str lr, [sp, #4] ; save lr
-
- pkhbt r6, r7, r8, lsl #16
- pkhbt lr, r9, r10, lsl #16
-
- ;transpose r12, r11, r6, lr to p1, p0, q0, q1
- TRANSPOSE_MATRIX r12, r11, r6, lr, r7, r8, r9, r10
-
- ;load back hev_mask r6 and filter_mask lr
- ldr r12, c0x80808080
- ldr r6, [sp]
- ldr lr, [sp, #4]
-
- eor r7, r7, r12 ; ps1
- eor r8, r8, r12 ; ps0
- eor r9, r9, r12 ; qs0
- eor r10, r10, r12 ; qs1
-
- qsub8 r12, r9, r8 ; vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
- str r7, [sp, #12] ; store ps1 temporarily
- qsub8 r7, r7, r10 ; vp8_signed_char_clamp(ps1-qs1)
- str r10, [sp, #8] ; store qs1 temporarily
- qadd8 r7, r7, r12
- str r9, [sp] ; store qs0 temporarily
- qadd8 r7, r7, r12
- str r8, [sp, #4] ; store ps0 temporarily
- qadd8 r7, r7, r12 ; vp8_filter: r7
-
- ldr r10, c0x03030303 ; r10 = 3 --modified for vp8
- ldr r9, c0x04040404
- ;mvn r11, #0 ; r11 == -1
-
- and r7, r7, lr ; vp8_filter &= mask (lr is free)
-
- mov r12, r7 ; Filter2: r12
- and r12, r12, r6 ; Filter2 &= hev
-
- ;modify code for vp8
- ;save bottom 3 bits so that we round one side +4 and the other +3
- qadd8 r8 , r12 , r9 ; Filter1 (r8) = vp8_signed_char_clamp(Filter2+4)
- qadd8 r12 , r12 , r10 ; Filter2 (r12) = vp8_signed_char_clamp(Filter2+3)
-
- mov r10, #0
- shadd8 r8 , r8 , r10 ; Filter1 >>= 3
- shadd8 r12 , r12 , r10 ; Filter2 >>= 3
- shadd8 r8 , r8 , r10
- shadd8 r12 , r12 , r10
- shadd8 r8 , r8 , r10 ; r8: Filter1
- shadd8 r12 , r12 , r10 ; r12: Filter2
-
- ldr r9, [sp] ; load qs0
- ldr r11, [sp, #4] ; load ps0
-
- qsub8 r9 , r9, r8 ; qs0 = vp8_signed_char_clamp(qs0 - Filter1)
- qadd8 r11, r11, r12 ; ps0 = vp8_signed_char_clamp(ps0 + Filter2)
-
- ;save bottom 3 bits so that we round one side +4 and the other +3
- ;and r8, r12, r10 ; s = Filter2 & 7 (s: r8)
- ;qadd8 r12 , r12 , r9 ; Filter2 = vp8_signed_char_clamp(Filter2+4)
- ;mov r10, #0
- ;shadd8 r12 , r12 , r10 ; Filter2 >>= 3
- ;usub8 lr, r8, r9 ; s = (s==4)*-1
- ;sel lr, r11, r10
- ;shadd8 r12 , r12 , r10
- ;usub8 r8, r9, r8
- ;sel r8, r11, r10
- ;ldr r9, [sp] ; load qs0
- ;ldr r11, [sp, #4] ; load ps0
- ;shadd8 r12 , r12 , r10
- ;and r8, r8, lr ; -1 for each element that equals 4
- ;qadd8 r10, r8, r12 ; u = vp8_signed_char_clamp(s + Filter2)
- ;qsub8 r9 , r9, r12 ; qs0 = vp8_signed_char_clamp(qs0 - Filter2)
- ;qadd8 r11, r11, r10 ; ps0 = vp8_signed_char_clamp(ps0 + u)
-
- ;end of modification for vp8
-
- bic r12, r7, r6 ;vp8_filter &= ~hev ( r6 is free)
- ;mov r12, r7
-
- ;roughly 3/7th difference across boundary
- mov lr, #0x1b ; 27
- mov r7, #0x3f ; 63
-
- sxtb16 r6, r12
- sxtb16 r10, r12, ror #8
- smlabb r8, r6, lr, r7
- smlatb r6, r6, lr, r7
- smlabb r7, r10, lr, r7
- smultb r10, r10, lr
- ssat r8, #8, r8, asr #7
- ssat r6, #8, r6, asr #7
- add r10, r10, #63
- ssat r7, #8, r7, asr #7
- ssat r10, #8, r10, asr #7
-
- ldr lr, c0x80808080
-
- pkhbt r6, r8, r6, lsl #16
- pkhbt r10, r7, r10, lsl #16
- uxtb16 r6, r6
- uxtb16 r10, r10
-
- sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
-
- orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
-
- qsub8 r8, r9, r10 ; s = vp8_signed_char_clamp(qs0 - u)
- qadd8 r10, r11, r10 ; s = vp8_signed_char_clamp(ps0 + u)
- eor r8, r8, lr ; *oq0 = s^0x80
- eor r10, r10, lr ; *op0 = s^0x80
-
- strb r10, [src, #-1] ; store op0 result
- strb r8, [src], pstep ; store oq0 result
- mov r10, r10, lsr #8
- mov r8, r8, lsr #8
- strb r10, [src, #-1]
- strb r8, [src], pstep
- mov r10, r10, lsr #8
- mov r8, r8, lsr #8
- strb r10, [src, #-1]
- strb r8, [src], pstep
- mov r10, r10, lsr #8
- mov r8, r8, lsr #8
- strb r10, [src, #-1]
- strb r8, [src], pstep
-
- ;roughly 2/7th difference across boundary
- mov lr, #0x12 ; 18
- mov r7, #0x3f ; 63
-
- sxtb16 r6, r12
- sxtb16 r10, r12, ror #8
- smlabb r8, r6, lr, r7
- smlatb r6, r6, lr, r7
- smlabb r9, r10, lr, r7
- smlatb r10, r10, lr, r7
- ssat r8, #8, r8, asr #7
- ssat r6, #8, r6, asr #7
- ssat r9, #8, r9, asr #7
- ssat r10, #8, r10, asr #7
-
- sub src, src, pstep, lsl #2 ; move src pointer down by 4 lines
-
- pkhbt r6, r8, r6, lsl #16
- pkhbt r10, r9, r10, lsl #16
-
- ldr r9, [sp, #8] ; load qs1
- ldr r11, [sp, #12] ; load ps1
- ldr lr, c0x80808080
-
- uxtb16 r6, r6
- uxtb16 r10, r10
-
- add src, src, #2
-
- orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
-
- qsub8 r8, r9, r10 ; s = vp8_signed_char_clamp(qs1 - u)
- qadd8 r10, r11, r10 ; s = vp8_signed_char_clamp(ps1 + u)
- eor r8, r8, lr ; *oq1 = s^0x80
- eor r10, r10, lr ; *op1 = s^0x80
-
- ldrb r11, [src, #-5] ; load p2 for 1/7th difference across boundary
- strb r10, [src, #-4] ; store op1
- strb r8, [src, #-1] ; store oq1
- ldrb r9, [src], pstep ; load q2 for 1/7th difference across boundary
-
- mov r10, r10, lsr #8
- mov r8, r8, lsr #8
-
- ldrb r6, [src, #-5]
- strb r10, [src, #-4]
- strb r8, [src, #-1]
- ldrb r7, [src], pstep
-
- mov r10, r10, lsr #8
- mov r8, r8, lsr #8
- orr r11, r11, r6, lsl #8
- orr r9, r9, r7, lsl #8
-
- ldrb r6, [src, #-5]
- strb r10, [src, #-4]
- strb r8, [src, #-1]
- ldrb r7, [src], pstep
-
- mov r10, r10, lsr #8
- mov r8, r8, lsr #8
- orr r11, r11, r6, lsl #16
- orr r9, r9, r7, lsl #16
-
- ldrb r6, [src, #-5]
- strb r10, [src, #-4]
- strb r8, [src, #-1]
- ldrb r7, [src], pstep
- orr r11, r11, r6, lsl #24
- orr r9, r9, r7, lsl #24
-
- ;roughly 1/7th difference across boundary
- eor r9, r9, lr
- eor r11, r11, lr
-
- mov lr, #0x9 ; 9
- mov r7, #0x3f ; 63
-
- sxtb16 r6, r12
- sxtb16 r10, r12, ror #8
- smlabb r8, r6, lr, r7
- smlatb r6, r6, lr, r7
- smlabb r12, r10, lr, r7
- smlatb r10, r10, lr, r7
- ssat r8, #8, r8, asr #7
- ssat r6, #8, r6, asr #7
- ssat r12, #8, r12, asr #7
- ssat r10, #8, r10, asr #7
-
- sub src, src, pstep, lsl #2
-
- pkhbt r6, r8, r6, lsl #16
- pkhbt r10, r12, r10, lsl #16
-
- uxtb16 r6, r6
- uxtb16 r10, r10
-
- ldr lr, c0x80808080
-
- orr r10, r6, r10, lsl #8 ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
-
- qadd8 r8, r11, r10 ; s = vp8_signed_char_clamp(ps2 + u)
- qsub8 r10, r9, r10 ; s = vp8_signed_char_clamp(qs2 - u)
- eor r8, r8, lr ; *op2 = s^0x80
- eor r10, r10, lr ; *oq2 = s^0x80
-
- strb r8, [src, #-5] ; store *op2
- strb r10, [src], pstep ; store *oq2
- mov r8, r8, lsr #8
- mov r10, r10, lsr #8
- strb r8, [src, #-5]
- strb r10, [src], pstep
- mov r8, r8, lsr #8
- mov r10, r10, lsr #8
- strb r8, [src, #-5]
- strb r10, [src], pstep
- mov r8, r8, lsr #8
- mov r10, r10, lsr #8
- strb r8, [src, #-5]
- strb r10, [src], pstep
-
- ;adjust src pointer for next loop
- sub src, src, #2
-
-|mbvskip_filter|
- sub src, src, #4
- subs count, count, #1
-
- ldrne r6, [src], pstep ; load source data
- ldrne r7, [src], pstep
- ldrne r8, [src], pstep
- ldrne lr, [src], pstep
-
- bne MBVnext8
-
- add sp, sp, #16
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_mbloop_filter_vertical_edge_armv6|
-
-; Constant Pool
-c0x80808080 DCD 0x80808080
-c0x03030303 DCD 0x03030303
-c0x04040404 DCD 0x04040404
-c0x01010101 DCD 0x01010101
-c0x7F7F7F7F DCD 0x7F7F7F7F
-
- END
diff --git a/vp8/common/arm/armv6/recon_v6.asm b/vp8/common/arm/armv6/recon_v6.asm
deleted file mode 100644
index 99c7bcf..0000000
--- a/vp8/common/arm/armv6/recon_v6.asm
+++ /dev/null
@@ -1,281 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_recon_b_armv6|
- EXPORT |vp8_recon2b_armv6|
- EXPORT |vp8_recon4b_armv6|
-
- AREA |.text|, CODE, READONLY ; name this block of code
-prd RN r0
-dif RN r1
-dst RN r2
-stride RN r3
-
-;void recon_b(unsigned char *pred_ptr, short *diff_ptr, unsigned char *dst_ptr, int stride)
-; R0 char* pred_ptr
-; R1 short * dif_ptr
-; R2 char * dst_ptr
-; R3 int stride
-
-; Description:
-; Loop through the block adding the Pred and Diff together. Clamp and then
-; store back into the Dst.
-
-; Restrictions :
-; all buffers are expected to be 4 byte aligned coming in and
-; going out.
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-;
-;
-;
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-|vp8_recon_b_armv6| PROC
- stmdb sp!, {r4 - r9, lr}
-
- ;0, 1, 2, 3
- ldr r4, [prd], #16 ; 3 | 2 | 1 | 0
- ldr r6, [dif, #0] ; 1 | 0
- ldr r7, [dif, #4] ; 3 | 2
-
- pkhbt r8, r6, r7, lsl #16 ; 2 | 0
- pkhtb r9, r7, r6, asr #16 ; 3 | 1
-
- uxtab16 r8, r8, r4 ; 2 | 0 + 3 | 2 | 2 | 0
- uxtab16 r9, r9, r4, ror #8 ; 3 | 1 + 0 | 3 | 2 | 1
-
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- add dif, dif, #32
- orr r8, r8, r9, lsl #8
-
- str r8, [dst], stride
-
- ;0, 1, 2, 3
- ldr r4, [prd], #16 ; 3 | 2 | 1 | 0
-;; ldr r6, [dif, #8] ; 1 | 0
-;; ldr r7, [dif, #12] ; 3 | 2
- ldr r6, [dif, #0] ; 1 | 0
- ldr r7, [dif, #4] ; 3 | 2
-
- pkhbt r8, r6, r7, lsl #16 ; 2 | 0
- pkhtb r9, r7, r6, asr #16 ; 3 | 1
-
- uxtab16 r8, r8, r4 ; 2 | 0 + 3 | 2 | 2 | 0
- uxtab16 r9, r9, r4, ror #8 ; 3 | 1 + 0 | 3 | 2 | 1
-
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- add dif, dif, #32
- orr r8, r8, r9, lsl #8
-
- str r8, [dst], stride
-
- ;0, 1, 2, 3
- ldr r4, [prd], #16 ; 3 | 2 | 1 | 0
-;; ldr r6, [dif, #16] ; 1 | 0
-;; ldr r7, [dif, #20] ; 3 | 2
- ldr r6, [dif, #0] ; 1 | 0
- ldr r7, [dif, #4] ; 3 | 2
-
- pkhbt r8, r6, r7, lsl #16 ; 2 | 0
- pkhtb r9, r7, r6, asr #16 ; 3 | 1
-
- uxtab16 r8, r8, r4 ; 2 | 0 + 3 | 2 | 2 | 0
- uxtab16 r9, r9, r4, ror #8 ; 3 | 1 + 0 | 3 | 2 | 1
-
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- add dif, dif, #32
- orr r8, r8, r9, lsl #8
-
- str r8, [dst], stride
-
- ;0, 1, 2, 3
- ldr r4, [prd], #16 ; 3 | 2 | 1 | 0
-;; ldr r6, [dif, #24] ; 1 | 0
-;; ldr r7, [dif, #28] ; 3 | 2
- ldr r6, [dif, #0] ; 1 | 0
- ldr r7, [dif, #4] ; 3 | 2
-
- pkhbt r8, r6, r7, lsl #16 ; 2 | 0
- pkhtb r9, r7, r6, asr #16 ; 3 | 1
-
- uxtab16 r8, r8, r4 ; 2 | 0 + 3 | 2 | 2 | 0
- uxtab16 r9, r9, r4, ror #8 ; 3 | 1 + 0 | 3 | 2 | 1
-
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- orr r8, r8, r9, lsl #8
-
- str r8, [dst], stride
-
- ldmia sp!, {r4 - r9, pc}
-
- ENDP ; |recon_b|
-
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-;
-;
-;
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-; R0 char *pred_ptr
-; R1 short *dif_ptr
-; R2 char *dst_ptr
-; R3 int stride
-|vp8_recon4b_armv6| PROC
- stmdb sp!, {r4 - r9, lr}
-
- mov lr, #4
-
-recon4b_loop
- ;0, 1, 2, 3
- ldr r4, [prd], #4 ; 3 | 2 | 1 | 0
- ldr r6, [dif, #0] ; 1 | 0
- ldr r7, [dif, #4] ; 3 | 2
-
- pkhbt r8, r6, r7, lsl #16 ; 2 | 0
- pkhtb r9, r7, r6, asr #16 ; 3 | 1
-
- uxtab16 r8, r8, r4 ; 2 | 0 + 3 | 2 | 2 | 0
- uxtab16 r9, r9, r4, ror #8 ; 3 | 1 + 0 | 3 | 2 | 1
-
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- orr r8, r8, r9, lsl #8
-
- str r8, [dst]
-
- ;4, 5, 6, 7
- ldr r4, [prd], #4
-;; ldr r6, [dif, #32]
-;; ldr r7, [dif, #36]
- ldr r6, [dif, #8]
- ldr r7, [dif, #12]
-
- pkhbt r8, r6, r7, lsl #16
- pkhtb r9, r7, r6, asr #16
-
- uxtab16 r8, r8, r4
- uxtab16 r9, r9, r4, ror #8
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- orr r8, r8, r9, lsl #8
-
- str r8, [dst, #4]
-
- ;8, 9, 10, 11
- ldr r4, [prd], #4
-;; ldr r6, [dif, #64]
-;; ldr r7, [dif, #68]
- ldr r6, [dif, #16]
- ldr r7, [dif, #20]
-
- pkhbt r8, r6, r7, lsl #16
- pkhtb r9, r7, r6, asr #16
-
- uxtab16 r8, r8, r4
- uxtab16 r9, r9, r4, ror #8
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- orr r8, r8, r9, lsl #8
-
- str r8, [dst, #8]
-
- ;12, 13, 14, 15
- ldr r4, [prd], #4
-;; ldr r6, [dif, #96]
-;; ldr r7, [dif, #100]
- ldr r6, [dif, #24]
- ldr r7, [dif, #28]
-
- pkhbt r8, r6, r7, lsl #16
- pkhtb r9, r7, r6, asr #16
-
- uxtab16 r8, r8, r4
- uxtab16 r9, r9, r4, ror #8
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- orr r8, r8, r9, lsl #8
-
- str r8, [dst, #12]
-
- add dst, dst, stride
-;; add dif, dif, #8
- add dif, dif, #32
-
- subs lr, lr, #1
- bne recon4b_loop
-
- ldmia sp!, {r4 - r9, pc}
-
- ENDP ; |Recon4B|
-
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-;
-;
-;
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-; R0 char *pred_ptr
-; R1 short *dif_ptr
-; R2 char *dst_ptr
-; R3 int stride
-|vp8_recon2b_armv6| PROC
- stmdb sp!, {r4 - r9, lr}
-
- mov lr, #4
-
-recon2b_loop
- ;0, 1, 2, 3
- ldr r4, [prd], #4
- ldr r6, [dif, #0]
- ldr r7, [dif, #4]
-
- pkhbt r8, r6, r7, lsl #16
- pkhtb r9, r7, r6, asr #16
-
- uxtab16 r8, r8, r4
- uxtab16 r9, r9, r4, ror #8
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- orr r8, r8, r9, lsl #8
-
- str r8, [dst]
-
- ;4, 5, 6, 7
- ldr r4, [prd], #4
-;; ldr r6, [dif, #32]
-;; ldr r7, [dif, #36]
- ldr r6, [dif, #8]
- ldr r7, [dif, #12]
-
- pkhbt r8, r6, r7, lsl #16
- pkhtb r9, r7, r6, asr #16
-
- uxtab16 r8, r8, r4
- uxtab16 r9, r9, r4, ror #8
- usat16 r8, #8, r8
- usat16 r9, #8, r9
- orr r8, r8, r9, lsl #8
-
- str r8, [dst, #4]
-
- add dst, dst, stride
-;; add dif, dif, #8
- add dif, dif, #16
-
- subs lr, lr, #1
- bne recon2b_loop
-
- ldmia sp!, {r4 - r9, pc}
-
- ENDP ; |Recon2B|
-
- END
diff --git a/vp8/common/arm/armv6/simpleloopfilter_v6.asm b/vp8/common/arm/armv6/simpleloopfilter_v6.asm
deleted file mode 100644
index 0137120..0000000
--- a/vp8/common/arm/armv6/simpleloopfilter_v6.asm
+++ /dev/null
@@ -1,287 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_loop_filter_simple_horizontal_edge_armv6|
- EXPORT |vp8_loop_filter_simple_vertical_edge_armv6|
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
- MACRO
- TRANSPOSE_MATRIX $a0, $a1, $a2, $a3, $b0, $b1, $b2, $b3
- ; input: $a0, $a1, $a2, $a3; output: $b0, $b1, $b2, $b3
- ; a0: 03 02 01 00
- ; a1: 13 12 11 10
- ; a2: 23 22 21 20
- ; a3: 33 32 31 30
- ; b3 b2 b1 b0
-
- uxtb16 $b1, $a1 ; xx 12 xx 10
- uxtb16 $b0, $a0 ; xx 02 xx 00
- uxtb16 $b3, $a3 ; xx 32 xx 30
- uxtb16 $b2, $a2 ; xx 22 xx 20
- orr $b1, $b0, $b1, lsl #8 ; 12 02 10 00
- orr $b3, $b2, $b3, lsl #8 ; 32 22 30 20
-
- uxtb16 $a1, $a1, ror #8 ; xx 13 xx 11
- uxtb16 $a3, $a3, ror #8 ; xx 33 xx 31
- uxtb16 $a0, $a0, ror #8 ; xx 03 xx 01
- uxtb16 $a2, $a2, ror #8 ; xx 23 xx 21
- orr $a0, $a0, $a1, lsl #8 ; 13 03 11 01
- orr $a2, $a2, $a3, lsl #8 ; 33 23 31 21
-
- pkhtb $b2, $b3, $b1, asr #16 ; 32 22 12 02 -- p1
- pkhbt $b0, $b1, $b3, lsl #16 ; 30 20 10 00 -- p3
-
- pkhtb $b3, $a2, $a0, asr #16 ; 33 23 13 03 -- p0
- pkhbt $b1, $a0, $a2, lsl #16 ; 31 21 11 01 -- p2
- MEND
-
-
-src RN r0
-pstep RN r1
-
-;r0 unsigned char *src_ptr,
-;r1 int src_pixel_step,
-;r2 const char *flimit,
-;r3 const char *limit,
-;stack const char *thresh,
-;stack int count
-
-; All 16 elements in flimit are equal. So, in the code, only one load is needed
-; for flimit. Same applies to limit. thresh is not used in simple looopfilter
-
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-|vp8_loop_filter_simple_horizontal_edge_armv6| PROC
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
- stmdb sp!, {r4 - r11, lr}
-
- ldr r12, [r3] ; limit
- ldr r3, [src, -pstep, lsl #1] ; p1
- ldr r4, [src, -pstep] ; p0
- ldr r5, [src] ; q0
- ldr r6, [src, pstep] ; q1
- ldr r7, [r2] ; flimit
- ldr r2, c0x80808080
- ldr r9, [sp, #40] ; count for 8-in-parallel
- uadd8 r7, r7, r7 ; flimit * 2
- mov r9, r9, lsl #1 ; double the count. we're doing 4 at a time
- uadd8 r12, r7, r12 ; flimit * 2 + limit
- mov lr, #0 ; need 0 in a couple places
-
-|simple_hnext8|
- ; vp8_simple_filter_mask()
-
- uqsub8 r7, r3, r6 ; p1 - q1
- uqsub8 r8, r6, r3 ; q1 - p1
- uqsub8 r10, r4, r5 ; p0 - q0
- uqsub8 r11, r5, r4 ; q0 - p0
- orr r8, r8, r7 ; abs(p1 - q1)
- orr r10, r10, r11 ; abs(p0 - q0)
- uqadd8 r10, r10, r10 ; abs(p0 - q0) * 2
- uhadd8 r8, r8, lr ; abs(p1 - q2) >> 1
- uqadd8 r10, r10, r8 ; abs(p0 - q0)*2 + abs(p1 - q1)/2
- mvn r8, #0
- usub8 r10, r12, r10 ; compare to flimit. usub8 sets GE flags
- sel r10, r8, lr ; filter mask: F or 0
- cmp r10, #0
- beq simple_hskip_filter ; skip filtering if all masks are 0x00
-
- ;vp8_simple_filter()
-
- eor r3, r3, r2 ; p1 offset to convert to a signed value
- eor r6, r6, r2 ; q1 offset to convert to a signed value
- eor r4, r4, r2 ; p0 offset to convert to a signed value
- eor r5, r5, r2 ; q0 offset to convert to a signed value
-
- qsub8 r3, r3, r6 ; vp8_filter = p1 - q1
- qsub8 r6, r5, r4 ; q0 - p0
- qadd8 r3, r3, r6 ; += q0 - p0
- ldr r7, c0x04040404
- qadd8 r3, r3, r6 ; += q0 - p0
- ldr r8, c0x03030303
- qadd8 r3, r3, r6 ; vp8_filter = p1-q1 + 3*(q0-p0))
- ;STALL
- and r3, r3, r10 ; vp8_filter &= mask
-
- qadd8 r7 , r3 , r7 ; Filter1 = vp8_filter + 4
- qadd8 r8 , r3 , r8 ; Filter2 = vp8_filter + 3
-
- shadd8 r7 , r7 , lr
- shadd8 r8 , r8 , lr
- shadd8 r7 , r7 , lr
- shadd8 r8 , r8 , lr
- shadd8 r7 , r7 , lr ; Filter1 >>= 3
- shadd8 r8 , r8 , lr ; Filter2 >>= 3
-
- qsub8 r5 ,r5, r7 ; u = q0 - Filter1
- qadd8 r4, r4, r8 ; u = p0 + Filter2
- eor r5, r5, r2 ; *oq0 = u^0x80
- str r5, [src] ; store oq0 result
- eor r4, r4, r2 ; *op0 = u^0x80
- str r4, [src, -pstep] ; store op0 result
-
-|simple_hskip_filter|
- subs r9, r9, #1
- addne src, src, #4 ; next row
-
- ldrne r3, [src, -pstep, lsl #1] ; p1
- ldrne r4, [src, -pstep] ; p0
- ldrne r5, [src] ; q0
- ldrne r6, [src, pstep] ; q1
-
- bne simple_hnext8
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_loop_filter_simple_horizontal_edge_armv6|
-
-
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-|vp8_loop_filter_simple_vertical_edge_armv6| PROC
-;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
- stmdb sp!, {r4 - r11, lr}
-
- ldr r12, [r2] ; r12: flimit
- ldr r2, c0x80808080
- ldr r7, [r3] ; limit
-
- ; load soure data to r7, r8, r9, r10
- ldrh r3, [src, #-2]
- ldrh r4, [src], pstep
- uadd8 r12, r12, r12 ; flimit * 2
-
- ldrh r5, [src, #-2]
- ldrh r6, [src], pstep
- uadd8 r12, r12, r7 ; flimit * 2 + limit
-
- pkhbt r7, r3, r4, lsl #16
-
- ldrh r3, [src, #-2]
- ldrh r4, [src], pstep
- ldr r11, [sp, #40] ; count (r11) for 8-in-parallel
-
- pkhbt r8, r5, r6, lsl #16
-
- ldrh r5, [src, #-2]
- ldrh r6, [src], pstep
- mov r11, r11, lsl #1 ; 4-in-parallel
-
-|simple_vnext8|
- ; vp8_simple_filter_mask() function
- pkhbt r9, r3, r4, lsl #16
- pkhbt r10, r5, r6, lsl #16
-
- ;transpose r7, r8, r9, r10 to r3, r4, r5, r6
- TRANSPOSE_MATRIX r7, r8, r9, r10, r3, r4, r5, r6
-
- uqsub8 r7, r3, r6 ; p1 - q1
- uqsub8 r8, r6, r3 ; q1 - p1
- uqsub8 r9, r4, r5 ; p0 - q0
- uqsub8 r10, r5, r4 ; q0 - p0
- orr r7, r7, r8 ; abs(p1 - q1)
- orr r9, r9, r10 ; abs(p0 - q0)
- mov r8, #0
- uqadd8 r9, r9, r9 ; abs(p0 - q0) * 2
- uhadd8 r7, r7, r8 ; abs(p1 - q1) / 2
- uqadd8 r7, r7, r9 ; abs(p0 - q0)*2 + abs(p1 - q1)/2
- mvn r10, #0 ; r10 == -1
-
- usub8 r7, r12, r7 ; compare to flimit
- sel lr, r10, r8 ; filter mask
-
- cmp lr, #0
- beq simple_vskip_filter ; skip filtering
-
- ;vp8_simple_filter() function
- eor r3, r3, r2 ; p1 offset to convert to a signed value
- eor r6, r6, r2 ; q1 offset to convert to a signed value
- eor r4, r4, r2 ; p0 offset to convert to a signed value
- eor r5, r5, r2 ; q0 offset to convert to a signed value
-
- qsub8 r3, r3, r6 ; vp8_filter = p1 - q1
- qsub8 r6, r5, r4 ; q0 - p0
-
- qadd8 r3, r3, r6 ; vp8_filter += q0 - p0
- ldr r9, c0x03030303 ; r9 = 3
-
- qadd8 r3, r3, r6 ; vp8_filter += q0 - p0
- ldr r7, c0x04040404
-
- qadd8 r3, r3, r6 ; vp8_filter = p1-q1 + 3*(q0-p0))
- ;STALL
- and r3, r3, lr ; vp8_filter &= mask
-
- qadd8 r9 , r3 , r9 ; Filter2 = vp8_filter + 3
- qadd8 r3 , r3 , r7 ; Filter1 = vp8_filter + 4
-
- shadd8 r9 , r9 , r8
- shadd8 r3 , r3 , r8
- shadd8 r9 , r9 , r8
- shadd8 r3 , r3 , r8
- shadd8 r9 , r9 , r8 ; Filter2 >>= 3
- shadd8 r3 , r3 , r8 ; Filter1 >>= 3
-
- ;calculate output
- sub src, src, pstep, lsl #2
-
- qadd8 r4, r4, r9 ; u = p0 + Filter2
- qsub8 r5, r5, r3 ; u = q0 - Filter1
- eor r4, r4, r2 ; *op0 = u^0x80
- eor r5, r5, r2 ; *oq0 = u^0x80
-
- strb r4, [src, #-1] ; store the result
- mov r4, r4, lsr #8
- strb r5, [src], pstep
- mov r5, r5, lsr #8
-
- strb r4, [src, #-1]
- mov r4, r4, lsr #8
- strb r5, [src], pstep
- mov r5, r5, lsr #8
-
- strb r4, [src, #-1]
- mov r4, r4, lsr #8
- strb r5, [src], pstep
- mov r5, r5, lsr #8
-
- strb r4, [src, #-1]
- strb r5, [src], pstep
-
-|simple_vskip_filter|
- subs r11, r11, #1
-
- ; load soure data to r7, r8, r9, r10
- ldrneh r3, [src, #-2]
- ldrneh r4, [src], pstep
-
- ldrneh r5, [src, #-2]
- ldrneh r6, [src], pstep
-
- pkhbt r7, r3, r4, lsl #16
-
- ldrneh r3, [src, #-2]
- ldrneh r4, [src], pstep
-
- pkhbt r8, r5, r6, lsl #16
-
- ldrneh r5, [src, #-2]
- ldrneh r6, [src], pstep
-
- bne simple_vnext8
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_loop_filter_simple_vertical_edge_armv6|
-
-; Constant Pool
-c0x80808080 DCD 0x80808080
-c0x03030303 DCD 0x03030303
-c0x04040404 DCD 0x04040404
-
- END
diff --git a/vp8/common/arm/armv6/sixtappredict8x4_v6.asm b/vp8/common/arm/armv6/sixtappredict8x4_v6.asm
deleted file mode 100644
index 17d70c3..0000000
--- a/vp8/common/arm/armv6/sixtappredict8x4_v6.asm
+++ /dev/null
@@ -1,268 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_sixtap_predict8x4_armv6|
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
-;-----------------
-;One word each is reserved. Label filter_coeff can be used to access the data.
-;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
-filter8_coeff
- DCD 0x00000000, 0x00000080, 0x00000000, 0x00000000
- DCD 0xfffa0000, 0x000c007b, 0x0000ffff, 0x00000000
- DCD 0xfff50002, 0x0024006c, 0x0001fff8, 0x00000000
- DCD 0xfff70000, 0x0032005d, 0x0000fffa, 0x00000000
- DCD 0xfff00003, 0x004d004d, 0x0003fff0, 0x00000000
- DCD 0xfffa0000, 0x005d0032, 0x0000fff7, 0x00000000
- DCD 0xfff80001, 0x006c0024, 0x0002fff5, 0x00000000
- DCD 0xffff0000, 0x007b000c, 0x0000fffa, 0x00000000
-
- ;DCD 0, 0, 128, 0, 0, 0
- ;DCD 0, -6, 123, 12, -1, 0
- ;DCD 2, -11, 108, 36, -8, 1
- ;DCD 0, -9, 93, 50, -6, 0
- ;DCD 3, -16, 77, 77, -16, 3
- ;DCD 0, -6, 50, 93, -9, 0
- ;DCD 1, -8, 36, 108, -11, 2
- ;DCD 0, -1, 12, 123, -6, 0
-
-;-------------------------------------
-; r0 unsigned char *src_ptr,
-; r1 int src_pixels_per_line,
-; r2 int xoffset,
-; r3 int yoffset,
-; stack unsigned char *dst_ptr,
-; stack int dst_pitch
-;-------------------------------------
-;note: In first pass, store the result in transpose(8linesx9columns) on stack. Temporary stack size is 184.
-;Line width is 20 that is 9 short data plus 2 to make it 4bytes aligned. In second pass, load data from stack,
-;and the result is stored in transpose.
-|vp8_sixtap_predict8x4_armv6| PROC
- stmdb sp!, {r4 - r11, lr}
- str r3, [sp, #-184]! ;reserve space on stack for temporary storage, store yoffset
-
- cmp r2, #0 ;skip first_pass filter if xoffset=0
- add lr, sp, #4 ;point to temporary buffer
- beq skip_firstpass_filter
-
-;first-pass filter
- adr r12, filter8_coeff
- sub r0, r0, r1, lsl #1
-
- add r2, r12, r2, lsl #4 ;calculate filter location
- add r0, r0, #3 ;adjust src only for loading convinience
-
- ldr r3, [r2] ; load up packed filter coefficients
- ldr r4, [r2, #4]
- ldr r5, [r2, #8]
-
- mov r2, #0x90000 ; height=9 is top part of counter
-
- sub r1, r1, #8
-
-|first_pass_hloop_v6|
- ldrb r6, [r0, #-5] ; load source data
- ldrb r7, [r0, #-4]
- ldrb r8, [r0, #-3]
- ldrb r9, [r0, #-2]
- ldrb r10, [r0, #-1]
-
- orr r2, r2, #0x4 ; construct loop counter. width=8=4x2
-
- pkhbt r6, r6, r7, lsl #16 ; r7 | r6
- pkhbt r7, r7, r8, lsl #16 ; r8 | r7
-
- pkhbt r8, r8, r9, lsl #16 ; r9 | r8
- pkhbt r9, r9, r10, lsl #16 ; r10 | r9
-
-|first_pass_wloop_v6|
- smuad r11, r6, r3 ; vp8_filter[0], vp8_filter[1]
- smuad r12, r7, r3
-
- ldrb r6, [r0], #1
-
- smlad r11, r8, r4, r11 ; vp8_filter[2], vp8_filter[3]
- ldrb r7, [r0], #1
- smlad r12, r9, r4, r12
-
- pkhbt r10, r10, r6, lsl #16 ; r10 | r9
- pkhbt r6, r6, r7, lsl #16 ; r11 | r10
- smlad r11, r10, r5, r11 ; vp8_filter[4], vp8_filter[5]
- smlad r12, r6, r5, r12
-
- sub r2, r2, #1
-
- add r11, r11, #0x40 ; round_shift_and_clamp
- tst r2, #0xff ; test loop counter
- usat r11, #8, r11, asr #7
- add r12, r12, #0x40
- strh r11, [lr], #20 ; result is transposed and stored, which
- usat r12, #8, r12, asr #7
-
- strh r12, [lr], #20
-
- movne r11, r6
- movne r12, r7
-
- movne r6, r8
- movne r7, r9
- movne r8, r10
- movne r9, r11
- movne r10, r12
-
- bne first_pass_wloop_v6
-
- ;;add r9, ppl, #30 ; attempt to load 2 adjacent cache lines
- ;;IF ARCHITECTURE=6
- ;pld [src, ppl]
- ;;pld [src, r9]
- ;;ENDIF
-
- subs r2, r2, #0x10000
-
- sub lr, lr, #158
-
- add r0, r0, r1 ; move to next input line
-
- bne first_pass_hloop_v6
-
-;second pass filter
-secondpass_filter
- ldr r3, [sp], #4 ; load back yoffset
- ldr r0, [sp, #216] ; load dst address from stack 180+36
- ldr r1, [sp, #220] ; load dst stride from stack 180+40
-
- cmp r3, #0
- beq skip_secondpass_filter
-
- adr r12, filter8_coeff
- add lr, r12, r3, lsl #4 ;calculate filter location
-
- mov r2, #0x00080000
-
- ldr r3, [lr] ; load up packed filter coefficients
- ldr r4, [lr, #4]
- ldr r5, [lr, #8]
-
- pkhbt r12, r4, r3 ; pack the filter differently
- pkhbt r11, r5, r4
-
-second_pass_hloop_v6
- ldr r6, [sp] ; load the data
- ldr r7, [sp, #4]
-
- orr r2, r2, #2 ; loop counter
-
-second_pass_wloop_v6
- smuad lr, r3, r6 ; apply filter
- smulbt r10, r3, r6
-
- ldr r8, [sp, #8]
-
- smlad lr, r4, r7, lr
- smladx r10, r12, r7, r10
-
- ldrh r9, [sp, #12]
-
- smlad lr, r5, r8, lr
- smladx r10, r11, r8, r10
-
- add sp, sp, #4
- smlatb r10, r5, r9, r10
-
- sub r2, r2, #1
-
- add lr, lr, #0x40 ; round_shift_and_clamp
- tst r2, #0xff
- usat lr, #8, lr, asr #7
- add r10, r10, #0x40
- strb lr, [r0], r1 ; the result is transposed back and stored
- usat r10, #8, r10, asr #7
-
- strb r10, [r0],r1
-
- movne r6, r7
- movne r7, r8
-
- bne second_pass_wloop_v6
-
- subs r2, r2, #0x10000
- add sp, sp, #12 ; updata src for next loop (20-8)
- sub r0, r0, r1, lsl #2
- add r0, r0, #1
-
- bne second_pass_hloop_v6
-
- add sp, sp, #20
- ldmia sp!, {r4 - r11, pc}
-
-;--------------------
-skip_firstpass_filter
- sub r0, r0, r1, lsl #1
- sub r1, r1, #8
- mov r2, #9
-
-skip_firstpass_hloop
- ldrb r4, [r0], #1 ; load data
- subs r2, r2, #1
- ldrb r5, [r0], #1
- strh r4, [lr], #20 ; store it to immediate buffer
- ldrb r6, [r0], #1 ; load data
- strh r5, [lr], #20
- ldrb r7, [r0], #1
- strh r6, [lr], #20
- ldrb r8, [r0], #1
- strh r7, [lr], #20
- ldrb r9, [r0], #1
- strh r8, [lr], #20
- ldrb r10, [r0], #1
- strh r9, [lr], #20
- ldrb r11, [r0], #1
- strh r10, [lr], #20
- add r0, r0, r1 ; move to next input line
- strh r11, [lr], #20
-
- sub lr, lr, #158 ; move over to next column
- bne skip_firstpass_hloop
-
- b secondpass_filter
-
-;--------------------
-skip_secondpass_filter
- mov r2, #8
- add sp, sp, #4 ;start from src[0] instead of src[-2]
-
-skip_secondpass_hloop
- ldr r6, [sp], #4
- subs r2, r2, #1
- ldr r8, [sp], #4
-
- mov r7, r6, lsr #16 ; unpack
- strb r6, [r0], r1
- mov r9, r8, lsr #16
- strb r7, [r0], r1
- add sp, sp, #12 ; 20-8
- strb r8, [r0], r1
- strb r9, [r0], r1
-
- sub r0, r0, r1, lsl #2
- add r0, r0, #1
-
- bne skip_secondpass_hloop
-
- add sp, sp, #16 ; 180 - (160 +4)
-
- ldmia sp!, {r4 - r11, pc}
-
- ENDP
-
- END
diff --git a/vp8/common/arm/bilinearfilter_arm.c b/vp8/common/arm/bilinearfilter_arm.c
deleted file mode 100644
index 7340e20..0000000
--- a/vp8/common/arm/bilinearfilter_arm.c
+++ /dev/null
@@ -1,113 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <math.h>
-#include "vp8/common/filter.h"
-#include "vp8/common/subpixel.h"
-#include "bilinearfilter_arm.h"
-
-void vp8_filter_block2d_bil_armv6
-(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- unsigned int src_pitch,
- unsigned int dst_pitch,
- const short *HFilter,
- const short *VFilter,
- int Width,
- int Height
-)
-{
- unsigned short FData[36*16]; /* Temp data buffer used in filtering */
-
- /* First filter 1-D horizontally... */
- vp8_filter_block2d_bil_first_pass_armv6(src_ptr, FData, src_pitch, Height + 1, Width, HFilter);
-
- /* then 1-D vertically... */
- vp8_filter_block2d_bil_second_pass_armv6(FData, dst_ptr, dst_pitch, Height, Width, VFilter);
-}
-
-
-void vp8_bilinear_predict4x4_armv6
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 4, 4);
-}
-
-void vp8_bilinear_predict8x8_armv6
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 8);
-}
-
-void vp8_bilinear_predict8x4_armv6
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 4);
-}
-
-void vp8_bilinear_predict16x16_armv6
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 16, 16);
-}
diff --git a/vp8/common/arm/filter_arm.c b/vp8/common/arm/filter_arm.c
deleted file mode 100644
index fe3c5a5..0000000
--- a/vp8/common/arm/filter_arm.c
+++ /dev/null
@@ -1,199 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include <math.h>
-#include "vp8/common/filter.h"
-#include "vp8/common/subpixel.h"
-#include "vpx_ports/mem.h"
-
-extern void vp8_filter_block2d_first_pass_armv6
-(
- unsigned char *src_ptr,
- short *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int output_width,
- unsigned int output_height,
- const short *vp8_filter
-);
-
-extern void vp8_filter_block2d_second_pass_armv6
-(
- short *src_ptr,
- unsigned char *output_ptr,
- unsigned int output_pitch,
- unsigned int cnt,
- const short *vp8_filter
-);
-
-extern void vp8_filter4_block2d_second_pass_armv6
-(
- short *src_ptr,
- unsigned char *output_ptr,
- unsigned int output_pitch,
- unsigned int cnt,
- const short *vp8_filter
-);
-
-extern void vp8_filter_block2d_first_pass_only_armv6
-(
- unsigned char *src_ptr,
- unsigned char *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int cnt,
- unsigned int output_pitch,
- const short *vp8_filter
-);
-
-
-extern void vp8_filter_block2d_second_pass_only_armv6
-(
- unsigned char *src_ptr,
- unsigned char *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int cnt,
- unsigned int output_pitch,
- const short *vp8_filter
-);
-
-#if HAVE_ARMV6
-void vp8_sixtap_predict_armv6
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
- DECLARE_ALIGNED_ARRAY(4, short, FData, 12*4); /* Temp data buffer used in filtering */
-
-
- HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
- VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
-
- /* Vfilter is null. First pass only */
- if (xoffset && !yoffset)
- {
- /*vp8_filter_block2d_first_pass_armv6 ( src_ptr, FData+2, src_pixels_per_line, 4, 4, HFilter );
- vp8_filter_block2d_second_pass_armv6 ( FData+2, dst_ptr, dst_pitch, 4, VFilter );*/
-
- vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, HFilter);
- }
- /* Hfilter is null. Second pass only */
- else if (!xoffset && yoffset)
- {
- vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, VFilter);
- }
- else
- {
- /* Vfilter is a 4 tap filter */
- if (yoffset & 0x1)
- {
- vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 4, 7, HFilter);
- vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
- }
- /* Vfilter is 6 tap filter */
- else
- {
- vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 4, 9, HFilter);
- vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
- }
- }
-}
-
-void vp8_sixtap_predict8x8_armv6
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
- DECLARE_ALIGNED_ARRAY(4, short, FData, 16*8); /* Temp data buffer used in filtering */
-
- HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
- VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
-
- if (xoffset && !yoffset)
- {
- vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, HFilter);
- }
- /* Hfilter is null. Second pass only */
- else if (!xoffset && yoffset)
- {
- vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, VFilter);
- }
- else
- {
- if (yoffset & 0x1)
- {
- vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 8, 11, HFilter);
- vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
- }
- else
- {
- vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 8, 13, HFilter);
- vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
- }
- }
-}
-
-
-void vp8_sixtap_predict16x16_armv6
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
- DECLARE_ALIGNED_ARRAY(4, short, FData, 24*16); /* Temp data buffer used in filtering */
-
- HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
- VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
-
- if (xoffset && !yoffset)
- {
- vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, HFilter);
- }
- /* Hfilter is null. Second pass only */
- else if (!xoffset && yoffset)
- {
- vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, VFilter);
- }
- else
- {
- if (yoffset & 0x1)
- {
- vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 16, 19, HFilter);
- vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
- }
- else
- {
- vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 16, 21, HFilter);
- vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
- }
- }
-
-}
-#endif
diff --git a/vp8/common/arm/idct_arm.h b/vp8/common/arm/idct_arm.h
deleted file mode 100644
index 8b8d179..0000000
--- a/vp8/common/arm/idct_arm.h
+++ /dev/null
@@ -1,65 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef IDCT_ARM_H
-#define IDCT_ARM_H
-
-#if HAVE_ARMV6
-extern prototype_idct(vp8_short_idct4x4llm_1_v6);
-extern prototype_idct(vp8_short_idct4x4llm_v6_dual);
-extern prototype_idct_scalar_add(vp8_dc_only_idct_add_v6);
-extern prototype_second_order(vp8_short_inv_walsh4x4_1_v6);
-extern prototype_second_order(vp8_short_inv_walsh4x4_v6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_idct_idct1
-#define vp8_idct_idct1 vp8_short_idct4x4llm_1_v6
-
-#undef vp8_idct_idct16
-#define vp8_idct_idct16 vp8_short_idct4x4llm_v6_dual
-
-#undef vp8_idct_idct1_scalar_add
-#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_v6
-
-#undef vp8_idct_iwalsh1
-#define vp8_idct_iwalsh1 vp8_short_inv_walsh4x4_1_v6
-
-#undef vp8_idct_iwalsh16
-#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_v6
-#endif
-#endif
-
-#if HAVE_ARMV7
-extern prototype_idct(vp8_short_idct4x4llm_1_neon);
-extern prototype_idct(vp8_short_idct4x4llm_neon);
-extern prototype_idct_scalar_add(vp8_dc_only_idct_add_neon);
-extern prototype_second_order(vp8_short_inv_walsh4x4_1_neon);
-extern prototype_second_order(vp8_short_inv_walsh4x4_neon);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_idct_idct1
-#define vp8_idct_idct1 vp8_short_idct4x4llm_1_neon
-
-#undef vp8_idct_idct16
-#define vp8_idct_idct16 vp8_short_idct4x4llm_neon
-
-#undef vp8_idct_idct1_scalar_add
-#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_neon
-
-#undef vp8_idct_iwalsh1
-#define vp8_idct_iwalsh1 vp8_short_inv_walsh4x4_1_neon
-
-#undef vp8_idct_iwalsh16
-#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_neon
-#endif
-#endif
-
-#endif
diff --git a/vp8/common/arm/loopfilter_arm.c b/vp8/common/arm/loopfilter_arm.c
deleted file mode 100644
index 3532a03..0000000
--- a/vp8/common/arm/loopfilter_arm.c
+++ /dev/null
@@ -1,237 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include <math.h>
-#include "vp8/common/loopfilter.h"
-#include "vp8/common/onyxc_int.h"
-
-extern prototype_loopfilter(vp8_loop_filter_horizontal_edge_armv6);
-extern prototype_loopfilter(vp8_loop_filter_vertical_edge_armv6);
-extern prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_armv6);
-extern prototype_loopfilter(vp8_mbloop_filter_vertical_edge_armv6);
-extern prototype_loopfilter(vp8_loop_filter_simple_horizontal_edge_armv6);
-extern prototype_loopfilter(vp8_loop_filter_simple_vertical_edge_armv6);
-
-extern prototype_loopfilter(vp8_loop_filter_horizontal_edge_y_neon);
-extern prototype_loopfilter(vp8_loop_filter_vertical_edge_y_neon);
-extern prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_y_neon);
-extern prototype_loopfilter(vp8_mbloop_filter_vertical_edge_y_neon);
-extern prototype_loopfilter(vp8_loop_filter_simple_horizontal_edge_neon);
-extern prototype_loopfilter(vp8_loop_filter_simple_vertical_edge_neon);
-
-extern loop_filter_uvfunction vp8_loop_filter_horizontal_edge_uv_neon;
-extern loop_filter_uvfunction vp8_loop_filter_vertical_edge_uv_neon;
-extern loop_filter_uvfunction vp8_mbloop_filter_horizontal_edge_uv_neon;
-extern loop_filter_uvfunction vp8_mbloop_filter_vertical_edge_uv_neon;
-
-
-#if HAVE_ARMV6
-/*ARMV6 loopfilter functions*/
-/* Horizontal MB filtering */
-void vp8_loop_filter_mbh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_armv6(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_mbloop_filter_horizontal_edge_armv6(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-}
-
-void vp8_loop_filter_mbhs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-/* Vertical MB Filtering */
-void vp8_loop_filter_mbv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_vertical_edge_armv6(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_mbloop_filter_vertical_edge_armv6(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-}
-
-void vp8_loop_filter_mbvs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-/* Horizontal B Filtering */
-void vp8_loop_filter_bh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_horizontal_edge_armv6(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_loop_filter_horizontal_edge_armv6(v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-}
-
-void vp8_loop_filter_bhs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-
-/* Vertical B Filtering */
-void vp8_loop_filter_bv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_vertical_edge_armv6(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_armv6(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_armv6(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_vertical_edge_armv6(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_loop_filter_vertical_edge_armv6(v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-}
-
-void vp8_loop_filter_bvs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-#endif
-
-#if HAVE_ARMV7
-/* NEON loopfilter functions */
-/* Horizontal MB filtering */
-void vp8_loop_filter_mbh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_y_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_uv_neon(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, v_ptr);
-}
-
-void vp8_loop_filter_mbhs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-/* Vertical MB Filtering */
-void vp8_loop_filter_mbv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_y_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_vertical_edge_uv_neon(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, v_ptr);
-}
-
-void vp8_loop_filter_mbvs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-/* Horizontal B Filtering */
-void vp8_loop_filter_bh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_horizontal_edge_uv_neon(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, v_ptr + 4 * uv_stride);
-}
-
-void vp8_loop_filter_bhs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_neon(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_neon(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_neon(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-
-/* Vertical B Filtering */
-void vp8_loop_filter_bv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_vertical_edge_y_neon(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_y_neon(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_y_neon(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_vertical_edge_uv_neon(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, v_ptr + 4);
-}
-
-void vp8_loop_filter_bvs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_neon(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_neon(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_neon(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-#endif
diff --git a/vp8/common/arm/loopfilter_arm.h b/vp8/common/arm/loopfilter_arm.h
deleted file mode 100644
index cd62207..0000000
--- a/vp8/common/arm/loopfilter_arm.h
+++ /dev/null
@@ -1,89 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef LOOPFILTER_ARM_H
-#define LOOPFILTER_ARM_H
-
-#if HAVE_ARMV6
-extern prototype_loopfilter_block(vp8_loop_filter_mbv_armv6);
-extern prototype_loopfilter_block(vp8_loop_filter_bv_armv6);
-extern prototype_loopfilter_block(vp8_loop_filter_mbh_armv6);
-extern prototype_loopfilter_block(vp8_loop_filter_bh_armv6);
-extern prototype_loopfilter_block(vp8_loop_filter_mbvs_armv6);
-extern prototype_loopfilter_block(vp8_loop_filter_bvs_armv6);
-extern prototype_loopfilter_block(vp8_loop_filter_mbhs_armv6);
-extern prototype_loopfilter_block(vp8_loop_filter_bhs_armv6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_lf_normal_mb_v
-#define vp8_lf_normal_mb_v vp8_loop_filter_mbv_armv6
-
-#undef vp8_lf_normal_b_v
-#define vp8_lf_normal_b_v vp8_loop_filter_bv_armv6
-
-#undef vp8_lf_normal_mb_h
-#define vp8_lf_normal_mb_h vp8_loop_filter_mbh_armv6
-
-#undef vp8_lf_normal_b_h
-#define vp8_lf_normal_b_h vp8_loop_filter_bh_armv6
-
-#undef vp8_lf_simple_mb_v
-#define vp8_lf_simple_mb_v vp8_loop_filter_mbvs_armv6
-
-#undef vp8_lf_simple_b_v
-#define vp8_lf_simple_b_v vp8_loop_filter_bvs_armv6
-
-#undef vp8_lf_simple_mb_h
-#define vp8_lf_simple_mb_h vp8_loop_filter_mbhs_armv6
-
-#undef vp8_lf_simple_b_h
-#define vp8_lf_simple_b_h vp8_loop_filter_bhs_armv6
-#endif
-#endif
-
-#if HAVE_ARMV7
-extern prototype_loopfilter_block(vp8_loop_filter_mbv_neon);
-extern prototype_loopfilter_block(vp8_loop_filter_bv_neon);
-extern prototype_loopfilter_block(vp8_loop_filter_mbh_neon);
-extern prototype_loopfilter_block(vp8_loop_filter_bh_neon);
-extern prototype_loopfilter_block(vp8_loop_filter_mbvs_neon);
-extern prototype_loopfilter_block(vp8_loop_filter_bvs_neon);
-extern prototype_loopfilter_block(vp8_loop_filter_mbhs_neon);
-extern prototype_loopfilter_block(vp8_loop_filter_bhs_neon);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_lf_normal_mb_v
-#define vp8_lf_normal_mb_v vp8_loop_filter_mbv_neon
-
-#undef vp8_lf_normal_b_v
-#define vp8_lf_normal_b_v vp8_loop_filter_bv_neon
-
-#undef vp8_lf_normal_mb_h
-#define vp8_lf_normal_mb_h vp8_loop_filter_mbh_neon
-
-#undef vp8_lf_normal_b_h
-#define vp8_lf_normal_b_h vp8_loop_filter_bh_neon
-
-#undef vp8_lf_simple_mb_v
-#define vp8_lf_simple_mb_v vp8_loop_filter_mbvs_neon
-
-#undef vp8_lf_simple_b_v
-#define vp8_lf_simple_b_v vp8_loop_filter_bvs_neon
-
-#undef vp8_lf_simple_mb_h
-#define vp8_lf_simple_mb_h vp8_loop_filter_mbhs_neon
-
-#undef vp8_lf_simple_b_h
-#define vp8_lf_simple_b_h vp8_loop_filter_bhs_neon
-#endif
-#endif
-
-#endif
diff --git a/vp8/common/arm/neon/dc_only_idct_add_neon.asm b/vp8/common/arm/neon/dc_only_idct_add_neon.asm
deleted file mode 100644
index 49ba05f..0000000
--- a/vp8/common/arm/neon/dc_only_idct_add_neon.asm
+++ /dev/null
@@ -1,49 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license and patent
-; grant that can be found in the LICENSE file in the root of the source
-; tree. All contributing project authors may be found in the AUTHORS
-; file in the root of the source tree.
-;
-
-
- EXPORT |vp8_dc_only_idct_add_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void vp8_dc_only_idct_add_neon(short input_dc, unsigned char *pred_ptr,
-; unsigned char *dst_ptr, int pitch, int stride)
-; r0 input_dc
-; r1 pred_ptr
-; r2 dst_ptr
-; r3 pitch
-; sp stride
-|vp8_dc_only_idct_add_neon| PROC
- add r0, r0, #4
- asr r0, r0, #3
- ldr r12, [sp]
- vdup.16 q0, r0
-
- vld1.32 {d2[0]}, [r1], r3
- vld1.32 {d2[1]}, [r1], r3
- vld1.32 {d4[0]}, [r1], r3
- vld1.32 {d4[1]}, [r1]
-
- vaddw.u8 q1, q0, d2
- vaddw.u8 q2, q0, d4
-
- vqmovun.s16 d2, q1
- vqmovun.s16 d4, q2
-
- vst1.32 {d2[0]}, [r2], r12
- vst1.32 {d2[1]}, [r2], r12
- vst1.32 {d4[0]}, [r2], r12
- vst1.32 {d4[1]}, [r2]
-
- bx lr
-
- ENDP
- END
diff --git a/vp8/common/arm/neon/iwalsh_neon.asm b/vp8/common/arm/neon/iwalsh_neon.asm
deleted file mode 100644
index 663bf39..0000000
--- a/vp8/common/arm/neon/iwalsh_neon.asm
+++ /dev/null
@@ -1,96 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
- EXPORT |vp8_short_inv_walsh4x4_neon|
- EXPORT |vp8_short_inv_walsh4x4_1_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
-;short vp8_short_inv_walsh4x4_neon(short *input, short *output)
-|vp8_short_inv_walsh4x4_neon| PROC
-
- ; read in all four lines of values: d0->d3
- vldm.64 r0, {q0, q1}
-
- ; first for loop
-
- vadd.s16 d4, d0, d3 ;a = [0] + [12]
- vadd.s16 d5, d1, d2 ;b = [4] + [8]
- vsub.s16 d6, d1, d2 ;c = [4] - [8]
- vsub.s16 d7, d0, d3 ;d = [0] - [12]
-
- vadd.s16 d0, d4, d5 ;a + b
- vadd.s16 d1, d6, d7 ;c + d
- vsub.s16 d2, d4, d5 ;a - b
- vsub.s16 d3, d7, d6 ;d - c
-
- vtrn.32 d0, d2 ;d0: 0 1 8 9
- ;d2: 2 3 10 11
- vtrn.32 d1, d3 ;d1: 4 5 12 13
- ;d3: 6 7 14 15
-
- vtrn.16 d0, d1 ;d0: 0 4 8 12
- ;d1: 1 5 9 13
- vtrn.16 d2, d3 ;d2: 2 6 10 14
- ;d3: 3 7 11 15
-
- ; second for loop
-
- vadd.s16 d4, d0, d3 ;a = [0] + [3]
- vadd.s16 d5, d1, d2 ;b = [1] + [2]
- vsub.s16 d6, d1, d2 ;c = [1] - [2]
- vsub.s16 d7, d0, d3 ;d = [0] - [3]
-
- vadd.s16 d0, d4, d5 ;e = a + b
- vadd.s16 d1, d6, d7 ;f = c + d
- vsub.s16 d2, d4, d5 ;g = a - b
- vsub.s16 d3, d7, d6 ;h = d - c
-
- vmov.i16 q2, #3
- vadd.i16 q0, q0, q2 ;e/f += 3
- vadd.i16 q1, q1, q2 ;g/h += 3
-
- vshr.s16 q0, q0, #3 ;e/f >> 3
- vshr.s16 q1, q1, #3 ;g/h >> 3
-
- vtrn.32 d0, d2
- vtrn.32 d1, d3
- vtrn.16 d0, d1
- vtrn.16 d2, d3
-
- vstmia.16 r1!, {q0}
- vstmia.16 r1!, {q1}
-
- bx lr
- ENDP ; |vp8_short_inv_walsh4x4_neon|
-
-
-;short vp8_short_inv_walsh4x4_1_neon(short *input, short *output)
-|vp8_short_inv_walsh4x4_1_neon| PROC
- ; load a full line into a neon register
- vld1.16 {q0}, [r0]
- ; extract first element and replicate
- vdup.16 q1, d0[0]
- ; add 3 to all values
- vmov.i16 q2, #3
- vadd.i16 q3, q1, q2
- ; right shift
- vshr.s16 q3, q3, #3
- ; write it back
- vstmia.16 r1!, {q3}
- vstmia.16 r1!, {q3}
-
- bx lr
- ENDP ; |vp8_short_inv_walsh4x4_1_neon|
-
- END
diff --git a/vp8/common/arm/neon/loopfilter_neon.asm b/vp8/common/arm/neon/loopfilter_neon.asm
deleted file mode 100644
index 3378013..0000000
--- a/vp8/common/arm/neon/loopfilter_neon.asm
+++ /dev/null
@@ -1,408 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_loop_filter_horizontal_edge_y_neon|
- EXPORT |vp8_loop_filter_horizontal_edge_uv_neon|
- EXPORT |vp8_loop_filter_vertical_edge_y_neon|
- EXPORT |vp8_loop_filter_vertical_edge_uv_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; flimit, limit, and thresh should be positive numbers.
-; All 16 elements in these variables are equal.
-
-; void vp8_loop_filter_horizontal_edge_y_neon(unsigned char *src, int pitch,
-; const signed char *flimit,
-; const signed char *limit,
-; const signed char *thresh,
-; int count)
-; r0 unsigned char *src
-; r1 int pitch
-; r2 const signed char *flimit
-; r3 const signed char *limit
-; sp const signed char *thresh,
-; sp+4 int count (unused)
-|vp8_loop_filter_horizontal_edge_y_neon| PROC
- stmdb sp!, {lr}
- vld1.s8 {d0[], d1[]}, [r2] ; flimit
- vld1.s8 {d2[], d3[]}, [r3] ; limit
- sub r2, r0, r1, lsl #2 ; move src pointer down by 4 lines
- ldr r12, [sp, #4] ; load thresh pointer
-
- vld1.u8 {q3}, [r2], r1 ; p3
- vld1.u8 {q4}, [r2], r1 ; p2
- vld1.u8 {q5}, [r2], r1 ; p1
- vld1.u8 {q6}, [r2], r1 ; p0
- vld1.u8 {q7}, [r2], r1 ; q0
- vld1.u8 {q8}, [r2], r1 ; q1
- vld1.u8 {q9}, [r2], r1 ; q2
- vld1.u8 {q10}, [r2] ; q3
- vld1.s8 {d4[], d5[]}, [r12] ; thresh
- sub r0, r0, r1, lsl #1
-
- bl vp8_loop_filter_neon
-
- vst1.u8 {q5}, [r0], r1 ; store op1
- vst1.u8 {q6}, [r0], r1 ; store op0
- vst1.u8 {q7}, [r0], r1 ; store oq0
- vst1.u8 {q8}, [r0], r1 ; store oq1
-
- ldmia sp!, {pc}
- ENDP ; |vp8_loop_filter_horizontal_edge_y_neon|
-
-; void vp8_loop_filter_horizontal_edge_uv_neon(unsigned char *u, int pitch
-; const signed char *flimit,
-; const signed char *limit,
-; const signed char *thresh,
-; unsigned char *v)
-; r0 unsigned char *u,
-; r1 int pitch,
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; sp const signed char *thresh,
-; sp+4 unsigned char *v
-|vp8_loop_filter_horizontal_edge_uv_neon| PROC
- stmdb sp!, {lr}
- vld1.s8 {d0[], d1[]}, [r2] ; flimit
- vld1.s8 {d2[], d3[]}, [r3] ; limit
- ldr r2, [sp, #8] ; load v ptr
-
- sub r3, r0, r1, lsl #2 ; move u pointer down by 4 lines
- vld1.u8 {d6}, [r3], r1 ; p3
- vld1.u8 {d8}, [r3], r1 ; p2
- vld1.u8 {d10}, [r3], r1 ; p1
- vld1.u8 {d12}, [r3], r1 ; p0
- vld1.u8 {d14}, [r3], r1 ; q0
- vld1.u8 {d16}, [r3], r1 ; q1
- vld1.u8 {d18}, [r3], r1 ; q2
- vld1.u8 {d20}, [r3] ; q3
-
- ldr r3, [sp, #4] ; load thresh pointer
-
- sub r12, r2, r1, lsl #2 ; move v pointer down by 4 lines
- vld1.u8 {d7}, [r12], r1 ; p3
- vld1.u8 {d9}, [r12], r1 ; p2
- vld1.u8 {d11}, [r12], r1 ; p1
- vld1.u8 {d13}, [r12], r1 ; p0
- vld1.u8 {d15}, [r12], r1 ; q0
- vld1.u8 {d17}, [r12], r1 ; q1
- vld1.u8 {d19}, [r12], r1 ; q2
- vld1.u8 {d21}, [r12] ; q3
-
- vld1.s8 {d4[], d5[]}, [r3] ; thresh
-
- bl vp8_loop_filter_neon
-
- sub r0, r0, r1, lsl #1
- sub r2, r2, r1, lsl #1
-
- vst1.u8 {d10}, [r0], r1 ; store u op1
- vst1.u8 {d11}, [r2], r1 ; store v op1
- vst1.u8 {d12}, [r0], r1 ; store u op0
- vst1.u8 {d13}, [r2], r1 ; store v op0
- vst1.u8 {d14}, [r0], r1 ; store u oq0
- vst1.u8 {d15}, [r2], r1 ; store v oq0
- vst1.u8 {d16}, [r0] ; store u oq1
- vst1.u8 {d17}, [r2] ; store v oq1
-
- ldmia sp!, {pc}
- ENDP ; |vp8_loop_filter_horizontal_edge_uv_neon|
-
-; void vp8_loop_filter_vertical_edge_y_neon(unsigned char *src, int pitch,
-; const signed char *flimit,
-; const signed char *limit,
-; const signed char *thresh,
-; int count)
-; r0 unsigned char *src,
-; r1 int pitch,
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; sp const signed char *thresh,
-; sp+4 int count (unused)
-|vp8_loop_filter_vertical_edge_y_neon| PROC
- stmdb sp!, {lr}
- vld1.s8 {d0[], d1[]}, [r2] ; flimit
- vld1.s8 {d2[], d3[]}, [r3] ; limit
- sub r2, r0, #4 ; src ptr down by 4 columns
- sub r0, r0, #2 ; dst ptr
- ldr r12, [sp, #4] ; load thresh pointer
-
- vld1.u8 {d6}, [r2], r1 ; load first 8-line src data
- vld1.u8 {d8}, [r2], r1
- vld1.u8 {d10}, [r2], r1
- vld1.u8 {d12}, [r2], r1
- vld1.u8 {d14}, [r2], r1
- vld1.u8 {d16}, [r2], r1
- vld1.u8 {d18}, [r2], r1
- vld1.u8 {d20}, [r2], r1
-
- vld1.s8 {d4[], d5[]}, [r12] ; thresh
-
- vld1.u8 {d7}, [r2], r1 ; load second 8-line src data
- vld1.u8 {d9}, [r2], r1
- vld1.u8 {d11}, [r2], r1
- vld1.u8 {d13}, [r2], r1
- vld1.u8 {d15}, [r2], r1
- vld1.u8 {d17}, [r2], r1
- vld1.u8 {d19}, [r2], r1
- vld1.u8 {d21}, [r2]
-
- ;transpose to 8x16 matrix
- vtrn.32 q3, q7
- vtrn.32 q4, q8
- vtrn.32 q5, q9
- vtrn.32 q6, q10
-
- vtrn.16 q3, q5
- vtrn.16 q4, q6
- vtrn.16 q7, q9
- vtrn.16 q8, q10
-
- vtrn.8 q3, q4
- vtrn.8 q5, q6
- vtrn.8 q7, q8
- vtrn.8 q9, q10
-
- bl vp8_loop_filter_neon
-
- vswp d12, d11
- vswp d16, d13
- vswp d14, d12
- vswp d16, d15
-
- ;store op1, op0, oq0, oq1
- vst4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r1
- vst4.8 {d10[1], d11[1], d12[1], d13[1]}, [r0], r1
- vst4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r1
- vst4.8 {d10[3], d11[3], d12[3], d13[3]}, [r0], r1
- vst4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r1
- vst4.8 {d10[5], d11[5], d12[5], d13[5]}, [r0], r1
- vst4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r1
- vst4.8 {d10[7], d11[7], d12[7], d13[7]}, [r0], r1
- vst4.8 {d14[0], d15[0], d16[0], d17[0]}, [r0], r1
- vst4.8 {d14[1], d15[1], d16[1], d17[1]}, [r0], r1
- vst4.8 {d14[2], d15[2], d16[2], d17[2]}, [r0], r1
- vst4.8 {d14[3], d15[3], d16[3], d17[3]}, [r0], r1
- vst4.8 {d14[4], d15[4], d16[4], d17[4]}, [r0], r1
- vst4.8 {d14[5], d15[5], d16[5], d17[5]}, [r0], r1
- vst4.8 {d14[6], d15[6], d16[6], d17[6]}, [r0], r1
- vst4.8 {d14[7], d15[7], d16[7], d17[7]}, [r0]
-
- ldmia sp!, {pc}
- ENDP ; |vp8_loop_filter_vertical_edge_y_neon|
-
-; void vp8_loop_filter_vertical_edge_uv_neon(unsigned char *u, int pitch
-; const signed char *flimit,
-; const signed char *limit,
-; const signed char *thresh,
-; unsigned char *v)
-; r0 unsigned char *u,
-; r1 int pitch,
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; sp const signed char *thresh,
-; sp+4 unsigned char *v
-|vp8_loop_filter_vertical_edge_uv_neon| PROC
- stmdb sp!, {lr}
- sub r12, r0, #4 ; move u pointer down by 4 columns
- vld1.s8 {d0[], d1[]}, [r2] ; flimit
- vld1.s8 {d2[], d3[]}, [r3] ; limit
-
- ldr r2, [sp, #8] ; load v ptr
-
- vld1.u8 {d6}, [r12], r1 ;load u data
- vld1.u8 {d8}, [r12], r1
- vld1.u8 {d10}, [r12], r1
- vld1.u8 {d12}, [r12], r1
- vld1.u8 {d14}, [r12], r1
- vld1.u8 {d16}, [r12], r1
- vld1.u8 {d18}, [r12], r1
- vld1.u8 {d20}, [r12]
-
- sub r3, r2, #4 ; move v pointer down by 4 columns
- vld1.u8 {d7}, [r3], r1 ;load v data
- vld1.u8 {d9}, [r3], r1
- vld1.u8 {d11}, [r3], r1
- vld1.u8 {d13}, [r3], r1
- vld1.u8 {d15}, [r3], r1
- vld1.u8 {d17}, [r3], r1
- vld1.u8 {d19}, [r3], r1
- vld1.u8 {d21}, [r3]
-
- ldr r12, [sp, #4] ; load thresh pointer
-
- ;transpose to 8x16 matrix
- vtrn.32 q3, q7
- vtrn.32 q4, q8
- vtrn.32 q5, q9
- vtrn.32 q6, q10
-
- vtrn.16 q3, q5
- vtrn.16 q4, q6
- vtrn.16 q7, q9
- vtrn.16 q8, q10
-
- vtrn.8 q3, q4
- vtrn.8 q5, q6
- vtrn.8 q7, q8
- vtrn.8 q9, q10
-
- vld1.s8 {d4[], d5[]}, [r12] ; thresh
-
- bl vp8_loop_filter_neon
-
- sub r0, r0, #2
- sub r2, r2, #2
-
- vswp d12, d11
- vswp d16, d13
- vswp d14, d12
- vswp d16, d15
-
- ;store op1, op0, oq0, oq1
- vst4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r1
- vst4.8 {d14[0], d15[0], d16[0], d17[0]}, [r2], r1
- vst4.8 {d10[1], d11[1], d12[1], d13[1]}, [r0], r1
- vst4.8 {d14[1], d15[1], d16[1], d17[1]}, [r2], r1
- vst4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r1
- vst4.8 {d14[2], d15[2], d16[2], d17[2]}, [r2], r1
- vst4.8 {d10[3], d11[3], d12[3], d13[3]}, [r0], r1
- vst4.8 {d14[3], d15[3], d16[3], d17[3]}, [r2], r1
- vst4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r1
- vst4.8 {d14[4], d15[4], d16[4], d17[4]}, [r2], r1
- vst4.8 {d10[5], d11[5], d12[5], d13[5]}, [r0], r1
- vst4.8 {d14[5], d15[5], d16[5], d17[5]}, [r2], r1
- vst4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r1
- vst4.8 {d14[6], d15[6], d16[6], d17[6]}, [r2], r1
- vst4.8 {d10[7], d11[7], d12[7], d13[7]}, [r0]
- vst4.8 {d14[7], d15[7], d16[7], d17[7]}, [r2]
-
- ldmia sp!, {pc}
- ENDP ; |vp8_loop_filter_vertical_edge_uv_neon|
-
-; void vp8_loop_filter_neon();
-; This is a helper function for the loopfilters. The invidual functions do the
-; necessary load, transpose (if necessary) and store.
-
-; r0-r3 PRESERVE
-; q0 flimit
-; q1 limit
-; q2 thresh
-; q3 p3
-; q4 p2
-; q5 p1
-; q6 p0
-; q7 q0
-; q8 q1
-; q9 q2
-; q10 q3
-|vp8_loop_filter_neon| PROC
- adr r12, lf_coeff
-
- ; vp8_filter_mask
- vabd.u8 q11, q3, q4 ; abs(p3 - p2)
- vabd.u8 q12, q4, q5 ; abs(p2 - p1)
- vabd.u8 q13, q5, q6 ; abs(p1 - p0)
- vabd.u8 q14, q8, q7 ; abs(q1 - q0)
- vabd.u8 q3, q9, q8 ; abs(q2 - q1)
- vabd.u8 q4, q10, q9 ; abs(q3 - q2)
- vabd.u8 q9, q6, q7 ; abs(p0 - q0)
-
- vmax.u8 q11, q11, q12
- vmax.u8 q12, q13, q14
- vmax.u8 q3, q3, q4
- vmax.u8 q15, q11, q12
-
- ; vp8_hevmask
- vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
- vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
- vmax.u8 q15, q15, q3
-
- vadd.u8 q0, q0, q0 ; flimit * 2
- vadd.u8 q0, q0, q1 ; flimit * 2 + limit
- vcge.u8 q15, q1, q15
-
- vabd.u8 q2, q5, q8 ; a = abs(p1 - q1)
- vqadd.u8 q9, q9, q9 ; b = abs(p0 - q0) * 2
- vshr.u8 q2, q2, #1 ; a = a / 2
- vqadd.u8 q9, q9, q2 ; a = b + a
- vcge.u8 q9, q0, q9 ; (a > flimit * 2 + limit) * -1
-
- vld1.u8 {q0}, [r12]!
-
- ; vp8_filter() function
- ; convert to signed
- veor q7, q7, q0 ; qs0
- veor q6, q6, q0 ; ps0
- veor q5, q5, q0 ; ps1
- veor q8, q8, q0 ; qs1
-
- vld1.u8 {q10}, [r12]!
-
- vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
- vsubl.s8 q11, d15, d13
-
- vmovl.u8 q4, d20
-
- vqsub.s8 q1, q5, q8 ; vp8_filter = clamp(ps1-qs1)
- vorr q14, q13, q14 ; vp8_hevmask
-
- vmul.i16 q2, q2, q4 ; 3 * ( qs0 - ps0)
- vmul.i16 q11, q11, q4
-
- vand q1, q1, q14 ; vp8_filter &= hev
- vand q15, q15, q9 ; vp8_filter_mask
-
- vaddw.s8 q2, q2, d2
- vaddw.s8 q11, q11, d3
-
- vld1.u8 {q9}, [r12]!
-
- ; vp8_filter = clamp(vp8_filter + 3 * ( qs0 - ps0))
- vqmovn.s16 d2, q2
- vqmovn.s16 d3, q11
- vand q1, q1, q15 ; vp8_filter &= mask
-
- vqadd.s8 q2, q1, q10 ; Filter2 = clamp(vp8_filter+3)
- vqadd.s8 q1, q1, q9 ; Filter1 = clamp(vp8_filter+4)
- vshr.s8 q2, q2, #3 ; Filter2 >>= 3
- vshr.s8 q1, q1, #3 ; Filter1 >>= 3
-
- vqadd.s8 q11, q6, q2 ; u = clamp(ps0 + Filter2)
- vqsub.s8 q10, q7, q1 ; u = clamp(qs0 - Filter1)
-
- ; outer tap adjustments: ++vp8_filter >> 1
- vrshr.s8 q1, q1, #1
- vbic q1, q1, q14 ; vp8_filter &= ~hev
-
- vqadd.s8 q13, q5, q1 ; u = clamp(ps1 + vp8_filter)
- vqsub.s8 q12, q8, q1 ; u = clamp(qs1 - vp8_filter)
-
- veor q5, q13, q0 ; *op1 = u^0x80
- veor q6, q11, q0 ; *op0 = u^0x80
- veor q7, q10, q0 ; *oq0 = u^0x80
- veor q8, q12, q0 ; *oq1 = u^0x80
-
- bx lr
- ENDP ; |vp8_loop_filter_horizontal_edge_y_neon|
-
-;-----------------
-
-lf_coeff
- DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
- DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
- DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
- DCD 0x01010101, 0x01010101, 0x01010101, 0x01010101
-
- END
diff --git a/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm b/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm
deleted file mode 100644
index cd3710d..0000000
--- a/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm
+++ /dev/null
@@ -1,113 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_loop_filter_simple_horizontal_edge_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
-;are equal. So, in the code, only one load is needed
-;for flimit. Same way applies to limit and thresh.
-; r0 unsigned char *s,
-; r1 int p, //pitch
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; stack(r4) const signed char *thresh,
-; //stack(r5) int count --unused
-
-|vp8_loop_filter_simple_horizontal_edge_neon| PROC
- sub r0, r0, r1, lsl #1 ; move src pointer down by 2 lines
-
- adr r12, lfhy_coeff
- vld1.u8 {q5}, [r0], r1 ; p1
- vld1.s8 {d2[], d3[]}, [r2] ; flimit
- vld1.s8 {d26[], d27[]}, [r3] ; limit -> q13
- vld1.u8 {q6}, [r0], r1 ; p0
- vld1.u8 {q0}, [r12]! ; 0x80
- vld1.u8 {q7}, [r0], r1 ; q0
- vld1.u8 {q10}, [r12]! ; 0x03
- vld1.u8 {q8}, [r0] ; q1
-
- ;vp8_filter_mask() function
- vabd.u8 q15, q6, q7 ; abs(p0 - q0)
- vabd.u8 q14, q5, q8 ; abs(p1 - q1)
- vqadd.u8 q15, q15, q15 ; abs(p0 - q0) * 2
- vshr.u8 q14, q14, #1 ; abs(p1 - q1) / 2
- vqadd.u8 q15, q15, q14 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
-
- ;vp8_filter() function
- veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
- veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
- veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
- veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
-
- vadd.u8 q1, q1, q1 ; flimit * 2
- vadd.u8 q1, q1, q13 ; flimit * 2 + limit
- vcge.u8 q15, q1, q15 ; (abs(p0 - q0)*2 + abs(p1-q1)/2 > flimit*2 + limit)*-1
-
-;;;;;;;;;;
- ;vqsub.s8 q2, q7, q6 ; ( qs0 - ps0)
- vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
- vsubl.s8 q3, d15, d13
-
- vqsub.s8 q4, q5, q8 ; q4: vp8_filter = vp8_signed_char_clamp(ps1-qs1)
-
- ;vmul.i8 q2, q2, q10 ; 3 * ( qs0 - ps0)
- vadd.s16 q11, q2, q2 ; 3 * ( qs0 - ps0)
- vadd.s16 q12, q3, q3
-
- vld1.u8 {q9}, [r12]! ; 0x04
-
- vadd.s16 q2, q2, q11
- vadd.s16 q3, q3, q12
-
- vaddw.s8 q2, q2, d8 ; vp8_filter + 3 * ( qs0 - ps0)
- vaddw.s8 q3, q3, d9
-
- ;vqadd.s8 q4, q4, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
- vqmovn.s16 d8, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
- vqmovn.s16 d9, q3
-;;;;;;;;;;;;;
-
- vand q4, q4, q15 ; vp8_filter &= mask
-
- vqadd.s8 q2, q4, q10 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
- vqadd.s8 q4, q4, q9 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
- vshr.s8 q2, q2, #3 ; Filter2 >>= 3
- vshr.s8 q4, q4, #3 ; Filter1 >>= 3
-
- sub r0, r0, r1, lsl #1
-
- ;calculate output
- vqadd.s8 q11, q6, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
- vqsub.s8 q10, q7, q4 ; u = vp8_signed_char_clamp(qs0 - Filter1)
-
- add r3, r0, r1
-
- veor q6, q11, q0 ; *op0 = u^0x80
- veor q7, q10, q0 ; *oq0 = u^0x80
-
- vst1.u8 {q6}, [r0] ; store op0
- vst1.u8 {q7}, [r3] ; store oq0
-
- bx lr
- ENDP ; |vp8_loop_filter_simple_horizontal_edge_neon|
-
-;-----------------
-
-lfhy_coeff
- DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
- DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
- DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
-
- END
diff --git a/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm b/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm
deleted file mode 100644
index ae5e093..0000000
--- a/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm
+++ /dev/null
@@ -1,154 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_loop_filter_simple_vertical_edge_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;Note: flimit, limit, and thresh should be positive numbers. All 16 elements in flimit
-;are equal. So, in the code, only one load is needed
-;for flimit. Same way applies to limit and thresh.
-; r0 unsigned char *s,
-; r1 int p, //pitch
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; stack(r4) const signed char *thresh,
-; //stack(r5) int count --unused
-
-|vp8_loop_filter_simple_vertical_edge_neon| PROC
- sub r0, r0, #2 ; move src pointer down by 2 columns
-
- vld4.8 {d6[0], d7[0], d8[0], d9[0]}, [r0], r1
- vld1.s8 {d2[], d3[]}, [r2] ; flimit
- vld1.s8 {d26[], d27[]}, [r3] ; limit -> q13
- vld4.8 {d6[1], d7[1], d8[1], d9[1]}, [r0], r1
- adr r12, vlfy_coeff
- vld4.8 {d6[2], d7[2], d8[2], d9[2]}, [r0], r1
- vld4.8 {d6[3], d7[3], d8[3], d9[3]}, [r0], r1
- vld4.8 {d6[4], d7[4], d8[4], d9[4]}, [r0], r1
- vld4.8 {d6[5], d7[5], d8[5], d9[5]}, [r0], r1
- vld4.8 {d6[6], d7[6], d8[6], d9[6]}, [r0], r1
- vld4.8 {d6[7], d7[7], d8[7], d9[7]}, [r0], r1
-
- vld4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r1
- vld1.u8 {q0}, [r12]! ; 0x80
- vld4.8 {d10[1], d11[1], d12[1], d13[1]}, [r0], r1
- vld1.u8 {q11}, [r12]! ; 0x03
- vld4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r1
- vld1.u8 {q12}, [r12]! ; 0x04
- vld4.8 {d10[3], d11[3], d12[3], d13[3]}, [r0], r1
- vld4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r1
- vld4.8 {d10[5], d11[5], d12[5], d13[5]}, [r0], r1
- vld4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r1
- vld4.8 {d10[7], d11[7], d12[7], d13[7]}, [r0], r1
-
- vswp d7, d10
- vswp d12, d9
- ;vswp q4, q5 ; p1:q3, p0:q5, q0:q4, q1:q6
-
- ;vp8_filter_mask() function
- ;vp8_hevmask() function
- sub r0, r0, r1, lsl #4
- vabd.u8 q15, q5, q4 ; abs(p0 - q0)
- vabd.u8 q14, q3, q6 ; abs(p1 - q1)
- vqadd.u8 q15, q15, q15 ; abs(p0 - q0) * 2
- vshr.u8 q14, q14, #1 ; abs(p1 - q1) / 2
- vqadd.u8 q15, q15, q14 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
-
- veor q4, q4, q0 ; qs0: q0 offset to convert to a signed value
- veor q5, q5, q0 ; ps0: p0 offset to convert to a signed value
- veor q3, q3, q0 ; ps1: p1 offset to convert to a signed value
- veor q6, q6, q0 ; qs1: q1 offset to convert to a signed value
-
- vadd.u8 q1, q1, q1 ; flimit * 2
- vadd.u8 q1, q1, q13 ; flimit * 2 + limit
- vcge.u8 q15, q1, q15 ; abs(p0 - q0)*2 + abs(p1-q1)/2 > flimit*2 + limit)*-1
-
- ;vp8_filter() function
-;;;;;;;;;;
- ;vqsub.s8 q2, q5, q4 ; ( qs0 - ps0)
- vsubl.s8 q2, d8, d10 ; ( qs0 - ps0)
- vsubl.s8 q13, d9, d11
-
- vqsub.s8 q1, q3, q6 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
-
- ;vmul.i8 q2, q2, q11 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
- vadd.s16 q10, q2, q2 ; 3 * ( qs0 - ps0)
- vadd.s16 q14, q13, q13
- vadd.s16 q2, q2, q10
- vadd.s16 q13, q13, q14
-
- ;vqadd.s8 q1, q1, q2
- vaddw.s8 q2, q2, d2 ; vp8_filter + 3 * ( qs0 - ps0)
- vaddw.s8 q13, q13, d3
-
- vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
- vqmovn.s16 d3, q13
-
- add r0, r0, #1
- add r2, r0, r1
-;;;;;;;;;;;
-
- vand q1, q1, q15 ; vp8_filter &= mask
-
- vqadd.s8 q2, q1, q11 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
- vqadd.s8 q1, q1, q12 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
- vshr.s8 q2, q2, #3 ; Filter2 >>= 3
- vshr.s8 q1, q1, #3 ; Filter1 >>= 3
-
- ;calculate output
- vqsub.s8 q10, q4, q1 ; u = vp8_signed_char_clamp(qs0 - Filter1)
- vqadd.s8 q11, q5, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
-
- veor q7, q10, q0 ; *oq0 = u^0x80
- veor q6, q11, q0 ; *op0 = u^0x80
-
- add r3, r2, r1
- vswp d13, d14
- add r12, r3, r1
-
- ;store op1, op0, oq0, oq1
- vst2.8 {d12[0], d13[0]}, [r0]
- vst2.8 {d12[1], d13[1]}, [r2]
- vst2.8 {d12[2], d13[2]}, [r3]
- vst2.8 {d12[3], d13[3]}, [r12], r1
- add r0, r12, r1
- vst2.8 {d12[4], d13[4]}, [r12]
- vst2.8 {d12[5], d13[5]}, [r0], r1
- add r2, r0, r1
- vst2.8 {d12[6], d13[6]}, [r0]
- vst2.8 {d12[7], d13[7]}, [r2], r1
- add r3, r2, r1
- vst2.8 {d14[0], d15[0]}, [r2]
- vst2.8 {d14[1], d15[1]}, [r3], r1
- add r12, r3, r1
- vst2.8 {d14[2], d15[2]}, [r3]
- vst2.8 {d14[3], d15[3]}, [r12], r1
- add r0, r12, r1
- vst2.8 {d14[4], d15[4]}, [r12]
- vst2.8 {d14[5], d15[5]}, [r0], r1
- add r2, r0, r1
- vst2.8 {d14[6], d15[6]}, [r0]
- vst2.8 {d14[7], d15[7]}, [r2]
-
- bx lr
- ENDP ; |vp8_loop_filter_simple_vertical_edge_neon|
-
-;-----------------
-
-vlfy_coeff
- DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
- DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
- DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
-
- END
diff --git a/vp8/common/arm/neon/mbloopfilter_neon.asm b/vp8/common/arm/neon/mbloopfilter_neon.asm
deleted file mode 100644
index c62830e..0000000
--- a/vp8/common/arm/neon/mbloopfilter_neon.asm
+++ /dev/null
@@ -1,518 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_mbloop_filter_horizontal_edge_y_neon|
- EXPORT |vp8_mbloop_filter_horizontal_edge_uv_neon|
- EXPORT |vp8_mbloop_filter_vertical_edge_y_neon|
- EXPORT |vp8_mbloop_filter_vertical_edge_uv_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; flimit, limit, and thresh should be positive numbers.
-; All 16 elements in these variables are equal.
-
-; void vp8_mbloop_filter_horizontal_edge_y_neon(unsigned char *src, int pitch,
-; const signed char *flimit,
-; const signed char *limit,
-; const signed char *thresh,
-; int count)
-; r0 unsigned char *src,
-; r1 int pitch,
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; sp const signed char *thresh,
-; sp+4 int count (unused)
-|vp8_mbloop_filter_horizontal_edge_y_neon| PROC
- stmdb sp!, {lr}
- sub r0, r0, r1, lsl #2 ; move src pointer down by 4 lines
- ldr r12, [sp, #4] ; load thresh pointer
-
- vld1.u8 {q3}, [r0], r1 ; p3
- vld1.s8 {d2[], d3[]}, [r3] ; limit
- vld1.u8 {q4}, [r0], r1 ; p2
- vld1.s8 {d4[], d5[]}, [r12] ; thresh
- vld1.u8 {q5}, [r0], r1 ; p1
- vld1.u8 {q6}, [r0], r1 ; p0
- vld1.u8 {q7}, [r0], r1 ; q0
- vld1.u8 {q8}, [r0], r1 ; q1
- vld1.u8 {q9}, [r0], r1 ; q2
- vld1.u8 {q10}, [r0], r1 ; q3
-
- bl vp8_mbloop_filter_neon
-
- sub r0, r0, r1, lsl #3
- add r0, r0, r1
- add r2, r0, r1
- add r3, r2, r1
-
- vst1.u8 {q4}, [r0] ; store op2
- vst1.u8 {q5}, [r2] ; store op1
- vst1.u8 {q6}, [r3], r1 ; store op0
- add r12, r3, r1
- vst1.u8 {q7}, [r3] ; store oq0
- vst1.u8 {q8}, [r12], r1 ; store oq1
- vst1.u8 {q9}, [r12] ; store oq2
-
- ldmia sp!, {pc}
- ENDP ; |vp8_mbloop_filter_horizontal_edge_y_neon|
-
-; void vp8_mbloop_filter_horizontal_edge_uv_neon(unsigned char *u, int pitch,
-; const signed char *flimit,
-; const signed char *limit,
-; const signed char *thresh,
-; unsigned char *v)
-; r0 unsigned char *u,
-; r1 int pitch,
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; sp const signed char *thresh,
-; sp+4 unsigned char *v
-|vp8_mbloop_filter_horizontal_edge_uv_neon| PROC
- stmdb sp!, {lr}
- sub r0, r0, r1, lsl #2 ; move u pointer down by 4 lines
- vld1.s8 {d2[], d3[]}, [r3] ; limit
- ldr r3, [sp, #8] ; load v ptr
- ldr r12, [sp, #4] ; load thresh pointer
- sub r3, r3, r1, lsl #2 ; move v pointer down by 4 lines
-
- vld1.u8 {d6}, [r0], r1 ; p3
- vld1.u8 {d7}, [r3], r1 ; p3
- vld1.u8 {d8}, [r0], r1 ; p2
- vld1.u8 {d9}, [r3], r1 ; p2
- vld1.u8 {d10}, [r0], r1 ; p1
- vld1.u8 {d11}, [r3], r1 ; p1
- vld1.u8 {d12}, [r0], r1 ; p0
- vld1.u8 {d13}, [r3], r1 ; p0
- vld1.u8 {d14}, [r0], r1 ; q0
- vld1.u8 {d15}, [r3], r1 ; q0
- vld1.u8 {d16}, [r0], r1 ; q1
- vld1.u8 {d17}, [r3], r1 ; q1
- vld1.u8 {d18}, [r0], r1 ; q2
- vld1.u8 {d19}, [r3], r1 ; q2
- vld1.u8 {d20}, [r0], r1 ; q3
- vld1.u8 {d21}, [r3], r1 ; q3
-
- vld1.s8 {d4[], d5[]}, [r12] ; thresh
-
- bl vp8_mbloop_filter_neon
-
- sub r0, r0, r1, lsl #3
- sub r3, r3, r1, lsl #3
-
- add r0, r0, r1
- add r3, r3, r1
-
- vst1.u8 {d8}, [r0], r1 ; store u op2
- vst1.u8 {d9}, [r3], r1 ; store v op2
- vst1.u8 {d10}, [r0], r1 ; store u op1
- vst1.u8 {d11}, [r3], r1 ; store v op1
- vst1.u8 {d12}, [r0], r1 ; store u op0
- vst1.u8 {d13}, [r3], r1 ; store v op0
- vst1.u8 {d14}, [r0], r1 ; store u oq0
- vst1.u8 {d15}, [r3], r1 ; store v oq0
- vst1.u8 {d16}, [r0], r1 ; store u oq1
- vst1.u8 {d17}, [r3], r1 ; store v oq1
- vst1.u8 {d18}, [r0], r1 ; store u oq2
- vst1.u8 {d19}, [r3], r1 ; store v oq2
-
- ldmia sp!, {pc}
- ENDP ; |vp8_mbloop_filter_horizontal_edge_uv_neon|
-
-; void vp8_mbloop_filter_vertical_edge_y_neon(unsigned char *src, int pitch,
-; const signed char *flimit,
-; const signed char *limit,
-; const signed char *thresh,
-; int count)
-; r0 unsigned char *src,
-; r1 int pitch,
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; sp const signed char *thresh,
-; sp+4 int count (unused)
-|vp8_mbloop_filter_vertical_edge_y_neon| PROC
- stmdb sp!, {lr}
- sub r0, r0, #4 ; move src pointer down by 4 columns
-
- vld1.u8 {d6}, [r0], r1 ; load first 8-line src data
- ldr r12, [sp, #4] ; load thresh pointer
- vld1.u8 {d8}, [r0], r1
- sub sp, sp, #32
- vld1.u8 {d10}, [r0], r1
- vld1.u8 {d12}, [r0], r1
- vld1.u8 {d14}, [r0], r1
- vld1.u8 {d16}, [r0], r1
- vld1.u8 {d18}, [r0], r1
- vld1.u8 {d20}, [r0], r1
-
- vld1.u8 {d7}, [r0], r1 ; load second 8-line src data
- vld1.u8 {d9}, [r0], r1
- vld1.u8 {d11}, [r0], r1
- vld1.u8 {d13}, [r0], r1
- vld1.u8 {d15}, [r0], r1
- vld1.u8 {d17}, [r0], r1
- vld1.u8 {d19}, [r0], r1
- vld1.u8 {d21}, [r0], r1
-
- ;transpose to 8x16 matrix
- vtrn.32 q3, q7
- vtrn.32 q4, q8
- vtrn.32 q5, q9
- vtrn.32 q6, q10
-
- vtrn.16 q3, q5
- vtrn.16 q4, q6
- vtrn.16 q7, q9
- vtrn.16 q8, q10
-
- vtrn.8 q3, q4
- vtrn.8 q5, q6
- vtrn.8 q7, q8
- vtrn.8 q9, q10
-
- vld1.s8 {d4[], d5[]}, [r12] ; thresh
- vld1.s8 {d2[], d3[]}, [r3] ; limit
- mov r12, sp
- vst1.u8 {q3}, [r12]!
- vst1.u8 {q10}, [r12]!
-
- bl vp8_mbloop_filter_neon
-
- sub r0, r0, r1, lsl #4
-
- add r2, r0, r1
-
- add r3, r2, r1
-
- vld1.u8 {q3}, [sp]!
- vld1.u8 {q10}, [sp]!
-
- ;transpose to 16x8 matrix
- vtrn.32 q3, q7
- vtrn.32 q4, q8
- vtrn.32 q5, q9
- vtrn.32 q6, q10
- add r12, r3, r1
-
- vtrn.16 q3, q5
- vtrn.16 q4, q6
- vtrn.16 q7, q9
- vtrn.16 q8, q10
-
- vtrn.8 q3, q4
- vtrn.8 q5, q6
- vtrn.8 q7, q8
- vtrn.8 q9, q10
-
- ;store op2, op1, op0, oq0, oq1, oq2
- vst1.8 {d6}, [r0]
- vst1.8 {d8}, [r2]
- vst1.8 {d10}, [r3]
- vst1.8 {d12}, [r12], r1
- add r0, r12, r1
- vst1.8 {d14}, [r12]
- vst1.8 {d16}, [r0], r1
- add r2, r0, r1
- vst1.8 {d18}, [r0]
- vst1.8 {d20}, [r2], r1
- add r3, r2, r1
- vst1.8 {d7}, [r2]
- vst1.8 {d9}, [r3], r1
- add r12, r3, r1
- vst1.8 {d11}, [r3]
- vst1.8 {d13}, [r12], r1
- add r0, r12, r1
- vst1.8 {d15}, [r12]
- vst1.8 {d17}, [r0], r1
- add r2, r0, r1
- vst1.8 {d19}, [r0]
- vst1.8 {d21}, [r2]
-
- ldmia sp!, {pc}
- ENDP ; |vp8_mbloop_filter_vertical_edge_y_neon|
-
-; void vp8_mbloop_filter_vertical_edge_uv_neon(unsigned char *u, int pitch,
-; const signed char *flimit,
-; const signed char *limit,
-; const signed char *thresh,
-; unsigned char *v)
-; r0 unsigned char *u,
-; r1 int pitch,
-; r2 const signed char *flimit,
-; r3 const signed char *limit,
-; sp const signed char *thresh,
-; sp+4 unsigned char *v
-|vp8_mbloop_filter_vertical_edge_uv_neon| PROC
- stmdb sp!, {lr}
- sub r0, r0, #4 ; move src pointer down by 4 columns
- vld1.s8 {d2[], d3[]}, [r3] ; limit
- ldr r3, [sp, #8] ; load v ptr
- ldr r12, [sp, #4] ; load thresh pointer
-
- sub r3, r3, #4 ; move v pointer down by 4 columns
-
- vld1.u8 {d6}, [r0], r1 ;load u data
- vld1.u8 {d7}, [r3], r1 ;load v data
- vld1.u8 {d8}, [r0], r1
- vld1.u8 {d9}, [r3], r1
- vld1.u8 {d10}, [r0], r1
- vld1.u8 {d11}, [r3], r1
- vld1.u8 {d12}, [r0], r1
- vld1.u8 {d13}, [r3], r1
- vld1.u8 {d14}, [r0], r1
- vld1.u8 {d15}, [r3], r1
- vld1.u8 {d16}, [r0], r1
- vld1.u8 {d17}, [r3], r1
- vld1.u8 {d18}, [r0], r1
- vld1.u8 {d19}, [r3], r1
- vld1.u8 {d20}, [r0], r1
- vld1.u8 {d21}, [r3], r1
-
- ;transpose to 8x16 matrix
- vtrn.32 q3, q7
- vtrn.32 q4, q8
- vtrn.32 q5, q9
- vtrn.32 q6, q10
-
- vtrn.16 q3, q5
- vtrn.16 q4, q6
- vtrn.16 q7, q9
- vtrn.16 q8, q10
-
- vtrn.8 q3, q4
- vtrn.8 q5, q6
- vtrn.8 q7, q8
- vtrn.8 q9, q10
-
- sub sp, sp, #32
- vld1.s8 {d4[], d5[]}, [r12] ; thresh
- mov r12, sp
- vst1.u8 {q3}, [r12]!
- vst1.u8 {q10}, [r12]!
-
- bl vp8_mbloop_filter_neon
-
- sub r0, r0, r1, lsl #3
- sub r3, r3, r1, lsl #3
-
- vld1.u8 {q3}, [sp]!
- vld1.u8 {q10}, [sp]!
-
- ;transpose to 16x8 matrix
- vtrn.32 q3, q7
- vtrn.32 q4, q8
- vtrn.32 q5, q9
- vtrn.32 q6, q10
-
- vtrn.16 q3, q5
- vtrn.16 q4, q6
- vtrn.16 q7, q9
- vtrn.16 q8, q10
-
- vtrn.8 q3, q4
- vtrn.8 q5, q6
- vtrn.8 q7, q8
- vtrn.8 q9, q10
-
- ;store op2, op1, op0, oq0, oq1, oq2
- vst1.8 {d6}, [r0], r1
- vst1.8 {d7}, [r3], r1
- vst1.8 {d8}, [r0], r1
- vst1.8 {d9}, [r3], r1
- vst1.8 {d10}, [r0], r1
- vst1.8 {d11}, [r3], r1
- vst1.8 {d12}, [r0], r1
- vst1.8 {d13}, [r3], r1
- vst1.8 {d14}, [r0], r1
- vst1.8 {d15}, [r3], r1
- vst1.8 {d16}, [r0], r1
- vst1.8 {d17}, [r3], r1
- vst1.8 {d18}, [r0], r1
- vst1.8 {d19}, [r3], r1
- vst1.8 {d20}, [r0], r1
- vst1.8 {d21}, [r3], r1
-
- ldmia sp!, {pc}
- ENDP ; |vp8_mbloop_filter_vertical_edge_uv_neon|
-
-; void vp8_mbloop_filter_neon()
-; This is a helper function for the macroblock loopfilters. The individual
-; functions do the necessary load, transpose (if necessary), preserve (if
-; necessary) and store.
-
-; TODO:
-; The vertical filter writes p3/q3 back out because two 4 element writes are
-; much simpler than ordering and writing two 3 element sets (or three 2 elements
-; sets, or whichever other combinations are possible).
-; If we can preserve q3 and q10, the vertical filter will be able to avoid
-; storing those values on the stack and reading them back after the filter.
-
-; r0,r1 PRESERVE
-; r2 flimit
-; r3 PRESERVE
-; q1 limit
-; q2 thresh
-; q3 p3
-; q4 p2
-; q5 p1
-; q6 p0
-; q7 q0
-; q8 q1
-; q9 q2
-; q10 q3
-
-|vp8_mbloop_filter_neon| PROC
- adr r12, mblf_coeff
-
- ; vp8_filter_mask
- vabd.u8 q11, q3, q4 ; abs(p3 - p2)
- vabd.u8 q12, q4, q5 ; abs(p2 - p1)
- vabd.u8 q13, q5, q6 ; abs(p1 - p0)
- vabd.u8 q14, q8, q7 ; abs(q1 - q0)
- vabd.u8 q3, q9, q8 ; abs(q2 - q1)
- vabd.u8 q0, q10, q9 ; abs(q3 - q2)
-
- vmax.u8 q11, q11, q12
- vmax.u8 q12, q13, q14
- vmax.u8 q3, q3, q0
- vmax.u8 q15, q11, q12
-
- vabd.u8 q12, q6, q7 ; abs(p0 - q0)
-
- ; vp8_hevmask
- vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh) * -1
- vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh) * -1
- vmax.u8 q15, q15, q3
-
- vld1.s8 {d4[], d5[]}, [r2] ; flimit
-
- vld1.u8 {q0}, [r12]!
-
- vadd.u8 q2, q2, q2 ; flimit * 2
- vadd.u8 q2, q2, q1 ; flimit * 2 + limit
- vcge.u8 q15, q1, q15
-
- vabd.u8 q1, q5, q8 ; a = abs(p1 - q1)
- vqadd.u8 q12, q12, q12 ; b = abs(p0 - q0) * 2
- vshr.u8 q1, q1, #1 ; a = a / 2
- vqadd.u8 q12, q12, q1 ; a = b + a
- vcge.u8 q12, q2, q12 ; (a > flimit * 2 + limit) * -1
-
- ; vp8_filter
- ; convert to signed
- veor q7, q7, q0 ; qs0
- veor q6, q6, q0 ; ps0
- veor q5, q5, q0 ; ps1
- veor q8, q8, q0 ; qs1
- veor q4, q4, q0 ; ps2
- veor q9, q9, q0 ; qs2
-
- vorr q14, q13, q14 ; vp8_hevmask
-
- vsubl.s8 q2, d14, d12 ; qs0 - ps0
- vsubl.s8 q13, d15, d13
-
- vqsub.s8 q1, q5, q8 ; vp8_filter = clamp(ps1-qs1)
-
- vadd.s16 q10, q2, q2 ; 3 * (qs0 - ps0)
- vadd.s16 q11, q13, q13
- vand q15, q15, q12 ; vp8_filter_mask
-
- vadd.s16 q2, q2, q10
- vadd.s16 q13, q13, q11
-
- vld1.u8 {q12}, [r12]! ; #3
-
- vaddw.s8 q2, q2, d2 ; vp8_filter + 3 * ( qs0 - ps0)
- vaddw.s8 q13, q13, d3
-
- vld1.u8 {q11}, [r12]! ; #4
-
- ; vp8_filter = clamp(vp8_filter + 3 * ( qs0 - ps0))
- vqmovn.s16 d2, q2
- vqmovn.s16 d3, q13
-
- vand q1, q1, q15 ; vp8_filter &= mask
-
- vld1.u8 {q15}, [r12]! ; #63
- ;
- vand q13, q1, q14 ; Filter2 &= hev
-
- vld1.u8 {d7}, [r12]! ; #9
-
- vqadd.s8 q2, q13, q11 ; Filter1 = clamp(Filter2+4)
- vqadd.s8 q13, q13, q12 ; Filter2 = clamp(Filter2+3)
-
- vld1.u8 {d6}, [r12]! ; #18
-
- vshr.s8 q2, q2, #3 ; Filter1 >>= 3
- vshr.s8 q13, q13, #3 ; Filter2 >>= 3
-
- vmov q10, q15
- vmov q12, q15
-
- vqsub.s8 q7, q7, q2 ; qs0 = clamp(qs0 - Filter1)
-
- vld1.u8 {d5}, [r12]! ; #27
-
- vqadd.s8 q6, q6, q13 ; ps0 = clamp(ps0 + Filter2)
-
- vbic q1, q1, q14 ; vp8_filter &= ~hev
-
- ; roughly 1/7th difference across boundary
- ; roughly 2/7th difference across boundary
- ; roughly 3/7th difference across boundary
- vmov q11, q15
- vmov q13, q15
- vmov q14, q15
-
- vmlal.s8 q10, d2, d7 ; Filter2 * 9
- vmlal.s8 q11, d3, d7
- vmlal.s8 q12, d2, d6 ; Filter2 * 18
- vmlal.s8 q13, d3, d6
- vmlal.s8 q14, d2, d5 ; Filter2 * 27
- vmlal.s8 q15, d3, d5
- vqshrn.s16 d20, q10, #7 ; u = clamp((63 + Filter2 * 9)>>7)
- vqshrn.s16 d21, q11, #7
- vqshrn.s16 d24, q12, #7 ; u = clamp((63 + Filter2 * 18)>>7)
- vqshrn.s16 d25, q13, #7
- vqshrn.s16 d28, q14, #7 ; u = clamp((63 + Filter2 * 27)>>7)
- vqshrn.s16 d29, q15, #7
-
- vqsub.s8 q11, q9, q10 ; s = clamp(qs2 - u)
- vqadd.s8 q10, q4, q10 ; s = clamp(ps2 + u)
- vqsub.s8 q13, q8, q12 ; s = clamp(qs1 - u)
- vqadd.s8 q12, q5, q12 ; s = clamp(ps1 + u)
- vqsub.s8 q15, q7, q14 ; s = clamp(qs0 - u)
- vqadd.s8 q14, q6, q14 ; s = clamp(ps0 + u)
- veor q9, q11, q0 ; *oq2 = s^0x80
- veor q4, q10, q0 ; *op2 = s^0x80
- veor q8, q13, q0 ; *oq1 = s^0x80
- veor q5, q12, q0 ; *op2 = s^0x80
- veor q7, q15, q0 ; *oq0 = s^0x80
- veor q6, q14, q0 ; *op0 = s^0x80
-
- bx lr
- ENDP ; |vp8_mbloop_filter_neon|
-
-;-----------------
-
-mblf_coeff
- DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
- DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
- DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
- DCD 0x003f003f, 0x003f003f, 0x003f003f, 0x003f003f
- DCD 0x09090909, 0x09090909, 0x12121212, 0x12121212
- DCD 0x1b1b1b1b, 0x1b1b1b1b
-
- END
diff --git a/vp8/common/arm/neon/recon16x16mb_neon.asm b/vp8/common/arm/neon/recon16x16mb_neon.asm
deleted file mode 100644
index 3f1a30f..0000000
--- a/vp8/common/arm/neon/recon16x16mb_neon.asm
+++ /dev/null
@@ -1,131 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_recon16x16mb_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *pred_ptr,
-; r1 short *diff_ptr,
-; r2 unsigned char *dst_ptr,
-; r3 int ystride,
-; stack unsigned char *udst_ptr,
-; stack unsigned char *vdst_ptr
-
-|vp8_recon16x16mb_neon| PROC
- mov r12, #4 ;loop counter for Y loop
-
-recon16x16mb_loop_y
- vld1.u8 {q12, q13}, [r0]! ;load data from pred_ptr
- vld1.16 {q8, q9}, [r1]! ;load data from diff_ptr
- vld1.u8 {q14, q15}, [r0]!
- vld1.16 {q10, q11}, [r1]!
-
- vmovl.u8 q0, d24 ;modify Pred data from 8 bits to 16 bits
- vmovl.u8 q1, d25
- vmovl.u8 q2, d26
- vmovl.u8 q3, d27
- vmovl.u8 q4, d28
- vmovl.u8 q5, d29
- vmovl.u8 q6, d30
- vld1.16 {q12, q13}, [r1]!
- vmovl.u8 q7, d31
- vld1.16 {q14, q15}, [r1]!
-
- pld [r0]
- pld [r1]
- pld [r1, #64]
-
- vadd.s16 q0, q0, q8 ;add Diff data and Pred data together
- vadd.s16 q1, q1, q9
- vadd.s16 q2, q2, q10
- vadd.s16 q3, q3, q11
- vadd.s16 q4, q4, q12
- vadd.s16 q5, q5, q13
- vadd.s16 q6, q6, q14
- vadd.s16 q7, q7, q15
-
- vqmovun.s16 d0, q0 ;CLAMP() saturation
- vqmovun.s16 d1, q1
- vqmovun.s16 d2, q2
- vqmovun.s16 d3, q3
- vqmovun.s16 d4, q4
- vqmovun.s16 d5, q5
- vst1.u8 {q0}, [r2], r3 ;store result
- vqmovun.s16 d6, q6
- vst1.u8 {q1}, [r2], r3
- vqmovun.s16 d7, q7
- vst1.u8 {q2}, [r2], r3
- subs r12, r12, #1
-
- moveq r12, #2 ;loop counter for UV loop
-
- vst1.u8 {q3}, [r2], r3
- bne recon16x16mb_loop_y
-
- mov r3, r3, lsr #1 ;uv_stride = ystride>>1
- ldr r2, [sp] ;load upred_ptr
-
-recon16x16mb_loop_uv
- vld1.u8 {q12, q13}, [r0]! ;load data from pred_ptr
- vld1.16 {q8, q9}, [r1]! ;load data from diff_ptr
- vld1.u8 {q14, q15}, [r0]!
- vld1.16 {q10, q11}, [r1]!
-
- vmovl.u8 q0, d24 ;modify Pred data from 8 bits to 16 bits
- vmovl.u8 q1, d25
- vmovl.u8 q2, d26
- vmovl.u8 q3, d27
- vmovl.u8 q4, d28
- vmovl.u8 q5, d29
- vmovl.u8 q6, d30
- vld1.16 {q12, q13}, [r1]!
- vmovl.u8 q7, d31
- vld1.16 {q14, q15}, [r1]!
-
- vadd.s16 q0, q0, q8 ;add Diff data and Pred data together
- vadd.s16 q1, q1, q9
- vadd.s16 q2, q2, q10
- vadd.s16 q3, q3, q11
- vadd.s16 q4, q4, q12
- vadd.s16 q5, q5, q13
- vadd.s16 q6, q6, q14
-
- vqmovun.s16 d0, q0 ;CLAMP() saturation
- vadd.s16 q7, q7, q15
- vqmovun.s16 d1, q1
- vqmovun.s16 d2, q2
- vqmovun.s16 d3, q3
- vst1.u8 {d0}, [r2], r3 ;store result
- vqmovun.s16 d4, q4
- vst1.u8 {d1}, [r2], r3
- vqmovun.s16 d5, q5
- vst1.u8 {d2}, [r2], r3
- vqmovun.s16 d6, q6
- vst1.u8 {d3}, [r2], r3
- vqmovun.s16 d7, q7
- vst1.u8 {d4}, [r2], r3
- subs r12, r12, #1
-
- vst1.u8 {d5}, [r2], r3
- vst1.u8 {d6}, [r2], r3
- vst1.u8 {d7}, [r2], r3
-
- ldrne r2, [sp, #4] ;load vpred_ptr
- bne recon16x16mb_loop_uv
-
- bx lr
-
- ENDP
- END
diff --git a/vp8/common/arm/neon/recon2b_neon.asm b/vp8/common/arm/neon/recon2b_neon.asm
deleted file mode 100644
index 99b251c..0000000
--- a/vp8/common/arm/neon/recon2b_neon.asm
+++ /dev/null
@@ -1,54 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_recon2b_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *pred_ptr,
-; r1 short *diff_ptr,
-; r2 unsigned char *dst_ptr,
-; r3 int stride
-
-|vp8_recon2b_neon| PROC
- vld1.u8 {q8, q9}, [r0] ;load data from pred_ptr
- vld1.16 {q4, q5}, [r1]! ;load data from diff_ptr
-
- vmovl.u8 q0, d16 ;modify Pred data from 8 bits to 16 bits
- vld1.16 {q6, q7}, [r1]!
- vmovl.u8 q1, d17
- vmovl.u8 q2, d18
- vmovl.u8 q3, d19
-
- vadd.s16 q0, q0, q4 ;add Diff data and Pred data together
- vadd.s16 q1, q1, q5
- vadd.s16 q2, q2, q6
- vadd.s16 q3, q3, q7
-
- vqmovun.s16 d0, q0 ;CLAMP() saturation
- vqmovun.s16 d1, q1
- vqmovun.s16 d2, q2
- vqmovun.s16 d3, q3
- add r0, r2, r3
-
- vst1.u8 {d0}, [r2] ;store result
- vst1.u8 {d1}, [r0], r3
- add r2, r0, r3
- vst1.u8 {d2}, [r0]
- vst1.u8 {d3}, [r2], r3
-
- bx lr
-
- ENDP
- END
diff --git a/vp8/common/arm/neon/recon4b_neon.asm b/vp8/common/arm/neon/recon4b_neon.asm
deleted file mode 100644
index 9917277..0000000
--- a/vp8/common/arm/neon/recon4b_neon.asm
+++ /dev/null
@@ -1,69 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_recon4b_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *pred_ptr,
-; r1 short *diff_ptr,
-; r2 unsigned char *dst_ptr,
-; r3 int stride
-
-|vp8_recon4b_neon| PROC
- vld1.u8 {q12, q13}, [r0]! ;load data from pred_ptr
- vld1.16 {q8, q9}, [r1]! ;load data from diff_ptr
- vld1.u8 {q14, q15}, [r0]
- vld1.16 {q10, q11}, [r1]!
-
- vmovl.u8 q0, d24 ;modify Pred data from 8 bits to 16 bits
- vmovl.u8 q1, d25
- vmovl.u8 q2, d26
- vmovl.u8 q3, d27
- vmovl.u8 q4, d28
- vmovl.u8 q5, d29
- vmovl.u8 q6, d30
- vld1.16 {q12, q13}, [r1]!
- vmovl.u8 q7, d31
- vld1.16 {q14, q15}, [r1]
-
- vadd.s16 q0, q0, q8 ;add Diff data and Pred data together
- vadd.s16 q1, q1, q9
- vadd.s16 q2, q2, q10
- vadd.s16 q3, q3, q11
- vadd.s16 q4, q4, q12
- vadd.s16 q5, q5, q13
- vadd.s16 q6, q6, q14
- vadd.s16 q7, q7, q15
-
- vqmovun.s16 d0, q0 ;CLAMP() saturation
- vqmovun.s16 d1, q1
- vqmovun.s16 d2, q2
- vqmovun.s16 d3, q3
- vqmovun.s16 d4, q4
- vqmovun.s16 d5, q5
- vqmovun.s16 d6, q6
- vqmovun.s16 d7, q7
- add r0, r2, r3
-
- vst1.u8 {q0}, [r2] ;store result
- vst1.u8 {q1}, [r0], r3
- add r2, r0, r3
- vst1.u8 {q2}, [r0]
- vst1.u8 {q3}, [r2], r3
-
- bx lr
-
- ENDP
- END
diff --git a/vp8/common/arm/neon/recon_neon.c b/vp8/common/arm/neon/recon_neon.c
deleted file mode 100644
index 09fd2a5..0000000
--- a/vp8/common/arm/neon/recon_neon.c
+++ /dev/null
@@ -1,29 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/common/recon.h"
-#include "vp8/common/blockd.h"
-
-extern void vp8_recon16x16mb_neon(unsigned char *pred_ptr, short *diff_ptr, unsigned char *dst_ptr, int ystride, unsigned char *udst_ptr, unsigned char *vdst_ptr);
-
-void vp8_recon_mb_neon(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
- unsigned char *pred_ptr = &x->predictor[0];
- short *diff_ptr = &x->diff[0];
- unsigned char *dst_ptr = x->dst.y_buffer;
- unsigned char *udst_ptr = x->dst.u_buffer;
- unsigned char *vdst_ptr = x->dst.v_buffer;
- int ystride = x->dst.y_stride;
- /*int uv_stride = x->dst.uv_stride;*/
-
- vp8_recon16x16mb_neon(pred_ptr, diff_ptr, dst_ptr, ystride, udst_ptr, vdst_ptr);
-}
diff --git a/vp8/common/arm/neon/reconb_neon.asm b/vp8/common/arm/neon/reconb_neon.asm
deleted file mode 100644
index 288c0ef..0000000
--- a/vp8/common/arm/neon/reconb_neon.asm
+++ /dev/null
@@ -1,61 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_recon_b_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *pred_ptr,
-; r1 short *diff_ptr,
-; r2 unsigned char *dst_ptr,
-; r3 int stride
-
-|vp8_recon_b_neon| PROC
- mov r12, #16
-
- vld1.u8 {d28}, [r0], r12 ;load 4 data/line from pred_ptr
- vld1.16 {q10, q11}, [r1]! ;load data from diff_ptr
- vld1.u8 {d29}, [r0], r12
- vld1.16 {q11, q12}, [r1]!
- vld1.u8 {d30}, [r0], r12
- vld1.16 {q12, q13}, [r1]!
- vld1.u8 {d31}, [r0], r12
- vld1.16 {q13}, [r1]
-
- vmovl.u8 q0, d28 ;modify Pred data from 8 bits to 16 bits
- vmovl.u8 q1, d29 ;Pred data in d0, d2, d4, d6
- vmovl.u8 q2, d30
- vmovl.u8 q3, d31
-
- vadd.s16 d0, d0, d20 ;add Diff data and Pred data together
- vadd.s16 d2, d2, d22
- vadd.s16 d4, d4, d24
- vadd.s16 d6, d6, d26
-
- vqmovun.s16 d0, q0 ;CLAMP() saturation
- vqmovun.s16 d1, q1
- vqmovun.s16 d2, q2
- vqmovun.s16 d3, q3
- add r1, r2, r3
-
- vst1.32 {d0[0]}, [r2] ;store result
- vst1.32 {d1[0]}, [r1], r3
- add r2, r1, r3
- vst1.32 {d2[0]}, [r1]
- vst1.32 {d3[0]}, [r2], r3
-
- bx lr
-
- ENDP
- END
diff --git a/vp8/common/arm/neon/shortidct4x4llm_1_neon.asm b/vp8/common/arm/neon/shortidct4x4llm_1_neon.asm
deleted file mode 100644
index d7bdbae..0000000
--- a/vp8/common/arm/neon/shortidct4x4llm_1_neon.asm
+++ /dev/null
@@ -1,67 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_short_idct4x4llm_1_neon|
- EXPORT |vp8_dc_only_idct_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
-;void vp8_short_idct4x4llm_1_c(short *input, short *output, int pitch);
-; r0 short *input;
-; r1 short *output;
-; r2 int pitch;
-;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
-|vp8_short_idct4x4llm_1_neon| PROC
- vld1.16 {d0[]}, [r0] ;load input[0]
-
- add r3, r1, r2
- add r12, r3, r2
-
- vrshr.s16 d0, d0, #3
-
- add r0, r12, r2
-
- vst1.16 {d0}, [r1]
- vst1.16 {d0}, [r3]
- vst1.16 {d0}, [r12]
- vst1.16 {d0}, [r0]
-
- bx lr
- ENDP
-
-;;;;;;;;;;;;;;;;;;;;;;;;;;;;
-;void vp8_dc_only_idct_c(short input_dc, short *output, int pitch);
-; r0 short input_dc;
-; r1 short *output;
-; r2 int pitch;
-;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
-|vp8_dc_only_idct_neon| PROC
- vdup.16 d0, r0
-
- add r3, r1, r2
- add r12, r3, r2
-
- vrshr.s16 d0, d0, #3
-
- add r0, r12, r2
-
- vst1.16 {d0}, [r1]
- vst1.16 {d0}, [r3]
- vst1.16 {d0}, [r12]
- vst1.16 {d0}, [r0]
-
- bx lr
-
- ENDP
- END
diff --git a/vp8/common/arm/neon/shortidct4x4llm_neon.asm b/vp8/common/arm/neon/shortidct4x4llm_neon.asm
deleted file mode 100644
index b74c315..0000000
--- a/vp8/common/arm/neon/shortidct4x4llm_neon.asm
+++ /dev/null
@@ -1,122 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_short_idct4x4llm_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-;*************************************************************
-;void vp8_short_idct4x4llm_c(short *input, short *output, int pitch)
-;r0 short * input
-;r1 short * output
-;r2 int pitch
-;*************************************************************
-;static const int cospi8sqrt2minus1=20091;
-;static const int sinpi8sqrt2 =35468;
-;static const int rounding = 0;
-;Optimization note: The resulted data from dequantization are signed 13-bit data that is
-;in the range of [-4096, 4095]. This allows to use "vqdmulh"(neon) instruction since
-;it won't go out of range (13+16+1=30bits<32bits). This instruction gives the high half
-;result of the multiplication that is needed in IDCT.
-
-|vp8_short_idct4x4llm_neon| PROC
- adr r12, idct_coeff
- vld1.16 {q1, q2}, [r0]
- vld1.16 {d0}, [r12]
-
- vswp d3, d4 ;q2(vp[4] vp[12])
-
- vqdmulh.s16 q3, q2, d0[2]
- vqdmulh.s16 q4, q2, d0[0]
-
- vqadd.s16 d12, d2, d3 ;a1
- vqsub.s16 d13, d2, d3 ;b1
-
- vshr.s16 q3, q3, #1
- vshr.s16 q4, q4, #1
-
- vqadd.s16 q3, q3, q2 ;modify since sinpi8sqrt2 > 65536/2 (negtive number)
- vqadd.s16 q4, q4, q2
-
- ;d6 - c1:temp1
- ;d7 - d1:temp2
- ;d8 - d1:temp1
- ;d9 - c1:temp2
-
- vqsub.s16 d10, d6, d9 ;c1
- vqadd.s16 d11, d7, d8 ;d1
-
- vqadd.s16 d2, d12, d11
- vqadd.s16 d3, d13, d10
- vqsub.s16 d4, d13, d10
- vqsub.s16 d5, d12, d11
-
- vtrn.32 d2, d4
- vtrn.32 d3, d5
- vtrn.16 d2, d3
- vtrn.16 d4, d5
-
- vswp d3, d4
-
- vqdmulh.s16 q3, q2, d0[2]
- vqdmulh.s16 q4, q2, d0[0]
-
- vqadd.s16 d12, d2, d3 ;a1
- vqsub.s16 d13, d2, d3 ;b1
-
- vshr.s16 q3, q3, #1
- vshr.s16 q4, q4, #1
-
- vqadd.s16 q3, q3, q2 ;modify since sinpi8sqrt2 > 65536/2 (negtive number)
- vqadd.s16 q4, q4, q2
-
- vqsub.s16 d10, d6, d9 ;c1
- vqadd.s16 d11, d7, d8 ;d1
-
- vqadd.s16 d2, d12, d11
- vqadd.s16 d3, d13, d10
- vqsub.s16 d4, d13, d10
- vqsub.s16 d5, d12, d11
-
- vrshr.s16 d2, d2, #3
- vrshr.s16 d3, d3, #3
- vrshr.s16 d4, d4, #3
- vrshr.s16 d5, d5, #3
-
- add r3, r1, r2
- add r12, r3, r2
- add r0, r12, r2
-
- vtrn.32 d2, d4
- vtrn.32 d3, d5
- vtrn.16 d2, d3
- vtrn.16 d4, d5
-
- vst1.16 {d2}, [r1]
- vst1.16 {d3}, [r3]
- vst1.16 {d4}, [r12]
- vst1.16 {d5}, [r0]
-
- bx lr
-
- ENDP
-
-;-----------------
-
-idct_coeff
- DCD 0x4e7b4e7b, 0x8a8c8a8c
-
-;20091, 20091, 35468, 35468
-
- END
diff --git a/vp8/common/arm/neon/sixtappredict16x16_neon.asm b/vp8/common/arm/neon/sixtappredict16x16_neon.asm
deleted file mode 100644
index 438999b..0000000
--- a/vp8/common/arm/neon/sixtappredict16x16_neon.asm
+++ /dev/null
@@ -1,490 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_sixtap_predict16x16_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-filter16_coeff
- DCD 0, 0, 128, 0, 0, 0, 0, 0
- DCD 0, -6, 123, 12, -1, 0, 0, 0
- DCD 2, -11, 108, 36, -8, 1, 0, 0
- DCD 0, -9, 93, 50, -6, 0, 0, 0
- DCD 3, -16, 77, 77, -16, 3, 0, 0
- DCD 0, -6, 50, 93, -9, 0, 0, 0
- DCD 1, -8, 36, 108, -11, 2, 0, 0
- DCD 0, -1, 12, 123, -6, 0, 0, 0
-
-;-----------------
-; r0 unsigned char *src_ptr,
-; r1 int src_pixels_per_line,
-; r2 int xoffset,
-; r3 int yoffset,
-; r4 unsigned char *dst_ptr,
-; stack(r5) int dst_pitch
-
-;Note: To take advantage of 8-bit mulplication instruction in NEON. First apply abs() to
-; filter coeffs to make them u8. Then, use vmlsl for negtive coeffs. After multiplication,
-; the result can be negtive. So, I treat the result as s16. But, since it is also possible
-; that the result can be a large positive number (> 2^15-1), which could be confused as a
-; negtive number. To avoid that error, apply filter coeffs in the order of 0, 1, 4 ,5 ,2,
-; which ensures that the result stays in s16 range. Finally, saturated add the result by
-; applying 3rd filter coeff. Same applys to other filter functions.
-
-|vp8_sixtap_predict16x16_neon| PROC
- push {r4-r5, lr}
-
- adr r12, filter16_coeff
- ldr r4, [sp, #12] ;load parameters from stack
- ldr r5, [sp, #16] ;load parameters from stack
-
- cmp r2, #0 ;skip first_pass filter if xoffset=0
- beq secondpass_filter16x16_only
-
- add r2, r12, r2, lsl #5 ;calculate filter location
-
- cmp r3, #0 ;skip second_pass filter if yoffset=0
-
- vld1.s32 {q14, q15}, [r2] ;load first_pass filter
-
- beq firstpass_filter16x16_only
-
- sub sp, sp, #336 ;reserve space on stack for temporary storage
- mov lr, sp
-
- vabs.s32 q12, q14
- vabs.s32 q13, q15
-
- mov r2, #7 ;loop counter
- sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
- sub r0, r0, r1, lsl #1
-
- vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
- vdup.8 d1, d24[4]
- vdup.8 d2, d25[0]
- vdup.8 d3, d25[4]
- vdup.8 d4, d26[0]
- vdup.8 d5, d26[4]
-
-;First Pass: output_height lines x output_width columns (21x16)
-filt_blk2d_fp16x16_loop_neon
- vld1.u8 {d6, d7, d8}, [r0], r1 ;load src data
- vld1.u8 {d9, d10, d11}, [r0], r1
- vld1.u8 {d12, d13, d14}, [r0], r1
-
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- vmull.u8 q8, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q9, d7, d0
- vmull.u8 q10, d9, d0
- vmull.u8 q11, d10, d0
- vmull.u8 q12, d12, d0
- vmull.u8 q13, d13, d0
-
- vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
- vext.8 d29, d9, d10, #1
- vext.8 d30, d12, d13, #1
-
- vmlsl.u8 q8, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q10, d29, d1
- vmlsl.u8 q12, d30, d1
-
- vext.8 d28, d7, d8, #1
- vext.8 d29, d10, d11, #1
- vext.8 d30, d13, d14, #1
-
- vmlsl.u8 q9, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q11, d29, d1
- vmlsl.u8 q13, d30, d1
-
- vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
- vext.8 d29, d9, d10, #4
- vext.8 d30, d12, d13, #4
-
- vmlsl.u8 q8, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q10, d29, d4
- vmlsl.u8 q12, d30, d4
-
- vext.8 d28, d7, d8, #4
- vext.8 d29, d10, d11, #4
- vext.8 d30, d13, d14, #4
-
- vmlsl.u8 q9, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q11, d29, d4
- vmlsl.u8 q13, d30, d4
-
- vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d29, d9, d10, #5
- vext.8 d30, d12, d13, #5
-
- vmlal.u8 q8, d28, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q10, d29, d5
- vmlal.u8 q12, d30, d5
-
- vext.8 d28, d7, d8, #5
- vext.8 d29, d10, d11, #5
- vext.8 d30, d13, d14, #5
-
- vmlal.u8 q9, d28, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q11, d29, d5
- vmlal.u8 q13, d30, d5
-
- vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
- vext.8 d29, d9, d10, #2
- vext.8 d30, d12, d13, #2
-
- vmlal.u8 q8, d28, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q10, d29, d2
- vmlal.u8 q12, d30, d2
-
- vext.8 d28, d7, d8, #2
- vext.8 d29, d10, d11, #2
- vext.8 d30, d13, d14, #2
-
- vmlal.u8 q9, d28, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q11, d29, d2
- vmlal.u8 q13, d30, d2
-
- vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
- vext.8 d29, d9, d10, #3
- vext.8 d30, d12, d13, #3
-
- vext.8 d15, d7, d8, #3
- vext.8 d31, d10, d11, #3
- vext.8 d6, d13, d14, #3
-
- vmull.u8 q4, d28, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q5, d29, d3
- vmull.u8 q6, d30, d3
-
- vqadd.s16 q8, q4 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q10, q5
- vqadd.s16 q12, q6
-
- vmull.u8 q6, d15, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q7, d31, d3
- vmull.u8 q3, d6, d3
-
- subs r2, r2, #1
-
- vqadd.s16 q9, q6
- vqadd.s16 q11, q7
- vqadd.s16 q13, q3
-
- vqrshrun.s16 d6, q8, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d7, q9, #7
- vqrshrun.s16 d8, q10, #7
- vqrshrun.s16 d9, q11, #7
- vqrshrun.s16 d10, q12, #7
- vqrshrun.s16 d11, q13, #7
-
- vst1.u8 {d6, d7, d8}, [lr]! ;store result
- vst1.u8 {d9, d10, d11}, [lr]!
-
- bne filt_blk2d_fp16x16_loop_neon
-
-;Second pass: 16x16
-;secondpass_filter - do first 8-columns and then second 8-columns
- add r3, r12, r3, lsl #5
- sub lr, lr, #336
-
- vld1.s32 {q5, q6}, [r3] ;load second_pass filter
- mov r3, #2 ;loop counter
-
- vabs.s32 q7, q5
- vabs.s32 q8, q6
-
- mov r2, #16
-
- vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
- vdup.8 d1, d14[4]
- vdup.8 d2, d15[0]
- vdup.8 d3, d15[4]
- vdup.8 d4, d16[0]
- vdup.8 d5, d16[4]
-
-filt_blk2d_sp16x16_outloop_neon
- vld1.u8 {d18}, [lr], r2 ;load src data
- vld1.u8 {d19}, [lr], r2
- vld1.u8 {d20}, [lr], r2
- vld1.u8 {d21}, [lr], r2
- mov r12, #4 ;loop counter
- vld1.u8 {d22}, [lr], r2
-
-secondpass_inner_loop_neon
- vld1.u8 {d23}, [lr], r2 ;load src data
- vld1.u8 {d24}, [lr], r2
- vld1.u8 {d25}, [lr], r2
- vld1.u8 {d26}, [lr], r2
-
- vmull.u8 q3, d18, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q4, d19, d0
- vmull.u8 q5, d20, d0
- vmull.u8 q6, d21, d0
-
- vmlsl.u8 q3, d19, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q4, d20, d1
- vmlsl.u8 q5, d21, d1
- vmlsl.u8 q6, d22, d1
-
- vmlsl.u8 q3, d22, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q4, d23, d4
- vmlsl.u8 q5, d24, d4
- vmlsl.u8 q6, d25, d4
-
- vmlal.u8 q3, d20, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q4, d21, d2
- vmlal.u8 q5, d22, d2
- vmlal.u8 q6, d23, d2
-
- vmlal.u8 q3, d23, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q4, d24, d5
- vmlal.u8 q5, d25, d5
- vmlal.u8 q6, d26, d5
-
- vmull.u8 q7, d21, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q8, d22, d3
- vmull.u8 q9, d23, d3
- vmull.u8 q10, d24, d3
-
- subs r12, r12, #1
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d7, q8, #7
- vqrshrun.s16 d8, q9, #7
- vqrshrun.s16 d9, q10, #7
-
- vst1.u8 {d6}, [r4], r5 ;store result
- vmov q9, q11
- vst1.u8 {d7}, [r4], r5
- vmov q10, q12
- vst1.u8 {d8}, [r4], r5
- vmov d22, d26
- vst1.u8 {d9}, [r4], r5
-
- bne secondpass_inner_loop_neon
-
- subs r3, r3, #1
- sub lr, lr, #336
- add lr, lr, #8
-
- sub r4, r4, r5, lsl #4
- add r4, r4, #8
-
- bne filt_blk2d_sp16x16_outloop_neon
-
- add sp, sp, #336
- pop {r4-r5,pc}
-
-;--------------------
-firstpass_filter16x16_only
- vabs.s32 q12, q14
- vabs.s32 q13, q15
-
- mov r2, #8 ;loop counter
- sub r0, r0, #2 ;move srcptr back to (column-2)
-
- vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
- vdup.8 d1, d24[4]
- vdup.8 d2, d25[0]
- vdup.8 d3, d25[4]
- vdup.8 d4, d26[0]
- vdup.8 d5, d26[4]
-
-;First Pass: output_height lines x output_width columns (16x16)
-filt_blk2d_fpo16x16_loop_neon
- vld1.u8 {d6, d7, d8}, [r0], r1 ;load src data
- vld1.u8 {d9, d10, d11}, [r0], r1
-
- pld [r0]
- pld [r0, r1]
-
- vmull.u8 q6, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q7, d7, d0
- vmull.u8 q8, d9, d0
- vmull.u8 q9, d10, d0
-
- vext.8 d20, d6, d7, #1 ;construct src_ptr[-1]
- vext.8 d21, d9, d10, #1
- vext.8 d22, d7, d8, #1
- vext.8 d23, d10, d11, #1
- vext.8 d24, d6, d7, #4 ;construct src_ptr[2]
- vext.8 d25, d9, d10, #4
- vext.8 d26, d7, d8, #4
- vext.8 d27, d10, d11, #4
- vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d29, d9, d10, #5
-
- vmlsl.u8 q6, d20, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q8, d21, d1
- vmlsl.u8 q7, d22, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q9, d23, d1
- vmlsl.u8 q6, d24, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q8, d25, d4
- vmlsl.u8 q7, d26, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q9, d27, d4
- vmlal.u8 q6, d28, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q8, d29, d5
-
- vext.8 d20, d7, d8, #5
- vext.8 d21, d10, d11, #5
- vext.8 d22, d6, d7, #2 ;construct src_ptr[0]
- vext.8 d23, d9, d10, #2
- vext.8 d24, d7, d8, #2
- vext.8 d25, d10, d11, #2
-
- vext.8 d26, d6, d7, #3 ;construct src_ptr[1]
- vext.8 d27, d9, d10, #3
- vext.8 d28, d7, d8, #3
- vext.8 d29, d10, d11, #3
-
- vmlal.u8 q7, d20, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q9, d21, d5
- vmlal.u8 q6, d22, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q8, d23, d2
- vmlal.u8 q7, d24, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q9, d25, d2
-
- vmull.u8 q10, d26, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q11, d27, d3
- vmull.u8 q12, d28, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q15, d29, d3
-
- vqadd.s16 q6, q10 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q11
- vqadd.s16 q7, q12
- vqadd.s16 q9, q15
-
- subs r2, r2, #1
-
- vqrshrun.s16 d6, q6, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d7, q7, #7
- vqrshrun.s16 d8, q8, #7
- vqrshrun.s16 d9, q9, #7
-
- vst1.u8 {q3}, [r4], r5 ;store result
- vst1.u8 {q4}, [r4], r5
-
- bne filt_blk2d_fpo16x16_loop_neon
-
- pop {r4-r5,pc}
-
-;--------------------
-secondpass_filter16x16_only
-;Second pass: 16x16
- add r3, r12, r3, lsl #5
- sub r0, r0, r1, lsl #1
-
- vld1.s32 {q5, q6}, [r3] ;load second_pass filter
- mov r3, #2 ;loop counter
-
- vabs.s32 q7, q5
- vabs.s32 q8, q6
-
- vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
- vdup.8 d1, d14[4]
- vdup.8 d2, d15[0]
- vdup.8 d3, d15[4]
- vdup.8 d4, d16[0]
- vdup.8 d5, d16[4]
-
-filt_blk2d_spo16x16_outloop_neon
- vld1.u8 {d18}, [r0], r1 ;load src data
- vld1.u8 {d19}, [r0], r1
- vld1.u8 {d20}, [r0], r1
- vld1.u8 {d21}, [r0], r1
- mov r12, #4 ;loop counter
- vld1.u8 {d22}, [r0], r1
-
-secondpass_only_inner_loop_neon
- vld1.u8 {d23}, [r0], r1 ;load src data
- vld1.u8 {d24}, [r0], r1
- vld1.u8 {d25}, [r0], r1
- vld1.u8 {d26}, [r0], r1
-
- vmull.u8 q3, d18, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q4, d19, d0
- vmull.u8 q5, d20, d0
- vmull.u8 q6, d21, d0
-
- vmlsl.u8 q3, d19, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q4, d20, d1
- vmlsl.u8 q5, d21, d1
- vmlsl.u8 q6, d22, d1
-
- vmlsl.u8 q3, d22, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q4, d23, d4
- vmlsl.u8 q5, d24, d4
- vmlsl.u8 q6, d25, d4
-
- vmlal.u8 q3, d20, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q4, d21, d2
- vmlal.u8 q5, d22, d2
- vmlal.u8 q6, d23, d2
-
- vmlal.u8 q3, d23, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q4, d24, d5
- vmlal.u8 q5, d25, d5
- vmlal.u8 q6, d26, d5
-
- vmull.u8 q7, d21, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q8, d22, d3
- vmull.u8 q9, d23, d3
- vmull.u8 q10, d24, d3
-
- subs r12, r12, #1
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d7, q8, #7
- vqrshrun.s16 d8, q9, #7
- vqrshrun.s16 d9, q10, #7
-
- vst1.u8 {d6}, [r4], r5 ;store result
- vmov q9, q11
- vst1.u8 {d7}, [r4], r5
- vmov q10, q12
- vst1.u8 {d8}, [r4], r5
- vmov d22, d26
- vst1.u8 {d9}, [r4], r5
-
- bne secondpass_only_inner_loop_neon
-
- subs r3, r3, #1
- sub r0, r0, r1, lsl #4
- sub r0, r0, r1, lsl #2
- sub r0, r0, r1
- add r0, r0, #8
-
- sub r4, r4, r5, lsl #4
- add r4, r4, #8
-
- bne filt_blk2d_spo16x16_outloop_neon
-
- pop {r4-r5,pc}
-
- ENDP
-
- END
diff --git a/vp8/common/arm/neon/sixtappredict4x4_neon.asm b/vp8/common/arm/neon/sixtappredict4x4_neon.asm
deleted file mode 100644
index 540987a..0000000
--- a/vp8/common/arm/neon/sixtappredict4x4_neon.asm
+++ /dev/null
@@ -1,421 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_sixtap_predict_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-filter4_coeff
- DCD 0, 0, 128, 0, 0, 0, 0, 0
- DCD 0, -6, 123, 12, -1, 0, 0, 0
- DCD 2, -11, 108, 36, -8, 1, 0, 0
- DCD 0, -9, 93, 50, -6, 0, 0, 0
- DCD 3, -16, 77, 77, -16, 3, 0, 0
- DCD 0, -6, 50, 93, -9, 0, 0, 0
- DCD 1, -8, 36, 108, -11, 2, 0, 0
- DCD 0, -1, 12, 123, -6, 0, 0, 0
-
-;-----------------
-; r0 unsigned char *src_ptr,
-; r1 int src_pixels_per_line,
-; r2 int xoffset,
-; r3 int yoffset,
-; stack(r4) unsigned char *dst_ptr,
-; stack(lr) int dst_pitch
-
-|vp8_sixtap_predict_neon| PROC
- push {r4, lr}
-
- adr r12, filter4_coeff
- ldr r4, [sp, #8] ;load parameters from stack
- ldr lr, [sp, #12] ;load parameters from stack
-
- cmp r2, #0 ;skip first_pass filter if xoffset=0
- beq secondpass_filter4x4_only
-
- add r2, r12, r2, lsl #5 ;calculate filter location
-
- cmp r3, #0 ;skip second_pass filter if yoffset=0
- vld1.s32 {q14, q15}, [r2] ;load first_pass filter
-
- beq firstpass_filter4x4_only
-
- vabs.s32 q12, q14 ;get abs(filer_parameters)
- vabs.s32 q13, q15
-
- sub r0, r0, #2 ;go back 2 columns of src data
- sub r0, r0, r1, lsl #1 ;go back 2 lines of src data
-
-;First pass: output_height lines x output_width columns (9x4)
- vld1.u8 {q3}, [r0], r1 ;load first 4-line src data
- vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
- vld1.u8 {q4}, [r0], r1
- vdup.8 d1, d24[4]
- vld1.u8 {q5}, [r0], r1
- vdup.8 d2, d25[0]
- vld1.u8 {q6}, [r0], r1
- vdup.8 d3, d25[4]
- vdup.8 d4, d26[0]
- vdup.8 d5, d26[4]
-
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- vext.8 d18, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d19, d8, d9, #5
- vext.8 d20, d10, d11, #5
- vext.8 d21, d12, d13, #5
-
- vswp d7, d8 ;discard 2nd half data after src_ptr[3] is done
- vswp d11, d12
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[3])
- vzip.32 d20, d21
- vmull.u8 q7, d18, d5 ;(src_ptr[3] * vp8_filter[5])
- vmull.u8 q8, d20, d5
-
- vmov q4, q3 ;keep original src data in q4 q6
- vmov q6, q5
-
- vzip.32 d6, d7 ;construct src_ptr[-2], and put 2-line data together
- vzip.32 d10, d11
- vshr.u64 q9, q4, #8 ;construct src_ptr[-1]
- vshr.u64 q10, q6, #8
- vmlal.u8 q7, d6, d0 ;+(src_ptr[-2] * vp8_filter[0])
- vmlal.u8 q8, d10, d0
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[-1])
- vzip.32 d20, d21
- vshr.u64 q3, q4, #32 ;construct src_ptr[2]
- vshr.u64 q5, q6, #32
- vmlsl.u8 q7, d18, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q8, d20, d1
-
- vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[2])
- vzip.32 d10, d11
- vshr.u64 q9, q4, #16 ;construct src_ptr[0]
- vshr.u64 q10, q6, #16
- vmlsl.u8 q7, d6, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q8, d10, d4
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[0])
- vzip.32 d20, d21
- vshr.u64 q3, q4, #24 ;construct src_ptr[1]
- vshr.u64 q5, q6, #24
- vmlal.u8 q7, d18, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q8, d20, d2
-
- vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[1])
- vzip.32 d10, d11
- vmull.u8 q9, d6, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q10, d10, d3
-
- vld1.u8 {q3}, [r0], r1 ;load rest 5-line src data
- vld1.u8 {q4}, [r0], r1
-
- vqadd.s16 q7, q9 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q10
-
- vld1.u8 {q5}, [r0], r1
- vld1.u8 {q6}, [r0], r1
-
- vqrshrun.s16 d27, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d28, q8, #7
-
- ;First Pass on rest 5-line data
- vld1.u8 {q11}, [r0], r1
-
- vext.8 d18, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d19, d8, d9, #5
- vext.8 d20, d10, d11, #5
- vext.8 d21, d12, d13, #5
-
- vswp d7, d8 ;discard 2nd half data after src_ptr[3] is done
- vswp d11, d12
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[3])
- vzip.32 d20, d21
- vext.8 d31, d22, d23, #5 ;construct src_ptr[3]
- vmull.u8 q7, d18, d5 ;(src_ptr[3] * vp8_filter[5])
- vmull.u8 q8, d20, d5
- vmull.u8 q12, d31, d5 ;(src_ptr[3] * vp8_filter[5])
-
- vmov q4, q3 ;keep original src data in q4 q6
- vmov q6, q5
-
- vzip.32 d6, d7 ;construct src_ptr[-2], and put 2-line data together
- vzip.32 d10, d11
- vshr.u64 q9, q4, #8 ;construct src_ptr[-1]
- vshr.u64 q10, q6, #8
-
- vmlal.u8 q7, d6, d0 ;+(src_ptr[-2] * vp8_filter[0])
- vmlal.u8 q8, d10, d0
- vmlal.u8 q12, d22, d0 ;(src_ptr[-2] * vp8_filter[0])
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[-1])
- vzip.32 d20, d21
- vshr.u64 q3, q4, #32 ;construct src_ptr[2]
- vshr.u64 q5, q6, #32
- vext.8 d31, d22, d23, #1 ;construct src_ptr[-1]
-
- vmlsl.u8 q7, d18, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q8, d20, d1
- vmlsl.u8 q12, d31, d1 ;-(src_ptr[-1] * vp8_filter[1])
-
- vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[2])
- vzip.32 d10, d11
- vshr.u64 q9, q4, #16 ;construct src_ptr[0]
- vshr.u64 q10, q6, #16
- vext.8 d31, d22, d23, #4 ;construct src_ptr[2]
-
- vmlsl.u8 q7, d6, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q8, d10, d4
- vmlsl.u8 q12, d31, d4 ;-(src_ptr[2] * vp8_filter[4])
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[0])
- vzip.32 d20, d21
- vshr.u64 q3, q4, #24 ;construct src_ptr[1]
- vshr.u64 q5, q6, #24
- vext.8 d31, d22, d23, #2 ;construct src_ptr[0]
-
- vmlal.u8 q7, d18, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q8, d20, d2
- vmlal.u8 q12, d31, d2 ;(src_ptr[0] * vp8_filter[2])
-
- vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[1])
- vzip.32 d10, d11
- vext.8 d31, d22, d23, #3 ;construct src_ptr[1]
- vmull.u8 q9, d6, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q10, d10, d3
- vmull.u8 q11, d31, d3 ;(src_ptr[1] * vp8_filter[3])
-
- add r3, r12, r3, lsl #5
-
- vqadd.s16 q7, q9 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q10
- vqadd.s16 q12, q11
-
- vext.8 d23, d27, d28, #4
- vld1.s32 {q5, q6}, [r3] ;load second_pass filter
-
- vqrshrun.s16 d29, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d30, q8, #7
- vqrshrun.s16 d31, q12, #7
-
-;Second pass: 4x4
- vabs.s32 q7, q5
- vabs.s32 q8, q6
-
- vext.8 d24, d28, d29, #4
- vext.8 d25, d29, d30, #4
- vext.8 d26, d30, d31, #4
-
- vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
- vdup.8 d1, d14[4]
- vdup.8 d2, d15[0]
- vdup.8 d3, d15[4]
- vdup.8 d4, d16[0]
- vdup.8 d5, d16[4]
-
- vmull.u8 q3, d27, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q4, d28, d0
-
- vmull.u8 q5, d25, d5 ;(src_ptr[3] * vp8_filter[5])
- vmull.u8 q6, d26, d5
-
- vmlsl.u8 q3, d29, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q4, d30, d4
-
- vmlsl.u8 q5, d23, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q6, d24, d1
-
- vmlal.u8 q3, d28, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q4, d29, d2
-
- vmlal.u8 q5, d24, d3 ;(src_ptr[1] * vp8_filter[3])
- vmlal.u8 q6, d25, d3
-
- add r0, r4, lr
- add r1, r0, lr
- add r2, r1, lr
-
- vqadd.s16 q5, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q6, q4
-
- vqrshrun.s16 d3, q5, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d4, q6, #7
-
- vst1.32 {d3[0]}, [r4] ;store result
- vst1.32 {d3[1]}, [r0]
- vst1.32 {d4[0]}, [r1]
- vst1.32 {d4[1]}, [r2]
-
- pop {r4, pc}
-
-
-;---------------------
-firstpass_filter4x4_only
- vabs.s32 q12, q14 ;get abs(filer_parameters)
- vabs.s32 q13, q15
-
- sub r0, r0, #2 ;go back 2 columns of src data
-
-;First pass: output_height lines x output_width columns (4x4)
- vld1.u8 {q3}, [r0], r1 ;load first 4-line src data
- vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
- vld1.u8 {q4}, [r0], r1
- vdup.8 d1, d24[4]
- vld1.u8 {q5}, [r0], r1
- vdup.8 d2, d25[0]
- vld1.u8 {q6}, [r0], r1
-
- vdup.8 d3, d25[4]
- vdup.8 d4, d26[0]
- vdup.8 d5, d26[4]
-
- vext.8 d18, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d19, d8, d9, #5
- vext.8 d20, d10, d11, #5
- vext.8 d21, d12, d13, #5
-
- vswp d7, d8 ;discard 2nd half data after src_ptr[3] is done
- vswp d11, d12
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[3])
- vzip.32 d20, d21
- vmull.u8 q7, d18, d5 ;(src_ptr[3] * vp8_filter[5])
- vmull.u8 q8, d20, d5
-
- vmov q4, q3 ;keep original src data in q4 q6
- vmov q6, q5
-
- vzip.32 d6, d7 ;construct src_ptr[-2], and put 2-line data together
- vzip.32 d10, d11
- vshr.u64 q9, q4, #8 ;construct src_ptr[-1]
- vshr.u64 q10, q6, #8
- vmlal.u8 q7, d6, d0 ;+(src_ptr[-2] * vp8_filter[0])
- vmlal.u8 q8, d10, d0
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[-1])
- vzip.32 d20, d21
- vshr.u64 q3, q4, #32 ;construct src_ptr[2]
- vshr.u64 q5, q6, #32
- vmlsl.u8 q7, d18, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q8, d20, d1
-
- vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[2])
- vzip.32 d10, d11
- vshr.u64 q9, q4, #16 ;construct src_ptr[0]
- vshr.u64 q10, q6, #16
- vmlsl.u8 q7, d6, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q8, d10, d4
-
- vzip.32 d18, d19 ;put 2-line data in 1 register (src_ptr[0])
- vzip.32 d20, d21
- vshr.u64 q3, q4, #24 ;construct src_ptr[1]
- vshr.u64 q5, q6, #24
- vmlal.u8 q7, d18, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q8, d20, d2
-
- vzip.32 d6, d7 ;put 2-line data in 1 register (src_ptr[1])
- vzip.32 d10, d11
- vmull.u8 q9, d6, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q10, d10, d3
-
- add r0, r4, lr
- add r1, r0, lr
- add r2, r1, lr
-
- vqadd.s16 q7, q9 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q10
-
- vqrshrun.s16 d27, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d28, q8, #7
-
- vst1.32 {d27[0]}, [r4] ;store result
- vst1.32 {d27[1]}, [r0]
- vst1.32 {d28[0]}, [r1]
- vst1.32 {d28[1]}, [r2]
-
- pop {r4, pc}
-
-
-;---------------------
-secondpass_filter4x4_only
- sub r0, r0, r1, lsl #1
- add r3, r12, r3, lsl #5
-
- vld1.32 {d27[0]}, [r0], r1 ;load src data
- vld1.s32 {q5, q6}, [r3] ;load second_pass filter
- vld1.32 {d27[1]}, [r0], r1
- vabs.s32 q7, q5
- vld1.32 {d28[0]}, [r0], r1
- vabs.s32 q8, q6
- vld1.32 {d28[1]}, [r0], r1
- vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
- vld1.32 {d29[0]}, [r0], r1
- vdup.8 d1, d14[4]
- vld1.32 {d29[1]}, [r0], r1
- vdup.8 d2, d15[0]
- vld1.32 {d30[0]}, [r0], r1
- vdup.8 d3, d15[4]
- vld1.32 {d30[1]}, [r0], r1
- vdup.8 d4, d16[0]
- vld1.32 {d31[0]}, [r0], r1
- vdup.8 d5, d16[4]
-
- vext.8 d23, d27, d28, #4
- vext.8 d24, d28, d29, #4
- vext.8 d25, d29, d30, #4
- vext.8 d26, d30, d31, #4
-
- vmull.u8 q3, d27, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q4, d28, d0
-
- vmull.u8 q5, d25, d5 ;(src_ptr[3] * vp8_filter[5])
- vmull.u8 q6, d26, d5
-
- vmlsl.u8 q3, d29, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q4, d30, d4
-
- vmlsl.u8 q5, d23, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q6, d24, d1
-
- vmlal.u8 q3, d28, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q4, d29, d2
-
- vmlal.u8 q5, d24, d3 ;(src_ptr[1] * vp8_filter[3])
- vmlal.u8 q6, d25, d3
-
- add r0, r4, lr
- add r1, r0, lr
- add r2, r1, lr
-
- vqadd.s16 q5, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q6, q4
-
- vqrshrun.s16 d3, q5, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d4, q6, #7
-
- vst1.32 {d3[0]}, [r4] ;store result
- vst1.32 {d3[1]}, [r0]
- vst1.32 {d4[0]}, [r1]
- vst1.32 {d4[1]}, [r2]
-
- pop {r4, pc}
-
- ENDP
-
- END
diff --git a/vp8/common/arm/neon/sixtappredict8x4_neon.asm b/vp8/common/arm/neon/sixtappredict8x4_neon.asm
deleted file mode 100644
index 2b09400..0000000
--- a/vp8/common/arm/neon/sixtappredict8x4_neon.asm
+++ /dev/null
@@ -1,472 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_sixtap_predict8x4_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-filter8_coeff
- DCD 0, 0, 128, 0, 0, 0, 0, 0
- DCD 0, -6, 123, 12, -1, 0, 0, 0
- DCD 2, -11, 108, 36, -8, 1, 0, 0
- DCD 0, -9, 93, 50, -6, 0, 0, 0
- DCD 3, -16, 77, 77, -16, 3, 0, 0
- DCD 0, -6, 50, 93, -9, 0, 0, 0
- DCD 1, -8, 36, 108, -11, 2, 0, 0
- DCD 0, -1, 12, 123, -6, 0, 0, 0
-
-;-----------------
-; r0 unsigned char *src_ptr,
-; r1 int src_pixels_per_line,
-; r2 int xoffset,
-; r3 int yoffset,
-; r4 unsigned char *dst_ptr,
-; stack(r5) int dst_pitch
-
-|vp8_sixtap_predict8x4_neon| PROC
- push {r4-r5, lr}
-
- adr r12, filter8_coeff
- ldr r4, [sp, #12] ;load parameters from stack
- ldr r5, [sp, #16] ;load parameters from stack
-
- cmp r2, #0 ;skip first_pass filter if xoffset=0
- beq secondpass_filter8x4_only
-
- add r2, r12, r2, lsl #5 ;calculate filter location
-
- cmp r3, #0 ;skip second_pass filter if yoffset=0
-
- vld1.s32 {q14, q15}, [r2] ;load first_pass filter
-
- beq firstpass_filter8x4_only
-
- sub sp, sp, #32 ;reserve space on stack for temporary storage
- vabs.s32 q12, q14
- vabs.s32 q13, q15
-
- sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
- mov lr, sp
- sub r0, r0, r1, lsl #1
-
- vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
- vdup.8 d1, d24[4]
- vdup.8 d2, d25[0]
-
-;First pass: output_height lines x output_width columns (9x8)
- vld1.u8 {q3}, [r0], r1 ;load src data
- vdup.8 d3, d25[4]
- vld1.u8 {q4}, [r0], r1
- vdup.8 d4, d26[0]
- vld1.u8 {q5}, [r0], r1
- vdup.8 d5, d26[4]
- vld1.u8 {q6}, [r0], r1
-
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- vmull.u8 q7, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q8, d8, d0
- vmull.u8 q9, d10, d0
- vmull.u8 q10, d12, d0
-
- vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
- vext.8 d29, d8, d9, #1
- vext.8 d30, d10, d11, #1
- vext.8 d31, d12, d13, #1
-
- vmlsl.u8 q7, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q8, d29, d1
- vmlsl.u8 q9, d30, d1
- vmlsl.u8 q10, d31, d1
-
- vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
- vext.8 d29, d8, d9, #4
- vext.8 d30, d10, d11, #4
- vext.8 d31, d12, d13, #4
-
- vmlsl.u8 q7, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q8, d29, d4
- vmlsl.u8 q9, d30, d4
- vmlsl.u8 q10, d31, d4
-
- vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
- vext.8 d29, d8, d9, #2
- vext.8 d30, d10, d11, #2
- vext.8 d31, d12, d13, #2
-
- vmlal.u8 q7, d28, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q8, d29, d2
- vmlal.u8 q9, d30, d2
- vmlal.u8 q10, d31, d2
-
- vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d29, d8, d9, #5
- vext.8 d30, d10, d11, #5
- vext.8 d31, d12, d13, #5
-
- vmlal.u8 q7, d28, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q8, d29, d5
- vmlal.u8 q9, d30, d5
- vmlal.u8 q10, d31, d5
-
- vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
- vext.8 d29, d8, d9, #3
- vext.8 d30, d10, d11, #3
- vext.8 d31, d12, d13, #3
-
- vmull.u8 q3, d28, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q4, d29, d3
- vmull.u8 q5, d30, d3
- vmull.u8 q6, d31, d3
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vld1.u8 {q3}, [r0], r1 ;load src data
-
- vqrshrun.s16 d22, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d23, q8, #7
- vqrshrun.s16 d24, q9, #7
- vqrshrun.s16 d25, q10, #7
-
- vld1.u8 {q4}, [r0], r1
- vst1.u8 {d22}, [lr]! ;store result
- vld1.u8 {q5}, [r0], r1
- vst1.u8 {d23}, [lr]!
- vld1.u8 {q6}, [r0], r1
- vst1.u8 {d24}, [lr]!
- vld1.u8 {q7}, [r0], r1
- vst1.u8 {d25}, [lr]!
-
- ;first_pass filtering on the rest 5-line data
- vmull.u8 q8, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q9, d8, d0
- vmull.u8 q10, d10, d0
- vmull.u8 q11, d12, d0
- vmull.u8 q12, d14, d0
-
- vext.8 d27, d6, d7, #1 ;construct src_ptr[-1]
- vext.8 d28, d8, d9, #1
- vext.8 d29, d10, d11, #1
- vext.8 d30, d12, d13, #1
- vext.8 d31, d14, d15, #1
-
- vmlsl.u8 q8, d27, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q9, d28, d1
- vmlsl.u8 q10, d29, d1
- vmlsl.u8 q11, d30, d1
- vmlsl.u8 q12, d31, d1
-
- vext.8 d27, d6, d7, #4 ;construct src_ptr[2]
- vext.8 d28, d8, d9, #4
- vext.8 d29, d10, d11, #4
- vext.8 d30, d12, d13, #4
- vext.8 d31, d14, d15, #4
-
- vmlsl.u8 q8, d27, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q9, d28, d4
- vmlsl.u8 q10, d29, d4
- vmlsl.u8 q11, d30, d4
- vmlsl.u8 q12, d31, d4
-
- vext.8 d27, d6, d7, #2 ;construct src_ptr[0]
- vext.8 d28, d8, d9, #2
- vext.8 d29, d10, d11, #2
- vext.8 d30, d12, d13, #2
- vext.8 d31, d14, d15, #2
-
- vmlal.u8 q8, d27, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q9, d28, d2
- vmlal.u8 q10, d29, d2
- vmlal.u8 q11, d30, d2
- vmlal.u8 q12, d31, d2
-
- vext.8 d27, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d28, d8, d9, #5
- vext.8 d29, d10, d11, #5
- vext.8 d30, d12, d13, #5
- vext.8 d31, d14, d15, #5
-
- vmlal.u8 q8, d27, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q9, d28, d5
- vmlal.u8 q10, d29, d5
- vmlal.u8 q11, d30, d5
- vmlal.u8 q12, d31, d5
-
- vext.8 d27, d6, d7, #3 ;construct src_ptr[1]
- vext.8 d28, d8, d9, #3
- vext.8 d29, d10, d11, #3
- vext.8 d30, d12, d13, #3
- vext.8 d31, d14, d15, #3
-
- vmull.u8 q3, d27, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q4, d28, d3
- vmull.u8 q5, d29, d3
- vmull.u8 q6, d30, d3
- vmull.u8 q7, d31, d3
-
- vqadd.s16 q8, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q9, q4
- vqadd.s16 q10, q5
- vqadd.s16 q11, q6
- vqadd.s16 q12, q7
-
- vqrshrun.s16 d26, q8, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d27, q9, #7
- vqrshrun.s16 d28, q10, #7
- vqrshrun.s16 d29, q11, #7 ;load intermediate data from stack
- vqrshrun.s16 d30, q12, #7
-
-;Second pass: 8x4
-;secondpass_filter
- add r3, r12, r3, lsl #5
- sub lr, lr, #32
-
- vld1.s32 {q5, q6}, [r3] ;load second_pass filter
- vld1.u8 {q11}, [lr]!
-
- vabs.s32 q7, q5
- vabs.s32 q8, q6
-
- vld1.u8 {q12}, [lr]!
-
- vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
- vdup.8 d1, d14[4]
- vdup.8 d2, d15[0]
- vdup.8 d3, d15[4]
- vdup.8 d4, d16[0]
- vdup.8 d5, d16[4]
-
- vmull.u8 q3, d22, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q4, d23, d0
- vmull.u8 q5, d24, d0
- vmull.u8 q6, d25, d0
-
- vmlsl.u8 q3, d23, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q4, d24, d1
- vmlsl.u8 q5, d25, d1
- vmlsl.u8 q6, d26, d1
-
- vmlsl.u8 q3, d26, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q4, d27, d4
- vmlsl.u8 q5, d28, d4
- vmlsl.u8 q6, d29, d4
-
- vmlal.u8 q3, d24, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q4, d25, d2
- vmlal.u8 q5, d26, d2
- vmlal.u8 q6, d27, d2
-
- vmlal.u8 q3, d27, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q4, d28, d5
- vmlal.u8 q5, d29, d5
- vmlal.u8 q6, d30, d5
-
- vmull.u8 q7, d25, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q8, d26, d3
- vmull.u8 q9, d27, d3
- vmull.u8 q10, d28, d3
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d7, q8, #7
- vqrshrun.s16 d8, q9, #7
- vqrshrun.s16 d9, q10, #7
-
- vst1.u8 {d6}, [r4], r5 ;store result
- vst1.u8 {d7}, [r4], r5
- vst1.u8 {d8}, [r4], r5
- vst1.u8 {d9}, [r4], r5
-
- add sp, sp, #32
- pop {r4-r5,pc}
-
-;--------------------
-firstpass_filter8x4_only
- vabs.s32 q12, q14
- vabs.s32 q13, q15
-
- sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
- vld1.u8 {q3}, [r0], r1 ;load src data
-
- vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
- vld1.u8 {q4}, [r0], r1
- vdup.8 d1, d24[4]
- vld1.u8 {q5}, [r0], r1
- vdup.8 d2, d25[0]
- vld1.u8 {q6}, [r0], r1
- vdup.8 d3, d25[4]
- vdup.8 d4, d26[0]
- vdup.8 d5, d26[4]
-
-;First pass: output_height lines x output_width columns (4x8)
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- vmull.u8 q7, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q8, d8, d0
- vmull.u8 q9, d10, d0
- vmull.u8 q10, d12, d0
-
- vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
- vext.8 d29, d8, d9, #1
- vext.8 d30, d10, d11, #1
- vext.8 d31, d12, d13, #1
-
- vmlsl.u8 q7, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q8, d29, d1
- vmlsl.u8 q9, d30, d1
- vmlsl.u8 q10, d31, d1
-
- vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
- vext.8 d29, d8, d9, #4
- vext.8 d30, d10, d11, #4
- vext.8 d31, d12, d13, #4
-
- vmlsl.u8 q7, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q8, d29, d4
- vmlsl.u8 q9, d30, d4
- vmlsl.u8 q10, d31, d4
-
- vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
- vext.8 d29, d8, d9, #2
- vext.8 d30, d10, d11, #2
- vext.8 d31, d12, d13, #2
-
- vmlal.u8 q7, d28, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q8, d29, d2
- vmlal.u8 q9, d30, d2
- vmlal.u8 q10, d31, d2
-
- vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d29, d8, d9, #5
- vext.8 d30, d10, d11, #5
- vext.8 d31, d12, d13, #5
-
- vmlal.u8 q7, d28, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q8, d29, d5
- vmlal.u8 q9, d30, d5
- vmlal.u8 q10, d31, d5
-
- vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
- vext.8 d29, d8, d9, #3
- vext.8 d30, d10, d11, #3
- vext.8 d31, d12, d13, #3
-
- vmull.u8 q3, d28, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q4, d29, d3
- vmull.u8 q5, d30, d3
- vmull.u8 q6, d31, d3
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vqrshrun.s16 d22, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d23, q8, #7
- vqrshrun.s16 d24, q9, #7
- vqrshrun.s16 d25, q10, #7
-
- vst1.u8 {d22}, [r4], r5 ;store result
- vst1.u8 {d23}, [r4], r5
- vst1.u8 {d24}, [r4], r5
- vst1.u8 {d25}, [r4], r5
-
- pop {r4-r5,pc}
-
-;---------------------
-secondpass_filter8x4_only
-;Second pass: 8x4
- add r3, r12, r3, lsl #5
- sub r0, r0, r1, lsl #1
- vld1.s32 {q5, q6}, [r3] ;load second_pass filter
- vabs.s32 q7, q5
- vabs.s32 q8, q6
-
- vld1.u8 {d22}, [r0], r1
- vld1.u8 {d23}, [r0], r1
- vld1.u8 {d24}, [r0], r1
- vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
- vld1.u8 {d25}, [r0], r1
- vdup.8 d1, d14[4]
- vld1.u8 {d26}, [r0], r1
- vdup.8 d2, d15[0]
- vld1.u8 {d27}, [r0], r1
- vdup.8 d3, d15[4]
- vld1.u8 {d28}, [r0], r1
- vdup.8 d4, d16[0]
- vld1.u8 {d29}, [r0], r1
- vdup.8 d5, d16[4]
- vld1.u8 {d30}, [r0], r1
-
- vmull.u8 q3, d22, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q4, d23, d0
- vmull.u8 q5, d24, d0
- vmull.u8 q6, d25, d0
-
- vmlsl.u8 q3, d23, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q4, d24, d1
- vmlsl.u8 q5, d25, d1
- vmlsl.u8 q6, d26, d1
-
- vmlsl.u8 q3, d26, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q4, d27, d4
- vmlsl.u8 q5, d28, d4
- vmlsl.u8 q6, d29, d4
-
- vmlal.u8 q3, d24, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q4, d25, d2
- vmlal.u8 q5, d26, d2
- vmlal.u8 q6, d27, d2
-
- vmlal.u8 q3, d27, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q4, d28, d5
- vmlal.u8 q5, d29, d5
- vmlal.u8 q6, d30, d5
-
- vmull.u8 q7, d25, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q8, d26, d3
- vmull.u8 q9, d27, d3
- vmull.u8 q10, d28, d3
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d7, q8, #7
- vqrshrun.s16 d8, q9, #7
- vqrshrun.s16 d9, q10, #7
-
- vst1.u8 {d6}, [r4], r5 ;store result
- vst1.u8 {d7}, [r4], r5
- vst1.u8 {d8}, [r4], r5
- vst1.u8 {d9}, [r4], r5
-
- pop {r4-r5,pc}
-
- ENDP
-
- END
diff --git a/vp8/common/arm/neon/sixtappredict8x8_neon.asm b/vp8/common/arm/neon/sixtappredict8x8_neon.asm
deleted file mode 100644
index bcbf63e..0000000
--- a/vp8/common/arm/neon/sixtappredict8x8_neon.asm
+++ /dev/null
@@ -1,523 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_sixtap_predict8x8_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-; r0 unsigned char *src_ptr,
-; r1 int src_pixels_per_line,
-; r2 int xoffset,
-; r3 int yoffset,
-; stack(r4) unsigned char *dst_ptr,
-; stack(r5) int dst_pitch
-
-|vp8_sixtap_predict8x8_neon| PROC
- push {r4-r5, lr}
-
- adrl r12, filter8_coeff
-
- ldr r4, [sp, #12] ;load parameters from stack
- ldr r5, [sp, #16] ;load parameters from stack
-
- cmp r2, #0 ;skip first_pass filter if xoffset=0
- beq secondpass_filter8x8_only
-
- add r2, r12, r2, lsl #5 ;calculate filter location
-
- cmp r3, #0 ;skip second_pass filter if yoffset=0
-
- vld1.s32 {q14, q15}, [r2] ;load first_pass filter
-
- beq firstpass_filter8x8_only
-
- sub sp, sp, #64 ;reserve space on stack for temporary storage
- mov lr, sp
-
- vabs.s32 q12, q14
- vabs.s32 q13, q15
-
- mov r2, #2 ;loop counter
- sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
- sub r0, r0, r1, lsl #1
-
- vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
- vdup.8 d1, d24[4]
- vdup.8 d2, d25[0]
-
-;First pass: output_height lines x output_width columns (13x8)
- vld1.u8 {q3}, [r0], r1 ;load src data
- vdup.8 d3, d25[4]
- vld1.u8 {q4}, [r0], r1
- vdup.8 d4, d26[0]
- vld1.u8 {q5}, [r0], r1
- vdup.8 d5, d26[4]
- vld1.u8 {q6}, [r0], r1
-
-filt_blk2d_fp8x8_loop_neon
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- vmull.u8 q7, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q8, d8, d0
- vmull.u8 q9, d10, d0
- vmull.u8 q10, d12, d0
-
- vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
- vext.8 d29, d8, d9, #1
- vext.8 d30, d10, d11, #1
- vext.8 d31, d12, d13, #1
-
- vmlsl.u8 q7, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q8, d29, d1
- vmlsl.u8 q9, d30, d1
- vmlsl.u8 q10, d31, d1
-
- vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
- vext.8 d29, d8, d9, #4
- vext.8 d30, d10, d11, #4
- vext.8 d31, d12, d13, #4
-
- vmlsl.u8 q7, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q8, d29, d4
- vmlsl.u8 q9, d30, d4
- vmlsl.u8 q10, d31, d4
-
- vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
- vext.8 d29, d8, d9, #2
- vext.8 d30, d10, d11, #2
- vext.8 d31, d12, d13, #2
-
- vmlal.u8 q7, d28, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q8, d29, d2
- vmlal.u8 q9, d30, d2
- vmlal.u8 q10, d31, d2
-
- vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d29, d8, d9, #5
- vext.8 d30, d10, d11, #5
- vext.8 d31, d12, d13, #5
-
- vmlal.u8 q7, d28, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q8, d29, d5
- vmlal.u8 q9, d30, d5
- vmlal.u8 q10, d31, d5
-
- vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
- vext.8 d29, d8, d9, #3
- vext.8 d30, d10, d11, #3
- vext.8 d31, d12, d13, #3
-
- vmull.u8 q3, d28, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q4, d29, d3
- vmull.u8 q5, d30, d3
- vmull.u8 q6, d31, d3
-
- subs r2, r2, #1
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vld1.u8 {q3}, [r0], r1 ;load src data
-
- vqrshrun.s16 d22, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d23, q8, #7
- vqrshrun.s16 d24, q9, #7
- vqrshrun.s16 d25, q10, #7
-
- vst1.u8 {d22}, [lr]! ;store result
- vld1.u8 {q4}, [r0], r1
- vst1.u8 {d23}, [lr]!
- vld1.u8 {q5}, [r0], r1
- vst1.u8 {d24}, [lr]!
- vld1.u8 {q6}, [r0], r1
- vst1.u8 {d25}, [lr]!
-
- bne filt_blk2d_fp8x8_loop_neon
-
- ;first_pass filtering on the rest 5-line data
- ;vld1.u8 {q3}, [r0], r1 ;load src data
- ;vld1.u8 {q4}, [r0], r1
- ;vld1.u8 {q5}, [r0], r1
- ;vld1.u8 {q6}, [r0], r1
- vld1.u8 {q7}, [r0], r1
-
- vmull.u8 q8, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q9, d8, d0
- vmull.u8 q10, d10, d0
- vmull.u8 q11, d12, d0
- vmull.u8 q12, d14, d0
-
- vext.8 d27, d6, d7, #1 ;construct src_ptr[-1]
- vext.8 d28, d8, d9, #1
- vext.8 d29, d10, d11, #1
- vext.8 d30, d12, d13, #1
- vext.8 d31, d14, d15, #1
-
- vmlsl.u8 q8, d27, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q9, d28, d1
- vmlsl.u8 q10, d29, d1
- vmlsl.u8 q11, d30, d1
- vmlsl.u8 q12, d31, d1
-
- vext.8 d27, d6, d7, #4 ;construct src_ptr[2]
- vext.8 d28, d8, d9, #4
- vext.8 d29, d10, d11, #4
- vext.8 d30, d12, d13, #4
- vext.8 d31, d14, d15, #4
-
- vmlsl.u8 q8, d27, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q9, d28, d4
- vmlsl.u8 q10, d29, d4
- vmlsl.u8 q11, d30, d4
- vmlsl.u8 q12, d31, d4
-
- vext.8 d27, d6, d7, #2 ;construct src_ptr[0]
- vext.8 d28, d8, d9, #2
- vext.8 d29, d10, d11, #2
- vext.8 d30, d12, d13, #2
- vext.8 d31, d14, d15, #2
-
- vmlal.u8 q8, d27, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q9, d28, d2
- vmlal.u8 q10, d29, d2
- vmlal.u8 q11, d30, d2
- vmlal.u8 q12, d31, d2
-
- vext.8 d27, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d28, d8, d9, #5
- vext.8 d29, d10, d11, #5
- vext.8 d30, d12, d13, #5
- vext.8 d31, d14, d15, #5
-
- vmlal.u8 q8, d27, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q9, d28, d5
- vmlal.u8 q10, d29, d5
- vmlal.u8 q11, d30, d5
- vmlal.u8 q12, d31, d5
-
- vext.8 d27, d6, d7, #3 ;construct src_ptr[1]
- vext.8 d28, d8, d9, #3
- vext.8 d29, d10, d11, #3
- vext.8 d30, d12, d13, #3
- vext.8 d31, d14, d15, #3
-
- vmull.u8 q3, d27, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q4, d28, d3
- vmull.u8 q5, d29, d3
- vmull.u8 q6, d30, d3
- vmull.u8 q7, d31, d3
-
- vqadd.s16 q8, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q9, q4
- vqadd.s16 q10, q5
- vqadd.s16 q11, q6
- vqadd.s16 q12, q7
-
- add r3, r12, r3, lsl #5
-
- vqrshrun.s16 d26, q8, #7 ;shift/round/saturate to u8
- sub lr, lr, #64
- vqrshrun.s16 d27, q9, #7
- vld1.u8 {q9}, [lr]! ;load intermediate data from stack
- vqrshrun.s16 d28, q10, #7
- vld1.u8 {q10}, [lr]!
-
- vld1.s32 {q5, q6}, [r3] ;load second_pass filter
-
- vqrshrun.s16 d29, q11, #7
- vld1.u8 {q11}, [lr]!
-
- vabs.s32 q7, q5
- vabs.s32 q8, q6
-
- vqrshrun.s16 d30, q12, #7
- vld1.u8 {q12}, [lr]!
-
-;Second pass: 8x8
- mov r3, #2 ;loop counter
-
- vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
- vdup.8 d1, d14[4]
- vdup.8 d2, d15[0]
- vdup.8 d3, d15[4]
- vdup.8 d4, d16[0]
- vdup.8 d5, d16[4]
-
-filt_blk2d_sp8x8_loop_neon
- vmull.u8 q3, d18, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q4, d19, d0
- vmull.u8 q5, d20, d0
- vmull.u8 q6, d21, d0
-
- vmlsl.u8 q3, d19, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q4, d20, d1
- vmlsl.u8 q5, d21, d1
- vmlsl.u8 q6, d22, d1
-
- vmlsl.u8 q3, d22, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q4, d23, d4
- vmlsl.u8 q5, d24, d4
- vmlsl.u8 q6, d25, d4
-
- vmlal.u8 q3, d20, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q4, d21, d2
- vmlal.u8 q5, d22, d2
- vmlal.u8 q6, d23, d2
-
- vmlal.u8 q3, d23, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q4, d24, d5
- vmlal.u8 q5, d25, d5
- vmlal.u8 q6, d26, d5
-
- vmull.u8 q7, d21, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q8, d22, d3
- vmull.u8 q9, d23, d3
- vmull.u8 q10, d24, d3
-
- subs r3, r3, #1
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d7, q8, #7
- vqrshrun.s16 d8, q9, #7
- vqrshrun.s16 d9, q10, #7
-
- vmov q9, q11
- vst1.u8 {d6}, [r4], r5 ;store result
- vmov q10, q12
- vst1.u8 {d7}, [r4], r5
- vmov q11, q13
- vst1.u8 {d8}, [r4], r5
- vmov q12, q14
- vst1.u8 {d9}, [r4], r5
- vmov d26, d30
-
- bne filt_blk2d_sp8x8_loop_neon
-
- add sp, sp, #64
- pop {r4-r5,pc}
-
-;---------------------
-firstpass_filter8x8_only
- ;add r2, r12, r2, lsl #5 ;calculate filter location
- ;vld1.s32 {q14, q15}, [r2] ;load first_pass filter
- vabs.s32 q12, q14
- vabs.s32 q13, q15
-
- mov r2, #2 ;loop counter
- sub r0, r0, #2 ;move srcptr back to (line-2) and (column-2)
-
- vdup.8 d0, d24[0] ;first_pass filter (d0-d5)
- vdup.8 d1, d24[4]
- vdup.8 d2, d25[0]
- vdup.8 d3, d25[4]
- vdup.8 d4, d26[0]
- vdup.8 d5, d26[4]
-
-;First pass: output_height lines x output_width columns (8x8)
-filt_blk2d_fpo8x8_loop_neon
- vld1.u8 {q3}, [r0], r1 ;load src data
- vld1.u8 {q4}, [r0], r1
- vld1.u8 {q5}, [r0], r1
- vld1.u8 {q6}, [r0], r1
-
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- vmull.u8 q7, d6, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q8, d8, d0
- vmull.u8 q9, d10, d0
- vmull.u8 q10, d12, d0
-
- vext.8 d28, d6, d7, #1 ;construct src_ptr[-1]
- vext.8 d29, d8, d9, #1
- vext.8 d30, d10, d11, #1
- vext.8 d31, d12, d13, #1
-
- vmlsl.u8 q7, d28, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q8, d29, d1
- vmlsl.u8 q9, d30, d1
- vmlsl.u8 q10, d31, d1
-
- vext.8 d28, d6, d7, #4 ;construct src_ptr[2]
- vext.8 d29, d8, d9, #4
- vext.8 d30, d10, d11, #4
- vext.8 d31, d12, d13, #4
-
- vmlsl.u8 q7, d28, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q8, d29, d4
- vmlsl.u8 q9, d30, d4
- vmlsl.u8 q10, d31, d4
-
- vext.8 d28, d6, d7, #2 ;construct src_ptr[0]
- vext.8 d29, d8, d9, #2
- vext.8 d30, d10, d11, #2
- vext.8 d31, d12, d13, #2
-
- vmlal.u8 q7, d28, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q8, d29, d2
- vmlal.u8 q9, d30, d2
- vmlal.u8 q10, d31, d2
-
- vext.8 d28, d6, d7, #5 ;construct src_ptr[3]
- vext.8 d29, d8, d9, #5
- vext.8 d30, d10, d11, #5
- vext.8 d31, d12, d13, #5
-
- vmlal.u8 q7, d28, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q8, d29, d5
- vmlal.u8 q9, d30, d5
- vmlal.u8 q10, d31, d5
-
- vext.8 d28, d6, d7, #3 ;construct src_ptr[1]
- vext.8 d29, d8, d9, #3
- vext.8 d30, d10, d11, #3
- vext.8 d31, d12, d13, #3
-
- vmull.u8 q3, d28, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q4, d29, d3
- vmull.u8 q5, d30, d3
- vmull.u8 q6, d31, d3
- ;
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- subs r2, r2, #1
-
- vqrshrun.s16 d22, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d23, q8, #7
- vqrshrun.s16 d24, q9, #7
- vqrshrun.s16 d25, q10, #7
-
- vst1.u8 {d22}, [r4], r5 ;store result
- vst1.u8 {d23}, [r4], r5
- vst1.u8 {d24}, [r4], r5
- vst1.u8 {d25}, [r4], r5
-
- bne filt_blk2d_fpo8x8_loop_neon
-
- pop {r4-r5,pc}
-
-;---------------------
-secondpass_filter8x8_only
- sub r0, r0, r1, lsl #1
- add r3, r12, r3, lsl #5
-
- vld1.u8 {d18}, [r0], r1 ;load src data
- vld1.s32 {q5, q6}, [r3] ;load second_pass filter
- vld1.u8 {d19}, [r0], r1
- vabs.s32 q7, q5
- vld1.u8 {d20}, [r0], r1
- vabs.s32 q8, q6
- vld1.u8 {d21}, [r0], r1
- mov r3, #2 ;loop counter
- vld1.u8 {d22}, [r0], r1
- vdup.8 d0, d14[0] ;second_pass filter parameters (d0-d5)
- vld1.u8 {d23}, [r0], r1
- vdup.8 d1, d14[4]
- vld1.u8 {d24}, [r0], r1
- vdup.8 d2, d15[0]
- vld1.u8 {d25}, [r0], r1
- vdup.8 d3, d15[4]
- vld1.u8 {d26}, [r0], r1
- vdup.8 d4, d16[0]
- vld1.u8 {d27}, [r0], r1
- vdup.8 d5, d16[4]
- vld1.u8 {d28}, [r0], r1
- vld1.u8 {d29}, [r0], r1
- vld1.u8 {d30}, [r0], r1
-
-;Second pass: 8x8
-filt_blk2d_spo8x8_loop_neon
- vmull.u8 q3, d18, d0 ;(src_ptr[-2] * vp8_filter[0])
- vmull.u8 q4, d19, d0
- vmull.u8 q5, d20, d0
- vmull.u8 q6, d21, d0
-
- vmlsl.u8 q3, d19, d1 ;-(src_ptr[-1] * vp8_filter[1])
- vmlsl.u8 q4, d20, d1
- vmlsl.u8 q5, d21, d1
- vmlsl.u8 q6, d22, d1
-
- vmlsl.u8 q3, d22, d4 ;-(src_ptr[2] * vp8_filter[4])
- vmlsl.u8 q4, d23, d4
- vmlsl.u8 q5, d24, d4
- vmlsl.u8 q6, d25, d4
-
- vmlal.u8 q3, d20, d2 ;(src_ptr[0] * vp8_filter[2])
- vmlal.u8 q4, d21, d2
- vmlal.u8 q5, d22, d2
- vmlal.u8 q6, d23, d2
-
- vmlal.u8 q3, d23, d5 ;(src_ptr[3] * vp8_filter[5])
- vmlal.u8 q4, d24, d5
- vmlal.u8 q5, d25, d5
- vmlal.u8 q6, d26, d5
-
- vmull.u8 q7, d21, d3 ;(src_ptr[1] * vp8_filter[3])
- vmull.u8 q8, d22, d3
- vmull.u8 q9, d23, d3
- vmull.u8 q10, d24, d3
-
- subs r3, r3, #1
-
- vqadd.s16 q7, q3 ;sum of all (src_data*filter_parameters)
- vqadd.s16 q8, q4
- vqadd.s16 q9, q5
- vqadd.s16 q10, q6
-
- vqrshrun.s16 d6, q7, #7 ;shift/round/saturate to u8
- vqrshrun.s16 d7, q8, #7
- vqrshrun.s16 d8, q9, #7
- vqrshrun.s16 d9, q10, #7
-
- vmov q9, q11
- vst1.u8 {d6}, [r4], r5 ;store result
- vmov q10, q12
- vst1.u8 {d7}, [r4], r5
- vmov q11, q13
- vst1.u8 {d8}, [r4], r5
- vmov q12, q14
- vst1.u8 {d9}, [r4], r5
- vmov d26, d30
-
- bne filt_blk2d_spo8x8_loop_neon
-
- pop {r4-r5,pc}
-
- ENDP
-
-;-----------------
-
-filter8_coeff
- DCD 0, 0, 128, 0, 0, 0, 0, 0
- DCD 0, -6, 123, 12, -1, 0, 0, 0
- DCD 2, -11, 108, 36, -8, 1, 0, 0
- DCD 0, -9, 93, 50, -6, 0, 0, 0
- DCD 3, -16, 77, 77, -16, 3, 0, 0
- DCD 0, -6, 50, 93, -9, 0, 0, 0
- DCD 1, -8, 36, 108, -11, 2, 0, 0
- DCD 0, -1, 12, 123, -6, 0, 0, 0
-
- END
diff --git a/vp8/common/arm/recon_arm.h b/vp8/common/arm/recon_arm.h
deleted file mode 100644
index 377cb2a..0000000
--- a/vp8/common/arm/recon_arm.h
+++ /dev/null
@@ -1,90 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef RECON_ARM_H
-#define RECON_ARM_H
-
-#if HAVE_ARMV6
-extern prototype_recon_block(vp8_recon_b_armv6);
-extern prototype_recon_block(vp8_recon2b_armv6);
-extern prototype_recon_block(vp8_recon4b_armv6);
-
-extern prototype_copy_block(vp8_copy_mem8x8_v6);
-extern prototype_copy_block(vp8_copy_mem8x4_v6);
-extern prototype_copy_block(vp8_copy_mem16x16_v6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_recon_recon
-#define vp8_recon_recon vp8_recon_b_armv6
-
-#undef vp8_recon_recon2
-#define vp8_recon_recon2 vp8_recon2b_armv6
-
-#undef vp8_recon_recon4
-#define vp8_recon_recon4 vp8_recon4b_armv6
-
-#undef vp8_recon_copy8x8
-#define vp8_recon_copy8x8 vp8_copy_mem8x8_v6
-
-#undef vp8_recon_copy8x4
-#define vp8_recon_copy8x4 vp8_copy_mem8x4_v6
-
-#undef vp8_recon_copy16x16
-#define vp8_recon_copy16x16 vp8_copy_mem16x16_v6
-#endif
-#endif
-
-#if HAVE_ARMV7
-extern prototype_recon_block(vp8_recon_b_neon);
-extern prototype_recon_block(vp8_recon2b_neon);
-extern prototype_recon_block(vp8_recon4b_neon);
-
-extern prototype_copy_block(vp8_copy_mem8x8_neon);
-extern prototype_copy_block(vp8_copy_mem8x4_neon);
-extern prototype_copy_block(vp8_copy_mem16x16_neon);
-
-extern prototype_recon_macroblock(vp8_recon_mb_neon);
-
-extern prototype_build_intra_predictors(vp8_build_intra_predictors_mby_neon);
-extern prototype_build_intra_predictors(vp8_build_intra_predictors_mby_s_neon);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_recon_recon
-#define vp8_recon_recon vp8_recon_b_neon
-
-#undef vp8_recon_recon2
-#define vp8_recon_recon2 vp8_recon2b_neon
-
-#undef vp8_recon_recon4
-#define vp8_recon_recon4 vp8_recon4b_neon
-
-#undef vp8_recon_copy8x8
-#define vp8_recon_copy8x8 vp8_copy_mem8x8_neon
-
-#undef vp8_recon_copy8x4
-#define vp8_recon_copy8x4 vp8_copy_mem8x4_neon
-
-#undef vp8_recon_copy16x16
-#define vp8_recon_copy16x16 vp8_copy_mem16x16_neon
-
-#undef vp8_recon_recon_mb
-#define vp8_recon_recon_mb vp8_recon_mb_neon
-
-#undef vp8_recon_build_intra_predictors_mby
-#define vp8_recon_build_intra_predictors_mby vp8_build_intra_predictors_mby_neon
-
-#undef vp8_recon_build_intra_predictors_mby_s
-#define vp8_recon_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s_neon
-
-#endif
-#endif
-
-#endif
diff --git a/vp8/common/arm/reconintra_arm.c b/vp8/common/arm/reconintra_arm.c
deleted file mode 100644
index f8f4dca..0000000
--- a/vp8/common/arm/reconintra_arm.c
+++ /dev/null
@@ -1,62 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/common/blockd.h"
-#include "vp8/common/reconintra.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vp8/common/recon.h"
-
-#if HAVE_ARMV7
-extern void vp8_build_intra_predictors_mby_neon_func(
- unsigned char *y_buffer,
- unsigned char *ypred_ptr,
- int y_stride,
- int mode,
- int Up,
- int Left);
-
-void vp8_build_intra_predictors_mby_neon(MACROBLOCKD *x)
-{
- unsigned char *y_buffer = x->dst.y_buffer;
- unsigned char *ypred_ptr = x->predictor;
- int y_stride = x->dst.y_stride;
- int mode = x->mode_info_context->mbmi.mode;
- int Up = x->up_available;
- int Left = x->left_available;
-
- vp8_build_intra_predictors_mby_neon_func(y_buffer, ypred_ptr, y_stride, mode, Up, Left);
-}
-#endif
-
-
-#if HAVE_ARMV7
-extern void vp8_build_intra_predictors_mby_s_neon_func(
- unsigned char *y_buffer,
- unsigned char *ypred_ptr,
- int y_stride,
- int mode,
- int Up,
- int Left);
-
-void vp8_build_intra_predictors_mby_s_neon(MACROBLOCKD *x)
-{
- unsigned char *y_buffer = x->dst.y_buffer;
- unsigned char *ypred_ptr = x->predictor;
- int y_stride = x->dst.y_stride;
- int mode = x->mode_info_context->mbmi.mode;
- int Up = x->up_available;
- int Left = x->left_available;
-
- vp8_build_intra_predictors_mby_s_neon_func(y_buffer, ypred_ptr, y_stride, mode, Up, Left);
-}
-
-#endif
diff --git a/vp8/common/arm/subpixel_arm.h b/vp8/common/arm/subpixel_arm.h
deleted file mode 100644
index 6288538..0000000
--- a/vp8/common/arm/subpixel_arm.h
+++ /dev/null
@@ -1,89 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef SUBPIXEL_ARM_H
-#define SUBPIXEL_ARM_H
-
-#if HAVE_ARMV6
-extern prototype_subpixel_predict(vp8_sixtap_predict16x16_armv6);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x8_armv6);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x4_armv6);
-extern prototype_subpixel_predict(vp8_sixtap_predict_armv6);
-extern prototype_subpixel_predict(vp8_bilinear_predict16x16_armv6);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x8_armv6);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x4_armv6);
-extern prototype_subpixel_predict(vp8_bilinear_predict4x4_armv6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_subpix_sixtap16x16
-#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_armv6
-
-#undef vp8_subpix_sixtap8x8
-#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_armv6
-
-#undef vp8_subpix_sixtap8x4
-#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_armv6
-
-#undef vp8_subpix_sixtap4x4
-#define vp8_subpix_sixtap4x4 vp8_sixtap_predict_armv6
-
-#undef vp8_subpix_bilinear16x16
-#define vp8_subpix_bilinear16x16 vp8_bilinear_predict16x16_armv6
-
-#undef vp8_subpix_bilinear8x8
-#define vp8_subpix_bilinear8x8 vp8_bilinear_predict8x8_armv6
-
-#undef vp8_subpix_bilinear8x4
-#define vp8_subpix_bilinear8x4 vp8_bilinear_predict8x4_armv6
-
-#undef vp8_subpix_bilinear4x4
-#define vp8_subpix_bilinear4x4 vp8_bilinear_predict4x4_armv6
-#endif
-#endif
-
-#if HAVE_ARMV7
-extern prototype_subpixel_predict(vp8_sixtap_predict16x16_neon);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x8_neon);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x4_neon);
-extern prototype_subpixel_predict(vp8_sixtap_predict_neon);
-extern prototype_subpixel_predict(vp8_bilinear_predict16x16_neon);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x8_neon);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x4_neon);
-extern prototype_subpixel_predict(vp8_bilinear_predict4x4_neon);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_subpix_sixtap16x16
-#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_neon
-
-#undef vp8_subpix_sixtap8x8
-#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_neon
-
-#undef vp8_subpix_sixtap8x4
-#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_neon
-
-#undef vp8_subpix_sixtap4x4
-#define vp8_subpix_sixtap4x4 vp8_sixtap_predict_neon
-
-#undef vp8_subpix_bilinear16x16
-#define vp8_subpix_bilinear16x16 vp8_bilinear_predict16x16_neon
-
-#undef vp8_subpix_bilinear8x8
-#define vp8_subpix_bilinear8x8 vp8_bilinear_predict8x8_neon
-
-#undef vp8_subpix_bilinear8x4
-#define vp8_subpix_bilinear8x4 vp8_bilinear_predict8x4_neon
-
-#undef vp8_subpix_bilinear4x4
-#define vp8_subpix_bilinear4x4 vp8_bilinear_predict4x4_neon
-#endif
-#endif
-
-#endif
diff --git a/vp8/common/asm_com_offsets.c b/vp8/common/asm_com_offsets.c
deleted file mode 100644
index d299dd2..0000000
--- a/vp8/common/asm_com_offsets.c
+++ /dev/null
@@ -1,49 +0,0 @@
-/*
- * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include <stddef.h>
-
-#include "vpx_scale/yv12config.h"
-
-#define ct_assert(name,cond) \
- static void assert_##name(void) UNUSED;\
- static void assert_##name(void) {switch(0){case 0:case !!(cond):;}}
-
-#define DEFINE(sym, val) int sym = val;
-
-/*
-#define BLANK() asm volatile("\n->" : : )
-*/
-
-/*
- * int main(void)
- * {
- */
-
-//vpx_scale
-DEFINE(yv12_buffer_config_y_width, offsetof(YV12_BUFFER_CONFIG, y_width));
-DEFINE(yv12_buffer_config_y_height, offsetof(YV12_BUFFER_CONFIG, y_height));
-DEFINE(yv12_buffer_config_y_stride, offsetof(YV12_BUFFER_CONFIG, y_stride));
-DEFINE(yv12_buffer_config_uv_width, offsetof(YV12_BUFFER_CONFIG, uv_width));
-DEFINE(yv12_buffer_config_uv_height, offsetof(YV12_BUFFER_CONFIG, uv_height));
-DEFINE(yv12_buffer_config_uv_stride, offsetof(YV12_BUFFER_CONFIG, uv_stride));
-DEFINE(yv12_buffer_config_y_buffer, offsetof(YV12_BUFFER_CONFIG, y_buffer));
-DEFINE(yv12_buffer_config_u_buffer, offsetof(YV12_BUFFER_CONFIG, u_buffer));
-DEFINE(yv12_buffer_config_v_buffer, offsetof(YV12_BUFFER_CONFIG, v_buffer));
-DEFINE(yv12_buffer_config_border, offsetof(YV12_BUFFER_CONFIG, border));
-
-//add asserts for any offset that is not supported by assembly code
-//add asserts for any size that is not supported by assembly code
-/*
- * return 0;
- * }
- */
diff --git a/vp8/common/bigend.h b/vp8/common/bigend.h
deleted file mode 100644
index 6ac3f8b..0000000
--- a/vp8/common/bigend.h
+++ /dev/null
@@ -1,32 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef _bigend_h
-#define _bigend_h
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-#define invert2(x) ( (((x)>>8)&0x00ff) | (((x)<<8)&0xff00) )
-#define invert4(x) ( ((invert2(x)&0x0000ffff)<<16) | (invert2((x>>16))&0x0000ffff) )
-
-#define high_byte(x) (unsigned char)x
-#define mid2Byte(x) (unsigned char)(x >> 8)
-#define mid1Byte(x) (unsigned char)(x >> 16)
-#define low_byte(x) (unsigned char)(x >> 24)
-
-#define SWAPENDS 1
-
-#if defined(__cplusplus)
-}
-#endif
-#endif
diff --git a/vp8/common/blockd.h b/vp8/common/blockd.h
deleted file mode 100644
index fc8e072..0000000
--- a/vp8/common/blockd.h
+++ /dev/null
@@ -1,294 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_BLOCKD_H
-#define __INC_BLOCKD_H
-
-void vpx_log(const char *format, ...);
-
-#include "vpx_ports/config.h"
-#include "vpx_scale/yv12config.h"
-#include "mv.h"
-#include "treecoder.h"
-#include "subpixel.h"
-#include "vpx_ports/mem.h"
-
-#define TRUE 1
-#define FALSE 0
-
-/*#define DCPRED 1*/
-#define DCPREDSIMTHRESH 0
-#define DCPREDCNTTHRESH 3
-
-#define MB_FEATURE_TREE_PROBS 3
-#define MAX_MB_SEGMENTS 4
-
-#define MAX_REF_LF_DELTAS 4
-#define MAX_MODE_LF_DELTAS 4
-
-/* Segment Feature Masks */
-#define SEGMENT_DELTADATA 0
-#define SEGMENT_ABSDATA 1
-
-typedef struct
-{
- int r, c;
-} POS;
-
-#define PLANE_TYPE_Y_NO_DC 0
-#define PLANE_TYPE_Y2 1
-#define PLANE_TYPE_UV 2
-#define PLANE_TYPE_Y_WITH_DC 3
-
-
-typedef char ENTROPY_CONTEXT;
-typedef struct
-{
- ENTROPY_CONTEXT y1[4];
- ENTROPY_CONTEXT u[2];
- ENTROPY_CONTEXT v[2];
- ENTROPY_CONTEXT y2;
-} ENTROPY_CONTEXT_PLANES;
-
-extern const unsigned char vp8_block2left[25];
-extern const unsigned char vp8_block2above[25];
-
-#define VP8_COMBINEENTROPYCONTEXTS( Dest, A, B) \
- Dest = ((A)!=0) + ((B)!=0);
-
-
-typedef enum
-{
- KEY_FRAME = 0,
- INTER_FRAME = 1
-} FRAME_TYPE;
-
-typedef enum
-{
- DC_PRED, /* average of above and left pixels */
- V_PRED, /* vertical prediction */
- H_PRED, /* horizontal prediction */
- TM_PRED, /* Truemotion prediction */
- B_PRED, /* block based prediction, each block has its own prediction mode */
-
- NEARESTMV,
- NEARMV,
- ZEROMV,
- NEWMV,
- SPLITMV,
-
- MB_MODE_COUNT
-} MB_PREDICTION_MODE;
-
-/* Macroblock level features */
-typedef enum
-{
- MB_LVL_ALT_Q = 0, /* Use alternate Quantizer .... */
- MB_LVL_ALT_LF = 1, /* Use alternate loop filter value... */
- MB_LVL_MAX = 2 /* Number of MB level features supported */
-
-} MB_LVL_FEATURES;
-
-/* Segment Feature Masks */
-#define SEGMENT_ALTQ 0x01
-#define SEGMENT_ALT_LF 0x02
-
-#define VP8_YMODES (B_PRED + 1)
-#define VP8_UV_MODES (TM_PRED + 1)
-
-#define VP8_MVREFS (1 + SPLITMV - NEARESTMV)
-
-typedef enum
-{
- B_DC_PRED, /* average of above and left pixels */
- B_TM_PRED,
-
- B_VE_PRED, /* vertical prediction */
- B_HE_PRED, /* horizontal prediction */
-
- B_LD_PRED,
- B_RD_PRED,
-
- B_VR_PRED,
- B_VL_PRED,
- B_HD_PRED,
- B_HU_PRED,
-
- LEFT4X4,
- ABOVE4X4,
- ZERO4X4,
- NEW4X4,
-
- B_MODE_COUNT
-} B_PREDICTION_MODE;
-
-#define VP8_BINTRAMODES (B_HU_PRED + 1) /* 10 */
-#define VP8_SUBMVREFS (1 + NEW4X4 - LEFT4X4)
-
-/* For keyframes, intra block modes are predicted by the (already decoded)
- modes for the Y blocks to the left and above us; for interframes, there
- is a single probability table. */
-
-typedef struct
-{
- B_PREDICTION_MODE mode;
- union
- {
- int as_int;
- MV as_mv;
- } mv;
-} B_MODE_INFO;
-
-
-typedef enum
-{
- INTRA_FRAME = 0,
- LAST_FRAME = 1,
- GOLDEN_FRAME = 2,
- ALTREF_FRAME = 3,
- MAX_REF_FRAMES = 4
-} MV_REFERENCE_FRAME;
-
-typedef struct
-{
- MB_PREDICTION_MODE mode, uv_mode;
- MV_REFERENCE_FRAME ref_frame;
- union
- {
- int as_int;
- MV as_mv;
- } mv;
-
- unsigned char partitioning;
- unsigned char mb_skip_coeff; /* does this mb has coefficients at all, 1=no coefficients, 0=need decode tokens */
- unsigned char dc_diff;
- unsigned char need_to_clamp_mvs;
-
- unsigned char segment_id; /* Which set of segmentation parameters should be used for this MB */
-
- unsigned char force_no_skip; /* encoder only */
-} MB_MODE_INFO;
-
-
-typedef struct
-{
- MB_MODE_INFO mbmi;
- B_MODE_INFO bmi[16];
-} MODE_INFO;
-
-
-typedef struct
-{
- short *qcoeff;
- short *dqcoeff;
- unsigned char *predictor;
- short *diff;
- short *reference;
-
- short *dequant;
-
- /* 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries */
- unsigned char **base_pre;
- int pre;
- int pre_stride;
-
- unsigned char **base_dst;
- int dst;
- int dst_stride;
-
- int eob;
-
- B_MODE_INFO bmi;
-
-} BLOCKD;
-
-typedef struct
-{
- DECLARE_ALIGNED(16, short, diff[400]); /* from idct diff */
- DECLARE_ALIGNED(16, unsigned char, predictor[384]);
-/* not used DECLARE_ALIGNED(16, short, reference[384]); */
- DECLARE_ALIGNED(16, short, qcoeff[400]);
- DECLARE_ALIGNED(16, short, dqcoeff[400]);
- DECLARE_ALIGNED(16, char, eobs[25]);
-
- /* 16 Y blocks, 4 U, 4 V, 1 DC 2nd order block, each with 16 entries. */
- BLOCKD block[25];
-
- YV12_BUFFER_CONFIG pre; /* Filtered copy of previous frame reconstruction */
- YV12_BUFFER_CONFIG dst;
-
- MODE_INFO *mode_info_context;
- int mode_info_stride;
-
- FRAME_TYPE frame_type;
-
- int up_available;
- int left_available;
-
- /* Y,U,V,Y2 */
- ENTROPY_CONTEXT_PLANES *above_context;
- ENTROPY_CONTEXT_PLANES *left_context;
-
- /* 0 indicates segmentation at MB level is not enabled. Otherwise the individual bits indicate which features are active. */
- unsigned char segmentation_enabled;
-
- /* 0 (do not update) 1 (update) the macroblock segmentation map. */
- unsigned char update_mb_segmentation_map;
-
- /* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
- unsigned char update_mb_segmentation_data;
-
- /* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
- unsigned char mb_segement_abs_delta;
-
- /* Per frame flags that define which MB level features (such as quantizer or loop filter level) */
- /* are enabled and when enabled the proabilities used to decode the per MB flags in MB_MODE_INFO */
- vp8_prob mb_segment_tree_probs[MB_FEATURE_TREE_PROBS]; /* Probability Tree used to code Segment number */
-
- signed char segment_feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS]; /* Segment parameters */
-
- /* mode_based Loop filter adjustment */
- unsigned char mode_ref_lf_delta_enabled;
- unsigned char mode_ref_lf_delta_update;
-
- /* Delta values have the range +/- MAX_LOOP_FILTER */
- signed char last_ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
- signed char ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
- signed char last_mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
- signed char mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
-
- /* Distance of MB away from frame edges */
- int mb_to_left_edge;
- int mb_to_right_edge;
- int mb_to_top_edge;
- int mb_to_bottom_edge;
-
- unsigned int frames_since_golden;
- unsigned int frames_till_alt_ref_frame;
- vp8_subpix_fn_t subpixel_predict;
- vp8_subpix_fn_t subpixel_predict8x4;
- vp8_subpix_fn_t subpixel_predict8x8;
- vp8_subpix_fn_t subpixel_predict16x16;
-
- void *current_bc;
-
- int corrupted;
-
-#if CONFIG_RUNTIME_CPU_DETECT
- struct VP8_COMMON_RTCD *rtcd;
-#endif
-} MACROBLOCKD;
-
-
-extern void vp8_build_block_doffsets(MACROBLOCKD *x);
-extern void vp8_setup_block_dptrs(MACROBLOCKD *x);
-
-#endif /* __INC_BLOCKD_H */
diff --git a/vp8/common/coefupdateprobs.h b/vp8/common/coefupdateprobs.h
deleted file mode 100644
index 785e3ff..0000000
--- a/vp8/common/coefupdateprobs.h
+++ /dev/null
@@ -1,185 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/* Update probabilities for the nodes in the token entropy tree.
- Generated file included by entropy.c */
-
-const vp8_prob vp8_coef_update_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1] =
-{
- {
- {
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {176, 246, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {223, 241, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
- {249, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 244, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
- {234, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 246, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {239, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {251, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {251, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 254, 253, 255, 254, 255, 255, 255, 255, 255, 255, },
- {250, 255, 254, 255, 254, 255, 255, 255, 255, 255, 255, },
- {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- },
- {
- {
- {217, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {225, 252, 241, 253, 255, 255, 254, 255, 255, 255, 255, },
- {234, 250, 241, 250, 253, 255, 253, 254, 255, 255, 255, },
- },
- {
- {255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {223, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {238, 253, 254, 254, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {249, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {247, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {252, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
- {250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- },
- {
- {
- {186, 251, 250, 255, 255, 255, 255, 255, 255, 255, 255, },
- {234, 251, 244, 254, 255, 255, 255, 255, 255, 255, 255, },
- {251, 251, 243, 253, 254, 255, 254, 255, 255, 255, 255, },
- },
- {
- {255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {236, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {251, 253, 253, 254, 254, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {254, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {254, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- },
- {
- {
- {248, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {250, 254, 252, 254, 255, 255, 255, 255, 255, 255, 255, },
- {248, 254, 249, 253, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
- {246, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
- {252, 254, 251, 254, 254, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 254, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
- {248, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
- {253, 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {245, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {253, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 251, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
- {252, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 252, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {249, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 255, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
- {250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- {
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- {255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
- },
- },
-};
diff --git a/vp8/common/common.h b/vp8/common/common.h
deleted file mode 100644
index 9a93da9..0000000
--- a/vp8/common/common.h
+++ /dev/null
@@ -1,42 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef common_h
-#define common_h 1
-
-#include <assert.h>
-
-/* Interface header for common constant data structures and lookup tables */
-
-#include "vpx_mem/vpx_mem.h"
-
-#include "common_types.h"
-
-/* Only need this for fixed-size arrays, for structs just assign. */
-
-#define vp8_copy( Dest, Src) { \
- assert( sizeof( Dest) == sizeof( Src)); \
- vpx_memcpy( Dest, Src, sizeof( Src)); \
- }
-
-/* Use this for variably-sized arrays. */
-
-#define vp8_copy_array( Dest, Src, N) { \
- assert( sizeof( *Dest) == sizeof( *Src)); \
- vpx_memcpy( Dest, Src, N * sizeof( *Src)); \
- }
-
-#define vp8_zero( Dest) vpx_memset( &Dest, 0, sizeof( Dest));
-
-#define vp8_zero_array( Dest, N) vpx_memset( Dest, 0, N * sizeof( *Dest));
-
-
-#endif /* common_h */
diff --git a/vp8/common/common_types.h b/vp8/common/common_types.h
deleted file mode 100644
index 4e62486..0000000
--- a/vp8/common/common_types.h
+++ /dev/null
@@ -1,18 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_COMMON_TYPES
-#define __INC_COMMON_TYPES
-
-#define TRUE 1
-#define FALSE 0
-
-#endif
diff --git a/vp8/common/debugmodes.c b/vp8/common/debugmodes.c
deleted file mode 100644
index 8c03480..0000000
--- a/vp8/common/debugmodes.c
+++ /dev/null
@@ -1,157 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdio.h>
-#include "blockd.h"
-
-
-void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols, int frame)
-{
-
- int mb_row;
- int mb_col;
- int mb_index = 0;
- FILE *mvs = fopen("mvs.stt", "a");
-
- /* print out the macroblock Y modes */
- mb_index = 0;
- fprintf(mvs, "Mb Modes for Frame %d\n", frame);
-
- for (mb_row = 0; mb_row < rows; mb_row++)
- {
- for (mb_col = 0; mb_col < cols; mb_col++)
- {
-
- fprintf(mvs, "%2d ", mi[mb_index].mbmi.mode);
-
- mb_index++;
- }
-
- fprintf(mvs, "\n");
- mb_index++;
- }
-
- fprintf(mvs, "\n");
-
- mb_index = 0;
- fprintf(mvs, "Mb mv ref for Frame %d\n", frame);
-
- for (mb_row = 0; mb_row < rows; mb_row++)
- {
- for (mb_col = 0; mb_col < cols; mb_col++)
- {
-
- fprintf(mvs, "%2d ", mi[mb_index].mbmi.ref_frame);
-
- mb_index++;
- }
-
- fprintf(mvs, "\n");
- mb_index++;
- }
-
- fprintf(mvs, "\n");
-
- /* print out the macroblock UV modes */
- mb_index = 0;
- fprintf(mvs, "UV Modes for Frame %d\n", frame);
-
- for (mb_row = 0; mb_row < rows; mb_row++)
- {
- for (mb_col = 0; mb_col < cols; mb_col++)
- {
-
- fprintf(mvs, "%2d ", mi[mb_index].mbmi.uv_mode);
-
- mb_index++;
- }
-
- mb_index++;
- fprintf(mvs, "\n");
- }
-
- fprintf(mvs, "\n");
-
- /* print out the block modes */
- mb_index = 0;
- fprintf(mvs, "Mbs for Frame %d\n", frame);
- {
- int b_row;
-
- for (b_row = 0; b_row < 4 * rows; b_row++)
- {
- int b_col;
- int bindex;
-
- for (b_col = 0; b_col < 4 * cols; b_col++)
- {
- mb_index = (b_row >> 2) * (cols + 1) + (b_col >> 2);
- bindex = (b_row & 3) * 4 + (b_col & 3);
-
- if (mi[mb_index].mbmi.mode == B_PRED)
- fprintf(mvs, "%2d ", mi[mb_index].bmi[bindex].mode);
- else
- fprintf(mvs, "xx ");
-
- }
-
- fprintf(mvs, "\n");
- }
- }
- fprintf(mvs, "\n");
-
- /* print out the macroblock mvs */
- mb_index = 0;
- fprintf(mvs, "MVs for Frame %d\n", frame);
-
- for (mb_row = 0; mb_row < rows; mb_row++)
- {
- for (mb_col = 0; mb_col < cols; mb_col++)
- {
- fprintf(mvs, "%5d:%-5d", mi[mb_index].mbmi.mv.as_mv.row / 2, mi[mb_index].mbmi.mv.as_mv.col / 2);
-
- mb_index++;
- }
-
- mb_index++;
- fprintf(mvs, "\n");
- }
-
- fprintf(mvs, "\n");
-
-
- /* print out the block modes */
- mb_index = 0;
- fprintf(mvs, "MVs for Frame %d\n", frame);
- {
- int b_row;
-
- for (b_row = 0; b_row < 4 * rows; b_row++)
- {
- int b_col;
- int bindex;
-
- for (b_col = 0; b_col < 4 * cols; b_col++)
- {
- mb_index = (b_row >> 2) * (cols + 1) + (b_col >> 2);
- bindex = (b_row & 3) * 4 + (b_col & 3);
- fprintf(mvs, "%3d:%-3d ", mi[mb_index].bmi[bindex].mv.as_mv.row, mi[mb_index].bmi[bindex].mv.as_mv.col);
-
- }
-
- fprintf(mvs, "\n");
- }
- }
- fprintf(mvs, "\n");
-
-
- fclose(mvs);
-}
diff --git a/vp8/common/defaultcoefcounts.h b/vp8/common/defaultcoefcounts.h
deleted file mode 100644
index ca58d56..0000000
--- a/vp8/common/defaultcoefcounts.h
+++ /dev/null
@@ -1,221 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/* Generated file, included by entropy.c */
-
-static const unsigned int default_coef_counts [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens] =
-{
-
- {
- /* Block Type ( 0 ) */
- {
- /* Coeff Band ( 0 ) */
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- },
- {
- /* Coeff Band ( 1 ) */
- {30190, 26544, 225, 24, 4, 0, 0, 0, 0, 0, 0, 4171593,},
- {26846, 25157, 1241, 130, 26, 6, 1, 0, 0, 0, 0, 149987,},
- {10484, 9538, 1006, 160, 36, 18, 0, 0, 0, 0, 0, 15104,},
- },
- {
- /* Coeff Band ( 2 ) */
- {25842, 40456, 1126, 83, 11, 2, 0, 0, 0, 0, 0, 0,},
- {9338, 8010, 512, 73, 7, 3, 2, 0, 0, 0, 0, 43294,},
- {1047, 751, 149, 31, 13, 6, 1, 0, 0, 0, 0, 879,},
- },
- {
- /* Coeff Band ( 3 ) */
- {26136, 9826, 252, 13, 0, 0, 0, 0, 0, 0, 0, 0,},
- {8134, 5574, 191, 14, 2, 0, 0, 0, 0, 0, 0, 35302,},
- { 605, 677, 116, 9, 1, 0, 0, 0, 0, 0, 0, 611,},
- },
- {
- /* Coeff Band ( 4 ) */
- {10263, 15463, 283, 17, 0, 0, 0, 0, 0, 0, 0, 0,},
- {2773, 2191, 128, 9, 2, 2, 0, 0, 0, 0, 0, 10073,},
- { 134, 125, 32, 4, 0, 2, 0, 0, 0, 0, 0, 50,},
- },
- {
- /* Coeff Band ( 5 ) */
- {10483, 2663, 23, 1, 0, 0, 0, 0, 0, 0, 0, 0,},
- {2137, 1251, 27, 1, 1, 0, 0, 0, 0, 0, 0, 14362,},
- { 116, 156, 14, 2, 1, 0, 0, 0, 0, 0, 0, 190,},
- },
- {
- /* Coeff Band ( 6 ) */
- {40977, 27614, 412, 28, 0, 0, 0, 0, 0, 0, 0, 0,},
- {6113, 5213, 261, 22, 3, 0, 0, 0, 0, 0, 0, 26164,},
- { 382, 312, 50, 14, 2, 0, 0, 0, 0, 0, 0, 345,},
- },
- {
- /* Coeff Band ( 7 ) */
- { 0, 26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 0, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 319,},
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8,},
- },
- },
- {
- /* Block Type ( 1 ) */
- {
- /* Coeff Band ( 0 ) */
- {3268, 19382, 1043, 250, 93, 82, 49, 26, 17, 8, 25, 82289,},
- {8758, 32110, 5436, 1832, 827, 668, 420, 153, 24, 0, 3, 52914,},
- {9337, 23725, 8487, 3954, 2107, 1836, 1069, 399, 59, 0, 0, 18620,},
- },
- {
- /* Coeff Band ( 1 ) */
- {12419, 8420, 452, 62, 9, 1, 0, 0, 0, 0, 0, 0,},
- {11715, 8705, 693, 92, 15, 7, 2, 0, 0, 0, 0, 53988,},
- {7603, 8585, 2306, 778, 270, 145, 39, 5, 0, 0, 0, 9136,},
- },
- {
- /* Coeff Band ( 2 ) */
- {15938, 14335, 1207, 184, 55, 13, 4, 1, 0, 0, 0, 0,},
- {7415, 6829, 1138, 244, 71, 26, 7, 0, 0, 0, 0, 9980,},
- {1580, 1824, 655, 241, 89, 46, 10, 2, 0, 0, 0, 429,},
- },
- {
- /* Coeff Band ( 3 ) */
- {19453, 5260, 201, 19, 0, 0, 0, 0, 0, 0, 0, 0,},
- {9173, 3758, 213, 22, 1, 1, 0, 0, 0, 0, 0, 9820,},
- {1689, 1277, 276, 51, 17, 4, 0, 0, 0, 0, 0, 679,},
- },
- {
- /* Coeff Band ( 4 ) */
- {12076, 10667, 620, 85, 19, 9, 5, 0, 0, 0, 0, 0,},
- {4665, 3625, 423, 55, 19, 9, 0, 0, 0, 0, 0, 5127,},
- { 415, 440, 143, 34, 20, 7, 2, 0, 0, 0, 0, 101,},
- },
- {
- /* Coeff Band ( 5 ) */
- {12183, 4846, 115, 11, 1, 0, 0, 0, 0, 0, 0, 0,},
- {4226, 3149, 177, 21, 2, 0, 0, 0, 0, 0, 0, 7157,},
- { 375, 621, 189, 51, 11, 4, 1, 0, 0, 0, 0, 198,},
- },
- {
- /* Coeff Band ( 6 ) */
- {61658, 37743, 1203, 94, 10, 3, 0, 0, 0, 0, 0, 0,},
- {15514, 11563, 903, 111, 14, 5, 0, 0, 0, 0, 0, 25195,},
- { 929, 1077, 291, 78, 14, 7, 1, 0, 0, 0, 0, 507,},
- },
- {
- /* Coeff Band ( 7 ) */
- { 0, 990, 15, 3, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 0, 412, 13, 0, 0, 0, 0, 0, 0, 0, 0, 1641,},
- { 0, 18, 7, 1, 0, 0, 0, 0, 0, 0, 0, 30,},
- },
- },
- {
- /* Block Type ( 2 ) */
- {
- /* Coeff Band ( 0 ) */
- { 953, 24519, 628, 120, 28, 12, 4, 0, 0, 0, 0, 2248798,},
- {1525, 25654, 2647, 617, 239, 143, 42, 5, 0, 0, 0, 66837,},
- {1180, 11011, 3001, 1237, 532, 448, 239, 54, 5, 0, 0, 7122,},
- },
- {
- /* Coeff Band ( 1 ) */
- {1356, 2220, 67, 10, 4, 1, 0, 0, 0, 0, 0, 0,},
- {1450, 2544, 102, 18, 4, 3, 0, 0, 0, 0, 0, 57063,},
- {1182, 2110, 470, 130, 41, 21, 0, 0, 0, 0, 0, 6047,},
- },
- {
- /* Coeff Band ( 2 ) */
- { 370, 3378, 200, 30, 5, 4, 1, 0, 0, 0, 0, 0,},
- { 293, 1006, 131, 29, 11, 0, 0, 0, 0, 0, 0, 5404,},
- { 114, 387, 98, 23, 4, 8, 1, 0, 0, 0, 0, 236,},
- },
- {
- /* Coeff Band ( 3 ) */
- { 579, 194, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 395, 213, 5, 1, 0, 0, 0, 0, 0, 0, 0, 4157,},
- { 119, 122, 4, 0, 0, 0, 0, 0, 0, 0, 0, 300,},
- },
- {
- /* Coeff Band ( 4 ) */
- { 38, 557, 19, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 21, 114, 12, 1, 0, 0, 0, 0, 0, 0, 0, 427,},
- { 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7,},
- },
- {
- /* Coeff Band ( 5 ) */
- { 52, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 18, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 652,},
- { 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 30,},
- },
- {
- /* Coeff Band ( 6 ) */
- { 640, 569, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 25, 77, 2, 0, 0, 0, 0, 0, 0, 0, 0, 517,},
- { 4, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3,},
- },
- {
- /* Coeff Band ( 7 ) */
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- },
- },
- {
- /* Block Type ( 3 ) */
- {
- /* Coeff Band ( 0 ) */
- {2506, 20161, 2707, 767, 261, 178, 107, 30, 14, 3, 0, 100694,},
- {8806, 36478, 8817, 3268, 1280, 850, 401, 114, 42, 0, 0, 58572,},
- {11003, 27214, 11798, 5716, 2482, 2072, 1048, 175, 32, 0, 0, 19284,},
- },
- {
- /* Coeff Band ( 1 ) */
- {9738, 11313, 959, 205, 70, 18, 11, 1, 0, 0, 0, 0,},
- {12628, 15085, 1507, 273, 52, 19, 9, 0, 0, 0, 0, 54280,},
- {10701, 15846, 5561, 1926, 813, 570, 249, 36, 0, 0, 0, 6460,},
- },
- {
- /* Coeff Band ( 2 ) */
- {6781, 22539, 2784, 634, 182, 123, 20, 4, 0, 0, 0, 0,},
- {6263, 11544, 2649, 790, 259, 168, 27, 5, 0, 0, 0, 20539,},
- {3109, 4075, 2031, 896, 457, 386, 158, 29, 0, 0, 0, 1138,},
- },
- {
- /* Coeff Band ( 3 ) */
- {11515, 4079, 465, 73, 5, 14, 2, 0, 0, 0, 0, 0,},
- {9361, 5834, 650, 96, 24, 8, 4, 0, 0, 0, 0, 22181,},
- {4343, 3974, 1360, 415, 132, 96, 14, 1, 0, 0, 0, 1267,},
- },
- {
- /* Coeff Band ( 4 ) */
- {4787, 9297, 823, 168, 44, 12, 4, 0, 0, 0, 0, 0,},
- {3619, 4472, 719, 198, 60, 31, 3, 0, 0, 0, 0, 8401,},
- {1157, 1175, 483, 182, 88, 31, 8, 0, 0, 0, 0, 268,},
- },
- {
- /* Coeff Band ( 5 ) */
- {8299, 1226, 32, 5, 1, 0, 0, 0, 0, 0, 0, 0,},
- {3502, 1568, 57, 4, 1, 1, 0, 0, 0, 0, 0, 9811,},
- {1055, 1070, 166, 29, 6, 1, 0, 0, 0, 0, 0, 527,},
- },
- {
- /* Coeff Band ( 6 ) */
- {27414, 27927, 1989, 347, 69, 26, 0, 0, 0, 0, 0, 0,},
- {5876, 10074, 1574, 341, 91, 24, 4, 0, 0, 0, 0, 21954,},
- {1571, 2171, 778, 324, 124, 65, 16, 0, 0, 0, 0, 979,},
- },
- {
- /* Coeff Band ( 7 ) */
- { 0, 29, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
- { 0, 23, 0, 0, 0, 0, 0, 0, 0, 0, 0, 459,},
- { 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 13,},
- },
- },
-};
diff --git a/vp8/common/dma_desc.h b/vp8/common/dma_desc.h
deleted file mode 100644
index b923da6..0000000
--- a/vp8/common/dma_desc.h
+++ /dev/null
@@ -1,125 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef _dma_desc_h
-#define _dma_desc_h
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-
-#define NDSIZE_LG 0x00000900 // Next Descriptor Size
-#define NDSIZE_SM 0x00000800 // Next Descriptor Size
-#define NDSIZE_7 0x00000700 // Next Descriptor Size
-#define NDSIZE_6 0x00000600 // Next Descriptor Size
-#define NDSIZE_5 0x00000500 // Next Descriptor Size
-#define NDSIZE_4 0x00000400 // Next Descriptor Size
-#define NDSIZE_3 0x00000300 // Next Descriptor Size
-#define NDSIZE_2 0x00000200 // Next Descriptor Size
-#define NDSIZE_1 0x00000100 // Next Descriptor Size
-
-#define FLOW_STOP 0x0000
-#define FLOW_AUTO 0x1000
-#define FLOW_DESC_AR 0x4000
-#define FLOW_DESC_SM 0x6000
-#define FLOW_DESC_LG 0x7000
-
- typedef struct
- {
- unsigned int ndp;
- //unsigned short ndpl;
- //unsigned short ndph;
- unsigned int sa;
- //unsigned short sal;
- //unsigned short sah;
-
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- unsigned short ycnt;
- unsigned short ymod;
-
- } LARGE_DESC;
-
- typedef struct
- {
- unsigned short ndpl;
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- unsigned short ycnt;
- unsigned short ymod;
- } SMALL_DESC;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- unsigned short ycnt;
- unsigned short ymod;
- } ARRAY_DESC_7;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- unsigned short ycnt;
- } ARRAY_DESC_6;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- } ARRAY_DESC_5;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- } ARRAY_DESC_4;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- } ARRAY_DESC_3;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- } ARRAY_DESC_2;
-
- typedef struct
- {
- unsigned short sal;
- } ARRAY_DESC_1;
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif //_dma_desc_h
diff --git a/vp8/common/duck_io.h b/vp8/common/duck_io.h
deleted file mode 100644
index 43daa65..0000000
--- a/vp8/common/duck_io.h
+++ /dev/null
@@ -1,116 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef _duck_io_h
-#define _duck_io_h
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-#if defined (_WIN32)
- typedef __int64 int64_t;
-#elif defined(__MWERKS__)
- typedef long long int64_t;
-#elif defined(__APPLE__) || defined(__POWERPC)
-#include <ppc/types.h>
-#else
- typedef long long int64_t;
-#endif
-
- typedef struct
- {
- int64_t offset; // offset to start from
- int blocking; // non-zero for blocking
- } re_open_t;
-
-
- typedef enum
- {
- SAL_ERR_MAX = -10,
- SAL_ERROR = -11, // Default error
- SAL_ERR_WSASTARTUP = -12,
- SAL_ERR_SOCKET_CREATE = -13,
- SAL_ERR_RESOLVING_HOSTNAME = -14,
- SAL_ERR_SERVER_CONNECTION = -15,
- SAL_ERR_SENDING_DATA = -16,
- SAL_ERR_RECEIVING_DATA = -17,
- SAL_ERR_404_FILE_NOT_FOUND = -18,
- SAL_ERR_PARSING_HTTP_HEADER = -19,
- SAL_ERR_PARSING_CONTENT_LEN = -20,
- SAL_ERR_CONNECTION_TIMEOUT = -21,
- SAL_ERR_FILE_OPEN_FAILED = -22,
- SAL_ERR_MIN = -23
- } SAL_ERR; /* EMH 1-15-03 */
-
-
- typedef struct sal_err_map_temp
- {
- SAL_ERR code;
- const char *decode;
-
- } sal_err_map_t;
-
-
- static char *sal_err_text(SAL_ERR e)
- {
- int t;
- const sal_err_map_t g_sal_err_map[] =
- {
- { SAL_ERR_WSASTARTUP, "Error with WSAStartup" },
- { SAL_ERR_SOCKET_CREATE, "Error creating socket" },
- { SAL_ERR_RESOLVING_HOSTNAME, "Error resolving hostname" },
- { SAL_ERR_SERVER_CONNECTION, "Error connecting to server" },
- { SAL_ERR_SENDING_DATA, "Error sending data" },
- { SAL_ERR_RECEIVING_DATA, "Error receiving data" },
- { SAL_ERR_404_FILE_NOT_FOUND, "Error file not found " },
- { SAL_ERR_PARSING_HTTP_HEADER, "Error parsing http header" },
- { SAL_ERR_PARSING_CONTENT_LEN, "Error parsing content length" },
- { SAL_ERR_CONNECTION_TIMEOUT, "Error Connection timed out" },
- { SAL_ERR_FILE_OPEN_FAILED, "Error opening file" }
- };
-
- for (t = 0; t < sizeof(g_sal_err_map) / sizeof(sal_err_map_t); t++)
- {
- if (e == g_sal_err_map[t].code)
- return (char *) g_sal_err_map[t].decode;
- }
-
- return 0;
- }
-
-
-
-
-
-
-
- int duck_open(const char *fname, unsigned long user_data);
-
- void duck_close(int ghndl);
-
- int duck_read(int ghndl, unsigned char *buf, int nbytes);
-
- int64_t duck_seek(int g_hndl, int64_t offs, int origin);
-
- int duck_read_finished(int han, int flag); /* FWG 7-9-99 */
-
- int duck_name(int handle, char name[], size_t max_len); /* EMH 9-23-03 */
-
- int duck_read_blocking(int handle, unsigned char *buffer, int bytes); /* EMH 9-23-03 */
-
- int64_t duck_available_data(int handle); /* EMH 10-23-03 */
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif
diff --git a/vp8/common/entropy.c b/vp8/common/entropy.c
deleted file mode 100644
index a1fe4f4..0000000
--- a/vp8/common/entropy.c
+++ /dev/null
@@ -1,167 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdio.h>
-
-#include "entropy.h"
-#include "string.h"
-#include "blockd.h"
-#include "onyxc_int.h"
-
-#define uchar unsigned char /* typedefs can clash */
-#define uint unsigned int
-
-typedef const uchar cuchar;
-typedef const uint cuint;
-
-typedef vp8_prob Prob;
-
-#include "coefupdateprobs.h"
-
-DECLARE_ALIGNED(16, cuchar, vp8_coef_bands[16]) = { 0, 1, 2, 3, 6, 4, 5, 6, 6, 6, 6, 6, 6, 6, 6, 7};
-DECLARE_ALIGNED(16, cuchar, vp8_prev_token_class[MAX_ENTROPY_TOKENS]) = { 0, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 0};
-DECLARE_ALIGNED(16, const int, vp8_default_zig_zag1d[16]) =
-{
- 0, 1, 4, 8,
- 5, 2, 3, 6,
- 9, 12, 13, 10,
- 7, 11, 14, 15,
-};
-
-DECLARE_ALIGNED(16, const short, vp8_default_inv_zig_zag[16]) =
-{
- 1, 2, 6, 7,
- 3, 5, 8, 13,
- 4, 9, 12, 14,
- 10, 11, 15, 16
-};
-
-DECLARE_ALIGNED(16, short, vp8_default_zig_zag_mask[16]);
-
-const int vp8_mb_feature_data_bits[MB_LVL_MAX] = {7, 6};
-
-/* Array indices are identical to previously-existing CONTEXT_NODE indices */
-
-const vp8_tree_index vp8_coef_tree[ 22] = /* corresponding _CONTEXT_NODEs */
-{
- -DCT_EOB_TOKEN, 2, /* 0 = EOB */
- -ZERO_TOKEN, 4, /* 1 = ZERO */
- -ONE_TOKEN, 6, /* 2 = ONE */
- 8, 12, /* 3 = LOW_VAL */
- -TWO_TOKEN, 10, /* 4 = TWO */
- -THREE_TOKEN, -FOUR_TOKEN, /* 5 = THREE */
- 14, 16, /* 6 = HIGH_LOW */
- -DCT_VAL_CATEGORY1, -DCT_VAL_CATEGORY2, /* 7 = CAT_ONE */
- 18, 20, /* 8 = CAT_THREEFOUR */
- -DCT_VAL_CATEGORY3, -DCT_VAL_CATEGORY4, /* 9 = CAT_THREE */
- -DCT_VAL_CATEGORY5, -DCT_VAL_CATEGORY6 /* 10 = CAT_FIVE */
-};
-
-struct vp8_token_struct vp8_coef_encodings[vp8_coef_tokens];
-
-/* Trees for extra bits. Probabilities are constant and
- do not depend on previously encoded bits */
-
-static const Prob Pcat1[] = { 159};
-static const Prob Pcat2[] = { 165, 145};
-static const Prob Pcat3[] = { 173, 148, 140};
-static const Prob Pcat4[] = { 176, 155, 140, 135};
-static const Prob Pcat5[] = { 180, 157, 141, 134, 130};
-static const Prob Pcat6[] =
-{ 254, 254, 243, 230, 196, 177, 153, 140, 133, 130, 129};
-
-static vp8_tree_index cat1[2], cat2[4], cat3[6], cat4[8], cat5[10], cat6[22];
-
-void vp8_init_scan_order_mask()
-{
- int i;
-
- for (i = 0; i < 16; i++)
- {
- vp8_default_zig_zag_mask[vp8_default_zig_zag1d[i]] = 1 << i;
- }
-
-}
-
-static void init_bit_tree(vp8_tree_index *p, int n)
-{
- int i = 0;
-
- while (++i < n)
- {
- p[0] = p[1] = i << 1;
- p += 2;
- }
-
- p[0] = p[1] = 0;
-}
-
-static void init_bit_trees()
-{
- init_bit_tree(cat1, 1);
- init_bit_tree(cat2, 2);
- init_bit_tree(cat3, 3);
- init_bit_tree(cat4, 4);
- init_bit_tree(cat5, 5);
- init_bit_tree(cat6, 11);
-}
-
-vp8_extra_bit_struct vp8_extra_bits[12] =
-{
- { 0, 0, 0, 0},
- { 0, 0, 0, 1},
- { 0, 0, 0, 2},
- { 0, 0, 0, 3},
- { 0, 0, 0, 4},
- { cat1, Pcat1, 1, 5},
- { cat2, Pcat2, 2, 7},
- { cat3, Pcat3, 3, 11},
- { cat4, Pcat4, 4, 19},
- { cat5, Pcat5, 5, 35},
- { cat6, Pcat6, 11, 67},
- { 0, 0, 0, 0}
-};
-#include "defaultcoefcounts.h"
-
-void vp8_default_coef_probs(VP8_COMMON *pc)
-{
- int h = 0;
-
- do
- {
- int i = 0;
-
- do
- {
- int k = 0;
-
- do
- {
- unsigned int branch_ct [vp8_coef_tokens-1] [2];
- vp8_tree_probs_from_distribution(
- vp8_coef_tokens, vp8_coef_encodings, vp8_coef_tree,
- pc->fc.coef_probs [h][i][k], branch_ct, default_coef_counts [h][i][k],
- 256, 1);
-
- }
- while (++k < PREV_COEF_CONTEXTS);
- }
- while (++i < COEF_BANDS);
- }
- while (++h < BLOCK_TYPES);
-}
-
-
-void vp8_coef_tree_initialize()
-{
- init_bit_trees();
- vp8_tokens_from_tree(vp8_coef_encodings, vp8_coef_tree);
-}
diff --git a/vp8/common/entropy.h b/vp8/common/entropy.h
deleted file mode 100644
index d174e45..0000000
--- a/vp8/common/entropy.h
+++ /dev/null
@@ -1,102 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_ENTROPY_H
-#define __INC_ENTROPY_H
-
-#include "treecoder.h"
-#include "blockd.h"
-
-/* Coefficient token alphabet */
-
-#define ZERO_TOKEN 0 /* 0 Extra Bits 0+0 */
-#define ONE_TOKEN 1 /* 1 Extra Bits 0+1 */
-#define TWO_TOKEN 2 /* 2 Extra Bits 0+1 */
-#define THREE_TOKEN 3 /* 3 Extra Bits 0+1 */
-#define FOUR_TOKEN 4 /* 4 Extra Bits 0+1 */
-#define DCT_VAL_CATEGORY1 5 /* 5-6 Extra Bits 1+1 */
-#define DCT_VAL_CATEGORY2 6 /* 7-10 Extra Bits 2+1 */
-#define DCT_VAL_CATEGORY3 7 /* 11-18 Extra Bits 3+1 */
-#define DCT_VAL_CATEGORY4 8 /* 19-34 Extra Bits 4+1 */
-#define DCT_VAL_CATEGORY5 9 /* 35-66 Extra Bits 5+1 */
-#define DCT_VAL_CATEGORY6 10 /* 67+ Extra Bits 11+1 */
-#define DCT_EOB_TOKEN 11 /* EOB Extra Bits 0+0 */
-
-#define vp8_coef_tokens 12
-#define MAX_ENTROPY_TOKENS vp8_coef_tokens
-#define ENTROPY_NODES 11
-
-extern const vp8_tree_index vp8_coef_tree[];
-
-extern struct vp8_token_struct vp8_coef_encodings[vp8_coef_tokens];
-
-typedef struct
-{
- vp8_tree_p tree;
- const vp8_prob *prob;
- int Len;
- int base_val;
-} vp8_extra_bit_struct;
-
-extern vp8_extra_bit_struct vp8_extra_bits[12]; /* indexed by token value */
-
-#define PROB_UPDATE_BASELINE_COST 7
-
-#define MAX_PROB 255
-#define DCT_MAX_VALUE 2048
-
-
-/* Coefficients are predicted via a 3-dimensional probability table. */
-
-/* Outside dimension. 0 = Y no DC, 1 = Y2, 2 = UV, 3 = Y with DC */
-
-#define BLOCK_TYPES 4
-
-/* Middle dimension is a coarsening of the coefficient's
- position within the 4x4 DCT. */
-
-#define COEF_BANDS 8
-extern DECLARE_ALIGNED(16, const unsigned char, vp8_coef_bands[16]);
-
-/* Inside dimension is 3-valued measure of nearby complexity, that is,
- the extent to which nearby coefficients are nonzero. For the first
- coefficient (DC, unless block type is 0), we look at the (already encoded)
- blocks above and to the left of the current block. The context index is
- then the number (0,1,or 2) of these blocks having nonzero coefficients.
- After decoding a coefficient, the measure is roughly the size of the
- most recently decoded coefficient (0 for 0, 1 for 1, 2 for >1).
- Note that the intuitive meaning of this measure changes as coefficients
- are decoded, e.g., prior to the first token, a zero means that my neighbors
- are empty while, after the first token, because of the use of end-of-block,
- a zero means we just decoded a zero and hence guarantees that a non-zero
- coefficient will appear later in this block. However, this shift
- in meaning is perfectly OK because our context depends also on the
- coefficient band (and since zigzag positions 0, 1, and 2 are in
- distinct bands). */
-
-/*# define DC_TOKEN_CONTEXTS 3*/ /* 00, 0!0, !0!0 */
-# define PREV_COEF_CONTEXTS 3
-
-extern DECLARE_ALIGNED(16, const unsigned char, vp8_prev_token_class[vp8_coef_tokens]);
-
-extern const vp8_prob vp8_coef_update_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1];
-
-
-struct VP8Common;
-void vp8_default_coef_probs(struct VP8Common *);
-
-extern DECLARE_ALIGNED(16, const int, vp8_default_zig_zag1d[16]);
-extern DECLARE_ALIGNED(16, const short, vp8_default_inv_zig_zag[16]);
-extern short vp8_default_zig_zag_mask[16];
-extern const int vp8_mb_feature_data_bits[MB_LVL_MAX];
-
-void vp8_coef_tree_initialize(void);
-#endif
diff --git a/vp8/common/entropymode.c b/vp8/common/entropymode.c
deleted file mode 100644
index e9dc668..0000000
--- a/vp8/common/entropymode.c
+++ /dev/null
@@ -1,273 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "entropymode.h"
-#include "entropy.h"
-#include "vpx_mem/vpx_mem.h"
-
-static const unsigned int kf_y_mode_cts[VP8_YMODES] = { 1607, 915, 812, 811, 5455};
-static const unsigned int y_mode_cts [VP8_YMODES] = { 8080, 1908, 1582, 1007, 5874};
-
-static const unsigned int uv_mode_cts [VP8_UV_MODES] = { 59483, 13605, 16492, 4230};
-static const unsigned int kf_uv_mode_cts[VP8_UV_MODES] = { 5319, 1904, 1703, 674};
-
-static const unsigned int bmode_cts[VP8_BINTRAMODES] =
-{
- 43891, 17694, 10036, 3920, 3363, 2546, 5119, 3221, 2471, 1723
-};
-
-typedef enum
-{
- SUBMVREF_NORMAL,
- SUBMVREF_LEFT_ZED,
- SUBMVREF_ABOVE_ZED,
- SUBMVREF_LEFT_ABOVE_SAME,
- SUBMVREF_LEFT_ABOVE_ZED
-} sumvfref_t;
-
-int vp8_mv_cont(const MV *l, const MV *a)
-{
- int lez = (l->row == 0 && l->col == 0);
- int aez = (a->row == 0 && a->col == 0);
- int lea = (l->row == a->row && l->col == a->col);
-
- if (lea && lez)
- return SUBMVREF_LEFT_ABOVE_ZED;
-
- if (lea)
- return SUBMVREF_LEFT_ABOVE_SAME;
-
- if (aez)
- return SUBMVREF_ABOVE_ZED;
-
- if (lez)
- return SUBMVREF_LEFT_ZED;
-
- return SUBMVREF_NORMAL;
-}
-
-static const vp8_prob sub_mv_ref_prob [VP8_SUBMVREFS-1] = { 180, 162, 25};
-
-const vp8_prob vp8_sub_mv_ref_prob2 [SUBMVREF_COUNT][VP8_SUBMVREFS-1] =
-{
- { 147, 136, 18 },
- { 106, 145, 1 },
- { 179, 121, 1 },
- { 223, 1 , 34 },
- { 208, 1 , 1 }
-};
-
-
-
-vp8_mbsplit vp8_mbsplits [VP8_NUMMBSPLITS] =
-{
- {
- 0, 0, 0, 0,
- 0, 0, 0, 0,
- 1, 1, 1, 1,
- 1, 1, 1, 1,
- },
- {
- 0, 0, 1, 1,
- 0, 0, 1, 1,
- 0, 0, 1, 1,
- 0, 0, 1, 1,
- },
- {
- 0, 0, 1, 1,
- 0, 0, 1, 1,
- 2, 2, 3, 3,
- 2, 2, 3, 3,
- },
- {
- 0, 1, 2, 3,
- 4, 5, 6, 7,
- 8, 9, 10, 11,
- 12, 13, 14, 15,
- },
-};
-
-const int vp8_mbsplit_count [VP8_NUMMBSPLITS] = { 2, 2, 4, 16};
-
-const vp8_prob vp8_mbsplit_probs [VP8_NUMMBSPLITS-1] = { 110, 111, 150};
-
-
-/* Array indices are identical to previously-existing INTRAMODECONTEXTNODES. */
-
-const vp8_tree_index vp8_bmode_tree[18] = /* INTRAMODECONTEXTNODE value */
-{
- -B_DC_PRED, 2, /* 0 = DC_NODE */
- -B_TM_PRED, 4, /* 1 = TM_NODE */
- -B_VE_PRED, 6, /* 2 = VE_NODE */
- 8, 12, /* 3 = COM_NODE */
- -B_HE_PRED, 10, /* 4 = HE_NODE */
- -B_RD_PRED, -B_VR_PRED, /* 5 = RD_NODE */
- -B_LD_PRED, 14, /* 6 = LD_NODE */
- -B_VL_PRED, 16, /* 7 = VL_NODE */
- -B_HD_PRED, -B_HU_PRED /* 8 = HD_NODE */
-};
-
-/* Again, these trees use the same probability indices as their
- explicitly-programmed predecessors. */
-
-const vp8_tree_index vp8_ymode_tree[8] =
-{
- -DC_PRED, 2,
- 4, 6,
- -V_PRED, -H_PRED,
- -TM_PRED, -B_PRED
-};
-
-const vp8_tree_index vp8_kf_ymode_tree[8] =
-{
- -B_PRED, 2,
- 4, 6,
- -DC_PRED, -V_PRED,
- -H_PRED, -TM_PRED
-};
-
-const vp8_tree_index vp8_uv_mode_tree[6] =
-{
- -DC_PRED, 2,
- -V_PRED, 4,
- -H_PRED, -TM_PRED
-};
-
-const vp8_tree_index vp8_mbsplit_tree[6] =
-{
- -3, 2,
- -2, 4,
- -0, -1
-};
-
-const vp8_tree_index vp8_mv_ref_tree[8] =
-{
- -ZEROMV, 2,
- -NEARESTMV, 4,
- -NEARMV, 6,
- -NEWMV, -SPLITMV
-};
-
-const vp8_tree_index vp8_sub_mv_ref_tree[6] =
-{
- -LEFT4X4, 2,
- -ABOVE4X4, 4,
- -ZERO4X4, -NEW4X4
-};
-
-
-struct vp8_token_struct vp8_bmode_encodings [VP8_BINTRAMODES];
-struct vp8_token_struct vp8_ymode_encodings [VP8_YMODES];
-struct vp8_token_struct vp8_kf_ymode_encodings [VP8_YMODES];
-struct vp8_token_struct vp8_uv_mode_encodings [VP8_UV_MODES];
-struct vp8_token_struct vp8_mbsplit_encodings [VP8_NUMMBSPLITS];
-
-struct vp8_token_struct vp8_mv_ref_encoding_array [VP8_MVREFS];
-struct vp8_token_struct vp8_sub_mv_ref_encoding_array [VP8_SUBMVREFS];
-
-
-const vp8_tree_index vp8_small_mvtree [14] =
-{
- 2, 8,
- 4, 6,
- -0, -1,
- -2, -3,
- 10, 12,
- -4, -5,
- -6, -7
-};
-
-struct vp8_token_struct vp8_small_mvencodings [8];
-
-void vp8_init_mbmode_probs(VP8_COMMON *x)
-{
- unsigned int bct [VP8_YMODES] [2]; /* num Ymodes > num UV modes */
-
- vp8_tree_probs_from_distribution(
- VP8_YMODES, vp8_ymode_encodings, vp8_ymode_tree,
- x->fc.ymode_prob, bct, y_mode_cts,
- 256, 1
- );
- vp8_tree_probs_from_distribution(
- VP8_YMODES, vp8_kf_ymode_encodings, vp8_kf_ymode_tree,
- x->kf_ymode_prob, bct, kf_y_mode_cts,
- 256, 1
- );
- vp8_tree_probs_from_distribution(
- VP8_UV_MODES, vp8_uv_mode_encodings, vp8_uv_mode_tree,
- x->fc.uv_mode_prob, bct, uv_mode_cts,
- 256, 1
- );
- vp8_tree_probs_from_distribution(
- VP8_UV_MODES, vp8_uv_mode_encodings, vp8_uv_mode_tree,
- x->kf_uv_mode_prob, bct, kf_uv_mode_cts,
- 256, 1
- );
- vpx_memcpy(x->fc.sub_mv_ref_prob, sub_mv_ref_prob, sizeof(sub_mv_ref_prob));
-}
-
-
-static void intra_bmode_probs_from_distribution(
- vp8_prob p [VP8_BINTRAMODES-1],
- unsigned int branch_ct [VP8_BINTRAMODES-1] [2],
- const unsigned int events [VP8_BINTRAMODES]
-)
-{
- vp8_tree_probs_from_distribution(
- VP8_BINTRAMODES, vp8_bmode_encodings, vp8_bmode_tree,
- p, branch_ct, events,
- 256, 1
- );
-}
-
-void vp8_default_bmode_probs(vp8_prob p [VP8_BINTRAMODES-1])
-{
- unsigned int branch_ct [VP8_BINTRAMODES-1] [2];
- intra_bmode_probs_from_distribution(p, branch_ct, bmode_cts);
-}
-
-void vp8_kf_default_bmode_probs(vp8_prob p [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES-1])
-{
- unsigned int branch_ct [VP8_BINTRAMODES-1] [2];
-
- int i = 0;
-
- do
- {
- int j = 0;
-
- do
- {
- intra_bmode_probs_from_distribution(
- p[i][j], branch_ct, vp8_kf_default_bmode_counts[i][j]);
-
- }
- while (++j < VP8_BINTRAMODES);
- }
- while (++i < VP8_BINTRAMODES);
-}
-
-
-void vp8_entropy_mode_init()
-{
- vp8_tokens_from_tree(vp8_bmode_encodings, vp8_bmode_tree);
- vp8_tokens_from_tree(vp8_ymode_encodings, vp8_ymode_tree);
- vp8_tokens_from_tree(vp8_kf_ymode_encodings, vp8_kf_ymode_tree);
- vp8_tokens_from_tree(vp8_uv_mode_encodings, vp8_uv_mode_tree);
- vp8_tokens_from_tree(vp8_mbsplit_encodings, vp8_mbsplit_tree);
-
- vp8_tokens_from_tree_offset(vp8_mv_ref_encoding_array,
- vp8_mv_ref_tree, NEARESTMV);
- vp8_tokens_from_tree_offset(vp8_sub_mv_ref_encoding_array,
- vp8_sub_mv_ref_tree, LEFT4X4);
-
- vp8_tokens_from_tree(vp8_small_mvencodings, vp8_small_mvtree);
-}
diff --git a/vp8/common/entropymode.h b/vp8/common/entropymode.h
deleted file mode 100644
index da6ae8e..0000000
--- a/vp8/common/entropymode.h
+++ /dev/null
@@ -1,68 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_ENTROPYMODE_H
-#define __INC_ENTROPYMODE_H
-
-#include "onyxc_int.h"
-#include "treecoder.h"
-
-typedef const int vp8_mbsplit[16];
-
-#define VP8_NUMMBSPLITS 4
-
-extern vp8_mbsplit vp8_mbsplits [VP8_NUMMBSPLITS];
-
-extern const int vp8_mbsplit_count [VP8_NUMMBSPLITS]; /* # of subsets */
-
-extern const vp8_prob vp8_mbsplit_probs [VP8_NUMMBSPLITS-1];
-
-extern int vp8_mv_cont(const MV *l, const MV *a);
-#define SUBMVREF_COUNT 5
-extern const vp8_prob vp8_sub_mv_ref_prob2 [SUBMVREF_COUNT][VP8_SUBMVREFS-1];
-
-
-extern const unsigned int vp8_kf_default_bmode_counts [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES];
-
-
-extern const vp8_tree_index vp8_bmode_tree[];
-
-extern const vp8_tree_index vp8_ymode_tree[];
-extern const vp8_tree_index vp8_kf_ymode_tree[];
-extern const vp8_tree_index vp8_uv_mode_tree[];
-
-extern const vp8_tree_index vp8_mbsplit_tree[];
-extern const vp8_tree_index vp8_mv_ref_tree[];
-extern const vp8_tree_index vp8_sub_mv_ref_tree[];
-
-extern struct vp8_token_struct vp8_bmode_encodings [VP8_BINTRAMODES];
-extern struct vp8_token_struct vp8_ymode_encodings [VP8_YMODES];
-extern struct vp8_token_struct vp8_kf_ymode_encodings [VP8_YMODES];
-extern struct vp8_token_struct vp8_uv_mode_encodings [VP8_UV_MODES];
-extern struct vp8_token_struct vp8_mbsplit_encodings [VP8_NUMMBSPLITS];
-
-/* Inter mode values do not start at zero */
-
-extern struct vp8_token_struct vp8_mv_ref_encoding_array [VP8_MVREFS];
-extern struct vp8_token_struct vp8_sub_mv_ref_encoding_array [VP8_SUBMVREFS];
-
-extern const vp8_tree_index vp8_small_mvtree[];
-
-extern struct vp8_token_struct vp8_small_mvencodings [8];
-
-void vp8_entropy_mode_init(void);
-
-void vp8_init_mbmode_probs(VP8_COMMON *x);
-
-void vp8_default_bmode_probs(vp8_prob dest [VP8_BINTRAMODES-1]);
-void vp8_kf_default_bmode_probs(vp8_prob dest [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES-1]);
-
-#endif
diff --git a/vp8/common/entropymv.h b/vp8/common/entropymv.h
deleted file mode 100644
index 911507d..0000000
--- a/vp8/common/entropymv.h
+++ /dev/null
@@ -1,42 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_ENTROPYMV_H
-#define __INC_ENTROPYMV_H
-
-#include "treecoder.h"
-
-enum
-{
- mv_max = 1023, /* max absolute value of a MV component */
- MVvals = (2 * mv_max) + 1, /* # possible values "" */
-
- mvlong_width = 10, /* Large MVs have 9 bit magnitudes */
- mvnum_short = 8, /* magnitudes 0 through 7 */
-
- /* probability offsets for coding each MV component */
-
- mvpis_short = 0, /* short (<= 7) vs long (>= 8) */
- MVPsign, /* sign for non-zero */
- MVPshort, /* 8 short values = 7-position tree */
-
- MVPbits = MVPshort + mvnum_short - 1, /* mvlong_width long value bits */
- MVPcount = MVPbits + mvlong_width /* (with independent probabilities) */
-};
-
-typedef struct mv_context
-{
- vp8_prob prob[MVPcount]; /* often come in row, col pairs */
-} MV_CONTEXT;
-
-extern const MV_CONTEXT vp8_mv_update_probs[2], vp8_default_mv_context[2];
-
-#endif
diff --git a/vp8/common/extend.c b/vp8/common/extend.c
deleted file mode 100644
index 47207fa..0000000
--- a/vp8/common/extend.c
+++ /dev/null
@@ -1,125 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "extend.h"
-#include "vpx_mem/vpx_mem.h"
-
-
-static void extend_plane_borders
-(
- unsigned char *s, /* source */
- int sp, /* pitch */
- int h, /* height */
- int w, /* width */
- int et, /* extend top border */
- int el, /* extend left border */
- int eb, /* extend bottom border */
- int er /* extend right border */
-)
-{
-
- int i;
- unsigned char *src_ptr1, *src_ptr2;
- unsigned char *dest_ptr1, *dest_ptr2;
- int linesize;
-
- /* copy the left and right most columns out */
- src_ptr1 = s;
- src_ptr2 = s + w - 1;
- dest_ptr1 = s - el;
- dest_ptr2 = s + w;
-
- for (i = 0; i < h - 0 + 1; i++)
- {
- /* Some linkers will complain if we call vpx_memset with el set to a
- * constant 0.
- */
- if (el)
- vpx_memset(dest_ptr1, src_ptr1[0], el);
- vpx_memset(dest_ptr2, src_ptr2[0], er);
- src_ptr1 += sp;
- src_ptr2 += sp;
- dest_ptr1 += sp;
- dest_ptr2 += sp;
- }
-
- /* Now copy the top and bottom source lines into each line of the respective borders */
- src_ptr1 = s - el;
- src_ptr2 = s + sp * (h - 1) - el;
- dest_ptr1 = s + sp * (-et) - el;
- dest_ptr2 = s + sp * (h) - el;
- linesize = el + er + w + 1;
-
- for (i = 0; i < (int)et; i++)
- {
- vpx_memcpy(dest_ptr1, src_ptr1, linesize);
- dest_ptr1 += sp;
- }
-
- for (i = 0; i < (int)eb; i++)
- {
- vpx_memcpy(dest_ptr2, src_ptr2, linesize);
- dest_ptr2 += sp;
- }
-}
-
-
-void vp8_extend_to_multiple_of16(YV12_BUFFER_CONFIG *ybf, int width, int height)
-{
- int er = 0xf & (16 - (width & 0xf));
- int eb = 0xf & (16 - (height & 0xf));
-
- /* check for non multiples of 16 */
- if (er != 0 || eb != 0)
- {
- extend_plane_borders(ybf->y_buffer, ybf->y_stride, height, width, 0, 0, eb, er);
-
- /* adjust for uv */
- height = (height + 1) >> 1;
- width = (width + 1) >> 1;
- er = 0x7 & (8 - (width & 0x7));
- eb = 0x7 & (8 - (height & 0x7));
-
- if (er || eb)
- {
- extend_plane_borders(ybf->u_buffer, ybf->uv_stride, height, width, 0, 0, eb, er);
- extend_plane_borders(ybf->v_buffer, ybf->uv_stride, height, width, 0, 0, eb, er);
- }
- }
-}
-
-/* note the extension is only for the last row, for intra prediction purpose */
-void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf, unsigned char *YPtr, unsigned char *UPtr, unsigned char *VPtr)
-{
- int i;
-
- YPtr += ybf->y_stride * 14;
- UPtr += ybf->uv_stride * 6;
- VPtr += ybf->uv_stride * 6;
-
- for (i = 0; i < 4; i++)
- {
- YPtr[i] = YPtr[-1];
- UPtr[i] = UPtr[-1];
- VPtr[i] = VPtr[-1];
- }
-
- YPtr += ybf->y_stride;
- UPtr += ybf->uv_stride;
- VPtr += ybf->uv_stride;
-
- for (i = 0; i < 4; i++)
- {
- YPtr[i] = YPtr[-1];
- UPtr[i] = UPtr[-1];
- VPtr[i] = VPtr[-1];
- }
-}
diff --git a/vp8/common/extend.h b/vp8/common/extend.h
deleted file mode 100644
index fd0a608..0000000
--- a/vp8/common/extend.h
+++ /dev/null
@@ -1,21 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_EXTEND_H
-#define __INC_EXTEND_H
-
-#include "vpx_scale/yv12config.h"
-
-void Extend(YV12_BUFFER_CONFIG *ybf);
-void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf, unsigned char *YPtr, unsigned char *UPtr, unsigned char *VPtr);
-void vp8_extend_to_multiple_of16(YV12_BUFFER_CONFIG *ybf, int width, int height);
-
-#endif
diff --git a/vp8/common/filter.c b/vp8/common/filter.c
deleted file mode 100644
index ae59529..0000000
--- a/vp8/common/filter.c
+++ /dev/null
@@ -1,494 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdlib.h>
-#include "filter.h"
-#include "vpx_ports/mem.h"
-
-DECLARE_ALIGNED(16, const short, vp8_bilinear_filters[8][2]) =
-{
- { 128, 0 },
- { 112, 16 },
- { 96, 32 },
- { 80, 48 },
- { 64, 64 },
- { 48, 80 },
- { 32, 96 },
- { 16, 112 }
-};
-
-DECLARE_ALIGNED(16, const short, vp8_sub_pel_filters[8][6]) =
-{
-
- { 0, 0, 128, 0, 0, 0 }, /* note that 1/8 pel positions are just as per alpha -0.5 bicubic */
- { 0, -6, 123, 12, -1, 0 },
- { 2, -11, 108, 36, -8, 1 }, /* New 1/4 pel 6 tap filter */
- { 0, -9, 93, 50, -6, 0 },
- { 3, -16, 77, 77, -16, 3 }, /* New 1/2 pel 6 tap filter */
- { 0, -6, 50, 93, -9, 0 },
- { 1, -8, 36, 108, -11, 2 }, /* New 1/4 pel 6 tap filter */
- { 0, -1, 12, 123, -6, 0 },
-};
-
-static void filter_block2d_first_pass
-(
- unsigned char *src_ptr,
- int *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-)
-{
- unsigned int i, j;
- int Temp;
-
- for (i = 0; i < output_height; i++)
- {
- for (j = 0; j < output_width; j++)
- {
- Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
- ((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
- ((int)src_ptr[0] * vp8_filter[2]) +
- ((int)src_ptr[pixel_step] * vp8_filter[3]) +
- ((int)src_ptr[2*pixel_step] * vp8_filter[4]) +
- ((int)src_ptr[3*pixel_step] * vp8_filter[5]) +
- (VP8_FILTER_WEIGHT >> 1); /* Rounding */
-
- /* Normalize back to 0-255 */
- Temp = Temp >> VP8_FILTER_SHIFT;
-
- if (Temp < 0)
- Temp = 0;
- else if (Temp > 255)
- Temp = 255;
-
- output_ptr[j] = Temp;
- src_ptr++;
- }
-
- /* Next row... */
- src_ptr += src_pixels_per_line - output_width;
- output_ptr += output_width;
- }
-}
-
-static void filter_block2d_second_pass
-(
- int *src_ptr,
- unsigned char *output_ptr,
- int output_pitch,
- unsigned int src_pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-)
-{
- unsigned int i, j;
- int Temp;
-
- for (i = 0; i < output_height; i++)
- {
- for (j = 0; j < output_width; j++)
- {
- /* Apply filter */
- Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
- ((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
- ((int)src_ptr[0] * vp8_filter[2]) +
- ((int)src_ptr[pixel_step] * vp8_filter[3]) +
- ((int)src_ptr[2*pixel_step] * vp8_filter[4]) +
- ((int)src_ptr[3*pixel_step] * vp8_filter[5]) +
- (VP8_FILTER_WEIGHT >> 1); /* Rounding */
-
- /* Normalize back to 0-255 */
- Temp = Temp >> VP8_FILTER_SHIFT;
-
- if (Temp < 0)
- Temp = 0;
- else if (Temp > 255)
- Temp = 255;
-
- output_ptr[j] = (unsigned char)Temp;
- src_ptr++;
- }
-
- /* Start next row */
- src_ptr += src_pixels_per_line - output_width;
- output_ptr += output_pitch;
- }
-}
-
-
-static void filter_block2d
-(
- unsigned char *src_ptr,
- unsigned char *output_ptr,
- unsigned int src_pixels_per_line,
- int output_pitch,
- const short *HFilter,
- const short *VFilter
-)
-{
- int FData[9*4]; /* Temp data buffer used in filtering */
-
- /* First filter 1-D horizontally... */
- filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 9, 4, HFilter);
-
- /* then filter verticaly... */
- filter_block2d_second_pass(FData + 8, output_ptr, output_pitch, 4, 4, 4, 4, VFilter);
-}
-
-
-void vp8_sixtap_predict_c
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
- VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
-
- filter_block2d(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter);
-}
-void vp8_sixtap_predict8x8_c
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
- int FData[13*16]; /* Temp data buffer used in filtering */
-
- HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
- VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
-
- /* First filter 1-D horizontally... */
- filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 13, 8, HFilter);
-
-
- /* then filter verticaly... */
- filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 8, 8, VFilter);
-
-}
-
-void vp8_sixtap_predict8x4_c
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
- int FData[13*16]; /* Temp data buffer used in filtering */
-
- HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
- VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
-
- /* First filter 1-D horizontally... */
- filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 9, 8, HFilter);
-
-
- /* then filter verticaly... */
- filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 4, 8, VFilter);
-
-}
-
-void vp8_sixtap_predict16x16_c
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
- int FData[21*24]; /* Temp data buffer used in filtering */
-
-
- HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
- VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
-
- /* First filter 1-D horizontally... */
- filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 21, 16, HFilter);
-
- /* then filter verticaly... */
- filter_block2d_second_pass(FData + 32, dst_ptr, dst_pitch, 16, 16, 16, 16, VFilter);
-
-}
-
-
-/****************************************************************************
- *
- * ROUTINE : filter_block2d_bil_first_pass
- *
- * INPUTS : UINT8 *src_ptr : Pointer to source block.
- * UINT32 src_stride : Stride of source block.
- * UINT32 height : Block height.
- * UINT32 width : Block width.
- * INT32 *vp8_filter : Array of 2 bi-linear filter taps.
- *
- * OUTPUTS : INT32 *dst_ptr : Pointer to filtered block.
- *
- * RETURNS : void
- *
- * FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block
- * in the horizontal direction to produce the filtered output
- * block. Used to implement first-pass of 2-D separable filter.
- *
- * SPECIAL NOTES : Produces INT32 output to retain precision for next pass.
- * Two filter taps should sum to VP8_FILTER_WEIGHT.
- *
- ****************************************************************************/
-static void filter_block2d_bil_first_pass
-(
- unsigned char *src_ptr,
- unsigned short *dst_ptr,
- unsigned int src_stride,
- unsigned int height,
- unsigned int width,
- const short *vp8_filter
-)
-{
- unsigned int i, j;
-
- for (i = 0; i < height; i++)
- {
- for (j = 0; j < width; j++)
- {
- /* Apply bilinear filter */
- dst_ptr[j] = (((int)src_ptr[0] * vp8_filter[0]) +
- ((int)src_ptr[1] * vp8_filter[1]) +
- (VP8_FILTER_WEIGHT / 2)) >> VP8_FILTER_SHIFT;
- src_ptr++;
- }
-
- /* Next row... */
- src_ptr += src_stride - width;
- dst_ptr += width;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : filter_block2d_bil_second_pass
- *
- * INPUTS : INT32 *src_ptr : Pointer to source block.
- * UINT32 dst_pitch : Destination block pitch.
- * UINT32 height : Block height.
- * UINT32 width : Block width.
- * INT32 *vp8_filter : Array of 2 bi-linear filter taps.
- *
- * OUTPUTS : UINT16 *dst_ptr : Pointer to filtered block.
- *
- * RETURNS : void
- *
- * FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block
- * in the vertical direction to produce the filtered output
- * block. Used to implement second-pass of 2-D separable filter.
- *
- * SPECIAL NOTES : Requires 32-bit input as produced by filter_block2d_bil_first_pass.
- * Two filter taps should sum to VP8_FILTER_WEIGHT.
- *
- ****************************************************************************/
-static void filter_block2d_bil_second_pass
-(
- unsigned short *src_ptr,
- unsigned char *dst_ptr,
- int dst_pitch,
- unsigned int height,
- unsigned int width,
- const short *vp8_filter
-)
-{
- unsigned int i, j;
- int Temp;
-
- for (i = 0; i < height; i++)
- {
- for (j = 0; j < width; j++)
- {
- /* Apply filter */
- Temp = ((int)src_ptr[0] * vp8_filter[0]) +
- ((int)src_ptr[width] * vp8_filter[1]) +
- (VP8_FILTER_WEIGHT / 2);
- dst_ptr[j] = (unsigned int)(Temp >> VP8_FILTER_SHIFT);
- src_ptr++;
- }
-
- /* Next row... */
- dst_ptr += dst_pitch;
- }
-}
-
-
-/****************************************************************************
- *
- * ROUTINE : filter_block2d_bil
- *
- * INPUTS : UINT8 *src_ptr : Pointer to source block.
- * UINT32 src_pitch : Stride of source block.
- * UINT32 dst_pitch : Stride of destination block.
- * INT32 *HFilter : Array of 2 horizontal filter taps.
- * INT32 *VFilter : Array of 2 vertical filter taps.
- * INT32 Width : Block width
- * INT32 Height : Block height
- *
- * OUTPUTS : UINT16 *dst_ptr : Pointer to filtered block.
- *
- * RETURNS : void
- *
- * FUNCTION : 2-D filters an input block by applying a 2-tap
- * bi-linear filter horizontally followed by a 2-tap
- * bi-linear filter vertically on the result.
- *
- * SPECIAL NOTES : The largest block size can be handled here is 16x16
- *
- ****************************************************************************/
-static void filter_block2d_bil
-(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- unsigned int src_pitch,
- unsigned int dst_pitch,
- const short *HFilter,
- const short *VFilter,
- int Width,
- int Height
-)
-{
-
- unsigned short FData[17*16]; /* Temp data buffer used in filtering */
-
- /* First filter 1-D horizontally... */
- filter_block2d_bil_first_pass(src_ptr, FData, src_pitch, Height + 1, Width, HFilter);
-
- /* then 1-D vertically... */
- filter_block2d_bil_second_pass(FData, dst_ptr, dst_pitch, Height, Width, VFilter);
-}
-
-
-void vp8_bilinear_predict4x4_c
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-#if 0
- {
- int i;
- unsigned char temp1[16];
- unsigned char temp2[16];
-
- bilinear_predict4x4_mmx(src_ptr, src_pixels_per_line, xoffset, yoffset, temp1, 4);
- filter_block2d_bil(src_ptr, temp2, src_pixels_per_line, 4, HFilter, VFilter, 4, 4);
-
- for (i = 0; i < 16; i++)
- {
- if (temp1[i] != temp2[i])
- {
- bilinear_predict4x4_mmx(src_ptr, src_pixels_per_line, xoffset, yoffset, temp1, 4);
- filter_block2d_bil(src_ptr, temp2, src_pixels_per_line, 4, HFilter, VFilter, 4, 4);
- }
- }
- }
-#endif
- filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 4, 4);
-
-}
-
-void vp8_bilinear_predict8x8_c
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 8);
-
-}
-
-void vp8_bilinear_predict8x4_c
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 4);
-
-}
-
-void vp8_bilinear_predict16x16_c
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- const short *HFilter;
- const short *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 16, 16);
-}
diff --git a/vp8/common/filter.h b/vp8/common/filter.h
deleted file mode 100644
index 0f225c2..0000000
--- a/vp8/common/filter.h
+++ /dev/null
@@ -1,22 +0,0 @@
-/*
- * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef FILTER_H
-#define FILTER_H
-
-#define BLOCK_HEIGHT_WIDTH 4
-#define VP8_FILTER_WEIGHT 128
-#define VP8_FILTER_SHIFT 7
-
-extern const short vp8_bilinear_filters[8][2];
-extern const short vp8_sub_pel_filters[8][6];
-
-#endif //FILTER_H
diff --git a/vp8/common/findnearmv.c b/vp8/common/findnearmv.c
deleted file mode 100644
index 2041afa..0000000
--- a/vp8/common/findnearmv.c
+++ /dev/null
@@ -1,177 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "findnearmv.h"
-
-const unsigned char vp8_mbsplit_offset[4][16] = {
- { 0, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
- { 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
- { 0, 2, 8, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
- { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}
-};
-
-/* Predict motion vectors using those from already-decoded nearby blocks.
- Note that we only consider one 4x4 subblock from each candidate 16x16
- macroblock. */
-void vp8_find_near_mvs
-(
- MACROBLOCKD *xd,
- const MODE_INFO *here,
- MV *nearest,
- MV *nearby,
- MV *best_mv,
- int cnt[4],
- int refframe,
- int *ref_frame_sign_bias
-)
-{
- const MODE_INFO *above = here - xd->mode_info_stride;
- const MODE_INFO *left = here - 1;
- const MODE_INFO *aboveleft = above - 1;
- int_mv near_mvs[4];
- int_mv *mv = near_mvs;
- int *cntx = cnt;
- enum {CNT_INTRA, CNT_NEAREST, CNT_NEAR, CNT_SPLITMV};
-
- /* Zero accumulators */
- mv[0].as_int = mv[1].as_int = mv[2].as_int = 0;
- cnt[0] = cnt[1] = cnt[2] = cnt[3] = 0;
-
- /* Process above */
- if (above->mbmi.ref_frame != INTRA_FRAME)
- {
- if (above->mbmi.mv.as_int)
- {
- (++mv)->as_int = above->mbmi.mv.as_int;
- mv_bias(ref_frame_sign_bias[above->mbmi.ref_frame], refframe, mv, ref_frame_sign_bias);
- ++cntx;
- }
-
- *cntx += 2;
- }
-
- /* Process left */
- if (left->mbmi.ref_frame != INTRA_FRAME)
- {
- if (left->mbmi.mv.as_int)
- {
- int_mv this_mv;
-
- this_mv.as_int = left->mbmi.mv.as_int;
- mv_bias(ref_frame_sign_bias[left->mbmi.ref_frame], refframe, &this_mv, ref_frame_sign_bias);
-
- if (this_mv.as_int != mv->as_int)
- {
- (++mv)->as_int = this_mv.as_int;
- ++cntx;
- }
-
- *cntx += 2;
- }
- else
- cnt[CNT_INTRA] += 2;
- }
-
- /* Process above left */
- if (aboveleft->mbmi.ref_frame != INTRA_FRAME)
- {
- if (aboveleft->mbmi.mv.as_int)
- {
- int_mv this_mv;
-
- this_mv.as_int = aboveleft->mbmi.mv.as_int;
- mv_bias(ref_frame_sign_bias[aboveleft->mbmi.ref_frame], refframe, &this_mv, ref_frame_sign_bias);
-
- if (this_mv.as_int != mv->as_int)
- {
- (++mv)->as_int = this_mv.as_int;
- ++cntx;
- }
-
- *cntx += 1;
- }
- else
- cnt[CNT_INTRA] += 1;
- }
-
- /* If we have three distinct MV's ... */
- if (cnt[CNT_SPLITMV])
- {
- /* See if above-left MV can be merged with NEAREST */
- if (mv->as_int == near_mvs[CNT_NEAREST].as_int)
- cnt[CNT_NEAREST] += 1;
- }
-
- cnt[CNT_SPLITMV] = ((above->mbmi.mode == SPLITMV)
- + (left->mbmi.mode == SPLITMV)) * 2
- + (aboveleft->mbmi.mode == SPLITMV);
-
- /* Swap near and nearest if necessary */
- if (cnt[CNT_NEAR] > cnt[CNT_NEAREST])
- {
- int tmp;
- tmp = cnt[CNT_NEAREST];
- cnt[CNT_NEAREST] = cnt[CNT_NEAR];
- cnt[CNT_NEAR] = tmp;
- tmp = near_mvs[CNT_NEAREST].as_int;
- near_mvs[CNT_NEAREST].as_int = near_mvs[CNT_NEAR].as_int;
- near_mvs[CNT_NEAR].as_int = tmp;
- }
-
- /* Use near_mvs[0] to store the "best" MV */
- if (cnt[CNT_NEAREST] >= cnt[CNT_INTRA])
- near_mvs[CNT_INTRA] = near_mvs[CNT_NEAREST];
-
- /* Set up return values */
- *best_mv = near_mvs[0].as_mv;
- *nearest = near_mvs[CNT_NEAREST].as_mv;
- *nearby = near_mvs[CNT_NEAR].as_mv;
-
- vp8_clamp_mv(nearest, xd);
- vp8_clamp_mv(nearby, xd);
- vp8_clamp_mv(best_mv, xd); /*TODO: move this up before the copy*/
-}
-
-vp8_prob *vp8_mv_ref_probs(
- vp8_prob p[VP8_MVREFS-1], const int near_mv_ref_ct[4]
-)
-{
- p[0] = vp8_mode_contexts [near_mv_ref_ct[0]] [0];
- p[1] = vp8_mode_contexts [near_mv_ref_ct[1]] [1];
- p[2] = vp8_mode_contexts [near_mv_ref_ct[2]] [2];
- p[3] = vp8_mode_contexts [near_mv_ref_ct[3]] [3];
- /*p[3] = vp8_mode_contexts [near_mv_ref_ct[1] + near_mv_ref_ct[2] + near_mv_ref_ct[3]] [3];*/
- return p;
-}
-
-const B_MODE_INFO *vp8_left_bmi(const MODE_INFO *cur_mb, int b)
-{
- if (!(b & 3))
- {
- /* On L edge, get from MB to left of us */
- --cur_mb;
- b += 4;
- }
-
- return cur_mb->bmi + b - 1;
-}
-
-const B_MODE_INFO *vp8_above_bmi(const MODE_INFO *cur_mb, int b, int mi_stride)
-{
- if (!(b >> 2))
- {
- /* On top edge, get from MB above us */
- cur_mb -= mi_stride;
- b += 16;
- }
-
- return cur_mb->bmi + b - 4;
-}
diff --git a/vp8/common/findnearmv.h b/vp8/common/findnearmv.h
deleted file mode 100644
index bdf9eca..0000000
--- a/vp8/common/findnearmv.h
+++ /dev/null
@@ -1,75 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_FINDNEARMV_H
-#define __INC_FINDNEARMV_H
-
-#include "mv.h"
-#include "blockd.h"
-#include "modecont.h"
-#include "treecoder.h"
-
-typedef union
-{
- unsigned int as_int;
- MV as_mv;
-} int_mv; /* facilitates rapid equality tests */
-
-static void mv_bias(int refmb_ref_frame_sign_bias, int refframe, int_mv *mvp, const int *ref_frame_sign_bias)
-{
- MV xmv;
- xmv = mvp->as_mv;
-
- if (refmb_ref_frame_sign_bias != ref_frame_sign_bias[refframe])
- {
- xmv.row *= -1;
- xmv.col *= -1;
- }
-
- mvp->as_mv = xmv;
-}
-
-#define LEFT_TOP_MARGIN (16 << 3)
-#define RIGHT_BOTTOM_MARGIN (16 << 3)
-static void vp8_clamp_mv(MV *mv, const MACROBLOCKD *xd)
-{
- if (mv->col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN))
- mv->col = xd->mb_to_left_edge - LEFT_TOP_MARGIN;
- else if (mv->col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN)
- mv->col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN;
-
- if (mv->row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN))
- mv->row = xd->mb_to_top_edge - LEFT_TOP_MARGIN;
- else if (mv->row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN)
- mv->row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN;
-}
-
-void vp8_find_near_mvs
-(
- MACROBLOCKD *xd,
- const MODE_INFO *here,
- MV *nearest, MV *nearby, MV *best,
- int near_mv_ref_cts[4],
- int refframe,
- int *ref_frame_sign_bias
-);
-
-vp8_prob *vp8_mv_ref_probs(
- vp8_prob p[VP8_MVREFS-1], const int near_mv_ref_ct[4]
-);
-
-const B_MODE_INFO *vp8_left_bmi(const MODE_INFO *cur_mb, int b);
-
-const B_MODE_INFO *vp8_above_bmi(const MODE_INFO *cur_mb, int b, int mi_stride);
-
-extern const unsigned char vp8_mbsplit_offset[4][16];
-
-#endif
diff --git a/vp8/common/g_common.h b/vp8/common/g_common.h
deleted file mode 100644
index 5f52398..0000000
--- a/vp8/common/g_common.h
+++ /dev/null
@@ -1,21 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-extern void (*vp8_clear_system_state)(void);
-extern void (*vp8_plane_add_noise)(unsigned char *Start, unsigned int Width, unsigned int Height, int Pitch, int DPitch, int q);
-extern void (*de_interlace)
-(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- int Width,
- int Height,
- int Stride
-);
diff --git a/vp8/common/generic/systemdependent.c b/vp8/common/generic/systemdependent.c
deleted file mode 100644
index 5c64647..0000000
--- a/vp8/common/generic/systemdependent.c
+++ /dev/null
@@ -1,85 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/common/g_common.h"
-#include "vp8/common/subpixel.h"
-#include "vp8/common/loopfilter.h"
-#include "vp8/common/recon.h"
-#include "vp8/common/idct.h"
-#include "vp8/common/onyxc_int.h"
-
-extern void vp8_arch_x86_common_init(VP8_COMMON *ctx);
-extern void vp8_arch_arm_common_init(VP8_COMMON *ctx);
-
-void vp8_machine_specific_config(VP8_COMMON *ctx)
-{
-#if CONFIG_RUNTIME_CPU_DETECT
- VP8_COMMON_RTCD *rtcd = &ctx->rtcd;
-
- rtcd->idct.idct1 = vp8_short_idct4x4llm_1_c;
- rtcd->idct.idct16 = vp8_short_idct4x4llm_c;
- rtcd->idct.idct1_scalar_add = vp8_dc_only_idct_add_c;
- rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_c;
- rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_c;
-
- rtcd->recon.copy16x16 = vp8_copy_mem16x16_c;
- rtcd->recon.copy8x8 = vp8_copy_mem8x8_c;
- rtcd->recon.copy8x4 = vp8_copy_mem8x4_c;
- rtcd->recon.recon = vp8_recon_b_c;
- rtcd->recon.recon2 = vp8_recon2b_c;
- rtcd->recon.recon4 = vp8_recon4b_c;
- rtcd->recon.recon_mb = vp8_recon_mb_c;
- rtcd->recon.recon_mby = vp8_recon_mby_c;
- rtcd->recon.build_intra_predictors_mby =
- vp8_build_intra_predictors_mby;
- rtcd->recon.build_intra_predictors_mby_s =
- vp8_build_intra_predictors_mby_s;
-
- rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_c;
- rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_c;
- rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_c;
- rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_c;
- rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_c;
- rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_c;
- rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_c;
- rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_c;
-
- rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_c;
- rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_c;
- rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_c;
- rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_c;
- rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_c;
- rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_c;
- rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_c;
- rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_c;
-
-#if CONFIG_POSTPROC || (CONFIG_VP8_ENCODER && CONFIG_PSNR)
- rtcd->postproc.down = vp8_mbpost_proc_down_c;
- rtcd->postproc.across = vp8_mbpost_proc_across_ip_c;
- rtcd->postproc.downacross = vp8_post_proc_down_and_across_c;
- rtcd->postproc.addnoise = vp8_plane_add_noise_c;
- rtcd->postproc.blend_mb_inner = vp8_blend_mb_inner_c;
- rtcd->postproc.blend_mb_outer = vp8_blend_mb_outer_c;
- rtcd->postproc.blend_b = vp8_blend_b_c;
-#endif
-
-#endif
-
-#if ARCH_X86 || ARCH_X86_64
- vp8_arch_x86_common_init(ctx);
-#endif
-
-#if ARCH_ARM
- vp8_arch_arm_common_init(ctx);
-#endif
-
-}
diff --git a/vp8/common/idct.h b/vp8/common/idct.h
deleted file mode 100644
index 6b65501..0000000
--- a/vp8/common/idct.h
+++ /dev/null
@@ -1,86 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_IDCT_H
-#define __INC_IDCT_H
-
-#define prototype_second_order(sym) \
- void sym(short *input, short *output)
-
-#define prototype_idct(sym) \
- void sym(short *input, short *output, int pitch)
-
-#define prototype_idct_scalar_add(sym) \
- void sym(short input, \
- unsigned char *pred, unsigned char *output, \
- int pitch, int stride)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/idct_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/idct_arm.h"
-#endif
-
-#if ARCH_MIPS
-#if defined(MIPS_DSP_REV) && MIPS_DSP_REV>=1
-#include "mips/idct_mips.h"
-#endif
-#endif
-
-#ifndef vp8_idct_idct1
-#define vp8_idct_idct1 vp8_short_idct4x4llm_1_c
-#endif
-extern prototype_idct(vp8_idct_idct1);
-
-#ifndef vp8_idct_idct16
-#define vp8_idct_idct16 vp8_short_idct4x4llm_c
-#endif
-extern prototype_idct(vp8_idct_idct16);
-
-#ifndef vp8_idct_idct1_scalar_add
-#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_c
-#endif
-extern prototype_idct_scalar_add(vp8_idct_idct1_scalar_add);
-
-
-#ifndef vp8_idct_iwalsh1
-#define vp8_idct_iwalsh1 vp8_short_inv_walsh4x4_1_c
-#endif
-extern prototype_second_order(vp8_idct_iwalsh1);
-
-#ifndef vp8_idct_iwalsh16
-#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_c
-#endif
-extern prototype_second_order(vp8_idct_iwalsh16);
-
-typedef prototype_idct((*vp8_idct_fn_t));
-typedef prototype_idct_scalar_add((*vp8_idct_scalar_add_fn_t));
-typedef prototype_second_order((*vp8_second_order_fn_t));
-
-typedef struct
-{
- vp8_idct_fn_t idct1;
- vp8_idct_fn_t idct16;
- vp8_idct_scalar_add_fn_t idct1_scalar_add;
-
- vp8_second_order_fn_t iwalsh1;
- vp8_second_order_fn_t iwalsh16;
-} vp8_idct_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IDCT_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define IDCT_INVOKE(ctx,fn) vp8_idct_##fn
-#endif
-
-#endif
diff --git a/vp8/common/idctllm.c b/vp8/common/idctllm.c
deleted file mode 100644
index 196062d..0000000
--- a/vp8/common/idctllm.c
+++ /dev/null
@@ -1,197 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/****************************************************************************
- * Notes:
- *
- * This implementation makes use of 16 bit fixed point verio of two multiply
- * constants:
- * 1. sqrt(2) * cos (pi/8)
- * 2. sqrt(2) * sin (pi/8)
- * Becuase the first constant is bigger than 1, to maintain the same 16 bit
- * fixed point precision as the second one, we use a trick of
- * x * a = x + x*(a-1)
- * so
- * x * sqrt(2) * cos (pi/8) = x + x * (sqrt(2) *cos(pi/8)-1).
- **************************************************************************/
-static const int cospi8sqrt2minus1 = 20091;
-static const int sinpi8sqrt2 = 35468;
-static const int rounding = 0;
-void vp8_short_idct4x4llm_c(short *input, short *output, int pitch)
-{
- int i;
- int a1, b1, c1, d1;
-
- short *ip = input;
- short *op = output;
- int temp1, temp2;
- int shortpitch = pitch >> 1;
-
- for (i = 0; i < 4; i++)
- {
- a1 = ip[0] + ip[8];
- b1 = ip[0] - ip[8];
-
- temp1 = (ip[4] * sinpi8sqrt2 + rounding) >> 16;
- temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1 + rounding) >> 16);
- c1 = temp1 - temp2;
-
- temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1 + rounding) >> 16);
- temp2 = (ip[12] * sinpi8sqrt2 + rounding) >> 16;
- d1 = temp1 + temp2;
-
- op[shortpitch*0] = a1 + d1;
- op[shortpitch*3] = a1 - d1;
-
- op[shortpitch*1] = b1 + c1;
- op[shortpitch*2] = b1 - c1;
-
- ip++;
- op++;
- }
-
- ip = output;
- op = output;
-
- for (i = 0; i < 4; i++)
- {
- a1 = ip[0] + ip[2];
- b1 = ip[0] - ip[2];
-
- temp1 = (ip[1] * sinpi8sqrt2 + rounding) >> 16;
- temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1 + rounding) >> 16);
- c1 = temp1 - temp2;
-
- temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1 + rounding) >> 16);
- temp2 = (ip[3] * sinpi8sqrt2 + rounding) >> 16;
- d1 = temp1 + temp2;
-
-
- op[0] = (a1 + d1 + 4) >> 3;
- op[3] = (a1 - d1 + 4) >> 3;
-
- op[1] = (b1 + c1 + 4) >> 3;
- op[2] = (b1 - c1 + 4) >> 3;
-
- ip += shortpitch;
- op += shortpitch;
- }
-}
-
-void vp8_short_idct4x4llm_1_c(short *input, short *output, int pitch)
-{
- int i;
- int a1;
- short *op = output;
- int shortpitch = pitch >> 1;
- a1 = ((input[0] + 4) >> 3);
-
- for (i = 0; i < 4; i++)
- {
- op[0] = a1;
- op[1] = a1;
- op[2] = a1;
- op[3] = a1;
- op += shortpitch;
- }
-}
-
-void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr, unsigned char *dst_ptr, int pitch, int stride)
-{
- int a1 = ((input_dc + 4) >> 3);
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- int a = a1 + pred_ptr[c] ;
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dst_ptr[c] = (unsigned char) a ;
- }
-
- dst_ptr += stride;
- pred_ptr += pitch;
- }
-
-}
-
-void vp8_short_inv_walsh4x4_c(short *input, short *output)
-{
- int i;
- int a1, b1, c1, d1;
- int a2, b2, c2, d2;
- short *ip = input;
- short *op = output;
-
- for (i = 0; i < 4; i++)
- {
- a1 = ip[0] + ip[12];
- b1 = ip[4] + ip[8];
- c1 = ip[4] - ip[8];
- d1 = ip[0] - ip[12];
-
- op[0] = a1 + b1;
- op[4] = c1 + d1;
- op[8] = a1 - b1;
- op[12] = d1 - c1;
- ip++;
- op++;
- }
-
- ip = output;
- op = output;
-
- for (i = 0; i < 4; i++)
- {
- a1 = ip[0] + ip[3];
- b1 = ip[1] + ip[2];
- c1 = ip[1] - ip[2];
- d1 = ip[0] - ip[3];
-
- a2 = a1 + b1;
- b2 = c1 + d1;
- c2 = a1 - b1;
- d2 = d1 - c1;
-
- op[0] = (a2 + 3) >> 3;
- op[1] = (b2 + 3) >> 3;
- op[2] = (c2 + 3) >> 3;
- op[3] = (d2 + 3) >> 3;
-
- ip += 4;
- op += 4;
- }
-}
-
-void vp8_short_inv_walsh4x4_1_c(short *input, short *output)
-{
- int i;
- int a1;
- short *op = output;
-
- a1 = ((input[0] + 3) >> 3);
-
- for (i = 0; i < 4; i++)
- {
- op[0] = a1;
- op[1] = a1;
- op[2] = a1;
- op[3] = a1;
- op += 4;
- }
-}
diff --git a/vp8/common/invtrans.c b/vp8/common/invtrans.c
deleted file mode 100644
index 81a3f2d..0000000
--- a/vp8/common/invtrans.c
+++ /dev/null
@@ -1,88 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "invtrans.h"
-
-
-
-static void recon_dcblock(MACROBLOCKD *x)
-{
- BLOCKD *b = &x->block[24];
- int i;
-
- for (i = 0; i < 16; i++)
- {
- x->block[i].dqcoeff[0] = b->diff[i];
- }
-
-}
-
-void vp8_inverse_transform_b(const vp8_idct_rtcd_vtable_t *rtcd, BLOCKD *b, int pitch)
-{
- if (b->eob > 1)
- IDCT_INVOKE(rtcd, idct16)(b->dqcoeff, b->diff, pitch);
- else
- IDCT_INVOKE(rtcd, idct1)(b->dqcoeff, b->diff, pitch);
-}
-
-
-void vp8_inverse_transform_mby(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
- int i;
-
- /* do 2nd order transform on the dc block */
- IDCT_INVOKE(rtcd, iwalsh16)(x->block[24].dqcoeff, x->block[24].diff);
-
- recon_dcblock(x);
-
- for (i = 0; i < 16; i++)
- {
- vp8_inverse_transform_b(rtcd, &x->block[i], 32);
- }
-
-}
-void vp8_inverse_transform_mbuv(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
- int i;
-
- for (i = 16; i < 24; i++)
- {
- vp8_inverse_transform_b(rtcd, &x->block[i], 16);
- }
-
-}
-
-
-void vp8_inverse_transform_mb(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
- int i;
-
- if (x->mode_info_context->mbmi.mode != B_PRED &&
- x->mode_info_context->mbmi.mode != SPLITMV)
- {
- /* do 2nd order transform on the dc block */
-
- IDCT_INVOKE(rtcd, iwalsh16)(&x->block[24].dqcoeff[0], x->block[24].diff);
- recon_dcblock(x);
- }
-
- for (i = 0; i < 16; i++)
- {
- vp8_inverse_transform_b(rtcd, &x->block[i], 32);
- }
-
-
- for (i = 16; i < 24; i++)
- {
- vp8_inverse_transform_b(rtcd, &x->block[i], 16);
- }
-
-}
diff --git a/vp8/common/invtrans.h b/vp8/common/invtrans.h
deleted file mode 100644
index b3ffb70..0000000
--- a/vp8/common/invtrans.h
+++ /dev/null
@@ -1,23 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_INVTRANS_H
-#define __INC_INVTRANS_H
-
-#include "vpx_ports/config.h"
-#include "idct.h"
-#include "blockd.h"
-extern void vp8_inverse_transform_b(const vp8_idct_rtcd_vtable_t *rtcd, BLOCKD *b, int pitch);
-extern void vp8_inverse_transform_mb(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
-extern void vp8_inverse_transform_mby(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
-extern void vp8_inverse_transform_mbuv(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
-
-#endif
diff --git a/vp8/common/littlend.h b/vp8/common/littlend.h
deleted file mode 100644
index 99df116..0000000
--- a/vp8/common/littlend.h
+++ /dev/null
@@ -1,33 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef _littlend_h
-#define _littlend_h
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-#define invert2(x) (x)
-#define invert4(x) (x)
-
-#define low_byte(x) (unsigned char)x
-#define mid1Byte(x) (unsigned char)(x >> 8)
-#define mid2Byte(x) (unsigned char)(x >> 16)
-#define high_byte(x) (unsigned char)(x >> 24)
-
-#define SWAPENDS 0
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif
diff --git a/vp8/common/loopfilter.c b/vp8/common/loopfilter.c
deleted file mode 100644
index 37c5b77..0000000
--- a/vp8/common/loopfilter.c
+++ /dev/null
@@ -1,591 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "loopfilter.h"
-#include "onyxc_int.h"
-
-typedef unsigned char uc;
-
-
-prototype_loopfilter(vp8_loop_filter_horizontal_edge_c);
-prototype_loopfilter(vp8_loop_filter_vertical_edge_c);
-prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_c);
-prototype_loopfilter(vp8_mbloop_filter_vertical_edge_c);
-prototype_loopfilter(vp8_loop_filter_simple_horizontal_edge_c);
-prototype_loopfilter(vp8_loop_filter_simple_vertical_edge_c);
-
-/* Horizontal MB filtering */
-void vp8_loop_filter_mbh_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_c(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_mbloop_filter_horizontal_edge_c(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-}
-
-void vp8_loop_filter_mbhs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-/* Vertical MB Filtering */
-void vp8_loop_filter_mbv_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_vertical_edge_c(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_mbloop_filter_vertical_edge_c(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-}
-
-void vp8_loop_filter_mbvs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-/* Horizontal B Filtering */
-void vp8_loop_filter_bh_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_horizontal_edge_c(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_loop_filter_horizontal_edge_c(v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-}
-
-void vp8_loop_filter_bhs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-
-/* Vertical B Filtering */
-void vp8_loop_filter_bv_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_vertical_edge_c(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_c(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_c(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_vertical_edge_c(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_loop_filter_vertical_edge_c(v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-}
-
-void vp8_loop_filter_bvs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_c(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_c(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_c(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-
-void vp8_init_loop_filter(VP8_COMMON *cm)
-{
- loop_filter_info *lfi = cm->lf_info;
- LOOPFILTERTYPE lft = cm->filter_type;
- int sharpness_lvl = cm->sharpness_level;
- int frame_type = cm->frame_type;
- int i, j;
-
- int block_inside_limit = 0;
- int HEVThresh;
-
- /* For each possible value for the loop filter fill out a "loop_filter_info" entry. */
- for (i = 0; i <= MAX_LOOP_FILTER; i++)
- {
- int filt_lvl = i;
-
- if (frame_type == KEY_FRAME)
- {
- if (filt_lvl >= 40)
- HEVThresh = 2;
- else if (filt_lvl >= 15)
- HEVThresh = 1;
- else
- HEVThresh = 0;
- }
- else
- {
- if (filt_lvl >= 40)
- HEVThresh = 3;
- else if (filt_lvl >= 20)
- HEVThresh = 2;
- else if (filt_lvl >= 15)
- HEVThresh = 1;
- else
- HEVThresh = 0;
- }
-
- /* Set loop filter paramaeters that control sharpness. */
- block_inside_limit = filt_lvl >> (sharpness_lvl > 0);
- block_inside_limit = block_inside_limit >> (sharpness_lvl > 4);
-
- if (sharpness_lvl > 0)
- {
- if (block_inside_limit > (9 - sharpness_lvl))
- block_inside_limit = (9 - sharpness_lvl);
- }
-
- if (block_inside_limit < 1)
- block_inside_limit = 1;
-
- for (j = 0; j < 16; j++)
- {
- lfi[i].lim[j] = block_inside_limit;
- lfi[i].mbflim[j] = filt_lvl + 2;
- lfi[i].flim[j] = filt_lvl;
- lfi[i].thr[j] = HEVThresh;
- }
-
- }
-
- /* Set up the function pointers depending on the type of loop filtering selected */
- if (lft == NORMAL_LOOPFILTER)
- {
- cm->lf_mbv = LF_INVOKE(&cm->rtcd.loopfilter, normal_mb_v);
- cm->lf_bv = LF_INVOKE(&cm->rtcd.loopfilter, normal_b_v);
- cm->lf_mbh = LF_INVOKE(&cm->rtcd.loopfilter, normal_mb_h);
- cm->lf_bh = LF_INVOKE(&cm->rtcd.loopfilter, normal_b_h);
- }
- else
- {
- cm->lf_mbv = LF_INVOKE(&cm->rtcd.loopfilter, simple_mb_v);
- cm->lf_bv = LF_INVOKE(&cm->rtcd.loopfilter, simple_b_v);
- cm->lf_mbh = LF_INVOKE(&cm->rtcd.loopfilter, simple_mb_h);
- cm->lf_bh = LF_INVOKE(&cm->rtcd.loopfilter, simple_b_h);
- }
-}
-
-/* Put vp8_init_loop_filter() in vp8dx_create_decompressor(). Only call vp8_frame_init_loop_filter() while decoding
- * each frame. Check last_frame_type to skip the function most of times.
- */
-void vp8_frame_init_loop_filter(loop_filter_info *lfi, int frame_type)
-{
- int HEVThresh;
- int i, j;
-
- /* For each possible value for the loop filter fill out a "loop_filter_info" entry. */
- for (i = 0; i <= MAX_LOOP_FILTER; i++)
- {
- int filt_lvl = i;
-
- if (frame_type == KEY_FRAME)
- {
- if (filt_lvl >= 40)
- HEVThresh = 2;
- else if (filt_lvl >= 15)
- HEVThresh = 1;
- else
- HEVThresh = 0;
- }
- else
- {
- if (filt_lvl >= 40)
- HEVThresh = 3;
- else if (filt_lvl >= 20)
- HEVThresh = 2;
- else if (filt_lvl >= 15)
- HEVThresh = 1;
- else
- HEVThresh = 0;
- }
-
- for (j = 0; j < 16; j++)
- {
- /*lfi[i].lim[j] = block_inside_limit;
- lfi[i].mbflim[j] = filt_lvl+2;*/
- /*lfi[i].flim[j] = filt_lvl;*/
- lfi[i].thr[j] = HEVThresh;
- }
- }
-}
-
-
-int vp8_adjust_mb_lf_value(MACROBLOCKD *mbd, int filter_level)
-{
- MB_MODE_INFO *mbmi = &mbd->mode_info_context->mbmi;
-
- if (mbd->mode_ref_lf_delta_enabled)
- {
- /* Apply delta for reference frame */
- filter_level += mbd->ref_lf_deltas[mbmi->ref_frame];
-
- /* Apply delta for mode */
- if (mbmi->ref_frame == INTRA_FRAME)
- {
- /* Only the split mode BPRED has a further special case */
- if (mbmi->mode == B_PRED)
- filter_level += mbd->mode_lf_deltas[0];
- }
- else
- {
- /* Zero motion mode */
- if (mbmi->mode == ZEROMV)
- filter_level += mbd->mode_lf_deltas[1];
-
- /* Split MB motion mode */
- else if (mbmi->mode == SPLITMV)
- filter_level += mbd->mode_lf_deltas[3];
-
- /* All other inter motion modes (Nearest, Near, New) */
- else
- filter_level += mbd->mode_lf_deltas[2];
- }
-
- /* Range check */
- if (filter_level > MAX_LOOP_FILTER)
- filter_level = MAX_LOOP_FILTER;
- else if (filter_level < 0)
- filter_level = 0;
- }
- return filter_level;
-}
-
-
-void vp8_loop_filter_frame
-(
- VP8_COMMON *cm,
- MACROBLOCKD *mbd,
- int default_filt_lvl
-)
-{
- YV12_BUFFER_CONFIG *post = cm->frame_to_show;
- loop_filter_info *lfi = cm->lf_info;
- FRAME_TYPE frame_type = cm->frame_type;
-
- int mb_row;
- int mb_col;
-
-
- int baseline_filter_level[MAX_MB_SEGMENTS];
- int filter_level;
- int alt_flt_enabled = mbd->segmentation_enabled;
-
- int i;
- unsigned char *y_ptr, *u_ptr, *v_ptr;
-
- mbd->mode_info_context = cm->mi; /* Point at base of Mb MODE_INFO list */
-
- /* Note the baseline filter values for each segment */
- if (alt_flt_enabled)
- {
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- {
- /* Abs value */
- if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
- baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i];
- /* Delta Value */
- else
- {
- baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i];
- baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; /* Clamp to valid range */
- }
- }
- }
- else
- {
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- baseline_filter_level[i] = default_filt_lvl;
- }
-
- /* Initialize the loop filter for this frame. */
- if ((cm->last_filter_type != cm->filter_type) || (cm->last_sharpness_level != cm->sharpness_level))
- vp8_init_loop_filter(cm);
- else if (frame_type != cm->last_frame_type)
- vp8_frame_init_loop_filter(lfi, frame_type);
-
- /* Set up the buffer pointers */
- y_ptr = post->y_buffer;
- u_ptr = post->u_buffer;
- v_ptr = post->v_buffer;
-
- /* vp8_filter each macro block */
- for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
- {
- for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
- {
- int Segment = (alt_flt_enabled) ? mbd->mode_info_context->mbmi.segment_id : 0;
-
- filter_level = baseline_filter_level[Segment];
-
- /* Distance of Mb to the various image edges.
- * These specified to 8th pel as they are always compared to values that are in 1/8th pel units
- * Apply any context driven MB level adjustment
- */
- filter_level = vp8_adjust_mb_lf_value(mbd, filter_level);
-
- if (filter_level)
- {
- if (mb_col > 0)
- cm->lf_mbv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
-
- if (mbd->mode_info_context->mbmi.dc_diff > 0)
- cm->lf_bv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
-
- /* don't apply across umv border */
- if (mb_row > 0)
- cm->lf_mbh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
-
- if (mbd->mode_info_context->mbmi.dc_diff > 0)
- cm->lf_bh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
- }
-
- y_ptr += 16;
- u_ptr += 8;
- v_ptr += 8;
-
- mbd->mode_info_context++; /* step to next MB */
- }
-
- y_ptr += post->y_stride * 16 - post->y_width;
- u_ptr += post->uv_stride * 8 - post->uv_width;
- v_ptr += post->uv_stride * 8 - post->uv_width;
-
- mbd->mode_info_context++; /* Skip border mb */
- }
-}
-
-
-void vp8_loop_filter_frame_yonly
-(
- VP8_COMMON *cm,
- MACROBLOCKD *mbd,
- int default_filt_lvl,
- int sharpness_lvl
-)
-{
- YV12_BUFFER_CONFIG *post = cm->frame_to_show;
-
- int i;
- unsigned char *y_ptr;
- int mb_row;
- int mb_col;
-
- loop_filter_info *lfi = cm->lf_info;
- int baseline_filter_level[MAX_MB_SEGMENTS];
- int filter_level;
- int alt_flt_enabled = mbd->segmentation_enabled;
- FRAME_TYPE frame_type = cm->frame_type;
-
- (void) sharpness_lvl;
-
- /*MODE_INFO * this_mb_mode_info = cm->mi;*/ /* Point at base of Mb MODE_INFO list */
- mbd->mode_info_context = cm->mi; /* Point at base of Mb MODE_INFO list */
-
- /* Note the baseline filter values for each segment */
- if (alt_flt_enabled)
- {
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- {
- /* Abs value */
- if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
- baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i];
- /* Delta Value */
- else
- {
- baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i];
- baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; /* Clamp to valid range */
- }
- }
- }
- else
- {
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- baseline_filter_level[i] = default_filt_lvl;
- }
-
- /* Initialize the loop filter for this frame. */
- if ((cm->last_filter_type != cm->filter_type) || (cm->last_sharpness_level != cm->sharpness_level))
- vp8_init_loop_filter(cm);
- else if (frame_type != cm->last_frame_type)
- vp8_frame_init_loop_filter(lfi, frame_type);
-
- /* Set up the buffer pointers */
- y_ptr = post->y_buffer;
-
- /* vp8_filter each macro block */
- for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
- {
- for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
- {
- int Segment = (alt_flt_enabled) ? mbd->mode_info_context->mbmi.segment_id : 0;
- filter_level = baseline_filter_level[Segment];
-
- /* Apply any context driven MB level adjustment */
- filter_level = vp8_adjust_mb_lf_value(mbd, filter_level);
-
- if (filter_level)
- {
- if (mb_col > 0)
- cm->lf_mbv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
-
- if (mbd->mode_info_context->mbmi.dc_diff > 0)
- cm->lf_bv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
-
- /* don't apply across umv border */
- if (mb_row > 0)
- cm->lf_mbh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
-
- if (mbd->mode_info_context->mbmi.dc_diff > 0)
- cm->lf_bh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
- }
-
- y_ptr += 16;
- mbd->mode_info_context ++; /* step to next MB */
-
- }
-
- y_ptr += post->y_stride * 16 - post->y_width;
- mbd->mode_info_context ++; /* Skip border mb */
- }
-
-}
-
-
-void vp8_loop_filter_partial_frame
-(
- VP8_COMMON *cm,
- MACROBLOCKD *mbd,
- int default_filt_lvl,
- int sharpness_lvl,
- int Fraction
-)
-{
- YV12_BUFFER_CONFIG *post = cm->frame_to_show;
-
- int i;
- unsigned char *y_ptr;
- int mb_row;
- int mb_col;
- /*int mb_rows = post->y_height >> 4;*/
- int mb_cols = post->y_width >> 4;
-
- int linestocopy;
-
- loop_filter_info *lfi = cm->lf_info;
- int baseline_filter_level[MAX_MB_SEGMENTS];
- int filter_level;
- int alt_flt_enabled = mbd->segmentation_enabled;
- FRAME_TYPE frame_type = cm->frame_type;
-
- (void) sharpness_lvl;
-
- /*MODE_INFO * this_mb_mode_info = cm->mi + (post->y_height>>5) * (mb_cols + 1);*/ /* Point at base of Mb MODE_INFO list */
- mbd->mode_info_context = cm->mi + (post->y_height >> 5) * (mb_cols + 1); /* Point at base of Mb MODE_INFO list */
-
- linestocopy = (post->y_height >> (4 + Fraction));
-
- if (linestocopy < 1)
- linestocopy = 1;
-
- linestocopy <<= 4;
-
- /* Note the baseline filter values for each segment */
- if (alt_flt_enabled)
- {
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- {
- /* Abs value */
- if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
- baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i];
- /* Delta Value */
- else
- {
- baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i];
- baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; /* Clamp to valid range */
- }
- }
- }
- else
- {
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- baseline_filter_level[i] = default_filt_lvl;
- }
-
- /* Initialize the loop filter for this frame. */
- if ((cm->last_filter_type != cm->filter_type) || (cm->last_sharpness_level != cm->sharpness_level))
- vp8_init_loop_filter(cm);
- else if (frame_type != cm->last_frame_type)
- vp8_frame_init_loop_filter(lfi, frame_type);
-
- /* Set up the buffer pointers */
- y_ptr = post->y_buffer + (post->y_height >> 5) * 16 * post->y_stride;
-
- /* vp8_filter each macro block */
- for (mb_row = 0; mb_row<(linestocopy >> 4); mb_row++)
- {
- for (mb_col = 0; mb_col < mb_cols; mb_col++)
- {
- int Segment = (alt_flt_enabled) ? mbd->mode_info_context->mbmi.segment_id : 0;
- filter_level = baseline_filter_level[Segment];
-
- if (filter_level)
- {
- if (mb_col > 0)
- cm->lf_mbv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
-
- if (mbd->mode_info_context->mbmi.dc_diff > 0)
- cm->lf_bv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
-
- cm->lf_mbh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
-
- if (mbd->mode_info_context->mbmi.dc_diff > 0)
- cm->lf_bh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
- }
-
- y_ptr += 16;
- mbd->mode_info_context += 1; /* step to next MB */
- }
-
- y_ptr += post->y_stride * 16 - post->y_width;
- mbd->mode_info_context += 1; /* Skip border mb */
- }
-}
diff --git a/vp8/common/loopfilter.h b/vp8/common/loopfilter.h
deleted file mode 100644
index f99c425..0000000
--- a/vp8/common/loopfilter.h
+++ /dev/null
@@ -1,130 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef loopfilter_h
-#define loopfilter_h
-
-#include "vpx_ports/mem.h"
-
-#define MAX_LOOP_FILTER 63
-
-typedef enum
-{
- NORMAL_LOOPFILTER = 0,
- SIMPLE_LOOPFILTER = 1
-} LOOPFILTERTYPE;
-
-/* FRK
- * Need to align this structure so when it is declared and
- * passed it can be loaded into vector registers.
- */
-typedef struct
-{
- DECLARE_ALIGNED(16, signed char, lim[16]);
- DECLARE_ALIGNED(16, signed char, flim[16]);
- DECLARE_ALIGNED(16, signed char, thr[16]);
- DECLARE_ALIGNED(16, signed char, mbflim[16]);
-} loop_filter_info;
-
-
-#define prototype_loopfilter(sym) \
- void sym(unsigned char *src, int pitch, const signed char *flimit,\
- const signed char *limit, const signed char *thresh, int count)
-
-#define prototype_loopfilter_block(sym) \
- void sym(unsigned char *y, unsigned char *u, unsigned char *v,\
- int ystride, int uv_stride, loop_filter_info *lfi, int simpler)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/loopfilter_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/loopfilter_arm.h"
-#endif
-
-#if ARCH_MIPS
-#if defined(MIPS_DSP_REV) && MIPS_DSP_REV>=2
-#include "mips/loopfilter_mips.h"
-#endif
-#endif
-
-#ifndef vp8_lf_normal_mb_v
-#define vp8_lf_normal_mb_v vp8_loop_filter_mbv_c
-#endif
-extern prototype_loopfilter_block(vp8_lf_normal_mb_v);
-
-#ifndef vp8_lf_normal_b_v
-#define vp8_lf_normal_b_v vp8_loop_filter_bv_c
-#endif
-extern prototype_loopfilter_block(vp8_lf_normal_b_v);
-
-#ifndef vp8_lf_normal_mb_h
-#define vp8_lf_normal_mb_h vp8_loop_filter_mbh_c
-#endif
-extern prototype_loopfilter_block(vp8_lf_normal_mb_h);
-
-#ifndef vp8_lf_normal_b_h
-#define vp8_lf_normal_b_h vp8_loop_filter_bh_c
-#endif
-extern prototype_loopfilter_block(vp8_lf_normal_b_h);
-
-
-#ifndef vp8_lf_simple_mb_v
-#define vp8_lf_simple_mb_v vp8_loop_filter_mbvs_c
-#endif
-extern prototype_loopfilter_block(vp8_lf_simple_mb_v);
-
-#ifndef vp8_lf_simple_b_v
-#define vp8_lf_simple_b_v vp8_loop_filter_bvs_c
-#endif
-extern prototype_loopfilter_block(vp8_lf_simple_b_v);
-
-#ifndef vp8_lf_simple_mb_h
-#define vp8_lf_simple_mb_h vp8_loop_filter_mbhs_c
-#endif
-extern prototype_loopfilter_block(vp8_lf_simple_mb_h);
-
-#ifndef vp8_lf_simple_b_h
-#define vp8_lf_simple_b_h vp8_loop_filter_bhs_c
-#endif
-extern prototype_loopfilter_block(vp8_lf_simple_b_h);
-
-typedef prototype_loopfilter_block((*vp8_lf_block_fn_t));
-typedef struct
-{
- vp8_lf_block_fn_t normal_mb_v;
- vp8_lf_block_fn_t normal_b_v;
- vp8_lf_block_fn_t normal_mb_h;
- vp8_lf_block_fn_t normal_b_h;
- vp8_lf_block_fn_t simple_mb_v;
- vp8_lf_block_fn_t simple_b_v;
- vp8_lf_block_fn_t simple_mb_h;
- vp8_lf_block_fn_t simple_b_h;
-} vp8_loopfilter_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define LF_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define LF_INVOKE(ctx,fn) vp8_lf_##fn
-#endif
-
-typedef void loop_filter_uvfunction
-(
- unsigned char *u, /* source pointer */
- int p, /* pitch */
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh,
- unsigned char *v
-);
-
-#endif
diff --git a/vp8/common/loopfilter_filters.c b/vp8/common/loopfilter_filters.c
deleted file mode 100644
index 6940529..0000000
--- a/vp8/common/loopfilter_filters.c
+++ /dev/null
@@ -1,363 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdlib.h>
-#include "loopfilter.h"
-#include "onyxc_int.h"
-
-typedef unsigned char uc;
-
-static __inline signed char vp8_signed_char_clamp(int t)
-{
- t = (t < -128 ? -128 : t);
- t = (t > 127 ? 127 : t);
- return (signed char) t;
-}
-
-
-/* should we apply any filter at all ( 11111111 yes, 00000000 no) */
-static __inline signed char vp8_filter_mask(signed char limit, signed char flimit,
- uc p3, uc p2, uc p1, uc p0, uc q0, uc q1, uc q2, uc q3)
-{
- signed char mask = 0;
- mask |= (abs(p3 - p2) > limit) * -1;
- mask |= (abs(p2 - p1) > limit) * -1;
- mask |= (abs(p1 - p0) > limit) * -1;
- mask |= (abs(q1 - q0) > limit) * -1;
- mask |= (abs(q2 - q1) > limit) * -1;
- mask |= (abs(q3 - q2) > limit) * -1;
- mask |= (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 > flimit * 2 + limit) * -1;
- mask = ~mask;
- return mask;
-}
-
-/* is there high variance internal edge ( 11111111 yes, 00000000 no) */
-static __inline signed char vp8_hevmask(signed char thresh, uc p1, uc p0, uc q0, uc q1)
-{
- signed char hev = 0;
- hev |= (abs(p1 - p0) > thresh) * -1;
- hev |= (abs(q1 - q0) > thresh) * -1;
- return hev;
-}
-
-static __inline void vp8_filter(signed char mask, signed char hev, uc *op1, uc *op0, uc *oq0, uc *oq1)
-
-{
- signed char ps0, qs0;
- signed char ps1, qs1;
- signed char vp8_filter, Filter1, Filter2;
- signed char u;
-
- ps1 = (signed char) * op1 ^ 0x80;
- ps0 = (signed char) * op0 ^ 0x80;
- qs0 = (signed char) * oq0 ^ 0x80;
- qs1 = (signed char) * oq1 ^ 0x80;
-
- /* add outer taps if we have high edge variance */
- vp8_filter = vp8_signed_char_clamp(ps1 - qs1);
- vp8_filter &= hev;
-
- /* inner taps */
- vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0));
- vp8_filter &= mask;
-
- /* save bottom 3 bits so that we round one side +4 and the other +3
- * if it equals 4 we'll set to adjust by -1 to account for the fact
- * we'd round 3 the other way
- */
- Filter1 = vp8_signed_char_clamp(vp8_filter + 4);
- Filter2 = vp8_signed_char_clamp(vp8_filter + 3);
- Filter1 >>= 3;
- Filter2 >>= 3;
- u = vp8_signed_char_clamp(qs0 - Filter1);
- *oq0 = u ^ 0x80;
- u = vp8_signed_char_clamp(ps0 + Filter2);
- *op0 = u ^ 0x80;
- vp8_filter = Filter1;
-
- /* outer tap adjustments */
- vp8_filter += 1;
- vp8_filter >>= 1;
- vp8_filter &= ~hev;
-
- u = vp8_signed_char_clamp(qs1 - vp8_filter);
- *oq1 = u ^ 0x80;
- u = vp8_signed_char_clamp(ps1 + vp8_filter);
- *op1 = u ^ 0x80;
-
-}
-void vp8_loop_filter_horizontal_edge_c
-(
- unsigned char *s,
- int p, /* pitch */
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh,
- int count
-)
-{
- int hev = 0; /* high edge variance */
- signed char mask = 0;
- int i = 0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
- do
- {
- mask = vp8_filter_mask(limit[i], flimit[i],
- s[-4*p], s[-3*p], s[-2*p], s[-1*p],
- s[0*p], s[1*p], s[2*p], s[3*p]);
-
- hev = vp8_hevmask(thresh[i], s[-2*p], s[-1*p], s[0*p], s[1*p]);
-
- vp8_filter(mask, hev, s - 2 * p, s - 1 * p, s, s + 1 * p);
-
- ++s;
- }
- while (++i < count * 8);
-}
-
-void vp8_loop_filter_vertical_edge_c
-(
- unsigned char *s,
- int p,
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh,
- int count
-)
-{
- int hev = 0; /* high edge variance */
- signed char mask = 0;
- int i = 0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
- do
- {
- mask = vp8_filter_mask(limit[i], flimit[i],
- s[-4], s[-3], s[-2], s[-1], s[0], s[1], s[2], s[3]);
-
- hev = vp8_hevmask(thresh[i], s[-2], s[-1], s[0], s[1]);
-
- vp8_filter(mask, hev, s - 2, s - 1, s, s + 1);
-
- s += p;
- }
- while (++i < count * 8);
-}
-
-static __inline void vp8_mbfilter(signed char mask, signed char hev,
- uc *op2, uc *op1, uc *op0, uc *oq0, uc *oq1, uc *oq2)
-{
- signed char s, u;
- signed char vp8_filter, Filter1, Filter2;
- signed char ps2 = (signed char) * op2 ^ 0x80;
- signed char ps1 = (signed char) * op1 ^ 0x80;
- signed char ps0 = (signed char) * op0 ^ 0x80;
- signed char qs0 = (signed char) * oq0 ^ 0x80;
- signed char qs1 = (signed char) * oq1 ^ 0x80;
- signed char qs2 = (signed char) * oq2 ^ 0x80;
-
- /* add outer taps if we have high edge variance */
- vp8_filter = vp8_signed_char_clamp(ps1 - qs1);
- vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0));
- vp8_filter &= mask;
-
- Filter2 = vp8_filter;
- Filter2 &= hev;
-
- /* save bottom 3 bits so that we round one side +4 and the other +3 */
- Filter1 = vp8_signed_char_clamp(Filter2 + 4);
- Filter2 = vp8_signed_char_clamp(Filter2 + 3);
- Filter1 >>= 3;
- Filter2 >>= 3;
- qs0 = vp8_signed_char_clamp(qs0 - Filter1);
- ps0 = vp8_signed_char_clamp(ps0 + Filter2);
-
-
- /* only apply wider filter if not high edge variance */
- vp8_filter &= ~hev;
- Filter2 = vp8_filter;
-
- /* roughly 3/7th difference across boundary */
- u = vp8_signed_char_clamp((63 + Filter2 * 27) >> 7);
- s = vp8_signed_char_clamp(qs0 - u);
- *oq0 = s ^ 0x80;
- s = vp8_signed_char_clamp(ps0 + u);
- *op0 = s ^ 0x80;
-
- /* roughly 2/7th difference across boundary */
- u = vp8_signed_char_clamp((63 + Filter2 * 18) >> 7);
- s = vp8_signed_char_clamp(qs1 - u);
- *oq1 = s ^ 0x80;
- s = vp8_signed_char_clamp(ps1 + u);
- *op1 = s ^ 0x80;
-
- /* roughly 1/7th difference across boundary */
- u = vp8_signed_char_clamp((63 + Filter2 * 9) >> 7);
- s = vp8_signed_char_clamp(qs2 - u);
- *oq2 = s ^ 0x80;
- s = vp8_signed_char_clamp(ps2 + u);
- *op2 = s ^ 0x80;
-}
-
-void vp8_mbloop_filter_horizontal_edge_c
-(
- unsigned char *s,
- int p,
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh,
- int count
-)
-{
- signed char hev = 0; /* high edge variance */
- signed char mask = 0;
- int i = 0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
- do
- {
-
- mask = vp8_filter_mask(limit[i], flimit[i],
- s[-4*p], s[-3*p], s[-2*p], s[-1*p],
- s[0*p], s[1*p], s[2*p], s[3*p]);
-
- hev = vp8_hevmask(thresh[i], s[-2*p], s[-1*p], s[0*p], s[1*p]);
-
- vp8_mbfilter(mask, hev, s - 3 * p, s - 2 * p, s - 1 * p, s, s + 1 * p, s + 2 * p);
-
- ++s;
- }
- while (++i < count * 8);
-
-}
-
-
-void vp8_mbloop_filter_vertical_edge_c
-(
- unsigned char *s,
- int p,
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh,
- int count
-)
-{
- signed char hev = 0; /* high edge variance */
- signed char mask = 0;
- int i = 0;
-
- do
- {
-
- mask = vp8_filter_mask(limit[i], flimit[i],
- s[-4], s[-3], s[-2], s[-1], s[0], s[1], s[2], s[3]);
-
- hev = vp8_hevmask(thresh[i], s[-2], s[-1], s[0], s[1]);
-
- vp8_mbfilter(mask, hev, s - 3, s - 2, s - 1, s, s + 1, s + 2);
-
- s += p;
- }
- while (++i < count * 8);
-
-}
-
-/* should we apply any filter at all ( 11111111 yes, 00000000 no) */
-static __inline signed char vp8_simple_filter_mask(signed char limit, signed char flimit, uc p1, uc p0, uc q0, uc q1)
-{
-/* Why does this cause problems for win32?
- * error C2143: syntax error : missing ';' before 'type'
- * (void) limit;
- */
- signed char mask = (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 <= flimit * 2 + limit) * -1;
- return mask;
-}
-
-static __inline void vp8_simple_filter(signed char mask, uc *op1, uc *op0, uc *oq0, uc *oq1)
-{
- signed char vp8_filter, Filter1, Filter2;
- signed char p1 = (signed char) * op1 ^ 0x80;
- signed char p0 = (signed char) * op0 ^ 0x80;
- signed char q0 = (signed char) * oq0 ^ 0x80;
- signed char q1 = (signed char) * oq1 ^ 0x80;
- signed char u;
-
- vp8_filter = vp8_signed_char_clamp(p1 - q1);
- vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (q0 - p0));
- vp8_filter &= mask;
-
- /* save bottom 3 bits so that we round one side +4 and the other +3 */
- Filter1 = vp8_signed_char_clamp(vp8_filter + 4);
- Filter1 >>= 3;
- u = vp8_signed_char_clamp(q0 - Filter1);
- *oq0 = u ^ 0x80;
-
- Filter2 = vp8_signed_char_clamp(vp8_filter + 3);
- Filter2 >>= 3;
- u = vp8_signed_char_clamp(p0 + Filter2);
- *op0 = u ^ 0x80;
-}
-
-void vp8_loop_filter_simple_horizontal_edge_c
-(
- unsigned char *s,
- int p,
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh,
- int count
-)
-{
- signed char mask = 0;
- int i = 0;
- (void) thresh;
-
- do
- {
- /*mask = vp8_simple_filter_mask( limit[i], flimit[i],s[-1*p],s[0*p]);*/
- mask = vp8_simple_filter_mask(limit[i], flimit[i], s[-2*p], s[-1*p], s[0*p], s[1*p]);
- vp8_simple_filter(mask, s - 2 * p, s - 1 * p, s, s + 1 * p);
- ++s;
- }
- while (++i < count * 8);
-}
-
-void vp8_loop_filter_simple_vertical_edge_c
-(
- unsigned char *s,
- int p,
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh,
- int count
-)
-{
- signed char mask = 0;
- int i = 0;
- (void) thresh;
-
- do
- {
- /*mask = vp8_simple_filter_mask( limit[i], flimit[i],s[-1],s[0]);*/
- mask = vp8_simple_filter_mask(limit[i], flimit[i], s[-2], s[-1], s[0], s[1]);
- vp8_simple_filter(mask, s - 2, s - 1, s, s + 1);
- s += p;
- }
- while (++i < count * 8);
-
-}
diff --git a/vp8/common/mbpitch.c b/vp8/common/mbpitch.c
deleted file mode 100644
index 054042c..0000000
--- a/vp8/common/mbpitch.c
+++ /dev/null
@@ -1,130 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "blockd.h"
-
-typedef enum
-{
- PRED = 0,
- DEST = 1
-} BLOCKSET;
-
-static void setup_block
-(
- BLOCKD *b,
- int mv_stride,
- unsigned char **base,
- int Stride,
- int offset,
- BLOCKSET bs
-)
-{
-
- if (bs == DEST)
- {
- b->dst_stride = Stride;
- b->dst = offset;
- b->base_dst = base;
- }
- else
- {
- b->pre_stride = Stride;
- b->pre = offset;
- b->base_pre = base;
- }
-
-}
-
-
-static void setup_macroblock(MACROBLOCKD *x, BLOCKSET bs)
-{
- int block;
-
- unsigned char **y, **u, **v;
-
- if (bs == DEST)
- {
- y = &x->dst.y_buffer;
- u = &x->dst.u_buffer;
- v = &x->dst.v_buffer;
- }
- else
- {
- y = &x->pre.y_buffer;
- u = &x->pre.u_buffer;
- v = &x->pre.v_buffer;
- }
-
- for (block = 0; block < 16; block++) /* y blocks */
- {
- setup_block(&x->block[block], x->dst.y_stride, y, x->dst.y_stride,
- (block >> 2) * 4 * x->dst.y_stride + (block & 3) * 4, bs);
- }
-
- for (block = 16; block < 20; block++) /* U and V blocks */
- {
- setup_block(&x->block[block], x->dst.uv_stride, u, x->dst.uv_stride,
- ((block - 16) >> 1) * 4 * x->dst.uv_stride + (block & 1) * 4, bs);
-
- setup_block(&x->block[block+4], x->dst.uv_stride, v, x->dst.uv_stride,
- ((block - 16) >> 1) * 4 * x->dst.uv_stride + (block & 1) * 4, bs);
- }
-}
-
-void vp8_setup_block_dptrs(MACROBLOCKD *x)
-{
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- x->block[r*4+c].diff = &x->diff[r * 4 * 16 + c * 4];
- x->block[r*4+c].predictor = x->predictor + r * 4 * 16 + c * 4;
- }
- }
-
- for (r = 0; r < 2; r++)
- {
- for (c = 0; c < 2; c++)
- {
- x->block[16+r*2+c].diff = &x->diff[256 + r * 4 * 8 + c * 4];
- x->block[16+r*2+c].predictor = x->predictor + 256 + r * 4 * 8 + c * 4;
-
- }
- }
-
- for (r = 0; r < 2; r++)
- {
- for (c = 0; c < 2; c++)
- {
- x->block[20+r*2+c].diff = &x->diff[320+ r * 4 * 8 + c * 4];
- x->block[20+r*2+c].predictor = x->predictor + 320 + r * 4 * 8 + c * 4;
-
- }
- }
-
- x->block[24].diff = &x->diff[384];
-
- for (r = 0; r < 25; r++)
- {
- x->block[r].qcoeff = x->qcoeff + r * 16;
- x->block[r].dqcoeff = x->dqcoeff + r * 16;
- }
-}
-
-void vp8_build_block_doffsets(MACROBLOCKD *x)
-{
-
- /* handle the destination pitch features */
- setup_macroblock(x, DEST);
- setup_macroblock(x, PRED);
-}
diff --git a/vp8/common/mips/idct_mips.c b/vp8/common/mips/idct_mips.c
deleted file mode 100755
index 4e7ec01..0000000
--- a/vp8/common/mips/idct_mips.c
+++ /dev/null
@@ -1,332 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_ports/config.h"
-#include "onyxc_int.h"
-
-static const int cospi8sqrt2minus1 = 20091;
-static const int sinpi8sqrt2 = 35468;
-
-
-inline void prefetch_load_short(short* src) {
- __asm__ __volatile__ (
- "pref 0, 0(%[src]) \n\t"
- :
- : [src] "r" (src)
- );
-}
-
-
-void vp8_short_idct4x4llm_mips(short *input, short *output, int pitch)
-{
- int i;
- int a1, b1, c1, d1, c2, d2;
-
- /* pitch has always value 4 */
-
- short *ip = input;
- short *op = output;
- int temp1, temp2, temp3, temp4;
-
- /* prepare data for load */
- prefetch_load_short(ip + 8);
-
- /* first loop is unrolled */
- a1 = ip[0] + ip[8];
- b1 = ip[0] - ip[8];
-
- temp1 = (ip[4] * sinpi8sqrt2) >> 16;
- temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
- c1 = temp1 - temp2;
-
- temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
- temp2 = (ip[12] * sinpi8sqrt2) >> 16;
- d1 = temp1 + temp2;
-
- temp3 = (ip[5] * sinpi8sqrt2) >> 16;
- temp4 = ip[13] + ((ip[13] * cospi8sqrt2minus1) >> 16);
- c2 = temp3 - temp4;
-
- temp3 = ip[5] + ((ip[5] * cospi8sqrt2minus1) >> 16);
- temp4 = (ip[13] * sinpi8sqrt2) >> 16;
- d2 = temp3 + temp4;
-
- op[0] = a1 + d1;
- op[12] = a1 - d1;
- op[4] = b1 + c1;
- op[8] = b1 - c1;
-
- a1 = ip[1] + ip[9];
- b1 = ip[1] - ip[9];
-
- op[1] = a1 + d2;
- op[13] = a1 - d2;
- op[5] = b1 + c2;
- op[9] = b1 - c2;
-
- a1 = ip[2] + ip[10];
- b1 = ip[2] - ip[10];
-
- temp1 = (ip[6] * sinpi8sqrt2) >> 16;
- temp2 = ip[14] + ((ip[14] * cospi8sqrt2minus1) >> 16);
- c1 = temp1 - temp2;
-
- temp1 = ip[6] + ((ip[6] * cospi8sqrt2minus1) >> 16);
- temp2 = (ip[14] * sinpi8sqrt2) >> 16;
- d1 = temp1 + temp2;
-
- temp3 = (ip[7] * sinpi8sqrt2) >> 16;
- temp4 = ip[15] + ((ip[15] * cospi8sqrt2minus1) >> 16);
- c2 = temp3 - temp4;
-
- temp3 = ip[7] + ((ip[7] * cospi8sqrt2minus1) >> 16);
- temp4 = (ip[15] * sinpi8sqrt2) >> 16;
- d2 = temp3 + temp4;
-
- op[2] = a1 + d1;
- op[14] = a1 - d1;
- op[6] = b1 + c1;
- op[10] = b1 - c1;
-
- a1 = ip[3] + ip[11];
- b1 = ip[3] - ip[11];
-
- op[3] = a1 + d2;
- op[15] = a1 - d2;
- op[7] = b1 + c2;
- op[11] = b1 - c2;
-
- ip = output;
-
- /* prepare data for load */
- prefetch_load_short(ip + pitch);
-
- /* second loop is unrolled */
- a1 = ip[0] + ip[2];
- b1 = ip[0] - ip[2];
-
- temp1 = (ip[1] * sinpi8sqrt2) >> 16;
- temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
- c1 = temp1 - temp2;
-
- temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
- temp2 = (ip[3] * sinpi8sqrt2) >> 16;
- d1 = temp1 + temp2;
-
- temp3 = (ip[5] * sinpi8sqrt2) >> 16;
- temp4 = ip[7] + ((ip[7] * cospi8sqrt2minus1) >> 16);
- c2 = temp3 - temp4;
-
- temp3 = ip[5] + ((ip[5] * cospi8sqrt2minus1) >> 16);
- temp4 = (ip[7] * sinpi8sqrt2) >> 16;
- d2 = temp3 + temp4;
-
- op[0] = (a1 + d1 + 4) >> 3;
- op[3] = (a1 - d1 + 4) >> 3;
- op[1] = (b1 + c1 + 4) >> 3;
- op[2] = (b1 - c1 + 4) >> 3;
-
- a1 = ip[4] + ip[6];
- b1 = ip[4] - ip[6];
-
- op[4] = (a1 + d2 + 4) >> 3;
- op[7] = (a1 - d2 + 4) >> 3;
- op[5] = (b1 + c2 + 4) >> 3;
- op[6] = (b1 - c2 + 4) >> 3;
-
- a1 = ip[8] + ip[10];
- b1 = ip[8] - ip[10];
-
- temp1 = (ip[9] * sinpi8sqrt2) >> 16;
- temp2 = ip[11] + ((ip[11] * cospi8sqrt2minus1) >> 16);
- c1 = temp1 - temp2;
-
- temp1 = ip[9] + ((ip[9] * cospi8sqrt2minus1) >> 16);
- temp2 = (ip[11] * sinpi8sqrt2) >> 16;
- d1 = temp1 + temp2;
-
- temp3 = (ip[13] * sinpi8sqrt2) >> 16;
- temp4 = ip[15] + ((ip[15] * cospi8sqrt2minus1) >> 16);
- c2 = temp3 - temp4;
-
- temp3 = ip[13] + ((ip[13] * cospi8sqrt2minus1) >> 16);
- temp4 = (ip[15] * sinpi8sqrt2) >> 16;
- d2 = temp3 + temp4;
-
- op[8] = (a1 + d1 + 4) >> 3;
- op[11] = (a1 - d1 + 4) >> 3;
- op[9] = (b1 + c1 + 4) >> 3;
- op[10] = (b1 - c1 + 4) >> 3;
-
- a1 = ip[12] + ip[14];
- b1 = ip[12] - ip[14];
-
- op[12] = (a1 + d2 + 4) >> 3;
- op[15] = (a1 - d2 + 4) >> 3;
- op[13] = (b1 + c2 + 4) >> 3;
- op[14] = (b1 - c2 + 4) >> 3;
-}
-
-
-void vp8_dc_only_idct_add_mips
-(
- short input_dc,
- unsigned char *pred_ptr,
- unsigned char *dst_ptr,
- int pitch,
- int stride
-)
-{
- int i, a1, absa1;
- int t2, vector_a1, vector_a;
-
- /* a1 = ((input_dc + 4) >> 3); */
- __asm__ __volatile__ (
- "addi %[a1], %[input_dc], 4 \n\t"
- "sra %[a1], %[a1], 3 \n\t"
- : [a1] "=r" (a1)
- : [input_dc] "r" (input_dc)
- );
-
- /* first for loop is unrolled
- * if (a1 < 0) then always (a1 + pred_ptr[c]) < 255
- */
- if (a1 < 0) {
- /* use quad-byte
- * input and output memory are four byte aligned
- */
- __asm__ __volatile__ (
- "abs %[absa1], %[a1] \n\t"
- "replv.qb %[vector_a1], %[absa1] \n\t"
- : [absa1] "=r" (absa1), [vector_a1] "=r" (vector_a1)
- : [a1] "r" (a1)
- );
-
- /* use (a1 - predptr[c]) instead a1 + predptr[c] */
- for (i = 4; i--;)
- {
- __asm__ __volatile__ (
- "lw %[t2], 0(%[pred_ptr]) \n\t"
- "add %[pred_ptr], %[pred_ptr], %[pitch] \n\t"
- "subu_s.qb %[vector_a], %[t2], %[vector_a1] \n\t"
- "sw %[vector_a], 0(%[dst_ptr]) \n\t"
- "add %[dst_ptr], %[dst_ptr], %[stride] \n\t"
- : [t2] "=&r" (t2), [vector_a] "=&r" (vector_a),
- [dst_ptr] "+&r" (dst_ptr), [pred_ptr] "+&r" (pred_ptr)
- : [stride] "r" (stride), [pitch] "r" (pitch), [vector_a1] "r" (vector_a1)
- );
- }
- }
- else {
- /* use quad-byte
- * input and output memory are four byte aligned
- */
- __asm__ __volatile__ (
- "replv.qb %[vector_a1], %[a1] \n\t"
- : [vector_a1] "=r" (vector_a1)
- : [a1] "r" (a1)
- );
-
- for (i = 4; i--;)
- {
- __asm__ __volatile__ (
- "lw %[t2], 0(%[pred_ptr]) \n\t"
- "add %[pred_ptr], %[pred_ptr], %[pitch] \n\t"
- "addu_s.qb %[vector_a], %[vector_a1], %[t2] \n\t"
- "sw %[vector_a], 0(%[dst_ptr]) \n\t"
- "add %[dst_ptr], %[dst_ptr], %[stride] \n\t"
- : [t2] "=&r" (t2), [vector_a] "=&r" (vector_a),
- [dst_ptr] "+&r" (dst_ptr), [pred_ptr] "+&r" (pred_ptr)
- : [stride] "r" (stride), [pitch] "r" (pitch), [vector_a1] "r" (vector_a1)
- );
- }
- }
-}
-
-
-void vp8_short_inv_walsh4x4_mips(short *input, short *output)
-{
- int i;
- int a1, b1, c1, d1;
- int a2, b2, c2, d2;
-
- short *ip = input;
- short *op = output;
-
- prefetch_load_short(ip);
-
- for (i = 4; i--;)
- {
- a1 = ip[0] + ip[12];
- b1 = ip[4] + ip[8];
- c1 = ip[4] - ip[8];
- d1 = ip[0] - ip[12];
-
- op[0] = a1 + b1;
- op[4] = c1 + d1;
- op[8] = a1 - b1;
- op[12] = d1 - c1;
-
- ip++;
- op++;
- }
-
- ip = output;
- op = output;
-
- prefetch_load_short(ip);
-
- for (i = 4; i--;)
- {
- a1 = ip[0] + ip[3] + 3;
- b1 = ip[1] + ip[2];
- c1 = ip[1] - ip[2];
- d1 = ip[0] - ip[3] + 3;
-
- a2 = a1 + b1;
- b2 = d1 + c1;
- c2 = a1 - b1;
- d2 = d1 - c1;
-
- op[0] = a2 >> 3;
- op[1] = b2 >> 3;
- op[2] = c2 >> 3;
- op[3] = d2 >> 3;
-
- ip += 4;
- op += 4;
- }
-}
-
-
-void vp8_short_inv_walsh4x4_1_mips(short *input, short *output)
-{
- int a1;
- int vect_a;
- unsigned int *op = (unsigned int *)output;
-
- a1 = ((input[0] + 3) >> 3);
-
- __asm__ __volatile__ (
- "replv.ph %[vect_a], %[a1] \n\t"
- : [vect_a] "=r" (vect_a)
- : [a1] "r" (a1)
- );
-
- /* output is 4 byte aligned */
- op[0] = vect_a;
- op[1] = vect_a;
- op[2] = vect_a;
- op[3] = vect_a;
- op[4] = vect_a;
- op[5] = vect_a;
- op[6] = vect_a;
- op[7] = vect_a;
-}
diff --git a/vp8/common/mips/idct_mips.h b/vp8/common/mips/idct_mips.h
deleted file mode 100755
index ae3844f..0000000
--- a/vp8/common/mips/idct_mips.h
+++ /dev/null
@@ -1,35 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#ifndef IDCT_MIPS_H
-#define IDCT_MIPS_H
-
-extern prototype_idct(vp8_short_idct4x4llm_mips);
-extern prototype_second_order(vp8_short_inv_walsh4x4_mips);
-extern prototype_idct_scalar_add(vp8_dc_only_idct_add_mips);
-extern prototype_second_order(vp8_short_inv_walsh4x4_1_mips);
-
-#undef vp8_idct_idct16
-#define vp8_idct_idct16 vp8_short_idct4x4llm_mips
-
-#undef vp8_idct_iwalsh16
-#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_mips
-
-#undef vp8_idct_idct1_scalar_add
-#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_mips
-
-#undef vp8_idct_iwalsh1
-#define vp8_idct_iwalsh1 vp8_short_inv_walsh4x4_1_mips
-
-#endif
-#endif
\ No newline at end of file
diff --git a/vp8/common/mips/loopfilter_filters_mips.c b/vp8/common/mips/loopfilter_filters_mips.c
deleted file mode 100755
index fa1a100..0000000
--- a/vp8/common/mips/loopfilter_filters_mips.c
+++ /dev/null
@@ -1,2420 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "onyxc_int.h"
-
-typedef unsigned char uc;
-
-
-/* prefetch data for load */
-inline void prefetch_load_lf(unsigned char* src) {
- __asm__ __volatile__ (
- "pref 0, 0(%[src]) \n\t"
- :
- : [src] "r" (src)
- );
-}
-
-
-/* prefetch data for store */
-inline void prefetch_store_lf(unsigned char* dst) {
- __asm__ __volatile__ (
- "pref 1, 0(%[dst]) \n\t"
- :
- : [dst] "r" (dst)
- );
-}
-
-
-/* processing 4 pixels at the same time
- * compute hev and mask in the same function
- */
-static __inline void vp8_filter_mask_vec_mips
-(
- uint32_t limit,
- uint32_t flimit,
- uint32_t p1,
- uint32_t p0,
- uint32_t p3,
- uint32_t p2,
- uint32_t q0,
- uint32_t q1,
- uint32_t q2,
- uint32_t q3,
- uint32_t thresh,
- uint32_t *hev,
- uint32_t *mask
-)
-{
- uint32_t c, r, r3, r_k;
- uint32_t s1, s2, s3;
- uint32_t ones = 0xFFFFFFFF;
- uint32_t hev1;
-
- __asm__ __volatile__ (
- /* mask |= (abs(p3 - p2) > limit) */
- "subu_s.qb %[c], %[p3], %[p2] \n\t"
- "subu_s.qb %[r_k], %[p2], %[p3] \n\t"
- "or %[r_k], %[r_k], %[c] \n\t"
- "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
- "or %[r], $0, %[c] \n\t"
-
- /* mask |= (abs(p2 - p1) > limit) */
- "subu_s.qb %[c], %[p2], %[p1] \n\t"
- "subu_s.qb %[r_k], %[p1], %[p2] \n\t"
- "or %[r_k], %[r_k], %[c] \n\t"
- "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
- "or %[r], %[r], %[c] \n\t"
-
- /* mask |= (abs(p1 - p0) > limit)
- * hev |= (abs(p1 - p0) > thresh)
- */
- "subu_s.qb %[c], %[p1], %[p0] \n\t"
- "subu_s.qb %[r_k], %[p0], %[p1] \n\t"
- "or %[r_k], %[r_k], %[c] \n\t"
- "cmpgu.lt.qb %[c], %[thresh], %[r_k] \n\t"
- "or %[r3], $0, %[c] \n\t"
- "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
- "or %[r], %[r], %[c] \n\t"
-
- /* mask |= (abs(q1 - q0) > limit)
- * hev |= (abs(q1 - q0) > thresh)
- */
- "subu_s.qb %[c], %[q1], %[q0] \n\t"
- "subu_s.qb %[r_k], %[q0], %[q1] \n\t"
- "or %[r_k], %[r_k], %[c] \n\t"
- "cmpgu.lt.qb %[c], %[thresh], %[r_k] \n\t"
- "or %[r3], %[r3], %[c] \n\t"
- "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
- "or %[r], %[r], %[c] \n\t"
-
- /* mask |= (abs(q2 - q1) > limit) */
- "subu_s.qb %[c], %[q2], %[q1] \n\t"
- "subu_s.qb %[r_k], %[q1], %[q2] \n\t"
- "or %[r_k], %[r_k], %[c] \n\t"
- "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
- "or %[r], %[r], %[c] \n\t"
- "sll %[r3], %[r3], 24 \n\t"
-
- /* mask |= (abs(q3 - q2) > limit) */
- "subu_s.qb %[c], %[q3], %[q2] \n\t"
- "subu_s.qb %[r_k], %[q2], %[q3] \n\t"
- "or %[r_k], %[r_k], %[c] \n\t"
- "cmpgu.lt.qb %[c], %[limit], %[r_k] \n\t"
- "or %[r], %[r], %[c] \n\t"
-
- : [c] "=&r" (c), [r_k] "=&r" (r_k),
- [r] "=&r" (r), [r3] "=&r" (r3)
- : [limit] "r" (limit), [p3] "r" (p3), [p2] "r" (p2),
- [p1] "r" (p1), [p0] "r" (p0), [q1] "r" (q1), [q0] "r" (q0),
- [q2] "r" (q2), [q3] "r" (q3), [thresh] "r" (thresh)
- );
-
- __asm__ __volatile__ (
- /* abs(p0 - q0) */
- "subu_s.qb %[c], %[p0], %[q0] \n\t"
- "subu_s.qb %[r_k], %[q0], %[p0] \n\t"
- "wrdsp %[r3] \n\t"
- "or %[s1], %[r_k], %[c] \n\t"
-
- /* abs(p1 - q1) */
- "subu_s.qb %[c], %[p1], %[q1] \n\t"
- "addu_s.qb %[s3], %[s1], %[s1] \n\t"
- "pick.qb %[hev1], %[ones], $0 \n\t"
- "subu_s.qb %[r_k], %[q1], %[p1] \n\t"
- "or %[s2], %[r_k], %[c] \n\t"
-
- /* abs(p0 - q0) * 2 + abs(p1 - q1) / 2 > flimit * 2 + limit */
- "shrl.qb %[s2], %[s2], 1 \n\t"
- "addu_s.qb %[s1], %[s2], %[s3] \n\t"
- "cmpgu.lt.qb %[c], %[flimit], %[s1] \n\t"
- "or %[r], %[r], %[c] \n\t"
- "sll %[r], %[r], 24 \n\t"
-
- "wrdsp %[r] \n\t"
- "pick.qb %[s2], $0, %[ones] \n\t"
-
- : [c] "=&r" (c), [r_k] "=&r" (r_k), [s1] "=&r" (s1), [hev1] "=&r" (hev1),
- [s2] "=&r" (s2), [r] "+r" (r), [s3] "=&r" (s3)
- : [p0] "r" (p0), [q0] "r" (q0), [p1] "r" (p1), [r3] "r" (r3),
- [q1] "r" (q1), [ones] "r" (ones), [flimit] "r" (flimit)
- );
-
- *hev = hev1;
- *mask = s2;
-}
-
-
-/* inputs & outputs are quad-byte vectors */
-static __inline void vp8_filter_mips
-(
- uint32_t mask,
- uint32_t hev,
- uint32_t *ps1,
- uint32_t *ps0,
- uint32_t *qs0,
- uint32_t *qs1
-)
-{
- int32_t vp8_filter_l, vp8_filter_r;
- int32_t Filter1_l, Filter1_r, Filter2_l, Filter2_r;
- int32_t subr_r, subr_l;
- uint32_t t1, t2, HWM, t3;
- uint32_t hev_l, hev_r, mask_l, mask_r, invhev_l, invhev_r;
-
- int32_t vps1, vps0, vqs0, vqs1;
- int32_t vps1_l, vps1_r, vps0_l, vps0_r, vqs0_l, vqs0_r, vqs1_l, vqs1_r;
- uint32_t N128;
-
- N128= 0x80808080;
- t1 = 0x03000300;
- t2 = 0x04000400;
- t3 = 0x01000100;
- HWM = 0xFF00FF00;
-
- vps0 = (*ps0) ^ N128;
- vps1 = (*ps1) ^ N128;
- vqs0 = (*qs0) ^ N128;
- vqs1 = (*qs1) ^ N128;
-
- /* use halfword pairs instead quad-bytes because of accuracy */
- vps0_l = vps0 & HWM;
- vps0_r = vps0 << 8;
- vps0_r = vps0_r & HWM;
-
- vps1_l = vps1 & HWM;
- vps1_r = vps1 << 8;
- vps1_r = vps1_r & HWM;
-
- vqs0_l = vqs0 & HWM;
- vqs0_r = vqs0 << 8;
- vqs0_r = vqs0_r & HWM;
-
- vqs1_l = vqs1 & HWM;
- vqs1_r = vqs1 << 8;
- vqs1_r = vqs1_r & HWM;
-
- mask_l = mask & HWM;
- mask_r = mask << 8;
- mask_r = mask_r & HWM;
-
- hev_l = hev & HWM;
- hev_r = hev << 8;
- hev_r = hev_r & HWM;
-
- __asm__ __volatile__ (
- /* vp8_filter = vp8_signed_char_clamp(ps1 - qs1); */
- "subq_s.ph %[vp8_filter_l], %[vps1_l], %[vqs1_l] \n\t"
- "subq_s.ph %[vp8_filter_r], %[vps1_r], %[vqs1_r] \n\t"
-
- /* qs0 - ps0 */
- "subq_s.ph %[subr_l], %[vqs0_l], %[vps0_l] \n\t"
- "subq_s.ph %[subr_r], %[vqs0_r], %[vps0_r] \n\t"
-
- /* vp8_filter &= hev; */
- "and %[vp8_filter_l], %[vp8_filter_l], %[hev_l] \n\t"
- "and %[vp8_filter_r], %[vp8_filter_r], %[hev_r] \n\t"
-
- /* vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0)); */
- "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
- "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
- "xor %[invhev_l], %[hev_l], %[HWM] \n\t"
- "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
- "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
- "xor %[invhev_r], %[hev_r], %[HWM] \n\t"
- "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
- "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
-
- /* vp8_filter &= mask; */
- "and %[vp8_filter_l], %[vp8_filter_l], %[mask_l] \n\t"
- "and %[vp8_filter_r], %[vp8_filter_r], %[mask_r] \n\t"
-
- : [vp8_filter_l] "=&r" (vp8_filter_l), [vp8_filter_r] "=&r" (vp8_filter_r),
- [subr_l] "=&r" (subr_l), [subr_r] "=&r" (subr_r),
- [invhev_l] "=&r" (invhev_l), [invhev_r] "=&r" (invhev_r)
-
- : [vps0_l] "r" (vps0_l), [vps0_r] "r" (vps0_r), [vps1_l] "r" (vps1_l),
- [vps1_r] "r" (vps1_r), [vqs0_l] "r" (vqs0_l), [vqs0_r] "r" (vqs0_r),
- [vqs1_l] "r" (vqs1_l), [vqs1_r] "r" (vqs1_r),
- [mask_l] "r" (mask_l), [mask_r] "r" (mask_r),
- [hev_l] "r" (hev_l), [hev_r] "r" (hev_r),
- [HWM] "r" (HWM)
- );
-
- /* save bottom 3 bits so that we round one side +4 and the other +3 */
- __asm__ __volatile__ (
- /* Filter2 = vp8_signed_char_clamp(vp8_filter + 3) >>= 3; */
- "addq_s.ph %[Filter1_l], %[vp8_filter_l], %[t2] \n\t"
- "addq_s.ph %[Filter1_r], %[vp8_filter_r], %[t2] \n\t"
-
- /* Filter1 = vp8_signed_char_clamp(vp8_filter + 4) >>= 3; */
- "addq_s.ph %[Filter2_l], %[vp8_filter_l], %[t1] \n\t"
- "addq_s.ph %[Filter2_r], %[vp8_filter_r], %[t1] \n\t"
- "shra.ph %[Filter1_r], %[Filter1_r], 3 \n\t"
- "shra.ph %[Filter1_l], %[Filter1_l], 3 \n\t"
-
- "shra.ph %[Filter2_l], %[Filter2_l], 3 \n\t"
- "shra.ph %[Filter2_r], %[Filter2_r], 3 \n\t"
-
- "and %[Filter1_l], %[Filter1_l], %[HWM] \n\t"
- "and %[Filter1_r], %[Filter1_r], %[HWM] \n\t"
-
- /* vps0 = vp8_signed_char_clamp(ps0 + Filter2); */
- "addq_s.ph %[vps0_l], %[vps0_l], %[Filter2_l] \n\t"
- "addq_s.ph %[vps0_r], %[vps0_r], %[Filter2_r] \n\t"
-
- /* vqs0 = vp8_signed_char_clamp(qs0 - Filter1); */
- "subq_s.ph %[vqs0_l], %[vqs0_l], %[Filter1_l] \n\t"
- "subq_s.ph %[vqs0_r], %[vqs0_r], %[Filter1_r] \n\t"
-
- : [Filter1_l] "=&r" (Filter1_l), [Filter1_r] "=&r" (Filter1_r),
- [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r),
- [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
- [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)
-
- : [t1] "r" (t1), [t2] "r" (t2),
- [vp8_filter_l] "r" (vp8_filter_l), [vp8_filter_r] "r" (vp8_filter_r),
- [HWM] "r" (HWM)
- );
-
- __asm__ __volatile__ (
- /* (vp8_filter += 1) >>= 1 */
- "addqh.ph %[Filter1_l], %[Filter1_l], %[t3] \n\t"
- "addqh.ph %[Filter1_r], %[Filter1_r], %[t3] \n\t"
-
- /* vp8_filter &= ~hev; */
- "and %[Filter1_l], %[Filter1_l], %[invhev_l] \n\t"
- "and %[Filter1_r], %[Filter1_r], %[invhev_r] \n\t"
-
- /* vps1 = vp8_signed_char_clamp(ps1 + vp8_filter); */
- "addq_s.ph %[vps1_l], %[vps1_l], %[Filter1_l] \n\t"
- "addq_s.ph %[vps1_r], %[vps1_r], %[Filter1_r] \n\t"
-
- /* vqs1 = vp8_signed_char_clamp(qs1 - vp8_filter); */
- "subq_s.ph %[vqs1_l], %[vqs1_l], %[Filter1_l] \n\t"
- "subq_s.ph %[vqs1_r], %[vqs1_r], %[Filter1_r] \n\t"
-
- : [Filter1_l] "+r" (Filter1_l), [Filter1_r] "+r" (Filter1_r),
- [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r),
- [vqs1_l] "+r" (vqs1_l), [vqs1_r] "+r" (vqs1_r)
-
- : [t3] "r" (t3), [invhev_l] "r" (invhev_l), [invhev_r] "r" (invhev_r)
- );
-
- /* Create quad-bytes from halfword pairs */
- vqs0_l = vqs0_l & HWM;
- vqs1_l = vqs1_l & HWM;
- vps0_l = vps0_l & HWM;
- vps1_l = vps1_l & HWM;
-
- __asm__ __volatile__ (
- "shrl.ph %[vqs0_r], %[vqs0_r], 8 \n\t"
- "shrl.ph %[vps0_r], %[vps0_r], 8 \n\t"
- "shrl.ph %[vqs1_r], %[vqs1_r], 8 \n\t"
- "shrl.ph %[vps1_r], %[vps1_r], 8 \n\t"
-
- : [vps1_r] "+r" (vps1_r), [vqs1_r] "+r" (vqs1_r),
- [vps0_r] "+r" (vps0_r), [vqs0_r] "+r" (vqs0_r)
- :
- );
-
- vqs0 = vqs0_l | vqs0_r;
- vqs1 = vqs1_l | vqs1_r;
- vps0 = vps0_l | vps0_r;
- vps1 = vps1_l | vps1_r;
-
- *ps0 = vps0 ^ N128;
- *ps1 = vps1 ^ N128;
- *qs0 = vqs0 ^ N128;
- *qs1 = vqs1 ^ N128;
-}
-
-
-void vp8_loop_filter_horizontal_edge_mips
-(
- unsigned char *s,
- int p,
- unsigned int flimit,
- unsigned int limit,
- unsigned int thresh,
- int count
-)
-{
- uint32_t mask;
- uint32_t hev;
- uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
- unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
-
- mask = 0;
- hev = 0;
- p1=0; p2=0; p3=0; p4=0;
-
- /* prefetch data for store */
- prefetch_store_lf(s);
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
-
- sm1 = s - (p<<2);
- s0 = s - p - p - p;
- s1 = s - p -p ;
- s2 = s - p;
- s3 = s;
- s4 = s + p;
- s5 = s + p + p;
- s6 = s + p + p + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood */
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- }
- }
-
- sm1 += 4;
- s0 += 4;
- s1 += 4;
- s2 += 4;
- s3 += 4;
- s4 += 4;
- s5 += 4;
- s6 += 4;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood */
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- }
- }
-
- sm1 += 4;
- s0 += 4;
- s1 += 4;
- s2 += 4;
- s3 += 4;
- s4 += 4;
- s5 += 4;
- s6 += 4;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood */
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- }
- }
-
- sm1 += 4;
- s0 += 4;
- s1 += 4;
- s2 += 4;
- s3 += 4;
- s4 += 4;
- s5 += 4;
- s6 += 4;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood */
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- }
- }
-}
-
-
-void vp8_loop_filter_uvhorizontal_edge_mips
-(
- unsigned char *s,
- int p,
- unsigned int flimit,
- unsigned int limit,
- unsigned int thresh,
- int count
-)
-{
- uint32_t mask;
- uint32_t hev;
- uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
- unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
-
- mask = 0;
- hev = 0;
- p1=0; p2=0; p3=0; p4=0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
-
- sm1 = s - (p<<2);
- s0 = s - p - p - p;
- s1 = s - p -p ;
- s2 = s - p;
- s3 = s;
- s4 = s + p;
- s5 = s + p + p;
- s6 = s + p + p + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood */
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- }
- }
-
- sm1 += 4;
- s0 += 4;
- s1 += 4;
- s2 += 4;
- s3 += 4;
- s4 += 4;
- s5 += 4;
- s6 += 4;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood */
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- }
- }
-}
-
-
-void vp8_loop_filter_vertical_edge_mips
-(
- unsigned char *s,
- int p,
- unsigned int flimit,
- unsigned int limit,
- unsigned int thresh,
- int count
-)
-{
- int i;
- uint32_t mask, hev;
- uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
- unsigned char *s1, *s2, *s3, *s4;
- uint32_t prim1, prim2, sec3, sec4, prim3, prim4;
-
- hev = 0;
- mask = 0;
- i = 0;
- pm1=0; p0=0; p1=0; p2=0; p3=0; p4=0; p5=0; p6=0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
-
- /* apply filter on 4 pixesl at the same time */
- do {
-
- /* prefetch data for store */
- prefetch_store_lf(s+p);
-
- s1 = s;
- s2 = s + p;
- s3 = s2 + p;
- s4 = s3 + p;
- s = s4 + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p2 = *((uint32_t*)(s1-4));
- p6 = *((uint32_t*)(s1));
- p1 = *((uint32_t*)(s2 - 4));
- p5 = *((uint32_t*)(s2));
- p0 = *((uint32_t*)(s3 - 4));
- p4 = *((uint32_t*)(s3));
- pm1 = *((uint32_t*)(s4 - 4));
- p3 = *((uint32_t*)(s4));
-
- /* transpose pm1, p0, p1, p2 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
- "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
- "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
- "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
-
- "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
- "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
- "append %[p1], %[sec3], 16 \n\t"
- "append %[pm1], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* transpose p3, p4, p5, p6 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
- "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
- "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
- "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
-
- "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
- "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
- "append %[p5], %[sec3], 16 \n\t"
- "append %[p3], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood
- * don't use transpose on output data
- * because memory isn't aligned
- */
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s4]) \n\t"
- "sb %[p3], 0(%[s4]) \n\t"
- "sb %[p2], -1(%[s4]) \n\t"
- "sb %[p1], -2(%[s4]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
- [p2] "r" (p2), [p1] "r" (p1)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s3]) \n\t"
- "sb %[p3], 0(%[s3]) \n\t"
- "sb %[p2], -1(%[s3]) \n\t"
- "sb %[p1], -2(%[s3]) \n\t"
- : [p1] "+r" (p1)
- : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s2]) \n\t"
- "sb %[p3], 0(%[s2]) \n\t"
- "sb %[p2], -1(%[s2]) \n\t"
- "sb %[p1], -2(%[s2]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
- [p2] "r" (p2), [p1] "r" (p1)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s1]) \n\t"
- "sb %[p3], 0(%[s1]) \n\t"
- "sb %[p2], -1(%[s1]) \n\t"
- "sb %[p1], -2(%[s1]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
- [p2] "r" (p2), [p1] "r" (p1)
- );
- }
- }
-
- s1 = s;
- s2 = s + p;
- s3 = s2 + p;
- s4 = s3 + p;
- s = s4 + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p2 = *((uint32_t*)(s1-4));
- p6 = *((uint32_t*)(s1));
- p1 = *((uint32_t*)(s2 - 4));
- p5 = *((uint32_t*)(s2));
- p0 = *((uint32_t*)(s3 - 4));
- p4 = *((uint32_t*)(s3));
- pm1 = *((uint32_t*)(s4 - 4));
- p3 = *((uint32_t*)(s4));
-
- /* transpose pm1, p0, p1, p2 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
- "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
- "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
- "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
-
- "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
- "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
- "append %[p1], %[sec3], 16 \n\t"
- "append %[pm1], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* transpose p3, p4, p5, p6 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
- "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
- "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
- "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
-
- "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
- "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
- "append %[p5], %[sec3], 16 \n\t"
- "append %[p3], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood
- * don't use transpose on output data
- * because memory isn't aligned
- */
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s4]) \n\t"
- "sb %[p3], 0(%[s4]) \n\t"
- "sb %[p2], -1(%[s4]) \n\t"
- "sb %[p1], -2(%[s4]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
- [p2] "r" (p2), [p1] "r" (p1)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s3]) \n\t"
- "sb %[p3], 0(%[s3]) \n\t"
- "sb %[p2], -1(%[s3]) \n\t"
- "sb %[p1], -2(%[s3]) \n\t"
- : [p1] "+r" (p1)
- : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s2]) \n\t"
- "sb %[p3], 0(%[s2]) \n\t"
- "sb %[p2], -1(%[s2]) \n\t"
- "sb %[p1], -2(%[s2]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
- [p2] "r" (p2), [p1] "r" (p1)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s1]) \n\t"
- "sb %[p3], 0(%[s1]) \n\t"
- "sb %[p2], -1(%[s1]) \n\t"
- "sb %[p1], -2(%[s1]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
- [p2] "r" (p2), [p1] "r" (p1)
- );
- }
- }
- i+=8;
- }
-
- while (i < count);
-}
-
-
-void vp8_loop_filter_uvvertical_edge_mips
-(
- unsigned char *s,
- int p,
- unsigned int flimit,
- unsigned int limit,
- unsigned int thresh,
- int count
-)
-{
- uint32_t mask, hev;
- uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
- unsigned char *s1, *s2, *s3, *s4;
- uint32_t temp, prim1, prim2, sec3, sec4, prim3, prim4;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
-
- /* apply filter on 4 pixesl at the same time */
-
- s1 = s;
- s2 = s + p;
- s3 = s2 + p;
- s4 = s3 + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p2 = *((uint32_t*)(s1-4));
- p6 = *((uint32_t*)(s1));
- p1 = *((uint32_t*)(s2 - 4));
- p5 = *((uint32_t*)(s2));
- p0 = *((uint32_t*)(s3 - 4));
- p4 = *((uint32_t*)(s3));
- pm1 = *((uint32_t*)(s4 - 4));
- p3 = *((uint32_t*)(s4));
-
- /* transpose pm1, p0, p1, p2 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
- "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
- "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
- "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
-
- "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
- "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
- "append %[p1], %[sec3], 16 \n\t"
- "append %[pm1], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* transpose p3, p4, p5, p6 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
- "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
- "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
- "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
-
- "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
- "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
- "append %[p5], %[sec3], 16 \n\t"
- "append %[p3], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood
- * don't use transpose on output data
- * because memory isn't aligned
- */
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s4]) \n\t"
- "sb %[p3], 0(%[s4]) \n\t"
- "sb %[p2], -1(%[s4]) \n\t"
- "sb %[p1], -2(%[s4]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
- [p2] "r" (p2), [p1] "r" (p1)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s3]) \n\t"
- "sb %[p3], 0(%[s3]) \n\t"
- "sb %[p2], -1(%[s3]) \n\t"
- "sb %[p1], -2(%[s3]) \n\t"
- : [p1] "+r" (p1)
- : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s2]) \n\t"
- "sb %[p3], 0(%[s2]) \n\t"
- "sb %[p2], -1(%[s2]) \n\t"
- "sb %[p1], -2(%[s2]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
- [p2] "r" (p2), [p1] "r" (p1)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s1]) \n\t"
- "sb %[p3], 0(%[s1]) \n\t"
- "sb %[p2], -1(%[s1]) \n\t"
- "sb %[p1], -2(%[s1]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1), [p2] "r" (p2), [p1] "r" (p1)
- );
- }
- }
-
- s1 = s4 +p;
- s2 = s1 + p;
- s3 = s2 + p;
- s4 = s3 + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p2 = *((uint32_t*)(s1-4));
- p6 = *((uint32_t*)(s1));
- p1 = *((uint32_t*)(s2 - 4));
- p5 = *((uint32_t*)(s2));
- p0 = *((uint32_t*)(s3 - 4));
- p4 = *((uint32_t*)(s3));
- pm1 = *((uint32_t*)(s4 - 4));
- p3 = *((uint32_t*)(s4));
-
- /* transpose pm1, p0, p1, p2 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
- "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
- "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
- "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
-
- "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
- "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
- "append %[p1], %[sec3], 16 \n\t"
- "append %[pm1], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* transpose p3, p4, p5, p6 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
- "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
- "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
- "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
-
- "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
- "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
- "append %[p5], %[sec3], 16 \n\t"
- "append %[p3], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_filter_mips(mask, hev, &p1, &p2, &p3, &p4);
-
- /* unpack processed 4x4 neighborhood
- * don't use transpose on output data
- * because memory isn't aligned
- */
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s4]) \n\t"
- "sb %[p3], 0(%[s4]) \n\t"
- "sb %[p2], -1(%[s4]) \n\t"
- "sb %[p1], -2(%[s4]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
- [p2] "r" (p2), [p1] "r" (p1)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s3]) \n\t"
- "sb %[p3], 0(%[s3]) \n\t"
- "sb %[p2], -1(%[s3]) \n\t"
- "sb %[p1], -2(%[s3]) \n\t"
- : [p1] "+r" (p1)
- : [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3), [p2] "r" (p2)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s2]) \n\t"
- "sb %[p3], 0(%[s2]) \n\t"
- "sb %[p2], -1(%[s2]) \n\t"
- "sb %[p1], -2(%[s2]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
- [p2] "r" (p2), [p1] "r" (p1)
- );
-
- __asm__ __volatile__ (
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- : [p4] "+r" (p4), [p3] "+r" (p3), [p2] "+r" (p2), [p1] "+r" (p1)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p4], 1(%[s1]) \n\t"
- "sb %[p3], 0(%[s1]) \n\t"
- "sb %[p2], -1(%[s1]) \n\t"
- "sb %[p1], -2(%[s1]) \n\t"
- :
- : [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
- [p2] "r" (p2), [p1] "r" (p1)
- );
- }
- }
-}
-
-
-/* inputs & outputs are quad-byte vectors */
-static __inline void vp8_mbfilter_mips
-(
- uint32_t mask,
- uint32_t hev,
- uint32_t *ps2,
- uint32_t *ps1,
- uint32_t *ps0,
- uint32_t *qs0,
- uint32_t *qs1,
- uint32_t *qs2
-)
-{
- int32_t vps2, vps1, vps0, vqs0, vqs1, vqs2;
- int32_t vps2_l, vps1_l, vps0_l, vqs0_l, vqs1_l, vqs2_l;
- int32_t vps2_r, vps1_r, vps0_r, vqs0_r, vqs1_r, vqs2_r;
- uint32_t HWM, vp8_filter_l, vp8_filter_r, mask_l, mask_r, hev_l, hev_r, subr_r, subr_l;
- uint32_t Filter2_l, Filter2_r, t1, t2, Filter1_l, Filter1_r, invhev_l, invhev_r;
- uint32_t N128, R63;
- uint32_t u1_l, u1_r, u2_l, u2_r, u3_l, u3_r;
-
- R63 = 0x003F003F;
- HWM = 0xFF00FF00;
- N128 = 0x80808080;
- t1 = 0x03000300;
- t2 = 0x04000400;
-
- vps0 = (*ps0) ^ N128;
- vps1 = (*ps1) ^ N128;
- vps2 = (*ps2) ^ N128;
- vqs0 = (*qs0) ^ N128;
- vqs1 = (*qs1) ^ N128;
- vqs2 = (*qs2) ^ N128;
-
- /* use halfword pairs instead quad-bytes because of accuracy */
- vps0_l = vps0 & HWM;
- vps0_r = vps0 << 8;
- vps0_r = vps0_r & HWM;
-
- vqs0_l = vqs0 & HWM;
- vqs0_r = vqs0 << 8;
- vqs0_r = vqs0_r & HWM;
-
- vps1_l = vps1 & HWM;
- vps1_r = vps1 << 8;
- vps1_r = vps1_r & HWM;
-
- vqs1_l = vqs1 & HWM;
- vqs1_r = vqs1 << 8;
- vqs1_r = vqs1_r & HWM;
-
- vqs2_l = vqs2 & HWM;
- vqs2_r = vqs2 << 8;
- vqs2_r = vqs2_r & HWM;
-
- __asm__ __volatile__ (
- /* qs0 - ps0 */
- "subq_s.ph %[subr_l], %[vqs0_l], %[vps0_l] \n\t"
- "subq_s.ph %[subr_r], %[vqs0_r], %[vps0_r] \n\t"
-
- /* vp8_filter = vp8_signed_char_clamp(ps1 - qs1); */
- "subq_s.ph %[vp8_filter_l], %[vps1_l], %[vqs1_l] \n\t"
- "subq_s.ph %[vp8_filter_r], %[vps1_r], %[vqs1_r] \n\t"
-
- : [vp8_filter_l] "=&r" (vp8_filter_l), [vp8_filter_r] "=r" (vp8_filter_r),
- [subr_l] "=&r" (subr_l), [subr_r] "=&r" (subr_r)
- : [vps0_l] "r" (vps0_l), [vps0_r] "r" (vps0_r), [vps1_l] "r" (vps1_l),
- [vps1_r] "r" (vps1_r), [vqs0_l] "r" (vqs0_l), [vqs0_r] "r" (vqs0_r),
- [vqs1_l] "r" (vqs1_l), [vqs1_r] "r" (vqs1_r)
- );
-
- vps2_l = vps2 & HWM;
- vps2_r = vps2 << 8;
- vps2_r = vps2_r & HWM;
-
- /* add outer taps if we have high edge variance */
- __asm__ __volatile__ (
- /* vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0)); */
- "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
- "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
- "and %[mask_l], %[HWM], %[mask] \n\t"
- "sll %[mask_r], %[mask], 8 \n\t"
- "and %[mask_r], %[HWM], %[mask_r] \n\t"
- "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
- "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
- "and %[hev_l], %[HWM], %[hev] \n\t"
- "sll %[hev_r], %[hev], 8 \n\t"
- "and %[hev_r], %[HWM], %[hev_r] \n\t"
- "addq_s.ph %[vp8_filter_l], %[vp8_filter_l], %[subr_l] \n\t"
- "addq_s.ph %[vp8_filter_r], %[vp8_filter_r], %[subr_r] \n\t"
-
- /* vp8_filter &= mask; */
- "and %[vp8_filter_l], %[vp8_filter_l], %[mask_l] \n\t"
- "and %[vp8_filter_r], %[vp8_filter_r], %[mask_r] \n\t"
-
- /* Filter2 = vp8_filter & hev; */
- "and %[Filter2_l], %[vp8_filter_l], %[hev_l] \n\t"
- "and %[Filter2_r], %[vp8_filter_r], %[hev_r] \n\t"
-
- : [vp8_filter_l] "+r" (vp8_filter_l), [vp8_filter_r] "+r" (vp8_filter_r),
- [hev_l] "=&r" (hev_l), [hev_r] "=&r" (hev_r),
- [mask_l] "=&r" (mask_l), [mask_r] "=&r" (mask_r),
- [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r)
- : [subr_l] "r" (subr_l), [subr_r] "r" (subr_r),
- [HWM] "r" (HWM), [hev] "r" (hev), [mask] "r" (mask)
- );
-
- /* save bottom 3 bits so that we round one side +4 and the other +3 */
- __asm__ __volatile__ (
- /* Filter1 = vp8_signed_char_clamp(Filter2 + 4) >>= 3; */
- "addq_s.ph %[Filter1_l], %[Filter2_l], %[t2] \n\t"
- "xor %[invhev_l], %[hev_l], %[HWM] \n\t"
- "addq_s.ph %[Filter1_r], %[Filter2_r], %[t2] \n\t"
-
- /* Filter2 = vp8_signed_char_clamp(Filter2 + 3) >>= 3; */
- "addq_s.ph %[Filter2_l], %[Filter2_l], %[t1] \n\t"
- "addq_s.ph %[Filter2_r], %[Filter2_r], %[t1] \n\t"
-
- "shra.ph %[Filter1_l], %[Filter1_l], 3 \n\t"
- "shra.ph %[Filter1_r], %[Filter1_r], 3 \n\t"
-
- "shra.ph %[Filter2_l], %[Filter2_l], 3 \n\t"
- "shra.ph %[Filter2_r], %[Filter2_r], 3 \n\t"
- "and %[Filter1_l], %[Filter1_l], %[HWM] \n\t"
- "and %[Filter1_r], %[Filter1_r], %[HWM] \n\t"
- "xor %[invhev_r], %[hev_r], %[HWM] \n\t"
-
- /* qs0 = vp8_signed_char_clamp(qs0 - Filter1); */
- "subq_s.ph %[vqs0_l], %[vqs0_l], %[Filter1_l] \n\t"
- "subq_s.ph %[vqs0_r], %[vqs0_r], %[Filter1_r] \n\t"
-
- /* ps0 = vp8_signed_char_clamp(ps0 + Filter2); */
- "addq_s.ph %[vps0_l], %[vps0_l], %[Filter2_l] \n\t"
- "addq_s.ph %[vps0_r], %[vps0_r], %[Filter2_r] \n\t"
-
- : [invhev_l] "=&r" (invhev_l), [invhev_r] "=&r" (invhev_r),
- [Filter1_l] "=&r" (Filter1_l), [Filter1_r] "=&r" (Filter1_r),
- [Filter2_l] "+r" (Filter2_l), [Filter2_r] "+r" (Filter2_r),
- [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
- [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)
- : [t1] "r" (t1), [t2] "r" (t2), [HWM] "r" (HWM),
- [hev_l] "r" (hev_l), [hev_r] "r" (hev_r)
- );
-
- /* only apply wider filter if not high edge variance */
- __asm__ __volatile__ (
- /* vp8_filter &= ~hev; */
- "and %[Filter2_l], %[vp8_filter_l], %[invhev_l] \n\t"
- "and %[Filter2_r], %[vp8_filter_r], %[invhev_r] \n\t"
-
- "shra.ph %[Filter2_l], %[Filter2_l], 8 \n\t"
- "shra.ph %[Filter2_r], %[Filter2_r], 8 \n\t"
-
- : [Filter2_l] "=&r" (Filter2_l), [Filter2_r] "=&r" (Filter2_r)
- : [vp8_filter_l] "r" (vp8_filter_l), [vp8_filter_r] "r" (vp8_filter_r),
- [invhev_l] "r" (invhev_l), [invhev_r] "r" (invhev_r)
- );
-
- /* roughly 3/7th difference across boundary */
- __asm__ __volatile__ (
- "shll.ph %[u3_l], %[Filter2_l], 3 \n\t"
- "shll.ph %[u3_r], %[Filter2_r], 3 \n\t"
-
- "addq.ph %[u3_l], %[u3_l], %[Filter2_l] \n\t"
- "addq.ph %[u3_r], %[u3_r], %[Filter2_r] \n\t"
-
- "shll.ph %[u2_l], %[u3_l], 1 \n\t"
- "shll.ph %[u2_r], %[u3_r], 1 \n\t"
-
- "addq.ph %[u1_l], %[u3_l], %[u2_l] \n\t"
- "addq.ph %[u1_r], %[u3_r], %[u2_r] \n\t"
-
- "addq.ph %[u2_l], %[u2_l], %[R63] \n\t"
- "addq.ph %[u2_r], %[u2_r], %[R63] \n\t"
-
- "addq.ph %[u3_l], %[u3_l], %[R63] \n\t"
- "addq.ph %[u3_r], %[u3_r], %[R63] \n\t"
-
- /* vp8_signed_char_clamp((63 + Filter2 * 27) >> 7)
- * vp8_signed_char_clamp((63 + Filter2 * 18) >> 7)
- */
- "addq.ph %[u1_l], %[u1_l], %[R63] \n\t"
- "addq.ph %[u1_r], %[u1_r], %[R63] \n\t"
- "shra.ph %[u1_l], %[u1_l], 7 \n\t"
- "shra.ph %[u1_r], %[u1_r], 7 \n\t"
- "shra.ph %[u2_l], %[u2_l], 7 \n\t"
- "shra.ph %[u2_r], %[u2_r], 7 \n\t"
- "shll.ph %[u1_l], %[u1_l], 8 \n\t"
- "shll.ph %[u1_r], %[u1_r], 8 \n\t"
- "shll.ph %[u2_l], %[u2_l], 8 \n\t"
- "shll.ph %[u2_r], %[u2_r], 8 \n\t"
-
- /* vqs0 = vp8_signed_char_clamp(qs0 - u); */
- "subq_s.ph %[vqs0_l], %[vqs0_l], %[u1_l] \n\t"
- "subq_s.ph %[vqs0_r], %[vqs0_r], %[u1_r] \n\t"
-
- /* vps0 = vp8_signed_char_clamp(ps0 + u); */
- "addq_s.ph %[vps0_l], %[vps0_l], %[u1_l] \n\t"
- "addq_s.ph %[vps0_r], %[vps0_r], %[u1_r] \n\t"
-
- : [u1_l] "=&r" (u1_l), [u1_r] "=&r" (u1_r), [u2_l] "=&r" (u2_l),
- [u2_r] "=&r" (u2_r), [u3_l] "=&r" (u3_l), [u3_r] "=&r" (u3_r),
- [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
- [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r)
- : [R63] "r" (R63),
- [Filter2_l] "r" (Filter2_l), [Filter2_r] "r" (Filter2_r)
- );
-
- __asm__ __volatile__ (
- /* vqs1 = vp8_signed_char_clamp(qs1 - u); */
- "subq_s.ph %[vqs1_l], %[vqs1_l], %[u2_l] \n\t"
- "addq_s.ph %[vps1_l], %[vps1_l], %[u2_l] \n\t"
-
- /* vps1 = vp8_signed_char_clamp(ps1 + u); */
- "addq_s.ph %[vps1_r], %[vps1_r], %[u2_r] \n\t"
- "subq_s.ph %[vqs1_r], %[vqs1_r], %[u2_r] \n\t"
-
- : [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r),
- [vqs1_l] "+r" (vqs1_l), [vqs1_r] "+r" (vqs1_r)
- : [u2_l] "r" (u2_l), [u2_r] "r" (u2_r)
- );
-
- /* roughly 1/7th difference across boundary */
- __asm__ __volatile__ (
- /* u = vp8_signed_char_clamp((63 + Filter2 * 9) >> 7); */
- "shra.ph %[u3_l], %[u3_l], 7 \n\t"
- "shra.ph %[u3_r], %[u3_r], 7 \n\t"
- "shll.ph %[u3_l], %[u3_l], 8 \n\t"
- "shll.ph %[u3_r], %[u3_r], 8 \n\t"
-
- /* vqs2 = vp8_signed_char_clamp(qs2 - u); */
- "subq_s.ph %[vqs2_l], %[vqs2_l], %[u3_l] \n\t"
- "subq_s.ph %[vqs2_r], %[vqs2_r], %[u3_r] \n\t"
-
- /* vps2 = vp8_signed_char_clamp(ps2 + u); */
- "addq_s.ph %[vps2_l], %[vps2_l], %[u3_l] \n\t"
- "addq_s.ph %[vps2_r], %[vps2_r], %[u3_r] \n\t"
-
- : [u3_l] "+r" (u3_l), [u3_r] "+r" (u3_r), [vps2_l] "+r" (vps2_l),
- [vps2_r] "+r" (vps2_r), [vqs2_l] "+r" (vqs2_l), [vqs2_r] "+r" (vqs2_r)
- :
- );
-
- /* Create quad-bytes from halfword pairs */
- __asm__ __volatile__ (
- "and %[vqs0_l], %[vqs0_l], %[HWM] \n\t"
- "shrl.ph %[vqs0_r], %[vqs0_r], 8 \n\t"
-
- "and %[vps0_l], %[vps0_l], %[HWM] \n\t"
- "shrl.ph %[vps0_r], %[vps0_r], 8 \n\t"
-
- "and %[vqs1_l], %[vqs1_l], %[HWM] \n\t"
- "shrl.ph %[vqs1_r], %[vqs1_r], 8 \n\t"
-
- "and %[vps1_l], %[vps1_l], %[HWM] \n\t"
- "shrl.ph %[vps1_r], %[vps1_r], 8 \n\t"
-
- "and %[vqs2_l], %[vqs2_l], %[HWM] \n\t"
- "shrl.ph %[vqs2_r], %[vqs2_r], 8 \n\t"
-
- "and %[vps2_l], %[vps2_l], %[HWM] \n\t"
- "shrl.ph %[vps2_r], %[vps2_r], 8 \n\t"
-
- "or %[vqs0_r], %[vqs0_l], %[vqs0_r] \n\t"
- "or %[vps0_r], %[vps0_l], %[vps0_r] \n\t"
- "or %[vqs1_r], %[vqs1_l], %[vqs1_r] \n\t"
- "or %[vps1_r], %[vps1_l], %[vps1_r] \n\t"
- "or %[vqs2_r], %[vqs2_l], %[vqs2_r] \n\t"
- "or %[vps2_r], %[vps2_l], %[vps2_r] \n\t"
-
- : [vps1_l] "+r" (vps1_l), [vps1_r] "+r" (vps1_r), [vqs1_l] "+r" (vqs1_l),
- [vqs1_r] "+r" (vqs1_r), [vps0_l] "+r" (vps0_l), [vps0_r] "+r" (vps0_r),
- [vqs0_l] "+r" (vqs0_l), [vqs0_r] "+r" (vqs0_r), [vqs2_l] "+r" (vqs2_l),
- [vqs2_r] "+r" (vqs2_r), [vps2_r] "+r" (vps2_r), [vps2_l] "+r" (vps2_l)
- : [HWM] "r" (HWM)
- );
-
- *ps0 = vps0_r ^ N128;
- *ps1 = vps1_r ^ N128;
- *ps2 = vps2_r ^ N128;
- *qs0 = vqs0_r ^ N128;
- *qs1 = vqs1_r ^ N128;
- *qs2 = vqs2_r ^ N128;
-}
-
-
-void vp8_mbloop_filter_horizontal_edge_mips
-(
- unsigned char *s,
- int p,
- unsigned int flimit,
- unsigned int limit,
- unsigned int thresh,
- int count
-)
-{
- int i;
- uint32_t mask, hev;
- uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
- unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
-
- mask = 0;
- hev = 0;
- i = 0;
- p1=0; p2=0; p3=0; p4=0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
-
- sm1 = s - (p<<2);
- s0 = s - p - p - p;
- s1 = s - p - p;
- s2 = s - p;
- s3 = s;
- s4 = s + p;
- s5 = s + p + p;
- s6 = s + p + p + p;
-
- /* prefetch data for load */
- prefetch_load_lf(s+p);
-
- /* apply filter on 4 pixesl at the same time */
- do
- {
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
-
- /* unpack processed 4x4 neighborhood
- * memory is 4 byte aligned
- */
- *((uint32_t*)s0) = p0;
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- *((uint32_t*)s5) = p5;
- }
- }
-
- sm1 += 4;
- s0 += 4;
- s1 += 4;
- s2 += 4;
- s3 += 4;
- s4 += 4;
- s5 += 4;
- s6 += 4;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
-
- /* unpack processed 4x4 neighborhood
- * memory is 4 byte aligned
- */
- *((uint32_t*)s0) = p0;
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- *((uint32_t*)s5) = p5;
- }
- }
-
- sm1 += 4;
- s0 += 4;
- s1 += 4;
- s2 += 4;
- s3 += 4;
- s4 += 4;
- s5 += 4;
- s6 += 4;
-
- i += 8;
- }
-
- while (i < count);
-}
-
-
-void vp8_mbloop_filter_uvhorizontal_edge_mips
-(
- unsigned char *s,
- int p,
- unsigned int flimit,
- unsigned int limit,
- unsigned int thresh,
- int count
-)
-{
- uint32_t mask, hev;
- uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
- unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
-
- mask = 0;
- hev = 0;
- p1=0; p2=0; p3=0; p4=0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
-
- sm1 = s - (p<<2);
- s0 = s - p - p - p;
- s1 = s - p - p;
- s2 = s - p;
- s3 = s;
- s4 = s + p;
- s5 = s + p + p;
- s6 = s + p + p + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- /* if mask == 0 do filtering is not needed */
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- if (mask) {
- /* filtering */
- vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
-
- /* unpack processed 4x4 neighborhood
- * memory is 4 byte aligned
- */
- *((uint32_t*)s0) = p0;
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- *((uint32_t*)s5) = p5;
- }
- }
-
- sm1 += 4;
- s0 += 4;
- s1 += 4;
- s2 += 4;
- s3 += 4;
- s4 += 4;
- s5 += 4;
- s6 += 4;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p1 = *((uint32_t*)(s1));
- p2 = *((uint32_t*)(s2));
- p3 = *((uint32_t*)(s3));
- p4 = *((uint32_t*)(s4));
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- pm1 = *((uint32_t*)(sm1));
- p0 = *((uint32_t*)(s0));
- p5 = *((uint32_t*)(s5));
- p6 = *((uint32_t*)(s6));
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
-
- /* unpack processed 4x4 neighborhood
- * memory is 4 byte aligned
- */
- *((uint32_t*)s0) = p0;
- *((uint32_t*)s1) = p1;
- *((uint32_t*)s2) = p2;
- *((uint32_t*)s3) = p3;
- *((uint32_t*)s4) = p4;
- *((uint32_t*)s5) = p5;
- }
- }
-}
-
-
-void vp8_mbloop_filter_vertical_edge_mips
-(
- unsigned char *s,
- int p,
- unsigned int flimit,
- unsigned int limit,
- unsigned int thresh,
- int count
-)
-{
-
- int i;
- uint32_t mask, hev;
- uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
- unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
- uint32_t temp, prim1, prim2, sec3, sec4, prim3, prim4;
-
- mask = 0;
- hev = 0;
- i = 0;
- pm1=0; p0=0; p1=0; p2=0; p3=0; p4=0; p5=0; p6=0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
-
- /* apply filter on 4 pixesl at the same time */
- do
- {
- s1 = s;
- s2 = s + p;
- s3 = s2 + p;
- s4 = s3 + p;
- s = s4 + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p2 = *((uint32_t*)(s1-4));
- p6 = *((uint32_t*)(s1));
- p1 = *((uint32_t*)(s2 - 4));
- p5 = *((uint32_t*)(s2));
- p0 = *((uint32_t*)(s3 - 4));
- p4 = *((uint32_t*)(s3));
- pm1 = *((uint32_t*)(s4 - 4));
- p3 = *((uint32_t*)(s4));
-
- /* transpose pm1, p0, p1, p2 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
- "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
- "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
- "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
-
- "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
- "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
- "append %[p1], %[sec3], 16 \n\t"
- "append %[pm1], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* transpose p3, p4, p5, p6 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
- "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
- "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
- "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
-
- "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
- "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
- "append %[p5], %[sec3], 16 \n\t"
- "append %[p3], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0)) ) {
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
-
- /* don't use transpose on output data
- * because memory isn't aligned
- */
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s4]) \n\t"
- "sb %[p4], 1(%[s4]) \n\t"
- "sb %[p3], 0(%[s4]) \n\t"
- "sb %[p2], -1(%[s4]) \n\t"
- "sb %[p1], -2(%[s4]) \n\t"
- "sb %[p0], -3(%[s4]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s3]) \n\t"
- "sb %[p4], 1(%[s3]) \n\t"
- "sb %[p3], 0(%[s3]) \n\t"
- "sb %[p2], -1(%[s3]) \n\t"
- "sb %[p1], -2(%[s3]) \n\t"
- "sb %[p0], -3(%[s3]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s2]) \n\t"
- "sb %[p4], 1(%[s2]) \n\t"
- "sb %[p3], 0(%[s2]) \n\t"
- "sb %[p2], -1(%[s2]) \n\t"
- "sb %[p1], -2(%[s2]) \n\t"
- "sb %[p0], -3(%[s2]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s1]) \n\t"
- "sb %[p4], 1(%[s1]) \n\t"
- "sb %[p3], 0(%[s1]) \n\t"
- "sb %[p2], -1(%[s1]) \n\t"
- "sb %[p1], -2(%[s1]) \n\t"
- "sb %[p0], -3(%[s1]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
- }
- }
- i+=4;
- }
-
- while (i < count);
-}
-
-
-void vp8_mbloop_filter_uvvertical_edge_mips
-(
- unsigned char *s,
- int p,
- unsigned int flimit,
- unsigned int limit,
- unsigned int thresh,
- int count
-)
-{
- uint32_t mask, hev;
- uint32_t pm1, p0, p1, p2, p3, p4, p5, p6;
- unsigned char *sm1, *s0, *s1, *s2, *s3, *s4, *s5, *s6;
- uint32_t temp, prim1, prim2, sec1, sec2, sec3, sec4, prim3, prim4;
-
- mask = 0;
- hev = 0;
- pm1=0; p0=0; p1=0; p2=0; p3=0; p4=0; p5=0; p6=0;
-
- /* loop filter designed to work using chars so that we can make maximum use
- * of 8 bit simd instructions.
- */
-
- /* apply filter on 4 pixesl at the same time */
-
- s1 = s;
- s2 = s + p;
- s3 = s2 + p;
- s4 = s3 + p;
-
- /* prefetch data for load */
- prefetch_load_lf(s + 2*p);
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p2 = *((uint32_t*)(s1-4));
- p6 = *((uint32_t*)(s1));
- p1 = *((uint32_t*)(s2 - 4));
- p5 = *((uint32_t*)(s2));
- p0 = *((uint32_t*)(s3 - 4));
- p4 = *((uint32_t*)(s3));
- pm1 = *((uint32_t*)(s4 - 4));
- p3 = *((uint32_t*)(s4));
-
- /* transpose pm1, p0, p1, p2 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
- "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
- "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
- "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
-
- "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
- "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
- "append %[p1], %[sec3], 16 \n\t"
- "append %[pm1], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* transpose p3, p4, p5, p6 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
- "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
- "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
- "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
-
- "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
- "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
- "append %[p5], %[sec3], 16 \n\t"
- "append %[p3], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6,
- thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
-
- /* don't use transpose on output data
- * because memory isn't aligned
- */
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s4]) \n\t"
- "sb %[p4], 1(%[s4]) \n\t"
- "sb %[p3], 0(%[s4]) \n\t"
- "sb %[p2], -1(%[s4]) \n\t"
- "sb %[p1], -2(%[s4]) \n\t"
- "sb %[p0], -3(%[s4]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s3]) \n\t"
- "sb %[p4], 1(%[s3]) \n\t"
- "sb %[p3], 0(%[s3]) \n\t"
- "sb %[p2], -1(%[s3]) \n\t"
- "sb %[p1], -2(%[s3]) \n\t"
- "sb %[p0], -3(%[s3]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s2]) \n\t"
- "sb %[p4], 1(%[s2]) \n\t"
- "sb %[p3], 0(%[s2]) \n\t"
- "sb %[p2], -1(%[s2]) \n\t"
- "sb %[p1], -2(%[s2]) \n\t"
- "sb %[p0], -3(%[s2]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s1]) \n\t"
- "sb %[p4], 1(%[s1]) \n\t"
- "sb %[p3], 0(%[s1]) \n\t"
- "sb %[p2], -1(%[s1]) \n\t"
- "sb %[p1], -2(%[s1]) \n\t"
- "sb %[p0], -3(%[s1]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
- }
- }
-
- s1 = s4 + p;
- s2 = s1 + p;
- s3 = s2 + p;
- s4 = s3 + p;
-
- /* load quad-byte vectors
- * memory is 4 byte aligned
- */
- p2 = *((uint32_t*)(s1-4));
- p6 = *((uint32_t*)(s1));
- p1 = *((uint32_t*)(s2 - 4));
- p5 = *((uint32_t*)(s2));
- p0 = *((uint32_t*)(s3 - 4));
- p4 = *((uint32_t*)(s3));
- pm1 = *((uint32_t*)(s4 - 4));
- p3 = *((uint32_t*)(s4));
-
- /* transpose pm1, p0, p1, p2 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p2], %[p1] \n\t"
- "precr.qb.ph %[prim2], %[p2], %[p1] \n\t"
- "precrq.qb.ph %[prim3], %[p0], %[pm1] \n\t"
- "precr.qb.ph %[prim4], %[p0], %[pm1] \n\t"
-
- "precrq.qb.ph %[p1], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[pm1], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p2], %[p1], %[sec3] \n\t"
- "precrq.ph.w %[p0], %[pm1], %[sec4] \n\t"
- "append %[p1], %[sec3], 16 \n\t"
- "append %[pm1], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0), [pm1] "+r" (pm1),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* transpose p3, p4, p5, p6 */
- __asm__ __volatile__ (
- "precrq.qb.ph %[prim1], %[p6], %[p5] \n\t"
- "precr.qb.ph %[prim2], %[p6], %[p5] \n\t"
- "precrq.qb.ph %[prim3], %[p4], %[p3] \n\t"
- "precr.qb.ph %[prim4], %[p4], %[p3] \n\t"
-
- "precrq.qb.ph %[p5], %[prim1], %[prim2] \n\t"
- "precr.qb.ph %[p3], %[prim1], %[prim2] \n\t"
- "precrq.qb.ph %[sec3], %[prim3], %[prim4] \n\t"
- "precr.qb.ph %[sec4], %[prim3], %[prim4] \n\t"
-
- "precrq.ph.w %[p6], %[p5], %[sec3] \n\t"
- "precrq.ph.w %[p4], %[p3], %[sec4] \n\t"
- "append %[p5], %[sec3], 16 \n\t"
- "append %[p3], %[sec4], 16 \n\t"
-
- : [prim1] "=&r" (prim1), [prim2] "=&r" (prim2),
- [prim3] "=&r" (prim3), [prim4] "=&r" (prim4),
- [p6] "+r" (p6), [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [sec3] "=&r" (sec3), [sec4] "=&r" (sec4)
- :
- );
-
- /* if (p1 - p4 == 0) and (p2 - p3 == 0)
- * mask will be zero and filtering is not needed
- */
- if (!(((p1-p4) == 0) && ((p2-p3) == 0))) {
-
- vp8_filter_mask_vec_mips(limit, flimit, p1, p2, pm1, p0, p3, p4, p5, p6, thresh, &hev, &mask);
-
- /* if mask == 0 do filtering is not needed */
- if (mask) {
- /* filtering */
- vp8_mbfilter_mips(mask, hev, &p0, &p1, &p2, &p3, &p4, &p5);
-
- /* don't use transpose on output data
- * because memory isn't aligned
- */
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s4]) \n\t"
- "sb %[p4], 1(%[s4]) \n\t"
- "sb %[p3], 0(%[s4]) \n\t"
- "sb %[p2], -1(%[s4]) \n\t"
- "sb %[p1], -2(%[s4]) \n\t"
- "sb %[p0], -3(%[s4]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s4] "r" (s4),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s3]) \n\t"
- "sb %[p4], 1(%[s3]) \n\t"
- "sb %[p3], 0(%[s3]) \n\t"
- "sb %[p2], -1(%[s3]) \n\t"
- "sb %[p1], -2(%[s3]) \n\t"
- "sb %[p0], -3(%[s3]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s3] "r" (s3),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s2]) \n\t"
- "sb %[p4], 1(%[s2]) \n\t"
- "sb %[p3], 0(%[s2]) \n\t"
- "sb %[p2], -1(%[s2]) \n\t"
- "sb %[p1], -2(%[s2]) \n\t"
- "sb %[p0], -3(%[s2]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s2] "r" (s2),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
-
- __asm__ __volatile__ (
- "srl %[p5], %[p5], 8 \n\t"
- "srl %[p4], %[p4], 8 \n\t"
- "srl %[p3], %[p3], 8 \n\t"
- "srl %[p2], %[p2], 8 \n\t"
- "srl %[p1], %[p1], 8 \n\t"
- "srl %[p0], %[p0], 8 \n\t"
- : [p5] "+r" (p5), [p4] "+r" (p4), [p3] "+r" (p3),
- [p2] "+r" (p2), [p1] "+r" (p1), [p0] "+r" (p0)
- :
- );
-
- __asm__ __volatile__ (
- "sb %[p5], 2(%[s1]) \n\t"
- "sb %[p4], 1(%[s1]) \n\t"
- "sb %[p3], 0(%[s1]) \n\t"
- "sb %[p2], -1(%[s1]) \n\t"
- "sb %[p1], -2(%[s1]) \n\t"
- "sb %[p0], -3(%[s1]) \n\t"
- :
- : [p5] "r" (p5), [p4] "r" (p4), [p3] "r" (p3), [s1] "r" (s1),
- [p2] "r" (p2), [p1] "r" (p1), [p0] "r" (p0)
- );
- }
- }
-}
diff --git a/vp8/common/mips/loopfilter_mips.c b/vp8/common/mips/loopfilter_mips.c
deleted file mode 100755
index b4ada4d..0000000
--- a/vp8/common/mips/loopfilter_mips.c
+++ /dev/null
@@ -1,165 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "loopfilter.h"
-#include "loopfilter_mips.h"
-#include "onyxc_int.h"
-
-extern prototype_loopfilter_mips(vp8_loop_filter_horizontal_edge_mips);
-extern prototype_loopfilter_mips(vp8_loop_filter_uvhorizontal_edge_mips);
-extern prototype_loopfilter_mips(vp8_loop_filter_vertical_edge_mips);
-extern prototype_loopfilter_mips(vp8_loop_filter_uvvertical_edge_mips);
-extern prototype_loopfilter_mips(vp8_mbloop_filter_horizontal_edge_mips);
-extern prototype_loopfilter_mips(vp8_mbloop_filter_uvhorizontal_edge_mips);
-extern prototype_loopfilter_mips(vp8_mbloop_filter_vertical_edge_mips);
-extern prototype_loopfilter_mips(vp8_mbloop_filter_uvvertical_edge_mips);
-
-/* Horizontal MB filtering */
-void vp8_loop_filter_mbh_mips(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- unsigned int thresh_vec, flimit_vec, limit_vec;
- unsigned char thresh, flimit, limit, flimit_temp;
-
- (void) simpler_lpf;
-
- /* use direct value instead pointers */
- limit = *(lfi->lim);
- flimit_temp = *(lfi->mbflim);
- thresh = *(lfi->thr);
- flimit = flimit_temp + flimit_temp + limit;
-
- /* create quad-byte */
- __asm__ __volatile__ (
- "replv.qb %[thresh_vec], %[thresh] \n\t"
- "replv.qb %[flimit_vec], %[flimit] \n\t"
- "replv.qb %[limit_vec], %[limit] \n\t"
- : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
- : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
- );
-
- vp8_mbloop_filter_horizontal_edge_mips(y_ptr, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
-
- if (u_ptr) {
- vp8_mbloop_filter_uvhorizontal_edge_mips(u_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
- }
-
- if (v_ptr) {
- vp8_mbloop_filter_uvhorizontal_edge_mips(v_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
- }
-}
-
-
-/* Vertical MB Filtering */
-void vp8_loop_filter_mbv_mips(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- unsigned int thresh_vec, flimit_vec, limit_vec;
- unsigned char thresh, flimit, limit, flimit_temp;
-
- (void) simpler_lpf;
-
- /* use direct value instead pointers */
- limit = *(lfi->lim);
- flimit_temp = *(lfi->mbflim);
- thresh = *(lfi->thr);
- flimit = flimit_temp + flimit_temp + limit;
-
- /* create quad-byte */
- __asm__ __volatile__ (
- "replv.qb %[thresh_vec], %[thresh] \n\t"
- "replv.qb %[flimit_vec], %[flimit] \n\t"
- "replv.qb %[limit_vec], %[limit] \n\t"
- : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
- : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
- );
-
- vp8_mbloop_filter_vertical_edge_mips(y_ptr, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
-
- if (u_ptr)
- vp8_mbloop_filter_uvvertical_edge_mips(u_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
-
- if (v_ptr)
- vp8_mbloop_filter_uvvertical_edge_mips(v_ptr, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
-}
-
-
-/* Horizontal B Filtering */
-void vp8_loop_filter_bh_mips(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- unsigned int thresh_vec, flimit_vec, limit_vec;
- unsigned char thresh, flimit, limit, flimit_temp;
-
- (void) simpler_lpf;
-
- /* use direct value instead pointers */
- limit = *(lfi->lim);
- flimit_temp = *(lfi->flim);
- thresh = *(lfi->thr);
- flimit = flimit_temp + flimit_temp + limit;
-
- /* create quad-byte */
- __asm__ __volatile__ (
- "replv.qb %[thresh_vec], %[thresh] \n\t"
- "replv.qb %[flimit_vec], %[flimit] \n\t"
- "replv.qb %[limit_vec], %[limit] \n\t"
- : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
- : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
- );
-
- vp8_loop_filter_horizontal_edge_mips(y_ptr + 4 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
- vp8_loop_filter_horizontal_edge_mips(y_ptr + 8 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
- vp8_loop_filter_horizontal_edge_mips(y_ptr + 12 * y_stride, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
-
- if (u_ptr)
- vp8_loop_filter_uvhorizontal_edge_mips(u_ptr + 4 * uv_stride, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
-
- if (v_ptr)
- vp8_loop_filter_uvhorizontal_edge_mips(v_ptr + 4 * uv_stride, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
-}
-
-
-/* Vertical B Filtering */
-void vp8_loop_filter_bv_mips(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- unsigned int thresh_vec, flimit_vec, limit_vec;
- unsigned char thresh, flimit, limit, flimit_temp;
-
- (void) simpler_lpf;
-
- /* use direct value instead pointers */
- limit = *(lfi->lim);
- flimit_temp = *(lfi->flim);
- thresh = *(lfi->thr);
- flimit = flimit_temp + flimit_temp + limit;
-
- /* create quad-byte */
- __asm__ __volatile__ (
- "replv.qb %[thresh_vec], %[thresh] \n\t"
- "replv.qb %[flimit_vec], %[flimit] \n\t"
- "replv.qb %[limit_vec], %[limit] \n\t"
- : [thresh_vec] "=&r" (thresh_vec), [flimit_vec] "=&r" (flimit_vec), [limit_vec] "=r" (limit_vec)
- : [thresh] "r" (thresh), [flimit] "r" (flimit), [limit] "r" (limit)
- );
-
- vp8_loop_filter_vertical_edge_mips(y_ptr + 4, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
- vp8_loop_filter_vertical_edge_mips(y_ptr + 8, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
- vp8_loop_filter_vertical_edge_mips(y_ptr + 12, y_stride, flimit_vec, limit_vec, thresh_vec, 16);
-
- if (u_ptr)
- vp8_loop_filter_uvvertical_edge_mips(u_ptr + 4, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
-
- if (v_ptr)
- vp8_loop_filter_uvvertical_edge_mips(v_ptr + 4, uv_stride, flimit_vec, limit_vec, thresh_vec, 0);
-}
\ No newline at end of file
diff --git a/vp8/common/mips/loopfilter_mips.h b/vp8/common/mips/loopfilter_mips.h
deleted file mode 100755
index 6d101c6..0000000
--- a/vp8/common/mips/loopfilter_mips.h
+++ /dev/null
@@ -1,39 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#ifndef LOOPFILTER_MIPS_H
-#define LOOPFILTER_MIPS_H
-
-extern prototype_loopfilter_block(vp8_loop_filter_mbv_mips);
-extern prototype_loopfilter_block(vp8_loop_filter_bv_mips);
-extern prototype_loopfilter_block(vp8_loop_filter_mbh_mips);
-extern prototype_loopfilter_block(vp8_loop_filter_bh_mips);
-
-#define prototype_loopfilter_mips(sym) \
- void sym(unsigned char *src, int pitch, const unsigned int flimit,\
- const unsigned int limit, const unsigned int thresh, int count)
-
-#undef vp8_lf_normal_mb_v
-#define vp8_lf_normal_mb_v vp8_loop_filter_mbv_mips
-
-#undef vp8_lf_normal_b_v
-#define vp8_lf_normal_b_v vp8_loop_filter_bv_mips
-
-#undef vp8_lf_normal_mb_h
-#define vp8_lf_normal_mb_h vp8_loop_filter_mbh_mips
-
-#undef vp8_lf_normal_b_h
-#define vp8_lf_normal_b_h vp8_loop_filter_bh_mips
-
-#endif
-#endif
diff --git a/vp8/common/mips/mips_systemdependent.c b/vp8/common/mips/mips_systemdependent.c
deleted file mode 100755
index df09064..0000000
--- a/vp8/common/mips/mips_systemdependent.c
+++ /dev/null
@@ -1,54 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "g_common.h"
-#include "subpixel.h"
-#include "loopfilter.h"
-#include "recon.h"
-#include "idct.h"
-#include "pragmas.h"
-#include "onyxc_int.h"
-
-void vp8_arch_mips_common_init(VP8_COMMON *ctx)
-{
-
-#if CONFIG_RUNTIME_CPU_DETECT
-
- VP8_COMMON_RTCD *rtcd = &ctx->rtcd;
-#ifdef MIPS_DSP_REV
-#if (MIPS_DSP_REV>=1)
- rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_mips;
- rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_mips;
- rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_mips;
- rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_mips;
-
- rtcd->recon.copy16x16 = vp8_copy_mem16x16_mips;
- rtcd->recon.copy8x8 = vp8_copy_mem8x8_mips;
- rtcd->recon.copy8x4 = vp8_copy_mem8x4_mips;
-
- rtcd->idct.idct1_scalar_add = vp8_dc_only_idct_add_mips;
- rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_mips;
- rtcd->idct.idct16 = vp8_short_idct4x4llm_mips;
- rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_mips;
-
-#if (MIPS_DSP_REV>=2)
- rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_mips;
- rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_mips;
- rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_mips;
- rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_mips;
-
-#endif
-#endif
-
-#endif
-#endif
-}
diff --git a/vp8/common/mips/recon_mips.h b/vp8/common/mips/recon_mips.h
deleted file mode 100755
index cad10a8..0000000
--- a/vp8/common/mips/recon_mips.h
+++ /dev/null
@@ -1,35 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#ifndef RECON_MIPS_H
-#define RECON_MIPS_H
-
-extern prototype_copy_block(vp8_copy_mem8x8_mips);
-extern prototype_copy_block(vp8_copy_mem8x4_mips);
-extern prototype_copy_block(vp8_copy_mem16x16_mips);
-
-#ifndef MUST_BE_ALIGNED
-#define MUST_BE_ALIGNED
-#endif
-
-#undef vp8_recon_copy8x8
-#define vp8_recon_copy8x8 vp8_copy_mem8x8_mips
-
-#undef vp8_recon_copy8x4
-#define vp8_recon_copy8x4 vp8_copy_mem8x4_mips
-
-#undef vp8_recon_copy16x16
-#define vp8_recon_copy16x16 vp8_copy_mem16x16_mips
-
-#endif
-#endif
\ No newline at end of file
diff --git a/vp8/common/mips/reconinter_mips.c b/vp8/common/mips/reconinter_mips.c
deleted file mode 100755
index 455c569..0000000
--- a/vp8/common/mips/reconinter_mips.c
+++ /dev/null
@@ -1,116 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_ports/config.h"
-#include "onyxc_int.h"
-
-
-inline void prefetch_load_int(unsigned char* src) {
- __asm__ __volatile__ (
- "pref 0, 0(%[src]) \n\t"
- :
- : [src] "r" (src)
- );
-}
-
-
-__inline void vp8_copy_mem16x16_mips(
- unsigned char * RESTRICT src,
- int src_stride,
- unsigned char * RESTRICT dst,
- int dst_stride)
-{
- int r;
- unsigned int a0, a1, a2, a3;
-
- for (r = 16; r--;)
- {
- /* load src data in cache memory */
- prefetch_load_int(src + src_stride);
-
- /* use unaligned memory load and store */
- __asm__ __volatile__ (
- "ulw %[a0], 0(%[src]) \n\t"
- "ulw %[a1], 4(%[src]) \n\t"
- "ulw %[a2], 8(%[src]) \n\t"
- "ulw %[a3], 12(%[src]) \n\t"
- "sw %[a0], 0(%[dst]) \n\t"
- "sw %[a1], 4(%[dst]) \n\t"
- "sw %[a2], 8(%[dst]) \n\t"
- "sw %[a3], 12(%[dst]) \n\t"
- : [a0] "=&r" (a0), [a1] "=&r" (a1),
- [a2] "=&r" (a2), [a3] "=&r" (a3)
- : [src] "r" (src), [dst] "r" (dst)
- );
-
- src += src_stride;
- dst += dst_stride;
- }
-}
-
-
-__inline void vp8_copy_mem8x8_mips(
- unsigned char * RESTRICT src,
- int src_stride,
- unsigned char * RESTRICT dst,
- int dst_stride)
-{
- int r;
- unsigned int a0, a1;
-
- /* load src data in cache memory */
- prefetch_load_int(src + src_stride);
-
- for (r = 8; r--;)
- {
- /* use unaligned memory load and store */
- __asm__ __volatile__ (
- "ulw %[a0], 0(%[src]) \n\t"
- "ulw %[a1], 4(%[src]) \n\t"
- "sw %[a0], 0(%[dst]) \n\t"
- "sw %[a1], 4(%[dst]) \n\t"
- : [a0] "=&r" (a0), [a1] "=&r" (a1)
- : [src] "r" (src), [dst] "r" (dst)
- );
-
- src += src_stride;
- dst += dst_stride;
- }
-}
-
-
-__inline void vp8_copy_mem8x4_mips(
- unsigned char * RESTRICT src,
- int src_stride,
- unsigned char * RESTRICT dst,
- int dst_stride)
-{
- int r;
- unsigned int a0, a1;
-
- /* load src data in cache memory */
- prefetch_load_int(src + src_stride);
-
- for (r = 4; r--;)
- {
- /* use unaligned memory load and store */
- __asm__ __volatile__ (
- "ulw %[a0], 0(%[src]) \n\t"
- "ulw %[a1], 4(%[src]) \n\t"
- "sw %[a0], 0(%[dst]) \n\t"
- "sw %[a1], 4(%[dst]) \n\t"
- : [a0] "=&r" (a0), [a1] "=&r" (a1)
- : [src] "r" (src), [dst] "r" (dst)
- );
-
- src += src_stride;
- dst += dst_stride;
- }
-}
diff --git a/vp8/common/mips/subpixel_mips.c b/vp8/common/mips/subpixel_mips.c
deleted file mode 100755
index 274b9cd..0000000
--- a/vp8/common/mips/subpixel_mips.c
+++ /dev/null
@@ -1,2786 +0,0 @@
-/*
- * Copyright (c) 2010 The VP8 project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "onyxc_int.h"
-
-#define CROP_WIDTH 256
-unsigned char ff_cropTbl[256 + 2 * CROP_WIDTH] = {0, };
-
-/* 6 tap filter coefficients */
-static const unsigned short sub_pel_filterss[8][3] =
-{
- { 0, 0, 0},
- { 0, 0x0601, 0x7b0c},
- { 0x0201, 0x0b08, 0x6c24},
- { 0, 0x0906, 0x5d32},
- { 0x0303, 0x1010, 0x4d4d},
- { 0, 0x0609, 0x325d},
- { 0x0102, 0x080b, 0x246c},
- { 0, 0x0106, 0x0c7b},
-};
-
-static const int sub_pel_filters_int[8][3] =
-{
- { 0, 0, 0},
- { 0x0000fffa, 0x007b000c, 0xffff0000},
- { 0x0002fff5, 0x006c0024, 0xfff80001},
- { 0x0000fff7, 0x005d0032, 0xfffa0000},
- { 0x0003fff0, 0x004d004d, 0xfff00003},
- { 0x0000fffa, 0x0032005d, 0xfff70000},
- { 0x0001fff8, 0x0024006c, 0xfff50002},
- { 0x0000ffff, 0x000c007b, 0xfffa0000},
-};
-
-static const int sub_pel_filters_inv[8][3] =
-{
- { 0, 0, 0},
- { 0xfffa0000, 0x000c007b, 0x0000ffff},
- { 0xfff50002, 0x0024006c, 0x0001fff8},
- { 0xfff70000, 0x0032005d, 0x0000fffa},
- { 0xfff00003, 0x004d004d, 0x0003fff0},
- { 0xfffa0000, 0x005d0032, 0x0000fff7},
- { 0xfff80001, 0x006c0024, 0x0002fff5},
- { 0xffff0000, 0x007b000c, 0x0000fffa},
-};
-
-/* 4 tap filter coefficients */
-static const int sub_pel_filters_int_tap_4[8][2] =
-{
- { 0, 0},
- { 0xfffa007b, 0x000cffff},
- { 0, 0},
- { 0xfff7005d, 0x0032fffa},
- { 0, 0},
- { 0xfffa0032, 0x005dfff7},
- { 0, 0},
- { 0xffff000c, 0x007bfffa},
-};
-
-static const int sub_pel_filters_inv_tap_4[8][2] =
-{
- { 0, 0},
- { 0x007bfffa, 0xffff000c},
- { 0, 0},
- { 0x005dfff7, 0xfffa0032},
- { 0, 0},
- { 0x0032fffa, 0xfff7005d},
- { 0, 0},
- { 0x000cffff, 0xfffa007b},
-};
-
-
-inline void prefetch_load(unsigned char* src) {
- __asm__ __volatile__ (
- "pref 0, 0(%[src]) \n\t"
- :
- : [src] "r" (src)
- );
-}
-
-
-inline void prefetch_store(unsigned char* dst) {
- __asm__ __volatile__ (
- "pref 1, 0(%[dst]) \n\t"
- :
- : [dst] "r" (dst)
- );
-}
-
-
-void dsputil_static_init(void)
-{
- int i;
-
- for(i=0;i<256;i++) ff_cropTbl[i + CROP_WIDTH] = i;
- for(i=0;i<CROP_WIDTH;i++) {
- ff_cropTbl[i] = 0;
- ff_cropTbl[i + CROP_WIDTH + 256] = 255;
- }
-}
-
-
-void vp8_filter_block2d_first_pass_4
-(
- unsigned char * RESTRICT src_ptr,
- unsigned char * RESTRICT dst_ptr,
- unsigned int src_pixels_per_line,
- unsigned int output_height,
- int xoffset,
- int pitch
-)
-{
- unsigned int i;
- int Temp1, Temp2, Temp3, Temp4;
-
- unsigned int vector4a = 64;
- int vector1b, vector2b, vector3b;
- unsigned int tp1, tp2, tn1, tn2;
- unsigned int p1, p2, p3;
- unsigned int n1, n2, n3;
- unsigned char *cm = ff_cropTbl + CROP_WIDTH;
-
- vector3b = sub_pel_filters_inv[xoffset][2];
-
- /* if (xoffset == 0) we don'n need any filtering */
- if (vector3b == 0) {
- for (i = 0; i < output_height; i++)
- {
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + src_pixels_per_line);
-
- dst_ptr[0] = src_ptr[0];
- dst_ptr[1] = src_ptr[1];
- dst_ptr[2] = src_ptr[2];
- dst_ptr[3] = src_ptr[3];
-
- /* next row... */
- src_ptr += src_pixels_per_line;
- dst_ptr += 4;
- }
- }
- else {
- if (vector3b>65536) {
- /* 6 tap filter */
-
- vector1b = sub_pel_filters_inv[xoffset][0];
- vector2b = sub_pel_filters_inv[xoffset][1];
-
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + src_pixels_per_line);
-
- for (i = output_height; i--;)
- {
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "ulw %[tp1], -2(%[src_ptr]) \n\t"
- "ulw %[tp2], 2(%[src_ptr]) \n\t"
-
- /* even 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p1], %[tp1] \n\t"
- "preceu.ph.qbl %[p2], %[tp1] \n\t"
- "preceu.ph.qbr %[p3], %[tp2] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[p3], %[vector3b] \n\t"
-
- /* even 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[p1], %[tp2] \n\t"
- "balign %[tp2], %[tp1], 3 \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[p1], %[vector3b] \n\t"
-
- /* odd 1. pixel */
- "ulw %[tn2], 3(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n1], %[tp2] \n\t"
- "preceu.ph.qbl %[n2], %[tp2] \n\t"
- "preceu.ph.qbr %[n3], %[tn2] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[n3], %[vector3b] \n\t"
-
- /* even 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[n1], %[tn2] \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[n1], %[vector3b] \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- /* clamp */
- "lbux %[tp1], %[Temp1](%[cm]) \n\t"
- "lbux %[tn1], %[Temp2](%[cm]) \n\t"
- "lbux %[tp2], %[Temp3](%[cm]) \n\t"
- "lbux %[n2], %[Temp4](%[cm]) \n\t"
-
- /* store bytes */
- "sb %[tp1], 0(%[dst_ptr]) \n\t"
- "sb %[tn1], 1(%[dst_ptr]) \n\t"
- "sb %[tp2], 2(%[dst_ptr]) \n\t"
- "sb %[n2], 3(%[dst_ptr]) \n\t"
-
- : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn1] "=&r" (tn1),
- [tn2] "=&r" (tn2), [p1] "=&r" (p1), [p2] "=&r" (p2),
- [p3] "=&r" (p3), [n1] "=&r" (n1), [n2] "=&r" (n2),
- [n3] "=&r" (n3), [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector4a] "r" (vector4a), [cm] "r" (cm), [dst_ptr] "r" (dst_ptr),
- [vector3b] "r" (vector3b), [src_ptr] "r" (src_ptr)
- );
-
- /* Next row... */
- src_ptr += src_pixels_per_line;
- dst_ptr += pitch;
- }
- }
- else {
- /* 4 tap filter */
-
- vector1b = sub_pel_filters_inv_tap_4[xoffset][0];
- vector2b = sub_pel_filters_inv_tap_4[xoffset][1];
-
- for (i = output_height; i--;)
- {
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "ulw %[tp1], -1(%[src_ptr]) \n\t"
- "ulw %[tp2], 3(%[src_ptr]) \n\t"
-
- /* even 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p1], %[tp1] \n\t"
- "preceu.ph.qbl %[p2], %[tp1] \n\t"
- "preceu.ph.qbr %[p3], %[tp2] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
-
- /* even 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
-
- /* odd 1. pixel */
- "srl %[tn1], %[tp2], 8 \n\t"
- "balign %[tp2], %[tp1], 3 \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n1], %[tp2] \n\t"
- "preceu.ph.qbl %[n2], %[tp2] \n\t"
- "preceu.ph.qbr %[n3], %[tn1] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
-
- /* odd 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- /* clamp and store results */
- "lbux %[tp1], %[Temp1](%[cm]) \n\t"
- "lbux %[tn1], %[Temp2](%[cm]) \n\t"
- "lbux %[tp2], %[Temp3](%[cm]) \n\t"
- "sb %[tp1], 0(%[dst_ptr]) \n\t"
- "sb %[tn1], 1(%[dst_ptr]) \n\t"
- "lbux %[n2], %[Temp4](%[cm]) \n\t"
- "sb %[tp2], 2(%[dst_ptr]) \n\t"
- "sb %[n2], 3(%[dst_ptr]) \n\t"
-
- : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn1] "=&r" (tn1),
- [p1] "=&r" (p1), [p2] "=&r" (p2), [p3] "=&r" (p3),
- [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector4a] "r" (vector4a), [cm] "r" (cm), [dst_ptr] "r" (dst_ptr),
- [src_ptr] "r" (src_ptr)
- );
- /* Next row... */
- src_ptr += src_pixels_per_line;
- dst_ptr += pitch;
- }
- }
- }
-}
-
-
-void vp8_filter_block2d_first_pass_8_all
-(
- unsigned char * RESTRICT src_ptr,
- unsigned char * RESTRICT dst_ptr,
- unsigned int src_pixels_per_line,
- unsigned int output_height,
- int xoffset,
- int pitch
-)
-{
- unsigned int i;
- int Temp1, Temp2, Temp3, Temp4;
-
- unsigned int vector1a, vector2a, vector3a;
- unsigned int vector4a=64;
- unsigned int vector1b, vector2b, vector3b;
- unsigned int tp1, tp2, tn1, tn2;
- unsigned int p1, p2, p3, p4;
- unsigned int n1, n2, n3, n4;
-
- unsigned char *cm = ff_cropTbl + CROP_WIDTH;
-
- /* if (xoffset == 0) we don'n need any filtering */
- if (xoffset == 0) {
- for (i = 0; i < output_height; i++)
- {
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + src_pixels_per_line);
-
- dst_ptr[0] = src_ptr[0];
- dst_ptr[1] = src_ptr[1];
- dst_ptr[2] = src_ptr[2];
- dst_ptr[3] = src_ptr[3];
- dst_ptr[4] = src_ptr[4];
- dst_ptr[5] = src_ptr[5];
- dst_ptr[6] = src_ptr[6];
- dst_ptr[7] = src_ptr[7];
-
- /* next row... */
- src_ptr += src_pixels_per_line;
- dst_ptr += 8;
- }
- }
- else {
- vector3b = sub_pel_filters_inv[xoffset][2];
-
- if (vector3b>65536) {
- /* 6 tap filter */
-
- vector1b = sub_pel_filters_inv[xoffset][0];
- vector2b = sub_pel_filters_inv[xoffset][1];
-
- for (i = output_height; i--;)
- {
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + src_pixels_per_line);
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "ulw %[tp1], -2(%[src_ptr]) \n\t"
- "ulw %[tp2], 2(%[src_ptr]) \n\t"
-
- /* even 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p1], %[tp1] \n\t"
- "preceu.ph.qbl %[p2], %[tp1] \n\t"
- "preceu.ph.qbr %[p3], %[tp2] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[p3], %[vector3b] \n\t"
-
- /* even 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[p1], %[tp2] \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[p1], %[vector3b] \n\t"
-
- "balign %[tp2], %[tp1], 3 \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
- "ulw %[tn2], 3(%[src_ptr]) \n\t"
-
- /* odd 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n1], %[tp2] \n\t"
- "preceu.ph.qbl %[n2], %[tp2] \n\t"
- "preceu.ph.qbr %[n3], %[tn2] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[n3], %[vector3b] \n\t"
-
- /* odd 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[n1], %[tn2] \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[n1], %[vector3b] \n\t"
- "ulw %[tp1], 6(%[src_ptr]) \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p2], %[tp1] \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn2] "=&r" (tn2),
- [p1] "=&r" (p1), [p2] "=&r" (p2), [p3] "=&r" (p3),
- [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector4a] "r" (vector4a), [vector3b] "r" (vector3b),
- [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- dst_ptr[0] = cm[Temp1];
- dst_ptr[1] = cm[Temp2];
- dst_ptr[2] = cm[Temp3];
- dst_ptr[3] = cm[Temp4];
-
- /* next 4 pixels */
- __asm__ __volatile__ (
- /* even 3. pixel */
- "dpa.w.ph $ac3, %[p3], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector3b] \n\t"
-
- /* even 4. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[p4], %[tp1] \n\t"
- "dpa.w.ph $ac2, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[p4], %[vector3b] \n\t"
-
- "ulw %[tn1], 7(%[src_ptr]) \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
-
- /* odd 3. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n2], %[tn1] \n\t"
- "dpa.w.ph $ac3, %[n3], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector3b] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
-
- /* odd 4. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[n4], %[tn1] \n\t"
- "dpa.w.ph $ac2, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[n4], %[vector3b] \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- : [tn1] "=&r" (tn1), [n2] "=&r" (n2),
- [p4] "=&r" (p4), [n4] "=&r" (n4),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
- : [tp1] "r" (tp1), [vector1b] "r" (vector1b), [p2] "r" (p2),
- [vector2b] "r" (vector2b), [n1] "r" (n1), [p1] "r" (p1),
- [vector4a] "r" (vector4a), [vector3b] "r" (vector3b),
- [p3] "r" (p3), [n3] "r" (n3), [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- dst_ptr[4] = cm[Temp1];
- dst_ptr[5] = cm[Temp2];
- dst_ptr[6] = cm[Temp3];
- dst_ptr[7] = cm[Temp4];
-
- src_ptr += src_pixels_per_line;
- dst_ptr += pitch;
- }
- }
- else {
- /* 4 tap filter */
-
- vector1b = sub_pel_filters_inv_tap_4[xoffset][0];
- vector2b = sub_pel_filters_inv_tap_4[xoffset][1];
-
- for (i = output_height; i--;)
- {
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + src_pixels_per_line);
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "ulw %[tp1], -1(%[src_ptr]) \n\t"
-
- /* even 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p1], %[tp1] \n\t"
- "preceu.ph.qbl %[p2], %[tp1] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
-
- "ulw %[tp2], 3(%[src_ptr]) \n\t"
-
- /* even 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbr %[p3], %[tp2] \n\t"
- "preceu.ph.qbl %[p4], %[tp2] \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
-
- "balign %[tp2], %[tp1], 3 \n\t"
-
- /* odd 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n1], %[tp2] \n\t"
- "preceu.ph.qbl %[n2], %[tp2] \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
-
- "ulw %[tn2], 4(%[src_ptr]) \n\t"
-
- /* odd 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbr %[n3], %[tn2] \n\t"
- "preceu.ph.qbl %[n4], %[tn2] \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
- "ulw %[tp1], 7(%[src_ptr]) \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2),
- [tn2] "=&r" (tn2), [p1] "=&r" (p1), [p2] "=&r" (p2),
- [p3] "=&r" (p3), [p4] "=&r" (p4), [n1] "=&r" (n1),
- [n2] "=&r" (n2), [n3] "=&r" (n3), [n4] "=&r" (n4),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- dst_ptr[0] = cm[Temp1];
- dst_ptr[1] = cm[Temp2];
- dst_ptr[2] = cm[Temp3];
- dst_ptr[3] = cm[Temp4];
-
- /* next 4 pixels */
- __asm__ __volatile__ (
- /* even 3. pixel */
- "dpa.w.ph $ac3, %[p3], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p4], %[vector2b] \n\t"
-
- /* even 4. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbr %[p2], %[tp1] \n\t"
- "dpa.w.ph $ac2, %[p4], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector2b] \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
-
- /* odd 3. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "dpa.w.ph $ac3, %[n3], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n4], %[vector2b] \n\t"
- "ulw %[tn1], 8(%[src_ptr]) \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
-
- /* odd 4. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbr %[n2], %[tn1] \n\t"
- "dpa.w.ph $ac2, %[n4], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector2b] \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- : [tn1] "=&r" (tn1), [p2] "=&r" (p2), [n2] "=&r" (n2),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
- : [tp1] "r" (tp1), [p3] "r" (p3), [p4] "r" (p4),
- [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr),
- [n3] "r" (n3), [n4] "r" (n4)
- );
-
- /* clamp and store results */
- dst_ptr[4] = cm[Temp1];
- dst_ptr[5] = cm[Temp2];
- dst_ptr[6] = cm[Temp3];
- dst_ptr[7] = cm[Temp4];
-
- /* next row... */
- src_ptr += src_pixels_per_line;
- dst_ptr += pitch;
- }
- }
- }
-}
-
-
-void vp8_filter_block2d_first_pass16_6tap
-(
- unsigned char * RESTRICT src_ptr,
- unsigned char * RESTRICT dst_ptr,
- unsigned int src_pixels_per_line,
- unsigned int output_height,
- int xoffset,
- int pitch
-)
-{
- unsigned int i;
- int Temp1, Temp2, Temp3, Temp4;
-
- unsigned int vector4a;
- unsigned int vector1b, vector2b, vector3b;
- unsigned int tp1, tp2, tn1, tn2;
- unsigned int p1, p2, p3, p4;
- unsigned int n1, n2, n3, n4;
- unsigned char *cm = ff_cropTbl + CROP_WIDTH;
-
- vector1b = sub_pel_filters_inv[xoffset][0];
- vector2b = sub_pel_filters_inv[xoffset][1];
- vector3b = sub_pel_filters_inv[xoffset][2];
- vector4a = 64;
-
- for (i = output_height; i--;)
- {
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + src_pixels_per_line);
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "ulw %[tp1], -2(%[src_ptr]) \n\t"
- "ulw %[tp2], 2(%[src_ptr]) \n\t"
-
- /* even 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p1], %[tp1] \n\t"
- "preceu.ph.qbl %[p2], %[tp1] \n\t"
- "preceu.ph.qbr %[p3], %[tp2] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[p3], %[vector3b] \n\t"
-
- /* even 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[p1], %[tp2] \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[p1], %[vector3b] \n\t"
-
- "balign %[tp2], %[tp1], 3 \n\t"
- "ulw %[tn2], 3(%[src_ptr]) \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
-
- /* odd 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n1], %[tp2] \n\t"
- "preceu.ph.qbl %[n2], %[tp2] \n\t"
- "preceu.ph.qbr %[n3], %[tn2] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[n3], %[vector3b] \n\t"
-
- /* odd 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[n1], %[tn2] \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[n1], %[vector3b] \n\t"
- "ulw %[tp1], 6(%[src_ptr]) \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p2], %[tp1] \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn2] "=&r" (tn2),
- [p1] "=&r" (p1), [p2] "=&r" (p2), [p3] "=&r" (p3),
- [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector4a] "r" (vector4a), [vector3b] "r" (vector3b),
- [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- dst_ptr[0] = cm[Temp1];
- dst_ptr[1] = cm[Temp2];
- dst_ptr[2] = cm[Temp3];
- dst_ptr[3] = cm[Temp4];
-
- /* next 4 pixels */
- __asm__ __volatile__ (
- /* even 3. pixel */
- "dpa.w.ph $ac3, %[p3], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector3b] \n\t"
-
- /* even 4. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[p4], %[tp1] \n\t"
- "dpa.w.ph $ac2, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[p4], %[vector3b] \n\t"
- "ulw %[tn1], 7(%[src_ptr]) \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
-
- /* odd 3. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n2], %[tn1] \n\t"
- "dpa.w.ph $ac3, %[n3], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector3b] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
-
- /* odd 4. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[n4], %[tn1] \n\t"
- "dpa.w.ph $ac2, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[n4], %[vector3b] \n\t"
- "ulw %[tp2], 10(%[src_ptr]) \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p1], %[tp2] \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- : [tn1] "=&r" (tn1), [tp2] "=&r" (tp2), [n2] "=&r" (n2),
- [p4] "=&r" (p4), [n4] "=&r" (n4),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [tp1] "r" (tp1), [n1] "r" (n1), [p1] "r" (p1),
- [vector4a] "r" (vector4a), [p2] "r" (p2), [vector3b] "r" (vector3b),
- [p3] "r" (p3), [n3] "r" (n3), [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- dst_ptr[4] = cm[Temp1];
- dst_ptr[5] = cm[Temp2];
- dst_ptr[6] = cm[Temp3];
- dst_ptr[7] = cm[Temp4];
-
- /* next 4 pixels */
- __asm__ __volatile__ (
- /* even 5. pixel */
- "dpa.w.ph $ac3, %[p2], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p4], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector3b] \n\t"
-
- /* even 6. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[p3], %[tp2] \n\t"
- "dpa.w.ph $ac2, %[p4], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p1], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector3b] \n\t"
-
- "ulw %[tn1], 11(%[src_ptr]) \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
-
- /* odd 5. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n1], %[tn1] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n4], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector3b] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
-
- /* odd 6. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[n3], %[tn1] \n\t"
- "dpa.w.ph $ac2, %[n4], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n1], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector3b] \n\t"
- "ulw %[tp1], 14(%[src_ptr]) \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[p4], %[tp1] \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- : [tn1] "=&r" (tn1), [tp1] "=&r" (tp1),
- [n1] "=&r" (n1), [p3] "=&r" (p3), [n3] "=&r" (n3),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [tp2] "r" (tp2), [p2] "r" (p2), [n2] "r" (n2),
- [p4] "r" (p4), [n4] "r" (n4), [p1] "r" (p1), [src_ptr] "r" (src_ptr),
- [vector4a] "r" (vector4a), [vector3b] "r" (vector3b)
- );
-
- /* clamp and store results */
- dst_ptr[8] = cm[Temp1];
- dst_ptr[9] = cm[Temp2];
- dst_ptr[10] = cm[Temp3];
- dst_ptr[11] = cm[Temp4];
-
- /* next 4 pixels */
- __asm__ __volatile__ (
- /* even 7. pixel */
- "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p3], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[p4], %[vector3b] \n\t"
-
- /* even 8. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[p2], %[tp1] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p4], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector3b] \n\t"
- "ulw %[tn1], 15(%[src_ptr]) \n\t"
- "extp %[Temp1], $ac3, 9 \n\t"
-
- /* odd 7. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "preceu.ph.qbr %[n4], %[tn1] \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n3], %[vector2b] \n\t"
- "dpa.w.ph $ac3, %[n4], %[vector3b] \n\t"
- "extp %[Temp3], $ac2, 9 \n\t"
-
- /* odd 8. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "preceu.ph.qbl %[n2], %[tn1] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n4], %[vector2b] \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector3b] \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
- "extp %[Temp4], $ac2, 9 \n\t"
-
- /* clamp and store results */
- "lbux %[tp1], %[Temp1](%[cm]) \n\t"
- "lbux %[tn1], %[Temp2](%[cm]) \n\t"
- "lbux %[p2], %[Temp3](%[cm]) \n\t"
- "sb %[tp1], 12(%[dst_ptr]) \n\t"
- "sb %[tn1], 13(%[dst_ptr]) \n\t"
- "lbux %[n2], %[Temp4](%[cm]) \n\t"
- "sb %[p2], 14(%[dst_ptr]) \n\t"
- "sb %[n2], 15(%[dst_ptr]) \n\t"
-
- : [tn1] "=&r" (tn1), [p2] "=&r" (p2), [n2] "=&r" (n2), [n4] "=&r" (n4),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [tp1] "r" (tp1), [p4] "r" (p4), [n1] "r" (n1), [p1] "r" (p1),
- [vector4a] "r" (vector4a), [vector3b] "r" (vector3b), [p3] "r" (p3),
- [n3] "r" (n3), [src_ptr] "r" (src_ptr),
- [cm] "r" (cm), [dst_ptr] "r" (dst_ptr)
- );
-
- src_ptr += src_pixels_per_line;
- dst_ptr += pitch;
- }
-}
-
-
-void vp8_filter_block2d_first_pass16_0
-(
- unsigned char * RESTRICT src_ptr,
- unsigned char * RESTRICT output_ptr,
- unsigned int src_pixels_per_line
-)
-{
- int Temp1, Temp2, Temp3, Temp4;
- int i;
-
- /* prefetch src_ptr data to cache memory */
- prefetch_store(output_ptr + 32);
-
- /* copy memory from src buffer to dst buffer */
- for (i = 0; i < 7; i++)
- {
-
- __asm__ __volatile__ (
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "ulw %[Temp3], 8(%[src_ptr]) \n\t"
- "ulw %[Temp4], 12(%[src_ptr]) \n\t"
- "sw %[Temp1], 0(%[output_ptr]) \n\t"
- "sw %[Temp2], 4(%[output_ptr]) \n\t"
- "sw %[Temp3], 8(%[output_ptr]) \n\t"
- "sw %[Temp4], 12(%[output_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [Temp3] "=&r" (Temp3),
- [Temp4] "=&r" (Temp4), [src_ptr] "+r" (src_ptr)
- : [src_pixels_per_line] "r" (src_pixels_per_line),
- [output_ptr] "r" (output_ptr)
- );
-
- __asm__ __volatile__ (
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "ulw %[Temp3], 8(%[src_ptr]) \n\t"
- "ulw %[Temp4], 12(%[src_ptr]) \n\t"
- "sw %[Temp1], 16(%[output_ptr]) \n\t"
- "sw %[Temp2], 20(%[output_ptr]) \n\t"
- "sw %[Temp3], 24(%[output_ptr]) \n\t"
- "sw %[Temp4], 28(%[output_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [Temp3] "=&r" (Temp3),
- [Temp4] "=&r" (Temp4), [src_ptr] "+r" (src_ptr)
- : [src_pixels_per_line] "r" (src_pixels_per_line),
- [output_ptr] "r" (output_ptr)
- );
-
- __asm__ __volatile__ (
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "ulw %[Temp3], 8(%[src_ptr]) \n\t"
- "ulw %[Temp4], 12(%[src_ptr]) \n\t"
- "sw %[Temp1], 32(%[output_ptr]) \n\t"
- "sw %[Temp2], 36(%[output_ptr]) \n\t"
- "sw %[Temp3], 40(%[output_ptr]) \n\t"
- "sw %[Temp4], 44(%[output_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [Temp3] "=&r" (Temp3),
- [Temp4] "=&r" (Temp4), [src_ptr] "+r" (src_ptr)
- : [src_pixels_per_line] "r" (src_pixels_per_line),
- [output_ptr] "r" (output_ptr)
- );
-
- output_ptr += 48;
- }
-}
-
-
-void vp8_filter_block2d_first_pass16_4tap
-(
- unsigned char * RESTRICT src_ptr,
- unsigned char * RESTRICT output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int output_width,
- unsigned int output_height,
- int xoffset,
- int yoffset,
- unsigned char * RESTRICT dst_ptr,
- int pitch
-)
-{
- unsigned int i, j;
- int Temp1, Temp2, Temp3, Temp4;
-
- unsigned int vector4a;
- int vector1b, vector2b;
- unsigned int tp1, tp2, tp3, tn1;
- unsigned int p1, p2, p3;
- unsigned int n1, n2, n3;
- unsigned char *cm = ff_cropTbl + CROP_WIDTH;
-
- vector4a = 64;
-
- vector1b = sub_pel_filters_inv_tap_4[xoffset][0];
- vector2b = sub_pel_filters_inv_tap_4[xoffset][1];
-
- /* if (yoffset == 0) don't need temp buffer, data will be stored in dst_ptr */
- if (yoffset == 0) {
- output_height -= 5;
- src_ptr += (src_pixels_per_line + src_pixels_per_line);
-
- for (i = output_height; i--;)
- {
- __asm__ __volatile__ (
- "ulw %[tp3], -1(%[src_ptr]) \n\t"
- : [tp3] "=&r" (tp3)
- : [src_ptr] "r" (src_ptr)
- );
-
- /* processing 4 adjacent pixels */
- for (j = 0; j < 16; j += 4)
- {
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "ulw %[tp2], 3(%[src_ptr]) \n\t"
- "move %[tp1], %[tp3] \n\t"
-
- /* even 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "mthi $0, $ac3 \n\t"
- "move %[tp3], %[tp2] \n\t"
- "preceu.ph.qbr %[p1], %[tp1] \n\t"
- "preceu.ph.qbl %[p2], %[tp1] \n\t"
- "preceu.ph.qbr %[p3], %[tp2] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
-
- /* even 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "mthi $0, $ac2 \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
- "extr.w %[Temp1], $ac3, 7 \n\t"
-
- /* odd 1. pixel */
- "ulw %[tn1], 4(%[src_ptr]) \n\t"
- "balign %[tp2], %[tp1], 3 \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "mthi $0, $ac3 \n\t"
- "preceu.ph.qbr %[n1], %[tp2] \n\t"
- "preceu.ph.qbl %[n2], %[tp2] \n\t"
- "preceu.ph.qbr %[n3], %[tn1] \n\t"
- "extr.w %[Temp3], $ac2, 7 \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
-
- /* odd 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "mthi $0, $ac2 \n\t"
- "extr.w %[Temp2], $ac3, 7 \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
- "extr.w %[Temp4], $ac2, 7 \n\t"
-
- /* clamp and store results */
- "lbux %[tp1], %[Temp1](%[cm]) \n\t"
- "lbux %[tn1], %[Temp2](%[cm]) \n\t"
- "lbux %[tp2], %[Temp3](%[cm]) \n\t"
- "sb %[tp1], 0(%[dst_ptr]) \n\t"
- "sb %[tn1], 1(%[dst_ptr]) \n\t"
- "lbux %[n2], %[Temp4](%[cm]) \n\t"
- "sb %[tp2], 2(%[dst_ptr]) \n\t"
- "sb %[n2], 3(%[dst_ptr]) \n\t"
-
- : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tp3] "=&r" (tp3),
- [tn1] "=&r" (tn1), [p1] "=&r" (p1), [p2] "=&r" (p2),
- [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [p3] "=&r" (p3),
- [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector4a] "r" (vector4a), [cm] "r" (cm), [dst_ptr] "r" (dst_ptr),
- [src_ptr] "r" (src_ptr)
- );
-
- src_ptr += 4;
- }
-
- /* Next row... */
- src_ptr += src_pixels_per_line - 16;
- dst_ptr += pitch;
- }
- }
- else {
- for (i = output_height; i--;)
- {
- /* processing 4 adjacent pixels */
- for (j = 0; j < 16; j += 4)
- {
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "ulw %[tp1], -1(%[src_ptr]) \n\t"
- "ulw %[tp2], 3(%[src_ptr]) \n\t"
-
- /* even 1. pixel */
- "mtlo %[vector4a], $ac3 \n\t"
- "mthi $0, $ac3 \n\t"
- "preceu.ph.qbr %[p1], %[tp1] \n\t"
- "preceu.ph.qbl %[p2], %[tp1] \n\t"
- "preceu.ph.qbr %[p3], %[tp2] \n\t"
- "dpa.w.ph $ac3, %[p1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[p2], %[vector2b] \n\t"
-
- /* even 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "mthi $0, $ac2 \n\t"
- "dpa.w.ph $ac2, %[p2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[p3], %[vector2b] \n\t"
- "extr.w %[Temp1], $ac3, 7 \n\t"
-
- /* odd 1. pixel */
- "ulw %[tn1], 4(%[src_ptr]) \n\t"
- "balign %[tp2], %[tp1], 3 \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "mthi $0, $ac3 \n\t"
- "preceu.ph.qbr %[n1], %[tp2] \n\t"
- "preceu.ph.qbl %[n2], %[tp2] \n\t"
- "preceu.ph.qbr %[n3], %[tn1] \n\t"
- "extr.w %[Temp3], $ac2, 7 \n\t"
- "dpa.w.ph $ac3, %[n1], %[vector1b] \n\t"
- "dpa.w.ph $ac3, %[n2], %[vector2b] \n\t"
-
- /* odd 2. pixel */
- "mtlo %[vector4a], $ac2 \n\t"
- "mthi $0, $ac2 \n\t"
- "extr.w %[Temp2], $ac3, 7 \n\t"
- "dpa.w.ph $ac2, %[n2], %[vector1b] \n\t"
- "dpa.w.ph $ac2, %[n3], %[vector2b] \n\t"
- "extr.w %[Temp4], $ac2, 7 \n\t"
-
- /* clamp and store results */
- "lbux %[tp1], %[Temp1](%[cm]) \n\t"
- "lbux %[tn1], %[Temp2](%[cm]) \n\t"
- "lbux %[tp2], %[Temp3](%[cm]) \n\t"
- "sb %[tp1], 0(%[output_ptr]) \n\t"
- "sb %[tn1], 1(%[output_ptr]) \n\t"
- "lbux %[n2], %[Temp4](%[cm]) \n\t"
- "sb %[tp2], 2(%[output_ptr]) \n\t"
- "sb %[n2], 3(%[output_ptr]) \n\t"
-
- : [tp1] "=&r" (tp1), [tp2] "=&r" (tp2), [tn1] "=&r" (tn1),
- [p1] "=&r" (p1), [p2] "=&r" (p2), [p3] "=&r" (p3),
- [n1] "=&r" (n1), [n2] "=&r" (n2), [n3] "=&r" (n3),
- [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector4a] "r" (vector4a), [cm] "r" (cm),
- [output_ptr] "r" (output_ptr), [src_ptr] "r" (src_ptr)
- );
-
- src_ptr += 4;
- }
-
- /* next row... */
- src_ptr += src_pixels_per_line;
- output_ptr += output_width;
- }
- }
-}
-
-
-void vp8_filter_block2d_second_pass4
-(
- unsigned char * RESTRICT src_ptr,
- unsigned char * RESTRICT output_ptr,
- int output_pitch,
- int yoffset
-)
-{
- unsigned int i;
-
- int Temp1, Temp2, Temp3, Temp4;
- unsigned int vector1b, vector2b, vector3b, vector4a;
-
- unsigned char src_ptr_l2;
- unsigned char src_ptr_l1;
- unsigned char src_ptr_0;
- unsigned char src_ptr_r1;
- unsigned char src_ptr_r2;
- unsigned char src_ptr_r3;
-
- unsigned char *cm = ff_cropTbl + CROP_WIDTH;
-
- vector4a = 64;
-
- /* load filter coefficients */
- vector1b = sub_pel_filterss[yoffset][0];
- vector2b = sub_pel_filterss[yoffset][2];
- vector3b = sub_pel_filterss[yoffset][1];
-
- if (vector1b) {
- /* 6 tap filter */
-
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr);
-
- for (i = 2; i--;)
- {
-
- /* do not allow compiler to reorder instructions */
- __asm__ __volatile__ (
- ".set noreorder \n\t"
- :
- :
- );
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l2], -8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 12(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 13(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 10(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 14(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 11(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 15(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp3], $ac0, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp4], $ac1, 9 \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
- [src_ptr_l2] "=&r" (src_ptr_l2), [src_ptr_r3] "=&r" (src_ptr_r3)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
- [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- output_ptr[0] = cm[Temp1];
- output_ptr[1] = cm[Temp2];
- output_ptr[2] = cm[Temp3];
- output_ptr[3] = cm[Temp4];
-
- output_ptr += output_pitch;
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l2], -4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], 0(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 12(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 16(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], 1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 13(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 17(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], 2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 10(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 14(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 18(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], 3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 11(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 15(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 19(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp3], $ac0, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp4], $ac1, 9 \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
- [src_ptr_l2] "=&r" (src_ptr_l2), [src_ptr_r3] "=&r" (src_ptr_r3)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
- [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- output_ptr[0] = cm[Temp1];
- output_ptr[1] = cm[Temp2];
- output_ptr[2] = cm[Temp3];
- output_ptr[3] = cm[Temp4];
-
- src_ptr += 8;
- output_ptr += output_pitch;
- }
- }
- else {
- /* 4 tap filter */
-
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr);
-
- for (i = 2; i--;)
- {
- /* do not allow compiler to reorder instructions */
- __asm__ __volatile__ (
- ".set noreorder \n\t"
- :
- :
- );
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l1], -4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 8(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 9(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 10(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 11(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp3], $ac0, 9 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp4], $ac1, 9 \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- output_ptr[0] = cm[Temp1];
- output_ptr[1] = cm[Temp2];
- output_ptr[2] = cm[Temp3];
- output_ptr[3] = cm[Temp4];
-
- output_ptr += output_pitch;
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l1], 0(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 12(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], 1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 13(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], 2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 10(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 14(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], 3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 11(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 15(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp3], $ac0, 9 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp4], $ac1, 9 \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=r" (Temp4),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- output_ptr[0] = cm[Temp1];
- output_ptr[1] = cm[Temp2];
- output_ptr[2] = cm[Temp3];
- output_ptr[3] = cm[Temp4];
-
- src_ptr += 8;
- output_ptr += output_pitch;
- }
- }
-}
-
-
-void vp8_filter_block2d_second_pass_8
-(
- unsigned char * RESTRICT src_ptr,
- unsigned char * RESTRICT output_ptr,
- int output_pitch,
- unsigned int output_height,
- unsigned int output_width,
- unsigned int yoffset
-)
-{
- unsigned int i, j;
-
- int Temp1, Temp2, Temp3, Temp4, Temp5, Temp6, Temp7, Temp8;
- unsigned int vector1b, vector2b, vector3b, vector4a;
-
- unsigned char src_ptr_l2;
- unsigned char src_ptr_l1;
- unsigned char src_ptr_0;
- unsigned char src_ptr_r1;
- unsigned char src_ptr_r2;
- unsigned char src_ptr_r3;
- unsigned char *cm = ff_cropTbl + CROP_WIDTH;
-
- vector4a = 64;
-
- vector1b = sub_pel_filterss[yoffset][0];
- vector2b = sub_pel_filterss[yoffset][2];
- vector3b = sub_pel_filterss[yoffset][1];
-
- if (vector1b) {
- /* 6 tap filter */
-
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr);
-
- for (i = output_height; i--;)
- {
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l2], -16(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 16(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 24(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -15(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 17(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 25(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -14(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 10(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 18(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 26(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -13(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 11(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 19(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 27(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp3], $ac0, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2), [Temp3] "=&r" (Temp3),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
- [src_ptr_l2] "=&r" (src_ptr_l2), [src_ptr_r3] "=&r" (src_ptr_r3)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
- [src_ptr] "r" (src_ptr)
- );
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l2], -12(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 12(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 20(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 28(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp4], $ac1, 9 \n\t"
-
- "lbu %[src_ptr_l2], -11(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 13(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 21(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 29(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp5], $ac2, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -10(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 14(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 22(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 30(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp6], $ac3, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 15(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 23(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 31(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp7], $ac0, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp8], $ac1, 9 \n\t"
-
- : [Temp4] "=&r" (Temp4), [Temp5] "=&r" (Temp5),
- [Temp6] "=&r" (Temp6), [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
- [src_ptr_l2] "=&r" (src_ptr_l2),[src_ptr_r3] "=&r" (src_ptr_r3)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
- [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- output_ptr[0] = cm[Temp1];
- output_ptr[1] = cm[Temp2];
- output_ptr[2] = cm[Temp3];
- output_ptr[3] = cm[Temp4];
- output_ptr[4] = cm[Temp5];
- output_ptr[5] = cm[Temp6];
- output_ptr[6] = cm[Temp7];
- output_ptr[7] = cm[Temp8];
-
- src_ptr += 8;
- output_ptr += output_pitch;
- }
- }
- else {
- /* 4 tap filter */
-
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr);
-
- for (i = output_height; i--;)
- {
- __asm__ __volatile__ (
- "lbu %[src_ptr_l1], -8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 16(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- : [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
- );
-
- __asm__ __volatile__ (
- "lbu %[src_ptr_l1], -7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 17(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- : [Temp1] "=r" (Temp1),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
- );
-
- src_ptr_l1 = src_ptr[-6];
- src_ptr_0 = src_ptr[2];
- src_ptr_r1 = src_ptr[10];
- src_ptr_r2 = src_ptr[18];
-
- __asm__ __volatile__ (
- "mtlo %[vector4a], $ac0 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
-
- : [Temp2] "=r" (Temp2)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
- [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
- [vector4a] "r" (vector4a)
- );
-
- src_ptr_l1 = src_ptr[-5];
- src_ptr_0 = src_ptr[3];
- src_ptr_r1 = src_ptr[11];
- src_ptr_r2 = src_ptr[19];
-
- __asm__ __volatile__ (
- "mtlo %[vector4a], $ac1 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp3], $ac0, 9 \n\t"
-
- : [Temp3] "=r" (Temp3)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
- [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
- [vector4a] "r" (vector4a)
- );
-
- src_ptr_l1 = src_ptr[-4];
- src_ptr_0 = src_ptr[4];
- src_ptr_r1 = src_ptr[12];
- src_ptr_r2 = src_ptr[20];
-
- __asm__ __volatile__ (
- "mtlo %[vector4a], $ac2 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp4], $ac1, 9 \n\t"
-
- : [Temp4] "=r" (Temp4)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
- [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
- [vector4a] "r" (vector4a)
- );
-
- src_ptr_l1 = src_ptr[-3];
- src_ptr_0 = src_ptr[5];
- src_ptr_r1 = src_ptr[13];
- src_ptr_r2 = src_ptr[21];
-
- __asm__ __volatile__ (
- "mtlo %[vector4a], $ac3 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp5], $ac2, 9 \n\t"
-
- : [Temp5] "=&r" (Temp5)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
- [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
- [vector4a] "r" (vector4a)
- );
-
- src_ptr_l1 = src_ptr[-2];
- src_ptr_0 = src_ptr[6];
- src_ptr_r1 = src_ptr[14];
- src_ptr_r2 = src_ptr[22];
-
- __asm__ __volatile__ (
- "mtlo %[vector4a], $ac0 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp6], $ac3, 9 \n\t"
-
- : [Temp6] "=r" (Temp6)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
- [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
- [vector4a] "r" (vector4a)
- );
-
- src_ptr_l1 = src_ptr[-1];
- src_ptr_0 = src_ptr[7];
- src_ptr_r1 = src_ptr[15];
- src_ptr_r2 = src_ptr[23];
-
- __asm__ __volatile__ (
- "mtlo %[vector4a], $ac1 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp7], $ac0, 9 \n\t"
- "extp %[Temp8], $ac1, 9 \n\t"
-
- : [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [src_ptr_l1] "r" (src_ptr_l1), [src_ptr_0] "r" (src_ptr_0),
- [src_ptr_r1] "r" (src_ptr_r1), [src_ptr_r2] "r" (src_ptr_r2),
- [vector4a] "r" (vector4a)
- );
-
- /* clamp and store results */
- output_ptr[0] = cm[Temp1];
- output_ptr[1] = cm[Temp2];
- output_ptr[2] = cm[Temp3];
- output_ptr[3] = cm[Temp4];
- output_ptr[4] = cm[Temp5];
- output_ptr[5] = cm[Temp6];
- output_ptr[6] = cm[Temp7];
- output_ptr[7] = cm[Temp8];
-
- src_ptr += 8;
- output_ptr += output_pitch;
- }
- }
-}
-
-
-void vp8_filter_block2d_second_pass161
-(
- unsigned char * RESTRICT src_ptr,
- unsigned char * RESTRICT output_ptr,
- int output_pitch,
- const unsigned short *vp8_filter
-)
-{
- unsigned int i, j;
-
- int Temp1, Temp2, Temp3, Temp4, Temp5, Temp6, Temp7, Temp8;
- unsigned int vector4a;
- unsigned int vector1b, vector2b, vector3b;
-
- unsigned char src_ptr_l2;
- unsigned char src_ptr_l1;
- unsigned char src_ptr_0;
- unsigned char src_ptr_r1;
- unsigned char src_ptr_r2;
- unsigned char src_ptr_r3;
- unsigned char *cm = ff_cropTbl + CROP_WIDTH;
-
- vector4a = 64;
-
- vector1b = vp8_filter[0];
- vector2b = vp8_filter[2];
- vector3b = vp8_filter[1];
-
- if (vector1b == 0) {
- /* 4 tap filter */
-
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + 16);
-
- for (i = 16; i--;)
- {
- /* unrolling for loop */
- for (j = 0; j < 16; j += 8)
- {
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l1], -16(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 16(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 32(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -15(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 17(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 33(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -14(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 18(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 34(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp2], $ac3, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -13(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 19(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 35(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp3], $ac1, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -12(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 20(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 36(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
- "extp %[Temp4], $ac3, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -11(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 21(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 37(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp5], $ac2, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -10(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 22(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 38(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp6], $ac3, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l1], -9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 23(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 39(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp7], $ac1, 9 \n\t"
-
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp8], $ac3, 9 \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4),
- [Temp5] "=&r" (Temp5), [Temp6] "=&r" (Temp6),
- [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2)
- : [vector2b] "r" (vector2b), [vector3b] "r" (vector3b),
- [vector4a] "r" (vector4a), [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- output_ptr[j] = cm[Temp1];
- output_ptr[j+1] = cm[Temp2];
- output_ptr[j+2] = cm[Temp3];
- output_ptr[j+3] = cm[Temp4];
- output_ptr[j+4] = cm[Temp5];
- output_ptr[j+5] = cm[Temp6];
- output_ptr[j+6] = cm[Temp7];
- output_ptr[j+7] = cm[Temp8];
-
- src_ptr += 8;
- }
-
- output_ptr += output_pitch;
- }
- }
- else {
- /* 4 tap filter */
-
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + 16);
-
- /* unroll for loop */
- for (i = 16; i--;)
- {
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l2], -32(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -16(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 0(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 16(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 32(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 48(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -31(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -15(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 17(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 33(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 49(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -30(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -14(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 18(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 34(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 50(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp2], $ac0, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -29(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -13(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 19(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 35(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 51(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp3], $ac1, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -28(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -12(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 20(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 36(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 52(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
- "extp %[Temp4], $ac3, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -27(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -11(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 21(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 37(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 53(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp5], $ac2, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -26(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -10(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 22(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 38(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 54(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp6], $ac0, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -25(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 23(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 39(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 55(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp7], $ac1, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp8], $ac3, 9 \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4),
- [Temp5] "=&r" (Temp5), [Temp6] "=&r" (Temp6),
- [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
- [src_ptr_l2] "=&r" (src_ptr_l2),[src_ptr_r3] "=&r" (src_ptr_r3)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
- [src_ptr] "r" (src_ptr)
- );
-
- /* clamp and store results */
- output_ptr[0] = cm[Temp1];
- output_ptr[1] = cm[Temp2];
- output_ptr[2] = cm[Temp3];
- output_ptr[3] = cm[Temp4];
- output_ptr[4] = cm[Temp5];
- output_ptr[5] = cm[Temp6];
- output_ptr[6] = cm[Temp7];
- output_ptr[7] = cm[Temp8];
-
- /* apply filter with vectors pairs */
- __asm__ __volatile__ (
- "lbu %[src_ptr_l2], -24(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 8(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 24(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 40(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 56(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -23(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -7(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 9(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 25(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 41(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 57(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp1], $ac2, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -22(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -6(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 10(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 26(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 42(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 58(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp2], $ac0, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -21(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -5(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 11(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 27(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 43(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 59(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp3], $ac1, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -20(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -4(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 12(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 28(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 44(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 60(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac2 \n\t"
- "extp %[Temp4], $ac3, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac2, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac2, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -19(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -3(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 13(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 29(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 45(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 61(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac0 \n\t"
- "extp %[Temp5], $ac2, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac0, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac0, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -18(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -2(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 14(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 30(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 46(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 62(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac1 \n\t"
- "extp %[Temp6], $ac0, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac1, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac1, %[src_ptr_l1], %[vector3b] \n\t"
-
- "lbu %[src_ptr_l2], -17(%[src_ptr]) \n\t"
- "lbu %[src_ptr_l1], -1(%[src_ptr]) \n\t"
- "lbu %[src_ptr_0], 15(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r1], 31(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r2], 47(%[src_ptr]) \n\t"
- "lbu %[src_ptr_r3], 63(%[src_ptr]) \n\t"
- "mtlo %[vector4a], $ac3 \n\t"
- "extp %[Temp7], $ac1, 9 \n\t"
-
- "append %[src_ptr_l2], %[src_ptr_r3], 8 \n\t"
- "append %[src_ptr_0], %[src_ptr_r1], 8 \n\t"
- "append %[src_ptr_l1], %[src_ptr_r2], 8 \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_l2], %[vector1b] \n\t"
- "dpau.h.qbr $ac3, %[src_ptr_0], %[vector2b] \n\t"
- "dpsu.h.qbr $ac3, %[src_ptr_l1], %[vector3b] \n\t"
- "extp %[Temp8], $ac3, 9 \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2),
- [Temp3] "=&r" (Temp3), [Temp4] "=&r" (Temp4),
- [Temp5] "=&r" (Temp5), [Temp6] "=&r" (Temp6),
- [Temp7] "=&r" (Temp7), [Temp8] "=r" (Temp8),
- [src_ptr_l1] "=&r" (src_ptr_l1), [src_ptr_0] "=&r" (src_ptr_0),
- [src_ptr_r1] "=&r" (src_ptr_r1), [src_ptr_r2] "=&r" (src_ptr_r2),
- [src_ptr_l2] "=&r" (src_ptr_l2), [src_ptr_r3] "=&r" (src_ptr_r3)
- : [vector1b] "r" (vector1b), [vector2b] "r" (vector2b),
- [vector3b] "r" (vector3b), [vector4a] "r" (vector4a),
- [src_ptr] "r" (src_ptr)
- );
-
- src_ptr += 16;
- output_ptr[8] = cm[Temp1];
- output_ptr[9] = cm[Temp2];
- output_ptr[10] = cm[Temp3];
- output_ptr[11] = cm[Temp4];
- output_ptr[12] = cm[Temp5];
- output_ptr[13] = cm[Temp6];
- output_ptr[14] = cm[Temp7];
- output_ptr[15] = cm[Temp8];
-
- output_ptr += output_pitch;
- }
- }
-}
-
-
-void vp8_sixtap_predict_mips
-(
- unsigned char * RESTRICT src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char * RESTRICT dst_ptr,
- int dst_pitch
-)
-{
- unsigned char FData[9*4]; /* Temp data bufffer used in filtering */
- unsigned int pos = 16;
-
- /* bit positon for extract from acc */
- __asm__ __volatile__ (
- "wrdsp %[pos], 1 \n\t"
- :
- : [pos] "r" (pos)
- );
-
- if (yoffset) {
- /* First filter 1-D horizontally... */
- vp8_filter_block2d_first_pass_4(src_ptr - (2 * src_pixels_per_line), FData,
- src_pixels_per_line, 9, xoffset, 4);
- /* then filter verticaly... */
- vp8_filter_block2d_second_pass4(FData + 8, dst_ptr, dst_pitch, yoffset);
- }
- else
- /* if (yoffsset == 0) vp8_filter_block2d_first_pass save data to dst_ptr */
- vp8_filter_block2d_first_pass_4(src_ptr, dst_ptr, src_pixels_per_line,
- 4, xoffset, dst_pitch);
-}
-
-
-void vp8_sixtap_predict8x8_mips
-(
- unsigned char * RESTRICT src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char * RESTRICT dst_ptr,
- int dst_pitch
-)
-{
-
- unsigned char FData[13*8]; /* Temp data bufffer used in filtering */
- unsigned int pos, i, Temp1, Temp2;
-
- pos = 16;
-
- /* bit positon for extract from acc */
- __asm__ __volatile__ (
- "wrdsp %[pos], 1 \n\t"
- :
- : [pos] "r" (pos)
- );
-
- if (yoffset) {
-
- src_ptr = src_ptr - (2 * src_pixels_per_line);
-
- if (xoffset)
- /* filter 1-D horizontally... */
- vp8_filter_block2d_first_pass_8_all(src_ptr, FData, src_pixels_per_line,
- 13, xoffset, 8);
-
- else {
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + 2*src_pixels_per_line);
-
- __asm__ __volatile__ (
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 0(%[FData]) \n\t"
- "sw %[Temp2], 4(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 8(%[FData]) \n\t"
- "sw %[Temp2], 12(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 16(%[FData]) \n\t"
- "sw %[Temp2], 20(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 24(%[FData]) \n\t"
- "sw %[Temp2], 28(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 32(%[FData]) \n\t"
- "sw %[Temp2], 36(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 40(%[FData]) \n\t"
- "sw %[Temp2], 44(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 48(%[FData]) \n\t"
- "sw %[Temp2], 52(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 56(%[FData]) \n\t"
- "sw %[Temp2], 60(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 64(%[FData]) \n\t"
- "sw %[Temp2], 68(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 72(%[FData]) \n\t"
- "sw %[Temp2], 76(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 80(%[FData]) \n\t"
- "sw %[Temp2], 84(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 88(%[FData]) \n\t"
- "sw %[Temp2], 92(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 96(%[FData]) \n\t"
- "sw %[Temp2], 100(%[FData]) \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2)
- : [FData] "r" (FData), [src_ptr] "r" (src_ptr),
- [src_pixels_per_line] "r" (src_pixels_per_line)
- );
- }
- /* filter verticaly... */
- vp8_filter_block2d_second_pass_8(FData + 16, dst_ptr, dst_pitch, 8, 8, yoffset);
- }
-
- /* if (yoffsset == 0) vp8_filter_block2d_first_pass save data to dst_ptr */
- else {
- if (xoffset)
- vp8_filter_block2d_first_pass_8_all(src_ptr, dst_ptr, src_pixels_per_line,
- 8, xoffset, dst_pitch);
-
- else {
- /* copy from src buffer to dst buffer */
- __asm__ __volatile__ (
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 0(%[dst_ptr]) \n\t"
- "sw %[Temp2], 4(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 8(%[dst_ptr]) \n\t"
- "sw %[Temp2], 12(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 16(%[dst_ptr]) \n\t"
- "sw %[Temp2], 20(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 24(%[dst_ptr]) \n\t"
- "sw %[Temp2], 28(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 32(%[dst_ptr]) \n\t"
- "sw %[Temp2], 36(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 40(%[dst_ptr]) \n\t"
- "sw %[Temp2], 44(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 48(%[dst_ptr]) \n\t"
- "sw %[Temp2], 52(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 56(%[dst_ptr]) \n\t"
- "sw %[Temp2], 60(%[dst_ptr]) \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2)
- : [dst_ptr] "r" (dst_ptr), [src_ptr] "r" (src_ptr),
- [src_pixels_per_line] "r" (src_pixels_per_line)
- );
- }
- }
-}
-
-
-void vp8_sixtap_predict8x4_mips
-(
- unsigned char * RESTRICT src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char * RESTRICT dst_ptr,
- int dst_pitch
-)
-{
- unsigned char FData[9*8]; /* Temp data bufffer used in filtering */
- unsigned int pos, i, Temp1, Temp2;
-
- pos = 16;
-
- /* bit positon for extract from acc */
- __asm__ __volatile__ (
- "wrdsp %[pos], 1 \n\t"
- :
- : [pos] "r" (pos)
- );
-
- if (yoffset) {
-
- src_ptr = src_ptr - (2 * src_pixels_per_line);
-
- if (xoffset)
- /* filter 1-D horizontally... */
- vp8_filter_block2d_first_pass_8_all(src_ptr, FData, src_pixels_per_line,
- 9, xoffset, 8);
-
- else {
- /* prefetch src_ptr data to cache memory */
- prefetch_load(src_ptr + 2*src_pixels_per_line);
-
- __asm__ __volatile__ (
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 0(%[FData]) \n\t"
- "sw %[Temp2], 4(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 8(%[FData]) \n\t"
- "sw %[Temp2], 12(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 16(%[FData]) \n\t"
- "sw %[Temp2], 20(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 24(%[FData]) \n\t"
- "sw %[Temp2], 28(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 32(%[FData]) \n\t"
- "sw %[Temp2], 36(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 40(%[FData]) \n\t"
- "sw %[Temp2], 44(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 48(%[FData]) \n\t"
- "sw %[Temp2], 52(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 56(%[FData]) \n\t"
- "sw %[Temp2], 60(%[FData]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 64(%[FData]) \n\t"
- "sw %[Temp2], 68(%[FData]) \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2)
- : [FData] "r" (FData), [src_ptr] "r" (src_ptr),
- [src_pixels_per_line] "r" (src_pixels_per_line)
- );
- }
- /* filter verticaly... */
- vp8_filter_block2d_second_pass_8(FData + 16, dst_ptr, dst_pitch, 4, 8, yoffset);
- }
-
- /* if (yoffsset == 0) vp8_filter_block2d_first_pass save data to dst_ptr */
- else {
- if (xoffset)
- vp8_filter_block2d_first_pass_8_all(src_ptr, dst_ptr, src_pixels_per_line,
- 4, xoffset, dst_pitch);
-
- else {
- /* copy from src buffer to dst buffer */
- __asm__ __volatile__ (
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 0(%[dst_ptr]) \n\t"
- "sw %[Temp2], 4(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 8(%[dst_ptr]) \n\t"
- "sw %[Temp2], 12(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 16(%[dst_ptr]) \n\t"
- "sw %[Temp2], 20(%[dst_ptr]) \n\t"
- "addu %[src_ptr], %[src_ptr], %[src_pixels_per_line] \n\t"
-
- "ulw %[Temp1], 0(%[src_ptr]) \n\t"
- "ulw %[Temp2], 4(%[src_ptr]) \n\t"
- "sw %[Temp1], 24(%[dst_ptr]) \n\t"
- "sw %[Temp2], 28(%[dst_ptr]) \n\t"
-
- : [Temp1] "=&r" (Temp1), [Temp2] "=&r" (Temp2)
- : [dst_ptr] "r" (dst_ptr), [src_ptr] "r" (src_ptr),
- [src_pixels_per_line] "r" (src_pixels_per_line)
- );
- }
- }
-}
-
-
-void vp8_sixtap_predict16x16_mips
-(
- unsigned char * RESTRICT src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char * RESTRICT dst_ptr,
- int dst_pitch
-)
-{
- const unsigned short *VFilter;
- unsigned char FData[21*16]; /* Temp data bufffer used in filtering */
- unsigned int pos;
-
- VFilter = sub_pel_filterss[yoffset];
-
- pos = 16;
-
- /* bit positon for extract from acc */
- __asm__ __volatile__ (
- "wrdsp %[pos], 1 \n\t"
- :
- : [pos] "r" (pos)
- );
-
- if (yoffset) {
-
- src_ptr = src_ptr - (2 * src_pixels_per_line);
-
- switch (xoffset) {
- /* filter 1-D horizontally... */
- case 2:
- case 4:
- case 6:
- /* 6 tap filter */
- vp8_filter_block2d_first_pass16_6tap(src_ptr, FData, src_pixels_per_line,
- 21, xoffset, 16);
- break;
- case 0:
- /* only copy buffer */
- vp8_filter_block2d_first_pass16_0(src_ptr, FData, src_pixels_per_line);
- break;
- case 1:
- case 3:
- case 5:
- case 7:
- /* 4 tap filter */
- vp8_filter_block2d_first_pass16_4tap(src_ptr, FData, src_pixels_per_line, 16,
- 21, xoffset, yoffset, dst_ptr, dst_pitch);
- break;
- }
- /* filter verticaly... */
- vp8_filter_block2d_second_pass161(FData + 32, dst_ptr, dst_pitch, VFilter);
- }
- else {
- /* if (yoffsset == 0) vp8_filter_block2d_first_pass save data to dst_ptr */
- switch (xoffset) {
- case 2:
- case 4:
- case 6:
- /* 6 tap filter */
- vp8_filter_block2d_first_pass16_6tap(src_ptr, dst_ptr, src_pixels_per_line,
- 16, xoffset, dst_pitch);
- break;
- case 1:
- case 3:
- case 5:
- case 7:
- /* 4 tap filter */
- vp8_filter_block2d_first_pass16_4tap(src_ptr, dst_ptr, src_pixels_per_line, 16,
- 21, xoffset, yoffset, dst_ptr, dst_pitch);
- break;
- }
- }
-}
diff --git a/vp8/common/mips/subpixel_mips.h b/vp8/common/mips/subpixel_mips.h
deleted file mode 100755
index 0ff1d3f..0000000
--- a/vp8/common/mips/subpixel_mips.h
+++ /dev/null
@@ -1,36 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#ifndef SUBPIXEL_MIPS_H
-#define SUBPIXEL_MIPS_H
-
-extern prototype_subpixel_predict(vp8_sixtap_predict16x16_mips);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x8_mips);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x4_mips);
-extern prototype_subpixel_predict(vp8_sixtap_predict_mips);
-extern void dsputil_static_init(void);
-
-#undef vp8_subpix_sixtap16x16
-#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_mips
-
-#undef vp8_subpix_sixtap8x8
-#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_mips
-
-#undef vp8_subpix_sixtap8x4
-#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_mips
-
-#undef vp8_subpix_sixtap4x4
-#define vp8_subpix_sixtap4x4 vp8_sixtap_predict_mips
-
-#endif
-#endif
\ No newline at end of file
diff --git a/vp8/common/modecontext.c b/vp8/common/modecontext.c
deleted file mode 100644
index a31a561..0000000
--- a/vp8/common/modecontext.c
+++ /dev/null
@@ -1,146 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "entropymode.h"
-
-const unsigned int vp8_kf_default_bmode_counts [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES] =
-{
- {
- /*Above Mode : 0*/
- { 43438, 2195, 470, 316, 615, 171, 217, 412, 124, 160, }, /* left_mode 0 */
- { 5722, 2751, 296, 291, 81, 68, 80, 101, 100, 170, }, /* left_mode 1 */
- { 1629, 201, 307, 25, 47, 16, 34, 72, 19, 28, }, /* left_mode 2 */
- { 332, 266, 36, 500, 20, 65, 23, 14, 154, 106, }, /* left_mode 3 */
- { 450, 97, 10, 24, 117, 10, 2, 12, 8, 71, }, /* left_mode 4 */
- { 384, 49, 29, 44, 12, 162, 51, 5, 87, 42, }, /* left_mode 5 */
- { 495, 53, 157, 27, 14, 57, 180, 17, 17, 34, }, /* left_mode 6 */
- { 695, 64, 62, 9, 27, 5, 3, 147, 10, 26, }, /* left_mode 7 */
- { 230, 54, 20, 124, 16, 125, 29, 12, 283, 37, }, /* left_mode 8 */
- { 260, 87, 21, 120, 32, 16, 33, 16, 33, 203, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 1*/
- { 3934, 2573, 355, 137, 128, 87, 133, 117, 37, 27, }, /* left_mode 0 */
- { 1036, 1929, 278, 135, 27, 37, 48, 55, 41, 91, }, /* left_mode 1 */
- { 223, 256, 253, 15, 13, 9, 28, 64, 3, 3, }, /* left_mode 2 */
- { 120, 129, 17, 316, 15, 11, 9, 4, 53, 74, }, /* left_mode 3 */
- { 129, 58, 6, 11, 38, 2, 0, 5, 2, 67, }, /* left_mode 4 */
- { 53, 22, 11, 16, 8, 26, 14, 3, 19, 12, }, /* left_mode 5 */
- { 59, 26, 61, 11, 4, 9, 35, 13, 8, 8, }, /* left_mode 6 */
- { 101, 52, 40, 8, 5, 2, 8, 59, 2, 20, }, /* left_mode 7 */
- { 48, 34, 10, 52, 8, 15, 6, 6, 63, 20, }, /* left_mode 8 */
- { 96, 48, 22, 63, 11, 14, 5, 8, 9, 96, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 2*/
- { 709, 461, 506, 36, 27, 33, 151, 98, 24, 6, }, /* left_mode 0 */
- { 201, 375, 442, 27, 13, 8, 46, 58, 6, 19, }, /* left_mode 1 */
- { 122, 140, 417, 4, 13, 3, 33, 59, 4, 2, }, /* left_mode 2 */
- { 36, 17, 22, 16, 6, 8, 12, 17, 9, 21, }, /* left_mode 3 */
- { 51, 15, 7, 1, 14, 0, 4, 5, 3, 22, }, /* left_mode 4 */
- { 18, 11, 30, 9, 7, 20, 11, 5, 2, 6, }, /* left_mode 5 */
- { 38, 21, 103, 9, 4, 12, 79, 13, 2, 5, }, /* left_mode 6 */
- { 64, 17, 66, 2, 12, 4, 2, 65, 4, 5, }, /* left_mode 7 */
- { 14, 7, 7, 16, 3, 11, 4, 13, 15, 16, }, /* left_mode 8 */
- { 36, 8, 32, 9, 9, 4, 14, 7, 6, 24, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 3*/
- { 1340, 173, 36, 119, 30, 10, 13, 10, 20, 26, }, /* left_mode 0 */
- { 156, 293, 26, 108, 5, 16, 2, 4, 23, 30, }, /* left_mode 1 */
- { 60, 34, 13, 7, 3, 3, 0, 8, 4, 5, }, /* left_mode 2 */
- { 72, 64, 1, 235, 3, 9, 2, 7, 28, 38, }, /* left_mode 3 */
- { 29, 14, 1, 3, 5, 0, 2, 2, 5, 13, }, /* left_mode 4 */
- { 22, 7, 4, 11, 2, 5, 1, 2, 6, 4, }, /* left_mode 5 */
- { 18, 14, 5, 6, 4, 3, 14, 0, 9, 2, }, /* left_mode 6 */
- { 41, 10, 7, 1, 2, 0, 0, 10, 2, 1, }, /* left_mode 7 */
- { 23, 19, 2, 33, 1, 5, 2, 0, 51, 8, }, /* left_mode 8 */
- { 33, 26, 7, 53, 3, 9, 3, 3, 9, 19, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 4*/
- { 410, 165, 43, 31, 66, 15, 30, 54, 8, 17, }, /* left_mode 0 */
- { 115, 64, 27, 18, 30, 7, 11, 15, 4, 19, }, /* left_mode 1 */
- { 31, 23, 25, 1, 7, 2, 2, 10, 0, 5, }, /* left_mode 2 */
- { 17, 4, 1, 6, 8, 2, 7, 5, 5, 21, }, /* left_mode 3 */
- { 120, 12, 1, 2, 83, 3, 0, 4, 1, 40, }, /* left_mode 4 */
- { 4, 3, 1, 2, 1, 2, 5, 0, 3, 6, }, /* left_mode 5 */
- { 10, 2, 13, 6, 6, 6, 8, 2, 4, 5, }, /* left_mode 6 */
- { 58, 10, 5, 1, 28, 1, 1, 33, 1, 9, }, /* left_mode 7 */
- { 8, 2, 1, 4, 2, 5, 1, 1, 2, 10, }, /* left_mode 8 */
- { 76, 7, 5, 7, 18, 2, 2, 0, 5, 45, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 5*/
- { 444, 46, 47, 20, 14, 110, 60, 14, 60, 7, }, /* left_mode 0 */
- { 59, 57, 25, 18, 3, 17, 21, 6, 14, 6, }, /* left_mode 1 */
- { 24, 17, 20, 6, 4, 13, 7, 2, 3, 2, }, /* left_mode 2 */
- { 13, 11, 5, 14, 4, 9, 2, 4, 15, 7, }, /* left_mode 3 */
- { 8, 5, 2, 1, 4, 0, 1, 1, 2, 12, }, /* left_mode 4 */
- { 19, 5, 5, 7, 4, 40, 6, 3, 10, 4, }, /* left_mode 5 */
- { 16, 5, 9, 1, 1, 16, 26, 2, 10, 4, }, /* left_mode 6 */
- { 11, 4, 8, 1, 1, 4, 4, 5, 4, 1, }, /* left_mode 7 */
- { 15, 1, 3, 7, 3, 21, 7, 1, 34, 5, }, /* left_mode 8 */
- { 18, 5, 1, 3, 4, 3, 7, 1, 2, 9, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 6*/
- { 476, 149, 94, 13, 14, 77, 291, 27, 23, 3, }, /* left_mode 0 */
- { 79, 83, 42, 14, 2, 12, 63, 2, 4, 14, }, /* left_mode 1 */
- { 43, 36, 55, 1, 3, 8, 42, 11, 5, 1, }, /* left_mode 2 */
- { 9, 9, 6, 16, 1, 5, 6, 3, 11, 10, }, /* left_mode 3 */
- { 10, 3, 1, 3, 10, 1, 0, 1, 1, 4, }, /* left_mode 4 */
- { 14, 6, 15, 5, 1, 20, 25, 2, 5, 0, }, /* left_mode 5 */
- { 28, 7, 51, 1, 0, 8, 127, 6, 2, 5, }, /* left_mode 6 */
- { 13, 3, 3, 2, 3, 1, 2, 8, 1, 2, }, /* left_mode 7 */
- { 10, 3, 3, 3, 3, 8, 2, 2, 9, 3, }, /* left_mode 8 */
- { 13, 7, 11, 4, 0, 4, 6, 2, 5, 8, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 7*/
- { 376, 135, 119, 6, 32, 8, 31, 224, 9, 3, }, /* left_mode 0 */
- { 93, 60, 54, 6, 13, 7, 8, 92, 2, 12, }, /* left_mode 1 */
- { 74, 36, 84, 0, 3, 2, 9, 67, 2, 1, }, /* left_mode 2 */
- { 19, 4, 4, 8, 8, 2, 4, 7, 6, 16, }, /* left_mode 3 */
- { 51, 7, 4, 1, 77, 3, 0, 14, 1, 15, }, /* left_mode 4 */
- { 7, 7, 5, 7, 4, 7, 4, 5, 0, 3, }, /* left_mode 5 */
- { 18, 2, 19, 2, 2, 4, 12, 11, 1, 2, }, /* left_mode 6 */
- { 129, 6, 27, 1, 21, 3, 0, 189, 0, 6, }, /* left_mode 7 */
- { 9, 1, 2, 8, 3, 7, 0, 5, 3, 3, }, /* left_mode 8 */
- { 20, 4, 5, 10, 4, 2, 7, 17, 3, 16, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 8*/
- { 617, 68, 34, 79, 11, 27, 25, 14, 75, 13, }, /* left_mode 0 */
- { 51, 82, 21, 26, 6, 12, 13, 1, 26, 16, }, /* left_mode 1 */
- { 29, 9, 12, 11, 3, 7, 1, 10, 2, 2, }, /* left_mode 2 */
- { 17, 19, 11, 74, 4, 3, 2, 0, 58, 13, }, /* left_mode 3 */
- { 10, 1, 1, 3, 4, 1, 0, 2, 1, 8, }, /* left_mode 4 */
- { 14, 4, 5, 5, 1, 13, 2, 0, 27, 8, }, /* left_mode 5 */
- { 10, 3, 5, 4, 1, 7, 6, 4, 5, 1, }, /* left_mode 6 */
- { 10, 2, 6, 2, 1, 1, 1, 4, 2, 1, }, /* left_mode 7 */
- { 14, 8, 5, 23, 2, 12, 6, 2, 117, 5, }, /* left_mode 8 */
- { 9, 6, 2, 19, 1, 6, 3, 2, 9, 9, }, /* left_mode 9 */
- },
- {
- /*Above Mode : 9*/
- { 680, 73, 22, 38, 42, 5, 11, 9, 6, 28, }, /* left_mode 0 */
- { 113, 112, 21, 22, 10, 2, 8, 4, 6, 42, }, /* left_mode 1 */
- { 44, 20, 24, 6, 5, 4, 3, 3, 1, 2, }, /* left_mode 2 */
- { 40, 23, 7, 71, 5, 2, 4, 1, 7, 22, }, /* left_mode 3 */
- { 85, 9, 4, 4, 17, 2, 0, 3, 2, 23, }, /* left_mode 4 */
- { 13, 4, 2, 6, 1, 7, 0, 1, 7, 6, }, /* left_mode 5 */
- { 26, 6, 8, 3, 2, 3, 8, 1, 5, 4, }, /* left_mode 6 */
- { 54, 8, 9, 6, 7, 0, 1, 11, 1, 3, }, /* left_mode 7 */
- { 9, 10, 4, 13, 2, 5, 4, 2, 14, 8, }, /* left_mode 8 */
- { 92, 9, 5, 19, 15, 3, 3, 1, 6, 58, }, /* left_mode 9 */
- },
-};
diff --git a/vp8/common/mv.h b/vp8/common/mv.h
deleted file mode 100644
index 73c91b9..0000000
--- a/vp8/common/mv.h
+++ /dev/null
@@ -1,21 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_MV_H
-#define __INC_MV_H
-
-typedef struct
-{
- short row;
- short col;
-} MV;
-
-#endif
diff --git a/vp8/common/onyx.h b/vp8/common/onyx.h
deleted file mode 100644
index 426b8fc..0000000
--- a/vp8/common/onyx.h
+++ /dev/null
@@ -1,227 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_VP8_H
-#define __INC_VP8_H
-
-#ifdef __cplusplus
-extern "C"
-{
-#endif
-
-#include "vpx/internal/vpx_codec_internal.h"
-#include "vpx/vp8cx.h"
-#include "vpx_scale/yv12config.h"
-#include "type_aliases.h"
-#include "ppflags.h"
- typedef int *VP8_PTR;
-
- /* Create/destroy static data structures. */
-
- typedef enum
- {
- NORMAL = 0,
- FOURFIVE = 1,
- THREEFIVE = 2,
- ONETWO = 3
-
- } VPX_SCALING;
-
- typedef enum
- {
- VP8_LAST_FLAG = 1,
- VP8_GOLD_FLAG = 2,
- VP8_ALT_FLAG = 4
- } VP8_REFFRAME;
-
-
- typedef enum
- {
- USAGE_STREAM_FROM_SERVER = 0x0,
- USAGE_LOCAL_FILE_PLAYBACK = 0x1,
- USAGE_CONSTRAINED_QUALITY = 0x2
- } END_USAGE;
-
-
- typedef enum
- {
- MODE_REALTIME = 0x0,
- MODE_GOODQUALITY = 0x1,
- MODE_BESTQUALITY = 0x2,
- MODE_FIRSTPASS = 0x3,
- MODE_SECONDPASS = 0x4,
- MODE_SECONDPASS_BEST = 0x5,
- } MODE;
-
- typedef enum
- {
- FRAMEFLAGS_KEY = 1,
- FRAMEFLAGS_GOLDEN = 2,
- FRAMEFLAGS_ALTREF = 4,
- } FRAMETYPE_FLAGS;
-
-
-#include <assert.h>
- static __inline void Scale2Ratio(int mode, int *hr, int *hs)
- {
- switch (mode)
- {
- case NORMAL:
- *hr = 1;
- *hs = 1;
- break;
- case FOURFIVE:
- *hr = 4;
- *hs = 5;
- break;
- case THREEFIVE:
- *hr = 3;
- *hs = 5;
- break;
- case ONETWO:
- *hr = 1;
- *hs = 2;
- break;
- default:
- *hr = 1;
- *hs = 1;
- assert(0);
- break;
- }
- }
-
- typedef struct
- {
- int Version; // 4 versions of bitstream defined 0 best quality/slowest decode, 3 lowest quality/fastest decode
- int Width; // width of data passed to the compressor
- int Height; // height of data passed to the compressor
- double frame_rate; // set to passed in framerate
- int target_bandwidth; // bandwidth to be used in kilobits per second
-
- int noise_sensitivity; // parameter used for applying pre processing blur: recommendation 0
- int Sharpness; // parameter used for sharpening output: recommendation 0:
- int cpu_used;
-
- // mode ->
- //(0)=Realtime/Live Encoding. This mode is optimized for realtim encoding (for example, capturing
- // a television signal or feed from a live camera). ( speed setting controls how fast )
- //(1)=Good Quality Fast Encoding. The encoder balances quality with the amount of time it takes to
- // encode the output. ( speed setting controls how fast )
- //(2)=One Pass - Best Quality. The encoder places priority on the quality of the output over encoding
- // speed. The output is compressed at the highest possible quality. This option takes the longest
- // amount of time to encode. ( speed setting ignored )
- //(3)=Two Pass - First Pass. The encoder generates a file of statistics for use in the second encoding
- // pass. ( speed setting controls how fast )
- //(4)=Two Pass - Second Pass. The encoder uses the statistics that were generated in the first encoding
- // pass to create the compressed output. ( speed setting controls how fast )
- //(5)=Two Pass - Second Pass Best. The encoder uses the statistics that were generated in the first
- // encoding pass to create the compressed output using the highest possible quality, and taking a
- // longer amount of time to encode.. ( speed setting ignored )
- int Mode; //
-
- // Key Framing Operations
- int auto_key; // automatically detect cut scenes and set the keyframes
- int key_freq; // maximum distance to key frame.
-
- int allow_lag; // allow lagged compression (if 0 lagin frames is ignored)
- int lag_in_frames; // how many frames lag before we start encoding
-
- //----------------------------------------------------------------
- // DATARATE CONTROL OPTIONS
-
- int end_usage; // vbr or cbr
-
- // shoot to keep buffer full at all times by undershooting a bit 95 recommended
- int under_shoot_pct;
-
- // buffering parameters
- int starting_buffer_level; // in seconds
- int optimal_buffer_level;
- int maximum_buffer_size;
-
- // controlling quality
- int fixed_q;
- int worst_allowed_q;
- int best_allowed_q;
- int cq_level;
-
- // allow internal resizing ( currently disabled in the build !!!!!)
- int allow_spatial_resampling;
- int resample_down_water_mark;
- int resample_up_water_mark;
-
- // allow internal frame rate alterations
- int allow_df;
- int drop_frames_water_mark;
-
- // two pass datarate control
- int two_pass_vbrbias; // two pass datarate control tweaks
- int two_pass_vbrmin_section;
- int two_pass_vbrmax_section;
- // END DATARATE CONTROL OPTIONS
- //----------------------------------------------------------------
-
-
- // these parameters aren't to be used in final build don't use!!!
- int play_alternate;
- int alt_freq;
- int alt_q;
- int key_q;
- int gold_q;
-
-
- int multi_threaded; // how many threads to run the encoder on
- int token_partitions; // how many token partitions to create for multi core decoding
- int encode_breakout; // early breakout encode threshold : for video conf recommend 800
-
- int error_resilient_mode; // if running over udp networks provides decodable frames after a
- // dropped packet
-
- int arnr_max_frames;
- int arnr_strength ;
- int arnr_type ;
-
- struct vpx_fixed_buf two_pass_stats_in;
- struct vpx_codec_pkt_list *output_pkt_list;
-
- vp8e_tuning tuning;
- } VP8_CONFIG;
-
-
- void vp8_initialize();
-
- VP8_PTR vp8_create_compressor(VP8_CONFIG *oxcf);
- void vp8_remove_compressor(VP8_PTR *comp);
-
- void vp8_init_config(VP8_PTR onyx, VP8_CONFIG *oxcf);
- void vp8_change_config(VP8_PTR onyx, VP8_CONFIG *oxcf);
-
-// receive a frames worth of data caller can assume that a copy of this frame is made
-// and not just a copy of the pointer..
- int vp8_receive_raw_frame(VP8_PTR comp, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, INT64 time_stamp, INT64 end_time_stamp);
- int vp8_get_compressed_data(VP8_PTR comp, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, INT64 *time_stamp, INT64 *time_end, int flush);
- int vp8_get_preview_raw_frame(VP8_PTR comp, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags);
-
- int vp8_use_as_reference(VP8_PTR comp, int ref_frame_flags);
- int vp8_update_reference(VP8_PTR comp, int ref_frame_flags);
- int vp8_get_reference(VP8_PTR comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
- int vp8_set_reference(VP8_PTR comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
- int vp8_update_entropy(VP8_PTR comp, int update);
- int vp8_set_roimap(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4]);
- int vp8_set_active_map(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols);
- int vp8_set_internal_size(VP8_PTR comp, VPX_SCALING horiz_mode, VPX_SCALING vert_mode);
- int vp8_get_quantizer(VP8_PTR c);
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif
diff --git a/vp8/common/onyxc_int.h b/vp8/common/onyxc_int.h
deleted file mode 100644
index c8c2277..0000000
--- a/vp8/common/onyxc_int.h
+++ /dev/null
@@ -1,208 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_VP8C_INT_H
-#define __INC_VP8C_INT_H
-
-#include "vpx_config.h"
-#include "vpx/internal/vpx_codec_internal.h"
-#include "loopfilter.h"
-#include "entropymv.h"
-#include "entropy.h"
-#include "idct.h"
-#include "recon.h"
-#include "postproc.h"
-
-/*#ifdef PACKET_TESTING*/
-#include "header.h"
-/*#endif*/
-
-/* Create/destroy static data structures. */
-
-void vp8_initialize_common(void);
-
-#define MINQ 0
-#define MAXQ 127
-#define QINDEX_RANGE (MAXQ + 1)
-
-#define NUM_YV12_BUFFERS 4
-
-typedef struct frame_contexts
-{
- vp8_prob bmode_prob [VP8_BINTRAMODES-1];
- vp8_prob ymode_prob [VP8_YMODES-1]; /* interframe intra mode probs */
- vp8_prob uv_mode_prob [VP8_UV_MODES-1];
- vp8_prob sub_mv_ref_prob [VP8_SUBMVREFS-1];
- vp8_prob coef_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1];
- MV_CONTEXT mvc[2];
- MV_CONTEXT pre_mvc[2]; /* not to caculate the mvcost for the frame if mvc doesn't change. */
-} FRAME_CONTEXT;
-
-typedef enum
-{
- ONE_PARTITION = 0,
- TWO_PARTITION = 1,
- FOUR_PARTITION = 2,
- EIGHT_PARTITION = 3
-} TOKEN_PARTITION;
-
-typedef enum
-{
- RECON_CLAMP_REQUIRED = 0,
- RECON_CLAMP_NOTREQUIRED = 1
-} CLAMP_TYPE;
-
-typedef enum
-{
- SIXTAP = 0,
- BILINEAR = 1
-} INTERPOLATIONFILTERTYPE;
-
-typedef struct VP8_COMMON_RTCD
-{
-#if CONFIG_RUNTIME_CPU_DETECT
- vp8_idct_rtcd_vtable_t idct;
- vp8_recon_rtcd_vtable_t recon;
- vp8_subpix_rtcd_vtable_t subpix;
- vp8_loopfilter_rtcd_vtable_t loopfilter;
- vp8_postproc_rtcd_vtable_t postproc;
- int flags;
-#else
- int unused;
-#endif
-} VP8_COMMON_RTCD;
-
-typedef struct VP8Common
-{
- struct vpx_internal_error_info error;
-
- DECLARE_ALIGNED(16, short, Y1dequant[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y2dequant[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, UVdequant[QINDEX_RANGE][16]);
-
- int Width;
- int Height;
- int horiz_scale;
- int vert_scale;
-
- YUV_TYPE clr_type;
- CLAMP_TYPE clamp_type;
-
- YV12_BUFFER_CONFIG *frame_to_show;
-
- YV12_BUFFER_CONFIG yv12_fb[NUM_YV12_BUFFERS];
- int fb_idx_ref_cnt[NUM_YV12_BUFFERS];
- int new_fb_idx, lst_fb_idx, gld_fb_idx, alt_fb_idx;
-
- YV12_BUFFER_CONFIG post_proc_buffer;
- YV12_BUFFER_CONFIG temp_scale_frame;
-
- FRAME_TYPE last_frame_type; /* Save last frame's frame type for loopfilter init checking and motion search. */
- FRAME_TYPE frame_type;
-
- int show_frame;
-
- int frame_flags;
- int MBs;
- int mb_rows;
- int mb_cols;
- int mode_info_stride;
-
- /* profile settings */
- int mb_no_coeff_skip;
- int no_lpf;
- int simpler_lpf;
- int use_bilinear_mc_filter;
- int full_pixel;
-
- int base_qindex;
- int last_kf_gf_q; /* Q used on the last GF or KF */
-
- int y1dc_delta_q;
- int y2dc_delta_q;
- int y2ac_delta_q;
- int uvdc_delta_q;
- int uvac_delta_q;
-
- unsigned int frames_since_golden;
- unsigned int frames_till_alt_ref_frame;
-
- /* We allocate a MODE_INFO struct for each macroblock, together with
- an extra row on top and column on the left to simplify prediction. */
-
- MODE_INFO *mip; /* Base of allocated array */
- MODE_INFO *mi; /* Corresponds to upper left visible macroblock */
-
-
- INTERPOLATIONFILTERTYPE mcomp_filter_type;
- LOOPFILTERTYPE last_filter_type;
- LOOPFILTERTYPE filter_type;
- loop_filter_info lf_info[MAX_LOOP_FILTER+1];
- prototype_loopfilter_block((*lf_mbv));
- prototype_loopfilter_block((*lf_mbh));
- prototype_loopfilter_block((*lf_bv));
- prototype_loopfilter_block((*lf_bh));
- int filter_level;
- int last_sharpness_level;
- int sharpness_level;
-
- int refresh_last_frame; /* Two state 0 = NO, 1 = YES */
- int refresh_golden_frame; /* Two state 0 = NO, 1 = YES */
- int refresh_alt_ref_frame; /* Two state 0 = NO, 1 = YES */
-
- int copy_buffer_to_gf; /* 0 none, 1 Last to GF, 2 ARF to GF */
- int copy_buffer_to_arf; /* 0 none, 1 Last to ARF, 2 GF to ARF */
-
- int refresh_entropy_probs; /* Two state 0 = NO, 1 = YES */
-
- int ref_frame_sign_bias[MAX_REF_FRAMES]; /* Two state 0, 1 */
-
- /* Y,U,V,Y2 */
- ENTROPY_CONTEXT_PLANES *above_context; /* row of context for each plane */
- ENTROPY_CONTEXT_PLANES left_context; /* (up to) 4 contexts "" */
-
-
- /* keyframe block modes are predicted by their above, left neighbors */
-
- vp8_prob kf_bmode_prob [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES-1];
- vp8_prob kf_ymode_prob [VP8_YMODES-1]; /* keyframe "" */
- vp8_prob kf_uv_mode_prob [VP8_UV_MODES-1];
-
-
- FRAME_CONTEXT lfc; /* last frame entropy */
- FRAME_CONTEXT fc; /* this frame entropy */
-
- unsigned int current_video_frame;
-
- int near_boffset[3];
- int version;
-
- TOKEN_PARTITION multi_token_partition;
-
-#ifdef PACKET_TESTING
- VP8_HEADER oh;
-#endif
- double bitrate;
- double framerate;
-
-#if CONFIG_RUNTIME_CPU_DETECT
- VP8_COMMON_RTCD rtcd;
-#endif
- struct postproc_state postproc_state;
-} VP8_COMMON;
-
-
-int vp8_adjust_mb_lf_value(MACROBLOCKD *mbd, int filter_level);
-void vp8_init_loop_filter(VP8_COMMON *cm);
-void vp8_frame_init_loop_filter(loop_filter_info *lfi, int frame_type);
-extern void vp8_loop_filter_frame(VP8_COMMON *cm, MACROBLOCKD *mbd, int filt_val);
-
-#endif
diff --git a/vp8/common/onyxd.h b/vp8/common/onyxd.h
deleted file mode 100644
index e53bc31..0000000
--- a/vp8/common/onyxd.h
+++ /dev/null
@@ -1,68 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_VP8D_H
-#define __INC_VP8D_H
-
-
-/* Create/destroy static data structures. */
-#ifdef __cplusplus
-extern "C"
-{
-#endif
-#include "type_aliases.h"
-#include "vpx_scale/yv12config.h"
-#include "ppflags.h"
-#include "vpx_ports/mem.h"
-
- typedef void *VP8D_PTR;
- typedef struct
- {
- int Width;
- int Height;
- int Version;
- int postprocess;
- int max_threads;
- } VP8D_CONFIG;
- typedef enum
- {
- VP8_LAST_FLAG = 1,
- VP8_GOLD_FLAG = 2,
- VP8_ALT_FLAG = 4
- } VP8_REFFRAME;
-
- typedef enum
- {
- VP8D_OK = 0
- } VP8D_SETTING;
-
- void vp8dx_initialize(void);
-
- void vp8dx_set_setting(VP8D_PTR comp, VP8D_SETTING oxst, int x);
-
- int vp8dx_get_setting(VP8D_PTR comp, VP8D_SETTING oxst);
-
- int vp8dx_receive_compressed_data(VP8D_PTR comp, unsigned long size, const unsigned char *dest, INT64 time_stamp);
- int vp8dx_get_raw_frame(VP8D_PTR comp, YV12_BUFFER_CONFIG *sd, INT64 *time_stamp, INT64 *time_end_stamp, vp8_ppflags_t *flags);
-
- int vp8dx_get_reference(VP8D_PTR comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
- int vp8dx_set_reference(VP8D_PTR comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
-
- VP8D_PTR vp8dx_create_decompressor(VP8D_CONFIG *oxcf);
-
- void vp8dx_remove_decompressor(VP8D_PTR comp);
-
-#ifdef __cplusplus
-}
-#endif
-
-
-#endif
diff --git a/vp8/common/postproc.c b/vp8/common/postproc.c
deleted file mode 100644
index 5bfc7d6..0000000
--- a/vp8/common/postproc.c
+++ /dev/null
@@ -1,1107 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_scale/yv12config.h"
-#include "postproc.h"
-#include "vpx_scale/yv12extend.h"
-#include "vpx_scale/vpxscale.h"
-#include "systemdependent.h"
-
-#include <math.h>
-#include <stdlib.h>
-#include <stdio.h>
-
-#define RGB_TO_YUV(t) \
- ( (0.257*(float)(t>>16)) + (0.504*(float)(t>>8&0xff)) + (0.098*(float)(t&0xff)) + 16), \
- (-(0.148*(float)(t>>16)) - (0.291*(float)(t>>8&0xff)) + (0.439*(float)(t&0xff)) + 128), \
- ( (0.439*(float)(t>>16)) - (0.368*(float)(t>>8&0xff)) - (0.071*(float)(t&0xff)) + 128)
-
-/* global constants */
-#if CONFIG_POSTPROC_VISUALIZER
-static const unsigned char MB_PREDICTION_MODE_colors[MB_MODE_COUNT][3] =
-{
- { RGB_TO_YUV(0x98FB98) }, /* PaleGreen */
- { RGB_TO_YUV(0x00FF00) }, /* Green */
- { RGB_TO_YUV(0xADFF2F) }, /* GreenYellow */
- { RGB_TO_YUV(0x228B22) }, /* ForestGreen */
- { RGB_TO_YUV(0x006400) }, /* DarkGreen */
- { RGB_TO_YUV(0x98F5FF) }, /* Cadet Blue */
- { RGB_TO_YUV(0x6CA6CD) }, /* Sky Blue */
- { RGB_TO_YUV(0x00008B) }, /* Dark blue */
- { RGB_TO_YUV(0x551A8B) }, /* Purple */
- { RGB_TO_YUV(0xFF0000) } /* Red */
-};
-
-static const unsigned char B_PREDICTION_MODE_colors[B_MODE_COUNT][3] =
-{
- { RGB_TO_YUV(0x6633ff) }, /* Purple */
- { RGB_TO_YUV(0xcc33ff) }, /* Magenta */
- { RGB_TO_YUV(0xff33cc) }, /* Pink */
- { RGB_TO_YUV(0xff3366) }, /* Coral */
- { RGB_TO_YUV(0x3366ff) }, /* Blue */
- { RGB_TO_YUV(0xed00f5) }, /* Dark Blue */
- { RGB_TO_YUV(0x2e00b8) }, /* Dark Purple */
- { RGB_TO_YUV(0xff6633) }, /* Orange */
- { RGB_TO_YUV(0x33ccff) }, /* Light Blue */
- { RGB_TO_YUV(0x8ab800) }, /* Green */
- { RGB_TO_YUV(0xffcc33) }, /* Light Orange */
- { RGB_TO_YUV(0x33ffcc) }, /* Aqua */
- { RGB_TO_YUV(0x66ff33) }, /* Light Green */
- { RGB_TO_YUV(0xccff33) }, /* Yellow */
-};
-
-static const unsigned char MV_REFERENCE_FRAME_colors[MAX_REF_FRAMES][3] =
-{
- { RGB_TO_YUV(0x00ff00) }, /* Blue */
- { RGB_TO_YUV(0x0000ff) }, /* Green */
- { RGB_TO_YUV(0xffff00) }, /* Yellow */
- { RGB_TO_YUV(0xff0000) }, /* Red */
-};
-#endif
-
-static const short kernel5[] =
-{
- 1, 1, 4, 1, 1
-};
-
-const short vp8_rv[] =
-{
- 8, 5, 2, 2, 8, 12, 4, 9, 8, 3,
- 0, 3, 9, 0, 0, 0, 8, 3, 14, 4,
- 10, 1, 11, 14, 1, 14, 9, 6, 12, 11,
- 8, 6, 10, 0, 0, 8, 9, 0, 3, 14,
- 8, 11, 13, 4, 2, 9, 0, 3, 9, 6,
- 1, 2, 3, 14, 13, 1, 8, 2, 9, 7,
- 3, 3, 1, 13, 13, 6, 6, 5, 2, 7,
- 11, 9, 11, 8, 7, 3, 2, 0, 13, 13,
- 14, 4, 12, 5, 12, 10, 8, 10, 13, 10,
- 4, 14, 4, 10, 0, 8, 11, 1, 13, 7,
- 7, 14, 6, 14, 13, 2, 13, 5, 4, 4,
- 0, 10, 0, 5, 13, 2, 12, 7, 11, 13,
- 8, 0, 4, 10, 7, 2, 7, 2, 2, 5,
- 3, 4, 7, 3, 3, 14, 14, 5, 9, 13,
- 3, 14, 3, 6, 3, 0, 11, 8, 13, 1,
- 13, 1, 12, 0, 10, 9, 7, 6, 2, 8,
- 5, 2, 13, 7, 1, 13, 14, 7, 6, 7,
- 9, 6, 10, 11, 7, 8, 7, 5, 14, 8,
- 4, 4, 0, 8, 7, 10, 0, 8, 14, 11,
- 3, 12, 5, 7, 14, 3, 14, 5, 2, 6,
- 11, 12, 12, 8, 0, 11, 13, 1, 2, 0,
- 5, 10, 14, 7, 8, 0, 4, 11, 0, 8,
- 0, 3, 10, 5, 8, 0, 11, 6, 7, 8,
- 10, 7, 13, 9, 2, 5, 1, 5, 10, 2,
- 4, 3, 5, 6, 10, 8, 9, 4, 11, 14,
- 0, 10, 0, 5, 13, 2, 12, 7, 11, 13,
- 8, 0, 4, 10, 7, 2, 7, 2, 2, 5,
- 3, 4, 7, 3, 3, 14, 14, 5, 9, 13,
- 3, 14, 3, 6, 3, 0, 11, 8, 13, 1,
- 13, 1, 12, 0, 10, 9, 7, 6, 2, 8,
- 5, 2, 13, 7, 1, 13, 14, 7, 6, 7,
- 9, 6, 10, 11, 7, 8, 7, 5, 14, 8,
- 4, 4, 0, 8, 7, 10, 0, 8, 14, 11,
- 3, 12, 5, 7, 14, 3, 14, 5, 2, 6,
- 11, 12, 12, 8, 0, 11, 13, 1, 2, 0,
- 5, 10, 14, 7, 8, 0, 4, 11, 0, 8,
- 0, 3, 10, 5, 8, 0, 11, 6, 7, 8,
- 10, 7, 13, 9, 2, 5, 1, 5, 10, 2,
- 4, 3, 5, 6, 10, 8, 9, 4, 11, 14,
- 3, 8, 3, 7, 8, 5, 11, 4, 12, 3,
- 11, 9, 14, 8, 14, 13, 4, 3, 1, 2,
- 14, 6, 5, 4, 4, 11, 4, 6, 2, 1,
- 5, 8, 8, 12, 13, 5, 14, 10, 12, 13,
- 0, 9, 5, 5, 11, 10, 13, 9, 10, 13,
-};
-
-
-extern void vp8_blit_text(const char *msg, unsigned char *address, const int pitch);
-extern void vp8_blit_line(int x0, int x1, int y0, int y1, unsigned char *image, const int pitch);
-/***********************************************************************************************************
- */
-void vp8_post_proc_down_and_across_c
-(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- int src_pixels_per_line,
- int dst_pixels_per_line,
- int rows,
- int cols,
- int flimit
-)
-{
- unsigned char *p_src, *p_dst;
- int row;
- int col;
- int i;
- int v;
- int pitch = src_pixels_per_line;
- unsigned char d[8];
- (void)dst_pixels_per_line;
-
- for (row = 0; row < rows; row++)
- {
- /* post_proc_down for one row */
- p_src = src_ptr;
- p_dst = dst_ptr;
-
- for (col = 0; col < cols; col++)
- {
-
- int kernel = 4;
- int v = p_src[col];
-
- for (i = -2; i <= 2; i++)
- {
- if (abs(v - p_src[col+i*pitch]) > flimit)
- goto down_skip_convolve;
-
- kernel += kernel5[2+i] * p_src[col+i*pitch];
- }
-
- v = (kernel >> 3);
- down_skip_convolve:
- p_dst[col] = v;
- }
-
- /* now post_proc_across */
- p_src = dst_ptr;
- p_dst = dst_ptr;
-
- for (i = 0; i < 8; i++)
- d[i] = p_src[i];
-
- for (col = 0; col < cols; col++)
- {
- int kernel = 4;
- v = p_src[col];
-
- d[col&7] = v;
-
- for (i = -2; i <= 2; i++)
- {
- if (abs(v - p_src[col+i]) > flimit)
- goto across_skip_convolve;
-
- kernel += kernel5[2+i] * p_src[col+i];
- }
-
- d[col&7] = (kernel >> 3);
- across_skip_convolve:
-
- if (col >= 2)
- p_dst[col-2] = d[(col-2)&7];
- }
-
- /* handle the last two pixels */
- p_dst[col-2] = d[(col-2)&7];
- p_dst[col-1] = d[(col-1)&7];
-
-
- /* next row */
- src_ptr += pitch;
- dst_ptr += pitch;
- }
-}
-
-static int q2mbl(int x)
-{
- if (x < 20) x = 20;
-
- x = 50 + (x - 50) * 10 / 8;
- return x * x / 3;
-}
-void vp8_mbpost_proc_across_ip_c(unsigned char *src, int pitch, int rows, int cols, int flimit)
-{
- int r, c, i;
-
- unsigned char *s = src;
- unsigned char d[16];
-
-
- for (r = 0; r < rows; r++)
- {
- int sumsq = 0;
- int sum = 0;
-
- for (i = -8; i <= 6; i++)
- {
- sumsq += s[i] * s[i];
- sum += s[i];
- d[i+8] = 0;
- }
-
- for (c = 0; c < cols + 8; c++)
- {
- int x = s[c+7] - s[c-8];
- int y = s[c+7] + s[c-8];
-
- sum += x;
- sumsq += x * y;
-
- d[c&15] = s[c];
-
- if (sumsq * 15 - sum * sum < flimit)
- {
- d[c&15] = (8 + sum + s[c]) >> 4;
- }
-
- s[c-8] = d[(c-8)&15];
- }
-
- s += pitch;
- }
-}
-
-
-
-
-
-void vp8_mbpost_proc_down_c(unsigned char *dst, int pitch, int rows, int cols, int flimit)
-{
- int r, c, i;
- const short *rv3 = &vp8_rv[63&rand()];
-
- for (c = 0; c < cols; c++)
- {
- unsigned char *s = &dst[c];
- int sumsq = 0;
- int sum = 0;
- unsigned char d[16];
- const short *rv2 = rv3 + ((c * 17) & 127);
-
- for (i = -8; i <= 6; i++)
- {
- sumsq += s[i*pitch] * s[i*pitch];
- sum += s[i*pitch];
- }
-
- for (r = 0; r < rows + 8; r++)
- {
- sumsq += s[7*pitch] * s[ 7*pitch] - s[-8*pitch] * s[-8*pitch];
- sum += s[7*pitch] - s[-8*pitch];
- d[r&15] = s[0];
-
- if (sumsq * 15 - sum * sum < flimit)
- {
- d[r&15] = (rv2[r&127] + sum + s[0]) >> 4;
- }
-
- s[-8*pitch] = d[(r-8)&15];
- s += pitch;
- }
- }
-}
-
-
-static void vp8_deblock_and_de_macro_block(YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *post,
- int q,
- int low_var_thresh,
- int flag,
- vp8_postproc_rtcd_vtable_t *rtcd)
-{
- double level = 6.0e-05 * q * q * q - .0067 * q * q + .306 * q + .0065;
- int ppl = (int)(level + .5);
- (void) low_var_thresh;
- (void) flag;
-
- POSTPROC_INVOKE(rtcd, downacross)(source->y_buffer, post->y_buffer, source->y_stride, post->y_stride, source->y_height, source->y_width, ppl);
- POSTPROC_INVOKE(rtcd, across)(post->y_buffer, post->y_stride, post->y_height, post->y_width, q2mbl(q));
- POSTPROC_INVOKE(rtcd, down)(post->y_buffer, post->y_stride, post->y_height, post->y_width, q2mbl(q));
-
- POSTPROC_INVOKE(rtcd, downacross)(source->u_buffer, post->u_buffer, source->uv_stride, post->uv_stride, source->uv_height, source->uv_width, ppl);
- POSTPROC_INVOKE(rtcd, downacross)(source->v_buffer, post->v_buffer, source->uv_stride, post->uv_stride, source->uv_height, source->uv_width, ppl);
-
-}
-
-void vp8_deblock(YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *post,
- int q,
- int low_var_thresh,
- int flag,
- vp8_postproc_rtcd_vtable_t *rtcd)
-{
- double level = 6.0e-05 * q * q * q - .0067 * q * q + .306 * q + .0065;
- int ppl = (int)(level + .5);
- (void) low_var_thresh;
- (void) flag;
-
- POSTPROC_INVOKE(rtcd, downacross)(source->y_buffer, post->y_buffer, source->y_stride, post->y_stride, source->y_height, source->y_width, ppl);
- POSTPROC_INVOKE(rtcd, downacross)(source->u_buffer, post->u_buffer, source->uv_stride, post->uv_stride, source->uv_height, source->uv_width, ppl);
- POSTPROC_INVOKE(rtcd, downacross)(source->v_buffer, post->v_buffer, source->uv_stride, post->uv_stride, source->uv_height, source->uv_width, ppl);
-}
-
-void vp8_de_noise(YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *post,
- int q,
- int low_var_thresh,
- int flag,
- vp8_postproc_rtcd_vtable_t *rtcd)
-{
- double level = 6.0e-05 * q * q * q - .0067 * q * q + .306 * q + .0065;
- int ppl = (int)(level + .5);
- (void) post;
- (void) low_var_thresh;
- (void) flag;
-
- POSTPROC_INVOKE(rtcd, downacross)(
- source->y_buffer + 2 * source->y_stride + 2,
- source->y_buffer + 2 * source->y_stride + 2,
- source->y_stride,
- source->y_stride,
- source->y_height - 4,
- source->y_width - 4,
- ppl);
- POSTPROC_INVOKE(rtcd, downacross)(
- source->u_buffer + 2 * source->uv_stride + 2,
- source->u_buffer + 2 * source->uv_stride + 2,
- source->uv_stride,
- source->uv_stride,
- source->uv_height - 4,
- source->uv_width - 4, ppl);
- POSTPROC_INVOKE(rtcd, downacross)(
- source->v_buffer + 2 * source->uv_stride + 2,
- source->v_buffer + 2 * source->uv_stride + 2,
- source->uv_stride,
- source->uv_stride,
- source->uv_height - 4,
- source->uv_width - 4, ppl);
-
-}
-
-double vp8_gaussian(double sigma, double mu, double x)
-{
- return 1 / (sigma * sqrt(2.0 * 3.14159265)) *
- (exp(-(x - mu) * (x - mu) / (2 * sigma * sigma)));
-}
-
-extern void (*vp8_clear_system_state)(void);
-
-
-static void fillrd(struct postproc_state *state, int q, int a)
-{
- char char_dist[300];
-
- double sigma;
- int ai = a, qi = q, i;
-
- vp8_clear_system_state();
-
-
- sigma = ai + .5 + .6 * (63 - qi) / 63.0;
-
- /* set up a lookup table of 256 entries that matches
- * a gaussian distribution with sigma determined by q.
- */
- {
- double i;
- int next, j;
-
- next = 0;
-
- for (i = -32; i < 32; i++)
- {
- int a = (int)(.5 + 256 * vp8_gaussian(sigma, 0, i));
-
- if (a)
- {
- for (j = 0; j < a; j++)
- {
- char_dist[next+j] = (char) i;
- }
-
- next = next + j;
- }
-
- }
-
- for (next = next; next < 256; next++)
- char_dist[next] = 0;
-
- }
-
- for (i = 0; i < 3072; i++)
- {
- state->noise[i] = char_dist[rand() & 0xff];
- }
-
- for (i = 0; i < 16; i++)
- {
- state->blackclamp[i] = -char_dist[0];
- state->whiteclamp[i] = -char_dist[0];
- state->bothclamp[i] = -2 * char_dist[0];
- }
-
- state->last_q = q;
- state->last_noise = a;
-}
-
-/****************************************************************************
- *
- * ROUTINE : plane_add_noise_c
- *
- * INPUTS : unsigned char *Start starting address of buffer to add gaussian
- * noise to
- * unsigned int Width width of plane
- * unsigned int Height height of plane
- * int Pitch distance between subsequent lines of frame
- * int q quantizer used to determine amount of noise
- * to add
- *
- * OUTPUTS : None.
- *
- * RETURNS : void.
- *
- * FUNCTION : adds gaussian noise to a plane of pixels
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8_plane_add_noise_c(unsigned char *Start, char *noise,
- char blackclamp[16],
- char whiteclamp[16],
- char bothclamp[16],
- unsigned int Width, unsigned int Height, int Pitch)
-{
- unsigned int i, j;
-
- for (i = 0; i < Height; i++)
- {
- unsigned char *Pos = Start + i * Pitch;
- char *Ref = (char *)(noise + (rand() & 0xff));
-
- for (j = 0; j < Width; j++)
- {
- if (Pos[j] < blackclamp[0])
- Pos[j] = blackclamp[0];
-
- if (Pos[j] > 255 + whiteclamp[0])
- Pos[j] = 255 + whiteclamp[0];
-
- Pos[j] += Ref[j];
- }
- }
-}
-
-/* Blend the macro block with a solid colored square. Leave the
- * edges unblended to give distinction to macro blocks in areas
- * filled with the same color block.
- */
-void vp8_blend_mb_inner_c (unsigned char *y, unsigned char *u, unsigned char *v,
- int y1, int u1, int v1, int alpha, int stride)
-{
- int i, j;
- int y1_const = y1*((1<<16)-alpha);
- int u1_const = u1*((1<<16)-alpha);
- int v1_const = v1*((1<<16)-alpha);
-
- y += 2*stride + 2;
- for (i = 0; i < 12; i++)
- {
- for (j = 0; j < 12; j++)
- {
- y[j] = (y[j]*alpha + y1_const)>>16;
- }
- y += stride;
- }
-
- stride >>= 1;
-
- u += stride + 1;
- v += stride + 1;
-
- for (i = 0; i < 6; i++)
- {
- for (j = 0; j < 6; j++)
- {
- u[j] = (u[j]*alpha + u1_const)>>16;
- v[j] = (v[j]*alpha + v1_const)>>16;
- }
- u += stride;
- v += stride;
- }
-}
-
-/* Blend only the edge of the macro block. Leave center
- * unblended to allow for other visualizations to be layered.
- */
-void vp8_blend_mb_outer_c (unsigned char *y, unsigned char *u, unsigned char *v,
- int y1, int u1, int v1, int alpha, int stride)
-{
- int i, j;
- int y1_const = y1*((1<<16)-alpha);
- int u1_const = u1*((1<<16)-alpha);
- int v1_const = v1*((1<<16)-alpha);
-
- for (i = 0; i < 2; i++)
- {
- for (j = 0; j < 16; j++)
- {
- y[j] = (y[j]*alpha + y1_const)>>16;
- }
- y += stride;
- }
-
- for (i = 0; i < 12; i++)
- {
- y[0] = (y[0]*alpha + y1_const)>>16;
- y[1] = (y[1]*alpha + y1_const)>>16;
- y[14] = (y[14]*alpha + y1_const)>>16;
- y[15] = (y[15]*alpha + y1_const)>>16;
- y += stride;
- }
-
- for (i = 0; i < 2; i++)
- {
- for (j = 0; j < 16; j++)
- {
- y[j] = (y[j]*alpha + y1_const)>>16;
- }
- y += stride;
- }
-
- stride >>= 1;
-
- for (j = 0; j < 8; j++)
- {
- u[j] = (u[j]*alpha + u1_const)>>16;
- v[j] = (v[j]*alpha + v1_const)>>16;
- }
- u += stride;
- v += stride;
-
- for (i = 0; i < 6; i++)
- {
- u[0] = (u[0]*alpha + u1_const)>>16;
- v[0] = (v[0]*alpha + v1_const)>>16;
-
- u[7] = (u[7]*alpha + u1_const)>>16;
- v[7] = (v[7]*alpha + v1_const)>>16;
-
- u += stride;
- v += stride;
- }
-
- for (j = 0; j < 8; j++)
- {
- u[j] = (u[j]*alpha + u1_const)>>16;
- v[j] = (v[j]*alpha + v1_const)>>16;
- }
-}
-
-void vp8_blend_b_c (unsigned char *y, unsigned char *u, unsigned char *v,
- int y1, int u1, int v1, int alpha, int stride)
-{
- int i, j;
- int y1_const = y1*((1<<16)-alpha);
- int u1_const = u1*((1<<16)-alpha);
- int v1_const = v1*((1<<16)-alpha);
-
- for (i = 0; i < 4; i++)
- {
- for (j = 0; j < 4; j++)
- {
- y[j] = (y[j]*alpha + y1_const)>>16;
- }
- y += stride;
- }
-
- stride >>= 1;
-
- for (i = 0; i < 2; i++)
- {
- for (j = 0; j < 2; j++)
- {
- u[j] = (u[j]*alpha + u1_const)>>16;
- v[j] = (v[j]*alpha + v1_const)>>16;
- }
- u += stride;
- v += stride;
- }
-}
-
-static void constrain_line (int x0, int *x1, int y0, int *y1, int width, int height)
-{
- int dx;
- int dy;
-
- if (*x1 > width)
- {
- dx = *x1 - x0;
- dy = *y1 - y0;
-
- *x1 = width;
- if (dx)
- *y1 = ((width-x0)*dy)/dx + y0;
- }
- if (*x1 < 0)
- {
- dx = *x1 - x0;
- dy = *y1 - y0;
-
- *x1 = 0;
- if (dx)
- *y1 = ((0-x0)*dy)/dx + y0;
- }
- if (*y1 > height)
- {
- dx = *x1 - x0;
- dy = *y1 - y0;
-
- *y1 = height;
- if (dy)
- *x1 = ((height-y0)*dx)/dy + x0;
- }
- if (*y1 < 0)
- {
- dx = *x1 - x0;
- dy = *y1 - y0;
-
- *y1 = 0;
- if (dy)
- *x1 = ((0-y0)*dx)/dy + x0;
- }
-}
-
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define RTCD_VTABLE(oci) (&(oci)->rtcd.postproc)
-#else
-#define RTCD_VTABLE(oci) NULL
-#endif
-
-int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *ppflags)
-{
- int q = oci->filter_level * 10 / 6;
- int flags = ppflags->post_proc_flag;
- int deblock_level = ppflags->deblocking_level;
- int noise_level = ppflags->noise_level;
-
- if (!oci->frame_to_show)
- return -1;
-
- if (q > 63)
- q = 63;
-
- if (!flags)
- {
- *dest = *oci->frame_to_show;
-
- /* handle problem with extending borders */
- dest->y_width = oci->Width;
- dest->y_height = oci->Height;
- dest->uv_height = dest->y_height / 2;
- return 0;
-
- }
-
-#if ARCH_X86||ARCH_X86_64
- vpx_reset_mmx_state();
-#endif
-
- if (flags & VP8D_DEMACROBLOCK)
- {
- vp8_deblock_and_de_macro_block(oci->frame_to_show, &oci->post_proc_buffer,
- q + (deblock_level - 5) * 10, 1, 0, RTCD_VTABLE(oci));
- }
- else if (flags & VP8D_DEBLOCK)
- {
- vp8_deblock(oci->frame_to_show, &oci->post_proc_buffer,
- q, 1, 0, RTCD_VTABLE(oci));
- }
- else
- {
- vp8_yv12_copy_frame_ptr(oci->frame_to_show, &oci->post_proc_buffer);
- }
-
- if (flags & VP8D_ADDNOISE)
- {
- if (oci->postproc_state.last_q != q
- || oci->postproc_state.last_noise != noise_level)
- {
- fillrd(&oci->postproc_state, 63 - q, noise_level);
- }
-
- POSTPROC_INVOKE(RTCD_VTABLE(oci), addnoise)
- (oci->post_proc_buffer.y_buffer,
- oci->postproc_state.noise,
- oci->postproc_state.blackclamp,
- oci->postproc_state.whiteclamp,
- oci->postproc_state.bothclamp,
- oci->post_proc_buffer.y_width, oci->post_proc_buffer.y_height,
- oci->post_proc_buffer.y_stride);
- }
-
-#if CONFIG_POSTPROC_VISUALIZER
- if (flags & VP8D_DEBUG_TXT_FRAME_INFO)
- {
- char message[512];
- sprintf(message, "F%1dG%1dQ%3dF%3dP%d_s%dx%d",
- (oci->frame_type == KEY_FRAME),
- oci->refresh_golden_frame,
- oci->base_qindex,
- oci->filter_level,
- flags,
- oci->mb_cols, oci->mb_rows);
- vp8_blit_text(message, oci->post_proc_buffer.y_buffer, oci->post_proc_buffer.y_stride);
- }
-
- if (flags & VP8D_DEBUG_TXT_MBLK_MODES)
- {
- int i, j;
- unsigned char *y_ptr;
- YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
- int mb_rows = post->y_height >> 4;
- int mb_cols = post->y_width >> 4;
- int mb_index = 0;
- MODE_INFO *mi = oci->mi;
-
- y_ptr = post->y_buffer + 4 * post->y_stride + 4;
-
- /* vp8_filter each macro block */
- for (i = 0; i < mb_rows; i++)
- {
- for (j = 0; j < mb_cols; j++)
- {
- char zz[4];
-
- sprintf(zz, "%c", mi[mb_index].mbmi.mode + 'a');
-
- vp8_blit_text(zz, y_ptr, post->y_stride);
- mb_index ++;
- y_ptr += 16;
- }
-
- mb_index ++; /* border */
- y_ptr += post->y_stride * 16 - post->y_width;
-
- }
- }
-
- if (flags & VP8D_DEBUG_TXT_DC_DIFF)
- {
- int i, j;
- unsigned char *y_ptr;
- YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
- int mb_rows = post->y_height >> 4;
- int mb_cols = post->y_width >> 4;
- int mb_index = 0;
- MODE_INFO *mi = oci->mi;
-
- y_ptr = post->y_buffer + 4 * post->y_stride + 4;
-
- /* vp8_filter each macro block */
- for (i = 0; i < mb_rows; i++)
- {
- for (j = 0; j < mb_cols; j++)
- {
- char zz[4];
-
- if (oci->frame_type == KEY_FRAME)
- sprintf(zz, "a");
- else
- sprintf(zz, "%c", mi[mb_index].mbmi.dc_diff + '0');
-
- vp8_blit_text(zz, y_ptr, post->y_stride);
- mb_index ++;
- y_ptr += 16;
- }
-
- mb_index ++; /* border */
- y_ptr += post->y_stride * 16 - post->y_width;
-
- }
- }
-
- if (flags & VP8D_DEBUG_TXT_RATE_INFO)
- {
- char message[512];
- sprintf(message, "Bitrate: %10.2f frame_rate: %10.2f ", oci->bitrate, oci->framerate);
- vp8_blit_text(message, oci->post_proc_buffer.y_buffer, oci->post_proc_buffer.y_stride);
- }
-
- /* Draw motion vectors */
- if ((flags & VP8D_DEBUG_DRAW_MV) && ppflags->display_mv_flag)
- {
- YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
- int width = post->y_width;
- int height = post->y_height;
- int mb_cols = width >> 4;
- unsigned char *y_buffer = oci->post_proc_buffer.y_buffer;
- int y_stride = oci->post_proc_buffer.y_stride;
- MODE_INFO *mi = oci->mi;
- int x0, y0;
-
- for (y0 = 0; y0 < height; y0 += 16)
- {
- for (x0 = 0; x0 < width; x0 += 16)
- {
- int x1, y1;
-
- if (!(ppflags->display_mv_flag & (1<<mi->mbmi.mode)))
- {
- mi++;
- continue;
- }
-
- if (mi->mbmi.mode == SPLITMV)
- {
- switch (mi->mbmi.partitioning)
- {
- case 0 : /* mv_top_bottom */
- {
- B_MODE_INFO *bmi = &mi->bmi[0];
- MV *mv = &bmi->mv.as_mv;
-
- x1 = x0 + 8 + (mv->col >> 3);
- y1 = y0 + 4 + (mv->row >> 3);
-
- constrain_line (x0+8, &x1, y0+4, &y1, width, height);
- vp8_blit_line (x0+8, x1, y0+4, y1, y_buffer, y_stride);
-
- bmi = &mi->bmi[8];
-
- x1 = x0 + 8 + (mv->col >> 3);
- y1 = y0 +12 + (mv->row >> 3);
-
- constrain_line (x0+8, &x1, y0+12, &y1, width, height);
- vp8_blit_line (x0+8, x1, y0+12, y1, y_buffer, y_stride);
-
- break;
- }
- case 1 : /* mv_left_right */
- {
- B_MODE_INFO *bmi = &mi->bmi[0];
- MV *mv = &bmi->mv.as_mv;
-
- x1 = x0 + 4 + (mv->col >> 3);
- y1 = y0 + 8 + (mv->row >> 3);
-
- constrain_line (x0+4, &x1, y0+8, &y1, width, height);
- vp8_blit_line (x0+4, x1, y0+8, y1, y_buffer, y_stride);
-
- bmi = &mi->bmi[2];
-
- x1 = x0 +12 + (mv->col >> 3);
- y1 = y0 + 8 + (mv->row >> 3);
-
- constrain_line (x0+12, &x1, y0+8, &y1, width, height);
- vp8_blit_line (x0+12, x1, y0+8, y1, y_buffer, y_stride);
-
- break;
- }
- case 2 : /* mv_quarters */
- {
- B_MODE_INFO *bmi = &mi->bmi[0];
- MV *mv = &bmi->mv.as_mv;
-
- x1 = x0 + 4 + (mv->col >> 3);
- y1 = y0 + 4 + (mv->row >> 3);
-
- constrain_line (x0+4, &x1, y0+4, &y1, width, height);
- vp8_blit_line (x0+4, x1, y0+4, y1, y_buffer, y_stride);
-
- bmi = &mi->bmi[2];
-
- x1 = x0 +12 + (mv->col >> 3);
- y1 = y0 + 4 + (mv->row >> 3);
-
- constrain_line (x0+12, &x1, y0+4, &y1, width, height);
- vp8_blit_line (x0+12, x1, y0+4, y1, y_buffer, y_stride);
-
- bmi = &mi->bmi[8];
-
- x1 = x0 + 4 + (mv->col >> 3);
- y1 = y0 +12 + (mv->row >> 3);
-
- constrain_line (x0+4, &x1, y0+12, &y1, width, height);
- vp8_blit_line (x0+4, x1, y0+12, y1, y_buffer, y_stride);
-
- bmi = &mi->bmi[10];
-
- x1 = x0 +12 + (mv->col >> 3);
- y1 = y0 +12 + (mv->row >> 3);
-
- constrain_line (x0+12, &x1, y0+12, &y1, width, height);
- vp8_blit_line (x0+12, x1, y0+12, y1, y_buffer, y_stride);
- break;
- }
- default :
- {
- B_MODE_INFO *bmi = mi->bmi;
- int bx0, by0;
-
- for (by0 = y0; by0 < (y0+16); by0 += 4)
- {
- for (bx0 = x0; bx0 < (x0+16); bx0 += 4)
- {
- MV *mv = &bmi->mv.as_mv;
-
- x1 = bx0 + 2 + (mv->col >> 3);
- y1 = by0 + 2 + (mv->row >> 3);
-
- constrain_line (bx0+2, &x1, by0+2, &y1, width, height);
- vp8_blit_line (bx0+2, x1, by0+2, y1, y_buffer, y_stride);
-
- bmi++;
- }
- }
- }
- }
- }
- else if (mi->mbmi.mode >= NEARESTMV)
- {
- MV *mv = &mi->mbmi.mv.as_mv;
- const int lx0 = x0 + 8;
- const int ly0 = y0 + 8;
-
- x1 = lx0 + (mv->col >> 3);
- y1 = ly0 + (mv->row >> 3);
-
- if (x1 != lx0 && y1 != ly0)
- {
- constrain_line (lx0, &x1, ly0-1, &y1, width, height);
- vp8_blit_line (lx0, x1, ly0-1, y1, y_buffer, y_stride);
-
- constrain_line (lx0, &x1, ly0+1, &y1, width, height);
- vp8_blit_line (lx0, x1, ly0+1, y1, y_buffer, y_stride);
- }
- else
- vp8_blit_line (lx0, x1, ly0, y1, y_buffer, y_stride);
- }
-
- mi++;
- }
- mi++;
- }
- }
-
- /* Color in block modes */
- if ((flags & VP8D_DEBUG_CLR_BLK_MODES)
- && (ppflags->display_mb_modes_flag || ppflags->display_b_modes_flag))
- {
- int y, x;
- YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
- int width = post->y_width;
- int height = post->y_height;
- unsigned char *y_ptr = oci->post_proc_buffer.y_buffer;
- unsigned char *u_ptr = oci->post_proc_buffer.u_buffer;
- unsigned char *v_ptr = oci->post_proc_buffer.v_buffer;
- int y_stride = oci->post_proc_buffer.y_stride;
- MODE_INFO *mi = oci->mi;
-
- for (y = 0; y < height; y += 16)
- {
- for (x = 0; x < width; x += 16)
- {
- int Y = 0, U = 0, V = 0;
-
- if (mi->mbmi.mode == B_PRED &&
- ((ppflags->display_mb_modes_flag & B_PRED) || ppflags->display_b_modes_flag))
- {
- int by, bx;
- unsigned char *yl, *ul, *vl;
- B_MODE_INFO *bmi = mi->bmi;
-
- yl = y_ptr + x;
- ul = u_ptr + (x>>1);
- vl = v_ptr + (x>>1);
-
- for (by = 0; by < 16; by += 4)
- {
- for (bx = 0; bx < 16; bx += 4)
- {
- if ((ppflags->display_b_modes_flag & (1<<mi->mbmi.mode))
- || (ppflags->display_mb_modes_flag & B_PRED))
- {
- Y = B_PREDICTION_MODE_colors[bmi->mode][0];
- U = B_PREDICTION_MODE_colors[bmi->mode][1];
- V = B_PREDICTION_MODE_colors[bmi->mode][2];
-
- POSTPROC_INVOKE(RTCD_VTABLE(oci), blend_b)
- (yl+bx, ul+(bx>>1), vl+(bx>>1), Y, U, V, 0xc000, y_stride);
- }
- bmi++;
- }
-
- yl += y_stride*4;
- ul += y_stride*1;
- vl += y_stride*1;
- }
- }
- else if (ppflags->display_mb_modes_flag & (1<<mi->mbmi.mode))
- {
- Y = MB_PREDICTION_MODE_colors[mi->mbmi.mode][0];
- U = MB_PREDICTION_MODE_colors[mi->mbmi.mode][1];
- V = MB_PREDICTION_MODE_colors[mi->mbmi.mode][2];
-
- POSTPROC_INVOKE(RTCD_VTABLE(oci), blend_mb_inner)
- (y_ptr+x, u_ptr+(x>>1), v_ptr+(x>>1), Y, U, V, 0xc000, y_stride);
- }
-
- mi++;
- }
- y_ptr += y_stride*16;
- u_ptr += y_stride*4;
- v_ptr += y_stride*4;
-
- mi++;
- }
- }
-
- /* Color in frame reference blocks */
- if ((flags & VP8D_DEBUG_CLR_FRM_REF_BLKS) && ppflags->display_ref_frame_flag)
- {
- int y, x;
- YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
- int width = post->y_width;
- int height = post->y_height;
- unsigned char *y_ptr = oci->post_proc_buffer.y_buffer;
- unsigned char *u_ptr = oci->post_proc_buffer.u_buffer;
- unsigned char *v_ptr = oci->post_proc_buffer.v_buffer;
- int y_stride = oci->post_proc_buffer.y_stride;
- MODE_INFO *mi = oci->mi;
-
- for (y = 0; y < height; y += 16)
- {
- for (x = 0; x < width; x +=16)
- {
- int Y = 0, U = 0, V = 0;
-
- if (ppflags->display_ref_frame_flag & (1<<mi->mbmi.ref_frame))
- {
- Y = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][0];
- U = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][1];
- V = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][2];
-
- POSTPROC_INVOKE(RTCD_VTABLE(oci), blend_mb_outer)
- (y_ptr+x, u_ptr+(x>>1), v_ptr+(x>>1), Y, U, V, 0xc000, y_stride);
- }
-
- mi++;
- }
- y_ptr += y_stride*16;
- u_ptr += y_stride*4;
- v_ptr += y_stride*4;
-
- mi++;
- }
- }
-#endif
-
- *dest = oci->post_proc_buffer;
-
- /* handle problem with extending borders */
- dest->y_width = oci->Width;
- dest->y_height = oci->Height;
- dest->uv_height = dest->y_height / 2;
- return 0;
-}
diff --git a/vp8/common/postproc.h b/vp8/common/postproc.h
deleted file mode 100644
index c641b9c..0000000
--- a/vp8/common/postproc.h
+++ /dev/null
@@ -1,130 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef POSTPROC_H
-#define POSTPROC_H
-
-#define prototype_postproc_inplace(sym)\
- void sym (unsigned char *dst, int pitch, int rows, int cols,int flimit)
-
-#define prototype_postproc(sym)\
- void sym (unsigned char *src, unsigned char *dst, int src_pitch,\
- int dst_pitch, int rows, int cols, int flimit)
-
-#define prototype_postproc_addnoise(sym) \
- void sym (unsigned char *s, char *noise, char blackclamp[16],\
- char whiteclamp[16], char bothclamp[16],\
- unsigned int w, unsigned int h, int pitch)
-
-#define prototype_postproc_blend_mb_inner(sym)\
- void sym (unsigned char *y, unsigned char *u, unsigned char *v,\
- int y1, int u1, int v1, int alpha, int stride)
-
-#define prototype_postproc_blend_mb_outer(sym)\
- void sym (unsigned char *y, unsigned char *u, unsigned char *v,\
- int y1, int u1, int v1, int alpha, int stride)
-
-#define prototype_postproc_blend_b(sym)\
- void sym (unsigned char *y, unsigned char *u, unsigned char *v,\
- int y1, int u1, int v1, int alpha, int stride)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/postproc_x86.h"
-#endif
-
-#ifndef vp8_postproc_down
-#define vp8_postproc_down vp8_mbpost_proc_down_c
-#endif
-extern prototype_postproc_inplace(vp8_postproc_down);
-
-#ifndef vp8_postproc_across
-#define vp8_postproc_across vp8_mbpost_proc_across_ip_c
-#endif
-extern prototype_postproc_inplace(vp8_postproc_across);
-
-#ifndef vp8_postproc_downacross
-#define vp8_postproc_downacross vp8_post_proc_down_and_across_c
-#endif
-extern prototype_postproc(vp8_postproc_downacross);
-
-#ifndef vp8_postproc_addnoise
-#define vp8_postproc_addnoise vp8_plane_add_noise_c
-#endif
-extern prototype_postproc_addnoise(vp8_postproc_addnoise);
-
-#ifndef vp8_postproc_blend_mb_inner
-#define vp8_postproc_blend_mb_inner vp8_blend_mb_inner_c
-#endif
-extern prototype_postproc_blend_mb_inner(vp8_postproc_blend_mb_inner);
-
-#ifndef vp8_postproc_blend_mb_outer
-#define vp8_postproc_blend_mb_outer vp8_blend_mb_outer_c
-#endif
-extern prototype_postproc_blend_mb_outer(vp8_postproc_blend_mb_outer);
-
-#ifndef vp8_postproc_blend_b
-#define vp8_postproc_blend_b vp8_blend_b_c
-#endif
-extern prototype_postproc_blend_b(vp8_postproc_blend_b);
-
-typedef prototype_postproc((*vp8_postproc_fn_t));
-typedef prototype_postproc_inplace((*vp8_postproc_inplace_fn_t));
-typedef prototype_postproc_addnoise((*vp8_postproc_addnoise_fn_t));
-typedef prototype_postproc_blend_mb_inner((*vp8_postproc_blend_mb_inner_fn_t));
-typedef prototype_postproc_blend_mb_outer((*vp8_postproc_blend_mb_outer_fn_t));
-typedef prototype_postproc_blend_b((*vp8_postproc_blend_b_fn_t));
-typedef struct
-{
- vp8_postproc_inplace_fn_t down;
- vp8_postproc_inplace_fn_t across;
- vp8_postproc_fn_t downacross;
- vp8_postproc_addnoise_fn_t addnoise;
- vp8_postproc_blend_mb_inner_fn_t blend_mb_inner;
- vp8_postproc_blend_mb_outer_fn_t blend_mb_outer;
- vp8_postproc_blend_b_fn_t blend_b;
-} vp8_postproc_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define POSTPROC_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define POSTPROC_INVOKE(ctx,fn) vp8_postproc_##fn
-#endif
-
-#include "vpx_ports/mem.h"
-struct postproc_state
-{
- int last_q;
- int last_noise;
- char noise[3072];
- DECLARE_ALIGNED(16, char, blackclamp[16]);
- DECLARE_ALIGNED(16, char, whiteclamp[16]);
- DECLARE_ALIGNED(16, char, bothclamp[16]);
-};
-#include "onyxc_int.h"
-#include "ppflags.h"
-int vp8_post_proc_frame(struct VP8Common *oci, YV12_BUFFER_CONFIG *dest,
- vp8_ppflags_t *flags);
-
-
-void vp8_de_noise(YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *post,
- int q,
- int low_var_thresh,
- int flag,
- vp8_postproc_rtcd_vtable_t *rtcd);
-
-void vp8_deblock(YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *post,
- int q,
- int low_var_thresh,
- int flag,
- vp8_postproc_rtcd_vtable_t *rtcd);
-#endif
diff --git a/vp8/common/ppc/loopfilter_altivec.c b/vp8/common/ppc/loopfilter_altivec.c
deleted file mode 100644
index e602fee..0000000
--- a/vp8/common/ppc/loopfilter_altivec.c
+++ /dev/null
@@ -1,143 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "loopfilter.h"
-#include "onyxc_int.h"
-
-typedef void loop_filter_function_y_ppc
-(
- unsigned char *s, // source pointer
- int p, // pitch
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh
-);
-
-typedef void loop_filter_function_uv_ppc
-(
- unsigned char *u, // source pointer
- unsigned char *v, // source pointer
- int p, // pitch
- const signed char *flimit,
- const signed char *limit,
- const signed char *thresh
-);
-
-typedef void loop_filter_function_s_ppc
-(
- unsigned char *s, // source pointer
- int p, // pitch
- const signed char *flimit
-);
-
-loop_filter_function_y_ppc mbloop_filter_horizontal_edge_y_ppc;
-loop_filter_function_y_ppc mbloop_filter_vertical_edge_y_ppc;
-loop_filter_function_y_ppc loop_filter_horizontal_edge_y_ppc;
-loop_filter_function_y_ppc loop_filter_vertical_edge_y_ppc;
-
-loop_filter_function_uv_ppc mbloop_filter_horizontal_edge_uv_ppc;
-loop_filter_function_uv_ppc mbloop_filter_vertical_edge_uv_ppc;
-loop_filter_function_uv_ppc loop_filter_horizontal_edge_uv_ppc;
-loop_filter_function_uv_ppc loop_filter_vertical_edge_uv_ppc;
-
-loop_filter_function_s_ppc loop_filter_simple_horizontal_edge_ppc;
-loop_filter_function_s_ppc loop_filter_simple_vertical_edge_ppc;
-
-// Horizontal MB filtering
-void loop_filter_mbh_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void)simpler_lpf;
- mbloop_filter_horizontal_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr);
-
- if (u_ptr)
- mbloop_filter_horizontal_edge_uv_ppc(u_ptr, v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr);
-}
-
-void loop_filter_mbhs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void)simpler_lpf;
- (void)u_ptr;
- (void)v_ptr;
- (void)uv_stride;
- loop_filter_simple_horizontal_edge_ppc(y_ptr, y_stride, lfi->mbflim);
-}
-
-// Vertical MB Filtering
-void loop_filter_mbv_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void)simpler_lpf;
- mbloop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr);
-
- if (u_ptr)
- mbloop_filter_vertical_edge_uv_ppc(u_ptr, v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr);
-}
-
-void loop_filter_mbvs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void)simpler_lpf;
- (void)u_ptr;
- (void)v_ptr;
- (void)uv_stride;
- loop_filter_simple_vertical_edge_ppc(y_ptr, y_stride, lfi->mbflim);
-}
-
-// Horizontal B Filtering
-void loop_filter_bh_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void)simpler_lpf;
- // These should all be done at once with one call, instead of 3
- loop_filter_horizontal_edge_y_ppc(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
- loop_filter_horizontal_edge_y_ppc(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
- loop_filter_horizontal_edge_y_ppc(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
-
- if (u_ptr)
- loop_filter_horizontal_edge_uv_ppc(u_ptr + 4 * uv_stride, v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr);
-}
-
-void loop_filter_bhs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void)simpler_lpf;
- (void)u_ptr;
- (void)v_ptr;
- (void)uv_stride;
- loop_filter_simple_horizontal_edge_ppc(y_ptr + 4 * y_stride, y_stride, lfi->flim);
- loop_filter_simple_horizontal_edge_ppc(y_ptr + 8 * y_stride, y_stride, lfi->flim);
- loop_filter_simple_horizontal_edge_ppc(y_ptr + 12 * y_stride, y_stride, lfi->flim);
-}
-
-// Vertical B Filtering
-void loop_filter_bv_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void)simpler_lpf;
- loop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->flim, lfi->lim, lfi->thr);
-
- if (u_ptr)
- loop_filter_vertical_edge_uv_ppc(u_ptr + 4, v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr);
-}
-
-void loop_filter_bvs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void)simpler_lpf;
- (void)u_ptr;
- (void)v_ptr;
- (void)uv_stride;
- loop_filter_simple_vertical_edge_ppc(y_ptr + 4, y_stride, lfi->flim);
- loop_filter_simple_vertical_edge_ppc(y_ptr + 8, y_stride, lfi->flim);
- loop_filter_simple_vertical_edge_ppc(y_ptr + 12, y_stride, lfi->flim);
-}
diff --git a/vp8/common/ppc/systemdependent.c b/vp8/common/ppc/systemdependent.c
deleted file mode 100644
index 1f5d790..0000000
--- a/vp8/common/ppc/systemdependent.c
+++ /dev/null
@@ -1,171 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "g_common.h"
-#include "subpixel.h"
-#include "loopfilter.h"
-#include "recon.h"
-#include "idct.h"
-#include "onyxc_int.h"
-
-void (*vp8_short_idct4x4)(short *input, short *output, int pitch);
-void (*vp8_short_idct4x4_1)(short *input, short *output, int pitch);
-void (*vp8_dc_only_idct)(short input_dc, short *output, int pitch);
-
-extern void (*vp8_post_proc_down_and_across)(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- int src_pixels_per_line,
- int dst_pixels_per_line,
- int rows,
- int cols,
- int flimit
-);
-
-extern void (*vp8_mbpost_proc_down)(unsigned char *dst, int pitch, int rows, int cols, int flimit);
-extern void vp8_mbpost_proc_down_c(unsigned char *dst, int pitch, int rows, int cols, int flimit);
-extern void (*vp8_mbpost_proc_across_ip)(unsigned char *src, int pitch, int rows, int cols, int flimit);
-extern void vp8_mbpost_proc_across_ip_c(unsigned char *src, int pitch, int rows, int cols, int flimit);
-
-extern void vp8_post_proc_down_and_across_c
-(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- int src_pixels_per_line,
- int dst_pixels_per_line,
- int rows,
- int cols,
- int flimit
-);
-void vp8_plane_add_noise_c(unsigned char *Start, unsigned int Width, unsigned int Height, int Pitch, int q, int a);
-
-extern copy_mem_block_function *vp8_copy_mem16x16;
-extern copy_mem_block_function *vp8_copy_mem8x8;
-extern copy_mem_block_function *vp8_copy_mem8x4;
-
-// PPC
-extern subpixel_predict_function sixtap_predict_ppc;
-extern subpixel_predict_function sixtap_predict8x4_ppc;
-extern subpixel_predict_function sixtap_predict8x8_ppc;
-extern subpixel_predict_function sixtap_predict16x16_ppc;
-extern subpixel_predict_function bilinear_predict4x4_ppc;
-extern subpixel_predict_function bilinear_predict8x4_ppc;
-extern subpixel_predict_function bilinear_predict8x8_ppc;
-extern subpixel_predict_function bilinear_predict16x16_ppc;
-
-extern copy_mem_block_function copy_mem16x16_ppc;
-
-void recon_b_ppc(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
-void recon2b_ppc(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
-void recon4b_ppc(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
-
-extern void short_idct4x4llm_ppc(short *input, short *output, int pitch);
-
-// Generic C
-extern subpixel_predict_function vp8_sixtap_predict_c;
-extern subpixel_predict_function vp8_sixtap_predict8x4_c;
-extern subpixel_predict_function vp8_sixtap_predict8x8_c;
-extern subpixel_predict_function vp8_sixtap_predict16x16_c;
-extern subpixel_predict_function vp8_bilinear_predict4x4_c;
-extern subpixel_predict_function vp8_bilinear_predict8x4_c;
-extern subpixel_predict_function vp8_bilinear_predict8x8_c;
-extern subpixel_predict_function vp8_bilinear_predict16x16_c;
-
-extern copy_mem_block_function vp8_copy_mem16x16_c;
-extern copy_mem_block_function vp8_copy_mem8x8_c;
-extern copy_mem_block_function vp8_copy_mem8x4_c;
-
-void vp8_recon_b_c(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
-void vp8_recon2b_c(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
-void vp8_recon4b_c(short *diff_ptr, unsigned char *pred_ptr, unsigned char *dst_ptr, int stride);
-
-extern void vp8_short_idct4x4llm_1_c(short *input, short *output, int pitch);
-extern void vp8_short_idct4x4llm_c(short *input, short *output, int pitch);
-extern void vp8_dc_only_idct_c(short input_dc, short *output, int pitch);
-
-// PPC
-extern loop_filter_block_function loop_filter_mbv_ppc;
-extern loop_filter_block_function loop_filter_bv_ppc;
-extern loop_filter_block_function loop_filter_mbh_ppc;
-extern loop_filter_block_function loop_filter_bh_ppc;
-
-extern loop_filter_block_function loop_filter_mbvs_ppc;
-extern loop_filter_block_function loop_filter_bvs_ppc;
-extern loop_filter_block_function loop_filter_mbhs_ppc;
-extern loop_filter_block_function loop_filter_bhs_ppc;
-
-// Generic C
-extern loop_filter_block_function vp8_loop_filter_mbv_c;
-extern loop_filter_block_function vp8_loop_filter_bv_c;
-extern loop_filter_block_function vp8_loop_filter_mbh_c;
-extern loop_filter_block_function vp8_loop_filter_bh_c;
-
-extern loop_filter_block_function vp8_loop_filter_mbvs_c;
-extern loop_filter_block_function vp8_loop_filter_bvs_c;
-extern loop_filter_block_function vp8_loop_filter_mbhs_c;
-extern loop_filter_block_function vp8_loop_filter_bhs_c;
-
-extern loop_filter_block_function *vp8_lf_mbvfull;
-extern loop_filter_block_function *vp8_lf_mbhfull;
-extern loop_filter_block_function *vp8_lf_bvfull;
-extern loop_filter_block_function *vp8_lf_bhfull;
-
-extern loop_filter_block_function *vp8_lf_mbvsimple;
-extern loop_filter_block_function *vp8_lf_mbhsimple;
-extern loop_filter_block_function *vp8_lf_bvsimple;
-extern loop_filter_block_function *vp8_lf_bhsimple;
-
-void vp8_clear_c(void)
-{
-}
-
-void vp8_machine_specific_config(void)
-{
- // Pure C:
- vp8_clear_system_state = vp8_clear_c;
- vp8_recon_b = vp8_recon_b_c;
- vp8_recon4b = vp8_recon4b_c;
- vp8_recon2b = vp8_recon2b_c;
-
- vp8_bilinear_predict16x16 = bilinear_predict16x16_ppc;
- vp8_bilinear_predict8x8 = bilinear_predict8x8_ppc;
- vp8_bilinear_predict8x4 = bilinear_predict8x4_ppc;
- vp8_bilinear_predict = bilinear_predict4x4_ppc;
-
- vp8_sixtap_predict16x16 = sixtap_predict16x16_ppc;
- vp8_sixtap_predict8x8 = sixtap_predict8x8_ppc;
- vp8_sixtap_predict8x4 = sixtap_predict8x4_ppc;
- vp8_sixtap_predict = sixtap_predict_ppc;
-
- vp8_short_idct4x4_1 = vp8_short_idct4x4llm_1_c;
- vp8_short_idct4x4 = short_idct4x4llm_ppc;
- vp8_dc_only_idct = vp8_dc_only_idct_c;
-
- vp8_lf_mbvfull = loop_filter_mbv_ppc;
- vp8_lf_bvfull = loop_filter_bv_ppc;
- vp8_lf_mbhfull = loop_filter_mbh_ppc;
- vp8_lf_bhfull = loop_filter_bh_ppc;
-
- vp8_lf_mbvsimple = loop_filter_mbvs_ppc;
- vp8_lf_bvsimple = loop_filter_bvs_ppc;
- vp8_lf_mbhsimple = loop_filter_mbhs_ppc;
- vp8_lf_bhsimple = loop_filter_bhs_ppc;
-
- vp8_post_proc_down_and_across = vp8_post_proc_down_and_across_c;
- vp8_mbpost_proc_down = vp8_mbpost_proc_down_c;
- vp8_mbpost_proc_across_ip = vp8_mbpost_proc_across_ip_c;
- vp8_plane_add_noise = vp8_plane_add_noise_c;
-
- vp8_copy_mem16x16 = copy_mem16x16_ppc;
- vp8_copy_mem8x8 = vp8_copy_mem8x8_c;
- vp8_copy_mem8x4 = vp8_copy_mem8x4_c;
-
-}
diff --git a/vp8/common/ppflags.h b/vp8/common/ppflags.h
deleted file mode 100644
index 65b0cab..0000000
--- a/vp8/common/ppflags.h
+++ /dev/null
@@ -1,40 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_PPFLAGS_H
-#define __INC_PPFLAGS_H
-enum
-{
- VP8D_NOFILTERING = 0,
- VP8D_DEBLOCK = 1<<0,
- VP8D_DEMACROBLOCK = 1<<1,
- VP8D_ADDNOISE = 1<<2,
- VP8D_DEBUG_TXT_FRAME_INFO = 1<<3,
- VP8D_DEBUG_TXT_MBLK_MODES = 1<<4,
- VP8D_DEBUG_TXT_DC_DIFF = 1<<5,
- VP8D_DEBUG_TXT_RATE_INFO = 1<<6,
- VP8D_DEBUG_DRAW_MV = 1<<7,
- VP8D_DEBUG_CLR_BLK_MODES = 1<<8,
- VP8D_DEBUG_CLR_FRM_REF_BLKS = 1<<9
-};
-
-typedef struct
-{
- int post_proc_flag;
- int deblocking_level;
- int noise_level;
- int display_ref_frame_flag;
- int display_mb_modes_flag;
- int display_b_modes_flag;
- int display_mv_flag;
-} vp8_ppflags_t;
-
-#endif
diff --git a/vp8/common/quant_common.c b/vp8/common/quant_common.c
deleted file mode 100644
index e9833fe..0000000
--- a/vp8/common/quant_common.c
+++ /dev/null
@@ -1,132 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "quant_common.h"
-
-static const int dc_qlookup[QINDEX_RANGE] =
-{
- 4, 5, 6, 7, 8, 9, 10, 10, 11, 12, 13, 14, 15, 16, 17, 17,
- 18, 19, 20, 20, 21, 21, 22, 22, 23, 23, 24, 25, 25, 26, 27, 28,
- 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43,
- 44, 45, 46, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58,
- 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74,
- 75, 76, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89,
- 91, 93, 95, 96, 98, 100, 101, 102, 104, 106, 108, 110, 112, 114, 116, 118,
- 122, 124, 126, 128, 130, 132, 134, 136, 138, 140, 143, 145, 148, 151, 154, 157,
-};
-
-static const int ac_qlookup[QINDEX_RANGE] =
-{
- 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
- 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,
- 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,
- 52, 53, 54, 55, 56, 57, 58, 60, 62, 64, 66, 68, 70, 72, 74, 76,
- 78, 80, 82, 84, 86, 88, 90, 92, 94, 96, 98, 100, 102, 104, 106, 108,
- 110, 112, 114, 116, 119, 122, 125, 128, 131, 134, 137, 140, 143, 146, 149, 152,
- 155, 158, 161, 164, 167, 170, 173, 177, 181, 185, 189, 193, 197, 201, 205, 209,
- 213, 217, 221, 225, 229, 234, 239, 245, 249, 254, 259, 264, 269, 274, 279, 284,
-};
-
-
-int vp8_dc_quant(int QIndex, int Delta)
-{
- int retval;
-
- QIndex = QIndex + Delta;
-
- if (QIndex > 127)
- QIndex = 127;
- else if (QIndex < 0)
- QIndex = 0;
-
- retval = dc_qlookup[ QIndex ];
- return retval;
-}
-
-int vp8_dc2quant(int QIndex, int Delta)
-{
- int retval;
-
- QIndex = QIndex + Delta;
-
- if (QIndex > 127)
- QIndex = 127;
- else if (QIndex < 0)
- QIndex = 0;
-
- retval = dc_qlookup[ QIndex ] * 2;
- return retval;
-
-}
-int vp8_dc_uv_quant(int QIndex, int Delta)
-{
- int retval;
-
- QIndex = QIndex + Delta;
-
- if (QIndex > 127)
- QIndex = 127;
- else if (QIndex < 0)
- QIndex = 0;
-
- retval = dc_qlookup[ QIndex ];
-
- if (retval > 132)
- retval = 132;
-
- return retval;
-}
-
-int vp8_ac_yquant(int QIndex)
-{
- int retval;
-
- if (QIndex > 127)
- QIndex = 127;
- else if (QIndex < 0)
- QIndex = 0;
-
- retval = ac_qlookup[ QIndex ];
- return retval;
-}
-
-int vp8_ac2quant(int QIndex, int Delta)
-{
- int retval;
-
- QIndex = QIndex + Delta;
-
- if (QIndex > 127)
- QIndex = 127;
- else if (QIndex < 0)
- QIndex = 0;
-
- retval = (ac_qlookup[ QIndex ] * 155) / 100;
-
- if (retval < 8)
- retval = 8;
-
- return retval;
-}
-int vp8_ac_uv_quant(int QIndex, int Delta)
-{
- int retval;
-
- QIndex = QIndex + Delta;
-
- if (QIndex > 127)
- QIndex = 127;
- else if (QIndex < 0)
- QIndex = 0;
-
- retval = ac_qlookup[ QIndex ];
- return retval;
-}
diff --git a/vp8/common/recon.c b/vp8/common/recon.c
deleted file mode 100644
index d72d6e4..0000000
--- a/vp8/common/recon.c
+++ /dev/null
@@ -1,181 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "recon.h"
-#include "blockd.h"
-
-void vp8_recon_b_c
-(
- unsigned char *pred_ptr,
- short *diff_ptr,
- unsigned char *dst_ptr,
- int stride
-)
-{
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- int a = diff_ptr[c] + pred_ptr[c] ;
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dst_ptr[c] = (unsigned char) a ;
- }
-
- dst_ptr += stride;
- diff_ptr += 16;
- pred_ptr += 16;
- }
-}
-
-void vp8_recon4b_c
-(
- unsigned char *pred_ptr,
- short *diff_ptr,
- unsigned char *dst_ptr,
- int stride
-)
-{
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 16; c++)
- {
- int a = diff_ptr[c] + pred_ptr[c] ;
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dst_ptr[c] = (unsigned char) a ;
- }
-
- dst_ptr += stride;
- diff_ptr += 16;
- pred_ptr += 16;
- }
-}
-
-void vp8_recon2b_c
-(
- unsigned char *pred_ptr,
- short *diff_ptr,
- unsigned char *dst_ptr,
- int stride
-)
-{
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 8; c++)
- {
- int a = diff_ptr[c] + pred_ptr[c] ;
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dst_ptr[c] = (unsigned char) a ;
- }
-
- dst_ptr += stride;
- diff_ptr += 8;
- pred_ptr += 8;
- }
-}
-
-void vp8_recon_mby_c(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
-#if ARCH_ARM
- BLOCKD *b = &x->block[0];
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-
- /*b = &x->block[4];*/
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-
- /*b = &x->block[8];*/
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-
- /*b = &x->block[12];*/
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-#else
- int i;
-
- for (i = 0; i < 16; i += 4)
- {
- BLOCKD *b = &x->block[i];
-
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- }
-#endif
-}
-
-void vp8_recon_mb_c(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
-#if ARCH_ARM
- BLOCKD *b = &x->block[0];
-
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b += 4;
-
- /*b = &x->block[16];*/
-
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b++;
- b++;
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b++;
- b++;
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b++;
- b++;
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-#else
- int i;
-
- for (i = 0; i < 16; i += 4)
- {
- BLOCKD *b = &x->block[i];
-
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- }
-
- for (i = 16; i < 24; i += 2)
- {
- BLOCKD *b = &x->block[i];
-
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- }
-#endif
-}
diff --git a/vp8/common/recon.h b/vp8/common/recon.h
deleted file mode 100644
index c7a1ca1..0000000
--- a/vp8/common/recon.h
+++ /dev/null
@@ -1,123 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_RECON_H
-#define __INC_RECON_H
-
-#include "blockd.h"
-
-#define prototype_copy_block(sym) \
- void sym(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch)
-
-#define prototype_recon_block(sym) \
- void sym(unsigned char *pred, short *diff, unsigned char *dst, int pitch)
-
-#define prototype_recon_macroblock(sym) \
- void sym(const struct vp8_recon_rtcd_vtable *rtcd, MACROBLOCKD *x)
-
-#define prototype_build_intra_predictors(sym) \
- void sym(MACROBLOCKD *x)
-
-struct vp8_recon_rtcd_vtable;
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/recon_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/recon_arm.h"
-#endif
-
-#if ARCH_MIPS
-#if defined(MIPS_DSP_REV) && MIPS_DSP_REV>=1
-#include "mips/recon_mips.h"
-#endif
-#endif
-
-#ifndef vp8_recon_copy16x16
-#define vp8_recon_copy16x16 vp8_copy_mem16x16_c
-#endif
-extern prototype_copy_block(vp8_recon_copy16x16);
-
-#ifndef vp8_recon_copy8x8
-#define vp8_recon_copy8x8 vp8_copy_mem8x8_c
-#endif
-extern prototype_copy_block(vp8_recon_copy8x8);
-
-#ifndef vp8_recon_copy8x4
-#define vp8_recon_copy8x4 vp8_copy_mem8x4_c
-#endif
-extern prototype_copy_block(vp8_recon_copy8x4);
-
-#ifndef vp8_recon_recon
-#define vp8_recon_recon vp8_recon_b_c
-#endif
-extern prototype_recon_block(vp8_recon_recon);
-
-#ifndef vp8_recon_recon2
-#define vp8_recon_recon2 vp8_recon2b_c
-#endif
-extern prototype_recon_block(vp8_recon_recon2);
-
-#ifndef vp8_recon_recon4
-#define vp8_recon_recon4 vp8_recon4b_c
-#endif
-extern prototype_recon_block(vp8_recon_recon4);
-
-#ifndef vp8_recon_recon_mb
-#define vp8_recon_recon_mb vp8_recon_mb_c
-#endif
-extern prototype_recon_macroblock(vp8_recon_recon_mb);
-
-#ifndef vp8_recon_recon_mby
-#define vp8_recon_recon_mby vp8_recon_mby_c
-#endif
-extern prototype_recon_macroblock(vp8_recon_recon_mby);
-
-#ifndef vp8_recon_build_intra_predictors_mby
-#define vp8_recon_build_intra_predictors_mby vp8_build_intra_predictors_mby
-#endif
-extern prototype_build_intra_predictors\
- (vp8_recon_build_intra_predictors_mby);
-
-#ifndef vp8_recon_build_intra_predictors_mby_s
-#define vp8_recon_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s
-#endif
-extern prototype_build_intra_predictors\
- (vp8_recon_build_intra_predictors_mby_s);
-
-
-typedef prototype_copy_block((*vp8_copy_block_fn_t));
-typedef prototype_recon_block((*vp8_recon_fn_t));
-typedef prototype_recon_macroblock((*vp8_recon_mb_fn_t));
-typedef prototype_build_intra_predictors((*vp8_build_intra_pred_fn_t));
-typedef struct vp8_recon_rtcd_vtable
-{
- vp8_copy_block_fn_t copy16x16;
- vp8_copy_block_fn_t copy8x8;
- vp8_copy_block_fn_t copy8x4;
- vp8_recon_fn_t recon;
- vp8_recon_fn_t recon2;
- vp8_recon_fn_t recon4;
- vp8_recon_mb_fn_t recon_mb;
- vp8_recon_mb_fn_t recon_mby;
- vp8_build_intra_pred_fn_t build_intra_predictors_mby_s;
- vp8_build_intra_pred_fn_t build_intra_predictors_mby;
-} vp8_recon_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define RECON_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define RECON_INVOKE(ctx,fn) vp8_recon_##fn
-#endif
-
-void vp8_recon_intra_mbuv(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
-#endif
diff --git a/vp8/common/reconinter.c b/vp8/common/reconinter.c
deleted file mode 100644
index 7cfab41..0000000
--- a/vp8/common/reconinter.c
+++ /dev/null
@@ -1,693 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "recon.h"
-#include "subpixel.h"
-#include "blockd.h"
-#include "reconinter.h"
-#if CONFIG_RUNTIME_CPU_DETECT
-#include "onyxc_int.h"
-#endif
-
-/* use this define on systems where unaligned int reads and writes are
- * not allowed, i.e. ARM architectures
- */
-/*#define MUST_BE_ALIGNED*/
-
-
-static const int bbb[4] = {0, 2, 8, 10};
-
-
-
-void vp8_copy_mem16x16_c(
- unsigned char *src,
- int src_stride,
- unsigned char *dst,
- int dst_stride)
-{
-
- int r;
-
- for (r = 0; r < 16; r++)
- {
-#ifdef MUST_BE_ALIGNED
- dst[0] = src[0];
- dst[1] = src[1];
- dst[2] = src[2];
- dst[3] = src[3];
- dst[4] = src[4];
- dst[5] = src[5];
- dst[6] = src[6];
- dst[7] = src[7];
- dst[8] = src[8];
- dst[9] = src[9];
- dst[10] = src[10];
- dst[11] = src[11];
- dst[12] = src[12];
- dst[13] = src[13];
- dst[14] = src[14];
- dst[15] = src[15];
-
-#else
- ((int *)dst)[0] = ((int *)src)[0] ;
- ((int *)dst)[1] = ((int *)src)[1] ;
- ((int *)dst)[2] = ((int *)src)[2] ;
- ((int *)dst)[3] = ((int *)src)[3] ;
-
-#endif
- src += src_stride;
- dst += dst_stride;
-
- }
-
-}
-
-void vp8_copy_mem8x8_c(
- unsigned char *src,
- int src_stride,
- unsigned char *dst,
- int dst_stride)
-{
- int r;
-
- for (r = 0; r < 8; r++)
- {
-#ifdef MUST_BE_ALIGNED
- dst[0] = src[0];
- dst[1] = src[1];
- dst[2] = src[2];
- dst[3] = src[3];
- dst[4] = src[4];
- dst[5] = src[5];
- dst[6] = src[6];
- dst[7] = src[7];
-#else
- ((int *)dst)[0] = ((int *)src)[0] ;
- ((int *)dst)[1] = ((int *)src)[1] ;
-#endif
- src += src_stride;
- dst += dst_stride;
-
- }
-
-}
-
-void vp8_copy_mem8x4_c(
- unsigned char *src,
- int src_stride,
- unsigned char *dst,
- int dst_stride)
-{
- int r;
-
- for (r = 0; r < 4; r++)
- {
-#ifdef MUST_BE_ALIGNED
- dst[0] = src[0];
- dst[1] = src[1];
- dst[2] = src[2];
- dst[3] = src[3];
- dst[4] = src[4];
- dst[5] = src[5];
- dst[6] = src[6];
- dst[7] = src[7];
-#else
- ((int *)dst)[0] = ((int *)src)[0] ;
- ((int *)dst)[1] = ((int *)src)[1] ;
-#endif
- src += src_stride;
- dst += dst_stride;
-
- }
-
-}
-
-
-
-void vp8_build_inter_predictors_b(BLOCKD *d, int pitch, vp8_subpix_fn_t sppf)
-{
- int r;
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *pred_ptr = d->predictor;
-
- ptr_base = *(d->base_pre);
-
- if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
- {
- ptr = ptr_base + d->pre + (d->bmi.mv.as_mv.row >> 3) * d->pre_stride + (d->bmi.mv.as_mv.col >> 3);
- sppf(ptr, d->pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, pred_ptr, pitch);
- }
- else
- {
- ptr_base += d->pre + (d->bmi.mv.as_mv.row >> 3) * d->pre_stride + (d->bmi.mv.as_mv.col >> 3);
- ptr = ptr_base;
-
- for (r = 0; r < 4; r++)
- {
-#ifdef MUST_BE_ALIGNED
- pred_ptr[0] = ptr[0];
- pred_ptr[1] = ptr[1];
- pred_ptr[2] = ptr[2];
- pred_ptr[3] = ptr[3];
-#else
- *(int *)pred_ptr = *(int *)ptr ;
-#endif
- pred_ptr += pitch;
- ptr += d->pre_stride;
- }
- }
-}
-
-static void build_inter_predictors4b(MACROBLOCKD *x, BLOCKD *d, int pitch)
-{
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *pred_ptr = d->predictor;
-
- ptr_base = *(d->base_pre);
- ptr = ptr_base + d->pre + (d->bmi.mv.as_mv.row >> 3) * d->pre_stride + (d->bmi.mv.as_mv.col >> 3);
-
- if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
- {
- x->subpixel_predict8x8(ptr, d->pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, pred_ptr, pitch);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(ptr, d->pre_stride, pred_ptr, pitch);
- }
-}
-
-static void build_inter_predictors2b(MACROBLOCKD *x, BLOCKD *d, int pitch)
-{
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *pred_ptr = d->predictor;
-
- ptr_base = *(d->base_pre);
- ptr = ptr_base + d->pre + (d->bmi.mv.as_mv.row >> 3) * d->pre_stride + (d->bmi.mv.as_mv.col >> 3);
-
- if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
- {
- x->subpixel_predict8x4(ptr, d->pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, pred_ptr, pitch);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x4)(ptr, d->pre_stride, pred_ptr, pitch);
- }
-}
-
-
-void vp8_build_inter_predictors_mbuv(MACROBLOCKD *x)
-{
- int i;
-
- if (x->mode_info_context->mbmi.ref_frame != INTRA_FRAME &&
- x->mode_info_context->mbmi.mode != SPLITMV)
- {
- unsigned char *uptr, *vptr;
- unsigned char *upred_ptr = &x->predictor[256];
- unsigned char *vpred_ptr = &x->predictor[320];
-
- int mv_row = x->block[16].bmi.mv.as_mv.row;
- int mv_col = x->block[16].bmi.mv.as_mv.col;
- int offset;
- int pre_stride = x->block[16].pre_stride;
-
- offset = (mv_row >> 3) * pre_stride + (mv_col >> 3);
- uptr = x->pre.u_buffer + offset;
- vptr = x->pre.v_buffer + offset;
-
- if ((mv_row | mv_col) & 7)
- {
- x->subpixel_predict8x8(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr, 8);
- x->subpixel_predict8x8(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr, 8);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(uptr, pre_stride, upred_ptr, 8);
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(vptr, pre_stride, vpred_ptr, 8);
- }
- }
- else
- {
- for (i = 16; i < 24; i += 2)
- {
- BLOCKD *d0 = &x->block[i];
- BLOCKD *d1 = &x->block[i+1];
-
- if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
- build_inter_predictors2b(x, d0, 8);
- else
- {
- vp8_build_inter_predictors_b(d0, 8, x->subpixel_predict);
- vp8_build_inter_predictors_b(d1, 8, x->subpixel_predict);
- }
- }
- }
-}
-
-/*encoder only*/
-void vp8_build_inter_predictors_mby(MACROBLOCKD *x)
-{
-
- if (x->mode_info_context->mbmi.ref_frame != INTRA_FRAME &&
- x->mode_info_context->mbmi.mode != SPLITMV)
- {
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *pred_ptr = x->predictor;
- int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
- int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
- int pre_stride = x->block[0].pre_stride;
-
- ptr_base = x->pre.y_buffer;
- ptr = ptr_base + (mv_row >> 3) * pre_stride + (mv_col >> 3);
-
- if ((mv_row | mv_col) & 7)
- {
- x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7, pred_ptr, 16);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy16x16)(ptr, pre_stride, pred_ptr, 16);
- }
- }
- else
- {
- int i;
-
- if (x->mode_info_context->mbmi.partitioning < 3)
- {
- for (i = 0; i < 4; i++)
- {
- BLOCKD *d = &x->block[bbb[i]];
- build_inter_predictors4b(x, d, 16);
- }
-
- }
- else
- {
- for (i = 0; i < 16; i += 2)
- {
- BLOCKD *d0 = &x->block[i];
- BLOCKD *d1 = &x->block[i+1];
-
- if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
- build_inter_predictors2b(x, d0, 16);
- else
- {
- vp8_build_inter_predictors_b(d0, 16, x->subpixel_predict);
- vp8_build_inter_predictors_b(d1, 16, x->subpixel_predict);
- }
-
- }
- }
- }
-}
-
-void vp8_build_inter_predictors_mb(MACROBLOCKD *x)
-{
-
- if (x->mode_info_context->mbmi.ref_frame != INTRA_FRAME &&
- x->mode_info_context->mbmi.mode != SPLITMV)
- {
- int offset;
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *uptr, *vptr;
- unsigned char *pred_ptr = x->predictor;
- unsigned char *upred_ptr = &x->predictor[256];
- unsigned char *vpred_ptr = &x->predictor[320];
-
- int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
- int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
- int pre_stride = x->block[0].pre_stride;
-
- ptr_base = x->pre.y_buffer;
- ptr = ptr_base + (mv_row >> 3) * pre_stride + (mv_col >> 3);
-
- if ((mv_row | mv_col) & 7)
- {
- x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7, pred_ptr, 16);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy16x16)(ptr, pre_stride, pred_ptr, 16);
- }
-
- mv_row = x->block[16].bmi.mv.as_mv.row;
- mv_col = x->block[16].bmi.mv.as_mv.col;
- pre_stride >>= 1;
- offset = (mv_row >> 3) * pre_stride + (mv_col >> 3);
- uptr = x->pre.u_buffer + offset;
- vptr = x->pre.v_buffer + offset;
-
- if ((mv_row | mv_col) & 7)
- {
- x->subpixel_predict8x8(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr, 8);
- x->subpixel_predict8x8(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr, 8);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(uptr, pre_stride, upred_ptr, 8);
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(vptr, pre_stride, vpred_ptr, 8);
- }
- }
- else
- {
- int i;
-
- if (x->mode_info_context->mbmi.partitioning < 3)
- {
- for (i = 0; i < 4; i++)
- {
- BLOCKD *d = &x->block[bbb[i]];
- build_inter_predictors4b(x, d, 16);
- }
- }
- else
- {
- for (i = 0; i < 16; i += 2)
- {
- BLOCKD *d0 = &x->block[i];
- BLOCKD *d1 = &x->block[i+1];
-
- if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
- build_inter_predictors2b(x, d0, 16);
- else
- {
- vp8_build_inter_predictors_b(d0, 16, x->subpixel_predict);
- vp8_build_inter_predictors_b(d1, 16, x->subpixel_predict);
- }
-
- }
-
- }
-
- for (i = 16; i < 24; i += 2)
- {
- BLOCKD *d0 = &x->block[i];
- BLOCKD *d1 = &x->block[i+1];
-
- if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
- build_inter_predictors2b(x, d0, 8);
- else
- {
- vp8_build_inter_predictors_b(d0, 8, x->subpixel_predict);
- vp8_build_inter_predictors_b(d1, 8, x->subpixel_predict);
- }
-
- }
-
- }
-}
-
-void vp8_build_uvmvs(MACROBLOCKD *x, int fullpixel)
-{
- int i, j;
-
- if (x->mode_info_context->mbmi.mode == SPLITMV)
- {
- for (i = 0; i < 2; i++)
- {
- for (j = 0; j < 2; j++)
- {
- int yoffset = i * 8 + j * 2;
- int uoffset = 16 + i * 2 + j;
- int voffset = 20 + i * 2 + j;
-
- int temp;
-
- temp = x->block[yoffset ].bmi.mv.as_mv.row
- + x->block[yoffset+1].bmi.mv.as_mv.row
- + x->block[yoffset+4].bmi.mv.as_mv.row
- + x->block[yoffset+5].bmi.mv.as_mv.row;
-
- if (temp < 0) temp -= 4;
- else temp += 4;
-
- x->block[uoffset].bmi.mv.as_mv.row = temp / 8;
-
- if (fullpixel)
- x->block[uoffset].bmi.mv.as_mv.row = (temp / 8) & 0xfffffff8;
-
- temp = x->block[yoffset ].bmi.mv.as_mv.col
- + x->block[yoffset+1].bmi.mv.as_mv.col
- + x->block[yoffset+4].bmi.mv.as_mv.col
- + x->block[yoffset+5].bmi.mv.as_mv.col;
-
- if (temp < 0) temp -= 4;
- else temp += 4;
-
- x->block[uoffset].bmi.mv.as_mv.col = temp / 8;
-
- if (fullpixel)
- x->block[uoffset].bmi.mv.as_mv.col = (temp / 8) & 0xfffffff8;
-
- x->block[voffset].bmi.mv.as_mv.row = x->block[uoffset].bmi.mv.as_mv.row ;
- x->block[voffset].bmi.mv.as_mv.col = x->block[uoffset].bmi.mv.as_mv.col ;
- }
- }
- }
- else
- {
- int mvrow = x->mode_info_context->mbmi.mv.as_mv.row;
- int mvcol = x->mode_info_context->mbmi.mv.as_mv.col;
-
- if (mvrow < 0)
- mvrow -= 1;
- else
- mvrow += 1;
-
- if (mvcol < 0)
- mvcol -= 1;
- else
- mvcol += 1;
-
- mvrow /= 2;
- mvcol /= 2;
-
- for (i = 0; i < 8; i++)
- {
- x->block[ 16 + i].bmi.mv.as_mv.row = mvrow;
- x->block[ 16 + i].bmi.mv.as_mv.col = mvcol;
-
- if (fullpixel)
- {
- x->block[ 16 + i].bmi.mv.as_mv.row = mvrow & 0xfffffff8;
- x->block[ 16 + i].bmi.mv.as_mv.col = mvcol & 0xfffffff8;
- }
- }
- }
-}
-
-
-/* The following functions are wriiten for skip_recon_mb() to call. Since there is no recon in this
- * situation, we can write the result directly to dst buffer instead of writing it to predictor
- * buffer and then copying it to dst buffer.
- */
-static void vp8_build_inter_predictors_b_s(BLOCKD *d, unsigned char *dst_ptr, vp8_subpix_fn_t sppf)
-{
- int r;
- unsigned char *ptr_base;
- unsigned char *ptr;
- /*unsigned char *pred_ptr = d->predictor;*/
- int dst_stride = d->dst_stride;
- int pre_stride = d->pre_stride;
-
- ptr_base = *(d->base_pre);
-
- if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
- {
- ptr = ptr_base + d->pre + (d->bmi.mv.as_mv.row >> 3) * d->pre_stride + (d->bmi.mv.as_mv.col >> 3);
- sppf(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst_ptr, dst_stride);
- }
- else
- {
- ptr_base += d->pre + (d->bmi.mv.as_mv.row >> 3) * d->pre_stride + (d->bmi.mv.as_mv.col >> 3);
- ptr = ptr_base;
-
- for (r = 0; r < 4; r++)
- {
-#ifdef MUST_BE_ALIGNED
- dst_ptr[0] = ptr[0];
- dst_ptr[1] = ptr[1];
- dst_ptr[2] = ptr[2];
- dst_ptr[3] = ptr[3];
-#else
- *(int *)dst_ptr = *(int *)ptr ;
-#endif
- dst_ptr += dst_stride;
- ptr += pre_stride;
- }
- }
-}
-
-
-
-void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
-{
- /*unsigned char *pred_ptr = x->block[0].predictor;
- unsigned char *dst_ptr = *(x->block[0].base_dst) + x->block[0].dst;*/
- unsigned char *pred_ptr = x->predictor;
- unsigned char *dst_ptr = x->dst.y_buffer;
-
- if (x->mode_info_context->mbmi.mode != SPLITMV)
- {
- int offset;
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *uptr, *vptr;
- /*unsigned char *pred_ptr = x->predictor;
- unsigned char *upred_ptr = &x->predictor[256];
- unsigned char *vpred_ptr = &x->predictor[320];*/
- unsigned char *udst_ptr = x->dst.u_buffer;
- unsigned char *vdst_ptr = x->dst.v_buffer;
-
- int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
- int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
- int pre_stride = x->dst.y_stride; /*x->block[0].pre_stride;*/
-
- ptr_base = x->pre.y_buffer;
- ptr = ptr_base + (mv_row >> 3) * pre_stride + (mv_col >> 3);
-
- if ((mv_row | mv_col) & 7)
- {
- x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7, dst_ptr, x->dst.y_stride); /*x->block[0].dst_stride);*/
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy16x16)(ptr, pre_stride, dst_ptr, x->dst.y_stride); /*x->block[0].dst_stride);*/
- }
-
- mv_row = x->block[16].bmi.mv.as_mv.row;
- mv_col = x->block[16].bmi.mv.as_mv.col;
- pre_stride >>= 1;
- offset = (mv_row >> 3) * pre_stride + (mv_col >> 3);
- uptr = x->pre.u_buffer + offset;
- vptr = x->pre.v_buffer + offset;
-
- if ((mv_row | mv_col) & 7)
- {
- x->subpixel_predict8x8(uptr, pre_stride, mv_col & 7, mv_row & 7, udst_ptr, x->dst.uv_stride);
- x->subpixel_predict8x8(vptr, pre_stride, mv_col & 7, mv_row & 7, vdst_ptr, x->dst.uv_stride);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(uptr, pre_stride, udst_ptr, x->dst.uv_stride);
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(vptr, pre_stride, vdst_ptr, x->dst.uv_stride);
- }
- }
- else
- {
- /* note: this whole ELSE part is not executed at all. So, no way to test the correctness of my modification. Later,
- * if sth is wrong, go back to what it is in build_inter_predictors_mb.
- */
- int i;
-
- if (x->mode_info_context->mbmi.partitioning < 3)
- {
- for (i = 0; i < 4; i++)
- {
- BLOCKD *d = &x->block[bbb[i]];
- /*build_inter_predictors4b(x, d, 16);*/
-
- {
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *pred_ptr = d->predictor;
-
- ptr_base = *(d->base_pre);
- ptr = ptr_base + d->pre + (d->bmi.mv.as_mv.row >> 3) * d->pre_stride + (d->bmi.mv.as_mv.col >> 3);
-
- if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
- {
- x->subpixel_predict8x8(ptr, d->pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst_ptr, x->dst.y_stride); /*x->block[0].dst_stride);*/
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(ptr, d->pre_stride, dst_ptr, x->dst.y_stride); /*x->block[0].dst_stride);*/
- }
- }
- }
- }
- else
- {
- for (i = 0; i < 16; i += 2)
- {
- BLOCKD *d0 = &x->block[i];
- BLOCKD *d1 = &x->block[i+1];
-
- if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
- {
- /*build_inter_predictors2b(x, d0, 16);*/
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *pred_ptr = d0->predictor;
-
- ptr_base = *(d0->base_pre);
- ptr = ptr_base + d0->pre + (d0->bmi.mv.as_mv.row >> 3) * d0->pre_stride + (d0->bmi.mv.as_mv.col >> 3);
-
- if (d0->bmi.mv.as_mv.row & 7 || d0->bmi.mv.as_mv.col & 7)
- {
- x->subpixel_predict8x4(ptr, d0->pre_stride, d0->bmi.mv.as_mv.col & 7, d0->bmi.mv.as_mv.row & 7, dst_ptr, x->dst.y_stride);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x4)(ptr, d0->pre_stride, dst_ptr, x->dst.y_stride);
- }
- }
- else
- {
- vp8_build_inter_predictors_b_s(d0, dst_ptr, x->subpixel_predict);
- vp8_build_inter_predictors_b_s(d1, dst_ptr, x->subpixel_predict);
- }
- }
- }
-
- for (i = 16; i < 24; i += 2)
- {
- BLOCKD *d0 = &x->block[i];
- BLOCKD *d1 = &x->block[i+1];
-
- if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
- {
- /*build_inter_predictors2b(x, d0, 8);*/
- unsigned char *ptr_base;
- unsigned char *ptr;
- unsigned char *pred_ptr = d0->predictor;
-
- ptr_base = *(d0->base_pre);
- ptr = ptr_base + d0->pre + (d0->bmi.mv.as_mv.row >> 3) * d0->pre_stride + (d0->bmi.mv.as_mv.col >> 3);
-
- if (d0->bmi.mv.as_mv.row & 7 || d0->bmi.mv.as_mv.col & 7)
- {
- x->subpixel_predict8x4(ptr, d0->pre_stride,
- d0->bmi.mv.as_mv.col & 7,
- d0->bmi.mv.as_mv.row & 7,
- dst_ptr, x->dst.uv_stride);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x4)(ptr,
- d0->pre_stride, dst_ptr, x->dst.uv_stride);
- }
- }
- else
- {
- vp8_build_inter_predictors_b_s(d0, dst_ptr, x->subpixel_predict);
- vp8_build_inter_predictors_b_s(d1, dst_ptr, x->subpixel_predict);
- }
- }
- }
-}
diff --git a/vp8/common/reconinter.h b/vp8/common/reconinter.h
deleted file mode 100644
index 7c1dee4..0000000
--- a/vp8/common/reconinter.h
+++ /dev/null
@@ -1,23 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_RECONINTER_H
-#define __INC_RECONINTER_H
-
-extern void vp8_build_inter_predictors_mb(MACROBLOCKD *x);
-extern void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x);
-
-extern void vp8_build_inter_predictors_mby(MACROBLOCKD *x);
-extern void vp8_build_uvmvs(MACROBLOCKD *x, int fullpixel);
-extern void vp8_build_inter_predictors_b(BLOCKD *d, int pitch, vp8_subpix_fn_t sppf);
-extern void vp8_build_inter_predictors_mbuv(MACROBLOCKD *x);
-
-#endif
diff --git a/vp8/common/reconintra.c b/vp8/common/reconintra.c
deleted file mode 100644
index 9cf5f6a..0000000
--- a/vp8/common/reconintra.c
+++ /dev/null
@@ -1,556 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "recon.h"
-#include "reconintra.h"
-#include "vpx_mem/vpx_mem.h"
-
-/* For skip_recon_mb(), add vp8_build_intra_predictors_mby_s(MACROBLOCKD *x) and
- * vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x).
- */
-void vp8_recon_intra_mbuv(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
- int i;
-
- for (i = 16; i < 24; i += 2)
- {
- BLOCKD *b = &x->block[i];
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- }
-}
-
-void vp8_build_intra_predictors_mby(MACROBLOCKD *x)
-{
-
- unsigned char *yabove_row = x->dst.y_buffer - x->dst.y_stride;
- unsigned char yleft_col[16];
- unsigned char ytop_left = yabove_row[-1];
- unsigned char *ypred_ptr = x->predictor;
- int r, c, i;
-
- for (i = 0; i < 16; i++)
- {
- yleft_col[i] = x->dst.y_buffer [i* x->dst.y_stride -1];
- }
-
- /* for Y */
- switch (x->mode_info_context->mbmi.mode)
- {
- case DC_PRED:
- {
- int expected_dc;
- int i;
- int shift;
- int average = 0;
-
-
- if (x->up_available || x->left_available)
- {
- if (x->up_available)
- {
- for (i = 0; i < 16; i++)
- {
- average += yabove_row[i];
- }
- }
-
- if (x->left_available)
- {
-
- for (i = 0; i < 16; i++)
- {
- average += yleft_col[i];
- }
-
- }
-
-
-
- shift = 3 + x->up_available + x->left_available;
- expected_dc = (average + (1 << (shift - 1))) >> shift;
- }
- else
- {
- expected_dc = 128;
- }
-
- vpx_memset(ypred_ptr, expected_dc, 256);
- }
- break;
- case V_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
-
- ((int *)ypred_ptr)[0] = ((int *)yabove_row)[0];
- ((int *)ypred_ptr)[1] = ((int *)yabove_row)[1];
- ((int *)ypred_ptr)[2] = ((int *)yabove_row)[2];
- ((int *)ypred_ptr)[3] = ((int *)yabove_row)[3];
- ypred_ptr += 16;
- }
- }
- break;
- case H_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
-
- vpx_memset(ypred_ptr, yleft_col[r], 16);
- ypred_ptr += 16;
- }
-
- }
- break;
- case TM_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
- for (c = 0; c < 16; c++)
- {
- int pred = yleft_col[r] + yabove_row[ c] - ytop_left;
-
- if (pred < 0)
- pred = 0;
-
- if (pred > 255)
- pred = 255;
-
- ypred_ptr[c] = pred;
- }
-
- ypred_ptr += 16;
- }
-
- }
- break;
- case B_PRED:
- case NEARESTMV:
- case NEARMV:
- case ZEROMV:
- case NEWMV:
- case SPLITMV:
- case MB_MODE_COUNT:
- break;
- }
-}
-
-void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x)
-{
-
- unsigned char *yabove_row = x->dst.y_buffer - x->dst.y_stride;
- unsigned char yleft_col[16];
- unsigned char ytop_left = yabove_row[-1];
- unsigned char *ypred_ptr = x->predictor;
- int r, c, i;
-
- int y_stride = x->dst.y_stride;
- ypred_ptr = x->dst.y_buffer; /*x->predictor;*/
-
- for (i = 0; i < 16; i++)
- {
- yleft_col[i] = x->dst.y_buffer [i* x->dst.y_stride -1];
- }
-
- /* for Y */
- switch (x->mode_info_context->mbmi.mode)
- {
- case DC_PRED:
- {
- int expected_dc;
- int i;
- int shift;
- int average = 0;
-
-
- if (x->up_available || x->left_available)
- {
- if (x->up_available)
- {
- for (i = 0; i < 16; i++)
- {
- average += yabove_row[i];
- }
- }
-
- if (x->left_available)
- {
-
- for (i = 0; i < 16; i++)
- {
- average += yleft_col[i];
- }
-
- }
-
-
-
- shift = 3 + x->up_available + x->left_available;
- expected_dc = (average + (1 << (shift - 1))) >> shift;
- }
- else
- {
- expected_dc = 128;
- }
-
- /*vpx_memset(ypred_ptr, expected_dc, 256);*/
- for (r = 0; r < 16; r++)
- {
- vpx_memset(ypred_ptr, expected_dc, 16);
- ypred_ptr += y_stride; /*16;*/
- }
- }
- break;
- case V_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
-
- ((int *)ypred_ptr)[0] = ((int *)yabove_row)[0];
- ((int *)ypred_ptr)[1] = ((int *)yabove_row)[1];
- ((int *)ypred_ptr)[2] = ((int *)yabove_row)[2];
- ((int *)ypred_ptr)[3] = ((int *)yabove_row)[3];
- ypred_ptr += y_stride; /*16;*/
- }
- }
- break;
- case H_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
-
- vpx_memset(ypred_ptr, yleft_col[r], 16);
- ypred_ptr += y_stride; /*16;*/
- }
-
- }
- break;
- case TM_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
- for (c = 0; c < 16; c++)
- {
- int pred = yleft_col[r] + yabove_row[ c] - ytop_left;
-
- if (pred < 0)
- pred = 0;
-
- if (pred > 255)
- pred = 255;
-
- ypred_ptr[c] = pred;
- }
-
- ypred_ptr += y_stride; /*16;*/
- }
-
- }
- break;
- case B_PRED:
- case NEARESTMV:
- case NEARMV:
- case ZEROMV:
- case NEWMV:
- case SPLITMV:
- case MB_MODE_COUNT:
- break;
- }
-}
-
-void vp8_build_intra_predictors_mbuv(MACROBLOCKD *x)
-{
- unsigned char *uabove_row = x->dst.u_buffer - x->dst.uv_stride;
- unsigned char uleft_col[16];
- unsigned char utop_left = uabove_row[-1];
- unsigned char *vabove_row = x->dst.v_buffer - x->dst.uv_stride;
- unsigned char vleft_col[20];
- unsigned char vtop_left = vabove_row[-1];
- unsigned char *upred_ptr = &x->predictor[256];
- unsigned char *vpred_ptr = &x->predictor[320];
- int i, j;
-
- for (i = 0; i < 8; i++)
- {
- uleft_col[i] = x->dst.u_buffer [i* x->dst.uv_stride -1];
- vleft_col[i] = x->dst.v_buffer [i* x->dst.uv_stride -1];
- }
-
- switch (x->mode_info_context->mbmi.uv_mode)
- {
- case DC_PRED:
- {
- int expected_udc;
- int expected_vdc;
- int i;
- int shift;
- int Uaverage = 0;
- int Vaverage = 0;
-
- if (x->up_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uabove_row[i];
- Vaverage += vabove_row[i];
- }
- }
-
- if (x->left_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uleft_col[i];
- Vaverage += vleft_col[i];
- }
- }
-
- if (!x->up_available && !x->left_available)
- {
- expected_udc = 128;
- expected_vdc = 128;
- }
- else
- {
- shift = 2 + x->up_available + x->left_available;
- expected_udc = (Uaverage + (1 << (shift - 1))) >> shift;
- expected_vdc = (Vaverage + (1 << (shift - 1))) >> shift;
- }
-
-
- vpx_memset(upred_ptr, expected_udc, 64);
- vpx_memset(vpred_ptr, expected_vdc, 64);
-
-
- }
- break;
- case V_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- vpx_memcpy(upred_ptr, uabove_row, 8);
- vpx_memcpy(vpred_ptr, vabove_row, 8);
- upred_ptr += 8;
- vpred_ptr += 8;
- }
-
- }
- break;
- case H_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- vpx_memset(upred_ptr, uleft_col[i], 8);
- vpx_memset(vpred_ptr, vleft_col[i], 8);
- upred_ptr += 8;
- vpred_ptr += 8;
- }
- }
-
- break;
- case TM_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- for (j = 0; j < 8; j++)
- {
- int predu = uleft_col[i] + uabove_row[j] - utop_left;
- int predv = vleft_col[i] + vabove_row[j] - vtop_left;
-
- if (predu < 0)
- predu = 0;
-
- if (predu > 255)
- predu = 255;
-
- if (predv < 0)
- predv = 0;
-
- if (predv > 255)
- predv = 255;
-
- upred_ptr[j] = predu;
- vpred_ptr[j] = predv;
- }
-
- upred_ptr += 8;
- vpred_ptr += 8;
- }
-
- }
- break;
- case B_PRED:
- case NEARESTMV:
- case NEARMV:
- case ZEROMV:
- case NEWMV:
- case SPLITMV:
- case MB_MODE_COUNT:
- break;
- }
-}
-
-void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x)
-{
- unsigned char *uabove_row = x->dst.u_buffer - x->dst.uv_stride;
- unsigned char uleft_col[16];
- unsigned char utop_left = uabove_row[-1];
- unsigned char *vabove_row = x->dst.v_buffer - x->dst.uv_stride;
- unsigned char vleft_col[20];
- unsigned char vtop_left = vabove_row[-1];
- unsigned char *upred_ptr = x->dst.u_buffer; /*&x->predictor[256];*/
- unsigned char *vpred_ptr = x->dst.v_buffer; /*&x->predictor[320];*/
- int uv_stride = x->dst.uv_stride;
-
- int i, j;
-
- for (i = 0; i < 8; i++)
- {
- uleft_col[i] = x->dst.u_buffer [i* x->dst.uv_stride -1];
- vleft_col[i] = x->dst.v_buffer [i* x->dst.uv_stride -1];
- }
-
- switch (x->mode_info_context->mbmi.uv_mode)
- {
- case DC_PRED:
- {
- int expected_udc;
- int expected_vdc;
- int i;
- int shift;
- int Uaverage = 0;
- int Vaverage = 0;
-
- if (x->up_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uabove_row[i];
- Vaverage += vabove_row[i];
- }
- }
-
- if (x->left_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uleft_col[i];
- Vaverage += vleft_col[i];
- }
- }
-
- if (!x->up_available && !x->left_available)
- {
- expected_udc = 128;
- expected_vdc = 128;
- }
- else
- {
- shift = 2 + x->up_available + x->left_available;
- expected_udc = (Uaverage + (1 << (shift - 1))) >> shift;
- expected_vdc = (Vaverage + (1 << (shift - 1))) >> shift;
- }
-
-
- /*vpx_memset(upred_ptr,expected_udc,64);*/
- /*vpx_memset(vpred_ptr,expected_vdc,64);*/
- for (i = 0; i < 8; i++)
- {
- vpx_memset(upred_ptr, expected_udc, 8);
- vpx_memset(vpred_ptr, expected_vdc, 8);
- upred_ptr += uv_stride; /*8;*/
- vpred_ptr += uv_stride; /*8;*/
- }
- }
- break;
- case V_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- vpx_memcpy(upred_ptr, uabove_row, 8);
- vpx_memcpy(vpred_ptr, vabove_row, 8);
- upred_ptr += uv_stride; /*8;*/
- vpred_ptr += uv_stride; /*8;*/
- }
-
- }
- break;
- case H_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- vpx_memset(upred_ptr, uleft_col[i], 8);
- vpx_memset(vpred_ptr, vleft_col[i], 8);
- upred_ptr += uv_stride; /*8;*/
- vpred_ptr += uv_stride; /*8;*/
- }
- }
-
- break;
- case TM_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- for (j = 0; j < 8; j++)
- {
- int predu = uleft_col[i] + uabove_row[j] - utop_left;
- int predv = vleft_col[i] + vabove_row[j] - vtop_left;
-
- if (predu < 0)
- predu = 0;
-
- if (predu > 255)
- predu = 255;
-
- if (predv < 0)
- predv = 0;
-
- if (predv > 255)
- predv = 255;
-
- upred_ptr[j] = predu;
- vpred_ptr[j] = predv;
- }
-
- upred_ptr += uv_stride; /*8;*/
- vpred_ptr += uv_stride; /*8;*/
- }
-
- }
- break;
- case B_PRED:
- case NEARESTMV:
- case NEARMV:
- case ZEROMV:
- case NEWMV:
- case SPLITMV:
- case MB_MODE_COUNT:
- break;
- }
-}
diff --git a/vp8/common/reconintra.h b/vp8/common/reconintra.h
deleted file mode 100644
index 4025a53..0000000
--- a/vp8/common/reconintra.h
+++ /dev/null
@@ -1,22 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_RECONINTRA_H
-#define __INC_RECONINTRA_H
-
-extern void init_intra_left_above_pixels(MACROBLOCKD *x);
-
-extern void vp8_build_intra_predictors_mbuv(MACROBLOCKD *x);
-extern void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x);
-
-extern void vp8_predict_intra4x4(BLOCKD *x, int b_mode, unsigned char *Predictor);
-
-#endif
diff --git a/vp8/common/reconintra4x4.c b/vp8/common/reconintra4x4.c
deleted file mode 100644
index cd70dca..0000000
--- a/vp8/common/reconintra4x4.c
+++ /dev/null
@@ -1,315 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "recon.h"
-#include "vpx_mem/vpx_mem.h"
-#include "reconintra.h"
-
-void vp8_predict_intra4x4(BLOCKD *x,
- int b_mode,
- unsigned char *predictor)
-{
- int i, r, c;
-
- unsigned char *Above = *(x->base_dst) + x->dst - x->dst_stride;
- unsigned char Left[4];
- unsigned char top_left = Above[-1];
-
- Left[0] = (*(x->base_dst))[x->dst - 1];
- Left[1] = (*(x->base_dst))[x->dst - 1 + x->dst_stride];
- Left[2] = (*(x->base_dst))[x->dst - 1 + 2 * x->dst_stride];
- Left[3] = (*(x->base_dst))[x->dst - 1 + 3 * x->dst_stride];
-
- switch (b_mode)
- {
- case B_DC_PRED:
- {
- int expected_dc = 0;
-
- for (i = 0; i < 4; i++)
- {
- expected_dc += Above[i];
- expected_dc += Left[i];
- }
-
- expected_dc = (expected_dc + 4) >> 3;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- predictor[c] = expected_dc;
- }
-
- predictor += 16;
- }
- }
- break;
- case B_TM_PRED:
- {
- /* prediction similar to true_motion prediction */
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- int pred = Above[c] - top_left + Left[r];
-
- if (pred < 0)
- pred = 0;
-
- if (pred > 255)
- pred = 255;
-
- predictor[c] = pred;
- }
-
- predictor += 16;
- }
- }
- break;
-
- case B_VE_PRED:
- {
-
- unsigned int ap[4];
- ap[0] = (top_left + 2 * Above[0] + Above[1] + 2) >> 2;
- ap[1] = (Above[0] + 2 * Above[1] + Above[2] + 2) >> 2;
- ap[2] = (Above[1] + 2 * Above[2] + Above[3] + 2) >> 2;
- ap[3] = (Above[2] + 2 * Above[3] + Above[4] + 2) >> 2;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
-
- predictor[c] = ap[c];
- }
-
- predictor += 16;
- }
-
- }
- break;
-
-
- case B_HE_PRED:
- {
-
- unsigned int lp[4];
- lp[0] = (top_left + 2 * Left[0] + Left[1] + 2) >> 2;
- lp[1] = (Left[0] + 2 * Left[1] + Left[2] + 2) >> 2;
- lp[2] = (Left[1] + 2 * Left[2] + Left[3] + 2) >> 2;
- lp[3] = (Left[2] + 2 * Left[3] + Left[3] + 2) >> 2;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- predictor[c] = lp[r];
- }
-
- predictor += 16;
- }
- }
- break;
- case B_LD_PRED:
- {
- unsigned char *ptr = Above;
- predictor[0 * 16 + 0] = (ptr[0] + ptr[1] * 2 + ptr[2] + 2) >> 2;
- predictor[0 * 16 + 1] =
- predictor[1 * 16 + 0] = (ptr[1] + ptr[2] * 2 + ptr[3] + 2) >> 2;
- predictor[0 * 16 + 2] =
- predictor[1 * 16 + 1] =
- predictor[2 * 16 + 0] = (ptr[2] + ptr[3] * 2 + ptr[4] + 2) >> 2;
- predictor[0 * 16 + 3] =
- predictor[1 * 16 + 2] =
- predictor[2 * 16 + 1] =
- predictor[3 * 16 + 0] = (ptr[3] + ptr[4] * 2 + ptr[5] + 2) >> 2;
- predictor[1 * 16 + 3] =
- predictor[2 * 16 + 2] =
- predictor[3 * 16 + 1] = (ptr[4] + ptr[5] * 2 + ptr[6] + 2) >> 2;
- predictor[2 * 16 + 3] =
- predictor[3 * 16 + 2] = (ptr[5] + ptr[6] * 2 + ptr[7] + 2) >> 2;
- predictor[3 * 16 + 3] = (ptr[6] + ptr[7] * 2 + ptr[7] + 2) >> 2;
-
- }
- break;
- case B_RD_PRED:
- {
-
- unsigned char pp[9];
-
- pp[0] = Left[3];
- pp[1] = Left[2];
- pp[2] = Left[1];
- pp[3] = Left[0];
- pp[4] = top_left;
- pp[5] = Above[0];
- pp[6] = Above[1];
- pp[7] = Above[2];
- pp[8] = Above[3];
-
- predictor[3 * 16 + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
- predictor[3 * 16 + 1] =
- predictor[2 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[3 * 16 + 2] =
- predictor[2 * 16 + 1] =
- predictor[1 * 16 + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
- predictor[3 * 16 + 3] =
- predictor[2 * 16 + 2] =
- predictor[1 * 16 + 1] =
- predictor[0 * 16 + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
- predictor[2 * 16 + 3] =
- predictor[1 * 16 + 2] =
- predictor[0 * 16 + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
- predictor[1 * 16 + 3] =
- predictor[0 * 16 + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
- predictor[0 * 16 + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
-
- }
- break;
- case B_VR_PRED:
- {
-
- unsigned char pp[9];
-
- pp[0] = Left[3];
- pp[1] = Left[2];
- pp[2] = Left[1];
- pp[3] = Left[0];
- pp[4] = top_left;
- pp[5] = Above[0];
- pp[6] = Above[1];
- pp[7] = Above[2];
- pp[8] = Above[3];
-
-
- predictor[3 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[2 * 16 + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
- predictor[3 * 16 + 1] =
- predictor[1 * 16 + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
- predictor[2 * 16 + 1] =
- predictor[0 * 16 + 0] = (pp[4] + pp[5] + 1) >> 1;
- predictor[3 * 16 + 2] =
- predictor[1 * 16 + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
- predictor[2 * 16 + 2] =
- predictor[0 * 16 + 1] = (pp[5] + pp[6] + 1) >> 1;
- predictor[3 * 16 + 3] =
- predictor[1 * 16 + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
- predictor[2 * 16 + 3] =
- predictor[0 * 16 + 2] = (pp[6] + pp[7] + 1) >> 1;
- predictor[1 * 16 + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
- predictor[0 * 16 + 3] = (pp[7] + pp[8] + 1) >> 1;
-
- }
- break;
- case B_VL_PRED:
- {
-
- unsigned char *pp = Above;
-
- predictor[0 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
- predictor[1 * 16 + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
- predictor[2 * 16 + 0] =
- predictor[0 * 16 + 1] = (pp[1] + pp[2] + 1) >> 1;
- predictor[1 * 16 + 1] =
- predictor[3 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[2 * 16 + 1] =
- predictor[0 * 16 + 2] = (pp[2] + pp[3] + 1) >> 1;
- predictor[3 * 16 + 1] =
- predictor[1 * 16 + 2] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
- predictor[0 * 16 + 3] =
- predictor[2 * 16 + 2] = (pp[3] + pp[4] + 1) >> 1;
- predictor[1 * 16 + 3] =
- predictor[3 * 16 + 2] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
- predictor[2 * 16 + 3] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
- predictor[3 * 16 + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
- }
- break;
-
- case B_HD_PRED:
- {
- unsigned char pp[9];
- pp[0] = Left[3];
- pp[1] = Left[2];
- pp[2] = Left[1];
- pp[3] = Left[0];
- pp[4] = top_left;
- pp[5] = Above[0];
- pp[6] = Above[1];
- pp[7] = Above[2];
- pp[8] = Above[3];
-
-
- predictor[3 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
- predictor[3 * 16 + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
- predictor[2 * 16 + 0] =
- predictor[3 * 16 + 2] = (pp[1] + pp[2] + 1) >> 1;
- predictor[2 * 16 + 1] =
- predictor[3 * 16 + 3] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[2 * 16 + 2] =
- predictor[1 * 16 + 0] = (pp[2] + pp[3] + 1) >> 1;
- predictor[2 * 16 + 3] =
- predictor[1 * 16 + 1] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
- predictor[1 * 16 + 2] =
- predictor[0 * 16 + 0] = (pp[3] + pp[4] + 1) >> 1;
- predictor[1 * 16 + 3] =
- predictor[0 * 16 + 1] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
- predictor[0 * 16 + 2] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
- predictor[0 * 16 + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
- }
- break;
-
-
- case B_HU_PRED:
- {
- unsigned char *pp = Left;
- predictor[0 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
- predictor[0 * 16 + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
- predictor[0 * 16 + 2] =
- predictor[1 * 16 + 0] = (pp[1] + pp[2] + 1) >> 1;
- predictor[0 * 16 + 3] =
- predictor[1 * 16 + 1] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[1 * 16 + 2] =
- predictor[2 * 16 + 0] = (pp[2] + pp[3] + 1) >> 1;
- predictor[1 * 16 + 3] =
- predictor[2 * 16 + 1] = (pp[2] + pp[3] * 2 + pp[3] + 2) >> 2;
- predictor[2 * 16 + 2] =
- predictor[2 * 16 + 3] =
- predictor[3 * 16 + 0] =
- predictor[3 * 16 + 1] =
- predictor[3 * 16 + 2] =
- predictor[3 * 16 + 3] = pp[3];
- }
- break;
-
-
- }
-}
-/* copy 4 bytes from the above right down so that the 4x4 prediction modes using pixels above and
- * to the right prediction have filled in pixels to use.
- */
-void vp8_intra_prediction_down_copy(MACROBLOCKD *x)
-{
- unsigned char *above_right = *(x->block[0].base_dst) + x->block[0].dst - x->block[0].dst_stride + 16;
-
- unsigned int *src_ptr = (unsigned int *)above_right;
- unsigned int *dst_ptr0 = (unsigned int *)(above_right + 4 * x->block[0].dst_stride);
- unsigned int *dst_ptr1 = (unsigned int *)(above_right + 8 * x->block[0].dst_stride);
- unsigned int *dst_ptr2 = (unsigned int *)(above_right + 12 * x->block[0].dst_stride);
-
- *dst_ptr0 = *src_ptr;
- *dst_ptr1 = *src_ptr;
- *dst_ptr2 = *src_ptr;
-}
-
-
diff --git a/vp8/common/reconintra4x4.h b/vp8/common/reconintra4x4.h
deleted file mode 100644
index 6ac2b71..0000000
--- a/vp8/common/reconintra4x4.h
+++ /dev/null
@@ -1,17 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_RECONINTRA4x4_H
-#define __INC_RECONINTRA4x4_H
-
-extern void vp8_intra_prediction_down_copy(MACROBLOCKD *x);
-
-#endif
diff --git a/vp8/common/setupintrarecon.c b/vp8/common/setupintrarecon.c
deleted file mode 100644
index 7976e25..0000000
--- a/vp8/common/setupintrarecon.c
+++ /dev/null
@@ -1,32 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "setupintrarecon.h"
-#include "vpx_mem/vpx_mem.h"
-
-void vp8_setup_intra_recon(YV12_BUFFER_CONFIG *ybf)
-{
- int i;
-
- /* set up frame new frame for intra coded blocks */
- vpx_memset(ybf->y_buffer - 1 - ybf->y_stride, 127, ybf->y_width + 5);
- for (i = 0; i < ybf->y_height; i++)
- ybf->y_buffer[ybf->y_stride *i - 1] = (unsigned char) 129;
-
- vpx_memset(ybf->u_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
- for (i = 0; i < ybf->uv_height; i++)
- ybf->u_buffer[ybf->uv_stride *i - 1] = (unsigned char) 129;
-
- vpx_memset(ybf->v_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
- for (i = 0; i < ybf->uv_height; i++)
- ybf->v_buffer[ybf->uv_stride *i - 1] = (unsigned char) 129;
-
-}
diff --git a/vp8/common/setupintrarecon.h b/vp8/common/setupintrarecon.h
deleted file mode 100644
index 5264fd0..0000000
--- a/vp8/common/setupintrarecon.h
+++ /dev/null
@@ -1,13 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_scale/yv12config.h"
-extern void vp8_setup_intra_recon(YV12_BUFFER_CONFIG *ybf);
diff --git a/vp8/common/subpixel.h b/vp8/common/subpixel.h
deleted file mode 100644
index 12bbdc9..0000000
--- a/vp8/common/subpixel.h
+++ /dev/null
@@ -1,94 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef SUBPIXEL_H
-#define SUBPIXEL_H
-
-#include "vpx_config.h"
-
-#define prototype_subpixel_predict(sym) \
- void sym(unsigned char *src, int src_pitch, int xofst, int yofst, \
- unsigned char *dst, int dst_pitch)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/subpixel_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/subpixel_arm.h"
-#endif
-
-#if ARCH_MIPS
-#if defined(MIPS_DSP_REV) && MIPS_DSP_REV>=2
-#include "mips/subpixel_mips.h"
-#endif
-#endif
-
-#ifndef vp8_subpix_sixtap16x16
-#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_c
-#endif
-extern prototype_subpixel_predict(vp8_subpix_sixtap16x16);
-
-#ifndef vp8_subpix_sixtap8x8
-#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_c
-#endif
-extern prototype_subpixel_predict(vp8_subpix_sixtap8x8);
-
-#ifndef vp8_subpix_sixtap8x4
-#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_c
-#endif
-extern prototype_subpixel_predict(vp8_subpix_sixtap8x4);
-
-#ifndef vp8_subpix_sixtap4x4
-#define vp8_subpix_sixtap4x4 vp8_sixtap_predict_c
-#endif
-extern prototype_subpixel_predict(vp8_subpix_sixtap4x4);
-
-#ifndef vp8_subpix_bilinear16x16
-#define vp8_subpix_bilinear16x16 vp8_bilinear_predict16x16_c
-#endif
-extern prototype_subpixel_predict(vp8_subpix_bilinear16x16);
-
-#ifndef vp8_subpix_bilinear8x8
-#define vp8_subpix_bilinear8x8 vp8_bilinear_predict8x8_c
-#endif
-extern prototype_subpixel_predict(vp8_subpix_bilinear8x8);
-
-#ifndef vp8_subpix_bilinear8x4
-#define vp8_subpix_bilinear8x4 vp8_bilinear_predict8x4_c
-#endif
-extern prototype_subpixel_predict(vp8_subpix_bilinear8x4);
-
-#ifndef vp8_subpix_bilinear4x4
-#define vp8_subpix_bilinear4x4 vp8_bilinear_predict4x4_c
-#endif
-extern prototype_subpixel_predict(vp8_subpix_bilinear4x4);
-
-typedef prototype_subpixel_predict((*vp8_subpix_fn_t));
-typedef struct
-{
- vp8_subpix_fn_t sixtap16x16;
- vp8_subpix_fn_t sixtap8x8;
- vp8_subpix_fn_t sixtap8x4;
- vp8_subpix_fn_t sixtap4x4;
- vp8_subpix_fn_t bilinear16x16;
- vp8_subpix_fn_t bilinear8x8;
- vp8_subpix_fn_t bilinear8x4;
- vp8_subpix_fn_t bilinear4x4;
-} vp8_subpix_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define SUBPIX_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define SUBPIX_INVOKE(ctx,fn) vp8_subpix_##fn
-#endif
-
-#endif
diff --git a/vp8/common/systemdependent.h b/vp8/common/systemdependent.h
deleted file mode 100644
index db99698..0000000
--- a/vp8/common/systemdependent.h
+++ /dev/null
@@ -1,21 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#if ARCH_X86 || ARCH_X86_64
-void vpx_reset_mmx_state(void);
-#define vp8_clear_system_state() vpx_reset_mmx_state()
-#else
-#define vp8_clear_system_state()
-#endif
-
-struct VP8Common;
-void vp8_machine_specific_config(struct VP8Common *);
diff --git a/vp8/common/threading.h b/vp8/common/threading.h
deleted file mode 100644
index 44eaf08..0000000
--- a/vp8/common/threading.h
+++ /dev/null
@@ -1,96 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef _PTHREAD_EMULATION
-#define _PTHREAD_EMULATION
-
-#define VPXINFINITE 10000 /* 10second. */
-
-#if CONFIG_OS_SUPPORT && CONFIG_MULTITHREAD
-
-/* Thread management macros */
-#ifdef _WIN32
-/* Win32 */
-#define _WIN32_WINNT 0x500 /* WINBASE.H - Enable signal_object_and_wait */
-#include <process.h>
-#include <windows.h>
-#define THREAD_FUNCTION DWORD WINAPI
-#define THREAD_FUNCTION_RETURN DWORD
-#define THREAD_SPECIFIC_INDEX DWORD
-#define pthread_t HANDLE
-#define pthread_attr_t DWORD
-#define pthread_create(thhandle,attr,thfunc,tharg) (int)((*thhandle=(HANDLE)_beginthreadex(NULL,0,(unsigned int (__stdcall *)(void *))thfunc,tharg,0,NULL))==NULL)
-#define pthread_join(thread, result) ((WaitForSingleObject((thread),VPXINFINITE)!=WAIT_OBJECT_0) || !CloseHandle(thread))
-#define pthread_detach(thread) if(thread!=NULL)CloseHandle(thread)
-#define thread_sleep(nms) Sleep(nms)
-#define pthread_cancel(thread) terminate_thread(thread,0)
-#define ts_key_create(ts_key, destructor) {ts_key = TlsAlloc();};
-#define pthread_getspecific(ts_key) TlsGetValue(ts_key)
-#define pthread_setspecific(ts_key, value) TlsSetValue(ts_key, (void *)value)
-#define pthread_self() GetCurrentThreadId()
-#else
-#ifdef __APPLE__
-#include <mach/mach_init.h>
-#include <mach/semaphore.h>
-#include <mach/task.h>
-#include <time.h>
-#include <unistd.h>
-
-#else
-#include <semaphore.h>
-#endif
-
-#include <pthread.h>
-/* pthreads */
-/* Nearly everything is already defined */
-#define THREAD_FUNCTION void *
-#define THREAD_FUNCTION_RETURN void *
-#define THREAD_SPECIFIC_INDEX pthread_key_t
-#define ts_key_create(ts_key, destructor) pthread_key_create (&(ts_key), destructor);
-#endif
-
-/* Syncrhronization macros: Win32 and Pthreads */
-#ifdef _WIN32
-#define sem_t HANDLE
-#define pause(voidpara) __asm PAUSE
-#define sem_init(sem, sem_attr1, sem_init_value) (int)((*sem = CreateEvent(NULL,FALSE,FALSE,NULL))==NULL)
-#define sem_wait(sem) (int)(WAIT_OBJECT_0 != WaitForSingleObject(*sem,VPXINFINITE))
-#define sem_post(sem) SetEvent(*sem)
-#define sem_destroy(sem) if(*sem)((int)(CloseHandle(*sem))==TRUE)
-#define thread_sleep(nms) Sleep(nms)
-
-#else
-
-#ifdef __APPLE__
-#define sem_t semaphore_t
-#define sem_init(X,Y,Z) semaphore_create(mach_task_self(), X, SYNC_POLICY_FIFO, Z)
-#define sem_wait(sem) (semaphore_wait(*sem) )
-#define sem_post(sem) semaphore_signal(*sem)
-#define sem_destroy(sem) semaphore_destroy(mach_task_self(),*sem)
-#define thread_sleep(nms) /* { struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} */
-#else
-#include <unistd.h>
-#include <sched.h>
-#define thread_sleep(nms) sched_yield();/* {struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} */
-#endif
-/* Not Windows. Assume pthreads */
-
-#endif
-
-#if ARCH_X86 || ARCH_X86_64
-#include "vpx_ports/x86.h"
-#else
-#define x86_pause_hint()
-#endif
-
-#endif /* CONFIG_OS_SUPPORT && CONFIG_MULTITHREAD */
-
-#endif
diff --git a/vp8/common/type_aliases.h b/vp8/common/type_aliases.h
deleted file mode 100644
index 22b531a..0000000
--- a/vp8/common/type_aliases.h
+++ /dev/null
@@ -1,117 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/****************************************************************************
-*
-* Module Title : type_aliases.h
-*
-* Description : Standard type aliases
-*
-****************************************************************************/
-#ifndef __INC_TYPE_ALIASES_H
-#define __INC_TYPE_ALIASES_H
-
-/****************************************************************************
-* Macros
-****************************************************************************/
-#define EXPORT
-#define IMPORT extern /* Used to declare imported data & routines */
-#define PRIVATE static /* Used to declare & define module-local data */
-#define LOCAL static /* Used to define all persistent routine-local data */
-#define STD_IN_PATH 0 /* Standard input path */
-#define STD_OUT_PATH 1 /* Standard output path */
-#define STD_ERR_PATH 2 /* Standard error path */
-#define STD_IN_FILE stdin /* Standard input file pointer */
-#define STD_OUT_FILE stdout /* Standard output file pointer */
-#define STD_ERR_FILE stderr /* Standard error file pointer */
-#define max_int 0x7FFFFFFF
-
-#define __export
-#define _export
-
-#define CCONV
-
-#ifndef NULL
-#ifdef __cplusplus
-#define NULL 0
-#else
-#define NULL ((void *)0)
-#endif
-#endif
-
-#ifndef FALSE
-#define FALSE 0
-#endif
-
-#ifndef TRUE
-#define TRUE 1
-#endif
-
-/****************************************************************************
-* Typedefs
-****************************************************************************/
-#ifndef TYPE_INT8
-#define TYPE_INT8
-typedef signed char INT8;
-#endif
-
-#ifndef TYPE_INT16
-/*#define TYPE_INT16*/
-typedef signed short INT16;
-#endif
-
-#ifndef TYPE_INT32
-/*#define TYPE_INT32*/
-typedef signed int INT32;
-#endif
-
-#ifndef TYPE_UINT8
-/*#define TYPE_UINT8*/
-typedef unsigned char UINT8;
-#endif
-
-#ifndef TYPE_UINT32
-/*#define TYPE_UINT32*/
-typedef unsigned int UINT32;
-#endif
-
-#ifndef TYPE_UINT16
-/*#define TYPE_UINT16*/
-typedef unsigned short UINT16;
-#endif
-
-#ifndef TYPE_BOOL
-/*#define TYPE_BOOL*/
-typedef int BOOL;
-#endif
-
-typedef unsigned char BOOLEAN;
-
-#ifdef _MSC_VER
-typedef __int64 INT64;
-#else
-
-#ifndef TYPE_INT64
-#ifdef _TMS320C6X
-/* for now we only have 40bits */
-typedef long INT64;
-#else
-typedef long long INT64;
-#endif
-#endif
-
-#endif
-
-/* Floating point */
-typedef double FLOAT64;
-typedef float FLOAT32;
-
-#endif
diff --git a/vp8/common/x86/boolcoder.cxx b/vp8/common/x86/boolcoder.cxx
deleted file mode 100644
index faddf1f..0000000
--- a/vp8/common/x86/boolcoder.cxx
+++ /dev/null
@@ -1,494 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-
-/* Arithmetic bool coder with largish probability range.
- Timothy S Murphy 6 August 2004 */
-
-#include <assert.h>
-#include <math.h>
-
-#include "bool_coder.h"
-
-#if tim_vp8
- extern "C" {
-# include "VP8cx/treewriter.h"
- }
-#endif
-
-int_types::~int_types() {}
-
-void bool_coder_spec::check_prec() const {
- assert( w && (r==Up || w > 1) && w < 24 && (ebias || w < 17));
-}
-
-bool bool_coder_spec::float_init( uint Ebits, uint Mbits) {
- uint b = (ebits = Ebits) + (mbits = Mbits);
- if( b) {
- assert( ebits < 6 && w + mbits < 31);
- assert( ebits + mbits < sizeof(Index) * 8);
- ebias = (1 << ebits) + 1 + mbits;
- mmask = (1 << mbits) - 1;
- max_index = ( ( half_index = 1 << b ) << 1) - 1;
- } else {
- ebias = 0;
- max_index = 255;
- half_index = 128;
- }
- check_prec();
- return b? 1:0;
-}
-
-void bool_coder_spec::cost_init()
-{
- static cdouble c = -(1 << 20)/log( 2.);
-
- FILE *f = fopen( "costs.txt", "w");
- assert( f);
-
- assert( sizeof(int) >= 4); /* for C interface */
- assert( max_index <= 255); /* size of Ctbl */
- uint i = 0; do {
- cdouble p = ( *this)( (Index) i);
- Ctbl[i] = (uint32) ( log( p) * c);
- fprintf(
- f, "cost( %d -> %10.7f) = %10d = %12.5f bits\n",
- i, p, Ctbl[i], (double) Ctbl[i] / (1<<20)
- );
- } while( ++i <= max_index);
- fclose( f);
-}
-
-bool_coder_spec_explicit_table::bool_coder_spec_explicit_table(
- cuint16 tbl[256], Rounding rr, uint prec
-)
- : bool_coder_spec( prec, rr)
-{
- check_prec();
- uint i = 0;
- if( tbl)
- do { Ptbl[i] = tbl[i];} while( ++i < 256);
- else
- do { Ptbl[i] = i << 8;} while( ++i < 256);
- cost_init();
-}
-
-
-bool_coder_spec_exponential_table::bool_coder_spec_exponential_table(
- uint x, Rounding rr, uint prec
-)
- : bool_coder_spec( prec, rr)
-{
- assert( x > 1 && x <= 16);
- check_prec();
- Ptbl[128] = 32768u;
- Ptbl[0] = (uint16) pow( 2., 16. - x);
- --x;
- int i=1; do {
- cdouble d = pow( .5, 1. + (1. - i/128.)*x) * 65536.;
- uint16 v = (uint16) d;
- if( v < i)
- v = i;
- Ptbl[256-i] = (uint16) ( 65536U - (Ptbl[i] = v));
- } while( ++i < 128);
- cost_init();
-}
-
-bool_coder_spec::bool_coder_spec( FILE *fp) {
- fscanf( fp, "%d", &w);
- int v;
- fscanf( fp, "%d", &v);
- assert( 0 <= v && v <= 2);
- r = (Rounding) v;
- fscanf( fp, "%d", &ebits);
- fscanf( fp, "%d", &mbits);
- if( float_init( ebits, mbits))
- return;
- int i=0; do {
- uint v;
- fscanf( fp, "%d", &v);
- assert( 0 <=v && v <= 65535U);
- Ptbl[i] = v;
- } while( ++i < 256);
- cost_init();
-}
-
-void bool_coder_spec::dump( FILE *fp) const {
- fprintf( fp, "%d %d %d %d\n", w, (int) r, ebits, mbits);
- if( ebits || mbits)
- return;
- int i=0; do { fprintf( fp, "%d\n", Ptbl[i]);} while( ++i < 256);
-}
-
-vp8bc_index_t bool_coder_spec::operator()( double p) const
-{
- if( p <= 0.)
- return 0;
- if( p >= 1.)
- return max_index;
- if( ebias) {
- if( p > .5)
- return max_index - ( *this)( 1. - p);
- int e;
- uint m = (uint) ldexp( frexp( p, &e), mbits + 2);
- uint x = 1 << (mbits + 1);
- assert( x <= m && m < x<<1);
- if( (m = (m >> 1) + (m & 1)) >= x) {
- m = x >> 1;
- ++e;
- }
- int y = 1 << ebits;
- if( (e += y) >= y)
- return half_index - 1;
- if( e < 0)
- return 0;
- return (Index) ( (e << mbits) + (m & mmask));
- }
-
- cuint16 v = (uint16) (p * 65536.);
- int i = 128;
- int j = 128;
- uint16 w;
- while( w = Ptbl[i], j >>= 1) {
- if( w < v)
- i += j;
- else if( w == v)
- return (uchar) i;
- else
- i -= j;
- }
- if( w > v) {
- cuint16 x = Ptbl[i-1];
- if( v <= x || w - v > v - x)
- --i;
- } else if( w < v && i < 255) {
- cuint16 x = Ptbl[i+1];
- if( x <= v || x - v < v - w)
- ++i;
- }
- return (Index) i;
-}
-
-double bool_coder_spec::operator()( Index i) const {
- if( !ebias)
- return Ptbl[i]/65536.;
- if( i >= half_index)
- return 1. - ( *this)( (Index) (max_index - i));
- return ldexp( (double)mantissa( i), - (int) exponent( i));
-}
-
-
-
-void bool_writer::carry() {
- uchar *p = B;
- assert( p > Bstart);
- while( *--p == 255) { assert( p > Bstart); *p = 0;}
- ++*p;
-}
-
-
-bool_writer::bool_writer( c_spec& s, uchar *Dest, size_t Len)
- : bool_coder( s),
- Bstart( Dest),
- Bend( Len? Dest+Len : 0),
- B( Dest)
-{
- assert( Dest);
- reset();
-}
-
-bool_writer::~bool_writer() { flush();}
-
-#if 1
- extern "C" { int bc_v = 0;}
-#else
-# define bc_v 0
-#endif
-
-
-void bool_writer::raw( bool value, uint32 s) {
- uint32 L = Low;
-
- assert( Range >= min_range && Range <= spec.max_range());
- assert( !is_toast && s && s < Range);
-
- if( bc_v) printf(
- "Writing a %d, B %x Low %x Range %x s %x blag %d ...\n",
- value? 1:0, B-Bstart, Low, Range, s, bit_lag
- );
- if( value) {
- L += s;
- s = Range - s;
- } else
- s -= rinc;
- if( s < min_range) {
- int ct = bit_lag; do {
- if( !--ct) {
- ct = 8;
- if( L & (1 << 31))
- carry();
- assert( !Bend || B < Bend);
- *B++ = (uchar) (L >> 23);
- L &= (1<<23) - 1;
- }
- } while( L += L, (s += s + rinc) < min_range);
- bit_lag = ct;
- }
- Low = L;
- Range = s;
- if( bc_v)
- printf(
- "...done, B %x Low %x Range %x blag %d \n",
- B-Bstart, Low, Range, bit_lag
- );
-}
-
-bool_writer& bool_writer::flush() {
- if( is_toast)
- return *this;
- int b = bit_lag;
- uint32 L = Low;
- assert( b);
- if( L & (1 << (32 - b)))
- carry();
- L <<= b & 7;
- b >>= 3;
- while( --b >= 0)
- L <<= 8;
- b = 4;
- assert( !Bend || B + 4 <= Bend);
- do {
- *B++ = (uchar) (L >> 24);
- L <<= 8;
- } while( --b);
- is_toast = 1;
- return *this;
-}
-
-
-bool_reader::bool_reader( c_spec& s, cuchar *src, size_t Len)
- : bool_coder( s),
- Bstart( src),
- B( src),
- Bend( Len? src+Len : 0),
- shf( 32 - s.w),
- bct( 8)
-{
- int i = 4; do { Low <<= 8; Low |= *B++;} while( --i);
-}
-
-
-bool bool_reader::raw( uint32 s) {
-
- bool val = 0;
- uint32 L = Low;
- cuint32 S = s << shf;
-
- assert( Range >= min_range && Range <= spec.max_range());
- assert( s && s < Range && (L >> shf) < Range);
-
- if( bc_v)
- printf(
- "Reading, B %x Low %x Range %x s %x bct %d ...\n",
- B-Bstart, Low, Range, s, bct
- );
-
- if( L >= S) {
- L -= S;
- s = Range - s;
- assert( L < (s << shf));
- val = 1;
- } else
- s -= rinc;
- if( s < min_range) {
- int ct = bct;
- do {
- assert( ~L & (1 << 31));
- L += L;
- if( !--ct) {
- ct = 8;
- if( !Bend || B < Bend)
- L |= *B++;
- }
- } while( (s += s + rinc) < min_range);
- bct = ct;
- }
- Low = L;
- Range = s;
- if( bc_v)
- printf(
- "...done, val %d B %x Low %x Range %x bct %d\n",
- val? 1:0, B-Bstart, Low, Range, bct
- );
- return val;
-}
-
-
-/* C interfaces */
-
-// spec interface
-
-struct NS : bool_coder_namespace {
- static Rounding r( vp8bc_c_prec *p, Rounding rr =down_full) {
- return p? (Rounding) p->r : rr;
- }
-};
-
-bool_coder_spec *vp8bc_vp6spec() {
- return new bool_coder_spec_explicit_table( 0, bool_coder_namespace::Down, 8);
-}
-bool_coder_spec *vp8bc_float_spec(
- unsigned int Ebits, unsigned int Mbits, vp8bc_c_prec *p
-) {
- return new bool_coder_spec_float( Ebits, Mbits, NS::r( p), p? p->prec : 12);
-}
-bool_coder_spec *vp8bc_literal_spec(
- const unsigned short m[256], vp8bc_c_prec *p
-) {
- return new bool_coder_spec_explicit_table( m, NS::r( p), p? p->prec : 16);
-}
-bool_coder_spec *vp8bc_exponential_spec( unsigned int x, vp8bc_c_prec *p)
-{
- return new bool_coder_spec_exponential_table( x, NS::r( p), p? p->prec : 16);
-}
-bool_coder_spec *vp8bc_spec_from_file( FILE *fp) {
- return new bool_coder_spec( fp);
-}
-void vp8bc_destroy_spec( c_bool_coder_spec *p) { delete p;}
-
-void vp8bc_spec_to_file( c_bool_coder_spec *p, FILE *fp) { p->dump( fp);}
-
-vp8bc_index_t vp8bc_index( c_bool_coder_spec *p, double x) {
- return ( *p)( x);
-}
-
-vp8bc_index_t vp8bc_index_from_counts(
- c_bool_coder_spec *p, unsigned int L, unsigned int R
-) {
- return ( *p)( (R += L)? (double) L/R : .5);
-}
-
-double vp8bc_probability( c_bool_coder_spec *p, vp8bc_index_t i) {
- return ( *p)( i);
-}
-
-vp8bc_index_t vp8bc_complement( c_bool_coder_spec *p, vp8bc_index_t i) {
- return p->complement( i);
-}
-unsigned int vp8bc_cost_zero( c_bool_coder_spec *p, vp8bc_index_t i) {
- return p->cost_zero( i);
-}
-unsigned int vp8bc_cost_one( c_bool_coder_spec *p, vp8bc_index_t i) {
- return p->cost_one( i);
-}
-unsigned int vp8bc_cost_bit( c_bool_coder_spec *p, vp8bc_index_t i, int v) {
- return p->cost_bit( i, v);
-}
-
-#if tim_vp8
- extern "C" int tok_verbose;
-
-# define dbg_l 1000000
-
- static vp8bc_index_t dbg_i [dbg_l];
- static char dbg_v [dbg_l];
- static size_t dbg_w = 0, dbg_r = 0;
-#endif
-
-// writer interface
-
-bool_writer *vp8bc_create_writer(
- c_bool_coder_spec *p, unsigned char *D, size_t L
-) {
- return new bool_writer( *p, D, L);
-}
-
-size_t vp8bc_destroy_writer( bool_writer *p) {
- const size_t s = p->flush().bytes_written();
- delete p;
- return s;
-}
-
-void vp8bc_write_bool( bool_writer *p, int v, vp8bc_index_t i)
-{
-# if tim_vp8
- // bc_v = dbg_w < 10;
- if( bc_v = tok_verbose)
- printf( " writing %d at prob %d\n", v? 1:0, i);
- accum_entropy_bc( &p->Spec(), i, v);
-
- ( *p)( i, (bool) v);
-
- if( dbg_w < dbg_l) {
- dbg_i [dbg_w] = i;
- dbg_v [dbg_w++] = v? 1:0;
- }
-# else
- ( *p)( i, (bool) v);
-# endif
-}
-
-void vp8bc_write_bits( bool_writer *p, unsigned int v, int n)
-{
-# if tim_vp8
- {
- c_bool_coder_spec * const s = & p->Spec();
- const vp8bc_index_t i = s->half_index();
- int m = n;
- while( --m >= 0)
- accum_entropy_bc( s, i, (v>>m) & 1);
- }
-# endif
-
- p->write_bits( n, v);
-}
-
-c_bool_coder_spec *vp8bc_writer_spec( c_bool_writer *w) { return & w->Spec();}
-
-// reader interface
-
-bool_reader *vp8bc_create_reader(
- c_bool_coder_spec *p, const unsigned char *S, size_t L
-) {
- return new bool_reader( *p, S, L);
-}
-
-void vp8bc_destroy_reader( bool_reader * p) { delete p;}
-
-int vp8bc_read_bool( bool_reader *p, vp8bc_index_t i)
-{
-# if tim_vp8
- // bc_v = dbg_r < 10;
- bc_v = tok_verbose;
- const int v = ( *p)( i)? 1:0;
- if( tok_verbose)
- printf( " reading %d at prob %d\n", v, i);
- if( dbg_r < dbg_l) {
- assert( dbg_r <= dbg_w);
- if( i != dbg_i[dbg_r] || v != dbg_v[dbg_r]) {
- printf(
- "Position %d: INCORRECTLY READING %d prob %d, wrote %d prob %d\n",
- dbg_r, v, i, dbg_v[dbg_r], dbg_i[dbg_r]
- );
- }
- ++dbg_r;
- }
- return v;
-# else
- return ( *p)( i)? 1:0;
-# endif
-}
-
-unsigned int vp8bc_read_bits( bool_reader *p, int n) { return p->read_bits( n);}
-
-c_bool_coder_spec *vp8bc_reader_spec( c_bool_reader *r) { return & r->Spec();}
-
-#undef bc_v
diff --git a/vp8/common/x86/idct_x86.h b/vp8/common/x86/idct_x86.h
deleted file mode 100644
index f6e568c..0000000
--- a/vp8/common/x86/idct_x86.h
+++ /dev/null
@@ -1,64 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef IDCT_X86_H
-#define IDCT_X86_H
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-
-#if HAVE_MMX
-extern prototype_idct(vp8_short_idct4x4llm_1_mmx);
-extern prototype_idct(vp8_short_idct4x4llm_mmx);
-extern prototype_idct_scalar_add(vp8_dc_only_idct_add_mmx);
-
-extern prototype_second_order(vp8_short_inv_walsh4x4_mmx);
-extern prototype_second_order(vp8_short_inv_walsh4x4_1_mmx);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_idct_idct1
-#define vp8_idct_idct1 vp8_short_idct4x4llm_1_mmx
-
-#undef vp8_idct_idct16
-#define vp8_idct_idct16 vp8_short_idct4x4llm_mmx
-
-#undef vp8_idct_idct1_scalar_add
-#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_mmx
-
-#undef vp8_idct_iwalsh16
-#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_mmx
-
-#undef vp8_idct_iwalsh1
-#define vp8_idct_iwalsh1 vp8_short_inv_walsh4x4_1_mmx
-
-#endif
-#endif
-
-#if HAVE_SSE2
-
-extern prototype_second_order(vp8_short_inv_walsh4x4_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_idct_iwalsh16
-#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_sse2
-
-#endif
-
-#endif
-
-
-
-#endif
diff --git a/vp8/common/x86/idctllm_mmx.asm b/vp8/common/x86/idctllm_mmx.asm
deleted file mode 100644
index 43735bc..0000000
--- a/vp8/common/x86/idctllm_mmx.asm
+++ /dev/null
@@ -1,292 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-; /****************************************************************************
-; * Notes:
-; *
-; * This implementation makes use of 16 bit fixed point verio of two multiply
-; * constants:
-; * 1. sqrt(2) * cos (pi/8)
-; * 2. sqrt(2) * sin (pi/8)
-; * Becuase the first constant is bigger than 1, to maintain the same 16 bit
-; * fixed point prrcision as the second one, we use a trick of
-; * x * a = x + x*(a-1)
-; * so
-; * x * sqrt(2) * cos (pi/8) = x + x * (sqrt(2) *cos(pi/8)-1).
-; *
-; * For the second constant, becuase of the 16bit version is 35468, which
-; * is bigger than 32768, in signed 16 bit multiply, it become a negative
-; * number.
-; * (x * (unsigned)35468 >> 16) = x * (signed)35468 >> 16 + x
-; *
-; **************************************************************************/
-
-
-;void short_idct4x4llm_mmx(short *input, short *output, int pitch)
-global sym(vp8_short_idct4x4llm_mmx)
-sym(vp8_short_idct4x4llm_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 3
- GET_GOT rbx
- ; end prolog
-
- mov rax, arg(0) ;input
- mov rdx, arg(1) ;output
-
- movq mm0, [rax ]
- movq mm1, [rax+ 8]
-
- movq mm2, [rax+16]
- movq mm3, [rax+24]
-
- movsxd rax, dword ptr arg(2) ;pitch
-
- psubw mm0, mm2 ; b1= 0-2
- paddw mm2, mm2 ;
-
- movq mm5, mm1
- paddw mm2, mm0 ; a1 =0+2
-
- pmulhw mm5, [GLOBAL(x_s1sqr2)] ;
- paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movq mm7, mm3 ;
- pmulhw mm7, [GLOBAL(x_c1sqr2less1)] ;
-
- paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw mm7, mm5 ; c1
-
- movq mm5, mm1
- movq mm4, mm3
-
- pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
- paddw mm5, mm1
-
- pmulhw mm3, [GLOBAL(x_s1sqr2)]
- paddw mm3, mm4
-
- paddw mm3, mm5 ; d1
- movq mm6, mm2 ; a1
-
- movq mm4, mm0 ; b1
- paddw mm2, mm3 ;0
-
- paddw mm4, mm7 ;1
- psubw mm0, mm7 ;2
-
- psubw mm6, mm3 ;3
-
- movq mm1, mm2 ; 03 02 01 00
- movq mm3, mm4 ; 23 22 21 20
-
- punpcklwd mm1, mm0 ; 11 01 10 00
- punpckhwd mm2, mm0 ; 13 03 12 02
-
- punpcklwd mm3, mm6 ; 31 21 30 20
- punpckhwd mm4, mm6 ; 33 23 32 22
-
- movq mm0, mm1 ; 11 01 10 00
- movq mm5, mm2 ; 13 03 12 02
-
- punpckldq mm0, mm3 ; 30 20 10 00
- punpckhdq mm1, mm3 ; 31 21 11 01
-
- punpckldq mm2, mm4 ; 32 22 12 02
- punpckhdq mm5, mm4 ; 33 23 13 03
-
- movq mm3, mm5 ; 33 23 13 03
-
- psubw mm0, mm2 ; b1= 0-2
- paddw mm2, mm2 ;
-
- movq mm5, mm1
- paddw mm2, mm0 ; a1 =0+2
-
- pmulhw mm5, [GLOBAL(x_s1sqr2)] ;
- paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movq mm7, mm3 ;
- pmulhw mm7, [GLOBAL(x_c1sqr2less1)] ;
-
- paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw mm7, mm5 ; c1
-
- movq mm5, mm1
- movq mm4, mm3
-
- pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
- paddw mm5, mm1
-
- pmulhw mm3, [GLOBAL(x_s1sqr2)]
- paddw mm3, mm4
-
- paddw mm3, mm5 ; d1
- paddw mm0, [GLOBAL(fours)]
-
- paddw mm2, [GLOBAL(fours)]
- movq mm6, mm2 ; a1
-
- movq mm4, mm0 ; b1
- paddw mm2, mm3 ;0
-
- paddw mm4, mm7 ;1
- psubw mm0, mm7 ;2
-
- psubw mm6, mm3 ;3
- psraw mm2, 3
-
- psraw mm0, 3
- psraw mm4, 3
-
- psraw mm6, 3
-
- movq mm1, mm2 ; 03 02 01 00
- movq mm3, mm4 ; 23 22 21 20
-
- punpcklwd mm1, mm0 ; 11 01 10 00
- punpckhwd mm2, mm0 ; 13 03 12 02
-
- punpcklwd mm3, mm6 ; 31 21 30 20
- punpckhwd mm4, mm6 ; 33 23 32 22
-
- movq mm0, mm1 ; 11 01 10 00
- movq mm5, mm2 ; 13 03 12 02
-
- punpckldq mm0, mm3 ; 30 20 10 00
- punpckhdq mm1, mm3 ; 31 21 11 01
-
- punpckldq mm2, mm4 ; 32 22 12 02
- punpckhdq mm5, mm4 ; 33 23 13 03
-
- movq [rdx], mm0
-
- movq [rdx+rax], mm1
- movq [rdx+rax*2], mm2
-
- add rdx, rax
- movq [rdx+rax*2], mm5
-
- ; begin epilog
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void short_idct4x4llm_1_mmx(short *input, short *output, int pitch)
-global sym(vp8_short_idct4x4llm_1_mmx)
-sym(vp8_short_idct4x4llm_1_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 3
- GET_GOT rbx
- ; end prolog
-
- mov rax, arg(0) ;input
- movd mm0, [rax]
-
- paddw mm0, [GLOBAL(fours)]
- mov rdx, arg(1) ;output
-
- psraw mm0, 3
- movsxd rax, dword ptr arg(2) ;pitch
-
- punpcklwd mm0, mm0
- punpckldq mm0, mm0
-
- movq [rdx], mm0
- movq [rdx+rax], mm0
-
- movq [rdx+rax*2], mm0
- add rdx, rax
-
- movq [rdx+rax*2], mm0
-
-
- ; begin epilog
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_dc_only_idct_add_mmx(short input_dc, unsigned char *pred_ptr, unsigned char *dst_ptr, int pitch, int stride)
-global sym(vp8_dc_only_idct_add_mmx)
-sym(vp8_dc_only_idct_add_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(1) ;s -- prediction
- mov rdi, arg(2) ;d -- destination
- movsxd rax, dword ptr arg(4) ;stride
- movsxd rdx, dword ptr arg(3) ;pitch
- pxor mm0, mm0
-
- movd mm5, arg(0) ;input_dc
-
- paddw mm5, [GLOBAL(fours)]
-
- psraw mm5, 3
-
- punpcklwd mm5, mm5
- punpckldq mm5, mm5
-
- movd mm1, [rsi]
- punpcklbw mm1, mm0
- paddsw mm1, mm5
- packuswb mm1, mm0 ; pack and unpack to saturate
- movd [rdi], mm1
-
- movd mm2, [rsi+rdx]
- punpcklbw mm2, mm0
- paddsw mm2, mm5
- packuswb mm2, mm0 ; pack and unpack to saturate
- movd [rdi+rax], mm2
-
- movd mm3, [rsi+2*rdx]
- punpcklbw mm3, mm0
- paddsw mm3, mm5
- packuswb mm3, mm0 ; pack and unpack to saturate
- movd [rdi+2*rax], mm3
-
- add rdi, rax
- add rsi, rdx
- movd mm4, [rsi+2*rdx]
- punpcklbw mm4, mm0
- paddsw mm4, mm5
- packuswb mm4, mm0 ; pack and unpack to saturate
- movd [rdi+2*rax], mm4
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-x_s1sqr2:
- times 4 dw 0x8A8C
-align 16
-x_c1sqr2less1:
- times 4 dw 0x4E7B
-align 16
-fours:
- times 4 dw 0x0004
diff --git a/vp8/common/x86/idctllm_sse2.asm b/vp8/common/x86/idctllm_sse2.asm
deleted file mode 100644
index edee157..0000000
--- a/vp8/common/x86/idctllm_sse2.asm
+++ /dev/null
@@ -1,708 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void idct_dequant_0_2x_sse2
-; (
-; short *qcoeff - 0
-; short *dequant - 1
-; unsigned char *pre - 2
-; unsigned char *dst - 3
-; int dst_stride - 4
-; int blk_stride - 5
-; )
-
-global sym(idct_dequant_0_2x_sse2)
-sym(idct_dequant_0_2x_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- ; end prolog
-
- mov rdx, arg(1) ; dequant
- mov rax, arg(0) ; qcoeff
-
- ; Zero out xmm7, for use unpacking
- pxor xmm7, xmm7
-
- movd xmm4, [rax]
- movd xmm5, [rdx]
-
- pinsrw xmm4, [rax+32], 4
- pinsrw xmm5, [rdx], 4
-
- pmullw xmm4, xmm5
-
- ; clear coeffs
- movd [rax], xmm7
- movd [rax+32], xmm7
-;pshufb
- pshuflw xmm4, xmm4, 00000000b
- pshufhw xmm4, xmm4, 00000000b
-
- mov rax, arg(2) ; pre
- paddw xmm4, [GLOBAL(fours)]
-
- movsxd rcx, dword ptr arg(5) ; blk_stride
- psraw xmm4, 3
-
- movq xmm0, [rax]
- movq xmm1, [rax+rcx]
- movq xmm2, [rax+2*rcx]
- lea rcx, [3*rcx]
- movq xmm3, [rax+rcx]
-
- punpcklbw xmm0, xmm7
- punpcklbw xmm1, xmm7
- punpcklbw xmm2, xmm7
- punpcklbw xmm3, xmm7
-
- mov rax, arg(3) ; dst
- movsxd rdx, dword ptr arg(4) ; dst_stride
-
- ; Add to predict buffer
- paddw xmm0, xmm4
- paddw xmm1, xmm4
- paddw xmm2, xmm4
- paddw xmm3, xmm4
-
- ; pack up before storing
- packuswb xmm0, xmm7
- packuswb xmm1, xmm7
- packuswb xmm2, xmm7
- packuswb xmm3, xmm7
-
- ; store blocks back out
- movq [rax], xmm0
- movq [rax + rdx], xmm1
-
- lea rax, [rax + 2*rdx]
-
- movq [rax], xmm2
- movq [rax + rdx], xmm3
-
- ; begin epilog
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-global sym(idct_dequant_full_2x_sse2)
-sym(idct_dequant_full_2x_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ; special case when 2 blocks have 0 or 1 coeffs
- ; dc is set as first coeff, so no need to load qcoeff
- mov rax, arg(0) ; qcoeff
- mov rsi, arg(2) ; pre
- mov rdi, arg(3) ; dst
- movsxd rcx, dword ptr arg(5) ; blk_stride
-
- ; Zero out xmm7, for use unpacking
- pxor xmm7, xmm7
-
- mov rdx, arg(1) ; dequant
-
- ; note the transpose of xmm1 and xmm2, necessary for shuffle
- ; to spit out sensicle data
- movdqa xmm0, [rax]
- movdqa xmm2, [rax+16]
- movdqa xmm1, [rax+32]
- movdqa xmm3, [rax+48]
-
- ; Clear out coeffs
- movdqa [rax], xmm7
- movdqa [rax+16], xmm7
- movdqa [rax+32], xmm7
- movdqa [rax+48], xmm7
-
- ; dequantize qcoeff buffer
- pmullw xmm0, [rdx]
- pmullw xmm2, [rdx+16]
- pmullw xmm1, [rdx]
- pmullw xmm3, [rdx+16]
-
- ; repack so block 0 row x and block 1 row x are together
- movdqa xmm4, xmm0
- punpckldq xmm0, xmm1
- punpckhdq xmm4, xmm1
-
- pshufd xmm0, xmm0, 11011000b
- pshufd xmm1, xmm4, 11011000b
-
- movdqa xmm4, xmm2
- punpckldq xmm2, xmm3
- punpckhdq xmm4, xmm3
-
- pshufd xmm2, xmm2, 11011000b
- pshufd xmm3, xmm4, 11011000b
-
- ; first pass
- psubw xmm0, xmm2 ; b1 = 0-2
- paddw xmm2, xmm2 ;
-
- movdqa xmm5, xmm1
- paddw xmm2, xmm0 ; a1 = 0+2
-
- pmulhw xmm5, [GLOBAL(x_s1sqr2)]
- paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movdqa xmm7, xmm3
- pmulhw xmm7, [GLOBAL(x_c1sqr2less1)]
-
- paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw xmm7, xmm5 ; c1
-
- movdqa xmm5, xmm1
- movdqa xmm4, xmm3
-
- pmulhw xmm5, [GLOBAL(x_c1sqr2less1)]
- paddw xmm5, xmm1
-
- pmulhw xmm3, [GLOBAL(x_s1sqr2)]
- paddw xmm3, xmm4
-
- paddw xmm3, xmm5 ; d1
- movdqa xmm6, xmm2 ; a1
-
- movdqa xmm4, xmm0 ; b1
- paddw xmm2, xmm3 ;0
-
- paddw xmm4, xmm7 ;1
- psubw xmm0, xmm7 ;2
-
- psubw xmm6, xmm3 ;3
-
- ; transpose for the second pass
- movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000
- punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000
- punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100
-
- movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008
- punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008
- punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108
-
-
- movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000
- punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000
- punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002
-
- movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100
- punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100
- punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102
-
-
- movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000
- punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000
- punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001
-
- movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002
- punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002
- punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003
-
- pshufd xmm0, xmm2, 11011000b
- pshufd xmm2, xmm1, 11011000b
-
- pshufd xmm1, xmm5, 11011000b
- pshufd xmm3, xmm7, 11011000b
-
- ; second pass
- psubw xmm0, xmm2 ; b1 = 0-2
- paddw xmm2, xmm2
-
- movdqa xmm5, xmm1
- paddw xmm2, xmm0 ; a1 = 0+2
-
- pmulhw xmm5, [GLOBAL(x_s1sqr2)]
- paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movdqa xmm7, xmm3
- pmulhw xmm7, [GLOBAL(x_c1sqr2less1)]
-
- paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw xmm7, xmm5 ; c1
-
- movdqa xmm5, xmm1
- movdqa xmm4, xmm3
-
- pmulhw xmm5, [GLOBAL(x_c1sqr2less1)]
- paddw xmm5, xmm1
-
- pmulhw xmm3, [GLOBAL(x_s1sqr2)]
- paddw xmm3, xmm4
-
- paddw xmm3, xmm5 ; d1
- paddw xmm0, [GLOBAL(fours)]
-
- paddw xmm2, [GLOBAL(fours)]
- movdqa xmm6, xmm2 ; a1
-
- movdqa xmm4, xmm0 ; b1
- paddw xmm2, xmm3 ;0
-
- paddw xmm4, xmm7 ;1
- psubw xmm0, xmm7 ;2
-
- psubw xmm6, xmm3 ;3
- psraw xmm2, 3
-
- psraw xmm0, 3
- psraw xmm4, 3
-
- psraw xmm6, 3
-
- ; transpose to save
- movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000
- punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000
- punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100
-
- movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008
- punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008
- punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108
-
-
- movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000
- punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000
- punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002
-
- movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100
- punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100
- punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102
-
-
- movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000
- punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000
- punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001
-
- movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002
- punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002
- punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003
-
- pshufd xmm0, xmm2, 11011000b
- pshufd xmm2, xmm1, 11011000b
-
- pshufd xmm1, xmm5, 11011000b
- pshufd xmm3, xmm7, 11011000b
-
- pxor xmm7, xmm7
-
- ; Load up predict blocks
- movq xmm4, [rsi]
- movq xmm5, [rsi+rcx]
-
- punpcklbw xmm4, xmm7
- punpcklbw xmm5, xmm7
-
- paddw xmm0, xmm4
- paddw xmm1, xmm5
-
- movq xmm4, [rsi+2*rcx]
- lea rcx, [3*rcx]
- movq xmm5, [rsi+rcx]
-
- punpcklbw xmm4, xmm7
- punpcklbw xmm5, xmm7
-
- paddw xmm2, xmm4
- paddw xmm3, xmm5
-
-.finish:
-
- ; pack up before storing
- packuswb xmm0, xmm7
- packuswb xmm1, xmm7
- packuswb xmm2, xmm7
- packuswb xmm3, xmm7
-
- ; Load destination stride before writing out,
- ; doesn't need to persist
- movsxd rdx, dword ptr arg(4) ; dst_stride
-
- ; store blocks back out
- movq [rdi], xmm0
- movq [rdi + rdx], xmm1
-
- lea rdi, [rdi + 2*rdx]
-
- movq [rdi], xmm2
- movq [rdi + rdx], xmm3
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void idct_dequant_dc_0_2x_sse2
-; (
-; short *qcoeff - 0
-; short *dequant - 1
-; unsigned char *pre - 2
-; unsigned char *dst - 3
-; int dst_stride - 4
-; short *dc - 5
-; )
-global sym(idct_dequant_dc_0_2x_sse2)
-sym(idct_dequant_dc_0_2x_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ; special case when 2 blocks have 0 or 1 coeffs
- ; dc is set as first coeff, so no need to load qcoeff
- mov rax, arg(0) ; qcoeff
- mov rsi, arg(2) ; pre
- mov rdi, arg(3) ; dst
- mov rdx, arg(5) ; dc
-
- ; Zero out xmm7, for use unpacking
- pxor xmm7, xmm7
-
- ; load up 2 dc words here == 2*16 = doubleword
- movd xmm4, [rdx]
-
- ; Load up predict blocks
- movq xmm0, [rsi]
- movq xmm1, [rsi+16]
- movq xmm2, [rsi+32]
- movq xmm3, [rsi+48]
-
- ; Duplicate and expand dc across
- punpcklwd xmm4, xmm4
- punpckldq xmm4, xmm4
-
- ; Rounding to dequant and downshift
- paddw xmm4, [GLOBAL(fours)]
- psraw xmm4, 3
-
- ; Predict buffer needs to be expanded from bytes to words
- punpcklbw xmm0, xmm7
- punpcklbw xmm1, xmm7
- punpcklbw xmm2, xmm7
- punpcklbw xmm3, xmm7
-
- ; Add to predict buffer
- paddw xmm0, xmm4
- paddw xmm1, xmm4
- paddw xmm2, xmm4
- paddw xmm3, xmm4
-
- ; pack up before storing
- packuswb xmm0, xmm7
- packuswb xmm1, xmm7
- packuswb xmm2, xmm7
- packuswb xmm3, xmm7
-
- ; Load destination stride before writing out,
- ; doesn't need to persist
- movsxd rdx, dword ptr arg(4) ; dst_stride
-
- ; store blocks back out
- movq [rdi], xmm0
- movq [rdi + rdx], xmm1
-
- lea rdi, [rdi + 2*rdx]
-
- movq [rdi], xmm2
- movq [rdi + rdx], xmm3
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-global sym(idct_dequant_dc_full_2x_sse2)
-sym(idct_dequant_dc_full_2x_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ; special case when 2 blocks have 0 or 1 coeffs
- ; dc is set as first coeff, so no need to load qcoeff
- mov rax, arg(0) ; qcoeff
- mov rsi, arg(2) ; pre
- mov rdi, arg(3) ; dst
-
- ; Zero out xmm7, for use unpacking
- pxor xmm7, xmm7
-
- mov rdx, arg(1) ; dequant
-
- ; note the transpose of xmm1 and xmm2, necessary for shuffle
- ; to spit out sensicle data
- movdqa xmm0, [rax]
- movdqa xmm2, [rax+16]
- movdqa xmm1, [rax+32]
- movdqa xmm3, [rax+48]
-
- ; Clear out coeffs
- movdqa [rax], xmm7
- movdqa [rax+16], xmm7
- movdqa [rax+32], xmm7
- movdqa [rax+48], xmm7
-
- ; dequantize qcoeff buffer
- pmullw xmm0, [rdx]
- pmullw xmm2, [rdx+16]
- pmullw xmm1, [rdx]
- pmullw xmm3, [rdx+16]
-
- ; DC component
- mov rdx, arg(5)
-
- ; repack so block 0 row x and block 1 row x are together
- movdqa xmm4, xmm0
- punpckldq xmm0, xmm1
- punpckhdq xmm4, xmm1
-
- pshufd xmm0, xmm0, 11011000b
- pshufd xmm1, xmm4, 11011000b
-
- movdqa xmm4, xmm2
- punpckldq xmm2, xmm3
- punpckhdq xmm4, xmm3
-
- pshufd xmm2, xmm2, 11011000b
- pshufd xmm3, xmm4, 11011000b
-
- ; insert DC component
- pinsrw xmm0, [rdx], 0
- pinsrw xmm0, [rdx+2], 4
-
- ; first pass
- psubw xmm0, xmm2 ; b1 = 0-2
- paddw xmm2, xmm2 ;
-
- movdqa xmm5, xmm1
- paddw xmm2, xmm0 ; a1 = 0+2
-
- pmulhw xmm5, [GLOBAL(x_s1sqr2)]
- paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movdqa xmm7, xmm3
- pmulhw xmm7, [GLOBAL(x_c1sqr2less1)]
-
- paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw xmm7, xmm5 ; c1
-
- movdqa xmm5, xmm1
- movdqa xmm4, xmm3
-
- pmulhw xmm5, [GLOBAL(x_c1sqr2less1)]
- paddw xmm5, xmm1
-
- pmulhw xmm3, [GLOBAL(x_s1sqr2)]
- paddw xmm3, xmm4
-
- paddw xmm3, xmm5 ; d1
- movdqa xmm6, xmm2 ; a1
-
- movdqa xmm4, xmm0 ; b1
- paddw xmm2, xmm3 ;0
-
- paddw xmm4, xmm7 ;1
- psubw xmm0, xmm7 ;2
-
- psubw xmm6, xmm3 ;3
-
- ; transpose for the second pass
- movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000
- punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000
- punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100
-
- movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008
- punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008
- punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108
-
-
- movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000
- punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000
- punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002
-
- movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100
- punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100
- punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102
-
-
- movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000
- punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000
- punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001
-
- movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002
- punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002
- punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003
-
- pshufd xmm0, xmm2, 11011000b
- pshufd xmm2, xmm1, 11011000b
-
- pshufd xmm1, xmm5, 11011000b
- pshufd xmm3, xmm7, 11011000b
-
- ; second pass
- psubw xmm0, xmm2 ; b1 = 0-2
- paddw xmm2, xmm2
-
- movdqa xmm5, xmm1
- paddw xmm2, xmm0 ; a1 = 0+2
-
- pmulhw xmm5, [GLOBAL(x_s1sqr2)]
- paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movdqa xmm7, xmm3
- pmulhw xmm7, [GLOBAL(x_c1sqr2less1)]
-
- paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw xmm7, xmm5 ; c1
-
- movdqa xmm5, xmm1
- movdqa xmm4, xmm3
-
- pmulhw xmm5, [GLOBAL(x_c1sqr2less1)]
- paddw xmm5, xmm1
-
- pmulhw xmm3, [GLOBAL(x_s1sqr2)]
- paddw xmm3, xmm4
-
- paddw xmm3, xmm5 ; d1
- paddw xmm0, [GLOBAL(fours)]
-
- paddw xmm2, [GLOBAL(fours)]
- movdqa xmm6, xmm2 ; a1
-
- movdqa xmm4, xmm0 ; b1
- paddw xmm2, xmm3 ;0
-
- paddw xmm4, xmm7 ;1
- psubw xmm0, xmm7 ;2
-
- psubw xmm6, xmm3 ;3
- psraw xmm2, 3
-
- psraw xmm0, 3
- psraw xmm4, 3
-
- psraw xmm6, 3
-
- ; transpose to save
- movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000
- punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000
- punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100
-
- movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008
- punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008
- punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108
-
-
- movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000
- punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000
- punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002
-
- movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100
- punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100
- punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102
-
-
- movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000
- punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000
- punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001
-
- movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002
- punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002
- punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003
-
- pshufd xmm0, xmm2, 11011000b
- pshufd xmm2, xmm1, 11011000b
-
- pshufd xmm1, xmm5, 11011000b
- pshufd xmm3, xmm7, 11011000b
-
- pxor xmm7, xmm7
-
- ; Load up predict blocks
- movq xmm4, [rsi]
- movq xmm5, [rsi+16]
-
- punpcklbw xmm4, xmm7
- punpcklbw xmm5, xmm7
-
- paddw xmm0, xmm4
- paddw xmm1, xmm5
-
- movq xmm4, [rsi+32]
- movq xmm5, [rsi+48]
-
- punpcklbw xmm4, xmm7
- punpcklbw xmm5, xmm7
-
- paddw xmm2, xmm4
- paddw xmm3, xmm5
-
-.finish:
-
- ; pack up before storing
- packuswb xmm0, xmm7
- packuswb xmm1, xmm7
- packuswb xmm2, xmm7
- packuswb xmm3, xmm7
-
- ; Load destination stride before writing out,
- ; doesn't need to persist
- movsxd rdx, dword ptr arg(4) ; dst_stride
-
- ; store blocks back out
- movq [rdi], xmm0
- movq [rdi + rdx], xmm1
-
- lea rdi, [rdi + 2*rdx]
-
- movq [rdi], xmm2
- movq [rdi + rdx], xmm3
-
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-fours:
- times 8 dw 0x0004
-align 16
-x_s1sqr2:
- times 8 dw 0x8A8C
-align 16
-x_c1sqr2less1:
- times 8 dw 0x4E7B
diff --git a/vp8/common/x86/iwalsh_mmx.asm b/vp8/common/x86/iwalsh_mmx.asm
deleted file mode 100644
index 10b5274..0000000
--- a/vp8/common/x86/iwalsh_mmx.asm
+++ /dev/null
@@ -1,173 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_short_inv_walsh4x4_1_mmx(short *input, short *output)
-global sym(vp8_short_inv_walsh4x4_1_mmx)
-sym(vp8_short_inv_walsh4x4_1_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 2
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0)
- mov rax, 3
-
- mov rdi, arg(1)
- add rax, [rsi] ;input[0] + 3
-
- movd mm0, eax
-
- punpcklwd mm0, mm0 ;x x val val
-
- punpckldq mm0, mm0 ;val val val val
-
- psraw mm0, 3 ;(input[0] + 3) >> 3
-
- movq [rdi + 0], mm0
- movq [rdi + 8], mm0
- movq [rdi + 16], mm0
- movq [rdi + 24], mm0
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_short_inv_walsh4x4_mmx(short *input, short *output)
-global sym(vp8_short_inv_walsh4x4_mmx)
-sym(vp8_short_inv_walsh4x4_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 2
- push rsi
- push rdi
- ; end prolog
-
- mov rax, 3
- mov rsi, arg(0)
- mov rdi, arg(1)
- shl rax, 16
-
- movq mm0, [rsi + 0] ;ip[0]
- movq mm1, [rsi + 8] ;ip[4]
- or rax, 3 ;00030003h
-
- movq mm2, [rsi + 16] ;ip[8]
- movq mm3, [rsi + 24] ;ip[12]
-
- movq mm7, rax
- movq mm4, mm0
-
- punpcklwd mm7, mm7 ;0003000300030003h
- movq mm5, mm1
-
- paddw mm4, mm3 ;ip[0] + ip[12] aka al
- paddw mm5, mm2 ;ip[4] + ip[8] aka bl
-
- movq mm6, mm4 ;temp al
-
- paddw mm4, mm5 ;al + bl
- psubw mm6, mm5 ;al - bl
-
- psubw mm0, mm3 ;ip[0] - ip[12] aka d1
- psubw mm1, mm2 ;ip[4] - ip[8] aka c1
-
- movq mm5, mm0 ;temp dl
-
- paddw mm0, mm1 ;dl + cl
- psubw mm5, mm1 ;dl - cl
-
- ; 03 02 01 00
- ; 13 12 11 10
- ; 23 22 21 20
- ; 33 32 31 30
-
- movq mm3, mm4 ; 03 02 01 00
- punpcklwd mm4, mm0 ; 11 01 10 00
- punpckhwd mm3, mm0 ; 13 03 12 02
-
- movq mm1, mm6 ; 23 22 21 20
- punpcklwd mm6, mm5 ; 31 21 30 20
- punpckhwd mm1, mm5 ; 33 23 32 22
-
- movq mm0, mm4 ; 11 01 10 00
- movq mm2, mm3 ; 13 03 12 02
-
- punpckldq mm0, mm6 ; 30 20 10 00 aka ip[0]
- punpckhdq mm4, mm6 ; 31 21 11 01 aka ip[4]
-
- punpckldq mm2, mm1 ; 32 22 12 02 aka ip[8]
- punpckhdq mm3, mm1 ; 33 23 13 03 aka ip[12]
-;~~~~~~~~~~~~~~~~~~~~~
- movq mm1, mm0
- movq mm5, mm4
-
- paddw mm1, mm3 ;ip[0] + ip[12] aka al
- paddw mm5, mm2 ;ip[4] + ip[8] aka bl
-
- movq mm6, mm1 ;temp al
-
- paddw mm1, mm5 ;al + bl
- psubw mm6, mm5 ;al - bl
-
- psubw mm0, mm3 ;ip[0] - ip[12] aka d1
- psubw mm4, mm2 ;ip[4] - ip[8] aka c1
-
- movq mm5, mm0 ;temp dl
-
- paddw mm0, mm4 ;dl + cl
- psubw mm5, mm4 ;dl - cl
-;~~~~~~~~~~~~~~~~~~~~~
- movq mm3, mm1 ; 03 02 01 00
- punpcklwd mm1, mm0 ; 11 01 10 00
- punpckhwd mm3, mm0 ; 13 03 12 02
-
- movq mm4, mm6 ; 23 22 21 20
- punpcklwd mm6, mm5 ; 31 21 30 20
- punpckhwd mm4, mm5 ; 33 23 32 22
-
- movq mm0, mm1 ; 11 01 10 00
- movq mm2, mm3 ; 13 03 12 02
-
- punpckldq mm0, mm6 ; 30 20 10 00 aka ip[0]
- punpckhdq mm1, mm6 ; 31 21 11 01 aka ip[4]
-
- punpckldq mm2, mm4 ; 32 22 12 02 aka ip[8]
- punpckhdq mm3, mm4 ; 33 23 13 03 aka ip[12]
-
- paddw mm0, mm7
- paddw mm1, mm7
- paddw mm2, mm7
- paddw mm3, mm7
-
- psraw mm0, 3
- psraw mm1, 3
- psraw mm2, 3
- psraw mm3, 3
-
- movq [rdi + 0], mm0
- movq [rdi + 8], mm1
- movq [rdi + 16], mm2
- movq [rdi + 24], mm3
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
diff --git a/vp8/common/x86/iwalsh_sse2.asm b/vp8/common/x86/iwalsh_sse2.asm
deleted file mode 100644
index 83c97df..0000000
--- a/vp8/common/x86/iwalsh_sse2.asm
+++ /dev/null
@@ -1,119 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_short_inv_walsh4x4_sse2(short *input, short *output)
-global sym(vp8_short_inv_walsh4x4_sse2)
-sym(vp8_short_inv_walsh4x4_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 2
- SAVE_XMM
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0)
- mov rdi, arg(1)
- mov rax, 3
-
- movdqa xmm0, [rsi + 0] ;ip[4] ip[0]
- movdqa xmm1, [rsi + 16] ;ip[12] ip[8]
-
- shl rax, 16
- or rax, 3 ;00030003h
-
- pshufd xmm2, xmm1, 4eh ;ip[8] ip[12]
- movdqa xmm3, xmm0 ;ip[4] ip[0]
-
- paddw xmm0, xmm2 ;ip[4]+ip[8] ip[0]+ip[12] aka b1 a1
- psubw xmm3, xmm2 ;ip[4]-ip[8] ip[0]-ip[12] aka c1 d1
-
- movdqa xmm4, xmm0
- punpcklqdq xmm0, xmm3 ;d1 a1
- punpckhqdq xmm4, xmm3 ;c1 b1
- movd xmm7, eax
-
- movdqa xmm1, xmm4 ;c1 b1
- paddw xmm4, xmm0 ;dl+cl a1+b1 aka op[4] op[0]
- psubw xmm0, xmm1 ;d1-c1 a1-b1 aka op[12] op[8]
-
-;;;temp output
-;; movdqu [rdi + 0], xmm4
-;; movdqu [rdi + 16], xmm3
-
-;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- ; 13 12 11 10 03 02 01 00
- ;
- ; 33 32 31 30 23 22 21 20
- ;
- movdqa xmm3, xmm4 ; 13 12 11 10 03 02 01 00
- punpcklwd xmm4, xmm0 ; 23 03 22 02 21 01 20 00
- punpckhwd xmm3, xmm0 ; 33 13 32 12 31 11 30 10
- movdqa xmm1, xmm4 ; 23 03 22 02 21 01 20 00
- punpcklwd xmm4, xmm3 ; 31 21 11 01 30 20 10 00
- punpckhwd xmm1, xmm3 ; 33 23 13 03 32 22 12 02
- ;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- pshufd xmm2, xmm1, 4eh ;ip[8] ip[12]
- movdqa xmm3, xmm4 ;ip[4] ip[0]
-
- pshufd xmm7, xmm7, 0 ;03 03 03 03 03 03 03 03
-
- paddw xmm4, xmm2 ;ip[4]+ip[8] ip[0]+ip[12] aka b1 a1
- psubw xmm3, xmm2 ;ip[4]-ip[8] ip[0]-ip[12] aka c1 d1
-
- movdqa xmm5, xmm4
- punpcklqdq xmm4, xmm3 ;d1 a1
- punpckhqdq xmm5, xmm3 ;c1 b1
-
- movdqa xmm1, xmm5 ;c1 b1
- paddw xmm5, xmm4 ;dl+cl a1+b1 aka op[4] op[0]
- psubw xmm4, xmm1 ;d1-c1 a1-b1 aka op[12] op[8]
-;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- ; 13 12 11 10 03 02 01 00
- ;
- ; 33 32 31 30 23 22 21 20
- ;
- movdqa xmm0, xmm5 ; 13 12 11 10 03 02 01 00
- punpcklwd xmm5, xmm4 ; 23 03 22 02 21 01 20 00
- punpckhwd xmm0, xmm4 ; 33 13 32 12 31 11 30 10
- movdqa xmm1, xmm5 ; 23 03 22 02 21 01 20 00
- punpcklwd xmm5, xmm0 ; 31 21 11 01 30 20 10 00
- punpckhwd xmm1, xmm0 ; 33 23 13 03 32 22 12 02
-;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- paddw xmm5, xmm7
- paddw xmm1, xmm7
-
- psraw xmm5, 3
- psraw xmm1, 3
-
- movdqa [rdi + 0], xmm5
- movdqa [rdi + 16], xmm1
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-x_s1sqr2:
- times 4 dw 0x8A8C
-align 16
-x_c1sqr2less1:
- times 4 dw 0x4E7B
-align 16
-fours:
- times 4 dw 0x0004
diff --git a/vp8/common/x86/loopfilter_mmx.asm b/vp8/common/x86/loopfilter_mmx.asm
deleted file mode 100644
index c6c215c..0000000
--- a/vp8/common/x86/loopfilter_mmx.asm
+++ /dev/null
@@ -1,1777 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-
-;void vp8_loop_filter_horizontal_edge_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_horizontal_edge_mmx)
-sym(vp8_loop_filter_horizontal_edge_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 32 ; reserve 32 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
-
- movsxd rcx, dword ptr arg(5) ;count
-next8_h:
- mov rdx, arg(3) ;limit
- movq mm7, [rdx]
- mov rdi, rsi ; rdi points to row +1 for indirect addressing
- add rdi, rax
-
- ; calculate breakout conditions
- movq mm2, [rdi+2*rax] ; q3
- movq mm1, [rsi+2*rax] ; q2
- movq mm6, mm1 ; q2
- psubusb mm1, mm2 ; q2-=q3
- psubusb mm2, mm6 ; q3-=q2
- por mm1, mm2 ; abs(q3-q2)
- psubusb mm1, mm7 ;
-
-
- movq mm4, [rsi+rax] ; q1
- movq mm3, mm4 ; q1
- psubusb mm4, mm6 ; q1-=q2
- psubusb mm6, mm3 ; q2-=q1
- por mm4, mm6 ; abs(q2-q1)
-
- psubusb mm4, mm7
- por mm1, mm4
-
- movq mm4, [rsi] ; q0
- movq mm0, mm4 ; q0
- psubusb mm4, mm3 ; q0-=q1
- psubusb mm3, mm0 ; q1-=q0
- por mm4, mm3 ; abs(q0-q1)
- movq t0, mm4 ; save to t0
- psubusb mm4, mm7
- por mm1, mm4
-
-
- neg rax ; negate pitch to deal with above border
-
- movq mm2, [rsi+4*rax] ; p3
- movq mm4, [rdi+4*rax] ; p2
- movq mm5, mm4 ; p2
- psubusb mm4, mm2 ; p2-=p3
- psubusb mm2, mm5 ; p3-=p2
- por mm4, mm2 ; abs(p3 - p2)
- psubusb mm4, mm7
- por mm1, mm4
-
-
- movq mm4, [rsi+2*rax] ; p1
- movq mm3, mm4 ; p1
- psubusb mm4, mm5 ; p1-=p2
- psubusb mm5, mm3 ; p2-=p1
- por mm4, mm5 ; abs(p2 - p1)
- psubusb mm4, mm7
- por mm1, mm4
-
- movq mm2, mm3 ; p1
-
- movq mm4, [rsi+rax] ; p0
- movq mm5, mm4 ; p0
- psubusb mm4, mm3 ; p0-=p1
- psubusb mm3, mm5 ; p1-=p0
- por mm4, mm3 ; abs(p1 - p0)
- movq t1, mm4 ; save to t1
- psubusb mm4, mm7
- por mm1, mm4
-
- movq mm3, [rdi] ; q1
- movq mm4, mm3 ; q1
- psubusb mm3, mm2 ; q1-=p1
- psubusb mm2, mm4 ; p1-=q1
- por mm2, mm3 ; abs(p1-q1)
- pand mm2, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psrlw mm2, 1 ; abs(p1-q1)/2
-
- movq mm6, mm5 ; p0
- movq mm3, [rsi] ; q0
- psubusb mm5, mm3 ; p0-=q0
- psubusb mm3, mm6 ; q0-=p0
- por mm5, mm3 ; abs(p0 - q0)
- paddusb mm5, mm5 ; abs(p0-q0)*2
- paddusb mm5, mm2 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- mov rdx, arg(2) ;flimit ; get flimit
- movq mm2, [rdx] ; flimit mm2
- paddb mm2, mm2 ; flimit*2 (less than 255)
- paddb mm7, mm2 ; flimit * 2 + limit (less than 255)
-
- psubusb mm5, mm7 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- por mm1, mm5
- pxor mm5, mm5
- pcmpeqb mm1, mm5 ; mask mm1
-
- ; calculate high edge variance
- mov rdx, arg(4) ;thresh ; get thresh
- movq mm7, [rdx] ;
- movq mm4, t0 ; get abs (q1 - q0)
- psubusb mm4, mm7
- movq mm3, t1 ; get abs (p1 - p0)
- psubusb mm3, mm7
- paddb mm4, mm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
-
- pcmpeqb mm4, mm5
-
- pcmpeqb mm5, mm5
- pxor mm4, mm5
-
-
- ; start work on filters
- movq mm2, [rsi+2*rax] ; p1
- movq mm7, [rdi] ; q1
- pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
- psubsb mm2, mm7 ; p1 - q1
- pand mm2, mm4 ; high var mask (hvm)(p1 - q1)
- pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
- pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
- movq mm3, mm0 ; q0
- psubsb mm0, mm6 ; q0 - p0
- paddsb mm2, mm0 ; 1 * (q0 - p0) + hvm(p1 - q1)
- paddsb mm2, mm0 ; 2 * (q0 - p0) + hvm(p1 - q1)
- paddsb mm2, mm0 ; 3 * (q0 - p0) + hvm(p1 - q1)
- pand mm1, mm2 ; mask filter values we don't care about
- movq mm2, mm1
- paddsb mm1, [GLOBAL(t4)] ; 3* (q0 - p0) + hvm(p1 - q1) + 4
- paddsb mm2, [GLOBAL(t3)] ; 3* (q0 - p0) + hvm(p1 - q1) + 3
-
- pxor mm0, mm0 ;
- pxor mm5, mm5
- punpcklbw mm0, mm2 ;
- punpckhbw mm5, mm2 ;
- psraw mm0, 11 ;
- psraw mm5, 11
- packsswb mm0, mm5
- movq mm2, mm0 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3;
-
- pxor mm0, mm0 ; 0
- movq mm5, mm1 ; abcdefgh
- punpcklbw mm0, mm1 ; e0f0g0h0
- psraw mm0, 11 ; sign extended shift right by 3
- pxor mm1, mm1 ; 0
- punpckhbw mm1, mm5 ; a0b0c0d0
- psraw mm1, 11 ; sign extended shift right by 3
- movq mm5, mm0 ; save results
-
- packsswb mm0, mm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3
- paddsw mm5, [GLOBAL(ones)]
- paddsw mm1, [GLOBAL(ones)]
- psraw mm5, 1 ; partial shifted one more time for 2nd tap
- psraw mm1, 1 ; partial shifted one more time for 2nd tap
- packsswb mm5, mm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4
- pandn mm4, mm5 ; high edge variance additive
-
- paddsb mm6, mm2 ; p0+= p0 add
- pxor mm6, [GLOBAL(t80)] ; unoffset
- movq [rsi+rax], mm6 ; write back
-
- movq mm6, [rsi+2*rax] ; p1
- pxor mm6, [GLOBAL(t80)] ; reoffset
- paddsb mm6, mm4 ; p1+= p1 add
- pxor mm6, [GLOBAL(t80)] ; unoffset
- movq [rsi+2*rax], mm6 ; write back
-
- psubsb mm3, mm0 ; q0-= q0 add
- pxor mm3, [GLOBAL(t80)] ; unoffset
- movq [rsi], mm3 ; write back
-
- psubsb mm7, mm4 ; q1-= q1 add
- pxor mm7, [GLOBAL(t80)] ; unoffset
- movq [rdi], mm7 ; write back
-
- add rsi,8
- neg rax
- dec rcx
- jnz next8_h
-
- add rsp, 32
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_loop_filter_vertical_edge_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_vertical_edge_mmx)
-sym(vp8_loop_filter_vertical_edge_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 64 ; reserve 64 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
- %define srct [rsp + 32] ;__declspec(align(16)) char srct[32];
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
-
- lea rsi, [rsi + rax*4 - 4]
-
- movsxd rcx, dword ptr arg(5) ;count
-next8_v:
- mov rdi, rsi ; rdi points to row +1 for indirect addressing
- add rdi, rax
-
-
- ;transpose
- movq mm6, [rsi+2*rax] ; 67 66 65 64 63 62 61 60
- movq mm7, mm6 ; 77 76 75 74 73 72 71 70
-
- punpckhbw mm7, [rdi+2*rax] ; 77 67 76 66 75 65 74 64
- punpcklbw mm6, [rdi+2*rax] ; 73 63 72 62 71 61 70 60
-
- movq mm4, [rsi] ; 47 46 45 44 43 42 41 40
- movq mm5, mm4 ; 47 46 45 44 43 42 41 40
-
- punpckhbw mm5, [rsi+rax] ; 57 47 56 46 55 45 54 44
- punpcklbw mm4, [rsi+rax] ; 53 43 52 42 51 41 50 40
-
- movq mm3, mm5 ; 57 47 56 46 55 45 54 44
- punpckhwd mm5, mm7 ; 77 67 57 47 76 66 56 46
-
- punpcklwd mm3, mm7 ; 75 65 55 45 74 64 54 44
- movq mm2, mm4 ; 53 43 52 42 51 41 50 40
-
- punpckhwd mm4, mm6 ; 73 63 53 43 72 62 52 42
- punpcklwd mm2, mm6 ; 71 61 51 41 70 60 50 40
-
- neg rax
- movq mm6, [rsi+rax*2] ; 27 26 25 24 23 22 21 20
-
- movq mm1, mm6 ; 27 26 25 24 23 22 21 20
- punpckhbw mm6, [rsi+rax] ; 37 27 36 36 35 25 34 24
-
- punpcklbw mm1, [rsi+rax] ; 33 23 32 22 31 21 30 20
- movq mm7, [rsi+rax*4]; ; 07 06 05 04 03 02 01 00
-
- punpckhbw mm7, [rdi+rax*4] ; 17 07 16 06 15 05 14 04
- movq mm0, mm7 ; 17 07 16 06 15 05 14 04
-
- punpckhwd mm7, mm6 ; 37 27 17 07 36 26 16 06
- punpcklwd mm0, mm6 ; 35 25 15 05 34 24 14 04
-
- movq mm6, mm7 ; 37 27 17 07 36 26 16 06
- punpckhdq mm7, mm5 ; 77 67 57 47 37 27 17 07 = q3
-
- punpckldq mm6, mm5 ; 76 66 56 46 36 26 16 06 = q2
-
- movq mm5, mm6 ; 76 66 56 46 36 26 16 06
- psubusb mm5, mm7 ; q2-q3
-
- psubusb mm7, mm6 ; q3-q2
- por mm7, mm5; ; mm7=abs (q3-q2)
-
- movq mm5, mm0 ; 35 25 15 05 34 24 14 04
- punpckhdq mm5, mm3 ; 75 65 55 45 35 25 15 05 = q1
-
- punpckldq mm0, mm3 ; 74 64 54 44 34 24 15 04 = q0
- movq mm3, mm5 ; 75 65 55 45 35 25 15 05 = q1
-
- psubusb mm3, mm6 ; q1-q2
- psubusb mm6, mm5 ; q2-q1
-
- por mm6, mm3 ; mm6=abs(q2-q1)
- lea rdx, srct
-
- movq [rdx+24], mm5 ; save q1
- movq [rdx+16], mm0 ; save q0
-
- movq mm3, [rsi+rax*4] ; 07 06 05 04 03 02 01 00
- punpcklbw mm3, [rdi+rax*4] ; 13 03 12 02 11 01 10 00
-
- movq mm0, mm3 ; 13 03 12 02 11 01 10 00
- punpcklwd mm0, mm1 ; 31 21 11 01 30 20 10 00
-
- punpckhwd mm3, mm1 ; 33 23 13 03 32 22 12 02
- movq mm1, mm0 ; 31 21 11 01 30 20 10 00
-
- punpckldq mm0, mm2 ; 70 60 50 40 30 20 10 00 =p3
- punpckhdq mm1, mm2 ; 71 61 51 41 31 21 11 01 =p2
-
- movq mm2, mm1 ; 71 61 51 41 31 21 11 01 =p2
- psubusb mm2, mm0 ; p2-p3
-
- psubusb mm0, mm1 ; p3-p2
- por mm0, mm2 ; mm0=abs(p3-p2)
-
- movq mm2, mm3 ; 33 23 13 03 32 22 12 02
- punpckldq mm2, mm4 ; 72 62 52 42 32 22 12 02 = p1
-
- punpckhdq mm3, mm4 ; 73 63 53 43 33 23 13 03 = p0
- movq [rdx+8], mm3 ; save p0
-
- movq [rdx], mm2 ; save p1
- movq mm5, mm2 ; mm5 = p1
-
- psubusb mm2, mm1 ; p1-p2
- psubusb mm1, mm5 ; p2-p1
-
- por mm1, mm2 ; mm1=abs(p2-p1)
- mov rdx, arg(3) ;limit
-
- movq mm4, [rdx] ; mm4 = limit
- psubusb mm7, mm4
-
- psubusb mm0, mm4
- psubusb mm1, mm4
-
- psubusb mm6, mm4
- por mm7, mm6
-
- por mm0, mm1
- por mm0, mm7 ; abs(q3-q2) > limit || abs(p3-p2) > limit ||abs(p2-p1) > limit || abs(q2-q1) > limit
-
- movq mm1, mm5 ; p1
-
- movq mm7, mm3 ; mm3=mm7=p0
- psubusb mm7, mm5 ; p0 - p1
-
- psubusb mm5, mm3 ; p1 - p0
- por mm5, mm7 ; abs(p1-p0)
-
- movq t0, mm5 ; save abs(p1-p0)
- lea rdx, srct
-
- psubusb mm5, mm4
- por mm0, mm5 ; mm0=mask
-
- movq mm5, [rdx+16] ; mm5=q0
- movq mm7, [rdx+24] ; mm7=q1
-
- movq mm6, mm5 ; mm6=q0
- movq mm2, mm7 ; q1
- psubusb mm5, mm7 ; q0-q1
-
- psubusb mm7, mm6 ; q1-q0
- por mm7, mm5 ; abs(q1-q0)
-
- movq t1, mm7 ; save abs(q1-q0)
- psubusb mm7, mm4
-
- por mm0, mm7 ; mask
-
- movq mm5, mm2 ; q1
- psubusb mm5, mm1 ; q1-=p1
- psubusb mm1, mm2 ; p1-=q1
- por mm5, mm1 ; abs(p1-q1)
- pand mm5, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psrlw mm5, 1 ; abs(p1-q1)/2
-
- mov rdx, arg(2) ;flimit ;
-
- movq mm2, [rdx] ;flimit mm2
- movq mm1, mm3 ; mm1=mm3=p0
-
- movq mm7, mm6 ; mm7=mm6=q0
- psubusb mm1, mm7 ; p0-q0
-
- psubusb mm7, mm3 ; q0-p0
- por mm1, mm7 ; abs(q0-p0)
- paddusb mm1, mm1 ; abs(q0-p0)*2
- paddusb mm1, mm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- paddb mm2, mm2 ; flimit*2 (less than 255)
- paddb mm4, mm2 ; flimit * 2 + limit (less than 255)
-
- psubusb mm1, mm4 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- por mm1, mm0; ; mask
-
- pxor mm0, mm0
- pcmpeqb mm1, mm0
-
- ; calculate high edge variance
- mov rdx, arg(4) ;thresh ; get thresh
- movq mm7, [rdx]
- ;
- movq mm4, t0 ; get abs (q1 - q0)
- psubusb mm4, mm7
-
- movq mm3, t1 ; get abs (p1 - p0)
- psubusb mm3, mm7
-
- por mm4, mm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
- pcmpeqb mm4, mm0
-
- pcmpeqb mm0, mm0
- pxor mm4, mm0
-
-
-
- ; start work on filters
- lea rdx, srct
-
- movq mm2, [rdx] ; p1
- movq mm7, [rdx+24] ; q1
-
- movq mm6, [rdx+8] ; p0
- movq mm0, [rdx+16] ; q0
-
- pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
-
- psubsb mm2, mm7 ; p1 - q1
- pand mm2, mm4 ; high var mask (hvm)(p1 - q1)
-
- pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
- pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
-
- movq mm3, mm0 ; q0
- psubsb mm0, mm6 ; q0 - p0
-
- paddsb mm2, mm0 ; 1 * (q0 - p0) + hvm(p1 - q1)
- paddsb mm2, mm0 ; 2 * (q0 - p0) + hvm(p1 - q1)
-
- paddsb mm2, mm0 ; 3 * (q0 - p0) + hvm(p1 - q1)
- pand mm1, mm2 ; mask filter values we don't care about
-
- movq mm2, mm1
- paddsb mm1, [GLOBAL(t4)] ; 3* (q0 - p0) + hvm(p1 - q1) + 4
-
- paddsb mm2, [GLOBAL(t3)] ; 3* (q0 - p0) + hvm(p1 - q1) + 3
- pxor mm0, mm0 ;
-
- pxor mm5, mm5
- punpcklbw mm0, mm2 ;
-
- punpckhbw mm5, mm2 ;
- psraw mm0, 11 ;
-
- psraw mm5, 11
- packsswb mm0, mm5
-
- movq mm2, mm0 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3;
-
- pxor mm0, mm0 ; 0
- movq mm5, mm1 ; abcdefgh
-
- punpcklbw mm0, mm1 ; e0f0g0h0
- psraw mm0, 11 ; sign extended shift right by 3
-
- pxor mm1, mm1 ; 0
- punpckhbw mm1, mm5 ; a0b0c0d0
-
- psraw mm1, 11 ; sign extended shift right by 3
- movq mm5, mm0 ; save results
-
- packsswb mm0, mm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3
- paddsw mm5, [GLOBAL(ones)]
-
- paddsw mm1, [GLOBAL(ones)]
- psraw mm5, 1 ; partial shifted one more time for 2nd tap
-
- psraw mm1, 1 ; partial shifted one more time for 2nd tap
- packsswb mm5, mm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4
-
- pandn mm4, mm5 ; high edge variance additive
-
- paddsb mm6, mm2 ; p0+= p0 add
- pxor mm6, [GLOBAL(t80)] ; unoffset
-
- ; mm6=p0 ;
- movq mm1, [rdx] ; p1
- pxor mm1, [GLOBAL(t80)] ; reoffset
-
- paddsb mm1, mm4 ; p1+= p1 add
- pxor mm1, [GLOBAL(t80)] ; unoffset
- ; mm6 = p0 mm1 = p1
-
- psubsb mm3, mm0 ; q0-= q0 add
- pxor mm3, [GLOBAL(t80)] ; unoffset
-
- ; mm3 = q0
- psubsb mm7, mm4 ; q1-= q1 add
- pxor mm7, [GLOBAL(t80)] ; unoffset
- ; mm7 = q1
-
- ; tranpose and write back
- ; mm1 = 72 62 52 42 32 22 12 02
- ; mm6 = 73 63 53 43 33 23 13 03
- ; mm3 = 74 64 54 44 34 24 14 04
- ; mm7 = 75 65 55 45 35 25 15 05
-
- movq mm2, mm1 ; 72 62 52 42 32 22 12 02
- punpcklbw mm2, mm6 ; 33 32 23 22 13 12 03 02
-
- movq mm4, mm3 ; 74 64 54 44 34 24 14 04
- punpckhbw mm1, mm6 ; 73 72 63 62 53 52 43 42
-
- punpcklbw mm4, mm7 ; 35 34 25 24 15 14 05 04
- punpckhbw mm3, mm7 ; 75 74 65 64 55 54 45 44
-
- movq mm6, mm2 ; 33 32 23 22 13 12 03 02
- punpcklwd mm2, mm4 ; 15 14 13 12 05 04 03 02
-
- punpckhwd mm6, mm4 ; 35 34 33 32 25 24 23 22
- movq mm5, mm1 ; 73 72 63 62 53 52 43 42
-
- punpcklwd mm1, mm3 ; 55 54 53 52 45 44 43 42
- punpckhwd mm5, mm3 ; 75 74 73 72 65 64 63 62
-
-
- ; mm2 = 15 14 13 12 05 04 03 02
- ; mm6 = 35 34 33 32 25 24 23 22
- ; mm5 = 55 54 53 52 45 44 43 42
- ; mm1 = 75 74 73 72 65 64 63 62
-
-
-
- movd [rsi+rax*4+2], mm2
- psrlq mm2, 32
-
- movd [rdi+rax*4+2], mm2
- movd [rsi+rax*2+2], mm6
-
- psrlq mm6, 32
- movd [rsi+rax+2],mm6
-
- movd [rsi+2], mm1
- psrlq mm1, 32
-
- movd [rdi+2], mm1
- neg rax
-
- movd [rdi+rax+2],mm5
- psrlq mm5, 32
-
- movd [rdi+rax*2+2], mm5
-
- lea rsi, [rsi+rax*8]
- dec rcx
- jnz next8_v
-
- add rsp, 64
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_mbloop_filter_horizontal_edge_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_mbloop_filter_horizontal_edge_mmx)
-sym(vp8_mbloop_filter_horizontal_edge_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 32 ; reserve 32 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
-
- movsxd rcx, dword ptr arg(5) ;count
-next8_mbh:
- mov rdx, arg(3) ;limit
- movq mm7, [rdx]
- mov rdi, rsi ; rdi points to row +1 for indirect addressing
- add rdi, rax
-
- ; calculate breakout conditions
- movq mm2, [rdi+2*rax] ; q3
-
- movq mm1, [rsi+2*rax] ; q2
- movq mm6, mm1 ; q2
- psubusb mm1, mm2 ; q2-=q3
- psubusb mm2, mm6 ; q3-=q2
- por mm1, mm2 ; abs(q3-q2)
- psubusb mm1, mm7
-
-
- ; mm1 = abs(q3-q2), mm6 =q2, mm7 = limit
- movq mm4, [rsi+rax] ; q1
- movq mm3, mm4 ; q1
- psubusb mm4, mm6 ; q1-=q2
- psubusb mm6, mm3 ; q2-=q1
- por mm4, mm6 ; abs(q2-q1)
- psubusb mm4, mm7
- por mm1, mm4
-
-
- ; mm1 = mask, mm3=q1, mm7 = limit
-
- movq mm4, [rsi] ; q0
- movq mm0, mm4 ; q0
- psubusb mm4, mm3 ; q0-=q1
- psubusb mm3, mm0 ; q1-=q0
- por mm4, mm3 ; abs(q0-q1)
- movq t0, mm4 ; save to t0
- psubusb mm4, mm7
- por mm1, mm4
-
-
- ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1)
-
- neg rax ; negate pitch to deal with above border
-
- movq mm2, [rsi+4*rax] ; p3
- movq mm4, [rdi+4*rax] ; p2
- movq mm5, mm4 ; p2
- psubusb mm4, mm2 ; p2-=p3
- psubusb mm2, mm5 ; p3-=p2
- por mm4, mm2 ; abs(p3 - p2)
- psubusb mm4, mm7
- por mm1, mm4
- ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1)
-
- movq mm4, [rsi+2*rax] ; p1
- movq mm3, mm4 ; p1
- psubusb mm4, mm5 ; p1-=p2
- psubusb mm5, mm3 ; p2-=p1
- por mm4, mm5 ; abs(p2 - p1)
- psubusb mm4, mm7
- por mm1, mm4
-
- movq mm2, mm3 ; p1
-
-
- ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1)
-
- movq mm4, [rsi+rax] ; p0
- movq mm5, mm4 ; p0
- psubusb mm4, mm3 ; p0-=p1
- psubusb mm3, mm5 ; p1-=p0
- por mm4, mm3 ; abs(p1 - p0)
- movq t1, mm4 ; save to t1
- psubusb mm4, mm7
- por mm1, mm4
- ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1) t1 = abs(p1-p0)
- ; mm5 = p0
- movq mm3, [rdi] ; q1
- movq mm4, mm3 ; q1
- psubusb mm3, mm2 ; q1-=p1
- psubusb mm2, mm4 ; p1-=q1
- por mm2, mm3 ; abs(p1-q1)
- pand mm2, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psrlw mm2, 1 ; abs(p1-q1)/2
-
- movq mm6, mm5 ; p0
- movq mm3, mm0 ; q0
- psubusb mm5, mm3 ; p0-=q0
- psubusb mm3, mm6 ; q0-=p0
- por mm5, mm3 ; abs(p0 - q0)
- paddusb mm5, mm5 ; abs(p0-q0)*2
- paddusb mm5, mm2 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- mov rdx, arg(2) ;flimit ; get flimit
- movq mm2, [rdx] ; flimit mm2
- paddb mm2, mm2 ; flimit*2 (less than 255)
- paddb mm7, mm2 ; flimit * 2 + limit (less than 255)
-
- psubusb mm5, mm7 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- por mm1, mm5
- pxor mm5, mm5
- pcmpeqb mm1, mm5 ; mask mm1
-
- ; mm1 = mask, mm0=q0, mm7 = flimit, t0 = abs(q0-q1) t1 = abs(p1-p0)
- ; mm6 = p0,
-
- ; calculate high edge variance
- mov rdx, arg(4) ;thresh ; get thresh
- movq mm7, [rdx] ;
- movq mm4, t0 ; get abs (q1 - q0)
- psubusb mm4, mm7
- movq mm3, t1 ; get abs (p1 - p0)
- psubusb mm3, mm7
- paddb mm4, mm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
-
- pcmpeqb mm4, mm5
-
- pcmpeqb mm5, mm5
- pxor mm4, mm5
-
-
-
- ; mm1 = mask, mm0=q0, mm7 = thresh, t0 = abs(q0-q1) t1 = abs(p1-p0)
- ; mm6 = p0, mm4=hev
- ; start work on filters
- movq mm2, [rsi+2*rax] ; p1
- movq mm7, [rdi] ; q1
- pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
- psubsb mm2, mm7 ; p1 - q1
-
- pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
- pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
- movq mm3, mm0 ; q0
- psubsb mm0, mm6 ; q0 - p0
- paddsb mm2, mm0 ; 1 * (q0 - p0) + (p1 - q1)
- paddsb mm2, mm0 ; 2 * (q0 - p0)
- paddsb mm2, mm0 ; 3 * (q0 - p0) + (p1 - q1)
- pand mm1, mm2 ; mask filter values we don't care about
-
-
- ; mm1 = vp8_filter, mm4=hev, mm6=ps0, mm3=qs0
- movq mm2, mm1 ; vp8_filter
- pand mm2, mm4; ; Filter2 = vp8_filter & hev
-
- movq mm5, mm2 ;
- paddsb mm5, [GLOBAL(t3)];
-
- pxor mm0, mm0 ; 0
- pxor mm7, mm7 ; 0
-
- punpcklbw mm0, mm5 ; e0f0g0h0
- psraw mm0, 11 ; sign extended shift right by 3
- punpckhbw mm7, mm5 ; a0b0c0d0
- psraw mm7, 11 ; sign extended shift right by 3
- packsswb mm0, mm7 ; Filter2 >>=3;
-
- movq mm5, mm0 ; Filter2
-
- paddsb mm2, [GLOBAL(t4)] ; vp8_signed_char_clamp(Filter2 + 4)
- pxor mm0, mm0 ; 0
- pxor mm7, mm7 ; 0
-
- punpcklbw mm0, mm2 ; e0f0g0h0
- psraw mm0, 11 ; sign extended shift right by 3
- punpckhbw mm7, mm2 ; a0b0c0d0
- psraw mm7, 11 ; sign extended shift right by 3
- packsswb mm0, mm7 ; Filter2 >>=3;
-
- ; mm0= filter2 mm1 = vp8_filter, mm3 =qs0 mm5=s mm4 =hev mm6=ps0
- psubsb mm3, mm0 ; qs0 =qs0 - filter1
- paddsb mm6, mm5 ; ps0 =ps0 + Fitler2
-
- ; mm1=vp8_filter, mm3=qs0, mm4 =hev mm6=ps0
- ; vp8_filter &= ~hev;
- ; Filter2 = vp8_filter;
- pandn mm4, mm1 ; vp8_filter&=~hev
-
-
- ; mm3=qs0, mm4=filter2, mm6=ps0
-
- ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7);
- ; s = vp8_signed_char_clamp(qs0 - u);
- ; *oq0 = s^0x80;
- ; s = vp8_signed_char_clamp(ps0 + u);
- ; *op0 = s^0x80;
- pxor mm0, mm0
-
- pxor mm1, mm1
- pxor mm2, mm2
- punpcklbw mm1, mm4
- punpckhbw mm2, mm4
- pmulhw mm1, [GLOBAL(s27)]
- pmulhw mm2, [GLOBAL(s27)]
- paddw mm1, [GLOBAL(s63)]
- paddw mm2, [GLOBAL(s63)]
- psraw mm1, 7
- psraw mm2, 7
- packsswb mm1, mm2
-
- psubsb mm3, mm1
- paddsb mm6, mm1
-
- pxor mm3, [GLOBAL(t80)]
- pxor mm6, [GLOBAL(t80)]
- movq [rsi+rax], mm6
- movq [rsi], mm3
-
- ; roughly 2/7th difference across boundary
- ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7);
- ; s = vp8_signed_char_clamp(qs1 - u);
- ; *oq1 = s^0x80;
- ; s = vp8_signed_char_clamp(ps1 + u);
- ; *op1 = s^0x80;
- pxor mm1, mm1
- pxor mm2, mm2
- punpcklbw mm1, mm4
- punpckhbw mm2, mm4
- pmulhw mm1, [GLOBAL(s18)]
- pmulhw mm2, [GLOBAL(s18)]
- paddw mm1, [GLOBAL(s63)]
- paddw mm2, [GLOBAL(s63)]
- psraw mm1, 7
- psraw mm2, 7
- packsswb mm1, mm2
-
- movq mm3, [rdi]
- movq mm6, [rsi+rax*2] ; p1
-
- pxor mm3, [GLOBAL(t80)]
- pxor mm6, [GLOBAL(t80)]
-
- paddsb mm6, mm1
- psubsb mm3, mm1
-
- pxor mm6, [GLOBAL(t80)]
- pxor mm3, [GLOBAL(t80)]
- movq [rdi], mm3
- movq [rsi+rax*2], mm6
-
- ; roughly 1/7th difference across boundary
- ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7);
- ; s = vp8_signed_char_clamp(qs2 - u);
- ; *oq2 = s^0x80;
- ; s = vp8_signed_char_clamp(ps2 + u);
- ; *op2 = s^0x80;
- pxor mm1, mm1
- pxor mm2, mm2
- punpcklbw mm1, mm4
- punpckhbw mm2, mm4
- pmulhw mm1, [GLOBAL(s9)]
- pmulhw mm2, [GLOBAL(s9)]
- paddw mm1, [GLOBAL(s63)]
- paddw mm2, [GLOBAL(s63)]
- psraw mm1, 7
- psraw mm2, 7
- packsswb mm1, mm2
-
-
- movq mm6, [rdi+rax*4]
- neg rax
- movq mm3, [rdi+rax ]
-
- pxor mm6, [GLOBAL(t80)]
- pxor mm3, [GLOBAL(t80)]
-
- paddsb mm6, mm1
- psubsb mm3, mm1
-
- pxor mm6, [GLOBAL(t80)]
- pxor mm3, [GLOBAL(t80)]
- movq [rdi+rax ], mm3
- neg rax
- movq [rdi+rax*4], mm6
-
-;EARLY_BREAK_OUT:
- neg rax
- add rsi,8
- dec rcx
- jnz next8_mbh
-
- add rsp, 32
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_mbloop_filter_vertical_edge_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_mbloop_filter_vertical_edge_mmx)
-sym(vp8_mbloop_filter_vertical_edge_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 96 ; reserve 96 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
- %define srct [rsp + 32] ;__declspec(align(16)) char srct[64];
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
-
- lea rsi, [rsi + rax*4 - 4]
-
- movsxd rcx, dword ptr arg(5) ;count
-next8_mbv:
- lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
-
- ;transpose
- movq mm0, [rdi+2*rax] ; 77 76 75 74 73 72 71 70
- movq mm6, [rsi+2*rax] ; 67 66 65 64 63 62 61 60
-
- movq mm7, mm6 ; 77 76 75 74 73 72 71 70
- punpckhbw mm7, mm0 ; 77 67 76 66 75 65 74 64
-
- punpcklbw mm6, mm0 ; 73 63 72 62 71 61 70 60
- movq mm0, [rsi+rax] ; 57 56 55 54 53 52 51 50
-
- movq mm4, [rsi] ; 47 46 45 44 43 42 41 40
- movq mm5, mm4 ; 47 46 45 44 43 42 41 40
-
- punpckhbw mm5, mm0 ; 57 47 56 46 55 45 54 44
- punpcklbw mm4, mm0 ; 53 43 52 42 51 41 50 40
-
- movq mm3, mm5 ; 57 47 56 46 55 45 54 44
- punpckhwd mm5, mm7 ; 77 67 57 47 76 66 56 46
-
- punpcklwd mm3, mm7 ; 75 65 55 45 74 64 54 44
- movq mm2, mm4 ; 53 43 52 42 51 41 50 40
-
- punpckhwd mm4, mm6 ; 73 63 53 43 72 62 52 42
- punpcklwd mm2, mm6 ; 71 61 51 41 70 60 50 40
-
- neg rax
-
- movq mm7, [rsi+rax] ; 37 36 35 34 33 32 31 30
- movq mm6, [rsi+rax*2] ; 27 26 25 24 23 22 21 20
-
- movq mm1, mm6 ; 27 26 25 24 23 22 21 20
- punpckhbw mm6, mm7 ; 37 27 36 36 35 25 34 24
-
- punpcklbw mm1, mm7 ; 33 23 32 22 31 21 30 20
-
- movq mm7, [rsi+rax*4]; ; 07 06 05 04 03 02 01 00
- punpckhbw mm7, [rdi+rax*4] ; 17 07 16 06 15 05 14 04
-
- movq mm0, mm7 ; 17 07 16 06 15 05 14 04
- punpckhwd mm7, mm6 ; 37 27 17 07 36 26 16 06
-
- punpcklwd mm0, mm6 ; 35 25 15 05 34 24 14 04
- movq mm6, mm7 ; 37 27 17 07 36 26 16 06
-
- punpckhdq mm7, mm5 ; 77 67 57 47 37 27 17 07 = q3
- punpckldq mm6, mm5 ; 76 66 56 46 36 26 16 06 = q2
-
- lea rdx, srct
- movq mm5, mm6 ; 76 66 56 46 36 26 16 06
-
- movq [rdx+56], mm7
- psubusb mm5, mm7 ; q2-q3
-
-
- movq [rdx+48], mm6
- psubusb mm7, mm6 ; q3-q2
-
- por mm7, mm5; ; mm7=abs (q3-q2)
- movq mm5, mm0 ; 35 25 15 05 34 24 14 04
-
- punpckhdq mm5, mm3 ; 75 65 55 45 35 25 15 05 = q1
- punpckldq mm0, mm3 ; 74 64 54 44 34 24 15 04 = q0
-
- movq mm3, mm5 ; 75 65 55 45 35 25 15 05 = q1
- psubusb mm3, mm6 ; q1-q2
-
- psubusb mm6, mm5 ; q2-q1
- por mm6, mm3 ; mm6=abs(q2-q1)
-
- movq [rdx+40], mm5 ; save q1
- movq [rdx+32], mm0 ; save q0
-
- movq mm3, [rsi+rax*4] ; 07 06 05 04 03 02 01 00
- punpcklbw mm3, [rdi+rax*4] ; 13 03 12 02 11 01 10 00
-
- movq mm0, mm3 ; 13 03 12 02 11 01 10 00
- punpcklwd mm0, mm1 ; 31 21 11 01 30 20 10 00
-
- punpckhwd mm3, mm1 ; 33 23 13 03 32 22 12 02
- movq mm1, mm0 ; 31 21 11 01 30 20 10 00
-
- punpckldq mm0, mm2 ; 70 60 50 40 30 20 10 00 =p3
- punpckhdq mm1, mm2 ; 71 61 51 41 31 21 11 01 =p2
-
- movq [rdx], mm0 ; save p3
- movq [rdx+8], mm1 ; save p2
-
- movq mm2, mm1 ; 71 61 51 41 31 21 11 01 =p2
- psubusb mm2, mm0 ; p2-p3
-
- psubusb mm0, mm1 ; p3-p2
- por mm0, mm2 ; mm0=abs(p3-p2)
-
- movq mm2, mm3 ; 33 23 13 03 32 22 12 02
- punpckldq mm2, mm4 ; 72 62 52 42 32 22 12 02 = p1
-
- punpckhdq mm3, mm4 ; 73 63 53 43 33 23 13 03 = p0
- movq [rdx+24], mm3 ; save p0
-
- movq [rdx+16], mm2 ; save p1
- movq mm5, mm2 ; mm5 = p1
-
- psubusb mm2, mm1 ; p1-p2
- psubusb mm1, mm5 ; p2-p1
-
- por mm1, mm2 ; mm1=abs(p2-p1)
- mov rdx, arg(3) ;limit
-
- movq mm4, [rdx] ; mm4 = limit
- psubusb mm7, mm4 ; abs(q3-q2) > limit
-
- psubusb mm0, mm4 ; abs(p3-p2) > limit
- psubusb mm1, mm4 ; abs(p2-p1) > limit
-
- psubusb mm6, mm4 ; abs(q2-q1) > limit
- por mm7, mm6 ; or
-
- por mm0, mm1 ;
- por mm0, mm7 ; abs(q3-q2) > limit || abs(p3-p2) > limit ||abs(p2-p1) > limit || abs(q2-q1) > limit
-
- movq mm1, mm5 ; p1
-
- movq mm7, mm3 ; mm3=mm7=p0
- psubusb mm7, mm5 ; p0 - p1
-
- psubusb mm5, mm3 ; p1 - p0
- por mm5, mm7 ; abs(p1-p0)
-
- movq t0, mm5 ; save abs(p1-p0)
- lea rdx, srct
-
- psubusb mm5, mm4 ; mm5 = abs(p1-p0) > limit
- por mm0, mm5 ; mm0=mask
-
- movq mm5, [rdx+32] ; mm5=q0
- movq mm7, [rdx+40] ; mm7=q1
-
- movq mm6, mm5 ; mm6=q0
- movq mm2, mm7 ; q1
- psubusb mm5, mm7 ; q0-q1
-
- psubusb mm7, mm6 ; q1-q0
- por mm7, mm5 ; abs(q1-q0)
-
- movq t1, mm7 ; save abs(q1-q0)
- psubusb mm7, mm4 ; mm7=abs(q1-q0)> limit
-
- por mm0, mm7 ; mask
-
- movq mm5, mm2 ; q1
- psubusb mm5, mm1 ; q1-=p1
- psubusb mm1, mm2 ; p1-=q1
- por mm5, mm1 ; abs(p1-q1)
- pand mm5, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psrlw mm5, 1 ; abs(p1-q1)/2
-
- mov rdx, arg(2) ;flimit ;
-
- movq mm2, [rdx] ;flimit mm2
- movq mm1, mm3 ; mm1=mm3=p0
-
- movq mm7, mm6 ; mm7=mm6=q0
- psubusb mm1, mm7 ; p0-q0
-
- psubusb mm7, mm3 ; q0-p0
- por mm1, mm7 ; abs(q0-p0)
- paddusb mm1, mm1 ; abs(q0-p0)*2
- paddusb mm1, mm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- paddb mm2, mm2 ; flimit*2 (less than 255)
- paddb mm4, mm2 ; flimit * 2 + limit (less than 255)
-
- psubusb mm1, mm4 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- por mm1, mm0; ; mask
-
- pxor mm0, mm0
- pcmpeqb mm1, mm0
-
- ; calculate high edge variance
- mov rdx, arg(4) ;thresh ; get thresh
- movq mm7, [rdx]
- ;
- movq mm4, t0 ; get abs (q1 - q0)
- psubusb mm4, mm7 ; abs(q1 - q0) > thresh
-
- movq mm3, t1 ; get abs (p1 - p0)
- psubusb mm3, mm7 ; abs(p1 - p0)> thresh
-
- por mm4, mm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
- pcmpeqb mm4, mm0
-
- pcmpeqb mm0, mm0
- pxor mm4, mm0
-
-
-
-
- ; start work on filters
- lea rdx, srct
-
- ; start work on filters
- movq mm2, [rdx+16] ; p1
- movq mm7, [rdx+40] ; q1
- pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
- psubsb mm2, mm7 ; p1 - q1
-
- movq mm6, [rdx+24] ; p0
- movq mm0, [rdx+32] ; q0
- pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
- pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
-
- movq mm3, mm0 ; q0
- psubsb mm0, mm6 ; q0 - p0
- paddsb mm2, mm0 ; 1 * (q0 - p0) + (p1 - q1)
- paddsb mm2, mm0 ; 2 * (q0 - p0)
- paddsb mm2, mm0 ; 3 * (q0 - p0) + (p1 - q1)
- pand mm1, mm2 ; mask filter values we don't care about
-
- ; mm1 = vp8_filter, mm4=hev, mm6=ps0, mm3=qs0
- movq mm2, mm1 ; vp8_filter
- pand mm2, mm4; ; Filter2 = vp8_filter & hev
-
- movq mm5, mm2 ;
- paddsb mm5, [GLOBAL(t3)];
-
- pxor mm0, mm0 ; 0
- pxor mm7, mm7 ; 0
-
- punpcklbw mm0, mm5 ; e0f0g0h0
- psraw mm0, 11 ; sign extended shift right by 3
- punpckhbw mm7, mm5 ; a0b0c0d0
- psraw mm7, 11 ; sign extended shift right by 3
- packsswb mm0, mm7 ; Filter2 >>=3;
-
- movq mm5, mm0 ; Filter2
-
- paddsb mm2, [GLOBAL(t4)] ; vp8_signed_char_clamp(Filter2 + 4)
- pxor mm0, mm0 ; 0
- pxor mm7, mm7 ; 0
-
- punpcklbw mm0, mm2 ; e0f0g0h0
- psraw mm0, 11 ; sign extended shift right by 3
- punpckhbw mm7, mm2 ; a0b0c0d0
- psraw mm7, 11 ; sign extended shift right by 3
- packsswb mm0, mm7 ; Filter2 >>=3;
-
- ; mm0= filter2 mm1 = vp8_filter, mm3 =qs0 mm5=s mm4 =hev mm6=ps0
- psubsb mm3, mm0 ; qs0 =qs0 - filter1
- paddsb mm6, mm5 ; ps0 =ps0 + Fitler2
-
- ; mm1=vp8_filter, mm3=qs0, mm4 =hev mm6=ps0
- ; vp8_filter &= ~hev;
- ; Filter2 = vp8_filter;
- pandn mm4, mm1 ; vp8_filter&=~hev
-
-
- ; mm3=qs0, mm4=filter2, mm6=ps0
-
- ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7);
- ; s = vp8_signed_char_clamp(qs0 - u);
- ; *oq0 = s^0x80;
- ; s = vp8_signed_char_clamp(ps0 + u);
- ; *op0 = s^0x80;
- pxor mm0, mm0
-
- pxor mm1, mm1
- pxor mm2, mm2
- punpcklbw mm1, mm4
- punpckhbw mm2, mm4
- pmulhw mm1, [GLOBAL(s27)]
- pmulhw mm2, [GLOBAL(s27)]
- paddw mm1, [GLOBAL(s63)]
- paddw mm2, [GLOBAL(s63)]
- psraw mm1, 7
- psraw mm2, 7
- packsswb mm1, mm2
-
- psubsb mm3, mm1
- paddsb mm6, mm1
-
- pxor mm3, [GLOBAL(t80)]
- pxor mm6, [GLOBAL(t80)]
- movq [rdx+24], mm6
- movq [rdx+32], mm3
-
- ; roughly 2/7th difference across boundary
- ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7);
- ; s = vp8_signed_char_clamp(qs1 - u);
- ; *oq1 = s^0x80;
- ; s = vp8_signed_char_clamp(ps1 + u);
- ; *op1 = s^0x80;
- pxor mm1, mm1
- pxor mm2, mm2
- punpcklbw mm1, mm4
- punpckhbw mm2, mm4
- pmulhw mm1, [GLOBAL(s18)]
- pmulhw mm2, [GLOBAL(s18)]
- paddw mm1, [GLOBAL(s63)]
- paddw mm2, [GLOBAL(s63)]
- psraw mm1, 7
- psraw mm2, 7
- packsswb mm1, mm2
-
- movq mm3, [rdx + 40]
- movq mm6, [rdx + 16] ; p1
- pxor mm3, [GLOBAL(t80)]
- pxor mm6, [GLOBAL(t80)]
-
- paddsb mm6, mm1
- psubsb mm3, mm1
-
- pxor mm6, [GLOBAL(t80)]
- pxor mm3, [GLOBAL(t80)]
- movq [rdx + 40], mm3
- movq [rdx + 16], mm6
-
- ; roughly 1/7th difference across boundary
- ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7);
- ; s = vp8_signed_char_clamp(qs2 - u);
- ; *oq2 = s^0x80;
- ; s = vp8_signed_char_clamp(ps2 + u);
- ; *op2 = s^0x80;
- pxor mm1, mm1
- pxor mm2, mm2
- punpcklbw mm1, mm4
- punpckhbw mm2, mm4
- pmulhw mm1, [GLOBAL(s9)]
- pmulhw mm2, [GLOBAL(s9)]
- paddw mm1, [GLOBAL(s63)]
- paddw mm2, [GLOBAL(s63)]
- psraw mm1, 7
- psraw mm2, 7
- packsswb mm1, mm2
-
- movq mm6, [rdx+ 8]
- movq mm3, [rdx+48]
-
- pxor mm6, [GLOBAL(t80)]
- pxor mm3, [GLOBAL(t80)]
-
- paddsb mm6, mm1
- psubsb mm3, mm1
-
- pxor mm6, [GLOBAL(t80)] ; mm6 = 71 61 51 41 31 21 11 01
- pxor mm3, [GLOBAL(t80)] ; mm3 = 76 66 56 46 36 26 15 06
-
- ; tranpose and write back
- movq mm0, [rdx] ; mm0 = 70 60 50 40 30 20 10 00
- movq mm1, mm0 ; mm0 = 70 60 50 40 30 20 10 00
-
- punpcklbw mm0, mm6 ; mm0 = 31 30 21 20 11 10 01 00
- punpckhbw mm1, mm6 ; mm3 = 71 70 61 60 51 50 41 40
-
- movq mm2, [rdx+16] ; mm2 = 72 62 52 42 32 22 12 02
- movq mm6, mm2 ; mm3 = 72 62 52 42 32 22 12 02
-
- punpcklbw mm2, [rdx+24] ; mm2 = 33 32 23 22 13 12 03 02
- punpckhbw mm6, [rdx+24] ; mm3 = 73 72 63 62 53 52 43 42
-
- movq mm5, mm0 ; mm5 = 31 30 21 20 11 10 01 00
- punpcklwd mm0, mm2 ; mm0 = 13 12 11 10 03 02 01 00
-
- punpckhwd mm5, mm2 ; mm5 = 33 32 31 30 23 22 21 20
- movq mm4, mm1 ; mm4 = 71 70 61 60 51 50 41 40
-
- punpcklwd mm1, mm6 ; mm1 = 53 52 51 50 43 42 41 40
- punpckhwd mm4, mm6 ; mm4 = 73 72 71 70 63 62 61 60
-
- movq mm2, [rdx+32] ; mm2 = 74 64 54 44 34 24 14 04
- punpcklbw mm2, [rdx+40] ; mm2 = 35 34 25 24 15 14 05 04
-
- movq mm6, mm3 ; mm6 = 76 66 56 46 36 26 15 06
- punpcklbw mm6, [rdx+56] ; mm6 = 37 36 27 26 17 16 07 06
-
- movq mm7, mm2 ; mm7 = 35 34 25 24 15 14 05 04
- punpcklwd mm2, mm6 ; mm2 = 17 16 15 14 07 06 05 04
-
- punpckhwd mm7, mm6 ; mm7 = 37 36 35 34 27 26 25 24
- movq mm6, mm0 ; mm6 = 13 12 11 10 03 02 01 00
-
- punpckldq mm0, mm2 ; mm0 = 07 06 05 04 03 02 01 00
- punpckhdq mm6, mm2 ; mm6 = 17 16 15 14 13 12 11 10
-
- movq [rsi+rax*4], mm0 ; write out
- movq [rdi+rax*4], mm6 ; write out
-
- movq mm0, mm5 ; mm0 = 33 32 31 30 23 22 21 20
- punpckldq mm0, mm7 ; mm0 = 27 26 25 24 23 22 20 20
-
- punpckhdq mm5, mm7 ; mm5 = 37 36 35 34 33 32 31 30
- movq [rsi+rax*2], mm0 ; write out
-
- movq [rdi+rax*2], mm5 ; write out
- movq mm2, [rdx+32] ; mm2 = 74 64 54 44 34 24 14 04
-
- punpckhbw mm2, [rdx+40] ; mm2 = 75 74 65 64 54 54 45 44
- punpckhbw mm3, [rdx+56] ; mm3 = 77 76 67 66 57 56 47 46
-
- movq mm5, mm2 ; mm5 = 75 74 65 64 54 54 45 44
- punpcklwd mm2, mm3 ; mm2 = 57 56 55 54 47 46 45 44
-
- punpckhwd mm5, mm3 ; mm5 = 77 76 75 74 67 66 65 64
- movq mm0, mm1 ; mm0= 53 52 51 50 43 42 41 40
-
- movq mm3, mm4 ; mm4 = 73 72 71 70 63 62 61 60
- punpckldq mm0, mm2 ; mm0 = 47 46 45 44 43 42 41 40
-
- punpckhdq mm1, mm2 ; mm1 = 57 56 55 54 53 52 51 50
- movq [rsi], mm0 ; write out
-
- movq [rdi], mm1 ; write out
- neg rax
-
- punpckldq mm3, mm5 ; mm3 = 67 66 65 64 63 62 61 60
- punpckhdq mm4, mm5 ; mm4 = 77 76 75 74 73 72 71 60
-
- movq [rsi+rax*2], mm3
- movq [rdi+rax*2], mm4
-
- lea rsi, [rsi+rax*8]
- dec rcx
-
- jnz next8_mbv
-
- add rsp, 96
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_loop_filter_simple_horizontal_edge_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_simple_horizontal_edge_mmx)
-sym(vp8_loop_filter_simple_horizontal_edge_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
-
- movsxd rcx, dword ptr arg(5) ;count
-nexts8_h:
- mov rdx, arg(3) ;limit
- movq mm7, [rdx]
- mov rdx, arg(2) ;flimit ; get flimit
- movq mm3, [rdx] ;
- paddb mm3, mm3 ; flimit*2 (less than 255)
- paddb mm3, mm7 ; flimit * 2 + limit (less than 255)
-
- mov rdi, rsi ; rdi points to row +1 for indirect addressing
- add rdi, rax
- neg rax
-
- ; calculate mask
- movq mm1, [rsi+2*rax] ; p1
- movq mm0, [rdi] ; q1
- movq mm2, mm1
- movq mm7, mm0
- movq mm4, mm0
- psubusb mm0, mm1 ; q1-=p1
- psubusb mm1, mm4 ; p1-=q1
- por mm1, mm0 ; abs(p1-q1)
- pand mm1, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psrlw mm1, 1 ; abs(p1-q1)/2
-
- movq mm5, [rsi+rax] ; p0
- movq mm4, [rsi] ; q0
- movq mm0, mm4 ; q0
- movq mm6, mm5 ; p0
- psubusb mm5, mm4 ; p0-=q0
- psubusb mm4, mm6 ; q0-=p0
- por mm5, mm4 ; abs(p0 - q0)
- paddusb mm5, mm5 ; abs(p0-q0)*2
- paddusb mm5, mm1 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- psubusb mm5, mm3 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- pxor mm3, mm3
- pcmpeqb mm5, mm3
-
- ; start work on filters
- pxor mm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor mm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
- psubsb mm2, mm7 ; p1 - q1
-
- pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
- pxor mm0, [GLOBAL(t80)] ; offset to convert to signed values
- movq mm3, mm0 ; q0
- psubsb mm0, mm6 ; q0 - p0
- paddsb mm2, mm0 ; p1 - q1 + 1 * (q0 - p0)
- paddsb mm2, mm0 ; p1 - q1 + 2 * (q0 - p0)
- paddsb mm2, mm0 ; p1 - q1 + 3 * (q0 - p0)
- pand mm5, mm2 ; mask filter values we don't care about
-
- ; do + 4 side
- paddsb mm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
-
- movq mm0, mm5 ; get a copy of filters
- psllw mm0, 8 ; shift left 8
- psraw mm0, 3 ; arithmetic shift right 11
- psrlw mm0, 8
- movq mm1, mm5 ; get a copy of filters
- psraw mm1, 11 ; arithmetic shift right 11
- psllw mm1, 8 ; shift left 8 to put it back
-
- por mm0, mm1 ; put the two together to get result
-
- psubsb mm3, mm0 ; q0-= q0 add
- pxor mm3, [GLOBAL(t80)] ; unoffset
- movq [rsi], mm3 ; write back
-
-
- ; now do +3 side
- psubsb mm5, [GLOBAL(t1s)] ; +3 instead of +4
-
- movq mm0, mm5 ; get a copy of filters
- psllw mm0, 8 ; shift left 8
- psraw mm0, 3 ; arithmetic shift right 11
- psrlw mm0, 8
- psraw mm5, 11 ; arithmetic shift right 11
- psllw mm5, 8 ; shift left 8 to put it back
- por mm0, mm5 ; put the two together to get result
-
-
- paddsb mm6, mm0 ; p0+= p0 add
- pxor mm6, [GLOBAL(t80)] ; unoffset
- movq [rsi+rax], mm6 ; write back
-
- add rsi,8
- neg rax
- dec rcx
- jnz nexts8_h
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_loop_filter_simple_vertical_edge_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_simple_vertical_edge_mmx)
-sym(vp8_loop_filter_simple_vertical_edge_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 32 ; reserve 32 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8];
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
-
- lea rsi, [rsi + rax*4- 2]; ;
- movsxd rcx, dword ptr arg(5) ;count
-nexts8_v:
-
- lea rdi, [rsi + rax];
- movd mm0, [rdi + rax * 2] ; xx xx xx xx 73 72 71 70
-
- movd mm6, [rsi + rax * 2] ; xx xx xx xx 63 62 61 60
- punpcklbw mm6, mm0 ; 73 63 72 62 71 61 70 60
-
- movd mm0, [rsi + rax] ; xx xx xx xx 53 52 51 50
- movd mm4, [rsi] ; xx xx xx xx 43 42 41 40
-
- punpcklbw mm4, mm0 ; 53 43 52 42 51 41 50 40
- movq mm5, mm4 ; 53 43 52 42 51 41 50 40
-
- punpcklwd mm4, mm6 ; 71 61 51 41 70 60 50 40
- punpckhwd mm5, mm6 ; 73 63 53 43 72 62 52 42
-
- neg rax
-
- movd mm7, [rsi + rax] ; xx xx xx xx 33 32 31 30
- movd mm6, [rsi + rax * 2] ; xx xx xx xx 23 22 21 20
-
- punpcklbw mm6, mm7 ; 33 23 32 22 31 21 30 20
- movd mm1, [rdi + rax * 4] ; xx xx xx xx 13 12 11 10
-
- movd mm0, [rsi + rax * 4] ; xx xx xx xx 03 02 01 00
- punpcklbw mm0, mm1 ; 13 03 12 02 11 01 10 00
-
- movq mm2, mm0 ; 13 03 12 02 11 01 10 00
- punpcklwd mm0, mm6 ; 31 21 11 01 30 20 10 00
-
- punpckhwd mm2, mm6 ; 33 23 13 03 32 22 12 02
- movq mm1, mm0 ; 13 03 12 02 11 01 10 00
-
- punpckldq mm0, mm4 ; 70 60 50 40 30 20 10 00 = p1
- movq mm3, mm2 ; 33 23 13 03 32 22 12 02
-
- punpckhdq mm1, mm4 ; 71 61 51 41 31 21 11 01 = p0
- punpckldq mm2, mm5 ; 72 62 52 42 32 22 12 02 = q0
-
- punpckhdq mm3, mm5 ; 73 63 53 43 33 23 13 03 = q1
-
-
- ; calculate mask
- movq mm6, mm0 ; p1
- movq mm7, mm3 ; q1
- psubusb mm7, mm6 ; q1-=p1
- psubusb mm6, mm3 ; p1-=q1
- por mm6, mm7 ; abs(p1-q1)
- pand mm6, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psrlw mm6, 1 ; abs(p1-q1)/2
-
- movq mm5, mm1 ; p0
- movq mm4, mm2 ; q0
-
- psubusb mm5, mm2 ; p0-=q0
- psubusb mm4, mm1 ; q0-=p0
-
- por mm5, mm4 ; abs(p0 - q0)
- paddusb mm5, mm5 ; abs(p0-q0)*2
- paddusb mm5, mm6 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- mov rdx, arg(2) ;flimit ; get flimit
- movq mm7, [rdx]
- mov rdx, arg(3) ; get limit
- movq mm6, [rdx]
- paddb mm7, mm7 ; flimit*2 (less than 255)
- paddb mm7, mm6 ; flimit * 2 + limit (less than 255)
-
- psubusb mm5, mm7 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- pxor mm7, mm7
- pcmpeqb mm5, mm7 ; mm5 = mask
-
- ; start work on filters
- movq t0, mm0
- movq t1, mm3
-
- pxor mm0, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor mm3, [GLOBAL(t80)] ; q1 offset to convert to signed values
-
- psubsb mm0, mm3 ; p1 - q1
- movq mm6, mm1 ; p0
-
- movq mm7, mm2 ; q0
- pxor mm6, [GLOBAL(t80)] ; offset to convert to signed values
-
- pxor mm7, [GLOBAL(t80)] ; offset to convert to signed values
- movq mm3, mm7 ; offseted ; q0
-
- psubsb mm7, mm6 ; q0 - p0
- paddsb mm0, mm7 ; p1 - q1 + 1 * (q0 - p0)
-
- paddsb mm0, mm7 ; p1 - q1 + 2 * (q0 - p0)
- paddsb mm0, mm7 ; p1 - q1 + 3 * (q0 - p0)
-
- pand mm5, mm0 ; mask filter values we don't care about
-
- paddsb mm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
-
- movq mm0, mm5 ; get a copy of filters
- psllw mm0, 8 ; shift left 8
- psraw mm0, 3 ; arithmetic shift right 11
- psrlw mm0, 8
-
- movq mm7, mm5 ; get a copy of filters
- psraw mm7, 11 ; arithmetic shift right 11
- psllw mm7, 8 ; shift left 8 to put it back
-
- por mm0, mm7 ; put the two together to get result
-
- psubsb mm3, mm0 ; q0-= q0sz add
- pxor mm3, [GLOBAL(t80)] ; unoffset
-
- ; now do +3 side
- psubsb mm5, [GLOBAL(t1s)] ; +3 instead of +4
-
- movq mm0, mm5 ; get a copy of filters
- psllw mm0, 8 ; shift left 8
- psraw mm0, 3 ; arithmetic shift right 11
- psrlw mm0, 8
-
- psraw mm5, 11 ; arithmetic shift right 11
- psllw mm5, 8 ; shift left 8 to put it back
- por mm0, mm5 ; put the two together to get result
-
- paddsb mm6, mm0 ; p0+= p0 add
- pxor mm6, [GLOBAL(t80)] ; unoffset
-
-
- movq mm0, t0
- movq mm4, t1
-
- ; mm0 = 70 60 50 40 30 20 10 00
- ; mm6 = 71 61 51 41 31 21 11 01
- ; mm3 = 72 62 52 42 32 22 12 02
- ; mm4 = 73 63 53 43 33 23 13 03
- ; transpose back to write out
-
- movq mm1, mm0 ;
- punpcklbw mm0, mm6 ; 31 30 21 20 11 10 01 00
-
- punpckhbw mm1, mm6 ; 71 70 61 60 51 50 41 40
- movq mm2, mm3 ;
-
- punpcklbw mm2, mm4 ; 33 32 23 22 13 12 03 02
- movq mm5, mm1 ; 71 70 61 60 51 50 41 40
-
- punpckhbw mm3, mm4 ; 73 72 63 62 53 52 43 42
- movq mm6, mm0 ; 31 30 21 20 11 10 01 00
-
- punpcklwd mm0, mm2 ; 13 12 11 10 03 02 01 00
- punpckhwd mm6, mm2 ; 33 32 31 30 23 22 21 20
-
- movd [rsi+rax*4], mm0 ; write 03 02 01 00
- punpcklwd mm1, mm3 ; 53 52 51 50 43 42 41 40
-
- psrlq mm0, 32 ; xx xx xx xx 13 12 11 10
- punpckhwd mm5, mm3 ; 73 72 71 70 63 62 61 60
-
- movd [rdi+rax*4], mm0 ; write 13 12 11 10
- movd [rsi+rax*2], mm6 ; write 23 22 21 20
-
- psrlq mm6, 32 ; 33 32 31 30
- movd [rsi], mm1 ; write 43 42 41 40
-
- movd [rsi + rax], mm6 ; write 33 32 31 30
- neg rax
-
- movd [rsi + rax*2], mm5 ; write 63 62 61 60
- psrlq mm1, 32 ; 53 52 51 50
-
- movd [rdi], mm1 ; write out 53 52 51 50
- psrlq mm5, 32 ; 73 72 71 70
-
- movd [rdi + rax*2], mm5 ; write 73 72 71 70
-
- lea rsi, [rsi+rax*8] ; next 8
-
- dec rcx
- jnz nexts8_v
-
- add rsp, 32
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-;void fast_loop_filter_vertical_edges_mmx(unsigned char *y_ptr,
-; int y_stride,
-; loop_filter_info *lfi)
-;{
-;
-;
-; vp8_loop_filter_simple_vertical_edge_mmx(y_ptr+4, y_stride, lfi->flim,lfi->lim,lfi->thr,2);
-; vp8_loop_filter_simple_vertical_edge_mmx(y_ptr+8, y_stride, lfi->flim,lfi->lim,lfi->thr,2);
-; vp8_loop_filter_simple_vertical_edge_mmx(y_ptr+12, y_stride, lfi->flim,lfi->lim,lfi->thr,2);
-;}
-
-SECTION_RODATA
-align 16
-tfe:
- times 8 db 0xfe
-align 16
-t80:
- times 8 db 0x80
-align 16
-t1s:
- times 8 db 0x01
-align 16
-t3:
- times 8 db 0x03
-align 16
-t4:
- times 8 db 0x04
-align 16
-ones:
- times 4 dw 0x0001
-align 16
-s27:
- times 4 dw 0x1b00
-align 16
-s18:
- times 4 dw 0x1200
-align 16
-s9:
- times 4 dw 0x0900
-align 16
-s63:
- times 4 dw 0x003f
diff --git a/vp8/common/x86/loopfilter_sse2.asm b/vp8/common/x86/loopfilter_sse2.asm
deleted file mode 100644
index 849133d..0000000
--- a/vp8/common/x86/loopfilter_sse2.asm
+++ /dev/null
@@ -1,1780 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-; Use of pmaxub instead of psubusb to compute filter mask was seen
-; in ffvp8
-
-%macro LFH_FILTER_AND_HEV_MASK 1
-%if %1
- movdqa xmm2, [rdi+2*rax] ; q3
- movdqa xmm1, [rsi+2*rax] ; q2
- movdqa xmm4, [rsi+rax] ; q1
- movdqa xmm5, [rsi] ; q0
- neg rax ; negate pitch to deal with above border
-%else
- movlps xmm2, [rsi + rcx*2] ; q3
- movlps xmm1, [rsi + rcx] ; q2
- movlps xmm4, [rsi] ; q1
- movlps xmm5, [rsi + rax] ; q0
-
- movhps xmm2, [rdi + rcx*2]
- movhps xmm1, [rdi + rcx]
- movhps xmm4, [rdi]
- movhps xmm5, [rdi + rax]
-
- lea rsi, [rsi + rax*4]
- lea rdi, [rdi + rax*4]
-
- movdqa XMMWORD PTR [rsp], xmm1 ; store q2
- movdqa XMMWORD PTR [rsp + 16], xmm4 ; store q1
-%endif
-
- movdqa xmm6, xmm1 ; q2
- movdqa xmm3, xmm4 ; q1
-
- psubusb xmm1, xmm2 ; q2-=q3
- psubusb xmm2, xmm6 ; q3-=q2
-
- psubusb xmm4, xmm6 ; q1-=q2
- psubusb xmm6, xmm3 ; q2-=q1
-
- por xmm4, xmm6 ; abs(q2-q1)
- por xmm1, xmm2 ; abs(q3-q2)
-
- movdqa xmm0, xmm5 ; q0
- pmaxub xmm1, xmm4
-
- psubusb xmm5, xmm3 ; q0-=q1
- psubusb xmm3, xmm0 ; q1-=q0
-
- por xmm5, xmm3 ; abs(q0-q1)
- movdqa t0, xmm5 ; save to t0
-
- pmaxub xmm1, xmm5
-
-%if %1
- movdqa xmm2, [rsi+4*rax] ; p3
- movdqa xmm4, [rdi+4*rax] ; p2
- movdqa xmm6, [rsi+2*rax] ; p1
-%else
- movlps xmm2, [rsi + rax] ; p3
- movlps xmm4, [rsi] ; p2
- movlps xmm6, [rsi + rcx] ; p1
-
- movhps xmm2, [rdi + rax]
- movhps xmm4, [rdi]
- movhps xmm6, [rdi + rcx]
-
- movdqa XMMWORD PTR [rsp + 32], xmm4 ; store p2
- movdqa XMMWORD PTR [rsp + 48], xmm6 ; store p1
-%endif
-
- movdqa xmm5, xmm4 ; p2
- movdqa xmm3, xmm6 ; p1
-
- psubusb xmm4, xmm2 ; p2-=p3
- psubusb xmm2, xmm5 ; p3-=p2
-
- psubusb xmm3, xmm5 ; p1-=p2
- pmaxub xmm1, xmm4 ; abs(p3 - p2)
-
- psubusb xmm5, xmm6 ; p2-=p1
- pmaxub xmm1, xmm2 ; abs(p3 - p2)
-
- pmaxub xmm1, xmm5 ; abs(p2 - p1)
- movdqa xmm2, xmm6 ; p1
-
- pmaxub xmm1, xmm3 ; abs(p2 - p1)
-%if %1
- movdqa xmm4, [rsi+rax] ; p0
- movdqa xmm3, [rdi] ; q1
-%else
- movlps xmm4, [rsi + rcx*2] ; p0
- movhps xmm4, [rdi + rcx*2]
- movdqa xmm3, q1 ; q1
-%endif
-
- movdqa xmm5, xmm4 ; p0
- psubusb xmm4, xmm6 ; p0-=p1
-
- psubusb xmm6, xmm5 ; p1-=p0
-
- por xmm6, xmm4 ; abs(p1 - p0)
- mov rdx, arg(2) ; get flimit
-
- movdqa t1, xmm6 ; save to t1
-
- movdqa xmm4, xmm3 ; q1
- pmaxub xmm1, xmm6
-
- psubusb xmm3, xmm2 ; q1-=p1
- psubusb xmm2, xmm4 ; p1-=q1
-
- psubusb xmm1, xmm7
- por xmm2, xmm3 ; abs(p1-q1)
-
- movdqa xmm4, XMMWORD PTR [rdx] ; flimit
-
- movdqa xmm3, xmm0 ; q0
- pand xmm2, [GLOBAL(tfe)] ; set lsb of each byte to zero
-
- mov rdx, arg(4) ; hev get thresh
-
- movdqa xmm6, xmm5 ; p0
- psrlw xmm2, 1 ; abs(p1-q1)/2
-
- psubusb xmm5, xmm3 ; p0-=q0
- paddb xmm4, xmm4 ; flimit*2 (less than 255)
-
- psubusb xmm3, xmm6 ; q0-=p0
- por xmm5, xmm3 ; abs(p0 - q0)
-
- paddusb xmm5, xmm5 ; abs(p0-q0)*2
- paddb xmm7, xmm4 ; flimit * 2 + limit (less than 255)
-
- movdqa xmm4, t0 ; hev get abs (q1 - q0)
-
- movdqa xmm3, t1 ; get abs (p1 - p0)
-
- paddusb xmm5, xmm2 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- movdqa xmm2, XMMWORD PTR [rdx] ; hev
-
- psubusb xmm5, xmm7 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- psubusb xmm4, xmm2 ; hev
-
- psubusb xmm3, xmm2 ; hev
- por xmm1, xmm5
-
- pxor xmm7, xmm7
- paddb xmm4, xmm3 ; hev abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
-
- pcmpeqb xmm4, xmm5 ; hev
- pcmpeqb xmm3, xmm3 ; hev
-
- pcmpeqb xmm1, xmm7 ; mask xmm1
- pxor xmm4, xmm3 ; hev
-%endmacro
-
-%macro B_FILTER 1
-%if %1 == 0
- movdqa xmm2, p1 ; p1
- movdqa xmm7, q1 ; q1
-%elif %1 == 1
- movdqa xmm2, [rsi+2*rax] ; p1
- movdqa xmm7, [rdi] ; q1
-%elif %1 == 2
- lea rdx, srct
-
- movdqa xmm2, [rdx] ; p1
- movdqa xmm7, [rdx+48] ; q1
- movdqa xmm6, [rdx+16] ; p0
- movdqa xmm0, [rdx+32] ; q0
-%endif
-
- pxor xmm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor xmm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
-
- psubsb xmm2, xmm7 ; p1 - q1
- pxor xmm6, [GLOBAL(t80)] ; offset to convert to signed values
-
- pand xmm2, xmm4 ; high var mask (hvm)(p1 - q1)
- pxor xmm0, [GLOBAL(t80)] ; offset to convert to signed values
-
- movdqa xmm3, xmm0 ; q0
- psubsb xmm0, xmm6 ; q0 - p0
-
- paddsb xmm2, xmm0 ; 1 * (q0 - p0) + hvm(p1 - q1)
-
- paddsb xmm2, xmm0 ; 2 * (q0 - p0) + hvm(p1 - q1)
-
- paddsb xmm2, xmm0 ; 3 * (q0 - p0) + hvm(p1 - q1)
-
- pand xmm1, xmm2 ; mask filter values we don't care about
-
- movdqa xmm2, xmm1
-
- paddsb xmm1, [GLOBAL(t4)] ; 3* (q0 - p0) + hvm(p1 - q1) + 4
- paddsb xmm2, [GLOBAL(t3)] ; 3* (q0 - p0) + hvm(p1 - q1) + 3
-
- punpckhbw xmm5, xmm2 ; axbxcxdx
- punpcklbw xmm2, xmm2 ; exfxgxhx
-
- punpcklbw xmm0, xmm1 ; exfxgxhx
- psraw xmm5, 11 ; sign extended shift right by 3
-
- punpckhbw xmm1, xmm1 ; axbxcxdx
- psraw xmm2, 11 ; sign extended shift right by 3
-
- packsswb xmm2, xmm5 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3;
- psraw xmm0, 11 ; sign extended shift right by 3
-
- psraw xmm1, 11 ; sign extended shift right by 3
- movdqa xmm5, xmm0 ; save results
-
- packsswb xmm0, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3
- paddsw xmm5, [GLOBAL(ones)]
-
- paddsw xmm1, [GLOBAL(ones)]
- psraw xmm5, 1 ; partial shifted one more time for 2nd tap
-
- psraw xmm1, 1 ; partial shifted one more time for 2nd tap
-
- paddsb xmm6, xmm2 ; p0+= p0 add
- packsswb xmm5, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4
-
-%if %1 == 0
- movdqa xmm1, p1 ; p1
-%elif %1 == 1
- movdqa xmm1, [rsi+2*rax] ; p1
-%elif %1 == 2
- movdqa xmm1, [rdx] ; p1
-%endif
- pandn xmm4, xmm5 ; high edge variance additive
- pxor xmm6, [GLOBAL(t80)] ; unoffset
-
- pxor xmm1, [GLOBAL(t80)] ; reoffset
- psubsb xmm3, xmm0 ; q0-= q0 add
-
- paddsb xmm1, xmm4 ; p1+= p1 add
- pxor xmm3, [GLOBAL(t80)] ; unoffset
-
- pxor xmm1, [GLOBAL(t80)] ; unoffset
- psubsb xmm7, xmm4 ; q1-= q1 add
-
- pxor xmm7, [GLOBAL(t80)] ; unoffset
-%if %1 == 0
- lea rsi, [rsi + rcx*2]
- lea rdi, [rdi + rcx*2]
- movq MMWORD PTR [rsi], xmm6 ; p0
- movhps MMWORD PTR [rdi], xmm6
- movq MMWORD PTR [rsi + rax], xmm1 ; p1
- movhps MMWORD PTR [rdi + rax], xmm1
- movq MMWORD PTR [rsi + rcx], xmm3 ; q0
- movhps MMWORD PTR [rdi + rcx], xmm3
- movq MMWORD PTR [rsi + rcx*2],xmm7 ; q1
- movhps MMWORD PTR [rdi + rcx*2],xmm7
-%elif %1 == 1
- movdqa [rsi+rax], xmm6 ; write back
- movdqa [rsi+2*rax], xmm1 ; write back
- movdqa [rsi], xmm3 ; write back
- movdqa [rdi], xmm7 ; write back
-%endif
-
-%endmacro
-
-
-;void vp8_loop_filter_horizontal_edge_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_horizontal_edge_sse2)
-sym(vp8_loop_filter_horizontal_edge_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 32 ; reserve 32 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16];
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step
-
- mov rdx, arg(3) ;limit
- movdqa xmm7, XMMWORD PTR [rdx]
-
- lea rdi, [rsi+rax] ; rdi points to row +1 for indirect addressing
-
- ; calculate breakout conditions and high edge variance
- LFH_FILTER_AND_HEV_MASK 1
- ; filter and write back the result
- B_FILTER 1
-
- add rsp, 32
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_loop_filter_horizontal_edge_uv_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_horizontal_edge_uv_sse2)
-sym(vp8_loop_filter_horizontal_edge_uv_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 96 ; reserve 96 bytes
- %define q2 [rsp + 0] ;__declspec(align(16)) char q2[16];
- %define q1 [rsp + 16] ;__declspec(align(16)) char q1[16];
- %define p2 [rsp + 32] ;__declspec(align(16)) char p2[16];
- %define p1 [rsp + 48] ;__declspec(align(16)) char p1[16];
- %define t0 [rsp + 64] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 80] ;__declspec(align(16)) char t1[16];
-
- mov rsi, arg(0) ; u
- mov rdi, arg(5) ; v
- movsxd rax, dword ptr arg(1) ; src_pixel_step
- mov rcx, rax
- neg rax ; negate pitch to deal with above border
-
- mov rdx, arg(3) ;limit
- movdqa xmm7, XMMWORD PTR [rdx]
-
- lea rsi, [rsi + rcx]
- lea rdi, [rdi + rcx]
-
- ; calculate breakout conditions and high edge variance
- LFH_FILTER_AND_HEV_MASK 0
- ; filter and write back the result
- B_FILTER 0
-
- add rsp, 96
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-%macro MB_FILTER_AND_WRITEBACK 1
-%if %1 == 0
- movdqa xmm2, p1 ; p1
- movdqa xmm7, q1 ; q1
-%elif %1 == 1
- movdqa xmm2, [rsi+2*rax] ; p1
- movdqa xmm7, [rdi] ; q1
-
- mov rcx, rax
- neg rcx
-%elif %1 == 2
- lea rdx, srct
-
- movdqa xmm2, [rdx+32] ; p1
- movdqa xmm7, [rdx+80] ; q1
- movdqa xmm6, [rdx+48] ; p0
- movdqa xmm0, [rdx+64] ; q0
-%endif
-
- pxor xmm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor xmm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
- pxor xmm6, [GLOBAL(t80)] ; offset to convert to signed values
- pxor xmm0, [GLOBAL(t80)] ; offset to convert to signed values
-
- psubsb xmm2, xmm7 ; p1 - q1
- movdqa xmm3, xmm0 ; q0
-
- psubsb xmm0, xmm6 ; q0 - p0
-
- paddsb xmm2, xmm0 ; 1 * (q0 - p0) + (p1 - q1)
-
- paddsb xmm2, xmm0 ; 2 * (q0 - p0)
-
- paddsb xmm2, xmm0 ; 3 * (q0 - p0) + (p1 - q1)
-
- pand xmm1, xmm2 ; mask filter values we don't care about
-
- movdqa xmm2, xmm1 ; vp8_filter
-
- pand xmm2, xmm4 ; Filter2 = vp8_filter & hev
- pxor xmm0, xmm0
-
- pandn xmm4, xmm1 ; vp8_filter&=~hev
- pxor xmm1, xmm1
-
- punpcklbw xmm0, xmm4 ; Filter 2 (hi)
- movdqa xmm5, xmm2
-
- punpckhbw xmm1, xmm4 ; Filter 2 (lo)
- paddsb xmm5, [GLOBAL(t3)] ; vp8_signed_char_clamp(Filter2 + 3)
-
- pmulhw xmm1, [GLOBAL(s9)] ; Filter 2 (lo) * 9
-
- pmulhw xmm0, [GLOBAL(s9)] ; Filter 2 (hi) * 9
-
- punpckhbw xmm7, xmm5 ; axbxcxdx
- paddsb xmm2, [GLOBAL(t4)] ; vp8_signed_char_clamp(Filter2 + 4)
-
- punpcklbw xmm5, xmm5 ; exfxgxhx
- psraw xmm7, 11 ; sign extended shift right by 3
-
- psraw xmm5, 11 ; sign extended shift right by 3
- punpckhbw xmm4, xmm2 ; axbxcxdx
-
- punpcklbw xmm2, xmm2 ; exfxgxhx
- psraw xmm4, 11 ; sign extended shift right by 3
-
- packsswb xmm5, xmm7 ; Filter2 >>=3;
- psraw xmm2, 11 ; sign extended shift right by 3
-
- packsswb xmm2, xmm4 ; Filter1 >>=3;
- movdqa xmm7, xmm1
-
- paddsb xmm6, xmm5 ; ps0 =ps0 + Fitler2
- movdqa xmm4, xmm1
-
- psubsb xmm3, xmm2 ; qs0 =qs0 - Filter1
- movdqa xmm5, xmm0
-
- movdqa xmm2, xmm5
- paddw xmm0, [GLOBAL(s63)] ; Filter 2 (hi) * 9 + 63
-
- paddw xmm1, [GLOBAL(s63)] ; Filter 2 (lo) * 9 + 63
- paddw xmm5, xmm5 ; Filter 2 (hi) * 18
-
- paddw xmm7, xmm7 ; Filter 2 (lo) * 18
- paddw xmm5, xmm0 ; Filter 2 (hi) * 27 + 63
-
- paddw xmm7, xmm1 ; Filter 2 (lo) * 27 + 63
- paddw xmm2, xmm0 ; Filter 2 (hi) * 18 + 63
-
- paddw xmm4, xmm1 ; Filter 2 (lo) * 18 + 63
- psraw xmm0, 7 ; (Filter 2 (hi) * 9 + 63) >> 7
-
- psraw xmm1, 7 ; (Filter 2 (lo) * 9 + 63) >> 7
- psraw xmm2, 7 ; (Filter 2 (hi) * 18 + 63) >> 7
-
- packsswb xmm0, xmm1 ; u1 = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
- psraw xmm4, 7 ; (Filter 2 (lo) * 18 + 63) >> 7
-
- psraw xmm5, 7 ; (Filter 2 (hi) * 27 + 63) >> 7
- packsswb xmm2, xmm4 ; u2 = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
-
- psraw xmm7, 7 ; (Filter 2 (lo) * 27 + 63) >> 7
-
- packsswb xmm5, xmm7 ; u3 = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
-
- psubsb xmm3, xmm5 ; sq = vp8_signed_char_clamp(qs0 - u3)
- paddsb xmm6, xmm5 ; sp = vp8_signed_char_clamp(ps0 - u3)
-
-%if %1 == 0
- movdqa xmm5, q2 ; q2
- movdqa xmm1, q1 ; q1
- movdqa xmm4, p1 ; p1
- movdqa xmm7, p2 ; p2
-
-%elif %1 == 1
- movdqa xmm5, XMMWORD PTR [rdi+rcx] ; q2
- movdqa xmm1, XMMWORD PTR [rdi] ; q1
- movdqa xmm4, XMMWORD PTR [rsi+rax*2] ; p1
- movdqa xmm7, XMMWORD PTR [rdi+rax*4] ; p2
-%elif %1 == 2
- movdqa xmm5, XMMWORD PTR [rdx+96] ; q2
- movdqa xmm1, XMMWORD PTR [rdx+80] ; q1
- movdqa xmm4, XMMWORD PTR [rdx+32] ; p1
- movdqa xmm7, XMMWORD PTR [rdx+16] ; p2
-%endif
-
- pxor xmm3, [GLOBAL(t80)] ; *oq0 = sq^0x80
- pxor xmm6, [GLOBAL(t80)] ; *oq0 = sp^0x80
-
- pxor xmm1, [GLOBAL(t80)]
- pxor xmm4, [GLOBAL(t80)]
-
- psubsb xmm1, xmm2 ; sq = vp8_signed_char_clamp(qs1 - u2)
- paddsb xmm4, xmm2 ; sp = vp8_signed_char_clamp(ps1 - u2)
-
- pxor xmm1, [GLOBAL(t80)] ; *oq1 = sq^0x80;
- pxor xmm4, [GLOBAL(t80)] ; *op1 = sp^0x80;
-
- pxor xmm7, [GLOBAL(t80)]
- pxor xmm5, [GLOBAL(t80)]
-
- paddsb xmm7, xmm0 ; sp = vp8_signed_char_clamp(ps2 - u)
- psubsb xmm5, xmm0 ; sq = vp8_signed_char_clamp(qs2 - u)
-
- pxor xmm7, [GLOBAL(t80)] ; *op2 = sp^0x80;
- pxor xmm5, [GLOBAL(t80)] ; *oq2 = sq^0x80;
-
-%if %1 == 0
- lea rsi, [rsi+rcx*2]
- lea rdi, [rdi+rcx*2]
-
- movq MMWORD PTR [rsi], xmm6 ; p0
- movhps MMWORD PTR [rdi], xmm6
- movq MMWORD PTR [rsi + rcx], xmm3 ; q0
- movhps MMWORD PTR [rdi + rcx], xmm3
-
- movq MMWORD PTR [rsi+rcx*2], xmm1 ; q1
- movhps MMWORD PTR [rdi+rcx*2], xmm1
-
- movq MMWORD PTR [rsi + rax], xmm4 ; p1
- movhps MMWORD PTR [rdi + rax], xmm4
-
- movq MMWORD PTR [rsi+rax*2], xmm7 ; p2
- movhps MMWORD PTR [rdi+rax*2], xmm7
-
- lea rsi, [rsi + rcx]
- lea rdi, [rdi + rcx]
- movq MMWORD PTR [rsi+rcx*2], xmm5 ; q2
- movhps MMWORD PTR [rdi+rcx*2], xmm5
-%elif %1 == 1
- movdqa XMMWORD PTR [rdi+rcx], xmm5 ; q2
- movdqa XMMWORD PTR [rdi], xmm1 ; q1
- movdqa XMMWORD PTR [rsi], xmm3 ; q0
- movdqa XMMWORD PTR [rsi+rax ],xmm6 ; p0
- movdqa XMMWORD PTR [rsi+rax*2],xmm4 ; p1
- movdqa XMMWORD PTR [rdi+rax*4],xmm7 ; p2
-%elif %1 == 2
- movdqa XMMWORD PTR [rdx+80], xmm1 ; q1
- movdqa XMMWORD PTR [rdx+64], xmm3 ; q0
- movdqa XMMWORD PTR [rdx+48], xmm6 ; p0
- movdqa XMMWORD PTR [rdx+32], xmm4 ; p1
-%endif
-
-%endmacro
-
-
-;void vp8_mbloop_filter_horizontal_edge_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_mbloop_filter_horizontal_edge_sse2)
-sym(vp8_mbloop_filter_horizontal_edge_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 32 ; reserve 32 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16];
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step
-
- mov rdx, arg(3) ;limit
- movdqa xmm7, XMMWORD PTR [rdx]
-
- lea rdi, [rsi+rax] ; rdi points to row +1 for indirect addressing
-
- ; calculate breakout conditions and high edge variance
- LFH_FILTER_AND_HEV_MASK 1
- ; filter and write back the results
- MB_FILTER_AND_WRITEBACK 1
-
- add rsp, 32
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_mbloop_filter_horizontal_edge_uv_sse2
-;(
-; unsigned char *u,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; unsigned char *v
-;)
-global sym(vp8_mbloop_filter_horizontal_edge_uv_sse2)
-sym(vp8_mbloop_filter_horizontal_edge_uv_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 96 ; reserve 96 bytes
- %define q2 [rsp + 0] ;__declspec(align(16)) char q2[16];
- %define q1 [rsp + 16] ;__declspec(align(16)) char q1[16];
- %define p2 [rsp + 32] ;__declspec(align(16)) char p2[16];
- %define p1 [rsp + 48] ;__declspec(align(16)) char p1[16];
- %define t0 [rsp + 64] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 80] ;__declspec(align(16)) char t1[16];
-
- mov rsi, arg(0) ; u
- mov rdi, arg(5) ; v
- movsxd rax, dword ptr arg(1) ; src_pixel_step
- mov rcx, rax
- neg rax ; negate pitch to deal with above border
-
- mov rdx, arg(3) ;limit
- movdqa xmm7, XMMWORD PTR [rdx]
-
- lea rsi, [rsi + rcx]
- lea rdi, [rdi + rcx]
-
- ; calculate breakout conditions and high edge variance
- LFH_FILTER_AND_HEV_MASK 0
- ; filter and write back the results
- MB_FILTER_AND_WRITEBACK 0
-
- add rsp, 96
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-%macro TRANSPOSE_16X8 2
- movq xmm4, QWORD PTR [rsi] ; xx xx xx xx xx xx xx xx 07 06 05 04 03 02 01 00
- movq xmm1, QWORD PTR [rdi] ; xx xx xx xx xx xx xx xx 17 16 15 14 13 12 11 10
- movq xmm0, QWORD PTR [rsi+2*rax] ; xx xx xx xx xx xx xx xx 27 26 25 24 23 22 21 20
- movq xmm7, QWORD PTR [rdi+2*rax] ; xx xx xx xx xx xx xx xx 37 36 35 34 33 32 31 30
- movq xmm5, QWORD PTR [rsi+4*rax] ; xx xx xx xx xx xx xx xx 47 46 45 44 43 42 41 40
- movq xmm2, QWORD PTR [rdi+4*rax] ; xx xx xx xx xx xx xx xx 57 56 55 54 53 52 51 50
-
- punpcklbw xmm4, xmm1 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00
-
- movq xmm1, QWORD PTR [rdi+2*rcx] ; xx xx xx xx xx xx xx xx 77 76 75 74 73 72 71 70
-
- movdqa xmm3, xmm4 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00
- punpcklbw xmm0, xmm7 ; 37 27 36 36 35 25 34 24 33 23 32 22 31 21 30 20
-
- movq xmm7, QWORD PTR [rsi+2*rcx] ; xx xx xx xx xx xx xx xx 67 66 65 64 63 62 61 60
-
- punpcklbw xmm5, xmm2 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40
-%if %1
- lea rsi, [rsi+rax*8]
-%else
- mov rsi, arg(5) ; v_ptr
-%endif
-
- movdqa xmm6, xmm5 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40
- punpcklbw xmm7, xmm1 ; 77 67 76 66 75 65 74 64 73 63 72 62 71 61 70 60
-
- punpcklwd xmm5, xmm7 ; 73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40
-
- punpckhwd xmm6, xmm7 ; 77 67 57 47 76 66 56 46 75 65 55 45 74 64 54 44
-%if %1
- lea rdi, [rdi+rax*8]
-%else
- lea rsi, [rsi - 4]
-%endif
-
- punpcklwd xmm3, xmm0 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00
-%if %1
- lea rdx, srct
-%else
- lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
-%endif
-
- movdqa xmm2, xmm3 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00
- punpckhwd xmm4, xmm0 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04
-
- movdqa xmm7, xmm4 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04
- punpckhdq xmm3, xmm5 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
-
- punpckhdq xmm7, xmm6 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06
-
- punpckldq xmm4, xmm6 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04
-
- punpckldq xmm2, xmm5 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
-
- movdqa t0, xmm2 ; save to free XMM2
- movq xmm2, QWORD PTR [rsi] ; xx xx xx xx xx xx xx xx 87 86 85 84 83 82 81 80
- movq xmm6, QWORD PTR [rdi] ; xx xx xx xx xx xx xx xx 97 96 95 94 93 92 91 90
- movq xmm0, QWORD PTR [rsi+2*rax] ; xx xx xx xx xx xx xx xx a7 a6 a5 a4 a3 a2 a1 a0
- movq xmm5, QWORD PTR [rdi+2*rax] ; xx xx xx xx xx xx xx xx b7 b6 b5 b4 b3 b2 b1 b0
- movq xmm1, QWORD PTR [rsi+4*rax] ; xx xx xx xx xx xx xx xx c7 c6 c5 c4 c3 c2 c1 c0
-
- punpcklbw xmm2, xmm6 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80
-
- movq xmm6, QWORD PTR [rdi+4*rax] ; xx xx xx xx xx xx xx xx d7 d6 d5 d4 d3 d2 d1 d0
-
- punpcklbw xmm0, xmm5 ; b7 a7 b6 a6 b5 a5 b4 a4 b3 a3 b2 a2 b1 a1 b0 a0
-
- movq xmm5, QWORD PTR [rsi+2*rcx] ; xx xx xx xx xx xx xx xx e7 e6 e5 e4 e3 e2 e1 e0
-
- punpcklbw xmm1, xmm6 ; d7 c7 d6 c6 d5 c5 d4 c4 d3 c3 d2 c2 d1 e1 d0 c0
-
- movq xmm6, QWORD PTR [rdi+2*rcx] ; xx xx xx xx xx xx xx xx f7 f6 f5 f4 f3 f2 f1 f0
-
- punpcklbw xmm5, xmm6 ; f7 e7 f6 e6 f5 e5 f4 e4 f3 e3 f2 e2 f1 e1 f0 e0
-
- movdqa xmm6, xmm1 ;
- punpckhwd xmm6, xmm5 ; f7 e7 d7 c7 f6 e6 d6 c6 f5 e5 d5 c5 f4 e4 d4 c4
-
- punpcklwd xmm1, xmm5 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0
- movdqa xmm5, xmm2 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80
-
- punpcklwd xmm5, xmm0 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80
-
- punpckhwd xmm2, xmm0 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84
-
- movdqa xmm0, xmm5
- punpckldq xmm0, xmm1 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80
-
- punpckhdq xmm5, xmm1 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82
- movdqa xmm1, xmm2 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84
-
- punpckldq xmm1, xmm6 ; f5 e5 d5 c5 b5 a5 95 85 f4 e4 d4 c4 b4 a4 94 84
-
- punpckhdq xmm2, xmm6 ; f7 e7 d7 c7 b7 a7 97 87 f6 e6 d6 c6 b6 a6 96 86
- movdqa xmm6, xmm7 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06
-
- punpcklqdq xmm6, xmm2 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06
-
- punpckhqdq xmm7, xmm2 ; f7 e7 d7 c7 b7 a7 97 87 77 67 57 47 37 27 17 07
-%if %2
- movdqa xmm2, xmm3 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
- punpcklqdq xmm2, xmm5 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
-
- punpckhqdq xmm3, xmm5 ; f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
-
- movdqa [rdx], xmm2 ; save 2
-
- movdqa xmm5, xmm4 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04
- punpcklqdq xmm4, xmm1 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
-
- movdqa [rdx+16], xmm3 ; save 3
-
- punpckhqdq xmm5, xmm1 ; f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05
-
- movdqa [rdx+32], xmm4 ; save 4
- movdqa [rdx+48], xmm5 ; save 5
- movdqa xmm1, t0 ; get
-
- movdqa xmm2, xmm1 ;
- punpckhqdq xmm1, xmm0 ; f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01
-
- punpcklqdq xmm2, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
-%else
- movdqa [rdx+112], xmm7 ; save 7
-
- movdqa [rdx+96], xmm6 ; save 6
-
- movdqa xmm2, xmm3 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
- punpckhqdq xmm3, xmm5 ; f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
-
- punpcklqdq xmm2, xmm5 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
-
- movdqa [rdx+32], xmm2 ; save 2
-
- movdqa xmm5, xmm4 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04
- punpcklqdq xmm4, xmm1 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
-
- movdqa [rdx+48], xmm3 ; save 3
-
- punpckhqdq xmm5, xmm1 ; f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05
-
- movdqa [rdx+64], xmm4 ; save 4
- movdqa [rdx+80], xmm5 ; save 5
- movdqa xmm1, t0 ; get
-
- movdqa xmm2, xmm1
- punpckhqdq xmm1, xmm0 ; f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01
-
- punpcklqdq xmm2, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
-
- movdqa [rdx+16], xmm1
-
- movdqa [rdx], xmm2
-%endif
-%endmacro
-
-%macro LFV_FILTER_MASK_HEV_MASK 1
- movdqa xmm0, xmm6 ; q2
- psubusb xmm0, xmm7 ; q2-q3
-
- psubusb xmm7, xmm6 ; q3-q2
- movdqa xmm4, xmm5 ; q1
-
- por xmm7, xmm0 ; abs (q3-q2)
- psubusb xmm4, xmm6 ; q1-q2
-
- movdqa xmm0, xmm1
- psubusb xmm6, xmm5 ; q2-q1
-
- por xmm6, xmm4 ; abs (q2-q1)
- psubusb xmm0, xmm2 ; p2 - p3;
-
- psubusb xmm2, xmm1 ; p3 - p2;
- por xmm0, xmm2 ; abs(p2-p3)
-%if %1
- movdqa xmm2, [rdx] ; p1
-%else
- movdqa xmm2, [rdx+32] ; p1
-%endif
- movdqa xmm5, xmm2 ; p1
- pmaxub xmm0, xmm7
-
- psubusb xmm5, xmm1 ; p1-p2
- psubusb xmm1, xmm2 ; p2-p1
-
- movdqa xmm7, xmm3 ; p0
- psubusb xmm7, xmm2 ; p0-p1
-
- por xmm1, xmm5 ; abs(p2-p1)
- pmaxub xmm0, xmm6
-
- pmaxub xmm0, xmm1
- movdqa xmm1, xmm2 ; p1
-
- psubusb xmm2, xmm3 ; p1-p0
- lea rdx, srct
-
- por xmm2, xmm7 ; abs(p1-p0)
-
- movdqa t0, xmm2 ; save abs(p1-p0)
-
- pmaxub xmm0, xmm2
-
-%if %1
- movdqa xmm5, [rdx+32] ; q0
- movdqa xmm7, [rdx+48] ; q1
-%else
- movdqa xmm5, [rdx+64] ; q0
- movdqa xmm7, [rdx+80] ; q1
-%endif
- mov rdx, arg(3) ; limit
-
- movdqa xmm6, xmm5 ; q0
- movdqa xmm2, xmm7 ; q1
-
- psubusb xmm5, xmm7 ; q0-q1
- psubusb xmm7, xmm6 ; q1-q0
-
- por xmm7, xmm5 ; abs(q1-q0)
-
- movdqa t1, xmm7 ; save abs(q1-q0)
-
- movdqa xmm4, XMMWORD PTR [rdx]; limit
-
- pmaxub xmm0, xmm7
- mov rdx, arg(2) ; flimit
-
- psubusb xmm0, xmm4
- movdqa xmm5, xmm2 ; q1
-
- psubusb xmm5, xmm1 ; q1-=p1
- psubusb xmm1, xmm2 ; p1-=q1
-
- por xmm5, xmm1 ; abs(p1-q1)
- movdqa xmm1, xmm3 ; p0
-
- pand xmm5, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psubusb xmm1, xmm6 ; p0-q0
-
- psrlw xmm5, 1 ; abs(p1-q1)/2
- psubusb xmm6, xmm3 ; q0-p0
-
- movdqa xmm2, XMMWORD PTR [rdx]; flimit
-
- mov rdx, arg(4) ; get thresh
-
- por xmm1, xmm6 ; abs(q0-p0)
- paddb xmm2, xmm2 ; flimit*2 (less than 255)
-
- movdqa xmm6, t0 ; get abs (q1 - q0)
-
- paddusb xmm1, xmm1 ; abs(q0-p0)*2
-
- movdqa xmm3, t1 ; get abs (p1 - p0)
-
- movdqa xmm7, XMMWORD PTR [rdx]
-
- paddusb xmm1, xmm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2
- psubusb xmm6, xmm7 ; abs(q1 - q0) > thresh
-
- paddb xmm4, xmm2 ; flimit * 2 + limit (less than 255)
- psubusb xmm3, xmm7 ; abs(p1 - p0)> thresh
-
- psubusb xmm1, xmm4 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- por xmm6, xmm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh
-
- por xmm1, xmm0 ; mask
- pcmpeqb xmm6, xmm0
-
- pxor xmm0, xmm0
- pcmpeqb xmm4, xmm4
-
- pcmpeqb xmm1, xmm0
- pxor xmm4, xmm6
-%endmacro
-
-%macro BV_TRANSPOSE 0
- ; xmm1 = f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
- ; xmm6 = f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
- ; xmm3 = f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
- ; xmm7 = f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05
- movdqa xmm2, xmm1 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
- punpcklbw xmm2, xmm6 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
-
- movdqa xmm4, xmm3 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
- punpckhbw xmm1, xmm6 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
-
- punpcklbw xmm4, xmm7 ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04
-
- punpckhbw xmm3, xmm7 ; f5 f4 e5 e4 d5 d4 c5 c4 b5 b4 a5 a4 95 94 85 84
-
- movdqa xmm6, xmm2 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
- punpcklwd xmm2, xmm4 ; 35 34 33 32 25 24 23 22 15 14 13 12 05 04 03 02
-
- punpckhwd xmm6, xmm4 ; 75 74 73 72 65 64 63 62 55 54 53 52 45 44 43 42
- movdqa xmm5, xmm1 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
-
- punpcklwd xmm1, xmm3 ; b5 b4 b3 b2 a5 a4 a3 a2 95 94 93 92 85 84 83 82
-
- punpckhwd xmm5, xmm3 ; f5 f4 f3 f2 e5 e4 e3 e2 d5 d4 d3 d2 c5 c4 c3 c2
- ; xmm2 = 35 34 33 32 25 24 23 22 15 14 13 12 05 04 03 02
- ; xmm6 = 75 74 73 72 65 64 63 62 55 54 53 52 45 44 43 42
- ; xmm1 = b5 b4 b3 b2 a5 a4 a3 a2 95 94 93 92 85 84 83 82
- ; xmm5 = f5 f4 f3 f2 e5 e4 e3 e2 d5 d4 d3 d2 c5 c4 c3 c2
-%endmacro
-
-%macro BV_WRITEBACK 2
- movd [rsi+2], %1
- psrldq %1, 4
-
- movd [rdi+2], %1
- psrldq %1, 4
-
- movd [rsi+2*rax+2], %1
- psrldq %1, 4
-
- movd [rdi+2*rax+2], %1
-
- movd [rsi+4*rax+2], %2
- psrldq %2, 4
-
- movd [rdi+4*rax+2], %2
- psrldq %2, 4
-
- movd [rsi+2*rcx+2], %2
- psrldq %2, 4
-
- movd [rdi+2*rcx+2], %2
-%endmacro
-
-
-;void vp8_loop_filter_vertical_edge_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_vertical_edge_sse2)
-sym(vp8_loop_filter_vertical_edge_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 96 ; reserve 96 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16];
- %define srct [rsp + 32] ;__declspec(align(16)) char srct[64];
-
- mov rsi, arg(0) ; src_ptr
- movsxd rax, dword ptr arg(1) ; src_pixel_step
-
- lea rsi, [rsi - 4]
- lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
- lea rcx, [rax*2+rax]
-
- ;transpose 16x8 to 8x16, and store the 8-line result on stack.
- TRANSPOSE_16X8 1, 1
-
- ; calculate filter mask and high edge variance
- LFV_FILTER_MASK_HEV_MASK 1
-
- ; start work on filters
- B_FILTER 2
-
- ; tranpose and write back - only work on q1, q0, p0, p1
- BV_TRANSPOSE
- ; store 16-line result
-
- lea rdx, [rax]
- neg rdx
-
- BV_WRITEBACK xmm1, xmm5
-
- lea rsi, [rsi+rdx*8]
- lea rdi, [rdi+rdx*8]
- BV_WRITEBACK xmm2, xmm6
-
- add rsp, 96
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_loop_filter_vertical_edge_uv_sse2
-;(
-; unsigned char *u,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; unsigned char *v
-;)
-global sym(vp8_loop_filter_vertical_edge_uv_sse2)
-sym(vp8_loop_filter_vertical_edge_uv_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 96 ; reserve 96 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16];
- %define srct [rsp + 32] ;__declspec(align(16)) char srct[64];
-
- mov rsi, arg(0) ; u_ptr
- movsxd rax, dword ptr arg(1) ; src_pixel_step
-
- lea rsi, [rsi - 4]
- lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
- lea rcx, [rax+2*rax]
-
- lea rdx, srct
-
- ;transpose 16x8 to 8x16, and store the 8-line result on stack.
- TRANSPOSE_16X8 0, 1
-
- ; calculate filter mask and high edge variance
- LFV_FILTER_MASK_HEV_MASK 1
-
- ; start work on filters
- B_FILTER 2
-
- ; tranpose and write back - only work on q1, q0, p0, p1
- BV_TRANSPOSE
-
- lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
-
- ; store 16-line result
- BV_WRITEBACK xmm1, xmm5
-
- mov rsi, arg(0) ; u_ptr
- lea rsi, [rsi - 4]
- lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
- BV_WRITEBACK xmm2, xmm6
-
- add rsp, 96
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-%macro MBV_TRANSPOSE 0
- movdqa xmm0, [rdx] ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
- movdqa xmm1, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
-
- punpcklbw xmm0, xmm7 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00
- punpckhbw xmm1, xmm7 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80
-
- movdqa xmm2, [rdx+32] ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
- movdqa xmm6, xmm2 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
-
- punpcklbw xmm2, [rdx+48] ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
- punpckhbw xmm6, [rdx+48] ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
-
- movdqa xmm3, xmm0 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00
- punpcklwd xmm0, xmm2 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00
-
- punpckhwd xmm3, xmm2 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40
- movdqa xmm4, xmm1 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80
-
- punpcklwd xmm1, xmm6 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80
- punpckhwd xmm4, xmm6 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0
-
- movdqa xmm2, [rdx+64] ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
- punpcklbw xmm2, [rdx+80] ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04
-
- movdqa xmm6, xmm5 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06
- punpcklbw xmm6, [rdx+112] ; 77 76 67 66 57 56 47 46 37 36 27 26 17 16 07 06
-
- movdqa xmm7, xmm2 ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04
- punpcklwd xmm2, xmm6 ; 37 36 35 34 27 26 25 24 17 16 15 14 07 06 05 04
-
- punpckhwd xmm7, xmm6 ; 77 76 75 74 67 66 65 64 57 56 55 54 47 46 45 44
- movdqa xmm6, xmm0 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00
-
- punpckldq xmm0, xmm2 ; 17 16 15 14 13 12 11 10 07 06 05 04 03 02 01 00
- punpckhdq xmm6, xmm2 ; 37 36 35 34 33 32 31 30 27 26 25 24 23 22 21 20
-%endmacro
-
-%macro MBV_WRITEBACK_1 0
- movq QWORD PTR [rsi], xmm0
- movhps MMWORD PTR [rdi], xmm0
-
- movq QWORD PTR [rsi+2*rax], xmm6
- movhps MMWORD PTR [rdi+2*rax], xmm6
-
- movdqa xmm0, xmm3 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40
- punpckldq xmm0, xmm7 ; 57 56 55 54 53 52 51 50 47 46 45 44 43 42 41 40
-
- punpckhdq xmm3, xmm7 ; 77 76 75 74 73 72 71 70 67 66 65 64 63 62 61 60
-
- movq QWORD PTR [rsi+4*rax], xmm0
- movhps MMWORD PTR [rdi+4*rax], xmm0
-
- movq QWORD PTR [rsi+2*rcx], xmm3
- movhps MMWORD PTR [rdi+2*rcx], xmm3
-
- movdqa xmm2, [rdx+64] ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04
- punpckhbw xmm2, [rdx+80] ; f5 f4 e5 e4 d5 d4 c5 c4 b5 b4 a5 a4 95 94 85 84
-
- punpckhbw xmm5, [rdx+112] ; f7 f6 e7 e6 d7 d6 c7 c6 b7 b6 a7 a6 97 96 87 86
- movdqa xmm0, xmm2
-
- punpcklwd xmm0, xmm5 ; b7 b6 b4 b4 a7 a6 a5 a4 97 96 95 94 87 86 85 84
- punpckhwd xmm2, xmm5 ; f7 f6 f5 f4 e7 e6 e5 e4 d7 d6 d5 d4 c7 c6 c5 c4
-
- movdqa xmm5, xmm1 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80
- punpckldq xmm1, xmm0 ; 97 96 95 94 93 92 91 90 87 86 85 83 84 82 81 80
-
- punpckhdq xmm5, xmm0 ; b7 b6 b5 b4 b3 b2 b1 b0 a7 a6 a5 a4 a3 a2 a1 a0
-%endmacro
-
-%macro MBV_WRITEBACK_2 0
- movq QWORD PTR [rsi], xmm1
- movhps MMWORD PTR [rdi], xmm1
-
- movq QWORD PTR [rsi+2*rax], xmm5
- movhps MMWORD PTR [rdi+2*rax], xmm5
-
- movdqa xmm1, xmm4 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0
- punpckldq xmm1, xmm2 ; d7 d6 d5 d4 d3 d2 d1 d0 c7 c6 c5 c4 c3 c2 c1 c0
- punpckhdq xmm4, xmm2 ; f7 f6 f4 f4 f3 f2 f1 f0 e7 e6 e5 e4 e3 e2 e1 e0
-
- movq QWORD PTR [rsi+4*rax], xmm1
- movhps MMWORD PTR [rdi+4*rax], xmm1
-
- movq QWORD PTR [rsi+2*rcx], xmm4
- movhps MMWORD PTR [rdi+2*rcx], xmm4
-%endmacro
-
-
-;void vp8_mbloop_filter_vertical_edge_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_mbloop_filter_vertical_edge_sse2)
-sym(vp8_mbloop_filter_vertical_edge_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 160 ; reserve 160 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16];
- %define srct [rsp + 32] ;__declspec(align(16)) char srct[128];
-
- mov rsi, arg(0) ; src_ptr
- movsxd rax, dword ptr arg(1) ; src_pixel_step
-
- lea rsi, [rsi - 4]
- lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
- lea rcx, [rax*2+rax]
-
- ; Transpose
- TRANSPOSE_16X8 1, 0
-
- ; calculate filter mask and high edge variance
- LFV_FILTER_MASK_HEV_MASK 0
-
- neg rax
- ; start work on filters
- MB_FILTER_AND_WRITEBACK 2
-
- lea rsi, [rsi+rax*8]
- lea rdi, [rdi+rax*8]
-
- ; transpose and write back
- MBV_TRANSPOSE
-
- neg rax
-
- MBV_WRITEBACK_1
-
- lea rsi, [rsi+rax*8]
- lea rdi, [rdi+rax*8]
- MBV_WRITEBACK_2
-
- add rsp, 160
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_mbloop_filter_vertical_edge_uv_sse2
-;(
-; unsigned char *u,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; unsigned char *v
-;)
-global sym(vp8_mbloop_filter_vertical_edge_uv_sse2)
-sym(vp8_mbloop_filter_vertical_edge_uv_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 160 ; reserve 160 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16];
- %define srct [rsp + 32] ;__declspec(align(16)) char srct[128];
-
- mov rsi, arg(0) ; u_ptr
- movsxd rax, dword ptr arg(1) ; src_pixel_step
-
- lea rsi, [rsi - 4]
- lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing
- lea rcx, [rax+2*rax]
-
- lea rdx, srct
-
- ; Transpose
- TRANSPOSE_16X8 0, 0
-
- ; calculate filter mask and high edge variance
- LFV_FILTER_MASK_HEV_MASK 0
-
- ; start work on filters
- MB_FILTER_AND_WRITEBACK 2
-
- ; transpose and write back
- MBV_TRANSPOSE
-
- mov rsi, arg(0) ;u_ptr
- lea rsi, [rsi - 4]
- lea rdi, [rsi + rax]
- MBV_WRITEBACK_1
- mov rsi, arg(5) ;v_ptr
- lea rsi, [rsi - 4]
- lea rdi, [rsi + rax]
- MBV_WRITEBACK_2
-
- add rsp, 160
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_loop_filter_simple_horizontal_edge_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_simple_horizontal_edge_sse2)
-sym(vp8_loop_filter_simple_horizontal_edge_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
- mov rdx, arg(2) ;flimit ; get flimit
- movdqa xmm3, XMMWORD PTR [rdx]
- mov rdx, arg(3) ;limit
- movdqa xmm7, XMMWORD PTR [rdx]
-
- paddb xmm3, xmm3 ; flimit*2 (less than 255)
- paddb xmm3, xmm7 ; flimit * 2 + limit (less than 255)
-
- mov rdi, rsi ; rdi points to row +1 for indirect addressing
- add rdi, rax
- neg rax
-
- ; calculate mask
- movdqu xmm1, [rsi+2*rax] ; p1
- movdqu xmm0, [rdi] ; q1
- movdqa xmm2, xmm1
- movdqa xmm7, xmm0
- movdqa xmm4, xmm0
- psubusb xmm0, xmm1 ; q1-=p1
- psubusb xmm1, xmm4 ; p1-=q1
- por xmm1, xmm0 ; abs(p1-q1)
- pand xmm1, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psrlw xmm1, 1 ; abs(p1-q1)/2
-
- movdqu xmm5, [rsi+rax] ; p0
- movdqu xmm4, [rsi] ; q0
- movdqa xmm0, xmm4 ; q0
- movdqa xmm6, xmm5 ; p0
- psubusb xmm5, xmm4 ; p0-=q0
- psubusb xmm4, xmm6 ; q0-=p0
- por xmm5, xmm4 ; abs(p0 - q0)
- paddusb xmm5, xmm5 ; abs(p0-q0)*2
- paddusb xmm5, xmm1 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- psubusb xmm5, xmm3 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- pxor xmm3, xmm3
- pcmpeqb xmm5, xmm3
-
- ; start work on filters
- pxor xmm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor xmm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
- psubsb xmm2, xmm7 ; p1 - q1
-
- pxor xmm6, [GLOBAL(t80)] ; offset to convert to signed values
- pxor xmm0, [GLOBAL(t80)] ; offset to convert to signed values
- movdqa xmm3, xmm0 ; q0
- psubsb xmm0, xmm6 ; q0 - p0
- paddsb xmm2, xmm0 ; p1 - q1 + 1 * (q0 - p0)
- paddsb xmm2, xmm0 ; p1 - q1 + 2 * (q0 - p0)
- paddsb xmm2, xmm0 ; p1 - q1 + 3 * (q0 - p0)
- pand xmm5, xmm2 ; mask filter values we don't care about
-
- ; do + 4 side
- paddsb xmm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
-
- movdqa xmm0, xmm5 ; get a copy of filters
- psllw xmm0, 8 ; shift left 8
- psraw xmm0, 3 ; arithmetic shift right 11
- psrlw xmm0, 8
- movdqa xmm1, xmm5 ; get a copy of filters
- psraw xmm1, 11 ; arithmetic shift right 11
- psllw xmm1, 8 ; shift left 8 to put it back
-
- por xmm0, xmm1 ; put the two together to get result
-
- psubsb xmm3, xmm0 ; q0-= q0 add
- pxor xmm3, [GLOBAL(t80)] ; unoffset
- movdqu [rsi], xmm3 ; write back
-
- ; now do +3 side
- psubsb xmm5, [GLOBAL(t1s)] ; +3 instead of +4
-
- movdqa xmm0, xmm5 ; get a copy of filters
- psllw xmm0, 8 ; shift left 8
- psraw xmm0, 3 ; arithmetic shift right 11
- psrlw xmm0, 8
- psraw xmm5, 11 ; arithmetic shift right 11
- psllw xmm5, 8 ; shift left 8 to put it back
- por xmm0, xmm5 ; put the two together to get result
-
-
- paddsb xmm6, xmm0 ; p0+= p0 add
- pxor xmm6, [GLOBAL(t80)] ; unoffset
- movdqu [rsi+rax], xmm6 ; write back
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_loop_filter_simple_vertical_edge_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixel_step,
-; const char *flimit,
-; const char *limit,
-; const char *thresh,
-; int count
-;)
-global sym(vp8_loop_filter_simple_vertical_edge_sse2)
-sym(vp8_loop_filter_simple_vertical_edge_sse2):
- push rbp ; save old base pointer value.
- mov rbp, rsp ; set new base pointer value.
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx ; save callee-saved reg
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 32 ; reserve 32 bytes
- %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16];
- %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16];
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
-
- lea rsi, [rsi - 2 ]
- lea rdi, [rsi + rax]
- lea rdx, [rsi + rax*4]
- lea rcx, [rdx + rax]
-
- movdqu xmm0, [rsi] ; (high 96 bits unused) 03 02 01 00
- movdqu xmm1, [rdx] ; (high 96 bits unused) 43 42 41 40
- movdqu xmm2, [rdi] ; 13 12 11 10
- movdqu xmm3, [rcx] ; 53 52 51 50
- punpckldq xmm0, xmm1 ; (high 64 bits unused) 43 42 41 40 03 02 01 00
- punpckldq xmm2, xmm3 ; 53 52 51 50 13 12 11 10
-
- movdqu xmm4, [rsi + rax*2] ; 23 22 21 20
- movdqu xmm5, [rdx + rax*2] ; 63 62 61 60
- movdqu xmm6, [rdi + rax*2] ; 33 32 31 30
- movdqu xmm7, [rcx + rax*2] ; 73 72 71 70
- punpckldq xmm4, xmm5 ; 63 62 61 60 23 22 21 20
- punpckldq xmm6, xmm7 ; 73 72 71 70 33 32 31 30
-
- punpcklbw xmm0, xmm2 ; 53 43 52 42 51 41 50 40 13 03 12 02 11 01 10 00
- punpcklbw xmm4, xmm6 ; 73 63 72 62 71 61 70 60 33 23 32 22 31 21 30 20
-
- movdqa xmm1, xmm0
- punpcklwd xmm0, xmm4 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00
- punpckhwd xmm1, xmm4 ; 73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40
-
- movdqa xmm2, xmm0
- punpckldq xmm0, xmm1 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
- punpckhdq xmm2, xmm1 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
-
- movdqa t0, xmm0 ; save to t0
- movdqa t1, xmm2 ; save to t1
-
- lea rsi, [rsi + rax*8]
- lea rdi, [rsi + rax]
- lea rdx, [rsi + rax*4]
- lea rcx, [rdx + rax]
-
- movdqu xmm4, [rsi] ; 83 82 81 80
- movdqu xmm1, [rdx] ; c3 c2 c1 c0
- movdqu xmm6, [rdi] ; 93 92 91 90
- movdqu xmm3, [rcx] ; d3 d2 d1 d0
- punpckldq xmm4, xmm1 ; c3 c2 c1 c0 83 82 81 80
- punpckldq xmm6, xmm3 ; d3 d2 d1 d0 93 92 91 90
-
- movdqu xmm0, [rsi + rax*2] ; a3 a2 a1 a0
- movdqu xmm5, [rdx + rax*2] ; e3 e2 e1 e0
- movdqu xmm2, [rdi + rax*2] ; b3 b2 b1 b0
- movdqu xmm7, [rcx + rax*2] ; f3 f2 f1 f0
- punpckldq xmm0, xmm5 ; e3 e2 e1 e0 a3 a2 a1 a0
- punpckldq xmm2, xmm7 ; f3 f2 f1 f0 b3 b2 b1 b0
-
- punpcklbw xmm4, xmm6 ; d3 c3 d2 c2 d1 c1 d0 c0 93 83 92 82 91 81 90 80
- punpcklbw xmm0, xmm2 ; f3 e3 f2 e2 f1 e1 f0 e0 b3 a3 b2 a2 b1 a1 b0 a0
-
- movdqa xmm1, xmm4
- punpcklwd xmm4, xmm0 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80
- punpckhwd xmm1, xmm0 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0
-
- movdqa xmm6, xmm4
- punpckldq xmm4, xmm1 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80
- punpckhdq xmm6, xmm1 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82
-
- movdqa xmm0, t0 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
- movdqa xmm2, t1 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
- movdqa xmm1, xmm0
- movdqa xmm3, xmm2
-
- punpcklqdq xmm0, xmm4 ; p1 f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
- punpckhqdq xmm1, xmm4 ; p0 f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01
- punpcklqdq xmm2, xmm6 ; q0 f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
- punpckhqdq xmm3, xmm6 ; q1 f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
-
- ; calculate mask
- movdqa xmm6, xmm0 ; p1
- movdqa xmm7, xmm3 ; q1
- psubusb xmm7, xmm0 ; q1-=p1
- psubusb xmm6, xmm3 ; p1-=q1
- por xmm6, xmm7 ; abs(p1-q1)
- pand xmm6, [GLOBAL(tfe)] ; set lsb of each byte to zero
- psrlw xmm6, 1 ; abs(p1-q1)/2
-
- movdqa xmm5, xmm1 ; p0
- movdqa xmm4, xmm2 ; q0
- psubusb xmm5, xmm2 ; p0-=q0
- psubusb xmm4, xmm1 ; q0-=p0
- por xmm5, xmm4 ; abs(p0 - q0)
- paddusb xmm5, xmm5 ; abs(p0-q0)*2
- paddusb xmm5, xmm6 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
- mov rdx, arg(2) ;flimit
- movdqa xmm7, XMMWORD PTR [rdx]
- mov rdx, arg(3) ; get limit
- movdqa xmm6, XMMWORD PTR [rdx]
- paddb xmm7, xmm7 ; flimit*2 (less than 255)
- paddb xmm7, xmm6 ; flimit * 2 + limit (less than 255)
-
- psubusb xmm5, xmm7 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit
- pxor xmm7, xmm7
- pcmpeqb xmm5, xmm7 ; mm5 = mask
-
- ; start work on filters
- movdqa t0, xmm0
- movdqa t1, xmm3
-
- pxor xmm0, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor xmm3, [GLOBAL(t80)] ; q1 offset to convert to signed values
-
- psubsb xmm0, xmm3 ; p1 - q1
- movdqa xmm6, xmm1 ; p0
-
- movdqa xmm7, xmm2 ; q0
- pxor xmm6, [GLOBAL(t80)] ; offset to convert to signed values
-
- pxor xmm7, [GLOBAL(t80)] ; offset to convert to signed values
- movdqa xmm3, xmm7 ; offseted ; q0
-
- psubsb xmm7, xmm6 ; q0 - p0
- paddsb xmm0, xmm7 ; p1 - q1 + 1 * (q0 - p0)
-
- paddsb xmm0, xmm7 ; p1 - q1 + 2 * (q0 - p0)
- paddsb xmm0, xmm7 ; p1 - q1 + 3 * (q0 - p0)
-
- pand xmm5, xmm0 ; mask filter values we don't care about
-
-
- paddsb xmm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
-
- movdqa xmm0, xmm5 ; get a copy of filters
- psllw xmm0, 8 ; shift left 8
-
- psraw xmm0, 3 ; arithmetic shift right 11
- psrlw xmm0, 8
-
- movdqa xmm7, xmm5 ; get a copy of filters
- psraw xmm7, 11 ; arithmetic shift right 11
-
- psllw xmm7, 8 ; shift left 8 to put it back
- por xmm0, xmm7 ; put the two together to get result
-
- psubsb xmm3, xmm0 ; q0-= q0sz add
- pxor xmm3, [GLOBAL(t80)] ; unoffset q0
-
- ; now do +3 side
- psubsb xmm5, [GLOBAL(t1s)] ; +3 instead of +4
- movdqa xmm0, xmm5 ; get a copy of filters
-
- psllw xmm0, 8 ; shift left 8
- psraw xmm0, 3 ; arithmetic shift right 11
-
- psrlw xmm0, 8
- psraw xmm5, 11 ; arithmetic shift right 11
-
- psllw xmm5, 8 ; shift left 8 to put it back
- por xmm0, xmm5 ; put the two together to get result
-
- paddsb xmm6, xmm0 ; p0+= p0 add
- pxor xmm6, [GLOBAL(t80)] ; unoffset p0
-
- movdqa xmm0, t0 ; p1
- movdqa xmm4, t1 ; q1
-
- ; transpose back to write out
- ; p1 f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
- ; p0 f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01
- ; q0 f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
- ; q1 f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
- movdqa xmm1, xmm0
- punpcklbw xmm0, xmm6 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00
- punpckhbw xmm1, xmm6 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80
-
- movdqa xmm5, xmm3
- punpcklbw xmm3, xmm4 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
- punpckhbw xmm5, xmm4 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
-
- movdqa xmm2, xmm0
- punpcklwd xmm0, xmm3 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00
- punpckhwd xmm2, xmm3 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40
-
- movdqa xmm3, xmm1
- punpcklwd xmm1, xmm5 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80
- punpckhwd xmm3, xmm5 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0
-
- ; write out order: xmm0 xmm2 xmm1 xmm3
- lea rdx, [rsi + rax*4]
-
- movd [rsi], xmm1 ; write the second 8-line result
- psrldq xmm1, 4
- movd [rdi], xmm1
- psrldq xmm1, 4
- movd [rsi + rax*2], xmm1
- psrldq xmm1, 4
- movd [rdi + rax*2], xmm1
-
- movd [rdx], xmm3
- psrldq xmm3, 4
- movd [rcx], xmm3
- psrldq xmm3, 4
- movd [rdx + rax*2], xmm3
- psrldq xmm3, 4
- movd [rcx + rax*2], xmm3
-
- neg rax
- lea rsi, [rsi + rax*8]
- neg rax
- lea rdi, [rsi + rax]
- lea rdx, [rsi + rax*4]
- lea rcx, [rdx + rax]
-
- movd [rsi], xmm0 ; write the first 8-line result
- psrldq xmm0, 4
- movd [rdi], xmm0
- psrldq xmm0, 4
- movd [rsi + rax*2], xmm0
- psrldq xmm0, 4
- movd [rdi + rax*2], xmm0
-
- movd [rdx], xmm2
- psrldq xmm2, 4
- movd [rcx], xmm2
- psrldq xmm2, 4
- movd [rdx + rax*2], xmm2
- psrldq xmm2, 4
- movd [rcx + rax*2], xmm2
-
- add rsp, 32
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-tfe:
- times 16 db 0xfe
-align 16
-t80:
- times 16 db 0x80
-align 16
-t1s:
- times 16 db 0x01
-align 16
-t3:
- times 16 db 0x03
-align 16
-t4:
- times 16 db 0x04
-align 16
-ones:
- times 8 dw 0x0001
-align 16
-s9:
- times 8 dw 0x0900
-align 16
-s63:
- times 8 dw 0x003f
diff --git a/vp8/common/x86/loopfilter_x86.c b/vp8/common/x86/loopfilter_x86.c
deleted file mode 100644
index 5837bc0..0000000
--- a/vp8/common/x86/loopfilter_x86.c
+++ /dev/null
@@ -1,268 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/common/loopfilter.h"
-
-prototype_loopfilter(vp8_loop_filter_horizontal_edge_c);
-prototype_loopfilter(vp8_loop_filter_vertical_edge_c);
-prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_c);
-prototype_loopfilter(vp8_mbloop_filter_vertical_edge_c);
-prototype_loopfilter(vp8_loop_filter_simple_horizontal_edge_c);
-prototype_loopfilter(vp8_loop_filter_simple_vertical_edge_c);
-
-prototype_loopfilter(vp8_mbloop_filter_vertical_edge_mmx);
-prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_mmx);
-prototype_loopfilter(vp8_loop_filter_vertical_edge_mmx);
-prototype_loopfilter(vp8_loop_filter_horizontal_edge_mmx);
-prototype_loopfilter(vp8_loop_filter_simple_vertical_edge_mmx);
-prototype_loopfilter(vp8_loop_filter_simple_horizontal_edge_mmx);
-
-prototype_loopfilter(vp8_loop_filter_vertical_edge_sse2);
-prototype_loopfilter(vp8_loop_filter_horizontal_edge_sse2);
-prototype_loopfilter(vp8_mbloop_filter_vertical_edge_sse2);
-prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_sse2);
-prototype_loopfilter(vp8_loop_filter_simple_vertical_edge_sse2);
-prototype_loopfilter(vp8_loop_filter_simple_horizontal_edge_sse2);
-prototype_loopfilter(vp8_fast_loop_filter_vertical_edges_sse2);
-
-extern loop_filter_uvfunction vp8_loop_filter_horizontal_edge_uv_sse2;
-extern loop_filter_uvfunction vp8_loop_filter_vertical_edge_uv_sse2;
-extern loop_filter_uvfunction vp8_mbloop_filter_horizontal_edge_uv_sse2;
-extern loop_filter_uvfunction vp8_mbloop_filter_vertical_edge_uv_sse2;
-
-#if HAVE_MMX
-/* Horizontal MB filtering */
-void vp8_loop_filter_mbh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_mmx(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_mbloop_filter_horizontal_edge_mmx(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-}
-
-
-void vp8_loop_filter_mbhs_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-
-/* Vertical MB Filtering */
-void vp8_loop_filter_mbv_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_vertical_edge_mmx(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_mbloop_filter_vertical_edge_mmx(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
-}
-
-
-void vp8_loop_filter_mbvs_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-
-/* Horizontal B Filtering */
-void vp8_loop_filter_bh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_horizontal_edge_mmx(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_loop_filter_horizontal_edge_mmx(v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-}
-
-
-void vp8_loop_filter_bhs_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-
-
-/* Vertical B Filtering */
-void vp8_loop_filter_bv_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_vertical_edge_mmx(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_mmx(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_mmx(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_vertical_edge_mmx(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-
- if (v_ptr)
- vp8_loop_filter_vertical_edge_mmx(v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
-}
-
-
-void vp8_loop_filter_bvs_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-#endif
-
-
-/* Horizontal MB filtering */
-#if HAVE_SSE2
-void vp8_loop_filter_mbh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_uv_sse2(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, v_ptr);
-}
-
-
-void vp8_loop_filter_mbhs_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-
-/* Vertical MB Filtering */
-void vp8_loop_filter_mbv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_mbloop_filter_vertical_edge_uv_sse2(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, v_ptr);
-}
-
-
-void vp8_loop_filter_mbvs_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
-}
-
-
-/* Horizontal B Filtering */
-void vp8_loop_filter_bh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_horizontal_edge_uv_sse2(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, v_ptr + 4 * uv_stride);
-}
-
-
-void vp8_loop_filter_bhs_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-
-
-/* Vertical B Filtering */
-void vp8_loop_filter_bv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) simpler_lpf;
- vp8_loop_filter_vertical_edge_sse2(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_sse2(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-
- if (u_ptr)
- vp8_loop_filter_vertical_edge_uv_sse2(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, v_ptr + 4);
-}
-
-
-void vp8_loop_filter_bvs_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
-{
- (void) u_ptr;
- (void) v_ptr;
- (void) uv_stride;
- (void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-
-#endif
-
-#if 0
-void vp8_fast_loop_filter_vertical_edges_sse(unsigned char *y_ptr,
- int y_stride,
- loop_filter_info *lfi)
-{
-
- vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
- vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
-}
-#endif
diff --git a/vp8/common/x86/loopfilter_x86.h b/vp8/common/x86/loopfilter_x86.h
deleted file mode 100644
index 80dbebc..0000000
--- a/vp8/common/x86/loopfilter_x86.h
+++ /dev/null
@@ -1,100 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef LOOPFILTER_X86_H
-#define LOOPFILTER_X86_H
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-
-#if HAVE_MMX
-extern prototype_loopfilter_block(vp8_loop_filter_mbv_mmx);
-extern prototype_loopfilter_block(vp8_loop_filter_bv_mmx);
-extern prototype_loopfilter_block(vp8_loop_filter_mbh_mmx);
-extern prototype_loopfilter_block(vp8_loop_filter_bh_mmx);
-extern prototype_loopfilter_block(vp8_loop_filter_mbvs_mmx);
-extern prototype_loopfilter_block(vp8_loop_filter_bvs_mmx);
-extern prototype_loopfilter_block(vp8_loop_filter_mbhs_mmx);
-extern prototype_loopfilter_block(vp8_loop_filter_bhs_mmx);
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_lf_normal_mb_v
-#define vp8_lf_normal_mb_v vp8_loop_filter_mbv_mmx
-
-#undef vp8_lf_normal_b_v
-#define vp8_lf_normal_b_v vp8_loop_filter_bv_mmx
-
-#undef vp8_lf_normal_mb_h
-#define vp8_lf_normal_mb_h vp8_loop_filter_mbh_mmx
-
-#undef vp8_lf_normal_b_h
-#define vp8_lf_normal_b_h vp8_loop_filter_bh_mmx
-
-#undef vp8_lf_simple_mb_v
-#define vp8_lf_simple_mb_v vp8_loop_filter_mbvs_mmx
-
-#undef vp8_lf_simple_b_v
-#define vp8_lf_simple_b_v vp8_loop_filter_bvs_mmx
-
-#undef vp8_lf_simple_mb_h
-#define vp8_lf_simple_mb_h vp8_loop_filter_mbhs_mmx
-
-#undef vp8_lf_simple_b_h
-#define vp8_lf_simple_b_h vp8_loop_filter_bhs_mmx
-#endif
-#endif
-
-
-#if HAVE_SSE2
-extern prototype_loopfilter_block(vp8_loop_filter_mbv_sse2);
-extern prototype_loopfilter_block(vp8_loop_filter_bv_sse2);
-extern prototype_loopfilter_block(vp8_loop_filter_mbh_sse2);
-extern prototype_loopfilter_block(vp8_loop_filter_bh_sse2);
-extern prototype_loopfilter_block(vp8_loop_filter_mbvs_sse2);
-extern prototype_loopfilter_block(vp8_loop_filter_bvs_sse2);
-extern prototype_loopfilter_block(vp8_loop_filter_mbhs_sse2);
-extern prototype_loopfilter_block(vp8_loop_filter_bhs_sse2);
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_lf_normal_mb_v
-#define vp8_lf_normal_mb_v vp8_loop_filter_mbv_sse2
-
-#undef vp8_lf_normal_b_v
-#define vp8_lf_normal_b_v vp8_loop_filter_bv_sse2
-
-#undef vp8_lf_normal_mb_h
-#define vp8_lf_normal_mb_h vp8_loop_filter_mbh_sse2
-
-#undef vp8_lf_normal_b_h
-#define vp8_lf_normal_b_h vp8_loop_filter_bh_sse2
-
-#undef vp8_lf_simple_mb_v
-#define vp8_lf_simple_mb_v vp8_loop_filter_mbvs_sse2
-
-#undef vp8_lf_simple_b_v
-#define vp8_lf_simple_b_v vp8_loop_filter_bvs_sse2
-
-#undef vp8_lf_simple_mb_h
-#define vp8_lf_simple_mb_h vp8_loop_filter_mbhs_sse2
-
-#undef vp8_lf_simple_b_h
-#define vp8_lf_simple_b_h vp8_loop_filter_bhs_sse2
-#endif
-#endif
-
-
-#endif
diff --git a/vp8/common/x86/postproc_mmx.asm b/vp8/common/x86/postproc_mmx.asm
deleted file mode 100644
index 787e832..0000000
--- a/vp8/common/x86/postproc_mmx.asm
+++ /dev/null
@@ -1,534 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%define VP8_FILTER_WEIGHT 128
-%define VP8_FILTER_SHIFT 7
-
-;void vp8_post_proc_down_and_across_mmx
-;(
-; unsigned char *src_ptr,
-; unsigned char *dst_ptr,
-; int src_pixels_per_line,
-; int dst_pixels_per_line,
-; int rows,
-; int cols,
-; int flimit
-;)
-global sym(vp8_post_proc_down_and_across_mmx)
-sym(vp8_post_proc_down_and_across_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
-%if ABI_IS_32BIT=1 && CONFIG_PIC=1
- ; move the global rd onto the stack, since we don't have enough registers
- ; to do PIC addressing
- movq mm0, [GLOBAL(rd)]
- sub rsp, 8
- movq [rsp], mm0
-%define RD [rsp]
-%else
-%define RD [GLOBAL(rd)]
-%endif
-
- push rbx
- lea rbx, [GLOBAL(Blur)]
- movd mm2, dword ptr arg(6) ;flimit
- punpcklwd mm2, mm2
- punpckldq mm2, mm2
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(1) ;dst_ptr
-
- movsxd rcx, DWORD PTR arg(4) ;rows
- movsxd rax, DWORD PTR arg(2) ;src_pixels_per_line ; destination pitch?
- pxor mm0, mm0 ; mm0 = 00000000
-
-nextrow:
-
- xor rdx, rdx ; clear out rdx for use as loop counter
-nextcol:
-
- pxor mm7, mm7 ; mm7 = 00000000
- movq mm6, [rbx + 32 ] ; mm6 = kernel 2 taps
- movq mm3, [rsi] ; mm4 = r0 p0..p7
- punpcklbw mm3, mm0 ; mm3 = p0..p3
- movq mm1, mm3 ; mm1 = p0..p3
- pmullw mm3, mm6 ; mm3 *= kernel 2 modifiers
-
- movq mm6, [rbx + 48] ; mm6 = kernel 3 taps
- movq mm5, [rsi + rax] ; mm4 = r1 p0..p7
- punpcklbw mm5, mm0 ; mm5 = r1 p0..p3
- pmullw mm6, mm5 ; mm6 *= p0..p3 * kernel 3 modifiers
- paddusw mm3, mm6 ; mm3 += mm6
-
- ; thresholding
- movq mm7, mm1 ; mm7 = r0 p0..p3
- psubusw mm7, mm5 ; mm7 = r0 p0..p3 - r1 p0..p3
- psubusw mm5, mm1 ; mm5 = r1 p0..p3 - r0 p0..p3
- paddusw mm7, mm5 ; mm7 = abs(r0 p0..p3 - r1 p0..p3)
- pcmpgtw mm7, mm2
-
- movq mm6, [rbx + 64 ] ; mm6 = kernel 4 modifiers
- movq mm5, [rsi + 2*rax] ; mm4 = r2 p0..p7
- punpcklbw mm5, mm0 ; mm5 = r2 p0..p3
- pmullw mm6, mm5 ; mm5 *= kernel 4 modifiers
- paddusw mm3, mm6 ; mm3 += mm5
-
- ; thresholding
- movq mm6, mm1 ; mm6 = r0 p0..p3
- psubusw mm6, mm5 ; mm6 = r0 p0..p3 - r2 p0..p3
- psubusw mm5, mm1 ; mm5 = r2 p0..p3 - r2 p0..p3
- paddusw mm6, mm5 ; mm6 = abs(r0 p0..p3 - r2 p0..p3)
- pcmpgtw mm6, mm2
- por mm7, mm6 ; accumulate thresholds
-
-
- neg rax
- movq mm6, [rbx ] ; kernel 0 taps
- movq mm5, [rsi+2*rax] ; mm4 = r-2 p0..p7
- punpcklbw mm5, mm0 ; mm5 = r-2 p0..p3
- pmullw mm6, mm5 ; mm5 *= kernel 0 modifiers
- paddusw mm3, mm6 ; mm3 += mm5
-
- ; thresholding
- movq mm6, mm1 ; mm6 = r0 p0..p3
- psubusw mm6, mm5 ; mm6 = p0..p3 - r-2 p0..p3
- psubusw mm5, mm1 ; mm5 = r-2 p0..p3 - p0..p3
- paddusw mm6, mm5 ; mm6 = abs(r0 p0..p3 - r-2 p0..p3)
- pcmpgtw mm6, mm2
- por mm7, mm6 ; accumulate thresholds
-
- movq mm6, [rbx + 16] ; kernel 1 taps
- movq mm4, [rsi+rax] ; mm4 = r-1 p0..p7
- punpcklbw mm4, mm0 ; mm4 = r-1 p0..p3
- pmullw mm6, mm4 ; mm4 *= kernel 1 modifiers.
- paddusw mm3, mm6 ; mm3 += mm5
-
- ; thresholding
- movq mm6, mm1 ; mm6 = r0 p0..p3
- psubusw mm6, mm4 ; mm6 = p0..p3 - r-2 p0..p3
- psubusw mm4, mm1 ; mm5 = r-1 p0..p3 - p0..p3
- paddusw mm6, mm4 ; mm6 = abs(r0 p0..p3 - r-1 p0..p3)
- pcmpgtw mm6, mm2
- por mm7, mm6 ; accumulate thresholds
-
-
- paddusw mm3, RD ; mm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; mm3 /= 128
-
- pand mm1, mm7 ; mm1 select vals > thresh from source
- pandn mm7, mm3 ; mm7 select vals < thresh from blurred result
- paddusw mm1, mm7 ; combination
-
- packuswb mm1, mm0 ; pack to bytes
-
- movd [rdi], mm1 ;
- neg rax ; pitch is positive
-
-
- add rsi, 4
- add rdi, 4
- add rdx, 4
-
- cmp edx, dword ptr arg(5) ;cols
- jl nextcol
- ; done with the all cols, start the across filtering in place
- sub rsi, rdx
- sub rdi, rdx
-
-
- push rax
- xor rdx, rdx
- mov rax, [rdi-4];
-
-acrossnextcol:
- pxor mm7, mm7 ; mm7 = 00000000
- movq mm6, [rbx + 32 ] ;
- movq mm4, [rdi+rdx] ; mm4 = p0..p7
- movq mm3, mm4 ; mm3 = p0..p7
- punpcklbw mm3, mm0 ; mm3 = p0..p3
- movq mm1, mm3 ; mm1 = p0..p3
- pmullw mm3, mm6 ; mm3 *= kernel 2 modifiers
-
- movq mm6, [rbx + 48]
- psrlq mm4, 8 ; mm4 = p1..p7
- movq mm5, mm4 ; mm5 = p1..p7
- punpcklbw mm5, mm0 ; mm5 = p1..p4
- pmullw mm6, mm5 ; mm6 *= p1..p4 * kernel 3 modifiers
- paddusw mm3, mm6 ; mm3 += mm6
-
- ; thresholding
- movq mm7, mm1 ; mm7 = p0..p3
- psubusw mm7, mm5 ; mm7 = p0..p3 - p1..p4
- psubusw mm5, mm1 ; mm5 = p1..p4 - p0..p3
- paddusw mm7, mm5 ; mm7 = abs(p0..p3 - p1..p4)
- pcmpgtw mm7, mm2
-
- movq mm6, [rbx + 64 ]
- psrlq mm4, 8 ; mm4 = p2..p7
- movq mm5, mm4 ; mm5 = p2..p7
- punpcklbw mm5, mm0 ; mm5 = p2..p5
- pmullw mm6, mm5 ; mm5 *= kernel 4 modifiers
- paddusw mm3, mm6 ; mm3 += mm5
-
- ; thresholding
- movq mm6, mm1 ; mm6 = p0..p3
- psubusw mm6, mm5 ; mm6 = p0..p3 - p1..p4
- psubusw mm5, mm1 ; mm5 = p1..p4 - p0..p3
- paddusw mm6, mm5 ; mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw mm6, mm2
- por mm7, mm6 ; accumulate thresholds
-
-
- movq mm6, [rbx ]
- movq mm4, [rdi+rdx-2] ; mm4 = p-2..p5
- movq mm5, mm4 ; mm5 = p-2..p5
- punpcklbw mm5, mm0 ; mm5 = p-2..p1
- pmullw mm6, mm5 ; mm5 *= kernel 0 modifiers
- paddusw mm3, mm6 ; mm3 += mm5
-
- ; thresholding
- movq mm6, mm1 ; mm6 = p0..p3
- psubusw mm6, mm5 ; mm6 = p0..p3 - p1..p4
- psubusw mm5, mm1 ; mm5 = p1..p4 - p0..p3
- paddusw mm6, mm5 ; mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw mm6, mm2
- por mm7, mm6 ; accumulate thresholds
-
- movq mm6, [rbx + 16]
- psrlq mm4, 8 ; mm4 = p-1..p5
- punpcklbw mm4, mm0 ; mm4 = p-1..p2
- pmullw mm6, mm4 ; mm4 *= kernel 1 modifiers.
- paddusw mm3, mm6 ; mm3 += mm5
-
- ; thresholding
- movq mm6, mm1 ; mm6 = p0..p3
- psubusw mm6, mm4 ; mm6 = p0..p3 - p1..p4
- psubusw mm4, mm1 ; mm5 = p1..p4 - p0..p3
- paddusw mm6, mm4 ; mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw mm6, mm2
- por mm7, mm6 ; accumulate thresholds
-
- paddusw mm3, RD ; mm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; mm3 /= 128
-
- pand mm1, mm7 ; mm1 select vals > thresh from source
- pandn mm7, mm3 ; mm7 select vals < thresh from blurred result
- paddusw mm1, mm7 ; combination
-
- packuswb mm1, mm0 ; pack to bytes
- mov DWORD PTR [rdi+rdx-4], eax ; store previous four bytes
- movd eax, mm1
-
- add rdx, 4
- cmp edx, dword ptr arg(5) ;cols
- jl acrossnextcol;
-
- mov DWORD PTR [rdi+rdx-4], eax
- pop rax
-
- ; done with this rwo
- add rsi,rax ; next line
- movsxd rax, dword ptr arg(3) ;dst_pixels_per_line ; destination pitch?
- add rdi,rax ; next destination
- movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; destination pitch?
-
- dec rcx ; decrement count
- jnz nextrow ; next row
- pop rbx
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-%undef RD
-
-
-;void vp8_mbpost_proc_down_mmx(unsigned char *dst,
-; int pitch, int rows, int cols,int flimit)
-extern sym(vp8_rv)
-global sym(vp8_mbpost_proc_down_mmx)
-sym(vp8_mbpost_proc_down_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 136
-
- ; unsigned char d[16][8] at [rsp]
- ; create flimit2 at [rsp+128]
- mov eax, dword ptr arg(4) ;flimit
- mov [rsp+128], eax
- mov [rsp+128+4], eax
-%define flimit2 [rsp+128]
-
-%if ABI_IS_32BIT=0
- lea r8, [GLOBAL(sym(vp8_rv))]
-%endif
-
- ;rows +=8;
- add dword ptr arg(2), 8
-
- ;for(c=0; c<cols; c+=4)
-loop_col:
- mov rsi, arg(0) ;s
- pxor mm0, mm0 ;
-
- movsxd rax, dword ptr arg(1) ;pitch ;
- neg rax ; rax = -pitch
-
- lea rsi, [rsi + rax*8]; ; rdi = s[-pitch*8]
- neg rax
-
-
- pxor mm5, mm5
- pxor mm6, mm6 ;
-
- pxor mm7, mm7 ;
- mov rdi, rsi
-
- mov rcx, 15 ;
-
-loop_initvar:
- movd mm1, DWORD PTR [rdi];
- punpcklbw mm1, mm0 ;
-
- paddw mm5, mm1 ;
- pmullw mm1, mm1 ;
-
- movq mm2, mm1 ;
- punpcklwd mm1, mm0 ;
-
- punpckhwd mm2, mm0 ;
- paddd mm6, mm1 ;
-
- paddd mm7, mm2 ;
- lea rdi, [rdi+rax] ;
-
- dec rcx
- jne loop_initvar
- ;save the var and sum
- xor rdx, rdx
-loop_row:
- movd mm1, DWORD PTR [rsi] ; [s-pitch*8]
- movd mm2, DWORD PTR [rdi] ; [s+pitch*7]
-
- punpcklbw mm1, mm0
- punpcklbw mm2, mm0
-
- paddw mm5, mm2
- psubw mm5, mm1
-
- pmullw mm2, mm2
- movq mm4, mm2
-
- punpcklwd mm2, mm0
- punpckhwd mm4, mm0
-
- paddd mm6, mm2
- paddd mm7, mm4
-
- pmullw mm1, mm1
- movq mm2, mm1
-
- punpcklwd mm1, mm0
- psubd mm6, mm1
-
- punpckhwd mm2, mm0
- psubd mm7, mm2
-
-
- movq mm3, mm6
- pslld mm3, 4
-
- psubd mm3, mm6
- movq mm1, mm5
-
- movq mm4, mm5
- pmullw mm1, mm1
-
- pmulhw mm4, mm4
- movq mm2, mm1
-
- punpcklwd mm1, mm4
- punpckhwd mm2, mm4
-
- movq mm4, mm7
- pslld mm4, 4
-
- psubd mm4, mm7
-
- psubd mm3, mm1
- psubd mm4, mm2
-
- psubd mm3, flimit2
- psubd mm4, flimit2
-
- psrad mm3, 31
- psrad mm4, 31
-
- packssdw mm3, mm4
- packsswb mm3, mm0
-
- movd mm1, DWORD PTR [rsi+rax*8]
-
- movq mm2, mm1
- punpcklbw mm1, mm0
-
- paddw mm1, mm5
- mov rcx, rdx
-
- and rcx, 127
-%if ABI_IS_32BIT=1 && CONFIG_PIC=1
- push rax
- lea rax, [GLOBAL(sym(vp8_rv))]
- movq mm4, [rax + rcx*2] ;vp8_rv[rcx*2]
- pop rax
-%elif ABI_IS_32BIT=0
- movq mm4, [r8 + rcx*2] ;vp8_rv[rcx*2]
-%else
- movq mm4, [sym(vp8_rv) + rcx*2]
-%endif
- paddw mm1, mm4
- ;paddw xmm1, eight8s
- psraw mm1, 4
-
- packuswb mm1, mm0
- pand mm1, mm3
-
- pandn mm3, mm2
- por mm1, mm3
-
- and rcx, 15
- movd DWORD PTR [rsp+rcx*4], mm1 ;d[rcx*4]
-
- mov rcx, rdx
- sub rcx, 8
-
- and rcx, 15
- movd mm1, DWORD PTR [rsp+rcx*4] ;d[rcx*4]
-
- movd [rsi], mm1
- lea rsi, [rsi+rax]
-
- lea rdi, [rdi+rax]
- add rdx, 1
-
- cmp edx, dword arg(2) ;rows
- jl loop_row
-
-
- add dword arg(0), 4 ; s += 4
- sub dword arg(3), 4 ; cols -= 4
- cmp dword arg(3), 0
- jg loop_col
-
- add rsp, 136
- pop rsp
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-%undef flimit2
-
-
-;void vp8_plane_add_noise_mmx (unsigned char *Start, unsigned char *noise,
-; unsigned char blackclamp[16],
-; unsigned char whiteclamp[16],
-; unsigned char bothclamp[16],
-; unsigned int Width, unsigned int Height, int Pitch)
-extern sym(rand)
-global sym(vp8_plane_add_noise_mmx)
-sym(vp8_plane_add_noise_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 8
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
-addnoise_loop:
- call sym(rand) WRT_PLT
- mov rcx, arg(1) ;noise
- and rax, 0xff
- add rcx, rax
-
- ; we rely on the fact that the clamping vectors are stored contiguously
- ; in black/white/both order. Note that we have to reload this here because
- ; rdx could be trashed by rand()
- mov rdx, arg(2) ; blackclamp
-
-
- mov rdi, rcx
- movsxd rcx, dword arg(5) ;[Width]
- mov rsi, arg(0) ;Pos
- xor rax,rax
-
-addnoise_nextset:
- movq mm1,[rsi+rax] ; get the source
-
- psubusb mm1, [rdx] ;blackclamp ; clamp both sides so we don't outrange adding noise
- paddusb mm1, [rdx+32] ;bothclamp
- psubusb mm1, [rdx+16] ;whiteclamp
-
- movq mm2,[rdi+rax] ; get the noise for this line
- paddb mm1,mm2 ; add it in
- movq [rsi+rax],mm1 ; store the result
-
- add rax,8 ; move to the next line
-
- cmp rax, rcx
- jl addnoise_nextset
-
- movsxd rax, dword arg(7) ; Pitch
- add arg(0), rax ; Start += Pitch
- sub dword arg(6), 1 ; Height -= 1
- jg addnoise_loop
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-SECTION_RODATA
-align 16
-Blur:
- times 16 dw 16
- times 8 dw 64
- times 16 dw 16
- times 8 dw 0
-
-rd:
- times 4 dw 0x40
diff --git a/vp8/common/x86/postproc_mmx.c b/vp8/common/x86/postproc_mmx.c
deleted file mode 100644
index 6b6321a..0000000
--- a/vp8/common/x86/postproc_mmx.c
+++ /dev/null
@@ -1,1508 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <math.h>
-#include <stdlib.h>
-#include "vpx_scale/yv12config.h"
-#include "pragmas.h"
-
-#define VP8_FILTER_WEIGHT 128
-#define VP8_FILTER_SHIFT 7
-
-
-
-/* static constants */
-__declspec(align(16))
-const static short Blur[48] =
-{
-
- 16, 16, 16, 16, 16, 16, 16, 16,
- 16, 16, 16, 16, 16, 16, 16, 16,
- 64, 64, 64, 64, 64, 64, 64, 64,
- 16, 16, 16, 16, 16, 16, 16, 16,
- 16, 16, 16, 16, 16, 16, 16, 16,
- 0, 0, 0, 0, 0, 0, 0, 0,
-
-};
-#define RD __declspec(align(16)) __int64 rd = 0x0040004000400040;
-#define R4D2 __declspec(align(16)) __int64 rd42[2] = {0x0004000400040004,0x0004000400040004};
-
-#ifndef RELOCATEABLE
-const static RD;
-const static R4D2;
-#endif
-
-
-/* external references */
-extern double vp8_gaussian(double sigma, double mu, double x);
-extern short vp8_rv[];
-extern int vp8_q2mbl(int x) ;
-
-
-
-void vp8_post_proc_down_and_across_mmx
-(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- int src_pixels_per_line,
- int dst_pixels_per_line,
- int rows,
- int cols,
- int flimit
-)
-{
-#ifdef RELOCATEABLE
- RD
- R4D2
-#endif
-
- __asm
- {
- push ebx
- lea ebx, Blur
- movd mm2, flimit
- punpcklwd mm2, mm2
- punpckldq mm2, mm2
-
- mov esi, src_ptr
- mov edi, dst_ptr
-
- mov ecx, DWORD PTR rows
- mov eax, src_pixels_per_line ;
- destination pitch?
- pxor mm0, mm0 ;
- mm0 = 00000000
-
- nextrow:
-
- xor edx, edx ;
-
- clear out edx for use as loop counter
- nextcol:
-
- pxor mm7, mm7 ;
-
- mm7 = 00000000
- movq mm6, [ebx + 32 ] ;
- mm6 = kernel 2 taps
- movq mm3, [esi] ;
- mm4 = r0 p0..p7
- punpcklbw mm3, mm0 ;
- mm3 = p0..p3
- movq mm1, mm3 ;
- mm1 = p0..p3
- pmullw mm3, mm6 ;
- mm3 *= kernel 2 modifiers
-
- movq mm6, [ebx + 48] ;
- mm6 = kernel 3 taps
- movq mm5, [esi + eax] ;
- mm4 = r1 p0..p7
- punpcklbw mm5, mm0 ;
- mm5 = r1 p0..p3
- pmullw mm6, mm5 ;
- mm6 *= p0..p3 * kernel 3 modifiers
- paddusw mm3, mm6 ;
- mm3 += mm6
-
- ;
- thresholding
- movq mm7, mm1 ;
- mm7 = r0 p0..p3
- psubusw mm7, mm5 ;
- mm7 = r0 p0..p3 - r1 p0..p3
- psubusw mm5, mm1 ;
- mm5 = r1 p0..p3 - r0 p0..p3
- paddusw mm7, mm5 ;
- mm7 = abs(r0 p0..p3 - r1 p0..p3)
- pcmpgtw mm7, mm2
-
- movq mm6, [ebx + 64 ] ;
- mm6 = kernel 4 modifiers
- movq mm5, [esi + 2*eax] ;
- mm4 = r2 p0..p7
- punpcklbw mm5, mm0 ;
- mm5 = r2 p0..p3
- pmullw mm6, mm5 ;
- mm5 *= kernel 4 modifiers
- paddusw mm3, mm6 ;
- mm3 += mm5
-
- ;
- thresholding
- movq mm6, mm1 ;
- mm6 = r0 p0..p3
- psubusw mm6, mm5 ;
- mm6 = r0 p0..p3 - r2 p0..p3
- psubusw mm5, mm1 ;
- mm5 = r2 p0..p3 - r2 p0..p3
- paddusw mm6, mm5 ;
- mm6 = abs(r0 p0..p3 - r2 p0..p3)
- pcmpgtw mm6, mm2
- por mm7, mm6 ;
- accumulate thresholds
-
-
- neg eax
- movq mm6, [ebx ] ;
- kernel 0 taps
- movq mm5, [esi+2*eax] ;
- mm4 = r-2 p0..p7
- punpcklbw mm5, mm0 ;
- mm5 = r-2 p0..p3
- pmullw mm6, mm5 ;
- mm5 *= kernel 0 modifiers
- paddusw mm3, mm6 ;
- mm3 += mm5
-
- ;
- thresholding
- movq mm6, mm1 ;
- mm6 = r0 p0..p3
- psubusw mm6, mm5 ;
- mm6 = p0..p3 - r-2 p0..p3
- psubusw mm5, mm1 ;
- mm5 = r-2 p0..p3 - p0..p3
- paddusw mm6, mm5 ;
- mm6 = abs(r0 p0..p3 - r-2 p0..p3)
- pcmpgtw mm6, mm2
- por mm7, mm6 ;
- accumulate thresholds
-
- movq mm6, [ebx + 16] ;
- kernel 1 taps
- movq mm4, [esi+eax] ;
- mm4 = r-1 p0..p7
- punpcklbw mm4, mm0 ;
- mm4 = r-1 p0..p3
- pmullw mm6, mm4 ;
- mm4 *= kernel 1 modifiers.
- paddusw mm3, mm6 ;
- mm3 += mm5
-
- ;
- thresholding
- movq mm6, mm1 ;
- mm6 = r0 p0..p3
- psubusw mm6, mm4 ;
- mm6 = p0..p3 - r-2 p0..p3
- psubusw mm4, mm1 ;
- mm5 = r-1 p0..p3 - p0..p3
- paddusw mm6, mm4 ;
- mm6 = abs(r0 p0..p3 - r-1 p0..p3)
- pcmpgtw mm6, mm2
- por mm7, mm6 ;
- accumulate thresholds
-
-
- paddusw mm3, rd ;
- mm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ;
- mm3 /= 128
-
- pand mm1, mm7 ;
- mm1 select vals > thresh from source
- pandn mm7, mm3 ;
- mm7 select vals < thresh from blurred result
- paddusw mm1, mm7 ;
- combination
-
- packuswb mm1, mm0 ;
- pack to bytes
-
- movd [edi], mm1 ;
- neg eax ;
- pitch is positive
-
-
- add esi, 4
- add edi, 4
- add edx, 4
-
- cmp edx, cols
- jl nextcol
- // done with the all cols, start the across filtering in place
- sub esi, edx
- sub edi, edx
-
-
- push eax
- xor edx, edx
- mov eax, [edi-4];
-
- acrossnextcol:
- pxor mm7, mm7 ;
- mm7 = 00000000
- movq mm6, [ebx + 32 ] ;
- movq mm4, [edi+edx] ;
- mm4 = p0..p7
- movq mm3, mm4 ;
- mm3 = p0..p7
- punpcklbw mm3, mm0 ;
- mm3 = p0..p3
- movq mm1, mm3 ;
- mm1 = p0..p3
- pmullw mm3, mm6 ;
- mm3 *= kernel 2 modifiers
-
- movq mm6, [ebx + 48]
- psrlq mm4, 8 ;
- mm4 = p1..p7
- movq mm5, mm4 ;
- mm5 = p1..p7
- punpcklbw mm5, mm0 ;
- mm5 = p1..p4
- pmullw mm6, mm5 ;
- mm6 *= p1..p4 * kernel 3 modifiers
- paddusw mm3, mm6 ;
- mm3 += mm6
-
- ;
- thresholding
- movq mm7, mm1 ;
- mm7 = p0..p3
- psubusw mm7, mm5 ;
- mm7 = p0..p3 - p1..p4
- psubusw mm5, mm1 ;
- mm5 = p1..p4 - p0..p3
- paddusw mm7, mm5 ;
- mm7 = abs(p0..p3 - p1..p4)
- pcmpgtw mm7, mm2
-
- movq mm6, [ebx + 64 ]
- psrlq mm4, 8 ;
- mm4 = p2..p7
- movq mm5, mm4 ;
- mm5 = p2..p7
- punpcklbw mm5, mm0 ;
- mm5 = p2..p5
- pmullw mm6, mm5 ;
- mm5 *= kernel 4 modifiers
- paddusw mm3, mm6 ;
- mm3 += mm5
-
- ;
- thresholding
- movq mm6, mm1 ;
- mm6 = p0..p3
- psubusw mm6, mm5 ;
- mm6 = p0..p3 - p1..p4
- psubusw mm5, mm1 ;
- mm5 = p1..p4 - p0..p3
- paddusw mm6, mm5 ;
- mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw mm6, mm2
- por mm7, mm6 ;
- accumulate thresholds
-
-
- movq mm6, [ebx ]
- movq mm4, [edi+edx-2] ;
- mm4 = p-2..p5
- movq mm5, mm4 ;
- mm5 = p-2..p5
- punpcklbw mm5, mm0 ;
- mm5 = p-2..p1
- pmullw mm6, mm5 ;
- mm5 *= kernel 0 modifiers
- paddusw mm3, mm6 ;
- mm3 += mm5
-
- ;
- thresholding
- movq mm6, mm1 ;
- mm6 = p0..p3
- psubusw mm6, mm5 ;
- mm6 = p0..p3 - p1..p4
- psubusw mm5, mm1 ;
- mm5 = p1..p4 - p0..p3
- paddusw mm6, mm5 ;
- mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw mm6, mm2
- por mm7, mm6 ;
- accumulate thresholds
-
- movq mm6, [ebx + 16]
- psrlq mm4, 8 ;
- mm4 = p-1..p5
- punpcklbw mm4, mm0 ;
- mm4 = p-1..p2
- pmullw mm6, mm4 ;
- mm4 *= kernel 1 modifiers.
- paddusw mm3, mm6 ;
- mm3 += mm5
-
- ;
- thresholding
- movq mm6, mm1 ;
- mm6 = p0..p3
- psubusw mm6, mm4 ;
- mm6 = p0..p3 - p1..p4
- psubusw mm4, mm1 ;
- mm5 = p1..p4 - p0..p3
- paddusw mm6, mm4 ;
- mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw mm6, mm2
- por mm7, mm6 ;
- accumulate thresholds
-
- paddusw mm3, rd ;
- mm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ;
- mm3 /= 128
-
- pand mm1, mm7 ;
- mm1 select vals > thresh from source
- pandn mm7, mm3 ;
- mm7 select vals < thresh from blurred result
- paddusw mm1, mm7 ;
- combination
-
- packuswb mm1, mm0 ;
- pack to bytes
- mov DWORD PTR [edi+edx-4], eax ;
- store previous four bytes
- movd eax, mm1
-
- add edx, 4
- cmp edx, cols
- jl acrossnextcol;
-
- mov DWORD PTR [edi+edx-4], eax
- pop eax
-
- // done with this rwo
- add esi, eax ;
- next line
- mov eax, dst_pixels_per_line ;
- destination pitch?
- add edi, eax ;
- next destination
- mov eax, src_pixels_per_line ;
- destination pitch?
-
- dec ecx ;
- decrement count
- jnz nextrow ;
- next row
- pop ebx
-
- }
-}
-
-
-
-void vp8_post_proc_down_and_across_xmm
-(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- int src_pixels_per_line,
- int dst_pixels_per_line,
- int rows,
- int cols,
- int flimit
-)
-{
-#ifdef RELOCATEABLE
- R4D2
-#endif
-
- __asm
- {
- movd xmm2, flimit
- punpcklwd xmm2, xmm2
- punpckldq xmm2, xmm2
- punpcklqdq xmm2, xmm2
-
- mov esi, src_ptr
- mov edi, dst_ptr
-
- mov ecx, DWORD PTR rows
- mov eax, src_pixels_per_line ;
- destination pitch?
- pxor xmm0, xmm0 ;
- mm0 = 00000000
-
- nextrow:
-
- xor edx, edx ;
-
- clear out edx for use as loop counter
- nextcol:
- movq xmm3, QWORD PTR [esi] ;
-
- mm4 = r0 p0..p7
- punpcklbw xmm3, xmm0 ;
- mm3 = p0..p3
- movdqa xmm1, xmm3 ;
- mm1 = p0..p3
- psllw xmm3, 2 ;
-
- movq xmm5, QWORD PTR [esi + eax] ;
- mm4 = r1 p0..p7
- punpcklbw xmm5, xmm0 ;
- mm5 = r1 p0..p3
- paddusw xmm3, xmm5 ;
- mm3 += mm6
-
- ;
- thresholding
- movdqa xmm7, xmm1 ;
- mm7 = r0 p0..p3
- psubusw xmm7, xmm5 ;
- mm7 = r0 p0..p3 - r1 p0..p3
- psubusw xmm5, xmm1 ;
- mm5 = r1 p0..p3 - r0 p0..p3
- paddusw xmm7, xmm5 ;
- mm7 = abs(r0 p0..p3 - r1 p0..p3)
- pcmpgtw xmm7, xmm2
-
- movq xmm5, QWORD PTR [esi + 2*eax] ;
- mm4 = r2 p0..p7
- punpcklbw xmm5, xmm0 ;
- mm5 = r2 p0..p3
- paddusw xmm3, xmm5 ;
- mm3 += mm5
-
- ;
- thresholding
- movdqa xmm6, xmm1 ;
- mm6 = r0 p0..p3
- psubusw xmm6, xmm5 ;
- mm6 = r0 p0..p3 - r2 p0..p3
- psubusw xmm5, xmm1 ;
- mm5 = r2 p0..p3 - r2 p0..p3
- paddusw xmm6, xmm5 ;
- mm6 = abs(r0 p0..p3 - r2 p0..p3)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ;
- accumulate thresholds
-
-
- neg eax
- movq xmm5, QWORD PTR [esi+2*eax] ;
- mm4 = r-2 p0..p7
- punpcklbw xmm5, xmm0 ;
- mm5 = r-2 p0..p3
- paddusw xmm3, xmm5 ;
- mm3 += mm5
-
- ;
- thresholding
- movdqa xmm6, xmm1 ;
- mm6 = r0 p0..p3
- psubusw xmm6, xmm5 ;
- mm6 = p0..p3 - r-2 p0..p3
- psubusw xmm5, xmm1 ;
- mm5 = r-2 p0..p3 - p0..p3
- paddusw xmm6, xmm5 ;
- mm6 = abs(r0 p0..p3 - r-2 p0..p3)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ;
- accumulate thresholds
-
- movq xmm4, QWORD PTR [esi+eax] ;
- mm4 = r-1 p0..p7
- punpcklbw xmm4, xmm0 ;
- mm4 = r-1 p0..p3
- paddusw xmm3, xmm4 ;
- mm3 += mm5
-
- ;
- thresholding
- movdqa xmm6, xmm1 ;
- mm6 = r0 p0..p3
- psubusw xmm6, xmm4 ;
- mm6 = p0..p3 - r-2 p0..p3
- psubusw xmm4, xmm1 ;
- mm5 = r-1 p0..p3 - p0..p3
- paddusw xmm6, xmm4 ;
- mm6 = abs(r0 p0..p3 - r-1 p0..p3)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ;
- accumulate thresholds
-
-
- paddusw xmm3, rd42 ;
- mm3 += round value
- psraw xmm3, 3 ;
- mm3 /= 8
-
- pand xmm1, xmm7 ;
- mm1 select vals > thresh from source
- pandn xmm7, xmm3 ;
- mm7 select vals < thresh from blurred result
- paddusw xmm1, xmm7 ;
- combination
-
- packuswb xmm1, xmm0 ;
- pack to bytes
- movq QWORD PTR [edi], xmm1 ;
-
- neg eax ;
- pitch is positive
- add esi, 8
- add edi, 8
-
- add edx, 8
- cmp edx, cols
-
- jl nextcol
-
- // done with the all cols, start the across filtering in place
- sub esi, edx
- sub edi, edx
-
- xor edx, edx
- movq mm0, QWORD PTR [edi-8];
-
- acrossnextcol:
- movq xmm7, QWORD PTR [edi +edx -2]
- movd xmm4, DWORD PTR [edi +edx +6]
-
- pslldq xmm4, 8
- por xmm4, xmm7
-
- movdqa xmm3, xmm4
- psrldq xmm3, 2
- punpcklbw xmm3, xmm0 ;
- mm3 = p0..p3
- movdqa xmm1, xmm3 ;
- mm1 = p0..p3
- psllw xmm3, 2
-
-
- movdqa xmm5, xmm4
- psrldq xmm5, 3
- punpcklbw xmm5, xmm0 ;
- mm5 = p1..p4
- paddusw xmm3, xmm5 ;
- mm3 += mm6
-
- ;
- thresholding
- movdqa xmm7, xmm1 ;
- mm7 = p0..p3
- psubusw xmm7, xmm5 ;
- mm7 = p0..p3 - p1..p4
- psubusw xmm5, xmm1 ;
- mm5 = p1..p4 - p0..p3
- paddusw xmm7, xmm5 ;
- mm7 = abs(p0..p3 - p1..p4)
- pcmpgtw xmm7, xmm2
-
- movdqa xmm5, xmm4
- psrldq xmm5, 4
- punpcklbw xmm5, xmm0 ;
- mm5 = p2..p5
- paddusw xmm3, xmm5 ;
- mm3 += mm5
-
- ;
- thresholding
- movdqa xmm6, xmm1 ;
- mm6 = p0..p3
- psubusw xmm6, xmm5 ;
- mm6 = p0..p3 - p1..p4
- psubusw xmm5, xmm1 ;
- mm5 = p1..p4 - p0..p3
- paddusw xmm6, xmm5 ;
- mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ;
- accumulate thresholds
-
-
- movdqa xmm5, xmm4 ;
- mm5 = p-2..p5
- punpcklbw xmm5, xmm0 ;
- mm5 = p-2..p1
- paddusw xmm3, xmm5 ;
- mm3 += mm5
-
- ;
- thresholding
- movdqa xmm6, xmm1 ;
- mm6 = p0..p3
- psubusw xmm6, xmm5 ;
- mm6 = p0..p3 - p1..p4
- psubusw xmm5, xmm1 ;
- mm5 = p1..p4 - p0..p3
- paddusw xmm6, xmm5 ;
- mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ;
- accumulate thresholds
-
- psrldq xmm4, 1 ;
- mm4 = p-1..p5
- punpcklbw xmm4, xmm0 ;
- mm4 = p-1..p2
- paddusw xmm3, xmm4 ;
- mm3 += mm5
-
- ;
- thresholding
- movdqa xmm6, xmm1 ;
- mm6 = p0..p3
- psubusw xmm6, xmm4 ;
- mm6 = p0..p3 - p1..p4
- psubusw xmm4, xmm1 ;
- mm5 = p1..p4 - p0..p3
- paddusw xmm6, xmm4 ;
- mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ;
- accumulate thresholds
-
- paddusw xmm3, rd42 ;
- mm3 += round value
- psraw xmm3, 3 ;
- mm3 /= 8
-
- pand xmm1, xmm7 ;
- mm1 select vals > thresh from source
- pandn xmm7, xmm3 ;
- mm7 select vals < thresh from blurred result
- paddusw xmm1, xmm7 ;
- combination
-
- packuswb xmm1, xmm0 ;
- pack to bytes
- movq QWORD PTR [edi+edx-8], mm0 ;
- store previous four bytes
- movdq2q mm0, xmm1
-
- add edx, 8
- cmp edx, cols
- jl acrossnextcol;
-
- // last 8 pixels
- movq QWORD PTR [edi+edx-8], mm0
-
- // done with this rwo
- add esi, eax ;
- next line
- mov eax, dst_pixels_per_line ;
- destination pitch?
- add edi, eax ;
- next destination
- mov eax, src_pixels_per_line ;
- destination pitch?
-
- dec ecx ;
- decrement count
- jnz nextrow ;
- next row
- }
-}
-
-
-void vp8_mbpost_proc_down_mmx(unsigned char *dst, int pitch, int rows, int cols, int flimit)
-{
- int c, i;
- __declspec(align(16))
- int flimit2[2];
- __declspec(align(16))
- unsigned char d[16][8];
-
- flimit = vp8_q2mbl(flimit);
-
- for (i = 0; i < 2; i++)
- flimit2[i] = flimit;
-
- rows += 8;
-
- for (c = 0; c < cols; c += 4)
- {
- unsigned char *s = &dst[c];
-
- __asm
- {
- mov esi, s ;
- pxor mm0, mm0 ;
-
- mov eax, pitch ;
- neg eax // eax = -pitch
-
- lea esi, [esi + eax*8]; // edi = s[-pitch*8]
- neg eax
-
-
- pxor mm5, mm5
- pxor mm6, mm6 ;
-
- pxor mm7, mm7 ;
- mov edi, esi
-
- mov ecx, 15 ;
-
- loop_initvar:
- movd mm1, DWORD PTR [edi];
- punpcklbw mm1, mm0 ;
-
- paddw mm5, mm1 ;
- pmullw mm1, mm1 ;
-
- movq mm2, mm1 ;
- punpcklwd mm1, mm0 ;
-
- punpckhwd mm2, mm0 ;
- paddd mm6, mm1 ;
-
- paddd mm7, mm2 ;
- lea edi, [edi+eax] ;
-
- dec ecx
- jne loop_initvar
- //save the var and sum
- xor edx, edx
- loop_row:
- movd mm1, DWORD PTR [esi] // [s-pitch*8]
- movd mm2, DWORD PTR [edi] // [s+pitch*7]
-
- punpcklbw mm1, mm0
- punpcklbw mm2, mm0
-
- paddw mm5, mm2
- psubw mm5, mm1
-
- pmullw mm2, mm2
- movq mm4, mm2
-
- punpcklwd mm2, mm0
- punpckhwd mm4, mm0
-
- paddd mm6, mm2
- paddd mm7, mm4
-
- pmullw mm1, mm1
- movq mm2, mm1
-
- punpcklwd mm1, mm0
- psubd mm6, mm1
-
- punpckhwd mm2, mm0
- psubd mm7, mm2
-
-
- movq mm3, mm6
- pslld mm3, 4
-
- psubd mm3, mm6
- movq mm1, mm5
-
- movq mm4, mm5
- pmullw mm1, mm1
-
- pmulhw mm4, mm4
- movq mm2, mm1
-
- punpcklwd mm1, mm4
- punpckhwd mm2, mm4
-
- movq mm4, mm7
- pslld mm4, 4
-
- psubd mm4, mm7
-
- psubd mm3, mm1
- psubd mm4, mm2
-
- psubd mm3, flimit2
- psubd mm4, flimit2
-
- psrad mm3, 31
- psrad mm4, 31
-
- packssdw mm3, mm4
- packsswb mm3, mm0
-
- movd mm1, DWORD PTR [esi+eax*8]
-
- movq mm2, mm1
- punpcklbw mm1, mm0
-
- paddw mm1, mm5
- mov ecx, edx
-
- and ecx, 127
- movq mm4, vp8_rv[ecx*2]
-
- paddw mm1, mm4
- //paddw xmm1, eight8s
- psraw mm1, 4
-
- packuswb mm1, mm0
- pand mm1, mm3
-
- pandn mm3, mm2
- por mm1, mm3
-
- and ecx, 15
- movd DWORD PTR d[ecx*4], mm1
-
- mov ecx, edx
- sub ecx, 8
-
- and ecx, 15
- movd mm1, DWORD PTR d[ecx*4]
-
- movd [esi], mm1
- lea esi, [esi+eax]
-
- lea edi, [edi+eax]
- add edx, 1
-
- cmp edx, rows
- jl loop_row
-
- }
-
- }
-}
-
-void vp8_mbpost_proc_down_xmm(unsigned char *dst, int pitch, int rows, int cols, int flimit)
-{
- int c, i;
- __declspec(align(16))
- int flimit4[4];
- __declspec(align(16))
- unsigned char d[16][8];
-
- flimit = vp8_q2mbl(flimit);
-
- for (i = 0; i < 4; i++)
- flimit4[i] = flimit;
-
- rows += 8;
-
- for (c = 0; c < cols; c += 8)
- {
- unsigned char *s = &dst[c];
-
- __asm
- {
- mov esi, s ;
- pxor xmm0, xmm0 ;
-
- mov eax, pitch ;
- neg eax // eax = -pitch
-
- lea esi, [esi + eax*8]; // edi = s[-pitch*8]
- neg eax
-
-
- pxor xmm5, xmm5
- pxor xmm6, xmm6 ;
-
- pxor xmm7, xmm7 ;
- mov edi, esi
-
- mov ecx, 15 ;
-
- loop_initvar:
- movq xmm1, QWORD PTR [edi];
- punpcklbw xmm1, xmm0 ;
-
- paddw xmm5, xmm1 ;
- pmullw xmm1, xmm1 ;
-
- movdqa xmm2, xmm1 ;
- punpcklwd xmm1, xmm0 ;
-
- punpckhwd xmm2, xmm0 ;
- paddd xmm6, xmm1 ;
-
- paddd xmm7, xmm2 ;
- lea edi, [edi+eax] ;
-
- dec ecx
- jne loop_initvar
- //save the var and sum
- xor edx, edx
- loop_row:
- movq xmm1, QWORD PTR [esi] // [s-pitch*8]
- movq xmm2, QWORD PTR [edi] // [s+pitch*7]
-
- punpcklbw xmm1, xmm0
- punpcklbw xmm2, xmm0
-
- paddw xmm5, xmm2
- psubw xmm5, xmm1
-
- pmullw xmm2, xmm2
- movdqa xmm4, xmm2
-
- punpcklwd xmm2, xmm0
- punpckhwd xmm4, xmm0
-
- paddd xmm6, xmm2
- paddd xmm7, xmm4
-
- pmullw xmm1, xmm1
- movdqa xmm2, xmm1
-
- punpcklwd xmm1, xmm0
- psubd xmm6, xmm1
-
- punpckhwd xmm2, xmm0
- psubd xmm7, xmm2
-
-
- movdqa xmm3, xmm6
- pslld xmm3, 4
-
- psubd xmm3, xmm6
- movdqa xmm1, xmm5
-
- movdqa xmm4, xmm5
- pmullw xmm1, xmm1
-
- pmulhw xmm4, xmm4
- movdqa xmm2, xmm1
-
- punpcklwd xmm1, xmm4
- punpckhwd xmm2, xmm4
-
- movdqa xmm4, xmm7
- pslld xmm4, 4
-
- psubd xmm4, xmm7
-
- psubd xmm3, xmm1
- psubd xmm4, xmm2
-
- psubd xmm3, flimit4
- psubd xmm4, flimit4
-
- psrad xmm3, 31
- psrad xmm4, 31
-
- packssdw xmm3, xmm4
- packsswb xmm3, xmm0
-
- movq xmm1, QWORD PTR [esi+eax*8]
-
- movq xmm2, xmm1
- punpcklbw xmm1, xmm0
-
- paddw xmm1, xmm5
- mov ecx, edx
-
- and ecx, 127
- movdqu xmm4, vp8_rv[ecx*2]
-
- paddw xmm1, xmm4
- //paddw xmm1, eight8s
- psraw xmm1, 4
-
- packuswb xmm1, xmm0
- pand xmm1, xmm3
-
- pandn xmm3, xmm2
- por xmm1, xmm3
-
- and ecx, 15
- movq QWORD PTR d[ecx*8], xmm1
-
- mov ecx, edx
- sub ecx, 8
-
- and ecx, 15
- movq mm0, d[ecx*8]
-
- movq [esi], mm0
- lea esi, [esi+eax]
-
- lea edi, [edi+eax]
- add edx, 1
-
- cmp edx, rows
- jl loop_row
-
- }
-
- }
-}
-#if 0
-/****************************************************************************
- *
- * ROUTINE : plane_add_noise_wmt
- *
- * INPUTS : unsigned char *Start starting address of buffer to add gaussian
- * noise to
- * unsigned int Width width of plane
- * unsigned int Height height of plane
- * int Pitch distance between subsequent lines of frame
- * int q quantizer used to determine amount of noise
- * to add
- *
- * OUTPUTS : None.
- *
- * RETURNS : void.
- *
- * FUNCTION : adds gaussian noise to a plane of pixels
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8_plane_add_noise_wmt(unsigned char *Start, unsigned int Width, unsigned int Height, int Pitch, int q, int a)
-{
- unsigned int i;
-
- __declspec(align(16)) unsigned char blackclamp[16];
- __declspec(align(16)) unsigned char whiteclamp[16];
- __declspec(align(16)) unsigned char bothclamp[16];
- char char_dist[300];
- char Rand[2048];
- double sigma;
-// return;
- __asm emms
- sigma = a + .5 + .6 * (63 - q) / 63.0;
-
- // set up a lookup table of 256 entries that matches
- // a gaussian distribution with sigma determined by q.
- //
- {
- double i;
- int next, j;
-
- next = 0;
-
- for (i = -32; i < 32; i++)
- {
- double g = 256 * vp8_gaussian(sigma, 0, 1.0 * i);
- int a = (int)(g + .5);
-
- if (a)
- {
- for (j = 0; j < a; j++)
- {
- char_dist[next+j] = (char) i;
- }
-
- next = next + j;
- }
-
- }
-
- for (next = next; next < 256; next++)
- char_dist[next] = 0;
-
- }
-
- for (i = 0; i < 2048; i++)
- {
- Rand[i] = char_dist[rand() & 0xff];
- }
-
- for (i = 0; i < 16; i++)
- {
- blackclamp[i] = -char_dist[0];
- whiteclamp[i] = -char_dist[0];
- bothclamp[i] = -2 * char_dist[0];
- }
-
- for (i = 0; i < Height; i++)
- {
- unsigned char *Pos = Start + i * Pitch;
- char *Ref = Rand + (rand() & 0xff);
-
- __asm
- {
- mov ecx, [Width]
- mov esi, Pos
- mov edi, Ref
- xor eax, eax
-
- nextset:
- movdqu xmm1, [esi+eax] // get the source
-
- psubusb xmm1, blackclamp // clamp both sides so we don't outrange adding noise
- paddusb xmm1, bothclamp
- psubusb xmm1, whiteclamp
-
- movdqu xmm2, [edi+eax] // get the noise for this line
- paddb xmm1, xmm2 // add it in
- movdqu [esi+eax], xmm1 // store the result
-
- add eax, 16 // move to the next line
-
- cmp eax, ecx
- jl nextset
-
-
- }
-
- }
-}
-#endif
-__declspec(align(16))
-static const int four8s[4] = { 8, 8, 8, 8};
-void vp8_mbpost_proc_across_ip_xmm(unsigned char *src, int pitch, int rows, int cols, int flimit)
-{
- int r, i;
- __declspec(align(16))
- int flimit4[4];
- unsigned char *s = src;
- int sumsq;
- int sum;
-
-
- flimit = vp8_q2mbl(flimit);
- flimit4[0] =
- flimit4[1] =
- flimit4[2] =
- flimit4[3] = flimit;
-
- for (r = 0; r < rows; r++)
- {
-
-
- sumsq = 0;
- sum = 0;
-
- for (i = -8; i <= 6; i++)
- {
- sumsq += s[i] * s[i];
- sum += s[i];
- }
-
- __asm
- {
- mov eax, sumsq
- movd xmm7, eax
-
- mov eax, sum
- movd xmm6, eax
-
- mov esi, s
- xor ecx, ecx
-
- mov edx, cols
- add edx, 8
- pxor mm0, mm0
- pxor mm1, mm1
-
- pxor xmm0, xmm0
- nextcol4:
-
- movd xmm1, DWORD PTR [esi+ecx-8] // -8 -7 -6 -5
- movd xmm2, DWORD PTR [esi+ecx+7] // +7 +8 +9 +10
-
- punpcklbw xmm1, xmm0 // expanding
- punpcklbw xmm2, xmm0 // expanding
-
- punpcklwd xmm1, xmm0 // expanding to dwords
- punpcklwd xmm2, xmm0 // expanding to dwords
-
- psubd xmm2, xmm1 // 7--8 8--7 9--6 10--5
- paddd xmm1, xmm1 // -8*2 -7*2 -6*2 -5*2
-
- paddd xmm1, xmm2 // 7+-8 8+-7 9+-6 10+-5
- pmaddwd xmm1, xmm2 // squared of 7+-8 8+-7 9+-6 10+-5
-
- paddd xmm6, xmm2
- paddd xmm7, xmm1
-
- pshufd xmm6, xmm6, 0 // duplicate the last ones
- pshufd xmm7, xmm7, 0 // duplicate the last ones
-
- psrldq xmm1, 4 // 8--7 9--6 10--5 0000
- psrldq xmm2, 4 // 8--7 9--6 10--5 0000
-
- pshufd xmm3, xmm1, 3 // 0000 8--7 8--7 8--7 squared
- pshufd xmm4, xmm2, 3 // 0000 8--7 8--7 8--7 squared
-
- paddd xmm6, xmm4
- paddd xmm7, xmm3
-
- pshufd xmm3, xmm1, 01011111b // 0000 0000 9--6 9--6 squared
- pshufd xmm4, xmm2, 01011111b // 0000 0000 9--6 9--6 squared
-
- paddd xmm7, xmm3
- paddd xmm6, xmm4
-
- pshufd xmm3, xmm1, 10111111b // 0000 0000 8--7 8--7 squared
- pshufd xmm4, xmm2, 10111111b // 0000 0000 8--7 8--7 squared
-
- paddd xmm7, xmm3
- paddd xmm6, xmm4
-
- movdqa xmm3, xmm6
- pmaddwd xmm3, xmm3
-
- movdqa xmm5, xmm7
- pslld xmm5, 4
-
- psubd xmm5, xmm7
- psubd xmm5, xmm3
-
- psubd xmm5, flimit4
- psrad xmm5, 31
-
- packssdw xmm5, xmm0
- packsswb xmm5, xmm0
-
- movd xmm1, DWORD PTR [esi+ecx]
- movq xmm2, xmm1
-
- punpcklbw xmm1, xmm0
- punpcklwd xmm1, xmm0
-
- paddd xmm1, xmm6
- paddd xmm1, four8s
-
- psrad xmm1, 4
- packssdw xmm1, xmm0
-
- packuswb xmm1, xmm0
- pand xmm1, xmm5
-
- pandn xmm5, xmm2
- por xmm5, xmm1
-
- movd [esi+ecx-8], mm0
- movq mm0, mm1
-
- movdq2q mm1, xmm5
- psrldq xmm7, 12
-
- psrldq xmm6, 12
- add ecx, 4
-
- cmp ecx, edx
- jl nextcol4
-
- }
- s += pitch;
- }
-}
-
-#if 0
-
-/****************************************************************************
- *
- * ROUTINE : plane_add_noise_mmx
- *
- * INPUTS : unsigned char *Start starting address of buffer to add gaussian
- * noise to
- * unsigned int Width width of plane
- * unsigned int Height height of plane
- * int Pitch distance between subsequent lines of frame
- * int q quantizer used to determine amount of noise
- * to add
- *
- * OUTPUTS : None.
- *
- * RETURNS : void.
- *
- * FUNCTION : adds gaussian noise to a plane of pixels
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8_plane_add_noise_mmx(unsigned char *Start, unsigned int Width, unsigned int Height, int Pitch, int q, int a)
-{
- unsigned int i;
- int Pitch4 = Pitch * 4;
- const int noise_amount = 2;
- const int noise_adder = 2 * noise_amount + 1;
-
- __declspec(align(16)) unsigned char blackclamp[16];
- __declspec(align(16)) unsigned char whiteclamp[16];
- __declspec(align(16)) unsigned char bothclamp[16];
-
- char char_dist[300];
- char Rand[2048];
-
- double sigma;
- __asm emms
- sigma = a + .5 + .6 * (63 - q) / 63.0;
-
- // set up a lookup table of 256 entries that matches
- // a gaussian distribution with sigma determined by q.
- //
- {
- double i, sum = 0;
- int next, j;
-
- next = 0;
-
- for (i = -32; i < 32; i++)
- {
- int a = (int)(.5 + 256 * vp8_gaussian(sigma, 0, i));
-
- if (a)
- {
- for (j = 0; j < a; j++)
- {
- char_dist[next+j] = (char) i;
- }
-
- next = next + j;
- }
-
- }
-
- for (next = next; next < 256; next++)
- char_dist[next] = 0;
-
- }
-
- for (i = 0; i < 2048; i++)
- {
- Rand[i] = char_dist[rand() & 0xff];
- }
-
- for (i = 0; i < 16; i++)
- {
- blackclamp[i] = -char_dist[0];
- whiteclamp[i] = -char_dist[0];
- bothclamp[i] = -2 * char_dist[0];
- }
-
- for (i = 0; i < Height; i++)
- {
- unsigned char *Pos = Start + i * Pitch;
- char *Ref = Rand + (rand() & 0xff);
-
- __asm
- {
- mov ecx, [Width]
- mov esi, Pos
- mov edi, Ref
- xor eax, eax
-
- nextset:
- movq mm1, [esi+eax] // get the source
-
- psubusb mm1, blackclamp // clamp both sides so we don't outrange adding noise
- paddusb mm1, bothclamp
- psubusb mm1, whiteclamp
-
- movq mm2, [edi+eax] // get the noise for this line
- paddb mm1, mm2 // add it in
- movq [esi+eax], mm1 // store the result
-
- add eax, 8 // move to the next line
-
- cmp eax, ecx
- jl nextset
-
-
- }
-
- }
-}
-#else
-extern char an[8][64][3072];
-extern int cd[8][64];
-
-void vp8_plane_add_noise_mmx(unsigned char *Start, unsigned int Width, unsigned int Height, int Pitch, int q, int a)
-{
- unsigned int i;
- __declspec(align(16)) unsigned char blackclamp[16];
- __declspec(align(16)) unsigned char whiteclamp[16];
- __declspec(align(16)) unsigned char bothclamp[16];
-
-
- __asm emms
-
- for (i = 0; i < 16; i++)
- {
- blackclamp[i] = -cd[a][q];
- whiteclamp[i] = -cd[a][q];
- bothclamp[i] = -2 * cd[a][q];
- }
-
- for (i = 0; i < Height; i++)
- {
- unsigned char *Pos = Start + i * Pitch;
- char *Ref = an[a][q] + (rand() & 0xff);
-
- __asm
- {
- mov ecx, [Width]
- mov esi, Pos
- mov edi, Ref
- xor eax, eax
-
- nextset:
- movq mm1, [esi+eax] // get the source
-
- psubusb mm1, blackclamp // clamp both sides so we don't outrange adding noise
- paddusb mm1, bothclamp
- psubusb mm1, whiteclamp
-
- movq mm2, [edi+eax] // get the noise for this line
- paddb mm1, mm2 // add it in
- movq [esi+eax], mm1 // store the result
-
- add eax, 8 // move to the next line
-
- cmp eax, ecx
- jl nextset
- }
- }
-}
-
-
-void vp8_plane_add_noise_wmt(unsigned char *Start, unsigned int Width, unsigned int Height, int Pitch, int q, int a)
-{
- unsigned int i;
-
- __declspec(align(16)) unsigned char blackclamp[16];
- __declspec(align(16)) unsigned char whiteclamp[16];
- __declspec(align(16)) unsigned char bothclamp[16];
-
- __asm emms
-
- for (i = 0; i < 16; i++)
- {
- blackclamp[i] = -cd[a][q];
- whiteclamp[i] = -cd[a][q];
- bothclamp[i] = -2 * cd[a][q];
- }
-
- for (i = 0; i < Height; i++)
- {
- unsigned char *Pos = Start + i * Pitch;
- char *Ref = an[a][q] + (rand() & 0xff);
-
- __asm
- {
- mov ecx, [Width]
- mov esi, Pos
- mov edi, Ref
- xor eax, eax
-
- nextset:
- movdqu xmm1, [esi+eax] // get the source
-
- psubusb xmm1, blackclamp // clamp both sides so we don't outrange adding noise
- paddusb xmm1, bothclamp
- psubusb xmm1, whiteclamp
-
- movdqu xmm2, [edi+eax] // get the noise for this line
- paddb xmm1, xmm2 // add it in
- movdqu [esi+eax], xmm1 // store the result
-
- add eax, 16 // move to the next line
-
- cmp eax, ecx
- jl nextset
- }
- }
-}
-
-#endif
diff --git a/vp8/common/x86/postproc_sse2.asm b/vp8/common/x86/postproc_sse2.asm
deleted file mode 100644
index 30b4bf5..0000000
--- a/vp8/common/x86/postproc_sse2.asm
+++ /dev/null
@@ -1,695 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_post_proc_down_and_across_xmm
-;(
-; unsigned char *src_ptr,
-; unsigned char *dst_ptr,
-; int src_pixels_per_line,
-; int dst_pixels_per_line,
-; int rows,
-; int cols,
-; int flimit
-;)
-global sym(vp8_post_proc_down_and_across_xmm)
-sym(vp8_post_proc_down_and_across_xmm):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
-%if ABI_IS_32BIT=1 && CONFIG_PIC=1
- ALIGN_STACK 16, rax
- ; move the global rd onto the stack, since we don't have enough registers
- ; to do PIC addressing
- movdqa xmm0, [GLOBAL(rd42)]
- sub rsp, 16
- movdqa [rsp], xmm0
-%define RD42 [rsp]
-%else
-%define RD42 [GLOBAL(rd42)]
-%endif
-
-
- movd xmm2, dword ptr arg(6) ;flimit
- punpcklwd xmm2, xmm2
- punpckldq xmm2, xmm2
- punpcklqdq xmm2, xmm2
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(1) ;dst_ptr
-
- movsxd rcx, DWORD PTR arg(4) ;rows
- movsxd rax, DWORD PTR arg(2) ;src_pixels_per_line ; destination pitch?
- pxor xmm0, xmm0 ; mm0 = 00000000
-
-nextrow:
-
- xor rdx, rdx ; clear out rdx for use as loop counter
-nextcol:
- movq xmm3, QWORD PTR [rsi] ; mm4 = r0 p0..p7
- punpcklbw xmm3, xmm0 ; mm3 = p0..p3
- movdqa xmm1, xmm3 ; mm1 = p0..p3
- psllw xmm3, 2 ;
-
- movq xmm5, QWORD PTR [rsi + rax] ; mm4 = r1 p0..p7
- punpcklbw xmm5, xmm0 ; mm5 = r1 p0..p3
- paddusw xmm3, xmm5 ; mm3 += mm6
-
- ; thresholding
- movdqa xmm7, xmm1 ; mm7 = r0 p0..p3
- psubusw xmm7, xmm5 ; mm7 = r0 p0..p3 - r1 p0..p3
- psubusw xmm5, xmm1 ; mm5 = r1 p0..p3 - r0 p0..p3
- paddusw xmm7, xmm5 ; mm7 = abs(r0 p0..p3 - r1 p0..p3)
- pcmpgtw xmm7, xmm2
-
- movq xmm5, QWORD PTR [rsi + 2*rax] ; mm4 = r2 p0..p7
- punpcklbw xmm5, xmm0 ; mm5 = r2 p0..p3
- paddusw xmm3, xmm5 ; mm3 += mm5
-
- ; thresholding
- movdqa xmm6, xmm1 ; mm6 = r0 p0..p3
- psubusw xmm6, xmm5 ; mm6 = r0 p0..p3 - r2 p0..p3
- psubusw xmm5, xmm1 ; mm5 = r2 p0..p3 - r2 p0..p3
- paddusw xmm6, xmm5 ; mm6 = abs(r0 p0..p3 - r2 p0..p3)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ; accumulate thresholds
-
-
- neg rax
- movq xmm5, QWORD PTR [rsi+2*rax] ; mm4 = r-2 p0..p7
- punpcklbw xmm5, xmm0 ; mm5 = r-2 p0..p3
- paddusw xmm3, xmm5 ; mm3 += mm5
-
- ; thresholding
- movdqa xmm6, xmm1 ; mm6 = r0 p0..p3
- psubusw xmm6, xmm5 ; mm6 = p0..p3 - r-2 p0..p3
- psubusw xmm5, xmm1 ; mm5 = r-2 p0..p3 - p0..p3
- paddusw xmm6, xmm5 ; mm6 = abs(r0 p0..p3 - r-2 p0..p3)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ; accumulate thresholds
-
- movq xmm4, QWORD PTR [rsi+rax] ; mm4 = r-1 p0..p7
- punpcklbw xmm4, xmm0 ; mm4 = r-1 p0..p3
- paddusw xmm3, xmm4 ; mm3 += mm5
-
- ; thresholding
- movdqa xmm6, xmm1 ; mm6 = r0 p0..p3
- psubusw xmm6, xmm4 ; mm6 = p0..p3 - r-2 p0..p3
- psubusw xmm4, xmm1 ; mm5 = r-1 p0..p3 - p0..p3
- paddusw xmm6, xmm4 ; mm6 = abs(r0 p0..p3 - r-1 p0..p3)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ; accumulate thresholds
-
-
- paddusw xmm3, RD42 ; mm3 += round value
- psraw xmm3, 3 ; mm3 /= 8
-
- pand xmm1, xmm7 ; mm1 select vals > thresh from source
- pandn xmm7, xmm3 ; mm7 select vals < thresh from blurred result
- paddusw xmm1, xmm7 ; combination
-
- packuswb xmm1, xmm0 ; pack to bytes
- movq QWORD PTR [rdi], xmm1 ;
-
- neg rax ; pitch is positive
- add rsi, 8
- add rdi, 8
-
- add rdx, 8
- cmp edx, dword arg(5) ;cols
-
- jl nextcol
-
- ; done with the all cols, start the across filtering in place
- sub rsi, rdx
- sub rdi, rdx
-
- xor rdx, rdx
- movq mm0, QWORD PTR [rdi-8];
-
-acrossnextcol:
- movq xmm7, QWORD PTR [rdi +rdx -2]
- movd xmm4, DWORD PTR [rdi +rdx +6]
-
- pslldq xmm4, 8
- por xmm4, xmm7
-
- movdqa xmm3, xmm4
- psrldq xmm3, 2
- punpcklbw xmm3, xmm0 ; mm3 = p0..p3
- movdqa xmm1, xmm3 ; mm1 = p0..p3
- psllw xmm3, 2
-
-
- movdqa xmm5, xmm4
- psrldq xmm5, 3
- punpcklbw xmm5, xmm0 ; mm5 = p1..p4
- paddusw xmm3, xmm5 ; mm3 += mm6
-
- ; thresholding
- movdqa xmm7, xmm1 ; mm7 = p0..p3
- psubusw xmm7, xmm5 ; mm7 = p0..p3 - p1..p4
- psubusw xmm5, xmm1 ; mm5 = p1..p4 - p0..p3
- paddusw xmm7, xmm5 ; mm7 = abs(p0..p3 - p1..p4)
- pcmpgtw xmm7, xmm2
-
- movdqa xmm5, xmm4
- psrldq xmm5, 4
- punpcklbw xmm5, xmm0 ; mm5 = p2..p5
- paddusw xmm3, xmm5 ; mm3 += mm5
-
- ; thresholding
- movdqa xmm6, xmm1 ; mm6 = p0..p3
- psubusw xmm6, xmm5 ; mm6 = p0..p3 - p1..p4
- psubusw xmm5, xmm1 ; mm5 = p1..p4 - p0..p3
- paddusw xmm6, xmm5 ; mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ; accumulate thresholds
-
-
- movdqa xmm5, xmm4 ; mm5 = p-2..p5
- punpcklbw xmm5, xmm0 ; mm5 = p-2..p1
- paddusw xmm3, xmm5 ; mm3 += mm5
-
- ; thresholding
- movdqa xmm6, xmm1 ; mm6 = p0..p3
- psubusw xmm6, xmm5 ; mm6 = p0..p3 - p1..p4
- psubusw xmm5, xmm1 ; mm5 = p1..p4 - p0..p3
- paddusw xmm6, xmm5 ; mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ; accumulate thresholds
-
- psrldq xmm4, 1 ; mm4 = p-1..p5
- punpcklbw xmm4, xmm0 ; mm4 = p-1..p2
- paddusw xmm3, xmm4 ; mm3 += mm5
-
- ; thresholding
- movdqa xmm6, xmm1 ; mm6 = p0..p3
- psubusw xmm6, xmm4 ; mm6 = p0..p3 - p1..p4
- psubusw xmm4, xmm1 ; mm5 = p1..p4 - p0..p3
- paddusw xmm6, xmm4 ; mm6 = abs(p0..p3 - p1..p4)
- pcmpgtw xmm6, xmm2
- por xmm7, xmm6 ; accumulate thresholds
-
- paddusw xmm3, RD42 ; mm3 += round value
- psraw xmm3, 3 ; mm3 /= 8
-
- pand xmm1, xmm7 ; mm1 select vals > thresh from source
- pandn xmm7, xmm3 ; mm7 select vals < thresh from blurred result
- paddusw xmm1, xmm7 ; combination
-
- packuswb xmm1, xmm0 ; pack to bytes
- movq QWORD PTR [rdi+rdx-8], mm0 ; store previous four bytes
- movdq2q mm0, xmm1
-
- add rdx, 8
- cmp edx, dword arg(5) ;cols
- jl acrossnextcol;
-
- ; last 8 pixels
- movq QWORD PTR [rdi+rdx-8], mm0
-
- ; done with this rwo
- add rsi,rax ; next line
- mov eax, dword arg(3) ;dst_pixels_per_line ; destination pitch?
- add rdi,rax ; next destination
- mov eax, dword arg(2) ;src_pixels_per_line ; destination pitch?
-
- dec rcx ; decrement count
- jnz nextrow ; next row
-
-%if ABI_IS_32BIT=1 && CONFIG_PIC=1
- add rsp,16
- pop rsp
-%endif
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-%undef RD42
-
-
-;void vp8_mbpost_proc_down_xmm(unsigned char *dst,
-; int pitch, int rows, int cols,int flimit)
-extern sym(vp8_rv)
-global sym(vp8_mbpost_proc_down_xmm)
-sym(vp8_mbpost_proc_down_xmm):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 128+16
-
- ; unsigned char d[16][8] at [rsp]
- ; create flimit2 at [rsp+128]
- mov eax, dword ptr arg(4) ;flimit
- mov [rsp+128], eax
- mov [rsp+128+4], eax
- mov [rsp+128+8], eax
- mov [rsp+128+12], eax
-%define flimit4 [rsp+128]
-
-%if ABI_IS_32BIT=0
- lea r8, [GLOBAL(sym(vp8_rv))]
-%endif
-
- ;rows +=8;
- add dword arg(2), 8
-
- ;for(c=0; c<cols; c+=8)
-loop_col:
- mov rsi, arg(0) ; s
- pxor xmm0, xmm0 ;
-
- movsxd rax, dword ptr arg(1) ;pitch ;
- neg rax ; rax = -pitch
-
- lea rsi, [rsi + rax*8]; ; rdi = s[-pitch*8]
- neg rax
-
-
- pxor xmm5, xmm5
- pxor xmm6, xmm6 ;
-
- pxor xmm7, xmm7 ;
- mov rdi, rsi
-
- mov rcx, 15 ;
-
-loop_initvar:
- movq xmm1, QWORD PTR [rdi];
- punpcklbw xmm1, xmm0 ;
-
- paddw xmm5, xmm1 ;
- pmullw xmm1, xmm1 ;
-
- movdqa xmm2, xmm1 ;
- punpcklwd xmm1, xmm0 ;
-
- punpckhwd xmm2, xmm0 ;
- paddd xmm6, xmm1 ;
-
- paddd xmm7, xmm2 ;
- lea rdi, [rdi+rax] ;
-
- dec rcx
- jne loop_initvar
- ;save the var and sum
- xor rdx, rdx
-loop_row:
- movq xmm1, QWORD PTR [rsi] ; [s-pitch*8]
- movq xmm2, QWORD PTR [rdi] ; [s+pitch*7]
-
- punpcklbw xmm1, xmm0
- punpcklbw xmm2, xmm0
-
- paddw xmm5, xmm2
- psubw xmm5, xmm1
-
- pmullw xmm2, xmm2
- movdqa xmm4, xmm2
-
- punpcklwd xmm2, xmm0
- punpckhwd xmm4, xmm0
-
- paddd xmm6, xmm2
- paddd xmm7, xmm4
-
- pmullw xmm1, xmm1
- movdqa xmm2, xmm1
-
- punpcklwd xmm1, xmm0
- psubd xmm6, xmm1
-
- punpckhwd xmm2, xmm0
- psubd xmm7, xmm2
-
-
- movdqa xmm3, xmm6
- pslld xmm3, 4
-
- psubd xmm3, xmm6
- movdqa xmm1, xmm5
-
- movdqa xmm4, xmm5
- pmullw xmm1, xmm1
-
- pmulhw xmm4, xmm4
- movdqa xmm2, xmm1
-
- punpcklwd xmm1, xmm4
- punpckhwd xmm2, xmm4
-
- movdqa xmm4, xmm7
- pslld xmm4, 4
-
- psubd xmm4, xmm7
-
- psubd xmm3, xmm1
- psubd xmm4, xmm2
-
- psubd xmm3, flimit4
- psubd xmm4, flimit4
-
- psrad xmm3, 31
- psrad xmm4, 31
-
- packssdw xmm3, xmm4
- packsswb xmm3, xmm0
-
- movq xmm1, QWORD PTR [rsi+rax*8]
-
- movq xmm2, xmm1
- punpcklbw xmm1, xmm0
-
- paddw xmm1, xmm5
- mov rcx, rdx
-
- and rcx, 127
-%if ABI_IS_32BIT=1 && CONFIG_PIC=1
- push rax
- lea rax, [GLOBAL(sym(vp8_rv))]
- movdqu xmm4, [rax + rcx*2] ;vp8_rv[rcx*2]
- pop rax
-%elif ABI_IS_32BIT=0
- movdqu xmm4, [r8 + rcx*2] ;vp8_rv[rcx*2]
-%else
- movdqu xmm4, [sym(vp8_rv) + rcx*2]
-%endif
-
- paddw xmm1, xmm4
- ;paddw xmm1, eight8s
- psraw xmm1, 4
-
- packuswb xmm1, xmm0
- pand xmm1, xmm3
-
- pandn xmm3, xmm2
- por xmm1, xmm3
-
- and rcx, 15
- movq QWORD PTR [rsp + rcx*8], xmm1 ;d[rcx*8]
-
- mov rcx, rdx
- sub rcx, 8
-
- and rcx, 15
- movq mm0, [rsp + rcx*8] ;d[rcx*8]
-
- movq [rsi], mm0
- lea rsi, [rsi+rax]
-
- lea rdi, [rdi+rax]
- add rdx, 1
-
- cmp edx, dword arg(2) ;rows
- jl loop_row
-
- add dword arg(0), 8 ; s += 8
- sub dword arg(3), 8 ; cols -= 8
- cmp dword arg(3), 0
- jg loop_col
-
- add rsp, 128+16
- pop rsp
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-%undef flimit4
-
-
-;void vp8_mbpost_proc_across_ip_xmm(unsigned char *src,
-; int pitch, int rows, int cols,int flimit)
-global sym(vp8_mbpost_proc_across_ip_xmm)
-sym(vp8_mbpost_proc_across_ip_xmm):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 16
-
- ; create flimit4 at [rsp]
- mov eax, dword ptr arg(4) ;flimit
- mov [rsp], eax
- mov [rsp+4], eax
- mov [rsp+8], eax
- mov [rsp+12], eax
-%define flimit4 [rsp]
-
-
- ;for(r=0;r<rows;r++)
-ip_row_loop:
-
- xor rdx, rdx ;sumsq=0;
- xor rcx, rcx ;sum=0;
- mov rsi, arg(0); s
- mov rdi, -8
-ip_var_loop:
- ;for(i=-8;i<=6;i++)
- ;{
- ; sumsq += s[i]*s[i];
- ; sum += s[i];
- ;}
- movzx eax, byte [rsi+rdi]
- add ecx, eax
- mul al
- add edx, eax
- add rdi, 1
- cmp rdi, 6
- jle ip_var_loop
-
-
- ;mov rax, sumsq
- ;movd xmm7, rax
- movd xmm7, edx
-
- ;mov rax, sum
- ;movd xmm6, rax
- movd xmm6, ecx
-
- mov rsi, arg(0) ;s
- xor rcx, rcx
-
- movsxd rdx, dword arg(3) ;cols
- add rdx, 8
- pxor mm0, mm0
- pxor mm1, mm1
-
- pxor xmm0, xmm0
-nextcol4:
-
- movd xmm1, DWORD PTR [rsi+rcx-8] ; -8 -7 -6 -5
- movd xmm2, DWORD PTR [rsi+rcx+7] ; +7 +8 +9 +10
-
- punpcklbw xmm1, xmm0 ; expanding
- punpcklbw xmm2, xmm0 ; expanding
-
- punpcklwd xmm1, xmm0 ; expanding to dwords
- punpcklwd xmm2, xmm0 ; expanding to dwords
-
- psubd xmm2, xmm1 ; 7--8 8--7 9--6 10--5
- paddd xmm1, xmm1 ; -8*2 -7*2 -6*2 -5*2
-
- paddd xmm1, xmm2 ; 7+-8 8+-7 9+-6 10+-5
- pmaddwd xmm1, xmm2 ; squared of 7+-8 8+-7 9+-6 10+-5
-
- paddd xmm6, xmm2
- paddd xmm7, xmm1
-
- pshufd xmm6, xmm6, 0 ; duplicate the last ones
- pshufd xmm7, xmm7, 0 ; duplicate the last ones
-
- psrldq xmm1, 4 ; 8--7 9--6 10--5 0000
- psrldq xmm2, 4 ; 8--7 9--6 10--5 0000
-
- pshufd xmm3, xmm1, 3 ; 0000 8--7 8--7 8--7 squared
- pshufd xmm4, xmm2, 3 ; 0000 8--7 8--7 8--7 squared
-
- paddd xmm6, xmm4
- paddd xmm7, xmm3
-
- pshufd xmm3, xmm1, 01011111b ; 0000 0000 9--6 9--6 squared
- pshufd xmm4, xmm2, 01011111b ; 0000 0000 9--6 9--6 squared
-
- paddd xmm7, xmm3
- paddd xmm6, xmm4
-
- pshufd xmm3, xmm1, 10111111b ; 0000 0000 8--7 8--7 squared
- pshufd xmm4, xmm2, 10111111b ; 0000 0000 8--7 8--7 squared
-
- paddd xmm7, xmm3
- paddd xmm6, xmm4
-
- movdqa xmm3, xmm6
- pmaddwd xmm3, xmm3
-
- movdqa xmm5, xmm7
- pslld xmm5, 4
-
- psubd xmm5, xmm7
- psubd xmm5, xmm3
-
- psubd xmm5, flimit4
- psrad xmm5, 31
-
- packssdw xmm5, xmm0
- packsswb xmm5, xmm0
-
- movd xmm1, DWORD PTR [rsi+rcx]
- movq xmm2, xmm1
-
- punpcklbw xmm1, xmm0
- punpcklwd xmm1, xmm0
-
- paddd xmm1, xmm6
- paddd xmm1, [GLOBAL(four8s)]
-
- psrad xmm1, 4
- packssdw xmm1, xmm0
-
- packuswb xmm1, xmm0
- pand xmm1, xmm5
-
- pandn xmm5, xmm2
- por xmm5, xmm1
-
- movd [rsi+rcx-8], mm0
- movq mm0, mm1
-
- movdq2q mm1, xmm5
- psrldq xmm7, 12
-
- psrldq xmm6, 12
- add rcx, 4
-
- cmp rcx, rdx
- jl nextcol4
-
- ;s+=pitch;
- movsxd rax, dword arg(1)
- add arg(0), rax
-
- sub dword arg(2), 1 ;rows-=1
- cmp dword arg(2), 0
- jg ip_row_loop
-
- add rsp, 16
- pop rsp
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-%undef flimit4
-
-
-;void vp8_plane_add_noise_wmt (unsigned char *Start, unsigned char *noise,
-; unsigned char blackclamp[16],
-; unsigned char whiteclamp[16],
-; unsigned char bothclamp[16],
-; unsigned int Width, unsigned int Height, int Pitch)
-extern sym(rand)
-global sym(vp8_plane_add_noise_wmt)
-sym(vp8_plane_add_noise_wmt):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 8
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
-addnoise_loop:
- call sym(rand) WRT_PLT
- mov rcx, arg(1) ;noise
- and rax, 0xff
- add rcx, rax
-
- ; we rely on the fact that the clamping vectors are stored contiguously
- ; in black/white/both order. Note that we have to reload this here because
- ; rdx could be trashed by rand()
- mov rdx, arg(2) ; blackclamp
-
-
- mov rdi, rcx
- movsxd rcx, dword arg(5) ;[Width]
- mov rsi, arg(0) ;Pos
- xor rax,rax
-
-addnoise_nextset:
- movdqu xmm1,[rsi+rax] ; get the source
-
- psubusb xmm1, [rdx] ;blackclamp ; clamp both sides so we don't outrange adding noise
- paddusb xmm1, [rdx+32] ;bothclamp
- psubusb xmm1, [rdx+16] ;whiteclamp
-
- movdqu xmm2,[rdi+rax] ; get the noise for this line
- paddb xmm1,xmm2 ; add it in
- movdqu [rsi+rax],xmm1 ; store the result
-
- add rax,16 ; move to the next line
-
- cmp rax, rcx
- jl addnoise_nextset
-
- movsxd rax, dword arg(7) ; Pitch
- add arg(0), rax ; Start += Pitch
- sub dword arg(6), 1 ; Height -= 1
- jg addnoise_loop
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-SECTION_RODATA
-align 16
-rd42:
- times 8 dw 0x04
-four8s:
- times 4 dd 8
diff --git a/vp8/common/x86/postproc_x86.h b/vp8/common/x86/postproc_x86.h
deleted file mode 100644
index 899dd2f..0000000
--- a/vp8/common/x86/postproc_x86.h
+++ /dev/null
@@ -1,64 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef POSTPROC_X86_H
-#define POSTPROC_X86_H
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-
-#if HAVE_MMX
-extern prototype_postproc_inplace(vp8_mbpost_proc_down_mmx);
-extern prototype_postproc(vp8_post_proc_down_and_across_mmx);
-extern prototype_postproc_addnoise(vp8_plane_add_noise_mmx);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_postproc_down
-#define vp8_postproc_down vp8_mbpost_proc_down_mmx
-
-#undef vp8_postproc_downacross
-#define vp8_postproc_downacross vp8_post_proc_down_and_across_mmx
-
-#undef vp8_postproc_addnoise
-#define vp8_postproc_addnoise vp8_plane_add_noise_mmx
-
-#endif
-#endif
-
-
-#if HAVE_SSE2
-extern prototype_postproc_inplace(vp8_mbpost_proc_down_xmm);
-extern prototype_postproc_inplace(vp8_mbpost_proc_across_ip_xmm);
-extern prototype_postproc(vp8_post_proc_down_and_across_xmm);
-extern prototype_postproc_addnoise(vp8_plane_add_noise_wmt);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_postproc_down
-#define vp8_postproc_down vp8_mbpost_proc_down_xmm
-
-#undef vp8_postproc_across
-#define vp8_postproc_across vp8_mbpost_proc_across_ip_xmm
-
-#undef vp8_postproc_downacross
-#define vp8_postproc_downacross vp8_post_proc_down_and_across_xmm
-
-#undef vp8_postproc_addnoise
-#define vp8_postproc_addnoise vp8_plane_add_noise_wmt
-
-
-#endif
-#endif
-
-#endif
diff --git a/vp8/common/x86/recon_mmx.asm b/vp8/common/x86/recon_mmx.asm
deleted file mode 100644
index e7211fc..0000000
--- a/vp8/common/x86/recon_mmx.asm
+++ /dev/null
@@ -1,321 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-;void vp8_recon_b_mmx(unsigned char *s, short *q, unsigned char *d, int stride)
-global sym(vp8_recon_b_mmx)
-sym(vp8_recon_b_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;s
- mov rdi, arg(2) ;d
- mov rdx, arg(1) ;q
- movsxd rax, dword ptr arg(3) ;stride
- pxor mm0, mm0
-
- movd mm1, [rsi]
- punpcklbw mm1, mm0
- paddsw mm1, [rdx]
- packuswb mm1, mm0 ; pack and unpack to saturate
- movd [rdi], mm1
-
- movd mm2, [rsi+16]
- punpcklbw mm2, mm0
- paddsw mm2, [rdx+32]
- packuswb mm2, mm0 ; pack and unpack to saturate
- movd [rdi+rax], mm2
-
- movd mm3, [rsi+32]
- punpcklbw mm3, mm0
- paddsw mm3, [rdx+64]
- packuswb mm3, mm0 ; pack and unpack to saturate
- movd [rdi+2*rax], mm3
-
- add rdi, rax
- movd mm4, [rsi+48]
- punpcklbw mm4, mm0
- paddsw mm4, [rdx+96]
- packuswb mm4, mm0 ; pack and unpack to saturate
- movd [rdi+2*rax], mm4
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void copy_mem8x8_mmx(
-; unsigned char *src,
-; int src_stride,
-; unsigned char *dst,
-; int dst_stride
-; )
-global sym(vp8_copy_mem8x8_mmx)
-sym(vp8_copy_mem8x8_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src;
- movq mm0, [rsi]
-
- movsxd rax, dword ptr arg(1) ;src_stride;
- mov rdi, arg(2) ;dst;
-
- movq mm1, [rsi+rax]
- movq mm2, [rsi+rax*2]
-
- movsxd rcx, dword ptr arg(3) ;dst_stride
- lea rsi, [rsi+rax*2]
-
- movq [rdi], mm0
- add rsi, rax
-
- movq [rdi+rcx], mm1
- movq [rdi+rcx*2], mm2
-
-
- lea rdi, [rdi+rcx*2]
- movq mm3, [rsi]
-
- add rdi, rcx
- movq mm4, [rsi+rax]
-
- movq mm5, [rsi+rax*2]
- movq [rdi], mm3
-
- lea rsi, [rsi+rax*2]
- movq [rdi+rcx], mm4
-
- movq [rdi+rcx*2], mm5
- lea rdi, [rdi+rcx*2]
-
- movq mm0, [rsi+rax]
- movq mm1, [rsi+rax*2]
-
- movq [rdi+rcx], mm0
- movq [rdi+rcx*2],mm1
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void copy_mem8x4_mmx(
-; unsigned char *src,
-; int src_stride,
-; unsigned char *dst,
-; int dst_stride
-; )
-global sym(vp8_copy_mem8x4_mmx)
-sym(vp8_copy_mem8x4_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src;
- movq mm0, [rsi]
-
- movsxd rax, dword ptr arg(1) ;src_stride;
- mov rdi, arg(2) ;dst;
-
- movq mm1, [rsi+rax]
- movq mm2, [rsi+rax*2]
-
- movsxd rcx, dword ptr arg(3) ;dst_stride
- lea rsi, [rsi+rax*2]
-
- movq [rdi], mm0
- movq [rdi+rcx], mm1
-
- movq [rdi+rcx*2], mm2
- lea rdi, [rdi+rcx*2]
-
- movq mm3, [rsi+rax]
- movq [rdi+rcx], mm3
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void copy_mem16x16_mmx(
-; unsigned char *src,
-; int src_stride,
-; unsigned char *dst,
-; int dst_stride
-; )
-global sym(vp8_copy_mem16x16_mmx)
-sym(vp8_copy_mem16x16_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src;
- movsxd rax, dword ptr arg(1) ;src_stride;
-
- mov rdi, arg(2) ;dst;
- movsxd rcx, dword ptr arg(3) ;dst_stride
-
- movq mm0, [rsi]
- movq mm3, [rsi+8];
-
- movq mm1, [rsi+rax]
- movq mm4, [rsi+rax+8]
-
- movq mm2, [rsi+rax*2]
- movq mm5, [rsi+rax*2+8]
-
- lea rsi, [rsi+rax*2]
- add rsi, rax
-
- movq [rdi], mm0
- movq [rdi+8], mm3
-
- movq [rdi+rcx], mm1
- movq [rdi+rcx+8], mm4
-
- movq [rdi+rcx*2], mm2
- movq [rdi+rcx*2+8], mm5
-
- lea rdi, [rdi+rcx*2]
- add rdi, rcx
-
- movq mm0, [rsi]
- movq mm3, [rsi+8];
-
- movq mm1, [rsi+rax]
- movq mm4, [rsi+rax+8]
-
- movq mm2, [rsi+rax*2]
- movq mm5, [rsi+rax*2+8]
-
- lea rsi, [rsi+rax*2]
- add rsi, rax
-
- movq [rdi], mm0
- movq [rdi+8], mm3
-
- movq [rdi+rcx], mm1
- movq [rdi+rcx+8], mm4
-
- movq [rdi+rcx*2], mm2
- movq [rdi+rcx*2+8], mm5
-
- lea rdi, [rdi+rcx*2]
- add rdi, rcx
-
- movq mm0, [rsi]
- movq mm3, [rsi+8];
-
- movq mm1, [rsi+rax]
- movq mm4, [rsi+rax+8]
-
- movq mm2, [rsi+rax*2]
- movq mm5, [rsi+rax*2+8]
-
- lea rsi, [rsi+rax*2]
- add rsi, rax
-
- movq [rdi], mm0
- movq [rdi+8], mm3
-
- movq [rdi+rcx], mm1
- movq [rdi+rcx+8], mm4
-
- movq [rdi+rcx*2], mm2
- movq [rdi+rcx*2+8], mm5
-
- lea rdi, [rdi+rcx*2]
- add rdi, rcx
-
- movq mm0, [rsi]
- movq mm3, [rsi+8];
-
- movq mm1, [rsi+rax]
- movq mm4, [rsi+rax+8]
-
- movq mm2, [rsi+rax*2]
- movq mm5, [rsi+rax*2+8]
-
- lea rsi, [rsi+rax*2]
- add rsi, rax
-
- movq [rdi], mm0
- movq [rdi+8], mm3
-
- movq [rdi+rcx], mm1
- movq [rdi+rcx+8], mm4
-
- movq [rdi+rcx*2], mm2
- movq [rdi+rcx*2+8], mm5
-
- lea rdi, [rdi+rcx*2]
- add rdi, rcx
-
- movq mm0, [rsi]
- movq mm3, [rsi+8];
-
- movq mm1, [rsi+rax]
- movq mm4, [rsi+rax+8]
-
- movq mm2, [rsi+rax*2]
- movq mm5, [rsi+rax*2+8]
-
- lea rsi, [rsi+rax*2]
- add rsi, rax
-
- movq [rdi], mm0
- movq [rdi+8], mm3
-
- movq [rdi+rcx], mm1
- movq [rdi+rcx+8], mm4
-
- movq [rdi+rcx*2], mm2
- movq [rdi+rcx*2+8], mm5
-
- lea rdi, [rdi+rcx*2]
- add rdi, rcx
-
- movq mm0, [rsi]
- movq mm3, [rsi+8];
-
- movq [rdi], mm0
- movq [rdi+8], mm3
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/common/x86/recon_sse2.asm b/vp8/common/x86/recon_sse2.asm
deleted file mode 100644
index 4ad3973..0000000
--- a/vp8/common/x86/recon_sse2.asm
+++ /dev/null
@@ -1,231 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-;void vp8_recon2b_sse2(unsigned char *s, short *q, unsigned char *d, int stride)
-global sym(vp8_recon2b_sse2)
-sym(vp8_recon2b_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;s
- mov rdi, arg(2) ;d
- mov rdx, arg(1) ;q
- movsxd rax, dword ptr arg(3) ;stride
- pxor xmm0, xmm0
-
- movq xmm1, MMWORD PTR [rsi]
- punpcklbw xmm1, xmm0
- paddsw xmm1, XMMWORD PTR [rdx]
- packuswb xmm1, xmm0 ; pack and unpack to saturate
- movq MMWORD PTR [rdi], xmm1
-
-
- movq xmm2, MMWORD PTR [rsi+8]
- punpcklbw xmm2, xmm0
- paddsw xmm2, XMMWORD PTR [rdx+16]
- packuswb xmm2, xmm0 ; pack and unpack to saturate
- movq MMWORD PTR [rdi+rax], xmm2
-
-
- movq xmm3, MMWORD PTR [rsi+16]
- punpcklbw xmm3, xmm0
- paddsw xmm3, XMMWORD PTR [rdx+32]
- packuswb xmm3, xmm0 ; pack and unpack to saturate
- movq MMWORD PTR [rdi+rax*2], xmm3
-
- add rdi, rax
- movq xmm4, MMWORD PTR [rsi+24]
- punpcklbw xmm4, xmm0
- paddsw xmm4, XMMWORD PTR [rdx+48]
- packuswb xmm4, xmm0 ; pack and unpack to saturate
- movq MMWORD PTR [rdi+rax*2], xmm4
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_recon4b_sse2(unsigned char *s, short *q, unsigned char *d, int stride)
-global sym(vp8_recon4b_sse2)
-sym(vp8_recon4b_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- SAVE_XMM
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;s
- mov rdi, arg(2) ;d
- mov rdx, arg(1) ;q
- movsxd rax, dword ptr arg(3) ;stride
- pxor xmm0, xmm0
-
- movdqa xmm1, XMMWORD PTR [rsi]
- movdqa xmm5, xmm1
- punpcklbw xmm1, xmm0
- punpckhbw xmm5, xmm0
- paddsw xmm1, XMMWORD PTR [rdx]
- paddsw xmm5, XMMWORD PTR [rdx+16]
- packuswb xmm1, xmm5 ; pack and unpack to saturate
- movdqa XMMWORD PTR [rdi], xmm1
-
-
- movdqa xmm2, XMMWORD PTR [rsi+16]
- movdqa xmm6, xmm2
- punpcklbw xmm2, xmm0
- punpckhbw xmm6, xmm0
- paddsw xmm2, XMMWORD PTR [rdx+32]
- paddsw xmm6, XMMWORD PTR [rdx+48]
- packuswb xmm2, xmm6 ; pack and unpack to saturate
- movdqa XMMWORD PTR [rdi+rax], xmm2
-
-
- movdqa xmm3, XMMWORD PTR [rsi+32]
- movdqa xmm7, xmm3
- punpcklbw xmm3, xmm0
- punpckhbw xmm7, xmm0
- paddsw xmm3, XMMWORD PTR [rdx+64]
- paddsw xmm7, XMMWORD PTR [rdx+80]
- packuswb xmm3, xmm7 ; pack and unpack to saturate
- movdqa XMMWORD PTR [rdi+rax*2], xmm3
-
- add rdi, rax
- movdqa xmm4, XMMWORD PTR [rsi+48]
- movdqa xmm5, xmm4
- punpcklbw xmm4, xmm0
- punpckhbw xmm5, xmm0
- paddsw xmm4, XMMWORD PTR [rdx+96]
- paddsw xmm5, XMMWORD PTR [rdx+112]
- packuswb xmm4, xmm5 ; pack and unpack to saturate
- movdqa XMMWORD PTR [rdi+rax*2], xmm4
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void copy_mem16x16_sse2(
-; unsigned char *src,
-; int src_stride,
-; unsigned char *dst,
-; int dst_stride
-; )
-global sym(vp8_copy_mem16x16_sse2)
-sym(vp8_copy_mem16x16_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src;
- movdqu xmm0, [rsi]
-
- movsxd rax, dword ptr arg(1) ;src_stride;
- mov rdi, arg(2) ;dst;
-
- movdqu xmm1, [rsi+rax]
- movdqu xmm2, [rsi+rax*2]
-
- movsxd rcx, dword ptr arg(3) ;dst_stride
- lea rsi, [rsi+rax*2]
-
- movdqa [rdi], xmm0
- add rsi, rax
-
- movdqa [rdi+rcx], xmm1
- movdqa [rdi+rcx*2],xmm2
-
- lea rdi, [rdi+rcx*2]
- movdqu xmm3, [rsi]
-
- add rdi, rcx
- movdqu xmm4, [rsi+rax]
-
- movdqu xmm5, [rsi+rax*2]
- lea rsi, [rsi+rax*2]
-
- movdqa [rdi], xmm3
- add rsi, rax
-
- movdqa [rdi+rcx], xmm4
- movdqa [rdi+rcx*2],xmm5
-
- lea rdi, [rdi+rcx*2]
- movdqu xmm0, [rsi]
-
- add rdi, rcx
- movdqu xmm1, [rsi+rax]
-
- movdqu xmm2, [rsi+rax*2]
- lea rsi, [rsi+rax*2]
-
- movdqa [rdi], xmm0
- add rsi, rax
-
- movdqa [rdi+rcx], xmm1
-
- movdqa [rdi+rcx*2], xmm2
- movdqu xmm3, [rsi]
-
- movdqu xmm4, [rsi+rax]
- lea rdi, [rdi+rcx*2]
-
- add rdi, rcx
- movdqu xmm5, [rsi+rax*2]
-
- lea rsi, [rsi+rax*2]
- movdqa [rdi], xmm3
-
- add rsi, rax
- movdqa [rdi+rcx], xmm4
-
- movdqa [rdi+rcx*2],xmm5
- movdqu xmm0, [rsi]
-
- lea rdi, [rdi+rcx*2]
- movdqu xmm1, [rsi+rax]
-
- add rdi, rcx
- movdqu xmm2, [rsi+rax*2]
-
- lea rsi, [rsi+rax*2]
- movdqa [rdi], xmm0
-
- movdqa [rdi+rcx], xmm1
- movdqa [rdi+rcx*2],xmm2
-
- movdqu xmm3, [rsi+rax]
- lea rdi, [rdi+rcx*2]
-
- movdqa [rdi+rcx], xmm3
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/common/x86/recon_x86.h b/vp8/common/x86/recon_x86.h
deleted file mode 100644
index 40ee65a..0000000
--- a/vp8/common/x86/recon_x86.h
+++ /dev/null
@@ -1,62 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef RECON_X86_H
-#define RECON_X86_H
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-
-#if HAVE_MMX
-extern prototype_recon_block(vp8_recon_b_mmx);
-extern prototype_copy_block(vp8_copy_mem8x8_mmx);
-extern prototype_copy_block(vp8_copy_mem8x4_mmx);
-extern prototype_copy_block(vp8_copy_mem16x16_mmx);
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_recon_recon
-#define vp8_recon_recon vp8_recon_b_mmx
-
-#undef vp8_recon_copy8x8
-#define vp8_recon_copy8x8 vp8_copy_mem8x8_mmx
-
-#undef vp8_recon_copy8x4
-#define vp8_recon_copy8x4 vp8_copy_mem8x4_mmx
-
-#undef vp8_recon_copy16x16
-#define vp8_recon_copy16x16 vp8_copy_mem16x16_mmx
-
-#endif
-#endif
-
-#if HAVE_SSE2
-extern prototype_recon_block(vp8_recon2b_sse2);
-extern prototype_recon_block(vp8_recon4b_sse2);
-extern prototype_copy_block(vp8_copy_mem16x16_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_recon_recon2
-#define vp8_recon_recon2 vp8_recon2b_sse2
-
-#undef vp8_recon_recon4
-#define vp8_recon_recon4 vp8_recon4b_sse2
-
-#undef vp8_recon_copy16x16
-#define vp8_recon_copy16x16 vp8_copy_mem16x16_sse2
-
-#endif
-#endif
-#endif
diff --git a/vp8/common/x86/subpixel_mmx.asm b/vp8/common/x86/subpixel_mmx.asm
deleted file mode 100644
index 9004b52..0000000
--- a/vp8/common/x86/subpixel_mmx.asm
+++ /dev/null
@@ -1,727 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-
-%define BLOCK_HEIGHT_WIDTH 4
-%define vp8_filter_weight 128
-%define VP8_FILTER_SHIFT 7
-
-
-;void vp8_filter_block1d_h6_mmx
-;(
-; unsigned char *src_ptr,
-; unsigned short *output_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned int pixel_step,
-; unsigned int output_height,
-; unsigned int output_width,
-; short * vp8_filter
-;)
-global sym(vp8_filter_block1d_h6_mmx)
-sym(vp8_filter_block1d_h6_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rdx, arg(6) ;vp8_filter
-
- movq mm1, [rdx + 16] ; do both the negative taps first!!!
- movq mm2, [rdx + 32] ;
- movq mm6, [rdx + 48] ;
- movq mm7, [rdx + 64] ;
-
- mov rdi, arg(1) ;output_ptr
- mov rsi, arg(0) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;output_height
- movsxd rax, dword ptr arg(5) ;output_width ; destination pitch?
- pxor mm0, mm0 ; mm0 = 00000000
-
-nextrow:
- movq mm3, [rsi-2] ; mm3 = p-2..p5
- movq mm4, mm3 ; mm4 = p-2..p5
- psrlq mm3, 8 ; mm3 = p-1..p5
- punpcklbw mm3, mm0 ; mm3 = p-1..p2
- pmullw mm3, mm1 ; mm3 *= kernel 1 modifiers.
-
- movq mm5, mm4 ; mm5 = p-2..p5
- punpckhbw mm4, mm0 ; mm5 = p2..p5
- pmullw mm4, mm7 ; mm5 *= kernel 4 modifiers
- paddsw mm3, mm4 ; mm3 += mm5
-
- movq mm4, mm5 ; mm4 = p-2..p5;
- psrlq mm5, 16 ; mm5 = p0..p5;
- punpcklbw mm5, mm0 ; mm5 = p0..p3
- pmullw mm5, mm2 ; mm5 *= kernel 2 modifiers
- paddsw mm3, mm5 ; mm3 += mm5
-
- movq mm5, mm4 ; mm5 = p-2..p5
- psrlq mm4, 24 ; mm4 = p1..p5
- punpcklbw mm4, mm0 ; mm4 = p1..p4
- pmullw mm4, mm6 ; mm5 *= kernel 3 modifiers
- paddsw mm3, mm4 ; mm3 += mm5
-
- ; do outer positive taps
- movd mm4, [rsi+3]
- punpcklbw mm4, mm0 ; mm5 = p3..p6
- pmullw mm4, [rdx+80] ; mm5 *= kernel 0 modifiers
- paddsw mm3, mm4 ; mm3 += mm5
-
- punpcklbw mm5, mm0 ; mm5 = p-2..p1
- pmullw mm5, [rdx] ; mm5 *= kernel 5 modifiers
- paddsw mm3, mm5 ; mm3 += mm5
-
- paddsw mm3, [GLOBAL(rd)] ; mm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; mm3 /= 128
- packuswb mm3, mm0 ; pack and unpack to saturate
- punpcklbw mm3, mm0 ;
-
- movq [rdi], mm3 ; store the results in the destination
-
-%if ABI_IS_32BIT
- add rsi, dword ptr arg(2) ;src_pixels_per_line ; next line
- add rdi, rax;
-%else
- movsxd r8, dword ptr arg(2) ;src_pixels_per_line
- add rdi, rax;
-
- add rsi, r8 ; next line
-%endif
-
- dec rcx ; decrement count
- jnz nextrow ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_filter_block1dc_v6_mmx
-;(
-; short *src_ptr,
-; unsigned char *output_ptr,
-; int output_pitch,
-; unsigned int pixels_per_line,
-; unsigned int pixel_step,
-; unsigned int output_height,
-; unsigned int output_width,
-; short * vp8_filter
-;)
-global sym(vp8_filter_block1dc_v6_mmx)
-sym(vp8_filter_block1dc_v6_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 8
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- movq mm5, [GLOBAL(rd)]
- push rbx
- mov rbx, arg(7) ;vp8_filter
- movq mm1, [rbx + 16] ; do both the negative taps first!!!
- movq mm2, [rbx + 32] ;
- movq mm6, [rbx + 48] ;
- movq mm7, [rbx + 64] ;
-
- movsxd rdx, dword ptr arg(3) ;pixels_per_line
- mov rdi, arg(1) ;output_ptr
- mov rsi, arg(0) ;src_ptr
- sub rsi, rdx
- sub rsi, rdx
- movsxd rcx, DWORD PTR arg(5) ;output_height
- movsxd rax, DWORD PTR arg(2) ;output_pitch ; destination pitch?
- pxor mm0, mm0 ; mm0 = 00000000
-
-
-nextrow_cv:
- movq mm3, [rsi+rdx] ; mm3 = p0..p8 = row -1
- pmullw mm3, mm1 ; mm3 *= kernel 1 modifiers.
-
-
- movq mm4, [rsi + 4*rdx] ; mm4 = p0..p3 = row 2
- pmullw mm4, mm7 ; mm4 *= kernel 4 modifiers.
- paddsw mm3, mm4 ; mm3 += mm4
-
- movq mm4, [rsi + 2*rdx] ; mm4 = p0..p3 = row 0
- pmullw mm4, mm2 ; mm4 *= kernel 2 modifiers.
- paddsw mm3, mm4 ; mm3 += mm4
-
- movq mm4, [rsi] ; mm4 = p0..p3 = row -2
- pmullw mm4, [rbx] ; mm4 *= kernel 0 modifiers.
- paddsw mm3, mm4 ; mm3 += mm4
-
-
- add rsi, rdx ; move source forward 1 line to avoid 3 * pitch
- movq mm4, [rsi + 2*rdx] ; mm4 = p0..p3 = row 1
- pmullw mm4, mm6 ; mm4 *= kernel 3 modifiers.
- paddsw mm3, mm4 ; mm3 += mm4
-
- movq mm4, [rsi + 4*rdx] ; mm4 = p0..p3 = row 3
- pmullw mm4, [rbx +80] ; mm4 *= kernel 3 modifiers.
- paddsw mm3, mm4 ; mm3 += mm4
-
-
- paddsw mm3, mm5 ; mm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; mm3 /= 128
- packuswb mm3, mm0 ; pack and saturate
-
- movd [rdi],mm3 ; store the results in the destination
- ; the subsequent iterations repeat 3 out of 4 of these reads. Since the
- ; recon block should be in cache this shouldn't cost much. Its obviously
- ; avoidable!!!.
- lea rdi, [rdi+rax] ;
- dec rcx ; decrement count
- jnz nextrow_cv ; next row
-
- pop rbx
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void bilinear_predict8x8_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; int xoffset,
-; int yoffset,
-; unsigned char *dst_ptr,
-; int dst_pitch
-;)
-global sym(vp8_bilinear_predict8x8_mmx)
-sym(vp8_bilinear_predict8x8_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ;const short *HFilter = bilinear_filters_mmx[xoffset];
- ;const short *VFilter = bilinear_filters_mmx[yoffset];
-
- movsxd rax, dword ptr arg(2) ;xoffset
- mov rdi, arg(4) ;dst_ptr ;
-
- shl rax, 5 ; offset * 32
- lea rcx, [GLOBAL(sym(vp8_bilinear_filters_mmx))]
-
- add rax, rcx ; HFilter
- mov rsi, arg(0) ;src_ptr ;
-
- movsxd rdx, dword ptr arg(5) ;dst_pitch
- movq mm1, [rax] ;
-
- movq mm2, [rax+16] ;
- movsxd rax, dword ptr arg(3) ;yoffset
-
- pxor mm0, mm0 ;
-
- shl rax, 5 ; offset*32
- add rax, rcx ; VFilter
-
- lea rcx, [rdi+rdx*8] ;
- movsxd rdx, dword ptr arg(1) ;src_pixels_per_line ;
-
-
-
- ; get the first horizontal line done ;
- movq mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- movq mm4, mm3 ; make a copy of current line
-
- punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
- punpckhbw mm4, mm0 ;
-
- pmullw mm3, mm1 ;
- pmullw mm4, mm1 ;
-
- movq mm5, [rsi+1] ;
- movq mm6, mm5 ;
-
- punpcklbw mm5, mm0 ;
- punpckhbw mm6, mm0 ;
-
- pmullw mm5, mm2 ;
- pmullw mm6, mm2 ;
-
- paddw mm3, mm5 ;
- paddw mm4, mm6 ;
-
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw mm4, [GLOBAL(rd)] ;
- psraw mm4, VP8_FILTER_SHIFT ;
-
- movq mm7, mm3 ;
- packuswb mm7, mm4 ;
-
- add rsi, rdx ; next line
-next_row_8x8:
- movq mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- movq mm4, mm3 ; make a copy of current line
-
- punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
- punpckhbw mm4, mm0 ;
-
- pmullw mm3, mm1 ;
- pmullw mm4, mm1 ;
-
- movq mm5, [rsi+1] ;
- movq mm6, mm5 ;
-
- punpcklbw mm5, mm0 ;
- punpckhbw mm6, mm0 ;
-
- pmullw mm5, mm2 ;
- pmullw mm6, mm2 ;
-
- paddw mm3, mm5 ;
- paddw mm4, mm6 ;
-
- movq mm5, mm7 ;
- movq mm6, mm7 ;
-
- punpcklbw mm5, mm0 ;
- punpckhbw mm6, mm0
-
- pmullw mm5, [rax] ;
- pmullw mm6, [rax] ;
-
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw mm4, [GLOBAL(rd)] ;
- psraw mm4, VP8_FILTER_SHIFT ;
-
- movq mm7, mm3 ;
- packuswb mm7, mm4 ;
-
-
- pmullw mm3, [rax+16] ;
- pmullw mm4, [rax+16] ;
-
- paddw mm3, mm5 ;
- paddw mm4, mm6 ;
-
-
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw mm4, [GLOBAL(rd)] ;
- psraw mm4, VP8_FILTER_SHIFT ;
-
- packuswb mm3, mm4
-
- movq [rdi], mm3 ; store the results in the destination
-
-%if ABI_IS_32BIT
- add rsi, rdx ; next line
- add rdi, dword ptr arg(5) ;dst_pitch ;
-%else
- movsxd r8, dword ptr arg(5) ;dst_pitch
- add rsi, rdx ; next line
- add rdi, r8 ;dst_pitch
-%endif
- cmp rdi, rcx ;
- jne next_row_8x8
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void bilinear_predict8x4_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; int xoffset,
-; int yoffset,
-; unsigned char *dst_ptr,
-; int dst_pitch
-;)
-global sym(vp8_bilinear_predict8x4_mmx)
-sym(vp8_bilinear_predict8x4_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ;const short *HFilter = bilinear_filters_mmx[xoffset];
- ;const short *VFilter = bilinear_filters_mmx[yoffset];
-
- movsxd rax, dword ptr arg(2) ;xoffset
- mov rdi, arg(4) ;dst_ptr ;
-
- lea rcx, [GLOBAL(sym(vp8_bilinear_filters_mmx))]
- shl rax, 5
-
- mov rsi, arg(0) ;src_ptr ;
- add rax, rcx
-
- movsxd rdx, dword ptr arg(5) ;dst_pitch
- movq mm1, [rax] ;
-
- movq mm2, [rax+16] ;
- movsxd rax, dword ptr arg(3) ;yoffset
-
- pxor mm0, mm0 ;
- shl rax, 5
-
- add rax, rcx
- lea rcx, [rdi+rdx*4] ;
-
- movsxd rdx, dword ptr arg(1) ;src_pixels_per_line ;
-
- ; get the first horizontal line done ;
- movq mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- movq mm4, mm3 ; make a copy of current line
-
- punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
- punpckhbw mm4, mm0 ;
-
- pmullw mm3, mm1 ;
- pmullw mm4, mm1 ;
-
- movq mm5, [rsi+1] ;
- movq mm6, mm5 ;
-
- punpcklbw mm5, mm0 ;
- punpckhbw mm6, mm0 ;
-
- pmullw mm5, mm2 ;
- pmullw mm6, mm2 ;
-
- paddw mm3, mm5 ;
- paddw mm4, mm6 ;
-
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw mm4, [GLOBAL(rd)] ;
- psraw mm4, VP8_FILTER_SHIFT ;
-
- movq mm7, mm3 ;
- packuswb mm7, mm4 ;
-
- add rsi, rdx ; next line
-next_row_8x4:
- movq mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- movq mm4, mm3 ; make a copy of current line
-
- punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
- punpckhbw mm4, mm0 ;
-
- pmullw mm3, mm1 ;
- pmullw mm4, mm1 ;
-
- movq mm5, [rsi+1] ;
- movq mm6, mm5 ;
-
- punpcklbw mm5, mm0 ;
- punpckhbw mm6, mm0 ;
-
- pmullw mm5, mm2 ;
- pmullw mm6, mm2 ;
-
- paddw mm3, mm5 ;
- paddw mm4, mm6 ;
-
- movq mm5, mm7 ;
- movq mm6, mm7 ;
-
- punpcklbw mm5, mm0 ;
- punpckhbw mm6, mm0
-
- pmullw mm5, [rax] ;
- pmullw mm6, [rax] ;
-
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw mm4, [GLOBAL(rd)] ;
- psraw mm4, VP8_FILTER_SHIFT ;
-
- movq mm7, mm3 ;
- packuswb mm7, mm4 ;
-
-
- pmullw mm3, [rax+16] ;
- pmullw mm4, [rax+16] ;
-
- paddw mm3, mm5 ;
- paddw mm4, mm6 ;
-
-
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw mm4, [GLOBAL(rd)] ;
- psraw mm4, VP8_FILTER_SHIFT ;
-
- packuswb mm3, mm4
-
- movq [rdi], mm3 ; store the results in the destination
-
-%if ABI_IS_32BIT
- add rsi, rdx ; next line
- add rdi, dword ptr arg(5) ;dst_pitch ;
-%else
- movsxd r8, dword ptr arg(5) ;dst_pitch
- add rsi, rdx ; next line
- add rdi, r8
-%endif
- cmp rdi, rcx ;
- jne next_row_8x4
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void bilinear_predict4x4_mmx
-;(
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; int xoffset,
-; int yoffset,
-; unsigned char *dst_ptr,
-; int dst_pitch
-;)
-global sym(vp8_bilinear_predict4x4_mmx)
-sym(vp8_bilinear_predict4x4_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ;const short *HFilter = bilinear_filters_mmx[xoffset];
- ;const short *VFilter = bilinear_filters_mmx[yoffset];
-
- movsxd rax, dword ptr arg(2) ;xoffset
- mov rdi, arg(4) ;dst_ptr ;
-
- lea rcx, [GLOBAL(sym(vp8_bilinear_filters_mmx))]
- shl rax, 5
-
- add rax, rcx ; HFilter
- mov rsi, arg(0) ;src_ptr ;
-
- movsxd rdx, dword ptr arg(5) ;ldst_pitch
- movq mm1, [rax] ;
-
- movq mm2, [rax+16] ;
- movsxd rax, dword ptr arg(3) ;yoffset
-
- pxor mm0, mm0 ;
- shl rax, 5
-
- add rax, rcx
- lea rcx, [rdi+rdx*4] ;
-
- movsxd rdx, dword ptr arg(1) ;src_pixels_per_line ;
-
- ; get the first horizontal line done ;
- movd mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
-
- pmullw mm3, mm1 ;
- movd mm5, [rsi+1] ;
-
- punpcklbw mm5, mm0 ;
- pmullw mm5, mm2 ;
-
- paddw mm3, mm5 ;
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
-
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- movq mm7, mm3 ;
- packuswb mm7, mm0 ;
-
- add rsi, rdx ; next line
-next_row_4x4:
- movd mm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- punpcklbw mm3, mm0 ; xx 00 01 02 03 04 05 06
-
- pmullw mm3, mm1 ;
- movd mm5, [rsi+1] ;
-
- punpcklbw mm5, mm0 ;
- pmullw mm5, mm2 ;
-
- paddw mm3, mm5 ;
-
- movq mm5, mm7 ;
- punpcklbw mm5, mm0 ;
-
- pmullw mm5, [rax] ;
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
-
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
- movq mm7, mm3 ;
-
- packuswb mm7, mm0 ;
-
- pmullw mm3, [rax+16] ;
- paddw mm3, mm5 ;
-
-
- paddw mm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw mm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- packuswb mm3, mm0
- movd [rdi], mm3 ; store the results in the destination
-
-%if ABI_IS_32BIT
- add rsi, rdx ; next line
- add rdi, dword ptr arg(5) ;dst_pitch ;
-%else
- movsxd r8, dword ptr arg(5) ;dst_pitch ;
- add rsi, rdx ; next line
- add rdi, r8
-%endif
-
- cmp rdi, rcx ;
- jne next_row_4x4
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-SECTION_RODATA
-align 16
-rd:
- times 4 dw 0x40
-
-align 16
-global HIDDEN_DATA(sym(vp8_six_tap_mmx))
-sym(vp8_six_tap_mmx):
- times 8 dw 0
- times 8 dw 0
- times 8 dw 128
- times 8 dw 0
- times 8 dw 0
- times 8 dw 0
-
- times 8 dw 0
- times 8 dw -6
- times 8 dw 123
- times 8 dw 12
- times 8 dw -1
- times 8 dw 0
-
- times 8 dw 2
- times 8 dw -11
- times 8 dw 108
- times 8 dw 36
- times 8 dw -8
- times 8 dw 1
-
- times 8 dw 0
- times 8 dw -9
- times 8 dw 93
- times 8 dw 50
- times 8 dw -6
- times 8 dw 0
-
- times 8 dw 3
- times 8 dw -16
- times 8 dw 77
- times 8 dw 77
- times 8 dw -16
- times 8 dw 3
-
- times 8 dw 0
- times 8 dw -6
- times 8 dw 50
- times 8 dw 93
- times 8 dw -9
- times 8 dw 0
-
- times 8 dw 1
- times 8 dw -8
- times 8 dw 36
- times 8 dw 108
- times 8 dw -11
- times 8 dw 2
-
- times 8 dw 0
- times 8 dw -1
- times 8 dw 12
- times 8 dw 123
- times 8 dw -6
- times 8 dw 0
-
-
-align 16
-global HIDDEN_DATA(sym(vp8_bilinear_filters_mmx))
-sym(vp8_bilinear_filters_mmx):
- times 8 dw 128
- times 8 dw 0
-
- times 8 dw 112
- times 8 dw 16
-
- times 8 dw 96
- times 8 dw 32
-
- times 8 dw 80
- times 8 dw 48
-
- times 8 dw 64
- times 8 dw 64
-
- times 8 dw 48
- times 8 dw 80
-
- times 8 dw 32
- times 8 dw 96
-
- times 8 dw 16
- times 8 dw 112
diff --git a/vp8/common/x86/subpixel_sse2.asm b/vp8/common/x86/subpixel_sse2.asm
deleted file mode 100644
index b87cad2..0000000
--- a/vp8/common/x86/subpixel_sse2.asm
+++ /dev/null
@@ -1,1374 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%define BLOCK_HEIGHT_WIDTH 4
-%define VP8_FILTER_WEIGHT 128
-%define VP8_FILTER_SHIFT 7
-
-
-;/************************************************************************************
-; Notes: filter_block1d_h6 applies a 6 tap filter horizontally to the input pixels. The
-; input pixel array has output_height rows. This routine assumes that output_height is an
-; even number. This function handles 8 pixels in horizontal direction, calculating ONE
-; rows each iteration to take advantage of the 128 bits operations.
-;*************************************************************************************/
-;void vp8_filter_block1d8_h6_sse2
-;(
-; unsigned char *src_ptr,
-; unsigned short *output_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned int pixel_step,
-; unsigned int output_height,
-; unsigned int output_width,
-; short *vp8_filter
-;)
-global sym(vp8_filter_block1d8_h6_sse2)
-sym(vp8_filter_block1d8_h6_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rdx, arg(6) ;vp8_filter
- mov rsi, arg(0) ;src_ptr
-
- mov rdi, arg(1) ;output_ptr
-
- movsxd rcx, dword ptr arg(4) ;output_height
- movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(5) ;output_width
-%endif
- pxor xmm0, xmm0 ; clear xmm0 for unpack
-
-filter_block1d8_h6_rowloop:
- movq xmm3, MMWORD PTR [rsi - 2]
- movq xmm1, MMWORD PTR [rsi + 6]
-
- prefetcht2 [rsi+rax-2]
-
- pslldq xmm1, 8
- por xmm1, xmm3
-
- movdqa xmm4, xmm1
- movdqa xmm5, xmm1
-
- movdqa xmm6, xmm1
- movdqa xmm7, xmm1
-
- punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
- psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
-
- pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
- punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
-
- psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
- pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
-
-
- punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
- psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
-
- pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
-
- punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
- psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
-
- pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
-
- punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
- psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
-
-
- pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
-
- punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
- pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
-
-
- paddsw xmm4, xmm7
- paddsw xmm4, xmm5
-
- paddsw xmm4, xmm3
- paddsw xmm4, xmm6
-
- paddsw xmm4, xmm1
- paddsw xmm4, [GLOBAL(rd)]
-
- psraw xmm4, 7
-
- packuswb xmm4, xmm0
- punpcklbw xmm4, xmm0
-
- movdqa XMMWORD Ptr [rdi], xmm4
- lea rsi, [rsi + rax]
-
-%if ABI_IS_32BIT
- add rdi, DWORD Ptr arg(5) ;[output_width]
-%else
- add rdi, r8
-%endif
- dec rcx
-
- jnz filter_block1d8_h6_rowloop ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_filter_block1d16_h6_sse2
-;(
-; unsigned char *src_ptr,
-; unsigned short *output_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned int pixel_step,
-; unsigned int output_height,
-; unsigned int output_width,
-; short *vp8_filter
-;)
-;/************************************************************************************
-; Notes: filter_block1d_h6 applies a 6 tap filter horizontally to the input pixels. The
-; input pixel array has output_height rows. This routine assumes that output_height is an
-; even number. This function handles 8 pixels in horizontal direction, calculating ONE
-; rows each iteration to take advantage of the 128 bits operations.
-;*************************************************************************************/
-global sym(vp8_filter_block1d16_h6_sse2)
-sym(vp8_filter_block1d16_h6_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rdx, arg(6) ;vp8_filter
- mov rsi, arg(0) ;src_ptr
-
- mov rdi, arg(1) ;output_ptr
-
- movsxd rcx, dword ptr arg(4) ;output_height
- movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(5) ;output_width
-%endif
-
- pxor xmm0, xmm0 ; clear xmm0 for unpack
-
-filter_block1d16_h6_sse2_rowloop:
- movq xmm3, MMWORD PTR [rsi - 2]
- movq xmm1, MMWORD PTR [rsi + 6]
-
- movq xmm2, MMWORD PTR [rsi +14]
- pslldq xmm2, 8
-
- por xmm2, xmm1
- prefetcht2 [rsi+rax-2]
-
- pslldq xmm1, 8
- por xmm1, xmm3
-
- movdqa xmm4, xmm1
- movdqa xmm5, xmm1
-
- movdqa xmm6, xmm1
- movdqa xmm7, xmm1
-
- punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
- psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
-
- pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
- punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
-
- psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
- pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
-
-
- punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
- psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
-
- pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
-
- punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
- psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
-
- pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
-
- punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
- psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
-
-
- pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
-
- punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
- pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
-
- paddsw xmm4, xmm7
- paddsw xmm4, xmm5
-
- paddsw xmm4, xmm3
- paddsw xmm4, xmm6
-
- paddsw xmm4, xmm1
- paddsw xmm4, [GLOBAL(rd)]
-
- psraw xmm4, 7
-
- packuswb xmm4, xmm0
- punpcklbw xmm4, xmm0
-
- movdqa XMMWORD Ptr [rdi], xmm4
-
- movdqa xmm3, xmm2
- movdqa xmm4, xmm2
-
- movdqa xmm5, xmm2
- movdqa xmm6, xmm2
-
- movdqa xmm7, xmm2
-
- punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
- psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
-
- pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
- punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
-
- psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
- pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
-
-
- punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
- psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
-
- pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
-
- punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
- psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
-
- pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
-
- punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
- psrldq xmm2, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
-
- pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
-
- punpcklbw xmm2, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
- pmullw xmm2, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
-
-
- paddsw xmm4, xmm7
- paddsw xmm4, xmm5
-
- paddsw xmm4, xmm3
- paddsw xmm4, xmm6
-
- paddsw xmm4, xmm2
- paddsw xmm4, [GLOBAL(rd)]
-
- psraw xmm4, 7
-
- packuswb xmm4, xmm0
- punpcklbw xmm4, xmm0
-
- movdqa XMMWORD Ptr [rdi+16], xmm4
-
- lea rsi, [rsi + rax]
-%if ABI_IS_32BIT
- add rdi, DWORD Ptr arg(5) ;[output_width]
-%else
- add rdi, r8
-%endif
-
- dec rcx
- jnz filter_block1d16_h6_sse2_rowloop ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_filter_block1d8_v6_sse2
-;(
-; short *src_ptr,
-; unsigned char *output_ptr,
-; int dst_ptich,
-; unsigned int pixels_per_line,
-; unsigned int pixel_step,
-; unsigned int output_height,
-; unsigned int output_width,
-; short * vp8_filter
-;)
-;/************************************************************************************
-; Notes: filter_block1d8_v6 applies a 6 tap filter vertically to the input pixels. The
-; input pixel array has output_height rows.
-;*************************************************************************************/
-global sym(vp8_filter_block1d8_v6_sse2)
-sym(vp8_filter_block1d8_v6_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 8
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rax, arg(7) ;vp8_filter
- movsxd rdx, dword ptr arg(3) ;pixels_per_line
-
- mov rdi, arg(1) ;output_ptr
- mov rsi, arg(0) ;src_ptr
-
- sub rsi, rdx
- sub rsi, rdx
-
- movsxd rcx, DWORD PTR arg(5) ;[output_height]
- pxor xmm0, xmm0 ; clear xmm0
-
- movdqa xmm7, XMMWORD PTR [GLOBAL(rd)]
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(2) ; dst_ptich
-%endif
-
-vp8_filter_block1d8_v6_sse2_loop:
- movdqa xmm1, XMMWORD PTR [rsi]
- pmullw xmm1, [rax]
-
- movdqa xmm2, XMMWORD PTR [rsi + rdx]
- pmullw xmm2, [rax + 16]
-
- movdqa xmm3, XMMWORD PTR [rsi + rdx * 2]
- pmullw xmm3, [rax + 32]
-
- movdqa xmm5, XMMWORD PTR [rsi + rdx * 4]
- pmullw xmm5, [rax + 64]
-
- add rsi, rdx
- movdqa xmm4, XMMWORD PTR [rsi + rdx * 2]
-
- pmullw xmm4, [rax + 48]
- movdqa xmm6, XMMWORD PTR [rsi + rdx * 4]
-
- pmullw xmm6, [rax + 80]
-
- paddsw xmm2, xmm5
- paddsw xmm2, xmm3
-
- paddsw xmm2, xmm1
- paddsw xmm2, xmm4
-
- paddsw xmm2, xmm6
- paddsw xmm2, xmm7
-
- psraw xmm2, 7
- packuswb xmm2, xmm0 ; pack and saturate
-
- movq QWORD PTR [rdi], xmm2 ; store the results in the destination
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(2) ;[dst_ptich]
-%else
- add rdi, r8
-%endif
- dec rcx ; decrement count
- jnz vp8_filter_block1d8_v6_sse2_loop ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_filter_block1d16_v6_sse2
-;(
-; unsigned short *src_ptr,
-; unsigned char *output_ptr,
-; int dst_ptich,
-; unsigned int pixels_per_line,
-; unsigned int pixel_step,
-; unsigned int output_height,
-; unsigned int output_width,
-; const short *vp8_filter
-;)
-;/************************************************************************************
-; Notes: filter_block1d16_v6 applies a 6 tap filter vertically to the input pixels. The
-; input pixel array has output_height rows.
-;*************************************************************************************/
-global sym(vp8_filter_block1d16_v6_sse2)
-sym(vp8_filter_block1d16_v6_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 8
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rax, arg(7) ;vp8_filter
- movsxd rdx, dword ptr arg(3) ;pixels_per_line
-
- mov rdi, arg(1) ;output_ptr
- mov rsi, arg(0) ;src_ptr
-
- sub rsi, rdx
- sub rsi, rdx
-
- movsxd rcx, DWORD PTR arg(5) ;[output_height]
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(2) ; dst_ptich
-%endif
-
-vp8_filter_block1d16_v6_sse2_loop:
-; The order for adding 6-tap is 2 5 3 1 4 6. Read in data in that order.
- movdqa xmm1, XMMWORD PTR [rsi + rdx] ; line 2
- movdqa xmm2, XMMWORD PTR [rsi + rdx + 16]
- pmullw xmm1, [rax + 16]
- pmullw xmm2, [rax + 16]
-
- movdqa xmm3, XMMWORD PTR [rsi + rdx * 4] ; line 5
- movdqa xmm4, XMMWORD PTR [rsi + rdx * 4 + 16]
- pmullw xmm3, [rax + 64]
- pmullw xmm4, [rax + 64]
-
- movdqa xmm5, XMMWORD PTR [rsi + rdx * 2] ; line 3
- movdqa xmm6, XMMWORD PTR [rsi + rdx * 2 + 16]
- pmullw xmm5, [rax + 32]
- pmullw xmm6, [rax + 32]
-
- movdqa xmm7, XMMWORD PTR [rsi] ; line 1
- movdqa xmm0, XMMWORD PTR [rsi + 16]
- pmullw xmm7, [rax]
- pmullw xmm0, [rax]
-
- paddsw xmm1, xmm3
- paddsw xmm2, xmm4
- paddsw xmm1, xmm5
- paddsw xmm2, xmm6
- paddsw xmm1, xmm7
- paddsw xmm2, xmm0
-
- add rsi, rdx
-
- movdqa xmm3, XMMWORD PTR [rsi + rdx * 2] ; line 4
- movdqa xmm4, XMMWORD PTR [rsi + rdx * 2 + 16]
- pmullw xmm3, [rax + 48]
- pmullw xmm4, [rax + 48]
-
- movdqa xmm5, XMMWORD PTR [rsi + rdx * 4] ; line 6
- movdqa xmm6, XMMWORD PTR [rsi + rdx * 4 + 16]
- pmullw xmm5, [rax + 80]
- pmullw xmm6, [rax + 80]
-
- movdqa xmm7, XMMWORD PTR [GLOBAL(rd)]
- pxor xmm0, xmm0 ; clear xmm0
-
- paddsw xmm1, xmm3
- paddsw xmm2, xmm4
- paddsw xmm1, xmm5
- paddsw xmm2, xmm6
-
- paddsw xmm1, xmm7
- paddsw xmm2, xmm7
-
- psraw xmm1, 7
- psraw xmm2, 7
-
- packuswb xmm1, xmm2 ; pack and saturate
- movdqa XMMWORD PTR [rdi], xmm1 ; store the results in the destination
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(2) ;[dst_ptich]
-%else
- add rdi, r8
-%endif
- dec rcx ; decrement count
- jnz vp8_filter_block1d16_v6_sse2_loop ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_filter_block1d8_h6_only_sse2
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned char *output_ptr,
-; int dst_ptich,
-; unsigned int output_height,
-; const short *vp8_filter
-;)
-; First-pass filter only when yoffset==0
-global sym(vp8_filter_block1d8_h6_only_sse2)
-sym(vp8_filter_block1d8_h6_only_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rdx, arg(5) ;vp8_filter
- mov rsi, arg(0) ;src_ptr
-
- mov rdi, arg(2) ;output_ptr
-
- movsxd rcx, dword ptr arg(4) ;output_height
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line ; Pitch for Source
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(3) ;dst_ptich
-%endif
- pxor xmm0, xmm0 ; clear xmm0 for unpack
-
-filter_block1d8_h6_only_rowloop:
- movq xmm3, MMWORD PTR [rsi - 2]
- movq xmm1, MMWORD PTR [rsi + 6]
-
- prefetcht2 [rsi+rax-2]
-
- pslldq xmm1, 8
- por xmm1, xmm3
-
- movdqa xmm4, xmm1
- movdqa xmm5, xmm1
-
- movdqa xmm6, xmm1
- movdqa xmm7, xmm1
-
- punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
- psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
-
- pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
- punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
-
- psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
- pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
-
-
- punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
- psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
-
- pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
-
- punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
- psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
-
- pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
-
- punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
- psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
-
-
- pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
-
- punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
- pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
-
-
- paddsw xmm4, xmm7
- paddsw xmm4, xmm5
-
- paddsw xmm4, xmm3
- paddsw xmm4, xmm6
-
- paddsw xmm4, xmm1
- paddsw xmm4, [GLOBAL(rd)]
-
- psraw xmm4, 7
-
- packuswb xmm4, xmm0
-
- movq QWORD PTR [rdi], xmm4 ; store the results in the destination
- lea rsi, [rsi + rax]
-
-%if ABI_IS_32BIT
- add rdi, DWORD Ptr arg(3) ;dst_ptich
-%else
- add rdi, r8
-%endif
- dec rcx
-
- jnz filter_block1d8_h6_only_rowloop ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_filter_block1d16_h6_only_sse2
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned char *output_ptr,
-; int dst_ptich,
-; unsigned int output_height,
-; const short *vp8_filter
-;)
-; First-pass filter only when yoffset==0
-global sym(vp8_filter_block1d16_h6_only_sse2)
-sym(vp8_filter_block1d16_h6_only_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rdx, arg(5) ;vp8_filter
- mov rsi, arg(0) ;src_ptr
-
- mov rdi, arg(2) ;output_ptr
-
- movsxd rcx, dword ptr arg(4) ;output_height
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line ; Pitch for Source
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(3) ;dst_ptich
-%endif
-
- pxor xmm0, xmm0 ; clear xmm0 for unpack
-
-filter_block1d16_h6_only_sse2_rowloop:
- movq xmm3, MMWORD PTR [rsi - 2]
- movq xmm1, MMWORD PTR [rsi + 6]
-
- movq xmm2, MMWORD PTR [rsi +14]
- pslldq xmm2, 8
-
- por xmm2, xmm1
- prefetcht2 [rsi+rax-2]
-
- pslldq xmm1, 8
- por xmm1, xmm3
-
- movdqa xmm4, xmm1
- movdqa xmm5, xmm1
-
- movdqa xmm6, xmm1
- movdqa xmm7, xmm1
-
- punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
- psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
-
- pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
- punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
-
- psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
- pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
-
- punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
- psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
-
- pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
-
- punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
- psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
-
- pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
-
- punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
- psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
-
- pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
-
- punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
- pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
-
- paddsw xmm4, xmm7
- paddsw xmm4, xmm5
-
- paddsw xmm4, xmm3
- paddsw xmm4, xmm6
-
- paddsw xmm4, xmm1
- paddsw xmm4, [GLOBAL(rd)]
-
- psraw xmm4, 7
-
- packuswb xmm4, xmm0 ; lower 8 bytes
-
- movq QWORD Ptr [rdi], xmm4 ; store the results in the destination
-
- movdqa xmm3, xmm2
- movdqa xmm4, xmm2
-
- movdqa xmm5, xmm2
- movdqa xmm6, xmm2
-
- movdqa xmm7, xmm2
-
- punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
- psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1
-
- pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1
- punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1
-
- psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
- pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2
-
- punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00
- psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01
-
- pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3
-
- punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01
- psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02
-
- pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4
-
- punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02
- psrldq xmm2, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03
-
- pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5
-
- punpcklbw xmm2, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03
- pmullw xmm2, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6
-
- paddsw xmm4, xmm7
- paddsw xmm4, xmm5
-
- paddsw xmm4, xmm3
- paddsw xmm4, xmm6
-
- paddsw xmm4, xmm2
- paddsw xmm4, [GLOBAL(rd)]
-
- psraw xmm4, 7
-
- packuswb xmm4, xmm0 ; higher 8 bytes
-
- movq QWORD Ptr [rdi+8], xmm4 ; store the results in the destination
-
- lea rsi, [rsi + rax]
-%if ABI_IS_32BIT
- add rdi, DWORD Ptr arg(3) ;dst_ptich
-%else
- add rdi, r8
-%endif
-
- dec rcx
- jnz filter_block1d16_h6_only_sse2_rowloop ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_filter_block1d8_v6_only_sse2
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned char *output_ptr,
-; int dst_ptich,
-; unsigned int output_height,
-; const short *vp8_filter
-;)
-; Second-pass filter only when xoffset==0
-global sym(vp8_filter_block1d8_v6_only_sse2)
-sym(vp8_filter_block1d8_v6_only_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;output_ptr
-
- movsxd rcx, dword ptr arg(4) ;output_height
- movsxd rdx, dword ptr arg(1) ;src_pixels_per_line
-
- mov rax, arg(5) ;vp8_filter
-
- pxor xmm0, xmm0 ; clear xmm0
-
- movdqa xmm7, XMMWORD PTR [GLOBAL(rd)]
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(3) ; dst_ptich
-%endif
-
-vp8_filter_block1d8_v6_only_sse2_loop:
- movq xmm1, MMWORD PTR [rsi]
- movq xmm2, MMWORD PTR [rsi + rdx]
- movq xmm3, MMWORD PTR [rsi + rdx * 2]
- movq xmm5, MMWORD PTR [rsi + rdx * 4]
- add rsi, rdx
- movq xmm4, MMWORD PTR [rsi + rdx * 2]
- movq xmm6, MMWORD PTR [rsi + rdx * 4]
-
- punpcklbw xmm1, xmm0
- pmullw xmm1, [rax]
-
- punpcklbw xmm2, xmm0
- pmullw xmm2, [rax + 16]
-
- punpcklbw xmm3, xmm0
- pmullw xmm3, [rax + 32]
-
- punpcklbw xmm5, xmm0
- pmullw xmm5, [rax + 64]
-
- punpcklbw xmm4, xmm0
- pmullw xmm4, [rax + 48]
-
- punpcklbw xmm6, xmm0
- pmullw xmm6, [rax + 80]
-
- paddsw xmm2, xmm5
- paddsw xmm2, xmm3
-
- paddsw xmm2, xmm1
- paddsw xmm2, xmm4
-
- paddsw xmm2, xmm6
- paddsw xmm2, xmm7
-
- psraw xmm2, 7
- packuswb xmm2, xmm0 ; pack and saturate
-
- movq QWORD PTR [rdi], xmm2 ; store the results in the destination
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(3) ;[dst_ptich]
-%else
- add rdi, r8
-%endif
- dec rcx ; decrement count
- jnz vp8_filter_block1d8_v6_only_sse2_loop ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_unpack_block1d16_h6_sse2
-;(
-; unsigned char *src_ptr,
-; unsigned short *output_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned int output_height,
-; unsigned int output_width
-;)
-global sym(vp8_unpack_block1d16_h6_sse2)
-sym(vp8_unpack_block1d16_h6_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- ;SAVE_XMM ;xmm6, xmm7 are not used here.
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(1) ;output_ptr
-
- movsxd rcx, dword ptr arg(3) ;output_height
- movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source
-
- pxor xmm0, xmm0 ; clear xmm0 for unpack
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(4) ;output_width ; Pitch for Source
-%endif
-
-unpack_block1d16_h6_sse2_rowloop:
- movq xmm1, MMWORD PTR [rsi] ; 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1 -2
- movq xmm3, MMWORD PTR [rsi+8] ; make copy of xmm1
-
- punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2
- punpcklbw xmm1, xmm0
-
- movdqa XMMWORD Ptr [rdi], xmm1
- movdqa XMMWORD Ptr [rdi + 16], xmm3
-
- lea rsi, [rsi + rax]
-%if ABI_IS_32BIT
- add rdi, DWORD Ptr arg(4) ;[output_width]
-%else
- add rdi, r8
-%endif
- dec rcx
- jnz unpack_block1d16_h6_sse2_rowloop ; next row
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- ;RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_bilinear_predict16x16_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; int xoffset,
-; int yoffset,
-; unsigned char *dst_ptr,
-; int dst_pitch
-;)
-extern sym(vp8_bilinear_filters_mmx)
-global sym(vp8_bilinear_predict16x16_sse2)
-sym(vp8_bilinear_predict16x16_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ;const short *HFilter = bilinear_filters_mmx[xoffset]
- ;const short *VFilter = bilinear_filters_mmx[yoffset]
-
- lea rcx, [GLOBAL(sym(vp8_bilinear_filters_mmx))]
- movsxd rax, dword ptr arg(2) ;xoffset
-
- cmp rax, 0 ;skip first_pass filter if xoffset=0
- je b16x16_sp_only
-
- shl rax, 5
- add rax, rcx ;HFilter
-
- mov rdi, arg(4) ;dst_ptr
- mov rsi, arg(0) ;src_ptr
- movsxd rdx, dword ptr arg(5) ;dst_pitch
-
- movdqa xmm1, [rax]
- movdqa xmm2, [rax+16]
-
- movsxd rax, dword ptr arg(3) ;yoffset
-
- cmp rax, 0 ;skip second_pass filter if yoffset=0
- je b16x16_fp_only
-
- shl rax, 5
- add rax, rcx ;VFilter
-
- lea rcx, [rdi+rdx*8]
- lea rcx, [rcx+rdx*8]
- movsxd rdx, dword ptr arg(1) ;src_pixels_per_line
-
- pxor xmm0, xmm0
-
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(5) ;dst_pitch
-%endif
- ; get the first horizontal line done
- movdqu xmm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- movdqa xmm4, xmm3 ; make a copy of current line
-
- punpcklbw xmm3, xmm0 ; xx 00 01 02 03 04 05 06
- punpckhbw xmm4, xmm0
-
- pmullw xmm3, xmm1
- pmullw xmm4, xmm1
-
- movdqu xmm5, [rsi+1]
- movdqa xmm6, xmm5
-
- punpcklbw xmm5, xmm0
- punpckhbw xmm6, xmm0
-
- pmullw xmm5, xmm2
- pmullw xmm6, xmm2
-
- paddw xmm3, xmm5
- paddw xmm4, xmm6
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw xmm4, [GLOBAL(rd)]
- psraw xmm4, VP8_FILTER_SHIFT
-
- movdqa xmm7, xmm3
- packuswb xmm7, xmm4
-
- add rsi, rdx ; next line
-next_row:
- movdqu xmm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- movdqa xmm4, xmm3 ; make a copy of current line
-
- punpcklbw xmm3, xmm0 ; xx 00 01 02 03 04 05 06
- punpckhbw xmm4, xmm0
-
- pmullw xmm3, xmm1
- pmullw xmm4, xmm1
-
- movdqu xmm5, [rsi+1]
- movdqa xmm6, xmm5
-
- punpcklbw xmm5, xmm0
- punpckhbw xmm6, xmm0
-
- pmullw xmm5, xmm2
- pmullw xmm6, xmm2
-
- paddw xmm3, xmm5
- paddw xmm4, xmm6
-
- movdqa xmm5, xmm7
- movdqa xmm6, xmm7
-
- punpcklbw xmm5, xmm0
- punpckhbw xmm6, xmm0
-
- pmullw xmm5, [rax]
- pmullw xmm6, [rax]
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw xmm4, [GLOBAL(rd)]
- psraw xmm4, VP8_FILTER_SHIFT
-
- movdqa xmm7, xmm3
- packuswb xmm7, xmm4
-
- pmullw xmm3, [rax+16]
- pmullw xmm4, [rax+16]
-
- paddw xmm3, xmm5
- paddw xmm4, xmm6
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw xmm4, [GLOBAL(rd)]
- psraw xmm4, VP8_FILTER_SHIFT
-
- packuswb xmm3, xmm4
- movdqa [rdi], xmm3 ; store the results in the destination
-
- add rsi, rdx ; next line
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(5) ;dst_pitch
-%else
- add rdi, r8
-%endif
-
- cmp rdi, rcx
- jne next_row
-
- jmp done
-
-b16x16_sp_only:
- movsxd rax, dword ptr arg(3) ;yoffset
- shl rax, 5
- add rax, rcx ;VFilter
-
- mov rdi, arg(4) ;dst_ptr
- mov rsi, arg(0) ;src_ptr
- movsxd rdx, dword ptr arg(5) ;dst_pitch
-
- movdqa xmm1, [rax]
- movdqa xmm2, [rax+16]
-
- lea rcx, [rdi+rdx*8]
- lea rcx, [rcx+rdx*8]
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line
-
- pxor xmm0, xmm0
-
- ; get the first horizontal line done
- movdqu xmm7, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
-
- add rsi, rax ; next line
-next_row_spo:
- movdqu xmm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
-
- movdqa xmm5, xmm7
- movdqa xmm6, xmm7
-
- movdqa xmm4, xmm3 ; make a copy of current line
- movdqa xmm7, xmm3
-
- punpcklbw xmm5, xmm0
- punpckhbw xmm6, xmm0
- punpcklbw xmm3, xmm0 ; xx 00 01 02 03 04 05 06
- punpckhbw xmm4, xmm0
-
- pmullw xmm5, xmm1
- pmullw xmm6, xmm1
- pmullw xmm3, xmm2
- pmullw xmm4, xmm2
-
- paddw xmm3, xmm5
- paddw xmm4, xmm6
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw xmm4, [GLOBAL(rd)]
- psraw xmm4, VP8_FILTER_SHIFT
-
- packuswb xmm3, xmm4
- movdqa [rdi], xmm3 ; store the results in the destination
-
- add rsi, rax ; next line
- add rdi, rdx ;dst_pitch
- cmp rdi, rcx
- jne next_row_spo
-
- jmp done
-
-b16x16_fp_only:
- lea rcx, [rdi+rdx*8]
- lea rcx, [rcx+rdx*8]
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line
- pxor xmm0, xmm0
-
-next_row_fpo:
- movdqu xmm3, [rsi] ; xx 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14
- movdqa xmm4, xmm3 ; make a copy of current line
-
- punpcklbw xmm3, xmm0 ; xx 00 01 02 03 04 05 06
- punpckhbw xmm4, xmm0
-
- pmullw xmm3, xmm1
- pmullw xmm4, xmm1
-
- movdqu xmm5, [rsi+1]
- movdqa xmm6, xmm5
-
- punpcklbw xmm5, xmm0
- punpckhbw xmm6, xmm0
-
- pmullw xmm5, xmm2
- pmullw xmm6, xmm2
-
- paddw xmm3, xmm5
- paddw xmm4, xmm6
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw xmm4, [GLOBAL(rd)]
- psraw xmm4, VP8_FILTER_SHIFT
-
- packuswb xmm3, xmm4
- movdqa [rdi], xmm3 ; store the results in the destination
-
- add rsi, rax ; next line
- add rdi, rdx ; dst_pitch
- cmp rdi, rcx
- jne next_row_fpo
-
-done:
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_bilinear_predict8x8_sse2
-;(
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; int xoffset,
-; int yoffset,
-; unsigned char *dst_ptr,
-; int dst_pitch
-;)
-extern sym(vp8_bilinear_filters_mmx)
-global sym(vp8_bilinear_predict8x8_sse2)
-sym(vp8_bilinear_predict8x8_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 144 ; reserve 144 bytes
-
- ;const short *HFilter = bilinear_filters_mmx[xoffset]
- ;const short *VFilter = bilinear_filters_mmx[yoffset]
- lea rcx, [GLOBAL(sym(vp8_bilinear_filters_mmx))]
-
- mov rsi, arg(0) ;src_ptr
- movsxd rdx, dword ptr arg(1) ;src_pixels_per_line
-
- ;Read 9-line unaligned data in and put them on stack. This gives a big
- ;performance boost.
- movdqu xmm0, [rsi]
- lea rax, [rdx + rdx*2]
- movdqu xmm1, [rsi+rdx]
- movdqu xmm2, [rsi+rdx*2]
- add rsi, rax
- movdqu xmm3, [rsi]
- movdqu xmm4, [rsi+rdx]
- movdqu xmm5, [rsi+rdx*2]
- add rsi, rax
- movdqu xmm6, [rsi]
- movdqu xmm7, [rsi+rdx]
-
- movdqa XMMWORD PTR [rsp], xmm0
-
- movdqu xmm0, [rsi+rdx*2]
-
- movdqa XMMWORD PTR [rsp+16], xmm1
- movdqa XMMWORD PTR [rsp+32], xmm2
- movdqa XMMWORD PTR [rsp+48], xmm3
- movdqa XMMWORD PTR [rsp+64], xmm4
- movdqa XMMWORD PTR [rsp+80], xmm5
- movdqa XMMWORD PTR [rsp+96], xmm6
- movdqa XMMWORD PTR [rsp+112], xmm7
- movdqa XMMWORD PTR [rsp+128], xmm0
-
- movsxd rax, dword ptr arg(2) ;xoffset
- shl rax, 5
- add rax, rcx ;HFilter
-
- mov rdi, arg(4) ;dst_ptr
- movsxd rdx, dword ptr arg(5) ;dst_pitch
-
- movdqa xmm1, [rax]
- movdqa xmm2, [rax+16]
-
- movsxd rax, dword ptr arg(3) ;yoffset
- shl rax, 5
- add rax, rcx ;VFilter
-
- lea rcx, [rdi+rdx*8]
-
- movdqa xmm5, [rax]
- movdqa xmm6, [rax+16]
-
- pxor xmm0, xmm0
-
- ; get the first horizontal line done
- movdqa xmm3, XMMWORD PTR [rsp]
- movdqa xmm4, xmm3 ; make a copy of current line
- psrldq xmm4, 1
-
- punpcklbw xmm3, xmm0 ; 00 01 02 03 04 05 06 07
- punpcklbw xmm4, xmm0 ; 01 02 03 04 05 06 07 08
-
- pmullw xmm3, xmm1
- pmullw xmm4, xmm2
-
- paddw xmm3, xmm4
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- movdqa xmm7, xmm3
- add rsp, 16 ; next line
-next_row8x8:
- movdqa xmm3, XMMWORD PTR [rsp] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
- movdqa xmm4, xmm3 ; make a copy of current line
- psrldq xmm4, 1
-
- punpcklbw xmm3, xmm0 ; 00 01 02 03 04 05 06 07
- punpcklbw xmm4, xmm0 ; 01 02 03 04 05 06 07 08
-
- pmullw xmm3, xmm1
- pmullw xmm4, xmm2
-
- paddw xmm3, xmm4
- pmullw xmm7, xmm5
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- movdqa xmm4, xmm3
-
- pmullw xmm3, xmm6
- paddw xmm3, xmm7
-
- movdqa xmm7, xmm4
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- packuswb xmm3, xmm0
- movq [rdi], xmm3 ; store the results in the destination
-
- add rsp, 16 ; next line
- add rdi, rdx
-
- cmp rdi, rcx
- jne next_row8x8
-
- ;add rsp, 144
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-SECTION_RODATA
-align 16
-rd:
- times 8 dw 0x40
diff --git a/vp8/common/x86/subpixel_ssse3.asm b/vp8/common/x86/subpixel_ssse3.asm
deleted file mode 100644
index 7f6fd93..0000000
--- a/vp8/common/x86/subpixel_ssse3.asm
+++ /dev/null
@@ -1,1554 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%define BLOCK_HEIGHT_WIDTH 4
-%define VP8_FILTER_WEIGHT 128
-%define VP8_FILTER_SHIFT 7
-
-
-;/************************************************************************************
-; Notes: filter_block1d_h6 applies a 6 tap filter horizontally to the input pixels. The
-; input pixel array has output_height rows. This routine assumes that output_height is an
-; even number. This function handles 8 pixels in horizontal direction, calculating ONE
-; rows each iteration to take advantage of the 128 bits operations.
-;
-; This is an implementation of some of the SSE optimizations first seen in ffvp8
-;
-;*************************************************************************************/
-;void vp8_filter_block1d8_h6_ssse3
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned char *output_ptr,
-; unsigned int output_pitch,
-; unsigned int output_height,
-; unsigned int vp8_filter_index
-;)
-global sym(vp8_filter_block1d8_h6_ssse3)
-sym(vp8_filter_block1d8_h6_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- movsxd rdx, DWORD PTR arg(5) ;table index
- xor rsi, rsi
- shl rdx, 4
-
- movdqa xmm7, [GLOBAL(rd)]
-
- lea rax, [GLOBAL(k0_k5)]
- add rax, rdx
- mov rdi, arg(2) ;output_ptr
-
- cmp esi, DWORD PTR [rax]
- je vp8_filter_block1d8_h4_ssse3
-
- movdqa xmm4, XMMWORD PTR [rax] ;k0_k5
- movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line
- movsxd rcx, dword ptr arg(4) ;output_height
-
- movsxd rdx, dword ptr arg(3) ;output_pitch
-
- sub rdi, rdx
-;xmm3 free
-filter_block1d8_h6_rowloop_ssse3:
- movq xmm0, MMWORD PTR [rsi - 2] ; -2 -1 0 1 2 3 4 5
-
- movq xmm2, MMWORD PTR [rsi + 3] ; 3 4 5 6 7 8 9 10
-
- punpcklbw xmm0, xmm2 ; -2 3 -1 4 0 5 1 6 2 7 3 8 4 9 5 10
-
- movdqa xmm1, xmm0
- pmaddubsw xmm0, xmm4
-
- movdqa xmm2, xmm1
- pshufb xmm1, [GLOBAL(shuf2bfrom1)]
-
- pshufb xmm2, [GLOBAL(shuf3bfrom1)]
- pmaddubsw xmm1, xmm5
-
- lea rdi, [rdi + rdx]
- pmaddubsw xmm2, xmm6
-
- lea rsi, [rsi + rax]
- dec rcx
-
- paddsw xmm0, xmm1
- paddsw xmm2, xmm7
-
- paddsw xmm0, xmm2
-
- psraw xmm0, 7
-
- packuswb xmm0, xmm0
-
- movq MMWORD Ptr [rdi], xmm0
- jnz filter_block1d8_h6_rowloop_ssse3
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-vp8_filter_block1d8_h4_ssse3:
- movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
-
- movdqa xmm3, XMMWORD PTR [GLOBAL(shuf2bfrom1)]
- movdqa xmm4, XMMWORD PTR [GLOBAL(shuf3bfrom1)]
-
- mov rsi, arg(0) ;src_ptr
-
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line
- movsxd rcx, dword ptr arg(4) ;output_height
-
- movsxd rdx, dword ptr arg(3) ;output_pitch
-
- sub rdi, rdx
-
-filter_block1d8_h4_rowloop_ssse3:
- movq xmm0, MMWORD PTR [rsi - 2] ; -2 -1 0 1 2 3 4 5
-
- movq xmm1, MMWORD PTR [rsi + 3] ; 3 4 5 6 7 8 9 10
-
- punpcklbw xmm0, xmm1 ; -2 3 -1 4 0 5 1 6 2 7 3 8 4 9 5 10
-
- movdqa xmm2, xmm0
- pshufb xmm0, xmm3
-
- pshufb xmm2, xmm4
- pmaddubsw xmm0, xmm5
-
- lea rdi, [rdi + rdx]
- pmaddubsw xmm2, xmm6
-
- lea rsi, [rsi + rax]
- dec rcx
-
- paddsw xmm0, xmm7
-
- paddsw xmm0, xmm2
-
- psraw xmm0, 7
-
- packuswb xmm0, xmm0
-
- movq MMWORD Ptr [rdi], xmm0
-
- jnz filter_block1d8_h4_rowloop_ssse3
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-;void vp8_filter_block1d16_h6_ssse3
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned char *output_ptr,
-; unsigned int output_pitch,
-; unsigned int output_height,
-; unsigned int vp8_filter_index
-;)
-global sym(vp8_filter_block1d16_h6_ssse3)
-sym(vp8_filter_block1d16_h6_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- movsxd rdx, DWORD PTR arg(5) ;table index
- xor rsi, rsi
- shl rdx, 4 ;
-
- lea rax, [GLOBAL(k0_k5)]
- add rax, rdx
-
- mov rdi, arg(2) ;output_ptr
-
-;;
-;; cmp esi, DWORD PTR [rax]
-;; je vp8_filter_block1d16_h4_ssse3
-
- mov rsi, arg(0) ;src_ptr
-
- movdqa xmm4, XMMWORD PTR [rax] ;k0_k5
- movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
-
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line
- movsxd rcx, dword ptr arg(4) ;output_height
- movsxd rdx, dword ptr arg(3) ;output_pitch
-
-filter_block1d16_h6_rowloop_ssse3:
- movq xmm0, MMWORD PTR [rsi - 2] ; -2 -1 0 1 2 3 4 5
-
- movq xmm3, MMWORD PTR [rsi + 3] ; 3 4 5 6 7 8 9 10
-
- punpcklbw xmm0, xmm3 ; -2 3 -1 4 0 5 1 6 2 7 3 8 4 9 5 10
-
- movdqa xmm1, xmm0
- pmaddubsw xmm0, xmm4
-
- movdqa xmm2, xmm1
- pshufb xmm1, [GLOBAL(shuf2bfrom1)]
-
- pshufb xmm2, [GLOBAL(shuf3bfrom1)]
- movq xmm3, MMWORD PTR [rsi + 6]
-
- pmaddubsw xmm1, xmm5
- movq xmm7, MMWORD PTR [rsi + 11]
-
- pmaddubsw xmm2, xmm6
- punpcklbw xmm3, xmm7
-
- paddsw xmm0, xmm1
- movdqa xmm1, xmm3
-
- pmaddubsw xmm3, xmm4
- paddsw xmm0, xmm2
-
- movdqa xmm2, xmm1
- paddsw xmm0, [GLOBAL(rd)]
-
- pshufb xmm1, [GLOBAL(shuf2bfrom1)]
- pshufb xmm2, [GLOBAL(shuf3bfrom1)]
-
- psraw xmm0, 7
- pmaddubsw xmm1, xmm5
-
- pmaddubsw xmm2, xmm6
- packuswb xmm0, xmm0
-
- lea rsi, [rsi + rax]
- paddsw xmm3, xmm1
-
- paddsw xmm3, xmm2
-
- paddsw xmm3, [GLOBAL(rd)]
-
- psraw xmm3, 7
-
- packuswb xmm3, xmm3
-
- punpcklqdq xmm0, xmm3
-
- movdqa XMMWORD Ptr [rdi], xmm0
-
- lea rdi, [rdi + rdx]
- dec rcx
- jnz filter_block1d16_h6_rowloop_ssse3
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-vp8_filter_block1d16_h4_ssse3:
- movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
-
- mov rsi, arg(0) ;src_ptr
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line
- movsxd rcx, dword ptr arg(4) ;output_height
- movsxd rdx, dword ptr arg(3) ;output_pitch
-
-filter_block1d16_h4_rowloop_ssse3:
- movdqu xmm1, XMMWORD PTR [rsi - 2]
-
- movdqa xmm2, xmm1
- pshufb xmm1, [GLOBAL(shuf2b)]
- pshufb xmm2, [GLOBAL(shuf3b)]
- pmaddubsw xmm1, xmm5
-
- movdqu xmm3, XMMWORD PTR [rsi + 6]
-
- pmaddubsw xmm2, xmm6
- movdqa xmm0, xmm3
- pshufb xmm3, [GLOBAL(shuf3b)]
- pshufb xmm0, [GLOBAL(shuf2b)]
-
- paddsw xmm1, [GLOBAL(rd)]
- paddsw xmm1, xmm2
-
- pmaddubsw xmm0, xmm5
- pmaddubsw xmm3, xmm6
-
- psraw xmm1, 7
- packuswb xmm1, xmm1
- lea rsi, [rsi + rax]
- paddsw xmm3, xmm0
- paddsw xmm3, [GLOBAL(rd)]
- psraw xmm3, 7
- packuswb xmm3, xmm3
-
- punpcklqdq xmm1, xmm3
-
- movdqa XMMWORD Ptr [rdi], xmm1
-
- add rdi, rdx
- dec rcx
- jnz filter_block1d16_h4_rowloop_ssse3
-
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_filter_block1d4_h6_ssse3
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pixels_per_line,
-; unsigned char *output_ptr,
-; unsigned int output_pitch,
-; unsigned int output_height,
-; unsigned int vp8_filter_index
-;)
-global sym(vp8_filter_block1d4_h6_ssse3)
-sym(vp8_filter_block1d4_h6_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- movsxd rdx, DWORD PTR arg(5) ;table index
- xor rsi, rsi
- shl rdx, 4 ;
-
- lea rax, [GLOBAL(k0_k5)]
- add rax, rdx
- movdqa xmm7, [GLOBAL(rd)]
-
- cmp esi, DWORD PTR [rax]
- je vp8_filter_block1d4_h4_ssse3
-
- movdqa xmm4, XMMWORD PTR [rax] ;k0_k5
- movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;output_ptr
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line
- movsxd rcx, dword ptr arg(4) ;output_height
-
- movsxd rdx, dword ptr arg(3) ;output_pitch
-
-;xmm3 free
-filter_block1d4_h6_rowloop_ssse3:
- movdqu xmm0, XMMWORD PTR [rsi - 2]
-
- movdqa xmm1, xmm0
- pshufb xmm0, [GLOBAL(shuf1b)]
-
- movdqa xmm2, xmm1
- pshufb xmm1, [GLOBAL(shuf2b)]
- pmaddubsw xmm0, xmm4
- pshufb xmm2, [GLOBAL(shuf3b)]
- pmaddubsw xmm1, xmm5
-
-;--
- pmaddubsw xmm2, xmm6
-
- lea rsi, [rsi + rax]
-;--
- paddsw xmm0, xmm1
- paddsw xmm0, xmm7
- pxor xmm1, xmm1
- paddsw xmm0, xmm2
- psraw xmm0, 7
- packuswb xmm0, xmm0
-
- movd DWORD PTR [rdi], xmm0
-
- add rdi, rdx
- dec rcx
- jnz filter_block1d4_h6_rowloop_ssse3
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-vp8_filter_block1d4_h4_ssse3:
- movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3
- movdqa xmm0, XMMWORD PTR [GLOBAL(shuf2b)]
- movdqa xmm3, XMMWORD PTR [GLOBAL(shuf3b)]
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;output_ptr
- movsxd rax, dword ptr arg(1) ;src_pixels_per_line
- movsxd rcx, dword ptr arg(4) ;output_height
-
- movsxd rdx, dword ptr arg(3) ;output_pitch
-
-filter_block1d4_h4_rowloop_ssse3:
- movdqu xmm1, XMMWORD PTR [rsi - 2]
-
- movdqa xmm2, xmm1
- pshufb xmm1, xmm0 ;;[GLOBAL(shuf2b)]
- pshufb xmm2, xmm3 ;;[GLOBAL(shuf3b)]
- pmaddubsw xmm1, xmm5
-
-;--
- pmaddubsw xmm2, xmm6
-
- lea rsi, [rsi + rax]
-;--
- paddsw xmm1, xmm7
- paddsw xmm1, xmm2
- psraw xmm1, 7
- packuswb xmm1, xmm1
-
- movd DWORD PTR [rdi], xmm1
-
- add rdi, rdx
- dec rcx
- jnz filter_block1d4_h4_rowloop_ssse3
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-;void vp8_filter_block1d16_v6_ssse3
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pitch,
-; unsigned char *output_ptr,
-; unsigned int out_pitch,
-; unsigned int output_height,
-; unsigned int vp8_filter_index
-;)
-global sym(vp8_filter_block1d16_v6_ssse3)
-sym(vp8_filter_block1d16_v6_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- movsxd rdx, DWORD PTR arg(5) ;table index
- xor rsi, rsi
- shl rdx, 4 ;
-
- lea rax, [GLOBAL(k0_k5)]
- add rax, rdx
-
- cmp esi, DWORD PTR [rax]
- je vp8_filter_block1d16_v4_ssse3
-
- movdqa xmm5, XMMWORD PTR [rax] ;k0_k5
- movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3
-
- mov rsi, arg(0) ;src_ptr
- movsxd rdx, DWORD PTR arg(1) ;pixels_per_line
- mov rdi, arg(2) ;output_ptr
-
-%if ABI_IS_32BIT=0
- movsxd r8, DWORD PTR arg(3) ;out_pitch
-%endif
- mov rax, rsi
- movsxd rcx, DWORD PTR arg(4) ;output_height
- add rax, rdx
-
-
-vp8_filter_block1d16_v6_ssse3_loop:
- movq xmm1, MMWORD PTR [rsi] ;A
- movq xmm2, MMWORD PTR [rsi + rdx] ;B
- movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C
- movq xmm4, MMWORD PTR [rax + rdx * 2] ;D
- movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E
-
- punpcklbw xmm2, xmm4 ;B D
- punpcklbw xmm3, xmm0 ;C E
-
- movq xmm0, MMWORD PTR [rax + rdx * 4] ;F
-
- pmaddubsw xmm3, xmm6
- punpcklbw xmm1, xmm0 ;A F
- pmaddubsw xmm2, xmm7
- pmaddubsw xmm1, xmm5
-
- paddsw xmm2, xmm3
- paddsw xmm2, xmm1
- paddsw xmm2, [GLOBAL(rd)]
- psraw xmm2, 7
- packuswb xmm2, xmm2
-
- movq MMWORD PTR [rdi], xmm2 ;store the results
-
- movq xmm1, MMWORD PTR [rsi + 8] ;A
- movq xmm2, MMWORD PTR [rsi + rdx + 8] ;B
- movq xmm3, MMWORD PTR [rsi + rdx * 2 + 8] ;C
- movq xmm4, MMWORD PTR [rax + rdx * 2 + 8] ;D
- movq xmm0, MMWORD PTR [rsi + rdx * 4 + 8] ;E
-
- punpcklbw xmm2, xmm4 ;B D
- punpcklbw xmm3, xmm0 ;C E
-
- movq xmm0, MMWORD PTR [rax + rdx * 4 + 8] ;F
- pmaddubsw xmm3, xmm6
- punpcklbw xmm1, xmm0 ;A F
- pmaddubsw xmm2, xmm7
- pmaddubsw xmm1, xmm5
-
- add rsi, rdx
- add rax, rdx
-;--
-;--
- paddsw xmm2, xmm3
- paddsw xmm2, xmm1
- paddsw xmm2, [GLOBAL(rd)]
- psraw xmm2, 7
- packuswb xmm2, xmm2
-
- movq MMWORD PTR [rdi+8], xmm2
-
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(3) ;out_pitch
-%else
- add rdi, r8
-%endif
- dec rcx
- jnz vp8_filter_block1d16_v6_ssse3_loop
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-vp8_filter_block1d16_v4_ssse3:
- movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3
-
- mov rsi, arg(0) ;src_ptr
- movsxd rdx, DWORD PTR arg(1) ;pixels_per_line
- mov rdi, arg(2) ;output_ptr
-
-%if ABI_IS_32BIT=0
- movsxd r8, DWORD PTR arg(3) ;out_pitch
-%endif
- mov rax, rsi
- movsxd rcx, DWORD PTR arg(4) ;output_height
- add rax, rdx
-
-vp8_filter_block1d16_v4_ssse3_loop:
- movq xmm2, MMWORD PTR [rsi + rdx] ;B
- movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C
- movq xmm4, MMWORD PTR [rax + rdx * 2] ;D
- movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E
-
- punpcklbw xmm2, xmm4 ;B D
- punpcklbw xmm3, xmm0 ;C E
-
- pmaddubsw xmm3, xmm6
- pmaddubsw xmm2, xmm7
- movq xmm5, MMWORD PTR [rsi + rdx + 8] ;B
- movq xmm1, MMWORD PTR [rsi + rdx * 2 + 8] ;C
- movq xmm4, MMWORD PTR [rax + rdx * 2 + 8] ;D
- movq xmm0, MMWORD PTR [rsi + rdx * 4 + 8] ;E
-
- paddsw xmm2, [GLOBAL(rd)]
- paddsw xmm2, xmm3
- psraw xmm2, 7
- packuswb xmm2, xmm2
-
- punpcklbw xmm5, xmm4 ;B D
- punpcklbw xmm1, xmm0 ;C E
-
- pmaddubsw xmm1, xmm6
- pmaddubsw xmm5, xmm7
-
- movdqa xmm4, [GLOBAL(rd)]
- add rsi, rdx
- add rax, rdx
-;--
-;--
- paddsw xmm5, xmm1
- paddsw xmm5, xmm4
- psraw xmm5, 7
- packuswb xmm5, xmm5
-
- punpcklqdq xmm2, xmm5
-
- movdqa XMMWORD PTR [rdi], xmm2
-
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(3) ;out_pitch
-%else
- add rdi, r8
-%endif
- dec rcx
- jnz vp8_filter_block1d16_v4_ssse3_loop
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_filter_block1d8_v6_ssse3
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pitch,
-; unsigned char *output_ptr,
-; unsigned int out_pitch,
-; unsigned int output_height,
-; unsigned int vp8_filter_index
-;)
-global sym(vp8_filter_block1d8_v6_ssse3)
-sym(vp8_filter_block1d8_v6_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- movsxd rdx, DWORD PTR arg(5) ;table index
- xor rsi, rsi
- shl rdx, 4 ;
-
- lea rax, [GLOBAL(k0_k5)]
- add rax, rdx
-
- movsxd rdx, DWORD PTR arg(1) ;pixels_per_line
- mov rdi, arg(2) ;output_ptr
-%if ABI_IS_32BIT=0
- movsxd r8, DWORD PTR arg(3) ; out_pitch
-%endif
- movsxd rcx, DWORD PTR arg(4) ;[output_height]
-
- cmp esi, DWORD PTR [rax]
- je vp8_filter_block1d8_v4_ssse3
-
- movdqa xmm5, XMMWORD PTR [rax] ;k0_k5
- movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3
-
- mov rsi, arg(0) ;src_ptr
-
- mov rax, rsi
- add rax, rdx
-
-vp8_filter_block1d8_v6_ssse3_loop:
- movq xmm1, MMWORD PTR [rsi] ;A
- movq xmm2, MMWORD PTR [rsi + rdx] ;B
- movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C
- movq xmm4, MMWORD PTR [rax + rdx * 2] ;D
- movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E
-
- punpcklbw xmm2, xmm4 ;B D
- punpcklbw xmm3, xmm0 ;C E
-
- movq xmm0, MMWORD PTR [rax + rdx * 4] ;F
- movdqa xmm4, [GLOBAL(rd)]
-
- pmaddubsw xmm3, xmm6
- punpcklbw xmm1, xmm0 ;A F
- pmaddubsw xmm2, xmm7
- pmaddubsw xmm1, xmm5
- add rsi, rdx
- add rax, rdx
-;--
-;--
- paddsw xmm2, xmm3
- paddsw xmm2, xmm1
- paddsw xmm2, xmm4
- psraw xmm2, 7
- packuswb xmm2, xmm2
-
- movq MMWORD PTR [rdi], xmm2
-
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(3) ;[out_pitch]
-%else
- add rdi, r8
-%endif
- dec rcx
- jnz vp8_filter_block1d8_v6_ssse3_loop
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-vp8_filter_block1d8_v4_ssse3:
- movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4
- movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3
- movdqa xmm5, [GLOBAL(rd)]
-
- mov rsi, arg(0) ;src_ptr
-
- mov rax, rsi
- add rax, rdx
-
-vp8_filter_block1d8_v4_ssse3_loop:
- movq xmm2, MMWORD PTR [rsi + rdx] ;B
- movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C
- movq xmm4, MMWORD PTR [rax + rdx * 2] ;D
- movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E
-
- punpcklbw xmm2, xmm4 ;B D
- punpcklbw xmm3, xmm0 ;C E
-
- pmaddubsw xmm3, xmm6
- pmaddubsw xmm2, xmm7
- add rsi, rdx
- add rax, rdx
-;--
-;--
- paddsw xmm2, xmm3
- paddsw xmm2, xmm5
- psraw xmm2, 7
- packuswb xmm2, xmm2
-
- movq MMWORD PTR [rdi], xmm2
-
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(3) ;[out_pitch]
-%else
- add rdi, r8
-%endif
- dec rcx
- jnz vp8_filter_block1d8_v4_ssse3_loop
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-;void vp8_filter_block1d4_v6_ssse3
-;(
-; unsigned char *src_ptr,
-; unsigned int src_pitch,
-; unsigned char *output_ptr,
-; unsigned int out_pitch,
-; unsigned int output_height,
-; unsigned int vp8_filter_index
-;)
-global sym(vp8_filter_block1d4_v6_ssse3)
-sym(vp8_filter_block1d4_v6_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- movsxd rdx, DWORD PTR arg(5) ;table index
- xor rsi, rsi
- shl rdx, 4 ;
-
- lea rax, [GLOBAL(k0_k5)]
- add rax, rdx
-
- movsxd rdx, DWORD PTR arg(1) ;pixels_per_line
- mov rdi, arg(2) ;output_ptr
-%if ABI_IS_32BIT=0
- movsxd r8, DWORD PTR arg(3) ; out_pitch
-%endif
- movsxd rcx, DWORD PTR arg(4) ;[output_height]
-
- cmp esi, DWORD PTR [rax]
- je vp8_filter_block1d4_v4_ssse3
-
- movq mm5, MMWORD PTR [rax] ;k0_k5
- movq mm6, MMWORD PTR [rax+256] ;k2_k4
- movq mm7, MMWORD PTR [rax+128] ;k1_k3
-
- mov rsi, arg(0) ;src_ptr
-
- mov rax, rsi
- add rax, rdx
-
-vp8_filter_block1d4_v6_ssse3_loop:
- movd mm1, DWORD PTR [rsi] ;A
- movd mm2, DWORD PTR [rsi + rdx] ;B
- movd mm3, DWORD PTR [rsi + rdx * 2] ;C
- movd mm4, DWORD PTR [rax + rdx * 2] ;D
- movd mm0, DWORD PTR [rsi + rdx * 4] ;E
-
- punpcklbw mm2, mm4 ;B D
- punpcklbw mm3, mm0 ;C E
-
- movd mm0, DWORD PTR [rax + rdx * 4] ;F
-
- movq mm4, [GLOBAL(rd)]
-
- pmaddubsw mm3, mm6
- punpcklbw mm1, mm0 ;A F
- pmaddubsw mm2, mm7
- pmaddubsw mm1, mm5
- add rsi, rdx
- add rax, rdx
-;--
-;--
- paddsw mm2, mm3
- paddsw mm2, mm1
- paddsw mm2, mm4
- psraw mm2, 7
- packuswb mm2, mm2
-
- movd DWORD PTR [rdi], mm2
-
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(3) ;[out_pitch]
-%else
- add rdi, r8
-%endif
- dec rcx
- jnz vp8_filter_block1d4_v6_ssse3_loop
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-vp8_filter_block1d4_v4_ssse3:
- movq mm6, MMWORD PTR [rax+256] ;k2_k4
- movq mm7, MMWORD PTR [rax+128] ;k1_k3
- movq mm5, MMWORD PTR [GLOBAL(rd)]
-
- mov rsi, arg(0) ;src_ptr
-
- mov rax, rsi
- add rax, rdx
-
-vp8_filter_block1d4_v4_ssse3_loop:
- movd mm2, DWORD PTR [rsi + rdx] ;B
- movd mm3, DWORD PTR [rsi + rdx * 2] ;C
- movd mm4, DWORD PTR [rax + rdx * 2] ;D
- movd mm0, DWORD PTR [rsi + rdx * 4] ;E
-
- punpcklbw mm2, mm4 ;B D
- punpcklbw mm3, mm0 ;C E
-
- pmaddubsw mm3, mm6
- pmaddubsw mm2, mm7
- add rsi, rdx
- add rax, rdx
-;--
-;--
- paddsw mm2, mm3
- paddsw mm2, mm5
- psraw mm2, 7
- packuswb mm2, mm2
-
- movd DWORD PTR [rdi], mm2
-
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(3) ;[out_pitch]
-%else
- add rdi, r8
-%endif
- dec rcx
- jnz vp8_filter_block1d4_v4_ssse3_loop
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_bilinear_predict16x16_ssse3
-;(
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; int xoffset,
-; int yoffset,
-; unsigned char *dst_ptr,
-; int dst_pitch
-;)
-global sym(vp8_bilinear_predict16x16_ssse3)
-sym(vp8_bilinear_predict16x16_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- lea rcx, [GLOBAL(vp8_bilinear_filters_ssse3)]
- movsxd rax, dword ptr arg(2) ; xoffset
-
- cmp rax, 0 ; skip first_pass filter if xoffset=0
- je b16x16_sp_only
-
- shl rax, 4
- lea rax, [rax + rcx] ; HFilter
-
- mov rdi, arg(4) ; dst_ptr
- mov rsi, arg(0) ; src_ptr
- movsxd rdx, dword ptr arg(5) ; dst_pitch
-
- movdqa xmm1, [rax]
-
- movsxd rax, dword ptr arg(3) ; yoffset
-
- cmp rax, 0 ; skip second_pass filter if yoffset=0
- je b16x16_fp_only
-
- shl rax, 4
- lea rax, [rax + rcx] ; VFilter
-
- lea rcx, [rdi+rdx*8]
- lea rcx, [rcx+rdx*8]
- movsxd rdx, dword ptr arg(1) ; src_pixels_per_line
-
- movdqa xmm2, [rax]
-
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(5) ; dst_pitch
-%endif
- movq xmm3, [rsi] ; 00 01 02 03 04 05 06 07
- movq xmm5, [rsi+1] ; 01 02 03 04 05 06 07 08
-
- punpcklbw xmm3, xmm5 ; 00 01 01 02 02 03 03 04 04 05 05 06 06 07 07 08
- movq xmm4, [rsi+8] ; 08 09 10 11 12 13 14 15
-
- movq xmm5, [rsi+9] ; 09 10 11 12 13 14 15 16
-
- lea rsi, [rsi + rdx] ; next line
-
- pmaddubsw xmm3, xmm1 ; 00 02 04 06 08 10 12 14
-
- punpcklbw xmm4, xmm5 ; 08 09 09 10 10 11 11 12 12 13 13 14 14 15 15 16
- pmaddubsw xmm4, xmm1 ; 01 03 05 07 09 11 13 15
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- paddw xmm4, [GLOBAL(rd)] ; xmm4 += round value
- psraw xmm4, VP8_FILTER_SHIFT ; xmm4 /= 128
-
- movdqa xmm7, xmm3
- packuswb xmm7, xmm4 ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
-
-.next_row:
- movq xmm6, [rsi] ; 00 01 02 03 04 05 06 07
- movq xmm5, [rsi+1] ; 01 02 03 04 05 06 07 08
-
- punpcklbw xmm6, xmm5
- movq xmm4, [rsi+8] ; 08 09 10 11 12 13 14 15
-
- movq xmm5, [rsi+9] ; 09 10 11 12 13 14 15 16
- lea rsi, [rsi + rdx] ; next line
-
- pmaddubsw xmm6, xmm1
-
- punpcklbw xmm4, xmm5
- pmaddubsw xmm4, xmm1
-
- paddw xmm6, [GLOBAL(rd)] ; xmm6 += round value
- psraw xmm6, VP8_FILTER_SHIFT ; xmm6 /= 128
-
- paddw xmm4, [GLOBAL(rd)] ; xmm4 += round value
- psraw xmm4, VP8_FILTER_SHIFT ; xmm4 /= 128
-
- packuswb xmm6, xmm4
- movdqa xmm5, xmm7
-
- punpcklbw xmm5, xmm6
- pmaddubsw xmm5, xmm2
-
- punpckhbw xmm7, xmm6
- pmaddubsw xmm7, xmm2
-
- paddw xmm5, [GLOBAL(rd)] ; xmm5 += round value
- psraw xmm5, VP8_FILTER_SHIFT ; xmm5 /= 128
-
- paddw xmm7, [GLOBAL(rd)] ; xmm7 += round value
- psraw xmm7, VP8_FILTER_SHIFT ; xmm7 /= 128
-
- packuswb xmm5, xmm7
- movdqa xmm7, xmm6
-
- movdqa [rdi], xmm5 ; store the results in the destination
-%if ABI_IS_32BIT
- add rdi, DWORD PTR arg(5) ; dst_pitch
-%else
- add rdi, r8
-%endif
-
- cmp rdi, rcx
- jne .next_row
-
- jmp done
-
-b16x16_sp_only:
- movsxd rax, dword ptr arg(3) ; yoffset
- shl rax, 4
- lea rax, [rax + rcx] ; VFilter
-
- mov rdi, arg(4) ; dst_ptr
- mov rsi, arg(0) ; src_ptr
- movsxd rdx, dword ptr arg(5) ; dst_pitch
-
- movdqa xmm1, [rax] ; VFilter
-
- lea rcx, [rdi+rdx*8]
- lea rcx, [rcx+rdx*8]
- movsxd rax, dword ptr arg(1) ; src_pixels_per_line
-
- ; get the first horizontal line done
- movq xmm4, [rsi] ; load row 0
- movq xmm2, [rsi + 8] ; load row 0
-
- lea rsi, [rsi + rax] ; next line
-.next_row:
- movq xmm3, [rsi] ; load row + 1
- movq xmm5, [rsi + 8] ; load row + 1
-
- punpcklbw xmm4, xmm3
- punpcklbw xmm2, xmm5
-
- pmaddubsw xmm4, xmm1
- movq xmm7, [rsi + rax] ; load row + 2
-
- pmaddubsw xmm2, xmm1
- movq xmm6, [rsi + rax + 8] ; load row + 2
-
- punpcklbw xmm3, xmm7
- punpcklbw xmm5, xmm6
-
- pmaddubsw xmm3, xmm1
- paddw xmm4, [GLOBAL(rd)]
-
- pmaddubsw xmm5, xmm1
- paddw xmm2, [GLOBAL(rd)]
-
- psraw xmm4, VP8_FILTER_SHIFT
- psraw xmm2, VP8_FILTER_SHIFT
-
- packuswb xmm4, xmm2
- paddw xmm3, [GLOBAL(rd)]
-
- movdqa [rdi], xmm4 ; store row 0
- paddw xmm5, [GLOBAL(rd)]
-
- psraw xmm3, VP8_FILTER_SHIFT
- psraw xmm5, VP8_FILTER_SHIFT
-
- packuswb xmm3, xmm5
- movdqa xmm4, xmm7
-
- movdqa [rdi + rdx],xmm3 ; store row 1
- lea rsi, [rsi + 2*rax]
-
- movdqa xmm2, xmm6
- lea rdi, [rdi + 2*rdx]
-
- cmp rdi, rcx
- jne .next_row
-
- jmp done
-
-b16x16_fp_only:
- lea rcx, [rdi+rdx*8]
- lea rcx, [rcx+rdx*8]
- movsxd rax, dword ptr arg(1) ; src_pixels_per_line
-
-.next_row:
- movq xmm2, [rsi] ; 00 01 02 03 04 05 06 07
- movq xmm4, [rsi+1] ; 01 02 03 04 05 06 07 08
-
- punpcklbw xmm2, xmm4
- movq xmm3, [rsi+8] ; 08 09 10 11 12 13 14 15
-
- pmaddubsw xmm2, xmm1
- movq xmm4, [rsi+9] ; 09 10 11 12 13 14 15 16
-
- lea rsi, [rsi + rax] ; next line
- punpcklbw xmm3, xmm4
-
- pmaddubsw xmm3, xmm1
- movq xmm5, [rsi]
-
- paddw xmm2, [GLOBAL(rd)]
- movq xmm7, [rsi+1]
-
- movq xmm6, [rsi+8]
- psraw xmm2, VP8_FILTER_SHIFT
-
- punpcklbw xmm5, xmm7
- movq xmm7, [rsi+9]
-
- paddw xmm3, [GLOBAL(rd)]
- pmaddubsw xmm5, xmm1
-
- psraw xmm3, VP8_FILTER_SHIFT
- punpcklbw xmm6, xmm7
-
- packuswb xmm2, xmm3
- pmaddubsw xmm6, xmm1
-
- movdqa [rdi], xmm2 ; store the results in the destination
- paddw xmm5, [GLOBAL(rd)]
-
- lea rdi, [rdi + rdx] ; dst_pitch
- psraw xmm5, VP8_FILTER_SHIFT
-
- paddw xmm6, [GLOBAL(rd)]
- psraw xmm6, VP8_FILTER_SHIFT
-
- packuswb xmm5, xmm6
- lea rsi, [rsi + rax] ; next line
-
- movdqa [rdi], xmm5 ; store the results in the destination
- lea rdi, [rdi + rdx] ; dst_pitch
-
- cmp rdi, rcx
-
- jne .next_row
-
-done:
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_bilinear_predict8x8_ssse3
-;(
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; int xoffset,
-; int yoffset,
-; unsigned char *dst_ptr,
-; int dst_pitch
-;)
-global sym(vp8_bilinear_predict8x8_ssse3)
-sym(vp8_bilinear_predict8x8_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- ALIGN_STACK 16, rax
- sub rsp, 144 ; reserve 144 bytes
-
- lea rcx, [GLOBAL(vp8_bilinear_filters_ssse3)]
-
- mov rsi, arg(0) ;src_ptr
- movsxd rdx, dword ptr arg(1) ;src_pixels_per_line
-
- ;Read 9-line unaligned data in and put them on stack. This gives a big
- ;performance boost.
- movdqu xmm0, [rsi]
- lea rax, [rdx + rdx*2]
- movdqu xmm1, [rsi+rdx]
- movdqu xmm2, [rsi+rdx*2]
- add rsi, rax
- movdqu xmm3, [rsi]
- movdqu xmm4, [rsi+rdx]
- movdqu xmm5, [rsi+rdx*2]
- add rsi, rax
- movdqu xmm6, [rsi]
- movdqu xmm7, [rsi+rdx]
-
- movdqa XMMWORD PTR [rsp], xmm0
-
- movdqu xmm0, [rsi+rdx*2]
-
- movdqa XMMWORD PTR [rsp+16], xmm1
- movdqa XMMWORD PTR [rsp+32], xmm2
- movdqa XMMWORD PTR [rsp+48], xmm3
- movdqa XMMWORD PTR [rsp+64], xmm4
- movdqa XMMWORD PTR [rsp+80], xmm5
- movdqa XMMWORD PTR [rsp+96], xmm6
- movdqa XMMWORD PTR [rsp+112], xmm7
- movdqa XMMWORD PTR [rsp+128], xmm0
-
- movsxd rax, dword ptr arg(2) ; xoffset
- cmp rax, 0 ; skip first_pass filter if xoffset=0
- je b8x8_sp_only
-
- shl rax, 4
- add rax, rcx ; HFilter
-
- mov rdi, arg(4) ; dst_ptr
- movsxd rdx, dword ptr arg(5) ; dst_pitch
-
- movdqa xmm0, [rax]
-
- movsxd rax, dword ptr arg(3) ; yoffset
- cmp rax, 0 ; skip second_pass filter if yoffset=0
- je b8x8_fp_only
-
- shl rax, 4
- lea rax, [rax + rcx] ; VFilter
-
- lea rcx, [rdi+rdx*8]
-
- movdqa xmm1, [rax]
-
- ; get the first horizontal line done
- movdqa xmm3, [rsp] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
- movdqa xmm5, xmm3 ; 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 xx
-
- psrldq xmm5, 1
- lea rsp, [rsp + 16] ; next line
-
- punpcklbw xmm3, xmm5 ; 00 01 01 02 02 03 03 04 04 05 05 06 06 07 07 08
- pmaddubsw xmm3, xmm0 ; 00 02 04 06 08 10 12 14
-
- paddw xmm3, [GLOBAL(rd)] ; xmm3 += round value
- psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128
-
- movdqa xmm7, xmm3
- packuswb xmm7, xmm7 ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
-
-.next_row:
- movdqa xmm6, [rsp] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
- lea rsp, [rsp + 16] ; next line
-
- movdqa xmm5, xmm6
-
- psrldq xmm5, 1
-
- punpcklbw xmm6, xmm5
- pmaddubsw xmm6, xmm0
-
- paddw xmm6, [GLOBAL(rd)] ; xmm6 += round value
- psraw xmm6, VP8_FILTER_SHIFT ; xmm6 /= 128
-
- packuswb xmm6, xmm6
-
- punpcklbw xmm7, xmm6
- pmaddubsw xmm7, xmm1
-
- paddw xmm7, [GLOBAL(rd)] ; xmm7 += round value
- psraw xmm7, VP8_FILTER_SHIFT ; xmm7 /= 128
-
- packuswb xmm7, xmm7
-
- movq [rdi], xmm7 ; store the results in the destination
- lea rdi, [rdi + rdx]
-
- movdqa xmm7, xmm6
-
- cmp rdi, rcx
- jne .next_row
-
- jmp done8x8
-
-b8x8_sp_only:
- movsxd rax, dword ptr arg(3) ; yoffset
- shl rax, 4
- lea rax, [rax + rcx] ; VFilter
-
- mov rdi, arg(4) ;dst_ptr
- movsxd rdx, dword ptr arg(5) ; dst_pitch
-
- movdqa xmm0, [rax] ; VFilter
-
- movq xmm1, XMMWORD PTR [rsp]
- movq xmm2, XMMWORD PTR [rsp+16]
-
- movq xmm3, XMMWORD PTR [rsp+32]
- punpcklbw xmm1, xmm2
-
- movq xmm4, XMMWORD PTR [rsp+48]
- punpcklbw xmm2, xmm3
-
- movq xmm5, XMMWORD PTR [rsp+64]
- punpcklbw xmm3, xmm4
-
- movq xmm6, XMMWORD PTR [rsp+80]
- punpcklbw xmm4, xmm5
-
- movq xmm7, XMMWORD PTR [rsp+96]
- punpcklbw xmm5, xmm6
-
- pmaddubsw xmm1, xmm0
- pmaddubsw xmm2, xmm0
-
- pmaddubsw xmm3, xmm0
- pmaddubsw xmm4, xmm0
-
- pmaddubsw xmm5, xmm0
- punpcklbw xmm6, xmm7
-
- pmaddubsw xmm6, xmm0
- paddw xmm1, [GLOBAL(rd)]
-
- paddw xmm2, [GLOBAL(rd)]
- psraw xmm1, VP8_FILTER_SHIFT
-
- paddw xmm3, [GLOBAL(rd)]
- psraw xmm2, VP8_FILTER_SHIFT
-
- paddw xmm4, [GLOBAL(rd)]
- psraw xmm3, VP8_FILTER_SHIFT
-
- paddw xmm5, [GLOBAL(rd)]
- psraw xmm4, VP8_FILTER_SHIFT
-
- paddw xmm6, [GLOBAL(rd)]
- psraw xmm5, VP8_FILTER_SHIFT
-
- psraw xmm6, VP8_FILTER_SHIFT
- packuswb xmm1, xmm1
-
- packuswb xmm2, xmm2
- movq [rdi], xmm1
-
- packuswb xmm3, xmm3
- movq [rdi+rdx], xmm2
-
- packuswb xmm4, xmm4
- movq xmm1, XMMWORD PTR [rsp+112]
-
- lea rdi, [rdi + 2*rdx]
- movq xmm2, XMMWORD PTR [rsp+128]
-
- packuswb xmm5, xmm5
- movq [rdi], xmm3
-
- packuswb xmm6, xmm6
- movq [rdi+rdx], xmm4
-
- lea rdi, [rdi + 2*rdx]
- punpcklbw xmm7, xmm1
-
- movq [rdi], xmm5
- pmaddubsw xmm7, xmm0
-
- movq [rdi+rdx], xmm6
- punpcklbw xmm1, xmm2
-
- pmaddubsw xmm1, xmm0
- paddw xmm7, [GLOBAL(rd)]
-
- psraw xmm7, VP8_FILTER_SHIFT
- paddw xmm1, [GLOBAL(rd)]
-
- psraw xmm1, VP8_FILTER_SHIFT
- packuswb xmm7, xmm7
-
- packuswb xmm1, xmm1
- lea rdi, [rdi + 2*rdx]
-
- movq [rdi], xmm7
-
- movq [rdi+rdx], xmm1
- lea rsp, [rsp + 144]
-
- jmp done8x8
-
-b8x8_fp_only:
- lea rcx, [rdi+rdx*8]
-
-.next_row:
- movdqa xmm1, XMMWORD PTR [rsp]
- movdqa xmm3, XMMWORD PTR [rsp+16]
-
- movdqa xmm2, xmm1
- movdqa xmm5, XMMWORD PTR [rsp+32]
-
- psrldq xmm2, 1
- movdqa xmm7, XMMWORD PTR [rsp+48]
-
- movdqa xmm4, xmm3
- psrldq xmm4, 1
-
- movdqa xmm6, xmm5
- psrldq xmm6, 1
-
- punpcklbw xmm1, xmm2
- pmaddubsw xmm1, xmm0
-
- punpcklbw xmm3, xmm4
- pmaddubsw xmm3, xmm0
-
- punpcklbw xmm5, xmm6
- pmaddubsw xmm5, xmm0
-
- movdqa xmm2, xmm7
- psrldq xmm2, 1
-
- punpcklbw xmm7, xmm2
- pmaddubsw xmm7, xmm0
-
- paddw xmm1, [GLOBAL(rd)]
- psraw xmm1, VP8_FILTER_SHIFT
-
- paddw xmm3, [GLOBAL(rd)]
- psraw xmm3, VP8_FILTER_SHIFT
-
- paddw xmm5, [GLOBAL(rd)]
- psraw xmm5, VP8_FILTER_SHIFT
-
- paddw xmm7, [GLOBAL(rd)]
- psraw xmm7, VP8_FILTER_SHIFT
-
- packuswb xmm1, xmm1
- packuswb xmm3, xmm3
-
- packuswb xmm5, xmm5
- movq [rdi], xmm1
-
- packuswb xmm7, xmm7
- movq [rdi+rdx], xmm3
-
- lea rdi, [rdi + 2*rdx]
- movq [rdi], xmm5
-
- lea rsp, [rsp + 4*16]
- movq [rdi+rdx], xmm7
-
- lea rdi, [rdi + 2*rdx]
- cmp rdi, rcx
-
- jne .next_row
-
- lea rsp, [rsp + 16]
-
-done8x8:
- ;add rsp, 144
- pop rsp
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-shuf1b:
- db 0, 5, 1, 6, 2, 7, 3, 8, 4, 9, 5, 10, 6, 11, 7, 12
-shuf2b:
- db 2, 4, 3, 5, 4, 6, 5, 7, 6, 8, 7, 9, 8, 10, 9, 11
-shuf3b:
- db 1, 3, 2, 4, 3, 5, 4, 6, 5, 7, 6, 8, 7, 9, 8, 10
-
-align 16
-shuf2bfrom1:
- db 4, 8, 6, 1, 8, 3, 1, 5, 3, 7, 5, 9, 7,11, 9,13
-align 16
-shuf3bfrom1:
- db 2, 6, 4, 8, 6, 1, 8, 3, 1, 5, 3, 7, 5, 9, 7,11
-
-align 16
-rd:
- times 8 dw 0x40
-
-align 16
-k0_k5:
- times 8 db 0, 0 ;placeholder
- times 8 db 0, 0
- times 8 db 2, 1
- times 8 db 0, 0
- times 8 db 3, 3
- times 8 db 0, 0
- times 8 db 1, 2
- times 8 db 0, 0
-k1_k3:
- times 8 db 0, 0 ;placeholder
- times 8 db -6, 12
- times 8 db -11, 36
- times 8 db -9, 50
- times 8 db -16, 77
- times 8 db -6, 93
- times 8 db -8, 108
- times 8 db -1, 123
-k2_k4:
- times 8 db 128, 0 ;placeholder
- times 8 db 123, -1
- times 8 db 108, -8
- times 8 db 93, -6
- times 8 db 77, -16
- times 8 db 50, -9
- times 8 db 36, -11
- times 8 db 12, -6
-align 16
-vp8_bilinear_filters_ssse3:
- times 8 db 128, 0
- times 8 db 112, 16
- times 8 db 96, 32
- times 8 db 80, 48
- times 8 db 64, 64
- times 8 db 48, 80
- times 8 db 32, 96
- times 8 db 16, 112
-
diff --git a/vp8/common/x86/subpixel_x86.h b/vp8/common/x86/subpixel_x86.h
deleted file mode 100644
index 75991cc..0000000
--- a/vp8/common/x86/subpixel_x86.h
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef SUBPIXEL_X86_H
-#define SUBPIXEL_X86_H
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-
-#if HAVE_MMX
-extern prototype_subpixel_predict(vp8_sixtap_predict16x16_mmx);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x8_mmx);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x4_mmx);
-extern prototype_subpixel_predict(vp8_sixtap_predict4x4_mmx);
-extern prototype_subpixel_predict(vp8_bilinear_predict16x16_mmx);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x8_mmx);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x4_mmx);
-extern prototype_subpixel_predict(vp8_bilinear_predict4x4_mmx);
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_subpix_sixtap16x16
-#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_mmx
-
-#undef vp8_subpix_sixtap8x8
-#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_mmx
-
-#undef vp8_subpix_sixtap8x4
-#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_mmx
-
-#undef vp8_subpix_sixtap4x4
-#define vp8_subpix_sixtap4x4 vp8_sixtap_predict4x4_mmx
-
-#undef vp8_subpix_bilinear16x16
-#define vp8_subpix_bilinear16x16 vp8_bilinear_predict16x16_mmx
-
-#undef vp8_subpix_bilinear8x8
-#define vp8_subpix_bilinear8x8 vp8_bilinear_predict8x8_mmx
-
-#undef vp8_subpix_bilinear8x4
-#define vp8_subpix_bilinear8x4 vp8_bilinear_predict8x4_mmx
-
-#undef vp8_subpix_bilinear4x4
-#define vp8_subpix_bilinear4x4 vp8_bilinear_predict4x4_mmx
-
-#endif
-#endif
-
-
-#if HAVE_SSE2
-extern prototype_subpixel_predict(vp8_sixtap_predict16x16_sse2);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x8_sse2);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x4_sse2);
-extern prototype_subpixel_predict(vp8_bilinear_predict16x16_sse2);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x8_sse2);
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_subpix_sixtap16x16
-#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_sse2
-
-#undef vp8_subpix_sixtap8x8
-#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_sse2
-
-#undef vp8_subpix_sixtap8x4
-#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_sse2
-
-#undef vp8_subpix_bilinear16x16
-#define vp8_subpix_bilinear16x16 vp8_bilinear_predict16x16_sse2
-
-#undef vp8_subpix_bilinear8x8
-#define vp8_subpix_bilinear8x8 vp8_bilinear_predict8x8_sse2
-
-#endif
-#endif
-
-#if HAVE_SSSE3
-extern prototype_subpixel_predict(vp8_sixtap_predict16x16_ssse3);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x8_ssse3);
-extern prototype_subpixel_predict(vp8_sixtap_predict8x4_ssse3);
-extern prototype_subpixel_predict(vp8_sixtap_predict4x4_ssse3);
-extern prototype_subpixel_predict(vp8_bilinear_predict16x16_ssse3);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x8_ssse3);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_subpix_sixtap16x16
-#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_ssse3
-
-#undef vp8_subpix_sixtap8x8
-#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_ssse3
-
-#undef vp8_subpix_sixtap8x4
-#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_ssse3
-
-#undef vp8_subpix_sixtap4x4
-#define vp8_subpix_sixtap4x4 vp8_sixtap_predict4x4_ssse3
-
-
-#undef vp8_subpix_bilinear16x16
-#define vp8_subpix_bilinear16x16 vp8_bilinear_predict16x16_ssse3
-
-#undef vp8_subpix_bilinear8x8
-#define vp8_subpix_bilinear8x8 vp8_bilinear_predict8x8_ssse3
-
-#endif
-#endif
-
-
-
-#endif
diff --git a/vp8/common/x86/vp8_asm_stubs.c b/vp8/common/x86/vp8_asm_stubs.c
deleted file mode 100644
index 7904006..0000000
--- a/vp8/common/x86/vp8_asm_stubs.c
+++ /dev/null
@@ -1,553 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_ports/mem.h"
-#include "vp8/common/subpixel.h"
-
-extern const short vp8_six_tap_mmx[8][6*8];
-extern const short vp8_bilinear_filters_mmx[8][2*8];
-
-extern void vp8_filter_block1d_h6_mmx
-(
- unsigned char *src_ptr,
- unsigned short *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-);
-extern void vp8_filter_block1dc_v6_mmx
-(
- unsigned short *src_ptr,
- unsigned char *output_ptr,
- int output_pitch,
- unsigned int pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-);
-extern void vp8_filter_block1d8_h6_sse2
-(
- unsigned char *src_ptr,
- unsigned short *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-);
-extern void vp8_filter_block1d16_h6_sse2
-(
- unsigned char *src_ptr,
- unsigned short *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-);
-extern void vp8_filter_block1d8_v6_sse2
-(
- unsigned short *src_ptr,
- unsigned char *output_ptr,
- int dst_ptich,
- unsigned int pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-);
-extern void vp8_filter_block1d16_v6_sse2
-(
- unsigned short *src_ptr,
- unsigned char *output_ptr,
- int dst_ptich,
- unsigned int pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-);
-extern void vp8_unpack_block1d16_h6_sse2
-(
- unsigned char *src_ptr,
- unsigned short *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int output_height,
- unsigned int output_width
-);
-extern void vp8_filter_block1d8_h6_only_sse2
-(
- unsigned char *src_ptr,
- unsigned int src_pixels_per_line,
- unsigned char *output_ptr,
- int dst_ptich,
- unsigned int output_height,
- const short *vp8_filter
-);
-extern void vp8_filter_block1d16_h6_only_sse2
-(
- unsigned char *src_ptr,
- unsigned int src_pixels_per_line,
- unsigned char *output_ptr,
- int dst_ptich,
- unsigned int output_height,
- const short *vp8_filter
-);
-extern void vp8_filter_block1d8_v6_only_sse2
-(
- unsigned char *src_ptr,
- unsigned int src_pixels_per_line,
- unsigned char *output_ptr,
- int dst_ptich,
- unsigned int output_height,
- const short *vp8_filter
-);
-extern prototype_subpixel_predict(vp8_bilinear_predict8x8_mmx);
-
-
-#if HAVE_MMX
-void vp8_sixtap_predict4x4_mmx
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 16*16); /* Temp data bufffer used in filtering */
- const short *HFilter, *VFilter;
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 9, 8, HFilter);
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1dc_v6_mmx(FData2 + 8, dst_ptr, dst_pitch, 8, 4 , 4, 4, VFilter);
-
-}
-
-
-void vp8_sixtap_predict16x16_mmx
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
-
- DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 24*24); /* Temp data bufffer used in filtering */
-
- const short *HFilter, *VFilter;
-
-
- HFilter = vp8_six_tap_mmx[xoffset];
-
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 21, 32, HFilter);
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 4, FData2 + 4, src_pixels_per_line, 1, 21, 32, HFilter);
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 8, FData2 + 8, src_pixels_per_line, 1, 21, 32, HFilter);
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 12, FData2 + 12, src_pixels_per_line, 1, 21, 32, HFilter);
-
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1dc_v6_mmx(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, 16, VFilter);
- vp8_filter_block1dc_v6_mmx(FData2 + 36, dst_ptr + 4, dst_pitch, 32, 16 , 16, 16, VFilter);
- vp8_filter_block1dc_v6_mmx(FData2 + 40, dst_ptr + 8, dst_pitch, 32, 16 , 16, 16, VFilter);
- vp8_filter_block1dc_v6_mmx(FData2 + 44, dst_ptr + 12, dst_pitch, 32, 16 , 16, 16, VFilter);
-
-}
-
-
-void vp8_sixtap_predict8x8_mmx
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
-
- DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); /* Temp data bufffer used in filtering */
-
- const short *HFilter, *VFilter;
-
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 13, 16, HFilter);
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 4, FData2 + 4, src_pixels_per_line, 1, 13, 16, HFilter);
-
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1dc_v6_mmx(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 8, 8, VFilter);
- vp8_filter_block1dc_v6_mmx(FData2 + 20, dst_ptr + 4, dst_pitch, 16, 8 , 8, 8, VFilter);
-
-}
-
-
-void vp8_sixtap_predict8x4_mmx
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
-
- DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); /* Temp data bufffer used in filtering */
-
- const short *HFilter, *VFilter;
-
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 9, 16, HFilter);
- vp8_filter_block1d_h6_mmx(src_ptr - (2 * src_pixels_per_line) + 4, FData2 + 4, src_pixels_per_line, 1, 9, 16, HFilter);
-
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1dc_v6_mmx(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 4, 8, VFilter);
- vp8_filter_block1dc_v6_mmx(FData2 + 20, dst_ptr + 4, dst_pitch, 16, 8 , 4, 8, VFilter);
-
-}
-
-
-
-void vp8_bilinear_predict16x16_mmx
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- vp8_bilinear_predict8x8_mmx(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pitch);
- vp8_bilinear_predict8x8_mmx(src_ptr + 8, src_pixels_per_line, xoffset, yoffset, dst_ptr + 8, dst_pitch);
- vp8_bilinear_predict8x8_mmx(src_ptr + 8 * src_pixels_per_line, src_pixels_per_line, xoffset, yoffset, dst_ptr + dst_pitch * 8, dst_pitch);
- vp8_bilinear_predict8x8_mmx(src_ptr + 8 * src_pixels_per_line + 8, src_pixels_per_line, xoffset, yoffset, dst_ptr + dst_pitch * 8 + 8, dst_pitch);
-}
-#endif
-
-
-#if HAVE_SSE2
-void vp8_sixtap_predict16x16_sse2
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-
-)
-{
- DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 24*24); /* Temp data bufffer used in filtering */
-
- const short *HFilter, *VFilter;
-
- if (xoffset)
- {
- if (yoffset)
- {
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d16_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 21, 32, HFilter);
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1d16_v6_sse2(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, dst_pitch, VFilter);
- }
- else
- {
- /* First-pass only */
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d16_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 16, HFilter);
- }
- }
- else
- {
- /* Second-pass only */
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_unpack_block1d16_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 21, 32);
- vp8_filter_block1d16_v6_sse2(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, dst_pitch, VFilter);
- }
-}
-
-
-void vp8_sixtap_predict8x8_sse2
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); /* Temp data bufffer used in filtering */
- const short *HFilter, *VFilter;
-
- if (xoffset)
- {
- if (yoffset)
- {
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 13, 16, HFilter);
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 8, dst_pitch, VFilter);
- }
- else
- {
- /* First-pass only */
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d8_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 8, HFilter);
- }
- }
- else
- {
- /* Second-pass only */
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1d8_v6_only_sse2(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 8, VFilter);
- }
-}
-
-
-void vp8_sixtap_predict8x4_sse2
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); /* Temp data bufffer used in filtering */
- const short *HFilter, *VFilter;
-
- if (xoffset)
- {
- if (yoffset)
- {
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 9, 16, HFilter);
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 4, dst_pitch, VFilter);
- }
- else
- {
- /* First-pass only */
- HFilter = vp8_six_tap_mmx[xoffset];
- vp8_filter_block1d8_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 4, HFilter);
- }
- }
- else
- {
- /* Second-pass only */
- VFilter = vp8_six_tap_mmx[yoffset];
- vp8_filter_block1d8_v6_only_sse2(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 4, VFilter);
- }
-}
-
-#endif
-
-#if HAVE_SSSE3
-
-extern void vp8_filter_block1d8_h6_ssse3
-(
- unsigned char *src_ptr,
- unsigned int src_pixels_per_line,
- unsigned char *output_ptr,
- unsigned int output_pitch,
- unsigned int output_height,
- unsigned int vp8_filter_index
-);
-
-extern void vp8_filter_block1d16_h6_ssse3
-(
- unsigned char *src_ptr,
- unsigned int src_pixels_per_line,
- unsigned char *output_ptr,
- unsigned int output_pitch,
- unsigned int output_height,
- unsigned int vp8_filter_index
-);
-
-extern void vp8_filter_block1d16_v6_ssse3
-(
- unsigned char *src_ptr,
- unsigned int src_pitch,
- unsigned char *output_ptr,
- unsigned int out_pitch,
- unsigned int output_height,
- unsigned int vp8_filter_index
-);
-
-extern void vp8_filter_block1d8_v6_ssse3
-(
- unsigned char *src_ptr,
- unsigned int src_pitch,
- unsigned char *output_ptr,
- unsigned int out_pitch,
- unsigned int output_height,
- unsigned int vp8_filter_index
-);
-
-extern void vp8_filter_block1d4_h6_ssse3
-(
- unsigned char *src_ptr,
- unsigned int src_pixels_per_line,
- unsigned char *output_ptr,
- unsigned int output_pitch,
- unsigned int output_height,
- unsigned int vp8_filter_index
-);
-
-extern void vp8_filter_block1d4_v6_ssse3
-(
- unsigned char *src_ptr,
- unsigned int src_pitch,
- unsigned char *output_ptr,
- unsigned int out_pitch,
- unsigned int output_height,
- unsigned int vp8_filter_index
-);
-
-void vp8_sixtap_predict16x16_ssse3
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-
-)
-{
- DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 24*24);
-
- if (xoffset)
- {
- if (yoffset)
- {
- vp8_filter_block1d16_h6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, FData2, 16, 21, xoffset);
- vp8_filter_block1d16_v6_ssse3(FData2 , 16, dst_ptr, dst_pitch, 16, yoffset);
- }
- else
- {
- /* First-pass only */
- vp8_filter_block1d16_h6_ssse3(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 16, xoffset);
- }
- }
- else
- {
- /* Second-pass only */
- vp8_filter_block1d16_v6_ssse3(src_ptr - (2 * src_pixels_per_line) , src_pixels_per_line, dst_ptr, dst_pitch, 16, yoffset);
- }
-}
-
-void vp8_sixtap_predict8x8_ssse3
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 256);
-
- if (xoffset)
- {
- if (yoffset)
- {
- vp8_filter_block1d8_h6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, FData2, 8, 13, xoffset);
- vp8_filter_block1d8_v6_ssse3(FData2, 8, dst_ptr, dst_pitch, 8, yoffset);
- }
- else
- {
- vp8_filter_block1d8_h6_ssse3(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 8, xoffset);
- }
- }
- else
- {
- /* Second-pass only */
- vp8_filter_block1d8_v6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 8, yoffset);
- }
-}
-
-
-void vp8_sixtap_predict8x4_ssse3
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 256);
-
- if (xoffset)
- {
- if (yoffset)
- {
- vp8_filter_block1d8_h6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, FData2, 8, 9, xoffset);
- vp8_filter_block1d8_v6_ssse3(FData2, 8, dst_ptr, dst_pitch, 4, yoffset);
- }
- else
- {
- /* First-pass only */
- vp8_filter_block1d8_h6_ssse3(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 4, xoffset);
- }
- }
- else
- {
- /* Second-pass only */
- vp8_filter_block1d8_v6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 4, yoffset);
- }
-}
-
-void vp8_sixtap_predict4x4_ssse3
-(
- unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- unsigned char *dst_ptr,
- int dst_pitch
-)
-{
- DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 4*9);
-
- if (xoffset)
- {
- if (yoffset)
- {
- vp8_filter_block1d4_h6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, FData2, 4, 9, xoffset);
- vp8_filter_block1d4_v6_ssse3(FData2, 4, dst_ptr, dst_pitch, 4, yoffset);
- }
- else
- {
- vp8_filter_block1d4_h6_ssse3(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 4, xoffset);
- }
- }
- else
- {
- vp8_filter_block1d4_v6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 4, yoffset);
- }
-
-}
-
-#endif
diff --git a/vp8/common/x86/x86_systemdependent.c b/vp8/common/x86/x86_systemdependent.c
deleted file mode 100644
index e89c07a..0000000
--- a/vp8/common/x86/x86_systemdependent.c
+++ /dev/null
@@ -1,133 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_ports/x86.h"
-#include "vp8/common/g_common.h"
-#include "vp8/common/subpixel.h"
-#include "vp8/common/loopfilter.h"
-#include "vp8/common/recon.h"
-#include "vp8/common/idct.h"
-#include "vp8/common/pragmas.h"
-#include "vp8/common/onyxc_int.h"
-
-void vp8_arch_x86_common_init(VP8_COMMON *ctx)
-{
-#if CONFIG_RUNTIME_CPU_DETECT
- VP8_COMMON_RTCD *rtcd = &ctx->rtcd;
- int flags = x86_simd_caps();
- int mmx_enabled = flags & HAS_MMX;
- int xmm_enabled = flags & HAS_SSE;
- int wmt_enabled = flags & HAS_SSE2;
- int SSSE3Enabled = flags & HAS_SSSE3;
-
- /* Note:
- *
- * This platform can be built without runtime CPU detection as well. If
- * you modify any of the function mappings present in this file, be sure
- * to also update them in static mapings (<arch>/filename_<arch>.h)
- */
-
- /* Override default functions with fastest ones for this CPU. */
-#if HAVE_MMX
-
- if (mmx_enabled)
- {
- rtcd->idct.idct1 = vp8_short_idct4x4llm_1_mmx;
- rtcd->idct.idct16 = vp8_short_idct4x4llm_mmx;
- rtcd->idct.idct1_scalar_add = vp8_dc_only_idct_add_mmx;
- rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_mmx;
- rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_mmx;
-
-
-
- rtcd->recon.recon = vp8_recon_b_mmx;
- rtcd->recon.copy8x8 = vp8_copy_mem8x8_mmx;
- rtcd->recon.copy8x4 = vp8_copy_mem8x4_mmx;
- rtcd->recon.copy16x16 = vp8_copy_mem16x16_mmx;
-
- rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_mmx;
- rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_mmx;
- rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_mmx;
- rtcd->subpix.sixtap4x4 = vp8_sixtap_predict4x4_mmx;
- rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_mmx;
- rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_mmx;
- rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_mmx;
- rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_mmx;
-
- rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_mmx;
- rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_mmx;
- rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_mmx;
- rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_mmx;
- rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_mmx;
- rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_mmx;
- rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_mmx;
- rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_mmx;
-
-#if CONFIG_POSTPROC
- rtcd->postproc.down = vp8_mbpost_proc_down_mmx;
- /*rtcd->postproc.across = vp8_mbpost_proc_across_ip_c;*/
- rtcd->postproc.downacross = vp8_post_proc_down_and_across_mmx;
- rtcd->postproc.addnoise = vp8_plane_add_noise_mmx;
-#endif
- }
-
-#endif
-#if HAVE_SSE2
-
- if (wmt_enabled)
- {
- rtcd->recon.recon2 = vp8_recon2b_sse2;
- rtcd->recon.recon4 = vp8_recon4b_sse2;
- rtcd->recon.copy16x16 = vp8_copy_mem16x16_sse2;
-
- rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_sse2;
-
- rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_sse2;
- rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_sse2;
- rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_sse2;
- rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_sse2;
- rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_sse2;
-
- rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_sse2;
- rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_sse2;
- rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_sse2;
- rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_sse2;
- rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_sse2;
- rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_sse2;
- rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_sse2;
- rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_sse2;
-
-#if CONFIG_POSTPROC
- rtcd->postproc.down = vp8_mbpost_proc_down_xmm;
- rtcd->postproc.across = vp8_mbpost_proc_across_ip_xmm;
- rtcd->postproc.downacross = vp8_post_proc_down_and_across_xmm;
- rtcd->postproc.addnoise = vp8_plane_add_noise_wmt;
-#endif
- }
-
-#endif
-
-#if HAVE_SSSE3
-
- if (SSSE3Enabled)
- {
- rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_ssse3;
- rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_ssse3;
- rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_ssse3;
- rtcd->subpix.sixtap4x4 = vp8_sixtap_predict4x4_ssse3;
- rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_ssse3;
- rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_ssse3;
- }
-#endif
-
-#endif
-}
diff --git a/vp8/decoder/arm/arm_dsystemdependent.c b/vp8/decoder/arm/arm_dsystemdependent.c
deleted file mode 100644
index 51e901d..0000000
--- a/vp8/decoder/arm/arm_dsystemdependent.c
+++ /dev/null
@@ -1,53 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_ports/arm.h"
-#include "vp8/common/blockd.h"
-#include "vp8/common/pragmas.h"
-#include "vp8/common/postproc.h"
-#include "vp8/decoder/dequantize.h"
-#include "vp8/decoder/onyxd_int.h"
-
-void vp8_arch_arm_decode_init(VP8D_COMP *pbi)
-{
-#if CONFIG_RUNTIME_CPU_DETECT
- int flags = pbi->common.rtcd.flags;
- int has_edsp = flags & HAS_EDSP;
- int has_media = flags & HAS_MEDIA;
- int has_neon = flags & HAS_NEON;
-
-#if HAVE_ARMV6
- if (has_media)
- {
- pbi->dequant.block = vp8_dequantize_b_v6;
- pbi->dequant.idct_add = vp8_dequant_idct_add_v6;
- pbi->dequant.dc_idct_add = vp8_dequant_dc_idct_add_v6;
- pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_v6;
- pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_v6;
- pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_v6;
- }
-#endif
-
-#if HAVE_ARMV7
- if (has_neon)
- {
- pbi->dequant.block = vp8_dequantize_b_neon;
- pbi->dequant.idct_add = vp8_dequant_idct_add_neon;
- /*This is not used: NEON always dequants two blocks at once.
- pbi->dequant.dc_idct_add = vp8_dequant_dc_idct_add_neon;*/
- pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_neon;
- pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_neon;
- pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_neon;
- }
-#endif
-#endif
-}
diff --git a/vp8/decoder/arm/armv5/dequantize_v5.asm b/vp8/decoder/arm/armv5/dequantize_v5.asm
deleted file mode 100644
index de3648a..0000000
--- a/vp8/decoder/arm/armv5/dequantize_v5.asm
+++ /dev/null
@@ -1,52 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_dequantize_b_armv5|
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
-q RN r0
-dqc RN r1
-cnt RN r2
-
-;void dequantize_b_armv5(short *Q, short *DQC)
-|vp8_dequantize_b_armv5| PROC
- stmdb sp!, {r4, lr}
- ldr r3, [q]
- ldr r4, [dqc], #8
-
- mov cnt, #4
-dequant_loop
- smulbb lr, r3, r4
- smultt r12, r3, r4
-
- ldr r3, [q, #4]
- ldr r4, [dqc, #-4]
-
- strh lr, [q], #2
- strh r12, [q], #2
-
- smulbb lr, r3, r4
- smultt r12, r3, r4
-
- subs cnt, cnt, #1
- ldrne r3, [q, #4]
- ldrne r4, [dqc], #8
-
- strh lr, [q], #2
- strh r12, [q], #2
-
- bne dequant_loop
-
- ldmia sp!, {r4, pc}
- ENDP ;|vp8_dequantize_b_arm|
-
- END
diff --git a/vp8/decoder/arm/armv6/dequant_dc_idct_v6.asm b/vp8/decoder/arm/armv6/dequant_dc_idct_v6.asm
deleted file mode 100644
index 6bebda2..0000000
--- a/vp8/decoder/arm/armv6/dequant_dc_idct_v6.asm
+++ /dev/null
@@ -1,218 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license and patent
-; grant that can be found in the LICENSE file in the root of the source
-; tree. All contributing project authors may be found in the AUTHORS
-; file in the root of the source tree.
-;
-
-
- EXPORT |vp8_dequant_dc_idct_add_v6|
-
- AREA |.text|, CODE, READONLY
-
-;void vp8_dequant_dc_idct_v6(short *input, short *dq, unsigned char *pred,
-; unsigned char *dest, int pitch, int stride, int Dc)
-; r0 = input
-; r1 = dq
-; r2 = pred
-; r3 = dest
-; sp + 36 = pitch ; +4 = 40
-; sp + 40 = stride ; +4 = 44
-; sp + 44 = Dc ; +4 = 48
-
-
-|vp8_dequant_dc_idct_add_v6| PROC
- stmdb sp!, {r4-r11, lr}
-
- ldr r6, [sp, #44]
-
- ldr r4, [r0] ;input
- ldr r5, [r1], #4 ;dq
-
- sub sp, sp, #4
- str r3, [sp]
-
- smultt r7, r4, r5
-
- ldr r4, [r0, #4] ;input
- ldr r5, [r1], #4 ;dq
-
- strh r6, [r0], #2
- strh r7, [r0], #2
-
- smulbb r6, r4, r5
- smultt r7, r4, r5
-
- ldr r4, [r0, #4] ;input
- ldr r5, [r1], #4 ;dq
-
- strh r6, [r0], #2
- strh r7, [r0], #2
-
- mov r12, #3
-
-vp8_dequant_dc_add_loop
- smulbb r6, r4, r5
- smultt r7, r4, r5
-
- ldr r4, [r0, #4] ;input
- ldr r5, [r1], #4 ;dq
-
- strh r6, [r0], #2
- strh r7, [r0], #2
-
- smulbb r6, r4, r5
- smultt r7, r4, r5
-
- subs r12, r12, #1
-
- ldrne r4, [r0, #4]
- ldrne r5, [r1], #4
-
- strh r6, [r0], #2
- strh r7, [r0], #2
-
- bne vp8_dequant_dc_add_loop
-
- sub r0, r0, #32
- mov r1, r0
-
-; short_idct4x4llm_v6_dual
- ldr r3, cospi8sqrt2minus1
- ldr r4, sinpi8sqrt2
- ldr r6, [r0, #8]
- mov r5, #2
-vp8_dequant_dc_idct_loop1_v6
- ldr r12, [r0, #24]
- ldr r14, [r0, #16]
- smulwt r9, r3, r6
- smulwb r7, r3, r6
- smulwt r10, r4, r6
- smulwb r8, r4, r6
- pkhbt r7, r7, r9, lsl #16
- smulwt r11, r3, r12
- pkhbt r8, r8, r10, lsl #16
- uadd16 r6, r6, r7
- smulwt r7, r4, r12
- smulwb r9, r3, r12
- smulwb r10, r4, r12
- subs r5, r5, #1
- pkhbt r9, r9, r11, lsl #16
- ldr r11, [r0], #4
- pkhbt r10, r10, r7, lsl #16
- uadd16 r7, r12, r9
- usub16 r7, r8, r7
- uadd16 r6, r6, r10
- uadd16 r10, r11, r14
- usub16 r8, r11, r14
- uadd16 r9, r10, r6
- usub16 r10, r10, r6
- uadd16 r6, r8, r7
- usub16 r7, r8, r7
- str r6, [r1, #8]
- ldrne r6, [r0, #8]
- str r7, [r1, #16]
- str r10, [r1, #24]
- str r9, [r1], #4
- bne vp8_dequant_dc_idct_loop1_v6
-
- mov r5, #2
- sub r0, r1, #8
-vp8_dequant_dc_idct_loop2_v6
- ldr r6, [r0], #4
- ldr r7, [r0], #4
- ldr r8, [r0], #4
- ldr r9, [r0], #4
- smulwt r1, r3, r6
- smulwt r12, r4, r6
- smulwt lr, r3, r8
- smulwt r10, r4, r8
- pkhbt r11, r8, r6, lsl #16
- pkhbt r1, lr, r1, lsl #16
- pkhbt r12, r10, r12, lsl #16
- pkhtb r6, r6, r8, asr #16
- uadd16 r6, r1, r6
- pkhbt lr, r9, r7, lsl #16
- uadd16 r10, r11, lr
- usub16 lr, r11, lr
- pkhtb r8, r7, r9, asr #16
- subs r5, r5, #1
- smulwt r1, r3, r8
- smulwb r7, r3, r8
- smulwt r11, r4, r8
- smulwb r9, r4, r8
- pkhbt r1, r7, r1, lsl #16
- uadd16 r8, r1, r8
- pkhbt r11, r9, r11, lsl #16
- usub16 r1, r12, r8
- uadd16 r8, r11, r6
- ldr r9, c0x00040004
- ldr r12, [sp, #40]
- uadd16 r6, r10, r8
- usub16 r7, r10, r8
- uadd16 r7, r7, r9
- uadd16 r6, r6, r9
- uadd16 r10, r14, r1
- usub16 r1, r14, r1
- uadd16 r10, r10, r9
- uadd16 r1, r1, r9
- ldr r11, [r2], r12
- mov r8, r7, asr #3
- pkhtb r9, r8, r10, asr #19
- mov r8, r1, asr #3
- pkhtb r8, r8, r6, asr #19
- uxtb16 lr, r11, ror #8
- qadd16 r9, r9, lr
- uxtb16 lr, r11
- qadd16 r8, r8, lr
- usat16 r9, #8, r9
- usat16 r8, #8, r8
- orr r9, r8, r9, lsl #8
- ldr r11, [r2], r12
- ldr lr, [sp]
- ldr r12, [sp, #44]
- mov r7, r7, lsl #16
- mov r1, r1, lsl #16
- mov r10, r10, lsl #16
- mov r6, r6, lsl #16
- mov r7, r7, asr #3
- pkhtb r7, r7, r10, asr #19
- mov r1, r1, asr #3
- pkhtb r1, r1, r6, asr #19
- uxtb16 r8, r11, ror #8
- qadd16 r7, r7, r8
- uxtb16 r8, r11
- qadd16 r1, r1, r8
- usat16 r7, #8, r7
- usat16 r1, #8, r1
- orr r1, r1, r7, lsl #8
- str r9, [lr], r12
- str r1, [lr], r12
- str lr, [sp]
- bne vp8_dequant_dc_idct_loop2_v6
-
-; vpx_memset
- sub r0, r0, #32
- add sp, sp, #4
-
- mov r12, #0
- str r12, [r0]
- str r12, [r0, #4]
- str r12, [r0, #8]
- str r12, [r0, #12]
- str r12, [r0, #16]
- str r12, [r0, #20]
- str r12, [r0, #24]
- str r12, [r0, #28]
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_dequant_dc_idct_add_v6|
-
-; Constant Pool
-cospi8sqrt2minus1 DCD 0x00004E7B
-sinpi8sqrt2 DCD 0x00008A8C
-c0x00040004 DCD 0x00040004
-
- END
diff --git a/vp8/decoder/arm/armv6/dequant_idct_v6.asm b/vp8/decoder/arm/armv6/dequant_idct_v6.asm
deleted file mode 100644
index 47b671c..0000000
--- a/vp8/decoder/arm/armv6/dequant_idct_v6.asm
+++ /dev/null
@@ -1,196 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license and patent
-; grant that can be found in the LICENSE file in the root of the source
-; tree. All contributing project authors may be found in the AUTHORS
-; file in the root of the source tree.
-;
-
- EXPORT |vp8_dequant_idct_add_v6|
-
- AREA |.text|, CODE, READONLY
-;void vp8_dequant_idct_v6(short *input, short *dq, unsigned char *pred,
-; unsigned char *dest, int pitch, int stride)
-; r0 = input
-; r1 = dq
-; r2 = pred
-; r3 = dest
-; sp + 36 = pitch ; +4 = 40
-; sp + 40 = stride ; +4 = 44
-
-
-|vp8_dequant_idct_add_v6| PROC
- stmdb sp!, {r4-r11, lr}
-
- ldr r4, [r0] ;input
- ldr r5, [r1], #4 ;dq
-
- sub sp, sp, #4
- str r3, [sp]
-
- mov r12, #4
-
-vp8_dequant_add_loop
- smulbb r6, r4, r5
- smultt r7, r4, r5
-
- ldr r4, [r0, #4] ;input
- ldr r5, [r1], #4 ;dq
-
- strh r6, [r0], #2
- strh r7, [r0], #2
-
- smulbb r6, r4, r5
- smultt r7, r4, r5
-
- subs r12, r12, #1
-
- ldrne r4, [r0, #4]
- ldrne r5, [r1], #4
-
- strh r6, [r0], #2
- strh r7, [r0], #2
-
- bne vp8_dequant_add_loop
-
- sub r0, r0, #32
- mov r1, r0
-
-; short_idct4x4llm_v6_dual
- ldr r3, cospi8sqrt2minus1
- ldr r4, sinpi8sqrt2
- ldr r6, [r0, #8]
- mov r5, #2
-vp8_dequant_idct_loop1_v6
- ldr r12, [r0, #24]
- ldr r14, [r0, #16]
- smulwt r9, r3, r6
- smulwb r7, r3, r6
- smulwt r10, r4, r6
- smulwb r8, r4, r6
- pkhbt r7, r7, r9, lsl #16
- smulwt r11, r3, r12
- pkhbt r8, r8, r10, lsl #16
- uadd16 r6, r6, r7
- smulwt r7, r4, r12
- smulwb r9, r3, r12
- smulwb r10, r4, r12
- subs r5, r5, #1
- pkhbt r9, r9, r11, lsl #16
- ldr r11, [r0], #4
- pkhbt r10, r10, r7, lsl #16
- uadd16 r7, r12, r9
- usub16 r7, r8, r7
- uadd16 r6, r6, r10
- uadd16 r10, r11, r14
- usub16 r8, r11, r14
- uadd16 r9, r10, r6
- usub16 r10, r10, r6
- uadd16 r6, r8, r7
- usub16 r7, r8, r7
- str r6, [r1, #8]
- ldrne r6, [r0, #8]
- str r7, [r1, #16]
- str r10, [r1, #24]
- str r9, [r1], #4
- bne vp8_dequant_idct_loop1_v6
-
- mov r5, #2
- sub r0, r1, #8
-vp8_dequant_idct_loop2_v6
- ldr r6, [r0], #4
- ldr r7, [r0], #4
- ldr r8, [r0], #4
- ldr r9, [r0], #4
- smulwt r1, r3, r6
- smulwt r12, r4, r6
- smulwt lr, r3, r8
- smulwt r10, r4, r8
- pkhbt r11, r8, r6, lsl #16
- pkhbt r1, lr, r1, lsl #16
- pkhbt r12, r10, r12, lsl #16
- pkhtb r6, r6, r8, asr #16
- uadd16 r6, r1, r6
- pkhbt lr, r9, r7, lsl #16
- uadd16 r10, r11, lr
- usub16 lr, r11, lr
- pkhtb r8, r7, r9, asr #16
- subs r5, r5, #1
- smulwt r1, r3, r8
- smulwb r7, r3, r8
- smulwt r11, r4, r8
- smulwb r9, r4, r8
- pkhbt r1, r7, r1, lsl #16
- uadd16 r8, r1, r8
- pkhbt r11, r9, r11, lsl #16
- usub16 r1, r12, r8
- uadd16 r8, r11, r6
- ldr r9, c0x00040004
- ldr r12, [sp, #40]
- uadd16 r6, r10, r8
- usub16 r7, r10, r8
- uadd16 r7, r7, r9
- uadd16 r6, r6, r9
- uadd16 r10, r14, r1
- usub16 r1, r14, r1
- uadd16 r10, r10, r9
- uadd16 r1, r1, r9
- ldr r11, [r2], r12
- mov r8, r7, asr #3
- pkhtb r9, r8, r10, asr #19
- mov r8, r1, asr #3
- pkhtb r8, r8, r6, asr #19
- uxtb16 lr, r11, ror #8
- qadd16 r9, r9, lr
- uxtb16 lr, r11
- qadd16 r8, r8, lr
- usat16 r9, #8, r9
- usat16 r8, #8, r8
- orr r9, r8, r9, lsl #8
- ldr r11, [r2], r12
- ldr lr, [sp]
- ldr r12, [sp, #44]
- mov r7, r7, lsl #16
- mov r1, r1, lsl #16
- mov r10, r10, lsl #16
- mov r6, r6, lsl #16
- mov r7, r7, asr #3
- pkhtb r7, r7, r10, asr #19
- mov r1, r1, asr #3
- pkhtb r1, r1, r6, asr #19
- uxtb16 r8, r11, ror #8
- qadd16 r7, r7, r8
- uxtb16 r8, r11
- qadd16 r1, r1, r8
- usat16 r7, #8, r7
- usat16 r1, #8, r1
- orr r1, r1, r7, lsl #8
- str r9, [lr], r12
- str r1, [lr], r12
- str lr, [sp]
- bne vp8_dequant_idct_loop2_v6
-
-; vpx_memset
- sub r0, r0, #32
- add sp, sp, #4
-
- mov r12, #0
- str r12, [r0]
- str r12, [r0, #4]
- str r12, [r0, #8]
- str r12, [r0, #12]
- str r12, [r0, #16]
- str r12, [r0, #20]
- str r12, [r0, #24]
- str r12, [r0, #28]
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_dequant_idct_add_v6|
-
-; Constant Pool
-cospi8sqrt2minus1 DCD 0x00004E7B
-sinpi8sqrt2 DCD 0x00008A8C
-c0x00040004 DCD 0x00040004
-
- END
diff --git a/vp8/decoder/arm/armv6/idct_blk_v6.c b/vp8/decoder/arm/armv6/idct_blk_v6.c
deleted file mode 100644
index 57c3446..0000000
--- a/vp8/decoder/arm/armv6/idct_blk_v6.c
+++ /dev/null
@@ -1,151 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_ports/config.h"
-#include "vp8/common/idct.h"
-#include "vp8/decoder/dequantize.h"
-
-void vp8_dequant_dc_idct_add_y_block_v6
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs, short *dc)
-{
- int i;
-
- for (i = 0; i < 4; i++)
- {
- if (eobs[0] > 1)
- vp8_dequant_dc_idct_add_v6 (q, dq, pre, dst, 16, stride, dc[0]);
- else
- vp8_dc_only_idct_add_v6 (dc[0], pre, dst, 16, stride);
-
- if (eobs[1] > 1)
- vp8_dequant_dc_idct_add_v6 (q+16, dq, pre+4, dst+4, 16, stride, dc[1]);
- else
- vp8_dc_only_idct_add_v6 (dc[1], pre+4, dst+4, 16, stride);
-
- if (eobs[2] > 1)
- vp8_dequant_dc_idct_add_v6 (q+32, dq, pre+8, dst+8, 16, stride, dc[2]);
- else
- vp8_dc_only_idct_add_v6 (dc[2], pre+8, dst+8, 16, stride);
-
- if (eobs[3] > 1)
- vp8_dequant_dc_idct_add_v6 (q+48, dq, pre+12, dst+12, 16, stride, dc[3]);
- else
- vp8_dc_only_idct_add_v6 (dc[3], pre+12, dst+12, 16, stride);
-
- q += 64;
- dc += 4;
- pre += 64;
- dst += 4*stride;
- eobs += 4;
- }
-}
-
-void vp8_dequant_idct_add_y_block_v6
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs)
-{
- int i;
-
- for (i = 0; i < 4; i++)
- {
- if (eobs[0] > 1)
- vp8_dequant_idct_add_v6 (q, dq, pre, dst, 16, stride);
- else
- {
- vp8_dc_only_idct_add_v6 (q[0]*dq[0], pre, dst, 16, stride);
- ((int *)q)[0] = 0;
- }
-
- if (eobs[1] > 1)
- vp8_dequant_idct_add_v6 (q+16, dq, pre+4, dst+4, 16, stride);
- else
- {
- vp8_dc_only_idct_add_v6 (q[16]*dq[0], pre+4, dst+4, 16, stride);
- ((int *)(q+16))[0] = 0;
- }
-
- if (eobs[2] > 1)
- vp8_dequant_idct_add_v6 (q+32, dq, pre+8, dst+8, 16, stride);
- else
- {
- vp8_dc_only_idct_add_v6 (q[32]*dq[0], pre+8, dst+8, 16, stride);
- ((int *)(q+32))[0] = 0;
- }
-
- if (eobs[3] > 1)
- vp8_dequant_idct_add_v6 (q+48, dq, pre+12, dst+12, 16, stride);
- else
- {
- vp8_dc_only_idct_add_v6 (q[48]*dq[0], pre+12, dst+12, 16, stride);
- ((int *)(q+48))[0] = 0;
- }
-
- q += 64;
- pre += 64;
- dst += 4*stride;
- eobs += 4;
- }
-}
-
-void vp8_dequant_idct_add_uv_block_v6
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
-{
- int i;
-
- for (i = 0; i < 2; i++)
- {
- if (eobs[0] > 1)
- vp8_dequant_idct_add_v6 (q, dq, pre, dstu, 8, stride);
- else
- {
- vp8_dc_only_idct_add_v6 (q[0]*dq[0], pre, dstu, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- if (eobs[1] > 1)
- vp8_dequant_idct_add_v6 (q+16, dq, pre+4, dstu+4, 8, stride);
- else
- {
- vp8_dc_only_idct_add_v6 (q[16]*dq[0], pre+4, dstu+4, 8, stride);
- ((int *)(q+16))[0] = 0;
- }
-
- q += 32;
- pre += 32;
- dstu += 4*stride;
- eobs += 2;
- }
-
- for (i = 0; i < 2; i++)
- {
- if (eobs[0] > 1)
- vp8_dequant_idct_add_v6 (q, dq, pre, dstv, 8, stride);
- else
- {
- vp8_dc_only_idct_add_v6 (q[0]*dq[0], pre, dstv, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- if (eobs[1] > 1)
- vp8_dequant_idct_add_v6 (q+16, dq, pre+4, dstv+4, 8, stride);
- else
- {
- vp8_dc_only_idct_add_v6 (q[16]*dq[0], pre+4, dstv+4, 8, stride);
- ((int *)(q+16))[0] = 0;
- }
-
- q += 32;
- pre += 32;
- dstv += 4*stride;
- eobs += 2;
- }
-}
diff --git a/vp8/decoder/arm/dequantize_arm.c b/vp8/decoder/arm/dequantize_arm.c
deleted file mode 100644
index d88adb7..0000000
--- a/vp8/decoder/arm/dequantize_arm.c
+++ /dev/null
@@ -1,48 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/decoder/dequantize.h"
-#include "vp8/common/idct.h"
-#include "vpx_mem/vpx_mem.h"
-
-#if HAVE_ARMV7
-extern void vp8_dequantize_b_loop_neon(short *Q, short *DQC, short *DQ);
-#endif
-
-#if HAVE_ARMV6
-extern void vp8_dequantize_b_loop_v6(short *Q, short *DQC, short *DQ);
-#endif
-
-#if HAVE_ARMV7
-
-void vp8_dequantize_b_neon(BLOCKD *d)
-{
- int i;
- short *DQ = d->dqcoeff;
- short *Q = d->qcoeff;
- short *DQC = d->dequant;
-
- vp8_dequantize_b_loop_neon(Q, DQC, DQ);
-}
-#endif
-
-#if HAVE_ARMV6
-void vp8_dequantize_b_v6(BLOCKD *d)
-{
- int i;
- short *DQ = d->dqcoeff;
- short *Q = d->qcoeff;
- short *DQC = d->dequant;
-
- vp8_dequantize_b_loop_v6(Q, DQC, DQ);
-}
-#endif
diff --git a/vp8/decoder/arm/dequantize_arm.h b/vp8/decoder/arm/dequantize_arm.h
deleted file mode 100644
index b7d800d..0000000
--- a/vp8/decoder/arm/dequantize_arm.h
+++ /dev/null
@@ -1,73 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef DEQUANTIZE_ARM_H
-#define DEQUANTIZE_ARM_H
-
-#if HAVE_ARMV6
-extern prototype_dequant_block(vp8_dequantize_b_v6);
-extern prototype_dequant_idct_add(vp8_dequant_idct_add_v6);
-extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add_v6);
-extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_v6);
-extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_v6);
-extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_v6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_dequant_block
-#define vp8_dequant_block vp8_dequantize_b_v6
-
-#undef vp8_dequant_idct_add
-#define vp8_dequant_idct_add vp8_dequant_idct_add_v6
-
-#undef vp8_dequant_dc_idct_add
-#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_v6
-
-#undef vp8_dequant_dc_idct_add_y_block
-#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_v6
-
-#undef vp8_dequant_idct_add_y_block
-#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_v6
-
-#undef vp8_dequant_idct_add_uv_block
-#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_v6
-#endif
-#endif
-
-#if HAVE_ARMV7
-extern prototype_dequant_block(vp8_dequantize_b_neon);
-extern prototype_dequant_idct_add(vp8_dequant_idct_add_neon);
-extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add_neon);
-extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_neon);
-extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_neon);
-extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_neon);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_dequant_block
-#define vp8_dequant_block vp8_dequantize_b_neon
-
-#undef vp8_dequant_idct_add
-#define vp8_dequant_idct_add vp8_dequant_idct_add_neon
-
-#undef vp8_dequant_dc_idct_add
-#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_neon
-
-#undef vp8_dequant_dc_idct_add_y_block
-#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_neon
-
-#undef vp8_dequant_idct_add_y_block
-#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_neon
-
-#undef vp8_dequant_idct_add_uv_block
-#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_neon
-#endif
-#endif
-
-#endif
diff --git a/vp8/decoder/arm/neon/dequant_idct_neon.asm b/vp8/decoder/arm/neon/dequant_idct_neon.asm
deleted file mode 100644
index 5ee2cc7..0000000
--- a/vp8/decoder/arm/neon/dequant_idct_neon.asm
+++ /dev/null
@@ -1,130 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_dequant_idct_add_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void vp8_dequant_idct_neon(short *input, short *dq, unsigned char *pred,
-; unsigned char *dest, int pitch, int stride)
-; r0 short *input,
-; r1 short *dq,
-; r2 unsigned char *pred
-; r3 unsigned char *dest
-; sp int pitch
-; sp+4 int stride
-
-|vp8_dequant_idct_add_neon| PROC
- vld1.16 {q3, q4}, [r0]
- vld1.16 {q5, q6}, [r1]
- ldr r1, [sp] ; pitch
- vld1.32 {d14[0]}, [r2], r1
- vld1.32 {d14[1]}, [r2], r1
- vld1.32 {d15[0]}, [r2], r1
- vld1.32 {d15[1]}, [r2]
-
- ldr r1, [sp, #4] ; stride
-
- adr r12, _CONSTANTS_
-
- vmul.i16 q1, q3, q5 ;input for short_idct4x4llm_neon
- vmul.i16 q2, q4, q6
-
-;|short_idct4x4llm_neon| PROC
- vld1.16 {d0}, [r12]
- vswp d3, d4 ;q2(vp[4] vp[12])
-
- vqdmulh.s16 q3, q2, d0[2]
- vqdmulh.s16 q4, q2, d0[0]
-
- vqadd.s16 d12, d2, d3 ;a1
- vqsub.s16 d13, d2, d3 ;b1
-
- vshr.s16 q3, q3, #1
- vshr.s16 q4, q4, #1
-
- vqadd.s16 q3, q3, q2
- vqadd.s16 q4, q4, q2
-
- vqsub.s16 d10, d6, d9 ;c1
- vqadd.s16 d11, d7, d8 ;d1
-
- vqadd.s16 d2, d12, d11
- vqadd.s16 d3, d13, d10
- vqsub.s16 d4, d13, d10
- vqsub.s16 d5, d12, d11
-
- vtrn.32 d2, d4
- vtrn.32 d3, d5
- vtrn.16 d2, d3
- vtrn.16 d4, d5
-
-; memset(input, 0, 32) -- 32bytes
- vmov.i16 q14, #0
-
- vswp d3, d4
- vqdmulh.s16 q3, q2, d0[2]
- vqdmulh.s16 q4, q2, d0[0]
-
- vqadd.s16 d12, d2, d3 ;a1
- vqsub.s16 d13, d2, d3 ;b1
-
- vmov q15, q14
-
- vshr.s16 q3, q3, #1
- vshr.s16 q4, q4, #1
-
- vqadd.s16 q3, q3, q2
- vqadd.s16 q4, q4, q2
-
- vqsub.s16 d10, d6, d9 ;c1
- vqadd.s16 d11, d7, d8 ;d1
-
- vqadd.s16 d2, d12, d11
- vqadd.s16 d3, d13, d10
- vqsub.s16 d4, d13, d10
- vqsub.s16 d5, d12, d11
-
- vst1.16 {q14, q15}, [r0]
-
- vrshr.s16 d2, d2, #3
- vrshr.s16 d3, d3, #3
- vrshr.s16 d4, d4, #3
- vrshr.s16 d5, d5, #3
-
- vtrn.32 d2, d4
- vtrn.32 d3, d5
- vtrn.16 d2, d3
- vtrn.16 d4, d5
-
- vaddw.u8 q1, q1, d14
- vaddw.u8 q2, q2, d15
-
- vqmovun.s16 d0, q1
- vqmovun.s16 d1, q2
-
- vst1.32 {d0[0]}, [r3], r1
- vst1.32 {d0[1]}, [r3], r1
- vst1.32 {d1[0]}, [r3], r1
- vst1.32 {d1[1]}, [r3]
-
- bx lr
-
- ENDP ; |vp8_dequant_idct_add_neon|
-
-; Constant Pool
-_CONSTANTS_
-cospi8sqrt2minus1 DCD 0x4e7b4e7b
-sinpi8sqrt2 DCD 0x8a8c8a8c
-
- END
diff --git a/vp8/decoder/arm/neon/idct_blk_neon.c b/vp8/decoder/arm/neon/idct_blk_neon.c
deleted file mode 100644
index ee35004..0000000
--- a/vp8/decoder/arm/neon/idct_blk_neon.c
+++ /dev/null
@@ -1,115 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_ports/config.h"
-#include "vp8/common/idct.h"
-#include "vp8/decoder/dequantize.h"
-
-/* place these declarations here because we don't want to maintain them
- * outside of this scope
- */
-void idct_dequant_dc_full_2x_neon
- (short *input, short *dq, unsigned char *pre, unsigned char *dst,
- int stride, short *dc);
-void idct_dequant_dc_0_2x_neon
- (short *dc, unsigned char *pre, unsigned char *dst, int stride);
-void idct_dequant_full_2x_neon
- (short *q, short *dq, unsigned char *pre, unsigned char *dst,
- int pitch, int stride);
-void idct_dequant_0_2x_neon
- (short *q, short dq, unsigned char *pre, int pitch,
- unsigned char *dst, int stride);
-
-void vp8_dequant_dc_idct_add_y_block_neon
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs, short *dc)
-{
- int i;
-
- for (i = 0; i < 4; i++)
- {
- if (((short *)eobs)[0] & 0xfefe)
- idct_dequant_dc_full_2x_neon (q, dq, pre, dst, stride, dc);
- else
- idct_dequant_dc_0_2x_neon(dc, pre, dst, stride);
-
- if (((short *)eobs)[1] & 0xfefe)
- idct_dequant_dc_full_2x_neon (q+32, dq, pre+8, dst+8, stride, dc+2);
- else
- idct_dequant_dc_0_2x_neon(dc+2, pre+8, dst+8, stride);
-
- q += 64;
- dc += 4;
- pre += 64;
- dst += 4*stride;
- eobs += 4;
- }
-}
-
-void vp8_dequant_idct_add_y_block_neon
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs)
-{
- int i;
-
- for (i = 0; i < 4; i++)
- {
- if (((short *)eobs)[0] & 0xfefe)
- idct_dequant_full_2x_neon (q, dq, pre, dst, 16, stride);
- else
- idct_dequant_0_2x_neon (q, dq[0], pre, 16, dst, stride);
-
- if (((short *)eobs)[1] & 0xfefe)
- idct_dequant_full_2x_neon (q+32, dq, pre+8, dst+8, 16, stride);
- else
- idct_dequant_0_2x_neon (q+32, dq[0], pre+8, 16, dst+8, stride);
-
- q += 64;
- pre += 64;
- dst += 4*stride;
- eobs += 4;
- }
-}
-
-void vp8_dequant_idct_add_uv_block_neon
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
-{
- if (((short *)eobs)[0] & 0xfefe)
- idct_dequant_full_2x_neon (q, dq, pre, dstu, 8, stride);
- else
- idct_dequant_0_2x_neon (q, dq[0], pre, 8, dstu, stride);
-
- q += 32;
- pre += 32;
- dstu += 4*stride;
-
- if (((short *)eobs)[1] & 0xfefe)
- idct_dequant_full_2x_neon (q, dq, pre, dstu, 8, stride);
- else
- idct_dequant_0_2x_neon (q, dq[0], pre, 8, dstu, stride);
-
- q += 32;
- pre += 32;
-
- if (((short *)eobs)[2] & 0xfefe)
- idct_dequant_full_2x_neon (q, dq, pre, dstv, 8, stride);
- else
- idct_dequant_0_2x_neon (q, dq[0], pre, 8, dstv, stride);
-
- q += 32;
- pre += 32;
- dstv += 4*stride;
-
- if (((short *)eobs)[3] & 0xfefe)
- idct_dequant_full_2x_neon (q, dq, pre, dstv, 8, stride);
- else
- idct_dequant_0_2x_neon (q, dq[0], pre, 8, dstv, stride);
-}
diff --git a/vp8/decoder/arm/neon/idct_dequant_0_2x_neon.asm b/vp8/decoder/arm/neon/idct_dequant_0_2x_neon.asm
deleted file mode 100644
index 456f8e1..0000000
--- a/vp8/decoder/arm/neon/idct_dequant_0_2x_neon.asm
+++ /dev/null
@@ -1,79 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license and patent
-; grant that can be found in the LICENSE file in the root of the source
-; tree. All contributing project authors may be found in the AUTHORS
-; file in the root of the source tree.
-;
-
-
- EXPORT |idct_dequant_0_2x_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void idct_dequant_0_2x_neon(short *q, short dq, unsigned char *pre,
-; int pitch, unsigned char *dst, int stride);
-; r0 *q
-; r1 dq
-; r2 *pre
-; r3 pitch
-; sp *dst
-; sp+4 stride
-|idct_dequant_0_2x_neon| PROC
- add r12, r2, #4
- vld1.32 {d2[0]}, [r2], r3
- vld1.32 {d2[1]}, [r2], r3
- vld1.32 {d4[0]}, [r2], r3
- vld1.32 {d4[1]}, [r2]
- vld1.32 {d8[0]}, [r12], r3
- vld1.32 {d8[1]}, [r12], r3
- vld1.32 {d10[0]}, [r12], r3
- vld1.32 {d10[1]}, [r12]
-
- ldrh r12, [r0] ; lo q
- ldrh r2, [r0, #32] ; hi q
- mov r3, #0
- strh r3, [r0]
- strh r3, [r0, #32]
-
- sxth r12, r12 ; lo
- mul r0, r12, r1
- add r0, r0, #4
- asr r0, r0, #3
- vdup.16 q0, r0
- sxth r2, r2 ; hi
- mul r0, r2, r1
- add r0, r0, #4
- asr r0, r0, #3
- vdup.16 q3, r0
-
- vaddw.u8 q1, q0, d2 ; lo
- vaddw.u8 q2, q0, d4
- vaddw.u8 q4, q3, d8 ; hi
- vaddw.u8 q5, q3, d10
-
- ldr r2, [sp] ; dst
- ldr r3, [sp, #4] ; stride
-
- vqmovun.s16 d2, q1 ; lo
- vqmovun.s16 d4, q2
- vqmovun.s16 d8, q4 ; hi
- vqmovun.s16 d10, q5
-
- add r0, r2, #4
- vst1.32 {d2[0]}, [r2], r3 ; lo
- vst1.32 {d2[1]}, [r2], r3
- vst1.32 {d4[0]}, [r2], r3
- vst1.32 {d4[1]}, [r2]
- vst1.32 {d8[0]}, [r0], r3 ; hi
- vst1.32 {d8[1]}, [r0], r3
- vst1.32 {d10[0]}, [r0], r3
- vst1.32 {d10[1]}, [r0]
-
- bx lr
-
- ENDP ; |idct_dequant_0_2x_neon|
- END
diff --git a/vp8/decoder/arm/neon/idct_dequant_dc_0_2x_neon.asm b/vp8/decoder/arm/neon/idct_dequant_dc_0_2x_neon.asm
deleted file mode 100644
index 0dc036a..0000000
--- a/vp8/decoder/arm/neon/idct_dequant_dc_0_2x_neon.asm
+++ /dev/null
@@ -1,69 +0,0 @@
-;
-; Copyright (c) 2010 The Webm project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license and patent
-; grant that can be found in the LICENSE file in the root of the source
-; tree. All contributing project authors may be found in the AUTHORS
-; file in the root of the source tree.
-;
-
-
- EXPORT |idct_dequant_dc_0_2x_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void idct_dequant_dc_0_2x_neon(short *dc, unsigned char *pre,
-; unsigned char *dst, int stride);
-; r0 *dc
-; r1 *pre
-; r2 *dst
-; r3 stride
-|idct_dequant_dc_0_2x_neon| PROC
- ldr r0, [r0] ; *dc
- mov r12, #16
-
- vld1.32 {d2[0]}, [r1], r12 ; lo
- vld1.32 {d2[1]}, [r1], r12
- vld1.32 {d4[0]}, [r1], r12
- vld1.32 {d4[1]}, [r1]
- sub r1, r1, #44
- vld1.32 {d8[0]}, [r1], r12 ; hi
- vld1.32 {d8[1]}, [r1], r12
- vld1.32 {d10[0]}, [r1], r12
- vld1.32 {d10[1]}, [r1]
-
- sxth r1, r0 ; lo *dc
- add r1, r1, #4
- asr r1, r1, #3
- vdup.16 q0, r1
- sxth r0, r0, ror #16 ; hi *dc
- add r0, r0, #4
- asr r0, r0, #3
- vdup.16 q3, r0
-
- vaddw.u8 q1, q0, d2 ; lo
- vaddw.u8 q2, q0, d4
- vaddw.u8 q4, q3, d8 ; hi
- vaddw.u8 q5, q3, d10
-
- vqmovun.s16 d2, q1 ; lo
- vqmovun.s16 d4, q2
- vqmovun.s16 d8, q4 ; hi
- vqmovun.s16 d10, q5
-
- add r0, r2, #4
- vst1.32 {d2[0]}, [r2], r3 ; lo
- vst1.32 {d2[1]}, [r2], r3
- vst1.32 {d4[0]}, [r2], r3
- vst1.32 {d4[1]}, [r2]
- vst1.32 {d8[0]}, [r0], r3 ; hi
- vst1.32 {d8[1]}, [r0], r3
- vst1.32 {d10[0]}, [r0], r3
- vst1.32 {d10[1]}, [r0]
-
- bx lr
-
- ENDP ;|idct_dequant_dc_0_2x_neon|
- END
diff --git a/vp8/decoder/arm/neon/idct_dequant_dc_full_2x_neon.asm b/vp8/decoder/arm/neon/idct_dequant_dc_full_2x_neon.asm
deleted file mode 100644
index 42ff35d..0000000
--- a/vp8/decoder/arm/neon/idct_dequant_dc_full_2x_neon.asm
+++ /dev/null
@@ -1,206 +0,0 @@
-;
-; Copyright (c) 2010 The Webm project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |idct_dequant_dc_full_2x_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void idct_dequant_dc_full_2x_neon(short *q, short *dq, unsigned char *pre,
-; unsigned char *dst, int stride, short *dc);
-; r0 *q,
-; r1 *dq,
-; r2 *pre
-; r3 *dst
-; sp stride
-; sp+4 *dc
-|idct_dequant_dc_full_2x_neon| PROC
- vld1.16 {q0, q1}, [r1] ; dq (same l/r)
- vld1.16 {q2, q3}, [r0] ; l q
- mov r1, #16 ; pitch
- add r0, r0, #32
- vld1.16 {q4, q5}, [r0] ; r q
- add r12, r2, #4
- ; interleave the predictors
- vld1.32 {d28[0]}, [r2], r1 ; l pre
- vld1.32 {d28[1]}, [r12], r1 ; r pre
- vld1.32 {d29[0]}, [r2], r1
- vld1.32 {d29[1]}, [r12], r1
- vld1.32 {d30[0]}, [r2], r1
- vld1.32 {d30[1]}, [r12], r1
- vld1.32 {d31[0]}, [r2]
- ldr r1, [sp, #4]
- vld1.32 {d31[1]}, [r12]
-
- adr r2, _CONSTANTS_
-
- ldrh r12, [r1], #2 ; lo *dc
- ldrh r1, [r1] ; hi *dc
-
- ; dequant: q[i] = q[i] * dq[i]
- vmul.i16 q2, q2, q0
- vmul.i16 q3, q3, q1
- vmul.i16 q4, q4, q0
- vmul.i16 q5, q5, q1
-
- ; move dc up to neon and overwrite first element
- vmov.16 d4[0], r12
- vmov.16 d8[0], r1
-
- vld1.16 {d0}, [r2]
-
- ; q2: l0r0 q3: l8r8
- ; q4: l4r4 q5: l12r12
- vswp d5, d8
- vswp d7, d10
-
- ; _CONSTANTS_ * 4,12 >> 16
- ; q6: 4 * sinpi : c1/temp1
- ; q7: 12 * sinpi : d1/temp2
- ; q8: 4 * cospi
- ; q9: 12 * cospi
- vqdmulh.s16 q6, q4, d0[2] ; sinpi8sqrt2
- vqdmulh.s16 q7, q5, d0[2]
- vqdmulh.s16 q8, q4, d0[0] ; cospi8sqrt2minus1
- vqdmulh.s16 q9, q5, d0[0]
-
- vqadd.s16 q10, q2, q3 ; a1 = 0 + 8
- vqsub.s16 q11, q2, q3 ; b1 = 0 - 8
-
- ; vqdmulh only accepts signed values. this was a problem because
- ; our constant had the high bit set, and was treated as a negative value.
- ; vqdmulh also doubles the value before it shifts by 16. we need to
- ; compensate for this. in the case of sinpi8sqrt2, the lowest bit is 0,
- ; so we can shift the constant without losing precision. this avoids
- ; shift again afterward, but also avoids the sign issue. win win!
- ; for cospi8sqrt2minus1 the lowest bit is 1, so we lose precision if we
- ; pre-shift it
- vshr.s16 q8, q8, #1
- vshr.s16 q9, q9, #1
-
- ; q4: 4 + 4 * cospi : d1/temp1
- ; q5: 12 + 12 * cospi : c1/temp2
- vqadd.s16 q4, q4, q8
- vqadd.s16 q5, q5, q9
-
- ; c1 = temp1 - temp2
- ; d1 = temp1 + temp2
- vqsub.s16 q2, q6, q5
- vqadd.s16 q3, q4, q7
-
- ; [0]: a1+d1
- ; [1]: b1+c1
- ; [2]: b1-c1
- ; [3]: a1-d1
- vqadd.s16 q4, q10, q3
- vqadd.s16 q5, q11, q2
- vqsub.s16 q6, q11, q2
- vqsub.s16 q7, q10, q3
-
- ; rotate
- vtrn.32 q4, q6
- vtrn.32 q5, q7
- vtrn.16 q4, q5
- vtrn.16 q6, q7
- ; idct loop 2
- ; q4: l 0, 4, 8,12 r 0, 4, 8,12
- ; q5: l 1, 5, 9,13 r 1, 5, 9,13
- ; q6: l 2, 6,10,14 r 2, 6,10,14
- ; q7: l 3, 7,11,15 r 3, 7,11,15
-
- ; q8: 1 * sinpi : c1/temp1
- ; q9: 3 * sinpi : d1/temp2
- ; q10: 1 * cospi
- ; q11: 3 * cospi
- vqdmulh.s16 q8, q5, d0[2] ; sinpi8sqrt2
- vqdmulh.s16 q9, q7, d0[2]
- vqdmulh.s16 q10, q5, d0[0] ; cospi8sqrt2minus1
- vqdmulh.s16 q11, q7, d0[0]
-
- vqadd.s16 q2, q4, q6 ; a1 = 0 + 2
- vqsub.s16 q3, q4, q6 ; b1 = 0 - 2
-
- ; see note on shifting above
- vshr.s16 q10, q10, #1
- vshr.s16 q11, q11, #1
-
- ; q10: 1 + 1 * cospi : d1/temp1
- ; q11: 3 + 3 * cospi : c1/temp2
- vqadd.s16 q10, q5, q10
- vqadd.s16 q11, q7, q11
-
- ; q8: c1 = temp1 - temp2
- ; q9: d1 = temp1 + temp2
- vqsub.s16 q8, q8, q11
- vqadd.s16 q9, q10, q9
-
- ; a1+d1
- ; b1+c1
- ; b1-c1
- ; a1-d1
- vqadd.s16 q4, q2, q9
- vqadd.s16 q5, q3, q8
- vqsub.s16 q6, q3, q8
- vqsub.s16 q7, q2, q9
-
- ; +4 >> 3 (rounding)
- vrshr.s16 q4, q4, #3 ; lo
- vrshr.s16 q5, q5, #3
- vrshr.s16 q6, q6, #3 ; hi
- vrshr.s16 q7, q7, #3
-
- vtrn.32 q4, q6
- vtrn.32 q5, q7
- vtrn.16 q4, q5
- vtrn.16 q6, q7
-
- ; adding pre
- ; input is still packed. pre was read interleaved
- vaddw.u8 q4, q4, d28
- vaddw.u8 q5, q5, d29
- vaddw.u8 q6, q6, d30
- vaddw.u8 q7, q7, d31
-
- vmov.i16 q14, #0
- vmov q15, q14
- vst1.16 {q14, q15}, [r0] ; write over high input
- sub r0, r0, #32
- vst1.16 {q14, q15}, [r0] ; write over low input
-
- ;saturate and narrow
- vqmovun.s16 d0, q4 ; lo
- vqmovun.s16 d1, q5
- vqmovun.s16 d2, q6 ; hi
- vqmovun.s16 d3, q7
-
- ldr r1, [sp] ; stride
- add r2, r3, #4 ; hi
- vst1.32 {d0[0]}, [r3], r1 ; lo
- vst1.32 {d0[1]}, [r2], r1 ; hi
- vst1.32 {d1[0]}, [r3], r1
- vst1.32 {d1[1]}, [r2], r1
- vst1.32 {d2[0]}, [r3], r1
- vst1.32 {d2[1]}, [r2], r1
- vst1.32 {d3[0]}, [r3]
- vst1.32 {d3[1]}, [r2]
-
- bx lr
-
- ENDP ; |idct_dequant_dc_full_2x_neon|
-
-; Constant Pool
-_CONSTANTS_
-cospi8sqrt2minus1 DCD 0x4e7b
-; because the lowest bit in 0x8a8c is 0, we can pre-shift this
-sinpi8sqrt2 DCD 0x4546
-
- END
diff --git a/vp8/decoder/arm/neon/idct_dequant_full_2x_neon.asm b/vp8/decoder/arm/neon/idct_dequant_full_2x_neon.asm
deleted file mode 100644
index ba27859..0000000
--- a/vp8/decoder/arm/neon/idct_dequant_full_2x_neon.asm
+++ /dev/null
@@ -1,198 +0,0 @@
-;
-; Copyright (c) 2010 The Webm project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |idct_dequant_full_2x_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void idct_dequant_full_2x_neon(short *q, short *dq, unsigned char *pre,
-; unsigned char *dst, int pitch, int stride);
-; r0 *q,
-; r1 *dq,
-; r2 *pre
-; r3 *dst
-; sp pitch
-; sp+4 stride
-|idct_dequant_full_2x_neon| PROC
- vld1.16 {q0, q1}, [r1] ; dq (same l/r)
- vld1.16 {q2, q3}, [r0] ; l q
- ldr r1, [sp] ; pitch
- add r0, r0, #32
- vld1.16 {q4, q5}, [r0] ; r q
- add r12, r2, #4
- ; interleave the predictors
- vld1.32 {d28[0]}, [r2], r1 ; l pre
- vld1.32 {d28[1]}, [r12], r1 ; r pre
- vld1.32 {d29[0]}, [r2], r1
- vld1.32 {d29[1]}, [r12], r1
- vld1.32 {d30[0]}, [r2], r1
- vld1.32 {d30[1]}, [r12], r1
- vld1.32 {d31[0]}, [r2]
- vld1.32 {d31[1]}, [r12]
-
- adr r2, _CONSTANTS_
-
- ; dequant: q[i] = q[i] * dq[i]
- vmul.i16 q2, q2, q0
- vmul.i16 q3, q3, q1
- vmul.i16 q4, q4, q0
- vmul.i16 q5, q5, q1
-
- vld1.16 {d0}, [r2]
-
- ; q2: l0r0 q3: l8r8
- ; q4: l4r4 q5: l12r12
- vswp d5, d8
- vswp d7, d10
-
- ; _CONSTANTS_ * 4,12 >> 16
- ; q6: 4 * sinpi : c1/temp1
- ; q7: 12 * sinpi : d1/temp2
- ; q8: 4 * cospi
- ; q9: 12 * cospi
- vqdmulh.s16 q6, q4, d0[2] ; sinpi8sqrt2
- vqdmulh.s16 q7, q5, d0[2]
- vqdmulh.s16 q8, q4, d0[0] ; cospi8sqrt2minus1
- vqdmulh.s16 q9, q5, d0[0]
-
- vqadd.s16 q10, q2, q3 ; a1 = 0 + 8
- vqsub.s16 q11, q2, q3 ; b1 = 0 - 8
-
- ; vqdmulh only accepts signed values. this was a problem because
- ; our constant had the high bit set, and was treated as a negative value.
- ; vqdmulh also doubles the value before it shifts by 16. we need to
- ; compensate for this. in the case of sinpi8sqrt2, the lowest bit is 0,
- ; so we can shift the constant without losing precision. this avoids
- ; shift again afterward, but also avoids the sign issue. win win!
- ; for cospi8sqrt2minus1 the lowest bit is 1, so we lose precision if we
- ; pre-shift it
- vshr.s16 q8, q8, #1
- vshr.s16 q9, q9, #1
-
- ; q4: 4 + 4 * cospi : d1/temp1
- ; q5: 12 + 12 * cospi : c1/temp2
- vqadd.s16 q4, q4, q8
- vqadd.s16 q5, q5, q9
-
- ; c1 = temp1 - temp2
- ; d1 = temp1 + temp2
- vqsub.s16 q2, q6, q5
- vqadd.s16 q3, q4, q7
-
- ; [0]: a1+d1
- ; [1]: b1+c1
- ; [2]: b1-c1
- ; [3]: a1-d1
- vqadd.s16 q4, q10, q3
- vqadd.s16 q5, q11, q2
- vqsub.s16 q6, q11, q2
- vqsub.s16 q7, q10, q3
-
- ; rotate
- vtrn.32 q4, q6
- vtrn.32 q5, q7
- vtrn.16 q4, q5
- vtrn.16 q6, q7
- ; idct loop 2
- ; q4: l 0, 4, 8,12 r 0, 4, 8,12
- ; q5: l 1, 5, 9,13 r 1, 5, 9,13
- ; q6: l 2, 6,10,14 r 2, 6,10,14
- ; q7: l 3, 7,11,15 r 3, 7,11,15
-
- ; q8: 1 * sinpi : c1/temp1
- ; q9: 3 * sinpi : d1/temp2
- ; q10: 1 * cospi
- ; q11: 3 * cospi
- vqdmulh.s16 q8, q5, d0[2] ; sinpi8sqrt2
- vqdmulh.s16 q9, q7, d0[2]
- vqdmulh.s16 q10, q5, d0[0] ; cospi8sqrt2minus1
- vqdmulh.s16 q11, q7, d0[0]
-
- vqadd.s16 q2, q4, q6 ; a1 = 0 + 2
- vqsub.s16 q3, q4, q6 ; b1 = 0 - 2
-
- ; see note on shifting above
- vshr.s16 q10, q10, #1
- vshr.s16 q11, q11, #1
-
- ; q10: 1 + 1 * cospi : d1/temp1
- ; q11: 3 + 3 * cospi : c1/temp2
- vqadd.s16 q10, q5, q10
- vqadd.s16 q11, q7, q11
-
- ; q8: c1 = temp1 - temp2
- ; q9: d1 = temp1 + temp2
- vqsub.s16 q8, q8, q11
- vqadd.s16 q9, q10, q9
-
- ; a1+d1
- ; b1+c1
- ; b1-c1
- ; a1-d1
- vqadd.s16 q4, q2, q9
- vqadd.s16 q5, q3, q8
- vqsub.s16 q6, q3, q8
- vqsub.s16 q7, q2, q9
-
- ; +4 >> 3 (rounding)
- vrshr.s16 q4, q4, #3 ; lo
- vrshr.s16 q5, q5, #3
- vrshr.s16 q6, q6, #3 ; hi
- vrshr.s16 q7, q7, #3
-
- vtrn.32 q4, q6
- vtrn.32 q5, q7
- vtrn.16 q4, q5
- vtrn.16 q6, q7
-
- ; adding pre
- ; input is still packed. pre was read interleaved
- vaddw.u8 q4, q4, d28
- vaddw.u8 q5, q5, d29
- vaddw.u8 q6, q6, d30
- vaddw.u8 q7, q7, d31
-
- vmov.i16 q14, #0
- vmov q15, q14
- vst1.16 {q14, q15}, [r0] ; write over high input
- sub r0, r0, #32
- vst1.16 {q14, q15}, [r0] ; write over low input
-
- ;saturate and narrow
- vqmovun.s16 d0, q4 ; lo
- vqmovun.s16 d1, q5
- vqmovun.s16 d2, q6 ; hi
- vqmovun.s16 d3, q7
-
- ldr r1, [sp, #4] ; stride
- add r2, r3, #4 ; hi
- vst1.32 {d0[0]}, [r3], r1 ; lo
- vst1.32 {d0[1]}, [r2], r1 ; hi
- vst1.32 {d1[0]}, [r3], r1
- vst1.32 {d1[1]}, [r2], r1
- vst1.32 {d2[0]}, [r3], r1
- vst1.32 {d2[1]}, [r2], r1
- vst1.32 {d3[0]}, [r3]
- vst1.32 {d3[1]}, [r2]
-
- bx lr
-
- ENDP ; |idct_dequant_full_2x_neon|
-
-; Constant Pool
-_CONSTANTS_
-cospi8sqrt2minus1 DCD 0x4e7b
-; because the lowest bit in 0x8a8c is 0, we can pre-shift this
-sinpi8sqrt2 DCD 0x4546
-
- END
diff --git a/vp8/decoder/asm_dec_offsets.c b/vp8/decoder/asm_dec_offsets.c
deleted file mode 100644
index e485cb4..0000000
--- a/vp8/decoder/asm_dec_offsets.c
+++ /dev/null
@@ -1,57 +0,0 @@
-/*
- * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include <stddef.h>
-
-#include "onyxd_int.h"
-
-#define DEFINE(sym, val) int sym = val;
-
-/*
-#define BLANK() asm volatile("\n->" : : )
-*/
-
-/*
- * int main(void)
- * {
- */
-
-DEFINE(detok_scan, offsetof(DETOK, scan));
-DEFINE(detok_ptr_block2leftabove, offsetof(DETOK, ptr_block2leftabove));
-DEFINE(detok_coef_tree_ptr, offsetof(DETOK, vp8_coef_tree_ptr));
-DEFINE(detok_teb_base_ptr, offsetof(DETOK, teb_base_ptr));
-DEFINE(detok_norm_ptr, offsetof(DETOK, norm_ptr));
-DEFINE(detok_ptr_coef_bands_x, offsetof(DETOK, ptr_coef_bands_x));
-
-DEFINE(detok_A, offsetof(DETOK, A));
-DEFINE(detok_L, offsetof(DETOK, L));
-
-DEFINE(detok_qcoeff_start_ptr, offsetof(DETOK, qcoeff_start_ptr));
-DEFINE(detok_current_bc, offsetof(DETOK, current_bc));
-DEFINE(detok_coef_probs, offsetof(DETOK, coef_probs));
-DEFINE(detok_eob, offsetof(DETOK, eob));
-
-DEFINE(bool_decoder_user_buffer_end, offsetof(BOOL_DECODER, user_buffer_end));
-DEFINE(bool_decoder_user_buffer, offsetof(BOOL_DECODER, user_buffer));
-DEFINE(bool_decoder_value, offsetof(BOOL_DECODER, value));
-DEFINE(bool_decoder_count, offsetof(BOOL_DECODER, count));
-DEFINE(bool_decoder_range, offsetof(BOOL_DECODER, range));
-
-DEFINE(tokenextrabits_min_val, offsetof(TOKENEXTRABITS, min_val));
-DEFINE(tokenextrabits_length, offsetof(TOKENEXTRABITS, Length));
-
-//add asserts for any offset that is not supported by assembly code
-//add asserts for any size that is not supported by assembly code
-/*
- * return 0;
- * }
- */
diff --git a/vp8/decoder/dboolhuff.c b/vp8/decoder/dboolhuff.c
deleted file mode 100644
index 8527d51..0000000
--- a/vp8/decoder/dboolhuff.c
+++ /dev/null
@@ -1,65 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "dboolhuff.h"
-#include "vpx_ports/mem.h"
-#include "vpx_mem/vpx_mem.h"
-
-DECLARE_ALIGNED(16, const unsigned char, vp8dx_bitreader_norm[256]) =
-{
- 0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
- 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
- 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
-};
-
-
-int vp8dx_start_decode(BOOL_DECODER *br,
- const unsigned char *source,
- unsigned int source_sz)
-{
- br->user_buffer_end = source+source_sz;
- br->user_buffer = source;
- br->value = 0;
- br->count = -8;
- br->range = 255;
-
- if (source_sz && !source)
- return 1;
-
- /* Populate the buffer */
- vp8dx_bool_decoder_fill(br);
-
- return 0;
-}
-
-
-void vp8dx_bool_decoder_fill(BOOL_DECODER *br)
-{
- const unsigned char *bufptr;
- const unsigned char *bufend;
- VP8_BD_VALUE value;
- int count;
- bufend = br->user_buffer_end;
- bufptr = br->user_buffer;
- value = br->value;
- count = br->count;
-
- VP8DX_BOOL_DECODER_FILL(count, value, bufptr, bufend);
-
- br->user_buffer = bufptr;
- br->value = value;
- br->count = count;
-}
diff --git a/vp8/decoder/dboolhuff.h b/vp8/decoder/dboolhuff.h
deleted file mode 100644
index a83e3f0..0000000
--- a/vp8/decoder/dboolhuff.h
+++ /dev/null
@@ -1,144 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef DBOOLHUFF_H
-#define DBOOLHUFF_H
-#include <stddef.h>
-#include <limits.h>
-#include "vpx_ports/config.h"
-#include "vpx_ports/mem.h"
-#include "vpx/vpx_integer.h"
-
-typedef size_t VP8_BD_VALUE;
-
-# define VP8_BD_VALUE_SIZE ((int)sizeof(VP8_BD_VALUE)*CHAR_BIT)
-/*This is meant to be a large, positive constant that can still be efficiently
- loaded as an immediate (on platforms like ARM, for example).
- Even relatively modest values like 100 would work fine.*/
-# define VP8_LOTS_OF_BITS (0x40000000)
-
-typedef struct
-{
- const unsigned char *user_buffer_end;
- const unsigned char *user_buffer;
- VP8_BD_VALUE value;
- int count;
- unsigned int range;
-} BOOL_DECODER;
-
-DECLARE_ALIGNED(16, extern const unsigned char, vp8dx_bitreader_norm[256]);
-
-int vp8dx_start_decode(BOOL_DECODER *br,
- const unsigned char *source,
- unsigned int source_sz);
-
-void vp8dx_bool_decoder_fill(BOOL_DECODER *br);
-
-/*The refill loop is used in several places, so define it in a macro to make
- sure they're all consistent.
- An inline function would be cleaner, but has a significant penalty, because
- multiple BOOL_DECODER fields must be modified, and the compiler is not smart
- enough to eliminate the stores to those fields and the subsequent reloads
- from them when inlining the function.*/
-#define VP8DX_BOOL_DECODER_FILL(_count,_value,_bufptr,_bufend) \
- do \
- { \
- int shift; \
- for(shift = VP8_BD_VALUE_SIZE - 8 - ((_count) + 8); shift >= 0; ) \
- { \
- if((_bufptr) >= (_bufend)) { \
- (_count) = VP8_LOTS_OF_BITS; \
- break; \
- } \
- (_count) += 8; \
- (_value) |= (VP8_BD_VALUE)*(_bufptr)++ << shift; \
- shift -= 8; \
- } \
- } \
- while(0)
-
-
-static int vp8dx_decode_bool(BOOL_DECODER *br, int probability) {
- unsigned int bit = 0;
- VP8_BD_VALUE value;
- unsigned int split;
- VP8_BD_VALUE bigsplit;
- int count;
- unsigned int range;
-
- value = br->value;
- count = br->count;
- range = br->range;
-
- split = 1 + (((range - 1) * probability) >> 8);
- bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8);
-
- range = split;
-
- if (value >= bigsplit)
- {
- range = br->range - split;
- value = value - bigsplit;
- bit = 1;
- }
-
- {
- register unsigned int shift = vp8dx_bitreader_norm[range];
- range <<= shift;
- value <<= shift;
- count -= shift;
- }
- br->value = value;
- br->count = count;
- br->range = range;
- if(count < 0)
- vp8dx_bool_decoder_fill(br);
- return bit;
-}
-
-static int vp8_decode_value(BOOL_DECODER *br, int bits)
-{
- int z = 0;
- int bit;
-
- for (bit = bits - 1; bit >= 0; bit--)
- {
- z |= (vp8dx_decode_bool(br, 0x80) << bit);
- }
-
- return z;
-}
-
-static int vp8dx_bool_error(BOOL_DECODER *br)
-{
- /* Check if we have reached the end of the buffer.
- *
- * Variable 'count' stores the number of bits in the 'value' buffer,
- * minus 8. So if count == 8, there are 16 bits available to be read.
- * Normally, count is filled with 8 and one byte is filled into the
- * value buffer. When we reach the end of the buffer, count is instead
- * filled with VP8_LOTS_OF_BITS, 8 of which represent the last 8 real
- * bits from the bitstream. So the last bit in the bitstream will be
- * represented by count == VP8_LOTS_OF_BITS - 16.
- */
- if ((br->count > VP8_BD_VALUE_SIZE)
- && (br->count <= VP8_LOTS_OF_BITS - 16))
- {
- /* We have tried to decode bits after the end of
- * stream was encountered.
- */
- return 1;
- }
-
- /* No error. */
- return 0;
-}
-#endif
diff --git a/vp8/decoder/decodemv.c b/vp8/decoder/decodemv.c
deleted file mode 100644
index e5830e8..0000000
--- a/vp8/decoder/decodemv.c
+++ /dev/null
@@ -1,551 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "treereader.h"
-#include "vp8/common/entropymv.h"
-#include "vp8/common/entropymode.h"
-#include "onyxd_int.h"
-#include "vp8/common/findnearmv.h"
-
-#if CONFIG_DEBUG
-#include <assert.h>
-#endif
-static int vp8_read_bmode(vp8_reader *bc, const vp8_prob *p)
-{
- const int i = vp8_treed_read(bc, vp8_bmode_tree, p);
-
- return i;
-}
-
-
-static int vp8_read_ymode(vp8_reader *bc, const vp8_prob *p)
-{
- const int i = vp8_treed_read(bc, vp8_ymode_tree, p);
-
- return i;
-}
-
-static int vp8_kfread_ymode(vp8_reader *bc, const vp8_prob *p)
-{
- const int i = vp8_treed_read(bc, vp8_kf_ymode_tree, p);
-
- return i;
-}
-
-
-
-static int vp8_read_uv_mode(vp8_reader *bc, const vp8_prob *p)
-{
- const int i = vp8_treed_read(bc, vp8_uv_mode_tree, p);
-
- return i;
-}
-
-static void vp8_read_mb_features(vp8_reader *r, MB_MODE_INFO *mi, MACROBLOCKD *x)
-{
- /* Is segmentation enabled */
- if (x->segmentation_enabled && x->update_mb_segmentation_map)
- {
- /* If so then read the segment id. */
- if (vp8_read(r, x->mb_segment_tree_probs[0]))
- mi->segment_id = (unsigned char)(2 + vp8_read(r, x->mb_segment_tree_probs[2]));
- else
- mi->segment_id = (unsigned char)(vp8_read(r, x->mb_segment_tree_probs[1]));
- }
-}
-
-static void vp8_kfread_modes(VP8D_COMP *pbi, MODE_INFO *m, int mb_row, int mb_col)
-{
- vp8_reader *const bc = & pbi->bc;
- const int mis = pbi->common.mode_info_stride;
-
- {
- MB_PREDICTION_MODE y_mode;
-
- /* Read the Macroblock segmentation map if it is being updated explicitly this frame (reset to 0 above by default)
- * By default on a key frame reset all MBs to segment 0
- */
- m->mbmi.segment_id = 0;
-
- if (pbi->mb.update_mb_segmentation_map)
- vp8_read_mb_features(bc, &m->mbmi, &pbi->mb);
-
- /* Read the macroblock coeff skip flag if this feature is in use, else default to 0 */
- if (pbi->common.mb_no_coeff_skip)
- m->mbmi.mb_skip_coeff = vp8_read(bc, pbi->prob_skip_false);
- else
- m->mbmi.mb_skip_coeff = 0;
-
- y_mode = (MB_PREDICTION_MODE) vp8_kfread_ymode(bc, pbi->common.kf_ymode_prob);
-
- m->mbmi.ref_frame = INTRA_FRAME;
-
- if ((m->mbmi.mode = y_mode) == B_PRED)
- {
- int i = 0;
-
- do
- {
- const B_PREDICTION_MODE A = vp8_above_bmi(m, i, mis)->mode;
- const B_PREDICTION_MODE L = vp8_left_bmi(m, i)->mode;
-
- m->bmi[i].mode = (B_PREDICTION_MODE) vp8_read_bmode(bc, pbi->common.kf_bmode_prob [A] [L]);
- }
- while (++i < 16);
- }
- else
- {
- int BMode;
- int i = 0;
-
- switch (y_mode)
- {
- case DC_PRED:
- BMode = B_DC_PRED;
- break;
- case V_PRED:
- BMode = B_VE_PRED;
- break;
- case H_PRED:
- BMode = B_HE_PRED;
- break;
- case TM_PRED:
- BMode = B_TM_PRED;
- break;
- default:
- BMode = B_DC_PRED;
- break;
- }
-
- do
- {
- m->bmi[i].mode = (B_PREDICTION_MODE)BMode;
- }
- while (++i < 16);
- }
-
- m->mbmi.uv_mode = (MB_PREDICTION_MODE)vp8_read_uv_mode(bc, pbi->common.kf_uv_mode_prob);
- }
-}
-
-static int read_mvcomponent(vp8_reader *r, const MV_CONTEXT *mvc)
-{
- const vp8_prob *const p = (const vp8_prob *) mvc;
- int x = 0;
-
- if (vp8_read(r, p [mvpis_short])) /* Large */
- {
- int i = 0;
-
- do
- {
- x += vp8_read(r, p [MVPbits + i]) << i;
- }
- while (++i < 3);
-
- i = mvlong_width - 1; /* Skip bit 3, which is sometimes implicit */
-
- do
- {
- x += vp8_read(r, p [MVPbits + i]) << i;
- }
- while (--i > 3);
-
- if (!(x & 0xFFF0) || vp8_read(r, p [MVPbits + 3]))
- x += 8;
- }
- else /* small */
- x = vp8_treed_read(r, vp8_small_mvtree, p + MVPshort);
-
- if (x && vp8_read(r, p [MVPsign]))
- x = -x;
-
- return x;
-}
-
-static void read_mv(vp8_reader *r, MV *mv, const MV_CONTEXT *mvc)
-{
- mv->row = (short)(read_mvcomponent(r, mvc) << 1);
- mv->col = (short)(read_mvcomponent(r, ++mvc) << 1);
-}
-
-
-static void read_mvcontexts(vp8_reader *bc, MV_CONTEXT *mvc)
-{
- int i = 0;
-
- do
- {
- const vp8_prob *up = vp8_mv_update_probs[i].prob;
- vp8_prob *p = (vp8_prob *)(mvc + i);
- vp8_prob *const pstop = p + MVPcount;
-
- do
- {
- if (vp8_read(bc, *up++))
- {
- const vp8_prob x = (vp8_prob)vp8_read_literal(bc, 7);
-
- *p = x ? x << 1 : 1;
- }
- }
- while (++p < pstop);
- }
- while (++i < 2);
-}
-
-
-static MB_PREDICTION_MODE read_mv_ref(vp8_reader *bc, const vp8_prob *p)
-{
- const int i = vp8_treed_read(bc, vp8_mv_ref_tree, p);
-
- return (MB_PREDICTION_MODE)i;
-}
-
-static MB_PREDICTION_MODE sub_mv_ref(vp8_reader *bc, const vp8_prob *p)
-{
- const int i = vp8_treed_read(bc, vp8_sub_mv_ref_tree, p);
-
- return (MB_PREDICTION_MODE)i;
-}
-
-#ifdef VPX_MODE_COUNT
-unsigned int vp8_mv_cont_count[5][4] =
-{
- { 0, 0, 0, 0 },
- { 0, 0, 0, 0 },
- { 0, 0, 0, 0 },
- { 0, 0, 0, 0 },
- { 0, 0, 0, 0 }
-};
-#endif
-
-static const unsigned char mbsplit_fill_count[4] = {8, 8, 4, 1};
-static const unsigned char mbsplit_fill_offset[4][16] = {
- { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15},
- { 0, 1, 4, 5, 8, 9, 12, 13, 2, 3, 6, 7, 10, 11, 14, 15},
- { 0, 1, 4, 5, 2, 3, 6, 7, 8, 9, 12, 13, 10, 11, 14, 15},
- { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}
-};
-
-
-
-
-static void mb_mode_mv_init(VP8D_COMP *pbi)
-{
- vp8_reader *const bc = & pbi->bc;
- MV_CONTEXT *const mvc = pbi->common.fc.mvc;
-
- pbi->prob_skip_false = 0;
- if (pbi->common.mb_no_coeff_skip)
- pbi->prob_skip_false = (vp8_prob)vp8_read_literal(bc, 8);
-
- if(pbi->common.frame_type != KEY_FRAME)
- {
- pbi->prob_intra = (vp8_prob)vp8_read_literal(bc, 8);
- pbi->prob_last = (vp8_prob)vp8_read_literal(bc, 8);
- pbi->prob_gf = (vp8_prob)vp8_read_literal(bc, 8);
-
- if (vp8_read_bit(bc))
- {
- int i = 0;
-
- do
- {
- pbi->common.fc.ymode_prob[i] = (vp8_prob) vp8_read_literal(bc, 8);
- }
- while (++i < 4);
- }
-
- if (vp8_read_bit(bc))
- {
- int i = 0;
-
- do
- {
- pbi->common.fc.uv_mode_prob[i] = (vp8_prob) vp8_read_literal(bc, 8);
- }
- while (++i < 3);
- }
-
- read_mvcontexts(bc, mvc);
- }
-}
-
-static void read_mb_modes_mv(VP8D_COMP *pbi, MODE_INFO *mi, MB_MODE_INFO *mbmi,
- int mb_row, int mb_col)
-{
- const MV Zero = { 0, 0};
- vp8_reader *const bc = & pbi->bc;
- MV_CONTEXT *const mvc = pbi->common.fc.mvc;
- const int mis = pbi->common.mode_info_stride;
-
- MV *const mv = & mbmi->mv.as_mv;
- int mb_to_left_edge;
- int mb_to_right_edge;
- int mb_to_top_edge;
- int mb_to_bottom_edge;
-
- mb_to_top_edge = pbi->mb.mb_to_top_edge;
- mb_to_bottom_edge = pbi->mb.mb_to_bottom_edge;
- mb_to_top_edge -= LEFT_TOP_MARGIN;
- mb_to_bottom_edge += RIGHT_BOTTOM_MARGIN;
-
- mbmi->need_to_clamp_mvs = 0;
- /* Distance of Mb to the various image edges.
- * These specified to 8th pel as they are always compared to MV values that are in 1/8th pel units
- */
- pbi->mb.mb_to_left_edge =
- mb_to_left_edge = -((mb_col * 16) << 3);
- mb_to_left_edge -= LEFT_TOP_MARGIN;
-
- pbi->mb.mb_to_right_edge =
- mb_to_right_edge = ((pbi->common.mb_cols - 1 - mb_col) * 16) << 3;
- mb_to_right_edge += RIGHT_BOTTOM_MARGIN;
-
- /* If required read in new segmentation data for this MB */
- if (pbi->mb.update_mb_segmentation_map)
- vp8_read_mb_features(bc, mbmi, &pbi->mb);
-
- /* Read the macroblock coeff skip flag if this feature is in use, else default to 0 */
- if (pbi->common.mb_no_coeff_skip)
- mbmi->mb_skip_coeff = vp8_read(bc, pbi->prob_skip_false);
- else
- mbmi->mb_skip_coeff = 0;
-
- if ((mbmi->ref_frame = (MV_REFERENCE_FRAME) vp8_read(bc, pbi->prob_intra))) /* inter MB */
- {
- int rct[4];
- vp8_prob mv_ref_p [VP8_MVREFS-1];
- MV nearest, nearby, best_mv;
-
- if (vp8_read(bc, pbi->prob_last))
- {
- mbmi->ref_frame = (MV_REFERENCE_FRAME)((int)mbmi->ref_frame + (int)(1 + vp8_read(bc, pbi->prob_gf)));
- }
-
- vp8_find_near_mvs(&pbi->mb, mi, &nearest, &nearby, &best_mv, rct, mbmi->ref_frame, pbi->common.ref_frame_sign_bias);
-
- vp8_mv_ref_probs(mv_ref_p, rct);
-
- mbmi->uv_mode = DC_PRED;
- switch (mbmi->mode = read_mv_ref(bc, mv_ref_p))
- {
- case SPLITMV:
- {
- const int s = mbmi->partitioning =
- vp8_treed_read(bc, vp8_mbsplit_tree, vp8_mbsplit_probs);
- const int num_p = vp8_mbsplit_count [s];
- int j = 0;
-
- do /* for each subset j */
- {
- B_MODE_INFO bmi;
- MV *const mv = & bmi.mv.as_mv;
-
- int k; /* first block in subset j */
- int mv_contz;
- k = vp8_mbsplit_offset[s][j];
-
- mv_contz = vp8_mv_cont(&(vp8_left_bmi(mi, k)->mv.as_mv), &(vp8_above_bmi(mi, k, mis)->mv.as_mv));
-
- switch (bmi.mode = (B_PREDICTION_MODE) sub_mv_ref(bc, vp8_sub_mv_ref_prob2 [mv_contz])) /*pc->fc.sub_mv_ref_prob))*/
- {
- case NEW4X4:
- read_mv(bc, mv, (const MV_CONTEXT *) mvc);
- mv->row += best_mv.row;
- mv->col += best_mv.col;
- #ifdef VPX_MODE_COUNT
- vp8_mv_cont_count[mv_contz][3]++;
- #endif
- break;
- case LEFT4X4:
- *mv = vp8_left_bmi(mi, k)->mv.as_mv;
- #ifdef VPX_MODE_COUNT
- vp8_mv_cont_count[mv_contz][0]++;
- #endif
- break;
- case ABOVE4X4:
- *mv = vp8_above_bmi(mi, k, mis)->mv.as_mv;
- #ifdef VPX_MODE_COUNT
- vp8_mv_cont_count[mv_contz][1]++;
- #endif
- break;
- case ZERO4X4:
- *mv = Zero;
- #ifdef VPX_MODE_COUNT
- vp8_mv_cont_count[mv_contz][2]++;
- #endif
- break;
- default:
- break;
- }
-
- mbmi->need_to_clamp_mvs |= (mv->col < mb_to_left_edge) ? 1 : 0;
- mbmi->need_to_clamp_mvs |= (mv->col > mb_to_right_edge) ? 1 : 0;
- mbmi->need_to_clamp_mvs |= (mv->row < mb_to_top_edge) ? 1 : 0;
- mbmi->need_to_clamp_mvs |= (mv->row > mb_to_bottom_edge) ? 1 : 0;
-
- {
- /* Fill (uniform) modes, mvs of jth subset.
- Must do it here because ensuing subsets can
- refer back to us via "left" or "above". */
- const unsigned char *fill_offset;
- unsigned int fill_count = mbsplit_fill_count[s];
-
- fill_offset = &mbsplit_fill_offset[s][(unsigned char)j * mbsplit_fill_count[s]];
-
- do {
- mi->bmi[ *fill_offset] = bmi;
- fill_offset++;
-
- }while (--fill_count);
- }
-
- }
- while (++j < num_p);
- }
-
- *mv = mi->bmi[15].mv.as_mv;
-
- break; /* done with SPLITMV */
-
- case NEARMV:
- *mv = nearby;
- /* Clip "next_nearest" so that it does not extend to far out of image */
- mv->col = (mv->col < mb_to_left_edge) ? mb_to_left_edge : mv->col;
- mv->col = (mv->col > mb_to_right_edge) ? mb_to_right_edge : mv->col;
- mv->row = (mv->row < mb_to_top_edge) ? mb_to_top_edge : mv->row;
- mv->row = (mv->row > mb_to_bottom_edge) ? mb_to_bottom_edge : mv->row;
- goto propagate_mv;
-
- case NEARESTMV:
- *mv = nearest;
- /* Clip "next_nearest" so that it does not extend to far out of image */
- mv->col = (mv->col < mb_to_left_edge) ? mb_to_left_edge : mv->col;
- mv->col = (mv->col > mb_to_right_edge) ? mb_to_right_edge : mv->col;
- mv->row = (mv->row < mb_to_top_edge) ? mb_to_top_edge : mv->row;
- mv->row = (mv->row > mb_to_bottom_edge) ? mb_to_bottom_edge : mv->row;
- goto propagate_mv;
-
- case ZEROMV:
- *mv = Zero;
- goto propagate_mv;
-
- case NEWMV:
- read_mv(bc, mv, (const MV_CONTEXT *) mvc);
- mv->row += best_mv.row;
- mv->col += best_mv.col;
-
- /* Don't need to check this on NEARMV and NEARESTMV modes
- * since those modes clamp the MV. The NEWMV mode does not,
- * so signal to the prediction stage whether special
- * handling may be required.
- */
- mbmi->need_to_clamp_mvs = (mv->col < mb_to_left_edge) ? 1 : 0;
- mbmi->need_to_clamp_mvs |= (mv->col > mb_to_right_edge) ? 1 : 0;
- mbmi->need_to_clamp_mvs |= (mv->row < mb_to_top_edge) ? 1 : 0;
- mbmi->need_to_clamp_mvs |= (mv->row > mb_to_bottom_edge) ? 1 : 0;
-
- propagate_mv: /* same MV throughout */
- {
- /*int i=0;
- do
- {
- mi->bmi[i].mv.as_mv = *mv;
- }
- while( ++i < 16);*/
-
- mi->bmi[0].mv.as_mv = *mv;
- mi->bmi[1].mv.as_mv = *mv;
- mi->bmi[2].mv.as_mv = *mv;
- mi->bmi[3].mv.as_mv = *mv;
- mi->bmi[4].mv.as_mv = *mv;
- mi->bmi[5].mv.as_mv = *mv;
- mi->bmi[6].mv.as_mv = *mv;
- mi->bmi[7].mv.as_mv = *mv;
- mi->bmi[8].mv.as_mv = *mv;
- mi->bmi[9].mv.as_mv = *mv;
- mi->bmi[10].mv.as_mv = *mv;
- mi->bmi[11].mv.as_mv = *mv;
- mi->bmi[12].mv.as_mv = *mv;
- mi->bmi[13].mv.as_mv = *mv;
- mi->bmi[14].mv.as_mv = *mv;
- mi->bmi[15].mv.as_mv = *mv;
- }
- break;
- default:;
- #if CONFIG_DEBUG
- assert(0);
- #endif
- }
- }
- else
- {
- /* MB is intra coded */
- int j = 0;
- do
- {
- mi->bmi[j].mv.as_mv = Zero;
- }
- while (++j < 16);
-
- if ((mbmi->mode = (MB_PREDICTION_MODE) vp8_read_ymode(bc, pbi->common.fc.ymode_prob)) == B_PRED)
- {
- j = 0;
- do
- {
- mi->bmi[j].mode = (B_PREDICTION_MODE)vp8_read_bmode(bc, pbi->common.fc.bmode_prob);
- }
- while (++j < 16);
- }
-
- mbmi->uv_mode = (MB_PREDICTION_MODE)vp8_read_uv_mode(bc, pbi->common.fc.uv_mode_prob);
- }
-
-}
-
-void vp8_decode_mode_mvs(VP8D_COMP *pbi)
-{
- MODE_INFO *mi = pbi->common.mi;
- int mb_row = -1;
-
- mb_mode_mv_init(pbi);
-
- while (++mb_row < pbi->common.mb_rows)
- {
- int mb_col = -1;
- int mb_to_top_edge;
- int mb_to_bottom_edge;
-
- pbi->mb.mb_to_top_edge =
- mb_to_top_edge = -((mb_row * 16)) << 3;
- mb_to_top_edge -= LEFT_TOP_MARGIN;
-
- pbi->mb.mb_to_bottom_edge =
- mb_to_bottom_edge = ((pbi->common.mb_rows - 1 - mb_row) * 16) << 3;
- mb_to_bottom_edge += RIGHT_BOTTOM_MARGIN;
-
- while (++mb_col < pbi->common.mb_cols)
- {
- /*read_mb_modes_mv(pbi, xd->mode_info_context, &xd->mode_info_context->mbmi, mb_row, mb_col);*/
- if(pbi->common.frame_type == KEY_FRAME)
- vp8_kfread_modes(pbi, mi, mb_row, mb_col);
- else
- read_mb_modes_mv(pbi, mi, &mi->mbmi, mb_row, mb_col);
-
- mi++; /* next macroblock */
- }
-
- mi++; /* skip left predictor each row */
- }
-}
-
diff --git a/vp8/decoder/decodframe.c b/vp8/decoder/decodframe.c
deleted file mode 100644
index 82841e8..0000000
--- a/vp8/decoder/decodframe.c
+++ /dev/null
@@ -1,942 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "onyxd_int.h"
-#include "vp8/common/header.h"
-#include "vp8/common/reconintra.h"
-#include "vp8/common/reconintra4x4.h"
-#include "vp8/common/recon.h"
-#include "vp8/common/reconinter.h"
-#include "dequantize.h"
-#include "detokenize.h"
-#include "vp8/common/invtrans.h"
-#include "vp8/common/alloccommon.h"
-#include "vp8/common/entropymode.h"
-#include "vp8/common/quant_common.h"
-#include "vpx_scale/vpxscale.h"
-#include "vpx_scale/yv12extend.h"
-#include "vp8/common/setupintrarecon.h"
-
-#include "decodemv.h"
-#include "vp8/common/extend.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vp8/common/idct.h"
-#include "dequantize.h"
-#include "vp8/common/threading.h"
-#include "decoderthreading.h"
-#include "dboolhuff.h"
-
-#include <assert.h>
-#include <stdio.h>
-
-void vp8cx_init_de_quantizer(VP8D_COMP *pbi)
-{
- int i;
- int Q;
- VP8_COMMON *const pc = & pbi->common;
-
- for (Q = 0; Q < QINDEX_RANGE; Q++)
- {
- pc->Y1dequant[Q][0] = (short)vp8_dc_quant(Q, pc->y1dc_delta_q);
- pc->Y2dequant[Q][0] = (short)vp8_dc2quant(Q, pc->y2dc_delta_q);
- pc->UVdequant[Q][0] = (short)vp8_dc_uv_quant(Q, pc->uvdc_delta_q);
-
- /* all the ac values = ; */
- for (i = 1; i < 16; i++)
- {
- int rc = vp8_default_zig_zag1d[i];
-
- pc->Y1dequant[Q][rc] = (short)vp8_ac_yquant(Q);
- pc->Y2dequant[Q][rc] = (short)vp8_ac2quant(Q, pc->y2ac_delta_q);
- pc->UVdequant[Q][rc] = (short)vp8_ac_uv_quant(Q, pc->uvac_delta_q);
- }
- }
-}
-
-void mb_init_dequantizer(VP8D_COMP *pbi, MACROBLOCKD *xd)
-{
- int i;
- int QIndex;
- MB_MODE_INFO *mbmi = &xd->mode_info_context->mbmi;
- VP8_COMMON *const pc = & pbi->common;
-
- /* Decide whether to use the default or alternate baseline Q value. */
- if (xd->segmentation_enabled)
- {
- /* Abs Value */
- if (xd->mb_segement_abs_delta == SEGMENT_ABSDATA)
- QIndex = xd->segment_feature_data[MB_LVL_ALT_Q][mbmi->segment_id];
-
- /* Delta Value */
- else
- {
- QIndex = pc->base_qindex + xd->segment_feature_data[MB_LVL_ALT_Q][mbmi->segment_id];
- QIndex = (QIndex >= 0) ? ((QIndex <= MAXQ) ? QIndex : MAXQ) : 0; /* Clamp to valid range */
- }
- }
- else
- QIndex = pc->base_qindex;
-
- /* Set up the block level dequant pointers */
- for (i = 0; i < 16; i++)
- {
- xd->block[i].dequant = pc->Y1dequant[QIndex];
- }
-
- for (i = 16; i < 24; i++)
- {
- xd->block[i].dequant = pc->UVdequant[QIndex];
- }
-
- xd->block[24].dequant = pc->Y2dequant[QIndex];
-
-}
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define RTCD_VTABLE(x) (&(pbi)->common.rtcd.x)
-#else
-#define RTCD_VTABLE(x) NULL
-#endif
-
-/* skip_recon_mb() is Modified: Instead of writing the result to predictor buffer and then copying it
- * to dst buffer, we can write the result directly to dst buffer. This eliminates unnecessary copy.
- */
-static void skip_recon_mb(VP8D_COMP *pbi, MACROBLOCKD *xd)
-{
- if (xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
- {
-
- vp8_build_intra_predictors_mbuv_s(xd);
- RECON_INVOKE(&pbi->common.rtcd.recon,
- build_intra_predictors_mby_s)(xd);
- }
- else
- {
- vp8_build_inter_predictors_mb_s(xd);
- }
-}
-
-static void clamp_mv_to_umv_border(MV *mv, const MACROBLOCKD *xd)
-{
- /* If the MV points so far into the UMV border that no visible pixels
- * are used for reconstruction, the subpel part of the MV can be
- * discarded and the MV limited to 16 pixels with equivalent results.
- *
- * This limit kicks in at 19 pixels for the top and left edges, for
- * the 16 pixels plus 3 taps right of the central pixel when subpel
- * filtering. The bottom and right edges use 16 pixels plus 2 pixels
- * left of the central pixel when filtering.
- */
- if (mv->col < (xd->mb_to_left_edge - (19 << 3)))
- mv->col = xd->mb_to_left_edge - (16 << 3);
- else if (mv->col > xd->mb_to_right_edge + (18 << 3))
- mv->col = xd->mb_to_right_edge + (16 << 3);
-
- if (mv->row < (xd->mb_to_top_edge - (19 << 3)))
- mv->row = xd->mb_to_top_edge - (16 << 3);
- else if (mv->row > xd->mb_to_bottom_edge + (18 << 3))
- mv->row = xd->mb_to_bottom_edge + (16 << 3);
-}
-
-/* A version of the above function for chroma block MVs.*/
-static void clamp_uvmv_to_umv_border(MV *mv, const MACROBLOCKD *xd)
-{
- mv->col = (2*mv->col < (xd->mb_to_left_edge - (19 << 3))) ? (xd->mb_to_left_edge - (16 << 3)) >> 1 : mv->col;
- mv->col = (2*mv->col > xd->mb_to_right_edge + (18 << 3)) ? (xd->mb_to_right_edge + (16 << 3)) >> 1 : mv->col;
-
- mv->row = (2*mv->row < (xd->mb_to_top_edge - (19 << 3))) ? (xd->mb_to_top_edge - (16 << 3)) >> 1 : mv->row;
- mv->row = (2*mv->row > xd->mb_to_bottom_edge + (18 << 3)) ? (xd->mb_to_bottom_edge + (16 << 3)) >> 1 : mv->row;
-}
-
-void clamp_mvs(MACROBLOCKD *xd)
-{
- if (xd->mode_info_context->mbmi.mode == SPLITMV)
- {
- int i;
-
- for (i=0; i<16; i++)
- clamp_mv_to_umv_border(&xd->block[i].bmi.mv.as_mv, xd);
- for (i=16; i<24; i++)
- clamp_uvmv_to_umv_border(&xd->block[i].bmi.mv.as_mv, xd);
- }
- else
- {
- clamp_mv_to_umv_border(&xd->mode_info_context->mbmi.mv.as_mv, xd);
- clamp_uvmv_to_umv_border(&xd->block[16].bmi.mv.as_mv, xd);
- }
-
-}
-
-static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd)
-{
- int eobtotal = 0;
- int i, do_clamp = xd->mode_info_context->mbmi.need_to_clamp_mvs;
-
- if (xd->mode_info_context->mbmi.mb_skip_coeff)
- {
- vp8_reset_mb_tokens_context(xd);
- }
- else
- {
- eobtotal = vp8_decode_mb_tokens(pbi, xd);
- }
-
- /* Perform temporary clamping of the MV to be used for prediction */
- if (do_clamp)
- {
- clamp_mvs(xd);
- }
-
- xd->mode_info_context->mbmi.dc_diff = 1;
-
- if (xd->mode_info_context->mbmi.mode != B_PRED && xd->mode_info_context->mbmi.mode != SPLITMV && eobtotal == 0)
- {
- xd->mode_info_context->mbmi.dc_diff = 0;
- skip_recon_mb(pbi, xd);
- return;
- }
-
- if (xd->segmentation_enabled)
- mb_init_dequantizer(pbi, xd);
-
- /* do prediction */
- if (xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
- {
- vp8_build_intra_predictors_mbuv(xd);
-
- if (xd->mode_info_context->mbmi.mode != B_PRED)
- {
- RECON_INVOKE(&pbi->common.rtcd.recon,
- build_intra_predictors_mby)(xd);
- } else {
- vp8_intra_prediction_down_copy(xd);
- }
- }
- else
- {
- vp8_build_inter_predictors_mb(xd);
- }
-
- /* dequantization and idct */
- if (xd->mode_info_context->mbmi.mode != B_PRED && xd->mode_info_context->mbmi.mode != SPLITMV)
- {
- BLOCKD *b = &xd->block[24];
- DEQUANT_INVOKE(&pbi->dequant, block)(b);
-
- /* do 2nd order transform on the dc block */
- if (xd->eobs[24] > 1)
- {
- IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh16)(&b->dqcoeff[0], b->diff);
- ((int *)b->qcoeff)[0] = 0;
- ((int *)b->qcoeff)[1] = 0;
- ((int *)b->qcoeff)[2] = 0;
- ((int *)b->qcoeff)[3] = 0;
- ((int *)b->qcoeff)[4] = 0;
- ((int *)b->qcoeff)[5] = 0;
- ((int *)b->qcoeff)[6] = 0;
- ((int *)b->qcoeff)[7] = 0;
- }
- else
- {
- IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh1)(&b->dqcoeff[0], b->diff);
- ((int *)b->qcoeff)[0] = 0;
- }
-
- DEQUANT_INVOKE (&pbi->dequant, dc_idct_add_y_block)
- (xd->qcoeff, xd->block[0].dequant,
- xd->predictor, xd->dst.y_buffer,
- xd->dst.y_stride, xd->eobs, xd->block[24].diff);
- }
- else if ((xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) && xd->mode_info_context->mbmi.mode == B_PRED)
- {
- for (i = 0; i < 16; i++)
- {
-
- BLOCKD *b = &xd->block[i];
- vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
-
- if (xd->eobs[i] > 1)
- {
- DEQUANT_INVOKE(&pbi->dequant, idct_add)
- (b->qcoeff, b->dequant, b->predictor,
- *(b->base_dst) + b->dst, 16, b->dst_stride);
- }
- else
- {
- IDCT_INVOKE(RTCD_VTABLE(idct), idct1_scalar_add)
- (b->qcoeff[0] * b->dequant[0], b->predictor,
- *(b->base_dst) + b->dst, 16, b->dst_stride);
- ((int *)b->qcoeff)[0] = 0;
- }
- }
-
- }
- else
- {
- DEQUANT_INVOKE (&pbi->dequant, idct_add_y_block)
- (xd->qcoeff, xd->block[0].dequant,
- xd->predictor, xd->dst.y_buffer,
- xd->dst.y_stride, xd->eobs);
- }
-
- DEQUANT_INVOKE (&pbi->dequant, idct_add_uv_block)
- (xd->qcoeff+16*16, xd->block[16].dequant,
- xd->predictor+16*16, xd->dst.u_buffer, xd->dst.v_buffer,
- xd->dst.uv_stride, xd->eobs+16);
-}
-
-
-static int get_delta_q(vp8_reader *bc, int prev, int *q_update)
-{
- int ret_val = 0;
-
- if (vp8_read_bit(bc))
- {
- ret_val = vp8_read_literal(bc, 4);
-
- if (vp8_read_bit(bc))
- ret_val = -ret_val;
- }
-
- /* Trigger a quantizer update if the delta-q value has changed */
- if (ret_val != prev)
- *q_update = 1;
-
- return ret_val;
-}
-
-#ifdef PACKET_TESTING
-#include <stdio.h>
-FILE *vpxlog = 0;
-#endif
-
-
-
-static void
-decode_mb_row(VP8D_COMP *pbi, VP8_COMMON *pc, int mb_row, MACROBLOCKD *xd)
-{
-
- int i;
- int recon_yoffset, recon_uvoffset;
- int mb_col;
- int ref_fb_idx = pc->lst_fb_idx;
- int dst_fb_idx = pc->new_fb_idx;
- int recon_y_stride = pc->yv12_fb[ref_fb_idx].y_stride;
- int recon_uv_stride = pc->yv12_fb[ref_fb_idx].uv_stride;
-
- vpx_memset(&pc->left_context, 0, sizeof(pc->left_context));
- recon_yoffset = mb_row * recon_y_stride * 16;
- recon_uvoffset = mb_row * recon_uv_stride * 8;
- /* reset above block coeffs */
-
- xd->above_context = pc->above_context;
- xd->up_available = (mb_row != 0);
-
- xd->mb_to_top_edge = -((mb_row * 16)) << 3;
- xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3;
-
- for (mb_col = 0; mb_col < pc->mb_cols; mb_col++)
- {
-
- if (xd->mode_info_context->mbmi.mode == SPLITMV || xd->mode_info_context->mbmi.mode == B_PRED)
- {
- for (i = 0; i < 16; i++)
- {
- BLOCKD *d = &xd->block[i];
- vpx_memcpy(&d->bmi, &xd->mode_info_context->bmi[i], sizeof(B_MODE_INFO));
- }
- }
-
- /* Distance of Mb to the various image edges.
- * These are specified to 8th pel as they are always compared to values that are in 1/8th pel units
- */
- xd->mb_to_left_edge = -((mb_col * 16) << 3);
- xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3;
-
- xd->dst.y_buffer = pc->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset;
- xd->dst.u_buffer = pc->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset;
- xd->dst.v_buffer = pc->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset;
-
- xd->left_available = (mb_col != 0);
-
- /* Select the appropriate reference frame for this MB */
- if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME)
- ref_fb_idx = pc->lst_fb_idx;
- else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME)
- ref_fb_idx = pc->gld_fb_idx;
- else
- ref_fb_idx = pc->alt_fb_idx;
-
- xd->pre.y_buffer = pc->yv12_fb[ref_fb_idx].y_buffer + recon_yoffset;
- xd->pre.u_buffer = pc->yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset;
- xd->pre.v_buffer = pc->yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset;
-
- if (xd->mode_info_context->mbmi.ref_frame != INTRA_FRAME)
- {
- /* propagate errors from reference frames */
- xd->corrupted |= pc->yv12_fb[ref_fb_idx].corrupted;
- }
-
- vp8_build_uvmvs(xd, pc->full_pixel);
-
- /*
- if(pc->current_video_frame==0 &&mb_col==1 && mb_row==0)
- pbi->debugoutput =1;
- else
- pbi->debugoutput =0;
- */
- decode_macroblock(pbi, xd);
-
- /* check if the boolean decoder has suffered an error */
- xd->corrupted |= vp8dx_bool_error(xd->current_bc);
-
- recon_yoffset += 16;
- recon_uvoffset += 8;
-
- ++xd->mode_info_context; /* next mb */
-
- xd->above_context++;
-
- }
-
- /* adjust to the next row of mbs */
- vp8_extend_mb_row(
- &pc->yv12_fb[dst_fb_idx],
- xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8
- );
-
- ++xd->mode_info_context; /* skip prediction column */
-}
-
-
-static unsigned int read_partition_size(const unsigned char *cx_size)
-{
- const unsigned int size =
- cx_size[0] + (cx_size[1] << 8) + (cx_size[2] << 16);
- return size;
-}
-
-
-static void setup_token_decoder(VP8D_COMP *pbi,
- const unsigned char *cx_data)
-{
- int num_part;
- int i;
- VP8_COMMON *pc = &pbi->common;
- const unsigned char *user_data_end = pbi->Source + pbi->source_sz;
- vp8_reader *bool_decoder;
- const unsigned char *partition;
-
- /* Parse number of token partitions to use */
- pc->multi_token_partition = (TOKEN_PARTITION)vp8_read_literal(&pbi->bc, 2);
- num_part = 1 << pc->multi_token_partition;
-
- /* Set up pointers to the first partition */
- partition = cx_data;
- bool_decoder = &pbi->bc2;
-
- if (num_part > 1)
- {
- CHECK_MEM_ERROR(pbi->mbc, vpx_malloc(num_part * sizeof(vp8_reader)));
- bool_decoder = pbi->mbc;
- partition += 3 * (num_part - 1);
- }
-
- for (i = 0; i < num_part; i++)
- {
- const unsigned char *partition_size_ptr = cx_data + i * 3;
- ptrdiff_t partition_size;
-
- /* Calculate the length of this partition. The last partition
- * size is implicit.
- */
- if (i < num_part - 1)
- {
- partition_size = read_partition_size(partition_size_ptr);
- }
- else
- {
- partition_size = user_data_end - partition;
- }
-
- if (partition + partition_size > user_data_end
- || partition + partition_size < partition)
- vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
- "Truncated packet or corrupt partition "
- "%d length", i + 1);
-
- if (vp8dx_start_decode(bool_decoder, partition, partition_size))
- vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate bool decoder %d", i + 1);
-
- /* Advance to the next partition */
- partition += partition_size;
- bool_decoder++;
- }
-
-#if CONFIG_MULTITHREAD
- /* Clamp number of decoder threads */
- if (pbi->decoding_thread_count > num_part - 1)
- pbi->decoding_thread_count = num_part - 1;
-#endif
-}
-
-
-static void stop_token_decoder(VP8D_COMP *pbi)
-{
- VP8_COMMON *pc = &pbi->common;
-
- if (pc->multi_token_partition != ONE_PARTITION)
- vpx_free(pbi->mbc);
-}
-
-static void init_frame(VP8D_COMP *pbi)
-{
- VP8_COMMON *const pc = & pbi->common;
- MACROBLOCKD *const xd = & pbi->mb;
-
- if (pc->frame_type == KEY_FRAME)
- {
- /* Various keyframe initializations */
- vpx_memcpy(pc->fc.mvc, vp8_default_mv_context, sizeof(vp8_default_mv_context));
-
- vp8_init_mbmode_probs(pc);
-
- vp8_default_coef_probs(pc);
- vp8_kf_default_bmode_probs(pc->kf_bmode_prob);
-
- /* reset the segment feature data to 0 with delta coding (Default state). */
- vpx_memset(xd->segment_feature_data, 0, sizeof(xd->segment_feature_data));
- xd->mb_segement_abs_delta = SEGMENT_DELTADATA;
-
- /* reset the mode ref deltasa for loop filter */
- vpx_memset(xd->ref_lf_deltas, 0, sizeof(xd->ref_lf_deltas));
- vpx_memset(xd->mode_lf_deltas, 0, sizeof(xd->mode_lf_deltas));
-
- /* All buffers are implicitly updated on key frames. */
- pc->refresh_golden_frame = 1;
- pc->refresh_alt_ref_frame = 1;
- pc->copy_buffer_to_gf = 0;
- pc->copy_buffer_to_arf = 0;
-
- /* Note that Golden and Altref modes cannot be used on a key frame so
- * ref_frame_sign_bias[] is undefined and meaningless
- */
- pc->ref_frame_sign_bias[GOLDEN_FRAME] = 0;
- pc->ref_frame_sign_bias[ALTREF_FRAME] = 0;
- }
- else
- {
- if (!pc->use_bilinear_mc_filter)
- pc->mcomp_filter_type = SIXTAP;
- else
- pc->mcomp_filter_type = BILINEAR;
-
- /* To enable choice of different interploation filters */
- if (pc->mcomp_filter_type == SIXTAP)
- {
- xd->subpixel_predict = SUBPIX_INVOKE(RTCD_VTABLE(subpix), sixtap4x4);
- xd->subpixel_predict8x4 = SUBPIX_INVOKE(RTCD_VTABLE(subpix), sixtap8x4);
- xd->subpixel_predict8x8 = SUBPIX_INVOKE(RTCD_VTABLE(subpix), sixtap8x8);
- xd->subpixel_predict16x16 = SUBPIX_INVOKE(RTCD_VTABLE(subpix), sixtap16x16);
- }
- else
- {
- xd->subpixel_predict = SUBPIX_INVOKE(RTCD_VTABLE(subpix), bilinear4x4);
- xd->subpixel_predict8x4 = SUBPIX_INVOKE(RTCD_VTABLE(subpix), bilinear8x4);
- xd->subpixel_predict8x8 = SUBPIX_INVOKE(RTCD_VTABLE(subpix), bilinear8x8);
- xd->subpixel_predict16x16 = SUBPIX_INVOKE(RTCD_VTABLE(subpix), bilinear16x16);
- }
- }
-
- xd->left_context = &pc->left_context;
- xd->mode_info_context = pc->mi;
- xd->frame_type = pc->frame_type;
- xd->mode_info_context->mbmi.mode = DC_PRED;
- xd->mode_info_stride = pc->mode_info_stride;
- xd->corrupted = 0; /* init without corruption */
-}
-
-int vp8_decode_frame(VP8D_COMP *pbi)
-{
- vp8_reader *const bc = & pbi->bc;
- VP8_COMMON *const pc = & pbi->common;
- MACROBLOCKD *const xd = & pbi->mb;
- const unsigned char *data = (const unsigned char *)pbi->Source;
- const unsigned char *const data_end = data + pbi->source_sz;
- ptrdiff_t first_partition_length_in_bytes;
-
- int mb_row;
- int i, j, k, l;
- const int *const mb_feature_data_bits = vp8_mb_feature_data_bits;
-
- /* start with no corruption of current frame */
- xd->corrupted = 0;
- pc->yv12_fb[pc->new_fb_idx].corrupted = 0;
-
- if (data_end - data < 3)
- vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
- "Truncated packet");
- pc->frame_type = (FRAME_TYPE)(data[0] & 1);
- pc->version = (data[0] >> 1) & 7;
- pc->show_frame = (data[0] >> 4) & 1;
- first_partition_length_in_bytes =
- (data[0] | (data[1] << 8) | (data[2] << 16)) >> 5;
- data += 3;
-
- if (data + first_partition_length_in_bytes > data_end
- || data + first_partition_length_in_bytes < data)
- vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
- "Truncated packet or corrupt partition 0 length");
- vp8_setup_version(pc);
-
- if (pc->frame_type == KEY_FRAME)
- {
- const int Width = pc->Width;
- const int Height = pc->Height;
-
- /* vet via sync code */
- if (data[0] != 0x9d || data[1] != 0x01 || data[2] != 0x2a)
- vpx_internal_error(&pc->error, VPX_CODEC_UNSUP_BITSTREAM,
- "Invalid frame sync code");
-
- pc->Width = (data[3] | (data[4] << 8)) & 0x3fff;
- pc->horiz_scale = data[4] >> 6;
- pc->Height = (data[5] | (data[6] << 8)) & 0x3fff;
- pc->vert_scale = data[6] >> 6;
- data += 7;
-
- if (Width != pc->Width || Height != pc->Height)
- {
- int prev_mb_rows = pc->mb_rows;
-
- if (pc->Width <= 0)
- {
- pc->Width = Width;
- vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
- "Invalid frame width");
- }
-
- if (pc->Height <= 0)
- {
- pc->Height = Height;
- vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
- "Invalid frame height");
- }
-
- if (vp8_alloc_frame_buffers(pc, pc->Width, pc->Height))
- vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate frame buffers");
-
-#if CONFIG_MULTITHREAD
- if (pbi->b_multithreaded_rd)
- vp8mt_alloc_temp_buffers(pbi, pc->Width, prev_mb_rows);
-#endif
- }
- }
-
- if (pc->Width == 0 || pc->Height == 0)
- {
- return -1;
- }
-
- init_frame(pbi);
-
- if (vp8dx_start_decode(bc, data, data_end - data))
- vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate bool decoder 0");
- if (pc->frame_type == KEY_FRAME) {
- pc->clr_type = (YUV_TYPE)vp8_read_bit(bc);
- pc->clamp_type = (CLAMP_TYPE)vp8_read_bit(bc);
- }
-
- /* Is segmentation enabled */
- xd->segmentation_enabled = (unsigned char)vp8_read_bit(bc);
-
- if (xd->segmentation_enabled)
- {
- /* Signal whether or not the segmentation map is being explicitly updated this frame. */
- xd->update_mb_segmentation_map = (unsigned char)vp8_read_bit(bc);
- xd->update_mb_segmentation_data = (unsigned char)vp8_read_bit(bc);
-
- if (xd->update_mb_segmentation_data)
- {
- xd->mb_segement_abs_delta = (unsigned char)vp8_read_bit(bc);
-
- vpx_memset(xd->segment_feature_data, 0, sizeof(xd->segment_feature_data));
-
- /* For each segmentation feature (Quant and loop filter level) */
- for (i = 0; i < MB_LVL_MAX; i++)
- {
- for (j = 0; j < MAX_MB_SEGMENTS; j++)
- {
- /* Frame level data */
- if (vp8_read_bit(bc))
- {
- xd->segment_feature_data[i][j] = (signed char)vp8_read_literal(bc, mb_feature_data_bits[i]);
-
- if (vp8_read_bit(bc))
- xd->segment_feature_data[i][j] = -xd->segment_feature_data[i][j];
- }
- else
- xd->segment_feature_data[i][j] = 0;
- }
- }
- }
-
- if (xd->update_mb_segmentation_map)
- {
- /* Which macro block level features are enabled */
- vpx_memset(xd->mb_segment_tree_probs, 255, sizeof(xd->mb_segment_tree_probs));
-
- /* Read the probs used to decode the segment id for each macro block. */
- for (i = 0; i < MB_FEATURE_TREE_PROBS; i++)
- {
- /* If not explicitly set value is defaulted to 255 by memset above */
- if (vp8_read_bit(bc))
- xd->mb_segment_tree_probs[i] = (vp8_prob)vp8_read_literal(bc, 8);
- }
- }
- }
-
- /* Read the loop filter level and type */
- pc->filter_type = (LOOPFILTERTYPE) vp8_read_bit(bc);
- pc->filter_level = vp8_read_literal(bc, 6);
- pc->sharpness_level = vp8_read_literal(bc, 3);
-
- /* Read in loop filter deltas applied at the MB level based on mode or ref frame. */
- xd->mode_ref_lf_delta_update = 0;
- xd->mode_ref_lf_delta_enabled = (unsigned char)vp8_read_bit(bc);
-
- if (xd->mode_ref_lf_delta_enabled)
- {
- /* Do the deltas need to be updated */
- xd->mode_ref_lf_delta_update = (unsigned char)vp8_read_bit(bc);
-
- if (xd->mode_ref_lf_delta_update)
- {
- /* Send update */
- for (i = 0; i < MAX_REF_LF_DELTAS; i++)
- {
- if (vp8_read_bit(bc))
- {
- /*sign = vp8_read_bit( bc );*/
- xd->ref_lf_deltas[i] = (signed char)vp8_read_literal(bc, 6);
-
- if (vp8_read_bit(bc)) /* Apply sign */
- xd->ref_lf_deltas[i] = xd->ref_lf_deltas[i] * -1;
- }
- }
-
- /* Send update */
- for (i = 0; i < MAX_MODE_LF_DELTAS; i++)
- {
- if (vp8_read_bit(bc))
- {
- /*sign = vp8_read_bit( bc );*/
- xd->mode_lf_deltas[i] = (signed char)vp8_read_literal(bc, 6);
-
- if (vp8_read_bit(bc)) /* Apply sign */
- xd->mode_lf_deltas[i] = xd->mode_lf_deltas[i] * -1;
- }
- }
- }
- }
-
- setup_token_decoder(pbi, data + first_partition_length_in_bytes);
- xd->current_bc = &pbi->bc2;
-
- /* Read the default quantizers. */
- {
- int Q, q_update;
-
- Q = vp8_read_literal(bc, 7); /* AC 1st order Q = default */
- pc->base_qindex = Q;
- q_update = 0;
- pc->y1dc_delta_q = get_delta_q(bc, pc->y1dc_delta_q, &q_update);
- pc->y2dc_delta_q = get_delta_q(bc, pc->y2dc_delta_q, &q_update);
- pc->y2ac_delta_q = get_delta_q(bc, pc->y2ac_delta_q, &q_update);
- pc->uvdc_delta_q = get_delta_q(bc, pc->uvdc_delta_q, &q_update);
- pc->uvac_delta_q = get_delta_q(bc, pc->uvac_delta_q, &q_update);
-
- if (q_update)
- vp8cx_init_de_quantizer(pbi);
-
- /* MB level dequantizer setup */
- mb_init_dequantizer(pbi, &pbi->mb);
- }
-
- /* Determine if the golden frame or ARF buffer should be updated and how.
- * For all non key frames the GF and ARF refresh flags and sign bias
- * flags must be set explicitly.
- */
- if (pc->frame_type != KEY_FRAME)
- {
- /* Should the GF or ARF be updated from the current frame */
- pc->refresh_golden_frame = vp8_read_bit(bc);
- pc->refresh_alt_ref_frame = vp8_read_bit(bc);
-
- /* Buffer to buffer copy flags. */
- pc->copy_buffer_to_gf = 0;
-
- if (!pc->refresh_golden_frame)
- pc->copy_buffer_to_gf = vp8_read_literal(bc, 2);
-
- pc->copy_buffer_to_arf = 0;
-
- if (!pc->refresh_alt_ref_frame)
- pc->copy_buffer_to_arf = vp8_read_literal(bc, 2);
-
- pc->ref_frame_sign_bias[GOLDEN_FRAME] = vp8_read_bit(bc);
- pc->ref_frame_sign_bias[ALTREF_FRAME] = vp8_read_bit(bc);
- }
-
- pc->refresh_entropy_probs = vp8_read_bit(bc);
- if (pc->refresh_entropy_probs == 0)
- {
- vpx_memcpy(&pc->lfc, &pc->fc, sizeof(pc->fc));
- }
-
- pc->refresh_last_frame = pc->frame_type == KEY_FRAME || vp8_read_bit(bc);
-
- if (0)
- {
- FILE *z = fopen("decodestats.stt", "a");
- fprintf(z, "%6d F:%d,G:%d,A:%d,L:%d,Q:%d\n",
- pc->current_video_frame,
- pc->frame_type,
- pc->refresh_golden_frame,
- pc->refresh_alt_ref_frame,
- pc->refresh_last_frame,
- pc->base_qindex);
- fclose(z);
- }
-
-
- {
- /* read coef probability tree */
-
- for (i = 0; i < BLOCK_TYPES; i++)
- for (j = 0; j < COEF_BANDS; j++)
- for (k = 0; k < PREV_COEF_CONTEXTS; k++)
- for (l = 0; l < MAX_ENTROPY_TOKENS - 1; l++)
- {
-
- vp8_prob *const p = pc->fc.coef_probs [i][j][k] + l;
-
- if (vp8_read(bc, vp8_coef_update_probs [i][j][k][l]))
- {
- *p = (vp8_prob)vp8_read_literal(bc, 8);
-
- }
- }
- }
-
- vpx_memcpy(&xd->pre, &pc->yv12_fb[pc->lst_fb_idx], sizeof(YV12_BUFFER_CONFIG));
- vpx_memcpy(&xd->dst, &pc->yv12_fb[pc->new_fb_idx], sizeof(YV12_BUFFER_CONFIG));
-
- /* set up frame new frame for intra coded blocks */
-#if CONFIG_MULTITHREAD
- if (!(pbi->b_multithreaded_rd) || pc->multi_token_partition == ONE_PARTITION || !(pc->filter_level))
-#endif
- vp8_setup_intra_recon(&pc->yv12_fb[pc->new_fb_idx]);
-
- vp8_setup_block_dptrs(xd);
-
- vp8_build_block_doffsets(xd);
-
- /* clear out the coeff buffer */
- vpx_memset(xd->qcoeff, 0, sizeof(xd->qcoeff));
-
- /* Read the mb_no_coeff_skip flag */
- pc->mb_no_coeff_skip = (int)vp8_read_bit(bc);
-
-
- vp8_decode_mode_mvs(pbi);
-
- vpx_memset(pc->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES) * pc->mb_cols);
-
- vpx_memcpy(&xd->block[0].bmi, &xd->mode_info_context->bmi[0], sizeof(B_MODE_INFO));
-
-#if CONFIG_MULTITHREAD
- if (pbi->b_multithreaded_rd && pc->multi_token_partition != ONE_PARTITION)
- {
- vp8mt_decode_mb_rows(pbi, xd);
- if(pbi->common.filter_level)
- {
- /*vp8_mt_loop_filter_frame(pbi);*/ /*cm, &pbi->mb, cm->filter_level);*/
-
- pc->last_frame_type = pc->frame_type;
- pc->last_filter_type = pc->filter_type;
- pc->last_sharpness_level = pc->sharpness_level;
- }
- vp8_yv12_extend_frame_borders_ptr(&pc->yv12_fb[pc->new_fb_idx]); /*cm->frame_to_show);*/
- }
- else
-#endif
- {
- int ibc = 0;
- int num_part = 1 << pc->multi_token_partition;
-
- /* Decode the individual macro block */
- for (mb_row = 0; mb_row < pc->mb_rows; mb_row++)
- {
-
- if (num_part > 1)
- {
- xd->current_bc = & pbi->mbc[ibc];
- ibc++;
-
- if (ibc == num_part)
- ibc = 0;
- }
-
- decode_mb_row(pbi, pc, mb_row, xd);
- }
- }
-
-
- stop_token_decoder(pbi);
-
- /* Collect information about decoder corruption. */
- /* 1. Check first boolean decoder for errors. */
- pc->yv12_fb[pc->new_fb_idx].corrupted =
- vp8dx_bool_error(bc);
- /* 2. Check the macroblock information */
- pc->yv12_fb[pc->new_fb_idx].corrupted |=
- xd->corrupted;
-
- /* vpx_log("Decoder: Frame Decoded, Size Roughly:%d bytes \n",bc->pos+pbi->bc2.pos); */
-
- /* If this was a kf or Gf note the Q used */
- if ((pc->frame_type == KEY_FRAME) ||
- pc->refresh_golden_frame || pc->refresh_alt_ref_frame)
- {
- pc->last_kf_gf_q = pc->base_qindex;
- }
-
- if (pc->refresh_entropy_probs == 0)
- {
- vpx_memcpy(&pc->fc, &pc->lfc, sizeof(pc->fc));
- }
-
-#ifdef PACKET_TESTING
- {
- FILE *f = fopen("decompressor.VP8", "ab");
- unsigned int size = pbi->bc2.pos + pbi->bc.pos + 8;
- fwrite((void *) &size, 4, 1, f);
- fwrite((void *) pbi->Source, size, 1, f);
- fclose(f);
- }
-#endif
-
- return 0;
-}
diff --git a/vp8/decoder/dequantize.c b/vp8/decoder/dequantize.c
deleted file mode 100644
index dd0c13b..0000000
--- a/vp8/decoder/dequantize.c
+++ /dev/null
@@ -1,113 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "dequantize.h"
-#include "vp8/common/idct.h"
-#include "vpx_mem/vpx_mem.h"
-
-extern void vp8_short_idct4x4llm_c(short *input, short *output, int pitch) ;
-extern void vp8_short_idct4x4llm_1_c(short *input, short *output, int pitch);
-
-
-void vp8_dequantize_b_c(BLOCKD *d)
-{
- int i;
- short *DQ = d->dqcoeff;
- short *Q = d->qcoeff;
- short *DQC = d->dequant;
-
- for (i = 0; i < 16; i++)
- {
- DQ[i] = Q[i] * DQC[i];
- }
-}
-
-void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *pred,
- unsigned char *dest, int pitch, int stride)
-{
- short output[16];
- short *diff_ptr = output;
- int r, c;
- int i;
-
- for (i = 0; i < 16; i++)
- {
- input[i] = dq[i] * input[i];
- }
-
- /* the idct halves ( >> 1) the pitch */
- vp8_short_idct4x4llm_c(input, output, 4 << 1);
-
- vpx_memset(input, 0, 32);
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- int a = diff_ptr[c] + pred[c];
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dest[c] = (unsigned char) a;
- }
-
- dest += stride;
- diff_ptr += 4;
- pred += pitch;
- }
-}
-
-void vp8_dequant_dc_idct_add_c(short *input, short *dq, unsigned char *pred,
- unsigned char *dest, int pitch, int stride,
- int Dc)
-{
- int i;
- short output[16];
- short *diff_ptr = output;
- int r, c;
-
- input[0] = (short)Dc;
-
- for (i = 1; i < 16; i++)
- {
- input[i] = dq[i] * input[i];
- }
-
- /* the idct halves ( >> 1) the pitch */
- vp8_short_idct4x4llm_c(input, output, 4 << 1);
-
- vpx_memset(input, 0, 32);
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- int a = diff_ptr[c] + pred[c];
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dest[c] = (unsigned char) a;
- }
-
- dest += stride;
- diff_ptr += 4;
- pred += pitch;
- }
-}
diff --git a/vp8/decoder/dequantize.h b/vp8/decoder/dequantize.h
deleted file mode 100644
index b3a78c5..0000000
--- a/vp8/decoder/dequantize.h
+++ /dev/null
@@ -1,118 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef DEQUANTIZE_H
-#define DEQUANTIZE_H
-#include "vp8/common/blockd.h"
-
-#define prototype_dequant_block(sym) \
- void sym(BLOCKD *x)
-
-#define prototype_dequant_idct_add(sym) \
- void sym(short *input, short *dq, \
- unsigned char *pred, unsigned char *output, \
- int pitch, int stride)
-
-#define prototype_dequant_dc_idct_add(sym) \
- void sym(short *input, short *dq, \
- unsigned char *pred, unsigned char *output, \
- int pitch, int stride, \
- int dc)
-
-#define prototype_dequant_dc_idct_add_y_block(sym) \
- void sym(short *q, short *dq, \
- unsigned char *pre, unsigned char *dst, \
- int stride, char *eobs, short *dc)
-
-#define prototype_dequant_idct_add_y_block(sym) \
- void sym(short *q, short *dq, \
- unsigned char *pre, unsigned char *dst, \
- int stride, char *eobs)
-
-#define prototype_dequant_idct_add_uv_block(sym) \
- void sym(short *q, short *dq, \
- unsigned char *pre, unsigned char *dst_u, \
- unsigned char *dst_v, int stride, char *eobs)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/dequantize_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/dequantize_arm.h"
-#endif
-
-#if ARCH_MIPS
-#if defined(MIPS_DSP_REV) && MIPS_DSP_REV>=2
-#include "mips/dequantize_mips.h"
-#endif
-#endif
-
-#ifndef vp8_dequant_block
-#define vp8_dequant_block vp8_dequantize_b_c
-#endif
-extern prototype_dequant_block(vp8_dequant_block);
-
-#ifndef vp8_dequant_idct_add
-#define vp8_dequant_idct_add vp8_dequant_idct_add_c
-#endif
-extern prototype_dequant_idct_add(vp8_dequant_idct_add);
-
-#ifndef vp8_dequant_dc_idct_add
-#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_c
-#endif
-extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add);
-
-#ifndef vp8_dequant_dc_idct_add_y_block
-#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_c
-#endif
-extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block);
-
-#ifndef vp8_dequant_idct_add_y_block
-#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_c
-#endif
-extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block);
-
-#ifndef vp8_dequant_idct_add_uv_block
-#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_c
-#endif
-extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block);
-
-
-typedef prototype_dequant_block((*vp8_dequant_block_fn_t));
-
-typedef prototype_dequant_idct_add((*vp8_dequant_idct_add_fn_t));
-
-typedef prototype_dequant_dc_idct_add((*vp8_dequant_dc_idct_add_fn_t));
-
-typedef prototype_dequant_dc_idct_add_y_block((*vp8_dequant_dc_idct_add_y_block_fn_t));
-
-typedef prototype_dequant_idct_add_y_block((*vp8_dequant_idct_add_y_block_fn_t));
-
-typedef prototype_dequant_idct_add_uv_block((*vp8_dequant_idct_add_uv_block_fn_t));
-
-typedef struct
-{
- vp8_dequant_block_fn_t block;
- vp8_dequant_idct_add_fn_t idct_add;
- vp8_dequant_dc_idct_add_fn_t dc_idct_add;
- vp8_dequant_dc_idct_add_y_block_fn_t dc_idct_add_y_block;
- vp8_dequant_idct_add_y_block_fn_t idct_add_y_block;
- vp8_dequant_idct_add_uv_block_fn_t idct_add_uv_block;
-} vp8_dequant_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define DEQUANT_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define DEQUANT_INVOKE(ctx,fn) vp8_dequant_##fn
-#endif
-
-#endif
diff --git a/vp8/decoder/detokenize.c b/vp8/decoder/detokenize.c
deleted file mode 100644
index b2e0819..0000000
--- a/vp8/decoder/detokenize.c
+++ /dev/null
@@ -1,371 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/type_aliases.h"
-#include "vp8/common/blockd.h"
-#include "onyxd_int.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vpx_ports/mem.h"
-#include "detokenize.h"
-
-#define BOOL_DATA UINT8
-
-#define OCB_X PREV_COEF_CONTEXTS * ENTROPY_NODES
-DECLARE_ALIGNED(16, static const unsigned char, coef_bands_x[16]) =
-{
- 0 * OCB_X, 1 * OCB_X, 2 * OCB_X, 3 * OCB_X,
- 6 * OCB_X, 4 * OCB_X, 5 * OCB_X, 6 * OCB_X,
- 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X,
- 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 7 * OCB_X
-};
-#define EOB_CONTEXT_NODE 0
-#define ZERO_CONTEXT_NODE 1
-#define ONE_CONTEXT_NODE 2
-#define LOW_VAL_CONTEXT_NODE 3
-#define TWO_CONTEXT_NODE 4
-#define THREE_CONTEXT_NODE 5
-#define HIGH_LOW_CONTEXT_NODE 6
-#define CAT_ONE_CONTEXT_NODE 7
-#define CAT_THREEFOUR_CONTEXT_NODE 8
-#define CAT_THREE_CONTEXT_NODE 9
-#define CAT_FIVE_CONTEXT_NODE 10
-
-/*
-//the definition is put in "onyxd_int.h"
-typedef struct
-{
- INT16 min_val;
- INT16 Length;
- UINT8 Probs[12];
-} TOKENEXTRABITS;
-*/
-
-DECLARE_ALIGNED(16, static const TOKENEXTRABITS, vp8d_token_extra_bits2[MAX_ENTROPY_TOKENS]) =
-{
- { 0, -1, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* ZERO_TOKEN */
- { 1, 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* ONE_TOKEN */
- { 2, 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* TWO_TOKEN */
- { 3, 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* THREE_TOKEN */
- { 4, 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* FOUR_TOKEN */
- { 5, 0, { 159, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* DCT_VAL_CATEGORY1 */
- { 7, 1, { 145, 165, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* DCT_VAL_CATEGORY2 */
- { 11, 2, { 140, 148, 173, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* DCT_VAL_CATEGORY3 */
- { 19, 3, { 135, 140, 155, 176, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* DCT_VAL_CATEGORY4 */
- { 35, 4, { 130, 134, 141, 157, 180, 0, 0, 0, 0, 0, 0, 0 } }, /* DCT_VAL_CATEGORY5 */
- { 67, 10, { 129, 130, 133, 140, 153, 177, 196, 230, 243, 254, 254, 0 } }, /* DCT_VAL_CATEGORY6 */
- { 0, -1, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, /* EOB TOKEN */
-};
-
-
-void vp8_reset_mb_tokens_context(MACROBLOCKD *x)
-{
- /* Clear entropy contexts for Y2 blocks */
- if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV)
- {
- vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES));
- }
- else
- {
- vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1);
- vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1);
- }
-}
-
-DECLARE_ALIGNED(16, extern const unsigned char, vp8dx_bitreader_norm[256]);
-#define FILL \
- if(count < 0) \
- VP8DX_BOOL_DECODER_FILL(count, value, bufptr, bufend);
-
-#define NORMALIZE \
- /*if(range < 0x80)*/ \
- { \
- shift = vp8dx_bitreader_norm[range]; \
- range <<= shift; \
- value <<= shift; \
- count -= shift; \
- }
-
-#define DECODE_AND_APPLYSIGN(value_to_sign) \
- split = (range + 1) >> 1; \
- bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); \
- FILL \
- if ( value < bigsplit ) \
- { \
- range = split; \
- v= value_to_sign; \
- } \
- else \
- { \
- range = range-split; \
- value = value-bigsplit; \
- v = -value_to_sign; \
- } \
- range +=range; \
- value +=value; \
- count--;
-
-#define DECODE_AND_BRANCH_IF_ZERO(probability,branch) \
- { \
- split = 1 + ((( probability*(range-1) ) )>> 8); \
- bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); \
- FILL \
- if ( value < bigsplit ) \
- { \
- range = split; \
- NORMALIZE \
- goto branch; \
- } \
- value -= bigsplit; \
- range = range - split; \
- NORMALIZE \
- }
-
-#define DECODE_AND_LOOP_IF_ZERO(probability,branch) \
- { \
- split = 1 + ((( probability*(range-1) ) ) >> 8); \
- bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); \
- FILL \
- if ( value < bigsplit ) \
- { \
- range = split; \
- NORMALIZE \
- Prob = coef_probs; \
- if(c<15) {\
- ++c; \
- Prob += coef_bands_x[c]; \
- goto branch; \
- } goto BLOCK_FINISHED; /*for malformed input */\
- } \
- value -= bigsplit; \
- range = range - split; \
- NORMALIZE \
- }
-
-#define DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val) \
- DECODE_AND_APPLYSIGN(val) \
- Prob = coef_probs + (ENTROPY_NODES*2); \
- if(c < 15){\
- qcoeff_ptr [ scan[c] ] = (INT16) v; \
- ++c; \
- goto DO_WHILE; }\
- qcoeff_ptr [ scan[15] ] = (INT16) v; \
- goto BLOCK_FINISHED;
-
-
-#define DECODE_EXTRABIT_AND_ADJUST_VAL(t,bits_count)\
- split = 1 + (((range-1) * vp8d_token_extra_bits2[t].Probs[bits_count]) >> 8); \
- bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); \
- FILL \
- if(value >= bigsplit)\
- {\
- range = range-split;\
- value = value-bigsplit;\
- val += ((UINT16)1<<bits_count);\
- }\
- else\
- {\
- range = split;\
- }\
- NORMALIZE
-
-int vp8_decode_mb_tokens(VP8D_COMP *dx, MACROBLOCKD *x)
-{
- ENTROPY_CONTEXT *A = (ENTROPY_CONTEXT *)x->above_context;
- ENTROPY_CONTEXT *L = (ENTROPY_CONTEXT *)x->left_context;
- const VP8_COMMON *const oc = & dx->common;
-
- BOOL_DECODER *bc = x->current_bc;
-
- char *eobs = x->eobs;
-
- ENTROPY_CONTEXT *a;
- ENTROPY_CONTEXT *l;
- int i;
-
- int eobtotal = 0;
-
- register int count;
-
- const BOOL_DATA *bufptr;
- const BOOL_DATA *bufend;
- register unsigned int range;
- VP8_BD_VALUE value;
- const int *scan;
- register unsigned int shift;
- UINT32 split;
- VP8_BD_VALUE bigsplit;
- INT16 *qcoeff_ptr;
-
- const vp8_prob *coef_probs;
- int type;
- int stop;
- INT16 val, bits_count;
- INT16 c;
- INT16 v;
- const vp8_prob *Prob;
-
- type = 3;
- i = 0;
- stop = 16;
-
- scan = vp8_default_zig_zag1d;
- qcoeff_ptr = &x->qcoeff[0];
-
- if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV)
- {
- i = 24;
- stop = 24;
- type = 1;
- qcoeff_ptr += 24*16;
- eobtotal -= 16;
- }
-
- bufend = bc->user_buffer_end;
- bufptr = bc->user_buffer;
- value = bc->value;
- count = bc->count;
- range = bc->range;
-
-
- coef_probs = oc->fc.coef_probs [type] [ 0 ] [0];
-
-BLOCK_LOOP:
- a = A + vp8_block2above[i];
- l = L + vp8_block2left[i];
-
- c = (INT16)(!type);
-
- /*Dest = ((A)!=0) + ((B)!=0);*/
- VP8_COMBINEENTROPYCONTEXTS(v, *a, *l);
- Prob = coef_probs;
- Prob += v * ENTROPY_NODES;
-
-DO_WHILE:
- Prob += coef_bands_x[c];
- DECODE_AND_BRANCH_IF_ZERO(Prob[EOB_CONTEXT_NODE], BLOCK_FINISHED);
-
-CHECK_0_:
- DECODE_AND_LOOP_IF_ZERO(Prob[ZERO_CONTEXT_NODE], CHECK_0_);
- DECODE_AND_BRANCH_IF_ZERO(Prob[ONE_CONTEXT_NODE], ONE_CONTEXT_NODE_0_);
- DECODE_AND_BRANCH_IF_ZERO(Prob[LOW_VAL_CONTEXT_NODE], LOW_VAL_CONTEXT_NODE_0_);
- DECODE_AND_BRANCH_IF_ZERO(Prob[HIGH_LOW_CONTEXT_NODE], HIGH_LOW_CONTEXT_NODE_0_);
- DECODE_AND_BRANCH_IF_ZERO(Prob[CAT_THREEFOUR_CONTEXT_NODE], CAT_THREEFOUR_CONTEXT_NODE_0_);
- DECODE_AND_BRANCH_IF_ZERO(Prob[CAT_FIVE_CONTEXT_NODE], CAT_FIVE_CONTEXT_NODE_0_);
- val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY6].min_val;
- bits_count = vp8d_token_extra_bits2[DCT_VAL_CATEGORY6].Length;
-
- do
- {
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY6, bits_count);
- bits_count -- ;
- }
- while (bits_count >= 0);
-
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val);
-
-CAT_FIVE_CONTEXT_NODE_0_:
- val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY5].min_val;
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 4);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 3);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 2);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 1);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 0);
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val);
-
-CAT_THREEFOUR_CONTEXT_NODE_0_:
- DECODE_AND_BRANCH_IF_ZERO(Prob[CAT_THREE_CONTEXT_NODE], CAT_THREE_CONTEXT_NODE_0_);
- val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY4].min_val;
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY4, 3);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY4, 2);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY4, 1);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY4, 0);
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val);
-
-CAT_THREE_CONTEXT_NODE_0_:
- val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY3].min_val;
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY3, 2);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY3, 1);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY3, 0);
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val);
-
-HIGH_LOW_CONTEXT_NODE_0_:
- DECODE_AND_BRANCH_IF_ZERO(Prob[CAT_ONE_CONTEXT_NODE], CAT_ONE_CONTEXT_NODE_0_);
-
- val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY2].min_val;
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY2, 1);
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY2, 0);
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val);
-
-CAT_ONE_CONTEXT_NODE_0_:
- val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY1].min_val;
- DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY1, 0);
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val);
-
-LOW_VAL_CONTEXT_NODE_0_:
- DECODE_AND_BRANCH_IF_ZERO(Prob[TWO_CONTEXT_NODE], TWO_CONTEXT_NODE_0_);
- DECODE_AND_BRANCH_IF_ZERO(Prob[THREE_CONTEXT_NODE], THREE_CONTEXT_NODE_0_);
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(4);
-
-THREE_CONTEXT_NODE_0_:
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(3);
-
-TWO_CONTEXT_NODE_0_:
- DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(2);
-
-ONE_CONTEXT_NODE_0_:
- DECODE_AND_APPLYSIGN(1);
- Prob = coef_probs + ENTROPY_NODES;
-
- if (c < 15)
- {
- qcoeff_ptr [ scan[c] ] = (INT16) v;
- ++c;
- goto DO_WHILE;
- }
-
- qcoeff_ptr [ scan[15] ] = (INT16) v;
-BLOCK_FINISHED:
- *a = *l = ((eobs[i] = c) != !type); /* any nonzero data? */
- eobtotal += c;
- qcoeff_ptr += 16;
-
- i++;
-
- if (i < stop)
- goto BLOCK_LOOP;
-
- if (i == 25)
- {
- type = 0;
- i = 0;
- stop = 16;
- coef_probs = oc->fc.coef_probs [type] [ 0 ] [0];
- qcoeff_ptr -= (24*16 + 16);
- goto BLOCK_LOOP;
- }
-
- if (i == 16)
- {
- type = 2;
- coef_probs = oc->fc.coef_probs [type] [ 0 ] [0];
- stop = 24;
- goto BLOCK_LOOP;
- }
-
- FILL
- bc->user_buffer = bufptr;
- bc->value = value;
- bc->count = count;
- bc->range = range;
- return eobtotal;
-
-}
diff --git a/vp8/decoder/generic/dsystemdependent.c b/vp8/decoder/generic/dsystemdependent.c
deleted file mode 100644
index 2406dea..0000000
--- a/vp8/decoder/generic/dsystemdependent.c
+++ /dev/null
@@ -1,39 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/decoder/dequantize.h"
-#include "vp8/decoder/onyxd_int.h"
-
-extern void vp8_arch_x86_decode_init(VP8D_COMP *pbi);
-extern void vp8_arch_arm_decode_init(VP8D_COMP *pbi);
-
-void vp8_dmachine_specific_config(VP8D_COMP *pbi)
-{
- /* Pure C: */
-#if CONFIG_RUNTIME_CPU_DETECT
- pbi->mb.rtcd = &pbi->common.rtcd;
- pbi->dequant.block = vp8_dequantize_b_c;
- pbi->dequant.idct_add = vp8_dequant_idct_add_c;
- pbi->dequant.dc_idct_add = vp8_dequant_dc_idct_add_c;
- pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_c;
- pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_c;
- pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_c;
-#endif
-
-#if ARCH_X86 || ARCH_X86_64
- vp8_arch_x86_decode_init(pbi);
-#endif
-
-#if ARCH_ARM
- vp8_arch_arm_decode_init(pbi);
-#endif
-}
diff --git a/vp8/decoder/idct_blk.c b/vp8/decoder/idct_blk.c
deleted file mode 100644
index df01923..0000000
--- a/vp8/decoder/idct_blk.c
+++ /dev/null
@@ -1,124 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_ports/config.h"
-#include "vp8/common/idct.h"
-#include "dequantize.h"
-
-void vp8_dequant_dc_idct_add_c(short *input, short *dq, unsigned char *pred,
- unsigned char *dest, int pitch, int stride,
- int Dc);
-void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *pred,
- unsigned char *dest, int pitch, int stride);
-void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr,
- unsigned char *dst_ptr, int pitch, int stride);
-
-void vp8_dequant_dc_idct_add_y_block_c
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs, short *dc)
-{
- int i, j;
-
- for (i = 0; i < 4; i++)
- {
- for (j = 0; j < 4; j++)
- {
- if (*eobs++ > 1)
- vp8_dequant_dc_idct_add_c (q, dq, pre, dst, 16, stride, dc[0]);
- else
- vp8_dc_only_idct_add_c (dc[0], pre, dst, 16, stride);
-
- q += 16;
- pre += 4;
- dst += 4;
- dc ++;
- }
-
- pre += 64 - 16;
- dst += 4*stride - 16;
- }
-}
-
-void vp8_dequant_idct_add_y_block_c
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs)
-{
- int i, j;
-
- for (i = 0; i < 4; i++)
- {
- for (j = 0; j < 4; j++)
- {
- if (*eobs++ > 1)
- vp8_dequant_idct_add_c (q, dq, pre, dst, 16, stride);
- else
- {
- vp8_dc_only_idct_add_c (q[0]*dq[0], pre, dst, 16, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dst += 4;
- }
-
- pre += 64 - 16;
- dst += 4*stride - 16;
- }
-}
-
-void vp8_dequant_idct_add_uv_block_c
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
-{
- int i, j;
-
- for (i = 0; i < 2; i++)
- {
- for (j = 0; j < 2; j++)
- {
- if (*eobs++ > 1)
- vp8_dequant_idct_add_c (q, dq, pre, dstu, 8, stride);
- else
- {
- vp8_dc_only_idct_add_c (q[0]*dq[0], pre, dstu, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dstu += 4;
- }
-
- pre += 32 - 8;
- dstu += 4*stride - 8;
- }
-
- for (i = 0; i < 2; i++)
- {
- for (j = 0; j < 2; j++)
- {
- if (*eobs++ > 1)
- vp8_dequant_idct_add_c (q, dq, pre, dstv, 8, stride);
- else
- {
- vp8_dc_only_idct_add_c (q[0]*dq[0], pre, dstv, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dstv += 4;
- }
-
- pre += 32 - 8;
- dstv += 4*stride - 8;
- }
-}
diff --git a/vp8/decoder/mips/dequantize_mips.c b/vp8/decoder/mips/dequantize_mips.c
deleted file mode 100755
index 49bf224..0000000
--- a/vp8/decoder/mips/dequantize_mips.c
+++ /dev/null
@@ -1,168 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "dequantize.h"
-#include "idct.h"
-#include "vpx_mem/vpx_mem.h"
-
-DECLARE_ALIGNED(8, const unsigned char, cma[512]) = {
-0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
-0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
-0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
-0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
-0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11,
-12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32,
-33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53,
-54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74,
-75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95,
-96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113,
-114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130,
-131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147,
-148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164,
-165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181,
-182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198,
-199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215,
-216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232,
-233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249,
-250, 251, 252, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
-255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
-255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
-255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
-255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
-255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
-255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
-255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255
-};
-
-extern void vp8_short_idct4x4llm_mips(short *input, short *output, int pitch) ;
-extern void vp8_short_idct4x4llm_1_mips(short *input, short *output, int pitch);
-
-
-void vp8_dequant_idct_add_mips(short *input, short *dq, unsigned char *pred,
- unsigned char *dest, int pitch, int stride)
-{
- short output[16];
- short *diff_ptr = output;
- int i;
- short a1, a2, a3, a0;
-
- short input_temp[16];
- unsigned int in1, dq1, x1, in2, dq2;
- const unsigned char *cm = &(cma[128]);
-
- __asm__ __volatile__ (
- "lh %[in1], 0(%[input]) \n\t"
- "lh %[dq1], 0(%[dq]) \n\t"
- "lh %[in2], 2(%[input]) \n\t"
- "lh %[dq2], 2(%[dq]) \n\t"
- "append %[in1], %[in2], 16 \n\t"
- "append %[dq1], %[dq2], 16 \n\t"
- "mul.ph %[x1], %[dq1], %[in1] \n\t"
- "sh %[x1], 2(%[input_temp]) \n\t"
- "srl %[x1], %[x1], 16 \n\t"
- "sh %[x1], 0(%[input_temp]) \n\t"
-
- : [x1] "=&r" (x1), [in1] "=&r" (in1), [dq1] "=&r" (dq1),
- [in2] "=&r" (in2), [dq2] "=&r" (dq2)
- : [dq] "r" (dq), [input] "r" (input),
- [input_temp] "r" (input_temp)
- );
-
- for (i = 2; i < 16; i++)
- {
- input_temp[i] = dq[i] * input[i];
- }
-
- /* the idct halves ( >> 1) the pitch */
- vp8_short_idct4x4llm_mips(input_temp, output, 4);
-
- vpx_memset(input, 0, 32);
-
- /* unroll the loop */
- for (i = 4; i--; )
- {
- a0 = diff_ptr[0] + pred[0];
- a1 = diff_ptr[1] + pred[1];
- a2 = diff_ptr[2] + pred[2];
- a3 = diff_ptr[3] + pred[3];
-
- dest[0] = cm[a0];
- dest[1] = cm[a1];
- dest[2] = cm[a2];
- dest[3] = cm[a3];
-
- dest += stride;
- diff_ptr += 4;
- pred += pitch;
- }
-}
-
-
-void vp8_dequant_dc_idct_add_mips(short *input, short *dq, unsigned char *pred,
- unsigned char *dest, int pitch, int stride,
- int Dc)
-{
- int i;
- short output[16];
- short *diff_ptr = output;
- short input_temp[16];
- short a1, a2, a3, a0;
-
- unsigned int in1, dq1, x1, in2, dq2;
- const unsigned char *cm = &(cma[128]);
-
- input_temp[0] = (short)Dc;
-
- __asm__ __volatile__ (
- "lh %[in1], 2(%[input]) \n\t"
- "lh %[dq1], 2(%[dq]) \n\t"
- "lh %[in2], 4(%[input]) \n\t"
- "lh %[dq2], 4(%[dq]) \n\t"
- "append %[in1], %[in2], 16 \n\t"
- "append %[dq1], %[dq2], 16 \n\t"
- "mul.ph %[x1], %[dq1], %[in1] \n\t"
- "sh %[x1], 4(%[input_temp]) \n\t"
- "srl %[x1], %[x1], 16 \n\t"
- "sh %[x1], 2(%[input_temp]) \n\t"
-
- : [x1] "=&r" (x1), [in1] "=&r" (in1), [dq1] "=&r" (dq1),
- [in2] "=&r" (in2), [dq2] "=&r" (dq2)
- : [dq] "r" (dq), [input] "r" (input),
- [input_temp] "r" (input_temp)
- );
-
- for (i = 3; i < 16; i++)
- {
- input_temp[i] = dq[i] * input[i];
- }
-
- vp8_short_idct4x4llm_mips(input_temp, output, 4);
-
- vpx_memset(input, 0, 32);
-
- for (i = 4; i--; )
- {
- a0 = diff_ptr[0] + pred[0];
- a1 = diff_ptr[1] + pred[1];
- a2 = diff_ptr[2] + pred[2];
- a3 = diff_ptr[3] + pred[3];
-
- dest[0] = cm[a0];
- dest[1] = cm[a1];
- dest[2] = cm[a2];
- dest[3] = cm[a3];
-
- dest += stride;
- diff_ptr += 4;
- pred += pitch;
- }
-}
\ No newline at end of file
diff --git a/vp8/decoder/mips/dequantize_mips.h b/vp8/decoder/mips/dequantize_mips.h
deleted file mode 100755
index 732bda7..0000000
--- a/vp8/decoder/mips/dequantize_mips.h
+++ /dev/null
@@ -1,38 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef DEQUANTIZE_MIPS_H
-#define DEQUANTIZE_MIPS_H
-
-extern prototype_dequant_idct_add(vp8_dequant_idct_add_mips);
-extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add_mips);
-extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_mips);
-extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_mips);
-extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_mips);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_dequant_idct_add
-#define vp8_dequant_idct_add vp8_dequant_idct_add_mips
-
-#undef vp8_dequant_dc_idct_add
-#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_mips
-
-#undef vp8_dequant_dc_idct_add_y_block
-#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_mips
-
-#undef vp8_dequant_idct_add_y_block
-#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_mips
-
-#undef vp8_dequant_idct_add_uv_block
-#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_mips
-
-#endif
-#endif
\ No newline at end of file
diff --git a/vp8/decoder/mips/idct_blk_mips.c b/vp8/decoder/mips/idct_blk_mips.c
deleted file mode 100755
index 1ae2a70..0000000
--- a/vp8/decoder/mips/idct_blk_mips.c
+++ /dev/null
@@ -1,225 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_ports/config.h"
-#include "idct.h"
-#include "dequantize.h"
-
-
-void vp8_dequant_dc_idct_add_y_block_mips
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs, short *dc)
-{
- int i, j;
- int higher = 4*stride - 12;
-
- /* unroll the loop */
- for (i = 0; i < 4; i++)
- {
- if (*eobs++ > 1)
- vp8_dequant_dc_idct_add_mips(q, dq, pre, dst, 16, stride, dc[0]);
- else
- vp8_dc_only_idct_add_mips(dc[0], pre, dst, 16, stride);
-
- q += 16;
- pre += 4;
- dst += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_dc_idct_add_mips(q, dq, pre, dst, 16, stride, dc[1]);
- else
- vp8_dc_only_idct_add_mips(dc[1], pre, dst, 16, stride);
-
- q += 16;
- pre += 4;
- dst += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_dc_idct_add_mips(q, dq, pre, dst, 16, stride, dc[2]);
- else
- vp8_dc_only_idct_add_mips(dc[2], pre, dst, 16, stride);
-
- q += 16;
- pre += 4;
- dst += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_dc_idct_add_mips(q, dq, pre, dst, 16, stride, dc[3]);
- else
- vp8_dc_only_idct_add_mips(dc[3], pre, dst, 16, stride);
-
- q += 16;
- dc += 4;
- pre += 52;
- dst += higher;
- }
-}
-
-
-void vp8_dequant_idct_add_y_block_mips
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs)
-{
- int i, j;
- int higher = 4*stride -12;
-
- /* unroll the loop */
- for (i = 4; i--; )
- {
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dst, 16, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dst, 16, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dst += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dst, 16, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dst, 16, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dst += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dst, 16, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dst, 16, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dst += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dst, 16, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dst, 16, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 52;
- dst += higher;
- }
-}
-
-
-void vp8_dequant_idct_add_uv_block_mips
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
-{
- int i, j;
- int higher = 4*stride -4;
-
- /* unroll the loops */
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dstu, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dstu, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dstu += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dstu, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dstu, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 28;
- dstu += higher;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dstu, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dstu, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dstu += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dstu, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dstu, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 28;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dstv, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dstv, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dstv += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dstv, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dstv, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 28;
- dstv += higher;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dstv, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dstv, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- q += 16;
- pre += 4;
- dstv += 4;
-
- if (*eobs++ > 1)
- vp8_dequant_idct_add_mips(q, dq, pre, dstv, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mips(q[0]*dq[0], pre, dstv, 8, stride);
- ((int *)q)[0] = 0;
- }
-}
\ No newline at end of file
diff --git a/vp8/decoder/mips/mips_dsystemdependent.c b/vp8/decoder/mips/mips_dsystemdependent.c
deleted file mode 100755
index bb20cc7..0000000
--- a/vp8/decoder/mips/mips_dsystemdependent.c
+++ /dev/null
@@ -1,36 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "blockd.h"
-#include "pragmas.h"
-#include "postproc.h"
-#include "dboolhuff.h"
-#include "dequantize.h"
-#include "onyxd_int.h"
-
-void vp8_arch_mips_decode_init(VP8D_COMP *pbi)
-{
-#if CONFIG_RUNTIME_CPU_DETECT
-
-#ifdef MIPS_DSP_REV
-#if (MIPS_DSP_REV>=2)
-
- pbi->dequant.idct_add = vp8_dequant_idct_add_mips;
- pbi->dequant.dc_idct_add = vp8_dequant_dc_idct_add_mips;
- pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_mips;
- pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_mips;
- pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_mips;
-
-#endif
-#endif
-#endif
-}
diff --git a/vp8/decoder/onyxd_if.c b/vp8/decoder/onyxd_if.c
deleted file mode 100644
index ef2e00d..0000000
--- a/vp8/decoder/onyxd_if.c
+++ /dev/null
@@ -1,496 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/onyxc_int.h"
-#if CONFIG_POSTPROC
-#include "vp8/common/postproc.h"
-#endif
-#include "vp8/common/onyxd.h"
-#include "onyxd_int.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vp8/common/alloccommon.h"
-#include "vpx_scale/yv12extend.h"
-#include "vp8/common/loopfilter.h"
-#include "vp8/common/swapyv12buffer.h"
-#include "vp8/common/g_common.h"
-#include "vp8/common/threading.h"
-#include "decoderthreading.h"
-#include <stdio.h>
-
-#include "vp8/common/quant_common.h"
-#include "vpx_scale/vpxscale.h"
-#include "vp8/common/systemdependent.h"
-#include "vpx_ports/vpx_timer.h"
-#include "detokenize.h"
-#if ARCH_ARM
-#include "vpx_ports/arm.h"
-#endif
-
-extern void vp8_init_loop_filter(VP8_COMMON *cm);
-extern void vp8cx_init_de_quantizer(VP8D_COMP *pbi);
-
-
-void vp8dx_initialize()
-{
- static int init_done = 0;
-
- if (!init_done)
- {
- vp8_initialize_common();
- vp8_scale_machine_specific_config();
- init_done = 1;
- }
-}
-
-
-VP8D_PTR vp8dx_create_decompressor(VP8D_CONFIG *oxcf)
-{
- VP8D_COMP *pbi = vpx_memalign(32, sizeof(VP8D_COMP));
-
- if (!pbi)
- return NULL;
-
- vpx_memset(pbi, 0, sizeof(VP8D_COMP));
-
- if (setjmp(pbi->common.error.jmp))
- {
- pbi->common.error.setjmp = 0;
- vp8dx_remove_decompressor(pbi);
- return 0;
- }
-
- pbi->common.error.setjmp = 1;
- vp8dx_initialize();
-
- vp8_create_common(&pbi->common);
- vp8_dmachine_specific_config(pbi);
-
- pbi->common.current_video_frame = 0;
- pbi->ready_for_new_data = 1;
-
- pbi->CPUFreq = 0; /*vp8_get_processor_freq();*/
-#if CONFIG_MULTITHREAD
- pbi->max_threads = oxcf->max_threads;
- vp8_decoder_create_threads(pbi);
-#endif
-
- /* vp8cx_init_de_quantizer() is first called here. Add check in frame_init_dequantizer() to avoid
- * unnecessary calling of vp8cx_init_de_quantizer() for every frame.
- */
- vp8cx_init_de_quantizer(pbi);
-
- {
- VP8_COMMON *cm = &pbi->common;
-
- vp8_init_loop_filter(cm);
- cm->last_frame_type = KEY_FRAME;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
- }
-
- pbi->common.error.setjmp = 0;
- return (VP8D_PTR) pbi;
-}
-
-
-void vp8dx_remove_decompressor(VP8D_PTR ptr)
-{
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
-
- if (!pbi)
- return;
-
-#if CONFIG_MULTITHREAD
- if (pbi->b_multithreaded_rd)
- vp8mt_de_alloc_temp_buffers(pbi, pbi->common.mb_rows);
- vp8_decoder_remove_threads(pbi);
-#endif
- vp8_remove_common(&pbi->common);
- vpx_free(pbi);
-}
-
-
-int vp8dx_get_reference(VP8D_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
-{
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
- VP8_COMMON *cm = &pbi->common;
- int ref_fb_idx;
-
- if (ref_frame_flag == VP8_LAST_FLAG)
- ref_fb_idx = cm->lst_fb_idx;
- else if (ref_frame_flag == VP8_GOLD_FLAG)
- ref_fb_idx = cm->gld_fb_idx;
- else if (ref_frame_flag == VP8_ALT_FLAG)
- ref_fb_idx = cm->alt_fb_idx;
- else
- return -1;
-
- vp8_yv12_copy_frame_ptr(&cm->yv12_fb[ref_fb_idx], sd);
-
- return 0;
-}
-
-
-int vp8dx_set_reference(VP8D_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
-{
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
- VP8_COMMON *cm = &pbi->common;
- int ref_fb_idx;
-
- if (ref_frame_flag == VP8_LAST_FLAG)
- ref_fb_idx = cm->lst_fb_idx;
- else if (ref_frame_flag == VP8_GOLD_FLAG)
- ref_fb_idx = cm->gld_fb_idx;
- else if (ref_frame_flag == VP8_ALT_FLAG)
- ref_fb_idx = cm->alt_fb_idx;
- else
- return -1;
-
- vp8_yv12_copy_frame_ptr(sd, &cm->yv12_fb[ref_fb_idx]);
-
- return 0;
-}
-
-/*For ARM NEON, d8-d15 are callee-saved registers, and need to be saved by us.*/
-#if HAVE_ARMV7
-extern void vp8_push_neon(INT64 *store);
-extern void vp8_pop_neon(INT64 *store);
-#endif
-
-static int get_free_fb (VP8_COMMON *cm)
-{
- int i;
- for (i = 0; i < NUM_YV12_BUFFERS; i++)
- if (cm->fb_idx_ref_cnt[i] == 0)
- break;
-
- cm->fb_idx_ref_cnt[i] = 1;
- return i;
-}
-
-static void ref_cnt_fb (int *buf, int *idx, int new_idx)
-{
- if (buf[*idx] > 0)
- buf[*idx]--;
-
- *idx = new_idx;
-
- buf[new_idx]++;
-}
-
-/* If any buffer copy / swapping is signalled it should be done here. */
-static int swap_frame_buffers (VP8_COMMON *cm)
-{
- int err = 0;
-
- /* The alternate reference frame or golden frame can be updated
- * using the new, last, or golden/alt ref frame. If it
- * is updated using the newly decoded frame it is a refresh.
- * An update using the last or golden/alt ref frame is a copy.
- */
- if (cm->copy_buffer_to_arf)
- {
- int new_fb = 0;
-
- if (cm->copy_buffer_to_arf == 1)
- new_fb = cm->lst_fb_idx;
- else if (cm->copy_buffer_to_arf == 2)
- new_fb = cm->gld_fb_idx;
- else
- err = -1;
-
- ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->alt_fb_idx, new_fb);
- }
-
- if (cm->copy_buffer_to_gf)
- {
- int new_fb = 0;
-
- if (cm->copy_buffer_to_gf == 1)
- new_fb = cm->lst_fb_idx;
- else if (cm->copy_buffer_to_gf == 2)
- new_fb = cm->alt_fb_idx;
- else
- err = -1;
-
- ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->gld_fb_idx, new_fb);
- }
-
- if (cm->refresh_golden_frame)
- ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->gld_fb_idx, cm->new_fb_idx);
-
- if (cm->refresh_alt_ref_frame)
- ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->alt_fb_idx, cm->new_fb_idx);
-
- if (cm->refresh_last_frame)
- {
- ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->lst_fb_idx, cm->new_fb_idx);
-
- cm->frame_to_show = &cm->yv12_fb[cm->lst_fb_idx];
- }
- else
- cm->frame_to_show = &cm->yv12_fb[cm->new_fb_idx];
-
- cm->fb_idx_ref_cnt[cm->new_fb_idx]--;
-
- return err;
-}
-
-int vp8dx_receive_compressed_data(VP8D_PTR ptr, unsigned long size, const unsigned char *source, INT64 time_stamp)
-{
-#if HAVE_ARMV7
- INT64 dx_store_reg[8];
-#endif
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
- VP8_COMMON *cm = &pbi->common;
- int retcode = 0;
- struct vpx_usec_timer timer;
-
- /*if(pbi->ready_for_new_data == 0)
- return -1;*/
-
- if (ptr == 0)
- {
- return -1;
- }
-
- pbi->common.error.error_code = VPX_CODEC_OK;
-
- if (size == 0)
- {
- /* This is used to signal that we are missing frames.
- * We do not know if the missing frame(s) was supposed to update
- * any of the reference buffers, but we act conservative and
- * mark only the last buffer as corrupted.
- */
- cm->yv12_fb[cm->lst_fb_idx].corrupted = 1;
-
- /* Signal that we have no frame to show. */
- cm->show_frame = 0;
-
- /* Nothing more to do. */
- return 0;
- }
-
-
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_push_neon(dx_store_reg);
- }
-#endif
-
- cm->new_fb_idx = get_free_fb (cm);
-
- if (setjmp(pbi->common.error.jmp))
- {
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(dx_store_reg);
- }
-#endif
- pbi->common.error.setjmp = 0;
-
- /* We do not know if the missing frame(s) was supposed to update
- * any of the reference buffers, but we act conservative and
- * mark only the last buffer as corrupted.
- */
- cm->yv12_fb[cm->lst_fb_idx].corrupted = 1;
-
- if (cm->fb_idx_ref_cnt[cm->new_fb_idx] > 0)
- cm->fb_idx_ref_cnt[cm->new_fb_idx]--;
- return -1;
- }
-
- pbi->common.error.setjmp = 1;
-
- vpx_usec_timer_start(&timer);
-
- /*cm->current_video_frame++;*/
- pbi->Source = source;
- pbi->source_sz = size;
-
- retcode = vp8_decode_frame(pbi);
-
- if (retcode < 0)
- {
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(dx_store_reg);
- }
-#endif
- pbi->common.error.error_code = VPX_CODEC_ERROR;
- pbi->common.error.setjmp = 0;
- if (cm->fb_idx_ref_cnt[cm->new_fb_idx] > 0)
- cm->fb_idx_ref_cnt[cm->new_fb_idx]--;
- return retcode;
- }
-
-#if CONFIG_MULTITHREAD
- if (pbi->b_multithreaded_rd && cm->multi_token_partition != ONE_PARTITION)
- {
- if (swap_frame_buffers (cm))
- {
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(dx_store_reg);
- }
-#endif
- pbi->common.error.error_code = VPX_CODEC_ERROR;
- pbi->common.error.setjmp = 0;
- return -1;
- }
- } else
-#endif
- {
- if (swap_frame_buffers (cm))
- {
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(dx_store_reg);
- }
-#endif
- pbi->common.error.error_code = VPX_CODEC_ERROR;
- pbi->common.error.setjmp = 0;
- return -1;
- }
-
- if(pbi->common.filter_level)
- {
- struct vpx_usec_timer lpftimer;
- vpx_usec_timer_start(&lpftimer);
- /* Apply the loop filter if appropriate. */
-
- vp8_loop_filter_frame(cm, &pbi->mb, cm->filter_level);
-
- vpx_usec_timer_mark(&lpftimer);
- pbi->time_loop_filtering += vpx_usec_timer_elapsed(&lpftimer);
-
- cm->last_frame_type = cm->frame_type;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
- }
- vp8_yv12_extend_frame_borders_ptr(cm->frame_to_show);
- }
-
-
- vp8_clear_system_state();
-
- vpx_usec_timer_mark(&timer);
- pbi->decode_microseconds = vpx_usec_timer_elapsed(&timer);
-
- pbi->time_decoding += pbi->decode_microseconds;
-
- /*vp8_print_modes_and_motion_vectors( cm->mi, cm->mb_rows,cm->mb_cols, cm->current_video_frame);*/
-
- if (cm->show_frame)
- cm->current_video_frame++;
-
- pbi->ready_for_new_data = 0;
- pbi->last_time_stamp = time_stamp;
-
-#if 0
- {
- int i;
- INT64 earliest_time = pbi->dr[0].time_stamp;
- INT64 latest_time = pbi->dr[0].time_stamp;
- INT64 time_diff = 0;
- int bytes = 0;
-
- pbi->dr[pbi->common.current_video_frame&0xf].size = pbi->bc.pos + pbi->bc2.pos + 4;;
- pbi->dr[pbi->common.current_video_frame&0xf].time_stamp = time_stamp;
-
- for (i = 0; i < 16; i++)
- {
-
- bytes += pbi->dr[i].size;
-
- if (pbi->dr[i].time_stamp < earliest_time)
- earliest_time = pbi->dr[i].time_stamp;
-
- if (pbi->dr[i].time_stamp > latest_time)
- latest_time = pbi->dr[i].time_stamp;
- }
-
- time_diff = latest_time - earliest_time;
-
- if (time_diff > 0)
- {
- pbi->common.bitrate = 80000.00 * bytes / time_diff ;
- pbi->common.framerate = 160000000.00 / time_diff ;
- }
-
- }
-#endif
-
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(dx_store_reg);
- }
-#endif
- pbi->common.error.setjmp = 0;
- return retcode;
-}
-int vp8dx_get_raw_frame(VP8D_PTR ptr, YV12_BUFFER_CONFIG *sd, INT64 *time_stamp, INT64 *time_end_stamp, vp8_ppflags_t *flags)
-{
- int ret = -1;
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
-
- if (pbi->ready_for_new_data == 1)
- return ret;
-
- /* ie no raw frame to show!!! */
- if (pbi->common.show_frame == 0)
- return ret;
-
- pbi->ready_for_new_data = 1;
- *time_stamp = pbi->last_time_stamp;
- *time_end_stamp = 0;
-
- sd->clrtype = pbi->common.clr_type;
-#if CONFIG_POSTPROC
- ret = vp8_post_proc_frame(&pbi->common, sd, flags);
-#else
-
- if (pbi->common.frame_to_show)
- {
- *sd = *pbi->common.frame_to_show;
- sd->y_width = pbi->common.Width;
- sd->y_height = pbi->common.Height;
- sd->uv_height = pbi->common.Height / 2;
- ret = 0;
- }
- else
- {
- ret = -1;
- }
-
-#endif /*!CONFIG_POSTPROC*/
- vp8_clear_system_state();
- return ret;
-}
diff --git a/vp8/decoder/onyxd_int.h b/vp8/decoder/onyxd_int.h
deleted file mode 100644
index ac1e332..0000000
--- a/vp8/decoder/onyxd_int.h
+++ /dev/null
@@ -1,160 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_VP8D_INT_H
-#define __INC_VP8D_INT_H
-#include "vpx_ports/config.h"
-#include "vp8/common/onyxd.h"
-#include "treereader.h"
-#include "vp8/common/onyxc_int.h"
-#include "vp8/common/threading.h"
-#include "dequantize.h"
-
-typedef struct
-{
- int ithread;
- void *ptr1;
- void *ptr2;
-} DECODETHREAD_DATA;
-
-typedef struct
-{
- MACROBLOCKD mbd;
- int mb_row;
- int current_mb_col;
- short *coef_ptr;
-} MB_ROW_DEC;
-
-typedef struct
-{
- INT64 time_stamp;
- int size;
-} DATARATE;
-
-typedef struct
-{
- INT16 min_val;
- INT16 Length;
- UINT8 Probs[12];
-} TOKENEXTRABITS;
-
-typedef struct
-{
- int const *scan;
- UINT8 const *ptr_block2leftabove;
- vp8_tree_index const *vp8_coef_tree_ptr;
- TOKENEXTRABITS const *teb_base_ptr;
- unsigned char *norm_ptr;
- UINT8 *ptr_coef_bands_x;
-
- ENTROPY_CONTEXT_PLANES *A;
- ENTROPY_CONTEXT_PLANES *L;
-
- INT16 *qcoeff_start_ptr;
- BOOL_DECODER *current_bc;
-
- vp8_prob const *coef_probs[4];
-
- UINT8 eob[25];
-
-} DETOK;
-
-typedef struct VP8Decompressor
-{
- DECLARE_ALIGNED(16, MACROBLOCKD, mb);
-
- DECLARE_ALIGNED(16, VP8_COMMON, common);
-
- vp8_reader bc, bc2;
-
- VP8D_CONFIG oxcf;
-
-
- const unsigned char *Source;
- unsigned int source_sz;
-
-
- unsigned int CPUFreq;
- unsigned int decode_microseconds;
- unsigned int time_decoding;
- unsigned int time_loop_filtering;
-
-#if CONFIG_MULTITHREAD
- /* variable for threading */
-
- volatile int b_multithreaded_rd;
- int max_threads;
- int current_mb_col_main;
- int decoding_thread_count;
- int allocated_decoding_thread_count;
-
- int mt_baseline_filter_level[MAX_MB_SEGMENTS];
- int sync_range;
- int *mt_current_mb_col; /* Each row remembers its already decoded column. */
-
- unsigned char **mt_yabove_row; /* mb_rows x width */
- unsigned char **mt_uabove_row;
- unsigned char **mt_vabove_row;
- unsigned char **mt_yleft_col; /* mb_rows x 16 */
- unsigned char **mt_uleft_col; /* mb_rows x 8 */
- unsigned char **mt_vleft_col; /* mb_rows x 8 */
-
- MB_ROW_DEC *mb_row_di;
- DECODETHREAD_DATA *de_thread_data;
-
- pthread_t *h_decoding_thread;
- sem_t *h_event_start_decoding;
- sem_t h_event_end_decoding;
- /* end of threading data */
-#endif
-
- vp8_reader *mbc;
- INT64 last_time_stamp;
- int ready_for_new_data;
-
- DATARATE dr[16];
-
- DETOK detoken;
-
-#if CONFIG_RUNTIME_CPU_DETECT
- vp8_dequant_rtcd_vtable_t dequant;
-#endif
-
-
- vp8_prob prob_intra;
- vp8_prob prob_last;
- vp8_prob prob_gf;
- vp8_prob prob_skip_false;
-
-} VP8D_COMP;
-
-int vp8_decode_frame(VP8D_COMP *cpi);
-void vp8_dmachine_specific_config(VP8D_COMP *pbi);
-
-
-#if CONFIG_DEBUG
-#define CHECK_MEM_ERROR(lval,expr) do {\
- lval = (expr); \
- if(!lval) \
- vpx_internal_error(&pbi->common.error, VPX_CODEC_MEM_ERROR,\
- "Failed to allocate "#lval" at %s:%d", \
- __FILE__,__LINE__);\
- } while(0)
-#else
-#define CHECK_MEM_ERROR(lval,expr) do {\
- lval = (expr); \
- if(!lval) \
- vpx_internal_error(&pbi->common.error, VPX_CODEC_MEM_ERROR,\
- "Failed to allocate "#lval);\
- } while(0)
-#endif
-
-#endif
diff --git a/vp8/decoder/reconintra_mt.c b/vp8/decoder/reconintra_mt.c
deleted file mode 100644
index b9d2b37..0000000
--- a/vp8/decoder/reconintra_mt.c
+++ /dev/null
@@ -1,937 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/common/recon.h"
-#include "vp8/common/reconintra.h"
-#include "vpx_mem/vpx_mem.h"
-#include "onyxd_int.h"
-
-/* For skip_recon_mb(), add vp8_build_intra_predictors_mby_s(MACROBLOCKD *x) and
- * vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x).
- */
-
-void vp8mt_build_intra_predictors_mby(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col)
-{
- unsigned char *yabove_row; /* = x->dst.y_buffer - x->dst.y_stride; */
- unsigned char *yleft_col;
- unsigned char yleft_buf[16];
- unsigned char ytop_left; /* = yabove_row[-1]; */
- unsigned char *ypred_ptr = x->predictor;
- int r, c, i;
-
- if (pbi->common.filter_level)
- {
- yabove_row = pbi->mt_yabove_row[mb_row] + mb_col*16 +32;
- yleft_col = pbi->mt_yleft_col[mb_row];
- } else
- {
- yabove_row = x->dst.y_buffer - x->dst.y_stride;
-
- for (i = 0; i < 16; i++)
- yleft_buf[i] = x->dst.y_buffer [i* x->dst.y_stride -1];
- yleft_col = yleft_buf;
- }
-
- ytop_left = yabove_row[-1];
-
- /* for Y */
- switch (x->mode_info_context->mbmi.mode)
- {
- case DC_PRED:
- {
- int expected_dc;
- int i;
- int shift;
- int average = 0;
-
-
- if (x->up_available || x->left_available)
- {
- if (x->up_available)
- {
- for (i = 0; i < 16; i++)
- {
- average += yabove_row[i];
- }
- }
-
- if (x->left_available)
- {
-
- for (i = 0; i < 16; i++)
- {
- average += yleft_col[i];
- }
-
- }
-
-
-
- shift = 3 + x->up_available + x->left_available;
- expected_dc = (average + (1 << (shift - 1))) >> shift;
- }
- else
- {
- expected_dc = 128;
- }
-
- vpx_memset(ypred_ptr, expected_dc, 256);
- }
- break;
- case V_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
-
- ((int *)ypred_ptr)[0] = ((int *)yabove_row)[0];
- ((int *)ypred_ptr)[1] = ((int *)yabove_row)[1];
- ((int *)ypred_ptr)[2] = ((int *)yabove_row)[2];
- ((int *)ypred_ptr)[3] = ((int *)yabove_row)[3];
- ypred_ptr += 16;
- }
- }
- break;
- case H_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
-
- vpx_memset(ypred_ptr, yleft_col[r], 16);
- ypred_ptr += 16;
- }
-
- }
- break;
- case TM_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
- for (c = 0; c < 16; c++)
- {
- int pred = yleft_col[r] + yabove_row[ c] - ytop_left;
-
- if (pred < 0)
- pred = 0;
-
- if (pred > 255)
- pred = 255;
-
- ypred_ptr[c] = pred;
- }
-
- ypred_ptr += 16;
- }
-
- }
- break;
- case B_PRED:
- case NEARESTMV:
- case NEARMV:
- case ZEROMV:
- case NEWMV:
- case SPLITMV:
- case MB_MODE_COUNT:
- break;
- }
-}
-
-void vp8mt_build_intra_predictors_mby_s(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col)
-{
- unsigned char *yabove_row; /* = x->dst.y_buffer - x->dst.y_stride; */
- unsigned char *yleft_col;
- unsigned char yleft_buf[16];
- unsigned char ytop_left; /* = yabove_row[-1]; */
- unsigned char *ypred_ptr = x->predictor;
- int r, c, i;
-
- int y_stride = x->dst.y_stride;
- ypred_ptr = x->dst.y_buffer; /*x->predictor;*/
-
- if (pbi->common.filter_level)
- {
- yabove_row = pbi->mt_yabove_row[mb_row] + mb_col*16 +32;
- yleft_col = pbi->mt_yleft_col[mb_row];
- } else
- {
- yabove_row = x->dst.y_buffer - x->dst.y_stride;
-
- for (i = 0; i < 16; i++)
- yleft_buf[i] = x->dst.y_buffer [i* x->dst.y_stride -1];
- yleft_col = yleft_buf;
- }
-
- ytop_left = yabove_row[-1];
-
- /* for Y */
- switch (x->mode_info_context->mbmi.mode)
- {
- case DC_PRED:
- {
- int expected_dc;
- int i;
- int shift;
- int average = 0;
-
-
- if (x->up_available || x->left_available)
- {
- if (x->up_available)
- {
- for (i = 0; i < 16; i++)
- {
- average += yabove_row[i];
- }
- }
-
- if (x->left_available)
- {
-
- for (i = 0; i < 16; i++)
- {
- average += yleft_col[i];
- }
-
- }
-
-
-
- shift = 3 + x->up_available + x->left_available;
- expected_dc = (average + (1 << (shift - 1))) >> shift;
- }
- else
- {
- expected_dc = 128;
- }
-
- /*vpx_memset(ypred_ptr, expected_dc, 256);*/
- for (r = 0; r < 16; r++)
- {
- vpx_memset(ypred_ptr, expected_dc, 16);
- ypred_ptr += y_stride; /*16;*/
- }
- }
- break;
- case V_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
-
- ((int *)ypred_ptr)[0] = ((int *)yabove_row)[0];
- ((int *)ypred_ptr)[1] = ((int *)yabove_row)[1];
- ((int *)ypred_ptr)[2] = ((int *)yabove_row)[2];
- ((int *)ypred_ptr)[3] = ((int *)yabove_row)[3];
- ypred_ptr += y_stride; /*16;*/
- }
- }
- break;
- case H_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
-
- vpx_memset(ypred_ptr, yleft_col[r], 16);
- ypred_ptr += y_stride; /*16;*/
- }
-
- }
- break;
- case TM_PRED:
- {
-
- for (r = 0; r < 16; r++)
- {
- for (c = 0; c < 16; c++)
- {
- int pred = yleft_col[r] + yabove_row[ c] - ytop_left;
-
- if (pred < 0)
- pred = 0;
-
- if (pred > 255)
- pred = 255;
-
- ypred_ptr[c] = pred;
- }
-
- ypred_ptr += y_stride; /*16;*/
- }
-
- }
- break;
- case B_PRED:
- case NEARESTMV:
- case NEARMV:
- case ZEROMV:
- case NEWMV:
- case SPLITMV:
- case MB_MODE_COUNT:
- break;
- }
-}
-
-void vp8mt_build_intra_predictors_mbuv(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col)
-{
- unsigned char *uabove_row; /* = x->dst.u_buffer - x->dst.uv_stride; */
- unsigned char *uleft_col; /*[16];*/
- unsigned char uleft_buf[8];
- unsigned char utop_left; /* = uabove_row[-1]; */
- unsigned char *vabove_row; /* = x->dst.v_buffer - x->dst.uv_stride; */
- unsigned char *vleft_col; /*[20];*/
- unsigned char vleft_buf[8];
- unsigned char vtop_left; /* = vabove_row[-1]; */
- unsigned char *upred_ptr = &x->predictor[256];
- unsigned char *vpred_ptr = &x->predictor[320];
- int i, j;
-
- if (pbi->common.filter_level)
- {
- uabove_row = pbi->mt_uabove_row[mb_row] + mb_col*8 +16;
- vabove_row = pbi->mt_vabove_row[mb_row] + mb_col*8 +16;
- uleft_col = pbi->mt_uleft_col[mb_row];
- vleft_col = pbi->mt_vleft_col[mb_row];
- } else
- {
- uabove_row = x->dst.u_buffer - x->dst.uv_stride;
- vabove_row = x->dst.v_buffer - x->dst.uv_stride;
-
- for (i = 0; i < 8; i++)
- {
- uleft_buf[i] = x->dst.u_buffer [i* x->dst.uv_stride -1];
- vleft_buf[i] = x->dst.v_buffer [i* x->dst.uv_stride -1];
- }
- uleft_col = uleft_buf;
- vleft_col = vleft_buf;
- }
- utop_left = uabove_row[-1];
- vtop_left = vabove_row[-1];
-
- switch (x->mode_info_context->mbmi.uv_mode)
- {
- case DC_PRED:
- {
- int expected_udc;
- int expected_vdc;
- int i;
- int shift;
- int Uaverage = 0;
- int Vaverage = 0;
-
- if (x->up_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uabove_row[i];
- Vaverage += vabove_row[i];
- }
- }
-
- if (x->left_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uleft_col[i];
- Vaverage += vleft_col[i];
- }
- }
-
- if (!x->up_available && !x->left_available)
- {
- expected_udc = 128;
- expected_vdc = 128;
- }
- else
- {
- shift = 2 + x->up_available + x->left_available;
- expected_udc = (Uaverage + (1 << (shift - 1))) >> shift;
- expected_vdc = (Vaverage + (1 << (shift - 1))) >> shift;
- }
-
-
- vpx_memset(upred_ptr, expected_udc, 64);
- vpx_memset(vpred_ptr, expected_vdc, 64);
-
-
- }
- break;
- case V_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- vpx_memcpy(upred_ptr, uabove_row, 8);
- vpx_memcpy(vpred_ptr, vabove_row, 8);
- upred_ptr += 8;
- vpred_ptr += 8;
- }
-
- }
- break;
- case H_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- vpx_memset(upred_ptr, uleft_col[i], 8);
- vpx_memset(vpred_ptr, vleft_col[i], 8);
- upred_ptr += 8;
- vpred_ptr += 8;
- }
- }
-
- break;
- case TM_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- for (j = 0; j < 8; j++)
- {
- int predu = uleft_col[i] + uabove_row[j] - utop_left;
- int predv = vleft_col[i] + vabove_row[j] - vtop_left;
-
- if (predu < 0)
- predu = 0;
-
- if (predu > 255)
- predu = 255;
-
- if (predv < 0)
- predv = 0;
-
- if (predv > 255)
- predv = 255;
-
- upred_ptr[j] = predu;
- vpred_ptr[j] = predv;
- }
-
- upred_ptr += 8;
- vpred_ptr += 8;
- }
-
- }
- break;
- case B_PRED:
- case NEARESTMV:
- case NEARMV:
- case ZEROMV:
- case NEWMV:
- case SPLITMV:
- case MB_MODE_COUNT:
- break;
- }
-}
-
-void vp8mt_build_intra_predictors_mbuv_s(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col)
-{
- unsigned char *uabove_row; /* = x->dst.u_buffer - x->dst.uv_stride; */
- unsigned char *uleft_col; /*[16];*/
- unsigned char uleft_buf[8];
- unsigned char utop_left; /* = uabove_row[-1]; */
- unsigned char *vabove_row; /* = x->dst.v_buffer - x->dst.uv_stride; */
- unsigned char *vleft_col; /*[20];*/
- unsigned char vleft_buf[8];
- unsigned char vtop_left; /* = vabove_row[-1]; */
- unsigned char *upred_ptr = x->dst.u_buffer; /*&x->predictor[256];*/
- unsigned char *vpred_ptr = x->dst.v_buffer; /*&x->predictor[320];*/
- int uv_stride = x->dst.uv_stride;
- int i, j;
-
- if (pbi->common.filter_level)
- {
- uabove_row = pbi->mt_uabove_row[mb_row] + mb_col*8 +16;
- vabove_row = pbi->mt_vabove_row[mb_row] + mb_col*8 +16;
- uleft_col = pbi->mt_uleft_col[mb_row];
- vleft_col = pbi->mt_vleft_col[mb_row];
- } else
- {
- uabove_row = x->dst.u_buffer - x->dst.uv_stride;
- vabove_row = x->dst.v_buffer - x->dst.uv_stride;
-
- for (i = 0; i < 8; i++)
- {
- uleft_buf[i] = x->dst.u_buffer [i* x->dst.uv_stride -1];
- vleft_buf[i] = x->dst.v_buffer [i* x->dst.uv_stride -1];
- }
- uleft_col = uleft_buf;
- vleft_col = vleft_buf;
- }
- utop_left = uabove_row[-1];
- vtop_left = vabove_row[-1];
-
- switch (x->mode_info_context->mbmi.uv_mode)
- {
- case DC_PRED:
- {
- int expected_udc;
- int expected_vdc;
- int i;
- int shift;
- int Uaverage = 0;
- int Vaverage = 0;
-
- if (x->up_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uabove_row[i];
- Vaverage += vabove_row[i];
- }
- }
-
- if (x->left_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uleft_col[i];
- Vaverage += vleft_col[i];
- }
- }
-
- if (!x->up_available && !x->left_available)
- {
- expected_udc = 128;
- expected_vdc = 128;
- }
- else
- {
- shift = 2 + x->up_available + x->left_available;
- expected_udc = (Uaverage + (1 << (shift - 1))) >> shift;
- expected_vdc = (Vaverage + (1 << (shift - 1))) >> shift;
- }
-
-
- /*vpx_memset(upred_ptr,expected_udc,64);
- vpx_memset(vpred_ptr,expected_vdc,64);*/
- for (i = 0; i < 8; i++)
- {
- vpx_memset(upred_ptr, expected_udc, 8);
- vpx_memset(vpred_ptr, expected_vdc, 8);
- upred_ptr += uv_stride; /*8;*/
- vpred_ptr += uv_stride; /*8;*/
- }
- }
- break;
- case V_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- vpx_memcpy(upred_ptr, uabove_row, 8);
- vpx_memcpy(vpred_ptr, vabove_row, 8);
- upred_ptr += uv_stride; /*8;*/
- vpred_ptr += uv_stride; /*8;*/
- }
-
- }
- break;
- case H_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- vpx_memset(upred_ptr, uleft_col[i], 8);
- vpx_memset(vpred_ptr, vleft_col[i], 8);
- upred_ptr += uv_stride; /*8;*/
- vpred_ptr += uv_stride; /*8;*/
- }
- }
-
- break;
- case TM_PRED:
- {
- int i;
-
- for (i = 0; i < 8; i++)
- {
- for (j = 0; j < 8; j++)
- {
- int predu = uleft_col[i] + uabove_row[j] - utop_left;
- int predv = vleft_col[i] + vabove_row[j] - vtop_left;
-
- if (predu < 0)
- predu = 0;
-
- if (predu > 255)
- predu = 255;
-
- if (predv < 0)
- predv = 0;
-
- if (predv > 255)
- predv = 255;
-
- upred_ptr[j] = predu;
- vpred_ptr[j] = predv;
- }
-
- upred_ptr += uv_stride; /*8;*/
- vpred_ptr += uv_stride; /*8;*/
- }
-
- }
- break;
- case B_PRED:
- case NEARESTMV:
- case NEARMV:
- case ZEROMV:
- case NEWMV:
- case SPLITMV:
- case MB_MODE_COUNT:
- break;
- }
-}
-
-
-void vp8mt_predict_intra4x4(VP8D_COMP *pbi,
- MACROBLOCKD *xd,
- int b_mode,
- unsigned char *predictor,
- int mb_row,
- int mb_col,
- int num)
-{
- int i, r, c;
-
- unsigned char *Above; /* = *(x->base_dst) + x->dst - x->dst_stride; */
- unsigned char Left[4];
- unsigned char top_left; /* = Above[-1]; */
-
- BLOCKD *x = &xd->block[num];
-
- /*Caution: For some b_mode, it needs 8 pixels (4 above + 4 above-right).*/
- if (num < 4 && pbi->common.filter_level)
- Above = pbi->mt_yabove_row[mb_row] + mb_col*16 + num*4 + 32;
- else
- Above = *(x->base_dst) + x->dst - x->dst_stride;
-
- if (num%4==0 && pbi->common.filter_level)
- {
- for (i=0; i<4; i++)
- Left[i] = pbi->mt_yleft_col[mb_row][num + i];
- }else
- {
- Left[0] = (*(x->base_dst))[x->dst - 1];
- Left[1] = (*(x->base_dst))[x->dst - 1 + x->dst_stride];
- Left[2] = (*(x->base_dst))[x->dst - 1 + 2 * x->dst_stride];
- Left[3] = (*(x->base_dst))[x->dst - 1 + 3 * x->dst_stride];
- }
-
- if ((num==4 || num==8 || num==12) && pbi->common.filter_level)
- top_left = pbi->mt_yleft_col[mb_row][num-1];
- else
- top_left = Above[-1];
-
- switch (b_mode)
- {
- case B_DC_PRED:
- {
- int expected_dc = 0;
-
- for (i = 0; i < 4; i++)
- {
- expected_dc += Above[i];
- expected_dc += Left[i];
- }
-
- expected_dc = (expected_dc + 4) >> 3;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- predictor[c] = expected_dc;
- }
-
- predictor += 16;
- }
- }
- break;
- case B_TM_PRED:
- {
- /* prediction similar to true_motion prediction */
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- int pred = Above[c] - top_left + Left[r];
-
- if (pred < 0)
- pred = 0;
-
- if (pred > 255)
- pred = 255;
-
- predictor[c] = pred;
- }
-
- predictor += 16;
- }
- }
- break;
-
- case B_VE_PRED:
- {
-
- unsigned int ap[4];
- ap[0] = (top_left + 2 * Above[0] + Above[1] + 2) >> 2;
- ap[1] = (Above[0] + 2 * Above[1] + Above[2] + 2) >> 2;
- ap[2] = (Above[1] + 2 * Above[2] + Above[3] + 2) >> 2;
- ap[3] = (Above[2] + 2 * Above[3] + Above[4] + 2) >> 2;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
-
- predictor[c] = ap[c];
- }
-
- predictor += 16;
- }
-
- }
- break;
-
-
- case B_HE_PRED:
- {
-
- unsigned int lp[4];
- lp[0] = (top_left + 2 * Left[0] + Left[1] + 2) >> 2;
- lp[1] = (Left[0] + 2 * Left[1] + Left[2] + 2) >> 2;
- lp[2] = (Left[1] + 2 * Left[2] + Left[3] + 2) >> 2;
- lp[3] = (Left[2] + 2 * Left[3] + Left[3] + 2) >> 2;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- predictor[c] = lp[r];
- }
-
- predictor += 16;
- }
- }
- break;
- case B_LD_PRED:
- {
- unsigned char *ptr = Above;
- predictor[0 * 16 + 0] = (ptr[0] + ptr[1] * 2 + ptr[2] + 2) >> 2;
- predictor[0 * 16 + 1] =
- predictor[1 * 16 + 0] = (ptr[1] + ptr[2] * 2 + ptr[3] + 2) >> 2;
- predictor[0 * 16 + 2] =
- predictor[1 * 16 + 1] =
- predictor[2 * 16 + 0] = (ptr[2] + ptr[3] * 2 + ptr[4] + 2) >> 2;
- predictor[0 * 16 + 3] =
- predictor[1 * 16 + 2] =
- predictor[2 * 16 + 1] =
- predictor[3 * 16 + 0] = (ptr[3] + ptr[4] * 2 + ptr[5] + 2) >> 2;
- predictor[1 * 16 + 3] =
- predictor[2 * 16 + 2] =
- predictor[3 * 16 + 1] = (ptr[4] + ptr[5] * 2 + ptr[6] + 2) >> 2;
- predictor[2 * 16 + 3] =
- predictor[3 * 16 + 2] = (ptr[5] + ptr[6] * 2 + ptr[7] + 2) >> 2;
- predictor[3 * 16 + 3] = (ptr[6] + ptr[7] * 2 + ptr[7] + 2) >> 2;
-
- }
- break;
- case B_RD_PRED:
- {
-
- unsigned char pp[9];
-
- pp[0] = Left[3];
- pp[1] = Left[2];
- pp[2] = Left[1];
- pp[3] = Left[0];
- pp[4] = top_left;
- pp[5] = Above[0];
- pp[6] = Above[1];
- pp[7] = Above[2];
- pp[8] = Above[3];
-
- predictor[3 * 16 + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
- predictor[3 * 16 + 1] =
- predictor[2 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[3 * 16 + 2] =
- predictor[2 * 16 + 1] =
- predictor[1 * 16 + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
- predictor[3 * 16 + 3] =
- predictor[2 * 16 + 2] =
- predictor[1 * 16 + 1] =
- predictor[0 * 16 + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
- predictor[2 * 16 + 3] =
- predictor[1 * 16 + 2] =
- predictor[0 * 16 + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
- predictor[1 * 16 + 3] =
- predictor[0 * 16 + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
- predictor[0 * 16 + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
-
- }
- break;
- case B_VR_PRED:
- {
-
- unsigned char pp[9];
-
- pp[0] = Left[3];
- pp[1] = Left[2];
- pp[2] = Left[1];
- pp[3] = Left[0];
- pp[4] = top_left;
- pp[5] = Above[0];
- pp[6] = Above[1];
- pp[7] = Above[2];
- pp[8] = Above[3];
-
-
- predictor[3 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[2 * 16 + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
- predictor[3 * 16 + 1] =
- predictor[1 * 16 + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
- predictor[2 * 16 + 1] =
- predictor[0 * 16 + 0] = (pp[4] + pp[5] + 1) >> 1;
- predictor[3 * 16 + 2] =
- predictor[1 * 16 + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
- predictor[2 * 16 + 2] =
- predictor[0 * 16 + 1] = (pp[5] + pp[6] + 1) >> 1;
- predictor[3 * 16 + 3] =
- predictor[1 * 16 + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
- predictor[2 * 16 + 3] =
- predictor[0 * 16 + 2] = (pp[6] + pp[7] + 1) >> 1;
- predictor[1 * 16 + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
- predictor[0 * 16 + 3] = (pp[7] + pp[8] + 1) >> 1;
-
- }
- break;
- case B_VL_PRED:
- {
-
- unsigned char *pp = Above;
-
- predictor[0 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
- predictor[1 * 16 + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
- predictor[2 * 16 + 0] =
- predictor[0 * 16 + 1] = (pp[1] + pp[2] + 1) >> 1;
- predictor[1 * 16 + 1] =
- predictor[3 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[2 * 16 + 1] =
- predictor[0 * 16 + 2] = (pp[2] + pp[3] + 1) >> 1;
- predictor[3 * 16 + 1] =
- predictor[1 * 16 + 2] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
- predictor[0 * 16 + 3] =
- predictor[2 * 16 + 2] = (pp[3] + pp[4] + 1) >> 1;
- predictor[1 * 16 + 3] =
- predictor[3 * 16 + 2] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
- predictor[2 * 16 + 3] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
- predictor[3 * 16 + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
- }
- break;
-
- case B_HD_PRED:
- {
- unsigned char pp[9];
- pp[0] = Left[3];
- pp[1] = Left[2];
- pp[2] = Left[1];
- pp[3] = Left[0];
- pp[4] = top_left;
- pp[5] = Above[0];
- pp[6] = Above[1];
- pp[7] = Above[2];
- pp[8] = Above[3];
-
-
- predictor[3 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
- predictor[3 * 16 + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
- predictor[2 * 16 + 0] =
- predictor[3 * 16 + 2] = (pp[1] + pp[2] + 1) >> 1;
- predictor[2 * 16 + 1] =
- predictor[3 * 16 + 3] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[2 * 16 + 2] =
- predictor[1 * 16 + 0] = (pp[2] + pp[3] + 1) >> 1;
- predictor[2 * 16 + 3] =
- predictor[1 * 16 + 1] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
- predictor[1 * 16 + 2] =
- predictor[0 * 16 + 0] = (pp[3] + pp[4] + 1) >> 1;
- predictor[1 * 16 + 3] =
- predictor[0 * 16 + 1] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
- predictor[0 * 16 + 2] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
- predictor[0 * 16 + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
- }
- break;
-
-
- case B_HU_PRED:
- {
- unsigned char *pp = Left;
- predictor[0 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
- predictor[0 * 16 + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
- predictor[0 * 16 + 2] =
- predictor[1 * 16 + 0] = (pp[1] + pp[2] + 1) >> 1;
- predictor[0 * 16 + 3] =
- predictor[1 * 16 + 1] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
- predictor[1 * 16 + 2] =
- predictor[2 * 16 + 0] = (pp[2] + pp[3] + 1) >> 1;
- predictor[1 * 16 + 3] =
- predictor[2 * 16 + 1] = (pp[2] + pp[3] * 2 + pp[3] + 2) >> 2;
- predictor[2 * 16 + 2] =
- predictor[2 * 16 + 3] =
- predictor[3 * 16 + 0] =
- predictor[3 * 16 + 1] =
- predictor[3 * 16 + 2] =
- predictor[3 * 16 + 3] = pp[3];
- }
- break;
-
-
- }
-}
-
-/* copy 4 bytes from the above right down so that the 4x4 prediction modes using pixels above and
- * to the right prediction have filled in pixels to use.
- */
-void vp8mt_intra_prediction_down_copy(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col)
-{
- unsigned char *above_right; /* = *(x->block[0].base_dst) + x->block[0].dst - x->block[0].dst_stride + 16; */
- unsigned int *src_ptr;
- unsigned int *dst_ptr0;
- unsigned int *dst_ptr1;
- unsigned int *dst_ptr2;
-
- if (pbi->common.filter_level)
- above_right = pbi->mt_yabove_row[mb_row] + mb_col*16 + 32 +16;
- else
- above_right = *(x->block[0].base_dst) + x->block[0].dst - x->block[0].dst_stride + 16;
-
- src_ptr = (unsigned int *)above_right;
- /*dst_ptr0 = (unsigned int *)(above_right + 4 * x->block[0].dst_stride);
- dst_ptr1 = (unsigned int *)(above_right + 8 * x->block[0].dst_stride);
- dst_ptr2 = (unsigned int *)(above_right + 12 * x->block[0].dst_stride);*/
- dst_ptr0 = (unsigned int *)(*(x->block[0].base_dst) + x->block[0].dst + 16 + 3 * x->block[0].dst_stride);
- dst_ptr1 = (unsigned int *)(*(x->block[0].base_dst) + x->block[0].dst + 16 + 7 * x->block[0].dst_stride);
- dst_ptr2 = (unsigned int *)(*(x->block[0].base_dst) + x->block[0].dst + 16 + 11 * x->block[0].dst_stride);
- *dst_ptr0 = *src_ptr;
- *dst_ptr1 = *src_ptr;
- *dst_ptr2 = *src_ptr;
-}
diff --git a/vp8/decoder/reconintra_mt.h b/vp8/decoder/reconintra_mt.h
deleted file mode 100644
index d401295..0000000
--- a/vp8/decoder/reconintra_mt.h
+++ /dev/null
@@ -1,26 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_RECONINTRA_MT_H
-#define __INC_RECONINTRA_MT_H
-
-/* reconintra functions used in multi-threaded decoder */
-#if CONFIG_MULTITHREAD
-extern void vp8mt_build_intra_predictors_mby(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col);
-extern void vp8mt_build_intra_predictors_mby_s(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col);
-extern void vp8mt_build_intra_predictors_mbuv(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col);
-extern void vp8mt_build_intra_predictors_mbuv_s(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col);
-
-extern void vp8mt_predict_intra4x4(VP8D_COMP *pbi, MACROBLOCKD *x, int b_mode, unsigned char *predictor, int mb_row, int mb_col, int num);
-extern void vp8mt_intra_prediction_down_copy(VP8D_COMP *pbi, MACROBLOCKD *x, int mb_row, int mb_col);
-#endif
-
-#endif
diff --git a/vp8/decoder/threading.c b/vp8/decoder/threading.c
deleted file mode 100644
index 3d9d428..0000000
--- a/vp8/decoder/threading.c
+++ /dev/null
@@ -1,899 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#if !defined(WIN32) && CONFIG_OS_SUPPORT == 1
-# include <unistd.h>
-#endif
-#include "onyxd_int.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vp8/common/threading.h"
-
-#include "vp8/common/loopfilter.h"
-#include "vp8/common/extend.h"
-#include "vpx_ports/vpx_timer.h"
-#include "detokenize.h"
-#include "vp8/common/reconinter.h"
-#include "reconintra_mt.h"
-
-extern void mb_init_dequantizer(VP8D_COMP *pbi, MACROBLOCKD *xd);
-extern void clamp_mvs(MACROBLOCKD *xd);
-extern void vp8_build_uvmvs(MACROBLOCKD *x, int fullpixel);
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define RTCD_VTABLE(x) (&(pbi)->common.rtcd.x)
-#else
-#define RTCD_VTABLE(x) NULL
-#endif
-
-static void setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC *mbrd, int count)
-{
- VP8_COMMON *const pc = & pbi->common;
- int i, j;
-
- for (i = 0; i < count; i++)
- {
- MACROBLOCKD *mbd = &mbrd[i].mbd;
-#if CONFIG_RUNTIME_CPU_DETECT
- mbd->rtcd = xd->rtcd;
-#endif
- mbd->subpixel_predict = xd->subpixel_predict;
- mbd->subpixel_predict8x4 = xd->subpixel_predict8x4;
- mbd->subpixel_predict8x8 = xd->subpixel_predict8x8;
- mbd->subpixel_predict16x16 = xd->subpixel_predict16x16;
-
- mbd->mode_info_context = pc->mi + pc->mode_info_stride * (i + 1);
- mbd->mode_info_stride = pc->mode_info_stride;
-
- mbd->frame_type = pc->frame_type;
- mbd->frames_since_golden = pc->frames_since_golden;
- mbd->frames_till_alt_ref_frame = pc->frames_till_alt_ref_frame;
-
- mbd->pre = pc->yv12_fb[pc->lst_fb_idx];
- mbd->dst = pc->yv12_fb[pc->new_fb_idx];
-
- vp8_setup_block_dptrs(mbd);
- vp8_build_block_doffsets(mbd);
- mbd->segmentation_enabled = xd->segmentation_enabled;
- mbd->mb_segement_abs_delta = xd->mb_segement_abs_delta;
- vpx_memcpy(mbd->segment_feature_data, xd->segment_feature_data, sizeof(xd->segment_feature_data));
-
- /*signed char ref_lf_deltas[MAX_REF_LF_DELTAS];*/
- vpx_memcpy(mbd->ref_lf_deltas, xd->ref_lf_deltas, sizeof(xd->ref_lf_deltas));
- /*signed char mode_lf_deltas[MAX_MODE_LF_DELTAS];*/
- vpx_memcpy(mbd->mode_lf_deltas, xd->mode_lf_deltas, sizeof(xd->mode_lf_deltas));
- /*unsigned char mode_ref_lf_delta_enabled;
- unsigned char mode_ref_lf_delta_update;*/
- mbd->mode_ref_lf_delta_enabled = xd->mode_ref_lf_delta_enabled;
- mbd->mode_ref_lf_delta_update = xd->mode_ref_lf_delta_update;
-
- mbd->current_bc = &pbi->bc2;
-
- for (j = 0; j < 25; j++)
- {
- mbd->block[j].dequant = xd->block[j].dequant;
- }
- }
-
- for (i=0; i< pc->mb_rows; i++)
- pbi->mt_current_mb_col[i]=-1;
-}
-
-
-static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd, int mb_row, int mb_col)
-{
- int eobtotal = 0;
- int i, do_clamp = xd->mode_info_context->mbmi.need_to_clamp_mvs;
- VP8_COMMON *pc = &pbi->common;
-
- if (xd->mode_info_context->mbmi.mb_skip_coeff)
- {
- vp8_reset_mb_tokens_context(xd);
- }
- else
- {
- eobtotal = vp8_decode_mb_tokens(pbi, xd);
- }
-
- /* Perform temporary clamping of the MV to be used for prediction */
- if (do_clamp)
- {
- clamp_mvs(xd);
- }
-
- xd->mode_info_context->mbmi.dc_diff = 1;
-
- if (xd->mode_info_context->mbmi.mode != B_PRED && xd->mode_info_context->mbmi.mode != SPLITMV && eobtotal == 0)
- {
- xd->mode_info_context->mbmi.dc_diff = 0;
-
- /*mt_skip_recon_mb(pbi, xd, mb_row, mb_col);*/
- if (xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
- {
- vp8mt_build_intra_predictors_mbuv_s(pbi, xd, mb_row, mb_col);
- vp8mt_build_intra_predictors_mby_s(pbi, xd, mb_row, mb_col);
- }
- else
- {
- vp8_build_inter_predictors_mb_s(xd);
- }
- return;
- }
-
- if (xd->segmentation_enabled)
- mb_init_dequantizer(pbi, xd);
-
- /* do prediction */
- if (xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
- {
- vp8mt_build_intra_predictors_mbuv(pbi, xd, mb_row, mb_col);
-
- if (xd->mode_info_context->mbmi.mode != B_PRED)
- {
- vp8mt_build_intra_predictors_mby(pbi, xd, mb_row, mb_col);
- } else {
- vp8mt_intra_prediction_down_copy(pbi, xd, mb_row, mb_col);
- }
- }
- else
- {
- vp8_build_inter_predictors_mb(xd);
- }
-
- /* dequantization and idct */
- if (xd->mode_info_context->mbmi.mode != B_PRED && xd->mode_info_context->mbmi.mode != SPLITMV)
- {
- BLOCKD *b = &xd->block[24];
- DEQUANT_INVOKE(&pbi->dequant, block)(b);
-
- /* do 2nd order transform on the dc block */
- if (xd->eobs[24] > 1)
- {
- IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh16)(&b->dqcoeff[0], b->diff);
- ((int *)b->qcoeff)[0] = 0;
- ((int *)b->qcoeff)[1] = 0;
- ((int *)b->qcoeff)[2] = 0;
- ((int *)b->qcoeff)[3] = 0;
- ((int *)b->qcoeff)[4] = 0;
- ((int *)b->qcoeff)[5] = 0;
- ((int *)b->qcoeff)[6] = 0;
- ((int *)b->qcoeff)[7] = 0;
- }
- else
- {
- IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh1)(&b->dqcoeff[0], b->diff);
- ((int *)b->qcoeff)[0] = 0;
- }
-
- DEQUANT_INVOKE (&pbi->dequant, dc_idct_add_y_block)
- (xd->qcoeff, xd->block[0].dequant,
- xd->predictor, xd->dst.y_buffer,
- xd->dst.y_stride, xd->eobs, xd->block[24].diff);
- }
- else if ((xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) && xd->mode_info_context->mbmi.mode == B_PRED)
- {
- for (i = 0; i < 16; i++)
- {
- BLOCKD *b = &xd->block[i];
- vp8mt_predict_intra4x4(pbi, xd, b->bmi.mode, b->predictor, mb_row, mb_col, i);
-
- if (xd->eobs[i] > 1)
- {
- DEQUANT_INVOKE(&pbi->dequant, idct_add)
- (b->qcoeff, b->dequant, b->predictor,
- *(b->base_dst) + b->dst, 16, b->dst_stride);
- }
- else
- {
- IDCT_INVOKE(RTCD_VTABLE(idct), idct1_scalar_add)
- (b->qcoeff[0] * b->dequant[0], b->predictor,
- *(b->base_dst) + b->dst, 16, b->dst_stride);
- ((int *)b->qcoeff)[0] = 0;
- }
- }
- }
- else
- {
- DEQUANT_INVOKE (&pbi->dequant, idct_add_y_block)
- (xd->qcoeff, xd->block[0].dequant,
- xd->predictor, xd->dst.y_buffer,
- xd->dst.y_stride, xd->eobs);
- }
-
- DEQUANT_INVOKE (&pbi->dequant, idct_add_uv_block)
- (xd->qcoeff+16*16, xd->block[16].dequant,
- xd->predictor+16*16, xd->dst.u_buffer, xd->dst.v_buffer,
- xd->dst.uv_stride, xd->eobs+16);
-}
-
-
-static THREAD_FUNCTION thread_decoding_proc(void *p_data)
-{
- int ithread = ((DECODETHREAD_DATA *)p_data)->ithread;
- VP8D_COMP *pbi = (VP8D_COMP *)(((DECODETHREAD_DATA *)p_data)->ptr1);
- MB_ROW_DEC *mbrd = (MB_ROW_DEC *)(((DECODETHREAD_DATA *)p_data)->ptr2);
- ENTROPY_CONTEXT_PLANES mb_row_left_context;
-
- while (1)
- {
- if (pbi->b_multithreaded_rd == 0)
- break;
-
- /*if(WaitForSingleObject(pbi->h_event_start_decoding[ithread], INFINITE) == WAIT_OBJECT_0)*/
- if (sem_wait(&pbi->h_event_start_decoding[ithread]) == 0)
- {
- if (pbi->b_multithreaded_rd == 0)
- break;
- else
- {
- VP8_COMMON *pc = &pbi->common;
- MACROBLOCKD *xd = &mbrd->mbd;
-
- int mb_row;
- int num_part = 1 << pbi->common.multi_token_partition;
- volatile int *last_row_current_mb_col;
- int nsync = pbi->sync_range;
-
- for (mb_row = ithread+1; mb_row < pc->mb_rows; mb_row += (pbi->decoding_thread_count + 1))
- {
- int i;
- int recon_yoffset, recon_uvoffset;
- int mb_col;
- int ref_fb_idx = pc->lst_fb_idx;
- int dst_fb_idx = pc->new_fb_idx;
- int recon_y_stride = pc->yv12_fb[ref_fb_idx].y_stride;
- int recon_uv_stride = pc->yv12_fb[ref_fb_idx].uv_stride;
-
- int filter_level;
- loop_filter_info *lfi = pc->lf_info;
- int alt_flt_enabled = xd->segmentation_enabled;
- int Segment;
-
- pbi->mb_row_di[ithread].mb_row = mb_row;
- pbi->mb_row_di[ithread].mbd.current_bc = &pbi->mbc[mb_row%num_part];
-
- last_row_current_mb_col = &pbi->mt_current_mb_col[mb_row -1];
-
- recon_yoffset = mb_row * recon_y_stride * 16;
- recon_uvoffset = mb_row * recon_uv_stride * 8;
- /* reset above block coeffs */
-
- xd->above_context = pc->above_context;
- xd->left_context = &mb_row_left_context;
- vpx_memset(&mb_row_left_context, 0, sizeof(mb_row_left_context));
- xd->up_available = (mb_row != 0);
-
- xd->mb_to_top_edge = -((mb_row * 16)) << 3;
- xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3;
-
- for (mb_col = 0; mb_col < pc->mb_cols; mb_col++)
- {
- if ((mb_col & (nsync-1)) == 0)
- {
- while (mb_col > (*last_row_current_mb_col - nsync) && *last_row_current_mb_col != pc->mb_cols - 1)
- {
- x86_pause_hint();
- thread_sleep(0);
- }
- }
-
- if (xd->mode_info_context->mbmi.mode == SPLITMV || xd->mode_info_context->mbmi.mode == B_PRED)
- {
- for (i = 0; i < 16; i++)
- {
- BLOCKD *d = &xd->block[i];
- vpx_memcpy(&d->bmi, &xd->mode_info_context->bmi[i], sizeof(B_MODE_INFO));
- }
- }
-
- /* Distance of Mb to the various image edges.
- * These are specified to 8th pel as they are always compared to values that are in 1/8th pel units
- */
- xd->mb_to_left_edge = -((mb_col * 16) << 3);
- xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3;
-
- xd->dst.y_buffer = pc->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset;
- xd->dst.u_buffer = pc->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset;
- xd->dst.v_buffer = pc->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset;
-
- xd->left_available = (mb_col != 0);
-
- /* Select the appropriate reference frame for this MB */
- if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME)
- ref_fb_idx = pc->lst_fb_idx;
- else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME)
- ref_fb_idx = pc->gld_fb_idx;
- else
- ref_fb_idx = pc->alt_fb_idx;
-
- xd->pre.y_buffer = pc->yv12_fb[ref_fb_idx].y_buffer + recon_yoffset;
- xd->pre.u_buffer = pc->yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset;
- xd->pre.v_buffer = pc->yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset;
-
- vp8_build_uvmvs(xd, pc->full_pixel);
- decode_macroblock(pbi, xd, mb_row, mb_col);
-
- if (pbi->common.filter_level)
- {
- if( mb_row != pc->mb_rows-1 )
- {
- /* Save decoded MB last row data for next-row decoding */
- vpx_memcpy((pbi->mt_yabove_row[mb_row + 1] + 32 + mb_col*16), (xd->dst.y_buffer + 15 * recon_y_stride), 16);
- vpx_memcpy((pbi->mt_uabove_row[mb_row + 1] + 16 + mb_col*8), (xd->dst.u_buffer + 7 * recon_uv_stride), 8);
- vpx_memcpy((pbi->mt_vabove_row[mb_row + 1] + 16 + mb_col*8), (xd->dst.v_buffer + 7 * recon_uv_stride), 8);
- }
-
- /* save left_col for next MB decoding */
- if(mb_col != pc->mb_cols-1)
- {
- MODE_INFO *next = xd->mode_info_context +1;
-
- if (xd->frame_type == KEY_FRAME || next->mbmi.ref_frame == INTRA_FRAME)
- {
- for (i = 0; i < 16; i++)
- pbi->mt_yleft_col[mb_row][i] = xd->dst.y_buffer [i* recon_y_stride + 15];
- for (i = 0; i < 8; i++)
- {
- pbi->mt_uleft_col[mb_row][i] = xd->dst.u_buffer [i* recon_uv_stride + 7];
- pbi->mt_vleft_col[mb_row][i] = xd->dst.v_buffer [i* recon_uv_stride + 7];
- }
- }
- }
-
- /* update loopfilter info */
- Segment = (alt_flt_enabled) ? xd->mode_info_context->mbmi.segment_id : 0;
- filter_level = pbi->mt_baseline_filter_level[Segment];
- /* Distance of Mb to the various image edges.
- * These are specified to 8th pel as they are always compared to values that are in 1/8th pel units
- * Apply any context driven MB level adjustment
- */
- filter_level = vp8_adjust_mb_lf_value(xd, filter_level);
-
- /* loopfilter on this macroblock. */
- if (filter_level)
- {
- if (mb_col > 0)
- pc->lf_mbv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
-
- if (xd->mode_info_context->mbmi.dc_diff > 0)
- pc->lf_bv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
-
- /* don't apply across umv border */
- if (mb_row > 0)
- pc->lf_mbh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
-
- if (xd->mode_info_context->mbmi.dc_diff > 0)
- pc->lf_bh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
- }
- }
-
- recon_yoffset += 16;
- recon_uvoffset += 8;
-
- ++xd->mode_info_context; /* next mb */
-
- xd->above_context++;
-
- /*pbi->mb_row_di[ithread].current_mb_col = mb_col;*/
- pbi->mt_current_mb_col[mb_row] = mb_col;
- }
-
- /* adjust to the next row of mbs */
- if (pbi->common.filter_level)
- {
- if(mb_row != pc->mb_rows-1)
- {
- int lasty = pc->yv12_fb[ref_fb_idx].y_width + VP8BORDERINPIXELS;
- int lastuv = (pc->yv12_fb[ref_fb_idx].y_width>>1) + (VP8BORDERINPIXELS>>1);
-
- for (i = 0; i < 4; i++)
- {
- pbi->mt_yabove_row[mb_row +1][lasty + i] = pbi->mt_yabove_row[mb_row +1][lasty -1];
- pbi->mt_uabove_row[mb_row +1][lastuv + i] = pbi->mt_uabove_row[mb_row +1][lastuv -1];
- pbi->mt_vabove_row[mb_row +1][lastuv + i] = pbi->mt_vabove_row[mb_row +1][lastuv -1];
- }
- }
- } else
- vp8_extend_mb_row(&pc->yv12_fb[dst_fb_idx], xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8);
-
- ++xd->mode_info_context; /* skip prediction column */
-
- /* since we have multithread */
- xd->mode_info_context += xd->mode_info_stride * pbi->decoding_thread_count;
- }
- }
- }
- /* add this to each frame */
- if ((mbrd->mb_row == pbi->common.mb_rows-1) || ((mbrd->mb_row == pbi->common.mb_rows-2) && (pbi->common.mb_rows % (pbi->decoding_thread_count+1))==1))
- {
- /*SetEvent(pbi->h_event_end_decoding);*/
- sem_post(&pbi->h_event_end_decoding);
- }
- }
-
- return 0 ;
-}
-
-
-void vp8_decoder_create_threads(VP8D_COMP *pbi)
-{
- int core_count = 0;
- int ithread;
-
- pbi->b_multithreaded_rd = 0;
- pbi->allocated_decoding_thread_count = 0;
- core_count = (pbi->max_threads > 16) ? 16 : pbi->max_threads;
-
- if (core_count > 1)
- {
- pbi->b_multithreaded_rd = 1;
- pbi->decoding_thread_count = core_count -1;
-
- CHECK_MEM_ERROR(pbi->h_decoding_thread, vpx_malloc(sizeof(pthread_t) * pbi->decoding_thread_count));
- CHECK_MEM_ERROR(pbi->h_event_start_decoding, vpx_malloc(sizeof(sem_t) * pbi->decoding_thread_count));
- CHECK_MEM_ERROR(pbi->mb_row_di, vpx_memalign(32, sizeof(MB_ROW_DEC) * pbi->decoding_thread_count));
- vpx_memset(pbi->mb_row_di, 0, sizeof(MB_ROW_DEC) * pbi->decoding_thread_count);
- CHECK_MEM_ERROR(pbi->de_thread_data, vpx_malloc(sizeof(DECODETHREAD_DATA) * pbi->decoding_thread_count));
-
- for (ithread = 0; ithread < pbi->decoding_thread_count; ithread++)
- {
- sem_init(&pbi->h_event_start_decoding[ithread], 0, 0);
-
- pbi->de_thread_data[ithread].ithread = ithread;
- pbi->de_thread_data[ithread].ptr1 = (void *)pbi;
- pbi->de_thread_data[ithread].ptr2 = (void *) &pbi->mb_row_di[ithread];
-
- pthread_create(&pbi->h_decoding_thread[ithread], 0, thread_decoding_proc, (&pbi->de_thread_data[ithread]));
- }
-
- sem_init(&pbi->h_event_end_decoding, 0, 0);
-
- pbi->allocated_decoding_thread_count = pbi->decoding_thread_count;
- }
-}
-
-
-void vp8mt_de_alloc_temp_buffers(VP8D_COMP *pbi, int mb_rows)
-{
- VP8_COMMON *const pc = & pbi->common;
- int i;
-
- if (pbi->b_multithreaded_rd)
- {
- vpx_free(pbi->mt_current_mb_col);
- pbi->mt_current_mb_col = NULL ;
-
- /* Free above_row buffers. */
- if (pbi->mt_yabove_row)
- {
- for (i=0; i< mb_rows; i++)
- {
- vpx_free(pbi->mt_yabove_row[i]);
- pbi->mt_yabove_row[i] = NULL ;
- }
- vpx_free(pbi->mt_yabove_row);
- pbi->mt_yabove_row = NULL ;
- }
-
- if (pbi->mt_uabove_row)
- {
- for (i=0; i< mb_rows; i++)
- {
- vpx_free(pbi->mt_uabove_row[i]);
- pbi->mt_uabove_row[i] = NULL ;
- }
- vpx_free(pbi->mt_uabove_row);
- pbi->mt_uabove_row = NULL ;
- }
-
- if (pbi->mt_vabove_row)
- {
- for (i=0; i< mb_rows; i++)
- {
- vpx_free(pbi->mt_vabove_row[i]);
- pbi->mt_vabove_row[i] = NULL ;
- }
- vpx_free(pbi->mt_vabove_row);
- pbi->mt_vabove_row = NULL ;
- }
-
- /* Free left_col buffers. */
- if (pbi->mt_yleft_col)
- {
- for (i=0; i< mb_rows; i++)
- {
- vpx_free(pbi->mt_yleft_col[i]);
- pbi->mt_yleft_col[i] = NULL ;
- }
- vpx_free(pbi->mt_yleft_col);
- pbi->mt_yleft_col = NULL ;
- }
-
- if (pbi->mt_uleft_col)
- {
- for (i=0; i< mb_rows; i++)
- {
- vpx_free(pbi->mt_uleft_col[i]);
- pbi->mt_uleft_col[i] = NULL ;
- }
- vpx_free(pbi->mt_uleft_col);
- pbi->mt_uleft_col = NULL ;
- }
-
- if (pbi->mt_vleft_col)
- {
- for (i=0; i< mb_rows; i++)
- {
- vpx_free(pbi->mt_vleft_col[i]);
- pbi->mt_vleft_col[i] = NULL ;
- }
- vpx_free(pbi->mt_vleft_col);
- pbi->mt_vleft_col = NULL ;
- }
- }
-}
-
-
-void vp8mt_alloc_temp_buffers(VP8D_COMP *pbi, int width, int prev_mb_rows)
-{
- VP8_COMMON *const pc = & pbi->common;
- int i;
- int uv_width;
-
- if (pbi->b_multithreaded_rd)
- {
- vp8mt_de_alloc_temp_buffers(pbi, prev_mb_rows);
-
- /* our internal buffers are always multiples of 16 */
- if ((width & 0xf) != 0)
- width += 16 - (width & 0xf);
-
- if (width < 640) pbi->sync_range = 1;
- else if (width <= 1280) pbi->sync_range = 8;
- else if (width <= 2560) pbi->sync_range =16;
- else pbi->sync_range = 32;
-
- uv_width = width >>1;
-
- /* Allocate an int for each mb row. */
- CHECK_MEM_ERROR(pbi->mt_current_mb_col, vpx_malloc(sizeof(int) * pc->mb_rows));
-
- /* Allocate memory for above_row buffers. */
- CHECK_MEM_ERROR(pbi->mt_yabove_row, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
- CHECK_MEM_ERROR(pbi->mt_yabove_row[i], vpx_calloc(sizeof(unsigned char) * (width + (VP8BORDERINPIXELS<<1)), 1));
-
- CHECK_MEM_ERROR(pbi->mt_uabove_row, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
- CHECK_MEM_ERROR(pbi->mt_uabove_row[i], vpx_calloc(sizeof(unsigned char) * (uv_width + VP8BORDERINPIXELS), 1));
-
- CHECK_MEM_ERROR(pbi->mt_vabove_row, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
- CHECK_MEM_ERROR(pbi->mt_vabove_row[i], vpx_calloc(sizeof(unsigned char) * (uv_width + VP8BORDERINPIXELS), 1));
-
- /* Allocate memory for left_col buffers. */
- CHECK_MEM_ERROR(pbi->mt_yleft_col, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
- CHECK_MEM_ERROR(pbi->mt_yleft_col[i], vpx_calloc(sizeof(unsigned char) * 16, 1));
-
- CHECK_MEM_ERROR(pbi->mt_uleft_col, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
- CHECK_MEM_ERROR(pbi->mt_uleft_col[i], vpx_calloc(sizeof(unsigned char) * 8, 1));
-
- CHECK_MEM_ERROR(pbi->mt_vleft_col, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
- CHECK_MEM_ERROR(pbi->mt_vleft_col[i], vpx_calloc(sizeof(unsigned char) * 8, 1));
- }
-}
-
-
-void vp8_decoder_remove_threads(VP8D_COMP *pbi)
-{
- /* shutdown MB Decoding thread; */
- if (pbi->b_multithreaded_rd)
- {
- int i;
-
- pbi->b_multithreaded_rd = 0;
-
- /* allow all threads to exit */
- for (i = 0; i < pbi->allocated_decoding_thread_count; i++)
- {
- sem_post(&pbi->h_event_start_decoding[i]);
- pthread_join(pbi->h_decoding_thread[i], NULL);
- }
-
- for (i = 0; i < pbi->allocated_decoding_thread_count; i++)
- {
- sem_destroy(&pbi->h_event_start_decoding[i]);
- }
-
- sem_destroy(&pbi->h_event_end_decoding);
-
- vpx_free(pbi->h_decoding_thread);
- pbi->h_decoding_thread = NULL;
-
- vpx_free(pbi->h_event_start_decoding);
- pbi->h_event_start_decoding = NULL;
-
- vpx_free(pbi->mb_row_di);
- pbi->mb_row_di = NULL ;
-
- vpx_free(pbi->de_thread_data);
- pbi->de_thread_data = NULL;
- }
-}
-
-
-static void lpf_init( VP8D_COMP *pbi, int default_filt_lvl)
-{
- VP8_COMMON *cm = &pbi->common;
- MACROBLOCKD *mbd = &pbi->mb;
- /*YV12_BUFFER_CONFIG *post = &cm->new_frame;*/ /*frame_to_show;*/
- loop_filter_info *lfi = cm->lf_info;
- FRAME_TYPE frame_type = cm->frame_type;
-
- /*int mb_row;
- int mb_col;
- int baseline_filter_level[MAX_MB_SEGMENTS];*/
- int alt_flt_enabled = mbd->segmentation_enabled;
-
- int i;
- /*unsigned char *y_ptr, *u_ptr, *v_ptr;*/
-
- /* Note the baseline filter values for each segment */
- if (alt_flt_enabled)
- {
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- {
- /* Abs value */
- if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
- pbi->mt_baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i];
- /* Delta Value */
- else
- {
- pbi->mt_baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i];
- pbi->mt_baseline_filter_level[i] = (pbi->mt_baseline_filter_level[i] >= 0) ? ((pbi->mt_baseline_filter_level[i] <= MAX_LOOP_FILTER) ? pbi->mt_baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; /* Clamp to valid range */
- }
- }
- }
- else
- {
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- pbi->mt_baseline_filter_level[i] = default_filt_lvl;
- }
-
- /* Initialize the loop filter for this frame. */
- if ((cm->last_filter_type != cm->filter_type) || (cm->last_sharpness_level != cm->sharpness_level))
- vp8_init_loop_filter(cm);
- else if (frame_type != cm->last_frame_type)
- vp8_frame_init_loop_filter(lfi, frame_type);
-}
-
-
-void vp8mt_decode_mb_rows( VP8D_COMP *pbi, MACROBLOCKD *xd)
-{
- int mb_row;
- VP8_COMMON *pc = &pbi->common;
-
- int ibc = 0;
- int num_part = 1 << pbi->common.multi_token_partition;
- int i;
- volatile int *last_row_current_mb_col = NULL;
- int nsync = pbi->sync_range;
-
- int filter_level;
- loop_filter_info *lfi = pc->lf_info;
- int alt_flt_enabled = xd->segmentation_enabled;
- int Segment;
-
- if(pbi->common.filter_level)
- {
- /* Set above_row buffer to 127 for decoding first MB row */
- vpx_memset(pbi->mt_yabove_row[0] + VP8BORDERINPIXELS-1, 127, pc->yv12_fb[pc->lst_fb_idx].y_width + 5);
- vpx_memset(pbi->mt_uabove_row[0] + (VP8BORDERINPIXELS>>1)-1, 127, (pc->yv12_fb[pc->lst_fb_idx].y_width>>1) +5);
- vpx_memset(pbi->mt_vabove_row[0] + (VP8BORDERINPIXELS>>1)-1, 127, (pc->yv12_fb[pc->lst_fb_idx].y_width>>1) +5);
-
- for (i=1; i<pc->mb_rows; i++)
- {
- vpx_memset(pbi->mt_yabove_row[i] + VP8BORDERINPIXELS-1, (unsigned char)129, 1);
- vpx_memset(pbi->mt_uabove_row[i] + (VP8BORDERINPIXELS>>1)-1, (unsigned char)129, 1);
- vpx_memset(pbi->mt_vabove_row[i] + (VP8BORDERINPIXELS>>1)-1, (unsigned char)129, 1);
- }
-
- /* Set left_col to 129 initially */
- for (i=0; i<pc->mb_rows; i++)
- {
- vpx_memset(pbi->mt_yleft_col[i], (unsigned char)129, 16);
- vpx_memset(pbi->mt_uleft_col[i], (unsigned char)129, 8);
- vpx_memset(pbi->mt_vleft_col[i], (unsigned char)129, 8);
- }
- lpf_init(pbi, pc->filter_level);
- }
-
- setup_decoding_thread_data(pbi, xd, pbi->mb_row_di, pbi->decoding_thread_count);
-
- for (i = 0; i < pbi->decoding_thread_count; i++)
- sem_post(&pbi->h_event_start_decoding[i]);
-
- for (mb_row = 0; mb_row < pc->mb_rows; mb_row += (pbi->decoding_thread_count + 1))
- {
-
- xd->current_bc = &pbi->mbc[mb_row%num_part];
-
- /* vp8_decode_mb_row(pbi, pc, mb_row, xd); */
- {
- int i;
- int recon_yoffset, recon_uvoffset;
- int mb_col;
- int ref_fb_idx = pc->lst_fb_idx;
- int dst_fb_idx = pc->new_fb_idx;
- int recon_y_stride = pc->yv12_fb[ref_fb_idx].y_stride;
- int recon_uv_stride = pc->yv12_fb[ref_fb_idx].uv_stride;
-
- /* volatile int *last_row_current_mb_col = NULL; */
- if (mb_row > 0)
- last_row_current_mb_col = &pbi->mt_current_mb_col[mb_row -1];
-
- vpx_memset(&pc->left_context, 0, sizeof(pc->left_context));
- recon_yoffset = mb_row * recon_y_stride * 16;
- recon_uvoffset = mb_row * recon_uv_stride * 8;
- /* reset above block coeffs */
-
- xd->above_context = pc->above_context;
- xd->up_available = (mb_row != 0);
-
- xd->mb_to_top_edge = -((mb_row * 16)) << 3;
- xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3;
-
- for (mb_col = 0; mb_col < pc->mb_cols; mb_col++)
- {
- if ( mb_row > 0 && (mb_col & (nsync-1)) == 0){
- while (mb_col > (*last_row_current_mb_col - nsync) && *last_row_current_mb_col != pc->mb_cols - 1)
- {
- x86_pause_hint();
- thread_sleep(0);
- }
- }
-
- if (xd->mode_info_context->mbmi.mode == SPLITMV || xd->mode_info_context->mbmi.mode == B_PRED)
- {
- for (i = 0; i < 16; i++)
- {
- BLOCKD *d = &xd->block[i];
- vpx_memcpy(&d->bmi, &xd->mode_info_context->bmi[i], sizeof(B_MODE_INFO));
- }
- }
-
- /* Distance of Mb to the various image edges.
- * These are specified to 8th pel as they are always compared to values that are in 1/8th pel units
- */
- xd->mb_to_left_edge = -((mb_col * 16) << 3);
- xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3;
-
- xd->dst.y_buffer = pc->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset;
- xd->dst.u_buffer = pc->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset;
- xd->dst.v_buffer = pc->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset;
-
- xd->left_available = (mb_col != 0);
-
- /* Select the appropriate reference frame for this MB */
- if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME)
- ref_fb_idx = pc->lst_fb_idx;
- else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME)
- ref_fb_idx = pc->gld_fb_idx;
- else
- ref_fb_idx = pc->alt_fb_idx;
-
- xd->pre.y_buffer = pc->yv12_fb[ref_fb_idx].y_buffer + recon_yoffset;
- xd->pre.u_buffer = pc->yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset;
- xd->pre.v_buffer = pc->yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset;
-
- if (xd->mode_info_context->mbmi.ref_frame != INTRA_FRAME)
- {
- /* propagate errors from reference frames */
- xd->corrupted |= pc->yv12_fb[ref_fb_idx].corrupted;
- }
-
- vp8_build_uvmvs(xd, pc->full_pixel);
- decode_macroblock(pbi, xd, mb_row, mb_col);
-
- /* check if the boolean decoder has suffered an error */
- xd->corrupted |= vp8dx_bool_error(xd->current_bc);
-
- if (pbi->common.filter_level)
- {
- /* Save decoded MB last row data for next-row decoding */
- if(mb_row != pc->mb_rows-1)
- {
- vpx_memcpy((pbi->mt_yabove_row[mb_row +1] + 32 + mb_col*16), (xd->dst.y_buffer + 15 * recon_y_stride), 16);
- vpx_memcpy((pbi->mt_uabove_row[mb_row +1] + 16 + mb_col*8), (xd->dst.u_buffer + 7 * recon_uv_stride), 8);
- vpx_memcpy((pbi->mt_vabove_row[mb_row +1] + 16 + mb_col*8), (xd->dst.v_buffer + 7 * recon_uv_stride), 8);
- }
-
- /* save left_col for next MB decoding */
- if(mb_col != pc->mb_cols-1)
- {
- MODE_INFO *next = xd->mode_info_context +1;
-
- if (xd->frame_type == KEY_FRAME || next->mbmi.ref_frame == INTRA_FRAME)
- {
- for (i = 0; i < 16; i++)
- pbi->mt_yleft_col[mb_row][i] = xd->dst.y_buffer [i* recon_y_stride + 15];
- for (i = 0; i < 8; i++)
- {
- pbi->mt_uleft_col[mb_row][i] = xd->dst.u_buffer [i* recon_uv_stride + 7];
- pbi->mt_vleft_col[mb_row][i] = xd->dst.v_buffer [i* recon_uv_stride + 7];
- }
- }
- }
-
- /* update loopfilter info */
- Segment = (alt_flt_enabled) ? xd->mode_info_context->mbmi.segment_id : 0;
- filter_level = pbi->mt_baseline_filter_level[Segment];
- /* Distance of Mb to the various image edges.
- * These are specified to 8th pel as they are always compared to values that are in 1/8th pel units
- * Apply any context driven MB level adjustment
- */
- filter_level = vp8_adjust_mb_lf_value(xd, filter_level);
-
- /* loopfilter on this macroblock. */
- if (filter_level)
- {
- if (mb_col > 0)
- pc->lf_mbv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
-
- if (xd->mode_info_context->mbmi.dc_diff > 0)
- pc->lf_bv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
-
- /* don't apply across umv border */
- if (mb_row > 0)
- pc->lf_mbh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
-
- if (xd->mode_info_context->mbmi.dc_diff > 0)
- pc->lf_bh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
- }
- }
-
- recon_yoffset += 16;
- recon_uvoffset += 8;
-
- ++xd->mode_info_context; /* next mb */
-
- xd->above_context++;
-
- pbi->mt_current_mb_col[mb_row] = mb_col;
- }
-
- /* adjust to the next row of mbs */
- if (pbi->common.filter_level)
- {
- if(mb_row != pc->mb_rows-1)
- {
- int lasty = pc->yv12_fb[ref_fb_idx].y_width + VP8BORDERINPIXELS;
- int lastuv = (pc->yv12_fb[ref_fb_idx].y_width>>1) + (VP8BORDERINPIXELS>>1);
-
- for (i = 0; i < 4; i++)
- {
- pbi->mt_yabove_row[mb_row +1][lasty + i] = pbi->mt_yabove_row[mb_row +1][lasty -1];
- pbi->mt_uabove_row[mb_row +1][lastuv + i] = pbi->mt_uabove_row[mb_row +1][lastuv -1];
- pbi->mt_vabove_row[mb_row +1][lastuv + i] = pbi->mt_vabove_row[mb_row +1][lastuv -1];
- }
- }
- }else
- vp8_extend_mb_row(&pc->yv12_fb[dst_fb_idx], xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8);
-
- ++xd->mode_info_context; /* skip prediction column */
- }
- xd->mode_info_context += xd->mode_info_stride * pbi->decoding_thread_count;
- }
-
- sem_wait(&pbi->h_event_end_decoding); /* add back for each frame */
-}
diff --git a/vp8/decoder/treereader.h b/vp8/decoder/treereader.h
deleted file mode 100644
index b50a4d2..0000000
--- a/vp8/decoder/treereader.h
+++ /dev/null
@@ -1,64 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef tree_reader_h
-#define tree_reader_h 1
-
-#include "vp8/common/treecoder.h"
-
-#include "dboolhuff.h"
-
-typedef BOOL_DECODER vp8_reader;
-
-#define vp8_read vp8dx_decode_bool
-#define vp8_read_literal vp8_decode_value
-#define vp8_read_bit( R) vp8_read( R, vp8_prob_half)
-
-
-/* Intent of tree data structure is to make decoding trivial. */
-
-static int vp8_treed_read(
- vp8_reader *const r, /* !!! must return a 0 or 1 !!! */
- vp8_tree t,
- const vp8_prob *const p
-)
-{
- register vp8_tree_index i = 0;
-
- while ((i = t[ i + vp8_read(r, p[i>>1])]) > 0) ;
-
- return -i;
-}
-
-
-/* Variant reads a binary number given distributions on each bit.
- Note that tree is arbitrary; probability of decoding a zero
- may or may not depend on previously decoded bits. */
-
-static int vp8_treed_read_num(
- vp8_reader *const r, /* !!! must return a 0 or 1 !!! */
- vp8_tree t,
- const vp8_prob *const p
-)
-{
- vp8_tree_index i = 0;
- int v = 0, b;
-
- do
- {
- b = vp8_read(r, p[i>>1]);
- v = (v << 1) + b;
- }
- while ((i = t[i+b]) > 0);
-
- return v;
-}
-#endif /* tree_reader_h */
diff --git a/vp8/decoder/x86/dequantize_mmx.asm b/vp8/decoder/x86/dequantize_mmx.asm
deleted file mode 100644
index 0d6133a..0000000
--- a/vp8/decoder/x86/dequantize_mmx.asm
+++ /dev/null
@@ -1,461 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-
-;void vp8_dequantize_b_impl_mmx(short *sq, short *dq, short *q)
-global sym(vp8_dequantize_b_impl_mmx)
-sym(vp8_dequantize_b_impl_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 3
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;sq
- mov rdi, arg(1) ;dq
- mov rax, arg(2) ;q
-
- movq mm1, [rsi]
- pmullw mm1, [rax+0] ; mm4 *= kernel 0 modifiers.
- movq [rdi], mm1
-
- movq mm1, [rsi+8]
- pmullw mm1, [rax+8] ; mm4 *= kernel 0 modifiers.
- movq [rdi+8], mm1
-
- movq mm1, [rsi+16]
- pmullw mm1, [rax+16] ; mm4 *= kernel 0 modifiers.
- movq [rdi+16], mm1
-
- movq mm1, [rsi+24]
- pmullw mm1, [rax+24] ; mm4 *= kernel 0 modifiers.
- movq [rdi+24], mm1
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void dequant_idct_add_mmx(short *input, short *dq, unsigned char *pred, unsigned char *dest, int pitch, int stride)
-global sym(vp8_dequant_idct_add_mmx)
-sym(vp8_dequant_idct_add_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rax, arg(0) ;input
- mov rdx, arg(1) ;dq
-
-
- movq mm0, [rax ]
- pmullw mm0, [rdx]
-
- movq mm1, [rax +8]
- pmullw mm1, [rdx +8]
-
- movq mm2, [rax+16]
- pmullw mm2, [rdx+16]
-
- movq mm3, [rax+24]
- pmullw mm3, [rdx+24]
-
- mov rdx, arg(3) ;dest
- mov rsi, arg(2) ;pred
- pxor mm7, mm7
-
-
- movq [rax], mm7
- movq [rax+8], mm7
-
- movq [rax+16],mm7
- movq [rax+24],mm7
-
-
- movsxd rax, dword ptr arg(4) ;pitch
- movsxd rdi, dword ptr arg(5) ;stride
-
- psubw mm0, mm2 ; b1= 0-2
- paddw mm2, mm2 ;
-
- movq mm5, mm1
- paddw mm2, mm0 ; a1 =0+2
-
- pmulhw mm5, [GLOBAL(x_s1sqr2)];
- paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movq mm7, mm3 ;
- pmulhw mm7, [GLOBAL(x_c1sqr2less1)];
-
- paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw mm7, mm5 ; c1
-
- movq mm5, mm1
- movq mm4, mm3
-
- pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
- paddw mm5, mm1
-
- pmulhw mm3, [GLOBAL(x_s1sqr2)]
- paddw mm3, mm4
-
- paddw mm3, mm5 ; d1
- movq mm6, mm2 ; a1
-
- movq mm4, mm0 ; b1
- paddw mm2, mm3 ;0
-
- paddw mm4, mm7 ;1
- psubw mm0, mm7 ;2
-
- psubw mm6, mm3 ;3
-
- movq mm1, mm2 ; 03 02 01 00
- movq mm3, mm4 ; 23 22 21 20
-
- punpcklwd mm1, mm0 ; 11 01 10 00
- punpckhwd mm2, mm0 ; 13 03 12 02
-
- punpcklwd mm3, mm6 ; 31 21 30 20
- punpckhwd mm4, mm6 ; 33 23 32 22
-
- movq mm0, mm1 ; 11 01 10 00
- movq mm5, mm2 ; 13 03 12 02
-
- punpckldq mm0, mm3 ; 30 20 10 00
- punpckhdq mm1, mm3 ; 31 21 11 01
-
- punpckldq mm2, mm4 ; 32 22 12 02
- punpckhdq mm5, mm4 ; 33 23 13 03
-
- movq mm3, mm5 ; 33 23 13 03
-
- psubw mm0, mm2 ; b1= 0-2
- paddw mm2, mm2 ;
-
- movq mm5, mm1
- paddw mm2, mm0 ; a1 =0+2
-
- pmulhw mm5, [GLOBAL(x_s1sqr2)];
- paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movq mm7, mm3 ;
- pmulhw mm7, [GLOBAL(x_c1sqr2less1)];
-
- paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw mm7, mm5 ; c1
-
- movq mm5, mm1
- movq mm4, mm3
-
- pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
- paddw mm5, mm1
-
- pmulhw mm3, [GLOBAL(x_s1sqr2)]
- paddw mm3, mm4
-
- paddw mm3, mm5 ; d1
- paddw mm0, [GLOBAL(fours)]
-
- paddw mm2, [GLOBAL(fours)]
- movq mm6, mm2 ; a1
-
- movq mm4, mm0 ; b1
- paddw mm2, mm3 ;0
-
- paddw mm4, mm7 ;1
- psubw mm0, mm7 ;2
-
- psubw mm6, mm3 ;3
- psraw mm2, 3
-
- psraw mm0, 3
- psraw mm4, 3
-
- psraw mm6, 3
-
- movq mm1, mm2 ; 03 02 01 00
- movq mm3, mm4 ; 23 22 21 20
-
- punpcklwd mm1, mm0 ; 11 01 10 00
- punpckhwd mm2, mm0 ; 13 03 12 02
-
- punpcklwd mm3, mm6 ; 31 21 30 20
- punpckhwd mm4, mm6 ; 33 23 32 22
-
- movq mm0, mm1 ; 11 01 10 00
- movq mm5, mm2 ; 13 03 12 02
-
- punpckldq mm0, mm3 ; 30 20 10 00
- punpckhdq mm1, mm3 ; 31 21 11 01
-
- punpckldq mm2, mm4 ; 32 22 12 02
- punpckhdq mm5, mm4 ; 33 23 13 03
-
- pxor mm7, mm7
-
- movd mm4, [rsi]
- punpcklbw mm4, mm7
- paddsw mm0, mm4
- packuswb mm0, mm7
- movd [rdx], mm0
-
- movd mm4, [rsi+rax]
- punpcklbw mm4, mm7
- paddsw mm1, mm4
- packuswb mm1, mm7
- movd [rdx+rdi], mm1
-
- movd mm4, [rsi+2*rax]
- punpcklbw mm4, mm7
- paddsw mm2, mm4
- packuswb mm2, mm7
- movd [rdx+rdi*2], mm2
-
- add rdx, rdi
- add rsi, rax
-
- movd mm4, [rsi+2*rax]
- punpcklbw mm4, mm7
- paddsw mm5, mm4
- packuswb mm5, mm7
- movd [rdx+rdi*2], mm5
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void dequant_dc_idct_add_mmx(short *input, short *dq, unsigned char *pred, unsigned char *dest, int pitch, int stride, int Dc)
-global sym(vp8_dequant_dc_idct_add_mmx)
-sym(vp8_dequant_dc_idct_add_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rax, arg(0) ;input
- mov rdx, arg(1) ;dq
-
- movq mm0, [rax ]
- pmullw mm0, [rdx]
-
- movq mm1, [rax +8]
- pmullw mm1, [rdx +8]
-
- movq mm2, [rax+16]
- pmullw mm2, [rdx+16]
-
- movq mm3, [rax+24]
- pmullw mm3, [rdx+24]
-
- mov rdx, arg(3) ;dest
- mov rsi, arg(2) ;pred
- pxor mm7, mm7
-
-
- movq [rax], mm7
- movq [rax+8], mm7
-
- movq [rax+16],mm7
- movq [rax+24],mm7
-
- ; move lower word of Dc to lower word of mm0
- psrlq mm0, 16
- movzx rcx, word ptr arg(6) ;Dc
- psllq mm0, 16
- movq mm7, rcx
- por mm0, mm7
-
- movsxd rax, dword ptr arg(4) ;pitch
- movsxd rdi, dword ptr arg(5) ;stride
-
- psubw mm0, mm2 ; b1= 0-2
- paddw mm2, mm2 ;
-
- movq mm5, mm1
- paddw mm2, mm0 ; a1 =0+2
-
- pmulhw mm5, [GLOBAL(x_s1sqr2)];
- paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movq mm7, mm3 ;
- pmulhw mm7, [GLOBAL(x_c1sqr2less1)];
-
- paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw mm7, mm5 ; c1
-
- movq mm5, mm1
- movq mm4, mm3
-
- pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
- paddw mm5, mm1
-
- pmulhw mm3, [GLOBAL(x_s1sqr2)]
- paddw mm3, mm4
-
- paddw mm3, mm5 ; d1
- movq mm6, mm2 ; a1
-
- movq mm4, mm0 ; b1
- paddw mm2, mm3 ;0
-
- paddw mm4, mm7 ;1
- psubw mm0, mm7 ;2
-
- psubw mm6, mm3 ;3
-
- movq mm1, mm2 ; 03 02 01 00
- movq mm3, mm4 ; 23 22 21 20
-
- punpcklwd mm1, mm0 ; 11 01 10 00
- punpckhwd mm2, mm0 ; 13 03 12 02
-
- punpcklwd mm3, mm6 ; 31 21 30 20
- punpckhwd mm4, mm6 ; 33 23 32 22
-
- movq mm0, mm1 ; 11 01 10 00
- movq mm5, mm2 ; 13 03 12 02
-
- punpckldq mm0, mm3 ; 30 20 10 00
- punpckhdq mm1, mm3 ; 31 21 11 01
-
- punpckldq mm2, mm4 ; 32 22 12 02
- punpckhdq mm5, mm4 ; 33 23 13 03
-
- movq mm3, mm5 ; 33 23 13 03
-
- psubw mm0, mm2 ; b1= 0-2
- paddw mm2, mm2 ;
-
- movq mm5, mm1
- paddw mm2, mm0 ; a1 =0+2
-
- pmulhw mm5, [GLOBAL(x_s1sqr2)];
- paddw mm5, mm1 ; ip1 * sin(pi/8) * sqrt(2)
-
- movq mm7, mm3 ;
- pmulhw mm7, [GLOBAL(x_c1sqr2less1)];
-
- paddw mm7, mm3 ; ip3 * cos(pi/8) * sqrt(2)
- psubw mm7, mm5 ; c1
-
- movq mm5, mm1
- movq mm4, mm3
-
- pmulhw mm5, [GLOBAL(x_c1sqr2less1)]
- paddw mm5, mm1
-
- pmulhw mm3, [GLOBAL(x_s1sqr2)]
- paddw mm3, mm4
-
- paddw mm3, mm5 ; d1
- paddw mm0, [GLOBAL(fours)]
-
- paddw mm2, [GLOBAL(fours)]
- movq mm6, mm2 ; a1
-
- movq mm4, mm0 ; b1
- paddw mm2, mm3 ;0
-
- paddw mm4, mm7 ;1
- psubw mm0, mm7 ;2
-
- psubw mm6, mm3 ;3
- psraw mm2, 3
-
- psraw mm0, 3
- psraw mm4, 3
-
- psraw mm6, 3
-
- movq mm1, mm2 ; 03 02 01 00
- movq mm3, mm4 ; 23 22 21 20
-
- punpcklwd mm1, mm0 ; 11 01 10 00
- punpckhwd mm2, mm0 ; 13 03 12 02
-
- punpcklwd mm3, mm6 ; 31 21 30 20
- punpckhwd mm4, mm6 ; 33 23 32 22
-
- movq mm0, mm1 ; 11 01 10 00
- movq mm5, mm2 ; 13 03 12 02
-
- punpckldq mm0, mm3 ; 30 20 10 00
- punpckhdq mm1, mm3 ; 31 21 11 01
-
- punpckldq mm2, mm4 ; 32 22 12 02
- punpckhdq mm5, mm4 ; 33 23 13 03
-
- pxor mm7, mm7
-
- movd mm4, [rsi]
- punpcklbw mm4, mm7
- paddsw mm0, mm4
- packuswb mm0, mm7
- movd [rdx], mm0
-
- movd mm4, [rsi+rax]
- punpcklbw mm4, mm7
- paddsw mm1, mm4
- packuswb mm1, mm7
- movd [rdx+rdi], mm1
-
- movd mm4, [rsi+2*rax]
- punpcklbw mm4, mm7
- paddsw mm2, mm4
- packuswb mm2, mm7
- movd [rdx+rdi*2], mm2
-
- add rdx, rdi
- add rsi, rax
-
- movd mm4, [rsi+2*rax]
- punpcklbw mm4, mm7
- paddsw mm5, mm4
- packuswb mm5, mm7
- movd [rdx+rdi*2], mm5
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-SECTION_RODATA
-align 16
-x_s1sqr2:
- times 4 dw 0x8A8C
-align 16
-x_c1sqr2less1:
- times 4 dw 0x4E7B
-align 16
-fours:
- times 4 dw 0x0004
diff --git a/vp8/decoder/x86/dequantize_x86.h b/vp8/decoder/x86/dequantize_x86.h
deleted file mode 100644
index dc68daa..0000000
--- a/vp8/decoder/x86/dequantize_x86.h
+++ /dev/null
@@ -1,70 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef DEQUANTIZE_X86_H
-#define DEQUANTIZE_X86_H
-
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-#if HAVE_MMX
-extern prototype_dequant_block(vp8_dequantize_b_mmx);
-extern prototype_dequant_idct_add(vp8_dequant_idct_add_mmx);
-extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add_mmx);
-extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_mmx);
-extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_mmx);
-extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_mmx);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_dequant_block
-#define vp8_dequant_block vp8_dequantize_b_mmx
-
-#undef vp8_dequant_idct_add
-#define vp8_dequant_idct_add vp8_dequant_idct_add_mmx
-
-#undef vp8_dequant_dc_idct_add
-#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_mmx
-
-#undef vp8_dequant_dc_idct_add_y_block
-#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_mmx
-
-#undef vp8_dequant_idct_add_y_block
-#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_mmx
-
-#undef vp8_dequant_idct_add_uv_block
-#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_mmx
-
-#endif
-#endif
-
-#if HAVE_SSE2
-extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_sse2);
-extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_sse2);
-extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_dequant_dc_idct_add_y_block
-#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_sse2
-
-#undef vp8_dequant_idct_add_y_block
-#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_sse2
-
-#undef vp8_dequant_idct_add_uv_block
-#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_sse2
-
-#endif
-#endif
-
-#endif
diff --git a/vp8/decoder/x86/idct_blk_mmx.c b/vp8/decoder/x86/idct_blk_mmx.c
deleted file mode 100644
index 8f1a363..0000000
--- a/vp8/decoder/x86/idct_blk_mmx.c
+++ /dev/null
@@ -1,151 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_ports/config.h"
-#include "vp8/common/idct.h"
-#include "vp8/decoder/dequantize.h"
-
-void vp8_dequant_dc_idct_add_y_block_mmx
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs, short *dc)
-{
- int i;
-
- for (i = 0; i < 4; i++)
- {
- if (eobs[0] > 1)
- vp8_dequant_dc_idct_add_mmx (q, dq, pre, dst, 16, stride, dc[0]);
- else
- vp8_dc_only_idct_add_mmx (dc[0], pre, dst, 16, stride);
-
- if (eobs[1] > 1)
- vp8_dequant_dc_idct_add_mmx (q+16, dq, pre+4, dst+4, 16, stride, dc[1]);
- else
- vp8_dc_only_idct_add_mmx (dc[1], pre+4, dst+4, 16, stride);
-
- if (eobs[2] > 1)
- vp8_dequant_dc_idct_add_mmx (q+32, dq, pre+8, dst+8, 16, stride, dc[2]);
- else
- vp8_dc_only_idct_add_mmx (dc[2], pre+8, dst+8, 16, stride);
-
- if (eobs[3] > 1)
- vp8_dequant_dc_idct_add_mmx (q+48, dq, pre+12, dst+12, 16, stride, dc[3]);
- else
- vp8_dc_only_idct_add_mmx (dc[3], pre+12, dst+12, 16, stride);
-
- q += 64;
- dc += 4;
- pre += 64;
- dst += 4*stride;
- eobs += 4;
- }
-}
-
-void vp8_dequant_idct_add_y_block_mmx
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs)
-{
- int i;
-
- for (i = 0; i < 4; i++)
- {
- if (eobs[0] > 1)
- vp8_dequant_idct_add_mmx (q, dq, pre, dst, 16, stride);
- else
- {
- vp8_dc_only_idct_add_mmx (q[0]*dq[0], pre, dst, 16, stride);
- ((int *)q)[0] = 0;
- }
-
- if (eobs[1] > 1)
- vp8_dequant_idct_add_mmx (q+16, dq, pre+4, dst+4, 16, stride);
- else
- {
- vp8_dc_only_idct_add_mmx (q[16]*dq[0], pre+4, dst+4, 16, stride);
- ((int *)(q+16))[0] = 0;
- }
-
- if (eobs[2] > 1)
- vp8_dequant_idct_add_mmx (q+32, dq, pre+8, dst+8, 16, stride);
- else
- {
- vp8_dc_only_idct_add_mmx (q[32]*dq[0], pre+8, dst+8, 16, stride);
- ((int *)(q+32))[0] = 0;
- }
-
- if (eobs[3] > 1)
- vp8_dequant_idct_add_mmx (q+48, dq, pre+12, dst+12, 16, stride);
- else
- {
- vp8_dc_only_idct_add_mmx (q[48]*dq[0], pre+12, dst+12, 16, stride);
- ((int *)(q+48))[0] = 0;
- }
-
- q += 64;
- pre += 64;
- dst += 4*stride;
- eobs += 4;
- }
-}
-
-void vp8_dequant_idct_add_uv_block_mmx
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
-{
- int i;
-
- for (i = 0; i < 2; i++)
- {
- if (eobs[0] > 1)
- vp8_dequant_idct_add_mmx (q, dq, pre, dstu, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mmx (q[0]*dq[0], pre, dstu, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- if (eobs[1] > 1)
- vp8_dequant_idct_add_mmx (q+16, dq, pre+4, dstu+4, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mmx (q[16]*dq[0], pre+4, dstu+4, 8, stride);
- ((int *)(q+16))[0] = 0;
- }
-
- q += 32;
- pre += 32;
- dstu += 4*stride;
- eobs += 2;
- }
-
- for (i = 0; i < 2; i++)
- {
- if (eobs[0] > 1)
- vp8_dequant_idct_add_mmx (q, dq, pre, dstv, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mmx (q[0]*dq[0], pre, dstv, 8, stride);
- ((int *)q)[0] = 0;
- }
-
- if (eobs[1] > 1)
- vp8_dequant_idct_add_mmx (q+16, dq, pre+4, dstv+4, 8, stride);
- else
- {
- vp8_dc_only_idct_add_mmx (q[16]*dq[0], pre+4, dstv+4, 8, stride);
- ((int *)(q+16))[0] = 0;
- }
-
- q += 32;
- pre += 32;
- dstv += 4*stride;
- eobs += 2;
- }
-}
diff --git a/vp8/decoder/x86/idct_blk_sse2.c b/vp8/decoder/x86/idct_blk_sse2.c
deleted file mode 100644
index 4c88db4..0000000
--- a/vp8/decoder/x86/idct_blk_sse2.c
+++ /dev/null
@@ -1,114 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_ports/config.h"
-#include "vp8/common/idct.h"
-#include "vp8/decoder/dequantize.h"
-
-void idct_dequant_dc_0_2x_sse2
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int dst_stride, short *dc);
-void idct_dequant_dc_full_2x_sse2
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int dst_stride, short *dc);
-
-void idct_dequant_0_2x_sse2
- (short *q, short *dq ,unsigned char *pre,
- unsigned char *dst, int dst_stride, int blk_stride);
-void idct_dequant_full_2x_sse2
- (short *q, short *dq ,unsigned char *pre,
- unsigned char *dst, int dst_stride, int blk_stride);
-
-void vp8_dequant_dc_idct_add_y_block_sse2
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs, short *dc)
-{
- int i;
-
- for (i = 0; i < 4; i++)
- {
- if (((short *)(eobs))[0] & 0xfefe)
- idct_dequant_dc_full_2x_sse2 (q, dq, pre, dst, stride, dc);
- else
- idct_dequant_dc_0_2x_sse2 (q, dq, pre, dst, stride, dc);
-
- if (((short *)(eobs))[1] & 0xfefe)
- idct_dequant_dc_full_2x_sse2 (q+32, dq, pre+8, dst+8, stride, dc+2);
- else
- idct_dequant_dc_0_2x_sse2 (q+32, dq, pre+8, dst+8, stride, dc+2);
-
- q += 64;
- dc += 4;
- pre += 64;
- dst += stride*4;
- eobs += 4;
- }
-}
-
-void vp8_dequant_idct_add_y_block_sse2
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dst, int stride, char *eobs)
-{
- int i;
-
- for (i = 0; i < 4; i++)
- {
- if (((short *)(eobs))[0] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dst, stride, 16);
- else
- idct_dequant_0_2x_sse2 (q, dq, pre, dst, stride, 16);
-
- if (((short *)(eobs))[1] & 0xfefe)
- idct_dequant_full_2x_sse2 (q+32, dq, pre+8, dst+8, stride, 16);
- else
- idct_dequant_0_2x_sse2 (q+32, dq, pre+8, dst+8, stride, 16);
-
- q += 64;
- pre += 64;
- dst += stride*4;
- eobs += 4;
- }
-}
-
-void vp8_dequant_idct_add_uv_block_sse2
- (short *q, short *dq, unsigned char *pre,
- unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
-{
- if (((short *)(eobs))[0] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dstu, stride, 8);
- else
- idct_dequant_0_2x_sse2 (q, dq, pre, dstu, stride, 8);
-
- q += 32;
- pre += 32;
- dstu += stride*4;
-
- if (((short *)(eobs))[1] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dstu, stride, 8);
- else
- idct_dequant_0_2x_sse2 (q, dq, pre, dstu, stride, 8);
-
- q += 32;
- pre += 32;
-
- if (((short *)(eobs))[2] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dstv, stride, 8);
- else
- idct_dequant_0_2x_sse2 (q, dq, pre, dstv, stride, 8);
-
- q += 32;
- pre += 32;
- dstv += stride*4;
-
- if (((short *)(eobs))[3] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dstv, stride, 8);
- else
- idct_dequant_0_2x_sse2 (q, dq, pre, dstv, stride, 8);
-}
diff --git a/vp8/decoder/x86/x86_dsystemdependent.c b/vp8/decoder/x86/x86_dsystemdependent.c
deleted file mode 100644
index 5c16842..0000000
--- a/vp8/decoder/x86/x86_dsystemdependent.c
+++ /dev/null
@@ -1,62 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_ports/x86.h"
-#include "vp8/decoder/onyxd_int.h"
-
-
-#if HAVE_MMX
-void vp8_dequantize_b_impl_mmx(short *sq, short *dq, short *q);
-
-static void dequantize_b_mmx(BLOCKD *d)
-{
- short *sq = (short *) d->qcoeff;
- short *dq = (short *) d->dqcoeff;
- short *q = (short *) d->dequant;
- vp8_dequantize_b_impl_mmx(sq, dq, q);
-}
-#endif
-
-void vp8_arch_x86_decode_init(VP8D_COMP *pbi)
-{
- int flags = x86_simd_caps();
-
- /* Note:
- *
- * This platform can be built without runtime CPU detection as well. If
- * you modify any of the function mappings present in this file, be sure
- * to also update them in static mapings (<arch>/filename_<arch>.h)
- */
-#if CONFIG_RUNTIME_CPU_DETECT
- /* Override default functions with fastest ones for this CPU. */
-#if HAVE_MMX
- if (flags & HAS_MMX)
- {
- pbi->dequant.block = dequantize_b_mmx;
- pbi->dequant.idct_add = vp8_dequant_idct_add_mmx;
- pbi->dequant.dc_idct_add = vp8_dequant_dc_idct_add_mmx;
- pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_mmx;
- pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_mmx;
- pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_mmx;
- }
-#endif
-#if HAVE_SSE2
- if (flags & HAS_SSE2)
- {
- pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_sse2;
- pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_sse2;
- pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_sse2;
- }
-#endif
-
-#endif
-}
diff --git a/vp8/encoder/arm/arm_csystemdependent.c b/vp8/encoder/arm/arm_csystemdependent.c
deleted file mode 100644
index f148b89..0000000
--- a/vp8/encoder/arm/arm_csystemdependent.c
+++ /dev/null
@@ -1,142 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_ports/arm.h"
-#include "vp8/encoder/variance.h"
-#include "vp8/encoder/onyx_int.h"
-
-extern void (*vp8_yv12_copy_partial_frame_ptr)(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc, int Fraction);
-extern void vp8_yv12_copy_partial_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc, int Fraction);
-extern void vpxyv12_copy_partial_frame_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc, int Fraction);
-
-void vp8_arch_arm_encoder_init(VP8_COMP *cpi)
-{
-#if CONFIG_RUNTIME_CPU_DETECT
- int flags = cpi->common.rtcd.flags;
- int has_edsp = flags & HAS_EDSP;
- int has_media = flags & HAS_MEDIA;
- int has_neon = flags & HAS_NEON;
-
-#if HAVE_ARMV6
- if (has_media)
- {
- cpi->rtcd.variance.sad16x16 = vp8_sad16x16_armv6;
- /*cpi->rtcd.variance.sad16x8 = vp8_sad16x8_c;
- cpi->rtcd.variance.sad8x16 = vp8_sad8x16_c;
- cpi->rtcd.variance.sad8x8 = vp8_sad8x8_c;
- cpi->rtcd.variance.sad4x4 = vp8_sad4x4_c;*/
-
- /*cpi->rtcd.variance.var4x4 = vp8_variance4x4_c;*/
- cpi->rtcd.variance.var8x8 = vp8_variance8x8_armv6;
- /*cpi->rtcd.variance.var8x16 = vp8_variance8x16_c;
- cpi->rtcd.variance.var16x8 = vp8_variance16x8_c;*/
- cpi->rtcd.variance.var16x16 = vp8_variance16x16_armv6;
-
- /*cpi->rtcd.variance.subpixvar4x4 = vp8_sub_pixel_variance4x4_c;*/
- cpi->rtcd.variance.subpixvar8x8 = vp8_sub_pixel_variance8x8_armv6;
- /*cpi->rtcd.variance.subpixvar8x16 = vp8_sub_pixel_variance8x16_c;
- cpi->rtcd.variance.subpixvar16x8 = vp8_sub_pixel_variance16x8_c;*/
- cpi->rtcd.variance.subpixvar16x16 = vp8_sub_pixel_variance16x16_armv6;
- cpi->rtcd.variance.halfpixvar16x16_h = vp8_variance_halfpixvar16x16_h_armv6;
- cpi->rtcd.variance.halfpixvar16x16_v = vp8_variance_halfpixvar16x16_v_armv6;
- cpi->rtcd.variance.halfpixvar16x16_hv = vp8_variance_halfpixvar16x16_hv_armv6;
-
- cpi->rtcd.variance.mse16x16 = vp8_mse16x16_armv6;
- /*cpi->rtcd.variance.getmbss = vp8_get_mb_ss_c;*/
-
- /*cpi->rtcd.variance.get16x16prederror = vp8_get16x16pred_error_c;
- cpi->rtcd.variance.get8x8var = vp8_get8x8var_c;
- cpi->rtcd.variance.get16x16var = vp8_get16x16var_c;;
- cpi->rtcd.variance.get4x4sse_cs = vp8_get4x4sse_cs_c;*/
-
- /*cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_c;
- cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_c;*/
- cpi->rtcd.fdct.fast4x4 = vp8_fast_fdct4x4_armv6;
- cpi->rtcd.fdct.fast8x4 = vp8_fast_fdct8x4_armv6;
- cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_armv6;
-
- /*cpi->rtcd.encodemb.berr = vp8_block_error_c;
- cpi->rtcd.encodemb.mberr = vp8_mbblock_error_c;
- cpi->rtcd.encodemb.mbuverr = vp8_mbuverror_c;*/
- cpi->rtcd.encodemb.subb = vp8_subtract_b_armv6;
- cpi->rtcd.encodemb.submby = vp8_subtract_mby_armv6;
- cpi->rtcd.encodemb.submbuv = vp8_subtract_mbuv_armv6;
-
- /*cpi->rtcd.quantize.quantb = vp8_regular_quantize_b;*/
- cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_armv6;
- }
-#endif
-
-#if HAVE_ARMV7
- if (has_neon)
- {
- cpi->rtcd.variance.sad16x16 = vp8_sad16x16_neon;
- cpi->rtcd.variance.sad16x8 = vp8_sad16x8_neon;
- cpi->rtcd.variance.sad8x16 = vp8_sad8x16_neon;
- cpi->rtcd.variance.sad8x8 = vp8_sad8x8_neon;
- cpi->rtcd.variance.sad4x4 = vp8_sad4x4_neon;
-
- /*cpi->rtcd.variance.var4x4 = vp8_variance4x4_c;*/
- cpi->rtcd.variance.var8x8 = vp8_variance8x8_neon;
- cpi->rtcd.variance.var8x16 = vp8_variance8x16_neon;
- cpi->rtcd.variance.var16x8 = vp8_variance16x8_neon;
- cpi->rtcd.variance.var16x16 = vp8_variance16x16_neon;
-
- /*cpi->rtcd.variance.subpixvar4x4 = vp8_sub_pixel_variance4x4_c;*/
- cpi->rtcd.variance.subpixvar8x8 = vp8_sub_pixel_variance8x8_neon;
- /*cpi->rtcd.variance.subpixvar8x16 = vp8_sub_pixel_variance8x16_c;
- cpi->rtcd.variance.subpixvar16x8 = vp8_sub_pixel_variance16x8_c;*/
- cpi->rtcd.variance.subpixvar16x16 = vp8_sub_pixel_variance16x16_neon;
- cpi->rtcd.variance.halfpixvar16x16_h = vp8_variance_halfpixvar16x16_h_neon;
- cpi->rtcd.variance.halfpixvar16x16_v = vp8_variance_halfpixvar16x16_v_neon;
- cpi->rtcd.variance.halfpixvar16x16_hv = vp8_variance_halfpixvar16x16_hv_neon;
-
- cpi->rtcd.variance.mse16x16 = vp8_mse16x16_neon;
- /*cpi->rtcd.variance.getmbss = vp8_get_mb_ss_c;*/
-
- cpi->rtcd.variance.get16x16prederror = vp8_get16x16pred_error_neon;
- /*cpi->rtcd.variance.get8x8var = vp8_get8x8var_c;
- cpi->rtcd.variance.get16x16var = vp8_get16x16var_c;*/
- cpi->rtcd.variance.get4x4sse_cs = vp8_get4x4sse_cs_neon;
-
- cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_neon;
- cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_neon;
- cpi->rtcd.fdct.fast4x4 = vp8_fast_fdct4x4_neon;
- cpi->rtcd.fdct.fast8x4 = vp8_fast_fdct8x4_neon;
- cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_neon;
-
- /*cpi->rtcd.encodemb.berr = vp8_block_error_c;
- cpi->rtcd.encodemb.mberr = vp8_mbblock_error_c;
- cpi->rtcd.encodemb.mbuverr = vp8_mbuverror_c;*/
- cpi->rtcd.encodemb.subb = vp8_subtract_b_neon;
- cpi->rtcd.encodemb.submby = vp8_subtract_mby_neon;
- cpi->rtcd.encodemb.submbuv = vp8_subtract_mbuv_neon;
-
- /*cpi->rtcd.quantize.quantb = vp8_regular_quantize_b;
- cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_c;*/
- /* The neon quantizer has not been updated to match the new exact
- * quantizer introduced in commit e04e2935
- */
- /*cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_neon;*/
- }
-#endif
-
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (has_neon)
-#endif
- {
- vp8_yv12_copy_partial_frame_ptr = vpxyv12_copy_partial_frame_neon;
- }
-#endif
-#endif
-}
diff --git a/vp8/encoder/arm/armv5te/boolhuff_armv5te.asm b/vp8/encoder/arm/armv5te/boolhuff_armv5te.asm
deleted file mode 100644
index 3c05f57..0000000
--- a/vp8/encoder/arm/armv5te/boolhuff_armv5te.asm
+++ /dev/null
@@ -1,286 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_start_encode|
- EXPORT |vp8_encode_bool|
- EXPORT |vp8_stop_encode|
- EXPORT |vp8_encode_value|
-
- INCLUDE asm_enc_offsets.asm
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA |.text|, CODE, READONLY
-
-; r0 BOOL_CODER *br
-; r1 unsigned char *source
-
-|vp8_start_encode| PROC
- mov r12, #0
- mov r3, #255
- mvn r2, #23
- str r12, [r0, #vp8_writer_lowvalue]
- str r3, [r0, #vp8_writer_range]
- str r12, [r0, #vp8_writer_value]
- str r2, [r0, #vp8_writer_count]
- str r12, [r0, #vp8_writer_pos]
- str r1, [r0, #vp8_writer_buffer]
- bx lr
- ENDP
-
-; r0 BOOL_CODER *br
-; r1 int bit
-; r2 int probability
-|vp8_encode_bool| PROC
- push {r4-r9, lr}
-
- mov r4, r2
-
- ldr r2, [r0, #vp8_writer_lowvalue]
- ldr r5, [r0, #vp8_writer_range]
- ldr r3, [r0, #vp8_writer_count]
-
- sub r7, r5, #1 ; range-1
-
- cmp r1, #0
- mul r4, r4, r7 ; ((range-1) * probability)
-
- mov r7, #1
- add r4, r7, r4, lsr #8 ; 1 + (((range-1) * probability) >> 8)
-
- addne r2, r2, r4 ; if (bit) lowvalue += split
- subne r4, r5, r4 ; if (bit) range = range-split
-
- ; Counting the leading zeros is used to normalize range.
- clz r6, r4
- sub r6, r6, #24 ; shift
-
- ; Flag is set on the sum of count. This flag is used later
- ; to determine if count >= 0
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi token_count_lt_zero ; if(count >= 0)
-
- sub r6, r6, r3 ; offset = shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl token_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos-1
- b token_zero_while_start
-token_zero_while_loop
- mov r9, #0
- strb r9, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-token_zero_while_start
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r1, [r7, r4]
- cmpge r1, #0xff
- beq token_zero_while_loop
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r9, [r7, r4] ; w->buffer[x]
- add r9, r9, #1
- strb r9, [r7, r4] ; w->buffer[x] + 1
-token_high_bit_not_set
- rsb r4, r6, #24 ; 24-offset
- ldr r9, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos] ; w->pos
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r1, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r1, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r9, r4] ; w->buffer[w->pos++]
-
-token_count_lt_zero
- lsl r2, r2, r6 ; lowvalue <<= shift
-
- str r2, [r0, #vp8_writer_lowvalue]
- str r5, [r0, #vp8_writer_range]
- str r3, [r0, #vp8_writer_count]
- pop {r4-r9, pc}
- ENDP
-
-; r0 BOOL_CODER *br
-|vp8_stop_encode| PROC
- push {r4-r10, lr}
-
- ldr r2, [r0, #vp8_writer_lowvalue]
- ldr r5, [r0, #vp8_writer_range]
- ldr r3, [r0, #vp8_writer_count]
-
- mov r10, #32
-
-stop_encode_loop
- sub r7, r5, #1 ; range-1
-
- mov r4, r7, lsl #7 ; ((range-1) * 128)
-
- mov r7, #1
- add r4, r7, r4, lsr #8 ; 1 + (((range-1) * 128) >> 8)
-
- ; Counting the leading zeros is used to normalize range.
- clz r6, r4
- sub r6, r6, #24 ; shift
-
- ; Flag is set on the sum of count. This flag is used later
- ; to determine if count >= 0
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi token_count_lt_zero_se ; if(count >= 0)
-
- sub r6, r6, r3 ; offset = shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl token_high_bit_not_set_se
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos-1
- b token_zero_while_start_se
-token_zero_while_loop_se
- mov r9, #0
- strb r9, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-token_zero_while_start_se
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r1, [r7, r4]
- cmpge r1, #0xff
- beq token_zero_while_loop_se
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r9, [r7, r4] ; w->buffer[x]
- add r9, r9, #1
- strb r9, [r7, r4] ; w->buffer[x] + 1
-token_high_bit_not_set_se
- rsb r4, r6, #24 ; 24-offset
- ldr r9, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos] ; w->pos
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r1, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r1, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r9, r4] ; w->buffer[w->pos++]
-
-token_count_lt_zero_se
- lsl r2, r2, r6 ; lowvalue <<= shift
-
- subs r10, r10, #1
- bne stop_encode_loop
-
- str r2, [r0, #vp8_writer_lowvalue]
- str r5, [r0, #vp8_writer_range]
- str r3, [r0, #vp8_writer_count]
- pop {r4-r10, pc}
-
- ENDP
-
-; r0 BOOL_CODER *br
-; r1 int data
-; r2 int bits
-|vp8_encode_value| PROC
- push {r4-r11, lr}
-
- mov r10, r2
-
- ldr r2, [r0, #vp8_writer_lowvalue]
- ldr r5, [r0, #vp8_writer_range]
- ldr r3, [r0, #vp8_writer_count]
-
- rsb r4, r10, #32 ; 32-n
-
- ; v is kept in r1 during the token pack loop
- lsl r1, r1, r4 ; r1 = v << 32 - n
-
-encode_value_loop
- sub r7, r5, #1 ; range-1
-
- ; Decisions are made based on the bit value shifted
- ; off of v, so set a flag here based on this.
- ; This value is refered to as "bb"
- lsls r1, r1, #1 ; bit = v >> n
- mov r4, r7, lsl #7 ; ((range-1) * 128)
-
- mov r7, #1
- add r4, r7, r4, lsr #8 ; 1 + (((range-1) * 128) >> 8)
-
- addcs r2, r2, r4 ; if (bit) lowvalue += split
- subcs r4, r5, r4 ; if (bit) range = range-split
-
- ; Counting the leading zeros is used to normalize range.
- clz r6, r4
- sub r6, r6, #24 ; shift
-
- ; Flag is set on the sum of count. This flag is used later
- ; to determine if count >= 0
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi token_count_lt_zero_ev ; if(count >= 0)
-
- sub r6, r6, r3 ; offset = shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl token_high_bit_not_set_ev
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos-1
- b token_zero_while_start_ev
-token_zero_while_loop_ev
- mov r9, #0
- strb r9, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-token_zero_while_start_ev
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r11, [r7, r4]
- cmpge r11, #0xff
- beq token_zero_while_loop_ev
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r9, [r7, r4] ; w->buffer[x]
- add r9, r9, #1
- strb r9, [r7, r4] ; w->buffer[x] + 1
-token_high_bit_not_set_ev
- rsb r4, r6, #24 ; 24-offset
- ldr r9, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos] ; w->pos
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r11, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r11, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r9, r4] ; w->buffer[w->pos++]
-
-token_count_lt_zero_ev
- lsl r2, r2, r6 ; lowvalue <<= shift
-
- subs r10, r10, #1
- bne encode_value_loop
-
- str r2, [r0, #vp8_writer_lowvalue]
- str r5, [r0, #vp8_writer_range]
- str r3, [r0, #vp8_writer_count]
- pop {r4-r11, pc}
- ENDP
-
- END
diff --git a/vp8/encoder/arm/armv5te/vp8_packtokens_armv5.asm b/vp8/encoder/arm/armv5te/vp8_packtokens_armv5.asm
deleted file mode 100644
index d939287..0000000
--- a/vp8/encoder/arm/armv5te/vp8_packtokens_armv5.asm
+++ /dev/null
@@ -1,291 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8cx_pack_tokens_armv5|
-
- INCLUDE asm_enc_offsets.asm
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA |.text|, CODE, READONLY
-
-; r0 vp8_writer *w
-; r1 const TOKENEXTRA *p
-; r2 int xcount
-; r3 vp8_coef_encodings
-; s0 vp8_extra_bits
-; s1 vp8_coef_tree
-|vp8cx_pack_tokens_armv5| PROC
- push {r4-r11, lr}
-
- ; Add size of xcount * sizeof (TOKENEXTRA) to get stop
- ; sizeof (TOKENEXTRA) is 8
- sub sp, sp, #12
- add r2, r1, r2, lsl #3 ; stop = p + xcount*sizeof(TOKENEXTRA)
- str r2, [sp, #0]
- str r3, [sp, #8] ; save vp8_coef_encodings
- ldr r2, [r0, #vp8_writer_lowvalue]
- ldr r5, [r0, #vp8_writer_range]
- ldr r3, [r0, #vp8_writer_count]
- b check_p_lt_stop
-
-while_p_lt_stop
- ldrb r6, [r1, #tokenextra_token] ; t
- ldr r4, [sp, #8] ; vp8_coef_encodings
- mov lr, #0
- add r4, r4, r6, lsl #3 ; a = vp8_coef_encodings + t
- ldr r9, [r1, #tokenextra_context_tree] ; pp
-
- ldrb r7, [r1, #tokenextra_skip_eob_node]
-
- ldr r6, [r4, #vp8_token_value] ; v
- ldr r8, [r4, #vp8_token_len] ; n
-
- ; vp8 specific skip_eob_node
- cmp r7, #0
- movne lr, #2 ; i = 2
- subne r8, r8, #1 ; --n
-
- rsb r4, r8, #32 ; 32-n
- ldr r10, [sp, #52] ; vp8_coef_tree
-
- ; v is kept in r12 during the token pack loop
- lsl r12, r6, r4 ; r12 = v << 32 - n
-
-; loop start
-token_loop
- ldrb r4, [r9, lr, asr #1] ; pp [i>>1]
- sub r7, r5, #1 ; range-1
-
- ; Decisions are made based on the bit value shifted
- ; off of v, so set a flag here based on this.
- ; This value is refered to as "bb"
- lsls r12, r12, #1 ; bb = v >> n
- mul r4, r4, r7 ; ((range-1) * pp[i>>1]))
-
- ; bb can only be 0 or 1. So only execute this statement
- ; if bb == 1, otherwise it will act like i + 0
- addcs lr, lr, #1 ; i + bb
-
- mov r7, #1
- ldrsb lr, [r10, lr] ; i = vp8_coef_tree[i+bb]
- add r4, r7, r4, lsr #8 ; 1 + (((range-1) * pp[i>>1]) >> 8)
-
- addcs r2, r2, r4 ; if (bb) lowvalue += split
- subcs r4, r5, r4 ; if (bb) range = range-split
-
- ; Counting the leading zeros is used to normalize range.
- clz r6, r4
- sub r6, r6, #24 ; shift
-
- ; Flag is set on the sum of count. This flag is used later
- ; to determine if count >= 0
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi token_count_lt_zero ; if(count >= 0)
-
- sub r6, r6, r3 ; offset = shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl token_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos-1
- b token_zero_while_start
-token_zero_while_loop
- mov r10, #0
- strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-token_zero_while_start
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r11, [r7, r4]
- cmpge r11, #0xff
- beq token_zero_while_loop
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r10, [r7, r4] ; w->buffer[x]
- add r10, r10, #1
- strb r10, [r7, r4] ; w->buffer[x] + 1
-token_high_bit_not_set
- rsb r4, r6, #24 ; 24-offset
- ldr r10, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos] ; w->pos
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r11, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r11, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r10, r4] ; w->buffer[w->pos++]
-
- ; r10 is used earlier in the loop, but r10 is used as
- ; temp variable here. So after r10 is used, reload
- ; vp8_coef_tree_dcd into r10
- ldr r10, [sp, #52] ; vp8_coef_tree
-
-token_count_lt_zero
- lsl r2, r2, r6 ; lowvalue <<= shift
-
- subs r8, r8, #1 ; --n
- bne token_loop
-
- ldrb r6, [r1, #tokenextra_token] ; t
- ldr r7, [sp, #48] ; vp8_extra_bits
- ; Add t * sizeof (vp8_extra_bit_struct) to get the desired
- ; element. Here vp8_extra_bit_struct == 16
- add r12, r7, r6, lsl #4 ; b = vp8_extra_bits + t
-
- ldr r4, [r12, #vp8_extra_bit_struct_base_val]
- cmp r4, #0
- beq skip_extra_bits
-
-; if( b->base_val)
- ldr r8, [r12, #vp8_extra_bit_struct_len] ; L
- ldrsh lr, [r1, #tokenextra_extra] ; e = p->Extra
- cmp r8, #0 ; if( L)
- beq no_extra_bits
-
- ldr r9, [r12, #vp8_extra_bit_struct_prob]
- asr r7, lr, #1 ; v=e>>1
-
- ldr r10, [r12, #vp8_extra_bit_struct_tree]
- str r10, [sp, #4] ; b->tree
-
- rsb r4, r8, #32
- lsl r12, r7, r4
-
- mov lr, #0 ; i = 0
-
-extra_bits_loop
- ldrb r4, [r9, lr, asr #1] ; pp[i>>1]
- sub r7, r5, #1 ; range-1
- lsls r12, r12, #1 ; v >> n
- mul r4, r4, r7 ; (range-1) * pp[i>>1]
- addcs lr, lr, #1 ; i + bb
-
- mov r7, #1
- ldrsb lr, [r10, lr] ; i = b->tree[i+bb]
- add r4, r7, r4, lsr #8 ; split = 1 + (((range-1) * pp[i>>1]) >> 8)
-
- addcs r2, r2, r4 ; if (bb) lowvalue += split
- subcs r4, r5, r4 ; if (bb) range = range-split
-
- clz r6, r4
- sub r6, r6, #24
-
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi extra_count_lt_zero ; if(count >= 0)
-
- sub r6, r6, r3 ; offset= shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl extra_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos - 1
- b extra_zero_while_start
-extra_zero_while_loop
- mov r10, #0
- strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-extra_zero_while_start
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r11, [r7, r4]
- cmpge r11, #0xff
- beq extra_zero_while_loop
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r10, [r7, r4]
- add r10, r10, #1
- strb r10, [r7, r4]
-extra_high_bit_not_set
- rsb r4, r6, #24 ; 24-offset
- ldr r10, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos]
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r11, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r11, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r10, r4] ; w->buffer[w->pos++]=(lowvalue >> (24-offset))
- ldr r10, [sp, #4] ; b->tree
-extra_count_lt_zero
- lsl r2, r2, r6
-
- subs r8, r8, #1 ; --n
- bne extra_bits_loop ; while (n)
-
-no_extra_bits
- ldr lr, [r1, #4] ; e = p->Extra
- add r4, r5, #1 ; range + 1
- tst lr, #1
- lsr r4, r4, #1 ; split = (range + 1) >> 1
- addne r2, r2, r4 ; lowvalue += split
- subne r4, r5, r4 ; range = range-split
- tst r2, #0x80000000 ; lowvalue & 0x80000000
- lsl r5, r4, #1 ; range <<= 1
- beq end_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos]
- mov r7, #0
- sub r4, r4, #1
- b end_zero_while_start
-end_zero_while_loop
- strb r7, [r6, r4]
- sub r4, r4, #1 ; x--
-end_zero_while_start
- cmp r4, #0
- ldrge r6, [r0, #vp8_writer_buffer]
- ldrb r12, [r6, r4]
- cmpge r12, #0xff
- beq end_zero_while_loop
-
- ldr r6, [r0, #vp8_writer_buffer]
- ldrb r7, [r6, r4]
- add r7, r7, #1
- strb r7, [r6, r4]
-end_high_bit_not_set
- adds r3, r3, #1 ; ++count
- lsl r2, r2, #1 ; lowvalue <<= 1
- bne end_count_zero
-
- ldr r4, [r0, #vp8_writer_pos]
- mvn r3, #7
- ldr r7, [r0, #vp8_writer_buffer]
- lsr r6, r2, #24 ; lowvalue >> 24
- add r12, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r12, [r0, #0x10]
- strb r6, [r7, r4]
-end_count_zero
-skip_extra_bits
- add r1, r1, #TOKENEXTRA_SZ ; ++p
-check_p_lt_stop
- ldr r4, [sp, #0] ; stop
- cmp r1, r4 ; while( p < stop)
- bcc while_p_lt_stop
-
- str r2, [r0, #vp8_writer_lowvalue]
- str r5, [r0, #vp8_writer_range]
- str r3, [r0, #vp8_writer_count]
- add sp, sp, #12
- pop {r4-r11, pc}
- ENDP
-
- END
diff --git a/vp8/encoder/arm/armv5te/vp8_packtokens_mbrow_armv5.asm b/vp8/encoder/arm/armv5te/vp8_packtokens_mbrow_armv5.asm
deleted file mode 100644
index ac2bba6..0000000
--- a/vp8/encoder/arm/armv5te/vp8_packtokens_mbrow_armv5.asm
+++ /dev/null
@@ -1,327 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8cx_pack_mb_row_tokens_armv5|
-
- INCLUDE asm_enc_offsets.asm
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA |.text|, CODE, READONLY
-
-; r0 VP8_COMP *cpi
-; r1 vp8_writer *w
-; r2 vp8_coef_encodings
-; r3 vp8_extra_bits
-; s0 vp8_coef_tree
-
-|vp8cx_pack_mb_row_tokens_armv5| PROC
- push {r4-r11, lr}
- sub sp, sp, #24
-
- ; Compute address of cpi->common.mb_rows
- ldr r4, _VP8_COMP_common_
- ldr r6, _VP8_COMMON_MBrows_
- add r4, r0, r4
-
- ldr r5, [r4, r6] ; load up mb_rows
-
- str r2, [sp, #20] ; save vp8_coef_encodings
- str r5, [sp, #12] ; save mb_rows
- str r3, [sp, #8] ; save vp8_extra_bits
-
- ldr r4, _VP8_COMP_tplist_
- add r4, r0, r4
- ldr r7, [r4, #0] ; dereference cpi->tp_list
-
- mov r0, r1 ; keep same as other loops
-
- ldr r2, [r0, #vp8_writer_lowvalue]
- ldr r5, [r0, #vp8_writer_range]
- ldr r3, [r0, #vp8_writer_count]
-
-mb_row_loop
-
- ldr r1, [r7, #tokenlist_start]
- ldr r9, [r7, #tokenlist_stop]
- str r9, [sp, #0] ; save stop for later comparison
- str r7, [sp, #16] ; tokenlist address for next time
-
- b check_p_lt_stop
-
- ; actuall work gets done here!
-
-while_p_lt_stop
- ldrb r6, [r1, #tokenextra_token] ; t
- ldr r4, [sp, #20] ; vp8_coef_encodings
- mov lr, #0
- add r4, r4, r6, lsl #3 ; a = vp8_coef_encodings + t
- ldr r9, [r1, #tokenextra_context_tree] ; pp
-
- ldrb r7, [r1, #tokenextra_skip_eob_node]
-
- ldr r6, [r4, #vp8_token_value] ; v
- ldr r8, [r4, #vp8_token_len] ; n
-
- ; vp8 specific skip_eob_node
- cmp r7, #0
- movne lr, #2 ; i = 2
- subne r8, r8, #1 ; --n
-
- rsb r4, r8, #32 ; 32-n
- ldr r10, [sp, #60] ; vp8_coef_tree
-
- ; v is kept in r12 during the token pack loop
- lsl r12, r6, r4 ; r12 = v << 32 - n
-
-; loop start
-token_loop
- ldrb r4, [r9, lr, asr #1] ; pp [i>>1]
- sub r7, r5, #1 ; range-1
-
- ; Decisions are made based on the bit value shifted
- ; off of v, so set a flag here based on this.
- ; This value is refered to as "bb"
- lsls r12, r12, #1 ; bb = v >> n
- mul r4, r4, r7 ; ((range-1) * pp[i>>1]))
-
- ; bb can only be 0 or 1. So only execute this statement
- ; if bb == 1, otherwise it will act like i + 0
- addcs lr, lr, #1 ; i + bb
-
- mov r7, #1
- ldrsb lr, [r10, lr] ; i = vp8_coef_tree[i+bb]
- add r4, r7, r4, lsr #8 ; 1 + (((range-1) * pp[i>>1]) >> 8)
-
- addcs r2, r2, r4 ; if (bb) lowvalue += split
- subcs r4, r5, r4 ; if (bb) range = range-split
-
- ; Counting the leading zeros is used to normalize range.
- clz r6, r4
- sub r6, r6, #24 ; shift
-
- ; Flag is set on the sum of count. This flag is used later
- ; to determine if count >= 0
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi token_count_lt_zero ; if(count >= 0)
-
- sub r6, r6, r3 ; offset = shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl token_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos-1
- b token_zero_while_start
-token_zero_while_loop
- mov r10, #0
- strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-token_zero_while_start
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r11, [r7, r4]
- cmpge r11, #0xff
- beq token_zero_while_loop
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r10, [r7, r4] ; w->buffer[x]
- add r10, r10, #1
- strb r10, [r7, r4] ; w->buffer[x] + 1
-token_high_bit_not_set
- rsb r4, r6, #24 ; 24-offset
- ldr r10, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos] ; w->pos
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r11, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r11, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r10, r4] ; w->buffer[w->pos++]
-
- ; r10 is used earlier in the loop, but r10 is used as
- ; temp variable here. So after r10 is used, reload
- ; vp8_coef_tree_dcd into r10
- ldr r10, [sp, #60] ; vp8_coef_tree
-
-token_count_lt_zero
- lsl r2, r2, r6 ; lowvalue <<= shift
-
- subs r8, r8, #1 ; --n
- bne token_loop
-
- ldrb r6, [r1, #tokenextra_token] ; t
- ldr r7, [sp, #8] ; vp8_extra_bits
- ; Add t * sizeof (vp8_extra_bit_struct) to get the desired
- ; element. Here vp8_extra_bit_struct == 16
- add r12, r7, r6, lsl #4 ; b = vp8_extra_bits + t
-
- ldr r4, [r12, #vp8_extra_bit_struct_base_val]
- cmp r4, #0
- beq skip_extra_bits
-
-; if( b->base_val)
- ldr r8, [r12, #vp8_extra_bit_struct_len] ; L
- ldrsh lr, [r1, #tokenextra_extra] ; e = p->Extra
- cmp r8, #0 ; if( L)
- beq no_extra_bits
-
- ldr r9, [r12, #vp8_extra_bit_struct_prob]
- asr r7, lr, #1 ; v=e>>1
-
- ldr r10, [r12, #vp8_extra_bit_struct_tree]
- str r10, [sp, #4] ; b->tree
-
- rsb r4, r8, #32
- lsl r12, r7, r4
-
- mov lr, #0 ; i = 0
-
-extra_bits_loop
- ldrb r4, [r9, lr, asr #1] ; pp[i>>1]
- sub r7, r5, #1 ; range-1
- lsls r12, r12, #1 ; v >> n
- mul r4, r4, r7 ; (range-1) * pp[i>>1]
- addcs lr, lr, #1 ; i + bb
-
- mov r7, #1
- ldrsb lr, [r10, lr] ; i = b->tree[i+bb]
- add r4, r7, r4, lsr #8 ; split = 1 + (((range-1) * pp[i>>1]) >> 8)
-
- addcs r2, r2, r4 ; if (bb) lowvalue += split
- subcs r4, r5, r4 ; if (bb) range = range-split
-
- clz r6, r4
- sub r6, r6, #24
-
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi extra_count_lt_zero ; if(count >= 0)
-
- sub r6, r6, r3 ; offset= shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl extra_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos - 1
- b extra_zero_while_start
-extra_zero_while_loop
- mov r10, #0
- strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-extra_zero_while_start
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r11, [r7, r4]
- cmpge r11, #0xff
- beq extra_zero_while_loop
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r10, [r7, r4]
- add r10, r10, #1
- strb r10, [r7, r4]
-extra_high_bit_not_set
- rsb r4, r6, #24 ; 24-offset
- ldr r10, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos]
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r11, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r11, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r10, r4] ; w->buffer[w->pos++]=(lowvalue >> (24-offset))
- ldr r10, [sp, #4] ; b->tree
-extra_count_lt_zero
- lsl r2, r2, r6
-
- subs r8, r8, #1 ; --n
- bne extra_bits_loop ; while (n)
-
-no_extra_bits
- ldr lr, [r1, #4] ; e = p->Extra
- add r4, r5, #1 ; range + 1
- tst lr, #1
- lsr r4, r4, #1 ; split = (range + 1) >> 1
- addne r2, r2, r4 ; lowvalue += split
- subne r4, r5, r4 ; range = range-split
- tst r2, #0x80000000 ; lowvalue & 0x80000000
- lsl r5, r4, #1 ; range <<= 1
- beq end_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos]
- mov r7, #0
- sub r4, r4, #1
- b end_zero_while_start
-end_zero_while_loop
- strb r7, [r6, r4]
- sub r4, r4, #1 ; x--
-end_zero_while_start
- cmp r4, #0
- ldrge r6, [r0, #vp8_writer_buffer]
- ldrb r12, [r6, r4]
- cmpge r12, #0xff
- beq end_zero_while_loop
-
- ldr r6, [r0, #vp8_writer_buffer]
- ldrb r7, [r6, r4]
- add r7, r7, #1
- strb r7, [r6, r4]
-end_high_bit_not_set
- adds r3, r3, #1 ; ++count
- lsl r2, r2, #1 ; lowvalue <<= 1
- bne end_count_zero
-
- ldr r4, [r0, #vp8_writer_pos]
- mvn r3, #7
- ldr r7, [r0, #vp8_writer_buffer]
- lsr r6, r2, #24 ; lowvalue >> 24
- add r12, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r12, [r0, #0x10]
- strb r6, [r7, r4]
-end_count_zero
-skip_extra_bits
- add r1, r1, #TOKENEXTRA_SZ ; ++p
-check_p_lt_stop
- ldr r4, [sp, #0] ; stop
- cmp r1, r4 ; while( p < stop)
- bcc while_p_lt_stop
-
- ldr r6, [sp, #12] ; mb_rows
- ldr r7, [sp, #16] ; tokenlist address
- subs r6, r6, #1
- add r7, r7, #TOKENLIST_SZ ; next element in the array
- str r6, [sp, #12]
- bne mb_row_loop
-
- str r2, [r0, #vp8_writer_lowvalue]
- str r5, [r0, #vp8_writer_range]
- str r3, [r0, #vp8_writer_count]
- add sp, sp, #24
- pop {r4-r11, pc}
- ENDP
-
-_VP8_COMP_common_
- DCD vp8_comp_common
-_VP8_COMMON_MBrows_
- DCD vp8_common_mb_rows
-_VP8_COMP_tplist_
- DCD vp8_comp_tplist
-
- END
diff --git a/vp8/encoder/arm/armv5te/vp8_packtokens_partitions_armv5.asm b/vp8/encoder/arm/armv5te/vp8_packtokens_partitions_armv5.asm
deleted file mode 100644
index c2eccdb..0000000
--- a/vp8/encoder/arm/armv5te/vp8_packtokens_partitions_armv5.asm
+++ /dev/null
@@ -1,465 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8cx_pack_tokens_into_partitions_armv5|
-
- INCLUDE asm_enc_offsets.asm
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA |.text|, CODE, READONLY
-
-; r0 VP8_COMP *cpi
-; r1 unsigned char *cx_data
-; r2 int num_part
-; r3 *size
-; s0 vp8_coef_encodings
-; s1 vp8_extra_bits,
-; s2 const vp8_tree_index *,
-
-|vp8cx_pack_tokens_into_partitions_armv5| PROC
- push {r4-r11, lr}
- sub sp, sp, #44
-
- ; Compute address of cpi->common.mb_rows
- ldr r4, _VP8_COMP_common_
- ldr r6, _VP8_COMMON_MBrows_
- add r4, r0, r4
-
- ldr r5, [r4, r6] ; load up mb_rows
-
- str r5, [sp, #36] ; save mb_rows
- str r1, [sp, #24] ; save cx_data
- str r2, [sp, #20] ; save num_part
- str r3, [sp, #8] ; save *size
-
- ; *size = 3*(num_part -1 );
- sub r2, r2, #1 ; num_part - 1
- add r2, r2, r2, lsl #1 ; 3*(num_part - 1)
- str r2, [r3]
-
- add r2, r2, r1 ; cx_data + *size
- str r2, [sp, #40] ; ptr
-
- ldr r4, _VP8_COMP_tplist_
- add r4, r0, r4
- ldr r7, [r4, #0] ; dereference cpi->tp_list
- str r7, [sp, #32] ; store start of cpi->tp_list
-
- ldr r11, _VP8_COMP_bc2_ ; load up vp8_writer out of cpi
- add r0, r0, r11
-
- mov r11, #0
- str r11, [sp, #28] ; i
-
-numparts_loop
- ldr r10, [sp, #40] ; ptr
- ldr r5, [sp, #36] ; move mb_rows to the counting section
- sub r5, r5, r11 ; move start point with each partition
- ; mb_rows starts at i
- str r5, [sp, #12]
-
- ; Reset all of the VP8 Writer data for each partition that
- ; is processed.
- ; start_encode
- mov r2, #0 ; vp8_writer_lowvalue
- mov r5, #255 ; vp8_writer_range
- mvn r3, #23 ; vp8_writer_count
-
- str r2, [r0, #vp8_writer_value]
- str r2, [r0, #vp8_writer_pos]
- str r10, [r0, #vp8_writer_buffer]
-
-mb_row_loop
-
- ldr r1, [r7, #tokenlist_start]
- ldr r9, [r7, #tokenlist_stop]
- str r9, [sp, #0] ; save stop for later comparison
- str r7, [sp, #16] ; tokenlist address for next time
-
- b check_p_lt_stop
-
- ; actual work gets done here!
-
-while_p_lt_stop
- ldrb r6, [r1, #tokenextra_token] ; t
- ldr r4, [sp, #80] ; vp8_coef_encodings
- mov lr, #0
- add r4, r4, r6, lsl #3 ; a = vp8_coef_encodings + t
- ldr r9, [r1, #tokenextra_context_tree] ; pp
-
- ldrb r7, [r1, #tokenextra_skip_eob_node]
-
- ldr r6, [r4, #vp8_token_value] ; v
- ldr r8, [r4, #vp8_token_len] ; n
-
- ; vp8 specific skip_eob_node
- cmp r7, #0
- movne lr, #2 ; i = 2
- subne r8, r8, #1 ; --n
-
- rsb r4, r8, #32 ; 32-n
- ldr r10, [sp, #88] ; vp8_coef_tree
-
- ; v is kept in r12 during the token pack loop
- lsl r12, r6, r4 ; r12 = v << 32 - n
-
-; loop start
-token_loop
- ldrb r4, [r9, lr, asr #1] ; pp [i>>1]
- sub r7, r5, #1 ; range-1
-
- ; Decisions are made based on the bit value shifted
- ; off of v, so set a flag here based on this.
- ; This value is refered to as "bb"
- lsls r12, r12, #1 ; bb = v >> n
- mul r4, r4, r7 ; ((range-1) * pp[i>>1]))
-
- ; bb can only be 0 or 1. So only execute this statement
- ; if bb == 1, otherwise it will act like i + 0
- addcs lr, lr, #1 ; i + bb
-
- mov r7, #1
- ldrsb lr, [r10, lr] ; i = vp8_coef_tree[i+bb]
- add r4, r7, r4, lsr #8 ; 1 + (((range-1) * pp[i>>1]) >> 8)
-
- addcs r2, r2, r4 ; if (bb) lowvalue += split
- subcs r4, r5, r4 ; if (bb) range = range-split
-
- ; Counting the leading zeros is used to normalize range.
- clz r6, r4
- sub r6, r6, #24 ; shift
-
- ; Flag is set on the sum of count. This flag is used later
- ; to determine if count >= 0
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi token_count_lt_zero ; if(count >= 0)
-
- sub r6, r6, r3 ; offset = shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl token_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos-1
- b token_zero_while_start
-token_zero_while_loop
- mov r10, #0
- strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-token_zero_while_start
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r11, [r7, r4]
- cmpge r11, #0xff
- beq token_zero_while_loop
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r10, [r7, r4] ; w->buffer[x]
- add r10, r10, #1
- strb r10, [r7, r4] ; w->buffer[x] + 1
-token_high_bit_not_set
- rsb r4, r6, #24 ; 24-offset
- ldr r10, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos] ; w->pos
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r11, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r11, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r10, r4] ; w->buffer[w->pos++]
-
- ; r10 is used earlier in the loop, but r10 is used as
- ; temp variable here. So after r10 is used, reload
- ; vp8_coef_tree_dcd into r10
- ldr r10, [sp, #88] ; vp8_coef_tree
-
-token_count_lt_zero
- lsl r2, r2, r6 ; lowvalue <<= shift
-
- subs r8, r8, #1 ; --n
- bne token_loop
-
- ldrb r6, [r1, #tokenextra_token] ; t
- ldr r7, [sp, #84] ; vp8_extra_bits
- ; Add t * sizeof (vp8_extra_bit_struct) to get the desired
- ; element. Here vp8_extra_bit_struct == 16
- add r12, r7, r6, lsl #4 ; b = vp8_extra_bits + t
-
- ldr r4, [r12, #vp8_extra_bit_struct_base_val]
- cmp r4, #0
- beq skip_extra_bits
-
-; if( b->base_val)
- ldr r8, [r12, #vp8_extra_bit_struct_len] ; L
- ldrsh lr, [r1, #tokenextra_extra] ; e = p->Extra
- cmp r8, #0 ; if( L)
- beq no_extra_bits
-
- ldr r9, [r12, #vp8_extra_bit_struct_prob]
- asr r7, lr, #1 ; v=e>>1
-
- ldr r10, [r12, #vp8_extra_bit_struct_tree]
- str r10, [sp, #4] ; b->tree
-
- rsb r4, r8, #32
- lsl r12, r7, r4
-
- mov lr, #0 ; i = 0
-
-extra_bits_loop
- ldrb r4, [r9, lr, asr #1] ; pp[i>>1]
- sub r7, r5, #1 ; range-1
- lsls r12, r12, #1 ; v >> n
- mul r4, r4, r7 ; (range-1) * pp[i>>1]
- addcs lr, lr, #1 ; i + bb
-
- mov r7, #1
- ldrsb lr, [r10, lr] ; i = b->tree[i+bb]
- add r4, r7, r4, lsr #8 ; split = 1 + (((range-1) * pp[i>>1]) >> 8)
-
- addcs r2, r2, r4 ; if (bb) lowvalue += split
- subcs r4, r5, r4 ; if (bb) range = range-split
-
- clz r6, r4
- sub r6, r6, #24
-
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi extra_count_lt_zero ; if(count >= 0)
-
- sub r6, r6, r3 ; offset= shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl extra_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos - 1
- b extra_zero_while_start
-extra_zero_while_loop
- mov r10, #0
- strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-extra_zero_while_start
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r11, [r7, r4]
- cmpge r11, #0xff
- beq extra_zero_while_loop
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r10, [r7, r4]
- add r10, r10, #1
- strb r10, [r7, r4]
-extra_high_bit_not_set
- rsb r4, r6, #24 ; 24-offset
- ldr r10, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos]
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r11, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r11, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r10, r4] ; w->buffer[w->pos++]=(lowvalue >> (24-offset))
- ldr r10, [sp, #4] ; b->tree
-extra_count_lt_zero
- lsl r2, r2, r6
-
- subs r8, r8, #1 ; --n
- bne extra_bits_loop ; while (n)
-
-no_extra_bits
- ldr lr, [r1, #4] ; e = p->Extra
- add r4, r5, #1 ; range + 1
- tst lr, #1
- lsr r4, r4, #1 ; split = (range + 1) >> 1
- addne r2, r2, r4 ; lowvalue += split
- subne r4, r5, r4 ; range = range-split
- tst r2, #0x80000000 ; lowvalue & 0x80000000
- lsl r5, r4, #1 ; range <<= 1
- beq end_high_bit_not_set
-
- ldr r4, [r0, #vp8_writer_pos]
- mov r7, #0
- sub r4, r4, #1
- b end_zero_while_start
-end_zero_while_loop
- strb r7, [r6, r4]
- sub r4, r4, #1 ; x--
-end_zero_while_start
- cmp r4, #0
- ldrge r6, [r0, #vp8_writer_buffer]
- ldrb r12, [r6, r4]
- cmpge r12, #0xff
- beq end_zero_while_loop
-
- ldr r6, [r0, #vp8_writer_buffer]
- ldrb r7, [r6, r4]
- add r7, r7, #1
- strb r7, [r6, r4]
-end_high_bit_not_set
- adds r3, r3, #1 ; ++count
- lsl r2, r2, #1 ; lowvalue <<= 1
- bne end_count_zero
-
- ldr r4, [r0, #vp8_writer_pos]
- mvn r3, #7
- ldr r7, [r0, #vp8_writer_buffer]
- lsr r6, r2, #24 ; lowvalue >> 24
- add r12, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r12, [r0, #0x10]
- strb r6, [r7, r4]
-end_count_zero
-skip_extra_bits
- add r1, r1, #TOKENEXTRA_SZ ; ++p
-check_p_lt_stop
- ldr r4, [sp, #0] ; stop
- cmp r1, r4 ; while( p < stop)
- bcc while_p_lt_stop
-
- ldr r10, [sp, #20] ; num_parts
- mov r1, #TOKENLIST_SZ
- mul r1, r10, r1
-
- ldr r6, [sp, #12] ; mb_rows
- ldr r7, [sp, #16] ; tokenlist address
- subs r6, r6, r10
- add r7, r7, r1 ; next element in the array
- str r6, [sp, #12]
- bgt mb_row_loop
-
- mov r12, #32
-
-stop_encode_loop
- sub r7, r5, #1 ; range-1
-
- mov r4, r7, lsl #7 ; ((range-1) * 128)
-
- mov r7, #1
- add r4, r7, r4, lsr #8 ; 1 + (((range-1) * 128) >> 8)
-
- ; Counting the leading zeros is used to normalize range.
- clz r6, r4
- sub r6, r6, #24 ; shift
-
- ; Flag is set on the sum of count. This flag is used later
- ; to determine if count >= 0
- adds r3, r3, r6 ; count += shift
- lsl r5, r4, r6 ; range <<= shift
- bmi token_count_lt_zero_se ; if(count >= 0)
-
- sub r6, r6, r3 ; offset = shift - count
- sub r4, r6, #1 ; offset-1
- lsls r4, r2, r4 ; if((lowvalue<<(offset-1)) & 0x80000000 )
- bpl token_high_bit_not_set_se
-
- ldr r4, [r0, #vp8_writer_pos] ; x
- sub r4, r4, #1 ; x = w->pos-1
- b token_zero_while_start_se
-token_zero_while_loop_se
- mov r10, #0
- strb r10, [r7, r4] ; w->buffer[x] =(unsigned char)0
- sub r4, r4, #1 ; x--
-token_zero_while_start_se
- cmp r4, #0
- ldrge r7, [r0, #vp8_writer_buffer]
- ldrb r11, [r7, r4]
- cmpge r11, #0xff
- beq token_zero_while_loop_se
-
- ldr r7, [r0, #vp8_writer_buffer]
- ldrb r10, [r7, r4] ; w->buffer[x]
- add r10, r10, #1
- strb r10, [r7, r4] ; w->buffer[x] + 1
-token_high_bit_not_set_se
- rsb r4, r6, #24 ; 24-offset
- ldr r10, [r0, #vp8_writer_buffer]
- lsr r7, r2, r4 ; lowvalue >> (24-offset)
- ldr r4, [r0, #vp8_writer_pos] ; w->pos
- lsl r2, r2, r6 ; lowvalue <<= offset
- mov r6, r3 ; shift = count
- add r11, r4, #1 ; w->pos++
- bic r2, r2, #0xff000000 ; lowvalue &= 0xffffff
- str r11, [r0, #vp8_writer_pos]
- sub r3, r3, #8 ; count -= 8
- strb r7, [r10, r4] ; w->buffer[w->pos++]
-
-token_count_lt_zero_se
- lsl r2, r2, r6 ; lowvalue <<= shift
-
- subs r12, r12, #1
- bne stop_encode_loop
-
- ldr r10, [sp, #8] ; *size
- ldr r11, [r10]
- ldr r4, [r0, #vp8_writer_pos] ; w->pos
- add r11, r11, r4 ; *size += w->pos
- str r11, [r10]
-
- ldr r9, [sp, #20] ; num_parts
- sub r9, r9, #1
- ldr r10, [sp, #28] ; i
- cmp r10, r9 ; if(i<(num_part - 1))
- bge skip_write_partition
-
- ldr r12, [sp, #40] ; ptr
- add r12, r12, r4 ; ptr += w->pos
- str r12, [sp, #40]
-
- ldr r9, [sp, #24] ; cx_data
- mov r8, r4, asr #8
- strb r4, [r9, #0]
- strb r8, [r9, #1]
- mov r4, r4, asr #16
- strb r4, [r9, #2]
-
- add r9, r9, #3 ; cx_data += 3
- str r9, [sp, #24]
-
-skip_write_partition
-
- ldr r11, [sp, #28] ; i
- ldr r10, [sp, #20] ; num_parts
-
- add r11, r11, #1 ; i++
- str r11, [sp, #28]
-
- ldr r7, [sp, #32] ; cpi->tp_list[i]
- mov r1, #TOKENLIST_SZ
- add r7, r7, r1 ; next element in cpi->tp_list
- str r7, [sp, #32] ; cpi->tp_list[i+1]
-
- cmp r10, r11
- bgt numparts_loop
-
-
- add sp, sp, #44
- pop {r4-r11, pc}
- ENDP
-
-_VP8_COMP_common_
- DCD vp8_comp_common
-_VP8_COMMON_MBrows_
- DCD vp8_common_mb_rows
-_VP8_COMP_tplist_
- DCD vp8_comp_tplist
-_VP8_COMP_bc2_
- DCD vp8_comp_bc2
-
- END
diff --git a/vp8/encoder/arm/armv6/vp8_fast_fdct4x4_armv6.asm b/vp8/encoder/arm/armv6/vp8_fast_fdct4x4_armv6.asm
deleted file mode 100644
index 65bd2b4..0000000
--- a/vp8/encoder/arm/armv6/vp8_fast_fdct4x4_armv6.asm
+++ /dev/null
@@ -1,262 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
- EXPORT |vp8_fast_fdct4x4_armv6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA |.text|, CODE, READONLY
-; void vp8_short_fdct4x4_c(short *input, short *output, int pitch)
-|vp8_fast_fdct4x4_armv6| PROC
-
- stmfd sp!, {r4 - r12, lr}
-
- ; PART 1
-
- ; coeffs 0-3
- ldrd r4, r5, [r0] ; [i1 | i0] [i3 | i2]
-
- ldr r10, c7500
- ldr r11, c14500
- ldr r12, c0x22a453a0 ; [2217*4 | 5352*4]
- ldr lr, c0x00080008
- ror r5, r5, #16 ; [i2 | i3]
-
- qadd16 r6, r4, r5 ; [i1+i2 | i0+i3] = [b1 | a1] without shift
- qsub16 r7, r4, r5 ; [i1-i2 | i0-i3] = [c1 | d1] without shift
-
- add r0, r0, r2 ; update input pointer
-
- qadd16 r7, r7, r7 ; 2*[c1|d1] --> we can use smlad and smlsd
- ; with 2217*4 and 5352*4 without losing the
- ; sign bit (overflow)
-
- smuad r4, r6, lr ; o0 = (i1+i2)*8 + (i0+i3)*8
- smusd r5, r6, lr ; o2 = (i1+i2)*8 - (i0+i3)*8
-
- smlad r6, r7, r12, r11 ; o1 = (c1 * 2217 + d1 * 5352 + 14500)
- smlsdx r7, r7, r12, r10 ; o3 = (d1 * 2217 - c1 * 5352 + 7500)
-
- ldrd r8, r9, [r0] ; [i5 | i4] [i7 | i6]
-
- pkhbt r3, r4, r6, lsl #4 ; [o1 | o0], keep in register for PART 2
- pkhbt r6, r5, r7, lsl #4 ; [o3 | o2]
-
- str r6, [r1, #4]
-
- ; coeffs 4-7
- ror r9, r9, #16 ; [i6 | i7]
-
- qadd16 r6, r8, r9 ; [i5+i6 | i4+i7] = [b1 | a1] without shift
- qsub16 r7, r8, r9 ; [i5-i6 | i4-i7] = [c1 | d1] without shift
-
- add r0, r0, r2 ; update input pointer
-
- qadd16 r7, r7, r7 ; 2x[c1|d1] --> we can use smlad and smlsd
- ; with 2217*4 and 5352*4 without losing the
- ; sign bit (overflow)
-
- smuad r9, r6, lr ; o4 = (i5+i6)*8 + (i4+i7)*8
- smusd r8, r6, lr ; o6 = (i5+i6)*8 - (i4+i7)*8
-
- smlad r6, r7, r12, r11 ; o5 = (c1 * 2217 + d1 * 5352 + 14500)
- smlsdx r7, r7, r12, r10 ; o7 = (d1 * 2217 - c1 * 5352 + 7500)
-
- ldrd r4, r5, [r0] ; [i9 | i8] [i11 | i10]
-
- pkhbt r9, r9, r6, lsl #4 ; [o5 | o4], keep in register for PART 2
- pkhbt r6, r8, r7, lsl #4 ; [o7 | o6]
-
- str r6, [r1, #12]
-
- ; coeffs 8-11
- ror r5, r5, #16 ; [i10 | i11]
-
- qadd16 r6, r4, r5 ; [i9+i10 | i8+i11]=[b1 | a1] without shift
- qsub16 r7, r4, r5 ; [i9-i10 | i8-i11]=[c1 | d1] without shift
-
- add r0, r0, r2 ; update input pointer
-
- qadd16 r7, r7, r7 ; 2x[c1|d1] --> we can use smlad and smlsd
- ; with 2217*4 and 5352*4 without losing the
- ; sign bit (overflow)
-
- smuad r2, r6, lr ; o8 = (i9+i10)*8 + (i8+i11)*8
- smusd r8, r6, lr ; o10 = (i9+i10)*8 - (i8+i11)*8
-
- smlad r6, r7, r12, r11 ; o9 = (c1 * 2217 + d1 * 5352 + 14500)
- smlsdx r7, r7, r12, r10 ; o11 = (d1 * 2217 - c1 * 5352 + 7500)
-
- ldrd r4, r5, [r0] ; [i13 | i12] [i15 | i14]
-
- pkhbt r2, r2, r6, lsl #4 ; [o9 | o8], keep in register for PART 2
- pkhbt r6, r8, r7, lsl #4 ; [o11 | o10]
-
- str r6, [r1, #20]
-
- ; coeffs 12-15
- ror r5, r5, #16 ; [i14 | i15]
-
- qadd16 r6, r4, r5 ; [i13+i14 | i12+i15]=[b1|a1] without shift
- qsub16 r7, r4, r5 ; [i13-i14 | i12-i15]=[c1|d1] without shift
-
- qadd16 r7, r7, r7 ; 2x[c1|d1] --> we can use smlad and smlsd
- ; with 2217*4 and 5352*4 without losing the
- ; sign bit (overflow)
-
- smuad r4, r6, lr ; o12 = (i13+i14)*8 + (i12+i15)*8
- smusd r5, r6, lr ; o14 = (i13+i14)*8 - (i12+i15)*8
-
- smlad r6, r7, r12, r11 ; o13 = (c1 * 2217 + d1 * 5352 + 14500)
- smlsdx r7, r7, r12, r10 ; o15 = (d1 * 2217 - c1 * 5352 + 7500)
-
- pkhbt r0, r4, r6, lsl #4 ; [o13 | o12], keep in register for PART 2
- pkhbt r6, r5, r7, lsl #4 ; [o15 | o14]
-
- str r6, [r1, #28]
-
-
- ; PART 2 -------------------------------------------------
- ldr r11, c12000
- ldr r10, c51000
- ldr lr, c0x00070007
-
- qadd16 r4, r3, r0 ; a1 = [i1+i13 | i0+i12]
- qadd16 r5, r9, r2 ; b1 = [i5+i9 | i4+i8]
- qsub16 r6, r9, r2 ; c1 = [i5-i9 | i4-i8]
- qsub16 r7, r3, r0 ; d1 = [i1-i13 | i0-i12]
-
- qadd16 r4, r4, lr ; a1 + 7
-
- add r0, r11, #0x10000 ; add (d!=0)
-
- qadd16 r2, r4, r5 ; a1 + b1 + 7
- qsub16 r3, r4, r5 ; a1 - b1 + 7
-
- ldr r12, c0x08a914e8 ; [2217 | 5352]
-
- lsl r8, r2, #16 ; prepare bottom halfword for scaling
- asr r2, r2, #4 ; scale top halfword
- lsl r9, r3, #16 ; prepare bottom halfword for scaling
- asr r3, r3, #4 ; scale top halfword
- pkhtb r4, r2, r8, asr #20 ; pack and scale bottom halfword
- pkhtb r5, r3, r9, asr #20 ; pack and scale bottom halfword
-
- smulbt r2, r6, r12 ; [ ------ | c1*2217]
- str r4, [r1, #0] ; [ o1 | o0]
- smultt r3, r6, r12 ; [c1*2217 | ------ ]
- str r5, [r1, #16] ; [ o9 | o8]
-
- smlabb r8, r7, r12, r2 ; [ ------ | d1*5352]
- smlatb r9, r7, r12, r3 ; [d1*5352 | ------ ]
-
- smulbb r2, r6, r12 ; [ ------ | c1*5352]
- smultb r3, r6, r12 ; [c1*5352 | ------ ]
-
- lsls r6, r7, #16 ; d1 != 0 ?
- addeq r8, r8, r11 ; c1_b*2217+d1_b*5352+12000 + (d==0)
- addne r8, r8, r0 ; c1_b*2217+d1_b*5352+12000 + (d!=0)
- asrs r6, r7, #16
- addeq r9, r9, r11 ; c1_t*2217+d1_t*5352+12000 + (d==0)
- addne r9, r9, r0 ; c1_t*2217+d1_t*5352+12000 + (d!=0)
-
- smlabt r4, r7, r12, r10 ; [ ------ | d1*2217] + 51000
- smlatt r5, r7, r12, r10 ; [d1*2217 | ------ ] + 51000
-
- pkhtb r9, r9, r8, asr #16
-
- sub r4, r4, r2
- sub r5, r5, r3
-
- ldr r3, [r1, #4] ; [i3 | i2]
-
- pkhtb r5, r5, r4, asr #16 ; [o13|o12]
-
- str r9, [r1, #8] ; [o5 | 04]
-
- ldr r9, [r1, #12] ; [i7 | i6]
- ldr r8, [r1, #28] ; [i15|i14]
- ldr r2, [r1, #20] ; [i11|i10]
- str r5, [r1, #24] ; [o13|o12]
-
- qadd16 r4, r3, r8 ; a1 = [i3+i15 | i2+i14]
- qadd16 r5, r9, r2 ; b1 = [i7+i11 | i6+i10]
-
- qadd16 r4, r4, lr ; a1 + 7
-
- qsub16 r6, r9, r2 ; c1 = [i7-i11 | i6-i10]
- qadd16 r2, r4, r5 ; a1 + b1 + 7
- qsub16 r7, r3, r8 ; d1 = [i3-i15 | i2-i14]
- qsub16 r3, r4, r5 ; a1 - b1 + 7
-
- lsl r8, r2, #16 ; prepare bottom halfword for scaling
- asr r2, r2, #4 ; scale top halfword
- lsl r9, r3, #16 ; prepare bottom halfword for scaling
- asr r3, r3, #4 ; scale top halfword
- pkhtb r4, r2, r8, asr #20 ; pack and scale bottom halfword
- pkhtb r5, r3, r9, asr #20 ; pack and scale bottom halfword
-
- smulbt r2, r6, r12 ; [ ------ | c1*2217]
- str r4, [r1, #4] ; [ o3 | o2]
- smultt r3, r6, r12 ; [c1*2217 | ------ ]
- str r5, [r1, #20] ; [ o11 | o10]
-
- smlabb r8, r7, r12, r2 ; [ ------ | d1*5352]
- smlatb r9, r7, r12, r3 ; [d1*5352 | ------ ]
-
- smulbb r2, r6, r12 ; [ ------ | c1*5352]
- smultb r3, r6, r12 ; [c1*5352 | ------ ]
-
- lsls r6, r7, #16 ; d1 != 0 ?
- addeq r8, r8, r11 ; c1_b*2217+d1_b*5352+12000 + (d==0)
- addne r8, r8, r0 ; c1_b*2217+d1_b*5352+12000 + (d!=0)
-
- asrs r6, r7, #16
- addeq r9, r9, r11 ; c1_t*2217+d1_t*5352+12000 + (d==0)
- addne r9, r9, r0 ; c1_t*2217+d1_t*5352+12000 + (d!=0)
-
- smlabt r4, r7, r12, r10 ; [ ------ | d1*2217] + 51000
- smlatt r5, r7, r12, r10 ; [d1*2217 | ------ ] + 51000
-
- pkhtb r9, r9, r8, asr #16
-
- sub r4, r4, r2
- sub r5, r5, r3
-
- str r9, [r1, #12] ; [o7 | o6]
- pkhtb r5, r5, r4, asr #16 ; [o15|o14]
-
- str r5, [r1, #28] ; [o15|o14]
-
- ldmfd sp!, {r4 - r12, pc}
-
- ENDP
-
-; Used constants
-c7500
- DCD 7500
-c14500
- DCD 14500
-c0x22a453a0
- DCD 0x22a453a0
-c0x00080008
- DCD 0x00080008
-c12000
- DCD 12000
-c51000
- DCD 51000
-c0x00070007
- DCD 0x00070007
-c0x08a914e8
- DCD 0x08a914e8
-
- END
diff --git a/vp8/encoder/arm/armv6/vp8_fast_quantize_b_armv6.asm b/vp8/encoder/arm/armv6/vp8_fast_quantize_b_armv6.asm
deleted file mode 100644
index ae2f603..0000000
--- a/vp8/encoder/arm/armv6/vp8_fast_quantize_b_armv6.asm
+++ /dev/null
@@ -1,224 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_fast_quantize_b_armv6|
-
- INCLUDE asm_enc_offsets.asm
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 BLOCK *b
-; r1 BLOCKD *d
-|vp8_fast_quantize_b_armv6| PROC
- stmfd sp!, {r1, r4-r11, lr}
-
- ldr r3, [r0, #vp8_block_coeff] ; coeff
- ldr r4, [r0, #vp8_block_quant_fast] ; quant_fast
- ldr r5, [r0, #vp8_block_round] ; round
- ldr r6, [r1, #vp8_blockd_qcoeff] ; qcoeff
- ldr r7, [r1, #vp8_blockd_dqcoeff] ; dqcoeff
- ldr r8, [r1, #vp8_blockd_dequant] ; dequant
-
- ldr r2, loop_count ; loop_count=0x1000000. 'lsls' instruction
- ; is used to update the counter so that
- ; it can be used to mark nonzero
- ; quantized coefficient pairs.
-
- mov r1, #0 ; flags for quantized coeffs
-
- ; PART 1: quantization and dequantization loop
-loop
- ldr r9, [r3], #4 ; [z1 | z0]
- ldr r10, [r5], #4 ; [r1 | r0]
- ldr r11, [r4], #4 ; [q1 | q0]
-
- ssat16 lr, #1, r9 ; [sz1 | sz0]
- eor r9, r9, lr ; [z1 ^ sz1 | z0 ^ sz0]
- ssub16 r9, r9, lr ; x = (z ^ sz) - sz
- sadd16 r9, r9, r10 ; [x1+r1 | x0+r0]
-
- ldr r12, [r3], #4 ; [z3 | z2]
-
- smulbb r0, r9, r11 ; [(x0+r0)*q0]
- smultt r9, r9, r11 ; [(x1+r1)*q1]
-
- ldr r10, [r5], #4 ; [r3 | r2]
-
- ssat16 r11, #1, r12 ; [sz3 | sz2]
- eor r12, r12, r11 ; [z3 ^ sz3 | z2 ^ sz2]
- pkhtb r0, r9, r0, asr #16 ; [y1 | y0]
- ldr r9, [r4], #4 ; [q3 | q2]
- ssub16 r12, r12, r11 ; x = (z ^ sz) - sz
-
- sadd16 r12, r12, r10 ; [x3+r3 | x2+r2]
-
- eor r0, r0, lr ; [(y1 ^ sz1) | (y0 ^ sz0)]
-
- smulbb r10, r12, r9 ; [(x2+r2)*q2]
- smultt r12, r12, r9 ; [(x3+r3)*q3]
-
- ssub16 r0, r0, lr ; x = (y ^ sz) - sz
-
- cmp r0, #0 ; check if zero
- orrne r1, r1, r2, lsr #24 ; add flag for nonzero coeffs
-
- str r0, [r6], #4 ; *qcoeff++ = x
- ldr r9, [r8], #4 ; [dq1 | dq0]
-
- pkhtb r10, r12, r10, asr #16 ; [y3 | y2]
- eor r10, r10, r11 ; [(y3 ^ sz3) | (y2 ^ sz2)]
- ssub16 r10, r10, r11 ; x = (y ^ sz) - sz
-
- cmp r10, #0 ; check if zero
- orrne r1, r1, r2, lsr #23 ; add flag for nonzero coeffs
-
- str r10, [r6], #4 ; *qcoeff++ = x
- ldr r11, [r8], #4 ; [dq3 | dq2]
-
- smulbb r12, r0, r9 ; [x0*dq0]
- smultt r0, r0, r9 ; [x1*dq1]
-
- smulbb r9, r10, r11 ; [x2*dq2]
- smultt r10, r10, r11 ; [x3*dq3]
-
- lsls r2, r2, #2 ; update loop counter
- strh r12, [r7, #0] ; dqcoeff[0] = [x0*dq0]
- strh r0, [r7, #2] ; dqcoeff[1] = [x1*dq1]
- strh r9, [r7, #4] ; dqcoeff[2] = [x2*dq2]
- strh r10, [r7, #6] ; dqcoeff[3] = [x3*dq3]
- add r7, r7, #8 ; dqcoeff += 8
- bne loop
-
- ; PART 2: check position for eob...
- mov lr, #0 ; init eob
- cmp r1, #0 ; coeffs after quantization?
- ldr r11, [sp, #0] ; restore BLOCKD pointer
- beq end ; skip eob calculations if all zero
-
- ldr r0, [r11, #vp8_blockd_qcoeff]
-
- ; check shortcut for nonzero qcoeffs
- tst r1, #0x80
- bne quant_coeff_15_14
- tst r1, #0x20
- bne quant_coeff_13_11
- tst r1, #0x8
- bne quant_coeff_12_7
- tst r1, #0x40
- bne quant_coeff_10_9
- tst r1, #0x10
- bne quant_coeff_8_3
- tst r1, #0x2
- bne quant_coeff_6_5
- tst r1, #0x4
- bne quant_coeff_4_2
- b quant_coeff_1_0
-
-quant_coeff_15_14
- ldrh r2, [r0, #30] ; rc=15, i=15
- mov lr, #16
- cmp r2, #0
- bne end
-
- ldrh r3, [r0, #28] ; rc=14, i=14
- mov lr, #15
- cmp r3, #0
- bne end
-
-quant_coeff_13_11
- ldrh r2, [r0, #22] ; rc=11, i=13
- mov lr, #14
- cmp r2, #0
- bne end
-
-quant_coeff_12_7
- ldrh r3, [r0, #14] ; rc=7, i=12
- mov lr, #13
- cmp r3, #0
- bne end
-
- ldrh r2, [r0, #20] ; rc=10, i=11
- mov lr, #12
- cmp r2, #0
- bne end
-
-quant_coeff_10_9
- ldrh r3, [r0, #26] ; rc=13, i=10
- mov lr, #11
- cmp r3, #0
- bne end
-
- ldrh r2, [r0, #24] ; rc=12, i=9
- mov lr, #10
- cmp r2, #0
- bne end
-
-quant_coeff_8_3
- ldrh r3, [r0, #18] ; rc=9, i=8
- mov lr, #9
- cmp r3, #0
- bne end
-
- ldrh r2, [r0, #12] ; rc=6, i=7
- mov lr, #8
- cmp r2, #0
- bne end
-
-quant_coeff_6_5
- ldrh r3, [r0, #6] ; rc=3, i=6
- mov lr, #7
- cmp r3, #0
- bne end
-
- ldrh r2, [r0, #4] ; rc=2, i=5
- mov lr, #6
- cmp r2, #0
- bne end
-
-quant_coeff_4_2
- ldrh r3, [r0, #10] ; rc=5, i=4
- mov lr, #5
- cmp r3, #0
- bne end
-
- ldrh r2, [r0, #16] ; rc=8, i=3
- mov lr, #4
- cmp r2, #0
- bne end
-
- ldrh r3, [r0, #8] ; rc=4, i=2
- mov lr, #3
- cmp r3, #0
- bne end
-
-quant_coeff_1_0
- ldrh r2, [r0, #2] ; rc=1, i=1
- mov lr, #2
- cmp r2, #0
- bne end
-
- mov lr, #1 ; rc=0, i=0
-
-end
- str lr, [r11, #vp8_blockd_eob]
- ldmfd sp!, {r1, r4-r11, pc}
-
- ENDP
-
-loop_count
- DCD 0x1000000
-
- END
-
diff --git a/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm b/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm
deleted file mode 100644
index a9060d7..0000000
--- a/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm
+++ /dev/null
@@ -1,133 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_mse16x16_armv6|
-
- ARM
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-;
-;note: Based on vp8_variance16x16_armv6. In this function, sum is never used.
-; So, we can remove this part of calculation.
-
-|vp8_mse16x16_armv6| PROC
-
- push {r4-r9, lr}
- mov r12, #16 ; set loop counter to 16 (=block height)
-
- mov r4, #0 ; initialize sse = 0
-
-loop
- ; 1st 4 pixels
- ldr r5, [r0, #0x0] ; load 4 src pixels
- ldr r6, [r2, #0x0] ; load 4 ref pixels
-
- mov lr, #0 ; constant zero
-
- usub8 r8, r5, r6 ; calculate difference
- sel r7, r8, lr ; select bytes with positive difference
- usub8 r9, r6, r5 ; calculate difference with reversed operands
- sel r8, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r5, r7, lr ; calculate sum of positive differences
- usad8 r6, r8, lr ; calculate sum of negative differences
- orr r8, r8, r7 ; differences of all 4 pixels
-
- ldr r5, [r0, #0x4] ; load 4 src pixels
-
- ; calculate sse
- uxtb16 r6, r8 ; byte (two pixels) to halfwords
- uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
- smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
-
- ; 2nd 4 pixels
- ldr r6, [r2, #0x4] ; load 4 ref pixels
- smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
-
- usub8 r8, r5, r6 ; calculate difference
- sel r7, r8, lr ; select bytes with positive difference
- usub8 r9, r6, r5 ; calculate difference with reversed operands
- sel r8, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r5, r7, lr ; calculate sum of positive differences
- usad8 r6, r8, lr ; calculate sum of negative differences
- orr r8, r8, r7 ; differences of all 4 pixels
- ldr r5, [r0, #0x8] ; load 4 src pixels
- ; calculate sse
- uxtb16 r6, r8 ; byte (two pixels) to halfwords
- uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
- smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
-
- ; 3rd 4 pixels
- ldr r6, [r2, #0x8] ; load 4 ref pixels
- smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
-
- usub8 r8, r5, r6 ; calculate difference
- sel r7, r8, lr ; select bytes with positive difference
- usub8 r9, r6, r5 ; calculate difference with reversed operands
- sel r8, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r5, r7, lr ; calculate sum of positive differences
- usad8 r6, r8, lr ; calculate sum of negative differences
- orr r8, r8, r7 ; differences of all 4 pixels
-
- ldr r5, [r0, #0xc] ; load 4 src pixels
-
- ; calculate sse
- uxtb16 r6, r8 ; byte (two pixels) to halfwords
- uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
- smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
-
- ; 4th 4 pixels
- ldr r6, [r2, #0xc] ; load 4 ref pixels
- smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
-
- usub8 r8, r5, r6 ; calculate difference
- add r0, r0, r1 ; set src_ptr to next row
- sel r7, r8, lr ; select bytes with positive difference
- usub8 r9, r6, r5 ; calculate difference with reversed operands
- add r2, r2, r3 ; set dst_ptr to next row
- sel r8, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r5, r7, lr ; calculate sum of positive differences
- usad8 r6, r8, lr ; calculate sum of negative differences
- orr r8, r8, r7 ; differences of all 4 pixels
-
- subs r12, r12, #1 ; next row
-
- ; calculate sse
- uxtb16 r6, r8 ; byte (two pixels) to halfwords
- uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
- smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
- smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
-
- bne loop
-
- ; return stuff
- ldr r1, [sp, #28] ; get address of sse
- mov r0, r4 ; return sse
- str r4, [r1] ; store sse
-
- pop {r4-r9, pc}
-
- ENDP
-
- END
diff --git a/vp8/encoder/arm/armv6/vp8_sad16x16_armv6.asm b/vp8/encoder/arm/armv6/vp8_sad16x16_armv6.asm
deleted file mode 100644
index c759f7c..0000000
--- a/vp8/encoder/arm/armv6/vp8_sad16x16_armv6.asm
+++ /dev/null
@@ -1,84 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_sad16x16_armv6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 const unsigned char *src_ptr
-; r1 int src_stride
-; r2 const unsigned char *ref_ptr
-; r3 int ref_stride
-; stack max_sad (not used)
-|vp8_sad16x16_armv6| PROC
- stmfd sp!, {r4-r12, lr}
- mov r4, #0 ; sad = 0;
- mov r5, #8 ; loop count
-
-loop
- ; 1st row
- ldr r6, [r0, #0x0] ; load 4 src pixels (1A)
- ldr r8, [r2, #0x0] ; load 4 ref pixels (1A)
- ldr r7, [r0, #0x4] ; load 4 src pixels (1A)
- ldr r9, [r2, #0x4] ; load 4 ref pixels (1A)
- ldr r10, [r0, #0x8] ; load 4 src pixels (1B)
- ldr r11, [r0, #0xC] ; load 4 src pixels (1B)
-
- usada8 r4, r8, r6, r4 ; calculate sad for 4 pixels
- usad8 r8, r7, r9 ; calculate sad for 4 pixels
-
- ldr r12, [r2, #0x8] ; load 4 ref pixels (1B)
- ldr lr, [r2, #0xC] ; load 4 ref pixels (1B)
-
- add r0, r0, r1 ; set src pointer to next row
- add r2, r2, r3 ; set dst pointer to next row
-
- usada8 r4, r10, r12, r4 ; calculate sad for 4 pixels
- usada8 r8, r11, lr, r8 ; calculate sad for 4 pixels
-
- ldr r6, [r0, #0x0] ; load 4 src pixels (2A)
- ldr r7, [r0, #0x4] ; load 4 src pixels (2A)
- add r4, r4, r8 ; add partial sad values
-
- ; 2nd row
- ldr r8, [r2, #0x0] ; load 4 ref pixels (2A)
- ldr r9, [r2, #0x4] ; load 4 ref pixels (2A)
- ldr r10, [r0, #0x8] ; load 4 src pixels (2B)
- ldr r11, [r0, #0xC] ; load 4 src pixels (2B)
-
- usada8 r4, r6, r8, r4 ; calculate sad for 4 pixels
- usad8 r8, r7, r9 ; calculate sad for 4 pixels
-
- ldr r12, [r2, #0x8] ; load 4 ref pixels (2B)
- ldr lr, [r2, #0xC] ; load 4 ref pixels (2B)
-
- add r0, r0, r1 ; set src pointer to next row
- add r2, r2, r3 ; set dst pointer to next row
-
- usada8 r4, r10, r12, r4 ; calculate sad for 4 pixels
- usada8 r8, r11, lr, r8 ; calculate sad for 4 pixels
-
- subs r5, r5, #1 ; decrement loop counter
- add r4, r4, r8 ; add partial sad values
-
- bne loop
-
- mov r0, r4 ; return sad
- ldmfd sp!, {r4-r12, pc}
-
- ENDP
-
- END
-
diff --git a/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm b/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm
deleted file mode 100644
index 0ca7438..0000000
--- a/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm
+++ /dev/null
@@ -1,265 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_subtract_mby_armv6|
- EXPORT |vp8_subtract_mbuv_armv6|
- EXPORT |vp8_subtract_b_armv6|
-
- INCLUDE asm_enc_offsets.asm
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 BLOCK *be
-; r1 BLOCKD *bd
-; r2 int pitch
-|vp8_subtract_b_armv6| PROC
-
- stmfd sp!, {r4-r9}
-
- ldr r4, [r0, #vp8_block_base_src]
- ldr r5, [r0, #vp8_block_src]
- ldr r6, [r0, #vp8_block_src_diff]
-
- ldr r3, [r4]
- ldr r7, [r0, #vp8_block_src_stride]
- add r3, r3, r5 ; src = *base_src + src
- ldr r8, [r1, #vp8_blockd_predictor]
-
- mov r9, #4 ; loop count
-
-loop_block
-
- ldr r0, [r3], r7 ; src
- ldr r1, [r8], r2 ; pred
-
- uxtb16 r4, r0 ; [s2 | s0]
- uxtb16 r5, r1 ; [p2 | p0]
- uxtb16 r0, r0, ror #8 ; [s3 | s1]
- uxtb16 r1, r1, ror #8 ; [p3 | p1]
-
- usub16 r4, r4, r5 ; [d2 | d0]
- usub16 r5, r0, r1 ; [d3 | d1]
-
- subs r9, r9, #1 ; decrement loop counter
-
- pkhbt r0, r4, r5, lsl #16 ; [d1 | d0]
- pkhtb r1, r5, r4, asr #16 ; [d3 | d2]
-
- str r0, [r6, #0] ; diff
- str r1, [r6, #4] ; diff
-
- add r6, r6, r2, lsl #1 ; update diff pointer
- bne loop_block
-
- ldmfd sp!, {r4-r9}
- mov pc, lr
-
- ENDP
-
-
-; r0 short *diff
-; r1 unsigned char *usrc
-; r2 unsigned char *vsrc
-; r3 unsigned char *pred
-; stack int stride
-|vp8_subtract_mbuv_armv6| PROC
-
- stmfd sp!, {r4-r12, lr}
-
- add r0, r0, #512 ; set *diff point to Cb
- add r3, r3, #256 ; set *pred point to Cb
-
- mov r4, #8 ; loop count
- ldr r5, [sp, #40] ; stride
-
- ; Subtract U block
-loop_u
- ldr r6, [r1] ; src (A)
- ldr r7, [r3], #4 ; pred (A)
-
- uxtb16 r8, r6 ; [s2 | s0] (A)
- uxtb16 r9, r7 ; [p2 | p0] (A)
- uxtb16 r10, r6, ror #8 ; [s3 | s1] (A)
- uxtb16 r11, r7, ror #8 ; [p3 | p1] (A)
-
- usub16 r6, r8, r9 ; [d2 | d0] (A)
- usub16 r7, r10, r11 ; [d3 | d1] (A)
-
- ldr r10, [r1, #4] ; src (B)
- ldr r11, [r3], #4 ; pred (B)
-
- pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (A)
- pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (A)
-
- str r8, [r0], #4 ; diff (A)
- uxtb16 r8, r10 ; [s2 | s0] (B)
- str r9, [r0], #4 ; diff (A)
-
- uxtb16 r9, r11 ; [p2 | p0] (B)
- uxtb16 r10, r10, ror #8 ; [s3 | s1] (B)
- uxtb16 r11, r11, ror #8 ; [p3 | p1] (B)
-
- usub16 r6, r8, r9 ; [d2 | d0] (B)
- usub16 r7, r10, r11 ; [d3 | d1] (B)
-
- add r1, r1, r5 ; update usrc pointer
-
- pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (B)
- pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (B)
-
- str r8, [r0], #4 ; diff (B)
- subs r4, r4, #1 ; update loop counter
- str r9, [r0], #4 ; diff (B)
-
- bne loop_u
-
- mov r4, #8 ; loop count
-
- ; Subtract V block
-loop_v
- ldr r6, [r2] ; src (A)
- ldr r7, [r3], #4 ; pred (A)
-
- uxtb16 r8, r6 ; [s2 | s0] (A)
- uxtb16 r9, r7 ; [p2 | p0] (A)
- uxtb16 r10, r6, ror #8 ; [s3 | s1] (A)
- uxtb16 r11, r7, ror #8 ; [p3 | p1] (A)
-
- usub16 r6, r8, r9 ; [d2 | d0] (A)
- usub16 r7, r10, r11 ; [d3 | d1] (A)
-
- ldr r10, [r2, #4] ; src (B)
- ldr r11, [r3], #4 ; pred (B)
-
- pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (A)
- pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (A)
-
- str r8, [r0], #4 ; diff (A)
- uxtb16 r8, r10 ; [s2 | s0] (B)
- str r9, [r0], #4 ; diff (A)
-
- uxtb16 r9, r11 ; [p2 | p0] (B)
- uxtb16 r10, r10, ror #8 ; [s3 | s1] (B)
- uxtb16 r11, r11, ror #8 ; [p3 | p1] (B)
-
- usub16 r6, r8, r9 ; [d2 | d0] (B)
- usub16 r7, r10, r11 ; [d3 | d1] (B)
-
- add r2, r2, r5 ; update vsrc pointer
-
- pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (B)
- pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (B)
-
- str r8, [r0], #4 ; diff (B)
- subs r4, r4, #1 ; update loop counter
- str r9, [r0], #4 ; diff (B)
-
- bne loop_v
-
- ldmfd sp!, {r4-r12, pc}
-
- ENDP
-
-
-; r0 short *diff
-; r1 unsigned char *src
-; r2 unsigned char *pred
-; r3 int stride
-|vp8_subtract_mby_armv6| PROC
-
- stmfd sp!, {r4-r11}
-
- mov r4, #16
-loop
- ldr r6, [r1] ; src (A)
- ldr r7, [r2], #4 ; pred (A)
-
- uxtb16 r8, r6 ; [s2 | s0] (A)
- uxtb16 r9, r7 ; [p2 | p0] (A)
- uxtb16 r10, r6, ror #8 ; [s3 | s1] (A)
- uxtb16 r11, r7, ror #8 ; [p3 | p1] (A)
-
- usub16 r6, r8, r9 ; [d2 | d0] (A)
- usub16 r7, r10, r11 ; [d3 | d1] (A)
-
- ldr r10, [r1, #4] ; src (B)
- ldr r11, [r2], #4 ; pred (B)
-
- pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (A)
- pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (A)
-
- str r8, [r0], #4 ; diff (A)
- uxtb16 r8, r10 ; [s2 | s0] (B)
- str r9, [r0], #4 ; diff (A)
-
- uxtb16 r9, r11 ; [p2 | p0] (B)
- uxtb16 r10, r10, ror #8 ; [s3 | s1] (B)
- uxtb16 r11, r11, ror #8 ; [p3 | p1] (B)
-
- usub16 r6, r8, r9 ; [d2 | d0] (B)
- usub16 r7, r10, r11 ; [d3 | d1] (B)
-
- ldr r10, [r1, #8] ; src (C)
- ldr r11, [r2], #4 ; pred (C)
-
- pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (B)
- pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (B)
-
- str r8, [r0], #4 ; diff (B)
- uxtb16 r8, r10 ; [s2 | s0] (C)
- str r9, [r0], #4 ; diff (B)
-
- uxtb16 r9, r11 ; [p2 | p0] (C)
- uxtb16 r10, r10, ror #8 ; [s3 | s1] (C)
- uxtb16 r11, r11, ror #8 ; [p3 | p1] (C)
-
- usub16 r6, r8, r9 ; [d2 | d0] (C)
- usub16 r7, r10, r11 ; [d3 | d1] (C)
-
- ldr r10, [r1, #12] ; src (D)
- ldr r11, [r2], #4 ; pred (D)
-
- pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (C)
- pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (C)
-
- str r8, [r0], #4 ; diff (C)
- uxtb16 r8, r10 ; [s2 | s0] (D)
- str r9, [r0], #4 ; diff (C)
-
- uxtb16 r9, r11 ; [p2 | p0] (D)
- uxtb16 r10, r10, ror #8 ; [s3 | s1] (D)
- uxtb16 r11, r11, ror #8 ; [p3 | p1] (D)
-
- usub16 r6, r8, r9 ; [d2 | d0] (D)
- usub16 r7, r10, r11 ; [d3 | d1] (D)
-
- add r1, r1, r3 ; update src pointer
-
- pkhbt r8, r6, r7, lsl #16 ; [d1 | d0] (D)
- pkhtb r9, r7, r6, asr #16 ; [d3 | d2] (D)
-
- str r8, [r0], #4 ; diff (D)
- subs r4, r4, #1 ; update loop counter
- str r9, [r0], #4 ; diff (D)
-
- bne loop
-
- ldmfd sp!, {r4-r11}
- mov pc, lr
-
- ENDP
-
- END
-
diff --git a/vp8/encoder/arm/armv6/vp8_variance16x16_armv6.asm b/vp8/encoder/arm/armv6/vp8_variance16x16_armv6.asm
deleted file mode 100644
index 9883763..0000000
--- a/vp8/encoder/arm/armv6/vp8_variance16x16_armv6.asm
+++ /dev/null
@@ -1,148 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_variance16x16_armv6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-|vp8_variance16x16_armv6| PROC
-
- stmfd sp!, {r4-r12, lr}
- mov r8, #0 ; initialize sum = 0
- mov r11, #0 ; initialize sse = 0
- mov r12, #16 ; set loop counter to 16 (=block height)
-
-loop
- ; 1st 4 pixels
- ldr r4, [r0, #0] ; load 4 src pixels
- ldr r5, [r2, #0] ; load 4 ref pixels
-
- mov lr, #0 ; constant zero
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r9, r5, r4 ; calculate difference with reversed operands
- sel r6, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
- ; calculate total sum
- adds r8, r8, r4 ; add positive differences to sum
- subs r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 2nd 4 pixels
- ldr r4, [r0, #4] ; load 4 src pixels
- ldr r5, [r2, #4] ; load 4 ref pixels
- smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r9, r5, r4 ; calculate difference with reversed operands
- sel r6, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 3rd 4 pixels
- ldr r4, [r0, #8] ; load 4 src pixels
- ldr r5, [r2, #8] ; load 4 ref pixels
- smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r9, r5, r4 ; calculate difference with reversed operands
- sel r6, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 4th 4 pixels
- ldr r4, [r0, #12] ; load 4 src pixels
- ldr r5, [r2, #12] ; load 4 ref pixels
- smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- add r0, r0, r1 ; set src_ptr to next row
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r9, r5, r4 ; calculate difference with reversed operands
- add r2, r2, r3 ; set dst_ptr to next row
- sel r6, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
- smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
-
-
- subs r12, r12, #1
-
- bne loop
-
- ; return stuff
- ldr r6, [sp, #40] ; get address of sse
- mul r0, r8, r8 ; sum * sum
- str r11, [r6] ; store sse
- sub r0, r11, r0, asr #8 ; return (sse - ((sum * sum) >> 8))
-
- ldmfd sp!, {r4-r12, pc}
-
- ENDP
-
- END
-
diff --git a/vp8/encoder/arm/armv6/vp8_variance8x8_armv6.asm b/vp8/encoder/arm/armv6/vp8_variance8x8_armv6.asm
deleted file mode 100644
index 7daecb9..0000000
--- a/vp8/encoder/arm/armv6/vp8_variance8x8_armv6.asm
+++ /dev/null
@@ -1,95 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_variance8x8_armv6|
-
- ARM
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-|vp8_variance8x8_armv6| PROC
-
- push {r4-r10, lr}
- mov r12, #8 ; set loop counter to 8 (=block height)
- mov r4, #0 ; initialize sum = 0
- mov r5, #0 ; initialize sse = 0
-
-loop
- ; 1st 4 pixels
- ldr r6, [r0, #0x0] ; load 4 src pixels
- ldr r7, [r2, #0x0] ; load 4 ref pixels
-
- mov lr, #0 ; constant zero
-
- usub8 r8, r6, r7 ; calculate difference
- sel r10, r8, lr ; select bytes with positive difference
- usub8 r9, r7, r6 ; calculate difference with reversed operands
- sel r8, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r6, r10, lr ; calculate sum of positive differences
- usad8 r7, r8, lr ; calculate sum of negative differences
- orr r8, r8, r10 ; differences of all 4 pixels
- ; calculate total sum
- add r4, r4, r6 ; add positive differences to sum
- sub r4, r4, r7 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r7, r8 ; byte (two pixels) to halfwords
- uxtb16 r10, r8, ror #8 ; another two pixels to halfwords
- smlad r5, r7, r7, r5 ; dual signed multiply, add and accumulate (1)
-
- ; 2nd 4 pixels
- ldr r6, [r0, #0x4] ; load 4 src pixels
- ldr r7, [r2, #0x4] ; load 4 ref pixels
- smlad r5, r10, r10, r5 ; dual signed multiply, add and accumulate (2)
-
- usub8 r8, r6, r7 ; calculate difference
- add r0, r0, r1 ; set src_ptr to next row
- sel r10, r8, lr ; select bytes with positive difference
- usub8 r9, r7, r6 ; calculate difference with reversed operands
- add r2, r2, r3 ; set dst_ptr to next row
- sel r8, r9, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r6, r10, lr ; calculate sum of positive differences
- usad8 r7, r8, lr ; calculate sum of negative differences
- orr r8, r8, r10 ; differences of all 4 pixels
-
- ; calculate total sum
- add r4, r4, r6 ; add positive differences to sum
- sub r4, r4, r7 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r7, r8 ; byte (two pixels) to halfwords
- uxtb16 r10, r8, ror #8 ; another two pixels to halfwords
- smlad r5, r7, r7, r5 ; dual signed multiply, add and accumulate (1)
- subs r12, r12, #1 ; next row
- smlad r5, r10, r10, r5 ; dual signed multiply, add and accumulate (2)
-
- bne loop
-
- ; return stuff
- ldr r8, [sp, #32] ; get address of sse
- mul r1, r4, r4 ; sum * sum
- str r5, [r8] ; store sse
- sub r0, r5, r1, ASR #6 ; return (sse - ((sum * sum) >> 6))
-
- pop {r4-r10, pc}
-
- ENDP
-
- END
diff --git a/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6.asm b/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6.asm
deleted file mode 100644
index 2350f3e..0000000
--- a/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6.asm
+++ /dev/null
@@ -1,176 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_variance_halfpixvar16x16_h_armv6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-|vp8_variance_halfpixvar16x16_h_armv6| PROC
-
- stmfd sp!, {r4-r12, lr}
- mov r8, #0 ; initialize sum = 0
- ldr r10, c80808080
- mov r11, #0 ; initialize sse = 0
- mov r12, #16 ; set loop counter to 16 (=block height)
- mov lr, #0 ; constant zero
-loop
- ; 1st 4 pixels
- ldr r4, [r0, #0] ; load 4 src pixels
- ldr r6, [r0, #1] ; load 4 src pixels with 1 byte offset
- ldr r5, [r2, #0] ; load 4 ref pixels
-
- ; bilinear interpolation
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
- ; calculate total sum
- adds r8, r8, r4 ; add positive differences to sum
- subs r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 2nd 4 pixels
- ldr r4, [r0, #4] ; load 4 src pixels
- ldr r6, [r0, #5] ; load 4 src pixels with 1 byte offset
- ldr r5, [r2, #4] ; load 4 ref pixels
-
- ; bilinear interpolation
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
-
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 3rd 4 pixels
- ldr r4, [r0, #8] ; load 4 src pixels
- ldr r6, [r0, #9] ; load 4 src pixels with 1 byte offset
- ldr r5, [r2, #8] ; load 4 ref pixels
-
- ; bilinear interpolation
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
-
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 4th 4 pixels
- ldr r4, [r0, #12] ; load 4 src pixels
- ldr r6, [r0, #13] ; load 4 src pixels with 1 byte offset
- ldr r5, [r2, #12] ; load 4 ref pixels
-
- ; bilinear interpolation
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
-
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- add r0, r0, r1 ; set src_ptr to next row
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- add r2, r2, r3 ; set dst_ptr to next row
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- subs r12, r12, #1
-
- bne loop
-
- ; return stuff
- ldr r6, [sp, #40] ; get address of sse
- mul r0, r8, r8 ; sum * sum
- str r11, [r6] ; store sse
- sub r0, r11, r0, asr #8 ; return (sse - ((sum * sum) >> 8))
-
- ldmfd sp!, {r4-r12, pc}
-
- ENDP
-
-c80808080
- DCD 0x80808080
-
- END
-
diff --git a/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6.asm b/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6.asm
deleted file mode 100644
index f9ae3b7..0000000
--- a/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6.asm
+++ /dev/null
@@ -1,216 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_variance_halfpixvar16x16_hv_armv6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-|vp8_variance_halfpixvar16x16_hv_armv6| PROC
-
- stmfd sp!, {r4-r12, lr}
- mov r8, #0 ; initialize sum = 0
- ldr r10, c80808080
- mov r11, #0 ; initialize sse = 0
- mov r12, #16 ; set loop counter to 16 (=block height)
- mov lr, #0 ; constant zero
-loop
- add r9, r0, r1 ; pointer to pixels on the next row
- ; 1st 4 pixels
- ldr r4, [r0, #0] ; load source pixels a, row N
- ldr r6, [r0, #1] ; load source pixels b, row N
- ldr r5, [r9, #0] ; load source pixels c, row N+1
- ldr r7, [r9, #1] ; load source pixels d, row N+1
-
- ; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
- ; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
- mvn r7, r7
- uhsub8 r5, r5, r7
- eor r5, r5, r10
- ; z = (x + y + 1) >> 1, interpolate half pixel values vertically
- mvn r5, r5
- uhsub8 r4, r4, r5
- ldr r5, [r2, #0] ; load 4 ref pixels
- eor r4, r4, r10
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
- ; calculate total sum
- adds r8, r8, r4 ; add positive differences to sum
- subs r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 2nd 4 pixels
- ldr r4, [r0, #4] ; load source pixels a, row N
- ldr r6, [r0, #5] ; load source pixels b, row N
- ldr r5, [r9, #4] ; load source pixels c, row N+1
-
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- ldr r7, [r9, #5] ; load source pixels d, row N+1
-
- ; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
- ; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
- mvn r7, r7
- uhsub8 r5, r5, r7
- eor r5, r5, r10
- ; z = (x + y + 1) >> 1, interpolate half pixel values vertically
- mvn r5, r5
- uhsub8 r4, r4, r5
- ldr r5, [r2, #4] ; load 4 ref pixels
- eor r4, r4, r10
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 3rd 4 pixels
- ldr r4, [r0, #8] ; load source pixels a, row N
- ldr r6, [r0, #9] ; load source pixels b, row N
- ldr r5, [r9, #8] ; load source pixels c, row N+1
-
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- ldr r7, [r9, #9] ; load source pixels d, row N+1
-
- ; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
- ; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
- mvn r7, r7
- uhsub8 r5, r5, r7
- eor r5, r5, r10
- ; z = (x + y + 1) >> 1, interpolate half pixel values vertically
- mvn r5, r5
- uhsub8 r4, r4, r5
- ldr r5, [r2, #8] ; load 4 ref pixels
- eor r4, r4, r10
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 4th 4 pixels
- ldr r4, [r0, #12] ; load source pixels a, row N
- ldr r6, [r0, #13] ; load source pixels b, row N
- ldr r5, [r9, #12] ; load source pixels c, row N+1
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
- ldr r7, [r9, #13] ; load source pixels d, row N+1
-
- ; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
- ; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
- mvn r7, r7
- uhsub8 r5, r5, r7
- eor r5, r5, r10
- ; z = (x + y + 1) >> 1, interpolate half pixel values vertically
- mvn r5, r5
- uhsub8 r4, r4, r5
- ldr r5, [r2, #12] ; load 4 ref pixels
- eor r4, r4, r10
-
- usub8 r6, r4, r5 ; calculate difference
- add r0, r0, r1 ; set src_ptr to next row
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- add r2, r2, r3 ; set dst_ptr to next row
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
- subs r12, r12, #1
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- bne loop
-
- ; return stuff
- ldr r6, [sp, #40] ; get address of sse
- mul r0, r8, r8 ; sum * sum
- str r11, [r6] ; store sse
- sub r0, r11, r0, asr #8 ; return (sse - ((sum * sum) >> 8))
-
- ldmfd sp!, {r4-r12, pc}
-
- ENDP
-
-c80808080
- DCD 0x80808080
-
- END
diff --git a/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6.asm b/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6.asm
deleted file mode 100644
index 9e0a035..0000000
--- a/vp8/encoder/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6.asm
+++ /dev/null
@@ -1,178 +0,0 @@
-;
-; Copyright (c) 2011 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_variance_halfpixvar16x16_v_armv6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-|vp8_variance_halfpixvar16x16_v_armv6| PROC
-
- stmfd sp!, {r4-r12, lr}
- mov r8, #0 ; initialize sum = 0
- ldr r10, c80808080
- mov r11, #0 ; initialize sse = 0
- mov r12, #16 ; set loop counter to 16 (=block height)
- mov lr, #0 ; constant zero
-loop
- add r9, r0, r1 ; set src pointer to next row
- ; 1st 4 pixels
- ldr r4, [r0, #0] ; load 4 src pixels
- ldr r6, [r9, #0] ; load 4 src pixels from next row
- ldr r5, [r2, #0] ; load 4 ref pixels
-
- ; bilinear interpolation
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
- ; calculate total sum
- adds r8, r8, r4 ; add positive differences to sum
- subs r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 2nd 4 pixels
- ldr r4, [r0, #4] ; load 4 src pixels
- ldr r6, [r9, #4] ; load 4 src pixels from next row
- ldr r5, [r2, #4] ; load 4 ref pixels
-
- ; bilinear interpolation
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
-
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 3rd 4 pixels
- ldr r4, [r0, #8] ; load 4 src pixels
- ldr r6, [r9, #8] ; load 4 src pixels from next row
- ldr r5, [r2, #8] ; load 4 ref pixels
-
- ; bilinear interpolation
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
-
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
-
- ; 4th 4 pixels
- ldr r4, [r0, #12] ; load 4 src pixels
- ldr r6, [r9, #12] ; load 4 src pixels from next row
- ldr r5, [r2, #12] ; load 4 ref pixels
-
- ; bilinear interpolation
- mvn r6, r6
- uhsub8 r4, r4, r6
- eor r4, r4, r10
-
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
- usub8 r6, r4, r5 ; calculate difference
- add r0, r0, r1 ; set src_ptr to next row
- sel r7, r6, lr ; select bytes with positive difference
- usub8 r6, r5, r4 ; calculate difference with reversed operands
- add r2, r2, r3 ; set dst_ptr to next row
- sel r6, r6, lr ; select bytes with negative difference
-
- ; calculate partial sums
- usad8 r4, r7, lr ; calculate sum of positive differences
- usad8 r5, r6, lr ; calculate sum of negative differences
- orr r6, r6, r7 ; differences of all 4 pixels
-
- ; calculate total sum
- add r8, r8, r4 ; add positive differences to sum
- sub r8, r8, r5 ; substract negative differences from sum
-
- ; calculate sse
- uxtb16 r5, r6 ; byte (two pixels) to halfwords
- uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
- smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
- smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
-
-
- subs r12, r12, #1
-
- bne loop
-
- ; return stuff
- ldr r6, [sp, #40] ; get address of sse
- mul r0, r8, r8 ; sum * sum
- str r11, [r6] ; store sse
- sub r0, r11, r0, asr #8 ; return (sse - ((sum * sum) >> 8))
-
- ldmfd sp!, {r4-r12, pc}
-
- ENDP
-
-c80808080
- DCD 0x80808080
-
- END
-
diff --git a/vp8/encoder/arm/armv6/walsh_v6.asm b/vp8/encoder/arm/armv6/walsh_v6.asm
deleted file mode 100644
index 61ffdb3..0000000
--- a/vp8/encoder/arm/armv6/walsh_v6.asm
+++ /dev/null
@@ -1,145 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
- EXPORT |vp8_short_walsh4x4_armv6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
-;short vp8_short_walsh4x4_armv6(short *input, short *output, int pitch)
-|vp8_short_walsh4x4_armv6| PROC
-
- stmdb sp!, {r4 - r11, lr}
-
- mov r12, r2 ; ugh. not clean
- ldr r2, [r0] ; [1 | 0]
- ldr r3, [r0, #4] ; [3 | 2]
- ldr r4, [r0, r12]! ; [5 | 4]
- ldr r5, [r0, #4] ; [7 | 6]
- ldr r6, [r0, r12]! ; [9 | 8]
- ldr r7, [r0, #4] ; [11 | 10]
- ldr r8, [r0, r12]! ; [13 | 12]
- ldr r9, [r0, #4] ; [15 | 14]
-
- qsubaddx r10, r2, r3 ; [c1|a1] [1-2 | 0+3]
- qaddsubx r11, r2, r3 ; [b1|d1] [1+2 | 0-3]
- qsubaddx r12, r4, r5 ; [c1|a1] [5-6 | 4+7]
- qaddsubx lr, r4, r5 ; [b1|d1] [5+6 | 4-7]
-
- qaddsubx r2, r10, r11 ; [1 | 2] [c1+d1 | a1-b1]
- qaddsubx r3, r11, r10 ; [0 | 3] [b1+a1 | d1-c1]
- qaddsubx r4, r12, lr ; [5 | 6] [c1+d1 | a1-b1]
- qaddsubx r5, lr, r12 ; [4 | 7] [b1+a1 | d1-c1]
-
- qsubaddx r10, r6, r7 ; [c1|a1] [9-10 | 8+11]
- qaddsubx r11, r6, r7 ; [b1|d1] [9+10 | 8-11]
- qsubaddx r12, r8, r9 ; [c1|a1] [13-14 | 12+15]
- qaddsubx lr, r8, r9 ; [b1|d1] [13+14 | 12-15]
-
- qaddsubx r6, r10, r11 ; [9 |10] [c1+d1 | a1-b1]
- qaddsubx r7, r11, r10 ; [8 |11] [b1+a1 | d1-c1]
- qaddsubx r8, r12, lr ; [13|14] [c1+d1 | a1-b1]
- qaddsubx r9, lr, r12 ; [12|15] [b1+a1 | d1-c1]
-
- ; first transform complete
-
- qadd16 r10, r3, r9 ; a1 [0+12 | 3+15]
- qadd16 r11, r5, r7 ; b1 [4+8 | 7+11]
- qsub16 r12, r5, r7 ; c1 [4-8 | 7-11]
- qsub16 lr, r3, r9 ; d1 [0-12 | 3-15]
-
- qadd16 r3, r10, r11 ; a2 [a1+b1] [0 | 3]
- qadd16 r5, r12, lr ; b2 [c1+d1] [4 | 7]
- qsub16 r7, r10, r11 ; c2 [a1-b1] [8 |11]
- qsub16 r9, lr, r12 ; d2 [d1-c1] [12|15]
-
- qadd16 r10, r2, r8 ; a1 [1+13 | 2+14]
- qadd16 r11, r4, r6 ; b1 [5+9 | 6+10]
- qsub16 r12, r4, r6 ; c1 [5-9 | 6-10]
- qsub16 lr, r2, r8 ; d1 [1-13 | 2-14]
-
- qadd16 r2, r10, r11 ; a2 [a1+b1] [1 | 2]
- qadd16 r4, r12, lr ; b2 [c1+d1] [5 | 6]
- qsub16 r6, r10, r11 ; c2 [a1-b1] [9 |10]
- qsub16 r8, lr, r12 ; d2 [d1-c1] [13|14]
-
- ; [a-d]2 += ([a-d]2 > 0)
-
- asrs r10, r3, #16
- addpl r10, r10, #1 ; [~0]
- asrs r11, r2, #16
- addpl r11, r11, #1 ; [~1]
- lsl r11, r11, #15 ; [1 | x]
- pkhtb r10, r11, r10, asr #1; [1 | 0]
- str r10, [r1], #4
-
- lsls r11, r2, #16
- addpl r11, r11, #0x10000 ; [~2]
- lsls r12, r3, #16
- addpl r12, r12, #0x10000 ; [~3]
- asr r12, r12, #1 ; [3 | x]
- pkhtb r11, r12, r11, asr #17; [3 | 2]
- str r11, [r1], #4
-
- asrs r2, r5, #16
- addpl r2, r2, #1 ; [~4]
- asrs r3, r4, #16
- addpl r3, r3, #1 ; [~5]
- lsl r3, r3, #15 ; [5 | x]
- pkhtb r2, r3, r2, asr #1 ; [5 | 4]
- str r2, [r1], #4
-
- lsls r2, r4, #16
- addpl r2, r2, #0x10000 ; [~6]
- lsls r3, r5, #16
- addpl r3, r3, #0x10000 ; [~7]
- asr r3, r3, #1 ; [7 | x]
- pkhtb r2, r3, r2, asr #17 ; [7 | 6]
- str r2, [r1], #4
-
- asrs r2, r7, #16
- addpl r2, r2, #1 ; [~8]
- asrs r3, r6, #16
- addpl r3, r3, #1 ; [~9]
- lsl r3, r3, #15 ; [9 | x]
- pkhtb r2, r3, r2, asr #1 ; [9 | 8]
- str r2, [r1], #4
-
- lsls r2, r6, #16
- addpl r2, r2, #0x10000 ; [~10]
- lsls r3, r7, #16
- addpl r3, r3, #0x10000 ; [~11]
- asr r3, r3, #1 ; [11 | x]
- pkhtb r2, r3, r2, asr #17 ; [11 | 10]
- str r2, [r1], #4
-
- asrs r2, r9, #16
- addpl r2, r2, #1 ; [~12]
- asrs r3, r8, #16
- addpl r3, r3, #1 ; [~13]
- lsl r3, r3, #15 ; [13 | x]
- pkhtb r2, r3, r2, asr #1 ; [13 | 12]
- str r2, [r1], #4
-
- lsls r2, r8, #16
- addpl r2, r2, #0x10000 ; [~14]
- lsls r3, r9, #16
- addpl r3, r3, #0x10000 ; [~15]
- asr r3, r3, #1 ; [15 | x]
- pkhtb r2, r3, r2, asr #17 ; [15 | 14]
- str r2, [r1]
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ; |vp8_short_walsh4x4_armv6|
-
- END
diff --git a/vp8/encoder/arm/boolhuff_arm.c b/vp8/encoder/arm/boolhuff_arm.c
deleted file mode 100644
index 9089663..0000000
--- a/vp8/encoder/arm/boolhuff_arm.c
+++ /dev/null
@@ -1,34 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/encoder/boolhuff.h"
-#include "vp8/common/blockd.h"
-
-const unsigned int vp8_prob_cost[256] =
-{
- 2047, 2047, 1791, 1641, 1535, 1452, 1385, 1328, 1279, 1235, 1196, 1161, 1129, 1099, 1072, 1046,
- 1023, 1000, 979, 959, 940, 922, 905, 889, 873, 858, 843, 829, 816, 803, 790, 778,
- 767, 755, 744, 733, 723, 713, 703, 693, 684, 675, 666, 657, 649, 641, 633, 625,
- 617, 609, 602, 594, 587, 580, 573, 567, 560, 553, 547, 541, 534, 528, 522, 516,
- 511, 505, 499, 494, 488, 483, 477, 472, 467, 462, 457, 452, 447, 442, 437, 433,
- 428, 424, 419, 415, 410, 406, 401, 397, 393, 389, 385, 381, 377, 373, 369, 365,
- 361, 357, 353, 349, 346, 342, 338, 335, 331, 328, 324, 321, 317, 314, 311, 307,
- 304, 301, 297, 294, 291, 288, 285, 281, 278, 275, 272, 269, 266, 263, 260, 257,
- 255, 252, 249, 246, 243, 240, 238, 235, 232, 229, 227, 224, 221, 219, 216, 214,
- 211, 208, 206, 203, 201, 198, 196, 194, 191, 189, 186, 184, 181, 179, 177, 174,
- 172, 170, 168, 165, 163, 161, 159, 156, 154, 152, 150, 148, 145, 143, 141, 139,
- 137, 135, 133, 131, 129, 127, 125, 123, 121, 119, 117, 115, 113, 111, 109, 107,
- 105, 103, 101, 99, 97, 95, 93, 92, 90, 88, 86, 84, 82, 81, 79, 77,
- 75, 73, 72, 70, 68, 66, 65, 63, 61, 60, 58, 56, 55, 53, 51, 50,
- 48, 46, 45, 43, 41, 40, 38, 37, 35, 33, 32, 30, 29, 27, 25, 24,
- 22, 21, 19, 18, 16, 15, 13, 12, 10, 9, 7, 6, 4, 3, 1, 1
-};
-
diff --git a/vp8/encoder/arm/dct_arm.c b/vp8/encoder/arm/dct_arm.c
deleted file mode 100644
index 60d649d..0000000
--- a/vp8/encoder/arm/dct_arm.c
+++ /dev/null
@@ -1,24 +0,0 @@
-/*
- * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_config.h"
-#include "vp8/encoder/dct.h"
-
-#if HAVE_ARMV6
-
-void vp8_fast_fdct8x4_armv6(short *input, short *output, int pitch)
-{
- vp8_fast_fdct4x4_armv6(input, output, pitch);
- vp8_fast_fdct4x4_armv6(input + 4, output + 16, pitch);
-}
-
-#endif /* HAVE_ARMV6 */
-
-
diff --git a/vp8/encoder/arm/dct_arm.h b/vp8/encoder/arm/dct_arm.h
deleted file mode 100644
index 769d5f4..0000000
--- a/vp8/encoder/arm/dct_arm.h
+++ /dev/null
@@ -1,59 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef DCT_ARM_H
-#define DCT_ARM_H
-
-#if HAVE_ARMV6
-extern prototype_fdct(vp8_short_walsh4x4_armv6);
-extern prototype_fdct(vp8_fast_fdct4x4_armv6);
-extern prototype_fdct(vp8_fast_fdct8x4_armv6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_fdct_walsh_short4x4
-#define vp8_fdct_walsh_short4x4 vp8_short_walsh4x4_armv6
-
-#undef vp8_fdct_fast4x4
-#define vp8_fdct_fast4x4 vp8_fast_fdct4x4_armv6
-
-#undef vp8_fdct_fast8x4
-#define vp8_fdct_fast8x4 vp8_fast_fdct8x4_armv6
-#endif
-
-#endif /* HAVE_ARMV6 */
-
-#if HAVE_ARMV7
-extern prototype_fdct(vp8_short_fdct4x4_neon);
-extern prototype_fdct(vp8_short_fdct8x4_neon);
-extern prototype_fdct(vp8_fast_fdct4x4_neon);
-extern prototype_fdct(vp8_fast_fdct8x4_neon);
-extern prototype_fdct(vp8_short_walsh4x4_neon);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_fdct_short4x4
-#define vp8_fdct_short4x4 vp8_short_fdct4x4_neon
-
-#undef vp8_fdct_short8x4
-#define vp8_fdct_short8x4 vp8_short_fdct8x4_neon
-
-#undef vp8_fdct_fast4x4
-#define vp8_fdct_fast4x4 vp8_fast_fdct4x4_neon
-
-#undef vp8_fdct_fast8x4
-#define vp8_fdct_fast8x4 vp8_fast_fdct8x4_neon
-
-#undef vp8_fdct_walsh_short4x4
-#define vp8_fdct_walsh_short4x4 vp8_short_walsh4x4_neon
-#endif
-
-#endif
-
-#endif
diff --git a/vp8/encoder/arm/encodemb_arm.h b/vp8/encoder/arm/encodemb_arm.h
deleted file mode 100644
index bf417fe..0000000
--- a/vp8/encoder/arm/encodemb_arm.h
+++ /dev/null
@@ -1,64 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef ENCODEMB_ARM_H
-#define ENCODEMB_ARM_H
-
-#if HAVE_ARMV6
-extern prototype_subb(vp8_subtract_b_armv6);
-extern prototype_submby(vp8_subtract_mby_armv6);
-extern prototype_submbuv(vp8_subtract_mbuv_armv6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_encodemb_subb
-#define vp8_encodemb_subb vp8_subtract_b_armv6
-
-#undef vp8_encodemb_submby
-#define vp8_encodemb_submby vp8_subtract_mby_armv6
-
-#undef vp8_encodemb_submbuv
-#define vp8_encodemb_submbuv vp8_subtract_mbuv_armv6
-#endif
-
-#endif /* HAVE_ARMV6 */
-
-#if HAVE_ARMV7
-//extern prototype_berr(vp8_block_error_c);
-//extern prototype_mberr(vp8_mbblock_error_c);
-//extern prototype_mbuverr(vp8_mbuverror_c);
-
-extern prototype_subb(vp8_subtract_b_neon);
-extern prototype_submby(vp8_subtract_mby_neon);
-extern prototype_submbuv(vp8_subtract_mbuv_neon);
-
-//#undef vp8_encodemb_berr
-//#define vp8_encodemb_berr vp8_block_error_c
-
-//#undef vp8_encodemb_mberr
-//#define vp8_encodemb_mberr vp8_mbblock_error_c
-
-//#undef vp8_encodemb_mbuverr
-//#define vp8_encodemb_mbuverr vp8_mbuverror_c
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_encodemb_subb
-#define vp8_encodemb_subb vp8_subtract_b_neon
-
-#undef vp8_encodemb_submby
-#define vp8_encodemb_submby vp8_subtract_mby_neon
-
-#undef vp8_encodemb_submbuv
-#define vp8_encodemb_submbuv vp8_subtract_mbuv_neon
-#endif
-
-#endif
-
-#endif
diff --git a/vp8/encoder/arm/neon/fastfdct4x4_neon.asm b/vp8/encoder/arm/neon/fastfdct4x4_neon.asm
deleted file mode 100644
index 757425d..0000000
--- a/vp8/encoder/arm/neon/fastfdct4x4_neon.asm
+++ /dev/null
@@ -1,122 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_fast_fdct4x4_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void vp8_fast_fdct4x4_c(short *input, short *output, int pitch);
-;NOTE:
-;The input *src_diff. src_diff is calculated as:
-;diff_ptr[c] = src_ptr[c] - pred_ptr[c]; (in Subtract* function)
-;In which *src_ptr and *pred_ptr both are unsigned char.
-;Therefore, *src_diff should be in the range of [-255, 255].
-;CAUTION:
-;The input values of 25th block are set in vp8_build_dcblock function, which are out of [-255, 255].
-;But, VP8 encoder only uses vp8_short_fdct4x4_c for 25th block, not vp8_fast_fdct4x4_c. That makes
-;it ok for assuming *input in [-255, 255] in vp8_fast_fdct4x4_c, but not ok in vp8_short_fdct4x4_c.
-
-|vp8_fast_fdct4x4_neon| PROC
- vld1.16 {d2}, [r0], r2 ;load input
- adr r12, ffdct_coeff
- vld1.16 {d3}, [r0], r2
- vld1.16 {d4}, [r0], r2
- vld1.16 {d0}, [r12]
- vld1.16 {d5}, [r0], r2
-
- ;First for-loop
- ;transpose d2, d3, d4, d5. Then, d2=ip[0], d3=ip[1], d4=ip[2], d5=ip[3]
- vtrn.32 d2, d4
- vtrn.32 d3, d5
- vtrn.16 d2, d3
- vtrn.16 d4, d5
-
- vadd.s16 d6, d2, d5 ;ip[0]+ip[3]
- vadd.s16 d7, d3, d4 ;ip[1]+ip[2]
- vsub.s16 d8, d3, d4 ;ip[1]-ip[2]
- vsub.s16 d9, d2, d5 ;ip[0]-ip[3]
- vshl.i16 q3, q3, #1 ; a1, b1
- vshl.i16 q4, q4, #1 ; c1, d1
-
- vadd.s16 d10, d6, d7 ;temp1 = a1 + b1
- vsub.s16 d11, d6, d7 ;temp2 = a1 - b1
-
- vqdmulh.s16 q6, q5, d0[1]
- vqdmulh.s16 q8, q4, d0[0]
- vqdmulh.s16 q7, q4, d0[2]
-
- vshr.s16 q6, q6, #1
- vshr.s16 q8, q8, #1
- vshr.s16 q7, q7, #1 ;d14:temp1 = ( c1 * x_c3)>>16; d15:temp1 = (d1 * x_c3)>>16
- vadd.s16 q8, q4, q8 ;d16:temp2 = ((c1 * x_c1)>>16) + c1; d17:temp2 = ((d1 * x_c1)>>16) + d1
-
- vadd.s16 d2, d10, d12 ;op[0] = ((temp1 * x_c2 )>>16) + temp1
- vadd.s16 d4, d11, d13 ;op[2] = ((temp2 * x_c2 )>>16) + temp2
- vadd.s16 d3, d14, d17 ;op[1] = temp1 + temp2 -- q is not necessary, just for protection
- vsub.s16 d5, d15, d16 ;op[3] = temp1 - temp2
-
- ;Second for-loop
- ;transpose d2, d3, d4, d5. Then, d2=ip[0], d3=ip[4], d4=ip[8], d5=ip[12]
- vtrn.32 d2, d4
- vtrn.32 d3, d5
- vtrn.16 d2, d3
- vtrn.16 d4, d5
-
- vadd.s16 d6, d2, d5 ;a1 = ip[0]+ip[12]
- vadd.s16 d7, d3, d4 ;b1 = ip[4]+ip[8]
- vsub.s16 d8, d3, d4 ;c1 = ip[4]-ip[8]
- vsub.s16 d9, d2, d5 ;d1 = ip[0]-ip[12]
-
- vadd.s16 d10, d6, d7 ;temp1 = a1 + b1
- vsub.s16 d11, d6, d7 ;temp2 = a1 - b1
-
-
- vqdmulh.s16 q6, q5, d0[1]
- vqdmulh.s16 q8, q4, d0[0]
- vqdmulh.s16 q7, q4, d0[2]
-
- vshr.s16 q6, q6, #1
- vshr.s16 q8, q8, #1
- vshr.s16 q7, q7, #1 ;d14:temp1 = ( c1 * x_c3)>>16; d15:temp1 = (d1 * x_c3)>>16
- vadd.s16 q8, q4, q8 ;d16:temp2 = ((c1 * x_c1)>>16) + c1; d17:temp2 = ((d1 * x_c1)>>16) + d1
-
- vadd.s16 d2, d10, d12 ;a2 = ((temp1 * x_c2 )>>16) + temp1
- vadd.s16 d4, d11, d13 ;c2 = ((temp2 * x_c2 )>>16) + temp2
- vadd.s16 d3, d14, d17 ;b2 = temp1 + temp2 -- q is not necessary, just for protection
- vsub.s16 d5, d15, d16 ;d2 = temp1 - temp2
-
- vclt.s16 q3, q1, #0
- vclt.s16 q4, q2, #0
-
- vsub.s16 q1, q1, q3
- vsub.s16 q2, q2, q4
-
- vshr.s16 q1, q1, #1
- vshr.s16 q2, q2, #1
-
- vst1.16 {q1, q2}, [r1]
-
- bx lr
-
- ENDP
-
-;-----------------
-
-ffdct_coeff
-; 60547 = 0xEC83
-; 46341 = 0xB505
-; 25080 = 0x61F8
- DCD 0xB505EC83, 0x000061F8
-
- END
diff --git a/vp8/encoder/arm/neon/fastfdct8x4_neon.asm b/vp8/encoder/arm/neon/fastfdct8x4_neon.asm
deleted file mode 100644
index 212c0cd..0000000
--- a/vp8/encoder/arm/neon/fastfdct8x4_neon.asm
+++ /dev/null
@@ -1,175 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_fast_fdct8x4_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void vp8_fast_fdct4x4_c(short *input, short *output, int pitch);
-;NOTE:
-;The input *src_diff. src_diff is calculated as:
-;diff_ptr[c] = src_ptr[c] - pred_ptr[c]; (in Subtract* function)
-;In which *src_ptr and *pred_ptr both are unsigned char.
-;Therefore, *src_diff should be in the range of [-255, 255].
-;CAUTION:
-;The input values of 25th block are set in vp8_build_dcblock function, which are out of [-255, 255].
-;But, VP8 encoder only uses vp8_short_fdct4x4_c for 25th block, not vp8_fast_fdct4x4_c. That makes
-;it ok for assuming *input in [-255, 255] in vp8_fast_fdct4x4_c, but not ok in vp8_short_fdct4x4_c.
-
-|vp8_fast_fdct8x4_neon| PROC
- vld1.16 {q1}, [r0], r2 ;load input
- adr r12, ffdct8_coeff
- vld1.16 {q2}, [r0], r2
- vld1.16 {q3}, [r0], r2
- vld1.16 {d0}, [r12]
- vld1.16 {q4}, [r0], r2
-
- ;First for-loop
- ;transpose d2, d4, d6, d8. Then, d2=ip[0], d4=ip[1], d6=ip[2], d8=ip[3]
- ;transpose d3, d5, d7, d9. Then, d3=ip[0], d5=ip[1], d7=ip[2], d9=ip[3]
- vtrn.32 d2, d6
- vtrn.32 d3, d7
- vtrn.32 d4, d8
- vtrn.32 d5, d9
- vtrn.16 d2, d4
- vtrn.16 d3, d5
- vtrn.16 d6, d8
- vtrn.16 d7, d9
-
- vadd.s16 d10, d2, d8 ;ip[0]+ip[3]
- vadd.s16 d11, d4, d6 ;ip[1]+ip[2]
- vsub.s16 d12, d4, d6 ;ip[1]-ip[2]
- vsub.s16 d13, d2, d8 ;ip[0]-ip[3]
- vadd.s16 d22, d3, d9
- vadd.s16 d23, d5, d7
- vsub.s16 d24, d5, d7
- vsub.s16 d25, d3, d9
-
- vshl.i16 q5, q5, #1 ; a1, b1
- vshl.i16 q6, q6, #1 ; c1, d1
- vshl.i16 q1, q11, #1
- vshl.i16 q2, q12, #1
-
- vadd.s16 d14, d10, d11 ;temp1 = a1 + b1
- vsub.s16 d15, d10, d11 ;temp2 = a1 - b1
- vadd.s16 d24, d2, d3
- vsub.s16 d25, d2, d3
-
- vqdmulh.s16 q8, q7, d0[1]
- vqdmulh.s16 q13, q12, d0[1]
- vqdmulh.s16 q10, q6, d0[0]
- vqdmulh.s16 q15, q2, d0[0]
- vqdmulh.s16 q9, q6, d0[2]
- vqdmulh.s16 q14, q2, d0[2]
-
- vshr.s16 q8, q8, #1
- vshr.s16 q13, q13, #1
- vshr.s16 q10, q10, #1
- vshr.s16 q15, q15, #1
- vshr.s16 q9, q9, #1 ;d18:temp1 = ( c1 * x_c3)>>16; d19:temp1 = (d1 * x_c3)>>16
- vshr.s16 q14, q14, #1 ;d28:temp1 = ( c1 * x_c3)>>16; d29:temp1 = (d1 * x_c3)>>16
- vadd.s16 q10, q6, q10 ;d20:temp2 = ((c1 * x_c1)>>16) + c1; d21:temp2 = ((d1 * x_c1)>>16) + d1
- vadd.s16 q15, q2, q15 ;d30:temp2 = ((c1 * x_c1)>>16) + c1; d31:temp2 = ((d1 * x_c1)>>16) + d1
-
- vadd.s16 d2, d14, d16 ;op[0] = ((temp1 * x_c2 )>>16) + temp1
- vadd.s16 d3, d24, d26 ;op[0] = ((temp1 * x_c2 )>>16) + temp1
- vadd.s16 d6, d15, d17 ;op[2] = ((temp2 * x_c2 )>>16) + temp2
- vadd.s16 d7, d25, d27 ;op[2] = ((temp2 * x_c2 )>>16) + temp2
- vadd.s16 d4, d18, d21 ;op[1] = temp1 + temp2 -- q is not necessary, just for protection
- vadd.s16 d5, d28, d31 ;op[1] = temp1 + temp2 -- q is not necessary, just for protection
- vsub.s16 d8, d19, d20 ;op[3] = temp1 - temp2
- vsub.s16 d9, d29, d30 ;op[3] = temp1 - temp2
-
- ;Second for-loop
- ;transpose d2, d4, d6, d8. Then, d2=ip[0], d4=ip[4], d6=ip[8], d8=ip[12]
- ;transpose d3, d5, d7, d9. Then, d3=ip[0], d5=ip[4], d7=ip[8], d9=ip[12]
- vtrn.32 d2, d6
- vtrn.32 d3, d7
- vtrn.32 d4, d8
- vtrn.32 d5, d9
- vtrn.16 d2, d4
- vtrn.16 d3, d5
- vtrn.16 d6, d8
- vtrn.16 d7, d9
-
- vadd.s16 d10, d2, d8 ;a1 = ip[0]+ip[12]
- vadd.s16 d11, d4, d6 ;b1 = ip[4]+ip[8]
- vsub.s16 d12, d4, d6 ;c1 = ip[4]-ip[8]
- vsub.s16 d13, d2, d8 ;d1 = ip[0]-ip[12]
- vadd.s16 d2, d3, d9
- vadd.s16 d4, d5, d7
- vsub.s16 d24, d5, d7
- vsub.s16 d25, d3, d9
-
- vadd.s16 d14, d10, d11 ;temp1 = a1 + b1
- vsub.s16 d15, d10, d11 ;temp2 = a1 - b1
- vadd.s16 d22, d2, d4
- vsub.s16 d23, d2, d4
-
- vqdmulh.s16 q8, q7, d0[1]
- vqdmulh.s16 q13, q11, d0[1]
- vqdmulh.s16 q10, q6, d0[0]
- vqdmulh.s16 q15, q12, d0[0]
- vqdmulh.s16 q9, q6, d0[2]
- vqdmulh.s16 q14, q12, d0[2]
-
- vshr.s16 q8, q8, #1
- vshr.s16 q13, q13, #1
- vshr.s16 q10, q10, #1
- vshr.s16 q15, q15, #1
- vshr.s16 q9, q9, #1 ;d18:temp1 = ( c1 * x_c3)>>16; d19:temp1 = (d1 * x_c3)>>16
- vshr.s16 q14, q14, #1 ;d28:temp1 = ( c1 * x_c3)>>16; d29:temp1 = (d1 * x_c3)>>16
- vadd.s16 q10, q6, q10 ;d20:temp2 = ((c1 * x_c1)>>16) + c1; d21:temp2 = ((d1 * x_c1)>>16) + d1
- vadd.s16 q15, q12, q15 ;d30:temp2 = ((c1 * x_c1)>>16) + c1; d31:temp2 = ((d1 * x_c1)>>16) + d1
-
- vadd.s16 d2, d14, d16 ;a2 = ((temp1 * x_c2 )>>16) + temp1
- vadd.s16 d6, d22, d26 ;a2 = ((temp1 * x_c2 )>>16) + temp1
- vadd.s16 d4, d15, d17 ;c2 = ((temp2 * x_c2 )>>16) + temp2
- vadd.s16 d8, d23, d27 ;c2 = ((temp2 * x_c2 )>>16) + temp2
- vadd.s16 d3, d18, d21 ;b2 = temp1 + temp2 -- q is not necessary, just for protection
- vadd.s16 d7, d28, d31 ;b2 = temp1 + temp2 -- q is not necessary, just for protection
- vsub.s16 d5, d19, d20 ;d2 = temp1 - temp2
- vsub.s16 d9, d29, d30 ;d2 = temp1 - temp2
-
- vclt.s16 q5, q1, #0
- vclt.s16 q6, q2, #0
- vclt.s16 q7, q3, #0
- vclt.s16 q8, q4, #0
-
- vsub.s16 q1, q1, q5
- vsub.s16 q2, q2, q6
- vsub.s16 q3, q3, q7
- vsub.s16 q4, q4, q8
-
- vshr.s16 q1, q1, #1
- vshr.s16 q2, q2, #1
- vshr.s16 q3, q3, #1
- vshr.s16 q4, q4, #1
-
- vst1.16 {q1, q2}, [r1]!
- vst1.16 {q3, q4}, [r1]
-
- bx lr
-
- ENDP
-
-;-----------------
-
-ffdct8_coeff
-; 60547 = 0xEC83
-; 46341 = 0xB505
-; 25080 = 0x61F8
- DCD 0xB505EC83, 0x000061F8
-
- END
diff --git a/vp8/encoder/arm/neon/fastquantizeb_neon.asm b/vp8/encoder/arm/neon/fastquantizeb_neon.asm
deleted file mode 100644
index ca1ea9c..0000000
--- a/vp8/encoder/arm/neon/fastquantizeb_neon.asm
+++ /dev/null
@@ -1,118 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_fast_quantize_b_neon_func|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 short *coeff_ptr
-; r1 short *zbin_ptr
-; r2 short *qcoeff_ptr
-; r3 short *dqcoeff_ptr
-; stack short *dequant_ptr
-; stack short *scan_mask
-; stack short *round_ptr
-; stack short *quant_ptr
-
-; return int * eob
-|vp8_fast_quantize_b_neon_func| PROC
- vld1.16 {q0, q1}, [r0] ;load z
- vld1.16 {q10, q11}, [r1] ;load zbin
-
- vabs.s16 q4, q0 ;calculate x = abs(z)
- vabs.s16 q5, q1
-
- vcge.s16 q10, q4, q10 ;x>=zbin
- vcge.s16 q11, q5, q11
-
- ;if x<zbin (q10 & q11 are all 0), go to zero_output
- vorr.s16 q6, q10, q11
- vorr.s16 d12, d12, d13
- vmov r0, r1, d12
- orr r0, r0, r1
- cmp r0, #0
- beq zero_output
-
- ldr r0, [sp, #8] ;load round_ptr
- ldr r12, [sp, #12] ;load quant_ptr
-
- ;right shift 15 to get sign, all 0 if it is positive, all 1 if it is negative
- vshr.s16 q2, q0, #15 ; sz
- vshr.s16 q3, q1, #15
-
- vld1.s16 {q6, q7}, [r0] ;load round_ptr [0-15]
- vld1.s16 {q8, q9}, [r12] ;load quant_ptr [0-15]
-
- vadd.s16 q4, q6 ;x + Round
- vadd.s16 q5, q7
-
- ldr r0, [sp, #4] ;load rvsplus1_scan_order ptr
-
- vqdmulh.s16 q4, q8 ;y = ((Round + abs(z)) * Quant) >> 16
- vqdmulh.s16 q5, q9
-
- vld1.16 {q0, q1}, [r0] ;load rvsplus1_scan_order
- vceq.s16 q8, q8 ;set q8 to all 1
-
- vshr.s16 q4, #1 ;right shift 1 after vqdmulh
- vshr.s16 q5, #1
-
- ;modify data to have its original sign
- veor.s16 q4, q2 ; y^sz
- veor.s16 q5, q3
-
- ldr r12, [sp] ;load dequant_ptr
-
- vsub.s16 q4, q2 ; x1 = (y^sz) - sz = (y^sz) - (-1) (two's complement)
- vsub.s16 q5, q3
-
- vand.s16 q4, q10 ;mask off x1 elements
- vand.s16 q5, q11
-
- vld1.s16 {q6, q7}, [r12] ;load dequant_ptr[i]
-
- vtst.16 q14, q4, q8 ;now find eob
- vtst.16 q15, q5, q8 ;non-zero element is set to all 1 in q4, q5
-
- vst1.s16 {q4, q5}, [r2] ;store: qcoeff = x1
-
- vand q0, q0, q14 ;get all valid number from rvsplus1_scan_order array
- vand q1, q1, q15
-
- vmax.u16 q0, q0, q1 ;find maximum value in q0, q1
- vmax.u16 d0, d0, d1
- vmovl.u16 q0, d0
-
- vmul.s16 q6, q4 ;x * Dequant
- vmul.s16 q7, q5
-
- vmax.u32 d0, d0, d1
- vpmax.u32 d0, d0, d0
-
- vst1.s16 {q6, q7}, [r3] ;store dqcoeff = x * Dequant
-
- vmov.32 r0, d0[0]
- bx lr
-
-zero_output
- vst1.s16 {q10, q11}, [r2] ; qcoeff = 0
- vst1.s16 {q10, q11}, [r3] ; dqcoeff = 0
- mov r0, #0
-
- bx lr
-
- ENDP
-
- END
diff --git a/vp8/encoder/arm/neon/shortfdct_neon.asm b/vp8/encoder/arm/neon/shortfdct_neon.asm
deleted file mode 100644
index 1b7f362..0000000
--- a/vp8/encoder/arm/neon/shortfdct_neon.asm
+++ /dev/null
@@ -1,144 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_short_fdct4x4_neon|
- EXPORT |vp8_short_fdct8x4_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 short *input
-; r1 short *output
-; r2 int pitch
-; Input has a pitch, output is contiguous
-|vp8_short_fdct4x4_neon| PROC
- ldr r12, _dct_matrix_
- vld1.16 d0, [r0], r2
- vld1.16 d1, [r0], r2
- vld1.16 d2, [r0], r2
- vld1.16 d3, [r0]
- vld1.16 {q2, q3}, [r12]
-
-;first stage
- vmull.s16 q11, d4, d0[0] ;i=0
- vmull.s16 q12, d4, d1[0] ;i=1
- vmull.s16 q13, d4, d2[0] ;i=2
- vmull.s16 q14, d4, d3[0] ;i=3
-
- vmlal.s16 q11, d5, d0[1]
- vmlal.s16 q12, d5, d1[1]
- vmlal.s16 q13, d5, d2[1]
- vmlal.s16 q14, d5, d3[1]
-
- vmlal.s16 q11, d6, d0[2]
- vmlal.s16 q12, d6, d1[2]
- vmlal.s16 q13, d6, d2[2]
- vmlal.s16 q14, d6, d3[2]
-
- vmlal.s16 q11, d7, d0[3] ;sumtemp for i=0
- vmlal.s16 q12, d7, d1[3] ;sumtemp for i=1
- vmlal.s16 q13, d7, d2[3] ;sumtemp for i=2
- vmlal.s16 q14, d7, d3[3] ;sumtemp for i=3
-
- ; rounding
- vrshrn.i32 d22, q11, #14
- vrshrn.i32 d24, q12, #14
- vrshrn.i32 d26, q13, #14
- vrshrn.i32 d28, q14, #14
-
-;second stage
- vmull.s16 q4, d22, d4[0] ;i=0
- vmull.s16 q5, d22, d4[1] ;i=1
- vmull.s16 q6, d22, d4[2] ;i=2
- vmull.s16 q7, d22, d4[3] ;i=3
-
- vmlal.s16 q4, d24, d5[0]
- vmlal.s16 q5, d24, d5[1]
- vmlal.s16 q6, d24, d5[2]
- vmlal.s16 q7, d24, d5[3]
-
- vmlal.s16 q4, d26, d6[0]
- vmlal.s16 q5, d26, d6[1]
- vmlal.s16 q6, d26, d6[2]
- vmlal.s16 q7, d26, d6[3]
-
- vmlal.s16 q4, d28, d7[0] ;sumtemp for i=0
- vmlal.s16 q5, d28, d7[1] ;sumtemp for i=1
- vmlal.s16 q6, d28, d7[2] ;sumtemp for i=2
- vmlal.s16 q7, d28, d7[3] ;sumtemp for i=3
-
- vrshr.s32 q0, q4, #16
- vrshr.s32 q1, q5, #16
- vrshr.s32 q2, q6, #16
- vrshr.s32 q3, q7, #16
-
- vmovn.i32 d0, q0
- vmovn.i32 d1, q1
- vmovn.i32 d2, q2
- vmovn.i32 d3, q3
-
- vst1.16 {q0, q1}, [r1]
-
- bx lr
-
- ENDP
-
-; r0 short *input
-; r1 short *output
-; r2 int pitch
-|vp8_short_fdct8x4_neon| PROC
- ; Store link register and input before calling
- ; first 4x4 fdct. Do not need to worry about
- ; output or pitch because those pointers are not
- ; touched in the 4x4 fdct function
- stmdb sp!, {r0, lr}
-
- bl vp8_short_fdct4x4_neon
-
- ldmia sp!, {r0, lr}
-
- ; Move to the next block of data.
- add r0, r0, #8
- add r1, r1, #32
-
- ; Second time through do not store off the
- ; link register, just return from the 4x4 fdtc
- b vp8_short_fdct4x4_neon
-
- ; Should never get to this.
- bx lr
-
- ENDP
-
-;-----------------
-
-_dct_matrix_
- DCD dct_matrix
-dct_matrix
-; DCW 23170, 30274, 23170, 12540
-; DCW 23170, 12540, -23170,-30274
-; DCW 23170, -12540, -23170, 30274
-; DCW 23170, -30274, 23170,-12540
-; 23170 = 0x5a82
-; -23170 = 0xa57e
-; 30274 = 0x7642
-; -30274 = 0x89be
-; 12540 = 0x30fc
-; -12540 = 0xcf04
- DCD 0x76425a82, 0x30fc5a82
- DCD 0x30fc5a82, 0x89bea57e
- DCD 0xcf045a82, 0x7642a57e
- DCD 0x89be5a82, 0xcf045a82
-
- END
diff --git a/vp8/encoder/arm/neon/subtract_neon.asm b/vp8/encoder/arm/neon/subtract_neon.asm
deleted file mode 100644
index 68c2950..0000000
--- a/vp8/encoder/arm/neon/subtract_neon.asm
+++ /dev/null
@@ -1,185 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
- EXPORT |vp8_subtract_b_neon|
- EXPORT |vp8_subtract_mby_neon|
- EXPORT |vp8_subtract_mbuv_neon|
-
- INCLUDE asm_enc_offsets.asm
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-;void vp8_subtract_b_neon(BLOCK *be, BLOCKD *bd, int pitch)
-|vp8_subtract_b_neon| PROC
-
- stmfd sp!, {r4-r7}
-
- ldr r3, [r0, #vp8_block_base_src]
- ldr r4, [r0, #vp8_block_src]
- ldr r5, [r0, #vp8_block_src_diff]
- ldr r3, [r3]
- ldr r6, [r0, #vp8_block_src_stride]
- add r3, r3, r4 ; src = *base_src + src
- ldr r7, [r1, #vp8_blockd_predictor]
-
- vld1.8 {d0}, [r3], r6 ;load src
- vld1.8 {d1}, [r7], r2 ;load pred
- vld1.8 {d2}, [r3], r6
- vld1.8 {d3}, [r7], r2
- vld1.8 {d4}, [r3], r6
- vld1.8 {d5}, [r7], r2
- vld1.8 {d6}, [r3], r6
- vld1.8 {d7}, [r7], r2
-
- vsubl.u8 q10, d0, d1
- vsubl.u8 q11, d2, d3
- vsubl.u8 q12, d4, d5
- vsubl.u8 q13, d6, d7
-
- mov r2, r2, lsl #1
-
- vst1.16 {d20}, [r5], r2 ;store diff
- vst1.16 {d22}, [r5], r2
- vst1.16 {d24}, [r5], r2
- vst1.16 {d26}, [r5], r2
-
- ldmfd sp!, {r4-r7}
- bx lr
-
- ENDP
-
-
-;==========================================
-;void vp8_subtract_mby_neon(short *diff, unsigned char *src, unsigned char *pred, int stride)
-|vp8_subtract_mby_neon| PROC
- mov r12, #4
-
-subtract_mby_loop
- vld1.8 {q0}, [r1], r3 ;load src
- vld1.8 {q1}, [r2]! ;load pred
- vld1.8 {q2}, [r1], r3
- vld1.8 {q3}, [r2]!
- vld1.8 {q4}, [r1], r3
- vld1.8 {q5}, [r2]!
- vld1.8 {q6}, [r1], r3
- vld1.8 {q7}, [r2]!
-
- vsubl.u8 q8, d0, d2
- vsubl.u8 q9, d1, d3
- vsubl.u8 q10, d4, d6
- vsubl.u8 q11, d5, d7
- vsubl.u8 q12, d8, d10
- vsubl.u8 q13, d9, d11
- vsubl.u8 q14, d12, d14
- vsubl.u8 q15, d13, d15
-
- vst1.16 {q8}, [r0]! ;store diff
- vst1.16 {q9}, [r0]!
- vst1.16 {q10}, [r0]!
- vst1.16 {q11}, [r0]!
- vst1.16 {q12}, [r0]!
- vst1.16 {q13}, [r0]!
- vst1.16 {q14}, [r0]!
- vst1.16 {q15}, [r0]!
-
- subs r12, r12, #1
- bne subtract_mby_loop
-
- bx lr
- ENDP
-
-;=================================
-;void vp8_subtract_mbuv_neon(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride)
-|vp8_subtract_mbuv_neon| PROC
- ldr r12, [sp]
-
-;u
- add r0, r0, #512 ; short *udiff = diff + 256;
- add r3, r3, #256 ; unsigned char *upred = pred + 256;
-
- vld1.8 {d0}, [r1], r12 ;load src
- vld1.8 {d1}, [r3]! ;load pred
- vld1.8 {d2}, [r1], r12
- vld1.8 {d3}, [r3]!
- vld1.8 {d4}, [r1], r12
- vld1.8 {d5}, [r3]!
- vld1.8 {d6}, [r1], r12
- vld1.8 {d7}, [r3]!
- vld1.8 {d8}, [r1], r12
- vld1.8 {d9}, [r3]!
- vld1.8 {d10}, [r1], r12
- vld1.8 {d11}, [r3]!
- vld1.8 {d12}, [r1], r12
- vld1.8 {d13}, [r3]!
- vld1.8 {d14}, [r1], r12
- vld1.8 {d15}, [r3]!
-
- vsubl.u8 q8, d0, d1
- vsubl.u8 q9, d2, d3
- vsubl.u8 q10, d4, d5
- vsubl.u8 q11, d6, d7
- vsubl.u8 q12, d8, d9
- vsubl.u8 q13, d10, d11
- vsubl.u8 q14, d12, d13
- vsubl.u8 q15, d14, d15
-
- vst1.16 {q8}, [r0]! ;store diff
- vst1.16 {q9}, [r0]!
- vst1.16 {q10}, [r0]!
- vst1.16 {q11}, [r0]!
- vst1.16 {q12}, [r0]!
- vst1.16 {q13}, [r0]!
- vst1.16 {q14}, [r0]!
- vst1.16 {q15}, [r0]!
-
-;v
- vld1.8 {d0}, [r2], r12 ;load src
- vld1.8 {d1}, [r3]! ;load pred
- vld1.8 {d2}, [r2], r12
- vld1.8 {d3}, [r3]!
- vld1.8 {d4}, [r2], r12
- vld1.8 {d5}, [r3]!
- vld1.8 {d6}, [r2], r12
- vld1.8 {d7}, [r3]!
- vld1.8 {d8}, [r2], r12
- vld1.8 {d9}, [r3]!
- vld1.8 {d10}, [r2], r12
- vld1.8 {d11}, [r3]!
- vld1.8 {d12}, [r2], r12
- vld1.8 {d13}, [r3]!
- vld1.8 {d14}, [r2], r12
- vld1.8 {d15}, [r3]!
-
- vsubl.u8 q8, d0, d1
- vsubl.u8 q9, d2, d3
- vsubl.u8 q10, d4, d5
- vsubl.u8 q11, d6, d7
- vsubl.u8 q12, d8, d9
- vsubl.u8 q13, d10, d11
- vsubl.u8 q14, d12, d13
- vsubl.u8 q15, d14, d15
-
- vst1.16 {q8}, [r0]! ;store diff
- vst1.16 {q9}, [r0]!
- vst1.16 {q10}, [r0]!
- vst1.16 {q11}, [r0]!
- vst1.16 {q12}, [r0]!
- vst1.16 {q13}, [r0]!
- vst1.16 {q14}, [r0]!
- vst1.16 {q15}, [r0]!
-
- bx lr
- ENDP
-
- END
diff --git a/vp8/encoder/arm/neon/variance_neon.asm b/vp8/encoder/arm/neon/variance_neon.asm
deleted file mode 100644
index e1a4686..0000000
--- a/vp8/encoder/arm/neon/variance_neon.asm
+++ /dev/null
@@ -1,276 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_variance16x16_neon|
- EXPORT |vp8_variance16x8_neon|
- EXPORT |vp8_variance8x16_neon|
- EXPORT |vp8_variance8x8_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-|vp8_variance16x16_neon| PROC
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
- mov r12, #8
-
-variance16x16_neon_loop
- vld1.8 {q0}, [r0], r1 ;Load up source and reference
- vld1.8 {q2}, [r2], r3
- vld1.8 {q1}, [r0], r1
- vld1.8 {q3}, [r2], r3
-
- vsubl.u8 q11, d0, d4 ;calculate diff
- vsubl.u8 q12, d1, d5
- vsubl.u8 q13, d2, d6
- vsubl.u8 q14, d3, d7
-
- ;VPADAL adds adjacent pairs of elements of a vector, and accumulates
- ;the results into the elements of the destination vector. The explanation
- ;in ARM guide is wrong.
- vpadal.s16 q8, q11 ;calculate sum
- vmlal.s16 q9, d22, d22 ;calculate sse
- vmlal.s16 q10, d23, d23
-
- subs r12, r12, #1
-
- vpadal.s16 q8, q12
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
- vpadal.s16 q8, q13
- vmlal.s16 q9, d26, d26
- vmlal.s16 q10, d27, d27
- vpadal.s16 q8, q14
- vmlal.s16 q9, d28, d28
- vmlal.s16 q10, d29, d29
-
- bne variance16x16_neon_loop
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- ldr r12, [sp] ;load *sse from stack
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- ;vmov.32 r0, d0[0] ;this instruction costs a lot
- ;vmov.32 r1, d1[0]
- ;mul r0, r0, r0
- ;str r1, [r12]
- ;sub r0, r1, r0, asr #8
-
- ;sum is in [-255x256, 255x256]. sumxsum is 32-bit. Shift to right should
- ;have sign-bit exension, which is vshr.s. Have to use s32 to make it right.
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [r12] ;store sse
- vshr.s32 d10, d10, #8
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0] ;return
- bx lr
-
- ENDP
-
-;================================
-;unsigned int vp8_variance16x8_c(
-; unsigned char *src_ptr,
-; int source_stride,
-; unsigned char *ref_ptr,
-; int recon_stride,
-; unsigned int *sse)
-|vp8_variance16x8_neon| PROC
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
- mov r12, #4
-
-variance16x8_neon_loop
- vld1.8 {q0}, [r0], r1 ;Load up source and reference
- vld1.8 {q2}, [r2], r3
- vld1.8 {q1}, [r0], r1
- vld1.8 {q3}, [r2], r3
-
- vsubl.u8 q11, d0, d4 ;calculate diff
- vsubl.u8 q12, d1, d5
- vsubl.u8 q13, d2, d6
- vsubl.u8 q14, d3, d7
-
- vpadal.s16 q8, q11 ;calculate sum
- vmlal.s16 q9, d22, d22 ;calculate sse
- vmlal.s16 q10, d23, d23
-
- subs r12, r12, #1
-
- vpadal.s16 q8, q12
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
- vpadal.s16 q8, q13
- vmlal.s16 q9, d26, d26
- vmlal.s16 q10, d27, d27
- vpadal.s16 q8, q14
- vmlal.s16 q9, d28, d28
- vmlal.s16 q10, d29, d29
-
- bne variance16x8_neon_loop
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- ldr r12, [sp] ;load *sse from stack
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [r12] ;store sse
- vshr.s32 d10, d10, #7
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0] ;return
- bx lr
-
- ENDP
-
-;=================================
-;unsigned int vp8_variance8x16_c(
-; unsigned char *src_ptr,
-; int source_stride,
-; unsigned char *ref_ptr,
-; int recon_stride,
-; unsigned int *sse)
-
-|vp8_variance8x16_neon| PROC
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
- mov r12, #8
-
-variance8x16_neon_loop
- vld1.8 {d0}, [r0], r1 ;Load up source and reference
- vld1.8 {d4}, [r2], r3
- vld1.8 {d2}, [r0], r1
- vld1.8 {d6}, [r2], r3
-
- vsubl.u8 q11, d0, d4 ;calculate diff
- vsubl.u8 q12, d2, d6
-
- vpadal.s16 q8, q11 ;calculate sum
- vmlal.s16 q9, d22, d22 ;calculate sse
- vmlal.s16 q10, d23, d23
-
- subs r12, r12, #1
-
- vpadal.s16 q8, q12
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
-
- bne variance8x16_neon_loop
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- ldr r12, [sp] ;load *sse from stack
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [r12] ;store sse
- vshr.s32 d10, d10, #7
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0] ;return
- bx lr
-
- ENDP
-
-;==================================
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-|vp8_variance8x8_neon| PROC
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
- mov r12, #2
-
-variance8x8_neon_loop
- vld1.8 {d0}, [r0], r1 ;Load up source and reference
- vld1.8 {d4}, [r2], r3
- vld1.8 {d1}, [r0], r1
- vld1.8 {d5}, [r2], r3
- vld1.8 {d2}, [r0], r1
- vld1.8 {d6}, [r2], r3
- vld1.8 {d3}, [r0], r1
- vld1.8 {d7}, [r2], r3
-
- vsubl.u8 q11, d0, d4 ;calculate diff
- vsubl.u8 q12, d1, d5
- vsubl.u8 q13, d2, d6
- vsubl.u8 q14, d3, d7
-
- vpadal.s16 q8, q11 ;calculate sum
- vmlal.s16 q9, d22, d22 ;calculate sse
- vmlal.s16 q10, d23, d23
-
- subs r12, r12, #1
-
- vpadal.s16 q8, q12
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
- vpadal.s16 q8, q13
- vmlal.s16 q9, d26, d26
- vmlal.s16 q10, d27, d27
- vpadal.s16 q8, q14
- vmlal.s16 q9, d28, d28
- vmlal.s16 q10, d29, d29
-
- bne variance8x8_neon_loop
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- ldr r12, [sp] ;load *sse from stack
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [r12] ;store sse
- vshr.s32 d10, d10, #6
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0] ;return
- bx lr
-
- ENDP
-
- END
diff --git a/vp8/encoder/arm/neon/vp8_memcpy_neon.asm b/vp8/encoder/arm/neon/vp8_memcpy_neon.asm
deleted file mode 100644
index b0450e5..0000000
--- a/vp8/encoder/arm/neon/vp8_memcpy_neon.asm
+++ /dev/null
@@ -1,68 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_memcpy_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;=========================================
-;void vp8_memcpy_neon(unsigned char *dst_ptr, unsigned char *src_ptr, int sz);
-|vp8_memcpy_neon| PROC
- ;pld [r1] ;preload pred data
- ;pld [r1, #128]
- ;pld [r1, #256]
- ;pld [r1, #384]
-
- mov r12, r2, lsr #8 ;copy 256 bytes data at one time
-
-memcpy_neon_loop
- vld1.8 {q0, q1}, [r1]! ;load src data
- subs r12, r12, #1
- vld1.8 {q2, q3}, [r1]!
- vst1.8 {q0, q1}, [r0]! ;copy to dst_ptr
- vld1.8 {q4, q5}, [r1]!
- vst1.8 {q2, q3}, [r0]!
- vld1.8 {q6, q7}, [r1]!
- vst1.8 {q4, q5}, [r0]!
- vld1.8 {q8, q9}, [r1]!
- vst1.8 {q6, q7}, [r0]!
- vld1.8 {q10, q11}, [r1]!
- vst1.8 {q8, q9}, [r0]!
- vld1.8 {q12, q13}, [r1]!
- vst1.8 {q10, q11}, [r0]!
- vld1.8 {q14, q15}, [r1]!
- vst1.8 {q12, q13}, [r0]!
- vst1.8 {q14, q15}, [r0]!
-
- ;pld [r1] ;preload pred data -- need to adjust for real device
- ;pld [r1, #128]
- ;pld [r1, #256]
- ;pld [r1, #384]
-
- bne memcpy_neon_loop
-
- ands r3, r2, #0xff ;extra copy
- beq done_copy_neon_loop
-
-extra_copy_neon_loop
- vld1.8 {q0}, [r1]! ;load src data
- subs r3, r3, #16
- vst1.8 {q0}, [r0]!
- bne extra_copy_neon_loop
-
-done_copy_neon_loop
- bx lr
- ENDP
-
- END
diff --git a/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm b/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm
deleted file mode 100644
index 6af4e87..0000000
--- a/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm
+++ /dev/null
@@ -1,173 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_mse16x16_neon|
- EXPORT |vp8_get16x16pred_error_neon|
- EXPORT |vp8_get4x4sse_cs_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;============================
-; r0 unsigned char *src_ptr
-; r1 int source_stride
-; r2 unsigned char *ref_ptr
-; r3 int recon_stride
-; stack unsigned int *sse
-;note: in this function, sum is never used. So, we can remove this part of calculation
-;from vp8_variance().
-
-|vp8_mse16x16_neon| PROC
- vmov.i8 q7, #0 ;q7, q8, q9, q10 - sse
- vmov.i8 q8, #0
- vmov.i8 q9, #0
- vmov.i8 q10, #0
-
- mov r12, #8
-
-mse16x16_neon_loop
- vld1.8 {q0}, [r0], r1 ;Load up source and reference
- vld1.8 {q2}, [r2], r3
- vld1.8 {q1}, [r0], r1
- vld1.8 {q3}, [r2], r3
-
- vsubl.u8 q11, d0, d4
- vsubl.u8 q12, d1, d5
- vsubl.u8 q13, d2, d6
- vsubl.u8 q14, d3, d7
-
- vmlal.s16 q7, d22, d22
- vmlal.s16 q8, d23, d23
-
- subs r12, r12, #1
-
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
- vmlal.s16 q7, d26, d26
- vmlal.s16 q8, d27, d27
- vmlal.s16 q9, d28, d28
- vmlal.s16 q10, d29, d29
-
- bne mse16x16_neon_loop
-
- vadd.u32 q7, q7, q8
- vadd.u32 q9, q9, q10
-
- ldr r12, [sp] ;load *sse from stack
-
- vadd.u32 q10, q7, q9
- vpaddl.u32 q1, q10
- vadd.u64 d0, d2, d3
-
- vst1.32 {d0[0]}, [r12]
- vmov.32 r0, d0[0]
-
- bx lr
-
- ENDP
-
-;============================
-; r0 unsigned char *src_ptr
-; r1 int src_stride
-; r2 unsigned char *ref_ptr
-; r3 int ref_stride
-|vp8_get16x16pred_error_neon| PROC
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - pred_error
- vmov.i8 q10, #0
-
- mov r12, #8
-
-get16x16pred_error_neon_loop
- vld1.8 {q0}, [r0], r1 ;Load up source and reference
- vld1.8 {q2}, [r2], r3
- vld1.8 {q1}, [r0], r1
- vld1.8 {q3}, [r2], r3
-
- vsubl.u8 q11, d0, d4
- vsubl.u8 q12, d1, d5
- vsubl.u8 q13, d2, d6
- vsubl.u8 q14, d3, d7
-
- vpadal.s16 q8, q11
- vmlal.s16 q9, d22, d22
- vmlal.s16 q10, d23, d23
-
- subs r12, r12, #1
-
- vpadal.s16 q8, q12
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
- vpadal.s16 q8, q13
- vmlal.s16 q9, d26, d26
- vmlal.s16 q10, d27, d27
- vpadal.s16 q8, q14
- vmlal.s16 q9, d28, d28
- vmlal.s16 q10, d29, d29
-
- bne get16x16pred_error_neon_loop
-
- vadd.u32 q10, q9, q10
- vpaddl.s32 q0, q8
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vshr.s32 d10, d10, #8
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0]
- bx lr
-
- ENDP
-
-;=============================
-; r0 unsigned char *src_ptr,
-; r1 int source_stride,
-; r2 unsigned char *ref_ptr,
-; r3 int recon_stride
-|vp8_get4x4sse_cs_neon| PROC
- vld1.8 {d0}, [r0], r1 ;Load up source and reference
- vld1.8 {d4}, [r2], r3
- vld1.8 {d1}, [r0], r1
- vld1.8 {d5}, [r2], r3
- vld1.8 {d2}, [r0], r1
- vld1.8 {d6}, [r2], r3
- vld1.8 {d3}, [r0], r1
- vld1.8 {d7}, [r2], r3
-
- vsubl.u8 q11, d0, d4
- vsubl.u8 q12, d1, d5
- vsubl.u8 q13, d2, d6
- vsubl.u8 q14, d3, d7
-
- vmull.s16 q7, d22, d22
- vmull.s16 q8, d24, d24
- vmull.s16 q9, d26, d26
- vmull.s16 q10, d28, d28
-
- vadd.u32 q7, q7, q8
- vadd.u32 q9, q9, q10
- vadd.u32 q9, q7, q9
-
- vpaddl.u32 q1, q9
- vadd.u64 d0, d2, d3
-
- vmov.32 r0, d0[0]
- bx lr
-
- ENDP
-
- END
diff --git a/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm b/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm
deleted file mode 100644
index ba3decf..0000000
--- a/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm
+++ /dev/null
@@ -1,76 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_short_walsh4x4_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void vp8_short_walsh4x4_c(short *input, short *output, int pitch)
-
-|vp8_short_walsh4x4_neon| PROC
- vld1.16 {d2}, [r0], r2 ;load input
- vld1.16 {d3}, [r0], r2
- vld1.16 {d4}, [r0], r2
- vld1.16 {d5}, [r0], r2
-
- ;First for-loop
- ;transpose d2, d3, d4, d5. Then, d2=ip[0], d3=ip[1], d4=ip[2], d5=ip[3]
- vtrn.32 d2, d4
- vtrn.32 d3, d5
- vtrn.16 d2, d3
- vtrn.16 d4, d5
-
- vadd.s16 d6, d2, d5 ;a1 = ip[0]+ip[3]
- vadd.s16 d7, d3, d4 ;b1 = ip[1]+ip[2]
- vsub.s16 d8, d3, d4 ;c1 = ip[1]-ip[2]
- vsub.s16 d9, d2, d5 ;d1 = ip[0]-ip[3]
-
- vadd.s16 d2, d6, d7 ;op[0] = a1 + b1
- vsub.s16 d4, d6, d7 ;op[2] = a1 - b1
- vadd.s16 d3, d8, d9 ;op[1] = c1 + d1
- vsub.s16 d5, d9, d8 ;op[3] = d1 - c1
-
- ;Second for-loop
- ;transpose d2, d3, d4, d5. Then, d2=ip[0], d3=ip[4], d4=ip[8], d5=ip[12]
- vtrn.32 d2, d4
- vtrn.32 d3, d5
- vtrn.16 d2, d3
- vtrn.16 d4, d5
-
- vadd.s16 d6, d2, d5 ;a1 = ip[0]+ip[12]
- vadd.s16 d7, d3, d4 ;b1 = ip[4]+ip[8]
- vsub.s16 d8, d3, d4 ;c1 = ip[4]-ip[8]
- vsub.s16 d9, d2, d5 ;d1 = ip[0]-ip[12]
-
- vadd.s16 d2, d6, d7 ;a2 = a1 + b1;
- vsub.s16 d4, d6, d7 ;c2 = a1 - b1;
- vadd.s16 d3, d8, d9 ;b2 = c1 + d1;
- vsub.s16 d5, d9, d8 ;d2 = d1 - c1;
-
- vcgt.s16 q3, q1, #0
- vcgt.s16 q4, q2, #0
-
- vsub.s16 q1, q1, q3
- vsub.s16 q2, q2, q4
-
- vshr.s16 q1, q1, #1
- vshr.s16 q2, q2, #1
-
- vst1.16 {q1, q2}, [r1]
-
- bx lr
-
- ENDP
-
- END
diff --git a/vp8/encoder/arm/neon/vp8_subpixelvariance16x16_neon.asm b/vp8/encoder/arm/neon/vp8_subpixelvariance16x16_neon.asm
deleted file mode 100644
index 05539f4..0000000
--- a/vp8/encoder/arm/neon/vp8_subpixelvariance16x16_neon.asm
+++ /dev/null
@@ -1,423 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_sub_pixel_variance16x16_neon_func|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-; r0 unsigned char *src_ptr,
-; r1 int src_pixels_per_line,
-; r2 int xoffset,
-; r3 int yoffset,
-; stack(r4) unsigned char *dst_ptr,
-; stack(r5) int dst_pixels_per_line,
-; stack(r6) unsigned int *sse
-;note: most of the code is copied from bilinear_predict16x16_neon and vp8_variance16x16_neon.
-
-|vp8_sub_pixel_variance16x16_neon_func| PROC
- push {r4-r6, lr}
-
- adr r12, BilinearTaps_coeff
- ldr r4, [sp, #16] ;load *dst_ptr from stack
- ldr r5, [sp, #20] ;load dst_pixels_per_line from stack
- ldr r6, [sp, #24] ;load *sse from stack
-
- cmp r2, #0 ;skip first_pass filter if xoffset=0
- beq secondpass_bfilter16x16_only
-
- add r2, r12, r2, lsl #3 ;calculate filter location
-
- cmp r3, #0 ;skip second_pass filter if yoffset=0
-
- vld1.s32 {d31}, [r2] ;load first_pass filter
-
- beq firstpass_bfilter16x16_only
-
- sub sp, sp, #272 ;reserve space on stack for temporary storage
- vld1.u8 {d2, d3, d4}, [r0], r1 ;load src data
- mov lr, sp
- vld1.u8 {d5, d6, d7}, [r0], r1
-
- mov r2, #3 ;loop counter
- vld1.u8 {d8, d9, d10}, [r0], r1
-
- vdup.8 d0, d31[0] ;first_pass filter (d0 d1)
- vld1.u8 {d11, d12, d13}, [r0], r1
-
- vdup.8 d1, d31[4]
-
-;First Pass: output_height lines x output_width columns (17x16)
-vp8e_filt_blk2d_fp16x16_loop_neon
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- vmull.u8 q7, d2, d0 ;(src_ptr[0] * Filter[0])
- vmull.u8 q8, d3, d0
- vmull.u8 q9, d5, d0
- vmull.u8 q10, d6, d0
- vmull.u8 q11, d8, d0
- vmull.u8 q12, d9, d0
- vmull.u8 q13, d11, d0
- vmull.u8 q14, d12, d0
-
- vext.8 d2, d2, d3, #1 ;construct src_ptr[1]
- vext.8 d5, d5, d6, #1
- vext.8 d8, d8, d9, #1
- vext.8 d11, d11, d12, #1
-
- vmlal.u8 q7, d2, d1 ;(src_ptr[0] * Filter[1])
- vmlal.u8 q9, d5, d1
- vmlal.u8 q11, d8, d1
- vmlal.u8 q13, d11, d1
-
- vext.8 d3, d3, d4, #1
- vext.8 d6, d6, d7, #1
- vext.8 d9, d9, d10, #1
- vext.8 d12, d12, d13, #1
-
- vmlal.u8 q8, d3, d1 ;(src_ptr[0] * Filter[1])
- vmlal.u8 q10, d6, d1
- vmlal.u8 q12, d9, d1
- vmlal.u8 q14, d12, d1
-
- subs r2, r2, #1
-
- vqrshrn.u16 d14, q7, #7 ;shift/round/saturate to u8
- vqrshrn.u16 d15, q8, #7
- vqrshrn.u16 d16, q9, #7
- vqrshrn.u16 d17, q10, #7
- vqrshrn.u16 d18, q11, #7
- vqrshrn.u16 d19, q12, #7
- vqrshrn.u16 d20, q13, #7
-
- vld1.u8 {d2, d3, d4}, [r0], r1 ;load src data
- vqrshrn.u16 d21, q14, #7
- vld1.u8 {d5, d6, d7}, [r0], r1
-
- vst1.u8 {d14, d15, d16, d17}, [lr]! ;store result
- vld1.u8 {d8, d9, d10}, [r0], r1
- vst1.u8 {d18, d19, d20, d21}, [lr]!
- vld1.u8 {d11, d12, d13}, [r0], r1
-
- bne vp8e_filt_blk2d_fp16x16_loop_neon
-
-;First-pass filtering for rest 5 lines
- vld1.u8 {d14, d15, d16}, [r0], r1
-
- vmull.u8 q9, d2, d0 ;(src_ptr[0] * Filter[0])
- vmull.u8 q10, d3, d0
- vmull.u8 q11, d5, d0
- vmull.u8 q12, d6, d0
- vmull.u8 q13, d8, d0
- vmull.u8 q14, d9, d0
-
- vext.8 d2, d2, d3, #1 ;construct src_ptr[1]
- vext.8 d5, d5, d6, #1
- vext.8 d8, d8, d9, #1
-
- vmlal.u8 q9, d2, d1 ;(src_ptr[0] * Filter[1])
- vmlal.u8 q11, d5, d1
- vmlal.u8 q13, d8, d1
-
- vext.8 d3, d3, d4, #1
- vext.8 d6, d6, d7, #1
- vext.8 d9, d9, d10, #1
-
- vmlal.u8 q10, d3, d1 ;(src_ptr[0] * Filter[1])
- vmlal.u8 q12, d6, d1
- vmlal.u8 q14, d9, d1
-
- vmull.u8 q1, d11, d0
- vmull.u8 q2, d12, d0
- vmull.u8 q3, d14, d0
- vmull.u8 q4, d15, d0
-
- vext.8 d11, d11, d12, #1 ;construct src_ptr[1]
- vext.8 d14, d14, d15, #1
-
- vmlal.u8 q1, d11, d1 ;(src_ptr[0] * Filter[1])
- vmlal.u8 q3, d14, d1
-
- vext.8 d12, d12, d13, #1
- vext.8 d15, d15, d16, #1
-
- vmlal.u8 q2, d12, d1 ;(src_ptr[0] * Filter[1])
- vmlal.u8 q4, d15, d1
-
- vqrshrn.u16 d10, q9, #7 ;shift/round/saturate to u8
- vqrshrn.u16 d11, q10, #7
- vqrshrn.u16 d12, q11, #7
- vqrshrn.u16 d13, q12, #7
- vqrshrn.u16 d14, q13, #7
- vqrshrn.u16 d15, q14, #7
- vqrshrn.u16 d16, q1, #7
- vqrshrn.u16 d17, q2, #7
- vqrshrn.u16 d18, q3, #7
- vqrshrn.u16 d19, q4, #7
-
- vst1.u8 {d10, d11, d12, d13}, [lr]! ;store result
- vst1.u8 {d14, d15, d16, d17}, [lr]!
- vst1.u8 {d18, d19}, [lr]!
-
-;Second pass: 16x16
-;secondpass_filter
- add r3, r12, r3, lsl #3
- sub lr, lr, #272
-
- vld1.u32 {d31}, [r3] ;load second_pass filter
-
- sub sp, sp, #256
- mov r3, sp
-
- vld1.u8 {d22, d23}, [lr]! ;load src data
-
- vdup.8 d0, d31[0] ;second_pass filter parameters (d0 d1)
- vdup.8 d1, d31[4]
- mov r12, #4 ;loop counter
-
-vp8e_filt_blk2d_sp16x16_loop_neon
- vld1.u8 {d24, d25}, [lr]!
- vmull.u8 q1, d22, d0 ;(src_ptr[0] * Filter[0])
- vld1.u8 {d26, d27}, [lr]!
- vmull.u8 q2, d23, d0
- vld1.u8 {d28, d29}, [lr]!
- vmull.u8 q3, d24, d0
- vld1.u8 {d30, d31}, [lr]!
-
- vmull.u8 q4, d25, d0
- vmull.u8 q5, d26, d0
- vmull.u8 q6, d27, d0
- vmull.u8 q7, d28, d0
- vmull.u8 q8, d29, d0
-
- vmlal.u8 q1, d24, d1 ;(src_ptr[pixel_step] * Filter[1])
- vmlal.u8 q2, d25, d1
- vmlal.u8 q3, d26, d1
- vmlal.u8 q4, d27, d1
- vmlal.u8 q5, d28, d1
- vmlal.u8 q6, d29, d1
- vmlal.u8 q7, d30, d1
- vmlal.u8 q8, d31, d1
-
- subs r12, r12, #1
-
- vqrshrn.u16 d2, q1, #7 ;shift/round/saturate to u8
- vqrshrn.u16 d3, q2, #7
- vqrshrn.u16 d4, q3, #7
- vqrshrn.u16 d5, q4, #7
- vqrshrn.u16 d6, q5, #7
- vqrshrn.u16 d7, q6, #7
- vqrshrn.u16 d8, q7, #7
- vqrshrn.u16 d9, q8, #7
-
- vst1.u8 {d2, d3}, [r3]! ;store result
- vst1.u8 {d4, d5}, [r3]!
- vst1.u8 {d6, d7}, [r3]!
- vmov q11, q15
- vst1.u8 {d8, d9}, [r3]!
-
- bne vp8e_filt_blk2d_sp16x16_loop_neon
-
- b sub_pixel_variance16x16_neon
-
-;--------------------
-firstpass_bfilter16x16_only
- mov r2, #4 ;loop counter
- sub sp, sp, #528 ;reserve space on stack for temporary storage
- vdup.8 d0, d31[0] ;first_pass filter (d0 d1)
- vdup.8 d1, d31[4]
- mov r3, sp
-
-;First Pass: output_height lines x output_width columns (16x16)
-vp8e_filt_blk2d_fpo16x16_loop_neon
- vld1.u8 {d2, d3, d4}, [r0], r1 ;load src data
- vld1.u8 {d5, d6, d7}, [r0], r1
- vld1.u8 {d8, d9, d10}, [r0], r1
- vld1.u8 {d11, d12, d13}, [r0], r1
-
- pld [r0]
- pld [r0, r1]
- pld [r0, r1, lsl #1]
-
- vmull.u8 q7, d2, d0 ;(src_ptr[0] * Filter[0])
- vmull.u8 q8, d3, d0
- vmull.u8 q9, d5, d0
- vmull.u8 q10, d6, d0
- vmull.u8 q11, d8, d0
- vmull.u8 q12, d9, d0
- vmull.u8 q13, d11, d0
- vmull.u8 q14, d12, d0
-
- vext.8 d2, d2, d3, #1 ;construct src_ptr[1]
- vext.8 d5, d5, d6, #1
- vext.8 d8, d8, d9, #1
- vext.8 d11, d11, d12, #1
-
- vmlal.u8 q7, d2, d1 ;(src_ptr[0] * Filter[1])
- vmlal.u8 q9, d5, d1
- vmlal.u8 q11, d8, d1
- vmlal.u8 q13, d11, d1
-
- vext.8 d3, d3, d4, #1
- vext.8 d6, d6, d7, #1
- vext.8 d9, d9, d10, #1
- vext.8 d12, d12, d13, #1
-
- vmlal.u8 q8, d3, d1 ;(src_ptr[0] * Filter[1])
- vmlal.u8 q10, d6, d1
- vmlal.u8 q12, d9, d1
- vmlal.u8 q14, d12, d1
-
- subs r2, r2, #1
-
- vqrshrn.u16 d14, q7, #7 ;shift/round/saturate to u8
- vqrshrn.u16 d15, q8, #7
- vqrshrn.u16 d16, q9, #7
- vqrshrn.u16 d17, q10, #7
- vqrshrn.u16 d18, q11, #7
- vqrshrn.u16 d19, q12, #7
- vqrshrn.u16 d20, q13, #7
- vst1.u8 {d14, d15}, [r3]! ;store result
- vqrshrn.u16 d21, q14, #7
-
- vst1.u8 {d16, d17}, [r3]!
- vst1.u8 {d18, d19}, [r3]!
- vst1.u8 {d20, d21}, [r3]!
-
- bne vp8e_filt_blk2d_fpo16x16_loop_neon
-
- b sub_pixel_variance16x16_neon
-
-;---------------------
-secondpass_bfilter16x16_only
-;Second pass: 16x16
-;secondpass_filter
- sub sp, sp, #528 ;reserve space on stack for temporary storage
- add r3, r12, r3, lsl #3
- mov r12, #4 ;loop counter
- vld1.u32 {d31}, [r3] ;load second_pass filter
- vld1.u8 {d22, d23}, [r0], r1 ;load src data
- mov r3, sp
-
- vdup.8 d0, d31[0] ;second_pass filter parameters (d0 d1)
- vdup.8 d1, d31[4]
-
-vp8e_filt_blk2d_spo16x16_loop_neon
- vld1.u8 {d24, d25}, [r0], r1
- vmull.u8 q1, d22, d0 ;(src_ptr[0] * Filter[0])
- vld1.u8 {d26, d27}, [r0], r1
- vmull.u8 q2, d23, d0
- vld1.u8 {d28, d29}, [r0], r1
- vmull.u8 q3, d24, d0
- vld1.u8 {d30, d31}, [r0], r1
-
- vmull.u8 q4, d25, d0
- vmull.u8 q5, d26, d0
- vmull.u8 q6, d27, d0
- vmull.u8 q7, d28, d0
- vmull.u8 q8, d29, d0
-
- vmlal.u8 q1, d24, d1 ;(src_ptr[pixel_step] * Filter[1])
- vmlal.u8 q2, d25, d1
- vmlal.u8 q3, d26, d1
- vmlal.u8 q4, d27, d1
- vmlal.u8 q5, d28, d1
- vmlal.u8 q6, d29, d1
- vmlal.u8 q7, d30, d1
- vmlal.u8 q8, d31, d1
-
- vqrshrn.u16 d2, q1, #7 ;shift/round/saturate to u8
- vqrshrn.u16 d3, q2, #7
- vqrshrn.u16 d4, q3, #7
- vqrshrn.u16 d5, q4, #7
- vqrshrn.u16 d6, q5, #7
- vqrshrn.u16 d7, q6, #7
- vqrshrn.u16 d8, q7, #7
- vqrshrn.u16 d9, q8, #7
-
- vst1.u8 {d2, d3}, [r3]! ;store result
- subs r12, r12, #1
- vst1.u8 {d4, d5}, [r3]!
- vmov q11, q15
- vst1.u8 {d6, d7}, [r3]!
- vst1.u8 {d8, d9}, [r3]!
-
- bne vp8e_filt_blk2d_spo16x16_loop_neon
-
- b sub_pixel_variance16x16_neon
-
-;----------------------------
-;variance16x16
-sub_pixel_variance16x16_neon
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
- sub r3, r3, #256
- mov r12, #8
-
-sub_pixel_variance16x16_neon_loop
- vld1.8 {q0}, [r3]! ;Load up source and reference
- vld1.8 {q2}, [r4], r5
- vld1.8 {q1}, [r3]!
- vld1.8 {q3}, [r4], r5
-
- vsubl.u8 q11, d0, d4 ;diff
- vsubl.u8 q12, d1, d5
- vsubl.u8 q13, d2, d6
- vsubl.u8 q14, d3, d7
-
- vpadal.s16 q8, q11 ;sum
- vmlal.s16 q9, d22, d22 ;sse
- vmlal.s16 q10, d23, d23
-
- subs r12, r12, #1
-
- vpadal.s16 q8, q12
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
- vpadal.s16 q8, q13
- vmlal.s16 q9, d26, d26
- vmlal.s16 q10, d27, d27
- vpadal.s16 q8, q14
- vmlal.s16 q9, d28, d28
- vmlal.s16 q10, d29, d29
-
- bne sub_pixel_variance16x16_neon_loop
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [r6] ;store sse
- vshr.s32 d10, d10, #8
- vsub.s32 d0, d1, d10
-
- add sp, sp, #528
- vmov.32 r0, d0[0] ;return
-
- pop {r4-r6,pc}
-
- ENDP
-
-;-----------------
-
-bilinear_taps_coeff
- DCD 128, 0, 112, 16, 96, 32, 80, 48, 64, 64, 48, 80, 32, 96, 16, 112
-
- END
diff --git a/vp8/encoder/arm/neon/vp8_subpixelvariance16x16s_neon.asm b/vp8/encoder/arm/neon/vp8_subpixelvariance16x16s_neon.asm
deleted file mode 100644
index 0a2b71c..0000000
--- a/vp8/encoder/arm/neon/vp8_subpixelvariance16x16s_neon.asm
+++ /dev/null
@@ -1,572 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_variance_halfpixvar16x16_h_neon|
- EXPORT |vp8_variance_halfpixvar16x16_v_neon|
- EXPORT |vp8_variance_halfpixvar16x16_hv_neon|
- EXPORT |vp8_sub_pixel_variance16x16s_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-;================================================
-;unsigned int vp8_variance_halfpixvar16x16_h_neon
-;(
-; unsigned char *src_ptr, r0
-; int src_pixels_per_line, r1
-; unsigned char *dst_ptr, r2
-; int dst_pixels_per_line, r3
-; unsigned int *sse
-;);
-;================================================
-|vp8_variance_halfpixvar16x16_h_neon| PROC
- push {lr}
-
- mov r12, #4 ;loop counter
- ldr lr, [sp, #4] ;load *sse from stack
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
-;First Pass: output_height lines x output_width columns (16x16)
-vp8_filt_fpo16x16s_4_0_loop_neon
- vld1.u8 {d0, d1, d2, d3}, [r0], r1 ;load src data
- vld1.8 {q11}, [r2], r3
- vld1.u8 {d4, d5, d6, d7}, [r0], r1
- vld1.8 {q12}, [r2], r3
- vld1.u8 {d8, d9, d10, d11}, [r0], r1
- vld1.8 {q13}, [r2], r3
- vld1.u8 {d12, d13, d14, d15}, [r0], r1
-
- ;pld [r0]
- ;pld [r0, r1]
- ;pld [r0, r1, lsl #1]
-
- vext.8 q1, q0, q1, #1 ;construct src_ptr[1]
- vext.8 q3, q2, q3, #1
- vext.8 q5, q4, q5, #1
- vext.8 q7, q6, q7, #1
-
- vrhadd.u8 q0, q0, q1 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
- vld1.8 {q14}, [r2], r3
- vrhadd.u8 q1, q2, q3
- vrhadd.u8 q2, q4, q5
- vrhadd.u8 q3, q6, q7
-
- vsubl.u8 q4, d0, d22 ;diff
- vsubl.u8 q5, d1, d23
- vsubl.u8 q6, d2, d24
- vsubl.u8 q7, d3, d25
- vsubl.u8 q0, d4, d26
- vsubl.u8 q1, d5, d27
- vsubl.u8 q2, d6, d28
- vsubl.u8 q3, d7, d29
-
- vpadal.s16 q8, q4 ;sum
- vmlal.s16 q9, d8, d8 ;sse
- vmlal.s16 q10, d9, d9
-
- subs r12, r12, #1
-
- vpadal.s16 q8, q5
- vmlal.s16 q9, d10, d10
- vmlal.s16 q10, d11, d11
- vpadal.s16 q8, q6
- vmlal.s16 q9, d12, d12
- vmlal.s16 q10, d13, d13
- vpadal.s16 q8, q7
- vmlal.s16 q9, d14, d14
- vmlal.s16 q10, d15, d15
-
- vpadal.s16 q8, q0 ;sum
- vmlal.s16 q9, d0, d0 ;sse
- vmlal.s16 q10, d1, d1
- vpadal.s16 q8, q1
- vmlal.s16 q9, d2, d2
- vmlal.s16 q10, d3, d3
- vpadal.s16 q8, q2
- vmlal.s16 q9, d4, d4
- vmlal.s16 q10, d5, d5
- vpadal.s16 q8, q3
- vmlal.s16 q9, d6, d6
- vmlal.s16 q10, d7, d7
-
- bne vp8_filt_fpo16x16s_4_0_loop_neon
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [lr] ;store sse
- vshr.s32 d10, d10, #8
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0] ;return
- pop {pc}
- ENDP
-
-;================================================
-;unsigned int vp8_variance_halfpixvar16x16_v_neon
-;(
-; unsigned char *src_ptr, r0
-; int src_pixels_per_line, r1
-; unsigned char *dst_ptr, r2
-; int dst_pixels_per_line, r3
-; unsigned int *sse
-;);
-;================================================
-|vp8_variance_halfpixvar16x16_v_neon| PROC
- push {lr}
-
- mov r12, #4 ;loop counter
-
- vld1.u8 {q0}, [r0], r1 ;load src data
- ldr lr, [sp, #4] ;load *sse from stack
-
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
-vp8_filt_spo16x16s_0_4_loop_neon
- vld1.u8 {q2}, [r0], r1
- vld1.8 {q1}, [r2], r3
- vld1.u8 {q4}, [r0], r1
- vld1.8 {q3}, [r2], r3
- vld1.u8 {q6}, [r0], r1
- vld1.8 {q5}, [r2], r3
- vld1.u8 {q15}, [r0], r1
-
- vrhadd.u8 q0, q0, q2
- vld1.8 {q7}, [r2], r3
- vrhadd.u8 q2, q2, q4
- vrhadd.u8 q4, q4, q6
- vrhadd.u8 q6, q6, q15
-
- vsubl.u8 q11, d0, d2 ;diff
- vsubl.u8 q12, d1, d3
- vsubl.u8 q13, d4, d6
- vsubl.u8 q14, d5, d7
- vsubl.u8 q0, d8, d10
- vsubl.u8 q1, d9, d11
- vsubl.u8 q2, d12, d14
- vsubl.u8 q3, d13, d15
-
- vpadal.s16 q8, q11 ;sum
- vmlal.s16 q9, d22, d22 ;sse
- vmlal.s16 q10, d23, d23
-
- subs r12, r12, #1
-
- vpadal.s16 q8, q12
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
- vpadal.s16 q8, q13
- vmlal.s16 q9, d26, d26
- vmlal.s16 q10, d27, d27
- vpadal.s16 q8, q14
- vmlal.s16 q9, d28, d28
- vmlal.s16 q10, d29, d29
-
- vpadal.s16 q8, q0 ;sum
- vmlal.s16 q9, d0, d0 ;sse
- vmlal.s16 q10, d1, d1
- vpadal.s16 q8, q1
- vmlal.s16 q9, d2, d2
- vmlal.s16 q10, d3, d3
- vpadal.s16 q8, q2
- vmlal.s16 q9, d4, d4
- vmlal.s16 q10, d5, d5
-
- vmov q0, q15
-
- vpadal.s16 q8, q3
- vmlal.s16 q9, d6, d6
- vmlal.s16 q10, d7, d7
-
- bne vp8_filt_spo16x16s_0_4_loop_neon
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [lr] ;store sse
- vshr.s32 d10, d10, #8
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0] ;return
- pop {pc}
- ENDP
-
-;================================================
-;unsigned int vp8_variance_halfpixvar16x16_hv_neon
-;(
-; unsigned char *src_ptr, r0
-; int src_pixels_per_line, r1
-; unsigned char *dst_ptr, r2
-; int dst_pixels_per_line, r3
-; unsigned int *sse
-;);
-;================================================
-|vp8_variance_halfpixvar16x16_hv_neon| PROC
- push {lr}
-
- vld1.u8 {d0, d1, d2, d3}, [r0], r1 ;load src data
-
- ldr lr, [sp, #4] ;load *sse from stack
- vmov.i8 q13, #0 ;q8 - sum
- vext.8 q1, q0, q1, #1 ;construct src_ptr[1]
-
- vmov.i8 q14, #0 ;q9, q10 - sse
- vmov.i8 q15, #0
-
- mov r12, #4 ;loop counter
- vrhadd.u8 q0, q0, q1 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
-
-;First Pass: output_height lines x output_width columns (17x16)
-vp8_filt16x16s_4_4_loop_neon
- vld1.u8 {d4, d5, d6, d7}, [r0], r1
- vld1.u8 {d8, d9, d10, d11}, [r0], r1
- vld1.u8 {d12, d13, d14, d15}, [r0], r1
- vld1.u8 {d16, d17, d18, d19}, [r0], r1
-
- ;pld [r0]
- ;pld [r0, r1]
- ;pld [r0, r1, lsl #1]
-
- vext.8 q3, q2, q3, #1 ;construct src_ptr[1]
- vext.8 q5, q4, q5, #1
- vext.8 q7, q6, q7, #1
- vext.8 q9, q8, q9, #1
-
- vrhadd.u8 q1, q2, q3 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
- vrhadd.u8 q2, q4, q5
- vrhadd.u8 q3, q6, q7
- vrhadd.u8 q4, q8, q9
-
- vld1.8 {q5}, [r2], r3
- vrhadd.u8 q0, q0, q1
- vld1.8 {q6}, [r2], r3
- vrhadd.u8 q1, q1, q2
- vld1.8 {q7}, [r2], r3
- vrhadd.u8 q2, q2, q3
- vld1.8 {q8}, [r2], r3
- vrhadd.u8 q3, q3, q4
-
- vsubl.u8 q9, d0, d10 ;diff
- vsubl.u8 q10, d1, d11
- vsubl.u8 q11, d2, d12
- vsubl.u8 q12, d3, d13
-
- vsubl.u8 q0, d4, d14 ;diff
- vsubl.u8 q1, d5, d15
- vsubl.u8 q5, d6, d16
- vsubl.u8 q6, d7, d17
-
- vpadal.s16 q13, q9 ;sum
- vmlal.s16 q14, d18, d18 ;sse
- vmlal.s16 q15, d19, d19
-
- vpadal.s16 q13, q10 ;sum
- vmlal.s16 q14, d20, d20 ;sse
- vmlal.s16 q15, d21, d21
-
- vpadal.s16 q13, q11 ;sum
- vmlal.s16 q14, d22, d22 ;sse
- vmlal.s16 q15, d23, d23
-
- vpadal.s16 q13, q12 ;sum
- vmlal.s16 q14, d24, d24 ;sse
- vmlal.s16 q15, d25, d25
-
- subs r12, r12, #1
-
- vpadal.s16 q13, q0 ;sum
- vmlal.s16 q14, d0, d0 ;sse
- vmlal.s16 q15, d1, d1
-
- vpadal.s16 q13, q1 ;sum
- vmlal.s16 q14, d2, d2 ;sse
- vmlal.s16 q15, d3, d3
-
- vpadal.s16 q13, q5 ;sum
- vmlal.s16 q14, d10, d10 ;sse
- vmlal.s16 q15, d11, d11
-
- vmov q0, q4
-
- vpadal.s16 q13, q6 ;sum
- vmlal.s16 q14, d12, d12 ;sse
- vmlal.s16 q15, d13, d13
-
- bne vp8_filt16x16s_4_4_loop_neon
-
- vadd.u32 q15, q14, q15 ;accumulate sse
- vpaddl.s32 q0, q13 ;accumulate sum
-
- vpaddl.u32 q1, q15
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [lr] ;store sse
- vshr.s32 d10, d10, #8
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0] ;return
- pop {pc}
- ENDP
-
-;==============================
-; r0 unsigned char *src_ptr,
-; r1 int src_pixels_per_line,
-; r2 int xoffset,
-; r3 int yoffset,
-; stack unsigned char *dst_ptr,
-; stack int dst_pixels_per_line,
-; stack unsigned int *sse
-;note: in vp8_find_best_half_pixel_step()(called when 8<Speed<15), and first call of vp8_find_best_sub_pixel_step()
-;(called when speed<=8). xoffset/yoffset can only be 4 or 0, which means either by pass the filter,
-;or filter coeff is {64, 64}. This simplified program only works in this situation.
-;note: It happens that both xoffset and yoffset are zero. This can be handled in c code later.
-
-|vp8_sub_pixel_variance16x16s_neon| PROC
- push {r4, lr}
-
- ldr r4, [sp, #8] ;load *dst_ptr from stack
- ldr r12, [sp, #12] ;load dst_pixels_per_line from stack
- ldr lr, [sp, #16] ;load *sse from stack
-
- cmp r2, #0 ;skip first_pass filter if xoffset=0
- beq secondpass_bfilter16x16s_only
-
- cmp r3, #0 ;skip second_pass filter if yoffset=0
- beq firstpass_bfilter16x16s_only
-
- vld1.u8 {d0, d1, d2, d3}, [r0], r1 ;load src data
- sub sp, sp, #256 ;reserve space on stack for temporary storage
- vext.8 q1, q0, q1, #1 ;construct src_ptr[1]
- mov r3, sp
- mov r2, #4 ;loop counter
- vrhadd.u8 q0, q0, q1 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
-
-;First Pass: output_height lines x output_width columns (17x16)
-vp8e_filt_blk2d_fp16x16s_loop_neon
- vld1.u8 {d4, d5, d6, d7}, [r0], r1
- vld1.u8 {d8, d9, d10, d11}, [r0], r1
- vld1.u8 {d12, d13, d14, d15}, [r0], r1
- vld1.u8 {d16, d17, d18, d19}, [r0], r1
-
- ;pld [r0]
- ;pld [r0, r1]
- ;pld [r0, r1, lsl #1]
-
- vext.8 q3, q2, q3, #1 ;construct src_ptr[1]
- vext.8 q5, q4, q5, #1
- vext.8 q7, q6, q7, #1
- vext.8 q9, q8, q9, #1
-
- vrhadd.u8 q1, q2, q3 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
- vrhadd.u8 q2, q4, q5
- vrhadd.u8 q3, q6, q7
- vrhadd.u8 q4, q8, q9
-
- vrhadd.u8 q0, q0, q1
- vrhadd.u8 q1, q1, q2
- vrhadd.u8 q2, q2, q3
- vrhadd.u8 q3, q3, q4
-
- subs r2, r2, #1
- vst1.u8 {d0, d1 ,d2, d3}, [r3]! ;store result
- vmov q0, q4
- vst1.u8 {d4, d5, d6, d7}, [r3]!
-
- bne vp8e_filt_blk2d_fp16x16s_loop_neon
-
- b sub_pixel_variance16x16s_neon
-
-;--------------------
-firstpass_bfilter16x16s_only
- mov r2, #2 ;loop counter
- sub sp, sp, #256 ;reserve space on stack for temporary storage
- mov r3, sp
-
-;First Pass: output_height lines x output_width columns (16x16)
-vp8e_filt_blk2d_fpo16x16s_loop_neon
- vld1.u8 {d0, d1, d2, d3}, [r0], r1 ;load src data
- vld1.u8 {d4, d5, d6, d7}, [r0], r1
- vld1.u8 {d8, d9, d10, d11}, [r0], r1
- vld1.u8 {d12, d13, d14, d15}, [r0], r1
-
- ;pld [r0]
- ;pld [r0, r1]
- ;pld [r0, r1, lsl #1]
-
- vext.8 q1, q0, q1, #1 ;construct src_ptr[1]
- vld1.u8 {d16, d17, d18, d19}, [r0], r1
- vext.8 q3, q2, q3, #1
- vld1.u8 {d20, d21, d22, d23}, [r0], r1
- vext.8 q5, q4, q5, #1
- vld1.u8 {d24, d25, d26, d27}, [r0], r1
- vext.8 q7, q6, q7, #1
- vld1.u8 {d28, d29, d30, d31}, [r0], r1
- vext.8 q9, q8, q9, #1
- vext.8 q11, q10, q11, #1
- vext.8 q13, q12, q13, #1
- vext.8 q15, q14, q15, #1
-
- vrhadd.u8 q0, q0, q1 ;(src_ptr[0]+src_ptr[1])/round/shift right 1
- vrhadd.u8 q1, q2, q3
- vrhadd.u8 q2, q4, q5
- vrhadd.u8 q3, q6, q7
- vrhadd.u8 q4, q8, q9
- vrhadd.u8 q5, q10, q11
- vrhadd.u8 q6, q12, q13
- vrhadd.u8 q7, q14, q15
-
- subs r2, r2, #1
-
- vst1.u8 {d0, d1, d2, d3}, [r3]! ;store result
- vst1.u8 {d4, d5, d6, d7}, [r3]!
- vst1.u8 {d8, d9, d10, d11}, [r3]!
- vst1.u8 {d12, d13, d14, d15}, [r3]!
-
- bne vp8e_filt_blk2d_fpo16x16s_loop_neon
-
- b sub_pixel_variance16x16s_neon
-
-;---------------------
-secondpass_bfilter16x16s_only
- sub sp, sp, #256 ;reserve space on stack for temporary storage
-
- mov r2, #2 ;loop counter
- vld1.u8 {d0, d1}, [r0], r1 ;load src data
- mov r3, sp
-
-vp8e_filt_blk2d_spo16x16s_loop_neon
- vld1.u8 {d2, d3}, [r0], r1
- vld1.u8 {d4, d5}, [r0], r1
- vld1.u8 {d6, d7}, [r0], r1
- vld1.u8 {d8, d9}, [r0], r1
-
- vrhadd.u8 q0, q0, q1
- vld1.u8 {d10, d11}, [r0], r1
- vrhadd.u8 q1, q1, q2
- vld1.u8 {d12, d13}, [r0], r1
- vrhadd.u8 q2, q2, q3
- vld1.u8 {d14, d15}, [r0], r1
- vrhadd.u8 q3, q3, q4
- vld1.u8 {d16, d17}, [r0], r1
- vrhadd.u8 q4, q4, q5
- vrhadd.u8 q5, q5, q6
- vrhadd.u8 q6, q6, q7
- vrhadd.u8 q7, q7, q8
-
- subs r2, r2, #1
-
- vst1.u8 {d0, d1, d2, d3}, [r3]! ;store result
- vmov q0, q8
- vst1.u8 {d4, d5, d6, d7}, [r3]!
- vst1.u8 {d8, d9, d10, d11}, [r3]! ;store result
- vst1.u8 {d12, d13, d14, d15}, [r3]!
-
- bne vp8e_filt_blk2d_spo16x16s_loop_neon
-
- b sub_pixel_variance16x16s_neon
-
-;----------------------------
-;variance16x16
-sub_pixel_variance16x16s_neon
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
- sub r3, r3, #256
- mov r2, #4
-
-sub_pixel_variance16x16s_neon_loop
- vld1.8 {q0}, [r3]! ;Load up source and reference
- vld1.8 {q1}, [r4], r12
- vld1.8 {q2}, [r3]!
- vld1.8 {q3}, [r4], r12
- vld1.8 {q4}, [r3]!
- vld1.8 {q5}, [r4], r12
- vld1.8 {q6}, [r3]!
- vld1.8 {q7}, [r4], r12
-
- vsubl.u8 q11, d0, d2 ;diff
- vsubl.u8 q12, d1, d3
- vsubl.u8 q13, d4, d6
- vsubl.u8 q14, d5, d7
- vsubl.u8 q0, d8, d10
- vsubl.u8 q1, d9, d11
- vsubl.u8 q2, d12, d14
- vsubl.u8 q3, d13, d15
-
- vpadal.s16 q8, q11 ;sum
- vmlal.s16 q9, d22, d22 ;sse
- vmlal.s16 q10, d23, d23
-
- subs r2, r2, #1
-
- vpadal.s16 q8, q12
- vmlal.s16 q9, d24, d24
- vmlal.s16 q10, d25, d25
- vpadal.s16 q8, q13
- vmlal.s16 q9, d26, d26
- vmlal.s16 q10, d27, d27
- vpadal.s16 q8, q14
- vmlal.s16 q9, d28, d28
- vmlal.s16 q10, d29, d29
-
- vpadal.s16 q8, q0 ;sum
- vmlal.s16 q9, d0, d0 ;sse
- vmlal.s16 q10, d1, d1
- vpadal.s16 q8, q1
- vmlal.s16 q9, d2, d2
- vmlal.s16 q10, d3, d3
- vpadal.s16 q8, q2
- vmlal.s16 q9, d4, d4
- vmlal.s16 q10, d5, d5
- vpadal.s16 q8, q3
- vmlal.s16 q9, d6, d6
- vmlal.s16 q10, d7, d7
-
- bne sub_pixel_variance16x16s_neon_loop
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [lr] ;store sse
- vshr.s32 d10, d10, #8
- vsub.s32 d0, d1, d10
-
- add sp, sp, #256
- vmov.32 r0, d0[0] ;return
-
- pop {r4, pc}
- ENDP
-
- END
diff --git a/vp8/encoder/arm/neon/vp8_subpixelvariance8x8_neon.asm b/vp8/encoder/arm/neon/vp8_subpixelvariance8x8_neon.asm
deleted file mode 100644
index 86cf505..0000000
--- a/vp8/encoder/arm/neon/vp8_subpixelvariance8x8_neon.asm
+++ /dev/null
@@ -1,222 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_sub_pixel_variance8x8_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-; r0 unsigned char *src_ptr,
-; r1 int src_pixels_per_line,
-; r2 int xoffset,
-; r3 int yoffset,
-; stack(r4) unsigned char *dst_ptr,
-; stack(r5) int dst_pixels_per_line,
-; stack(r6) unsigned int *sse
-;note: most of the code is copied from bilinear_predict8x8_neon and vp8_variance8x8_neon.
-
-|vp8_sub_pixel_variance8x8_neon| PROC
- push {r4-r5, lr}
-
- adr r12, BilinearTaps_coeff
- ldr r4, [sp, #12] ;load *dst_ptr from stack
- ldr r5, [sp, #16] ;load dst_pixels_per_line from stack
- ldr lr, [sp, #20] ;load *sse from stack
-
- cmp r2, #0 ;skip first_pass filter if xoffset=0
- beq skip_firstpass_filter
-
-;First pass: output_height lines x output_width columns (9x8)
- add r2, r12, r2, lsl #3 ;calculate filter location
-
- vld1.u8 {q1}, [r0], r1 ;load src data
- vld1.u32 {d31}, [r2] ;load first_pass filter
- vld1.u8 {q2}, [r0], r1
- vdup.8 d0, d31[0] ;first_pass filter (d0 d1)
- vld1.u8 {q3}, [r0], r1
- vdup.8 d1, d31[4]
- vld1.u8 {q4}, [r0], r1
-
- vmull.u8 q6, d2, d0 ;(src_ptr[0] * Filter[0])
- vmull.u8 q7, d4, d0
- vmull.u8 q8, d6, d0
- vmull.u8 q9, d8, d0
-
- vext.8 d3, d2, d3, #1 ;construct src_ptr[-1]
- vext.8 d5, d4, d5, #1
- vext.8 d7, d6, d7, #1
- vext.8 d9, d8, d9, #1
-
- vmlal.u8 q6, d3, d1 ;(src_ptr[1] * Filter[1])
- vmlal.u8 q7, d5, d1
- vmlal.u8 q8, d7, d1
- vmlal.u8 q9, d9, d1
-
- vld1.u8 {q1}, [r0], r1 ;load src data
- vqrshrn.u16 d22, q6, #7 ;shift/round/saturate to u8
- vld1.u8 {q2}, [r0], r1
- vqrshrn.u16 d23, q7, #7
- vld1.u8 {q3}, [r0], r1
- vqrshrn.u16 d24, q8, #7
- vld1.u8 {q4}, [r0], r1
- vqrshrn.u16 d25, q9, #7
-
- ;first_pass filtering on the rest 5-line data
- vld1.u8 {q5}, [r0], r1
-
- vmull.u8 q6, d2, d0 ;(src_ptr[0] * Filter[0])
- vmull.u8 q7, d4, d0
- vmull.u8 q8, d6, d0
- vmull.u8 q9, d8, d0
- vmull.u8 q10, d10, d0
-
- vext.8 d3, d2, d3, #1 ;construct src_ptr[-1]
- vext.8 d5, d4, d5, #1
- vext.8 d7, d6, d7, #1
- vext.8 d9, d8, d9, #1
- vext.8 d11, d10, d11, #1
-
- vmlal.u8 q6, d3, d1 ;(src_ptr[1] * Filter[1])
- vmlal.u8 q7, d5, d1
- vmlal.u8 q8, d7, d1
- vmlal.u8 q9, d9, d1
- vmlal.u8 q10, d11, d1
-
- vqrshrn.u16 d26, q6, #7 ;shift/round/saturate to u8
- vqrshrn.u16 d27, q7, #7
- vqrshrn.u16 d28, q8, #7
- vqrshrn.u16 d29, q9, #7
- vqrshrn.u16 d30, q10, #7
-
-;Second pass: 8x8
-secondpass_filter
- cmp r3, #0 ;skip second_pass filter if yoffset=0
- ;skip_secondpass_filter
- beq sub_pixel_variance8x8_neon
-
- add r3, r12, r3, lsl #3
-
- vld1.u32 {d31}, [r3] ;load second_pass filter
-
- vdup.8 d0, d31[0] ;second_pass filter parameters (d0 d1)
- vdup.8 d1, d31[4]
-
- vmull.u8 q1, d22, d0 ;(src_ptr[0] * Filter[0])
- vmull.u8 q2, d23, d0
- vmull.u8 q3, d24, d0
- vmull.u8 q4, d25, d0
- vmull.u8 q5, d26, d0
- vmull.u8 q6, d27, d0
- vmull.u8 q7, d28, d0
- vmull.u8 q8, d29, d0
-
- vmlal.u8 q1, d23, d1 ;(src_ptr[pixel_step] * Filter[1])
- vmlal.u8 q2, d24, d1
- vmlal.u8 q3, d25, d1
- vmlal.u8 q4, d26, d1
- vmlal.u8 q5, d27, d1
- vmlal.u8 q6, d28, d1
- vmlal.u8 q7, d29, d1
- vmlal.u8 q8, d30, d1
-
- vqrshrn.u16 d22, q1, #7 ;shift/round/saturate to u8
- vqrshrn.u16 d23, q2, #7
- vqrshrn.u16 d24, q3, #7
- vqrshrn.u16 d25, q4, #7
- vqrshrn.u16 d26, q5, #7
- vqrshrn.u16 d27, q6, #7
- vqrshrn.u16 d28, q7, #7
- vqrshrn.u16 d29, q8, #7
-
- b sub_pixel_variance8x8_neon
-
-;--------------------
-skip_firstpass_filter
- vld1.u8 {d22}, [r0], r1 ;load src data
- vld1.u8 {d23}, [r0], r1
- vld1.u8 {d24}, [r0], r1
- vld1.u8 {d25}, [r0], r1
- vld1.u8 {d26}, [r0], r1
- vld1.u8 {d27}, [r0], r1
- vld1.u8 {d28}, [r0], r1
- vld1.u8 {d29}, [r0], r1
- vld1.u8 {d30}, [r0], r1
-
- b secondpass_filter
-
-;----------------------
-;vp8_variance8x8_neon
-sub_pixel_variance8x8_neon
- vmov.i8 q8, #0 ;q8 - sum
- vmov.i8 q9, #0 ;q9, q10 - sse
- vmov.i8 q10, #0
-
- mov r12, #2
-
-sub_pixel_variance8x8_neon_loop
- vld1.8 {d0}, [r4], r5 ;load dst data
- subs r12, r12, #1
- vld1.8 {d1}, [r4], r5
- vld1.8 {d2}, [r4], r5
- vsubl.u8 q4, d22, d0 ;calculate diff
- vld1.8 {d3}, [r4], r5
-
- vsubl.u8 q5, d23, d1
- vsubl.u8 q6, d24, d2
-
- vpadal.s16 q8, q4 ;sum
- vmlal.s16 q9, d8, d8 ;sse
- vmlal.s16 q10, d9, d9
-
- vsubl.u8 q7, d25, d3
-
- vpadal.s16 q8, q5
- vmlal.s16 q9, d10, d10
- vmlal.s16 q10, d11, d11
-
- vmov q11, q13
-
- vpadal.s16 q8, q6
- vmlal.s16 q9, d12, d12
- vmlal.s16 q10, d13, d13
-
- vmov q12, q14
-
- vpadal.s16 q8, q7
- vmlal.s16 q9, d14, d14
- vmlal.s16 q10, d15, d15
-
- bne sub_pixel_variance8x8_neon_loop
-
- vadd.u32 q10, q9, q10 ;accumulate sse
- vpaddl.s32 q0, q8 ;accumulate sum
-
- vpaddl.u32 q1, q10
- vadd.s64 d0, d0, d1
- vadd.u64 d1, d2, d3
-
- vmull.s32 q5, d0, d0
- vst1.32 {d1[0]}, [lr] ;store sse
- vshr.s32 d10, d10, #6
- vsub.s32 d0, d1, d10
-
- vmov.32 r0, d0[0] ;return
- pop {r4-r5, pc}
-
- ENDP
-
-;-----------------
-
-bilinear_taps_coeff
- DCD 128, 0, 112, 16, 96, 32, 80, 48, 64, 64, 48, 80, 32, 96, 16, 112
-
- END
diff --git a/vp8/encoder/arm/picklpf_arm.c b/vp8/encoder/arm/picklpf_arm.c
deleted file mode 100644
index 3fb370c..0000000
--- a/vp8/encoder/arm/picklpf_arm.c
+++ /dev/null
@@ -1,50 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/onyxc_int.h"
-#include "vp8/encoder/onyx_int.h"
-#include "vp8/encoder/quantize.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vpx_scale/yv12extend.h"
-#include "vpx_scale/vpxscale.h"
-#include "vp8/common/alloccommon.h"
-
-extern void vp8_memcpy_neon(unsigned char *dst_ptr, unsigned char *src_ptr, int sz);
-
-
-void
-vpxyv12_copy_partial_frame_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc, int Fraction)
-{
- unsigned char *src_y, *dst_y;
- int yheight;
- int ystride;
- int border;
- int yoffset;
- int linestocopy;
-
- border = src_ybc->border;
- yheight = src_ybc->y_height;
- ystride = src_ybc->y_stride;
-
- linestocopy = (yheight >> (Fraction + 4));
-
- if (linestocopy < 1)
- linestocopy = 1;
-
- linestocopy <<= 4;
-
- yoffset = ystride * ((yheight >> 5) * 16 - 8);
- src_y = src_ybc->y_buffer + yoffset;
- dst_y = dst_ybc->y_buffer + yoffset;
-
- //vpx_memcpy (dst_y, src_y, ystride * (linestocopy +16));
- vp8_memcpy_neon((unsigned char *)dst_y, (unsigned char *)src_y, (int)(ystride *(linestocopy + 16)));
-}
diff --git a/vp8/encoder/arm/quantize_arm.c b/vp8/encoder/arm/quantize_arm.c
deleted file mode 100644
index 0e3334a..0000000
--- a/vp8/encoder/arm/quantize_arm.c
+++ /dev/null
@@ -1,79 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <math.h>
-#include "vpx_mem/vpx_mem.h"
-
-#include "vp8/encoder/quantize.h"
-#include "vp8/common/entropy.h"
-
-DECLARE_ALIGNED(16, const short, vp8_rvsplus1_default_zig_zag1d[16]) =
-{
- 1, 2, 6, 7,
- 3, 5, 8, 13,
- 4, 9, 12, 14,
- 10, 11, 15, 16,
-};
-
-
-extern int vp8_fast_quantize_b_neon_func(short *coeff_ptr, short *zbin_ptr, short *qcoeff_ptr, short *dqcoeff_ptr, short *dequant_ptr, const short *scan_mask, short *round_ptr, short *quant_ptr);
-
-void vp8_fast_quantize_b_neon(BLOCK *b, BLOCKD *d)
-{
- d->eob = vp8_fast_quantize_b_neon_func(b->coeff, b->zbin, d->qcoeff, d->dqcoeff, d->dequant, vp8_rvsplus1_default_zig_zag1d, b->round, b->quant_fast);
-}
-
-/*
-//neon code is written according to the following rewritten c code
-void vp8_fast_quantize_b_neon(BLOCK *b,BLOCKD *d)
-{
- int i, rc, eob;
- int zbin;
- int x, x1, y, z, sz;
- short *coeff_ptr = &b->Coeff[0];
- short *zbin_ptr = &b->Zbin[0][0];
- short *round_ptr = &b->Round[0][0];
- short *quant_ptr = &b->Quant[0][0];
- short *qcoeff_ptr = d->qcoeff;
- short *dqcoeff_ptr= d->dqcoeff;
- short *dequant_ptr= &d->Dequant[0][0];
-
- eob = 0;
-
- for(i=0;i<16;i++)
- {
- z = coeff_ptr[i];
- zbin = zbin_ptr[i] ;
- x = abs(z); // x = abs(z)
-
- if(x>=zbin)
- {
- sz = (z>>31); // sign of z
- y = ((x+round_ptr[i])*quant_ptr[i])>>16; // quantize (x)
- x1 = (y^sz) - sz; // get the sign back
-
- qcoeff_ptr[i] = x1; // write to destination
- dqcoeff_ptr[i] = x1 * dequant_ptr[i]; // dequantized value
-
- if(y)
- {
- if(eob<vp8_rvsplus1_default_zig_zag1d[i])
- eob=(int)vp8_rvsplus1_default_zig_zag1d[i]; // last nonzero coeffs
- }
- }else
- {
- qcoeff_ptr[i] = 0; // write to destination
- dqcoeff_ptr[i] = 0; // dequantized value
- }
- }
- d->eob = eob;
-}
-*/
diff --git a/vp8/encoder/arm/quantize_arm.h b/vp8/encoder/arm/quantize_arm.h
deleted file mode 100644
index 0c6adf4..0000000
--- a/vp8/encoder/arm/quantize_arm.h
+++ /dev/null
@@ -1,36 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef QUANTIZE_ARM_H
-#define QUANTIZE_ARM_H
-
-#if HAVE_ARMV6
-
-extern prototype_quantize_block(vp8_fast_quantize_b_armv6);
-
-#undef vp8_quantize_fastquantb
-#define vp8_quantize_fastquantb vp8_fast_quantize_b_armv6
-
-#endif /* HAVE_ARMV6 */
-
-
-#if HAVE_ARMV7
-extern prototype_quantize_block(vp8_fast_quantize_b_neon);
-
-/* The neon quantizer has not been updated to match the new exact
- * quantizer introduced in commit e04e2935
- */
-//#undef vp8_quantize_fastquantb
-//#define vp8_quantize_fastquantb vp8_fast_quantize_b_neon
-
-#endif
-
-#endif
diff --git a/vp8/encoder/arm/variance_arm.c b/vp8/encoder/arm/variance_arm.c
deleted file mode 100644
index e77be9f..0000000
--- a/vp8/encoder/arm/variance_arm.c
+++ /dev/null
@@ -1,120 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "vpx_config.h"
-#include "vp8/encoder/variance.h"
-#include "vp8/common/filter.h"
-#include "vp8/common/arm/bilinearfilter_arm.h"
-
-#if HAVE_ARMV6
-
-unsigned int vp8_sub_pixel_variance8x8_armv6
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- unsigned short first_pass[10*8];
- unsigned char second_pass[8*8];
- const short *HFilter, *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- vp8_filter_block2d_bil_first_pass_armv6(src_ptr, first_pass,
- src_pixels_per_line,
- 9, 8, HFilter);
- vp8_filter_block2d_bil_second_pass_armv6(first_pass, second_pass,
- 8, 8, 8, VFilter);
-
- return vp8_variance8x8_armv6(second_pass, 8, dst_ptr,
- dst_pixels_per_line, sse);
-}
-
-unsigned int vp8_sub_pixel_variance16x16_armv6
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- unsigned short first_pass[36*16];
- unsigned char second_pass[20*16];
- const short *HFilter, *VFilter;
- unsigned int var;
-
- if (xoffset == 4 && yoffset == 0)
- {
- var = vp8_variance_halfpixvar16x16_h_armv6(src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, sse);
- }
- else if (xoffset == 0 && yoffset == 4)
- {
- var = vp8_variance_halfpixvar16x16_v_armv6(src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, sse);
- }
- else if (xoffset == 4 && yoffset == 4)
- {
- var = vp8_variance_halfpixvar16x16_hv_armv6(src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, sse);
- }
- else
- {
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- vp8_filter_block2d_bil_first_pass_armv6(src_ptr, first_pass,
- src_pixels_per_line,
- 17, 16, HFilter);
- vp8_filter_block2d_bil_second_pass_armv6(first_pass, second_pass,
- 16, 16, 16, VFilter);
-
- var = vp8_variance16x16_armv6(second_pass, 16, dst_ptr,
- dst_pixels_per_line, sse);
- }
- return var;
-}
-
-#endif /* HAVE_ARMV6 */
-
-
-#if HAVE_ARMV7
-
-unsigned int vp8_sub_pixel_variance16x16_neon
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- if (xoffset == 4 && yoffset == 0)
- return vp8_variance_halfpixvar16x16_h_neon(src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
- else if (xoffset == 0 && yoffset == 4)
- return vp8_variance_halfpixvar16x16_v_neon(src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
- else if (xoffset == 4 && yoffset == 4)
- return vp8_variance_halfpixvar16x16_hv_neon(src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
- else
- return vp8_sub_pixel_variance16x16_neon_func(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
-}
-
-#endif
diff --git a/vp8/encoder/arm/variance_arm.h b/vp8/encoder/arm/variance_arm.h
deleted file mode 100644
index 86de274..0000000
--- a/vp8/encoder/arm/variance_arm.h
+++ /dev/null
@@ -1,167 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VARIANCE_ARM_H
-#define VARIANCE_ARM_H
-
-#if HAVE_ARMV6
-
-extern prototype_sad(vp8_sad16x16_armv6);
-extern prototype_variance(vp8_variance16x16_armv6);
-extern prototype_variance(vp8_variance8x8_armv6);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_armv6);
-extern prototype_subpixvariance(vp8_sub_pixel_variance8x8_armv6);
-extern prototype_variance(vp8_variance_halfpixvar16x16_h_armv6);
-extern prototype_variance(vp8_variance_halfpixvar16x16_v_armv6);
-extern prototype_variance(vp8_variance_halfpixvar16x16_hv_armv6);
-extern prototype_variance(vp8_mse16x16_armv6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_variance_sad16x16
-#define vp8_variance_sad16x16 vp8_sad16x16_armv6
-
-#undef vp8_variance_subpixvar16x16
-#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_armv6
-
-#undef vp8_variance_subpixvar8x8
-#define vp8_variance_subpixvar8x8 vp8_sub_pixel_variance8x8_armv6
-
-#undef vp8_variance_var16x16
-#define vp8_variance_var16x16 vp8_variance16x16_armv6
-
-#undef vp8_variance_mse16x16
-#define vp8_variance_mse16x16 vp8_mse16x16_armv6
-
-#undef vp8_variance_var8x8
-#define vp8_variance_var8x8 vp8_variance8x8_armv6
-
-#undef vp8_variance_halfpixvar16x16_h
-#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_armv6
-
-#undef vp8_variance_halfpixvar16x16_v
-#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_armv6
-
-#undef vp8_variance_halfpixvar16x16_hv
-#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_armv6
-
-#endif /* !CONFIG_RUNTIME_CPU_DETECT */
-
-#endif /* HAVE_ARMV6 */
-
-
-#if HAVE_ARMV7
-extern prototype_sad(vp8_sad4x4_neon);
-extern prototype_sad(vp8_sad8x8_neon);
-extern prototype_sad(vp8_sad8x16_neon);
-extern prototype_sad(vp8_sad16x8_neon);
-extern prototype_sad(vp8_sad16x16_neon);
-
-//extern prototype_variance(vp8_variance4x4_c);
-extern prototype_variance(vp8_variance8x8_neon);
-extern prototype_variance(vp8_variance8x16_neon);
-extern prototype_variance(vp8_variance16x8_neon);
-extern prototype_variance(vp8_variance16x16_neon);
-
-//extern prototype_subpixvariance(vp8_sub_pixel_variance4x4_c);
-extern prototype_subpixvariance(vp8_sub_pixel_variance8x8_neon);
-//extern prototype_subpixvariance(vp8_sub_pixel_variance8x16_c);
-//extern prototype_subpixvariance(vp8_sub_pixel_variance16x8_c);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_neon);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_neon_func);
-extern prototype_variance(vp8_variance_halfpixvar16x16_h_neon);
-extern prototype_variance(vp8_variance_halfpixvar16x16_v_neon);
-extern prototype_variance(vp8_variance_halfpixvar16x16_hv_neon);
-
-//extern prototype_getmbss(vp8_get_mb_ss_c);
-extern prototype_variance(vp8_mse16x16_neon);
-extern prototype_sad(vp8_get16x16pred_error_neon);
-//extern prototype_variance2(vp8_get8x8var_c);
-//extern prototype_variance2(vp8_get16x16var_c);
-extern prototype_sad(vp8_get4x4sse_cs_neon);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_variance_sad4x4
-#define vp8_variance_sad4x4 vp8_sad4x4_neon
-
-#undef vp8_variance_sad8x8
-#define vp8_variance_sad8x8 vp8_sad8x8_neon
-
-#undef vp8_variance_sad8x16
-#define vp8_variance_sad8x16 vp8_sad8x16_neon
-
-#undef vp8_variance_sad16x8
-#define vp8_variance_sad16x8 vp8_sad16x8_neon
-
-#undef vp8_variance_sad16x16
-#define vp8_variance_sad16x16 vp8_sad16x16_neon
-
-//#undef vp8_variance_var4x4
-//#define vp8_variance_var4x4 vp8_variance4x4_c
-
-#undef vp8_variance_var8x8
-#define vp8_variance_var8x8 vp8_variance8x8_neon
-
-#undef vp8_variance_var8x16
-#define vp8_variance_var8x16 vp8_variance8x16_neon
-
-#undef vp8_variance_var16x8
-#define vp8_variance_var16x8 vp8_variance16x8_neon
-
-#undef vp8_variance_var16x16
-#define vp8_variance_var16x16 vp8_variance16x16_neon
-
-//#undef vp8_variance_subpixvar4x4
-//#define vp8_variance_subpixvar4x4 vp8_sub_pixel_variance4x4_c
-
-#undef vp8_variance_subpixvar8x8
-#define vp8_variance_subpixvar8x8 vp8_sub_pixel_variance8x8_neon
-
-//#undef vp8_variance_subpixvar8x16
-//#define vp8_variance_subpixvar8x16 vp8_sub_pixel_variance8x16_c
-
-//#undef vp8_variance_subpixvar16x8
-//#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_c
-
-#undef vp8_variance_subpixvar16x16
-#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_neon
-
-#undef vp8_variance_halfpixvar16x16_h
-#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_neon
-
-#undef vp8_variance_halfpixvar16x16_v
-#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_neon
-
-#undef vp8_variance_halfpixvar16x16_hv
-#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_neon
-
-//#undef vp8_variance_getmbss
-//#define vp8_variance_getmbss vp8_get_mb_ss_c
-
-#undef vp8_variance_mse16x16
-#define vp8_variance_mse16x16 vp8_mse16x16_neon
-
-#undef vp8_variance_get16x16prederror
-#define vp8_variance_get16x16prederror vp8_get16x16pred_error_neon
-
-//#undef vp8_variance_get8x8var
-//#define vp8_variance_get8x8var vp8_get8x8var_c
-
-//#undef vp8_variance_get16x16var
-//#define vp8_variance_get16x16var vp8_get16x16var_c
-
-#undef vp8_variance_get4x4sse_cs
-#define vp8_variance_get4x4sse_cs vp8_get4x4sse_cs_neon
-#endif
-
-#endif
-
-#endif
diff --git a/vp8/encoder/asm_enc_offsets.c b/vp8/encoder/asm_enc_offsets.c
deleted file mode 100644
index 9c81c8d..0000000
--- a/vp8/encoder/asm_enc_offsets.c
+++ /dev/null
@@ -1,105 +0,0 @@
-/*
- * Copyright (c) 2011 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include <stddef.h>
-
-#include "block.h"
-#include "vp8/common/blockd.h"
-#include "onyx_int.h"
-#include "treewriter.h"
-#include "tokenize.h"
-
-#define ct_assert(name,cond) \
- static void assert_##name(void) UNUSED;\
- static void assert_##name(void) {switch(0){case 0:case !!(cond):;}}
-
-#define DEFINE(sym, val) int sym = val;
-
-/*
-#define BLANK() asm volatile("\n->" : : )
-*/
-
-/*
- * int main(void)
- * {
- */
-
-//regular quantize
-DEFINE(vp8_block_coeff, offsetof(BLOCK, coeff));
-DEFINE(vp8_block_zbin, offsetof(BLOCK, zbin));
-DEFINE(vp8_block_round, offsetof(BLOCK, round));
-DEFINE(vp8_block_quant, offsetof(BLOCK, quant));
-DEFINE(vp8_block_quant_fast, offsetof(BLOCK, quant_fast));
-DEFINE(vp8_block_zbin_extra, offsetof(BLOCK, zbin_extra));
-DEFINE(vp8_block_zrun_zbin_boost, offsetof(BLOCK, zrun_zbin_boost));
-DEFINE(vp8_block_quant_shift, offsetof(BLOCK, quant_shift));
-
-DEFINE(vp8_blockd_qcoeff, offsetof(BLOCKD, qcoeff));
-DEFINE(vp8_blockd_dequant, offsetof(BLOCKD, dequant));
-DEFINE(vp8_blockd_dqcoeff, offsetof(BLOCKD, dqcoeff));
-DEFINE(vp8_blockd_eob, offsetof(BLOCKD, eob));
-
-// subtract
-DEFINE(vp8_block_base_src, offsetof(BLOCK, base_src));
-DEFINE(vp8_block_src, offsetof(BLOCK, src));
-DEFINE(vp8_block_src_diff, offsetof(BLOCK, src_diff));
-DEFINE(vp8_block_src_stride, offsetof(BLOCK, src_stride));
-
-DEFINE(vp8_blockd_predictor, offsetof(BLOCKD, predictor));
-
-//pack tokens
-DEFINE(vp8_writer_lowvalue, offsetof(vp8_writer, lowvalue));
-DEFINE(vp8_writer_range, offsetof(vp8_writer, range));
-DEFINE(vp8_writer_value, offsetof(vp8_writer, value));
-DEFINE(vp8_writer_count, offsetof(vp8_writer, count));
-DEFINE(vp8_writer_pos, offsetof(vp8_writer, pos));
-DEFINE(vp8_writer_buffer, offsetof(vp8_writer, buffer));
-
-DEFINE(tokenextra_token, offsetof(TOKENEXTRA, Token));
-DEFINE(tokenextra_extra, offsetof(TOKENEXTRA, Extra));
-DEFINE(tokenextra_context_tree, offsetof(TOKENEXTRA, context_tree));
-DEFINE(tokenextra_skip_eob_node, offsetof(TOKENEXTRA, skip_eob_node));
-DEFINE(TOKENEXTRA_SZ, sizeof(TOKENEXTRA));
-
-DEFINE(vp8_extra_bit_struct_sz, sizeof(vp8_extra_bit_struct));
-
-DEFINE(vp8_token_value, offsetof(vp8_token, value));
-DEFINE(vp8_token_len, offsetof(vp8_token, Len));
-
-DEFINE(vp8_extra_bit_struct_tree, offsetof(vp8_extra_bit_struct, tree));
-DEFINE(vp8_extra_bit_struct_prob, offsetof(vp8_extra_bit_struct, prob));
-DEFINE(vp8_extra_bit_struct_len, offsetof(vp8_extra_bit_struct, Len));
-DEFINE(vp8_extra_bit_struct_base_val, offsetof(vp8_extra_bit_struct, base_val));
-
-DEFINE(vp8_comp_tplist, offsetof(VP8_COMP, tplist));
-DEFINE(vp8_comp_common, offsetof(VP8_COMP, common));
-DEFINE(vp8_comp_bc2, offsetof(VP8_COMP, bc2));
-
-DEFINE(tokenlist_start, offsetof(TOKENLIST, start));
-DEFINE(tokenlist_stop, offsetof(TOKENLIST, stop));
-DEFINE(TOKENLIST_SZ, sizeof(TOKENLIST));
-
-DEFINE(vp8_common_mb_rows, offsetof(VP8_COMMON, mb_rows));
-
-// These two sizes are used in vp8cx_pack_tokens. They are hard coded
-// so if the size changes this will have to be adjusted.
-#if HAVE_ARMV5TE
-ct_assert(TOKENEXTRA_SZ, sizeof(TOKENEXTRA) == 8)
-ct_assert(vp8_extra_bit_struct_sz, sizeof(vp8_extra_bit_struct) == 16)
-#endif
-
-//add asserts for any offset that is not supported by assembly code
-//add asserts for any size that is not supported by assembly code
-/*
- * return 0;
- * }
- */
diff --git a/vp8/encoder/bitstream.c b/vp8/encoder/bitstream.c
deleted file mode 100644
index adbd106..0000000
--- a/vp8/encoder/bitstream.c
+++ /dev/null
@@ -1,1729 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/header.h"
-#include "encodemv.h"
-#include "vp8/common/entropymode.h"
-#include "vp8/common/findnearmv.h"
-#include "mcomp.h"
-#include "vp8/common/systemdependent.h"
-#include <assert.h>
-#include <stdio.h>
-#include "vp8/common/pragmas.h"
-#include "vpx_mem/vpx_mem.h"
-#include "bitstream.h"
-
-const int vp8cx_base_skip_false_prob[128] =
-{
- 255, 255, 255, 255, 255, 255, 255, 255,
- 255, 255, 255, 255, 255, 255, 255, 255,
- 255, 255, 255, 255, 255, 255, 255, 255,
- 255, 255, 255, 255, 255, 255, 255, 255,
- 255, 255, 255, 255, 255, 255, 255, 255,
- 255, 255, 255, 255, 255, 255, 255, 255,
- 255, 255, 255, 255, 255, 255, 255, 255,
- 251, 248, 244, 240, 236, 232, 229, 225,
- 221, 217, 213, 208, 204, 199, 194, 190,
- 187, 183, 179, 175, 172, 168, 164, 160,
- 157, 153, 149, 145, 142, 138, 134, 130,
- 127, 124, 120, 117, 114, 110, 107, 104,
- 101, 98, 95, 92, 89, 86, 83, 80,
- 77, 74, 71, 68, 65, 62, 59, 56,
- 53, 50, 47, 44, 41, 38, 35, 32,
- 30, 28, 26, 24, 22, 20, 18, 16,
-};
-#ifdef VP8REF
-#define __int64 long long
-#endif
-
-#if defined(SECTIONBITS_OUTPUT)
-unsigned __int64 Sectionbits[500];
-#endif
-
-#ifdef ENTROPY_STATS
-int intra_mode_stats[10][10][10];
-static unsigned int tree_update_hist [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1] [2];
-extern unsigned int active_section;
-#endif
-
-#ifdef MODE_STATS
-int count_mb_seg[4] = { 0, 0, 0, 0 };
-#endif
-
-
-static void update_mode(
- vp8_writer *const w,
- int n,
- vp8_token tok [/* n */],
- vp8_tree tree,
- vp8_prob Pnew [/* n-1 */],
- vp8_prob Pcur [/* n-1 */],
- unsigned int bct [/* n-1 */] [2],
- const unsigned int num_events[/* n */]
-)
-{
- unsigned int new_b = 0, old_b = 0;
- int i = 0;
-
- vp8_tree_probs_from_distribution(
- n--, tok, tree,
- Pnew, bct, num_events,
- 256, 1
- );
-
- do
- {
- new_b += vp8_cost_branch(bct[i], Pnew[i]);
- old_b += vp8_cost_branch(bct[i], Pcur[i]);
- }
- while (++i < n);
-
- if (new_b + (n << 8) < old_b)
- {
- int i = 0;
-
- vp8_write_bit(w, 1);
-
- do
- {
- const vp8_prob p = Pnew[i];
-
- vp8_write_literal(w, Pcur[i] = p ? p : 1, 8);
- }
- while (++i < n);
- }
- else
- vp8_write_bit(w, 0);
-}
-
-static void update_mbintra_mode_probs(VP8_COMP *cpi)
-{
- VP8_COMMON *const x = & cpi->common;
-
- vp8_writer *const w = & cpi->bc;
-
- {
- vp8_prob Pnew [VP8_YMODES-1];
- unsigned int bct [VP8_YMODES-1] [2];
-
- update_mode(
- w, VP8_YMODES, vp8_ymode_encodings, vp8_ymode_tree,
- Pnew, x->fc.ymode_prob, bct, (unsigned int *)cpi->ymode_count
- );
- }
- {
- vp8_prob Pnew [VP8_UV_MODES-1];
- unsigned int bct [VP8_UV_MODES-1] [2];
-
- update_mode(
- w, VP8_UV_MODES, vp8_uv_mode_encodings, vp8_uv_mode_tree,
- Pnew, x->fc.uv_mode_prob, bct, (unsigned int *)cpi->uv_mode_count
- );
- }
-}
-
-static void write_ymode(vp8_writer *bc, int m, const vp8_prob *p)
-{
- vp8_write_token(bc, vp8_ymode_tree, p, vp8_ymode_encodings + m);
-}
-
-static void kfwrite_ymode(vp8_writer *bc, int m, const vp8_prob *p)
-{
- vp8_write_token(bc, vp8_kf_ymode_tree, p, vp8_kf_ymode_encodings + m);
-}
-
-static void write_uv_mode(vp8_writer *bc, int m, const vp8_prob *p)
-{
- vp8_write_token(bc, vp8_uv_mode_tree, p, vp8_uv_mode_encodings + m);
-}
-
-
-static void write_bmode(vp8_writer *bc, int m, const vp8_prob *p)
-{
- vp8_write_token(bc, vp8_bmode_tree, p, vp8_bmode_encodings + m);
-}
-
-static void write_split(vp8_writer *bc, int x)
-{
- vp8_write_token(
- bc, vp8_mbsplit_tree, vp8_mbsplit_probs, vp8_mbsplit_encodings + x
- );
-}
-
-static const unsigned int norm[256] =
-{
- 0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
- 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
- 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
-};
-
-static void pack_tokens_c(vp8_writer *w, const TOKENEXTRA *p, int xcount)
-{
- const TOKENEXTRA *const stop = p + xcount;
- unsigned int split;
- unsigned int shift;
- int count = w->count;
- unsigned int range = w->range;
- unsigned int lowvalue = w->lowvalue;
-
- while (p < stop)
- {
- const int t = p->Token;
- vp8_token *const a = vp8_coef_encodings + t;
- const vp8_extra_bit_struct *const b = vp8_extra_bits + t;
- int i = 0;
- const unsigned char *pp = p->context_tree;
- int v = a->value;
- int n = a->Len;
-
- if (p->skip_eob_node)
- {
- n--;
- i = 2;
- }
-
- do
- {
- const int bb = (v >> --n) & 1;
- split = 1 + (((range - 1) * pp[i>>1]) >> 8);
- i = vp8_coef_tree[i+bb];
-
- if (bb)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- shift = norm[range];
- range <<= shift;
- count += shift;
-
- if (count >= 0)
- {
- int offset = shift - count;
-
- if ((lowvalue << (offset - 1)) & 0x80000000)
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
- }
-
- w->buffer[w->pos++] = (lowvalue >> (24 - offset));
- lowvalue <<= offset;
- shift = count;
- lowvalue &= 0xffffff;
- count -= 8 ;
- }
-
- lowvalue <<= shift;
- }
- while (n);
-
-
- if (b->base_val)
- {
- const int e = p->Extra, L = b->Len;
-
- if (L)
- {
- const unsigned char *pp = b->prob;
- int v = e >> 1;
- int n = L; /* number of bits in v, assumed nonzero */
- int i = 0;
-
- do
- {
- const int bb = (v >> --n) & 1;
- split = 1 + (((range - 1) * pp[i>>1]) >> 8);
- i = b->tree[i+bb];
-
- if (bb)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- shift = norm[range];
- range <<= shift;
- count += shift;
-
- if (count >= 0)
- {
- int offset = shift - count;
-
- if ((lowvalue << (offset - 1)) & 0x80000000)
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
- }
-
- w->buffer[w->pos++] = (lowvalue >> (24 - offset));
- lowvalue <<= offset;
- shift = count;
- lowvalue &= 0xffffff;
- count -= 8 ;
- }
-
- lowvalue <<= shift;
- }
- while (n);
- }
-
-
- {
-
- split = (range + 1) >> 1;
-
- if (e & 1)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- range <<= 1;
-
- if ((lowvalue & 0x80000000))
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
-
- }
-
- lowvalue <<= 1;
-
- if (!++count)
- {
- count = -8;
- w->buffer[w->pos++] = (lowvalue >> 24);
- lowvalue &= 0xffffff;
- }
- }
-
- }
-
- ++p;
- }
-
- w->count = count;
- w->lowvalue = lowvalue;
- w->range = range;
-
-}
-
-static void write_partition_size(unsigned char *cx_data, int size)
-{
- signed char csize;
-
- csize = size & 0xff;
- *cx_data = csize;
- csize = (size >> 8) & 0xff;
- *(cx_data + 1) = csize;
- csize = (size >> 16) & 0xff;
- *(cx_data + 2) = csize;
-
-}
-
-static void pack_tokens_into_partitions_c(VP8_COMP *cpi, unsigned char *cx_data, int num_part, int *size)
-{
-
- int i;
- unsigned char *ptr = cx_data;
- unsigned int shift;
- vp8_writer *w = &cpi->bc2;
- *size = 3 * (num_part - 1);
- ptr = cx_data + (*size);
-
- for (i = 0; i < num_part; i++)
- {
- vp8_start_encode(w, ptr);
- {
- unsigned int split;
- int count = w->count;
- unsigned int range = w->range;
- unsigned int lowvalue = w->lowvalue;
- int mb_row;
-
- for (mb_row = i; mb_row < cpi->common.mb_rows; mb_row += num_part)
- {
- TOKENEXTRA *p = cpi->tplist[mb_row].start;
- TOKENEXTRA *stop = cpi->tplist[mb_row].stop;
-
- while (p < stop)
- {
- const int t = p->Token;
- vp8_token *const a = vp8_coef_encodings + t;
- const vp8_extra_bit_struct *const b = vp8_extra_bits + t;
- int i = 0;
- const unsigned char *pp = p->context_tree;
- int v = a->value;
- int n = a->Len;
-
- if (p->skip_eob_node)
- {
- n--;
- i = 2;
- }
-
- do
- {
- const int bb = (v >> --n) & 1;
- split = 1 + (((range - 1) * pp[i>>1]) >> 8);
- i = vp8_coef_tree[i+bb];
-
- if (bb)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- shift = norm[range];
- range <<= shift;
- count += shift;
-
- if (count >= 0)
- {
- int offset = shift - count;
-
- if ((lowvalue << (offset - 1)) & 0x80000000)
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
- }
-
- w->buffer[w->pos++] = (lowvalue >> (24 - offset));
- lowvalue <<= offset;
- shift = count;
- lowvalue &= 0xffffff;
- count -= 8 ;
- }
-
- lowvalue <<= shift;
- }
- while (n);
-
-
- if (b->base_val)
- {
- const int e = p->Extra, L = b->Len;
-
- if (L)
- {
- const unsigned char *pp = b->prob;
- int v = e >> 1;
- int n = L; /* number of bits in v, assumed nonzero */
- int i = 0;
-
- do
- {
- const int bb = (v >> --n) & 1;
- split = 1 + (((range - 1) * pp[i>>1]) >> 8);
- i = b->tree[i+bb];
-
- if (bb)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- shift = norm[range];
- range <<= shift;
- count += shift;
-
- if (count >= 0)
- {
- int offset = shift - count;
-
- if ((lowvalue << (offset - 1)) & 0x80000000)
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
- }
-
- w->buffer[w->pos++] = (lowvalue >> (24 - offset));
- lowvalue <<= offset;
- shift = count;
- lowvalue &= 0xffffff;
- count -= 8 ;
- }
-
- lowvalue <<= shift;
- }
- while (n);
- }
-
- {
- split = (range + 1) >> 1;
-
- if (e & 1)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- range <<= 1;
-
- if ((lowvalue & 0x80000000))
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
-
- }
-
- lowvalue <<= 1;
-
- if (!++count)
- {
- count = -8;
- w->buffer[w->pos++] = (lowvalue >> 24);
- lowvalue &= 0xffffff;
- }
- }
-
- }
-
- ++p;
- }
- }
-
- w->count = count;
- w->lowvalue = lowvalue;
- w->range = range;
-
- }
-
- vp8_stop_encode(w);
- *size += w->pos;
-
- if (i < (num_part - 1))
- {
- write_partition_size(cx_data, w->pos);
- cx_data += 3;
- ptr += w->pos;
- }
- }
-}
-
-
-static void pack_mb_row_tokens_c(VP8_COMP *cpi, vp8_writer *w)
-{
-
- unsigned int split;
- int count = w->count;
- unsigned int range = w->range;
- unsigned int lowvalue = w->lowvalue;
- unsigned int shift;
- int mb_row;
-
- for (mb_row = 0; mb_row < cpi->common.mb_rows; mb_row++)
- {
- TOKENEXTRA *p = cpi->tplist[mb_row].start;
- TOKENEXTRA *stop = cpi->tplist[mb_row].stop;
-
- while (p < stop)
- {
- const int t = p->Token;
- vp8_token *const a = vp8_coef_encodings + t;
- const vp8_extra_bit_struct *const b = vp8_extra_bits + t;
- int i = 0;
- const unsigned char *pp = p->context_tree;
- int v = a->value;
- int n = a->Len;
-
- if (p->skip_eob_node)
- {
- n--;
- i = 2;
- }
-
- do
- {
- const int bb = (v >> --n) & 1;
- split = 1 + (((range - 1) * pp[i>>1]) >> 8);
- i = vp8_coef_tree[i+bb];
-
- if (bb)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- shift = norm[range];
- range <<= shift;
- count += shift;
-
- if (count >= 0)
- {
- int offset = shift - count;
-
- if ((lowvalue << (offset - 1)) & 0x80000000)
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
- }
-
- w->buffer[w->pos++] = (lowvalue >> (24 - offset));
- lowvalue <<= offset;
- shift = count;
- lowvalue &= 0xffffff;
- count -= 8 ;
- }
-
- lowvalue <<= shift;
- }
- while (n);
-
-
- if (b->base_val)
- {
- const int e = p->Extra, L = b->Len;
-
- if (L)
- {
- const unsigned char *pp = b->prob;
- int v = e >> 1;
- int n = L; /* number of bits in v, assumed nonzero */
- int i = 0;
-
- do
- {
- const int bb = (v >> --n) & 1;
- split = 1 + (((range - 1) * pp[i>>1]) >> 8);
- i = b->tree[i+bb];
-
- if (bb)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- shift = norm[range];
- range <<= shift;
- count += shift;
-
- if (count >= 0)
- {
- int offset = shift - count;
-
- if ((lowvalue << (offset - 1)) & 0x80000000)
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
- }
-
- w->buffer[w->pos++] = (lowvalue >> (24 - offset));
- lowvalue <<= offset;
- shift = count;
- lowvalue &= 0xffffff;
- count -= 8 ;
- }
-
- lowvalue <<= shift;
- }
- while (n);
- }
-
- {
- split = (range + 1) >> 1;
-
- if (e & 1)
- {
- lowvalue += split;
- range = range - split;
- }
- else
- {
- range = split;
- }
-
- range <<= 1;
-
- if ((lowvalue & 0x80000000))
- {
- int x = w->pos - 1;
-
- while (x >= 0 && w->buffer[x] == 0xff)
- {
- w->buffer[x] = (unsigned char)0;
- x--;
- }
-
- w->buffer[x] += 1;
-
- }
-
- lowvalue <<= 1;
-
- if (!++count)
- {
- count = -8;
- w->buffer[w->pos++] = (lowvalue >> 24);
- lowvalue &= 0xffffff;
- }
- }
-
- }
-
- ++p;
- }
- }
-
- w->count = count;
- w->lowvalue = lowvalue;
- w->range = range;
-
-}
-
-static void write_mv_ref
-(
- vp8_writer *w, MB_PREDICTION_MODE m, const vp8_prob *p
-)
-{
-
- assert(NEARESTMV <= m && m <= SPLITMV);
-
- vp8_write_token(w, vp8_mv_ref_tree, p,
- vp8_mv_ref_encoding_array - NEARESTMV + m);
-}
-
-static void write_sub_mv_ref
-(
- vp8_writer *w, B_PREDICTION_MODE m, const vp8_prob *p
-)
-{
- assert(LEFT4X4 <= m && m <= NEW4X4);
-
- vp8_write_token(w, vp8_sub_mv_ref_tree, p,
- vp8_sub_mv_ref_encoding_array - LEFT4X4 + m);
-}
-
-static void write_mv
-(
- vp8_writer *w, const MV *mv, const MV *ref, const MV_CONTEXT *mvc
-)
-{
- MV e;
- e.row = mv->row - ref->row;
- e.col = mv->col - ref->col;
-
- vp8_encode_motion_vector(w, &e, mvc);
-}
-
-static void write_mb_features(vp8_writer *w, const MB_MODE_INFO *mi, const MACROBLOCKD *x)
-{
- // Encode the MB segment id.
- if (x->segmentation_enabled && x->update_mb_segmentation_map)
- {
- switch (mi->segment_id)
- {
- case 0:
- vp8_write(w, 0, x->mb_segment_tree_probs[0]);
- vp8_write(w, 0, x->mb_segment_tree_probs[1]);
- break;
- case 1:
- vp8_write(w, 0, x->mb_segment_tree_probs[0]);
- vp8_write(w, 1, x->mb_segment_tree_probs[1]);
- break;
- case 2:
- vp8_write(w, 1, x->mb_segment_tree_probs[0]);
- vp8_write(w, 0, x->mb_segment_tree_probs[2]);
- break;
- case 3:
- vp8_write(w, 1, x->mb_segment_tree_probs[0]);
- vp8_write(w, 1, x->mb_segment_tree_probs[2]);
- break;
-
- // TRAP.. This should not happen
- default:
- vp8_write(w, 0, x->mb_segment_tree_probs[0]);
- vp8_write(w, 0, x->mb_segment_tree_probs[1]);
- break;
- }
- }
-}
-
-
-static void pack_inter_mode_mvs(VP8_COMP *const cpi)
-{
- VP8_COMMON *const pc = & cpi->common;
- vp8_writer *const w = & cpi->bc;
- const MV_CONTEXT *mvc = pc->fc.mvc;
-
- const int *const rfct = cpi->count_mb_ref_frame_usage;
- const int rf_intra = rfct[INTRA_FRAME];
- const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
-
- MODE_INFO *m = pc->mi, *ms;
- const int mis = pc->mode_info_stride;
- int mb_row = -1;
-
- int prob_last_coded;
- int prob_gf_coded;
- int prob_skip_false = 0;
- ms = pc->mi - 1;
-
- cpi->mb.partition_info = cpi->mb.pi;
-
- // Calculate the probabilities to be used to code the reference frame based on actual useage this frame
- if (!(cpi->prob_intra_coded = rf_intra * 255 / (rf_intra + rf_inter)))
- cpi->prob_intra_coded = 1;
-
- prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- if (!prob_last_coded)
- prob_last_coded = 1;
-
- prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- if (!prob_gf_coded)
- prob_gf_coded = 1;
-
-
-#ifdef ENTROPY_STATS
- active_section = 1;
-#endif
-
- if (pc->mb_no_coeff_skip)
- {
- prob_skip_false = cpi->skip_false_count * 256 / (cpi->skip_false_count + cpi->skip_true_count);
-
- if (prob_skip_false <= 1)
- prob_skip_false = 1;
-
- if (prob_skip_false > 255)
- prob_skip_false = 255;
-
- cpi->prob_skip_false = prob_skip_false;
- vp8_write_literal(w, prob_skip_false, 8);
- }
-
- vp8_write_literal(w, cpi->prob_intra_coded, 8);
- vp8_write_literal(w, prob_last_coded, 8);
- vp8_write_literal(w, prob_gf_coded, 8);
-
- update_mbintra_mode_probs(cpi);
-
- vp8_write_mvprobs(cpi);
-
- while (++mb_row < pc->mb_rows)
- {
- int mb_col = -1;
-
- while (++mb_col < pc->mb_cols)
- {
- const MB_MODE_INFO *const mi = & m->mbmi;
- const MV_REFERENCE_FRAME rf = mi->ref_frame;
- const MB_PREDICTION_MODE mode = mi->mode;
-
- MACROBLOCKD *xd = &cpi->mb.e_mbd;
-
- // Distance of Mb to the various image edges.
- // These specified to 8th pel as they are always compared to MV values that are in 1/8th pel units
- xd->mb_to_left_edge = -((mb_col * 16) << 3);
- xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3;
- xd->mb_to_top_edge = -((mb_row * 16)) << 3;
- xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3;
-
-#ifdef ENTROPY_STATS
- active_section = 9;
-#endif
-
- if (cpi->mb.e_mbd.update_mb_segmentation_map)
- write_mb_features(w, mi, &cpi->mb.e_mbd);
-
- if (pc->mb_no_coeff_skip)
- vp8_encode_bool(w, m->mbmi.mb_skip_coeff, prob_skip_false);
-
- if (rf == INTRA_FRAME)
- {
- vp8_write(w, 0, cpi->prob_intra_coded);
-#ifdef ENTROPY_STATS
- active_section = 6;
-#endif
- write_ymode(w, mode, pc->fc.ymode_prob);
-
- if (mode == B_PRED)
- {
- int j = 0;
-
- do
- write_bmode(w, m->bmi[j].mode, pc->fc.bmode_prob);
-
- while (++j < 16);
- }
-
- write_uv_mode(w, mi->uv_mode, pc->fc.uv_mode_prob);
- }
- else /* inter coded */
- {
- MV best_mv;
- vp8_prob mv_ref_p [VP8_MVREFS-1];
-
- vp8_write(w, 1, cpi->prob_intra_coded);
-
- if (rf == LAST_FRAME)
- vp8_write(w, 0, prob_last_coded);
- else
- {
- vp8_write(w, 1, prob_last_coded);
- vp8_write(w, (rf == GOLDEN_FRAME) ? 0 : 1, prob_gf_coded);
- }
-
- {
- MV n1, n2;
- int ct[4];
-
- vp8_find_near_mvs(xd, m, &n1, &n2, &best_mv, ct, rf, cpi->common.ref_frame_sign_bias);
- vp8_mv_ref_probs(mv_ref_p, ct);
-
-#ifdef ENTROPY_STATS
- accum_mv_refs(mode, ct);
-#endif
-
- }
-
-#ifdef ENTROPY_STATS
- active_section = 3;
-#endif
-
- write_mv_ref(w, mode, mv_ref_p);
-
- switch (mode) /* new, split require MVs */
- {
- case NEWMV:
-
-#ifdef ENTROPY_STATS
- active_section = 5;
-#endif
-
- write_mv(w, &mi->mv.as_mv, &best_mv, mvc);
- break;
-
- case SPLITMV:
- {
- int j = 0;
-
-#ifdef MODE_STATS
- ++count_mb_seg [mi->partitioning];
-#endif
-
- write_split(w, mi->partitioning);
-
- do
- {
- const B_MODE_INFO *const b = cpi->mb.partition_info->bmi + j;
- const int *const L = vp8_mbsplits [mi->partitioning];
- int k = -1; /* first block in subset j */
- int mv_contz;
-
- while (j != L[++k])
- if (k >= 16)
- assert(0);
-
- mv_contz = vp8_mv_cont
- (&(vp8_left_bmi(m, k)->mv.as_mv),
- &(vp8_above_bmi(m, k, mis)->mv.as_mv));
- write_sub_mv_ref(w, b->mode, vp8_sub_mv_ref_prob2 [mv_contz]); //pc->fc.sub_mv_ref_prob);
-
- if (b->mode == NEW4X4)
- {
-#ifdef ENTROPY_STATS
- active_section = 11;
-#endif
- write_mv(w, &b->mv.as_mv, &best_mv, (const MV_CONTEXT *) mvc);
- }
- }
- while (++j < cpi->mb.partition_info->count);
- }
- break;
- default:
- break;
- }
- }
-
- ++m;
- cpi->mb.partition_info++;
- }
-
- ++m; /* skip L prediction border */
- cpi->mb.partition_info++;
- }
-}
-
-
-static void write_kfmodes(VP8_COMP *cpi)
-{
- vp8_writer *const bc = & cpi->bc;
- const VP8_COMMON *const c = & cpi->common;
- /* const */
- MODE_INFO *m = c->mi;
-
- int mb_row = -1;
- int prob_skip_false = 0;
-
- if (c->mb_no_coeff_skip)
- {
- prob_skip_false = cpi->skip_false_count * 256 / (cpi->skip_false_count + cpi->skip_true_count);
-
- if (prob_skip_false <= 1)
- prob_skip_false = 1;
-
- if (prob_skip_false >= 255)
- prob_skip_false = 255;
-
- cpi->prob_skip_false = prob_skip_false;
- vp8_write_literal(bc, prob_skip_false, 8);
- }
-
- while (++mb_row < c->mb_rows)
- {
- int mb_col = -1;
-
- while (++mb_col < c->mb_cols)
- {
- const int ym = m->mbmi.mode;
-
- if (cpi->mb.e_mbd.update_mb_segmentation_map)
- write_mb_features(bc, &m->mbmi, &cpi->mb.e_mbd);
-
- if (c->mb_no_coeff_skip)
- vp8_encode_bool(bc, m->mbmi.mb_skip_coeff, prob_skip_false);
-
- kfwrite_ymode(bc, ym, c->kf_ymode_prob);
-
- if (ym == B_PRED)
- {
- const int mis = c->mode_info_stride;
- int i = 0;
-
- do
- {
- const B_PREDICTION_MODE A = vp8_above_bmi(m, i, mis)->mode;
- const B_PREDICTION_MODE L = vp8_left_bmi(m, i)->mode;
- const int bm = m->bmi[i].mode;
-
-#ifdef ENTROPY_STATS
- ++intra_mode_stats [A] [L] [bm];
-#endif
-
- write_bmode(bc, bm, c->kf_bmode_prob [A] [L]);
- }
- while (++i < 16);
- }
-
- write_uv_mode(bc, (m++)->mbmi.uv_mode, c->kf_uv_mode_prob);
- }
-
- m++; // skip L prediction border
- }
-}
-int vp8_estimate_entropy_savings(VP8_COMP *cpi)
-{
- int i = 0;
- int savings = 0;
-
- const int *const rfct = cpi->count_mb_ref_frame_usage;
- const int rf_intra = rfct[INTRA_FRAME];
- const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
- int new_intra, new_last, gf_last, oldtotal, newtotal;
- int ref_frame_cost[MAX_REF_FRAMES];
-
- vp8_clear_system_state(); //__asm emms;
-
- if (cpi->common.frame_type != KEY_FRAME)
- {
- if (!(new_intra = rf_intra * 255 / (rf_intra + rf_inter)))
- new_intra = 1;
-
- new_last = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- gf_last = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- // new costs
- ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(new_intra);
- ref_frame_cost[LAST_FRAME] = vp8_cost_one(new_intra)
- + vp8_cost_zero(new_last);
- ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(new_intra)
- + vp8_cost_one(new_last)
- + vp8_cost_zero(gf_last);
- ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(new_intra)
- + vp8_cost_one(new_last)
- + vp8_cost_one(gf_last);
-
- newtotal =
- rfct[INTRA_FRAME] * ref_frame_cost[INTRA_FRAME] +
- rfct[LAST_FRAME] * ref_frame_cost[LAST_FRAME] +
- rfct[GOLDEN_FRAME] * ref_frame_cost[GOLDEN_FRAME] +
- rfct[ALTREF_FRAME] * ref_frame_cost[ALTREF_FRAME];
-
-
- // old costs
- ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(cpi->prob_intra_coded);
- ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(cpi->prob_last_coded);
- ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_zero(cpi->prob_gf_coded);
- ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_one(cpi->prob_gf_coded);
-
- oldtotal =
- rfct[INTRA_FRAME] * ref_frame_cost[INTRA_FRAME] +
- rfct[LAST_FRAME] * ref_frame_cost[LAST_FRAME] +
- rfct[GOLDEN_FRAME] * ref_frame_cost[GOLDEN_FRAME] +
- rfct[ALTREF_FRAME] * ref_frame_cost[ALTREF_FRAME];
-
- savings += (oldtotal - newtotal) / 256;
- }
-
-
- do
- {
- int j = 0;
-
- do
- {
- int k = 0;
-
- do
- {
- /* at every context */
-
- /* calc probs and branch cts for this frame only */
- //vp8_prob new_p [vp8_coef_tokens-1];
- //unsigned int branch_ct [vp8_coef_tokens-1] [2];
-
- int t = 0; /* token/prob index */
-
- vp8_tree_probs_from_distribution(
- vp8_coef_tokens, vp8_coef_encodings, vp8_coef_tree,
- cpi->frame_coef_probs [i][j][k], cpi->frame_branch_ct [i][j][k], cpi->coef_counts [i][j][k],
- 256, 1
- );
-
- do
- {
- const unsigned int *ct = cpi->frame_branch_ct [i][j][k][t];
- const vp8_prob newp = cpi->frame_coef_probs [i][j][k][t];
-
- const vp8_prob old = cpi->common.fc.coef_probs [i][j][k][t];
- const vp8_prob upd = vp8_coef_update_probs [i][j][k][t];
-
- const int old_b = vp8_cost_branch(ct, old);
- const int new_b = vp8_cost_branch(ct, newp);
-
- const int update_b = 8 +
- ((vp8_cost_one(upd) - vp8_cost_zero(upd)) >> 8);
-
- const int s = old_b - new_b - update_b;
-
- if (s > 0)
- savings += s;
-
-
- }
- while (++t < vp8_coef_tokens - 1);
-
-
- }
- while (++k < PREV_COEF_CONTEXTS);
- }
- while (++j < COEF_BANDS);
- }
- while (++i < BLOCK_TYPES);
-
- return savings;
-}
-
-static void update_coef_probs(VP8_COMP *cpi)
-{
- int i = 0;
- vp8_writer *const w = & cpi->bc;
- int savings = 0;
-
- vp8_clear_system_state(); //__asm emms;
-
-
- do
- {
- int j = 0;
-
- do
- {
- int k = 0;
-
- do
- {
- //note: use result from vp8_estimate_entropy_savings, so no need to call vp8_tree_probs_from_distribution here.
- /* at every context */
-
- /* calc probs and branch cts for this frame only */
- //vp8_prob new_p [vp8_coef_tokens-1];
- //unsigned int branch_ct [vp8_coef_tokens-1] [2];
-
- int t = 0; /* token/prob index */
-
- //vp8_tree_probs_from_distribution(
- // vp8_coef_tokens, vp8_coef_encodings, vp8_coef_tree,
- // new_p, branch_ct, (unsigned int *)cpi->coef_counts [i][j][k],
- // 256, 1
- // );
-
- do
- {
- const unsigned int *ct = cpi->frame_branch_ct [i][j][k][t];
- const vp8_prob newp = cpi->frame_coef_probs [i][j][k][t];
-
- vp8_prob *Pold = cpi->common.fc.coef_probs [i][j][k] + t;
- const vp8_prob old = *Pold;
- const vp8_prob upd = vp8_coef_update_probs [i][j][k][t];
-
- const int old_b = vp8_cost_branch(ct, old);
- const int new_b = vp8_cost_branch(ct, newp);
-
- const int update_b = 8 +
- ((vp8_cost_one(upd) - vp8_cost_zero(upd)) >> 8);
-
- const int s = old_b - new_b - update_b;
- const int u = s > 0 ? 1 : 0;
-
- vp8_write(w, u, upd);
-
-
-#ifdef ENTROPY_STATS
- ++ tree_update_hist [i][j][k][t] [u];
-#endif
-
- if (u)
- {
- /* send/use new probability */
-
- *Pold = newp;
- vp8_write_literal(w, newp, 8);
-
- savings += s;
-
- }
-
- }
- while (++t < vp8_coef_tokens - 1);
-
- /* Accum token counts for generation of default statistics */
-#ifdef ENTROPY_STATS
- t = 0;
-
- do
- {
- context_counters [i][j][k][t] += cpi->coef_counts [i][j][k][t];
- }
- while (++t < vp8_coef_tokens);
-
-#endif
-
- }
- while (++k < PREV_COEF_CONTEXTS);
- }
- while (++j < COEF_BANDS);
- }
- while (++i < BLOCK_TYPES);
-
-}
-#ifdef PACKET_TESTING
-FILE *vpxlogc = 0;
-#endif
-
-static void put_delta_q(vp8_writer *bc, int delta_q)
-{
- if (delta_q != 0)
- {
- vp8_write_bit(bc, 1);
- vp8_write_literal(bc, abs(delta_q), 4);
-
- if (delta_q < 0)
- vp8_write_bit(bc, 1);
- else
- vp8_write_bit(bc, 0);
- }
- else
- vp8_write_bit(bc, 0);
-}
-
-void vp8_pack_bitstream(VP8_COMP *cpi, unsigned char *dest, unsigned long *size)
-{
- int i, j;
- VP8_HEADER oh;
- VP8_COMMON *const pc = & cpi->common;
- vp8_writer *const bc = & cpi->bc;
- MACROBLOCKD *const xd = & cpi->mb.e_mbd;
- int extra_bytes_packed = 0;
-
- unsigned char *cx_data = dest;
- const int *mb_feature_data_bits;
-
- oh.show_frame = (int) pc->show_frame;
- oh.type = (int)pc->frame_type;
- oh.version = pc->version;
- oh.first_partition_length_in_bytes = 0;
-
- mb_feature_data_bits = vp8_mb_feature_data_bits;
- cx_data += 3;
-
-#if defined(SECTIONBITS_OUTPUT)
- Sectionbits[active_section = 1] += sizeof(VP8_HEADER) * 8 * 256;
-#endif
-
- //vp8_kf_default_bmode_probs() is called in vp8_setup_key_frame() once for each
- //K frame before encode frame. pc->kf_bmode_prob doesn't get changed anywhere
- //else. No need to call it again here. --yw
- //vp8_kf_default_bmode_probs( pc->kf_bmode_prob);
-
- // every keyframe send startcode, width, height, scale factor, clamp and color type
- if (oh.type == KEY_FRAME)
- {
- int v;
-
- // Start / synch code
- cx_data[0] = 0x9D;
- cx_data[1] = 0x01;
- cx_data[2] = 0x2a;
-
- v = (pc->horiz_scale << 14) | pc->Width;
- cx_data[3] = v;
- cx_data[4] = v >> 8;
-
- v = (pc->vert_scale << 14) | pc->Height;
- cx_data[5] = v;
- cx_data[6] = v >> 8;
-
- extra_bytes_packed = 7;
- cx_data += extra_bytes_packed ;
-
- vp8_start_encode(bc, cx_data);
-
- // signal clr type
- vp8_write_bit(bc, pc->clr_type);
- vp8_write_bit(bc, pc->clamp_type);
-
- }
- else
- vp8_start_encode(bc, cx_data);
-
-
- // Signal whether or not Segmentation is enabled
- vp8_write_bit(bc, (xd->segmentation_enabled) ? 1 : 0);
-
- // Indicate which features are enabled
- if (xd->segmentation_enabled)
- {
- // Signal whether or not the segmentation map is being updated.
- vp8_write_bit(bc, (xd->update_mb_segmentation_map) ? 1 : 0);
- vp8_write_bit(bc, (xd->update_mb_segmentation_data) ? 1 : 0);
-
- if (xd->update_mb_segmentation_data)
- {
- signed char Data;
-
- vp8_write_bit(bc, (xd->mb_segement_abs_delta) ? 1 : 0);
-
- // For each segmentation feature (Quant and loop filter level)
- for (i = 0; i < MB_LVL_MAX; i++)
- {
- // For each of the segments
- for (j = 0; j < MAX_MB_SEGMENTS; j++)
- {
- Data = xd->segment_feature_data[i][j];
-
- // Frame level data
- if (Data)
- {
- vp8_write_bit(bc, 1);
-
- if (Data < 0)
- {
- Data = - Data;
- vp8_write_literal(bc, Data, mb_feature_data_bits[i]);
- vp8_write_bit(bc, 1);
- }
- else
- {
- vp8_write_literal(bc, Data, mb_feature_data_bits[i]);
- vp8_write_bit(bc, 0);
- }
- }
- else
- vp8_write_bit(bc, 0);
- }
- }
- }
-
- if (xd->update_mb_segmentation_map)
- {
- // Write the probs used to decode the segment id for each macro block.
- for (i = 0; i < MB_FEATURE_TREE_PROBS; i++)
- {
- int Data = xd->mb_segment_tree_probs[i];
-
- if (Data != 255)
- {
- vp8_write_bit(bc, 1);
- vp8_write_literal(bc, Data, 8);
- }
- else
- vp8_write_bit(bc, 0);
- }
- }
- }
-
- // Code to determine whether or not to update the scan order.
- vp8_write_bit(bc, pc->filter_type);
- vp8_write_literal(bc, pc->filter_level, 6);
- vp8_write_literal(bc, pc->sharpness_level, 3);
-
- // Write out loop filter deltas applied at the MB level based on mode or ref frame (if they are enabled).
- vp8_write_bit(bc, (xd->mode_ref_lf_delta_enabled) ? 1 : 0);
-
- if (xd->mode_ref_lf_delta_enabled)
- {
- // Do the deltas need to be updated
- int send_update = xd->mode_ref_lf_delta_update
- || cpi->oxcf.error_resilient_mode;
-
- vp8_write_bit(bc, send_update);
- if (send_update)
- {
- int Data;
-
- // Send update
- for (i = 0; i < MAX_REF_LF_DELTAS; i++)
- {
- Data = xd->ref_lf_deltas[i];
-
- // Frame level data
- if (xd->ref_lf_deltas[i] != xd->last_ref_lf_deltas[i]
- || cpi->oxcf.error_resilient_mode)
- {
- xd->last_ref_lf_deltas[i] = xd->ref_lf_deltas[i];
- vp8_write_bit(bc, 1);
-
- if (Data > 0)
- {
- vp8_write_literal(bc, (Data & 0x3F), 6);
- vp8_write_bit(bc, 0); // sign
- }
- else
- {
- Data = -Data;
- vp8_write_literal(bc, (Data & 0x3F), 6);
- vp8_write_bit(bc, 1); // sign
- }
- }
- else
- vp8_write_bit(bc, 0);
- }
-
- // Send update
- for (i = 0; i < MAX_MODE_LF_DELTAS; i++)
- {
- Data = xd->mode_lf_deltas[i];
-
- if (xd->mode_lf_deltas[i] != xd->last_mode_lf_deltas[i]
- || cpi->oxcf.error_resilient_mode)
- {
- xd->last_mode_lf_deltas[i] = xd->mode_lf_deltas[i];
- vp8_write_bit(bc, 1);
-
- if (Data > 0)
- {
- vp8_write_literal(bc, (Data & 0x3F), 6);
- vp8_write_bit(bc, 0); // sign
- }
- else
- {
- Data = -Data;
- vp8_write_literal(bc, (Data & 0x3F), 6);
- vp8_write_bit(bc, 1); // sign
- }
- }
- else
- vp8_write_bit(bc, 0);
- }
- }
- }
-
- //signal here is multi token partition is enabled
- vp8_write_literal(bc, pc->multi_token_partition, 2);
-
- // Frame Qbaseline quantizer index
- vp8_write_literal(bc, pc->base_qindex, 7);
-
- // Transmit Dc, Second order and Uv quantizer delta information
- put_delta_q(bc, pc->y1dc_delta_q);
- put_delta_q(bc, pc->y2dc_delta_q);
- put_delta_q(bc, pc->y2ac_delta_q);
- put_delta_q(bc, pc->uvdc_delta_q);
- put_delta_q(bc, pc->uvac_delta_q);
-
- // When there is a key frame all reference buffers are updated using the new key frame
- if (pc->frame_type != KEY_FRAME)
- {
- // Should the GF or ARF be updated using the transmitted frame or buffer
- vp8_write_bit(bc, pc->refresh_golden_frame);
- vp8_write_bit(bc, pc->refresh_alt_ref_frame);
-
- // If not being updated from current frame should either GF or ARF be updated from another buffer
- if (!pc->refresh_golden_frame)
- vp8_write_literal(bc, pc->copy_buffer_to_gf, 2);
-
- if (!pc->refresh_alt_ref_frame)
- vp8_write_literal(bc, pc->copy_buffer_to_arf, 2);
-
- // Indicate reference frame sign bias for Golden and ARF frames (always 0 for last frame buffer)
- vp8_write_bit(bc, pc->ref_frame_sign_bias[GOLDEN_FRAME]);
- vp8_write_bit(bc, pc->ref_frame_sign_bias[ALTREF_FRAME]);
- }
-
- vp8_write_bit(bc, pc->refresh_entropy_probs);
-
- if (pc->frame_type != KEY_FRAME)
- vp8_write_bit(bc, pc->refresh_last_frame);
-
-#ifdef ENTROPY_STATS
-
- if (pc->frame_type == INTER_FRAME)
- active_section = 0;
- else
- active_section = 7;
-
-#endif
-
- vp8_clear_system_state(); //__asm emms;
-
- //************************************************
- // save a copy for later refresh
- {
- vpx_memcpy(&cpi->common.lfc, &cpi->common.fc, sizeof(cpi->common.fc));
- }
-
- update_coef_probs(cpi);
-
-#ifdef ENTROPY_STATS
- active_section = 2;
-#endif
-
- // Write out the mb_no_coeff_skip flag
- vp8_write_bit(bc, pc->mb_no_coeff_skip);
-
- if (pc->frame_type == KEY_FRAME)
- {
- write_kfmodes(cpi);
-
-#ifdef ENTROPY_STATS
- active_section = 8;
-#endif
- }
- else
- {
- pack_inter_mode_mvs(cpi);
-
-#ifdef ENTROPY_STATS
- active_section = 1;
-#endif
- }
-
- vp8_stop_encode(bc);
-
- oh.first_partition_length_in_bytes = cpi->bc.pos;
-
- /* update frame tag */
- {
- int v = (oh.first_partition_length_in_bytes << 5) |
- (oh.show_frame << 4) |
- (oh.version << 1) |
- oh.type;
-
- dest[0] = v;
- dest[1] = v >> 8;
- dest[2] = v >> 16;
- }
-
- *size = VP8_HEADER_SIZE + extra_bytes_packed + cpi->bc.pos;
-
- if (pc->multi_token_partition != ONE_PARTITION)
- {
- int num_part;
- int asize;
- num_part = 1 << pc->multi_token_partition;
-
- pack_tokens_into_partitions(cpi, cx_data + bc->pos, num_part, &asize);
-
- *size += asize;
- }
- else
- {
- vp8_start_encode(&cpi->bc2, cx_data + bc->pos);
-
-#if CONFIG_MULTITHREAD
- if (cpi->b_multi_threaded)
- pack_mb_row_tokens(cpi, &cpi->bc2);
- else
-#endif
- pack_tokens(&cpi->bc2, cpi->tok, cpi->tok_count);
-
- vp8_stop_encode(&cpi->bc2);
-
- *size += cpi->bc2.pos;
- }
-}
-
-#ifdef ENTROPY_STATS
-void print_tree_update_probs()
-{
- int i, j, k, l;
- FILE *f = fopen("context.c", "a");
- int Sum;
- fprintf(f, "\n/* Update probabilities for token entropy tree. */\n\n");
- fprintf(f, "const vp8_prob tree_update_probs[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1] = {\n");
-
- for (i = 0; i < BLOCK_TYPES; i++)
- {
- fprintf(f, " { \n");
-
- for (j = 0; j < COEF_BANDS; j++)
- {
- fprintf(f, " {\n");
-
- for (k = 0; k < PREV_COEF_CONTEXTS; k++)
- {
- fprintf(f, " {");
-
- for (l = 0; l < MAX_ENTROPY_TOKENS - 1; l++)
- {
- Sum = tree_update_hist[i][j][k][l][0] + tree_update_hist[i][j][k][l][1];
-
- if (Sum > 0)
- {
- if (((tree_update_hist[i][j][k][l][0] * 255) / Sum) > 0)
- fprintf(f, "%3ld, ", (tree_update_hist[i][j][k][l][0] * 255) / Sum);
- else
- fprintf(f, "%3ld, ", 1);
- }
- else
- fprintf(f, "%3ld, ", 128);
- }
-
- fprintf(f, "},\n");
- }
-
- fprintf(f, " },\n");
- }
-
- fprintf(f, " },\n");
- }
-
- fprintf(f, "};\n");
- fclose(f);
-}
-#endif
diff --git a/vp8/encoder/bitstream.h b/vp8/encoder/bitstream.h
deleted file mode 100644
index f5d148e..0000000
--- a/vp8/encoder/bitstream.h
+++ /dev/null
@@ -1,39 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_BITSTREAM_H
-#define __INC_BITSTREAM_H
-
-#if HAVE_ARMV5TE
-void vp8cx_pack_tokens_armv5(vp8_writer *w, const TOKENEXTRA *p, int xcount,
- vp8_token *,
- vp8_extra_bit_struct *,
- const vp8_tree_index *);
-void vp8cx_pack_tokens_into_partitions_armv5(VP8_COMP *, unsigned char *, int , int *,
- vp8_token *,
- vp8_extra_bit_struct *,
- const vp8_tree_index *);
-void vp8cx_pack_mb_row_tokens_armv5(VP8_COMP *cpi, vp8_writer *w,
- vp8_token *,
- vp8_extra_bit_struct *,
- const vp8_tree_index *);
-# define pack_tokens(a,b,c) \
- vp8cx_pack_tokens_armv5(a,b,c,vp8_coef_encodings,vp8_extra_bits,vp8_coef_tree)
-# define pack_tokens_into_partitions(a,b,c,d) \
- vp8cx_pack_tokens_into_partitions_armv5(a,b,c,d,vp8_coef_encodings,vp8_extra_bits,vp8_coef_tree)
-# define pack_mb_row_tokens(a,b) \
- vp8cx_pack_mb_row_tokens_armv5(a,b,vp8_coef_encodings,vp8_extra_bits,vp8_coef_tree)
-#else
-# define pack_tokens(a,b,c) pack_tokens_c(a,b,c)
-# define pack_tokens_into_partitions(a,b,c,d) pack_tokens_into_partitions_c(a,b,c,d)
-# define pack_mb_row_tokens(a,b) pack_mb_row_tokens_c(a,b)
-#endif
-#endif
diff --git a/vp8/encoder/block.h b/vp8/encoder/block.h
deleted file mode 100644
index 2fd6782..0000000
--- a/vp8/encoder/block.h
+++ /dev/null
@@ -1,125 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_BLOCK_H
-#define __INC_BLOCK_H
-
-#include "vp8/common/onyx.h"
-#include "vp8/common/blockd.h"
-#include "vp8/common/entropymv.h"
-#include "vp8/common/entropy.h"
-#include "vpx_ports/mem.h"
-
-// motion search site
-typedef struct
-{
- MV mv;
- int offset;
-} search_site;
-
-typedef struct
-{
- // 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries
- short *src_diff;
- short *coeff;
-
- // 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries
- short *quant;
- short *quant_fast;
- short *quant_shift;
- short *zbin;
- short *zrun_zbin_boost;
- short *round;
-
- // Zbin Over Quant value
- short zbin_extra;
-
- unsigned char **base_src;
- int src;
- int src_stride;
-
-// MV enc_mv;
- int force_empty;
-
-} BLOCK;
-
-typedef struct
-{
- int count;
- B_MODE_INFO bmi[16];
-} PARTITION_INFO;
-
-typedef struct
-{
- DECLARE_ALIGNED(16, short, src_diff[400]); // 16x16 Y 8x8 U 8x8 V 4x4 2nd Y
- DECLARE_ALIGNED(16, short, coeff[400]); // 16x16 Y 8x8 U 8x8 V 4x4 2nd Y
-
- // 16 Y blocks, 4 U blocks, 4 V blocks, 1 DC 2nd order block each with 16 entries
- BLOCK block[25];
-
- YV12_BUFFER_CONFIG src;
-
- MACROBLOCKD e_mbd;
- PARTITION_INFO *partition_info; /* work pointer */
- PARTITION_INFO *pi; /* Corresponds to upper left visible macroblock */
- PARTITION_INFO *pip; /* Base of allocated array */
-
- search_site *ss;
- int ss_count;
- int searches_per_step;
-
- int errorperbit;
- int sadperbit16;
- int sadperbit4;
- int errthresh;
- int rddiv;
- int rdmult;
- INT64 activity_sum;
-
- int mvcosts[2][MVvals+1];
- int *mvcost[2];
- int mvsadcosts[2][MVvals+1];
- int *mvsadcost[2];
- int mbmode_cost[2][MB_MODE_COUNT];
- int intra_uv_mode_cost[2][MB_MODE_COUNT];
- unsigned int bmode_costs[10][10][10];
- unsigned int inter_bmode_costs[B_MODE_COUNT];
-
- // These define limits to motion vector components to prevent them from extending outside the UMV borders
- int mv_col_min;
- int mv_col_max;
- int mv_row_min;
- int mv_row_max;
-
- int vector_range; // Used to monitor limiting range of recent vectors to guide search.
- int skip;
-
- int encode_breakout;
-
- //char * gf_active_ptr;
- signed char *gf_active_ptr;
-
- unsigned char *active_ptr;
- MV_CONTEXT *mvc;
-
- unsigned int token_costs[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens];
- int optimize;
- int q_index;
-
- void (*vp8_short_fdct4x4)(short *input, short *output, int pitch);
- void (*vp8_short_fdct8x4)(short *input, short *output, int pitch);
- void (*short_walsh4x4)(short *input, short *output, int pitch);
- void (*quantize_b)(BLOCK *b, BLOCKD *d);
-
-} MACROBLOCK;
-
-
-#endif
diff --git a/vp8/encoder/boolhuff.c b/vp8/encoder/boolhuff.c
deleted file mode 100644
index 788d2b0..0000000
--- a/vp8/encoder/boolhuff.c
+++ /dev/null
@@ -1,148 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "boolhuff.h"
-#include "vp8/common/blockd.h"
-
-
-
-#if defined(SECTIONBITS_OUTPUT)
-unsigned __int64 Sectionbits[500];
-
-#endif
-
-#ifdef ENTROPY_STATS
-unsigned int active_section = 0;
-#endif
-
-const unsigned int vp8_prob_cost[256] =
-{
- 2047, 2047, 1791, 1641, 1535, 1452, 1385, 1328, 1279, 1235, 1196, 1161, 1129, 1099, 1072, 1046,
- 1023, 1000, 979, 959, 940, 922, 905, 889, 873, 858, 843, 829, 816, 803, 790, 778,
- 767, 755, 744, 733, 723, 713, 703, 693, 684, 675, 666, 657, 649, 641, 633, 625,
- 617, 609, 602, 594, 587, 580, 573, 567, 560, 553, 547, 541, 534, 528, 522, 516,
- 511, 505, 499, 494, 488, 483, 477, 472, 467, 462, 457, 452, 447, 442, 437, 433,
- 428, 424, 419, 415, 410, 406, 401, 397, 393, 389, 385, 381, 377, 373, 369, 365,
- 361, 357, 353, 349, 346, 342, 338, 335, 331, 328, 324, 321, 317, 314, 311, 307,
- 304, 301, 297, 294, 291, 288, 285, 281, 278, 275, 272, 269, 266, 263, 260, 257,
- 255, 252, 249, 246, 243, 240, 238, 235, 232, 229, 227, 224, 221, 219, 216, 214,
- 211, 208, 206, 203, 201, 198, 196, 194, 191, 189, 186, 184, 181, 179, 177, 174,
- 172, 170, 168, 165, 163, 161, 159, 156, 154, 152, 150, 148, 145, 143, 141, 139,
- 137, 135, 133, 131, 129, 127, 125, 123, 121, 119, 117, 115, 113, 111, 109, 107,
- 105, 103, 101, 99, 97, 95, 93, 92, 90, 88, 86, 84, 82, 81, 79, 77,
- 75, 73, 72, 70, 68, 66, 65, 63, 61, 60, 58, 56, 55, 53, 51, 50,
- 48, 46, 45, 43, 41, 40, 38, 37, 35, 33, 32, 30, 29, 27, 25, 24,
- 22, 21, 19, 18, 16, 15, 13, 12, 10, 9, 7, 6, 4, 3, 1, 1
-};
-
-void vp8_start_encode(BOOL_CODER *br, unsigned char *source)
-{
-
- br->lowvalue = 0;
- br->range = 255;
- br->value = 0;
- br->count = -24;
- br->buffer = source;
- br->pos = 0;
-}
-
-void vp8_stop_encode(BOOL_CODER *br)
-{
- int i;
-
- for (i = 0; i < 32; i++)
- vp8_encode_bool(br, 0, 128);
-}
-
-DECLARE_ALIGNED(16, static const unsigned int, norm[256]) =
-{
- 0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
- 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
- 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
-};
-
-void vp8_encode_bool(BOOL_CODER *br, int bit, int probability)
-{
- unsigned int split;
- int count = br->count;
- unsigned int range = br->range;
- unsigned int lowvalue = br->lowvalue;
- register unsigned int shift;
-
-#ifdef ENTROPY_STATS
-#if defined(SECTIONBITS_OUTPUT)
-
- if (bit)
- Sectionbits[active_section] += vp8_prob_cost[255-probability];
- else
- Sectionbits[active_section] += vp8_prob_cost[probability];
-
-#endif
-#endif
-
- split = 1 + (((range - 1) * probability) >> 8);
-
- range = split;
-
- if (bit)
- {
- lowvalue += split;
- range = br->range - split;
- }
-
- shift = norm[range];
-
- range <<= shift;
- count += shift;
-
- if (count >= 0)
- {
- int offset = shift - count;
-
- if ((lowvalue << (offset - 1)) & 0x80000000)
- {
- int x = br->pos - 1;
-
- while (x >= 0 && br->buffer[x] == 0xff)
- {
- br->buffer[x] = (unsigned char)0;
- x--;
- }
-
- br->buffer[x] += 1;
- }
-
- br->buffer[br->pos++] = (lowvalue >> (24 - offset));
- lowvalue <<= offset;
- shift = count;
- lowvalue &= 0xffffff;
- count -= 8 ;
- }
-
- lowvalue <<= shift;
- br->count = count;
- br->lowvalue = lowvalue;
- br->range = range;
-}
-
-void vp8_encode_value(BOOL_CODER *br, int data, int bits)
-{
- int bit;
-
- for (bit = bits - 1; bit >= 0; bit--)
- vp8_encode_bool(br, (1 & (data >> bit)), 0x80);
-
-}
diff --git a/vp8/encoder/boolhuff.h b/vp8/encoder/boolhuff.h
deleted file mode 100644
index f723da3..0000000
--- a/vp8/encoder/boolhuff.h
+++ /dev/null
@@ -1,43 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/****************************************************************************
-*
-* Module Title : boolhuff.h
-*
-* Description : Bool Coder header file.
-*
-****************************************************************************/
-#ifndef __INC_BOOLHUFF_H
-#define __INC_BOOLHUFF_H
-
-
-typedef struct
-{
- unsigned int lowvalue;
- unsigned int range;
- unsigned int value;
- int count;
- unsigned int pos;
- unsigned char *buffer;
-
- // Variables used to track bit costs without outputing to the bitstream
- unsigned int measure_cost;
- unsigned long bit_counter;
-} BOOL_CODER;
-
-extern void vp8_start_encode(BOOL_CODER *bc, unsigned char *buffer);
-extern void vp8_encode_bool(BOOL_CODER *bc, int x, int context);
-extern void vp8_encode_value(BOOL_CODER *br, int data, int bits);
-extern void vp8_stop_encode(BOOL_CODER *bc);
-extern const unsigned int vp8_prob_cost[256];
-
-#endif
diff --git a/vp8/encoder/dct.h b/vp8/encoder/dct.h
deleted file mode 100644
index fec3b4c..0000000
--- a/vp8/encoder/dct.h
+++ /dev/null
@@ -1,65 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_DCT_H
-#define __INC_DCT_H
-
-#define prototype_fdct(sym) void (sym)(short *input, short *output, int pitch)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/dct_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/dct_arm.h"
-#endif
-
-#ifndef vp8_fdct_short4x4
-#define vp8_fdct_short4x4 vp8_short_fdct4x4_c
-#endif
-extern prototype_fdct(vp8_fdct_short4x4);
-
-#ifndef vp8_fdct_short8x4
-#define vp8_fdct_short8x4 vp8_short_fdct8x4_c
-#endif
-extern prototype_fdct(vp8_fdct_short8x4);
-
-// There is no fast4x4 (for now)
-#ifndef vp8_fdct_fast4x4
-#define vp8_fdct_fast4x4 vp8_short_fdct4x4_c
-#endif
-
-#ifndef vp8_fdct_fast8x4
-#define vp8_fdct_fast8x4 vp8_short_fdct8x4_c
-#endif
-
-#ifndef vp8_fdct_walsh_short4x4
-#define vp8_fdct_walsh_short4x4 vp8_short_walsh4x4_c
-#endif
-extern prototype_fdct(vp8_fdct_walsh_short4x4);
-
-typedef prototype_fdct(*vp8_fdct_fn_t);
-typedef struct
-{
- vp8_fdct_fn_t short4x4;
- vp8_fdct_fn_t short8x4;
- vp8_fdct_fn_t fast4x4;
- vp8_fdct_fn_t fast8x4;
- vp8_fdct_fn_t walsh_short4x4;
-} vp8_fdct_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define FDCT_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define FDCT_INVOKE(ctx,fn) vp8_fdct_##fn
-#endif
-
-#endif
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
deleted file mode 100644
index fa01d20..0000000
--- a/vp8/encoder/encodeframe.c
+++ /dev/null
@@ -1,1428 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "encodemb.h"
-#include "encodemv.h"
-#include "vp8/common/common.h"
-#include "onyx_int.h"
-#include "vp8/common/extend.h"
-#include "vp8/common/entropymode.h"
-#include "vp8/common/quant_common.h"
-#include "segmentation.h"
-#include "vp8/common/setupintrarecon.h"
-#include "encodeintra.h"
-#include "vp8/common/reconinter.h"
-#include "rdopt.h"
-#include "pickinter.h"
-#include "vp8/common/findnearmv.h"
-#include "vp8/common/reconintra.h"
-#include <stdio.h>
-#include <limits.h>
-#include "vp8/common/subpixel.h"
-#include "vpx_ports/vpx_timer.h"
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define RTCD(x) &cpi->common.rtcd.x
-#define IF_RTCD(x) (x)
-#else
-#define RTCD(x) NULL
-#define IF_RTCD(x) NULL
-#endif
-extern void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) ;
-
-extern void vp8cx_initialize_me_consts(VP8_COMP *cpi, int QIndex);
-extern void vp8_auto_select_speed(VP8_COMP *cpi);
-extern void vp8cx_init_mbrthread_data(VP8_COMP *cpi,
- MACROBLOCK *x,
- MB_ROW_COMP *mbr_ei,
- int mb_row,
- int count);
-void vp8_build_block_offsets(MACROBLOCK *x);
-void vp8_setup_block_ptrs(MACROBLOCK *x);
-int vp8cx_encode_inter_macroblock(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t, int recon_yoffset, int recon_uvoffset);
-int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t);
-
-#ifdef MODE_STATS
-unsigned int inter_y_modes[10] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
-unsigned int inter_uv_modes[4] = {0, 0, 0, 0};
-unsigned int inter_b_modes[15] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
-unsigned int y_modes[5] = {0, 0, 0, 0, 0};
-unsigned int uv_modes[4] = {0, 0, 0, 0};
-unsigned int b_modes[14] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
-#endif
-
-static const int qrounding_factors[129] =
-{
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48
-};
-
-static const int qzbin_factors[129] =
-{
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80
-};
-
-static const int qrounding_factors_y2[129] =
-{
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48, 48, 48, 48, 48, 48, 48, 48,
- 48
-};
-
-static const int qzbin_factors_y2[129] =
-{
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 84, 84, 84, 84, 84, 84, 84, 84,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80, 80, 80, 80, 80, 80, 80, 80,
- 80
-};
-
-#define EXACT_QUANT
-#ifdef EXACT_QUANT
-static void vp8cx_invert_quant(int improved_quant, short *quant,
- short *shift, short d)
-{
- if(improved_quant)
- {
- unsigned t;
- int l;
- t = d;
- for(l = 0; t > 1; l++)
- t>>=1;
- t = 1 + (1<<(16+l))/d;
- *quant = (short)(t - (1<<16));
- *shift = l;
- }
- else
- {
- *quant = (1 << 16) / d;
- *shift = 0;
- }
-}
-
-void vp8cx_init_quantizer(VP8_COMP *cpi)
-{
- int i;
- int quant_val;
- int Q;
-
- int zbin_boost[16] = {0, 0, 8, 10, 12, 14, 16, 20, 24, 28, 32, 36, 40, 44, 44, 44};
-
- for (Q = 0; Q < QINDEX_RANGE; Q++)
- {
- // dc values
- quant_val = vp8_dc_quant(Q, cpi->common.y1dc_delta_q);
- cpi->Y1quant_fast[Q][0] = (1 << 16) / quant_val;
- vp8cx_invert_quant(cpi->sf.improved_quant, cpi->Y1quant[Q] + 0,
- cpi->Y1quant_shift[Q] + 0, quant_val);
- cpi->Y1zbin[Q][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
- cpi->Y1round[Q][0] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.Y1dequant[Q][0] = quant_val;
- cpi->zrun_zbin_boost_y1[Q][0] = (quant_val * zbin_boost[0]) >> 7;
-
- quant_val = vp8_dc2quant(Q, cpi->common.y2dc_delta_q);
- cpi->Y2quant_fast[Q][0] = (1 << 16) / quant_val;
- vp8cx_invert_quant(cpi->sf.improved_quant, cpi->Y2quant[Q] + 0,
- cpi->Y2quant_shift[Q] + 0, quant_val);
- cpi->Y2zbin[Q][0] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
- cpi->Y2round[Q][0] = (qrounding_factors_y2[Q] * quant_val) >> 7;
- cpi->common.Y2dequant[Q][0] = quant_val;
- cpi->zrun_zbin_boost_y2[Q][0] = (quant_val * zbin_boost[0]) >> 7;
-
- quant_val = vp8_dc_uv_quant(Q, cpi->common.uvdc_delta_q);
- cpi->UVquant_fast[Q][0] = (1 << 16) / quant_val;
- vp8cx_invert_quant(cpi->sf.improved_quant, cpi->UVquant[Q] + 0,
- cpi->UVquant_shift[Q] + 0, quant_val);
- cpi->UVzbin[Q][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;;
- cpi->UVround[Q][0] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.UVdequant[Q][0] = quant_val;
- cpi->zrun_zbin_boost_uv[Q][0] = (quant_val * zbin_boost[0]) >> 7;
-
- // all the ac values = ;
- for (i = 1; i < 16; i++)
- {
- int rc = vp8_default_zig_zag1d[i];
-
- quant_val = vp8_ac_yquant(Q);
- cpi->Y1quant_fast[Q][rc] = (1 << 16) / quant_val;
- vp8cx_invert_quant(cpi->sf.improved_quant, cpi->Y1quant[Q] + rc,
- cpi->Y1quant_shift[Q] + rc, quant_val);
- cpi->Y1zbin[Q][rc] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
- cpi->Y1round[Q][rc] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.Y1dequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_y1[Q][i] = (quant_val * zbin_boost[i]) >> 7;
-
- quant_val = vp8_ac2quant(Q, cpi->common.y2ac_delta_q);
- cpi->Y2quant_fast[Q][rc] = (1 << 16) / quant_val;
- vp8cx_invert_quant(cpi->sf.improved_quant, cpi->Y2quant[Q] + rc,
- cpi->Y2quant_shift[Q] + rc, quant_val);
- cpi->Y2zbin[Q][rc] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
- cpi->Y2round[Q][rc] = (qrounding_factors_y2[Q] * quant_val) >> 7;
- cpi->common.Y2dequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_y2[Q][i] = (quant_val * zbin_boost[i]) >> 7;
-
- quant_val = vp8_ac_uv_quant(Q, cpi->common.uvac_delta_q);
- cpi->UVquant_fast[Q][rc] = (1 << 16) / quant_val;
- vp8cx_invert_quant(cpi->sf.improved_quant, cpi->UVquant[Q] + rc,
- cpi->UVquant_shift[Q] + rc, quant_val);
- cpi->UVzbin[Q][rc] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
- cpi->UVround[Q][rc] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.UVdequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_uv[Q][i] = (quant_val * zbin_boost[i]) >> 7;
- }
- }
-}
-#else
-void vp8cx_init_quantizer(VP8_COMP *cpi)
-{
- int i;
- int quant_val;
- int Q;
-
- int zbin_boost[16] = {0, 0, 8, 10, 12, 14, 16, 20, 24, 28, 32, 36, 40, 44, 44, 44};
-
- for (Q = 0; Q < QINDEX_RANGE; Q++)
- {
- // dc values
- quant_val = vp8_dc_quant(Q, cpi->common.y1dc_delta_q);
- cpi->Y1quant[Q][0] = (1 << 16) / quant_val;
- cpi->Y1zbin[Q][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
- cpi->Y1round[Q][0] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.Y1dequant[Q][0] = quant_val;
- cpi->zrun_zbin_boost_y1[Q][0] = (quant_val * zbin_boost[0]) >> 7;
-
- quant_val = vp8_dc2quant(Q, cpi->common.y2dc_delta_q);
- cpi->Y2quant[Q][0] = (1 << 16) / quant_val;
- cpi->Y2zbin[Q][0] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
- cpi->Y2round[Q][0] = (qrounding_factors_y2[Q] * quant_val) >> 7;
- cpi->common.Y2dequant[Q][0] = quant_val;
- cpi->zrun_zbin_boost_y2[Q][0] = (quant_val * zbin_boost[0]) >> 7;
-
- quant_val = vp8_dc_uv_quant(Q, cpi->common.uvdc_delta_q);
- cpi->UVquant[Q][0] = (1 << 16) / quant_val;
- cpi->UVzbin[Q][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;;
- cpi->UVround[Q][0] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.UVdequant[Q][0] = quant_val;
- cpi->zrun_zbin_boost_uv[Q][0] = (quant_val * zbin_boost[0]) >> 7;
-
- // all the ac values = ;
- for (i = 1; i < 16; i++)
- {
- int rc = vp8_default_zig_zag1d[i];
-
- quant_val = vp8_ac_yquant(Q);
- cpi->Y1quant[Q][rc] = (1 << 16) / quant_val;
- cpi->Y1zbin[Q][rc] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
- cpi->Y1round[Q][rc] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.Y1dequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_y1[Q][i] = (quant_val * zbin_boost[i]) >> 7;
-
- quant_val = vp8_ac2quant(Q, cpi->common.y2ac_delta_q);
- cpi->Y2quant[Q][rc] = (1 << 16) / quant_val;
- cpi->Y2zbin[Q][rc] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
- cpi->Y2round[Q][rc] = (qrounding_factors_y2[Q] * quant_val) >> 7;
- cpi->common.Y2dequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_y2[Q][i] = (quant_val * zbin_boost[i]) >> 7;
-
- quant_val = vp8_ac_uv_quant(Q, cpi->common.uvac_delta_q);
- cpi->UVquant[Q][rc] = (1 << 16) / quant_val;
- cpi->UVzbin[Q][rc] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
- cpi->UVround[Q][rc] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.UVdequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_uv[Q][i] = (quant_val * zbin_boost[i]) >> 7;
- }
- }
-}
-#endif
-void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x)
-{
- int i;
- int QIndex;
- MACROBLOCKD *xd = &x->e_mbd;
- int zbin_extra;
-
- // Select the baseline MB Q index.
- if (xd->segmentation_enabled)
- {
- // Abs Value
- if (xd->mb_segement_abs_delta == SEGMENT_ABSDATA)
-
- QIndex = xd->segment_feature_data[MB_LVL_ALT_Q][xd->mode_info_context->mbmi.segment_id];
- // Delta Value
- else
- {
- QIndex = cpi->common.base_qindex + xd->segment_feature_data[MB_LVL_ALT_Q][xd->mode_info_context->mbmi.segment_id];
- QIndex = (QIndex >= 0) ? ((QIndex <= MAXQ) ? QIndex : MAXQ) : 0; // Clamp to valid range
- }
- }
- else
- QIndex = cpi->common.base_qindex;
-
- // Y
- zbin_extra = (cpi->common.Y1dequant[QIndex][1] * (cpi->zbin_over_quant + cpi->zbin_mode_boost)) >> 7;
-
- for (i = 0; i < 16; i++)
- {
- x->block[i].quant = cpi->Y1quant[QIndex];
- x->block[i].quant_fast = cpi->Y1quant_fast[QIndex];
- x->block[i].quant_shift = cpi->Y1quant_shift[QIndex];
- x->block[i].zbin = cpi->Y1zbin[QIndex];
- x->block[i].round = cpi->Y1round[QIndex];
- x->e_mbd.block[i].dequant = cpi->common.Y1dequant[QIndex];
- x->block[i].zrun_zbin_boost = cpi->zrun_zbin_boost_y1[QIndex];
- x->block[i].zbin_extra = (short)zbin_extra;
- }
-
- // UV
- zbin_extra = (cpi->common.UVdequant[QIndex][1] * (cpi->zbin_over_quant + cpi->zbin_mode_boost)) >> 7;
-
- for (i = 16; i < 24; i++)
- {
- x->block[i].quant = cpi->UVquant[QIndex];
- x->block[i].quant_fast = cpi->UVquant_fast[QIndex];
- x->block[i].quant_shift = cpi->UVquant_shift[QIndex];
- x->block[i].zbin = cpi->UVzbin[QIndex];
- x->block[i].round = cpi->UVround[QIndex];
- x->e_mbd.block[i].dequant = cpi->common.UVdequant[QIndex];
- x->block[i].zrun_zbin_boost = cpi->zrun_zbin_boost_uv[QIndex];
- x->block[i].zbin_extra = (short)zbin_extra;
- }
-
- // Y2
- zbin_extra = (cpi->common.Y2dequant[QIndex][1] * ((cpi->zbin_over_quant / 2) + cpi->zbin_mode_boost)) >> 7;
- x->block[24].quant_fast = cpi->Y2quant_fast[QIndex];
- x->block[24].quant = cpi->Y2quant[QIndex];
- x->block[24].quant_shift = cpi->Y2quant_shift[QIndex];
- x->block[24].zbin = cpi->Y2zbin[QIndex];
- x->block[24].round = cpi->Y2round[QIndex];
- x->e_mbd.block[24].dequant = cpi->common.Y2dequant[QIndex];
- x->block[24].zrun_zbin_boost = cpi->zrun_zbin_boost_y2[QIndex];
- x->block[24].zbin_extra = (short)zbin_extra;
-
- /* save this macroblock QIndex for vp8_update_zbin_extra() */
- x->q_index = QIndex;
-}
-void vp8_update_zbin_extra(VP8_COMP *cpi, MACROBLOCK *x)
-{
- int i;
- int QIndex = x->q_index;
- int zbin_extra;
-
- // Y
- zbin_extra = (cpi->common.Y1dequant[QIndex][1] * (cpi->zbin_over_quant + cpi->zbin_mode_boost)) >> 7;
- for (i = 0; i < 16; i++)
- {
- x->block[i].zbin_extra = (short)zbin_extra;
- }
-
- // UV
- zbin_extra = (cpi->common.UVdequant[QIndex][1] * (cpi->zbin_over_quant + cpi->zbin_mode_boost)) >> 7;
- for (i = 16; i < 24; i++)
- {
- x->block[i].zbin_extra = (short)zbin_extra;
- }
-
- // Y2
- zbin_extra = (cpi->common.Y2dequant[QIndex][1] * ((cpi->zbin_over_quant / 2) + cpi->zbin_mode_boost)) >> 7;
- x->block[24].zbin_extra = (short)zbin_extra;
-}
-
-void vp8cx_frame_init_quantizer(VP8_COMP *cpi)
-{
- // Clear Zbin mode boost for default case
- cpi->zbin_mode_boost = 0;
-
- // MB level quantizer setup
- vp8cx_mb_init_quantizer(cpi, &cpi->mb);
-}
-
-
-/* activity_avg must be positive, or flat regions could get a zero weight
- * (infinite lambda), which confounds analysis.
- * This also avoids the need for divide by zero checks in
- * vp8_activity_masking().
- */
-#define VP8_ACTIVITY_AVG_MIN (64)
-
-/* This is used as a reference when computing the source variance for the
- * purposes of activity masking.
- * Eventually this should be replaced by custom no-reference routines,
- * which will be faster.
- */
-static const unsigned char VP8_VAR_OFFS[16]=
-{
- 128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128
-};
-
-unsigned int vp8_activity_masking(VP8_COMP *cpi, MACROBLOCK *x)
-{
- unsigned int act;
- unsigned int sse;
- int sum;
- unsigned int a;
- unsigned int b;
- /* TODO: This could also be done over smaller areas (8x8), but that would
- * require extensive changes elsewhere, as lambda is assumed to be fixed
- * over an entire MB in most of the code.
- * Another option is to compute four 8x8 variances, and pick a single
- * lambda using a non-linear combination (e.g., the smallest, or second
- * smallest, etc.).
- */
- VARIANCE_INVOKE(&cpi->rtcd.variance, get16x16var)(x->src.y_buffer,
- x->src.y_stride, VP8_VAR_OFFS, 0, &sse, &sum);
- /* This requires a full 32 bits of precision. */
- act = (sse<<8) - sum*sum;
- /* Drop 4 to give us some headroom to work with. */
- act = (act + 8) >> 4;
- /* If the region is flat, lower the activity some more. */
- if (act < 8<<12)
- act = act < 5<<12 ? act : 5<<12;
- /* TODO: For non-flat regions, edge regions should receive less masking
- * than textured regions, but identifying edge regions quickly and
- * reliably enough is still a subject of experimentation.
- * This will be most noticable near edges with a complex shape (e.g.,
- * text), but the 4x4 transform size should make this less of a problem
- * than it would be for an 8x8 transform.
- */
- /* Apply the masking to the RD multiplier. */
- a = act + 4*cpi->activity_avg;
- b = 4*act + cpi->activity_avg;
- x->rdmult = (unsigned int)(((INT64)x->rdmult*b + (a>>1))/a);
- return act;
-}
-
-
-
-static
-void encode_mb_row(VP8_COMP *cpi,
- VP8_COMMON *cm,
- int mb_row,
- MACROBLOCK *x,
- MACROBLOCKD *xd,
- TOKENEXTRA **tp,
- int *segment_counts,
- int *totalrate)
-{
- INT64 activity_sum = 0;
- int i;
- int recon_yoffset, recon_uvoffset;
- int mb_col;
- int ref_fb_idx = cm->lst_fb_idx;
- int dst_fb_idx = cm->new_fb_idx;
- int recon_y_stride = cm->yv12_fb[ref_fb_idx].y_stride;
- int recon_uv_stride = cm->yv12_fb[ref_fb_idx].uv_stride;
- int seg_map_index = (mb_row * cpi->common.mb_cols);
-
-#if CONFIG_MULTITHREAD
- const int nsync = cpi->mt_sync_range;
- const int rightmost_col = cm->mb_cols - 1;
- volatile const int *last_row_current_mb_col;
-
- if ((cpi->b_multi_threaded != 0) && (mb_row != 0))
- last_row_current_mb_col = &cpi->mt_current_mb_col[mb_row - 1];
- else
- last_row_current_mb_col = &rightmost_col;
-#endif
-
- // reset above block coeffs
- xd->above_context = cm->above_context;
-
- xd->up_available = (mb_row != 0);
- recon_yoffset = (mb_row * recon_y_stride * 16);
- recon_uvoffset = (mb_row * recon_uv_stride * 8);
-
- cpi->tplist[mb_row].start = *tp;
- //printf("Main mb_row = %d\n", mb_row);
-
- // Distance of Mb to the top & bottom edges, specified in 1/8th pel
- // units as they are always compared to values that are in 1/8th pel units
- xd->mb_to_top_edge = -((mb_row * 16) << 3);
- xd->mb_to_bottom_edge = ((cm->mb_rows - 1 - mb_row) * 16) << 3;
-
- // Set up limit values for vertical motion vector components
- // to prevent them extending beyond the UMV borders
- x->mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 16));
- x->mv_row_max = ((cm->mb_rows - 1 - mb_row) * 16)
- + (VP8BORDERINPIXELS - 16);
-
- // for each macroblock col in image
- for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
- {
- // Distance of Mb to the left & right edges, specified in
- // 1/8th pel units as they are always compared to values
- // that are in 1/8th pel units
- xd->mb_to_left_edge = -((mb_col * 16) << 3);
- xd->mb_to_right_edge = ((cm->mb_cols - 1 - mb_col) * 16) << 3;
-
- // Set up limit values for horizontal motion vector components
- // to prevent them extending beyond the UMV borders
- x->mv_col_min = -((mb_col * 16) + (VP8BORDERINPIXELS - 16));
- x->mv_col_max = ((cm->mb_cols - 1 - mb_col) * 16)
- + (VP8BORDERINPIXELS - 16);
-
- xd->dst.y_buffer = cm->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset;
- xd->dst.u_buffer = cm->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset;
- xd->dst.v_buffer = cm->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset;
- xd->left_available = (mb_col != 0);
-
- x->rddiv = cpi->RDDIV;
- x->rdmult = cpi->RDMULT;
-
-#if CONFIG_MULTITHREAD
- if ((cpi->b_multi_threaded != 0) && (mb_row != 0))
- {
- if ((mb_col & (nsync - 1)) == 0)
- {
- while (mb_col > (*last_row_current_mb_col - nsync)
- && (*last_row_current_mb_col) != (cm->mb_cols - 1))
- {
- x86_pause_hint();
- thread_sleep(0);
- }
- }
- }
-#endif
-
- if(cpi->oxcf.tuning == VP8_TUNE_SSIM)
- activity_sum += vp8_activity_masking(cpi, x);
-
- // Is segmentation enabled
- // MB level adjutment to quantizer
- if (xd->segmentation_enabled)
- {
- // Code to set segment id in xd->mbmi.segment_id for current MB (with range checking)
- if (cpi->segmentation_map[seg_map_index+mb_col] <= 3)
- xd->mode_info_context->mbmi.segment_id = cpi->segmentation_map[seg_map_index+mb_col];
- else
- xd->mode_info_context->mbmi.segment_id = 0;
-
- vp8cx_mb_init_quantizer(cpi, x);
- }
- else
- xd->mode_info_context->mbmi.segment_id = 0; // Set to Segment 0 by default
-
- x->active_ptr = cpi->active_map + seg_map_index + mb_col;
-
- if (cm->frame_type == KEY_FRAME)
- {
- *totalrate += vp8cx_encode_intra_macro_block(cpi, x, tp);
-#ifdef MODE_STATS
- y_modes[xd->mbmi.mode] ++;
-#endif
- }
- else
- {
- *totalrate += vp8cx_encode_inter_macroblock(cpi, x, tp, recon_yoffset, recon_uvoffset);
-
-#ifdef MODE_STATS
- inter_y_modes[xd->mbmi.mode] ++;
-
- if (xd->mbmi.mode == SPLITMV)
- {
- int b;
-
- for (b = 0; b < xd->mbmi.partition_count; b++)
- {
- inter_b_modes[x->partition->bmi[b].mode] ++;
- }
- }
-
-#endif
-
- // Count of last ref frame 0,0 useage
- if ((xd->mode_info_context->mbmi.mode == ZEROMV) && (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME))
- cpi->inter_zz_count ++;
-
- // Special case code for cyclic refresh
- // If cyclic update enabled then copy xd->mbmi.segment_id; (which may have been updated based on mode
- // during vp8cx_encode_inter_macroblock()) back into the global sgmentation map
- if (cpi->cyclic_refresh_mode_enabled && xd->segmentation_enabled)
- {
- cpi->segmentation_map[seg_map_index+mb_col] = xd->mode_info_context->mbmi.segment_id;
-
- // If the block has been refreshed mark it as clean (the magnitude of the -ve influences how long it will be before we consider another refresh):
- // Else if it was coded (last frame 0,0) and has not already been refreshed then mark it as a candidate for cleanup next time (marked 0)
- // else mark it as dirty (1).
- if (xd->mode_info_context->mbmi.segment_id)
- cpi->cyclic_refresh_map[seg_map_index+mb_col] = -1;
- else if ((xd->mode_info_context->mbmi.mode == ZEROMV) && (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME))
- {
- if (cpi->cyclic_refresh_map[seg_map_index+mb_col] == 1)
- cpi->cyclic_refresh_map[seg_map_index+mb_col] = 0;
- }
- else
- cpi->cyclic_refresh_map[seg_map_index+mb_col] = 1;
-
- }
- }
-
- cpi->tplist[mb_row].stop = *tp;
-
- x->gf_active_ptr++; // Increment pointer into gf useage flags structure for next mb
-
- for (i = 0; i < 16; i++)
- vpx_memcpy(&xd->mode_info_context->bmi[i], &xd->block[i].bmi, sizeof(xd->block[i].bmi));
-
- // adjust to the next column of macroblocks
- x->src.y_buffer += 16;
- x->src.u_buffer += 8;
- x->src.v_buffer += 8;
-
- recon_yoffset += 16;
- recon_uvoffset += 8;
-
- // Keep track of segment useage
- segment_counts[xd->mode_info_context->mbmi.segment_id] ++;
-
- // skip to next mb
- xd->mode_info_context++;
- x->partition_info++;
-
- xd->above_context++;
-#if CONFIG_MULTITHREAD
- if (cpi->b_multi_threaded != 0)
- {
- cpi->mt_current_mb_col[mb_row] = mb_col;
- }
-#endif
- }
-
- //extend the recon for intra prediction
- vp8_extend_mb_row(
- &cm->yv12_fb[dst_fb_idx],
- xd->dst.y_buffer + 16,
- xd->dst.u_buffer + 8,
- xd->dst.v_buffer + 8);
-
- // this is to account for the border
- xd->mode_info_context++;
- x->partition_info++;
- x->activity_sum += activity_sum;
-
-#if CONFIG_MULTITHREAD
- if ((cpi->b_multi_threaded != 0) && (mb_row == cm->mb_rows - 1))
- {
- sem_post(&cpi->h_event_end_encoding); /* signal frame encoding end */
- }
-#endif
-}
-
-void vp8_encode_frame(VP8_COMP *cpi)
-{
- int mb_row;
- MACROBLOCK *const x = & cpi->mb;
- VP8_COMMON *const cm = & cpi->common;
- MACROBLOCKD *const xd = & x->e_mbd;
-
- TOKENEXTRA *tp = cpi->tok;
- int segment_counts[MAX_MB_SEGMENTS];
- int totalrate;
-
- // Functions setup for all frame types so we can use MC in AltRef
- if (cm->mcomp_filter_type == SIXTAP)
- {
- xd->subpixel_predict = SUBPIX_INVOKE(
- &cpi->common.rtcd.subpix, sixtap4x4);
- xd->subpixel_predict8x4 = SUBPIX_INVOKE(
- &cpi->common.rtcd.subpix, sixtap8x4);
- xd->subpixel_predict8x8 = SUBPIX_INVOKE(
- &cpi->common.rtcd.subpix, sixtap8x8);
- xd->subpixel_predict16x16 = SUBPIX_INVOKE(
- &cpi->common.rtcd.subpix, sixtap16x16);
- }
- else
- {
- xd->subpixel_predict = SUBPIX_INVOKE(
- &cpi->common.rtcd.subpix, bilinear4x4);
- xd->subpixel_predict8x4 = SUBPIX_INVOKE(
- &cpi->common.rtcd.subpix, bilinear8x4);
- xd->subpixel_predict8x8 = SUBPIX_INVOKE(
- &cpi->common.rtcd.subpix, bilinear8x8);
- xd->subpixel_predict16x16 = SUBPIX_INVOKE(
- &cpi->common.rtcd.subpix, bilinear16x16);
- }
-
- x->gf_active_ptr = (signed char *)cpi->gf_active_flags; // Point to base of GF active flags data structure
-
- x->vector_range = 32;
-
- // Count of MBs using the alternate Q if any
- cpi->alt_qcount = 0;
-
- // Reset frame count of inter 0,0 motion vector useage.
- cpi->inter_zz_count = 0;
-
- vpx_memset(segment_counts, 0, sizeof(segment_counts));
-
- cpi->prediction_error = 0;
- cpi->intra_error = 0;
- cpi->skip_true_count = 0;
- cpi->skip_false_count = 0;
-
-#if 0
- // Experimental code
- cpi->frame_distortion = 0;
- cpi->last_mb_distortion = 0;
-#endif
-
- totalrate = 0;
-
- x->partition_info = x->pi;
-
- xd->mode_info_context = cm->mi;
- xd->mode_info_stride = cm->mode_info_stride;
-
- xd->frame_type = cm->frame_type;
-
- xd->frames_since_golden = cm->frames_since_golden;
- xd->frames_till_alt_ref_frame = cm->frames_till_alt_ref_frame;
- vp8_zero(cpi->MVcount);
- // vp8_zero( Contexts)
- vp8_zero(cpi->coef_counts);
-
- // reset intra mode contexts
- if (cm->frame_type == KEY_FRAME)
- vp8_init_mbmode_probs(cm);
-
-
- vp8cx_frame_init_quantizer(cpi);
-
- if (cpi->compressor_speed == 2)
- {
- if (cpi->oxcf.cpu_used < 0)
- cpi->Speed = -(cpi->oxcf.cpu_used);
- else
- vp8_auto_select_speed(cpi);
- }
-
- vp8_initialize_rd_consts(cpi, vp8_dc_quant(cm->base_qindex, cm->y1dc_delta_q));
- vp8cx_initialize_me_consts(cpi, cm->base_qindex);
-
- // Copy data over into macro block data sturctures.
-
- x->src = * cpi->Source;
- xd->pre = cm->yv12_fb[cm->lst_fb_idx];
- xd->dst = cm->yv12_fb[cm->new_fb_idx];
-
- // set up frame new frame for intra coded blocks
-
- vp8_setup_intra_recon(&cm->yv12_fb[cm->new_fb_idx]);
-
- vp8_build_block_offsets(x);
-
- vp8_setup_block_dptrs(&x->e_mbd);
-
- vp8_setup_block_ptrs(x);
-
- x->activity_sum = 0;
-
- xd->mode_info_context->mbmi.mode = DC_PRED;
- xd->mode_info_context->mbmi.uv_mode = DC_PRED;
-
- xd->left_context = &cm->left_context;
-
- vp8_zero(cpi->count_mb_ref_frame_usage)
- vp8_zero(cpi->ymode_count)
- vp8_zero(cpi->uv_mode_count)
-
- x->mvc = cm->fc.mvc;
-
- vpx_memset(cm->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES) * cm->mb_cols);
-
- {
- struct vpx_usec_timer emr_timer;
- vpx_usec_timer_start(&emr_timer);
-
-#if CONFIG_MULTITHREAD
- if (cpi->b_multi_threaded)
- {
- int i;
-
- vp8cx_init_mbrthread_data(cpi, x, cpi->mb_row_ei, 1, cpi->encoding_thread_count);
-
- for (i = 0; i < cm->mb_rows; i++)
- cpi->mt_current_mb_col[i] = -1;
-
- for (i = 0; i < cpi->encoding_thread_count; i++)
- {
- sem_post(&cpi->h_event_start_encoding[i]);
- }
-
- for (mb_row = 0; mb_row < cm->mb_rows; mb_row += (cpi->encoding_thread_count + 1))
- {
- vp8_zero(cm->left_context)
-
- tp = cpi->tok + mb_row * (cm->mb_cols * 16 * 24);
-
- encode_mb_row(cpi, cm, mb_row, x, xd, &tp, segment_counts, &totalrate);
-
- // adjust to the next row of mbs
- x->src.y_buffer += 16 * x->src.y_stride * (cpi->encoding_thread_count + 1) - 16 * cm->mb_cols;
- x->src.u_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols;
- x->src.v_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols;
-
- xd->mode_info_context += xd->mode_info_stride * cpi->encoding_thread_count;
- x->partition_info += xd->mode_info_stride * cpi->encoding_thread_count;
-
- }
-
- sem_wait(&cpi->h_event_end_encoding); /* wait for other threads to finish */
-
- cpi->tok_count = 0;
-
- for (mb_row = 0; mb_row < cm->mb_rows; mb_row ++)
- {
- cpi->tok_count += cpi->tplist[mb_row].stop - cpi->tplist[mb_row].start;
- }
-
- if (xd->segmentation_enabled)
- {
- int i, j;
-
- if (xd->segmentation_enabled)
- {
-
- for (i = 0; i < cpi->encoding_thread_count; i++)
- {
- for (j = 0; j < 4; j++)
- segment_counts[j] += cpi->mb_row_ei[i].segment_counts[j];
- }
- }
- }
-
- for (i = 0; i < cpi->encoding_thread_count; i++)
- {
- totalrate += cpi->mb_row_ei[i].totalrate;
- }
-
- for (i = 0; i < cpi->encoding_thread_count; i++)
- {
- x->activity_sum += cpi->mb_row_ei[i].mb.activity_sum;
- }
-
- }
- else
-#endif
- {
- // for each macroblock row in image
- for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
- {
-
- vp8_zero(cm->left_context)
-
- encode_mb_row(cpi, cm, mb_row, x, xd, &tp, segment_counts, &totalrate);
-
- // adjust to the next row of mbs
- x->src.y_buffer += 16 * x->src.y_stride - 16 * cm->mb_cols;
- x->src.u_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols;
- x->src.v_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols;
- }
-
- cpi->tok_count = tp - cpi->tok;
-
- }
-
- vpx_usec_timer_mark(&emr_timer);
- cpi->time_encode_mb_row += vpx_usec_timer_elapsed(&emr_timer);
-
- }
-
-
- // Work out the segment probabilites if segmentation is enabled
- if (xd->segmentation_enabled)
- {
- int tot_count;
- int i;
-
- // Set to defaults
- vpx_memset(xd->mb_segment_tree_probs, 255 , sizeof(xd->mb_segment_tree_probs));
-
- tot_count = segment_counts[0] + segment_counts[1] + segment_counts[2] + segment_counts[3];
-
- if (tot_count)
- {
- xd->mb_segment_tree_probs[0] = ((segment_counts[0] + segment_counts[1]) * 255) / tot_count;
-
- tot_count = segment_counts[0] + segment_counts[1];
-
- if (tot_count > 0)
- {
- xd->mb_segment_tree_probs[1] = (segment_counts[0] * 255) / tot_count;
- }
-
- tot_count = segment_counts[2] + segment_counts[3];
-
- if (tot_count > 0)
- xd->mb_segment_tree_probs[2] = (segment_counts[2] * 255) / tot_count;
-
- // Zero probabilities not allowed
- for (i = 0; i < MB_FEATURE_TREE_PROBS; i ++)
- {
- if (xd->mb_segment_tree_probs[i] == 0)
- xd->mb_segment_tree_probs[i] = 1;
- }
- }
- }
-
- // 256 rate units to the bit
- cpi->projected_frame_size = totalrate >> 8; // projected_frame_size in units of BYTES
-
- // Make a note of the percentage MBs coded Intra.
- if (cm->frame_type == KEY_FRAME)
- {
- cpi->this_frame_percent_intra = 100;
- }
- else
- {
- int tot_modes;
-
- tot_modes = cpi->count_mb_ref_frame_usage[INTRA_FRAME]
- + cpi->count_mb_ref_frame_usage[LAST_FRAME]
- + cpi->count_mb_ref_frame_usage[GOLDEN_FRAME]
- + cpi->count_mb_ref_frame_usage[ALTREF_FRAME];
-
- if (tot_modes)
- cpi->this_frame_percent_intra = cpi->count_mb_ref_frame_usage[INTRA_FRAME] * 100 / tot_modes;
-
- }
-
-#if 0
- {
- int cnt = 0;
- int flag[2] = {0, 0};
-
- for (cnt = 0; cnt < MVPcount; cnt++)
- {
- if (cm->fc.pre_mvc[0][cnt] != cm->fc.mvc[0][cnt])
- {
- flag[0] = 1;
- vpx_memcpy(cm->fc.pre_mvc[0], cm->fc.mvc[0], MVPcount);
- break;
- }
- }
-
- for (cnt = 0; cnt < MVPcount; cnt++)
- {
- if (cm->fc.pre_mvc[1][cnt] != cm->fc.mvc[1][cnt])
- {
- flag[1] = 1;
- vpx_memcpy(cm->fc.pre_mvc[1], cm->fc.mvc[1], MVPcount);
- break;
- }
- }
-
- if (flag[0] || flag[1])
- vp8_build_component_cost_table(cpi->mb.mvcost, cpi->mb.mvsadcost, (const MV_CONTEXT *) cm->fc.mvc, flag);
- }
-#endif
-
- // Adjust the projected reference frame useage probability numbers to reflect
- // what we have just seen. This may be usefull when we make multiple itterations
- // of the recode loop rather than continuing to use values from the previous frame.
- if ((cm->frame_type != KEY_FRAME) && !cm->refresh_alt_ref_frame && !cm->refresh_golden_frame)
- {
- const int *const rfct = cpi->count_mb_ref_frame_usage;
- const int rf_intra = rfct[INTRA_FRAME];
- const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
-
- if ((rf_intra + rf_inter) > 0)
- {
- cpi->prob_intra_coded = (rf_intra * 255) / (rf_intra + rf_inter);
-
- if (cpi->prob_intra_coded < 1)
- cpi->prob_intra_coded = 1;
-
- if ((cm->frames_since_golden > 0) || cpi->source_alt_ref_active)
- {
- cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- if (cpi->prob_last_coded < 1)
- cpi->prob_last_coded = 1;
-
- cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- if (cpi->prob_gf_coded < 1)
- cpi->prob_gf_coded = 1;
- }
- }
- }
-
-#if 0
- // Keep record of the total distortion this time around for future use
- cpi->last_frame_distortion = cpi->frame_distortion;
-#endif
-
- /* Update the average activity for the next frame.
- * This is feed-forward for now; it could also be saved in two-pass, or
- * done during lookahead when that is eventually added.
- */
- cpi->activity_avg = (unsigned int )(x->activity_sum/cpi->common.MBs);
- if (cpi->activity_avg < VP8_ACTIVITY_AVG_MIN)
- cpi->activity_avg = VP8_ACTIVITY_AVG_MIN;
-
-}
-void vp8_setup_block_ptrs(MACROBLOCK *x)
-{
- int r, c;
- int i;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- x->block[r*4+c].src_diff = x->src_diff + r * 4 * 16 + c * 4;
- }
- }
-
- for (r = 0; r < 2; r++)
- {
- for (c = 0; c < 2; c++)
- {
- x->block[16 + r*2+c].src_diff = x->src_diff + 256 + r * 4 * 8 + c * 4;
- }
- }
-
-
- for (r = 0; r < 2; r++)
- {
- for (c = 0; c < 2; c++)
- {
- x->block[20 + r*2+c].src_diff = x->src_diff + 320 + r * 4 * 8 + c * 4;
- }
- }
-
- x->block[24].src_diff = x->src_diff + 384;
-
-
- for (i = 0; i < 25; i++)
- {
- x->block[i].coeff = x->coeff + i * 16;
- }
-}
-
-void vp8_build_block_offsets(MACROBLOCK *x)
-{
- int block = 0;
- int br, bc;
-
- vp8_build_block_doffsets(&x->e_mbd);
-
- // y blocks
- for (br = 0; br < 4; br++)
- {
- for (bc = 0; bc < 4; bc++)
- {
- BLOCK *this_block = &x->block[block];
- this_block->base_src = &x->src.y_buffer;
- this_block->src_stride = x->src.y_stride;
- this_block->src = 4 * br * this_block->src_stride + 4 * bc;
- ++block;
- }
- }
-
- // u blocks
- for (br = 0; br < 2; br++)
- {
- for (bc = 0; bc < 2; bc++)
- {
- BLOCK *this_block = &x->block[block];
- this_block->base_src = &x->src.u_buffer;
- this_block->src_stride = x->src.uv_stride;
- this_block->src = 4 * br * this_block->src_stride + 4 * bc;
- ++block;
- }
- }
-
- // v blocks
- for (br = 0; br < 2; br++)
- {
- for (bc = 0; bc < 2; bc++)
- {
- BLOCK *this_block = &x->block[block];
- this_block->base_src = &x->src.v_buffer;
- this_block->src_stride = x->src.uv_stride;
- this_block->src = 4 * br * this_block->src_stride + 4 * bc;
- ++block;
- }
- }
-}
-
-static void sum_intra_stats(VP8_COMP *cpi, MACROBLOCK *x)
-{
- const MACROBLOCKD *xd = & x->e_mbd;
- const MB_PREDICTION_MODE m = xd->mode_info_context->mbmi.mode;
- const MB_PREDICTION_MODE uvm = xd->mode_info_context->mbmi.uv_mode;
-
-#ifdef MODE_STATS
- const int is_key = cpi->common.frame_type == KEY_FRAME;
-
- ++ (is_key ? uv_modes : inter_uv_modes)[uvm];
-
- if (m == B_PRED)
- {
- unsigned int *const bct = is_key ? b_modes : inter_b_modes;
-
- int b = 0;
-
- do
- {
- ++ bct[xd->block[b].bmi.mode];
- }
- while (++b < 16);
- }
-
-#endif
-
- ++cpi->ymode_count[m];
- ++cpi->uv_mode_count[uvm];
-
-}
-int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t)
-{
- int Error4x4, Error16x16, error_uv;
- int rate4x4, rate16x16, rateuv;
- int dist4x4, dist16x16, distuv;
- int rate = 0;
- int rate4x4_tokenonly = 0;
- int rate16x16_tokenonly = 0;
- int rateuv_tokenonly = 0;
-
- x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
-
-#if !(CONFIG_REALTIME_ONLY)
- if (cpi->sf.RD && cpi->compressor_speed != 2)
- {
- error_uv = vp8_rd_pick_intra_mbuv_mode(cpi, x, &rateuv, &rateuv_tokenonly, &distuv);
- rate += rateuv;
-
- Error16x16 = vp8_rd_pick_intra16x16mby_mode(cpi, x, &rate16x16, &rate16x16_tokenonly, &dist16x16);
-
- Error4x4 = vp8_rd_pick_intra4x4mby_modes(cpi, x, &rate4x4, &rate4x4_tokenonly, &dist4x4, Error16x16);
-
- rate += (Error4x4 < Error16x16) ? rate4x4 : rate16x16;
- }
- else
-#endif
- {
- int rate2, best_distortion;
- MB_PREDICTION_MODE mode, best_mode = DC_PRED;
- int this_rd;
- Error16x16 = INT_MAX;
-
- vp8_pick_intra_mbuv_mode(x);
-
- for (mode = DC_PRED; mode <= TM_PRED; mode ++)
- {
- int distortion2;
-
- x->e_mbd.mode_info_context->mbmi.mode = mode;
- RECON_INVOKE(&cpi->common.rtcd.recon, build_intra_predictors_mby)
- (&x->e_mbd);
- distortion2 = VARIANCE_INVOKE(&cpi->rtcd.variance, get16x16prederror)(x->src.y_buffer, x->src.y_stride, x->e_mbd.predictor, 16, 0x7fffffff);
- rate2 = x->mbmode_cost[x->e_mbd.frame_type][mode];
- this_rd = RD_ESTIMATE(x->rdmult, x->rddiv, rate2, distortion2);
-
- if (Error16x16 > this_rd)
- {
- Error16x16 = this_rd;
- best_mode = mode;
- best_distortion = distortion2;
- }
- }
- x->e_mbd.mode_info_context->mbmi.mode = best_mode;
-
- Error4x4 = vp8_pick_intra4x4mby_modes(IF_RTCD(&cpi->rtcd), x, &rate2, &best_distortion);
- }
-
- if (Error4x4 < Error16x16)
- {
- x->e_mbd.mode_info_context->mbmi.mode = B_PRED;
- vp8_encode_intra4x4mby(IF_RTCD(&cpi->rtcd), x);
- }
- else
- {
- vp8_encode_intra16x16mby(IF_RTCD(&cpi->rtcd), x);
- }
-
- vp8_encode_intra16x16mbuv(IF_RTCD(&cpi->rtcd), x);
- sum_intra_stats(cpi, x);
- vp8_tokenize_mb(cpi, &x->e_mbd, t);
-
- return rate;
-}
-#ifdef SPEEDSTATS
-extern int cnt_pm;
-#endif
-
-extern void vp8_fix_contexts(MACROBLOCKD *x);
-
-int vp8cx_encode_inter_macroblock
-(
- VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t,
- int recon_yoffset, int recon_uvoffset
-)
-{
- MACROBLOCKD *const xd = &x->e_mbd;
- int inter_error;
- int intra_error = 0;
- int rate;
- int distortion;
-
- x->skip = 0;
-
- if (xd->segmentation_enabled)
- x->encode_breakout = cpi->segment_encode_breakout[xd->mode_info_context->mbmi.segment_id];
- else
- x->encode_breakout = cpi->oxcf.encode_breakout;
-
-#if !(CONFIG_REALTIME_ONLY)
-
- if (cpi->sf.RD)
- {
- int zbin_mode_boost_enabled = cpi->zbin_mode_boost_enabled;
-
- /* Are we using the fast quantizer for the mode selection? */
- if(cpi->sf.use_fastquant_for_pick)
- {
- cpi->mb.quantize_b = QUANTIZE_INVOKE(&cpi->rtcd.quantize, fastquantb);
-
- /* the fast quantizer does not use zbin_extra, so
- * do not recalculate */
- cpi->zbin_mode_boost_enabled = 0;
- }
- inter_error = vp8_rd_pick_inter_mode(cpi, x, recon_yoffset, recon_uvoffset, &rate, &distortion, &intra_error);
-
- /* switch back to the regular quantizer for the encode */
- if (cpi->sf.improved_quant)
- {
- cpi->mb.quantize_b = QUANTIZE_INVOKE(&cpi->rtcd.quantize, quantb);
- }
-
- /* restore cpi->zbin_mode_boost_enabled */
- cpi->zbin_mode_boost_enabled = zbin_mode_boost_enabled;
-
- }
- else
-#endif
- inter_error = vp8_pick_inter_mode(cpi, x, recon_yoffset, recon_uvoffset, &rate, &distortion, &intra_error);
-
-
- cpi->prediction_error += inter_error;
- cpi->intra_error += intra_error;
-
-#if 0
- // Experimental RD code
- cpi->frame_distortion += distortion;
- cpi->last_mb_distortion = distortion;
-#endif
-
- // MB level adjutment to quantizer setup
- if (xd->segmentation_enabled)
- {
- // If cyclic update enabled
- if (cpi->cyclic_refresh_mode_enabled)
- {
- // Clear segment_id back to 0 if not coded (last frame 0,0)
- if ((xd->mode_info_context->mbmi.segment_id == 1) &&
- ((xd->mode_info_context->mbmi.ref_frame != LAST_FRAME) || (xd->mode_info_context->mbmi.mode != ZEROMV)))
- {
- xd->mode_info_context->mbmi.segment_id = 0;
-
- /* segment_id changed, so update */
- vp8cx_mb_init_quantizer(cpi, x);
- }
- }
- }
-
- {
- // Experimental code. Special case for gf and arf zeromv modes. Increase zbin size to supress noise
- if (cpi->zbin_mode_boost_enabled)
- {
- if ( xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME )
- cpi->zbin_mode_boost = 0;
- else
- {
- if (xd->mode_info_context->mbmi.mode == ZEROMV)
- {
- if (xd->mode_info_context->mbmi.ref_frame != LAST_FRAME)
- cpi->zbin_mode_boost = GF_ZEROMV_ZBIN_BOOST;
- else
- cpi->zbin_mode_boost = LF_ZEROMV_ZBIN_BOOST;
- }
- else if (xd->mode_info_context->mbmi.mode == SPLITMV)
- cpi->zbin_mode_boost = 0;
- else
- cpi->zbin_mode_boost = MV_ZBIN_BOOST;
- }
- }
- else
- cpi->zbin_mode_boost = 0;
-
- vp8_update_zbin_extra(cpi, x);
- }
-
- cpi->count_mb_ref_frame_usage[xd->mode_info_context->mbmi.ref_frame] ++;
-
- if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
- {
- vp8_encode_intra16x16mbuv(IF_RTCD(&cpi->rtcd), x);
-
- if (xd->mode_info_context->mbmi.mode == B_PRED)
- {
- vp8_encode_intra4x4mby(IF_RTCD(&cpi->rtcd), x);
- }
- else
- {
- vp8_encode_intra16x16mby(IF_RTCD(&cpi->rtcd), x);
- }
-
- sum_intra_stats(cpi, x);
- }
- else
- {
- MV best_ref_mv;
- MV nearest, nearby;
- int mdcounts[4];
- int ref_fb_idx;
-
- vp8_find_near_mvs(xd, xd->mode_info_context,
- &nearest, &nearby, &best_ref_mv, mdcounts, xd->mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias);
-
- vp8_build_uvmvs(xd, cpi->common.full_pixel);
-
- if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME)
- ref_fb_idx = cpi->common.lst_fb_idx;
- else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME)
- ref_fb_idx = cpi->common.gld_fb_idx;
- else
- ref_fb_idx = cpi->common.alt_fb_idx;
-
- xd->pre.y_buffer = cpi->common.yv12_fb[ref_fb_idx].y_buffer + recon_yoffset;
- xd->pre.u_buffer = cpi->common.yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset;
- xd->pre.v_buffer = cpi->common.yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset;
-
- if (xd->mode_info_context->mbmi.mode == SPLITMV)
- {
- int i;
-
- for (i = 0; i < 16; i++)
- {
- if (xd->block[i].bmi.mode == NEW4X4)
- {
- cpi->MVcount[0][mv_max+((xd->block[i].bmi.mv.as_mv.row - best_ref_mv.row) >> 1)]++;
- cpi->MVcount[1][mv_max+((xd->block[i].bmi.mv.as_mv.col - best_ref_mv.col) >> 1)]++;
- }
- }
- }
- else if (xd->mode_info_context->mbmi.mode == NEWMV)
- {
- cpi->MVcount[0][mv_max+((xd->block[0].bmi.mv.as_mv.row - best_ref_mv.row) >> 1)]++;
- cpi->MVcount[1][mv_max+((xd->block[0].bmi.mv.as_mv.col - best_ref_mv.col) >> 1)]++;
- }
-
- if (!x->skip && !x->e_mbd.mode_info_context->mbmi.force_no_skip)
- {
- vp8_encode_inter16x16(IF_RTCD(&cpi->rtcd), x);
-
- // Clear mb_skip_coeff if mb_no_coeff_skip is not set
- if (!cpi->common.mb_no_coeff_skip)
- xd->mode_info_context->mbmi.mb_skip_coeff = 0;
-
- }
- else
- vp8_stuff_inter16x16(x);
- }
-
- if (!x->skip)
- vp8_tokenize_mb(cpi, xd, t);
- else
- {
- if (cpi->common.mb_no_coeff_skip)
- {
- if (xd->mode_info_context->mbmi.mode != B_PRED && xd->mode_info_context->mbmi.mode != SPLITMV)
- xd->mode_info_context->mbmi.dc_diff = 0;
- else
- xd->mode_info_context->mbmi.dc_diff = 1;
-
- xd->mode_info_context->mbmi.mb_skip_coeff = 1;
- cpi->skip_true_count ++;
- vp8_fix_contexts(xd);
- }
- else
- {
- vp8_stuff_mb(cpi, xd, t);
- xd->mode_info_context->mbmi.mb_skip_coeff = 0;
- cpi->skip_false_count ++;
- }
- }
-
- return rate;
-}
diff --git a/vp8/encoder/encodeintra.c b/vp8/encoder/encodeintra.c
deleted file mode 100644
index 4400006..0000000
--- a/vp8/encoder/encodeintra.c
+++ /dev/null
@@ -1,142 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/common/idct.h"
-#include "quantize.h"
-#include "vp8/common/reconintra.h"
-#include "vp8/common/reconintra4x4.h"
-#include "encodemb.h"
-#include "vp8/common/invtrans.h"
-#include "vp8/common/recon.h"
-#include "dct.h"
-#include "vp8/common/g_common.h"
-#include "encodeintra.h"
-
-#define intra4x4ibias_rate 128
-#define intra4x4pbias_rate 256
-
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IF_RTCD(x) (x)
-#else
-#define IF_RTCD(x) NULL
-#endif
-void vp8_encode_intra4x4block(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x, BLOCK *be, BLOCKD *b, int best_mode)
-{
- vp8_predict_intra4x4(b, best_mode, b->predictor);
-
- ENCODEMB_INVOKE(&rtcd->encodemb, subb)(be, b, 16);
-
- x->vp8_short_fdct4x4(be->src_diff, be->coeff, 32);
-
- x->quantize_b(be, b);
-
- vp8_inverse_transform_b(IF_RTCD(&rtcd->common->idct), b, 32);
-
- RECON_INVOKE(&rtcd->common->recon, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-}
-
-void vp8_encode_intra4x4mby(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *mb)
-{
- int i;
-
- MACROBLOCKD *x = &mb->e_mbd;
- vp8_intra_prediction_down_copy(x);
-
- for (i = 0; i < 16; i++)
- {
- BLOCK *be = &mb->block[i];
- BLOCKD *b = &x->block[i];
-
- vp8_encode_intra4x4block(rtcd, mb, be, b, b->bmi.mode);
- }
-
- return;
-}
-
-void vp8_encode_intra16x16mby(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
-{
- int b;
-
- RECON_INVOKE(&rtcd->common->recon, build_intra_predictors_mby)(&x->e_mbd);
-
- ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, x->src.y_buffer, x->e_mbd.predictor, x->src.y_stride);
-
- vp8_transform_intra_mby(x);
-
- vp8_quantize_mby(x);
-
-#if !(CONFIG_REALTIME_ONLY)
-#if 1
- if (x->optimize)
- vp8_optimize_mby(x, rtcd);
-
-#endif
-#endif
-
- vp8_inverse_transform_mby(IF_RTCD(&rtcd->common->idct), &x->e_mbd);
-
- RECON_INVOKE(&rtcd->common->recon, recon_mby)
- (IF_RTCD(&rtcd->common->recon), &x->e_mbd);
-
- // make sure block modes are set the way we want them for context updates
- for (b = 0; b < 16; b++)
- {
- BLOCKD *d = &x->e_mbd.block[b];
-
- switch (x->e_mbd.mode_info_context->mbmi.mode)
- {
-
- case DC_PRED:
- d->bmi.mode = B_DC_PRED;
- break;
- case V_PRED:
- d->bmi.mode = B_VE_PRED;
- break;
- case H_PRED:
- d->bmi.mode = B_HE_PRED;
- break;
- case TM_PRED:
- d->bmi.mode = B_TM_PRED;
- break;
- default:
- d->bmi.mode = B_DC_PRED;
- break;
-
- }
- }
-}
-
-void vp8_encode_intra16x16mbuv(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
-{
- vp8_build_intra_predictors_mbuv(&x->e_mbd);
-
- ENCODEMB_INVOKE(&rtcd->encodemb, submbuv)(x->src_diff, x->src.u_buffer, x->src.v_buffer, x->e_mbd.predictor, x->src.uv_stride);
-
- vp8_transform_mbuv(x);
-
- vp8_quantize_mbuv(x);
-
-#if !(CONFIG_REALTIME_ONLY)
-#if 1
-
- if (x->optimize==2 ||(x->optimize && x->rddiv > 1))
- vp8_optimize_mbuv(x, rtcd);
-
-#endif
-#endif
-
- vp8_inverse_transform_mbuv(IF_RTCD(&rtcd->common->idct), &x->e_mbd);
-
- vp8_recon_intra_mbuv(IF_RTCD(&rtcd->common->recon), &x->e_mbd);
-}
-
diff --git a/vp8/encoder/encodeintra.h b/vp8/encoder/encodeintra.h
deleted file mode 100644
index 40930bc..0000000
--- a/vp8/encoder/encodeintra.h
+++ /dev/null
@@ -1,22 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef _ENCODEINTRA_H_
-#define _ENCODEINTRA_H_
-#include "onyx_int.h"
-
-void vp8_encode_intra16x16mby(const VP8_ENCODER_RTCD *, MACROBLOCK *x);
-void vp8_encode_intra16x16mbuv(const VP8_ENCODER_RTCD *, MACROBLOCK *x);
-void vp8_encode_intra4x4mby(const VP8_ENCODER_RTCD *, MACROBLOCK *mb);
-void vp8_encode_intra4x4block(const VP8_ENCODER_RTCD *, MACROBLOCK *x, BLOCK *be, BLOCKD *b, int best_mode);
-void vp8_encode_intra4x4block_rd(const VP8_ENCODER_RTCD *, MACROBLOCK *x, BLOCK *be, BLOCKD *b, int best_mode);
-
-#endif
diff --git a/vp8/encoder/encodemb.c b/vp8/encoder/encodemb.c
deleted file mode 100644
index 463dbca..0000000
--- a/vp8/encoder/encodemb.c
+++ /dev/null
@@ -1,662 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "encodemb.h"
-#include "vp8/common/reconinter.h"
-#include "quantize.h"
-#include "tokenize.h"
-#include "vp8/common/invtrans.h"
-#include "vp8/common/recon.h"
-#include "vp8/common/reconintra.h"
-#include "dct.h"
-#include "vpx_mem/vpx_mem.h"
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IF_RTCD(x) (x)
-#else
-#define IF_RTCD(x) NULL
-#endif
-void vp8_subtract_b_c(BLOCK *be, BLOCKD *bd, int pitch)
-{
- unsigned char *src_ptr = (*(be->base_src) + be->src);
- short *diff_ptr = be->src_diff;
- unsigned char *pred_ptr = bd->predictor;
- int src_stride = be->src_stride;
-
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- diff_ptr[c] = src_ptr[c] - pred_ptr[c];
- }
-
- diff_ptr += pitch;
- pred_ptr += pitch;
- src_ptr += src_stride;
- }
-}
-
-void vp8_subtract_mbuv_c(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride)
-{
- short *udiff = diff + 256;
- short *vdiff = diff + 320;
- unsigned char *upred = pred + 256;
- unsigned char *vpred = pred + 320;
-
- int r, c;
-
- for (r = 0; r < 8; r++)
- {
- for (c = 0; c < 8; c++)
- {
- udiff[c] = usrc[c] - upred[c];
- }
-
- udiff += 8;
- upred += 8;
- usrc += stride;
- }
-
- for (r = 0; r < 8; r++)
- {
- for (c = 0; c < 8; c++)
- {
- vdiff[c] = vsrc[c] - vpred[c];
- }
-
- vdiff += 8;
- vpred += 8;
- vsrc += stride;
- }
-}
-
-void vp8_subtract_mby_c(short *diff, unsigned char *src, unsigned char *pred, int stride)
-{
- int r, c;
-
- for (r = 0; r < 16; r++)
- {
- for (c = 0; c < 16; c++)
- {
- diff[c] = src[c] - pred[c];
- }
-
- diff += 16;
- pred += 16;
- src += stride;
- }
-}
-
-static void vp8_subtract_mb(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
-{
- ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, x->src.y_buffer, x->e_mbd.predictor, x->src.y_stride);
- ENCODEMB_INVOKE(&rtcd->encodemb, submbuv)(x->src_diff, x->src.u_buffer, x->src.v_buffer, x->e_mbd.predictor, x->src.uv_stride);
-}
-
-static void build_dcblock(MACROBLOCK *x)
-{
- short *src_diff_ptr = &x->src_diff[384];
- int i;
-
- for (i = 0; i < 16; i++)
- {
- src_diff_ptr[i] = x->coeff[i * 16];
- }
-}
-
-void vp8_transform_mbuv(MACROBLOCK *x)
-{
- int i;
-
- for (i = 16; i < 24; i += 2)
- {
- x->vp8_short_fdct8x4(&x->block[i].src_diff[0],
- &x->block[i].coeff[0], 16);
- }
-}
-
-
-void vp8_transform_intra_mby(MACROBLOCK *x)
-{
- int i;
-
- for (i = 0; i < 16; i += 2)
- {
- x->vp8_short_fdct8x4(&x->block[i].src_diff[0],
- &x->block[i].coeff[0], 32);
- }
-
- // build dc block from 16 y dc values
- build_dcblock(x);
-
- // do 2nd order transform on the dc block
- x->short_walsh4x4(&x->block[24].src_diff[0],
- &x->block[24].coeff[0], 8);
-
-}
-
-
-static void transform_mb(MACROBLOCK *x)
-{
- int i;
-
- for (i = 0; i < 16; i += 2)
- {
- x->vp8_short_fdct8x4(&x->block[i].src_diff[0],
- &x->block[i].coeff[0], 32);
- }
-
- // build dc block from 16 y dc values
- if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV)
- build_dcblock(x);
-
- for (i = 16; i < 24; i += 2)
- {
- x->vp8_short_fdct8x4(&x->block[i].src_diff[0],
- &x->block[i].coeff[0], 16);
- }
-
- // do 2nd order transform on the dc block
- if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV)
- x->short_walsh4x4(&x->block[24].src_diff[0],
- &x->block[24].coeff[0], 8);
-
-}
-
-
-static void transform_mby(MACROBLOCK *x)
-{
- int i;
-
- for (i = 0; i < 16; i += 2)
- {
- x->vp8_short_fdct8x4(&x->block[i].src_diff[0],
- &x->block[i].coeff[0], 32);
- }
-
- // build dc block from 16 y dc values
- if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV)
- {
- build_dcblock(x);
- x->short_walsh4x4(&x->block[24].src_diff[0],
- &x->block[24].coeff[0], 8);
- }
-}
-
-
-void vp8_stuff_inter16x16(MACROBLOCK *x)
-{
- vp8_build_inter_predictors_mb_s(&x->e_mbd);
- /*
- // recon = copy from predictors to destination
- {
- BLOCKD *b = &x->e_mbd.block[0];
- unsigned char *pred_ptr = b->predictor;
- unsigned char *dst_ptr = *(b->base_dst) + b->dst;
- int stride = b->dst_stride;
-
- int i;
- for(i=0;i<16;i++)
- vpx_memcpy(dst_ptr+i*stride,pred_ptr+16*i,16);
-
- b = &x->e_mbd.block[16];
- pred_ptr = b->predictor;
- dst_ptr = *(b->base_dst) + b->dst;
- stride = b->dst_stride;
-
- for(i=0;i<8;i++)
- vpx_memcpy(dst_ptr+i*stride,pred_ptr+8*i,8);
-
- b = &x->e_mbd.block[20];
- pred_ptr = b->predictor;
- dst_ptr = *(b->base_dst) + b->dst;
- stride = b->dst_stride;
-
- for(i=0;i<8;i++)
- vpx_memcpy(dst_ptr+i*stride,pred_ptr+8*i,8);
- }
- */
-}
-
-#if !(CONFIG_REALTIME_ONLY)
-#define RDCOST(RM,DM,R,D) ( ((128+(R)*(RM)) >> 8) + (DM)*(D) )
-#define RDTRUNC(RM,DM,R,D) ( (128+(R)*(RM)) & 0xFF )
-
-typedef struct vp8_token_state vp8_token_state;
-
-struct vp8_token_state{
- int rate;
- int error;
- signed char next;
- signed char token;
- short qc;
-};
-
-// TODO: experiments to find optimal multiple numbers
-#define Y1_RD_MULT 4
-#define UV_RD_MULT 2
-#define Y2_RD_MULT 16
-
-static const int plane_rd_mult[4]=
-{
- Y1_RD_MULT,
- Y2_RD_MULT,
- UV_RD_MULT,
- Y1_RD_MULT
-};
-
-static void optimize_b(MACROBLOCK *mb, int ib, int type,
- ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l,
- const VP8_ENCODER_RTCD *rtcd)
-{
- BLOCK *b;
- BLOCKD *d;
- vp8_token_state tokens[17][2];
- unsigned best_mask[2];
- const short *dequant_ptr;
- const short *coeff_ptr;
- short *qcoeff_ptr;
- short *dqcoeff_ptr;
- int eob;
- int i0;
- int rc;
- int x;
- int sz;
- int next;
- int rdmult;
- int rddiv;
- int final_eob;
- int rd_cost0;
- int rd_cost1;
- int rate0;
- int rate1;
- int error0;
- int error1;
- int t0;
- int t1;
- int best;
- int band;
- int pt;
- int i;
- int err_mult = plane_rd_mult[type];
-
- b = &mb->block[ib];
- d = &mb->e_mbd.block[ib];
-
- /* Enable this to test the effect of RDO as a replacement for the dynamic
- * zero bin instead of an augmentation of it.
- */
-#if 0
- vp8_strict_quantize_b(b, d);
-#endif
-
- dequant_ptr = d->dequant;
- coeff_ptr = b->coeff;
- qcoeff_ptr = d->qcoeff;
- dqcoeff_ptr = d->dqcoeff;
- i0 = !type;
- eob = d->eob;
-
- /* Now set up a Viterbi trellis to evaluate alternative roundings. */
- rdmult = mb->rdmult * err_mult;
- if(mb->e_mbd.mode_info_context->mbmi.ref_frame==INTRA_FRAME)
- rdmult = (rdmult * 9)>>4;
-
- rddiv = mb->rddiv;
- best_mask[0] = best_mask[1] = 0;
- /* Initialize the sentinel node of the trellis. */
- tokens[eob][0].rate = 0;
- tokens[eob][0].error = 0;
- tokens[eob][0].next = 16;
- tokens[eob][0].token = DCT_EOB_TOKEN;
- tokens[eob][0].qc = 0;
- *(tokens[eob] + 1) = *(tokens[eob] + 0);
- next = eob;
- for (i = eob; i-- > i0;)
- {
- int base_bits;
- int d2;
- int dx;
-
- rc = vp8_default_zig_zag1d[i];
- x = qcoeff_ptr[rc];
- /* Only add a trellis state for non-zero coefficients. */
- if (x)
- {
- int shortcut=0;
- error0 = tokens[next][0].error;
- error1 = tokens[next][1].error;
- /* Evaluate the first possibility for this state. */
- rate0 = tokens[next][0].rate;
- rate1 = tokens[next][1].rate;
- t0 = (vp8_dct_value_tokens_ptr + x)->Token;
- /* Consider both possible successor states. */
- if (next < 16)
- {
- band = vp8_coef_bands[i + 1];
- pt = vp8_prev_token_class[t0];
- rate0 +=
- mb->token_costs[type][band][pt][tokens[next][0].token];
- rate1 +=
- mb->token_costs[type][band][pt][tokens[next][1].token];
- }
- rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0);
- rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1);
- if (rd_cost0 == rd_cost1)
- {
- rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0);
- rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1);
- }
- /* And pick the best. */
- best = rd_cost1 < rd_cost0;
- base_bits = *(vp8_dct_value_cost_ptr + x);
- dx = dqcoeff_ptr[rc] - coeff_ptr[rc];
- d2 = dx*dx;
- tokens[i][0].rate = base_bits + (best ? rate1 : rate0);
- tokens[i][0].error = d2 + (best ? error1 : error0);
- tokens[i][0].next = next;
- tokens[i][0].token = t0;
- tokens[i][0].qc = x;
- best_mask[0] |= best << i;
- /* Evaluate the second possibility for this state. */
- rate0 = tokens[next][0].rate;
- rate1 = tokens[next][1].rate;
-
- if((abs(x)*dequant_ptr[rc]>abs(coeff_ptr[rc])) &&
- (abs(x)*dequant_ptr[rc]<abs(coeff_ptr[rc])+dequant_ptr[rc]))
- shortcut = 1;
- else
- shortcut = 0;
-
- if(shortcut)
- {
- sz = -(x < 0);
- x -= 2*sz + 1;
- }
-
- /* Consider both possible successor states. */
- if (!x)
- {
- /* If we reduced this coefficient to zero, check to see if
- * we need to move the EOB back here.
- */
- t0 = tokens[next][0].token == DCT_EOB_TOKEN ?
- DCT_EOB_TOKEN : ZERO_TOKEN;
- t1 = tokens[next][1].token == DCT_EOB_TOKEN ?
- DCT_EOB_TOKEN : ZERO_TOKEN;
- }
- else
- {
- t0=t1 = (vp8_dct_value_tokens_ptr + x)->Token;
- }
- if (next < 16)
- {
- band = vp8_coef_bands[i + 1];
- if(t0!=DCT_EOB_TOKEN)
- {
- pt = vp8_prev_token_class[t0];
- rate0 += mb->token_costs[type][band][pt][
- tokens[next][0].token];
- }
- if(t1!=DCT_EOB_TOKEN)
- {
- pt = vp8_prev_token_class[t1];
- rate1 += mb->token_costs[type][band][pt][
- tokens[next][1].token];
- }
- }
-
- rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0);
- rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1);
- if (rd_cost0 == rd_cost1)
- {
- rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0);
- rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1);
- }
- /* And pick the best. */
- best = rd_cost1 < rd_cost0;
- base_bits = *(vp8_dct_value_cost_ptr + x);
-
- if(shortcut)
- {
- dx -= (dequant_ptr[rc] + sz) ^ sz;
- d2 = dx*dx;
- }
- tokens[i][1].rate = base_bits + (best ? rate1 : rate0);
- tokens[i][1].error = d2 + (best ? error1 : error0);
- tokens[i][1].next = next;
- tokens[i][1].token =best?t1:t0;
- tokens[i][1].qc = x;
- best_mask[1] |= best << i;
- /* Finally, make this the new head of the trellis. */
- next = i;
- }
- /* There's no choice to make for a zero coefficient, so we don't
- * add a new trellis node, but we do need to update the costs.
- */
- else
- {
- band = vp8_coef_bands[i + 1];
- t0 = tokens[next][0].token;
- t1 = tokens[next][1].token;
- /* Update the cost of each path if we're past the EOB token. */
- if (t0 != DCT_EOB_TOKEN)
- {
- tokens[next][0].rate += mb->token_costs[type][band][0][t0];
- tokens[next][0].token = ZERO_TOKEN;
- }
- if (t1 != DCT_EOB_TOKEN)
- {
- tokens[next][1].rate += mb->token_costs[type][band][0][t1];
- tokens[next][1].token = ZERO_TOKEN;
- }
- /* Don't update next, because we didn't add a new node. */
- }
- }
-
- /* Now pick the best path through the whole trellis. */
- band = vp8_coef_bands[i + 1];
- VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
- rate0 = tokens[next][0].rate;
- rate1 = tokens[next][1].rate;
- error0 = tokens[next][0].error;
- error1 = tokens[next][1].error;
- t0 = tokens[next][0].token;
- t1 = tokens[next][1].token;
- rate0 += mb->token_costs[type][band][pt][t0];
- rate1 += mb->token_costs[type][band][pt][t1];
- rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0);
- rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1);
- if (rd_cost0 == rd_cost1)
- {
- rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0);
- rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1);
- }
- best = rd_cost1 < rd_cost0;
- final_eob = i0 - 1;
- for (i = next; i < eob; i = next)
- {
- x = tokens[i][best].qc;
- if (x)
- final_eob = i;
- rc = vp8_default_zig_zag1d[i];
- qcoeff_ptr[rc] = x;
- dqcoeff_ptr[rc] = x * dequant_ptr[rc];
- next = tokens[i][best].next;
- best = (best_mask[best] >> i) & 1;
- }
- final_eob++;
-
- d->eob = final_eob;
- *a = *l = (d->eob != !type);
-}
-
-static void optimize_mb(MACROBLOCK *x, const VP8_ENCODER_RTCD *rtcd)
-{
- int b;
- int type;
- int has_2nd_order;
- ENTROPY_CONTEXT_PLANES t_above, t_left;
- ENTROPY_CONTEXT *ta;
- ENTROPY_CONTEXT *tl;
-
- vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
-
- ta = (ENTROPY_CONTEXT *)&t_above;
- tl = (ENTROPY_CONTEXT *)&t_left;
-
- has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED
- && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
- type = has_2nd_order ? PLANE_TYPE_Y_NO_DC : PLANE_TYPE_Y_WITH_DC;
-
- for (b = 0; b < 16; b++)
- {
- optimize_b(x, b, type,
- ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd);
- }
-
- for (b = 16; b < 24; b++)
- {
- optimize_b(x, b, PLANE_TYPE_UV,
- ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd);
- }
-
- if (has_2nd_order)
- {
- b=24;
- optimize_b(x, b, PLANE_TYPE_Y2,
- ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd);
- }
-}
-
-
-void vp8_optimize_mby(MACROBLOCK *x, const VP8_ENCODER_RTCD *rtcd)
-{
- int b;
- int type;
- int has_2nd_order;
-
- ENTROPY_CONTEXT_PLANES t_above, t_left;
- ENTROPY_CONTEXT *ta;
- ENTROPY_CONTEXT *tl;
-
- if (!x->e_mbd.above_context)
- return;
-
- if (!x->e_mbd.left_context)
- return;
-
- vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
-
- ta = (ENTROPY_CONTEXT *)&t_above;
- tl = (ENTROPY_CONTEXT *)&t_left;
-
- has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED
- && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
- type = has_2nd_order ? PLANE_TYPE_Y_NO_DC : PLANE_TYPE_Y_WITH_DC;
-
- for (b = 0; b < 16; b++)
- {
- optimize_b(x, b, type,
- ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd);
- }
-
-
- if (has_2nd_order)
- {
- b=24;
- optimize_b(x, b, PLANE_TYPE_Y2,
- ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd);
- }
-}
-
-void vp8_optimize_mbuv(MACROBLOCK *x, const VP8_ENCODER_RTCD *rtcd)
-{
- int b;
- ENTROPY_CONTEXT_PLANES t_above, t_left;
- ENTROPY_CONTEXT *ta;
- ENTROPY_CONTEXT *tl;
-
- if (!x->e_mbd.above_context)
- return;
-
- if (!x->e_mbd.left_context)
- return;
-
- vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
-
- ta = (ENTROPY_CONTEXT *)&t_above;
- tl = (ENTROPY_CONTEXT *)&t_left;
-
- for (b = 16; b < 24; b++)
- {
- optimize_b(x, b, PLANE_TYPE_UV,
- ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd);
- }
-}
-#endif
-
-void vp8_encode_inter16x16(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
-{
- vp8_build_inter_predictors_mb(&x->e_mbd);
-
- vp8_subtract_mb(rtcd, x);
-
- transform_mb(x);
-
- vp8_quantize_mb(x);
-
-#if !(CONFIG_REALTIME_ONLY)
- if (x->optimize)
- optimize_mb(x, rtcd);
-#endif
-
- vp8_inverse_transform_mb(IF_RTCD(&rtcd->common->idct), &x->e_mbd);
-
- RECON_INVOKE(&rtcd->common->recon, recon_mb)
- (IF_RTCD(&rtcd->common->recon), &x->e_mbd);
-}
-
-
-/* this funciton is used by first pass only */
-void vp8_encode_inter16x16y(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
-{
- vp8_build_inter_predictors_mby(&x->e_mbd);
-
- ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, x->src.y_buffer, x->e_mbd.predictor, x->src.y_stride);
-
- transform_mby(x);
-
- vp8_quantize_mby(x);
-
- vp8_inverse_transform_mby(IF_RTCD(&rtcd->common->idct), &x->e_mbd);
-
- RECON_INVOKE(&rtcd->common->recon, recon_mby)
- (IF_RTCD(&rtcd->common->recon), &x->e_mbd);
-}
-
-
-void vp8_encode_inter16x16uvrd(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
-{
- vp8_build_inter_predictors_mbuv(&x->e_mbd);
- ENCODEMB_INVOKE(&rtcd->encodemb, submbuv)(x->src_diff, x->src.u_buffer, x->src.v_buffer, x->e_mbd.predictor, x->src.uv_stride);
-
- vp8_transform_mbuv(x);
-
- vp8_quantize_mbuv(x);
-
-}
diff --git a/vp8/encoder/encodemb.h b/vp8/encoder/encodemb.h
deleted file mode 100644
index 8c93aa1..0000000
--- a/vp8/encoder/encodemb.h
+++ /dev/null
@@ -1,108 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_ENCODEMB_H
-#define __INC_ENCODEMB_H
-
-#include "vpx_ports/config.h"
-#include "block.h"
-
-#define prototype_mberr(sym) \
- int (sym)(MACROBLOCK *mb, int dc)
-
-#define prototype_berr(sym) \
- int (sym)(short *coeff, short *dqcoeff)
-
-#define prototype_mbuverr(sym) \
- int (sym)(MACROBLOCK *mb)
-
-#define prototype_subb(sym) \
- void (sym)(BLOCK *be,BLOCKD *bd, int pitch)
-
-#define prototype_submby(sym) \
- void (sym)(short *diff, unsigned char *src, unsigned char *pred, int stride)
-
-#define prototype_submbuv(sym) \
- void (sym)(short *diff, unsigned char *usrc, unsigned char *vsrc,\
- unsigned char *pred, int stride)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/encodemb_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/encodemb_arm.h"
-#endif
-
-#ifndef vp8_encodemb_berr
-#define vp8_encodemb_berr vp8_block_error_c
-#endif
-extern prototype_berr(vp8_encodemb_berr);
-
-#ifndef vp8_encodemb_mberr
-#define vp8_encodemb_mberr vp8_mbblock_error_c
-#endif
-extern prototype_mberr(vp8_encodemb_mberr);
-
-#ifndef vp8_encodemb_mbuverr
-#define vp8_encodemb_mbuverr vp8_mbuverror_c
-#endif
-extern prototype_mbuverr(vp8_encodemb_mbuverr);
-
-#ifndef vp8_encodemb_subb
-#define vp8_encodemb_subb vp8_subtract_b_c
-#endif
-extern prototype_subb(vp8_encodemb_subb);
-
-#ifndef vp8_encodemb_submby
-#define vp8_encodemb_submby vp8_subtract_mby_c
-#endif
-extern prototype_submby(vp8_encodemb_submby);
-
-#ifndef vp8_encodemb_submbuv
-#define vp8_encodemb_submbuv vp8_subtract_mbuv_c
-#endif
-extern prototype_submbuv(vp8_encodemb_submbuv);
-
-
-typedef struct
-{
- prototype_berr(*berr);
- prototype_mberr(*mberr);
- prototype_mbuverr(*mbuverr);
- prototype_subb(*subb);
- prototype_submby(*submby);
- prototype_submbuv(*submbuv);
-} vp8_encodemb_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define ENCODEMB_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define ENCODEMB_INVOKE(ctx,fn) vp8_encodemb_##fn
-#endif
-
-
-
-#include "onyx_int.h"
-struct VP8_ENCODER_RTCD;
-void vp8_encode_inter16x16(const struct VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x);
-
-extern void vp8_stuff_inter16x16(MACROBLOCK *x);
-
-void vp8_build_dcblock(MACROBLOCK *b);
-void vp8_transform_mb(MACROBLOCK *mb);
-void vp8_transform_mbuv(MACROBLOCK *x);
-void vp8_transform_intra_mby(MACROBLOCK *x);
-void vp8_encode_inter16x16uvrd(const struct VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x);
-void vp8_optimize_mby(MACROBLOCK *x, const struct VP8_ENCODER_RTCD *rtcd);
-void vp8_optimize_mbuv(MACROBLOCK *x, const struct VP8_ENCODER_RTCD *rtcd);
-void vp8_encode_inter16x16y(const struct VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x);
-#endif
diff --git a/vp8/encoder/encodemv.c b/vp8/encoder/encodemv.c
deleted file mode 100644
index 6b1e6f9..0000000
--- a/vp8/encoder/encodemv.c
+++ /dev/null
@@ -1,444 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/common.h"
-#include "encodemv.h"
-#include "vp8/common/entropymode.h"
-#include "vp8/common/systemdependent.h"
-
-#include <math.h>
-
-#ifdef ENTROPY_STATS
-extern unsigned int active_section;
-#endif
-
-static void encode_mvcomponent(
- vp8_writer *const w,
- const int v,
- const struct mv_context *mvc
-)
-{
- const vp8_prob *p = mvc->prob;
- const int x = v < 0 ? -v : v;
-
- if (x < mvnum_short) // Small
- {
- vp8_write(w, 0, p [mvpis_short]);
- vp8_treed_write(w, vp8_small_mvtree, p + MVPshort, x, 3);
-
- if (!x)
- return; // no sign bit
- }
- else // Large
- {
- int i = 0;
-
- vp8_write(w, 1, p [mvpis_short]);
-
- do
- vp8_write(w, (x >> i) & 1, p [MVPbits + i]);
-
- while (++i < 3);
-
- i = mvlong_width - 1; /* Skip bit 3, which is sometimes implicit */
-
- do
- vp8_write(w, (x >> i) & 1, p [MVPbits + i]);
-
- while (--i > 3);
-
- if (x & 0xFFF0)
- vp8_write(w, (x >> 3) & 1, p [MVPbits + 3]);
- }
-
- vp8_write(w, v < 0, p [MVPsign]);
-}
-#if 0
-static int max_mv_r = 0;
-static int max_mv_c = 0;
-#endif
-void vp8_encode_motion_vector(vp8_writer *w, const MV *mv, const MV_CONTEXT *mvc)
-{
-
-#if 0
- {
- if (abs(mv->row >> 1) > max_mv_r)
- {
- FILE *f = fopen("maxmv.stt", "a");
- max_mv_r = abs(mv->row >> 1);
- fprintf(f, "New Mv Row Max %6d\n", (mv->row >> 1));
-
- if ((abs(mv->row) / 2) != max_mv_r)
- fprintf(f, "MV Row conversion error %6d\n", abs(mv->row) / 2);
-
- fclose(f);
- }
-
- if (abs(mv->col >> 1) > max_mv_c)
- {
- FILE *f = fopen("maxmv.stt", "a");
- fprintf(f, "New Mv Col Max %6d\n", (mv->col >> 1));
- max_mv_c = abs(mv->col >> 1);
- fclose(f);
- }
- }
-#endif
-
- encode_mvcomponent(w, mv->row >> 1, &mvc[0]);
- encode_mvcomponent(w, mv->col >> 1, &mvc[1]);
-}
-
-
-static unsigned int cost_mvcomponent(const int v, const struct mv_context *mvc)
-{
- const vp8_prob *p = mvc->prob;
- const int x = v; //v<0? -v:v;
- unsigned int cost;
-
- if (x < mvnum_short)
- {
- cost = vp8_cost_zero(p [mvpis_short])
- + vp8_treed_cost(vp8_small_mvtree, p + MVPshort, x, 3);
-
- if (!x)
- return cost;
- }
- else
- {
- int i = 0;
- cost = vp8_cost_one(p [mvpis_short]);
-
- do
- cost += vp8_cost_bit(p [MVPbits + i], (x >> i) & 1);
-
- while (++i < 3);
-
- i = mvlong_width - 1; /* Skip bit 3, which is sometimes implicit */
-
- do
- cost += vp8_cost_bit(p [MVPbits + i], (x >> i) & 1);
-
- while (--i > 3);
-
- if (x & 0xFFF0)
- cost += vp8_cost_bit(p [MVPbits + 3], (x >> 3) & 1);
- }
-
- return cost; // + vp8_cost_bit( p [MVPsign], v < 0);
-}
-//#define M_LOG2_E 0.693147180559945309417
-//#define log2f(x) (log (x) / (float) M_LOG2_E)
-
-void vp8_build_component_cost_table(int *mvcost[2], int *mvsadcost[2], const MV_CONTEXT *mvc, int mvc_flag[2])
-{
- int i = 1; //-mv_max;
- unsigned int cost0 = 0;
- unsigned int cost1 = 0;
-
- vp8_clear_system_state();
-#if 0
- mvsadcost [0] [0] = 300;
- mvsadcost [1] [0] = 300;
-
- do
- {
- double z = 256 * (2 * (log2f(2 * i) + .6));
- mvsadcost [0][i] = (int) z;
- mvsadcost [1][i] = (int) z;
- mvsadcost [0][-i] = (int) z;
- mvsadcost [1][-i] = (int) z;
- }
- while (++i <= mv_max);
-
-#endif
-
- i = 1;
-
- if (mvc_flag[0])
- {
- mvcost [0] [0] = cost_mvcomponent(0, &mvc[0]);
-
- do
- {
- //mvcost [0] [i] = cost_mvcomponent( i, &mvc[0]);
- cost0 = cost_mvcomponent(i, &mvc[0]);
-
- mvcost [0] [i] = cost0 + vp8_cost_zero(mvc[0].prob[MVPsign]);
- mvcost [0] [-i] = cost0 + vp8_cost_one(mvc[0].prob[MVPsign]);
- }
- while (++i <= mv_max);
- }
-
- i = 1;
-
- if (mvc_flag[1])
- {
- mvcost [1] [0] = cost_mvcomponent(0, &mvc[1]);
-
- do
- {
- //mvcost [1] [i] = cost_mvcomponent( i, mvc[1]);
- cost1 = cost_mvcomponent(i, &mvc[1]);
-
- mvcost [1] [i] = cost1 + vp8_cost_zero(mvc[1].prob[MVPsign]);
- mvcost [1] [-i] = cost1 + vp8_cost_one(mvc[1].prob[MVPsign]);
- }
- while (++i <= mv_max);
- }
-
- /*
- i=-mv_max;
- do
- {
- mvcost [0] [i] = cost_mvcomponent( i, mvc[0]);
- mvcost [1] [i] = cost_mvcomponent( i, mvc[1]);
- }
- while( ++i <= mv_max);
- */
-}
-
-
-// Motion vector probability table update depends on benefit.
-// Small correction allows for the fact that an update to an MV probability
-// may have benefit in subsequent frames as well as the current one.
-
-#define MV_PROB_UPDATE_CORRECTION -1
-
-
-__inline static void calc_prob(vp8_prob *p, const unsigned int ct[2])
-{
- const unsigned int tot = ct[0] + ct[1];
-
- if (tot)
- {
- const vp8_prob x = ((ct[0] * 255) / tot) & -2;
- *p = x ? x : 1;
- }
-}
-
-static void update(
- vp8_writer *const w,
- const unsigned int ct[2],
- vp8_prob *const cur_p,
- const vp8_prob new_p,
- const vp8_prob update_p,
- int *updated
-)
-{
- const int cur_b = vp8_cost_branch(ct, *cur_p);
- const int new_b = vp8_cost_branch(ct, new_p);
- const int cost = 7 + MV_PROB_UPDATE_CORRECTION + ((vp8_cost_one(update_p) - vp8_cost_zero(update_p) + 128) >> 8);
-
- if (cur_b - new_b > cost)
- {
- *cur_p = new_p;
- vp8_write(w, 1, update_p);
- vp8_write_literal(w, new_p >> 1, 7);
- *updated = 1;
-
- }
- else
- vp8_write(w, 0, update_p);
-}
-
-static void write_component_probs(
- vp8_writer *const w,
- struct mv_context *cur_mvc,
- const struct mv_context *default_mvc_,
- const struct mv_context *update_mvc,
- const unsigned int events [MVvals],
- unsigned int rc,
- int *updated
-)
-{
- vp8_prob *Pcur = cur_mvc->prob;
- const vp8_prob *default_mvc = default_mvc_->prob;
- const vp8_prob *Pupdate = update_mvc->prob;
- unsigned int is_short_ct[2], sign_ct[2];
-
- unsigned int bit_ct [mvlong_width] [2];
-
- unsigned int short_ct [mvnum_short];
- unsigned int short_bct [mvnum_short-1] [2];
-
- vp8_prob Pnew [MVPcount];
-
- (void) rc;
- vp8_copy_array(Pnew, default_mvc, MVPcount);
-
- vp8_zero(is_short_ct)
- vp8_zero(sign_ct)
- vp8_zero(bit_ct)
- vp8_zero(short_ct)
- vp8_zero(short_bct)
-
-
- //j=0
- {
- const int c = events [mv_max];
-
- is_short_ct [0] += c; // Short vector
- short_ct [0] += c; // Magnitude distribution
- }
-
- //j: 1 ~ mv_max (1023)
- {
- int j = 1;
-
- do
- {
- const int c1 = events [mv_max + j]; //positive
- const int c2 = events [mv_max - j]; //negative
- const int c = c1 + c2;
- int a = j;
-
- sign_ct [0] += c1;
- sign_ct [1] += c2;
-
- if (a < mvnum_short)
- {
- is_short_ct [0] += c; // Short vector
- short_ct [a] += c; // Magnitude distribution
- }
- else
- {
- int k = mvlong_width - 1;
- is_short_ct [1] += c; // Long vector
-
- /* bit 3 not always encoded. */
- do
- bit_ct [k] [(a >> k) & 1] += c;
-
- while (--k >= 0);
- }
- }
- while (++j <= mv_max);
- }
-
- /*
- {
- int j = -mv_max;
- do
- {
-
- const int c = events [mv_max + j];
- int a = j;
-
- if( j < 0)
- {
- sign_ct [1] += c;
- a = -j;
- }
- else if( j)
- sign_ct [0] += c;
-
- if( a < mvnum_short)
- {
- is_short_ct [0] += c; // Short vector
- short_ct [a] += c; // Magnitude distribution
- }
- else
- {
- int k = mvlong_width - 1;
- is_short_ct [1] += c; // Long vector
-
- // bit 3 not always encoded.
-
- do
- bit_ct [k] [(a >> k) & 1] += c;
- while( --k >= 0);
- }
- } while( ++j <= mv_max);
- }
- */
-
- calc_prob(Pnew + mvpis_short, is_short_ct);
-
- calc_prob(Pnew + MVPsign, sign_ct);
-
- {
- vp8_prob p [mvnum_short - 1]; /* actually only need branch ct */
- int j = 0;
-
- vp8_tree_probs_from_distribution(
- 8, vp8_small_mvencodings, vp8_small_mvtree,
- p, short_bct, short_ct,
- 256, 1
- );
-
- do
- calc_prob(Pnew + MVPshort + j, short_bct[j]);
-
- while (++j < mvnum_short - 1);
- }
-
- {
- int j = 0;
-
- do
- calc_prob(Pnew + MVPbits + j, bit_ct[j]);
-
- while (++j < mvlong_width);
- }
-
- update(w, is_short_ct, Pcur + mvpis_short, Pnew[mvpis_short], *Pupdate++, updated);
-
- update(w, sign_ct, Pcur + MVPsign, Pnew[MVPsign], *Pupdate++, updated);
-
- {
- const vp8_prob *const new_p = Pnew + MVPshort;
- vp8_prob *const cur_p = Pcur + MVPshort;
-
- int j = 0;
-
- do
-
- update(w, short_bct[j], cur_p + j, new_p[j], *Pupdate++, updated);
-
- while (++j < mvnum_short - 1);
- }
-
- {
- const vp8_prob *const new_p = Pnew + MVPbits;
- vp8_prob *const cur_p = Pcur + MVPbits;
-
- int j = 0;
-
- do
-
- update(w, bit_ct[j], cur_p + j, new_p[j], *Pupdate++, updated);
-
- while (++j < mvlong_width);
- }
-}
-
-void vp8_write_mvprobs(VP8_COMP *cpi)
-{
- vp8_writer *const w = & cpi->bc;
- MV_CONTEXT *mvc = cpi->common.fc.mvc;
- int flags[2] = {0, 0};
-#ifdef ENTROPY_STATS
- active_section = 4;
-#endif
- write_component_probs(
- w, &mvc[0], &vp8_default_mv_context[0], &vp8_mv_update_probs[0], cpi->MVcount[0], 0, &flags[0]
- );
- write_component_probs(
- w, &mvc[1], &vp8_default_mv_context[1], &vp8_mv_update_probs[1], cpi->MVcount[1], 1, &flags[1]
- );
-
- if (flags[0] || flags[1])
- vp8_build_component_cost_table(cpi->mb.mvcost, cpi->mb.mvsadcost, (const MV_CONTEXT *) cpi->common.fc.mvc, flags);
-
-#ifdef ENTROPY_STATS
- active_section = 5;
-#endif
-}
diff --git a/vp8/encoder/encodemv.h b/vp8/encoder/encodemv.h
deleted file mode 100644
index e4481bf..0000000
--- a/vp8/encoder/encodemv.h
+++ /dev/null
@@ -1,21 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_ENCODEMV_H
-#define __INC_ENCODEMV_H
-
-#include "onyx_int.h"
-
-void vp8_write_mvprobs(VP8_COMP *);
-void vp8_encode_motion_vector(vp8_writer *, const MV *, const MV_CONTEXT *);
-void vp8_build_component_cost_table(int *mvcost[2], int *mvsadcost[2], const MV_CONTEXT *mvc, int mvc_flag[2]);
-
-#endif
diff --git a/vp8/encoder/ethreading.c b/vp8/encoder/ethreading.c
deleted file mode 100644
index 71da103..0000000
--- a/vp8/encoder/ethreading.c
+++ /dev/null
@@ -1,560 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "onyx_int.h"
-#include "vp8/common/threading.h"
-#include "vp8/common/common.h"
-#include "vp8/common/extend.h"
-
-#if CONFIG_MULTITHREAD
-
-extern int vp8cx_encode_inter_macroblock(VP8_COMP *cpi, MACROBLOCK *x,
- TOKENEXTRA **t, int recon_yoffset,
- int recon_uvoffset);
-extern int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x,
- TOKENEXTRA **t);
-extern void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x);
-extern void vp8_build_block_offsets(MACROBLOCK *x);
-extern void vp8_setup_block_ptrs(MACROBLOCK *x);
-
-extern void loopfilter_frame(VP8_COMP *cpi, VP8_COMMON *cm);
-
-static THREAD_FUNCTION loopfilter_thread(void *p_data)
-{
- VP8_COMP *cpi = (VP8_COMP *)(((LPFTHREAD_DATA *)p_data)->ptr1);
- VP8_COMMON *cm = &cpi->common;
-
- while (1)
- {
- if (cpi->b_multi_threaded == 0)
- break;
-
- if (sem_wait(&cpi->h_event_start_lpf) == 0)
- {
- if (cpi->b_multi_threaded == FALSE) // we're shutting down
- break;
-
- loopfilter_frame(cpi, cm);
-
- sem_post(&cpi->h_event_end_lpf);
- }
- }
-
- return 0;
-}
-
-static
-THREAD_FUNCTION thread_encoding_proc(void *p_data)
-{
- int ithread = ((ENCODETHREAD_DATA *)p_data)->ithread;
- VP8_COMP *cpi = (VP8_COMP *)(((ENCODETHREAD_DATA *)p_data)->ptr1);
- MB_ROW_COMP *mbri = (MB_ROW_COMP *)(((ENCODETHREAD_DATA *)p_data)->ptr2);
- ENTROPY_CONTEXT_PLANES mb_row_left_context;
-
- const int nsync = cpi->mt_sync_range;
- //printf("Started thread %d\n", ithread);
-
- while (1)
- {
- if (cpi->b_multi_threaded == 0)
- break;
-
- //if(WaitForSingleObject(cpi->h_event_mbrencoding[ithread], INFINITE) == WAIT_OBJECT_0)
- if (sem_wait(&cpi->h_event_start_encoding[ithread]) == 0)
- {
- VP8_COMMON *cm = &cpi->common;
- int mb_row;
- MACROBLOCK *x = &mbri->mb;
- MACROBLOCKD *xd = &x->e_mbd;
- TOKENEXTRA *tp ;
-
- int *segment_counts = mbri->segment_counts;
- int *totalrate = &mbri->totalrate;
-
- if (cpi->b_multi_threaded == FALSE) // we're shutting down
- break;
-
- for (mb_row = ithread + 1; mb_row < cm->mb_rows; mb_row += (cpi->encoding_thread_count + 1))
- {
-
- int i;
- int recon_yoffset, recon_uvoffset;
- int mb_col;
- int ref_fb_idx = cm->lst_fb_idx;
- int dst_fb_idx = cm->new_fb_idx;
- int recon_y_stride = cm->yv12_fb[ref_fb_idx].y_stride;
- int recon_uv_stride = cm->yv12_fb[ref_fb_idx].uv_stride;
- volatile int *last_row_current_mb_col;
- INT64 activity_sum = 0;
-
- tp = cpi->tok + (mb_row * (cm->mb_cols * 16 * 24));
-
- last_row_current_mb_col = &cpi->mt_current_mb_col[mb_row - 1];
-
- // reset above block coeffs
- xd->above_context = cm->above_context;
- xd->left_context = &mb_row_left_context;
-
- vp8_zero(mb_row_left_context);
-
- xd->up_available = (mb_row != 0);
- recon_yoffset = (mb_row * recon_y_stride * 16);
- recon_uvoffset = (mb_row * recon_uv_stride * 8);
-
- cpi->tplist[mb_row].start = tp;
-
- //printf("Thread mb_row = %d\n", mb_row);
-
- // for each macroblock col in image
- for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
- {
- int seg_map_index = (mb_row * cm->mb_cols);
-
- if ((mb_col & (nsync - 1)) == 0)
- {
- while (mb_col > (*last_row_current_mb_col - nsync) && *last_row_current_mb_col != cm->mb_cols - 1)
- {
- x86_pause_hint();
- thread_sleep(0);
- }
- }
-
- // Distance of Mb to the various image edges.
- // These specified to 8th pel as they are always compared to values that are in 1/8th pel units
- xd->mb_to_left_edge = -((mb_col * 16) << 3);
- xd->mb_to_right_edge = ((cm->mb_cols - 1 - mb_col) * 16) << 3;
- xd->mb_to_top_edge = -((mb_row * 16) << 3);
- xd->mb_to_bottom_edge = ((cm->mb_rows - 1 - mb_row) * 16) << 3;
-
- // Set up limit values for motion vectors used to prevent them extending outside the UMV borders
- x->mv_col_min = -((mb_col * 16) + (VP8BORDERINPIXELS - 16));
- x->mv_col_max = ((cm->mb_cols - 1 - mb_col) * 16) + (VP8BORDERINPIXELS - 16);
- x->mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 16));
- x->mv_row_max = ((cm->mb_rows - 1 - mb_row) * 16) + (VP8BORDERINPIXELS - 16);
-
- xd->dst.y_buffer = cm->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset;
- xd->dst.u_buffer = cm->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset;
- xd->dst.v_buffer = cm->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset;
- xd->left_available = (mb_col != 0);
-
- x->rddiv = cpi->RDDIV;
- x->rdmult = cpi->RDMULT;
-
- if (cpi->oxcf.tuning == VP8_TUNE_SSIM)
- activity_sum += vp8_activity_masking(cpi, x);
-
- // Is segmentation enabled
- // MB level adjutment to quantizer
- if (xd->segmentation_enabled)
- {
- // Code to set segment id in xd->mbmi.segment_id for current MB (with range checking)
- if (cpi->segmentation_map[seg_map_index + mb_col] <= 3)
- xd->mode_info_context->mbmi.segment_id = cpi->segmentation_map[seg_map_index + mb_col];
- else
- xd->mode_info_context->mbmi.segment_id = 0;
-
- vp8cx_mb_init_quantizer(cpi, x);
- }
- else
- xd->mode_info_context->mbmi.segment_id = 0; // Set to Segment 0 by default
-
- x->active_ptr = cpi->active_map + seg_map_index + mb_col;
-
- if (cm->frame_type == KEY_FRAME)
- {
- *totalrate += vp8cx_encode_intra_macro_block(cpi, x, &tp);
-#ifdef MODE_STATS
- y_modes[xd->mbmi.mode] ++;
-#endif
- }
- else
- {
- *totalrate += vp8cx_encode_inter_macroblock(cpi, x, &tp, recon_yoffset, recon_uvoffset);
-
-#ifdef MODE_STATS
- inter_y_modes[xd->mbmi.mode] ++;
-
- if (xd->mbmi.mode == SPLITMV)
- {
- int b;
-
- for (b = 0; b < xd->mbmi.partition_count; b++)
- {
- inter_b_modes[x->partition->bmi[b].mode] ++;
- }
- }
-
-#endif
-
- // Count of last ref frame 0,0 useage
- if ((xd->mode_info_context->mbmi.mode == ZEROMV) && (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME))
- cpi->inter_zz_count++;
-
- // Special case code for cyclic refresh
- // If cyclic update enabled then copy xd->mbmi.segment_id; (which may have been updated based on mode
- // during vp8cx_encode_inter_macroblock()) back into the global sgmentation map
- if (cpi->cyclic_refresh_mode_enabled && xd->segmentation_enabled)
- {
- const MB_MODE_INFO * mbmi = &xd->mode_info_context->mbmi;
- cpi->segmentation_map[seg_map_index + mb_col] = mbmi->segment_id;
-
- // If the block has been refreshed mark it as clean (the magnitude of the -ve influences how long it will be before we consider another refresh):
- // Else if it was coded (last frame 0,0) and has not already been refreshed then mark it as a candidate for cleanup next time (marked 0)
- // else mark it as dirty (1).
- if (mbmi->segment_id)
- cpi->cyclic_refresh_map[seg_map_index + mb_col] = -1;
- else if ((mbmi->mode == ZEROMV) && (mbmi->ref_frame == LAST_FRAME))
- {
- if (cpi->cyclic_refresh_map[seg_map_index + mb_col] == 1)
- cpi->cyclic_refresh_map[seg_map_index + mb_col] = 0;
- }
- else
- cpi->cyclic_refresh_map[seg_map_index + mb_col] = 1;
-
- }
- }
- cpi->tplist[mb_row].stop = tp;
-
- x->gf_active_ptr++; // Increment pointer into gf useage flags structure for next mb
-
- for (i = 0; i < 16; i++)
- vpx_memcpy(&xd->mode_info_context->bmi[i], &xd->block[i].bmi, sizeof(xd->block[i].bmi));
-
- // adjust to the next column of macroblocks
- x->src.y_buffer += 16;
- x->src.u_buffer += 8;
- x->src.v_buffer += 8;
-
- recon_yoffset += 16;
- recon_uvoffset += 8;
-
- // Keep track of segment useage
- segment_counts[xd->mode_info_context->mbmi.segment_id]++;
-
- // skip to next mb
- xd->mode_info_context++;
- x->partition_info++;
- xd->above_context++;
-
- cpi->mt_current_mb_col[mb_row] = mb_col;
- }
-
- //extend the recon for intra prediction
- vp8_extend_mb_row(
- &cm->yv12_fb[dst_fb_idx],
- xd->dst.y_buffer + 16,
- xd->dst.u_buffer + 8,
- xd->dst.v_buffer + 8);
-
- // this is to account for the border
- xd->mode_info_context++;
- x->partition_info++;
- x->activity_sum += activity_sum;
-
- x->src.y_buffer += 16 * x->src.y_stride * (cpi->encoding_thread_count + 1) - 16 * cm->mb_cols;
- x->src.u_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols;
- x->src.v_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols;
-
- xd->mode_info_context += xd->mode_info_stride * cpi->encoding_thread_count;
- x->partition_info += xd->mode_info_stride * cpi->encoding_thread_count;
-
- if (mb_row == cm->mb_rows - 1)
- {
- //SetEvent(cpi->h_event_main);
- sem_post(&cpi->h_event_end_encoding); /* signal frame encoding end */
- }
- }
- }
- }
-
- //printf("exit thread %d\n", ithread);
- return 0;
-}
-
-static void setup_mbby_copy(MACROBLOCK *mbdst, MACROBLOCK *mbsrc)
-{
-
- MACROBLOCK *x = mbsrc;
- MACROBLOCK *z = mbdst;
- int i;
-
- z->ss = x->ss;
- z->ss_count = x->ss_count;
- z->searches_per_step = x->searches_per_step;
- z->errorperbit = x->errorperbit;
-
- z->sadperbit16 = x->sadperbit16;
- z->sadperbit4 = x->sadperbit4;
- z->errthresh = x->errthresh;
-
- /*
- z->mv_col_min = x->mv_col_min;
- z->mv_col_max = x->mv_col_max;
- z->mv_row_min = x->mv_row_min;
- z->mv_row_max = x->mv_row_max;
- z->vector_range = x->vector_range ;
- */
-
- z->vp8_short_fdct4x4 = x->vp8_short_fdct4x4;
- z->vp8_short_fdct8x4 = x->vp8_short_fdct8x4;
- z->short_walsh4x4 = x->short_walsh4x4;
- z->quantize_b = x->quantize_b;
- z->optimize = x->optimize;
-
- /*
- z->mvc = x->mvc;
- z->src.y_buffer = x->src.y_buffer;
- z->src.u_buffer = x->src.u_buffer;
- z->src.v_buffer = x->src.v_buffer;
- */
-
-
- vpx_memcpy(z->mvcosts, x->mvcosts, sizeof(x->mvcosts));
- z->mvcost[0] = &z->mvcosts[0][mv_max+1];
- z->mvcost[1] = &z->mvcosts[1][mv_max+1];
- z->mvsadcost[0] = &z->mvsadcosts[0][mv_max+1];
- z->mvsadcost[1] = &z->mvsadcosts[1][mv_max+1];
-
-
- vpx_memcpy(z->token_costs, x->token_costs, sizeof(x->token_costs));
- vpx_memcpy(z->inter_bmode_costs, x->inter_bmode_costs, sizeof(x->inter_bmode_costs));
- //memcpy(z->mvcosts, x->mvcosts, sizeof(x->mvcosts));
- //memcpy(z->mvcost, x->mvcost, sizeof(x->mvcost));
- vpx_memcpy(z->mbmode_cost, x->mbmode_cost, sizeof(x->mbmode_cost));
- vpx_memcpy(z->intra_uv_mode_cost, x->intra_uv_mode_cost, sizeof(x->intra_uv_mode_cost));
- vpx_memcpy(z->bmode_costs, x->bmode_costs, sizeof(x->bmode_costs));
-
- for (i = 0; i < 25; i++)
- {
- z->block[i].quant = x->block[i].quant;
- z->block[i].quant_fast = x->block[i].quant_fast;
- z->block[i].quant_shift = x->block[i].quant_shift;
- z->block[i].zbin = x->block[i].zbin;
- z->block[i].zrun_zbin_boost = x->block[i].zrun_zbin_boost;
- z->block[i].round = x->block[i].round;
- /*
- z->block[i].src = x->block[i].src;
- */
- z->block[i].src_stride = x->block[i].src_stride;
- z->block[i].force_empty = x->block[i].force_empty;
-
- }
-
- {
- MACROBLOCKD *xd = &x->e_mbd;
- MACROBLOCKD *zd = &z->e_mbd;
-
- /*
- zd->mode_info_context = xd->mode_info_context;
- zd->mode_info = xd->mode_info;
-
- zd->mode_info_stride = xd->mode_info_stride;
- zd->frame_type = xd->frame_type;
- zd->up_available = xd->up_available ;
- zd->left_available = xd->left_available;
- zd->left_context = xd->left_context;
- zd->last_frame_dc = xd->last_frame_dc;
- zd->last_frame_dccons = xd->last_frame_dccons;
- zd->gold_frame_dc = xd->gold_frame_dc;
- zd->gold_frame_dccons = xd->gold_frame_dccons;
- zd->mb_to_left_edge = xd->mb_to_left_edge;
- zd->mb_to_right_edge = xd->mb_to_right_edge;
- zd->mb_to_top_edge = xd->mb_to_top_edge ;
- zd->mb_to_bottom_edge = xd->mb_to_bottom_edge;
- zd->gf_active_ptr = xd->gf_active_ptr;
- zd->frames_since_golden = xd->frames_since_golden;
- zd->frames_till_alt_ref_frame = xd->frames_till_alt_ref_frame;
- */
- zd->subpixel_predict = xd->subpixel_predict;
- zd->subpixel_predict8x4 = xd->subpixel_predict8x4;
- zd->subpixel_predict8x8 = xd->subpixel_predict8x8;
- zd->subpixel_predict16x16 = xd->subpixel_predict16x16;
- zd->segmentation_enabled = xd->segmentation_enabled;
- zd->mb_segement_abs_delta = xd->mb_segement_abs_delta;
- vpx_memcpy(zd->segment_feature_data, xd->segment_feature_data, sizeof(xd->segment_feature_data));
-
- for (i = 0; i < 25; i++)
- {
- zd->block[i].dequant = xd->block[i].dequant;
- }
- }
-}
-
-void vp8cx_init_mbrthread_data(VP8_COMP *cpi,
- MACROBLOCK *x,
- MB_ROW_COMP *mbr_ei,
- int mb_row,
- int count
- )
-{
-
- VP8_COMMON *const cm = & cpi->common;
- MACROBLOCKD *const xd = & x->e_mbd;
- int i;
- (void) mb_row;
-
- for (i = 0; i < count; i++)
- {
- MACROBLOCK *mb = & mbr_ei[i].mb;
- MACROBLOCKD *mbd = &mb->e_mbd;
-
- mbd->subpixel_predict = xd->subpixel_predict;
- mbd->subpixel_predict8x4 = xd->subpixel_predict8x4;
- mbd->subpixel_predict8x8 = xd->subpixel_predict8x8;
- mbd->subpixel_predict16x16 = xd->subpixel_predict16x16;
-#if CONFIG_RUNTIME_CPU_DETECT
- mbd->rtcd = xd->rtcd;
-#endif
- mb->gf_active_ptr = x->gf_active_ptr;
-
- mb->vector_range = 32;
-
- vpx_memset(mbr_ei[i].segment_counts, 0, sizeof(mbr_ei[i].segment_counts));
- mbr_ei[i].totalrate = 0;
-
- mb->partition_info = x->pi + x->e_mbd.mode_info_stride * (i + 1);
-
- mbd->mode_info_context = cm->mi + x->e_mbd.mode_info_stride * (i + 1);
- mbd->mode_info_stride = cm->mode_info_stride;
-
- mbd->frame_type = cm->frame_type;
-
- mbd->frames_since_golden = cm->frames_since_golden;
- mbd->frames_till_alt_ref_frame = cm->frames_till_alt_ref_frame;
-
- mb->src = * cpi->Source;
- mbd->pre = cm->yv12_fb[cm->lst_fb_idx];
- mbd->dst = cm->yv12_fb[cm->new_fb_idx];
-
- mb->src.y_buffer += 16 * x->src.y_stride * (i + 1);
- mb->src.u_buffer += 8 * x->src.uv_stride * (i + 1);
- mb->src.v_buffer += 8 * x->src.uv_stride * (i + 1);
-
- vp8_build_block_offsets(mb);
-
- vp8_setup_block_dptrs(mbd);
-
- vp8_setup_block_ptrs(mb);
-
- mb->activity_sum = 0;
-
- mbd->left_context = &cm->left_context;
- mb->mvc = cm->fc.mvc;
-
- setup_mbby_copy(&mbr_ei[i].mb, x);
-
- }
-}
-
-void vp8cx_create_encoder_threads(VP8_COMP *cpi)
-{
- const VP8_COMMON * cm = &cpi->common;
-
- cpi->b_multi_threaded = 0;
- cpi->encoding_thread_count = 0;
- cpi->processor_core_count = 32; //vp8_get_proc_core_count();
-
- if (cpi->processor_core_count > 1 && cpi->oxcf.multi_threaded > 1)
- {
- int ithread;
- int th_count = cpi->oxcf.multi_threaded - 1;
-
- if (cpi->oxcf.multi_threaded > cpi->processor_core_count)
- th_count = cpi->processor_core_count - 1;
-
- /* we have th_count + 1 (main) threads processing one row each */
- /* no point to have more threads than the sync range allows */
- if(th_count > ((cm->mb_cols / cpi->mt_sync_range) - 1))
- {
- th_count = (cm->mb_cols / cpi->mt_sync_range) - 1;
- }
-
- if(th_count == 0)
- return;
-
- CHECK_MEM_ERROR(cpi->h_encoding_thread, vpx_malloc(sizeof(pthread_t) * th_count));
- CHECK_MEM_ERROR(cpi->h_event_start_encoding, vpx_malloc(sizeof(sem_t) * th_count));
- CHECK_MEM_ERROR(cpi->mb_row_ei, vpx_memalign(32, sizeof(MB_ROW_COMP) * th_count));
- vpx_memset(cpi->mb_row_ei, 0, sizeof(MB_ROW_COMP) * th_count);
- CHECK_MEM_ERROR(cpi->en_thread_data,
- vpx_malloc(sizeof(ENCODETHREAD_DATA) * th_count));
- CHECK_MEM_ERROR(cpi->mt_current_mb_col,
- vpx_malloc(sizeof(*cpi->mt_current_mb_col) * cm->mb_rows));
-
- sem_init(&cpi->h_event_end_encoding, 0, 0);
-
- cpi->b_multi_threaded = 1;
- cpi->encoding_thread_count = th_count;
-
- /*
- printf("[VP8:] multi_threaded encoding is enabled with %d threads\n\n",
- (cpi->encoding_thread_count +1));
- */
-
- for (ithread = 0; ithread < th_count; ithread++)
- {
- ENCODETHREAD_DATA * ethd = &cpi->en_thread_data[ithread];
-
- sem_init(&cpi->h_event_start_encoding[ithread], 0, 0);
- ethd->ithread = ithread;
- ethd->ptr1 = (void *)cpi;
- ethd->ptr2 = (void *)&cpi->mb_row_ei[ithread];
-
- pthread_create(&cpi->h_encoding_thread[ithread], 0, thread_encoding_proc, ethd);
- }
-
- {
- LPFTHREAD_DATA * lpfthd = &cpi->lpf_thread_data;
-
- sem_init(&cpi->h_event_start_lpf, 0, 0);
- sem_init(&cpi->h_event_end_lpf, 0, 0);
-
- lpfthd->ptr1 = (void *)cpi;
- pthread_create(&cpi->h_filter_thread, 0, loopfilter_thread, lpfthd);
- }
- }
-
-}
-
-void vp8cx_remove_encoder_threads(VP8_COMP *cpi)
-{
- if (cpi->b_multi_threaded)
- {
- //shutdown other threads
- cpi->b_multi_threaded = 0;
- {
- int i;
-
- for (i = 0; i < cpi->encoding_thread_count; i++)
- {
- //SetEvent(cpi->h_event_mbrencoding[i]);
- sem_post(&cpi->h_event_start_encoding[i]);
- pthread_join(cpi->h_encoding_thread[i], 0);
-
- sem_destroy(&cpi->h_event_start_encoding[i]);
- }
-
- sem_post(&cpi->h_event_start_lpf);
- pthread_join(cpi->h_filter_thread, 0);
- }
-
- sem_destroy(&cpi->h_event_end_encoding);
- sem_destroy(&cpi->h_event_end_lpf);
- sem_destroy(&cpi->h_event_start_lpf);
-
- //free thread related resources
- vpx_free(cpi->h_event_start_encoding);
- vpx_free(cpi->h_encoding_thread);
- vpx_free(cpi->mb_row_ei);
- vpx_free(cpi->en_thread_data);
- vpx_free(cpi->mt_current_mb_col);
- }
-}
-#endif
diff --git a/vp8/encoder/firstpass.c b/vp8/encoder/firstpass.c
deleted file mode 100644
index 6c9433b..0000000
--- a/vp8/encoder/firstpass.c
+++ /dev/null
@@ -1,2851 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "math.h"
-#include "limits.h"
-#include "block.h"
-#include "onyx_int.h"
-#include "variance.h"
-#include "encodeintra.h"
-#include "vp8/common/setupintrarecon.h"
-#include "mcomp.h"
-#include "vpx_scale/vpxscale.h"
-#include "encodemb.h"
-#include "vp8/common/extend.h"
-#include "vp8/common/systemdependent.h"
-#include "vpx_scale/yv12extend.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vp8/common/swapyv12buffer.h"
-#include <stdio.h>
-#include "rdopt.h"
-#include "vp8/common/quant_common.h"
-#include "encodemv.h"
-
-//#define OUTPUT_FPF 1
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IF_RTCD(x) (x)
-#else
-#define IF_RTCD(x) NULL
-#endif
-
-extern void vp8_build_block_offsets(MACROBLOCK *x);
-extern void vp8_setup_block_ptrs(MACROBLOCK *x);
-extern void vp8cx_frame_init_quantizer(VP8_COMP *cpi);
-extern void vp8_set_mbmode_and_mvs(MACROBLOCK *x, MB_PREDICTION_MODE mb, MV *mv);
-extern void vp8_alloc_compressor_data(VP8_COMP *cpi);
-
-//#define GFQ_ADJUSTMENT (40 + ((15*Q)/10))
-//#define GFQ_ADJUSTMENT (80 + ((15*Q)/10))
-#define GFQ_ADJUSTMENT vp8_gf_boost_qadjustment[Q]
-extern int vp8_kf_boost_qadjustment[QINDEX_RANGE];
-
-extern const int vp8_gf_boost_qadjustment[QINDEX_RANGE];
-
-#define IIFACTOR 1.4
-#define IIKFACTOR1 1.40
-#define IIKFACTOR2 1.5
-#define RMAX 14.0
-#define GF_RMAX 48.0
-
-#define KF_MB_INTRA_MIN 300
-#define GF_MB_INTRA_MIN 200
-
-#define DOUBLE_DIVIDE_CHECK(X) ((X)<0?(X)-.000001:(X)+.000001)
-
-#define POW1 (double)cpi->oxcf.two_pass_vbrbias/100.0
-#define POW2 (double)cpi->oxcf.two_pass_vbrbias/100.0
-
-static int vscale_lookup[7] = {0, 1, 1, 2, 2, 3, 3};
-static int hscale_lookup[7] = {0, 0, 1, 1, 2, 2, 3};
-
-
-static const int cq_level[QINDEX_RANGE] =
-{
- 0,0,1,1,2,3,3,4,4,5,6,6,7,8,8,9,
- 9,10,11,11,12,13,13,14,15,15,16,17,17,18,19,20,
- 20,21,22,22,23,24,24,25,26,27,27,28,29,30,30,31,
- 32,33,33,34,35,36,36,37,38,39,39,40,41,42,42,43,
- 44,45,46,46,47,48,49,50,50,51,52,53,54,55,55,56,
- 57,58,59,60,60,61,62,63,64,65,66,67,67,68,69,70,
- 71,72,73,74,75,75,76,77,78,79,80,81,82,83,84,85,
- 86,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100
-};
-
-static void find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame);
-
-static int encode_intra(VP8_COMP *cpi, MACROBLOCK *x, int use_dc_pred)
-{
-
- int i;
- int intra_pred_var = 0;
- (void) cpi;
-
- if (use_dc_pred)
- {
- x->e_mbd.mode_info_context->mbmi.mode = DC_PRED;
- x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
- x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
-
- vp8_encode_intra16x16mby(IF_RTCD(&cpi->rtcd), x);
- }
- else
- {
- for (i = 0; i < 16; i++)
- {
- BLOCKD *b = &x->e_mbd.block[i];
- BLOCK *be = &x->block[i];
-
- vp8_encode_intra4x4block(IF_RTCD(&cpi->rtcd), x, be, b, B_DC_PRED);
- }
- }
-
- intra_pred_var = VARIANCE_INVOKE(&cpi->rtcd.variance, getmbss)(x->src_diff);
-
- return intra_pred_var;
-}
-
-// Resets the first pass file to the given position using a relative seek from the current position
-static void reset_fpf_position(VP8_COMP *cpi, FIRSTPASS_STATS *Position)
-{
- cpi->stats_in = Position;
-}
-
-static int lookup_next_frame_stats(VP8_COMP *cpi, FIRSTPASS_STATS *next_frame)
-{
- if (cpi->stats_in >= cpi->stats_in_end)
- return EOF;
-
- *next_frame = *cpi->stats_in;
- return 1;
-}
-
-// Calculate a modified Error used in distributing bits between easier and harder frames
-static double calculate_modified_err(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
-{
- double av_err = cpi->total_stats->ssim_weighted_pred_err;
- double this_err = this_frame->ssim_weighted_pred_err;
- double modified_err;
-
- //double relative_next_iiratio;
- //double next_iiratio;
- //double sum_iiratio;
- //int i;
-
- //FIRSTPASS_STATS next_frame;
- //FIRSTPASS_STATS *start_pos;
-
- /*start_pos = cpi->stats_in;
- sum_iiratio = 0.0;
- i = 0;
- while ( (i < 1) && input_stats(cpi,&next_frame) != EOF )
- {
-
- next_iiratio = next_frame.intra_error / DOUBLE_DIVIDE_CHECK(next_frame.coded_error);
- next_iiratio = ( next_iiratio < 1.0 ) ? 1.0 : (next_iiratio > 20.0) ? 20.0 : next_iiratio;
- sum_iiratio += next_iiratio;
- i++;
- }
- if ( i > 0 )
- {
- relative_next_iiratio = sum_iiratio / DOUBLE_DIVIDE_CHECK(cpi->avg_iiratio * (double)i);
- }
- else
- {
- relative_next_iiratio = 1.0;
- }
- reset_fpf_position(cpi, start_pos);*/
-
- if (this_err > av_err)
- modified_err = av_err * pow((this_err / DOUBLE_DIVIDE_CHECK(av_err)), POW1);
- else
- modified_err = av_err * pow((this_err / DOUBLE_DIVIDE_CHECK(av_err)), POW2);
-
- /*
- relative_next_iiratio = pow(relative_next_iiratio,0.25);
- modified_err = modified_err * relative_next_iiratio;
- */
-
- return modified_err;
-}
-
-static const double weight_table[256] = {
-0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000,
-0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000,
-0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000,
-0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000, 0.020000,
-0.020000, 0.031250, 0.062500, 0.093750, 0.125000, 0.156250, 0.187500, 0.218750,
-0.250000, 0.281250, 0.312500, 0.343750, 0.375000, 0.406250, 0.437500, 0.468750,
-0.500000, 0.531250, 0.562500, 0.593750, 0.625000, 0.656250, 0.687500, 0.718750,
-0.750000, 0.781250, 0.812500, 0.843750, 0.875000, 0.906250, 0.937500, 0.968750,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000,
-1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000, 1.000000
-};
-
-static double simple_weight(YV12_BUFFER_CONFIG *source)
-{
- int i, j;
-
- unsigned char *src = source->y_buffer;
- double sum_weights = 0.0;
-
- // Loop throught the Y plane raw examining levels and creating a weight for the image
- i = source->y_height;
- do
- {
- j = source->y_width;
- do
- {
- sum_weights += weight_table[ *src];
- src++;
- }while(--j);
- src -= source->y_width;
- src += source->y_stride;
- }while(--i);
-
- sum_weights /= (source->y_height * source->y_width);
-
- return sum_weights;
-}
-
-
-// This function returns the current per frame maximum bitrate target
-static int frame_max_bits(VP8_COMP *cpi)
-{
- // Max allocation for a single frame based on the max section guidelines passed in and how many bits are left
- int max_bits;
-
- // For CBR we need to also consider buffer fullness.
- // If we are running below the optimal level then we need to gradually tighten up on max_bits.
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- double buffer_fullness_ratio = (double)cpi->buffer_level / DOUBLE_DIVIDE_CHECK((double)cpi->oxcf.optimal_buffer_level);
-
- // For CBR base this on the target average bits per frame plus the maximum sedction rate passed in by the user
- max_bits = (int)(cpi->av_per_frame_bandwidth * ((double)cpi->oxcf.two_pass_vbrmax_section / 100.0));
-
- // If our buffer is below the optimum level
- if (buffer_fullness_ratio < 1.0)
- {
- // The lower of max_bits / 4 or cpi->av_per_frame_bandwidth / 4.
- int min_max_bits = ((cpi->av_per_frame_bandwidth >> 2) < (max_bits >> 2)) ? cpi->av_per_frame_bandwidth >> 2 : max_bits >> 2;
-
- max_bits = (int)(max_bits * buffer_fullness_ratio);
-
- if (max_bits < min_max_bits)
- max_bits = min_max_bits; // Lowest value we will set ... which should allow the buffer to refil.
- }
- }
- // VBR
- else
- {
- // For VBR base this on the bits and frames left plus the two_pass_vbrmax_section rate passed in by the user
- max_bits = (int)(((double)cpi->bits_left / (cpi->total_stats->count - (double)cpi->common.current_video_frame)) * ((double)cpi->oxcf.two_pass_vbrmax_section / 100.0));
- }
-
- // Trap case where we are out of bits
- if (max_bits < 0)
- max_bits = 0;
-
- return max_bits;
-}
-
-
-static void output_stats(const VP8_COMP *cpi,
- struct vpx_codec_pkt_list *pktlist,
- FIRSTPASS_STATS *stats)
-{
- struct vpx_codec_cx_pkt pkt;
- pkt.kind = VPX_CODEC_STATS_PKT;
- pkt.data.twopass_stats.buf = stats;
- pkt.data.twopass_stats.sz = sizeof(FIRSTPASS_STATS);
- vpx_codec_pkt_list_add(pktlist, &pkt);
-
-// TEMP debug code
-#if OUTPUT_FPF
-
- {
- FILE *fpfile;
- fpfile = fopen("firstpass.stt", "a");
-
- fprintf(fpfile, "%12.0f %12.0f %12.0f %12.4f %12.4f %12.4f %12.4f"
- " %12.4f %12.4f %12.4f %12.4f %12.4f %12.4f %12.4f %12.4f"
- " %12.0f %12.4f\n",
- stats->frame,
- stats->intra_error,
- stats->coded_error,
- stats->ssim_weighted_pred_err,
- stats->pcnt_inter,
- stats->pcnt_motion,
- stats->pcnt_second_ref,
- stats->pcnt_neutral,
- stats->MVr,
- stats->mvr_abs,
- stats->MVc,
- stats->mvc_abs,
- stats->MVrv,
- stats->MVcv,
- stats->mv_in_out_count,
- stats->count,
- stats->duration);
- fclose(fpfile);
- }
-#endif
-}
-
-static int input_stats(VP8_COMP *cpi, FIRSTPASS_STATS *fps)
-{
- if (cpi->stats_in >= cpi->stats_in_end)
- return EOF;
-
- *fps = *cpi->stats_in;
- cpi->stats_in = (void*)((char *)cpi->stats_in + sizeof(FIRSTPASS_STATS));
- return 1;
-}
-
-static void zero_stats(FIRSTPASS_STATS *section)
-{
- section->frame = 0.0;
- section->intra_error = 0.0;
- section->coded_error = 0.0;
- section->ssim_weighted_pred_err = 0.0;
- section->pcnt_inter = 0.0;
- section->pcnt_motion = 0.0;
- section->pcnt_second_ref = 0.0;
- section->pcnt_neutral = 0.0;
- section->MVr = 0.0;
- section->mvr_abs = 0.0;
- section->MVc = 0.0;
- section->mvc_abs = 0.0;
- section->MVrv = 0.0;
- section->MVcv = 0.0;
- section->mv_in_out_count = 0.0;
- section->count = 0.0;
- section->duration = 1.0;
-}
-static void accumulate_stats(FIRSTPASS_STATS *section, FIRSTPASS_STATS *frame)
-{
- section->frame += frame->frame;
- section->intra_error += frame->intra_error;
- section->coded_error += frame->coded_error;
- section->ssim_weighted_pred_err += frame->ssim_weighted_pred_err;
- section->pcnt_inter += frame->pcnt_inter;
- section->pcnt_motion += frame->pcnt_motion;
- section->pcnt_second_ref += frame->pcnt_second_ref;
- section->pcnt_neutral += frame->pcnt_neutral;
- section->MVr += frame->MVr;
- section->mvr_abs += frame->mvr_abs;
- section->MVc += frame->MVc;
- section->mvc_abs += frame->mvc_abs;
- section->MVrv += frame->MVrv;
- section->MVcv += frame->MVcv;
- section->mv_in_out_count += frame->mv_in_out_count;
- section->count += frame->count;
- section->duration += frame->duration;
-}
-static void avg_stats(FIRSTPASS_STATS *section)
-{
- if (section->count < 1.0)
- return;
-
- section->intra_error /= section->count;
- section->coded_error /= section->count;
- section->ssim_weighted_pred_err /= section->count;
- section->pcnt_inter /= section->count;
- section->pcnt_second_ref /= section->count;
- section->pcnt_neutral /= section->count;
- section->pcnt_motion /= section->count;
- section->MVr /= section->count;
- section->mvr_abs /= section->count;
- section->MVc /= section->count;
- section->mvc_abs /= section->count;
- section->MVrv /= section->count;
- section->MVcv /= section->count;
- section->mv_in_out_count /= section->count;
- section->duration /= section->count;
-}
-
-void vp8_init_first_pass(VP8_COMP *cpi)
-{
- zero_stats(cpi->total_stats);
-}
-
-void vp8_end_first_pass(VP8_COMP *cpi)
-{
- output_stats(cpi, cpi->output_pkt_list, cpi->total_stats);
-}
-
-static void zz_motion_search( VP8_COMP *cpi, MACROBLOCK * x, YV12_BUFFER_CONFIG * recon_buffer, int * best_motion_err, int recon_yoffset )
-{
- MACROBLOCKD * const xd = & x->e_mbd;
- BLOCK *b = &x->block[0];
- BLOCKD *d = &x->e_mbd.block[0];
-
- unsigned char *src_ptr = (*(b->base_src) + b->src);
- int src_stride = b->src_stride;
- unsigned char *ref_ptr;
- int ref_stride=d->pre_stride;
-
- // Set up pointers for this macro block recon buffer
- xd->pre.y_buffer = recon_buffer->y_buffer + recon_yoffset;
-
- ref_ptr = (unsigned char *)(*(d->base_pre) + d->pre );
-
- VARIANCE_INVOKE(IF_RTCD(&cpi->rtcd.variance), mse16x16) ( src_ptr, src_stride, ref_ptr, ref_stride, (unsigned int *)(best_motion_err));
-}
-
-static void first_pass_motion_search(VP8_COMP *cpi, MACROBLOCK *x, MV *ref_mv, MV *best_mv, YV12_BUFFER_CONFIG *recon_buffer, int *best_motion_err, int recon_yoffset )
-{
- MACROBLOCKD *const xd = & x->e_mbd;
- BLOCK *b = &x->block[0];
- BLOCKD *d = &x->e_mbd.block[0];
- int num00;
-
- MV tmp_mv = {0, 0};
-
- int tmp_err;
- int step_param = 3; //3; // Dont search over full range for first pass
- int further_steps = (MAX_MVSEARCH_STEPS - 1) - step_param; //3;
- int n;
- vp8_variance_fn_ptr_t v_fn_ptr = cpi->fn_ptr[BLOCK_16X16];
- int new_mv_mode_penalty = 256;
-
- // override the default variance function to use MSE
- v_fn_ptr.vf = VARIANCE_INVOKE(IF_RTCD(&cpi->rtcd.variance), mse16x16);
-
- // Set up pointers for this macro block recon buffer
- xd->pre.y_buffer = recon_buffer->y_buffer + recon_yoffset;
-
- // Initial step/diamond search centred on best mv
- tmp_err = cpi->diamond_search_sad(x, b, d, ref_mv, &tmp_mv, step_param, x->errorperbit, &num00, &v_fn_ptr, x->mvsadcost, x->mvcost, ref_mv);
- if ( tmp_err < INT_MAX-new_mv_mode_penalty )
- tmp_err += new_mv_mode_penalty;
-
- if (tmp_err < *best_motion_err)
- {
- *best_motion_err = tmp_err;
- best_mv->row = tmp_mv.row;
- best_mv->col = tmp_mv.col;
- }
-
- // Further step/diamond searches as necessary
- n = num00;
- num00 = 0;
-
- while (n < further_steps)
- {
- n++;
-
- if (num00)
- num00--;
- else
- {
- tmp_err = cpi->diamond_search_sad(x, b, d, ref_mv, &tmp_mv, step_param + n, x->errorperbit, &num00, &v_fn_ptr, x->mvsadcost, x->mvcost, ref_mv);
- if ( tmp_err < INT_MAX-new_mv_mode_penalty )
- tmp_err += new_mv_mode_penalty;
-
- if (tmp_err < *best_motion_err)
- {
- *best_motion_err = tmp_err;
- best_mv->row = tmp_mv.row;
- best_mv->col = tmp_mv.col;
- }
- }
- }
-}
-
-void vp8_first_pass(VP8_COMP *cpi)
-{
- int mb_row, mb_col;
- MACROBLOCK *const x = & cpi->mb;
- VP8_COMMON *const cm = & cpi->common;
- MACROBLOCKD *const xd = & x->e_mbd;
-
- int col_blocks = 4 * cm->mb_cols;
- int recon_yoffset, recon_uvoffset;
- YV12_BUFFER_CONFIG *lst_yv12 = &cm->yv12_fb[cm->lst_fb_idx];
- YV12_BUFFER_CONFIG *new_yv12 = &cm->yv12_fb[cm->new_fb_idx];
- YV12_BUFFER_CONFIG *gld_yv12 = &cm->yv12_fb[cm->gld_fb_idx];
- int recon_y_stride = lst_yv12->y_stride;
- int recon_uv_stride = lst_yv12->uv_stride;
- long long intra_error = 0;
- long long coded_error = 0;
-
- int sum_mvr = 0, sum_mvc = 0;
- int sum_mvr_abs = 0, sum_mvc_abs = 0;
- int sum_mvrs = 0, sum_mvcs = 0;
- int mvcount = 0;
- int intercount = 0;
- int second_ref_count = 0;
- int intrapenalty = 256;
- int neutral_count = 0;
-
- int sum_in_vectors = 0;
-
- MV zero_ref_mv = {0, 0};
-
- vp8_clear_system_state(); //__asm emms;
-
- x->src = * cpi->Source;
- xd->pre = *lst_yv12;
- xd->dst = *new_yv12;
-
- x->partition_info = x->pi;
-
- xd->mode_info_context = cm->mi;
-
- vp8_build_block_offsets(x);
-
- vp8_setup_block_dptrs(&x->e_mbd);
-
- vp8_setup_block_ptrs(x);
-
- // set up frame new frame for intra coded blocks
- vp8_setup_intra_recon(new_yv12);
- vp8cx_frame_init_quantizer(cpi);
-
- // Initialise the MV cost table to the defaults
- //if( cm->current_video_frame == 0)
- //if ( 0 )
- {
- int flag[2] = {1, 1};
- vp8_initialize_rd_consts(cpi, vp8_dc_quant(cm->base_qindex, cm->y1dc_delta_q));
- vpx_memcpy(cm->fc.mvc, vp8_default_mv_context, sizeof(vp8_default_mv_context));
- vp8_build_component_cost_table(cpi->mb.mvcost, cpi->mb.mvsadcost, (const MV_CONTEXT *) cm->fc.mvc, flag);
- }
-
- // for each macroblock row in image
- for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
- {
- int_mv best_ref_mv;
-
- best_ref_mv.as_int = 0;
-
- // reset above block coeffs
- xd->up_available = (mb_row != 0);
- recon_yoffset = (mb_row * recon_y_stride * 16);
- recon_uvoffset = (mb_row * recon_uv_stride * 8);
-
- // Set up limit values for motion vectors to prevent them extending outside the UMV borders
- x->mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 16));
- x->mv_row_max = ((cm->mb_rows - 1 - mb_row) * 16) + (VP8BORDERINPIXELS - 16);
-
-
- // for each macroblock col in image
- for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
- {
- int this_error;
- int zz_to_best_ratio;
- int gf_motion_error = INT_MAX;
- int use_dc_pred = (mb_col || mb_row) && (!mb_col || !mb_row);
-
- xd->dst.y_buffer = new_yv12->y_buffer + recon_yoffset;
- xd->dst.u_buffer = new_yv12->u_buffer + recon_uvoffset;
- xd->dst.v_buffer = new_yv12->v_buffer + recon_uvoffset;
- xd->left_available = (mb_col != 0);
-
- // do intra 16x16 prediction
- this_error = encode_intra(cpi, x, use_dc_pred);
-
- // "intrapenalty" below deals with situations where the intra and inter error scores are very low (eg a plain black frame)
- // We do not have special cases in first pass for 0,0 and nearest etc so all inter modes carry an overhead cost estimate fot the mv.
- // When the error score is very low this causes us to pick all or lots of INTRA modes and throw lots of key frames.
- // This penalty adds a cost matching that of a 0,0 mv to the intra case.
- this_error += intrapenalty;
-
- // Cumulative intra error total
- intra_error += (long long)this_error;
-
- // Set up limit values for motion vectors to prevent them extending outside the UMV borders
- x->mv_col_min = -((mb_col * 16) + (VP8BORDERINPIXELS - 16));
- x->mv_col_max = ((cm->mb_cols - 1 - mb_col) * 16) + (VP8BORDERINPIXELS - 16);
-
- // Other than for the first frame do a motion search
- if (cm->current_video_frame > 0)
- {
- BLOCK *b = &x->block[0];
- BLOCKD *d = &x->e_mbd.block[0];
- MV tmp_mv = {0, 0};
- int tmp_err;
- int motion_error = INT_MAX;
-
- // Simple 0,0 motion with no mv overhead
- zz_motion_search( cpi, x, lst_yv12, &motion_error, recon_yoffset );
- d->bmi.mv.as_mv.row = 0;
- d->bmi.mv.as_mv.col = 0;
-
- // Test last reference frame using the previous best mv as the
- // starting point (best reference) for the search
- first_pass_motion_search(cpi, x, &best_ref_mv.as_mv,
- &d->bmi.mv.as_mv, lst_yv12,
- &motion_error, recon_yoffset);
-
- // If the current best reference mv is not centred on 0,0 then do a 0,0 based search as well
- if (best_ref_mv.as_int)
- {
- tmp_err = INT_MAX;
- first_pass_motion_search(cpi, x, &zero_ref_mv, &tmp_mv,
- lst_yv12, &tmp_err, recon_yoffset);
-
- if ( tmp_err < motion_error )
- {
- motion_error = tmp_err;
- d->bmi.mv.as_mv.row = tmp_mv.row;
- d->bmi.mv.as_mv.col = tmp_mv.col;
- }
- }
-
- // Experimental search in a second reference frame ((0,0) based only)
- if (cm->current_video_frame > 1)
- {
- first_pass_motion_search(cpi, x, &zero_ref_mv, &tmp_mv, gld_yv12, &gf_motion_error, recon_yoffset);
-
- if ((gf_motion_error < motion_error) && (gf_motion_error < this_error))
- {
- second_ref_count++;
- //motion_error = gf_motion_error;
- //d->bmi.mv.as_mv.row = tmp_mv.row;
- //d->bmi.mv.as_mv.col = tmp_mv.col;
- }
- /*else
- {
- xd->pre.y_buffer = cm->last_frame.y_buffer + recon_yoffset;
- xd->pre.u_buffer = cm->last_frame.u_buffer + recon_uvoffset;
- xd->pre.v_buffer = cm->last_frame.v_buffer + recon_uvoffset;
- }*/
-
-
- // Reset to last frame as reference buffer
- xd->pre.y_buffer = lst_yv12->y_buffer + recon_yoffset;
- xd->pre.u_buffer = lst_yv12->u_buffer + recon_uvoffset;
- xd->pre.v_buffer = lst_yv12->v_buffer + recon_uvoffset;
- }
-
- /* Intra assumed best */
- best_ref_mv.as_int = 0;
-
- if (motion_error <= this_error)
- {
- // Keep a count of cases where the inter and intra were
- // very close and very low. This helps with scene cut
- // detection for example in cropped clips with black bars
- // at the sides or top and bottom.
- if( (((this_error-intrapenalty) * 9) <=
- (motion_error*10)) &&
- (this_error < (2*intrapenalty)) )
- {
- neutral_count++;
- }
-
- d->bmi.mv.as_mv.row <<= 3;
- d->bmi.mv.as_mv.col <<= 3;
- this_error = motion_error;
- vp8_set_mbmode_and_mvs(x, NEWMV, &d->bmi.mv.as_mv);
- vp8_encode_inter16x16y(IF_RTCD(&cpi->rtcd), x);
- sum_mvr += d->bmi.mv.as_mv.row;
- sum_mvr_abs += abs(d->bmi.mv.as_mv.row);
- sum_mvc += d->bmi.mv.as_mv.col;
- sum_mvc_abs += abs(d->bmi.mv.as_mv.col);
- sum_mvrs += d->bmi.mv.as_mv.row * d->bmi.mv.as_mv.row;
- sum_mvcs += d->bmi.mv.as_mv.col * d->bmi.mv.as_mv.col;
- intercount++;
-
- best_ref_mv.as_int = d->bmi.mv.as_int;
-
- // Was the vector non-zero
- if (d->bmi.mv.as_int)
- {
- mvcount++;
-
- // Does the Row vector point inwards or outwards
- if (mb_row < cm->mb_rows / 2)
- {
- if (d->bmi.mv.as_mv.row > 0)
- sum_in_vectors--;
- else if (d->bmi.mv.as_mv.row < 0)
- sum_in_vectors++;
- }
- else if (mb_row > cm->mb_rows / 2)
- {
- if (d->bmi.mv.as_mv.row > 0)
- sum_in_vectors++;
- else if (d->bmi.mv.as_mv.row < 0)
- sum_in_vectors--;
- }
-
- // Does the Row vector point inwards or outwards
- if (mb_col < cm->mb_cols / 2)
- {
- if (d->bmi.mv.as_mv.col > 0)
- sum_in_vectors--;
- else if (d->bmi.mv.as_mv.col < 0)
- sum_in_vectors++;
- }
- else if (mb_col > cm->mb_cols / 2)
- {
- if (d->bmi.mv.as_mv.col > 0)
- sum_in_vectors++;
- else if (d->bmi.mv.as_mv.col < 0)
- sum_in_vectors--;
- }
- }
- }
- }
-
- coded_error += (long long)this_error;
-
- // adjust to the next column of macroblocks
- x->src.y_buffer += 16;
- x->src.u_buffer += 8;
- x->src.v_buffer += 8;
-
- recon_yoffset += 16;
- recon_uvoffset += 8;
- }
-
- // adjust to the next row of mbs
- x->src.y_buffer += 16 * x->src.y_stride - 16 * cm->mb_cols;
- x->src.u_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols;
- x->src.v_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols;
-
- //extend the recon for intra prediction
- vp8_extend_mb_row(new_yv12, xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8);
- vp8_clear_system_state(); //__asm emms;
- }
-
- vp8_clear_system_state(); //__asm emms;
- {
- double weight = 0.0;
-
- FIRSTPASS_STATS fps;
-
- fps.frame = cm->current_video_frame ;
- fps.intra_error = intra_error >> 8;
- fps.coded_error = coded_error >> 8;
- weight = simple_weight(cpi->Source);
-
-
- if (weight < 0.1)
- weight = 0.1;
-
- fps.ssim_weighted_pred_err = fps.coded_error * weight;
-
- fps.pcnt_inter = 0.0;
- fps.pcnt_motion = 0.0;
- fps.MVr = 0.0;
- fps.mvr_abs = 0.0;
- fps.MVc = 0.0;
- fps.mvc_abs = 0.0;
- fps.MVrv = 0.0;
- fps.MVcv = 0.0;
- fps.mv_in_out_count = 0.0;
- fps.count = 1.0;
-
- fps.pcnt_inter = 1.0 * (double)intercount / cm->MBs;
- fps.pcnt_second_ref = 1.0 * (double)second_ref_count / cm->MBs;
- fps.pcnt_neutral = 1.0 * (double)neutral_count / cm->MBs;
-
- if (mvcount > 0)
- {
- fps.MVr = (double)sum_mvr / (double)mvcount;
- fps.mvr_abs = (double)sum_mvr_abs / (double)mvcount;
- fps.MVc = (double)sum_mvc / (double)mvcount;
- fps.mvc_abs = (double)sum_mvc_abs / (double)mvcount;
- fps.MVrv = ((double)sum_mvrs - (fps.MVr * fps.MVr / (double)mvcount)) / (double)mvcount;
- fps.MVcv = ((double)sum_mvcs - (fps.MVc * fps.MVc / (double)mvcount)) / (double)mvcount;
- fps.mv_in_out_count = (double)sum_in_vectors / (double)(mvcount * 2);
-
- fps.pcnt_motion = 1.0 * (double)mvcount / cpi->common.MBs;
- }
-
- // TODO: handle the case when duration is set to 0, or something less
- // than the full time between subsequent cpi->source_time_stamp s .
- fps.duration = cpi->source_end_time_stamp - cpi->source_time_stamp;
-
- // don't want to do output stats with a stack variable!
- memcpy(cpi->this_frame_stats,
- &fps,
- sizeof(FIRSTPASS_STATS));
- output_stats(cpi, cpi->output_pkt_list, cpi->this_frame_stats);
- accumulate_stats(cpi->total_stats, &fps);
- }
-
- // Copy the previous Last Frame into the GF buffer if specific conditions for doing so are met
- if ((cm->current_video_frame > 0) &&
- (cpi->this_frame_stats->pcnt_inter > 0.20) &&
- ((cpi->this_frame_stats->intra_error / cpi->this_frame_stats->coded_error) > 2.0))
- {
- vp8_yv12_copy_frame_ptr(lst_yv12, gld_yv12);
- }
-
- // swap frame pointers so last frame refers to the frame we just compressed
- vp8_swap_yv12_buffer(lst_yv12, new_yv12);
- vp8_yv12_extend_frame_borders(lst_yv12);
-
- // Special case for the first frame. Copy into the GF buffer as a second reference.
- if (cm->current_video_frame == 0)
- {
- vp8_yv12_copy_frame_ptr(lst_yv12, gld_yv12);
- }
-
-
- // use this to see what the first pass reconstruction looks like
- if (0)
- {
- char filename[512];
- FILE *recon_file;
- sprintf(filename, "enc%04d.yuv", (int) cm->current_video_frame);
-
- if (cm->current_video_frame == 0)
- recon_file = fopen(filename, "wb");
- else
- recon_file = fopen(filename, "ab");
-
- if(fwrite(lst_yv12->buffer_alloc, lst_yv12->frame_size, 1, recon_file));
- fclose(recon_file);
- }
-
- cm->current_video_frame++;
-
-}
-extern const int vp8_bits_per_mb[2][QINDEX_RANGE];
-
-#define BASE_ERRPERMB 150
-static int estimate_max_q(VP8_COMP *cpi, double section_err, int section_target_bandwitdh)
-{
- int Q;
- int num_mbs = cpi->common.MBs;
- int target_norm_bits_per_mb;
-
- double err_per_mb = section_err / num_mbs;
- double correction_factor;
- double corr_high;
- double speed_correction = 1.0;
- double rolling_ratio;
-
- double pow_highq = 0.90;
- double pow_lowq = 0.40;
-
- if (section_target_bandwitdh <= 0)
- return cpi->maxq_max_limit; // Highest value allowed
-
- target_norm_bits_per_mb = (section_target_bandwitdh < (1 << 20)) ? (512 * section_target_bandwitdh) / num_mbs : 512 * (section_target_bandwitdh / num_mbs);
-
- // Calculate a corrective factor based on a rolling ratio of bits spent vs target bits
- if ((cpi->rolling_target_bits > 0.0) && (cpi->active_worst_quality < cpi->worst_quality))
- {
- //double adjustment_rate = 0.985 + (0.00005 * cpi->active_worst_quality);
- double adjustment_rate = 0.99;
-
- rolling_ratio = (double)cpi->rolling_actual_bits / (double)cpi->rolling_target_bits;
-
- //if ( cpi->est_max_qcorrection_factor > rolling_ratio )
- if (rolling_ratio < 0.95)
- //cpi->est_max_qcorrection_factor *= adjustment_rate;
- cpi->est_max_qcorrection_factor -= 0.005;
- //else if ( cpi->est_max_qcorrection_factor < rolling_ratio )
- else if (rolling_ratio > 1.05)
- cpi->est_max_qcorrection_factor += 0.005;
-
- //cpi->est_max_qcorrection_factor /= adjustment_rate;
-
- cpi->est_max_qcorrection_factor = (cpi->est_max_qcorrection_factor < 0.1) ? 0.1 : (cpi->est_max_qcorrection_factor > 10.0) ? 10.0 : cpi->est_max_qcorrection_factor;
- }
-
- // Corrections for higher compression speed settings (reduced compression expected)
- if ((cpi->compressor_speed == 3) || (cpi->compressor_speed == 1))
- {
- if (cpi->oxcf.cpu_used <= 5)
- speed_correction = 1.04 + (cpi->oxcf.cpu_used * 0.04);
- else
- speed_correction = 1.25;
- }
-
- // Correction factor used for Q values >= 20
- corr_high = pow(err_per_mb / BASE_ERRPERMB, pow_highq);
- corr_high = (corr_high < 0.05)
- ? 0.05 : (corr_high > 5.0) ? 5.0 : corr_high;
-
- // Try and pick a max Q that will be high enough to encode the
- // content at the given rate.
- for (Q = cpi->maxq_min_limit; Q < cpi->maxq_max_limit; Q++)
- {
- int bits_per_mb_at_this_q;
-
- if (Q < 50)
- {
- correction_factor = pow(err_per_mb / BASE_ERRPERMB, (pow_lowq + Q * 0.01));
- correction_factor = (correction_factor < 0.05) ? 0.05 : (correction_factor > 5.0) ? 5.0 : correction_factor;
- }
- else
- correction_factor = corr_high;
-
- bits_per_mb_at_this_q = (int)(.5 + correction_factor * speed_correction * cpi->est_max_qcorrection_factor * cpi->section_max_qfactor * (double)vp8_bits_per_mb[INTER_FRAME][Q] / 1.0);
- //bits_per_mb_at_this_q = (int)(.5 + correction_factor * speed_correction * cpi->est_max_qcorrection_factor * (double)vp8_bits_per_mb[INTER_FRAME][Q] / 1.0);
-
- if (bits_per_mb_at_this_q <= target_norm_bits_per_mb)
- break;
- }
-
- // Restriction on active max q for constrained quality mode.
- if ( (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
- (Q < cpi->cq_target_quality) )
- //(Q < cpi->oxcf.cq_level;) )
- {
- Q = cpi->cq_target_quality;
- //Q = cpi->oxcf.cq_level;
- }
-
- // Adjust maxq_min_limit and maxq_max_limit limits based on
- // averaga q observed in clip for non kf/gf.arf frames
- // Give average a chance to settle though.
- if ( (cpi->ni_frames >
- ((unsigned int)cpi->total_stats->count >> 8)) &&
- (cpi->ni_frames > 150) )
- {
- cpi->maxq_max_limit = ((cpi->ni_av_qi + 32) < cpi->worst_quality)
- ? (cpi->ni_av_qi + 32) : cpi->worst_quality;
- cpi->maxq_min_limit = ((cpi->ni_av_qi - 32) > cpi->best_quality)
- ? (cpi->ni_av_qi - 32) : cpi->best_quality;
- }
-
- return Q;
-}
-static int estimate_q(VP8_COMP *cpi, double section_err, int section_target_bandwitdh)
-{
- int Q;
- int num_mbs = cpi->common.MBs;
- int target_norm_bits_per_mb;
-
- double err_per_mb = section_err / num_mbs;
- double correction_factor;
- double corr_high;
- double speed_correction = 1.0;
- double pow_highq = 0.90;
- double pow_lowq = 0.40;
-
- target_norm_bits_per_mb = (section_target_bandwitdh < (1 << 20)) ? (512 * section_target_bandwitdh) / num_mbs : 512 * (section_target_bandwitdh / num_mbs);
-
- // Corrections for higher compression speed settings (reduced compression expected)
- if ((cpi->compressor_speed == 3) || (cpi->compressor_speed == 1))
- {
- if (cpi->oxcf.cpu_used <= 5)
- speed_correction = 1.04 + (cpi->oxcf.cpu_used * 0.04);
- else
- speed_correction = 1.25;
- }
-
- // Correction factor used for Q values >= 20
- corr_high = pow(err_per_mb / BASE_ERRPERMB, pow_highq);
- corr_high = (corr_high < 0.05) ? 0.05 : (corr_high > 5.0) ? 5.0 : corr_high;
-
- // Try and pick a Q that can encode the content at the given rate.
- for (Q = 0; Q < MAXQ; Q++)
- {
- int bits_per_mb_at_this_q;
-
- if (Q < 50)
- {
- correction_factor = pow(err_per_mb / BASE_ERRPERMB, (pow_lowq + Q * 0.01));
- correction_factor = (correction_factor < 0.05) ? 0.05 : (correction_factor > 5.0) ? 5.0 : correction_factor;
- }
- else
- correction_factor = corr_high;
-
- bits_per_mb_at_this_q = (int)(.5 + correction_factor * speed_correction * cpi->est_max_qcorrection_factor * (double)vp8_bits_per_mb[INTER_FRAME][Q] / 1.0);
-
- if (bits_per_mb_at_this_q <= target_norm_bits_per_mb)
- break;
- }
-
- return Q;
-}
-
-// Estimate a worst case Q for a KF group
-static int estimate_kf_group_q(VP8_COMP *cpi, double section_err, int section_target_bandwitdh, double group_iiratio)
-{
- int Q;
- int num_mbs = cpi->common.MBs;
- int target_norm_bits_per_mb = (512 * section_target_bandwitdh) / num_mbs;
- int bits_per_mb_at_this_q;
-
- double err_per_mb = section_err / num_mbs;
- double err_correction_factor;
- double corr_high;
- double speed_correction = 1.0;
- double current_spend_ratio = 1.0;
-
- double pow_highq = (POW1 < 0.6) ? POW1 + 0.3 : 0.90;
- double pow_lowq = (POW1 < 0.7) ? POW1 + 0.1 : 0.80;
-
- double iiratio_correction_factor = 1.0;
-
- double combined_correction_factor;
-
- // Trap special case where the target is <= 0
- if (target_norm_bits_per_mb <= 0)
- return MAXQ * 2;
-
- // Calculate a corrective factor based on a rolling ratio of bits spent vs target bits
- // This is clamped to the range 0.1 to 10.0
- if (cpi->long_rolling_target_bits <= 0)
- current_spend_ratio = 10.0;
- else
- {
- current_spend_ratio = (double)cpi->long_rolling_actual_bits / (double)cpi->long_rolling_target_bits;
- current_spend_ratio = (current_spend_ratio > 10.0) ? 10.0 : (current_spend_ratio < 0.1) ? 0.1 : current_spend_ratio;
- }
-
- // Calculate a correction factor based on the quality of prediction in the sequence as indicated by intra_inter error score ratio (IIRatio)
- // The idea here is to favour subsampling in the hardest sections vs the easyest.
- iiratio_correction_factor = 1.0 - ((group_iiratio - 6.0) * 0.1);
-
- if (iiratio_correction_factor < 0.5)
- iiratio_correction_factor = 0.5;
-
- // Corrections for higher compression speed settings (reduced compression expected)
- if ((cpi->compressor_speed == 3) || (cpi->compressor_speed == 1))
- {
- if (cpi->oxcf.cpu_used <= 5)
- speed_correction = 1.04 + (cpi->oxcf.cpu_used * 0.04);
- else
- speed_correction = 1.25;
- }
-
- // Combine the various factors calculated above
- combined_correction_factor = speed_correction * iiratio_correction_factor * current_spend_ratio;
-
- // Correction factor used for Q values >= 20
- corr_high = pow(err_per_mb / BASE_ERRPERMB, pow_highq);
- corr_high = (corr_high < 0.05) ? 0.05 : (corr_high > 5.0) ? 5.0 : corr_high;
-
- // Try and pick a Q that should be high enough to encode the content at the given rate.
- for (Q = 0; Q < MAXQ; Q++)
- {
- // Q values < 20 treated as a special case
- if (Q < 20)
- {
- err_correction_factor = pow(err_per_mb / BASE_ERRPERMB, (pow_lowq + Q * 0.01));
- err_correction_factor = (err_correction_factor < 0.05) ? 0.05 : (err_correction_factor > 5.0) ? 5.0 : err_correction_factor;
- }
- else
- err_correction_factor = corr_high;
-
- bits_per_mb_at_this_q = (int)(.5 + err_correction_factor * combined_correction_factor * (double)vp8_bits_per_mb[INTER_FRAME][Q]);
-
- if (bits_per_mb_at_this_q <= target_norm_bits_per_mb)
- break;
- }
-
- // If we could not hit the target even at Max Q then estimate what Q would have bee required
- while ((bits_per_mb_at_this_q > target_norm_bits_per_mb) && (Q < (MAXQ * 2)))
- {
-
- bits_per_mb_at_this_q = (int)(0.96 * bits_per_mb_at_this_q);
- Q++;
- }
-
- if (0)
- {
- FILE *f = fopen("estkf_q.stt", "a");
- fprintf(f, "%8d %8d %8d %8.2f %8.3f %8.2f %8.3f %8.3f %8.3f %8d\n", cpi->common.current_video_frame, bits_per_mb_at_this_q,
- target_norm_bits_per_mb, err_per_mb, err_correction_factor,
- current_spend_ratio, group_iiratio, iiratio_correction_factor,
- (double)cpi->buffer_level / (double)cpi->oxcf.optimal_buffer_level, Q);
- fclose(f);
- }
-
- return Q;
-}
-
-// For cq mode estimate a cq level that matches the observed
-// complexity and data rate.
-static int estimate_cq(VP8_COMP *cpi, double section_err, int section_target_bandwitdh)
-{
- int Q;
- int num_mbs = cpi->common.MBs;
- int target_norm_bits_per_mb;
-
- double err_per_mb = section_err / num_mbs;
- double correction_factor;
- double corr_high;
- double speed_correction = 1.0;
- double pow_highq = 0.90;
- double pow_lowq = 0.40;
- double clip_iiratio;
- double clip_iifactor;
-
- target_norm_bits_per_mb = (section_target_bandwitdh < (1 << 20))
- ? (512 * section_target_bandwitdh) / num_mbs
- : 512 * (section_target_bandwitdh / num_mbs);
-
- // Corrections for higher compression speed settings
- // (reduced compression expected)
- if ((cpi->compressor_speed == 3) || (cpi->compressor_speed == 1))
- {
- if (cpi->oxcf.cpu_used <= 5)
- speed_correction = 1.04 + (cpi->oxcf.cpu_used * 0.04);
- else
- speed_correction = 1.25;
- }
- // II ratio correction factor for clip as a whole
- clip_iiratio = cpi->total_stats->intra_error /
- DOUBLE_DIVIDE_CHECK(cpi->total_stats->coded_error);
- clip_iifactor = 1.0 - ((clip_iiratio - 10.0) * 0.025);
- if (clip_iifactor < 0.80)
- clip_iifactor = 0.80;
-
- // Correction factor used for Q values >= 20
- corr_high = pow(err_per_mb / BASE_ERRPERMB, pow_highq);
- corr_high = (corr_high < 0.05) ? 0.05 : (corr_high > 5.0) ? 5.0 : corr_high;
-
- // Try and pick a Q that can encode the content at the given rate.
- for (Q = 0; Q < MAXQ; Q++)
- {
- int bits_per_mb_at_this_q;
-
- if (Q < 50)
- {
- correction_factor =
- pow( err_per_mb / BASE_ERRPERMB, (pow_lowq + Q * 0.01));
-
- correction_factor = (correction_factor < 0.05) ? 0.05
- : (correction_factor > 5.0) ? 5.0
- : correction_factor;
- }
- else
- correction_factor = corr_high;
-
- bits_per_mb_at_this_q =
- (int)( .5 + correction_factor *
- speed_correction *
- clip_iifactor *
- (double)vp8_bits_per_mb[INTER_FRAME][Q] / 1.0);
-
- if (bits_per_mb_at_this_q <= target_norm_bits_per_mb)
- break;
- }
-
- return cq_level[Q];
-}
-
-extern void vp8_new_frame_rate(VP8_COMP *cpi, double framerate);
-
-void vp8_init_second_pass(VP8_COMP *cpi)
-{
- FIRSTPASS_STATS this_frame;
- FIRSTPASS_STATS *start_pos;
-
- double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100);
-
- zero_stats(cpi->total_stats);
-
- if (!cpi->stats_in_end)
- return;
-
- *cpi->total_stats = *cpi->stats_in_end;
-
- cpi->total_error_left = cpi->total_stats->ssim_weighted_pred_err;
- cpi->total_intra_error_left = cpi->total_stats->intra_error;
- cpi->total_coded_error_left = cpi->total_stats->coded_error;
- cpi->start_tot_err_left = cpi->total_error_left;
-
- //cpi->bits_left = (long long)(cpi->total_stats->count * cpi->oxcf.target_bandwidth / DOUBLE_DIVIDE_CHECK((double)cpi->oxcf.frame_rate));
- //cpi->bits_left -= (long long)(cpi->total_stats->count * two_pass_min_rate / DOUBLE_DIVIDE_CHECK((double)cpi->oxcf.frame_rate));
-
- // each frame can have a different duration, as the frame rate in the source
- // isn't guaranteed to be constant. The frame rate prior to the first frame
- // encoded in the second pass is a guess. However the sum duration is not.
- // Its calculated based on the actual durations of all frames from the first
- // pass.
- vp8_new_frame_rate(cpi, 10000000.0 * cpi->total_stats->count / cpi->total_stats->duration);
-
- cpi->output_frame_rate = cpi->oxcf.frame_rate;
- cpi->bits_left = (long long)(cpi->total_stats->duration * cpi->oxcf.target_bandwidth / 10000000.0) ;
- cpi->bits_left -= (long long)(cpi->total_stats->duration * two_pass_min_rate / 10000000.0);
- cpi->clip_bits_total = cpi->bits_left;
-
- // Calculate a minimum intra value to be used in determining the IIratio
- // scores used in the second pass. We have this minimum to make sure
- // that clips that are static but "low complexity" in the intra domain
- // are still boosted appropriately for KF/GF/ARF
- cpi->kf_intra_err_min = KF_MB_INTRA_MIN * cpi->common.MBs;
- cpi->gf_intra_err_min = GF_MB_INTRA_MIN * cpi->common.MBs;
-
- avg_stats(cpi->total_stats);
-
- // Scan the first pass file and calculate an average Intra / Inter error score ratio for the sequence
- {
- double sum_iiratio = 0.0;
- double IIRatio;
-
- start_pos = cpi->stats_in; // Note starting "file" position
-
- while (input_stats(cpi, &this_frame) != EOF)
- {
- IIRatio = this_frame.intra_error / DOUBLE_DIVIDE_CHECK(this_frame.coded_error);
- IIRatio = (IIRatio < 1.0) ? 1.0 : (IIRatio > 20.0) ? 20.0 : IIRatio;
- sum_iiratio += IIRatio;
- }
-
- cpi->avg_iiratio = sum_iiratio / DOUBLE_DIVIDE_CHECK((double)cpi->total_stats->count);
-
- // Reset file position
- reset_fpf_position(cpi, start_pos);
- }
-
- // Scan the first pass file and calculate a modified total error based upon the bias/power function
- // used to allocate bits
- {
- start_pos = cpi->stats_in; // Note starting "file" position
-
- cpi->modified_error_total = 0.0;
- cpi->modified_error_used = 0.0;
-
- while (input_stats(cpi, &this_frame) != EOF)
- {
- cpi->modified_error_total += calculate_modified_err(cpi, &this_frame);
- }
- cpi->modified_error_left = cpi->modified_error_total;
-
- reset_fpf_position(cpi, start_pos); // Reset file position
-
- }
-
- // Calculate the clip target modified bits per error
- // The observed bpe starts as the same number.
- cpi->clip_bpe = cpi->bits_left /
- DOUBLE_DIVIDE_CHECK(cpi->modified_error_total);
- cpi->observed_bpe = cpi->clip_bpe;
-}
-
-void vp8_end_second_pass(VP8_COMP *cpi)
-{
-}
-
-// This function gives and estimate of how badly we believe
-// the prediction quality is decaying from frame to frame.
-static double get_prediction_decay_rate(VP8_COMP *cpi, FIRSTPASS_STATS *next_frame)
-{
- double prediction_decay_rate;
- double motion_decay;
- double motion_pct = next_frame->pcnt_motion;
-
-
- // Initial basis is the % mbs inter coded
- prediction_decay_rate = next_frame->pcnt_inter;
-
- // High % motion -> somewhat higher decay rate
- motion_decay = (1.0 - (motion_pct / 20.0));
- if (motion_decay < prediction_decay_rate)
- prediction_decay_rate = motion_decay;
-
- // Adjustment to decay rate based on speed of motion
- {
- double this_mv_rabs;
- double this_mv_cabs;
- double distance_factor;
-
- this_mv_rabs = fabs(next_frame->mvr_abs * motion_pct);
- this_mv_cabs = fabs(next_frame->mvc_abs * motion_pct);
-
- distance_factor = sqrt((this_mv_rabs * this_mv_rabs) +
- (this_mv_cabs * this_mv_cabs)) / 250.0;
- distance_factor = ((distance_factor > 1.0)
- ? 0.0 : (1.0 - distance_factor));
- if (distance_factor < prediction_decay_rate)
- prediction_decay_rate = distance_factor;
- }
-
- return prediction_decay_rate;
-}
-
-// Function to test for a condition where a complex transition is followed
-// by a static section. For example in slide shows where there is a fade
-// between slides. This is to help with more optimal kf and gf positioning.
-static int detect_transition_to_still(
- VP8_COMP *cpi,
- int frame_interval,
- int still_interval,
- double loop_decay_rate,
- double decay_accumulator )
-{
- BOOL trans_to_still = FALSE;
-
- // Break clause to detect very still sections after motion
- // For example a static image after a fade or other transition
- // instead of a clean scene cut.
- if ( (frame_interval > MIN_GF_INTERVAL) &&
- (loop_decay_rate >= 0.999) &&
- (decay_accumulator < 0.9) )
- {
- int j;
- FIRSTPASS_STATS * position = cpi->stats_in;
- FIRSTPASS_STATS tmp_next_frame;
- double decay_rate;
-
- // Look ahead a few frames to see if static condition
- // persists...
- for ( j = 0; j < still_interval; j++ )
- {
- if (EOF == input_stats(cpi, &tmp_next_frame))
- break;
-
- decay_rate = get_prediction_decay_rate(cpi, &tmp_next_frame);
- if ( decay_rate < 0.999 )
- break;
- }
- // Reset file position
- reset_fpf_position(cpi, position);
-
- // Only if it does do we signal a transition to still
- if ( j == still_interval )
- trans_to_still = TRUE;
- }
-
- return trans_to_still;
-}
-
-// Analyse and define a gf/arf group .
-static void define_gf_group(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
-{
- FIRSTPASS_STATS next_frame;
- FIRSTPASS_STATS *start_pos;
- int i;
- int y_width = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_width;
- int y_height = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_height;
- int image_size = y_width * y_height;
- double boost_score = 0.0;
- double old_boost_score = 0.0;
- double gf_group_err = 0.0;
- double gf_first_frame_err = 0.0;
- double mod_frame_err = 0.0;
-
- double mv_accumulator_rabs = 0.0;
- double mv_accumulator_cabs = 0.0;
- double mv_ratio_accumulator = 0.0;
- double decay_accumulator = 1.0;
-
- double boost_factor = IIFACTOR;
- double loop_decay_rate = 1.00; // Starting decay rate
-
- double this_frame_mv_in_out = 0.0;
- double mv_in_out_accumulator = 0.0;
- double abs_mv_in_out_accumulator = 0.0;
- double mod_err_per_mb_accumulator = 0.0;
-
- int max_bits = frame_max_bits(cpi); // Max for a single frame
-
- unsigned int allow_alt_ref =
- cpi->oxcf.play_alternate && cpi->oxcf.lag_in_frames;
-
- cpi->gf_group_bits = 0;
- cpi->gf_decay_rate = 0;
-
- vp8_clear_system_state(); //__asm emms;
-
- start_pos = cpi->stats_in;
-
- vpx_memset(&next_frame, 0, sizeof(next_frame)); // assure clean
-
- // Preload the stats for the next frame.
- mod_frame_err = calculate_modified_err(cpi, this_frame);
-
- // Note the error of the frame at the start of the group (this will be
- // the GF frame error if we code a normal gf
- gf_first_frame_err = mod_frame_err;
-
- // Special treatment if the current frame is a key frame (which is also
- // a gf). If it is then its error score (and hence bit allocation) need
- // to be subtracted out from the calculation for the GF group
- if (cpi->common.frame_type == KEY_FRAME)
- gf_group_err -= gf_first_frame_err;
-
- // Scan forward to try and work out how many frames the next gf group
- // should contain and what level of boost is appropriate for the GF
- // or ARF that will be coded with the group
- i = 0;
-
- while (((i < cpi->static_scene_max_gf_interval) ||
- ((cpi->frames_to_key - i) < MIN_GF_INTERVAL)) &&
- (i < cpi->frames_to_key))
- {
- double r;
- double this_frame_mvr_ratio;
- double this_frame_mvc_ratio;
- double motion_decay;
- //double motion_pct = next_frame.pcnt_motion;
- double motion_pct;
-
- i++; // Increment the loop counter
-
- // Accumulate error score of frames in this gf group
- mod_frame_err = calculate_modified_err(cpi, this_frame);
-
- gf_group_err += mod_frame_err;
-
- mod_err_per_mb_accumulator +=
- mod_frame_err / DOUBLE_DIVIDE_CHECK((double)cpi->common.MBs);
-
- if (EOF == input_stats(cpi, &next_frame))
- break;
-
- // Accumulate motion stats.
- motion_pct = next_frame.pcnt_motion;
- mv_accumulator_rabs += fabs(next_frame.mvr_abs * motion_pct);
- mv_accumulator_cabs += fabs(next_frame.mvc_abs * motion_pct);
-
- //Accumulate Motion In/Out of frame stats
- this_frame_mv_in_out =
- next_frame.mv_in_out_count * motion_pct;
- mv_in_out_accumulator +=
- next_frame.mv_in_out_count * motion_pct;
- abs_mv_in_out_accumulator +=
- fabs(next_frame.mv_in_out_count * motion_pct);
-
- // If there is a significant amount of motion
- if (motion_pct > 0.05)
- {
- this_frame_mvr_ratio = fabs(next_frame.mvr_abs) /
- DOUBLE_DIVIDE_CHECK(fabs(next_frame.MVr));
-
- this_frame_mvc_ratio = fabs(next_frame.mvc_abs) /
- DOUBLE_DIVIDE_CHECK(fabs(next_frame.MVc));
-
- mv_ratio_accumulator +=
- (this_frame_mvr_ratio < next_frame.mvr_abs)
- ? (this_frame_mvr_ratio * motion_pct)
- : next_frame.mvr_abs * motion_pct;
-
- mv_ratio_accumulator +=
- (this_frame_mvc_ratio < next_frame.mvc_abs)
- ? (this_frame_mvc_ratio * motion_pct)
- : next_frame.mvc_abs * motion_pct;
- }
- else
- {
- mv_ratio_accumulator += 0.0;
- this_frame_mvr_ratio = 1.0;
- this_frame_mvc_ratio = 1.0;
- }
-
- // Underlying boost factor is based on inter intra error ratio
- r = ( boost_factor *
- ( next_frame.intra_error /
- DOUBLE_DIVIDE_CHECK(next_frame.coded_error)));
-
- if (next_frame.intra_error > cpi->gf_intra_err_min)
- r = (IIKFACTOR2 * next_frame.intra_error /
- DOUBLE_DIVIDE_CHECK(next_frame.coded_error));
- else
- r = (IIKFACTOR2 * cpi->gf_intra_err_min /
- DOUBLE_DIVIDE_CHECK(next_frame.coded_error));
-
- // Increase boost for frames where new data coming into frame
- // (eg zoom out). Slightly reduce boost if there is a net balance
- // of motion out of the frame (zoom in).
- // The range for this_frame_mv_in_out is -1.0 to +1.0
- if (this_frame_mv_in_out > 0.0)
- r += r * (this_frame_mv_in_out * 2.0);
- // In extreme case boost is halved
- else
- r += r * (this_frame_mv_in_out / 2.0);
-
- if (r > GF_RMAX)
- r = GF_RMAX;
-
- loop_decay_rate = get_prediction_decay_rate(cpi, &next_frame);
-
- // Cumulative effect of decay
- decay_accumulator = decay_accumulator * loop_decay_rate;
- decay_accumulator = decay_accumulator < 0.1 ? 0.1 : decay_accumulator;
-
- boost_score += (decay_accumulator * r);
-
- // Break clause to detect very still sections after motion
- // For example a staic image after a fade or other transition.
- if ( detect_transition_to_still( cpi, i, 5,
- loop_decay_rate, decay_accumulator ) )
- {
- allow_alt_ref = FALSE;
- boost_score = old_boost_score;
- break;
- }
-
- // Break out conditions.
- if ( /* i>4 || */
- // Break at cpi->max_gf_interval unless almost totally static
- (i >= cpi->max_gf_interval && (decay_accumulator < 0.995)) ||
- (
- // Dont break out with a very short interval
- (i > MIN_GF_INTERVAL) &&
- // Dont break out very close to a key frame
- ((cpi->frames_to_key - i) >= MIN_GF_INTERVAL) &&
- ((boost_score > 20.0) || (next_frame.pcnt_inter < 0.75)) &&
- ((mv_ratio_accumulator > 100.0) ||
- (abs_mv_in_out_accumulator > 3.0) ||
- (mv_in_out_accumulator < -2.0) ||
- ((boost_score - old_boost_score) < 2.0))
- ) )
- {
- boost_score = old_boost_score;
- break;
- }
-
- vpx_memcpy(this_frame, &next_frame, sizeof(*this_frame));
-
- old_boost_score = boost_score;
- }
-
- cpi->gf_decay_rate =
- (i > 0) ? (int)(100.0 * (1.0 - decay_accumulator)) / i : 0;
-
- // When using CBR apply additional buffer related upper limits
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- double max_boost;
-
- // For cbr apply buffer related limits
- if (cpi->drop_frames_allowed)
- {
- int df_buffer_level = cpi->oxcf.drop_frames_water_mark *
- (cpi->oxcf.optimal_buffer_level / 100);
-
- if (cpi->buffer_level > df_buffer_level)
- max_boost = ((double)((cpi->buffer_level - df_buffer_level) * 2 / 3) * 16.0) / DOUBLE_DIVIDE_CHECK((double)cpi->av_per_frame_bandwidth);
- else
- max_boost = 0.0;
- }
- else if (cpi->buffer_level > 0)
- {
- max_boost = ((double)(cpi->buffer_level * 2 / 3) * 16.0) / DOUBLE_DIVIDE_CHECK((double)cpi->av_per_frame_bandwidth);
- }
- else
- {
- max_boost = 0.0;
- }
-
- if (boost_score > max_boost)
- boost_score = max_boost;
- }
-
- cpi->gfu_boost = (int)(boost_score * 100.0) >> 4;
-
- // Should we use the alternate refernce frame
- if (allow_alt_ref &&
- (i >= MIN_GF_INTERVAL) &&
- // dont use ARF very near next kf
- (i <= (cpi->frames_to_key - MIN_GF_INTERVAL)) &&
- (((next_frame.pcnt_inter > 0.75) &&
- ((mv_in_out_accumulator / (double)i > -0.2) || (mv_in_out_accumulator > -2.0)) &&
- //(cpi->gfu_boost>150) &&
- (cpi->gfu_boost > 100) &&
- //(cpi->gfu_boost>AF_THRESH2) &&
- //((cpi->gfu_boost/i)>AF_THRESH) &&
- //(decay_accumulator > 0.5) &&
- (cpi->gf_decay_rate <= (ARF_DECAY_THRESH + (cpi->gfu_boost / 200)))
- )
- )
- )
- {
- int Boost;
- int allocation_chunks;
- int Q = (cpi->oxcf.fixed_q < 0) ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
- int tmp_q;
- int arf_frame_bits = 0;
- int group_bits;
-
- // Estimate the bits to be allocated to the group as a whole
- if ((cpi->kf_group_bits > 0) && (cpi->kf_group_error_left > 0))
- group_bits = (int)((double)cpi->kf_group_bits * (gf_group_err / (double)cpi->kf_group_error_left));
- else
- group_bits = 0;
-
- // Boost for arf frame
- Boost = (cpi->gfu_boost * 3 * GFQ_ADJUSTMENT) / (2 * 100);
- Boost += (i * 50);
- allocation_chunks = (i * 100) + Boost;
-
- // Normalize Altboost and allocations chunck down to prevent overflow
- while (Boost > 1000)
- {
- Boost /= 2;
- allocation_chunks /= 2;
- }
-
- // Calculate the number of bits to be spent on the arf based on the boost number
- arf_frame_bits = (int)((double)Boost * (group_bits / (double)allocation_chunks));
-
- // Estimate if there are enough bits available to make worthwhile use of an arf.
- tmp_q = estimate_q(cpi, mod_frame_err, (int)arf_frame_bits);
-
- // Only use an arf if it is likely we will be able to code it at a lower Q than the surrounding frames.
- if (tmp_q < cpi->worst_quality)
- {
- int half_gf_int;
- int frames_after_arf;
- int frames_bwd = cpi->oxcf.arnr_max_frames - 1;
- int frames_fwd = cpi->oxcf.arnr_max_frames - 1;
-
- cpi->source_alt_ref_pending = TRUE;
-
- // For alt ref frames the error score for the end frame of the group (the alt ref frame) should not contribute to the group total and hence
- // the number of bit allocated to the group. Rather it forms part of the next group (it is the GF at the start of the next group)
- gf_group_err -= mod_frame_err;
-
- // Set the interval till the next gf or arf. For ARFs this is the number of frames to be coded before the future frame that is coded as an ARF.
- // The future frame itself is part of the next group
- cpi->baseline_gf_interval = i - 1;
-
- // Define the arnr filter width for this group of frames:
- // We only filter frames that lie within a distance of half
- // the GF interval from the ARF frame. We also have to trap
- // cases where the filter extends beyond the end of clip.
- // Note: this_frame->frame has been updated in the loop
- // so it now points at the ARF frame.
- half_gf_int = cpi->baseline_gf_interval >> 1;
- frames_after_arf = cpi->total_stats->count - this_frame->frame - 1;
-
- switch (cpi->oxcf.arnr_type)
- {
- case 1: // Backward filter
- frames_fwd = 0;
- if (frames_bwd > half_gf_int)
- frames_bwd = half_gf_int;
- break;
-
- case 2: // Forward filter
- if (frames_fwd > half_gf_int)
- frames_fwd = half_gf_int;
- if (frames_fwd > frames_after_arf)
- frames_fwd = frames_after_arf;
- frames_bwd = 0;
- break;
-
- case 3: // Centered filter
- default:
- frames_fwd >>= 1;
- if (frames_fwd > frames_after_arf)
- frames_fwd = frames_after_arf;
- if (frames_fwd > half_gf_int)
- frames_fwd = half_gf_int;
-
- frames_bwd = frames_fwd;
-
- // For even length filter there is one more frame backward
- // than forward: e.g. len=6 ==> bbbAff, len=7 ==> bbbAfff.
- if (frames_bwd < half_gf_int)
- frames_bwd += (cpi->oxcf.arnr_max_frames+1) & 0x1;
- break;
- }
-
- cpi->active_arnr_frames = frames_bwd + 1 + frames_fwd;
- }
- else
- {
- cpi->source_alt_ref_pending = FALSE;
- cpi->baseline_gf_interval = i;
- }
- }
- else
- {
- cpi->source_alt_ref_pending = FALSE;
- cpi->baseline_gf_interval = i;
- }
-
- // Conventional GF
- if (!cpi->source_alt_ref_pending)
- {
- // Dont allow conventional gf too near the next kf
- if ((cpi->frames_to_key - cpi->baseline_gf_interval) < MIN_GF_INTERVAL)
- {
- while (cpi->baseline_gf_interval < cpi->frames_to_key)
- {
- if (EOF == input_stats(cpi, this_frame))
- break;
-
- cpi->baseline_gf_interval++;
-
- if (cpi->baseline_gf_interval < cpi->frames_to_key)
- gf_group_err += calculate_modified_err(cpi, this_frame);
- }
- }
- }
-
- // Now decide how many bits should be allocated to the GF group as a proportion of those remaining in the kf group.
- // The final key frame group in the clip is treated as a special case where cpi->kf_group_bits is tied to cpi->bits_left.
- // This is also important for short clips where there may only be one key frame.
- if (cpi->frames_to_key >= (int)(cpi->total_stats->count - cpi->common.current_video_frame))
- {
- cpi->kf_group_bits = (cpi->bits_left > 0) ? cpi->bits_left : 0;
- }
-
- // Calculate the bits to be allocated to the group as a whole
- if ((cpi->kf_group_bits > 0) && (cpi->kf_group_error_left > 0))
- cpi->gf_group_bits = (int)((double)cpi->kf_group_bits * (gf_group_err / (double)cpi->kf_group_error_left));
- else
- cpi->gf_group_bits = 0;
-
- cpi->gf_group_bits = (cpi->gf_group_bits < 0) ? 0 : (cpi->gf_group_bits > cpi->kf_group_bits) ? cpi->kf_group_bits : cpi->gf_group_bits;
-
- // Clip cpi->gf_group_bits based on user supplied data rate variability limit (cpi->oxcf.two_pass_vbrmax_section)
- if (cpi->gf_group_bits > max_bits * cpi->baseline_gf_interval)
- cpi->gf_group_bits = max_bits * cpi->baseline_gf_interval;
-
- // Reset the file position
- reset_fpf_position(cpi, start_pos);
-
- // Update the record of error used so far (only done once per gf group)
- cpi->modified_error_used += gf_group_err;
-
- // Assign bits to the arf or gf.
- {
- int Boost;
- int frames_in_section;
- int allocation_chunks;
- int Q = (cpi->oxcf.fixed_q < 0) ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
-
- // For ARF frames
- if (cpi->source_alt_ref_pending)
- {
- Boost = (cpi->gfu_boost * 3 * GFQ_ADJUSTMENT) / (2 * 100);
- //Boost += (cpi->baseline_gf_interval * 25);
- Boost += (cpi->baseline_gf_interval * 50);
-
- // Set max and minimum boost and hence minimum allocation
- if (Boost > ((cpi->baseline_gf_interval + 1) * 200))
- Boost = ((cpi->baseline_gf_interval + 1) * 200);
- else if (Boost < 125)
- Boost = 125;
-
- frames_in_section = cpi->baseline_gf_interval + 1;
- allocation_chunks = (frames_in_section * 100) + Boost;
- }
- // Else for standard golden frames
- else
- {
- // boost based on inter / intra ratio of subsequent frames
- Boost = (cpi->gfu_boost * GFQ_ADJUSTMENT) / 100;
-
- // Set max and minimum boost and hence minimum allocation
- if (Boost > (cpi->baseline_gf_interval * 150))
- Boost = (cpi->baseline_gf_interval * 150);
- else if (Boost < 125)
- Boost = 125;
-
- frames_in_section = cpi->baseline_gf_interval;
- allocation_chunks = (frames_in_section * 100) + (Boost - 100);
- }
-
- // Normalize Altboost and allocations chunck down to prevent overflow
- while (Boost > 1000)
- {
- Boost /= 2;
- allocation_chunks /= 2;
- }
-
- // Calculate the number of bits to be spent on the gf or arf based on the boost number
- cpi->gf_bits = (int)((double)Boost * (cpi->gf_group_bits / (double)allocation_chunks));
-
- // If the frame that is to be boosted is simpler than the average for
- // the gf/arf group then use an alternative calculation
- // based on the error score of the frame itself
- if (mod_frame_err < gf_group_err / (double)cpi->baseline_gf_interval)
- {
- double alt_gf_grp_bits;
- int alt_gf_bits;
-
- alt_gf_grp_bits =
- (double)cpi->kf_group_bits *
- (mod_frame_err * (double)cpi->baseline_gf_interval) /
- DOUBLE_DIVIDE_CHECK((double)cpi->kf_group_error_left);
-
- alt_gf_bits = (int)((double)Boost * (alt_gf_grp_bits /
- (double)allocation_chunks));
-
- if (cpi->gf_bits > alt_gf_bits)
- {
- cpi->gf_bits = alt_gf_bits;
- }
- }
- // Else if it is harder than other frames in the group make sure it at
- // least receives an allocation in keeping with its relative error
- // score, otherwise it may be worse off than an "un-boosted" frame
- else
- {
- int alt_gf_bits =
- (int)((double)cpi->kf_group_bits *
- mod_frame_err /
- DOUBLE_DIVIDE_CHECK((double)cpi->kf_group_error_left));
-
- if (alt_gf_bits > cpi->gf_bits)
- {
- cpi->gf_bits = alt_gf_bits;
- }
- }
-
- // Apply an additional limit for CBR
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- if (cpi->gf_bits > (cpi->buffer_level >> 1))
- cpi->gf_bits = cpi->buffer_level >> 1;
- }
-
- // Dont allow a negative value for gf_bits
- if (cpi->gf_bits < 0)
- cpi->gf_bits = 0;
-
- // Adjust KF group bits and error remainin
- cpi->kf_group_error_left -= gf_group_err;
- cpi->kf_group_bits -= cpi->gf_group_bits;
-
- if (cpi->kf_group_bits < 0)
- cpi->kf_group_bits = 0;
-
- // Note the error score left in the remaining frames of the group.
- // For normal GFs we want to remove the error score for the first frame of the group (except in Key frame case where this has already happened)
- if (!cpi->source_alt_ref_pending && cpi->common.frame_type != KEY_FRAME)
- cpi->gf_group_error_left = gf_group_err - gf_first_frame_err;
- else
- cpi->gf_group_error_left = gf_group_err;
-
- cpi->gf_group_bits -= cpi->gf_bits;
-
- if (cpi->gf_group_bits < 0)
- cpi->gf_group_bits = 0;
-
- // Set aside some bits for a mid gf sequence boost
- if ((cpi->gfu_boost > 150) && (cpi->baseline_gf_interval > 5))
- {
- int pct_extra = (cpi->gfu_boost - 100) / 50;
- pct_extra = (pct_extra > 10) ? 10 : pct_extra;
-
- cpi->mid_gf_extra_bits = (cpi->gf_group_bits * pct_extra) / 100;
- cpi->gf_group_bits -= cpi->mid_gf_extra_bits;
- }
- else
- cpi->mid_gf_extra_bits = 0;
-
- cpi->gf_bits += cpi->min_frame_bandwidth; // Add in minimum for a frame
- }
-
- if (!cpi->source_alt_ref_pending && (cpi->common.frame_type != KEY_FRAME)) // Normal GF and not a KF
- {
- cpi->per_frame_bandwidth = cpi->gf_bits; // Per frame bit target for this frame
- }
-
- // Adjustment to estimate_max_q based on a measure of complexity of the section
- if (cpi->common.frame_type != KEY_FRAME)
- {
- FIRSTPASS_STATS sectionstats;
- double Ratio;
-
- zero_stats(§ionstats);
- reset_fpf_position(cpi, start_pos);
-
- for (i = 0 ; i < cpi->baseline_gf_interval ; i++)
- {
- input_stats(cpi, &next_frame);
- accumulate_stats(§ionstats, &next_frame);
- }
-
- avg_stats(§ionstats);
-
- cpi->section_intra_rating =
- sectionstats.intra_error /
- DOUBLE_DIVIDE_CHECK(sectionstats.coded_error);
-
- Ratio = sectionstats.intra_error / DOUBLE_DIVIDE_CHECK(sectionstats.coded_error);
- //if( (Ratio > 11) ) //&& (sectionstats.pcnt_second_ref < .20) )
- //{
- cpi->section_max_qfactor = 1.0 - ((Ratio - 10.0) * 0.025);
-
- if (cpi->section_max_qfactor < 0.80)
- cpi->section_max_qfactor = 0.80;
-
- //}
- //else
- // cpi->section_max_qfactor = 1.0;
-
- reset_fpf_position(cpi, start_pos);
- }
-}
-
-// Allocate bits to a normal frame that is neither a gf an arf or a key frame.
-static void assign_std_frame_bits(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
-{
- int target_frame_size; // gf_group_error_left
-
- double modified_err;
- double err_fraction; // What portion of the remaining GF group error is used by this frame
-
- int max_bits = frame_max_bits(cpi); // Max for a single frame
-
- // The final few frames have special treatment
- if (cpi->frames_till_gf_update_due >= (int)(cpi->total_stats->count - cpi->common.current_video_frame))
- {
- cpi->gf_group_bits = (cpi->bits_left > 0) ? cpi->bits_left : 0;;
- }
-
- // Calculate modified prediction error used in bit allocation
- modified_err = calculate_modified_err(cpi, this_frame);
-
- if (cpi->gf_group_error_left > 0)
- err_fraction = modified_err / cpi->gf_group_error_left; // What portion of the remaining GF group error is used by this frame
- else
- err_fraction = 0.0;
-
- target_frame_size = (int)((double)cpi->gf_group_bits * err_fraction); // How many of those bits available for allocation should we give it?
-
- // Clip to target size to 0 - max_bits (or cpi->gf_group_bits) at the top end.
- if (target_frame_size < 0)
- target_frame_size = 0;
- else
- {
- if (target_frame_size > max_bits)
- target_frame_size = max_bits;
-
- if (target_frame_size > cpi->gf_group_bits)
- target_frame_size = cpi->gf_group_bits;
- }
-
- cpi->gf_group_error_left -= modified_err; // Adjust error remaining
- cpi->gf_group_bits -= target_frame_size; // Adjust bits remaining
-
- if (cpi->gf_group_bits < 0)
- cpi->gf_group_bits = 0;
-
- target_frame_size += cpi->min_frame_bandwidth; // Add in the minimum number of bits that is set aside for every frame.
-
- // Special case for the frame that lies half way between two gfs
- if (cpi->common.frames_since_golden == cpi->baseline_gf_interval / 2)
- target_frame_size += cpi->mid_gf_extra_bits;
-
- cpi->per_frame_bandwidth = target_frame_size; // Per frame bit target for this frame
-}
-
-void vp8_second_pass(VP8_COMP *cpi)
-{
- int tmp_q;
- int frames_left = (int)(cpi->total_stats->count - cpi->common.current_video_frame);
-
- FIRSTPASS_STATS this_frame;
- FIRSTPASS_STATS this_frame_copy;
-
- VP8_COMMON *cm = &cpi->common;
-
- double this_frame_error;
- double this_frame_intra_error;
- double this_frame_coded_error;
-
- FIRSTPASS_STATS *start_pos;
-
- if (!cpi->stats_in)
- {
- return ;
- }
-
- vp8_clear_system_state();
-
- if (EOF == input_stats(cpi, &this_frame))
- return;
-
- this_frame_error = this_frame.ssim_weighted_pred_err;
- this_frame_intra_error = this_frame.intra_error;
- this_frame_coded_error = this_frame.coded_error;
-
- // Store information regarding level of motion etc for use mode decisions.
- cpi->motion_speed = (int)(fabs(this_frame.MVr) + fabs(this_frame.MVc));
- cpi->motion_var = (int)(fabs(this_frame.MVrv) + fabs(this_frame.MVcv));
- cpi->inter_lvl = (int)(this_frame.pcnt_inter * 100);
- cpi->intra_lvl = (int)((1.0 - this_frame.pcnt_inter) * 100);
- cpi->motion_lvl = (int)(this_frame.pcnt_motion * 100);
-
- start_pos = cpi->stats_in;
-
- // keyframe and section processing !
- if (cpi->frames_to_key == 0)
- {
- // Define next KF group and assign bits to it
- vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
- find_next_key_frame(cpi, &this_frame_copy);
-
- // Special case: Error error_resilient_mode mode does not make much sense for two pass but with its current meaning but this code is designed to stop
- // outlandish behaviour if someone does set it when using two pass. It effectively disables GF groups.
- // This is temporary code till we decide what should really happen in this case.
- if (cpi->oxcf.error_resilient_mode)
- {
- cpi->gf_group_bits = cpi->kf_group_bits;
- cpi->gf_group_error_left = cpi->kf_group_error_left;
- cpi->baseline_gf_interval = cpi->frames_to_key;
- cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
- cpi->source_alt_ref_pending = FALSE;
- }
-
- }
-
- // Is this a GF / ARF (Note that a KF is always also a GF)
- if (cpi->frames_till_gf_update_due == 0)
- {
- // Update monitor of the bits per error observed so far.
- // Done once per gf group based on what has gone before
- // so do nothing if this is the first frame.
- if (cpi->common.current_video_frame > 0)
- {
- cpi->observed_bpe =
- (double)(cpi->clip_bits_total - cpi->bits_left) /
- cpi->modified_error_used;
- }
-
- // Define next gf group and assign bits to it
- vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
- define_gf_group(cpi, &this_frame_copy);
-
- // If we are going to code an altref frame at the end of the group and the current frame is not a key frame....
- // If the previous group used an arf this frame has already benefited from that arf boost and it should not be given extra bits
- // If the previous group was NOT coded using arf we may want to apply some boost to this GF as well
- if (cpi->source_alt_ref_pending && (cpi->common.frame_type != KEY_FRAME))
- {
- // Assign a standard frames worth of bits from those allocated to the GF group
- vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
- assign_std_frame_bits(cpi, &this_frame_copy);
-
- // If appropriate (we are switching into ARF active but it was not previously active) apply a boost for the gf at the start of the group.
- //if ( !cpi->source_alt_ref_active && (cpi->gfu_boost > 150) )
- if (FALSE)
- {
- int extra_bits;
- int pct_extra = (cpi->gfu_boost - 100) / 50;
-
- pct_extra = (pct_extra > 20) ? 20 : pct_extra;
-
- extra_bits = (cpi->gf_group_bits * pct_extra) / 100;
- cpi->gf_group_bits -= extra_bits;
- cpi->per_frame_bandwidth += extra_bits;
- }
- }
- }
-
- // Otherwise this is an ordinary frame
- else
- {
- // Special case: Error error_resilient_mode mode does not make much sense for two pass but with its current meaning but this code is designed to stop
- // outlandish behaviour if someone does set it when using two pass. It effectively disables GF groups.
- // This is temporary code till we decide what should really happen in this case.
- if (cpi->oxcf.error_resilient_mode)
- {
- cpi->frames_till_gf_update_due = cpi->frames_to_key;
-
- if (cpi->common.frame_type != KEY_FRAME)
- {
- // Assign bits from those allocated to the GF group
- vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
- assign_std_frame_bits(cpi, &this_frame_copy);
- }
- }
- else
- {
- // Assign bits from those allocated to the GF group
- vpx_memcpy(&this_frame_copy, &this_frame, sizeof(this_frame));
- assign_std_frame_bits(cpi, &this_frame_copy);
- }
- }
-
- // Keep a globally available copy of this and the next frame's iiratio.
- cpi->this_iiratio = this_frame_intra_error /
- DOUBLE_DIVIDE_CHECK(this_frame_coded_error);
- {
- FIRSTPASS_STATS next_frame;
- if ( lookup_next_frame_stats(cpi, &next_frame) != EOF )
- {
- cpi->next_iiratio = next_frame.intra_error /
- DOUBLE_DIVIDE_CHECK(next_frame.coded_error);
- }
- }
-
- // Set nominal per second bandwidth for this frame
- cpi->target_bandwidth = cpi->per_frame_bandwidth * cpi->output_frame_rate;
- if (cpi->target_bandwidth < 0)
- cpi->target_bandwidth = 0;
-
- if (cpi->common.current_video_frame == 0)
- {
- cpi->est_max_qcorrection_factor = 1.0;
-
- // Experimental code to try and set a cq_level in constrained
- // quality mode.
- if ( cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY )
- {
- int est_cq;
-
- est_cq =
- estimate_cq( cpi,
- (cpi->total_coded_error_left / frames_left),
- (int)(cpi->bits_left / frames_left));
-
- cpi->cq_target_quality = cpi->oxcf.cq_level;
- if ( est_cq > cpi->cq_target_quality )
- cpi->cq_target_quality = est_cq;
- }
-
- // guess at maxq needed in 2nd pass
- cpi->maxq_max_limit = cpi->worst_quality;
- cpi->maxq_min_limit = cpi->best_quality;
- tmp_q = estimate_max_q( cpi,
- (cpi->total_coded_error_left / frames_left),
- (int)(cpi->bits_left / frames_left));
-
- // Limit the maxq value returned subsequently.
- // This increases the risk of overspend or underspend if the initial
- // estimate for the clip is bad, but helps prevent excessive
- // variation in Q, especially near the end of a clip
- // where for example a small overspend may cause Q to crash
- cpi->maxq_max_limit = ((tmp_q + 32) < cpi->worst_quality)
- ? (tmp_q + 32) : cpi->worst_quality;
- cpi->maxq_min_limit = ((tmp_q - 32) > cpi->best_quality)
- ? (tmp_q - 32) : cpi->best_quality;
-
- cpi->active_worst_quality = tmp_q;
- cpi->ni_av_qi = tmp_q;
- }
-
- // The last few frames of a clip almost always have to few or too many
- // bits and for the sake of over exact rate control we dont want to make
- // radical adjustments to the allowed quantizer range just to use up a
- // few surplus bits or get beneath the target rate.
- else if ( (cpi->common.current_video_frame <
- (((unsigned int)cpi->total_stats->count * 255)>>8)) &&
- ((cpi->common.current_video_frame + cpi->baseline_gf_interval) <
- (unsigned int)cpi->total_stats->count) )
- {
- if (frames_left < 1)
- frames_left = 1;
-
- tmp_q = estimate_max_q(cpi, (cpi->total_coded_error_left / frames_left), (int)(cpi->bits_left / frames_left));
-
- // Move active_worst_quality but in a damped way
- if (tmp_q > cpi->active_worst_quality)
- cpi->active_worst_quality ++;
- else if (tmp_q < cpi->active_worst_quality)
- cpi->active_worst_quality --;
-
- cpi->active_worst_quality = ((cpi->active_worst_quality * 3) + tmp_q + 2) / 4;
- }
-
- cpi->frames_to_key --;
- cpi->total_error_left -= this_frame_error;
- cpi->total_intra_error_left -= this_frame_intra_error;
- cpi->total_coded_error_left -= this_frame_coded_error;
-}
-
-
-static BOOL test_candidate_kf(VP8_COMP *cpi, FIRSTPASS_STATS *last_frame, FIRSTPASS_STATS *this_frame, FIRSTPASS_STATS *next_frame)
-{
- BOOL is_viable_kf = FALSE;
-
- // Does the frame satisfy the primary criteria of a key frame
- // If so, then examine how well it predicts subsequent frames
- if ((this_frame->pcnt_second_ref < 0.10) &&
- (next_frame->pcnt_second_ref < 0.10) &&
- ((this_frame->pcnt_inter < 0.05) ||
- (
- ((this_frame->pcnt_inter - this_frame->pcnt_neutral) < .25) &&
- ((this_frame->intra_error / DOUBLE_DIVIDE_CHECK(this_frame->coded_error)) < 2.5) &&
- ((fabs(last_frame->coded_error - this_frame->coded_error) / DOUBLE_DIVIDE_CHECK(this_frame->coded_error) > .40) ||
- (fabs(last_frame->intra_error - this_frame->intra_error) / DOUBLE_DIVIDE_CHECK(this_frame->intra_error) > .40) ||
- ((next_frame->intra_error / DOUBLE_DIVIDE_CHECK(next_frame->coded_error)) > 3.5)
- )
- )
- )
- )
- {
- int i;
- FIRSTPASS_STATS *start_pos;
-
- FIRSTPASS_STATS local_next_frame;
-
- double boost_score = 0.0;
- double old_boost_score = 0.0;
- double decay_accumulator = 1.0;
- double next_iiratio;
-
- vpx_memcpy(&local_next_frame, next_frame, sizeof(*next_frame));
-
- // Note the starting file position so we can reset to it
- start_pos = cpi->stats_in;
-
- // Examine how well the key frame predicts subsequent frames
- for (i = 0 ; i < 16; i++)
- {
- next_iiratio = (IIKFACTOR1 * local_next_frame.intra_error / DOUBLE_DIVIDE_CHECK(local_next_frame.coded_error)) ;
-
- if (next_iiratio > RMAX)
- next_iiratio = RMAX;
-
- // Cumulative effect of decay in prediction quality
- if (local_next_frame.pcnt_inter > 0.85)
- decay_accumulator = decay_accumulator * local_next_frame.pcnt_inter;
- else
- decay_accumulator = decay_accumulator * ((0.85 + local_next_frame.pcnt_inter) / 2.0);
-
- //decay_accumulator = decay_accumulator * local_next_frame.pcnt_inter;
-
- // Keep a running total
- boost_score += (decay_accumulator * next_iiratio);
-
- // Test various breakout clauses
- if ((local_next_frame.pcnt_inter < 0.05) ||
- (next_iiratio < 1.5) ||
- (((local_next_frame.pcnt_inter -
- local_next_frame.pcnt_neutral) < 0.20) &&
- (next_iiratio < 3.0)) ||
- ((boost_score - old_boost_score) < 0.5) ||
- (local_next_frame.intra_error < 200)
- )
- {
- break;
- }
-
- old_boost_score = boost_score;
-
- // Get the next frame details
- if (EOF == input_stats(cpi, &local_next_frame))
- break;
- }
-
- // If there is tolerable prediction for at least the next 3 frames then break out else discard this pottential key frame and move on
- if (boost_score > 5.0 && (i > 3))
- is_viable_kf = TRUE;
- else
- {
- // Reset the file position
- reset_fpf_position(cpi, start_pos);
-
- is_viable_kf = FALSE;
- }
- }
-
- return is_viable_kf;
-}
-static void find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
-{
- int i,j;
- FIRSTPASS_STATS last_frame;
- FIRSTPASS_STATS first_frame;
- FIRSTPASS_STATS next_frame;
- FIRSTPASS_STATS *start_position;
-
- double decay_accumulator = 1.0;
- double boost_score = 0;
- double old_boost_score = 0.0;
- double loop_decay_rate;
-
- double kf_mod_err = 0.0;
- double kf_group_err = 0.0;
- double kf_group_intra_err = 0.0;
- double kf_group_coded_err = 0.0;
- double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100);
- double recent_loop_decay[8] = {1.0,1.0,1.0,1.0,1.0,1.0,1.0,1.0};
-
- vpx_memset(&next_frame, 0, sizeof(next_frame)); // assure clean
-
- vp8_clear_system_state(); //__asm emms;
- start_position = cpi->stats_in;
-
- cpi->common.frame_type = KEY_FRAME;
-
- // is this a forced key frame by interval
- cpi->this_key_frame_forced = cpi->next_key_frame_forced;
-
- // Clear the alt ref active flag as this can never be active on a key frame
- cpi->source_alt_ref_active = FALSE;
-
- // Kf is always a gf so clear frames till next gf counter
- cpi->frames_till_gf_update_due = 0;
-
- cpi->frames_to_key = 1;
-
- // Take a copy of the initial frame details
- vpx_memcpy(&first_frame, this_frame, sizeof(*this_frame));
-
- cpi->kf_group_bits = 0; // Total bits avaialable to kf group
- cpi->kf_group_error_left = 0; // Group modified error score.
-
- kf_mod_err = calculate_modified_err(cpi, this_frame);
-
- // find the next keyframe
- i = 0;
- while (cpi->stats_in < cpi->stats_in_end)
- {
- // Accumulate kf group error
- kf_group_err += calculate_modified_err(cpi, this_frame);
-
- // These figures keep intra and coded error counts for all frames including key frames in the group.
- // The effect of the key frame itself can be subtracted out using the first_frame data collected above
- kf_group_intra_err += this_frame->intra_error;
- kf_group_coded_err += this_frame->coded_error;
-
- // load a the next frame's stats
- vpx_memcpy(&last_frame, this_frame, sizeof(*this_frame));
- input_stats(cpi, this_frame);
-
- // Provided that we are not at the end of the file...
- if (cpi->oxcf.auto_key
- && lookup_next_frame_stats(cpi, &next_frame) != EOF)
- {
- // Normal scene cut check
- if (test_candidate_kf(cpi, &last_frame, this_frame, &next_frame))
- break;
-
- // How fast is prediction quality decaying
- loop_decay_rate = get_prediction_decay_rate(cpi, &next_frame);
-
- // We want to know something about the recent past... rather than
- // as used elsewhere where we are concened with decay in prediction
- // quality since the last GF or KF.
- recent_loop_decay[i%8] = loop_decay_rate;
- decay_accumulator = 1.0;
- for (j = 0; j < 8; j++)
- {
- decay_accumulator = decay_accumulator * recent_loop_decay[j];
- }
-
- // Special check for transition or high motion followed by a
- // to a static scene.
- if ( detect_transition_to_still( cpi, i,
- (cpi->key_frame_frequency-i),
- loop_decay_rate,
- decay_accumulator ) )
- {
- break;
- }
-
-
- // Step on to the next frame
- cpi->frames_to_key ++;
-
- // If we don't have a real key frame within the next two
- // forcekeyframeevery intervals then break out of the loop.
- if (cpi->frames_to_key >= 2 *(int)cpi->key_frame_frequency)
- break;
- } else
- cpi->frames_to_key ++;
-
- i++;
- }
-
- // If there is a max kf interval set by the user we must obey it.
- // We already breakout of the loop above at 2x max.
- // This code centers the extra kf if the actual natural
- // interval is between 1x and 2x
- if (cpi->oxcf.auto_key
- && cpi->frames_to_key > (int)cpi->key_frame_frequency )
- {
- FIRSTPASS_STATS *current_pos = cpi->stats_in;
- FIRSTPASS_STATS tmp_frame;
-
- cpi->frames_to_key /= 2;
-
- // Copy first frame details
- vpx_memcpy(&tmp_frame, &first_frame, sizeof(first_frame));
-
- // Reset to the start of the group
- reset_fpf_position(cpi, start_position);
-
- kf_group_err = 0;
- kf_group_intra_err = 0;
- kf_group_coded_err = 0;
-
- // Rescan to get the correct error data for the forced kf group
- for( i = 0; i < cpi->frames_to_key; i++ )
- {
- // Accumulate kf group errors
- kf_group_err += calculate_modified_err(cpi, &tmp_frame);
- kf_group_intra_err += tmp_frame.intra_error;
- kf_group_coded_err += tmp_frame.coded_error;
-
- // Load a the next frame's stats
- input_stats(cpi, &tmp_frame);
- }
-
- // Reset to the start of the group
- reset_fpf_position(cpi, current_pos);
-
- cpi->next_key_frame_forced = TRUE;
- }
- else
- cpi->next_key_frame_forced = FALSE;
-
- // Special case for the last frame of the file
- if (cpi->stats_in >= cpi->stats_in_end)
- {
- // Accumulate kf group error
- kf_group_err += calculate_modified_err(cpi, this_frame);
-
- // These figures keep intra and coded error counts for all frames including key frames in the group.
- // The effect of the key frame itself can be subtracted out using the first_frame data collected above
- kf_group_intra_err += this_frame->intra_error;
- kf_group_coded_err += this_frame->coded_error;
- }
-
- // Calculate the number of bits that should be assigned to the kf group.
- if ((cpi->bits_left > 0) && (cpi->modified_error_left > 0.0))
- {
- // Max for a single normal frame (not key frame)
- int max_bits = frame_max_bits(cpi);
-
- // Maximum bits for the kf group
- long long max_grp_bits;
-
- // Default allocation based on bits left and relative
- // complexity of the section
- cpi->kf_group_bits = (long long)( cpi->bits_left *
- ( kf_group_err /
- cpi->modified_error_left ));
-
- // Clip based on maximum per frame rate defined by the user.
- max_grp_bits = (long long)max_bits * (long long)cpi->frames_to_key;
- if (cpi->kf_group_bits > max_grp_bits)
- cpi->kf_group_bits = max_grp_bits;
-
- // Additional special case for CBR if buffer is getting full.
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- int opt_buffer_lvl = cpi->oxcf.optimal_buffer_level;
- int buffer_lvl = cpi->buffer_level;
-
- // If the buffer is near or above the optimal and this kf group is
- // not being allocated much then increase the allocation a bit.
- if (buffer_lvl >= opt_buffer_lvl)
- {
- int high_water_mark = (opt_buffer_lvl +
- cpi->oxcf.maximum_buffer_size) >> 1;
-
- long long av_group_bits;
-
- // Av bits per frame * number of frames
- av_group_bits = (long long)cpi->av_per_frame_bandwidth *
- (long long)cpi->frames_to_key;
-
- // We are at or above the maximum.
- if (cpi->buffer_level >= high_water_mark)
- {
- long long min_group_bits;
-
- min_group_bits = av_group_bits +
- (long long)(buffer_lvl -
- high_water_mark);
-
- if (cpi->kf_group_bits < min_group_bits)
- cpi->kf_group_bits = min_group_bits;
- }
- // We are above optimal but below the maximum
- else if (cpi->kf_group_bits < av_group_bits)
- {
- long long bits_below_av = av_group_bits -
- cpi->kf_group_bits;
-
- cpi->kf_group_bits +=
- (long long)((double)bits_below_av *
- (double)(buffer_lvl - opt_buffer_lvl) /
- (double)(high_water_mark - opt_buffer_lvl));
- }
- }
- }
- }
- else
- cpi->kf_group_bits = 0;
-
- // Reset the first pass file position
- reset_fpf_position(cpi, start_position);
-
- // determine how big to make this keyframe based on how well the subsequent frames use inter blocks
- decay_accumulator = 1.0;
- boost_score = 0.0;
- loop_decay_rate = 1.00; // Starting decay rate
-
- for (i = 0 ; i < cpi->frames_to_key ; i++)
- {
- double r;
- double motion_decay;
- double motion_pct;
-
- if (EOF == input_stats(cpi, &next_frame))
- break;
-
- if (next_frame.intra_error > cpi->kf_intra_err_min)
- r = (IIKFACTOR2 * next_frame.intra_error /
- DOUBLE_DIVIDE_CHECK(next_frame.coded_error));
- else
- r = (IIKFACTOR2 * cpi->kf_intra_err_min /
- DOUBLE_DIVIDE_CHECK(next_frame.coded_error));
-
- if (r > RMAX)
- r = RMAX;
-
- // How fast is prediction quality decaying
- loop_decay_rate = get_prediction_decay_rate(cpi, &next_frame);
-
- decay_accumulator = decay_accumulator * loop_decay_rate;
- decay_accumulator = decay_accumulator < 0.1 ? 0.1 : decay_accumulator;
-
- boost_score += (decay_accumulator * r);
-
- if ((i > MIN_GF_INTERVAL) &&
- ((boost_score - old_boost_score) < 1.0))
- {
- break;
- }
-
- old_boost_score = boost_score;
- }
-
- if (1)
- {
- FIRSTPASS_STATS sectionstats;
- double Ratio;
-
- zero_stats(§ionstats);
- reset_fpf_position(cpi, start_position);
-
- for (i = 0 ; i < cpi->frames_to_key ; i++)
- {
- input_stats(cpi, &next_frame);
- accumulate_stats(§ionstats, &next_frame);
- }
-
- avg_stats(§ionstats);
-
- cpi->section_intra_rating = sectionstats.intra_error / DOUBLE_DIVIDE_CHECK(sectionstats.coded_error);
-
- Ratio = sectionstats.intra_error / DOUBLE_DIVIDE_CHECK(sectionstats.coded_error);
- // if( (Ratio > 11) ) //&& (sectionstats.pcnt_second_ref < .20) )
- //{
- cpi->section_max_qfactor = 1.0 - ((Ratio - 10.0) * 0.025);
-
- if (cpi->section_max_qfactor < 0.80)
- cpi->section_max_qfactor = 0.80;
-
- //}
- //else
- // cpi->section_max_qfactor = 1.0;
- }
-
- // When using CBR apply additional buffer fullness related upper limits
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- double max_boost;
-
- if (cpi->drop_frames_allowed)
- {
- int df_buffer_level = cpi->oxcf.drop_frames_water_mark * (cpi->oxcf.optimal_buffer_level / 100);
-
- if (cpi->buffer_level > df_buffer_level)
- max_boost = ((double)((cpi->buffer_level - df_buffer_level) * 2 / 3) * 16.0) / DOUBLE_DIVIDE_CHECK((double)cpi->av_per_frame_bandwidth);
- else
- max_boost = 0.0;
- }
- else if (cpi->buffer_level > 0)
- {
- max_boost = ((double)(cpi->buffer_level * 2 / 3) * 16.0) / DOUBLE_DIVIDE_CHECK((double)cpi->av_per_frame_bandwidth);
- }
- else
- {
- max_boost = 0.0;
- }
-
- if (boost_score > max_boost)
- boost_score = max_boost;
- }
-
- // Reset the first pass file position
- reset_fpf_position(cpi, start_position);
-
- // Work out how many bits to allocate for the key frame itself
- if (1)
- {
- int kf_boost = boost_score;
- int allocation_chunks;
- int Counter = cpi->frames_to_key;
- int alt_kf_bits;
- YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx];
- // Min boost based on kf interval
-#if 0
-
- while ((kf_boost < 48) && (Counter > 0))
- {
- Counter -= 2;
- kf_boost ++;
- }
-
-#endif
-
- if (kf_boost < 48)
- {
- kf_boost += ((Counter + 1) >> 1);
-
- if (kf_boost > 48) kf_boost = 48;
- }
-
- // bigger frame sizes need larger kf boosts, smaller frames smaller boosts...
- if ((lst_yv12->y_width * lst_yv12->y_height) > (320 * 240))
- kf_boost += 2 * (lst_yv12->y_width * lst_yv12->y_height) / (320 * 240);
- else if ((lst_yv12->y_width * lst_yv12->y_height) < (320 * 240))
- kf_boost -= 4 * (320 * 240) / (lst_yv12->y_width * lst_yv12->y_height);
-
- kf_boost = (int)((double)kf_boost * 100.0) >> 4; // Scale 16 to 100
-
- // Adjustment to boost based on recent average q
- //kf_boost = kf_boost * vp8_kf_boost_qadjustment[cpi->ni_av_qi] / 100;
-
- if (kf_boost < 250) // Min KF boost
- kf_boost = 250;
-
- // We do three calculations for kf size.
- // The first is based on the error score for the whole kf group.
- // The second (optionaly) on the key frames own error if this is smaller than the average for the group.
- // The final one insures that the frame receives at least the allocation it would have received based on its own error score vs the error score remaining
-
- allocation_chunks = ((cpi->frames_to_key - 1) * 100) + kf_boost; // cpi->frames_to_key-1 because key frame itself is taken care of by kf_boost
-
- // Normalize Altboost and allocations chunck down to prevent overflow
- while (kf_boost > 1000)
- {
- kf_boost /= 2;
- allocation_chunks /= 2;
- }
-
- cpi->kf_group_bits = (cpi->kf_group_bits < 0) ? 0 : cpi->kf_group_bits;
-
- // Calculate the number of bits to be spent on the key frame
- cpi->kf_bits = (int)((double)kf_boost * ((double)cpi->kf_group_bits / (double)allocation_chunks));
-
- // Apply an additional limit for CBR
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- if (cpi->kf_bits > ((3 * cpi->buffer_level) >> 2))
- cpi->kf_bits = (3 * cpi->buffer_level) >> 2;
- }
-
- // If the key frame is actually easier than the average for the
- // kf group (which does sometimes happen... eg a blank intro frame)
- // Then use an alternate calculation based on the kf error score
- // which should give a smaller key frame.
- if (kf_mod_err < kf_group_err / cpi->frames_to_key)
- {
- double alt_kf_grp_bits =
- ((double)cpi->bits_left *
- (kf_mod_err * (double)cpi->frames_to_key) /
- DOUBLE_DIVIDE_CHECK(cpi->modified_error_left));
-
- alt_kf_bits = (int)((double)kf_boost *
- (alt_kf_grp_bits / (double)allocation_chunks));
-
- if (cpi->kf_bits > alt_kf_bits)
- {
- cpi->kf_bits = alt_kf_bits;
- }
- }
- // Else if it is much harder than other frames in the group make sure
- // it at least receives an allocation in keeping with its relative
- // error score
- else
- {
- alt_kf_bits =
- (int)((double)cpi->bits_left *
- (kf_mod_err /
- DOUBLE_DIVIDE_CHECK(cpi->modified_error_left)));
-
- if (alt_kf_bits > cpi->kf_bits)
- {
- cpi->kf_bits = alt_kf_bits;
- }
- }
-
- cpi->kf_group_bits -= cpi->kf_bits;
- cpi->kf_bits += cpi->min_frame_bandwidth; // Add in the minimum frame allowance
-
- cpi->per_frame_bandwidth = cpi->kf_bits; // Peer frame bit target for this frame
- cpi->target_bandwidth = cpi->kf_bits * cpi->output_frame_rate; // Convert to a per second bitrate
- }
-
- // Note the total error score of the kf group minus the key frame itself
- cpi->kf_group_error_left = (int)(kf_group_err - kf_mod_err);
-
- // Adjust the count of total modified error left.
- // The count of bits left is adjusted elsewhere based on real coded frame sizes
- cpi->modified_error_left -= kf_group_err;
-
- if (cpi->oxcf.allow_spatial_resampling)
- {
- int resample_trigger = FALSE;
- int last_kf_resampled = FALSE;
- int kf_q;
- int scale_val = 0;
- int hr, hs, vr, vs;
- int new_width = cpi->oxcf.Width;
- int new_height = cpi->oxcf.Height;
-
- int projected_buffer_level = cpi->buffer_level;
- int tmp_q;
-
- double projected_bits_perframe;
- double group_iiratio = (kf_group_intra_err - first_frame.intra_error) / (kf_group_coded_err - first_frame.coded_error);
- double err_per_frame = kf_group_err / cpi->frames_to_key;
- double bits_per_frame;
- double av_bits_per_frame;
- double effective_size_ratio;
-
- if ((cpi->common.Width != cpi->oxcf.Width) || (cpi->common.Height != cpi->oxcf.Height))
- last_kf_resampled = TRUE;
-
- // Set back to unscaled by defaults
- cpi->common.horiz_scale = NORMAL;
- cpi->common.vert_scale = NORMAL;
-
- // Calculate Average bits per frame.
- //av_bits_per_frame = cpi->bits_left/(double)(cpi->total_stats->count - cpi->common.current_video_frame);
- av_bits_per_frame = cpi->oxcf.target_bandwidth / DOUBLE_DIVIDE_CHECK((double)cpi->oxcf.frame_rate);
- //if ( av_bits_per_frame < 0.0 )
- // av_bits_per_frame = 0.0
-
- // CBR... Use the clip average as the target for deciding resample
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- bits_per_frame = av_bits_per_frame;
- }
-
- // In VBR we want to avoid downsampling in easy section unless we are under extreme pressure
- // So use the larger of target bitrate for this sectoion or average bitrate for sequence
- else
- {
- bits_per_frame = cpi->kf_group_bits / cpi->frames_to_key; // This accounts for how hard the section is...
-
- if (bits_per_frame < av_bits_per_frame) // Dont turn to resampling in easy sections just because they have been assigned a small number of bits
- bits_per_frame = av_bits_per_frame;
- }
-
- // bits_per_frame should comply with our minimum
- if (bits_per_frame < (cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100))
- bits_per_frame = (cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100);
-
- // Work out if spatial resampling is necessary
- kf_q = estimate_kf_group_q(cpi, err_per_frame, bits_per_frame, group_iiratio);
-
- // If we project a required Q higher than the maximum allowed Q then make a guess at the actual size of frames in this section
- projected_bits_perframe = bits_per_frame;
- tmp_q = kf_q;
-
- while (tmp_q > cpi->worst_quality)
- {
- projected_bits_perframe *= 1.04;
- tmp_q--;
- }
-
- // Guess at buffer level at the end of the section
- projected_buffer_level = cpi->buffer_level - (int)((projected_bits_perframe - av_bits_per_frame) * cpi->frames_to_key);
-
- if (0)
- {
- FILE *f = fopen("Subsamle.stt", "a");
- fprintf(f, " %8d %8d %8d %8d %12.0f %8d %8d %8d\n", cpi->common.current_video_frame, kf_q, cpi->common.horiz_scale, cpi->common.vert_scale, kf_group_err / cpi->frames_to_key, (int)(cpi->kf_group_bits / cpi->frames_to_key), new_height, new_width);
- fclose(f);
- }
-
- // The trigger for spatial resampling depends on the various parameters such as whether we are streaming (CBR) or VBR.
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- // Trigger resample if we are projected to fall below down sample level or
- // resampled last time and are projected to remain below the up sample level
- if ((projected_buffer_level < (cpi->oxcf.resample_down_water_mark * cpi->oxcf.optimal_buffer_level / 100)) ||
- (last_kf_resampled && (projected_buffer_level < (cpi->oxcf.resample_up_water_mark * cpi->oxcf.optimal_buffer_level / 100))))
- //( ((cpi->buffer_level < (cpi->oxcf.resample_down_water_mark * cpi->oxcf.optimal_buffer_level / 100))) &&
- // ((projected_buffer_level < (cpi->oxcf.resample_up_water_mark * cpi->oxcf.optimal_buffer_level / 100))) ))
- resample_trigger = TRUE;
- else
- resample_trigger = FALSE;
- }
- else
- {
- long long clip_bits = (long long)(cpi->total_stats->count * cpi->oxcf.target_bandwidth / DOUBLE_DIVIDE_CHECK((double)cpi->oxcf.frame_rate));
- long long over_spend = cpi->oxcf.starting_buffer_level - cpi->buffer_level;
- long long over_spend2 = cpi->oxcf.starting_buffer_level - projected_buffer_level;
-
- if ((last_kf_resampled && (kf_q > cpi->worst_quality)) || // If triggered last time the threshold for triggering again is reduced
- ((kf_q > cpi->worst_quality) && // Projected Q higher than allowed and ...
- (over_spend > clip_bits / 20))) // ... Overspend > 5% of total bits
- resample_trigger = TRUE;
- else
- resample_trigger = FALSE;
-
- }
-
- if (resample_trigger)
- {
- while ((kf_q >= cpi->worst_quality) && (scale_val < 6))
- {
- scale_val ++;
-
- cpi->common.vert_scale = vscale_lookup[scale_val];
- cpi->common.horiz_scale = hscale_lookup[scale_val];
-
- Scale2Ratio(cpi->common.horiz_scale, &hr, &hs);
- Scale2Ratio(cpi->common.vert_scale, &vr, &vs);
-
- new_width = ((hs - 1) + (cpi->oxcf.Width * hr)) / hs;
- new_height = ((vs - 1) + (cpi->oxcf.Height * vr)) / vs;
-
- // Reducing the area to 1/4 does not reduce the complexity (err_per_frame) to 1/4...
- // effective_sizeratio attempts to provide a crude correction for this
- effective_size_ratio = (double)(new_width * new_height) / (double)(cpi->oxcf.Width * cpi->oxcf.Height);
- effective_size_ratio = (1.0 + (3.0 * effective_size_ratio)) / 4.0;
-
- // Now try again and see what Q we get with the smaller image size
- kf_q = estimate_kf_group_q(cpi, err_per_frame * effective_size_ratio, bits_per_frame, group_iiratio);
-
- if (0)
- {
- FILE *f = fopen("Subsamle.stt", "a");
- fprintf(f, "******** %8d %8d %8d %12.0f %8d %8d %8d\n", kf_q, cpi->common.horiz_scale, cpi->common.vert_scale, kf_group_err / cpi->frames_to_key, (int)(cpi->kf_group_bits / cpi->frames_to_key), new_height, new_width);
- fclose(f);
- }
- }
- }
-
- if ((cpi->common.Width != new_width) || (cpi->common.Height != new_height))
- {
- cpi->common.Width = new_width;
- cpi->common.Height = new_height;
- vp8_alloc_compressor_data(cpi);
- }
- }
-}
diff --git a/vp8/encoder/generic/csystemdependent.c b/vp8/encoder/generic/csystemdependent.c
deleted file mode 100644
index 1d672be..0000000
--- a/vp8/encoder/generic/csystemdependent.c
+++ /dev/null
@@ -1,117 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vp8/encoder/variance.h"
-#include "vp8/encoder/onyx_int.h"
-
-
-void vp8_arch_x86_encoder_init(VP8_COMP *cpi);
-void vp8_arch_arm_encoder_init(VP8_COMP *cpi);
-
-extern void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d);
-
-void (*vp8_yv12_copy_partial_frame_ptr)(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc, int Fraction);
-extern void vp8_yv12_copy_partial_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc, int Fraction);
-
-void vp8_cmachine_specific_config(VP8_COMP *cpi)
-{
-#if CONFIG_RUNTIME_CPU_DETECT
- cpi->rtcd.common = &cpi->common.rtcd;
- cpi->rtcd.variance.sad16x16 = vp8_sad16x16_c;
- cpi->rtcd.variance.sad16x8 = vp8_sad16x8_c;
- cpi->rtcd.variance.sad8x16 = vp8_sad8x16_c;
- cpi->rtcd.variance.sad8x8 = vp8_sad8x8_c;
- cpi->rtcd.variance.sad4x4 = vp8_sad4x4_c;
-
- cpi->rtcd.variance.sad16x16x3 = vp8_sad16x16x3_c;
- cpi->rtcd.variance.sad16x8x3 = vp8_sad16x8x3_c;
- cpi->rtcd.variance.sad8x16x3 = vp8_sad8x16x3_c;
- cpi->rtcd.variance.sad8x8x3 = vp8_sad8x8x3_c;
- cpi->rtcd.variance.sad4x4x3 = vp8_sad4x4x3_c;
-
- cpi->rtcd.variance.sad16x16x8 = vp8_sad16x16x8_c;
- cpi->rtcd.variance.sad16x8x8 = vp8_sad16x8x8_c;
- cpi->rtcd.variance.sad8x16x8 = vp8_sad8x16x8_c;
- cpi->rtcd.variance.sad8x8x8 = vp8_sad8x8x8_c;
- cpi->rtcd.variance.sad4x4x8 = vp8_sad4x4x8_c;
-
- cpi->rtcd.variance.sad16x16x4d = vp8_sad16x16x4d_c;
- cpi->rtcd.variance.sad16x8x4d = vp8_sad16x8x4d_c;
- cpi->rtcd.variance.sad8x16x4d = vp8_sad8x16x4d_c;
- cpi->rtcd.variance.sad8x8x4d = vp8_sad8x8x4d_c;
- cpi->rtcd.variance.sad4x4x4d = vp8_sad4x4x4d_c;
-
- cpi->rtcd.variance.var4x4 = vp8_variance4x4_c;
- cpi->rtcd.variance.var8x8 = vp8_variance8x8_c;
- cpi->rtcd.variance.var8x16 = vp8_variance8x16_c;
- cpi->rtcd.variance.var16x8 = vp8_variance16x8_c;
- cpi->rtcd.variance.var16x16 = vp8_variance16x16_c;
-
- cpi->rtcd.variance.subpixvar4x4 = vp8_sub_pixel_variance4x4_c;
- cpi->rtcd.variance.subpixvar8x8 = vp8_sub_pixel_variance8x8_c;
- cpi->rtcd.variance.subpixvar8x16 = vp8_sub_pixel_variance8x16_c;
- cpi->rtcd.variance.subpixvar16x8 = vp8_sub_pixel_variance16x8_c;
- cpi->rtcd.variance.subpixvar16x16 = vp8_sub_pixel_variance16x16_c;
- cpi->rtcd.variance.halfpixvar16x16_h = vp8_variance_halfpixvar16x16_h_c;
- cpi->rtcd.variance.halfpixvar16x16_v = vp8_variance_halfpixvar16x16_v_c;
- cpi->rtcd.variance.halfpixvar16x16_hv = vp8_variance_halfpixvar16x16_hv_c;
- cpi->rtcd.variance.subpixmse16x16 = vp8_sub_pixel_mse16x16_c;
-
- cpi->rtcd.variance.mse16x16 = vp8_mse16x16_c;
- cpi->rtcd.variance.getmbss = vp8_get_mb_ss_c;
-
- cpi->rtcd.variance.get16x16prederror = vp8_get16x16pred_error_c;
- cpi->rtcd.variance.get8x8var = vp8_get8x8var_c;
- cpi->rtcd.variance.get16x16var = vp8_get16x16var_c;;
- cpi->rtcd.variance.get4x4sse_cs = vp8_get4x4sse_cs_c;
-
- cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_c;
- cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_c;
- cpi->rtcd.fdct.fast4x4 = vp8_short_fdct4x4_c;
- cpi->rtcd.fdct.fast8x4 = vp8_short_fdct8x4_c;
- cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_c;
-
- cpi->rtcd.encodemb.berr = vp8_block_error_c;
- cpi->rtcd.encodemb.mberr = vp8_mbblock_error_c;
- cpi->rtcd.encodemb.mbuverr = vp8_mbuverror_c;
- cpi->rtcd.encodemb.subb = vp8_subtract_b_c;
- cpi->rtcd.encodemb.submby = vp8_subtract_mby_c;
- cpi->rtcd.encodemb.submbuv = vp8_subtract_mbuv_c;
-
- cpi->rtcd.quantize.quantb = vp8_regular_quantize_b;
- cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_c;
-#if !(CONFIG_REALTIME_ONLY)
- cpi->rtcd.search.full_search = vp8_full_search_sad;
-#endif
- cpi->rtcd.search.diamond_search = vp8_diamond_search_sad;
-#if !(CONFIG_REALTIME_ONLY)
- cpi->rtcd.temporal.apply = vp8_temporal_filter_apply_c;
-#endif
-#endif
-
- // Pure C:
- vp8_yv12_copy_partial_frame_ptr = vp8_yv12_copy_partial_frame;
-
-#if CONFIG_PSNR
- cpi->rtcd.variance.ssimpf_8x8 = ssim_parms_8x8_c;
- cpi->rtcd.variance.ssimpf = ssim_parms_c;
-#endif
-
-#if ARCH_X86 || ARCH_X86_64
- vp8_arch_x86_encoder_init(cpi);
-#endif
-
-#if ARCH_ARM
- vp8_arch_arm_encoder_init(cpi);
-#endif
-
-}
diff --git a/vp8/encoder/mcomp.c b/vp8/encoder/mcomp.c
deleted file mode 100644
index de6642b..0000000
--- a/vp8/encoder/mcomp.c
+++ /dev/null
@@ -1,1659 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "mcomp.h"
-#include "vpx_mem/vpx_mem.h"
-
-#include <stdio.h>
-#include <limits.h>
-#include <math.h>
-
-#ifdef ENTROPY_STATS
-static int mv_ref_ct [31] [4] [2];
-static int mv_mode_cts [4] [2];
-#endif
-
-static int mv_bits_sadcost[256];
-
-void vp8cx_init_mv_bits_sadcost()
-{
- int i;
-
- for (i = 0; i < 256; i++)
- {
- mv_bits_sadcost[i] = (int)sqrt(i * 16);
- }
-}
-
-
-int vp8_mv_bit_cost(MV *mv, MV *ref, int *mvcost[2], int Weight)
-{
- // MV costing is based on the distribution of vectors in the previous frame and as such will tend to
- // over state the cost of vectors. In addition coding a new vector can have a knock on effect on the
- // cost of subsequent vectors and the quality of prediction from NEAR and NEAREST for subsequent blocks.
- // The "Weight" parameter allows, to a limited extent, for some account to be taken of these factors.
- return ((mvcost[0][(mv->row - ref->row) >> 1] + mvcost[1][(mv->col - ref->col) >> 1]) * Weight) >> 7;
-}
-
-static int mv_err_cost(MV *mv, MV *ref, int *mvcost[2], int error_per_bit)
-{
- //int i;
- //return ((mvcost[0][(mv->row - ref->row)>>1] + mvcost[1][(mv->col - ref->col)>>1] + 128) * error_per_bit) >> 8;
- //return ( (vp8_mv_bit_cost(mv, ref, mvcost, 100) + 128) * error_per_bit) >> 8;
-
- //i = (vp8_mv_bit_cost(mv, ref, mvcost, 100) * error_per_bit + 128) >> 8;
- return ((mvcost[0][(mv->row - ref->row) >> 1] + mvcost[1][(mv->col - ref->col) >> 1]) * error_per_bit + 128) >> 8;
- //return (vp8_mv_bit_cost(mv, ref, mvcost, 128) * error_per_bit + 128) >> 8;
-}
-
-
-static int mv_bits(MV *mv, MV *ref, int *mvcost[2])
-{
- // get the estimated number of bits for a motion vector, to be used for costing in SAD based
- // motion estimation
- return ((mvcost[0][(mv->row - ref->row) >> 1] + mvcost[1][(mv->col - ref->col)>> 1]) + 128) >> 8;
-}
-
-void vp8_init_dsmotion_compensation(MACROBLOCK *x, int stride)
-{
- int Len;
- int search_site_count = 0;
-
-
- // Generate offsets for 4 search sites per step.
- Len = MAX_FIRST_STEP;
- x->ss[search_site_count].mv.col = 0;
- x->ss[search_site_count].mv.row = 0;
- x->ss[search_site_count].offset = 0;
- search_site_count++;
-
- while (Len > 0)
- {
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = 0;
- x->ss[search_site_count].mv.row = -Len;
- x->ss[search_site_count].offset = -Len * stride;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = 0;
- x->ss[search_site_count].mv.row = Len;
- x->ss[search_site_count].offset = Len * stride;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = -Len;
- x->ss[search_site_count].mv.row = 0;
- x->ss[search_site_count].offset = -Len;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = Len;
- x->ss[search_site_count].mv.row = 0;
- x->ss[search_site_count].offset = Len;
- search_site_count++;
-
- // Contract.
- Len /= 2;
- }
-
- x->ss_count = search_site_count;
- x->searches_per_step = 4;
-}
-
-void vp8_init3smotion_compensation(MACROBLOCK *x, int stride)
-{
- int Len;
- int search_site_count = 0;
-
- // Generate offsets for 8 search sites per step.
- Len = MAX_FIRST_STEP;
- x->ss[search_site_count].mv.col = 0;
- x->ss[search_site_count].mv.row = 0;
- x->ss[search_site_count].offset = 0;
- search_site_count++;
-
- while (Len > 0)
- {
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = 0;
- x->ss[search_site_count].mv.row = -Len;
- x->ss[search_site_count].offset = -Len * stride;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = 0;
- x->ss[search_site_count].mv.row = Len;
- x->ss[search_site_count].offset = Len * stride;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = -Len;
- x->ss[search_site_count].mv.row = 0;
- x->ss[search_site_count].offset = -Len;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = Len;
- x->ss[search_site_count].mv.row = 0;
- x->ss[search_site_count].offset = Len;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = -Len;
- x->ss[search_site_count].mv.row = -Len;
- x->ss[search_site_count].offset = -Len * stride - Len;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = Len;
- x->ss[search_site_count].mv.row = -Len;
- x->ss[search_site_count].offset = -Len * stride + Len;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = -Len;
- x->ss[search_site_count].mv.row = Len;
- x->ss[search_site_count].offset = Len * stride - Len;
- search_site_count++;
-
- // Compute offsets for search sites.
- x->ss[search_site_count].mv.col = Len;
- x->ss[search_site_count].mv.row = Len;
- x->ss[search_site_count].offset = Len * stride + Len;
- search_site_count++;
-
-
- // Contract.
- Len /= 2;
- }
-
- x->ss_count = search_site_count;
- x->searches_per_step = 8;
-}
-
-
-#define MVC(r,c) (((mvcost[0][(r)-rr] + mvcost[1][(c) - rc]) * error_per_bit + 128 )>>8 ) // estimated cost of a motion vector (r,c)
-#define PRE(r,c) (*(d->base_pre) + d->pre + ((r)>>2) * d->pre_stride + ((c)>>2)) // pointer to predictor base of a motionvector
-#define SP(x) (((x)&3)<<1) // convert motion vector component to offset for svf calc
-#define DIST(r,c) vfp->svf( PRE(r,c), d->pre_stride, SP(c),SP(r), z,b->src_stride,&sse) // returns subpixel variance error function.
-#define IFMVCV(r,c,s,e) if ( c >= minc && c <= maxc && r >= minr && r <= maxr) s else e;
-#define ERR(r,c) (MVC(r,c)+DIST(r,c)) // returns distortion + motion vector cost
-#define CHECK_BETTER(v,r,c) IFMVCV(r,c,{if((v = ERR(r,c)) < besterr) { besterr = v; br=r; bc=c; }}, v=INT_MAX;)// checks if (r,c) has better score than previous best
-#define MIN(x,y) (((x)<(y))?(x):(y))
-#define MAX(x,y) (((x)>(y))?(x):(y))
-
-//#define CHECK_BETTER(v,r,c) if((v = ERR(r,c)) < besterr) { besterr = v; br=r; bc=c; }
-
-int vp8_find_best_sub_pixel_step_iteratively(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *bestmv, MV *ref_mv, int error_per_bit, const vp8_variance_fn_ptr_t *vfp, int *mvcost[2])
-{
- unsigned char *y = *(d->base_pre) + d->pre + (bestmv->row) * d->pre_stride + bestmv->col;
- unsigned char *z = (*(b->base_src) + b->src);
-
- int rr = ref_mv->row >> 1, rc = ref_mv->col >> 1;
- int br = bestmv->row << 2, bc = bestmv->col << 2;
- int tr = br, tc = bc;
- unsigned int besterr = INT_MAX;
- unsigned int left, right, up, down, diag;
- unsigned int sse;
- unsigned int whichdir;
- unsigned int halfiters = 4;
- unsigned int quarteriters = 4;
-
- int minc = MAX(x->mv_col_min << 2, (ref_mv->col >> 1) - ((1 << mvlong_width) - 1));
- int maxc = MIN(x->mv_col_max << 2, (ref_mv->col >> 1) + ((1 << mvlong_width) - 1));
- int minr = MAX(x->mv_row_min << 2, (ref_mv->row >> 1) - ((1 << mvlong_width) - 1));
- int maxr = MIN(x->mv_row_max << 2, (ref_mv->row >> 1) + ((1 << mvlong_width) - 1));
-
- // central mv
- bestmv->row <<= 3;
- bestmv->col <<= 3;
-
- // calculate central point error
- besterr = vfp->vf(y, d->pre_stride, z, b->src_stride, &sse);
- besterr += mv_err_cost(bestmv, ref_mv, mvcost, error_per_bit);
-
- // TODO: Each subsequent iteration checks at least one point in common with the last iteration could be 2 ( if diag selected)
- while (--halfiters)
- {
- // 1/2 pel
- CHECK_BETTER(left, tr, tc - 2);
- CHECK_BETTER(right, tr, tc + 2);
- CHECK_BETTER(up, tr - 2, tc);
- CHECK_BETTER(down, tr + 2, tc);
-
- whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
-
- switch (whichdir)
- {
- case 0:
- CHECK_BETTER(diag, tr - 2, tc - 2);
- break;
- case 1:
- CHECK_BETTER(diag, tr - 2, tc + 2);
- break;
- case 2:
- CHECK_BETTER(diag, tr + 2, tc - 2);
- break;
- case 3:
- CHECK_BETTER(diag, tr + 2, tc + 2);
- break;
- }
-
- // no reason to check the same one again.
- if (tr == br && tc == bc)
- break;
-
- tr = br;
- tc = bc;
- }
-
- // TODO: Each subsequent iteration checks at least one point in common with the last iteration could be 2 ( if diag selected)
- // 1/4 pel
- while (--quarteriters)
- {
- CHECK_BETTER(left, tr, tc - 1);
- CHECK_BETTER(right, tr, tc + 1);
- CHECK_BETTER(up, tr - 1, tc);
- CHECK_BETTER(down, tr + 1, tc);
-
- whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
-
- switch (whichdir)
- {
- case 0:
- CHECK_BETTER(diag, tr - 1, tc - 1);
- break;
- case 1:
- CHECK_BETTER(diag, tr - 1, tc + 1);
- break;
- case 2:
- CHECK_BETTER(diag, tr + 1, tc - 1);
- break;
- case 3:
- CHECK_BETTER(diag, tr + 1, tc + 1);
- break;
- }
-
- // no reason to check the same one again.
- if (tr == br && tc == bc)
- break;
-
- tr = br;
- tc = bc;
- }
-
- bestmv->row = br << 1;
- bestmv->col = bc << 1;
-
- if ((abs(bestmv->col - ref_mv->col) > MAX_FULL_PEL_VAL) || (abs(bestmv->row - ref_mv->row) > MAX_FULL_PEL_VAL))
- return INT_MAX;
-
- return besterr;
-}
-#undef MVC
-#undef PRE
-#undef SP
-#undef DIST
-#undef ERR
-#undef CHECK_BETTER
-#undef MIN
-#undef MAX
-int vp8_find_best_sub_pixel_step(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *bestmv, MV *ref_mv, int error_per_bit, const vp8_variance_fn_ptr_t *vfp, int *mvcost[2])
-{
- int bestmse = INT_MAX;
- MV startmv;
- //MV this_mv;
- MV this_mv;
- unsigned char *y = *(d->base_pre) + d->pre + (bestmv->row) * d->pre_stride + bestmv->col;
- unsigned char *z = (*(b->base_src) + b->src);
- int left, right, up, down, diag;
- unsigned int sse;
- int whichdir ;
-
-
- // Trap uncodable vectors
- if ((abs((bestmv->col << 3) - ref_mv->col) > MAX_FULL_PEL_VAL) || (abs((bestmv->row << 3) - ref_mv->row) > MAX_FULL_PEL_VAL))
- {
- bestmv->row <<= 3;
- bestmv->col <<= 3;
- return INT_MAX;
- }
-
- // central mv
- bestmv->row <<= 3;
- bestmv->col <<= 3;
- startmv = *bestmv;
-
- // calculate central point error
- bestmse = vfp->vf(y, d->pre_stride, z, b->src_stride, &sse);
- bestmse += mv_err_cost(bestmv, ref_mv, mvcost, error_per_bit);
-
- // go left then right and check error
- this_mv.row = startmv.row;
- this_mv.col = ((startmv.col - 8) | 4);
- left = vfp->svf_halfpix_h(y - 1, d->pre_stride, z, b->src_stride, &sse);
- left += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (left < bestmse)
- {
- *bestmv = this_mv;
- bestmse = left;
- }
-
- this_mv.col += 8;
- right = vfp->svf_halfpix_h(y, d->pre_stride, z, b->src_stride, &sse);
- right += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (right < bestmse)
- {
- *bestmv = this_mv;
- bestmse = right;
- }
-
- // go up then down and check error
- this_mv.col = startmv.col;
- this_mv.row = ((startmv.row - 8) | 4);
- up = vfp->svf_halfpix_v(y - d->pre_stride, d->pre_stride, z, b->src_stride, &sse);
- up += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (up < bestmse)
- {
- *bestmv = this_mv;
- bestmse = up;
- }
-
- this_mv.row += 8;
- down = vfp->svf_halfpix_v(y, d->pre_stride, z, b->src_stride, &sse);
- down += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (down < bestmse)
- {
- *bestmv = this_mv;
- bestmse = down;
- }
-
-
- // now check 1 more diagonal
- whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
- //for(whichdir =0;whichdir<4;whichdir++)
- //{
- this_mv = startmv;
-
- switch (whichdir)
- {
- case 0:
- this_mv.col = (this_mv.col - 8) | 4;
- this_mv.row = (this_mv.row - 8) | 4;
- diag = vfp->svf_halfpix_hv(y - 1 - d->pre_stride, d->pre_stride, z, b->src_stride, &sse);
- break;
- case 1:
- this_mv.col += 4;
- this_mv.row = (this_mv.row - 8) | 4;
- diag = vfp->svf_halfpix_hv(y - d->pre_stride, d->pre_stride, z, b->src_stride, &sse);
- break;
- case 2:
- this_mv.col = (this_mv.col - 8) | 4;
- this_mv.row += 4;
- diag = vfp->svf_halfpix_hv(y - 1, d->pre_stride, z, b->src_stride, &sse);
- break;
- case 3:
- default:
- this_mv.col += 4;
- this_mv.row += 4;
- diag = vfp->svf_halfpix_hv(y, d->pre_stride, z, b->src_stride, &sse);
- break;
- }
-
- diag += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (diag < bestmse)
- {
- *bestmv = this_mv;
- bestmse = diag;
- }
-
-// }
-
-
- // time to check quarter pels.
- if (bestmv->row < startmv.row)
- y -= d->pre_stride;
-
- if (bestmv->col < startmv.col)
- y--;
-
- startmv = *bestmv;
-
-
-
- // go left then right and check error
- this_mv.row = startmv.row;
-
- if (startmv.col & 7)
- {
- this_mv.col = startmv.col - 2;
- left = vfp->svf(y, d->pre_stride, this_mv.col & 7, this_mv.row & 7, z, b->src_stride, &sse);
- }
- else
- {
- this_mv.col = (startmv.col - 8) | 6;
- left = vfp->svf(y - 1, d->pre_stride, 6, this_mv.row & 7, z, b->src_stride, &sse);
- }
-
- left += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (left < bestmse)
- {
- *bestmv = this_mv;
- bestmse = left;
- }
-
- this_mv.col += 4;
- right = vfp->svf(y, d->pre_stride, this_mv.col & 7, this_mv.row & 7, z, b->src_stride, &sse);
- right += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (right < bestmse)
- {
- *bestmv = this_mv;
- bestmse = right;
- }
-
- // go up then down and check error
- this_mv.col = startmv.col;
-
- if (startmv.row & 7)
- {
- this_mv.row = startmv.row - 2;
- up = vfp->svf(y, d->pre_stride, this_mv.col & 7, this_mv.row & 7, z, b->src_stride, &sse);
- }
- else
- {
- this_mv.row = (startmv.row - 8) | 6;
- up = vfp->svf(y - d->pre_stride, d->pre_stride, this_mv.col & 7, 6, z, b->src_stride, &sse);
- }
-
- up += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (up < bestmse)
- {
- *bestmv = this_mv;
- bestmse = up;
- }
-
- this_mv.row += 4;
- down = vfp->svf(y, d->pre_stride, this_mv.col & 7, this_mv.row & 7, z, b->src_stride, &sse);
- down += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (down < bestmse)
- {
- *bestmv = this_mv;
- bestmse = down;
- }
-
-
- // now check 1 more diagonal
- whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
-
-// for(whichdir=0;whichdir<4;whichdir++)
-// {
- this_mv = startmv;
-
- switch (whichdir)
- {
- case 0:
-
- if (startmv.row & 7)
- {
- this_mv.row -= 2;
-
- if (startmv.col & 7)
- {
- this_mv.col -= 2;
- diag = vfp->svf(y, d->pre_stride, this_mv.col & 7, this_mv.row & 7, z, b->src_stride, &sse);
- }
- else
- {
- this_mv.col = (startmv.col - 8) | 6;
- diag = vfp->svf(y - 1, d->pre_stride, 6, this_mv.row & 7, z, b->src_stride, &sse);;
- }
- }
- else
- {
- this_mv.row = (startmv.row - 8) | 6;
-
- if (startmv.col & 7)
- {
- this_mv.col -= 2;
- diag = vfp->svf(y - d->pre_stride, d->pre_stride, this_mv.col & 7, 6, z, b->src_stride, &sse);
- }
- else
- {
- this_mv.col = (startmv.col - 8) | 6;
- diag = vfp->svf(y - d->pre_stride - 1, d->pre_stride, 6, 6, z, b->src_stride, &sse);
- }
- }
-
- break;
- case 1:
- this_mv.col += 2;
-
- if (startmv.row & 7)
- {
- this_mv.row -= 2;
- diag = vfp->svf(y, d->pre_stride, this_mv.col & 7, this_mv.row & 7, z, b->src_stride, &sse);
- }
- else
- {
- this_mv.row = (startmv.row - 8) | 6;
- diag = vfp->svf(y - d->pre_stride, d->pre_stride, this_mv.col & 7, 6, z, b->src_stride, &sse);
- }
-
- break;
- case 2:
- this_mv.row += 2;
-
- if (startmv.col & 7)
- {
- this_mv.col -= 2;
- diag = vfp->svf(y, d->pre_stride, this_mv.col & 7, this_mv.row & 7, z, b->src_stride, &sse);
- }
- else
- {
- this_mv.col = (startmv.col - 8) | 6;
- diag = vfp->svf(y - 1, d->pre_stride, 6, this_mv.row & 7, z, b->src_stride, &sse);;
- }
-
- break;
- case 3:
- this_mv.col += 2;
- this_mv.row += 2;
- diag = vfp->svf(y, d->pre_stride, this_mv.col & 7, this_mv.row & 7, z, b->src_stride, &sse);
- break;
- }
-
- diag += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (diag < bestmse)
- {
- *bestmv = this_mv;
- bestmse = diag;
- }
-
-// }
-
- return bestmse;
-}
-
-int vp8_find_best_half_pixel_step(MACROBLOCK *mb, BLOCK *b, BLOCKD *d, MV *bestmv, MV *ref_mv, int error_per_bit, const vp8_variance_fn_ptr_t *vfp, int *mvcost[2])
-{
- int bestmse = INT_MAX;
- MV startmv;
- //MV this_mv;
- MV this_mv;
- unsigned char *y = *(d->base_pre) + d->pre + (bestmv->row) * d->pre_stride + bestmv->col;
- unsigned char *z = (*(b->base_src) + b->src);
- int left, right, up, down, diag;
- unsigned int sse;
-
- // Trap uncodable vectors
- if ((abs((bestmv->col << 3) - ref_mv->col) > MAX_FULL_PEL_VAL) || (abs((bestmv->row << 3) - ref_mv->row) > MAX_FULL_PEL_VAL))
- {
- bestmv->row <<= 3;
- bestmv->col <<= 3;
- return INT_MAX;
- }
-
- // central mv
- bestmv->row <<= 3;
- bestmv->col <<= 3;
- startmv = *bestmv;
-
- // calculate central point error
- bestmse = vfp->vf(y, d->pre_stride, z, b->src_stride, &sse);
- bestmse += mv_err_cost(bestmv, ref_mv, mvcost, error_per_bit);
-
- // go left then right and check error
- this_mv.row = startmv.row;
- this_mv.col = ((startmv.col - 8) | 4);
- left = vfp->svf_halfpix_h(y - 1, d->pre_stride, z, b->src_stride, &sse);
- left += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (left < bestmse)
- {
- *bestmv = this_mv;
- bestmse = left;
- }
-
- this_mv.col += 8;
- right = vfp->svf_halfpix_h(y, d->pre_stride, z, b->src_stride, &sse);
- right += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (right < bestmse)
- {
- *bestmv = this_mv;
- bestmse = right;
- }
-
- // go up then down and check error
- this_mv.col = startmv.col;
- this_mv.row = ((startmv.row - 8) | 4);
- up = vfp->svf_halfpix_v(y - d->pre_stride, d->pre_stride, z, b->src_stride, &sse);
- up += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (up < bestmse)
- {
- *bestmv = this_mv;
- bestmse = up;
- }
-
- this_mv.row += 8;
- down = vfp->svf_halfpix_v(y, d->pre_stride, z, b->src_stride, &sse);
- down += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (down < bestmse)
- {
- *bestmv = this_mv;
- bestmse = down;
- }
-
- // somewhat strangely not doing all the diagonals for half pel is slower than doing them.
-#if 0
- // now check 1 more diagonal -
- whichdir = (left < right ? 0 : 1) + (up < down ? 0 : 2);
- this_mv = startmv;
-
- switch (whichdir)
- {
- case 0:
- this_mv.col = (this_mv.col - 8) | 4;
- this_mv.row = (this_mv.row - 8) | 4;
- diag = vfp->svf(y - 1 - d->pre_stride, d->pre_stride, 4, 4, z, b->src_stride, &sse);
- break;
- case 1:
- this_mv.col += 4;
- this_mv.row = (this_mv.row - 8) | 4;
- diag = vfp->svf(y - d->pre_stride, d->pre_stride, 4, 4, z, b->src_stride, &sse);
- break;
- case 2:
- this_mv.col = (this_mv.col - 8) | 4;
- this_mv.row += 4;
- diag = vfp->svf(y - 1, d->pre_stride, 4, 4, z, b->src_stride, &sse);
- break;
- case 3:
- this_mv.col += 4;
- this_mv.row += 4;
- diag = vfp->svf(y, d->pre_stride, 4, 4, z, b->src_stride, &sse);
- break;
- }
-
- diag += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (diag < bestmse)
- {
- *bestmv = this_mv;
- bestmse = diag;
- }
-
-#else
- this_mv.col = (this_mv.col - 8) | 4;
- this_mv.row = (this_mv.row - 8) | 4;
- diag = vfp->svf_halfpix_hv(y - 1 - d->pre_stride, d->pre_stride, z, b->src_stride, &sse);
- diag += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (diag < bestmse)
- {
- *bestmv = this_mv;
- bestmse = diag;
- }
-
- this_mv.col += 8;
- diag = vfp->svf_halfpix_hv(y - d->pre_stride, d->pre_stride, z, b->src_stride, &sse);
- diag += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (diag < bestmse)
- {
- *bestmv = this_mv;
- bestmse = diag;
- }
-
- this_mv.col = (this_mv.col - 8) | 4;
- this_mv.row = startmv.row + 4;
- diag = vfp->svf_halfpix_hv(y - 1, d->pre_stride, z, b->src_stride, &sse);
- diag += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (diag < bestmse)
- {
- *bestmv = this_mv;
- bestmse = diag;
- }
-
- this_mv.col += 8;
- diag = vfp->svf_halfpix_hv(y, d->pre_stride, z, b->src_stride, &sse);
- diag += mv_err_cost(&this_mv, ref_mv, mvcost, error_per_bit);
-
- if (diag < bestmse)
- {
- *bestmv = this_mv;
- bestmse = diag;
- }
-
-#endif
- return bestmse;
-}
-
-
-#define MVC(r,c) (((mvsadcost[0][((r)<<2)-rr] + mvsadcost[1][((c)<<2) - rc]) * error_per_bit + 128 )>>8 ) // estimated cost of a motion vector (r,c)
-#define PRE(r,c) (*(d->base_pre) + d->pre + (r) * d->pre_stride + (c)) // pointer to predictor base of a motionvector
-#define DIST(r,c,v) vfp->sdf( src,src_stride,PRE(r,c),d->pre_stride, v) // returns sad error score.
-#define ERR(r,c,v) (MVC(r,c)+DIST(r,c,v)) // returns distortion + motion vector cost
-#define CHECK_BETTER(v,r,c) if ((v = ERR(r,c,besterr)) < besterr) { besterr = v; br=r; bc=c; } // checks if (r,c) has better score than previous best
-static const MV next_chkpts[6][3] =
-{
- {{ -2, 0}, { -1, -2}, {1, -2}},
- {{ -1, -2}, {1, -2}, {2, 0}},
- {{1, -2}, {2, 0}, {1, 2}},
- {{2, 0}, {1, 2}, { -1, 2}},
- {{1, 2}, { -1, 2}, { -2, 0}},
- {{ -1, 2}, { -2, 0}, { -1, -2}}
-};
-int vp8_hex_search
-(
- MACROBLOCK *x,
- BLOCK *b,
- BLOCKD *d,
- MV *ref_mv,
- MV *best_mv,
- int search_param,
- int error_per_bit,
- int *num00,
- const vp8_variance_fn_ptr_t *vfp,
- int *mvsadcost[2],
- int *mvcost[2],
- MV *center_mv
-)
-{
- MV hex[6] = { { -1, -2}, {1, -2}, {2, 0}, {1, 2}, { -1, 2}, { -2, 0} } ;
- MV neighbors[8] = { { -1, -1}, {0, -1}, {1, -1}, { -1, 0}, {1, 0}, { -1, 1}, {0, 1}, {1, 1} } ;
- int i, j;
- unsigned char *src = (*(b->base_src) + b->src);
- int src_stride = b->src_stride;
- int rr = center_mv->row, rc = center_mv->col;
- int br = ref_mv->row >> 3, bc = ref_mv->col >> 3, tr, tc;
- unsigned int besterr, thiserr = 0x7fffffff;
- int k = -1, tk;
-
- if (bc < x->mv_col_min) bc = x->mv_col_min;
-
- if (bc > x->mv_col_max) bc = x->mv_col_max;
-
- if (br < x->mv_row_min) br = x->mv_row_min;
-
- if (br > x->mv_row_max) br = x->mv_row_max;
-
- rr >>= 1;
- rc >>= 1;
-
- besterr = ERR(br, bc, thiserr);
-
- // hex search
- //j=0
- tr = br;
- tc = bc;
-
- for (i = 0; i < 6; i++)
- {
- int nr = tr + hex[i].row, nc = tc + hex[i].col;
-
- if (nc < x->mv_col_min) continue;
-
- if (nc > x->mv_col_max) continue;
-
- if (nr < x->mv_row_min) continue;
-
- if (nr > x->mv_row_max) continue;
-
- //CHECK_BETTER(thiserr,nr,nc);
- if ((thiserr = ERR(nr, nc, besterr)) < besterr)
- {
- besterr = thiserr;
- br = nr;
- bc = nc;
- k = i;
- }
- }
-
- if (tr == br && tc == bc)
- goto cal_neighbors;
-
- for (j = 1; j < 127; j++)
- {
- tr = br;
- tc = bc;
- tk = k;
-
- for (i = 0; i < 3; i++)
- {
- int nr = tr + next_chkpts[tk][i].row, nc = tc + next_chkpts[tk][i].col;
-
- if (nc < x->mv_col_min) continue;
-
- if (nc > x->mv_col_max) continue;
-
- if (nr < x->mv_row_min) continue;
-
- if (nr > x->mv_row_max) continue;
-
- //CHECK_BETTER(thiserr,nr,nc);
- if ((thiserr = ERR(nr, nc, besterr)) < besterr)
- {
- besterr = thiserr;
- br = nr;
- bc = nc; //k=(tk+5+i)%6;}
- k = tk + 5 + i;
-
- if (k >= 12) k -= 12;
- else if (k >= 6) k -= 6;
- }
- }
-
- if (tr == br && tc == bc)
- break;
- }
-
- // check 8 1 away neighbors
-cal_neighbors:
- tr = br;
- tc = bc;
-
- for (i = 0; i < 8; i++)
- {
- int nr = tr + neighbors[i].row, nc = tc + neighbors[i].col;
-
- if (nc < x->mv_col_min) continue;
-
- if (nc > x->mv_col_max) continue;
-
- if (nr < x->mv_row_min) continue;
-
- if (nr > x->mv_row_max) continue;
-
- CHECK_BETTER(thiserr, nr, nc);
- }
-
- best_mv->row = br;
- best_mv->col = bc;
-
- return vfp->vf(src, src_stride, PRE(br, bc), d->pre_stride, &thiserr) + mv_err_cost(best_mv, center_mv, mvcost, error_per_bit) ;
-}
-#undef MVC
-#undef PRE
-#undef SP
-#undef DIST
-#undef ERR
-#undef CHECK_BETTER
-
-
-int vp8_diamond_search_sad
-(
- MACROBLOCK *x,
- BLOCK *b,
- BLOCKD *d,
- MV *ref_mv,
- MV *best_mv,
- int search_param,
- int error_per_bit,
- int *num00,
- vp8_variance_fn_ptr_t *fn_ptr,
- int *mvsadcost[2],
- int *mvcost[2],
- MV *center_mv
-)
-{
- int i, j, step;
-
- unsigned char *what = (*(b->base_src) + b->src);
- int what_stride = b->src_stride;
- unsigned char *in_what;
- int in_what_stride = d->pre_stride;
- unsigned char *best_address;
-
- int tot_steps;
- MV this_mv;
-
- int bestsad = INT_MAX;
- int best_site = 0;
- int last_site = 0;
-
- int ref_row = ref_mv->row >> 3;
- int ref_col = ref_mv->col >> 3;
- int this_row_offset;
- int this_col_offset;
- search_site *ss;
-
- unsigned char *check_here;
- int thissad;
-
- *num00 = 0;
-
- // Work out the start point for the search
- in_what = (unsigned char *)(*(d->base_pre) + d->pre + (ref_row * (d->pre_stride)) + ref_col);
- best_address = in_what;
-
- // We need to check that the starting point for the search (as indicated by ref_mv) is within the buffer limits
- if ((ref_col > x->mv_col_min) && (ref_col < x->mv_col_max) &&
- (ref_row > x->mv_row_min) && (ref_row < x->mv_row_max))
- {
- // Check the starting position
- bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride, 0x7fffffff) + mv_err_cost(ref_mv, center_mv, mvsadcost, error_per_bit);
- }
-
- // search_param determines the length of the initial step and hence the number of iterations
- // 0 = initial step (MAX_FIRST_STEP) pel : 1 = (MAX_FIRST_STEP/2) pel, 2 = (MAX_FIRST_STEP/4) pel... etc.
- ss = &x->ss[search_param * x->searches_per_step];
- tot_steps = (x->ss_count / x->searches_per_step) - search_param;
-
- i = 1;
- best_mv->row = ref_row;
- best_mv->col = ref_col;
-
- for (step = 0; step < tot_steps ; step++)
- {
- for (j = 0 ; j < x->searches_per_step ; j++)
- {
- // Trap illegal vectors
- this_row_offset = best_mv->row + ss[i].mv.row;
- this_col_offset = best_mv->col + ss[i].mv.col;
-
- if ((this_col_offset > x->mv_col_min) && (this_col_offset < x->mv_col_max) &&
- (this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
-
- {
- check_here = ss[i].offset + best_address;
- thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
-
- if (thissad < bestsad)
- {
- this_mv.row = this_row_offset << 3;
- this_mv.col = this_col_offset << 3;
- thissad += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit);
-
- if (thissad < bestsad)
- {
- bestsad = thissad;
- best_site = i;
- }
- }
- }
-
- i++;
- }
-
- if (best_site != last_site)
- {
- best_mv->row += ss[best_site].mv.row;
- best_mv->col += ss[best_site].mv.col;
- best_address += ss[best_site].offset;
- last_site = best_site;
- }
- else if (best_address == in_what)
- (*num00)++;
- }
-
- this_mv.row = best_mv->row << 3;
- this_mv.col = best_mv->col << 3;
-
- if (bestsad == INT_MAX)
- return INT_MAX;
-
- return fn_ptr->vf(what, what_stride, best_address, in_what_stride, (unsigned int *)(&thissad))
- + mv_err_cost(&this_mv, center_mv, mvcost, error_per_bit);
-}
-
-int vp8_diamond_search_sadx4
-(
- MACROBLOCK *x,
- BLOCK *b,
- BLOCKD *d,
- MV *ref_mv,
- MV *best_mv,
- int search_param,
- int error_per_bit,
- int *num00,
- vp8_variance_fn_ptr_t *fn_ptr,
- int *mvsadcost[2],
- int *mvcost[2],
- MV *center_mv
-)
-{
- int i, j, step;
-
- unsigned char *what = (*(b->base_src) + b->src);
- int what_stride = b->src_stride;
- unsigned char *in_what;
- int in_what_stride = d->pre_stride;
- unsigned char *best_address;
-
- int tot_steps;
- MV this_mv;
-
- int bestsad = INT_MAX;
- int best_site = 0;
- int last_site = 0;
-
- int ref_row = ref_mv->row >> 3;
- int ref_col = ref_mv->col >> 3;
- int this_row_offset;
- int this_col_offset;
- search_site *ss;
-
- unsigned char *check_here;
- unsigned int thissad;
-
- *num00 = 0;
-
- // Work out the start point for the search
- in_what = (unsigned char *)(*(d->base_pre) + d->pre + (ref_row * (d->pre_stride)) + ref_col);
- best_address = in_what;
-
- // We need to check that the starting point for the search (as indicated by ref_mv) is within the buffer limits
- if ((ref_col > x->mv_col_min) && (ref_col < x->mv_col_max) &&
- (ref_row > x->mv_row_min) && (ref_row < x->mv_row_max))
- {
- // Check the starting position
- bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride, 0x7fffffff) + mv_err_cost(ref_mv, center_mv, mvsadcost, error_per_bit);
- }
-
- // search_param determines the length of the initial step and hence the number of iterations
- // 0 = initial step (MAX_FIRST_STEP) pel : 1 = (MAX_FIRST_STEP/2) pel, 2 = (MAX_FIRST_STEP/4) pel... etc.
- ss = &x->ss[search_param * x->searches_per_step];
- tot_steps = (x->ss_count / x->searches_per_step) - search_param;
-
- i = 1;
- best_mv->row = ref_row;
- best_mv->col = ref_col;
-
- for (step = 0; step < tot_steps ; step++)
- {
- int all_in = 1, t;
-
- // To know if all neighbor points are within the bounds, 4 bounds checking are enough instead of
- // checking 4 bounds for each points.
- all_in &= ((best_mv->row + ss[i].mv.row)> x->mv_row_min);
- all_in &= ((best_mv->row + ss[i+1].mv.row) < x->mv_row_max);
- all_in &= ((best_mv->col + ss[i+2].mv.col) > x->mv_col_min);
- all_in &= ((best_mv->col + ss[i+3].mv.col) < x->mv_col_max);
-
- if (all_in)
- {
- unsigned int sad_array[4];
-
- for (j = 0 ; j < x->searches_per_step ; j += 4)
- {
- unsigned char *block_offset[4];
-
- for (t = 0; t < 4; t++)
- block_offset[t] = ss[i+t].offset + best_address;
-
- fn_ptr->sdx4df(what, what_stride, block_offset, in_what_stride, sad_array);
-
- for (t = 0; t < 4; t++, i++)
- {
- if (sad_array[t] < bestsad)
- {
- this_mv.row = (best_mv->row + ss[i].mv.row) << 3;
- this_mv.col = (best_mv->col + ss[i].mv.col) << 3;
- sad_array[t] += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit);
-
- if (sad_array[t] < bestsad)
- {
- bestsad = sad_array[t];
- best_site = i;
- }
- }
- }
- }
- }
- else
- {
- for (j = 0 ; j < x->searches_per_step ; j++)
- {
- // Trap illegal vectors
- this_row_offset = best_mv->row + ss[i].mv.row;
- this_col_offset = best_mv->col + ss[i].mv.col;
-
- if ((this_col_offset > x->mv_col_min) && (this_col_offset < x->mv_col_max) &&
- (this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
- {
- check_here = ss[i].offset + best_address;
- thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
-
- if (thissad < bestsad)
- {
- this_mv.row = this_row_offset << 3;
- this_mv.col = this_col_offset << 3;
- thissad += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit);
-
- if (thissad < bestsad)
- {
- bestsad = thissad;
- best_site = i;
- }
- }
- }
- i++;
- }
- }
-
- if (best_site != last_site)
- {
- best_mv->row += ss[best_site].mv.row;
- best_mv->col += ss[best_site].mv.col;
- best_address += ss[best_site].offset;
- last_site = best_site;
- }
- else if (best_address == in_what)
- (*num00)++;
- }
-
- this_mv.row = best_mv->row << 3;
- this_mv.col = best_mv->col << 3;
-
- if (bestsad == INT_MAX)
- return INT_MAX;
-
- return fn_ptr->vf(what, what_stride, best_address, in_what_stride, (unsigned int *)(&thissad))
- + mv_err_cost(&this_mv, center_mv, mvcost, error_per_bit);
-}
-
-
-#if !(CONFIG_REALTIME_ONLY)
-int vp8_full_search_sad(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *ref_mv, int error_per_bit, int distance, vp8_variance_fn_ptr_t *fn_ptr, int *mvcost[2], int *mvsadcost[2], MV *center_mv)
-{
- unsigned char *what = (*(b->base_src) + b->src);
- int what_stride = b->src_stride;
- unsigned char *in_what;
- int in_what_stride = d->pre_stride;
- int mv_stride = d->pre_stride;
- unsigned char *bestaddress;
- MV *best_mv = &d->bmi.mv.as_mv;
- MV this_mv;
- int bestsad = INT_MAX;
- int r, c;
-
- unsigned char *check_here;
- int thissad;
-
- int ref_row = ref_mv->row >> 3;
- int ref_col = ref_mv->col >> 3;
-
- int row_min = ref_row - distance;
- int row_max = ref_row + distance;
- int col_min = ref_col - distance;
- int col_max = ref_col + distance;
-
- // Work out the mid point for the search
- in_what = *(d->base_pre) + d->pre;
- bestaddress = in_what + (ref_row * d->pre_stride) + ref_col;
-
- best_mv->row = ref_row;
- best_mv->col = ref_col;
-
- // We need to check that the starting point for the search (as indicated by ref_mv) is within the buffer limits
- if ((ref_col > x->mv_col_min) && (ref_col < x->mv_col_max) &&
- (ref_row > x->mv_row_min) && (ref_row < x->mv_row_max))
- {
- // Baseline value at the centre
-
- //bestsad = fn_ptr->sf( what,what_stride,bestaddress,in_what_stride) + (int)sqrt(mv_err_cost(ref_mv,ref_mv, mvcost,error_per_bit*14));
- bestsad = fn_ptr->sdf(what, what_stride, bestaddress, in_what_stride, 0x7fffffff) + mv_err_cost(ref_mv, center_mv, mvsadcost, error_per_bit);
- }
-
- // Apply further limits to prevent us looking using vectors that stretch beyiond the UMV border
- if (col_min < x->mv_col_min)
- col_min = x->mv_col_min;
-
- if (col_max > x->mv_col_max)
- col_max = x->mv_col_max;
-
- if (row_min < x->mv_row_min)
- row_min = x->mv_row_min;
-
- if (row_max > x->mv_row_max)
- row_max = x->mv_row_max;
-
- for (r = row_min; r < row_max ; r++)
- {
- this_mv.row = r << 3;
- check_here = r * mv_stride + in_what + col_min;
-
- for (c = col_min; c < col_max; c++)
- {
- thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
-
- this_mv.col = c << 3;
- //thissad += (int)sqrt(mv_err_cost(&this_mv,ref_mv, mvcost,error_per_bit*14));
- //thissad += error_per_bit * mv_bits_sadcost[mv_bits(&this_mv, ref_mv, mvcost)];
- thissad += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit); //mv_bits(error_per_bit, &this_mv, ref_mv, mvsadcost);
-
- if (thissad < bestsad)
- {
- bestsad = thissad;
- best_mv->row = r;
- best_mv->col = c;
- bestaddress = check_here;
- }
-
- check_here++;
- }
- }
-
- this_mv.row = best_mv->row << 3;
- this_mv.col = best_mv->col << 3;
-
- if (bestsad < INT_MAX)
- return fn_ptr->vf(what, what_stride, bestaddress, in_what_stride, (unsigned int *)(&thissad))
- + mv_err_cost(&this_mv, center_mv, mvcost, error_per_bit);
- else
- return INT_MAX;
-}
-
-int vp8_full_search_sadx3(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *ref_mv, int error_per_bit, int distance, vp8_variance_fn_ptr_t *fn_ptr, int *mvcost[2], int *mvsadcost[2], MV *center_mv)
-{
- unsigned char *what = (*(b->base_src) + b->src);
- int what_stride = b->src_stride;
- unsigned char *in_what;
- int in_what_stride = d->pre_stride;
- int mv_stride = d->pre_stride;
- unsigned char *bestaddress;
- MV *best_mv = &d->bmi.mv.as_mv;
- MV this_mv;
- int bestsad = INT_MAX;
- int r, c;
-
- unsigned char *check_here;
- unsigned int thissad;
-
- int ref_row = ref_mv->row >> 3;
- int ref_col = ref_mv->col >> 3;
-
- int row_min = ref_row - distance;
- int row_max = ref_row + distance;
- int col_min = ref_col - distance;
- int col_max = ref_col + distance;
-
- unsigned int sad_array[3];
-
- // Work out the mid point for the search
- in_what = *(d->base_pre) + d->pre;
- bestaddress = in_what + (ref_row * d->pre_stride) + ref_col;
-
- best_mv->row = ref_row;
- best_mv->col = ref_col;
-
- // We need to check that the starting point for the search (as indicated by ref_mv) is within the buffer limits
- if ((ref_col > x->mv_col_min) && (ref_col < x->mv_col_max) &&
- (ref_row > x->mv_row_min) && (ref_row < x->mv_row_max))
- {
- // Baseline value at the centre
- bestsad = fn_ptr->sdf(what, what_stride, bestaddress, in_what_stride, 0x7fffffff) + mv_err_cost(ref_mv, center_mv, mvsadcost, error_per_bit);
- }
-
- // Apply further limits to prevent us looking using vectors that stretch beyiond the UMV border
- if (col_min < x->mv_col_min)
- col_min = x->mv_col_min;
-
- if (col_max > x->mv_col_max)
- col_max = x->mv_col_max;
-
- if (row_min < x->mv_row_min)
- row_min = x->mv_row_min;
-
- if (row_max > x->mv_row_max)
- row_max = x->mv_row_max;
-
- for (r = row_min; r < row_max ; r++)
- {
- this_mv.row = r << 3;
- check_here = r * mv_stride + in_what + col_min;
- c = col_min;
-
- while ((c + 2) < col_max)
- {
- int i;
-
- fn_ptr->sdx3f(what, what_stride, check_here , in_what_stride, sad_array);
-
- for (i = 0; i < 3; i++)
- {
- thissad = sad_array[i];
-
- if (thissad < bestsad)
- {
- this_mv.col = c << 3;
- thissad += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit);
-
- if (thissad < bestsad)
- {
- bestsad = thissad;
- best_mv->row = r;
- best_mv->col = c;
- bestaddress = check_here;
- }
- }
-
- check_here++;
- c++;
- }
- }
-
- while (c < col_max)
- {
- thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
-
- if (thissad < bestsad)
- {
- this_mv.col = c << 3;
- thissad += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit);
-
- if (thissad < bestsad)
- {
- bestsad = thissad;
- best_mv->row = r;
- best_mv->col = c;
- bestaddress = check_here;
- }
- }
-
- check_here ++;
- c ++;
- }
-
- }
-
- this_mv.row = best_mv->row << 3;
- this_mv.col = best_mv->col << 3;
-
- if (bestsad < INT_MAX)
- return fn_ptr->vf(what, what_stride, bestaddress, in_what_stride, (unsigned int *)(&thissad))
- + mv_err_cost(&this_mv, center_mv, mvcost, error_per_bit);
- else
- return INT_MAX;
-}
-
-int vp8_full_search_sadx8(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *ref_mv, int error_per_bit, int distance, vp8_variance_fn_ptr_t *fn_ptr, int *mvcost[2], int *mvsadcost[2], MV *center_mv)
-{
- unsigned char *what = (*(b->base_src) + b->src);
- int what_stride = b->src_stride;
- unsigned char *in_what;
- int in_what_stride = d->pre_stride;
- int mv_stride = d->pre_stride;
- unsigned char *bestaddress;
- MV *best_mv = &d->bmi.mv.as_mv;
- MV this_mv;
- int bestsad = INT_MAX;
- int r, c;
-
- unsigned char *check_here;
- unsigned int thissad;
-
- int ref_row = ref_mv->row >> 3;
- int ref_col = ref_mv->col >> 3;
-
- int row_min = ref_row - distance;
- int row_max = ref_row + distance;
- int col_min = ref_col - distance;
- int col_max = ref_col + distance;
-
- DECLARE_ALIGNED_ARRAY(16, unsigned short, sad_array8, 8);
- unsigned int sad_array[3];
-
- // Work out the mid point for the search
- in_what = *(d->base_pre) + d->pre;
- bestaddress = in_what + (ref_row * d->pre_stride) + ref_col;
-
- best_mv->row = ref_row;
- best_mv->col = ref_col;
-
- // We need to check that the starting point for the search (as indicated by ref_mv) is within the buffer limits
- if ((ref_col > x->mv_col_min) && (ref_col < x->mv_col_max) &&
- (ref_row > x->mv_row_min) && (ref_row < x->mv_row_max))
- {
- // Baseline value at the centre
- bestsad = fn_ptr->sdf(what, what_stride, bestaddress, in_what_stride, 0x7fffffff) + mv_err_cost(ref_mv, center_mv, mvsadcost, error_per_bit);
- }
-
- // Apply further limits to prevent us looking using vectors that stretch beyiond the UMV border
- if (col_min < x->mv_col_min)
- col_min = x->mv_col_min;
-
- if (col_max > x->mv_col_max)
- col_max = x->mv_col_max;
-
- if (row_min < x->mv_row_min)
- row_min = x->mv_row_min;
-
- if (row_max > x->mv_row_max)
- row_max = x->mv_row_max;
-
- for (r = row_min; r < row_max ; r++)
- {
- this_mv.row = r << 3;
- check_here = r * mv_stride + in_what + col_min;
- c = col_min;
-
- while ((c + 7) < col_max)
- {
- int i;
-
- fn_ptr->sdx8f(what, what_stride, check_here , in_what_stride, sad_array8);
-
- for (i = 0; i < 8; i++)
- {
- thissad = (unsigned int)sad_array8[i];
-
- if (thissad < bestsad)
- {
- this_mv.col = c << 3;
- thissad += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit);
-
- if (thissad < bestsad)
- {
- bestsad = thissad;
- best_mv->row = r;
- best_mv->col = c;
- bestaddress = check_here;
- }
- }
-
- check_here++;
- c++;
- }
- }
-
- while ((c + 2) < col_max)
- {
- int i;
-
- fn_ptr->sdx3f(what, what_stride, check_here , in_what_stride, sad_array);
-
- for (i = 0; i < 3; i++)
- {
- thissad = sad_array[i];
-
- if (thissad < bestsad)
- {
- this_mv.col = c << 3;
- thissad += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit);
-
- if (thissad < bestsad)
- {
- bestsad = thissad;
- best_mv->row = r;
- best_mv->col = c;
- bestaddress = check_here;
- }
- }
-
- check_here++;
- c++;
- }
- }
-
- while (c < col_max)
- {
- thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
-
- if (thissad < bestsad)
- {
- this_mv.col = c << 3;
- thissad += mv_err_cost(&this_mv, center_mv, mvsadcost, error_per_bit);
-
- if (thissad < bestsad)
- {
- bestsad = thissad;
- best_mv->row = r;
- best_mv->col = c;
- bestaddress = check_here;
- }
- }
-
- check_here ++;
- c ++;
- }
- }
-
- this_mv.row = best_mv->row << 3;
- this_mv.col = best_mv->col << 3;
-
- if (bestsad < INT_MAX)
- return fn_ptr->vf(what, what_stride, bestaddress, in_what_stride, (unsigned int *)(&thissad))
- + mv_err_cost(&this_mv, center_mv, mvcost, error_per_bit);
- else
- return INT_MAX;
-}
-#endif /* !(CONFIG_REALTIME_ONLY) */
-
-#ifdef ENTROPY_STATS
-void print_mode_context(void)
-{
- FILE *f = fopen("modecont.c", "w");
- int i, j;
-
- fprintf(f, "#include \"entropy.h\"\n");
- fprintf(f, "const int vp8_mode_contexts[6][4] =\n");
- fprintf(f, "{\n");
-
- for (j = 0; j < 6; j++)
- {
- fprintf(f, " { // %d \n", j);
- fprintf(f, " ");
-
- for (i = 0; i < 4; i++)
- {
- int overal_prob;
- int this_prob;
- int count; // = mv_ref_ct[j][i][0]+mv_ref_ct[j][i][1];
-
- // Overall probs
- count = mv_mode_cts[i][0] + mv_mode_cts[i][1];
-
- if (count)
- overal_prob = 256 * mv_mode_cts[i][0] / count;
- else
- overal_prob = 128;
-
- if (overal_prob == 0)
- overal_prob = 1;
-
- // context probs
- count = mv_ref_ct[j][i][0] + mv_ref_ct[j][i][1];
-
- if (count)
- this_prob = 256 * mv_ref_ct[j][i][0] / count;
- else
- this_prob = 128;
-
- if (this_prob == 0)
- this_prob = 1;
-
- fprintf(f, "%5d, ", this_prob);
- //fprintf(f,"%5d, %5d, %8d,", this_prob, overal_prob, (this_prob << 10)/overal_prob);
- //fprintf(f,"%8d, ", (this_prob << 10)/overal_prob);
- }
-
- fprintf(f, " },\n");
- }
-
- fprintf(f, "};\n");
- fclose(f);
-}
-
-/* MV ref count ENTROPY_STATS stats code */
-#ifdef ENTROPY_STATS
-void init_mv_ref_counts()
-{
- vpx_memset(mv_ref_ct, 0, sizeof(mv_ref_ct));
- vpx_memset(mv_mode_cts, 0, sizeof(mv_mode_cts));
-}
-
-void accum_mv_refs(MB_PREDICTION_MODE m, const int ct[4])
-{
- if (m == ZEROMV)
- {
- ++mv_ref_ct [ct[0]] [0] [0];
- ++mv_mode_cts[0][0];
- }
- else
- {
- ++mv_ref_ct [ct[0]] [0] [1];
- ++mv_mode_cts[0][1];
-
- if (m == NEARESTMV)
- {
- ++mv_ref_ct [ct[1]] [1] [0];
- ++mv_mode_cts[1][0];
- }
- else
- {
- ++mv_ref_ct [ct[1]] [1] [1];
- ++mv_mode_cts[1][1];
-
- if (m == NEARMV)
- {
- ++mv_ref_ct [ct[2]] [2] [0];
- ++mv_mode_cts[2][0];
- }
- else
- {
- ++mv_ref_ct [ct[2]] [2] [1];
- ++mv_mode_cts[2][1];
-
- if (m == NEWMV)
- {
- ++mv_ref_ct [ct[3]] [3] [0];
- ++mv_mode_cts[3][0];
- }
- else
- {
- ++mv_ref_ct [ct[3]] [3] [1];
- ++mv_mode_cts[3][1];
- }
- }
- }
- }
-}
-
-#endif/* END MV ref count ENTROPY_STATS stats code */
-
-#endif
diff --git a/vp8/encoder/mcomp.h b/vp8/encoder/mcomp.h
deleted file mode 100644
index 83f95c6..0000000
--- a/vp8/encoder/mcomp.h
+++ /dev/null
@@ -1,125 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_MCOMP_H
-#define __INC_MCOMP_H
-
-#include "block.h"
-#include "variance.h"
-
-#ifdef ENTROPY_STATS
-extern void init_mv_ref_counts();
-extern void accum_mv_refs(MB_PREDICTION_MODE, const int near_mv_ref_cts[4]);
-#endif
-
-
-#define MAX_MVSEARCH_STEPS 8 // The maximum number of steps in a step search given the largest allowed initial step
-#define MAX_FULL_PEL_VAL ((1 << (MAX_MVSEARCH_STEPS+3)) - 8) // Max full pel mv specified in 1/8 pel units
-#define MAX_FIRST_STEP (1 << (MAX_MVSEARCH_STEPS-1)) // Maximum size of the first step in full pel units
-
-extern void print_mode_context(void);
-extern int vp8_mv_bit_cost(MV *mv, MV *ref, int *mvcost[2], int Weight);
-extern void vp8_init_dsmotion_compensation(MACROBLOCK *x, int stride);
-extern void vp8_init3smotion_compensation(MACROBLOCK *x, int stride);
-
-
-extern int vp8_hex_search
-(
- MACROBLOCK *x,
- BLOCK *b,
- BLOCKD *d,
- MV *ref_mv,
- MV *best_mv,
- int search_param,
- int error_per_bit,
- int *num00,
- const vp8_variance_fn_ptr_t *vf,
- int *mvsadcost[2],
- int *mvcost[2],
- MV *center_mv
-);
-
-typedef int (fractional_mv_step_fp)
- (MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *bestmv, MV *ref_mv,
- int error_per_bit, const vp8_variance_fn_ptr_t *vfp, int *mvcost[2]);
-extern fractional_mv_step_fp vp8_find_best_sub_pixel_step_iteratively;
-extern fractional_mv_step_fp vp8_find_best_sub_pixel_step;
-extern fractional_mv_step_fp vp8_find_best_half_pixel_step;
-extern fractional_mv_step_fp vp8_skip_fractional_mv_step;
-
-#define prototype_full_search_sad(sym)\
- int (sym)\
- (\
- MACROBLOCK *x, \
- BLOCK *b, \
- BLOCKD *d, \
- MV *ref_mv, \
- int error_per_bit, \
- int distance, \
- vp8_variance_fn_ptr_t *fn_ptr, \
- int *mvcost[2], \
- int *mvsadcost[2], \
- MV *center_mv \
- )
-
-#define prototype_diamond_search_sad(sym)\
- int (sym)\
- (\
- MACROBLOCK *x, \
- BLOCK *b, \
- BLOCKD *d, \
- MV *ref_mv, \
- MV *best_mv, \
- int search_param, \
- int error_per_bit, \
- int *num00, \
- vp8_variance_fn_ptr_t *fn_ptr, \
- int *mvsadcost[2], \
- int *mvcost[2], \
- MV *center_mv \
- )
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/mcomp_x86.h"
-#endif
-
-typedef prototype_full_search_sad(*vp8_full_search_fn_t);
-extern prototype_full_search_sad(vp8_full_search_sad);
-extern prototype_full_search_sad(vp8_full_search_sadx3);
-extern prototype_full_search_sad(vp8_full_search_sadx8);
-
-typedef prototype_diamond_search_sad(*vp8_diamond_search_fn_t);
-extern prototype_diamond_search_sad(vp8_diamond_search_sad);
-extern prototype_diamond_search_sad(vp8_diamond_search_sadx4);
-
-#ifndef vp8_search_full_search
-#define vp8_search_full_search vp8_full_search_sad
-#endif
-extern prototype_full_search_sad(vp8_search_full_search);
-
-#ifndef vp8_search_diamond_search
-#define vp8_search_diamond_search vp8_diamond_search_sad
-#endif
-extern prototype_diamond_search_sad(vp8_search_diamond_search);
-
-typedef struct
-{
- prototype_full_search_sad(*full_search);
- prototype_diamond_search_sad(*diamond_search);
-} vp8_search_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define SEARCH_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define SEARCH_INVOKE(ctx,fn) vp8_search_##fn
-#endif
-
-#endif
diff --git a/vp8/encoder/modecosts.c b/vp8/encoder/modecosts.c
deleted file mode 100644
index c636c48..0000000
--- a/vp8/encoder/modecosts.c
+++ /dev/null
@@ -1,47 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/blockd.h"
-#include "onyx_int.h"
-#include "treewriter.h"
-#include "vp8/common/entropymode.h"
-
-
-void vp8_init_mode_costs(VP8_COMP *c)
-{
- VP8_COMMON *x = &c->common;
- {
- const vp8_tree_p T = vp8_bmode_tree;
-
- int i = 0;
-
- do
- {
- int j = 0;
-
- do
- {
- vp8_cost_tokens((int *)c->mb.bmode_costs[i][j], x->kf_bmode_prob[i][j], T);
- }
- while (++j < VP8_BINTRAMODES);
- }
- while (++i < VP8_BINTRAMODES);
-
- vp8_cost_tokens((int *)c->mb.inter_bmode_costs, x->fc.bmode_prob, T);
- }
- vp8_cost_tokens((int *)c->mb.inter_bmode_costs, x->fc.sub_mv_ref_prob, vp8_sub_mv_ref_tree);
-
- vp8_cost_tokens(c->mb.mbmode_cost[1], x->fc.ymode_prob, vp8_ymode_tree);
- vp8_cost_tokens(c->mb.mbmode_cost[0], x->kf_ymode_prob, vp8_kf_ymode_tree);
-
- vp8_cost_tokens(c->mb.intra_uv_mode_cost[1], x->fc.uv_mode_prob, vp8_uv_mode_tree);
- vp8_cost_tokens(c->mb.intra_uv_mode_cost[0], x->kf_uv_mode_prob, vp8_uv_mode_tree);
-}
diff --git a/vp8/encoder/onyx_if.c b/vp8/encoder/onyx_if.c
deleted file mode 100644
index be65545..0000000
--- a/vp8/encoder/onyx_if.c
+++ /dev/null
@@ -1,5412 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/onyxc_int.h"
-#include "onyx_int.h"
-#include "vp8/common/systemdependent.h"
-#include "quantize.h"
-#include "vp8/common/alloccommon.h"
-#include "mcomp.h"
-#include "firstpass.h"
-#include "psnr.h"
-#include "vpx_scale/vpxscale.h"
-#include "vp8/common/extend.h"
-#include "ratectrl.h"
-#include "vp8/common/quant_common.h"
-#include "segmentation.h"
-#include "vp8/common/g_common.h"
-#include "vpx_scale/yv12extend.h"
-#include "vp8/common/postproc.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vp8/common/swapyv12buffer.h"
-#include "vp8/common/threading.h"
-#include "vpx_ports/vpx_timer.h"
-#include "temporal_filter.h"
-#if ARCH_ARM
-#include "vpx_ports/arm.h"
-#endif
-
-#include <math.h>
-#include <stdio.h>
-#include <limits.h>
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IF_RTCD(x) (x)
-#define RTCD(x) &cpi->common.rtcd.x
-#else
-#define IF_RTCD(x) NULL
-#define RTCD(x) NULL
-#endif
-
-extern void vp8cx_init_mv_bits_sadcost();
-extern void vp8cx_pick_filter_level_fast(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi);
-extern void vp8cx_set_alt_lf_level(VP8_COMP *cpi, int filt_val);
-extern void vp8cx_pick_filter_level(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi);
-
-extern void vp8_init_loop_filter(VP8_COMMON *cm);
-extern void vp8_loop_filter_frame(VP8_COMMON *cm, MACROBLOCKD *mbd, int filt_val);
-extern void vp8_loop_filter_frame_yonly(VP8_COMMON *cm, MACROBLOCKD *mbd, int filt_val, int sharpness_lvl);
-extern void vp8_dmachine_specific_config(VP8_COMP *cpi);
-extern void vp8_cmachine_specific_config(VP8_COMP *cpi);
-extern void vp8_calc_auto_iframe_target_size(VP8_COMP *cpi);
-extern void vp8_deblock_frame(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *post, int filt_lvl, int low_var_thresh, int flag);
-extern void print_parms(VP8_CONFIG *ocf, char *filenam);
-extern unsigned int vp8_get_processor_freq();
-extern void print_tree_update_probs();
-extern void vp8cx_create_encoder_threads(VP8_COMP *cpi);
-extern void vp8cx_remove_encoder_threads(VP8_COMP *cpi);
-#if HAVE_ARMV7
-extern void vp8_yv12_copy_frame_func_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_src_frame_func_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-#endif
-
-int vp8_estimate_entropy_savings(VP8_COMP *cpi);
-int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, const vp8_variance_rtcd_vtable_t *rtcd);
-
-extern void vp8_temporal_filter_prepare_c(VP8_COMP *cpi);
-
-static void set_default_lf_deltas(VP8_COMP *cpi);
-
-extern const int vp8_gf_interval_table[101];
-
-#if CONFIG_PSNR
-#include "math.h"
-
-extern double vp8_calc_ssim
-(
- YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *dest,
- int lumamask,
- double *weight,
- const vp8_variance_rtcd_vtable_t *rtcd
-);
-
-
-extern double vp8_calc_ssimg
-(
- YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *dest,
- double *ssim_y,
- double *ssim_u,
- double *ssim_v
-);
-
-
-#endif
-
-
-#ifdef OUTPUT_YUV_SRC
-FILE *yuv_file;
-#endif
-
-#if 0
-FILE *framepsnr;
-FILE *kf_list;
-FILE *keyfile;
-#endif
-
-#if 0
-extern int skip_true_count;
-extern int skip_false_count;
-#endif
-
-
-#ifdef ENTROPY_STATS
-extern int intra_mode_stats[10][10][10];
-#endif
-
-#ifdef SPEEDSTATS
-unsigned int frames_at_speed[16] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
-unsigned int tot_pm = 0;
-unsigned int cnt_pm = 0;
-unsigned int tot_ef = 0;
-unsigned int cnt_ef = 0;
-#endif
-
-#ifdef MODE_STATS
-extern unsigned __int64 Sectionbits[50];
-extern int y_modes[5] ;
-extern int uv_modes[4] ;
-extern int b_modes[10] ;
-
-extern int inter_y_modes[10] ;
-extern int inter_uv_modes[4] ;
-extern unsigned int inter_b_modes[15];
-#endif
-
-extern void (*vp8_short_fdct4x4)(short *input, short *output, int pitch);
-extern void (*vp8_short_fdct8x4)(short *input, short *output, int pitch);
-
-extern const int vp8_bits_per_mb[2][QINDEX_RANGE];
-
-extern const int qrounding_factors[129];
-extern const int qzbin_factors[129];
-extern void vp8cx_init_quantizer(VP8_COMP *cpi);
-extern const int vp8cx_base_skip_false_prob[128];
-
-// Tables relating active max Q to active min Q
-static const int kf_low_motion_minq[QINDEX_RANGE] =
-{
- 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
- 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
- 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
- 0,0,0,0,1,1,1,1,1,1,1,1,2,2,2,2,
- 3,3,3,3,3,3,4,4,4,5,5,5,5,5,6,6,
- 6,6,7,7,8,8,8,8,9,9,10,10,10,10,11,11,
- 11,11,12,12,13,13,13,13,14,14,15,15,15,15,16,16,
- 16,16,17,17,18,18,18,18,19,20,20,21,21,22,23,23
-};
-static const int kf_high_motion_minq[QINDEX_RANGE] =
-{
- 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
- 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
- 1,1,1,1,1,1,1,1,2,2,2,2,3,3,3,3,
- 3,3,3,3,4,4,4,4,5,5,5,5,5,5,6,6,
- 6,6,7,7,8,8,8,8,9,9,10,10,10,10,11,11,
- 11,11,12,12,13,13,13,13,14,14,15,15,15,15,16,16,
- 16,16,17,17,18,18,18,18,19,19,20,20,20,20,21,21,
- 21,21,22,22,23,23,24,25,25,26,26,27,28,28,29,30
-};
-static const int gf_low_motion_minq[QINDEX_RANGE] =
-{
- 0,0,0,0,1,1,1,1,1,1,1,1,2,2,2,2,
- 3,3,3,3,4,4,4,4,5,5,5,5,6,6,6,6,
- 7,7,7,7,8,8,8,8,9,9,9,9,10,10,10,10,
- 11,11,12,12,13,13,14,14,15,15,16,16,17,17,18,18,
- 19,19,20,20,21,21,22,22,23,23,24,24,25,25,26,26,
- 27,27,28,28,29,29,30,30,31,31,32,32,33,33,34,34,
- 35,35,36,36,37,37,38,38,39,39,40,40,41,41,42,42,
- 43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58
-};
-static const int gf_mid_motion_minq[QINDEX_RANGE] =
-{
- 0,0,0,0,1,1,1,1,1,1,2,2,3,3,3,4,
- 4,4,5,5,5,6,6,6,7,7,7,8,8,8,9,9,
- 9,10,10,10,10,11,11,11,12,12,12,12,13,13,13,14,
- 14,14,15,15,16,16,17,17,18,18,19,19,20,20,21,21,
- 22,22,23,23,24,24,25,25,26,26,27,27,28,28,29,29,
- 30,30,31,31,32,32,33,33,34,34,35,35,36,36,37,37,
- 38,39,39,40,40,41,41,42,42,43,43,44,45,46,47,48,
- 49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64
-};
-static const int gf_high_motion_minq[QINDEX_RANGE] =
-{
- 0,0,0,0,1,1,1,1,1,2,2,2,3,3,3,4,
- 4,4,5,5,5,6,6,6,7,7,7,8,8,8,9,9,
- 9,10,10,10,11,11,12,12,13,13,14,14,15,15,16,16,
- 17,17,18,18,19,19,20,20,21,21,22,22,23,23,24,24,
- 25,25,26,26,27,27,28,28,29,29,30,30,31,31,32,32,
- 33,33,34,34,35,35,36,36,37,37,38,38,39,39,40,40,
- 41,41,42,42,43,44,45,46,47,48,49,50,51,52,53,54,
- 55,56,57,58,59,60,62,64,66,68,70,72,74,76,78,80
-};
-static const int inter_minq[QINDEX_RANGE] =
-{
- 0,0,1,1,2,3,3,4,4,5,6,6,7,8,8,9,
- 9,10,11,11,12,13,13,14,15,15,16,17,17,18,19,20,
- 20,21,22,22,23,24,24,25,26,27,27,28,29,30,30,31,
- 32,33,33,34,35,36,36,37,38,39,39,40,41,42,42,43,
- 44,45,46,46,47,48,49,50,50,51,52,53,54,55,55,56,
- 57,58,59,60,60,61,62,63,64,65,66,67,67,68,69,70,
- 71,72,73,74,75,75,76,77,78,79,80,81,82,83,84,85,
- 86,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100
-};
-
-void vp8_initialize()
-{
- static int init_done = 0;
-
- if (!init_done)
- {
- vp8_scale_machine_specific_config();
- vp8_initialize_common();
- //vp8_dmachine_specific_config();
- vp8_tokenize_initialize();
-
- vp8cx_init_mv_bits_sadcost();
- init_done = 1;
- }
-}
-#ifdef PACKET_TESTING
-extern FILE *vpxlogc;
-#endif
-
-static void setup_features(VP8_COMP *cpi)
-{
- // Set up default state for MB feature flags
- cpi->mb.e_mbd.segmentation_enabled = 0;
- cpi->mb.e_mbd.update_mb_segmentation_map = 0;
- cpi->mb.e_mbd.update_mb_segmentation_data = 0;
- vpx_memset(cpi->mb.e_mbd.mb_segment_tree_probs, 255, sizeof(cpi->mb.e_mbd.mb_segment_tree_probs));
- vpx_memset(cpi->mb.e_mbd.segment_feature_data, 0, sizeof(cpi->mb.e_mbd.segment_feature_data));
-
- cpi->mb.e_mbd.mode_ref_lf_delta_enabled = 0;
- cpi->mb.e_mbd.mode_ref_lf_delta_update = 0;
- vpx_memset(cpi->mb.e_mbd.ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
- vpx_memset(cpi->mb.e_mbd.mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
- vpx_memset(cpi->mb.e_mbd.last_ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
- vpx_memset(cpi->mb.e_mbd.last_mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
-
- set_default_lf_deltas(cpi);
-
-}
-
-
-static void dealloc_compressor_data(VP8_COMP *cpi)
-{
- vpx_free(cpi->tplist);
- cpi->tplist = NULL;
-
- // Delete last frame MV storage buffers
- vpx_free(cpi->lfmv);
- cpi->lfmv = 0;
-
- vpx_free(cpi->lf_ref_frame_sign_bias);
- cpi->lf_ref_frame_sign_bias = 0;
-
- vpx_free(cpi->lf_ref_frame);
- cpi->lf_ref_frame = 0;
-
- // Delete sementation map
- vpx_free(cpi->segmentation_map);
- cpi->segmentation_map = 0;
-
- vpx_free(cpi->active_map);
- cpi->active_map = 0;
-
- vp8_de_alloc_frame_buffers(&cpi->common);
-
- vp8_yv12_de_alloc_frame_buffer(&cpi->last_frame_uf);
- vp8_yv12_de_alloc_frame_buffer(&cpi->scaled_source);
-#if VP8_TEMPORAL_ALT_REF
- vp8_yv12_de_alloc_frame_buffer(&cpi->alt_ref_buffer.source_buffer);
-#endif
- {
- int i;
-
- for (i = 0; i < MAX_LAG_BUFFERS; i++)
- vp8_yv12_de_alloc_frame_buffer(&cpi->src_buffer[i].source_buffer);
-
- cpi->source_buffer_count = 0;
- }
-
- vpx_free(cpi->tok);
- cpi->tok = 0;
-
- // Structure used to monitor GF usage
- vpx_free(cpi->gf_active_flags);
- cpi->gf_active_flags = 0;
-
- vpx_free(cpi->mb.pip);
- cpi->mb.pip = 0;
-
-#if !(CONFIG_REALTIME_ONLY)
- vpx_free(cpi->total_stats);
- cpi->total_stats = 0;
-
- vpx_free(cpi->this_frame_stats);
- cpi->this_frame_stats = 0;
-#endif
-}
-
-static void enable_segmentation(VP8_PTR ptr)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
- // Set the appropriate feature bit
- cpi->mb.e_mbd.segmentation_enabled = 1;
- cpi->mb.e_mbd.update_mb_segmentation_map = 1;
- cpi->mb.e_mbd.update_mb_segmentation_data = 1;
-}
-static void disable_segmentation(VP8_PTR ptr)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
- // Clear the appropriate feature bit
- cpi->mb.e_mbd.segmentation_enabled = 0;
-}
-
-// Valid values for a segment are 0 to 3
-// Segmentation map is arrange as [Rows][Columns]
-static void set_segmentation_map(VP8_PTR ptr, unsigned char *segmentation_map)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
- // Copy in the new segmentation map
- vpx_memcpy(cpi->segmentation_map, segmentation_map, (cpi->common.mb_rows * cpi->common.mb_cols));
-
- // Signal that the map should be updated.
- cpi->mb.e_mbd.update_mb_segmentation_map = 1;
- cpi->mb.e_mbd.update_mb_segmentation_data = 1;
-}
-
-// The values given for each segment can be either deltas (from the default value chosen for the frame) or absolute values.
-//
-// Valid range for abs values is (0-127 for MB_LVL_ALT_Q) , (0-63 for SEGMENT_ALT_LF)
-// Valid range for delta values are (+/-127 for MB_LVL_ALT_Q) , (+/-63 for SEGMENT_ALT_LF)
-//
-// abs_delta = SEGMENT_DELTADATA (deltas) abs_delta = SEGMENT_ABSDATA (use the absolute values given).
-//
-//
-static void set_segment_data(VP8_PTR ptr, signed char *feature_data, unsigned char abs_delta)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
- cpi->mb.e_mbd.mb_segement_abs_delta = abs_delta;
- vpx_memcpy(cpi->segment_feature_data, feature_data, sizeof(cpi->segment_feature_data));
-}
-
-
-static void segmentation_test_function(VP8_PTR ptr)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
- unsigned char *seg_map;
- signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
-
- // Create a temporary map for segmentation data.
- CHECK_MEM_ERROR(seg_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
-
- // MB loop to set local segmentation map
- /*for ( i = 0; i < cpi->common.mb_rows; i++ )
- {
- for ( j = 0; j < cpi->common.mb_cols; j++ )
- {
- //seg_map[(i*cpi->common.mb_cols) + j] = (j % 2) + ((i%2)* 2);
- //if ( j < cpi->common.mb_cols/2 )
-
- // Segment 1 around the edge else 0
- if ( (i == 0) || (j == 0) || (i == (cpi->common.mb_rows-1)) || (j == (cpi->common.mb_cols-1)) )
- seg_map[(i*cpi->common.mb_cols) + j] = 1;
- //else if ( (i < 2) || (j < 2) || (i > (cpi->common.mb_rows-3)) || (j > (cpi->common.mb_cols-3)) )
- // seg_map[(i*cpi->common.mb_cols) + j] = 2;
- //else if ( (i < 5) || (j < 5) || (i > (cpi->common.mb_rows-6)) || (j > (cpi->common.mb_cols-6)) )
- // seg_map[(i*cpi->common.mb_cols) + j] = 3;
- else
- seg_map[(i*cpi->common.mb_cols) + j] = 0;
- }
- }*/
-
- // Set the segmentation Map
- set_segmentation_map(ptr, seg_map);
-
- // Activate segmentation.
- enable_segmentation(ptr);
-
- // Set up the quant segment data
- feature_data[MB_LVL_ALT_Q][0] = 0;
- feature_data[MB_LVL_ALT_Q][1] = 4;
- feature_data[MB_LVL_ALT_Q][2] = 0;
- feature_data[MB_LVL_ALT_Q][3] = 0;
- // Set up the loop segment data
- feature_data[MB_LVL_ALT_LF][0] = 0;
- feature_data[MB_LVL_ALT_LF][1] = 0;
- feature_data[MB_LVL_ALT_LF][2] = 0;
- feature_data[MB_LVL_ALT_LF][3] = 0;
-
- // Initialise the feature data structure
- // SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
- set_segment_data(ptr, &feature_data[0][0], SEGMENT_DELTADATA);
-
- // Delete sementation map
- vpx_free(seg_map);
-
- seg_map = 0;
-
-}
-
-// A simple function to cyclically refresh the background at a lower Q
-static void cyclic_background_refresh(VP8_COMP *cpi, int Q, int lf_adjustment)
-{
- unsigned char *seg_map;
- signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
- int i;
- int block_count = cpi->cyclic_refresh_mode_max_mbs_perframe;
- int mbs_in_frame = cpi->common.mb_rows * cpi->common.mb_cols;
-
- // Create a temporary map for segmentation data.
- CHECK_MEM_ERROR(seg_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
-
- cpi->cyclic_refresh_q = Q;
-
- for (i = Q; i > 0; i--)
- {
- if (vp8_bits_per_mb[cpi->common.frame_type][i] >= ((vp8_bits_per_mb[cpi->common.frame_type][Q]*(Q + 128)) / 64))
- //if ( vp8_bits_per_mb[cpi->common.frame_type][i] >= ((vp8_bits_per_mb[cpi->common.frame_type][Q]*((2*Q)+96))/64) )
- {
- break;
- }
- }
-
- cpi->cyclic_refresh_q = i;
-
- // Only update for inter frames
- if (cpi->common.frame_type != KEY_FRAME)
- {
- // Cycle through the macro_block rows
- // MB loop to set local segmentation map
- for (i = cpi->cyclic_refresh_mode_index; i < mbs_in_frame; i++)
- {
- // If the MB is as a candidate for clean up then mark it for possible boost/refresh (segment 1)
- // The segment id may get reset to 0 later if the MB gets coded anything other than last frame 0,0
- // as only (last frame 0,0) MBs are eligable for refresh : that is to say Mbs likely to be background blocks.
- if (cpi->cyclic_refresh_map[i] == 0)
- {
- seg_map[i] = 1;
- }
- else
- {
- seg_map[i] = 0;
-
- // Skip blocks that have been refreshed recently anyway.
- if (cpi->cyclic_refresh_map[i] < 0)
- //cpi->cyclic_refresh_map[i] = cpi->cyclic_refresh_map[i] / 16;
- cpi->cyclic_refresh_map[i]++;
- }
-
-
- if (block_count > 0)
- block_count--;
- else
- break;
-
- }
-
- // If we have gone through the frame reset to the start
- cpi->cyclic_refresh_mode_index = i;
-
- if (cpi->cyclic_refresh_mode_index >= mbs_in_frame)
- cpi->cyclic_refresh_mode_index = 0;
- }
-
- // Set the segmentation Map
- set_segmentation_map((VP8_PTR)cpi, seg_map);
-
- // Activate segmentation.
- enable_segmentation((VP8_PTR)cpi);
-
- // Set up the quant segment data
- feature_data[MB_LVL_ALT_Q][0] = 0;
- feature_data[MB_LVL_ALT_Q][1] = (cpi->cyclic_refresh_q - Q);
- feature_data[MB_LVL_ALT_Q][2] = 0;
- feature_data[MB_LVL_ALT_Q][3] = 0;
-
- // Set up the loop segment data
- feature_data[MB_LVL_ALT_LF][0] = 0;
- feature_data[MB_LVL_ALT_LF][1] = lf_adjustment;
- feature_data[MB_LVL_ALT_LF][2] = 0;
- feature_data[MB_LVL_ALT_LF][3] = 0;
-
- // Initialise the feature data structure
- // SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
- set_segment_data((VP8_PTR)cpi, &feature_data[0][0], SEGMENT_DELTADATA);
-
- // Delete sementation map
- vpx_free(seg_map);
-
- seg_map = 0;
-
-}
-
-static void set_default_lf_deltas(VP8_COMP *cpi)
-{
- cpi->mb.e_mbd.mode_ref_lf_delta_enabled = 1;
- cpi->mb.e_mbd.mode_ref_lf_delta_update = 1;
-
- vpx_memset(cpi->mb.e_mbd.ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
- vpx_memset(cpi->mb.e_mbd.mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
-
- // Test of ref frame deltas
- cpi->mb.e_mbd.ref_lf_deltas[INTRA_FRAME] = 2;
- cpi->mb.e_mbd.ref_lf_deltas[LAST_FRAME] = 0;
- cpi->mb.e_mbd.ref_lf_deltas[GOLDEN_FRAME] = -2;
- cpi->mb.e_mbd.ref_lf_deltas[ALTREF_FRAME] = -2;
-
- cpi->mb.e_mbd.mode_lf_deltas[0] = 4; // BPRED
- cpi->mb.e_mbd.mode_lf_deltas[1] = -2; // Zero
- cpi->mb.e_mbd.mode_lf_deltas[2] = 2; // New mv
- cpi->mb.e_mbd.mode_lf_deltas[3] = 4; // Split mv
-}
-
-void vp8_set_speed_features(VP8_COMP *cpi)
-{
- SPEED_FEATURES *sf = &cpi->sf;
- int Mode = cpi->compressor_speed;
- int Speed = cpi->Speed;
- int i;
- VP8_COMMON *cm = &cpi->common;
- int last_improved_quant = sf->improved_quant;
-
- // Initialise default mode frequency sampling variables
- for (i = 0; i < MAX_MODES; i ++)
- {
- cpi->mode_check_freq[i] = 0;
- cpi->mode_test_hit_counts[i] = 0;
- cpi->mode_chosen_counts[i] = 0;
- }
-
- cpi->mbs_tested_so_far = 0;
-
- // best quality defaults
- sf->RD = 1;
- sf->search_method = NSTEP;
- sf->improved_quant = 1;
- sf->improved_dct = 1;
- sf->auto_filter = 1;
- sf->recode_loop = 1;
- sf->quarter_pixel_search = 1;
- sf->half_pixel_search = 1;
- sf->full_freq[0] = 7;
- sf->full_freq[1] = 7;
- sf->min_fs_radius = 8;
- sf->max_fs_radius = 32;
- sf->iterative_sub_pixel = 1;
- sf->optimize_coefficients = 1;
- sf->use_fastquant_for_pick = 0;
- sf->no_skip_block4x4_search = 1;
-
- sf->first_step = 0;
- sf->max_step_search_steps = MAX_MVSEARCH_STEPS;
- sf->improved_mv_pred = 1;
-
- cpi->do_full[0] = 0;
- cpi->do_full[1] = 0;
-
- // default thresholds to 0
- for (i = 0; i < MAX_MODES; i++)
- sf->thresh_mult[i] = 0;
-
- switch (Mode)
- {
-#if !(CONFIG_REALTIME_ONLY)
- case 0: // best quality mode
- sf->thresh_mult[THR_ZEROMV ] = 0;
- sf->thresh_mult[THR_ZEROG ] = 0;
- sf->thresh_mult[THR_ZEROA ] = 0;
- sf->thresh_mult[THR_NEARESTMV] = 0;
- sf->thresh_mult[THR_NEARESTG ] = 0;
- sf->thresh_mult[THR_NEARESTA ] = 0;
- sf->thresh_mult[THR_NEARMV ] = 0;
- sf->thresh_mult[THR_NEARG ] = 0;
- sf->thresh_mult[THR_NEARA ] = 0;
-
- sf->thresh_mult[THR_DC ] = 0;
-
- sf->thresh_mult[THR_V_PRED ] = 1000;
- sf->thresh_mult[THR_H_PRED ] = 1000;
- sf->thresh_mult[THR_B_PRED ] = 2000;
- sf->thresh_mult[THR_TM ] = 1000;
-
- sf->thresh_mult[THR_NEWMV ] = 1000;
- sf->thresh_mult[THR_NEWG ] = 1000;
- sf->thresh_mult[THR_NEWA ] = 1000;
-
- sf->thresh_mult[THR_SPLITMV ] = 2500;
- sf->thresh_mult[THR_SPLITG ] = 5000;
- sf->thresh_mult[THR_SPLITA ] = 5000;
-
- sf->full_freq[0] = 7;
- sf->full_freq[1] = 15;
-
- sf->first_step = 0;
- sf->max_step_search_steps = MAX_MVSEARCH_STEPS;
- break;
- case 1:
- case 3:
- sf->thresh_mult[THR_NEARESTMV] = 0;
- sf->thresh_mult[THR_ZEROMV ] = 0;
- sf->thresh_mult[THR_DC ] = 0;
- sf->thresh_mult[THR_NEARMV ] = 0;
- sf->thresh_mult[THR_V_PRED ] = 1000;
- sf->thresh_mult[THR_H_PRED ] = 1000;
- sf->thresh_mult[THR_B_PRED ] = 2500;
- sf->thresh_mult[THR_TM ] = 1000;
-
- sf->thresh_mult[THR_NEARESTG ] = 1000;
- sf->thresh_mult[THR_NEARESTA ] = 1000;
-
- sf->thresh_mult[THR_ZEROG ] = 1000;
- sf->thresh_mult[THR_ZEROA ] = 1000;
- sf->thresh_mult[THR_NEARG ] = 1000;
- sf->thresh_mult[THR_NEARA ] = 1000;
-
-#if 1
- sf->thresh_mult[THR_ZEROMV ] = 0;
- sf->thresh_mult[THR_ZEROG ] = 0;
- sf->thresh_mult[THR_ZEROA ] = 0;
- sf->thresh_mult[THR_NEARESTMV] = 0;
- sf->thresh_mult[THR_NEARESTG ] = 0;
- sf->thresh_mult[THR_NEARESTA ] = 0;
- sf->thresh_mult[THR_NEARMV ] = 0;
- sf->thresh_mult[THR_NEARG ] = 0;
- sf->thresh_mult[THR_NEARA ] = 0;
-
-// sf->thresh_mult[THR_DC ] = 0;
-
-// sf->thresh_mult[THR_V_PRED ] = 1000;
-// sf->thresh_mult[THR_H_PRED ] = 1000;
-// sf->thresh_mult[THR_B_PRED ] = 2000;
-// sf->thresh_mult[THR_TM ] = 1000;
-
- sf->thresh_mult[THR_NEWMV ] = 1000;
- sf->thresh_mult[THR_NEWG ] = 1000;
- sf->thresh_mult[THR_NEWA ] = 1000;
-
- sf->thresh_mult[THR_SPLITMV ] = 1700;
- sf->thresh_mult[THR_SPLITG ] = 4500;
- sf->thresh_mult[THR_SPLITA ] = 4500;
-#else
- sf->thresh_mult[THR_NEWMV ] = 1500;
- sf->thresh_mult[THR_NEWG ] = 1500;
- sf->thresh_mult[THR_NEWA ] = 1500;
-
- sf->thresh_mult[THR_SPLITMV ] = 5000;
- sf->thresh_mult[THR_SPLITG ] = 10000;
- sf->thresh_mult[THR_SPLITA ] = 10000;
-#endif
- sf->full_freq[0] = 15;
- sf->full_freq[1] = 31;
-
- if (Speed > 0)
- {
- /* Disable coefficient optimization above speed 0 */
- sf->optimize_coefficients = 0;
- sf->use_fastquant_for_pick = 1;
- sf->no_skip_block4x4_search = 0;
-
- sf->first_step = 1;
-
- cpi->mode_check_freq[THR_SPLITG] = 2;
- cpi->mode_check_freq[THR_SPLITA] = 2;
- cpi->mode_check_freq[THR_SPLITMV] = 0;
- }
-
- if (Speed > 1)
- {
- cpi->mode_check_freq[THR_SPLITG] = 4;
- cpi->mode_check_freq[THR_SPLITA] = 4;
- cpi->mode_check_freq[THR_SPLITMV] = 2;
-
- sf->thresh_mult[THR_TM ] = 1500;
- sf->thresh_mult[THR_V_PRED ] = 1500;
- sf->thresh_mult[THR_H_PRED ] = 1500;
- sf->thresh_mult[THR_B_PRED ] = 5000;
-
- if (cpi->ref_frame_flags & VP8_LAST_FLAG)
- {
- sf->thresh_mult[THR_NEWMV ] = 2000;
- sf->thresh_mult[THR_SPLITMV ] = 10000;
- }
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- sf->thresh_mult[THR_NEARESTG ] = 1500;
- sf->thresh_mult[THR_ZEROG ] = 1500;
- sf->thresh_mult[THR_NEARG ] = 1500;
- sf->thresh_mult[THR_NEWG ] = 2000;
- sf->thresh_mult[THR_SPLITG ] = 20000;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- sf->thresh_mult[THR_NEARESTA ] = 1500;
- sf->thresh_mult[THR_ZEROA ] = 1500;
- sf->thresh_mult[THR_NEARA ] = 1500;
- sf->thresh_mult[THR_NEWA ] = 2000;
- sf->thresh_mult[THR_SPLITA ] = 20000;
- }
- }
-
- if (Speed > 2)
- {
- cpi->mode_check_freq[THR_SPLITG] = 15;
- cpi->mode_check_freq[THR_SPLITA] = 15;
- cpi->mode_check_freq[THR_SPLITMV] = 7;
-
- sf->thresh_mult[THR_TM ] = 2000;
- sf->thresh_mult[THR_V_PRED ] = 2000;
- sf->thresh_mult[THR_H_PRED ] = 2000;
- sf->thresh_mult[THR_B_PRED ] = 7500;
-
- if (cpi->ref_frame_flags & VP8_LAST_FLAG)
- {
- sf->thresh_mult[THR_NEWMV ] = 2000;
- sf->thresh_mult[THR_SPLITMV ] = 25000;
- }
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- sf->thresh_mult[THR_NEARESTG ] = 2000;
- sf->thresh_mult[THR_ZEROG ] = 2000;
- sf->thresh_mult[THR_NEARG ] = 2000;
- sf->thresh_mult[THR_NEWG ] = 2500;
- sf->thresh_mult[THR_SPLITG ] = 50000;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- sf->thresh_mult[THR_NEARESTA ] = 2000;
- sf->thresh_mult[THR_ZEROA ] = 2000;
- sf->thresh_mult[THR_NEARA ] = 2000;
- sf->thresh_mult[THR_NEWA ] = 2500;
- sf->thresh_mult[THR_SPLITA ] = 50000;
- }
-
- sf->improved_quant = 0;
- sf->improved_dct = 0;
-
- // Only do recode loop on key frames, golden frames and
- // alt ref frames
- sf->recode_loop = 2;
-
- sf->full_freq[0] = 31;
- sf->full_freq[1] = 63;
- }
-
- if (Speed > 3)
- {
- sf->thresh_mult[THR_SPLITA ] = INT_MAX;
- sf->thresh_mult[THR_SPLITG ] = INT_MAX;
- sf->thresh_mult[THR_SPLITMV ] = INT_MAX;
-
- cpi->mode_check_freq[THR_V_PRED] = 0;
- cpi->mode_check_freq[THR_H_PRED] = 0;
- cpi->mode_check_freq[THR_B_PRED] = 0;
- cpi->mode_check_freq[THR_NEARG] = 0;
- cpi->mode_check_freq[THR_NEWG] = 0;
- cpi->mode_check_freq[THR_NEARA] = 0;
- cpi->mode_check_freq[THR_NEWA] = 0;
-
- sf->auto_filter = 1;
- sf->recode_loop = 0; // recode loop off
- sf->RD = 0; // Turn rd off
-
- sf->full_freq[0] = 63;
- sf->full_freq[1] = 127;
- }
-
- if (Speed > 4)
- {
- sf->auto_filter = 0; // Faster selection of loop filter
- sf->full_freq[0] = INT_MAX;
- sf->full_freq[1] = INT_MAX;
-
- cpi->mode_check_freq[THR_V_PRED] = 2;
- cpi->mode_check_freq[THR_H_PRED] = 2;
- cpi->mode_check_freq[THR_B_PRED] = 2;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- cpi->mode_check_freq[THR_NEARG] = 2;
- cpi->mode_check_freq[THR_NEWG] = 4;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- cpi->mode_check_freq[THR_NEARA] = 2;
- cpi->mode_check_freq[THR_NEWA] = 4;
- }
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- sf->thresh_mult[THR_NEARESTG ] = 2000;
- sf->thresh_mult[THR_ZEROG ] = 2000;
- sf->thresh_mult[THR_NEARG ] = 2000;
- sf->thresh_mult[THR_NEWG ] = 4000;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- sf->thresh_mult[THR_NEARESTA ] = 2000;
- sf->thresh_mult[THR_ZEROA ] = 2000;
- sf->thresh_mult[THR_NEARA ] = 2000;
- sf->thresh_mult[THR_NEWA ] = 4000;
- }
- }
-
- break;
-#endif
- case 2:
- sf->optimize_coefficients = 0;
- sf->recode_loop = 0;
- sf->auto_filter = 1;
- sf->iterative_sub_pixel = 1;
- sf->thresh_mult[THR_NEARESTMV] = 0;
- sf->thresh_mult[THR_ZEROMV ] = 0;
- sf->thresh_mult[THR_DC ] = 0;
- sf->thresh_mult[THR_TM ] = 0;
- sf->thresh_mult[THR_NEARMV ] = 0;
- sf->thresh_mult[THR_V_PRED ] = 1000;
- sf->thresh_mult[THR_H_PRED ] = 1000;
- sf->thresh_mult[THR_B_PRED ] = 2500;
- sf->thresh_mult[THR_NEARESTG ] = 1000;
- sf->thresh_mult[THR_ZEROG ] = 1000;
- sf->thresh_mult[THR_NEARG ] = 1000;
- sf->thresh_mult[THR_NEARESTA ] = 1000;
- sf->thresh_mult[THR_ZEROA ] = 1000;
- sf->thresh_mult[THR_NEARA ] = 1000;
- sf->thresh_mult[THR_NEWMV ] = 2000;
- sf->thresh_mult[THR_NEWG ] = 2000;
- sf->thresh_mult[THR_NEWA ] = 2000;
- sf->thresh_mult[THR_SPLITMV ] = 5000;
- sf->thresh_mult[THR_SPLITG ] = 10000;
- sf->thresh_mult[THR_SPLITA ] = 10000;
- sf->full_freq[0] = 15;
- sf->full_freq[1] = 31;
- sf->search_method = NSTEP;
-
- if (Speed > 0)
- {
- cpi->mode_check_freq[THR_SPLITG] = 4;
- cpi->mode_check_freq[THR_SPLITA] = 4;
- cpi->mode_check_freq[THR_SPLITMV] = 2;
-
- sf->thresh_mult[THR_DC ] = 0;
- sf->thresh_mult[THR_TM ] = 1000;
- sf->thresh_mult[THR_V_PRED ] = 2000;
- sf->thresh_mult[THR_H_PRED ] = 2000;
- sf->thresh_mult[THR_B_PRED ] = 5000;
-
- if (cpi->ref_frame_flags & VP8_LAST_FLAG)
- {
- sf->thresh_mult[THR_NEARESTMV] = 0;
- sf->thresh_mult[THR_ZEROMV ] = 0;
- sf->thresh_mult[THR_NEARMV ] = 0;
- sf->thresh_mult[THR_NEWMV ] = 2000;
- sf->thresh_mult[THR_SPLITMV ] = 10000;
- }
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- sf->thresh_mult[THR_NEARESTG ] = 1000;
- sf->thresh_mult[THR_ZEROG ] = 1000;
- sf->thresh_mult[THR_NEARG ] = 1000;
- sf->thresh_mult[THR_NEWG ] = 2000;
- sf->thresh_mult[THR_SPLITG ] = 20000;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- sf->thresh_mult[THR_NEARESTA ] = 1000;
- sf->thresh_mult[THR_ZEROA ] = 1000;
- sf->thresh_mult[THR_NEARA ] = 1000;
- sf->thresh_mult[THR_NEWA ] = 2000;
- sf->thresh_mult[THR_SPLITA ] = 20000;
- }
-
- sf->improved_quant = 0;
- sf->improved_dct = 0;
- }
-
- if (Speed > 1)
- {
- cpi->mode_check_freq[THR_SPLITMV] = 7;
- cpi->mode_check_freq[THR_SPLITG] = 15;
- cpi->mode_check_freq[THR_SPLITA] = 15;
-
- sf->thresh_mult[THR_TM ] = 2000;
- sf->thresh_mult[THR_V_PRED ] = 2000;
- sf->thresh_mult[THR_H_PRED ] = 2000;
- sf->thresh_mult[THR_B_PRED ] = 5000;
-
- if (cpi->ref_frame_flags & VP8_LAST_FLAG)
- {
- sf->thresh_mult[THR_NEWMV ] = 2000;
- sf->thresh_mult[THR_SPLITMV ] = 25000;
- }
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- sf->thresh_mult[THR_NEARESTG ] = 2000;
- sf->thresh_mult[THR_ZEROG ] = 2000;
- sf->thresh_mult[THR_NEARG ] = 2000;
- sf->thresh_mult[THR_NEWG ] = 2500;
- sf->thresh_mult[THR_SPLITG ] = 50000;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- sf->thresh_mult[THR_NEARESTA ] = 2000;
- sf->thresh_mult[THR_ZEROA ] = 2000;
- sf->thresh_mult[THR_NEARA ] = 2000;
- sf->thresh_mult[THR_NEWA ] = 2500;
- sf->thresh_mult[THR_SPLITA ] = 50000;
- }
-
- sf->full_freq[0] = 31;
- sf->full_freq[1] = 63;
- }
-
- if (Speed > 2)
- {
- sf->auto_filter = 0; // Faster selection of loop filter
-
- cpi->mode_check_freq[THR_V_PRED] = 2;
- cpi->mode_check_freq[THR_H_PRED] = 2;
- cpi->mode_check_freq[THR_B_PRED] = 2;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- cpi->mode_check_freq[THR_NEARG] = 2;
- cpi->mode_check_freq[THR_NEWG] = 4;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- cpi->mode_check_freq[THR_NEARA] = 2;
- cpi->mode_check_freq[THR_NEWA] = 4;
- }
-
- sf->thresh_mult[THR_SPLITMV ] = INT_MAX;
- sf->thresh_mult[THR_SPLITG ] = INT_MAX;
- sf->thresh_mult[THR_SPLITA ] = INT_MAX;
-
- sf->full_freq[0] = 63;
- sf->full_freq[1] = 127;
- }
-
- if (Speed > 3)
- {
- sf->RD = 0;
- sf->full_freq[0] = INT_MAX;
- sf->full_freq[1] = INT_MAX;
-
- sf->auto_filter = 1;
- }
-
- if (Speed > 4)
- {
- sf->auto_filter = 0; // Faster selection of loop filter
-
-#if CONFIG_REALTIME_ONLY
- sf->search_method = HEX;
-#else
- sf->search_method = DIAMOND;
-#endif
- sf->iterative_sub_pixel = 0;
-
- cpi->mode_check_freq[THR_V_PRED] = 4;
- cpi->mode_check_freq[THR_H_PRED] = 4;
- cpi->mode_check_freq[THR_B_PRED] = 4;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- cpi->mode_check_freq[THR_NEARG] = 2;
- cpi->mode_check_freq[THR_NEWG] = 4;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- cpi->mode_check_freq[THR_NEARA] = 2;
- cpi->mode_check_freq[THR_NEWA] = 4;
- }
-
- sf->thresh_mult[THR_TM ] = 2000;
- sf->thresh_mult[THR_B_PRED ] = 5000;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- sf->thresh_mult[THR_NEARESTG ] = 2000;
- sf->thresh_mult[THR_ZEROG ] = 2000;
- sf->thresh_mult[THR_NEARG ] = 2000;
- sf->thresh_mult[THR_NEWG ] = 4000;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- sf->thresh_mult[THR_NEARESTA ] = 2000;
- sf->thresh_mult[THR_ZEROA ] = 2000;
- sf->thresh_mult[THR_NEARA ] = 2000;
- sf->thresh_mult[THR_NEWA ] = 4000;
- }
- }
-
- if (Speed > 5)
- {
- // Disable split MB intra prediction mode
- sf->thresh_mult[THR_B_PRED] = INT_MAX;
- }
-
- if (Speed > 6)
- {
- unsigned int i, sum = 0;
- unsigned int total_mbs = cm->MBs;
- int thresh;
- int total_skip;
-
- int min = 2000;
-
- if (cpi->oxcf.encode_breakout > 2000)
- min = cpi->oxcf.encode_breakout;
-
- min >>= 7;
-
- for (i = 0; i < min; i++)
- {
- sum += cpi->error_bins[i];
- }
-
- total_skip = sum;
- sum = 0;
-
- // i starts from 2 to make sure thresh started from 2048
- for (; i < 1024; i++)
- {
- sum += cpi->error_bins[i];
-
- if (10 * sum >= (unsigned int)(cpi->Speed - 6)*(total_mbs - total_skip))
- break;
- }
-
- i--;
- thresh = (i << 7);
-
- if (thresh < 2000)
- thresh = 2000;
-
- if (cpi->ref_frame_flags & VP8_LAST_FLAG)
- {
- sf->thresh_mult[THR_NEWMV] = thresh;
- sf->thresh_mult[THR_NEARESTMV ] = thresh >> 1;
- sf->thresh_mult[THR_NEARMV ] = thresh >> 1;
- }
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- sf->thresh_mult[THR_NEWG] = thresh << 1;
- sf->thresh_mult[THR_NEARESTG ] = thresh;
- sf->thresh_mult[THR_NEARG ] = thresh;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- sf->thresh_mult[THR_NEWA] = thresh << 1;
- sf->thresh_mult[THR_NEARESTA ] = thresh;
- sf->thresh_mult[THR_NEARA ] = thresh;
- }
-
- // Disable other intra prediction modes
- sf->thresh_mult[THR_TM] = INT_MAX;
- sf->thresh_mult[THR_V_PRED] = INT_MAX;
- sf->thresh_mult[THR_H_PRED] = INT_MAX;
-
- sf->improved_mv_pred = 0;
- }
-
- if (Speed > 8)
- {
- sf->quarter_pixel_search = 0;
- }
-
- if (Speed > 9)
- {
- int Tmp = cpi->Speed - 8;
-
- if (Tmp > 4)
- Tmp = 4;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- cpi->mode_check_freq[THR_ZEROG] = 1 << (Tmp - 1);
- cpi->mode_check_freq[THR_NEARESTG] = 1 << (Tmp - 1);
- cpi->mode_check_freq[THR_NEARG] = 1 << Tmp;
- cpi->mode_check_freq[THR_NEWG] = 1 << (Tmp + 1);
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- cpi->mode_check_freq[THR_ZEROA] = 1 << (Tmp - 1);
- cpi->mode_check_freq[THR_NEARESTA] = 1 << (Tmp - 1);
- cpi->mode_check_freq[THR_NEARA] = 1 << Tmp;
- cpi->mode_check_freq[THR_NEWA] = 1 << (Tmp + 1);
- }
-
- cpi->mode_check_freq[THR_NEWMV] = 1 << (Tmp - 1);
- }
-
- cm->filter_type = NORMAL_LOOPFILTER;
-
- if (Speed >= 14)
- cm->filter_type = SIMPLE_LOOPFILTER;
-
- if (Speed >= 15)
- {
- sf->half_pixel_search = 0; // This has a big hit on quality. Last resort
- }
-
- vpx_memset(cpi->error_bins, 0, sizeof(cpi->error_bins));
-
- }; /* switch */
-
- /* disable frame modes if flags not set */
- if (!(cpi->ref_frame_flags & VP8_LAST_FLAG))
- {
- sf->thresh_mult[THR_NEWMV ] = INT_MAX;
- sf->thresh_mult[THR_NEARESTMV] = INT_MAX;
- sf->thresh_mult[THR_ZEROMV ] = INT_MAX;
- sf->thresh_mult[THR_NEARMV ] = INT_MAX;
- sf->thresh_mult[THR_SPLITMV ] = INT_MAX;
- }
-
- if (!(cpi->ref_frame_flags & VP8_GOLD_FLAG))
- {
- sf->thresh_mult[THR_NEARESTG ] = INT_MAX;
- sf->thresh_mult[THR_ZEROG ] = INT_MAX;
- sf->thresh_mult[THR_NEARG ] = INT_MAX;
- sf->thresh_mult[THR_NEWG ] = INT_MAX;
- sf->thresh_mult[THR_SPLITG ] = INT_MAX;
- }
-
- if (!(cpi->ref_frame_flags & VP8_ALT_FLAG))
- {
- sf->thresh_mult[THR_NEARESTA ] = INT_MAX;
- sf->thresh_mult[THR_ZEROA ] = INT_MAX;
- sf->thresh_mult[THR_NEARA ] = INT_MAX;
- sf->thresh_mult[THR_NEWA ] = INT_MAX;
- sf->thresh_mult[THR_SPLITA ] = INT_MAX;
- }
-
-
- // Slow quant, dct and trellis not worthwhile for first pass
- // so make sure they are always turned off.
- if ( cpi->pass == 1 )
- {
- sf->improved_quant = 0;
- sf->optimize_coefficients = 0;
- sf->improved_dct = 0;
- }
-
- if (cpi->sf.search_method == NSTEP)
- {
- vp8_init3smotion_compensation(&cpi->mb, cm->yv12_fb[cm->lst_fb_idx].y_stride);
- }
- else if (cpi->sf.search_method == DIAMOND)
- {
- vp8_init_dsmotion_compensation(&cpi->mb, cm->yv12_fb[cm->lst_fb_idx].y_stride);
- }
-
- if (cpi->sf.improved_dct)
- {
- cpi->mb.vp8_short_fdct8x4 = FDCT_INVOKE(&cpi->rtcd.fdct, short8x4);
- cpi->mb.vp8_short_fdct4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, short4x4);
- }
- else
- {
- cpi->mb.vp8_short_fdct8x4 = FDCT_INVOKE(&cpi->rtcd.fdct, fast8x4);
- cpi->mb.vp8_short_fdct4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, fast4x4);
- }
-
- cpi->mb.short_walsh4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, walsh_short4x4);
-
- if (cpi->sf.improved_quant)
- {
- cpi->mb.quantize_b = QUANTIZE_INVOKE(&cpi->rtcd.quantize, quantb);
- }
- else
- {
- cpi->mb.quantize_b = QUANTIZE_INVOKE(&cpi->rtcd.quantize, fastquantb);
- }
- if (cpi->sf.improved_quant != last_improved_quant)
- vp8cx_init_quantizer(cpi);
-
-#if CONFIG_RUNTIME_CPU_DETECT
- cpi->mb.e_mbd.rtcd = &cpi->common.rtcd;
-#endif
-
- if (cpi->sf.iterative_sub_pixel == 1)
- {
- cpi->find_fractional_mv_step = vp8_find_best_sub_pixel_step_iteratively;
- }
- else if (cpi->sf.quarter_pixel_search)
- {
- cpi->find_fractional_mv_step = vp8_find_best_sub_pixel_step;
- }
- else if (cpi->sf.half_pixel_search)
- {
- cpi->find_fractional_mv_step = vp8_find_best_half_pixel_step;
- }
- else
- {
- cpi->find_fractional_mv_step = vp8_skip_fractional_mv_step;
- }
-
- if (cpi->sf.optimize_coefficients == 1)
- cpi->mb.optimize = 1 + cpi->is_next_src_alt_ref;
- else
- cpi->mb.optimize = 0;
-
- if (cpi->common.full_pixel)
- cpi->find_fractional_mv_step = vp8_skip_fractional_mv_step;
-
-#ifdef SPEEDSTATS
- frames_at_speed[cpi->Speed]++;
-#endif
-}
-static void alloc_raw_frame_buffers(VP8_COMP *cpi)
-{
- int i, buffers;
- /* allocate source_buffer to be multiples of 16 */
- int width = (cpi->oxcf.Width + 15) & ~15;
-
- buffers = cpi->oxcf.lag_in_frames;
-
- if (buffers > MAX_LAG_BUFFERS)
- buffers = MAX_LAG_BUFFERS;
-
- if (buffers < 1)
- buffers = 1;
-
- for (i = 0; i < buffers; i++)
- if (vp8_yv12_alloc_frame_buffer(&cpi->src_buffer[i].source_buffer,
- width, cpi->oxcf.Height,
- 16))
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate lag buffer");
-
-#if VP8_TEMPORAL_ALT_REF
-
- if (vp8_yv12_alloc_frame_buffer(&cpi->alt_ref_buffer.source_buffer,
- width, cpi->oxcf.Height, 16))
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate altref buffer");
-
-#endif
-
- cpi->source_buffer_count = 0;
-}
-
-static int vp8_alloc_partition_data(VP8_COMP *cpi)
-{
- vpx_free(cpi->mb.pip);
-
- cpi->mb.pip = vpx_calloc((cpi->common.mb_cols + 1) *
- (cpi->common.mb_rows + 1),
- sizeof(PARTITION_INFO));
- if(!cpi->mb.pip)
- return 1;
-
- cpi->mb.pi = cpi->mb.pip + cpi->common.mode_info_stride + 1;
-
- return 0;
-}
-
-void vp8_alloc_compressor_data(VP8_COMP *cpi)
-{
- VP8_COMMON *cm = & cpi->common;
-
- int width = cm->Width;
- int height = cm->Height;
-
- if (vp8_alloc_frame_buffers(cm, width, height))
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate frame buffers");
-
- if (vp8_alloc_partition_data(cpi))
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate partition data");
-
-
- if ((width & 0xf) != 0)
- width += 16 - (width & 0xf);
-
- if ((height & 0xf) != 0)
- height += 16 - (height & 0xf);
-
-
- if (vp8_yv12_alloc_frame_buffer(&cpi->last_frame_uf,
- width, height, VP8BORDERINPIXELS))
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate last frame buffer");
-
- if (vp8_yv12_alloc_frame_buffer(&cpi->scaled_source, width, height, 16))
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate scaled source buffer");
-
-
- vpx_free(cpi->tok);
-
- {
- unsigned int tokens = cm->mb_rows * cm->mb_cols * 24 * 16;
-
- CHECK_MEM_ERROR(cpi->tok, vpx_calloc(tokens, sizeof(*cpi->tok)));
- }
-
- // Data used for real time vc mode to see if gf needs refreshing
- cpi->inter_zz_count = 0;
- cpi->gf_bad_count = 0;
- cpi->gf_update_recommended = 0;
-
-
- // Structures used to minitor GF usage
- vpx_free(cpi->gf_active_flags);
-
- CHECK_MEM_ERROR(cpi->gf_active_flags, vpx_calloc(1, cm->mb_rows * cm->mb_cols));
-
- cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
-
-#if !(CONFIG_REALTIME_ONLY)
- vpx_free(cpi->total_stats);
-
- cpi->total_stats = vpx_calloc(1, sizeof(FIRSTPASS_STATS));
-
- vpx_free(cpi->this_frame_stats);
-
- cpi->this_frame_stats = vpx_calloc(1, sizeof(FIRSTPASS_STATS));
-
- if(!cpi->total_stats || !cpi->this_frame_stats)
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate firstpass stats");
-#endif
-
-#if CONFIG_MULTITHREAD
- if (width < 640)
- cpi->mt_sync_range = 1;
- else if (width <= 1280)
- cpi->mt_sync_range = 4;
- else if (width <= 2560)
- cpi->mt_sync_range = 8;
- else
- cpi->mt_sync_range = 16;
-#endif
-
- vpx_free(cpi->tplist);
-
- CHECK_MEM_ERROR(cpi->tplist, vpx_malloc(sizeof(TOKENLIST) * cpi->common.mb_rows));
-}
-
-
-// Quant MOD
-static const int q_trans[] =
-{
- 0, 1, 2, 3, 4, 5, 7, 8,
- 9, 10, 12, 13, 15, 17, 18, 19,
- 20, 21, 23, 24, 25, 26, 27, 28,
- 29, 30, 31, 33, 35, 37, 39, 41,
- 43, 45, 47, 49, 51, 53, 55, 57,
- 59, 61, 64, 67, 70, 73, 76, 79,
- 82, 85, 88, 91, 94, 97, 100, 103,
- 106, 109, 112, 115, 118, 121, 124, 127,
-};
-
-int vp8_reverse_trans(int x)
-{
- int i;
-
- for (i = 0; i < 64; i++)
- if (q_trans[i] >= x)
- return i;
-
- return 63;
-};
-void vp8_new_frame_rate(VP8_COMP *cpi, double framerate)
-{
- if(framerate < .1)
- framerate = 30;
-
- cpi->oxcf.frame_rate = framerate;
- cpi->output_frame_rate = cpi->oxcf.frame_rate;
- cpi->per_frame_bandwidth = (int)(cpi->oxcf.target_bandwidth / cpi->output_frame_rate);
- cpi->av_per_frame_bandwidth = (int)(cpi->oxcf.target_bandwidth / cpi->output_frame_rate);
- cpi->min_frame_bandwidth = (int)(cpi->av_per_frame_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100);
-
- // Set Maximum gf/arf interval
- cpi->max_gf_interval = ((int)(cpi->output_frame_rate / 2.0) + 2);
-
- if(cpi->max_gf_interval < 12)
- cpi->max_gf_interval = 12;
-
- // Extended interval for genuinely static scenes
- cpi->static_scene_max_gf_interval = cpi->key_frame_frequency >> 1;
-
- // Special conditions when altr ref frame enabled in lagged compress mode
- if (cpi->oxcf.play_alternate && cpi->oxcf.lag_in_frames)
- {
- if (cpi->max_gf_interval > cpi->oxcf.lag_in_frames - 1)
- cpi->max_gf_interval = cpi->oxcf.lag_in_frames - 1;
-
- if (cpi->static_scene_max_gf_interval > cpi->oxcf.lag_in_frames - 1)
- cpi->static_scene_max_gf_interval = cpi->oxcf.lag_in_frames - 1;
- }
-
- if ( cpi->max_gf_interval > cpi->static_scene_max_gf_interval )
- cpi->max_gf_interval = cpi->static_scene_max_gf_interval;
-}
-
-
-static int
-rescale(int val, int num, int denom)
-{
- int64_t llnum = num;
- int64_t llden = denom;
- int64_t llval = val;
-
- return llval * llnum / llden;
-}
-
-
-static void init_config(VP8_PTR ptr, VP8_CONFIG *oxcf)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
- VP8_COMMON *cm = &cpi->common;
-
- cpi->oxcf = *oxcf;
-
- cpi->auto_gold = 1;
- cpi->auto_adjust_gold_quantizer = 1;
- cpi->goldquantizer = 1;
- cpi->goldfreq = 7;
- cpi->auto_adjust_key_quantizer = 1;
- cpi->keyquantizer = 1;
-
- cm->version = oxcf->Version;
- vp8_setup_version(cm);
-
- // change includes all joint functionality
- vp8_change_config(ptr, oxcf);
-
- // Initialize active best and worst q and average q values.
- cpi->active_worst_quality = cpi->oxcf.worst_allowed_q;
- cpi->active_best_quality = cpi->oxcf.best_allowed_q;
- cpi->avg_frame_qindex = cpi->oxcf.worst_allowed_q;
-
- // Initialise the starting buffer levels
- cpi->oxcf.starting_buffer_level =
- rescale(cpi->oxcf.starting_buffer_level,
- cpi->oxcf.target_bandwidth, 1000);
-
- cpi->buffer_level = cpi->oxcf.starting_buffer_level;
- cpi->bits_off_target = cpi->oxcf.starting_buffer_level;
-
- cpi->rolling_target_bits = cpi->av_per_frame_bandwidth;
- cpi->rolling_actual_bits = cpi->av_per_frame_bandwidth;
- cpi->long_rolling_target_bits = cpi->av_per_frame_bandwidth;
- cpi->long_rolling_actual_bits = cpi->av_per_frame_bandwidth;
-
- cpi->total_actual_bits = 0;
- cpi->total_target_vs_actual = 0;
-
-#if VP8_TEMPORAL_ALT_REF
- {
- int i;
-
- cpi->fixed_divide[0] = 0;
-
- for (i = 1; i < 512; i++)
- cpi->fixed_divide[i] = 0x80000 / i;
- }
-#endif
-}
-
-
-void vp8_change_config(VP8_PTR ptr, VP8_CONFIG *oxcf)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
- VP8_COMMON *cm = &cpi->common;
-
- if (!cpi)
- return;
-
- if (!oxcf)
- return;
-
- if (cm->version != oxcf->Version)
- {
- cm->version = oxcf->Version;
- vp8_setup_version(cm);
- }
-
- cpi->oxcf = *oxcf;
-
- switch (cpi->oxcf.Mode)
- {
-
- case MODE_REALTIME:
- cpi->pass = 0;
- cpi->compressor_speed = 2;
-
- if (cpi->oxcf.cpu_used < -16)
- {
- cpi->oxcf.cpu_used = -16;
- }
-
- if (cpi->oxcf.cpu_used > 16)
- cpi->oxcf.cpu_used = 16;
-
- break;
-
-#if !(CONFIG_REALTIME_ONLY)
- case MODE_GOODQUALITY:
- cpi->pass = 0;
- cpi->compressor_speed = 1;
-
- if (cpi->oxcf.cpu_used < -5)
- {
- cpi->oxcf.cpu_used = -5;
- }
-
- if (cpi->oxcf.cpu_used > 5)
- cpi->oxcf.cpu_used = 5;
-
- break;
-
- case MODE_BESTQUALITY:
- cpi->pass = 0;
- cpi->compressor_speed = 0;
- break;
-
- case MODE_FIRSTPASS:
- cpi->pass = 1;
- cpi->compressor_speed = 1;
- break;
- case MODE_SECONDPASS:
- cpi->pass = 2;
- cpi->compressor_speed = 1;
-
- if (cpi->oxcf.cpu_used < -5)
- {
- cpi->oxcf.cpu_used = -5;
- }
-
- if (cpi->oxcf.cpu_used > 5)
- cpi->oxcf.cpu_used = 5;
-
- break;
- case MODE_SECONDPASS_BEST:
- cpi->pass = 2;
- cpi->compressor_speed = 0;
- break;
-#endif
- }
-
- if (cpi->pass == 0)
- cpi->auto_worst_q = 1;
-
- cpi->oxcf.worst_allowed_q = q_trans[oxcf->worst_allowed_q];
- cpi->oxcf.best_allowed_q = q_trans[oxcf->best_allowed_q];
- cpi->oxcf.cq_level = q_trans[cpi->oxcf.cq_level];
-
- if (oxcf->fixed_q >= 0)
- {
- if (oxcf->worst_allowed_q < 0)
- cpi->oxcf.fixed_q = q_trans[0];
- else
- cpi->oxcf.fixed_q = q_trans[oxcf->worst_allowed_q];
-
- if (oxcf->alt_q < 0)
- cpi->oxcf.alt_q = q_trans[0];
- else
- cpi->oxcf.alt_q = q_trans[oxcf->alt_q];
-
- if (oxcf->key_q < 0)
- cpi->oxcf.key_q = q_trans[0];
- else
- cpi->oxcf.key_q = q_trans[oxcf->key_q];
-
- if (oxcf->gold_q < 0)
- cpi->oxcf.gold_q = q_trans[0];
- else
- cpi->oxcf.gold_q = q_trans[oxcf->gold_q];
-
- }
-
- cpi->baseline_gf_interval =
- cpi->oxcf.alt_freq ? cpi->oxcf.alt_freq : DEFAULT_GF_INTERVAL;
-
- cpi->ref_frame_flags = VP8_ALT_FLAG | VP8_GOLD_FLAG | VP8_LAST_FLAG;
-
- //cpi->use_golden_frame_only = 0;
- //cpi->use_last_frame_only = 0;
- cm->refresh_golden_frame = 0;
- cm->refresh_last_frame = 1;
- cm->refresh_entropy_probs = 1;
-
- if (cpi->oxcf.token_partitions >= 0 && cpi->oxcf.token_partitions <= 3)
- cm->multi_token_partition =
- (TOKEN_PARTITION) cpi->oxcf.token_partitions;
-
- setup_features(cpi);
-
- {
- int i;
-
- for (i = 0; i < MAX_MB_SEGMENTS; i++)
- cpi->segment_encode_breakout[i] = cpi->oxcf.encode_breakout;
- }
-
- // At the moment the first order values may not be > MAXQ
- if (cpi->oxcf.fixed_q > MAXQ)
- cpi->oxcf.fixed_q = MAXQ;
-
- // local file playback mode == really big buffer
- if (cpi->oxcf.end_usage == USAGE_LOCAL_FILE_PLAYBACK)
- {
- cpi->oxcf.starting_buffer_level = 60000;
- cpi->oxcf.optimal_buffer_level = 60000;
- cpi->oxcf.maximum_buffer_size = 240000;
- }
-
- // Convert target bandwidth from Kbit/s to Bit/s
- cpi->oxcf.target_bandwidth *= 1000;
-
- // Set or reset optimal and maximum buffer levels.
- if (cpi->oxcf.optimal_buffer_level == 0)
- cpi->oxcf.optimal_buffer_level = cpi->oxcf.target_bandwidth / 8;
- else
- cpi->oxcf.optimal_buffer_level =
- rescale(cpi->oxcf.optimal_buffer_level,
- cpi->oxcf.target_bandwidth, 1000);
-
- if (cpi->oxcf.maximum_buffer_size == 0)
- cpi->oxcf.maximum_buffer_size = cpi->oxcf.target_bandwidth / 8;
- else
- cpi->oxcf.maximum_buffer_size =
- rescale(cpi->oxcf.maximum_buffer_size,
- cpi->oxcf.target_bandwidth, 1000);
-
- // Set up frame rate and related parameters rate control values.
- vp8_new_frame_rate(cpi, cpi->oxcf.frame_rate);
-
- // Set absolute upper and lower quality limits
- cpi->worst_quality = cpi->oxcf.worst_allowed_q;
- cpi->best_quality = cpi->oxcf.best_allowed_q;
-
- // active values should only be modified if out of new range
- if (cpi->active_worst_quality > cpi->oxcf.worst_allowed_q)
- {
- cpi->active_worst_quality = cpi->oxcf.worst_allowed_q;
- }
- // less likely
- else if (cpi->active_worst_quality < cpi->oxcf.best_allowed_q)
- {
- cpi->active_worst_quality = cpi->oxcf.best_allowed_q;
- }
- if (cpi->active_best_quality < cpi->oxcf.best_allowed_q)
- {
- cpi->active_best_quality = cpi->oxcf.best_allowed_q;
- }
- // less likely
- else if (cpi->active_best_quality > cpi->oxcf.worst_allowed_q)
- {
- cpi->active_best_quality = cpi->oxcf.worst_allowed_q;
- }
-
- cpi->buffered_mode = (cpi->oxcf.optimal_buffer_level > 0) ? TRUE : FALSE;
-
- cpi->cq_target_quality = cpi->oxcf.cq_level;
-
- // Only allow dropped frames in buffered mode
- cpi->drop_frames_allowed = cpi->oxcf.allow_df && cpi->buffered_mode;
-
- cm->filter_type = (LOOPFILTERTYPE) cpi->filter_type;
-
- if (!cm->use_bilinear_mc_filter)
- cm->mcomp_filter_type = SIXTAP;
- else
- cm->mcomp_filter_type = BILINEAR;
-
- cpi->target_bandwidth = cpi->oxcf.target_bandwidth;
-
- cm->Width = cpi->oxcf.Width ;
- cm->Height = cpi->oxcf.Height ;
-
- cm->horiz_scale = cpi->horiz_scale;
- cm->vert_scale = cpi->vert_scale ;
-
- // As per VP8
- cpi->intra_frame_target = (4 * (cm->Width + cm->Height) / 15) * 1000;
-
- // VP8 sharpness level mapping 0-7 (vs 0-10 in general VPx dialogs)
- if (cpi->oxcf.Sharpness > 7)
- cpi->oxcf.Sharpness = 7;
-
- cm->sharpness_level = cpi->oxcf.Sharpness;
-
- if (cm->horiz_scale != NORMAL || cm->vert_scale != NORMAL)
- {
- int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
- int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
-
- Scale2Ratio(cm->horiz_scale, &hr, &hs);
- Scale2Ratio(cm->vert_scale, &vr, &vs);
-
- // always go to the next whole number
- cm->Width = (hs - 1 + cpi->oxcf.Width * hr) / hs;
- cm->Height = (vs - 1 + cpi->oxcf.Height * vr) / vs;
- }
-
- if (((cm->Width + 15) & 0xfffffff0) !=
- cm->yv12_fb[cm->lst_fb_idx].y_width ||
- ((cm->Height + 15) & 0xfffffff0) !=
- cm->yv12_fb[cm->lst_fb_idx].y_height ||
- cm->yv12_fb[cm->lst_fb_idx].y_width == 0)
- {
- alloc_raw_frame_buffers(cpi);
- vp8_alloc_compressor_data(cpi);
- }
-
- // Clamp KF frame size to quarter of data rate
- if (cpi->intra_frame_target > cpi->target_bandwidth >> 2)
- cpi->intra_frame_target = cpi->target_bandwidth >> 2;
-
- if (cpi->oxcf.fixed_q >= 0)
- {
- cpi->last_q[0] = cpi->oxcf.fixed_q;
- cpi->last_q[1] = cpi->oxcf.fixed_q;
- }
-
- cpi->Speed = cpi->oxcf.cpu_used;
-
- // force to allowlag to 0 if lag_in_frames is 0;
- if (cpi->oxcf.lag_in_frames == 0)
- {
- cpi->oxcf.allow_lag = 0;
- }
- // Limit on lag buffers as these are not currently dynamically allocated
- else if (cpi->oxcf.lag_in_frames > MAX_LAG_BUFFERS)
- cpi->oxcf.lag_in_frames = MAX_LAG_BUFFERS;
-
- // YX Temp
- cpi->last_alt_ref_sei = -1;
- cpi->is_src_frame_alt_ref = 0;
- cpi->is_next_src_alt_ref = 0;
-
-#if 0
- // Experimental RD Code
- cpi->frame_distortion = 0;
- cpi->last_frame_distortion = 0;
-#endif
-
-}
-
-#define M_LOG2_E 0.693147180559945309417
-#define log2f(x) (log (x) / (float) M_LOG2_E)
-static void cal_mvsadcosts(int *mvsadcost[2])
-{
- int i = 1;
-
- mvsadcost [0] [0] = 300;
- mvsadcost [1] [0] = 300;
-
- do
- {
- double z = 256 * (2 * (log2f(2 * i) + .6));
- mvsadcost [0][i] = (int) z;
- mvsadcost [1][i] = (int) z;
- mvsadcost [0][-i] = (int) z;
- mvsadcost [1][-i] = (int) z;
- }
- while (++i <= mv_max);
-}
-
-VP8_PTR vp8_create_compressor(VP8_CONFIG *oxcf)
-{
- int i;
- volatile union
- {
- VP8_COMP *cpi;
- VP8_PTR ptr;
- } ctx;
-
- VP8_COMP *cpi;
- VP8_COMMON *cm;
-
- cpi = ctx.cpi = vpx_memalign(32, sizeof(VP8_COMP));
- // Check that the CPI instance is valid
- if (!cpi)
- return 0;
-
- cm = &cpi->common;
-
- vpx_memset(cpi, 0, sizeof(VP8_COMP));
-
- if (setjmp(cm->error.jmp))
- {
- VP8_PTR ptr = ctx.ptr;
-
- ctx.cpi->common.error.setjmp = 0;
- vp8_remove_compressor(&ptr);
- return 0;
- }
-
- cpi->common.error.setjmp = 1;
-
- CHECK_MEM_ERROR(cpi->mb.ss, vpx_calloc(sizeof(search_site), (MAX_MVSEARCH_STEPS * 8) + 1));
-
- vp8_create_common(&cpi->common);
- vp8_cmachine_specific_config(cpi);
-
- init_config((VP8_PTR)cpi, oxcf);
-
- memcpy(cpi->base_skip_false_prob, vp8cx_base_skip_false_prob, sizeof(vp8cx_base_skip_false_prob));
- cpi->common.current_video_frame = 0;
- cpi->kf_overspend_bits = 0;
- cpi->kf_bitrate_adjustment = 0;
- cpi->frames_till_gf_update_due = 0;
- cpi->gf_overspend_bits = 0;
- cpi->non_gf_bitrate_adjustment = 0;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- cpi->prob_intra_coded = 63;
-
- // Prime the recent reference frame useage counters.
- // Hereafter they will be maintained as a sort of moving average
- cpi->recent_ref_frame_usage[INTRA_FRAME] = 1;
- cpi->recent_ref_frame_usage[LAST_FRAME] = 1;
- cpi->recent_ref_frame_usage[GOLDEN_FRAME] = 1;
- cpi->recent_ref_frame_usage[ALTREF_FRAME] = 1;
-
- // Set reference frame sign bias for ALTREF frame to 1 (for now)
- cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 1;
-
- cpi->gf_decay_rate = 0;
- cpi->baseline_gf_interval = DEFAULT_GF_INTERVAL;
-
- cpi->gold_is_last = 0 ;
- cpi->alt_is_last = 0 ;
- cpi->gold_is_alt = 0 ;
-
- // allocate memory for storing last frame's MVs for MV prediction.
- CHECK_MEM_ERROR(cpi->lfmv, vpx_calloc((cpi->common.mb_rows+2) * (cpi->common.mb_cols+2), sizeof(int_mv)));
- CHECK_MEM_ERROR(cpi->lf_ref_frame_sign_bias, vpx_calloc((cpi->common.mb_rows+2) * (cpi->common.mb_cols+2), sizeof(int)));
- CHECK_MEM_ERROR(cpi->lf_ref_frame, vpx_calloc((cpi->common.mb_rows+2) * (cpi->common.mb_cols+2), sizeof(int)));
-
- // Create the encoder segmentation map and set all entries to 0
- CHECK_MEM_ERROR(cpi->segmentation_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
- CHECK_MEM_ERROR(cpi->active_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
- vpx_memset(cpi->active_map , 1, (cpi->common.mb_rows * cpi->common.mb_cols));
- cpi->active_map_enabled = 0;
-
-#if 0
- // Experimental code for lagged and one pass
- // Initialise one_pass GF frames stats
- // Update stats used for GF selection
- if (cpi->pass == 0)
- {
- cpi->one_pass_frame_index = 0;
-
- for (i = 0; i < MAX_LAG_BUFFERS; i++)
- {
- cpi->one_pass_frame_stats[i].frames_so_far = 0;
- cpi->one_pass_frame_stats[i].frame_intra_error = 0.0;
- cpi->one_pass_frame_stats[i].frame_coded_error = 0.0;
- cpi->one_pass_frame_stats[i].frame_pcnt_inter = 0.0;
- cpi->one_pass_frame_stats[i].frame_pcnt_motion = 0.0;
- cpi->one_pass_frame_stats[i].frame_mvr = 0.0;
- cpi->one_pass_frame_stats[i].frame_mvr_abs = 0.0;
- cpi->one_pass_frame_stats[i].frame_mvc = 0.0;
- cpi->one_pass_frame_stats[i].frame_mvc_abs = 0.0;
- }
- }
-#endif
-
- // Should we use the cyclic refresh method.
- // Currently this is tied to error resilliant mode
- cpi->cyclic_refresh_mode_enabled = cpi->oxcf.error_resilient_mode;
- cpi->cyclic_refresh_mode_max_mbs_perframe = (cpi->common.mb_rows * cpi->common.mb_cols) / 40;
- cpi->cyclic_refresh_mode_index = 0;
- cpi->cyclic_refresh_q = 32;
-
- if (cpi->cyclic_refresh_mode_enabled)
- {
- CHECK_MEM_ERROR(cpi->cyclic_refresh_map, vpx_calloc((cpi->common.mb_rows * cpi->common.mb_cols), 1));
- }
- else
- cpi->cyclic_refresh_map = (signed char *) NULL;
-
- // Test function for segmentation
- //segmentation_test_function((VP8_PTR) cpi);
-
-#ifdef ENTROPY_STATS
- init_context_counters();
-#endif
-
- /*Initialize the feed-forward activity masking.*/
- cpi->activity_avg = 90<<12;
-
- cpi->frames_since_key = 8; // Give a sensible default for the first frame.
- cpi->key_frame_frequency = cpi->oxcf.key_freq;
- cpi->this_key_frame_forced = FALSE;
- cpi->next_key_frame_forced = FALSE;
-
- cpi->source_alt_ref_pending = FALSE;
- cpi->source_alt_ref_active = FALSE;
- cpi->common.refresh_alt_ref_frame = 0;
-
- cpi->b_calculate_psnr = CONFIG_PSNR;
-#if CONFIG_PSNR
- cpi->b_calculate_ssimg = 0;
-
- cpi->count = 0;
- cpi->bytes = 0;
-
- if (cpi->b_calculate_psnr)
- {
- cpi->total_sq_error = 0.0;
- cpi->total_sq_error2 = 0.0;
- cpi->total_y = 0.0;
- cpi->total_u = 0.0;
- cpi->total_v = 0.0;
- cpi->total = 0.0;
- cpi->totalp_y = 0.0;
- cpi->totalp_u = 0.0;
- cpi->totalp_v = 0.0;
- cpi->totalp = 0.0;
- cpi->tot_recode_hits = 0;
- cpi->summed_quality = 0;
- cpi->summed_weights = 0;
- }
-
- if (cpi->b_calculate_ssimg)
- {
- cpi->total_ssimg_y = 0;
- cpi->total_ssimg_u = 0;
- cpi->total_ssimg_v = 0;
- cpi->total_ssimg_all = 0;
- }
-
-#ifndef LLONG_MAX
-#define LLONG_MAX 9223372036854775807LL
-#endif
- cpi->first_time_stamp_ever = LLONG_MAX;
-
-#endif
-
- cpi->frames_till_gf_update_due = 0;
- cpi->key_frame_count = 1;
- cpi->tot_key_frame_bits = 0;
-
- cpi->ni_av_qi = cpi->oxcf.worst_allowed_q;
- cpi->ni_tot_qi = 0;
- cpi->ni_frames = 0;
- cpi->total_byte_count = 0;
-
- cpi->drop_frame = 0;
- cpi->drop_count = 0;
- cpi->max_drop_count = 0;
- cpi->max_consec_dropped_frames = 4;
-
- cpi->rate_correction_factor = 1.0;
- cpi->key_frame_rate_correction_factor = 1.0;
- cpi->gf_rate_correction_factor = 1.0;
- cpi->est_max_qcorrection_factor = 1.0;
-
- cpi->mb.mvcost[0] = &cpi->mb.mvcosts[0][mv_max+1];
- cpi->mb.mvcost[1] = &cpi->mb.mvcosts[1][mv_max+1];
- cpi->mb.mvsadcost[0] = &cpi->mb.mvsadcosts[0][mv_max+1];
- cpi->mb.mvsadcost[1] = &cpi->mb.mvsadcosts[1][mv_max+1];
-
- cal_mvsadcosts(cpi->mb.mvsadcost);
-
- for (i = 0; i < KEY_FRAME_CONTEXT; i++)
- {
- cpi->prior_key_frame_size[i] = cpi->intra_frame_target;
- cpi->prior_key_frame_distance[i] = (int)cpi->output_frame_rate;
- }
-
- cpi->check_freq[0] = 15;
- cpi->check_freq[1] = 15;
-
-#ifdef OUTPUT_YUV_SRC
- yuv_file = fopen("bd.yuv", "ab");
-#endif
-
-#if 0
- framepsnr = fopen("framepsnr.stt", "a");
- kf_list = fopen("kf_list.stt", "w");
-#endif
-
- cpi->output_pkt_list = oxcf->output_pkt_list;
-
-#if !(CONFIG_REALTIME_ONLY)
-
- if (cpi->pass == 1)
- {
- vp8_init_first_pass(cpi);
- }
- else if (cpi->pass == 2)
- {
- size_t packet_sz = sizeof(FIRSTPASS_STATS);
- int packets = oxcf->two_pass_stats_in.sz / packet_sz;
-
- cpi->stats_in = oxcf->two_pass_stats_in.buf;
- cpi->stats_in_end = (void*)((char *)cpi->stats_in
- + (packets - 1) * packet_sz);
- vp8_init_second_pass(cpi);
- }
-
-#endif
-
- if (cpi->compressor_speed == 2)
- {
- cpi->cpu_freq = 0; //vp8_get_processor_freq();
- cpi->avg_encode_time = 0;
- cpi->avg_pick_mode_time = 0;
- }
-
- vp8_set_speed_features(cpi);
-
- // Set starting values of RD threshold multipliers (128 = *1)
- for (i = 0; i < MAX_MODES; i++)
- {
- cpi->rd_thresh_mult[i] = 128;
- }
-
-#ifdef ENTROPY_STATS
- init_mv_ref_counts();
-#endif
-
-#if CONFIG_MULTITHREAD
- vp8cx_create_encoder_threads(cpi);
-#endif
-
- cpi->fn_ptr[BLOCK_16X16].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x16);
- cpi->fn_ptr[BLOCK_16X16].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x16);
- cpi->fn_ptr[BLOCK_16X16].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar16x16);
- cpi->fn_ptr[BLOCK_16X16].svf_halfpix_h = VARIANCE_INVOKE(&cpi->rtcd.variance, halfpixvar16x16_h);
- cpi->fn_ptr[BLOCK_16X16].svf_halfpix_v = VARIANCE_INVOKE(&cpi->rtcd.variance, halfpixvar16x16_v);
- cpi->fn_ptr[BLOCK_16X16].svf_halfpix_hv = VARIANCE_INVOKE(&cpi->rtcd.variance, halfpixvar16x16_hv);
- cpi->fn_ptr[BLOCK_16X16].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x16x3);
- cpi->fn_ptr[BLOCK_16X16].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x16x8);
- cpi->fn_ptr[BLOCK_16X16].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x16x4d);
-
- cpi->fn_ptr[BLOCK_16X8].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x8);
- cpi->fn_ptr[BLOCK_16X8].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x8);
- cpi->fn_ptr[BLOCK_16X8].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar16x8);
- cpi->fn_ptr[BLOCK_16X8].svf_halfpix_h = NULL;
- cpi->fn_ptr[BLOCK_16X8].svf_halfpix_v = NULL;
- cpi->fn_ptr[BLOCK_16X8].svf_halfpix_hv = NULL;
- cpi->fn_ptr[BLOCK_16X8].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x8x3);
- cpi->fn_ptr[BLOCK_16X8].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x8x8);
- cpi->fn_ptr[BLOCK_16X8].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x8x4d);
-
- cpi->fn_ptr[BLOCK_8X16].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x16);
- cpi->fn_ptr[BLOCK_8X16].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var8x16);
- cpi->fn_ptr[BLOCK_8X16].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar8x16);
- cpi->fn_ptr[BLOCK_8X16].svf_halfpix_h = NULL;
- cpi->fn_ptr[BLOCK_8X16].svf_halfpix_v = NULL;
- cpi->fn_ptr[BLOCK_8X16].svf_halfpix_hv = NULL;
- cpi->fn_ptr[BLOCK_8X16].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x16x3);
- cpi->fn_ptr[BLOCK_8X16].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x16x8);
- cpi->fn_ptr[BLOCK_8X16].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x16x4d);
-
- cpi->fn_ptr[BLOCK_8X8].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x8);
- cpi->fn_ptr[BLOCK_8X8].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var8x8);
- cpi->fn_ptr[BLOCK_8X8].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar8x8);
- cpi->fn_ptr[BLOCK_8X8].svf_halfpix_h = NULL;
- cpi->fn_ptr[BLOCK_8X8].svf_halfpix_v = NULL;
- cpi->fn_ptr[BLOCK_8X8].svf_halfpix_hv = NULL;
- cpi->fn_ptr[BLOCK_8X8].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x8x3);
- cpi->fn_ptr[BLOCK_8X8].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x8x8);
- cpi->fn_ptr[BLOCK_8X8].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x8x4d);
-
- cpi->fn_ptr[BLOCK_4X4].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad4x4);
- cpi->fn_ptr[BLOCK_4X4].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var4x4);
- cpi->fn_ptr[BLOCK_4X4].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar4x4);
- cpi->fn_ptr[BLOCK_4X4].svf_halfpix_h = NULL;
- cpi->fn_ptr[BLOCK_4X4].svf_halfpix_v = NULL;
- cpi->fn_ptr[BLOCK_4X4].svf_halfpix_hv = NULL;
- cpi->fn_ptr[BLOCK_4X4].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad4x4x3);
- cpi->fn_ptr[BLOCK_4X4].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad4x4x8);
- cpi->fn_ptr[BLOCK_4X4].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad4x4x4d);
-
-#if !(CONFIG_REALTIME_ONLY)
- cpi->full_search_sad = SEARCH_INVOKE(&cpi->rtcd.search, full_search);
-#endif
- cpi->diamond_search_sad = SEARCH_INVOKE(&cpi->rtcd.search, diamond_search);
-
- cpi->ready_for_new_frame = 1;
-
- cpi->source_encode_index = 0;
-
- // make sure frame 1 is okay
- cpi->error_bins[0] = cpi->common.MBs;
-
- //vp8cx_init_quantizer() is first called here. Add check in vp8cx_frame_init_quantizer() so that vp8cx_init_quantizer is only called later
- //when needed. This will avoid unnecessary calls of vp8cx_init_quantizer() for every frame.
- vp8cx_init_quantizer(cpi);
- {
- vp8_init_loop_filter(cm);
- cm->last_frame_type = KEY_FRAME;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
- }
- cpi->common.error.setjmp = 0;
- return (VP8_PTR) cpi;
-
-}
-
-
-void vp8_remove_compressor(VP8_PTR *ptr)
-{
- VP8_COMP *cpi = (VP8_COMP *)(*ptr);
-
- if (!cpi)
- return;
-
- if (cpi && (cpi->common.current_video_frame > 0))
- {
-#if !(CONFIG_REALTIME_ONLY)
-
- if (cpi->pass == 2)
- {
- vp8_end_second_pass(cpi);
- }
-
-#endif
-
-#ifdef ENTROPY_STATS
- print_context_counters();
- print_tree_update_probs();
- print_mode_context();
-#endif
-
-#if CONFIG_PSNR
-
- if (cpi->pass != 1)
- {
- FILE *f = fopen("opsnr.stt", "a");
- double time_encoded = (cpi->source_end_time_stamp - cpi->first_time_stamp_ever) / 10000000.000;
- double total_encode_time = (cpi->time_receive_data + cpi->time_compress_data) / 1000.000;
- double dr = (double)cpi->bytes * (double) 8 / (double)1000 / time_encoded;
-
- if (cpi->b_calculate_psnr)
- {
- YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx];
- double samples = 3.0 / 2 * cpi->count * lst_yv12->y_width * lst_yv12->y_height;
- double total_psnr = vp8_mse2psnr(samples, 255.0, cpi->total_sq_error);
- double total_psnr2 = vp8_mse2psnr(samples, 255.0, cpi->total_sq_error2);
- double total_ssim = 100 * pow(cpi->summed_quality / cpi->summed_weights, 8.0);
-
- fprintf(f, "Bitrate\tAVGPsnr\tGLBPsnr\tAVPsnrP\tGLPsnrP\tVPXSSIM\t Time(us)\n");
- fprintf(f, "%7.3f\t%7.3f\t%7.3f\t%7.3f\t%7.3f\t%7.3f %8.0f\n",
- dr, cpi->total / cpi->count, total_psnr, cpi->totalp / cpi->count, total_psnr2, total_ssim,
- total_encode_time);
- }
-
- if (cpi->b_calculate_ssimg)
- {
- fprintf(f, "BitRate\tSSIM_Y\tSSIM_U\tSSIM_V\tSSIM_A\t Time(us)\n");
- fprintf(f, "%7.3f\t%6.4f\t%6.4f\t%6.4f\t%6.4f\t%8.0f\n", dr,
- cpi->total_ssimg_y / cpi->count, cpi->total_ssimg_u / cpi->count,
- cpi->total_ssimg_v / cpi->count, cpi->total_ssimg_all / cpi->count, total_encode_time);
- }
-
- fclose(f);
-#if 0
- f = fopen("qskip.stt", "a");
- fprintf(f, "minq:%d -maxq:%d skipture:skipfalse = %d:%d\n", cpi->oxcf.best_allowed_q, cpi->oxcf.worst_allowed_q, skiptruecount, skipfalsecount);
- fclose(f);
-#endif
-
- }
-
-#endif
-
-
-#ifdef SPEEDSTATS
-
- if (cpi->compressor_speed == 2)
- {
- int i;
- FILE *f = fopen("cxspeed.stt", "a");
- cnt_pm /= cpi->common.MBs;
-
- for (i = 0; i < 16; i++)
- fprintf(f, "%5d", frames_at_speed[i]);
-
- fprintf(f, "\n");
- //fprintf(f, "%10d PM %10d %10d %10d EF %10d %10d %10d\n", cpi->Speed, cpi->avg_pick_mode_time, (tot_pm/cnt_pm), cnt_pm, cpi->avg_encode_time, 0, 0);
- fclose(f);
- }
-
-#endif
-
-
-#ifdef MODE_STATS
- {
- extern int count_mb_seg[4];
- FILE *f = fopen("modes.stt", "a");
- double dr = (double)cpi->oxcf.frame_rate * (double)bytes * (double)8 / (double)count / (double)1000 ;
- fprintf(f, "intra_mode in Intra Frames:\n");
- fprintf(f, "Y: %8d, %8d, %8d, %8d, %8d\n", y_modes[0], y_modes[1], y_modes[2], y_modes[3], y_modes[4]);
- fprintf(f, "UV:%8d, %8d, %8d, %8d\n", uv_modes[0], uv_modes[1], uv_modes[2], uv_modes[3]);
- fprintf(f, "B: ");
- {
- int i;
-
- for (i = 0; i < 10; i++)
- fprintf(f, "%8d, ", b_modes[i]);
-
- fprintf(f, "\n");
-
- }
-
- fprintf(f, "Modes in Inter Frames:\n");
- fprintf(f, "Y: %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d\n",
- inter_y_modes[0], inter_y_modes[1], inter_y_modes[2], inter_y_modes[3], inter_y_modes[4],
- inter_y_modes[5], inter_y_modes[6], inter_y_modes[7], inter_y_modes[8], inter_y_modes[9]);
- fprintf(f, "UV:%8d, %8d, %8d, %8d\n", inter_uv_modes[0], inter_uv_modes[1], inter_uv_modes[2], inter_uv_modes[3]);
- fprintf(f, "B: ");
- {
- int i;
-
- for (i = 0; i < 15; i++)
- fprintf(f, "%8d, ", inter_b_modes[i]);
-
- fprintf(f, "\n");
-
- }
- fprintf(f, "P:%8d, %8d, %8d, %8d\n", count_mb_seg[0], count_mb_seg[1], count_mb_seg[2], count_mb_seg[3]);
- fprintf(f, "PB:%8d, %8d, %8d, %8d\n", inter_b_modes[LEFT4X4], inter_b_modes[ABOVE4X4], inter_b_modes[ZERO4X4], inter_b_modes[NEW4X4]);
-
-
-
- fclose(f);
- }
-#endif
-
-#ifdef ENTROPY_STATS
- {
- int i, j, k;
- FILE *fmode = fopen("modecontext.c", "w");
-
- fprintf(fmode, "\n#include \"entropymode.h\"\n\n");
- fprintf(fmode, "const unsigned int vp8_kf_default_bmode_counts ");
- fprintf(fmode, "[VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES] =\n{\n");
-
- for (i = 0; i < 10; i++)
- {
-
- fprintf(fmode, " { //Above Mode : %d\n", i);
-
- for (j = 0; j < 10; j++)
- {
-
- fprintf(fmode, " {");
-
- for (k = 0; k < 10; k++)
- {
- if (!intra_mode_stats[i][j][k])
- fprintf(fmode, " %5d, ", 1);
- else
- fprintf(fmode, " %5d, ", intra_mode_stats[i][j][k]);
- }
-
- fprintf(fmode, "}, // left_mode %d\n", j);
-
- }
-
- fprintf(fmode, " },\n");
-
- }
-
- fprintf(fmode, "};\n");
- fclose(fmode);
- }
-#endif
-
-
-#if defined(SECTIONBITS_OUTPUT)
-
- if (0)
- {
- int i;
- FILE *f = fopen("tokenbits.stt", "a");
-
- for (i = 0; i < 28; i++)
- fprintf(f, "%8d", (int)(Sectionbits[i] / 256));
-
- fprintf(f, "\n");
- fclose(f);
- }
-
-#endif
-
-#if 0
- {
- printf("\n_pick_loop_filter_level:%d\n", cpi->time_pick_lpf / 1000);
- printf("\n_frames recive_data encod_mb_row compress_frame Total\n");
- printf("%6d %10ld %10ld %10ld %10ld\n", cpi->common.current_video_frame, cpi->time_receive_data / 1000, cpi->time_encode_mb_row / 1000, cpi->time_compress_data / 1000, (cpi->time_receive_data + cpi->time_compress_data) / 1000);
- }
-#endif
-
- }
-
-#if CONFIG_MULTITHREAD
- vp8cx_remove_encoder_threads(cpi);
-#endif
-
- dealloc_compressor_data(cpi);
- vpx_free(cpi->mb.ss);
- vpx_free(cpi->tok);
- vpx_free(cpi->cyclic_refresh_map);
-
- vp8_remove_common(&cpi->common);
- vpx_free(cpi);
- *ptr = 0;
-
-#ifdef OUTPUT_YUV_SRC
- fclose(yuv_file);
-#endif
-
-#if 0
-
- if (keyfile)
- fclose(keyfile);
-
- if (framepsnr)
- fclose(framepsnr);
-
- if (kf_list)
- fclose(kf_list);
-
-#endif
-
-}
-
-
-static uint64_t calc_plane_error(unsigned char *orig, int orig_stride,
- unsigned char *recon, int recon_stride,
- unsigned int cols, unsigned int rows,
- vp8_variance_rtcd_vtable_t *rtcd)
-{
- unsigned int row, col;
- uint64_t total_sse = 0;
- int diff;
-
- for (row = 0; row + 16 <= rows; row += 16)
- {
- for (col = 0; col + 16 <= cols; col += 16)
- {
- unsigned int sse;
-
- VARIANCE_INVOKE(rtcd, mse16x16)(orig + col, orig_stride,
- recon + col, recon_stride,
- &sse);
- total_sse += sse;
- }
-
- /* Handle odd-sized width */
- if (col < cols)
- {
- unsigned int border_row, border_col;
- unsigned char *border_orig = orig;
- unsigned char *border_recon = recon;
-
- for (border_row = 0; border_row < 16; border_row++)
- {
- for (border_col = col; border_col < cols; border_col++)
- {
- diff = border_orig[border_col] - border_recon[border_col];
- total_sse += diff * diff;
- }
-
- border_orig += orig_stride;
- border_recon += recon_stride;
- }
- }
-
- orig += orig_stride * 16;
- recon += recon_stride * 16;
- }
-
- /* Handle odd-sized height */
- for (; row < rows; row++)
- {
- for (col = 0; col < cols; col++)
- {
- diff = orig[col] - recon[col];
- total_sse += diff * diff;
- }
-
- orig += orig_stride;
- recon += recon_stride;
- }
-
- return total_sse;
-}
-
-
-static void generate_psnr_packet(VP8_COMP *cpi)
-{
- YV12_BUFFER_CONFIG *orig = cpi->Source;
- YV12_BUFFER_CONFIG *recon = cpi->common.frame_to_show;
- struct vpx_codec_cx_pkt pkt;
- uint64_t sse;
- int i;
- unsigned int width = cpi->common.Width;
- unsigned int height = cpi->common.Height;
-
- pkt.kind = VPX_CODEC_PSNR_PKT;
- sse = calc_plane_error(orig->y_buffer, orig->y_stride,
- recon->y_buffer, recon->y_stride,
- width, height,
- IF_RTCD(&cpi->rtcd.variance));
- pkt.data.psnr.sse[0] = sse;
- pkt.data.psnr.sse[1] = sse;
- pkt.data.psnr.samples[0] = width * height;
- pkt.data.psnr.samples[1] = width * height;
-
- width = (width + 1) / 2;
- height = (height + 1) / 2;
-
- sse = calc_plane_error(orig->u_buffer, orig->uv_stride,
- recon->u_buffer, recon->uv_stride,
- width, height,
- IF_RTCD(&cpi->rtcd.variance));
- pkt.data.psnr.sse[0] += sse;
- pkt.data.psnr.sse[2] = sse;
- pkt.data.psnr.samples[0] += width * height;
- pkt.data.psnr.samples[2] = width * height;
-
- sse = calc_plane_error(orig->v_buffer, orig->uv_stride,
- recon->v_buffer, recon->uv_stride,
- width, height,
- IF_RTCD(&cpi->rtcd.variance));
- pkt.data.psnr.sse[0] += sse;
- pkt.data.psnr.sse[3] = sse;
- pkt.data.psnr.samples[0] += width * height;
- pkt.data.psnr.samples[3] = width * height;
-
- for (i = 0; i < 4; i++)
- pkt.data.psnr.psnr[i] = vp8_mse2psnr(pkt.data.psnr.samples[i], 255.0,
- pkt.data.psnr.sse[i]);
-
- vpx_codec_pkt_list_add(cpi->output_pkt_list, &pkt);
-}
-
-
-int vp8_use_as_reference(VP8_PTR ptr, int ref_frame_flags)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
- if (ref_frame_flags > 7)
- return -1 ;
-
- cpi->ref_frame_flags = ref_frame_flags;
- return 0;
-}
-int vp8_update_reference(VP8_PTR ptr, int ref_frame_flags)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
- if (ref_frame_flags > 7)
- return -1 ;
-
- cpi->common.refresh_golden_frame = 0;
- cpi->common.refresh_alt_ref_frame = 0;
- cpi->common.refresh_last_frame = 0;
-
- if (ref_frame_flags & VP8_LAST_FLAG)
- cpi->common.refresh_last_frame = 1;
-
- if (ref_frame_flags & VP8_GOLD_FLAG)
- cpi->common.refresh_golden_frame = 1;
-
- if (ref_frame_flags & VP8_ALT_FLAG)
- cpi->common.refresh_alt_ref_frame = 1;
-
- return 0;
-}
-
-int vp8_get_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
- VP8_COMMON *cm = &cpi->common;
- int ref_fb_idx;
-
- if (ref_frame_flag == VP8_LAST_FLAG)
- ref_fb_idx = cm->lst_fb_idx;
- else if (ref_frame_flag == VP8_GOLD_FLAG)
- ref_fb_idx = cm->gld_fb_idx;
- else if (ref_frame_flag == VP8_ALT_FLAG)
- ref_fb_idx = cm->alt_fb_idx;
- else
- return -1;
-
- vp8_yv12_copy_frame_ptr(&cm->yv12_fb[ref_fb_idx], sd);
-
- return 0;
-}
-int vp8_set_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
-{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
- VP8_COMMON *cm = &cpi->common;
-
- int ref_fb_idx;
-
- if (ref_frame_flag == VP8_LAST_FLAG)
- ref_fb_idx = cm->lst_fb_idx;
- else if (ref_frame_flag == VP8_GOLD_FLAG)
- ref_fb_idx = cm->gld_fb_idx;
- else if (ref_frame_flag == VP8_ALT_FLAG)
- ref_fb_idx = cm->alt_fb_idx;
- else
- return -1;
-
- vp8_yv12_copy_frame_ptr(sd, &cm->yv12_fb[ref_fb_idx]);
-
- return 0;
-}
-int vp8_update_entropy(VP8_PTR comp, int update)
-{
- VP8_COMP *cpi = (VP8_COMP *) comp;
- VP8_COMMON *cm = &cpi->common;
- cm->refresh_entropy_probs = update;
-
- return 0;
-}
-
-
-#if OUTPUT_YUV_SRC
-void vp8_write_yuv_frame(const char *name, YV12_BUFFER_CONFIG *s)
-{
- FILE *yuv_file = fopen(name, "ab");
- unsigned char *src = s->y_buffer;
- int h = s->y_height;
-
- do
- {
- fwrite(src, s->y_width, 1, yuv_file);
- src += s->y_stride;
- }
- while (--h);
-
- src = s->u_buffer;
- h = s->uv_height;
-
- do
- {
- fwrite(src, s->uv_width, 1, yuv_file);
- src += s->uv_stride;
- }
- while (--h);
-
- src = s->v_buffer;
- h = s->uv_height;
-
- do
- {
- fwrite(src, s->uv_width, 1, yuv_file);
- src += s->uv_stride;
- }
- while (--h);
-
- fclose(yuv_file);
-}
-#endif
-
-
-static void scale_and_extend_source(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
-
- // are we resizing the image
- if (cm->horiz_scale != 0 || cm->vert_scale != 0)
- {
-#if CONFIG_SPATIAL_RESAMPLING
- int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
- int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
- int tmp_height;
-
- if (cm->vert_scale == 3)
- tmp_height = 9;
- else
- tmp_height = 11;
-
- Scale2Ratio(cm->horiz_scale, &hr, &hs);
- Scale2Ratio(cm->vert_scale, &vr, &vs);
-
- vp8_scale_frame(sd, &cpi->scaled_source, cm->temp_scale_frame.y_buffer,
- tmp_height, hs, hr, vs, vr, 0);
-
- cpi->Source = &cpi->scaled_source;
-#endif
- }
- // we may need to copy to a buffer so we can extend the image...
- else if (cm->Width != cm->yv12_fb[cm->lst_fb_idx].y_width ||
- cm->Height != cm->yv12_fb[cm->lst_fb_idx].y_height)
- {
- //vp8_yv12_copy_frame_ptr(sd, &cpi->scaled_source);
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_yv12_copy_src_frame_func_neon(sd, &cpi->scaled_source);
- }
-#if CONFIG_RUNTIME_CPU_DETECT
- else
-#endif
-#endif
-#if !HAVE_ARMV7 || CONFIG_RUNTIME_CPU_DETECT
- {
- vp8_yv12_copy_frame_ptr(sd, &cpi->scaled_source);
- }
-#endif
-
- cpi->Source = &cpi->scaled_source;
- }
-
- vp8_extend_to_multiple_of16(cpi->Source, cm->Width, cm->Height);
-
-}
-static void resize_key_frame(VP8_COMP *cpi)
-{
-#if CONFIG_SPATIAL_RESAMPLING
- VP8_COMMON *cm = &cpi->common;
-
- // Do we need to apply resampling for one pass cbr.
- // In one pass this is more limited than in two pass cbr
- // The test and any change is only made one per key frame sequence
- if (cpi->oxcf.allow_spatial_resampling && (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER))
- {
- int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
- int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
- int new_width, new_height;
-
- // If we are below the resample DOWN watermark then scale down a notch.
- if (cpi->buffer_level < (cpi->oxcf.resample_down_water_mark * cpi->oxcf.optimal_buffer_level / 100))
- {
- cm->horiz_scale = (cm->horiz_scale < ONETWO) ? cm->horiz_scale + 1 : ONETWO;
- cm->vert_scale = (cm->vert_scale < ONETWO) ? cm->vert_scale + 1 : ONETWO;
- }
- // Should we now start scaling back up
- else if (cpi->buffer_level > (cpi->oxcf.resample_up_water_mark * cpi->oxcf.optimal_buffer_level / 100))
- {
- cm->horiz_scale = (cm->horiz_scale > NORMAL) ? cm->horiz_scale - 1 : NORMAL;
- cm->vert_scale = (cm->vert_scale > NORMAL) ? cm->vert_scale - 1 : NORMAL;
- }
-
- // Get the new hieght and width
- Scale2Ratio(cm->horiz_scale, &hr, &hs);
- Scale2Ratio(cm->vert_scale, &vr, &vs);
- new_width = ((hs - 1) + (cpi->oxcf.Width * hr)) / hs;
- new_height = ((vs - 1) + (cpi->oxcf.Height * vr)) / vs;
-
- // If the image size has changed we need to reallocate the buffers
- // and resample the source image
- if ((cm->Width != new_width) || (cm->Height != new_height))
- {
- cm->Width = new_width;
- cm->Height = new_height;
- vp8_alloc_compressor_data(cpi);
- scale_and_extend_source(cpi->un_scaled_source, cpi);
- }
- }
-
-#endif
-}
-// return of 0 means drop frame
-static int pick_frame_size(VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
-
- // First Frame is a special case
- if (cm->current_video_frame == 0)
- {
-#if !(CONFIG_REALTIME_ONLY)
-
- if (cpi->pass == 2)
- vp8_calc_auto_iframe_target_size(cpi);
-
- // 1 Pass there is no information on which to base size so use bandwidth per second * fixed fraction
- else
-#endif
- cpi->this_frame_target = cpi->oxcf.target_bandwidth / 2;
-
- // in error resilient mode the first frame is bigger since it likely contains
- // all the static background
- if (cpi->oxcf.error_resilient_mode == 1 || (cpi->compressor_speed == 2))
- {
- cpi->this_frame_target *= 3; // 5;
- }
-
- // Key frame from VFW/auto-keyframe/first frame
- cm->frame_type = KEY_FRAME;
-
- }
- // Special case for forced key frames
- // The frame sizing here is still far from ideal for 2 pass.
- else if (cm->frame_flags & FRAMEFLAGS_KEY)
- {
- cm->frame_type = KEY_FRAME;
- resize_key_frame(cpi);
- vp8_calc_iframe_target_size(cpi);
- }
- else if (cm->frame_type == KEY_FRAME)
- {
- vp8_calc_auto_iframe_target_size(cpi);
- }
- else
- {
- // INTER frame: compute target frame size
- cm->frame_type = INTER_FRAME;
- vp8_calc_pframe_target_size(cpi);
-
- // Check if we're dropping the frame:
- if (cpi->drop_frame)
- {
- cpi->drop_frame = FALSE;
- cpi->drop_count++;
- return 0;
- }
- }
-
- return 1;
-}
-
-static void set_quantizer(VP8_COMP *cpi, int Q)
-{
- VP8_COMMON *cm = &cpi->common;
- MACROBLOCKD *mbd = &cpi->mb.e_mbd;
- int update = 0;
- int new_delta_q;
- cm->base_qindex = Q;
-
- /* if any of the delta_q values are changing update flag has to be set */
- /* currently only y2dc_delta_q may change */
-
- cm->y1dc_delta_q = 0;
- cm->y2ac_delta_q = 0;
- cm->uvdc_delta_q = 0;
- cm->uvac_delta_q = 0;
-
- if (Q < 4)
- {
- new_delta_q = 4-Q;
- }
- else
- new_delta_q = 0;
-
- update |= cm->y2dc_delta_q != new_delta_q;
- cm->y2dc_delta_q = new_delta_q;
-
-
- // Set Segment specific quatizers
- mbd->segment_feature_data[MB_LVL_ALT_Q][0] = cpi->segment_feature_data[MB_LVL_ALT_Q][0];
- mbd->segment_feature_data[MB_LVL_ALT_Q][1] = cpi->segment_feature_data[MB_LVL_ALT_Q][1];
- mbd->segment_feature_data[MB_LVL_ALT_Q][2] = cpi->segment_feature_data[MB_LVL_ALT_Q][2];
- mbd->segment_feature_data[MB_LVL_ALT_Q][3] = cpi->segment_feature_data[MB_LVL_ALT_Q][3];
-
- /* quantizer has to be reinitialized for any delta_q changes */
- if(update)
- vp8cx_init_quantizer(cpi);
-
-}
-
-static void update_alt_ref_frame_and_stats(VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
-
- // Update the golden frame buffer
- vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->yv12_fb[cm->alt_fb_idx]);
-
- // Select an interval before next GF or altref
- if (!cpi->auto_gold)
- cpi->frames_till_gf_update_due = cpi->goldfreq;
-
- if ((cpi->pass != 2) && cpi->frames_till_gf_update_due)
- {
- cpi->current_gf_interval = cpi->frames_till_gf_update_due;
-
- // Set the bits per frame that we should try and recover in subsequent inter frames
- // to account for the extra GF spend... note that his does not apply for GF updates
- // that occur coincident with a key frame as the extra cost of key frames is dealt
- // with elsewhere.
-
- cpi->gf_overspend_bits += cpi->projected_frame_size;
- cpi->non_gf_bitrate_adjustment = cpi->gf_overspend_bits / cpi->frames_till_gf_update_due;
- }
-
- // Update data structure that monitors level of reference to last GF
- vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols));
- cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
-
- // this frame refreshes means next frames don't unless specified by user
- cpi->common.frames_since_golden = 0;
-
- // Clear the alternate reference update pending flag.
- cpi->source_alt_ref_pending = FALSE;
-
- // Set the alternate refernce frame active flag
- cpi->source_alt_ref_active = TRUE;
-
-
-}
-static void update_golden_frame_and_stats(VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
-
- // Update the Golden frame reconstruction buffer if signalled and the GF usage counts.
- if (cm->refresh_golden_frame)
- {
- if (cm->frame_type != KEY_FRAME)
- {
- // Update the golden frame buffer
- vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->yv12_fb[cm->gld_fb_idx]);
- }
-
- // Select an interval before next GF
- if (!cpi->auto_gold)
- cpi->frames_till_gf_update_due = cpi->goldfreq;
-
- if ((cpi->pass != 2) && (cpi->frames_till_gf_update_due > 0))
- {
- cpi->current_gf_interval = cpi->frames_till_gf_update_due;
-
- // Set the bits per frame that we should try and recover in subsequent inter frames
- // to account for the extra GF spend... note that his does not apply for GF updates
- // that occur coincident with a key frame as the extra cost of key frames is dealt
- // with elsewhere.
- if ((cm->frame_type != KEY_FRAME) && !cpi->source_alt_ref_active)
- {
- // Calcluate GF bits to be recovered
- // Projected size - av frame bits available for inter frames for clip as a whole
- cpi->gf_overspend_bits += (cpi->projected_frame_size - cpi->inter_frame_target);
- }
-
- cpi->non_gf_bitrate_adjustment = cpi->gf_overspend_bits / cpi->frames_till_gf_update_due;
-
- }
-
- // Update data structure that monitors level of reference to last GF
- vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols));
- cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
-
- // this frame refreshes means next frames don't unless specified by user
- cm->refresh_golden_frame = 0;
- cpi->common.frames_since_golden = 0;
-
- //if ( cm->frame_type == KEY_FRAME )
- //{
- cpi->recent_ref_frame_usage[INTRA_FRAME] = 1;
- cpi->recent_ref_frame_usage[LAST_FRAME] = 1;
- cpi->recent_ref_frame_usage[GOLDEN_FRAME] = 1;
- cpi->recent_ref_frame_usage[ALTREF_FRAME] = 1;
- //}
- //else
- //{
- // // Carry a potrtion of count over to begining of next gf sequence
- // cpi->recent_ref_frame_usage[INTRA_FRAME] >>= 5;
- // cpi->recent_ref_frame_usage[LAST_FRAME] >>= 5;
- // cpi->recent_ref_frame_usage[GOLDEN_FRAME] >>= 5;
- // cpi->recent_ref_frame_usage[ALTREF_FRAME] >>= 5;
- //}
-
- // ******** Fixed Q test code only ************
- // If we are going to use the ALT reference for the next group of frames set a flag to say so.
- if (cpi->oxcf.fixed_q >= 0 &&
- cpi->oxcf.play_alternate && !cpi->common.refresh_alt_ref_frame)
- {
- cpi->source_alt_ref_pending = TRUE;
- cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
- }
-
- if (!cpi->source_alt_ref_pending)
- cpi->source_alt_ref_active = FALSE;
-
- // Decrement count down till next gf
- if (cpi->frames_till_gf_update_due > 0)
- cpi->frames_till_gf_update_due--;
-
- }
- else if (!cpi->common.refresh_alt_ref_frame)
- {
- // Decrement count down till next gf
- if (cpi->frames_till_gf_update_due > 0)
- cpi->frames_till_gf_update_due--;
-
- if (cpi->common.frames_till_alt_ref_frame)
- cpi->common.frames_till_alt_ref_frame --;
-
- cpi->common.frames_since_golden ++;
-
- if (cpi->common.frames_since_golden > 1)
- {
- cpi->recent_ref_frame_usage[INTRA_FRAME] += cpi->count_mb_ref_frame_usage[INTRA_FRAME];
- cpi->recent_ref_frame_usage[LAST_FRAME] += cpi->count_mb_ref_frame_usage[LAST_FRAME];
- cpi->recent_ref_frame_usage[GOLDEN_FRAME] += cpi->count_mb_ref_frame_usage[GOLDEN_FRAME];
- cpi->recent_ref_frame_usage[ALTREF_FRAME] += cpi->count_mb_ref_frame_usage[ALTREF_FRAME];
- }
- }
-}
-
-// This function updates the reference frame probability estimates that
-// will be used during mode selection
-static void update_rd_ref_frame_probs(VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
-
-#if 0
- const int *const rfct = cpi->recent_ref_frame_usage;
- const int rf_intra = rfct[INTRA_FRAME];
- const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
-
- if (cm->frame_type == KEY_FRAME)
- {
- cpi->prob_intra_coded = 255;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- }
- else if (!(rf_intra + rf_inter))
- {
- // This is a trap in case this function is called with cpi->recent_ref_frame_usage[] blank.
- cpi->prob_intra_coded = 63;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- }
- else
- {
- cpi->prob_intra_coded = (rf_intra * 255) / (rf_intra + rf_inter);
-
- if (cpi->prob_intra_coded < 1)
- cpi->prob_intra_coded = 1;
-
- if ((cm->frames_since_golden > 0) || cpi->source_alt_ref_active)
- {
- cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- if (cpi->prob_last_coded < 1)
- cpi->prob_last_coded = 1;
-
- cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- if (cpi->prob_gf_coded < 1)
- cpi->prob_gf_coded = 1;
- }
- }
-
-#else
- const int *const rfct = cpi->count_mb_ref_frame_usage;
- const int rf_intra = rfct[INTRA_FRAME];
- const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
-
- if (cm->frame_type == KEY_FRAME)
- {
- cpi->prob_intra_coded = 255;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- }
- else if (!(rf_intra + rf_inter))
- {
- // This is a trap in case this function is called with cpi->recent_ref_frame_usage[] blank.
- cpi->prob_intra_coded = 63;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- }
- else
- {
- cpi->prob_intra_coded = (rf_intra * 255) / (rf_intra + rf_inter);
-
- if (cpi->prob_intra_coded < 1)
- cpi->prob_intra_coded = 1;
-
- cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- if (cpi->prob_last_coded < 1)
- cpi->prob_last_coded = 1;
-
- cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- if (cpi->prob_gf_coded < 1)
- cpi->prob_gf_coded = 1;
- }
-
- // update reference frame costs since we can do better than what we got last frame.
-
- if (cpi->common.refresh_alt_ref_frame)
- {
- cpi->prob_intra_coded += 40;
- cpi->prob_last_coded = 200;
- cpi->prob_gf_coded = 1;
- }
- else if (cpi->common.frames_since_golden == 0)
- {
- cpi->prob_last_coded = 214;
- cpi->prob_gf_coded = 1;
- }
- else if (cpi->common.frames_since_golden == 1)
- {
- cpi->prob_last_coded = 192;
- cpi->prob_gf_coded = 220;
- }
- else if (cpi->source_alt_ref_active)
- {
- //int dist = cpi->common.frames_till_alt_ref_frame + cpi->common.frames_since_golden;
- cpi->prob_gf_coded -= 20;
-
- if (cpi->prob_gf_coded < 10)
- cpi->prob_gf_coded = 10;
- }
-
-#endif
-}
-
-
-// 1 = key, 0 = inter
-static int decide_key_frame(VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
-
- int code_key_frame = FALSE;
-
- cpi->kf_boost = 0;
-
- if (cpi->Speed > 11)
- return FALSE;
-
- // Clear down mmx registers
- vp8_clear_system_state(); //__asm emms;
-
- if ((cpi->compressor_speed == 2) && (cpi->Speed >= 5) && (cpi->sf.RD == 0))
- {
- double change = 1.0 * abs((int)(cpi->intra_error - cpi->last_intra_error)) / (1 + cpi->last_intra_error);
- double change2 = 1.0 * abs((int)(cpi->prediction_error - cpi->last_prediction_error)) / (1 + cpi->last_prediction_error);
- double minerror = cm->MBs * 256;
-
-#if 0
-
- if (10 * cpi->intra_error / (1 + cpi->prediction_error) < 15
- && cpi->prediction_error > minerror
- && (change > .25 || change2 > .25))
- {
- FILE *f = fopen("intra_inter.stt", "a");
-
- if (cpi->prediction_error <= 0)
- cpi->prediction_error = 1;
-
- fprintf(f, "%d %d %d %d %14.4f\n",
- cm->current_video_frame,
- (int) cpi->prediction_error,
- (int) cpi->intra_error,
- (int)((10 * cpi->intra_error) / cpi->prediction_error),
- change);
-
- fclose(f);
- }
-
-#endif
-
- cpi->last_intra_error = cpi->intra_error;
- cpi->last_prediction_error = cpi->prediction_error;
-
- if (10 * cpi->intra_error / (1 + cpi->prediction_error) < 15
- && cpi->prediction_error > minerror
- && (change > .25 || change2 > .25))
- {
- /*(change > 1.4 || change < .75)&& cpi->this_frame_percent_intra > cpi->last_frame_percent_intra + 3*/
- return TRUE;
- }
-
- return FALSE;
-
- }
-
- // If the following are true we might as well code a key frame
- if (((cpi->this_frame_percent_intra == 100) &&
- (cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra + 2))) ||
- ((cpi->this_frame_percent_intra > 95) &&
- (cpi->this_frame_percent_intra >= (cpi->last_frame_percent_intra + 5))))
- {
- code_key_frame = TRUE;
- }
- // in addition if the following are true and this is not a golden frame then code a key frame
- // Note that on golden frames there often seems to be a pop in intra useage anyway hence this
- // restriction is designed to prevent spurious key frames. The Intra pop needs to be investigated.
- else if (((cpi->this_frame_percent_intra > 60) &&
- (cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra * 2))) ||
- ((cpi->this_frame_percent_intra > 75) &&
- (cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra * 3 / 2))) ||
- ((cpi->this_frame_percent_intra > 90) &&
- (cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra + 10))))
- {
- if (!cm->refresh_golden_frame)
- code_key_frame = TRUE;
- }
-
- return code_key_frame;
-
-}
-
-#if !(CONFIG_REALTIME_ONLY)
-static void Pass1Encode(VP8_COMP *cpi, unsigned long *size, unsigned char *dest, unsigned int *frame_flags)
-{
- (void) size;
- (void) dest;
- (void) frame_flags;
- set_quantizer(cpi, 26);
-
- scale_and_extend_source(cpi->un_scaled_source, cpi);
- vp8_first_pass(cpi);
-}
-#endif
-
-#if 0
-void write_cx_frame_to_file(YV12_BUFFER_CONFIG *frame, int this_frame)
-{
-
- // write the frame
- FILE *yframe;
- int i;
- char filename[255];
-
- sprintf(filename, "cx\\y%04d.raw", this_frame);
- yframe = fopen(filename, "wb");
-
- for (i = 0; i < frame->y_height; i++)
- fwrite(frame->y_buffer + i * frame->y_stride, frame->y_width, 1, yframe);
-
- fclose(yframe);
- sprintf(filename, "cx\\u%04d.raw", this_frame);
- yframe = fopen(filename, "wb");
-
- for (i = 0; i < frame->uv_height; i++)
- fwrite(frame->u_buffer + i * frame->uv_stride, frame->uv_width, 1, yframe);
-
- fclose(yframe);
- sprintf(filename, "cx\\v%04d.raw", this_frame);
- yframe = fopen(filename, "wb");
-
- for (i = 0; i < frame->uv_height; i++)
- fwrite(frame->v_buffer + i * frame->uv_stride, frame->uv_width, 1, yframe);
-
- fclose(yframe);
-}
-#endif
-// return of 0 means drop frame
-
-// Function to test for conditions that indeicate we should loop
-// back and recode a frame.
-static BOOL recode_loop_test( VP8_COMP *cpi,
- int high_limit, int low_limit,
- int q, int maxq, int minq )
-{
- BOOL force_recode = FALSE;
- VP8_COMMON *cm = &cpi->common;
-
- // Is frame recode allowed at all
- // Yes if either recode mode 1 is selected or mode two is selcted
- // and the frame is a key frame. golden frame or alt_ref_frame
- if ( (cpi->sf.recode_loop == 1) ||
- ( (cpi->sf.recode_loop == 2) &&
- ( (cm->frame_type == KEY_FRAME) ||
- cm->refresh_golden_frame ||
- cm->refresh_alt_ref_frame ) ) )
- {
- // General over and under shoot tests
- if ( ((cpi->projected_frame_size > high_limit) && (q < maxq)) ||
- ((cpi->projected_frame_size < low_limit) && (q > minq)) )
- {
- force_recode = TRUE;
- }
- // Special Constrained quality tests
- else if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
- {
- // Undershoot and below auto cq level
- if ( (q > cpi->cq_target_quality) &&
- (cpi->projected_frame_size <
- ((cpi->this_frame_target * 7) >> 3)))
- {
- force_recode = TRUE;
- }
- // Severe undershoot and between auto and user cq level
- else if ( (q > cpi->oxcf.cq_level) &&
- (cpi->projected_frame_size < cpi->min_frame_bandwidth) &&
- (cpi->active_best_quality > cpi->oxcf.cq_level))
- {
- force_recode = TRUE;
- cpi->active_best_quality = cpi->oxcf.cq_level;
- }
- }
- }
-
- return force_recode;
-}
-
-void loopfilter_frame(VP8_COMP *cpi, VP8_COMMON *cm)
-{
- if (cm->no_lpf)
- {
- cm->filter_level = 0;
- }
- else
- {
- struct vpx_usec_timer timer;
-
- vp8_clear_system_state();
-
- vpx_usec_timer_start(&timer);
- if (cpi->sf.auto_filter == 0)
- vp8cx_pick_filter_level_fast(cpi->Source, cpi);
-
- else
- vp8cx_pick_filter_level(cpi->Source, cpi);
-
- vpx_usec_timer_mark(&timer);
- cpi->time_pick_lpf += vpx_usec_timer_elapsed(&timer);
- }
-
-#if CONFIG_MULTITHREAD
- if (cpi->b_multi_threaded)
- sem_post(&cpi->h_event_end_lpf); /* signal that we have set filter_level */
-#endif
-
- if (cm->filter_level > 0)
- {
- vp8cx_set_alt_lf_level(cpi, cm->filter_level);
- vp8_loop_filter_frame(cm, &cpi->mb.e_mbd, cm->filter_level);
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
- }
-
- vp8_yv12_extend_frame_borders_ptr(cm->frame_to_show);
-
- {
- YV12_BUFFER_CONFIG *lst_yv12 = &cm->yv12_fb[cm->lst_fb_idx];
- YV12_BUFFER_CONFIG *new_yv12 = &cm->yv12_fb[cm->new_fb_idx];
- YV12_BUFFER_CONFIG *gld_yv12 = &cm->yv12_fb[cm->gld_fb_idx];
- YV12_BUFFER_CONFIG *alt_yv12 = &cm->yv12_fb[cm->alt_fb_idx];
- // At this point the new frame has been encoded.
- // If any buffer copy / swapping is signaled it should be done here.
- if (cm->frame_type == KEY_FRAME)
- {
- vp8_yv12_copy_frame_ptr(cm->frame_to_show, gld_yv12);
- vp8_yv12_copy_frame_ptr(cm->frame_to_show, alt_yv12);
- }
- else // For non key frames
- {
- // Code to copy between reference buffers
- if (cm->copy_buffer_to_arf)
- {
- if (cm->copy_buffer_to_arf == 1)
- {
- if (cm->refresh_last_frame)
- // We copy new_frame here because last and new buffers will already have been swapped if cm->refresh_last_frame is set.
- vp8_yv12_copy_frame_ptr(new_yv12, alt_yv12);
- else
- vp8_yv12_copy_frame_ptr(lst_yv12, alt_yv12);
- }
- else if (cm->copy_buffer_to_arf == 2)
- vp8_yv12_copy_frame_ptr(gld_yv12, alt_yv12);
- }
-
- if (cm->copy_buffer_to_gf)
- {
- if (cm->copy_buffer_to_gf == 1)
- {
- if (cm->refresh_last_frame)
- // We copy new_frame here because last and new buffers will already have been swapped if cm->refresh_last_frame is set.
- vp8_yv12_copy_frame_ptr(new_yv12, gld_yv12);
- else
- vp8_yv12_copy_frame_ptr(lst_yv12, gld_yv12);
- }
- else if (cm->copy_buffer_to_gf == 2)
- vp8_yv12_copy_frame_ptr(alt_yv12, gld_yv12);
- }
- }
- }
-}
-
-static void encode_frame_to_data_rate
-(
- VP8_COMP *cpi,
- unsigned long *size,
- unsigned char *dest,
- unsigned int *frame_flags
-)
-{
- int Q;
- int frame_over_shoot_limit;
- int frame_under_shoot_limit;
-
- int Loop = FALSE;
- int loop_count;
- int this_q;
- int last_zbin_oq;
-
- int q_low;
- int q_high;
- int zbin_oq_high;
- int zbin_oq_low = 0;
- int top_index;
- int bottom_index;
- VP8_COMMON *cm = &cpi->common;
- int active_worst_qchanged = FALSE;
-
- int overshoot_seen = FALSE;
- int undershoot_seen = FALSE;
- int drop_mark = cpi->oxcf.drop_frames_water_mark * cpi->oxcf.optimal_buffer_level / 100;
- int drop_mark75 = drop_mark * 2 / 3;
- int drop_mark50 = drop_mark / 4;
- int drop_mark25 = drop_mark / 8;
-
-
- // Clear down mmx registers to allow floating point in what follows
- vp8_clear_system_state();
-
- // Test code for segmentation of gf/arf (0,0)
- //segmentation_test_function((VP8_PTR) cpi);
-
-#if CONFIG_REALTIME_ONLY
- if(cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME)
- {
- if(cpi->force_next_frame_intra)
- {
- cm->frame_type = KEY_FRAME; /* delayed intra frame */
- }
- }
- cpi->force_next_frame_intra = 0;
-#endif
-
- // For an alt ref frame in 2 pass we skip the call to the second pass function that sets the target bandwidth
-#if !(CONFIG_REALTIME_ONLY)
-
- if (cpi->pass == 2)
- {
- if (cpi->common.refresh_alt_ref_frame)
- {
- cpi->per_frame_bandwidth = cpi->gf_bits; // Per frame bit target for the alt ref frame
- cpi->target_bandwidth = cpi->gf_bits * cpi->output_frame_rate; // per second target bitrate
- }
- }
- else
-#endif
- cpi->per_frame_bandwidth = (int)(cpi->target_bandwidth / cpi->output_frame_rate);
-
- // Default turn off buffer to buffer copying
- cm->copy_buffer_to_gf = 0;
- cm->copy_buffer_to_arf = 0;
-
- // Clear zbin over-quant value and mode boost values.
- cpi->zbin_over_quant = 0;
- cpi->zbin_mode_boost = 0;
-
- // Enable or disable mode based tweaking of the zbin
- // For 2 Pass Only used where GF/ARF prediction quality
- // is above a threshold
- cpi->zbin_mode_boost = 0;
- cpi->zbin_mode_boost_enabled = TRUE;
- if (cpi->pass == 2)
- {
- if ( cpi->gfu_boost <= 400 )
- {
- cpi->zbin_mode_boost_enabled = FALSE;
- }
- }
-
- // Current default encoder behaviour for the altref sign bias
- if (cpi->source_alt_ref_active)
- cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 1;
- else
- cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 0;
-
- // Check to see if a key frame is signalled
- // For two pass with auto key frame enabled cm->frame_type may already be set, but not for one pass.
- if ((cm->current_video_frame == 0) ||
- (cm->frame_flags & FRAMEFLAGS_KEY) ||
- (cpi->oxcf.auto_key && (cpi->frames_since_key % cpi->key_frame_frequency == 0)))
- {
- // Key frame from VFW/auto-keyframe/first frame
- cm->frame_type = KEY_FRAME;
- }
-
- // Set default state for segment and mode based loop filter update flags
- cpi->mb.e_mbd.update_mb_segmentation_map = 0;
- cpi->mb.e_mbd.update_mb_segmentation_data = 0;
- cpi->mb.e_mbd.mode_ref_lf_delta_update = 0;
-
- // Set various flags etc to special state if it is a key frame
- if (cm->frame_type == KEY_FRAME)
- {
- int i;
-
- // Reset the loop filter deltas and segmentation map
- setup_features(cpi);
-
- // If segmentation is enabled force a map update for key frames
- if (cpi->mb.e_mbd.segmentation_enabled)
- {
- cpi->mb.e_mbd.update_mb_segmentation_map = 1;
- cpi->mb.e_mbd.update_mb_segmentation_data = 1;
- }
-
- // The alternate reference frame cannot be active for a key frame
- cpi->source_alt_ref_active = FALSE;
-
- // Reset the RD threshold multipliers to default of * 1 (128)
- for (i = 0; i < MAX_MODES; i++)
- {
- cpi->rd_thresh_mult[i] = 128;
- }
- }
-
- // Test code for segmentation
- //if ( (cm->frame_type == KEY_FRAME) || ((cm->current_video_frame % 2) == 0))
- //if ( (cm->current_video_frame % 2) == 0 )
- // enable_segmentation((VP8_PTR)cpi);
- //else
- // disable_segmentation((VP8_PTR)cpi);
-
-#if 0
- // Experimental code for lagged compress and one pass
- // Initialise one_pass GF frames stats
- // Update stats used for GF selection
- //if ( cpi->pass == 0 )
- {
- cpi->one_pass_frame_index = cm->current_video_frame % MAX_LAG_BUFFERS;
-
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frames_so_far = 0;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_intra_error = 0.0;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_coded_error = 0.0;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_pcnt_inter = 0.0;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_pcnt_motion = 0.0;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvr = 0.0;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvr_abs = 0.0;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvc = 0.0;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvc_abs = 0.0;
- }
-#endif
-
- update_rd_ref_frame_probs(cpi);
-
- if (cpi->drop_frames_allowed)
- {
- // The reset to decimation 0 is only done here for one pass.
- // Once it is set two pass leaves decimation on till the next kf.
- if ((cpi->buffer_level > drop_mark) && (cpi->decimation_factor > 0))
- cpi->decimation_factor --;
-
- if (cpi->buffer_level > drop_mark75 && cpi->decimation_factor > 0)
- cpi->decimation_factor = 1;
-
- else if (cpi->buffer_level < drop_mark25 && (cpi->decimation_factor == 2 || cpi->decimation_factor == 3))
- {
- cpi->decimation_factor = 3;
- }
- else if (cpi->buffer_level < drop_mark50 && (cpi->decimation_factor == 1 || cpi->decimation_factor == 2))
- {
- cpi->decimation_factor = 2;
- }
- else if (cpi->buffer_level < drop_mark75 && (cpi->decimation_factor == 0 || cpi->decimation_factor == 1))
- {
- cpi->decimation_factor = 1;
- }
-
- //vpx_log("Encoder: Decimation Factor: %d \n",cpi->decimation_factor);
- }
-
- // The following decimates the frame rate according to a regular pattern (i.e. to 1/2 or 2/3 frame rate)
- // This can be used to help prevent buffer under-run in CBR mode. Alternatively it might be desirable in
- // some situations to drop frame rate but throw more bits at each frame.
- //
- // Note that dropping a key frame can be problematic if spatial resampling is also active
- if (cpi->decimation_factor > 0)
- {
- switch (cpi->decimation_factor)
- {
- case 1:
- cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 3 / 2;
- break;
- case 2:
- cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 5 / 4;
- break;
- case 3:
- cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 5 / 4;
- break;
- }
-
- // Note that we should not throw out a key frame (especially when spatial resampling is enabled).
- if ((cm->frame_type == KEY_FRAME)) // && cpi->oxcf.allow_spatial_resampling )
- {
- cpi->decimation_count = cpi->decimation_factor;
- }
- else if (cpi->decimation_count > 0)
- {
- cpi->decimation_count --;
- cpi->bits_off_target += cpi->av_per_frame_bandwidth;
- cm->current_video_frame++;
- cpi->frames_since_key++;
-
-#if CONFIG_PSNR
- cpi->count ++;
-#endif
-
- cpi->buffer_level = cpi->bits_off_target;
-
- return;
- }
- else
- cpi->decimation_count = cpi->decimation_factor;
- }
-
- // Decide how big to make the frame
- if (!pick_frame_size(cpi))
- {
- cm->current_video_frame++;
- cpi->frames_since_key++;
- return;
- }
-
- // Reduce active_worst_allowed_q for CBR if our buffer is getting too full.
- // This has a knock on effect on active best quality as well.
- // For CBR if the buffer reaches its maximum level then we can no longer
- // save up bits for later frames so we might as well use them up
- // on the current frame.
- if ((cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
- (cpi->buffer_level >= cpi->oxcf.optimal_buffer_level) && cpi->buffered_mode)
- {
- int Adjustment = cpi->active_worst_quality / 4; // Max adjustment is 1/4
-
- if (Adjustment)
- {
- int buff_lvl_step;
- int tmp_lvl = cpi->buffer_level;
-
- if (cpi->buffer_level < cpi->oxcf.maximum_buffer_size)
- {
- buff_lvl_step = (cpi->oxcf.maximum_buffer_size - cpi->oxcf.optimal_buffer_level) / Adjustment;
-
- if (buff_lvl_step)
- {
- Adjustment = (cpi->buffer_level - cpi->oxcf.optimal_buffer_level) / buff_lvl_step;
- cpi->active_worst_quality -= Adjustment;
- }
- }
- else
- {
- cpi->active_worst_quality -= Adjustment;
- }
- }
- }
-
- // Set an active best quality and if necessary active worst quality
- // There is some odd behaviour for one pass here that needs attention.
- if ( (cpi->pass == 2) || (cpi->ni_frames > 150))
- {
- vp8_clear_system_state();
-
- Q = cpi->active_worst_quality;
-
- if ( cm->frame_type == KEY_FRAME )
- {
- if ( cpi->pass == 2 )
- {
- if (cpi->gfu_boost > 600)
- cpi->active_best_quality = kf_low_motion_minq[Q];
- else
- cpi->active_best_quality = kf_high_motion_minq[Q];
-
- // Special case for key frames forced because we have reached
- // the maximum key frame interval. Here force the Q to a range
- // based on the ambient Q to reduce the risk of popping
- if ( cpi->this_key_frame_forced )
- {
- if ( cpi->active_best_quality > cpi->avg_frame_qindex * 7/8)
- cpi->active_best_quality = cpi->avg_frame_qindex * 7/8;
- else if ( cpi->active_best_quality < cpi->avg_frame_qindex >> 2 )
- cpi->active_best_quality = cpi->avg_frame_qindex >> 2;
- }
- }
- // One pass more conservative
- else
- cpi->active_best_quality = kf_high_motion_minq[Q];
- }
-
- else if (cm->refresh_golden_frame || cpi->common.refresh_alt_ref_frame)
- {
- // Use the lower of cpi->active_worst_quality and recent
- // average Q as basis for GF/ARF Q limit unless last frame was
- // a key frame.
- if ( (cpi->frames_since_key > 1) &&
- (cpi->avg_frame_qindex < cpi->active_worst_quality) )
- {
- Q = cpi->avg_frame_qindex;
-
- if ( (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
- (Q < cpi->oxcf.cq_level) )
- {
- Q = cpi->oxcf.cq_level;
- }
- }
-
- if ( cpi->pass == 2 )
- {
- if ( cpi->gfu_boost > 1000 )
- cpi->active_best_quality = gf_low_motion_minq[Q];
- else if ( cpi->gfu_boost < 400 )
- cpi->active_best_quality = gf_high_motion_minq[Q];
- else
- cpi->active_best_quality = gf_mid_motion_minq[Q];
- }
- // One pass more conservative
- else
- cpi->active_best_quality = gf_high_motion_minq[Q];
- }
- else
- {
- cpi->active_best_quality = inter_minq[Q];
-
- // For the constant/constrained quality mode we dont want
- // the quality to rise above the cq level.
- if ((cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
- (cpi->active_best_quality < cpi->cq_target_quality) )
- {
- // If we are strongly undershooting the target rate in the last
- // frames then use the user passed in cq value not the auto
- // cq value.
- if ( cpi->rolling_actual_bits < cpi->min_frame_bandwidth )
- cpi->active_best_quality = cpi->oxcf.cq_level;
- else
- cpi->active_best_quality = cpi->cq_target_quality;
- }
- }
-
- // If CBR and the buffer is as full then it is reasonable to allow
- // higher quality on the frames to prevent bits just going to waste.
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- // Note that the use of >= here elliminates the risk of a devide
- // by 0 error in the else if clause
- if (cpi->buffer_level >= cpi->oxcf.maximum_buffer_size)
- cpi->active_best_quality = cpi->best_quality;
-
- else if (cpi->buffer_level > cpi->oxcf.optimal_buffer_level)
- {
- int Fraction = ((cpi->buffer_level - cpi->oxcf.optimal_buffer_level) * 128) / (cpi->oxcf.maximum_buffer_size - cpi->oxcf.optimal_buffer_level);
- int min_qadjustment = ((cpi->active_best_quality - cpi->best_quality) * Fraction) / 128;
-
- cpi->active_best_quality -= min_qadjustment;
- }
- }
- }
- // Make sure constrained quality mode limits are adhered to for the first
- // few frames of one pass encodes
- else if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
- {
- if ( (cm->frame_type == KEY_FRAME) ||
- cm->refresh_golden_frame || cpi->common.refresh_alt_ref_frame )
- {
- cpi->active_best_quality = cpi->best_quality;
- }
- else if (cpi->active_best_quality < cpi->cq_target_quality)
- {
- cpi->active_best_quality = cpi->cq_target_quality;
- }
- }
-
- // Clip the active best and worst quality values to limits
- if (cpi->active_worst_quality > cpi->worst_quality)
- cpi->active_worst_quality = cpi->worst_quality;
-
- if (cpi->active_best_quality < cpi->best_quality)
- cpi->active_best_quality = cpi->best_quality;
- else if (cpi->active_best_quality > cpi->active_worst_quality)
- cpi->active_best_quality = cpi->active_worst_quality;
-
- // Determine initial Q to try
- Q = vp8_regulate_q(cpi, cpi->this_frame_target);
- last_zbin_oq = cpi->zbin_over_quant;
-
- // Set highest allowed value for Zbin over quant
- if (cm->frame_type == KEY_FRAME)
- zbin_oq_high = 0; //ZBIN_OQ_MAX/16
- else if (cm->refresh_alt_ref_frame || (cm->refresh_golden_frame && !cpi->source_alt_ref_active))
- zbin_oq_high = 16;
- else
- zbin_oq_high = ZBIN_OQ_MAX;
-
- // Setup background Q adjustment for error resilliant mode
- if (cpi->cyclic_refresh_mode_enabled)
- cyclic_background_refresh(cpi, Q, 0);
-
- vp8_compute_frame_size_bounds(cpi, &frame_under_shoot_limit, &frame_over_shoot_limit);
-
- // Limit Q range for the adaptive loop.
- bottom_index = cpi->active_best_quality;
- top_index = cpi->active_worst_quality;
- q_low = cpi->active_best_quality;
- q_high = cpi->active_worst_quality;
-
- vp8_save_coding_context(cpi);
-
- loop_count = 0;
-
-
- scale_and_extend_source(cpi->un_scaled_source, cpi);
-#if !(CONFIG_REALTIME_ONLY) && CONFIG_POSTPROC
-
- if (cpi->oxcf.noise_sensitivity > 0)
- {
- unsigned char *src;
- int l = 0;
-
- switch (cpi->oxcf.noise_sensitivity)
- {
- case 1:
- l = 20;
- break;
- case 2:
- l = 40;
- break;
- case 3:
- l = 60;
- break;
- case 4:
- l = 80;
- break;
- case 5:
- l = 100;
- break;
- case 6:
- l = 150;
- break;
- }
-
-
- if (cm->frame_type == KEY_FRAME)
- {
- vp8_de_noise(cpi->Source, cpi->Source, l , 1, 0, RTCD(postproc));
- }
- else
- {
- vp8_de_noise(cpi->Source, cpi->Source, l , 1, 0, RTCD(postproc));
-
- src = cpi->Source->y_buffer;
-
- if (cpi->Source->y_stride < 0)
- {
- src += cpi->Source->y_stride * (cpi->Source->y_height - 1);
- }
- }
- }
-
-#endif
-
-#ifdef OUTPUT_YUV_SRC
- vp8_write_yuv_frame(cpi->Source);
-#endif
-
- do
- {
- vp8_clear_system_state(); //__asm emms;
-
- /*
- if(cpi->is_src_frame_alt_ref)
- Q = 127;
- */
-
- set_quantizer(cpi, Q);
- this_q = Q;
-
- // setup skip prob for costing in mode/mv decision
- if (cpi->common.mb_no_coeff_skip)
- {
- cpi->prob_skip_false = cpi->base_skip_false_prob[Q];
-
- if (cm->frame_type != KEY_FRAME)
- {
- if (cpi->common.refresh_alt_ref_frame)
- {
- if (cpi->last_skip_false_probs[2] != 0)
- cpi->prob_skip_false = cpi->last_skip_false_probs[2];
-
- /*
- if(cpi->last_skip_false_probs[2]!=0 && abs(Q- cpi->last_skip_probs_q[2])<=16 )
- cpi->prob_skip_false = cpi->last_skip_false_probs[2];
- else if (cpi->last_skip_false_probs[2]!=0)
- cpi->prob_skip_false = (cpi->last_skip_false_probs[2] + cpi->prob_skip_false ) / 2;
- */
- }
- else if (cpi->common.refresh_golden_frame)
- {
- if (cpi->last_skip_false_probs[1] != 0)
- cpi->prob_skip_false = cpi->last_skip_false_probs[1];
-
- /*
- if(cpi->last_skip_false_probs[1]!=0 && abs(Q- cpi->last_skip_probs_q[1])<=16 )
- cpi->prob_skip_false = cpi->last_skip_false_probs[1];
- else if (cpi->last_skip_false_probs[1]!=0)
- cpi->prob_skip_false = (cpi->last_skip_false_probs[1] + cpi->prob_skip_false ) / 2;
- */
- }
- else
- {
- if (cpi->last_skip_false_probs[0] != 0)
- cpi->prob_skip_false = cpi->last_skip_false_probs[0];
-
- /*
- if(cpi->last_skip_false_probs[0]!=0 && abs(Q- cpi->last_skip_probs_q[0])<=16 )
- cpi->prob_skip_false = cpi->last_skip_false_probs[0];
- else if(cpi->last_skip_false_probs[0]!=0)
- cpi->prob_skip_false = (cpi->last_skip_false_probs[0] + cpi->prob_skip_false ) / 2;
- */
- }
-
- //as this is for cost estimate, let's make sure it does not go extreme eitehr way
- if (cpi->prob_skip_false < 5)
- cpi->prob_skip_false = 5;
-
- if (cpi->prob_skip_false > 250)
- cpi->prob_skip_false = 250;
-
- if (cpi->is_src_frame_alt_ref)
- cpi->prob_skip_false = 1;
-
-
- }
-
-#if 0
-
- if (cpi->pass != 1)
- {
- FILE *f = fopen("skip.stt", "a");
- fprintf(f, "%d, %d, %4d ", cpi->common.refresh_golden_frame, cpi->common.refresh_alt_ref_frame, cpi->prob_skip_false);
- fclose(f);
- }
-
-#endif
-
- }
-
- if (cm->frame_type == KEY_FRAME)
- vp8_setup_key_frame(cpi);
-
- // transform / motion compensation build reconstruction frame
- vp8_encode_frame(cpi);
-
- cpi->projected_frame_size -= vp8_estimate_entropy_savings(cpi);
- cpi->projected_frame_size = (cpi->projected_frame_size > 0) ? cpi->projected_frame_size : 0;
-
- vp8_clear_system_state(); //__asm emms;
-
- // Test to see if the stats generated for this frame indicate that we should have coded a key frame
- // (assuming that we didn't)!
- if (cpi->pass != 2 && cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME)
- {
-
-#if CONFIG_REALTIME_ONLY
- {
- /* we don't do re-encoding in realtime mode
- * if key frame is decided than we force it on next frame */
- cpi->force_next_frame_intra = decide_key_frame(cpi);
- }
-#else
- if (decide_key_frame(cpi))
- {
- vp8_calc_auto_iframe_target_size(cpi);
-
- // Reset all our sizing numbers and recode
- cm->frame_type = KEY_FRAME;
-
- // Clear the Alt reference frame active flag when we have a key frame
- cpi->source_alt_ref_active = FALSE;
-
- // Reset the loop filter deltas and segmentation map
- setup_features(cpi);
-
- // If segmentation is enabled force a map update for key frames
- if (cpi->mb.e_mbd.segmentation_enabled)
- {
- cpi->mb.e_mbd.update_mb_segmentation_map = 1;
- cpi->mb.e_mbd.update_mb_segmentation_data = 1;
- }
-
- vp8_restore_coding_context(cpi);
-
- Q = vp8_regulate_q(cpi, cpi->this_frame_target);
-
- vp8_compute_frame_size_bounds(cpi, &frame_under_shoot_limit, &frame_over_shoot_limit);
-
- // Limit Q range for the adaptive loop.
- bottom_index = cpi->active_best_quality;
- top_index = cpi->active_worst_quality;
- q_low = cpi->active_best_quality;
- q_high = cpi->active_worst_quality;
-
- loop_count++;
- Loop = TRUE;
-
- resize_key_frame(cpi);
- continue;
- }
-#endif
- }
-
- vp8_clear_system_state();
-
- if (frame_over_shoot_limit == 0)
- frame_over_shoot_limit = 1;
-
- // Are we are overshooting and up against the limit of active max Q.
- if (((cpi->pass != 2) || (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)) &&
- (Q == cpi->active_worst_quality) &&
- (cpi->active_worst_quality < cpi->worst_quality) &&
- (cpi->projected_frame_size > frame_over_shoot_limit))
- {
- int over_size_percent = ((cpi->projected_frame_size - frame_over_shoot_limit) * 100) / frame_over_shoot_limit;
-
- // If so is there any scope for relaxing it
- while ((cpi->active_worst_quality < cpi->worst_quality) && (over_size_percent > 0))
- {
- cpi->active_worst_quality++;
- top_index = cpi->active_worst_quality;
- over_size_percent = (int)(over_size_percent * 0.96); // Assume 1 qstep = about 4% on frame size.
- }
-
- // If we have updated the active max Q do not call vp8_update_rate_correction_factors() this loop.
- active_worst_qchanged = TRUE;
- }
- else
- active_worst_qchanged = FALSE;
-
-#if !(CONFIG_REALTIME_ONLY)
- // Special case handling for forced key frames
- if ( (cm->frame_type == KEY_FRAME) && cpi->this_key_frame_forced )
- {
- int last_q = Q;
- int kf_err = vp8_calc_ss_err(cpi->Source,
- &cm->yv12_fb[cm->new_fb_idx],
- IF_RTCD(&cpi->rtcd.variance));
-
- // The key frame is not good enough
- if ( kf_err > ((cpi->ambient_err * 7) >> 3) )
- {
- // Lower q_high
- q_high = (Q > q_low) ? (Q - 1) : q_low;
-
- // Adjust Q
- Q = (q_high + q_low) >> 1;
- }
- // The key frame is much better than the previous frame
- else if ( kf_err < (cpi->ambient_err >> 1) )
- {
- // Raise q_low
- q_low = (Q < q_high) ? (Q + 1) : q_high;
-
- // Adjust Q
- Q = (q_high + q_low + 1) >> 1;
- }
-
- // Clamp Q to upper and lower limits:
- if (Q > q_high)
- Q = q_high;
- else if (Q < q_low)
- Q = q_low;
-
- Loop = ((Q != last_q)) ? TRUE : FALSE;
- }
-
- // Is the projected frame size out of range and are we allowed to attempt to recode.
- else if ( recode_loop_test( cpi,
- frame_over_shoot_limit, frame_under_shoot_limit,
- Q, top_index, bottom_index ) )
- {
- int last_q = Q;
- int Retries = 0;
-
- // Frame size out of permitted range:
- // Update correction factor & compute new Q to try...
-
- // Frame is too large
- if (cpi->projected_frame_size > cpi->this_frame_target)
- {
- //if ( cpi->zbin_over_quant == 0 )
- q_low = (Q < q_high) ? (Q + 1) : q_high; // Raise Qlow as to at least the current value
-
- if (cpi->zbin_over_quant > 0) // If we are using over quant do the same for zbin_oq_low
- zbin_oq_low = (cpi->zbin_over_quant < zbin_oq_high) ? (cpi->zbin_over_quant + 1) : zbin_oq_high;
-
- //if ( undershoot_seen || (Q == MAXQ) )
- if (undershoot_seen)
- {
- // Update rate_correction_factor unless cpi->active_worst_quality has changed.
- if (!active_worst_qchanged)
- vp8_update_rate_correction_factors(cpi, 1);
-
- Q = (q_high + q_low + 1) / 2;
-
- // Adjust cpi->zbin_over_quant (only allowed when Q is max)
- if (Q < MAXQ)
- cpi->zbin_over_quant = 0;
- else
- {
- zbin_oq_low = (cpi->zbin_over_quant < zbin_oq_high) ? (cpi->zbin_over_quant + 1) : zbin_oq_high;
- cpi->zbin_over_quant = (zbin_oq_high + zbin_oq_low) / 2;
- }
- }
- else
- {
- // Update rate_correction_factor unless cpi->active_worst_quality has changed.
- if (!active_worst_qchanged)
- vp8_update_rate_correction_factors(cpi, 0);
-
- Q = vp8_regulate_q(cpi, cpi->this_frame_target);
-
- while (((Q < q_low) || (cpi->zbin_over_quant < zbin_oq_low)) && (Retries < 10))
- {
- vp8_update_rate_correction_factors(cpi, 0);
- Q = vp8_regulate_q(cpi, cpi->this_frame_target);
- Retries ++;
- }
- }
-
- overshoot_seen = TRUE;
- }
- // Frame is too small
- else
- {
- if (cpi->zbin_over_quant == 0)
- q_high = (Q > q_low) ? (Q - 1) : q_low; // Lower q_high if not using over quant
- else // else lower zbin_oq_high
- zbin_oq_high = (cpi->zbin_over_quant > zbin_oq_low) ? (cpi->zbin_over_quant - 1) : zbin_oq_low;
-
- if (overshoot_seen)
- {
- // Update rate_correction_factor unless cpi->active_worst_quality has changed.
- if (!active_worst_qchanged)
- vp8_update_rate_correction_factors(cpi, 1);
-
- Q = (q_high + q_low) / 2;
-
- // Adjust cpi->zbin_over_quant (only allowed when Q is max)
- if (Q < MAXQ)
- cpi->zbin_over_quant = 0;
- else
- cpi->zbin_over_quant = (zbin_oq_high + zbin_oq_low) / 2;
- }
- else
- {
- // Update rate_correction_factor unless cpi->active_worst_quality has changed.
- if (!active_worst_qchanged)
- vp8_update_rate_correction_factors(cpi, 0);
-
- Q = vp8_regulate_q(cpi, cpi->this_frame_target);
-
- // Special case reset for qlow for constrained quality.
- // This should only trigger where there is very substantial
- // undershoot on a frame and the auto cq level is above
- // the user passsed in value.
- if ( (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
- (Q < q_low) )
- {
- q_low = Q;
- }
-
- while (((Q > q_high) || (cpi->zbin_over_quant > zbin_oq_high)) && (Retries < 10))
- {
- vp8_update_rate_correction_factors(cpi, 0);
- Q = vp8_regulate_q(cpi, cpi->this_frame_target);
- Retries ++;
- }
- }
-
- undershoot_seen = TRUE;
- }
-
- // Clamp Q to upper and lower limits:
- if (Q > q_high)
- Q = q_high;
- else if (Q < q_low)
- Q = q_low;
-
- // Clamp cpi->zbin_over_quant
- cpi->zbin_over_quant = (cpi->zbin_over_quant < zbin_oq_low) ? zbin_oq_low : (cpi->zbin_over_quant > zbin_oq_high) ? zbin_oq_high : cpi->zbin_over_quant;
-
- //Loop = ((Q != last_q) || (last_zbin_oq != cpi->zbin_over_quant)) ? TRUE : FALSE;
- Loop = ((Q != last_q)) ? TRUE : FALSE;
- last_zbin_oq = cpi->zbin_over_quant;
- }
- else
-#endif
- Loop = FALSE;
-
- if (cpi->is_src_frame_alt_ref)
- Loop = FALSE;
-
- if (Loop == TRUE)
- {
- vp8_restore_coding_context(cpi);
- loop_count++;
-#if CONFIG_PSNR
- cpi->tot_recode_hits++;
-#endif
- }
- }
- while (Loop == TRUE);
-
-#if 0
- // Experimental code for lagged and one pass
- // Update stats used for one pass GF selection
- {
- /*
- int frames_so_far;
- double frame_intra_error;
- double frame_coded_error;
- double frame_pcnt_inter;
- double frame_pcnt_motion;
- double frame_mvr;
- double frame_mvr_abs;
- double frame_mvc;
- double frame_mvc_abs;
- */
-
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_coded_error = (double)cpi->prediction_error;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_intra_error = (double)cpi->intra_error;
- cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_pcnt_inter = (double)(100 - cpi->this_frame_percent_intra) / 100.0;
- }
-#endif
-
- // Special case code to reduce pulsing when key frames are forced at a
- // fixed interval. Note the reconstruction error if it is the frame before
- // the force key frame
- if ( cpi->next_key_frame_forced && (cpi->frames_to_key == 0) )
- {
- cpi->ambient_err = vp8_calc_ss_err(cpi->Source,
- &cm->yv12_fb[cm->new_fb_idx],
- IF_RTCD(&cpi->rtcd.variance));
- }
-
- // This frame's MVs are saved and will be used in next frame's MV prediction.
- // Last frame has one more line(add to bottom) and one more column(add to right) than cm->mip. The edge elements are initialized to 0.
- if(cm->show_frame) //do not save for altref frame
- {
- int mb_row;
- int mb_col;
- MODE_INFO *tmp = cm->mip; //point to beginning of allocated MODE_INFO arrays.
-
- if(cm->frame_type != KEY_FRAME)
- {
- for (mb_row = 0; mb_row < cm->mb_rows+1; mb_row ++)
- {
- for (mb_col = 0; mb_col < cm->mb_cols+1; mb_col ++)
- {
- if(tmp->mbmi.ref_frame != INTRA_FRAME)
- cpi->lfmv[mb_col + mb_row*(cm->mode_info_stride+1)].as_int = tmp->mbmi.mv.as_int;
-
- cpi->lf_ref_frame_sign_bias[mb_col + mb_row*(cm->mode_info_stride+1)] = cm->ref_frame_sign_bias[tmp->mbmi.ref_frame];
- cpi->lf_ref_frame[mb_col + mb_row*(cm->mode_info_stride+1)] = tmp->mbmi.ref_frame;
- tmp++;
- }
- }
- }
- }
-
- // Update the GF useage maps.
- // This is done after completing the compression of a frame when all modes etc. are finalized but before loop filter
- // This is done after completing the compression of a frame when all modes etc. are finalized but before loop filter
- vp8_update_gf_useage_maps(cpi, cm, &cpi->mb);
-
- if (cm->frame_type == KEY_FRAME)
- cm->refresh_last_frame = 1;
-
-#if 0
- {
- FILE *f = fopen("gfactive.stt", "a");
- fprintf(f, "%8d %8d %8d %8d %8d\n", cm->current_video_frame, (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols), cpi->this_iiratio, cpi->next_iiratio, cm->refresh_golden_frame);
- fclose(f);
- }
-#endif
-
- // For inter frames the current default behaviour is that when cm->refresh_golden_frame is set we copy the old GF over to the ARF buffer
- // This is purely an encoder descision at present.
- if (!cpi->oxcf.error_resilient_mode && cm->refresh_golden_frame)
- cm->copy_buffer_to_arf = 2;
- else
- cm->copy_buffer_to_arf = 0;
-
- if (cm->refresh_last_frame)
- {
- vp8_swap_yv12_buffer(&cm->yv12_fb[cm->lst_fb_idx], &cm->yv12_fb[cm->new_fb_idx]);
- cm->frame_to_show = &cm->yv12_fb[cm->lst_fb_idx];
- }
- else
- cm->frame_to_show = &cm->yv12_fb[cm->new_fb_idx];
-
-
-#if CONFIG_MULTITHREAD
- if (cpi->b_multi_threaded)
- {
- sem_post(&cpi->h_event_start_lpf); /* start loopfilter in separate thread */
- }
- else
-#endif
- {
- loopfilter_frame(cpi, cm);
- }
-
- if (cpi->oxcf.error_resilient_mode == 1)
- {
- cm->refresh_entropy_probs = 0;
- }
-
-#if CONFIG_MULTITHREAD
- /* wait that filter_level is picked so that we can continue with stream packing */
- if (cpi->b_multi_threaded)
- sem_wait(&cpi->h_event_end_lpf);
-#endif
-
- // build the bitstream
- vp8_pack_bitstream(cpi, dest, size);
-
-#if CONFIG_MULTITHREAD
- /* wait for loopfilter thread done */
- if (cpi->b_multi_threaded)
- {
- sem_wait(&cpi->h_event_end_lpf);
- }
-#endif
-
- /* Move storing frame_type out of the above loop since it is also
- * needed in motion search besides loopfilter */
- cm->last_frame_type = cm->frame_type;
-
- // Update rate control heuristics
- cpi->total_byte_count += (*size);
- cpi->projected_frame_size = (*size) << 3;
-
- if (!active_worst_qchanged)
- vp8_update_rate_correction_factors(cpi, 2);
-
- cpi->last_q[cm->frame_type] = cm->base_qindex;
-
- if (cm->frame_type == KEY_FRAME)
- {
- vp8_adjust_key_frame_context(cpi);
- }
-
- // Keep a record of ambient average Q.
- if (cm->frame_type != KEY_FRAME)
- cpi->avg_frame_qindex = (2 + 3 * cpi->avg_frame_qindex + cm->base_qindex) >> 2;
-
- // Keep a record from which we can calculate the average Q excluding GF updates and key frames
- if ((cm->frame_type != KEY_FRAME) && !cm->refresh_golden_frame && !cm->refresh_alt_ref_frame)
- {
- cpi->ni_frames++;
-
- // Calculate the average Q for normal inter frames (not key or GFU
- // frames).
- if ( cpi->pass == 2 )
- {
- cpi->ni_tot_qi += Q;
- cpi->ni_av_qi = (cpi->ni_tot_qi / cpi->ni_frames);
- }
- else
- {
- // Damp value for first few frames
- if (cpi->ni_frames > 150 )
- {
- cpi->ni_tot_qi += Q;
- cpi->ni_av_qi = (cpi->ni_tot_qi / cpi->ni_frames);
- }
- // For one pass, early in the clip ... average the current frame Q
- // value with the worstq entered by the user as a dampening measure
- else
- {
- cpi->ni_tot_qi += Q;
- cpi->ni_av_qi = ((cpi->ni_tot_qi / cpi->ni_frames) + cpi->worst_quality + 1) / 2;
- }
-
- // If the average Q is higher than what was used in the last frame
- // (after going through the recode loop to keep the frame size within range)
- // then use the last frame value - 1.
- // The -1 is designed to stop Q and hence the data rate, from progressively
- // falling away during difficult sections, but at the same time reduce the number of
- // itterations around the recode loop.
- if (Q > cpi->ni_av_qi)
- cpi->ni_av_qi = Q - 1;
- }
- }
-
-#if 0
-
- // If the frame was massively oversize and we are below optimal buffer level drop next frame
- if ((cpi->drop_frames_allowed) &&
- (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
- (cpi->buffer_level < cpi->oxcf.drop_frames_water_mark * cpi->oxcf.optimal_buffer_level / 100) &&
- (cpi->projected_frame_size > (4 * cpi->this_frame_target)))
- {
- cpi->drop_frame = TRUE;
- }
-
-#endif
-
- // Set the count for maximum consequative dropped frames based upon the ratio of
- // this frame size to the target average per frame bandwidth.
- // (cpi->av_per_frame_bandwidth > 0) is just a sanity check to prevent / 0.
- if (cpi->drop_frames_allowed && (cpi->av_per_frame_bandwidth > 0))
- {
- cpi->max_drop_count = cpi->projected_frame_size / cpi->av_per_frame_bandwidth;
-
- if (cpi->max_drop_count > cpi->max_consec_dropped_frames)
- cpi->max_drop_count = cpi->max_consec_dropped_frames;
- }
-
- // Update the buffer level variable.
- // Non-viewable frames are a special case and are treated as pure overhead.
- if ( !cm->show_frame )
- cpi->bits_off_target -= cpi->projected_frame_size;
- else
- cpi->bits_off_target += cpi->av_per_frame_bandwidth - cpi->projected_frame_size;
-
- // Rolling monitors of whether we are over or underspending used to help regulate min and Max Q in two pass.
- cpi->rolling_target_bits = ((cpi->rolling_target_bits * 3) + cpi->this_frame_target + 2) / 4;
- cpi->rolling_actual_bits = ((cpi->rolling_actual_bits * 3) + cpi->projected_frame_size + 2) / 4;
- cpi->long_rolling_target_bits = ((cpi->long_rolling_target_bits * 31) + cpi->this_frame_target + 16) / 32;
- cpi->long_rolling_actual_bits = ((cpi->long_rolling_actual_bits * 31) + cpi->projected_frame_size + 16) / 32;
-
- // Actual bits spent
- cpi->total_actual_bits += cpi->projected_frame_size;
-
- // Debug stats
- cpi->total_target_vs_actual += (cpi->this_frame_target - cpi->projected_frame_size);
-
- cpi->buffer_level = cpi->bits_off_target;
-
- // Update bits left to the kf and gf groups to account for overshoot or undershoot on these frames
- if (cm->frame_type == KEY_FRAME)
- {
- cpi->kf_group_bits += cpi->this_frame_target - cpi->projected_frame_size;
-
- if (cpi->kf_group_bits < 0)
- cpi->kf_group_bits = 0 ;
- }
- else if (cm->refresh_golden_frame || cm->refresh_alt_ref_frame)
- {
- cpi->gf_group_bits += cpi->this_frame_target - cpi->projected_frame_size;
-
- if (cpi->gf_group_bits < 0)
- cpi->gf_group_bits = 0 ;
- }
-
- if (cm->frame_type != KEY_FRAME)
- {
- if (cpi->common.refresh_alt_ref_frame)
- {
- cpi->last_skip_false_probs[2] = cpi->prob_skip_false;
- cpi->last_skip_probs_q[2] = cm->base_qindex;
- }
- else if (cpi->common.refresh_golden_frame)
- {
- cpi->last_skip_false_probs[1] = cpi->prob_skip_false;
- cpi->last_skip_probs_q[1] = cm->base_qindex;
- }
- else
- {
- cpi->last_skip_false_probs[0] = cpi->prob_skip_false;
- cpi->last_skip_probs_q[0] = cm->base_qindex;
-
- //update the baseline
- cpi->base_skip_false_prob[cm->base_qindex] = cpi->prob_skip_false;
-
- }
- }
-
-#if 0 && CONFIG_PSNR
- {
- FILE *f = fopen("tmp.stt", "a");
-
- vp8_clear_system_state(); //__asm emms;
-
- if (cpi->total_coded_error_left != 0.0)
- fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %6ld %6ld"
- "%6ld %6ld %6ld %5ld %5ld %5ld %8ld %8.2f %10d %10.3f"
- "%10.3f %8ld\n",
- cpi->common.current_video_frame, cpi->this_frame_target,
- cpi->projected_frame_size,
- (cpi->projected_frame_size - cpi->this_frame_target),
- (int)cpi->total_target_vs_actual,
- (cpi->oxcf.starting_buffer_level-cpi->bits_off_target),
- (int)cpi->total_actual_bits, cm->base_qindex,
- cpi->active_best_quality, cpi->active_worst_quality,
- cpi->ni_av_qi, cpi->cq_target_quality, cpi->zbin_over_quant,
- //cpi->avg_frame_qindex, cpi->zbin_over_quant,
- cm->refresh_golden_frame, cm->refresh_alt_ref_frame,
- cm->frame_type, cpi->gfu_boost,
- cpi->est_max_qcorrection_factor, (int)cpi->bits_left,
- cpi->total_coded_error_left,
- (double)cpi->bits_left / cpi->total_coded_error_left,
- cpi->tot_recode_hits);
- else
- fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %6ld %6ld"
- "%6ld %6ld %6ld %5ld %5ld %5ld %8ld %8.2f %10d %10.3f"
- "%8ld\n",
- cpi->common.current_video_frame,
- cpi->this_frame_target, cpi->projected_frame_size,
- (cpi->projected_frame_size - cpi->this_frame_target),
- (int)cpi->total_target_vs_actual,
- (cpi->oxcf.starting_buffer_level-cpi->bits_off_target),
- (int)cpi->total_actual_bits, cm->base_qindex,
- cpi->active_best_quality, cpi->active_worst_quality,
- cpi->ni_av_qi, cpi->cq_target_quality, cpi->zbin_over_quant,
- //cpi->avg_frame_qindex, cpi->zbin_over_quant,
- cm->refresh_golden_frame, cm->refresh_alt_ref_frame,
- cm->frame_type, cpi->gfu_boost,
- cpi->est_max_qcorrection_factor, (int)cpi->bits_left,
- cpi->total_coded_error_left, cpi->tot_recode_hits);
-
- fclose(f);
-
- {
- FILE *fmodes = fopen("Modes.stt", "a");
- int i;
-
- fprintf(fmodes, "%6d:%1d:%1d:%1d ",
- cpi->common.current_video_frame,
- cm->frame_type, cm->refresh_golden_frame,
- cm->refresh_alt_ref_frame);
-
- for (i = 0; i < MAX_MODES; i++)
- fprintf(fmodes, "%5d ", cpi->mode_chosen_counts[i]);
-
- fprintf(fmodes, "\n");
-
- fclose(fmodes);
- }
- }
-
-#endif
-
- // If this was a kf or Gf note the Q
- if ((cm->frame_type == KEY_FRAME) || cm->refresh_golden_frame || cm->refresh_alt_ref_frame)
- cm->last_kf_gf_q = cm->base_qindex;
-
- if (cm->refresh_golden_frame == 1)
- cm->frame_flags = cm->frame_flags | FRAMEFLAGS_GOLDEN;
- else
- cm->frame_flags = cm->frame_flags&~FRAMEFLAGS_GOLDEN;
-
- if (cm->refresh_alt_ref_frame == 1)
- cm->frame_flags = cm->frame_flags | FRAMEFLAGS_ALTREF;
- else
- cm->frame_flags = cm->frame_flags&~FRAMEFLAGS_ALTREF;
-
-
- if (cm->refresh_last_frame & cm->refresh_golden_frame) // both refreshed
- cpi->gold_is_last = 1;
- else if (cm->refresh_last_frame ^ cm->refresh_golden_frame) // 1 refreshed but not the other
- cpi->gold_is_last = 0;
-
- if (cm->refresh_last_frame & cm->refresh_alt_ref_frame) // both refreshed
- cpi->alt_is_last = 1;
- else if (cm->refresh_last_frame ^ cm->refresh_alt_ref_frame) // 1 refreshed but not the other
- cpi->alt_is_last = 0;
-
- if (cm->refresh_alt_ref_frame & cm->refresh_golden_frame) // both refreshed
- cpi->gold_is_alt = 1;
- else if (cm->refresh_alt_ref_frame ^ cm->refresh_golden_frame) // 1 refreshed but not the other
- cpi->gold_is_alt = 0;
-
- cpi->ref_frame_flags = VP8_ALT_FLAG | VP8_GOLD_FLAG | VP8_LAST_FLAG;
-
- if (cpi->gold_is_last)
- cpi->ref_frame_flags &= ~VP8_GOLD_FLAG;
-
- if (cpi->alt_is_last)
- cpi->ref_frame_flags &= ~VP8_ALT_FLAG;
-
- if (cpi->gold_is_alt)
- cpi->ref_frame_flags &= ~VP8_ALT_FLAG;
-
-
- if (cpi->oxcf.error_resilient_mode)
- {
- if (cm->frame_type != KEY_FRAME)
- {
- // Is this an alternate reference update
- if (cm->refresh_alt_ref_frame)
- vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->yv12_fb[cm->alt_fb_idx]);
-
- if (cm->refresh_golden_frame)
- vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->yv12_fb[cm->gld_fb_idx]);
- }
- }
- else
- {
- if (cpi->oxcf.play_alternate && cm->refresh_alt_ref_frame && (cm->frame_type != KEY_FRAME))
- // Update the alternate reference frame and stats as appropriate.
- update_alt_ref_frame_and_stats(cpi);
- else
- // Update the Golden frame and golden frame and stats as appropriate.
- update_golden_frame_and_stats(cpi);
- }
-
- if (cm->frame_type == KEY_FRAME)
- {
- // Tell the caller that the frame was coded as a key frame
- *frame_flags = cm->frame_flags | FRAMEFLAGS_KEY;
-
- // As this frame is a key frame the next defaults to an inter frame.
- cm->frame_type = INTER_FRAME;
-
- cpi->last_frame_percent_intra = 100;
- }
- else
- {
- *frame_flags = cm->frame_flags&~FRAMEFLAGS_KEY;
-
- cpi->last_frame_percent_intra = cpi->this_frame_percent_intra;
- }
-
- // Clear the one shot update flags for segmentation map and mode/ref loop filter deltas.
- cpi->mb.e_mbd.update_mb_segmentation_map = 0;
- cpi->mb.e_mbd.update_mb_segmentation_data = 0;
- cpi->mb.e_mbd.mode_ref_lf_delta_update = 0;
-
-
- // Dont increment frame counters if this was an altref buffer update not a real frame
- if (cm->show_frame)
- {
- cm->current_video_frame++;
- cpi->frames_since_key++;
- }
-
- // reset to normal state now that we are done.
-
-
-
-#if 0
- {
- char filename[512];
- FILE *recon_file;
- sprintf(filename, "enc%04d.yuv", (int) cm->current_video_frame);
- recon_file = fopen(filename, "wb");
- fwrite(cm->yv12_fb[cm->lst_fb_idx].buffer_alloc,
- cm->yv12_fb[cm->lst_fb_idx].frame_size, 1, recon_file);
- fclose(recon_file);
- }
-#endif
-
- // DEBUG
- //vp8_write_yuv_frame("encoder_recon.yuv", cm->frame_to_show);
-
-
-}
-
-
-static void check_gf_quality(VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
- int gf_active_pct = (100 * cpi->gf_active_count) / (cm->mb_rows * cm->mb_cols);
- int gf_ref_usage_pct = (cpi->count_mb_ref_frame_usage[GOLDEN_FRAME] * 100) / (cm->mb_rows * cm->mb_cols);
- int last_ref_zz_useage = (cpi->inter_zz_count * 100) / (cm->mb_rows * cm->mb_cols);
-
- // Gf refresh is not currently being signalled
- if (cpi->gf_update_recommended == 0)
- {
- if (cpi->common.frames_since_golden > 7)
- {
- // Low use of gf
- if ((gf_active_pct < 10) || ((gf_active_pct + gf_ref_usage_pct) < 15))
- {
- // ...but last frame zero zero usage is reasonbable so a new gf might be appropriate
- if (last_ref_zz_useage >= 25)
- {
- cpi->gf_bad_count ++;
-
- if (cpi->gf_bad_count >= 8) // Check that the condition is stable
- {
- cpi->gf_update_recommended = 1;
- cpi->gf_bad_count = 0;
- }
- }
- else
- cpi->gf_bad_count = 0; // Restart count as the background is not stable enough
- }
- else
- cpi->gf_bad_count = 0; // Gf useage has picked up so reset count
- }
- }
- // If the signal is set but has not been read should we cancel it.
- else if (last_ref_zz_useage < 15)
- {
- cpi->gf_update_recommended = 0;
- cpi->gf_bad_count = 0;
- }
-
-#if 0
- {
- FILE *f = fopen("gfneeded.stt", "a");
- fprintf(f, "%10d %10d %10d %10d %10ld \n",
- cm->current_video_frame,
- cpi->common.frames_since_golden,
- gf_active_pct, gf_ref_usage_pct,
- cpi->gf_update_recommended);
- fclose(f);
- }
-
-#endif
-}
-
-#if !(CONFIG_REALTIME_ONLY)
-static void Pass2Encode(VP8_COMP *cpi, unsigned long *size, unsigned char *dest, unsigned int *frame_flags)
-{
-
- if (!cpi->common.refresh_alt_ref_frame)
- vp8_second_pass(cpi);
-
- encode_frame_to_data_rate(cpi, size, dest, frame_flags);
- cpi->bits_left -= 8 * *size;
-
- if (!cpi->common.refresh_alt_ref_frame)
- {
- double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth
- *cpi->oxcf.two_pass_vbrmin_section / 100);
- cpi->bits_left += (long long)(two_pass_min_rate / cpi->oxcf.frame_rate);
- }
-}
-#endif
-
-//For ARM NEON, d8-d15 are callee-saved registers, and need to be saved by us.
-#if HAVE_ARMV7
-extern void vp8_push_neon(INT64 *store);
-extern void vp8_pop_neon(INT64 *store);
-#endif
-int vp8_receive_raw_frame(VP8_PTR ptr, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, INT64 time_stamp, INT64 end_time)
-{
-#if HAVE_ARMV7
- INT64 store_reg[8];
-#endif
- VP8_COMP *cpi = (VP8_COMP *) ptr;
- VP8_COMMON *cm = &cpi->common;
- struct vpx_usec_timer timer;
-
- if (!cpi)
- return -1;
-
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_push_neon(store_reg);
- }
-#endif
-
- vpx_usec_timer_start(&timer);
-
- // no more room for frames;
- if (cpi->source_buffer_count != 0 && cpi->source_buffer_count >= cpi->oxcf.lag_in_frames)
- {
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(store_reg);
- }
-#endif
- return -1;
- }
-
- //printf("in-cpi->source_buffer_count: %d\n", cpi->source_buffer_count);
-
- cm->clr_type = sd->clrtype;
-
- // make a copy of the frame for use later...
-#if !(CONFIG_REALTIME_ONLY)
-
- if (cpi->oxcf.allow_lag)
- {
- int which_buffer = cpi->source_encode_index - 1;
- SOURCE_SAMPLE *s;
-
- if (which_buffer == -1)
- which_buffer = cpi->oxcf.lag_in_frames - 1;
-
- if (cpi->source_buffer_count < cpi->oxcf.lag_in_frames - 1)
- which_buffer = cpi->source_buffer_count;
-
- s = &cpi->src_buffer[which_buffer];
-
- s->source_time_stamp = time_stamp;
- s->source_end_time_stamp = end_time;
- s->source_frame_flags = frame_flags;
- vp8_yv12_copy_frame_ptr(sd, &s->source_buffer);
-
- cpi->source_buffer_count ++;
- }
- else
-#endif
- {
- SOURCE_SAMPLE *s;
- s = &cpi->src_buffer[0];
- s->source_end_time_stamp = end_time;
- s->source_time_stamp = time_stamp;
- s->source_frame_flags = frame_flags;
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_yv12_copy_src_frame_func_neon(sd, &s->source_buffer);
- }
-#if CONFIG_RUNTIME_CPU_DETECT
- else
-#endif
-#endif
-#if !HAVE_ARMV7 || CONFIG_RUNTIME_CPU_DETECT
- {
- vp8_yv12_copy_frame_ptr(sd, &s->source_buffer);
- }
-#endif
- cpi->source_buffer_count = 1;
- }
-
- vpx_usec_timer_mark(&timer);
- cpi->time_receive_data += vpx_usec_timer_elapsed(&timer);
-
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(store_reg);
- }
-#endif
-
- return 0;
-}
-int vp8_get_compressed_data(VP8_PTR ptr, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, INT64 *time_stamp, INT64 *time_end, int flush)
-{
-#if HAVE_ARMV7
- INT64 store_reg[8];
-#endif
- VP8_COMP *cpi = (VP8_COMP *) ptr;
- VP8_COMMON *cm = &cpi->common;
- struct vpx_usec_timer tsctimer;
- struct vpx_usec_timer ticktimer;
- struct vpx_usec_timer cmptimer;
-
- if (!cpi)
- return -1;
-
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_push_neon(store_reg);
- }
-#endif
-
- vpx_usec_timer_start(&cmptimer);
-
-
- // flush variable tells us that even though we have less than 10 frames
- // in our buffer we need to start producing compressed frames.
- // Probably because we are at the end of a file....
- if ((cpi->source_buffer_count == cpi->oxcf.lag_in_frames && cpi->oxcf.lag_in_frames > 0)
- || (!cpi->oxcf.allow_lag && cpi->source_buffer_count > 0)
- || (flush && cpi->source_buffer_count > 0))
- {
-
- SOURCE_SAMPLE *s;
-
- s = &cpi->src_buffer[cpi->source_encode_index];
- cpi->source_time_stamp = s->source_time_stamp;
- cpi->source_end_time_stamp = s->source_end_time_stamp;
-
-#if !(CONFIG_REALTIME_ONLY)
-
- // Should we code an alternate reference frame
- if (cpi->oxcf.error_resilient_mode == 0 &&
- cpi->oxcf.play_alternate &&
- cpi->source_alt_ref_pending &&
- (cpi->frames_till_gf_update_due < cpi->source_buffer_count) &&
- cpi->oxcf.lag_in_frames != 0)
- {
- cpi->last_alt_ref_sei = (cpi->source_encode_index + cpi->frames_till_gf_update_due) % cpi->oxcf.lag_in_frames;
-
-#if VP8_TEMPORAL_ALT_REF
-
- if (cpi->oxcf.arnr_max_frames > 0)
- {
-#if 0
- // my attempt at a loop that tests the results of strength filter.
- int start_frame = cpi->last_alt_ref_sei - 3;
-
- int i, besti = -1, pastin = cpi->oxcf.arnr_strength;
-
- int besterr;
-
- if (start_frame < 0)
- start_frame += cpi->oxcf.lag_in_frames;
-
- besterr = calc_low_ss_err(&cpi->src_buffer[cpi->last_alt_ref_sei].source_buffer,
- &cpi->src_buffer[start_frame].source_buffer, IF_RTCD(&cpi->rtcd.variance));
-
- for (i = 0; i < 7; i++)
- {
- int thiserr;
- cpi->oxcf.arnr_strength = i;
- vp8_temporal_filter_prepare_c(cpi);
-
- thiserr = calc_low_ss_err(&cpi->alt_ref_buffer.source_buffer,
- &cpi->src_buffer[start_frame].source_buffer, IF_RTCD(&cpi->rtcd.variance));
-
- if (10 * thiserr < besterr * 8)
- {
- besterr = thiserr;
- besti = i;
- }
- }
-
- if (besti != -1)
- {
- cpi->oxcf.arnr_strength = besti;
- vp8_temporal_filter_prepare_c(cpi);
- s = &cpi->alt_ref_buffer;
-
- // FWG not sure if I need to copy this data for the Alt Ref frame
- s->source_time_stamp = cpi->src_buffer[cpi->last_alt_ref_sei].source_time_stamp;
- s->source_end_time_stamp = cpi->src_buffer[cpi->last_alt_ref_sei].source_end_time_stamp;
- s->source_frame_flags = cpi->src_buffer[cpi->last_alt_ref_sei].source_frame_flags;
- }
- else
- s = &cpi->src_buffer[cpi->last_alt_ref_sei];
-
-#else
- vp8_temporal_filter_prepare_c(cpi);
- s = &cpi->alt_ref_buffer;
-
- // FWG not sure if I need to copy this data for the Alt Ref frame
- s->source_time_stamp = cpi->src_buffer[cpi->last_alt_ref_sei].source_time_stamp;
- s->source_end_time_stamp = cpi->src_buffer[cpi->last_alt_ref_sei].source_end_time_stamp;
- s->source_frame_flags = cpi->src_buffer[cpi->last_alt_ref_sei].source_frame_flags;
-
-#endif
- }
- else
-#endif
- s = &cpi->src_buffer[cpi->last_alt_ref_sei];
-
- cm->frames_till_alt_ref_frame = cpi->frames_till_gf_update_due;
- cm->refresh_alt_ref_frame = 1;
- cm->refresh_golden_frame = 0;
- cm->refresh_last_frame = 0;
- cm->show_frame = 0;
- cpi->source_alt_ref_pending = FALSE; // Clear Pending altf Ref flag.
- cpi->is_src_frame_alt_ref = 0;
- cpi->is_next_src_alt_ref = 0;
- }
- else
-#endif
- {
- cm->show_frame = 1;
-#if !(CONFIG_REALTIME_ONLY)
-
- if (cpi->oxcf.allow_lag)
- {
- if (cpi->source_encode_index == cpi->last_alt_ref_sei)
- {
- cpi->is_src_frame_alt_ref = 1;
- cpi->last_alt_ref_sei = -1;
- }
- else
- cpi->is_src_frame_alt_ref = 0;
-
- cpi->source_encode_index = (cpi->source_encode_index + 1) % cpi->oxcf.lag_in_frames;
-
- if(cpi->source_encode_index == cpi->last_alt_ref_sei)
- cpi->is_next_src_alt_ref = 1;
- else
- cpi->is_next_src_alt_ref = 0;
- }
-
-#endif
- cpi->source_buffer_count--;
- }
-
- cpi->un_scaled_source = &s->source_buffer;
- cpi->Source = &s->source_buffer;
- cpi->source_frame_flags = s->source_frame_flags;
-
- *time_stamp = cpi->source_time_stamp;
- *time_end = cpi->source_end_time_stamp;
- }
- else
- {
- *size = 0;
-#if !(CONFIG_REALTIME_ONLY)
-
- if (flush && cpi->pass == 1 && !cpi->first_pass_done)
- {
- vp8_end_first_pass(cpi); /* get last stats packet */
- cpi->first_pass_done = 1;
- }
-
-#endif
-
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(store_reg);
- }
-#endif
- return -1;
- }
-
- *frame_flags = cpi->source_frame_flags;
-
- if (cpi->source_time_stamp < cpi->first_time_stamp_ever)
- {
- cpi->first_time_stamp_ever = cpi->source_time_stamp;
- cpi->last_end_time_stamp_seen = cpi->source_time_stamp;
- }
-
- // adjust frame rates based on timestamps given
- if (!cm->refresh_alt_ref_frame)
- {
- if (cpi->source_time_stamp == cpi->first_time_stamp_ever)
- {
- double this_fps = 10000000.000 / (cpi->source_end_time_stamp - cpi->source_time_stamp);
-
- vp8_new_frame_rate(cpi, this_fps);
- }
- else
- {
- long long nanosecs = cpi->source_end_time_stamp
- - cpi->last_end_time_stamp_seen;
-
- if (nanosecs > 0)
- {
- double this_fps = 10000000.000 / nanosecs;
- vp8_new_frame_rate(cpi, (7 * cpi->oxcf.frame_rate + this_fps) / 8);
- }
-
- }
-
- cpi->last_time_stamp_seen = cpi->source_time_stamp;
- cpi->last_end_time_stamp_seen = cpi->source_end_time_stamp;
- }
-
- if (cpi->compressor_speed == 2)
- {
- check_gf_quality(cpi);
- vpx_usec_timer_start(&tsctimer);
- vpx_usec_timer_start(&ticktimer);
- }
-
- // start with a 0 size frame
- *size = 0;
-
- // Clear down mmx registers
- vp8_clear_system_state(); //__asm emms;
-
- cm->frame_type = INTER_FRAME;
- cm->frame_flags = *frame_flags;
-
-#if 0
-
- if (cm->refresh_alt_ref_frame)
- {
- //cm->refresh_golden_frame = 1;
- cm->refresh_golden_frame = 0;
- cm->refresh_last_frame = 0;
- }
- else
- {
- cm->refresh_golden_frame = 0;
- cm->refresh_last_frame = 1;
- }
-
-#endif
-
-#if !(CONFIG_REALTIME_ONLY)
-
- if (cpi->pass == 1)
- {
- Pass1Encode(cpi, size, dest, frame_flags);
- }
- else if (cpi->pass == 2)
- {
- Pass2Encode(cpi, size, dest, frame_flags);
- }
- else
-#endif
- encode_frame_to_data_rate(cpi, size, dest, frame_flags);
-
- if (cpi->compressor_speed == 2)
- {
- unsigned int duration, duration2;
- vpx_usec_timer_mark(&tsctimer);
- vpx_usec_timer_mark(&ticktimer);
-
- duration = vpx_usec_timer_elapsed(&ticktimer);
- duration2 = (unsigned int)((double)duration / 2);
-
- if (cm->frame_type != KEY_FRAME)
- {
- if (cpi->avg_encode_time == 0)
- cpi->avg_encode_time = duration;
- else
- cpi->avg_encode_time = (7 * cpi->avg_encode_time + duration) >> 3;
- }
-
- if (duration2)
- {
- //if(*frame_flags!=1)
- {
-
- if (cpi->avg_pick_mode_time == 0)
- cpi->avg_pick_mode_time = duration2;
- else
- cpi->avg_pick_mode_time = (7 * cpi->avg_pick_mode_time + duration2) >> 3;
- }
- }
-
- }
-
- if (cm->refresh_entropy_probs == 0)
- {
- vpx_memcpy(&cm->fc, &cm->lfc, sizeof(cm->fc));
- }
-
- // if its a dropped frame honor the requests on subsequent frames
- if (*size > 0)
- {
-
- // return to normal state
- cm->refresh_entropy_probs = 1;
- cm->refresh_alt_ref_frame = 0;
- cm->refresh_golden_frame = 0;
- cm->refresh_last_frame = 1;
- cm->frame_type = INTER_FRAME;
-
- }
-
- cpi->ready_for_new_frame = 1;
-
- vpx_usec_timer_mark(&cmptimer);
- cpi->time_compress_data += vpx_usec_timer_elapsed(&cmptimer);
-
- if (cpi->b_calculate_psnr && cpi->pass != 1 && cm->show_frame)
- {
- generate_psnr_packet(cpi);
- }
-
-#if CONFIG_PSNR
-
- if (cpi->pass != 1)
- {
- cpi->bytes += *size;
-
- if (cm->show_frame)
- {
-
- cpi->count ++;
-
- if (cpi->b_calculate_psnr)
- {
- double y, u, v;
- double ye,ue,ve;
- double frame_psnr;
- YV12_BUFFER_CONFIG *orig = cpi->Source;
- YV12_BUFFER_CONFIG *recon = cpi->common.frame_to_show;
- YV12_BUFFER_CONFIG *pp = &cm->post_proc_buffer;
- int y_samples = orig->y_height * orig->y_width ;
- int uv_samples = orig->uv_height * orig->uv_width ;
- int t_samples = y_samples + 2 * uv_samples;
- long long sq_error;
-
- ye = calc_plane_error(orig->y_buffer, orig->y_stride,
- recon->y_buffer, recon->y_stride, orig->y_width, orig->y_height,
- IF_RTCD(&cpi->rtcd.variance));
-
- ue = calc_plane_error(orig->u_buffer, orig->uv_stride,
- recon->u_buffer, recon->uv_stride, orig->uv_width, orig->uv_height,
- IF_RTCD(&cpi->rtcd.variance));
-
- ve = calc_plane_error(orig->v_buffer, orig->uv_stride,
- recon->v_buffer, recon->uv_stride, orig->uv_width, orig->uv_height,
- IF_RTCD(&cpi->rtcd.variance));
-
- sq_error = ye + ue + ve;
-
- frame_psnr = vp8_mse2psnr(t_samples, 255.0, sq_error);
-
- cpi->total_y += vp8_mse2psnr(y_samples, 255.0, ye);
- cpi->total_u += vp8_mse2psnr(uv_samples, 255.0, ue);
- cpi->total_v += vp8_mse2psnr(uv_samples, 255.0, ve);
- cpi->total_sq_error += sq_error;
- cpi->total += frame_psnr;
- {
- double y2, u2, v2, frame_psnr2, frame_ssim2 = 0;
- double weight = 0;
-
- vp8_deblock(cm->frame_to_show, &cm->post_proc_buffer, cm->filter_level * 10 / 6, 1, 0, IF_RTCD(&cm->rtcd.postproc));
- vp8_clear_system_state();
-
- ye = calc_plane_error(orig->y_buffer, orig->y_stride,
- pp->y_buffer, pp->y_stride, orig->y_width, orig->y_height,
- IF_RTCD(&cpi->rtcd.variance));
-
- ue = calc_plane_error(orig->u_buffer, orig->uv_stride,
- pp->u_buffer, pp->uv_stride, orig->uv_width, orig->uv_height,
- IF_RTCD(&cpi->rtcd.variance));
-
- ve = calc_plane_error(orig->v_buffer, orig->uv_stride,
- pp->v_buffer, pp->uv_stride, orig->uv_width, orig->uv_height,
- IF_RTCD(&cpi->rtcd.variance));
-
- sq_error = ye + ue + ve;
-
- frame_psnr2 = vp8_mse2psnr(t_samples, 255.0, sq_error);
-
- cpi->totalp_y += vp8_mse2psnr(y_samples, 255.0, ye);
- cpi->totalp_u += vp8_mse2psnr(uv_samples, 255.0, ue);
- cpi->totalp_v += vp8_mse2psnr(uv_samples, 255.0, ve);
- cpi->total_sq_error2 += sq_error;
- cpi->totalp += frame_psnr2;
-
- frame_ssim2 = vp8_calc_ssim(cpi->Source,
- &cm->post_proc_buffer, 1, &weight,
- IF_RTCD(&cpi->rtcd.variance));
-
- cpi->summed_quality += frame_ssim2 * weight;
- cpi->summed_weights += weight;
-
- }
- }
-
- if (cpi->b_calculate_ssimg)
- {
- double y, u, v, frame_all;
- frame_all = vp8_calc_ssimg(cpi->Source, cm->frame_to_show, &y, &u, &v);
- cpi->total_ssimg_y += y;
- cpi->total_ssimg_u += u;
- cpi->total_ssimg_v += v;
- cpi->total_ssimg_all += frame_all;
- }
-
- }
- }
-
-#if 0
-
- if (cpi->common.frame_type != 0 && cpi->common.base_qindex == cpi->oxcf.worst_allowed_q)
- {
- skiptruecount += cpi->skip_true_count;
- skipfalsecount += cpi->skip_false_count;
- }
-
-#endif
-#if 0
-
- if (cpi->pass != 1)
- {
- FILE *f = fopen("skip.stt", "a");
- fprintf(f, "frame:%4d flags:%4x Q:%4d P:%4d Size:%5d\n", cpi->common.current_video_frame, *frame_flags, cpi->common.base_qindex, cpi->prob_skip_false, *size);
-
- if (cpi->is_src_frame_alt_ref == 1)
- fprintf(f, "skipcount: %4d framesize: %d\n", cpi->skip_true_count , *size);
-
- fclose(f);
- }
-
-#endif
-#endif
-
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_pop_neon(store_reg);
- }
-#endif
-
- return 0;
-}
-
-int vp8_get_preview_raw_frame(VP8_PTR comp, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags)
-{
- VP8_COMP *cpi = (VP8_COMP *) comp;
-
- if (cpi->common.refresh_alt_ref_frame)
- return -1;
- else
- {
- int ret;
-#if CONFIG_POSTPROC
- ret = vp8_post_proc_frame(&cpi->common, dest, flags);
-#else
-
- if (cpi->common.frame_to_show)
- {
- *dest = *cpi->common.frame_to_show;
- dest->y_width = cpi->common.Width;
- dest->y_height = cpi->common.Height;
- dest->uv_height = cpi->common.Height / 2;
- ret = 0;
- }
- else
- {
- ret = -1;
- }
-
-#endif //!CONFIG_POSTPROC
- vp8_clear_system_state();
- return ret;
- }
-}
-
-int vp8_set_roimap(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4])
-{
- VP8_COMP *cpi = (VP8_COMP *) comp;
- signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
-
- if (cpi->common.mb_rows != rows || cpi->common.mb_cols != cols)
- return -1;
-
- if (!map)
- {
- disable_segmentation((VP8_PTR)cpi);
- return 0;
- }
-
- // Set the segmentation Map
- set_segmentation_map((VP8_PTR)cpi, map);
-
- // Activate segmentation.
- enable_segmentation((VP8_PTR)cpi);
-
- // Set up the quant segment data
- feature_data[MB_LVL_ALT_Q][0] = delta_q[0];
- feature_data[MB_LVL_ALT_Q][1] = delta_q[1];
- feature_data[MB_LVL_ALT_Q][2] = delta_q[2];
- feature_data[MB_LVL_ALT_Q][3] = delta_q[3];
-
- // Set up the loop segment data s
- feature_data[MB_LVL_ALT_LF][0] = delta_lf[0];
- feature_data[MB_LVL_ALT_LF][1] = delta_lf[1];
- feature_data[MB_LVL_ALT_LF][2] = delta_lf[2];
- feature_data[MB_LVL_ALT_LF][3] = delta_lf[3];
-
- cpi->segment_encode_breakout[0] = threshold[0];
- cpi->segment_encode_breakout[1] = threshold[1];
- cpi->segment_encode_breakout[2] = threshold[2];
- cpi->segment_encode_breakout[3] = threshold[3];
-
- // Initialise the feature data structure
- // SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
- set_segment_data((VP8_PTR)cpi, &feature_data[0][0], SEGMENT_DELTADATA);
-
- return 0;
-}
-
-int vp8_set_active_map(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols)
-{
- VP8_COMP *cpi = (VP8_COMP *) comp;
-
- if (rows == cpi->common.mb_rows && cols == cpi->common.mb_cols)
- {
- if (map)
- {
- vpx_memcpy(cpi->active_map, map, rows * cols);
- cpi->active_map_enabled = 1;
- }
- else
- cpi->active_map_enabled = 0;
-
- return 0;
- }
- else
- {
- //cpi->active_map_enabled = 0;
- return -1 ;
- }
-}
-
-int vp8_set_internal_size(VP8_PTR comp, VPX_SCALING horiz_mode, VPX_SCALING vert_mode)
-{
- VP8_COMP *cpi = (VP8_COMP *) comp;
-
- if (horiz_mode <= ONETWO)
- cpi->common.horiz_scale = horiz_mode;
- else
- return -1;
-
- if (vert_mode <= ONETWO)
- cpi->common.vert_scale = vert_mode;
- else
- return -1;
-
- return 0;
-}
-
-
-
-int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, const vp8_variance_rtcd_vtable_t *rtcd)
-{
- int i, j;
- int Total = 0;
-
- unsigned char *src = source->y_buffer;
- unsigned char *dst = dest->y_buffer;
- (void)rtcd;
-
- // Loop through the Y plane raw and reconstruction data summing (square differences)
- for (i = 0; i < source->y_height; i += 16)
- {
- for (j = 0; j < source->y_width; j += 16)
- {
- unsigned int sse;
- Total += VARIANCE_INVOKE(rtcd, mse16x16)(src + j, source->y_stride, dst + j, dest->y_stride, &sse);
- }
-
- src += 16 * source->y_stride;
- dst += 16 * dest->y_stride;
- }
-
- return Total;
-}
-
-
-static int calc_low_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, const vp8_variance_rtcd_vtable_t *rtcd)
-{
- int i, j;
- int Total = 0;
-
- unsigned char *src = source->y_buffer;
- unsigned char *dst = dest->y_buffer;
- (void)rtcd;
-
- // Loop through the Y plane raw and reconstruction data summing (square differences)
- for (i = 0; i < source->y_height; i += 16)
- {
- for (j = 0; j < source->y_width; j += 16)
- {
- unsigned int sse;
- VARIANCE_INVOKE(rtcd, mse16x16)(src + j, source->y_stride, dst + j, dest->y_stride, &sse);
-
- if (sse < 8096)
- Total += sse;
- }
-
- src += 16 * source->y_stride;
- dst += 16 * dest->y_stride;
- }
-
- return Total;
-}
-
-
-int vp8_get_quantizer(VP8_PTR c)
-{
- VP8_COMP *cpi = (VP8_COMP *) c;
- return cpi->common.base_qindex;
-}
diff --git a/vp8/encoder/onyx_int.h b/vp8/encoder/onyx_int.h
deleted file mode 100644
index 0e53f68..0000000
--- a/vp8/encoder/onyx_int.h
+++ /dev/null
@@ -1,717 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_VP8_INT_H
-#define __INC_VP8_INT_H
-
-#include <stdio.h>
-#include "vpx_ports/config.h"
-#include "vp8/common/onyx.h"
-#include "treewriter.h"
-#include "tokenize.h"
-#include "vp8/common/onyxc_int.h"
-#include "variance.h"
-#include "dct.h"
-#include "encodemb.h"
-#include "quantize.h"
-#include "vp8/common/entropy.h"
-#include "vp8/common/threading.h"
-#include "vpx_ports/mem.h"
-#include "vpx/internal/vpx_codec_internal.h"
-#include "mcomp.h"
-#include "temporal_filter.h"
-#include "vp8/common/findnearmv.h"
-
-//#define SPEEDSTATS 1
-#define MIN_GF_INTERVAL 4
-#define DEFAULT_GF_INTERVAL 7
-
-#define KEY_FRAME_CONTEXT 5
-
-#define MAX_LAG_BUFFERS (CONFIG_REALTIME_ONLY? 1 : 25)
-
-#define AF_THRESH 25
-#define AF_THRESH2 100
-#define ARF_DECAY_THRESH 12
-#define MAX_MODES 20
-
-#define MIN_THRESHMULT 32
-#define MAX_THRESHMULT 512
-
-#define GF_ZEROMV_ZBIN_BOOST 24
-#define LF_ZEROMV_ZBIN_BOOST 12
-#define MV_ZBIN_BOOST 4
-#define ZBIN_OQ_MAX 192
-
-#if !(CONFIG_REALTIME_ONLY)
-#define VP8_TEMPORAL_ALT_REF 1
-#endif
-
-typedef struct
-{
- int kf_indicated;
- unsigned int frames_since_key;
- unsigned int frames_since_golden;
- int filter_level;
- int frames_till_gf_update_due;
- int recent_ref_frame_usage[MAX_REF_FRAMES];
-
- MV_CONTEXT mvc[2];
- int mvcosts[2][MVvals+1];
-
-#ifdef MODE_STATS
- // Stats
- int y_modes[5];
- int uv_modes[4];
- int b_modes[10];
- int inter_y_modes[10];
- int inter_uv_modes[4];
- int inter_b_modes[10];
-#endif
-
- vp8_prob ymode_prob[4], uv_mode_prob[3]; /* interframe intra mode probs */
- vp8_prob kf_ymode_prob[4], kf_uv_mode_prob[3]; /* keyframe "" */
-
- int ymode_count[5], uv_mode_count[4]; /* intra MB type cts this frame */
-
- int count_mb_ref_frame_usage[MAX_REF_FRAMES];
-
- int this_frame_percent_intra;
- int last_frame_percent_intra;
-
-
-} CODING_CONTEXT;
-
-typedef struct
-{
- double frame;
- double intra_error;
- double coded_error;
- double ssim_weighted_pred_err;
- double pcnt_inter;
- double pcnt_motion;
- double pcnt_second_ref;
- double pcnt_neutral;
- double MVr;
- double mvr_abs;
- double MVc;
- double mvc_abs;
- double MVrv;
- double MVcv;
- double mv_in_out_count;
- double duration;
- double count;
-}
-FIRSTPASS_STATS;
-
-typedef struct
-{
- int frames_so_far;
- double frame_intra_error;
- double frame_coded_error;
- double frame_pcnt_inter;
- double frame_pcnt_motion;
- double frame_mvr;
- double frame_mvr_abs;
- double frame_mvc;
- double frame_mvc_abs;
-
-} ONEPASS_FRAMESTATS;
-
-
-typedef enum
-{
- THR_ZEROMV = 0,
- THR_DC = 1,
-
- THR_NEARESTMV = 2,
- THR_NEARMV = 3,
-
- THR_ZEROG = 4,
- THR_NEARESTG = 5,
-
- THR_ZEROA = 6,
- THR_NEARESTA = 7,
-
- THR_NEARG = 8,
- THR_NEARA = 9,
-
- THR_V_PRED = 10,
- THR_H_PRED = 11,
- THR_TM = 12,
-
- THR_NEWMV = 13,
- THR_NEWG = 14,
- THR_NEWA = 15,
-
- THR_SPLITMV = 16,
- THR_SPLITG = 17,
- THR_SPLITA = 18,
-
- THR_B_PRED = 19,
-}
-THR_MODES;
-
-typedef enum
-{
- DIAMOND = 0,
- NSTEP = 1,
- HEX = 2
-} SEARCH_METHODS;
-
-typedef struct
-{
- int RD;
- SEARCH_METHODS search_method;
- int improved_quant;
- int improved_dct;
- int auto_filter;
- int recode_loop;
- int iterative_sub_pixel;
- int half_pixel_search;
- int quarter_pixel_search;
- int thresh_mult[MAX_MODES];
- int full_freq[2];
- int min_fs_radius;
- int max_fs_radius;
- int max_step_search_steps;
- int first_step;
- int optimize_coefficients;
-
- int use_fastquant_for_pick;
- int no_skip_block4x4_search;
- int improved_mv_pred;
-
-} SPEED_FEATURES;
-
-typedef struct
-{
- MACROBLOCK mb;
- int segment_counts[MAX_MB_SEGMENTS];
- int totalrate;
-} MB_ROW_COMP;
-
-typedef struct
-{
- TOKENEXTRA *start;
- TOKENEXTRA *stop;
-} TOKENLIST;
-
-typedef struct
-{
- int ithread;
- void *ptr1;
- void *ptr2;
-} ENCODETHREAD_DATA;
-typedef struct
-{
- int ithread;
- void *ptr1;
-} LPFTHREAD_DATA;
-
-typedef struct
-{
- INT64 source_time_stamp;
- INT64 source_end_time_stamp;
-
- DECLARE_ALIGNED(16, YV12_BUFFER_CONFIG, source_buffer);
- unsigned int source_frame_flags;
-} SOURCE_SAMPLE;
-
-typedef struct VP8_ENCODER_RTCD
-{
- VP8_COMMON_RTCD *common;
- vp8_variance_rtcd_vtable_t variance;
- vp8_fdct_rtcd_vtable_t fdct;
- vp8_encodemb_rtcd_vtable_t encodemb;
- vp8_quantize_rtcd_vtable_t quantize;
- vp8_search_rtcd_vtable_t search;
- vp8_temporal_rtcd_vtable_t temporal;
-} VP8_ENCODER_RTCD;
-
-enum
-{
- BLOCK_16X8,
- BLOCK_8X16,
- BLOCK_8X8,
- BLOCK_4X4,
- BLOCK_16X16,
- BLOCK_MAX_SEGMENTS
-};
-
-typedef struct
-{
-
- DECLARE_ALIGNED(16, short, Y1quant[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y1quant_shift[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y1zbin[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y1round[QINDEX_RANGE][16]);
-
- DECLARE_ALIGNED(16, short, Y2quant[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y2quant_shift[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y2zbin[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y2round[QINDEX_RANGE][16]);
-
- DECLARE_ALIGNED(16, short, UVquant[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, UVquant_shift[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, UVzbin[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, UVround[QINDEX_RANGE][16]);
-
- DECLARE_ALIGNED(16, short, zrun_zbin_boost_y1[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, zrun_zbin_boost_y2[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, zrun_zbin_boost_uv[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y1quant_fast[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, Y2quant_fast[QINDEX_RANGE][16]);
- DECLARE_ALIGNED(16, short, UVquant_fast[QINDEX_RANGE][16]);
-
-
- MACROBLOCK mb;
- VP8_COMMON common;
- vp8_writer bc, bc2;
- // bool_writer *bc2;
-
- VP8_CONFIG oxcf;
-
- YV12_BUFFER_CONFIG *Source;
- YV12_BUFFER_CONFIG *un_scaled_source;
- INT64 source_time_stamp;
- INT64 source_end_time_stamp;
- unsigned int source_frame_flags;
- YV12_BUFFER_CONFIG scaled_source;
-
- int source_buffer_count; // number of src_buffers in use for lagged encoding
- int source_encode_index; // index of buffer in src_buffer to encode
- int source_alt_ref_pending; // frame in src_buffers has been identified to be encoded as an alt ref
- int source_alt_ref_active; // an alt ref frame has been encoded and is usable
-
- int last_alt_ref_sei; // index into src_buffers of frame used as alt reference
- int is_src_frame_alt_ref; // source of frame to encode is an exact copy of an alt ref frame
- int is_next_src_alt_ref; // source of next frame to encode is an exact copy of an alt ref frame
-
- int gold_is_last; // golden frame same as last frame ( short circuit gold searches)
- int alt_is_last; // Alt reference frame same as last ( short circuit altref search)
- int gold_is_alt; // don't do both alt and gold search ( just do gold).
-
- //int refresh_alt_ref_frame;
- SOURCE_SAMPLE src_buffer[MAX_LAG_BUFFERS];
-
- YV12_BUFFER_CONFIG last_frame_uf;
-
- TOKENEXTRA *tok;
- unsigned int tok_count;
-
-
- unsigned int frames_since_key;
- unsigned int key_frame_frequency;
- unsigned int this_key_frame_forced;
- unsigned int next_key_frame_forced;
-
- // Ambient reconstruction err target for force key frames
- int ambient_err;
-
- unsigned int mode_check_freq[MAX_MODES];
- unsigned int mode_test_hit_counts[MAX_MODES];
- unsigned int mode_chosen_counts[MAX_MODES];
- unsigned int mbs_tested_so_far;
-
- unsigned int check_freq[2];
- unsigned int do_full[2];
-
- int rd_thresh_mult[MAX_MODES];
- int rd_baseline_thresh[MAX_MODES];
- int rd_threshes[MAX_MODES];
- int mvcostbase;
- int mvcostmultiplier;
- int subseqblockweight;
- int errthresh;
- unsigned int activity_avg;
-
- int RDMULT;
- int RDDIV ;
-
- CODING_CONTEXT coding_context;
-
- // Rate targetting variables
- long long prediction_error;
- long long last_prediction_error;
- long long intra_error;
- long long last_intra_error;
-
-#if 0
- // Experimental RD code
- long long frame_distortion;
- long long last_frame_distortion;
-#endif
-
- int last_mb_distortion;
-
- int frames_since_auto_filter;
-
- int this_frame_target;
- int projected_frame_size;
- int last_q[2]; // Separate values for Intra/Inter
-
- double rate_correction_factor;
- double key_frame_rate_correction_factor;
- double gf_rate_correction_factor;
- double est_max_qcorrection_factor;
-
- int frames_till_gf_update_due; // Count down till next GF
- int current_gf_interval; // GF interval chosen when we coded the last GF
-
- int gf_overspend_bits; // Total bits overspent becasue of GF boost (cumulative)
-
- int gf_group_bits; // Projected Bits available for a group of frames including 1 GF or ARF
- int gf_bits; // Bits for the golden frame or ARF - 2 pass only
- int mid_gf_extra_bits; // A few extra bits for the frame half way between two gfs.
-
- // Projected total bits available for a key frame group of frames
- long long kf_group_bits;
-
- // Error score of frames still to be coded in kf group
- long long kf_group_error_left;
-
- // Bits for the key frame in a key frame group - 2 pass only
- int kf_bits;
-
- int non_gf_bitrate_adjustment; // Used in the few frames following a GF to recover the extra bits spent in that GF
- int initial_gf_use; // percentage use of gf 2 frames after gf
-
- int gf_group_error_left; // Remaining error from uncoded frames in a gf group. Two pass use only
-
- int kf_overspend_bits; // Extra bits spent on key frames that need to be recovered on inter frames
- int kf_bitrate_adjustment; // Current number of bit s to try and recover on each inter frame.
- int max_gf_interval;
- int static_scene_max_gf_interval;
- int baseline_gf_interval;
- int gf_decay_rate;
- int active_arnr_frames; // <= cpi->oxcf.arnr_max_frames
-
- INT64 key_frame_count;
- INT64 tot_key_frame_bits;
- int prior_key_frame_size[KEY_FRAME_CONTEXT];
- int prior_key_frame_distance[KEY_FRAME_CONTEXT];
- int per_frame_bandwidth; // Current section per frame bandwidth target
- int av_per_frame_bandwidth; // Average frame size target for clip
- int min_frame_bandwidth; // Minimum allocation that should be used for any frame
- int last_key_frame_size;
- int intra_frame_target;
- int inter_frame_target;
- double output_frame_rate;
- long long last_time_stamp_seen;
- long long last_end_time_stamp_seen;
- long long first_time_stamp_ever;
-
- int ni_av_qi;
- int ni_tot_qi;
- int ni_frames;
- int avg_frame_qindex;
-
- int zbin_over_quant;
- int zbin_mode_boost;
- int zbin_mode_boost_enabled;
-
- INT64 total_byte_count;
-
- int buffered_mode;
-
- int buffer_level;
- int bits_off_target;
-
- int rolling_target_bits;
- int rolling_actual_bits;
-
- int long_rolling_target_bits;
- int long_rolling_actual_bits;
-
- long long total_actual_bits;
- int total_target_vs_actual; // debug stats
-
- int worst_quality;
- int active_worst_quality;
- int best_quality;
- int active_best_quality;
-
- int cq_target_quality;
- int maxq_max_limit;
- int maxq_min_limit;
-
- int drop_frames_allowed; // Are we permitted to drop frames?
- int drop_frame; // Drop this frame?
- int drop_count; // How many frames have we dropped?
- int max_drop_count; // How many frames should we drop?
- int max_consec_dropped_frames; // Limit number of consecutive frames that can be dropped.
-
-
- int ymode_count [VP8_YMODES]; /* intra MB type cts this frame */
- int uv_mode_count[VP8_UV_MODES]; /* intra MB type cts this frame */
-
- unsigned int MVcount [2] [MVvals]; /* (row,col) MV cts this frame */
-
- unsigned int coef_counts [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens]; /* for this frame */
- //DECLARE_ALIGNED(16, int, coef_counts_backup [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens]); //not used any more
- //save vp8_tree_probs_from_distribution result for each frame to avoid repeat calculation
- vp8_prob frame_coef_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1];
- unsigned int frame_branch_ct [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1][2];
-
- /* Second compressed data partition contains coefficient data. */
-
- unsigned char *output_partition2;
- size_t output_partition2size;
-
- int frames_to_key;
- int gfu_boost;
- int kf_boost;
- int last_boost;
- double total_error_left;
- double total_intra_error_left;
- double total_coded_error_left;
- double start_tot_err_left;
- double min_error;
- double kf_intra_err_min;
- double gf_intra_err_min;
-
- double modified_error_total;
- double modified_error_used;
- double modified_error_left;
- double clip_bpe;
- double observed_bpe;
-
- double avg_iiratio;
-
- int target_bandwidth;
- long long bits_left;
- long long clip_bits_total;
- FIRSTPASS_STATS *total_stats;
- FIRSTPASS_STATS *this_frame_stats;
- FIRSTPASS_STATS *stats_in, *stats_in_end;
- struct vpx_codec_pkt_list *output_pkt_list;
- int first_pass_done;
-
-#if 0
- // Experimental code for lagged and one pass
- ONEPASS_FRAMESTATS one_pass_frame_stats[MAX_LAG_BUFFERS];
- int one_pass_frame_index;
-#endif
-
- int decimation_factor;
- int decimation_count;
-
- // for real time encoding
- int avg_encode_time; //microsecond
- int avg_pick_mode_time; //microsecond
- int Speed;
- unsigned int cpu_freq; //Mhz
- int compressor_speed;
-
- int interquantizer;
- int auto_gold;
- int auto_adjust_gold_quantizer;
- int goldquantizer;
- int goldfreq;
- int auto_adjust_key_quantizer;
- int keyquantizer;
- int auto_worst_q;
- int filter_type;
- int cpu_used;
- int chroma_boost;
- int horiz_scale;
- int vert_scale;
- int pass;
-
-
- int prob_intra_coded;
- int prob_last_coded;
- int prob_gf_coded;
- int prob_skip_false;
- int last_skip_false_probs[3];
- int last_skip_probs_q[3];
- int recent_ref_frame_usage[MAX_REF_FRAMES];
-
- int count_mb_ref_frame_usage[MAX_REF_FRAMES];
- int this_frame_percent_intra;
- int last_frame_percent_intra;
-
- int last_key_frame_q;
- int last_kffilt_lvl;
-
- int ref_frame_flags;
-
- SPEED_FEATURES sf;
- int error_bins[1024];
-
- int inter_lvl;
- int intra_lvl;
- int motion_lvl;
- int motion_speed;
- int motion_var;
- unsigned int next_iiratio;
- unsigned int this_iiratio;
- int this_frame_modified_error;
-
- double norm_intra_err_per_mb;
- double norm_inter_err_per_mb;
- double norm_iidiff_per_mb;
-
- int last_best_mode_index; // Record of mode index chosen for previous macro block.
- int last_auto_filt_val;
- int last_auto_filt_q;
-
- // Data used for real time conferencing mode to help determine if it would be good to update the gf
- int inter_zz_count;
- int gf_bad_count;
- int gf_update_recommended;
- int skip_true_count;
- int skip_false_count;
-
- int alt_qcount;
-
- int ready_for_new_frame;
-
- unsigned char *segmentation_map;
- signed char segment_feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS]; // Segment data (can be deltas or absolute values)
- int segment_encode_breakout[MAX_MB_SEGMENTS]; // segment threashold for encode breakout
-
- unsigned char *active_map;
- unsigned int active_map_enabled;
- // Video conferencing cyclic refresh mode flags etc
- // This is a mode designed to clean up the background over time in live encoding scenarious. It uses segmentation
- int cyclic_refresh_mode_enabled;
- int cyclic_refresh_mode_max_mbs_perframe;
- int cyclic_refresh_mode_index;
- int cyclic_refresh_q;
- signed char *cyclic_refresh_map;
-
-#if CONFIG_MULTITHREAD
- // multithread data
- int * mt_current_mb_col;
- int mt_sync_range;
- int processor_core_count;
- int b_multi_threaded;
- int encoding_thread_count;
-
- pthread_t *h_encoding_thread;
- pthread_t h_filter_thread;
-
- MB_ROW_COMP *mb_row_ei;
- ENCODETHREAD_DATA *en_thread_data;
- LPFTHREAD_DATA lpf_thread_data;
-
- //events
- sem_t *h_event_start_encoding;
- sem_t h_event_end_encoding;
- sem_t h_event_start_lpf;
- sem_t h_event_end_lpf;
-#endif
-
- TOKENLIST *tplist;
- // end of multithread data
-
-
- fractional_mv_step_fp *find_fractional_mv_step;
- vp8_full_search_fn_t full_search_sad;
- vp8_diamond_search_fn_t diamond_search_sad;
- vp8_variance_fn_ptr_t fn_ptr[BLOCK_MAX_SEGMENTS];
- unsigned int time_receive_data;
- unsigned int time_compress_data;
- unsigned int time_pick_lpf;
- unsigned int time_encode_mb_row;
-
- unsigned int tempdata1;
- unsigned int tempdata2;
-
- int base_skip_false_prob[128];
- unsigned int section_intra_rating;
-
- double section_max_qfactor;
-
-
-#if CONFIG_RUNTIME_CPU_DETECT
- VP8_ENCODER_RTCD rtcd;
-#endif
-#if VP8_TEMPORAL_ALT_REF
- SOURCE_SAMPLE alt_ref_buffer;
- YV12_BUFFER_CONFIG *frames[MAX_LAG_BUFFERS];
- int fixed_divide[512];
-#endif
-
-#if CONFIG_PSNR
- int count;
- double total_y;
- double total_u;
- double total_v;
- double total ;
- double total_sq_error;
- double totalp_y;
- double totalp_u;
- double totalp_v;
- double totalp;
- double total_sq_error2;
- int bytes;
- double summed_quality;
- double summed_weights;
- unsigned int tot_recode_hits;
-
-
- double total_ssimg_y;
- double total_ssimg_u;
- double total_ssimg_v;
- double total_ssimg_all;
-
- int b_calculate_ssimg;
-#endif
- int b_calculate_psnr;
-
-
- unsigned char *gf_active_flags; // Record of which MBs still refer to last golden frame either directly or through 0,0
- int gf_active_count;
-
- //Store last frame's MV info for next frame MV prediction
- int_mv *lfmv;
- int *lf_ref_frame_sign_bias;
- int *lf_ref_frame;
-
-#if CONFIG_REALTIME_ONLY
- int force_next_frame_intra; /* force next frame to intra when kf_auto says so */
-#endif
-} VP8_COMP;
-
-void control_data_rate(VP8_COMP *cpi);
-
-void vp8_encode_frame(VP8_COMP *cpi);
-
-void vp8_pack_bitstream(VP8_COMP *cpi, unsigned char *dest, unsigned long *size);
-
-unsigned int vp8_activity_masking(VP8_COMP *cpi, MACROBLOCK *x);
-
-int rd_cost_intra_mb(MACROBLOCKD *x);
-
-void vp8_tokenize_mb(VP8_COMP *, MACROBLOCKD *, TOKENEXTRA **);
-
-void vp8_set_speed_features(VP8_COMP *cpi);
-
-#if CONFIG_DEBUG
-#define CHECK_MEM_ERROR(lval,expr) do {\
- lval = (expr); \
- if(!lval) \
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,\
- "Failed to allocate "#lval" at %s:%d", \
- __FILE__,__LINE__);\
- } while(0)
-#else
-#define CHECK_MEM_ERROR(lval,expr) do {\
- lval = (expr); \
- if(!lval) \
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,\
- "Failed to allocate "#lval);\
- } while(0)
-#endif
-#endif
diff --git a/vp8/encoder/parms.cpp b/vp8/encoder/parms.cpp
deleted file mode 100644
index 2a39b2c..0000000
--- a/vp8/encoder/parms.cpp
+++ /dev/null
@@ -1,107 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#if 0
-
-#include <map>
-#include <string>
-#include <fstream>
-extern "C"
-{
- #include "vp8/common/onyx.h"
-}
-
-
-using namespace std;
-
-typedef map<string,int> Parms;
-
-#define ALLPARMS(O,DOTHIS) \
- DOTHIS(O, interquantizer )\
- DOTHIS(O, auto_gold )\
- DOTHIS(O, auto_adjust_gold_quantizer )\
- DOTHIS(O, goldquantizer )\
- DOTHIS(O, goldfreq )\
- DOTHIS(O, auto_key )\
- DOTHIS(O, auto_adjust_key_quantizer )\
- DOTHIS(O, keyquantizer )\
- DOTHIS(O, keyfreq )\
- DOTHIS(O, pass )\
- DOTHIS(O, fixed_q )\
- DOTHIS(O, target_bandwidth )\
- DOTHIS(O, auto_worst_q )\
- DOTHIS(O, worst_quality )\
- DOTHIS(O, best_allowed_q )\
- DOTHIS(O, end_usage )\
- DOTHIS(O, starting_buffer_level )\
- DOTHIS(O, optimal_buffer_level )\
- DOTHIS(O, maximum_buffer_size )\
- DOTHIS(O, under_shoot_pct )\
- DOTHIS(O, allow_df )\
- DOTHIS(O, drop_frames_water_mark )\
- DOTHIS(O, max_allowed_datarate )\
- DOTHIS(O, two_pass_vbrbias )\
- DOTHIS(O, two_pass_vbrmin_section )\
- DOTHIS(O, two_pass_vbrmax_section )\
- DOTHIS(O, filter_type )\
- DOTHIS(O, compressor_speed )\
- DOTHIS(O, mbpitch_feature )\
- DOTHIS(O, allow_spatial_resampling )\
- DOTHIS(O, resample_down_water_mark )\
- DOTHIS(O, resample_up_water_mark )\
- DOTHIS(O, noise_sensitivity )\
- DOTHIS(O, horiz_scale )\
- DOTHIS(O, vert_scale )
-
-
-#define GET(O,V) O->V = x[#V];
-#define PUT(O,V) x[#V] = O->V;
-
-
-extern "C" void get_parms(VP8_CONFIG *ocf,char *filename)
-{
-
- Parms x;
- int value;
- string variable;
- string equal;
-
- ifstream config_file(filename);
-
- ALLPARMS(ocf, PUT);
-
- // store all the parms in a map (really simple parsing)
- while(!config_file.eof() && config_file.is_open())
- {
- config_file >> variable;
- config_file >> equal;
-
- if(equal != "=")
- continue;
-
- config_file >> value;
-
- x[variable] = value;
- }
-
- ALLPARMS(ocf, GET);
-
-}
-
-#define PRINT(O,V) debug_file<<#V <<" = " << O->V <<"\n";
-extern "C" void print_parms(VP8_CONFIG *ocf,char *filename)
-{
- ofstream debug_file(filename,ios_base::app);
- ALLPARMS(ocf, PRINT);
- debug_file << "=============================================="<<"\n";
-}
-
-#endif
diff --git a/vp8/encoder/pickinter.c b/vp8/encoder/pickinter.c
deleted file mode 100644
index 0790d35..0000000
--- a/vp8/encoder/pickinter.c
+++ /dev/null
@@ -1,960 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <limits.h>
-#include "vpx_ports/config.h"
-#include "onyx_int.h"
-#include "modecosts.h"
-#include "encodeintra.h"
-#include "vp8/common/entropymode.h"
-#include "pickinter.h"
-#include "vp8/common/findnearmv.h"
-#include "encodemb.h"
-#include "vp8/common/reconinter.h"
-#include "vp8/common/reconintra.h"
-#include "vp8/common/reconintra4x4.h"
-#include "vp8/common/g_common.h"
-#include "variance.h"
-#include "mcomp.h"
-#include "rdopt.h"
-#include "vpx_mem/vpx_mem.h"
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IF_RTCD(x) (x)
-#else
-#define IF_RTCD(x) NULL
-#endif
-
-extern int VP8_UVSSE(MACROBLOCK *x, const vp8_variance_rtcd_vtable_t *rtcd);
-
-#ifdef SPEEDSTATS
-extern unsigned int cnt_pm;
-#endif
-
-extern const MV_REFERENCE_FRAME vp8_ref_frame_order[MAX_MODES];
-extern const MB_PREDICTION_MODE vp8_mode_order[MAX_MODES];
-
-
-extern unsigned int (*vp8_get16x16pred_error)(unsigned char *src_ptr, int src_stride, unsigned char *ref_ptr, int ref_stride);
-extern unsigned int (*vp8_get4x4sse_cs)(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride);
-extern int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, MV *best_ref_mv, int best_rd, int *, int *, int *, int, int *mvcost[2], int, int fullpixel);
-extern int vp8_cost_mv_ref(MB_PREDICTION_MODE m, const int near_mv_ref_ct[4]);
-extern void vp8_set_mbmode_and_mvs(MACROBLOCK *x, MB_PREDICTION_MODE mb, MV *mv);
-
-
-int vp8_skip_fractional_mv_step(MACROBLOCK *mb, BLOCK *b, BLOCKD *d, MV *bestmv, MV *ref_mv, int error_per_bit, const vp8_variance_fn_ptr_t *vfp, int *mvcost[2])
-{
- (void) b;
- (void) d;
- (void) ref_mv;
- (void) error_per_bit;
- (void) vfp;
- (void) mvcost;
- bestmv->row <<= 3;
- bestmv->col <<= 3;
- return 0;
-}
-
-
-static int get_inter_mbpred_error(MACROBLOCK *mb, const vp8_variance_fn_ptr_t *vfp, unsigned int *sse)
-{
-
- BLOCK *b = &mb->block[0];
- BLOCKD *d = &mb->e_mbd.block[0];
- unsigned char *what = (*(b->base_src) + b->src);
- int what_stride = b->src_stride;
- unsigned char *in_what = *(d->base_pre) + d->pre ;
- int in_what_stride = d->pre_stride;
- int xoffset = d->bmi.mv.as_mv.col & 7;
- int yoffset = d->bmi.mv.as_mv.row & 7;
-
- in_what += (d->bmi.mv.as_mv.row >> 3) * d->pre_stride + (d->bmi.mv.as_mv.col >> 3);
-
- if (xoffset | yoffset)
- {
- return vfp->svf(in_what, in_what_stride, xoffset, yoffset, what, what_stride, sse);
- }
- else
- {
- return vfp->vf(what, what_stride, in_what, in_what_stride, sse);
- }
-
-}
-
-unsigned int vp8_get16x16pred_error_c
-(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- int max_sad
-)
-{
- unsigned pred_error = 0;
- int i, j;
- int sum = 0;
-
- for (i = 0; i < 16; i++)
- {
- int diff;
-
- for (j = 0; j < 16; j++)
- {
- diff = src_ptr[j] - ref_ptr[j];
- sum += diff;
- pred_error += diff * diff;
- }
-
- src_ptr += src_stride;
- ref_ptr += ref_stride;
- }
-
- pred_error -= sum * sum / 256;
- return pred_error;
-}
-
-
-unsigned int vp8_get4x4sse_cs_c
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- int max_sad
-)
-{
- int distortion = 0;
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- int diff = src_ptr[c] - ref_ptr[c];
- distortion += diff * diff;
- }
-
- src_ptr += source_stride;
- ref_ptr += recon_stride;
- }
-
- return distortion;
-}
-
-static int get_prediction_error(BLOCK *be, BLOCKD *b, const vp8_variance_rtcd_vtable_t *rtcd)
-{
- unsigned char *sptr;
- unsigned char *dptr;
- sptr = (*(be->base_src) + be->src);
- dptr = b->predictor;
-
- return VARIANCE_INVOKE(rtcd, get4x4sse_cs)(sptr, be->src_stride, dptr, 16, 0x7fffffff);
-
-}
-
-static int pick_intra4x4block(
- const VP8_ENCODER_RTCD *rtcd,
- MACROBLOCK *x,
- BLOCK *be,
- BLOCKD *b,
- B_PREDICTION_MODE *best_mode,
- B_PREDICTION_MODE above,
- B_PREDICTION_MODE left,
-
- int *bestrate,
- int *bestdistortion)
-{
- B_PREDICTION_MODE mode;
- int best_rd = INT_MAX; // 1<<30
- int rate;
- int distortion;
- unsigned int *mode_costs;
-
- if (x->e_mbd.frame_type == KEY_FRAME)
- {
- mode_costs = x->bmode_costs[above][left];
- }
- else
- {
- mode_costs = x->inter_bmode_costs;
- }
-
- for (mode = B_DC_PRED; mode <= B_HE_PRED /*B_HU_PRED*/; mode++)
- {
- int this_rd;
-
- rate = mode_costs[mode];
- vp8_predict_intra4x4(b, mode, b->predictor);
- distortion = get_prediction_error(be, b, &rtcd->variance);
- this_rd = RD_ESTIMATE(x->rdmult, x->rddiv, rate, distortion);
-
- if (this_rd < best_rd)
- {
- *bestrate = rate;
- *bestdistortion = distortion;
- best_rd = this_rd;
- *best_mode = mode;
- }
- }
-
- b->bmi.mode = (B_PREDICTION_MODE)(*best_mode);
- vp8_encode_intra4x4block(rtcd, x, be, b, b->bmi.mode);
-
- return best_rd;
-}
-
-
-int vp8_pick_intra4x4mby_modes(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *mb, int *Rate, int *best_dist)
-{
- MACROBLOCKD *const xd = &mb->e_mbd;
- int i;
- int cost = mb->mbmode_cost [xd->frame_type] [B_PRED];
- int error;
- int distortion = 0;
-
- vp8_intra_prediction_down_copy(xd);
-
- for (i = 0; i < 16; i++)
- {
- MODE_INFO *const mic = xd->mode_info_context;
- const int mis = xd->mode_info_stride;
- const B_PREDICTION_MODE A = vp8_above_bmi(mic, i, mis)->mode;
- const B_PREDICTION_MODE L = vp8_left_bmi(mic, i)->mode;
- B_PREDICTION_MODE UNINITIALIZED_IS_SAFE(best_mode);
- int UNINITIALIZED_IS_SAFE(r), UNINITIALIZED_IS_SAFE(d);
-
- pick_intra4x4block(rtcd, mb, mb->block + i, xd->block + i,
- &best_mode, A, L, &r, &d);
-
- cost += r;
- distortion += d;
-
- mic->bmi[i].mode = xd->block[i].bmi.mode = best_mode;
-
- // Break out case where we have already exceeded best so far value that was bassed in
- if (distortion > *best_dist)
- break;
- }
-
- for (i = 0; i < 16; i++)
- xd->block[i].bmi.mv.as_int = 0;
-
- *Rate = cost;
-
- if (i == 16)
- {
- *best_dist = distortion;
- error = RD_ESTIMATE(mb->rdmult, mb->rddiv, cost, distortion);
- }
- else
- {
- *best_dist = INT_MAX;
- error = INT_MAX;
- }
-
- return error;
-}
-
-int vp8_pick_intra_mbuv_mode(MACROBLOCK *mb)
-{
-
- MACROBLOCKD *x = &mb->e_mbd;
- unsigned char *uabove_row = x->dst.u_buffer - x->dst.uv_stride;
- unsigned char *vabove_row = x->dst.v_buffer - x->dst.uv_stride;
- unsigned char *usrc_ptr = (mb->block[16].src + *mb->block[16].base_src);
- unsigned char *vsrc_ptr = (mb->block[20].src + *mb->block[20].base_src);
- int uvsrc_stride = mb->block[16].src_stride;
- unsigned char uleft_col[8];
- unsigned char vleft_col[8];
- unsigned char utop_left = uabove_row[-1];
- unsigned char vtop_left = vabove_row[-1];
- int i, j;
- int expected_udc;
- int expected_vdc;
- int shift;
- int Uaverage = 0;
- int Vaverage = 0;
- int diff;
- int pred_error[4] = {0, 0, 0, 0}, best_error = INT_MAX;
- MB_PREDICTION_MODE UNINITIALIZED_IS_SAFE(best_mode);
-
-
- for (i = 0; i < 8; i++)
- {
- uleft_col[i] = x->dst.u_buffer [i* x->dst.uv_stride -1];
- vleft_col[i] = x->dst.v_buffer [i* x->dst.uv_stride -1];
- }
-
- if (!x->up_available && !x->left_available)
- {
- expected_udc = 128;
- expected_vdc = 128;
- }
- else
- {
- shift = 2;
-
- if (x->up_available)
- {
-
- for (i = 0; i < 8; i++)
- {
- Uaverage += uabove_row[i];
- Vaverage += vabove_row[i];
- }
-
- shift ++;
-
- }
-
- if (x->left_available)
- {
- for (i = 0; i < 8; i++)
- {
- Uaverage += uleft_col[i];
- Vaverage += vleft_col[i];
- }
-
- shift ++;
-
- }
-
- expected_udc = (Uaverage + (1 << (shift - 1))) >> shift;
- expected_vdc = (Vaverage + (1 << (shift - 1))) >> shift;
- }
-
-
- for (i = 0; i < 8; i++)
- {
- for (j = 0; j < 8; j++)
- {
-
- int predu = uleft_col[i] + uabove_row[j] - utop_left;
- int predv = vleft_col[i] + vabove_row[j] - vtop_left;
- int u_p, v_p;
-
- u_p = usrc_ptr[j];
- v_p = vsrc_ptr[j];
-
- if (predu < 0)
- predu = 0;
-
- if (predu > 255)
- predu = 255;
-
- if (predv < 0)
- predv = 0;
-
- if (predv > 255)
- predv = 255;
-
-
- diff = u_p - expected_udc;
- pred_error[DC_PRED] += diff * diff;
- diff = v_p - expected_vdc;
- pred_error[DC_PRED] += diff * diff;
-
-
- diff = u_p - uabove_row[j];
- pred_error[V_PRED] += diff * diff;
- diff = v_p - vabove_row[j];
- pred_error[V_PRED] += diff * diff;
-
-
- diff = u_p - uleft_col[i];
- pred_error[H_PRED] += diff * diff;
- diff = v_p - vleft_col[i];
- pred_error[H_PRED] += diff * diff;
-
-
- diff = u_p - predu;
- pred_error[TM_PRED] += diff * diff;
- diff = v_p - predv;
- pred_error[TM_PRED] += diff * diff;
-
-
- }
-
- usrc_ptr += uvsrc_stride;
- vsrc_ptr += uvsrc_stride;
-
- if (i == 3)
- {
- usrc_ptr = (mb->block[18].src + *mb->block[18].base_src);
- vsrc_ptr = (mb->block[22].src + *mb->block[22].base_src);
- }
-
-
-
- }
-
-
- for (i = DC_PRED; i <= TM_PRED; i++)
- {
- if (best_error > pred_error[i])
- {
- best_error = pred_error[i];
- best_mode = (MB_PREDICTION_MODE)i;
- }
- }
-
-
- mb->e_mbd.mode_info_context->mbmi.uv_mode = best_mode;
- return best_error;
-
-}
-
-int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra)
-{
- BLOCK *b = &x->block[0];
- BLOCKD *d = &x->e_mbd.block[0];
- MACROBLOCKD *xd = &x->e_mbd;
- B_MODE_INFO best_bmodes[16];
- MB_MODE_INFO best_mbmode;
- PARTITION_INFO best_partition;
- MV best_ref_mv;
- MV mode_mv[MB_MODE_COUNT];
- MB_PREDICTION_MODE this_mode;
- int num00;
- int i;
- int mdcounts[4];
- int best_rd = INT_MAX; // 1 << 30;
- int best_intra_rd = INT_MAX;
- int mode_index;
- int ref_frame_cost[MAX_REF_FRAMES];
- int rate;
- int rate2;
- int distortion2;
- int bestsme;
- //int all_rds[MAX_MODES]; // Experimental debug code.
- int best_mode_index = 0;
- int sse = INT_MAX;
-
- MV mvp;
- int near_sadidx[8] = {0, 1, 2, 3, 4, 5, 6, 7};
- int saddone=0;
- int sr=0; //search range got from mv_pred(). It uses step_param levels. (0-7)
-
- MV nearest_mv[4];
- MV near_mv[4];
- MV frame_best_ref_mv[4];
- int MDCounts[4][4];
- unsigned char *y_buffer[4];
- unsigned char *u_buffer[4];
- unsigned char *v_buffer[4];
-
- int skip_mode[4] = {0, 0, 0, 0};
-
- vpx_memset(mode_mv, 0, sizeof(mode_mv));
- vpx_memset(nearest_mv, 0, sizeof(nearest_mv));
- vpx_memset(near_mv, 0, sizeof(near_mv));
- vpx_memset(&best_mbmode, 0, sizeof(best_mbmode));
-
-
- // set up all the refframe dependent pointers.
- if (cpi->ref_frame_flags & VP8_LAST_FLAG)
- {
- YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx];
-
- vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[LAST_FRAME], &near_mv[LAST_FRAME],
- &frame_best_ref_mv[LAST_FRAME], MDCounts[LAST_FRAME], LAST_FRAME, cpi->common.ref_frame_sign_bias);
-
- y_buffer[LAST_FRAME] = lst_yv12->y_buffer + recon_yoffset;
- u_buffer[LAST_FRAME] = lst_yv12->u_buffer + recon_uvoffset;
- v_buffer[LAST_FRAME] = lst_yv12->v_buffer + recon_uvoffset;
- }
- else
- skip_mode[LAST_FRAME] = 1;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- YV12_BUFFER_CONFIG *gld_yv12 = &cpi->common.yv12_fb[cpi->common.gld_fb_idx];
-
- vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[GOLDEN_FRAME], &near_mv[GOLDEN_FRAME],
- &frame_best_ref_mv[GOLDEN_FRAME], MDCounts[GOLDEN_FRAME], GOLDEN_FRAME, cpi->common.ref_frame_sign_bias);
-
- y_buffer[GOLDEN_FRAME] = gld_yv12->y_buffer + recon_yoffset;
- u_buffer[GOLDEN_FRAME] = gld_yv12->u_buffer + recon_uvoffset;
- v_buffer[GOLDEN_FRAME] = gld_yv12->v_buffer + recon_uvoffset;
- }
- else
- skip_mode[GOLDEN_FRAME] = 1;
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG && cpi->source_alt_ref_active)
- {
- YV12_BUFFER_CONFIG *alt_yv12 = &cpi->common.yv12_fb[cpi->common.alt_fb_idx];
-
- vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[ALTREF_FRAME], &near_mv[ALTREF_FRAME],
- &frame_best_ref_mv[ALTREF_FRAME], MDCounts[ALTREF_FRAME], ALTREF_FRAME, cpi->common.ref_frame_sign_bias);
-
- y_buffer[ALTREF_FRAME] = alt_yv12->y_buffer + recon_yoffset;
- u_buffer[ALTREF_FRAME] = alt_yv12->u_buffer + recon_uvoffset;
- v_buffer[ALTREF_FRAME] = alt_yv12->v_buffer + recon_uvoffset;
- }
- else
- skip_mode[ALTREF_FRAME] = 1;
-
- cpi->mbs_tested_so_far++; // Count of the number of MBs tested so far this frame
-
- *returnintra = best_intra_rd;
- x->skip = 0;
-
- ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(cpi->prob_intra_coded);
-
- // Special case treatment when GF and ARF are not sensible options for reference
- if (cpi->ref_frame_flags == VP8_LAST_FLAG)
- {
- ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(255);
- ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(255)
- + vp8_cost_zero(128);
- ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(255)
- + vp8_cost_one(128);
- }
- else
- {
- ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(cpi->prob_last_coded);
- ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_zero(cpi->prob_gf_coded);
- ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_one(cpi->prob_gf_coded);
- }
-
- x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
-
- // if we encode a new mv this is important
- // find the best new motion vector
- for (mode_index = 0; mode_index < MAX_MODES; mode_index++)
- {
- int frame_cost;
- int this_rd = INT_MAX;
-
- if (best_rd <= cpi->rd_threshes[mode_index])
- continue;
-
- x->e_mbd.mode_info_context->mbmi.ref_frame = vp8_ref_frame_order[mode_index];
-
- if (skip_mode[x->e_mbd.mode_info_context->mbmi.ref_frame])
- continue;
-
- // Check to see if the testing frequency for this mode is at its max
- // If so then prevent it from being tested and increase the threshold for its testing
- if (cpi->mode_test_hit_counts[mode_index] && (cpi->mode_check_freq[mode_index] > 1))
- {
- //if ( (cpi->mbs_tested_so_far / cpi->mode_test_hit_counts[mode_index]) <= cpi->mode_check_freq[mode_index] )
- if (cpi->mbs_tested_so_far <= (cpi->mode_check_freq[mode_index] * cpi->mode_test_hit_counts[mode_index]))
- {
- // Increase the threshold for coding this mode to make it less likely to be chosen
- cpi->rd_thresh_mult[mode_index] += 4;
-
- if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
- cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
-
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
-
- continue;
- }
- }
-
- // We have now reached the point where we are going to test the current mode so increment the counter for the number of times it has been tested
- cpi->mode_test_hit_counts[mode_index] ++;
-
- rate2 = 0;
- distortion2 = 0;
-
- this_mode = vp8_mode_order[mode_index];
-
- // Experimental debug code.
- //all_rds[mode_index] = -1;
-
- x->e_mbd.mode_info_context->mbmi.mode = this_mode;
- x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
-
- // Work out the cost assosciated with selecting the reference frame
- frame_cost = ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
- rate2 += frame_cost;
-
- // everything but intra
- if (x->e_mbd.mode_info_context->mbmi.ref_frame)
- {
- x->e_mbd.pre.y_buffer = y_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- x->e_mbd.pre.u_buffer = u_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- x->e_mbd.pre.v_buffer = v_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- mode_mv[NEARESTMV] = nearest_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
- mode_mv[NEARMV] = near_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
- best_ref_mv = frame_best_ref_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
- memcpy(mdcounts, MDCounts[x->e_mbd.mode_info_context->mbmi.ref_frame], sizeof(mdcounts));
- }
-
- // Only consider ZEROMV/ALTREF_FRAME for alt ref frame,
- // unless ARNR filtering is enabled in which case we want
- // an unfiltered alternative
- if (cpi->is_src_frame_alt_ref && (cpi->oxcf.arnr_max_frames == 0))
- {
- if (this_mode != ZEROMV || x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME)
- continue;
- }
-
- if(cpi->sf.improved_mv_pred && x->e_mbd.mode_info_context->mbmi.mode == NEWMV)
- {
- if(!saddone)
- {
- vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] );
- saddone = 1;
- }
-
- vp8_mv_pred(cpi, &x->e_mbd, x->e_mbd.mode_info_context, &mvp,
- x->e_mbd.mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias, &sr, &near_sadidx[0]);
-
- /* adjust mvp to make sure it is within MV range */
- if(mvp.row > best_ref_mv.row + MAX_FULL_PEL_VAL)
- mvp.row = best_ref_mv.row + MAX_FULL_PEL_VAL;
- else if(mvp.row < best_ref_mv.row - MAX_FULL_PEL_VAL)
- mvp.row = best_ref_mv.row - MAX_FULL_PEL_VAL;
- if(mvp.col > best_ref_mv.col + MAX_FULL_PEL_VAL)
- mvp.col = best_ref_mv.col + MAX_FULL_PEL_VAL;
- else if(mvp.col < best_ref_mv.col - MAX_FULL_PEL_VAL)
- mvp.col = best_ref_mv.col - MAX_FULL_PEL_VAL;
- }
-
- switch (this_mode)
- {
- case B_PRED:
- distortion2 = *returndistortion; // Best so far passed in as breakout value to vp8_pick_intra4x4mby_modes
- vp8_pick_intra4x4mby_modes(IF_RTCD(&cpi->rtcd), x, &rate, &distortion2);
- rate2 += rate;
- distortion2 = VARIANCE_INVOKE(&cpi->rtcd.variance, get16x16prederror)(x->src.y_buffer, x->src.y_stride, x->e_mbd.predictor, 16, 0x7fffffff);
-
- if (distortion2 == INT_MAX)
- {
- this_rd = INT_MAX;
- }
- else
- {
- this_rd = RD_ESTIMATE(x->rdmult, x->rddiv, rate2, distortion2);
-
- if (this_rd < best_intra_rd)
- {
- best_intra_rd = this_rd;
- *returnintra = best_intra_rd ;
- }
- }
-
- break;
-
- case SPLITMV:
-
- // Split MV modes currently not supported when RD is nopt enabled.
- break;
-
- case DC_PRED:
- case V_PRED:
- case H_PRED:
- case TM_PRED:
- RECON_INVOKE(&cpi->common.rtcd.recon, build_intra_predictors_mby)
- (&x->e_mbd);
- distortion2 = VARIANCE_INVOKE(&cpi->rtcd.variance, get16x16prederror)(x->src.y_buffer, x->src.y_stride, x->e_mbd.predictor, 16, 0x7fffffff);
- rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode];
- this_rd = RD_ESTIMATE(x->rdmult, x->rddiv, rate2, distortion2);
-
- if (this_rd < best_intra_rd)
- {
- best_intra_rd = this_rd;
- *returnintra = best_intra_rd ;
- }
-
- break;
-
- case NEWMV:
- {
- int thissme;
- int step_param;
- int further_steps;
- int n = 0;
- int sadpb = x->sadperbit16;
-
- int col_min;
- int col_max;
- int row_min;
- int row_max;
-
- int tmp_col_min = x->mv_col_min;
- int tmp_col_max = x->mv_col_max;
- int tmp_row_min = x->mv_row_min;
- int tmp_row_max = x->mv_row_max;
-
- int speed_adjust = (cpi->Speed > 5) ? ((cpi->Speed >= 8)? 3 : 2) : 1;
-
- // Further step/diamond searches as necessary
- step_param = cpi->sf.first_step + speed_adjust;
-
- if(cpi->sf.improved_mv_pred)
- {
- sr += speed_adjust;
- //adjust search range according to sr from mv prediction
- if(sr > step_param)
- step_param = sr;
-
- col_min = (best_ref_mv.col - MAX_FULL_PEL_VAL) >>3;
- col_max = (best_ref_mv.col + MAX_FULL_PEL_VAL) >>3;
- row_min = (best_ref_mv.row - MAX_FULL_PEL_VAL) >>3;
- row_max = (best_ref_mv.row + MAX_FULL_PEL_VAL) >>3;
-
- // Get intersection of UMV window and valid MV window to reduce # of checks in diamond search.
- if (x->mv_col_min < col_min )
- x->mv_col_min = col_min;
- if (x->mv_col_max > col_max )
- x->mv_col_max = col_max;
- if (x->mv_row_min < row_min )
- x->mv_row_min = row_min;
- if (x->mv_row_max > row_max )
- x->mv_row_max = row_max;
- }else
- {
- mvp.row = best_ref_mv.row;
- mvp.col = best_ref_mv.col;
- }
-
- further_steps = (cpi->Speed >= 8)? 0: (cpi->sf.max_step_search_steps - 1 - step_param);
-
- if (cpi->sf.search_method == HEX)
- {
- bestsme = vp8_hex_search(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param, sadpb/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv);
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
- }
- else
- {
- bestsme = cpi->diamond_search_sad(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param, sadpb / 2/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); //sadpb < 9
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
-
- // Further step/diamond searches as necessary
- n = 0;
- //further_steps = (cpi->sf.max_step_search_steps - 1) - step_param;
-
- n = num00;
- num00 = 0;
-
- while (n < further_steps)
- {
- n++;
-
- if (num00)
- num00--;
- else
- {
- thissme = cpi->diamond_search_sad(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param + n, sadpb / 4/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); //sadpb = 9
-
- if (thissme < bestsme)
- {
- bestsme = thissme;
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
- }
- else
- {
- d->bmi.mv.as_mv.row = mode_mv[NEWMV].row;
- d->bmi.mv.as_mv.col = mode_mv[NEWMV].col;
- }
- }
- }
- }
-
- if(cpi->sf.improved_mv_pred)
- {
- x->mv_col_min = tmp_col_min;
- x->mv_col_max = tmp_col_max;
- x->mv_row_min = tmp_row_min;
- x->mv_row_max = tmp_row_max;
- }
-
- if (bestsme < INT_MAX)
- cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv.as_mv, &best_ref_mv, x->errorperbit, &cpi->fn_ptr[BLOCK_16X16], cpi->mb.mvcost);
-
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
-
- // mv cost;
- rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv, cpi->mb.mvcost, 128);
- }
-
- case NEARESTMV:
- case NEARMV:
-
- if (mode_mv[this_mode].row == 0 && mode_mv[this_mode].col == 0)
- continue;
-
- case ZEROMV:
-
- // Trap vectors that reach beyond the UMV borders
- // Note that ALL New MV, Nearest MV Near MV and Zero MV code drops through to this point
- // because of the lack of break statements in the previous two cases.
- if (((mode_mv[this_mode].row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].row >> 3) > x->mv_row_max) ||
- ((mode_mv[this_mode].col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].col >> 3) > x->mv_col_max))
- continue;
-
- rate2 += vp8_cost_mv_ref(this_mode, mdcounts);
- x->e_mbd.mode_info_context->mbmi.mode = this_mode;
- x->e_mbd.mode_info_context->mbmi.mv.as_mv = mode_mv[this_mode];
- x->e_mbd.block[0].bmi.mode = this_mode;
- x->e_mbd.block[0].bmi.mv.as_int = x->e_mbd.mode_info_context->mbmi.mv.as_int;
-
- distortion2 = get_inter_mbpred_error(x, &cpi->fn_ptr[BLOCK_16X16], (unsigned int *)(&sse));
-
- this_rd = RD_ESTIMATE(x->rdmult, x->rddiv, rate2, distortion2);
-
- if (cpi->active_map_enabled && x->active_ptr[0] == 0)
- {
- x->skip = 1;
- }
- else if (sse < x->encode_breakout)
- {
- // Check u and v to make sure skip is ok
- int sse2 = 0;
-
- sse2 = VP8_UVSSE(x, IF_RTCD(&cpi->rtcd.variance));
-
- if (sse2 * 2 < x->encode_breakout)
- x->skip = 1;
- else
- x->skip = 0;
- }
-
- break;
- default:
- break;
- }
-
- // Experimental debug code.
- //all_rds[mode_index] = this_rd;
-
- if (this_rd < best_rd || x->skip)
- {
- // Note index of best mode
- best_mode_index = mode_index;
-
- *returnrate = rate2;
- *returndistortion = distortion2;
- best_rd = this_rd;
- vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi, sizeof(MB_MODE_INFO));
- vpx_memcpy(&best_partition, x->partition_info, sizeof(PARTITION_INFO));
-
- if (this_mode == B_PRED || this_mode == SPLITMV)
- for (i = 0; i < 16; i++)
- {
- vpx_memcpy(&best_bmodes[i], &x->e_mbd.block[i].bmi, sizeof(B_MODE_INFO));
- }
- else
- {
- best_bmodes[0].mv = x->e_mbd.block[0].bmi.mv;
- }
-
- // Testing this mode gave rise to an improvement in best error score. Lower threshold a bit for next time
- cpi->rd_thresh_mult[mode_index] = (cpi->rd_thresh_mult[mode_index] >= (MIN_THRESHMULT + 2)) ? cpi->rd_thresh_mult[mode_index] - 2 : MIN_THRESHMULT;
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
- }
-
- // If the mode did not help improve the best error case then raise the threshold for testing that mode next time around.
- else
- {
- cpi->rd_thresh_mult[mode_index] += 4;
-
- if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
- cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
-
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
- }
-
- if (x->skip)
- break;
- }
-
- // Reduce the activation RD thresholds for the best choice mode
- if ((cpi->rd_baseline_thresh[best_mode_index] > 0) && (cpi->rd_baseline_thresh[best_mode_index] < (INT_MAX >> 2)))
- {
- int best_adjustment = (cpi->rd_thresh_mult[best_mode_index] >> 3);
-
- cpi->rd_thresh_mult[best_mode_index] = (cpi->rd_thresh_mult[best_mode_index] >= (MIN_THRESHMULT + best_adjustment)) ? cpi->rd_thresh_mult[best_mode_index] - best_adjustment : MIN_THRESHMULT;
- cpi->rd_threshes[best_mode_index] = (cpi->rd_baseline_thresh[best_mode_index] >> 7) * cpi->rd_thresh_mult[best_mode_index];
- }
-
- // Keep a record of best mode index for use in next loop
- cpi->last_best_mode_index = best_mode_index;
-
- if (best_mbmode.mode <= B_PRED)
- {
- x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
- vp8_pick_intra_mbuv_mode(x);
- best_mbmode.uv_mode = x->e_mbd.mode_info_context->mbmi.uv_mode;
- }
-
-
- {
- int this_rdbin = (*returndistortion >> 7);
-
- if (this_rdbin >= 1024)
- {
- this_rdbin = 1023;
- }
-
- cpi->error_bins[this_rdbin] ++;
- }
-
-
- if (cpi->is_src_frame_alt_ref && (best_mbmode.mode != ZEROMV || best_mbmode.ref_frame != ALTREF_FRAME))
- {
- best_mbmode.mode = ZEROMV;
- best_mbmode.ref_frame = ALTREF_FRAME;
- best_mbmode.mv.as_int = 0;
- best_mbmode.uv_mode = 0;
- best_mbmode.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0;
- best_mbmode.partitioning = 0;
- best_mbmode.dc_diff = 0;
-
- vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO));
- vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO));
-
- for (i = 0; i < 16; i++)
- {
- vpx_memset(&x->e_mbd.block[i].bmi, 0, sizeof(B_MODE_INFO));
- }
-
- x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
-
- return best_rd;
- }
-
-
- // macroblock modes
- vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO));
- vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO));
-
- if (x->e_mbd.mode_info_context->mbmi.mode == B_PRED || x->e_mbd.mode_info_context->mbmi.mode == SPLITMV)
- for (i = 0; i < 16; i++)
- {
- vpx_memcpy(&x->e_mbd.block[i].bmi, &best_bmodes[i], sizeof(B_MODE_INFO));
-
- }
- else
- {
- vp8_set_mbmode_and_mvs(x, x->e_mbd.mode_info_context->mbmi.mode, &best_bmodes[0].mv.as_mv);
- }
-
- x->e_mbd.mode_info_context->mbmi.mv.as_mv = x->e_mbd.block[15].bmi.mv.as_mv;
-
- return best_rd;
-}
diff --git a/vp8/encoder/pickinter.h b/vp8/encoder/pickinter.h
deleted file mode 100644
index af060d7..0000000
--- a/vp8/encoder/pickinter.h
+++ /dev/null
@@ -1,21 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_PICKINTER_H
-#define __INC_PICKINTER_H
-#include "vpx_ports/config.h"
-#include "vp8/common/onyxc_int.h"
-
-#define RD_ESTIMATE(RM,DM,R,D) ( ((128+(R)*(RM)) >> 8) + (DM)*(D) )
-extern int vp8_pick_intra4x4mby_modes(const VP8_ENCODER_RTCD *, MACROBLOCK *mb, int *Rate, int *Distortion);
-extern int vp8_pick_intra_mbuv_mode(MACROBLOCK *mb);
-extern int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra);
-#endif
diff --git a/vp8/encoder/picklpf.c b/vp8/encoder/picklpf.c
deleted file mode 100644
index d294af6..0000000
--- a/vp8/encoder/picklpf.c
+++ /dev/null
@@ -1,481 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/onyxc_int.h"
-#include "onyx_int.h"
-#include "quantize.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vpx_scale/yv12extend.h"
-#include "vpx_scale/vpxscale.h"
-#include "vp8/common/alloccommon.h"
-#if ARCH_ARM
-#include "vpx_ports/arm.h"
-#endif
-
-extern void vp8_loop_filter_frame(VP8_COMMON *cm, MACROBLOCKD *mbd, int filt_val);
-extern void vp8_loop_filter_frame_yonly(VP8_COMMON *cm, MACROBLOCKD *mbd, int filt_val, int sharpness_lvl);
-extern int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, const vp8_variance_rtcd_vtable_t *rtcd);
-#if HAVE_ARMV7
-extern void vp8_yv12_copy_frame_yonly_no_extend_frame_borders_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-#endif
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IF_RTCD(x) (x)
-#else
-#define IF_RTCD(x) NULL
-#endif
-
-extern void
-(*vp8_yv12_copy_partial_frame_ptr)(YV12_BUFFER_CONFIG *src_ybc,
- YV12_BUFFER_CONFIG *dst_ybc,
- int Fraction);
-void
-vp8_yv12_copy_partial_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc, int Fraction)
-{
- unsigned char *src_y, *dst_y;
- int yheight;
- int ystride;
- int border;
- int yoffset;
- int linestocopy;
-
- border = src_ybc->border;
- yheight = src_ybc->y_height;
- ystride = src_ybc->y_stride;
-
- linestocopy = (yheight >> (Fraction + 4));
-
- if (linestocopy < 1)
- linestocopy = 1;
-
- linestocopy <<= 4;
-
- yoffset = ystride * ((yheight >> 5) * 16 - 8);
- src_y = src_ybc->y_buffer + yoffset;
- dst_y = dst_ybc->y_buffer + yoffset;
-
- vpx_memcpy(dst_y, src_y, ystride *(linestocopy + 16));
-}
-
-static int vp8_calc_partial_ssl_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, int Fraction, const vp8_variance_rtcd_vtable_t *rtcd)
-{
- int i, j;
- int Total = 0;
- int srcoffset, dstoffset;
- unsigned char *src = source->y_buffer;
- unsigned char *dst = dest->y_buffer;
-
- int linestocopy = (source->y_height >> (Fraction + 4));
- (void)rtcd;
-
- if (linestocopy < 1)
- linestocopy = 1;
-
- linestocopy <<= 4;
-
-
- srcoffset = source->y_stride * (dest->y_height >> 5) * 16;
- dstoffset = dest->y_stride * (dest->y_height >> 5) * 16;
-
- src += srcoffset;
- dst += dstoffset;
-
- // Loop through the Y plane raw and reconstruction data summing (square differences)
- for (i = 0; i < linestocopy; i += 16)
- {
- for (j = 0; j < source->y_width; j += 16)
- {
- unsigned int sse;
- Total += VARIANCE_INVOKE(rtcd, mse16x16)(src + j, source->y_stride, dst + j, dest->y_stride, &sse);
- }
-
- src += 16 * source->y_stride;
- dst += 16 * dest->y_stride;
- }
-
- return Total;
-}
-
-extern void vp8_loop_filter_partial_frame
-(
- VP8_COMMON *cm,
- MACROBLOCKD *mbd,
- int default_filt_lvl,
- int sharpness_lvl,
- int Fraction
-);
-
-// Enforce a minimum filter level based upon baseline Q
-static int get_min_filter_level(VP8_COMP *cpi, int base_qindex)
-{
- int min_filter_level;
-
- if (cpi->source_alt_ref_active && cpi->common.refresh_golden_frame && !cpi->common.refresh_alt_ref_frame)
- min_filter_level = 0;
- else
- {
- if (base_qindex <= 6)
- min_filter_level = 0;
- else if (base_qindex <= 16)
- min_filter_level = 1;
- else
- min_filter_level = (base_qindex / 8);
- }
-
- return min_filter_level;
-}
-
-// Enforce a maximum filter level based upon baseline Q
-static int get_max_filter_level(VP8_COMP *cpi, int base_qindex)
-{
- // PGW August 2006: Highest filter values almost always a bad idea
-
- // jbb chg: 20100118 - not so any more with this overquant stuff allow high values
- // with lots of intra coming in.
- int max_filter_level = MAX_LOOP_FILTER ;//* 3 / 4;
-
- if (cpi->section_intra_rating > 8)
- max_filter_level = MAX_LOOP_FILTER * 3 / 4;
-
- (void) cpi;
- (void) base_qindex;
-
- return max_filter_level;
-}
-
-void vp8cx_pick_filter_level_fast(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
-
- int best_err = 0;
- int filt_err = 0;
- int min_filter_level = 0;
- int max_filter_level = MAX_LOOP_FILTER * 3 / 4; // PGW August 2006: Highest filter values almost always a bad idea
- int filt_val;
- int best_filt_val = cm->filter_level;
-
- // Make a copy of the unfiltered / processed recon buffer
- //vp8_yv12_copy_frame_ptr( cm->frame_to_show, &cpi->last_frame_uf );
- vp8_yv12_copy_partial_frame_ptr(cm->frame_to_show, &cpi->last_frame_uf, 3);
-
- if (cm->frame_type == KEY_FRAME)
- cm->sharpness_level = 0;
- else
- cm->sharpness_level = cpi->oxcf.Sharpness;
-
- // Enforce a minimum filter level based upon Q
- min_filter_level = get_min_filter_level(cpi, cm->base_qindex);
- max_filter_level = get_max_filter_level(cpi, cm->base_qindex);
-
- // Start the search at the previous frame filter level unless it is now out of range.
- if (cm->filter_level < min_filter_level)
- cm->filter_level = min_filter_level;
- else if (cm->filter_level > max_filter_level)
- cm->filter_level = max_filter_level;
-
- filt_val = cm->filter_level;
- best_filt_val = filt_val;
-
- // Set up alternate filter values
-
- // Get the err using the previous frame's filter value.
- vp8_loop_filter_partial_frame(cm, &cpi->mb.e_mbd, filt_val, 0 , 3);
- cm->last_frame_type = cm->frame_type;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
-
- best_err = vp8_calc_partial_ssl_err(sd, cm->frame_to_show, 3, IF_RTCD(&cpi->rtcd.variance));
-
- // Re-instate the unfiltered frame
- vp8_yv12_copy_partial_frame_ptr(&cpi->last_frame_uf, cm->frame_to_show, 3);
-
- filt_val -= (1 + ((filt_val > 10) ? 1 : 0));
-
- // Search lower filter levels
- while (filt_val >= min_filter_level)
- {
- // Apply the loop filter
- vp8_loop_filter_partial_frame(cm, &cpi->mb.e_mbd, filt_val, 0, 3);
- cm->last_frame_type = cm->frame_type;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
-
- // Get the err for filtered frame
- filt_err = vp8_calc_partial_ssl_err(sd, cm->frame_to_show, 3, IF_RTCD(&cpi->rtcd.variance));
-
-
- // Re-instate the unfiltered frame
- vp8_yv12_copy_partial_frame_ptr(&cpi->last_frame_uf, cm->frame_to_show, 3);
-
-
- // Update the best case record or exit loop.
- if (filt_err < best_err)
- {
- best_err = filt_err;
- best_filt_val = filt_val;
- }
- else
- break;
-
- // Adjust filter level
- filt_val -= (1 + ((filt_val > 10) ? 1 : 0));
- }
-
- // Search up (note that we have already done filt_val = cm->filter_level)
- filt_val = cm->filter_level + (1 + ((filt_val > 10) ? 1 : 0));
-
- if (best_filt_val == cm->filter_level)
- {
- // Resist raising filter level for very small gains
- best_err -= (best_err >> 10);
-
- while (filt_val < max_filter_level)
- {
- // Apply the loop filter
- vp8_loop_filter_partial_frame(cm, &cpi->mb.e_mbd, filt_val, 0, 3);
- cm->last_frame_type = cm->frame_type;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
-
- // Get the err for filtered frame
- filt_err = vp8_calc_partial_ssl_err(sd, cm->frame_to_show, 3, IF_RTCD(&cpi->rtcd.variance));
-
- // Re-instate the unfiltered frame
- vp8_yv12_copy_partial_frame_ptr(&cpi->last_frame_uf, cm->frame_to_show, 3);
-
- // Update the best case record or exit loop.
- if (filt_err < best_err)
- {
- // Do not raise filter level if improvement is < 1 part in 4096
- best_err = filt_err - (filt_err >> 10);
-
- best_filt_val = filt_val;
- }
- else
- break;
-
- // Adjust filter level
- filt_val += (1 + ((filt_val > 10) ? 1 : 0));
- }
- }
-
- cm->filter_level = best_filt_val;
-
- if (cm->filter_level < min_filter_level)
- cm->filter_level = min_filter_level;
-
- if (cm->filter_level > max_filter_level)
- cm->filter_level = max_filter_level;
-}
-
-// Stub function for now Alt LF not used
-void vp8cx_set_alt_lf_level(VP8_COMP *cpi, int filt_val)
-{
- MACROBLOCKD *mbd = &cpi->mb.e_mbd;
- (void) filt_val;
-
- mbd->segment_feature_data[MB_LVL_ALT_LF][0] = cpi->segment_feature_data[MB_LVL_ALT_LF][0];
- mbd->segment_feature_data[MB_LVL_ALT_LF][1] = cpi->segment_feature_data[MB_LVL_ALT_LF][1];
- mbd->segment_feature_data[MB_LVL_ALT_LF][2] = cpi->segment_feature_data[MB_LVL_ALT_LF][2];
- mbd->segment_feature_data[MB_LVL_ALT_LF][3] = cpi->segment_feature_data[MB_LVL_ALT_LF][3];
-}
-
-void vp8cx_pick_filter_level(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi)
-{
- VP8_COMMON *cm = &cpi->common;
-
- int best_err = 0;
- int filt_err = 0;
- int min_filter_level;
- int max_filter_level;
-
- int filter_step;
- int filt_high = 0;
- int filt_mid = cm->filter_level; // Start search at previous frame filter level
- int filt_low = 0;
- int filt_best;
- int filt_direction = 0;
-
- int Bias = 0; // Bias against raising loop filter and in favour of lowering it
-
- // Make a copy of the unfiltered / processed recon buffer
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_yv12_copy_frame_yonly_no_extend_frame_borders_neon(cm->frame_to_show, &cpi->last_frame_uf);
- }
-#if CONFIG_RUNTIME_CPU_DETECT
- else
-#endif
-#endif
-#if !HAVE_ARMV7 || CONFIG_RUNTIME_CPU_DETECT
- {
- vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cpi->last_frame_uf);
- }
-#endif
-
- if (cm->frame_type == KEY_FRAME)
- cm->sharpness_level = 0;
- else
- cm->sharpness_level = cpi->oxcf.Sharpness;
-
- // Enforce a minimum filter level based upon Q
- min_filter_level = get_min_filter_level(cpi, cm->base_qindex);
- max_filter_level = get_max_filter_level(cpi, cm->base_qindex);
-
- // Start the search at the previous frame filter level unless it is now out of range.
- filt_mid = cm->filter_level;
-
- if (filt_mid < min_filter_level)
- filt_mid = min_filter_level;
- else if (filt_mid > max_filter_level)
- filt_mid = max_filter_level;
-
- // Define the initial step size
- filter_step = (filt_mid < 16) ? 4 : filt_mid / 4;
-
- // Get baseline error score
- vp8cx_set_alt_lf_level(cpi, filt_mid);
- vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_mid, 0);
- cm->last_frame_type = cm->frame_type;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
-
- best_err = vp8_calc_ss_err(sd, cm->frame_to_show, IF_RTCD(&cpi->rtcd.variance));
- filt_best = filt_mid;
-
- // Re-instate the unfiltered frame
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_yv12_copy_frame_yonly_no_extend_frame_borders_neon(&cpi->last_frame_uf, cm->frame_to_show);
- }
-#if CONFIG_RUNTIME_CPU_DETECT
- else
-#endif
-#endif
-#if !HAVE_ARMV7 || CONFIG_RUNTIME_CPU_DETECT
- {
- vp8_yv12_copy_frame_yonly_ptr(&cpi->last_frame_uf, cm->frame_to_show);
- }
-#endif
-
- while (filter_step > 0)
- {
- Bias = (best_err >> (15 - (filt_mid / 8))) * filter_step; //PGW change 12/12/06 for small images
-
- // jbb chg: 20100118 - in sections with lots of new material coming in don't bias as much to a low filter value
- if (cpi->section_intra_rating < 20)
- Bias = Bias * cpi->section_intra_rating / 20;
-
- filt_high = ((filt_mid + filter_step) > max_filter_level) ? max_filter_level : (filt_mid + filter_step);
- filt_low = ((filt_mid - filter_step) < min_filter_level) ? min_filter_level : (filt_mid - filter_step);
-
- if ((filt_direction <= 0) && (filt_low != filt_mid))
- {
- // Get Low filter error score
- vp8cx_set_alt_lf_level(cpi, filt_low);
- vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_low, 0);
- cm->last_frame_type = cm->frame_type;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
-
- filt_err = vp8_calc_ss_err(sd, cm->frame_to_show, IF_RTCD(&cpi->rtcd.variance));
-
- // Re-instate the unfiltered frame
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_yv12_copy_frame_yonly_no_extend_frame_borders_neon(&cpi->last_frame_uf, cm->frame_to_show);
- }
-#if CONFIG_RUNTIME_CPU_DETECT
- else
-#endif
-#endif
-#if !HAVE_ARMV7 || CONFIG_RUNTIME_CPU_DETECT
- {
- vp8_yv12_copy_frame_yonly_ptr(&cpi->last_frame_uf, cm->frame_to_show);
- }
-#endif
-
- // If value is close to the best so far then bias towards a lower loop filter value.
- if ((filt_err - Bias) < best_err)
- {
- // Was it actually better than the previous best?
- if (filt_err < best_err)
- best_err = filt_err;
-
- filt_best = filt_low;
- }
- }
-
- // Now look at filt_high
- if ((filt_direction >= 0) && (filt_high != filt_mid))
- {
- vp8cx_set_alt_lf_level(cpi, filt_high);
- vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_high, 0);
- cm->last_frame_type = cm->frame_type;
- cm->last_filter_type = cm->filter_type;
- cm->last_sharpness_level = cm->sharpness_level;
-
- filt_err = vp8_calc_ss_err(sd, cm->frame_to_show, IF_RTCD(&cpi->rtcd.variance));
-
- // Re-instate the unfiltered frame
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- if (cm->rtcd.flags & HAS_NEON)
-#endif
- {
- vp8_yv12_copy_frame_yonly_no_extend_frame_borders_neon(&cpi->last_frame_uf, cm->frame_to_show);
- }
-#if CONFIG_RUNTIME_CPU_DETECT
- else
-#endif
-#endif
-#if !HAVE_ARMV7 || CONFIG_RUNTIME_CPU_DETECT
- {
- vp8_yv12_copy_frame_yonly_ptr(&cpi->last_frame_uf, cm->frame_to_show);
- }
-#endif
-
- // Was it better than the previous best?
- if (filt_err < (best_err - Bias))
- {
- best_err = filt_err;
- filt_best = filt_high;
- }
- }
-
- // Half the step distance if the best filter value was the same as last time
- if (filt_best == filt_mid)
- {
- filter_step = filter_step / 2;
- filt_direction = 0;
- }
- else
- {
- filt_direction = (filt_best < filt_mid) ? -1 : 1;
- filt_mid = filt_best;
- }
- }
-
- cm->filter_level = filt_best;
- cpi->last_auto_filt_val = filt_best;
- cpi->last_auto_filt_q = cm->base_qindex;
-
- cpi->frames_since_auto_filter = 0;
-}
diff --git a/vp8/encoder/ppc/csystemdependent.c b/vp8/encoder/ppc/csystemdependent.c
deleted file mode 100644
index 8dfd2a5..0000000
--- a/vp8/encoder/ppc/csystemdependent.c
+++ /dev/null
@@ -1,169 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/encoder/variance.h"
-#include "vp8/encoder/onyx_int.h"
-
-SADFunction *vp8_sad16x16;
-SADFunction *vp8_sad16x8;
-SADFunction *vp8_sad8x16;
-SADFunction *vp8_sad8x8;
-SADFunction *vp8_sad4x4;
-
-variance_function *vp8_variance4x4;
-variance_function *vp8_variance8x8;
-variance_function *vp8_variance8x16;
-variance_function *vp8_variance16x8;
-variance_function *vp8_variance16x16;
-
-variance_function *vp8_mse16x16;
-
-sub_pixel_variance_function *vp8_sub_pixel_variance4x4;
-sub_pixel_variance_function *vp8_sub_pixel_variance8x8;
-sub_pixel_variance_function *vp8_sub_pixel_variance8x16;
-sub_pixel_variance_function *vp8_sub_pixel_variance16x8;
-sub_pixel_variance_function *vp8_sub_pixel_variance16x16;
-
-int (*vp8_block_error)(short *coeff, short *dqcoeff);
-int (*vp8_mbblock_error)(MACROBLOCK *mb, int dc);
-
-int (*vp8_mbuverror)(MACROBLOCK *mb);
-unsigned int (*vp8_get_mb_ss)(short *);
-void (*vp8_short_fdct4x4)(short *input, short *output, int pitch);
-void (*vp8_short_fdct8x4)(short *input, short *output, int pitch);
-void (*vp8_fast_fdct4x4)(short *input, short *output, int pitch);
-void (*vp8_fast_fdct8x4)(short *input, short *output, int pitch);
-void (*short_walsh4x4)(short *input, short *output, int pitch);
-
-void (*vp8_subtract_b)(BLOCK *be, BLOCKD *bd, int pitch);
-void (*vp8_subtract_mby)(short *diff, unsigned char *src, unsigned char *pred, int stride);
-void (*vp8_subtract_mbuv)(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride);
-void (*vp8_fast_quantize_b)(BLOCK *b, BLOCKD *d);
-
-unsigned int (*vp8_get16x16pred_error)(unsigned char *src_ptr, int src_stride, unsigned char *ref_ptr, int ref_stride);
-unsigned int (*vp8_get8x8var)(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
-unsigned int (*vp8_get16x16var)(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
-unsigned int (*vp8_get4x4sse_cs)(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride);
-
-// c imports
-extern int block_error_c(short *coeff, short *dqcoeff);
-extern int vp8_mbblock_error_c(MACROBLOCK *mb, int dc);
-
-extern int vp8_mbuverror_c(MACROBLOCK *mb);
-extern unsigned int vp8_get8x8var_c(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
-extern void short_fdct4x4_c(short *input, short *output, int pitch);
-extern void short_fdct8x4_c(short *input, short *output, int pitch);
-extern void vp8_short_walsh4x4_c(short *input, short *output, int pitch);
-
-extern void vp8_subtract_b_c(BLOCK *be, BLOCKD *bd, int pitch);
-extern void subtract_mby_c(short *diff, unsigned char *src, unsigned char *pred, int stride);
-extern void subtract_mbuv_c(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride);
-extern void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d);
-
-extern SADFunction sad16x16_c;
-extern SADFunction sad16x8_c;
-extern SADFunction sad8x16_c;
-extern SADFunction sad8x8_c;
-extern SADFunction sad4x4_c;
-
-extern variance_function variance16x16_c;
-extern variance_function variance8x16_c;
-extern variance_function variance16x8_c;
-extern variance_function variance8x8_c;
-extern variance_function variance4x4_c;
-extern variance_function mse16x16_c;
-
-extern sub_pixel_variance_function sub_pixel_variance4x4_c;
-extern sub_pixel_variance_function sub_pixel_variance8x8_c;
-extern sub_pixel_variance_function sub_pixel_variance8x16_c;
-extern sub_pixel_variance_function sub_pixel_variance16x8_c;
-extern sub_pixel_variance_function sub_pixel_variance16x16_c;
-
-extern unsigned int vp8_get_mb_ss_c(short *);
-extern unsigned int vp8_get16x16pred_error_c(unsigned char *src_ptr, int src_stride, unsigned char *ref_ptr, int ref_stride);
-extern unsigned int vp8_get8x8var_c(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
-extern unsigned int vp8_get16x16var_c(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
-extern unsigned int vp8_get4x4sse_cs_c(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride);
-
-// ppc
-extern int vp8_block_error_ppc(short *coeff, short *dqcoeff);
-
-extern void vp8_short_fdct4x4_ppc(short *input, short *output, int pitch);
-extern void vp8_short_fdct8x4_ppc(short *input, short *output, int pitch);
-
-extern void vp8_subtract_mby_ppc(short *diff, unsigned char *src, unsigned char *pred, int stride);
-extern void vp8_subtract_mbuv_ppc(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride);
-
-extern SADFunction vp8_sad16x16_ppc;
-extern SADFunction vp8_sad16x8_ppc;
-extern SADFunction vp8_sad8x16_ppc;
-extern SADFunction vp8_sad8x8_ppc;
-extern SADFunction vp8_sad4x4_ppc;
-
-extern variance_function vp8_variance16x16_ppc;
-extern variance_function vp8_variance8x16_ppc;
-extern variance_function vp8_variance16x8_ppc;
-extern variance_function vp8_variance8x8_ppc;
-extern variance_function vp8_variance4x4_ppc;
-extern variance_function vp8_mse16x16_ppc;
-
-extern sub_pixel_variance_function vp8_sub_pixel_variance4x4_ppc;
-extern sub_pixel_variance_function vp8_sub_pixel_variance8x8_ppc;
-extern sub_pixel_variance_function vp8_sub_pixel_variance8x16_ppc;
-extern sub_pixel_variance_function vp8_sub_pixel_variance16x8_ppc;
-extern sub_pixel_variance_function vp8_sub_pixel_variance16x16_ppc;
-
-extern unsigned int vp8_get8x8var_ppc(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
-extern unsigned int vp8_get16x16var_ppc(unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr, int recon_stride, unsigned int *SSE, int *Sum);
-
-void vp8_cmachine_specific_config(void)
-{
- // Pure C:
- vp8_mbuverror = vp8_mbuverror_c;
- vp8_fast_quantize_b = vp8_fast_quantize_b_c;
- vp8_short_fdct4x4 = vp8_short_fdct4x4_ppc;
- vp8_short_fdct8x4 = vp8_short_fdct8x4_ppc;
- vp8_fast_fdct4x4 = vp8_short_fdct4x4_ppc;
- vp8_fast_fdct8x4 = vp8_short_fdct8x4_ppc;
- short_walsh4x4 = vp8_short_walsh4x4_c;
-
- vp8_variance4x4 = vp8_variance4x4_ppc;
- vp8_variance8x8 = vp8_variance8x8_ppc;
- vp8_variance8x16 = vp8_variance8x16_ppc;
- vp8_variance16x8 = vp8_variance16x8_ppc;
- vp8_variance16x16 = vp8_variance16x16_ppc;
- vp8_mse16x16 = vp8_mse16x16_ppc;
-
- vp8_sub_pixel_variance4x4 = vp8_sub_pixel_variance4x4_ppc;
- vp8_sub_pixel_variance8x8 = vp8_sub_pixel_variance8x8_ppc;
- vp8_sub_pixel_variance8x16 = vp8_sub_pixel_variance8x16_ppc;
- vp8_sub_pixel_variance16x8 = vp8_sub_pixel_variance16x8_ppc;
- vp8_sub_pixel_variance16x16 = vp8_sub_pixel_variance16x16_ppc;
-
- vp8_get_mb_ss = vp8_get_mb_ss_c;
- vp8_get16x16pred_error = vp8_get16x16pred_error_c;
- vp8_get8x8var = vp8_get8x8var_ppc;
- vp8_get16x16var = vp8_get16x16var_ppc;
- vp8_get4x4sse_cs = vp8_get4x4sse_cs_c;
-
- vp8_sad16x16 = vp8_sad16x16_ppc;
- vp8_sad16x8 = vp8_sad16x8_ppc;
- vp8_sad8x16 = vp8_sad8x16_ppc;
- vp8_sad8x8 = vp8_sad8x8_ppc;
- vp8_sad4x4 = vp8_sad4x4_ppc;
-
- vp8_block_error = vp8_block_error_ppc;
- vp8_mbblock_error = vp8_mbblock_error_c;
-
- vp8_subtract_b = vp8_subtract_b_c;
- vp8_subtract_mby = vp8_subtract_mby_ppc;
- vp8_subtract_mbuv = vp8_subtract_mbuv_ppc;
-}
diff --git a/vp8/encoder/ppc/variance_altivec.asm b/vp8/encoder/ppc/variance_altivec.asm
deleted file mode 100644
index a1ebf66..0000000
--- a/vp8/encoder/ppc/variance_altivec.asm
+++ /dev/null
@@ -1,375 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- .globl vp8_get8x8var_ppc
- .globl vp8_get16x16var_ppc
- .globl vp8_mse16x16_ppc
- .globl vp8_variance16x16_ppc
- .globl vp8_variance16x8_ppc
- .globl vp8_variance8x16_ppc
- .globl vp8_variance8x8_ppc
- .globl vp8_variance4x4_ppc
-
-.macro load_aligned_16 V R O
- lvsl v3, 0, \R ;# permutate value for alignment
-
- lvx v1, 0, \R
- lvx v2, \O, \R
-
- vperm \V, v1, v2, v3
-.endm
-
-.macro prologue
- mfspr r11, 256 ;# get old VRSAVE
- oris r12, r11, 0xffc0
- mtspr 256, r12 ;# set VRSAVE
-
- stwu r1, -32(r1) ;# create space on the stack
-
- li r10, 16 ;# load offset and loop counter
-
- vspltisw v7, 0 ;# zero for merging
- vspltisw v8, 0 ;# zero out total to start
- vspltisw v9, 0 ;# zero out total for dif^2
-.endm
-
-.macro epilogue
- addi r1, r1, 32 ;# recover stack
-
- mtspr 256, r11 ;# reset old VRSAVE
-.endm
-
-.macro compute_sum_sse
- ;# Compute sum first. Unpack to so signed subract
- ;# can be used. Only have a half word signed
- ;# subract. Do high, then low.
- vmrghb v2, v7, v4
- vmrghb v3, v7, v5
- vsubshs v2, v2, v3
- vsum4shs v8, v2, v8
-
- vmrglb v2, v7, v4
- vmrglb v3, v7, v5
- vsubshs v2, v2, v3
- vsum4shs v8, v2, v8
-
- ;# Now compute sse.
- vsububs v2, v4, v5
- vsububs v3, v5, v4
- vor v2, v2, v3
-
- vmsumubm v9, v2, v2, v9
-.endm
-
-.macro variance_16 DS loop_label store_sum
-\loop_label:
- ;# only one of the inputs should need to be aligned.
- load_aligned_16 v4, r3, r10
- load_aligned_16 v5, r5, r10
-
- ;# move onto the next line
- add r3, r3, r4
- add r5, r5, r6
-
- compute_sum_sse
-
- bdnz \loop_label
-
- vsumsws v8, v8, v7
- vsumsws v9, v9, v7
-
- stvx v8, 0, r1
- lwz r3, 12(r1)
-
- stvx v9, 0, r1
- lwz r4, 12(r1)
-
-.if \store_sum
- stw r3, 0(r8) ;# sum
-.endif
- stw r4, 0(r7) ;# sse
-
- mullw r3, r3, r3 ;# sum*sum
- srawi r3, r3, \DS ;# (sum*sum) >> DS
- subf r3, r3, r4 ;# sse - ((sum*sum) >> DS)
-.endm
-
-.macro variance_8 DS loop_label store_sum
-\loop_label:
- ;# only one of the inputs should need to be aligned.
- load_aligned_16 v4, r3, r10
- load_aligned_16 v5, r5, r10
-
- ;# move onto the next line
- add r3, r3, r4
- add r5, r5, r6
-
- ;# only one of the inputs should need to be aligned.
- load_aligned_16 v6, r3, r10
- load_aligned_16 v0, r5, r10
-
- ;# move onto the next line
- add r3, r3, r4
- add r5, r5, r6
-
- vmrghb v4, v4, v6
- vmrghb v5, v5, v0
-
- compute_sum_sse
-
- bdnz \loop_label
-
- vsumsws v8, v8, v7
- vsumsws v9, v9, v7
-
- stvx v8, 0, r1
- lwz r3, 12(r1)
-
- stvx v9, 0, r1
- lwz r4, 12(r1)
-
-.if \store_sum
- stw r3, 0(r8) ;# sum
-.endif
- stw r4, 0(r7) ;# sse
-
- mullw r3, r3, r3 ;# sum*sum
- srawi r3, r3, \DS ;# (sum*sum) >> 8
- subf r3, r3, r4 ;# sse - ((sum*sum) >> 8)
-.endm
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int source_stride
-;# r5 unsigned char *ref_ptr
-;# r6 int recon_stride
-;# r7 unsigned int *SSE
-;# r8 int *Sum
-;#
-;# r3 return value
-vp8_get8x8var_ppc:
-
- prologue
-
- li r9, 4
- mtctr r9
-
- variance_8 6, get8x8var_loop, 1
-
- epilogue
-
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int source_stride
-;# r5 unsigned char *ref_ptr
-;# r6 int recon_stride
-;# r7 unsigned int *SSE
-;# r8 int *Sum
-;#
-;# r3 return value
-vp8_get16x16var_ppc:
-
- prologue
-
- mtctr r10
-
- variance_16 8, get16x16var_loop, 1
-
- epilogue
-
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int source_stride
-;# r5 unsigned char *ref_ptr
-;# r6 int recon_stride
-;# r7 unsigned int *sse
-;#
-;# r 3 return value
-vp8_mse16x16_ppc:
- prologue
-
- mtctr r10
-
-mse16x16_loop:
- ;# only one of the inputs should need to be aligned.
- load_aligned_16 v4, r3, r10
- load_aligned_16 v5, r5, r10
-
- ;# move onto the next line
- add r3, r3, r4
- add r5, r5, r6
-
- ;# Now compute sse.
- vsububs v2, v4, v5
- vsububs v3, v5, v4
- vor v2, v2, v3
-
- vmsumubm v9, v2, v2, v9
-
- bdnz mse16x16_loop
-
- vsumsws v9, v9, v7
-
- stvx v9, 0, r1
- lwz r3, 12(r1)
-
- stvx v9, 0, r1
- lwz r3, 12(r1)
-
- stw r3, 0(r7) ;# sse
-
- epilogue
-
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int source_stride
-;# r5 unsigned char *ref_ptr
-;# r6 int recon_stride
-;# r7 unsigned int *sse
-;#
-;# r3 return value
-vp8_variance16x16_ppc:
-
- prologue
-
- mtctr r10
-
- variance_16 8, variance16x16_loop, 0
-
- epilogue
-
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int source_stride
-;# r5 unsigned char *ref_ptr
-;# r6 int recon_stride
-;# r7 unsigned int *sse
-;#
-;# r3 return value
-vp8_variance16x8_ppc:
-
- prologue
-
- li r9, 8
- mtctr r9
-
- variance_16 7, variance16x8_loop, 0
-
- epilogue
-
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int source_stride
-;# r5 unsigned char *ref_ptr
-;# r6 int recon_stride
-;# r7 unsigned int *sse
-;#
-;# r3 return value
-vp8_variance8x16_ppc:
-
- prologue
-
- li r9, 8
- mtctr r9
-
- variance_8 7, variance8x16_loop, 0
-
- epilogue
-
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int source_stride
-;# r5 unsigned char *ref_ptr
-;# r6 int recon_stride
-;# r7 unsigned int *sse
-;#
-;# r3 return value
-vp8_variance8x8_ppc:
-
- prologue
-
- li r9, 4
- mtctr r9
-
- variance_8 6, variance8x8_loop, 0
-
- epilogue
-
- blr
-
-.macro transfer_4x4 I P
- lwz r0, 0(\I)
- add \I, \I, \P
-
- lwz r10,0(\I)
- add \I, \I, \P
-
- lwz r8, 0(\I)
- add \I, \I, \P
-
- lwz r9, 0(\I)
-
- stw r0, 0(r1)
- stw r10, 4(r1)
- stw r8, 8(r1)
- stw r9, 12(r1)
-.endm
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int source_stride
-;# r5 unsigned char *ref_ptr
-;# r6 int recon_stride
-;# r7 unsigned int *sse
-;#
-;# r3 return value
-vp8_variance4x4_ppc:
-
- prologue
-
- transfer_4x4 r3, r4
- lvx v4, 0, r1
-
- transfer_4x4 r5, r6
- lvx v5, 0, r1
-
- compute_sum_sse
-
- vsumsws v8, v8, v7
- vsumsws v9, v9, v7
-
- stvx v8, 0, r1
- lwz r3, 12(r1)
-
- stvx v9, 0, r1
- lwz r4, 12(r1)
-
- stw r4, 0(r7) ;# sse
-
- mullw r3, r3, r3 ;# sum*sum
- srawi r3, r3, 4 ;# (sum*sum) >> 4
- subf r3, r3, r4 ;# sse - ((sum*sum) >> 4)
-
- epilogue
-
- blr
diff --git a/vp8/encoder/ppc/variance_subpixel_altivec.asm b/vp8/encoder/ppc/variance_subpixel_altivec.asm
deleted file mode 100644
index 301360b..0000000
--- a/vp8/encoder/ppc/variance_subpixel_altivec.asm
+++ /dev/null
@@ -1,865 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- .globl vp8_sub_pixel_variance4x4_ppc
- .globl vp8_sub_pixel_variance8x8_ppc
- .globl vp8_sub_pixel_variance8x16_ppc
- .globl vp8_sub_pixel_variance16x8_ppc
- .globl vp8_sub_pixel_variance16x16_ppc
-
-.macro load_c V, LABEL, OFF, R0, R1
- lis \R0, \LABEL@ha
- la \R1, \LABEL@l(\R0)
- lvx \V, \OFF, \R1
-.endm
-
-.macro load_vfilter V0, V1
- load_c \V0, vfilter_b, r6, r12, r10
-
- addi r6, r6, 16
- lvx \V1, r6, r10
-.endm
-
-.macro HProlog jump_label
- ;# load up horizontal filter
- slwi. r5, r5, 4 ;# index into horizontal filter array
-
- ;# index to the next set of vectors in the row.
- li r10, 16
-
- ;# downshift by 7 ( divide by 128 ) at the end
- vspltish v19, 7
-
- ;# If there isn't any filtering to be done for the horizontal, then
- ;# just skip to the second pass.
- beq \jump_label
-
- load_c v20, hfilter_b, r5, r12, r0
-
- ;# setup constants
- ;# v14 permutation value for alignment
- load_c v28, b_hperm_b, 0, r12, r0
-
- ;# index to the next set of vectors in the row.
- li r12, 32
-
- ;# rounding added in on the multiply
- vspltisw v21, 8
- vspltisw v18, 3
- vslw v18, v21, v18 ;# 0x00000040000000400000004000000040
-
- slwi. r6, r6, 5 ;# index into vertical filter array
-.endm
-
-;# Filters a horizontal line
-;# expects:
-;# r3 src_ptr
-;# r4 pitch
-;# r10 16
-;# r12 32
-;# v17 perm intput
-;# v18 rounding
-;# v19 shift
-;# v20 filter taps
-;# v21 tmp
-;# v22 tmp
-;# v23 tmp
-;# v24 tmp
-;# v25 tmp
-;# v26 tmp
-;# v27 tmp
-;# v28 perm output
-;#
-
-.macro hfilter_8 V, hp, lp, increment_counter
- lvsl v17, 0, r3 ;# permutate value for alignment
-
- ;# input to filter is 9 bytes wide, output is 8 bytes.
- lvx v21, 0, r3
- lvx v22, r10, r3
-
-.if \increment_counter
- add r3, r3, r4
-.endif
- vperm v21, v21, v22, v17
-
- vperm v24, v21, v21, \hp ;# v20 = 0123 1234 2345 3456
- vperm v25, v21, v21, \lp ;# v21 = 4567 5678 6789 789A
-
- vmsummbm v24, v20, v24, v18
- vmsummbm v25, v20, v25, v18
-
- vpkswus v24, v24, v25 ;# v24 = 0 4 8 C 1 5 9 D (16-bit)
-
- vsrh v24, v24, v19 ;# divide v0, v1 by 128
-
- vpkuhus \V, v24, v24 ;# \V = scrambled 8-bit result
-.endm
-
-.macro vfilter_16 P0 P1
- vmuleub v22, \P0, v20 ;# 64 + 4 positive taps
- vadduhm v22, v18, v22
- vmuloub v23, \P0, v20
- vadduhm v23, v18, v23
-
- vmuleub v24, \P1, v21
- vadduhm v22, v22, v24 ;# Re = evens, saturation unnecessary
- vmuloub v25, \P1, v21
- vadduhm v23, v23, v25 ;# Ro = odds
-
- vsrh v22, v22, v19 ;# divide by 128
- vsrh v23, v23, v19 ;# v16 v17 = evens, odds
- vmrghh \P0, v22, v23 ;# v18 v19 = 16-bit result in order
- vmrglh v23, v22, v23
- vpkuhus \P0, \P0, v23 ;# P0 = 8-bit result
-.endm
-
-.macro compute_sum_sse src, ref, sum, sse, t1, t2, z0
- ;# Compute sum first. Unpack to so signed subract
- ;# can be used. Only have a half word signed
- ;# subract. Do high, then low.
- vmrghb \t1, \z0, \src
- vmrghb \t2, \z0, \ref
- vsubshs \t1, \t1, \t2
- vsum4shs \sum, \t1, \sum
-
- vmrglb \t1, \z0, \src
- vmrglb \t2, \z0, \ref
- vsubshs \t1, \t1, \t2
- vsum4shs \sum, \t1, \sum
-
- ;# Now compute sse.
- vsububs \t1, \src, \ref
- vsububs \t2, \ref, \src
- vor \t1, \t1, \t2
-
- vmsumubm \sse, \t1, \t1, \sse
-.endm
-
-.macro variance_final sum, sse, z0, DS
- vsumsws \sum, \sum, \z0
- vsumsws \sse, \sse, \z0
-
- stvx \sum, 0, r1
- lwz r3, 12(r1)
-
- stvx \sse, 0, r1
- lwz r4, 12(r1)
-
- stw r4, 0(r9) ;# sse
-
- mullw r3, r3, r3 ;# sum*sum
- srawi r3, r3, \DS ;# (sum*sum) >> 8
- subf r3, r3, r4 ;# sse - ((sum*sum) >> 8)
-.endm
-
-.macro compute_sum_sse_16 V, increment_counter
- load_and_align_16 v16, r7, r8, \increment_counter
- compute_sum_sse \V, v16, v18, v19, v20, v21, v23
-.endm
-
-.macro load_and_align_16 V, R, P, increment_counter
- lvsl v17, 0, \R ;# permutate value for alignment
-
- ;# input to filter is 21 bytes wide, output is 16 bytes.
- ;# input will can span three vectors if not aligned correctly.
- lvx v21, 0, \R
- lvx v22, r10, \R
-
-.if \increment_counter
- add \R, \R, \P
-.endif
-
- vperm \V, v21, v22, v17
-.endm
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int src_pixels_per_line
-;# r5 int xoffset
-;# r6 int yoffset
-;# r7 unsigned char *dst_ptr
-;# r8 int dst_pixels_per_line
-;# r9 unsigned int *sse
-;#
-;# r3 return value
-vp8_sub_pixel_variance4x4_ppc:
- mfspr r11, 256 ;# get old VRSAVE
- oris r12, r11, 0xf830
- ori r12, r12, 0xfff8
- mtspr 256, r12 ;# set VRSAVE
-
- stwu r1,-32(r1) ;# create space on the stack
-
- HProlog second_pass_4x4_pre_copy_b
-
- ;# Load up permutation constants
- load_c v10, b_0123_b, 0, r12, r0
- load_c v11, b_4567_b, 0, r12, r0
-
- hfilter_8 v0, v10, v11, 1
- hfilter_8 v1, v10, v11, 1
- hfilter_8 v2, v10, v11, 1
- hfilter_8 v3, v10, v11, 1
-
- ;# Finished filtering main horizontal block. If there is no
- ;# vertical filtering, jump to storing the data. Otherwise
- ;# load up and filter the additional line that is needed
- ;# for the vertical filter.
- beq compute_sum_sse_4x4_b
-
- hfilter_8 v4, v10, v11, 0
-
- b second_pass_4x4_b
-
-second_pass_4x4_pre_copy_b:
- slwi r6, r6, 5 ;# index into vertical filter array
-
- load_and_align_16 v0, r3, r4, 1
- load_and_align_16 v1, r3, r4, 1
- load_and_align_16 v2, r3, r4, 1
- load_and_align_16 v3, r3, r4, 1
- load_and_align_16 v4, r3, r4, 0
-
-second_pass_4x4_b:
- vspltish v20, 8
- vspltish v18, 3
- vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
-
- load_vfilter v20, v21
-
- vfilter_16 v0, v1
- vfilter_16 v1, v2
- vfilter_16 v2, v3
- vfilter_16 v3, v4
-
-compute_sum_sse_4x4_b:
- vspltish v18, 0 ;# sum
- vspltish v19, 0 ;# sse
- vspltish v23, 0 ;# unpack
- li r10, 16
-
- load_and_align_16 v4, r7, r8, 1
- load_and_align_16 v5, r7, r8, 1
- load_and_align_16 v6, r7, r8, 1
- load_and_align_16 v7, r7, r8, 1
-
- vmrghb v0, v0, v1
- vmrghb v1, v2, v3
-
- vmrghb v2, v4, v5
- vmrghb v3, v6, v7
-
- load_c v10, b_hilo_b, 0, r12, r0
-
- vperm v0, v0, v1, v10
- vperm v1, v2, v3, v10
-
- compute_sum_sse v0, v1, v18, v19, v20, v21, v23
-
- variance_final v18, v19, v23, 4
-
- addi r1, r1, 32 ;# recover stack
- mtspr 256, r11 ;# reset old VRSAVE
-
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int src_pixels_per_line
-;# r5 int xoffset
-;# r6 int yoffset
-;# r7 unsigned char *dst_ptr
-;# r8 int dst_pixels_per_line
-;# r9 unsigned int *sse
-;#
-;# r3 return value
-vp8_sub_pixel_variance8x8_ppc:
- mfspr r11, 256 ;# get old VRSAVE
- oris r12, r11, 0xfff0
- ori r12, r12, 0xffff
- mtspr 256, r12 ;# set VRSAVE
-
- stwu r1,-32(r1) ;# create space on the stack
-
- HProlog second_pass_8x8_pre_copy_b
-
- ;# Load up permutation constants
- load_c v10, b_0123_b, 0, r12, r0
- load_c v11, b_4567_b, 0, r12, r0
-
- hfilter_8 v0, v10, v11, 1
- hfilter_8 v1, v10, v11, 1
- hfilter_8 v2, v10, v11, 1
- hfilter_8 v3, v10, v11, 1
- hfilter_8 v4, v10, v11, 1
- hfilter_8 v5, v10, v11, 1
- hfilter_8 v6, v10, v11, 1
- hfilter_8 v7, v10, v11, 1
-
- ;# Finished filtering main horizontal block. If there is no
- ;# vertical filtering, jump to storing the data. Otherwise
- ;# load up and filter the additional line that is needed
- ;# for the vertical filter.
- beq compute_sum_sse_8x8_b
-
- hfilter_8 v8, v10, v11, 0
-
- b second_pass_8x8_b
-
-second_pass_8x8_pre_copy_b:
- slwi. r6, r6, 5 ;# index into vertical filter array
-
- load_and_align_16 v0, r3, r4, 1
- load_and_align_16 v1, r3, r4, 1
- load_and_align_16 v2, r3, r4, 1
- load_and_align_16 v3, r3, r4, 1
- load_and_align_16 v4, r3, r4, 1
- load_and_align_16 v5, r3, r4, 1
- load_and_align_16 v6, r3, r4, 1
- load_and_align_16 v7, r3, r4, 1
- load_and_align_16 v8, r3, r4, 0
-
- beq compute_sum_sse_8x8_b
-
-second_pass_8x8_b:
- vspltish v20, 8
- vspltish v18, 3
- vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
-
- load_vfilter v20, v21
-
- vfilter_16 v0, v1
- vfilter_16 v1, v2
- vfilter_16 v2, v3
- vfilter_16 v3, v4
- vfilter_16 v4, v5
- vfilter_16 v5, v6
- vfilter_16 v6, v7
- vfilter_16 v7, v8
-
-compute_sum_sse_8x8_b:
- vspltish v18, 0 ;# sum
- vspltish v19, 0 ;# sse
- vspltish v23, 0 ;# unpack
- li r10, 16
-
- vmrghb v0, v0, v1
- vmrghb v1, v2, v3
- vmrghb v2, v4, v5
- vmrghb v3, v6, v7
-
- load_and_align_16 v4, r7, r8, 1
- load_and_align_16 v5, r7, r8, 1
- load_and_align_16 v6, r7, r8, 1
- load_and_align_16 v7, r7, r8, 1
- load_and_align_16 v8, r7, r8, 1
- load_and_align_16 v9, r7, r8, 1
- load_and_align_16 v10, r7, r8, 1
- load_and_align_16 v11, r7, r8, 0
-
- vmrghb v4, v4, v5
- vmrghb v5, v6, v7
- vmrghb v6, v8, v9
- vmrghb v7, v10, v11
-
- compute_sum_sse v0, v4, v18, v19, v20, v21, v23
- compute_sum_sse v1, v5, v18, v19, v20, v21, v23
- compute_sum_sse v2, v6, v18, v19, v20, v21, v23
- compute_sum_sse v3, v7, v18, v19, v20, v21, v23
-
- variance_final v18, v19, v23, 6
-
- addi r1, r1, 32 ;# recover stack
- mtspr 256, r11 ;# reset old VRSAVE
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int src_pixels_per_line
-;# r5 int xoffset
-;# r6 int yoffset
-;# r7 unsigned char *dst_ptr
-;# r8 int dst_pixels_per_line
-;# r9 unsigned int *sse
-;#
-;# r3 return value
-vp8_sub_pixel_variance8x16_ppc:
- mfspr r11, 256 ;# get old VRSAVE
- oris r12, r11, 0xffff
- ori r12, r12, 0xfffc
- mtspr 256, r12 ;# set VRSAVE
-
- stwu r1,-32(r1) ;# create space on the stack
-
- HProlog second_pass_8x16_pre_copy_b
-
- ;# Load up permutation constants
- load_c v29, b_0123_b, 0, r12, r0
- load_c v30, b_4567_b, 0, r12, r0
-
- hfilter_8 v0, v29, v30, 1
- hfilter_8 v1, v29, v30, 1
- hfilter_8 v2, v29, v30, 1
- hfilter_8 v3, v29, v30, 1
- hfilter_8 v4, v29, v30, 1
- hfilter_8 v5, v29, v30, 1
- hfilter_8 v6, v29, v30, 1
- hfilter_8 v7, v29, v30, 1
- hfilter_8 v8, v29, v30, 1
- hfilter_8 v9, v29, v30, 1
- hfilter_8 v10, v29, v30, 1
- hfilter_8 v11, v29, v30, 1
- hfilter_8 v12, v29, v30, 1
- hfilter_8 v13, v29, v30, 1
- hfilter_8 v14, v29, v30, 1
- hfilter_8 v15, v29, v30, 1
-
- ;# Finished filtering main horizontal block. If there is no
- ;# vertical filtering, jump to storing the data. Otherwise
- ;# load up and filter the additional line that is needed
- ;# for the vertical filter.
- beq compute_sum_sse_8x16_b
-
- hfilter_8 v16, v29, v30, 0
-
- b second_pass_8x16_b
-
-second_pass_8x16_pre_copy_b:
- slwi. r6, r6, 5 ;# index into vertical filter array
-
- load_and_align_16 v0, r3, r4, 1
- load_and_align_16 v1, r3, r4, 1
- load_and_align_16 v2, r3, r4, 1
- load_and_align_16 v3, r3, r4, 1
- load_and_align_16 v4, r3, r4, 1
- load_and_align_16 v5, r3, r4, 1
- load_and_align_16 v6, r3, r4, 1
- load_and_align_16 v7, r3, r4, 1
- load_and_align_16 v8, r3, r4, 1
- load_and_align_16 v9, r3, r4, 1
- load_and_align_16 v10, r3, r4, 1
- load_and_align_16 v11, r3, r4, 1
- load_and_align_16 v12, r3, r4, 1
- load_and_align_16 v13, r3, r4, 1
- load_and_align_16 v14, r3, r4, 1
- load_and_align_16 v15, r3, r4, 1
- load_and_align_16 v16, r3, r4, 0
-
- beq compute_sum_sse_8x16_b
-
-second_pass_8x16_b:
- vspltish v20, 8
- vspltish v18, 3
- vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
-
- load_vfilter v20, v21
-
- vfilter_16 v0, v1
- vfilter_16 v1, v2
- vfilter_16 v2, v3
- vfilter_16 v3, v4
- vfilter_16 v4, v5
- vfilter_16 v5, v6
- vfilter_16 v6, v7
- vfilter_16 v7, v8
- vfilter_16 v8, v9
- vfilter_16 v9, v10
- vfilter_16 v10, v11
- vfilter_16 v11, v12
- vfilter_16 v12, v13
- vfilter_16 v13, v14
- vfilter_16 v14, v15
- vfilter_16 v15, v16
-
-compute_sum_sse_8x16_b:
- vspltish v18, 0 ;# sum
- vspltish v19, 0 ;# sse
- vspltish v23, 0 ;# unpack
- li r10, 16
-
- vmrghb v0, v0, v1
- vmrghb v1, v2, v3
- vmrghb v2, v4, v5
- vmrghb v3, v6, v7
- vmrghb v4, v8, v9
- vmrghb v5, v10, v11
- vmrghb v6, v12, v13
- vmrghb v7, v14, v15
-
- load_and_align_16 v8, r7, r8, 1
- load_and_align_16 v9, r7, r8, 1
- load_and_align_16 v10, r7, r8, 1
- load_and_align_16 v11, r7, r8, 1
- load_and_align_16 v12, r7, r8, 1
- load_and_align_16 v13, r7, r8, 1
- load_and_align_16 v14, r7, r8, 1
- load_and_align_16 v15, r7, r8, 1
-
- vmrghb v8, v8, v9
- vmrghb v9, v10, v11
- vmrghb v10, v12, v13
- vmrghb v11, v14, v15
-
- compute_sum_sse v0, v8, v18, v19, v20, v21, v23
- compute_sum_sse v1, v9, v18, v19, v20, v21, v23
- compute_sum_sse v2, v10, v18, v19, v20, v21, v23
- compute_sum_sse v3, v11, v18, v19, v20, v21, v23
-
- load_and_align_16 v8, r7, r8, 1
- load_and_align_16 v9, r7, r8, 1
- load_and_align_16 v10, r7, r8, 1
- load_and_align_16 v11, r7, r8, 1
- load_and_align_16 v12, r7, r8, 1
- load_and_align_16 v13, r7, r8, 1
- load_and_align_16 v14, r7, r8, 1
- load_and_align_16 v15, r7, r8, 0
-
- vmrghb v8, v8, v9
- vmrghb v9, v10, v11
- vmrghb v10, v12, v13
- vmrghb v11, v14, v15
-
- compute_sum_sse v4, v8, v18, v19, v20, v21, v23
- compute_sum_sse v5, v9, v18, v19, v20, v21, v23
- compute_sum_sse v6, v10, v18, v19, v20, v21, v23
- compute_sum_sse v7, v11, v18, v19, v20, v21, v23
-
- variance_final v18, v19, v23, 7
-
- addi r1, r1, 32 ;# recover stack
- mtspr 256, r11 ;# reset old VRSAVE
- blr
-
-;# Filters a horizontal line
-;# expects:
-;# r3 src_ptr
-;# r4 pitch
-;# r10 16
-;# r12 32
-;# v17 perm intput
-;# v18 rounding
-;# v19 shift
-;# v20 filter taps
-;# v21 tmp
-;# v22 tmp
-;# v23 tmp
-;# v24 tmp
-;# v25 tmp
-;# v26 tmp
-;# v27 tmp
-;# v28 perm output
-;#
-.macro hfilter_16 V, increment_counter
-
- lvsl v17, 0, r3 ;# permutate value for alignment
-
- ;# input to filter is 21 bytes wide, output is 16 bytes.
- ;# input will can span three vectors if not aligned correctly.
- lvx v21, 0, r3
- lvx v22, r10, r3
- lvx v23, r12, r3
-
-.if \increment_counter
- add r3, r3, r4
-.endif
- vperm v21, v21, v22, v17
- vperm v22, v22, v23, v17 ;# v8 v9 = 21 input pixels left-justified
-
- ;# set 0
- vmsummbm v24, v20, v21, v18 ;# taps times elements
-
- ;# set 1
- vsldoi v23, v21, v22, 1
- vmsummbm v25, v20, v23, v18
-
- ;# set 2
- vsldoi v23, v21, v22, 2
- vmsummbm v26, v20, v23, v18
-
- ;# set 3
- vsldoi v23, v21, v22, 3
- vmsummbm v27, v20, v23, v18
-
- vpkswus v24, v24, v25 ;# v24 = 0 4 8 C 1 5 9 D (16-bit)
- vpkswus v25, v26, v27 ;# v25 = 2 6 A E 3 7 B F
-
- vsrh v24, v24, v19 ;# divide v0, v1 by 128
- vsrh v25, v25, v19
-
- vpkuhus \V, v24, v25 ;# \V = scrambled 8-bit result
- vperm \V, \V, v0, v28 ;# \V = correctly-ordered result
-.endm
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int src_pixels_per_line
-;# r5 int xoffset
-;# r6 int yoffset
-;# r7 unsigned char *dst_ptr
-;# r8 int dst_pixels_per_line
-;# r9 unsigned int *sse
-;#
-;# r3 return value
-vp8_sub_pixel_variance16x8_ppc:
- mfspr r11, 256 ;# get old VRSAVE
- oris r12, r11, 0xffff
- ori r12, r12, 0xfff8
- mtspr 256, r12 ;# set VRSAVE
-
- stwu r1, -32(r1) ;# create space on the stack
-
- HProlog second_pass_16x8_pre_copy_b
-
- hfilter_16 v0, 1
- hfilter_16 v1, 1
- hfilter_16 v2, 1
- hfilter_16 v3, 1
- hfilter_16 v4, 1
- hfilter_16 v5, 1
- hfilter_16 v6, 1
- hfilter_16 v7, 1
-
- ;# Finished filtering main horizontal block. If there is no
- ;# vertical filtering, jump to storing the data. Otherwise
- ;# load up and filter the additional line that is needed
- ;# for the vertical filter.
- beq compute_sum_sse_16x8_b
-
- hfilter_16 v8, 0
-
- b second_pass_16x8_b
-
-second_pass_16x8_pre_copy_b:
- slwi. r6, r6, 5 ;# index into vertical filter array
-
- load_and_align_16 v0, r3, r4, 1
- load_and_align_16 v1, r3, r4, 1
- load_and_align_16 v2, r3, r4, 1
- load_and_align_16 v3, r3, r4, 1
- load_and_align_16 v4, r3, r4, 1
- load_and_align_16 v5, r3, r4, 1
- load_and_align_16 v6, r3, r4, 1
- load_and_align_16 v7, r3, r4, 1
- load_and_align_16 v8, r3, r4, 1
-
- beq compute_sum_sse_16x8_b
-
-second_pass_16x8_b:
- vspltish v20, 8
- vspltish v18, 3
- vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
-
- load_vfilter v20, v21
-
- vfilter_16 v0, v1
- vfilter_16 v1, v2
- vfilter_16 v2, v3
- vfilter_16 v3, v4
- vfilter_16 v4, v5
- vfilter_16 v5, v6
- vfilter_16 v6, v7
- vfilter_16 v7, v8
-
-compute_sum_sse_16x8_b:
- vspltish v18, 0 ;# sum
- vspltish v19, 0 ;# sse
- vspltish v23, 0 ;# unpack
- li r10, 16
-
- compute_sum_sse_16 v0, 1
- compute_sum_sse_16 v1, 1
- compute_sum_sse_16 v2, 1
- compute_sum_sse_16 v3, 1
- compute_sum_sse_16 v4, 1
- compute_sum_sse_16 v5, 1
- compute_sum_sse_16 v6, 1
- compute_sum_sse_16 v7, 0
-
- variance_final v18, v19, v23, 7
-
- addi r1, r1, 32 ;# recover stack
-
- mtspr 256, r11 ;# reset old VRSAVE
-
- blr
-
- .align 2
-;# r3 unsigned char *src_ptr
-;# r4 int src_pixels_per_line
-;# r5 int xoffset
-;# r6 int yoffset
-;# r7 unsigned char *dst_ptr
-;# r8 int dst_pixels_per_line
-;# r9 unsigned int *sse
-;#
-;# r3 return value
-vp8_sub_pixel_variance16x16_ppc:
- mfspr r11, 256 ;# get old VRSAVE
- oris r12, r11, 0xffff
- ori r12, r12, 0xfff8
- mtspr 256, r12 ;# set VRSAVE
-
- stwu r1, -32(r1) ;# create space on the stack
-
- HProlog second_pass_16x16_pre_copy_b
-
- hfilter_16 v0, 1
- hfilter_16 v1, 1
- hfilter_16 v2, 1
- hfilter_16 v3, 1
- hfilter_16 v4, 1
- hfilter_16 v5, 1
- hfilter_16 v6, 1
- hfilter_16 v7, 1
- hfilter_16 v8, 1
- hfilter_16 v9, 1
- hfilter_16 v10, 1
- hfilter_16 v11, 1
- hfilter_16 v12, 1
- hfilter_16 v13, 1
- hfilter_16 v14, 1
- hfilter_16 v15, 1
-
- ;# Finished filtering main horizontal block. If there is no
- ;# vertical filtering, jump to storing the data. Otherwise
- ;# load up and filter the additional line that is needed
- ;# for the vertical filter.
- beq compute_sum_sse_16x16_b
-
- hfilter_16 v16, 0
-
- b second_pass_16x16_b
-
-second_pass_16x16_pre_copy_b:
- slwi. r6, r6, 5 ;# index into vertical filter array
-
- load_and_align_16 v0, r3, r4, 1
- load_and_align_16 v1, r3, r4, 1
- load_and_align_16 v2, r3, r4, 1
- load_and_align_16 v3, r3, r4, 1
- load_and_align_16 v4, r3, r4, 1
- load_and_align_16 v5, r3, r4, 1
- load_and_align_16 v6, r3, r4, 1
- load_and_align_16 v7, r3, r4, 1
- load_and_align_16 v8, r3, r4, 1
- load_and_align_16 v9, r3, r4, 1
- load_and_align_16 v10, r3, r4, 1
- load_and_align_16 v11, r3, r4, 1
- load_and_align_16 v12, r3, r4, 1
- load_and_align_16 v13, r3, r4, 1
- load_and_align_16 v14, r3, r4, 1
- load_and_align_16 v15, r3, r4, 1
- load_and_align_16 v16, r3, r4, 0
-
- beq compute_sum_sse_16x16_b
-
-second_pass_16x16_b:
- vspltish v20, 8
- vspltish v18, 3
- vslh v18, v20, v18 ;# 0x0040 0040 0040 0040 0040 0040 0040 0040
-
- load_vfilter v20, v21
-
- vfilter_16 v0, v1
- vfilter_16 v1, v2
- vfilter_16 v2, v3
- vfilter_16 v3, v4
- vfilter_16 v4, v5
- vfilter_16 v5, v6
- vfilter_16 v6, v7
- vfilter_16 v7, v8
- vfilter_16 v8, v9
- vfilter_16 v9, v10
- vfilter_16 v10, v11
- vfilter_16 v11, v12
- vfilter_16 v12, v13
- vfilter_16 v13, v14
- vfilter_16 v14, v15
- vfilter_16 v15, v16
-
-compute_sum_sse_16x16_b:
- vspltish v18, 0 ;# sum
- vspltish v19, 0 ;# sse
- vspltish v23, 0 ;# unpack
- li r10, 16
-
- compute_sum_sse_16 v0, 1
- compute_sum_sse_16 v1, 1
- compute_sum_sse_16 v2, 1
- compute_sum_sse_16 v3, 1
- compute_sum_sse_16 v4, 1
- compute_sum_sse_16 v5, 1
- compute_sum_sse_16 v6, 1
- compute_sum_sse_16 v7, 1
- compute_sum_sse_16 v8, 1
- compute_sum_sse_16 v9, 1
- compute_sum_sse_16 v10, 1
- compute_sum_sse_16 v11, 1
- compute_sum_sse_16 v12, 1
- compute_sum_sse_16 v13, 1
- compute_sum_sse_16 v14, 1
- compute_sum_sse_16 v15, 0
-
- variance_final v18, v19, v23, 8
-
- addi r1, r1, 32 ;# recover stack
-
- mtspr 256, r11 ;# reset old VRSAVE
-
- blr
-
- .data
-
- .align 4
-hfilter_b:
- .byte 128, 0, 0, 0,128, 0, 0, 0,128, 0, 0, 0,128, 0, 0, 0
- .byte 112, 16, 0, 0,112, 16, 0, 0,112, 16, 0, 0,112, 16, 0, 0
- .byte 96, 32, 0, 0, 96, 32, 0, 0, 96, 32, 0, 0, 96, 32, 0, 0
- .byte 80, 48, 0, 0, 80, 48, 0, 0, 80, 48, 0, 0, 80, 48, 0, 0
- .byte 64, 64, 0, 0, 64, 64, 0, 0, 64, 64, 0, 0, 64, 64, 0, 0
- .byte 48, 80, 0, 0, 48, 80, 0, 0, 48, 80, 0, 0, 48, 80, 0, 0
- .byte 32, 96, 0, 0, 32, 96, 0, 0, 32, 96, 0, 0, 32, 96, 0, 0
- .byte 16,112, 0, 0, 16,112, 0, 0, 16,112, 0, 0, 16,112, 0, 0
-
- .align 4
-vfilter_b:
- .byte 128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128
- .byte 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
- .byte 112,112,112,112,112,112,112,112,112,112,112,112,112,112,112,112
- .byte 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16
- .byte 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96
- .byte 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32
- .byte 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80
- .byte 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48
- .byte 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
- .byte 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
- .byte 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48
- .byte 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80
- .byte 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32
- .byte 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96, 96
- .byte 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16
- .byte 112,112,112,112,112,112,112,112,112,112,112,112,112,112,112,112
-
- .align 4
-b_hperm_b:
- .byte 0, 4, 8, 12, 1, 5, 9, 13, 2, 6, 10, 14, 3, 7, 11, 15
-
- .align 4
-b_0123_b:
- .byte 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6
-
- .align 4
-b_4567_b:
- .byte 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10
-
-b_hilo_b:
- .byte 0, 1, 2, 3, 4, 5, 6, 7, 16, 17, 18, 19, 20, 21, 22, 23
diff --git a/vp8/encoder/psnr.c b/vp8/encoder/psnr.c
deleted file mode 100644
index 5119bb8..0000000
--- a/vp8/encoder/psnr.c
+++ /dev/null
@@ -1,31 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_scale/yv12config.h"
-#include "math.h"
-#include "vp8/common/systemdependent.h" /* for vp8_clear_system_state() */
-
-#define MAX_PSNR 60
-
-double vp8_mse2psnr(double Samples, double Peak, double Mse)
-{
- double psnr;
-
- if ((double)Mse > 0.0)
- psnr = 10.0 * log10(Peak * Peak * Samples / Mse);
- else
- psnr = MAX_PSNR; // Limit to prevent / 0
-
- if (psnr > MAX_PSNR)
- psnr = MAX_PSNR;
-
- return psnr;
-}
diff --git a/vp8/encoder/quantize.c b/vp8/encoder/quantize.c
deleted file mode 100644
index 803e3a5..0000000
--- a/vp8/encoder/quantize.c
+++ /dev/null
@@ -1,301 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <math.h>
-#include "vpx_mem/vpx_mem.h"
-
-#include "quantize.h"
-#include "vp8/common/entropy.h"
-
-#define EXACT_QUANT
-
-#ifdef EXACT_FASTQUANT
-void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d)
-{
- int i, rc, eob;
- int zbin;
- int x, y, z, sz;
- short *coeff_ptr = b->coeff;
- short *zbin_ptr = b->zbin;
- short *round_ptr = b->round;
- short *quant_ptr = b->quant_fast;
- short *quant_shift_ptr = b->quant_shift;
- short *qcoeff_ptr = d->qcoeff;
- short *dqcoeff_ptr = d->dqcoeff;
- short *dequant_ptr = d->dequant;
-
- vpx_memset(qcoeff_ptr, 0, 32);
- vpx_memset(dqcoeff_ptr, 0, 32);
-
- eob = -1;
-
- for (i = 0; i < 16; i++)
- {
- rc = vp8_default_zig_zag1d[i];
- z = coeff_ptr[rc];
- zbin = zbin_ptr[rc] ;
-
- sz = (z >> 31); // sign of z
- x = (z ^ sz) - sz; // x = abs(z)
-
- if (x >= zbin)
- {
- x += round_ptr[rc];
- y = (((x * quant_ptr[rc]) >> 16) + x)
- >> quant_shift_ptr[rc]; // quantize (x)
- x = (y ^ sz) - sz; // get the sign back
- qcoeff_ptr[rc] = x; // write to destination
- dqcoeff_ptr[rc] = x * dequant_ptr[rc]; // dequantized value
-
- if (y)
- {
- eob = i; // last nonzero coeffs
- }
- }
- }
- d->eob = eob + 1;
-}
-
-#else
-
-void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d)
-{
- int i, rc, eob;
- int x, y, z, sz;
- short *coeff_ptr = b->coeff;
- short *round_ptr = b->round;
- short *quant_ptr = b->quant_fast;
- short *qcoeff_ptr = d->qcoeff;
- short *dqcoeff_ptr = d->dqcoeff;
- short *dequant_ptr = d->dequant;
-
- eob = -1;
- for (i = 0; i < 16; i++)
- {
- rc = vp8_default_zig_zag1d[i];
- z = coeff_ptr[rc];
-
- sz = (z >> 31); // sign of z
- x = (z ^ sz) - sz; // x = abs(z)
-
- y = ((x + round_ptr[rc]) * quant_ptr[rc]) >> 16; // quantize (x)
- x = (y ^ sz) - sz; // get the sign back
- qcoeff_ptr[rc] = x; // write to destination
- dqcoeff_ptr[rc] = x * dequant_ptr[rc]; // dequantized value
-
- if (y)
- {
- eob = i; // last nonzero coeffs
- }
- }
- d->eob = eob + 1;
-}
-
-#endif
-
-#ifdef EXACT_QUANT
-void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d)
-{
- int i, rc, eob;
- int zbin;
- int x, y, z, sz;
- short *zbin_boost_ptr = b->zrun_zbin_boost;
- short *coeff_ptr = b->coeff;
- short *zbin_ptr = b->zbin;
- short *round_ptr = b->round;
- short *quant_ptr = b->quant;
- short *quant_shift_ptr = b->quant_shift;
- short *qcoeff_ptr = d->qcoeff;
- short *dqcoeff_ptr = d->dqcoeff;
- short *dequant_ptr = d->dequant;
- short zbin_oq_value = b->zbin_extra;
-
- vpx_memset(qcoeff_ptr, 0, 32);
- vpx_memset(dqcoeff_ptr, 0, 32);
-
- eob = -1;
-
- for (i = 0; i < 16; i++)
- {
- rc = vp8_default_zig_zag1d[i];
- z = coeff_ptr[rc];
-
- zbin = zbin_ptr[rc] + *zbin_boost_ptr + zbin_oq_value;
-
- zbin_boost_ptr ++;
- sz = (z >> 31); // sign of z
- x = (z ^ sz) - sz; // x = abs(z)
-
- if (x >= zbin)
- {
- x += round_ptr[rc];
- y = (((x * quant_ptr[rc]) >> 16) + x)
- >> quant_shift_ptr[rc]; // quantize (x)
- x = (y ^ sz) - sz; // get the sign back
- qcoeff_ptr[rc] = x; // write to destination
- dqcoeff_ptr[rc] = x * dequant_ptr[rc]; // dequantized value
-
- if (y)
- {
- eob = i; // last nonzero coeffs
- zbin_boost_ptr = b->zrun_zbin_boost; // reset zero runlength
- }
- }
- }
-
- d->eob = eob + 1;
-}
-
-/* Perform regular quantization, with unbiased rounding and no zero bin. */
-void vp8_strict_quantize_b(BLOCK *b, BLOCKD *d)
-{
- int i;
- int rc;
- int eob;
- int x;
- int y;
- int z;
- int sz;
- short *coeff_ptr;
- short *quant_ptr;
- short *quant_shift_ptr;
- short *qcoeff_ptr;
- short *dqcoeff_ptr;
- short *dequant_ptr;
-
- coeff_ptr = b->coeff;
- quant_ptr = b->quant;
- quant_shift_ptr = b->quant_shift;
- qcoeff_ptr = d->qcoeff;
- dqcoeff_ptr = d->dqcoeff;
- dequant_ptr = d->dequant;
- eob = - 1;
- vpx_memset(qcoeff_ptr, 0, 32);
- vpx_memset(dqcoeff_ptr, 0, 32);
- for (i = 0; i < 16; i++)
- {
- int dq;
- int round;
-
- /*TODO: These arrays should be stored in zig-zag order.*/
- rc = vp8_default_zig_zag1d[i];
- z = coeff_ptr[rc];
- dq = dequant_ptr[rc];
- round = dq >> 1;
- /* Sign of z. */
- sz = -(z < 0);
- x = (z + sz) ^ sz;
- x += round;
- if (x >= dq)
- {
- /* Quantize x. */
- y = (((x * quant_ptr[rc]) >> 16) + x) >> quant_shift_ptr[rc];
- /* Put the sign back. */
- x = (y + sz) ^ sz;
- /* Save the coefficient and its dequantized value. */
- qcoeff_ptr[rc] = x;
- dqcoeff_ptr[rc] = x * dq;
- /* Remember the last non-zero coefficient. */
- if (y)
- eob = i;
- }
- }
-
- d->eob = eob + 1;
-}
-
-#else
-
-void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d)
-{
- int i, rc, eob;
- int zbin;
- int x, y, z, sz;
- short *zbin_boost_ptr = b->zrun_zbin_boost;
- short *coeff_ptr = b->coeff;
- short *zbin_ptr = b->zbin;
- short *round_ptr = b->round;
- short *quant_ptr = b->quant;
- short *qcoeff_ptr = d->qcoeff;
- short *dqcoeff_ptr = d->dqcoeff;
- short *dequant_ptr = d->dequant;
- short zbin_oq_value = b->zbin_extra;
-
- vpx_memset(qcoeff_ptr, 0, 32);
- vpx_memset(dqcoeff_ptr, 0, 32);
-
- eob = -1;
-
- for (i = 0; i < 16; i++)
- {
- rc = vp8_default_zig_zag1d[i];
- z = coeff_ptr[rc];
-
- //if ( i == 0 )
- // zbin = zbin_ptr[rc] + *zbin_boost_ptr + zbin_oq_value/2;
- //else
- zbin = zbin_ptr[rc] + *zbin_boost_ptr + zbin_oq_value;
-
- zbin_boost_ptr ++;
- sz = (z >> 31); // sign of z
- x = (z ^ sz) - sz; // x = abs(z)
-
- if (x >= zbin)
- {
- y = ((x + round_ptr[rc]) * quant_ptr[rc]) >> 16; // quantize (x)
- x = (y ^ sz) - sz; // get the sign back
- qcoeff_ptr[rc] = x; // write to destination
- dqcoeff_ptr[rc] = x * dequant_ptr[rc]; // dequantized value
-
- if (y)
- {
- eob = i; // last nonzero coeffs
- zbin_boost_ptr = &b->zrun_zbin_boost[0]; // reset zero runlength
- }
- }
- }
-
- d->eob = eob + 1;
-}
-
-#endif
-
-void vp8_quantize_mby(MACROBLOCK *x)
-{
- int i;
- int has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED
- && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
-
- for (i = 0; i < 16; i++)
- x->quantize_b(&x->block[i], &x->e_mbd.block[i]);
-
- if(has_2nd_order)
- x->quantize_b(&x->block[24], &x->e_mbd.block[24]);
-}
-
-void vp8_quantize_mb(MACROBLOCK *x)
-{
- int i;
- int has_2nd_order=(x->e_mbd.mode_info_context->mbmi.mode != B_PRED
- && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV);
-
- for (i = 0; i < 24+has_2nd_order; i++)
- x->quantize_b(&x->block[i], &x->e_mbd.block[i]);
-}
-
-
-void vp8_quantize_mbuv(MACROBLOCK *x)
-{
- int i;
-
- for (i = 16; i < 24; i++)
- x->quantize_b(&x->block[i], &x->e_mbd.block[i]);
-}
diff --git a/vp8/encoder/quantize.h b/vp8/encoder/quantize.h
deleted file mode 100644
index b74718b..0000000
--- a/vp8/encoder/quantize.h
+++ /dev/null
@@ -1,56 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_QUANTIZE_H
-#define __INC_QUANTIZE_H
-
-#include "block.h"
-
-#define prototype_quantize_block(sym) \
- void (sym)(BLOCK *b,BLOCKD *d)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/quantize_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/quantize_arm.h"
-#endif
-
-#ifndef vp8_quantize_quantb
-#define vp8_quantize_quantb vp8_regular_quantize_b
-#endif
-extern prototype_quantize_block(vp8_quantize_quantb);
-
-#ifndef vp8_quantize_fastquantb
-#define vp8_quantize_fastquantb vp8_fast_quantize_b_c
-#endif
-extern prototype_quantize_block(vp8_quantize_fastquantb);
-
-typedef struct
-{
- prototype_quantize_block(*quantb);
- prototype_quantize_block(*fastquantb);
-} vp8_quantize_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define QUANTIZE_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define QUANTIZE_INVOKE(ctx,fn) vp8_quantize_##fn
-#endif
-
-extern void vp8_strict_quantize_b(BLOCK *b,BLOCKD *d);
-
-extern void vp8_quantize_mb(MACROBLOCK *x);
-extern void vp8_quantize_mbuv(MACROBLOCK *x);
-extern void vp8_quantize_mby(MACROBLOCK *x);
-
-#endif
diff --git a/vp8/encoder/ratectrl.c b/vp8/encoder/ratectrl.c
deleted file mode 100644
index e2c6327..0000000
--- a/vp8/encoder/ratectrl.c
+++ /dev/null
@@ -1,1607 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdlib.h>
-#include <stdio.h>
-#include <string.h>
-#include <limits.h>
-#include <assert.h>
-
-#include "math.h"
-#include "vp8/common/common.h"
-#include "ratectrl.h"
-#include "vp8/common/entropymode.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vp8/common/systemdependent.h"
-#include "encodemv.h"
-
-
-#define MIN_BPB_FACTOR 0.01
-#define MAX_BPB_FACTOR 50
-
-extern const MB_PREDICTION_MODE vp8_mode_order[MAX_MODES];
-extern const MV_REFERENCE_FRAME vp8_ref_frame_order[MAX_MODES];
-
-
-
-#ifdef MODE_STATS
-extern int y_modes[5];
-extern int uv_modes[4];
-extern int b_modes[10];
-
-extern int inter_y_modes[10];
-extern int inter_uv_modes[4];
-extern int inter_b_modes[10];
-#endif
-
-// Bits Per MB at different Q (Multiplied by 512)
-#define BPER_MB_NORMBITS 9
-
-// Work in progress recalibration of baseline rate tables based on
-// the assumption that bits per mb is inversely proportional to the
-// quantizer value.
-const int vp8_bits_per_mb[2][QINDEX_RANGE] =
-{
- // Intra case 450000/Qintra
- {
- 1125000,900000, 750000, 642857, 562500, 500000, 450000, 450000,
- 409090, 375000, 346153, 321428, 300000, 281250, 264705, 264705,
- 250000, 236842, 225000, 225000, 214285, 214285, 204545, 204545,
- 195652, 195652, 187500, 180000, 180000, 173076, 166666, 160714,
- 155172, 150000, 145161, 140625, 136363, 132352, 128571, 125000,
- 121621, 121621, 118421, 115384, 112500, 109756, 107142, 104651,
- 102272, 100000, 97826, 97826, 95744, 93750, 91836, 90000,
- 88235, 86538, 84905, 83333, 81818, 80357, 78947, 77586,
- 76271, 75000, 73770, 72580, 71428, 70312, 69230, 68181,
- 67164, 66176, 65217, 64285, 63380, 62500, 61643, 60810,
- 60000, 59210, 59210, 58441, 57692, 56962, 56250, 55555,
- 54878, 54216, 53571, 52941, 52325, 51724, 51136, 50561,
- 49450, 48387, 47368, 46875, 45918, 45000, 44554, 44117,
- 43269, 42452, 41666, 40909, 40178, 39473, 38793, 38135,
- 36885, 36290, 35714, 35156, 34615, 34090, 33582, 33088,
- 32608, 32142, 31468, 31034, 30405, 29801, 29220, 28662,
- },
- // Inter case 285000/Qinter
- {
- 712500, 570000, 475000, 407142, 356250, 316666, 285000, 259090,
- 237500, 219230, 203571, 190000, 178125, 167647, 158333, 150000,
- 142500, 135714, 129545, 123913, 118750, 114000, 109615, 105555,
- 101785, 98275, 95000, 91935, 89062, 86363, 83823, 81428,
- 79166, 77027, 75000, 73076, 71250, 69512, 67857, 66279,
- 64772, 63333, 61956, 60638, 59375, 58163, 57000, 55882,
- 54807, 53773, 52777, 51818, 50892, 50000, 49137, 47500,
- 45967, 44531, 43181, 41911, 40714, 39583, 38513, 37500,
- 36538, 35625, 34756, 33928, 33139, 32386, 31666, 30978,
- 30319, 29687, 29081, 28500, 27941, 27403, 26886, 26388,
- 25909, 25446, 25000, 24568, 23949, 23360, 22800, 22265,
- 21755, 21268, 20802, 20357, 19930, 19520, 19127, 18750,
- 18387, 18037, 17701, 17378, 17065, 16764, 16473, 16101,
- 15745, 15405, 15079, 14766, 14467, 14179, 13902, 13636,
- 13380, 13133, 12895, 12666, 12445, 12179, 11924, 11632,
- 11445, 11220, 11003, 10795, 10594, 10401, 10215, 10035,
- }
-};
-
-static const int kf_boost_qadjustment[QINDEX_RANGE] =
-{
- 128, 129, 130, 131, 132, 133, 134, 135,
- 136, 137, 138, 139, 140, 141, 142, 143,
- 144, 145, 146, 147, 148, 149, 150, 151,
- 152, 153, 154, 155, 156, 157, 158, 159,
- 160, 161, 162, 163, 164, 165, 166, 167,
- 168, 169, 170, 171, 172, 173, 174, 175,
- 176, 177, 178, 179, 180, 181, 182, 183,
- 184, 185, 186, 187, 188, 189, 190, 191,
- 192, 193, 194, 195, 196, 197, 198, 199,
- 200, 200, 201, 201, 202, 203, 203, 203,
- 204, 204, 205, 205, 206, 206, 207, 207,
- 208, 208, 209, 209, 210, 210, 211, 211,
- 212, 212, 213, 213, 214, 214, 215, 215,
- 216, 216, 217, 217, 218, 218, 219, 219,
- 220, 220, 220, 220, 220, 220, 220, 220,
- 220, 220, 220, 220, 220, 220, 220, 220,
-};
-
-//#define GFQ_ADJUSTMENT (Q+100)
-#define GFQ_ADJUSTMENT vp8_gf_boost_qadjustment[Q]
-const int vp8_gf_boost_qadjustment[QINDEX_RANGE] =
-{
- 80, 82, 84, 86, 88, 90, 92, 94,
- 96, 97, 98, 99, 100, 101, 102, 103,
- 104, 105, 106, 107, 108, 109, 110, 111,
- 112, 113, 114, 115, 116, 117, 118, 119,
- 120, 121, 122, 123, 124, 125, 126, 127,
- 128, 129, 130, 131, 132, 133, 134, 135,
- 136, 137, 138, 139, 140, 141, 142, 143,
- 144, 145, 146, 147, 148, 149, 150, 151,
- 152, 153, 154, 155, 156, 157, 158, 159,
- 160, 161, 162, 163, 164, 165, 166, 167,
- 168, 169, 170, 171, 172, 173, 174, 175,
- 176, 177, 178, 179, 180, 181, 182, 183,
- 184, 184, 185, 185, 186, 186, 187, 187,
- 188, 188, 189, 189, 190, 190, 191, 191,
- 192, 192, 193, 193, 194, 194, 194, 194,
- 195, 195, 196, 196, 197, 197, 198, 198
-};
-
-/*
-const int vp8_gf_boost_qadjustment[QINDEX_RANGE] =
-{
- 100,101,102,103,104,105,105,106,
- 106,107,107,108,109,109,110,111,
- 112,113,114,115,116,117,118,119,
- 120,121,122,123,124,125,126,127,
- 128,129,130,131,132,133,134,135,
- 136,137,138,139,140,141,142,143,
- 144,145,146,147,148,149,150,151,
- 152,153,154,155,156,157,158,159,
- 160,161,162,163,164,165,166,167,
- 168,169,170,170,171,171,172,172,
- 173,173,173,174,174,174,175,175,
- 175,176,176,176,177,177,177,177,
- 178,178,179,179,180,180,181,181,
- 182,182,183,183,184,184,185,185,
- 186,186,187,187,188,188,189,189,
- 190,190,191,191,192,192,193,193,
-};
-*/
-
-static const int kf_gf_boost_qlimits[QINDEX_RANGE] =
-{
- 150, 155, 160, 165, 170, 175, 180, 185,
- 190, 195, 200, 205, 210, 215, 220, 225,
- 230, 235, 240, 245, 250, 255, 260, 265,
- 270, 275, 280, 285, 290, 295, 300, 305,
- 310, 320, 330, 340, 350, 360, 370, 380,
- 390, 400, 410, 420, 430, 440, 450, 460,
- 470, 480, 490, 500, 510, 520, 530, 540,
- 550, 560, 570, 580, 590, 600, 600, 600,
- 600, 600, 600, 600, 600, 600, 600, 600,
- 600, 600, 600, 600, 600, 600, 600, 600,
- 600, 600, 600, 600, 600, 600, 600, 600,
- 600, 600, 600, 600, 600, 600, 600, 600,
- 600, 600, 600, 600, 600, 600, 600, 600,
- 600, 600, 600, 600, 600, 600, 600, 600,
- 600, 600, 600, 600, 600, 600, 600, 600,
- 600, 600, 600, 600, 600, 600, 600, 600,
-};
-
-// % adjustment to target kf size based on seperation from previous frame
-static const int kf_boost_seperation_adjustment[16] =
-{
- 30, 40, 50, 55, 60, 65, 70, 75,
- 80, 85, 90, 95, 100, 100, 100, 100,
-};
-
-
-static const int gf_adjust_table[101] =
-{
- 100,
- 115, 130, 145, 160, 175, 190, 200, 210, 220, 230,
- 240, 260, 270, 280, 290, 300, 310, 320, 330, 340,
- 350, 360, 370, 380, 390, 400, 400, 400, 400, 400,
- 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
- 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
- 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
- 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
- 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
- 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
- 400, 400, 400, 400, 400, 400, 400, 400, 400, 400,
-};
-
-static const int gf_intra_usage_adjustment[20] =
-{
- 125, 120, 115, 110, 105, 100, 95, 85, 80, 75,
- 70, 65, 60, 55, 50, 50, 50, 50, 50, 50,
-};
-
-static const int gf_interval_table[101] =
-{
- 7,
- 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
- 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
- 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
- 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
- 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
- 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
- 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
- 10, 10, 10, 10, 10, 10, 10, 10, 10, 10,
- 10, 10, 10, 10, 10, 10, 10, 10, 10, 10,
- 11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
-};
-
-static const unsigned int prior_key_frame_weight[KEY_FRAME_CONTEXT] = { 1, 2, 3, 4, 5 };
-
-
-void vp8_save_coding_context(VP8_COMP *cpi)
-{
- CODING_CONTEXT *const cc = & cpi->coding_context;
-
- // Stores a snapshot of key state variables which can subsequently be
- // restored with a call to vp8_restore_coding_context. These functions are
- // intended for use in a re-code loop in vp8_compress_frame where the
- // quantizer value is adjusted between loop iterations.
-
- cc->frames_since_key = cpi->frames_since_key;
- cc->filter_level = cpi->common.filter_level;
- cc->frames_till_gf_update_due = cpi->frames_till_gf_update_due;
- cc->frames_since_golden = cpi->common.frames_since_golden;
-
- vp8_copy(cc->mvc, cpi->common.fc.mvc);
- vp8_copy(cc->mvcosts, cpi->mb.mvcosts);
-
- vp8_copy(cc->kf_ymode_prob, cpi->common.kf_ymode_prob);
- vp8_copy(cc->ymode_prob, cpi->common.fc.ymode_prob);
- vp8_copy(cc->kf_uv_mode_prob, cpi->common.kf_uv_mode_prob);
- vp8_copy(cc->uv_mode_prob, cpi->common.fc.uv_mode_prob);
-
- vp8_copy(cc->ymode_count, cpi->ymode_count);
- vp8_copy(cc->uv_mode_count, cpi->uv_mode_count);
-
-
- // Stats
-#ifdef MODE_STATS
- vp8_copy(cc->y_modes, y_modes);
- vp8_copy(cc->uv_modes, uv_modes);
- vp8_copy(cc->b_modes, b_modes);
- vp8_copy(cc->inter_y_modes, inter_y_modes);
- vp8_copy(cc->inter_uv_modes, inter_uv_modes);
- vp8_copy(cc->inter_b_modes, inter_b_modes);
-#endif
-
- cc->this_frame_percent_intra = cpi->this_frame_percent_intra;
-}
-
-
-void vp8_restore_coding_context(VP8_COMP *cpi)
-{
- CODING_CONTEXT *const cc = & cpi->coding_context;
-
- // Restore key state variables to the snapshot state stored in the
- // previous call to vp8_save_coding_context.
-
- cpi->frames_since_key = cc->frames_since_key;
- cpi->common.filter_level = cc->filter_level;
- cpi->frames_till_gf_update_due = cc->frames_till_gf_update_due;
- cpi->common.frames_since_golden = cc->frames_since_golden;
-
- vp8_copy(cpi->common.fc.mvc, cc->mvc);
-
- vp8_copy(cpi->mb.mvcosts, cc->mvcosts);
-
- vp8_copy(cpi->common.kf_ymode_prob, cc->kf_ymode_prob);
- vp8_copy(cpi->common.fc.ymode_prob, cc->ymode_prob);
- vp8_copy(cpi->common.kf_uv_mode_prob, cc->kf_uv_mode_prob);
- vp8_copy(cpi->common.fc.uv_mode_prob, cc->uv_mode_prob);
-
- vp8_copy(cpi->ymode_count, cc->ymode_count);
- vp8_copy(cpi->uv_mode_count, cc->uv_mode_count);
-
- // Stats
-#ifdef MODE_STATS
- vp8_copy(y_modes, cc->y_modes);
- vp8_copy(uv_modes, cc->uv_modes);
- vp8_copy(b_modes, cc->b_modes);
- vp8_copy(inter_y_modes, cc->inter_y_modes);
- vp8_copy(inter_uv_modes, cc->inter_uv_modes);
- vp8_copy(inter_b_modes, cc->inter_b_modes);
-#endif
-
-
- cpi->this_frame_percent_intra = cc->this_frame_percent_intra;
-}
-
-
-void vp8_setup_key_frame(VP8_COMP *cpi)
-{
- // Setup for Key frame:
-
- vp8_default_coef_probs(& cpi->common);
- vp8_kf_default_bmode_probs(cpi->common.kf_bmode_prob);
-
- vpx_memcpy(cpi->common.fc.mvc, vp8_default_mv_context, sizeof(vp8_default_mv_context));
- {
- int flag[2] = {1, 1};
- vp8_build_component_cost_table(cpi->mb.mvcost, cpi->mb.mvsadcost, (const MV_CONTEXT *) cpi->common.fc.mvc, flag);
- }
-
- vpx_memset(cpi->common.fc.pre_mvc, 0, sizeof(cpi->common.fc.pre_mvc)); //initialize pre_mvc to all zero.
-
- //cpi->common.filter_level = 0; // Reset every key frame.
- cpi->common.filter_level = cpi->common.base_qindex * 3 / 8 ;
-
- // Provisional interval before next GF
- if (cpi->auto_gold)
- //cpi->frames_till_gf_update_due = DEFAULT_GF_INTERVAL;
- cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
- else
- cpi->frames_till_gf_update_due = cpi->goldfreq;
-
- cpi->common.refresh_golden_frame = TRUE;
- cpi->common.refresh_alt_ref_frame = TRUE;
-}
-
-void vp8_calc_auto_iframe_target_size(VP8_COMP *cpi)
-{
- // boost defaults to half second
- int kf_boost;
-
- // Clear down mmx registers to allow floating point in what follows
- vp8_clear_system_state(); //__asm emms;
-
- if (cpi->oxcf.fixed_q >= 0)
- {
- vp8_calc_iframe_target_size(cpi);
- return;
- }
-
- if (cpi->pass == 2)
- {
- cpi->this_frame_target = cpi->per_frame_bandwidth; // New Two pass RC
- }
- else
- {
- // Boost depends somewhat on frame rate
- kf_boost = (int)(2 * cpi->output_frame_rate - 16);
-
- // adjustment up based on q
- kf_boost = kf_boost * kf_boost_qadjustment[cpi->ni_av_qi] / 100;
-
- // frame separation adjustment ( down)
- if (cpi->frames_since_key < cpi->output_frame_rate / 2)
- kf_boost = (int)(kf_boost * cpi->frames_since_key / (cpi->output_frame_rate / 2));
-
- if (kf_boost < 16)
- kf_boost = 16;
-
- // Reset the active worst quality to the baseline value for key frames.
- cpi->active_worst_quality = cpi->worst_quality;
-
- cpi->this_frame_target = ((16 + kf_boost) * cpi->per_frame_bandwidth) >> 4;
- }
-
-
- // Should the next frame be an altref frame
- if (cpi->pass != 2)
- {
- // For now Alt ref is not allowed except in 2 pass modes.
- cpi->source_alt_ref_pending = FALSE;
-
- /*if ( cpi->oxcf.fixed_q == -1)
- {
- if ( cpi->oxcf.play_alternate && ( (cpi->last_boost/2) > (100+(AF_THRESH*cpi->frames_till_gf_update_due)) ) )
- cpi->source_alt_ref_pending = TRUE;
- else
- cpi->source_alt_ref_pending = FALSE;
- }*/
- }
-
- if (0)
- {
- FILE *f;
-
- f = fopen("kf_boost.stt", "a");
- //fprintf(f, " %8d %10d %10d %10d %10d %10d %10d\n",
- // cpi->common.current_video_frame, cpi->target_bandwidth, cpi->frames_to_key, kf_boost_qadjustment[cpi->ni_av_qi], cpi->kf_boost, (cpi->this_frame_target *100 / cpi->per_frame_bandwidth), cpi->this_frame_target );
-
- fprintf(f, " %8u %10d %10d %10d\n",
- cpi->common.current_video_frame, cpi->gfu_boost, cpi->baseline_gf_interval, cpi->source_alt_ref_pending);
-
- fclose(f);
- }
-}
-
-// Do the best we can to define the parameteres for the next GF based on what information we have available.
-static void calc_gf_params(VP8_COMP *cpi)
-{
- int Q = (cpi->oxcf.fixed_q < 0) ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
- int Boost = 0;
-
- int gf_frame_useage = 0; // Golden frame useage since last GF
- int tot_mbs = cpi->recent_ref_frame_usage[INTRA_FRAME] +
- cpi->recent_ref_frame_usage[LAST_FRAME] +
- cpi->recent_ref_frame_usage[GOLDEN_FRAME] +
- cpi->recent_ref_frame_usage[ALTREF_FRAME];
-
- int pct_gf_active = (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols);
-
- // Reset the last boost indicator
- //cpi->last_boost = 100;
-
- if (tot_mbs)
- gf_frame_useage = (cpi->recent_ref_frame_usage[GOLDEN_FRAME] + cpi->recent_ref_frame_usage[ALTREF_FRAME]) * 100 / tot_mbs;
-
- if (pct_gf_active > gf_frame_useage)
- gf_frame_useage = pct_gf_active;
-
- // Not two pass
- if (cpi->pass != 2)
- {
- // Single Pass lagged mode: TBD
- if (FALSE)
- {
- }
-
- // Single Pass compression: Has to use current and historical data
- else
- {
-#if 0
- // Experimental code
- int index = cpi->one_pass_frame_index;
- int frames_to_scan = (cpi->max_gf_interval <= MAX_LAG_BUFFERS) ? cpi->max_gf_interval : MAX_LAG_BUFFERS;
-
- /*
- // *************** Experimental code - incomplete
- double decay_val = 1.0;
- double IIAccumulator = 0.0;
- double last_iiaccumulator = 0.0;
- double IIRatio;
-
- cpi->one_pass_frame_index = cpi->common.current_video_frame%MAX_LAG_BUFFERS;
-
- for ( i = 0; i < (frames_to_scan - 1); i++ )
- {
- if ( index < 0 )
- index = MAX_LAG_BUFFERS;
- index --;
-
- if ( cpi->one_pass_frame_stats[index].frame_coded_error > 0.0 )
- {
- IIRatio = cpi->one_pass_frame_stats[index].frame_intra_error / cpi->one_pass_frame_stats[index].frame_coded_error;
-
- if ( IIRatio > 30.0 )
- IIRatio = 30.0;
- }
- else
- IIRatio = 30.0;
-
- IIAccumulator += IIRatio * decay_val;
-
- decay_val = decay_val * cpi->one_pass_frame_stats[index].frame_pcnt_inter;
-
- if ( (i > MIN_GF_INTERVAL) &&
- ((IIAccumulator - last_iiaccumulator) < 2.0) )
- {
- break;
- }
- last_iiaccumulator = IIAccumulator;
- }
-
- Boost = IIAccumulator*100.0/16.0;
- cpi->baseline_gf_interval = i;
-
- */
-#else
-
- /*************************************************************/
- // OLD code
-
- // Adjust boost based upon ambient Q
- Boost = GFQ_ADJUSTMENT;
-
- // Adjust based upon most recently measure intra useage
- Boost = Boost * gf_intra_usage_adjustment[(cpi->this_frame_percent_intra < 15) ? cpi->this_frame_percent_intra : 14] / 100;
-
- // Adjust gf boost based upon GF usage since last GF
- Boost = Boost * gf_adjust_table[gf_frame_useage] / 100;
-#endif
- }
-
- // golden frame boost without recode loop often goes awry. be safe by keeping numbers down.
- if (!cpi->sf.recode_loop)
- {
- if (cpi->compressor_speed == 2)
- Boost = Boost / 2;
- }
-
- // Apply an upper limit based on Q for 1 pass encodes
- if (Boost > kf_gf_boost_qlimits[Q] && (cpi->pass == 0))
- Boost = kf_gf_boost_qlimits[Q];
-
- // Apply lower limits to boost.
- else if (Boost < 110)
- Boost = 110;
-
- // Note the boost used
- cpi->last_boost = Boost;
-
- }
-
- // Estimate next interval
- // This is updated once the real frame size/boost is known.
- if (cpi->oxcf.fixed_q == -1)
- {
- if (cpi->pass == 2) // 2 Pass
- {
- cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
- }
- else // 1 Pass
- {
- cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
-
- if (cpi->last_boost > 750)
- cpi->frames_till_gf_update_due++;
-
- if (cpi->last_boost > 1000)
- cpi->frames_till_gf_update_due++;
-
- if (cpi->last_boost > 1250)
- cpi->frames_till_gf_update_due++;
-
- if (cpi->last_boost >= 1500)
- cpi->frames_till_gf_update_due ++;
-
- if (gf_interval_table[gf_frame_useage] > cpi->frames_till_gf_update_due)
- cpi->frames_till_gf_update_due = gf_interval_table[gf_frame_useage];
-
- if (cpi->frames_till_gf_update_due > cpi->max_gf_interval)
- cpi->frames_till_gf_update_due = cpi->max_gf_interval;
- }
- }
- else
- cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
-
- // ARF on or off
- if (cpi->pass != 2)
- {
- // For now Alt ref is not allowed except in 2 pass modes.
- cpi->source_alt_ref_pending = FALSE;
-
- /*if ( cpi->oxcf.fixed_q == -1)
- {
- if ( cpi->oxcf.play_alternate && (cpi->last_boost > (100 + (AF_THRESH*cpi->frames_till_gf_update_due)) ) )
- cpi->source_alt_ref_pending = TRUE;
- else
- cpi->source_alt_ref_pending = FALSE;
- }*/
- }
-}
-/* This is equvialent to estimate_bits_at_q without the rate_correction_factor. */
-static int baseline_bits_at_q(int frame_kind, int Q, int MBs)
-{
- int Bpm = vp8_bits_per_mb[frame_kind][Q];
-
- /* Attempt to retain reasonable accuracy without overflow. The cutoff is
- * chosen such that the maximum product of Bpm and MBs fits 31 bits. The
- * largest Bpm takes 20 bits.
- */
- if (MBs > (1 << 11))
- return (Bpm >> BPER_MB_NORMBITS) * MBs;
- else
- return (Bpm * MBs) >> BPER_MB_NORMBITS;
-}
-
-void vp8_calc_iframe_target_size(VP8_COMP *cpi)
-{
- int Q;
- int Boost = 100;
-
- Q = (cpi->oxcf.fixed_q >= 0) ? cpi->oxcf.fixed_q : cpi->avg_frame_qindex;
-
- if (cpi->auto_adjust_key_quantizer == 1)
- {
- // If (auto_adjust_key_quantizer==1) then a lower Q is selected for key-frames.
- // The enhanced Q is calculated so as to boost the key frame size by a factor
- // specified in kf_boost_qadjustment. Also, can adjust based on distance
- // between key frames.
-
- // Adjust boost based upon ambient Q
- Boost = kf_boost_qadjustment[Q];
-
- // Make the Key frame boost less if the seperation from the previous key frame is small
- if (cpi->frames_since_key < 16)
- Boost = Boost * kf_boost_seperation_adjustment[cpi->frames_since_key] / 100;
- else
- Boost = Boost * kf_boost_seperation_adjustment[15] / 100;
-
- // Apply limits on boost
- if (Boost > kf_gf_boost_qlimits[Q])
- Boost = kf_gf_boost_qlimits[Q];
- else if (Boost < 120)
- Boost = 120;
- }
-
- // Keep a record of the boost that was used
- cpi->last_boost = Boost;
-
- // Should the next frame be an altref frame
- if (cpi->pass != 2)
- {
- // For now Alt ref is not allowed except in 2 pass modes.
- cpi->source_alt_ref_pending = FALSE;
-
- /*if ( cpi->oxcf.fixed_q == -1)
- {
- if ( cpi->oxcf.play_alternate && ( (cpi->last_boost/2) > (100+(AF_THRESH*cpi->frames_till_gf_update_due)) ) )
- cpi->source_alt_ref_pending = TRUE;
- else
- cpi->source_alt_ref_pending = FALSE;
- }*/
- }
-
- if (cpi->oxcf.fixed_q >= 0)
- {
- cpi->this_frame_target = (baseline_bits_at_q(0, Q, cpi->common.MBs) * Boost) / 100;
- }
- else
- {
-
- int bits_per_mb_at_this_q ;
-
- if (cpi->oxcf.error_resilient_mode == 1)
- {
- cpi->this_frame_target = 2 * cpi->av_per_frame_bandwidth;
- return;
- }
-
- // Rate targetted scenario:
- // Be careful of 32-bit OVERFLOW if restructuring the caluclation of cpi->this_frame_target
- bits_per_mb_at_this_q = (int)(.5 +
- cpi->key_frame_rate_correction_factor * vp8_bits_per_mb[0][Q]);
-
- cpi->this_frame_target = (((bits_per_mb_at_this_q * cpi->common.MBs) >> BPER_MB_NORMBITS) * Boost) / 100;
-
- // Reset the active worst quality to the baseline value for key frames.
- if (cpi->pass < 2)
- cpi->active_worst_quality = cpi->worst_quality;
- }
-}
-
-
-
-void vp8_calc_pframe_target_size(VP8_COMP *cpi)
-{
- int min_frame_target;
- int Adjustment;
-
- // Set the min frame bandwidth.
- //min_frame_target = estimate_min_frame_size( cpi );
- min_frame_target = 0;
-
- if (cpi->pass == 2)
- {
- min_frame_target = cpi->min_frame_bandwidth;
-
- if (min_frame_target < (cpi->av_per_frame_bandwidth >> 5))
- min_frame_target = cpi->av_per_frame_bandwidth >> 5;
- }
- else if (min_frame_target < cpi->per_frame_bandwidth / 4)
- min_frame_target = cpi->per_frame_bandwidth / 4;
-
-
- // Special alt reference frame case
- if (cpi->common.refresh_alt_ref_frame)
- {
- if (cpi->pass == 2)
- {
- cpi->per_frame_bandwidth = cpi->gf_bits; // Per frame bit target for the alt ref frame
- cpi->this_frame_target = cpi->per_frame_bandwidth;
- }
-
- /* One Pass ??? TBD */
- /*else
- {
- int frames_in_section;
- int allocation_chunks;
- int Q = (cpi->oxcf.fixed_q < 0) ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
- int alt_boost;
- int max_arf_rate;
-
- alt_boost = (cpi->gfu_boost * 3 * GFQ_ADJUSTMENT) / (2 * 100);
- alt_boost += (cpi->frames_till_gf_update_due * 50);
-
- // If alt ref is not currently active then we have a pottential double hit with GF and ARF so reduce the boost a bit.
- // A similar thing is done on GFs that preceed a arf update.
- if ( !cpi->source_alt_ref_active )
- alt_boost = alt_boost * 3 / 4;
-
- frames_in_section = cpi->frames_till_gf_update_due+1; // Standard frames + GF
- allocation_chunks = (frames_in_section * 100) + alt_boost;
-
- // Normalize Altboost and allocations chunck down to prevent overflow
- while ( alt_boost > 1000 )
- {
- alt_boost /= 2;
- allocation_chunks /= 2;
- }
-
- else
- {
- int bits_in_section;
-
- if ( cpi->kf_overspend_bits > 0 )
- {
- Adjustment = (cpi->kf_bitrate_adjustment <= cpi->kf_overspend_bits) ? cpi->kf_bitrate_adjustment : cpi->kf_overspend_bits;
-
- if ( Adjustment > (cpi->per_frame_bandwidth - min_frame_target) )
- Adjustment = (cpi->per_frame_bandwidth - min_frame_target);
-
- cpi->kf_overspend_bits -= Adjustment;
-
- // Calculate an inter frame bandwidth target for the next few frames designed to recover
- // any extra bits spent on the key frame.
- cpi->inter_frame_target = cpi->per_frame_bandwidth - Adjustment;
- if ( cpi->inter_frame_target < min_frame_target )
- cpi->inter_frame_target = min_frame_target;
- }
- else
- cpi->inter_frame_target = cpi->per_frame_bandwidth;
-
- bits_in_section = cpi->inter_frame_target * frames_in_section;
-
- // Avoid loss of precision but avoid overflow
- if ( (bits_in_section>>7) > allocation_chunks )
- cpi->this_frame_target = alt_boost * (bits_in_section / allocation_chunks);
- else
- cpi->this_frame_target = (alt_boost * bits_in_section) / allocation_chunks;
- }
- }
- */
- }
-
- // Normal frames (gf,and inter)
- else
- {
- // 2 pass
- if (cpi->pass == 2)
- {
- cpi->this_frame_target = cpi->per_frame_bandwidth;
- }
- // 1 pass
- else
- {
- // Make rate adjustment to recover bits spent in key frame
- // Test to see if the key frame inter data rate correction should still be in force
- if (cpi->kf_overspend_bits > 0)
- {
- Adjustment = (cpi->kf_bitrate_adjustment <= cpi->kf_overspend_bits) ? cpi->kf_bitrate_adjustment : cpi->kf_overspend_bits;
-
- if (Adjustment > (cpi->per_frame_bandwidth - min_frame_target))
- Adjustment = (cpi->per_frame_bandwidth - min_frame_target);
-
- cpi->kf_overspend_bits -= Adjustment;
-
- // Calculate an inter frame bandwidth target for the next few frames designed to recover
- // any extra bits spent on the key frame.
- cpi->this_frame_target = cpi->per_frame_bandwidth - Adjustment;
-
- if (cpi->this_frame_target < min_frame_target)
- cpi->this_frame_target = min_frame_target;
- }
- else
- cpi->this_frame_target = cpi->per_frame_bandwidth;
-
- // If appropriate make an adjustment to recover bits spent on a recent GF
- if ((cpi->gf_overspend_bits > 0) && (cpi->this_frame_target > min_frame_target))
- {
- int Adjustment = (cpi->non_gf_bitrate_adjustment <= cpi->gf_overspend_bits) ? cpi->non_gf_bitrate_adjustment : cpi->gf_overspend_bits;
-
- if (Adjustment > (cpi->this_frame_target - min_frame_target))
- Adjustment = (cpi->this_frame_target - min_frame_target);
-
- cpi->gf_overspend_bits -= Adjustment;
- cpi->this_frame_target -= Adjustment;
- }
-
- // Apply small + and - boosts for non gf frames
- if ((cpi->last_boost > 150) && (cpi->frames_till_gf_update_due > 0) &&
- (cpi->current_gf_interval >= (MIN_GF_INTERVAL << 1)))
- {
- // % Adjustment limited to the range 1% to 10%
- Adjustment = (cpi->last_boost - 100) >> 5;
-
- if (Adjustment < 1)
- Adjustment = 1;
- else if (Adjustment > 10)
- Adjustment = 10;
-
- // Convert to bits
- Adjustment = (cpi->this_frame_target * Adjustment) / 100;
-
- if (Adjustment > (cpi->this_frame_target - min_frame_target))
- Adjustment = (cpi->this_frame_target - min_frame_target);
-
- if (cpi->common.frames_since_golden == (cpi->current_gf_interval >> 1))
- cpi->this_frame_target += ((cpi->current_gf_interval - 1) * Adjustment);
- else
- cpi->this_frame_target -= Adjustment;
- }
- }
- }
-
- // Set a reduced data rate target for our initial Q calculation.
- // This should help to save bits during earier sections.
- if ((cpi->oxcf.under_shoot_pct > 0) && (cpi->oxcf.under_shoot_pct <= 100))
- cpi->this_frame_target = (cpi->this_frame_target * cpi->oxcf.under_shoot_pct) / 100;
-
- // Sanity check that the total sum of adjustments is not above the maximum allowed
- // That is that having allowed for KF and GF penalties we have not pushed the
- // current interframe target to low. If the adjustment we apply here is not capable of recovering
- // all the extra bits we have spent in the KF or GF then the remainder will have to be recovered over
- // a longer time span via other buffer / rate control mechanisms.
- if (cpi->this_frame_target < min_frame_target)
- cpi->this_frame_target = min_frame_target;
-
- if (!cpi->common.refresh_alt_ref_frame)
- // Note the baseline target data rate for this inter frame.
- cpi->inter_frame_target = cpi->this_frame_target;
-
- // One Pass specific code
- if (cpi->pass == 0)
- {
- // Adapt target frame size with respect to any buffering constraints:
- if (cpi->buffered_mode)
- {
- int one_percent_bits = 1 + cpi->oxcf.optimal_buffer_level / 100;
-
- if ((cpi->buffer_level < cpi->oxcf.optimal_buffer_level) ||
- (cpi->bits_off_target < cpi->oxcf.optimal_buffer_level))
- {
- int percent_low = 0;
-
- // Decide whether or not we need to adjust the frame data rate target.
- //
- // If we are are below the optimal buffer fullness level and adherence
- // to buffering contraints is important to the end useage then adjust
- // the per frame target.
- if ((cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
- (cpi->buffer_level < cpi->oxcf.optimal_buffer_level))
- {
- percent_low =
- (cpi->oxcf.optimal_buffer_level - cpi->buffer_level) /
- one_percent_bits;
-
- if (percent_low > 100)
- percent_low = 100;
- else if (percent_low < 0)
- percent_low = 0;
- }
- // Are we overshooting the long term clip data rate...
- else if (cpi->bits_off_target < 0)
- {
- // Adjust per frame data target downwards to compensate.
- percent_low = (int)(100 * -cpi->bits_off_target /
- (cpi->total_byte_count * 8));
-
- if (percent_low > 100)
- percent_low = 100;
- else if (percent_low < 0)
- percent_low = 0;
- }
-
- // lower the target bandwidth for this frame.
- cpi->this_frame_target =
- (cpi->this_frame_target * (100 - (percent_low / 2))) / 100;
-
- // Are we using allowing control of active_worst_allowed_q
- // according to buffer level.
- if (cpi->auto_worst_q)
- {
- int critical_buffer_level;
-
- // For streaming applications the most important factor is
- // cpi->buffer_level as this takes into account the
- // specified short term buffering constraints. However,
- // hitting the long term clip data rate target is also
- // important.
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- // Take the smaller of cpi->buffer_level and
- // cpi->bits_off_target
- critical_buffer_level =
- (cpi->buffer_level < cpi->bits_off_target)
- ? cpi->buffer_level : cpi->bits_off_target;
- }
- // For local file playback short term buffering contraints
- // are less of an issue
- else
- {
- // Consider only how we are doing for the clip as a
- // whole
- critical_buffer_level = cpi->bits_off_target;
- }
-
- // Set the active worst quality based upon the selected
- // buffer fullness number.
- if (critical_buffer_level < cpi->oxcf.optimal_buffer_level)
- {
- if ( critical_buffer_level >
- (cpi->oxcf.optimal_buffer_level >> 2) )
- {
- INT64 qadjustment_range =
- cpi->worst_quality - cpi->ni_av_qi;
- INT64 above_base =
- (critical_buffer_level -
- (cpi->oxcf.optimal_buffer_level >> 2));
-
- // Step active worst quality down from
- // cpi->ni_av_qi when (critical_buffer_level ==
- // cpi->optimal_buffer_level) to
- // cpi->worst_quality when
- // (critical_buffer_level ==
- // cpi->optimal_buffer_level >> 2)
- cpi->active_worst_quality =
- cpi->worst_quality -
- ((qadjustment_range * above_base) /
- (cpi->oxcf.optimal_buffer_level*3>>2));
- }
- else
- {
- cpi->active_worst_quality = cpi->worst_quality;
- }
- }
- else
- {
- cpi->active_worst_quality = cpi->ni_av_qi;
- }
- }
- else
- {
- cpi->active_worst_quality = cpi->worst_quality;
- }
- }
- else
- {
- int percent_high;
-
- if (cpi->bits_off_target > cpi->oxcf.optimal_buffer_level)
- {
- percent_high = (int)(100 * (cpi->bits_off_target - cpi->oxcf.optimal_buffer_level) / (cpi->total_byte_count * 8));
-
- if (percent_high > 100)
- percent_high = 100;
- else if (percent_high < 0)
- percent_high = 0;
-
- cpi->this_frame_target = (cpi->this_frame_target * (100 + (percent_high / 2))) / 100;
-
- }
-
- // Are we allowing control of active_worst_allowed_q according to bufferl level.
- if (cpi->auto_worst_q)
- {
- // When using the relaxed buffer model stick to the user specified value
- cpi->active_worst_quality = cpi->ni_av_qi;
- }
- else
- {
- cpi->active_worst_quality = cpi->worst_quality;
- }
- }
-
- // Set active_best_quality to prevent quality rising too high
- cpi->active_best_quality = cpi->best_quality;
-
- // Worst quality obviously must not be better than best quality
- if (cpi->active_worst_quality <= cpi->active_best_quality)
- cpi->active_worst_quality = cpi->active_best_quality + 1;
-
- }
- // Unbuffered mode (eg. video conferencing)
- else
- {
- // Set the active worst quality
- cpi->active_worst_quality = cpi->worst_quality;
- }
-
- // Special trap for constrained quality mode
- // "active_worst_quality" may never drop below cq level
- // for any frame type.
- if ( cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY &&
- cpi->active_worst_quality < cpi->cq_target_quality)
- {
- cpi->active_worst_quality = cpi->cq_target_quality;
- }
- }
-
- // Test to see if we have to drop a frame
- // The auto-drop frame code is only used in buffered mode.
- // In unbufferd mode (eg vide conferencing) the descision to
- // code or drop a frame is made outside the codec in response to real
- // world comms or buffer considerations.
- if (cpi->drop_frames_allowed && cpi->buffered_mode &&
- (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
- ((cpi->common.frame_type != KEY_FRAME))) //|| !cpi->oxcf.allow_spatial_resampling) )
- {
- // Check for a buffer underun-crisis in which case we have to drop a frame
- if ((cpi->buffer_level < 0))
- {
-#if 0
- FILE *f = fopen("dec.stt", "a");
- fprintf(f, "%10d %10d %10d %10d ***** BUFFER EMPTY\n",
- (int) cpi->common.current_video_frame,
- cpi->decimation_factor, cpi->common.horiz_scale,
- (cpi->buffer_level * 100) / cpi->oxcf.optimal_buffer_level);
- fclose(f);
-#endif
- //vpx_log("Decoder: Drop frame due to bandwidth: %d \n",cpi->buffer_level, cpi->av_per_frame_bandwidth);
-
- cpi->drop_frame = TRUE;
- }
-
-#if 0
- // Check for other drop frame crtieria (Note 2 pass cbr uses decimation on whole KF sections)
- else if ((cpi->buffer_level < cpi->oxcf.drop_frames_water_mark * cpi->oxcf.optimal_buffer_level / 100) &&
- (cpi->drop_count < cpi->max_drop_count) && (cpi->pass == 0))
- {
- cpi->drop_frame = TRUE;
- }
-
-#endif
-
- if (cpi->drop_frame)
- {
- // Update the buffer level variable.
- cpi->bits_off_target += cpi->av_per_frame_bandwidth;
- cpi->buffer_level = cpi->bits_off_target;
- }
- else
- cpi->drop_count = 0;
- }
-
- // Adjust target frame size for Golden Frames:
- if (cpi->oxcf.error_resilient_mode == 0 &&
- (cpi->frames_till_gf_update_due == 0) && !cpi->drop_frame)
- {
- //int Boost = 0;
- int Q = (cpi->oxcf.fixed_q < 0) ? cpi->last_q[INTER_FRAME] : cpi->oxcf.fixed_q;
-
- int gf_frame_useage = 0; // Golden frame useage since last GF
- int tot_mbs = cpi->recent_ref_frame_usage[INTRA_FRAME] +
- cpi->recent_ref_frame_usage[LAST_FRAME] +
- cpi->recent_ref_frame_usage[GOLDEN_FRAME] +
- cpi->recent_ref_frame_usage[ALTREF_FRAME];
-
- int pct_gf_active = (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols);
-
- // Reset the last boost indicator
- //cpi->last_boost = 100;
-
- if (tot_mbs)
- gf_frame_useage = (cpi->recent_ref_frame_usage[GOLDEN_FRAME] + cpi->recent_ref_frame_usage[ALTREF_FRAME]) * 100 / tot_mbs;
-
- if (pct_gf_active > gf_frame_useage)
- gf_frame_useage = pct_gf_active;
-
- // Is a fixed manual GF frequency being used
- if (cpi->auto_gold)
- {
- // For one pass throw a GF if recent frame intra useage is low or the GF useage is high
- if ((cpi->pass == 0) && (cpi->this_frame_percent_intra < 15 || gf_frame_useage >= 5))
- cpi->common.refresh_golden_frame = TRUE;
-
- // Two pass GF descision
- else if (cpi->pass == 2)
- cpi->common.refresh_golden_frame = TRUE;
- }
-
-#if 0
-
- // Debug stats
- if (0)
- {
- FILE *f;
-
- f = fopen("gf_useaget.stt", "a");
- fprintf(f, " %8ld %10ld %10ld %10ld %10ld\n",
- cpi->common.current_video_frame, cpi->gfu_boost, GFQ_ADJUSTMENT, cpi->gfu_boost, gf_frame_useage);
- fclose(f);
- }
-
-#endif
-
- if (cpi->common.refresh_golden_frame == TRUE)
- {
-#if 0
-
- if (0) // p_gw
- {
- FILE *f;
-
- f = fopen("GFexit.stt", "a");
- fprintf(f, "%8ld GF coded\n", cpi->common.current_video_frame);
- fclose(f);
- }
-
-#endif
- cpi->initial_gf_use = 0;
-
- if (cpi->auto_adjust_gold_quantizer)
- {
- calc_gf_params(cpi);
- }
-
- // If we are using alternate ref instead of gf then do not apply the boost
- // It will instead be applied to the altref update
- // Jims modified boost
- if (!cpi->source_alt_ref_active)
- {
- if (cpi->oxcf.fixed_q < 0)
- {
- if (cpi->pass == 2)
- {
- cpi->this_frame_target = cpi->per_frame_bandwidth; // The spend on the GF is defined in the two pass code for two pass encodes
- }
- else
- {
- int Boost = cpi->last_boost;
- int frames_in_section = cpi->frames_till_gf_update_due + 1;
- int allocation_chunks = (frames_in_section * 100) + (Boost - 100);
- int bits_in_section = cpi->inter_frame_target * frames_in_section;
-
- // Normalize Altboost and allocations chunck down to prevent overflow
- while (Boost > 1000)
- {
- Boost /= 2;
- allocation_chunks /= 2;
- }
-
- // Avoid loss of precision but avoid overflow
- if ((bits_in_section >> 7) > allocation_chunks)
- cpi->this_frame_target = Boost * (bits_in_section / allocation_chunks);
- else
- cpi->this_frame_target = (Boost * bits_in_section) / allocation_chunks;
- }
- }
- else
- cpi->this_frame_target = (baseline_bits_at_q(1, Q, cpi->common.MBs) * cpi->last_boost) / 100;
-
- }
- // If there is an active ARF at this location use the minimum
- // bits on this frame even if it is a contructed arf.
- // The active maximum quantizer insures that an appropriate
- // number of bits will be spent if needed for contstructed ARFs.
- else
- {
- cpi->this_frame_target = 0;
- }
-
- cpi->current_gf_interval = cpi->frames_till_gf_update_due;
-
- }
- }
-}
-
-
-void vp8_update_rate_correction_factors(VP8_COMP *cpi, int damp_var)
-{
- int Q = cpi->common.base_qindex;
- int correction_factor = 100;
- double rate_correction_factor;
- double adjustment_limit;
-
- int projected_size_based_on_q = 0;
-
- // Clear down mmx registers to allow floating point in what follows
- vp8_clear_system_state(); //__asm emms;
-
- if (cpi->common.frame_type == KEY_FRAME)
- {
- rate_correction_factor = cpi->key_frame_rate_correction_factor;
- }
- else
- {
- if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
- rate_correction_factor = cpi->gf_rate_correction_factor;
- else
- rate_correction_factor = cpi->rate_correction_factor;
- }
-
- // Work out how big we would have expected the frame to be at this Q given the current correction factor.
- // Stay in double to avoid int overflow when values are large
- //projected_size_based_on_q = ((int)(.5 + rate_correction_factor * vp8_bits_per_mb[cpi->common.frame_type][Q]) * cpi->common.MBs) >> BPER_MB_NORMBITS;
- projected_size_based_on_q = (int)(((.5 + rate_correction_factor * vp8_bits_per_mb[cpi->common.frame_type][Q]) * cpi->common.MBs) / (1 << BPER_MB_NORMBITS));
-
- // Make some allowance for cpi->zbin_over_quant
- if (cpi->zbin_over_quant > 0)
- {
- int Z = cpi->zbin_over_quant;
- double Factor = 0.99;
- double factor_adjustment = 0.01 / 256.0; //(double)ZBIN_OQ_MAX;
-
- while (Z > 0)
- {
- Z --;
- projected_size_based_on_q =
- (int)(Factor * projected_size_based_on_q);
- Factor += factor_adjustment;
-
- if (Factor >= 0.999)
- Factor = 0.999;
- }
- }
-
- // Work out a size correction factor.
- //if ( cpi->this_frame_target > 0 )
- // correction_factor = (100 * cpi->projected_frame_size) / cpi->this_frame_target;
- if (projected_size_based_on_q > 0)
- correction_factor = (100 * cpi->projected_frame_size) / projected_size_based_on_q;
-
- // More heavily damped adjustment used if we have been oscillating either side of target
- switch (damp_var)
- {
- case 0:
- adjustment_limit = 0.75;
- break;
- case 1:
- adjustment_limit = 0.375;
- break;
- case 2:
- default:
- adjustment_limit = 0.25;
- break;
- }
-
- //if ( (correction_factor > 102) && (Q < cpi->active_worst_quality) )
- if (correction_factor > 102)
- {
- // We are not already at the worst allowable quality
- correction_factor = (int)(100.5 + ((correction_factor - 100) * adjustment_limit));
- rate_correction_factor = ((rate_correction_factor * correction_factor) / 100);
-
- // Keep rate_correction_factor within limits
- if (rate_correction_factor > MAX_BPB_FACTOR)
- rate_correction_factor = MAX_BPB_FACTOR;
- }
- //else if ( (correction_factor < 99) && (Q > cpi->active_best_quality) )
- else if (correction_factor < 99)
- {
- // We are not already at the best allowable quality
- correction_factor = (int)(100.5 - ((100 - correction_factor) * adjustment_limit));
- rate_correction_factor = ((rate_correction_factor * correction_factor) / 100);
-
- // Keep rate_correction_factor within limits
- if (rate_correction_factor < MIN_BPB_FACTOR)
- rate_correction_factor = MIN_BPB_FACTOR;
- }
-
- if (cpi->common.frame_type == KEY_FRAME)
- cpi->key_frame_rate_correction_factor = rate_correction_factor;
- else
- {
- if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
- cpi->gf_rate_correction_factor = rate_correction_factor;
- else
- cpi->rate_correction_factor = rate_correction_factor;
- }
-}
-
-static int estimate_bits_at_q(VP8_COMP *cpi, int Q)
-{
- int Bpm = (int)(.5 + cpi->rate_correction_factor * vp8_bits_per_mb[INTER_FRAME][Q]);
-
- /* Attempt to retain reasonable accuracy without overflow. The cutoff is
- * chosen such that the maximum product of Bpm and MBs fits 31 bits. The
- * largest Bpm takes 20 bits.
- */
- if (cpi->common.MBs > (1 << 11))
- return (Bpm >> BPER_MB_NORMBITS) * cpi->common.MBs;
- else
- return (Bpm * cpi->common.MBs) >> BPER_MB_NORMBITS;
-
-}
-
-
-int vp8_regulate_q(VP8_COMP *cpi, int target_bits_per_frame)
-{
- int Q = cpi->active_worst_quality;
-
- // Reset Zbin OQ value
- cpi->zbin_over_quant = 0;
-
- if (cpi->oxcf.fixed_q >= 0)
- {
- Q = cpi->oxcf.fixed_q;
-
- if (cpi->common.frame_type == KEY_FRAME)
- {
- Q = cpi->oxcf.key_q;
- }
- else if (cpi->common.refresh_alt_ref_frame)
- {
- Q = cpi->oxcf.alt_q;
- }
- else if (cpi->common.refresh_golden_frame)
- {
- Q = cpi->oxcf.gold_q;
- }
-
- }
- else
- {
- int i;
- int last_error = INT_MAX;
- int target_bits_per_mb;
- int bits_per_mb_at_this_q;
- double correction_factor;
-
- // Select the appropriate correction factor based upon type of frame.
- if (cpi->common.frame_type == KEY_FRAME)
- correction_factor = cpi->key_frame_rate_correction_factor;
- else
- {
- if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
- correction_factor = cpi->gf_rate_correction_factor;
- else
- correction_factor = cpi->rate_correction_factor;
- }
-
- // Calculate required scaling factor based on target frame size and size of frame produced using previous Q
- if (target_bits_per_frame >= (INT_MAX >> BPER_MB_NORMBITS))
- target_bits_per_mb = (target_bits_per_frame / cpi->common.MBs) << BPER_MB_NORMBITS; // Case where we would overflow int
- else
- target_bits_per_mb = (target_bits_per_frame << BPER_MB_NORMBITS) / cpi->common.MBs;
-
- i = cpi->active_best_quality;
-
- do
- {
- bits_per_mb_at_this_q = (int)(.5 + correction_factor * vp8_bits_per_mb[cpi->common.frame_type][i]);
-
- if (bits_per_mb_at_this_q <= target_bits_per_mb)
- {
- if ((target_bits_per_mb - bits_per_mb_at_this_q) <= last_error)
- Q = i;
- else
- Q = i - 1;
-
- break;
- }
- else
- last_error = bits_per_mb_at_this_q - target_bits_per_mb;
- }
- while (++i <= cpi->active_worst_quality);
-
-
- // If we are at MAXQ then enable Q over-run which seeks to claw back additional bits through things like
- // the RD multiplier and zero bin size.
- if (Q >= MAXQ)
- {
- int zbin_oqmax;
-
- double Factor = 0.99;
- double factor_adjustment = 0.01 / 256.0; //(double)ZBIN_OQ_MAX;
-
- if (cpi->common.frame_type == KEY_FRAME)
- zbin_oqmax = 0; //ZBIN_OQ_MAX/16
- else if (cpi->common.refresh_alt_ref_frame || (cpi->common.refresh_golden_frame && !cpi->source_alt_ref_active))
- zbin_oqmax = 16;
- else
- zbin_oqmax = ZBIN_OQ_MAX;
-
- /*{
- double Factor = (double)target_bits_per_mb/(double)bits_per_mb_at_this_q;
- double Oq;
-
- Factor = Factor/1.2683;
-
- Oq = pow( Factor, (1.0/-0.165) );
-
- if ( Oq > zbin_oqmax )
- Oq = zbin_oqmax;
-
- cpi->zbin_over_quant = (int)Oq;
- }*/
-
- // Each incrment in the zbin is assumed to have a fixed effect on bitrate. This is not of course true.
- // The effect will be highly clip dependent and may well have sudden steps.
- // The idea here is to acheive higher effective quantizers than the normal maximum by expanding the zero
- // bin and hence decreasing the number of low magnitude non zero coefficients.
- while (cpi->zbin_over_quant < zbin_oqmax)
- {
- cpi->zbin_over_quant ++;
-
- if (cpi->zbin_over_quant > zbin_oqmax)
- cpi->zbin_over_quant = zbin_oqmax;
-
- // Adjust bits_per_mb_at_this_q estimate
- bits_per_mb_at_this_q = (int)(Factor * bits_per_mb_at_this_q);
- Factor += factor_adjustment;
-
- if (Factor >= 0.999)
- Factor = 0.999;
-
- if (bits_per_mb_at_this_q <= target_bits_per_mb) // Break out if we get down to the target rate
- break;
- }
-
- }
- }
-
- return Q;
-}
-
-static int estimate_min_frame_size(VP8_COMP *cpi)
-{
- double correction_factor;
- int bits_per_mb_at_max_q;
-
- // This funtion returns a default value for the first few frames untill the correction factor has had time to adapt.
- if (cpi->common.current_video_frame < 10)
- {
- if (cpi->pass == 2)
- return (cpi->min_frame_bandwidth);
- else
- return cpi->per_frame_bandwidth / 3;
- }
-
- /* // Select the appropriate correction factor based upon type of frame.
- if ( cpi->common.frame_type == KEY_FRAME )
- correction_factor = cpi->key_frame_rate_correction_factor;
- else
- {
- if ( cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame )
- correction_factor = cpi->gf_rate_correction_factor;
- else
- correction_factor = cpi->rate_correction_factor;
- }*/
-
- // We estimate at half the value we get from vp8_bits_per_mb
- correction_factor = cpi->rate_correction_factor / 2.0;
-
- bits_per_mb_at_max_q = (int)(.5 + correction_factor * vp8_bits_per_mb[cpi->common.frame_type][MAXQ]);
-
- return (bits_per_mb_at_max_q * cpi->common.MBs) >> BPER_MB_NORMBITS;
-}
-
-void vp8_adjust_key_frame_context(VP8_COMP *cpi)
-{
- int i;
- int av_key_frames_per_second;
-
- // Average key frame frequency and size
- unsigned int total_weight = 0;
- unsigned int av_key_frame_frequency = 0;
- unsigned int av_key_frame_bits = 0;
-
- unsigned int output_frame_rate = (unsigned int)(100 * cpi->output_frame_rate);
- unsigned int target_bandwidth = (unsigned int)(100 * cpi->target_bandwidth);
-
- // Clear down mmx registers to allow floating point in what follows
- vp8_clear_system_state(); //__asm emms;
-
- // Update the count of total key frame bits
- cpi->tot_key_frame_bits += cpi->projected_frame_size;
-
- // First key frame at start of sequence is a special case. We have no frequency data.
- if (cpi->key_frame_count == 1)
- {
- av_key_frame_frequency = (int)cpi->output_frame_rate * 2; // Assume a default of 1 kf every 2 seconds
- av_key_frame_bits = cpi->projected_frame_size;
- av_key_frames_per_second = output_frame_rate / av_key_frame_frequency; // Note output_frame_rate not cpi->output_frame_rate
- }
- else
- {
- int last_kf_interval =
- (cpi->frames_since_key > 0) ? cpi->frames_since_key : 1;
-
- // reset keyframe context and calculate weighted average of last KEY_FRAME_CONTEXT keyframes
- for (i = 0; i < KEY_FRAME_CONTEXT; i++)
- {
- if (i < KEY_FRAME_CONTEXT - 1)
- {
- cpi->prior_key_frame_size[i] = cpi->prior_key_frame_size[i+1];
- cpi->prior_key_frame_distance[i] = cpi->prior_key_frame_distance[i+1];
- }
- else
- {
- cpi->prior_key_frame_size[i] = cpi->projected_frame_size;
- cpi->prior_key_frame_distance[i] = last_kf_interval;
- }
-
- av_key_frame_bits += prior_key_frame_weight[i] * cpi->prior_key_frame_size[i];
- av_key_frame_frequency += prior_key_frame_weight[i] * cpi->prior_key_frame_distance[i];
- total_weight += prior_key_frame_weight[i];
- }
-
- av_key_frame_bits /= total_weight;
- av_key_frame_frequency /= total_weight;
- av_key_frames_per_second = output_frame_rate / av_key_frame_frequency;
-
- }
-
- // Do we have any key frame overspend to recover?
- if ((cpi->pass != 2) && (cpi->projected_frame_size > cpi->per_frame_bandwidth))
- {
- // Update the count of key frame overspend to be recovered in subsequent frames
- // A portion of the KF overspend is treated as gf overspend (and hence recovered more quickly)
- // as the kf is also a gf. Otherwise the few frames following each kf tend to get more bits
- // allocated than those following other gfs.
- cpi->kf_overspend_bits += (cpi->projected_frame_size - cpi->per_frame_bandwidth) * 7 / 8;
- cpi->gf_overspend_bits += (cpi->projected_frame_size - cpi->per_frame_bandwidth) * 1 / 8;
- if(!av_key_frame_frequency)
- av_key_frame_frequency = 60;
-
- // Work out how much to try and recover per frame.
- // For one pass we estimate the number of frames to spread it over based upon past history.
- // For two pass we know how many frames there will be till the next kf.
- if (cpi->pass == 2)
- {
- if (cpi->frames_to_key > 16)
- cpi->kf_bitrate_adjustment = cpi->kf_overspend_bits / (int)cpi->frames_to_key;
- else
- cpi->kf_bitrate_adjustment = cpi->kf_overspend_bits / 16;
- }
- else
- cpi->kf_bitrate_adjustment = cpi->kf_overspend_bits / (int)av_key_frame_frequency;
- }
-
- cpi->frames_since_key = 0;
- cpi->last_key_frame_size = cpi->projected_frame_size;
- cpi->key_frame_count++;
-}
-
-void vp8_compute_frame_size_bounds(VP8_COMP *cpi, int *frame_under_shoot_limit, int *frame_over_shoot_limit)
-{
- // Set-up bounds on acceptable frame size:
- if (cpi->oxcf.fixed_q >= 0)
- {
- // Fixed Q scenario: frame size never outranges target (there is no target!)
- *frame_under_shoot_limit = 0;
- *frame_over_shoot_limit = INT_MAX;
- }
- else
- {
- if (cpi->common.frame_type == KEY_FRAME)
- {
- *frame_over_shoot_limit = cpi->this_frame_target * 9 / 8;
- *frame_under_shoot_limit = cpi->this_frame_target * 7 / 8;
- }
- else
- {
- if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
- {
- *frame_over_shoot_limit = cpi->this_frame_target * 9 / 8;
- *frame_under_shoot_limit = cpi->this_frame_target * 7 / 8;
- }
- else
- {
- // For CBR take buffer fullness into account
- if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
- {
- if (cpi->buffer_level >= ((cpi->oxcf.optimal_buffer_level + cpi->oxcf.maximum_buffer_size) >> 1))
- {
- // Buffer is too full so relax overshoot and tighten undershoot
- *frame_over_shoot_limit = cpi->this_frame_target * 12 / 8;
- *frame_under_shoot_limit = cpi->this_frame_target * 6 / 8;
- }
- else if (cpi->buffer_level <= (cpi->oxcf.optimal_buffer_level >> 1))
- {
- // Buffer is too low so relax undershoot and tighten overshoot
- *frame_over_shoot_limit = cpi->this_frame_target * 10 / 8;
- *frame_under_shoot_limit = cpi->this_frame_target * 4 / 8;
- }
- else
- {
- *frame_over_shoot_limit = cpi->this_frame_target * 11 / 8;
- *frame_under_shoot_limit = cpi->this_frame_target * 5 / 8;
- }
- }
- // VBR and CQ mode
- // Note that tighter restrictions here can help quality but hurt encode speed
- else
- {
- // Stron overshoot limit for constrained quality
- if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
- {
- *frame_over_shoot_limit = cpi->this_frame_target * 11 / 8;
- *frame_under_shoot_limit = cpi->this_frame_target * 2 / 8;
- }
- else
- {
- *frame_over_shoot_limit = cpi->this_frame_target * 11 / 8;
- *frame_under_shoot_limit = cpi->this_frame_target * 5 / 8;
- }
- }
- }
- }
- }
-}
diff --git a/vp8/encoder/ratectrl.h b/vp8/encoder/ratectrl.h
deleted file mode 100644
index 766dfdf..0000000
--- a/vp8/encoder/ratectrl.h
+++ /dev/null
@@ -1,27 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#if !defined __INC_RATECTRL_H
-
-#include "onyx_int.h"
-
-extern void vp8_save_coding_context(VP8_COMP *cpi);
-extern void vp8_restore_coding_context(VP8_COMP *cpi);
-
-extern void vp8_setup_key_frame(VP8_COMP *cpi);
-extern void vp8_calc_iframe_target_size(VP8_COMP *cpi);
-extern void vp8_calc_pframe_target_size(VP8_COMP *cpi);
-extern void vp8_update_rate_correction_factors(VP8_COMP *cpi, int damp_var);
-extern int vp8_regulate_q(VP8_COMP *cpi, int target_bits_per_frame);
-extern void vp8_adjust_key_frame_context(VP8_COMP *cpi);
-extern void vp8_compute_frame_size_bounds(VP8_COMP *cpi, int *frame_under_shoot_limit, int *frame_over_shoot_limit);
-
-#endif
diff --git a/vp8/encoder/rdopt.c b/vp8/encoder/rdopt.c
deleted file mode 100644
index a125cc4..0000000
--- a/vp8/encoder/rdopt.c
+++ /dev/null
@@ -1,2509 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdio.h>
-#include <math.h>
-#include <limits.h>
-#include <assert.h>
-#include "vp8/common/pragmas.h"
-
-#include "tokenize.h"
-#include "treewriter.h"
-#include "onyx_int.h"
-#include "modecosts.h"
-#include "encodeintra.h"
-#include "vp8/common/entropymode.h"
-#include "vp8/common/reconinter.h"
-#include "vp8/common/reconintra.h"
-#include "vp8/common/reconintra4x4.h"
-#include "vp8/common/findnearmv.h"
-#include "encodemb.h"
-#include "quantize.h"
-#include "vp8/common/idct.h"
-#include "vp8/common/g_common.h"
-#include "variance.h"
-#include "mcomp.h"
-
-#include "vpx_mem/vpx_mem.h"
-#include "dct.h"
-#include "vp8/common/systemdependent.h"
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IF_RTCD(x) (x)
-#else
-#define IF_RTCD(x) NULL
-#endif
-
-
-extern void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x);
-extern void vp8_update_zbin_extra(VP8_COMP *cpi, MACROBLOCK *x);
-
-
-#define RDCOST(RM,DM,R,D) ( ((128+(R)*(RM)) >> 8) + (DM)*(D) )
-
-#define MAXF(a,b) (((a) > (b)) ? (a) : (b))
-
-
-
-static const int auto_speed_thresh[17] =
-{
- 1000,
- 200,
- 150,
- 130,
- 150,
- 125,
- 120,
- 115,
- 115,
- 115,
- 115,
- 115,
- 115,
- 115,
- 115,
- 115,
- 105
-};
-
-const MB_PREDICTION_MODE vp8_mode_order[MAX_MODES] =
-{
- ZEROMV,
- DC_PRED,
-
- NEARESTMV,
- NEARMV,
-
- ZEROMV,
- NEARESTMV,
-
- ZEROMV,
- NEARESTMV,
-
- NEARMV,
- NEARMV,
-
- V_PRED,
- H_PRED,
- TM_PRED,
-
- NEWMV,
- NEWMV,
- NEWMV,
-
- SPLITMV,
- SPLITMV,
- SPLITMV,
-
- B_PRED,
-};
-
-const MV_REFERENCE_FRAME vp8_ref_frame_order[MAX_MODES] =
-{
- LAST_FRAME,
- INTRA_FRAME,
-
- LAST_FRAME,
- LAST_FRAME,
-
- GOLDEN_FRAME,
- GOLDEN_FRAME,
-
- ALTREF_FRAME,
- ALTREF_FRAME,
-
- GOLDEN_FRAME,
- ALTREF_FRAME,
-
- INTRA_FRAME,
- INTRA_FRAME,
- INTRA_FRAME,
-
- LAST_FRAME,
- GOLDEN_FRAME,
- ALTREF_FRAME,
-
- LAST_FRAME,
- GOLDEN_FRAME,
- ALTREF_FRAME,
-
- INTRA_FRAME,
-};
-
-static void fill_token_costs(
- unsigned int c [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens],
- const vp8_prob p [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1]
-)
-{
- int i, j, k;
-
-
- for (i = 0; i < BLOCK_TYPES; i++)
- for (j = 0; j < COEF_BANDS; j++)
- for (k = 0; k < PREV_COEF_CONTEXTS; k++)
-
- vp8_cost_tokens((int *)(c [i][j][k]), p [i][j][k], vp8_coef_tree);
-
-}
-
-static int rd_iifactor [ 32 ] = { 4, 4, 3, 2, 1, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- };
-
-
-/* values are now correlated to quantizer */
-static int sad_per_bit16lut[QINDEX_RANGE] =
-{
- 5, 5, 5, 5, 5, 5, 6, 6,
- 6, 6, 6, 6, 6, 7, 7, 7,
- 7, 7, 7, 7, 8, 8, 8, 8,
- 8, 8, 8, 8, 8, 8, 9, 9,
- 9, 9, 9, 9, 10, 10, 10, 10,
- 10, 10, 11, 11, 11, 11, 11, 11,
- 12, 12, 12, 12, 12, 12, 12, 13,
- 13, 13, 13, 13, 13, 14, 14, 14,
- 14, 14, 15, 15, 15, 15, 15, 15,
- 16, 16, 16, 16, 16, 16, 17, 17,
- 17, 17, 17, 17, 17, 18, 18, 18,
- 18, 18, 19, 19, 19, 19, 19, 19,
- 20, 20, 20, 21, 21, 21, 21, 22,
- 22, 22, 23, 23, 23, 24, 24, 24,
- 25, 25, 26, 26, 27, 27, 27, 28,
- 28, 28, 29, 29, 30, 30, 31, 31
-};
-static int sad_per_bit4lut[QINDEX_RANGE] =
-{
- 5, 5, 5, 5, 5, 5, 7, 7,
- 7, 7, 7, 7, 7, 8, 8, 8,
- 8, 8, 8, 8, 10, 10, 10, 10,
- 10, 10, 10, 10, 10, 10, 11, 11,
- 11, 11, 11, 11, 13, 13, 13, 13,
- 13, 13, 14, 14, 14, 14, 14, 14,
- 16, 16, 16, 16, 16, 16, 16, 17,
- 17, 17, 17, 17, 17, 19, 19, 19,
- 19, 19, 20, 20, 20, 20, 20, 20,
- 22, 22, 22, 22, 22, 22, 23, 23,
- 23, 23, 23, 23, 23, 25, 25, 25,
- 25, 25, 26, 26, 26, 26, 26, 26,
- 28, 28, 28, 29, 29, 29, 29, 31,
- 31, 31, 32, 32, 32, 34, 34, 34,
- 35, 35, 37, 37, 38, 38, 38, 40,
- 40, 40, 41, 41, 43, 43, 44, 44,
-};
-
-void vp8cx_initialize_me_consts(VP8_COMP *cpi, int QIndex)
-{
- cpi->mb.sadperbit16 = sad_per_bit16lut[QIndex];
- cpi->mb.sadperbit4 = sad_per_bit4lut[QIndex];
-}
-
-void vp8_initialize_rd_consts(VP8_COMP *cpi, int Qvalue)
-{
- int q;
- int i;
- double capped_q = (Qvalue < 160) ? (double)Qvalue : 160.0;
- double rdconst = 3.00;
-
- vp8_clear_system_state(); //__asm emms;
-
- // Further tests required to see if optimum is different
- // for key frames, golden frames and arf frames.
- // if (cpi->common.refresh_golden_frame ||
- // cpi->common.refresh_alt_ref_frame)
- cpi->RDMULT = (int)(rdconst * (capped_q * capped_q));
-
- // Extend rate multiplier along side quantizer zbin increases
- if (cpi->zbin_over_quant > 0)
- {
- double oq_factor;
- double modq;
-
- // Experimental code using the same basic equation as used for Q above
- // The units of cpi->zbin_over_quant are 1/128 of Q bin size
- oq_factor = 1.0 + ((double)0.0015625 * cpi->zbin_over_quant);
- modq = (int)((double)capped_q * oq_factor);
- cpi->RDMULT = (int)(rdconst * (modq * modq));
- }
-
- if (cpi->pass == 2 && (cpi->common.frame_type != KEY_FRAME))
- {
- if (cpi->next_iiratio > 31)
- cpi->RDMULT += (cpi->RDMULT * rd_iifactor[31]) >> 4;
- else
- cpi->RDMULT += (cpi->RDMULT * rd_iifactor[cpi->next_iiratio]) >> 4;
- }
-
- cpi->mb.errorperbit = (cpi->RDMULT / 100);
- cpi->mb.errorperbit += (cpi->mb.errorperbit==0);
-
- vp8_set_speed_features(cpi);
-
- if (cpi->common.simpler_lpf)
- cpi->common.filter_type = SIMPLE_LOOPFILTER;
-
- q = (int)pow(Qvalue, 1.25);
-
- if (q < 8)
- q = 8;
-
- if (cpi->RDMULT > 1000)
- {
- cpi->RDDIV = 1;
- cpi->RDMULT /= 100;
-
- for (i = 0; i < MAX_MODES; i++)
- {
- if (cpi->sf.thresh_mult[i] < INT_MAX)
- {
- cpi->rd_threshes[i] = cpi->sf.thresh_mult[i] * q / 100;
- }
- else
- {
- cpi->rd_threshes[i] = INT_MAX;
- }
-
- cpi->rd_baseline_thresh[i] = cpi->rd_threshes[i];
- }
- }
- else
- {
- cpi->RDDIV = 100;
-
- for (i = 0; i < MAX_MODES; i++)
- {
- if (cpi->sf.thresh_mult[i] < (INT_MAX / q))
- {
- cpi->rd_threshes[i] = cpi->sf.thresh_mult[i] * q;
- }
- else
- {
- cpi->rd_threshes[i] = INT_MAX;
- }
-
- cpi->rd_baseline_thresh[i] = cpi->rd_threshes[i];
- }
- }
-
- fill_token_costs(
- cpi->mb.token_costs,
- (const vp8_prob( *)[8][3][11]) cpi->common.fc.coef_probs
- );
-
- vp8_init_mode_costs(cpi);
-
-}
-
-void vp8_auto_select_speed(VP8_COMP *cpi)
-{
- int used = cpi->oxcf.cpu_used;
-
- int milliseconds_for_compress = (int)(1000000 / cpi->oxcf.frame_rate);
-
- milliseconds_for_compress = milliseconds_for_compress * (16 - cpi->oxcf.cpu_used) / 16;
-
-#if 0
-
- if (0)
- {
- FILE *f;
-
- f = fopen("speed.stt", "a");
- fprintf(f, " %8ld %10ld %10ld %10ld\n",
- cpi->common.current_video_frame, cpi->Speed, milliseconds_for_compress, cpi->avg_pick_mode_time);
- fclose(f);
- }
-
-#endif
-
- /*
- // this is done during parameter valid check
- if( used > 16)
- used = 16;
- if( used < -16)
- used = -16;
- */
-
- if (cpi->avg_pick_mode_time < milliseconds_for_compress && (cpi->avg_encode_time - cpi->avg_pick_mode_time) < milliseconds_for_compress)
- {
- if (cpi->avg_pick_mode_time == 0)
- {
- cpi->Speed = 4;
- }
- else
- {
- if (milliseconds_for_compress * 100 < cpi->avg_encode_time * 95)
- {
- cpi->Speed += 2;
- cpi->avg_pick_mode_time = 0;
- cpi->avg_encode_time = 0;
-
- if (cpi->Speed > 16)
- {
- cpi->Speed = 16;
- }
- }
-
- if (milliseconds_for_compress * 100 > cpi->avg_encode_time * auto_speed_thresh[cpi->Speed])
- {
- cpi->Speed -= 1;
- cpi->avg_pick_mode_time = 0;
- cpi->avg_encode_time = 0;
-
- // In real-time mode, cpi->speed is in [4, 16].
- if (cpi->Speed < 4) //if ( cpi->Speed < 0 )
- {
- cpi->Speed = 4; //cpi->Speed = 0;
- }
- }
- }
- }
- else
- {
- cpi->Speed += 4;
-
- if (cpi->Speed > 16)
- cpi->Speed = 16;
-
-
- cpi->avg_pick_mode_time = 0;
- cpi->avg_encode_time = 0;
- }
-}
-
-int vp8_block_error_c(short *coeff, short *dqcoeff)
-{
- int i;
- int error = 0;
-
- for (i = 0; i < 16; i++)
- {
- int this_diff = coeff[i] - dqcoeff[i];
- error += this_diff * this_diff;
- }
-
- return error;
-}
-
-int vp8_mbblock_error_c(MACROBLOCK *mb, int dc)
-{
- BLOCK *be;
- BLOCKD *bd;
- int i, j;
- int berror, error = 0;
-
- for (i = 0; i < 16; i++)
- {
- be = &mb->block[i];
- bd = &mb->e_mbd.block[i];
-
- berror = 0;
-
- for (j = dc; j < 16; j++)
- {
- int this_diff = be->coeff[j] - bd->dqcoeff[j];
- berror += this_diff * this_diff;
- }
-
- error += berror;
- }
-
- return error;
-}
-
-int vp8_mbuverror_c(MACROBLOCK *mb)
-{
-
- BLOCK *be;
- BLOCKD *bd;
-
-
- int i;
- int error = 0;
-
- for (i = 16; i < 24; i++)
- {
- be = &mb->block[i];
- bd = &mb->e_mbd.block[i];
-
- error += vp8_block_error_c(be->coeff, bd->dqcoeff);
- }
-
- return error;
-}
-
-int VP8_UVSSE(MACROBLOCK *x, const vp8_variance_rtcd_vtable_t *rtcd)
-{
- unsigned char *uptr, *vptr;
- unsigned char *upred_ptr = (*(x->block[16].base_src) + x->block[16].src);
- unsigned char *vpred_ptr = (*(x->block[20].base_src) + x->block[20].src);
- int uv_stride = x->block[16].src_stride;
-
- unsigned int sse1 = 0;
- unsigned int sse2 = 0;
- int mv_row;
- int mv_col;
- int offset;
- int pre_stride = x->e_mbd.block[16].pre_stride;
-
- vp8_build_uvmvs(&x->e_mbd, 0);
- mv_row = x->e_mbd.block[16].bmi.mv.as_mv.row;
- mv_col = x->e_mbd.block[16].bmi.mv.as_mv.col;
-
- offset = (mv_row >> 3) * pre_stride + (mv_col >> 3);
- uptr = x->e_mbd.pre.u_buffer + offset;
- vptr = x->e_mbd.pre.v_buffer + offset;
-
- if ((mv_row | mv_col) & 7)
- {
- VARIANCE_INVOKE(rtcd, subpixvar8x8)(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr, uv_stride, &sse2);
- VARIANCE_INVOKE(rtcd, subpixvar8x8)(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr, uv_stride, &sse1);
- sse2 += sse1;
- }
- else
- {
- VARIANCE_INVOKE(rtcd, subpixvar8x8)(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr, uv_stride, &sse2);
- VARIANCE_INVOKE(rtcd, subpixvar8x8)(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr, uv_stride, &sse1);
- sse2 += sse1;
- }
-
- return sse2;
-
-}
-
-#if !(CONFIG_REALTIME_ONLY)
-static int cost_coeffs(MACROBLOCK *mb, BLOCKD *b, int type, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l)
-{
- int c = !type; /* start at coef 0, unless Y with Y2 */
- int eob = b->eob;
- int pt ; /* surrounding block/prev coef predictor */
- int cost = 0;
- short *qcoeff_ptr = b->qcoeff;
-
- VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
-
-# define QC( I) ( qcoeff_ptr [vp8_default_zig_zag1d[I]] )
-
- for (; c < eob; c++)
- {
- int v = QC(c);
- int t = vp8_dct_value_tokens_ptr[v].Token;
- cost += mb->token_costs [type] [vp8_coef_bands[c]] [pt] [t];
- cost += vp8_dct_value_cost_ptr[v];
- pt = vp8_prev_token_class[t];
- }
-
-# undef QC
-
- if (c < 16)
- cost += mb->token_costs [type] [vp8_coef_bands[c]] [pt] [DCT_EOB_TOKEN];
-
- pt = (c != !type); // is eob first coefficient;
- *a = *l = pt;
-
- return cost;
-}
-
-static int vp8_rdcost_mby(MACROBLOCK *mb)
-{
- int cost = 0;
- int b;
- MACROBLOCKD *x = &mb->e_mbd;
- ENTROPY_CONTEXT_PLANES t_above, t_left;
- ENTROPY_CONTEXT *ta;
- ENTROPY_CONTEXT *tl;
-
- vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
-
- ta = (ENTROPY_CONTEXT *)&t_above;
- tl = (ENTROPY_CONTEXT *)&t_left;
-
- for (b = 0; b < 16; b++)
- cost += cost_coeffs(mb, x->block + b, PLANE_TYPE_Y_NO_DC,
- ta + vp8_block2above[b], tl + vp8_block2left[b]);
-
- cost += cost_coeffs(mb, x->block + 24, PLANE_TYPE_Y2,
- ta + vp8_block2above[24], tl + vp8_block2left[24]);
-
- return cost;
-}
-
-static void macro_block_yrd( MACROBLOCK *mb,
- int *Rate,
- int *Distortion,
- const vp8_encodemb_rtcd_vtable_t *rtcd)
-{
- int b;
- MACROBLOCKD *const x = &mb->e_mbd;
- BLOCK *const mb_y2 = mb->block + 24;
- BLOCKD *const x_y2 = x->block + 24;
- short *Y2DCPtr = mb_y2->src_diff;
- BLOCK *beptr;
- int d;
-
- ENCODEMB_INVOKE(rtcd, submby)( mb->src_diff, mb->src.y_buffer,
- mb->e_mbd.predictor, mb->src.y_stride );
-
- // Fdct and building the 2nd order block
- for (beptr = mb->block; beptr < mb->block + 16; beptr += 2)
- {
- mb->vp8_short_fdct8x4(beptr->src_diff, beptr->coeff, 32);
- *Y2DCPtr++ = beptr->coeff[0];
- *Y2DCPtr++ = beptr->coeff[16];
- }
-
- // 2nd order fdct
- mb->short_walsh4x4(mb_y2->src_diff, mb_y2->coeff, 8);
-
- // Quantization
- for (b = 0; b < 16; b++)
- {
- mb->quantize_b(&mb->block[b], &mb->e_mbd.block[b]);
- }
-
- // DC predication and Quantization of 2nd Order block
- mb->quantize_b(mb_y2, x_y2);
-
- // Distortion
- d = ENCODEMB_INVOKE(rtcd, mberr)(mb, 1) << 2;
- d += ENCODEMB_INVOKE(rtcd, berr)(mb_y2->coeff, x_y2->dqcoeff);
-
- *Distortion = (d >> 4);
-
- // rate
- *Rate = vp8_rdcost_mby(mb);
-}
-
-static void copy_predictor(unsigned char *dst, const unsigned char *predictor)
-{
- const unsigned int *p = (const unsigned int *)predictor;
- unsigned int *d = (unsigned int *)dst;
- d[0] = p[0];
- d[4] = p[4];
- d[8] = p[8];
- d[12] = p[12];
-}
-static int rd_pick_intra4x4block(
- VP8_COMP *cpi,
- MACROBLOCK *x,
- BLOCK *be,
- BLOCKD *b,
- B_PREDICTION_MODE *best_mode,
- unsigned int *bmode_costs,
- ENTROPY_CONTEXT *a,
- ENTROPY_CONTEXT *l,
-
- int *bestrate,
- int *bestratey,
- int *bestdistortion)
-{
- B_PREDICTION_MODE mode;
- int best_rd = INT_MAX;
- int rate = 0;
- int distortion;
-
- ENTROPY_CONTEXT ta = *a, tempa = *a;
- ENTROPY_CONTEXT tl = *l, templ = *l;
- /*
- * The predictor buffer is a 2d buffer with a stride of 16. Create
- * a temp buffer that meets the stride requirements, but we are only
- * interested in the left 4x4 block
- * */
- DECLARE_ALIGNED_ARRAY(16, unsigned char, best_predictor, 16*4);
- DECLARE_ALIGNED_ARRAY(16, short, best_dqcoeff, 16);
-
- for (mode = B_DC_PRED; mode <= B_HU_PRED; mode++)
- {
- int this_rd;
- int ratey;
-
- rate = bmode_costs[mode];
-
- vp8_predict_intra4x4(b, mode, b->predictor);
- ENCODEMB_INVOKE(IF_RTCD(&cpi->rtcd.encodemb), subb)(be, b, 16);
- x->vp8_short_fdct4x4(be->src_diff, be->coeff, 32);
- x->quantize_b(be, b);
-
- tempa = ta;
- templ = tl;
-
- ratey = cost_coeffs(x, b, PLANE_TYPE_Y_WITH_DC, &tempa, &templ);
- rate += ratey;
- distortion = ENCODEMB_INVOKE(IF_RTCD(&cpi->rtcd.encodemb), berr)(be->coeff, b->dqcoeff) >> 2;
-
- this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
-
- if (this_rd < best_rd)
- {
- *bestrate = rate;
- *bestratey = ratey;
- *bestdistortion = distortion;
- best_rd = this_rd;
- *best_mode = mode;
- *a = tempa;
- *l = templ;
- copy_predictor(best_predictor, b->predictor);
- vpx_memcpy(best_dqcoeff, b->dqcoeff, 32);
- }
- }
-
- b->bmi.mode = (B_PREDICTION_MODE)(*best_mode);
-
- IDCT_INVOKE(IF_RTCD(&cpi->rtcd.common->idct), idct16)(best_dqcoeff, b->diff, 32);
- RECON_INVOKE(IF_RTCD(&cpi->rtcd.common->recon), recon)(best_predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-
- return best_rd;
-}
-
-int vp8_rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *Rate,
- int *rate_y, int *Distortion, int best_rd)
-{
- MACROBLOCKD *const xd = &mb->e_mbd;
- int i;
- int cost = mb->mbmode_cost [xd->frame_type] [B_PRED];
- int distortion = 0;
- int tot_rate_y = 0;
- long long total_rd = 0;
- ENTROPY_CONTEXT_PLANES t_above, t_left;
- ENTROPY_CONTEXT *ta;
- ENTROPY_CONTEXT *tl;
- unsigned int *bmode_costs;
-
- vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
-
- ta = (ENTROPY_CONTEXT *)&t_above;
- tl = (ENTROPY_CONTEXT *)&t_left;
-
- vp8_intra_prediction_down_copy(xd);
-
- bmode_costs = mb->inter_bmode_costs;
-
- for (i = 0; i < 16; i++)
- {
- MODE_INFO *const mic = xd->mode_info_context;
- const int mis = xd->mode_info_stride;
- B_PREDICTION_MODE UNINITIALIZED_IS_SAFE(best_mode);
- int UNINITIALIZED_IS_SAFE(r), UNINITIALIZED_IS_SAFE(ry), UNINITIALIZED_IS_SAFE(d);
-
- if (mb->e_mbd.frame_type == KEY_FRAME)
- {
- const B_PREDICTION_MODE A = vp8_above_bmi(mic, i, mis)->mode;
- const B_PREDICTION_MODE L = vp8_left_bmi(mic, i)->mode;
-
- bmode_costs = mb->bmode_costs[A][L];
- }
-
- total_rd += rd_pick_intra4x4block(
- cpi, mb, mb->block + i, xd->block + i, &best_mode, bmode_costs,
- ta + vp8_block2above[i],
- tl + vp8_block2left[i], &r, &ry, &d);
-
- cost += r;
- distortion += d;
- tot_rate_y += ry;
- mic->bmi[i].mode = xd->block[i].bmi.mode = best_mode;
-
- if(total_rd >= (long long)best_rd)
- break;
- }
-
- if(total_rd >= (long long)best_rd)
- return INT_MAX;
-
- *Rate = cost;
- *rate_y += tot_rate_y;
- *Distortion = distortion;
-
- return RDCOST(mb->rdmult, mb->rddiv, cost, distortion);
-}
-int vp8_rd_pick_intra16x16mby_mode(VP8_COMP *cpi,
- MACROBLOCK *x,
- int *Rate,
- int *rate_y,
- int *Distortion)
-{
- MB_PREDICTION_MODE mode;
- MB_PREDICTION_MODE UNINITIALIZED_IS_SAFE(mode_selected);
- int rate, ratey;
- int distortion;
- int best_rd = INT_MAX;
- int this_rd;
-
- //Y Search for 16x16 intra prediction mode
- for (mode = DC_PRED; mode <= TM_PRED; mode++)
- {
- x->e_mbd.mode_info_context->mbmi.mode = mode;
-
- RECON_INVOKE(&cpi->common.rtcd.recon, build_intra_predictors_mby)
- (&x->e_mbd);
-
- macro_block_yrd(x, &ratey, &distortion, IF_RTCD(&cpi->rtcd.encodemb));
- rate = ratey + x->mbmode_cost[x->e_mbd.frame_type]
- [x->e_mbd.mode_info_context->mbmi.mode];
-
- this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
-
- if (this_rd < best_rd)
- {
- mode_selected = mode;
- best_rd = this_rd;
- *Rate = rate;
- *rate_y = ratey;
- *Distortion = distortion;
- }
- }
-
- x->e_mbd.mode_info_context->mbmi.mode = mode_selected;
- return best_rd;
-}
-
-static int rd_cost_mbuv(MACROBLOCK *mb)
-{
- int b;
- int cost = 0;
- MACROBLOCKD *x = &mb->e_mbd;
- ENTROPY_CONTEXT_PLANES t_above, t_left;
- ENTROPY_CONTEXT *ta;
- ENTROPY_CONTEXT *tl;
-
- vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
-
- ta = (ENTROPY_CONTEXT *)&t_above;
- tl = (ENTROPY_CONTEXT *)&t_left;
-
- for (b = 16; b < 24; b++)
- cost += cost_coeffs(mb, x->block + b, PLANE_TYPE_UV,
- ta + vp8_block2above[b], tl + vp8_block2left[b]);
-
- return cost;
-}
-
-
-static int vp8_rd_inter_uv(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *distortion, int fullpixel)
-{
- vp8_build_uvmvs(&x->e_mbd, fullpixel);
- vp8_encode_inter16x16uvrd(IF_RTCD(&cpi->rtcd), x);
-
-
- *rate = rd_cost_mbuv(x);
- *distortion = ENCODEMB_INVOKE(&cpi->rtcd.encodemb, mbuverr)(x) / 4;
-
- return RDCOST(x->rdmult, x->rddiv, *rate, *distortion);
-}
-
-int vp8_rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *rate_tokenonly, int *distortion)
-{
- MB_PREDICTION_MODE mode;
- MB_PREDICTION_MODE UNINITIALIZED_IS_SAFE(mode_selected);
- int best_rd = INT_MAX;
- int UNINITIALIZED_IS_SAFE(d), UNINITIALIZED_IS_SAFE(r);
- int rate_to;
-
- for (mode = DC_PRED; mode <= TM_PRED; mode++)
- {
- int rate;
- int distortion;
- int this_rd;
-
- x->e_mbd.mode_info_context->mbmi.uv_mode = mode;
- vp8_build_intra_predictors_mbuv(&x->e_mbd);
- ENCODEMB_INVOKE(IF_RTCD(&cpi->rtcd.encodemb), submbuv)(x->src_diff,
- x->src.u_buffer, x->src.v_buffer, x->e_mbd.predictor,
- x->src.uv_stride);
- vp8_transform_mbuv(x);
- vp8_quantize_mbuv(x);
-
- rate_to = rd_cost_mbuv(x);
- rate = rate_to + x->intra_uv_mode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.uv_mode];
-
- distortion = ENCODEMB_INVOKE(&cpi->rtcd.encodemb, mbuverr)(x) / 4;
-
- this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
-
- if (this_rd < best_rd)
- {
- best_rd = this_rd;
- d = distortion;
- r = rate;
- *rate_tokenonly = rate_to;
- mode_selected = mode;
- }
- }
-
- *rate = r;
- *distortion = d;
-
- x->e_mbd.mode_info_context->mbmi.uv_mode = mode_selected;
- return best_rd;
-}
-#endif
-
-int vp8_cost_mv_ref(MB_PREDICTION_MODE m, const int near_mv_ref_ct[4])
-{
- vp8_prob p [VP8_MVREFS-1];
- assert(NEARESTMV <= m && m <= SPLITMV);
- vp8_mv_ref_probs(p, near_mv_ref_ct);
- return vp8_cost_token(vp8_mv_ref_tree, p,
- vp8_mv_ref_encoding_array - NEARESTMV + m);
-}
-
-void vp8_set_mbmode_and_mvs(MACROBLOCK *x, MB_PREDICTION_MODE mb, MV *mv)
-{
- int i;
-
- x->e_mbd.mode_info_context->mbmi.mode = mb;
- x->e_mbd.mode_info_context->mbmi.mv.as_mv.row = mv->row;
- x->e_mbd.mode_info_context->mbmi.mv.as_mv.col = mv->col;
-
- for (i = 0; i < 16; i++)
- {
- B_MODE_INFO *bmi = &x->e_mbd.block[i].bmi;
- bmi->mode = (B_PREDICTION_MODE) mb;
- bmi->mv.as_mv.row = mv->row;
- bmi->mv.as_mv.col = mv->col;
- }
-}
-
-#if !(CONFIG_REALTIME_ONLY)
-static int labels2mode(
- MACROBLOCK *x,
- int const *labelings, int which_label,
- B_PREDICTION_MODE this_mode,
- MV *this_mv, MV *best_ref_mv,
- int *mvcost[2]
-)
-{
- MACROBLOCKD *const xd = & x->e_mbd;
- MODE_INFO *const mic = xd->mode_info_context;
- const int mis = xd->mode_info_stride;
-
- int cost = 0;
- int thismvcost = 0;
-
- /* We have to be careful retrieving previously-encoded motion vectors.
- Ones from this macroblock have to be pulled from the BLOCKD array
- as they have not yet made it to the bmi array in our MB_MODE_INFO. */
-
- int i = 0;
-
- do
- {
- BLOCKD *const d = xd->block + i;
- const int row = i >> 2, col = i & 3;
-
- B_PREDICTION_MODE m;
-
- if (labelings[i] != which_label)
- continue;
-
- if (col && labelings[i] == labelings[i-1])
- m = LEFT4X4;
- else if (row && labelings[i] == labelings[i-4])
- m = ABOVE4X4;
- else
- {
- // the only time we should do costing for new motion vector or mode
- // is when we are on a new label (jbb May 08, 2007)
- switch (m = this_mode)
- {
- case NEW4X4 :
- thismvcost = vp8_mv_bit_cost(this_mv, best_ref_mv, mvcost, 102);
- break;
- case LEFT4X4:
- *this_mv = col ? d[-1].bmi.mv.as_mv : vp8_left_bmi(mic, i)->mv.as_mv;
- break;
- case ABOVE4X4:
- *this_mv = row ? d[-4].bmi.mv.as_mv : vp8_above_bmi(mic, i, mis)->mv.as_mv;
- break;
- case ZERO4X4:
- this_mv->row = this_mv->col = 0;
- break;
- default:
- break;
- }
-
- if (m == ABOVE4X4) // replace above with left if same
- {
- const MV mv = col ? d[-1].bmi.mv.as_mv : vp8_left_bmi(mic, i)->mv.as_mv;
-
- if (mv.row == this_mv->row && mv.col == this_mv->col)
- m = LEFT4X4;
- }
-
- cost = x->inter_bmode_costs[ m];
- }
-
- d->bmi.mode = m;
- d->bmi.mv.as_mv = *this_mv;
-
- }
- while (++i < 16);
-
- cost += thismvcost ;
- return cost;
-}
-
-static int rdcost_mbsegment_y(MACROBLOCK *mb, const int *labels,
- int which_label, ENTROPY_CONTEXT *ta,
- ENTROPY_CONTEXT *tl)
-{
- int cost = 0;
- int b;
- MACROBLOCKD *x = &mb->e_mbd;
-
- for (b = 0; b < 16; b++)
- if (labels[ b] == which_label)
- cost += cost_coeffs(mb, x->block + b, PLANE_TYPE_Y_WITH_DC,
- ta + vp8_block2above[b],
- tl + vp8_block2left[b]);
-
- return cost;
-
-}
-static unsigned int vp8_encode_inter_mb_segment(MACROBLOCK *x, int const *labels, int which_label, const vp8_encodemb_rtcd_vtable_t *rtcd)
-{
- int i;
- unsigned int distortion = 0;
-
- for (i = 0; i < 16; i++)
- {
- if (labels[i] == which_label)
- {
- BLOCKD *bd = &x->e_mbd.block[i];
- BLOCK *be = &x->block[i];
-
-
- vp8_build_inter_predictors_b(bd, 16, x->e_mbd.subpixel_predict);
- ENCODEMB_INVOKE(rtcd, subb)(be, bd, 16);
- x->vp8_short_fdct4x4(be->src_diff, be->coeff, 32);
-
- // set to 0 no way to account for 2nd order DC so discount
- //be->coeff[0] = 0;
- x->quantize_b(be, bd);
-
- distortion += ENCODEMB_INVOKE(rtcd, berr)(be->coeff, bd->dqcoeff);
- }
- }
-
- return distortion;
-}
-
-
-static const unsigned int segmentation_to_sseshift[4] = {3, 3, 2, 0};
-
-
-typedef struct
-{
- MV *ref_mv;
- MV *mvp;
-
- int segment_rd;
- int segment_num;
- int r;
- int d;
- int segment_yrate;
- B_PREDICTION_MODE modes[16];
- int_mv mvs[16];
- unsigned char eobs[16];
-
- int mvthresh;
- int *mdcounts;
-
- MV sv_mvp[4]; // save 4 mvp from 8x8
- int sv_istep[2]; // save 2 initial step_param for 16x8/8x16
-
-} BEST_SEG_INFO;
-
-
-static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x,
- BEST_SEG_INFO *bsi, unsigned int segmentation)
-{
- int i;
- int const *labels;
- int br = 0;
- int bd = 0;
- B_PREDICTION_MODE this_mode;
-
-
- int label_count;
- int this_segment_rd = 0;
- int label_mv_thresh;
- int rate = 0;
- int sbr = 0;
- int sbd = 0;
- int segmentyrate = 0;
-
- vp8_variance_fn_ptr_t *v_fn_ptr;
-
- ENTROPY_CONTEXT_PLANES t_above, t_left;
- ENTROPY_CONTEXT *ta;
- ENTROPY_CONTEXT *tl;
- ENTROPY_CONTEXT_PLANES t_above_b, t_left_b;
- ENTROPY_CONTEXT *ta_b;
- ENTROPY_CONTEXT *tl_b;
-
- vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES));
-
- ta = (ENTROPY_CONTEXT *)&t_above;
- tl = (ENTROPY_CONTEXT *)&t_left;
- ta_b = (ENTROPY_CONTEXT *)&t_above_b;
- tl_b = (ENTROPY_CONTEXT *)&t_left_b;
-
- br = 0;
- bd = 0;
-
- v_fn_ptr = &cpi->fn_ptr[segmentation];
- labels = vp8_mbsplits[segmentation];
- label_count = vp8_mbsplit_count[segmentation];
-
- // 64 makes this threshold really big effectively
- // making it so that we very rarely check mvs on
- // segments. setting this to 1 would make mv thresh
- // roughly equal to what it is for macroblocks
- label_mv_thresh = 1 * bsi->mvthresh / label_count ;
-
- // Segmentation method overheads
- rate = vp8_cost_token(vp8_mbsplit_tree, vp8_mbsplit_probs, vp8_mbsplit_encodings + segmentation);
- rate += vp8_cost_mv_ref(SPLITMV, bsi->mdcounts);
- this_segment_rd += RDCOST(x->rdmult, x->rddiv, rate, 0);
- br += rate;
-
- for (i = 0; i < label_count; i++)
- {
- MV mode_mv[B_MODE_COUNT];
- int best_label_rd = INT_MAX;
- B_PREDICTION_MODE mode_selected = ZERO4X4;
- int bestlabelyrate = 0;
-
- // search for the best motion vector on this segment
- for (this_mode = LEFT4X4; this_mode <= NEW4X4 ; this_mode ++)
- {
- int this_rd;
- int distortion;
- int labelyrate;
- ENTROPY_CONTEXT_PLANES t_above_s, t_left_s;
- ENTROPY_CONTEXT *ta_s;
- ENTROPY_CONTEXT *tl_s;
-
- vpx_memcpy(&t_above_s, &t_above, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left_s, &t_left, sizeof(ENTROPY_CONTEXT_PLANES));
-
- ta_s = (ENTROPY_CONTEXT *)&t_above_s;
- tl_s = (ENTROPY_CONTEXT *)&t_left_s;
-
- if (this_mode == NEW4X4)
- {
- int sseshift;
- int num00;
- int step_param = 0;
- int further_steps;
- int n;
- int thissme;
- int bestsme = INT_MAX;
- MV temp_mv;
- BLOCK *c;
- BLOCKD *e;
-
- // Is the best so far sufficiently good that we cant justify doing and new motion search.
- if (best_label_rd < label_mv_thresh)
- break;
-
- if(cpi->compressor_speed)
- {
- if (segmentation == BLOCK_8X16 || segmentation == BLOCK_16X8)
- {
- bsi->mvp = &bsi->sv_mvp[i];
- if (i==1 && segmentation == BLOCK_16X8) bsi->mvp = &bsi->sv_mvp[2];
-
- step_param = bsi->sv_istep[i];
- }
-
- // use previous block's result as next block's MV predictor.
- if (segmentation == BLOCK_4X4 && i>0)
- {
- bsi->mvp = &(x->e_mbd.block[i-1].bmi.mv.as_mv);
- if (i==4 || i==8 || i==12) bsi->mvp = &(x->e_mbd.block[i-4].bmi.mv.as_mv);
- step_param = 2;
- }
- }
-
- further_steps = (MAX_MVSEARCH_STEPS - 1) - step_param;
-
- {
- int sadpb = x->sadperbit4;
-
- // find first label
- n = vp8_mbsplit_offset[segmentation][i];
-
- c = &x->block[n];
- e = &x->e_mbd.block[n];
-
- if (cpi->sf.search_method == HEX)
- bestsme = vp8_hex_search(x, c, e, bsi->ref_mv,
- &mode_mv[NEW4X4], step_param, sadpb, &num00, v_fn_ptr, x->mvsadcost, x->mvcost, bsi->ref_mv);
-
- else
- {
- bestsme = cpi->diamond_search_sad(x, c, e, bsi->mvp,
- &mode_mv[NEW4X4], step_param,
- sadpb / 2, &num00, v_fn_ptr, x->mvsadcost, x->mvcost, bsi->ref_mv);
-
- n = num00;
- num00 = 0;
-
- while (n < further_steps)
- {
- n++;
-
- if (num00)
- num00--;
- else
- {
- thissme = cpi->diamond_search_sad(x, c, e, bsi->mvp,
- &temp_mv, step_param + n,
- sadpb / 2, &num00, v_fn_ptr, x->mvsadcost, x->mvcost, bsi->ref_mv);
-
- if (thissme < bestsme)
- {
- bestsme = thissme;
- mode_mv[NEW4X4].row = temp_mv.row;
- mode_mv[NEW4X4].col = temp_mv.col;
- }
- }
- }
- }
-
- sseshift = segmentation_to_sseshift[segmentation];
-
- // Should we do a full search (best quality only)
- if ((cpi->compressor_speed == 0) && (bestsme >> sseshift) > 4000)
- {
- thissme = cpi->full_search_sad(x, c, e, bsi->mvp,
- sadpb / 4, 16, v_fn_ptr, x->mvcost, x->mvsadcost,bsi->ref_mv);
-
- if (thissme < bestsme)
- {
- bestsme = thissme;
- mode_mv[NEW4X4] = e->bmi.mv.as_mv;
- }
- else
- {
- // The full search result is actually worse so re-instate the previous best vector
- e->bmi.mv.as_mv = mode_mv[NEW4X4];
- }
- }
- }
-
- if (bestsme < INT_MAX)
- {
- if (!cpi->common.full_pixel)
- cpi->find_fractional_mv_step(x, c, e, &mode_mv[NEW4X4],
- bsi->ref_mv, x->errorperbit / 2, v_fn_ptr, x->mvcost);
- else
- vp8_skip_fractional_mv_step(x, c, e, &mode_mv[NEW4X4],
- bsi->ref_mv, x->errorperbit, v_fn_ptr, x->mvcost);
- }
- } /* NEW4X4 */
-
- rate = labels2mode(x, labels, i, this_mode, &mode_mv[this_mode],
- bsi->ref_mv, x->mvcost);
-
- // Trap vectors that reach beyond the UMV borders
- if (((mode_mv[this_mode].row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].row >> 3) > x->mv_row_max) ||
- ((mode_mv[this_mode].col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].col >> 3) > x->mv_col_max))
- {
- continue;
- }
-
- distortion = vp8_encode_inter_mb_segment(x, labels, i, IF_RTCD(&cpi->rtcd.encodemb)) / 4;
-
- labelyrate = rdcost_mbsegment_y(x, labels, i, ta_s, tl_s);
- rate += labelyrate;
-
- this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
-
- if (this_rd < best_label_rd)
- {
- sbr = rate;
- sbd = distortion;
- bestlabelyrate = labelyrate;
- mode_selected = this_mode;
- best_label_rd = this_rd;
-
- vpx_memcpy(ta_b, ta_s, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(tl_b, tl_s, sizeof(ENTROPY_CONTEXT_PLANES));
-
- }
- } /*for each 4x4 mode*/
-
- vpx_memcpy(ta, ta_b, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(tl, tl_b, sizeof(ENTROPY_CONTEXT_PLANES));
-
- labels2mode(x, labels, i, mode_selected, &mode_mv[mode_selected],
- bsi->ref_mv, x->mvcost);
-
- br += sbr;
- bd += sbd;
- segmentyrate += bestlabelyrate;
- this_segment_rd += best_label_rd;
-
- if (this_segment_rd >= bsi->segment_rd)
- break;
-
- } /* for each label */
-
- if (this_segment_rd < bsi->segment_rd)
- {
- bsi->r = br;
- bsi->d = bd;
- bsi->segment_yrate = segmentyrate;
- bsi->segment_rd = this_segment_rd;
- bsi->segment_num = segmentation;
-
- // store everything needed to come back to this!!
- for (i = 0; i < 16; i++)
- {
- BLOCKD *bd = &x->e_mbd.block[i];
-
- bsi->mvs[i].as_mv = bd->bmi.mv.as_mv;
- bsi->modes[i] = bd->bmi.mode;
- bsi->eobs[i] = bd->eob;
- }
- }
-}
-
-static __inline
-void vp8_cal_step_param(int sr, int *sp)
-{
- int step = 0;
-
- if (sr > MAX_FIRST_STEP) sr = MAX_FIRST_STEP;
- else if (sr < 1) sr = 1;
-
- while (sr>>=1)
- step++;
-
- *sp = MAX_MVSEARCH_STEPS - 1 - step;
-}
-
-static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x,
- MV *best_ref_mv, int best_rd,
- int *mdcounts, int *returntotrate,
- int *returnyrate, int *returndistortion,
- int mvthresh)
-{
- int i;
- BEST_SEG_INFO bsi;
-
- vpx_memset(&bsi, 0, sizeof(bsi));
-
- bsi.segment_rd = best_rd;
- bsi.ref_mv = best_ref_mv;
- bsi.mvp = best_ref_mv;
- bsi.mvthresh = mvthresh;
- bsi.mdcounts = mdcounts;
-
- for(i = 0; i < 16; i++)
- {
- bsi.modes[i] = ZERO4X4;
- }
-
- if(cpi->compressor_speed == 0)
- {
- /* for now, we will keep the original segmentation order
- when in best quality mode */
- rd_check_segment(cpi, x, &bsi, BLOCK_16X8);
- rd_check_segment(cpi, x, &bsi, BLOCK_8X16);
- rd_check_segment(cpi, x, &bsi, BLOCK_8X8);
- rd_check_segment(cpi, x, &bsi, BLOCK_4X4);
- }
- else
- {
- int sr;
-
- rd_check_segment(cpi, x, &bsi, BLOCK_8X8);
-
- if (bsi.segment_rd < best_rd)
- {
- int col_min = (best_ref_mv->col - MAX_FULL_PEL_VAL) >>3;
- int col_max = (best_ref_mv->col + MAX_FULL_PEL_VAL) >>3;
- int row_min = (best_ref_mv->row - MAX_FULL_PEL_VAL) >>3;
- int row_max = (best_ref_mv->row + MAX_FULL_PEL_VAL) >>3;
-
- int tmp_col_min = x->mv_col_min;
- int tmp_col_max = x->mv_col_max;
- int tmp_row_min = x->mv_row_min;
- int tmp_row_max = x->mv_row_max;
-
- /* Get intersection of UMV window and valid MV window to reduce # of checks in diamond search. */
- if (x->mv_col_min < col_min )
- x->mv_col_min = col_min;
- if (x->mv_col_max > col_max )
- x->mv_col_max = col_max;
- if (x->mv_row_min < row_min )
- x->mv_row_min = row_min;
- if (x->mv_row_max > row_max )
- x->mv_row_max = row_max;
-
- /* Get 8x8 result */
- bsi.sv_mvp[0] = bsi.mvs[0].as_mv;
- bsi.sv_mvp[1] = bsi.mvs[2].as_mv;
- bsi.sv_mvp[2] = bsi.mvs[8].as_mv;
- bsi.sv_mvp[3] = bsi.mvs[10].as_mv;
-
- /* Use 8x8 result as 16x8/8x16's predictor MV. Adjust search range according to the closeness of 2 MV. */
- /* block 8X16 */
- {
- sr = MAXF((abs(bsi.sv_mvp[0].row - bsi.sv_mvp[2].row))>>3, (abs(bsi.sv_mvp[0].col - bsi.sv_mvp[2].col))>>3);
- vp8_cal_step_param(sr, &bsi.sv_istep[0]);
-
- sr = MAXF((abs(bsi.sv_mvp[1].row - bsi.sv_mvp[3].row))>>3, (abs(bsi.sv_mvp[1].col - bsi.sv_mvp[3].col))>>3);
- vp8_cal_step_param(sr, &bsi.sv_istep[1]);
-
- rd_check_segment(cpi, x, &bsi, BLOCK_8X16);
- }
-
- /* block 16X8 */
- {
- sr = MAXF((abs(bsi.sv_mvp[0].row - bsi.sv_mvp[1].row))>>3, (abs(bsi.sv_mvp[0].col - bsi.sv_mvp[1].col))>>3);
- vp8_cal_step_param(sr, &bsi.sv_istep[0]);
-
- sr = MAXF((abs(bsi.sv_mvp[2].row - bsi.sv_mvp[3].row))>>3, (abs(bsi.sv_mvp[2].col - bsi.sv_mvp[3].col))>>3);
- vp8_cal_step_param(sr, &bsi.sv_istep[1]);
-
- rd_check_segment(cpi, x, &bsi, BLOCK_16X8);
- }
-
- /* If 8x8 is better than 16x8/8x16, then do 4x4 search */
- /* Not skip 4x4 if speed=0 (good quality) */
- if (cpi->sf.no_skip_block4x4_search || bsi.segment_num == BLOCK_8X8) /* || (sv_segment_rd8x8-bsi.segment_rd) < sv_segment_rd8x8>>5) */
- {
- bsi.mvp = &bsi.sv_mvp[0];
- rd_check_segment(cpi, x, &bsi, BLOCK_4X4);
- }
-
- /* restore UMV window */
- x->mv_col_min = tmp_col_min;
- x->mv_col_max = tmp_col_max;
- x->mv_row_min = tmp_row_min;
- x->mv_row_max = tmp_row_max;
- }
- }
-
- /* set it to the best */
- for (i = 0; i < 16; i++)
- {
- BLOCKD *bd = &x->e_mbd.block[i];
-
- bd->bmi.mv.as_mv = bsi.mvs[i].as_mv;
- bd->bmi.mode = bsi.modes[i];
- bd->eob = bsi.eobs[i];
- }
-
- *returntotrate = bsi.r;
- *returndistortion = bsi.d;
- *returnyrate = bsi.segment_yrate;
-
- /* save partitions */
- x->e_mbd.mode_info_context->mbmi.partitioning = bsi.segment_num;
- x->partition_info->count = vp8_mbsplit_count[bsi.segment_num];
-
- for (i = 0; i < x->partition_info->count; i++)
- {
- int j;
-
- j = vp8_mbsplit_offset[bsi.segment_num][i];
-
- x->partition_info->bmi[i].mode = x->e_mbd.block[j].bmi.mode;
- x->partition_info->bmi[i].mv.as_mv = x->e_mbd.block[j].bmi.mv.as_mv;
- }
-
- return bsi.segment_rd;
-}
-#endif
-
-static void swap(int *x,int *y)
-{
- int tmp;
-
- tmp = *x;
- *x = *y;
- *y = tmp;
-}
-
-static void quicksortmv(int arr[],int left, int right)
-{
- int lidx,ridx,pivot;
-
- lidx = left;
- ridx = right;
-
- if( left < right)
- {
- pivot = (left + right)/2;
-
- while(lidx <=pivot && ridx >=pivot)
- {
- while(arr[lidx] < arr[pivot] && lidx <= pivot)
- lidx++;
- while(arr[ridx] > arr[pivot] && ridx >= pivot)
- ridx--;
- swap(&arr[lidx], &arr[ridx]);
- lidx++;
- ridx--;
- if(lidx-1 == pivot)
- {
- ridx++;
- pivot = ridx;
- }
- else if(ridx+1 == pivot)
- {
- lidx--;
- pivot = lidx;
- }
- }
- quicksortmv(arr, left, pivot - 1);
- quicksortmv(arr, pivot + 1, right);
- }
-}
-
-static void quicksortsad(int arr[],int idx[], int left, int right)
-{
- int lidx,ridx,pivot;
-
- lidx = left;
- ridx = right;
-
- if( left < right)
- {
- pivot = (left + right)/2;
-
- while(lidx <=pivot && ridx >=pivot)
- {
- while(arr[lidx] < arr[pivot] && lidx <= pivot)
- lidx++;
- while(arr[ridx] > arr[pivot] && ridx >= pivot)
- ridx--;
- swap(&arr[lidx], &arr[ridx]);
- swap(&idx[lidx], &idx[ridx]);
- lidx++;
- ridx--;
- if(lidx-1 == pivot)
- {
- ridx++;
- pivot = ridx;
- }
- else if(ridx+1 == pivot)
- {
- lidx--;
- pivot = lidx;
- }
- }
- quicksortsad(arr, idx, left, pivot - 1);
- quicksortsad(arr, idx, pivot + 1, right);
- }
-}
-
-//The improved MV prediction
-void vp8_mv_pred
-(
- VP8_COMP *cpi,
- MACROBLOCKD *xd,
- const MODE_INFO *here,
- MV *mvp,
- int refframe,
- int *ref_frame_sign_bias,
- int *sr,
- int near_sadidx[]
-)
-{
- const MODE_INFO *above = here - xd->mode_info_stride;
- const MODE_INFO *left = here - 1;
- const MODE_INFO *aboveleft = above - 1;
- int_mv near_mvs[8];
- int near_ref[8];
- int_mv mv;
- int vcnt=0;
- int find=0;
- int mb_offset;
-
- int mvx[8];
- int mvy[8];
- int i;
-
- mv.as_int = 0;
-
- if(here->mbmi.ref_frame != INTRA_FRAME)
- {
- near_mvs[0].as_int = near_mvs[1].as_int = near_mvs[2].as_int = near_mvs[3].as_int = near_mvs[4].as_int = near_mvs[5].as_int = near_mvs[6].as_int = near_mvs[7].as_int = 0;
- near_ref[0] = near_ref[1] = near_ref[2] = near_ref[3] = near_ref[4] = near_ref[5] = near_ref[6] = near_ref[7] = 0;
-
- // read in 3 nearby block's MVs from current frame as prediction candidates.
- if (above->mbmi.ref_frame != INTRA_FRAME)
- {
- near_mvs[vcnt].as_int = above->mbmi.mv.as_int;
- mv_bias(ref_frame_sign_bias[above->mbmi.ref_frame], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
- near_ref[vcnt] = above->mbmi.ref_frame;
- }
- vcnt++;
- if (left->mbmi.ref_frame != INTRA_FRAME)
- {
- near_mvs[vcnt].as_int = left->mbmi.mv.as_int;
- mv_bias(ref_frame_sign_bias[left->mbmi.ref_frame], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
- near_ref[vcnt] = left->mbmi.ref_frame;
- }
- vcnt++;
- if (aboveleft->mbmi.ref_frame != INTRA_FRAME)
- {
- near_mvs[vcnt].as_int = aboveleft->mbmi.mv.as_int;
- mv_bias(ref_frame_sign_bias[aboveleft->mbmi.ref_frame], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
- near_ref[vcnt] = aboveleft->mbmi.ref_frame;
- }
- vcnt++;
-
- // read in 5 nearby block's MVs from last frame.
- if(cpi->common.last_frame_type != KEY_FRAME)
- {
- mb_offset = (-xd->mb_to_top_edge/128 + 1) * (xd->mode_info_stride +1) + (-xd->mb_to_left_edge/128 +1) ;
-
- // current in last frame
- if (cpi->lf_ref_frame[mb_offset] != INTRA_FRAME)
- {
- near_mvs[vcnt].as_int = cpi->lfmv[mb_offset].as_int;
- mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
- near_ref[vcnt] = cpi->lf_ref_frame[mb_offset];
- }
- vcnt++;
-
- // above in last frame
- if (cpi->lf_ref_frame[mb_offset - xd->mode_info_stride-1] != INTRA_FRAME)
- {
- near_mvs[vcnt].as_int = cpi->lfmv[mb_offset - xd->mode_info_stride-1].as_int;
- mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset - xd->mode_info_stride-1], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
- near_ref[vcnt] = cpi->lf_ref_frame[mb_offset - xd->mode_info_stride-1];
- }
- vcnt++;
-
- // left in last frame
- if (cpi->lf_ref_frame[mb_offset-1] != INTRA_FRAME)
- {
- near_mvs[vcnt].as_int = cpi->lfmv[mb_offset -1].as_int;
- mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset -1], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
- near_ref[vcnt] = cpi->lf_ref_frame[mb_offset - 1];
- }
- vcnt++;
-
- // right in last frame
- if (cpi->lf_ref_frame[mb_offset +1] != INTRA_FRAME)
- {
- near_mvs[vcnt].as_int = cpi->lfmv[mb_offset +1].as_int;
- mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset +1], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
- near_ref[vcnt] = cpi->lf_ref_frame[mb_offset +1];
- }
- vcnt++;
-
- // below in last frame
- if (cpi->lf_ref_frame[mb_offset + xd->mode_info_stride +1] != INTRA_FRAME)
- {
- near_mvs[vcnt].as_int = cpi->lfmv[mb_offset + xd->mode_info_stride +1].as_int;
- mv_bias(cpi->lf_ref_frame_sign_bias[mb_offset + xd->mode_info_stride +1], refframe, &near_mvs[vcnt], ref_frame_sign_bias);
- near_ref[vcnt] = cpi->lf_ref_frame[mb_offset + xd->mode_info_stride +1];
- }
- vcnt++;
- }
-
- for(i=0; i< vcnt; i++)
- {
- if(near_ref[near_sadidx[i]] != INTRA_FRAME)
- {
- if(here->mbmi.ref_frame == near_ref[near_sadidx[i]])
- {
- mv.as_int = near_mvs[near_sadidx[i]].as_int;
- find = 1;
- if (i < 3)
- *sr = 3;
- else
- *sr = 2;
- break;
- }
- }
- }
-
- if(!find)
- {
- for(i=0; i<vcnt; i++)
- {
- mvx[i] = near_mvs[i].as_mv.row;
- mvy[i] = near_mvs[i].as_mv.col;
- }
-
- quicksortmv (mvx, 0, vcnt-1);
- quicksortmv (mvy, 0, vcnt-1);
- mv.as_mv.row = mvx[vcnt/2];
- mv.as_mv.col = mvy[vcnt/2];
-
- find = 1;
- //sr is set to 0 to allow calling function to decide the search range.
- *sr = 0;
- }
- }
-
- /* Set up return values */
- *mvp = mv.as_mv;
- vp8_clamp_mv(mvp, xd);
-}
-
-void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffset, int near_sadidx[])
-{
-
- int near_sad[8] = {0}; // 0-cf above, 1-cf left, 2-cf aboveleft, 3-lf current, 4-lf above, 5-lf left, 6-lf right, 7-lf below
-
- //calculate sad for current frame 3 nearby MBs.
- if( xd->mb_to_top_edge==0 && xd->mb_to_left_edge ==0)
- {
- near_sad[0] = near_sad[1] = near_sad[2] = INT_MAX;
- }else if(xd->mb_to_top_edge==0)
- { //only has left MB for sad calculation.
- near_sad[0] = near_sad[2] = INT_MAX;
- near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
- }else if(xd->mb_to_left_edge ==0)
- { //only has left MB for sad calculation.
- near_sad[1] = near_sad[2] = INT_MAX;
- near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
- }else
- {
- near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
- near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
- near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, 0x7fffffff);
- }
-
- if(cpi->common.last_frame_type != KEY_FRAME)
- {
- //calculate sad for last frame 5 nearby MBs.
- unsigned char *pre_y_buffer = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_buffer + recon_yoffset;
- int pre_y_stride = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_stride;
-
- if(xd->mb_to_top_edge==0) near_sad[4] = INT_MAX;
- if(xd->mb_to_left_edge ==0) near_sad[5] = INT_MAX;
- if(xd->mb_to_right_edge ==0) near_sad[6] = INT_MAX;
- if(xd->mb_to_bottom_edge==0) near_sad[7] = INT_MAX;
-
- if(near_sad[4] != INT_MAX)
- near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, 0x7fffffff);
- if(near_sad[5] != INT_MAX)
- near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - 16, pre_y_stride, 0x7fffffff);
- near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer, pre_y_stride, 0x7fffffff);
- if(near_sad[6] != INT_MAX)
- near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + 16, pre_y_stride, 0x7fffffff);
- if(near_sad[7] != INT_MAX)
- near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, 0x7fffffff);
- }
-
- if(cpi->common.last_frame_type != KEY_FRAME)
- {
- quicksortsad(near_sad, near_sadidx, 0, 7);
- }else
- {
- quicksortsad(near_sad, near_sadidx, 0, 2);
- }
-}
-
-#if !(CONFIG_REALTIME_ONLY)
-int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra)
-{
- BLOCK *b = &x->block[0];
- BLOCKD *d = &x->e_mbd.block[0];
- MACROBLOCKD *xd = &x->e_mbd;
- B_MODE_INFO best_bmodes[16];
- MB_MODE_INFO best_mbmode;
- PARTITION_INFO best_partition;
- MV best_ref_mv;
- MV mode_mv[MB_MODE_COUNT];
- MB_PREDICTION_MODE this_mode;
- int num00;
- int best_mode_index = 0;
-
- int i;
- int mode_index;
- int mdcounts[4];
- int rate;
- int distortion;
- int best_rd = INT_MAX; // 1 << 30;
- int ref_frame_cost[MAX_REF_FRAMES];
- int rate2, distortion2;
- int uv_intra_rate, uv_intra_distortion, uv_intra_rate_tokenonly;
- int rate_y, UNINITIALIZED_IS_SAFE(rate_uv);
- int distortion_uv;
- int best_yrd = INT_MAX;
-
- //int all_rds[MAX_MODES]; // Experimental debug code.
- //int all_rates[MAX_MODES];
- //int all_dist[MAX_MODES];
- //int intermodecost[MAX_MODES];
-
- MB_PREDICTION_MODE uv_intra_mode;
-
- int force_no_skip = 0;
-
- MV mvp;
- int near_sadidx[8] = {0, 1, 2, 3, 4, 5, 6, 7};
- int saddone=0;
- int sr=0; //search range got from mv_pred(). It uses step_param levels. (0-7)
-
- MV frame_nearest_mv[4];
- MV frame_near_mv[4];
- MV frame_best_ref_mv[4];
- int frame_mdcounts[4][4];
- int frame_lf_or_gf[4];
- unsigned char *y_buffer[4];
- unsigned char *u_buffer[4];
- unsigned char *v_buffer[4];
-
- vpx_memset(&best_mbmode, 0, sizeof(best_mbmode));
-
- if (cpi->ref_frame_flags & VP8_LAST_FLAG)
- {
- YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx];
-
- vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &frame_nearest_mv[LAST_FRAME], &frame_near_mv[LAST_FRAME],
- &frame_best_ref_mv[LAST_FRAME], frame_mdcounts[LAST_FRAME], LAST_FRAME, cpi->common.ref_frame_sign_bias);
-
- y_buffer[LAST_FRAME] = lst_yv12->y_buffer + recon_yoffset;
- u_buffer[LAST_FRAME] = lst_yv12->u_buffer + recon_uvoffset;
- v_buffer[LAST_FRAME] = lst_yv12->v_buffer + recon_uvoffset;
-
- frame_lf_or_gf[LAST_FRAME] = 0;
- }
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- YV12_BUFFER_CONFIG *gld_yv12 = &cpi->common.yv12_fb[cpi->common.gld_fb_idx];
-
- vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &frame_nearest_mv[GOLDEN_FRAME], &frame_near_mv[GOLDEN_FRAME],
- &frame_best_ref_mv[GOLDEN_FRAME], frame_mdcounts[GOLDEN_FRAME], GOLDEN_FRAME, cpi->common.ref_frame_sign_bias);
-
- y_buffer[GOLDEN_FRAME] = gld_yv12->y_buffer + recon_yoffset;
- u_buffer[GOLDEN_FRAME] = gld_yv12->u_buffer + recon_uvoffset;
- v_buffer[GOLDEN_FRAME] = gld_yv12->v_buffer + recon_uvoffset;
-
- frame_lf_or_gf[GOLDEN_FRAME] = 1;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- YV12_BUFFER_CONFIG *alt_yv12 = &cpi->common.yv12_fb[cpi->common.alt_fb_idx];
-
- vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &frame_nearest_mv[ALTREF_FRAME], &frame_near_mv[ALTREF_FRAME],
- &frame_best_ref_mv[ALTREF_FRAME], frame_mdcounts[ALTREF_FRAME], ALTREF_FRAME, cpi->common.ref_frame_sign_bias);
-
- y_buffer[ALTREF_FRAME] = alt_yv12->y_buffer + recon_yoffset;
- u_buffer[ALTREF_FRAME] = alt_yv12->u_buffer + recon_uvoffset;
- v_buffer[ALTREF_FRAME] = alt_yv12->v_buffer + recon_uvoffset;
-
- frame_lf_or_gf[ALTREF_FRAME] = 1;
- }
-
- *returnintra = INT_MAX;
- cpi->mbs_tested_so_far++; // Count of the number of MBs tested so far this frame
-
- x->skip = 0;
-
- ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(cpi->prob_intra_coded);
-
- // Special case treatment when GF and ARF are not sensible options for reference
- if (cpi->ref_frame_flags == VP8_LAST_FLAG)
- {
- ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(255);
- ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(255)
- + vp8_cost_zero(128);
- ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(255)
- + vp8_cost_one(128);
- }
- else
- {
- ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(cpi->prob_last_coded);
- ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_zero(cpi->prob_gf_coded);
- ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_one(cpi->prob_gf_coded);
- }
-
- vpx_memset(mode_mv, 0, sizeof(mode_mv));
-
- x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
- vp8_rd_pick_intra_mbuv_mode(cpi, x, &uv_intra_rate, &uv_intra_rate_tokenonly, &uv_intra_distortion);
- uv_intra_mode = x->e_mbd.mode_info_context->mbmi.uv_mode;
-
- for (mode_index = 0; mode_index < MAX_MODES; mode_index++)
- {
- int this_rd = INT_MAX;
- int lf_or_gf = 0; // Lat Frame (01) or gf/arf (1)
- int disable_skip = 0;
- int other_cost = 0;
-
- force_no_skip = 0;
-
- // Experimental debug code.
- // Record of rd values recorded for this MB. -1 indicates not measured
- //all_rds[mode_index] = -1;
- //all_rates[mode_index] = -1;
- //all_dist[mode_index] = -1;
- //intermodecost[mode_index] = -1;
-
- // Test best rd so far against threshold for trying this mode.
- if (best_rd <= cpi->rd_threshes[mode_index])
- continue;
-
- // These variables hold are rolling total cost and distortion for this mode
- rate2 = 0;
- distortion2 = 0;
-
- this_mode = vp8_mode_order[mode_index];
-
- x->e_mbd.mode_info_context->mbmi.mode = this_mode;
- x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
- x->e_mbd.mode_info_context->mbmi.ref_frame = vp8_ref_frame_order[mode_index];
-
- // Only consider ZEROMV/ALTREF_FRAME for alt ref frame,
- // unless ARNR filtering is enabled in which case we want
- // an unfiltered alternative
- if (cpi->is_src_frame_alt_ref && (cpi->oxcf.arnr_max_frames == 0))
- {
- if (this_mode != ZEROMV || x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME)
- continue;
- }
-
- /* everything but intra */
- if (x->e_mbd.mode_info_context->mbmi.ref_frame)
- {
- x->e_mbd.pre.y_buffer = y_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- x->e_mbd.pre.u_buffer = u_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- x->e_mbd.pre.v_buffer = v_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- mode_mv[NEARESTMV] = frame_nearest_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
- mode_mv[NEARMV] = frame_near_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
- best_ref_mv = frame_best_ref_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
- vpx_memcpy(mdcounts, frame_mdcounts[x->e_mbd.mode_info_context->mbmi.ref_frame], sizeof(mdcounts));
- lf_or_gf = frame_lf_or_gf[x->e_mbd.mode_info_context->mbmi.ref_frame];
- }
-
- if(x->e_mbd.mode_info_context->mbmi.mode == NEWMV)
- {
- if(!saddone)
- {
- vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] );
- saddone = 1;
- }
-
- vp8_mv_pred(cpi, &x->e_mbd, x->e_mbd.mode_info_context, &mvp,
- x->e_mbd.mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias, &sr, &near_sadidx[0]);
-
- /* adjust mvp to make sure it is within MV range */
- if(mvp.row > best_ref_mv.row + MAX_FULL_PEL_VAL)
- mvp.row = best_ref_mv.row + MAX_FULL_PEL_VAL;
- else if(mvp.row < best_ref_mv.row - MAX_FULL_PEL_VAL)
- mvp.row = best_ref_mv.row - MAX_FULL_PEL_VAL;
- if(mvp.col > best_ref_mv.col + MAX_FULL_PEL_VAL)
- mvp.col = best_ref_mv.col + MAX_FULL_PEL_VAL;
- else if(mvp.col < best_ref_mv.col - MAX_FULL_PEL_VAL)
- mvp.col = best_ref_mv.col - MAX_FULL_PEL_VAL;
- }
-
- // Check to see if the testing frequency for this mode is at its max
- // If so then prevent it from being tested and increase the threshold for its testing
- if (cpi->mode_test_hit_counts[mode_index] && (cpi->mode_check_freq[mode_index] > 1))
- {
- if (cpi->mbs_tested_so_far <= cpi->mode_check_freq[mode_index] * cpi->mode_test_hit_counts[mode_index])
- {
- // Increase the threshold for coding this mode to make it less likely to be chosen
- cpi->rd_thresh_mult[mode_index] += 4;
-
- if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
- cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
-
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
-
- continue;
- }
- }
-
- // We have now reached the point where we are going to test the current mode so increment the counter for the number of times it has been tested
- cpi->mode_test_hit_counts[mode_index] ++;
-
- // Experimental code. Special case for gf and arf zeromv modes. Increase zbin size to supress noise
- if (cpi->zbin_mode_boost_enabled)
- {
- if ( vp8_ref_frame_order[mode_index] == INTRA_FRAME )
- cpi->zbin_mode_boost = 0;
- else
- {
- if (vp8_mode_order[mode_index] == ZEROMV)
- {
- if (vp8_ref_frame_order[mode_index] != LAST_FRAME)
- cpi->zbin_mode_boost = GF_ZEROMV_ZBIN_BOOST;
- else
- cpi->zbin_mode_boost = LF_ZEROMV_ZBIN_BOOST;
- }
- else if (vp8_mode_order[mode_index] == SPLITMV)
- cpi->zbin_mode_boost = 0;
- else
- cpi->zbin_mode_boost = MV_ZBIN_BOOST;
- }
-
- vp8_update_zbin_extra(cpi, x);
- }
-
- switch (this_mode)
- {
- case B_PRED:
- {
- int tmp_rd;
-
- // Note the rate value returned here includes the cost of coding the BPRED mode : x->mbmode_cost[x->e_mbd.frame_type][BPRED];
- tmp_rd = vp8_rd_pick_intra4x4mby_modes(cpi, x, &rate, &rate_y, &distortion, best_yrd);
- rate2 += rate;
- distortion2 += distortion;
-
- if(tmp_rd < best_yrd)
- {
- rate2 += uv_intra_rate;
- rate_uv = uv_intra_rate_tokenonly;
- distortion2 += uv_intra_distortion;
- distortion_uv = uv_intra_distortion;
- }
- else
- {
- this_rd = INT_MAX;
- disable_skip = 1;
- }
- }
- break;
-
- case SPLITMV:
- {
- int tmp_rd;
- int this_rd_thresh;
-
- this_rd_thresh = (x->e_mbd.mode_info_context->mbmi.ref_frame == LAST_FRAME) ? cpi->rd_threshes[THR_NEWMV] : cpi->rd_threshes[THR_NEWA];
- this_rd_thresh = (x->e_mbd.mode_info_context->mbmi.ref_frame == GOLDEN_FRAME) ? cpi->rd_threshes[THR_NEWG]: this_rd_thresh;
-
- tmp_rd = vp8_rd_pick_best_mbsegmentation(cpi, x, &best_ref_mv,
- best_yrd, mdcounts,
- &rate, &rate_y, &distortion, this_rd_thresh) ;
-
- rate2 += rate;
- distortion2 += distortion;
-
- // If even the 'Y' rd value of split is higher than best so far then dont bother looking at UV
- if (tmp_rd < best_yrd)
- {
- // Now work out UV cost and add it in
- vp8_rd_inter_uv(cpi, x, &rate_uv, &distortion_uv, cpi->common.full_pixel);
- rate2 += rate_uv;
- distortion2 += distortion_uv;
- }
- else
- {
- this_rd = INT_MAX;
- disable_skip = 1;
- }
- }
- break;
- case DC_PRED:
- case V_PRED:
- case H_PRED:
- case TM_PRED:
- x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
- RECON_INVOKE(&cpi->common.rtcd.recon, build_intra_predictors_mby)
- (&x->e_mbd);
- macro_block_yrd(x, &rate_y, &distortion, IF_RTCD(&cpi->rtcd.encodemb)) ;
- rate2 += rate_y;
- distortion2 += distortion;
- rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode];
- rate2 += uv_intra_rate;
- rate_uv = uv_intra_rate_tokenonly;
- distortion2 += uv_intra_distortion;
- distortion_uv = uv_intra_distortion;
- break;
-
- case NEWMV:
-
- // Decrement full search counter
- if (cpi->check_freq[lf_or_gf] > 0)
- cpi->check_freq[lf_or_gf] --;
-
- {
- int thissme;
- int bestsme = INT_MAX;
- int step_param = cpi->sf.first_step;
- int search_range;
- int further_steps;
- int n;
-
- int col_min = (best_ref_mv.col - MAX_FULL_PEL_VAL) >>3;
- int col_max = (best_ref_mv.col + MAX_FULL_PEL_VAL) >>3;
- int row_min = (best_ref_mv.row - MAX_FULL_PEL_VAL) >>3;
- int row_max = (best_ref_mv.row + MAX_FULL_PEL_VAL) >>3;
-
- int tmp_col_min = x->mv_col_min;
- int tmp_col_max = x->mv_col_max;
- int tmp_row_min = x->mv_row_min;
- int tmp_row_max = x->mv_row_max;
-
- // Get intersection of UMV window and valid MV window to reduce # of checks in diamond search.
- if (x->mv_col_min < col_min )
- x->mv_col_min = col_min;
- if (x->mv_col_max > col_max )
- x->mv_col_max = col_max;
- if (x->mv_row_min < row_min )
- x->mv_row_min = row_min;
- if (x->mv_row_max > row_max )
- x->mv_row_max = row_max;
-
- //adjust search range according to sr from mv prediction
- if(sr > step_param)
- step_param = sr;
-
- // Work out how long a search we should do
- search_range = MAXF(abs(best_ref_mv.col), abs(best_ref_mv.row)) >> 3;
-
- if (search_range >= x->vector_range)
- x->vector_range = search_range;
- else if (x->vector_range > cpi->sf.min_fs_radius)
- x->vector_range--;
-
- // Initial step/diamond search
- {
- int sadpb = x->sadperbit16;
-
- if (cpi->sf.search_method == HEX)
- {
- bestsme = vp8_hex_search(x, b, d, &best_ref_mv, &d->bmi.mv.as_mv, step_param, sadpb/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv);
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
- }
- else
- {
- bestsme = cpi->diamond_search_sad(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param, sadpb / 2/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); //sadpb < 9
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
-
- // Further step/diamond searches as necessary
- n = 0;
- further_steps = (cpi->sf.max_step_search_steps - 1) - step_param;
-
- n = num00;
- num00 = 0;
-
- while (n < further_steps)
- {
- n++;
-
- if (num00)
- num00--;
- else
- {
- thissme = cpi->diamond_search_sad(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param + n, sadpb / 4/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); //sadpb = 9
-
- if (thissme < bestsme)
- {
- bestsme = thissme;
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
- }
- else
- {
- d->bmi.mv.as_mv.row = mode_mv[NEWMV].row;
- d->bmi.mv.as_mv.col = mode_mv[NEWMV].col;
- }
- }
- }
- }
-
- }
-
- // Should we do a full search
- if (!cpi->check_freq[lf_or_gf] || cpi->do_full[lf_or_gf])
- {
- int thissme;
- int full_flag_thresh = 0;
- MV full_mvp;
-
- full_mvp.row = d->bmi.mv.as_mv.row <<3; // use diamond search result as full search staring point
- full_mvp.col = d->bmi.mv.as_mv.col <<3;
-
- // Update x->vector_range based on best vector found in step search
- search_range = MAXF(abs((mvp.row>>3) - d->bmi.mv.as_mv.row), abs((mvp.col>>3) - d->bmi.mv.as_mv.col));
- //search_range *= 1.4; //didn't improve PSNR
-
- if (search_range > x->vector_range)
- x->vector_range = search_range;
- else
- search_range = x->vector_range;
-
- // Apply limits
- search_range = (search_range > cpi->sf.max_fs_radius) ? cpi->sf.max_fs_radius : search_range;
-
- //add this to reduce full search range.
- if(sr<=3 && search_range > 8) search_range = 8;
-
- {
- int sadpb = x->sadperbit16 >> 2;
- thissme = cpi->full_search_sad(x, b, d, &full_mvp, sadpb, search_range, &cpi->fn_ptr[BLOCK_16X16], x->mvcost, x->mvsadcost,&best_ref_mv);
- }
-
- // Barrier threshold to initiating full search
- // full_flag_thresh = 10 + (thissme >> 7);
- if ((thissme + full_flag_thresh) < bestsme)
- {
- cpi->do_full[lf_or_gf] ++;
- bestsme = thissme;
- }
- else if (thissme < bestsme)
- bestsme = thissme;
- else
- {
- cpi->do_full[lf_or_gf] = cpi->do_full[lf_or_gf] >> 1;
- cpi->check_freq[lf_or_gf] = cpi->sf.full_freq[lf_or_gf];
-
- // The full search result is actually worse so re-instate the previous best vector
- d->bmi.mv.as_mv.row = mode_mv[NEWMV].row;
- d->bmi.mv.as_mv.col = mode_mv[NEWMV].col;
- }
- }
-
- x->mv_col_min = tmp_col_min;
- x->mv_col_max = tmp_col_max;
- x->mv_row_min = tmp_row_min;
- x->mv_row_max = tmp_row_max;
-
- if (bestsme < INT_MAX)
- // cpi->find_fractional_mv_step(x,b,d,&d->bmi.mv.as_mv,&best_ref_mv,x->errorperbit/2,cpi->fn_ptr.svf,cpi->fn_ptr.vf,x->mvcost); // normal mvc=11
- cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv.as_mv, &best_ref_mv, x->errorperbit / 4, &cpi->fn_ptr[BLOCK_16X16], x->mvcost);
-
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
-
- // Add the new motion vector cost to our rolling cost variable
- rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv, x->mvcost, 96);
-
- }
-
- case NEARESTMV:
- case NEARMV:
-
- // Clip "next_nearest" so that it does not extend to far out of image
- if (mode_mv[this_mode].col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN))
- mode_mv[this_mode].col = xd->mb_to_left_edge - LEFT_TOP_MARGIN;
- else if (mode_mv[this_mode].col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN)
- mode_mv[this_mode].col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN;
-
- if (mode_mv[this_mode].row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN))
- mode_mv[this_mode].row = xd->mb_to_top_edge - LEFT_TOP_MARGIN;
- else if (mode_mv[this_mode].row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN)
- mode_mv[this_mode].row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN;
-
- // Do not bother proceeding if the vector (from newmv,nearest or near) is 0,0 as this should then be coded using the zeromv mode.
- if (((this_mode == NEARMV) || (this_mode == NEARESTMV)) &&
- ((mode_mv[this_mode].row == 0) && (mode_mv[this_mode].col == 0)))
- continue;
-
- case ZEROMV:
-
- mv_selected:
-
- // Trap vectors that reach beyond the UMV borders
- // Note that ALL New MV, Nearest MV Near MV and Zero MV code drops through to this point
- // because of the lack of break statements in the previous two cases.
- if (((mode_mv[this_mode].row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].row >> 3) > x->mv_row_max) ||
- ((mode_mv[this_mode].col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].col >> 3) > x->mv_col_max))
- continue;
-
- vp8_set_mbmode_and_mvs(x, this_mode, &mode_mv[this_mode]);
- vp8_build_inter_predictors_mby(&x->e_mbd);
-
- if (cpi->active_map_enabled && x->active_ptr[0] == 0) {
- x->skip = 1;
- }
- else if (x->encode_breakout)
- {
- int sum, sse;
- int threshold = (xd->block[0].dequant[1]
- * xd->block[0].dequant[1] >>4);
-
- if(threshold < x->encode_breakout)
- threshold = x->encode_breakout;
-
- VARIANCE_INVOKE(&cpi->rtcd.variance, get16x16var)
- (x->src.y_buffer, x->src.y_stride,
- x->e_mbd.predictor, 16, (unsigned int *)(&sse), &sum);
-
- if (sse < threshold)
- {
- // Check u and v to make sure skip is ok
- int sse2 = 0;
- /* If theres is no codeable 2nd order dc
- or a very small uniform pixel change change */
- if (abs(sum) < (xd->block[24].dequant[0]<<2)||
- ((sum * sum>>8) > sse && abs(sum) <128))
- {
- sse2 = VP8_UVSSE(x, IF_RTCD(&cpi->rtcd.variance));
-
- if (sse2 * 2 < threshold)
- {
- x->skip = 1;
- distortion2 = sse + sse2;
- rate2 = 500;
-
- /* for best_yrd calculation */
- rate_uv = 0;
- distortion_uv = sse2;
-
- disable_skip = 1;
- this_rd = RDCOST(x->rdmult, x->rddiv, rate2,
- distortion2);
-
- break;
- }
- }
- }
- }
-
-
- //intermodecost[mode_index] = vp8_cost_mv_ref(this_mode, mdcounts); // Experimental debug code
-
- // Add in the Mv/mode cost
- rate2 += vp8_cost_mv_ref(this_mode, mdcounts);
-
- // Y cost and distortion
- macro_block_yrd(x, &rate_y, &distortion, IF_RTCD(&cpi->rtcd.encodemb));
- rate2 += rate_y;
- distortion2 += distortion;
-
- // UV cost and distortion
- vp8_rd_inter_uv(cpi, x, &rate_uv, &distortion_uv, cpi->common.full_pixel);
- rate2 += rate_uv;
- distortion2 += distortion_uv;
- break;
-
- default:
- break;
- }
-
- // Where skip is allowable add in the default per mb cost for the no skip case.
- // where we then decide to skip we have to delete this and replace it with the
- // cost of signallying a skip
- if (cpi->common.mb_no_coeff_skip)
- {
- other_cost += vp8_cost_bit(cpi->prob_skip_false, 0);
- rate2 += other_cost;
- }
-
- // Estimate the reference frame signaling cost and add it to the rolling cost variable.
- rate2 += ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
-
- if (!disable_skip)
- {
- // Test for the condition where skip block will be activated because there are no non zero coefficients and make any necessary adjustment for rate
- if (cpi->common.mb_no_coeff_skip)
- {
- int tteob;
-
- tteob = 0;
-
- for (i = 0; i <= 24; i++)
- {
- tteob += x->e_mbd.block[i].eob;
- }
-
- if (tteob == 0)
- {
- rate2 -= (rate_y + rate_uv);
- //for best_yrd calculation
- rate_uv = 0;
-
- // Back out no skip flag costing and add in skip flag costing
- if (cpi->prob_skip_false)
- {
- int prob_skip_cost;
-
- prob_skip_cost = vp8_cost_bit(cpi->prob_skip_false, 1);
- prob_skip_cost -= vp8_cost_bit(cpi->prob_skip_false, 0);
- rate2 += prob_skip_cost;
- other_cost += prob_skip_cost;
- }
- }
- }
- // Calculate the final RD estimate for this mode
- this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
- }
-
- // Experimental debug code.
- //all_rds[mode_index] = this_rd;
- //all_rates[mode_index] = rate2;
- //all_dist[mode_index] = distortion2;
-
- if ((x->e_mbd.mode_info_context->mbmi.ref_frame == INTRA_FRAME) && (this_rd < *returnintra))
- {
- *returnintra = this_rd ;
- }
-
- // Did this mode help.. i.i is it the new best mode
- if (this_rd < best_rd || x->skip)
- {
- // Note index of best mode so far
- best_mode_index = mode_index;
- x->e_mbd.mode_info_context->mbmi.force_no_skip = force_no_skip;
-
- if (this_mode <= B_PRED)
- {
- x->e_mbd.mode_info_context->mbmi.uv_mode = uv_intra_mode;
- }
-
- other_cost += ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
-
- /* Calculate the final y RD estimate for this mode */
- best_yrd = RDCOST(x->rdmult, x->rddiv, (rate2-rate_uv-other_cost),
- (distortion2-distortion_uv));
-
- *returnrate = rate2;
- *returndistortion = distortion2;
- best_rd = this_rd;
- vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi, sizeof(MB_MODE_INFO));
- vpx_memcpy(&best_partition, x->partition_info, sizeof(PARTITION_INFO));
-
- for (i = 0; i < 16; i++)
- {
- vpx_memcpy(&best_bmodes[i], &x->e_mbd.block[i].bmi, sizeof(B_MODE_INFO));
- }
-
- // Testing this mode gave rise to an improvement in best error score. Lower threshold a bit for next time
- cpi->rd_thresh_mult[mode_index] = (cpi->rd_thresh_mult[mode_index] >= (MIN_THRESHMULT + 2)) ? cpi->rd_thresh_mult[mode_index] - 2 : MIN_THRESHMULT;
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
- }
-
- // If the mode did not help improve the best error case then raise the threshold for testing that mode next time around.
- else
- {
- cpi->rd_thresh_mult[mode_index] += 4;
-
- if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
- cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
-
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
- }
-
- if (x->skip)
- break;
-
- }
-
- // Reduce the activation RD thresholds for the best choice mode
- if ((cpi->rd_baseline_thresh[best_mode_index] > 0) && (cpi->rd_baseline_thresh[best_mode_index] < (INT_MAX >> 2)))
- {
- int best_adjustment = (cpi->rd_thresh_mult[best_mode_index] >> 2);
-
- cpi->rd_thresh_mult[best_mode_index] = (cpi->rd_thresh_mult[best_mode_index] >= (MIN_THRESHMULT + best_adjustment)) ? cpi->rd_thresh_mult[best_mode_index] - best_adjustment : MIN_THRESHMULT;
- cpi->rd_threshes[best_mode_index] = (cpi->rd_baseline_thresh[best_mode_index] >> 7) * cpi->rd_thresh_mult[best_mode_index];
-
- // If we chose a split mode then reset the new MV thresholds as well
- /*if ( vp8_mode_order[best_mode_index] == SPLITMV )
- {
- best_adjustment = 4; //(cpi->rd_thresh_mult[THR_NEWMV] >> 4);
- cpi->rd_thresh_mult[THR_NEWMV] = (cpi->rd_thresh_mult[THR_NEWMV] >= (MIN_THRESHMULT+best_adjustment)) ? cpi->rd_thresh_mult[THR_NEWMV]-best_adjustment: MIN_THRESHMULT;
- cpi->rd_threshes[THR_NEWMV] = (cpi->rd_baseline_thresh[THR_NEWMV] >> 7) * cpi->rd_thresh_mult[THR_NEWMV];
-
- best_adjustment = 4; //(cpi->rd_thresh_mult[THR_NEWG] >> 4);
- cpi->rd_thresh_mult[THR_NEWG] = (cpi->rd_thresh_mult[THR_NEWG] >= (MIN_THRESHMULT+best_adjustment)) ? cpi->rd_thresh_mult[THR_NEWG]-best_adjustment: MIN_THRESHMULT;
- cpi->rd_threshes[THR_NEWG] = (cpi->rd_baseline_thresh[THR_NEWG] >> 7) * cpi->rd_thresh_mult[THR_NEWG];
-
- best_adjustment = 4; //(cpi->rd_thresh_mult[THR_NEWA] >> 4);
- cpi->rd_thresh_mult[THR_NEWA] = (cpi->rd_thresh_mult[THR_NEWA] >= (MIN_THRESHMULT+best_adjustment)) ? cpi->rd_thresh_mult[THR_NEWA]-best_adjustment: MIN_THRESHMULT;
- cpi->rd_threshes[THR_NEWA] = (cpi->rd_baseline_thresh[THR_NEWA] >> 7) * cpi->rd_thresh_mult[THR_NEWA];
- }*/
-
- }
-
- // If we have chosen new mv or split then decay the full search check count more quickly.
- if ((vp8_mode_order[best_mode_index] == NEWMV) || (vp8_mode_order[best_mode_index] == SPLITMV))
- {
- int lf_or_gf = (vp8_ref_frame_order[best_mode_index] == LAST_FRAME) ? 0 : 1;
-
- if (cpi->check_freq[lf_or_gf] && !cpi->do_full[lf_or_gf])
- {
- cpi->check_freq[lf_or_gf] --;
- }
- }
-
- // Keep a record of best mode index that we chose
- cpi->last_best_mode_index = best_mode_index;
-
- // Note how often each mode chosen as best
- cpi->mode_chosen_counts[best_mode_index] ++;
-
-
- if (cpi->is_src_frame_alt_ref && (best_mbmode.mode != ZEROMV || best_mbmode.ref_frame != ALTREF_FRAME))
- {
- best_mbmode.mode = ZEROMV;
- best_mbmode.ref_frame = ALTREF_FRAME;
- best_mbmode.mv.as_int = 0;
- best_mbmode.uv_mode = 0;
- best_mbmode.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0;
- best_mbmode.partitioning = 0;
- best_mbmode.dc_diff = 0;
-
- vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO));
- vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO));
-
- for (i = 0; i < 16; i++)
- {
- vpx_memset(&x->e_mbd.block[i].bmi, 0, sizeof(B_MODE_INFO));
- }
-
- x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
-
- return best_rd;
- }
-
-
- if(best_mbmode.mode <= B_PRED)
- {
- int i;
- for (i = 0; i < 16; i++)
- {
- best_bmodes[i].mv.as_int = 0;
- }
- }
-
- // macroblock modes
- vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO));
- vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO));
-
- for (i = 0; i < 16; i++)
- {
- vpx_memcpy(&x->e_mbd.block[i].bmi, &best_bmodes[i], sizeof(B_MODE_INFO));
- }
-
- x->e_mbd.mode_info_context->mbmi.mv.as_mv = x->e_mbd.block[15].bmi.mv.as_mv;
-
- return best_rd;
-}
-#endif
diff --git a/vp8/encoder/rdopt.h b/vp8/encoder/rdopt.h
deleted file mode 100644
index 72ba9a0..0000000
--- a/vp8/encoder/rdopt.h
+++ /dev/null
@@ -1,33 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_RDOPT_H
-#define __INC_RDOPT_H
-void vp8_initialize_rd_consts(VP8_COMP *cpi, int Qvalue);
-int vp8_rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *rate, int *rate_to, int *distortion, int best_rd);
-int vp8_rd_pick_intra16x16mby_mode(VP8_COMP *cpi, MACROBLOCK *x, int *returnrate, int *rate_to, int *returndistortion);
-int vp8_rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *rate_to, int *distortion);
-extern int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra);
-
-extern void vp8_mv_pred
-(
- VP8_COMP *cpi,
- MACROBLOCKD *xd,
- const MODE_INFO *here,
- MV *mvp,
- int refframe,
- int *ref_frame_sign_bias,
- int *sr,
- int near_sadidx[]
-);
-void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffset, int near_sadidx[]);
-
-#endif
diff --git a/vp8/encoder/sad_c.c b/vp8/encoder/sad_c.c
deleted file mode 100644
index 5eaca59..0000000
--- a/vp8/encoder/sad_c.c
+++ /dev/null
@@ -1,339 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdlib.h>
-
-unsigned int vp8_sad16x16_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- int max_sad)
-{
-
- int r, c;
- unsigned int sad = 0;
-
- for (r = 0; r < 16; r++)
- {
- for (c = 0; c < 16; c++)
- {
- sad += abs(src_ptr[c] - ref_ptr[c]);
- }
-
- src_ptr += src_stride;
- ref_ptr += ref_stride;
- }
-
- return sad;
-}
-
-
-static __inline
-unsigned int sad_mx_n_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- int m,
- int n)
-{
-
- int r, c;
- unsigned int sad = 0;
-
- for (r = 0; r < n; r++)
- {
- for (c = 0; c < m; c++)
- {
- sad += abs(src_ptr[c] - ref_ptr[c]);
- }
-
- src_ptr += src_stride;
- ref_ptr += ref_stride;
- }
-
- return sad;
-}
-
-
-unsigned int vp8_sad8x8_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- int max_sad)
-{
-
- return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 8, 8);
-}
-
-
-unsigned int vp8_sad16x8_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- int max_sad)
-{
-
- return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 16, 8);
-
-}
-
-
-unsigned int vp8_sad8x16_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- int max_sad)
-{
-
- return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 8, 16);
-}
-
-
-unsigned int vp8_sad4x4_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- int max_sad)
-{
-
- return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 4, 4);
-}
-
-void vp8_sad16x16x3_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad16x16x8_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned short *sad_array
-)
-{
- sad_array[0] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
- sad_array[3] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 3 , ref_stride, 0x7fffffff);
- sad_array[4] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
- sad_array[5] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
- sad_array[6] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 6 , ref_stride, 0x7fffffff);
- sad_array[7] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad16x8x3_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad16x8x8_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned short *sad_array
-)
-{
- sad_array[0] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
- sad_array[3] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 3 , ref_stride, 0x7fffffff);
- sad_array[4] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
- sad_array[5] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
- sad_array[6] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 6 , ref_stride, 0x7fffffff);
- sad_array[7] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad8x8x3_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad8x8x8_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned short *sad_array
-)
-{
- sad_array[0] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
- sad_array[3] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 3 , ref_stride, 0x7fffffff);
- sad_array[4] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
- sad_array[5] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
- sad_array[6] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 6 , ref_stride, 0x7fffffff);
- sad_array[7] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad8x16x3_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad8x16x8_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned short *sad_array
-)
-{
- sad_array[0] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
- sad_array[3] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 3 , ref_stride, 0x7fffffff);
- sad_array[4] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
- sad_array[5] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
- sad_array[6] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 6 , ref_stride, 0x7fffffff);
- sad_array[7] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad4x4x3_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad4x4x8_c(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride,
- unsigned short *sad_array
-)
-{
- sad_array[0] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr , ref_stride, 0x7fffffff);
- sad_array[1] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
- sad_array[2] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
- sad_array[3] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 3 , ref_stride, 0x7fffffff);
- sad_array[4] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
- sad_array[5] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
- sad_array[6] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 6 , ref_stride, 0x7fffffff);
- sad_array[7] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, 0x7fffffff);
-}
-
-void vp8_sad16x16x4d_c(
- const unsigned char *src_ptr,
- int src_stride,
- unsigned char *ref_ptr[],
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[0], ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
- sad_array[3] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
-}
-
-void vp8_sad16x8x4d_c(
- const unsigned char *src_ptr,
- int src_stride,
- unsigned char *ref_ptr[],
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[0], ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
- sad_array[3] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
-}
-
-void vp8_sad8x8x4d_c(
- const unsigned char *src_ptr,
- int src_stride,
- unsigned char *ref_ptr[],
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[0], ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
- sad_array[3] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
-}
-
-void vp8_sad8x16x4d_c(
- const unsigned char *src_ptr,
- int src_stride,
- unsigned char *ref_ptr[],
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[0], ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
- sad_array[3] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
-}
-
-void vp8_sad4x4x4d_c(
- const unsigned char *src_ptr,
- int src_stride,
- unsigned char *ref_ptr[],
- int ref_stride,
- unsigned int *sad_array
-)
-{
- sad_array[0] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[0], ref_stride, 0x7fffffff);
- sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
- sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
- sad_array[3] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
-}
diff --git a/vp8/encoder/segmentation.c b/vp8/encoder/segmentation.c
deleted file mode 100644
index fc0967d..0000000
--- a/vp8/encoder/segmentation.c
+++ /dev/null
@@ -1,64 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "segmentation.h"
-#include "vpx_mem/vpx_mem.h"
-
-void vp8_update_gf_useage_maps(VP8_COMP *cpi, VP8_COMMON *cm, MACROBLOCK *x)
-{
- int mb_row, mb_col;
-
- MODE_INFO *this_mb_mode_info = cm->mi;
-
- x->gf_active_ptr = (signed char *)cpi->gf_active_flags;
-
- if ((cm->frame_type == KEY_FRAME) || (cm->refresh_golden_frame))
- {
- // Reset Gf useage monitors
- vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols));
- cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
- }
- else
- {
- // for each macroblock row in image
- for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
- {
- // for each macroblock col in image
- for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
- {
-
- // If using golden then set GF active flag if not already set.
- // If using last frame 0,0 mode then leave flag as it is
- // else if using non 0,0 motion or intra modes then clear flag if it is currently set
- if ((this_mb_mode_info->mbmi.ref_frame == GOLDEN_FRAME) || (this_mb_mode_info->mbmi.ref_frame == ALTREF_FRAME))
- {
- if (*(x->gf_active_ptr) == 0)
- {
- *(x->gf_active_ptr) = 1;
- cpi->gf_active_count ++;
- }
- }
- else if ((this_mb_mode_info->mbmi.mode != ZEROMV) && *(x->gf_active_ptr))
- {
- *(x->gf_active_ptr) = 0;
- cpi->gf_active_count--;
- }
-
- x->gf_active_ptr++; // Step onto next entry
- this_mb_mode_info++; // skip to next mb
-
- }
-
- // this is to account for the border
- this_mb_mode_info++;
- }
- }
-}
diff --git a/vp8/encoder/ssim.c b/vp8/encoder/ssim.c
deleted file mode 100644
index 64d67c6..0000000
--- a/vp8/encoder/ssim.c
+++ /dev/null
@@ -1,408 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_scale/yv12config.h"
-#include "math.h"
-#include "onyx_int.h"
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define IF_RTCD(x) (x)
-#else
-#define IF_RTCD(x) NULL
-#endif
-// Google version of SSIM
-// SSIM
-#define KERNEL 3
-#define KERNEL_SIZE (2 * KERNEL + 1)
-
-typedef unsigned char uint8;
-typedef unsigned int uint32;
-
-static const int K[KERNEL_SIZE] =
-{
- 1, 4, 11, 16, 11, 4, 1 // 16 * exp(-0.3 * i * i)
-};
-static const double ki_w = 1. / 2304.; // 1 / sum(i:0..6, j..6) K[i]*K[j]
-double get_ssimg(const uint8 *org, const uint8 *rec,
- int xo, int yo, int W, int H,
- const int stride1, const int stride2
- )
-{
- // TODO(skal): use summed tables
- int y, x;
-
- const int ymin = (yo - KERNEL < 0) ? 0 : yo - KERNEL;
- const int ymax = (yo + KERNEL > H - 1) ? H - 1 : yo + KERNEL;
- const int xmin = (xo - KERNEL < 0) ? 0 : xo - KERNEL;
- const int xmax = (xo + KERNEL > W - 1) ? W - 1 : xo + KERNEL;
- // worst case of accumulation is a weight of 48 = 16 + 2 * (11 + 4 + 1)
- // with a diff of 255, squares. That would a max error of 0x8ee0900,
- // which fits into 32 bits integers.
- uint32 w = 0, xm = 0, ym = 0, xxm = 0, xym = 0, yym = 0;
- org += ymin * stride1;
- rec += ymin * stride2;
-
- for (y = ymin; y <= ymax; ++y, org += stride1, rec += stride2)
- {
- const int Wy = K[KERNEL + y - yo];
-
- for (x = xmin; x <= xmax; ++x)
- {
- const int Wxy = Wy * K[KERNEL + x - xo];
- // TODO(skal): inlined assembly
- w += Wxy;
- xm += Wxy * org[x];
- ym += Wxy * rec[x];
- xxm += Wxy * org[x] * org[x];
- xym += Wxy * org[x] * rec[x];
- yym += Wxy * rec[x] * rec[x];
- }
- }
-
- {
- const double iw = 1. / w;
- const double iwx = xm * iw;
- const double iwy = ym * iw;
- double sxx = xxm * iw - iwx * iwx;
- double syy = yym * iw - iwy * iwy;
-
- // small errors are possible, due to rounding. Clamp to zero.
- if (sxx < 0.) sxx = 0.;
-
- if (syy < 0.) syy = 0.;
-
- {
- const double sxsy = sqrt(sxx * syy);
- const double sxy = xym * iw - iwx * iwy;
- static const double C11 = (0.01 * 0.01) * (255 * 255);
- static const double C22 = (0.03 * 0.03) * (255 * 255);
- static const double C33 = (0.015 * 0.015) * (255 * 255);
- const double l = (2. * iwx * iwy + C11) / (iwx * iwx + iwy * iwy + C11);
- const double c = (2. * sxsy + C22) / (sxx + syy + C22);
-
- const double s = (sxy + C33) / (sxsy + C33);
- return l * c * s;
-
- }
- }
-
-}
-
-double get_ssimfull_kernelg(const uint8 *org, const uint8 *rec,
- int xo, int yo, int W, int H,
- const int stride1, const int stride2)
-{
- // TODO(skal): use summed tables
- // worst case of accumulation is a weight of 48 = 16 + 2 * (11 + 4 + 1)
- // with a diff of 255, squares. That would a max error of 0x8ee0900,
- // which fits into 32 bits integers.
- int y_, x_;
- uint32 xm = 0, ym = 0, xxm = 0, xym = 0, yym = 0;
- org += (yo - KERNEL) * stride1;
- org += (xo - KERNEL);
- rec += (yo - KERNEL) * stride2;
- rec += (xo - KERNEL);
-
- for (y_ = 0; y_ < KERNEL_SIZE; ++y_, org += stride1, rec += stride2)
- {
- const int Wy = K[y_];
-
- for (x_ = 0; x_ < KERNEL_SIZE; ++x_)
- {
- const int Wxy = Wy * K[x_];
- // TODO(skal): inlined assembly
- const int org_x = org[x_];
- const int rec_x = rec[x_];
- xm += Wxy * org_x;
- ym += Wxy * rec_x;
- xxm += Wxy * org_x * org_x;
- xym += Wxy * org_x * rec_x;
- yym += Wxy * rec_x * rec_x;
- }
- }
-
- {
- const double iw = ki_w;
- const double iwx = xm * iw;
- const double iwy = ym * iw;
- double sxx = xxm * iw - iwx * iwx;
- double syy = yym * iw - iwy * iwy;
-
- // small errors are possible, due to rounding. Clamp to zero.
- if (sxx < 0.) sxx = 0.;
-
- if (syy < 0.) syy = 0.;
-
- {
- const double sxsy = sqrt(sxx * syy);
- const double sxy = xym * iw - iwx * iwy;
- static const double C11 = (0.01 * 0.01) * (255 * 255);
- static const double C22 = (0.03 * 0.03) * (255 * 255);
- static const double C33 = (0.015 * 0.015) * (255 * 255);
- const double l = (2. * iwx * iwy + C11) / (iwx * iwx + iwy * iwy + C11);
- const double c = (2. * sxsy + C22) / (sxx + syy + C22);
- const double s = (sxy + C33) / (sxsy + C33);
- return l * c * s;
- }
- }
-}
-
-double calc_ssimg(const uint8 *org, const uint8 *rec,
- const int image_width, const int image_height,
- const int stride1, const int stride2
- )
-{
- int j, i;
- double SSIM = 0.;
-
- for (j = 0; j < KERNEL; ++j)
- {
- for (i = 0; i < image_width; ++i)
- {
- SSIM += get_ssimg(org, rec, i, j, image_width, image_height, stride1, stride2);
- }
- }
-
- for (j = KERNEL; j < image_height - KERNEL; ++j)
- {
- for (i = 0; i < KERNEL; ++i)
- {
- SSIM += get_ssimg(org, rec, i, j, image_width, image_height, stride1, stride2);
- }
-
- for (i = KERNEL; i < image_width - KERNEL; ++i)
- {
- SSIM += get_ssimfull_kernelg(org, rec, i, j,
- image_width, image_height, stride1, stride2);
- }
-
- for (i = image_width - KERNEL; i < image_width; ++i)
- {
- SSIM += get_ssimg(org, rec, i, j, image_width, image_height, stride1, stride2);
- }
- }
-
- for (j = image_height - KERNEL; j < image_height; ++j)
- {
- for (i = 0; i < image_width; ++i)
- {
- SSIM += get_ssimg(org, rec, i, j, image_width, image_height, stride1, stride2);
- }
- }
-
- return SSIM;
-}
-
-
-double vp8_calc_ssimg
-(
- YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *dest,
- double *ssim_y,
- double *ssim_u,
- double *ssim_v
-)
-{
- double ssim_all = 0;
- int ysize = source->y_width * source->y_height;
- int uvsize = ysize / 4;
-
- *ssim_y = calc_ssimg(source->y_buffer, dest->y_buffer,
- source->y_width, source->y_height,
- source->y_stride, dest->y_stride);
-
-
- *ssim_u = calc_ssimg(source->u_buffer, dest->u_buffer,
- source->uv_width, source->uv_height,
- source->uv_stride, dest->uv_stride);
-
-
- *ssim_v = calc_ssimg(source->v_buffer, dest->v_buffer,
- source->uv_width, source->uv_height,
- source->uv_stride, dest->uv_stride);
-
- ssim_all = (*ssim_y + *ssim_u + *ssim_v) / (ysize + uvsize + uvsize);
- *ssim_y /= ysize;
- *ssim_u /= uvsize;
- *ssim_v /= uvsize;
- return ssim_all;
-}
-
-
-void ssim_parms_c
-(
- unsigned char *s,
- int sp,
- unsigned char *r,
- int rp,
- unsigned long *sum_s,
- unsigned long *sum_r,
- unsigned long *sum_sq_s,
- unsigned long *sum_sq_r,
- unsigned long *sum_sxr
-)
-{
- int i,j;
- for(i=0;i<16;i++,s+=sp,r+=rp)
- {
- for(j=0;j<16;j++)
- {
- *sum_s += s[j];
- *sum_r += r[j];
- *sum_sq_s += s[j] * s[j];
- *sum_sq_r += r[j] * r[j];
- *sum_sxr += s[j] * r[j];
- }
- }
-}
-void ssim_parms_8x8_c
-(
- unsigned char *s,
- int sp,
- unsigned char *r,
- int rp,
- unsigned long *sum_s,
- unsigned long *sum_r,
- unsigned long *sum_sq_s,
- unsigned long *sum_sq_r,
- unsigned long *sum_sxr
-)
-{
- int i,j;
- for(i=0;i<8;i++,s+=sp,r+=rp)
- {
- for(j=0;j<8;j++)
- {
- *sum_s += s[j];
- *sum_r += r[j];
- *sum_sq_s += s[j] * s[j];
- *sum_sq_r += r[j] * r[j];
- *sum_sxr += s[j] * r[j];
- }
- }
-}
-
-const static long long c1 = 426148; // (256^2*(.01*255)^2
-const static long long c2 = 3835331; //(256^2*(.03*255)^2
-
-static double similarity
-(
- unsigned long sum_s,
- unsigned long sum_r,
- unsigned long sum_sq_s,
- unsigned long sum_sq_r,
- unsigned long sum_sxr,
- int count
-)
-{
- long long ssim_n = (2*sum_s*sum_r+ c1)*(2*count*sum_sxr-2*sum_s*sum_r+c2);
-
- long long ssim_d = (sum_s*sum_s +sum_r*sum_r+c1)*
- (count*sum_sq_s-sum_s*sum_s + count*sum_sq_r-sum_r*sum_r +c2) ;
-
- return ssim_n * 1.0 / ssim_d;
-}
-
-static double ssim_16x16(unsigned char *s,int sp, unsigned char *r,int rp,
- const vp8_variance_rtcd_vtable_t *rtcd)
-{
- unsigned long sum_s=0,sum_r=0,sum_sq_s=0,sum_sq_r=0,sum_sxr=0;
- rtcd->ssimpf(s, sp, r, rp, &sum_s, &sum_r, &sum_sq_s, &sum_sq_r, &sum_sxr);
- return similarity(sum_s, sum_r, sum_sq_s, sum_sq_r, sum_sxr, 256);
-}
-static double ssim_8x8(unsigned char *s,int sp, unsigned char *r,int rp,
- const vp8_variance_rtcd_vtable_t *rtcd)
-{
- unsigned long sum_s=0,sum_r=0,sum_sq_s=0,sum_sq_r=0,sum_sxr=0;
- rtcd->ssimpf_8x8(s, sp, r, rp, &sum_s, &sum_r, &sum_sq_s, &sum_sq_r, &sum_sxr);
- return similarity(sum_s, sum_r, sum_sq_s, sum_sq_r, sum_sxr, 64);
-}
-
-// TODO: (jbb) tried to scale this function such that we may be able to use it
-// for distortion metric in mode selection code ( provided we do a reconstruction)
-long dssim(unsigned char *s,int sp, unsigned char *r,int rp,
- const vp8_variance_rtcd_vtable_t *rtcd)
-{
- unsigned long sum_s=0,sum_r=0,sum_sq_s=0,sum_sq_r=0,sum_sxr=0;
- double ssim3;
- long long ssim_n;
- long long ssim_d;
-
- rtcd->ssimpf(s, sp, r, rp, &sum_s, &sum_r, &sum_sq_s, &sum_sq_r, &sum_sxr);
- ssim_n = (2*sum_s*sum_r+ c1)*(2*256*sum_sxr-2*sum_s*sum_r+c2);
-
- ssim_d = (sum_s*sum_s +sum_r*sum_r+c1)*
- (256*sum_sq_s-sum_s*sum_s + 256*sum_sq_r-sum_r*sum_r +c2) ;
-
- ssim3 = 256 * (ssim_d-ssim_n) / ssim_d;
- return (long)( 256*ssim3 * ssim3 );
-}
-// TODO: (jbb) this 8x8 window might be too big + we may want to pick pixels
-// such that the window regions overlap block boundaries to penalize blocking
-// artifacts.
-
-double vp8_ssim2
-(
- unsigned char *img1,
- unsigned char *img2,
- int stride_img1,
- int stride_img2,
- int width,
- int height,
- const vp8_variance_rtcd_vtable_t *rtcd
-)
-{
- int i,j;
-
- double ssim_total=0;
-
- // we can sample points as frequently as we like start with 1 per 8x8
- for(i=0; i < height; i+=8, img1 += stride_img1*8, img2 += stride_img2*8)
- {
- for(j=0; j < width; j+=8 )
- {
- ssim_total += ssim_8x8(img1, stride_img1, img2, stride_img2, rtcd);
- }
- }
- ssim_total /= (width/8 * height /8);
- return ssim_total;
-
-}
-double vp8_calc_ssim
-(
- YV12_BUFFER_CONFIG *source,
- YV12_BUFFER_CONFIG *dest,
- int lumamask,
- double *weight,
- const vp8_variance_rtcd_vtable_t *rtcd
-)
-{
- double a, b, c;
- double ssimv;
-
- a = vp8_ssim2(source->y_buffer, dest->y_buffer,
- source->y_stride, dest->y_stride, source->y_width,
- source->y_height, rtcd);
-
- b = vp8_ssim2(source->u_buffer, dest->u_buffer,
- source->uv_stride, dest->uv_stride, source->uv_width,
- source->uv_height, rtcd);
-
- c = vp8_ssim2(source->v_buffer, dest->v_buffer,
- source->uv_stride, dest->uv_stride, source->uv_width,
- source->uv_height, rtcd);
-
- ssimv = a * .8 + .1 * (b + c);
-
- *weight = 1;
-
- return ssimv;
-}
diff --git a/vp8/encoder/temporal_filter.c b/vp8/encoder/temporal_filter.c
deleted file mode 100644
index fd36b22..0000000
--- a/vp8/encoder/temporal_filter.c
+++ /dev/null
@@ -1,586 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/common/onyxc_int.h"
-#include "onyx_int.h"
-#include "vp8/common/systemdependent.h"
-#include "quantize.h"
-#include "vp8/common/alloccommon.h"
-#include "mcomp.h"
-#include "firstpass.h"
-#include "psnr.h"
-#include "vpx_scale/vpxscale.h"
-#include "vp8/common/extend.h"
-#include "ratectrl.h"
-#include "vp8/common/quant_common.h"
-#include "segmentation.h"
-#include "vp8/common/g_common.h"
-#include "vpx_scale/yv12extend.h"
-#include "vp8/common/postproc.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vp8/common/swapyv12buffer.h"
-#include "vp8/common/threading.h"
-#include "vpx_ports/vpx_timer.h"
-
-#include <math.h>
-#include <limits.h>
-
-#define ALT_REF_MC_ENABLED 1 // dis/enable MC in AltRef filtering
-#define ALT_REF_SUBPEL_ENABLED 1 // dis/enable subpel in MC AltRef filtering
-
-#if VP8_TEMPORAL_ALT_REF
-
-static void vp8_temporal_filter_predictors_mb_c
-(
- MACROBLOCKD *x,
- unsigned char *y_mb_ptr,
- unsigned char *u_mb_ptr,
- unsigned char *v_mb_ptr,
- int stride,
- int mv_row,
- int mv_col,
- unsigned char *pred
-)
-{
- int offset;
- unsigned char *yptr, *uptr, *vptr;
-
- // Y
- yptr = y_mb_ptr + (mv_row >> 3) * stride + (mv_col >> 3);
-
- if ((mv_row | mv_col) & 7)
- {
- x->subpixel_predict16x16(yptr, stride,
- mv_col & 7, mv_row & 7, &pred[0], 16);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy16x16)(yptr, stride, &pred[0], 16);
- }
-
- // U & V
- mv_row >>= 1;
- mv_col >>= 1;
- stride = (stride + 1) >> 1;
- offset = (mv_row >> 3) * stride + (mv_col >> 3);
- uptr = u_mb_ptr + offset;
- vptr = v_mb_ptr + offset;
-
- if ((mv_row | mv_col) & 7)
- {
- x->subpixel_predict8x8(uptr, stride,
- mv_col & 7, mv_row & 7, &pred[256], 8);
- x->subpixel_predict8x8(vptr, stride,
- mv_col & 7, mv_row & 7, &pred[320], 8);
- }
- else
- {
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(uptr, stride, &pred[256], 8);
- RECON_INVOKE(&x->rtcd->recon, copy8x8)(vptr, stride, &pred[320], 8);
- }
-}
-void vp8_temporal_filter_apply_c
-(
- unsigned char *frame1,
- unsigned int stride,
- unsigned char *frame2,
- unsigned int block_size,
- int strength,
- int filter_weight,
- unsigned int *accumulator,
- unsigned short *count
-)
-{
- int i, j, k;
- int modifier;
- int byte = 0;
-
- for (i = 0,k = 0; i < block_size; i++)
- {
- for (j = 0; j < block_size; j++, k++)
- {
-
- int src_byte = frame1[byte];
- int pixel_value = *frame2++;
-
- modifier = src_byte - pixel_value;
- // This is an integer approximation of:
- // float coeff = (3.0 * modifer * modifier) / pow(2, strength);
- // modifier = (int)roundf(coeff > 16 ? 0 : 16-coeff);
- modifier *= modifier;
- modifier *= 3;
- modifier += 1 << (strength - 1);
- modifier >>= strength;
-
- if (modifier > 16)
- modifier = 16;
-
- modifier = 16 - modifier;
- modifier *= filter_weight;
-
- count[k] += modifier;
- accumulator[k] += modifier * pixel_value;
-
- byte++;
- }
-
- byte += stride - block_size;
- }
-}
-
-#if ALT_REF_MC_ENABLED
-static int dummy_cost[2*mv_max+1];
-
-static int vp8_temporal_filter_find_matching_mb_c
-(
- VP8_COMP *cpi,
- YV12_BUFFER_CONFIG *arf_frame,
- YV12_BUFFER_CONFIG *frame_ptr,
- int mb_offset,
- int error_thresh
-)
-{
- MACROBLOCK *x = &cpi->mb;
- int thissme;
- int step_param;
- int further_steps;
- int n = 0;
- int sadpb = x->sadperbit16;
- int bestsme = INT_MAX;
- int num00 = 0;
-
- BLOCK *b = &x->block[0];
- BLOCKD *d = &x->e_mbd.block[0];
- MV best_ref_mv1 = {0,0};
-
- int *mvcost[2] = { &dummy_cost[mv_max+1], &dummy_cost[mv_max+1] };
- int *mvsadcost[2] = { &dummy_cost[mv_max+1], &dummy_cost[mv_max+1] };
-
- // Save input state
- unsigned char **base_src = b->base_src;
- int src = b->src;
- int src_stride = b->src_stride;
- unsigned char **base_pre = d->base_pre;
- int pre = d->pre;
- int pre_stride = d->pre_stride;
-
- // Setup frame pointers
- b->base_src = &arf_frame->y_buffer;
- b->src_stride = arf_frame->y_stride;
- b->src = mb_offset;
-
- d->base_pre = &frame_ptr->y_buffer;
- d->pre_stride = frame_ptr->y_stride;
- d->pre = mb_offset;
-
- // Further step/diamond searches as necessary
- if (cpi->Speed < 8)
- {
- step_param = cpi->sf.first_step +
- ((cpi->Speed > 5) ? 1 : 0);
- further_steps =
- (cpi->sf.max_step_search_steps - 1)-step_param;
- }
- else
- {
- step_param = cpi->sf.first_step + 2;
- further_steps = 0;
- }
-
- if (1/*cpi->sf.search_method == HEX*/)
- {
- // TODO Check that the 16x16 vf & sdf are selected here
- bestsme = vp8_hex_search(x, b, d,
- &best_ref_mv1, &d->bmi.mv.as_mv,
- step_param,
- sadpb/*x->errorperbit*/,
- &num00, &cpi->fn_ptr[BLOCK_16X16],
- mvsadcost, mvcost, &best_ref_mv1);
- }
- else
- {
- int mv_x, mv_y;
-
- bestsme = cpi->diamond_search_sad(x, b, d,
- &best_ref_mv1, &d->bmi.mv.as_mv,
- step_param,
- sadpb / 2/*x->errorperbit*/,
- &num00, &cpi->fn_ptr[BLOCK_16X16],
- mvsadcost, mvcost, &best_ref_mv1); //sadpb < 9
-
- // Further step/diamond searches as necessary
- n = 0;
- //further_steps = (cpi->sf.max_step_search_steps - 1) - step_param;
-
- n = num00;
- num00 = 0;
-
- while (n < further_steps)
- {
- n++;
-
- if (num00)
- num00--;
- else
- {
- thissme = cpi->diamond_search_sad(x, b, d,
- &best_ref_mv1, &d->bmi.mv.as_mv,
- step_param + n,
- sadpb / 4/*x->errorperbit*/,
- &num00, &cpi->fn_ptr[BLOCK_16X16],
- mvsadcost, mvcost, &best_ref_mv1); //sadpb = 9
-
- if (thissme < bestsme)
- {
- bestsme = thissme;
- mv_y = d->bmi.mv.as_mv.row;
- mv_x = d->bmi.mv.as_mv.col;
- }
- else
- {
- d->bmi.mv.as_mv.row = mv_y;
- d->bmi.mv.as_mv.col = mv_x;
- }
- }
- }
- }
-
-#if ALT_REF_SUBPEL_ENABLED
- // Try sub-pixel MC?
- //if (bestsme > error_thresh && bestsme < INT_MAX)
- {
- bestsme = cpi->find_fractional_mv_step(x, b, d,
- &d->bmi.mv.as_mv, &best_ref_mv1,
- x->errorperbit, &cpi->fn_ptr[BLOCK_16X16],
- mvcost);
- }
-#endif
-
- // Save input state
- b->base_src = base_src;
- b->src = src;
- b->src_stride = src_stride;
- d->base_pre = base_pre;
- d->pre = pre;
- d->pre_stride = pre_stride;
-
- return bestsme;
-}
-#endif
-
-static void vp8_temporal_filter_iterate_c
-(
- VP8_COMP *cpi,
- int frame_count,
- int alt_ref_index,
- int strength
-)
-{
- int byte;
- int frame;
- int mb_col, mb_row;
- unsigned int filter_weight;
- int mb_cols = cpi->common.mb_cols;
- int mb_rows = cpi->common.mb_rows;
- int MBs = cpi->common.MBs;
- int mb_y_offset = 0;
- int mb_uv_offset = 0;
- DECLARE_ALIGNED_ARRAY(16, unsigned int, accumulator, 16*16 + 8*8 + 8*8);
- DECLARE_ALIGNED_ARRAY(16, unsigned short, count, 16*16 + 8*8 + 8*8);
- MACROBLOCKD *mbd = &cpi->mb.e_mbd;
- YV12_BUFFER_CONFIG *f = cpi->frames[alt_ref_index];
- unsigned char *dst1, *dst2;
- DECLARE_ALIGNED_ARRAY(16, unsigned char, predictor, 16*16 + 8*8 + 8*8);
-
- // Save input state
- unsigned char *y_buffer = mbd->pre.y_buffer;
- unsigned char *u_buffer = mbd->pre.u_buffer;
- unsigned char *v_buffer = mbd->pre.v_buffer;
-
- for (mb_row = 0; mb_row < mb_rows; mb_row++)
- {
-#if ALT_REF_MC_ENABLED
- // Reduced search extent by 3 for 6-tap filter & smaller UMV border
- cpi->mb.mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 19));
- cpi->mb.mv_row_max = ((cpi->common.mb_rows - 1 - mb_row) * 16)
- + (VP8BORDERINPIXELS - 19);
-#endif
-
- for (mb_col = 0; mb_col < mb_cols; mb_col++)
- {
- int i, j, k, w;
- int weight_cap;
- int stride;
-
- vpx_memset(accumulator, 0, 384*sizeof(unsigned int));
- vpx_memset(count, 0, 384*sizeof(unsigned short));
-
-#if ALT_REF_MC_ENABLED
- // Reduced search extent by 3 for 6-tap filter & smaller UMV border
- cpi->mb.mv_col_min = -((mb_col * 16) + (VP8BORDERINPIXELS - 19));
- cpi->mb.mv_col_max = ((cpi->common.mb_cols - 1 - mb_col) * 16)
- + (VP8BORDERINPIXELS - 19);
-#endif
-
- for (frame = 0; frame < frame_count; frame++)
- {
- int err = 0;
-
- if (cpi->frames[frame] == NULL)
- continue;
-
- mbd->block[0].bmi.mv.as_mv.row = 0;
- mbd->block[0].bmi.mv.as_mv.col = 0;
-
-#if ALT_REF_MC_ENABLED
-#define THRESH_LOW 10000
-#define THRESH_HIGH 20000
-
- // Find best match in this frame by MC
- err = vp8_temporal_filter_find_matching_mb_c
- (cpi,
- cpi->frames[alt_ref_index],
- cpi->frames[frame],
- mb_y_offset,
- THRESH_LOW);
-
-#endif
- // Assign higher weight to matching MB if it's error
- // score is lower. If not applying MC default behavior
- // is to weight all MBs equal.
- filter_weight = err<THRESH_LOW
- ? 2 : err<THRESH_HIGH ? 1 : 0;
-
- if (filter_weight != 0)
- {
- // Construct the predictors
- vp8_temporal_filter_predictors_mb_c
- (mbd,
- cpi->frames[frame]->y_buffer + mb_y_offset,
- cpi->frames[frame]->u_buffer + mb_uv_offset,
- cpi->frames[frame]->v_buffer + mb_uv_offset,
- cpi->frames[frame]->y_stride,
- mbd->block[0].bmi.mv.as_mv.row,
- mbd->block[0].bmi.mv.as_mv.col,
- predictor);
-
- // Apply the filter (YUV)
- TEMPORAL_INVOKE(&cpi->rtcd.temporal, apply)
- (f->y_buffer + mb_y_offset,
- f->y_stride,
- predictor,
- 16,
- strength,
- filter_weight,
- accumulator,
- count);
-
- TEMPORAL_INVOKE(&cpi->rtcd.temporal, apply)
- (f->u_buffer + mb_uv_offset,
- f->uv_stride,
- predictor + 256,
- 8,
- strength,
- filter_weight,
- accumulator + 256,
- count + 256);
-
- TEMPORAL_INVOKE(&cpi->rtcd.temporal, apply)
- (f->v_buffer + mb_uv_offset,
- f->uv_stride,
- predictor + 320,
- 8,
- strength,
- filter_weight,
- accumulator + 320,
- count + 320);
- }
- }
-
- // Normalize filter output to produce AltRef frame
- dst1 = cpi->alt_ref_buffer.source_buffer.y_buffer;
- stride = cpi->alt_ref_buffer.source_buffer.y_stride;
- byte = mb_y_offset;
- for (i = 0,k = 0; i < 16; i++)
- {
- for (j = 0; j < 16; j++, k++)
- {
- unsigned int pval = accumulator[k] + (count[k] >> 1);
- pval *= cpi->fixed_divide[count[k]];
- pval >>= 19;
-
- dst1[byte] = (unsigned char)pval;
-
- // move to next pixel
- byte++;
- }
-
- byte += stride - 16;
- }
-
- dst1 = cpi->alt_ref_buffer.source_buffer.u_buffer;
- dst2 = cpi->alt_ref_buffer.source_buffer.v_buffer;
- stride = cpi->alt_ref_buffer.source_buffer.uv_stride;
- byte = mb_uv_offset;
- for (i = 0,k = 256; i < 8; i++)
- {
- for (j = 0; j < 8; j++, k++)
- {
- int m=k+64;
-
- // U
- unsigned int pval = accumulator[k] + (count[k] >> 1);
- pval *= cpi->fixed_divide[count[k]];
- pval >>= 19;
- dst1[byte] = (unsigned char)pval;
-
- // V
- pval = accumulator[m] + (count[m] >> 1);
- pval *= cpi->fixed_divide[count[m]];
- pval >>= 19;
- dst2[byte] = (unsigned char)pval;
-
- // move to next pixel
- byte++;
- }
-
- byte += stride - 8;
- }
-
- mb_y_offset += 16;
- mb_uv_offset += 8;
- }
-
- mb_y_offset += 16*(f->y_stride-mb_cols);
- mb_uv_offset += 8*(f->uv_stride-mb_cols);
- }
-
- // Restore input state
- mbd->pre.y_buffer = y_buffer;
- mbd->pre.u_buffer = u_buffer;
- mbd->pre.v_buffer = v_buffer;
-}
-
-void vp8_temporal_filter_prepare_c
-(
- VP8_COMP *cpi
-)
-{
- int frame = 0;
-
- int num_frames_backward = 0;
- int num_frames_forward = 0;
- int frames_to_blur_backward = 0;
- int frames_to_blur_forward = 0;
- int frames_to_blur = 0;
- int start_frame = 0;
- unsigned int filtered = 0;
-
- int strength = cpi->oxcf.arnr_strength;
-
- int blur_type = cpi->oxcf.arnr_type;
-
- int max_frames = cpi->active_arnr_frames;
-
- num_frames_backward = cpi->last_alt_ref_sei - cpi->source_encode_index;
-
- if (num_frames_backward < 0)
- num_frames_backward += cpi->oxcf.lag_in_frames;
-
- num_frames_forward = cpi->oxcf.lag_in_frames - (num_frames_backward + 1);
-
- switch (blur_type)
- {
- case 1:
- /////////////////////////////////////////
- // Backward Blur
-
- frames_to_blur_backward = num_frames_backward;
-
- if (frames_to_blur_backward >= max_frames)
- frames_to_blur_backward = max_frames - 1;
-
- frames_to_blur = frames_to_blur_backward + 1;
- break;
-
- case 2:
- /////////////////////////////////////////
- // Forward Blur
-
- frames_to_blur_forward = num_frames_forward;
-
- if (frames_to_blur_forward >= max_frames)
- frames_to_blur_forward = max_frames - 1;
-
- frames_to_blur = frames_to_blur_forward + 1;
- break;
-
- case 3:
- default:
- /////////////////////////////////////////
- // Center Blur
- frames_to_blur_forward = num_frames_forward;
- frames_to_blur_backward = num_frames_backward;
-
- if (frames_to_blur_forward > frames_to_blur_backward)
- frames_to_blur_forward = frames_to_blur_backward;
-
- if (frames_to_blur_backward > frames_to_blur_forward)
- frames_to_blur_backward = frames_to_blur_forward;
-
- // When max_frames is even we have 1 more frame backward than forward
- if (frames_to_blur_forward > (max_frames - 1) / 2)
- frames_to_blur_forward = ((max_frames - 1) / 2);
-
- if (frames_to_blur_backward > (max_frames / 2))
- frames_to_blur_backward = (max_frames / 2);
-
- frames_to_blur = frames_to_blur_backward + frames_to_blur_forward + 1;
- break;
- }
-
- start_frame = (cpi->last_alt_ref_sei
- + frames_to_blur_forward) % cpi->oxcf.lag_in_frames;
-
-#ifdef DEBUGFWG
- // DEBUG FWG
- printf("max:%d FBCK:%d FFWD:%d ftb:%d ftbbck:%d ftbfwd:%d sei:%d lasei:%d start:%d"
- , max_frames
- , num_frames_backward
- , num_frames_forward
- , frames_to_blur
- , frames_to_blur_backward
- , frames_to_blur_forward
- , cpi->source_encode_index
- , cpi->last_alt_ref_sei
- , start_frame);
-#endif
-
- // Setup frame pointers, NULL indicates frame not included in filter
- vpx_memset(cpi->frames, 0, max_frames*sizeof(YV12_BUFFER_CONFIG *));
- for (frame = 0; frame < frames_to_blur; frame++)
- {
- int which_buffer = start_frame - frame;
-
- if (which_buffer < 0)
- which_buffer += cpi->oxcf.lag_in_frames;
-
- cpi->frames[frames_to_blur-1-frame]
- = &cpi->src_buffer[which_buffer].source_buffer;
- }
-
- vp8_temporal_filter_iterate_c (
- cpi,
- frames_to_blur,
- frames_to_blur_backward,
- strength );
-}
-#endif
diff --git a/vp8/encoder/temporal_filter.h b/vp8/encoder/temporal_filter.h
deleted file mode 100644
index 740037a..0000000
--- a/vp8/encoder/temporal_filter.h
+++ /dev/null
@@ -1,48 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_VP8_TEMPORAL_FILTER_H
-#define __INC_VP8_TEMPORAL_FILTER_H
-
-#define prototype_apply(sym)\
- void (sym) \
- ( \
- unsigned char *frame1, \
- unsigned int stride, \
- unsigned char *frame2, \
- unsigned int block_size, \
- int strength, \
- int filter_weight, \
- unsigned int *accumulator, \
- unsigned short *count \
- )
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/temporal_filter_x86.h"
-#endif
-
-#ifndef vp8_temporal_filter_apply
-#define vp8_temporal_filter_apply vp8_temporal_filter_apply_c
-#endif
-extern prototype_apply(vp8_temporal_filter_apply);
-
-typedef struct
-{
- prototype_apply(*apply);
-} vp8_temporal_rtcd_vtable_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define TEMPORAL_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define TEMPORAL_INVOKE(ctx,fn) vp8_temporal_filter_##fn
-#endif
-
-#endif // __INC_VP8_TEMPORAL_FILTER_H
diff --git a/vp8/encoder/tokenize.c b/vp8/encoder/tokenize.c
deleted file mode 100644
index e3f423f..0000000
--- a/vp8/encoder/tokenize.c
+++ /dev/null
@@ -1,545 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <math.h>
-#include <stdio.h>
-#include <string.h>
-#include <assert.h>
-#include "onyx_int.h"
-#include "tokenize.h"
-#include "vpx_mem/vpx_mem.h"
-
-/* Global event counters used for accumulating statistics across several
- compressions, then generating context.c = initial stats. */
-
-#ifdef ENTROPY_STATS
-_int64 context_counters[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens];
-#endif
-void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) ;
-void vp8_fix_contexts(MACROBLOCKD *x);
-
-static TOKENVALUE dct_value_tokens[DCT_MAX_VALUE*2];
-const TOKENVALUE *vp8_dct_value_tokens_ptr;
-static int dct_value_cost[DCT_MAX_VALUE*2];
-const int *vp8_dct_value_cost_ptr;
-#if 0
-int skip_true_count = 0;
-int skip_false_count = 0;
-#endif
-static void fill_value_tokens()
-{
-
- TOKENVALUE *const t = dct_value_tokens + DCT_MAX_VALUE;
- vp8_extra_bit_struct *const e = vp8_extra_bits;
-
- int i = -DCT_MAX_VALUE;
- int sign = 1;
-
- do
- {
- if (!i)
- sign = 0;
-
- {
- const int a = sign ? -i : i;
- int eb = sign;
-
- if (a > 4)
- {
- int j = 4;
-
- while (++j < 11 && e[j].base_val <= a) {}
-
- t[i].Token = --j;
- eb |= (a - e[j].base_val) << 1;
- }
- else
- t[i].Token = a;
-
- t[i].Extra = eb;
- }
-
- // initialize the cost for extra bits for all possible coefficient value.
- {
- int cost = 0;
- vp8_extra_bit_struct *p = vp8_extra_bits + t[i].Token;
-
- if (p->base_val)
- {
- const int extra = t[i].Extra;
- const int Length = p->Len;
-
- if (Length)
- cost += vp8_treed_cost(p->tree, p->prob, extra >> 1, Length);
-
- cost += vp8_cost_bit(vp8_prob_half, extra & 1); /* sign */
- dct_value_cost[i + DCT_MAX_VALUE] = cost;
- }
-
- }
-
- }
- while (++i < DCT_MAX_VALUE);
-
- vp8_dct_value_tokens_ptr = dct_value_tokens + DCT_MAX_VALUE;
- vp8_dct_value_cost_ptr = dct_value_cost + DCT_MAX_VALUE;
-}
-
-static void tokenize2nd_order_b
-(
- const BLOCKD *const b,
- TOKENEXTRA **tp,
- const int type, /* which plane: 0=Y no DC, 1=Y2, 2=UV, 3=Y with DC */
- const FRAME_TYPE frametype,
- ENTROPY_CONTEXT *a,
- ENTROPY_CONTEXT *l,
- VP8_COMP *cpi
-)
-{
- int pt; /* near block/prev token context index */
- int c = 0; /* start at DC */
- const int eob = b->eob; /* one beyond last nonzero coeff */
- TOKENEXTRA *t = *tp; /* store tokens starting here */
- int x;
- const short *qcoeff_ptr = b->qcoeff;
- VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
-
- do
- {
- const int band = vp8_coef_bands[c];
-
- if (c < eob)
- {
- int rc = vp8_default_zig_zag1d[c];
- const int v = qcoeff_ptr[rc];
-
- assert(-DCT_MAX_VALUE <= v && v < (DCT_MAX_VALUE));
-
- t->Extra = vp8_dct_value_tokens_ptr[v].Extra;
- x = vp8_dct_value_tokens_ptr[v].Token;
- }
- else
- x = DCT_EOB_TOKEN;
-
- t->Token = x;
- t->context_tree = cpi->common.fc.coef_probs [type] [band] [pt];
-
- t->skip_eob_node = pt == 0 && ((band > 0 && type > 0) || (band > 1 && type == 0));
-
- ++cpi->coef_counts [type] [band] [pt] [x];
- }
- while (pt = vp8_prev_token_class[x], ++t, c < eob && ++c < 16);
-
- *tp = t;
- pt = (c != !type); /* 0 <-> all coeff data is zero */
- *a = *l = pt;
-
-}
-
-static void tokenize1st_order_b
-(
- const BLOCKD *const b,
- TOKENEXTRA **tp,
- const int type, /* which plane: 0=Y no DC, 1=Y2, 2=UV, 3=Y with DC */
- const FRAME_TYPE frametype,
- ENTROPY_CONTEXT *a,
- ENTROPY_CONTEXT *l,
- VP8_COMP *cpi
-)
-{
- int pt; /* near block/prev token context index */
- int c = type ? 0 : 1; /* start at DC unless type 0 */
- const int eob = b->eob; /* one beyond last nonzero coeff */
- TOKENEXTRA *t = *tp; /* store tokens starting here */
- int x;
- const short *qcoeff_ptr = b->qcoeff;
- VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
-
- do
- {
- const int band = vp8_coef_bands[c];
-
- x = DCT_EOB_TOKEN;
-
- if (c < eob)
- {
- int rc = vp8_default_zig_zag1d[c];
- const int v = qcoeff_ptr[rc];
-
- assert(-DCT_MAX_VALUE <= v && v < (DCT_MAX_VALUE));
-
- t->Extra = vp8_dct_value_tokens_ptr[v].Extra;
- x = vp8_dct_value_tokens_ptr[v].Token;
- }
-
- t->Token = x;
- t->context_tree = cpi->common.fc.coef_probs [type] [band] [pt];
-
- t->skip_eob_node = pt == 0 && ((band > 0 && type > 0) || (band > 1 && type == 0));
-
- ++cpi->coef_counts [type] [band] [pt] [x];
- }
- while (pt = vp8_prev_token_class[x], ++t, c < eob && ++c < 16);
-
- *tp = t;
- pt = (c != !type); /* 0 <-> all coeff data is zero */
- *a = *l = pt;
-
-}
-
-
-static int mb_is_skippable(MACROBLOCKD *x)
-{
- int has_y2_block;
- int skip = 1;
- int i = 0;
-
- has_y2_block = (x->mode_info_context->mbmi.mode != B_PRED
- && x->mode_info_context->mbmi.mode != SPLITMV);
- if (has_y2_block)
- {
- for (i = 0; i < 16; i++)
- skip &= (x->block[i].eob < 2);
- }
-
- for (; i < 24 + has_y2_block; i++)
- skip &= (!x->block[i].eob);
-
- return skip;
-}
-
-
-void vp8_tokenize_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t)
-{
- ENTROPY_CONTEXT * A = (ENTROPY_CONTEXT *)x->above_context;
- ENTROPY_CONTEXT * L = (ENTROPY_CONTEXT *)x->left_context;
- int plane_type;
- int b;
-
- TOKENEXTRA *start = *t;
- TOKENEXTRA *tp = *t;
-
- x->mode_info_context->mbmi.dc_diff = 1;
-
-#if 0
-
- if (x->mbmi.force_no_skip)
- {
- x->mbmi.mb_skip_coeff = 1;
- //reset for next_mb.
- x->mbmi.force_no_skip = 0;
- }
-
-#endif
-
-#if 1
-
- x->mode_info_context->mbmi.mb_skip_coeff = mb_is_skippable(x);
- if (x->mode_info_context->mbmi.mb_skip_coeff)
- {
-
- cpi->skip_true_count++;
-
- if (!cpi->common.mb_no_coeff_skip)
- vp8_stuff_mb(cpi, x, t) ;
- else
- {
- vp8_fix_contexts(x);
- }
-
- if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV)
- x->mode_info_context->mbmi.dc_diff = 0;
- else
- x->mode_info_context->mbmi.dc_diff = 1;
-
-
- return;
- }
-
- cpi->skip_false_count++;
-#endif
-#if 0
- vpx_memcpy(cpi->coef_counts_backup, cpi->coef_counts, sizeof(cpi->coef_counts));
-#endif
-
- if (x->mode_info_context->mbmi.mode == B_PRED || x->mode_info_context->mbmi.mode == SPLITMV)
- {
- plane_type = 3;
- }
- else
- {
- tokenize2nd_order_b(x->block + 24, t, 1, x->frame_type,
- A + vp8_block2above[24], L + vp8_block2left[24], cpi);
- plane_type = 0;
-
- }
-
- for (b = 0; b < 16; b++)
- tokenize1st_order_b(x->block + b, t, plane_type, x->frame_type,
- A + vp8_block2above[b],
- L + vp8_block2left[b], cpi);
-
- for (b = 16; b < 24; b++)
- tokenize1st_order_b(x->block + b, t, 2, x->frame_type,
- A + vp8_block2above[b],
- L + vp8_block2left[b], cpi);
-
-#if 0
-
- if (cpi->common.mb_no_coeff_skip)
- {
- int skip = 1;
-
- while ((tp != *t) && skip)
- {
- skip = (skip && (tp->Token == DCT_EOB_TOKEN));
- tp ++;
- }
-
- if (skip != x->mbmi.mb_skip_coeff)
- skip += 0;
-
- x->mbmi.mb_skip_coeff = skip;
-
- if (x->mbmi.mb_skip_coeff == 1)
- {
- x->mbmi.dc_diff = 0;
- //redo the coutnts
- vpx_memcpy(cpi->coef_counts, cpi->coef_counts_backup, sizeof(cpi->coef_counts));
-
- *t = start;
- cpi->skip_true_count++;
- //skip_true_count++;
- }
- else
- {
-
- cpi->skip_false_count++;
- //skip_false_count++;
- }
- }
-
-#endif
-}
-
-
-#ifdef ENTROPY_STATS
-
-void init_context_counters(void)
-{
- vpx_memset(context_counters, 0, sizeof(context_counters));
-}
-
-void print_context_counters()
-{
-
- int type, band, pt, t;
-
- FILE *const f = fopen("context.c", "w");
-
- fprintf(f, "#include \"entropy.h\"\n");
-
- fprintf(f, "\n/* *** GENERATED FILE: DO NOT EDIT *** */\n\n");
-
- fprintf(f, "int Contexts[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens];\n\n");
-
- fprintf(f, "const int default_contexts[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens] = {");
-
-# define Comma( X) (X? ",":"")
-
- type = 0;
-
- do
- {
- fprintf(f, "%s\n { /* block Type %d */", Comma(type), type);
-
- band = 0;
-
- do
- {
- fprintf(f, "%s\n { /* Coeff Band %d */", Comma(band), band);
-
- pt = 0;
-
- do
- {
- fprintf(f, "%s\n {", Comma(pt));
-
- t = 0;
-
- do
- {
- const _int64 x = context_counters [type] [band] [pt] [t];
- const int y = (int) x;
-
- assert(x == (_int64) y); /* no overflow handling yet */
- fprintf(f, "%s %d", Comma(t), y);
-
- }
- while (++t < vp8_coef_tokens);
-
- fprintf(f, "}");
- }
- while (++pt < PREV_COEF_CONTEXTS);
-
- fprintf(f, "\n }");
-
- }
- while (++band < COEF_BANDS);
-
- fprintf(f, "\n }");
- }
- while (++type < BLOCK_TYPES);
-
- fprintf(f, "\n};\n");
- fclose(f);
-}
-#endif
-
-
-void vp8_tokenize_initialize()
-{
- fill_value_tokens();
-}
-
-
-static __inline void stuff2nd_order_b
-(
- const BLOCKD *const b,
- TOKENEXTRA **tp,
- const int type, /* which plane: 0=Y no DC, 1=Y2, 2=UV, 3=Y with DC */
- const FRAME_TYPE frametype,
- ENTROPY_CONTEXT *a,
- ENTROPY_CONTEXT *l,
- VP8_COMP *cpi
-)
-{
- int pt; /* near block/prev token context index */
- TOKENEXTRA *t = *tp; /* store tokens starting here */
- VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
- (void) frametype;
- (void) type;
- (void) b;
-
- t->Token = DCT_EOB_TOKEN;
- t->context_tree = cpi->common.fc.coef_probs [1] [0] [pt];
- t->skip_eob_node = 0;
- ++cpi->coef_counts [1] [0] [pt] [DCT_EOB_TOKEN];
- ++t;
-
- *tp = t;
- pt = 0;
- *a = *l = pt;
-
-}
-
-static __inline void stuff1st_order_b
-(
- const BLOCKD *const b,
- TOKENEXTRA **tp,
- const int type, /* which plane: 0=Y no DC, 1=Y2, 2=UV, 3=Y with DC */
- const FRAME_TYPE frametype,
- ENTROPY_CONTEXT *a,
- ENTROPY_CONTEXT *l,
- VP8_COMP *cpi
-)
-{
- int pt; /* near block/prev token context index */
- TOKENEXTRA *t = *tp; /* store tokens starting here */
- VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
- (void) frametype;
- (void) type;
- (void) b;
-
- t->Token = DCT_EOB_TOKEN;
- t->context_tree = cpi->common.fc.coef_probs [0] [1] [pt];
- t->skip_eob_node = 0;
- ++cpi->coef_counts [0] [1] [pt] [DCT_EOB_TOKEN];
- ++t;
- *tp = t;
- pt = 0; /* 0 <-> all coeff data is zero */
- *a = *l = pt;
-
-}
-static __inline
-void stuff1st_order_buv
-(
- const BLOCKD *const b,
- TOKENEXTRA **tp,
- const int type, /* which plane: 0=Y no DC, 1=Y2, 2=UV, 3=Y with DC */
- const FRAME_TYPE frametype,
- ENTROPY_CONTEXT *a,
- ENTROPY_CONTEXT *l,
- VP8_COMP *cpi
-)
-{
- int pt; /* near block/prev token context index */
- TOKENEXTRA *t = *tp; /* store tokens starting here */
- VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
- (void) frametype;
- (void) type;
- (void) b;
-
- t->Token = DCT_EOB_TOKEN;
- t->context_tree = cpi->common.fc.coef_probs [2] [0] [pt];
- t->skip_eob_node = 0;
- ++cpi->coef_counts[2] [0] [pt] [DCT_EOB_TOKEN];
- ++t;
- *tp = t;
- pt = 0; /* 0 <-> all coeff data is zero */
- *a = *l = pt;
-
-}
-
-void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t)
-{
- ENTROPY_CONTEXT * A = (ENTROPY_CONTEXT *)x->above_context;
- ENTROPY_CONTEXT * L = (ENTROPY_CONTEXT *)x->left_context;
- int plane_type;
- int b;
-
- stuff2nd_order_b(x->block + 24, t, 1, x->frame_type,
- A + vp8_block2above[24], L + vp8_block2left[24], cpi);
- plane_type = 0;
-
-
- if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV)
- x->mode_info_context->mbmi.dc_diff = 0;
- else
- x->mode_info_context->mbmi.dc_diff = 1;
-
-
- for (b = 0; b < 16; b++)
- stuff1st_order_b(x->block + b, t, plane_type, x->frame_type,
- A + vp8_block2above[b],
- L + vp8_block2left[b], cpi);
-
- for (b = 16; b < 24; b++)
- stuff1st_order_buv(x->block + b, t, 2, x->frame_type,
- A + vp8_block2above[b],
- L + vp8_block2left[b], cpi);
-
-}
-void vp8_fix_contexts(MACROBLOCKD *x)
-{
- /* Clear entropy contexts for Y2 blocks */
- if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV)
- {
- vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES));
- }
- else
- {
- vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1);
- vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1);
- }
-
-}
diff --git a/vp8/encoder/tokenize.h b/vp8/encoder/tokenize.h
deleted file mode 100644
index d87c1a3..0000000
--- a/vp8/encoder/tokenize.h
+++ /dev/null
@@ -1,50 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef tokenize_h
-#define tokenize_h
-
-#include "vp8/common/entropy.h"
-#include "block.h"
-
-void vp8_tokenize_initialize();
-
-typedef struct
-{
- short Token;
- short Extra;
-} TOKENVALUE;
-
-typedef struct
-{
- const vp8_prob *context_tree;
- short Extra;
- unsigned char Token;
- unsigned char skip_eob_node;
-} TOKENEXTRA;
-
-int rd_cost_mby(MACROBLOCKD *);
-
-#ifdef ENTROPY_STATS
-void init_context_counters();
-void print_context_counters();
-
-extern _int64 context_counters[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens];
-#endif
-
-extern const int *vp8_dct_value_cost_ptr;
-/* TODO: The Token field should be broken out into a separate char array to
- * improve cache locality, since it's needed for costing when the rest of the
- * fields are not.
- */
-extern const TOKENVALUE *vp8_dct_value_tokens_ptr;
-
-#endif /* tokenize_h */
diff --git a/vp8/encoder/treewriter.c b/vp8/encoder/treewriter.c
deleted file mode 100644
index 03967c8..0000000
--- a/vp8/encoder/treewriter.c
+++ /dev/null
@@ -1,39 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "treewriter.h"
-
-static void cost(
- int *const C,
- vp8_tree T,
- const vp8_prob *const P,
- int i,
- int c
-)
-{
- const vp8_prob p = P [i>>1];
-
- do
- {
- const vp8_tree_index j = T[i];
- const int d = c + vp8_cost_bit(p, i & 1);
-
- if (j <= 0)
- C[-j] = d;
- else
- cost(C, T, P, j, d);
- }
- while (++i & 1);
-}
-void vp8_cost_tokens(int *c, const vp8_prob *p, vp8_tree t)
-{
- cost(c, t, p, 0, 0);
-}
diff --git a/vp8/encoder/treewriter.h b/vp8/encoder/treewriter.h
deleted file mode 100644
index c28a0fa..0000000
--- a/vp8/encoder/treewriter.h
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_TREEWRITER_H
-#define __INC_TREEWRITER_H
-
-/* Trees map alphabets into huffman-like codes suitable for an arithmetic
- bit coder. Timothy S Murphy 11 October 2004 */
-
-#include "vp8/common/treecoder.h"
-
-#include "boolhuff.h" /* for now */
-
-typedef BOOL_CODER vp8_writer;
-
-#define vp8_write vp8_encode_bool
-#define vp8_write_literal vp8_encode_value
-#define vp8_write_bit( W, V) vp8_write( W, V, vp8_prob_half)
-
-#define vp8bc_write vp8bc_write_bool
-#define vp8bc_write_literal vp8bc_write_bits
-#define vp8bc_write_bit( W, V) vp8bc_write_bits( W, V, 1)
-
-
-/* Approximate length of an encoded bool in 256ths of a bit at given prob */
-
-#define vp8_cost_zero( x) ( vp8_prob_cost[x])
-#define vp8_cost_one( x) vp8_cost_zero( vp8_complement(x))
-
-#define vp8_cost_bit( x, b) vp8_cost_zero( (b)? vp8_complement(x) : (x) )
-
-/* VP8BC version is scaled by 2^20 rather than 2^8; see bool_coder.h */
-
-
-/* Both of these return bits, not scaled bits. */
-
-static __inline unsigned int vp8_cost_branch(const unsigned int ct[2], vp8_prob p)
-{
- /* Imitate existing calculation */
-
- return ((ct[0] * vp8_cost_zero(p))
- + (ct[1] * vp8_cost_one(p))) >> 8;
-}
-
-/* Small functions to write explicit values and tokens, as well as
- estimate their lengths. */
-
-static __inline void vp8_treed_write
-(
- vp8_writer *const w,
- vp8_tree t,
- const vp8_prob *const p,
- int v,
- int n /* number of bits in v, assumed nonzero */
-)
-{
- vp8_tree_index i = 0;
-
- do
- {
- const int b = (v >> --n) & 1;
- vp8_write(w, b, p[i>>1]);
- i = t[i+b];
- }
- while (n);
-}
-static __inline void vp8_write_token
-(
- vp8_writer *const w,
- vp8_tree t,
- const vp8_prob *const p,
- vp8_token *const x
-)
-{
- vp8_treed_write(w, t, p, x->value, x->Len);
-}
-
-static __inline int vp8_treed_cost(
- vp8_tree t,
- const vp8_prob *const p,
- int v,
- int n /* number of bits in v, assumed nonzero */
-)
-{
- int c = 0;
- vp8_tree_index i = 0;
-
- do
- {
- const int b = (v >> --n) & 1;
- c += vp8_cost_bit(p[i>>1], b);
- i = t[i+b];
- }
- while (n);
-
- return c;
-}
-static __inline int vp8_cost_token
-(
- vp8_tree t,
- const vp8_prob *const p,
- vp8_token *const x
-)
-{
- return vp8_treed_cost(t, p, x->value, x->Len);
-}
-
-/* Fill array of costs for all possible token values. */
-
-void vp8_cost_tokens(
- int *Costs, const vp8_prob *, vp8_tree
-);
-
-#endif
diff --git a/vp8/encoder/variance.h b/vp8/encoder/variance.h
deleted file mode 100644
index bf17ea8..0000000
--- a/vp8/encoder/variance.h
+++ /dev/null
@@ -1,421 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VARIANCE_H
-#define VARIANCE_H
-
-#define prototype_sad(sym)\
- unsigned int (sym)\
- (\
- const unsigned char *src_ptr, \
- int source_stride, \
- const unsigned char *ref_ptr, \
- int ref_stride, \
- int max_sad\
- )
-
-#define prototype_sad_multi_same_address(sym)\
- void (sym)\
- (\
- const unsigned char *src_ptr, \
- int source_stride, \
- const unsigned char *ref_ptr, \
- int ref_stride, \
- unsigned int *sad_array\
- )
-
-#define prototype_sad_multi_same_address_1(sym)\
- void (sym)\
- (\
- const unsigned char *src_ptr, \
- int source_stride, \
- const unsigned char *ref_ptr, \
- int ref_stride, \
- unsigned short *sad_array\
- )
-
-#define prototype_sad_multi_dif_address(sym)\
- void (sym)\
- (\
- const unsigned char *src_ptr, \
- int source_stride, \
- unsigned char *ref_ptr[4], \
- int ref_stride, \
- unsigned int *sad_array\
- )
-
-#define prototype_variance(sym) \
- unsigned int (sym) \
- (\
- const unsigned char *src_ptr, \
- int source_stride, \
- const unsigned char *ref_ptr, \
- int ref_stride, \
- unsigned int *sse\
- )
-
-#define prototype_variance2(sym) \
- unsigned int (sym) \
- (\
- const unsigned char *src_ptr, \
- int source_stride, \
- const unsigned char *ref_ptr, \
- int ref_stride, \
- unsigned int *sse,\
- int *sum\
- )
-
-#define prototype_subpixvariance(sym) \
- unsigned int (sym) \
- ( \
- const unsigned char *src_ptr, \
- int source_stride, \
- int xoffset, \
- int yoffset, \
- const unsigned char *ref_ptr, \
- int Refstride, \
- unsigned int *sse \
- );
-
-#define prototype_ssimpf(sym) \
- void (sym) \
- ( \
- unsigned char *s, \
- int sp, \
- unsigned char *r, \
- int rp, \
- unsigned long *sum_s, \
- unsigned long *sum_r, \
- unsigned long *sum_sq_s, \
- unsigned long *sum_sq_r, \
- unsigned long *sum_sxr \
- );
-
-#define prototype_getmbss(sym) unsigned int (sym)(const short *)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/variance_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/variance_arm.h"
-#endif
-
-#ifndef vp8_variance_sad4x4
-#define vp8_variance_sad4x4 vp8_sad4x4_c
-#endif
-extern prototype_sad(vp8_variance_sad4x4);
-
-#ifndef vp8_variance_sad8x8
-#define vp8_variance_sad8x8 vp8_sad8x8_c
-#endif
-extern prototype_sad(vp8_variance_sad8x8);
-
-#ifndef vp8_variance_sad8x16
-#define vp8_variance_sad8x16 vp8_sad8x16_c
-#endif
-extern prototype_sad(vp8_variance_sad8x16);
-
-#ifndef vp8_variance_sad16x8
-#define vp8_variance_sad16x8 vp8_sad16x8_c
-#endif
-extern prototype_sad(vp8_variance_sad16x8);
-
-#ifndef vp8_variance_sad16x16
-#define vp8_variance_sad16x16 vp8_sad16x16_c
-#endif
-extern prototype_sad(vp8_variance_sad16x16);
-
-//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-
-#ifndef vp8_variance_sad16x16x3
-#define vp8_variance_sad16x16x3 vp8_sad16x16x3_c
-#endif
-extern prototype_sad_multi_same_address(vp8_variance_sad16x16x3);
-
-#ifndef vp8_variance_sad16x8x3
-#define vp8_variance_sad16x8x3 vp8_sad16x8x3_c
-#endif
-extern prototype_sad_multi_same_address(vp8_variance_sad16x8x3);
-
-#ifndef vp8_variance_sad8x8x3
-#define vp8_variance_sad8x8x3 vp8_sad8x8x3_c
-#endif
-extern prototype_sad_multi_same_address(vp8_variance_sad8x8x3);
-
-#ifndef vp8_variance_sad8x16x3
-#define vp8_variance_sad8x16x3 vp8_sad8x16x3_c
-#endif
-extern prototype_sad_multi_same_address(vp8_variance_sad8x16x3);
-
-#ifndef vp8_variance_sad4x4x3
-#define vp8_variance_sad4x4x3 vp8_sad4x4x3_c
-#endif
-extern prototype_sad_multi_same_address(vp8_variance_sad4x4x3);
-
-#ifndef vp8_variance_sad16x16x8
-#define vp8_variance_sad16x16x8 vp8_sad16x16x8_c
-#endif
-extern prototype_sad_multi_same_address_1(vp8_variance_sad16x16x8);
-
-#ifndef vp8_variance_sad16x8x8
-#define vp8_variance_sad16x8x8 vp8_sad16x8x8_c
-#endif
-extern prototype_sad_multi_same_address_1(vp8_variance_sad16x8x8);
-
-#ifndef vp8_variance_sad8x8x8
-#define vp8_variance_sad8x8x8 vp8_sad8x8x8_c
-#endif
-extern prototype_sad_multi_same_address_1(vp8_variance_sad8x8x8);
-
-#ifndef vp8_variance_sad8x16x8
-#define vp8_variance_sad8x16x8 vp8_sad8x16x8_c
-#endif
-extern prototype_sad_multi_same_address_1(vp8_variance_sad8x16x8);
-
-#ifndef vp8_variance_sad4x4x8
-#define vp8_variance_sad4x4x8 vp8_sad4x4x8_c
-#endif
-extern prototype_sad_multi_same_address_1(vp8_variance_sad4x4x8);
-
-//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-
-#ifndef vp8_variance_sad16x16x4d
-#define vp8_variance_sad16x16x4d vp8_sad16x16x4d_c
-#endif
-extern prototype_sad_multi_dif_address(vp8_variance_sad16x16x4d);
-
-#ifndef vp8_variance_sad16x8x4d
-#define vp8_variance_sad16x8x4d vp8_sad16x8x4d_c
-#endif
-extern prototype_sad_multi_dif_address(vp8_variance_sad16x8x4d);
-
-#ifndef vp8_variance_sad8x8x4d
-#define vp8_variance_sad8x8x4d vp8_sad8x8x4d_c
-#endif
-extern prototype_sad_multi_dif_address(vp8_variance_sad8x8x4d);
-
-#ifndef vp8_variance_sad8x16x4d
-#define vp8_variance_sad8x16x4d vp8_sad8x16x4d_c
-#endif
-extern prototype_sad_multi_dif_address(vp8_variance_sad8x16x4d);
-
-#ifndef vp8_variance_sad4x4x4d
-#define vp8_variance_sad4x4x4d vp8_sad4x4x4d_c
-#endif
-extern prototype_sad_multi_dif_address(vp8_variance_sad4x4x4d);
-
-//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-
-#ifndef vp8_variance_var4x4
-#define vp8_variance_var4x4 vp8_variance4x4_c
-#endif
-extern prototype_variance(vp8_variance_var4x4);
-
-#ifndef vp8_variance_var8x8
-#define vp8_variance_var8x8 vp8_variance8x8_c
-#endif
-extern prototype_variance(vp8_variance_var8x8);
-
-#ifndef vp8_variance_var8x16
-#define vp8_variance_var8x16 vp8_variance8x16_c
-#endif
-extern prototype_variance(vp8_variance_var8x16);
-
-#ifndef vp8_variance_var16x8
-#define vp8_variance_var16x8 vp8_variance16x8_c
-#endif
-extern prototype_variance(vp8_variance_var16x8);
-
-#ifndef vp8_variance_var16x16
-#define vp8_variance_var16x16 vp8_variance16x16_c
-#endif
-extern prototype_variance(vp8_variance_var16x16);
-
-//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-
-#ifndef vp8_variance_subpixvar4x4
-#define vp8_variance_subpixvar4x4 vp8_sub_pixel_variance4x4_c
-#endif
-extern prototype_subpixvariance(vp8_variance_subpixvar4x4);
-
-#ifndef vp8_variance_subpixvar8x8
-#define vp8_variance_subpixvar8x8 vp8_sub_pixel_variance8x8_c
-#endif
-extern prototype_subpixvariance(vp8_variance_subpixvar8x8);
-
-#ifndef vp8_variance_subpixvar8x16
-#define vp8_variance_subpixvar8x16 vp8_sub_pixel_variance8x16_c
-#endif
-extern prototype_subpixvariance(vp8_variance_subpixvar8x16);
-
-#ifndef vp8_variance_subpixvar16x8
-#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_c
-#endif
-extern prototype_subpixvariance(vp8_variance_subpixvar16x8);
-
-#ifndef vp8_variance_subpixvar16x16
-#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_c
-#endif
-extern prototype_subpixvariance(vp8_variance_subpixvar16x16);
-
-#ifndef vp8_variance_halfpixvar16x16_h
-#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_c
-#endif
-extern prototype_variance(vp8_variance_halfpixvar16x16_h);
-
-#ifndef vp8_variance_halfpixvar16x16_v
-#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_c
-#endif
-extern prototype_variance(vp8_variance_halfpixvar16x16_v);
-
-#ifndef vp8_variance_halfpixvar16x16_hv
-#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_c
-#endif
-extern prototype_variance(vp8_variance_halfpixvar16x16_hv);
-
-#ifndef vp8_variance_subpixmse16x16
-#define vp8_variance_subpixmse16x16 vp8_sub_pixel_mse16x16_c
-#endif
-extern prototype_subpixvariance(vp8_variance_subpixmse16x16);
-
-//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-
-#ifndef vp8_variance_getmbss
-#define vp8_variance_getmbss vp8_get_mb_ss_c
-#endif
-extern prototype_getmbss(vp8_variance_getmbss);
-
-#ifndef vp8_variance_mse16x16
-#define vp8_variance_mse16x16 vp8_mse16x16_c
-#endif
-extern prototype_variance(vp8_variance_mse16x16);
-
-#ifndef vp8_variance_get16x16prederror
-#define vp8_variance_get16x16prederror vp8_get16x16pred_error_c
-#endif
-extern prototype_sad(vp8_variance_get16x16prederror);
-
-#ifndef vp8_variance_get8x8var
-#define vp8_variance_get8x8var vp8_get8x8var_c
-#endif
-extern prototype_variance2(vp8_variance_get8x8var);
-
-#ifndef vp8_variance_get16x16var
-#define vp8_variance_get16x16var vp8_get16x16var_c
-#endif
-extern prototype_variance2(vp8_variance_get16x16var);
-
-#ifndef vp8_variance_get4x4sse_cs
-#define vp8_variance_get4x4sse_cs vp8_get4x4sse_cs_c
-#endif
-extern prototype_sad(vp8_variance_get4x4sse_cs);
-
-#ifndef vp8_ssimpf
-#define vp8_ssimpf ssim_parms_c
-#endif
-extern prototype_ssimpf(vp8_ssimpf)
-
-#ifndef vp8_ssimpf_8x8
-#define vp8_ssimpf_8x8 ssim_parms_8x8_c
-#endif
-extern prototype_ssimpf(vp8_ssimpf_8x8)
-
-typedef prototype_sad(*vp8_sad_fn_t);
-typedef prototype_sad_multi_same_address(*vp8_sad_multi_fn_t);
-typedef prototype_sad_multi_same_address_1(*vp8_sad_multi1_fn_t);
-typedef prototype_sad_multi_dif_address(*vp8_sad_multi_d_fn_t);
-typedef prototype_variance(*vp8_variance_fn_t);
-typedef prototype_variance2(*vp8_variance2_fn_t);
-typedef prototype_subpixvariance(*vp8_subpixvariance_fn_t);
-typedef prototype_getmbss(*vp8_getmbss_fn_t);
-
-typedef prototype_ssimpf(*vp8_ssimpf_fn_t)
-
-
-typedef struct
-{
- vp8_sad_fn_t sad4x4;
- vp8_sad_fn_t sad8x8;
- vp8_sad_fn_t sad8x16;
- vp8_sad_fn_t sad16x8;
- vp8_sad_fn_t sad16x16;
-
- vp8_variance_fn_t var4x4;
- vp8_variance_fn_t var8x8;
- vp8_variance_fn_t var8x16;
- vp8_variance_fn_t var16x8;
- vp8_variance_fn_t var16x16;
-
- vp8_subpixvariance_fn_t subpixvar4x4;
- vp8_subpixvariance_fn_t subpixvar8x8;
- vp8_subpixvariance_fn_t subpixvar8x16;
- vp8_subpixvariance_fn_t subpixvar16x8;
- vp8_subpixvariance_fn_t subpixvar16x16;
- vp8_variance_fn_t halfpixvar16x16_h;
- vp8_variance_fn_t halfpixvar16x16_v;
- vp8_variance_fn_t halfpixvar16x16_hv;
- vp8_subpixvariance_fn_t subpixmse16x16;
-
- vp8_getmbss_fn_t getmbss;
- vp8_variance_fn_t mse16x16;
-
- vp8_sad_fn_t get16x16prederror;
- vp8_variance2_fn_t get8x8var;
- vp8_variance2_fn_t get16x16var;
- vp8_sad_fn_t get4x4sse_cs;
-
- vp8_sad_multi_fn_t sad16x16x3;
- vp8_sad_multi_fn_t sad16x8x3;
- vp8_sad_multi_fn_t sad8x16x3;
- vp8_sad_multi_fn_t sad8x8x3;
- vp8_sad_multi_fn_t sad4x4x3;
-
- vp8_sad_multi1_fn_t sad16x16x8;
- vp8_sad_multi1_fn_t sad16x8x8;
- vp8_sad_multi1_fn_t sad8x16x8;
- vp8_sad_multi1_fn_t sad8x8x8;
- vp8_sad_multi1_fn_t sad4x4x8;
-
- vp8_sad_multi_d_fn_t sad16x16x4d;
- vp8_sad_multi_d_fn_t sad16x8x4d;
- vp8_sad_multi_d_fn_t sad8x16x4d;
- vp8_sad_multi_d_fn_t sad8x8x4d;
- vp8_sad_multi_d_fn_t sad4x4x4d;
-
-#if CONFIG_PSNR
- vp8_ssimpf_fn_t ssimpf_8x8;
- vp8_ssimpf_fn_t ssimpf;
-#endif
-
-} vp8_variance_rtcd_vtable_t;
-
-typedef struct
-{
- vp8_sad_fn_t sdf;
- vp8_variance_fn_t vf;
- vp8_subpixvariance_fn_t svf;
- vp8_variance_fn_t svf_halfpix_h;
- vp8_variance_fn_t svf_halfpix_v;
- vp8_variance_fn_t svf_halfpix_hv;
- vp8_sad_multi_fn_t sdx3f;
- vp8_sad_multi1_fn_t sdx8f;
- vp8_sad_multi_d_fn_t sdx4df;
-
-} vp8_variance_fn_ptr_t;
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define VARIANCE_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define VARIANCE_INVOKE(ctx,fn) vp8_variance_##fn
-#endif
-
-#endif
diff --git a/vp8/encoder/variance_c.c b/vp8/encoder/variance_c.c
deleted file mode 100644
index ede07c8..0000000
--- a/vp8/encoder/variance_c.c
+++ /dev/null
@@ -1,492 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "variance.h"
-#include "vp8/common/filter.h"
-
-
-unsigned int vp8_get_mb_ss_c
-(
- const short *src_ptr
-)
-{
- unsigned int i = 0, sum = 0;
-
- do
- {
- sum += (src_ptr[i] * src_ptr[i]);
- i++;
- }
- while (i < 256);
-
- return sum;
-}
-
-
-static void variance(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- int w,
- int h,
- unsigned int *sse,
- int *sum)
-{
- int i, j;
- int diff;
-
- *sum = 0;
- *sse = 0;
-
- for (i = 0; i < h; i++)
- {
- for (j = 0; j < w; j++)
- {
- diff = src_ptr[j] - ref_ptr[j];
- *sum += diff;
- *sse += diff * diff;
- }
-
- src_ptr += source_stride;
- ref_ptr += recon_stride;
- }
-}
-
-unsigned int
-vp8_get8x8var_c
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *SSE,
- int *Sum
-)
-{
-
- variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 8, SSE, Sum);
- return (*SSE - (((*Sum) * (*Sum)) >> 6));
-}
-
-unsigned int
-vp8_get16x16var_c
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *SSE,
- int *Sum
-)
-{
-
- variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, SSE, Sum);
- return (*SSE - (((*Sum) * (*Sum)) >> 8));
-
-}
-
-
-
-unsigned int vp8_variance16x16_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int var;
- int avg;
-
-
- variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, &var, &avg);
- *sse = var;
- return (var - ((avg * avg) >> 8));
-}
-
-unsigned int vp8_variance8x16_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int var;
- int avg;
-
-
- variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 16, &var, &avg);
- *sse = var;
- return (var - ((avg * avg) >> 7));
-}
-
-unsigned int vp8_variance16x8_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int var;
- int avg;
-
-
- variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 8, &var, &avg);
- *sse = var;
- return (var - ((avg * avg) >> 7));
-}
-
-
-unsigned int vp8_variance8x8_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int var;
- int avg;
-
-
- variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 8, &var, &avg);
- *sse = var;
- return (var - ((avg * avg) >> 6));
-}
-
-unsigned int vp8_variance4x4_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int var;
- int avg;
-
-
- variance(src_ptr, source_stride, ref_ptr, recon_stride, 4, 4, &var, &avg);
- *sse = var;
- return (var - ((avg * avg) >> 4));
-}
-
-
-unsigned int vp8_mse16x16_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int var;
- int avg;
-
- variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, &var, &avg);
- *sse = var;
- return var;
-}
-
-
-/****************************************************************************
- *
- * ROUTINE : filter_block2d_bil_first_pass
- *
- * INPUTS : UINT8 *src_ptr : Pointer to source block.
- * UINT32 src_pixels_per_line : Stride of input block.
- * UINT32 pixel_step : Offset between filter input samples (see notes).
- * UINT32 output_height : Input block height.
- * UINT32 output_width : Input block width.
- * INT32 *vp8_filter : Array of 2 bi-linear filter taps.
- *
- * OUTPUTS : INT32 *output_ptr : Pointer to filtered block.
- *
- * RETURNS : void
- *
- * FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block in
- * either horizontal or vertical direction to produce the
- * filtered output block. Used to implement first-pass
- * of 2-D separable filter.
- *
- * SPECIAL NOTES : Produces INT32 output to retain precision for next pass.
- * Two filter taps should sum to VP8_FILTER_WEIGHT.
- * pixel_step defines whether the filter is applied
- * horizontally (pixel_step=1) or vertically (pixel_step=stride).
- * It defines the offset required to move from one input
- * to the next.
- *
- ****************************************************************************/
-static void var_filter_block2d_bil_first_pass
-(
- const unsigned char *src_ptr,
- unsigned short *output_ptr,
- unsigned int src_pixels_per_line,
- int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-)
-{
- unsigned int i, j;
-
- for (i = 0; i < output_height; i++)
- {
- for (j = 0; j < output_width; j++)
- {
- // Apply bilinear filter
- output_ptr[j] = (((int)src_ptr[0] * vp8_filter[0]) +
- ((int)src_ptr[pixel_step] * vp8_filter[1]) +
- (VP8_FILTER_WEIGHT / 2)) >> VP8_FILTER_SHIFT;
- src_ptr++;
- }
-
- // Next row...
- src_ptr += src_pixels_per_line - output_width;
- output_ptr += output_width;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : filter_block2d_bil_second_pass
- *
- * INPUTS : INT32 *src_ptr : Pointer to source block.
- * UINT32 src_pixels_per_line : Stride of input block.
- * UINT32 pixel_step : Offset between filter input samples (see notes).
- * UINT32 output_height : Input block height.
- * UINT32 output_width : Input block width.
- * INT32 *vp8_filter : Array of 2 bi-linear filter taps.
- *
- * OUTPUTS : UINT16 *output_ptr : Pointer to filtered block.
- *
- * RETURNS : void
- *
- * FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block in
- * either horizontal or vertical direction to produce the
- * filtered output block. Used to implement second-pass
- * of 2-D separable filter.
- *
- * SPECIAL NOTES : Requires 32-bit input as produced by filter_block2d_bil_first_pass.
- * Two filter taps should sum to VP8_FILTER_WEIGHT.
- * pixel_step defines whether the filter is applied
- * horizontally (pixel_step=1) or vertically (pixel_step=stride).
- * It defines the offset required to move from one input
- * to the next.
- *
- ****************************************************************************/
-static void var_filter_block2d_bil_second_pass
-(
- const unsigned short *src_ptr,
- unsigned char *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- const short *vp8_filter
-)
-{
- unsigned int i, j;
- int Temp;
-
- for (i = 0; i < output_height; i++)
- {
- for (j = 0; j < output_width; j++)
- {
- // Apply filter
- Temp = ((int)src_ptr[0] * vp8_filter[0]) +
- ((int)src_ptr[pixel_step] * vp8_filter[1]) +
- (VP8_FILTER_WEIGHT / 2);
- output_ptr[j] = (unsigned int)(Temp >> VP8_FILTER_SHIFT);
- src_ptr++;
- }
-
- // Next row...
- src_ptr += src_pixels_per_line - output_width;
- output_ptr += output_width;
- }
-}
-
-
-unsigned int vp8_sub_pixel_variance4x4_c
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- unsigned char temp2[20*16];
- const short *HFilter, *VFilter;
- unsigned short FData3[5*4]; // Temp data bufffer used in filtering
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- // First filter 1d Horizontal
- var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 5, 4, HFilter);
-
- // Now filter Verticaly
- var_filter_block2d_bil_second_pass(FData3, temp2, 4, 4, 4, 4, VFilter);
-
- return vp8_variance4x4_c(temp2, 4, dst_ptr, dst_pixels_per_line, sse);
-}
-
-
-unsigned int vp8_sub_pixel_variance8x8_c
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- unsigned short FData3[9*8]; // Temp data bufffer used in filtering
- unsigned char temp2[20*16];
- const short *HFilter, *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 9, 8, HFilter);
- var_filter_block2d_bil_second_pass(FData3, temp2, 8, 8, 8, 8, VFilter);
-
- return vp8_variance8x8_c(temp2, 8, dst_ptr, dst_pixels_per_line, sse);
-}
-
-unsigned int vp8_sub_pixel_variance16x16_c
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- unsigned short FData3[17*16]; // Temp data bufffer used in filtering
- unsigned char temp2[20*16];
- const short *HFilter, *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 17, 16, HFilter);
- var_filter_block2d_bil_second_pass(FData3, temp2, 16, 16, 16, 16, VFilter);
-
- return vp8_variance16x16_c(temp2, 16, dst_ptr, dst_pixels_per_line, sse);
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_h_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 4, 0,
- ref_ptr, recon_stride, sse);
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_v_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 0, 4,
- ref_ptr, recon_stride, sse);
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_hv_c(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 4, 4,
- ref_ptr, recon_stride, sse);
-}
-
-
-unsigned int vp8_sub_pixel_mse16x16_c
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- vp8_sub_pixel_variance16x16_c(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
- return *sse;
-}
-
-unsigned int vp8_sub_pixel_variance16x8_c
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- unsigned short FData3[16*9]; // Temp data bufffer used in filtering
- unsigned char temp2[20*16];
- const short *HFilter, *VFilter;
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
- var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 9, 16, HFilter);
- var_filter_block2d_bil_second_pass(FData3, temp2, 16, 16, 8, 16, VFilter);
-
- return vp8_variance16x8_c(temp2, 16, dst_ptr, dst_pixels_per_line, sse);
-}
-
-unsigned int vp8_sub_pixel_variance8x16_c
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- unsigned short FData3[9*16]; // Temp data bufffer used in filtering
- unsigned char temp2[20*16];
- const short *HFilter, *VFilter;
-
-
- HFilter = vp8_bilinear_filters[xoffset];
- VFilter = vp8_bilinear_filters[yoffset];
-
-
- var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 17, 8, HFilter);
- var_filter_block2d_bil_second_pass(FData3, temp2, 8, 8, 16, 8, VFilter);
-
- return vp8_variance8x16_c(temp2, 8, dst_ptr, dst_pixels_per_line, sse);
-}
diff --git a/vp8/encoder/x86/dct_mmx.asm b/vp8/encoder/x86/dct_mmx.asm
deleted file mode 100644
index f07b030..0000000
--- a/vp8/encoder/x86/dct_mmx.asm
+++ /dev/null
@@ -1,241 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_short_fdct4x4_mmx(short *input, short *output, int pitch)
-global sym(vp8_short_fdct4x4_mmx)
-sym(vp8_short_fdct4x4_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 3
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ; input
- mov rdi, arg(1) ; output
-
- movsxd rax, dword ptr arg(2) ;pitch
-
- lea rcx, [rsi + rax*2]
- ; read the input data
- movq mm0, [rsi]
- movq mm1, [rsi + rax]
-
- movq mm2, [rcx]
- movq mm4, [rcx + rax]
-
- ; transpose for the first stage
- movq mm3, mm0 ; 00 01 02 03
- movq mm5, mm2 ; 20 21 22 23
-
- punpcklwd mm0, mm1 ; 00 10 01 11
- punpckhwd mm3, mm1 ; 02 12 03 13
-
- punpcklwd mm2, mm4 ; 20 30 21 31
- punpckhwd mm5, mm4 ; 22 32 23 33
-
- movq mm1, mm0 ; 00 10 01 11
- punpckldq mm0, mm2 ; 00 10 20 30
-
- punpckhdq mm1, mm2 ; 01 11 21 31
-
- movq mm2, mm3 ; 02 12 03 13
- punpckldq mm2, mm5 ; 02 12 22 32
-
- punpckhdq mm3, mm5 ; 03 13 23 33
-
- ; mm0 0
- ; mm1 1
- ; mm2 2
- ; mm3 3
-
- ; first stage
- movq mm5, mm0
- movq mm4, mm1
-
- paddw mm0, mm3 ; a1 = 0 + 3
- paddw mm1, mm2 ; b1 = 1 + 2
-
- psubw mm4, mm2 ; c1 = 1 - 2
- psubw mm5, mm3 ; d1 = 0 - 3
-
- psllw mm5, 3
- psllw mm4, 3
-
- psllw mm0, 3
- psllw mm1, 3
-
- ; output 0 and 2
- movq mm2, mm0 ; a1
-
- paddw mm0, mm1 ; op[0] = a1 + b1
- psubw mm2, mm1 ; op[2] = a1 - b1
-
- ; output 1 and 3
- ; interleave c1, d1
- movq mm1, mm5 ; d1
- punpcklwd mm1, mm4 ; c1 d1
- punpckhwd mm5, mm4 ; c1 d1
-
- movq mm3, mm1
- movq mm4, mm5
-
- pmaddwd mm1, MMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
- pmaddwd mm4, MMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
-
- pmaddwd mm3, MMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
- pmaddwd mm5, MMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
-
- paddd mm1, MMWORD PTR[GLOBAL(_14500)]
- paddd mm4, MMWORD PTR[GLOBAL(_14500)]
- paddd mm3, MMWORD PTR[GLOBAL(_7500)]
- paddd mm5, MMWORD PTR[GLOBAL(_7500)]
-
- psrad mm1, 12 ; (c1 * 2217 + d1 * 5352 + 14500)>>12
- psrad mm4, 12 ; (c1 * 2217 + d1 * 5352 + 14500)>>12
- psrad mm3, 12 ; (d1 * 2217 - c1 * 5352 + 7500)>>12
- psrad mm5, 12 ; (d1 * 2217 - c1 * 5352 + 7500)>>12
-
- packssdw mm1, mm4 ; op[1]
- packssdw mm3, mm5 ; op[3]
-
- ; done with vertical
- ; transpose for the second stage
- movq mm4, mm0 ; 00 10 20 30
- movq mm5, mm2 ; 02 12 22 32
-
- punpcklwd mm0, mm1 ; 00 01 10 11
- punpckhwd mm4, mm1 ; 20 21 30 31
-
- punpcklwd mm2, mm3 ; 02 03 12 13
- punpckhwd mm5, mm3 ; 22 23 32 33
-
- movq mm1, mm0 ; 00 01 10 11
- punpckldq mm0, mm2 ; 00 01 02 03
-
- punpckhdq mm1, mm2 ; 01 22 12 13
-
- movq mm2, mm4 ; 20 31 30 31
- punpckldq mm2, mm5 ; 20 21 22 23
-
- punpckhdq mm4, mm5 ; 30 31 32 33
-
- ; mm0 0
- ; mm1 1
- ; mm2 2
- ; mm3 4
-
- movq mm5, mm0
- movq mm3, mm1
-
- paddw mm0, mm4 ; a1 = 0 + 3
- paddw mm1, mm2 ; b1 = 1 + 2
-
- psubw mm3, mm2 ; c1 = 1 - 2
- psubw mm5, mm4 ; d1 = 0 - 3
-
- pxor mm6, mm6 ; zero out for compare
-
- pcmpeqw mm6, mm5 ; d1 != 0
-
- pandn mm6, MMWORD PTR[GLOBAL(_cmp_mask)] ; clear upper,
- ; and keep bit 0 of lower
-
- ; output 0 and 2
- movq mm2, mm0 ; a1
-
- paddw mm0, mm1 ; a1 + b1
- psubw mm2, mm1 ; a1 - b1
-
- paddw mm0, MMWORD PTR[GLOBAL(_7w)]
- paddw mm2, MMWORD PTR[GLOBAL(_7w)]
-
- psraw mm0, 4 ; op[0] = (a1 + b1 + 7)>>4
- psraw mm2, 4 ; op[8] = (a1 - b1 + 7)>>4
-
- movq MMWORD PTR[rdi + 0 ], mm0
- movq MMWORD PTR[rdi + 16], mm2
-
- ; output 1 and 3
- ; interleave c1, d1
- movq mm1, mm5 ; d1
- punpcklwd mm1, mm3 ; c1 d1
- punpckhwd mm5, mm3 ; c1 d1
-
- movq mm3, mm1
- movq mm4, mm5
-
- pmaddwd mm1, MMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
- pmaddwd mm4, MMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
-
- pmaddwd mm3, MMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
- pmaddwd mm5, MMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
-
- paddd mm1, MMWORD PTR[GLOBAL(_12000)]
- paddd mm4, MMWORD PTR[GLOBAL(_12000)]
- paddd mm3, MMWORD PTR[GLOBAL(_51000)]
- paddd mm5, MMWORD PTR[GLOBAL(_51000)]
-
- psrad mm1, 16 ; (c1 * 2217 + d1 * 5352 + 14500)>>16
- psrad mm4, 16 ; (c1 * 2217 + d1 * 5352 + 14500)>>16
- psrad mm3, 16 ; (d1 * 2217 - c1 * 5352 + 7500)>>16
- psrad mm5, 16 ; (d1 * 2217 - c1 * 5352 + 7500)>>16
-
- packssdw mm1, mm4 ; op[4]
- packssdw mm3, mm5 ; op[12]
-
- paddw mm1, mm6 ; op[4] += (d1!=0)
-
- movq MMWORD PTR[rdi + 8 ], mm1
- movq MMWORD PTR[rdi + 24], mm3
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 8
-_5352_2217:
- dw 5352
- dw 2217
- dw 5352
- dw 2217
-align 8
-_2217_neg5352:
- dw 2217
- dw -5352
- dw 2217
- dw -5352
-align 8
-_cmp_mask:
- times 4 dw 1
-align 8
-_7w:
- times 4 dw 7
-align 8
-_14500:
- times 2 dd 14500
-align 8
-_7500:
- times 2 dd 7500
-align 8
-_12000:
- times 2 dd 12000
-align 8
-_51000:
- times 2 dd 51000
diff --git a/vp8/encoder/x86/dct_sse2.asm b/vp8/encoder/x86/dct_sse2.asm
deleted file mode 100644
index 652dd98..0000000
--- a/vp8/encoder/x86/dct_sse2.asm
+++ /dev/null
@@ -1,430 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%macro STACK_FRAME_CREATE 0
-%if ABI_IS_32BIT
- %define input rsi
- %define output rdi
- %define pitch rax
- push rbp
- mov rbp, rsp
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0)
- mov rdi, arg(1)
-
- movsxd rax, dword ptr arg(2)
- lea rcx, [rsi + rax*2]
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- %define input rcx
- %define output rdx
- %define pitch r8
- %else
- %define input rdi
- %define output rsi
- %define pitch rdx
- %endif
-%endif
-%endmacro
-
-%macro STACK_FRAME_DESTROY 0
- %define input
- %define output
- %define pitch
-
-%if ABI_IS_32BIT
- pop rdi
- pop rsi
- RESTORE_GOT
- pop rbp
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- %endif
-%endif
- ret
-%endmacro
-
-;void vp8_short_fdct4x4_sse2(short *input, short *output, int pitch)
-global sym(vp8_short_fdct4x4_sse2)
-sym(vp8_short_fdct4x4_sse2):
-
- STACK_FRAME_CREATE
-
- movq xmm0, MMWORD PTR[input ] ;03 02 01 00
- movq xmm2, MMWORD PTR[input+ pitch] ;13 12 11 10
- lea input, [input+2*pitch]
- movq xmm1, MMWORD PTR[input ] ;23 22 21 20
- movq xmm3, MMWORD PTR[input+ pitch] ;33 32 31 30
-
- punpcklqdq xmm0, xmm2 ;13 12 11 10 03 02 01 00
- punpcklqdq xmm1, xmm3 ;33 32 31 30 23 22 21 20
-
- movdqa xmm2, xmm0
- punpckldq xmm0, xmm1 ;23 22 03 02 21 20 01 00
- punpckhdq xmm2, xmm1 ;33 32 13 12 31 30 11 10
- movdqa xmm1, xmm0
- punpckldq xmm0, xmm2 ;31 21 30 20 11 10 01 00
- pshufhw xmm1, xmm1, 0b1h ;22 23 02 03 xx xx xx xx
- pshufhw xmm2, xmm2, 0b1h ;32 33 12 13 xx xx xx xx
-
- punpckhdq xmm1, xmm2 ;32 33 22 23 12 13 02 03
- movdqa xmm3, xmm0
- paddw xmm0, xmm1 ;b1 a1 b1 a1 b1 a1 b1 a1
- psubw xmm3, xmm1 ;c1 d1 c1 d1 c1 d1 c1 d1
- psllw xmm0, 3 ;b1 <<= 3 a1 <<= 3
- psllw xmm3, 3 ;c1 <<= 3 d1 <<= 3
-
- movdqa xmm1, xmm0
- pmaddwd xmm0, XMMWORD PTR[GLOBAL(_mult_add)] ;a1 + b1
- pmaddwd xmm1, XMMWORD PTR[GLOBAL(_mult_sub)] ;a1 - b1
- movdqa xmm4, xmm3
- pmaddwd xmm3, XMMWORD PTR[GLOBAL(_5352_2217)] ;c1*2217 + d1*5352
- pmaddwd xmm4, XMMWORD PTR[GLOBAL(_2217_neg5352)];d1*2217 - c1*5352
-
- paddd xmm3, XMMWORD PTR[GLOBAL(_14500)]
- paddd xmm4, XMMWORD PTR[GLOBAL(_7500)]
- psrad xmm3, 12 ;(c1 * 2217 + d1 * 5352 + 14500)>>12
- psrad xmm4, 12 ;(d1 * 2217 - c1 * 5352 + 7500)>>12
-
- packssdw xmm0, xmm1 ;op[2] op[0]
- packssdw xmm3, xmm4 ;op[3] op[1]
- ; 23 22 21 20 03 02 01 00
- ;
- ; 33 32 31 30 13 12 11 10
- ;
- movdqa xmm2, xmm0
- punpcklqdq xmm0, xmm3 ;13 12 11 10 03 02 01 00
- punpckhqdq xmm2, xmm3 ;23 22 21 20 33 32 31 30
-
- movdqa xmm3, xmm0
- punpcklwd xmm0, xmm2 ;32 30 22 20 12 10 02 00
- punpckhwd xmm3, xmm2 ;33 31 23 21 13 11 03 01
- movdqa xmm2, xmm0
- punpcklwd xmm0, xmm3 ;13 12 11 10 03 02 01 00
- punpckhwd xmm2, xmm3 ;33 32 31 30 23 22 21 20
-
- movdqa xmm5, XMMWORD PTR[GLOBAL(_7)]
- pshufd xmm2, xmm2, 04eh
- movdqa xmm3, xmm0
- paddw xmm0, xmm2 ;b1 b1 b1 b1 a1 a1 a1 a1
- psubw xmm3, xmm2 ;c1 c1 c1 c1 d1 d1 d1 d1
-
- pshufd xmm0, xmm0, 0d8h ;b1 b1 a1 a1 b1 b1 a1 a1
- movdqa xmm2, xmm3 ;save d1 for compare
- pshufd xmm3, xmm3, 0d8h ;c1 c1 d1 d1 c1 c1 d1 d1
- pshuflw xmm0, xmm0, 0d8h ;b1 b1 a1 a1 b1 a1 b1 a1
- pshuflw xmm3, xmm3, 0d8h ;c1 c1 d1 d1 c1 d1 c1 d1
- pshufhw xmm0, xmm0, 0d8h ;b1 a1 b1 a1 b1 a1 b1 a1
- pshufhw xmm3, xmm3, 0d8h ;c1 d1 c1 d1 c1 d1 c1 d1
- movdqa xmm1, xmm0
- pmaddwd xmm0, XMMWORD PTR[GLOBAL(_mult_add)] ;a1 + b1
- pmaddwd xmm1, XMMWORD PTR[GLOBAL(_mult_sub)] ;a1 - b1
-
- pxor xmm4, xmm4 ;zero out for compare
- paddd xmm0, xmm5
- paddd xmm1, xmm5
- pcmpeqw xmm2, xmm4
- psrad xmm0, 4 ;(a1 + b1 + 7)>>4
- psrad xmm1, 4 ;(a1 - b1 + 7)>>4
- pandn xmm2, XMMWORD PTR[GLOBAL(_cmp_mask)] ;clear upper,
- ;and keep bit 0 of lower
-
- movdqa xmm4, xmm3
- pmaddwd xmm3, XMMWORD PTR[GLOBAL(_5352_2217)] ;c1*2217 + d1*5352
- pmaddwd xmm4, XMMWORD PTR[GLOBAL(_2217_neg5352)] ;d1*2217 - c1*5352
- paddd xmm3, XMMWORD PTR[GLOBAL(_12000)]
- paddd xmm4, XMMWORD PTR[GLOBAL(_51000)]
- packssdw xmm0, xmm1 ;op[8] op[0]
- psrad xmm3, 16 ;(c1 * 2217 + d1 * 5352 + 12000)>>16
- psrad xmm4, 16 ;(d1 * 2217 - c1 * 5352 + 51000)>>16
-
- packssdw xmm3, xmm4 ;op[12] op[4]
- movdqa xmm1, xmm0
- paddw xmm3, xmm2 ;op[4] += (d1!=0)
- punpcklqdq xmm0, xmm3 ;op[4] op[0]
- punpckhqdq xmm1, xmm3 ;op[12] op[8]
-
- movdqa XMMWORD PTR[output + 0], xmm0
- movdqa XMMWORD PTR[output + 16], xmm1
-
- STACK_FRAME_DESTROY
-
-;void vp8_short_fdct8x4_sse2(short *input, short *output, int pitch)
-global sym(vp8_short_fdct8x4_sse2)
-sym(vp8_short_fdct8x4_sse2):
-
- STACK_FRAME_CREATE
-
- ; read the input data
- movdqa xmm0, [input ]
- movdqa xmm2, [input+ pitch]
- lea input, [input+2*pitch]
- movdqa xmm4, [input ]
- movdqa xmm3, [input+ pitch]
-
- ; transpose for the first stage
- movdqa xmm1, xmm0 ; 00 01 02 03 04 05 06 07
- movdqa xmm5, xmm4 ; 20 21 22 23 24 25 26 27
-
- punpcklwd xmm0, xmm2 ; 00 10 01 11 02 12 03 13
- punpckhwd xmm1, xmm2 ; 04 14 05 15 06 16 07 17
-
- punpcklwd xmm4, xmm3 ; 20 30 21 31 22 32 23 33
- punpckhwd xmm5, xmm3 ; 24 34 25 35 26 36 27 37
-
- movdqa xmm2, xmm0 ; 00 10 01 11 02 12 03 13
- punpckldq xmm0, xmm4 ; 00 10 20 30 01 11 21 31
-
- punpckhdq xmm2, xmm4 ; 02 12 22 32 03 13 23 33
-
- movdqa xmm4, xmm1 ; 04 14 05 15 06 16 07 17
- punpckldq xmm4, xmm5 ; 04 14 24 34 05 15 25 35
-
- punpckhdq xmm1, xmm5 ; 06 16 26 36 07 17 27 37
- movdqa xmm3, xmm2 ; 02 12 22 32 03 13 23 33
-
- punpckhqdq xmm3, xmm1 ; 03 13 23 33 07 17 27 37
- punpcklqdq xmm2, xmm1 ; 02 12 22 32 06 16 26 36
-
- movdqa xmm1, xmm0 ; 00 10 20 30 01 11 21 31
- punpcklqdq xmm0, xmm4 ; 00 10 20 30 04 14 24 34
-
- punpckhqdq xmm1, xmm4 ; 01 11 21 32 05 15 25 35
-
- ; xmm0 0
- ; xmm1 1
- ; xmm2 2
- ; xmm3 3
-
- ; first stage
- movdqa xmm5, xmm0
- movdqa xmm4, xmm1
-
- paddw xmm0, xmm3 ; a1 = 0 + 3
- paddw xmm1, xmm2 ; b1 = 1 + 2
-
- psubw xmm4, xmm2 ; c1 = 1 - 2
- psubw xmm5, xmm3 ; d1 = 0 - 3
-
- psllw xmm5, 3
- psllw xmm4, 3
-
- psllw xmm0, 3
- psllw xmm1, 3
-
- ; output 0 and 2
- movdqa xmm2, xmm0 ; a1
-
- paddw xmm0, xmm1 ; op[0] = a1 + b1
- psubw xmm2, xmm1 ; op[2] = a1 - b1
-
- ; output 1 and 3
- ; interleave c1, d1
- movdqa xmm1, xmm5 ; d1
- punpcklwd xmm1, xmm4 ; c1 d1
- punpckhwd xmm5, xmm4 ; c1 d1
-
- movdqa xmm3, xmm1
- movdqa xmm4, xmm5
-
- pmaddwd xmm1, XMMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
- pmaddwd xmm4, XMMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
-
- pmaddwd xmm3, XMMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
- pmaddwd xmm5, XMMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
-
- paddd xmm1, XMMWORD PTR[GLOBAL(_14500)]
- paddd xmm4, XMMWORD PTR[GLOBAL(_14500)]
- paddd xmm3, XMMWORD PTR[GLOBAL(_7500)]
- paddd xmm5, XMMWORD PTR[GLOBAL(_7500)]
-
- psrad xmm1, 12 ; (c1 * 2217 + d1 * 5352 + 14500)>>12
- psrad xmm4, 12 ; (c1 * 2217 + d1 * 5352 + 14500)>>12
- psrad xmm3, 12 ; (d1 * 2217 - c1 * 5352 + 7500)>>12
- psrad xmm5, 12 ; (d1 * 2217 - c1 * 5352 + 7500)>>12
-
- packssdw xmm1, xmm4 ; op[1]
- packssdw xmm3, xmm5 ; op[3]
-
- ; done with vertical
- ; transpose for the second stage
- movdqa xmm4, xmm0 ; 00 10 20 30 04 14 24 34
- movdqa xmm5, xmm2 ; 02 12 22 32 06 16 26 36
-
- punpcklwd xmm0, xmm1 ; 00 01 10 11 20 21 30 31
- punpckhwd xmm4, xmm1 ; 04 05 14 15 24 25 34 35
-
- punpcklwd xmm2, xmm3 ; 02 03 12 13 22 23 32 33
- punpckhwd xmm5, xmm3 ; 06 07 16 17 26 27 36 37
-
- movdqa xmm1, xmm0 ; 00 01 10 11 20 21 30 31
- punpckldq xmm0, xmm2 ; 00 01 02 03 10 11 12 13
-
- punpckhdq xmm1, xmm2 ; 20 21 22 23 30 31 32 33
-
- movdqa xmm2, xmm4 ; 04 05 14 15 24 25 34 35
- punpckldq xmm2, xmm5 ; 04 05 06 07 14 15 16 17
-
- punpckhdq xmm4, xmm5 ; 24 25 26 27 34 35 36 37
- movdqa xmm3, xmm1 ; 20 21 22 23 30 31 32 33
-
- punpckhqdq xmm3, xmm4 ; 30 31 32 33 34 35 36 37
- punpcklqdq xmm1, xmm4 ; 20 21 22 23 24 25 26 27
-
- movdqa xmm4, xmm0 ; 00 01 02 03 10 11 12 13
- punpcklqdq xmm0, xmm2 ; 00 01 02 03 04 05 06 07
-
- punpckhqdq xmm4, xmm2 ; 10 11 12 13 14 15 16 17
-
- ; xmm0 0
- ; xmm1 4
- ; xmm2 1
- ; xmm3 3
-
- movdqa xmm5, xmm0
- movdqa xmm2, xmm1
-
- paddw xmm0, xmm3 ; a1 = 0 + 3
- paddw xmm1, xmm4 ; b1 = 1 + 2
-
- psubw xmm4, xmm2 ; c1 = 1 - 2
- psubw xmm5, xmm3 ; d1 = 0 - 3
-
- pxor xmm6, xmm6 ; zero out for compare
-
- pcmpeqw xmm6, xmm5 ; d1 != 0
-
- pandn xmm6, XMMWORD PTR[GLOBAL(_cmp_mask8x4)] ; clear upper,
- ; and keep bit 0 of lower
-
- ; output 0 and 2
- movdqa xmm2, xmm0 ; a1
-
- paddw xmm0, xmm1 ; a1 + b1
- psubw xmm2, xmm1 ; a1 - b1
-
- paddw xmm0, XMMWORD PTR[GLOBAL(_7w)]
- paddw xmm2, XMMWORD PTR[GLOBAL(_7w)]
-
- psraw xmm0, 4 ; op[0] = (a1 + b1 + 7)>>4
- psraw xmm2, 4 ; op[8] = (a1 - b1 + 7)>>4
-
- ; output 1 and 3
- ; interleave c1, d1
- movdqa xmm1, xmm5 ; d1
- punpcklwd xmm1, xmm4 ; c1 d1
- punpckhwd xmm5, xmm4 ; c1 d1
-
- movdqa xmm3, xmm1
- movdqa xmm4, xmm5
-
- pmaddwd xmm1, XMMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
- pmaddwd xmm4, XMMWORD PTR[GLOBAL (_5352_2217)] ; c1*2217 + d1*5352
-
- pmaddwd xmm3, XMMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
- pmaddwd xmm5, XMMWORD PTR[GLOBAL(_2217_neg5352)] ; d1*2217 - c1*5352
-
- paddd xmm1, XMMWORD PTR[GLOBAL(_12000)]
- paddd xmm4, XMMWORD PTR[GLOBAL(_12000)]
- paddd xmm3, XMMWORD PTR[GLOBAL(_51000)]
- paddd xmm5, XMMWORD PTR[GLOBAL(_51000)]
-
- psrad xmm1, 16 ; (c1 * 2217 + d1 * 5352 + 14500)>>16
- psrad xmm4, 16 ; (c1 * 2217 + d1 * 5352 + 14500)>>16
- psrad xmm3, 16 ; (d1 * 2217 - c1 * 5352 + 7500)>>16
- psrad xmm5, 16 ; (d1 * 2217 - c1 * 5352 + 7500)>>16
-
- packssdw xmm1, xmm4 ; op[4]
- packssdw xmm3, xmm5 ; op[12]
-
- paddw xmm1, xmm6 ; op[4] += (d1!=0)
-
- movdqa xmm4, xmm0
- movdqa xmm5, xmm2
-
- punpcklqdq xmm0, xmm1
- punpckhqdq xmm4, xmm1
-
- punpcklqdq xmm2, xmm3
- punpckhqdq xmm5, xmm3
-
- movdqa XMMWORD PTR[output + 0 ], xmm0
- movdqa XMMWORD PTR[output + 16], xmm2
- movdqa XMMWORD PTR[output + 32], xmm4
- movdqa XMMWORD PTR[output + 48], xmm5
-
- STACK_FRAME_DESTROY
-
-SECTION_RODATA
-align 16
-_5352_2217:
- dw 5352
- dw 2217
- dw 5352
- dw 2217
- dw 5352
- dw 2217
- dw 5352
- dw 2217
-align 16
-_2217_neg5352:
- dw 2217
- dw -5352
- dw 2217
- dw -5352
- dw 2217
- dw -5352
- dw 2217
- dw -5352
-align 16
-_mult_add:
- times 8 dw 1
-align 16
-_cmp_mask:
- times 4 dw 1
- times 4 dw 0
-align 16
-_cmp_mask8x4:
- times 8 dw 1
-align 16
-_mult_sub:
- dw 1
- dw -1
- dw 1
- dw -1
- dw 1
- dw -1
- dw 1
- dw -1
-align 16
-_7:
- times 4 dd 7
-align 16
-_7w:
- times 8 dw 7
-align 16
-_14500:
- times 4 dd 14500
-align 16
-_7500:
- times 4 dd 7500
-align 16
-_12000:
- times 4 dd 12000
-align 16
-_51000:
- times 4 dd 51000
diff --git a/vp8/encoder/x86/dct_x86.h b/vp8/encoder/x86/dct_x86.h
deleted file mode 100644
index 59a5cb1..0000000
--- a/vp8/encoder/x86/dct_x86.h
+++ /dev/null
@@ -1,67 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef DCT_X86_H
-#define DCT_X86_H
-
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-#if HAVE_MMX
-extern prototype_fdct(vp8_short_fdct4x4_mmx);
-extern prototype_fdct(vp8_short_fdct8x4_mmx);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_fdct_short4x4
-#define vp8_fdct_short4x4 vp8_short_fdct4x4_mmx
-
-#undef vp8_fdct_short8x4
-#define vp8_fdct_short8x4 vp8_short_fdct8x4_mmx
-
-#endif
-
-#endif
-
-
-#if HAVE_SSE2
-extern prototype_fdct(vp8_short_fdct8x4_sse2);
-extern prototype_fdct(vp8_short_walsh4x4_sse2);
-
-extern prototype_fdct(vp8_short_fdct4x4_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_fdct_short4x4
-#define vp8_fdct_short4x4 vp8_short_fdct4x4_sse2
-
-#undef vp8_fdct_short8x4
-#define vp8_fdct_short8x4 vp8_short_fdct8x4_sse2
-
-#undef vp8_fdct_fast4x4
-#define vp8_fdct_fast4x4 vp8_short_fdct4x4_sse2
-
-#undef vp8_fdct_fast8x4
-#define vp8_fdct_fast8x4 vp8_short_fdct8x4_sse2
-
-#undef vp8_fdct_walsh_short4x4
-#define vp8_fdct_walsh_short4x4 vp8_short_walsh4x4_sse2
-
-#endif
-
-
-#endif
-
-#endif
diff --git a/vp8/encoder/x86/encodemb_x86.h b/vp8/encoder/x86/encodemb_x86.h
deleted file mode 100644
index 69b3edd..0000000
--- a/vp8/encoder/x86/encodemb_x86.h
+++ /dev/null
@@ -1,85 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef ENCODEMB_X86_H
-#define ENCODEMB_X86_H
-
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-#if HAVE_MMX
-extern prototype_berr(vp8_block_error_mmx);
-extern prototype_mberr(vp8_mbblock_error_mmx);
-extern prototype_mbuverr(vp8_mbuverror_mmx);
-extern prototype_subb(vp8_subtract_b_mmx);
-extern prototype_submby(vp8_subtract_mby_mmx);
-extern prototype_submbuv(vp8_subtract_mbuv_mmx);
-
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_encodemb_berr
-#define vp8_encodemb_berr vp8_block_error_mmx
-
-#undef vp8_encodemb_mberr
-#define vp8_encodemb_mberr vp8_mbblock_error_mmx
-
-#undef vp8_encodemb_mbuverr
-#define vp8_encodemb_mbuverr vp8_mbuverror_mmx
-
-#undef vp8_encodemb_subb
-#define vp8_encodemb_subb vp8_subtract_b_mmx
-
-#undef vp8_encodemb_submby
-#define vp8_encodemb_submby vp8_subtract_mby_mmx
-
-#undef vp8_encodemb_submbuv
-#define vp8_encodemb_submbuv vp8_subtract_mbuv_mmx
-
-#endif
-#endif
-
-
-#if HAVE_SSE2
-extern prototype_berr(vp8_block_error_xmm);
-extern prototype_mberr(vp8_mbblock_error_xmm);
-extern prototype_mbuverr(vp8_mbuverror_xmm);
-extern prototype_subb(vp8_subtract_b_sse2);
-extern prototype_submby(vp8_subtract_mby_sse2);
-extern prototype_submbuv(vp8_subtract_mbuv_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_encodemb_berr
-#define vp8_encodemb_berr vp8_block_error_xmm
-
-#undef vp8_encodemb_mberr
-#define vp8_encodemb_mberr vp8_mbblock_error_xmm
-
-#undef vp8_encodemb_mbuverr
-#define vp8_encodemb_mbuverr vp8_mbuverror_xmm
-
-#undef vp8_encodemb_subb
-#define vp8_encodemb_subb vp8_subtract_b_sse2
-
-#undef vp8_encodemb_submby
-#define vp8_encodemb_submby vp8_subtract_mby_sse2
-
-#undef vp8_encodemb_submbuv
-#define vp8_encodemb_submbuv vp8_subtract_mbuv_sse2
-
-#endif
-#endif
-
-
-#endif
diff --git a/vp8/encoder/x86/encodeopt.asm b/vp8/encoder/x86/encodeopt.asm
deleted file mode 100644
index c0f06bb..0000000
--- a/vp8/encoder/x86/encodeopt.asm
+++ /dev/null
@@ -1,384 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;int vp8_block_error_xmm(short *coeff_ptr, short *dcoef_ptr)
-global sym(vp8_block_error_xmm)
-sym(vp8_block_error_xmm):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 2
- push rsi
- push rdi
- ; end prologue
-
- mov rsi, arg(0) ;coeff_ptr
-
- mov rdi, arg(1) ;dcoef_ptr
- movdqa xmm3, [rsi]
-
- movdqa xmm4, [rdi]
- movdqa xmm5, [rsi+16]
-
- movdqa xmm6, [rdi+16]
- psubw xmm3, xmm4
-
- psubw xmm5, xmm6
- pmaddwd xmm3, xmm3
- pmaddwd xmm5, xmm5
-
- paddd xmm3, xmm5
-
- pxor xmm7, xmm7
- movdqa xmm0, xmm3
-
- punpckldq xmm0, xmm7
- punpckhdq xmm3, xmm7
-
- paddd xmm0, xmm3
- movdqa xmm3, xmm0
-
- psrldq xmm0, 8
- paddd xmm0, xmm3
-
- movq rax, xmm0
-
- pop rdi
- pop rsi
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;int vp8_block_error_mmx(short *coeff_ptr, short *dcoef_ptr)
-global sym(vp8_block_error_mmx)
-sym(vp8_block_error_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 2
- push rsi
- push rdi
- ; end prolog
-
-
- mov rsi, arg(0) ;coeff_ptr
- pxor mm7, mm7
-
- mov rdi, arg(1) ;dcoef_ptr
- movq mm3, [rsi]
-
- movq mm4, [rdi]
- movq mm5, [rsi+8]
-
- movq mm6, [rdi+8]
- pxor mm1, mm1 ; from movd mm1, dc ; dc =0
-
- movq mm2, mm7
- psubw mm5, mm6
-
- por mm1, mm2
- pmaddwd mm5, mm5
-
- pcmpeqw mm1, mm7
- psubw mm3, mm4
-
- pand mm1, mm3
- pmaddwd mm1, mm1
-
- paddd mm1, mm5
- movq mm3, [rsi+16]
-
- movq mm4, [rdi+16]
- movq mm5, [rsi+24]
-
- movq mm6, [rdi+24]
- psubw mm5, mm6
-
- pmaddwd mm5, mm5
- psubw mm3, mm4
-
- pmaddwd mm3, mm3
- paddd mm3, mm5
-
- paddd mm1, mm3
- movq mm0, mm1
-
- psrlq mm1, 32
- paddd mm0, mm1
-
- movq rax, mm0
-
- pop rdi
- pop rsi
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;int vp8_mbblock_error_mmx_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
-global sym(vp8_mbblock_error_mmx_impl)
-sym(vp8_mbblock_error_mmx_impl):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 3
- push rsi
- push rdi
- ; end prolog
-
-
- mov rsi, arg(0) ;coeff_ptr
- pxor mm7, mm7
-
- mov rdi, arg(1) ;dcoef_ptr
- pxor mm2, mm2
-
- movd mm1, dword ptr arg(2) ;dc
- por mm1, mm2
-
- pcmpeqw mm1, mm7
- mov rcx, 16
-
-mberror_loop_mmx:
- movq mm3, [rsi]
- movq mm4, [rdi]
-
- movq mm5, [rsi+8]
- movq mm6, [rdi+8]
-
-
- psubw mm5, mm6
- pmaddwd mm5, mm5
-
- psubw mm3, mm4
- pand mm3, mm1
-
- pmaddwd mm3, mm3
- paddd mm2, mm5
-
- paddd mm2, mm3
- movq mm3, [rsi+16]
-
- movq mm4, [rdi+16]
- movq mm5, [rsi+24]
-
- movq mm6, [rdi+24]
- psubw mm5, mm6
-
- pmaddwd mm5, mm5
- psubw mm3, mm4
-
- pmaddwd mm3, mm3
- paddd mm2, mm5
-
- paddd mm2, mm3
- add rsi, 32
-
- add rdi, 32
- sub rcx, 1
-
- jnz mberror_loop_mmx
-
- movq mm0, mm2
- psrlq mm2, 32
-
- paddd mm0, mm2
- movq rax, mm0
-
- pop rdi
- pop rsi
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;int vp8_mbblock_error_xmm_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
-global sym(vp8_mbblock_error_xmm_impl)
-sym(vp8_mbblock_error_xmm_impl):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 3
- push rsi
- push rdi
- ; end prolog
-
-
- mov rsi, arg(0) ;coeff_ptr
- pxor xmm7, xmm7
-
- mov rdi, arg(1) ;dcoef_ptr
- pxor xmm2, xmm2
-
- movd xmm1, dword ptr arg(2) ;dc
- por xmm1, xmm2
-
- pcmpeqw xmm1, xmm7
- mov rcx, 16
-
-mberror_loop:
- movdqa xmm3, [rsi]
- movdqa xmm4, [rdi]
-
- movdqa xmm5, [rsi+16]
- movdqa xmm6, [rdi+16]
-
-
- psubw xmm5, xmm6
- pmaddwd xmm5, xmm5
-
- psubw xmm3, xmm4
- pand xmm3, xmm1
-
- pmaddwd xmm3, xmm3
- add rsi, 32
-
- add rdi, 32
-
- sub rcx, 1
- paddd xmm2, xmm5
-
- paddd xmm2, xmm3
- jnz mberror_loop
-
- movdqa xmm0, xmm2
- punpckldq xmm0, xmm7
-
- punpckhdq xmm2, xmm7
- paddd xmm0, xmm2
-
- movdqa xmm1, xmm0
- psrldq xmm0, 8
-
- paddd xmm0, xmm1
- movq rax, xmm0
-
- pop rdi
- pop rsi
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;int vp8_mbuverror_mmx_impl(short *s_ptr, short *d_ptr);
-global sym(vp8_mbuverror_mmx_impl)
-sym(vp8_mbuverror_mmx_impl):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 2
- push rsi
- push rdi
- ; end prolog
-
-
- mov rsi, arg(0) ;s_ptr
- mov rdi, arg(1) ;d_ptr
-
- mov rcx, 16
- pxor mm7, mm7
-
-mbuverror_loop_mmx:
-
- movq mm1, [rsi]
- movq mm2, [rdi]
-
- psubw mm1, mm2
- pmaddwd mm1, mm1
-
-
- movq mm3, [rsi+8]
- movq mm4, [rdi+8]
-
- psubw mm3, mm4
- pmaddwd mm3, mm3
-
-
- paddd mm7, mm1
- paddd mm7, mm3
-
-
- add rsi, 16
- add rdi, 16
-
- dec rcx
- jnz mbuverror_loop_mmx
-
- movq mm0, mm7
- psrlq mm7, 32
-
- paddd mm0, mm7
- movq rax, mm0
-
- pop rdi
- pop rsi
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;int vp8_mbuverror_xmm_impl(short *s_ptr, short *d_ptr);
-global sym(vp8_mbuverror_xmm_impl)
-sym(vp8_mbuverror_xmm_impl):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 2
- push rsi
- push rdi
- ; end prolog
-
-
- mov rsi, arg(0) ;s_ptr
- mov rdi, arg(1) ;d_ptr
-
- mov rcx, 16
- pxor xmm7, xmm7
-
-mbuverror_loop:
-
- movdqa xmm1, [rsi]
- movdqa xmm2, [rdi]
-
- psubw xmm1, xmm2
- pmaddwd xmm1, xmm1
-
- paddd xmm7, xmm1
-
- add rsi, 16
- add rdi, 16
-
- dec rcx
- jnz mbuverror_loop
-
- pxor xmm0, xmm0
- movdqa xmm1, xmm7
-
- movdqa xmm2, xmm1
- punpckldq xmm1, xmm0
-
- punpckhdq xmm2, xmm0
- paddd xmm1, xmm2
-
- movdqa xmm2, xmm1
-
- psrldq xmm1, 8
- paddd xmm1, xmm2
-
- movq rax, xmm1
-
- pop rdi
- pop rsi
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/encoder/x86/fwalsh_sse2.asm b/vp8/encoder/x86/fwalsh_sse2.asm
deleted file mode 100644
index 39439f0..0000000
--- a/vp8/encoder/x86/fwalsh_sse2.asm
+++ /dev/null
@@ -1,164 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_short_walsh4x4_sse2(short *input, short *output, int pitch)
-global sym(vp8_short_walsh4x4_sse2)
-sym(vp8_short_walsh4x4_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 3
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ; input
- mov rdi, arg(1) ; output
- movsxd rdx, dword ptr arg(2) ; pitch
-
- ; first for loop
- movq xmm0, MMWORD PTR [rsi] ; load input
- movq xmm1, MMWORD PTR [rsi + rdx]
- lea rsi, [rsi + rdx*2]
- movq xmm2, MMWORD PTR [rsi]
- movq xmm3, MMWORD PTR [rsi + rdx]
-
- punpcklwd xmm0, xmm1
- punpcklwd xmm2, xmm3
-
- movdqa xmm1, xmm0
- punpckldq xmm0, xmm2 ; ip[1] ip[0]
- punpckhdq xmm1, xmm2 ; ip[3] ip[2]
-
- movdqa xmm2, xmm0
- paddw xmm0, xmm1
- psubw xmm2, xmm1
-
- psllw xmm0, 2 ; d1 a1
- psllw xmm2, 2 ; c1 b1
-
- movdqa xmm1, xmm0
- punpcklqdq xmm0, xmm2 ; b1 a1
- punpckhqdq xmm1, xmm2 ; c1 d1
-
- pxor xmm6, xmm6
- movq xmm6, xmm0
- pxor xmm7, xmm7
- pcmpeqw xmm7, xmm6
- paddw xmm7, [GLOBAL(c1)]
-
- movdqa xmm2, xmm0
- paddw xmm0, xmm1 ; b1+c1 a1+d1
- psubw xmm2, xmm1 ; b1-c1 a1-d1
- paddw xmm0, xmm7 ; b1+c1 a1+d1+(a1!=0)
-
- ; second for loop
- ; input: 13 9 5 1 12 8 4 0 (xmm0)
- ; 14 10 6 2 15 11 7 3 (xmm2)
- ; after shuffle:
- ; 13 5 9 1 12 4 8 0 (xmm0)
- ; 14 6 10 2 15 7 11 3 (xmm1)
- pshuflw xmm3, xmm0, 0xd8
- pshufhw xmm0, xmm3, 0xd8
- pshuflw xmm3, xmm2, 0xd8
- pshufhw xmm1, xmm3, 0xd8
-
- movdqa xmm2, xmm0
- pmaddwd xmm0, [GLOBAL(c1)] ; d11 a11 d10 a10
- pmaddwd xmm2, [GLOBAL(cn1)] ; c11 b11 c10 b10
- movdqa xmm3, xmm1
- pmaddwd xmm1, [GLOBAL(c1)] ; d12 a12 d13 a13
- pmaddwd xmm3, [GLOBAL(cn1)] ; c12 b12 c13 b13
-
- pshufd xmm4, xmm0, 0xd8 ; d11 d10 a11 a10
- pshufd xmm5, xmm2, 0xd8 ; c11 c10 b11 b10
- pshufd xmm6, xmm1, 0x72 ; d13 d12 a13 a12
- pshufd xmm7, xmm3, 0x72 ; c13 c12 b13 b12
-
- movdqa xmm0, xmm4
- punpcklqdq xmm0, xmm5 ; b11 b10 a11 a10
- punpckhqdq xmm4, xmm5 ; c11 c10 d11 d10
- movdqa xmm1, xmm6
- punpcklqdq xmm1, xmm7 ; b13 b12 a13 a12
- punpckhqdq xmm6, xmm7 ; c13 c12 d13 d12
-
- movdqa xmm2, xmm0
- paddd xmm0, xmm4 ; b21 b20 a21 a20
- psubd xmm2, xmm4 ; c21 c20 d21 d20
- movdqa xmm3, xmm1
- paddd xmm1, xmm6 ; b23 b22 a23 a22
- psubd xmm3, xmm6 ; c23 c22 d23 d22
-
- pxor xmm4, xmm4
- movdqa xmm5, xmm4
- pcmpgtd xmm4, xmm0
- pcmpgtd xmm5, xmm2
- pand xmm4, [GLOBAL(cd1)]
- pand xmm5, [GLOBAL(cd1)]
-
- pxor xmm6, xmm6
- movdqa xmm7, xmm6
- pcmpgtd xmm6, xmm1
- pcmpgtd xmm7, xmm3
- pand xmm6, [GLOBAL(cd1)]
- pand xmm7, [GLOBAL(cd1)]
-
- paddd xmm0, xmm4
- paddd xmm2, xmm5
- paddd xmm0, [GLOBAL(cd3)]
- paddd xmm2, [GLOBAL(cd3)]
- paddd xmm1, xmm6
- paddd xmm3, xmm7
- paddd xmm1, [GLOBAL(cd3)]
- paddd xmm3, [GLOBAL(cd3)]
-
- psrad xmm0, 3
- psrad xmm1, 3
- psrad xmm2, 3
- psrad xmm3, 3
- movdqa xmm4, xmm0
- punpcklqdq xmm0, xmm1 ; a23 a22 a21 a20
- punpckhqdq xmm4, xmm1 ; b23 b22 b21 b20
- movdqa xmm5, xmm2
- punpckhqdq xmm2, xmm3 ; c23 c22 c21 c20
- punpcklqdq xmm5, xmm3 ; d23 d22 d21 d20
-
- packssdw xmm0, xmm4 ; b23 b22 b21 b20 a23 a22 a21 a20
- packssdw xmm2, xmm5 ; d23 d22 d21 d20 c23 c22 c21 c20
-
- movdqa XMMWORD PTR [rdi], xmm0
- movdqa XMMWORD PTR [rdi + 16], xmm2
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-c1:
- dw 0x0001, 0x0001, 0x0001, 0x0001, 0x0001, 0x0001, 0x0001, 0x0001
-align 16
-cn1:
- dw 0x0001, 0xffff, 0x0001, 0xffff, 0x0001, 0xffff, 0x0001, 0xffff
-align 16
-cd1:
- dd 0x00000001, 0x00000001, 0x00000001, 0x00000001
-align 16
-cd3:
- dd 0x00000003, 0x00000003, 0x00000003, 0x00000003
diff --git a/vp8/encoder/x86/mcomp_x86.h b/vp8/encoder/x86/mcomp_x86.h
deleted file mode 100644
index 3b7b29c..0000000
--- a/vp8/encoder/x86/mcomp_x86.h
+++ /dev/null
@@ -1,37 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef MCOMP_X86_H
-#define MCOMP_X86_H
-
-#if HAVE_SSE3
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_search_full_search
-#define vp8_search_full_search vp8_full_search_sadx3
-
-#undef vp8_search_diamond_search
-#define vp8_search_diamond_search vp8_diamond_search_sadx4
-
-#endif
-#endif
-
-#if HAVE_SSE4_1
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_search_full_search
-#define vp8_search_full_search vp8_full_search_sadx8
-
-#endif
-#endif
-
-#endif
-
diff --git a/vp8/encoder/x86/quantize_mmx.asm b/vp8/encoder/x86/quantize_mmx.asm
deleted file mode 100644
index f29a54e..0000000
--- a/vp8/encoder/x86/quantize_mmx.asm
+++ /dev/null
@@ -1,286 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;int vp8_fast_quantize_b_impl_mmx(short *coeff_ptr, short *zbin_ptr,
-; short *qcoeff_ptr,short *dequant_ptr,
-; short *scan_mask, short *round_ptr,
-; short *quant_ptr, short *dqcoeff_ptr);
-global sym(vp8_fast_quantize_b_impl_mmx)
-sym(vp8_fast_quantize_b_impl_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 8
- push rsi
- push rdi
- ; end prolog
-
-
- mov rsi, arg(0) ;coeff_ptr
- movq mm0, [rsi]
-
- mov rax, arg(1) ;zbin_ptr
- movq mm1, [rax]
-
- movq mm3, mm0
- psraw mm0, 15
-
- pxor mm3, mm0
- psubw mm3, mm0 ; abs
-
- movq mm2, mm3
- pcmpgtw mm1, mm2
-
- pandn mm1, mm2
- movq mm3, mm1
-
- mov rdx, arg(6) ;quant_ptr
- movq mm1, [rdx]
-
- mov rcx, arg(5) ;round_ptr
- movq mm2, [rcx]
-
- paddw mm3, mm2
- pmulhuw mm3, mm1
-
- pxor mm3, mm0
- psubw mm3, mm0 ;gain the sign back
-
- mov rdi, arg(2) ;qcoeff_ptr
- movq mm0, mm3
-
- movq [rdi], mm3
-
- mov rax, arg(3) ;dequant_ptr
- movq mm2, [rax]
-
- pmullw mm3, mm2
- mov rax, arg(7) ;dqcoeff_ptr
-
- movq [rax], mm3
-
- ; next 8
- movq mm4, [rsi+8]
-
- mov rax, arg(1) ;zbin_ptr
- movq mm5, [rax+8]
-
- movq mm7, mm4
- psraw mm4, 15
-
- pxor mm7, mm4
- psubw mm7, mm4 ; abs
-
- movq mm6, mm7
- pcmpgtw mm5, mm6
-
- pandn mm5, mm6
- movq mm7, mm5
-
- movq mm5, [rdx+8]
- movq mm6, [rcx+8]
-
- paddw mm7, mm6
- pmulhuw mm7, mm5
-
- pxor mm7, mm4
- psubw mm7, mm4;gain the sign back
-
- mov rdi, arg(2) ;qcoeff_ptr
-
- movq mm1, mm7
- movq [rdi+8], mm7
-
- mov rax, arg(3) ;dequant_ptr
- movq mm6, [rax+8]
-
- pmullw mm7, mm6
- mov rax, arg(7) ;dqcoeff_ptr
-
- movq [rax+8], mm7
-
-
- ; next 8
- movq mm4, [rsi+16]
-
- mov rax, arg(1) ;zbin_ptr
- movq mm5, [rax+16]
-
- movq mm7, mm4
- psraw mm4, 15
-
- pxor mm7, mm4
- psubw mm7, mm4 ; abs
-
- movq mm6, mm7
- pcmpgtw mm5, mm6
-
- pandn mm5, mm6
- movq mm7, mm5
-
- movq mm5, [rdx+16]
- movq mm6, [rcx+16]
-
- paddw mm7, mm6
- pmulhuw mm7, mm5
-
- pxor mm7, mm4
- psubw mm7, mm4;gain the sign back
-
- mov rdi, arg(2) ;qcoeff_ptr
-
- movq mm1, mm7
- movq [rdi+16], mm7
-
- mov rax, arg(3) ;dequant_ptr
- movq mm6, [rax+16]
-
- pmullw mm7, mm6
- mov rax, arg(7) ;dqcoeff_ptr
-
- movq [rax+16], mm7
-
-
- ; next 8
- movq mm4, [rsi+24]
-
- mov rax, arg(1) ;zbin_ptr
- movq mm5, [rax+24]
-
- movq mm7, mm4
- psraw mm4, 15
-
- pxor mm7, mm4
- psubw mm7, mm4 ; abs
-
- movq mm6, mm7
- pcmpgtw mm5, mm6
-
- pandn mm5, mm6
- movq mm7, mm5
-
- movq mm5, [rdx+24]
- movq mm6, [rcx+24]
-
- paddw mm7, mm6
- pmulhuw mm7, mm5
-
- pxor mm7, mm4
- psubw mm7, mm4;gain the sign back
-
- mov rdi, arg(2) ;qcoeff_ptr
-
- movq mm1, mm7
- movq [rdi+24], mm7
-
- mov rax, arg(3) ;dequant_ptr
- movq mm6, [rax+24]
-
- pmullw mm7, mm6
- mov rax, arg(7) ;dqcoeff_ptr
-
- movq [rax+24], mm7
-
-
-
- mov rdi, arg(4) ;scan_mask
- mov rsi, arg(2) ;qcoeff_ptr
-
- pxor mm5, mm5
- pxor mm7, mm7
-
- movq mm0, [rsi]
- movq mm1, [rsi+8]
-
- movq mm2, [rdi]
- movq mm3, [rdi+8];
-
- pcmpeqw mm0, mm7
- pcmpeqw mm1, mm7
-
- pcmpeqw mm6, mm6
- pxor mm0, mm6
-
- pxor mm1, mm6
- psrlw mm0, 15
-
- psrlw mm1, 15
- pmaddwd mm0, mm2
-
- pmaddwd mm1, mm3
- movq mm5, mm0
-
- paddd mm5, mm1
-
- movq mm0, [rsi+16]
- movq mm1, [rsi+24]
-
- movq mm2, [rdi+16]
- movq mm3, [rdi+24];
-
- pcmpeqw mm0, mm7
- pcmpeqw mm1, mm7
-
- pcmpeqw mm6, mm6
- pxor mm0, mm6
-
- pxor mm1, mm6
- psrlw mm0, 15
-
- psrlw mm1, 15
- pmaddwd mm0, mm2
-
- pmaddwd mm1, mm3
- paddd mm5, mm0
-
- paddd mm5, mm1
- movq mm0, mm5
-
- psrlq mm5, 32
- paddd mm0, mm5
-
- ; eob adjustment begins here
- movq rcx, mm0
- and rcx, 0xffff
-
- xor rdx, rdx
- sub rdx, rcx ; rdx=-rcx
-
- bsr rax, rcx
- inc rax
-
- sar rdx, 31
- and rax, rdx
- ; Substitute the sse assembly for the old mmx mixed assembly/C. The
- ; following is kept as reference
- ; movq rcx, mm0
- ; bsr rax, rcx
- ;
- ; mov eob, rax
- ; mov eee, rcx
- ;
- ;if(eee==0)
- ;{
- ; eob=-1;
- ;}
- ;else if(eee<0)
- ;{
- ; eob=15;
- ;}
- ;d->eob = eob+1;
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/encoder/x86/quantize_sse2.asm b/vp8/encoder/x86/quantize_sse2.asm
deleted file mode 100644
index 9a15840..0000000
--- a/vp8/encoder/x86/quantize_sse2.asm
+++ /dev/null
@@ -1,359 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license and patent
-; grant that can be found in the LICENSE file in the root of the source
-; tree. All contributing project authors may be found in the AUTHORS
-; file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-%include "asm_enc_offsets.asm"
-
-
-; void vp8_regular_quantize_b_sse2 | arg
-; (BLOCK *b, | 0
-; BLOCKD *d) | 1
-
-global sym(vp8_regular_quantize_b_sse2)
-sym(vp8_regular_quantize_b_sse2):
- push rbp
- mov rbp, rsp
- SAVE_XMM
- GET_GOT rbx
- push rsi
-
-%if ABI_IS_32BIT
- push rdi
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- push rdi
- %endif
-%endif
-
- ALIGN_STACK 16, rax
- %define BLOCKD_d 0 ; 8
- %define zrun_zbin_boost 8 ; 8
- %define abs_minus_zbin 16 ; 32
- %define temp_qcoeff 48 ; 32
- %define qcoeff 80 ; 32
- %define stack_size 112
- sub rsp, stack_size
- ; end prolog
-
-%if ABI_IS_32BIT
- mov rdi, arg(0)
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- mov rdi, rcx ; BLOCK *b
- mov [rsp + BLOCKD_d], rdx
- %else
- ;mov rdi, rdi ; BLOCK *b
- mov [rsp + BLOCKD_d], rsi
- %endif
-%endif
-
- mov rdx, [rdi + vp8_block_coeff] ; coeff_ptr
- mov rcx, [rdi + vp8_block_zbin] ; zbin_ptr
- movd xmm7, [rdi + vp8_block_zbin_extra] ; zbin_oq_value
-
- ; z
- movdqa xmm0, [rdx]
- movdqa xmm4, [rdx + 16]
- mov rdx, [rdi + vp8_block_round] ; round_ptr
-
- pshuflw xmm7, xmm7, 0
- punpcklwd xmm7, xmm7 ; duplicated zbin_oq_value
-
- movdqa xmm1, xmm0
- movdqa xmm5, xmm4
-
- ; sz
- psraw xmm0, 15
- psraw xmm4, 15
-
- ; (z ^ sz)
- pxor xmm1, xmm0
- pxor xmm5, xmm4
-
- ; x = abs(z)
- psubw xmm1, xmm0
- psubw xmm5, xmm4
-
- movdqa xmm2, [rcx]
- movdqa xmm3, [rcx + 16]
- mov rcx, [rdi + vp8_block_quant] ; quant_ptr
-
- ; *zbin_ptr + zbin_oq_value
- paddw xmm2, xmm7
- paddw xmm3, xmm7
-
- ; x - (*zbin_ptr + zbin_oq_value)
- psubw xmm1, xmm2
- psubw xmm5, xmm3
- movdqa [rsp + abs_minus_zbin], xmm1
- movdqa [rsp + abs_minus_zbin + 16], xmm5
-
- ; add (zbin_ptr + zbin_oq_value) back
- paddw xmm1, xmm2
- paddw xmm5, xmm3
-
- movdqa xmm2, [rdx]
- movdqa xmm6, [rdx + 16]
-
- movdqa xmm3, [rcx]
- movdqa xmm7, [rcx + 16]
-
- ; x + round
- paddw xmm1, xmm2
- paddw xmm5, xmm6
-
- ; y = x * quant_ptr >> 16
- pmulhw xmm3, xmm1
- pmulhw xmm7, xmm5
-
- ; y += x
- paddw xmm1, xmm3
- paddw xmm5, xmm7
-
- movdqa [rsp + temp_qcoeff], xmm1
- movdqa [rsp + temp_qcoeff + 16], xmm5
-
- pxor xmm6, xmm6
- ; zero qcoeff
- movdqa [rsp + qcoeff], xmm6
- movdqa [rsp + qcoeff + 16], xmm6
-
- mov rsi, [rdi + vp8_block_zrun_zbin_boost] ; zbin_boost_ptr
- mov rax, [rdi + vp8_block_quant_shift] ; quant_shift_ptr
- mov [rsp + zrun_zbin_boost], rsi
-
-%macro ZIGZAG_LOOP 1
- movsx edx, WORD PTR[GLOBAL(zig_zag + (%1 * 2))] ; rc
-
- ; x
- movsx ecx, WORD PTR[rsp + abs_minus_zbin + rdx *2]
-
- ; if (x >= zbin)
- sub cx, WORD PTR[rsi] ; x - zbin
- lea rsi, [rsi + 2] ; zbin_boost_ptr++
- jl rq_zigzag_loop_%1 ; x < zbin
-
- movsx edi, WORD PTR[rsp + temp_qcoeff + rdx *2]
-
- ; downshift by quant_shift[rdx]
- movsx ecx, WORD PTR[rax + rdx*2] ; quant_shift_ptr[rc]
- sar edi, cl ; also sets Z bit
- je rq_zigzag_loop_%1 ; !y
- mov WORD PTR[rsp + qcoeff + rdx*2], di ;qcoeff_ptr[rc] = temp_qcoeff[rc]
- mov rsi, [rsp + zrun_zbin_boost] ; reset to b->zrun_zbin_boost
-rq_zigzag_loop_%1:
-%endmacro
-ZIGZAG_LOOP 0
-ZIGZAG_LOOP 1
-ZIGZAG_LOOP 2
-ZIGZAG_LOOP 3
-ZIGZAG_LOOP 4
-ZIGZAG_LOOP 5
-ZIGZAG_LOOP 6
-ZIGZAG_LOOP 7
-ZIGZAG_LOOP 8
-ZIGZAG_LOOP 9
-ZIGZAG_LOOP 10
-ZIGZAG_LOOP 11
-ZIGZAG_LOOP 12
-ZIGZAG_LOOP 13
-ZIGZAG_LOOP 14
-ZIGZAG_LOOP 15
-
- movdqa xmm2, [rsp + qcoeff]
- movdqa xmm3, [rsp + qcoeff + 16]
-
-%if ABI_IS_32BIT
- mov rdi, arg(1)
-%else
- mov rdi, [rsp + BLOCKD_d]
-%endif
-
- mov rcx, [rdi + vp8_blockd_dequant] ; dequant_ptr
- mov rsi, [rdi + vp8_blockd_dqcoeff] ; dqcoeff_ptr
-
- ; y ^ sz
- pxor xmm2, xmm0
- pxor xmm3, xmm4
- ; x = (y ^ sz) - sz
- psubw xmm2, xmm0
- psubw xmm3, xmm4
-
- ; dequant
- movdqa xmm0, [rcx]
- movdqa xmm1, [rcx + 16]
-
- mov rcx, [rdi + vp8_blockd_qcoeff] ; qcoeff_ptr
-
- pmullw xmm0, xmm2
- pmullw xmm1, xmm3
-
- movdqa [rcx], xmm2 ; store qcoeff
- movdqa [rcx + 16], xmm3
- movdqa [rsi], xmm0 ; store dqcoeff
- movdqa [rsi + 16], xmm1
-
- ; select the last value (in zig_zag order) for EOB
- pcmpeqw xmm2, xmm6
- pcmpeqw xmm3, xmm6
- ; !
- pcmpeqw xmm6, xmm6
- pxor xmm2, xmm6
- pxor xmm3, xmm6
- ; mask inv_zig_zag
- pand xmm2, [GLOBAL(inv_zig_zag)]
- pand xmm3, [GLOBAL(inv_zig_zag + 16)]
- ; select the max value
- pmaxsw xmm2, xmm3
- pshufd xmm3, xmm2, 00001110b
- pmaxsw xmm2, xmm3
- pshuflw xmm3, xmm2, 00001110b
- pmaxsw xmm2, xmm3
- pshuflw xmm3, xmm2, 00000001b
- pmaxsw xmm2, xmm3
- movd eax, xmm2
- and eax, 0xff
- mov [rdi + vp8_blockd_eob], eax
-
- ; begin epilog
- add rsp, stack_size
- pop rsp
-%if ABI_IS_32BIT
- pop rdi
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- pop rdi
- %endif
-%endif
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- pop rbp
- ret
-
-; int vp8_fast_quantize_b_impl_sse2 | arg
-; (short *coeff_ptr, | 0
-; short *qcoeff_ptr, | 1
-; short *dequant_ptr, | 2
-; short *inv_scan_order, | 3
-; short *round_ptr, | 4
-; short *quant_ptr, | 5
-; short *dqcoeff_ptr) | 6
-
-global sym(vp8_fast_quantize_b_impl_sse2)
-sym(vp8_fast_quantize_b_impl_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- push rsi
- push rdi
- ; end prolog
-
- mov rdx, arg(0) ;coeff_ptr
- mov rcx, arg(2) ;dequant_ptr
- mov rdi, arg(4) ;round_ptr
- mov rsi, arg(5) ;quant_ptr
-
- movdqa xmm0, XMMWORD PTR[rdx]
- movdqa xmm4, XMMWORD PTR[rdx + 16]
-
- movdqa xmm2, XMMWORD PTR[rdi] ;round lo
- movdqa xmm3, XMMWORD PTR[rdi + 16] ;round hi
-
- movdqa xmm1, xmm0
- movdqa xmm5, xmm4
-
- psraw xmm0, 15 ;sign of z (aka sz)
- psraw xmm4, 15 ;sign of z (aka sz)
-
- pxor xmm1, xmm0
- pxor xmm5, xmm4
- psubw xmm1, xmm0 ;x = abs(z)
- psubw xmm5, xmm4 ;x = abs(z)
-
- paddw xmm1, xmm2
- paddw xmm5, xmm3
-
- pmulhw xmm1, XMMWORD PTR[rsi]
- pmulhw xmm5, XMMWORD PTR[rsi + 16]
-
- mov rdi, arg(1) ;qcoeff_ptr
- mov rsi, arg(6) ;dqcoeff_ptr
-
- movdqa xmm2, XMMWORD PTR[rcx]
- movdqa xmm3, XMMWORD PTR[rcx + 16]
-
- pxor xmm1, xmm0
- pxor xmm5, xmm4
- psubw xmm1, xmm0
- psubw xmm5, xmm4
-
- movdqa XMMWORD PTR[rdi], xmm1
- movdqa XMMWORD PTR[rdi + 16], xmm5
-
- pmullw xmm2, xmm1
- pmullw xmm3, xmm5
-
- mov rdi, arg(3) ;inv_scan_order
-
- ; Start with 16
- pxor xmm4, xmm4 ;clear all bits
- pcmpeqw xmm1, xmm4
- pcmpeqw xmm5, xmm4
-
- pcmpeqw xmm4, xmm4 ;set all bits
- pxor xmm1, xmm4
- pxor xmm5, xmm4
-
- pand xmm1, XMMWORD PTR[rdi]
- pand xmm5, XMMWORD PTR[rdi+16]
-
- pmaxsw xmm1, xmm5
-
- ; now down to 8
- pshufd xmm5, xmm1, 00001110b
-
- pmaxsw xmm1, xmm5
-
- ; only 4 left
- pshuflw xmm5, xmm1, 00001110b
-
- pmaxsw xmm1, xmm5
-
- ; okay, just 2!
- pshuflw xmm5, xmm1, 00000001b
-
- pmaxsw xmm1, xmm5
-
- movd rax, xmm1
- and rax, 0xff
-
- movdqa XMMWORD PTR[rsi], xmm2 ;store dqcoeff
- movdqa XMMWORD PTR[rsi + 16], xmm3 ;store dqcoeff
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-zig_zag:
- dw 0x0000, 0x0001, 0x0004, 0x0008
- dw 0x0005, 0x0002, 0x0003, 0x0006
- dw 0x0009, 0x000c, 0x000d, 0x000a
- dw 0x0007, 0x000b, 0x000e, 0x000f
-inv_zig_zag:
- dw 0x0001, 0x0002, 0x0006, 0x0007
- dw 0x0003, 0x0005, 0x0008, 0x000d
- dw 0x0004, 0x0009, 0x000c, 0x000e
- dw 0x000a, 0x000b, 0x000f, 0x0010
diff --git a/vp8/encoder/x86/quantize_ssse3.asm b/vp8/encoder/x86/quantize_ssse3.asm
deleted file mode 100644
index 2f33199..0000000
--- a/vp8/encoder/x86/quantize_ssse3.asm
+++ /dev/null
@@ -1,114 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license and patent
-; grant that can be found in the LICENSE file in the root of the source
-; tree. All contributing project authors may be found in the AUTHORS
-; file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-
-;int vp8_fast_quantize_b_impl_ssse3(short *coeff_ptr
-; short *qcoeff_ptr,short *dequant_ptr,
-; short *round_ptr,
-; short *quant_ptr, short *dqcoeff_ptr);
-;
-global sym(vp8_fast_quantize_b_impl_ssse3)
-sym(vp8_fast_quantize_b_impl_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rdx, arg(0) ;coeff_ptr
- mov rdi, arg(3) ;round_ptr
- mov rsi, arg(4) ;quant_ptr
-
- movdqa xmm0, [rdx]
- movdqa xmm4, [rdx + 16]
-
- movdqa xmm2, [rdi] ;round lo
- movdqa xmm3, [rdi + 16] ;round hi
-
- movdqa xmm1, xmm0
- movdqa xmm5, xmm4
-
- psraw xmm0, 15 ;sign of z (aka sz)
- psraw xmm4, 15 ;sign of z (aka sz)
-
- pabsw xmm1, xmm1
- pabsw xmm5, xmm5
-
- paddw xmm1, xmm2
- paddw xmm5, xmm3
-
- pmulhw xmm1, [rsi]
- pmulhw xmm5, [rsi + 16]
-
- mov rdi, arg(1) ;qcoeff_ptr
- mov rcx, arg(2) ;dequant_ptr
- mov rsi, arg(5) ;dqcoeff_ptr
-
- pxor xmm1, xmm0
- pxor xmm5, xmm4
- psubw xmm1, xmm0
- psubw xmm5, xmm4
-
- movdqa [rdi], xmm1
- movdqa [rdi + 16], xmm5
-
- movdqa xmm2, [rcx]
- movdqa xmm3, [rcx + 16]
-
- pxor xmm4, xmm4
- pmullw xmm2, xmm1
- pmullw xmm3, xmm5
-
- pcmpeqw xmm1, xmm4 ;non zero mask
- pcmpeqw xmm5, xmm4 ;non zero mask
- packsswb xmm1, xmm5
- pshufb xmm1, [ GLOBAL(zz_shuf)]
-
- pmovmskb edx, xmm1
-
-; xor ecx, ecx
-; mov eax, -1
-;find_eob_loop:
-; shr edx, 1
-; jc fq_skip
-; mov eax, ecx
-;fq_skip:
-; inc ecx
-; cmp ecx, 16
-; jne find_eob_loop
- xor rdi, rdi
- mov eax, -1
- xor dx, ax ;flip the bits for bsr
- bsr eax, edx
-
- movdqa [rsi], xmm2 ;store dqcoeff
- movdqa [rsi + 16], xmm3 ;store dqcoeff
-
- sub edi, edx ;check for all zeros in bit mask
- sar edi, 31 ;0 or -1
- add eax, 1
- and eax, edi ;if the bit mask was all zero,
- ;then eob = 0
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-zz_shuf:
- db 0, 1, 4, 8, 5, 2, 3, 6, 9, 12, 13, 10, 7, 11, 14, 15
diff --git a/vp8/encoder/x86/quantize_x86.h b/vp8/encoder/x86/quantize_x86.h
deleted file mode 100644
index 6f54bec..0000000
--- a/vp8/encoder/x86/quantize_x86.h
+++ /dev/null
@@ -1,38 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license and patent
- * grant that can be found in the LICENSE file in the root of the source
- * tree. All contributing project authors may be found in the AUTHORS
- * file in the root of the source tree.
- */
-
-#ifndef QUANTIZE_X86_H
-#define QUANTIZE_X86_H
-
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-#if HAVE_MMX
-
-#endif
-
-
-#if HAVE_SSE2
-extern prototype_quantize_block(vp8_regular_quantize_b_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_quantize_quantb
-#define vp8_quantize_quantb vp8_regular_quantize_b_sse2
-
-#endif
-
-#endif
-
-
-#endif
diff --git a/vp8/encoder/x86/sad_mmx.asm b/vp8/encoder/x86/sad_mmx.asm
deleted file mode 100644
index 85cb023..0000000
--- a/vp8/encoder/x86/sad_mmx.asm
+++ /dev/null
@@ -1,427 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-global sym(vp8_sad16x16_mmx)
-global sym(vp8_sad8x16_mmx)
-global sym(vp8_sad8x8_mmx)
-global sym(vp8_sad4x4_mmx)
-global sym(vp8_sad16x8_mmx)
-
-;unsigned int vp8_sad16x16_mmx(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-sym(vp8_sad16x16_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- lea rcx, [rsi+rax*8]
-
- lea rcx, [rcx+rax*8]
- pxor mm7, mm7
-
- pxor mm6, mm6
-
-x16x16sad_mmx_loop:
-
- movq mm0, QWORD PTR [rsi]
- movq mm2, QWORD PTR [rsi+8]
-
- movq mm1, QWORD PTR [rdi]
- movq mm3, QWORD PTR [rdi+8]
-
- movq mm4, mm0
- movq mm5, mm2
-
- psubusb mm0, mm1
- psubusb mm1, mm4
-
- psubusb mm2, mm3
- psubusb mm3, mm5
-
- por mm0, mm1
- por mm2, mm3
-
- movq mm1, mm0
- movq mm3, mm2
-
- punpcklbw mm0, mm6
- punpcklbw mm2, mm6
-
- punpckhbw mm1, mm6
- punpckhbw mm3, mm6
-
- paddw mm0, mm2
- paddw mm1, mm3
-
-
- lea rsi, [rsi+rax]
- add rdi, rdx
-
- paddw mm7, mm0
- paddw mm7, mm1
-
- cmp rsi, rcx
- jne x16x16sad_mmx_loop
-
-
- movq mm0, mm7
-
- punpcklwd mm0, mm6
- punpckhwd mm7, mm6
-
- paddw mm0, mm7
- movq mm7, mm0
-
-
- psrlq mm0, 32
- paddw mm7, mm0
-
- movq rax, mm7
-
- pop rdi
- pop rsi
- mov rsp, rbp
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_sad8x16_mmx(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-sym(vp8_sad8x16_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- lea rcx, [rsi+rax*8]
-
- lea rcx, [rcx+rax*8]
- pxor mm7, mm7
-
- pxor mm6, mm6
-
-x8x16sad_mmx_loop:
-
- movq mm0, QWORD PTR [rsi]
- movq mm1, QWORD PTR [rdi]
-
- movq mm2, mm0
- psubusb mm0, mm1
-
- psubusb mm1, mm2
- por mm0, mm1
-
- movq mm2, mm0
- punpcklbw mm0, mm6
-
- punpckhbw mm2, mm6
- lea rsi, [rsi+rax]
-
- add rdi, rdx
- paddw mm7, mm0
-
- paddw mm7, mm2
- cmp rsi, rcx
-
- jne x8x16sad_mmx_loop
-
- movq mm0, mm7
- punpcklwd mm0, mm6
-
- punpckhwd mm7, mm6
- paddw mm0, mm7
-
- movq mm7, mm0
- psrlq mm0, 32
-
- paddw mm7, mm0
- movq rax, mm7
-
- pop rdi
- pop rsi
- mov rsp, rbp
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_sad8x8_mmx(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-sym(vp8_sad8x8_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- lea rcx, [rsi+rax*8]
- pxor mm7, mm7
-
- pxor mm6, mm6
-
-x8x8sad_mmx_loop:
-
- movq mm0, QWORD PTR [rsi]
- movq mm1, QWORD PTR [rdi]
-
- movq mm2, mm0
- psubusb mm0, mm1
-
- psubusb mm1, mm2
- por mm0, mm1
-
- movq mm2, mm0
- punpcklbw mm0, mm6
-
- punpckhbw mm2, mm6
- paddw mm0, mm2
-
- lea rsi, [rsi+rax]
- add rdi, rdx
-
- paddw mm7, mm0
- cmp rsi, rcx
-
- jne x8x8sad_mmx_loop
-
- movq mm0, mm7
- punpcklwd mm0, mm6
-
- punpckhwd mm7, mm6
- paddw mm0, mm7
-
- movq mm7, mm0
- psrlq mm0, 32
-
- paddw mm7, mm0
- movq rax, mm7
-
- pop rdi
- pop rsi
- mov rsp, rbp
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_sad4x4_mmx(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-sym(vp8_sad4x4_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- movd mm0, DWORD PTR [rsi]
- movd mm1, DWORD PTR [rdi]
-
- movd mm2, DWORD PTR [rsi+rax]
- movd mm3, DWORD PTR [rdi+rdx]
-
- punpcklbw mm0, mm2
- punpcklbw mm1, mm3
-
- movq mm2, mm0
- psubusb mm0, mm1
-
- psubusb mm1, mm2
- por mm0, mm1
-
- movq mm2, mm0
- pxor mm3, mm3
-
- punpcklbw mm0, mm3
- punpckhbw mm2, mm3
-
- paddw mm0, mm2
-
- lea rsi, [rsi+rax*2]
- lea rdi, [rdi+rdx*2]
-
- movd mm4, DWORD PTR [rsi]
- movd mm5, DWORD PTR [rdi]
-
- movd mm6, DWORD PTR [rsi+rax]
- movd mm7, DWORD PTR [rdi+rdx]
-
- punpcklbw mm4, mm6
- punpcklbw mm5, mm7
-
- movq mm6, mm4
- psubusb mm4, mm5
-
- psubusb mm5, mm6
- por mm4, mm5
-
- movq mm5, mm4
- punpcklbw mm4, mm3
-
- punpckhbw mm5, mm3
- paddw mm4, mm5
-
- paddw mm0, mm4
- movq mm1, mm0
-
- punpcklwd mm0, mm3
- punpckhwd mm1, mm3
-
- paddw mm0, mm1
- movq mm1, mm0
-
- psrlq mm0, 32
- paddw mm0, mm1
-
- movq rax, mm0
-
- pop rdi
- pop rsi
- mov rsp, rbp
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_sad16x8_mmx(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-sym(vp8_sad16x8_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- lea rcx, [rsi+rax*8]
- pxor mm7, mm7
-
- pxor mm6, mm6
-
-x16x8sad_mmx_loop:
-
- movq mm0, [rsi]
- movq mm1, [rdi]
-
- movq mm2, [rsi+8]
- movq mm3, [rdi+8]
-
- movq mm4, mm0
- movq mm5, mm2
-
- psubusb mm0, mm1
- psubusb mm1, mm4
-
- psubusb mm2, mm3
- psubusb mm3, mm5
-
- por mm0, mm1
- por mm2, mm3
-
- movq mm1, mm0
- movq mm3, mm2
-
- punpcklbw mm0, mm6
- punpckhbw mm1, mm6
-
- punpcklbw mm2, mm6
- punpckhbw mm3, mm6
-
-
- paddw mm0, mm2
- paddw mm1, mm3
-
- paddw mm0, mm1
- lea rsi, [rsi+rax]
-
- add rdi, rdx
- paddw mm7, mm0
-
- cmp rsi, rcx
- jne x16x8sad_mmx_loop
-
- movq mm0, mm7
- punpcklwd mm0, mm6
-
- punpckhwd mm7, mm6
- paddw mm0, mm7
-
- movq mm7, mm0
- psrlq mm0, 32
-
- paddw mm7, mm0
- movq rax, mm7
-
- pop rdi
- pop rsi
- mov rsp, rbp
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/encoder/x86/sad_sse2.asm b/vp8/encoder/x86/sad_sse2.asm
deleted file mode 100644
index 39ed796..0000000
--- a/vp8/encoder/x86/sad_sse2.asm
+++ /dev/null
@@ -1,328 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;unsigned int vp8_sad16x16_wmt(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-global sym(vp8_sad16x16_wmt)
-sym(vp8_sad16x16_wmt):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- lea rcx, [rsi+rax*8]
-
- lea rcx, [rcx+rax*8]
- pxor xmm7, xmm7
-
-x16x16sad_wmt_loop:
-
- movq xmm0, QWORD PTR [rsi]
- movq xmm2, QWORD PTR [rsi+8]
-
- movq xmm1, QWORD PTR [rdi]
- movq xmm3, QWORD PTR [rdi+8]
-
- movq xmm4, QWORD PTR [rsi+rax]
- movq xmm5, QWORD PTR [rdi+rdx]
-
-
- punpcklbw xmm0, xmm2
- punpcklbw xmm1, xmm3
-
- psadbw xmm0, xmm1
- movq xmm6, QWORD PTR [rsi+rax+8]
-
- movq xmm3, QWORD PTR [rdi+rdx+8]
- lea rsi, [rsi+rax*2]
-
- lea rdi, [rdi+rdx*2]
- punpcklbw xmm4, xmm6
-
- punpcklbw xmm5, xmm3
- psadbw xmm4, xmm5
-
- paddw xmm7, xmm0
- paddw xmm7, xmm4
-
- cmp rsi, rcx
- jne x16x16sad_wmt_loop
-
- movq xmm0, xmm7
- psrldq xmm7, 8
-
- paddw xmm0, xmm7
- movq rax, xmm0
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;unsigned int vp8_sad8x16_wmt(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int max_err)
-global sym(vp8_sad8x16_wmt)
-sym(vp8_sad8x16_wmt):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rbx, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- lea rcx, [rsi+rbx*8]
-
- lea rcx, [rcx+rbx*8]
- pxor mm7, mm7
-
-x8x16sad_wmt_loop:
-
- movq rax, mm7
- cmp rax, arg(4)
- jg x8x16sad_wmt_early_exit
-
- movq mm0, QWORD PTR [rsi]
- movq mm1, QWORD PTR [rdi]
-
- movq mm2, QWORD PTR [rsi+rbx]
- movq mm3, QWORD PTR [rdi+rdx]
-
- psadbw mm0, mm1
- psadbw mm2, mm3
-
- lea rsi, [rsi+rbx*2]
- lea rdi, [rdi+rdx*2]
-
- paddw mm7, mm0
- paddw mm7, mm2
-
- cmp rsi, rcx
- jne x8x16sad_wmt_loop
-
- movq rax, mm7
-
-x8x16sad_wmt_early_exit:
-
- ; begin epilog
- pop rdi
- pop rsi
- pop rbx
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_sad8x8_wmt(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-global sym(vp8_sad8x8_wmt)
-sym(vp8_sad8x8_wmt):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rbx, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- lea rcx, [rsi+rbx*8]
- pxor mm7, mm7
-
-x8x8sad_wmt_loop:
-
- movq rax, mm7
- cmp rax, arg(4)
- jg x8x8sad_wmt_early_exit
-
- movq mm0, QWORD PTR [rsi]
- movq mm1, QWORD PTR [rdi]
-
- psadbw mm0, mm1
- lea rsi, [rsi+rbx]
-
- add rdi, rdx
- paddw mm7, mm0
-
- cmp rsi, rcx
- jne x8x8sad_wmt_loop
-
- movq rax, mm7
-x8x8sad_wmt_early_exit:
-
- ; begin epilog
- pop rdi
- pop rsi
- pop rbx
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;unsigned int vp8_sad4x4_wmt(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-global sym(vp8_sad4x4_wmt)
-sym(vp8_sad4x4_wmt):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- movd mm0, DWORD PTR [rsi]
- movd mm1, DWORD PTR [rdi]
-
- movd mm2, DWORD PTR [rsi+rax]
- movd mm3, DWORD PTR [rdi+rdx]
-
- punpcklbw mm0, mm2
- punpcklbw mm1, mm3
-
- psadbw mm0, mm1
- lea rsi, [rsi+rax*2]
-
- lea rdi, [rdi+rdx*2]
- movd mm4, DWORD PTR [rsi]
-
- movd mm5, DWORD PTR [rdi]
- movd mm6, DWORD PTR [rsi+rax]
-
- movd mm7, DWORD PTR [rdi+rdx]
- punpcklbw mm4, mm6
-
- punpcklbw mm5, mm7
- psadbw mm4, mm5
-
- paddw mm0, mm4
- movq rax, mm0
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_sad16x8_wmt(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride)
-global sym(vp8_sad16x8_wmt)
-sym(vp8_sad16x8_wmt):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rbx
- push rsi
- push rdi
- ; end prolog
-
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rbx, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- lea rcx, [rsi+rbx*8]
- pxor mm7, mm7
-
-x16x8sad_wmt_loop:
-
- movq rax, mm7
- cmp rax, arg(4)
- jg x16x8sad_wmt_early_exit
-
- movq mm0, QWORD PTR [rsi]
- movq mm2, QWORD PTR [rsi+8]
-
- movq mm1, QWORD PTR [rdi]
- movq mm3, QWORD PTR [rdi+8]
-
- movq mm4, QWORD PTR [rsi+rbx]
- movq mm5, QWORD PTR [rdi+rdx]
-
- psadbw mm0, mm1
- psadbw mm2, mm3
-
- movq mm1, QWORD PTR [rsi+rbx+8]
- movq mm3, QWORD PTR [rdi+rdx+8]
-
- psadbw mm4, mm5
- psadbw mm1, mm3
-
- lea rsi, [rsi+rbx*2]
- lea rdi, [rdi+rdx*2]
-
- paddw mm0, mm2
- paddw mm4, mm1
-
- paddw mm7, mm0
- paddw mm7, mm4
-
- cmp rsi, rcx
- jne x16x8sad_wmt_loop
-
- movq rax, mm7
-
-x16x8sad_wmt_early_exit:
-
- ; begin epilog
- pop rdi
- pop rsi
- pop rbx
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/encoder/x86/sad_sse3.asm b/vp8/encoder/x86/sad_sse3.asm
deleted file mode 100644
index f0336ab..0000000
--- a/vp8/encoder/x86/sad_sse3.asm
+++ /dev/null
@@ -1,890 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%macro STACK_FRAME_CREATE_X3 0
-%if ABI_IS_32BIT
- %define src_ptr rsi
- %define src_stride rax
- %define ref_ptr rdi
- %define ref_stride rdx
- %define end_ptr rcx
- %define ret_var rbx
- %define result_ptr arg(4)
- %define max_err arg(4)
- push rbp
- mov rbp, rsp
- push rsi
- push rdi
- push rbx
-
- mov rsi, arg(0) ; src_ptr
- mov rdi, arg(2) ; ref_ptr
-
- movsxd rax, dword ptr arg(1) ; src_stride
- movsxd rdx, dword ptr arg(3) ; ref_stride
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- %define src_ptr rcx
- %define src_stride rdx
- %define ref_ptr r8
- %define ref_stride r9
- %define end_ptr r10
- %define ret_var r11
- %define result_ptr [rsp+8+4*8]
- %define max_err [rsp+8+4*8]
- %else
- %define src_ptr rdi
- %define src_stride rsi
- %define ref_ptr rdx
- %define ref_stride rcx
- %define end_ptr r9
- %define ret_var r10
- %define result_ptr r8
- %define max_err r8
- %endif
-%endif
-
-%endmacro
-
-%macro STACK_FRAME_DESTROY_X3 0
- %define src_ptr
- %define src_stride
- %define ref_ptr
- %define ref_stride
- %define end_ptr
- %define ret_var
- %define result_ptr
- %define max_err
-
-%if ABI_IS_32BIT
- pop rbx
- pop rdi
- pop rsi
- pop rbp
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- %endif
-%endif
- ret
-%endmacro
-
-%macro STACK_FRAME_CREATE_X4 0
-%if ABI_IS_32BIT
- %define src_ptr rsi
- %define src_stride rax
- %define r0_ptr rcx
- %define r1_ptr rdx
- %define r2_ptr rbx
- %define r3_ptr rdi
- %define ref_stride rbp
- %define result_ptr arg(4)
- push rbp
- mov rbp, rsp
- push rsi
- push rdi
- push rbx
-
- push rbp
- mov rdi, arg(2) ; ref_ptr_base
-
- LOAD_X4_ADDRESSES rdi, rcx, rdx, rax, rdi
-
- mov rsi, arg(0) ; src_ptr
-
- movsxd rbx, dword ptr arg(1) ; src_stride
- movsxd rbp, dword ptr arg(3) ; ref_stride
-
- xchg rbx, rax
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- %define src_ptr rcx
- %define src_stride rdx
- %define r0_ptr rsi
- %define r1_ptr r10
- %define r2_ptr r11
- %define r3_ptr r8
- %define ref_stride r9
- %define result_ptr [rsp+16+4*8]
- push rsi
-
- LOAD_X4_ADDRESSES r8, r0_ptr, r1_ptr, r2_ptr, r3_ptr
- %else
- %define src_ptr rdi
- %define src_stride rsi
- %define r0_ptr r9
- %define r1_ptr r10
- %define r2_ptr r11
- %define r3_ptr rdx
- %define ref_stride rcx
- %define result_ptr r8
-
- LOAD_X4_ADDRESSES rdx, r0_ptr, r1_ptr, r2_ptr, r3_ptr
-
- %endif
-%endif
-%endmacro
-
-%macro STACK_FRAME_DESTROY_X4 0
- %define src_ptr
- %define src_stride
- %define r0_ptr
- %define r1_ptr
- %define r2_ptr
- %define r3_ptr
- %define ref_stride
- %define result_ptr
-
-%if ABI_IS_32BIT
- pop rbx
- pop rdi
- pop rsi
- pop rbp
-%else
- %ifidn __OUTPUT_FORMAT__,x64
- pop rsi
- %endif
-%endif
- ret
-%endmacro
-
-%macro PROCESS_16X2X3 5
-%if %1==0
- movdqa xmm0, XMMWORD PTR [%2]
- lddqu xmm5, XMMWORD PTR [%3]
- lddqu xmm6, XMMWORD PTR [%3+1]
- lddqu xmm7, XMMWORD PTR [%3+2]
-
- psadbw xmm5, xmm0
- psadbw xmm6, xmm0
- psadbw xmm7, xmm0
-%else
- movdqa xmm0, XMMWORD PTR [%2]
- lddqu xmm1, XMMWORD PTR [%3]
- lddqu xmm2, XMMWORD PTR [%3+1]
- lddqu xmm3, XMMWORD PTR [%3+2]
-
- psadbw xmm1, xmm0
- psadbw xmm2, xmm0
- psadbw xmm3, xmm0
-
- paddw xmm5, xmm1
- paddw xmm6, xmm2
- paddw xmm7, xmm3
-%endif
- movdqa xmm0, XMMWORD PTR [%2+%4]
- lddqu xmm1, XMMWORD PTR [%3+%5]
- lddqu xmm2, XMMWORD PTR [%3+%5+1]
- lddqu xmm3, XMMWORD PTR [%3+%5+2]
-
-%if %1==0 || %1==1
- lea %2, [%2+%4*2]
- lea %3, [%3+%5*2]
-%endif
-
- psadbw xmm1, xmm0
- psadbw xmm2, xmm0
- psadbw xmm3, xmm0
-
- paddw xmm5, xmm1
- paddw xmm6, xmm2
- paddw xmm7, xmm3
-%endmacro
-
-%macro PROCESS_8X2X3 5
-%if %1==0
- movq mm0, QWORD PTR [%2]
- movq mm5, QWORD PTR [%3]
- movq mm6, QWORD PTR [%3+1]
- movq mm7, QWORD PTR [%3+2]
-
- psadbw mm5, mm0
- psadbw mm6, mm0
- psadbw mm7, mm0
-%else
- movq mm0, QWORD PTR [%2]
- movq mm1, QWORD PTR [%3]
- movq mm2, QWORD PTR [%3+1]
- movq mm3, QWORD PTR [%3+2]
-
- psadbw mm1, mm0
- psadbw mm2, mm0
- psadbw mm3, mm0
-
- paddw mm5, mm1
- paddw mm6, mm2
- paddw mm7, mm3
-%endif
- movq mm0, QWORD PTR [%2+%4]
- movq mm1, QWORD PTR [%3+%5]
- movq mm2, QWORD PTR [%3+%5+1]
- movq mm3, QWORD PTR [%3+%5+2]
-
-%if %1==0 || %1==1
- lea %2, [%2+%4*2]
- lea %3, [%3+%5*2]
-%endif
-
- psadbw mm1, mm0
- psadbw mm2, mm0
- psadbw mm3, mm0
-
- paddw mm5, mm1
- paddw mm6, mm2
- paddw mm7, mm3
-%endmacro
-
-%macro LOAD_X4_ADDRESSES 5
- mov %2, [%1+REG_SZ_BYTES*0]
- mov %3, [%1+REG_SZ_BYTES*1]
-
- mov %4, [%1+REG_SZ_BYTES*2]
- mov %5, [%1+REG_SZ_BYTES*3]
-%endmacro
-
-%macro PROCESS_16X2X4 8
-%if %1==0
- movdqa xmm0, XMMWORD PTR [%2]
- lddqu xmm4, XMMWORD PTR [%3]
- lddqu xmm5, XMMWORD PTR [%4]
- lddqu xmm6, XMMWORD PTR [%5]
- lddqu xmm7, XMMWORD PTR [%6]
-
- psadbw xmm4, xmm0
- psadbw xmm5, xmm0
- psadbw xmm6, xmm0
- psadbw xmm7, xmm0
-%else
- movdqa xmm0, XMMWORD PTR [%2]
- lddqu xmm1, XMMWORD PTR [%3]
- lddqu xmm2, XMMWORD PTR [%4]
- lddqu xmm3, XMMWORD PTR [%5]
-
- psadbw xmm1, xmm0
- psadbw xmm2, xmm0
- psadbw xmm3, xmm0
-
- paddw xmm4, xmm1
- lddqu xmm1, XMMWORD PTR [%6]
- paddw xmm5, xmm2
- paddw xmm6, xmm3
-
- psadbw xmm1, xmm0
- paddw xmm7, xmm1
-%endif
- movdqa xmm0, XMMWORD PTR [%2+%7]
- lddqu xmm1, XMMWORD PTR [%3+%8]
- lddqu xmm2, XMMWORD PTR [%4+%8]
- lddqu xmm3, XMMWORD PTR [%5+%8]
-
- psadbw xmm1, xmm0
- psadbw xmm2, xmm0
- psadbw xmm3, xmm0
-
- paddw xmm4, xmm1
- lddqu xmm1, XMMWORD PTR [%6+%8]
- paddw xmm5, xmm2
- paddw xmm6, xmm3
-
-%if %1==0 || %1==1
- lea %2, [%2+%7*2]
- lea %3, [%3+%8*2]
-
- lea %4, [%4+%8*2]
- lea %5, [%5+%8*2]
-
- lea %6, [%6+%8*2]
-%endif
- psadbw xmm1, xmm0
- paddw xmm7, xmm1
-
-%endmacro
-
-%macro PROCESS_8X2X4 8
-%if %1==0
- movq mm0, QWORD PTR [%2]
- movq mm4, QWORD PTR [%3]
- movq mm5, QWORD PTR [%4]
- movq mm6, QWORD PTR [%5]
- movq mm7, QWORD PTR [%6]
-
- psadbw mm4, mm0
- psadbw mm5, mm0
- psadbw mm6, mm0
- psadbw mm7, mm0
-%else
- movq mm0, QWORD PTR [%2]
- movq mm1, QWORD PTR [%3]
- movq mm2, QWORD PTR [%4]
- movq mm3, QWORD PTR [%5]
-
- psadbw mm1, mm0
- psadbw mm2, mm0
- psadbw mm3, mm0
-
- paddw mm4, mm1
- movq mm1, QWORD PTR [%6]
- paddw mm5, mm2
- paddw mm6, mm3
-
- psadbw mm1, mm0
- paddw mm7, mm1
-%endif
- movq mm0, QWORD PTR [%2+%7]
- movq mm1, QWORD PTR [%3+%8]
- movq mm2, QWORD PTR [%4+%8]
- movq mm3, QWORD PTR [%5+%8]
-
- psadbw mm1, mm0
- psadbw mm2, mm0
- psadbw mm3, mm0
-
- paddw mm4, mm1
- movq mm1, QWORD PTR [%6+%8]
- paddw mm5, mm2
- paddw mm6, mm3
-
-%if %1==0 || %1==1
- lea %2, [%2+%7*2]
- lea %3, [%3+%8*2]
-
- lea %4, [%4+%8*2]
- lea %5, [%5+%8*2]
-
- lea %6, [%6+%8*2]
-%endif
- psadbw mm1, mm0
- paddw mm7, mm1
-
-%endmacro
-
-;void int vp8_sad16x16x3_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad16x16x3_sse3)
-sym(vp8_sad16x16x3_sse3):
-
- STACK_FRAME_CREATE_X3
-
- PROCESS_16X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride
-
- mov rcx, result_ptr
-
- movq xmm0, xmm5
- psrldq xmm5, 8
-
- paddw xmm0, xmm5
- movd [rcx], xmm0
-;-
- movq xmm0, xmm6
- psrldq xmm6, 8
-
- paddw xmm0, xmm6
- movd [rcx+4], xmm0
-;-
- movq xmm0, xmm7
- psrldq xmm7, 8
-
- paddw xmm0, xmm7
- movd [rcx+8], xmm0
-
- STACK_FRAME_DESTROY_X3
-
-;void int vp8_sad16x8x3_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad16x8x3_sse3)
-sym(vp8_sad16x8x3_sse3):
-
- STACK_FRAME_CREATE_X3
-
- PROCESS_16X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_16X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride
-
- mov rcx, result_ptr
-
- movq xmm0, xmm5
- psrldq xmm5, 8
-
- paddw xmm0, xmm5
- movd [rcx], xmm0
-;-
- movq xmm0, xmm6
- psrldq xmm6, 8
-
- paddw xmm0, xmm6
- movd [rcx+4], xmm0
-;-
- movq xmm0, xmm7
- psrldq xmm7, 8
-
- paddw xmm0, xmm7
- movd [rcx+8], xmm0
-
- STACK_FRAME_DESTROY_X3
-
-;void int vp8_sad8x16x3_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad8x16x3_sse3)
-sym(vp8_sad8x16x3_sse3):
-
- STACK_FRAME_CREATE_X3
-
- PROCESS_8X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride
-
- mov rcx, result_ptr
-
- punpckldq mm5, mm6
-
- movq [rcx], mm5
- movd [rcx+8], mm7
-
- STACK_FRAME_DESTROY_X3
-
-;void int vp8_sad8x8x3_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad8x8x3_sse3)
-sym(vp8_sad8x8x3_sse3):
-
- STACK_FRAME_CREATE_X3
-
- PROCESS_8X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride
- PROCESS_8X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride
-
- mov rcx, result_ptr
-
- punpckldq mm5, mm6
-
- movq [rcx], mm5
- movd [rcx+8], mm7
-
- STACK_FRAME_DESTROY_X3
-
-;void int vp8_sad4x4x3_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad4x4x3_sse3)
-sym(vp8_sad4x4x3_sse3):
-
- STACK_FRAME_CREATE_X3
-
- movd mm0, DWORD PTR [src_ptr]
- movd mm1, DWORD PTR [ref_ptr]
-
- movd mm2, DWORD PTR [src_ptr+src_stride]
- movd mm3, DWORD PTR [ref_ptr+ref_stride]
-
- punpcklbw mm0, mm2
- punpcklbw mm1, mm3
-
- movd mm4, DWORD PTR [ref_ptr+1]
- movd mm5, DWORD PTR [ref_ptr+2]
-
- movd mm2, DWORD PTR [ref_ptr+ref_stride+1]
- movd mm3, DWORD PTR [ref_ptr+ref_stride+2]
-
- psadbw mm1, mm0
-
- punpcklbw mm4, mm2
- punpcklbw mm5, mm3
-
- psadbw mm4, mm0
- psadbw mm5, mm0
-
- lea src_ptr, [src_ptr+src_stride*2]
- lea ref_ptr, [ref_ptr+ref_stride*2]
-
- movd mm0, DWORD PTR [src_ptr]
- movd mm2, DWORD PTR [ref_ptr]
-
- movd mm3, DWORD PTR [src_ptr+src_stride]
- movd mm6, DWORD PTR [ref_ptr+ref_stride]
-
- punpcklbw mm0, mm3
- punpcklbw mm2, mm6
-
- movd mm3, DWORD PTR [ref_ptr+1]
- movd mm7, DWORD PTR [ref_ptr+2]
-
- psadbw mm2, mm0
-
- paddw mm1, mm2
-
- movd mm2, DWORD PTR [ref_ptr+ref_stride+1]
- movd mm6, DWORD PTR [ref_ptr+ref_stride+2]
-
- punpcklbw mm3, mm2
- punpcklbw mm7, mm6
-
- psadbw mm3, mm0
- psadbw mm7, mm0
-
- paddw mm3, mm4
- paddw mm7, mm5
-
- mov rcx, result_ptr
-
- punpckldq mm1, mm3
-
- movq [rcx], mm1
- movd [rcx+8], mm7
-
- STACK_FRAME_DESTROY_X3
-
-;unsigned int vp8_sad16x16_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int max_err)
-;%define lddqu movdqu
-global sym(vp8_sad16x16_sse3)
-sym(vp8_sad16x16_sse3):
-
- STACK_FRAME_CREATE_X3
-
- mov end_ptr, 4
- pxor xmm7, xmm7
-
-.vp8_sad16x16_sse3_loop:
- movdqa xmm0, XMMWORD PTR [src_ptr]
- movdqu xmm1, XMMWORD PTR [ref_ptr]
- movdqa xmm2, XMMWORD PTR [src_ptr+src_stride]
- movdqu xmm3, XMMWORD PTR [ref_ptr+ref_stride]
-
- lea src_ptr, [src_ptr+src_stride*2]
- lea ref_ptr, [ref_ptr+ref_stride*2]
-
- movdqa xmm4, XMMWORD PTR [src_ptr]
- movdqu xmm5, XMMWORD PTR [ref_ptr]
- movdqa xmm6, XMMWORD PTR [src_ptr+src_stride]
-
- psadbw xmm0, xmm1
-
- movdqu xmm1, XMMWORD PTR [ref_ptr+ref_stride]
-
- psadbw xmm2, xmm3
- psadbw xmm4, xmm5
- psadbw xmm6, xmm1
-
- lea src_ptr, [src_ptr+src_stride*2]
- lea ref_ptr, [ref_ptr+ref_stride*2]
-
- paddw xmm7, xmm0
- paddw xmm7, xmm2
- paddw xmm7, xmm4
- paddw xmm7, xmm6
-
- sub end_ptr, 1
- jne .vp8_sad16x16_sse3_loop
-
- movq xmm0, xmm7
- psrldq xmm7, 8
- paddw xmm0, xmm7
- movq rax, xmm0
-
- STACK_FRAME_DESTROY_X3
-
-;void vp8_sad16x16x4d_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr_base,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad16x16x4d_sse3)
-sym(vp8_sad16x16x4d_sse3):
-
- STACK_FRAME_CREATE_X4
-
- PROCESS_16X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
-
-%if ABI_IS_32BIT
- pop rbp
-%endif
- mov rcx, result_ptr
-
- movq xmm0, xmm4
- psrldq xmm4, 8
-
- paddw xmm0, xmm4
- movd [rcx], xmm0
-;-
- movq xmm0, xmm5
- psrldq xmm5, 8
-
- paddw xmm0, xmm5
- movd [rcx+4], xmm0
-;-
- movq xmm0, xmm6
- psrldq xmm6, 8
-
- paddw xmm0, xmm6
- movd [rcx+8], xmm0
-;-
- movq xmm0, xmm7
- psrldq xmm7, 8
-
- paddw xmm0, xmm7
- movd [rcx+12], xmm0
-
- STACK_FRAME_DESTROY_X4
-
-;void vp8_sad16x8x4d_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr_base,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad16x8x4d_sse3)
-sym(vp8_sad16x8x4d_sse3):
-
- STACK_FRAME_CREATE_X4
-
- PROCESS_16X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_16X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
-
-%if ABI_IS_32BIT
- pop rbp
-%endif
- mov rcx, result_ptr
-
- movq xmm0, xmm4
- psrldq xmm4, 8
-
- paddw xmm0, xmm4
- movd [rcx], xmm0
-;-
- movq xmm0, xmm5
- psrldq xmm5, 8
-
- paddw xmm0, xmm5
- movd [rcx+4], xmm0
-;-
- movq xmm0, xmm6
- psrldq xmm6, 8
-
- paddw xmm0, xmm6
- movd [rcx+8], xmm0
-;-
- movq xmm0, xmm7
- psrldq xmm7, 8
-
- paddw xmm0, xmm7
- movd [rcx+12], xmm0
-
- STACK_FRAME_DESTROY_X4
-
-;void int vp8_sad8x16x4d_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad8x16x4d_sse3)
-sym(vp8_sad8x16x4d_sse3):
-
- STACK_FRAME_CREATE_X4
-
- PROCESS_8X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
-
-%if ABI_IS_32BIT
- pop rbp
-%endif
- mov rcx, result_ptr
-
- punpckldq mm4, mm5
- punpckldq mm6, mm7
-
- movq [rcx], mm4
- movq [rcx+8], mm6
-
- STACK_FRAME_DESTROY_X4
-
-;void int vp8_sad8x8x4d_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad8x8x4d_sse3)
-sym(vp8_sad8x8x4d_sse3):
-
- STACK_FRAME_CREATE_X4
-
- PROCESS_8X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
- PROCESS_8X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
-
-%if ABI_IS_32BIT
- pop rbp
-%endif
- mov rcx, result_ptr
-
- punpckldq mm4, mm5
- punpckldq mm6, mm7
-
- movq [rcx], mm4
- movq [rcx+8], mm6
-
- STACK_FRAME_DESTROY_X4
-
-;void int vp8_sad4x4x4d_sse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad4x4x4d_sse3)
-sym(vp8_sad4x4x4d_sse3):
-
- STACK_FRAME_CREATE_X4
-
- movd mm0, DWORD PTR [src_ptr]
- movd mm1, DWORD PTR [r0_ptr]
-
- movd mm2, DWORD PTR [src_ptr+src_stride]
- movd mm3, DWORD PTR [r0_ptr+ref_stride]
-
- punpcklbw mm0, mm2
- punpcklbw mm1, mm3
-
- movd mm4, DWORD PTR [r1_ptr]
- movd mm5, DWORD PTR [r2_ptr]
-
- movd mm6, DWORD PTR [r3_ptr]
- movd mm2, DWORD PTR [r1_ptr+ref_stride]
-
- movd mm3, DWORD PTR [r2_ptr+ref_stride]
- movd mm7, DWORD PTR [r3_ptr+ref_stride]
-
- psadbw mm1, mm0
-
- punpcklbw mm4, mm2
- punpcklbw mm5, mm3
-
- punpcklbw mm6, mm7
- psadbw mm4, mm0
-
- psadbw mm5, mm0
- psadbw mm6, mm0
-
-
-
- lea src_ptr, [src_ptr+src_stride*2]
- lea r0_ptr, [r0_ptr+ref_stride*2]
-
- lea r1_ptr, [r1_ptr+ref_stride*2]
- lea r2_ptr, [r2_ptr+ref_stride*2]
-
- lea r3_ptr, [r3_ptr+ref_stride*2]
-
- movd mm0, DWORD PTR [src_ptr]
- movd mm2, DWORD PTR [r0_ptr]
-
- movd mm3, DWORD PTR [src_ptr+src_stride]
- movd mm7, DWORD PTR [r0_ptr+ref_stride]
-
- punpcklbw mm0, mm3
- punpcklbw mm2, mm7
-
- movd mm3, DWORD PTR [r1_ptr]
- movd mm7, DWORD PTR [r2_ptr]
-
- psadbw mm2, mm0
-%if ABI_IS_32BIT
- mov rax, rbp
-
- pop rbp
-%define ref_stride rax
-%endif
- mov rsi, result_ptr
-
- paddw mm1, mm2
- movd [rsi], mm1
-
- movd mm2, DWORD PTR [r1_ptr+ref_stride]
- movd mm1, DWORD PTR [r2_ptr+ref_stride]
-
- punpcklbw mm3, mm2
- punpcklbw mm7, mm1
-
- psadbw mm3, mm0
- psadbw mm7, mm0
-
- movd mm2, DWORD PTR [r3_ptr]
- movd mm1, DWORD PTR [r3_ptr+ref_stride]
-
- paddw mm3, mm4
- paddw mm7, mm5
-
- movd [rsi+4], mm3
- punpcklbw mm2, mm1
-
- movd [rsi+8], mm7
- psadbw mm2, mm0
-
- paddw mm2, mm6
- movd [rsi+12], mm2
-
-
- STACK_FRAME_DESTROY_X4
diff --git a/vp8/encoder/x86/sad_sse4.asm b/vp8/encoder/x86/sad_sse4.asm
deleted file mode 100644
index 03ecec4..0000000
--- a/vp8/encoder/x86/sad_sse4.asm
+++ /dev/null
@@ -1,353 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%macro PROCESS_16X2X8 1
-%if %1
- movdqa xmm0, XMMWORD PTR [rsi]
- movq xmm1, MMWORD PTR [rdi]
- movq xmm3, MMWORD PTR [rdi+8]
- movq xmm2, MMWORD PTR [rdi+16]
- punpcklqdq xmm1, xmm3
- punpcklqdq xmm3, xmm2
-
- movdqa xmm2, xmm1
- mpsadbw xmm1, xmm0, 0x0
- mpsadbw xmm2, xmm0, 0x5
-
- psrldq xmm0, 8
-
- movdqa xmm4, xmm3
- mpsadbw xmm3, xmm0, 0x0
- mpsadbw xmm4, xmm0, 0x5
-
- paddw xmm1, xmm2
- paddw xmm1, xmm3
- paddw xmm1, xmm4
-%else
- movdqa xmm0, XMMWORD PTR [rsi]
- movq xmm5, MMWORD PTR [rdi]
- movq xmm3, MMWORD PTR [rdi+8]
- movq xmm2, MMWORD PTR [rdi+16]
- punpcklqdq xmm5, xmm3
- punpcklqdq xmm3, xmm2
-
- movdqa xmm2, xmm5
- mpsadbw xmm5, xmm0, 0x0
- mpsadbw xmm2, xmm0, 0x5
-
- psrldq xmm0, 8
-
- movdqa xmm4, xmm3
- mpsadbw xmm3, xmm0, 0x0
- mpsadbw xmm4, xmm0, 0x5
-
- paddw xmm5, xmm2
- paddw xmm5, xmm3
- paddw xmm5, xmm4
-
- paddw xmm1, xmm5
-%endif
- movdqa xmm0, XMMWORD PTR [rsi + rax]
- movq xmm5, MMWORD PTR [rdi+ rdx]
- movq xmm3, MMWORD PTR [rdi+ rdx+8]
- movq xmm2, MMWORD PTR [rdi+ rdx+16]
- punpcklqdq xmm5, xmm3
- punpcklqdq xmm3, xmm2
-
- lea rsi, [rsi+rax*2]
- lea rdi, [rdi+rdx*2]
-
- movdqa xmm2, xmm5
- mpsadbw xmm5, xmm0, 0x0
- mpsadbw xmm2, xmm0, 0x5
-
- psrldq xmm0, 8
- movdqa xmm4, xmm3
- mpsadbw xmm3, xmm0, 0x0
- mpsadbw xmm4, xmm0, 0x5
-
- paddw xmm5, xmm2
- paddw xmm5, xmm3
- paddw xmm5, xmm4
-
- paddw xmm1, xmm5
-%endmacro
-
-%macro PROCESS_8X2X8 1
-%if %1
- movq xmm0, MMWORD PTR [rsi]
- movq xmm1, MMWORD PTR [rdi]
- movq xmm3, MMWORD PTR [rdi+8]
- punpcklqdq xmm1, xmm3
-
- movdqa xmm2, xmm1
- mpsadbw xmm1, xmm0, 0x0
- mpsadbw xmm2, xmm0, 0x5
- paddw xmm1, xmm2
-%else
- movq xmm0, MMWORD PTR [rsi]
- movq xmm5, MMWORD PTR [rdi]
- movq xmm3, MMWORD PTR [rdi+8]
- punpcklqdq xmm5, xmm3
-
- movdqa xmm2, xmm5
- mpsadbw xmm5, xmm0, 0x0
- mpsadbw xmm2, xmm0, 0x5
- paddw xmm5, xmm2
-
- paddw xmm1, xmm5
-%endif
- movq xmm0, MMWORD PTR [rsi + rax]
- movq xmm5, MMWORD PTR [rdi+ rdx]
- movq xmm3, MMWORD PTR [rdi+ rdx+8]
- punpcklqdq xmm5, xmm3
-
- lea rsi, [rsi+rax*2]
- lea rdi, [rdi+rdx*2]
-
- movdqa xmm2, xmm5
- mpsadbw xmm5, xmm0, 0x0
- mpsadbw xmm2, xmm0, 0x5
- paddw xmm5, xmm2
-
- paddw xmm1, xmm5
-%endmacro
-
-%macro PROCESS_4X2X8 1
-%if %1
- movd xmm0, [rsi]
- movq xmm1, MMWORD PTR [rdi]
- movq xmm3, MMWORD PTR [rdi+8]
- punpcklqdq xmm1, xmm3
-
- mpsadbw xmm1, xmm0, 0x0
-%else
- movd xmm0, [rsi]
- movq xmm5, MMWORD PTR [rdi]
- movq xmm3, MMWORD PTR [rdi+8]
- punpcklqdq xmm5, xmm3
-
- mpsadbw xmm5, xmm0, 0x0
-
- paddw xmm1, xmm5
-%endif
- movd xmm0, [rsi + rax]
- movq xmm5, MMWORD PTR [rdi+ rdx]
- movq xmm3, MMWORD PTR [rdi+ rdx+8]
- punpcklqdq xmm5, xmm3
-
- lea rsi, [rsi+rax*2]
- lea rdi, [rdi+rdx*2]
-
- mpsadbw xmm5, xmm0, 0x0
-
- paddw xmm1, xmm5
-%endmacro
-
-
-;void vp8_sad16x16x8_sse4(
-; const unsigned char *src_ptr,
-; int src_stride,
-; const unsigned char *ref_ptr,
-; int ref_stride,
-; unsigned short *sad_array);
-global sym(vp8_sad16x16x8_sse4)
-sym(vp8_sad16x16x8_sse4):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- PROCESS_16X2X8 1
- PROCESS_16X2X8 0
- PROCESS_16X2X8 0
- PROCESS_16X2X8 0
- PROCESS_16X2X8 0
- PROCESS_16X2X8 0
- PROCESS_16X2X8 0
- PROCESS_16X2X8 0
-
- mov rdi, arg(4) ;Results
- movdqa XMMWORD PTR [rdi], xmm1
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_sad16x8x8_sse4(
-; const unsigned char *src_ptr,
-; int src_stride,
-; const unsigned char *ref_ptr,
-; int ref_stride,
-; unsigned short *sad_array
-;);
-global sym(vp8_sad16x8x8_sse4)
-sym(vp8_sad16x8x8_sse4):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- PROCESS_16X2X8 1
- PROCESS_16X2X8 0
- PROCESS_16X2X8 0
- PROCESS_16X2X8 0
-
- mov rdi, arg(4) ;Results
- movdqa XMMWORD PTR [rdi], xmm1
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_sad8x8x8_sse4(
-; const unsigned char *src_ptr,
-; int src_stride,
-; const unsigned char *ref_ptr,
-; int ref_stride,
-; unsigned short *sad_array
-;);
-global sym(vp8_sad8x8x8_sse4)
-sym(vp8_sad8x8x8_sse4):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- PROCESS_8X2X8 1
- PROCESS_8X2X8 0
- PROCESS_8X2X8 0
- PROCESS_8X2X8 0
-
- mov rdi, arg(4) ;Results
- movdqa XMMWORD PTR [rdi], xmm1
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_sad8x16x8_sse4(
-; const unsigned char *src_ptr,
-; int src_stride,
-; const unsigned char *ref_ptr,
-; int ref_stride,
-; unsigned short *sad_array
-;);
-global sym(vp8_sad8x16x8_sse4)
-sym(vp8_sad8x16x8_sse4):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- PROCESS_8X2X8 1
- PROCESS_8X2X8 0
- PROCESS_8X2X8 0
- PROCESS_8X2X8 0
- PROCESS_8X2X8 0
- PROCESS_8X2X8 0
- PROCESS_8X2X8 0
- PROCESS_8X2X8 0
- mov rdi, arg(4) ;Results
- movdqa XMMWORD PTR [rdi], xmm1
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_sad4x4x8_c(
-; const unsigned char *src_ptr,
-; int src_stride,
-; const unsigned char *ref_ptr,
-; int ref_stride,
-; unsigned short *sad_array
-;);
-global sym(vp8_sad4x4x8_sse4)
-sym(vp8_sad4x4x8_sse4):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- PROCESS_4X2X8 1
- PROCESS_4X2X8 0
-
- mov rdi, arg(4) ;Results
- movdqa XMMWORD PTR [rdi], xmm1
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-
diff --git a/vp8/encoder/x86/sad_ssse3.asm b/vp8/encoder/x86/sad_ssse3.asm
deleted file mode 100644
index 69c5eae..0000000
--- a/vp8/encoder/x86/sad_ssse3.asm
+++ /dev/null
@@ -1,366 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%macro PROCESS_16X2X3 1
-%if %1
- movdqa xmm0, XMMWORD PTR [rsi]
- lddqu xmm5, XMMWORD PTR [rdi]
- lddqu xmm6, XMMWORD PTR [rdi+1]
- lddqu xmm7, XMMWORD PTR [rdi+2]
-
- psadbw xmm5, xmm0
- psadbw xmm6, xmm0
- psadbw xmm7, xmm0
-%else
- movdqa xmm0, XMMWORD PTR [rsi]
- lddqu xmm1, XMMWORD PTR [rdi]
- lddqu xmm2, XMMWORD PTR [rdi+1]
- lddqu xmm3, XMMWORD PTR [rdi+2]
-
- psadbw xmm1, xmm0
- psadbw xmm2, xmm0
- psadbw xmm3, xmm0
-
- paddw xmm5, xmm1
- paddw xmm6, xmm2
- paddw xmm7, xmm3
-%endif
- movdqa xmm0, XMMWORD PTR [rsi+rax]
- lddqu xmm1, XMMWORD PTR [rdi+rdx]
- lddqu xmm2, XMMWORD PTR [rdi+rdx+1]
- lddqu xmm3, XMMWORD PTR [rdi+rdx+2]
-
- lea rsi, [rsi+rax*2]
- lea rdi, [rdi+rdx*2]
-
- psadbw xmm1, xmm0
- psadbw xmm2, xmm0
- psadbw xmm3, xmm0
-
- paddw xmm5, xmm1
- paddw xmm6, xmm2
- paddw xmm7, xmm3
-%endmacro
-
-%macro PROCESS_16X2X3_OFFSET 2
-%if %1
- movdqa xmm0, XMMWORD PTR [rsi]
- movdqa xmm4, XMMWORD PTR [rdi]
- movdqa xmm7, XMMWORD PTR [rdi+16]
-
- movdqa xmm5, xmm7
- palignr xmm5, xmm4, %2
-
- movdqa xmm6, xmm7
- palignr xmm6, xmm4, (%2+1)
-
- palignr xmm7, xmm4, (%2+2)
-
- psadbw xmm5, xmm0
- psadbw xmm6, xmm0
- psadbw xmm7, xmm0
-%else
- movdqa xmm0, XMMWORD PTR [rsi]
- movdqa xmm4, XMMWORD PTR [rdi]
- movdqa xmm3, XMMWORD PTR [rdi+16]
-
- movdqa xmm1, xmm3
- palignr xmm1, xmm4, %2
-
- movdqa xmm2, xmm3
- palignr xmm2, xmm4, (%2+1)
-
- palignr xmm3, xmm4, (%2+2)
-
- psadbw xmm1, xmm0
- psadbw xmm2, xmm0
- psadbw xmm3, xmm0
-
- paddw xmm5, xmm1
- paddw xmm6, xmm2
- paddw xmm7, xmm3
-%endif
- movdqa xmm0, XMMWORD PTR [rsi+rax]
- movdqa xmm4, XMMWORD PTR [rdi+rdx]
- movdqa xmm3, XMMWORD PTR [rdi+rdx+16]
-
- movdqa xmm1, xmm3
- palignr xmm1, xmm4, %2
-
- movdqa xmm2, xmm3
- palignr xmm2, xmm4, (%2+1)
-
- palignr xmm3, xmm4, (%2+2)
-
- lea rsi, [rsi+rax*2]
- lea rdi, [rdi+rdx*2]
-
- psadbw xmm1, xmm0
- psadbw xmm2, xmm0
- psadbw xmm3, xmm0
-
- paddw xmm5, xmm1
- paddw xmm6, xmm2
- paddw xmm7, xmm3
-%endmacro
-
-%macro PROCESS_16X16X3_OFFSET 2
-%2_aligned_by_%1:
-
- sub rdi, %1
-
- PROCESS_16X2X3_OFFSET 1, %1
- PROCESS_16X2X3_OFFSET 0, %1
- PROCESS_16X2X3_OFFSET 0, %1
- PROCESS_16X2X3_OFFSET 0, %1
- PROCESS_16X2X3_OFFSET 0, %1
- PROCESS_16X2X3_OFFSET 0, %1
- PROCESS_16X2X3_OFFSET 0, %1
- PROCESS_16X2X3_OFFSET 0, %1
-
- jmp %2_store_off
-
-%endmacro
-
-%macro PROCESS_16X8X3_OFFSET 2
-%2_aligned_by_%1:
-
- sub rdi, %1
-
- PROCESS_16X2X3_OFFSET 1, %1
- PROCESS_16X2X3_OFFSET 0, %1
- PROCESS_16X2X3_OFFSET 0, %1
- PROCESS_16X2X3_OFFSET 0, %1
-
- jmp %2_store_off
-
-%endmacro
-
-;void int vp8_sad16x16x3_ssse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad16x16x3_ssse3)
-sym(vp8_sad16x16x3_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- push rcx
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- mov rdx, 0xf
- and rdx, rdi
-
- jmp vp8_sad16x16x3_ssse3_skiptable
-vp8_sad16x16x3_ssse3_jumptable:
- dd vp8_sad16x16x3_ssse3_aligned_by_0 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_1 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_2 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_3 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_4 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_5 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_6 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_7 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_8 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_9 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_10 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_11 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_12 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_13 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_14 - vp8_sad16x16x3_ssse3_do_jump
- dd vp8_sad16x16x3_ssse3_aligned_by_15 - vp8_sad16x16x3_ssse3_do_jump
-vp8_sad16x16x3_ssse3_skiptable:
-
- call vp8_sad16x16x3_ssse3_do_jump
-vp8_sad16x16x3_ssse3_do_jump:
- pop rcx ; get the address of do_jump
- mov rax, vp8_sad16x16x3_ssse3_jumptable - vp8_sad16x16x3_ssse3_do_jump
- add rax, rcx ; get the absolute address of vp8_sad16x16x3_ssse3_jumptable
-
- movsxd rax, dword [rax + 4*rdx] ; get the 32 bit offset from the jumptable
- add rcx, rax
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- jmp rcx
-
- PROCESS_16X16X3_OFFSET 0, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 1, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 2, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 3, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 4, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 5, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 6, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 7, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 8, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 9, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 10, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 11, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 12, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 13, vp8_sad16x16x3_ssse3
- PROCESS_16X16X3_OFFSET 14, vp8_sad16x16x3_ssse3
-
-vp8_sad16x16x3_ssse3_aligned_by_15:
- PROCESS_16X2X3 1
- PROCESS_16X2X3 0
- PROCESS_16X2X3 0
- PROCESS_16X2X3 0
- PROCESS_16X2X3 0
- PROCESS_16X2X3 0
- PROCESS_16X2X3 0
- PROCESS_16X2X3 0
-
-vp8_sad16x16x3_ssse3_store_off:
- mov rdi, arg(4) ;Results
-
- movq xmm0, xmm5
- psrldq xmm5, 8
-
- paddw xmm0, xmm5
- movd [rdi], xmm0
-;-
- movq xmm0, xmm6
- psrldq xmm6, 8
-
- paddw xmm0, xmm6
- movd [rdi+4], xmm0
-;-
- movq xmm0, xmm7
- psrldq xmm7, 8
-
- paddw xmm0, xmm7
- movd [rdi+8], xmm0
-
- ; begin epilog
- pop rcx
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void int vp8_sad16x8x3_ssse3(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride,
-; int *results)
-global sym(vp8_sad16x8x3_ssse3)
-sym(vp8_sad16x8x3_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- push rcx
- ; end prolog
-
- mov rsi, arg(0) ;src_ptr
- mov rdi, arg(2) ;ref_ptr
-
- mov rdx, 0xf
- and rdx, rdi
-
- jmp vp8_sad16x8x3_ssse3_skiptable
-vp8_sad16x8x3_ssse3_jumptable:
- dd vp8_sad16x8x3_ssse3_aligned_by_0 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_1 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_2 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_3 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_4 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_5 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_6 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_7 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_8 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_9 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_10 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_11 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_12 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_13 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_14 - vp8_sad16x8x3_ssse3_do_jump
- dd vp8_sad16x8x3_ssse3_aligned_by_15 - vp8_sad16x8x3_ssse3_do_jump
-vp8_sad16x8x3_ssse3_skiptable:
-
- call vp8_sad16x8x3_ssse3_do_jump
-vp8_sad16x8x3_ssse3_do_jump:
- pop rcx ; get the address of do_jump
- mov rax, vp8_sad16x8x3_ssse3_jumptable - vp8_sad16x8x3_ssse3_do_jump
- add rax, rcx ; get the absolute address of vp8_sad16x8x3_ssse3_jumptable
-
- movsxd rax, dword [rax + 4*rdx] ; get the 32 bit offset from the jumptable
- add rcx, rax
-
- movsxd rax, dword ptr arg(1) ;src_stride
- movsxd rdx, dword ptr arg(3) ;ref_stride
-
- jmp rcx
-
- PROCESS_16X8X3_OFFSET 0, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 1, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 2, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 3, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 4, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 5, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 6, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 7, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 8, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 9, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 10, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 11, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 12, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 13, vp8_sad16x8x3_ssse3
- PROCESS_16X8X3_OFFSET 14, vp8_sad16x8x3_ssse3
-
-vp8_sad16x8x3_ssse3_aligned_by_15:
-
- PROCESS_16X2X3 1
- PROCESS_16X2X3 0
- PROCESS_16X2X3 0
- PROCESS_16X2X3 0
-
-vp8_sad16x8x3_ssse3_store_off:
- mov rdi, arg(4) ;Results
-
- movq xmm0, xmm5
- psrldq xmm5, 8
-
- paddw xmm0, xmm5
- movd [rdi], xmm0
-;-
- movq xmm0, xmm6
- psrldq xmm6, 8
-
- paddw xmm0, xmm6
- movd [rdi+4], xmm0
-;-
- movq xmm0, xmm7
- psrldq xmm7, 8
-
- paddw xmm0, xmm7
- movd [rdi+8], xmm0
-
- ; begin epilog
- pop rcx
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/encoder/x86/ssim_opt.asm b/vp8/encoder/x86/ssim_opt.asm
deleted file mode 100644
index c267cdb..0000000
--- a/vp8/encoder/x86/ssim_opt.asm
+++ /dev/null
@@ -1,215 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-%include "vpx_ports/x86_abi_support.asm"
-
-; tabulate_ssim - sums sum_s,sum_r,sum_sq_s,sum_sq_r, sum_sxr
-%macro TABULATE_SSIM 0
- paddusw xmm15, xmm3 ; sum_s
- paddusw xmm14, xmm4 ; sum_r
- movdqa xmm1, xmm3
- pmaddwd xmm1, xmm1
- paddq xmm13, xmm1 ; sum_sq_s
- movdqa xmm2, xmm4
- pmaddwd xmm2, xmm2
- paddq xmm12, xmm2 ; sum_sq_r
- pmaddwd xmm3, xmm4
- paddq xmm11, xmm3 ; sum_sxr
-%endmacro
-
-; Sum across the register %1 starting with q words
-%macro SUM_ACROSS_Q 1
- movdqa xmm2,%1
- punpckldq %1,xmm0
- punpckhdq xmm2,xmm0
- paddq %1,xmm2
- movdqa xmm2,%1
- punpcklqdq %1,xmm0
- punpckhqdq xmm2,xmm0
- paddq %1,xmm2
-%endmacro
-
-; Sum across the register %1 starting with q words
-%macro SUM_ACROSS_W 1
- movdqa xmm1, %1
- punpcklwd %1,xmm0
- punpckhwd xmm1,xmm0
- paddd %1, xmm1
- SUM_ACROSS_Q %1
-%endmacro
-;void ssim_parms_sse3(
-; unsigned char *s,
-; int sp,
-; unsigned char *r,
-; int rp
-; unsigned long *sum_s,
-; unsigned long *sum_r,
-; unsigned long *sum_sq_s,
-; unsigned long *sum_sq_r,
-; unsigned long *sum_sxr);
-;
-; TODO: Use parm passing through structure, probably don't need the pxors
-; ( calling app will initialize to 0 ) could easily fit everything in sse2
-; without too much hastle, and can probably do better estimates with psadw
-; or pavgb At this point this is just meant to be first pass for calculating
-; all the parms needed for 16x16 ssim so we can play with dssim as distortion
-; in mode selection code.
-global sym(vp8_ssim_parms_16x16_sse3)
-sym(vp8_ssim_parms_16x16_sse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 9
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;s
- mov rcx, arg(1) ;sp
- mov rdi, arg(2) ;r
- mov rax, arg(3) ;rp
-
- pxor xmm0, xmm0
- pxor xmm15,xmm15 ;sum_s
- pxor xmm14,xmm14 ;sum_r
- pxor xmm13,xmm13 ;sum_sq_s
- pxor xmm12,xmm12 ;sum_sq_r
- pxor xmm11,xmm11 ;sum_sxr
-
- mov rdx, 16 ;row counter
-NextRow:
-
- ;grab source and reference pixels
- movdqu xmm5, [rsi]
- movdqu xmm6, [rdi]
- movdqa xmm3, xmm5
- movdqa xmm4, xmm6
- punpckhbw xmm3, xmm0 ; high_s
- punpckhbw xmm4, xmm0 ; high_r
-
- TABULATE_SSIM
-
- movdqa xmm3, xmm5
- movdqa xmm4, xmm6
- punpcklbw xmm3, xmm0 ; low_s
- punpcklbw xmm4, xmm0 ; low_r
-
- TABULATE_SSIM
-
- add rsi, rcx ; next s row
- add rdi, rax ; next r row
-
- dec rdx ; counter
- jnz NextRow
-
- SUM_ACROSS_W xmm15
- SUM_ACROSS_W xmm14
- SUM_ACROSS_Q xmm13
- SUM_ACROSS_Q xmm12
- SUM_ACROSS_Q xmm11
-
- mov rdi,arg(4)
- movq [rdi], xmm15;
- mov rdi,arg(5)
- movq [rdi], xmm14;
- mov rdi,arg(6)
- movq [rdi], xmm13;
- mov rdi,arg(7)
- movq [rdi], xmm12;
- mov rdi,arg(8)
- movq [rdi], xmm11;
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void ssim_parms_sse3(
-; unsigned char *s,
-; int sp,
-; unsigned char *r,
-; int rp
-; unsigned long *sum_s,
-; unsigned long *sum_r,
-; unsigned long *sum_sq_s,
-; unsigned long *sum_sq_r,
-; unsigned long *sum_sxr);
-;
-; TODO: Use parm passing through structure, probably don't need the pxors
-; ( calling app will initialize to 0 ) could easily fit everything in sse2
-; without too much hastle, and can probably do better estimates with psadw
-; or pavgb At this point this is just meant to be first pass for calculating
-; all the parms needed for 16x16 ssim so we can play with dssim as distortion
-; in mode selection code.
-global sym(vp8_ssim_parms_8x8_sse3)
-sym(vp8_ssim_parms_8x8_sse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 9
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;s
- mov rcx, arg(1) ;sp
- mov rdi, arg(2) ;r
- mov rax, arg(3) ;rp
-
- pxor xmm0, xmm0
- pxor xmm15,xmm15 ;sum_s
- pxor xmm14,xmm14 ;sum_r
- pxor xmm13,xmm13 ;sum_sq_s
- pxor xmm12,xmm12 ;sum_sq_r
- pxor xmm11,xmm11 ;sum_sxr
-
- mov rdx, 8 ;row counter
-NextRow2:
-
- ;grab source and reference pixels
- movq xmm5, [rsi]
- movq xmm6, [rdi]
-
- movdqa xmm3, xmm5
- movdqa xmm4, xmm6
- punpcklbw xmm3, xmm0 ; low_s
- punpcklbw xmm4, xmm0 ; low_r
-
- TABULATE_SSIM
-
- add rsi, rcx ; next s row
- add rdi, rax ; next r row
-
- dec rdx ; counter
- jnz NextRow2
-
- SUM_ACROSS_W xmm15
- SUM_ACROSS_W xmm14
- SUM_ACROSS_Q xmm13
- SUM_ACROSS_Q xmm12
- SUM_ACROSS_Q xmm11
-
- mov rdi,arg(4)
- movq [rdi], xmm15;
- mov rdi,arg(5)
- movq [rdi], xmm14;
- mov rdi,arg(6)
- movq [rdi], xmm13;
- mov rdi,arg(7)
- movq [rdi], xmm12;
- mov rdi,arg(8)
- movq [rdi], xmm11;
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/encoder/x86/subtract_mmx.asm b/vp8/encoder/x86/subtract_mmx.asm
deleted file mode 100644
index a47e1f0..0000000
--- a/vp8/encoder/x86/subtract_mmx.asm
+++ /dev/null
@@ -1,432 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_subtract_b_mmx_impl(unsigned char *z, int src_stride,
-; short *diff, unsigned char *Predictor,
-; int pitch);
-global sym(vp8_subtract_b_mmx_impl)
-sym(vp8_subtract_b_mmx_impl):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- ; end prolog
-
-
- mov rdi, arg(2) ;diff
- mov rax, arg(3) ;Predictor
- mov rsi, arg(0) ;z
- movsxd rdx, dword ptr arg(1);src_stride;
- movsxd rcx, dword ptr arg(4);pitch
- pxor mm7, mm7
-
- movd mm0, [rsi]
- movd mm1, [rax]
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- psubw mm0, mm1
- movq [rdi], mm0
-
-
- movd mm0, [rsi+rdx]
- movd mm1, [rax+rcx]
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- psubw mm0, mm1
- movq [rdi+rcx*2],mm0
-
-
- movd mm0, [rsi+rdx*2]
- movd mm1, [rax+rcx*2]
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- psubw mm0, mm1
- movq [rdi+rcx*4], mm0
-
- lea rsi, [rsi+rdx*2]
- lea rcx, [rcx+rcx*2]
-
-
-
- movd mm0, [rsi+rdx]
- movd mm1, [rax+rcx]
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- psubw mm0, mm1
- movq [rdi+rcx*2], mm0
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_subtract_mby_mmx(short *diff, unsigned char *src, unsigned char *pred, int stride)
-global sym(vp8_subtract_mby_mmx)
-sym(vp8_subtract_mby_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- ; end prolog
-
-
- mov rsi, arg(1) ;src
- mov rdi, arg(0) ;diff
-
- mov rax, arg(2) ;pred
- movsxd rdx, dword ptr arg(3) ;stride
-
- mov rcx, 16
- pxor mm0, mm0
-
-submby_loop:
-
- movq mm1, [rsi]
- movq mm3, [rax]
-
- movq mm2, mm1
- movq mm4, mm3
-
- punpcklbw mm1, mm0
- punpcklbw mm3, mm0
-
- punpckhbw mm2, mm0
- punpckhbw mm4, mm0
-
- psubw mm1, mm3
- psubw mm2, mm4
-
- movq [rdi], mm1
- movq [rdi+8], mm2
-
-
- movq mm1, [rsi+8]
- movq mm3, [rax+8]
-
- movq mm2, mm1
- movq mm4, mm3
-
- punpcklbw mm1, mm0
- punpcklbw mm3, mm0
-
- punpckhbw mm2, mm0
- punpckhbw mm4, mm0
-
- psubw mm1, mm3
- psubw mm2, mm4
-
- movq [rdi+16], mm1
- movq [rdi+24], mm2
-
-
- add rdi, 32
- add rax, 16
-
- lea rsi, [rsi+rdx]
-
- sub rcx, 1
- jnz submby_loop
-
- pop rdi
- pop rsi
- ; begin epilog
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_subtract_mbuv_mmx(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride)
-global sym(vp8_subtract_mbuv_mmx)
-sym(vp8_subtract_mbuv_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- push rsi
- push rdi
- ; end prolog
-
- ;short *udiff = diff + 256;
- ;short *vdiff = diff + 320;
- ;unsigned char *upred = pred + 256;
- ;unsigned char *vpred = pred + 320;
-
- ;unsigned char *z = usrc;
- ;unsigned short *diff = udiff;
- ;unsigned char *Predictor= upred;
-
- mov rdi, arg(0) ;diff
- mov rax, arg(3) ;pred
- mov rsi, arg(1) ;z = usrc
- add rdi, 256*2 ;diff = diff + 256 (shorts)
- add rax, 256 ;Predictor = pred + 256
- movsxd rdx, dword ptr arg(4) ;stride;
- pxor mm7, mm7
-
- movq mm0, [rsi]
- movq mm1, [rax]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi], mm0
- movq [rdi+8], mm3
-
-
- movq mm0, [rsi+rdx]
- movq mm1, [rax+8]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi+16], mm0
- movq [rdi+24], mm3
-
- movq mm0, [rsi+rdx*2]
- movq mm1, [rax+16]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi+32], mm0
- movq [rdi+40], mm3
- lea rsi, [rsi+rdx*2]
-
-
- movq mm0, [rsi+rdx]
- movq mm1, [rax+24]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
-
- movq [rdi+48], mm0
- movq [rdi+56], mm3
-
-
- add rdi, 64
- add rax, 32
- lea rsi, [rsi+rdx*2]
-
-
- movq mm0, [rsi]
- movq mm1, [rax]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi], mm0
- movq [rdi+8], mm3
-
-
- movq mm0, [rsi+rdx]
- movq mm1, [rax+8]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi+16], mm0
- movq [rdi+24], mm3
-
- movq mm0, [rsi+rdx*2]
- movq mm1, [rax+16]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi+32], mm0
- movq [rdi+40], mm3
- lea rsi, [rsi+rdx*2]
-
-
- movq mm0, [rsi+rdx]
- movq mm1, [rax+24]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
-
- movq [rdi+48], mm0
- movq [rdi+56], mm3
-
- ;unsigned char *z = vsrc;
- ;unsigned short *diff = vdiff;
- ;unsigned char *Predictor= vpred;
-
- mov rdi, arg(0) ;diff
- mov rax, arg(3) ;pred
- mov rsi, arg(2) ;z = usrc
- add rdi, 320*2 ;diff = diff + 320 (shorts)
- add rax, 320 ;Predictor = pred + 320
- movsxd rdx, dword ptr arg(4) ;stride;
- pxor mm7, mm7
-
- movq mm0, [rsi]
- movq mm1, [rax]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi], mm0
- movq [rdi+8], mm3
-
-
- movq mm0, [rsi+rdx]
- movq mm1, [rax+8]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi+16], mm0
- movq [rdi+24], mm3
-
- movq mm0, [rsi+rdx*2]
- movq mm1, [rax+16]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi+32], mm0
- movq [rdi+40], mm3
- lea rsi, [rsi+rdx*2]
-
-
- movq mm0, [rsi+rdx]
- movq mm1, [rax+24]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
-
- movq [rdi+48], mm0
- movq [rdi+56], mm3
-
-
- add rdi, 64
- add rax, 32
- lea rsi, [rsi+rdx*2]
-
-
- movq mm0, [rsi]
- movq mm1, [rax]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi], mm0
- movq [rdi+8], mm3
-
-
- movq mm0, [rsi+rdx]
- movq mm1, [rax+8]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi+16], mm0
- movq [rdi+24], mm3
-
- movq mm0, [rsi+rdx*2]
- movq mm1, [rax+16]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
- movq [rdi+32], mm0
- movq [rdi+40], mm3
- lea rsi, [rsi+rdx*2]
-
-
- movq mm0, [rsi+rdx]
- movq mm1, [rax+24]
- movq mm3, mm0
- movq mm4, mm1
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- punpckhbw mm3, mm7
- punpckhbw mm4, mm7
- psubw mm0, mm1
- psubw mm3, mm4
-
- movq [rdi+48], mm0
- movq [rdi+56], mm3
-
- ; begin epilog
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
diff --git a/vp8/encoder/x86/subtract_sse2.asm b/vp8/encoder/x86/subtract_sse2.asm
deleted file mode 100644
index 3fb23d0..0000000
--- a/vp8/encoder/x86/subtract_sse2.asm
+++ /dev/null
@@ -1,356 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_subtract_b_sse2_impl(unsigned char *z, int src_stride,
-; short *diff, unsigned char *Predictor,
-; int pitch);
-global sym(vp8_subtract_b_sse2_impl)
-sym(vp8_subtract_b_sse2_impl):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rdi, arg(2) ;diff
- mov rax, arg(3) ;Predictor
- mov rsi, arg(0) ;z
- movsxd rdx, dword ptr arg(1);src_stride;
- movsxd rcx, dword ptr arg(4);pitch
- pxor mm7, mm7
-
- movd mm0, [rsi]
- movd mm1, [rax]
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- psubw mm0, mm1
- movq MMWORD PTR [rdi], mm0
-
- movd mm0, [rsi+rdx]
- movd mm1, [rax+rcx]
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- psubw mm0, mm1
- movq MMWORD PTR [rdi+rcx*2], mm0
-
- movd mm0, [rsi+rdx*2]
- movd mm1, [rax+rcx*2]
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- psubw mm0, mm1
- movq MMWORD PTR [rdi+rcx*4], mm0
-
- lea rsi, [rsi+rdx*2]
- lea rcx, [rcx+rcx*2]
-
- movd mm0, [rsi+rdx]
- movd mm1, [rax+rcx]
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
- psubw mm0, mm1
- movq MMWORD PTR [rdi+rcx*2], mm0
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_subtract_mby_sse2(short *diff, unsigned char *src, unsigned char *pred, int stride)
-global sym(vp8_subtract_mby_sse2)
-sym(vp8_subtract_mby_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(1) ;src
- mov rdi, arg(0) ;diff
-
- mov rax, arg(2) ;pred
- movsxd rdx, dword ptr arg(3) ;stride
-
- mov rcx, 8 ; do two lines at one time
-
-submby_loop:
- movdqa xmm0, XMMWORD PTR [rsi] ; src
- movdqa xmm1, XMMWORD PTR [rax] ; pred
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi], xmm0
- movdqa XMMWORD PTR [rdi +16], xmm2
-
- movdqa xmm4, XMMWORD PTR [rsi + rdx]
- movdqa xmm5, XMMWORD PTR [rax + 16]
-
- movdqa xmm6, xmm4
- psubb xmm4, xmm5
-
- pxor xmm5, [GLOBAL(t80)] ;convert to signed values
- pxor xmm6, [GLOBAL(t80)]
- pcmpgtb xmm5, xmm6 ; obtain sign information
-
- movdqa xmm6, xmm4
- movdqa xmm7, xmm5
- punpcklbw xmm4, xmm5 ; put sign back to subtraction
- punpckhbw xmm6, xmm7 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi +32], xmm4
- movdqa XMMWORD PTR [rdi +48], xmm6
-
- add rdi, 64
- add rax, 32
- lea rsi, [rsi+rdx*2]
-
- sub rcx, 1
- jnz submby_loop
-
- pop rdi
- pop rsi
- ; begin epilog
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_subtract_mbuv_sse2(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride)
-global sym(vp8_subtract_mbuv_sse2)
-sym(vp8_subtract_mbuv_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 5
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rdi, arg(0) ;diff
- mov rax, arg(3) ;pred
- mov rsi, arg(1) ;z = usrc
- add rdi, 256*2 ;diff = diff + 256 (shorts)
- add rax, 256 ;Predictor = pred + 256
- movsxd rdx, dword ptr arg(4) ;stride;
- lea rcx, [rdx + rdx*2]
-
- ;u
- ;line 0 1
- movq xmm0, MMWORD PTR [rsi] ; src
- movq xmm2, MMWORD PTR [rsi+rdx]
- movdqa xmm1, XMMWORD PTR [rax] ; pred
- punpcklqdq xmm0, xmm2
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1 ; subtraction with sign missed
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi], xmm0
- movdqa XMMWORD PTR [rdi +16], xmm2
-
- ;line 2 3
- movq xmm0, MMWORD PTR [rsi+rdx*2] ; src
- movq xmm2, MMWORD PTR [rsi+rcx]
- movdqa xmm1, XMMWORD PTR [rax+16] ; pred
- punpcklqdq xmm0, xmm2
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1 ; subtraction with sign missed
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi + 32], xmm0
- movdqa XMMWORD PTR [rdi + 48], xmm2
-
- ;line 4 5
- lea rsi, [rsi + rdx*4]
-
- movq xmm0, MMWORD PTR [rsi] ; src
- movq xmm2, MMWORD PTR [rsi+rdx]
- movdqa xmm1, XMMWORD PTR [rax + 32] ; pred
- punpcklqdq xmm0, xmm2
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1 ; subtraction with sign missed
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi + 64], xmm0
- movdqa XMMWORD PTR [rdi + 80], xmm2
-
- ;line 6 7
- movq xmm0, MMWORD PTR [rsi+rdx*2] ; src
- movq xmm2, MMWORD PTR [rsi+rcx]
- movdqa xmm1, XMMWORD PTR [rax+ 48] ; pred
- punpcklqdq xmm0, xmm2
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1 ; subtraction with sign missed
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi + 96], xmm0
- movdqa XMMWORD PTR [rdi + 112], xmm2
-
- ;v
- mov rsi, arg(2) ;z = vsrc
- add rdi, 64*2 ;diff = diff + 320 (shorts)
- add rax, 64 ;Predictor = pred + 320
-
- ;line 0 1
- movq xmm0, MMWORD PTR [rsi] ; src
- movq xmm2, MMWORD PTR [rsi+rdx]
- movdqa xmm1, XMMWORD PTR [rax] ; pred
- punpcklqdq xmm0, xmm2
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1 ; subtraction with sign missed
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi], xmm0
- movdqa XMMWORD PTR [rdi +16], xmm2
-
- ;line 2 3
- movq xmm0, MMWORD PTR [rsi+rdx*2] ; src
- movq xmm2, MMWORD PTR [rsi+rcx]
- movdqa xmm1, XMMWORD PTR [rax+16] ; pred
- punpcklqdq xmm0, xmm2
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1 ; subtraction with sign missed
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi + 32], xmm0
- movdqa XMMWORD PTR [rdi + 48], xmm2
-
- ;line 4 5
- lea rsi, [rsi + rdx*4]
-
- movq xmm0, MMWORD PTR [rsi] ; src
- movq xmm2, MMWORD PTR [rsi+rdx]
- movdqa xmm1, XMMWORD PTR [rax + 32] ; pred
- punpcklqdq xmm0, xmm2
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1 ; subtraction with sign missed
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi + 64], xmm0
- movdqa XMMWORD PTR [rdi + 80], xmm2
-
- ;line 6 7
- movq xmm0, MMWORD PTR [rsi+rdx*2] ; src
- movq xmm2, MMWORD PTR [rsi+rcx]
- movdqa xmm1, XMMWORD PTR [rax+ 48] ; pred
- punpcklqdq xmm0, xmm2
-
- movdqa xmm2, xmm0
- psubb xmm0, xmm1 ; subtraction with sign missed
-
- pxor xmm1, [GLOBAL(t80)] ;convert to signed values
- pxor xmm2, [GLOBAL(t80)]
- pcmpgtb xmm1, xmm2 ; obtain sign information
-
- movdqa xmm2, xmm0
- movdqa xmm3, xmm1
- punpcklbw xmm0, xmm1 ; put sign back to subtraction
- punpckhbw xmm2, xmm3 ; put sign back to subtraction
-
- movdqa XMMWORD PTR [rdi + 96], xmm0
- movdqa XMMWORD PTR [rdi + 112], xmm2
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-t80:
- times 16 db 0x80
diff --git a/vp8/encoder/x86/temporal_filter_apply_sse2.asm b/vp8/encoder/x86/temporal_filter_apply_sse2.asm
deleted file mode 100644
index 0127b01..0000000
--- a/vp8/encoder/x86/temporal_filter_apply_sse2.asm
+++ /dev/null
@@ -1,207 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-; void vp8_temporal_filter_apply_sse2 | arg
-; (unsigned char *frame1, | 0
-; unsigned int stride, | 1
-; unsigned char *frame2, | 2
-; unsigned int block_size, | 3
-; int strength, | 4
-; int filter_weight, | 5
-; unsigned int *accumulator, | 6
-; unsigned short *count) | 7
-global sym(vp8_temporal_filter_apply_sse2)
-sym(vp8_temporal_filter_apply_sse2):
-
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 8
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ALIGN_STACK 16, rax
- %define block_size 0
- %define strength 16
- %define filter_weight 32
- %define rounding_bit 48
- %define rbp_backup 64
- %define stack_size 80
- sub rsp, stack_size
- mov [rsp + rbp_backup], rbp
- ; end prolog
-
- mov rdx, arg(3)
- mov [rsp + block_size], rdx
- movd xmm6, arg(4)
- movdqa [rsp + strength], xmm6 ; where strength is used, all 16 bytes are read
-
- ; calculate the rounding bit outside the loop
- ; 0x8000 >> (16 - strength)
- mov rdx, 16
- sub rdx, arg(4) ; 16 - strength
- movd xmm4, rdx ; can't use rdx w/ shift
- movdqa xmm5, [GLOBAL(_const_top_bit)]
- psrlw xmm5, xmm4
- movdqa [rsp + rounding_bit], xmm5
-
- mov rsi, arg(0) ; src/frame1
- mov rdx, arg(2) ; predictor frame
- mov rdi, arg(6) ; accumulator
- mov rax, arg(7) ; count
-
- ; dup the filter weight and store for later
- movd xmm0, arg(5) ; filter_weight
- pshuflw xmm0, xmm0, 0
- punpcklwd xmm0, xmm0
- movdqa [rsp + filter_weight], xmm0
-
- mov rbp, arg(1) ; stride
- pxor xmm7, xmm7 ; zero for extraction
-
- lea rcx, [rdx + 16*16*1]
- cmp dword ptr [rsp + block_size], 8
- jne temporal_filter_apply_load_16
- lea rcx, [rdx + 8*8*1]
-
-temporal_filter_apply_load_8:
- movq xmm0, [rsi] ; first row
- lea rsi, [rsi + rbp] ; += stride
- punpcklbw xmm0, xmm7 ; src[ 0- 7]
- movq xmm1, [rsi] ; second row
- lea rsi, [rsi + rbp] ; += stride
- punpcklbw xmm1, xmm7 ; src[ 8-15]
- jmp temporal_filter_apply_load_finished
-
-temporal_filter_apply_load_16:
- movdqa xmm0, [rsi] ; src (frame1)
- lea rsi, [rsi + rbp] ; += stride
- movdqa xmm1, xmm0
- punpcklbw xmm0, xmm7 ; src[ 0- 7]
- punpckhbw xmm1, xmm7 ; src[ 8-15]
-
-temporal_filter_apply_load_finished:
- movdqa xmm2, [rdx] ; predictor (frame2)
- movdqa xmm3, xmm2
- punpcklbw xmm2, xmm7 ; pred[ 0- 7]
- punpckhbw xmm3, xmm7 ; pred[ 8-15]
-
- ; modifier = src_byte - pixel_value
- psubw xmm0, xmm2 ; src - pred[ 0- 7]
- psubw xmm1, xmm3 ; src - pred[ 8-15]
-
- ; modifier *= modifier
- pmullw xmm0, xmm0 ; modifer[ 0- 7]^2
- pmullw xmm1, xmm1 ; modifer[ 8-15]^2
-
- ; modifier *= 3
- pmullw xmm0, [GLOBAL(_const_3w)]
- pmullw xmm1, [GLOBAL(_const_3w)]
-
- ; modifer += 0x8000 >> (16 - strength)
- paddw xmm0, [rsp + rounding_bit]
- paddw xmm1, [rsp + rounding_bit]
-
- ; modifier >>= strength
- psrlw xmm0, [rsp + strength]
- psrlw xmm1, [rsp + strength]
-
- ; modifier = 16 - modifier
- ; saturation takes care of modifier > 16
- movdqa xmm3, [GLOBAL(_const_16w)]
- movdqa xmm2, [GLOBAL(_const_16w)]
- psubusw xmm3, xmm1
- psubusw xmm2, xmm0
-
- ; modifier *= filter_weight
- pmullw xmm2, [rsp + filter_weight]
- pmullw xmm3, [rsp + filter_weight]
-
- ; count
- movdqa xmm4, [rax]
- movdqa xmm5, [rax+16]
- ; += modifier
- paddw xmm4, xmm2
- paddw xmm5, xmm3
- ; write back
- movdqa [rax], xmm4
- movdqa [rax+16], xmm5
- lea rax, [rax + 16*2] ; count += 16*(sizeof(short))
-
- ; load and extract the predictor up to shorts
- pxor xmm7, xmm7
- movdqa xmm0, [rdx]
- lea rdx, [rdx + 16*1] ; pred += 16*(sizeof(char))
- movdqa xmm1, xmm0
- punpcklbw xmm0, xmm7 ; pred[ 0- 7]
- punpckhbw xmm1, xmm7 ; pred[ 8-15]
-
- ; modifier *= pixel_value
- pmullw xmm0, xmm2
- pmullw xmm1, xmm3
-
- ; expand to double words
- movdqa xmm2, xmm0
- punpcklwd xmm0, xmm7 ; [ 0- 3]
- punpckhwd xmm2, xmm7 ; [ 4- 7]
- movdqa xmm3, xmm1
- punpcklwd xmm1, xmm7 ; [ 8-11]
- punpckhwd xmm3, xmm7 ; [12-15]
-
- ; accumulator
- movdqa xmm4, [rdi]
- movdqa xmm5, [rdi+16]
- movdqa xmm6, [rdi+32]
- movdqa xmm7, [rdi+48]
- ; += modifier
- paddw xmm4, xmm0
- paddw xmm5, xmm2
- paddw xmm6, xmm1
- paddw xmm7, xmm3
- ; write back
- movdqa [rdi], xmm4
- movdqa [rdi+16], xmm5
- movdqa [rdi+32], xmm6
- movdqa [rdi+48], xmm7
- lea rdi, [rdi + 16*4] ; accumulator += 16*(sizeof(int))
-
- cmp rdx, rcx
- je temporal_filter_apply_epilog
- pxor xmm7, xmm7 ; zero for extraction
- cmp dword ptr [rsp + block_size], 16
- je temporal_filter_apply_load_16
- jmp temporal_filter_apply_load_8
-
-temporal_filter_apply_epilog:
- ; begin epilog
- mov rbp, [rsp + rbp_backup]
- add rsp, stack_size
- pop rsp
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-align 16
-_const_3w:
- times 8 dw 3
-align 16
-_const_top_bit:
- times 8 dw 1<<15
-align 16
-_const_16w
- times 8 dw 16
diff --git a/vp8/encoder/x86/temporal_filter_x86.h b/vp8/encoder/x86/temporal_filter_x86.h
deleted file mode 100644
index 2daa140..0000000
--- a/vp8/encoder/x86/temporal_filter_x86.h
+++ /dev/null
@@ -1,27 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __INC_VP8_TEMPORAL_FILTER_X86_H
-#define __INC_VP8_TEMPORAL_FILTER_X86_H
-
-#if HAVE_SSE2
-extern prototype_apply(vp8_temporal_filter_apply_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_temporal_filter_apply
-#define vp8_temporal_filter_apply vp8_temporal_filter_apply_sse2
-
-#endif
-
-#endif
-
-#endif // __INC_VP8_TEMPORAL_FILTER_X86_H
diff --git a/vp8/encoder/x86/variance_impl_mmx.asm b/vp8/encoder/x86/variance_impl_mmx.asm
deleted file mode 100644
index 67a9b4d..0000000
--- a/vp8/encoder/x86/variance_impl_mmx.asm
+++ /dev/null
@@ -1,981 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;unsigned int vp8_get_mb_ss_mmx( short *src_ptr )
-global sym(vp8_get_mb_ss_mmx)
-sym(vp8_get_mb_ss_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- sub rsp, 8
- ; end prolog
-
- mov rax, arg(0) ;src_ptr
- mov rcx, 16
- pxor mm4, mm4
-
-NEXTROW:
- movq mm0, [rax]
- movq mm1, [rax+8]
- movq mm2, [rax+16]
- movq mm3, [rax+24]
- pmaddwd mm0, mm0
- pmaddwd mm1, mm1
- pmaddwd mm2, mm2
- pmaddwd mm3, mm3
-
- paddd mm4, mm0
- paddd mm4, mm1
- paddd mm4, mm2
- paddd mm4, mm3
-
- add rax, 32
- dec rcx
- ja NEXTROW
- movq QWORD PTR [rsp], mm4
-
- ;return sum[0]+sum[1];
- movsxd rax, dword ptr [rsp]
- movsxd rcx, dword ptr [rsp+4]
- add rax, rcx
-
-
- ; begin epilog
- add rsp, 8
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_get8x8var_mmx
-;(
-; unsigned char *src_ptr,
-; int source_stride,
-; unsigned char *ref_ptr,
-; int recon_stride,
-; unsigned int *SSE,
-; int *Sum
-;)
-global sym(vp8_get8x8var_mmx)
-sym(vp8_get8x8var_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- push rsi
- push rdi
- push rbx
- sub rsp, 16
- ; end prolog
-
-
- pxor mm5, mm5 ; Blank mmx6
- pxor mm6, mm6 ; Blank mmx7
- pxor mm7, mm7 ; Blank mmx7
-
- mov rax, arg(0) ;[src_ptr] ; Load base addresses
- mov rbx, arg(2) ;[ref_ptr]
- movsxd rcx, dword ptr arg(1) ;[source_stride]
- movsxd rdx, dword ptr arg(3) ;[recon_stride]
-
- ; Row 1
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm1, [rbx] ; Copy eight bytes to mm1
- movq mm2, mm0 ; Take copies
- movq mm3, mm1 ; Take copies
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- punpckhbw mm2, mm6 ; unpack to higher prrcision
- punpckhbw mm3, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- psubsw mm2, mm3 ; A-B (high order) to MM2
-
- paddw mm5, mm0 ; accumulate differences in mm5
- paddw mm5, mm2 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- pmaddwd mm2, mm2 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
- paddd mm7, mm2 ; accumulate in mm7
-
-
- ; Row 2
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm2, mm0 ; Take copies
- movq mm3, mm1 ; Take copies
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- punpckhbw mm2, mm6 ; unpack to higher prrcision
- punpckhbw mm3, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- psubsw mm2, mm3 ; A-B (high order) to MM2
-
- paddw mm5, mm0 ; accumulate differences in mm5
- paddw mm5, mm2 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- pmaddwd mm2, mm2 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
- paddd mm7, mm2 ; accumulate in mm7
-
- ; Row 3
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm2, mm0 ; Take copies
- movq mm3, mm1 ; Take copies
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- punpckhbw mm2, mm6 ; unpack to higher prrcision
- punpckhbw mm3, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- psubsw mm2, mm3 ; A-B (high order) to MM2
-
- paddw mm5, mm0 ; accumulate differences in mm5
- paddw mm5, mm2 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- pmaddwd mm2, mm2 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
- paddd mm7, mm2 ; accumulate in mm7
-
- ; Row 4
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm2, mm0 ; Take copies
- movq mm3, mm1 ; Take copies
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- punpckhbw mm2, mm6 ; unpack to higher prrcision
- punpckhbw mm3, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- psubsw mm2, mm3 ; A-B (high order) to MM2
-
- paddw mm5, mm0 ; accumulate differences in mm5
- paddw mm5, mm2 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- pmaddwd mm2, mm2 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
- paddd mm7, mm2 ; accumulate in mm7
-
- ; Row 5
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm2, mm0 ; Take copies
- movq mm3, mm1 ; Take copies
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- punpckhbw mm2, mm6 ; unpack to higher prrcision
- punpckhbw mm3, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- psubsw mm2, mm3 ; A-B (high order) to MM2
-
- paddw mm5, mm0 ; accumulate differences in mm5
- paddw mm5, mm2 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- pmaddwd mm2, mm2 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- ; movq mm4, [rbx + rdx]
- paddd mm7, mm0 ; accumulate in mm7
- paddd mm7, mm2 ; accumulate in mm7
-
- ; Row 6
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm2, mm0 ; Take copies
- movq mm3, mm1 ; Take copies
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- punpckhbw mm2, mm6 ; unpack to higher prrcision
- punpckhbw mm3, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- psubsw mm2, mm3 ; A-B (high order) to MM2
-
- paddw mm5, mm0 ; accumulate differences in mm5
- paddw mm5, mm2 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- pmaddwd mm2, mm2 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
- paddd mm7, mm2 ; accumulate in mm7
-
- ; Row 7
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm2, mm0 ; Take copies
- movq mm3, mm1 ; Take copies
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- punpckhbw mm2, mm6 ; unpack to higher prrcision
- punpckhbw mm3, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- psubsw mm2, mm3 ; A-B (high order) to MM2
-
- paddw mm5, mm0 ; accumulate differences in mm5
- paddw mm5, mm2 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- pmaddwd mm2, mm2 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
- paddd mm7, mm2 ; accumulate in mm7
-
- ; Row 8
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm2, mm0 ; Take copies
- movq mm3, mm1 ; Take copies
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- punpckhbw mm2, mm6 ; unpack to higher prrcision
- punpckhbw mm3, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- psubsw mm2, mm3 ; A-B (high order) to MM2
-
- paddw mm5, mm0 ; accumulate differences in mm5
- paddw mm5, mm2 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- pmaddwd mm2, mm2 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- paddd mm7, mm0 ; accumulate in mm7
- paddd mm7, mm2 ; accumulate in mm7
-
- ; Now accumulate the final results.
- movq QWORD PTR [rsp+8], mm5 ; copy back accumulated results into normal memory
- movq QWORD PTR [rsp], mm7 ; copy back accumulated results into normal memory
- movsx rdx, WORD PTR [rsp+8]
- movsx rcx, WORD PTR [rsp+10]
- movsx rbx, WORD PTR [rsp+12]
- movsx rax, WORD PTR [rsp+14]
- add rdx, rcx
- add rbx, rax
- add rdx, rbx ;XSum
- movsxd rax, DWORD PTR [rsp]
- movsxd rcx, DWORD PTR [rsp+4]
- add rax, rcx ;XXSum
- mov rsi, arg(4) ;SSE
- mov rdi, arg(5) ;Sum
- mov dword ptr [rsi], eax
- mov dword ptr [rdi], edx
- xor rax, rax ; return 0
-
-
- ; begin epilog
- add rsp, 16
- pop rbx
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-;unsigned int
-;vp8_get4x4var_mmx
-;(
-; unsigned char *src_ptr,
-; int source_stride,
-; unsigned char *ref_ptr,
-; int recon_stride,
-; unsigned int *SSE,
-; int *Sum
-;)
-global sym(vp8_get4x4var_mmx)
-sym(vp8_get4x4var_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- push rsi
- push rdi
- push rbx
- sub rsp, 16
- ; end prolog
-
-
- pxor mm5, mm5 ; Blank mmx6
- pxor mm6, mm6 ; Blank mmx7
- pxor mm7, mm7 ; Blank mmx7
-
- mov rax, arg(0) ;[src_ptr] ; Load base addresses
- mov rbx, arg(2) ;[ref_ptr]
- movsxd rcx, dword ptr arg(1) ;[source_stride]
- movsxd rdx, dword ptr arg(3) ;[recon_stride]
-
- ; Row 1
- movq mm0, [rax] ; Copy eight bytes to mm0
- movq mm1, [rbx] ; Copy eight bytes to mm1
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- paddw mm5, mm0 ; accumulate differences in mm5
- pmaddwd mm0, mm0 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
-
-
- ; Row 2
- movq mm0, [rax] ; Copy eight bytes to mm0
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- paddw mm5, mm0 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
-
- ; Row 3
- movq mm0, [rax] ; Copy eight bytes to mm0
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- paddw mm5, mm0 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movq mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
-
- ; Row 4
- movq mm0, [rax] ; Copy eight bytes to mm0
-
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
-
- paddw mm5, mm0 ; accumulate differences in mm5
-
- pmaddwd mm0, mm0 ; square and accumulate
- paddd mm7, mm0 ; accumulate in mm7
-
-
- ; Now accumulate the final results.
- movq QWORD PTR [rsp+8], mm5 ; copy back accumulated results into normal memory
- movq QWORD PTR [rsp], mm7 ; copy back accumulated results into normal memory
- movsx rdx, WORD PTR [rsp+8]
- movsx rcx, WORD PTR [rsp+10]
- movsx rbx, WORD PTR [rsp+12]
- movsx rax, WORD PTR [rsp+14]
- add rdx, rcx
- add rbx, rax
- add rdx, rbx ;XSum
- movsxd rax, DWORD PTR [rsp]
- movsxd rcx, DWORD PTR [rsp+4]
- add rax, rcx ;XXSum
- mov rsi, arg(4) ;SSE
- mov rdi, arg(5) ;Sum
- mov dword ptr [rsi], eax
- mov dword ptr [rdi], edx
- xor rax, rax ; return 0
-
-
- ; begin epilog
- add rsp, 16
- pop rbx
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-;unsigned int
-;vp8_get4x4sse_cs_mmx
-;(
-; unsigned char *src_ptr,
-; int source_stride,
-; unsigned char *ref_ptr,
-; int recon_stride
-;)
-global sym(vp8_get4x4sse_cs_mmx)
-sym(vp8_get4x4sse_cs_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- push rsi
- push rdi
- push rbx
- ; end prolog
-
-
- pxor mm6, mm6 ; Blank mmx7
- pxor mm7, mm7 ; Blank mmx7
-
- mov rax, arg(0) ;[src_ptr] ; Load base addresses
- mov rbx, arg(2) ;[ref_ptr]
- movsxd rcx, dword ptr arg(1) ;[source_stride]
- movsxd rdx, dword ptr arg(3) ;[recon_stride]
- ; Row 1
- movd mm0, [rax] ; Copy eight bytes to mm0
- movd mm1, [rbx] ; Copy eight bytes to mm1
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- pmaddwd mm0, mm0 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movd mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
-
- ; Row 2
- movd mm0, [rax] ; Copy eight bytes to mm0
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- pmaddwd mm0, mm0 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movd mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
-
- ; Row 3
- movd mm0, [rax] ; Copy eight bytes to mm0
- punpcklbw mm1, mm6
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- psubsw mm0, mm1 ; A-B (low order) to MM0
-
- pmaddwd mm0, mm0 ; square and accumulate
- add rbx,rdx ; Inc pointer into ref data
- add rax,rcx ; Inc pointer into the new data
- movd mm1, [rbx] ; Copy eight bytes to mm1
- paddd mm7, mm0 ; accumulate in mm7
-
- ; Row 4
- movd mm0, [rax] ; Copy eight bytes to mm0
- punpcklbw mm0, mm6 ; unpack to higher prrcision
- punpcklbw mm1, mm6
- psubsw mm0, mm1 ; A-B (low order) to MM0
- pmaddwd mm0, mm0 ; square and accumulate
- paddd mm7, mm0 ; accumulate in mm7
-
- movq mm0, mm7 ;
- psrlq mm7, 32
-
- paddd mm0, mm7
- movq rax, mm0
-
-
- ; begin epilog
- pop rbx
- pop rdi
- pop rsi
- UNSHADOW_ARGS
- pop rbp
- ret
-
-%define mmx_filter_shift 7
-
-;void vp8_filter_block2d_bil4x4_var_mmx
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned short *HFilter,
-; unsigned short *VFilter,
-; int *sum,
-; unsigned int *sumsquared
-;)
-global sym(vp8_filter_block2d_bil4x4_var_mmx)
-sym(vp8_filter_block2d_bil4x4_var_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 8
- GET_GOT rbx
- push rsi
- push rdi
- sub rsp, 16
- ; end prolog
-
-
- pxor mm6, mm6 ;
- pxor mm7, mm7 ;
-
- mov rax, arg(4) ;HFilter ;
- mov rdx, arg(5) ;VFilter ;
-
- mov rsi, arg(0) ;ref_ptr ;
- mov rdi, arg(2) ;src_ptr ;
-
- mov rcx, 4 ;
- pxor mm0, mm0 ;
-
- movd mm1, [rsi] ;
- movd mm3, [rsi+1] ;
-
- punpcklbw mm1, mm0 ;
- pmullw mm1, [rax] ;
-
- punpcklbw mm3, mm0 ;
- pmullw mm3, [rax+8] ;
-
- paddw mm1, mm3 ;
- paddw mm1, [GLOBAL(mmx_bi_rd)] ;
-
- psraw mm1, mmx_filter_shift ;
- movq mm5, mm1
-
-%if ABI_IS_32BIT
- add rsi, dword ptr arg(1) ;ref_pixels_per_line ;
-%else
- movsxd r8, dword ptr arg(1) ;ref_pixels_per_line ;
- add rsi, r8
-%endif
-
-filter_block2d_bil4x4_var_mmx_loop:
-
- movd mm1, [rsi] ;
- movd mm3, [rsi+1] ;
-
- punpcklbw mm1, mm0 ;
- pmullw mm1, [rax] ;
-
- punpcklbw mm3, mm0 ;
- pmullw mm3, [rax+8] ;
-
- paddw mm1, mm3 ;
- paddw mm1, [GLOBAL(mmx_bi_rd)] ;
-
- psraw mm1, mmx_filter_shift ;
- movq mm3, mm5 ;
-
- movq mm5, mm1 ;
- pmullw mm3, [rdx] ;
-
- pmullw mm1, [rdx+8] ;
- paddw mm1, mm3 ;
-
-
- paddw mm1, [GLOBAL(mmx_bi_rd)] ;
- psraw mm1, mmx_filter_shift ;
-
- movd mm3, [rdi] ;
- punpcklbw mm3, mm0 ;
-
- psubw mm1, mm3 ;
- paddw mm6, mm1 ;
-
- pmaddwd mm1, mm1 ;
- paddd mm7, mm1 ;
-
-%if ABI_IS_32BIT
- add rsi, dword ptr arg(1) ;ref_pixels_per_line ;
- add rdi, dword ptr arg(3) ;src_pixels_per_line ;
-%else
- movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line
- add rsi, r8
- add rdi, r9
-%endif
- sub rcx, 1 ;
- jnz filter_block2d_bil4x4_var_mmx_loop ;
-
-
- pxor mm3, mm3 ;
- pxor mm2, mm2 ;
-
- punpcklwd mm2, mm6 ;
- punpckhwd mm3, mm6 ;
-
- paddd mm2, mm3 ;
- movq mm6, mm2 ;
-
- psrlq mm6, 32 ;
- paddd mm2, mm6 ;
-
- psrad mm2, 16 ;
- movq mm4, mm7 ;
-
- psrlq mm4, 32 ;
- paddd mm4, mm7 ;
-
- mov rdi, arg(6) ;sum
- mov rsi, arg(7) ;sumsquared
-
- movd dword ptr [rdi], mm2 ;
- movd dword ptr [rsi], mm4 ;
-
-
-
- ; begin epilog
- add rsp, 16
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-
-;void vp8_filter_block2d_bil_var_mmx
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; unsigned short *HFilter,
-; unsigned short *VFilter,
-; int *sum,
-; unsigned int *sumsquared
-;)
-global sym(vp8_filter_block2d_bil_var_mmx)
-sym(vp8_filter_block2d_bil_var_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 9
- GET_GOT rbx
- push rsi
- push rdi
- sub rsp, 16
- ; end prolog
-
- pxor mm6, mm6 ;
- pxor mm7, mm7 ;
- mov rax, arg(5) ;HFilter ;
-
- mov rdx, arg(6) ;VFilter ;
- mov rsi, arg(0) ;ref_ptr ;
-
- mov rdi, arg(2) ;src_ptr ;
- movsxd rcx, dword ptr arg(4) ;Height ;
-
- pxor mm0, mm0 ;
- movq mm1, [rsi] ;
-
- movq mm3, [rsi+1] ;
- movq mm2, mm1 ;
-
- movq mm4, mm3 ;
- punpcklbw mm1, mm0 ;
-
- punpckhbw mm2, mm0 ;
- pmullw mm1, [rax] ;
-
- pmullw mm2, [rax] ;
- punpcklbw mm3, mm0 ;
-
- punpckhbw mm4, mm0 ;
- pmullw mm3, [rax+8] ;
-
- pmullw mm4, [rax+8] ;
- paddw mm1, mm3 ;
-
- paddw mm2, mm4 ;
- paddw mm1, [GLOBAL(mmx_bi_rd)] ;
-
- psraw mm1, mmx_filter_shift ;
- paddw mm2, [GLOBAL(mmx_bi_rd)] ;
-
- psraw mm2, mmx_filter_shift ;
- movq mm5, mm1
-
- packuswb mm5, mm2 ;
-%if ABI_IS_32BIT
- add rsi, dword ptr arg(1) ;ref_pixels_per_line
-%else
- movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
- add rsi, r8
-%endif
-
-filter_block2d_bil_var_mmx_loop:
-
- movq mm1, [rsi] ;
- movq mm3, [rsi+1] ;
-
- movq mm2, mm1 ;
- movq mm4, mm3 ;
-
- punpcklbw mm1, mm0 ;
- punpckhbw mm2, mm0 ;
-
- pmullw mm1, [rax] ;
- pmullw mm2, [rax] ;
-
- punpcklbw mm3, mm0 ;
- punpckhbw mm4, mm0 ;
-
- pmullw mm3, [rax+8] ;
- pmullw mm4, [rax+8] ;
-
- paddw mm1, mm3 ;
- paddw mm2, mm4 ;
-
- paddw mm1, [GLOBAL(mmx_bi_rd)] ;
- psraw mm1, mmx_filter_shift ;
-
- paddw mm2, [GLOBAL(mmx_bi_rd)] ;
- psraw mm2, mmx_filter_shift ;
-
- movq mm3, mm5 ;
- movq mm4, mm5 ;
-
- punpcklbw mm3, mm0 ;
- punpckhbw mm4, mm0 ;
-
- movq mm5, mm1 ;
- packuswb mm5, mm2 ;
-
- pmullw mm3, [rdx] ;
- pmullw mm4, [rdx] ;
-
- pmullw mm1, [rdx+8] ;
- pmullw mm2, [rdx+8] ;
-
- paddw mm1, mm3 ;
- paddw mm2, mm4 ;
-
- paddw mm1, [GLOBAL(mmx_bi_rd)] ;
- paddw mm2, [GLOBAL(mmx_bi_rd)] ;
-
- psraw mm1, mmx_filter_shift ;
- psraw mm2, mmx_filter_shift ;
-
- movq mm3, [rdi] ;
- movq mm4, mm3 ;
-
- punpcklbw mm3, mm0 ;
- punpckhbw mm4, mm0 ;
-
- psubw mm1, mm3 ;
- psubw mm2, mm4 ;
-
- paddw mm6, mm1 ;
- pmaddwd mm1, mm1 ;
-
- paddw mm6, mm2 ;
- pmaddwd mm2, mm2 ;
-
- paddd mm7, mm1 ;
- paddd mm7, mm2 ;
-
-%if ABI_IS_32BIT
- add rsi, dword ptr arg(1) ;ref_pixels_per_line ;
- add rdi, dword ptr arg(3) ;src_pixels_per_line ;
-%else
- movsxd r8, dword ptr arg(1) ;ref_pixels_per_line ;
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line ;
- add rsi, r8
- add rdi, r9
-%endif
- sub rcx, 1 ;
- jnz filter_block2d_bil_var_mmx_loop ;
-
-
- pxor mm3, mm3 ;
- pxor mm2, mm2 ;
-
- punpcklwd mm2, mm6 ;
- punpckhwd mm3, mm6 ;
-
- paddd mm2, mm3 ;
- movq mm6, mm2 ;
-
- psrlq mm6, 32 ;
- paddd mm2, mm6 ;
-
- psrad mm2, 16 ;
- movq mm4, mm7 ;
-
- psrlq mm4, 32 ;
- paddd mm4, mm7 ;
-
- mov rdi, arg(7) ;sum
- mov rsi, arg(8) ;sumsquared
-
- movd dword ptr [rdi], mm2 ;
- movd dword ptr [rsi], mm4 ;
-
- ; begin epilog
- add rsp, 16
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;unsigned int vp8_get16x16pred_error_mmx
-;(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride
-;)
-global sym(vp8_get16x16pred_error_mmx)
-sym(vp8_get16x16pred_error_mmx):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- GET_GOT rbx
- push rsi
- push rdi
- sub rsp, 16
- ; end prolog
-
- mov rsi, arg(0) ;DWORD PTR [src_ptr]
- mov rdi, arg(2) ;DWORD PTR [ref_ptr]
-
- movsxd rax, DWORD PTR arg(1) ;[src_stride]
- movsxd rdx, DWORD PTR arg(3) ;[ref_stride]
-
- pxor mm0, mm0 ; clear xmm0 for unpack
- pxor mm7, mm7 ; clear xmm7 for accumulating diffs
-
- pxor mm6, mm6 ; clear xmm6 for accumulating sse
- mov rcx, 16
-
-var16loop:
-
- movq mm1, [rsi]
- movq mm2, [rdi]
-
- movq mm3, mm1
- movq mm4, mm2
-
- punpcklbw mm1, mm0
- punpckhbw mm3, mm0
-
- punpcklbw mm2, mm0
- punpckhbw mm4, mm0
-
- psubw mm1, mm2
- psubw mm3, mm4
-
- paddw mm7, mm1
- pmaddwd mm1, mm1
-
- paddw mm7, mm3
- pmaddwd mm3, mm3
-
- paddd mm6, mm1
- paddd mm6, mm3
-
-
- movq mm1, [rsi+8]
- movq mm2, [rdi+8]
-
- movq mm3, mm1
- movq mm4, mm2
-
- punpcklbw mm1, mm0
- punpckhbw mm3, mm0
-
- punpcklbw mm2, mm0
- punpckhbw mm4, mm0
-
- psubw mm1, mm2
- psubw mm3, mm4
-
- paddw mm7, mm1
- pmaddwd mm1, mm1
-
- paddw mm7, mm3
- pmaddwd mm3, mm3
-
- paddd mm6, mm1
- paddd mm6, mm3
-
- add rsi, rax
- add rdi, rdx
-
- sub rcx, 1
- jnz var16loop
-
-
- movq mm1, mm6
- pxor mm6, mm6
-
- pxor mm5, mm5
- punpcklwd mm6, mm7
-
- punpckhwd mm5, mm7
- psrad mm5, 16
-
- psrad mm6, 16
- paddd mm6, mm5
-
- movq mm2, mm1
- psrlq mm1, 32
-
- paddd mm2, mm1
- movq mm7, mm6
-
- psrlq mm6, 32
- paddd mm6, mm7
-
- movd DWORD PTR [rsp], mm6 ;Sum
- movd DWORD PTR [rsp+4], mm2 ;SSE
-
- ; return (SSE-((Sum*Sum)>>8));
- movsxd rdx, dword ptr [rsp]
- imul rdx, rdx
- sar rdx, 8
- movsxd rax, dword ptr [rsp + 4]
- sub rax, rdx
-
-
- ; begin epilog
- add rsp, 16
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-SECTION_RODATA
-;short mmx_bi_rd[4] = { 64, 64, 64, 64};
-align 16
-mmx_bi_rd:
- times 4 dw 64
diff --git a/vp8/encoder/x86/variance_impl_sse2.asm b/vp8/encoder/x86/variance_impl_sse2.asm
deleted file mode 100644
index c2c30de..0000000
--- a/vp8/encoder/x86/variance_impl_sse2.asm
+++ /dev/null
@@ -1,1463 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%define xmm_filter_shift 7
-
-;unsigned int vp8_get_mb_ss_sse2
-;(
-; short *src_ptr
-;)
-global sym(vp8_get_mb_ss_sse2)
-sym(vp8_get_mb_ss_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 1
- GET_GOT rbx
- push rsi
- push rdi
- sub rsp, 16
- ; end prolog
-
-
- mov rax, arg(0) ;[src_ptr]
- mov rcx, 8
- pxor xmm4, xmm4
-
-NEXTROW:
- movdqa xmm0, [rax]
- movdqa xmm1, [rax+16]
- movdqa xmm2, [rax+32]
- movdqa xmm3, [rax+48]
- pmaddwd xmm0, xmm0
- pmaddwd xmm1, xmm1
- pmaddwd xmm2, xmm2
- pmaddwd xmm3, xmm3
-
- paddd xmm0, xmm1
- paddd xmm2, xmm3
- paddd xmm4, xmm0
- paddd xmm4, xmm2
-
- add rax, 0x40
- dec rcx
- ja NEXTROW
-
- movdqa xmm3,xmm4
- psrldq xmm4,8
- paddd xmm4,xmm3
- movdqa xmm3,xmm4
- psrldq xmm4,4
- paddd xmm4,xmm3
- movq rax,xmm4
-
-
- ; begin epilog
- add rsp, 16
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_get16x16var_sse2
-;(
-; unsigned char * src_ptr,
-; int source_stride,
-; unsigned char * ref_ptr,
-; int recon_stride,
-; unsigned int * SSE,
-; int * Sum
-;)
-global sym(vp8_get16x16var_sse2)
-sym(vp8_get16x16var_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- push rbx
- push rsi
- push rdi
- ; end prolog
-
- mov rsi, arg(0) ;[src_ptr]
- mov rdi, arg(2) ;[ref_ptr]
-
- movsxd rax, DWORD PTR arg(1) ;[source_stride]
- movsxd rdx, DWORD PTR arg(3) ;[recon_stride]
-
- ; Prefetch data
- lea rcx, [rax+rax*2]
- prefetcht0 [rsi]
- prefetcht0 [rsi+rax]
- prefetcht0 [rsi+rax*2]
- prefetcht0 [rsi+rcx]
- lea rbx, [rsi+rax*4]
- prefetcht0 [rbx]
- prefetcht0 [rbx+rax]
- prefetcht0 [rbx+rax*2]
- prefetcht0 [rbx+rcx]
-
- lea rcx, [rdx+rdx*2]
- prefetcht0 [rdi]
- prefetcht0 [rdi+rdx]
- prefetcht0 [rdi+rdx*2]
- prefetcht0 [rdi+rcx]
- lea rbx, [rdi+rdx*4]
- prefetcht0 [rbx]
- prefetcht0 [rbx+rdx]
- prefetcht0 [rbx+rdx*2]
- prefetcht0 [rbx+rcx]
-
- pxor xmm0, xmm0 ; clear xmm0 for unpack
- pxor xmm7, xmm7 ; clear xmm7 for accumulating diffs
-
- pxor xmm6, xmm6 ; clear xmm6 for accumulating sse
- mov rcx, 16
-
-var16loop:
- movdqu xmm1, XMMWORD PTR [rsi]
- movdqu xmm2, XMMWORD PTR [rdi]
-
- prefetcht0 [rsi+rax*8]
- prefetcht0 [rdi+rdx*8]
-
- movdqa xmm3, xmm1
- movdqa xmm4, xmm2
-
-
- punpcklbw xmm1, xmm0
- punpckhbw xmm3, xmm0
-
- punpcklbw xmm2, xmm0
- punpckhbw xmm4, xmm0
-
-
- psubw xmm1, xmm2
- psubw xmm3, xmm4
-
- paddw xmm7, xmm1
- pmaddwd xmm1, xmm1
-
- paddw xmm7, xmm3
- pmaddwd xmm3, xmm3
-
- paddd xmm6, xmm1
- paddd xmm6, xmm3
-
- add rsi, rax
- add rdi, rdx
-
- sub rcx, 1
- jnz var16loop
-
-
- movdqa xmm1, xmm6
- pxor xmm6, xmm6
-
- pxor xmm5, xmm5
- punpcklwd xmm6, xmm7
-
- punpckhwd xmm5, xmm7
- psrad xmm5, 16
-
- psrad xmm6, 16
- paddd xmm6, xmm5
-
- movdqa xmm2, xmm1
- punpckldq xmm1, xmm0
-
- punpckhdq xmm2, xmm0
- movdqa xmm7, xmm6
-
- paddd xmm1, xmm2
- punpckldq xmm6, xmm0
-
- punpckhdq xmm7, xmm0
- paddd xmm6, xmm7
-
- movdqa xmm2, xmm1
- movdqa xmm7, xmm6
-
- psrldq xmm1, 8
- psrldq xmm6, 8
-
- paddd xmm7, xmm6
- paddd xmm1, xmm2
-
- mov rax, arg(5) ;[Sum]
- mov rdi, arg(4) ;[SSE]
-
- movd DWORD PTR [rax], xmm7
- movd DWORD PTR [rdi], xmm1
-
-
- ; begin epilog
- pop rdi
- pop rsi
- pop rbx
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;unsigned int vp8_get16x16pred_error_sse2
-;(
-; unsigned char *src_ptr,
-; int src_stride,
-; unsigned char *ref_ptr,
-; int ref_stride
-;)
-global sym(vp8_get16x16pred_error_sse2)
-sym(vp8_get16x16pred_error_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 4
- GET_GOT rbx
- push rsi
- push rdi
- sub rsp, 16
- ; end prolog
-
- mov rsi, arg(0) ;[src_ptr]
- mov rdi, arg(2) ;[ref_ptr]
-
- movsxd rax, DWORD PTR arg(1) ;[src_stride]
- movsxd rdx, DWORD PTR arg(3) ;[ref_stride]
-
- pxor xmm0, xmm0 ; clear xmm0 for unpack
- pxor xmm7, xmm7 ; clear xmm7 for accumulating diffs
-
- pxor xmm6, xmm6 ; clear xmm6 for accumulating sse
- mov rcx, 16
-
-var16peloop:
- movdqu xmm1, XMMWORD PTR [rsi]
- movdqu xmm2, XMMWORD PTR [rdi]
-
- movdqa xmm3, xmm1
- movdqa xmm4, xmm2
-
- punpcklbw xmm1, xmm0
- punpckhbw xmm3, xmm0
-
- punpcklbw xmm2, xmm0
- punpckhbw xmm4, xmm0
-
- psubw xmm1, xmm2
- psubw xmm3, xmm4
-
- paddw xmm7, xmm1
- pmaddwd xmm1, xmm1
-
- paddw xmm7, xmm3
- pmaddwd xmm3, xmm3
-
- paddd xmm6, xmm1
- paddd xmm6, xmm3
-
- add rsi, rax
- add rdi, rdx
-
- sub rcx, 1
- jnz var16peloop
-
-
- movdqa xmm1, xmm6
- pxor xmm6, xmm6
-
- pxor xmm5, xmm5
- punpcklwd xmm6, xmm7
-
- punpckhwd xmm5, xmm7
- psrad xmm5, 16
-
- psrad xmm6, 16
- paddd xmm6, xmm5
-
- movdqa xmm2, xmm1
- punpckldq xmm1, xmm0
-
- punpckhdq xmm2, xmm0
- movdqa xmm7, xmm6
-
- paddd xmm1, xmm2
- punpckldq xmm6, xmm0
-
- punpckhdq xmm7, xmm0
- paddd xmm6, xmm7
-
- movdqa xmm2, xmm1
- movdqa xmm7, xmm6
-
- psrldq xmm1, 8
- psrldq xmm6, 8
-
- paddd xmm7, xmm6
- paddd xmm1, xmm2
-
- movd DWORD PTR [rsp], xmm7 ;Sum
- movd DWORD PTR [rsp+4], xmm1 ;SSE
-
- ; return (SSE-((Sum*Sum)>>8));
- movsxd rdx, dword ptr [rsp]
- imul rdx, rdx
- sar rdx, 8
- movsxd rax, dword ptr [rsp + 4]
- sub rax, rdx
-
- ; begin epilog
- add rsp, 16
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-
-;unsigned int vp8_get8x8var_sse2
-;(
-; unsigned char * src_ptr,
-; int source_stride,
-; unsigned char * ref_ptr,
-; int recon_stride,
-; unsigned int * SSE,
-; int * Sum
-;)
-global sym(vp8_get8x8var_sse2)
-sym(vp8_get8x8var_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 6
- GET_GOT rbx
- push rsi
- push rdi
- sub rsp, 16
- ; end prolog
-
- mov rsi, arg(0) ;[src_ptr]
- mov rdi, arg(2) ;[ref_ptr]
-
- movsxd rax, DWORD PTR arg(1) ;[source_stride]
- movsxd rdx, DWORD PTR arg(3) ;[recon_stride]
-
- pxor xmm0, xmm0 ; clear xmm0 for unpack
- pxor xmm7, xmm7 ; clear xmm7 for accumulating diffs
-
- movq xmm1, QWORD PTR [rsi]
- movq xmm2, QWORD PTR [rdi]
-
- punpcklbw xmm1, xmm0
- punpcklbw xmm2, xmm0
-
- psubsw xmm1, xmm2
- paddw xmm7, xmm1
-
- pmaddwd xmm1, xmm1
-
- movq xmm2, QWORD PTR[rsi + rax]
- movq xmm3, QWORD PTR[rdi + rdx]
-
- punpcklbw xmm2, xmm0
- punpcklbw xmm3, xmm0
-
- psubsw xmm2, xmm3
- paddw xmm7, xmm2
-
- pmaddwd xmm2, xmm2
- paddd xmm1, xmm2
-
-
- movq xmm2, QWORD PTR[rsi + rax * 2]
- movq xmm3, QWORD PTR[rdi + rdx * 2]
-
- punpcklbw xmm2, xmm0
- punpcklbw xmm3, xmm0
-
- psubsw xmm2, xmm3
- paddw xmm7, xmm2
-
- pmaddwd xmm2, xmm2
- paddd xmm1, xmm2
-
-
- lea rsi, [rsi + rax * 2]
- lea rdi, [rdi + rdx * 2]
- movq xmm2, QWORD PTR[rsi + rax]
- movq xmm3, QWORD PTR[rdi + rdx]
-
- punpcklbw xmm2, xmm0
- punpcklbw xmm3, xmm0
-
- psubsw xmm2, xmm3
- paddw xmm7, xmm2
-
- pmaddwd xmm2, xmm2
- paddd xmm1, xmm2
-
- movq xmm2, QWORD PTR[rsi + rax *2]
- movq xmm3, QWORD PTR[rdi + rdx *2]
-
- punpcklbw xmm2, xmm0
- punpcklbw xmm3, xmm0
-
- psubsw xmm2, xmm3
- paddw xmm7, xmm2
-
- pmaddwd xmm2, xmm2
- paddd xmm1, xmm2
-
-
- lea rsi, [rsi + rax * 2]
- lea rdi, [rdi + rdx * 2]
-
-
- movq xmm2, QWORD PTR[rsi + rax]
- movq xmm3, QWORD PTR[rdi + rdx]
-
- punpcklbw xmm2, xmm0
- punpcklbw xmm3, xmm0
-
- psubsw xmm2, xmm3
- paddw xmm7, xmm2
-
- pmaddwd xmm2, xmm2
- paddd xmm1, xmm2
-
- movq xmm2, QWORD PTR[rsi + rax *2]
- movq xmm3, QWORD PTR[rdi + rdx *2]
-
- punpcklbw xmm2, xmm0
- punpcklbw xmm3, xmm0
-
- psubsw xmm2, xmm3
- paddw xmm7, xmm2
-
- pmaddwd xmm2, xmm2
- paddd xmm1, xmm2
-
-
- lea rsi, [rsi + rax * 2]
- lea rdi, [rdi + rdx * 2]
-
- movq xmm2, QWORD PTR[rsi + rax]
- movq xmm3, QWORD PTR[rdi + rdx]
-
- punpcklbw xmm2, xmm0
- punpcklbw xmm3, xmm0
-
- psubsw xmm2, xmm3
- paddw xmm7, xmm2
-
- pmaddwd xmm2, xmm2
- paddd xmm1, xmm2
-
-
- movdqa xmm6, xmm7
- punpcklwd xmm6, xmm0
-
- punpckhwd xmm7, xmm0
- movdqa xmm2, xmm1
-
- paddw xmm6, xmm7
- punpckldq xmm1, xmm0
-
- punpckhdq xmm2, xmm0
- movdqa xmm7, xmm6
-
- paddd xmm1, xmm2
- punpckldq xmm6, xmm0
-
- punpckhdq xmm7, xmm0
- paddw xmm6, xmm7
-
- movdqa xmm2, xmm1
- movdqa xmm7, xmm6
-
- psrldq xmm1, 8
- psrldq xmm6, 8
-
- paddw xmm7, xmm6
- paddd xmm1, xmm2
-
- mov rax, arg(5) ;[Sum]
- mov rdi, arg(4) ;[SSE]
-
- movq rdx, xmm7
- movsx rcx, dx
-
- mov dword ptr [rax], ecx
- movd DWORD PTR [rdi], xmm1
-
- ; begin epilog
- add rsp, 16
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_filter_block2d_bil_var_sse2
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; int xoffset,
-; int yoffset,
-; int *sum,
-; unsigned int *sumsquared;;
-;
-;)
-global sym(vp8_filter_block2d_bil_var_sse2)
-sym(vp8_filter_block2d_bil_var_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 9
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- push rbx
- ; end prolog
-
- pxor xmm6, xmm6 ;
- pxor xmm7, xmm7 ;
-
- lea rsi, [GLOBAL(xmm_bi_rd)] ; rounding
- movdqa xmm4, XMMWORD PTR [rsi]
-
- lea rcx, [GLOBAL(vp8_bilinear_filters_sse2)]
- movsxd rax, dword ptr arg(5) ; xoffset
-
- cmp rax, 0 ; skip first_pass filter if xoffset=0
- je filter_block2d_bil_var_sse2_sp_only
-
- shl rax, 5 ; point to filter coeff with xoffset
- lea rax, [rax + rcx] ; HFilter
-
- movsxd rdx, dword ptr arg(6) ; yoffset
-
- cmp rdx, 0 ; skip second_pass filter if yoffset=0
- je filter_block2d_bil_var_sse2_fp_only
-
- shl rdx, 5
- lea rdx, [rdx + rcx] ; VFilter
-
- mov rsi, arg(0) ;ref_ptr
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
-
- pxor xmm0, xmm0 ;
- movq xmm1, QWORD PTR [rsi] ;
- movq xmm3, QWORD PTR [rsi+1] ;
-
- punpcklbw xmm1, xmm0 ;
- pmullw xmm1, [rax] ;
- punpcklbw xmm3, xmm0
- pmullw xmm3, [rax+16] ;
-
- paddw xmm1, xmm3 ;
- paddw xmm1, xmm4 ;
- psraw xmm1, xmm_filter_shift ;
- movdqa xmm5, xmm1
-
- movsxd rbx, dword ptr arg(1) ;ref_pixels_per_line
- lea rsi, [rsi + rbx]
-%if ABI_IS_32BIT=0
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line
-%endif
-
-filter_block2d_bil_var_sse2_loop:
- movq xmm1, QWORD PTR [rsi] ;
- movq xmm3, QWORD PTR [rsi+1] ;
-
- punpcklbw xmm1, xmm0 ;
- pmullw xmm1, [rax] ;
- punpcklbw xmm3, xmm0 ;
- pmullw xmm3, [rax+16] ;
-
- paddw xmm1, xmm3 ;
- paddw xmm1, xmm4 ;
- psraw xmm1, xmm_filter_shift ;
-
- movdqa xmm3, xmm5 ;
- movdqa xmm5, xmm1 ;
-
- pmullw xmm3, [rdx] ;
- pmullw xmm1, [rdx+16] ;
- paddw xmm1, xmm3 ;
- paddw xmm1, xmm4 ;
- psraw xmm1, xmm_filter_shift ;
-
- movq xmm3, QWORD PTR [rdi] ;
- punpcklbw xmm3, xmm0 ;
-
- psubw xmm1, xmm3 ;
- paddw xmm6, xmm1 ;
-
- pmaddwd xmm1, xmm1 ;
- paddd xmm7, xmm1 ;
-
- lea rsi, [rsi + rbx] ;ref_pixels_per_line
-%if ABI_IS_32BIT
- add rdi, dword ptr arg(3) ;src_pixels_per_line
-%else
- lea rdi, [rdi + r9]
-%endif
-
- sub rcx, 1 ;
- jnz filter_block2d_bil_var_sse2_loop ;
-
- jmp filter_block2d_bil_variance
-
-filter_block2d_bil_var_sse2_sp_only:
- movsxd rdx, dword ptr arg(6) ; yoffset
-
- cmp rdx, 0 ; skip all if both xoffset=0 and yoffset=0
- je filter_block2d_bil_var_sse2_full_pixel
-
- shl rdx, 5
- lea rdx, [rdx + rcx] ; VFilter
-
- mov rsi, arg(0) ;ref_ptr
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
-
- pxor xmm0, xmm0 ;
- movq xmm1, QWORD PTR [rsi] ;
- punpcklbw xmm1, xmm0 ;
-
- movsxd rbx, dword ptr arg(3) ;src_pixels_per_line
- lea rsi, [rsi + rax]
-
-filter_block2d_bil_sp_only_loop:
- movq xmm3, QWORD PTR [rsi] ;
- punpcklbw xmm3, xmm0 ;
- movdqa xmm5, xmm3
-
- pmullw xmm1, [rdx] ;
- pmullw xmm3, [rdx+16] ;
- paddw xmm1, xmm3 ;
- paddw xmm1, xmm4 ;
- psraw xmm1, xmm_filter_shift ;
-
- movq xmm3, QWORD PTR [rdi] ;
- punpcklbw xmm3, xmm0 ;
-
- psubw xmm1, xmm3 ;
- paddw xmm6, xmm1 ;
-
- pmaddwd xmm1, xmm1 ;
- paddd xmm7, xmm1 ;
-
- movdqa xmm1, xmm5 ;
- lea rsi, [rsi + rax] ;ref_pixels_per_line
- lea rdi, [rdi + rbx] ;src_pixels_per_line
-
- sub rcx, 1 ;
- jnz filter_block2d_bil_sp_only_loop ;
-
- jmp filter_block2d_bil_variance
-
-filter_block2d_bil_var_sse2_full_pixel:
- mov rsi, arg(0) ;ref_ptr
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
- movsxd rbx, dword ptr arg(3) ;src_pixels_per_line
- pxor xmm0, xmm0 ;
-
-filter_block2d_bil_full_pixel_loop:
- movq xmm1, QWORD PTR [rsi] ;
- punpcklbw xmm1, xmm0 ;
-
- movq xmm2, QWORD PTR [rdi] ;
- punpcklbw xmm2, xmm0 ;
-
- psubw xmm1, xmm2 ;
- paddw xmm6, xmm1 ;
-
- pmaddwd xmm1, xmm1 ;
- paddd xmm7, xmm1 ;
-
- lea rsi, [rsi + rax] ;ref_pixels_per_line
- lea rdi, [rdi + rbx] ;src_pixels_per_line
-
- sub rcx, 1 ;
- jnz filter_block2d_bil_full_pixel_loop ;
-
- jmp filter_block2d_bil_variance
-
-filter_block2d_bil_var_sse2_fp_only:
- mov rsi, arg(0) ;ref_ptr
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
- movsxd rdx, dword ptr arg(1) ;ref_pixels_per_line
-
- pxor xmm0, xmm0 ;
- movsxd rbx, dword ptr arg(3) ;src_pixels_per_line
-
-filter_block2d_bil_fp_only_loop:
- movq xmm1, QWORD PTR [rsi] ;
- movq xmm3, QWORD PTR [rsi+1] ;
-
- punpcklbw xmm1, xmm0 ;
- pmullw xmm1, [rax] ;
- punpcklbw xmm3, xmm0 ;
- pmullw xmm3, [rax+16] ;
-
- paddw xmm1, xmm3 ;
- paddw xmm1, xmm4 ;
- psraw xmm1, xmm_filter_shift ;
-
- movq xmm3, QWORD PTR [rdi] ;
- punpcklbw xmm3, xmm0 ;
-
- psubw xmm1, xmm3 ;
- paddw xmm6, xmm1 ;
-
- pmaddwd xmm1, xmm1 ;
- paddd xmm7, xmm1 ;
- lea rsi, [rsi + rdx]
- lea rdi, [rdi + rbx] ;src_pixels_per_line
-
- sub rcx, 1 ;
- jnz filter_block2d_bil_fp_only_loop ;
-
- jmp filter_block2d_bil_variance
-
-filter_block2d_bil_variance:
- movdq2q mm6, xmm6 ;
- movdq2q mm7, xmm7 ;
-
- psrldq xmm6, 8
- psrldq xmm7, 8
-
- movdq2q mm2, xmm6
- movdq2q mm3, xmm7
-
- paddw mm6, mm2
- paddd mm7, mm3
-
- pxor mm3, mm3 ;
- pxor mm2, mm2 ;
-
- punpcklwd mm2, mm6 ;
- punpckhwd mm3, mm6 ;
-
- paddd mm2, mm3 ;
- movq mm6, mm2 ;
-
- psrlq mm6, 32 ;
- paddd mm2, mm6 ;
-
- psrad mm2, 16 ;
- movq mm4, mm7 ;
-
- psrlq mm4, 32 ;
- paddd mm4, mm7 ;
-
- mov rsi, arg(7) ; sum
- mov rdi, arg(8) ; sumsquared
-
- movd [rsi], mm2 ; xsum
- movd [rdi], mm4 ; xxsum
-
- ; begin epilog
- pop rbx
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_half_horiz_vert_variance8x_h_sse2
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; int *sum,
-; unsigned int *sumsquared
-;)
-global sym(vp8_half_horiz_vert_variance8x_h_sse2)
-sym(vp8_half_horiz_vert_variance8x_h_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line
-%endif
-
- pxor xmm6, xmm6 ; error accumulator
- pxor xmm7, xmm7 ; sse eaccumulator
- mov rsi, arg(0) ;ref_ptr ;
-
- mov rdi, arg(2) ;src_ptr ;
- movsxd rcx, dword ptr arg(4) ;Height ;
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
-
- pxor xmm0, xmm0 ;
-
- movq xmm5, QWORD PTR [rsi] ; xmm5 = s0,s1,s2..s8
- movq xmm3, QWORD PTR [rsi+1] ; xmm3 = s1,s2,s3..s9
- pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3) horizontal line 1
-
-%if ABI_IS_32BIT
- add rsi, dword ptr arg(1) ;ref_pixels_per_line ; next source
-%else
- add rsi, r8
-%endif
-
-vp8_half_horiz_vert_variance8x_h_1:
-
- movq xmm1, QWORD PTR [rsi] ;
- movq xmm2, QWORD PTR [rsi+1] ;
- pavgb xmm1, xmm2 ; xmm1 = avg(xmm1,xmm3) horizontal line i+1
-
- pavgb xmm5, xmm1 ; xmm = vertical average of the above
- punpcklbw xmm5, xmm0 ; xmm5 = words of above
-
- movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d8
- punpcklbw xmm3, xmm0 ; xmm3 = words of above
-
- psubw xmm5, xmm3 ; xmm5 -= xmm3
- paddw xmm6, xmm5 ; xmm6 += accumulated column differences
- pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
- paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
-
- movdqa xmm5, xmm1 ; save xmm1 for use on the next row
-
-%if ABI_IS_32BIT
- add esi, dword ptr arg(1) ;ref_pixels_per_line ; next source
- add edi, dword ptr arg(3) ;src_pixels_per_line ; next destination
-%else
- add rsi, r8
- add rdi, r9
-%endif
-
- sub rcx, 1 ;
- jnz vp8_half_horiz_vert_variance8x_h_1 ;
-
- movdq2q mm6, xmm6 ;
- movdq2q mm7, xmm7 ;
-
- psrldq xmm6, 8
- psrldq xmm7, 8
-
- movdq2q mm2, xmm6
- movdq2q mm3, xmm7
-
- paddw mm6, mm2
- paddd mm7, mm3
-
- pxor mm3, mm3 ;
- pxor mm2, mm2 ;
-
- punpcklwd mm2, mm6 ;
- punpckhwd mm3, mm6 ;
-
- paddd mm2, mm3 ;
- movq mm6, mm2 ;
-
- psrlq mm6, 32 ;
- paddd mm2, mm6 ;
-
- psrad mm2, 16 ;
- movq mm4, mm7 ;
-
- psrlq mm4, 32 ;
- paddd mm4, mm7 ;
-
- mov rsi, arg(5) ; sum
- mov rdi, arg(6) ; sumsquared
-
- movd [rsi], mm2 ;
- movd [rdi], mm4 ;
-
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_half_horiz_vert_variance16x_h_sse2
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; int *sum,
-; unsigned int *sumsquared
-;)
-global sym(vp8_half_horiz_vert_variance16x_h_sse2)
-sym(vp8_half_horiz_vert_variance16x_h_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- pxor xmm6, xmm6 ; error accumulator
- pxor xmm7, xmm7 ; sse eaccumulator
- mov rsi, arg(0) ;ref_ptr ;
-
- mov rdi, arg(2) ;src_ptr ;
- movsxd rcx, dword ptr arg(4) ;Height ;
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
- movsxd rdx, dword ptr arg(3) ;src_pixels_per_line
-
- pxor xmm0, xmm0 ;
-
- movdqu xmm5, XMMWORD PTR [rsi]
- movdqu xmm3, XMMWORD PTR [rsi+1]
- pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3) horizontal line 1
-
- lea rsi, [rsi + rax]
-
-vp8_half_horiz_vert_variance16x_h_1:
- movdqu xmm1, XMMWORD PTR [rsi] ;
- movdqu xmm2, XMMWORD PTR [rsi+1] ;
- pavgb xmm1, xmm2 ; xmm1 = avg(xmm1,xmm3) horizontal line i+1
-
- pavgb xmm5, xmm1 ; xmm = vertical average of the above
-
- movdqa xmm4, xmm5
- punpcklbw xmm5, xmm0 ; xmm5 = words of above
- punpckhbw xmm4, xmm0
-
- movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d7
- punpcklbw xmm3, xmm0 ; xmm3 = words of above
- psubw xmm5, xmm3 ; xmm5 -= xmm3
-
- movq xmm3, QWORD PTR [rdi+8]
- punpcklbw xmm3, xmm0
- psubw xmm4, xmm3
-
- paddw xmm6, xmm5 ; xmm6 += accumulated column differences
- paddw xmm6, xmm4
- pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
- pmaddwd xmm4, xmm4
- paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
- paddd xmm7, xmm4
-
- movdqa xmm5, xmm1 ; save xmm1 for use on the next row
-
- lea rsi, [rsi + rax]
- lea rdi, [rdi + rdx]
-
- sub rcx, 1 ;
- jnz vp8_half_horiz_vert_variance16x_h_1 ;
-
- pxor xmm1, xmm1
- pxor xmm5, xmm5
-
- punpcklwd xmm0, xmm6
- punpckhwd xmm1, xmm6
- psrad xmm0, 16
- psrad xmm1, 16
- paddd xmm0, xmm1
- movdqa xmm1, xmm0
-
- movdqa xmm6, xmm7
- punpckldq xmm6, xmm5
- punpckhdq xmm7, xmm5
- paddd xmm6, xmm7
-
- punpckldq xmm0, xmm5
- punpckhdq xmm1, xmm5
- paddd xmm0, xmm1
-
- movdqa xmm7, xmm6
- movdqa xmm1, xmm0
-
- psrldq xmm7, 8
- psrldq xmm1, 8
-
- paddd xmm6, xmm7
- paddd xmm0, xmm1
-
- mov rsi, arg(5) ;[Sum]
- mov rdi, arg(6) ;[SSE]
-
- movd [rsi], xmm0
- movd [rdi], xmm6
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_half_vert_variance8x_h_sse2
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; int *sum,
-; unsigned int *sumsquared
-;)
-global sym(vp8_half_vert_variance8x_h_sse2)
-sym(vp8_half_vert_variance8x_h_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line
-%endif
-
- pxor xmm6, xmm6 ; error accumulator
- pxor xmm7, xmm7 ; sse eaccumulator
- mov rsi, arg(0) ;ref_ptr ;
-
- mov rdi, arg(2) ;src_ptr ;
- movsxd rcx, dword ptr arg(4) ;Height ;
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
-
- pxor xmm0, xmm0 ;
-vp8_half_vert_variance8x_h_1:
- movq xmm5, QWORD PTR [rsi] ; xmm5 = s0,s1,s2..s8
- movq xmm3, QWORD PTR [rsi+rax] ; xmm3 = s1,s2,s3..s9
-
- pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3)
- punpcklbw xmm5, xmm0 ; xmm5 = words of above
-
- movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d8
- punpcklbw xmm3, xmm0 ; xmm3 = words of above
-
- psubw xmm5, xmm3 ; xmm5 -= xmm3
- paddw xmm6, xmm5 ; xmm6 += accumulated column differences
- pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
- paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
-
-%if ABI_IS_32BIT
- add esi, dword ptr arg(1) ;ref_pixels_per_line ; next source
- add edi, dword ptr arg(3) ;src_pixels_per_line ; next destination
-%else
- add rsi, r8
- add rdi, r9
-%endif
-
- sub rcx, 1 ;
- jnz vp8_half_vert_variance8x_h_1 ;
-
- movdq2q mm6, xmm6 ;
- movdq2q mm7, xmm7 ;
-
- psrldq xmm6, 8
- psrldq xmm7, 8
-
- movdq2q mm2, xmm6
- movdq2q mm3, xmm7
-
- paddw mm6, mm2
- paddd mm7, mm3
-
- pxor mm3, mm3 ;
- pxor mm2, mm2 ;
-
- punpcklwd mm2, mm6 ;
- punpckhwd mm3, mm6 ;
-
- paddd mm2, mm3 ;
- movq mm6, mm2 ;
-
- psrlq mm6, 32 ;
- paddd mm2, mm6 ;
-
- psrad mm2, 16 ;
- movq mm4, mm7 ;
-
- psrlq mm4, 32 ;
- paddd mm4, mm7 ;
-
- mov rsi, arg(5) ; sum
- mov rdi, arg(6) ; sumsquared
-
- movd [rsi], mm2 ;
- movd [rdi], mm4 ;
-
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_half_vert_variance16x_h_sse2
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; int *sum,
-; unsigned int *sumsquared
-;)
-global sym(vp8_half_vert_variance16x_h_sse2)
-sym(vp8_half_vert_variance16x_h_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- pxor xmm6, xmm6 ; error accumulator
- pxor xmm7, xmm7 ; sse eaccumulator
- mov rsi, arg(0) ;ref_ptr
-
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
- movsxd rdx, dword ptr arg(3) ;src_pixels_per_line
-
- movdqu xmm5, XMMWORD PTR [rsi]
- lea rsi, [rsi + rax ]
- pxor xmm0, xmm0
-
-vp8_half_vert_variance16x_h_1:
- movdqu xmm3, XMMWORD PTR [rsi]
-
- pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3)
- movdqa xmm4, xmm5
- punpcklbw xmm5, xmm0
- punpckhbw xmm4, xmm0
-
- movq xmm2, QWORD PTR [rdi]
- punpcklbw xmm2, xmm0
- psubw xmm5, xmm2
- movq xmm2, QWORD PTR [rdi+8]
- punpcklbw xmm2, xmm0
- psubw xmm4, xmm2
-
- paddw xmm6, xmm5 ; xmm6 += accumulated column differences
- paddw xmm6, xmm4
- pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
- pmaddwd xmm4, xmm4
- paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
- paddd xmm7, xmm4
-
- movdqa xmm5, xmm3
-
- lea rsi, [rsi + rax]
- lea rdi, [rdi + rdx]
-
- sub rcx, 1
- jnz vp8_half_vert_variance16x_h_1
-
- pxor xmm1, xmm1
- pxor xmm5, xmm5
-
- punpcklwd xmm0, xmm6
- punpckhwd xmm1, xmm6
- psrad xmm0, 16
- psrad xmm1, 16
- paddd xmm0, xmm1
- movdqa xmm1, xmm0
-
- movdqa xmm6, xmm7
- punpckldq xmm6, xmm5
- punpckhdq xmm7, xmm5
- paddd xmm6, xmm7
-
- punpckldq xmm0, xmm5
- punpckhdq xmm1, xmm5
- paddd xmm0, xmm1
-
- movdqa xmm7, xmm6
- movdqa xmm1, xmm0
-
- psrldq xmm7, 8
- psrldq xmm1, 8
-
- paddd xmm6, xmm7
- paddd xmm0, xmm1
-
- mov rsi, arg(5) ;[Sum]
- mov rdi, arg(6) ;[SSE]
-
- movd [rsi], xmm0
- movd [rdi], xmm6
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-;void vp8_half_horiz_variance8x_h_sse2
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; int *sum,
-; unsigned int *sumsquared
-;)
-global sym(vp8_half_horiz_variance8x_h_sse2)
-sym(vp8_half_horiz_variance8x_h_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
-%if ABI_IS_32BIT=0
- movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line
-%endif
-
- pxor xmm6, xmm6 ; error accumulator
- pxor xmm7, xmm7 ; sse eaccumulator
- mov rsi, arg(0) ;ref_ptr ;
-
- mov rdi, arg(2) ;src_ptr ;
- movsxd rcx, dword ptr arg(4) ;Height ;
-
- pxor xmm0, xmm0 ;
-vp8_half_horiz_variance8x_h_1:
- movq xmm5, QWORD PTR [rsi] ; xmm5 = s0,s1,s2..s8
- movq xmm3, QWORD PTR [rsi+1] ; xmm3 = s1,s2,s3..s9
-
- pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3)
- punpcklbw xmm5, xmm0 ; xmm5 = words of above
-
- movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d8
- punpcklbw xmm3, xmm0 ; xmm3 = words of above
-
- psubw xmm5, xmm3 ; xmm5 -= xmm3
- paddw xmm6, xmm5 ; xmm6 += accumulated column differences
- pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
- paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
-
-%if ABI_IS_32BIT
- add esi, dword ptr arg(1) ;ref_pixels_per_line ; next source
- add edi, dword ptr arg(3) ;src_pixels_per_line ; next destination
-%else
- add rsi, r8
- add rdi, r9
-%endif
- sub rcx, 1 ;
- jnz vp8_half_horiz_variance8x_h_1 ;
-
- movdq2q mm6, xmm6 ;
- movdq2q mm7, xmm7 ;
-
- psrldq xmm6, 8
- psrldq xmm7, 8
-
- movdq2q mm2, xmm6
- movdq2q mm3, xmm7
-
- paddw mm6, mm2
- paddd mm7, mm3
-
- pxor mm3, mm3 ;
- pxor mm2, mm2 ;
-
- punpcklwd mm2, mm6 ;
- punpckhwd mm3, mm6 ;
-
- paddd mm2, mm3 ;
- movq mm6, mm2 ;
-
- psrlq mm6, 32 ;
- paddd mm2, mm6 ;
-
- psrad mm2, 16 ;
- movq mm4, mm7 ;
-
- psrlq mm4, 32 ;
- paddd mm4, mm7 ;
-
- mov rsi, arg(5) ; sum
- mov rdi, arg(6) ; sumsquared
-
- movd [rsi], mm2 ;
- movd [rdi], mm4 ;
-
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- UNSHADOW_ARGS
- pop rbp
- ret
-
-;void vp8_half_horiz_variance16x_h_sse2
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; int *sum,
-; unsigned int *sumsquared
-;)
-global sym(vp8_half_horiz_variance16x_h_sse2)
-sym(vp8_half_horiz_variance16x_h_sse2):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 7
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- pxor xmm6, xmm6 ; error accumulator
- pxor xmm7, xmm7 ; sse eaccumulator
- mov rsi, arg(0) ;ref_ptr ;
-
- mov rdi, arg(2) ;src_ptr ;
- movsxd rcx, dword ptr arg(4) ;Height ;
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
- movsxd rdx, dword ptr arg(3) ;src_pixels_per_line
-
- pxor xmm0, xmm0 ;
-
-vp8_half_horiz_variance16x_h_1:
- movdqu xmm5, XMMWORD PTR [rsi] ; xmm5 = s0,s1,s2..s15
- movdqu xmm3, XMMWORD PTR [rsi+1] ; xmm3 = s1,s2,s3..s16
-
- pavgb xmm5, xmm3 ; xmm5 = avg(xmm1,xmm3)
- movdqa xmm1, xmm5
- punpcklbw xmm5, xmm0 ; xmm5 = words of above
- punpckhbw xmm1, xmm0
-
- movq xmm3, QWORD PTR [rdi] ; xmm3 = d0,d1,d2..d7
- punpcklbw xmm3, xmm0 ; xmm3 = words of above
- movq xmm2, QWORD PTR [rdi+8]
- punpcklbw xmm2, xmm0
-
- psubw xmm5, xmm3 ; xmm5 -= xmm3
- psubw xmm1, xmm2
- paddw xmm6, xmm5 ; xmm6 += accumulated column differences
- paddw xmm6, xmm1
- pmaddwd xmm5, xmm5 ; xmm5 *= xmm5
- pmaddwd xmm1, xmm1
- paddd xmm7, xmm5 ; xmm7 += accumulated square column differences
- paddd xmm7, xmm1
-
- lea rsi, [rsi + rax]
- lea rdi, [rdi + rdx]
-
- sub rcx, 1 ;
- jnz vp8_half_horiz_variance16x_h_1 ;
-
- pxor xmm1, xmm1
- pxor xmm5, xmm5
-
- punpcklwd xmm0, xmm6
- punpckhwd xmm1, xmm6
- psrad xmm0, 16
- psrad xmm1, 16
- paddd xmm0, xmm1
- movdqa xmm1, xmm0
-
- movdqa xmm6, xmm7
- punpckldq xmm6, xmm5
- punpckhdq xmm7, xmm5
- paddd xmm6, xmm7
-
- punpckldq xmm0, xmm5
- punpckhdq xmm1, xmm5
- paddd xmm0, xmm1
-
- movdqa xmm7, xmm6
- movdqa xmm1, xmm0
-
- psrldq xmm7, 8
- psrldq xmm1, 8
-
- paddd xmm6, xmm7
- paddd xmm0, xmm1
-
- mov rsi, arg(5) ;[Sum]
- mov rdi, arg(6) ;[SSE]
-
- movd [rsi], xmm0
- movd [rdi], xmm6
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-SECTION_RODATA
-; short xmm_bi_rd[8] = { 64, 64, 64, 64,64, 64, 64, 64};
-align 16
-xmm_bi_rd:
- times 8 dw 64
-align 16
-vp8_bilinear_filters_sse2:
- dw 128, 128, 128, 128, 128, 128, 128, 128, 0, 0, 0, 0, 0, 0, 0, 0
- dw 112, 112, 112, 112, 112, 112, 112, 112, 16, 16, 16, 16, 16, 16, 16, 16
- dw 96, 96, 96, 96, 96, 96, 96, 96, 32, 32, 32, 32, 32, 32, 32, 32
- dw 80, 80, 80, 80, 80, 80, 80, 80, 48, 48, 48, 48, 48, 48, 48, 48
- dw 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
- dw 48, 48, 48, 48, 48, 48, 48, 48, 80, 80, 80, 80, 80, 80, 80, 80
- dw 32, 32, 32, 32, 32, 32, 32, 32, 96, 96, 96, 96, 96, 96, 96, 96
- dw 16, 16, 16, 16, 16, 16, 16, 16, 112, 112, 112, 112, 112, 112, 112, 112
diff --git a/vp8/encoder/x86/variance_impl_ssse3.asm b/vp8/encoder/x86/variance_impl_ssse3.asm
deleted file mode 100644
index 3c0fef9..0000000
--- a/vp8/encoder/x86/variance_impl_ssse3.asm
+++ /dev/null
@@ -1,364 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-%define xmm_filter_shift 7
-
-
-;void vp8_filter_block2d_bil_var_ssse3
-;(
-; unsigned char *ref_ptr,
-; int ref_pixels_per_line,
-; unsigned char *src_ptr,
-; int src_pixels_per_line,
-; unsigned int Height,
-; int xoffset,
-; int yoffset,
-; int *sum,
-; unsigned int *sumsquared;;
-;
-;)
-;Note: The filter coefficient at offset=0 is 128. Since the second register
-;for Pmaddubsw is signed bytes, we must calculate zero offset seperately.
-global sym(vp8_filter_block2d_bil_var_ssse3)
-sym(vp8_filter_block2d_bil_var_ssse3):
- push rbp
- mov rbp, rsp
- SHADOW_ARGS_TO_STACK 9
- SAVE_XMM
- GET_GOT rbx
- push rsi
- push rdi
- ; end prolog
-
- pxor xmm6, xmm6
- pxor xmm7, xmm7
-
- lea rcx, [GLOBAL(vp8_bilinear_filters_ssse3)]
- movsxd rax, dword ptr arg(5) ; xoffset
-
- cmp rax, 0 ; skip first_pass filter if xoffset=0
- je filter_block2d_bil_var_ssse3_sp_only
-
- shl rax, 4 ; point to filter coeff with xoffset
- lea rax, [rax + rcx] ; HFilter
-
- movsxd rdx, dword ptr arg(6) ; yoffset
-
- cmp rdx, 0 ; skip second_pass filter if yoffset=0
- je filter_block2d_bil_var_ssse3_fp_only
-
- shl rdx, 4
- lea rdx, [rdx + rcx] ; VFilter
-
- mov rsi, arg(0) ;ref_ptr
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
-
- movdqu xmm0, XMMWORD PTR [rsi]
- movdqu xmm1, XMMWORD PTR [rsi+1]
- movdqa xmm2, xmm0
-
- punpcklbw xmm0, xmm1
- punpckhbw xmm2, xmm1
- pmaddubsw xmm0, [rax]
- pmaddubsw xmm2, [rax]
-
- paddw xmm0, [GLOBAL(xmm_bi_rd)]
- paddw xmm2, [GLOBAL(xmm_bi_rd)]
- psraw xmm0, xmm_filter_shift
- psraw xmm2, xmm_filter_shift
-
- packuswb xmm0, xmm2
-
-%if ABI_IS_32BIT
- add rsi, dword ptr arg(1) ;ref_pixels_per_line
-%else
- movsxd r8, dword ptr arg(1) ;ref_pixels_per_line
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line
- lea rsi, [rsi + r8]
-%endif
-
-filter_block2d_bil_var_ssse3_loop:
- movdqu xmm1, XMMWORD PTR [rsi]
- movdqu xmm2, XMMWORD PTR [rsi+1]
- movdqa xmm3, xmm1
-
- punpcklbw xmm1, xmm2
- punpckhbw xmm3, xmm2
- pmaddubsw xmm1, [rax]
- pmaddubsw xmm3, [rax]
-
- paddw xmm1, [GLOBAL(xmm_bi_rd)]
- paddw xmm3, [GLOBAL(xmm_bi_rd)]
- psraw xmm1, xmm_filter_shift
- psraw xmm3, xmm_filter_shift
- packuswb xmm1, xmm3
-
- movdqa xmm2, xmm0
- movdqa xmm0, xmm1
- movdqa xmm3, xmm2
-
- punpcklbw xmm2, xmm1
- punpckhbw xmm3, xmm1
- pmaddubsw xmm2, [rdx]
- pmaddubsw xmm3, [rdx]
-
- paddw xmm2, [GLOBAL(xmm_bi_rd)]
- paddw xmm3, [GLOBAL(xmm_bi_rd)]
- psraw xmm2, xmm_filter_shift
- psraw xmm3, xmm_filter_shift
-
- movq xmm1, QWORD PTR [rdi]
- pxor xmm4, xmm4
- punpcklbw xmm1, xmm4
- movq xmm5, QWORD PTR [rdi+8]
- punpcklbw xmm5, xmm4
-
- psubw xmm2, xmm1
- psubw xmm3, xmm5
- paddw xmm6, xmm2
- paddw xmm6, xmm3
- pmaddwd xmm2, xmm2
- pmaddwd xmm3, xmm3
- paddd xmm7, xmm2
- paddd xmm7, xmm3
-
-%if ABI_IS_32BIT
- add rsi, dword ptr arg(1) ;ref_pixels_per_line
- add rdi, dword ptr arg(3) ;src_pixels_per_line
-%else
- lea rsi, [rsi + r8]
- lea rdi, [rdi + r9]
-%endif
-
- sub rcx, 1
- jnz filter_block2d_bil_var_ssse3_loop
-
- jmp filter_block2d_bil_variance
-
-filter_block2d_bil_var_ssse3_sp_only:
- movsxd rdx, dword ptr arg(6) ; yoffset
-
- cmp rdx, 0 ; Both xoffset =0 and yoffset=0
- je filter_block2d_bil_var_ssse3_full_pixel
-
- shl rdx, 4
- lea rdx, [rdx + rcx] ; VFilter
-
- mov rsi, arg(0) ;ref_ptr
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
-
- movdqu xmm1, XMMWORD PTR [rsi]
- movdqa xmm0, xmm1
-
-%if ABI_IS_32BIT=0
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line
-%endif
-
- lea rsi, [rsi + rax]
-
-filter_block2d_bil_sp_only_loop:
- movdqu xmm3, XMMWORD PTR [rsi]
- movdqa xmm2, xmm1
- movdqa xmm0, xmm3
-
- punpcklbw xmm1, xmm3
- punpckhbw xmm2, xmm3
- pmaddubsw xmm1, [rdx]
- pmaddubsw xmm2, [rdx]
-
- paddw xmm1, [GLOBAL(xmm_bi_rd)]
- paddw xmm2, [GLOBAL(xmm_bi_rd)]
- psraw xmm1, xmm_filter_shift
- psraw xmm2, xmm_filter_shift
-
- movq xmm3, QWORD PTR [rdi]
- pxor xmm4, xmm4
- punpcklbw xmm3, xmm4
- movq xmm5, QWORD PTR [rdi+8]
- punpcklbw xmm5, xmm4
-
- psubw xmm1, xmm3
- psubw xmm2, xmm5
- paddw xmm6, xmm1
- paddw xmm6, xmm2
- pmaddwd xmm1, xmm1
- pmaddwd xmm2, xmm2
- paddd xmm7, xmm1
- paddd xmm7, xmm2
-
- movdqa xmm1, xmm0
- lea rsi, [rsi + rax] ;ref_pixels_per_line
-
-%if ABI_IS_32BIT
- add rdi, dword ptr arg(3) ;src_pixels_per_line
-%else
- lea rdi, [rdi + r9]
-%endif
-
- sub rcx, 1
- jnz filter_block2d_bil_sp_only_loop
-
- jmp filter_block2d_bil_variance
-
-filter_block2d_bil_var_ssse3_full_pixel:
- mov rsi, arg(0) ;ref_ptr
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
- movsxd rax, dword ptr arg(1) ;ref_pixels_per_line
- movsxd rdx, dword ptr arg(3) ;src_pixels_per_line
- pxor xmm0, xmm0
-
-filter_block2d_bil_full_pixel_loop:
- movq xmm1, QWORD PTR [rsi]
- punpcklbw xmm1, xmm0
- movq xmm2, QWORD PTR [rsi+8]
- punpcklbw xmm2, xmm0
-
- movq xmm3, QWORD PTR [rdi]
- punpcklbw xmm3, xmm0
- movq xmm4, QWORD PTR [rdi+8]
- punpcklbw xmm4, xmm0
-
- psubw xmm1, xmm3
- psubw xmm2, xmm4
- paddw xmm6, xmm1
- paddw xmm6, xmm2
- pmaddwd xmm1, xmm1
- pmaddwd xmm2, xmm2
- paddd xmm7, xmm1
- paddd xmm7, xmm2
-
- lea rsi, [rsi + rax] ;ref_pixels_per_line
- lea rdi, [rdi + rdx] ;src_pixels_per_line
- sub rcx, 1
- jnz filter_block2d_bil_full_pixel_loop
-
- jmp filter_block2d_bil_variance
-
-filter_block2d_bil_var_ssse3_fp_only:
- mov rsi, arg(0) ;ref_ptr
- mov rdi, arg(2) ;src_ptr
- movsxd rcx, dword ptr arg(4) ;Height
- movsxd rdx, dword ptr arg(1) ;ref_pixels_per_line
-
- pxor xmm0, xmm0
-
-%if ABI_IS_32BIT=0
- movsxd r9, dword ptr arg(3) ;src_pixels_per_line
-%endif
-
-filter_block2d_bil_fp_only_loop:
- movdqu xmm1, XMMWORD PTR [rsi]
- movdqu xmm2, XMMWORD PTR [rsi+1]
- movdqa xmm3, xmm1
-
- punpcklbw xmm1, xmm2
- punpckhbw xmm3, xmm2
- pmaddubsw xmm1, [rax]
- pmaddubsw xmm3, [rax]
-
- paddw xmm1, [GLOBAL(xmm_bi_rd)]
- paddw xmm3, [GLOBAL(xmm_bi_rd)]
- psraw xmm1, xmm_filter_shift
- psraw xmm3, xmm_filter_shift
-
- movq xmm2, XMMWORD PTR [rdi]
- pxor xmm4, xmm4
- punpcklbw xmm2, xmm4
- movq xmm5, QWORD PTR [rdi+8]
- punpcklbw xmm5, xmm4
-
- psubw xmm1, xmm2
- psubw xmm3, xmm5
- paddw xmm6, xmm1
- paddw xmm6, xmm3
- pmaddwd xmm1, xmm1
- pmaddwd xmm3, xmm3
- paddd xmm7, xmm1
- paddd xmm7, xmm3
-
- lea rsi, [rsi + rdx]
-%if ABI_IS_32BIT
- add rdi, dword ptr arg(3) ;src_pixels_per_line
-%else
- lea rdi, [rdi + r9]
-%endif
-
- sub rcx, 1
- jnz filter_block2d_bil_fp_only_loop
-
- jmp filter_block2d_bil_variance
-
-filter_block2d_bil_variance:
- pxor xmm0, xmm0
- pxor xmm1, xmm1
- pxor xmm5, xmm5
-
- punpcklwd xmm0, xmm6
- punpckhwd xmm1, xmm6
- psrad xmm0, 16
- psrad xmm1, 16
- paddd xmm0, xmm1
- movdqa xmm1, xmm0
-
- movdqa xmm6, xmm7
- punpckldq xmm6, xmm5
- punpckhdq xmm7, xmm5
- paddd xmm6, xmm7
-
- punpckldq xmm0, xmm5
- punpckhdq xmm1, xmm5
- paddd xmm0, xmm1
-
- movdqa xmm7, xmm6
- movdqa xmm1, xmm0
-
- psrldq xmm7, 8
- psrldq xmm1, 8
-
- paddd xmm6, xmm7
- paddd xmm0, xmm1
-
- mov rsi, arg(7) ;[Sum]
- mov rdi, arg(8) ;[SSE]
-
- movd [rsi], xmm0
- movd [rdi], xmm6
-
- ; begin epilog
- pop rdi
- pop rsi
- RESTORE_GOT
- RESTORE_XMM
- UNSHADOW_ARGS
- pop rbp
- ret
-
-
-SECTION_RODATA
-align 16
-xmm_bi_rd:
- times 8 dw 64
-align 16
-vp8_bilinear_filters_ssse3:
- times 8 db 128, 0
- times 8 db 112, 16
- times 8 db 96, 32
- times 8 db 80, 48
- times 8 db 64, 64
- times 8 db 48, 80
- times 8 db 32, 96
- times 8 db 16, 112
diff --git a/vp8/encoder/x86/variance_mmx.c b/vp8/encoder/x86/variance_mmx.c
deleted file mode 100644
index cc8c1d0..0000000
--- a/vp8/encoder/x86/variance_mmx.c
+++ /dev/null
@@ -1,453 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/encoder/variance.h"
-#include "vp8/common/pragmas.h"
-#include "vpx_ports/mem.h"
-
-extern void filter_block1d_h6_mmx
-(
- const unsigned char *src_ptr,
- unsigned short *output_ptr,
- unsigned int src_pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- short *vp7_filter
-);
-extern void filter_block1d_v6_mmx
-(
- const short *src_ptr,
- unsigned char *output_ptr,
- unsigned int pixels_per_line,
- unsigned int pixel_step,
- unsigned int output_height,
- unsigned int output_width,
- short *vp7_filter
-);
-
-extern unsigned int vp8_get_mb_ss_mmx(short *src_ptr);
-extern unsigned int vp8_get8x8var_mmx
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *SSE,
- int *Sum
-);
-extern unsigned int vp8_get4x4var_mmx
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *SSE,
- int *Sum
-);
-extern void vp8_filter_block2d_bil4x4_var_mmx
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- const short *HFilter,
- const short *VFilter,
- int *sum,
- unsigned int *sumsquared
-);
-extern void vp8_filter_block2d_bil_var_mmx
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- const short *HFilter,
- const short *VFilter,
- int *sum,
- unsigned int *sumsquared
-);
-extern unsigned int vp8_get16x16pred_error_mmx
-(
- unsigned char *src_ptr,
- int src_stride,
- unsigned char *ref_ptr,
- int ref_stride
-);
-
-
-unsigned int vp8_get16x16var_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned *SSE,
- unsigned *SUM
-)
-{
- unsigned int sse0, sse1, sse2, sse3, var;
- int sum0, sum1, sum2, sum3, avg;
-
-
- vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
- vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse2, &sum2) ;
- vp8_get8x8var_mmx(src_ptr + 8 * source_stride + 8, source_stride, ref_ptr + 8 * recon_stride + 8, recon_stride, &sse3, &sum3);
-
- var = sse0 + sse1 + sse2 + sse3;
- avg = sum0 + sum1 + sum2 + sum3;
-
- *SSE = var;
- *SUM = avg;
- return (var - ((avg * avg) >> 8));
-
-}
-
-
-
-
-
-unsigned int vp8_variance4x4_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int var;
- int avg;
-
- vp8_get4x4var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
- *sse = var;
- return (var - ((avg * avg) >> 4));
-
-}
-
-unsigned int vp8_variance8x8_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int var;
- int avg;
-
- vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
- *sse = var;
-
- return (var - ((avg * avg) >> 6));
-
-}
-
-unsigned int vp8_mse16x16_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int sse0, sse1, sse2, sse3, var;
- int sum0, sum1, sum2, sum3;
-
-
- vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
- vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse2, &sum2) ;
- vp8_get8x8var_mmx(src_ptr + 8 * source_stride + 8, source_stride, ref_ptr + 8 * recon_stride + 8, recon_stride, &sse3, &sum3);
-
- var = sse0 + sse1 + sse2 + sse3;
- *sse = var;
- return var;
-}
-
-
-unsigned int vp8_variance16x16_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- int *sse)
-{
- unsigned int sse0, sse1, sse2, sse3, var;
- int sum0, sum1, sum2, sum3, avg;
-
-
- vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
- vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse2, &sum2) ;
- vp8_get8x8var_mmx(src_ptr + 8 * source_stride + 8, source_stride, ref_ptr + 8 * recon_stride + 8, recon_stride, &sse3, &sum3);
-
- var = sse0 + sse1 + sse2 + sse3;
- avg = sum0 + sum1 + sum2 + sum3;
- *sse = var;
- return (var - ((avg * avg) >> 8));
-}
-
-unsigned int vp8_variance16x8_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int sse0, sse1, var;
- int sum0, sum1, avg;
-
- vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
-
- var = sse0 + sse1;
- avg = sum0 + sum1;
- *sse = var;
- return (var - ((avg * avg) >> 7));
-
-}
-
-
-unsigned int vp8_variance8x16_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int sse0, sse1, var;
- int sum0, sum1, avg;
-
- vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse1, &sum1) ;
-
- var = sse0 + sse1;
- avg = sum0 + sum1;
- *sse = var;
-
- return (var - ((avg * avg) >> 7));
-
-}
-
-
-
-
-///////////////////////////////////////////////////////////////////////////
-// the mmx function that does the bilinear filtering and var calculation //
-// int one pass //
-///////////////////////////////////////////////////////////////////////////
-DECLARE_ALIGNED(16, const short, vp8_vp7_bilinear_filters_mmx[8][8]) =
-{
- { 128, 128, 128, 128, 0, 0, 0, 0 },
- { 112, 112, 112, 112, 16, 16, 16, 16 },
- { 96, 96, 96, 96, 32, 32, 32, 32 },
- { 80, 80, 80, 80, 48, 48, 48, 48 },
- { 64, 64, 64, 64, 64, 64, 64, 64 },
- { 48, 48, 48, 48, 80, 80, 80, 80 },
- { 32, 32, 32, 32, 96, 96, 96, 96 },
- { 16, 16, 16, 16, 112, 112, 112, 112 }
-};
-
-unsigned int vp8_sub_pixel_variance4x4_mmx
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse)
-
-{
- int xsum;
- unsigned int xxsum;
- vp8_filter_block2d_bil4x4_var_mmx(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line,
- vp8_vp7_bilinear_filters_mmx[xoffset], vp8_vp7_bilinear_filters_mmx[yoffset],
- &xsum, &xxsum
- );
- *sse = xxsum;
- return (xxsum - ((xsum * xsum) >> 4));
-}
-
-
-unsigned int vp8_sub_pixel_variance8x8_mmx
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
-
- int xsum;
- unsigned int xxsum;
- vp8_filter_block2d_bil_var_mmx(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- vp8_vp7_bilinear_filters_mmx[xoffset], vp8_vp7_bilinear_filters_mmx[yoffset],
- &xsum, &xxsum
- );
- *sse = xxsum;
- return (xxsum - ((xsum * xsum) >> 6));
-}
-
-unsigned int vp8_sub_pixel_variance16x16_mmx
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
-
- int xsum0, xsum1;
- unsigned int xxsum0, xxsum1;
-
-
- vp8_filter_block2d_bil_var_mmx(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- vp8_vp7_bilinear_filters_mmx[xoffset], vp8_vp7_bilinear_filters_mmx[yoffset],
- &xsum0, &xxsum0
- );
-
-
- vp8_filter_block2d_bil_var_mmx(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 16,
- vp8_vp7_bilinear_filters_mmx[xoffset], vp8_vp7_bilinear_filters_mmx[yoffset],
- &xsum1, &xxsum1
- );
-
- xsum0 += xsum1;
- xxsum0 += xxsum1;
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 8));
-
-
-}
-
-unsigned int vp8_sub_pixel_mse16x16_mmx(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- vp8_sub_pixel_variance16x16_mmx(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
- return *sse;
-}
-
-unsigned int vp8_sub_pixel_variance16x8_mmx
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- int xsum0, xsum1;
- unsigned int xxsum0, xxsum1;
-
-
- vp8_filter_block2d_bil_var_mmx(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- vp8_vp7_bilinear_filters_mmx[xoffset], vp8_vp7_bilinear_filters_mmx[yoffset],
- &xsum0, &xxsum0
- );
-
-
- vp8_filter_block2d_bil_var_mmx(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 8,
- vp8_vp7_bilinear_filters_mmx[xoffset], vp8_vp7_bilinear_filters_mmx[yoffset],
- &xsum1, &xxsum1
- );
-
- xsum0 += xsum1;
- xxsum0 += xxsum1;
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 7));
-}
-
-unsigned int vp8_sub_pixel_variance8x16_mmx
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- int *sse
-)
-{
- int xsum;
- unsigned int xxsum;
- vp8_filter_block2d_bil_var_mmx(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- vp8_vp7_bilinear_filters_mmx[xoffset], vp8_vp7_bilinear_filters_mmx[yoffset],
- &xsum, &xxsum
- );
- *sse = xxsum;
- return (xxsum - ((xsum * xsum) >> 7));
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_h_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- return vp8_sub_pixel_variance16x16_mmx(src_ptr, source_stride, 4, 0,
- ref_ptr, recon_stride, sse);
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_v_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- return vp8_sub_pixel_variance16x16_mmx(src_ptr, source_stride, 0, 4,
- ref_ptr, recon_stride, sse);
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_hv_mmx(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- return vp8_sub_pixel_variance16x16_mmx(src_ptr, source_stride, 4, 4,
- ref_ptr, recon_stride, sse);
-}
diff --git a/vp8/encoder/x86/variance_sse2.c b/vp8/encoder/x86/variance_sse2.c
deleted file mode 100644
index 0edda30..0000000
--- a/vp8/encoder/x86/variance_sse2.c
+++ /dev/null
@@ -1,564 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/encoder/variance.h"
-#include "vp8/common/pragmas.h"
-#include "vpx_ports/mem.h"
-
-extern void filter_block1d_h6_mmx(const unsigned char *src_ptr, unsigned short *output_ptr, unsigned int src_pixels_per_line, unsigned int pixel_step, unsigned int output_height, unsigned int output_width, short *vp7_filter);
-extern void filter_block1d_v6_mmx(const short *src_ptr, unsigned char *output_ptr, unsigned int pixels_per_line, unsigned int pixel_step, unsigned int output_height, unsigned int output_width, short *vp7_filter);
-extern void filter_block1d8_h6_sse2(const unsigned char *src_ptr, unsigned short *output_ptr, unsigned int src_pixels_per_line, unsigned int pixel_step, unsigned int output_height, unsigned int output_width, short *vp7_filter);
-extern void filter_block1d8_v6_sse2(const short *src_ptr, unsigned char *output_ptr, unsigned int pixels_per_line, unsigned int pixel_step, unsigned int output_height, unsigned int output_width, short *vp7_filter);
-
-extern void vp8_filter_block2d_bil4x4_var_mmx
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- const short *HFilter,
- const short *VFilter,
- int *sum,
- unsigned int *sumsquared
-);
-
-extern unsigned int vp8_get4x4var_mmx
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *SSE,
- int *Sum
-);
-
-unsigned int vp8_get_mb_ss_sse2
-(
- const short *src_ptr
-);
-unsigned int vp8_get16x16var_sse2
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *SSE,
- int *Sum
-);
-unsigned int vp8_get16x16pred_error_sse2
-(
- const unsigned char *src_ptr,
- int src_stride,
- const unsigned char *ref_ptr,
- int ref_stride
-);
-unsigned int vp8_get8x8var_sse2
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *SSE,
- int *Sum
-);
-void vp8_filter_block2d_bil_var_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int xoffset,
- int yoffset,
- int *sum,
- unsigned int *sumsquared
-);
-void vp8_half_horiz_vert_variance8x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-void vp8_half_horiz_vert_variance16x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-void vp8_half_horiz_variance8x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-void vp8_half_horiz_variance16x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-void vp8_half_vert_variance8x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-void vp8_half_vert_variance16x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-
-DECLARE_ALIGNED(16, extern short, vp8_vp7_bilinear_filters_mmx[8][8]);
-
-unsigned int vp8_variance4x4_wmt(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride)
-{
- unsigned int var;
- int avg;
-
- vp8_get4x4var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
- return (var - ((avg * avg) >> 4));
-
-}
-
-
-
-unsigned int vp8_variance8x8_wmt
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride)
-{
- unsigned int var;
- int avg;
-
- vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
-
- return (var - ((avg * avg) >> 6));
-
-}
-
-
-unsigned int vp8_variance16x16_wmt
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int sse0;
- int sum0;
-
-
- vp8_get16x16var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- *sse = sse0;
- return (sse0 - ((sum0 * sum0) >> 8));
-}
-unsigned int vp8_mse16x16_wmt(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
-
- unsigned int sse0;
- int sum0;
- vp8_get16x16var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- *sse = sse0;
- return sse0;
-
-}
-
-
-unsigned int vp8_variance16x8_wmt
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int sse0, sse1, var;
- int sum0, sum1, avg;
-
- vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_sse2(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
-
- var = sse0 + sse1;
- avg = sum0 + sum1;
- *sse = var;
- return (var - ((avg * avg) >> 7));
-
-}
-
-unsigned int vp8_variance8x16_wmt
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int sse0, sse1, var;
- int sum0, sum1, avg;
-
- vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_sse2(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse1, &sum1) ;
-
- var = sse0 + sse1;
- avg = sum0 + sum1;
- *sse = var;
- return (var - ((avg * avg) >> 7));
-
-}
-
-unsigned int vp8_sub_pixel_variance4x4_wmt
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- int xsum;
- unsigned int xxsum;
- vp8_filter_block2d_bil4x4_var_mmx(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line,
- vp8_vp7_bilinear_filters_mmx[xoffset], vp8_vp7_bilinear_filters_mmx[yoffset],
- &xsum, &xxsum
- );
- *sse = xxsum;
- return (xxsum - ((xsum * xsum) >> 4));
-}
-
-
-unsigned int vp8_sub_pixel_variance8x8_wmt
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- int xsum;
- unsigned int xxsum;
-
- if (xoffset == 4 && yoffset == 0)
- {
- vp8_half_horiz_variance8x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum, &xxsum);
- }
- else if (xoffset == 0 && yoffset == 4)
- {
- vp8_half_vert_variance8x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum, &xxsum);
- }
- else if (xoffset == 4 && yoffset == 4)
- {
- vp8_half_horiz_vert_variance8x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum, &xxsum);
- }
- else
- {
- vp8_filter_block2d_bil_var_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- xoffset, yoffset,
- &xsum, &xxsum);
- }
-
- *sse = xxsum;
- return (xxsum - ((xsum * xsum) >> 6));
-}
-
-unsigned int vp8_sub_pixel_variance16x16_wmt
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- int xsum0, xsum1;
- unsigned int xxsum0, xxsum1;
-
-
- // note we could avoid these if statements if the calling function
- // just called the appropriate functions inside.
- if (xoffset == 4 && yoffset == 0)
- {
- vp8_half_horiz_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
- }
- else if (xoffset == 0 && yoffset == 4)
- {
- vp8_half_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
- }
- else if (xoffset == 4 && yoffset == 4)
- {
- vp8_half_horiz_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
- }
- else
- {
- vp8_filter_block2d_bil_var_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- xoffset, yoffset,
- &xsum0, &xxsum0
- );
-
- vp8_filter_block2d_bil_var_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 16,
- xoffset, yoffset,
- &xsum1, &xxsum1
- );
- xsum0 += xsum1;
- xxsum0 += xxsum1;
- }
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 8));
-}
-
-unsigned int vp8_sub_pixel_mse16x16_wmt(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- vp8_sub_pixel_variance16x16_wmt(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
- return *sse;
-}
-
-unsigned int vp8_sub_pixel_variance16x8_wmt
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-
-)
-{
- int xsum0, xsum1;
- unsigned int xxsum0, xxsum1;
-
- if (xoffset == 4 && yoffset == 0)
- {
- vp8_half_horiz_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum0, &xxsum0);
- }
- else if (xoffset == 0 && yoffset == 4)
- {
- vp8_half_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum0, &xxsum0);
- }
- else if (xoffset == 4 && yoffset == 4)
- {
- vp8_half_horiz_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum0, &xxsum0);
- }
- else
- {
- vp8_filter_block2d_bil_var_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- xoffset, yoffset,
- &xsum0, &xxsum0);
-
- vp8_filter_block2d_bil_var_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 8,
- xoffset, yoffset,
- &xsum1, &xxsum1);
- xsum0 += xsum1;
- xxsum0 += xxsum1;
- }
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 7));
-}
-
-unsigned int vp8_sub_pixel_variance8x16_wmt
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- int xsum;
- unsigned int xxsum;
-
- if (xoffset == 4 && yoffset == 0)
- {
- vp8_half_horiz_variance8x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum, &xxsum);
- }
- else if (xoffset == 0 && yoffset == 4)
- {
- vp8_half_vert_variance8x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum, &xxsum);
- }
- else if (xoffset == 4 && yoffset == 4)
- {
- vp8_half_horiz_vert_variance8x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum, &xxsum);
- }
- else
- {
- vp8_filter_block2d_bil_var_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- xoffset, yoffset,
- &xsum, &xxsum);
- }
-
- *sse = xxsum;
- return (xxsum - ((xsum * xsum) >> 7));
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_h_wmt(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse)
-{
- int xsum0;
- unsigned int xxsum0;
-
- vp8_half_horiz_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 8));
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_v_wmt(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse)
-{
- int xsum0;
- unsigned int xxsum0;
- vp8_half_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 8));
-}
-
-
-unsigned int vp8_variance_halfpixvar16x16_hv_wmt(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse)
-{
- int xsum0;
- unsigned int xxsum0;
-
- vp8_half_horiz_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 8));
-}
diff --git a/vp8/encoder/x86/variance_ssse3.c b/vp8/encoder/x86/variance_ssse3.c
deleted file mode 100644
index eb5d486..0000000
--- a/vp8/encoder/x86/variance_ssse3.c
+++ /dev/null
@@ -1,165 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8/encoder/variance.h"
-#include "vp8/common/pragmas.h"
-#include "vpx_ports/mem.h"
-
-extern unsigned int vp8_get16x16var_sse2
-(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *SSE,
- int *Sum
-);
-extern void vp8_half_horiz_vert_variance16x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-extern void vp8_half_horiz_variance16x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-extern void vp8_half_vert_variance16x_h_sse2
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int *sum,
- unsigned int *sumsquared
-);
-extern void vp8_filter_block2d_bil_var_ssse3
-(
- const unsigned char *ref_ptr,
- int ref_pixels_per_line,
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- unsigned int Height,
- int xoffset,
- int yoffset,
- int *sum,
- unsigned int *sumsquared
-);
-
-unsigned int vp8_sub_pixel_variance16x16_ssse3
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- int xsum0;
- unsigned int xxsum0;
-
- // note we could avoid these if statements if the calling function
- // just called the appropriate functions inside.
- if (xoffset == 4 && yoffset == 0)
- {
- vp8_half_horiz_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
- }
- else if (xoffset == 0 && yoffset == 4)
- {
- vp8_half_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
- }
- else if (xoffset == 4 && yoffset == 4)
- {
- vp8_half_horiz_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- &xsum0, &xxsum0);
- }
- else
- {
- vp8_filter_block2d_bil_var_ssse3(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 16,
- xoffset, yoffset,
- &xsum0, &xxsum0);
- }
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 8));
-}
-
-unsigned int vp8_sub_pixel_variance16x8_ssse3
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-
-)
-{
- int xsum0;
- unsigned int xxsum0;
-
- if (xoffset == 4 && yoffset == 0)
- {
- vp8_half_horiz_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum0, &xxsum0);
- }
- else if (xoffset == 0 && yoffset == 4)
- {
- vp8_half_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum0, &xxsum0);
- }
- else if (xoffset == 4 && yoffset == 4)
- {
- vp8_half_horiz_vert_variance16x_h_sse2(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- &xsum0, &xxsum0);
- }
- else
- {
- vp8_filter_block2d_bil_var_ssse3(
- src_ptr, src_pixels_per_line,
- dst_ptr, dst_pixels_per_line, 8,
- xoffset, yoffset,
- &xsum0, &xxsum0);
- }
-
- *sse = xxsum0;
- return (xxsum0 - ((xsum0 * xsum0) >> 7));
-}
diff --git a/vp8/encoder/x86/variance_x86.h b/vp8/encoder/x86/variance_x86.h
deleted file mode 100644
index 3560f74..0000000
--- a/vp8/encoder/x86/variance_x86.h
+++ /dev/null
@@ -1,335 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VARIANCE_X86_H
-#define VARIANCE_X86_H
-
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-#if HAVE_MMX
-extern prototype_sad(vp8_sad4x4_mmx);
-extern prototype_sad(vp8_sad8x8_mmx);
-extern prototype_sad(vp8_sad8x16_mmx);
-extern prototype_sad(vp8_sad16x8_mmx);
-extern prototype_sad(vp8_sad16x16_mmx);
-extern prototype_variance(vp8_variance4x4_mmx);
-extern prototype_variance(vp8_variance8x8_mmx);
-extern prototype_variance(vp8_variance8x16_mmx);
-extern prototype_variance(vp8_variance16x8_mmx);
-extern prototype_variance(vp8_variance16x16_mmx);
-extern prototype_subpixvariance(vp8_sub_pixel_variance4x4_mmx);
-extern prototype_subpixvariance(vp8_sub_pixel_variance8x8_mmx);
-extern prototype_subpixvariance(vp8_sub_pixel_variance8x16_mmx);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x8_mmx);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_mmx);
-extern prototype_variance(vp8_variance_halfpixvar16x16_h_mmx);
-extern prototype_variance(vp8_variance_halfpixvar16x16_v_mmx);
-extern prototype_variance(vp8_variance_halfpixvar16x16_hv_mmx);
-extern prototype_subpixvariance(vp8_sub_pixel_mse16x16_mmx);
-extern prototype_getmbss(vp8_get_mb_ss_mmx);
-extern prototype_variance(vp8_mse16x16_mmx);
-extern prototype_sad(vp8_get16x16pred_error_mmx);
-extern prototype_variance2(vp8_get8x8var_mmx);
-extern prototype_variance2(vp8_get16x16var_mmx);
-extern prototype_sad(vp8_get4x4sse_cs_mmx);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_variance_sad4x4
-#define vp8_variance_sad4x4 vp8_sad4x4_mmx
-
-#undef vp8_variance_sad8x8
-#define vp8_variance_sad8x8 vp8_sad8x8_mmx
-
-#undef vp8_variance_sad8x16
-#define vp8_variance_sad8x16 vp8_sad8x16_mmx
-
-#undef vp8_variance_sad16x8
-#define vp8_variance_sad16x8 vp8_sad16x8_mmx
-
-#undef vp8_variance_sad16x16
-#define vp8_variance_sad16x16 vp8_sad16x16_mmx
-
-#undef vp8_variance_var4x4
-#define vp8_variance_var4x4 vp8_variance4x4_mmx
-
-#undef vp8_variance_var8x8
-#define vp8_variance_var8x8 vp8_variance8x8_mmx
-
-#undef vp8_variance_var8x16
-#define vp8_variance_var8x16 vp8_variance8x16_mmx
-
-#undef vp8_variance_var16x8
-#define vp8_variance_var16x8 vp8_variance16x8_mmx
-
-#undef vp8_variance_var16x16
-#define vp8_variance_var16x16 vp8_variance16x16_mmx
-
-#undef vp8_variance_subpixvar4x4
-#define vp8_variance_subpixvar4x4 vp8_sub_pixel_variance4x4_mmx
-
-#undef vp8_variance_subpixvar8x8
-#define vp8_variance_subpixvar8x8 vp8_sub_pixel_variance8x8_mmx
-
-#undef vp8_variance_subpixvar8x16
-#define vp8_variance_subpixvar8x16 vp8_sub_pixel_variance8x16_mmx
-
-#undef vp8_variance_subpixvar16x8
-#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_mmx
-
-#undef vp8_variance_subpixvar16x16
-#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_mmx
-
-#undef vp8_variance_halfpixvar16x16_h
-#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_mmx
-
-#undef vp8_variance_halfpixvar16x16_v
-#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_mmx
-
-#undef vp8_variance_halfpixvar16x16_hv
-#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_mmx
-
-#undef vp8_variance_subpixmse16x16
-#define vp8_variance_subpixmse16x16 vp8_sub_pixel_mse16x16_mmx
-
-#undef vp8_variance_getmbss
-#define vp8_variance_getmbss vp8_get_mb_ss_mmx
-
-#undef vp8_variance_mse16x16
-#define vp8_variance_mse16x16 vp8_mse16x16_mmx
-
-#undef vp8_variance_get16x16prederror
-#define vp8_variance_get16x16prederror vp8_get16x16pred_error_mmx
-
-#undef vp8_variance_get8x8var
-#define vp8_variance_get8x8var vp8_get8x8var_mmx
-
-#undef vp8_variance_get16x16var
-#define vp8_variance_get16x16var vp8_get16x16var_mmx
-
-#undef vp8_variance_get4x4sse_cs
-#define vp8_variance_get4x4sse_cs vp8_get4x4sse_cs_mmx
-
-#endif
-#endif
-
-
-#if HAVE_SSE2
-extern prototype_sad(vp8_sad4x4_wmt);
-extern prototype_sad(vp8_sad8x8_wmt);
-extern prototype_sad(vp8_sad8x16_wmt);
-extern prototype_sad(vp8_sad16x8_wmt);
-extern prototype_sad(vp8_sad16x16_wmt);
-extern prototype_variance(vp8_variance4x4_wmt);
-extern prototype_variance(vp8_variance8x8_wmt);
-extern prototype_variance(vp8_variance8x16_wmt);
-extern prototype_variance(vp8_variance16x8_wmt);
-extern prototype_variance(vp8_variance16x16_wmt);
-extern prototype_subpixvariance(vp8_sub_pixel_variance4x4_wmt);
-extern prototype_subpixvariance(vp8_sub_pixel_variance8x8_wmt);
-extern prototype_subpixvariance(vp8_sub_pixel_variance8x16_wmt);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x8_wmt);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_wmt);
-extern prototype_variance(vp8_variance_halfpixvar16x16_h_wmt);
-extern prototype_variance(vp8_variance_halfpixvar16x16_v_wmt);
-extern prototype_variance(vp8_variance_halfpixvar16x16_hv_wmt);
-extern prototype_subpixvariance(vp8_sub_pixel_mse16x16_wmt);
-extern prototype_getmbss(vp8_get_mb_ss_sse2);
-extern prototype_variance(vp8_mse16x16_wmt);
-extern prototype_sad(vp8_get16x16pred_error_sse2);
-extern prototype_variance2(vp8_get8x8var_sse2);
-extern prototype_variance2(vp8_get16x16var_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_variance_sad4x4
-#define vp8_variance_sad4x4 vp8_sad4x4_wmt
-
-#undef vp8_variance_sad8x8
-#define vp8_variance_sad8x8 vp8_sad8x8_wmt
-
-#undef vp8_variance_sad8x16
-#define vp8_variance_sad8x16 vp8_sad8x16_wmt
-
-#undef vp8_variance_sad16x8
-#define vp8_variance_sad16x8 vp8_sad16x8_wmt
-
-#undef vp8_variance_sad16x16
-#define vp8_variance_sad16x16 vp8_sad16x16_wmt
-
-#undef vp8_variance_var4x4
-#define vp8_variance_var4x4 vp8_variance4x4_wmt
-
-#undef vp8_variance_var8x8
-#define vp8_variance_var8x8 vp8_variance8x8_wmt
-
-#undef vp8_variance_var8x16
-#define vp8_variance_var8x16 vp8_variance8x16_wmt
-
-#undef vp8_variance_var16x8
-#define vp8_variance_var16x8 vp8_variance16x8_wmt
-
-#undef vp8_variance_var16x16
-#define vp8_variance_var16x16 vp8_variance16x16_wmt
-
-#undef vp8_variance_subpixvar4x4
-#define vp8_variance_subpixvar4x4 vp8_sub_pixel_variance4x4_wmt
-
-#undef vp8_variance_subpixvar8x8
-#define vp8_variance_subpixvar8x8 vp8_sub_pixel_variance8x8_wmt
-
-#undef vp8_variance_subpixvar8x16
-#define vp8_variance_subpixvar8x16 vp8_sub_pixel_variance8x16_wmt
-
-#undef vp8_variance_subpixvar16x8
-#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_wmt
-
-#undef vp8_variance_subpixvar16x16
-#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_wmt
-
-#undef vp8_variance_halfpixvar16x16_h
-#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_wmt
-
-#undef vp8_variance_halfpixvar16x16_v
-#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_wmt
-
-#undef vp8_variance_halfpixvar16x16_hv
-#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_wmt
-
-#undef vp8_variance_subpixmse16x16
-#define vp8_variance_subpixmse16x16 vp8_sub_pixel_mse16x16_wmt
-
-#undef vp8_variance_getmbss
-#define vp8_variance_getmbss vp8_get_mb_ss_sse2
-
-#undef vp8_variance_mse16x16
-#define vp8_variance_mse16x16 vp8_mse16x16_wmt
-
-#undef vp8_variance_get16x16prederror
-#define vp8_variance_get16x16prederror vp8_get16x16pred_error_sse2
-
-#undef vp8_variance_get8x8var
-#define vp8_variance_get8x8var vp8_get8x8var_sse2
-
-#undef vp8_variance_get16x16var
-#define vp8_variance_get16x16var vp8_get16x16var_sse2
-
-#endif
-#endif
-
-
-#if HAVE_SSE3
-extern prototype_sad(vp8_sad16x16_sse3);
-extern prototype_sad(vp8_sad16x8_sse3);
-extern prototype_sad_multi_same_address(vp8_sad16x16x3_sse3);
-extern prototype_sad_multi_same_address(vp8_sad16x8x3_sse3);
-extern prototype_sad_multi_same_address(vp8_sad8x16x3_sse3);
-extern prototype_sad_multi_same_address(vp8_sad8x8x3_sse3);
-extern prototype_sad_multi_same_address(vp8_sad4x4x3_sse3);
-
-extern prototype_sad_multi_dif_address(vp8_sad16x16x4d_sse3);
-extern prototype_sad_multi_dif_address(vp8_sad16x8x4d_sse3);
-extern prototype_sad_multi_dif_address(vp8_sad8x16x4d_sse3);
-extern prototype_sad_multi_dif_address(vp8_sad8x8x4d_sse3);
-extern prototype_sad_multi_dif_address(vp8_sad4x4x4d_sse3);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_variance_sad16x16
-#define vp8_variance_sad16x16 vp8_sad16x16_sse3
-
-#undef vp8_variance_sad16x16x3
-#define vp8_variance_sad16x16x3 vp8_sad16x16x3_sse3
-
-#undef vp8_variance_sad16x8x3
-#define vp8_variance_sad16x8x3 vp8_sad16x8x3_sse3
-
-#undef vp8_variance_sad8x16x3
-#define vp8_variance_sad8x16x3 vp8_sad8x16x3_sse3
-
-#undef vp8_variance_sad8x8x3
-#define vp8_variance_sad8x8x3 vp8_sad8x8x3_sse3
-
-#undef vp8_variance_sad4x4x3
-#define vp8_variance_sad4x4x3 vp8_sad4x4x3_sse3
-
-#undef vp8_variance_sad16x16x4d
-#define vp8_variance_sad16x16x4d vp8_sad16x16x4d_sse3
-
-#undef vp8_variance_sad16x8x4d
-#define vp8_variance_sad16x8x4d vp8_sad16x8x4d_sse3
-
-#undef vp8_variance_sad8x16x4d
-#define vp8_variance_sad8x16x4d vp8_sad8x16x4d_sse3
-
-#undef vp8_variance_sad8x8x4d
-#define vp8_variance_sad8x8x4d vp8_sad8x8x4d_sse3
-
-#undef vp8_variance_sad4x4x4d
-#define vp8_variance_sad4x4x4d vp8_sad4x4x4d_sse3
-
-#endif
-#endif
-
-
-#if HAVE_SSSE3
-extern prototype_sad_multi_same_address(vp8_sad16x16x3_ssse3);
-extern prototype_sad_multi_same_address(vp8_sad16x8x3_ssse3);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x8_ssse3);
-extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_ssse3);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_variance_sad16x16x3
-#define vp8_variance_sad16x16x3 vp8_sad16x16x3_ssse3
-
-#undef vp8_variance_sad16x8x3
-#define vp8_variance_sad16x8x3 vp8_sad16x8x3_ssse3
-
-#undef vp8_variance_subpixvar16x8
-#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_ssse3
-
-#undef vp8_variance_subpixvar16x16
-#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_ssse3
-
-#endif
-#endif
-
-
-#if HAVE_SSE4_1
-extern prototype_sad_multi_same_address_1(vp8_sad16x16x8_sse4);
-extern prototype_sad_multi_same_address_1(vp8_sad16x8x8_sse4);
-extern prototype_sad_multi_same_address_1(vp8_sad8x16x8_sse4);
-extern prototype_sad_multi_same_address_1(vp8_sad8x8x8_sse4);
-extern prototype_sad_multi_same_address_1(vp8_sad4x4x8_sse4);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef vp8_variance_sad16x16x8
-#define vp8_variance_sad16x16x8 vp8_sad16x16x8_sse4
-
-#undef vp8_variance_sad16x8x8
-#define vp8_variance_sad16x8x8 vp8_sad16x8x8_sse4
-
-#undef vp8_variance_sad8x16x8
-#define vp8_variance_sad8x16x8 vp8_sad8x16x8_sse4
-
-#undef vp8_variance_sad8x8x8
-#define vp8_variance_sad8x8x8 vp8_sad8x8x8_sse4
-
-#undef vp8_variance_sad4x4x8
-#define vp8_variance_sad4x4x8 vp8_sad4x4x8_sse4
-
-#endif
-#endif
-
-#endif
diff --git a/vp8/encoder/x86/x86_csystemdependent.c b/vp8/encoder/x86/x86_csystemdependent.c
deleted file mode 100644
index 2b6bd98..0000000
--- a/vp8/encoder/x86/x86_csystemdependent.c
+++ /dev/null
@@ -1,364 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_ports/x86.h"
-#include "vp8/encoder/variance.h"
-#include "vp8/encoder/onyx_int.h"
-
-
-#if HAVE_MMX
-static void short_fdct8x4_mmx(short *input, short *output, int pitch)
-{
- vp8_short_fdct4x4_mmx(input, output, pitch);
- vp8_short_fdct4x4_mmx(input + 4, output + 16, pitch);
-}
-
-int vp8_fast_quantize_b_impl_mmx(short *coeff_ptr, short *zbin_ptr,
- short *qcoeff_ptr, short *dequant_ptr,
- short *scan_mask, short *round_ptr,
- short *quant_ptr, short *dqcoeff_ptr);
-static void fast_quantize_b_mmx(BLOCK *b, BLOCKD *d)
-{
- short *scan_mask = vp8_default_zig_zag_mask;//d->scan_order_mask_ptr;
- short *coeff_ptr = b->coeff;
- short *zbin_ptr = b->zbin;
- short *round_ptr = b->round;
- short *quant_ptr = b->quant_fast;
- short *qcoeff_ptr = d->qcoeff;
- short *dqcoeff_ptr = d->dqcoeff;
- short *dequant_ptr = d->dequant;
-
- d->eob = vp8_fast_quantize_b_impl_mmx(
- coeff_ptr,
- zbin_ptr,
- qcoeff_ptr,
- dequant_ptr,
- scan_mask,
-
- round_ptr,
- quant_ptr,
- dqcoeff_ptr
- );
-}
-
-int vp8_mbblock_error_mmx_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
-static int mbblock_error_mmx(MACROBLOCK *mb, int dc)
-{
- short *coeff_ptr = mb->block[0].coeff;
- short *dcoef_ptr = mb->e_mbd.block[0].dqcoeff;
- return vp8_mbblock_error_mmx_impl(coeff_ptr, dcoef_ptr, dc);
-}
-
-int vp8_mbuverror_mmx_impl(short *s_ptr, short *d_ptr);
-static int mbuverror_mmx(MACROBLOCK *mb)
-{
- short *s_ptr = &mb->coeff[256];
- short *d_ptr = &mb->e_mbd.dqcoeff[256];
- return vp8_mbuverror_mmx_impl(s_ptr, d_ptr);
-}
-
-void vp8_subtract_b_mmx_impl(unsigned char *z, int src_stride,
- short *diff, unsigned char *predictor,
- int pitch);
-static void subtract_b_mmx(BLOCK *be, BLOCKD *bd, int pitch)
-{
- unsigned char *z = *(be->base_src) + be->src;
- unsigned int src_stride = be->src_stride;
- short *diff = &be->src_diff[0];
- unsigned char *predictor = &bd->predictor[0];
- vp8_subtract_b_mmx_impl(z, src_stride, diff, predictor, pitch);
-}
-
-#endif
-
-#if HAVE_SSE2
-int vp8_fast_quantize_b_impl_sse2(short *coeff_ptr,
- short *qcoeff_ptr, short *dequant_ptr,
- const short *inv_scan_order, short *round_ptr,
- short *quant_ptr, short *dqcoeff_ptr);
-static void fast_quantize_b_sse2(BLOCK *b, BLOCKD *d)
-{
- short *scan_mask = vp8_default_zig_zag_mask;//d->scan_order_mask_ptr;
- short *coeff_ptr = b->coeff;
- short *round_ptr = b->round;
- short *quant_ptr = b->quant_fast;
- short *qcoeff_ptr = d->qcoeff;
- short *dqcoeff_ptr = d->dqcoeff;
- short *dequant_ptr = d->dequant;
-
- d->eob = vp8_fast_quantize_b_impl_sse2(
- coeff_ptr,
- qcoeff_ptr,
- dequant_ptr,
- vp8_default_inv_zig_zag,
- round_ptr,
- quant_ptr,
- dqcoeff_ptr
- );
-}
-
-int vp8_mbblock_error_xmm_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
-static int mbblock_error_xmm(MACROBLOCK *mb, int dc)
-{
- short *coeff_ptr = mb->block[0].coeff;
- short *dcoef_ptr = mb->e_mbd.block[0].dqcoeff;
- return vp8_mbblock_error_xmm_impl(coeff_ptr, dcoef_ptr, dc);
-}
-
-int vp8_mbuverror_xmm_impl(short *s_ptr, short *d_ptr);
-static int mbuverror_xmm(MACROBLOCK *mb)
-{
- short *s_ptr = &mb->coeff[256];
- short *d_ptr = &mb->e_mbd.dqcoeff[256];
- return vp8_mbuverror_xmm_impl(s_ptr, d_ptr);
-}
-
-void vp8_subtract_b_sse2_impl(unsigned char *z, int src_stride,
- short *diff, unsigned char *predictor,
- int pitch);
-static void subtract_b_sse2(BLOCK *be, BLOCKD *bd, int pitch)
-{
- unsigned char *z = *(be->base_src) + be->src;
- unsigned int src_stride = be->src_stride;
- short *diff = &be->src_diff[0];
- unsigned char *predictor = &bd->predictor[0];
- vp8_subtract_b_sse2_impl(z, src_stride, diff, predictor, pitch);
-}
-
-#endif
-
-#if HAVE_SSSE3
-int vp8_fast_quantize_b_impl_ssse3(short *coeff_ptr,
- short *qcoeff_ptr, short *dequant_ptr,
- short *round_ptr,
- short *quant_ptr, short *dqcoeff_ptr);
-static void fast_quantize_b_ssse3(BLOCK *b, BLOCKD *d)
-{
- d->eob = vp8_fast_quantize_b_impl_ssse3(
- b->coeff,
- d->qcoeff,
- d->dequant,
- b->round,
- b->quant_fast,
- d->dqcoeff
- );
-}
-#if CONFIG_PSNR
-#if ARCH_X86_64
-typedef void ssimpf
-(
- unsigned char *s,
- int sp,
- unsigned char *r,
- int rp,
- unsigned long *sum_s,
- unsigned long *sum_r,
- unsigned long *sum_sq_s,
- unsigned long *sum_sq_r,
- unsigned long *sum_sxr
-);
-
-extern ssimpf vp8_ssim_parms_16x16_sse3;
-extern ssimpf vp8_ssim_parms_8x8_sse3;
-#endif
-#endif
-#endif
-
-
-void vp8_arch_x86_encoder_init(VP8_COMP *cpi)
-{
-#if CONFIG_RUNTIME_CPU_DETECT
- int flags = x86_simd_caps();
- int mmx_enabled = flags & HAS_MMX;
- int xmm_enabled = flags & HAS_SSE;
- int wmt_enabled = flags & HAS_SSE2;
- int SSE3Enabled = flags & HAS_SSE3;
- int SSSE3Enabled = flags & HAS_SSSE3;
- int SSE4_1Enabled = flags & HAS_SSE4_1;
-
- /* Note:
- *
- * This platform can be built without runtime CPU detection as well. If
- * you modify any of the function mappings present in this file, be sure
- * to also update them in static mapings (<arch>/filename_<arch>.h)
- */
-
- /* Override default functions with fastest ones for this CPU. */
-#if HAVE_MMX
- if (mmx_enabled)
- {
- cpi->rtcd.variance.sad16x16 = vp8_sad16x16_mmx;
- cpi->rtcd.variance.sad16x8 = vp8_sad16x8_mmx;
- cpi->rtcd.variance.sad8x16 = vp8_sad8x16_mmx;
- cpi->rtcd.variance.sad8x8 = vp8_sad8x8_mmx;
- cpi->rtcd.variance.sad4x4 = vp8_sad4x4_mmx;
-
- cpi->rtcd.variance.var4x4 = vp8_variance4x4_mmx;
- cpi->rtcd.variance.var8x8 = vp8_variance8x8_mmx;
- cpi->rtcd.variance.var8x16 = vp8_variance8x16_mmx;
- cpi->rtcd.variance.var16x8 = vp8_variance16x8_mmx;
- cpi->rtcd.variance.var16x16 = vp8_variance16x16_mmx;
-
- cpi->rtcd.variance.subpixvar4x4 = vp8_sub_pixel_variance4x4_mmx;
- cpi->rtcd.variance.subpixvar8x8 = vp8_sub_pixel_variance8x8_mmx;
- cpi->rtcd.variance.subpixvar8x16 = vp8_sub_pixel_variance8x16_mmx;
- cpi->rtcd.variance.subpixvar16x8 = vp8_sub_pixel_variance16x8_mmx;
- cpi->rtcd.variance.subpixvar16x16 = vp8_sub_pixel_variance16x16_mmx;
- cpi->rtcd.variance.halfpixvar16x16_h = vp8_variance_halfpixvar16x16_h_mmx;
- cpi->rtcd.variance.halfpixvar16x16_v = vp8_variance_halfpixvar16x16_v_mmx;
- cpi->rtcd.variance.halfpixvar16x16_hv = vp8_variance_halfpixvar16x16_hv_mmx;
- cpi->rtcd.variance.subpixmse16x16 = vp8_sub_pixel_mse16x16_mmx;
-
- cpi->rtcd.variance.mse16x16 = vp8_mse16x16_mmx;
- cpi->rtcd.variance.getmbss = vp8_get_mb_ss_mmx;
-
- cpi->rtcd.variance.get16x16prederror = vp8_get16x16pred_error_mmx;
- cpi->rtcd.variance.get8x8var = vp8_get8x8var_mmx;
- cpi->rtcd.variance.get16x16var = vp8_get16x16var_mmx;
- cpi->rtcd.variance.get4x4sse_cs = vp8_get4x4sse_cs_mmx;
-
- cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_mmx;
- cpi->rtcd.fdct.short8x4 = short_fdct8x4_mmx;
- cpi->rtcd.fdct.fast4x4 = vp8_short_fdct4x4_mmx;
- cpi->rtcd.fdct.fast8x4 = short_fdct8x4_mmx;
-
- cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_c;
-
- cpi->rtcd.encodemb.berr = vp8_block_error_mmx;
- cpi->rtcd.encodemb.mberr = mbblock_error_mmx;
- cpi->rtcd.encodemb.mbuverr = mbuverror_mmx;
- cpi->rtcd.encodemb.subb = subtract_b_mmx;
- cpi->rtcd.encodemb.submby = vp8_subtract_mby_mmx;
- cpi->rtcd.encodemb.submbuv = vp8_subtract_mbuv_mmx;
-
- /*cpi->rtcd.quantize.fastquantb = fast_quantize_b_mmx;*/
- }
-#endif
-
-#if HAVE_SSE2
- if (wmt_enabled)
- {
- cpi->rtcd.variance.sad16x16 = vp8_sad16x16_wmt;
- cpi->rtcd.variance.sad16x8 = vp8_sad16x8_wmt;
- cpi->rtcd.variance.sad8x16 = vp8_sad8x16_wmt;
- cpi->rtcd.variance.sad8x8 = vp8_sad8x8_wmt;
- cpi->rtcd.variance.sad4x4 = vp8_sad4x4_wmt;
-
- cpi->rtcd.variance.var4x4 = vp8_variance4x4_wmt;
- cpi->rtcd.variance.var8x8 = vp8_variance8x8_wmt;
- cpi->rtcd.variance.var8x16 = vp8_variance8x16_wmt;
- cpi->rtcd.variance.var16x8 = vp8_variance16x8_wmt;
- cpi->rtcd.variance.var16x16 = vp8_variance16x16_wmt;
-
- cpi->rtcd.variance.subpixvar4x4 = vp8_sub_pixel_variance4x4_wmt;
- cpi->rtcd.variance.subpixvar8x8 = vp8_sub_pixel_variance8x8_wmt;
- cpi->rtcd.variance.subpixvar8x16 = vp8_sub_pixel_variance8x16_wmt;
- cpi->rtcd.variance.subpixvar16x8 = vp8_sub_pixel_variance16x8_wmt;
- cpi->rtcd.variance.subpixvar16x16 = vp8_sub_pixel_variance16x16_wmt;
- cpi->rtcd.variance.halfpixvar16x16_h = vp8_variance_halfpixvar16x16_h_wmt;
- cpi->rtcd.variance.halfpixvar16x16_v = vp8_variance_halfpixvar16x16_v_wmt;
- cpi->rtcd.variance.halfpixvar16x16_hv = vp8_variance_halfpixvar16x16_hv_wmt;
- cpi->rtcd.variance.subpixmse16x16 = vp8_sub_pixel_mse16x16_wmt;
-
- cpi->rtcd.variance.mse16x16 = vp8_mse16x16_wmt;
- cpi->rtcd.variance.getmbss = vp8_get_mb_ss_sse2;
-
- cpi->rtcd.variance.get16x16prederror = vp8_get16x16pred_error_sse2;
- cpi->rtcd.variance.get8x8var = vp8_get8x8var_sse2;
- cpi->rtcd.variance.get16x16var = vp8_get16x16var_sse2;
-
-
- /* cpi->rtcd.variance.get4x4sse_cs not implemented for wmt */;
-
- cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_sse2;
- cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_sse2;
- cpi->rtcd.fdct.fast4x4 = vp8_short_fdct4x4_sse2;
- cpi->rtcd.fdct.fast8x4 = vp8_short_fdct8x4_sse2;
-
- cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_sse2 ;
-
- cpi->rtcd.encodemb.berr = vp8_block_error_xmm;
- cpi->rtcd.encodemb.mberr = mbblock_error_xmm;
- cpi->rtcd.encodemb.mbuverr = mbuverror_xmm;
- cpi->rtcd.encodemb.subb = subtract_b_sse2;
- cpi->rtcd.encodemb.submby = vp8_subtract_mby_sse2;
- cpi->rtcd.encodemb.submbuv = vp8_subtract_mbuv_sse2;
-
- cpi->rtcd.quantize.quantb = vp8_regular_quantize_b_sse2;
- cpi->rtcd.quantize.fastquantb = fast_quantize_b_sse2;
-
-#if !(CONFIG_REALTIME_ONLY)
- cpi->rtcd.temporal.apply = vp8_temporal_filter_apply_sse2;
-#endif
- }
-#endif
-
-#if HAVE_SSE3
- if (SSE3Enabled)
- {
- cpi->rtcd.variance.sad16x16 = vp8_sad16x16_sse3;
- cpi->rtcd.variance.sad16x16x3 = vp8_sad16x16x3_sse3;
- cpi->rtcd.variance.sad16x8x3 = vp8_sad16x8x3_sse3;
- cpi->rtcd.variance.sad8x16x3 = vp8_sad8x16x3_sse3;
- cpi->rtcd.variance.sad8x8x3 = vp8_sad8x8x3_sse3;
- cpi->rtcd.variance.sad4x4x3 = vp8_sad4x4x3_sse3;
-#if !(CONFIG_REALTIME_ONLY)
- cpi->rtcd.search.full_search = vp8_full_search_sadx3;
-#endif
- cpi->rtcd.variance.sad16x16x4d = vp8_sad16x16x4d_sse3;
- cpi->rtcd.variance.sad16x8x4d = vp8_sad16x8x4d_sse3;
- cpi->rtcd.variance.sad8x16x4d = vp8_sad8x16x4d_sse3;
- cpi->rtcd.variance.sad8x8x4d = vp8_sad8x8x4d_sse3;
- cpi->rtcd.variance.sad4x4x4d = vp8_sad4x4x4d_sse3;
- cpi->rtcd.search.diamond_search = vp8_diamond_search_sadx4;
- }
-#endif
-
-#if HAVE_SSSE3
- if (SSSE3Enabled)
- {
- cpi->rtcd.variance.sad16x16x3 = vp8_sad16x16x3_ssse3;
- cpi->rtcd.variance.sad16x8x3 = vp8_sad16x8x3_ssse3;
-
- cpi->rtcd.variance.subpixvar16x8 = vp8_sub_pixel_variance16x8_ssse3;
- cpi->rtcd.variance.subpixvar16x16 = vp8_sub_pixel_variance16x16_ssse3;
-
- cpi->rtcd.quantize.fastquantb = fast_quantize_b_ssse3;
-
-#if CONFIG_PSNR
-#if ARCH_X86_64
- cpi->rtcd.variance.ssimpf_8x8 = vp8_ssim_parms_8x8_sse3;
- cpi->rtcd.variance.ssimpf = vp8_ssim_parms_16x16_sse3;
-#endif
-#endif
-
- }
-#endif
-
-
-
-#if HAVE_SSE4_1
- if (SSE4_1Enabled)
- {
- cpi->rtcd.variance.sad16x16x8 = vp8_sad16x16x8_sse4;
- cpi->rtcd.variance.sad16x8x8 = vp8_sad16x8x8_sse4;
- cpi->rtcd.variance.sad8x16x8 = vp8_sad8x16x8_sse4;
- cpi->rtcd.variance.sad8x8x8 = vp8_sad8x8x8_sse4;
- cpi->rtcd.variance.sad4x4x8 = vp8_sad4x4x8_sse4;
-#if !(CONFIG_REALTIME_ONLY)
- cpi->rtcd.search.full_search = vp8_full_search_sadx8;
-#endif
- }
-#endif
-
-#endif
-}
diff --git a/vp8/vp8_common.mk b/vp8/vp8_common.mk
deleted file mode 100644
index 2a2f0cf..0000000
--- a/vp8/vp8_common.mk
+++ /dev/null
@@ -1,151 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-VP8_COMMON_SRCS-yes += vp8_common.mk
-VP8_COMMON_SRCS-yes += common/type_aliases.h
-VP8_COMMON_SRCS-yes += common/pragmas.h
-VP8_COMMON_SRCS-yes += common/ppflags.h
-VP8_COMMON_SRCS-yes += common/onyx.h
-VP8_COMMON_SRCS-yes += common/onyxd.h
-VP8_COMMON_SRCS-yes += common/alloccommon.c
-VP8_COMMON_SRCS-yes += common/blockd.c
-VP8_COMMON_SRCS-yes += common/coefupdateprobs.h
-VP8_COMMON_SRCS-yes += common/debugmodes.c
-VP8_COMMON_SRCS-yes += common/defaultcoefcounts.h
-VP8_COMMON_SRCS-yes += common/entropy.c
-VP8_COMMON_SRCS-yes += common/entropymode.c
-VP8_COMMON_SRCS-yes += common/entropymv.c
-VP8_COMMON_SRCS-yes += common/extend.c
-VP8_COMMON_SRCS-yes += common/filter.c
-VP8_COMMON_SRCS-yes += common/filter.h
-VP8_COMMON_SRCS-yes += common/findnearmv.c
-VP8_COMMON_SRCS-yes += common/generic/systemdependent.c
-VP8_COMMON_SRCS-yes += common/idctllm.c
-VP8_COMMON_SRCS-yes += common/alloccommon.h
-VP8_COMMON_SRCS-yes += common/blockd.h
-VP8_COMMON_SRCS-yes += common/common.h
-VP8_COMMON_SRCS-yes += common/common_types.h
-VP8_COMMON_SRCS-yes += common/entropy.h
-VP8_COMMON_SRCS-yes += common/entropymode.h
-VP8_COMMON_SRCS-yes += common/entropymv.h
-VP8_COMMON_SRCS-yes += common/extend.h
-VP8_COMMON_SRCS-yes += common/findnearmv.h
-VP8_COMMON_SRCS-yes += common/g_common.h
-VP8_COMMON_SRCS-yes += common/header.h
-VP8_COMMON_SRCS-yes += common/idct.h
-VP8_COMMON_SRCS-yes += common/invtrans.h
-VP8_COMMON_SRCS-yes += common/loopfilter.h
-VP8_COMMON_SRCS-yes += common/modecont.h
-VP8_COMMON_SRCS-yes += common/mv.h
-VP8_COMMON_SRCS-yes += common/onyxc_int.h
-VP8_COMMON_SRCS-yes += common/quant_common.h
-VP8_COMMON_SRCS-yes += common/recon.h
-VP8_COMMON_SRCS-yes += common/reconinter.h
-VP8_COMMON_SRCS-yes += common/reconintra.h
-VP8_COMMON_SRCS-yes += common/reconintra4x4.h
-VP8_COMMON_SRCS-yes += common/setupintrarecon.h
-VP8_COMMON_SRCS-yes += common/subpixel.h
-VP8_COMMON_SRCS-yes += common/swapyv12buffer.h
-VP8_COMMON_SRCS-yes += common/systemdependent.h
-VP8_COMMON_SRCS-yes += common/threading.h
-VP8_COMMON_SRCS-yes += common/treecoder.h
-VP8_COMMON_SRCS-yes += common/invtrans.c
-VP8_COMMON_SRCS-yes += common/loopfilter.c
-VP8_COMMON_SRCS-yes += common/loopfilter_filters.c
-VP8_COMMON_SRCS-yes += common/mbpitch.c
-VP8_COMMON_SRCS-yes += common/modecont.c
-VP8_COMMON_SRCS-yes += common/modecontext.c
-VP8_COMMON_SRCS-yes += common/quant_common.c
-VP8_COMMON_SRCS-yes += common/recon.c
-VP8_COMMON_SRCS-yes += common/reconinter.c
-VP8_COMMON_SRCS-yes += common/reconintra.c
-VP8_COMMON_SRCS-yes += common/reconintra4x4.c
-VP8_COMMON_SRCS-yes += common/setupintrarecon.c
-VP8_COMMON_SRCS-yes += common/swapyv12buffer.c
-VP8_COMMON_SRCS-$(CONFIG_POSTPROC_VISUALIZER) += common/textblit.c
-VP8_COMMON_SRCS-yes += common/treecoder.c
-
-VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/idct_x86.h
-VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/subpixel_x86.h
-VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/recon_x86.h
-VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/loopfilter_x86.h
-VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/postproc_x86.h
-VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/x86_systemdependent.c
-VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/vp8_asm_stubs.c
-VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/loopfilter_x86.c
-VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/postproc.h
-VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/postproc.c
-VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/idctllm_mmx.asm
-VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/iwalsh_mmx.asm
-VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/recon_mmx.asm
-VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/subpixel_mmx.asm
-VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/loopfilter_mmx.asm
-VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/idctllm_sse2.asm
-VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/recon_sse2.asm
-VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/subpixel_sse2.asm
-VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/loopfilter_sse2.asm
-VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/iwalsh_sse2.asm
-VP8_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/subpixel_ssse3.asm
-ifeq ($(CONFIG_POSTPROC),yes)
-VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/postproc_mmx.asm
-VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/postproc_sse2.asm
-endif
-
-VP8_COMMON_SRCS-$(ARCH_ARM) += common/asm_com_offsets.c
-VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/arm_systemdependent.c
-
-# common (c)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/bilinearfilter_arm.c
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/bilinearfilter_arm.h
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/filter_arm.c
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/loopfilter_arm.c
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/reconintra_arm.c
-
-# common (armv6)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/bilinearfilter_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/copymem8x4_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/copymem8x8_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/copymem16x16_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/dc_only_idct_add_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/iwalsh_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/filter_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/idct_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/loopfilter_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/recon_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/simpleloopfilter_v6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/sixtappredict8x4_v6$(ASM)
-
-# common (neon)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/bilinearpredict4x4_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/bilinearpredict8x4_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/bilinearpredict8x8_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/bilinearpredict16x16_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/copymem8x4_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/copymem8x8_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/copymem16x16_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/dc_only_idct_add_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/iwalsh_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/loopfilter_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/loopfiltersimplehorizontaledge_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/loopfiltersimpleverticaledge_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/mbloopfilter_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/recon2b_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/recon4b_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/reconb_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/shortidct4x4llm_1_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/shortidct4x4llm_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/sixtappredict4x4_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/sixtappredict8x4_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/sixtappredict8x8_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/sixtappredict16x16_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/recon16x16mb_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/buildintrapredictorsmby_neon$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/save_neon_reg$(ASM)
-VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/recon_neon.c
diff --git a/vp8/vp8_cx_iface.c b/vp8/vp8_cx_iface.c
deleted file mode 100644
index 2622738..0000000
--- a/vp8/vp8_cx_iface.c
+++ /dev/null
@@ -1,1231 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx/vpx_codec.h"
-#include "vpx/internal/vpx_codec_internal.h"
-#include "vpx_version.h"
-#include "vp8/encoder/onyx_int.h"
-#include "vpx/vp8e.h"
-#include "vp8/encoder/firstpass.h"
-#include "vp8/common/onyx.h"
-#include <stdlib.h>
-#include <string.h>
-
-/* This value is a sentinel for determining whether the user has set a mode
- * directly through the deprecated VP8E_SET_ENCODING_MODE control.
- */
-#define NO_MODE_SET 255
-
-struct vp8_extracfg
-{
- struct vpx_codec_pkt_list *pkt_list;
- vp8e_encoding_mode encoding_mode; /** best, good, realtime */
- int cpu_used; /** available cpu percentage in 1/16*/
- unsigned int enable_auto_alt_ref; /** if encoder decides to uses alternate reference frame */
- unsigned int noise_sensitivity;
- unsigned int Sharpness;
- unsigned int static_thresh;
- unsigned int token_partitions;
- unsigned int arnr_max_frames; /* alt_ref Noise Reduction Max Frame Count */
- unsigned int arnr_strength; /* alt_ref Noise Reduction Strength */
- unsigned int arnr_type; /* alt_ref filter type */
- vp8e_tuning tuning;
- unsigned int cq_level; /* constrained quality level */
-
-};
-
-struct extraconfig_map
-{
- int usage;
- struct vp8_extracfg cfg;
-};
-
-static const struct extraconfig_map extracfg_map[] =
-{
- {
- 0,
- {
- NULL,
-#if !(CONFIG_REALTIME_ONLY)
- VP8_BEST_QUALITY_ENCODING, /* Encoding Mode */
- 0, /* cpu_used */
-#else
- VP8_REAL_TIME_ENCODING, /* Encoding Mode */
- 4, /* cpu_used */
-#endif
- 0, /* enable_auto_alt_ref */
- 0, /* noise_sensitivity */
- 0, /* Sharpness */
- 0, /* static_thresh */
- VP8_ONE_TOKENPARTITION, /* token_partitions */
- 0, /* arnr_max_frames */
- 3, /* arnr_strength */
- 3, /* arnr_type*/
- 0, /* tuning*/
- 10, /* cq_level */
- }
- }
-};
-
-struct vpx_codec_alg_priv
-{
- vpx_codec_priv_t base;
- vpx_codec_enc_cfg_t cfg;
- struct vp8_extracfg vp8_cfg;
- VP8_CONFIG oxcf;
- VP8_PTR cpi;
- unsigned char *cx_data;
- unsigned int cx_data_sz;
- vpx_image_t preview_img;
- unsigned int next_frame_flag;
- vp8_postproc_cfg_t preview_ppcfg;
- vpx_codec_pkt_list_decl(64) pkt_list; // changed to accomendate the maximum number of lagged frames allowed
- int deprecated_mode;
- unsigned int fixed_kf_cntr;
-};
-
-
-static vpx_codec_err_t
-update_error_state(vpx_codec_alg_priv_t *ctx,
- const struct vpx_internal_error_info *error)
-{
- vpx_codec_err_t res;
-
- if ((res = error->error_code))
- ctx->base.err_detail = error->has_detail
- ? error->detail
- : NULL;
-
- return res;
-}
-
-
-#undef ERROR
-#define ERROR(str) do {\
- ctx->base.err_detail = str;\
- return VPX_CODEC_INVALID_PARAM;\
- } while(0)
-
-#define RANGE_CHECK(p,memb,lo,hi) do {\
- if(!(((p)->memb == lo || (p)->memb > (lo)) && (p)->memb <= hi)) \
- ERROR(#memb " out of range ["#lo".."#hi"]");\
- } while(0)
-
-#define RANGE_CHECK_HI(p,memb,hi) do {\
- if(!((p)->memb <= (hi))) \
- ERROR(#memb " out of range [.."#hi"]");\
- } while(0)
-
-#define RANGE_CHECK_LO(p,memb,lo) do {\
- if(!((p)->memb >= (lo))) \
- ERROR(#memb " out of range ["#lo"..]");\
- } while(0)
-
-#define RANGE_CHECK_BOOL(p,memb) do {\
- if(!!((p)->memb) != (p)->memb) ERROR(#memb " expected boolean");\
- } while(0)
-
-static vpx_codec_err_t validate_config(vpx_codec_alg_priv_t *ctx,
- const vpx_codec_enc_cfg_t *cfg,
- const struct vp8_extracfg *vp8_cfg)
-{
- RANGE_CHECK(cfg, g_w, 1, 16384);
- RANGE_CHECK(cfg, g_h, 1, 16384);
- RANGE_CHECK(cfg, g_timebase.den, 1, 1000000000);
- RANGE_CHECK(cfg, g_timebase.num, 1, cfg->g_timebase.den);
- RANGE_CHECK_HI(cfg, g_profile, 3);
- RANGE_CHECK_HI(cfg, rc_max_quantizer, 63);
- RANGE_CHECK_HI(cfg, rc_min_quantizer, cfg->rc_max_quantizer);
- RANGE_CHECK_HI(cfg, g_threads, 64);
-#if !(CONFIG_REALTIME_ONLY)
- RANGE_CHECK_HI(cfg, g_lag_in_frames, 25);
-#else
- RANGE_CHECK_HI(cfg, g_lag_in_frames, 0);
-#endif
- RANGE_CHECK(cfg, rc_end_usage, VPX_VBR, VPX_CQ);
- RANGE_CHECK_HI(cfg, rc_undershoot_pct, 100);
- RANGE_CHECK_HI(cfg, rc_2pass_vbr_bias_pct, 100);
- RANGE_CHECK(cfg, kf_mode, VPX_KF_DISABLED, VPX_KF_AUTO);
- //RANGE_CHECK_BOOL(cfg, g_delete_firstpassfile);
- RANGE_CHECK_BOOL(cfg, rc_resize_allowed);
- RANGE_CHECK_HI(cfg, rc_dropframe_thresh, 100);
- RANGE_CHECK_HI(cfg, rc_resize_up_thresh, 100);
- RANGE_CHECK_HI(cfg, rc_resize_down_thresh, 100);
-#if !(CONFIG_REALTIME_ONLY)
- RANGE_CHECK(cfg, g_pass, VPX_RC_ONE_PASS, VPX_RC_LAST_PASS);
-#else
- RANGE_CHECK(cfg, g_pass, VPX_RC_ONE_PASS, VPX_RC_ONE_PASS);
-#endif
-
- /* VP8 does not support a lower bound on the keyframe interval in
- * automatic keyframe placement mode.
- */
- if (cfg->kf_mode != VPX_KF_DISABLED && cfg->kf_min_dist != cfg->kf_max_dist
- && cfg->kf_min_dist > 0)
- ERROR("kf_min_dist not supported in auto mode, use 0 "
- "or kf_max_dist instead.");
-
- RANGE_CHECK_BOOL(vp8_cfg, enable_auto_alt_ref);
-#if !(CONFIG_REALTIME_ONLY)
- RANGE_CHECK(vp8_cfg, encoding_mode, VP8_BEST_QUALITY_ENCODING, VP8_REAL_TIME_ENCODING);
- RANGE_CHECK(vp8_cfg, cpu_used, -16, 16);
- RANGE_CHECK_HI(vp8_cfg, noise_sensitivity, 6);
-#else
- RANGE_CHECK(vp8_cfg, encoding_mode, VP8_REAL_TIME_ENCODING, VP8_REAL_TIME_ENCODING);
-
- if (!((vp8_cfg->cpu_used >= -16 && vp8_cfg->cpu_used <= -4) || (vp8_cfg->cpu_used >= 4 && vp8_cfg->cpu_used <= 16)))
- ERROR("cpu_used out of range [-16..-4] or [4..16]");
-
- RANGE_CHECK(vp8_cfg, noise_sensitivity, 0, 0);
-#endif
-
- RANGE_CHECK(vp8_cfg, token_partitions, VP8_ONE_TOKENPARTITION, VP8_EIGHT_TOKENPARTITION);
- RANGE_CHECK_HI(vp8_cfg, Sharpness, 7);
- RANGE_CHECK(vp8_cfg, arnr_max_frames, 0, 15);
- RANGE_CHECK_HI(vp8_cfg, arnr_strength, 6);
- RANGE_CHECK(vp8_cfg, arnr_type, 1, 3);
- RANGE_CHECK(vp8_cfg, cq_level, 0, 63);
-
-#if !(CONFIG_REALTIME_ONLY)
- if (cfg->g_pass == VPX_RC_LAST_PASS)
- {
- int mb_r = (cfg->g_h + 15) / 16;
- int mb_c = (cfg->g_w + 15) / 16;
- size_t packet_sz = sizeof(FIRSTPASS_STATS);
- int n_packets = cfg->rc_twopass_stats_in.sz / packet_sz;
- FIRSTPASS_STATS *stats;
-
- if (!cfg->rc_twopass_stats_in.buf)
- ERROR("rc_twopass_stats_in.buf not set.");
-
- if (cfg->rc_twopass_stats_in.sz % packet_sz)
- ERROR("rc_twopass_stats_in.sz indicates truncated packet.");
-
- if (cfg->rc_twopass_stats_in.sz < 2 * packet_sz)
- ERROR("rc_twopass_stats_in requires at least two packets.");
-
- stats = (void*)((char *)cfg->rc_twopass_stats_in.buf
- + (n_packets - 1) * packet_sz);
-
- if ((int)(stats->count + 0.5) != n_packets - 1)
- ERROR("rc_twopass_stats_in missing EOS stats packet");
- }
-#endif
-
- return VPX_CODEC_OK;
-}
-
-
-static vpx_codec_err_t validate_img(vpx_codec_alg_priv_t *ctx,
- const vpx_image_t *img)
-{
- switch (img->fmt)
- {
- case VPX_IMG_FMT_YV12:
- case VPX_IMG_FMT_I420:
- case VPX_IMG_FMT_VPXI420:
- case VPX_IMG_FMT_VPXYV12:
- break;
- default:
- ERROR("Invalid image format. Only YV12 and I420 images are supported");
- }
-
- if ((img->d_w != ctx->cfg.g_w) || (img->d_h != ctx->cfg.g_h))
- ERROR("Image size must match encoder init configuration size");
-
- return VPX_CODEC_OK;
-}
-
-
-static vpx_codec_err_t set_vp8e_config(VP8_CONFIG *oxcf,
- vpx_codec_enc_cfg_t cfg,
- struct vp8_extracfg vp8_cfg)
-{
- oxcf->multi_threaded = cfg.g_threads;
- oxcf->Version = cfg.g_profile;
-
- oxcf->Width = cfg.g_w;
- oxcf->Height = cfg.g_h;
- /* guess a frame rate if out of whack, use 30 */
- oxcf->frame_rate = (double)(cfg.g_timebase.den) / (double)(cfg.g_timebase.num);
-
- if (oxcf->frame_rate > 180)
- {
- oxcf->frame_rate = 30;
- }
-
- oxcf->error_resilient_mode = cfg.g_error_resilient;
-
- switch (cfg.g_pass)
- {
- case VPX_RC_ONE_PASS:
- oxcf->Mode = MODE_BESTQUALITY;
- break;
- case VPX_RC_FIRST_PASS:
- oxcf->Mode = MODE_FIRSTPASS;
- break;
- case VPX_RC_LAST_PASS:
- oxcf->Mode = MODE_SECONDPASS_BEST;
- break;
- }
-
- if (cfg.g_pass == VPX_RC_FIRST_PASS)
- {
- oxcf->allow_lag = 0;
- oxcf->lag_in_frames = 0;
- }
- else
- {
- oxcf->allow_lag = (cfg.g_lag_in_frames) > 0;
- oxcf->lag_in_frames = cfg.g_lag_in_frames;
- }
-
- oxcf->allow_df = (cfg.rc_dropframe_thresh > 0);
- oxcf->drop_frames_water_mark = cfg.rc_dropframe_thresh;
-
- oxcf->allow_spatial_resampling = cfg.rc_resize_allowed;
- oxcf->resample_up_water_mark = cfg.rc_resize_up_thresh;
- oxcf->resample_down_water_mark = cfg.rc_resize_down_thresh;
-
- if (cfg.rc_end_usage == VPX_VBR)
- {
- oxcf->end_usage = USAGE_LOCAL_FILE_PLAYBACK;
- }
- else if (cfg.rc_end_usage == VPX_CBR)
- {
- oxcf->end_usage = USAGE_STREAM_FROM_SERVER;
- }
- else if (cfg.rc_end_usage == VPX_CQ)
- {
- oxcf->end_usage = USAGE_CONSTRAINED_QUALITY;
- }
-
- oxcf->target_bandwidth = cfg.rc_target_bitrate;
-
- oxcf->best_allowed_q = cfg.rc_min_quantizer;
- oxcf->worst_allowed_q = cfg.rc_max_quantizer;
- oxcf->cq_level = vp8_cfg.cq_level;
- oxcf->fixed_q = -1;
-
- oxcf->under_shoot_pct = cfg.rc_undershoot_pct;
- //oxcf->over_shoot_pct = cfg.rc_overshoot_pct;
-
- oxcf->maximum_buffer_size = cfg.rc_buf_sz;
- oxcf->starting_buffer_level = cfg.rc_buf_initial_sz;
- oxcf->optimal_buffer_level = cfg.rc_buf_optimal_sz;
-
- oxcf->two_pass_vbrbias = cfg.rc_2pass_vbr_bias_pct;
- oxcf->two_pass_vbrmin_section = cfg.rc_2pass_vbr_minsection_pct;
- oxcf->two_pass_vbrmax_section = cfg.rc_2pass_vbr_maxsection_pct;
-
- oxcf->auto_key = cfg.kf_mode == VPX_KF_AUTO
- && cfg.kf_min_dist != cfg.kf_max_dist;
- //oxcf->kf_min_dist = cfg.kf_min_dis;
- oxcf->key_freq = cfg.kf_max_dist;
-
- //oxcf->delete_first_pass_file = cfg.g_delete_firstpassfile;
- //strcpy(oxcf->first_pass_file, cfg.g_firstpass_file);
-
- oxcf->cpu_used = vp8_cfg.cpu_used;
- oxcf->encode_breakout = vp8_cfg.static_thresh;
- oxcf->play_alternate = vp8_cfg.enable_auto_alt_ref;
- oxcf->noise_sensitivity = vp8_cfg.noise_sensitivity;
- oxcf->Sharpness = vp8_cfg.Sharpness;
- oxcf->token_partitions = vp8_cfg.token_partitions;
-
- oxcf->two_pass_stats_in = cfg.rc_twopass_stats_in;
- oxcf->output_pkt_list = vp8_cfg.pkt_list;
-
- oxcf->arnr_max_frames = vp8_cfg.arnr_max_frames;
- oxcf->arnr_strength = vp8_cfg.arnr_strength;
- oxcf->arnr_type = vp8_cfg.arnr_type;
-
- oxcf->tuning = vp8_cfg.tuning;
-
- /*
- printf("Current VP8 Settings: \n");
- printf("target_bandwidth: %d\n", oxcf->target_bandwidth);
- printf("noise_sensitivity: %d\n", oxcf->noise_sensitivity);
- printf("Sharpness: %d\n", oxcf->Sharpness);
- printf("cpu_used: %d\n", oxcf->cpu_used);
- printf("Mode: %d\n", oxcf->Mode);
- printf("delete_first_pass_file: %d\n", oxcf->delete_first_pass_file);
- printf("auto_key: %d\n", oxcf->auto_key);
- printf("key_freq: %d\n", oxcf->key_freq);
- printf("end_usage: %d\n", oxcf->end_usage);
- printf("under_shoot_pct: %d\n", oxcf->under_shoot_pct);
- printf("starting_buffer_level: %d\n", oxcf->starting_buffer_level);
- printf("optimal_buffer_level: %d\n", oxcf->optimal_buffer_level);
- printf("maximum_buffer_size: %d\n", oxcf->maximum_buffer_size);
- printf("fixed_q: %d\n", oxcf->fixed_q);
- printf("worst_allowed_q: %d\n", oxcf->worst_allowed_q);
- printf("best_allowed_q: %d\n", oxcf->best_allowed_q);
- printf("allow_spatial_resampling: %d\n", oxcf->allow_spatial_resampling);
- printf("resample_down_water_mark: %d\n", oxcf->resample_down_water_mark);
- printf("resample_up_water_mark: %d\n", oxcf->resample_up_water_mark);
- printf("allow_df: %d\n", oxcf->allow_df);
- printf("drop_frames_water_mark: %d\n", oxcf->drop_frames_water_mark);
- printf("two_pass_vbrbias: %d\n", oxcf->two_pass_vbrbias);
- printf("two_pass_vbrmin_section: %d\n", oxcf->two_pass_vbrmin_section);
- printf("two_pass_vbrmax_section: %d\n", oxcf->two_pass_vbrmax_section);
- printf("allow_lag: %d\n", oxcf->allow_lag);
- printf("lag_in_frames: %d\n", oxcf->lag_in_frames);
- printf("play_alternate: %d\n", oxcf->play_alternate);
- printf("Version: %d\n", oxcf->Version);
- printf("multi_threaded: %d\n", oxcf->multi_threaded);
- printf("encode_breakout: %d\n", oxcf->encode_breakout);
- */
- return VPX_CODEC_OK;
-}
-
-static vpx_codec_err_t vp8e_set_config(vpx_codec_alg_priv_t *ctx,
- const vpx_codec_enc_cfg_t *cfg)
-{
- vpx_codec_err_t res;
-
- if ((cfg->g_w != ctx->cfg.g_w) || (cfg->g_h != ctx->cfg.g_h))
- ERROR("Cannot change width or height after initialization");
-
- /* Prevent increasing lag_in_frames. This check is stricter than it needs
- * to be -- the limit is not increasing past the first lag_in_frames
- * value, but we don't track the initial config, only the last successful
- * config.
- */
- if ((cfg->g_lag_in_frames > ctx->cfg.g_lag_in_frames))
- ERROR("Cannot increase lag_in_frames");
-
- res = validate_config(ctx, cfg, &ctx->vp8_cfg);
-
- if (!res)
- {
- ctx->cfg = *cfg;
- set_vp8e_config(&ctx->oxcf, ctx->cfg, ctx->vp8_cfg);
- vp8_change_config(ctx->cpi, &ctx->oxcf);
- }
-
- return res;
-}
-
-
-int vp8_reverse_trans(int);
-
-
-static vpx_codec_err_t get_param(vpx_codec_alg_priv_t *ctx,
- int ctrl_id,
- va_list args)
-{
- void *arg = va_arg(args, void *);
-
-#define MAP(id, var) case id: *(RECAST(id, arg)) = var; break
-
- if (!arg)
- return VPX_CODEC_INVALID_PARAM;
-
- switch (ctrl_id)
- {
- MAP(VP8E_GET_LAST_QUANTIZER, vp8_get_quantizer(ctx->cpi));
- MAP(VP8E_GET_LAST_QUANTIZER_64, vp8_reverse_trans(vp8_get_quantizer(ctx->cpi)));
- }
-
- return VPX_CODEC_OK;
-#undef MAP
-}
-
-
-static vpx_codec_err_t set_param(vpx_codec_alg_priv_t *ctx,
- int ctrl_id,
- va_list args)
-{
- vpx_codec_err_t res = VPX_CODEC_OK;
- struct vp8_extracfg xcfg = ctx->vp8_cfg;
-
-#define MAP(id, var) case id: var = CAST(id, args); break;
-
- switch (ctrl_id)
- {
- MAP(VP8E_SET_ENCODING_MODE, ctx->deprecated_mode);
- MAP(VP8E_SET_CPUUSED, xcfg.cpu_used);
- MAP(VP8E_SET_ENABLEAUTOALTREF, xcfg.enable_auto_alt_ref);
- MAP(VP8E_SET_NOISE_SENSITIVITY, xcfg.noise_sensitivity);
- MAP(VP8E_SET_SHARPNESS, xcfg.Sharpness);
- MAP(VP8E_SET_STATIC_THRESHOLD, xcfg.static_thresh);
- MAP(VP8E_SET_TOKEN_PARTITIONS, xcfg.token_partitions);
-
- MAP(VP8E_SET_ARNR_MAXFRAMES, xcfg.arnr_max_frames);
- MAP(VP8E_SET_ARNR_STRENGTH , xcfg.arnr_strength);
- MAP(VP8E_SET_ARNR_TYPE , xcfg.arnr_type);
- MAP(VP8E_SET_TUNING, xcfg.tuning);
- MAP(VP8E_SET_CQ_LEVEL, xcfg.cq_level);
-
- }
-
- res = validate_config(ctx, &ctx->cfg, &xcfg);
-
- if (!res)
- {
- ctx->vp8_cfg = xcfg;
- set_vp8e_config(&ctx->oxcf, ctx->cfg, ctx->vp8_cfg);
- vp8_change_config(ctx->cpi, &ctx->oxcf);
- }
-
- return res;
-#undef MAP
-}
-static vpx_codec_err_t vp8e_init(vpx_codec_ctx_t *ctx)
-{
- vpx_codec_err_t res = VPX_DEC_OK;
- struct vpx_codec_alg_priv *priv;
- vpx_codec_enc_cfg_t *cfg;
- unsigned int i;
-
- VP8_PTR optr;
-
- if (!ctx->priv)
- {
- priv = calloc(1, sizeof(struct vpx_codec_alg_priv));
-
- if (!priv)
- {
- return VPX_CODEC_MEM_ERROR;
- }
-
- ctx->priv = &priv->base;
- ctx->priv->sz = sizeof(*ctx->priv);
- ctx->priv->iface = ctx->iface;
- ctx->priv->alg_priv = priv;
- ctx->priv->init_flags = ctx->init_flags;
-
- if (ctx->config.enc)
- {
- /* Update the reference to the config structure to an
- * internal copy.
- */
- ctx->priv->alg_priv->cfg = *ctx->config.enc;
- ctx->config.enc = &ctx->priv->alg_priv->cfg;
- }
-
- cfg = &ctx->priv->alg_priv->cfg;
-
- /* Select the extra vp6 configuration table based on the current
- * usage value. If the current usage value isn't found, use the
- * values for usage case 0.
- */
- for (i = 0;
- extracfg_map[i].usage && extracfg_map[i].usage != cfg->g_usage;
- i++);
-
- priv->vp8_cfg = extracfg_map[i].cfg;
- priv->vp8_cfg.pkt_list = &priv->pkt_list.head;
-
- priv->cx_data_sz = priv->cfg.g_w * priv->cfg.g_h * 3 / 2 * 2;
-
- if (priv->cx_data_sz < 4096) priv->cx_data_sz = 4096;
-
- priv->cx_data = malloc(priv->cx_data_sz);
-
- if (!priv->cx_data)
- {
- return VPX_CODEC_MEM_ERROR;
- }
-
- priv->deprecated_mode = NO_MODE_SET;
-
- vp8_initialize();
-
- res = validate_config(priv, &priv->cfg, &priv->vp8_cfg);
-
- if (!res)
- {
- set_vp8e_config(&ctx->priv->alg_priv->oxcf,
- ctx->priv->alg_priv->cfg,
- ctx->priv->alg_priv->vp8_cfg);
- optr = vp8_create_compressor(&ctx->priv->alg_priv->oxcf);
-
- if (!optr)
- res = VPX_CODEC_MEM_ERROR;
- else
- ctx->priv->alg_priv->cpi = optr;
- }
- }
-
- return res;
-}
-
-static vpx_codec_err_t vp8e_destroy(vpx_codec_alg_priv_t *ctx)
-{
-
- free(ctx->cx_data);
- vp8_remove_compressor(&ctx->cpi);
- free(ctx);
- return VPX_CODEC_OK;
-}
-
-static vpx_codec_err_t image2yuvconfig(const vpx_image_t *img,
- YV12_BUFFER_CONFIG *yv12)
-{
- vpx_codec_err_t res = VPX_CODEC_OK;
- yv12->y_buffer = img->planes[VPX_PLANE_Y];
- yv12->u_buffer = img->planes[VPX_PLANE_U];
- yv12->v_buffer = img->planes[VPX_PLANE_V];
-
- yv12->y_width = img->d_w;
- yv12->y_height = img->d_h;
- yv12->uv_width = (1 + yv12->y_width) / 2;
- yv12->uv_height = (1 + yv12->y_height) / 2;
-
- yv12->y_stride = img->stride[VPX_PLANE_Y];
- yv12->uv_stride = img->stride[VPX_PLANE_U];
-
- yv12->border = (img->stride[VPX_PLANE_Y] - img->w) / 2;
- yv12->clrtype = (img->fmt == VPX_IMG_FMT_VPXI420 || img->fmt == VPX_IMG_FMT_VPXYV12); //REG_YUV = 0
- return res;
-}
-
-static void pick_quickcompress_mode(vpx_codec_alg_priv_t *ctx,
- unsigned long duration,
- unsigned long deadline)
-{
- unsigned int new_qc;
-
-#if !(CONFIG_REALTIME_ONLY)
- /* Use best quality mode if no deadline is given. */
- new_qc = MODE_BESTQUALITY;
-
- if (deadline)
- {
- uint64_t duration_us;
-
- /* Convert duration parameter from stream timebase to microseconds */
- duration_us = (uint64_t)duration * 1000000
- * (uint64_t)ctx->cfg.g_timebase.num
- / (uint64_t)ctx->cfg.g_timebase.den;
-
- /* If the deadline is more that the duration this frame is to be shown,
- * use good quality mode. Otherwise use realtime mode.
- */
- new_qc = (deadline > duration_us) ? MODE_GOODQUALITY : MODE_REALTIME;
- }
-
-#else
- new_qc = MODE_REALTIME;
-#endif
-
- switch (ctx->deprecated_mode)
- {
- case VP8_BEST_QUALITY_ENCODING:
- new_qc = MODE_BESTQUALITY;
- break;
- case VP8_GOOD_QUALITY_ENCODING:
- new_qc = MODE_GOODQUALITY;
- break;
- case VP8_REAL_TIME_ENCODING:
- new_qc = MODE_REALTIME;
- break;
- }
-
- if (ctx->cfg.g_pass == VPX_RC_FIRST_PASS)
- new_qc = MODE_FIRSTPASS;
- else if (ctx->cfg.g_pass == VPX_RC_LAST_PASS)
- new_qc = (new_qc == MODE_BESTQUALITY)
- ? MODE_SECONDPASS_BEST
- : MODE_SECONDPASS;
-
- if (ctx->oxcf.Mode != new_qc)
- {
- ctx->oxcf.Mode = new_qc;
- vp8_change_config(ctx->cpi, &ctx->oxcf);
- }
-}
-
-
-static vpx_codec_err_t vp8e_encode(vpx_codec_alg_priv_t *ctx,
- const vpx_image_t *img,
- vpx_codec_pts_t pts,
- unsigned long duration,
- vpx_enc_frame_flags_t flags,
- unsigned long deadline)
-{
- vpx_codec_err_t res = VPX_CODEC_OK;
-
- if (img)
- res = validate_img(ctx, img);
-
- pick_quickcompress_mode(ctx, duration, deadline);
- vpx_codec_pkt_list_init(&ctx->pkt_list);
-
- /* Handle Flags */
- if (((flags & VP8_EFLAG_NO_UPD_GF) && (flags & VP8_EFLAG_FORCE_GF))
- || ((flags & VP8_EFLAG_NO_UPD_ARF) && (flags & VP8_EFLAG_FORCE_ARF)))
- {
- ctx->base.err_detail = "Conflicting flags.";
- return VPX_CODEC_INVALID_PARAM;
- }
-
- if (flags & (VP8_EFLAG_NO_REF_LAST | VP8_EFLAG_NO_REF_GF
- | VP8_EFLAG_NO_REF_ARF))
- {
- int ref = 7;
-
- if (flags & VP8_EFLAG_NO_REF_LAST)
- ref ^= VP8_LAST_FLAG;
-
- if (flags & VP8_EFLAG_NO_REF_GF)
- ref ^= VP8_GOLD_FLAG;
-
- if (flags & VP8_EFLAG_NO_REF_ARF)
- ref ^= VP8_ALT_FLAG;
-
- vp8_use_as_reference(ctx->cpi, ref);
- }
-
- if (flags & (VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF
- | VP8_EFLAG_NO_UPD_ARF | VP8_EFLAG_FORCE_GF
- | VP8_EFLAG_FORCE_ARF))
- {
- int upd = 7;
-
- if (flags & VP8_EFLAG_NO_UPD_LAST)
- upd ^= VP8_LAST_FLAG;
-
- if (flags & VP8_EFLAG_NO_UPD_GF)
- upd ^= VP8_GOLD_FLAG;
-
- if (flags & VP8_EFLAG_NO_UPD_ARF)
- upd ^= VP8_ALT_FLAG;
-
- vp8_update_reference(ctx->cpi, upd);
- }
-
- if (flags & VP8_EFLAG_NO_UPD_ENTROPY)
- {
- vp8_update_entropy(ctx->cpi, 0);
- }
-
- /* Handle fixed keyframe intervals */
- if (ctx->cfg.kf_mode == VPX_KF_AUTO
- && ctx->cfg.kf_min_dist == ctx->cfg.kf_max_dist)
- {
- if (++ctx->fixed_kf_cntr > ctx->cfg.kf_min_dist)
- {
- flags |= VPX_EFLAG_FORCE_KF;
- ctx->fixed_kf_cntr = 1;
- }
- }
-
- /* Initialize the encoder instance on the first frame*/
- if (!res && ctx->cpi)
- {
- unsigned int lib_flags;
- YV12_BUFFER_CONFIG sd;
- INT64 dst_time_stamp, dst_end_time_stamp;
- unsigned long size, cx_data_sz;
- unsigned char *cx_data;
-
- /* Set up internal flags */
- if (ctx->base.init_flags & VPX_CODEC_USE_PSNR)
- ((VP8_COMP *)ctx->cpi)->b_calculate_psnr = 1;
-
- /* Convert API flags to internal codec lib flags */
- lib_flags = (flags & VPX_EFLAG_FORCE_KF) ? FRAMEFLAGS_KEY : 0;
-
- /* vp8 use 10,000,000 ticks/second as time stamp */
- dst_time_stamp = pts * 10000000 * ctx->cfg.g_timebase.num / ctx->cfg.g_timebase.den;
- dst_end_time_stamp = (pts + duration) * 10000000 * ctx->cfg.g_timebase.num / ctx->cfg.g_timebase.den;
-
- if (img != NULL)
- {
- res = image2yuvconfig(img, &sd);
-
- if (vp8_receive_raw_frame(ctx->cpi, ctx->next_frame_flag | lib_flags,
- &sd, dst_time_stamp, dst_end_time_stamp))
- {
- VP8_COMP *cpi = (VP8_COMP *)ctx->cpi;
- res = update_error_state(ctx, &cpi->common.error);
- }
-
- /* reset for next frame */
- ctx->next_frame_flag = 0;
- }
-
- cx_data = ctx->cx_data;
- cx_data_sz = ctx->cx_data_sz;
- lib_flags = 0;
-
- while (cx_data_sz >= ctx->cx_data_sz / 2
- && -1 != vp8_get_compressed_data(ctx->cpi, &lib_flags, &size, cx_data, &dst_time_stamp, &dst_end_time_stamp, !img))
- {
- if (size)
- {
- vpx_codec_pts_t round, delta;
- vpx_codec_cx_pkt_t pkt;
- VP8_COMP *cpi = (VP8_COMP *)ctx->cpi;
-
- /* Add the frame packet to the list of returned packets. */
- round = 1000000 * ctx->cfg.g_timebase.num / 2 - 1;
- delta = (dst_end_time_stamp - dst_time_stamp);
- pkt.kind = VPX_CODEC_CX_FRAME_PKT;
- pkt.data.frame.buf = cx_data;
- pkt.data.frame.sz = size;
- pkt.data.frame.pts =
- (dst_time_stamp * ctx->cfg.g_timebase.den + round)
- / ctx->cfg.g_timebase.num / 10000000;
- pkt.data.frame.duration =
- (delta * ctx->cfg.g_timebase.den + round)
- / ctx->cfg.g_timebase.num / 10000000;
- pkt.data.frame.flags = lib_flags << 16;
-
- if (lib_flags & FRAMEFLAGS_KEY)
- pkt.data.frame.flags |= VPX_FRAME_IS_KEY;
-
- if (!cpi->common.show_frame)
- {
- pkt.data.frame.flags |= VPX_FRAME_IS_INVISIBLE;
-
- // This timestamp should be as close as possible to the
- // prior PTS so that if a decoder uses pts to schedule when
- // to do this, we start right after last frame was decoded.
- // Invisible frames have no duration.
- pkt.data.frame.pts = ((cpi->last_time_stamp_seen
- * ctx->cfg.g_timebase.den + round)
- / ctx->cfg.g_timebase.num / 10000000) + 1;
- pkt.data.frame.duration = 0;
- }
-
- vpx_codec_pkt_list_add(&ctx->pkt_list.head, &pkt);
-
- //printf("timestamp: %lld, duration: %d\n", pkt->data.frame.pts, pkt->data.frame.duration);
- cx_data += size;
- cx_data_sz -= size;
- }
- }
- }
-
- return res;
-}
-
-
-static const vpx_codec_cx_pkt_t *vp8e_get_cxdata(vpx_codec_alg_priv_t *ctx,
- vpx_codec_iter_t *iter)
-{
- return vpx_codec_pkt_list_get(&ctx->pkt_list.head, iter);
-}
-
-static vpx_codec_err_t vp8e_set_reference(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
- vpx_ref_frame_t *data = va_arg(args, vpx_ref_frame_t *);
-
- if (data)
- {
- vpx_ref_frame_t *frame = (vpx_ref_frame_t *)data;
- YV12_BUFFER_CONFIG sd;
-
- image2yuvconfig(&frame->img, &sd);
- vp8_set_reference(ctx->cpi, frame->frame_type, &sd);
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-
-}
-
-static vpx_codec_err_t vp8e_get_reference(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
-
- vpx_ref_frame_t *data = va_arg(args, vpx_ref_frame_t *);
-
- if (data)
- {
- vpx_ref_frame_t *frame = (vpx_ref_frame_t *)data;
- YV12_BUFFER_CONFIG sd;
-
- image2yuvconfig(&frame->img, &sd);
- vp8_get_reference(ctx->cpi, frame->frame_type, &sd);
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-}
-
-static vpx_codec_err_t vp8e_set_previewpp(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
-#if CONFIG_POSTPROC
- vp8_postproc_cfg_t *data = va_arg(args, vp8_postproc_cfg_t *);
- (void)ctr_id;
-
- if (data)
- {
- ctx->preview_ppcfg = *((vp8_postproc_cfg_t *)data);
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-#else
- (void)ctx;
- (void)ctr_id;
- (void)args;
- return VPX_CODEC_INCAPABLE;
-#endif
-}
-
-
-static vpx_image_t *vp8e_get_preview(vpx_codec_alg_priv_t *ctx)
-{
-
- YV12_BUFFER_CONFIG sd;
- vp8_ppflags_t flags = {0};
-
- if (ctx->preview_ppcfg.post_proc_flag)
- {
- flags.post_proc_flag = ctx->preview_ppcfg.post_proc_flag;
- flags.deblocking_level = ctx->preview_ppcfg.deblocking_level;
- flags.noise_level = ctx->preview_ppcfg.noise_level;
- }
-
- if (0 == vp8_get_preview_raw_frame(ctx->cpi, &sd, &flags))
- {
-
- /*
- vpx_img_wrap(&ctx->preview_img, VPX_IMG_FMT_YV12,
- sd.y_width + 2*VP8BORDERINPIXELS,
- sd.y_height + 2*VP8BORDERINPIXELS,
- 1,
- sd.buffer_alloc);
- vpx_img_set_rect(&ctx->preview_img,
- VP8BORDERINPIXELS, VP8BORDERINPIXELS,
- sd.y_width, sd.y_height);
- */
-
- ctx->preview_img.bps = 12;
- ctx->preview_img.planes[VPX_PLANE_Y] = sd.y_buffer;
- ctx->preview_img.planes[VPX_PLANE_U] = sd.u_buffer;
- ctx->preview_img.planes[VPX_PLANE_V] = sd.v_buffer;
-
- if (sd.clrtype == REG_YUV)
- ctx->preview_img.fmt = VPX_IMG_FMT_I420;
- else
- ctx->preview_img.fmt = VPX_IMG_FMT_VPXI420;
-
- ctx->preview_img.x_chroma_shift = 1;
- ctx->preview_img.y_chroma_shift = 1;
-
- ctx->preview_img.d_w = sd.y_width;
- ctx->preview_img.d_h = sd.y_height;
- ctx->preview_img.stride[VPX_PLANE_Y] = sd.y_stride;
- ctx->preview_img.stride[VPX_PLANE_U] = sd.uv_stride;
- ctx->preview_img.stride[VPX_PLANE_V] = sd.uv_stride;
- ctx->preview_img.w = sd.y_width;
- ctx->preview_img.h = sd.y_height;
-
- return &ctx->preview_img;
- }
- else
- return NULL;
-}
-
-static vpx_codec_err_t vp8e_update_entropy(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
- int update = va_arg(args, int);
- vp8_update_entropy(ctx->cpi, update);
- return VPX_CODEC_OK;
-
-}
-
-static vpx_codec_err_t vp8e_update_reference(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
- int update = va_arg(args, int);
- vp8_update_reference(ctx->cpi, update);
- return VPX_CODEC_OK;
-}
-
-static vpx_codec_err_t vp8e_use_reference(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
- int reference_flag = va_arg(args, int);
- vp8_use_as_reference(ctx->cpi, reference_flag);
- return VPX_CODEC_OK;
-}
-
-static vpx_codec_err_t vp8e_set_roi_map(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
- vpx_roi_map_t *data = va_arg(args, vpx_roi_map_t *);
-
- if (data)
- {
- vpx_roi_map_t *roi = (vpx_roi_map_t *)data;
-
- if (!vp8_set_roimap(ctx->cpi, roi->roi_map, roi->rows, roi->cols, roi->delta_q, roi->delta_lf, roi->static_threshold))
- return VPX_CODEC_OK;
- else
- return VPX_CODEC_INVALID_PARAM;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-}
-
-
-static vpx_codec_err_t vp8e_set_activemap(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
- vpx_active_map_t *data = va_arg(args, vpx_active_map_t *);
-
- if (data)
- {
-
- vpx_active_map_t *map = (vpx_active_map_t *)data;
-
- if (!vp8_set_active_map(ctx->cpi, map->active_map, map->rows, map->cols))
- return VPX_CODEC_OK;
- else
- return VPX_CODEC_INVALID_PARAM;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-}
-
-static vpx_codec_err_t vp8e_set_scalemode(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
-
- vpx_scaling_mode_t *data = va_arg(args, vpx_scaling_mode_t *);
-
- if (data)
- {
- int res;
- vpx_scaling_mode_t scalemode = *(vpx_scaling_mode_t *)data ;
- res = vp8_set_internal_size(ctx->cpi, scalemode.h_scaling_mode, scalemode.v_scaling_mode);
-
- if (!res)
- {
- /*force next frame a key frame to effect scaling mode */
- ctx->next_frame_flag |= FRAMEFLAGS_KEY;
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-}
-
-
-static vpx_codec_ctrl_fn_map_t vp8e_ctf_maps[] =
-{
- {VP8_SET_REFERENCE, vp8e_set_reference},
- {VP8_COPY_REFERENCE, vp8e_get_reference},
- {VP8_SET_POSTPROC, vp8e_set_previewpp},
- {VP8E_UPD_ENTROPY, vp8e_update_entropy},
- {VP8E_UPD_REFERENCE, vp8e_update_reference},
- {VP8E_USE_REFERENCE, vp8e_use_reference},
- {VP8E_SET_ROI_MAP, vp8e_set_roi_map},
- {VP8E_SET_ACTIVEMAP, vp8e_set_activemap},
- {VP8E_SET_SCALEMODE, vp8e_set_scalemode},
- {VP8E_SET_ENCODING_MODE, set_param},
- {VP8E_SET_CPUUSED, set_param},
- {VP8E_SET_NOISE_SENSITIVITY, set_param},
- {VP8E_SET_ENABLEAUTOALTREF, set_param},
- {VP8E_SET_SHARPNESS, set_param},
- {VP8E_SET_STATIC_THRESHOLD, set_param},
- {VP8E_SET_TOKEN_PARTITIONS, set_param},
- {VP8E_GET_LAST_QUANTIZER, get_param},
- {VP8E_GET_LAST_QUANTIZER_64, get_param},
- {VP8E_SET_ARNR_MAXFRAMES, set_param},
- {VP8E_SET_ARNR_STRENGTH , set_param},
- {VP8E_SET_ARNR_TYPE , set_param},
- {VP8E_SET_TUNING, set_param},
- {VP8E_SET_CQ_LEVEL, set_param},
- { -1, NULL},
-};
-
-static vpx_codec_enc_cfg_map_t vp8e_usage_cfg_map[] =
-{
- {
- 0,
- {
- 0, /* g_usage */
- 0, /* g_threads */
- 0, /* g_profile */
-
- 320, /* g_width */
- 240, /* g_height */
- {1, 30}, /* g_timebase */
-
- 0, /* g_error_resilient */
-
- VPX_RC_ONE_PASS, /* g_pass */
-
- 0, /* g_lag_in_frames */
-
- 0, /* rc_dropframe_thresh */
- 0, /* rc_resize_allowed */
- 60, /* rc_resize_down_thresold */
- 30, /* rc_resize_up_thresold */
-
- VPX_VBR, /* rc_end_usage */
-#if VPX_ENCODER_ABI_VERSION > (1 + VPX_CODEC_ABI_VERSION)
- {0}, /* rc_twopass_stats_in */
-#endif
- 256, /* rc_target_bandwidth */
-
- 4, /* rc_min_quantizer */
- 63, /* rc_max_quantizer */
- 95, /* rc_undershoot_pct */
- 200, /* rc_overshoot_pct */
-
- 6000, /* rc_max_buffer_size */
- 4000, /* rc_buffer_initial_size; */
- 5000, /* rc_buffer_optimal_size; */
-
- 50, /* rc_two_pass_vbrbias */
- 0, /* rc_two_pass_vbrmin_section */
- 400, /* rc_two_pass_vbrmax_section */
-
- /* keyframing settings (kf) */
- VPX_KF_AUTO, /* g_kfmode*/
- 0, /* kf_min_dist */
- 9999, /* kf_max_dist */
-
-#if VPX_ENCODER_ABI_VERSION == (1 + VPX_CODEC_ABI_VERSION)
- 1, /* g_delete_first_pass_file */
- "vp8.fpf" /* first pass filename */
-#endif
- }},
- { -1, {NOT_IMPLEMENTED}}
-};
-
-
-#ifndef VERSION_STRING
-#define VERSION_STRING
-#endif
-CODEC_INTERFACE(vpx_codec_vp8_cx) =
-{
- "WebM Project VP8 Encoder" VERSION_STRING,
- VPX_CODEC_INTERNAL_ABI_VERSION,
- VPX_CODEC_CAP_ENCODER | VPX_CODEC_CAP_PSNR,
- /* vpx_codec_caps_t caps; */
- vp8e_init, /* vpx_codec_init_fn_t init; */
- vp8e_destroy, /* vpx_codec_destroy_fn_t destroy; */
- vp8e_ctf_maps, /* vpx_codec_ctrl_fn_map_t *ctrl_maps; */
- NOT_IMPLEMENTED, /* vpx_codec_get_mmap_fn_t get_mmap; */
- NOT_IMPLEMENTED, /* vpx_codec_set_mmap_fn_t set_mmap; */
- {
- NOT_IMPLEMENTED, /* vpx_codec_peek_si_fn_t peek_si; */
- NOT_IMPLEMENTED, /* vpx_codec_get_si_fn_t get_si; */
- NOT_IMPLEMENTED, /* vpx_codec_decode_fn_t decode; */
- NOT_IMPLEMENTED, /* vpx_codec_frame_get_fn_t frame_get; */
- },
- {
- vp8e_usage_cfg_map, /* vpx_codec_enc_cfg_map_t peek_si; */
- vp8e_encode, /* vpx_codec_encode_fn_t encode; */
- vp8e_get_cxdata, /* vpx_codec_get_cx_data_fn_t frame_get; */
- vp8e_set_config,
- NOT_IMPLEMENTED,
- vp8e_get_preview,
- } /* encoder functions */
-};
-
-
-/*
- * BEGIN BACKWARDS COMPATIBILITY SHIM.
- */
-#define FORCE_KEY 2
-static vpx_codec_err_t api1_control(vpx_codec_alg_priv_t *ctx,
- int ctrl_id,
- va_list args)
-{
- vpx_codec_ctrl_fn_map_t *entry;
-
- switch (ctrl_id)
- {
- case VP8E_SET_FLUSHFLAG:
- /* VP8 sample code did VP8E_SET_FLUSHFLAG followed by
- * vpx_codec_get_cx_data() rather than vpx_codec_encode().
- */
- return vp8e_encode(ctx, NULL, 0, 0, 0, 0);
- case VP8E_SET_FRAMETYPE:
- ctx->base.enc.tbd |= FORCE_KEY;
- return VPX_CODEC_OK;
- }
-
- for (entry = vp8e_ctf_maps; entry && entry->fn; entry++)
- {
- if (!entry->ctrl_id || entry->ctrl_id == ctrl_id)
- {
- return entry->fn(ctx, ctrl_id, args);
- }
- }
-
- return VPX_CODEC_ERROR;
-}
-
-
-static vpx_codec_ctrl_fn_map_t api1_ctrl_maps[] =
-{
- {0, api1_control},
- { -1, NULL}
-};
-
-
-static vpx_codec_err_t api1_encode(vpx_codec_alg_priv_t *ctx,
- const vpx_image_t *img,
- vpx_codec_pts_t pts,
- unsigned long duration,
- vpx_enc_frame_flags_t flags,
- unsigned long deadline)
-{
- int force = ctx->base.enc.tbd;
-
- ctx->base.enc.tbd = 0;
- return vp8e_encode
- (ctx,
- img,
- pts,
- duration,
- flags | ((force & FORCE_KEY) ? VPX_EFLAG_FORCE_KF : 0),
- deadline);
-}
-
-
-vpx_codec_iface_t vpx_enc_vp8_algo =
-{
- "WebM Project VP8 Encoder (Deprecated API)" VERSION_STRING,
- VPX_CODEC_INTERNAL_ABI_VERSION,
- VPX_CODEC_CAP_ENCODER,
- /* vpx_codec_caps_t caps; */
- vp8e_init, /* vpx_codec_init_fn_t init; */
- vp8e_destroy, /* vpx_codec_destroy_fn_t destroy; */
- api1_ctrl_maps, /* vpx_codec_ctrl_fn_map_t *ctrl_maps; */
- NOT_IMPLEMENTED, /* vpx_codec_get_mmap_fn_t get_mmap; */
- NOT_IMPLEMENTED, /* vpx_codec_set_mmap_fn_t set_mmap; */
- {NOT_IMPLEMENTED}, /* decoder functions */
- {
- vp8e_usage_cfg_map, /* vpx_codec_enc_cfg_map_t peek_si; */
- api1_encode, /* vpx_codec_encode_fn_t encode; */
- vp8e_get_cxdata, /* vpx_codec_get_cx_data_fn_t frame_get; */
- vp8e_set_config,
- NOT_IMPLEMENTED,
- vp8e_get_preview,
- } /* encoder functions */
-};
diff --git a/vp8/vp8_dx_iface.c b/vp8/vp8_dx_iface.c
deleted file mode 100644
index 99657e0..0000000
--- a/vp8/vp8_dx_iface.c
+++ /dev/null
@@ -1,773 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdlib.h>
-#include <string.h>
-#include "vpx/vpx_decoder.h"
-#include "vpx/vp8dx.h"
-#include "vpx/internal/vpx_codec_internal.h"
-#include "vpx_version.h"
-#include "common/onyxd.h"
-#include "decoder/onyxd_int.h"
-
-#define VP8_CAP_POSTPROC (CONFIG_POSTPROC ? VPX_CODEC_CAP_POSTPROC : 0)
-
-typedef vpx_codec_stream_info_t vp8_stream_info_t;
-
-/* Structures for handling memory allocations */
-typedef enum
-{
- VP8_SEG_ALG_PRIV = 256,
- VP8_SEG_MAX
-} mem_seg_id_t;
-#define NELEMENTS(x) ((int)(sizeof(x)/sizeof(x[0])))
-
-static unsigned long vp8_priv_sz(const vpx_codec_dec_cfg_t *si, vpx_codec_flags_t);
-
-typedef struct
-{
- unsigned int id;
- unsigned long sz;
- unsigned int align;
- unsigned int flags;
- unsigned long(*calc_sz)(const vpx_codec_dec_cfg_t *, vpx_codec_flags_t);
-} mem_req_t;
-
-static const mem_req_t vp8_mem_req_segs[] =
-{
- {VP8_SEG_ALG_PRIV, 0, 8, VPX_CODEC_MEM_ZERO, vp8_priv_sz},
- {VP8_SEG_MAX, 0, 0, 0, NULL}
-};
-
-struct vpx_codec_alg_priv
-{
- vpx_codec_priv_t base;
- vpx_codec_mmap_t mmaps[NELEMENTS(vp8_mem_req_segs)-1];
- vpx_codec_dec_cfg_t cfg;
- vp8_stream_info_t si;
- int defer_alloc;
- int decoder_init;
- VP8D_PTR pbi;
- int postproc_cfg_set;
- vp8_postproc_cfg_t postproc_cfg;
-#if CONFIG_POSTPROC_VISUALIZER
- unsigned int dbg_postproc_flag;
- int dbg_color_ref_frame_flag;
- int dbg_color_mb_modes_flag;
- int dbg_color_b_modes_flag;
- int dbg_display_mv_flag;
-#endif
- vpx_image_t img;
- int img_setup;
- int img_avail;
-};
-
-static unsigned long vp8_priv_sz(const vpx_codec_dec_cfg_t *si, vpx_codec_flags_t flags)
-{
- /* Although this declaration is constant, we can't use it in the requested
- * segments list because we want to define the requested segments list
- * before defining the private type (so that the number of memory maps is
- * known)
- */
- (void)si;
- return sizeof(vpx_codec_alg_priv_t);
-}
-
-
-static void vp8_mmap_dtor(vpx_codec_mmap_t *mmap)
-{
- free(mmap->priv);
-}
-
-static vpx_codec_err_t vp8_mmap_alloc(vpx_codec_mmap_t *mmap)
-{
- vpx_codec_err_t res;
- unsigned int align;
-
- align = mmap->align ? mmap->align - 1 : 0;
-
- if (mmap->flags & VPX_CODEC_MEM_ZERO)
- mmap->priv = calloc(1, mmap->sz + align);
- else
- mmap->priv = malloc(mmap->sz + align);
-
- res = (mmap->priv) ? VPX_CODEC_OK : VPX_CODEC_MEM_ERROR;
- mmap->base = (void *)((((uintptr_t)mmap->priv) + align) & ~(uintptr_t)align);
- mmap->dtor = vp8_mmap_dtor;
- return res;
-}
-
-static vpx_codec_err_t vp8_validate_mmaps(const vp8_stream_info_t *si,
- const vpx_codec_mmap_t *mmaps,
- vpx_codec_flags_t init_flags)
-{
- int i;
- vpx_codec_err_t res = VPX_CODEC_OK;
-
- for (i = 0; i < NELEMENTS(vp8_mem_req_segs) - 1; i++)
- {
- /* Ensure the segment has been allocated */
- if (!mmaps[i].base)
- {
- res = VPX_CODEC_MEM_ERROR;
- break;
- }
-
- /* Verify variable size segment is big enough for the current si. */
- if (vp8_mem_req_segs[i].calc_sz)
- {
- vpx_codec_dec_cfg_t cfg;
-
- cfg.w = si->w;
- cfg.h = si->h;
-
- if (mmaps[i].sz < vp8_mem_req_segs[i].calc_sz(&cfg, init_flags))
- {
- res = VPX_CODEC_MEM_ERROR;
- break;
- }
- }
- }
-
- return res;
-}
-
-static void vp8_init_ctx(vpx_codec_ctx_t *ctx, const vpx_codec_mmap_t *mmap)
-{
- int i;
-
- ctx->priv = mmap->base;
- ctx->priv->sz = sizeof(*ctx->priv);
- ctx->priv->iface = ctx->iface;
- ctx->priv->alg_priv = mmap->base;
-
- for (i = 0; i < NELEMENTS(ctx->priv->alg_priv->mmaps); i++)
- ctx->priv->alg_priv->mmaps[i].id = vp8_mem_req_segs[i].id;
-
- ctx->priv->alg_priv->mmaps[0] = *mmap;
- ctx->priv->alg_priv->si.sz = sizeof(ctx->priv->alg_priv->si);
- ctx->priv->init_flags = ctx->init_flags;
-
- if (ctx->config.dec)
- {
- /* Update the reference to the config structure to an internal copy. */
- ctx->priv->alg_priv->cfg = *ctx->config.dec;
- ctx->config.dec = &ctx->priv->alg_priv->cfg;
- }
-}
-
-static void *mmap_lkup(vpx_codec_alg_priv_t *ctx, unsigned int id)
-{
- int i;
-
- for (i = 0; i < NELEMENTS(ctx->mmaps); i++)
- if (ctx->mmaps[i].id == id)
- return ctx->mmaps[i].base;
-
- return NULL;
-}
-static void vp8_finalize_mmaps(vpx_codec_alg_priv_t *ctx)
-{
- /* nothing to clean up */
-}
-
-static vpx_codec_err_t vp8_init(vpx_codec_ctx_t *ctx)
-{
- vpx_codec_err_t res = VPX_CODEC_OK;
-
- /* This function only allocates space for the vpx_codec_alg_priv_t
- * structure. More memory may be required at the time the stream
- * information becomes known.
- */
- if (!ctx->priv)
- {
- vpx_codec_mmap_t mmap;
-
- mmap.id = vp8_mem_req_segs[0].id;
- mmap.sz = sizeof(vpx_codec_alg_priv_t);
- mmap.align = vp8_mem_req_segs[0].align;
- mmap.flags = vp8_mem_req_segs[0].flags;
-
- res = vp8_mmap_alloc(&mmap);
-
- if (!res)
- {
- vp8_init_ctx(ctx, &mmap);
-
- ctx->priv->alg_priv->defer_alloc = 1;
- /*post processing level initialized to do nothing */
- }
- }
-
- return res;
-}
-
-static vpx_codec_err_t vp8_destroy(vpx_codec_alg_priv_t *ctx)
-{
- int i;
-
- vp8dx_remove_decompressor(ctx->pbi);
-
- for (i = NELEMENTS(ctx->mmaps) - 1; i >= 0; i--)
- {
- if (ctx->mmaps[i].dtor)
- ctx->mmaps[i].dtor(&ctx->mmaps[i]);
- }
-
- return VPX_CODEC_OK;
-}
-
-static vpx_codec_err_t vp8_peek_si(const uint8_t *data,
- unsigned int data_sz,
- vpx_codec_stream_info_t *si)
-{
- vpx_codec_err_t res = VPX_CODEC_OK;
-
- if(data + data_sz <= data)
- res = VPX_CODEC_INVALID_PARAM;
- else
- {
- /* Parse uncompresssed part of key frame header.
- * 3 bytes:- including version, frame type and an offset
- * 3 bytes:- sync code (0x9d, 0x01, 0x2a)
- * 4 bytes:- including image width and height in the lowest 14 bits
- * of each 2-byte value.
- */
- si->is_kf = 0;
-
- if (data_sz >= 10 && !(data[0] & 0x01)) /* I-Frame */
- {
- const uint8_t *c = data + 3;
- si->is_kf = 1;
-
- /* vet via sync code */
- if (c[0] != 0x9d || c[1] != 0x01 || c[2] != 0x2a)
- res = VPX_CODEC_UNSUP_BITSTREAM;
-
- si->w = (c[3] | (c[4] << 8)) & 0x3fff;
- si->h = (c[5] | (c[6] << 8)) & 0x3fff;
-
- /*printf("w=%d, h=%d\n", si->w, si->h);*/
- if (!(si->h | si->w))
- res = VPX_CODEC_UNSUP_BITSTREAM;
- }
- else
- res = VPX_CODEC_UNSUP_BITSTREAM;
- }
-
- return res;
-
-}
-
-static vpx_codec_err_t vp8_get_si(vpx_codec_alg_priv_t *ctx,
- vpx_codec_stream_info_t *si)
-{
-
- unsigned int sz;
-
- if (si->sz >= sizeof(vp8_stream_info_t))
- sz = sizeof(vp8_stream_info_t);
- else
- sz = sizeof(vpx_codec_stream_info_t);
-
- memcpy(si, &ctx->si, sz);
- si->sz = sz;
-
- return VPX_CODEC_OK;
-}
-
-
-static vpx_codec_err_t
-update_error_state(vpx_codec_alg_priv_t *ctx,
- const struct vpx_internal_error_info *error)
-{
- vpx_codec_err_t res;
-
- if ((res = error->error_code))
- ctx->base.err_detail = error->has_detail
- ? error->detail
- : NULL;
-
- return res;
-}
-
-
-static vpx_codec_err_t vp8_decode(vpx_codec_alg_priv_t *ctx,
- const uint8_t *data,
- unsigned int data_sz,
- void *user_priv,
- long deadline)
-{
- vpx_codec_err_t res = VPX_CODEC_OK;
-
- ctx->img_avail = 0;
-
- /* Determine the stream parameters. Note that we rely on peek_si to
- * validate that we have a buffer that does not wrap around the top
- * of the heap.
- */
- if (!ctx->si.h)
- res = ctx->base.iface->dec.peek_si(data, data_sz, &ctx->si);
-
-
- /* Perform deferred allocations, if required */
- if (!res && ctx->defer_alloc)
- {
- int i;
-
- for (i = 1; !res && i < NELEMENTS(ctx->mmaps); i++)
- {
- vpx_codec_dec_cfg_t cfg;
-
- cfg.w = ctx->si.w;
- cfg.h = ctx->si.h;
- ctx->mmaps[i].id = vp8_mem_req_segs[i].id;
- ctx->mmaps[i].sz = vp8_mem_req_segs[i].sz;
- ctx->mmaps[i].align = vp8_mem_req_segs[i].align;
- ctx->mmaps[i].flags = vp8_mem_req_segs[i].flags;
-
- if (!ctx->mmaps[i].sz)
- ctx->mmaps[i].sz = vp8_mem_req_segs[i].calc_sz(&cfg,
- ctx->base.init_flags);
-
- res = vp8_mmap_alloc(&ctx->mmaps[i]);
- }
-
- if (!res)
- vp8_finalize_mmaps(ctx);
-
- ctx->defer_alloc = 0;
- }
-
- /* Initialize the decoder instance on the first frame*/
- if (!res && !ctx->decoder_init)
- {
- res = vp8_validate_mmaps(&ctx->si, ctx->mmaps, ctx->base.init_flags);
-
- if (!res)
- {
- VP8D_CONFIG oxcf;
- VP8D_PTR optr;
-
- vp8dx_initialize();
-
- oxcf.Width = ctx->si.w;
- oxcf.Height = ctx->si.h;
- oxcf.Version = 9;
- oxcf.postprocess = 0;
- oxcf.max_threads = ctx->cfg.threads;
-
- optr = vp8dx_create_decompressor(&oxcf);
-
- /* If postprocessing was enabled by the application and a
- * configuration has not been provided, default it.
- */
- if (!ctx->postproc_cfg_set
- && (ctx->base.init_flags & VPX_CODEC_USE_POSTPROC))
- {
- ctx->postproc_cfg.post_proc_flag =
- VP8_DEBLOCK | VP8_DEMACROBLOCK;
- ctx->postproc_cfg.deblocking_level = 4;
- ctx->postproc_cfg.noise_level = 0;
- }
-
- if (!optr)
- res = VPX_CODEC_ERROR;
- else
- ctx->pbi = optr;
- }
-
- ctx->decoder_init = 1;
- }
-
- if (!res && ctx->pbi)
- {
- YV12_BUFFER_CONFIG sd;
- INT64 time_stamp = 0, time_end_stamp = 0;
- vp8_ppflags_t flags = {0};
-
- if (ctx->base.init_flags & VPX_CODEC_USE_POSTPROC)
- {
- flags.post_proc_flag= ctx->postproc_cfg.post_proc_flag
-#if CONFIG_POSTPROC_VISUALIZER
-
- | ((ctx->dbg_color_ref_frame_flag != 0) ? VP8D_DEBUG_CLR_FRM_REF_BLKS : 0)
- | ((ctx->dbg_color_mb_modes_flag != 0) ? VP8D_DEBUG_CLR_BLK_MODES : 0)
- | ((ctx->dbg_color_b_modes_flag != 0) ? VP8D_DEBUG_CLR_BLK_MODES : 0)
- | ((ctx->dbg_display_mv_flag != 0) ? VP8D_DEBUG_DRAW_MV : 0)
-#endif
- ;
- flags.deblocking_level = ctx->postproc_cfg.deblocking_level;
- flags.noise_level = ctx->postproc_cfg.noise_level;
-#if CONFIG_POSTPROC_VISUALIZER
- flags.display_ref_frame_flag= ctx->dbg_color_ref_frame_flag;
- flags.display_mb_modes_flag = ctx->dbg_color_mb_modes_flag;
- flags.display_b_modes_flag = ctx->dbg_color_b_modes_flag;
- flags.display_mv_flag = ctx->dbg_display_mv_flag;
-#endif
- }
-
- if (vp8dx_receive_compressed_data(ctx->pbi, data_sz, data, deadline))
- {
- VP8D_COMP *pbi = (VP8D_COMP *)ctx->pbi;
- res = update_error_state(ctx, &pbi->common.error);
- }
-
- if (!res && 0 == vp8dx_get_raw_frame(ctx->pbi, &sd, &time_stamp, &time_end_stamp, &flags))
- {
- /* Align width/height */
- unsigned int a_w = (sd.y_width + 15) & ~15;
- unsigned int a_h = (sd.y_height + 15) & ~15;
-
- vpx_img_wrap(&ctx->img, VPX_IMG_FMT_I420,
- a_w + 2 * VP8BORDERINPIXELS,
- a_h + 2 * VP8BORDERINPIXELS,
- 1,
- sd.buffer_alloc);
- vpx_img_set_rect(&ctx->img,
- VP8BORDERINPIXELS, VP8BORDERINPIXELS,
- sd.y_width, sd.y_height);
- ctx->img.user_priv = user_priv;
- ctx->img_avail = 1;
-
- }
- }
-
- return res;
-}
-
-static vpx_image_t *vp8_get_frame(vpx_codec_alg_priv_t *ctx,
- vpx_codec_iter_t *iter)
-{
- vpx_image_t *img = NULL;
-
- if (ctx->img_avail)
- {
- /* iter acts as a flip flop, so an image is only returned on the first
- * call to get_frame.
- */
- if (!(*iter))
- {
- img = &ctx->img;
- *iter = img;
- }
- }
-
- return img;
-}
-
-
-static
-vpx_codec_err_t vp8_xma_get_mmap(const vpx_codec_ctx_t *ctx,
- vpx_codec_mmap_t *mmap,
- vpx_codec_iter_t *iter)
-{
- vpx_codec_err_t res;
- const mem_req_t *seg_iter = *iter;
-
- /* Get address of next segment request */
- do
- {
- if (!seg_iter)
- seg_iter = vp8_mem_req_segs;
- else if (seg_iter->id != VP8_SEG_MAX)
- seg_iter++;
-
- *iter = (vpx_codec_iter_t)seg_iter;
-
- if (seg_iter->id != VP8_SEG_MAX)
- {
- mmap->id = seg_iter->id;
- mmap->sz = seg_iter->sz;
- mmap->align = seg_iter->align;
- mmap->flags = seg_iter->flags;
-
- if (!seg_iter->sz)
- mmap->sz = seg_iter->calc_sz(ctx->config.dec, ctx->init_flags);
-
- res = VPX_CODEC_OK;
- }
- else
- res = VPX_CODEC_LIST_END;
- }
- while (!mmap->sz && res != VPX_CODEC_LIST_END);
-
- return res;
-}
-
-static vpx_codec_err_t vp8_xma_set_mmap(vpx_codec_ctx_t *ctx,
- const vpx_codec_mmap_t *mmap)
-{
- vpx_codec_err_t res = VPX_CODEC_MEM_ERROR;
- int i, done;
-
- if (!ctx->priv)
- {
- if (mmap->id == VP8_SEG_ALG_PRIV)
- {
- if (!ctx->priv)
- {
- vp8_init_ctx(ctx, mmap);
- res = VPX_CODEC_OK;
- }
- }
- }
-
- done = 1;
-
- if (!res && ctx->priv->alg_priv)
- {
- for (i = 0; i < NELEMENTS(ctx->priv->alg_priv->mmaps); i++)
- {
- if (ctx->priv->alg_priv->mmaps[i].id == mmap->id)
- if (!ctx->priv->alg_priv->mmaps[i].base)
- {
- ctx->priv->alg_priv->mmaps[i] = *mmap;
- res = VPX_CODEC_OK;
- }
-
- done &= (ctx->priv->alg_priv->mmaps[i].base != NULL);
- }
- }
-
- if (done && !res)
- {
- vp8_finalize_mmaps(ctx->priv->alg_priv);
- res = ctx->iface->init(ctx);
- }
-
- return res;
-}
-
-static vpx_codec_err_t image2yuvconfig(const vpx_image_t *img,
- YV12_BUFFER_CONFIG *yv12)
-{
- vpx_codec_err_t res = VPX_CODEC_OK;
- yv12->y_buffer = img->planes[VPX_PLANE_Y];
- yv12->u_buffer = img->planes[VPX_PLANE_U];
- yv12->v_buffer = img->planes[VPX_PLANE_V];
-
- yv12->y_width = img->d_w;
- yv12->y_height = img->d_h;
- yv12->uv_width = yv12->y_width / 2;
- yv12->uv_height = yv12->y_height / 2;
-
- yv12->y_stride = img->stride[VPX_PLANE_Y];
- yv12->uv_stride = img->stride[VPX_PLANE_U];
-
- yv12->border = (img->stride[VPX_PLANE_Y] - img->d_w) / 2;
- yv12->clrtype = (img->fmt == VPX_IMG_FMT_VPXI420 || img->fmt == VPX_IMG_FMT_VPXYV12);
-
- return res;
-}
-
-
-static vpx_codec_err_t vp8_set_reference(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
-
- vpx_ref_frame_t *data = va_arg(args, vpx_ref_frame_t *);
-
- if (data)
- {
- vpx_ref_frame_t *frame = (vpx_ref_frame_t *)data;
- YV12_BUFFER_CONFIG sd;
-
- image2yuvconfig(&frame->img, &sd);
-
- vp8dx_set_reference(ctx->pbi, frame->frame_type, &sd);
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-
-}
-
-static vpx_codec_err_t vp8_get_reference(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
-
- vpx_ref_frame_t *data = va_arg(args, vpx_ref_frame_t *);
-
- if (data)
- {
- vpx_ref_frame_t *frame = (vpx_ref_frame_t *)data;
- YV12_BUFFER_CONFIG sd;
-
- image2yuvconfig(&frame->img, &sd);
-
- vp8dx_get_reference(ctx->pbi, frame->frame_type, &sd);
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-
-}
-
-static vpx_codec_err_t vp8_set_postproc(vpx_codec_alg_priv_t *ctx,
- int ctr_id,
- va_list args)
-{
- vp8_postproc_cfg_t *data = va_arg(args, vp8_postproc_cfg_t *);
-#if CONFIG_POSTPROC
-
- if (data)
- {
- ctx->postproc_cfg_set = 1;
- ctx->postproc_cfg = *((vp8_postproc_cfg_t *)data);
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-
-#else
- return VPX_CODEC_INCAPABLE;
-#endif
-}
-
-static vpx_codec_err_t vp8_set_dbg_options(vpx_codec_alg_priv_t *ctx,
- int ctrl_id,
- va_list args)
-{
-#if CONFIG_POSTPROC_VISUALIZER && CONFIG_POSTPROC
- int data = va_arg(args, int);
-
-#define MAP(id, var) case id: var = data; break;
-
- switch (ctrl_id)
- {
- MAP (VP8_SET_DBG_COLOR_REF_FRAME, ctx->dbg_color_ref_frame_flag);
- MAP (VP8_SET_DBG_COLOR_MB_MODES, ctx->dbg_color_mb_modes_flag);
- MAP (VP8_SET_DBG_COLOR_B_MODES, ctx->dbg_color_b_modes_flag);
- MAP (VP8_SET_DBG_DISPLAY_MV, ctx->dbg_display_mv_flag);
- }
-
- return VPX_CODEC_OK;
-#else
- return VPX_CODEC_INCAPABLE;
-#endif
-}
-
-static vpx_codec_err_t vp8_get_last_ref_updates(vpx_codec_alg_priv_t *ctx,
- int ctrl_id,
- va_list args)
-{
- int *update_info = va_arg(args, int *);
- VP8D_COMP *pbi = (VP8D_COMP *)ctx->pbi;
-
- if (update_info)
- {
- *update_info = pbi->common.refresh_alt_ref_frame * (int) VP8_ALTR_FRAME
- + pbi->common.refresh_golden_frame * (int) VP8_GOLD_FRAME
- + pbi->common.refresh_last_frame * (int) VP8_LAST_FRAME;
-
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-}
-
-
-static vpx_codec_err_t vp8_get_frame_corrupted(vpx_codec_alg_priv_t *ctx,
- int ctrl_id,
- va_list args)
-{
-
- int *corrupted = va_arg(args, int *);
-
- if (corrupted)
- {
- VP8D_COMP *pbi = (VP8D_COMP *)ctx->pbi;
- *corrupted = pbi->common.frame_to_show->corrupted;
-
- return VPX_CODEC_OK;
- }
- else
- return VPX_CODEC_INVALID_PARAM;
-
-}
-
-vpx_codec_ctrl_fn_map_t vp8_ctf_maps[] =
-{
- {VP8_SET_REFERENCE, vp8_set_reference},
- {VP8_COPY_REFERENCE, vp8_get_reference},
- {VP8_SET_POSTPROC, vp8_set_postproc},
- {VP8_SET_DBG_COLOR_REF_FRAME, vp8_set_dbg_options},
- {VP8_SET_DBG_COLOR_MB_MODES, vp8_set_dbg_options},
- {VP8_SET_DBG_COLOR_B_MODES, vp8_set_dbg_options},
- {VP8_SET_DBG_DISPLAY_MV, vp8_set_dbg_options},
- {VP8D_GET_LAST_REF_UPDATES, vp8_get_last_ref_updates},
- {VP8D_GET_FRAME_CORRUPTED, vp8_get_frame_corrupted},
- { -1, NULL},
-};
-
-
-#ifndef VERSION_STRING
-#define VERSION_STRING
-#endif
-CODEC_INTERFACE(vpx_codec_vp8_dx) =
-{
- "WebM Project VP8 Decoder" VERSION_STRING,
- VPX_CODEC_INTERNAL_ABI_VERSION,
- VPX_CODEC_CAP_DECODER | VP8_CAP_POSTPROC,
- /* vpx_codec_caps_t caps; */
- vp8_init, /* vpx_codec_init_fn_t init; */
- vp8_destroy, /* vpx_codec_destroy_fn_t destroy; */
- vp8_ctf_maps, /* vpx_codec_ctrl_fn_map_t *ctrl_maps; */
- vp8_xma_get_mmap, /* vpx_codec_get_mmap_fn_t get_mmap; */
- vp8_xma_set_mmap, /* vpx_codec_set_mmap_fn_t set_mmap; */
- {
- vp8_peek_si, /* vpx_codec_peek_si_fn_t peek_si; */
- vp8_get_si, /* vpx_codec_get_si_fn_t get_si; */
- vp8_decode, /* vpx_codec_decode_fn_t decode; */
- vp8_get_frame, /* vpx_codec_frame_get_fn_t frame_get; */
- },
- { /* encoder functions */
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED
- }
-};
-
-/*
- * BEGIN BACKWARDS COMPATIBILITY SHIM.
- */
-vpx_codec_iface_t vpx_codec_vp8_algo =
-{
- "WebM Project VP8 Decoder (Deprecated API)" VERSION_STRING,
- VPX_CODEC_INTERNAL_ABI_VERSION,
- VPX_CODEC_CAP_DECODER | VP8_CAP_POSTPROC,
- /* vpx_codec_caps_t caps; */
- vp8_init, /* vpx_codec_init_fn_t init; */
- vp8_destroy, /* vpx_codec_destroy_fn_t destroy; */
- vp8_ctf_maps, /* vpx_codec_ctrl_fn_map_t *ctrl_maps; */
- vp8_xma_get_mmap, /* vpx_codec_get_mmap_fn_t get_mmap; */
- vp8_xma_set_mmap, /* vpx_codec_set_mmap_fn_t set_mmap; */
- {
- vp8_peek_si, /* vpx_codec_peek_si_fn_t peek_si; */
- vp8_get_si, /* vpx_codec_get_si_fn_t get_si; */
- vp8_decode, /* vpx_codec_decode_fn_t decode; */
- vp8_get_frame, /* vpx_codec_frame_get_fn_t frame_get; */
- },
- { /* encoder functions */
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED,
- NOT_IMPLEMENTED
- }
-};
diff --git a/vp8/vp8cx.mk b/vp8/vp8cx.mk
deleted file mode 100644
index 8037f9a..0000000
--- a/vp8/vp8cx.mk
+++ /dev/null
@@ -1,127 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8_common.mk
-
-VP8_CX_EXPORTS += exports_enc
-
-VP8_CX_SRCS-yes += $(VP8_COMMON_SRCS-yes)
-VP8_CX_SRCS-no += $(VP8_COMMON_SRCS-no)
-VP8_CX_SRCS_REMOVE-yes += $(VP8_COMMON_SRCS_REMOVE-yes)
-VP8_CX_SRCS_REMOVE-no += $(VP8_COMMON_SRCS_REMOVE-no)
-
-ifeq ($(ARCH_ARM),yes)
- include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8cx_arm.mk
-endif
-
-VP8_CX_SRCS-yes += vp8_cx_iface.c
-
-# encoder
-#INCLUDES += algo/vpx_common/vpx_mem/include
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += algo/vpx_ref/cpu_id/include
-#INCLUDES += common
-#INCLUDES += encoder
-
-VP8_CX_SRCS-yes += encoder/asm_enc_offsets.c
-VP8_CX_SRCS-yes += encoder/bitstream.c
-VP8_CX_SRCS-yes += encoder/boolhuff.c
-VP8_CX_SRCS-yes += encoder/dct.c
-VP8_CX_SRCS-yes += encoder/encodeframe.c
-VP8_CX_SRCS-yes += encoder/encodeintra.c
-VP8_CX_SRCS-yes += encoder/encodemb.c
-VP8_CX_SRCS-yes += encoder/encodemv.c
-VP8_CX_SRCS-$(CONFIG_MULTITHREAD) += encoder/ethreading.c
-VP8_CX_SRCS-yes += encoder/firstpass.c
-VP8_CX_SRCS-yes += encoder/generic/csystemdependent.c
-VP8_CX_SRCS-yes += encoder/block.h
-VP8_CX_SRCS-yes += encoder/boolhuff.h
-VP8_CX_SRCS-yes += encoder/bitstream.h
-VP8_CX_SRCS-yes += encoder/dct.h
-VP8_CX_SRCS-yes += encoder/encodeintra.h
-VP8_CX_SRCS-yes += encoder/encodemb.h
-VP8_CX_SRCS-yes += encoder/encodemv.h
-VP8_CX_SRCS-yes += encoder/firstpass.h
-VP8_CX_SRCS-yes += encoder/mcomp.h
-VP8_CX_SRCS-yes += encoder/modecosts.h
-VP8_CX_SRCS-yes += encoder/onyx_int.h
-VP8_CX_SRCS-yes += encoder/pickinter.h
-VP8_CX_SRCS-yes += encoder/psnr.h
-VP8_CX_SRCS-yes += encoder/quantize.h
-VP8_CX_SRCS-yes += encoder/ratectrl.h
-VP8_CX_SRCS-yes += encoder/rdopt.h
-VP8_CX_SRCS-yes += encoder/tokenize.h
-VP8_CX_SRCS-yes += encoder/treewriter.h
-VP8_CX_SRCS-yes += encoder/variance.h
-VP8_CX_SRCS-yes += encoder/mcomp.c
-VP8_CX_SRCS-yes += encoder/modecosts.c
-VP8_CX_SRCS-yes += encoder/onyx_if.c
-VP8_CX_SRCS-yes += encoder/pickinter.c
-VP8_CX_SRCS-yes += encoder/picklpf.c
-VP8_CX_SRCS-yes += encoder/psnr.c
-VP8_CX_SRCS-yes += encoder/quantize.c
-VP8_CX_SRCS-yes += encoder/ratectrl.c
-VP8_CX_SRCS-yes += encoder/rdopt.c
-VP8_CX_SRCS-yes += encoder/sad_c.c
-VP8_CX_SRCS-yes += encoder/segmentation.c
-VP8_CX_SRCS-yes += encoder/segmentation.h
-VP8_CX_SRCS-$(CONFIG_PSNR) += encoder/ssim.c
-VP8_CX_SRCS-yes += encoder/tokenize.c
-VP8_CX_SRCS-yes += encoder/treewriter.c
-VP8_CX_SRCS-yes += encoder/variance_c.c
-VP8_CX_SRCS-$(CONFIG_PSNR) += common/postproc.h
-VP8_CX_SRCS-$(CONFIG_PSNR) += common/postproc.c
-VP8_CX_SRCS-yes += encoder/temporal_filter.c
-VP8_CX_SRCS-yes += encoder/temporal_filter.h
-
-ifeq ($(CONFIG_REALTIME_ONLY),yes)
-VP8_CX_SRCS_REMOVE-yes += encoder/firstpass.c
-VP8_CX_SRCS_REMOVE-yes += encoder/temporal_filter.c
-endif
-
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/encodemb_x86.h
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/dct_x86.h
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/mcomp_x86.h
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/variance_x86.h
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/quantize_x86.h
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/temporal_filter_x86.h
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/x86_csystemdependent.c
-VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/variance_mmx.c
-VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/variance_impl_mmx.asm
-VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/sad_mmx.asm
-VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/dct_mmx.asm
-VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/subtract_mmx.asm
-VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/dct_sse2.asm
-VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/variance_sse2.c
-VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/variance_impl_sse2.asm
-VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/sad_sse2.asm
-VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/fwalsh_sse2.asm
-VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/quantize_sse2.asm
-VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/subtract_sse2.asm
-VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/temporal_filter_apply_sse2.asm
-VP8_CX_SRCS-$(HAVE_SSE3) += encoder/x86/sad_sse3.asm
-VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/sad_ssse3.asm
-VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/variance_ssse3.c
-VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/variance_impl_ssse3.asm
-VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/quantize_ssse3.asm
-VP8_CX_SRCS-$(HAVE_SSE4_1) += encoder/x86/sad_sse4.asm
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/quantize_mmx.asm
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/encodeopt.asm
-VP8_CX_SRCS-$(ARCH_X86_64) += encoder/x86/ssim_opt.asm
-
-ifeq ($(CONFIG_REALTIME_ONLY),yes)
-VP8_CX_SRCS_REMOVE-$(HAVE_SSE2) += encoder/x86/temporal_filter_apply_sse2.asm
-endif
-
-
-VP8_CX_SRCS-yes := $(filter-out $(VP8_CX_SRCS_REMOVE-yes),$(VP8_CX_SRCS-yes))
diff --git a/vp8/vp8cx_arm.mk b/vp8/vp8cx_arm.mk
deleted file mode 100644
index 165dada..0000000
--- a/vp8/vp8cx_arm.mk
+++ /dev/null
@@ -1,63 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-#VP8_CX_SRCS list is modified according to different platforms.
-
-#File list for arm
-# encoder
-VP8_CX_SRCS-$(ARCH_ARM) += encoder/arm/arm_csystemdependent.c
-
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/quantize_arm.c
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/picklpf_arm.c
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/dct_arm.c
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/variance_arm.c
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/variance_arm.h
-VP8_CX_SRCS-$(HAVE_ARMV5TE) += encoder/arm/boolhuff_arm.c
-
-VP8_CX_SRCS_REMOVE-$(HAVE_ARMV5TE) += encoder/boolhuff.c
-
-#File list for armv5te
-# encoder
-VP8_CX_SRCS-$(HAVE_ARMV5TE) += encoder/arm/armv5te/boolhuff_armv5te$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV5TE) += encoder/arm/armv5te/vp8_packtokens_armv5$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV5TE) += encoder/arm/armv5te/vp8_packtokens_mbrow_armv5$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV5TE) += encoder/arm/armv5te/vp8_packtokens_partitions_armv5$(ASM)
-
-#File list for armv6
-# encoder
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_subtract_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_fast_fdct4x4_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_fast_quantize_b_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_sad16x16_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_variance16x16_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_mse16x16_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/vp8_variance8x8_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV6) += encoder/arm/armv6/walsh_v6$(ASM)
-
-#File list for neon
-# encoder
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/fastfdct4x4_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/fastfdct8x4_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/fastquantizeb_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/sad8_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/sad16_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/shortfdct_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/subtract_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/variance_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/vp8_mse16x16_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/vp8_subpixelvariance8x8_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/vp8_subpixelvariance16x16_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/vp8_subpixelvariance16x16s_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/vp8_memcpy_neon$(ASM)
-VP8_CX_SRCS-$(HAVE_ARMV7) += encoder/arm/neon/vp8_shortwalsh4x4_neon$(ASM)
diff --git a/vp8/vp8dx.mk b/vp8/vp8dx.mk
deleted file mode 100644
index 5649671..0000000
--- a/vp8/vp8dx.mk
+++ /dev/null
@@ -1,76 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8_common.mk
-
-VP8_DX_EXPORTS += exports_dec
-
-VP8_DX_SRCS-yes += $(VP8_COMMON_SRCS-yes)
-VP8_DX_SRCS-no += $(VP8_COMMON_SRCS-no)
-VP8_DX_SRCS_REMOVE-yes += $(VP8_COMMON_SRCS_REMOVE-yes)
-VP8_DX_SRCS_REMOVE-no += $(VP8_COMMON_SRCS_REMOVE-no)
-
-ifeq ($(ARCH_ARM),yes)
- include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8dx_arm.mk
-endif
-
-VP8_DX_SRCS-yes += vp8_dx_iface.c
-
-# common
-#define ARM
-#define DISABLE_THREAD
-
-#INCLUDES += algo/vpx_common/vpx_mem/include
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += decoder
-
-
-
-# decoder
-#define ARM
-#define DISABLE_THREAD
-
-#INCLUDES += algo/vpx_common/vpx_mem/include
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += decoder
-
-VP8_DX_SRCS-yes += decoder/dboolhuff.c
-VP8_DX_SRCS-yes += decoder/decodemv.c
-VP8_DX_SRCS-yes += decoder/decodframe.c
-VP8_DX_SRCS-yes += decoder/dequantize.c
-VP8_DX_SRCS-yes += decoder/detokenize.c
-VP8_DX_SRCS-yes += decoder/generic/dsystemdependent.c
-VP8_DX_SRCS-yes += decoder/dboolhuff.h
-VP8_DX_SRCS-yes += decoder/decodemv.h
-VP8_DX_SRCS-yes += decoder/decoderthreading.h
-VP8_DX_SRCS-yes += decoder/dequantize.h
-VP8_DX_SRCS-yes += decoder/detokenize.h
-VP8_DX_SRCS-yes += decoder/onyxd_int.h
-VP8_DX_SRCS-yes += decoder/treereader.h
-VP8_DX_SRCS-yes += decoder/onyxd_if.c
-VP8_DX_SRCS-$(CONFIG_MULTITHREAD) += decoder/threading.c
-VP8_DX_SRCS-yes += decoder/idct_blk.c
-VP8_DX_SRCS-$(CONFIG_MULTITHREAD) += decoder/reconintra_mt.h
-VP8_DX_SRCS-$(CONFIG_MULTITHREAD) += decoder/reconintra_mt.c
-
-VP8_DX_SRCS-yes := $(filter-out $(VP8_DX_SRCS_REMOVE-yes),$(VP8_DX_SRCS-yes))
-
-VP8_DX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += decoder/x86/dequantize_x86.h
-VP8_DX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += decoder/x86/x86_dsystemdependent.c
-VP8_DX_SRCS-$(HAVE_MMX) += decoder/x86/dequantize_mmx.asm
-VP8_DX_SRCS-$(HAVE_MMX) += decoder/x86/idct_blk_mmx.c
-VP8_DX_SRCS-$(HAVE_SSE2) += decoder/x86/idct_blk_sse2.c
diff --git a/vp8/vp8dx_arm.mk b/vp8/vp8dx_arm.mk
deleted file mode 100644
index 03084c5..0000000
--- a/vp8/vp8dx_arm.mk
+++ /dev/null
@@ -1,32 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-#VP8_DX_SRCS list is modified according to different platforms.
-
-VP8_DX_SRCS-$(ARCH_ARM) += decoder/arm/arm_dsystemdependent.c
-VP8_CX_SRCS-$(ARCH_ARM) += decoder/asm_dec_offsets.c
-
-VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/dequantize_arm.c
-
-#File list for armv6
-VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequant_dc_idct_v6$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequant_idct_v6$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequantize_v6$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/idct_blk_v6.c
-
-#File list for neon
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_dequant_dc_full_2x_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_dequant_dc_0_2x_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequant_idct_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_dequant_full_2x_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_dequant_0_2x_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequantizeb_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_blk_neon.c
diff --git a/vp8_api1_migration.txt b/vp8_api1_migration.txt
deleted file mode 100644
index 47b7981..0000000
--- a/vp8_api1_migration.txt
+++ /dev/null
@@ -1,198 +0,0 @@
-Version 2.x of this library has deprecated or removed a number of interfaces to
-the VP8 codec. Where possible, the old interfaces have been left in place in a
-deprecated state, and will generate compiler warnings when they are referenced.
-All users are encouraged to update their code to the new interfaces as soon as
-possible. To assist in this effort, the `VPX_CODEC_DISABLE_COMPAT` symbol can
-be #defined to 1 prior to including vpx headers. This will disable the
-backwards compatability workarounds and ensure that you are using only the
-latest API.
-
-The *TWO-PASS STATISTICS* sections detail the one section of code which is not
-backwards compatable and will require code changes.
-
-
-HEADER FILES
-============
-The following header files were renamed:
-
- vp8.h -> vp8dx.h
- vp8e.h -> vp8cx.h
-
-
-INTERFACE SYMBOLS
-=================
-The following interface symbols were renamed:
-
- vpx_codec_vp8_algo -> vpx_codec_vp8_dx_algo
- vpx_enc_vp8_algo -> vpx_codec_vp8_cx_algo
-
-
-TWO-PASS STATISTICS
-===================
-Two-pass statistics are handled significantly differently. The version 1 API
-stored statistics in a file, and the application passed the name of that file
-in the `vpx_codec_enc_cfg` structure. In this version, statistics are returned
-though the application though the `vpx_codec_get_cx_data()` interface. The
-application must concatenate these packets into a contiguous buffer and then
-pass that buffer to the encoder through the `vpx_codec_enc_cfg` structure on
-the second pass initialization. The application may choose to keep these packets
-in memory or write them to disk. Statistics packets are approximately 112 bytes
-per frame. See the example code for more detailed examples.
-
-
-ENCODER CONTROLS
-================
-
-Renames
--------
-The following controls are duplicated between the encoder and the decoder, but
-the encoder unnecessarily introduced unique identifiers for them. These
-identifiers were removed in favor of the ones used by the decoder:
-
- VP8E_SET_REFERENCE -> VP8_SET_REFERENCE
- VP8E_COPY_REFERENCE -> VP8_COPY_REFERENCE
- VP8E_SET_PREVIEWPP -> VP8_SET_POSTPROC
-
-
-VP8E_SET_FRAMETYPE
-------------------
-This control was removed in favor of the `flags` parameter to
-`vpx_codec_encode()`. Existing code such as:
-
-~~~
- vpx_codec_control(&encoder, VP8E_SET_FRAMETYPE, KEY_FRAME);
- ...
- vpx_codec_encode(&encoder, img, pts, 1, 0, 0);
-~~~
-
-becomes:
-
-~~~
- vpx_codec_encode(&encoder, img, pts, 1, VPX_EFLAG_FORCE_KF,
- VPX_DL_REALTIME);
-~~~
-
-
-
-VP8E_SET_FLUSHFLAG
-------------------
-Flush is handled by passing `NULL` to the `img` parameter of
-`vpx_codec_encode()`. You must do this at least once, regardless of your encoder
-configuration. i.e. it's not specific to g_lag_in_frames. This control was
-removed.
-
-~~~
- while(...) {
- ...
- vpx_codec_encode(&encoder, img, pts, 1, 0, 0);
- while( (pkt = vpx_codec_get_cx_data(&encoder, &iter)) ) {
- ...
- }
- }
- vpx_codec_control(&encoder, VP8E_SET_FLUSHFLAG, 1);
- while( (pkt = vpx_codec_get_cx_data(&encoder, &iter)) ) {
- ...
- }
- vpx_codec_encode(&encoder, img, pts, 1, 0, 0);
-~~~
-
-becomes
-
-~~~
- while(new_image && ...) {
- ...
- vpx_codec_encode(&encoder, new_image?img:NULL, pts, 1, 0, 0);
- while( (pkt = vpx_codec_get_cx_data(&encoder, &iter)) ) {
- ...
- }
- }
-~~~
-
-
-
-VP8E_SET_ENCODING_MODE
-----------------------
-This control was removed in favor of the `deadline` parameter to
-`vpx_codec_encode()`. There are three macros that can be used to get the
-equivalent behavior: VPX_DL_REALTIME, VPX_DL_GOOD_QUALITY,
-VPX_DL_BEST_QUALITY. Existing code such as:
-
-~~~
- vpx_codec_control(&encoder, VP8E_SET_ENCODING_MODE, VP8_REAL_TIME_ENCODING);
- ...
- vpx_codec_encode(&encoder, img, pts, 1, 0, 0);
-~~~
-
-becomes:
-
-~~~
- vpx_codec_encode(&encoder, img, pts, 1, 0, VPX_DL_REALTIME);
-~~~
-
-
-VP8E_UPD_ENTROPY
-------------------
-This control was deprecated in favor of the `flags` parameter to
-`vpx_codec_encode()`. Existing code such as:
-
-~~~
- vpx_codec_control(&encoder, VP8E_UPD_ENTROPY, 0);
-~~~
-
-becomes:
-
-~~~
- vpx_codec_encode(&encoder, img, pts, 1, VP8_EFLAG_NO_UPD_ENTROPY,
- VPX_DL_REALTIME);
-~~~
-
-
-VP8E_UPD_REFERENCE
-------------------
-This control was deprecated in favor of the `flags` parameter to
-`vpx_codec_encode()`. A set bit on the VP8E_UPD_REFERENCE bitfield is
-analogous to setting the VP8_EFLAG_FORCE_* flag. A cleared bit is analogous
-to setting the VP8_EFLAG_NO_UPD_* flag. If neither the FORCE or NO_UPD bit
-is set, the encoder will make its decision automatically, as usual. Setting
-both bits will result in an error being returned. Existing code such as:
-
-~~~
- vpx_codec_control(&encoder, VP8E_UPD_REFERENCE,
- VP8_LAST_FRAME | VP8_GOLD_FRAME);
- vpx_codec_control(&encoder, VP8E_UPD_REFERENCE, 0);
- ...
- vpx_codec_encode(&encoder, img, pts, 1, 0, VPX_DL_REALTIME);
-~~~
-
-becomes:
-
-~~~
- vpx_codec_encode(&encoder, img, pts, 1, VP8_EFLAG_FORCE_GF,
- VPX_DL_REALTIME);
- vpx_codec_encode(&encoder, img, pts, 1, VP8_EFLAG_NO_UPD_LAST
- | VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF,
- VPX_DL_REALTIME);
-~~~
-
-
-VP8E_USE_REFERENCE
-------------------
-This control was deprecated in favor of the `flags` parameter to
-`vpx_codec_encode()`. A cleared bit on the VP8E_USE_REFERENCE bitfield is
-analogous to setting the VP8_EFLAG_NO_REF* flag. A set bit indicates that
-the encoder will make its decision automatically, as usual.
-Existing code such as:
-
-~~~
- vpx_codec_control(&encoder, VP8E_USE_REFERENCE,
- VP8_ALTR_FRAME | VP8_GOLD_FRAME);
- ...
- vpx_codec_encode(&encoder, img, pts, 1, 0, VPX_DL_REALTIME);
-~~~
-
-becomes
-
-~~~
- vpx_codec_encode(&encoder, img, pts, 1, VP8_EFLAG_NO_REF_LAST,
- VPX_DL_REALTIME);
-~~~
diff --git a/vpx/internal/vpx_codec_internal.h b/vpx/internal/vpx_codec_internal.h
deleted file mode 100644
index a1ff192..0000000
--- a/vpx/internal/vpx_codec_internal.h
+++ /dev/null
@@ -1,471 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\file
- * \brief Describes the decoder algorithm interface for algorithm
- * implementations.
- *
- * This file defines the private structures and data types that are only
- * relevant to implementing an algorithm, as opposed to using it.
- *
- * To create a decoder algorithm class, an interface structure is put
- * into the global namespace:
- * <pre>
- * my_codec.c:
- * vpx_codec_iface_t my_codec = {
- * "My Codec v1.0",
- * VPX_CODEC_ALG_ABI_VERSION,
- * ...
- * };
- * </pre>
- *
- * An application instantiates a specific decoder instance by using
- * vpx_codec_init() and a pointer to the algorithm's interface structure:
- * <pre>
- * my_app.c:
- * extern vpx_codec_iface_t my_codec;
- * {
- * vpx_codec_ctx_t algo;
- * res = vpx_codec_init(&algo, &my_codec);
- * }
- * </pre>
- *
- * Once initialized, the instance is manged using other functions from
- * the vpx_codec_* family.
- */
-#ifndef VPX_CODEC_INTERNAL_H
-#define VPX_CODEC_INTERNAL_H
-#include "../vpx_decoder.h"
-#include "../vpx_encoder.h"
-#include <stdarg.h>
-
-
-/*!\brief Current ABI version number
- *
- * \internal
- * If this file is altered in any way that changes the ABI, this value
- * must be bumped. Examples include, but are not limited to, changing
- * types, removing or reassigning enums, adding/removing/rearranging
- * fields to structures
- */
-#define VPX_CODEC_INTERNAL_ABI_VERSION (3) /**<\hideinitializer*/
-
-typedef struct vpx_codec_alg_priv vpx_codec_alg_priv_t;
-
-/*!\brief init function pointer prototype
- *
- * Performs algorithm-specific initialization of the decoder context. This
- * function is called by the generic vpx_codec_init() wrapper function, so
- * plugins implementing this interface may trust the input parameters to be
- * properly initialized.
- *
- * \param[in] ctx Pointer to this instance's context
- * \retval #VPX_CODEC_OK
- * The input stream was recognized and decoder initialized.
- * \retval #VPX_CODEC_MEM_ERROR
- * Memory operation failed.
- */
-typedef vpx_codec_err_t (*vpx_codec_init_fn_t)(vpx_codec_ctx_t *ctx);
-
-/*!\brief destroy function pointer prototype
- *
- * Performs algorithm-specific destruction of the decoder context. This
- * function is called by the generic vpx_codec_destroy() wrapper function,
- * so plugins implementing this interface may trust the input parameters
- * to be properly initialized.
- *
- * \param[in] ctx Pointer to this instance's context
- * \retval #VPX_CODEC_OK
- * The input stream was recognized and decoder initialized.
- * \retval #VPX_CODEC_MEM_ERROR
- * Memory operation failed.
- */
-typedef vpx_codec_err_t (*vpx_codec_destroy_fn_t)(vpx_codec_alg_priv_t *ctx);
-
-/*!\brief parse stream info function pointer prototype
- *
- * Performs high level parsing of the bitstream. This function is called by
- * the generic vpx_codec_parse_stream() wrapper function, so plugins implementing
- * this interface may trust the input parameters to be properly initialized.
- *
- * \param[in] data Pointer to a block of data to parse
- * \param[in] data_sz Size of the data buffer
- * \param[in,out] si Pointer to stream info to update. The size member
- * \ref MUST be properly initialized, but \ref MAY be
- * clobbered by the algorithm. This parameter \ref MAY
- * be NULL.
- *
- * \retval #VPX_CODEC_OK
- * Bitstream is parsable and stream information updated
- */
-typedef vpx_codec_err_t (*vpx_codec_peek_si_fn_t)(const uint8_t *data,
- unsigned int data_sz,
- vpx_codec_stream_info_t *si);
-
-/*!\brief Return information about the current stream.
- *
- * Returns information about the stream that has been parsed during decoding.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in,out] si Pointer to stream info to update. The size member
- * \ref MUST be properly initialized, but \ref MAY be
- * clobbered by the algorithm. This parameter \ref MAY
- * be NULL.
- *
- * \retval #VPX_CODEC_OK
- * Bitstream is parsable and stream information updated
- */
-typedef vpx_codec_err_t (*vpx_codec_get_si_fn_t)(vpx_codec_alg_priv_t *ctx,
- vpx_codec_stream_info_t *si);
-
-/*!\brief control function pointer prototype
- *
- * This function is used to exchange algorithm specific data with the decoder
- * instance. This can be used to implement features specific to a particular
- * algorithm.
- *
- * This function is called by the generic vpx_codec_control() wrapper
- * function, so plugins implementing this interface may trust the input
- * parameters to be properly initialized. However, this interface does not
- * provide type safety for the exchanged data or assign meanings to the
- * control codes. Those details should be specified in the algorithm's
- * header file. In particular, the ctrl_id parameter is guaranteed to exist
- * in the algorithm's control mapping table, and the data parameter may be NULL.
- *
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] ctrl_id Algorithm specific control identifier
- * \param[in,out] data Data to exchange with algorithm instance.
- *
- * \retval #VPX_CODEC_OK
- * The internal state data was deserialized.
- */
-typedef vpx_codec_err_t (*vpx_codec_control_fn_t)(vpx_codec_alg_priv_t *ctx,
- int ctrl_id,
- va_list ap);
-
-/*!\brief control function pointer mapping
- *
- * This structure stores the mapping between control identifiers and
- * implementing functions. Each algorithm provides a list of these
- * mappings. This list is searched by the vpx_codec_control() wrapper
- * function to determine which function to invoke. The special
- * value {0, NULL} is used to indicate end-of-list, and must be
- * present. The special value {0, <non-null>} can be used as a catch-all
- * mapping. This implies that ctrl_id values chosen by the algorithm
- * \ref MUST be non-zero.
- */
-typedef const struct
-{
- int ctrl_id;
- vpx_codec_control_fn_t fn;
-} vpx_codec_ctrl_fn_map_t;
-
-/*!\brief decode data function pointer prototype
- *
- * Processes a buffer of coded data. If the processing results in a new
- * decoded frame becoming available, #VPX_CODEC_CB_PUT_SLICE and
- * #VPX_CODEC_CB_PUT_FRAME events are generated as appropriate. This
- * function is called by the generic vpx_codec_decode() wrapper function,
- * so plugins implementing this interface may trust the input parameters
- * to be properly initialized.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] data Pointer to this block of new coded data. If
- * NULL, a #VPX_CODEC_CB_PUT_FRAME event is posted
- * for the previously decoded frame.
- * \param[in] data_sz Size of the coded data, in bytes.
- *
- * \return Returns #VPX_CODEC_OK if the coded data was processed completely
- * and future pictures can be decoded without error. Otherwise,
- * see the descriptions of the other error codes in ::vpx_codec_err_t
- * for recoverability capabilities.
- */
-typedef vpx_codec_err_t (*vpx_codec_decode_fn_t)(vpx_codec_alg_priv_t *ctx,
- const uint8_t *data,
- unsigned int data_sz,
- void *user_priv,
- long deadline);
-
-/*!\brief Decoded frames iterator
- *
- * Iterates over a list of the frames available for display. The iterator
- * storage should be initialized to NULL to start the iteration. Iteration is
- * complete when this function returns NULL.
- *
- * The list of available frames becomes valid upon completion of the
- * vpx_codec_decode call, and remains valid until the next call to vpx_codec_decode.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in out] iter Iterator storage, initialized to NULL
- *
- * \return Returns a pointer to an image, if one is ready for display. Frames
- * produced will always be in PTS (presentation time stamp) order.
- */
-typedef vpx_image_t*(*vpx_codec_get_frame_fn_t)(vpx_codec_alg_priv_t *ctx,
- vpx_codec_iter_t *iter);
-
-
-/*\brief eXternal Memory Allocation memory map get iterator
- *
- * Iterates over a list of the memory maps requested by the decoder. The
- * iterator storage should be initialized to NULL to start the iteration.
- * Iteration is complete when this function returns NULL.
- *
- * \param[in out] iter Iterator storage, initialized to NULL
- *
- * \return Returns a pointer to an memory segment descriptor, or NULL to
- * indicate end-of-list.
- */
-typedef vpx_codec_err_t (*vpx_codec_get_mmap_fn_t)(const vpx_codec_ctx_t *ctx,
- vpx_codec_mmap_t *mmap,
- vpx_codec_iter_t *iter);
-
-
-/*\brief eXternal Memory Allocation memory map set iterator
- *
- * Sets a memory descriptor inside the decoder instance.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] mmap Memory map to store.
- *
- * \retval #VPX_CODEC_OK
- * The memory map was accepted and stored.
- * \retval #VPX_CODEC_MEM_ERROR
- * The memory map was rejected.
- */
-typedef vpx_codec_err_t (*vpx_codec_set_mmap_fn_t)(vpx_codec_ctx_t *ctx,
- const vpx_codec_mmap_t *mmap);
-
-
-typedef vpx_codec_err_t (*vpx_codec_encode_fn_t)(vpx_codec_alg_priv_t *ctx,
- const vpx_image_t *img,
- vpx_codec_pts_t pts,
- unsigned long duration,
- vpx_enc_frame_flags_t flags,
- unsigned long deadline);
-typedef const vpx_codec_cx_pkt_t*(*vpx_codec_get_cx_data_fn_t)(vpx_codec_alg_priv_t *ctx,
- vpx_codec_iter_t *iter);
-
-typedef vpx_codec_err_t
-(*vpx_codec_enc_config_set_fn_t)(vpx_codec_alg_priv_t *ctx,
- const vpx_codec_enc_cfg_t *cfg);
-typedef vpx_fixed_buf_t *
-(*vpx_codec_get_global_headers_fn_t)(vpx_codec_alg_priv_t *ctx);
-
-typedef vpx_image_t *
-(*vpx_codec_get_preview_frame_fn_t)(vpx_codec_alg_priv_t *ctx);
-
-/*!\brief usage configuration mapping
- *
- * This structure stores the mapping between usage identifiers and
- * configuration structures. Each algorithm provides a list of these
- * mappings. This list is searched by the vpx_codec_enc_config_default()
- * wrapper function to determine which config to return. The special value
- * {-1, {0}} is used to indicate end-of-list, and must be present. At least
- * one mapping must be present, in addition to the end-of-list.
- *
- */
-typedef const struct
-{
- int usage;
- vpx_codec_enc_cfg_t cfg;
-} vpx_codec_enc_cfg_map_t;
-
-#define NOT_IMPLEMENTED 0
-
-/*!\brief Decoder algorithm interface interface
- *
- * All decoders \ref MUST expose a variable of this type.
- */
-struct vpx_codec_iface
-{
- const char *name; /**< Identification String */
- int abi_version; /**< Implemented ABI version */
- vpx_codec_caps_t caps; /**< Decoder capabilities */
- vpx_codec_init_fn_t init; /**< \copydoc ::vpx_codec_init_fn_t */
- vpx_codec_destroy_fn_t destroy; /**< \copydoc ::vpx_codec_destroy_fn_t */
- vpx_codec_ctrl_fn_map_t *ctrl_maps; /**< \copydoc ::vpx_codec_ctrl_fn_map_t */
- vpx_codec_get_mmap_fn_t get_mmap; /**< \copydoc ::vpx_codec_get_mmap_fn_t */
- vpx_codec_set_mmap_fn_t set_mmap; /**< \copydoc ::vpx_codec_set_mmap_fn_t */
- struct
- {
- vpx_codec_peek_si_fn_t peek_si; /**< \copydoc ::vpx_codec_peek_si_fn_t */
- vpx_codec_get_si_fn_t get_si; /**< \copydoc ::vpx_codec_peek_si_fn_t */
- vpx_codec_decode_fn_t decode; /**< \copydoc ::vpx_codec_decode_fn_t */
- vpx_codec_get_frame_fn_t get_frame; /**< \copydoc ::vpx_codec_get_frame_fn_t */
- } dec;
- struct
- {
- vpx_codec_enc_cfg_map_t *cfg_maps; /**< \copydoc ::vpx_codec_enc_cfg_map_t */
- vpx_codec_encode_fn_t encode; /**< \copydoc ::vpx_codec_encode_fn_t */
- vpx_codec_get_cx_data_fn_t get_cx_data; /**< \copydoc ::vpx_codec_get_cx_data_fn_t */
- vpx_codec_enc_config_set_fn_t cfg_set; /**< \copydoc ::vpx_codec_enc_config_set_fn_t */
- vpx_codec_get_global_headers_fn_t get_glob_hdrs; /**< \copydoc ::vpx_codec_enc_config_set_fn_t */
- vpx_codec_get_preview_frame_fn_t get_preview; /**< \copydoc ::vpx_codec_get_preview_frame_fn_t */
- } enc;
-};
-
-/*!\brief Callback function pointer / user data pair storage */
-typedef struct vpx_codec_priv_cb_pair
-{
- union
- {
- vpx_codec_put_frame_cb_fn_t put_frame;
- vpx_codec_put_slice_cb_fn_t put_slice;
- } u;
- void *user_priv;
-} vpx_codec_priv_cb_pair_t;
-
-
-/*!\brief Instance private storage
- *
- * This structure is allocated by the algorithm's init function. It can be
- * extended in one of two ways. First, a second, algorithm specific structure
- * can be allocated and the priv member pointed to it. Alternatively, this
- * structure can be made the first member of the algorithm specific structure,
- * and the pointer cast to the proper type.
- */
-struct vpx_codec_priv
-{
- unsigned int sz;
- vpx_codec_iface_t *iface;
- struct vpx_codec_alg_priv *alg_priv;
- const char *err_detail;
- vpx_codec_flags_t init_flags;
- struct
- {
- vpx_codec_priv_cb_pair_t put_frame_cb;
- vpx_codec_priv_cb_pair_t put_slice_cb;
- } dec;
- struct
- {
- int tbd;
- struct vpx_fixed_buf cx_data_dst_buf;
- unsigned int cx_data_pad_before;
- unsigned int cx_data_pad_after;
- vpx_codec_cx_pkt_t cx_data_pkt;
- } enc;
-};
-
-#undef VPX_CTRL_USE_TYPE
-#define VPX_CTRL_USE_TYPE(id, typ) \
- static typ id##__value(va_list args) {return va_arg(args, typ);} \
- static typ id##__convert(void *x)\
- {\
- union\
- {\
- void *x;\
- typ d;\
- } u;\
- u.x = x;\
- return u.d;\
- }
-
-
-#undef VPX_CTRL_USE_TYPE_DEPRECATED
-#define VPX_CTRL_USE_TYPE_DEPRECATED(id, typ) \
- static typ id##__value(va_list args) {return va_arg(args, typ);} \
- static typ id##__convert(void *x)\
- {\
- union\
- {\
- void *x;\
- typ d;\
- } u;\
- u.x = x;\
- return u.d;\
- }
-
-#define CAST(id, arg) id##__value(arg)
-#define RECAST(id, x) id##__convert(x)
-
-
-/* CODEC_INTERFACE convenience macro
- *
- * By convention, each codec interface is a struct with extern linkage, where
- * the symbol is suffixed with _algo. A getter function is also defined to
- * return a pointer to the struct, since in some cases it's easier to work
- * with text symbols than data symbols (see issue #169). This function has
- * the same name as the struct, less the _algo suffix. The CODEC_INTERFACE
- * macro is provided to define this getter function automatically.
- */
-#define CODEC_INTERFACE(id)\
-vpx_codec_iface_t* id(void) { return &id##_algo; }\
-vpx_codec_iface_t id##_algo
-
-
-/* Internal Utility Functions
- *
- * The following functions are intended to be used inside algorithms as
- * utilities for manipulating vpx_codec_* data structures.
- */
-struct vpx_codec_pkt_list
-{
- unsigned int cnt;
- unsigned int max;
- struct vpx_codec_cx_pkt pkts[1];
-};
-
-#define vpx_codec_pkt_list_decl(n)\
- union {struct vpx_codec_pkt_list head;\
- struct {struct vpx_codec_pkt_list head;\
- struct vpx_codec_cx_pkt pkts[n];} alloc;}
-
-#define vpx_codec_pkt_list_init(m)\
- (m)->alloc.head.cnt = 0,\
- (m)->alloc.head.max = sizeof((m)->alloc.pkts) / sizeof((m)->alloc.pkts[0])
-
-int
-vpx_codec_pkt_list_add(struct vpx_codec_pkt_list *,
- const struct vpx_codec_cx_pkt *);
-
-const vpx_codec_cx_pkt_t*
-vpx_codec_pkt_list_get(struct vpx_codec_pkt_list *list,
- vpx_codec_iter_t *iter);
-
-
-#include <stdio.h>
-#include <setjmp.h>
-struct vpx_internal_error_info
-{
- vpx_codec_err_t error_code;
- int has_detail;
- char detail[80];
- int setjmp;
- jmp_buf jmp;
-};
-
-static void vpx_internal_error(struct vpx_internal_error_info *info,
- vpx_codec_err_t error,
- const char *fmt,
- ...)
-{
- va_list ap;
-
- info->error_code = error;
- info->has_detail = 0;
-
- if (fmt)
- {
- size_t sz = sizeof(info->detail);
-
- info->has_detail = 1;
- va_start(ap, fmt);
- vsnprintf(info->detail, sz - 1, fmt, ap);
- va_end(ap);
- info->detail[sz-1] = '\0';
- }
-
- if (info->setjmp)
- longjmp(info->jmp, info->error_code);
-}
-#endif
diff --git a/vpx/src/vpx_decoder.c b/vpx/src/vpx_decoder.c
deleted file mode 100644
index e0be55a..0000000
--- a/vpx/src/vpx_decoder.c
+++ /dev/null
@@ -1,247 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\file
- * \brief Provides the high level interface to wrap decoder algorithms.
- *
- */
-#include <string.h>
-#include "vpx/internal/vpx_codec_internal.h"
-
-#define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var)
-
-vpx_codec_err_t vpx_codec_dec_init_ver(vpx_codec_ctx_t *ctx,
- vpx_codec_iface_t *iface,
- vpx_codec_dec_cfg_t *cfg,
- vpx_codec_flags_t flags,
- int ver)
-{
- vpx_codec_err_t res;
-
-#ifdef MIPS_DSP_REV
-#if (MIPS_DSP_REV>=2)
- dsputil_static_init();
-#endif
-#endif
-
- if (ver != VPX_DECODER_ABI_VERSION)
- res = VPX_CODEC_ABI_MISMATCH;
- else if (!ctx || !iface)
- res = VPX_CODEC_INVALID_PARAM;
- else if (iface->abi_version != VPX_CODEC_INTERNAL_ABI_VERSION)
- res = VPX_CODEC_ABI_MISMATCH;
- else if ((flags & VPX_CODEC_USE_XMA) && !(iface->caps & VPX_CODEC_CAP_XMA))
- res = VPX_CODEC_INCAPABLE;
- else if ((flags & VPX_CODEC_USE_POSTPROC) && !(iface->caps & VPX_CODEC_CAP_POSTPROC))
- res = VPX_CODEC_INCAPABLE;
- else if (!(iface->caps & VPX_CODEC_CAP_DECODER))
- res = VPX_CODEC_INCAPABLE;
- else
- {
- memset(ctx, 0, sizeof(*ctx));
- ctx->iface = iface;
- ctx->name = iface->name;
- ctx->priv = NULL;
- ctx->init_flags = flags;
- ctx->config.dec = cfg;
- res = VPX_CODEC_OK;
-
- if (!(flags & VPX_CODEC_USE_XMA))
- {
- res = ctx->iface->init(ctx);
-
- if (res)
- {
- ctx->err_detail = ctx->priv ? ctx->priv->err_detail : NULL;
- vpx_codec_destroy(ctx);
- }
-
- if (ctx->priv)
- ctx->priv->iface = ctx->iface;
- }
- }
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-vpx_codec_err_t vpx_codec_peek_stream_info(vpx_codec_iface_t *iface,
- const uint8_t *data,
- unsigned int data_sz,
- vpx_codec_stream_info_t *si)
-{
- vpx_codec_err_t res;
-
- if (!iface || !data || !data_sz || !si
- || si->sz < sizeof(vpx_codec_stream_info_t))
- res = VPX_CODEC_INVALID_PARAM;
- else
- {
- /* Set default/unknown values */
- si->w = 0;
- si->h = 0;
-
- res = iface->dec.peek_si(data, data_sz, si);
- }
-
- return res;
-}
-
-
-vpx_codec_err_t vpx_codec_get_stream_info(vpx_codec_ctx_t *ctx,
- vpx_codec_stream_info_t *si)
-{
- vpx_codec_err_t res;
-
- if (!ctx || !si || si->sz < sizeof(vpx_codec_stream_info_t))
- res = VPX_CODEC_INVALID_PARAM;
- else if (!ctx->iface || !ctx->priv)
- res = VPX_CODEC_ERROR;
- else
- {
- /* Set default/unknown values */
- si->w = 0;
- si->h = 0;
-
- res = ctx->iface->dec.get_si(ctx->priv->alg_priv, si);
- }
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-vpx_codec_err_t vpx_codec_decode(vpx_codec_ctx_t *ctx,
- const uint8_t *data,
- unsigned int data_sz,
- void *user_priv,
- long deadline)
-{
- vpx_codec_err_t res;
-
- /* Sanity checks */
- /* NULL data ptr allowed if data_sz is 0 too */
- if (!ctx || (!data && data_sz))
- res = VPX_CODEC_INVALID_PARAM;
- else if (!ctx->iface || !ctx->priv)
- res = VPX_CODEC_ERROR;
- else
- {
- res = ctx->iface->dec.decode(ctx->priv->alg_priv, data, data_sz,
- user_priv, deadline);
- }
-
- return SAVE_STATUS(ctx, res);
-}
-
-vpx_image_t *vpx_codec_get_frame(vpx_codec_ctx_t *ctx,
- vpx_codec_iter_t *iter)
-{
- vpx_image_t *img;
-
- if (!ctx || !iter || !ctx->iface || !ctx->priv)
- img = NULL;
- else
- img = ctx->iface->dec.get_frame(ctx->priv->alg_priv, iter);
-
- return img;
-}
-
-
-vpx_codec_err_t vpx_codec_register_put_frame_cb(vpx_codec_ctx_t *ctx,
- vpx_codec_put_frame_cb_fn_t cb,
- void *user_priv)
-{
- vpx_codec_err_t res;
-
- if (!ctx || !cb)
- res = VPX_CODEC_INVALID_PARAM;
- else if (!ctx->iface || !ctx->priv
- || !(ctx->iface->caps & VPX_CODEC_CAP_PUT_FRAME))
- res = VPX_CODEC_ERROR;
- else
- {
- ctx->priv->dec.put_frame_cb.u.put_frame = cb;
- ctx->priv->dec.put_frame_cb.user_priv = user_priv;
- res = VPX_CODEC_OK;
- }
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-vpx_codec_err_t vpx_codec_register_put_slice_cb(vpx_codec_ctx_t *ctx,
- vpx_codec_put_slice_cb_fn_t cb,
- void *user_priv)
-{
- vpx_codec_err_t res;
-
- if (!ctx || !cb)
- res = VPX_CODEC_INVALID_PARAM;
- else if (!ctx->iface || !ctx->priv
- || !(ctx->iface->caps & VPX_CODEC_CAP_PUT_FRAME))
- res = VPX_CODEC_ERROR;
- else
- {
- ctx->priv->dec.put_slice_cb.u.put_slice = cb;
- ctx->priv->dec.put_slice_cb.user_priv = user_priv;
- res = VPX_CODEC_OK;
- }
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-vpx_codec_err_t vpx_codec_get_mem_map(vpx_codec_ctx_t *ctx,
- vpx_codec_mmap_t *mmap,
- vpx_codec_iter_t *iter)
-{
- vpx_codec_err_t res = VPX_CODEC_OK;
-
- if (!ctx || !mmap || !iter || !ctx->iface)
- res = VPX_CODEC_INVALID_PARAM;
- else if (!(ctx->iface->caps & VPX_CODEC_CAP_XMA))
- res = VPX_CODEC_ERROR;
- else
- res = ctx->iface->get_mmap(ctx, mmap, iter);
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-vpx_codec_err_t vpx_codec_set_mem_map(vpx_codec_ctx_t *ctx,
- vpx_codec_mmap_t *mmap,
- unsigned int num_maps)
-{
- vpx_codec_err_t res = VPX_CODEC_MEM_ERROR;
-
- if (!ctx || !mmap || !ctx->iface)
- res = VPX_CODEC_INVALID_PARAM;
- else if (!(ctx->iface->caps & VPX_CODEC_CAP_XMA))
- res = VPX_CODEC_ERROR;
- else
- {
- unsigned int i;
-
- for (i = 0; i < num_maps; i++, mmap++)
- {
- if (!mmap->base)
- break;
-
- /* Everything look ok, set the mmap in the decoder */
- res = ctx->iface->set_mmap(ctx, mmap);
-
- if (res)
- break;
- }
- }
-
- return SAVE_STATUS(ctx, res);
-}
diff --git a/vpx/src/vpx_decoder_compat.c b/vpx/src/vpx_decoder_compat.c
deleted file mode 100644
index 4fe00ce..0000000
--- a/vpx/src/vpx_decoder_compat.c
+++ /dev/null
@@ -1,201 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\file
- * \brief Provides the high level interface to wrap decoder algorithms.
- *
- */
-#include <stdlib.h>
-#include <string.h>
-#include "vpx/vpx_decoder.h"
-#include "vpx/internal/vpx_codec_internal.h"
-
-#define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var)
-
-const char *vpx_dec_iface_name(vpx_dec_iface_t *iface)
-{
- return vpx_codec_iface_name((vpx_codec_iface_t *)iface);
-}
-
-const char *vpx_dec_err_to_string(vpx_dec_err_t err)
-{
- return vpx_codec_err_to_string(err);
-}
-
-const char *vpx_dec_error(vpx_dec_ctx_t *ctx)
-{
- return vpx_codec_error((vpx_codec_ctx_t *)ctx);
-}
-
-const char *vpx_dec_error_detail(vpx_dec_ctx_t *ctx)
-{
- return vpx_codec_error_detail((vpx_codec_ctx_t *)ctx);
-}
-
-
-vpx_dec_err_t vpx_dec_init_ver(vpx_dec_ctx_t *ctx,
- vpx_dec_iface_t *iface,
- int ver)
-{
- return vpx_codec_dec_init_ver((vpx_codec_ctx_t *)ctx,
- (vpx_codec_iface_t *)iface,
- NULL,
- 0,
- ver);
-}
-
-
-vpx_dec_err_t vpx_dec_destroy(vpx_dec_ctx_t *ctx)
-{
- return vpx_codec_destroy((vpx_codec_ctx_t *)ctx);
-}
-
-
-vpx_dec_caps_t vpx_dec_get_caps(vpx_dec_iface_t *iface)
-{
- return vpx_codec_get_caps((vpx_codec_iface_t *)iface);
-}
-
-
-vpx_dec_err_t vpx_dec_peek_stream_info(vpx_dec_iface_t *iface,
- const uint8_t *data,
- unsigned int data_sz,
- vpx_dec_stream_info_t *si)
-{
- return vpx_codec_peek_stream_info((vpx_codec_iface_t *)iface, data, data_sz,
- (vpx_codec_stream_info_t *)si);
-}
-
-
-vpx_dec_err_t vpx_dec_get_stream_info(vpx_dec_ctx_t *ctx,
- vpx_dec_stream_info_t *si)
-{
- return vpx_codec_get_stream_info((vpx_codec_ctx_t *)ctx,
- (vpx_codec_stream_info_t *)si);
-}
-
-
-vpx_dec_err_t vpx_dec_control(vpx_dec_ctx_t *ctx,
- int ctrl_id,
- void *data)
-{
- return vpx_codec_control_((vpx_codec_ctx_t *)ctx, ctrl_id, data);
-}
-
-
-vpx_dec_err_t vpx_dec_decode(vpx_dec_ctx_t *ctx,
- uint8_t *data,
- unsigned int data_sz,
- void *user_priv,
- int rel_pts)
-{
- (void)rel_pts;
- return vpx_codec_decode((vpx_codec_ctx_t *)ctx, data, data_sz, user_priv,
- 0);
-}
-
-vpx_image_t *vpx_dec_get_frame(vpx_dec_ctx_t *ctx,
- vpx_dec_iter_t *iter)
-{
- return vpx_codec_get_frame((vpx_codec_ctx_t *)ctx, iter);
-}
-
-
-vpx_dec_err_t vpx_dec_register_put_frame_cb(vpx_dec_ctx_t *ctx,
- vpx_dec_put_frame_cb_fn_t cb,
- void *user_priv)
-{
- return vpx_codec_register_put_frame_cb((vpx_codec_ctx_t *)ctx, cb,
- user_priv);
-}
-
-
-vpx_dec_err_t vpx_dec_register_put_slice_cb(vpx_dec_ctx_t *ctx,
- vpx_dec_put_slice_cb_fn_t cb,
- void *user_priv)
-{
- return vpx_codec_register_put_slice_cb((vpx_codec_ctx_t *)ctx, cb,
- user_priv);
-}
-
-
-vpx_dec_err_t vpx_dec_xma_init_ver(vpx_dec_ctx_t *ctx,
- vpx_dec_iface_t *iface,
- int ver)
-{
- return vpx_codec_dec_init_ver((vpx_codec_ctx_t *)ctx,
- (vpx_codec_iface_t *)iface,
- NULL,
- VPX_CODEC_USE_XMA,
- ver);
-}
-
-vpx_dec_err_t vpx_dec_get_mem_map(vpx_dec_ctx_t *ctx_,
- vpx_dec_mmap_t *mmap,
- const vpx_dec_stream_info_t *si,
- vpx_dec_iter_t *iter)
-{
- vpx_codec_ctx_t *ctx = (vpx_codec_ctx_t *)ctx_;
- vpx_dec_err_t res = VPX_DEC_OK;
-
- if (!ctx || !mmap || !si || !iter || !ctx->iface)
- res = VPX_DEC_INVALID_PARAM;
- else if (!(ctx->iface->caps & VPX_DEC_CAP_XMA))
- res = VPX_DEC_ERROR;
- else
- {
- if (!ctx->config.dec)
- {
- ctx->config.dec = malloc(sizeof(vpx_codec_dec_cfg_t));
- ctx->config.dec->w = si->w;
- ctx->config.dec->h = si->h;
- }
-
- res = ctx->iface->get_mmap(ctx, mmap, iter);
- }
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-vpx_dec_err_t vpx_dec_set_mem_map(vpx_dec_ctx_t *ctx_,
- vpx_dec_mmap_t *mmap,
- unsigned int num_maps)
-{
- vpx_codec_ctx_t *ctx = (vpx_codec_ctx_t *)ctx_;
- vpx_dec_err_t res = VPX_DEC_MEM_ERROR;
-
- if (!ctx || !mmap || !ctx->iface)
- res = VPX_DEC_INVALID_PARAM;
- else if (!(ctx->iface->caps & VPX_DEC_CAP_XMA))
- res = VPX_DEC_ERROR;
- else
- {
- void *save = (ctx->priv) ? NULL : ctx->config.dec;
- unsigned int i;
-
- for (i = 0; i < num_maps; i++, mmap++)
- {
- if (!mmap->base)
- break;
-
- /* Everything look ok, set the mmap in the decoder */
- res = ctx->iface->set_mmap(ctx, mmap);
-
- if (res)
- break;
- }
-
- if (save) free(save);
- }
-
- return SAVE_STATUS(ctx, res);
-}
diff --git a/vpx/src/vpx_encoder.c b/vpx/src/vpx_encoder.c
deleted file mode 100644
index db778ff..0000000
--- a/vpx/src/vpx_encoder.c
+++ /dev/null
@@ -1,312 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\file
- * \brief Provides the high level interface to wrap encoder algorithms.
- *
- */
-#include <limits.h>
-#include <string.h>
-#include "vpx/internal/vpx_codec_internal.h"
-
-#define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var)
-
-vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx,
- vpx_codec_iface_t *iface,
- vpx_codec_enc_cfg_t *cfg,
- vpx_codec_flags_t flags,
- int ver)
-{
- vpx_codec_err_t res;
-
- if (ver != VPX_ENCODER_ABI_VERSION)
- res = VPX_CODEC_ABI_MISMATCH;
- else if (!ctx || !iface || !cfg)
- res = VPX_CODEC_INVALID_PARAM;
- else if (iface->abi_version != VPX_CODEC_INTERNAL_ABI_VERSION)
- res = VPX_CODEC_ABI_MISMATCH;
- else if (!(iface->caps & VPX_CODEC_CAP_ENCODER))
- res = VPX_CODEC_INCAPABLE;
- else if ((flags & VPX_CODEC_USE_XMA) && !(iface->caps & VPX_CODEC_CAP_XMA))
- res = VPX_CODEC_INCAPABLE;
- else if ((flags & VPX_CODEC_USE_PSNR)
- && !(iface->caps & VPX_CODEC_CAP_PSNR))
- res = VPX_CODEC_INCAPABLE;
- else
- {
- ctx->iface = iface;
- ctx->name = iface->name;
- ctx->priv = NULL;
- ctx->init_flags = flags;
- ctx->config.enc = cfg;
- res = ctx->iface->init(ctx);
-
- if (res)
- {
- ctx->err_detail = ctx->priv ? ctx->priv->err_detail : NULL;
- vpx_codec_destroy(ctx);
- }
-
- if (ctx->priv)
- ctx->priv->iface = ctx->iface;
- }
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-
-vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface,
- vpx_codec_enc_cfg_t *cfg,
- unsigned int usage)
-{
- vpx_codec_err_t res;
- vpx_codec_enc_cfg_map_t *map;
-
- if (!iface || !cfg || usage > INT_MAX)
- res = VPX_CODEC_INVALID_PARAM;
- else if (!(iface->caps & VPX_CODEC_CAP_ENCODER))
- res = VPX_CODEC_INCAPABLE;
- else
- {
- res = VPX_CODEC_INVALID_PARAM;
-
- for (map = iface->enc.cfg_maps; map->usage >= 0; map++)
- {
- if (map->usage == (int)usage)
- {
- *cfg = map->cfg;
- cfg->g_usage = usage;
- res = VPX_CODEC_OK;
- break;
- }
- }
- }
-
- return res;
-}
-
-
-#if ARCH_X86 || ARCH_X86_64
-/* On X86, disable the x87 unit's internal 80 bit precision for better
- * consistency with the SSE unit's 64 bit precision.
- */
-#include "vpx_ports/x86.h"
-#define FLOATING_POINT_INIT() do {\
- unsigned short x87_orig_mode = x87_set_double_precision();
-#define FLOATING_POINT_RESTORE() \
- x87_set_control_word(x87_orig_mode); }while(0)
-
-
-#else
-static void FLOATING_POINT_INIT() {}
-static void FLOATING_POINT_RESTORE() {}
-#endif
-
-
-vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx,
- const vpx_image_t *img,
- vpx_codec_pts_t pts,
- unsigned long duration,
- vpx_enc_frame_flags_t flags,
- unsigned long deadline)
-{
- vpx_codec_err_t res;
-
- if (!ctx || (img && !duration))
- res = VPX_CODEC_INVALID_PARAM;
- else if (!ctx->iface || !ctx->priv)
- res = VPX_CODEC_ERROR;
- else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
- res = VPX_CODEC_INCAPABLE;
- else
- {
- /* Execute in a normalized floating point environment, if the platform
- * requires it.
- */
- FLOATING_POINT_INIT();
- res = ctx->iface->enc.encode(ctx->priv->alg_priv, img, pts,
- duration, flags, deadline);
- FLOATING_POINT_RESTORE();
- }
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx,
- vpx_codec_iter_t *iter)
-{
- const vpx_codec_cx_pkt_t *pkt = NULL;
-
- if (ctx)
- {
- if (!iter)
- ctx->err = VPX_CODEC_INVALID_PARAM;
- else if (!ctx->iface || !ctx->priv)
- ctx->err = VPX_CODEC_ERROR;
- else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
- ctx->err = VPX_CODEC_INCAPABLE;
- else
- pkt = ctx->iface->enc.get_cx_data(ctx->priv->alg_priv, iter);
- }
-
- if (pkt && pkt->kind == VPX_CODEC_CX_FRAME_PKT)
- {
- /* If the application has specified a destination area for the
- * compressed data, and the codec has not placed the data there,
- * and it fits, copy it.
- */
- char *dst_buf = ctx->priv->enc.cx_data_dst_buf.buf;
-
- if (dst_buf
- && pkt->data.raw.buf != dst_buf
- && pkt->data.raw.sz
- + ctx->priv->enc.cx_data_pad_before
- + ctx->priv->enc.cx_data_pad_after
- <= ctx->priv->enc.cx_data_dst_buf.sz)
- {
- vpx_codec_cx_pkt_t *modified_pkt = &ctx->priv->enc.cx_data_pkt;
-
- memcpy(dst_buf + ctx->priv->enc.cx_data_pad_before,
- pkt->data.raw.buf, pkt->data.raw.sz);
- *modified_pkt = *pkt;
- modified_pkt->data.raw.buf = dst_buf;
- modified_pkt->data.raw.sz += ctx->priv->enc.cx_data_pad_before
- + ctx->priv->enc.cx_data_pad_after;
- pkt = modified_pkt;
- }
-
- if (dst_buf == pkt->data.raw.buf)
- {
- ctx->priv->enc.cx_data_dst_buf.buf = dst_buf + pkt->data.raw.sz;
- ctx->priv->enc.cx_data_dst_buf.sz -= pkt->data.raw.sz;
- }
- }
-
- return pkt;
-}
-
-
-vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx,
- const vpx_fixed_buf_t *buf,
- unsigned int pad_before,
- unsigned int pad_after)
-{
- if (!ctx || !ctx->priv)
- return VPX_CODEC_INVALID_PARAM;
-
- if (buf)
- {
- ctx->priv->enc.cx_data_dst_buf = *buf;
- ctx->priv->enc.cx_data_pad_before = pad_before;
- ctx->priv->enc.cx_data_pad_after = pad_after;
- }
- else
- {
- ctx->priv->enc.cx_data_dst_buf.buf = NULL;
- ctx->priv->enc.cx_data_dst_buf.sz = 0;
- ctx->priv->enc.cx_data_pad_before = 0;
- ctx->priv->enc.cx_data_pad_after = 0;
- }
-
- return VPX_CODEC_OK;
-}
-
-
-const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx)
-{
- vpx_image_t *img = NULL;
-
- if (ctx)
- {
- if (!ctx->iface || !ctx->priv)
- ctx->err = VPX_CODEC_ERROR;
- else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
- ctx->err = VPX_CODEC_INCAPABLE;
- else if (!ctx->iface->enc.get_preview)
- ctx->err = VPX_CODEC_INCAPABLE;
- else
- img = ctx->iface->enc.get_preview(ctx->priv->alg_priv);
- }
-
- return img;
-}
-
-
-vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx)
-{
- vpx_fixed_buf_t *buf = NULL;
-
- if (ctx)
- {
- if (!ctx->iface || !ctx->priv)
- ctx->err = VPX_CODEC_ERROR;
- else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
- ctx->err = VPX_CODEC_INCAPABLE;
- else if (!ctx->iface->enc.get_glob_hdrs)
- ctx->err = VPX_CODEC_INCAPABLE;
- else
- buf = ctx->iface->enc.get_glob_hdrs(ctx->priv->alg_priv);
- }
-
- return buf;
-}
-
-
-vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx,
- const vpx_codec_enc_cfg_t *cfg)
-{
- vpx_codec_err_t res;
-
- if (!ctx || !ctx->iface || !ctx->priv || !cfg)
- res = VPX_CODEC_INVALID_PARAM;
- else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER))
- res = VPX_CODEC_INCAPABLE;
- else
- res = ctx->iface->enc.cfg_set(ctx->priv->alg_priv, cfg);
-
- return SAVE_STATUS(ctx, res);
-}
-
-
-int vpx_codec_pkt_list_add(struct vpx_codec_pkt_list *list,
- const struct vpx_codec_cx_pkt *pkt)
-{
- if (list->cnt < list->max)
- {
- list->pkts[list->cnt++] = *pkt;
- return 0;
- }
-
- return 1;
-}
-
-
-const vpx_codec_cx_pkt_t *vpx_codec_pkt_list_get(struct vpx_codec_pkt_list *list,
- vpx_codec_iter_t *iter)
-{
- const vpx_codec_cx_pkt_t *pkt;
-
- if (!(*iter))
- {
- *iter = list->pkts;
- }
-
- pkt = (const void *) * iter;
-
- if ((size_t)(pkt - list->pkts) < list->cnt)
- *iter = pkt + 1;
- else
- pkt = NULL;
-
- return pkt;
-}
diff --git a/vpx/src/vpx_image.c b/vpx/src/vpx_image.c
deleted file mode 100644
index 7a4e270..0000000
--- a/vpx/src/vpx_image.c
+++ /dev/null
@@ -1,262 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <stdlib.h>
-#include <string.h>
-#include "vpx/vpx_image.h"
-
-static vpx_image_t *img_alloc_helper(vpx_image_t *img,
- vpx_img_fmt_t fmt,
- unsigned int d_w,
- unsigned int d_h,
- unsigned int stride_align,
- unsigned char *img_data)
-{
-
- unsigned int h, w, s, xcs, ycs, bps;
- int align;
-
- /* Treat align==0 like align==1 */
- if (!stride_align)
- stride_align = 1;
-
- /* Validate alignment (must be power of 2) */
- if (stride_align & (stride_align - 1))
- goto fail;
-
- /* Get sample size for this format */
- switch (fmt)
- {
- case VPX_IMG_FMT_RGB32:
- case VPX_IMG_FMT_RGB32_LE:
- case VPX_IMG_FMT_ARGB:
- case VPX_IMG_FMT_ARGB_LE:
- bps = 32;
- break;
- case VPX_IMG_FMT_RGB24:
- case VPX_IMG_FMT_BGR24:
- bps = 24;
- break;
- case VPX_IMG_FMT_RGB565:
- case VPX_IMG_FMT_RGB565_LE:
- case VPX_IMG_FMT_RGB555:
- case VPX_IMG_FMT_RGB555_LE:
- case VPX_IMG_FMT_UYVY:
- case VPX_IMG_FMT_YUY2:
- case VPX_IMG_FMT_YVYU:
- bps = 16;
- break;
- case VPX_IMG_FMT_I420:
- case VPX_IMG_FMT_YV12:
- case VPX_IMG_FMT_VPXI420:
- case VPX_IMG_FMT_VPXYV12:
- bps = 12;
- break;
- default:
- bps = 16;
- break;
- }
-
- /* Get chroma shift values for this format */
- switch (fmt)
- {
- case VPX_IMG_FMT_I420:
- case VPX_IMG_FMT_YV12:
- case VPX_IMG_FMT_VPXI420:
- case VPX_IMG_FMT_VPXYV12:
- xcs = 1;
- break;
- default:
- xcs = 0;
- break;
- }
-
- switch (fmt)
- {
- case VPX_IMG_FMT_I420:
- case VPX_IMG_FMT_YV12:
- case VPX_IMG_FMT_VPXI420:
- case VPX_IMG_FMT_VPXYV12:
- ycs = 1;
- break;
- default:
- ycs = 0;
- break;
- }
-
- /* Calculate storage sizes given the chroma subsampling */
- align = (1 << xcs) - 1;
- w = (d_w + align) & ~align;
- align = (1 << ycs) - 1;
- h = (d_h + align) & ~align;
- s = (fmt & VPX_IMG_FMT_PLANAR) ? w : bps * w / 8;
- s = (s + stride_align - 1) & ~(stride_align - 1);
-
- /* Allocate the new image */
- if (!img)
- {
- img = (vpx_image_t *)calloc(1, sizeof(vpx_image_t));
-
- if (!img)
- goto fail;
-
- img->self_allocd = 1;
- }
- else
- {
- memset(img, 0, sizeof(vpx_image_t));
- }
-
- img->img_data = img_data;
-
- if (!img_data)
- {
- img->img_data = malloc((fmt & VPX_IMG_FMT_PLANAR) ? h * w * bps / 8 : h * s);
- img->img_data_owner = 1;
- }
-
- if (!img->img_data)
- goto fail;
-
- img->fmt = fmt;
- img->w = w;
- img->h = h;
- img->x_chroma_shift = xcs;
- img->y_chroma_shift = ycs;
- img->bps = bps;
-
- /* Calculate strides */
- img->stride[VPX_PLANE_Y] = img->stride[VPX_PLANE_ALPHA] = s;
- img->stride[VPX_PLANE_U] = img->stride[VPX_PLANE_V] = s >> xcs;
-
- /* Default viewport to entire image */
- if (!vpx_img_set_rect(img, 0, 0, d_w, d_h))
- return img;
-
-fail:
- vpx_img_free(img);
- return NULL;
-}
-
-vpx_image_t *vpx_img_alloc(vpx_image_t *img,
- vpx_img_fmt_t fmt,
- unsigned int d_w,
- unsigned int d_h,
- unsigned int stride_align)
-{
- return img_alloc_helper(img, fmt, d_w, d_h, stride_align, NULL);
-}
-
-vpx_image_t *vpx_img_wrap(vpx_image_t *img,
- vpx_img_fmt_t fmt,
- unsigned int d_w,
- unsigned int d_h,
- unsigned int stride_align,
- unsigned char *img_data)
-{
- return img_alloc_helper(img, fmt, d_w, d_h, stride_align, img_data);
-}
-
-int vpx_img_set_rect(vpx_image_t *img,
- unsigned int x,
- unsigned int y,
- unsigned int w,
- unsigned int h)
-{
- unsigned char *data;
-
- if (x + w <= img->w && y + h <= img->h)
- {
- img->d_w = w;
- img->d_h = h;
-
- /* Calculate plane pointers */
- if (!(img->fmt & VPX_IMG_FMT_PLANAR))
- {
- img->planes[VPX_PLANE_PACKED] =
- img->img_data + x * img->bps / 8 + y * img->stride[VPX_PLANE_PACKED];
- }
- else
- {
- data = img->img_data;
-
- if (img->fmt & VPX_IMG_FMT_HAS_ALPHA)
- {
- img->planes[VPX_PLANE_ALPHA] =
- data + x + y * img->stride[VPX_PLANE_ALPHA];
- data += img->h * img->stride[VPX_PLANE_ALPHA];
- }
-
- img->planes[VPX_PLANE_Y] = data + x + y * img->stride[VPX_PLANE_Y];
- data += img->h * img->stride[VPX_PLANE_Y];
-
- if (!(img->fmt & VPX_IMG_FMT_UV_FLIP))
- {
- img->planes[VPX_PLANE_U] = data
- + (x >> img->x_chroma_shift)
- + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U];
- data += (img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_U];
- img->planes[VPX_PLANE_V] = data
- + (x >> img->x_chroma_shift)
- + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_V];
- }
- else
- {
- img->planes[VPX_PLANE_V] = data
- + (x >> img->x_chroma_shift)
- + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_V];
- data += (img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_V];
- img->planes[VPX_PLANE_U] = data
- + (x >> img->x_chroma_shift)
- + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U];
- }
- }
-
- return 0;
- }
-
- return -1;
-}
-
-void vpx_img_flip(vpx_image_t *img)
-{
- /* Note: In the calculation pointer adjustment calculation, we want the
- * rhs to be promoted to a signed type. Section 6.3.1.8 of the ISO C99
- * standard indicates that if the adjustment parameter is unsigned, the
- * stride parameter will be promoted to unsigned, causing errors when
- * the lhs is a larger type than the rhs.
- */
- img->planes[VPX_PLANE_Y] += (signed)(img->d_h - 1) * img->stride[VPX_PLANE_Y];
- img->stride[VPX_PLANE_Y] = -img->stride[VPX_PLANE_Y];
-
- img->planes[VPX_PLANE_U] += (signed)((img->d_h >> img->y_chroma_shift) - 1)
- * img->stride[VPX_PLANE_U];
- img->stride[VPX_PLANE_U] = -img->stride[VPX_PLANE_U];
-
- img->planes[VPX_PLANE_V] += (signed)((img->d_h >> img->y_chroma_shift) - 1)
- * img->stride[VPX_PLANE_V];
- img->stride[VPX_PLANE_V] = -img->stride[VPX_PLANE_V];
-
- img->planes[VPX_PLANE_ALPHA] += (signed)(img->d_h - 1) * img->stride[VPX_PLANE_ALPHA];
- img->stride[VPX_PLANE_ALPHA] = -img->stride[VPX_PLANE_ALPHA];
-}
-
-void vpx_img_free(vpx_image_t *img)
-{
- if (img)
- {
- if (img->img_data && img->img_data_owner)
- free(img->img_data);
-
- if (img->self_allocd)
- free(img);
- }
-}
diff --git a/vpx/vp8.h b/vpx/vp8.h
deleted file mode 100644
index 7ca5c6e..0000000
--- a/vpx/vp8.h
+++ /dev/null
@@ -1,130 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\defgroup vp8 VP8
- * \ingroup codecs
- * VP8 is vpx's newest video compression algorithm that uses motion
- * compensated prediction, Discrete Cosine Transform (DCT) coding of the
- * prediction error signal and context dependent entropy coding techniques
- * based on arithmetic principles. It features:
- * - YUV 4:2:0 image format
- * - Macro-block based coding (16x16 luma plus two 8x8 chroma)
- * - 1/4 (1/8) pixel accuracy motion compensated prediction
- * - 4x4 DCT transform
- * - 128 level linear quantizer
- * - In loop deblocking filter
- * - Context-based entropy coding
- *
- * @{
- */
-/*!\file
- * \brief Provides controls common to both the VP8 encoder and decoder.
- */
-#ifndef VP8_H
-#define VP8_H
-#include "vpx/vpx_codec_impl_top.h"
-
-/*!\brief Control functions
- *
- * The set of macros define the control functions of VP8 interface
- */
-enum vp8_com_control_id
-{
- VP8_SET_REFERENCE = 1, /**< pass in an external frame into decoder to be used as reference frame */
- VP8_COPY_REFERENCE = 2, /**< get a copy of reference frame from the decoder */
- VP8_SET_POSTPROC = 3, /**< set the decoder's post processing settings */
- VP8_SET_DBG_COLOR_REF_FRAME = 4, /**< set the reference frames to color for each macroblock */
- VP8_SET_DBG_COLOR_MB_MODES = 5, /**< set which macro block modes to color */
- VP8_SET_DBG_COLOR_B_MODES = 6, /**< set which blocks modes to color */
- VP8_SET_DBG_DISPLAY_MV = 7, /**< set which motion vector modes to draw */
- VP8_COMMON_CTRL_ID_MAX,
- VP8_DECODER_CTRL_ID_START = 256,
-};
-
-/*!\brief post process flags
- *
- * The set of macros define VP8 decoder post processing flags
- */
-enum vp8_postproc_level
-{
- VP8_NOFILTERING = 0,
- VP8_DEBLOCK = 1<<0,
- VP8_DEMACROBLOCK = 1<<1,
- VP8_ADDNOISE = 1<<2,
- VP8_DEBUG_TXT_FRAME_INFO = 1<<3, /**< print frame information */
- VP8_DEBUG_TXT_MBLK_MODES = 1<<4, /**< print macro block modes over each macro block */
- VP8_DEBUG_TXT_DC_DIFF = 1<<5, /**< print dc diff for each macro block */
- VP8_DEBUG_TXT_RATE_INFO = 1<<6, /**< print video rate info (encoder only) */
-};
-
-/*!\brief post process flags
- *
- * This define a structure that describe the post processing settings. For
- * the best objective measure (using the PSNR metric) set post_proc_flag
- * to VP8_DEBLOCK and deblocking_level to 1.
- */
-
-typedef struct vp8_postproc_cfg
-{
- int post_proc_flag; /**< the types of post processing to be done, should be combination of "vp8_postproc_level" */
- int deblocking_level; /**< the strength of deblocking, valid range [0, 16] */
- int noise_level; /**< the strength of additive noise, valid range [0, 16] */
-} vp8_postproc_cfg_t;
-
-/*!\brief reference frame type
- *
- * The set of macros define the type of VP8 reference frames
- */
-typedef enum vpx_ref_frame_type
-{
- VP8_LAST_FRAME = 1,
- VP8_GOLD_FRAME = 2,
- VP8_ALTR_FRAME = 4
-} vpx_ref_frame_type_t;
-
-/*!\brief reference frame data struct
- *
- * define the data struct to access vp8 reference frames
- */
-
-typedef struct vpx_ref_frame
-{
- vpx_ref_frame_type_t frame_type; /**< which reference frame */
- vpx_image_t img; /**< reference frame data in image format */
-} vpx_ref_frame_t;
-
-
-/*!\brief vp8 decoder control function parameter type
- *
- * defines the data type for each of VP8 decoder control function requires
- */
-
-VPX_CTRL_USE_TYPE(VP8_SET_REFERENCE, vpx_ref_frame_t *)
-VPX_CTRL_USE_TYPE(VP8_COPY_REFERENCE, vpx_ref_frame_t *)
-VPX_CTRL_USE_TYPE(VP8_SET_POSTPROC, vp8_postproc_cfg_t *)
-VPX_CTRL_USE_TYPE(VP8_SET_DBG_COLOR_REF_FRAME, int)
-VPX_CTRL_USE_TYPE(VP8_SET_DBG_COLOR_MB_MODES, int)
-VPX_CTRL_USE_TYPE(VP8_SET_DBG_COLOR_B_MODES, int)
-VPX_CTRL_USE_TYPE(VP8_SET_DBG_DISPLAY_MV, int)
-
-
-/*! @} - end defgroup vp8 */
-
-#if !defined(VPX_CODEC_DISABLE_COMPAT) || !VPX_CODEC_DISABLE_COMPAT
-/* The following definitions are provided for backward compatibility with
- * the VP8 1.0.x SDK. USE IN PRODUCTION CODE IS NOT RECOMMENDED.
- */
-
-DECLSPEC_DEPRECATED extern vpx_codec_iface_t vpx_codec_vp8_algo DEPRECATED;
-#endif
-
-#include "vpx/vpx_codec_impl_bottom.h"
-#endif
diff --git a/vpx/vp8cx.h b/vpx/vp8cx.h
deleted file mode 100644
index 5ff6bdc..0000000
--- a/vpx/vp8cx.h
+++ /dev/null
@@ -1,300 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\defgroup vp8_encoder WebM VP8 Encoder
- * \ingroup vp8
- *
- * @{
- */
-#include "vp8.h"
-
-/*!\file
- * \brief Provides definitions for using the VP8 encoder algorithm within the
- * vpx Codec Interface.
- */
-#ifndef VP8CX_H
-#define VP8CX_H
-#include "vpx/vpx_codec_impl_top.h"
-
-/*!\name Algorithm interface for VP8
- *
- * This interface provides the capability to encode raw VP8 streams, as would
- * be found in AVI files.
- * @{
- */
-extern vpx_codec_iface_t vpx_codec_vp8_cx_algo;
-extern vpx_codec_iface_t* vpx_codec_vp8_cx(void);
-/*!@} - end algorithm interface member group*/
-
-
-/*
- * Algorithm Flags
- */
-
-/*!\brief Don't reference the last frame
- *
- * When this flag is set, the encoder will not use the last frame as a
- * predictor. When not set, the encoder will choose whether to use the
- * last frame or not automatically.
- */
-#define VP8_EFLAG_NO_REF_LAST (1<<16)
-
-
-/*!\brief Don't reference the golden frame
- *
- * When this flag is set, the encoder will not use the golden frame as a
- * predictor. When not set, the encoder will choose whether to use the
- * golden frame or not automatically.
- */
-#define VP8_EFLAG_NO_REF_GF (1<<17)
-
-
-/*!\brief Don't reference the alternate reference frame
- *
- * When this flag is set, the encoder will not use the alt ref frame as a
- * predictor. When not set, the encoder will choose whether to use the
- * alt ref frame or not automatically.
- */
-#define VP8_EFLAG_NO_REF_ARF (1<<21)
-
-
-/*!\brief Don't update the last frame
- *
- * When this flag is set, the encoder will not update the last frame with
- * the contents of the current frame.
- */
-#define VP8_EFLAG_NO_UPD_LAST (1<<18)
-
-
-/*!\brief Don't update the golden frame
- *
- * When this flag is set, the encoder will not update the golden frame with
- * the contents of the current frame.
- */
-#define VP8_EFLAG_NO_UPD_GF (1<<22)
-
-
-/*!\brief Don't update the alternate reference frame
- *
- * When this flag is set, the encoder will not update the alt ref frame with
- * the contents of the current frame.
- */
-#define VP8_EFLAG_NO_UPD_ARF (1<<23)
-
-
-/*!\brief Force golden frame update
- *
- * When this flag is set, the encoder copy the contents of the current frame
- * to the golden frame buffer.
- */
-#define VP8_EFLAG_FORCE_GF (1<<19)
-
-
-/*!\brief Force alternate reference frame update
- *
- * When this flag is set, the encoder copy the contents of the current frame
- * to the alternate reference frame buffer.
- */
-#define VP8_EFLAG_FORCE_ARF (1<<24)
-
-
-/*!\brief Disable entropy update
- *
- * When this flag is set, the encoder will not update its internal entropy
- * model based on the entropy of this frame.
- */
-#define VP8_EFLAG_NO_UPD_ENTROPY (1<<20)
-
-
-/*!\brief VP8 encoder control functions
- *
- * This set of macros define the control functions available for the VP8
- * encoder interface.
- *
- * \sa #vpx_codec_control
- */
-enum vp8e_enc_control_id
-{
- VP8E_UPD_ENTROPY = 5, /**< control function to set mode of entropy update in encoder */
- VP8E_UPD_REFERENCE, /**< control function to set reference update mode in encoder */
- VP8E_USE_REFERENCE, /**< control function to set which reference frame encoder can use */
- VP8E_SET_ROI_MAP, /**< control function to pass an ROI map to encoder */
- VP8E_SET_ACTIVEMAP, /**< control function to pass an Active map to encoder */
- VP8E_SET_SCALEMODE = 11, /**< control function to set encoder scaling mode */
- /*!\brief control function to set vp8 encoder cpuused
- *
- * Changes in this value influences, among others, the encoder's selection
- * of motion estimation methods. Values greater than 0 will increase encoder
- * speed at the expense of quality.
- * The full set of adjustments can be found in
- * onyx_if.c:vp8_set_speed_features().
- * \todo List highlights of the changes at various levels.
- *
- * \note Valid range: -16..16 or {-16..-4, 4..16} w/CONFIG_REALTIME_ONLY
- */
- VP8E_SET_CPUUSED = 13,
- VP8E_SET_ENABLEAUTOALTREF, /**< control function to enable vp8 to automatic set and use altref frame */
- VP8E_SET_NOISE_SENSITIVITY, /**< control function to set noise sensitivity */
- VP8E_SET_SHARPNESS, /**< control function to set sharpness */
- VP8E_SET_STATIC_THRESHOLD, /**< control function to set the threshold for macroblocks treated static */
- VP8E_SET_TOKEN_PARTITIONS, /**< control function to set the number of token partitions */
- VP8E_GET_LAST_QUANTIZER, /**< return the quantizer chosen by the
- encoder for the last frame using the internal
- scale */
- VP8E_GET_LAST_QUANTIZER_64, /**< return the quantizer chosen by the
- encoder for the last frame, using the 0..63
- scale as used by the rc_*_quantizer config
- parameters */
- VP8E_SET_ARNR_MAXFRAMES, /**< control function to set the max number of frames blurred creating arf*/
- VP8E_SET_ARNR_STRENGTH , /**< control function to set the filter strength for the arf */
- VP8E_SET_ARNR_TYPE , /**< control function to set the type of filter to use for the arf*/
- VP8E_SET_TUNING, /**< control function to set visual tuning */
- /*!\brief control function to set constrained quality level
- *
- * \attention For this value to be used vpx_codec_enc_cfg_t::g_usage must be
- * set to #VPX_CQ.
- * \note Valid range: 0..63
- */
- VP8E_SET_CQ_LEVEL,
-};
-
-/*!\brief vpx 1-D scaling mode
- *
- * This set of constants define 1-D vpx scaling modes
- */
-typedef enum vpx_scaling_mode_1d
-{
- VP8E_NORMAL = 0,
- VP8E_FOURFIVE = 1,
- VP8E_THREEFIVE = 2,
- VP8E_ONETWO = 3
-} VPX_SCALING_MODE;
-
-
-/*!\brief vpx region of interest map
- *
- * These defines the data structures for the region of interest map
- *
- */
-
-typedef struct vpx_roi_map
-{
- unsigned char *roi_map; /**< specify an id between 0 and 3 for each 16x16 region within a frame */
- unsigned int rows; /**< number of rows */
- unsigned int cols; /**< number of cols */
- int delta_q[4]; /**< quantizer delta [-64, 64] off baseline for regions with id between 0 and 3*/
- int delta_lf[4]; /**< loop filter strength delta [-32, 32] for regions with id between 0 and 3 */
- unsigned int static_threshold[4];/**< threshold for region to be treated as static */
-} vpx_roi_map_t;
-
-/*!\brief vpx active region map
- *
- * These defines the data structures for active region map
- *
- */
-
-
-typedef struct vpx_active_map
-{
- unsigned char *active_map; /**< specify an on (1) or off (0) each 16x16 region within a frame */
- unsigned int rows; /**< number of rows */
- unsigned int cols; /**< number of cols */
-} vpx_active_map_t;
-
-/*!\brief vpx image scaling mode
- *
- * This defines the data structure for image scaling mode
- *
- */
-typedef struct vpx_scaling_mode
-{
- VPX_SCALING_MODE h_scaling_mode; /**< horizontal scaling mode */
- VPX_SCALING_MODE v_scaling_mode; /**< vertical scaling mode */
-} vpx_scaling_mode_t;
-
-/*!\brief VP8 encoding mode
- *
- * This defines VP8 encoding mode
- *
- */
-typedef enum
-{
- VP8_BEST_QUALITY_ENCODING,
- VP8_GOOD_QUALITY_ENCODING,
- VP8_REAL_TIME_ENCODING
-} vp8e_encoding_mode;
-
-/*!\brief VP8 token partition mode
- *
- * This defines VP8 partitioning mode for compressed data, i.e., the number of
- * sub-streams in the bitstream. Used for parallelized decoding.
- *
- */
-
-typedef enum
-{
- VP8_ONE_TOKENPARTITION = 0,
- VP8_TWO_TOKENPARTITION = 1,
- VP8_FOUR_TOKENPARTITION = 2,
- VP8_EIGHT_TOKENPARTITION = 3,
-} vp8e_token_partitions;
-
-
-/*!\brief VP8 model tuning parameters
- *
- * Changes the encoder to tune for certain types of input material.
- *
- */
-typedef enum
-{
- VP8_TUNE_PSNR,
- VP8_TUNE_SSIM
-} vp8e_tuning;
-
-
-/*!\brief VP8 encoder control function parameter type
- *
- * Defines the data types that VP8E control functions take. Note that
- * additional common controls are defined in vp8.h
- *
- */
-
-
-/* These controls have been deprecated in favor of the flags parameter to
- * vpx_codec_encode(). See the definition of VP8_EFLAG_* above.
- */
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_UPD_ENTROPY, int)
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_UPD_REFERENCE, int)
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_USE_REFERENCE, int)
-
-VPX_CTRL_USE_TYPE(VP8E_SET_ROI_MAP, vpx_roi_map_t *)
-VPX_CTRL_USE_TYPE(VP8E_SET_ACTIVEMAP, vpx_active_map_t *)
-VPX_CTRL_USE_TYPE(VP8E_SET_SCALEMODE, vpx_scaling_mode_t *)
-
-VPX_CTRL_USE_TYPE(VP8E_SET_CPUUSED, int)
-VPX_CTRL_USE_TYPE(VP8E_SET_ENABLEAUTOALTREF, unsigned int)
-VPX_CTRL_USE_TYPE(VP8E_SET_NOISE_SENSITIVITY, unsigned int)
-VPX_CTRL_USE_TYPE(VP8E_SET_SHARPNESS, unsigned int)
-VPX_CTRL_USE_TYPE(VP8E_SET_STATIC_THRESHOLD, unsigned int)
-VPX_CTRL_USE_TYPE(VP8E_SET_TOKEN_PARTITIONS, vp8e_token_partitions)
-
-VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_MAXFRAMES, unsigned int)
-VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_STRENGTH , unsigned int)
-VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_TYPE , unsigned int)
-VPX_CTRL_USE_TYPE(VP8E_SET_TUNING, vp8e_tuning)
-VPX_CTRL_USE_TYPE(VP8E_SET_CQ_LEVEL , unsigned int)
-
-VPX_CTRL_USE_TYPE(VP8E_GET_LAST_QUANTIZER, int *)
-VPX_CTRL_USE_TYPE(VP8E_GET_LAST_QUANTIZER_64, int *)
-
-/*! @} - end defgroup vp8_encoder */
-#include "vpx/vpx_codec_impl_bottom.h"
-#endif
diff --git a/vpx/vp8dx.h b/vpx/vp8dx.h
deleted file mode 100644
index 4a3aef7..0000000
--- a/vpx/vp8dx.h
+++ /dev/null
@@ -1,78 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vp8.h"
-
-/*!\defgroup vp8_decoder WebM VP8 Decoder
- * \ingroup vp8
- *
- * @{
- */
-/*!\file
- * \brief Provides definitions for using the VP8 algorithm within the vpx Decoder
- * interface.
- */
-#ifndef VP8DX_H
-#define VP8DX_H
-#include "vpx/vpx_codec_impl_top.h"
-
-/*!\name Algorithm interface for VP8
- *
- * This interface provides the capability to decode raw VP8 streams, as would
- * be found in AVI files and other non-Flash uses.
- * @{
- */
-extern vpx_codec_iface_t vpx_codec_vp8_dx_algo;
-extern vpx_codec_iface_t* vpx_codec_vp8_dx(void);
-/*!@} - end algorithm interface member group*/
-
-/* Include controls common to both the encoder and decoder */
-#include "vp8.h"
-
-
-/*!\brief VP8 decoder control functions
- *
- * This set of macros define the control functions available for the VP8
- * decoder interface.
- *
- * \sa #vpx_codec_control
- */
-enum vp8_dec_control_id
-{
- /** control function to get info on which reference frames were updated
- * by the last decode
- */
- VP8D_GET_LAST_REF_UPDATES = VP8_DECODER_CTRL_ID_START,
-
- /** check if the indicated frame is corrupted */
- VP8D_GET_FRAME_CORRUPTED,
-
- VP8_DECODER_CTRL_ID_MAX
-} ;
-
-
-/*!\brief VP8 decoder control function parameter type
- *
- * Defines the data types that VP8D control functions take. Note that
- * additional common controls are defined in vp8.h
- *
- */
-
-
-VPX_CTRL_USE_TYPE(VP8D_GET_LAST_REF_UPDATES, int *)
-VPX_CTRL_USE_TYPE(VP8D_GET_FRAME_CORRUPTED, int *)
-
-
-/*! @} - end defgroup vp8_decoder */
-
-
-#include "vpx/vpx_codec_impl_bottom.h"
-#endif
diff --git a/vpx/vp8e.h b/vpx/vp8e.h
deleted file mode 100644
index abfce33..0000000
--- a/vpx/vp8e.h
+++ /dev/null
@@ -1,63 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/* This file contains backwards compatibility stubs for applications using
- * the VP8 version 1.0 API.
- */
-#ifndef VP8E_H
-#define VP8E_H
-#include "vpx/vpx_codec_impl_top.h"
-
-#if defined(VPX_CODEC_DISABLE_COMPAT) && VPX_CODEC_DISABLE_COMPAT
-#error "Backwards compatibility disabled: don't include vp8e.h"
-#endif
-
-#include "vp8cx.h"
-DECLSPEC_DEPRECATED extern vpx_codec_iface_t vpx_enc_vp8_algo DEPRECATED;
-
-
-enum
-{
- VP8E_SET_REFERENCE = VP8_SET_REFERENCE,
- VP8E_COPY_REFERENCE = VP8_COPY_REFERENCE,
- VP8E_SET_PREVIEWPP = VP8_SET_POSTPROC,
- VP8E_SET_FLUSHFLAG = 4,
- VP8E_SET_FRAMETYPE = 10,
- VP8E_SET_ENCODING_MODE = 12
-};
-
-#define NORMAL_FRAME (0)
-#define KEY_FRAME (1)
-
-/* Change VP8E to VP8 to get the undeprecated version of these (defined in
- * vp8.h)
- */
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_SET_REFERENCE, vpx_ref_frame_t *)
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_COPY_REFERENCE, vpx_ref_frame_t *)
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_SET_PREVIEWPP, vp8_postproc_cfg_t *)
-
-
-/* Flush is done by calling vpx_codec_encode with a NULL input image. */
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_SET_FLUSHFLAG, int)
-
-
-/* Frame type is set with a flag to vpx_codec_control. See VPX_EFLAG_FORCE_KF
- */
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_SET_FRAMETYPE, int)
-
-
-/* This control has been deprecated in favor of the duration parameter to
- * vpx_codec_encode(). Use the #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY,
- * #VPX_DL_BEST_QUALITY constants to that parameter instead.
- */
-VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_SET_ENCODING_MODE, vp8e_encoding_mode)
-#include "vpx/vpx_codec_impl_bottom.h"
-#endif
diff --git a/vpx/vpx_codec.mk b/vpx/vpx_codec.mk
deleted file mode 100644
index 4f1d74b..0000000
--- a/vpx/vpx_codec.mk
+++ /dev/null
@@ -1,27 +0,0 @@
-##
-## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-##
-## Use of this source code is governed by a BSD-style license
-## that can be found in the LICENSE file in the root of the source
-## tree. An additional intellectual property rights grant can be found
-## in the file PATENTS. All contributing project authors may
-## be found in the AUTHORS file in the root of the source tree.
-##
-
-
-API_EXPORTS += exports
-
-API_SRCS-$(CONFIG_DECODERS) += src/vpx_decoder.c
-API_SRCS-$(CONFIG_DECODERS) += src/vpx_decoder_compat.c
-API_SRCS-$(CONFIG_DECODERS) += vpx_decoder.h
-API_SRCS-$(CONFIG_DECODERS) += vpx_decoder_compat.h
-API_SRCS-$(CONFIG_ENCODERS) += src/vpx_encoder.c
-API_SRCS-$(CONFIG_ENCODERS) += vpx_encoder.h
-API_SRCS-yes += internal/vpx_codec_internal.h
-API_SRCS-yes += src/vpx_codec.c
-API_SRCS-yes += src/vpx_image.c
-API_SRCS-yes += vpx_codec.h
-API_SRCS-yes += vpx_codec.mk
-API_SRCS-yes += vpx_codec_impl_bottom.h
-API_SRCS-yes += vpx_codec_impl_top.h
-API_SRCS-yes += vpx_image.h
diff --git a/vpx/vpx_decoder.h b/vpx/vpx_decoder.h
deleted file mode 100644
index 003675a..0000000
--- a/vpx/vpx_decoder.h
+++ /dev/null
@@ -1,323 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\defgroup decoder Decoder Algorithm Interface
- * \ingroup codec
- * This abstraction allows applications using this decoder to easily support
- * multiple video formats with minimal code duplication. This section describes
- * the interface common to all decoders.
- * @{
- */
-
-/*!\file
- * \brief Describes the decoder algorithm interface to applications.
- *
- * This file describes the interface between an application and a
- * video decoder algorithm.
- *
- */
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#ifndef VPX_DECODER_H
-#define VPX_DECODER_H
-#include "vpx_codec.h"
-
- /*!\brief Current ABI version number
- *
- * \internal
- * If this file is altered in any way that changes the ABI, this value
- * must be bumped. Examples include, but are not limited to, changing
- * types, removing or reassigning enums, adding/removing/rearranging
- * fields to structures
- */
-#define VPX_DECODER_ABI_VERSION (2 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/
-
- /*! \brief Decoder capabilities bitfield
- *
- * Each decoder advertises the capabilities it supports as part of its
- * ::vpx_codec_iface_t interface structure. Capabilities are extra interfaces
- * or functionality, and are not required to be supported by a decoder.
- *
- * The available flags are specified by VPX_CODEC_CAP_* defines.
- */
-#define VPX_CODEC_CAP_PUT_SLICE 0x10000 /**< Will issue put_slice callbacks */
-#define VPX_CODEC_CAP_PUT_FRAME 0x20000 /**< Will issue put_frame callbacks */
-#define VPX_CODEC_CAP_POSTPROC 0x40000 /**< Can postprocess decoded frame */
-
- /*! \brief Initialization-time Feature Enabling
- *
- * Certain codec features must be known at initialization time, to allow for
- * proper memory allocation.
- *
- * The available flags are specified by VPX_CODEC_USE_* defines.
- */
-#define VPX_CODEC_USE_POSTPROC 0x10000 /**< Postprocess decoded frame */
-
- /*!\brief Stream properties
- *
- * This structure is used to query or set properties of the decoded
- * stream. Algorithms may extend this structure with data specific
- * to their bitstream by setting the sz member appropriately.
- */
- typedef struct vpx_codec_stream_info
- {
- unsigned int sz; /**< Size of this structure */
- unsigned int w; /**< Width (or 0 for unknown/default) */
- unsigned int h; /**< Height (or 0 for unknown/default) */
- unsigned int is_kf; /**< Current frame is a keyframe */
- } vpx_codec_stream_info_t;
-
- /* REQUIRED FUNCTIONS
- *
- * The following functions are required to be implemented for all decoders.
- * They represent the base case functionality expected of all decoders.
- */
-
-
- /*!\brief Initialization Configurations
- *
- * This structure is used to pass init time configuration options to the
- * decoder.
- */
- typedef struct vpx_codec_dec_cfg
- {
- unsigned int threads; /**< Maximum number of threads to use, default 1 */
- unsigned int w; /**< Width */
- unsigned int h; /**< Height */
- } vpx_codec_dec_cfg_t; /**< alias for struct vpx_codec_dec_cfg */
-
-
- /*!\brief Initialize a decoder instance
- *
- * Initializes a decoder context using the given interface. Applications
- * should call the vpx_codec_dec_init convenience macro instead of this
- * function directly, to ensure that the ABI version number parameter
- * is properly initialized.
- *
- * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
- * parameter), the storage pointed to by the cfg parameter must be
- * kept readable and stable until all memory maps have been set.
- *
- * \param[in] ctx Pointer to this instance's context.
- * \param[in] iface Pointer to the algorithm interface to use.
- * \param[in] cfg Configuration to use, if known. May be NULL.
- * \param[in] flags Bitfield of VPX_CODEC_USE_* flags
- * \param[in] ver ABI version number. Must be set to
- * VPX_DECODER_ABI_VERSION
- * \retval #VPX_CODEC_OK
- * The decoder algorithm initialized.
- * \retval #VPX_CODEC_MEM_ERROR
- * Memory allocation failed.
- */
- vpx_codec_err_t vpx_codec_dec_init_ver(vpx_codec_ctx_t *ctx,
- vpx_codec_iface_t *iface,
- vpx_codec_dec_cfg_t *cfg,
- vpx_codec_flags_t flags,
- int ver);
-
- /*!\brief Convenience macro for vpx_codec_dec_init_ver()
- *
- * Ensures the ABI version parameter is properly set.
- */
-#define vpx_codec_dec_init(ctx, iface, cfg, flags) \
- vpx_codec_dec_init_ver(ctx, iface, cfg, flags, VPX_DECODER_ABI_VERSION)
-
-
- /*!\brief Parse stream info from a buffer
- *
- * Performs high level parsing of the bitstream. Construction of a decoder
- * context is not necessary. Can be used to determine if the bitstream is
- * of the proper format, and to extract information from the stream.
- *
- * \param[in] iface Pointer to the algorithm interface
- * \param[in] data Pointer to a block of data to parse
- * \param[in] data_sz Size of the data buffer
- * \param[in,out] si Pointer to stream info to update. The size member
- * \ref MUST be properly initialized, but \ref MAY be
- * clobbered by the algorithm. This parameter \ref MAY
- * be NULL.
- *
- * \retval #VPX_CODEC_OK
- * Bitstream is parsable and stream information updated
- */
- vpx_codec_err_t vpx_codec_peek_stream_info(vpx_codec_iface_t *iface,
- const uint8_t *data,
- unsigned int data_sz,
- vpx_codec_stream_info_t *si);
-
-
- /*!\brief Return information about the current stream.
- *
- * Returns information about the stream that has been parsed during decoding.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in,out] si Pointer to stream info to update. The size member
- * \ref MUST be properly initialized, but \ref MAY be
- * clobbered by the algorithm. This parameter \ref MAY
- * be NULL.
- *
- * \retval #VPX_CODEC_OK
- * Bitstream is parsable and stream information updated
- */
- vpx_codec_err_t vpx_codec_get_stream_info(vpx_codec_ctx_t *ctx,
- vpx_codec_stream_info_t *si);
-
-
- /*!\brief Decode data
- *
- * Processes a buffer of coded data. If the processing results in a new
- * decoded frame becoming available, PUT_SLICE and PUT_FRAME events may be
- * generated, as appropriate. Encoded data \ref MUST be passed in DTS (decode
- * time stamp) order. Frames produced will always be in PTS (presentation
- * time stamp) order.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] data Pointer to this block of new coded data. If
- * NULL, a VPX_CODEC_CB_PUT_FRAME event is posted
- * for the previously decoded frame.
- * \param[in] data_sz Size of the coded data, in bytes.
- * \param[in] user_priv Application specific data to associate with
- * this frame.
- * \param[in] deadline Soft deadline the decoder should attempt to meet,
- * in us. Set to zero for unlimited.
- *
- * \return Returns #VPX_CODEC_OK if the coded data was processed completely
- * and future pictures can be decoded without error. Otherwise,
- * see the descriptions of the other error codes in ::vpx_codec_err_t
- * for recoverability capabilities.
- */
- vpx_codec_err_t vpx_codec_decode(vpx_codec_ctx_t *ctx,
- const uint8_t *data,
- unsigned int data_sz,
- void *user_priv,
- long deadline);
-
-
- /*!\brief Decoded frames iterator
- *
- * Iterates over a list of the frames available for display. The iterator
- * storage should be initialized to NULL to start the iteration. Iteration is
- * complete when this function returns NULL.
- *
- * The list of available frames becomes valid upon completion of the
- * vpx_codec_decode call, and remains valid until the next call to vpx_codec_decode.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in,out] iter Iterator storage, initialized to NULL
- *
- * \return Returns a pointer to an image, if one is ready for display. Frames
- * produced will always be in PTS (presentation time stamp) order.
- */
- vpx_image_t *vpx_codec_get_frame(vpx_codec_ctx_t *ctx,
- vpx_codec_iter_t *iter);
-
-
- /*!\defgroup cap_put_frame Frame-Based Decoding Functions
- *
- * The following functions are required to be implemented for all decoders
- * that advertise the VPX_CODEC_CAP_PUT_FRAME capability. Calling these functions
- * for codecs that don't advertise this capability will result in an error
- * code being returned, usually VPX_CODEC_ERROR
- * @{
- */
-
- /*!\brief put frame callback prototype
- *
- * This callback is invoked by the decoder to notify the application of
- * the availability of decoded image data.
- */
- typedef void (*vpx_codec_put_frame_cb_fn_t)(void *user_priv,
- const vpx_image_t *img);
-
-
- /*!\brief Register for notification of frame completion.
- *
- * Registers a given function to be called when a decoded frame is
- * available.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] cb Pointer to the callback function
- * \param[in] user_priv User's private data
- *
- * \retval #VPX_CODEC_OK
- * Callback successfully registered.
- * \retval #VPX_CODEC_ERROR
- * Decoder context not initialized, or algorithm not capable of
- * posting slice completion.
- */
- vpx_codec_err_t vpx_codec_register_put_frame_cb(vpx_codec_ctx_t *ctx,
- vpx_codec_put_frame_cb_fn_t cb,
- void *user_priv);
-
-
- /*!@} - end defgroup cap_put_frame */
-
- /*!\defgroup cap_put_slice Slice-Based Decoding Functions
- *
- * The following functions are required to be implemented for all decoders
- * that advertise the VPX_CODEC_CAP_PUT_SLICE capability. Calling these functions
- * for codecs that don't advertise this capability will result in an error
- * code being returned, usually VPX_CODEC_ERROR
- * @{
- */
-
- /*!\brief put slice callback prototype
- *
- * This callback is invoked by the decoder to notify the application of
- * the availability of partially decoded image data. The
- */
- typedef void (*vpx_codec_put_slice_cb_fn_t)(void *user_priv,
- const vpx_image_t *img,
- const vpx_image_rect_t *valid,
- const vpx_image_rect_t *update);
-
-
- /*!\brief Register for notification of slice completion.
- *
- * Registers a given function to be called when a decoded slice is
- * available.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] cb Pointer to the callback function
- * \param[in] user_priv User's private data
- *
- * \retval #VPX_CODEC_OK
- * Callback successfully registered.
- * \retval #VPX_CODEC_ERROR
- * Decoder context not initialized, or algorithm not capable of
- * posting slice completion.
- */
- vpx_codec_err_t vpx_codec_register_put_slice_cb(vpx_codec_ctx_t *ctx,
- vpx_codec_put_slice_cb_fn_t cb,
- void *user_priv);
-
-
-#ifdef MIPS_DSP_REV
-#if (MIPS_DSP_REV>=2)
- void dsputil_static_init();
-#endif
-#endif
-
- /*!@} - end defgroup cap_put_slice*/
-
- /*!@} - end defgroup decoder*/
-
-#endif
-
-#ifdef __cplusplus
-}
-#endif
-
-#if !defined(VPX_CODEC_DISABLE_COMPAT) || !VPX_CODEC_DISABLE_COMPAT
-#include "vpx_decoder_compat.h"
-#endif
diff --git a/vpx/vpx_decoder_compat.h b/vpx/vpx_decoder_compat.h
deleted file mode 100644
index 8adc1b9..0000000
--- a/vpx/vpx_decoder_compat.h
+++ /dev/null
@@ -1,587 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\defgroup decoder Common Decoder Algorithm Interface
- * This abstraction allows applications using this decoder to easily support
- * multiple video formats with minimal code duplication. This section describes
- * the interface common to all codecs.
- * @{
- */
-
-/*!\file
- * \brief Provides a compatibility layer between version 1 and 2 of this API.
- *
- * This interface has been deprecated. Only existing code should make use
- * of this interface, and therefore, it is only thinly documented. Existing
- * code should be ported to the vpx_codec_* API.
- */
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#ifndef VPX_DECODER_COMPAT_H
-#define VPX_DECODER_COMPAT_H
-
- /*!\brief Decoder algorithm return codes */
- typedef enum {
- /*!\brief Operation completed without error */
- VPX_DEC_OK = VPX_CODEC_OK,
-
- /*!\brief Unspecified error */
- VPX_DEC_ERROR = VPX_CODEC_ERROR,
-
- /*!\brief Memory operation failed */
- VPX_DEC_MEM_ERROR = VPX_CODEC_MEM_ERROR,
-
- /*!\brief ABI version mismatch */
- VPX_DEC_ABI_MISMATCH = VPX_CODEC_ABI_MISMATCH,
-
- /*!\brief The given bitstream is not supported.
- *
- * The bitstream was unable to be parsed at the highest level. The decoder
- * is unable to proceed. This error \ref SHOULD be treated as fatal to the
- * stream. */
- VPX_DEC_UNSUP_BITSTREAM = VPX_CODEC_UNSUP_BITSTREAM,
-
- /*!\brief Encoded bitstream uses an unsupported feature
- *
- * The decoder does not implement a feature required by the encoder. This
- * return code should only be used for features that prevent future
- * pictures from being properly decoded. This error \ref MAY be treated as
- * fatal to the stream or \ref MAY be treated as fatal to the current GOP.
- */
- VPX_DEC_UNSUP_FEATURE = VPX_CODEC_UNSUP_FEATURE,
-
- /*!\brief The coded data for this stream is corrupt or incomplete
- *
- * There was a problem decoding the current frame. This return code
- * should only be used for failures that prevent future pictures from
- * being properly decoded. This error \ref MAY be treated as fatal to the
- * stream or \ref MAY be treated as fatal to the current GOP. If decoding
- * is continued for the current GOP, artifacts may be present.
- */
- VPX_DEC_CORRUPT_FRAME = VPX_CODEC_CORRUPT_FRAME,
-
- /*!\brief An application-supplied parameter is not valid.
- *
- */
- VPX_DEC_INVALID_PARAM = VPX_CODEC_INVALID_PARAM,
-
- /*!\brief An iterator reached the end of list.
- *
- */
- VPX_DEC_LIST_END = VPX_CODEC_LIST_END
-
- }
- vpx_dec_err_t;
-
- /*! \brief Decoder capabilities bitfield
- *
- * Each decoder advertises the capabilities it supports as part of its
- * ::vpx_dec_iface_t interface structure. Capabilities are extra interfaces
- * or functionality, and are not required to be supported by a decoder.
- *
- * The available flags are specified by VPX_DEC_CAP_* defines.
- */
- typedef int vpx_dec_caps_t;
-#define VPX_DEC_CAP_PUT_SLICE 0x0001 /**< Will issue put_slice callbacks */
-#define VPX_DEC_CAP_PUT_FRAME 0x0002 /**< Will issue put_frame callbacks */
-#define VPX_DEC_CAP_XMA 0x0004 /**< Supports eXternal Memory Allocation */
-
- /*!\brief Stream properties
- *
- * This structure is used to query or set properties of the decoded
- * stream. Algorithms may extend this structure with data specific
- * to their bitstream by setting the sz member appropriately.
- */
-#if 1
- typedef vpx_codec_stream_info_t vpx_dec_stream_info_t;
-#else
- typedef struct
- {
- unsigned int sz; /**< Size of this structure */
- unsigned int w; /**< Width (or 0 for unknown/default) */
- unsigned int h; /**< Height (or 0 for unknown/default) */
- unsigned int is_kf; /**< Current frame is a keyframe */
- } vpx_dec_stream_info_t;
-#endif
-
-
- /*!\brief Decoder interface structure.
- *
- * Contains function pointers and other data private to the decoder
- * implementation. This structure is opaque to the application.
- */
- typedef const struct vpx_codec_iface vpx_dec_iface_t;
- typedef struct vpx_codec_priv vpx_dec_priv_t;
-
- /*!\brief Iterator
- *
- * Opaque storage used for iterating over lists.
- */
- typedef vpx_codec_iter_t vpx_dec_iter_t;
-
- /*!\brief Decoder context structure
- *
- * All decoders \ref MUST support this context structure fully. In general,
- * this data should be considered private to the decoder algorithm, and
- * not be manipulated or examined by the calling application. Applications
- * may reference the 'name' member to get a printable description of the
- * algorithm.
- */
-#if 1
- typedef vpx_codec_ctx_t vpx_dec_ctx_t;
-#else
- typedef struct
- {
- const char *name; /**< Printable interface name */
- vpx_dec_iface_t *iface; /**< Interface pointers */
- vpx_dec_err_t err; /**< Last returned error */
- vpx_dec_priv_t *priv; /**< Algorithm private storage */
- } vpx_dec_ctx_t;
-#endif
-
-
- /*!\brief Return the build configuration
- *
- * Returns a printable string containing an encoded version of the build
- * configuration. This may be useful to vpx support.
- *
- */
- const char *vpx_dec_build_config(void) DEPRECATED;
-
- /*!\brief Return the name for a given interface
- *
- * Returns a human readable string for name of the given decoder interface.
- *
- * \param[in] iface Interface pointer
- *
- */
- const char *vpx_dec_iface_name(vpx_dec_iface_t *iface) DEPRECATED;
-
-
- /*!\brief Convert error number to printable string
- *
- * Returns a human readable string for the last error returned by the
- * algorithm. The returned error will be one line and will not contain
- * any newline characters.
- *
- *
- * \param[in] err Error number.
- *
- */
- const char *vpx_dec_err_to_string(vpx_dec_err_t err) DEPRECATED;
-
-
- /*!\brief Retrieve error synopsis for decoder context
- *
- * Returns a human readable string for the last error returned by the
- * algorithm. The returned error will be one line and will not contain
- * any newline characters.
- *
- *
- * \param[in] ctx Pointer to this instance's context.
- *
- */
- const char *vpx_dec_error(vpx_dec_ctx_t *ctx) DEPRECATED;
-
-
- /*!\brief Retrieve detailed error information for decoder context
- *
- * Returns a human readable string providing detailed information about
- * the last error.
- *
- * \param[in] ctx Pointer to this instance's context.
- *
- * \retval NULL
- * No detailed information is available.
- */
- const char *vpx_dec_error_detail(vpx_dec_ctx_t *ctx) DEPRECATED;
-
-
- /* REQUIRED FUNCTIONS
- *
- * The following functions are required to be implemented for all decoders.
- * They represent the base case functionality expected of all decoders.
- */
-
-
- /*!\brief Initialize a decoder instance
- *
- * Initializes a decoder context using the given interface. Applications
- * should call the vpx_dec_init convenience macro instead of this
- * function directly, to ensure that the ABI version number parameter
- * is properly initialized.
- *
- * \param[in] ctx Pointer to this instance's context.
- * \param[in] iface Pointer to the algorithm interface to use.
- * \param[in] ver ABI version number. Must be set to
- * VPX_DECODER_ABI_VERSION
- * \retval #VPX_DEC_OK
- * The decoder algorithm initialized.
- * \retval #VPX_DEC_MEM_ERROR
- * Memory allocation failed.
- */
- vpx_dec_err_t vpx_dec_init_ver(vpx_dec_ctx_t *ctx,
- vpx_dec_iface_t *iface,
- int ver) DEPRECATED;
-#define vpx_dec_init(ctx, iface) \
- vpx_dec_init_ver(ctx, iface, VPX_DECODER_ABI_VERSION)
-
-
- /*!\brief Destroy a decoder instance
- *
- * Destroys a decoder context, freeing any associated memory buffers.
- *
- * \param[in] ctx Pointer to this instance's context
- *
- * \retval #VPX_DEC_OK
- * The decoder algorithm initialized.
- * \retval #VPX_DEC_MEM_ERROR
- * Memory allocation failed.
- */
- vpx_dec_err_t vpx_dec_destroy(vpx_dec_ctx_t *ctx) DEPRECATED;
-
-
- /*!\brief Get the capabilities of an algorithm.
- *
- * Retrieves the capabilities bitfield from the algorithm's interface.
- *
- * \param[in] iface Pointer to the algorithm interface
- *
- */
- vpx_dec_caps_t vpx_dec_get_caps(vpx_dec_iface_t *iface) DEPRECATED;
-
-
- /*!\brief Parse stream info from a buffer
- *
- * Performs high level parsing of the bitstream. Construction of a decoder
- * context is not necessary. Can be used to determine if the bitstream is
- * of the proper format, and to extract information from the stream.
- *
- * \param[in] iface Pointer to the algorithm interface
- * \param[in] data Pointer to a block of data to parse
- * \param[in] data_sz Size of the data buffer
- * \param[in,out] si Pointer to stream info to update. The size member
- * \ref MUST be properly initialized, but \ref MAY be
- * clobbered by the algorithm. This parameter \ref MAY
- * be NULL.
- *
- * \retval #VPX_DEC_OK
- * Bitstream is parsable and stream information updated
- */
- vpx_dec_err_t vpx_dec_peek_stream_info(vpx_dec_iface_t *iface,
- const uint8_t *data,
- unsigned int data_sz,
- vpx_dec_stream_info_t *si) DEPRECATED;
-
-
- /*!\brief Return information about the current stream.
- *
- * Returns information about the stream that has been parsed during decoding.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in,out] si Pointer to stream info to update. The size member
- * \ref MUST be properly initialized, but \ref MAY be
- * clobbered by the algorithm. This parameter \ref MAY
- * be NULL.
- *
- * \retval #VPX_DEC_OK
- * Bitstream is parsable and stream information updated
- */
- vpx_dec_err_t vpx_dec_get_stream_info(vpx_dec_ctx_t *ctx,
- vpx_dec_stream_info_t *si) DEPRECATED;
-
-
- /*!\brief Control algorithm
- *
- * This function is used to exchange algorithm specific data with the decoder
- * instance. This can be used to implement features specific to a particular
- * algorithm.
- *
- * This wrapper function dispatches the request to the helper function
- * associated with the given ctrl_id. It tries to call this function
- * transparently, but will return #VPX_DEC_ERROR if the request could not
- * be dispatched.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] ctrl_id Algorithm specific control identifier
- * \param[in,out] data Data to exchange with algorithm instance.
- *
- * \retval #VPX_DEC_OK
- * The control request was processed.
- * \retval #VPX_DEC_ERROR
- * The control request was not processed.
- * \retval #VPX_DEC_INVALID_PARAM
- * The data was not valid.
- */
- vpx_dec_err_t vpx_dec_control(vpx_dec_ctx_t *ctx,
- int ctrl_id,
- void *data) DEPRECATED;
-
- /*!\brief Decode data
- *
- * Processes a buffer of coded data. If the processing results in a new
- * decoded frame becoming available, #VPX_DEC_CB_PUT_SLICE and
- * #VPX_DEC_CB_PUT_FRAME events may be generated, as appropriate. Encoded data
- * \ref MUST be passed in DTS (decode time stamp) order. Frames produced will
- * always be in PTS (presentation time stamp) order.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] data Pointer to this block of new coded data. If
- * NULL, a VPX_DEC_CB_PUT_FRAME event is posted
- * for the previously decoded frame.
- * \param[in] data_sz Size of the coded data, in bytes.
- * \param[in] user_priv Application specific data to associate with
- * this frame.
- * \param[in] rel_pts PTS relative to the previous frame, in us. If
- * unknown or unavailable, set to zero.
- *
- * \return Returns #VPX_DEC_OK if the coded data was processed completely
- * and future pictures can be decoded without error. Otherwise,
- * see the descriptions of the other error codes in ::vpx_dec_err_t
- * for recoverability capabilities.
- */
- vpx_dec_err_t vpx_dec_decode(vpx_dec_ctx_t *ctx,
- uint8_t *data,
- unsigned int data_sz,
- void *user_priv,
- int rel_pts) DEPRECATED;
-
-
- /*!\brief Decoded frames iterator
- *
- * Iterates over a list of the frames available for display. The iterator
- * storage should be initialized to NULL to start the iteration. Iteration is
- * complete when this function returns NULL.
- *
- * The list of available frames becomes valid upon completion of the
- * vpx_dec_decode call, and remains valid until the next call to vpx_dec_decode.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in out] iter Iterator storage, initialized to NULL
- *
- * \return Returns a pointer to an image, if one is ready for display. Frames
- * produced will always be in PTS (presentation time stamp) order.
- */
- vpx_image_t *vpx_dec_get_frame(vpx_dec_ctx_t *ctx,
- vpx_dec_iter_t *iter) DEPRECATED;
-
-
- /*!\defgroup cap_put_frame Frame-Based Decoding Functions
- *
- * The following functions are required to be implemented for all decoders
- * that advertise the VPX_DEC_CAP_PUT_FRAME capability. Calling these functions
- * for codecs that don't advertise this capability will result in an error
- * code being returned, usually VPX_DEC_ERROR
- * @{
- */
-
- /*!\brief put frame callback prototype
- *
- * This callback is invoked by the decoder to notify the application of
- * the availability of decoded image data.
- */
- typedef void (*vpx_dec_put_frame_cb_fn_t)(void *user_priv,
- const vpx_image_t *img);
-
-
- /*!\brief Register for notification of frame completion.
- *
- * Registers a given function to be called when a decoded frame is
- * available.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] cb Pointer to the callback function
- * \param[in] user_priv User's private data
- *
- * \retval #VPX_DEC_OK
- * Callback successfully registered.
- * \retval #VPX_DEC_ERROR
- * Decoder context not initialized, or algorithm not capable of
- * posting slice completion.
- */
- vpx_dec_err_t vpx_dec_register_put_frame_cb(vpx_dec_ctx_t *ctx,
- vpx_dec_put_frame_cb_fn_t cb,
- void *user_priv) DEPRECATED;
-
-
- /*!@} - end defgroup cap_put_frame */
-
- /*!\defgroup cap_put_slice Slice-Based Decoding Functions
- *
- * The following functions are required to be implemented for all decoders
- * that advertise the VPX_DEC_CAP_PUT_SLICE capability. Calling these functions
- * for codecs that don't advertise this capability will result in an error
- * code being returned, usually VPX_DEC_ERROR
- * @{
- */
-
- /*!\brief put slice callback prototype
- *
- * This callback is invoked by the decoder to notify the application of
- * the availability of partially decoded image data. The
- */
- typedef void (*vpx_dec_put_slice_cb_fn_t)(void *user_priv,
- const vpx_image_t *img,
- const vpx_image_rect_t *valid,
- const vpx_image_rect_t *update);
-
-
- /*!\brief Register for notification of slice completion.
- *
- * Registers a given function to be called when a decoded slice is
- * available.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] cb Pointer to the callback function
- * \param[in] user_priv User's private data
- *
- * \retval #VPX_DEC_OK
- * Callback successfully registered.
- * \retval #VPX_DEC_ERROR
- * Decoder context not initialized, or algorithm not capable of
- * posting slice completion.
- */
- vpx_dec_err_t vpx_dec_register_put_slice_cb(vpx_dec_ctx_t *ctx,
- vpx_dec_put_slice_cb_fn_t cb,
- void *user_priv) DEPRECATED;
-
-
- /*!@} - end defgroup cap_put_slice*/
-
- /*!\defgroup cap_xma External Memory Allocation Functions
- *
- * The following functions are required to be implemented for all decoders
- * that advertise the VPX_DEC_CAP_XMA capability. Calling these functions
- * for codecs that don't advertise this capability will result in an error
- * code being returned, usually VPX_DEC_ERROR
- * @{
- */
-
- /*!\brief Memory Map Entry
- *
- * This structure is used to contain the properties of a memory segment. It
- * is populated by the decoder in the request phase, and by the calling
- * application once the requested allocation has been performed.
- */
-#if 1
-#define VPX_DEC_MEM_ZERO 0x1 /**< Segment must be zeroed by allocation */
-#define VPX_DEC_MEM_WRONLY 0x2 /**< Segment need not be readable */
-#define VPX_DEC_MEM_FAST 0x4 /**< Place in fast memory, if available */
- typedef struct vpx_codec_mmap vpx_dec_mmap_t;
-#else
- typedef struct vpx_dec_mmap
- {
- /*
- * The following members are set by the codec when requesting a segment
- */
- unsigned int id; /**< identifier for the segment's contents */
- unsigned long sz; /**< size of the segment, in bytes */
- unsigned int align; /**< required alignment of the segment, in bytes */
- unsigned int flags; /**< bitfield containing segment properties */
-#define VPX_DEC_MEM_ZERO 0x1 /**< Segment must be zeroed by allocation */
-#define VPX_DEC_MEM_WRONLY 0x2 /**< Segment need not be readable */
-#define VPX_DEC_MEM_FAST 0x4 /**< Place in fast memory, if available */
-
- /* The following members are to be filled in by the allocation function */
- void *base; /**< pointer to the allocated segment */
- void (*dtor)(struct vpx_dec_mmap *map); /**< destructor to call */
- void *priv; /**< allocator private storage */
- } vpx_dec_mmap_t;
-#endif
-
- /*!\brief Initialize a decoder instance in external allocation mode
- *
- * Initializes a decoder context using the given interface. Applications
- * should call the vpx_dec_xma_init convenience macro instead of this
- * function directly, to ensure that the ABI version number parameter
- * is properly initialized.
- *
- * \param[in] ctx Pointer to this instance's context.
- * \param[in] iface Pointer to the algorithm interface to use.
- * \param[in] ver ABI version number. Must be set to
- * VPX_DECODER_ABI_VERSION
- * \retval #VPX_DEC_OK
- * The decoder algorithm initialized.
- * \retval #VPX_DEC_ERROR
- * Decoder does not support XMA mode.
- */
- vpx_dec_err_t vpx_dec_xma_init_ver(vpx_dec_ctx_t *ctx,
- vpx_dec_iface_t *iface,
- int ver) DEPRECATED;
-#define vpx_dec_xma_init(ctx, iface) \
- vpx_dec_xma_init_ver(ctx, iface, VPX_DECODER_ABI_VERSION)
-
-
- /*!\brief Iterate over the list of segments to allocate.
- *
- * Iterates over a list of the segments to allocate. The iterator storage
- * should be initialized to NULL to start the iteration. Iteration is complete
- * when this function returns VPX_DEC_LIST_END. The amount of memory needed to
- * allocate is dependent upon the size of the encoded stream. This means that
- * the stream info structure must be known at allocation time. It can be
- * populated with the vpx_dec_peek_stream_info() function. In cases where the
- * stream to be decoded is not available at allocation time, a fixed size must
- * be requested. The decoder will not be able to decode streams larger than
- * the size used at allocation time.
- *
- * \param[in] ctx Pointer to this instance's context.
- * \param[out] mmap Pointer to the memory map entry to populate.
- * \param[in] si Pointer to the stream info.
- * \param[in out] iter Iterator storage, initialized to NULL
- *
- * \retval #VPX_DEC_OK
- * The memory map entry was populated.
- * \retval #VPX_DEC_ERROR
- * Decoder does not support XMA mode.
- * \retval #VPX_DEC_MEM_ERROR
- * Unable to determine segment size from stream info.
- */
- vpx_dec_err_t vpx_dec_get_mem_map(vpx_dec_ctx_t *ctx,
- vpx_dec_mmap_t *mmap,
- const vpx_dec_stream_info_t *si,
- vpx_dec_iter_t *iter) DEPRECATED;
-
-
- /*!\brief Identify allocated segments to decoder instance
- *
- * Stores a list of allocated segments in the decoder. Segments \ref MUST be
- * passed in the order they are read from vpx_dec_get_mem_map(), but may be
- * passed in groups of any size. Segments \ref MUST be set only once. The
- * allocation function \ref MUST ensure that the vpx_dec_mmap_t::base member
- * is non-NULL. If the segment requires cleanup handling (e.g., calling free()
- * or close()) then the vpx_dec_mmap_t::dtor member \ref MUST be populated.
- *
- * \param[in] ctx Pointer to this instance's context.
- * \param[in] mmaps Pointer to the first memory map entry in the list.
- * \param[in] num_maps Number of entries being set at this time
- *
- * \retval #VPX_DEC_OK
- * The segment was stored in the decoder context.
- * \retval #VPX_DEC_ERROR
- * Decoder does not support XMA mode.
- * \retval #VPX_DEC_MEM_ERROR
- * Segment base address was not set, or segment was already stored.
-
- */
- vpx_dec_err_t vpx_dec_set_mem_map(vpx_dec_ctx_t *ctx,
- vpx_dec_mmap_t *mmaps,
- unsigned int num_maps) DEPRECATED;
-
- /*!@} - end defgroup cap_xma*/
- /*!@} - end defgroup decoder*/
-
-
-#endif
-#ifdef __cplusplus
-}
-#endif
diff --git a/vpx/vpx_encoder.h b/vpx/vpx_encoder.h
deleted file mode 100644
index 9c44414..0000000
--- a/vpx/vpx_encoder.h
+++ /dev/null
@@ -1,794 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\defgroup encoder Encoder Algorithm Interface
- * \ingroup codec
- * This abstraction allows applications using this encoder to easily support
- * multiple video formats with minimal code duplication. This section describes
- * the interface common to all encoders.
- * @{
- */
-
-/*!\file
- * \brief Describes the encoder algorithm interface to applications.
- *
- * This file describes the interface between an application and a
- * video encoder algorithm.
- *
- */
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#ifndef VPX_ENCODER_H
-#define VPX_ENCODER_H
-#include "vpx_codec.h"
-
-
- /*!\brief Current ABI version number
- *
- * \internal
- * If this file is altered in any way that changes the ABI, this value
- * must be bumped. Examples include, but are not limited to, changing
- * types, removing or reassigning enums, adding/removing/rearranging
- * fields to structures
- */
-#define VPX_ENCODER_ABI_VERSION (2 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/
-
-
- /*! \brief Encoder capabilities bitfield
- *
- * Each encoder advertises the capabilities it supports as part of its
- * ::vpx_codec_iface_t interface structure. Capabilities are extra
- * interfaces or functionality, and are not required to be supported
- * by an encoder.
- *
- * The available flags are specified by VPX_CODEC_CAP_* defines.
- */
-#define VPX_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */
-
-
- /*! \brief Initialization-time Feature Enabling
- *
- * Certain codec features must be known at initialization time, to allow
- * for proper memory allocation.
- *
- * The available flags are specified by VPX_CODEC_USE_* defines.
- */
-#define VPX_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */
-
-
- /*!\brief Generic fixed size buffer structure
- *
- * This structure is able to hold a reference to any fixed size buffer.
- */
- typedef struct vpx_fixed_buf
- {
- void *buf; /**< Pointer to the data */
- size_t sz; /**< Length of the buffer, in chars */
- } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */
-
-
- /*!\brief Time Stamp Type
- *
- * An integer, which when multiplied by the stream's time base, provides
- * the absolute time of a sample.
- */
- typedef int64_t vpx_codec_pts_t;
-
-
- /*!\brief Compressed Frame Flags
- *
- * This type represents a bitfield containing information about a compressed
- * frame that may be useful to an application. The most significant 16 bits
- * can be used by an algorithm to provide additional detail, for example to
- * support frame types that are codec specific (MPEG-1 D-frames for example)
- */
- typedef uint32_t vpx_codec_frame_flags_t;
-#define VPX_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */
-#define VPX_FRAME_IS_DROPPABLE 0x2 /**< frame can be dropped without affecting
- the stream (no future frame depends on
- this one) */
-#define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not
- be shown */
-
-
- /*!\brief Encoder output packet variants
- *
- * This enumeration lists the different kinds of data packets that can be
- * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY
- * extend this list to provide additional functionality.
- */
- enum vpx_codec_cx_pkt_kind
- {
- VPX_CODEC_CX_FRAME_PKT, /**< Compressed video frame */
- VPX_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */
- VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */
- VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */
- };
-
-
- /*!\brief Encoder output packet
- *
- * This structure contains the different kinds of output data the encoder
- * may produce while compressing a frame.
- */
- typedef struct vpx_codec_cx_pkt
- {
- enum vpx_codec_cx_pkt_kind kind; /**< packet variant */
- union
- {
- struct
- {
- void *buf; /**< compressed data buffer */
- size_t sz; /**< length of compressed data */
- vpx_codec_pts_t pts; /**< time stamp to show frame
- (in timebase units) */
- unsigned long duration; /**< duration to show frame
- (in timebase units) */
- vpx_codec_frame_flags_t flags; /**< flags for this frame */
- } frame; /**< data for compressed frame packet */
- struct vpx_fixed_buf twopass_stats; /**< data for two-pass packet */
- struct vpx_psnr_pkt
- {
- unsigned int samples[4]; /**< Number of samples, total/y/u/v */
- uint64_t sse[4]; /**< sum squared error, total/y/u/v */
- double psnr[4]; /**< PSNR, total/y/u/v */
- } psnr; /**< data for PSNR packet */
- struct vpx_fixed_buf raw; /**< data for arbitrary packets */
-
- /* This packet size is fixed to allow codecs to extend this
- * interface without having to manage storage for raw packets,
- * i.e., if it's smaller than 128 bytes, you can store in the
- * packet list directly.
- */
- char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */
- } data; /**< packet data */
- } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */
-
-
- /*!\brief Rational Number
- *
- * This structure holds a fractional value.
- */
- typedef struct vpx_rational
- {
- int num; /**< fraction numerator */
- int den; /**< fraction denominator */
- } vpx_rational_t; /**< alias for struct vpx_rational */
-
-
- /*!\brief Multi-pass Encoding Pass */
- enum vpx_enc_pass
- {
- VPX_RC_ONE_PASS, /**< Single pass mode */
- VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */
- VPX_RC_LAST_PASS /**< Final pass of multi-pass mode */
- };
-
-
- /*!\brief Rate control mode */
- enum vpx_rc_mode
- {
- VPX_VBR, /**< Variable Bit Rate (VBR) mode */
- VPX_CBR, /**< Constant Bit Rate (CBR) mode */
- VPX_CQ /**< Constant Quality (CQ) mode */
- };
-
-
- /*!\brief Keyframe placement mode.
- *
- * This enumeration determines whether keyframes are placed automatically by
- * the encoder or whether this behavior is disabled. Older releases of this
- * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled.
- * This name is confusing for this behavior, so the new symbols to be used
- * are VPX_KF_AUTO and VPX_KF_DISABLED.
- */
- enum vpx_kf_mode
- {
- VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */
- VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */
- VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */
- };
-
-
- /*!\brief Encoded Frame Flags
- *
- * This type indicates a bitfield to be passed to vpx_codec_encode(), defining
- * per-frame boolean values. By convention, bits common to all codecs will be
- * named VPX_EFLAG_*, and bits specific to an algorithm will be named
- * /algo/_eflag_*. The lower order 16 bits are reserved for common use.
- */
- typedef long vpx_enc_frame_flags_t;
-#define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */
-
-
- /*!\brief Encoder configuration structure
- *
- * This structure contains the encoder settings that have common representations
- * across all codecs. This doesn't imply that all codecs support all features,
- * however.
- */
- typedef struct vpx_codec_enc_cfg
- {
- /*
- * generic settings (g)
- */
-
- /*!\brief Algorithm specific "usage" value
- *
- * Algorithms may define multiple values for usage, which may convey the
- * intent of how the application intends to use the stream. If this value
- * is non-zero, consult the documentation for the codec to determine its
- * meaning.
- */
- unsigned int g_usage;
-
-
- /*!\brief Maximum number of threads to use
- *
- * For multi-threaded implementations, use no more than this number of
- * threads. The codec may use fewer threads than allowed. The value
- * 0 is equivalent to the value 1.
- */
- unsigned int g_threads;
-
-
- /*!\brief Bitstream profile to use
- *
- * Some codecs support a notion of multiple bitstream profiles. Typically
- * this maps to a set of features that are turned on or off. Often the
- * profile to use is determined by the features of the intended decoder.
- * Consult the documentation for the codec to determine the valid values
- * for this parameter, or set to zero for a sane default.
- */
- unsigned int g_profile; /**< profile of bitstream to use */
-
-
-
- /*!\brief Width of the frame
- *
- * This value identifies the presentation resolution of the frame,
- * in pixels. Note that the frames passed as input to the encoder must
- * have this resolution. Frames will be presented by the decoder in this
- * resolution, independent of any spatial resampling the encoder may do.
- */
- unsigned int g_w;
-
-
- /*!\brief Height of the frame
- *
- * This value identifies the presentation resolution of the frame,
- * in pixels. Note that the frames passed as input to the encoder must
- * have this resolution. Frames will be presented by the decoder in this
- * resolution, independent of any spatial resampling the encoder may do.
- */
- unsigned int g_h;
-
-
- /*!\brief Stream timebase units
- *
- * Indicates the smallest interval of time, in seconds, used by the stream.
- * For fixed frame rate material, or variable frame rate material where
- * frames are timed at a multiple of a given clock (ex: video capture),
- * the \ref RECOMMENDED method is to set the timebase to the reciprocal
- * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the
- * pts to correspond to the frame number, which can be handy. For
- * re-encoding video from containers with absolute time timestamps, the
- * \ref RECOMMENDED method is to set the timebase to that of the parent
- * container or multimedia framework (ex: 1/1000 for ms, as in FLV).
- */
- struct vpx_rational g_timebase;
-
-
- /*!\brief Enable error resilient mode.
- *
- * Error resilient mode indicates to the encoder that it should take
- * measures appropriate for streaming over lossy or noisy links, if
- * possible. Set to 1 to enable this feature, 0 to disable it.
- */
- unsigned int g_error_resilient;
-
-
- /*!\brief Multi-pass Encoding Mode
- *
- * This value should be set to the current phase for multi-pass encoding.
- * For single pass, set to #VPX_RC_ONE_PASS.
- */
- enum vpx_enc_pass g_pass;
-
-
- /*!\brief Allow lagged encoding
- *
- * If set, this value allows the encoder to consume a number of input
- * frames before producing output frames. This allows the encoder to
- * base decisions for the current frame on future frames. This does
- * increase the latency of the encoding pipeline, so it is not appropriate
- * in all situations (ex: realtime encoding).
- *
- * Note that this is a maximum value -- the encoder may produce frames
- * sooner than the given limit. Set this value to 0 to disable this
- * feature.
- */
- unsigned int g_lag_in_frames;
-
-
- /*
- * rate control settings (rc)
- */
-
- /*!\brief Temporal resampling configuration, if supported by the codec.
- *
- * Temporal resampling allows the codec to "drop" frames as a strategy to
- * meet its target data rate. This can cause temporal discontinuities in
- * the encoded video, which may appear as stuttering during playback. This
- * trade-off is often acceptable, but for many applications is not. It can
- * be disabled in these cases.
- *
- * Note that not all codecs support this feature. All vpx VPx codecs do.
- * For other codecs, consult the documentation for that algorithm.
- *
- * This threshold is described as a percentage of the target data buffer.
- * When the data buffer falls below this percentage of fullness, a
- * dropped frame is indicated. Set the threshold to zero (0) to disable
- * this feature.
- */
- unsigned int rc_dropframe_thresh;
-
-
- /*!\brief Enable/disable spatial resampling, if supported by the codec.
- *
- * Spatial resampling allows the codec to compress a lower resolution
- * version of the frame, which is then upscaled by the encoder to the
- * correct presentation resolution. This increases visual quality at
- * low data rates, at the expense of CPU time on the encoder/decoder.
- */
- unsigned int rc_resize_allowed;
-
-
- /*!\brief Spatial resampling up watermark.
- *
- * This threshold is described as a percentage of the target data buffer.
- * When the data buffer rises above this percentage of fullness, the
- * encoder will step up to a higher resolution version of the frame.
- */
- unsigned int rc_resize_up_thresh;
-
-
- /*!\brief Spatial resampling down watermark.
- *
- * This threshold is described as a percentage of the target data buffer.
- * When the data buffer falls below this percentage of fullness, the
- * encoder will step down to a lower resolution version of the frame.
- */
- unsigned int rc_resize_down_thresh;
-
-
- /*!\brief Rate control algorithm to use.
- *
- * Indicates whether the end usage of this stream is to be streamed over
- * a bandwidth constrained link, indicating that Constant Bit Rate (CBR)
- * mode should be used, or whether it will be played back on a high
- * bandwidth link, as from a local disk, where higher variations in
- * bitrate are acceptable.
- */
- enum vpx_rc_mode rc_end_usage;
-
-
- /*!\brief Two-pass stats buffer.
- *
- * A buffer containing all of the stats packets produced in the first
- * pass, concatenated.
- */
- struct vpx_fixed_buf rc_twopass_stats_in;
-
-
- /*!\brief Target data rate
- *
- * Target bandwidth to use for this stream, in kilobits per second.
- */
- unsigned int rc_target_bitrate;
-
-
- /*
- * quantizer settings
- */
-
-
- /*!\brief Minimum (Best Quality) Quantizer
- *
- * The quantizer is the most direct control over the quality of the
- * encoded image. The range of valid values for the quantizer is codec
- * specific. Consult the documentation for the codec to determine the
- * values to use. To determine the range programmatically, call
- * vpx_codec_enc_config_default() with a usage value of 0.
- */
- unsigned int rc_min_quantizer;
-
-
- /*!\brief Maximum (Worst Quality) Quantizer
- *
- * The quantizer is the most direct control over the quality of the
- * encoded image. The range of valid values for the quantizer is codec
- * specific. Consult the documentation for the codec to determine the
- * values to use. To determine the range programmatically, call
- * vpx_codec_enc_config_default() with a usage value of 0.
- */
- unsigned int rc_max_quantizer;
-
-
- /*
- * bitrate tolerance
- */
-
-
- /*!\brief Rate control undershoot tolerance
- *
- * This value, expressed as a percentage of the target bitrate, describes
- * the target bitrate for easier frames, allowing bits to be saved for
- * harder frames. Set to zero to use the codec default.
- */
- unsigned int rc_undershoot_pct;
-
-
- /*!\brief Rate control overshoot tolerance
- *
- * This value, expressed as a percentage of the target bitrate, describes
- * the maximum allowed bitrate for a given frame. Set to zero to use the
- * codec default.
- */
- unsigned int rc_overshoot_pct;
-
-
- /*
- * decoder buffer model parameters
- */
-
-
- /*!\brief Decoder Buffer Size
- *
- * This value indicates the amount of data that may be buffered by the
- * decoding application. Note that this value is expressed in units of
- * time (milliseconds). For example, a value of 5000 indicates that the
- * client will buffer (at least) 5000ms worth of encoded data. Use the
- * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if
- * necessary.
- */
- unsigned int rc_buf_sz;
-
-
- /*!\brief Decoder Buffer Initial Size
- *
- * This value indicates the amount of data that will be buffered by the
- * decoding application prior to beginning playback. This value is
- * expressed in units of time (milliseconds). Use the target bitrate
- * (#rc_target_bitrate) to convert to bits/bytes, if necessary.
- */
- unsigned int rc_buf_initial_sz;
-
-
- /*!\brief Decoder Buffer Optimal Size
- *
- * This value indicates the amount of data that the encoder should try
- * to maintain in the decoder's buffer. This value is expressed in units
- * of time (milliseconds). Use the target bitrate (#rc_target_bitrate)
- * to convert to bits/bytes, if necessary.
- */
- unsigned int rc_buf_optimal_sz;
-
-
- /*
- * 2 pass rate control parameters
- */
-
-
- /*!\brief Two-pass mode CBR/VBR bias
- *
- * Bias, expressed on a scale of 0 to 100, for determining target size
- * for the current frame. The value 0 indicates the optimal CBR mode
- * value should be used. The value 100 indicates the optimal VBR mode
- * value should be used. Values in between indicate which way the
- * encoder should "lean."
- */
- unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */
-
-
- /*!\brief Two-pass mode per-GOP minimum bitrate
- *
- * This value, expressed as a percentage of the target bitrate, indicates
- * the minimum bitrate to be used for a single GOP (aka "section")
- */
- unsigned int rc_2pass_vbr_minsection_pct;
-
-
- /*!\brief Two-pass mode per-GOP maximum bitrate
- *
- * This value, expressed as a percentage of the target bitrate, indicates
- * the maximum bitrate to be used for a single GOP (aka "section")
- */
- unsigned int rc_2pass_vbr_maxsection_pct;
-
-
- /*
- * keyframing settings (kf)
- */
-
- /*!\brief Keyframe placement mode
- *
- * This value indicates whether the encoder should place keyframes at a
- * fixed interval, or determine the optimal placement automatically
- * (as governed by the #kf_min_dist and #kf_max_dist parameters)
- */
- enum vpx_kf_mode kf_mode;
-
-
- /*!\brief Keyframe minimum interval
- *
- * This value, expressed as a number of frames, prevents the encoder from
- * placing a keyframe nearer than kf_min_dist to the previous keyframe. At
- * least kf_min_dist frames non-keyframes will be coded before the next
- * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval.
- */
- unsigned int kf_min_dist;
-
-
- /*!\brief Keyframe maximum interval
- *
- * This value, expressed as a number of frames, forces the encoder to code
- * a keyframe if one has not been coded in the last kf_max_dist frames.
- * A value of 0 implies all frames will be keyframes. Set kf_min_dist
- * equal to kf_max_dist for a fixed interval.
- */
- unsigned int kf_max_dist;
-
- } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */
-
-
- /*!\brief Initialize an encoder instance
- *
- * Initializes a encoder context using the given interface. Applications
- * should call the vpx_codec_enc_init convenience macro instead of this
- * function directly, to ensure that the ABI version number parameter
- * is properly initialized.
- *
- * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
- * parameter), the storage pointed to by the cfg parameter must be
- * kept readable and stable until all memory maps have been set.
- *
- * \param[in] ctx Pointer to this instance's context.
- * \param[in] iface Pointer to the algorithm interface to use.
- * \param[in] cfg Configuration to use, if known. May be NULL.
- * \param[in] flags Bitfield of VPX_CODEC_USE_* flags
- * \param[in] ver ABI version number. Must be set to
- * VPX_ENCODER_ABI_VERSION
- * \retval #VPX_CODEC_OK
- * The decoder algorithm initialized.
- * \retval #VPX_CODEC_MEM_ERROR
- * Memory allocation failed.
- */
- vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx,
- vpx_codec_iface_t *iface,
- vpx_codec_enc_cfg_t *cfg,
- vpx_codec_flags_t flags,
- int ver);
-
-
- /*!\brief Convenience macro for vpx_codec_enc_init_ver()
- *
- * Ensures the ABI version parameter is properly set.
- */
-#define vpx_codec_enc_init(ctx, iface, cfg, flags) \
- vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION)
-
-
- /*!\brief Get a default configuration
- *
- * Initializes a encoder configuration structure with default values. Supports
- * the notion of "usages" so that an algorithm may offer different default
- * settings depending on the user's intended goal. This function \ref SHOULD
- * be called by all applications to initialize the configuration structure
- * before specializing the configuration with application specific values.
- *
- * \param[in] iface Pointer to the algorithm interface to use.
- * \param[out] cfg Configuration buffer to populate
- * \param[in] usage End usage. Set to 0 or use codec specific values.
- *
- * \retval #VPX_CODEC_OK
- * The configuration was populated.
- * \retval #VPX_CODEC_INCAPABLE
- * Interface is not an encoder interface.
- * \retval #VPX_CODEC_INVALID_PARAM
- * A parameter was NULL, or the usage value was not recognized.
- */
- vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface,
- vpx_codec_enc_cfg_t *cfg,
- unsigned int usage);
-
-
- /*!\brief Set or change configuration
- *
- * Reconfigures an encoder instance according to the given configuration.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] cfg Configuration buffer to use
- *
- * \retval #VPX_CODEC_OK
- * The configuration was populated.
- * \retval #VPX_CODEC_INCAPABLE
- * Interface is not an encoder interface.
- * \retval #VPX_CODEC_INVALID_PARAM
- * A parameter was NULL, or the usage value was not recognized.
- */
- vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx,
- const vpx_codec_enc_cfg_t *cfg);
-
-
- /*!\brief Get global stream headers
- *
- * Retrieves a stream level global header packet, if supported by the codec.
- *
- * \param[in] ctx Pointer to this instance's context
- *
- * \retval NULL
- * Encoder does not support global header
- * \retval Non-NULL
- * Pointer to buffer containing global header packet
- */
- vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx);
-
-
-#define VPX_DL_REALTIME (1) /**< deadline parameter analogous to
- * VPx REALTIME mode. */
-#define VPX_DL_GOOD_QUALITY (1000000) /**< deadline parameter analogous to
- * VPx GOOD QUALITY mode. */
-#define VPX_DL_BEST_QUALITY (0) /**< deadline parameter analogous to
- * VPx BEST QUALITY mode. */
- /*!\brief Encode a frame
- *
- * Encodes a video frame at the given "presentation time." The presentation
- * time stamp (PTS) \ref MUST be strictly increasing.
- *
- * The encoder supports the notion of a soft real-time deadline. Given a
- * non-zero value to the deadline parameter, the encoder will make a "best
- * effort" guarantee to return before the given time slice expires. It is
- * implicit that limiting the available time to encode will degrade the
- * output quality. The encoder can be given an unlimited time to produce the
- * best possible frame by specifying a deadline of '0'. This deadline
- * supercedes the VPx notion of "best quality, good quality, realtime".
- * Applications that wish to map these former settings to the new deadline
- * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY,
- * and #VPX_DL_BEST_QUALITY.
- *
- * When the last frame has been passed to the encoder, this function should
- * continue to be called, with the img parameter set to NULL. This will
- * signal the end-of-stream condition to the encoder and allow it to encode
- * any held buffers. Encoding is complete when vpx_codec_encode() is called
- * and vpx_codec_get_cx_data() returns no data.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] img Image data to encode, NULL to flush.
- * \param[in] pts Presentation time stamp, in timebase units.
- * \param[in] duration Duration to show frame, in timebase units.
- * \param[in] flags Flags to use for encoding this frame.
- * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite)
- *
- * \retval #VPX_CODEC_OK
- * The configuration was populated.
- * \retval #VPX_CODEC_INCAPABLE
- * Interface is not an encoder interface.
- * \retval #VPX_CODEC_INVALID_PARAM
- * A parameter was NULL, the image format is unsupported, etc.
- */
- vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx,
- const vpx_image_t *img,
- vpx_codec_pts_t pts,
- unsigned long duration,
- vpx_enc_frame_flags_t flags,
- unsigned long deadline);
-
-
- /*!\brief Set compressed data output buffer
- *
- * Sets the buffer that the codec should output the compressed data
- * into. This call effectively sets the buffer pointer returned in the
- * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be
- * appended into this buffer. The buffer is preserved across frames,
- * so applications must periodically call this function after flushing
- * the accumulated compressed data to disk or to the network to reset
- * the pointer to the buffer's head.
- *
- * `pad_before` bytes will be skipped before writing the compressed
- * data, and `pad_after` bytes will be appended to the packet. The size
- * of the packet will be the sum of the size of the actual compressed
- * data, pad_before, and pad_after. The padding bytes will be preserved
- * (not overwritten).
- *
- * Note that calling this function does not guarantee that the returned
- * compressed data will be placed into the specified buffer. In the
- * event that the encoded data will not fit into the buffer provided,
- * the returned packet \ref MAY point to an internal buffer, as it would
- * if this call were never used. In this event, the output packet will
- * NOT have any padding, and the application must free space and copy it
- * to the proper place. This is of particular note in configurations
- * that may output multiple packets for a single encoded frame (e.g., lagged
- * encoding) or if the application does not reset the buffer periodically.
- *
- * Applications may restore the default behavior of the codec providing
- * the compressed data buffer by calling this function with a NULL
- * buffer.
- *
- * Applications \ref MUSTNOT call this function during iteration of
- * vpx_codec_get_cx_data().
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in] buf Buffer to store compressed data into
- * \param[in] pad_before Bytes to skip before writing compressed data
- * \param[in] pad_after Bytes to skip after writing compressed data
- *
- * \retval #VPX_CODEC_OK
- * The buffer was set successfully.
- * \retval #VPX_CODEC_INVALID_PARAM
- * A parameter was NULL, the image format is unsupported, etc.
- */
- vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx,
- const vpx_fixed_buf_t *buf,
- unsigned int pad_before,
- unsigned int pad_after);
-
-
- /*!\brief Encoded data iterator
- *
- * Iterates over a list of data packets to be passed from the encoder to the
- * application. The different kinds of packets available are enumerated in
- * #vpx_codec_cx_pkt_kind.
- *
- * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's
- * muxer. Multiple compressed frames may be in the list.
- * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer.
- *
- * The application \ref MUST silently ignore any packet kinds that it does
- * not recognize or support.
- *
- * The data buffers returned from this function are only guaranteed to be
- * valid until the application makes another call to any vpx_codec_* function.
- *
- * \param[in] ctx Pointer to this instance's context
- * \param[in,out] iter Iterator storage, initialized to NULL
- *
- * \return Returns a pointer to an output data packet (compressed frame data,
- * two-pass statistics, etc.) or NULL to signal end-of-list.
- *
- */
- const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx,
- vpx_codec_iter_t *iter);
-
-
- /*!\brief Get Preview Frame
- *
- * Returns an image that can be used as a preview. Shows the image as it would
- * exist at the decompressor. The application \ref MUST NOT write into this
- * image buffer.
- *
- * \param[in] ctx Pointer to this instance's context
- *
- * \return Returns a pointer to a preview image, or NULL if no image is
- * available.
- *
- */
- const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx);
-
-
- /*!@} - end defgroup encoder*/
-
-#endif
-#ifdef __cplusplus
-}
-#endif
diff --git a/vpx/vpx_image.h b/vpx/vpx_image.h
deleted file mode 100644
index 8e08b36..0000000
--- a/vpx/vpx_image.h
+++ /dev/null
@@ -1,242 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*!\file
- * \brief Describes the vpx image descriptor and associated operations
- *
- */
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#ifndef VPX_IMAGE_H
-#define VPX_IMAGE_H
-
- /*!\brief Current ABI version number
- *
- * \internal
- * If this file is altered in any way that changes the ABI, this value
- * must be bumped. Examples include, but are not limited to, changing
- * types, removing or reassigning enums, adding/removing/rearranging
- * fields to structures
- */
-#define VPX_IMAGE_ABI_VERSION (1) /**<\hideinitializer*/
-
-
-#define VPX_IMG_FMT_PLANAR 0x100 /**< Image is a planar format */
-#define VPX_IMG_FMT_UV_FLIP 0x200 /**< V plane precedes U plane in memory */
-#define VPX_IMG_FMT_HAS_ALPHA 0x400 /**< Image has an alpha channel component */
-
-
- /*!\brief List of supported image formats */
- typedef enum vpx_img_fmt {
- VPX_IMG_FMT_NONE,
- VPX_IMG_FMT_RGB24, /**< 24 bit per pixel packed RGB */
- VPX_IMG_FMT_RGB32, /**< 32 bit per pixel packed 0RGB */
- VPX_IMG_FMT_RGB565, /**< 16 bit per pixel, 565 */
- VPX_IMG_FMT_RGB555, /**< 16 bit per pixel, 555 */
- VPX_IMG_FMT_UYVY, /**< UYVY packed YUV */
- VPX_IMG_FMT_YUY2, /**< YUYV packed YUV */
- VPX_IMG_FMT_YVYU, /**< YVYU packed YUV */
- VPX_IMG_FMT_BGR24, /**< 24 bit per pixel packed BGR */
- VPX_IMG_FMT_RGB32_LE, /**< 32 bit packed BGR0 */
- VPX_IMG_FMT_ARGB, /**< 32 bit packed ARGB, alpha=255 */
- VPX_IMG_FMT_ARGB_LE, /**< 32 bit packed BGRA, alpha=255 */
- VPX_IMG_FMT_RGB565_LE, /**< 16 bit per pixel, gggbbbbb rrrrrggg */
- VPX_IMG_FMT_RGB555_LE, /**< 16 bit per pixel, gggbbbbb 0rrrrrgg */
- VPX_IMG_FMT_YV12 = VPX_IMG_FMT_PLANAR | VPX_IMG_FMT_UV_FLIP | 1, /**< planar YVU */
- VPX_IMG_FMT_I420 = VPX_IMG_FMT_PLANAR | 2,
- VPX_IMG_FMT_VPXYV12 = VPX_IMG_FMT_PLANAR | VPX_IMG_FMT_UV_FLIP | 3, /** < planar 4:2:0 format with vpx color space */
- VPX_IMG_FMT_VPXI420 = VPX_IMG_FMT_PLANAR | 4 /** < planar 4:2:0 format with vpx color space */
- }
- vpx_img_fmt_t; /**< alias for enum vpx_img_fmt */
-
-#if !defined(VPX_CODEC_DISABLE_COMPAT) || !VPX_CODEC_DISABLE_COMPAT
-#define IMG_FMT_PLANAR VPX_IMG_FMT_PLANAR /**< \deprecated Use #VPX_IMG_FMT_PLANAR */
-#define IMG_FMT_UV_FLIP VPX_IMG_FMT_UV_FLIP /**< \deprecated Use #VPX_IMG_FMT_UV_FLIP */
-#define IMG_FMT_HAS_ALPHA VPX_IMG_FMT_HAS_ALPHA /**< \deprecated Use #VPX_IMG_FMT_HAS_ALPHA */
-
- /*!\brief Deprecated list of supported image formats
- * \deprecated New code should use #vpx_img_fmt
- */
-#define img_fmt vpx_img_fmt
- /*!\brief alias for enum img_fmt.
- * \deprecated New code should use #vpx_img_fmt_t
- */
-#define img_fmt_t vpx_img_fmt_t
-
-#define IMG_FMT_NONE VPX_IMG_FMT_NONE /**< \deprecated Use #VPX_IMG_FMT_NONE */
-#define IMG_FMT_RGB24 VPX_IMG_FMT_RGB24 /**< \deprecated Use #VPX_IMG_FMT_RGB24 */
-#define IMG_FMT_RGB32 VPX_IMG_FMT_RGB32 /**< \deprecated Use #VPX_IMG_FMT_RGB32 */
-#define IMG_FMT_RGB565 VPX_IMG_FMT_RGB565 /**< \deprecated Use #VPX_IMG_FMT_RGB565 */
-#define IMG_FMT_RGB555 VPX_IMG_FMT_RGB555 /**< \deprecated Use #VPX_IMG_FMT_RGB555 */
-#define IMG_FMT_UYVY VPX_IMG_FMT_UYVY /**< \deprecated Use #VPX_IMG_FMT_UYVY */
-#define IMG_FMT_YUY2 VPX_IMG_FMT_YUY2 /**< \deprecated Use #VPX_IMG_FMT_YUY2 */
-#define IMG_FMT_YVYU VPX_IMG_FMT_YVYU /**< \deprecated Use #VPX_IMG_FMT_YVYU */
-#define IMG_FMT_BGR24 VPX_IMG_FMT_BGR24 /**< \deprecated Use #VPX_IMG_FMT_BGR24 */
-#define IMG_FMT_RGB32_LE VPX_IMG_FMT_RGB32_LE /**< \deprecated Use #VPX_IMG_FMT_RGB32_LE */
-#define IMG_FMT_ARGB VPX_IMG_FMT_ARGB /**< \deprecated Use #VPX_IMG_FMT_ARGB */
-#define IMG_FMT_ARGB_LE VPX_IMG_FMT_ARGB_LE /**< \deprecated Use #VPX_IMG_FMT_ARGB_LE */
-#define IMG_FMT_RGB565_LE VPX_IMG_FMT_RGB565_LE /**< \deprecated Use #VPX_IMG_FMT_RGB565_LE */
-#define IMG_FMT_RGB555_LE VPX_IMG_FMT_RGB555_LE /**< \deprecated Use #VPX_IMG_FMT_RGB555_LE */
-#define IMG_FMT_YV12 VPX_IMG_FMT_YV12 /**< \deprecated Use #VPX_IMG_FMT_YV12 */
-#define IMG_FMT_I420 VPX_IMG_FMT_I420 /**< \deprecated Use #VPX_IMG_FMT_I420 */
-#define IMG_FMT_VPXYV12 VPX_IMG_FMT_VPXYV12 /**< \deprecated Use #VPX_IMG_FMT_VPXYV12 */
-#define IMG_FMT_VPXI420 VPX_IMG_FMT_VPXI420 /**< \deprecated Use #VPX_IMG_FMT_VPXI420 */
-#endif /* VPX_CODEC_DISABLE_COMPAT */
-
- /**\brief Image Descriptor */
- typedef struct vpx_image
- {
- vpx_img_fmt_t fmt; /**< Image Format */
-
- /* Image storage dimensions */
- unsigned int w; /**< Stored image width */
- unsigned int h; /**< Stored image height */
-
- /* Image display dimensions */
- unsigned int d_w; /**< Displayed image width */
- unsigned int d_h; /**< Displayed image height */
-
- /* Chroma subsampling info */
- unsigned int x_chroma_shift; /**< subsampling order, X */
- unsigned int y_chroma_shift; /**< subsampling order, Y */
-
- /* Image data pointers. */
-#define VPX_PLANE_PACKED 0 /**< To be used for all packed formats */
-#define VPX_PLANE_Y 0 /**< Y (Luminance) plane */
-#define VPX_PLANE_U 1 /**< U (Chroma) plane */
-#define VPX_PLANE_V 2 /**< V (Chroma) plane */
-#define VPX_PLANE_ALPHA 3 /**< A (Transparency) plane */
-#if !defined(VPX_CODEC_DISABLE_COMPAT) || !VPX_CODEC_DISABLE_COMPAT
-#define PLANE_PACKED VPX_PLANE_PACKED
-#define PLANE_Y VPX_PLANE_Y
-#define PLANE_U VPX_PLANE_U
-#define PLANE_V VPX_PLANE_V
-#define PLANE_ALPHA VPX_PLANE_ALPHA
-#endif
- unsigned char *planes[4]; /**< pointer to the top left pixel for each plane */
- int stride[4]; /**< stride between rows for each plane */
-
- int bps; /**< bits per sample (for packed formats) */
-
- /* The following member may be set by the application to associate data
- * with this image.
- */
- void *user_priv; /**< may be set by the application to associate data
- * with this image. */
-
- /* The following members should be treated as private. */
- unsigned char *img_data; /**< private */
- int img_data_owner; /**< private */
- int self_allocd; /**< private */
- } vpx_image_t; /**< alias for struct vpx_image */
-
- /**\brief Representation of a rectangle on a surface */
- typedef struct vpx_image_rect
- {
- unsigned int x; /**< leftmost column */
- unsigned int y; /**< topmost row */
- unsigned int w; /**< width */
- unsigned int h; /**< height */
- } vpx_image_rect_t; /**< alias for struct vpx_image_rect */
-
- /*!\brief Open a descriptor, allocating storage for the underlying image
- *
- * Returns a descriptor for storing an image of the given format. The
- * storage for the descriptor is allocated on the heap.
- *
- * \param[in] img Pointer to storage for descriptor. If this parameter
- * is NULL, the storage for the descriptor will be
- * allocated on the heap.
- * \param[in] fmt Format for the image
- * \param[in] d_w Width of the image
- * \param[in] d_h Height of the image
- * \param[in] align Alignment, in bytes, of each row in the image.
- *
- * \return Returns a pointer to the initialized image descriptor. If the img
- * parameter is non-null, the value of the img parameter will be
- * returned.
- */
- vpx_image_t *vpx_img_alloc(vpx_image_t *img,
- vpx_img_fmt_t fmt,
- unsigned int d_w,
- unsigned int d_h,
- unsigned int align);
-
- /*!\brief Open a descriptor, using existing storage for the underlying image
- *
- * Returns a descriptor for storing an image of the given format. The
- * storage for descriptor has been allocated elsewhere, and a descriptor is
- * desired to "wrap" that storage.
- *
- * \param[in] img Pointer to storage for descriptor. If this parameter
- * is NULL, the storage for the descriptor will be
- * allocated on the heap.
- * \param[in] fmt Format for the image
- * \param[in] d_w Width of the image
- * \param[in] d_h Height of the image
- * \param[in] align Alignment, in bytes, of each row in the image.
- * \param[in] img_data Storage to use for the image
- *
- * \return Returns a pointer to the initialized image descriptor. If the img
- * parameter is non-null, the value of the img parameter will be
- * returned.
- */
- vpx_image_t *vpx_img_wrap(vpx_image_t *img,
- vpx_img_fmt_t fmt,
- unsigned int d_w,
- unsigned int d_h,
- unsigned int align,
- unsigned char *img_data);
-
-
- /*!\brief Set the rectangle identifying the displayed portion of the image
- *
- * Updates the displayed rectangle (aka viewport) on the image surface to
- * match the specified coordinates and size.
- *
- * \param[in] img Image descriptor
- * \param[in] x leftmost column
- * \param[in] y topmost row
- * \param[in] w width
- * \param[in] h height
- *
- * \return 0 if the requested rectangle is valid, nonzero otherwise.
- */
- int vpx_img_set_rect(vpx_image_t *img,
- unsigned int x,
- unsigned int y,
- unsigned int w,
- unsigned int h);
-
-
- /*!\brief Flip the image vertically (top for bottom)
- *
- * Adjusts the image descriptor's pointers and strides to make the image
- * be referenced upside-down.
- *
- * \param[in] img Image descriptor
- */
- void vpx_img_flip(vpx_image_t *img);
-
- /*!\brief Close an image descriptor
- *
- * Frees all allocated storage associated with an image descriptor.
- *
- * \param[in] img Image descriptor
- */
- void vpx_img_free(vpx_image_t *img);
-
-#endif
-#ifdef __cplusplus
-}
-#endif
diff --git a/vpx/vpx_integer.h b/vpx/vpx_integer.h
deleted file mode 100644
index 9a06c1a..0000000
--- a/vpx/vpx_integer.h
+++ /dev/null
@@ -1,57 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VPX_INTEGER_H
-#define VPX_INTEGER_H
-
-/* get ptrdiff_t, size_t, wchar_t, NULL */
-#include <stddef.h>
-
-#if defined(_MSC_VER) || defined(VPX_EMULATE_INTTYPES)
-typedef signed char int8_t;
-typedef signed short int16_t;
-typedef signed int int32_t;
-
-typedef unsigned char uint8_t;
-typedef unsigned short uint16_t;
-typedef unsigned int uint32_t;
-
-#if defined(_MSC_VER)
-typedef signed __int64 int64_t;
-typedef unsigned __int64 uint64_t;
-#define PRId64 "I64d"
-#endif
-
-#ifdef HAVE_ARMV6
-typedef unsigned int int_fast16_t;
-#else
-typedef signed short int_fast16_t;
-#endif
-typedef signed char int_fast8_t;
-typedef unsigned char uint_fast8_t;
-
-#ifndef _UINTPTR_T_DEFINED
-typedef unsigned int uintptr_t;
-#endif
-
-#else
-
-/* Most platforms have the C99 standard integer types. */
-
-#if defined(__cplusplus) && !defined(__STDC_FORMAT_MACROS)
-#define __STDC_FORMAT_MACROS
-#endif
-#include <stdint.h>
-#include <inttypes.h>
-
-#endif
-
-#endif
diff --git a/vpx_config.c b/vpx_config.c
deleted file mode 100644
index ad2775b..0000000
--- a/vpx_config.c
+++ /dev/null
@@ -1,2 +0,0 @@
-static const char* const cfg = "--target=generic-gnu";
-const char *vpx_codec_build_config(void) {return cfg;}
diff --git a/vpx_config.h b/vpx_config.h
deleted file mode 100644
index 71d9a09..0000000
--- a/vpx_config.h
+++ /dev/null
@@ -1,87 +0,0 @@
-/* This file automatically generated by configure. Do not edit! */
-#define INLINE
-#define FORCEINLINE
-#define RESTRICT
-
-#if defined(__arm__)
-
-#define ARCH_ARM 1
-#define HAVE_ARMV5TE 0
-#else
-#define ARCH_ARM 0
-#endif
-
-#if defined(__ARM_HAVE_NEON)
-#define HAVE_ARMV7 1
-#else
-#define HAVE_ARMV7 0
-#endif
-
-#if defined(__ARM_HAVE_ARMV6)
-#define HAVE_ARMV6 1
-#else
-#define HAVE_ARMV6 0
-#endif
-
-#if defined(__mips__)
-#define ARCH_MIPS 1
-#else
-#define ARCH_MIPS 0
-#endif
-
-#define ARCH_X86 0
-#define ARCH_X86_64 0
-#define ARCH_PPC32 0
-#define ARCH_PPC64 0
-
-#define HAVE_IWMMXT 0
-#define HAVE_IWMMXT2 0
-#define HAVE_MMX 0
-#define HAVE_SSE 0
-#define HAVE_SSE2 0
-#define HAVE_SSE3 0
-#define HAVE_SSSE3 0
-#define HAVE_ALTIVEC 0
-#define HAVE_VPX_PORTS 1
-#define HAVE_STDINT_H 1
-#define HAVE_ALT_TREE_LAYOUT 0
-#define HAVE_PTHREAD_H 1
-#define HAVE_SYS_MMAN_H 1
-#define CONFIG_EXTERNAL_BUILD 0
-#define CONFIG_INSTALL_DOCS 0
-#define CONFIG_INSTALL_BINS 1
-#define CONFIG_INSTALL_LIBS 1
-#define CONFIG_INSTALL_SRCS 1
-#define CONFIG_DEBUG 0
-#define CONFIG_GPROF 0
-#define CONFIG_GCOV 0
-#define CONFIG_RVCT 0
-#define CONFIG_GCC 0
-#define CONFIG_MSVS 0
-#define CONFIG_PIC 0
-#define CONFIG_BIG_ENDIAN 0
-#define CONFIG_CODEC_SRCS 0
-#define CONFIG_DEBUG_LIBS 0
-#define CONFIG_FAST_UNALIGNED 1
-#define CONFIG_MEM_MANAGER 0
-#define CONFIG_MEM_TRACKER 0
-#define CONFIG_MEM_CHECKS 0
-#define CONFIG_MD5 1
-#define CONFIG_DEQUANT_TOKENS 0
-#define CONFIG_DC_RECON 0
-#define CONFIG_NEW_TOKENS 0
-#define CONFIG_EVAL_LIMIT 0
-#define CONFIG_RUNTIME_CPU_DETECT 0
-#define CONFIG_POSTPROC 0
-#define CONFIG_POSTPROC_GENERIC 0
-#define CONFIG_OS_SUPPORT 1
-#define CONFIG_MULTITHREAD 1
-#define CONFIG_PSNR 0
-#define CONFIG_VP8_ENCODER 1
-#define CONFIG_VP8_DECODER 1
-#define CONFIG_VP8 1
-#define CONFIG_ENCODERS 1
-#define CONFIG_DECODERS 1
-#define CONFIG_STATIC_MSVCRT 0
-#define CONFIG_SPATIAL_RESAMPLING 1
-#define CONFIG_REALTIME_ONLY 0
diff --git a/vpx_mem/include/nds/vpx_mem_nds.h b/vpx_mem/include/nds/vpx_mem_nds.h
deleted file mode 100644
index e54f54d..0000000
--- a/vpx_mem/include/nds/vpx_mem_nds.h
+++ /dev/null
@@ -1,30 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __VPX_MEM_NDS_H__
-#define __VPX_MEM_NDS_H__
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-#include <nitro.h>
-#include <nitro/os.h>
-
- void *vpx_mem_nds_alloc(osarena_id id, osheap_handle handle, size_t size, size_t align);
- void vpx_mem_nds_free(osarena_id id, osheap_handle handle, void *mem);
- int vpx_nds_alloc_heap(osarena_id id, u32 size);
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif /*__VPX_MEM_NDS_H__*/
diff --git a/vpx_mem/include/vpx_mem_intrnl.h b/vpx_mem/include/vpx_mem_intrnl.h
deleted file mode 100644
index 6e261ba..0000000
--- a/vpx_mem/include/vpx_mem_intrnl.h
+++ /dev/null
@@ -1,99 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __VPX_MEM_INTRNL_H__
-#define __VPX_MEM_INTRNL_H__
-#include "vpx_ports/config.h"
-
-#ifndef CONFIG_MEM_MANAGER
-# if defined(VXWORKS)
-# define CONFIG_MEM_MANAGER 1 /*include heap manager functionality,*/
-/*default: enabled on vxworks*/
-# else
-# define CONFIG_MEM_MANAGER 0 /*include heap manager functionality*/
-# endif
-#endif /*CONFIG_MEM_MANAGER*/
-
-#ifndef CONFIG_MEM_TRACKER
-# define CONFIG_MEM_TRACKER 1 /*include xvpx_* calls in the lib*/
-#endif
-
-#ifndef CONFIG_MEM_CHECKS
-# define CONFIG_MEM_CHECKS 0 /*include some basic safety checks in
-vpx_memcpy, _memset, and _memmove*/
-#endif
-
-#ifndef USE_GLOBAL_FUNCTION_POINTERS
-# define USE_GLOBAL_FUNCTION_POINTERS 0 /*use function pointers instead of compiled functions.*/
-#endif
-
-#if CONFIG_MEM_TRACKER
-# include "vpx_mem_tracker.h"
-# if VPX_MEM_TRACKER_VERSION_CHIEF != 2 || VPX_MEM_TRACKER_VERSION_MAJOR != 5
-# error "vpx_mem requires memory tracker version 2.5 to track memory usage"
-# endif
-#endif
-
-#define ADDRESS_STORAGE_SIZE sizeof(size_t)
-
-#ifndef DEFAULT_ALIGNMENT
-# if defined(VXWORKS)
-# define DEFAULT_ALIGNMENT 32 /*default addr alignment to use in
- calls to vpx_* functions other
- than vpx_memalign*/
-# else
-# define DEFAULT_ALIGNMENT 1
-# endif
-#endif
-
-#if DEFAULT_ALIGNMENT < 1
-# error "DEFAULT_ALIGNMENT must be >= 1!"
-#endif
-
-#if CONFIG_MEM_TRACKER
-# define TRY_BOUNDS_CHECK 1 /*when set to 1 pads each allocation,
- integrity can be checked using
- vpx_memory_tracker_check_integrity
- or on free by defining*/
-/*TRY_BOUNDS_CHECK_ON_FREE*/
-#else
-# define TRY_BOUNDS_CHECK 0
-#endif /*CONFIG_MEM_TRACKER*/
-
-#if TRY_BOUNDS_CHECK
-# define TRY_BOUNDS_CHECK_ON_FREE 0 /*checks mem integrity on every
- free, very expensive*/
-# define BOUNDS_CHECK_VALUE 0xdeadbeef /*value stored before/after ea.
- mem addr for bounds checking*/
-# define BOUNDS_CHECK_PAD_SIZE 32 /*size of the padding before and
- after ea allocation to be filled
- with BOUNDS_CHECK_VALUE.
- this should be a multiple of 4*/
-#else
-# define BOUNDS_CHECK_VALUE 0
-# define BOUNDS_CHECK_PAD_SIZE 0
-#endif /*TRY_BOUNDS_CHECK*/
-
-#ifndef REMOVE_PRINTFS
-# define REMOVE_PRINTFS 0
-#endif
-
-/* Should probably use a vpx_mem logger function. */
-#if REMOVE_PRINTFS
-# define _P(x)
-#else
-# define _P(x) x
-#endif
-
-/*returns an addr aligned to the byte boundary specified by align*/
-#define align_addr(addr,align) (void*)(((size_t)(addr) + ((align) - 1)) & (size_t)-(align))
-
-#endif /*__VPX_MEM_INTRNL_H__*/
diff --git a/vpx_mem/vpx_mem_tracker.c b/vpx_mem/vpx_mem_tracker.c
deleted file mode 100644
index 938ad07..0000000
--- a/vpx_mem/vpx_mem_tracker.c
+++ /dev/null
@@ -1,823 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/*
- vpx_mem_tracker.c
-
- jwz 2003-09-30:
- Stores a list of addreses, their size, and file and line they came from.
- All exposed lib functions are prefaced by vpx_ and allow the global list
- to be thread safe.
- Current supported platforms are:
- Linux, Win32, win_ce and vx_works
- Further support can be added by defining the platform specific mutex
- in the memory_tracker struct as well as calls to create/destroy/lock/unlock
- the mutex in vpx_memory_tracker_init/Destroy and memory_tracker_lock_mutex/unlock_mutex
-*/
-#include "vpx_ports/config.h"
-
-#if defined(__uClinux__)
-# include <lddk.h>
-#endif
-
-#if HAVE_PTHREAD_H
-# include <pthread.h>
-#elif defined(WIN32) || defined(_WIN32_WCE)
-# define WIN32_LEAN_AND_MEAN
-# include <windows.h>
-# include <winbase.h>
-#elif defined(VXWORKS)
-# include <sem_lib.h>
-#elif defined(NDS_NITRO)
-# include <nitro.h>
-# include <nitro/os.h>
-#endif
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h> //VXWORKS doesn't have a malloc/memory.h file,
-//this should pull in malloc,free,etc.
-#include <stdarg.h>
-
-#include "include/vpx_mem_tracker.h"
-
-#undef vpx_malloc //undefine any vpx_mem macros that may affect calls to
-#undef vpx_free //memory functions in this file
-#undef vpx_memcpy
-#undef vpx_memset
-
-
-#ifndef USE_GLOBAL_FUNCTION_POINTERS
-# define USE_GLOBAL_FUNCTION_POINTERS 0 //use function pointers instead of compiled functions.
-#endif
-
-#if USE_GLOBAL_FUNCTION_POINTERS
-static mem_track_malloc_func g_malloc = malloc;
-static mem_track_calloc_func g_calloc = calloc;
-static mem_track_realloc_func g_realloc = realloc;
-static mem_track_free_func g_free = free;
-static mem_track_memcpy_func g_memcpy = memcpy;
-static mem_track_memset_func g_memset = memset;
-static mem_track_memmove_func g_memmove = memmove;
-# define MEM_TRACK_MALLOC g_malloc
-# define MEM_TRACK_FREE g_free
-# define MEM_TRACK_MEMCPY g_memcpy
-# define MEM_TRACK_MEMSET g_memset
-#else
-# define MEM_TRACK_MALLOC vpx_malloc
-# define MEM_TRACK_FREE vpx_free
-# define MEM_TRACK_MEMCPY vpx_memcpy
-# define MEM_TRACK_MEMSET vpx_memset
-#endif // USE_GLOBAL_FUNCTION_POINTERS
-
-/* prototypes for internal library functions */
-static void memtrack_log(const char *fmt, ...);
-static void memory_tracker_dump();
-static void memory_tracker_check_integrity(char *file, unsigned int line);
-static void memory_tracker_add(size_t addr, unsigned int size,
- char *file, unsigned int line,
- int padded);
-static int memory_tracker_remove(size_t addr);
-static struct mem_block *memory_tracker_find(size_t addr);
-
-#if defined(NO_MUTEX)
-# define memory_tracker_lock_mutex() (!g_b_mem_tracker_inited)
-# define memory_tracker_unlock_mutex()
-#else
-static int memory_tracker_lock_mutex();
-static int memory_tracker_unlock_mutex();
-#endif
-
-#ifndef VPX_NO_GLOBALS
-struct memory_tracker
-{
- struct mem_block *head,
- * tail;
- int len,
- totalsize;
- unsigned int current_allocated,
- max_allocated;
-
-#if HAVE_PTHREAD_H
- pthread_mutex_t mutex;
-#elif defined(WIN32) || defined(_WIN32_WCE)
- HANDLE mutex;
-#elif defined(VXWORKS)
- SEM_ID mutex;
-#elif defined(NDS_NITRO)
- OSMutex mutex;
-#elif defined(NO_MUTEX)
-#else
-#error "No mutex type defined for this platform!"
-#endif
-
- int padding_size,
- pad_value;
-};
-
-static struct memory_tracker memtrack; //our global memory allocation list
-static int g_b_mem_tracker_inited = 0; //indicates whether the global list has
-//been initialized (1:yes/0:no)
-static struct
-{
- FILE *file;
- int type;
- void (*func)(void *userdata, const char *fmt, va_list args);
- void *userdata;
-} g_logging = {NULL, 0, NULL, NULL};
-#else
-# include "vpx_global_handling.h"
-#define g_b_mem_tracker_inited vpxglobalm(vpxmem,g_b_mem_tracker_inited)
-#define g_logging vpxglobalm(vpxmem,g_logging)
-#define memtrack vpxglobalm(vpxmem,memtrack)
-#endif // #ifndef VPX_NO_GLOBALS
-
-extern void *vpx_malloc(size_t size);
-extern void vpx_free(void *memblk);
-extern void *vpx_memcpy(void *dest, const void *src, size_t length);
-extern void *vpx_memset(void *dest, int val, size_t length);
-
-/*
- *
- * Exposed library functions
- *
-*/
-
-/*
- vpx_memory_tracker_init(int padding_size, int pad_value)
- padding_size - the size of the padding before and after each mem addr.
- Values > 0 indicate that integrity checks can be performed
- by inspecting these areas.
- pad_value - the initial value within the padding area before and after
- each mem addr.
-
- Initializes global memory tracker structure
- Allocates the head of the list
-*/
-int vpx_memory_tracker_init(int padding_size, int pad_value)
-{
- if (!g_b_mem_tracker_inited)
- {
- if ((memtrack.head = (struct mem_block *)
- MEM_TRACK_MALLOC(sizeof(struct mem_block))))
- {
- int ret;
-
- MEM_TRACK_MEMSET(memtrack.head, 0, sizeof(struct mem_block));
-
- memtrack.tail = memtrack.head;
-
- memtrack.current_allocated = 0;
- memtrack.max_allocated = 0;
-
- memtrack.padding_size = padding_size;
- memtrack.pad_value = pad_value;
-
-#if HAVE_PTHREAD_H
- ret = pthread_mutex_init(&memtrack.mutex,
- NULL); /*mutex attributes (NULL=default)*/
-#elif defined(WIN32) || defined(_WIN32_WCE)
- memtrack.mutex = CreateMutex(NULL, /*security attributes*/
- FALSE, /*we don't want initial ownership*/
- NULL); /*mutex name*/
- ret = !memtrack.mutex;
-#elif defined(VXWORKS)
- memtrack.mutex = sem_bcreate(SEM_Q_FIFO, /*SEM_Q_FIFO non-priority based mutex*/
- SEM_FULL); /*SEM_FULL initial state is unlocked*/
- ret = !memtrack.mutex;
-#elif defined(NDS_NITRO)
- os_init_mutex(&memtrack.mutex);
- ret = 0;
-#elif defined(NO_MUTEX)
- ret = 0;
-#endif
-
- if (ret)
- {
- memtrack_log("vpx_memory_tracker_init: Error creating mutex!\n");
-
- MEM_TRACK_FREE(memtrack.head);
- memtrack.head = NULL;
- }
- else
- {
- memtrack_log("Memory Tracker init'd, v."vpx_mem_tracker_version" pad_size:%d pad_val:0x%x %d\n"
- , padding_size
- , pad_value
- , pad_value);
- g_b_mem_tracker_inited = 1;
- }
- }
- }
-
- return g_b_mem_tracker_inited;
-}
-
-/*
- vpx_memory_tracker_destroy()
- If our global struct was initialized zeros out all its members,
- frees memory and destroys it's mutex
-*/
-void vpx_memory_tracker_destroy()
-{
- if (!memory_tracker_lock_mutex())
- {
- struct mem_block *p = memtrack.head,
- * p2 = memtrack.head;
-
- memory_tracker_dump();
-
- while (p)
- {
- p2 = p;
- p = p->next;
-
- MEM_TRACK_FREE(p2);
- }
-
- memtrack.head = NULL;
- memtrack.tail = NULL;
- memtrack.len = 0;
- memtrack.current_allocated = 0;
- memtrack.max_allocated = 0;
-
- if (!g_logging.type && g_logging.file && g_logging.file != stderr)
- {
-#if !defined(NDS_NITRO)
- fclose(g_logging.file);
-#endif
- g_logging.file = NULL;
- }
-
- memory_tracker_unlock_mutex();
-
- g_b_mem_tracker_inited = 0;
- }
-}
-
-/*
- vpx_memory_tracker_add(size_t addr, unsigned int size,
- char * file, unsigned int line)
- addr - memory address to be added to list
- size - size of addr
- file - the file addr was referenced from
- line - the line in file addr was referenced from
- Adds memory address addr, it's size, file and line it came from
- to the global list via the thread safe internal library function
-*/
-void vpx_memory_tracker_add(size_t addr, unsigned int size,
- char *file, unsigned int line,
- int padded)
-{
- memory_tracker_add(addr, size, file, line, padded);
-}
-
-/*
- vpx_memory_tracker_remove(size_t addr)
- addr - memory address to be removed from list
- Removes addr from the global list via the thread safe
- internal remove function
- Return:
- Same as described for memory_tracker_remove
-*/
-int vpx_memory_tracker_remove(size_t addr)
-{
- return memory_tracker_remove(addr);
-}
-
-/*
- vpx_memory_tracker_find(size_t addr)
- addr - address to be found in list
- Return:
- If found, pointer to the memory block that matches addr
- NULL otherwise
-*/
-struct mem_block *vpx_memory_tracker_find(size_t addr)
-{
- struct mem_block *p = NULL;
-
- if (!memory_tracker_lock_mutex())
- {
- p = memory_tracker_find(addr);
- memory_tracker_unlock_mutex();
- }
-
- return p;
-}
-
-/*
- vpx_memory_tracker_dump()
- Locks the memory tracker's mutex and calls the internal
- library function to dump the current contents of the
- global memory allocation list
-*/
-void vpx_memory_tracker_dump()
-{
- if (!memory_tracker_lock_mutex())
- {
- memory_tracker_dump();
- memory_tracker_unlock_mutex();
- }
-}
-
-/*
- vpx_memory_tracker_check_integrity(char* file, unsigned int line)
- file - The file name where the check was placed
- line - The line in file where the check was placed
- Locks the memory tracker's mutex and calls the internal
- integrity check function to inspect every address in the global
- memory allocation list
-*/
-void vpx_memory_tracker_check_integrity(char *file, unsigned int line)
-{
- if (!memory_tracker_lock_mutex())
- {
- memory_tracker_check_integrity(file, line);
- memory_tracker_unlock_mutex();
- }
-}
-
-/*
- vpx_memory_tracker_set_log_type
- Sets the logging type for the memory tracker. Based on the value it will
- direct its output to the appropriate place.
- Return:
- 0: on success
- -1: if the logging type could not be set, because the value was invalid
- or because a file could not be opened
-*/
-int vpx_memory_tracker_set_log_type(int type, char *option)
-{
- int ret = -1;
-
- switch (type)
- {
- case 0:
- g_logging.type = 0;
-
- if (!option)
- {
- g_logging.file = stderr;
- ret = 0;
- }
-
-#if !defined(NDS_NITRO)
- else
- {
- if ((g_logging.file = fopen((char *)option, "w")))
- ret = 0;
- }
-
-#endif
- break;
-#if defined(WIN32) && !defined(_WIN32_WCE)
- case 1:
- g_logging.type = type;
- ret = 0;
- break;
-#endif
- default:
- break;
- }
-
- //output the version to the new logging destination
- if (!ret)
- memtrack_log("Memory Tracker logging initialized, "
- "Memory Tracker v."vpx_mem_tracker_version"\n");
-
- return ret;
-}
-
-/*
- vpx_memory_tracker_set_log_func
- Sets a logging function to be used by the memory tracker.
- Return:
- 0: on success
- -1: if the logging type could not be set because logfunc was NULL
-*/
-int vpx_memory_tracker_set_log_func(void *userdata,
- void(*logfunc)(void *userdata,
- const char *fmt, va_list args))
-{
- int ret = -1;
-
- if (logfunc)
- {
- g_logging.type = -1;
- g_logging.userdata = userdata;
- g_logging.func = logfunc;
- ret = 0;
- }
-
- //output the version to the new logging destination
- if (!ret)
- memtrack_log("Memory Tracker logging initialized, "
- "Memory Tracker v."vpx_mem_tracker_version"\n");
-
- return ret;
-}
-
-/*
- *
- * END - Exposed library functions
- *
-*/
-
-
-/*
- *
- * Internal library functions
- *
-*/
-
-static void memtrack_log(const char *fmt, ...)
-{
- va_list list;
-
- va_start(list, fmt);
-
- switch (g_logging.type)
- {
- case -1:
-
- if (g_logging.func)
- g_logging.func(g_logging.userdata, fmt, list);
-
- break;
- case 0:
-
- if (g_logging.file)
- {
- vfprintf(g_logging.file, fmt, list);
- fflush(g_logging.file);
- }
-
- break;
-#if defined(WIN32) && !defined(_WIN32_WCE)
- case 1:
- {
- char temp[1024];
- _vsnprintf(temp, sizeof(temp) / sizeof(char) - 1, fmt, list);
- OutputDebugString(temp);
- }
- break;
-#endif
- default:
- break;
- }
-
- va_end(list);
-}
-
-/*
- memory_tracker_dump()
- Dumps the current contents of the global memory allocation list
-*/
-static void memory_tracker_dump()
-{
- int i = 0;
- struct mem_block *p = (memtrack.head ? memtrack.head->next : NULL);
-
- memtrack_log("\n_currently Allocated= %d; Max allocated= %d\n",
- memtrack.current_allocated, memtrack.max_allocated);
-
- while (p)
- {
-#if defined(WIN32) && !defined(_WIN32_WCE)
-
- /*when using outputdebugstring, output filenames so they
- can be clicked to be opened in visual studio*/
- if (g_logging.type == 1)
- memtrack_log("memblocks[%d].addr= 0x%.8x, memblocks[%d].size= %d, file:\n"
- " %s(%d):\n", i,
- p->addr, i, p->size,
- p->file, p->line);
- else
-#endif
- memtrack_log("memblocks[%d].addr= 0x%.8x, memblocks[%d].size= %d, file: %s, line: %d\n", i,
- p->addr, i, p->size,
- p->file, p->line);
-
-#ifdef NDS_NITRO
-
- if (!(i % 20)) os_sleep(500);
-
-#endif
-
- p = p->next;
- ++i;
- }
-
- memtrack_log("\n");
-}
-
-/*
- memory_tracker_check_integrity(char* file, unsigned int file)
- file - the file name where the check was placed
- line - the line in file where the check was placed
- If a padding_size was supplied to vpx_memory_tracker_init()
- this function will check ea. addr in the list verifying that
- addr-padding_size and addr+padding_size is filled with pad_value
-*/
-static void memory_tracker_check_integrity(char *file, unsigned int line)
-{
- if (memtrack.padding_size)
- {
- int i,
- index = 0;
- unsigned char *p_show_me,
- * p_show_me2;
- unsigned int tempme = memtrack.pad_value,
- dead1,
- dead2;
- unsigned char *x_bounds;
- struct mem_block *p = memtrack.head->next;
-
- while (p)
- {
- //x_bounds = (unsigned char*)p->addr;
- //back up VPX_BYTE_ALIGNMENT
- //x_bounds -= memtrack.padding_size;
-
- if (p->padded) // can the bounds be checked?
- {
- /*yes, move to the address that was actually allocated
- by the vpx_* calls*/
- x_bounds = (unsigned char *)(((size_t *)p->addr)[-1]);
-
- for (i = 0; i < memtrack.padding_size; i += sizeof(unsigned int))
- {
- p_show_me = (x_bounds + i);
- p_show_me2 = (unsigned char *)(p->addr + p->size + i);
-
- MEM_TRACK_MEMCPY(&dead1, p_show_me, sizeof(unsigned int));
- MEM_TRACK_MEMCPY(&dead2, p_show_me2, sizeof(unsigned int));
-
- if ((dead1 != tempme) || (dead2 != tempme))
- {
- memtrack_log("\n[vpx_mem integrity check failed]:\n"
- " index[%d,%d] {%s:%d} addr=0x%x, size=%d,"
- " file: %s, line: %d c0:0x%x c1:0x%x\n",
- index, i, file, line, p->addr, p->size, p->file,
- p->line, dead1, dead2);
- }
- }
- }
-
- ++index;
- p = p->next;
- }
- }
-}
-
-/*
- memory_tracker_add(size_t addr, unsigned int size,
- char * file, unsigned int line)
- Adds an address (addr), it's size, file and line number to our list.
- Adjusts the total bytes allocated and max bytes allocated if necessary.
- If memory cannot be allocated the list will be destroyed.
-*/
-void memory_tracker_add(size_t addr, unsigned int size,
- char *file, unsigned int line,
- int padded)
-{
- if (!memory_tracker_lock_mutex())
- {
- struct mem_block *p;
-
- p = MEM_TRACK_MALLOC(sizeof(struct mem_block));
-
- if (p)
- {
- p->prev = memtrack.tail;
- p->prev->next = p;
- p->addr = addr;
- p->size = size;
- p->line = line;
- p->file = file;
- p->padded = padded;
- p->next = NULL;
-
- memtrack.tail = p;
-
- memtrack.current_allocated += size;
-
- if (memtrack.current_allocated > memtrack.max_allocated)
- memtrack.max_allocated = memtrack.current_allocated;
-
- //memtrack_log("memory_tracker_add: added addr=0x%.8x\n", addr);
-
- memory_tracker_unlock_mutex();
- }
- else
- {
- memtrack_log("memory_tracker_add: error allocating memory!\n");
- memory_tracker_unlock_mutex();
- vpx_memory_tracker_destroy();
- }
- }
-}
-
-/*
- memory_tracker_remove(size_t addr)
- Removes an address and its corresponding size (if they exist)
- from the memory tracker list and adjusts the current number
- of bytes allocated.
- Return:
- 0: on success
- -1: if the mutex could not be locked
- -2: if the addr was not found in the list
-*/
-int memory_tracker_remove(size_t addr)
-{
- int ret = -1;
-
- if (!memory_tracker_lock_mutex())
- {
- struct mem_block *p;
-
- if ((p = memory_tracker_find(addr)))
- {
- memtrack.current_allocated -= p->size;
-
- p->prev->next = p->next;
-
- if (p->next)
- p->next->prev = p->prev;
- else
- memtrack.tail = p->prev;
-
- ret = 0;
- MEM_TRACK_FREE(p);
- }
- else
- {
- if (addr)
- memtrack_log("memory_tracker_remove(): addr not found in list,"
- " 0x%.8x\n", addr);
-
- ret = -2;
- }
-
- memory_tracker_unlock_mutex();
- }
-
- return ret;
-}
-
-/*
- memory_tracker_find(size_t addr)
- Finds an address in our addrs list
- NOTE: the mutex MUST be locked in the other internal
- functions before calling this one. This avoids
- the need for repeated locking and unlocking as in Remove
- Returns: pointer to the mem block if found, NULL otherwise
-*/
-static struct mem_block *memory_tracker_find(size_t addr)
-{
- struct mem_block *p = NULL;
-
- if (memtrack.head)
- {
- p = memtrack.head->next;
-
- while (p && (p->addr != addr))
- p = p->next;
- }
-
- return p;
-}
-
-
-#if !defined(NO_MUTEX)
-/*
- memory_tracker_lock_mutex()
- Locks the memory tracker mutex with a platform specific call
- Returns:
- 0: Success
- <0: Failure, either the mutex was not initialized
- or the call to lock the mutex failed
-*/
-static int memory_tracker_lock_mutex()
-{
- int ret = -1;
-
- if (g_b_mem_tracker_inited)
- {
-
-#if HAVE_PTHREAD_H
- ret = pthread_mutex_lock(&memtrack.mutex);
-#elif defined(WIN32) || defined(_WIN32_WCE)
- ret = WaitForSingleObject(memtrack.mutex, INFINITE);
-#elif defined(VXWORKS)
- ret = sem_take(memtrack.mutex, WAIT_FOREVER);
-#elif defined(NDS_NITRO)
- os_lock_mutex(&memtrack.mutex);
- ret = 0;
-#endif
-
- if (ret)
- {
- memtrack_log("memory_tracker_lock_mutex: mutex lock failed\n");
- }
- }
-
- return ret;
-}
-
-/*
- memory_tracker_unlock_mutex()
- Unlocks the memory tracker mutex with a platform specific call
- Returns:
- 0: Success
- <0: Failure, either the mutex was not initialized
- or the call to unlock the mutex failed
-*/
-static int memory_tracker_unlock_mutex()
-{
- int ret = -1;
-
- if (g_b_mem_tracker_inited)
- {
-
-#if HAVE_PTHREAD_H
- ret = pthread_mutex_unlock(&memtrack.mutex);
-#elif defined(WIN32) || defined(_WIN32_WCE)
- ret = !ReleaseMutex(memtrack.mutex);
-#elif defined(VXWORKS)
- ret = sem_give(memtrack.mutex);
-#elif defined(NDS_NITRO)
- os_unlock_mutex(&memtrack.mutex);
- ret = 0;
-#endif
-
- if (ret)
- {
- memtrack_log("memory_tracker_unlock_mutex: mutex unlock failed\n");
- }
- }
-
- return ret;
-}
-#endif
-
-/*
- vpx_memory_tracker_set_functions
-
- Sets the function pointers for the standard library functions.
-
- Return:
- 0: on success
- -1: if the use global function pointers is not set.
-*/
-int vpx_memory_tracker_set_functions(mem_track_malloc_func g_malloc_l
- , mem_track_calloc_func g_calloc_l
- , mem_track_realloc_func g_realloc_l
- , mem_track_free_func g_free_l
- , mem_track_memcpy_func g_memcpy_l
- , mem_track_memset_func g_memset_l
- , mem_track_memmove_func g_memmove_l)
-{
-#if USE_GLOBAL_FUNCTION_POINTERS
-
- if (g_malloc_l)
- g_malloc = g_malloc_l;
-
- if (g_calloc_l)
- g_calloc = g_calloc_l;
-
- if (g_realloc_l)
- g_realloc = g_realloc_l;
-
- if (g_free_l)
- g_free = g_free_l;
-
- if (g_memcpy_l)
- g_memcpy = g_memcpy_l;
-
- if (g_memset_l)
- g_memset = g_memset_l;
-
- if (g_memmove_l)
- g_memmove = g_memmove_l;
-
- return 0;
-#else
- (void)g_malloc_l;
- (void)g_calloc_l;
- (void)g_realloc_l;
- (void)g_free_l;
- (void)g_memcpy_l;
- (void)g_memset_l;
- (void)g_memmove_l;
- return -1;
-#endif
-}
diff --git a/vpx_ports/arm.h b/vpx_ports/arm.h
deleted file mode 100644
index 81af1f1..0000000
--- a/vpx_ports/arm.h
+++ /dev/null
@@ -1,27 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VPX_PORTS_ARM_H
-#define VPX_PORTS_ARM_H
-#include <stdlib.h>
-#include "config.h"
-
-/*ARMv5TE "Enhanced DSP" instructions.*/
-#define HAS_EDSP 0x01
-/*ARMv6 "Parallel" or "Media" instructions.*/
-#define HAS_MEDIA 0x02
-/*ARMv7 optional NEON instructions.*/
-#define HAS_NEON 0x04
-
-int arm_cpu_caps(void);
-
-#endif
-
diff --git a/vpx_ports/arm_cpudetect.c b/vpx_ports/arm_cpudetect.c
deleted file mode 100644
index 4109924..0000000
--- a/vpx_ports/arm_cpudetect.c
+++ /dev/null
@@ -1,190 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-#include <stdlib.h>
-#include <string.h>
-#include "arm.h"
-
-static int arm_cpu_env_flags(int *flags)
-{
- char *env;
- env = getenv("VPX_SIMD_CAPS");
- if (env && *env)
- {
- *flags = (int)strtol(env, NULL, 0);
- return 0;
- }
- *flags = 0;
- return -1;
-}
-
-static int arm_cpu_env_mask(void)
-{
- char *env;
- env = getenv("VPX_SIMD_CAPS_MASK");
- return env && *env ? (int)strtol(env, NULL, 0) : ~0;
-}
-
-
-#if defined(_MSC_VER)
-/*For GetExceptionCode() and EXCEPTION_ILLEGAL_INSTRUCTION.*/
-#define WIN32_LEAN_AND_MEAN
-#define WIN32_EXTRA_LEAN
-#include <windows.h>
-
-int arm_cpu_caps(void)
-{
- int flags;
- int mask;
- if (!arm_cpu_env_flags(&flags))
- {
- return flags;
- }
- mask = arm_cpu_env_mask();
- /* MSVC has no inline __asm support for ARM, but it does let you __emit
- * instructions via their assembled hex code.
- * All of these instructions should be essentially nops.
- */
-#if defined(HAVE_ARMV5TE)
- if (mask & HAS_EDSP)
- {
- __try
- {
- /*PLD [r13]*/
- __emit(0xF5DDF000);
- flags |= HAS_EDSP;
- }
- __except(GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION)
- {
- /*Ignore exception.*/
- }
- }
-#if defined(HAVE_ARMV6)
- if (mask & HAS_MEDIA)
- __try
- {
- /*SHADD8 r3,r3,r3*/
- __emit(0xE6333F93);
- flags |= HAS_MEDIA;
- }
- __except(GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION)
- {
- /*Ignore exception.*/
- }
- }
-#if defined(HAVE_ARMV7)
- if (mask & HAS_NEON)
- {
- __try
- {
- /*VORR q0,q0,q0*/
- __emit(0xF2200150);
- flags |= HAS_NEON;
- }
- __except(GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION)
- {
- /*Ignore exception.*/
- }
- }
-#endif
-#endif
-#endif
- return flags & mask;
-}
-
-#elif defined(__linux__)
-#include <stdio.h>
-
-int arm_cpu_caps(void)
-{
- FILE *fin;
- int flags;
- int mask;
- if (!arm_cpu_env_flags(&flags))
- {
- return flags;
- }
- mask = arm_cpu_env_mask();
- /* Reading /proc/self/auxv would be easier, but that doesn't work reliably
- * on Android.
- * This also means that detection will fail in Scratchbox.
- */
- fin = fopen("/proc/cpuinfo","r");
- if(fin != NULL)
- {
- /* 512 should be enough for anybody (it's even enough for all the flags
- * that x86 has accumulated... so far).
- */
- char buf[512];
- while (fgets(buf, 511, fin) != NULL)
- {
-#if defined(HAVE_ARMV5TE) || defined(HAVE_ARMV7)
- if (memcmp(buf, "Features", 8) == 0)
- {
- char *p;
-#if defined(HAVE_ARMV5TE)
- p=strstr(buf, " edsp");
- if (p != NULL && (p[5] == ' ' || p[5] == '\n'))
- {
- flags |= HAS_EDSP;
- }
-#if defined(HAVE_ARMV7)
- p = strstr(buf, " neon");
- if (p != NULL && (p[5] == ' ' || p[5] == '\n'))
- {
- flags |= HAS_NEON;
- }
-#endif
-#endif
- }
-#endif
-#if defined(HAVE_ARMV6)
- if (memcmp(buf, "CPU architecture:",17) == 0){
- int version;
- version = atoi(buf+17);
- if (version >= 6)
- {
- flags |= HAS_MEDIA;
- }
- }
-#endif
- }
- fclose(fin);
- }
- return flags & mask;
-}
-
-#elif !CONFIG_RUNTIME_CPU_DETECT
-
-int arm_cpu_caps(void)
-{
- int flags;
- int mask;
- if (!arm_cpu_env_flags(&flags))
- {
- return flags;
- }
- mask = arm_cpu_env_mask();
-#if defined(HAVE_ARMV5TE)
- flags |= HAS_EDSP;
-#endif
-#if defined(HAVE_ARMV6)
- flags |= HAS_MEDIA;
-#endif
-#if defined(HAVE_ARMV7)
- flags |= HAS_NEON;
-#endif
- return flags & mask;
-}
-
-#else
-#error "--enable-runtime-cpu-detect selected, but no CPU detection method " \
- "available for your platform. Reconfigure without --enable-runtime-cpu-detect."
-#endif
diff --git a/vpx_ports/config.h b/vpx_ports/config.h
deleted file mode 100644
index 1abe70d..0000000
--- a/vpx_ports/config.h
+++ /dev/null
@@ -1,10 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-#include "vpx_config.h"
diff --git a/vpx_ports/emms.asm b/vpx_ports/emms.asm
deleted file mode 100644
index 306e235..0000000
--- a/vpx_ports/emms.asm
+++ /dev/null
@@ -1,38 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-section .text
- global sym(vpx_reset_mmx_state)
-sym(vpx_reset_mmx_state):
- emms
- ret
-
-
-%ifidn __OUTPUT_FORMAT__,x64
-global sym(vpx_winx64_fldcw)
-sym(vpx_winx64_fldcw):
- sub rsp, 8
- mov [rsp], rcx ; win x64 specific
- fldcw [rsp]
- add rsp, 8
- ret
-
-
-global sym(vpx_winx64_fstcw)
-sym(vpx_winx64_fstcw):
- sub rsp, 8
- fstcw [rsp]
- mov rax, [rsp]
- add rsp, 8
- ret
-%endif
diff --git a/vpx_ports/mem.h b/vpx_ports/mem.h
deleted file mode 100644
index 9ec34fe..0000000
--- a/vpx_ports/mem.h
+++ /dev/null
@@ -1,46 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VPX_PORTS_MEM_H
-#define VPX_PORTS_MEM_H
-#include "vpx_config.h"
-#include "vpx/vpx_integer.h"
-
-#if defined(__GNUC__) && __GNUC__
-#define DECLARE_ALIGNED(n,typ,val) typ val __attribute__ ((aligned (n)))
-#elif defined(_MSC_VER)
-#define DECLARE_ALIGNED(n,typ,val) __declspec(align(n)) typ val
-#else
-#warning No alignment directives known for this compiler.
-#define DECLARE_ALIGNED(n,typ,val) typ val
-#endif
-#endif
-
-
-/* Declare an aligned array on the stack, for situations where the stack
- * pointer may not have the alignment we expect. Creates an array with a
- * modified name, then defines val to be a pointer, and aligns that pointer
- * within the array.
- */
-#define DECLARE_ALIGNED_ARRAY(a,typ,val,n)\
-typ val##_[(n)+(a)/sizeof(typ)+1];\
-typ *val = (typ*)((((intptr_t)val##_)+(a)-1)&((intptr_t)-(a)))
-
-
-/* Indicates that the usage of the specified variable has been audited to assure
- * that it's safe to use uninitialized. Silences 'may be used uninitialized'
- * warnings on gcc.
- */
-#if defined(__GNUC__) && __GNUC__
-#define UNINITIALIZED_IS_SAFE(x) x=x
-#else
-#define UNINITIALIZED_IS_SAFE(x) x
-#endif
diff --git a/vpx_ports/mem_ops.h b/vpx_ports/mem_ops.h
deleted file mode 100644
index 0e52368..0000000
--- a/vpx_ports/mem_ops.h
+++ /dev/null
@@ -1,234 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/* \file
- * \brief Provides portable memory access primitives
- *
- * This function provides portable primitives for getting and setting of
- * signed and unsigned integers in 16, 24, and 32 bit sizes. The operations
- * can be performed on unaligned data regardless of hardware support for
- * unaligned accesses.
- *
- * The type used to pass the integral values may be changed by defining
- * MEM_VALUE_T with the appropriate type. The type given must be an integral
- * numeric type.
- *
- * The actual functions instantiated have the MEM_VALUE_T type name pasted
- * on to the symbol name. This allows the developer to instantiate these
- * operations for multiple types within the same translation unit. This is
- * of somewhat questionable utility, but the capability exists nonetheless.
- * Users not making use of this functionality should call the functions
- * without the type name appended, and the preprocessor will take care of
- * it.
- *
- * NOTE: This code is not supported on platforms where char > 1 octet ATM.
- */
-
-#ifndef MAU_T
-/* Minimum Access Unit for this target */
-#define MAU_T unsigned char
-#endif
-
-#ifndef MEM_VALUE_T
-#define MEM_VALUE_T int
-#endif
-
-#undef MEM_VALUE_T_SZ_BITS
-#define MEM_VALUE_T_SZ_BITS (sizeof(MEM_VALUE_T) << 3)
-
-#undef mem_ops_wrap_symbol
-#define mem_ops_wrap_symbol(fn) mem_ops_wrap_symbol2(fn, MEM_VALUE_T)
-#undef mem_ops_wrap_symbol2
-#define mem_ops_wrap_symbol2(fn,typ) mem_ops_wrap_symbol3(fn,typ)
-#undef mem_ops_wrap_symbol3
-#define mem_ops_wrap_symbol3(fn,typ) fn##_as_##typ
-
-/*
- * Include aligned access routines
- */
-#define INCLUDED_BY_MEM_OPS_H
-#include "mem_ops_aligned.h"
-#undef INCLUDED_BY_MEM_OPS_H
-
-#undef mem_get_be16
-#define mem_get_be16 mem_ops_wrap_symbol(mem_get_be16)
-static unsigned MEM_VALUE_T mem_get_be16(const void *vmem)
-{
- unsigned MEM_VALUE_T val;
- const MAU_T *mem = (const MAU_T *)vmem;
-
- val = mem[0] << 8;
- val |= mem[1];
- return val;
-}
-
-#undef mem_get_be24
-#define mem_get_be24 mem_ops_wrap_symbol(mem_get_be24)
-static unsigned MEM_VALUE_T mem_get_be24(const void *vmem)
-{
- unsigned MEM_VALUE_T val;
- const MAU_T *mem = (const MAU_T *)vmem;
-
- val = mem[0] << 16;
- val |= mem[1] << 8;
- val |= mem[2];
- return val;
-}
-
-#undef mem_get_be32
-#define mem_get_be32 mem_ops_wrap_symbol(mem_get_be32)
-static unsigned MEM_VALUE_T mem_get_be32(const void *vmem)
-{
- unsigned MEM_VALUE_T val;
- const MAU_T *mem = (const MAU_T *)vmem;
-
- val = mem[0] << 24;
- val |= mem[1] << 16;
- val |= mem[2] << 8;
- val |= mem[3];
- return val;
-}
-
-#undef mem_get_le16
-#define mem_get_le16 mem_ops_wrap_symbol(mem_get_le16)
-static unsigned MEM_VALUE_T mem_get_le16(const void *vmem)
-{
- unsigned MEM_VALUE_T val;
- const MAU_T *mem = (const MAU_T *)vmem;
-
- val = mem[1] << 8;
- val |= mem[0];
- return val;
-}
-
-#undef mem_get_le24
-#define mem_get_le24 mem_ops_wrap_symbol(mem_get_le24)
-static unsigned MEM_VALUE_T mem_get_le24(const void *vmem)
-{
- unsigned MEM_VALUE_T val;
- const MAU_T *mem = (const MAU_T *)vmem;
-
- val = mem[2] << 16;
- val |= mem[1] << 8;
- val |= mem[0];
- return val;
-}
-
-#undef mem_get_le32
-#define mem_get_le32 mem_ops_wrap_symbol(mem_get_le32)
-static unsigned MEM_VALUE_T mem_get_le32(const void *vmem)
-{
- unsigned MEM_VALUE_T val;
- const MAU_T *mem = (const MAU_T *)vmem;
-
- val = mem[3] << 24;
- val |= mem[2] << 16;
- val |= mem[1] << 8;
- val |= mem[0];
- return val;
-}
-
-#define mem_get_s_generic(end,sz) \
- static signed MEM_VALUE_T mem_get_s##end##sz(const void *vmem) {\
- const MAU_T *mem = (const MAU_T*)vmem;\
- signed MEM_VALUE_T val = mem_get_##end##sz(mem);\
- return (val << (MEM_VALUE_T_SZ_BITS - sz)) >> (MEM_VALUE_T_SZ_BITS - sz);\
- }
-
-#undef mem_get_sbe16
-#define mem_get_sbe16 mem_ops_wrap_symbol(mem_get_sbe16)
-mem_get_s_generic(be, 16);
-
-#undef mem_get_sbe24
-#define mem_get_sbe24 mem_ops_wrap_symbol(mem_get_sbe24)
-mem_get_s_generic(be, 24);
-
-#undef mem_get_sbe32
-#define mem_get_sbe32 mem_ops_wrap_symbol(mem_get_sbe32)
-mem_get_s_generic(be, 32);
-
-#undef mem_get_sle16
-#define mem_get_sle16 mem_ops_wrap_symbol(mem_get_sle16)
-mem_get_s_generic(le, 16);
-
-#undef mem_get_sle24
-#define mem_get_sle24 mem_ops_wrap_symbol(mem_get_sle24)
-mem_get_s_generic(le, 24);
-
-#undef mem_get_sle32
-#define mem_get_sle32 mem_ops_wrap_symbol(mem_get_sle32)
-mem_get_s_generic(le, 32);
-
-#undef mem_put_be16
-#define mem_put_be16 mem_ops_wrap_symbol(mem_put_be16)
-static void mem_put_be16(void *vmem, MEM_VALUE_T val)
-{
- MAU_T *mem = (MAU_T *)vmem;
-
- mem[0] = (val >> 8) & 0xff;
- mem[1] = (val >> 0) & 0xff;
-}
-
-#undef mem_put_be24
-#define mem_put_be24 mem_ops_wrap_symbol(mem_put_be24)
-static void mem_put_be24(void *vmem, MEM_VALUE_T val)
-{
- MAU_T *mem = (MAU_T *)vmem;
-
- mem[0] = (val >> 16) & 0xff;
- mem[1] = (val >> 8) & 0xff;
- mem[2] = (val >> 0) & 0xff;
-}
-
-#undef mem_put_be32
-#define mem_put_be32 mem_ops_wrap_symbol(mem_put_be32)
-static void mem_put_be32(void *vmem, MEM_VALUE_T val)
-{
- MAU_T *mem = (MAU_T *)vmem;
-
- mem[0] = (val >> 24) & 0xff;
- mem[1] = (val >> 16) & 0xff;
- mem[2] = (val >> 8) & 0xff;
- mem[3] = (val >> 0) & 0xff;
-}
-
-#undef mem_put_le16
-#define mem_put_le16 mem_ops_wrap_symbol(mem_put_le16)
-static void mem_put_le16(void *vmem, MEM_VALUE_T val)
-{
- MAU_T *mem = (MAU_T *)vmem;
-
- mem[0] = (val >> 0) & 0xff;
- mem[1] = (val >> 8) & 0xff;
-}
-
-#undef mem_put_le24
-#define mem_put_le24 mem_ops_wrap_symbol(mem_put_le24)
-static void mem_put_le24(void *vmem, MEM_VALUE_T val)
-{
- MAU_T *mem = (MAU_T *)vmem;
-
- mem[0] = (val >> 0) & 0xff;
- mem[1] = (val >> 8) & 0xff;
- mem[2] = (val >> 16) & 0xff;
-}
-
-#undef mem_put_le32
-#define mem_put_le32 mem_ops_wrap_symbol(mem_put_le32)
-static void mem_put_le32(void *vmem, MEM_VALUE_T val)
-{
- MAU_T *mem = (MAU_T *)vmem;
-
- mem[0] = (val >> 0) & 0xff;
- mem[1] = (val >> 8) & 0xff;
- mem[2] = (val >> 16) & 0xff;
- mem[3] = (val >> 24) & 0xff;
-}
diff --git a/vpx_ports/mem_ops_aligned.h b/vpx_ports/mem_ops_aligned.h
deleted file mode 100644
index 82a18b2..0000000
--- a/vpx_ports/mem_ops_aligned.h
+++ /dev/null
@@ -1,157 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/* \file
- * \brief Provides portable memory access primitives for operating on aligned
- * data
- *
- * This file is split from mem_ops.h for easier maintenance. See mem_ops.h
- * for a more detailed description of these primitives.
- */
-#ifndef INCLUDED_BY_MEM_OPS_H
-#error Include mem_ops.h, not mem_ops_aligned.h directly.
-#endif
-
-/* Architectures that provide instructions for doing this byte swapping
- * could redefine these macros.
- */
-#define swap_endian_16(val,raw) do {\
- val = ((raw>>8) & 0x00ff) \
- | ((raw<<8) & 0xff00);\
- } while(0)
-#define swap_endian_32(val,raw) do {\
- val = ((raw>>24) & 0x000000ff) \
- | ((raw>>8) & 0x0000ff00) \
- | ((raw<<8) & 0x00ff0000) \
- | ((raw<<24) & 0xff000000); \
- } while(0)
-#define swap_endian_16_se(val,raw) do {\
- swap_endian_16(val,raw);\
- val = ((val << 16) >> 16);\
- } while(0)
-#define swap_endian_32_se(val,raw) swap_endian_32(val,raw)
-
-#define mem_get_ne_aligned_generic(end,sz) \
- static unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\
- const uint##sz##_t *mem = (const uint##sz##_t *)vmem;\
- return *mem;\
- }
-
-#define mem_get_sne_aligned_generic(end,sz) \
- static signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\
- const int##sz##_t *mem = (const int##sz##_t *)vmem;\
- return *mem;\
- }
-
-#define mem_get_se_aligned_generic(end,sz) \
- static unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\
- const uint##sz##_t *mem = (const uint##sz##_t *)vmem;\
- unsigned MEM_VALUE_T val, raw = *mem;\
- swap_endian_##sz(val,raw);\
- return val;\
- }
-
-#define mem_get_sse_aligned_generic(end,sz) \
- static signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\
- const int##sz##_t *mem = (const int##sz##_t *)vmem;\
- unsigned MEM_VALUE_T val, raw = *mem;\
- swap_endian_##sz##_se(val,raw);\
- return val;\
- }
-
-#define mem_put_ne_aligned_generic(end,sz) \
- static void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\
- uint##sz##_t *mem = (uint##sz##_t *)vmem;\
- *mem = (uint##sz##_t)val;\
- }
-
-#define mem_put_se_aligned_generic(end,sz) \
- static void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\
- uint##sz##_t *mem = (uint##sz##_t *)vmem, raw;\
- swap_endian_##sz(raw,val);\
- *mem = (uint##sz##_t)raw;\
- }
-
-#include "config.h"
-#if CONFIG_BIG_ENDIAN
-#define mem_get_be_aligned_generic(sz) mem_get_ne_aligned_generic(be,sz)
-#define mem_get_sbe_aligned_generic(sz) mem_get_sne_aligned_generic(be,sz)
-#define mem_get_le_aligned_generic(sz) mem_get_se_aligned_generic(le,sz)
-#define mem_get_sle_aligned_generic(sz) mem_get_sse_aligned_generic(le,sz)
-#define mem_put_be_aligned_generic(sz) mem_put_ne_aligned_generic(be,sz)
-#define mem_put_le_aligned_generic(sz) mem_put_se_aligned_generic(le,sz)
-#else
-#define mem_get_be_aligned_generic(sz) mem_get_se_aligned_generic(be,sz)
-#define mem_get_sbe_aligned_generic(sz) mem_get_sse_aligned_generic(be,sz)
-#define mem_get_le_aligned_generic(sz) mem_get_ne_aligned_generic(le,sz)
-#define mem_get_sle_aligned_generic(sz) mem_get_sne_aligned_generic(le,sz)
-#define mem_put_be_aligned_generic(sz) mem_put_se_aligned_generic(be,sz)
-#define mem_put_le_aligned_generic(sz) mem_put_ne_aligned_generic(le,sz)
-#endif
-
-#undef mem_get_be16_aligned
-#define mem_get_be16_aligned mem_ops_wrap_symbol(mem_get_be16_aligned)
-mem_get_be_aligned_generic(16);
-
-#undef mem_get_be32_aligned
-#define mem_get_be32_aligned mem_ops_wrap_symbol(mem_get_be32_aligned)
-mem_get_be_aligned_generic(32);
-
-#undef mem_get_le16_aligned
-#define mem_get_le16_aligned mem_ops_wrap_symbol(mem_get_le16_aligned)
-mem_get_le_aligned_generic(16);
-
-#undef mem_get_le32_aligned
-#define mem_get_le32_aligned mem_ops_wrap_symbol(mem_get_le32_aligned)
-mem_get_le_aligned_generic(32);
-
-#undef mem_get_sbe16_aligned
-#define mem_get_sbe16_aligned mem_ops_wrap_symbol(mem_get_sbe16_aligned)
-mem_get_sbe_aligned_generic(16);
-
-#undef mem_get_sbe32_aligned
-#define mem_get_sbe32_aligned mem_ops_wrap_symbol(mem_get_sbe32_aligned)
-mem_get_sbe_aligned_generic(32);
-
-#undef mem_get_sle16_aligned
-#define mem_get_sle16_aligned mem_ops_wrap_symbol(mem_get_sle16_aligned)
-mem_get_sle_aligned_generic(16);
-
-#undef mem_get_sle32_aligned
-#define mem_get_sle32_aligned mem_ops_wrap_symbol(mem_get_sle32_aligned)
-mem_get_sle_aligned_generic(32);
-
-#undef mem_put_be16_aligned
-#define mem_put_be16_aligned mem_ops_wrap_symbol(mem_put_be16_aligned)
-mem_put_be_aligned_generic(16);
-
-#undef mem_put_be32_aligned
-#define mem_put_be32_aligned mem_ops_wrap_symbol(mem_put_be32_aligned)
-mem_put_be_aligned_generic(32);
-
-#undef mem_put_le16_aligned
-#define mem_put_le16_aligned mem_ops_wrap_symbol(mem_put_le16_aligned)
-mem_put_le_aligned_generic(16);
-
-#undef mem_put_le32_aligned
-#define mem_put_le32_aligned mem_ops_wrap_symbol(mem_put_le32_aligned)
-mem_put_le_aligned_generic(32);
-
-#undef mem_get_ne_aligned_generic
-#undef mem_get_se_aligned_generic
-#undef mem_get_sne_aligned_generic
-#undef mem_get_sse_aligned_generic
-#undef mem_put_ne_aligned_generic
-#undef mem_put_se_aligned_generic
-#undef swap_endian_16
-#undef swap_endian_32
-#undef swap_endian_16_se
-#undef swap_endian_32_se
diff --git a/vpx_ports/vpx_timer.h b/vpx_ports/vpx_timer.h
deleted file mode 100644
index c8335a0..0000000
--- a/vpx_ports/vpx_timer.h
+++ /dev/null
@@ -1,121 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VPX_TIMER_H
-#define VPX_TIMER_H
-
-#if CONFIG_OS_SUPPORT
-
-#if defined(_WIN32)
-/*
- * Win32 specific includes
- */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif
-#include <windows.h>
-#else
-/*
- * POSIX specific includes
- */
-#include <sys/time.h>
-
-/* timersub is not provided by msys at this time. */
-#ifndef timersub
-#define timersub(a, b, result) \
- do { \
- (result)->tv_sec = (a)->tv_sec - (b)->tv_sec; \
- (result)->tv_usec = (a)->tv_usec - (b)->tv_usec; \
- if ((result)->tv_usec < 0) { \
- --(result)->tv_sec; \
- (result)->tv_usec += 1000000; \
- } \
- } while (0)
-#endif
-#endif
-
-
-struct vpx_usec_timer
-{
-#if defined(_WIN32)
- LARGE_INTEGER begin, end;
-#else
- struct timeval begin, end;
-#endif
-};
-
-
-static void
-vpx_usec_timer_start(struct vpx_usec_timer *t)
-{
-#if defined(_WIN32)
- QueryPerformanceCounter(&t->begin);
-#else
- gettimeofday(&t->begin, NULL);
-#endif
-}
-
-
-static void
-vpx_usec_timer_mark(struct vpx_usec_timer *t)
-{
-#if defined(_WIN32)
- QueryPerformanceCounter(&t->end);
-#else
- gettimeofday(&t->end, NULL);
-#endif
-}
-
-
-static long
-vpx_usec_timer_elapsed(struct vpx_usec_timer *t)
-{
-#if defined(_WIN32)
- LARGE_INTEGER freq, diff;
-
- diff.QuadPart = t->end.QuadPart - t->begin.QuadPart;
-
- if (QueryPerformanceFrequency(&freq) && diff.QuadPart < freq.QuadPart)
- return (long)(diff.QuadPart * 1000000 / freq.QuadPart);
-
- return 1000000;
-#else
- struct timeval diff;
-
- timersub(&t->end, &t->begin, &diff);
- return diff.tv_sec ? 1000000 : diff.tv_usec;
-#endif
-}
-
-#else /* CONFIG_OS_SUPPORT = 0*/
-
-/* Empty timer functions if CONFIG_OS_SUPPORT = 0 */
-#ifndef timersub
-#define timersub(a, b, result)
-#endif
-
-struct vpx_usec_timer
-{
- void *dummy;
-};
-
-static void
-vpx_usec_timer_start(struct vpx_usec_timer *t) { }
-
-static void
-vpx_usec_timer_mark(struct vpx_usec_timer *t) { }
-
-static long
-vpx_usec_timer_elapsed(struct vpx_usec_timer *t) { return 0; }
-
-#endif /* CONFIG_OS_SUPPORT */
-
-#endif
diff --git a/vpx_ports/vpxtypes.h b/vpx_ports/vpxtypes.h
deleted file mode 100644
index 2ab66b1..0000000
--- a/vpx_ports/vpxtypes.h
+++ /dev/null
@@ -1,172 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __VPXTYPES_H__
-#define __VPXTYPES_H__
-
-#include "vpx_ports/config.h"
-
-//#include <sys/types.h>
-#ifdef _MSC_VER
-# include <basetsd.h>
-typedef SSIZE_T ssize_t;
-#endif
-
-#if defined(HAVE_STDINT_H) && HAVE_STDINT_H
-/* C99 types are preferred to vpx integer types */
-# include <stdint.h>
-#endif
-
-/*!\defgroup basetypes Base Types
- @{*/
-#if !defined(HAVE_STDINT_H) && !defined(INT_T_DEFINED)
-# ifdef STRICTTYPES
-typedef signed char int8_t;
-typedef signed short int16_t;
-typedef signed int int32_t;
-# else
-typedef char int8_t;
-typedef short int16_t;
-typedef int int32_t;
-# endif
-typedef unsigned char uint8_t;
-typedef unsigned short uint16_t;
-typedef unsigned int uint32_t;
-#endif
-
-typedef int8_t vpxs8;
-typedef uint8_t vpxu8;
-typedef int16_t vpxs16;
-typedef uint16_t vpxu16;
-typedef int32_t vpxs32;
-typedef uint32_t vpxu32;
-typedef int32_t vpxbool;
-
-enum {vpxfalse, vpxtrue};
-
-/*!\def OTC
- \brief a macro suitable for declaring a constant #vpxtc*/
-/*!\def VPXTC
- \brief printf format string suitable for printing an #vpxtc*/
-#ifdef UNICODE
-# ifdef NO_WCHAR
-# error "no non-wchar support added yet"
-# else
-# include <wchar.h>
-typedef wchar_t vpxtc;
-# define OTC(str) L ## str
-# define VPXTC "ls"
-# endif /*NO_WCHAR*/
-#else
-typedef char vpxtc;
-# define OTC(str) (vpxtc*)str
-# define VPXTC "s"
-#endif /*UNICODE*/
-/*@} end - base types*/
-
-/*!\addtogroup basetypes
- @{*/
-/*!\def VPX64
- \brief printf format string suitable for printing an #vpxs64*/
-#if defined(HAVE_STDINT_H)
-# define VPX64 PRId64
-typedef int64_t vpxs64;
-#elif defined(HASLONGLONG)
-# undef PRId64
-# define PRId64 "lld"
-# define VPX64 PRId64
-typedef long long vpxs64;
-#elif defined(WIN32) || defined(_WIN32_WCE)
-# undef PRId64
-# define PRId64 "I64d"
-# define VPX64 PRId64
-typedef __int64 vpxs64;
-typedef unsigned __int64 vpxu64;
-#elif defined(__uClinux__) && defined(CHIP_DM642)
-# include <lddk.h>
-# undef PRId64
-# define PRId64 "lld"
-# define VPX64 PRId64
-typedef long vpxs64;
-#elif defined(__SYMBIAN32__)
-# undef PRId64
-# define PRId64 "u"
-# define VPX64 PRId64
-typedef unsigned int vpxs64;
-#else
-# error "64 bit integer type undefined for this platform!"
-#endif
-#if !defined(HAVE_STDINT_H) && !defined(INT_T_DEFINED)
-typedef vpxs64 int64_t;
-typedef vpxu64 uint64_t;
-#endif
-/*!@} end - base types*/
-
-/*!\ingroup basetypes
- \brief Common return type*/
-typedef enum
-{
- VPX_NOT_FOUND = -404,
- VPX_BUFFER_EMPTY = -202,
- VPX_BUFFER_FULL = -201,
-
- VPX_CONNREFUSED = -102,
- VPX_TIMEDOUT = -101,
- VPX_WOULDBLOCK = -100,
-
- VPX_NET_ERROR = -9,
- VPX_INVALID_VERSION = -8,
- VPX_INPROGRESS = -7,
- VPX_NOT_SUPP = -6,
- VPX_NO_MEM = -3,
- VPX_INVALID_PARAMS = -2,
- VPX_ERROR = -1,
- VPX_OK = 0,
- VPX_DONE = 1
-} vpxsc;
-
-#if defined(WIN32) || defined(_WIN32_WCE)
-# define DLLIMPORT __declspec(dllimport)
-# define DLLEXPORT __declspec(dllexport)
-# define DLLLOCAL
-#elif defined(LINUX)
-# define DLLIMPORT
-/*visibility attribute support is available in 3.4 and later.
- see: http://gcc.gnu.org/wiki/Visibility for more info*/
-# if defined(__GNUC__) && ((__GNUC__<<16|(__GNUC_MINOR__&0xff)) >= (3<<16|4))
-# define GCC_HASCLASSVISIBILITY
-# endif /*defined(__GNUC__) && __GNUC_PREREQ(3,4)*/
-# ifdef GCC_HASCLASSVISIBILITY
-# define DLLEXPORT __attribute__ ((visibility("default")))
-# define DLLLOCAL __attribute__ ((visibility("hidden")))
-# else
-# define DLLEXPORT
-# define DLLLOCAL
-# endif /*GCC_HASCLASSVISIBILITY*/
-#endif /*platform ifdefs*/
-
-#endif /*__VPXTYPES_H__*/
-
-#undef VPXAPI
-/*!\def VPXAPI
- \brief library calling convention/storage class attributes.
-
- Specifies whether the function is imported through a dll
- or is from a static library.*/
-#ifdef VPXDLL
-# ifdef VPXDLLEXPORT
-# define VPXAPI DLLEXPORT
-# else
-# define VPXAPI DLLIMPORT
-# endif /*VPXDLLEXPORT*/
-#else
-# define VPXAPI
-#endif /*VPXDLL*/
diff --git a/vpx_ports/x86.h b/vpx_ports/x86.h
deleted file mode 100644
index 190c864..0000000
--- a/vpx_ports/x86.h
+++ /dev/null
@@ -1,207 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VPX_PORTS_X86_H
-#define VPX_PORTS_X86_H
-#include <stdlib.h>
-#include "config.h"
-
-typedef enum
-{
- VPX_CPU_UNKNOWN = -1,
- VPX_CPU_AMD,
- VPX_CPU_AMD_OLD,
- VPX_CPU_CENTAUR,
- VPX_CPU_CYRIX,
- VPX_CPU_INTEL,
- VPX_CPU_NEXGEN,
- VPX_CPU_NSC,
- VPX_CPU_RISE,
- VPX_CPU_SIS,
- VPX_CPU_TRANSMETA,
- VPX_CPU_TRANSMETA_OLD,
- VPX_CPU_UMC,
- VPX_CPU_VIA,
-
- VPX_CPU_LAST
-} vpx_cpu_t;
-
-#if defined(__GNUC__) && __GNUC__
-#if ARCH_X86_64
-#define cpuid(func,ax,bx,cx,dx)\
- __asm__ __volatile__ (\
- "cpuid \n\t" \
- : "=a" (ax), "=b" (bx), "=c" (cx), "=d" (dx) \
- : "a" (func));
-#else
-#define cpuid(func,ax,bx,cx,dx)\
- __asm__ __volatile__ (\
- "mov %%ebx, %%edi \n\t" \
- "cpuid \n\t" \
- "xchg %%edi, %%ebx \n\t" \
- : "=a" (ax), "=D" (bx), "=c" (cx), "=d" (dx) \
- : "a" (func));
-#endif
-#else
-#if ARCH_X86_64
-void __cpuid(int CPUInfo[4], int info_type);
-#pragma intrinsic(__cpuid)
-#define cpuid(func,a,b,c,d) do{\
- int regs[4];\
- __cpuid(regs,func); a=regs[0]; b=regs[1]; c=regs[2]; d=regs[3];\
- } while(0)
-#else
-#define cpuid(func,a,b,c,d)\
- __asm mov eax, func\
- __asm cpuid\
- __asm mov a, eax\
- __asm mov b, ebx\
- __asm mov c, ecx\
- __asm mov d, edx
-#endif
-#endif
-
-#define HAS_MMX 0x01
-#define HAS_SSE 0x02
-#define HAS_SSE2 0x04
-#define HAS_SSE3 0x08
-#define HAS_SSSE3 0x10
-#define HAS_SSE4_1 0x20
-#ifndef BIT
-#define BIT(n) (1<<n)
-#endif
-
-static int
-x86_simd_caps(void)
-{
- unsigned int flags = 0;
- unsigned int mask = ~0;
- unsigned int reg_eax, reg_ebx, reg_ecx, reg_edx;
- char *env;
- (void)reg_ebx;
-
- /* See if the CPU capabilities are being overridden by the environment */
- env = getenv("VPX_SIMD_CAPS");
-
- if (env && *env)
- return (int)strtol(env, NULL, 0);
-
- env = getenv("VPX_SIMD_CAPS_MASK");
-
- if (env && *env)
- mask = strtol(env, NULL, 0);
-
- /* Ensure that the CPUID instruction supports extended features */
- cpuid(0, reg_eax, reg_ebx, reg_ecx, reg_edx);
-
- if (reg_eax < 1)
- return 0;
-
- /* Get the standard feature flags */
- cpuid(1, reg_eax, reg_ebx, reg_ecx, reg_edx);
-
- if (reg_edx & BIT(23)) flags |= HAS_MMX;
-
- if (reg_edx & BIT(25)) flags |= HAS_SSE; /* aka xmm */
-
- if (reg_edx & BIT(26)) flags |= HAS_SSE2; /* aka wmt */
-
- if (reg_ecx & BIT(0)) flags |= HAS_SSE3;
-
- if (reg_ecx & BIT(9)) flags |= HAS_SSSE3;
-
- if (reg_ecx & BIT(19)) flags |= HAS_SSE4_1;
-
- return flags & mask;
-}
-
-vpx_cpu_t vpx_x86_vendor(void);
-
-#if ARCH_X86_64 && defined(_MSC_VER)
-unsigned __int64 __rdtsc(void);
-#pragma intrinsic(__rdtsc)
-#endif
-static unsigned int
-x86_readtsc(void)
-{
-#if defined(__GNUC__) && __GNUC__
- unsigned int tsc;
- __asm__ __volatile__("rdtsc\n\t":"=a"(tsc):);
- return tsc;
-#else
-#if ARCH_X86_64
- return __rdtsc();
-#else
- __asm rdtsc;
-#endif
-#endif
-}
-
-
-#if defined(__GNUC__) && __GNUC__
-#define x86_pause_hint()\
- __asm__ __volatile__ ("pause \n\t")
-#else
-#if ARCH_X86_64
-/* No pause intrinsic for windows x64 */
-#define x86_pause_hint()
-#else
-#define x86_pause_hint()\
- __asm pause
-#endif
-#endif
-
-#if defined(__GNUC__) && __GNUC__
-static void
-x87_set_control_word(unsigned short mode)
-{
- __asm__ __volatile__("fldcw %0" : : "m"(*&mode));
-}
-static unsigned short
-x87_get_control_word(void)
-{
- unsigned short mode;
- __asm__ __volatile__("fstcw %0\n\t":"=m"(*&mode):);
- return mode;
-}
-#elif ARCH_X86_64
-/* No fldcw intrinsics on Windows x64, punt to external asm */
-extern void vpx_winx64_fldcw(unsigned short mode);
-extern unsigned short vpx_winx64_fstcw(void);
-#define x87_set_control_word vpx_winx64_fldcw
-#define x87_get_control_word vpx_winx64_fstcw
-#else
-static void
-x87_set_control_word(unsigned short mode)
-{
- __asm { fldcw mode }
-}
-static unsigned short
-x87_get_control_word(void)
-{
- unsigned short mode;
- __asm { fstcw mode }
- return mode;
-}
-#endif
-
-static unsigned short
-x87_set_double_precision(void)
-{
- unsigned short mode = x87_get_control_word();
- x87_set_control_word((mode&~0x300) | 0x200);
- return mode;
-}
-
-
-extern void vpx_reset_mmx_state(void);
-#endif
-
diff --git a/vpx_ports/x86_abi_support.asm b/vpx_ports/x86_abi_support.asm
deleted file mode 100644
index be64cd7..0000000
--- a/vpx_ports/x86_abi_support.asm
+++ /dev/null
@@ -1,301 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
-%include "vpx_config.asm"
-
-; 32/64 bit compatibility macros
-;
-; In general, we make the source use 64 bit syntax, then twiddle with it using
-; the preprocessor to get the 32 bit syntax on 32 bit platforms.
-;
-%ifidn __OUTPUT_FORMAT__,elf32
-%define ABI_IS_32BIT 1
-%elifidn __OUTPUT_FORMAT__,macho32
-%define ABI_IS_32BIT 1
-%elifidn __OUTPUT_FORMAT__,win32
-%define ABI_IS_32BIT 1
-%else
-%define ABI_IS_32BIT 0
-%endif
-
-%if ABI_IS_32BIT
-%define rax eax
-%define rbx ebx
-%define rcx ecx
-%define rdx edx
-%define rsi esi
-%define rdi edi
-%define rsp esp
-%define rbp ebp
-%define movsxd mov
-%macro movq 2
- %ifidn %1,eax
- movd %1,%2
- %elifidn %2,eax
- movd %1,%2
- %elifidn %1,ebx
- movd %1,%2
- %elifidn %2,ebx
- movd %1,%2
- %elifidn %1,ecx
- movd %1,%2
- %elifidn %2,ecx
- movd %1,%2
- %elifidn %1,edx
- movd %1,%2
- %elifidn %2,edx
- movd %1,%2
- %elifidn %1,esi
- movd %1,%2
- %elifidn %2,esi
- movd %1,%2
- %elifidn %1,edi
- movd %1,%2
- %elifidn %2,edi
- movd %1,%2
- %elifidn %1,esp
- movd %1,%2
- %elifidn %2,esp
- movd %1,%2
- %elifidn %1,ebp
- movd %1,%2
- %elifidn %2,ebp
- movd %1,%2
- %else
- movq %1,%2
- %endif
-%endmacro
-%endif
-
-
-; sym()
-; Return the proper symbol name for the target ABI.
-;
-; Certain ABIs, notably MS COFF and Darwin MACH-O, require that symbols
-; with C linkage be prefixed with an underscore.
-;
-%ifidn __OUTPUT_FORMAT__,elf32
-%define sym(x) x
-%elifidn __OUTPUT_FORMAT__,elf64
-%define sym(x) x
-%elifidn __OUTPUT_FORMAT__,x64
-%define sym(x) x
-%else
-%define sym(x) _ %+ x
-%endif
-
-; arg()
-; Return the address specification of the given argument
-;
-%if ABI_IS_32BIT
- %define arg(x) [ebp+8+4*x]
-%else
- ; 64 bit ABI passes arguments in registers. This is a workaround to get up
- ; and running quickly. Relies on SHADOW_ARGS_TO_STACK
- %ifidn __OUTPUT_FORMAT__,x64
- %define arg(x) [rbp+16+8*x]
- %else
- %define arg(x) [rbp-8-8*x]
- %endif
-%endif
-
-; REG_SZ_BYTES, REG_SZ_BITS
-; Size of a register
-%if ABI_IS_32BIT
-%define REG_SZ_BYTES 4
-%define REG_SZ_BITS 32
-%else
-%define REG_SZ_BYTES 8
-%define REG_SZ_BITS 64
-%endif
-
-
-; ALIGN_STACK <alignment> <register>
-; This macro aligns the stack to the given alignment (in bytes). The stack
-; is left such that the previous value of the stack pointer is the first
-; argument on the stack (ie, the inverse of this macro is 'pop rsp.')
-; This macro uses one temporary register, which is not preserved, and thus
-; must be specified as an argument.
-%macro ALIGN_STACK 2
- mov %2, rsp
- and rsp, -%1
- lea rsp, [rsp - (%1 - REG_SZ_BYTES)]
- push %2
-%endmacro
-
-
-;
-; The Microsoft assembler tries to impose a certain amount of type safety in
-; its register usage. YASM doesn't recognize these directives, so we just
-; %define them away to maintain as much compatibility as possible with the
-; original inline assembler we're porting from.
-;
-%idefine PTR
-%idefine XMMWORD
-%idefine MMWORD
-
-; PIC macros
-;
-%if ABI_IS_32BIT
- %if CONFIG_PIC=1
- %ifidn __OUTPUT_FORMAT__,elf32
- %define WRT_PLT wrt ..plt
- %macro GET_GOT 1
- extern _GLOBAL_OFFSET_TABLE_
- push %1
- call %%get_got
- %%sub_offset:
- jmp %%exitGG
- %%get_got:
- mov %1, [esp]
- add %1, _GLOBAL_OFFSET_TABLE_ + $$ - %%sub_offset wrt ..gotpc
- ret
- %%exitGG:
- %undef GLOBAL
- %define GLOBAL(x) x + %1 wrt ..gotoff
- %undef RESTORE_GOT
- %define RESTORE_GOT pop %1
- %endmacro
- %elifidn __OUTPUT_FORMAT__,macho32
- %macro GET_GOT 1
- push %1
- call %%get_got
- %%get_got:
- pop %1
- %undef GLOBAL
- %define GLOBAL(x) x + %1 - %%get_got
- %undef RESTORE_GOT
- %define RESTORE_GOT pop %1
- %endmacro
- %endif
- %endif
- %define HIDDEN_DATA(x) x
-%else
- %macro GET_GOT 1
- %endmacro
- %define GLOBAL(x) rel x
- %ifidn __OUTPUT_FORMAT__,elf64
- %define WRT_PLT wrt ..plt
- %define HIDDEN_DATA(x) x:data hidden
- %else
- %define HIDDEN_DATA(x) x
- %endif
-%endif
-%ifnmacro GET_GOT
- %macro GET_GOT 1
- %endmacro
- %define GLOBAL(x) x
-%endif
-%ifndef RESTORE_GOT
-%define RESTORE_GOT
-%endif
-%ifndef WRT_PLT
-%define WRT_PLT
-%endif
-
-%if ABI_IS_32BIT
- %macro SHADOW_ARGS_TO_STACK 1
- %endm
- %define UNSHADOW_ARGS
-%else
-%ifidn __OUTPUT_FORMAT__,x64
- %macro SHADOW_ARGS_TO_STACK 1 ; argc
- %if %1 > 0
- mov arg(0),rcx
- %endif
- %if %1 > 1
- mov arg(1),rdx
- %endif
- %if %1 > 2
- mov arg(2),r8
- %endif
- %if %1 > 3
- mov arg(3),r9
- %endif
- %endm
-%else
- %macro SHADOW_ARGS_TO_STACK 1 ; argc
- %if %1 > 0
- push rdi
- %endif
- %if %1 > 1
- push rsi
- %endif
- %if %1 > 2
- push rdx
- %endif
- %if %1 > 3
- push rcx
- %endif
- %if %1 > 4
- push r8
- %endif
- %if %1 > 5
- push r9
- %endif
- %if %1 > 6
- %assign i %1-6
- %assign off 16
- %rep i
- mov rax,[rbp+off]
- push rax
- %assign off off+8
- %endrep
- %endif
- %endm
-%endif
- %define UNSHADOW_ARGS mov rsp, rbp
-%endif
-
-; must keep XMM6:XMM15 (libvpx uses XMM6 and XMM7) on Win64 ABI
-; rsp register has to be aligned
-%ifidn __OUTPUT_FORMAT__,x64
-%macro SAVE_XMM 0
- sub rsp, 32
- movdqa XMMWORD PTR [rsp], xmm6
- movdqa XMMWORD PTR [rsp+16], xmm7
-%endmacro
-%macro RESTORE_XMM 0
- movdqa xmm6, XMMWORD PTR [rsp]
- movdqa xmm7, XMMWORD PTR [rsp+16]
- add rsp, 32
-%endmacro
-%else
-%macro SAVE_XMM 0
-%endmacro
-%macro RESTORE_XMM 0
-%endmacro
-%endif
-
-; Name of the rodata section
-;
-; .rodata seems to be an elf-ism, as it doesn't work on OSX.
-;
-%ifidn __OUTPUT_FORMAT__,macho64
-%define SECTION_RODATA section .text
-%elifidn __OUTPUT_FORMAT__,macho32
-%macro SECTION_RODATA 0
-section .text
-%endmacro
-%else
-%define SECTION_RODATA section .rodata
-%endif
-
-
-; Tell GNU ld that we don't require an executable stack.
-%ifidn __OUTPUT_FORMAT__,elf32
-section .note.GNU-stack noalloc noexec nowrite progbits
-section .text
-%elifidn __OUTPUT_FORMAT__,elf64
-section .note.GNU-stack noalloc noexec nowrite progbits
-section .text
-%endif
-
diff --git a/vpx_scale/arm/armv4/gen_scalers_armv4.asm b/vpx_scale/arm/armv4/gen_scalers_armv4.asm
deleted file mode 100644
index e495184..0000000
--- a/vpx_scale/arm/armv4/gen_scalers_armv4.asm
+++ /dev/null
@@ -1,774 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |horizontal_line_4_5_scale_armv4|
- EXPORT |vertical_band_4_5_scale_armv4|
- EXPORT |horizontal_line_2_3_scale_armv4|
- EXPORT |vertical_band_2_3_scale_armv4|
- EXPORT |horizontal_line_3_5_scale_armv4|
- EXPORT |vertical_band_3_5_scale_armv4|
- EXPORT |horizontal_line_3_4_scale_armv4|
- EXPORT |vertical_band_3_4_scale_armv4|
- EXPORT |horizontal_line_1_2_scale_armv4|
- EXPORT |vertical_band_1_2_scale_armv4|
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
-src RN r0
-srcw RN r1
-dest RN r2
-mask RN r12
-c51_205 RN r10
-c102_154 RN r11
-;/****************************************************************************
-; *
-; * ROUTINE : horizontal_line_4_5_scale_armv4
-; *
-; * INPUTS : const unsigned char *source : Pointer to source data.
-; * unsigned int source_width : Stride of source.
-; * unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_width : Stride of destination (NOT USED).
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Copies horizontal line of pixels from source to
-; * destination scaling up by 4 to 5.
-; *
-; * SPECIAL NOTES : None.
-; *
-; ****************************************************************************/
-;void horizontal_line_4_5_scale_armv4
-;(
-; r0 = UINT8 *source
-; r1 = UINT32 source_width
-; r2 = UINT8 *dest
-; r3 = UINT32 dest_width
-;)
-|horizontal_line_4_5_scale_armv4| PROC
- stmdb sp!, {r4 - r11, lr}
-
- mov mask, #255 ; mask for selection
- ldr c51_205, =0x3300cd
- ldr c102_154, =0x66009a
-
- ldr r3, [src], #4
-
-hl45_loop
-
- and r4, r3, mask ; a = src[0]
- and r5, mask, r3, lsr #8 ; b = src[1]
- strb r4, [dest], #1
-
- orr r6, r4, r5, lsl #16 ; b | a
- and r7, mask, r3, lsr #16 ; c = src[2]
- mul r6, c51_205, r6 ; a * 51 + 205 * b
-
- orr r5, r5, r7, lsl #16 ; c | b
- mul r5, c102_154, r5 ; b * 102 + 154 * c
- add r6, r6, #0x8000
- and r8, mask, r3, lsr #24 ; d = src[3]
- mov r6, r6, lsr #24
- strb r6, [dest], #1
-
- orr r7, r8, r7, lsl #16 ; c | d
- mul r7, c102_154, r7 ; c * 154 + 102 * d
- add r5, r5, #0x8000
- ldr r3, [src], #4
- mov r5, r5, lsr #24
- strb r5, [dest], #1
-
- add r7, r7, #0x8000
- and r9, mask, r3 ; e = src[4]
- orr r9, r9, r8, lsl #16 ; d | e
- mul r9, c51_205, r9 ; d * 205 + 51 * e
- mov r7, r7, lsr #24
- strb r7, [dest], #1
-
- add r9, r9, #0x8000
- subs srcw, srcw, #4
- mov r9, r9, lsr #24
- strb r9, [dest], #1
-
- bne hl45_loop
-
- and r4, r3, mask
- and r5, mask, r3, lsl #8
- strb r4, [dest], #1
-
- orr r6, r4, r5, lsl #16 ; b | a
- mul r6, c51_205, r6
-
- and r7, mask, r3, lsl #16
- orr r5, r5, r7, lsl #16 ; c | b
- mul r5, c102_154, r5
- add r6, r6, #0x8000
- and r8, mask, r3, lsl #24
- mov r6, r6, lsr #24
- strb r6, [dest], #1
-
- orr r7, r8, r7, lsl #16 ; c | d
- mul r7, c102_154, r7
- add r5, r5, #0x8000
- mov r5, r5, lsr #24
- strb r5, [dest], #1
-
- add r7, r7, #0x8000
- mov r7, r7, lsr #24
- strb r7, [dest], #1
-
- ldrb r3, [src]
- strb r3, [dest], #1
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ;|vp8cx_horizontal_line_4_5_scale_c|
-
-;/****************************************************************************
-; *
-; * ROUTINE : vertical_band_4_5_scale_armv4
-; *
-; * INPUTS : unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_pitch : Stride of destination data.
-; * unsigned int dest_width : Width of destination data.
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Scales vertical band of pixels by scale 4 to 5. The
-; * height of the band scaled is 4-pixels.
-; *
-; * SPECIAL NOTES : The routine uses the first line of the band below
-; * the current band.
-; *
-; ****************************************************************************/
-;void vertical_band_4_5_scale_armv4
-;(
-; r0 = UINT8 *dest
-; r1 = UINT32 dest_pitch
-; r2 = UINT32 dest_width
-;)
-|vertical_band_4_5_scale_armv4| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr c51_205, =0x3300cd
- ldr c102_154, =0x66009a
-
-vl45_loop
- mov r3, src
- ldrb r4, [r3], r1 ; a = des [0]
- ldrb r5, [r3], r1 ; b = des [dest_pitch]
- ldrb r7, [r3], r1 ; c = des[dest_pitch*2]
- add lr, src, r1
-
- orr r6, r4, r5, lsl #16 ; b | a
- mul r6, c51_205, r6 ; a * 51 + 205 * b
-
- ldrb r8, [r3], r1 ; d = des[dest_pitch*3]
- orr r5, r5, r7, lsl #16 ; c | b
- mul r5, c102_154, r5 ; b * 102 + 154 * c
- add r6, r6, #0x8000
- orr r7, r8, r7, lsl #16 ; c | d
- mov r6, r6, lsr #24
- strb r6, [lr], r1
-
- ldrb r9, [r3, r1] ; e = des [dest_pitch * 5]
- mul r7, c102_154, r7 ; c * 154 + 102 * d
- add r5, r5, #0x8000
- orr r9, r9, r8, lsl #16 ; d | e
- mov r5, r5, lsr #24
- strb r5, [lr], r1
-
- mul r9, c51_205, r9 ; d * 205 + 51 * e
- add r7, r7, #0x8000
- add src, src, #1
- mov r7, r7, lsr #24
- strb r7, [lr], r1
-
- add r9, r9, #0x8000
- subs r2, r2, #1
- mov r9, r9, lsr #24
- strb r9, [lr], r1
-
- bne vl45_loop
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ;|vertical_band_4_5_scale_armv4|
-
-;/****************************************************************************
-; *
-; * ROUTINE : horizontal_line_2_3_scale_armv4
-; *
-; * INPUTS : const unsigned char *source : Pointer to source data.
-; * unsigned int source_width : Stride of source.
-; * unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_width : Stride of destination (NOT USED).
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Copies horizontal line of pixels from source to
-; * destination scaling up by 2 to 3.
-; *
-; * SPECIAL NOTES : None.
-; *
-; *
-; ****************************************************************************/
-;void horizontal_line_2_3_scale_armv4
-;(
-; const unsigned char *source,
-; unsigned int source_width,
-; unsigned char *dest,
-; unsigned int dest_width
-;)
-|horizontal_line_2_3_scale_armv4| PROC
- stmdb sp!, {r4 - r11, lr}
- ldr lr, =85
- ldr r12, =171
-
-hl23_loop
-
- ldrb r3, [src], #1 ; a
- ldrb r4, [src], #1 ; b
- ldrb r5, [src] ; c
-
- strb r3, [dest], #1
- mul r4, r12, r4 ; b * 171
- mla r6, lr, r3, r4 ; a * 85
- mla r7, lr, r5, r4 ; c * 85
-
- add r6, r6, #128
- mov r6, r6, lsr #8
- strb r6, [dest], #1
-
- add r7, r7, #128
- mov r7, r7, lsr #8
- strb r7, [dest], #1
-
- subs srcw, srcw, #2
- bne hl23_loop
-
- ldrb r4, [src, #1] ; b
- strb r5, [dest], #1
- strb r4, [dest, #1]
-
- mul r4, r12, r4 ; b * 171
- mla r6, lr, r5, r4 ; a * 85 + b *171
-
- add r6, r6, #128
- mov r6, r6, lsr #8
- strb r6, [dest]
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ;|horizontal_line_2_3_scale_armv4|
-
-;/****************************************************************************
-; *
-; * ROUTINE : vertical_band_2_3_scale_armv4
-; *
-; * INPUTS : unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_pitch : Stride of destination data.
-; * unsigned int dest_width : Width of destination data.
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Scales vertical band of pixels by scale 2 to 3. The
-; * height of the band scaled is 2-pixels.
-; *
-; * SPECIAL NOTES : The routine uses the first line of the band below
-; * the current band.
-; *
-; ****************************************************************************/
-;void vertical_band_2_3_scale_armv4
-;(
-; r0 = UINT8 *dest
-; r1 = UINT32 dest_pitch
-; r2 = UINT32 dest_width
-;)
-|vertical_band_2_3_scale_armv4| PROC
- stmdb sp!, {r4 - r8, lr}
- ldr lr, =85
- ldr r12, =171
- add r3, r1, r1, lsl #1 ; 3 * dest_pitch
-
-vl23_loop
- ldrb r4, [src] ; a = des [0]
- ldrb r5, [src, r1] ; b = des [dest_pitch]
- ldrb r7, [src, r3] ; c = des [dest_pitch*3]
- subs r2, r2, #1
-
- mul r5, r12, r5 ; b * 171
- mla r6, lr, r4, r5 ; a * 85
- mla r8, lr, r7, r5 ; c * 85
-
- add r6, r6, #128
- mov r6, r6, lsr #8
- strb r6, [src, r1]
-
- add r8, r8, #128
- mov r8, r8, lsr #8
- strb r8, [src, r1, lsl #1]
-
- add src, src, #1
-
- bne vl23_loop
-
- ldmia sp!, {r4 - r8, pc}
- ENDP ;|vertical_band_2_3_scale_armv4|
-
-;/****************************************************************************
-; *
-; * ROUTINE : vp8cx_horizontal_line_3_5_scale_c
-; *
-; * INPUTS : const unsigned char *source : Pointer to source data.
-; * unsigned int source_width : Stride of source.
-; * unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_width : Stride of destination (NOT USED).
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Copies horizontal line of pixels from source to
-; * destination scaling up by 3 to 5.
-; *
-; * SPECIAL NOTES : None.
-; *
-; *
-; ****************************************************************************/
-;void vp8cx_horizontal_line_3_5_scale_c
-;(
-; const unsigned char *source,
-; unsigned int source_width,
-; unsigned char *dest,
-; unsigned int dest_width
-;)
-|horizontal_line_3_5_scale_armv4| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr c51_205, =0x3300cd
- ldr c102_154, =0x66009a
-
- ldrb r4, [src], #1 ; a = src[0]
-
-hl35_loop
-
- ldrb r8, [src], #1 ; b = src[1]
- strb r4, [dest], #1
-
- orr r6, r4, r8, lsl #16 ; b | a
- ldrb r9, [src], #1 ; c = src[2]
- mul r6, c102_154, r6 ; a * 102 + 154 * b
-
- orr r5, r9, r8, lsl #16 ; b | c
- mul r5, c51_205, r5 ; b * 205 + 51 * c
- add r6, r6, #0x8000
- ldrb r4, [src], #1 ; d = src[3]
- mov r6, r6, lsr #24
- strb r6, [dest], #1
-
- orr r7, r8, r9, lsl #16 ; c | b
- mul r7, c51_205, r7 ; c * 205 + 154 * b
- add r5, r5, #0x8000
- mov r5, r5, lsr #24
- strb r5, [dest], #1
-
- orr r9, r4, r9, lsl #16 ; c | d
- mul r9, c102_154, r9 ; c * 154 + 102 * d
- add r7, r7, #0x8000
- mov r7, r7, lsr #24
- strb r7, [dest], #1
-
- add r9, r9, #0x8000
- subs srcw, srcw, #3
- mov r9, r9, lsr #24
- strb r9, [dest], #1
-
- bpl hl35_loop
-
- ldrb r5, [src], #1 ; b = src[1]
- strb r4, [dest], #1
-
- orr r6, r4, r8, lsl #16 ; b | a
- ldrb r9, [src], #1 ; c = src[2]
- mul r6, c102_154, r6 ; a * 102 + 154 * b
-
- orr r5, r9, r8, lsl #16 ; b | c
- mul r5, c51_205, r5 ; b * 205 + 51 * c
- add r6, r6, #0x8000
- mov r6, r6, lsr #24
- strb r6, [dest], #1
-
- orr r7, r8, r9, lsl #16 ; c | b
- mul r7, c51_205, r7 ; c * 205 + 154 * b
- add r5, r5, #0x8000
- mov r5, r5, lsr #24
- strb r5, [dest], #1
-
- add r7, r7, #0x8000
- mov r7, r7, lsr #24
- strb r7, [dest], #1
- strb r9, [dest], #1
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ;|vp8cx_horizontal_line_3_5_scale_c|
-
-
-;/****************************************************************************
-; *
-; * ROUTINE : vp8cx_vertical_band_3_5_scale_c
-; *
-; * INPUTS : unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_pitch : Stride of destination data.
-; * unsigned int dest_width : Width of destination data.
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Scales vertical band of pixels by scale 3 to 5. The
-; * height of the band scaled is 3-pixels.
-; *
-; * SPECIAL NOTES : The routine uses the first line of the band below
-; * the current band.
-; *
-; ****************************************************************************/
-;void vertical_band_4_5_scale_armv4
-;(
-; r0 = UINT8 *dest
-; r1 = UINT32 dest_pitch
-; r2 = UINT32 dest_width
-;)
-|vertical_band_3_5_scale_armv4| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr c51_205, =0x3300cd
- ldr c102_154, =0x66009a
-
-vl35_loop
- mov r3, src
- ldrb r4, [r3], r1 ; a = des [0]
- ldrb r5, [r3], r1 ; b = des [dest_pitch]
- ldrb r7, [r3], r1 ; c = des[dest_pitch*2]
- add lr, src, r1
-
- orr r8, r4, r5, lsl #16 ; b | a
- mul r6, c102_154, r8 ; a * 102 + 154 * b
-
- ldrb r8, [r3, r1, lsl #1] ; d = des[dest_pitch*5]
- orr r3, r7, r5, lsl #16 ; b | c
- mul r9, c51_205, r3 ; b * 205 + 51 * c
- add r6, r6, #0x8000
- orr r3, r5, r7, lsl #16 ; c | b
- mov r6, r6, lsr #24
- strb r6, [lr], r1
-
- mul r5, c51_205, r3 ; c * 205 + 154 * b
- add r9, r9, #0x8000
- orr r3, r8, r7, lsl #16 ; c | d
- mov r9, r9, lsr #24
- strb r9, [lr], r1
-
- mul r7, c102_154, r3 ; c * 154 + 102 * d
- add r5, r5, #0x8000
- add src, src, #1
- mov r5, r5, lsr #24
- strb r5, [lr], r1
-
- add r7, r7, #0x8000
- subs r2, r2, #1
- mov r7, r7, lsr #24
- strb r7, [lr], r1
-
-
- bne vl35_loop
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ;|vertical_band_3_5_scale_armv4|
-
-;/****************************************************************************
-; *
-; * ROUTINE : horizontal_line_3_4_scale_armv4
-; *
-; * INPUTS : const unsigned char *source : Pointer to source data.
-; * unsigned int source_width : Stride of source.
-; * unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_width : Stride of destination (NOT USED).
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Copies horizontal line of pixels from source to
-; * destination scaling up by 3 to 4.
-; *
-; * SPECIAL NOTES : None.
-; *
-; *
-; ****************************************************************************/
-;void horizontal_line_3_4_scale_armv4
-;(
-; const unsigned char *source,
-; unsigned int source_width,
-; unsigned char *dest,
-; unsigned int dest_width
-;)
-|horizontal_line_3_4_scale_armv4| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r10, =64
- ldr r11, =192
- mov r9, #128
-
- ldrb r4, [src], #1 ; a = src[0]
-
-hl34_loop
-
- ldrb r8, [src], #1 ; b = src[1]
- ldrb r7, [src], #1 ; c = src[2]
- strb r4, [dest], #1
-
- mla r4, r10, r4, r9 ; a*64 + 128
- mla r4, r11, r8, r4 ; a*64 + b*192 + 1
-
- add r8, r8, #1 ; b + 1
- add r8, r8, r7 ; b + c + 1
- mov r8, r8, asr #1 ; (b + c + 1) >> 1
-
- mov r4, r4, asr #8 ; (a*64 + b*192 + 1) >> 8
- strb r4, [dest], #1
-
- strb r8, [dest], #1
-
- ldrb r4, [src], #1 ; [a+1]
-
- mla r7, r11, r7, r9 ; c*192 + 128
- mla r7, r4, r10, r7 ; a*64 + b*192 + 128
-
- subs srcw, srcw, #3
-
- mov r7, r7, asr #8 ; (a*64 + b*192 + 128) >> 8
- strb r7, [dest], #1
-
- bpl hl34_loop
-
- ldrb r8, [src], #1 ; b = src[1]
- ldrb r7, [src], #1 ; c = src[2]
- strb r4, [dest], #1
-
- mla r4, r10, r4, r9 ; a*64 + 128
- mla r4, r11, r8, r4 ; a*64 + b*192 + 1
- mov r4, r4, asr #8 ; (a*64 + b*192 + 1) >> 8
- strb r4, [dest], #1
-
- add r8, r8, #1 ; b + 1
- add r8, r8, r7 ; b + c + 1
- mov r8, r8, asr #1 ; (b + c + 1) >> 1
- strb r8, [dest], #1
- strb r7, [dest], #1
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ;|vp8cx_horizontal_line_3_4_scale_c|
-
-
-;/****************************************************************************
-; *
-; * ROUTINE : vertical_band_3_4_scale_armv4
-; *
-; * INPUTS : unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_pitch : Stride of destination data.
-; * unsigned int dest_width : Width of destination data.
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Scales vertical band of pixels by scale 3 to 4. The
-; * height of the band scaled is 3-pixels.
-; *
-; * SPECIAL NOTES : The routine uses the first line of the band below
-; * the current band.
-; *
-; ****************************************************************************/
-;void vertical_band_3_4_scale_armv4
-;(
-; r0 = UINT8 *dest
-; r1 = UINT32 dest_pitch
-; r2 = UINT32 dest_width
-;)
-|vertical_band_3_4_scale_armv4| PROC
- stmdb sp!, {r4 - r11, lr}
-
- ldr r10, =64
- ldr r11, =192
- mov r9, #128
-
-; ldr r1,[r1]
-vl34_loop
- mov r3, src
- ldrb r4, [r3], r1 ; a = des [0]
- ldrb r5, [r3], r1 ; b = des [dest_pitch]
- ldrb r7, [r3], r1 ; c = des [dest_pitch*2]
- add lr, src, r1
-
- mla r4, r10, r4, r9 ; a*64 + 128
- mla r4, r11, r5, r4 ; a*64 + b*192 + 1
-
- add r5, r5, #1 ; b + 1
- add r5, r5, r7 ; b + c + 1
- mov r5, r5, asr #1 ; (b + c + 1) >> 1
-
- mov r4, r4, asr #8 ; (a*64 + b*192 + 1) >> 8
- strb r4, [lr], r1
-
- ldrb r4, [r3, r1] ; a = des [dest_pitch*4]
-
- strb r5, [lr], r1
-
- mla r7, r11, r7, r9 ; c*192 + 128
- mla r7, r4, r10, r7 ; a*64 + b*192 + 128
- mov r7, r7, asr #8 ; (a*64 + b*192 + 128) >> 8
-
- add src, src, #1
- subs r2, r2, #1
-
- strb r7, [lr]
-
- bne vl34_loop
-
- ldmia sp!, {r4 - r11, pc}
- ENDP ;|vertical_band_3_4_scale_armv4|
-
-;/****************************************************************************
-; *
-; * ROUTINE : vp8cx_horizontal_line_1_2_scale_c
-; *
-; * INPUTS : const unsigned char *source : Pointer to source data.
-; * unsigned int source_width : Stride of source.
-; * unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_width : Stride of destination (NOT USED).
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Copies horizontal line of pixels from source to
-; * destination scaling up by 1 to 2.
-; *
-; * SPECIAL NOTES : None.
-; *
-; ****************************************************************************/
-;void vp8cx_horizontal_line_1_2_scale_c
-;(
-; const unsigned char *source,
-; unsigned int source_width,
-; unsigned char *dest,
-; unsigned int dest_width
-;)
-|horizontal_line_1_2_scale_armv4| PROC
- stmdb sp!, {r4 - r5, lr}
-
- sub srcw, srcw, #1
-
- ldrb r3, [src], #1
- ldrb r4, [src], #1
-hl12_loop
- subs srcw, srcw, #1
-
- add r5, r3, r4
- add r5, r5, #1
- mov r5, r5, lsr #1
-
- orr r5, r3, r5, lsl #8
- strh r5, [dest], #2
-
- mov r3, r4
-
- ldrneb r4, [src], #1
- bne hl12_loop
-
- orr r5, r4, r4, lsl #8
- strh r5, [dest]
-
- ldmia sp!, {r4 - r5, pc}
- ENDP ;|vertical_band_3_5_scale_armv4|
-
-;/****************************************************************************
-; *
-; * ROUTINE : vp8cx_vertical_band_1_2_scale_c
-; *
-; * INPUTS : unsigned char *dest : Pointer to destination data.
-; * unsigned int dest_pitch : Stride of destination data.
-; * unsigned int dest_width : Width of destination data.
-; *
-; * OUTPUTS : None.
-; *
-; * RETURNS : void
-; *
-; * FUNCTION : Scales vertical band of pixels by scale 1 to 2. The
-; * height of the band scaled is 1-pixel.
-; *
-; * SPECIAL NOTES : The routine uses the first line of the band below
-; * the current band.
-; *
-; ****************************************************************************/
-;void vp8cx_vertical_band_1_2_scale_c
-;(
-; r0 = UINT8 *dest
-; r1 = UINT32 dest_pitch
-; r2 = UINT32 dest_width
-;)
-|vertical_band_1_2_scale_armv4| PROC
- stmdb sp!, {r4 - r7, lr}
-
- ldr mask, =0xff00ff ; mask for selection
- ldr lr, = 0x010001
-
-vl12_loop
- mov r3, src
- ldr r4, [r3], r1
- ldr r5, [r3, r1]
-
- add src, src, #4
- subs r2, r2, #4
-
- and r6, r4, mask
- and r7, r5, mask
-
- add r6, r7, r6
- add r6, r6, lr
-
- and r4, mask, r4, lsr #8
- and r5, mask, r5, lsr #8
-
- mov r6, r6, lsr #1
- and r6, r6, mask
-
- add r4, r5, r4
- add r4, r4, lr
-
- mov r4, r4, lsr #1
- and r4, r4, mask
-
- orr r5, r6, r4, lsl #8
-
- str r5, [r3]
-
- bpl vl12_loop
-
- ldmia sp!, {r4 - r7, pc}
- ENDP ;|vertical_band_3_5_scale_armv4|
-
- END
diff --git a/vpx_scale/arm/nds/yv12extend.c b/vpx_scale/arm/nds/yv12extend.c
deleted file mode 100644
index 48c0dfb..0000000
--- a/vpx_scale/arm/nds/yv12extend.c
+++ /dev/null
@@ -1,221 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/****************************************************************************
-*
-* Module Title : yv12extend.c
-*
-* Description :
-*
-***************************************************************************/
-
-/****************************************************************************
-* Header Files
-****************************************************************************/
-#include "vpx_scale/yv12config.h"
-#include "vpx_mem/vpx_mem.h"
-#include <nitro.h>
-#include <nitro/mi.h>
-#include <nitro/itcm_begin.h>
-
-//---- DMA Number
-#define DMA_NO 3
-
-/****************************************************************************
-* Exports
-****************************************************************************/
-
-/****************************************************************************
-*
-****************************************************************************/
-void
-vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf)
-{
- int i;
- unsigned char *src_ptr1, *src_ptr2;
- unsigned char *dest_ptr1, *dest_ptr2;
-
- unsigned int Border;
- int plane_stride;
- int plane_height;
- int plane_width;
-
- /***********/
- /* Y Plane */
- /***********/
- Border = ybf->border;
- plane_stride = ybf->y_stride;
- plane_height = ybf->y_height;
- plane_width = ybf->y_width;
-
- // copy the left and right most columns out
- src_ptr1 = ybf->y_buffer;
- src_ptr2 = src_ptr1 + plane_width - 1;
- dest_ptr1 = src_ptr1 - Border;
- dest_ptr2 = src_ptr2 + 1;
-
- for (i = 0; i < plane_height; i++)
- {
- mi_cpu_fill8(dest_ptr1, src_ptr1[0], Border);
- mi_cpu_fill8(dest_ptr2, src_ptr2[0], Border);
- src_ptr1 += plane_stride;
- src_ptr2 += plane_stride;
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- // Now copy the top and bottom source lines into each line of the respective borders
- src_ptr1 = ybf->y_buffer - Border;
- src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
- dest_ptr1 = src_ptr1 - (Border * plane_stride);
- dest_ptr2 = src_ptr2 + plane_stride;
-
- for (i = 0; i < (int)Border; i++)
- {
- mi_cpu_copy_fast(src_ptr1, dest_ptr1, plane_stride);
- mi_cpu_copy_fast(src_ptr2, dest_ptr2, plane_stride);
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- plane_stride /= 2;
- plane_height /= 2;
- plane_width /= 2;
- Border /= 2;
-
- /***********/
- /* U Plane */
- /***********/
-
- // copy the left and right most columns out
- src_ptr1 = ybf->u_buffer;
- src_ptr2 = src_ptr1 + plane_width - 1;
- dest_ptr1 = src_ptr1 - Border;
- dest_ptr2 = src_ptr2 + 1;
-
- for (i = 0; i < plane_height; i++)
- {
- mi_cpu_fill8(dest_ptr1, src_ptr1[0], Border);
- mi_cpu_fill8(dest_ptr2, src_ptr2[0], Border);
- src_ptr1 += plane_stride;
- src_ptr2 += plane_stride;
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- // Now copy the top and bottom source lines into each line of the respective borders
- src_ptr1 = ybf->u_buffer - Border;
- src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
- dest_ptr1 = src_ptr1 - (Border * plane_stride);
- dest_ptr2 = src_ptr2 + plane_stride;
-
- for (i = 0; i < (int)(Border); i++)
- {
- mi_cpu_copy_fast(src_ptr1, dest_ptr1, plane_stride);
- mi_cpu_copy_fast(src_ptr2, dest_ptr2, plane_stride);
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- /***********/
- /* V Plane */
- /***********/
-
- // copy the left and right most columns out
- src_ptr1 = ybf->v_buffer;
- src_ptr2 = src_ptr1 + plane_width - 1;
- dest_ptr1 = src_ptr1 - Border;
- dest_ptr2 = src_ptr2 + 1;
-
- for (i = 0; i < plane_height; i++)
- {
- mi_cpu_fill8(dest_ptr1, src_ptr1[0], Border);
- mi_cpu_fill8(dest_ptr2, src_ptr2[0], Border);
- src_ptr1 += plane_stride;
- src_ptr2 += plane_stride;
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- // Now copy the top and bottom source lines into each line of the respective borders
- src_ptr1 = ybf->v_buffer - Border;
- src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
- dest_ptr1 = src_ptr1 - (Border * plane_stride);
- dest_ptr2 = src_ptr2 + plane_stride;
-
- for (i = 0; i < (int)(Border); i++)
- {
- mi_cpu_copy_fast(src_ptr1, dest_ptr1, plane_stride);
- mi_cpu_copy_fast(src_ptr2, dest_ptr2, plane_stride);
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-}
-
-
-
-/****************************************************************************
-*
-* ROUTINE : vp8_yv12_copy_frame
-*
-* INPUTS :
-*
-* OUTPUTS : None.
-*
-* RETURNS : void
-*
-* FUNCTION : Copies the source image into the destination image and
-* updates the destination's UMV borders.
-*
-* SPECIAL NOTES : The frames are assumed to be identical in size.
-*
-****************************************************************************/
-void
-vp8_yv12_copy_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc)
-{
- int yplane_size = (src_ybc->y_height + 2 * src_ybc->border) * (src_ybc->y_stride);
- int mem_size = (yplane_size * 3 / 2) + (src_ybc->y_stride * 2);
-
- mi_cpu_copy_fast(src_ybc->buffer_alloc, dst_ybc->buffer_alloc, mem_size);
-
- /* unsigned char *src_y, *dst_y;
- unsigned char *src_u, *dst_u;
- unsigned char *src_v, *dst_v;
-
- int yheight, uv_height;
- int ystride, uv_stride;
- int border;
- int yoffset, uvoffset;
-
- border = src_ybc->border;
- yheight = src_ybc->y_height;
- uv_height = src_ybc->uv_height;
-
- ystride = src_ybc->y_stride;
- uv_stride = src_ybc->uv_stride;
-
- yoffset = border * (ystride + 1);
- uvoffset = border/2 * (uv_stride + 1);
-
- src_y = src_ybc->y_buffer - yoffset;
- dst_y = dst_ybc->y_buffer - yoffset;
- src_u = src_ybc->u_buffer - uvoffset;
- dst_u = dst_ybc->u_buffer - uvoffset;
- src_v = src_ybc->v_buffer - uvoffset;
- dst_v = dst_ybc->v_buffer - uvoffset;
-
- mi_cpu_copy_fast (src_y, dst_y, ystride * (yheight + 2 * border));
- mi_cpu_copy_fast (src_u, dst_u, uv_stride * (uv_height + border));
- mi_cpu_copy_fast (src_v, dst_v, uv_stride * (uv_height + border));
- */
-}
-
-#include <nitro/itcm_end.h>
diff --git a/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm b/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm
deleted file mode 100644
index e6bb486..0000000
--- a/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm
+++ /dev/null
@@ -1,228 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_yv12_copy_frame_func_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- INCLUDE asm_com_offsets.asm
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-
-;void vp8_yv12_copy_frame_func_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-|vp8_yv12_copy_frame_func_neon| PROC
- push {r4 - r11, lr}
- vpush {d8 - d15}
-
- sub sp, sp, #16
-
- ;Copy Y plane
- ldr r8, [r0, #yv12_buffer_config_u_buffer] ;srcptr1
- ldr r9, [r1, #yv12_buffer_config_u_buffer] ;srcptr1
- ldr r10, [r0, #yv12_buffer_config_v_buffer] ;srcptr1
- ldr r11, [r1, #yv12_buffer_config_v_buffer] ;srcptr1
-
- ldr r4, [r0, #yv12_buffer_config_y_height]
- ldr r5, [r0, #yv12_buffer_config_y_width]
- ldr r6, [r0, #yv12_buffer_config_y_stride]
- ldr r7, [r1, #yv12_buffer_config_y_stride]
- ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
-
- str r8, [sp]
- str r9, [sp, #4]
- str r10, [sp, #8]
- str r11, [sp, #12]
-
- ; copy two rows at one time
- mov lr, r4, lsr #1
-
-cp_src_to_dst_height_loop
- mov r8, r2
- mov r9, r3
- add r10, r2, r6
- add r11, r3, r7
- mov r12, r5, lsr #7
-
-cp_src_to_dst_width_loop
- vld1.8 {q0, q1}, [r8]!
- vld1.8 {q8, q9}, [r10]!
- vld1.8 {q2, q3}, [r8]!
- vld1.8 {q10, q11}, [r10]!
- vld1.8 {q4, q5}, [r8]!
- vld1.8 {q12, q13}, [r10]!
- vld1.8 {q6, q7}, [r8]!
- vld1.8 {q14, q15}, [r10]!
-
- subs r12, r12, #1
-
- vst1.8 {q0, q1}, [r9]!
- vst1.8 {q8, q9}, [r11]!
- vst1.8 {q2, q3}, [r9]!
- vst1.8 {q10, q11}, [r11]!
- vst1.8 {q4, q5}, [r9]!
- vst1.8 {q12, q13}, [r11]!
- vst1.8 {q6, q7}, [r9]!
- vst1.8 {q14, q15}, [r11]!
-
- bne cp_src_to_dst_width_loop
-
- subs lr, lr, #1
- add r2, r2, r6, lsl #1
- add r3, r3, r7, lsl #1
-
- bne cp_src_to_dst_height_loop
-
- ands r10, r5, #0x7f ;check to see if extra copy is needed
- sub r11, r5, r10
- ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
- bne extra_cp_src_to_dst_width
-end_of_cp_src_to_dst
-
-;Copy U & V planes
- ldr r2, [sp] ;srcptr1
- ldr r3, [sp, #4] ;dstptr1
- mov r4, r4, lsr #1 ;src uv_height
- mov r5, r5, lsr #1 ;src uv_width
- mov r6, r6, lsr #1 ;src uv_stride
- mov r7, r7, lsr #1 ;dst uv_stride
-
- mov r1, #2
-
-cp_uv_loop
-
- ;copy two rows at one time
- mov lr, r4, lsr #1
-
-cp_src_to_dst_height_uv_loop
- mov r8, r2
- mov r9, r3
- add r10, r2, r6
- add r11, r3, r7
- mov r12, r5, lsr #6
-
-cp_src_to_dst_width_uv_loop
- vld1.8 {q0, q1}, [r8]!
- vld1.8 {q8, q9}, [r10]!
- vld1.8 {q2, q3}, [r8]!
- vld1.8 {q10, q11}, [r10]!
-
- subs r12, r12, #1
-
- vst1.8 {q0, q1}, [r9]!
- vst1.8 {q8, q9}, [r11]!
- vst1.8 {q2, q3}, [r9]!
- vst1.8 {q10, q11}, [r11]!
-
- bne cp_src_to_dst_width_uv_loop
-
- subs lr, lr, #1
- add r2, r2, r6, lsl #1
- add r3, r3, r7, lsl #1
-
- bne cp_src_to_dst_height_uv_loop
-
- ands r10, r5, #0x3f ;check to see if extra copy is needed
- sub r11, r5, r10
- ldr r2, [sp] ;srcptr1
- ldr r3, [sp, #4] ;dstptr1
- bne extra_cp_src_to_dst_uv_width
-end_of_cp_src_to_dst_uv
-
- subs r1, r1, #1
-
- addne sp, sp, #8
-
- ldrne r2, [sp] ;srcptr1
- ldrne r3, [sp, #4] ;dstptr1
-
- bne cp_uv_loop
-
- add sp, sp, #8
-
- vpop {d8 - d15}
- pop {r4 - r11, pc}
-
-;=============================
-extra_cp_src_to_dst_width
- add r2, r2, r11
- add r3, r3, r11
- add r0, r8, r6
- add r11, r9, r7
-
- mov lr, r4, lsr #1
-extra_cp_src_to_dst_height_loop
- mov r8, r2
- mov r9, r3
- add r0, r8, r6
- add r11, r9, r7
-
- mov r12, r10
-
-extra_cp_src_to_dst_width_loop
- vld1.8 {q0}, [r8]!
- vld1.8 {q1}, [r0]!
-
- subs r12, r12, #16
-
- vst1.8 {q0}, [r9]!
- vst1.8 {q1}, [r11]!
- bne extra_cp_src_to_dst_width_loop
-
- subs lr, lr, #1
-
- add r2, r2, r6, lsl #1
- add r3, r3, r7, lsl #1
-
- bne extra_cp_src_to_dst_height_loop
-
- b end_of_cp_src_to_dst
-
-;=================================
-extra_cp_src_to_dst_uv_width
- add r2, r2, r11
- add r3, r3, r11
- add r0, r8, r6
- add r11, r9, r7
-
- mov lr, r4, lsr #1
-extra_cp_src_to_dst_height_uv_loop
- mov r8, r2
- mov r9, r3
- add r0, r8, r6
- add r11, r9, r7
-
- mov r12, r10
-
-extra_cp_src_to_dst_width_uv_loop
- vld1.8 {d0}, [r8]!
- vld1.8 {d1}, [r0]!
-
- subs r12, r12, #8
-
- vst1.8 {d0}, [r9]!
- vst1.8 {d1}, [r11]!
- bne extra_cp_src_to_dst_width_uv_loop
-
- subs lr, lr, #1
-
- add r2, r2, r6, lsl #1
- add r3, r3, r7, lsl #1
-
- bne extra_cp_src_to_dst_height_uv_loop
-
- b end_of_cp_src_to_dst_uv
-
- ENDP
- END
diff --git a/vpx_scale/arm/neon/vp8_vpxyv12_copyframeyonly_neon.asm b/vpx_scale/arm/neon/vp8_vpxyv12_copyframeyonly_neon.asm
deleted file mode 100644
index febccc2..0000000
--- a/vpx_scale/arm/neon/vp8_vpxyv12_copyframeyonly_neon.asm
+++ /dev/null
@@ -1,500 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_yv12_copy_frame_yonly_neon|
- EXPORT |vp8_yv12_copy_frame_yonly_no_extend_frame_borders_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- INCLUDE asm_com_offsets.asm
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void vpxyv12_copy_frame_yonly(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-; Note: this is VP8 function, which has border=32 and 16. Internal y_width and y_height
-; are always multiples of 16.
-
-|vp8_yv12_copy_frame_yonly_neon| PROC
- push {r4 - r11, lr}
- vpush {d8 - d15}
-
- ldr r4, [r0, #yv12_buffer_config_y_height]
- ldr r5, [r0, #yv12_buffer_config_y_width]
- ldr r6, [r0, #yv12_buffer_config_y_stride]
- ldr r7, [r1, #yv12_buffer_config_y_stride]
- ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
-
- ; copy two rows at one time
- mov lr, r4, lsr #1
-
-cp_src_to_dst_height_loop
- mov r8, r2
- mov r9, r3
- add r10, r2, r6
- add r11, r3, r7
- mov r12, r5, lsr #7
-
-cp_src_to_dst_width_loop
- vld1.8 {q0, q1}, [r8]!
- vld1.8 {q8, q9}, [r10]!
- vld1.8 {q2, q3}, [r8]!
- vld1.8 {q10, q11}, [r10]!
- vld1.8 {q4, q5}, [r8]!
- vld1.8 {q12, q13}, [r10]!
- vld1.8 {q6, q7}, [r8]!
- vld1.8 {q14, q15}, [r10]!
-
- subs r12, r12, #1
-
- vst1.8 {q0, q1}, [r9]!
- vst1.8 {q8, q9}, [r11]!
- vst1.8 {q2, q3}, [r9]!
- vst1.8 {q10, q11}, [r11]!
- vst1.8 {q4, q5}, [r9]!
- vst1.8 {q12, q13}, [r11]!
- vst1.8 {q6, q7}, [r9]!
- vst1.8 {q14, q15}, [r11]!
-
- bne cp_src_to_dst_width_loop
-
- subs lr, lr, #1
- add r2, r2, r6, lsl #1
- add r3, r3, r7, lsl #1
-
- bne cp_src_to_dst_height_loop
-
- ands r10, r5, #0x7f ;check to see if extra copy is needed
- sub r11, r5, r10
- ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
- bne extra_cp_src_to_dst_width
-end_of_cp_src_to_dst
-
-
- ;vpxyv12_extend_frame_borders_yonly
- mov r0, r1
- ;Not need to load y_width, since: y_width = y_stride - 2*border
- ldr r3, [r0, #yv12_buffer_config_border]
- ldr r1, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- ldr r4, [r0, #yv12_buffer_config_y_height]
- ldr lr, [r0, #yv12_buffer_config_y_stride]
-
- cmp r3, #16
- beq b16_extend_frame_borders
-
-;=======================
-b32_extend_frame_borders
-;border = 32
-;=======================
-;Border copy for Y plane
-;copy the left and right most columns out
- sub r5, r1, r3 ;destptr1
- add r6, r1, lr
- sub r6, r6, r3, lsl #1 ;destptr2
- sub r2, r6, #1 ;srcptr2
-
- ;Do four rows at one time
- mov r12, r4, lsr #2
-
-copy_left_right_y
- vld1.8 {d0[], d1[]}, [r1], lr
- vld1.8 {d4[], d5[]}, [r2], lr
- vld1.8 {d8[], d9[]}, [r1], lr
- vld1.8 {d12[], d13[]}, [r2], lr
- vld1.8 {d16[], d17[]}, [r1], lr
- vld1.8 {d20[], d21[]}, [r2], lr
- vld1.8 {d24[], d25[]}, [r1], lr
- vld1.8 {d28[], d29[]}, [r2], lr
-
- vmov q1, q0
- vmov q3, q2
- vmov q5, q4
- vmov q7, q6
- vmov q9, q8
- vmov q11, q10
- vmov q13, q12
- vmov q15, q14
-
- subs r12, r12, #1
-
- vst1.8 {q0, q1}, [r5], lr
- vst1.8 {q2, q3}, [r6], lr
- vst1.8 {q4, q5}, [r5], lr
- vst1.8 {q6, q7}, [r6], lr
- vst1.8 {q8, q9}, [r5], lr
- vst1.8 {q10, q11}, [r6], lr
- vst1.8 {q12, q13}, [r5], lr
- vst1.8 {q14, q15}, [r6], lr
-
- bne copy_left_right_y
-
-;Now copy the top and bottom source lines into each line of the respective borders
- ldr r7, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- mul r8, r3, lr
-
- mov r12, lr, lsr #7
-
- sub r6, r1, r3 ;destptr2
- sub r2, r6, lr ;srcptr2
- sub r1, r7, r3 ;srcptr1
- sub r5, r1, r8 ;destptr1
-
-copy_top_bottom_y
- vld1.8 {q0, q1}, [r1]!
- vld1.8 {q8, q9}, [r2]!
- vld1.8 {q2, q3}, [r1]!
- vld1.8 {q10, q11}, [r2]!
- vld1.8 {q4, q5}, [r1]!
- vld1.8 {q12, q13}, [r2]!
- vld1.8 {q6, q7}, [r1]!
- vld1.8 {q14, q15}, [r2]!
-
- mov r7, r3
-
-top_bottom_32
- subs r7, r7, #1
-
- vst1.8 {q0, q1}, [r5]!
- vst1.8 {q8, q9}, [r6]!
- vst1.8 {q2, q3}, [r5]!
- vst1.8 {q10, q11}, [r6]!
- vst1.8 {q4, q5}, [r5]!
- vst1.8 {q12, q13}, [r6]!
- vst1.8 {q6, q7}, [r5]!
- vst1.8 {q14, q15}, [r6]!
-
- add r5, r5, lr
- sub r5, r5, #128
- add r6, r6, lr
- sub r6, r6, #128
-
- bne top_bottom_32
-
- sub r5, r1, r8
- add r6, r2, lr
-
- subs r12, r12, #1
- bne copy_top_bottom_y
-
- mov r7, lr, lsr #4 ;check to see if extra copy is needed
- ands r7, r7, #0x7
- bne extra_top_bottom_y
-end_of_border_copy_y
-
- vpop {d8 - d15}
- pop {r4 - r11, pc}
-
-;=====================
-;extra copy part for Y
-extra_top_bottom_y
- vld1.8 {q0}, [r1]!
- vld1.8 {q2}, [r2]!
-
- mov r9, r3, lsr #3
-
-extra_top_bottom_32
- subs r9, r9, #1
-
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- bne extra_top_bottom_32
-
- sub r5, r1, r8
- add r6, r2, lr
- subs r7, r7, #1
- bne extra_top_bottom_y
-
- b end_of_border_copy_y
-
-
-;=======================
-b16_extend_frame_borders
-;border = 16
-;=======================
-;Border copy for Y plane
-;copy the left and right most columns out
- sub r5, r1, r3 ;destptr1
- add r6, r1, lr
- sub r6, r6, r3, lsl #1 ;destptr2
- sub r2, r6, #1 ;srcptr2
-
- ;Do four rows at one time
- mov r12, r4, lsr #2
-
-copy_left_right_y_b16
- vld1.8 {d0[], d1[]}, [r1], lr
- vld1.8 {d4[], d5[]}, [r2], lr
- vld1.8 {d8[], d9[]}, [r1], lr
- vld1.8 {d12[], d13[]}, [r2], lr
- vld1.8 {d16[], d17[]}, [r1], lr
- vld1.8 {d20[], d21[]}, [r2], lr
- vld1.8 {d24[], d25[]}, [r1], lr
- vld1.8 {d28[], d29[]}, [r2], lr
-
- subs r12, r12, #1
-
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q4}, [r5], lr
- vst1.8 {q6}, [r6], lr
- vst1.8 {q8}, [r5], lr
- vst1.8 {q10}, [r6], lr
- vst1.8 {q12}, [r5], lr
- vst1.8 {q14}, [r6], lr
-
- bne copy_left_right_y_b16
-
-;Now copy the top and bottom source lines into each line of the respective borders
- ldr r7, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- mul r8, r3, lr
-
- mov r12, lr, lsr #7
-
- sub r6, r1, r3 ;destptr2
- sub r2, r6, lr ;srcptr2
- sub r1, r7, r3 ;srcptr1
- sub r5, r1, r8 ;destptr1
-
-copy_top_bottom_y_b16
- vld1.8 {q0, q1}, [r1]!
- vld1.8 {q8, q9}, [r2]!
- vld1.8 {q2, q3}, [r1]!
- vld1.8 {q10, q11}, [r2]!
- vld1.8 {q4, q5}, [r1]!
- vld1.8 {q12, q13}, [r2]!
- vld1.8 {q6, q7}, [r1]!
- vld1.8 {q14, q15}, [r2]!
-
- mov r7, r3
-
-top_bottom_16_b16
- subs r7, r7, #1
-
- vst1.8 {q0, q1}, [r5]!
- vst1.8 {q8, q9}, [r6]!
- vst1.8 {q2, q3}, [r5]!
- vst1.8 {q10, q11}, [r6]!
- vst1.8 {q4, q5}, [r5]!
- vst1.8 {q12, q13}, [r6]!
- vst1.8 {q6, q7}, [r5]!
- vst1.8 {q14, q15}, [r6]!
-
- add r5, r5, lr
- sub r5, r5, #128
- add r6, r6, lr
- sub r6, r6, #128
-
- bne top_bottom_16_b16
-
- sub r5, r1, r8
- add r6, r2, lr
-
- subs r12, r12, #1
- bne copy_top_bottom_y_b16
-
- mov r7, lr, lsr #4 ;check to see if extra copy is needed
- ands r7, r7, #0x7
- bne extra_top_bottom_y_b16
-end_of_border_copy_y_b16
-
- vpop {d8 - d15}
- pop {r4 - r11, pc}
-
-;=====================
-;extra copy part for Y
-extra_top_bottom_y_b16
- vld1.8 {q0}, [r1]!
- vld1.8 {q2}, [r2]!
-
- mov r9, r3, lsr #3
-
-extra_top_bottom_16_b16
- subs r9, r9, #1
-
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- bne extra_top_bottom_16_b16
-
- sub r5, r1, r8
- add r6, r2, lr
- subs r7, r7, #1
- bne extra_top_bottom_y_b16
-
- b end_of_border_copy_y_b16
-
-;=============================
-extra_cp_src_to_dst_width
- add r2, r2, r11
- add r3, r3, r11
- add r0, r8, r6
- add r11, r9, r7
-
- mov lr, r4, lsr #1
-extra_cp_src_to_dst_height_loop
- mov r8, r2
- mov r9, r3
- add r0, r8, r6
- add r11, r9, r7
-
- mov r12, r10
-
-extra_cp_src_to_dst_width_loop
- vld1.8 {q0}, [r8]!
- vld1.8 {q1}, [r0]!
-
- subs r12, r12, #16
-
- vst1.8 {q0}, [r9]!
- vst1.8 {q1}, [r11]!
- bne extra_cp_src_to_dst_width_loop
-
- subs lr, lr, #1
-
- add r2, r2, r6, lsl #1
- add r3, r3, r7, lsl #1
-
- bne extra_cp_src_to_dst_height_loop
-
- b end_of_cp_src_to_dst
-
- ENDP
-
-;===========================================================
-;In vp8cx_pick_filter_level(), call vp8_yv12_copy_frame_yonly
-;without extend_frame_borders.
-|vp8_yv12_copy_frame_yonly_no_extend_frame_borders_neon| PROC
- push {r4 - r11, lr}
- vpush {d8-d15}
-
- ldr r4, [r0, #yv12_buffer_config_y_height]
- ldr r5, [r0, #yv12_buffer_config_y_width]
- ldr r6, [r0, #yv12_buffer_config_y_stride]
- ldr r7, [r1, #yv12_buffer_config_y_stride]
- ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
-
- ; copy two rows at one time
- mov lr, r4, lsr #1
-
-cp_src_to_dst_height_loop1
- mov r8, r2
- mov r9, r3
- add r10, r2, r6
- add r11, r3, r7
- mov r12, r5, lsr #7
-
-cp_src_to_dst_width_loop1
- vld1.8 {q0, q1}, [r8]!
- vld1.8 {q8, q9}, [r10]!
- vld1.8 {q2, q3}, [r8]!
- vld1.8 {q10, q11}, [r10]!
- vld1.8 {q4, q5}, [r8]!
- vld1.8 {q12, q13}, [r10]!
- vld1.8 {q6, q7}, [r8]!
- vld1.8 {q14, q15}, [r10]!
-
- subs r12, r12, #1
-
- vst1.8 {q0, q1}, [r9]!
- vst1.8 {q8, q9}, [r11]!
- vst1.8 {q2, q3}, [r9]!
- vst1.8 {q10, q11}, [r11]!
- vst1.8 {q4, q5}, [r9]!
- vst1.8 {q12, q13}, [r11]!
- vst1.8 {q6, q7}, [r9]!
- vst1.8 {q14, q15}, [r11]!
-
- bne cp_src_to_dst_width_loop1
-
- subs lr, lr, #1
- add r2, r2, r6, lsl #1
- add r3, r3, r7, lsl #1
-
- bne cp_src_to_dst_height_loop1
-
- ands r10, r5, #0x7f ;check to see if extra copy is needed
- sub r11, r5, r10
- ldr r2, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- ldr r3, [r1, #yv12_buffer_config_y_buffer] ;dstptr1
- bne extra_cp_src_to_dst_width1
-end_of_cp_src_to_dst1
-
- vpop {d8 - d15}
- pop {r4-r11, pc}
-
-;=============================
-extra_cp_src_to_dst_width1
- add r2, r2, r11
- add r3, r3, r11
- add r0, r8, r6
- add r11, r9, r7
-
- mov lr, r4, lsr #1
-extra_cp_src_to_dst_height_loop1
- mov r8, r2
- mov r9, r3
- add r0, r8, r6
- add r11, r9, r7
-
- mov r12, r10
-
-extra_cp_src_to_dst_width_loop1
- vld1.8 {q0}, [r8]!
- vld1.8 {q1}, [r0]!
-
- subs r12, r12, #16
-
- vst1.8 {q0}, [r9]!
- vst1.8 {q1}, [r11]!
- bne extra_cp_src_to_dst_width_loop1
-
- subs lr, lr, #1
-
- add r2, r2, r6, lsl #1
- add r3, r3, r7, lsl #1
-
- bne extra_cp_src_to_dst_height_loop1
-
- b end_of_cp_src_to_dst1
-
- ENDP
-
- END
diff --git a/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm b/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm
deleted file mode 100644
index b0a3b93..0000000
--- a/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm
+++ /dev/null
@@ -1,588 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_yv12_extend_frame_borders_neon|
- ARM
- REQUIRE8
- PRESERVE8
-
- INCLUDE asm_com_offsets.asm
-
- AREA ||.text||, CODE, READONLY, ALIGN=2
-;void vp8_yv12_extend_frame_borders_neon (YV12_BUFFER_CONFIG *ybf);
-;Note: this is VP8 function, which has border=32 and 16. Internal y_width and y_height
-; are always multiples of 16.
-
-|vp8_yv12_extend_frame_borders_neon| PROC
- push {r4 - r10, lr}
- vpush {d8 - d15}
-
- ;Not need to load y_width, since: y_width = y_stride - 2*border
- ldr r3, [r0, #yv12_buffer_config_border]
- ldr r1, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- ldr r4, [r0, #yv12_buffer_config_y_height]
- ldr lr, [r0, #yv12_buffer_config_y_stride]
-
- cmp r3, #16
- beq b16_extend_frame_borders
-
-;=======================
-b32_extend_frame_borders
-;border = 32
-;=======================
-;Border copy for Y plane
-;copy the left and right most columns out
- sub r5, r1, r3 ;destptr1
- add r6, r1, lr
- sub r6, r6, r3, lsl #1 ;destptr2
- sub r2, r6, #1 ;srcptr2
-
- ;Do four rows at one time
- mov r12, r4, lsr #2
-
-copy_left_right_y
- vld1.8 {d0[], d1[]}, [r1], lr
- vld1.8 {d4[], d5[]}, [r2], lr
- vld1.8 {d8[], d9[]}, [r1], lr
- vld1.8 {d12[], d13[]}, [r2], lr
- vld1.8 {d16[], d17[]}, [r1], lr
- vld1.8 {d20[], d21[]}, [r2], lr
- vld1.8 {d24[], d25[]}, [r1], lr
- vld1.8 {d28[], d29[]}, [r2], lr
-
- vmov q1, q0
- vmov q3, q2
- vmov q5, q4
- vmov q7, q6
- vmov q9, q8
- vmov q11, q10
- vmov q13, q12
- vmov q15, q14
-
- subs r12, r12, #1
-
- vst1.8 {q0, q1}, [r5], lr
- vst1.8 {q2, q3}, [r6], lr
- vst1.8 {q4, q5}, [r5], lr
- vst1.8 {q6, q7}, [r6], lr
- vst1.8 {q8, q9}, [r5], lr
- vst1.8 {q10, q11}, [r6], lr
- vst1.8 {q12, q13}, [r5], lr
- vst1.8 {q14, q15}, [r6], lr
-
- bne copy_left_right_y
-
-;Now copy the top and bottom source lines into each line of the respective borders
- ldr r7, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- mul r8, r3, lr
-
- mov r12, lr, lsr #7
-
- sub r6, r1, r3 ;destptr2
- sub r2, r6, lr ;srcptr2
- sub r1, r7, r3 ;srcptr1
- sub r5, r1, r8 ;destptr1
-
-copy_top_bottom_y
- vld1.8 {q0, q1}, [r1]!
- vld1.8 {q8, q9}, [r2]!
- vld1.8 {q2, q3}, [r1]!
- vld1.8 {q10, q11}, [r2]!
- vld1.8 {q4, q5}, [r1]!
- vld1.8 {q12, q13}, [r2]!
- vld1.8 {q6, q7}, [r1]!
- vld1.8 {q14, q15}, [r2]!
-
- mov r7, r3
-
-top_bottom_32
- subs r7, r7, #1
-
- vst1.8 {q0, q1}, [r5]!
- vst1.8 {q8, q9}, [r6]!
- vst1.8 {q2, q3}, [r5]!
- vst1.8 {q10, q11}, [r6]!
- vst1.8 {q4, q5}, [r5]!
- vst1.8 {q12, q13}, [r6]!
- vst1.8 {q6, q7}, [r5]!
- vst1.8 {q14, q15}, [r6]!
-
- add r5, r5, lr
- sub r5, r5, #128
- add r6, r6, lr
- sub r6, r6, #128
-
- bne top_bottom_32
-
- sub r5, r1, r8
- add r6, r2, lr
-
- subs r12, r12, #1
- bne copy_top_bottom_y
-
- mov r7, lr, lsr #4 ;check to see if extra copy is needed
- ands r7, r7, #0x7
- bne extra_top_bottom_y
-end_of_border_copy_y
-
-;Border copy for U, V planes
- ldr r1, [r0, #yv12_buffer_config_u_buffer] ;srcptr1
- mov lr, lr, lsr #1 ;uv_stride
- mov r3, r3, lsr #1 ;border
- mov r4, r4, lsr #1 ;uv_height
- mov r8, r8, lsr #2
-
- mov r10, #2
-
-;copy the left and right most columns out
-border_copy_uv
- sub r5, r1, r3 ;destptr1
- add r6, r1, lr
- sub r6, r6, r3, lsl #1 ;destptr2
- sub r2, r6, #1 ;srcptr2
-
- mov r7, r1
-
- ;Do eight rows at one time
- mov r12, r4, lsr #3
-
-copy_left_right_uv
- vld1.8 {d0[], d1[]}, [r1], lr
- vld1.8 {d2[], d3[]}, [r2], lr
- vld1.8 {d4[], d5[]}, [r1], lr
- vld1.8 {d6[], d7[]}, [r2], lr
- vld1.8 {d8[], d9[]}, [r1], lr
- vld1.8 {d10[], d11[]}, [r2], lr
- vld1.8 {d12[], d13[]}, [r1], lr
- vld1.8 {d14[], d15[]}, [r2], lr
- vld1.8 {d16[], d17[]}, [r1], lr
- vld1.8 {d18[], d19[]}, [r2], lr
- vld1.8 {d20[], d21[]}, [r1], lr
- vld1.8 {d22[], d23[]}, [r2], lr
- vld1.8 {d24[], d25[]}, [r1], lr
- vld1.8 {d26[], d27[]}, [r2], lr
- vld1.8 {d28[], d29[]}, [r1], lr
- vld1.8 {d30[], d31[]}, [r2], lr
-
- subs r12, r12, #1
-
- vst1.8 {q0}, [r5], lr
- vst1.8 {q1}, [r6], lr
- vst1.8 {q2}, [r5], lr
- vst1.8 {q3}, [r6], lr
- vst1.8 {q4}, [r5], lr
- vst1.8 {q5}, [r6], lr
- vst1.8 {q6}, [r5], lr
- vst1.8 {q7}, [r6], lr
- vst1.8 {q8}, [r5], lr
- vst1.8 {q9}, [r6], lr
- vst1.8 {q10}, [r5], lr
- vst1.8 {q11}, [r6], lr
- vst1.8 {q12}, [r5], lr
- vst1.8 {q13}, [r6], lr
- vst1.8 {q14}, [r5], lr
- vst1.8 {q15}, [r6], lr
-
- bne copy_left_right_uv
-
-;Now copy the top and bottom source lines into each line of the respective borders
- mov r12, lr, lsr #6
-
- sub r6, r1, r3 ;destptr2
- sub r2, r6, lr ;srcptr2
- sub r1, r7, r3 ;srcptr1
- sub r5, r1, r8 ;destptr1
-
-copy_top_bottom_uv
- vld1.8 {q0, q1}, [r1]!
- vld1.8 {q8, q9}, [r2]!
- vld1.8 {q2, q3}, [r1]!
- vld1.8 {q10, q11}, [r2]!
-
- mov r7, r3
-
-top_bottom_16
- subs r7, r7, #1
-
- vst1.8 {q0, q1}, [r5]!
- vst1.8 {q8, q9}, [r6]!
- vst1.8 {q2, q3}, [r5]!
- vst1.8 {q10, q11}, [r6]!
-
- add r5, r5, lr
- sub r5, r5, #64
- add r6, r6, lr
- sub r6, r6, #64
-
- bne top_bottom_16
-
- sub r5, r1, r8
- add r6, r2, lr
-
- subs r12, r12, #1
- bne copy_top_bottom_uv
-
- mov r7, lr, lsr #3 ;check to see if extra copy is needed
- ands r7, r7, #0x7
- bne extra_top_bottom_uv
-
-end_of_border_copy_uv
- subs r10, r10, #1
- ldrne r1, [r0, #yv12_buffer_config_v_buffer] ;srcptr1
- bne border_copy_uv
-
- vpop {d8 - d15}
- pop {r4 - r10, pc}
-
-;;;;;;;;;;;;;;;;;;;;;;
-;extra copy part for Y
-extra_top_bottom_y
- vld1.8 {q0}, [r1]!
- vld1.8 {q2}, [r2]!
-
- mov r9, r3, lsr #3
-
-extra_top_bottom_32
- subs r9, r9, #1
-
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- bne extra_top_bottom_32
-
- sub r5, r1, r8
- add r6, r2, lr
- subs r7, r7, #1
- bne extra_top_bottom_y
-
- b end_of_border_copy_y
-
-;extra copy part for UV
-extra_top_bottom_uv
- vld1.8 {d0}, [r1]!
- vld1.8 {d8}, [r2]!
-
- mov r9, r3, lsr #3
-
-extra_top_bottom_16
- subs r9, r9, #1
-
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- bne extra_top_bottom_16
-
- sub r5, r1, r8
- add r6, r2, lr
- subs r7, r7, #1
- bne extra_top_bottom_uv
-
- b end_of_border_copy_uv
-
-
-;=======================
-b16_extend_frame_borders
-;border = 16
-;=======================
-;Border copy for Y plane
-;copy the left and right most columns out
- sub r5, r1, r3 ;destptr1
- add r6, r1, lr
- sub r6, r6, r3, lsl #1 ;destptr2
- sub r2, r6, #1 ;srcptr2
-
- ;Do four rows at one time
- mov r12, r4, lsr #2
-
-copy_left_right_y_b16
- vld1.8 {d0[], d1[]}, [r1], lr
- vld1.8 {d4[], d5[]}, [r2], lr
- vld1.8 {d8[], d9[]}, [r1], lr
- vld1.8 {d12[], d13[]}, [r2], lr
- vld1.8 {d16[], d17[]}, [r1], lr
- vld1.8 {d20[], d21[]}, [r2], lr
- vld1.8 {d24[], d25[]}, [r1], lr
- vld1.8 {d28[], d29[]}, [r2], lr
-
- subs r12, r12, #1
-
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q4}, [r5], lr
- vst1.8 {q6}, [r6], lr
- vst1.8 {q8}, [r5], lr
- vst1.8 {q10}, [r6], lr
- vst1.8 {q12}, [r5], lr
- vst1.8 {q14}, [r6], lr
-
- bne copy_left_right_y_b16
-
-;Now copy the top and bottom source lines into each line of the respective borders
- ldr r7, [r0, #yv12_buffer_config_y_buffer] ;srcptr1
- mul r8, r3, lr
-
- mov r12, lr, lsr #7
-
- sub r6, r1, r3 ;destptr2
- sub r2, r6, lr ;srcptr2
- sub r1, r7, r3 ;srcptr1
- sub r5, r1, r8 ;destptr1
-
-copy_top_bottom_y_b16
- vld1.8 {q0, q1}, [r1]!
- vld1.8 {q8, q9}, [r2]!
- vld1.8 {q2, q3}, [r1]!
- vld1.8 {q10, q11}, [r2]!
- vld1.8 {q4, q5}, [r1]!
- vld1.8 {q12, q13}, [r2]!
- vld1.8 {q6, q7}, [r1]!
- vld1.8 {q14, q15}, [r2]!
-
- mov r7, r3
-
-top_bottom_16_b16
- subs r7, r7, #1
-
- vst1.8 {q0, q1}, [r5]!
- vst1.8 {q8, q9}, [r6]!
- vst1.8 {q2, q3}, [r5]!
- vst1.8 {q10, q11}, [r6]!
- vst1.8 {q4, q5}, [r5]!
- vst1.8 {q12, q13}, [r6]!
- vst1.8 {q6, q7}, [r5]!
- vst1.8 {q14, q15}, [r6]!
-
- add r5, r5, lr
- sub r5, r5, #128
- add r6, r6, lr
- sub r6, r6, #128
-
- bne top_bottom_16_b16
-
- sub r5, r1, r8
- add r6, r2, lr
-
- subs r12, r12, #1
- bne copy_top_bottom_y_b16
-
- mov r7, lr, lsr #4 ;check to see if extra copy is needed
- ands r7, r7, #0x7
- bne extra_top_bottom_y_b16
-end_of_border_copy_y_b16
-
-;Border copy for U, V planes
- ldr r1, [r0, #yv12_buffer_config_u_buffer] ;srcptr1
- mov lr, lr, lsr #1 ;uv_stride
- mov r3, r3, lsr #1 ;border
- mov r4, r4, lsr #1 ;uv_height
- mov r8, r8, lsr #2
-
- mov r10, #2
-
-;copy the left and right most columns out
-border_copy_uv_b16
- sub r5, r1, r3 ;destptr1
- add r6, r1, lr
- sub r6, r6, r3, lsl #1 ;destptr2
- sub r2, r6, #1 ;srcptr2
-
- mov r7, r1
-
- ;Do eight rows at one time
- mov r12, r4, lsr #3
-
-copy_left_right_uv_b16
- vld1.8 {d0[]}, [r1], lr
- vld1.8 {d2[]}, [r2], lr
- vld1.8 {d4[]}, [r1], lr
- vld1.8 {d6[]}, [r2], lr
- vld1.8 {d8[]}, [r1], lr
- vld1.8 {d10[]}, [r2], lr
- vld1.8 {d12[]}, [r1], lr
- vld1.8 {d14[]}, [r2], lr
- vld1.8 {d16[]}, [r1], lr
- vld1.8 {d18[]}, [r2], lr
- vld1.8 {d20[]}, [r1], lr
- vld1.8 {d22[]}, [r2], lr
- vld1.8 {d24[]}, [r1], lr
- vld1.8 {d26[]}, [r2], lr
- vld1.8 {d28[]}, [r1], lr
- vld1.8 {d30[]}, [r2], lr
-
- subs r12, r12, #1
-
- vst1.8 {d0}, [r5], lr
- vst1.8 {d2}, [r6], lr
- vst1.8 {d4}, [r5], lr
- vst1.8 {d6}, [r6], lr
- vst1.8 {d8}, [r5], lr
- vst1.8 {d10}, [r6], lr
- vst1.8 {d12}, [r5], lr
- vst1.8 {d14}, [r6], lr
- vst1.8 {d16}, [r5], lr
- vst1.8 {d18}, [r6], lr
- vst1.8 {d20}, [r5], lr
- vst1.8 {d22}, [r6], lr
- vst1.8 {d24}, [r5], lr
- vst1.8 {d26}, [r6], lr
- vst1.8 {d28}, [r5], lr
- vst1.8 {d30}, [r6], lr
-
- bne copy_left_right_uv_b16
-
-;Now copy the top and bottom source lines into each line of the respective borders
- mov r12, lr, lsr #6
-
- sub r6, r1, r3 ;destptr2
- sub r2, r6, lr ;srcptr2
- sub r1, r7, r3 ;srcptr1
- sub r5, r1, r8 ;destptr1
-
-copy_top_bottom_uv_b16
- vld1.8 {q0, q1}, [r1]!
- vld1.8 {q8, q9}, [r2]!
- vld1.8 {q2, q3}, [r1]!
- vld1.8 {q10, q11}, [r2]!
-
- mov r7, r3
-
-top_bottom_8_b16
- subs r7, r7, #1
-
- vst1.8 {q0, q1}, [r5]!
- vst1.8 {q8, q9}, [r6]!
- vst1.8 {q2, q3}, [r5]!
- vst1.8 {q10, q11}, [r6]!
-
- add r5, r5, lr
- sub r5, r5, #64
- add r6, r6, lr
- sub r6, r6, #64
-
- bne top_bottom_8_b16
-
- sub r5, r1, r8
- add r6, r2, lr
-
- subs r12, r12, #1
- bne copy_top_bottom_uv_b16
-
- mov r7, lr, lsr #3 ;check to see if extra copy is needed
- ands r7, r7, #0x7
- bne extra_top_bottom_uv_b16
-
-end_of_border_copy_uv_b16
- subs r10, r10, #1
- ldrne r1, [r0, #yv12_buffer_config_v_buffer] ;srcptr1
- bne border_copy_uv_b16
-
- vpop {d8-d15}
- pop {r4 - r10, pc}
-
-;;;;;;;;;;;;;;;;;;;;;;
-;extra copy part for Y
-extra_top_bottom_y_b16
- vld1.8 {q0}, [r1]!
- vld1.8 {q2}, [r2]!
-
- mov r9, r3, lsr #3
-
-extra_top_bottom_16_b16
- subs r9, r9, #1
-
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- vst1.8 {q0}, [r5], lr
- vst1.8 {q2}, [r6], lr
- bne extra_top_bottom_16_b16
-
- sub r5, r1, r8
- add r6, r2, lr
- subs r7, r7, #1
- bne extra_top_bottom_y_b16
-
- b end_of_border_copy_y_b16
-
-;extra copy part for UV
-extra_top_bottom_uv_b16
- vld1.8 {d0}, [r1]!
- vld1.8 {d8}, [r2]!
-
- mov r9, r3, lsr #3
-
-extra_top_bottom_8_b16
- subs r9, r9, #1
-
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- vst1.8 {d0}, [r5], lr
- vst1.8 {d8}, [r6], lr
- bne extra_top_bottom_8_b16
-
- sub r5, r1, r8
- add r6, r2, lr
- subs r7, r7, #1
- bne extra_top_bottom_uv_b16
-
- b end_of_border_copy_uv_b16
-
- ENDP
- END
diff --git a/vpx_scale/arm/scalesystemdependent.c b/vpx_scale/arm/scalesystemdependent.c
deleted file mode 100644
index fee76ff..0000000
--- a/vpx_scale/arm/scalesystemdependent.c
+++ /dev/null
@@ -1,96 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_ports/arm.h"
-#include "vpx_scale/vpxscale.h"
-
-
-void (*vp8_yv12_extend_frame_borders_ptr)(YV12_BUFFER_CONFIG *ybf);
-extern void vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf);
-extern void vp8_yv12_extend_frame_borders_neon(YV12_BUFFER_CONFIG *ybf);
-
-void (*vp8_yv12_copy_frame_yonly_ptr)(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame_yonly(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame_yonly_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-void (*vp8_yv12_copy_frame_ptr)(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-/****************************************************************************
-* Imports
-*****************************************************************************/
-
-/****************************************************************************
- *
- * ROUTINE : vp8_scale_machine_specific_config
- *
- * INPUTS : UINT32 Version : Codec version number.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Checks for machine specifc features such as MMX support
- * sets appropriate flags and function pointers.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8_scale_machine_specific_config()
-{
-#if HAVE_ARMV7 && CONFIG_RUNTIME_CPU_DETECT
- int flags;
-#endif
- /*
- vp8_horizontal_line_1_2_scale = horizontal_line_1_2_scale_armv4;
- vp8_vertical_band_1_2_scale = vertical_band_1_2_scale_armv4;
- vp8_last_vertical_band_1_2_scale = vp8cx_last_vertical_band_1_2_scale_c;
- vp8_horizontal_line_3_5_scale = horizontal_line_3_5_scale_armv4;
- vp8_vertical_band_3_5_scale = vertical_band_3_5_scale_armv4;
- vp8_last_vertical_band_3_5_scale = vp8cx_last_vertical_band_3_5_scale_c;
- vp8_horizontal_line_3_4_scale = horizontal_line_3_4_scale_armv4;
- vp8_vertical_band_3_4_scale = vertical_band_3_4_scale_armv4;
- vp8_last_vertical_band_3_4_scale = vp8cx_last_vertical_band_3_4_scale_c;
- vp8_horizontal_line_2_3_scale = horizontal_line_2_3_scale_armv4;
- vp8_vertical_band_2_3_scale = vertical_band_2_3_scale_armv4;
- vp8_last_vertical_band_2_3_scale = vp8cx_last_vertical_band_2_3_scale_c;
- vp8_horizontal_line_4_5_scale = horizontal_line_4_5_scale_armv4;
- vp8_vertical_band_4_5_scale = vertical_band_4_5_scale_armv4;
- vp8_last_vertical_band_4_5_scale = vp8cx_last_vertical_band_4_5_scale_c;
-
- vp8_vertical_band_5_4_scale = vp8cx_vertical_band_5_4_scale_c;
- vp8_vertical_band_5_3_scale = vp8cx_vertical_band_5_3_scale_c;
- vp8_vertical_band_2_1_scale = vp8cx_vertical_band_2_1_scale_c;
- vp8_vertical_band_2_1_scale_i = vp8cx_vertical_band_2_1_scale_i_c;
- vp8_horizontal_line_2_1_scale = vp8cx_horizontal_line_2_1_scale_c;
- vp8_horizontal_line_5_3_scale = vp8cx_horizontal_line_5_3_scale_c;
- vp8_horizontal_line_5_4_scale = vp8cx_horizontal_line_5_4_scale_c;
- */
-
-#if !HAVE_ARMV7 || CONFIG_RUNTIME_CPU_DETECT
- vp8_yv12_extend_frame_borders_ptr = vp8_yv12_extend_frame_borders;
- vp8_yv12_copy_frame_yonly_ptr = vp8_yv12_copy_frame_yonly;
- vp8_yv12_copy_frame_ptr = vp8_yv12_copy_frame;
-#endif
-#if HAVE_ARMV7
-#if CONFIG_RUNTIME_CPU_DETECT
- flags = arm_cpu_caps();
- if (flags & HAS_NEON)
-#endif
- {
- vp8_yv12_extend_frame_borders_ptr = vp8_yv12_extend_frame_borders_neon;
- vp8_yv12_copy_frame_yonly_ptr = vp8_yv12_copy_frame_yonly_neon;
- vp8_yv12_copy_frame_ptr = vp8_yv12_copy_frame_neon;
- }
-#endif
-}
diff --git a/vpx_scale/arm/yv12extend_arm.c b/vpx_scale/arm/yv12extend_arm.c
deleted file mode 100644
index d7a8289..0000000
--- a/vpx_scale/arm/yv12extend_arm.c
+++ /dev/null
@@ -1,25 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_scale/yv12config.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vpx_scale/vpxscale.h"
-
-void vp8_yv12_copy_frame_func_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-void
-vp8_yv12_copy_frame_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc)
-{
- vp8_yv12_copy_frame_func_neon(src_ybc, dst_ybc);
- //printf("Border:%d; plane_stride:%d; plane_height:%d; plane_width:%d\n",dst_ybc->border,dst_ybc->y_stride,dst_ybc->y_height,dst_ybc->y_width);
-
- vp8_yv12_extend_frame_borders_ptr(dst_ybc);
-}
diff --git a/vpx_scale/generic/bicubic_scaler.c b/vpx_scale/generic/bicubic_scaler.c
deleted file mode 100644
index 4468e9d..0000000
--- a/vpx_scale/generic/bicubic_scaler.c
+++ /dev/null
@@ -1,602 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include <float.h>
-#include <math.h>
-#include <stdio.h>
-#include "vpx_mem/vpx_mem.h"
-#include "vpxscale_arbitrary.h"
-
-#define FIXED_POINT
-
-#define MAX_IN_WIDTH 800
-#define MAX_IN_HEIGHT 600
-#define MAX_OUT_WIDTH 800
-#define MAX_OUT_HEIGHT 600
-#define MAX_OUT_DIMENSION ((MAX_OUT_WIDTH > MAX_OUT_HEIGHT) ? \
- MAX_OUT_WIDTH : MAX_OUT_HEIGHT)
-
-BICUBIC_SCALER_STRUCT g_b_scaler;
-static int g_first_time = 1;
-
-#pragma DATA_SECTION(g_hbuf, "VP6_HEAP")
-#pragma DATA_ALIGN (g_hbuf, 32);
-unsigned char g_hbuf[MAX_OUT_DIMENSION];
-
-#pragma DATA_SECTION(g_hbuf_uv, "VP6_HEAP")
-#pragma DATA_ALIGN (g_hbuf_uv, 32);
-unsigned char g_hbuf_uv[MAX_OUT_DIMENSION];
-
-
-#ifdef FIXED_POINT
-static int a_i = 0.6 * 65536;
-#else
-static float a = -0.6;
-#endif
-
-#ifdef FIXED_POINT
-// 3 2
-// C0 = a*t - a*t
-//
-static short c0_fixed(unsigned int t)
-{
- // put t in Q16 notation
- unsigned short v1, v2;
-
- // Q16
- v1 = (a_i * t) >> 16;
- v1 = (v1 * t) >> 16;
-
- // Q16
- v2 = (a_i * t) >> 16;
- v2 = (v2 * t) >> 16;
- v2 = (v2 * t) >> 16;
-
- // Q12
- return -((v1 - v2) >> 4);
-}
-
-// 2 3
-// C1 = a*t + (3-2*a)*t - (2-a)*t
-//
-static short c1_fixed(unsigned int t)
-{
- unsigned short v1, v2, v3;
- unsigned short two, three;
-
- // Q16
- v1 = (a_i * t) >> 16;
-
- // Q13
- two = 2 << 13;
- v2 = two - (a_i >> 3);
- v2 = (v2 * t) >> 16;
- v2 = (v2 * t) >> 16;
- v2 = (v2 * t) >> 16;
-
- // Q13
- three = 3 << 13;
- v3 = three - (2 * (a_i >> 3));
- v3 = (v3 * t) >> 16;
- v3 = (v3 * t) >> 16;
-
- // Q12
- return (((v1 >> 3) - v2 + v3) >> 1);
-
-}
-
-// 2 3
-// C2 = 1 - (3-a)*t + (2-a)*t
-//
-static short c2_fixed(unsigned int t)
-{
- unsigned short v1, v2, v3;
- unsigned short two, three;
-
- // Q13
- v1 = 1 << 13;
-
- // Q13
- three = 3 << 13;
- v2 = three - (a_i >> 3);
- v2 = (v2 * t) >> 16;
- v2 = (v2 * t) >> 16;
-
- // Q13
- two = 2 << 13;
- v3 = two - (a_i >> 3);
- v3 = (v3 * t) >> 16;
- v3 = (v3 * t) >> 16;
- v3 = (v3 * t) >> 16;
-
- // Q12
- return (v1 - v2 + v3) >> 1;
-}
-
-// 2 3
-// C3 = a*t - 2*a*t + a*t
-//
-static short c3_fixed(unsigned int t)
-{
- int v1, v2, v3;
-
- // Q16
- v1 = (a_i * t) >> 16;
-
- // Q15
- v2 = 2 * (a_i >> 1);
- v2 = (v2 * t) >> 16;
- v2 = (v2 * t) >> 16;
-
- // Q16
- v3 = (a_i * t) >> 16;
- v3 = (v3 * t) >> 16;
- v3 = (v3 * t) >> 16;
-
- // Q12
- return ((v2 - (v1 >> 1) - (v3 >> 1)) >> 3);
-}
-#else
-// 3 2
-// C0 = -a*t + a*t
-//
-float C0(float t)
-{
- return -a * t * t * t + a * t * t;
-}
-
-// 2 3
-// C1 = -a*t + (2*a+3)*t - (a+2)*t
-//
-float C1(float t)
-{
- return -(a + 2.0f) * t * t * t + (2.0f * a + 3.0f) * t * t - a * t;
-}
-
-// 2 3
-// C2 = 1 - (a+3)*t + (a+2)*t
-//
-float C2(float t)
-{
- return (a + 2.0f) * t * t * t - (a + 3.0f) * t * t + 1.0f;
-}
-
-// 2 3
-// C3 = a*t - 2*a*t + a*t
-//
-float C3(float t)
-{
- return a * t * t * t - 2.0f * a * t * t + a * t;
-}
-#endif
-
-#if 0
-int compare_real_fixed()
-{
- int i, errors = 0;
- float mult = 1.0 / 10000.0;
- unsigned int fixed_mult = mult * 4294967296;//65536;
- unsigned int phase_offset_int;
- float phase_offset_real;
-
- for (i = 0; i < 10000; i++)
- {
- int fixed0, fixed1, fixed2, fixed3, fixed_total;
- int real0, real1, real2, real3, real_total;
-
- phase_offset_real = (float)i * mult;
- phase_offset_int = (fixed_mult * i) >> 16;
-// phase_offset_int = phase_offset_real * 65536;
-
- fixed0 = c0_fixed(phase_offset_int);
- real0 = C0(phase_offset_real) * 4096.0;
-
- if ((abs(fixed0) > (abs(real0) + 1)) || (abs(fixed0) < (abs(real0) - 1)))
- errors++;
-
- fixed1 = c1_fixed(phase_offset_int);
- real1 = C1(phase_offset_real) * 4096.0;
-
- if ((abs(fixed1) > (abs(real1) + 1)) || (abs(fixed1) < (abs(real1) - 1)))
- errors++;
-
- fixed2 = c2_fixed(phase_offset_int);
- real2 = C2(phase_offset_real) * 4096.0;
-
- if ((abs(fixed2) > (abs(real2) + 1)) || (abs(fixed2) < (abs(real2) - 1)))
- errors++;
-
- fixed3 = c3_fixed(phase_offset_int);
- real3 = C3(phase_offset_real) * 4096.0;
-
- if ((abs(fixed3) > (abs(real3) + 1)) || (abs(fixed3) < (abs(real3) - 1)))
- errors++;
-
- fixed_total = fixed0 + fixed1 + fixed2 + fixed3;
- real_total = real0 + real1 + real2 + real3;
-
- if ((fixed_total > 4097) || (fixed_total < 4094))
- errors ++;
-
- if ((real_total > 4097) || (real_total < 4095))
- errors ++;
- }
-
- return errors;
-}
-#endif
-
-// Find greatest common denominator between two integers. Method used here is
-// slow compared to Euclid's algorithm, but does not require any division.
-int gcd(int a, int b)
-{
- // Problem with this algorithm is that if a or b = 0 this function
- // will never exit. Don't want to return 0 because any computation
- // that was based on a common denoninator and tried to reduce by
- // dividing by 0 would fail. Best solution that could be thought of
- // would to be fail by returing a 1;
- if (a <= 0 || b <= 0)
- return 1;
-
- while (a != b)
- {
- if (b > a)
- b = b - a;
- else
- {
- int tmp = a;//swap large and
- a = b; //small
- b = tmp;
- }
- }
-
- return b;
-}
-
-void bicubic_coefficient_init()
-{
- vpx_memset(&g_b_scaler, 0, sizeof(BICUBIC_SCALER_STRUCT));
- g_first_time = 0;
-}
-
-void bicubic_coefficient_destroy()
-{
- if (!g_first_time)
- {
- vpx_free(g_b_scaler.l_w);
-
- vpx_free(g_b_scaler.l_h);
-
- vpx_free(g_b_scaler.l_h_uv);
-
- vpx_free(g_b_scaler.c_w);
-
- vpx_free(g_b_scaler.c_h);
-
- vpx_free(g_b_scaler.c_h_uv);
-
- vpx_memset(&g_b_scaler, 0, sizeof(BICUBIC_SCALER_STRUCT));
- }
-}
-
-// Create the coeffients that will be used for the cubic interpolation.
-// Because scaling does not have to be equal in the vertical and horizontal
-// regimes the phase offsets will be different. There are 4 coefficents
-// for each point, two on each side. The layout is that there are the
-// 4 coefficents for each phase in the array and then the next phase.
-int bicubic_coefficient_setup(int in_width, int in_height, int out_width, int out_height)
-{
- int i;
-#ifdef FIXED_POINT
- int phase_offset_int;
- unsigned int fixed_mult;
- int product_val = 0;
-#else
- float phase_offset;
-#endif
- int gcd_w, gcd_h, gcd_h_uv, d_w, d_h, d_h_uv;
-
- if (g_first_time)
- bicubic_coefficient_init();
-
-
- // check to see if the coefficents have already been set up correctly
- if ((in_width == g_b_scaler.in_width) && (in_height == g_b_scaler.in_height)
- && (out_width == g_b_scaler.out_width) && (out_height == g_b_scaler.out_height))
- return 0;
-
- g_b_scaler.in_width = in_width;
- g_b_scaler.in_height = in_height;
- g_b_scaler.out_width = out_width;
- g_b_scaler.out_height = out_height;
-
- // Don't want to allow crazy scaling, just try and prevent a catastrophic
- // failure here. Want to fail after setting the member functions so if
- // if the scaler is called the member functions will not scale.
- if (out_width <= 0 || out_height <= 0)
- return -1;
-
- // reduce in/out width and height ratios using the gcd
- gcd_w = gcd(out_width, in_width);
- gcd_h = gcd(out_height, in_height);
- gcd_h_uv = gcd(out_height, in_height / 2);
-
- // the numerator width and height are to be saved in
- // globals so they can be used during the scaling process
- // without having to be recalculated.
- g_b_scaler.nw = out_width / gcd_w;
- d_w = in_width / gcd_w;
-
- g_b_scaler.nh = out_height / gcd_h;
- d_h = in_height / gcd_h;
-
- g_b_scaler.nh_uv = out_height / gcd_h_uv;
- d_h_uv = (in_height / 2) / gcd_h_uv;
-
- // allocate memory for the coefficents
- vpx_free(g_b_scaler.l_w);
-
- vpx_free(g_b_scaler.l_h);
-
- vpx_free(g_b_scaler.l_h_uv);
-
- g_b_scaler.l_w = (short *)vpx_memalign(32, out_width * 2);
- g_b_scaler.l_h = (short *)vpx_memalign(32, out_height * 2);
- g_b_scaler.l_h_uv = (short *)vpx_memalign(32, out_height * 2);
-
- vpx_free(g_b_scaler.c_w);
-
- vpx_free(g_b_scaler.c_h);
-
- vpx_free(g_b_scaler.c_h_uv);
-
- g_b_scaler.c_w = (short *)vpx_memalign(32, g_b_scaler.nw * 4 * 2);
- g_b_scaler.c_h = (short *)vpx_memalign(32, g_b_scaler.nh * 4 * 2);
- g_b_scaler.c_h_uv = (short *)vpx_memalign(32, g_b_scaler.nh_uv * 4 * 2);
-
- g_b_scaler.hbuf = g_hbuf;
- g_b_scaler.hbuf_uv = g_hbuf_uv;
-
- // Set up polyphase filter taps. This needs to be done before
- // the scaling because of the floating point math required. The
- // coefficients are multiplied by 2^12 so that fixed point math
- // can be used in the main scaling loop.
-#ifdef FIXED_POINT
- fixed_mult = (1.0 / (float)g_b_scaler.nw) * 4294967296;
-
- product_val = 0;
-
- for (i = 0; i < g_b_scaler.nw; i++)
- {
- if (product_val > g_b_scaler.nw)
- product_val -= g_b_scaler.nw;
-
- phase_offset_int = (fixed_mult * product_val) >> 16;
-
- g_b_scaler.c_w[i*4] = c3_fixed(phase_offset_int);
- g_b_scaler.c_w[i*4+1] = c2_fixed(phase_offset_int);
- g_b_scaler.c_w[i*4+2] = c1_fixed(phase_offset_int);
- g_b_scaler.c_w[i*4+3] = c0_fixed(phase_offset_int);
-
- product_val += d_w;
- }
-
-
- fixed_mult = (1.0 / (float)g_b_scaler.nh) * 4294967296;
-
- product_val = 0;
-
- for (i = 0; i < g_b_scaler.nh; i++)
- {
- if (product_val > g_b_scaler.nh)
- product_val -= g_b_scaler.nh;
-
- phase_offset_int = (fixed_mult * product_val) >> 16;
-
- g_b_scaler.c_h[i*4] = c0_fixed(phase_offset_int);
- g_b_scaler.c_h[i*4+1] = c1_fixed(phase_offset_int);
- g_b_scaler.c_h[i*4+2] = c2_fixed(phase_offset_int);
- g_b_scaler.c_h[i*4+3] = c3_fixed(phase_offset_int);
-
- product_val += d_h;
- }
-
- fixed_mult = (1.0 / (float)g_b_scaler.nh_uv) * 4294967296;
-
- product_val = 0;
-
- for (i = 0; i < g_b_scaler.nh_uv; i++)
- {
- if (product_val > g_b_scaler.nh_uv)
- product_val -= g_b_scaler.nh_uv;
-
- phase_offset_int = (fixed_mult * product_val) >> 16;
-
- g_b_scaler.c_h_uv[i*4] = c0_fixed(phase_offset_int);
- g_b_scaler.c_h_uv[i*4+1] = c1_fixed(phase_offset_int);
- g_b_scaler.c_h_uv[i*4+2] = c2_fixed(phase_offset_int);
- g_b_scaler.c_h_uv[i*4+3] = c3_fixed(phase_offset_int);
-
- product_val += d_h_uv;
- }
-
-#else
-
- for (i = 0; i < g_nw; i++)
- {
- phase_offset = (float)((i * d_w) % g_nw) / (float)g_nw;
- g_c_w[i*4] = (C3(phase_offset) * 4096.0);
- g_c_w[i*4+1] = (C2(phase_offset) * 4096.0);
- g_c_w[i*4+2] = (C1(phase_offset) * 4096.0);
- g_c_w[i*4+3] = (C0(phase_offset) * 4096.0);
- }
-
- for (i = 0; i < g_nh; i++)
- {
- phase_offset = (float)((i * d_h) % g_nh) / (float)g_nh;
- g_c_h[i*4] = (C0(phase_offset) * 4096.0);
- g_c_h[i*4+1] = (C1(phase_offset) * 4096.0);
- g_c_h[i*4+2] = (C2(phase_offset) * 4096.0);
- g_c_h[i*4+3] = (C3(phase_offset) * 4096.0);
- }
-
- for (i = 0; i < g_nh_uv; i++)
- {
- phase_offset = (float)((i * d_h_uv) % g_nh_uv) / (float)g_nh_uv;
- g_c_h_uv[i*4] = (C0(phase_offset) * 4096.0);
- g_c_h_uv[i*4+1] = (C1(phase_offset) * 4096.0);
- g_c_h_uv[i*4+2] = (C2(phase_offset) * 4096.0);
- g_c_h_uv[i*4+3] = (C3(phase_offset) * 4096.0);
- }
-
-#endif
-
- // Create an array that corresponds input lines to output lines.
- // This doesn't require floating point math, but it does require
- // a division and because hardware division is not present that
- // is a call.
- for (i = 0; i < out_width; i++)
- {
- g_b_scaler.l_w[i] = (i * d_w) / g_b_scaler.nw;
-
- if ((g_b_scaler.l_w[i] + 2) <= in_width)
- g_b_scaler.max_usable_out_width = i;
-
- }
-
- for (i = 0; i < out_height + 1; i++)
- {
- g_b_scaler.l_h[i] = (i * d_h) / g_b_scaler.nh;
- g_b_scaler.l_h_uv[i] = (i * d_h_uv) / g_b_scaler.nh_uv;
- }
-
- return 0;
-}
-
-int bicubic_scale(int in_width, int in_height, int in_stride,
- int out_width, int out_height, int out_stride,
- unsigned char *input_image, unsigned char *output_image)
-{
- short *RESTRICT l_w, * RESTRICT l_h;
- short *RESTRICT c_w, * RESTRICT c_h;
- unsigned char *RESTRICT ip, * RESTRICT op;
- unsigned char *RESTRICT hbuf;
- int h, w, lw, lh;
- int temp_sum;
- int phase_offset_w, phase_offset_h;
-
- c_w = g_b_scaler.c_w;
- c_h = g_b_scaler.c_h;
-
- op = output_image;
-
- l_w = g_b_scaler.l_w;
- l_h = g_b_scaler.l_h;
-
- phase_offset_h = 0;
-
- for (h = 0; h < out_height; h++)
- {
- // select the row to work on
- lh = l_h[h];
- ip = input_image + (in_stride * lh);
-
- // vp8_filter the row vertically into an temporary buffer.
- // If the phase offset == 0 then all the multiplication
- // is going to result in the output equalling the input.
- // So instead point the temporary buffer to the input.
- // Also handle the boundry condition of not being able to
- // filter that last lines.
- if (phase_offset_h && (lh < in_height - 2))
- {
- hbuf = g_b_scaler.hbuf;
-
- for (w = 0; w < in_width; w++)
- {
- temp_sum = c_h[phase_offset_h*4+3] * ip[w - in_stride];
- temp_sum += c_h[phase_offset_h*4+2] * ip[w];
- temp_sum += c_h[phase_offset_h*4+1] * ip[w + in_stride];
- temp_sum += c_h[phase_offset_h*4] * ip[w + 2*in_stride];
-
- hbuf[w] = temp_sum >> 12;
- }
- }
- else
- hbuf = ip;
-
- // increase the phase offset for the next time around.
- if (++phase_offset_h >= g_b_scaler.nh)
- phase_offset_h = 0;
-
- // now filter and expand it horizontally into the final
- // output buffer
- phase_offset_w = 0;
-
- for (w = 0; w < out_width; w++)
- {
- // get the index to use to expand the image
- lw = l_w[w];
-
- temp_sum = c_w[phase_offset_w*4] * hbuf[lw - 1];
- temp_sum += c_w[phase_offset_w*4+1] * hbuf[lw];
- temp_sum += c_w[phase_offset_w*4+2] * hbuf[lw + 1];
- temp_sum += c_w[phase_offset_w*4+3] * hbuf[lw + 2];
- temp_sum = temp_sum >> 12;
-
- if (++phase_offset_w >= g_b_scaler.nw)
- phase_offset_w = 0;
-
- // boundry conditions
- if ((lw + 2) >= in_width)
- temp_sum = hbuf[lw];
-
- if (lw == 0)
- temp_sum = hbuf[0];
-
- op[w] = temp_sum;
- }
-
- op += out_stride;
- }
-
- return 0;
-}
-
-void bicubic_scale_frame_reset()
-{
- g_b_scaler.out_width = 0;
- g_b_scaler.out_height = 0;
-}
-
-void bicubic_scale_frame(YV12_BUFFER_CONFIG *src, YV12_BUFFER_CONFIG *dst,
- int new_width, int new_height)
-{
-
- dst->y_width = new_width;
- dst->y_height = new_height;
- dst->uv_width = new_width / 2;
- dst->uv_height = new_height / 2;
-
- dst->y_stride = dst->y_width;
- dst->uv_stride = dst->uv_width;
-
- bicubic_scale(src->y_width, src->y_height, src->y_stride,
- new_width, new_height, dst->y_stride,
- src->y_buffer, dst->y_buffer);
-
- bicubic_scale(src->uv_width, src->uv_height, src->uv_stride,
- new_width / 2, new_height / 2, dst->uv_stride,
- src->u_buffer, dst->u_buffer);
-
- bicubic_scale(src->uv_width, src->uv_height, src->uv_stride,
- new_width / 2, new_height / 2, dst->uv_stride,
- src->v_buffer, dst->v_buffer);
-}
diff --git a/vpx_scale/generic/gen_scalers.c b/vpx_scale/generic/gen_scalers.c
deleted file mode 100644
index b54e334..0000000
--- a/vpx_scale/generic/gen_scalers.c
+++ /dev/null
@@ -1,956 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_scale/vpxscale.h"
-#include "vpx_mem/vpx_mem.h"
-/****************************************************************************
-* Imports
-****************************************************************************/
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_horizontal_line_4_5_scale_c
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 4 to 5.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8cx_horizontal_line_4_5_scale_c
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- unsigned i;
- unsigned int a, b, c;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for (i = 0; i < source_width - 4; i += 4)
- {
- a = src[0];
- b = src[1];
- des [0] = (unsigned char) a;
- des [1] = (unsigned char)((a * 51 + 205 * b + 128) >> 8);
- c = src[2] * 154;
- a = src[3];
- des [2] = (unsigned char)((b * 102 + c + 128) >> 8);
- des [3] = (unsigned char)((c + 102 * a + 128) >> 8);
- b = src[4];
- des [4] = (unsigned char)((a * 205 + 51 * b + 128) >> 8);
-
- src += 4;
- des += 5;
- }
-
- a = src[0];
- b = src[1];
- des [0] = (unsigned char)(a);
- des [1] = (unsigned char)((a * 51 + 205 * b + 128) >> 8);
- c = src[2] * 154;
- a = src[3];
- des [2] = (unsigned char)((b * 102 + c + 128) >> 8);
- des [3] = (unsigned char)((c + 102 * a + 128) >> 8);
- des [4] = (unsigned char)(a);
-
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_vertical_band_4_5_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales vertical band of pixels by scale 4 to 5. The
- * height of the band scaled is 4-pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band.
- *
- ****************************************************************************/
-void vp8cx_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c, d;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; i++)
- {
- a = des [0];
- b = des [dest_pitch];
-
- des[dest_pitch] = (unsigned char)((a * 51 + 205 * b + 128) >> 8);
-
- c = des[dest_pitch*2] * 154;
- d = des[dest_pitch*3];
-
- des [dest_pitch*2] = (unsigned char)((b * 102 + c + 128) >> 8);
- des [dest_pitch*3] = (unsigned char)((c + 102 * d + 128) >> 8);
-
- /* First line in next band */
- a = des [dest_pitch * 5];
- des [dest_pitch * 4] = (unsigned char)((d * 205 + 51 * a + 128) >> 8);
-
- des ++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_last_vertical_band_4_5_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales last vertical band of pixels by scale 4 to 5. The
- * height of the band scaled is 4-pixels.
- *
- * SPECIAL NOTES : The routine does not have available the first line of
- * the band below the current band, since this is the
- * last band.
- *
- ****************************************************************************/
-void vp8cx_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c, d;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; ++i)
- {
- a = des[0];
- b = des[dest_pitch];
-
- des[dest_pitch] = (unsigned char)((a * 51 + 205 * b + 128) >> 8);
-
- c = des[dest_pitch*2] * 154;
- d = des[dest_pitch*3];
-
- des [dest_pitch*2] = (unsigned char)((b * 102 + c + 128) >> 8);
- des [dest_pitch*3] = (unsigned char)((c + 102 * d + 128) >> 8);
-
- /* No other line for interplation of this line, so .. */
- des[dest_pitch*4] = (unsigned char) d;
-
- des++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_horizontal_line_2_3_scale_c
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 2 to 3.
- *
- * SPECIAL NOTES : None.
- *
- *
- ****************************************************************************/
-void vp8cx_horizontal_line_2_3_scale_c
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- unsigned int i;
- unsigned int a, b, c;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for (i = 0; i < source_width - 2; i += 2)
- {
- a = src[0];
- b = src[1];
- c = src[2];
-
- des [0] = (unsigned char)(a);
- des [1] = (unsigned char)((a * 85 + 171 * b + 128) >> 8);
- des [2] = (unsigned char)((b * 171 + 85 * c + 128) >> 8);
-
- src += 2;
- des += 3;
- }
-
- a = src[0];
- b = src[1];
- des [0] = (unsigned char)(a);
- des [1] = (unsigned char)((a * 85 + 171 * b + 128) >> 8);
- des [2] = (unsigned char)(b);
-}
-
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_vertical_band_2_3_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales vertical band of pixels by scale 2 to 3. The
- * height of the band scaled is 2-pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band.
- *
- ****************************************************************************/
-void vp8cx_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; i++)
- {
- a = des [0];
- b = des [dest_pitch];
- c = des[dest_pitch*3];
- des [dest_pitch ] = (unsigned char)((a * 85 + 171 * b + 128) >> 8);
- des [dest_pitch*2] = (unsigned char)((b * 171 + 85 * c + 128) >> 8);
-
- des++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_last_vertical_band_2_3_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales last vertical band of pixels by scale 2 to 3. The
- * height of the band scaled is 2-pixels.
- *
- * SPECIAL NOTES : The routine does not have available the first line of
- * the band below the current band, since this is the
- * last band.
- *
- ****************************************************************************/
-void vp8cx_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; ++i)
- {
- a = des [0];
- b = des [dest_pitch];
-
- des [dest_pitch ] = (unsigned char)((a * 85 + 171 * b + 128) >> 8);
- des [dest_pitch*2] = (unsigned char)(b);
- des++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_horizontal_line_3_5_scale_c
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 3 to 5.
- *
- * SPECIAL NOTES : None.
- *
- *
- ****************************************************************************/
-void vp8cx_horizontal_line_3_5_scale_c
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- unsigned int i;
- unsigned int a, b, c;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for (i = 0; i < source_width - 3; i += 3)
- {
- a = src[0];
- b = src[1];
- des [0] = (unsigned char)(a);
- des [1] = (unsigned char)((a * 102 + 154 * b + 128) >> 8);
-
- c = src[2] ;
- des [2] = (unsigned char)((b * 205 + c * 51 + 128) >> 8);
- des [3] = (unsigned char)((b * 51 + c * 205 + 128) >> 8);
-
- a = src[3];
- des [4] = (unsigned char)((c * 154 + a * 102 + 128) >> 8);
-
- src += 3;
- des += 5;
- }
-
- a = src[0];
- b = src[1];
- des [0] = (unsigned char)(a);
-
- des [1] = (unsigned char)((a * 102 + 154 * b + 128) >> 8);
- c = src[2] ;
- des [2] = (unsigned char)((b * 205 + c * 51 + 128) >> 8);
- des [3] = (unsigned char)((b * 51 + c * 205 + 128) >> 8);
-
- des [4] = (unsigned char)(c);
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_vertical_band_3_5_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales vertical band of pixels by scale 3 to 5. The
- * height of the band scaled is 3-pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band.
- *
- ****************************************************************************/
-void vp8cx_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; i++)
- {
- a = des [0];
- b = des [dest_pitch];
- des [dest_pitch] = (unsigned char)((a * 102 + 154 * b + 128) >> 8);
-
- c = des[dest_pitch*2];
- des [dest_pitch*2] = (unsigned char)((b * 205 + c * 51 + 128) >> 8);
- des [dest_pitch*3] = (unsigned char)((b * 51 + c * 205 + 128) >> 8);
-
- /* First line in next band... */
- a = des [dest_pitch * 5];
- des [dest_pitch * 4] = (unsigned char)((c * 154 + a * 102 + 128) >> 8);
-
- des++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_last_vertical_band_3_5_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales last vertical band of pixels by scale 3 to 5. The
- * height of the band scaled is 3-pixels.
- *
- * SPECIAL NOTES : The routine does not have available the first line of
- * the band below the current band, since this is the
- * last band.
- *
- ****************************************************************************/
-void vp8cx_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; ++i)
- {
- a = des [0];
- b = des [dest_pitch];
-
- des [ dest_pitch ] = (unsigned char)((a * 102 + 154 * b + 128) >> 8);
-
- c = des[dest_pitch*2];
- des [dest_pitch*2] = (unsigned char)((b * 205 + c * 51 + 128) >> 8);
- des [dest_pitch*3] = (unsigned char)((b * 51 + c * 205 + 128) >> 8);
-
- /* No other line for interplation of this line, so .. */
- des [ dest_pitch * 4 ] = (unsigned char)(c) ;
-
- des++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_horizontal_line_3_4_scale_c
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 3 to 4.
- *
- * SPECIAL NOTES : None.
- *
- *
- ****************************************************************************/
-void vp8cx_horizontal_line_3_4_scale_c
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- unsigned int i;
- unsigned int a, b, c;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for (i = 0; i < source_width - 3; i += 3)
- {
- a = src[0];
- b = src[1];
- des [0] = (unsigned char)(a);
- des [1] = (unsigned char)((a * 64 + b * 192 + 128) >> 8);
-
- c = src[2];
- des [2] = (unsigned char)((b + c + 1) >> 1);
-
- a = src[3];
- des [3] = (unsigned char)((c * 192 + a * 64 + 128) >> 8);
-
- src += 3;
- des += 4;
- }
-
- a = src[0];
- b = src[1];
- des [0] = (unsigned char)(a);
- des [1] = (unsigned char)((a * 64 + b * 192 + 128) >> 8);
-
- c = src[2] ;
- des [2] = (unsigned char)((b + c + 1) >> 1);
- des [3] = (unsigned char)(c);
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_vertical_band_3_4_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales vertical band of pixels by scale 3 to 4. The
- * height of the band scaled is 3-pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band.
- *
- ****************************************************************************/
-void vp8cx_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; i++)
- {
- a = des [0];
- b = des [dest_pitch];
- des [dest_pitch] = (unsigned char)((a * 64 + b * 192 + 128) >> 8);
-
- c = des[dest_pitch*2];
- des [dest_pitch*2] = (unsigned char)((b + c + 1) >> 1);
-
- /* First line in next band... */
- a = des [dest_pitch*4];
- des [dest_pitch*3] = (unsigned char)((c * 192 + a * 64 + 128) >> 8);
-
- des++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_last_vertical_band_3_4_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales last vertical band of pixels by scale 3 to 4. The
- * height of the band scaled is 3-pixels.
- *
- * SPECIAL NOTES : The routine does not have available the first line of
- * the band below the current band, since this is the
- * last band.
- *
- ****************************************************************************/
-void vp8cx_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; ++i)
- {
- a = des [0];
- b = des [dest_pitch];
-
- des [dest_pitch] = (unsigned char)((a * 64 + b * 192 + 128) >> 8);
-
- c = des[dest_pitch*2];
- des [dest_pitch*2] = (unsigned char)((b + c + 1) >> 1);
-
- /* No other line for interplation of this line, so .. */
- des [dest_pitch*3] = (unsigned char)(c);
-
- des++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_horizontal_line_1_2_scale_c
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 1 to 2.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8cx_horizontal_line_1_2_scale_c
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- unsigned int i;
- unsigned int a, b;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for (i = 0; i < source_width - 1; i += 1)
- {
- a = src[0];
- b = src[1];
- des [0] = (unsigned char)(a);
- des [1] = (unsigned char)((a + b + 1) >> 1);
- src += 1;
- des += 2;
- }
-
- a = src[0];
- des [0] = (unsigned char)(a);
- des [1] = (unsigned char)(a);
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_vertical_band_1_2_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales vertical band of pixels by scale 1 to 2. The
- * height of the band scaled is 1-pixel.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band.
- *
- ****************************************************************************/
-void vp8cx_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; i++)
- {
- a = des [0];
- b = des [dest_pitch * 2];
-
- des[dest_pitch] = (unsigned char)((a + b + 1) >> 1);
-
- des++;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_last_vertical_band_1_2_scale_c
- *
- * INPUTS : unsigned char *dest : Pointer to destination data.
- * unsigned int dest_pitch : Stride of destination data.
- * unsigned int dest_width : Width of destination data.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Scales last vertical band of pixels by scale 1 to 2. The
- * height of the band scaled is 1-pixel.
- *
- * SPECIAL NOTES : The routine does not have available the first line of
- * the band below the current band, since this is the
- * last band.
- *
- ****************************************************************************/
-void vp8cx_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned char *des = dest;
-
- for (i = 0; i < dest_width; ++i)
- {
- des[dest_pitch] = des[0];
- des++;
- }
-}
-
-
-
-
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_horizontal_line_4_5_scale_c
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 4 to 5.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8cx_horizontal_line_5_4_scale_c
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- unsigned i;
- unsigned int a, b, c, d, e;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for (i = 0; i < source_width; i += 5)
- {
- a = src[0];
- b = src[1];
- c = src[2];
- d = src[3];
- e = src[4];
-
- des[0] = (unsigned char) a;
- des[1] = (unsigned char)((b * 192 + c * 64 + 128) >> 8);
- des[2] = (unsigned char)((c * 128 + d * 128 + 128) >> 8);
- des[3] = (unsigned char)((d * 64 + e * 192 + 128) >> 8);
-
- src += 5;
- des += 4;
- }
-}
-
-
-
-
-void vp8cx_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c, d, e;
- unsigned char *des = dest;
- unsigned char *src = source;
-
- for (i = 0; i < dest_width; i++)
- {
-
- a = src[0 * src_pitch];
- b = src[1 * src_pitch];
- c = src[2 * src_pitch];
- d = src[3 * src_pitch];
- e = src[4 * src_pitch];
-
- des[0 * dest_pitch] = (unsigned char) a;
- des[1 * dest_pitch] = (unsigned char)((b * 192 + c * 64 + 128) >> 8);
- des[2 * dest_pitch] = (unsigned char)((c * 128 + d * 128 + 128) >> 8);
- des[3 * dest_pitch] = (unsigned char)((d * 64 + e * 192 + 128) >> 8);
-
- src ++;
- des ++;
-
- }
-}
-
-
-/*7***************************************************************************
- *
- * ROUTINE : vp8cx_horizontal_line_3_5_scale_c
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 3 to 5.
- *
- * SPECIAL NOTES : None.
- *
- *
- ****************************************************************************/
-void vp8cx_horizontal_line_5_3_scale_c
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- unsigned int i;
- unsigned int a, b, c, d , e;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for (i = 0; i < source_width; i += 5)
- {
- a = src[0];
- b = src[1];
- c = src[2];
- d = src[3];
- e = src[4];
-
- des[0] = (unsigned char) a;
- des[1] = (unsigned char)((b * 85 + c * 171 + 128) >> 8);
- des[2] = (unsigned char)((d * 171 + e * 85 + 128) >> 8);
-
- src += 5;
- des += 3;
- }
-
-}
-
-void vp8cx_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- unsigned int i;
- unsigned int a, b, c, d, e;
- unsigned char *des = dest;
- unsigned char *src = source;
-
- for (i = 0; i < dest_width; i++)
- {
-
- a = src[0 * src_pitch];
- b = src[1 * src_pitch];
- c = src[2 * src_pitch];
- d = src[3 * src_pitch];
- e = src[4 * src_pitch];
-
- des[0 * dest_pitch] = (unsigned char) a;
- des[1 * dest_pitch] = (unsigned char)((b * 85 + c * 171 + 128) >> 8);
- des[2 * dest_pitch] = (unsigned char)((d * 171 + e * 85 + 128) >> 8);
-
- src ++;
- des ++;
-
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vp8cx_horizontal_line_1_2_scale_c
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 1 to 2.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8cx_horizontal_line_2_1_scale_c
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- unsigned int i;
- unsigned int a;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for (i = 0; i < source_width; i += 2)
- {
- a = src[0];
- des [0] = (unsigned char)(a);
- src += 2;
- des += 1;
- }
-
-
-
-}
-void vp8cx_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- (void) dest_pitch;
- (void) src_pitch;
- vpx_memcpy(dest, source, dest_width);
-}
-
-void vp8cx_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- int i;
- int temp;
- int width = dest_width;
-
- (void) dest_pitch;
-
- for (i = 0; i < width; i++)
- {
- temp = 8;
- temp += source[i-(int)src_pitch] * 3;
- temp += source[i] * 10;
- temp += source[i+src_pitch] * 3;
- temp >>= 4 ;
- dest[i] = (unsigned char)(temp);
- }
-
-}
diff --git a/vpx_scale/generic/scalesystemdependent.c b/vpx_scale/generic/scalesystemdependent.c
deleted file mode 100644
index 926feb7..0000000
--- a/vpx_scale/generic/scalesystemdependent.c
+++ /dev/null
@@ -1,78 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_ports/config.h"
-#include "vpx_scale/vpxscale.h"
-
-
-void (*vp8_yv12_extend_frame_borders_ptr)(YV12_BUFFER_CONFIG *ybf);
-extern void vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf);
-
-void (*vp8_yv12_copy_frame_yonly_ptr)(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame_yonly(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-void (*vp8_yv12_copy_frame_ptr)(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-/****************************************************************************
-* Imports
-*****************************************************************************/
-
-/****************************************************************************
- *
- * ROUTINE : vp8_scale_machine_specific_config
- *
- * INPUTS : UINT32 Version : Codec version number.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Checks for machine specifc features such as MMX support
- * sets appropriate flags and function pointers.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void vp8_scale_machine_specific_config()
-{
-#if CONFIG_SPATIAL_RESAMPLING
- vp8_horizontal_line_1_2_scale = vp8cx_horizontal_line_1_2_scale_c;
- vp8_vertical_band_1_2_scale = vp8cx_vertical_band_1_2_scale_c;
- vp8_last_vertical_band_1_2_scale = vp8cx_last_vertical_band_1_2_scale_c;
- vp8_horizontal_line_3_5_scale = vp8cx_horizontal_line_3_5_scale_c;
- vp8_vertical_band_3_5_scale = vp8cx_vertical_band_3_5_scale_c;
- vp8_last_vertical_band_3_5_scale = vp8cx_last_vertical_band_3_5_scale_c;
- vp8_horizontal_line_3_4_scale = vp8cx_horizontal_line_3_4_scale_c;
- vp8_vertical_band_3_4_scale = vp8cx_vertical_band_3_4_scale_c;
- vp8_last_vertical_band_3_4_scale = vp8cx_last_vertical_band_3_4_scale_c;
- vp8_horizontal_line_2_3_scale = vp8cx_horizontal_line_2_3_scale_c;
- vp8_vertical_band_2_3_scale = vp8cx_vertical_band_2_3_scale_c;
- vp8_last_vertical_band_2_3_scale = vp8cx_last_vertical_band_2_3_scale_c;
- vp8_horizontal_line_4_5_scale = vp8cx_horizontal_line_4_5_scale_c;
- vp8_vertical_band_4_5_scale = vp8cx_vertical_band_4_5_scale_c;
- vp8_last_vertical_band_4_5_scale = vp8cx_last_vertical_band_4_5_scale_c;
-
-
- vp8_vertical_band_5_4_scale = vp8cx_vertical_band_5_4_scale_c;
- vp8_vertical_band_5_3_scale = vp8cx_vertical_band_5_3_scale_c;
- vp8_vertical_band_2_1_scale = vp8cx_vertical_band_2_1_scale_c;
- vp8_vertical_band_2_1_scale_i = vp8cx_vertical_band_2_1_scale_i_c;
- vp8_horizontal_line_2_1_scale = vp8cx_horizontal_line_2_1_scale_c;
- vp8_horizontal_line_5_3_scale = vp8cx_horizontal_line_5_3_scale_c;
- vp8_horizontal_line_5_4_scale = vp8cx_horizontal_line_5_4_scale_c;
-#endif
-
- vp8_yv12_extend_frame_borders_ptr = vp8_yv12_extend_frame_borders;
- vp8_yv12_copy_frame_yonly_ptr = vp8_yv12_copy_frame_yonly;
- vp8_yv12_copy_frame_ptr = vp8_yv12_copy_frame;
-
-}
diff --git a/vpx_scale/generic/vpxscale.c b/vpx_scale/generic/vpxscale.c
deleted file mode 100644
index 13c9122..0000000
--- a/vpx_scale/generic/vpxscale.c
+++ /dev/null
@@ -1,1089 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/****************************************************************************
- *
- * Module Title : scale.c
- *
- * Description : Image scaling functions.
- *
- ***************************************************************************/
-
-/****************************************************************************
-* Header Files
-****************************************************************************/
-#include "vpx_mem/vpx_mem.h"
-#include "vpx_scale/yv12config.h"
-#include "vpx_scale/scale_mode.h"
-
-/****************************************************************************
-* Exports
-****************************************************************************/
-#ifndef VPX_NO_GLOBALS
-void (*vp8_vertical_band_4_5_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_last_vertical_band_4_5_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_vertical_band_2_3_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_last_vertical_band_2_3_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_vertical_band_3_5_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_last_vertical_band_3_5_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_vertical_band_3_4_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_last_vertical_band_3_4_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_horizontal_line_1_2_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width) = 0;
-void (*vp8_horizontal_line_3_5_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width) = 0;
-void (*vp8_horizontal_line_3_4_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width) = 0;
-void (*vp8_horizontal_line_2_3_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width) = 0;
-void (*vp8_horizontal_line_4_5_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width) = 0;
-void (*vp8_vertical_band_1_2_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_last_vertical_band_1_2_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-
-void (*vp8_vertical_band_5_4_scale)(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_vertical_band_5_3_scale)(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_vertical_band_2_1_scale)(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_vertical_band_2_1_scale_i)(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) = 0;
-void (*vp8_horizontal_line_2_1_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width) = 0;
-void (*vp8_horizontal_line_5_3_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width) = 0;
-void (*vp8_horizontal_line_5_4_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width) = 0;
-#else
-# include "vpxscale_nofp.h"
-#endif
-
-typedef struct
-{
- int expanded_frame_width;
- int expanded_frame_height;
-
- int HScale;
- int HRatio;
- int VScale;
- int VRatio;
-
- YV12_BUFFER_CONFIG *src_yuv_config;
- YV12_BUFFER_CONFIG *dst_yuv_config;
-
-} SCALE_VARS;
-
-/****************************************************************************
- *
- * ROUTINE : horizontal_line_copy
- *
- * INPUTS : None
- *
- *
- * OUTPUTS : None.
- *
- * RETURNS : None
- *
- * FUNCTION : 1 to 1 scaling up for a horizontal line of pixles
- *
- * SPECIAL NOTES : None.
- *
- * ERRORS : None.
- *
- ****************************************************************************/
-static
-void horizontal_line_copy(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- (void) dest_width;
-
- duck_memcpy(dest, source, source_width);
-}
-/****************************************************************************
- *
- * ROUTINE : null_scale
- *
- * INPUTS : None
- *
- *
- * OUTPUTS : None.
- *
- * RETURNS : None
- *
- * FUNCTION : 1 to 1 scaling up for a vertical band
- *
- * SPECIAL NOTES : None.
- *
- * ERRORS : None.
- *
- ****************************************************************************/
-static
-void null_scale(
- unsigned char *dest,
- unsigned int dest_pitch,
- unsigned int dest_width
-)
-{
- (void) dest;
- (void) dest_pitch;
- (void) dest_width;
-
- return;
-}
-
-/****************************************************************************
- *
- * ROUTINE : scale1d_2t1_i
- *
- * INPUTS : const unsigned char *source : Pointer to data to be scaled.
- * int source_step : Number of pixels to step on in source.
- * unsigned int source_scale : Scale for source (UNUSED).
- * unsigned int source_length : Length of source (UNUSED).
- * unsigned char *dest : Pointer to output data array.
- * int dest_step : Number of pixels to step on in destination.
- * unsigned int dest_scale : Scale for destination (UNUSED).
- * unsigned int dest_length : Length of destination.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Performs 2-to-1 interpolated scaling.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-void scale1d_2t1_i
-(
- const unsigned char *source,
- int source_step,
- unsigned int source_scale,
- unsigned int source_length,
- unsigned char *dest,
- int dest_step,
- unsigned int dest_scale,
- unsigned int dest_length
-)
-{
- unsigned int i, j;
- unsigned int temp;
- int source_pitch = source_step;
- (void) source_length;
- (void) source_scale;
- (void) dest_scale;
-
- source_step *= 2;
- dest[0] = source[0];
-
- for (i = dest_step, j = source_step; i < dest_length * dest_step; i += dest_step, j += source_step)
- {
- temp = 8;
- temp += 3 * source[j-source_pitch];
- temp += 10 * source[j];
- temp += 3 * source[j+source_pitch];
- temp >>= 4;
- dest[i] = (char)(temp);
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : scale1d_2t1_ps
- *
- * INPUTS : const unsigned char *source : Pointer to data to be scaled.
- * int source_step : Number of pixels to step on in source.
- * unsigned int source_scale : Scale for source (UNUSED).
- * unsigned int source_length : Length of source (UNUSED).
- * unsigned char *dest : Pointer to output data array.
- * int dest_step : Number of pixels to step on in destination.
- * unsigned int dest_scale : Scale for destination (UNUSED).
- * unsigned int dest_length : Length of destination.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Performs 2-to-1 point subsampled scaling.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-void scale1d_2t1_ps
-(
- const unsigned char *source,
- int source_step,
- unsigned int source_scale,
- unsigned int source_length,
- unsigned char *dest,
- int dest_step,
- unsigned int dest_scale,
- unsigned int dest_length
-)
-{
- unsigned int i, j;
-
- (void) source_length;
- (void) source_scale;
- (void) dest_scale;
-
- source_step *= 2;
- j = 0;
-
- for (i = 0; i < dest_length * dest_step; i += dest_step, j += source_step)
- dest[i] = source[j];
-}
-/****************************************************************************
- *
- * ROUTINE : scale1d_c
- *
- * INPUTS : const unsigned char *source : Pointer to data to be scaled.
- * int source_step : Number of pixels to step on in source.
- * unsigned int source_scale : Scale for source.
- * unsigned int source_length : Length of source (UNUSED).
- * unsigned char *dest : Pointer to output data array.
- * int dest_step : Number of pixels to step on in destination.
- * unsigned int dest_scale : Scale for destination.
- * unsigned int dest_length : Length of destination.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Performs linear interpolation in one dimension.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-void scale1d_c
-(
- const unsigned char *source,
- int source_step,
- unsigned int source_scale,
- unsigned int source_length,
- unsigned char *dest,
- int dest_step,
- unsigned int dest_scale,
- unsigned int dest_length
-)
-{
- unsigned int i;
- unsigned int round_value = dest_scale / 2;
- unsigned int left_modifier = dest_scale;
- unsigned int right_modifier = 0;
- unsigned char left_pixel = *source;
- unsigned char right_pixel = *(source + source_step);
-
- (void) source_length;
-
- /* These asserts are needed if there are boundary issues... */
- /*assert ( dest_scale > source_scale );*/
- /*assert ( (source_length-1) * dest_scale >= (dest_length-1) * source_scale );*/
-
- for (i = 0; i < dest_length * dest_step; i += dest_step)
- {
- dest[i] = (char)((left_modifier * left_pixel + right_modifier * right_pixel + round_value) / dest_scale);
-
- right_modifier += source_scale;
-
- while (right_modifier > dest_scale)
- {
- right_modifier -= dest_scale;
- source += source_step;
- left_pixel = *source;
- right_pixel = *(source + source_step);
- }
-
- left_modifier = dest_scale - right_modifier;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : Scale2D
- *
- * INPUTS : const unsigned char *source : Pointer to data to be scaled.
- * int source_pitch : Stride of source image.
- * unsigned int source_width : Width of input image.
- * unsigned int source_height : Height of input image.
- * unsigned char *dest : Pointer to output data array.
- * int dest_pitch : Stride of destination image.
- * unsigned int dest_width : Width of destination image.
- * unsigned int dest_height : Height of destination image.
- * unsigned char *temp_area : Pointer to temp work area.
- * unsigned char temp_area_height : Height of temp work area.
- * unsigned int hscale : Horizontal scale factor numerator.
- * unsigned int hratio : Horizontal scale factor denominator.
- * unsigned int vscale : Vertical scale factor numerator.
- * unsigned int vratio : Vertical scale factor denominator.
- * unsigned int interlaced : Interlace flag.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Performs 2-tap linear interpolation in two dimensions.
- *
- * SPECIAL NOTES : Expansion is performed one band at a time to help with
- * caching.
- *
- ****************************************************************************/
-static
-void Scale2D
-(
- /*const*/
- unsigned char *source,
- int source_pitch,
- unsigned int source_width,
- unsigned int source_height,
- unsigned char *dest,
- int dest_pitch,
- unsigned int dest_width,
- unsigned int dest_height,
- unsigned char *temp_area,
- unsigned char temp_area_height,
- unsigned int hscale,
- unsigned int hratio,
- unsigned int vscale,
- unsigned int vratio,
- unsigned int interlaced
-)
-{
- /*unsigned*/
- int i, j, k;
- int bands;
- int dest_band_height;
- int source_band_height;
-
- typedef void (*Scale1D)(const unsigned char * source, int source_step, unsigned int source_scale, unsigned int source_length,
- unsigned char * dest, int dest_step, unsigned int dest_scale, unsigned int dest_length);
-
- Scale1D Scale1Dv = scale1d_c;
- Scale1D Scale1Dh = scale1d_c;
-
- void (*horiz_line_scale)(const unsigned char *, unsigned int, unsigned char *, unsigned int) = NULL;
- void (*vert_band_scale)(unsigned char *, unsigned int, unsigned char *, unsigned int, unsigned int) = NULL;
-
- int ratio_scalable = 1;
- int interpolation = 0;
-
- unsigned char *source_base; /* = (unsigned char *) ((source_pitch >= 0) ? source : (source + ((source_height-1) * source_pitch))); */
- unsigned char *line_src;
-
-
- source_base = (unsigned char *)source;
-
- if (source_pitch < 0)
- {
- int offset;
-
- offset = (source_height - 1);
- offset *= source_pitch;
-
- source_base += offset;
- }
-
- /* find out the ratio for each direction */
- switch (hratio * 10 / hscale)
- {
- case 8:
- /* 4-5 Scale in Width direction */
- horiz_line_scale = vp8_horizontal_line_5_4_scale;
- break;
- case 6:
- /* 3-5 Scale in Width direction */
- horiz_line_scale = vp8_horizontal_line_5_3_scale;
- break;
- case 5:
- /* 1-2 Scale in Width direction */
- horiz_line_scale = vp8_horizontal_line_2_1_scale;
- break;
- default:
- /* The ratio is not acceptable now */
- /* throw("The ratio is not acceptable for now!"); */
- ratio_scalable = 0;
- break;
- }
-
- switch (vratio * 10 / vscale)
- {
- case 8:
- /* 4-5 Scale in vertical direction */
- vert_band_scale = vp8_vertical_band_5_4_scale;
- source_band_height = 5;
- dest_band_height = 4;
- break;
- case 6:
- /* 3-5 Scale in vertical direction */
- vert_band_scale = vp8_vertical_band_5_3_scale;
- source_band_height = 5;
- dest_band_height = 3;
- break;
- case 5:
- /* 1-2 Scale in vertical direction */
-
- if (interlaced)
- {
- /* if the content is interlaced, point sampling is used */
- vert_band_scale = vp8_vertical_band_2_1_scale;
- }
- else
- {
-
- interpolation = 1;
- /* if the content is progressive, interplo */
- vert_band_scale = vp8_vertical_band_2_1_scale_i;
-
- }
-
- source_band_height = 2;
- dest_band_height = 1;
- break;
- default:
- /* The ratio is not acceptable now */
- /* throw("The ratio is not acceptable for now!"); */
- ratio_scalable = 0;
- break;
- }
-
- if (ratio_scalable)
- {
- if (source_height == dest_height)
- {
- /* for each band of the image */
- for (k = 0; k < (int)dest_height; k++)
- {
- horiz_line_scale(source, source_width, dest, dest_width);
- source += source_pitch;
- dest += dest_pitch;
- }
-
- return;
- }
-
- if (interpolation)
- {
- if (source < source_base)
- source = source_base;
-
- horiz_line_scale(source, source_width, temp_area, dest_width);
- }
-
- for (k = 0; k < (int)(dest_height + dest_band_height - 1) / dest_band_height; k++)
- {
- /* scale one band horizontally */
- for (i = 0; i < source_band_height; i++)
- {
- /* Trap case where we could read off the base of the source buffer */
-
- line_src = (unsigned char *)source + i * source_pitch;
-
- if (line_src < source_base)
- line_src = source_base;
-
- horiz_line_scale(line_src, source_width,
- temp_area + (i + 1)*dest_pitch, dest_width);
- }
-
- /* Vertical scaling is in place */
- vert_band_scale(temp_area + dest_pitch, dest_pitch, dest, dest_pitch, dest_width);
-
- if (interpolation)
- vpx_memcpy(temp_area, temp_area + source_band_height * dest_pitch, dest_width);
-
- /* Next band... */
- source += (unsigned long) source_band_height * source_pitch;
- dest += (unsigned long) dest_band_height * dest_pitch;
- }
-
- return;
- }
-
- if (hscale == 2 && hratio == 1)
- Scale1Dh = scale1d_2t1_ps;
-
- if (vscale == 2 && vratio == 1)
- {
- if (interlaced)
- Scale1Dv = scale1d_2t1_ps;
- else
- Scale1Dv = scale1d_2t1_i;
- }
-
- if (source_height == dest_height)
- {
- /* for each band of the image */
- for (k = 0; k < (int)dest_height; k++)
- {
- Scale1Dh(source, 1, hscale, source_width + 1, dest, 1, hratio, dest_width);
- source += source_pitch;
- dest += dest_pitch;
- }
-
- return;
- }
-
- if (dest_height > source_height)
- {
- dest_band_height = temp_area_height - 1;
- source_band_height = dest_band_height * source_height / dest_height;
- }
- else
- {
- source_band_height = temp_area_height - 1;
- dest_band_height = source_band_height * vratio / vscale;
- }
-
- /* first row needs to be done so that we can stay one row ahead for vertical zoom */
- Scale1Dh(source, 1, hscale, source_width + 1, temp_area, 1, hratio, dest_width);
-
- /* for each band of the image */
- bands = (dest_height + dest_band_height - 1) / dest_band_height;
-
- for (k = 0; k < bands; k++)
- {
- /* scale one band horizontally */
- for (i = 1; i < source_band_height + 1; i++)
- {
- if (k * source_band_height + i < (int) source_height)
- {
- Scale1Dh(source + i * source_pitch, 1, hscale, source_width + 1,
- temp_area + i * dest_pitch, 1, hratio, dest_width);
- }
- else /* Duplicate the last row */
- {
- /* copy temp_area row 0 over from last row in the past */
- duck_memcpy(temp_area + i * dest_pitch, temp_area + (i - 1)*dest_pitch, dest_pitch);
- }
- }
-
- /* scale one band vertically */
- for (j = 0; j < (int)dest_width; j++)
- {
- Scale1Dv(&temp_area[j], dest_pitch, vscale, source_band_height + 1,
- &dest[j], dest_pitch, vratio, dest_band_height);
- }
-
- /* copy temp_area row 0 over from last row in the past */
- duck_memcpy(temp_area, temp_area + source_band_height * dest_pitch, dest_pitch);
-
- /* move to the next band */
- source += source_band_height * source_pitch;
- dest += dest_band_height * dest_pitch;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE :
- *
- * INPUTS : YV12_BUFFER_CONFIG *src : Pointer to frame to be scaled.
- * YV12_BUFFER_CONFIG *dst : Pointer to buffer to hold scaled frame.
- * unsigned char *temp_area : Pointer to temp work area.
- * unsigned char temp_area_height : Height of temp work area.
- * unsigned int hscale : Horizontal scale factor numerator.
- * unsigned int hratio : Horizontal scale factor denominator.
- * unsigned int vscale : Vertical scale factor numerator.
- * unsigned int vratio : Vertical scale factor denominator.
- * unsigned int interlaced : Interlace flag.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Performs 2-tap linear interpolation in two dimensions.
- *
- * SPECIAL NOTES : Expansion is performed one band at a time to help with
- * caching.
- *
- ****************************************************************************/
-void vp8_scale_frame
-(
- YV12_BUFFER_CONFIG *src,
- YV12_BUFFER_CONFIG *dst,
- unsigned char *temp_area,
- unsigned char temp_height,
- unsigned int hscale,
- unsigned int hratio,
- unsigned int vscale,
- unsigned int vratio,
- unsigned int interlaced
-)
-{
- int i;
- int dw = (hscale - 1 + src->y_width * hratio) / hscale;
- int dh = (vscale - 1 + src->y_height * vratio) / vscale;
-
- /* call our internal scaling routines!! */
- Scale2D((unsigned char *) src->y_buffer, src->y_stride, src->y_width, src->y_height,
- (unsigned char *) dst->y_buffer, dst->y_stride, dw, dh,
- temp_area, temp_height, hscale, hratio, vscale, vratio, interlaced);
-
- if (dw < (int)dst->y_width)
- for (i = 0; i < dh; i++)
- duck_memset(dst->y_buffer + i * dst->y_stride + dw - 1, dst->y_buffer[i*dst->y_stride+dw-2], dst->y_width - dw + 1);
-
- if (dh < (int)dst->y_height)
- for (i = dh - 1; i < (int)dst->y_height; i++)
- duck_memcpy(dst->y_buffer + i * dst->y_stride, dst->y_buffer + (dh - 2) * dst->y_stride, dst->y_width + 1);
-
- Scale2D((unsigned char *) src->u_buffer, src->uv_stride, src->uv_width, src->uv_height,
- (unsigned char *) dst->u_buffer, dst->uv_stride, dw / 2, dh / 2,
- temp_area, temp_height, hscale, hratio, vscale, vratio, interlaced);
-
- if (dw / 2 < (int)dst->uv_width)
- for (i = 0; i < dst->uv_height; i++)
- duck_memset(dst->u_buffer + i * dst->uv_stride + dw / 2 - 1, dst->u_buffer[i*dst->uv_stride+dw/2-2], dst->uv_width - dw / 2 + 1);
-
- if (dh / 2 < (int)dst->uv_height)
- for (i = dh / 2 - 1; i < (int)dst->y_height / 2; i++)
- duck_memcpy(dst->u_buffer + i * dst->uv_stride, dst->u_buffer + (dh / 2 - 2)*dst->uv_stride, dst->uv_width);
-
- Scale2D((unsigned char *) src->v_buffer, src->uv_stride, src->uv_width, src->uv_height,
- (unsigned char *) dst->v_buffer, dst->uv_stride, dw / 2, dh / 2,
- temp_area, temp_height, hscale, hratio, vscale, vratio, interlaced);
-
- if (dw / 2 < (int)dst->uv_width)
- for (i = 0; i < dst->uv_height; i++)
- duck_memset(dst->v_buffer + i * dst->uv_stride + dw / 2 - 1, dst->v_buffer[i*dst->uv_stride+dw/2-2], dst->uv_width - dw / 2 + 1);
-
- if (dh / 2 < (int) dst->uv_height)
- for (i = dh / 2 - 1; i < (int)dst->y_height / 2; i++)
- duck_memcpy(dst->v_buffer + i * dst->uv_stride, dst->v_buffer + (dh / 2 - 2)*dst->uv_stride, dst->uv_width);
-}
-/****************************************************************************
- *
- * ROUTINE : any_ratio_2d_scale
- *
- * INPUTS : SCALE_INSTANCE *si : Pointer to post-processor instance (NOT USED).
- * const unsigned char *source : Pointer to source image.
- * unsigned int source_pitch : Stride of source image.
- * unsigned int source_width : Width of source image.
- * unsigned int source_height : Height of source image (NOT USED).
- * unsigned char *dest : Pointer to destination image.
- * unsigned int dest_pitch : Stride of destination image.
- * unsigned int dest_width : Width of destination image.
- * unsigned int dest_height : Height of destination image.
- *
- * OUTPUTS : None.
- *
- * RETURNS : int: 1 if image scaled, 0 if image could not be scaled.
- *
- * FUNCTION : Scale the image with changing apect ratio.
- *
- * SPECIAL NOTES : This scaling is a bi-linear scaling. Need to re-work the
- * whole function for new scaling algorithm.
- *
- ****************************************************************************/
-static
-int any_ratio_2d_scale
-(
- SCALE_VARS *si,
- const unsigned char *source,
- int source_pitch,
- unsigned int source_width,
- unsigned int source_height,
- unsigned char *dest,
- unsigned int dest_pitch,
- unsigned int dest_width,
- unsigned int dest_height
-)
-{
- unsigned int i, k;
- unsigned int src_band_height = 0;
- unsigned int dest_band_height = 0;
-
- /* suggested scale factors */
- int hs = si->HScale;
- int hr = si->HRatio;
- int vs = si->VScale;
- int vr = si->VRatio;
-
- /* assume the ratios are scalable instead of should be centered */
- int ratio_scalable = 1;
-
- const unsigned char *source_base = ((source_pitch >= 0) ? source : (source + ((source_height - 1) * source_pitch)));
- const unsigned char *line_src;
-
- void (*horiz_line_scale)(const unsigned char *, unsigned int, unsigned char *, unsigned int) = NULL;
- void (*vert_band_scale)(unsigned char *, unsigned int, unsigned int) = NULL;
- void (*last_vert_band_scale)(unsigned char *, unsigned int, unsigned int) = NULL;
-
- (void) si;
-
- /* find out the ratio for each direction */
- switch (hr * 30 / hs)
- {
- case 24:
- /* 4-5 Scale in Width direction */
- horiz_line_scale = vp8_horizontal_line_4_5_scale;
- break;
- case 22:
- /* 3-4 Scale in Width direction */
- horiz_line_scale = vp8_horizontal_line_3_4_scale;
- break;
-
- case 20:
- /* 4-5 Scale in Width direction */
- horiz_line_scale = vp8_horizontal_line_2_3_scale;
- break;
- case 18:
- /* 3-5 Scale in Width direction */
- horiz_line_scale = vp8_horizontal_line_3_5_scale;
- break;
- case 15:
- /* 1-2 Scale in Width direction */
- horiz_line_scale = vp8_horizontal_line_1_2_scale;
- break;
- case 30:
- /* no scale in Width direction */
- horiz_line_scale = horizontal_line_copy;
- break;
- default:
- /* The ratio is not acceptable now */
- /* throw("The ratio is not acceptable for now!"); */
- ratio_scalable = 0;
- break;
- }
-
- switch (vr * 30 / vs)
- {
- case 24:
- /* 4-5 Scale in vertical direction */
- vert_band_scale = vp8_vertical_band_4_5_scale;
- last_vert_band_scale = vp8_last_vertical_band_4_5_scale;
- src_band_height = 4;
- dest_band_height = 5;
- break;
- case 22:
- /* 3-4 Scale in vertical direction */
- vert_band_scale = vp8_vertical_band_3_4_scale;
- last_vert_band_scale = vp8_last_vertical_band_3_4_scale;
- src_band_height = 3;
- dest_band_height = 4;
- break;
- case 20:
- /* 2-3 Scale in vertical direction */
- vert_band_scale = vp8_vertical_band_2_3_scale;
- last_vert_band_scale = vp8_last_vertical_band_2_3_scale;
- src_band_height = 2;
- dest_band_height = 3;
- break;
- case 18:
- /* 3-5 Scale in vertical direction */
- vert_band_scale = vp8_vertical_band_3_5_scale;
- last_vert_band_scale = vp8_last_vertical_band_3_5_scale;
- src_band_height = 3;
- dest_band_height = 5;
- break;
- case 15:
- /* 1-2 Scale in vertical direction */
- vert_band_scale = vp8_vertical_band_1_2_scale;
- last_vert_band_scale = vp8_last_vertical_band_1_2_scale;
- src_band_height = 1;
- dest_band_height = 2;
- break;
- case 30:
- /* no scale in Width direction */
- vert_band_scale = null_scale;
- last_vert_band_scale = null_scale;
- src_band_height = 4;
- dest_band_height = 4;
- break;
- default:
- /* The ratio is not acceptable now */
- /* throw("The ratio is not acceptable for now!"); */
- ratio_scalable = 0;
- break;
- }
-
- if (ratio_scalable == 0)
- return ratio_scalable;
-
- horiz_line_scale(source, source_width, dest, dest_width);
-
- /* except last band */
- for (k = 0; k < (dest_height + dest_band_height - 1) / dest_band_height - 1; k++)
- {
- /* scale one band horizontally */
- for (i = 1; i < src_band_height; i++)
- {
- /* Trap case where we could read off the base of the source buffer */
- line_src = source + i * source_pitch;
-
- if (line_src < source_base)
- line_src = source_base;
-
- horiz_line_scale(line_src, source_width,
- dest + i * dest_pitch, dest_width);
- }
-
- /* first line of next band */
- /* Trap case where we could read off the base of the source buffer */
- line_src = source + src_band_height * source_pitch;
-
- if (line_src < source_base)
- line_src = source_base;
-
- horiz_line_scale(line_src, source_width,
- dest + dest_band_height * dest_pitch,
- dest_width);
-
- /* Vertical scaling is in place */
- vert_band_scale(dest, dest_pitch, dest_width);
-
- /* Next band... */
- source += src_band_height * source_pitch;
- dest += dest_band_height * dest_pitch;
- }
-
- /* scale one band horizontally */
- for (i = 1; i < src_band_height; i++)
- {
- /* Trap case where we could read off the base of the source buffer */
- line_src = source + i * source_pitch;
-
- if (line_src < source_base)
- line_src = source_base;
-
- horiz_line_scale(line_src, source_width,
- dest + i * dest_pitch,
- dest_width);
- }
-
- /* Vertical scaling is in place */
- last_vert_band_scale(dest, dest_pitch, dest_width);
-
- return ratio_scalable;
-}
-
-/****************************************************************************
- *
- * ROUTINE : any_ratio_frame_scale
- *
- * INPUTS : SCALE_INSTANCE *si : Pointer to post-processor instance (NOT USED).
- * unsigned char *frame_buffer : Pointer to source image.
- * int YOffset : Offset from start of buffer to Y samples.
- * int UVOffset : Offset from start of buffer to UV samples.
- *
- * OUTPUTS : None.
- *
- * RETURNS : int: 1 if image scaled, 0 if image could not be scaled.
- *
- * FUNCTION : Scale the image with changing apect ratio.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-int any_ratio_frame_scale(SCALE_VARS *scale_vars, int YOffset, int UVOffset)
-{
- int i;
- int ew;
- int eh;
-
- /* suggested scale factors */
- int hs = scale_vars->HScale;
- int hr = scale_vars->HRatio;
- int vs = scale_vars->VScale;
- int vr = scale_vars->VRatio;
-
- int ratio_scalable = 1;
-
- int sw = (scale_vars->expanded_frame_width * hr + hs - 1) / hs;
- int sh = (scale_vars->expanded_frame_height * vr + vs - 1) / vs;
- int dw = scale_vars->expanded_frame_width;
- int dh = scale_vars->expanded_frame_height;
- YV12_BUFFER_CONFIG *src_yuv_config = scale_vars->src_yuv_config;
- YV12_BUFFER_CONFIG *dst_yuv_config = scale_vars->dst_yuv_config;
-
- if (hr == 3)
- ew = (sw + 2) / 3 * 3 * hs / hr;
- else
- ew = (sw + 7) / 8 * 8 * hs / hr;
-
- if (vr == 3)
- eh = (sh + 2) / 3 * 3 * vs / vr;
- else
- eh = (sh + 7) / 8 * 8 * vs / vr;
-
- ratio_scalable = any_ratio_2d_scale(scale_vars,
- (const unsigned char *)src_yuv_config->y_buffer,
- src_yuv_config->y_stride, sw, sh,
- (unsigned char *) dst_yuv_config->y_buffer + YOffset,
- dst_yuv_config->y_stride, dw, dh);
-
- for (i = 0; i < eh; i++)
- duck_memset(dst_yuv_config->y_buffer + YOffset + i * dst_yuv_config->y_stride + dw, 0, ew - dw);
-
- for (i = dh; i < eh; i++)
- duck_memset(dst_yuv_config->y_buffer + YOffset + i * dst_yuv_config->y_stride, 0, ew);
-
- if (ratio_scalable == 0)
- return ratio_scalable;
-
- sw = (sw + 1) >> 1;
- sh = (sh + 1) >> 1;
- dw = (dw + 1) >> 1;
- dh = (dh + 1) >> 1;
-
- any_ratio_2d_scale(scale_vars,
- (const unsigned char *)src_yuv_config->u_buffer,
- src_yuv_config->y_stride / 2, sw, sh,
- (unsigned char *)dst_yuv_config->u_buffer + UVOffset,
- dst_yuv_config->uv_stride, dw, dh);
-
- any_ratio_2d_scale(scale_vars,
- (const unsigned char *)src_yuv_config->v_buffer,
- src_yuv_config->y_stride / 2, sw, sh,
- (unsigned char *)dst_yuv_config->v_buffer + UVOffset,
- dst_yuv_config->uv_stride, dw, dh);
-
- return ratio_scalable;
-}
-
-/****************************************************************************
- *
- * ROUTINE : center_image
- *
- * INPUTS : SCALE_INSTANCE *si : Pointer to post-processor instance.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Centers the image without scaling in the output buffer.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static void
-center_image(YV12_BUFFER_CONFIG *src_yuv_config, YV12_BUFFER_CONFIG *dst_yuv_config)
-{
- int i;
- int row_offset, col_offset;
- unsigned char *src_data_pointer;
- unsigned char *dst_data_pointer;
-
- /* center values */
- row_offset = (dst_yuv_config->y_height - src_yuv_config->y_height) / 2;
- col_offset = (dst_yuv_config->y_width - src_yuv_config->y_width) / 2;
-
- /* Y's */
- src_data_pointer = src_yuv_config->y_buffer;
- dst_data_pointer = (unsigned char *)dst_yuv_config->y_buffer + (row_offset * dst_yuv_config->y_stride) + col_offset;
-
- for (i = 0; i < src_yuv_config->y_height; i++)
- {
- duck_memcpy(dst_data_pointer, src_data_pointer, src_yuv_config->y_width);
- dst_data_pointer += dst_yuv_config->y_stride;
- src_data_pointer += src_yuv_config->y_stride;
- }
-
- row_offset /= 2;
- col_offset /= 2;
-
- /* U's */
- src_data_pointer = src_yuv_config->u_buffer;
- dst_data_pointer = (unsigned char *)dst_yuv_config->u_buffer + (row_offset * dst_yuv_config->uv_stride) + col_offset;
-
- for (i = 0; i < src_yuv_config->uv_height; i++)
- {
- duck_memcpy(dst_data_pointer, src_data_pointer, src_yuv_config->uv_width);
- dst_data_pointer += dst_yuv_config->uv_stride;
- src_data_pointer += src_yuv_config->uv_stride;
- }
-
- /* V's */
- src_data_pointer = src_yuv_config->v_buffer;
- dst_data_pointer = (unsigned char *)dst_yuv_config->v_buffer + (row_offset * dst_yuv_config->uv_stride) + col_offset;
-
- for (i = 0; i < src_yuv_config->uv_height; i++)
- {
- duck_memcpy(dst_data_pointer, src_data_pointer, src_yuv_config->uv_width);
- dst_data_pointer += dst_yuv_config->uv_stride;
- src_data_pointer += src_yuv_config->uv_stride;
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : scale_or_center
- *
- * INPUTS : SCALE_INSTANCE *si : Pointer to post-processor instance.
- *
- *
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Decides to scale or center image in scale buffer for blit
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void
-vp8_yv12_scale_or_center
-(
- YV12_BUFFER_CONFIG *src_yuv_config,
- YV12_BUFFER_CONFIG *dst_yuv_config,
- int expanded_frame_width,
- int expanded_frame_height,
- int scaling_mode,
- int HScale,
- int HRatio,
- int VScale,
- int VRatio
-)
-{
- /*if ( ppi->post_processing_level )
- update_umvborder ( ppi, frame_buffer );*/
-
-
- switch (scaling_mode)
- {
- case SCALE_TO_FIT:
- case MAINTAIN_ASPECT_RATIO:
- {
- SCALE_VARS scale_vars;
- /* center values */
-#if 1
- int row = (dst_yuv_config->y_height - expanded_frame_height) / 2;
- int col = (dst_yuv_config->y_width - expanded_frame_width) / 2;
- /*int YOffset = row * dst_yuv_config->y_width + col;
- int UVOffset = (row>>1) * dst_yuv_config->uv_width + (col>>1);*/
- int YOffset = row * dst_yuv_config->y_stride + col;
- int UVOffset = (row >> 1) * dst_yuv_config->uv_stride + (col >> 1);
-#else
- int row = (src_yuv_config->y_height - expanded_frame_height) / 2;
- int col = (src_yuv_config->y_width - expanded_frame_width) / 2;
- int YOffset = row * src_yuv_config->y_width + col;
- int UVOffset = (row >> 1) * src_yuv_config->uv_width + (col >> 1);
-#endif
-
- scale_vars.dst_yuv_config = dst_yuv_config;
- scale_vars.src_yuv_config = src_yuv_config;
- scale_vars.HScale = HScale;
- scale_vars.HRatio = HRatio;
- scale_vars.VScale = VScale;
- scale_vars.VRatio = VRatio;
- scale_vars.expanded_frame_width = expanded_frame_width;
- scale_vars.expanded_frame_height = expanded_frame_height;
-
- /* perform center and scale */
- any_ratio_frame_scale(&scale_vars, YOffset, UVOffset);
-
- break;
- }
- case CENTER:
- center_image(src_yuv_config, dst_yuv_config);
- break;
-
- default:
- break;
- }
-}
diff --git a/vpx_scale/generic/yv12config.c b/vpx_scale/generic/yv12config.c
deleted file mode 100644
index cb0ab94..0000000
--- a/vpx_scale/generic/yv12config.c
+++ /dev/null
@@ -1,90 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_scale/yv12config.h"
-#include "vpx_mem/vpx_mem.h"
-
-/****************************************************************************
-* Exports
-****************************************************************************/
-
-/****************************************************************************
- *
- ****************************************************************************/
-int
-vp8_yv12_de_alloc_frame_buffer(YV12_BUFFER_CONFIG *ybf)
-{
- if (ybf)
- {
- duck_free(ybf->buffer_alloc);
-
- ybf->buffer_alloc = 0;
- }
- else
- {
- return -1;
- }
-
- return 0;
-}
-
-/****************************************************************************
- *
- ****************************************************************************/
-int
-vp8_yv12_alloc_frame_buffer(YV12_BUFFER_CONFIG *ybf, int width, int height, int border)
-{
-/*NOTE:*/
-
- int yplane_size = (height + 2 * border) * (width + 2 * border);
- int uvplane_size = ((1 + height) / 2 + border) * ((1 + width) / 2 + border);
-
- if (ybf)
- {
- vp8_yv12_de_alloc_frame_buffer(ybf);
-
- ybf->y_width = width;
- ybf->y_height = height;
- ybf->y_stride = width + 2 * border;
-
- ybf->uv_width = (1 + width) / 2;
- ybf->uv_height = (1 + height) / 2;
- ybf->uv_stride = ybf->uv_width + border;
-
- ybf->border = border;
- ybf->frame_size = yplane_size + 2 * uvplane_size;
-
- /* Added 2 extra lines to framebuffer so that copy12x12 doesn't fail
- * when we have a large motion vector in V on the last v block.
- * Note : We never use these pixels anyway so this doesn't hurt.
- */
- ybf->buffer_alloc = (unsigned char *) duck_memalign(32, ybf->frame_size + (ybf->y_stride * 2) + 32, 0);
-
- if (ybf->buffer_alloc == NULL)
- return -1;
-
- ybf->y_buffer = ybf->buffer_alloc + (border * ybf->y_stride) + border;
-
- if (yplane_size & 0xf)
- yplane_size += 16 - (yplane_size & 0xf);
-
- ybf->u_buffer = ybf->buffer_alloc + yplane_size + (border / 2 * ybf->uv_stride) + border / 2;
- ybf->v_buffer = ybf->buffer_alloc + yplane_size + uvplane_size + (border / 2 * ybf->uv_stride) + border / 2;
-
- ybf->corrupted = 0; /* assume not currupted by errors */
- }
- else
- {
- return -2;
- }
-
- return 0;
-}
diff --git a/vpx_scale/generic/yv12extend.c b/vpx_scale/generic/yv12extend.c
deleted file mode 100644
index c087bdd..0000000
--- a/vpx_scale/generic/yv12extend.c
+++ /dev/null
@@ -1,280 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "vpx_scale/yv12config.h"
-#include "vpx_mem/vpx_mem.h"
-#include "vpx_scale/vpxscale.h"
-
-/****************************************************************************
-* Exports
-****************************************************************************/
-
-/****************************************************************************
- *
- ****************************************************************************/
-void
-vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf)
-{
- int i;
- unsigned char *src_ptr1, *src_ptr2;
- unsigned char *dest_ptr1, *dest_ptr2;
-
- unsigned int Border;
- int plane_stride;
- int plane_height;
- int plane_width;
-
- /***********/
- /* Y Plane */
- /***********/
- Border = ybf->border;
- plane_stride = ybf->y_stride;
- plane_height = ybf->y_height;
- plane_width = ybf->y_width;
-
- /* copy the left and right most columns out */
- src_ptr1 = ybf->y_buffer;
- src_ptr2 = src_ptr1 + plane_width - 1;
- dest_ptr1 = src_ptr1 - Border;
- dest_ptr2 = src_ptr2 + 1;
-
- for (i = 0; i < plane_height; i++)
- {
- vpx_memset(dest_ptr1, src_ptr1[0], Border);
- vpx_memset(dest_ptr2, src_ptr2[0], Border);
- src_ptr1 += plane_stride;
- src_ptr2 += plane_stride;
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- /* Now copy the top and bottom source lines into each line of the respective borders */
- src_ptr1 = ybf->y_buffer - Border;
- src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
- dest_ptr1 = src_ptr1 - (Border * plane_stride);
- dest_ptr2 = src_ptr2 + plane_stride;
-
- for (i = 0; i < (int)Border; i++)
- {
- vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
- vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
-
- /***********/
- /* U Plane */
- /***********/
- plane_stride = ybf->uv_stride;
- plane_height = ybf->uv_height;
- plane_width = ybf->uv_width;
- Border /= 2;
-
- /* copy the left and right most columns out */
- src_ptr1 = ybf->u_buffer;
- src_ptr2 = src_ptr1 + plane_width - 1;
- dest_ptr1 = src_ptr1 - Border;
- dest_ptr2 = src_ptr2 + 1;
-
- for (i = 0; i < plane_height; i++)
- {
- vpx_memset(dest_ptr1, src_ptr1[0], Border);
- vpx_memset(dest_ptr2, src_ptr2[0], Border);
- src_ptr1 += plane_stride;
- src_ptr2 += plane_stride;
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- /* Now copy the top and bottom source lines into each line of the respective borders */
- src_ptr1 = ybf->u_buffer - Border;
- src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
- dest_ptr1 = src_ptr1 - (Border * plane_stride);
- dest_ptr2 = src_ptr2 + plane_stride;
-
- for (i = 0; i < (int)(Border); i++)
- {
- vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
- vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- /***********/
- /* V Plane */
- /***********/
-
- /* copy the left and right most columns out */
- src_ptr1 = ybf->v_buffer;
- src_ptr2 = src_ptr1 + plane_width - 1;
- dest_ptr1 = src_ptr1 - Border;
- dest_ptr2 = src_ptr2 + 1;
-
- for (i = 0; i < plane_height; i++)
- {
- vpx_memset(dest_ptr1, src_ptr1[0], Border);
- vpx_memset(dest_ptr2, src_ptr2[0], Border);
- src_ptr1 += plane_stride;
- src_ptr2 += plane_stride;
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- /* Now copy the top and bottom source lines into each line of the respective borders */
- src_ptr1 = ybf->v_buffer - Border;
- src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
- dest_ptr1 = src_ptr1 - (Border * plane_stride);
- dest_ptr2 = src_ptr2 + plane_stride;
-
- for (i = 0; i < (int)(Border); i++)
- {
- vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
- vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-}
-
-
-static void
-extend_frame_borders_yonly(YV12_BUFFER_CONFIG *ybf)
-{
- int i;
- unsigned char *src_ptr1, *src_ptr2;
- unsigned char *dest_ptr1, *dest_ptr2;
-
- unsigned int Border;
- int plane_stride;
- int plane_height;
- int plane_width;
-
- /***********/
- /* Y Plane */
- /***********/
- Border = ybf->border;
- plane_stride = ybf->y_stride;
- plane_height = ybf->y_height;
- plane_width = ybf->y_width;
-
- /* copy the left and right most columns out */
- src_ptr1 = ybf->y_buffer;
- src_ptr2 = src_ptr1 + plane_width - 1;
- dest_ptr1 = src_ptr1 - Border;
- dest_ptr2 = src_ptr2 + 1;
-
- for (i = 0; i < plane_height; i++)
- {
- vpx_memset(dest_ptr1, src_ptr1[0], Border);
- vpx_memset(dest_ptr2, src_ptr2[0], Border);
- src_ptr1 += plane_stride;
- src_ptr2 += plane_stride;
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- /* Now copy the top and bottom source lines into each line of the respective borders */
- src_ptr1 = ybf->y_buffer - Border;
- src_ptr2 = src_ptr1 + (plane_height * plane_stride) - plane_stride;
- dest_ptr1 = src_ptr1 - (Border * plane_stride);
- dest_ptr2 = src_ptr2 + plane_stride;
-
- for (i = 0; i < (int)Border; i++)
- {
- vpx_memcpy(dest_ptr1, src_ptr1, plane_stride);
- vpx_memcpy(dest_ptr2, src_ptr2, plane_stride);
- dest_ptr1 += plane_stride;
- dest_ptr2 += plane_stride;
- }
-
- plane_stride /= 2;
- plane_height /= 2;
- plane_width /= 2;
- Border /= 2;
-
-}
-
-
-
-/****************************************************************************
- *
- * ROUTINE : vp8_yv12_copy_frame
- *
- * INPUTS :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies the source image into the destination image and
- * updates the destination's UMV borders.
- *
- * SPECIAL NOTES : The frames are assumed to be identical in size.
- *
- ****************************************************************************/
-void
-vp8_yv12_copy_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc)
-{
- int row;
- unsigned char *source, *dest;
-
- source = src_ybc->y_buffer;
- dest = dst_ybc->y_buffer;
-
- for (row = 0; row < src_ybc->y_height; row++)
- {
- vpx_memcpy(dest, source, src_ybc->y_width);
- source += src_ybc->y_stride;
- dest += dst_ybc->y_stride;
- }
-
- source = src_ybc->u_buffer;
- dest = dst_ybc->u_buffer;
-
- for (row = 0; row < src_ybc->uv_height; row++)
- {
- vpx_memcpy(dest, source, src_ybc->uv_width);
- source += src_ybc->uv_stride;
- dest += dst_ybc->uv_stride;
- }
-
- source = src_ybc->v_buffer;
- dest = dst_ybc->v_buffer;
-
- for (row = 0; row < src_ybc->uv_height; row++)
- {
- vpx_memcpy(dest, source, src_ybc->uv_width);
- source += src_ybc->uv_stride;
- dest += dst_ybc->uv_stride;
- }
-
- vp8_yv12_extend_frame_borders_ptr(dst_ybc);
-}
-
-void
-vp8_yv12_copy_frame_yonly(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc)
-{
- int row;
- unsigned char *source, *dest;
-
-
- source = src_ybc->y_buffer;
- dest = dst_ybc->y_buffer;
-
- for (row = 0; row < src_ybc->y_height; row++)
- {
- vpx_memcpy(dest, source, src_ybc->y_width);
- source += src_ybc->y_stride;
- dest += dst_ybc->y_stride;
- }
-
- extend_frame_borders_yonly(dst_ybc);
-}
diff --git a/vpx_scale/include/arm/vpxscale_nofp.h b/vpx_scale/include/arm/vpxscale_nofp.h
deleted file mode 100644
index 3e1a9fa..0000000
--- a/vpx_scale/include/arm/vpxscale_nofp.h
+++ /dev/null
@@ -1,68 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-void vp8cx_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_3_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-
-void vp8cx_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-
-void horizontal_line_4_5_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_2_3_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_3_5_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_3_4_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_1_2_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vertical_band_4_5_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_2_3_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_3_5_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_3_4_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_1_2_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-
-#define vp8_vertical_band_4_5_scale vertical_band_4_5_scale_armv4
-#define vp8_last_vertical_band_4_5_scale vp8cx_last_vertical_band_4_5_scale_c
-#define vp8_vertical_band_2_3_scale vertical_band_2_3_scale_armv4
-#define vp8_last_vertical_band_2_3_scale vp8cx_last_vertical_band_2_3_scale_c
-#define vp8_vertical_band_3_5_scale vertical_band_3_5_scale_armv4
-#define vp8_last_vertical_band_3_5_scale vp8cx_last_vertical_band_3_5_scale_c
-#define vp8_vertical_band_3_4_scale vertical_band_3_4_scale_armv4
-#define vp8_last_vertical_band_3_4_scale vp8cx_last_vertical_band_3_4_scale_c
-#define vp8_horizontal_line_1_2_scale horizontal_line_1_2_scale_armv4
-#define vp8_horizontal_line_3_5_scale horizontal_line_3_5_scale_armv4
-#define vp8_horizontal_line_3_4_scale horizontal_line_3_4_scale_armv4
-#define vp8_horizontal_line_4_5_scale horizontal_line_4_5_scale_armv4
-#define vp8_horizontal_line_2_3_scale horizontal_line_2_3_scale_armv4
-#define vp8_vertical_band_1_2_scale vertical_band_1_2_scale_armv4
-#define vp8_last_vertical_band_1_2_scale vp8cx_last_vertical_band_1_2_scale_c
-#define vp8_vertical_band_5_4_scale vp8cx_vertical_band_5_4_scale_c
-#define vp8_vertical_band_5_3_scale vp8cx_vertical_band_5_3_scale_c
-#define vp8_vertical_band_2_1_scale vp8cx_vertical_band_2_1_scale_c
-#define vp8_vertical_band_2_1_scale_i vp8cx_vertical_band_2_1_scale_i_c
-#define vp8_horizontal_line_2_1_scale vp8cx_horizontal_line_2_1_scale_c
-#define vp8_horizontal_line_5_3_scale vp8cx_horizontal_line_5_3_scale_c
-#define vp8_horizontal_line_5_4_scale vp8cx_horizontal_line_5_4_scale_c
diff --git a/vpx_scale/include/generic/vpxscale_arbitrary.h b/vpx_scale/include/generic/vpxscale_arbitrary.h
deleted file mode 100644
index 39de181..0000000
--- a/vpx_scale/include/generic/vpxscale_arbitrary.h
+++ /dev/null
@@ -1,56 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef __VPX_SCALE_ARBITRARY_H__
-#define __VPX_SCALE_ARBITRARY_H__
-
-#include "vpx_scale/yv12config.h"
-
-typedef struct
-{
- int in_width;
- int in_height;
-
- int out_width;
- int out_height;
- int max_usable_out_width;
-
- // numerator for the width and height
- int nw;
- int nh;
- int nh_uv;
-
- // output to input correspondance array
- short *l_w;
- short *l_h;
- short *l_h_uv;
-
- // polyphase coefficients
- short *c_w;
- short *c_h;
- short *c_h_uv;
-
- // buffer for horizontal filtering.
- unsigned char *hbuf;
- unsigned char *hbuf_uv;
-} BICUBIC_SCALER_STRUCT;
-
-int bicubic_coefficient_setup(int in_width, int in_height, int out_width, int out_height);
-int bicubic_scale(int in_width, int in_height, int in_stride,
- int out_width, int out_height, int out_stride,
- unsigned char *input_image, unsigned char *output_image);
-void bicubic_scale_frame_reset();
-void bicubic_scale_frame(YV12_BUFFER_CONFIG *src, YV12_BUFFER_CONFIG *dst,
- int new_width, int new_height);
-void bicubic_coefficient_init();
-void bicubic_coefficient_destroy();
-
-#endif /* __VPX_SCALE_ARBITRARY_H__ */
diff --git a/vpx_scale/include/generic/vpxscale_nofp.h b/vpx_scale/include/generic/vpxscale_nofp.h
deleted file mode 100644
index 7b8205a..0000000
--- a/vpx_scale/include/generic/vpxscale_nofp.h
+++ /dev/null
@@ -1,51 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-void vp8cx_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-
-void vp8cx_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-
-#define vp8_vertical_band_4_5_scale vp8cx_vertical_band_4_5_scale_c
-#define vp8_last_vertical_band_4_5_scale vp8cx_last_vertical_band_4_5_scale_c
-#define vp8_vertical_band_2_3_scale vp8cx_vertical_band_2_3_scale_c
-#define vp8_last_vertical_band_2_3_scale vp8cx_last_vertical_band_2_3_scale_c
-#define vp8_vertical_band_3_5_scale vp8cx_vertical_band_3_5_scale_c
-#define vp8_last_vertical_band_3_5_scale vp8cx_last_vertical_band_3_5_scale_c
-#define vp8_horizontal_line_1_2_scale vp8cx_horizontal_line_1_2_scale_c
-#define vp8_horizontal_line_3_5_scale vp8cx_horizontal_line_3_5_scale_c
-#define vp8_horizontal_line_4_5_scale vp8cx_horizontal_line_4_5_scale_c
-#define vp8_horizontal_line_2_3_scale vp8cx_horizontal_line_2_3_scale_c
-#define vp8_vertical_band_1_2_scale vp8cx_vertical_band_1_2_scale_c
-#define vp8_last_vertical_band_1_2_scale vp8cx_last_vertical_band_1_2_scale_c
-#define vp8_vertical_band_5_4_scale vp8cx_vertical_band_5_4_scale_c
-#define vp8_vertical_band_5_3_scale vp8cx_vertical_band_5_3_scale_c
-#define vp8_vertical_band_2_1_scale vp8cx_vertical_band_2_1_scale_c
-#define vp8_vertical_band_2_1_scale_i vp8cx_vertical_band_2_1_scale_i_c
-#define vp8_horizontal_line_2_1_scale vp8cx_horizontal_line_2_1_scale_c
-#define vp8_horizontal_line_5_3_scale vp8cx_horizontal_line_5_3_scale_c
-#define vp8_horizontal_line_5_4_scale vp8cx_horizontal_line_5_4_scale_c
diff --git a/vpx_scale/include/symbian/vpxscale_nofp.h b/vpx_scale/include/symbian/vpxscale_nofp.h
deleted file mode 100644
index 3e1a9fa..0000000
--- a/vpx_scale/include/symbian/vpxscale_nofp.h
+++ /dev/null
@@ -1,68 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-void vp8cx_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_3_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-
-void vp8cx_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-
-void horizontal_line_4_5_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_2_3_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_3_5_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_3_4_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_1_2_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vertical_band_4_5_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_2_3_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_3_5_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_3_4_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_1_2_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-
-#define vp8_vertical_band_4_5_scale vertical_band_4_5_scale_armv4
-#define vp8_last_vertical_band_4_5_scale vp8cx_last_vertical_band_4_5_scale_c
-#define vp8_vertical_band_2_3_scale vertical_band_2_3_scale_armv4
-#define vp8_last_vertical_band_2_3_scale vp8cx_last_vertical_band_2_3_scale_c
-#define vp8_vertical_band_3_5_scale vertical_band_3_5_scale_armv4
-#define vp8_last_vertical_band_3_5_scale vp8cx_last_vertical_band_3_5_scale_c
-#define vp8_vertical_band_3_4_scale vertical_band_3_4_scale_armv4
-#define vp8_last_vertical_band_3_4_scale vp8cx_last_vertical_band_3_4_scale_c
-#define vp8_horizontal_line_1_2_scale horizontal_line_1_2_scale_armv4
-#define vp8_horizontal_line_3_5_scale horizontal_line_3_5_scale_armv4
-#define vp8_horizontal_line_3_4_scale horizontal_line_3_4_scale_armv4
-#define vp8_horizontal_line_4_5_scale horizontal_line_4_5_scale_armv4
-#define vp8_horizontal_line_2_3_scale horizontal_line_2_3_scale_armv4
-#define vp8_vertical_band_1_2_scale vertical_band_1_2_scale_armv4
-#define vp8_last_vertical_band_1_2_scale vp8cx_last_vertical_band_1_2_scale_c
-#define vp8_vertical_band_5_4_scale vp8cx_vertical_band_5_4_scale_c
-#define vp8_vertical_band_5_3_scale vp8cx_vertical_band_5_3_scale_c
-#define vp8_vertical_band_2_1_scale vp8cx_vertical_band_2_1_scale_c
-#define vp8_vertical_band_2_1_scale_i vp8cx_vertical_band_2_1_scale_i_c
-#define vp8_horizontal_line_2_1_scale vp8cx_horizontal_line_2_1_scale_c
-#define vp8_horizontal_line_5_3_scale vp8cx_horizontal_line_5_3_scale_c
-#define vp8_horizontal_line_5_4_scale vp8cx_horizontal_line_5_4_scale_c
diff --git a/vpx_scale/include/vpxscale_nofp.h b/vpx_scale/include/vpxscale_nofp.h
deleted file mode 100644
index a704bd9..0000000
--- a/vpx_scale/include/vpxscale_nofp.h
+++ /dev/null
@@ -1,16 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#if defined(__S60_V20__) || defined(__SYMBIAN32__) && !defined(__WINS__)
-#include "symbian\vpxscale_nofp.h"
-#else
-#include "generic\vpxscale_nofp.h"
-#endif
diff --git a/vpx_scale/scale_mode.h b/vpx_scale/scale_mode.h
deleted file mode 100644
index 1476e64..0000000
--- a/vpx_scale/scale_mode.h
+++ /dev/null
@@ -1,29 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/****************************************************************************
-*
-*****************************************************************************
-*/
-
-#ifndef SCALE_MODE_H
-#define SCALE_MODE_H
-
-typedef enum
-{
- MAINTAIN_ASPECT_RATIO = 0x0,
- SCALE_TO_FIT = 0x1,
- CENTER = 0x2,
- OTHER = 0x3
-} SCALE_MODE;
-
-
-#endif
diff --git a/vpx_scale/vpx_scale.mk b/vpx_scale/vpx_scale.mk
deleted file mode 100644
index edb5419..0000000
--- a/vpx_scale/vpx_scale.mk
+++ /dev/null
@@ -1,23 +0,0 @@
-SCALE_SRCS-yes += vpx_scale.mk
-SCALE_SRCS-yes += scale_mode.h
-SCALE_SRCS-yes += yv12extend.h
-SCALE_SRCS-yes += yv12config.h
-SCALE_SRCS-yes += vpxscale.h
-SCALE_SRCS-yes += generic/vpxscale.c
-SCALE_SRCS-yes += generic/yv12config.c
-SCALE_SRCS-yes += generic/yv12extend.c
-SCALE_SRCS-yes += generic/scalesystemdependent.c
-SCALE_SRCS-$(CONFIG_SPATIAL_RESAMPLING) += generic/gen_scalers.c
-
-#arm
-SCALE_SRCS-$(HAVE_ARMV7) += arm/scalesystemdependent.c
-SCALE_SRCS-$(HAVE_ARMV7) += arm/yv12extend_arm.c
-SCALE_SRCS_REMOVE-$(HAVE_ARMV7) += generic/scalesystemdependent.c
-
-#neon
-SCALE_SRCS-$(HAVE_ARMV7) += arm/neon/vp8_vpxyv12_copyframe_func_neon$(ASM)
-SCALE_SRCS-$(HAVE_ARMV7) += arm/neon/vp8_vpxyv12_copyframeyonly_neon$(ASM)
-SCALE_SRCS-$(HAVE_ARMV7) += arm/neon/vp8_vpxyv12_copysrcframe_func_neon$(ASM)
-SCALE_SRCS-$(HAVE_ARMV7) += arm/neon/vp8_vpxyv12_extendframeborders_neon$(ASM)
-
-SCALE_SRCS-no += $(SCALE_SRCS_REMOVE-yes)
diff --git a/vpx_scale/vpxscale.h b/vpx_scale/vpxscale.h
deleted file mode 100644
index 86fc128..0000000
--- a/vpx_scale/vpxscale.h
+++ /dev/null
@@ -1,114 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef VPXSCALE_H
-#define VPXSCALE_H
-
-#include "vpx_scale/yv12config.h"
-void vp8cx_horizontal_line_4_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_4_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_2_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_2_3_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_3_5_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_3_5_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_3_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_3_4_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_1_2_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_last_vertical_band_1_2_scale_c(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vp8cx_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vp8cx_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-
-
-extern void (*vp8_vertical_band_4_5_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_last_vertical_band_4_5_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_vertical_band_2_3_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_last_vertical_band_2_3_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_vertical_band_3_5_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_last_vertical_band_3_5_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_vertical_band_3_4_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_last_vertical_band_3_4_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_horizontal_line_1_2_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-extern void (*vp8_horizontal_line_3_4_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-extern void (*vp8_horizontal_line_3_5_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-extern void (*vp8_horizontal_line_2_3_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-extern void (*vp8_horizontal_line_4_5_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-extern void (*vp8_vertical_band_1_2_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_last_vertical_band_1_2_scale)(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_vertical_band_5_4_scale)(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_vertical_band_5_3_scale)(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_vertical_band_2_1_scale)(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_vertical_band_2_1_scale_i)(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-extern void (*vp8_horizontal_line_2_1_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-extern void (*vp8_horizontal_line_5_3_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-extern void (*vp8_horizontal_line_5_4_scale)(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-
-void horizontal_line_4_5_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_2_3_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_3_5_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_3_4_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void horizontal_line_1_2_scale_armv4(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width);
-void vertical_band_4_5_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_2_3_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_3_5_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_3_4_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-void vertical_band_1_2_scale_armv4(unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width);
-
-
-extern void dmachine_specific_config(int mmx_enabled, int xmm_enabled, int wmt_enabled);
-extern void vp8_yv12_scale_or_center
-(
- YV12_BUFFER_CONFIG *src_yuv_config,
- YV12_BUFFER_CONFIG *dst_yuv_config,
- int expanded_frame_width,
- int expanded_frame_height,
- int scaling_mode,
- int HScale,
- int HRatio,
- int VScale,
- int VRatio
-);
-extern void vp8_scale_frame
-(
- YV12_BUFFER_CONFIG *src,
- YV12_BUFFER_CONFIG *dst,
- unsigned char *temp_area,
- unsigned char temp_height,
- unsigned int hscale,
- unsigned int hratio,
- unsigned int vscale,
- unsigned int vratio,
- unsigned int interlaced
-);
-extern void vp8_scale_machine_specific_config(void);
-
-extern void (*vp8_yv12_extend_frame_borders_ptr)(YV12_BUFFER_CONFIG *ybf);
-extern void vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf);
-extern void vp8_yv12_extend_frame_borders_neon(YV12_BUFFER_CONFIG *ybf);
-
-extern void (*vp8_yv12_copy_frame_yonly_ptr)(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame_yonly(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame_yonly_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-extern void (*vp8_yv12_copy_frame_ptr)(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-extern void vp8_yv12_copy_frame_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-#endif
diff --git a/vpx_scale/win32/scaleopt.c b/vpx_scale/win32/scaleopt.c
deleted file mode 100644
index 3711fe5..0000000
--- a/vpx_scale/win32/scaleopt.c
+++ /dev/null
@@ -1,1750 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/****************************************************************************
-*
-* Module Title : scaleopt.cpp
-*
-* Description : Optimized scaling functions
-*
-****************************************************************************/
-#include "pragmas.h"
-
-
-
-/****************************************************************************
-* Module Statics
-****************************************************************************/
-__declspec(align(16)) const static unsigned short one_fifth[] = { 51, 51, 51, 51 };
-__declspec(align(16)) const static unsigned short two_fifths[] = { 102, 102, 102, 102 };
-__declspec(align(16)) const static unsigned short three_fifths[] = { 154, 154, 154, 154 };
-__declspec(align(16)) const static unsigned short four_fifths[] = { 205, 205, 205, 205 };
-__declspec(align(16)) const static unsigned short round_values[] = { 128, 128, 128, 128 };
-__declspec(align(16)) const static unsigned short four_ones[] = { 1, 1, 1, 1};
-__declspec(align(16)) const static unsigned short const45_2[] = {205, 154, 102, 51 };
-__declspec(align(16)) const static unsigned short const45_1[] = { 51, 102, 154, 205 };
-__declspec(align(16)) const static unsigned char mask45[] = { 0, 0, 0, 0, 0, 0, 255, 0};
-__declspec(align(16)) const static unsigned short const35_2[] = { 154, 51, 205, 102 };
-__declspec(align(16)) const static unsigned short const35_1[] = { 102, 205, 51, 154 };
-
-
-
-#include "vpx_scale/vpxscale.h"
-#include "vpx_mem/vpx_mem.h"
-
-/****************************************************************************
- *
- * ROUTINE : horizontal_line_3_5_scale_mmx
- *
- * INPUTS : const unsigned char *source :
- * unsigned int source_width :
- * unsigned char *dest :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 3 to 5 up-scaling of a horizontal line of pixels.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-void horizontal_line_3_5_scale_mmx
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- (void) dest_width;
-
- __asm
- {
-
- push ebx
-
- mov esi, source
- mov edi, dest
-
- mov ecx, source_width
- lea edx, [esi+ecx-3];
-
- movq mm5, const35_1 // mm5 = 66 xx cd xx 33 xx 9a xx
- movq mm6, const35_2 // mm6 = 9a xx 33 xx cd xx 66 xx
-
- movq mm4, round_values // mm4 = 80 xx 80 xx 80 xx 80 xx
- pxor mm7, mm7 // clear mm7
-
- horiz_line_3_5_loop:
-
- mov eax, DWORD PTR [esi] // eax = 00 01 02 03
- mov ebx, eax
-
- and ebx, 0xffff00 // ebx = xx 01 02 xx
- mov ecx, eax // ecx = 00 01 02 03
-
- and eax, 0xffff0000 // eax = xx xx 02 03
- xor ecx, eax // ecx = 00 01 xx xx
-
- shr ebx, 8 // ebx = 01 02 xx xx
- or eax, ebx // eax = 01 02 02 03
-
- shl ebx, 16 // ebx = xx xx 01 02
- movd mm1, eax // mm1 = 01 02 02 03 xx xx xx xx
-
- or ebx, ecx // ebx = 00 01 01 02
- punpcklbw mm1, mm7 // mm1 = 01 xx 02 xx 02 xx 03 xx
-
- movd mm0, ebx // mm0 = 00 01 01 02
- pmullw mm1, mm6 //
-
- punpcklbw mm0, mm7 // mm0 = 00 xx 01 xx 01 xx 02 xx
- pmullw mm0, mm5 //
-
- mov [edi], ebx // writeoutput 00 xx xx xx
- add esi, 3
-
- add edi, 5
- paddw mm0, mm1
-
- paddw mm0, mm4
- psrlw mm0, 8
-
- cmp esi, edx
- packuswb mm0, mm7
-
- movd DWORD Ptr [edi-4], mm0
- jl horiz_line_3_5_loop
-
-//Exit:
- mov eax, DWORD PTR [esi] // eax = 00 01 02 03
- mov ebx, eax
-
- and ebx, 0xffff00 // ebx = xx 01 02 xx
- mov ecx, eax // ecx = 00 01 02 03
-
- and eax, 0xffff0000 // eax = xx xx 02 03
- xor ecx, eax // ecx = 00 01 xx xx
-
- shr ebx, 8 // ebx = 01 02 xx xx
- or eax, ebx // eax = 01 02 02 03
-
- shl eax, 8 // eax = xx 01 02 02
- and eax, 0xffff0000 // eax = xx xx 02 02
-
- or eax, ebx // eax = 01 02 02 02
-
- shl ebx, 16 // ebx = xx xx 01 02
- movd mm1, eax // mm1 = 01 02 02 02 xx xx xx xx
-
- or ebx, ecx // ebx = 00 01 01 02
- punpcklbw mm1, mm7 // mm1 = 01 xx 02 xx 02 xx 02 xx
-
- movd mm0, ebx // mm0 = 00 01 01 02
- pmullw mm1, mm6 //
-
- punpcklbw mm0, mm7 // mm0 = 00 xx 01 xx 01 xx 02 xx
- pmullw mm0, mm5 //
-
- mov [edi], ebx // writeoutput 00 xx xx xx
- paddw mm0, mm1
-
- paddw mm0, mm4
- psrlw mm0, 8
-
- packuswb mm0, mm7
- movd DWORD Ptr [edi+1], mm0
-
- pop ebx
-
- }
-
-}
-
-
-/****************************************************************************
- *
- * ROUTINE : horizontal_line_4_5_scale_mmx
- *
- * INPUTS : const unsigned char *source :
- * unsigned int source_width :
- * unsigned char *dest :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 4 to 5 up-scaling of a horizontal line of pixels.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-void horizontal_line_4_5_scale_mmx
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- (void)dest_width;
-
- __asm
- {
-
- mov esi, source
- mov edi, dest
-
- mov ecx, source_width
- lea edx, [esi+ecx-8];
-
- movq mm5, const45_1 // mm5 = 33 xx 66 xx 9a xx cd xx
- movq mm6, const45_2 // mm6 = cd xx 9a xx 66 xx 33 xx
-
- movq mm4, round_values // mm4 = 80 xx 80 xx 80 xx 80 xx
- pxor mm7, mm7 // clear mm7
-
- horiz_line_4_5_loop:
-
- movq mm0, QWORD PTR [esi] // mm0 = 00 01 02 03 04 05 06 07
- movq mm1, QWORD PTR [esi+1]; // mm1 = 01 02 03 04 05 06 07 08
-
- movq mm2, mm0 // mm2 = 00 01 02 03 04 05 06 07
- movq mm3, mm1 // mm3 = 01 02 03 04 05 06 07 08
-
- movd DWORD PTR [edi], mm0 // write output 00 xx xx xx
- punpcklbw mm0, mm7 // mm0 = 00 xx 01 xx 02 xx 03 xx
-
- punpcklbw mm1, mm7 // mm1 = 01 xx 02 xx 03 xx 04 xx
- pmullw mm0, mm5 // 00* 51 01*102 02*154 03*205
-
- pmullw mm1, mm6 // 01*205 02*154 03*102 04* 51
- punpckhbw mm2, mm7 // mm2 = 04 xx 05 xx 06 xx 07 xx
-
- movd DWORD PTR [edi+5], mm2 // write ouput 05 xx xx xx
- pmullw mm2, mm5 // 04* 51 05*102 06*154 07*205
-
- punpckhbw mm3, mm7 // mm3 = 05 xx 06 xx 07 xx 08 xx
- pmullw mm3, mm6 // 05*205 06*154 07*102 08* 51
-
- paddw mm0, mm1 // added round values
- paddw mm0, mm4
-
- psrlw mm0, 8 // output: 01 xx 02 xx 03 xx 04 xx
- packuswb mm0, mm7
-
- movd DWORD PTR [edi+1], mm0 // write output 01 02 03 04
- add edi, 10
-
- add esi, 8
- paddw mm2, mm3 //
-
- paddw mm2, mm4 // added round values
- cmp esi, edx
-
- psrlw mm2, 8
- packuswb mm2, mm7
-
- movd DWORD PTR [edi-4], mm2 // writeoutput 06 07 08 09
- jl horiz_line_4_5_loop
-
-//Exit:
- movq mm0, [esi] // mm0 = 00 01 02 03 04 05 06 07
- movq mm1, mm0 // mm1 = 00 01 02 03 04 05 06 07
-
- movq mm2, mm0 // mm2 = 00 01 02 03 04 05 06 07
- psrlq mm1, 8 // mm1 = 01 02 03 04 05 06 07 00
-
- movq mm3, mask45 // mm3 = 00 00 00 00 00 00 ff 00
- pand mm3, mm1 // mm3 = 00 00 00 00 00 00 07 00
-
- psllq mm3, 8 // mm3 = 00 00 00 00 00 00 00 07
- por mm1, mm3 // mm1 = 01 02 03 04 05 06 07 07
-
- movq mm3, mm1
-
- movd DWORD PTR [edi], mm0 // write output 00 xx xx xx
- punpcklbw mm0, mm7 // mm0 = 00 xx 01 xx 02 xx 03 xx
-
- punpcklbw mm1, mm7 // mm1 = 01 xx 02 xx 03 xx 04 xx
- pmullw mm0, mm5 // 00* 51 01*102 02*154 03*205
-
- pmullw mm1, mm6 // 01*205 02*154 03*102 04* 51
- punpckhbw mm2, mm7 // mm2 = 04 xx 05 xx 06 xx 07 xx
-
- movd DWORD PTR [edi+5], mm2 // write ouput 05 xx xx xx
- pmullw mm2, mm5 // 04* 51 05*102 06*154 07*205
-
- punpckhbw mm3, mm7 // mm3 = 05 xx 06 xx 07 xx 08 xx
- pmullw mm3, mm6 // 05*205 06*154 07*102 07* 51
-
- paddw mm0, mm1 // added round values
- paddw mm0, mm4
-
- psrlw mm0, 8 // output: 01 xx 02 xx 03 xx 04 xx
- packuswb mm0, mm7 // 01 02 03 04 xx xx xx xx
-
- movd DWORD PTR [edi+1], mm0 // write output 01 02 03 04
- paddw mm2, mm3 //
-
- paddw mm2, mm4 // added round values
- psrlw mm2, 8
-
- packuswb mm2, mm7
- movd DWORD PTR [edi+6], mm2 // writeoutput 06 07 08 09
-
-
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vertical_band_4_5_scale_mmx
- *
- * INPUTS : unsigned char *dest :
- * unsigned int dest_pitch :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 4 to 5 up-scaling of a 4 pixel high band of pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band. The function also has a "C" only
- * version.
- *
- ****************************************************************************/
-static
-void vertical_band_4_5_scale_mmx
-(
- unsigned char *dest,
- unsigned int dest_pitch,
- unsigned int dest_width
-)
-{
- __asm
- {
-
- mov esi, dest // Get the source and destination pointer
- mov ecx, dest_pitch // Get the pitch size
-
- lea edi, [esi+ecx*2] // tow lines below
- add edi, ecx // three lines below
-
- pxor mm7, mm7 // clear out mm7
- mov edx, dest_width // Loop counter
-
- vs_4_5_loop:
-
- movq mm0, QWORD ptr [esi] // src[0];
- movq mm1, QWORD ptr [esi+ecx] // src[1];
-
- movq mm2, mm0 // Make a copy
- punpcklbw mm0, mm7 // unpack low to word
-
- movq mm5, one_fifth
- punpckhbw mm2, mm7 // unpack high to word
-
- pmullw mm0, mm5 // a * 1/5
-
- movq mm3, mm1 // make a copy
- punpcklbw mm1, mm7 // unpack low to word
-
- pmullw mm2, mm5 // a * 1/5
- movq mm6, four_fifths // constan
-
- movq mm4, mm1 // copy of low b
- pmullw mm4, mm6 // b * 4/5
-
- punpckhbw mm3, mm7 // unpack high to word
- movq mm5, mm3 // copy of high b
-
- pmullw mm5, mm6 // b * 4/5
- paddw mm0, mm4 // a * 1/5 + b * 4/5
-
- paddw mm2, mm5 // a * 1/5 + b * 4/5
- paddw mm0, round_values // + 128
-
- paddw mm2, round_values // + 128
- psrlw mm0, 8
-
- psrlw mm2, 8
- packuswb mm0, mm2 // des [1]
-
- movq QWORD ptr [esi+ecx], mm0 // write des[1]
- movq mm0, [esi+ecx*2] // mm0 = src[2]
-
- // mm1, mm3 --- Src[1]
- // mm0 --- Src[2]
- // mm7 for unpacking
-
- movq mm5, two_fifths
- movq mm2, mm0 // make a copy
-
- pmullw mm1, mm5 // b * 2/5
- movq mm6, three_fifths
-
-
- punpcklbw mm0, mm7 // unpack low to word
- pmullw mm3, mm5 // b * 2/5
-
- movq mm4, mm0 // make copy of c
- punpckhbw mm2, mm7 // unpack high to word
-
- pmullw mm4, mm6 // c * 3/5
- movq mm5, mm2
-
- pmullw mm5, mm6 // c * 3/5
- paddw mm1, mm4 // b * 2/5 + c * 3/5
-
- paddw mm3, mm5 // b * 2/5 + c * 3/5
- paddw mm1, round_values // + 128
-
- paddw mm3, round_values // + 128
- psrlw mm1, 8
-
- psrlw mm3, 8
- packuswb mm1, mm3 // des[2]
-
- movq QWORD ptr [esi+ecx*2], mm1 // write des[2]
- movq mm1, [edi] // mm1=Src[3];
-
- // mm0, mm2 --- Src[2]
- // mm1 --- Src[3]
- // mm6 --- 3/5
- // mm7 for unpacking
-
- pmullw mm0, mm6 // c * 3/5
- movq mm5, two_fifths // mm5 = 2/5
-
- movq mm3, mm1 // make a copy
- pmullw mm2, mm6 // c * 3/5
-
- punpcklbw mm1, mm7 // unpack low
- movq mm4, mm1 // make a copy
-
- punpckhbw mm3, mm7 // unpack high
- pmullw mm4, mm5 // d * 2/5
-
- movq mm6, mm3 // make a copy
- pmullw mm6, mm5 // d * 2/5
-
- paddw mm0, mm4 // c * 3/5 + d * 2/5
- paddw mm2, mm6 // c * 3/5 + d * 2/5
-
- paddw mm0, round_values // + 128
- paddw mm2, round_values // + 128
-
- psrlw mm0, 8
- psrlw mm2, 8
-
- packuswb mm0, mm2 // des[3]
- movq QWORD ptr [edi], mm0 // write des[3]
-
- // mm1, mm3 --- Src[3]
- // mm7 -- cleared for unpacking
-
- movq mm0, [edi+ecx*2] // mm0, Src[0] of the next group
-
- movq mm5, four_fifths // mm5 = 4/5
- pmullw mm1, mm5 // d * 4/5
-
- movq mm6, one_fifth // mm6 = 1/5
- movq mm2, mm0 // make a copy
-
- pmullw mm3, mm5 // d * 4/5
- punpcklbw mm0, mm7 // unpack low
-
- pmullw mm0, mm6 // an * 1/5
- punpckhbw mm2, mm7 // unpack high
-
- paddw mm1, mm0 // d * 4/5 + an * 1/5
- pmullw mm2, mm6 // an * 1/5
-
- paddw mm3, mm2 // d * 4/5 + an * 1/5
- paddw mm1, round_values // + 128
-
- paddw mm3, round_values // + 128
- psrlw mm1, 8
-
- psrlw mm3, 8
- packuswb mm1, mm3 // des[4]
-
- movq QWORD ptr [edi+ecx], mm1 // write des[4]
-
- add edi, 8
- add esi, 8
-
- sub edx, 8
- jg vs_4_5_loop
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : last_vertical_band_4_5_scale_mmx
- *
- * INPUTS : unsigned char *dest :
- * unsigned int dest_pitch :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : None
- *
- * FUNCTION : 4 to 5 up-scaling of the last 4-pixel high band in an image.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band. The function also has an "C" only
- * version.
- *
- ****************************************************************************/
-static
-void last_vertical_band_4_5_scale_mmx
-(
- unsigned char *dest,
- unsigned int dest_pitch,
- unsigned int dest_width
-)
-{
- __asm
- {
- mov esi, dest // Get the source and destination pointer
- mov ecx, dest_pitch // Get the pitch size
-
- lea edi, [esi+ecx*2] // tow lines below
- add edi, ecx // three lines below
-
- pxor mm7, mm7 // clear out mm7
- mov edx, dest_width // Loop counter
-
- last_vs_4_5_loop:
-
- movq mm0, QWORD ptr [esi] // src[0];
- movq mm1, QWORD ptr [esi+ecx] // src[1];
-
- movq mm2, mm0 // Make a copy
- punpcklbw mm0, mm7 // unpack low to word
-
- movq mm5, one_fifth
- punpckhbw mm2, mm7 // unpack high to word
-
- pmullw mm0, mm5 // a * 1/5
-
- movq mm3, mm1 // make a copy
- punpcklbw mm1, mm7 // unpack low to word
-
- pmullw mm2, mm5 // a * 1/5
- movq mm6, four_fifths // constan
-
- movq mm4, mm1 // copy of low b
- pmullw mm4, mm6 // b * 4/5
-
- punpckhbw mm3, mm7 // unpack high to word
- movq mm5, mm3 // copy of high b
-
- pmullw mm5, mm6 // b * 4/5
- paddw mm0, mm4 // a * 1/5 + b * 4/5
-
- paddw mm2, mm5 // a * 1/5 + b * 4/5
- paddw mm0, round_values // + 128
-
- paddw mm2, round_values // + 128
- psrlw mm0, 8
-
- psrlw mm2, 8
- packuswb mm0, mm2 // des [1]
-
- movq QWORD ptr [esi+ecx], mm0 // write des[1]
- movq mm0, [esi+ecx*2] // mm0 = src[2]
-
- // mm1, mm3 --- Src[1]
- // mm0 --- Src[2]
- // mm7 for unpacking
-
- movq mm5, two_fifths
- movq mm2, mm0 // make a copy
-
- pmullw mm1, mm5 // b * 2/5
- movq mm6, three_fifths
-
-
- punpcklbw mm0, mm7 // unpack low to word
- pmullw mm3, mm5 // b * 2/5
-
- movq mm4, mm0 // make copy of c
- punpckhbw mm2, mm7 // unpack high to word
-
- pmullw mm4, mm6 // c * 3/5
- movq mm5, mm2
-
- pmullw mm5, mm6 // c * 3/5
- paddw mm1, mm4 // b * 2/5 + c * 3/5
-
- paddw mm3, mm5 // b * 2/5 + c * 3/5
- paddw mm1, round_values // + 128
-
- paddw mm3, round_values // + 128
- psrlw mm1, 8
-
- psrlw mm3, 8
- packuswb mm1, mm3 // des[2]
-
- movq QWORD ptr [esi+ecx*2], mm1 // write des[2]
- movq mm1, [edi] // mm1=Src[3];
-
- movq QWORD ptr [edi+ecx], mm1 // write des[4];
-
- // mm0, mm2 --- Src[2]
- // mm1 --- Src[3]
- // mm6 --- 3/5
- // mm7 for unpacking
-
- pmullw mm0, mm6 // c * 3/5
- movq mm5, two_fifths // mm5 = 2/5
-
- movq mm3, mm1 // make a copy
- pmullw mm2, mm6 // c * 3/5
-
- punpcklbw mm1, mm7 // unpack low
- movq mm4, mm1 // make a copy
-
- punpckhbw mm3, mm7 // unpack high
- pmullw mm4, mm5 // d * 2/5
-
- movq mm6, mm3 // make a copy
- pmullw mm6, mm5 // d * 2/5
-
- paddw mm0, mm4 // c * 3/5 + d * 2/5
- paddw mm2, mm6 // c * 3/5 + d * 2/5
-
- paddw mm0, round_values // + 128
- paddw mm2, round_values // + 128
-
- psrlw mm0, 8
- psrlw mm2, 8
-
- packuswb mm0, mm2 // des[3]
- movq QWORD ptr [edi], mm0 // write des[3]
-
- // mm1, mm3 --- Src[3]
- // mm7 -- cleared for unpacking
- add edi, 8
- add esi, 8
-
- sub edx, 8
- jg last_vs_4_5_loop
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vertical_band_3_5_scale_mmx
- *
- * INPUTS : unsigned char *dest :
- * unsigned int dest_pitch :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 3 to 5 up-scaling of a 3-pixel high band of pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band. The function also has an "C" only
- * version.
- *
- ****************************************************************************/
-static
-void vertical_band_3_5_scale_mmx
-(
- unsigned char *dest,
- unsigned int dest_pitch,
- unsigned int dest_width
-)
-{
- __asm
- {
- mov esi, dest // Get the source and destination pointer
- mov ecx, dest_pitch // Get the pitch size
-
- lea edi, [esi+ecx*2] // tow lines below
- add edi, ecx // three lines below
-
- pxor mm7, mm7 // clear out mm7
- mov edx, dest_width // Loop counter
-
- vs_3_5_loop:
-
- movq mm0, QWORD ptr [esi] // src[0];
- movq mm1, QWORD ptr [esi+ecx] // src[1];
-
- movq mm2, mm0 // Make a copy
- punpcklbw mm0, mm7 // unpack low to word
-
- movq mm5, two_fifths // mm5 = 2/5
- punpckhbw mm2, mm7 // unpack high to word
-
- pmullw mm0, mm5 // a * 2/5
-
- movq mm3, mm1 // make a copy
- punpcklbw mm1, mm7 // unpack low to word
-
- pmullw mm2, mm5 // a * 2/5
- movq mm6, three_fifths // mm6 = 3/5
-
- movq mm4, mm1 // copy of low b
- pmullw mm4, mm6 // b * 3/5
-
- punpckhbw mm3, mm7 // unpack high to word
- movq mm5, mm3 // copy of high b
-
- pmullw mm5, mm6 // b * 3/5
- paddw mm0, mm4 // a * 2/5 + b * 3/5
-
- paddw mm2, mm5 // a * 2/5 + b * 3/5
- paddw mm0, round_values // + 128
-
- paddw mm2, round_values // + 128
- psrlw mm0, 8
-
- psrlw mm2, 8
- packuswb mm0, mm2 // des [1]
-
- movq QWORD ptr [esi+ecx], mm0 // write des[1]
- movq mm0, [esi+ecx*2] // mm0 = src[2]
-
- // mm1, mm3 --- Src[1]
- // mm0 --- Src[2]
- // mm7 for unpacking
-
- movq mm4, mm1 // b low
- pmullw mm1, four_fifths // b * 4/5 low
-
- movq mm5, mm3 // b high
- pmullw mm3, four_fifths // b * 4/5 high
-
- movq mm2, mm0 // c
- pmullw mm4, one_fifth // b * 1/5
-
- punpcklbw mm0, mm7 // c low
- pmullw mm5, one_fifth // b * 1/5
-
- movq mm6, mm0 // make copy of c low
- punpckhbw mm2, mm7 // c high
-
- pmullw mm6, one_fifth // c * 1/5 low
- movq mm7, mm2 // make copy of c high
-
- pmullw mm7, one_fifth // c * 1/5 high
- paddw mm1, mm6 // b * 4/5 + c * 1/5 low
-
- paddw mm3, mm7 // b * 4/5 + c * 1/5 high
- movq mm6, mm0 // make copy of c low
-
- pmullw mm6, four_fifths // c * 4/5 low
- movq mm7, mm2 // make copy of c high
-
- pmullw mm7, four_fifths // c * 4/5 high
-
- paddw mm4, mm6 // b * 1/5 + c * 4/5 low
- paddw mm5, mm7 // b * 1/5 + c * 4/5 high
-
- paddw mm1, round_values // + 128
- paddw mm3, round_values // + 128
-
- psrlw mm1, 8
- psrlw mm3, 8
-
- packuswb mm1, mm3 // des[2]
- movq QWORD ptr [esi+ecx*2], mm1 // write des[2]
-
- paddw mm4, round_values // + 128
- paddw mm5, round_values // + 128
-
- psrlw mm4, 8
- psrlw mm5, 8
-
- packuswb mm4, mm5 // des[3]
- movq QWORD ptr [edi], mm4 // write des[3]
-
- // mm0, mm2 --- Src[3]
-
- pxor mm7, mm7 // clear mm7 for unpacking
- movq mm1, [edi+ecx*2] // mm1 = Src[0] of the next group
-
- movq mm5, three_fifths // mm5 = 3/5
- pmullw mm0, mm5 // d * 3/5
-
- movq mm6, two_fifths // mm6 = 2/5
- movq mm3, mm1 // make a copy
-
- pmullw mm2, mm5 // d * 3/5
- punpcklbw mm1, mm7 // unpack low
-
- pmullw mm1, mm6 // an * 2/5
- punpckhbw mm3, mm7 // unpack high
-
- paddw mm0, mm1 // d * 3/5 + an * 2/5
- pmullw mm3, mm6 // an * 2/5
-
- paddw mm2, mm3 // d * 3/5 + an * 2/5
- paddw mm0, round_values // + 128
-
- paddw mm2, round_values // + 128
- psrlw mm0, 8
-
- psrlw mm2, 8
- packuswb mm0, mm2 // des[4]
-
- movq QWORD ptr [edi+ecx], mm0 // write des[4]
-
- add edi, 8
- add esi, 8
-
- sub edx, 8
- jg vs_3_5_loop
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : last_vertical_band_3_5_scale_mmx
- *
- * INPUTS : unsigned char *dest :
- * unsigned int dest_pitch :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 3 to 5 up-scaling of a 3-pixel high band of pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band. The function also has an "C" only
- * version.
- *
- ****************************************************************************/
-static
-void last_vertical_band_3_5_scale_mmx
-(
- unsigned char *dest,
- unsigned int dest_pitch,
- unsigned int dest_width
-)
-{
- __asm
- {
- mov esi, dest // Get the source and destination pointer
- mov ecx, dest_pitch // Get the pitch size
-
- lea edi, [esi+ecx*2] // tow lines below
- add edi, ecx // three lines below
-
- pxor mm7, mm7 // clear out mm7
- mov edx, dest_width // Loop counter
-
-
- last_vs_3_5_loop:
-
- movq mm0, QWORD ptr [esi] // src[0];
- movq mm1, QWORD ptr [esi+ecx] // src[1];
-
- movq mm2, mm0 // Make a copy
- punpcklbw mm0, mm7 // unpack low to word
-
- movq mm5, two_fifths // mm5 = 2/5
- punpckhbw mm2, mm7 // unpack high to word
-
- pmullw mm0, mm5 // a * 2/5
-
- movq mm3, mm1 // make a copy
- punpcklbw mm1, mm7 // unpack low to word
-
- pmullw mm2, mm5 // a * 2/5
- movq mm6, three_fifths // mm6 = 3/5
-
- movq mm4, mm1 // copy of low b
- pmullw mm4, mm6 // b * 3/5
-
- punpckhbw mm3, mm7 // unpack high to word
- movq mm5, mm3 // copy of high b
-
- pmullw mm5, mm6 // b * 3/5
- paddw mm0, mm4 // a * 2/5 + b * 3/5
-
- paddw mm2, mm5 // a * 2/5 + b * 3/5
- paddw mm0, round_values // + 128
-
- paddw mm2, round_values // + 128
- psrlw mm0, 8
-
- psrlw mm2, 8
- packuswb mm0, mm2 // des [1]
-
- movq QWORD ptr [esi+ecx], mm0 // write des[1]
- movq mm0, [esi+ecx*2] // mm0 = src[2]
-
-
-
- // mm1, mm3 --- Src[1]
- // mm0 --- Src[2]
- // mm7 for unpacking
-
- movq mm4, mm1 // b low
- pmullw mm1, four_fifths // b * 4/5 low
-
- movq QWORD ptr [edi+ecx], mm0 // write des[4]
-
- movq mm5, mm3 // b high
- pmullw mm3, four_fifths // b * 4/5 high
-
- movq mm2, mm0 // c
- pmullw mm4, one_fifth // b * 1/5
-
- punpcklbw mm0, mm7 // c low
- pmullw mm5, one_fifth // b * 1/5
-
- movq mm6, mm0 // make copy of c low
- punpckhbw mm2, mm7 // c high
-
- pmullw mm6, one_fifth // c * 1/5 low
- movq mm7, mm2 // make copy of c high
-
- pmullw mm7, one_fifth // c * 1/5 high
- paddw mm1, mm6 // b * 4/5 + c * 1/5 low
-
- paddw mm3, mm7 // b * 4/5 + c * 1/5 high
- movq mm6, mm0 // make copy of c low
-
- pmullw mm6, four_fifths // c * 4/5 low
- movq mm7, mm2 // make copy of c high
-
- pmullw mm7, four_fifths // c * 4/5 high
-
- paddw mm4, mm6 // b * 1/5 + c * 4/5 low
- paddw mm5, mm7 // b * 1/5 + c * 4/5 high
-
- paddw mm1, round_values // + 128
- paddw mm3, round_values // + 128
-
- psrlw mm1, 8
- psrlw mm3, 8
-
- packuswb mm1, mm3 // des[2]
- movq QWORD ptr [esi+ecx*2], mm1 // write des[2]
-
- paddw mm4, round_values // + 128
- paddw mm5, round_values // + 128
-
- psrlw mm4, 8
- psrlw mm5, 8
-
- packuswb mm4, mm5 // des[3]
- movq QWORD ptr [edi], mm4 // write des[3]
-
- // mm0, mm2 --- Src[3]
-
- add edi, 8
- add esi, 8
-
- sub edx, 8
- jg last_vs_3_5_loop
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : vertical_band_1_2_scale_mmx
- *
- * INPUTS : unsigned char *dest :
- * unsigned int dest_pitch :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 1 to 2 up-scaling of a band of pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band. The function also has an "C" only
- * version.
- *
- ****************************************************************************/
-static
-void vertical_band_1_2_scale_mmx
-(
- unsigned char *dest,
- unsigned int dest_pitch,
- unsigned int dest_width
-)
-{
- __asm
- {
-
- mov esi, dest // Get the source and destination pointer
- mov ecx, dest_pitch // Get the pitch size
-
- pxor mm7, mm7 // clear out mm7
- mov edx, dest_width // Loop counter
-
- vs_1_2_loop:
-
- movq mm0, [esi] // get Src[0]
- movq mm1, [esi + ecx * 2] // get Src[1]
-
- movq mm2, mm0 // make copy before unpack
- movq mm3, mm1 // make copy before unpack
-
- punpcklbw mm0, mm7 // low Src[0]
- movq mm6, four_ones // mm6= 1, 1, 1, 1
-
- punpcklbw mm1, mm7 // low Src[1]
- paddw mm0, mm1 // low (a + b)
-
- punpckhbw mm2, mm7 // high Src[0]
- paddw mm0, mm6 // low (a + b + 1)
-
- punpckhbw mm3, mm7
- paddw mm2, mm3 // high (a + b )
-
- psraw mm0, 1 // low (a + b +1 )/2
- paddw mm2, mm6 // high (a + b + 1)
-
- psraw mm2, 1 // high (a + b + 1)/2
- packuswb mm0, mm2 // pack results
-
- movq [esi+ecx], mm0 // write out eight bytes
- add esi, 8
-
- sub edx, 8
- jg vs_1_2_loop
- }
-
-}
-
-/****************************************************************************
- *
- * ROUTINE : last_vertical_band_1_2_scale_mmx
- *
- * INPUTS : unsigned char *dest :
- * unsigned int dest_pitch :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 1 to 2 up-scaling of band of pixels.
- *
- * SPECIAL NOTES : The routine uses the first line of the band below
- * the current band. The function also has an "C" only
- * version.
- *
- ****************************************************************************/
-static
-void last_vertical_band_1_2_scale_mmx
-(
- unsigned char *dest,
- unsigned int dest_pitch,
- unsigned int dest_width
-)
-{
- __asm
- {
- mov esi, dest // Get the source and destination pointer
- mov ecx, dest_pitch // Get the pitch size
-
- mov edx, dest_width // Loop counter
-
- last_vs_1_2_loop:
-
- movq mm0, [esi] // get Src[0]
- movq [esi+ecx], mm0 // write out eight bytes
-
- add esi, 8
- sub edx, 8
-
- jg last_vs_1_2_loop
- }
-}
-
-/****************************************************************************
- *
- * ROUTINE : horizontal_line_1_2_scale
- *
- * INPUTS : const unsigned char *source :
- * unsigned int source_width :
- * unsigned char *dest :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 1 to 2 up-scaling of a horizontal line of pixels.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-void horizontal_line_1_2_scale_mmx
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- (void) dest_width;
-
- __asm
- {
- mov esi, source
- mov edi, dest
-
- pxor mm7, mm7
- movq mm6, four_ones
-
- mov ecx, source_width
-
- hs_1_2_loop:
-
- movq mm0, [esi]
- movq mm1, [esi+1]
-
- movq mm2, mm0
- movq mm3, mm1
-
- movq mm4, mm0
- punpcklbw mm0, mm7
-
- punpcklbw mm1, mm7
- paddw mm0, mm1
-
- paddw mm0, mm6
- punpckhbw mm2, mm7
-
- punpckhbw mm3, mm7
- paddw mm2, mm3
-
- paddw mm2, mm6
- psraw mm0, 1
-
- psraw mm2, 1
- packuswb mm0, mm2
-
- movq mm2, mm4
- punpcklbw mm2, mm0
-
- movq [edi], mm2
- punpckhbw mm4, mm0
-
- movq [edi+8], mm4
- add esi, 8
-
- add edi, 16
- sub ecx, 8
-
- cmp ecx, 8
- jg hs_1_2_loop
-
-// last eight pixel
-
- movq mm0, [esi]
- movq mm1, mm0
-
- movq mm2, mm0
- movq mm3, mm1
-
- psrlq mm1, 8
- psrlq mm3, 56
-
- psllq mm3, 56
- por mm1, mm3
-
- movq mm3, mm1
- movq mm4, mm0
-
- punpcklbw mm0, mm7
- punpcklbw mm1, mm7
-
- paddw mm0, mm1
- paddw mm0, mm6
-
- punpckhbw mm2, mm7
- punpckhbw mm3, mm7
-
- paddw mm2, mm3
- paddw mm2, mm6
-
- psraw mm0, 1
- psraw mm2, 1
-
- packuswb mm0, mm2
- movq mm2, mm4
-
- punpcklbw mm2, mm0
- movq [edi], mm2
-
- punpckhbw mm4, mm0
- movq [edi+8], mm4
- }
-}
-
-
-
-
-
-__declspec(align(16)) const static unsigned short const54_2[] = { 0, 64, 128, 192 };
-__declspec(align(16)) const static unsigned short const54_1[] = {256, 192, 128, 64 };
-
-
-/****************************************************************************
- *
- * ROUTINE : horizontal_line_5_4_scale_mmx
- *
- * INPUTS : const unsigned char *source : Pointer to source data.
- * unsigned int source_width : Stride of source.
- * unsigned char *dest : Pointer to destination data.
- * unsigned int dest_width : Stride of destination (NOT USED).
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Copies horizontal line of pixels from source to
- * destination scaling up by 4 to 5.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-void horizontal_line_5_4_scale_mmx
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- /*
- unsigned i;
- unsigned int a, b, c, d, e;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
- (void) dest_width;
-
- for ( i=0; i<source_width; i+=5 )
- {
- a = src[0];
- b = src[1];
- c = src[2];
- d = src[3];
- e = src[4];
-
- des[0] = a;
- des[1] = ((b*192 + c* 64 + 128)>>8);
- des[2] = ((c*128 + d*128 + 128)>>8);
- des[3] = ((d* 64 + e*192 + 128)>>8);
-
- src += 5;
- des += 4;
- }
- */
- (void) dest_width;
-
- __asm
- {
-
- mov esi, source ;
- mov edi, dest ;
-
- mov ecx, source_width ;
- movq mm5, const54_1 ;
-
- pxor mm7, mm7 ;
- movq mm6, const54_2 ;
-
- movq mm4, round_values ;
- lea edx, [esi+ecx] ;
- horizontal_line_5_4_loop:
-
- movq mm0, QWORD PTR [esi] ;
- 00 01 02 03 04 05 06 07
- movq mm1, mm0 ;
- 00 01 02 03 04 05 06 07
-
- psrlq mm0, 8 ;
- 01 02 03 04 05 06 07 xx
- punpcklbw mm1, mm7 ;
- xx 00 xx 01 xx 02 xx 03
-
- punpcklbw mm0, mm7 ;
- xx 01 xx 02 xx 03 xx 04
- pmullw mm1, mm5
-
- pmullw mm0, mm6
- add esi, 5
-
- add edi, 4
- paddw mm1, mm0
-
- paddw mm1, mm4
- psrlw mm1, 8
-
- cmp esi, edx
- packuswb mm1, mm7
-
- movd DWORD PTR [edi-4], mm1
-
- jl horizontal_line_5_4_loop
-
- }
-
-}
-__declspec(align(16)) const static unsigned short one_fourths[] = { 64, 64, 64, 64 };
-__declspec(align(16)) const static unsigned short two_fourths[] = { 128, 128, 128, 128 };
-__declspec(align(16)) const static unsigned short three_fourths[] = { 192, 192, 192, 192 };
-
-static
-void vertical_band_5_4_scale_mmx(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
-
- __asm
- {
- push ebx
-
- mov esi, source // Get the source and destination pointer
- mov ecx, src_pitch // Get the pitch size
-
- mov edi, dest // tow lines below
- pxor mm7, mm7 // clear out mm7
-
- mov edx, dest_pitch // Loop counter
- mov ebx, dest_width
-
- vs_5_4_loop:
-
- movd mm0, DWORD ptr [esi] // src[0];
- movd mm1, DWORD ptr [esi+ecx] // src[1];
-
- movd mm2, DWORD ptr [esi+ecx*2]
- lea eax, [esi+ecx*2] //
-
- punpcklbw mm1, mm7
- punpcklbw mm2, mm7
-
- movq mm3, mm2
- pmullw mm1, three_fourths
-
- pmullw mm2, one_fourths
- movd mm4, [eax+ecx]
-
- pmullw mm3, two_fourths
- punpcklbw mm4, mm7
-
- movq mm5, mm4
- pmullw mm4, two_fourths
-
- paddw mm1, mm2
- movd mm6, [eax+ecx*2]
-
- pmullw mm5, one_fourths
- paddw mm1, round_values;
-
- paddw mm3, mm4
- psrlw mm1, 8
-
- punpcklbw mm6, mm7
- paddw mm3, round_values
-
- pmullw mm6, three_fourths
- psrlw mm3, 8
-
- packuswb mm1, mm7
- packuswb mm3, mm7
-
- movd DWORD PTR [edi], mm0
- movd DWORD PTR [edi+edx], mm1
-
-
- paddw mm5, mm6
- movd DWORD PTR [edi+edx*2], mm3
-
- lea eax, [edi+edx*2]
- paddw mm5, round_values
-
- psrlw mm5, 8
- add edi, 4
-
- packuswb mm5, mm7
- movd DWORD PTR [eax+edx], mm5
-
- add esi, 4
- sub ebx, 4
-
- jg vs_5_4_loop
-
- pop ebx
- }
-}
-
-
-__declspec(align(16)) const static unsigned short const53_1[] = { 0, 85, 171, 0 };
-__declspec(align(16)) const static unsigned short const53_2[] = {256, 171, 85, 0 };
-
-
-static
-void horizontal_line_5_3_scale_mmx
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
-
- (void) dest_width;
- __asm
- {
-
- mov esi, source ;
- mov edi, dest ;
-
- mov ecx, source_width ;
- movq mm5, const53_1 ;
-
- pxor mm7, mm7 ;
- movq mm6, const53_2 ;
-
- movq mm4, round_values ;
- lea edx, [esi+ecx-5] ;
- horizontal_line_5_3_loop:
-
- movq mm0, QWORD PTR [esi] ;
- 00 01 02 03 04 05 06 07
- movq mm1, mm0 ;
- 00 01 02 03 04 05 06 07
-
- psllw mm0, 8 ;
- xx 00 xx 02 xx 04 xx 06
- psrlw mm1, 8 ;
- 01 xx 03 xx 05 xx 07 xx
-
- psrlw mm0, 8 ;
- 00 xx 02 xx 04 xx 06 xx
- psllq mm1, 16 ;
- xx xx 01 xx 03 xx 05 xx
-
- pmullw mm0, mm6
-
- pmullw mm1, mm5
- add esi, 5
-
- add edi, 3
- paddw mm1, mm0
-
- paddw mm1, mm4
- psrlw mm1, 8
-
- cmp esi, edx
- packuswb mm1, mm7
-
- movd DWORD PTR [edi-3], mm1
- jl horizontal_line_5_3_loop
-
-//exit condition
- movq mm0, QWORD PTR [esi] ;
- 00 01 02 03 04 05 06 07
- movq mm1, mm0 ;
- 00 01 02 03 04 05 06 07
-
- psllw mm0, 8 ;
- xx 00 xx 02 xx 04 xx 06
- psrlw mm1, 8 ;
- 01 xx 03 xx 05 xx 07 xx
-
- psrlw mm0, 8 ;
- 00 xx 02 xx 04 xx 06 xx
- psllq mm1, 16 ;
- xx xx 01 xx 03 xx 05 xx
-
- pmullw mm0, mm6
-
- pmullw mm1, mm5
- paddw mm1, mm0
-
- paddw mm1, mm4
- psrlw mm1, 8
-
- packuswb mm1, mm7
- movd eax, mm1
-
- mov edx, eax
- shr edx, 16
-
- mov WORD PTR[edi], ax
- mov BYTE PTR[edi+2], dl
-
- }
-
-}
-
-__declspec(align(16)) const static unsigned short one_thirds[] = { 85, 85, 85, 85 };
-__declspec(align(16)) const static unsigned short two_thirds[] = { 171, 171, 171, 171 };
-
-static
-void vertical_band_5_3_scale_mmx(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
-
- __asm
- {
- push ebx
-
- mov esi, source // Get the source and destination pointer
- mov ecx, src_pitch // Get the pitch size
-
- mov edi, dest // tow lines below
- pxor mm7, mm7 // clear out mm7
-
- mov edx, dest_pitch // Loop counter
- movq mm5, one_thirds
-
- movq mm6, two_thirds
- mov ebx, dest_width;
-
- vs_5_3_loop:
-
- movd mm0, DWORD ptr [esi] // src[0];
- movd mm1, DWORD ptr [esi+ecx] // src[1];
-
- movd mm2, DWORD ptr [esi+ecx*2]
- lea eax, [esi+ecx*2] //
-
- punpcklbw mm1, mm7
- punpcklbw mm2, mm7
-
- pmullw mm1, mm5
- pmullw mm2, mm6
-
- movd mm3, DWORD ptr [eax+ecx]
- movd mm4, DWORD ptr [eax+ecx*2]
-
- punpcklbw mm3, mm7
- punpcklbw mm4, mm7
-
- pmullw mm3, mm6
- pmullw mm4, mm5
-
-
- movd DWORD PTR [edi], mm0
- paddw mm1, mm2
-
- paddw mm1, round_values
- psrlw mm1, 8
-
- packuswb mm1, mm7
- paddw mm3, mm4
-
- paddw mm3, round_values
- movd DWORD PTR [edi+edx], mm1
-
- psrlw mm3, 8
- packuswb mm3, mm7
-
- movd DWORD PTR [edi+edx*2], mm3
-
-
- add edi, 4
- add esi, 4
-
- sub ebx, 4
- jg vs_5_3_loop
-
- pop ebx
- }
-}
-
-
-
-
-/****************************************************************************
- *
- * ROUTINE : horizontal_line_2_1_scale
- *
- * INPUTS : const unsigned char *source :
- * unsigned int source_width :
- * unsigned char *dest :
- * unsigned int dest_width :
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : 1 to 2 up-scaling of a horizontal line of pixels.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-static
-void horizontal_line_2_1_scale_mmx
-(
- const unsigned char *source,
- unsigned int source_width,
- unsigned char *dest,
- unsigned int dest_width
-)
-{
- (void) dest_width;
- (void) source_width;
- __asm
- {
- mov esi, source
- mov edi, dest
-
- pxor mm7, mm7
- mov ecx, dest_width
-
- xor edx, edx
- hs_2_1_loop:
-
- movq mm0, [esi+edx*2]
- psllw mm0, 8
-
- psrlw mm0, 8
- packuswb mm0, mm7
-
- movd DWORD Ptr [edi+edx], mm0;
- add edx, 4
-
- cmp edx, ecx
- jl hs_2_1_loop
-
- }
-}
-
-
-
-static
-void vertical_band_2_1_scale_mmx(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
- (void) dest_pitch;
- (void) src_pitch;
- vpx_memcpy(dest, source, dest_width);
-}
-
-
-__declspec(align(16)) const static unsigned short three_sixteenths[] = { 48, 48, 48, 48 };
-__declspec(align(16)) const static unsigned short ten_sixteenths[] = { 160, 160, 160, 160 };
-
-static
-void vertical_band_2_1_scale_i_mmx(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width)
-{
-
- (void) dest_pitch;
- __asm
- {
- mov esi, source
- mov edi, dest
-
- mov eax, src_pitch
- mov edx, dest_width
-
- pxor mm7, mm7
- sub esi, eax //back one line
-
-
- lea ecx, [esi+edx];
- movq mm6, round_values;
-
- movq mm5, three_sixteenths;
- movq mm4, ten_sixteenths;
-
- vs_2_1_i_loop:
- movd mm0, [esi] //
- movd mm1, [esi+eax] //
-
- movd mm2, [esi+eax*2] //
- punpcklbw mm0, mm7
-
- pmullw mm0, mm5
- punpcklbw mm1, mm7
-
- pmullw mm1, mm4
- punpcklbw mm2, mm7
-
- pmullw mm2, mm5
- paddw mm0, round_values
-
- paddw mm1, mm2
- paddw mm0, mm1
-
- psrlw mm0, 8
- packuswb mm0, mm7
-
- movd DWORD PTR [edi], mm0
- add esi, 4
-
- add edi, 4;
- cmp esi, ecx
- jl vs_2_1_i_loop
-
- }
-}
-
-
-
-void
-register_mmxscalers(void)
-{
- vp8_horizontal_line_1_2_scale = horizontal_line_1_2_scale_mmx;
- vp8_vertical_band_1_2_scale = vertical_band_1_2_scale_mmx;
- vp8_last_vertical_band_1_2_scale = last_vertical_band_1_2_scale_mmx;
- vp8_horizontal_line_3_5_scale = horizontal_line_3_5_scale_mmx;
- vp8_vertical_band_3_5_scale = vertical_band_3_5_scale_mmx;
- vp8_last_vertical_band_3_5_scale = last_vertical_band_3_5_scale_mmx;
- vp8_horizontal_line_4_5_scale = horizontal_line_4_5_scale_mmx;
- vp8_vertical_band_4_5_scale = vertical_band_4_5_scale_mmx;
- vp8_last_vertical_band_4_5_scale = last_vertical_band_4_5_scale_mmx;
-
- vp8_horizontal_line_3_4_scale = vp8cx_horizontal_line_3_4_scale_c;
- vp8_vertical_band_3_4_scale = vp8cx_vertical_band_3_4_scale_c;
- vp8_last_vertical_band_3_4_scale = vp8cx_last_vertical_band_3_4_scale_c;
- vp8_horizontal_line_2_3_scale = vp8cx_horizontal_line_2_3_scale_c;
- vp8_vertical_band_2_3_scale = vp8cx_vertical_band_2_3_scale_c;
- vp8_last_vertical_band_2_3_scale = vp8cx_last_vertical_band_2_3_scale_c;
-
-
-
- vp8_vertical_band_5_4_scale = vertical_band_5_4_scale_mmx;
- vp8_vertical_band_5_3_scale = vertical_band_5_3_scale_mmx;
- vp8_vertical_band_2_1_scale = vertical_band_2_1_scale_mmx;
- vp8_vertical_band_2_1_scale_i = vertical_band_2_1_scale_i_mmx;
- vp8_horizontal_line_2_1_scale = horizontal_line_2_1_scale_mmx;
- vp8_horizontal_line_5_3_scale = horizontal_line_5_3_scale_mmx;
- vp8_horizontal_line_5_4_scale = horizontal_line_5_4_scale_mmx;
-
-
-
-
-}
diff --git a/vpx_scale/win32/scalesystemdependent.c b/vpx_scale/win32/scalesystemdependent.c
deleted file mode 100644
index 19e61c3..0000000
--- a/vpx_scale/win32/scalesystemdependent.c
+++ /dev/null
@@ -1,91 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/****************************************************************************
-*
-* Module Title : system_dependent.c
-*
-* Description : Miscellaneous system dependent functions
-*
-****************************************************************************/
-
-/****************************************************************************
-* Header Files
-****************************************************************************/
-#include "vpx_scale/vpxscale.h"
-#include "cpuidlib.h"
-
-/****************************************************************************
-* Imports
-*****************************************************************************/
-extern void register_generic_scalers(void);
-extern void register_mmxscalers(void);
-
-/****************************************************************************
- *
- * ROUTINE : post_proc_machine_specific_config
- *
- * INPUTS : UINT32 Version : Codec version number.
- *
- * OUTPUTS : None.
- *
- * RETURNS : void
- *
- * FUNCTION : Checks for machine specifc features such as MMX support
- * sets appropriate flags and function pointers.
- *
- * SPECIAL NOTES : None.
- *
- ****************************************************************************/
-void
-vp8_scale_machine_specific_config(void)
-{
- // If MMX supported then set to use MMX versions of functions else
- // use original 'C' versions.
- int mmx_enabled;
- int xmm_enabled;
- int wmt_enabled;
-
- vpx_get_processor_flags(&mmx_enabled, &xmm_enabled, &wmt_enabled);
-
- if (mmx_enabled || xmm_enabled || wmt_enabled)
- {
- register_mmxscalers();
- }
- else
- {
- vp8_horizontal_line_1_2_scale = vp8cx_horizontal_line_1_2_scale_c;
- vp8_vertical_band_1_2_scale = vp8cx_vertical_band_1_2_scale_c;
- vp8_last_vertical_band_1_2_scale = vp8cx_last_vertical_band_1_2_scale_c;
- vp8_horizontal_line_3_5_scale = vp8cx_horizontal_line_3_5_scale_c;
- vp8_vertical_band_3_5_scale = vp8cx_vertical_band_3_5_scale_c;
- vp8_last_vertical_band_3_5_scale = vp8cx_last_vertical_band_3_5_scale_c;
- vp8_horizontal_line_3_4_scale = vp8cx_horizontal_line_3_4_scale_c;
- vp8_vertical_band_3_4_scale = vp8cx_vertical_band_3_4_scale_c;
- vp8_last_vertical_band_3_4_scale = vp8cx_last_vertical_band_3_4_scale_c;
- vp8_horizontal_line_2_3_scale = vp8cx_horizontal_line_2_3_scale_c;
- vp8_vertical_band_2_3_scale = vp8cx_vertical_band_2_3_scale_c;
- vp8_last_vertical_band_2_3_scale = vp8cx_last_vertical_band_2_3_scale_c;
- vp8_horizontal_line_4_5_scale = vp8cx_horizontal_line_4_5_scale_c;
- vp8_vertical_band_4_5_scale = vp8cx_vertical_band_4_5_scale_c;
- vp8_last_vertical_band_4_5_scale = vp8cx_last_vertical_band_4_5_scale_c;
-
-
- vp8_vertical_band_5_4_scale = vp8cx_vertical_band_5_4_scale_c;
- vp8_vertical_band_5_3_scale = vp8cx_vertical_band_5_3_scale_c;
- vp8_vertical_band_2_1_scale = vp8cx_vertical_band_2_1_scale_c;
- vp8_vertical_band_2_1_scale_i = vp8cx_vertical_band_2_1_scale_i_c;
- vp8_horizontal_line_2_1_scale = vp8cx_horizontal_line_2_1_scale_c;
- vp8_horizontal_line_5_3_scale = vp8cx_horizontal_line_5_3_scale_c;
- vp8_horizontal_line_5_4_scale = vp8cx_horizontal_line_5_4_scale_c;
-
- }
-}
diff --git a/vpx_scale/yv12config.h b/vpx_scale/yv12config.h
deleted file mode 100644
index e10db34..0000000
--- a/vpx_scale/yv12config.h
+++ /dev/null
@@ -1,72 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef YV12_CONFIG_H
-#define YV12_CONFIG_H
-#ifdef __cplusplus
-extern "C"
-{
-#endif
-
-#define VP7BORDERINPIXELS 48
-#define VP8BORDERINPIXELS 32
-
- /*************************************
- For INT_YUV:
-
- Y = (R+G*2+B)/4;
- U = (R-B)/2;
- V = (G*2 - R - B)/4;
- And
- R = Y+U-V;
- G = Y+V;
- B = Y-U-V;
- ************************************/
- typedef enum
- {
- REG_YUV = 0, /* Regular yuv */
- INT_YUV = 1 /* The type of yuv that can be tranfer to and from RGB through integer transform */
- }
- YUV_TYPE;
-
- typedef struct
- {
- int y_width;
- int y_height;
- int y_stride;
-/* int yinternal_width; */
-
- int uv_width;
- int uv_height;
- int uv_stride;
-/* int uvinternal_width; */
-
- unsigned char *y_buffer;
- unsigned char *u_buffer;
- unsigned char *v_buffer;
-
- unsigned char *buffer_alloc;
- int border;
- int frame_size;
- YUV_TYPE clrtype;
-
- int corrupted;
- } YV12_BUFFER_CONFIG;
-
- int vp8_yv12_alloc_frame_buffer(YV12_BUFFER_CONFIG *ybf, int width, int height, int border);
- int vp8_yv12_de_alloc_frame_buffer(YV12_BUFFER_CONFIG *ybf);
-
-#ifdef __cplusplus
-}
-#endif
-
-
-#endif /*YV12_CONFIG_H*/
diff --git a/vpx_scale/yv12extend.h b/vpx_scale/yv12extend.h
deleted file mode 100644
index 0c239b9..0000000
--- a/vpx_scale/yv12extend.h
+++ /dev/null
@@ -1,33 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef YV12_EXTEND_H
-#define YV12_EXTEND_H
-
-#include "vpx_scale/yv12config.h"
-
-#ifdef __cplusplus
-extern "C"
-{
-#endif
-
- void vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf);
-
- /* Copy Y,U,V buffer data from src to dst, filling border of dst as well. */
-
- void vp8_yv12_copy_frame(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
- void vp8_yv12_copy_frame_yonly(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif
diff --git a/vpx_version.h b/vpx_version.h
deleted file mode 100644
index 1d8ba96..0000000
--- a/vpx_version.h
+++ /dev/null
@@ -1,7 +0,0 @@
-#define VERSION_MAJOR 0
-#define VERSION_MINOR 9
-#define VERSION_PATCH 6
-#define VERSION_EXTRA ""
-#define VERSION_PACKED ((VERSION_MAJOR<<16)|(VERSION_MINOR<<8)|(VERSION_PATCH))
-#define VERSION_STRING_NOSP "v0.9.6"
-#define VERSION_STRING " v0.9.6"
diff --git a/vpxdec.c b/vpxdec.c
deleted file mode 100644
index ca9af1e..0000000
--- a/vpxdec.c
+++ /dev/null
@@ -1,1124 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/* This is a simple program that reads ivf files and decodes them
- * using the new interface. Decoded frames are output as YV12 raw.
- */
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdarg.h>
-#include <string.h>
-#include <limits.h>
-
-#define VPX_CODEC_DISABLE_COMPAT 1
-#include "vpx_config.h"
-#include "vpx/vpx_decoder.h"
-#include "vpx_ports/vpx_timer.h"
-#if CONFIG_VP8_DECODER
-#include "vpx/vp8dx.h"
-#endif
-#if CONFIG_MD5
-#include "md5_utils.h"
-#endif
-#include "tools_common.h"
-#include "nestegg/include/nestegg/nestegg.h"
-
-#if CONFIG_OS_SUPPORT
-#if defined(_WIN32)
-#include <io.h>
-#define snprintf _snprintf
-#define isatty _isatty
-#define fileno _fileno
-#else
-#include <unistd.h>
-#endif
-#endif
-
-#ifndef PATH_MAX
-#define PATH_MAX 256
-#endif
-
-static const char *exec_name;
-
-#define VP8_FOURCC (0x00385056)
-static const struct
-{
- char const *name;
- const vpx_codec_iface_t *iface;
- unsigned int fourcc;
- unsigned int fourcc_mask;
-} ifaces[] =
-{
-#if CONFIG_VP8_DECODER
- {"vp8", &vpx_codec_vp8_dx_algo, VP8_FOURCC, 0x00FFFFFF},
-#endif
-};
-
-#include "args.h"
-static const arg_def_t codecarg = ARG_DEF(NULL, "codec", 1,
- "Codec to use");
-static const arg_def_t use_yv12 = ARG_DEF(NULL, "yv12", 0,
- "Output raw YV12 frames");
-static const arg_def_t use_i420 = ARG_DEF(NULL, "i420", 0,
- "Output raw I420 frames");
-static const arg_def_t flipuvarg = ARG_DEF(NULL, "flipuv", 0,
- "Flip the chroma planes in the output");
-static const arg_def_t noblitarg = ARG_DEF(NULL, "noblit", 0,
- "Don't process the decoded frames");
-static const arg_def_t progressarg = ARG_DEF(NULL, "progress", 0,
- "Show progress after each frame decodes");
-static const arg_def_t limitarg = ARG_DEF(NULL, "limit", 1,
- "Stop decoding after n frames");
-static const arg_def_t postprocarg = ARG_DEF(NULL, "postproc", 0,
- "Postprocess decoded frames");
-static const arg_def_t summaryarg = ARG_DEF(NULL, "summary", 0,
- "Show timing summary");
-static const arg_def_t outputfile = ARG_DEF("o", "output", 1,
- "Output file name pattern (see below)");
-static const arg_def_t threadsarg = ARG_DEF("t", "threads", 1,
- "Max threads to use");
-static const arg_def_t verbosearg = ARG_DEF("v", "verbose", 0,
- "Show version string");
-
-#if CONFIG_MD5
-static const arg_def_t md5arg = ARG_DEF(NULL, "md5", 0,
- "Compute the MD5 sum of the decoded frame");
-#endif
-static const arg_def_t *all_args[] =
-{
- &codecarg, &use_yv12, &use_i420, &flipuvarg, &noblitarg,
- &progressarg, &limitarg, &postprocarg, &summaryarg, &outputfile,
- &threadsarg, &verbosearg,
-#if CONFIG_MD5
- &md5arg,
-#endif
- NULL
-};
-
-#if CONFIG_VP8_DECODER
-static const arg_def_t addnoise_level = ARG_DEF(NULL, "noise-level", 1,
- "Enable VP8 postproc add noise");
-static const arg_def_t deblock = ARG_DEF(NULL, "deblock", 0,
- "Enable VP8 deblocking");
-static const arg_def_t demacroblock_level = ARG_DEF(NULL, "demacroblock-level", 1,
- "Enable VP8 demacroblocking, w/ level");
-static const arg_def_t pp_debug_info = ARG_DEF(NULL, "pp-debug-info", 1,
- "Enable VP8 visible debug info");
-static const arg_def_t pp_disp_ref_frame = ARG_DEF(NULL, "pp-dbg-ref-frame", 1,
- "Display only selected reference frame per macro block");
-static const arg_def_t pp_disp_mb_modes = ARG_DEF(NULL, "pp-dbg-mb-modes", 1,
- "Display only selected macro block modes");
-static const arg_def_t pp_disp_b_modes = ARG_DEF(NULL, "pp-dbg-b-modes", 1,
- "Display only selected block modes");
-static const arg_def_t pp_disp_mvs = ARG_DEF(NULL, "pp-dbg-mvs", 1,
- "Draw only selected motion vectors");
-
-static const arg_def_t *vp8_pp_args[] =
-{
- &addnoise_level, &deblock, &demacroblock_level, &pp_debug_info,
- &pp_disp_ref_frame, &pp_disp_mb_modes, &pp_disp_b_modes, &pp_disp_mvs,
- NULL
-};
-#endif
-
-static void usage_exit()
-{
- int i;
-
- fprintf(stderr, "Usage: %s <options> filename\n\n"
- "Options:\n", exec_name);
- arg_show_usage(stderr, all_args);
-#if CONFIG_VP8_DECODER
- fprintf(stderr, "\nVP8 Postprocessing Options:\n");
- arg_show_usage(stderr, vp8_pp_args);
-#endif
- fprintf(stderr,
- "\nOutput File Patterns:\n\n"
- " The -o argument specifies the name of the file(s) to "
- "write to. If the\n argument does not include any escape "
- "characters, the output will be\n written to a single file. "
- "Otherwise, the filename will be calculated by\n expanding "
- "the following escape characters:\n"
- "\n\t%%w - Frame width"
- "\n\t%%h - Frame height"
- "\n\t%%<n> - Frame number, zero padded to <n> places (1..9)"
- "\n\n Pattern arguments are only supported in conjunction "
- "with the --yv12 and\n --i420 options. If the -o option is "
- "not specified, the output will be\n directed to stdout.\n"
- );
- fprintf(stderr, "\nIncluded decoders:\n\n");
-
- for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++)
- fprintf(stderr, " %-6s - %s\n",
- ifaces[i].name,
- vpx_codec_iface_name(ifaces[i].iface));
-
- exit(EXIT_FAILURE);
-}
-
-void die(const char *fmt, ...)
-{
- va_list ap;
- va_start(ap, fmt);
- vfprintf(stderr, fmt, ap);
- fprintf(stderr, "\n");
- usage_exit();
-}
-
-static unsigned int mem_get_le16(const void *vmem)
-{
- unsigned int val;
- const unsigned char *mem = (const unsigned char *)vmem;
-
- val = mem[1] << 8;
- val |= mem[0];
- return val;
-}
-
-static unsigned int mem_get_le32(const void *vmem)
-{
- unsigned int val;
- const unsigned char *mem = (const unsigned char *)vmem;
-
- val = mem[3] << 24;
- val |= mem[2] << 16;
- val |= mem[1] << 8;
- val |= mem[0];
- return val;
-}
-
-enum file_kind
-{
- RAW_FILE,
- IVF_FILE,
- WEBM_FILE
-};
-
-struct input_ctx
-{
- enum file_kind kind;
- FILE *infile;
- nestegg *nestegg_ctx;
- nestegg_packet *pkt;
- unsigned int chunk;
- unsigned int chunks;
- unsigned int video_track;
-};
-
-#define IVF_FRAME_HDR_SZ (sizeof(uint32_t) + sizeof(uint64_t))
-#define RAW_FRAME_HDR_SZ (sizeof(uint32_t))
-static int read_frame(struct input_ctx *input,
- uint8_t **buf,
- size_t *buf_sz,
- size_t *buf_alloc_sz)
-{
- char raw_hdr[IVF_FRAME_HDR_SZ];
- size_t new_buf_sz;
- FILE *infile = input->infile;
- enum file_kind kind = input->kind;
- if(kind == WEBM_FILE)
- {
- if(input->chunk >= input->chunks)
- {
- unsigned int track;
-
- do
- {
- /* End of this packet, get another. */
- if(input->pkt)
- nestegg_free_packet(input->pkt);
-
- if(nestegg_read_packet(input->nestegg_ctx, &input->pkt) <= 0
- || nestegg_packet_track(input->pkt, &track))
- return 1;
-
- } while(track != input->video_track);
-
- if(nestegg_packet_count(input->pkt, &input->chunks))
- return 1;
- input->chunk = 0;
- }
-
- if(nestegg_packet_data(input->pkt, input->chunk, buf, buf_sz))
- return 1;
- input->chunk++;
-
- return 0;
- }
- /* For both the raw and ivf formats, the frame size is the first 4 bytes
- * of the frame header. We just need to special case on the header
- * size.
- */
- else if (fread(raw_hdr, kind==IVF_FILE
- ? IVF_FRAME_HDR_SZ : RAW_FRAME_HDR_SZ, 1, infile) != 1)
- {
- if (!feof(infile))
- fprintf(stderr, "Failed to read frame size\n");
-
- new_buf_sz = 0;
- }
- else
- {
- new_buf_sz = mem_get_le32(raw_hdr);
-
- if (new_buf_sz > 256 * 1024 * 1024)
- {
- fprintf(stderr, "Error: Read invalid frame size (%u)\n",
- (unsigned int)new_buf_sz);
- new_buf_sz = 0;
- }
-
- if (kind == RAW_FILE && new_buf_sz > 256 * 1024)
- fprintf(stderr, "Warning: Read invalid frame size (%u)"
- " - not a raw file?\n", (unsigned int)new_buf_sz);
-
- if (new_buf_sz > *buf_alloc_sz)
- {
- uint8_t *new_buf = realloc(*buf, 2 * new_buf_sz);
-
- if (new_buf)
- {
- *buf = new_buf;
- *buf_alloc_sz = 2 * new_buf_sz;
- }
- else
- {
- fprintf(stderr, "Failed to allocate compressed data buffer\n");
- new_buf_sz = 0;
- }
- }
- }
-
- *buf_sz = new_buf_sz;
-
- if (*buf_sz)
- {
- if (fread(*buf, 1, *buf_sz, infile) != *buf_sz)
- {
- fprintf(stderr, "Failed to read full frame\n");
- return 1;
- }
-
- return 0;
- }
-
- return 1;
-}
-
-void *out_open(const char *out_fn, int do_md5)
-{
- void *out = NULL;
-
- if (do_md5)
- {
-#if CONFIG_MD5
- MD5Context *md5_ctx = out = malloc(sizeof(MD5Context));
- (void)out_fn;
- MD5Init(md5_ctx);
-#endif
- }
- else
- {
- FILE *outfile = out = strcmp("-", out_fn) ? fopen(out_fn, "wb")
- : set_binary_mode(stdout);
-
- if (!outfile)
- {
- fprintf(stderr, "Failed to output file");
- exit(EXIT_FAILURE);
- }
- }
-
- return out;
-}
-
-void out_put(void *out, const uint8_t *buf, unsigned int len, int do_md5)
-{
- if (do_md5)
- {
-#if CONFIG_MD5
- MD5Update(out, buf, len);
-#endif
- }
- else
- {
- if(fwrite(buf, 1, len, out));
- }
-}
-
-void out_close(void *out, const char *out_fn, int do_md5)
-{
- if (do_md5)
- {
-#if CONFIG_MD5
- uint8_t md5[16];
- int i;
-
- MD5Final(md5, out);
- free(out);
-
- for (i = 0; i < 16; i++)
- printf("%02x", md5[i]);
-
- printf(" %s\n", out_fn);
-#endif
- }
- else
- {
- fclose(out);
- }
-}
-
-unsigned int file_is_ivf(FILE *infile,
- unsigned int *fourcc,
- unsigned int *width,
- unsigned int *height,
- unsigned int *fps_den,
- unsigned int *fps_num)
-{
- char raw_hdr[32];
- int is_ivf = 0;
-
- if (fread(raw_hdr, 1, 32, infile) == 32)
- {
- if (raw_hdr[0] == 'D' && raw_hdr[1] == 'K'
- && raw_hdr[2] == 'I' && raw_hdr[3] == 'F')
- {
- is_ivf = 1;
-
- if (mem_get_le16(raw_hdr + 4) != 0)
- fprintf(stderr, "Error: Unrecognized IVF version! This file may not"
- " decode properly.");
-
- *fourcc = mem_get_le32(raw_hdr + 8);
- *width = mem_get_le16(raw_hdr + 12);
- *height = mem_get_le16(raw_hdr + 14);
- *fps_num = mem_get_le32(raw_hdr + 16);
- *fps_den = mem_get_le32(raw_hdr + 20);
-
- /* Some versions of vpxenc used 1/(2*fps) for the timebase, so
- * we can guess the framerate using only the timebase in this
- * case. Other files would require reading ahead to guess the
- * timebase, like we do for webm.
- */
- if(*fps_num < 1000)
- {
- /* Correct for the factor of 2 applied to the timebase in the
- * encoder.
- */
- if(*fps_num&1)*fps_den<<=1;
- else *fps_num>>=1;
- }
- else
- {
- /* Don't know FPS for sure, and don't have readahead code
- * (yet?), so just default to 30fps.
- */
- *fps_num = 30;
- *fps_den = 1;
- }
- }
- }
-
- if (!is_ivf)
- rewind(infile);
-
- return is_ivf;
-}
-
-
-unsigned int file_is_raw(FILE *infile,
- unsigned int *fourcc,
- unsigned int *width,
- unsigned int *height,
- unsigned int *fps_den,
- unsigned int *fps_num)
-{
- unsigned char buf[32];
- int is_raw = 0;
- vpx_codec_stream_info_t si;
-
- si.sz = sizeof(si);
-
- if (fread(buf, 1, 32, infile) == 32)
- {
- int i;
-
- if(mem_get_le32(buf) < 256 * 1024 * 1024)
- for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++)
- if(!vpx_codec_peek_stream_info(ifaces[i].iface,
- buf + 4, 32 - 4, &si))
- {
- is_raw = 1;
- *fourcc = ifaces[i].fourcc;
- *width = si.w;
- *height = si.h;
- *fps_num = 30;
- *fps_den = 1;
- break;
- }
- }
-
- rewind(infile);
- return is_raw;
-}
-
-
-static int
-nestegg_read_cb(void *buffer, size_t length, void *userdata)
-{
- FILE *f = userdata;
-
- if(fread(buffer, 1, length, f) < length)
- {
- if (ferror(f))
- return -1;
- if (feof(f))
- return 0;
- }
- return 1;
-}
-
-
-static int
-nestegg_seek_cb(int64_t offset, int whence, void * userdata)
-{
- switch(whence) {
- case NESTEGG_SEEK_SET: whence = SEEK_SET; break;
- case NESTEGG_SEEK_CUR: whence = SEEK_CUR; break;
- case NESTEGG_SEEK_END: whence = SEEK_END; break;
- };
- return fseek(userdata, offset, whence)? -1 : 0;
-}
-
-
-static int64_t
-nestegg_tell_cb(void * userdata)
-{
- return ftell(userdata);
-}
-
-
-static void
-nestegg_log_cb(nestegg * context, unsigned int severity, char const * format,
- ...)
-{
- va_list ap;
-
- va_start(ap, format);
- vfprintf(stderr, format, ap);
- fprintf(stderr, "\n");
- va_end(ap);
-}
-
-
-static int
-webm_guess_framerate(struct input_ctx *input,
- unsigned int *fps_den,
- unsigned int *fps_num)
-{
- unsigned int i;
- uint64_t tstamp=0;
-
- /* Guess the framerate. Read up to 1 second, or 50 video packets,
- * whichever comes first.
- */
- for(i=0; tstamp < 1000000000 && i < 50;)
- {
- nestegg_packet * pkt;
- unsigned int track;
-
- if(nestegg_read_packet(input->nestegg_ctx, &pkt) <= 0)
- break;
-
- nestegg_packet_track(pkt, &track);
- if(track == input->video_track)
- {
- nestegg_packet_tstamp(pkt, &tstamp);
- i++;
- }
-
- nestegg_free_packet(pkt);
- }
-
- if(nestegg_track_seek(input->nestegg_ctx, input->video_track, 0))
- goto fail;
-
- *fps_num = (i - 1) * 1000000;
- *fps_den = tstamp / 1000;
- return 0;
-fail:
- nestegg_destroy(input->nestegg_ctx);
- input->nestegg_ctx = NULL;
- rewind(input->infile);
- return 1;
-}
-
-
-static int
-file_is_webm(struct input_ctx *input,
- unsigned int *fourcc,
- unsigned int *width,
- unsigned int *height,
- unsigned int *fps_den,
- unsigned int *fps_num)
-{
- unsigned int i, n;
- int track_type = -1;
- uint64_t tstamp=0;
-
- nestegg_io io = {nestegg_read_cb, nestegg_seek_cb, nestegg_tell_cb,
- input->infile};
- nestegg_video_params params;
- nestegg_packet * pkt;
-
- if(nestegg_init(&input->nestegg_ctx, io, NULL))
- goto fail;
-
- if(nestegg_track_count(input->nestegg_ctx, &n))
- goto fail;
-
- for(i=0; i<n; i++)
- {
- track_type = nestegg_track_type(input->nestegg_ctx, i);
-
- if(track_type == NESTEGG_TRACK_VIDEO)
- break;
- else if(track_type < 0)
- goto fail;
- }
-
- if(nestegg_track_codec_id(input->nestegg_ctx, i) != NESTEGG_CODEC_VP8)
- {
- fprintf(stderr, "Not VP8 video, quitting.\n");
- exit(1);
- }
-
- input->video_track = i;
-
- if(nestegg_track_video_params(input->nestegg_ctx, i, ¶ms))
- goto fail;
-
- *fps_den = 0;
- *fps_num = 0;
- *fourcc = VP8_FOURCC;
- *width = params.width;
- *height = params.height;
- return 1;
-fail:
- input->nestegg_ctx = NULL;
- rewind(input->infile);
- return 0;
-}
-
-
-void show_progress(int frame_in, int frame_out, unsigned long dx_time)
-{
- fprintf(stderr, "%d decoded frames/%d showed frames in %lu us (%.2f fps)\r",
- frame_in, frame_out, dx_time,
- (float)frame_out * 1000000.0 / (float)dx_time);
-}
-
-
-void generate_filename(const char *pattern, char *out, size_t q_len,
- unsigned int d_w, unsigned int d_h,
- unsigned int frame_in)
-{
- const char *p = pattern;
- char *q = out;
-
- do
- {
- char *next_pat = strchr(p, '%');
-
- if(p == next_pat)
- {
- size_t pat_len;
-
- // parse the pattern
- q[q_len - 1] = '\0';
- switch(p[1])
- {
- case 'w': snprintf(q, q_len - 1, "%d", d_w); break;
- case 'h': snprintf(q, q_len - 1, "%d", d_h); break;
- case '1': snprintf(q, q_len - 1, "%d", frame_in); break;
- case '2': snprintf(q, q_len - 1, "%02d", frame_in); break;
- case '3': snprintf(q, q_len - 1, "%03d", frame_in); break;
- case '4': snprintf(q, q_len - 1, "%04d", frame_in); break;
- case '5': snprintf(q, q_len - 1, "%05d", frame_in); break;
- case '6': snprintf(q, q_len - 1, "%06d", frame_in); break;
- case '7': snprintf(q, q_len - 1, "%07d", frame_in); break;
- case '8': snprintf(q, q_len - 1, "%08d", frame_in); break;
- case '9': snprintf(q, q_len - 1, "%09d", frame_in); break;
- default:
- die("Unrecognized pattern %%%c\n", p[1]);
- }
-
- pat_len = strlen(q);
- if(pat_len >= q_len - 1)
- die("Output filename too long.\n");
- q += pat_len;
- p += 2;
- q_len -= pat_len;
- }
- else
- {
- size_t copy_len;
-
- // copy the next segment
- if(!next_pat)
- copy_len = strlen(p);
- else
- copy_len = next_pat - p;
-
- if(copy_len >= q_len - 1)
- die("Output filename too long.\n");
-
- memcpy(q, p, copy_len);
- q[copy_len] = '\0';
- q += copy_len;
- p += copy_len;
- q_len -= copy_len;
- }
- } while(*p);
-}
-
-
-int main(int argc, const char **argv_)
-{
- vpx_codec_ctx_t decoder;
- char *fn = NULL;
- int i;
- uint8_t *buf = NULL;
- size_t buf_sz = 0, buf_alloc_sz = 0;
- FILE *infile;
- int frame_in = 0, frame_out = 0, flipuv = 0, noblit = 0, do_md5 = 0, progress = 0;
- int stop_after = 0, postproc = 0, summary = 0, quiet = 1;
- vpx_codec_iface_t *iface = NULL;
- unsigned int fourcc;
- unsigned long dx_time = 0;
- struct arg arg;
- char **argv, **argi, **argj;
- const char *outfile_pattern = 0;
- char outfile[PATH_MAX];
- int single_file;
- int use_y4m = 1;
- unsigned int width;
- unsigned int height;
- unsigned int fps_den;
- unsigned int fps_num;
- void *out = NULL;
- vpx_codec_dec_cfg_t cfg = {0};
-#if CONFIG_VP8_DECODER
- vp8_postproc_cfg_t vp8_pp_cfg = {0};
- int vp8_dbg_color_ref_frame = 0;
- int vp8_dbg_color_mb_modes = 0;
- int vp8_dbg_color_b_modes = 0;
- int vp8_dbg_display_mv = 0;
-#endif
- struct input_ctx input = {0};
-
- /* Parse command line */
- exec_name = argv_[0];
- argv = argv_dup(argc - 1, argv_ + 1);
-
- for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step)
- {
- memset(&arg, 0, sizeof(arg));
- arg.argv_step = 1;
-
- if (arg_match(&arg, &codecarg, argi))
- {
- int j, k = -1;
-
- for (j = 0; j < sizeof(ifaces) / sizeof(ifaces[0]); j++)
- if (!strcmp(ifaces[j].name, arg.val))
- k = j;
-
- if (k >= 0)
- iface = ifaces[k].iface;
- else
- die("Error: Unrecognized argument (%s) to --codec\n",
- arg.val);
- }
- else if (arg_match(&arg, &outputfile, argi))
- outfile_pattern = arg.val;
- else if (arg_match(&arg, &use_yv12, argi))
- {
- use_y4m = 0;
- flipuv = 1;
- }
- else if (arg_match(&arg, &use_i420, argi))
- {
- use_y4m = 0;
- flipuv = 0;
- }
- else if (arg_match(&arg, &flipuvarg, argi))
- flipuv = 1;
- else if (arg_match(&arg, &noblitarg, argi))
- noblit = 1;
- else if (arg_match(&arg, &progressarg, argi))
- progress = 1;
- else if (arg_match(&arg, &limitarg, argi))
- stop_after = arg_parse_uint(&arg);
- else if (arg_match(&arg, &postprocarg, argi))
- postproc = 1;
- else if (arg_match(&arg, &md5arg, argi))
- do_md5 = 1;
- else if (arg_match(&arg, &summaryarg, argi))
- summary = 1;
- else if (arg_match(&arg, &threadsarg, argi))
- cfg.threads = arg_parse_uint(&arg);
- else if (arg_match(&arg, &verbosearg, argi))
- quiet = 0;
-
-#if CONFIG_VP8_DECODER
- else if (arg_match(&arg, &addnoise_level, argi))
- {
- postproc = 1;
- vp8_pp_cfg.post_proc_flag |= VP8_ADDNOISE;
- vp8_pp_cfg.noise_level = arg_parse_uint(&arg);
- }
- else if (arg_match(&arg, &demacroblock_level, argi))
- {
- postproc = 1;
- vp8_pp_cfg.post_proc_flag |= VP8_DEMACROBLOCK;
- vp8_pp_cfg.deblocking_level = arg_parse_uint(&arg);
- }
- else if (arg_match(&arg, &deblock, argi))
- {
- postproc = 1;
- vp8_pp_cfg.post_proc_flag |= VP8_DEBLOCK;
- }
- else if (arg_match(&arg, &pp_debug_info, argi))
- {
- unsigned int level = arg_parse_uint(&arg);
-
- postproc = 1;
- vp8_pp_cfg.post_proc_flag &= ~0x7;
-
- if (level)
- vp8_pp_cfg.post_proc_flag |= level;
- }
- else if (arg_match(&arg, &pp_disp_ref_frame, argi))
- {
- unsigned int flags = arg_parse_int(&arg);
- if (flags)
- {
- postproc = 1;
- vp8_dbg_color_ref_frame = flags;
- }
- }
- else if (arg_match(&arg, &pp_disp_mb_modes, argi))
- {
- unsigned int flags = arg_parse_int(&arg);
- if (flags)
- {
- postproc = 1;
- vp8_dbg_color_mb_modes = flags;
- }
- }
- else if (arg_match(&arg, &pp_disp_b_modes, argi))
- {
- unsigned int flags = arg_parse_int(&arg);
- if (flags)
- {
- postproc = 1;
- vp8_dbg_color_b_modes = flags;
- }
- }
- else if (arg_match(&arg, &pp_disp_mvs, argi))
- {
- unsigned int flags = arg_parse_int(&arg);
- if (flags)
- {
- postproc = 1;
- vp8_dbg_display_mv = flags;
- }
- }
-
-#endif
- else
- argj++;
- }
-
- /* Check for unrecognized options */
- for (argi = argv; *argi; argi++)
- if (argi[0][0] == '-' && strlen(argi[0]) > 1)
- die("Error: Unrecognized option %s\n", *argi);
-
- /* Handle non-option arguments */
- fn = argv[0];
-
- if (!fn)
- usage_exit();
-
- /* Open file */
- infile = strcmp(fn, "-") ? fopen(fn, "rb") : set_binary_mode(stdin);
-
- if (!infile)
- {
- fprintf(stderr, "Failed to open file '%s'",
- strcmp(fn, "-") ? fn : "stdin");
- return EXIT_FAILURE;
- }
-#if CONFIG_OS_SUPPORT
- /* Make sure we don't dump to the terminal, unless forced to with -o - */
- if(!outfile_pattern && isatty(fileno(stdout)) && !do_md5 && !noblit)
- {
- fprintf(stderr,
- "Not dumping raw video to your terminal. Use '-o -' to "
- "override.\n");
- return EXIT_FAILURE;
- }
-#endif
- input.infile = infile;
- if(file_is_ivf(infile, &fourcc, &width, &height, &fps_den,
- &fps_num))
- input.kind = IVF_FILE;
- else if(file_is_webm(&input, &fourcc, &width, &height, &fps_den, &fps_num))
- input.kind = WEBM_FILE;
- else if(file_is_raw(infile, &fourcc, &width, &height, &fps_den, &fps_num))
- input.kind = RAW_FILE;
- else
- {
- fprintf(stderr, "Unrecognized input file type.\n");
- return EXIT_FAILURE;
- }
-
- /* If the output file is not set or doesn't have a sequence number in
- * it, then we only open it once.
- */
- outfile_pattern = outfile_pattern ? outfile_pattern : "-";
- single_file = 1;
- {
- const char *p = outfile_pattern;
- do
- {
- p = strchr(p, '%');
- if(p && p[1] >= '1' && p[1] <= '9')
- {
- // pattern contains sequence number, so it's not unique.
- single_file = 0;
- break;
- }
- if(p)
- p++;
- } while(p);
- }
-
- if(single_file && !noblit)
- {
- generate_filename(outfile_pattern, outfile, sizeof(outfile)-1,
- width, height, 0);
- out = out_open(outfile, do_md5);
- }
-
- if (use_y4m && !noblit)
- {
- char buffer[128];
- if (!single_file)
- {
- fprintf(stderr, "YUV4MPEG2 not supported with output patterns,"
- " try --i420 or --yv12.\n");
- return EXIT_FAILURE;
- }
-
- if(input.kind == WEBM_FILE)
- if(webm_guess_framerate(&input, &fps_den, &fps_num))
- {
- fprintf(stderr, "Failed to guess framerate -- error parsing "
- "webm file?\n");
- return EXIT_FAILURE;
- }
-
-
- /*Note: We can't output an aspect ratio here because IVF doesn't
- store one, and neither does VP8.
- That will have to wait until these tools support WebM natively.*/
- sprintf(buffer, "YUV4MPEG2 C%s W%u H%u F%u:%u I%c\n",
- "420jpeg", width, height, fps_num, fps_den, 'p');
- out_put(out, (unsigned char *)buffer, strlen(buffer), do_md5);
- }
-
- /* Try to determine the codec from the fourcc. */
- for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++)
- if ((fourcc & ifaces[i].fourcc_mask) == ifaces[i].fourcc)
- {
- vpx_codec_iface_t *ivf_iface = ifaces[i].iface;
-
- if (iface && iface != ivf_iface)
- fprintf(stderr, "Notice -- IVF header indicates codec: %s\n",
- ifaces[i].name);
- else
- iface = ivf_iface;
-
- break;
- }
-
- if (vpx_codec_dec_init(&decoder, iface ? iface : ifaces[0].iface, &cfg,
- postproc ? VPX_CODEC_USE_POSTPROC : 0))
- {
- fprintf(stderr, "Failed to initialize decoder: %s\n", vpx_codec_error(&decoder));
- return EXIT_FAILURE;
- }
-
- if (!quiet)
- fprintf(stderr, "%s\n", decoder.name);
-
-#if CONFIG_VP8_DECODER
-
- if (vp8_pp_cfg.post_proc_flag
- && vpx_codec_control(&decoder, VP8_SET_POSTPROC, &vp8_pp_cfg))
- {
- fprintf(stderr, "Failed to configure postproc: %s\n", vpx_codec_error(&decoder));
- return EXIT_FAILURE;
- }
-
- if (vp8_dbg_color_ref_frame
- && vpx_codec_control(&decoder, VP8_SET_DBG_COLOR_REF_FRAME, vp8_dbg_color_ref_frame))
- {
- fprintf(stderr, "Failed to configure reference block visualizer: %s\n", vpx_codec_error(&decoder));
- return EXIT_FAILURE;
- }
-
- if (vp8_dbg_color_mb_modes
- && vpx_codec_control(&decoder, VP8_SET_DBG_COLOR_MB_MODES, vp8_dbg_color_mb_modes))
- {
- fprintf(stderr, "Failed to configure macro block visualizer: %s\n", vpx_codec_error(&decoder));
- return EXIT_FAILURE;
- }
-
- if (vp8_dbg_color_b_modes
- && vpx_codec_control(&decoder, VP8_SET_DBG_COLOR_B_MODES, vp8_dbg_color_b_modes))
- {
- fprintf(stderr, "Failed to configure block visualizer: %s\n", vpx_codec_error(&decoder));
- return EXIT_FAILURE;
- }
-
- if (vp8_dbg_display_mv
- && vpx_codec_control(&decoder, VP8_SET_DBG_DISPLAY_MV, vp8_dbg_display_mv))
- {
- fprintf(stderr, "Failed to configure motion vector visualizer: %s\n", vpx_codec_error(&decoder));
- return EXIT_FAILURE;
- }
-#endif
-
- /* Decode file */
- while (!read_frame(&input, &buf, &buf_sz, &buf_alloc_sz))
- {
- vpx_codec_iter_t iter = NULL;
- vpx_image_t *img;
- struct vpx_usec_timer timer;
-
- vpx_usec_timer_start(&timer);
-
- if (vpx_codec_decode(&decoder, buf, buf_sz, NULL, 0))
- {
- const char *detail = vpx_codec_error_detail(&decoder);
- fprintf(stderr, "Failed to decode frame: %s\n", vpx_codec_error(&decoder));
-
- if (detail)
- fprintf(stderr, " Additional information: %s\n", detail);
-
- goto fail;
- }
-
- vpx_usec_timer_mark(&timer);
- dx_time += vpx_usec_timer_elapsed(&timer);
-
- ++frame_in;
-
- if ((img = vpx_codec_get_frame(&decoder, &iter)))
- ++frame_out;
-
- if (progress)
- show_progress(frame_in, frame_out, dx_time);
-
- if (!noblit)
- {
- if (img)
- {
- unsigned int y;
- char out_fn[PATH_MAX];
- uint8_t *buf;
-
- if (!single_file)
- {
- size_t len = sizeof(out_fn)-1;
-
- out_fn[len] = '\0';
- generate_filename(outfile_pattern, out_fn, len-1,
- img->d_w, img->d_h, frame_in);
- out = out_open(out_fn, do_md5);
- }
- else if(use_y4m)
- out_put(out, (unsigned char *)"FRAME\n", 6, do_md5);
-
- buf = img->planes[VPX_PLANE_Y];
-
- for (y = 0; y < img->d_h; y++)
- {
- out_put(out, buf, img->d_w, do_md5);
- buf += img->stride[VPX_PLANE_Y];
- }
-
- buf = img->planes[flipuv?VPX_PLANE_V:VPX_PLANE_U];
-
- for (y = 0; y < (1 + img->d_h) / 2; y++)
- {
- out_put(out, buf, (1 + img->d_w) / 2, do_md5);
- buf += img->stride[VPX_PLANE_U];
- }
-
- buf = img->planes[flipuv?VPX_PLANE_U:VPX_PLANE_V];
-
- for (y = 0; y < (1 + img->d_h) / 2; y++)
- {
- out_put(out, buf, (1 + img->d_w) / 2, do_md5);
- buf += img->stride[VPX_PLANE_V];
- }
-
- if (!single_file)
- out_close(out, out_fn, do_md5);
- }
- }
-
- if (stop_after && frame_in >= stop_after)
- break;
- }
-
- if (summary || progress)
- {
- show_progress(frame_in, frame_out, dx_time);
- fprintf(stderr, "\n");
- }
-
-fail:
-
- if (vpx_codec_destroy(&decoder))
- {
- fprintf(stderr, "Failed to destroy decoder: %s\n", vpx_codec_error(&decoder));
- return EXIT_FAILURE;
- }
-
- if (single_file && !noblit)
- out_close(out, outfile, do_md5);
-
- if(input.nestegg_ctx)
- nestegg_destroy(input.nestegg_ctx);
- if(input.kind != WEBM_FILE)
- free(buf);
- fclose(infile);
- free(argv);
-
- return EXIT_SUCCESS;
-}
diff --git a/vpxenc.c b/vpxenc.c
deleted file mode 100755
index 39256b6..0000000
--- a/vpxenc.c
+++ /dev/null
@@ -1,1735 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
-
-
-/* This is a simple program that encodes YV12 files and generates ivf
- * files using the new interface.
- */
-#if defined(_WIN32) || !CONFIG_OS_SUPPORT
-#define USE_POSIX_MMAP 0
-#else
-#define USE_POSIX_MMAP 1
-#endif
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdarg.h>
-#include <string.h>
-#include <limits.h>
-#include "vpx/vpx_encoder.h"
-#if USE_POSIX_MMAP
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <sys/mman.h>
-#include <fcntl.h>
-#include <unistd.h>
-#endif
-#include "vpx_version.h"
-#include "vpx/vp8cx.h"
-#include "vpx_ports/mem_ops.h"
-#include "vpx_ports/vpx_timer.h"
-#include "tools_common.h"
-#include "y4minput.h"
-#include "libmkv/EbmlWriter.h"
-#include "libmkv/EbmlIDs.h"
-
-/* Need special handling of these functions on Windows */
-#if defined(_MSC_VER)
-/* MSVS doesn't define off_t, and uses _f{seek,tell}i64 */
-typedef __int64 off_t;
-#define fseeko _fseeki64
-#define ftello _ftelli64
-#elif defined(_WIN32)
-/* MinGW defines off_t, and uses f{seek,tell}o64 */
-#define fseeko fseeko64
-#define ftello ftello64
-#endif
-
-#if defined(_MSC_VER)
-#define LITERALU64(n) n
-#else
-#define LITERALU64(n) n##LLU
-#endif
-
-/* We should use 32-bit file operations in WebM file format
- * when building ARM executable file (.axf) with RVCT */
-#if !CONFIG_OS_SUPPORT
-typedef long off_t;
-#define fseeko fseek
-#define ftello ftell
-#endif
-
-static const char *exec_name;
-
-static const struct codec_item
-{
- char const *name;
- const vpx_codec_iface_t *iface;
- unsigned int fourcc;
-} codecs[] =
-{
-#if CONFIG_VP8_ENCODER
- {"vp8", &vpx_codec_vp8_cx_algo, 0x30385056},
-#endif
-};
-
-static void usage_exit();
-
-void die(const char *fmt, ...)
-{
- va_list ap;
- va_start(ap, fmt);
- vfprintf(stderr, fmt, ap);
- fprintf(stderr, "\n");
- usage_exit();
-}
-
-static void ctx_exit_on_error(vpx_codec_ctx_t *ctx, const char *s)
-{
- if (ctx->err)
- {
- const char *detail = vpx_codec_error_detail(ctx);
-
- fprintf(stderr, "%s: %s\n", s, vpx_codec_error(ctx));
-
- if (detail)
- fprintf(stderr, " %s\n", detail);
-
- exit(EXIT_FAILURE);
- }
-}
-
-/* This structure is used to abstract the different ways of handling
- * first pass statistics.
- */
-typedef struct
-{
- vpx_fixed_buf_t buf;
- int pass;
- FILE *file;
- char *buf_ptr;
- size_t buf_alloc_sz;
-} stats_io_t;
-
-int stats_open_file(stats_io_t *stats, const char *fpf, int pass)
-{
- int res;
-
- stats->pass = pass;
-
- if (pass == 0)
- {
- stats->file = fopen(fpf, "wb");
- stats->buf.sz = 0;
- stats->buf.buf = NULL,
- res = (stats->file != NULL);
- }
- else
- {
-#if 0
-#elif USE_POSIX_MMAP
- struct stat stat_buf;
- int fd;
-
- fd = open(fpf, O_RDONLY);
- stats->file = fdopen(fd, "rb");
- fstat(fd, &stat_buf);
- stats->buf.sz = stat_buf.st_size;
- stats->buf.buf = mmap(NULL, stats->buf.sz, PROT_READ, MAP_PRIVATE,
- fd, 0);
- res = (stats->buf.buf != NULL);
-#else
- size_t nbytes;
-
- stats->file = fopen(fpf, "rb");
-
- if (fseek(stats->file, 0, SEEK_END))
- {
- fprintf(stderr, "First-pass stats file must be seekable!\n");
- exit(EXIT_FAILURE);
- }
-
- stats->buf.sz = stats->buf_alloc_sz = ftell(stats->file);
- rewind(stats->file);
-
- stats->buf.buf = malloc(stats->buf_alloc_sz);
-
- if (!stats->buf.buf)
- {
- fprintf(stderr, "Failed to allocate first-pass stats buffer (%lu bytes)\n",
- (unsigned long)stats->buf_alloc_sz);
- exit(EXIT_FAILURE);
- }
-
- nbytes = fread(stats->buf.buf, 1, stats->buf.sz, stats->file);
- res = (nbytes == stats->buf.sz);
-#endif
- }
-
- return res;
-}
-
-int stats_open_mem(stats_io_t *stats, int pass)
-{
- int res;
- stats->pass = pass;
-
- if (!pass)
- {
- stats->buf.sz = 0;
- stats->buf_alloc_sz = 64 * 1024;
- stats->buf.buf = malloc(stats->buf_alloc_sz);
- }
-
- stats->buf_ptr = stats->buf.buf;
- res = (stats->buf.buf != NULL);
- return res;
-}
-
-
-void stats_close(stats_io_t *stats, int last_pass)
-{
- if (stats->file)
- {
- if (stats->pass == last_pass)
- {
-#if 0
-#elif USE_POSIX_MMAP
- munmap(stats->buf.buf, stats->buf.sz);
-#else
- free(stats->buf.buf);
-#endif
- }
-
- fclose(stats->file);
- stats->file = NULL;
- }
- else
- {
- if (stats->pass == last_pass)
- free(stats->buf.buf);
- }
-}
-
-void stats_write(stats_io_t *stats, const void *pkt, size_t len)
-{
- if (stats->file)
- {
- if(fwrite(pkt, 1, len, stats->file));
- }
- else
- {
- if (stats->buf.sz + len > stats->buf_alloc_sz)
- {
- size_t new_sz = stats->buf_alloc_sz + 64 * 1024;
- char *new_ptr = realloc(stats->buf.buf, new_sz);
-
- if (new_ptr)
- {
- stats->buf_ptr = new_ptr + (stats->buf_ptr - (char *)stats->buf.buf);
- stats->buf.buf = new_ptr;
- stats->buf_alloc_sz = new_sz;
- }
- else
- {
- fprintf(stderr,
- "\nFailed to realloc firstpass stats buffer.\n");
- exit(EXIT_FAILURE);
- }
- }
-
- memcpy(stats->buf_ptr, pkt, len);
- stats->buf.sz += len;
- stats->buf_ptr += len;
- }
-}
-
-vpx_fixed_buf_t stats_get(stats_io_t *stats)
-{
- return stats->buf;
-}
-
-enum video_file_type
-{
- FILE_TYPE_RAW,
- FILE_TYPE_IVF,
- FILE_TYPE_Y4M
-};
-
-struct detect_buffer {
- char buf[4];
- size_t buf_read;
- size_t position;
-};
-
-
-#define IVF_FRAME_HDR_SZ (4+8) /* 4 byte size + 8 byte timestamp */
-static int read_frame(FILE *f, vpx_image_t *img, unsigned int file_type,
- y4m_input *y4m, struct detect_buffer *detect)
-{
- int plane = 0;
- int shortread = 0;
-
- if (file_type == FILE_TYPE_Y4M)
- {
- if (y4m_input_fetch_frame(y4m, f, img) < 1)
- return 0;
- }
- else
- {
- if (file_type == FILE_TYPE_IVF)
- {
- char junk[IVF_FRAME_HDR_SZ];
-
- /* Skip the frame header. We know how big the frame should be. See
- * write_ivf_frame_header() for documentation on the frame header
- * layout.
- */
- if(fread(junk, 1, IVF_FRAME_HDR_SZ, f));
- }
-
- for (plane = 0; plane < 3; plane++)
- {
- unsigned char *ptr;
- int w = (plane ? (1 + img->d_w) / 2 : img->d_w);
- int h = (plane ? (1 + img->d_h) / 2 : img->d_h);
- int r;
-
- /* Determine the correct plane based on the image format. The for-loop
- * always counts in Y,U,V order, but this may not match the order of
- * the data on disk.
- */
- switch (plane)
- {
- case 1:
- ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12? VPX_PLANE_V : VPX_PLANE_U];
- break;
- case 2:
- ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12?VPX_PLANE_U : VPX_PLANE_V];
- break;
- default:
- ptr = img->planes[plane];
- }
-
- for (r = 0; r < h; r++)
- {
- size_t needed = w;
- size_t buf_position = 0;
- const size_t left = detect->buf_read - detect->position;
- if (left > 0)
- {
- const size_t more = (left < needed) ? left : needed;
- memcpy(ptr, detect->buf + detect->position, more);
- buf_position = more;
- needed -= more;
- detect->position += more;
- }
- if (needed > 0)
- {
- shortread |= (fread(ptr + buf_position, 1, needed, f) < needed);
- }
-
- ptr += img->stride[plane];
- }
- }
- }
-
- return !shortread;
-}
-
-
-unsigned int file_is_y4m(FILE *infile,
- y4m_input *y4m,
- char detect[4])
-{
- if(memcmp(detect, "YUV4", 4) == 0)
- {
- return 1;
- }
- return 0;
-}
-
-#define IVF_FILE_HDR_SZ (32)
-unsigned int file_is_ivf(FILE *infile,
- unsigned int *fourcc,
- unsigned int *width,
- unsigned int *height,
- struct detect_buffer *detect)
-{
- char raw_hdr[IVF_FILE_HDR_SZ];
- int is_ivf = 0;
-
- if(memcmp(detect->buf, "DKIF", 4) != 0)
- return 0;
-
- /* See write_ivf_file_header() for more documentation on the file header
- * layout.
- */
- if (fread(raw_hdr + 4, 1, IVF_FILE_HDR_SZ - 4, infile)
- == IVF_FILE_HDR_SZ - 4)
- {
- {
- is_ivf = 1;
-
- if (mem_get_le16(raw_hdr + 4) != 0)
- fprintf(stderr, "Error: Unrecognized IVF version! This file may not"
- " decode properly.");
-
- *fourcc = mem_get_le32(raw_hdr + 8);
- }
- }
-
- if (is_ivf)
- {
- *width = mem_get_le16(raw_hdr + 12);
- *height = mem_get_le16(raw_hdr + 14);
- detect->position = 4;
- }
-
- return is_ivf;
-}
-
-
-static void write_ivf_file_header(FILE *outfile,
- const vpx_codec_enc_cfg_t *cfg,
- unsigned int fourcc,
- int frame_cnt)
-{
- char header[32];
-
- if (cfg->g_pass != VPX_RC_ONE_PASS && cfg->g_pass != VPX_RC_LAST_PASS)
- return;
-
- header[0] = 'D';
- header[1] = 'K';
- header[2] = 'I';
- header[3] = 'F';
- mem_put_le16(header + 4, 0); /* version */
- mem_put_le16(header + 6, 32); /* headersize */
- mem_put_le32(header + 8, fourcc); /* headersize */
- mem_put_le16(header + 12, cfg->g_w); /* width */
- mem_put_le16(header + 14, cfg->g_h); /* height */
- mem_put_le32(header + 16, cfg->g_timebase.den); /* rate */
- mem_put_le32(header + 20, cfg->g_timebase.num); /* scale */
- mem_put_le32(header + 24, frame_cnt); /* length */
- mem_put_le32(header + 28, 0); /* unused */
-
- if(fwrite(header, 1, 32, outfile));
-}
-
-
-static void write_ivf_frame_header(FILE *outfile,
- const vpx_codec_cx_pkt_t *pkt)
-{
- char header[12];
- vpx_codec_pts_t pts;
-
- if (pkt->kind != VPX_CODEC_CX_FRAME_PKT)
- return;
-
- pts = pkt->data.frame.pts;
- mem_put_le32(header, pkt->data.frame.sz);
- mem_put_le32(header + 4, pts & 0xFFFFFFFF);
- mem_put_le32(header + 8, pts >> 32);
-
- if(fwrite(header, 1, 12, outfile));
-}
-
-
-typedef off_t EbmlLoc;
-
-
-struct cue_entry
-{
- unsigned int time;
- uint64_t loc;
-};
-
-
-struct EbmlGlobal
-{
- int debug;
-
- FILE *stream;
- int64_t last_pts_ms;
- vpx_rational_t framerate;
-
- /* These pointers are to the start of an element */
- off_t position_reference;
- off_t seek_info_pos;
- off_t segment_info_pos;
- off_t track_pos;
- off_t cue_pos;
- off_t cluster_pos;
-
- /* This pointer is to a specific element to be serialized */
- off_t track_id_pos;
-
- /* These pointers are to the size field of the element */
- EbmlLoc startSegment;
- EbmlLoc startCluster;
-
- uint32_t cluster_timecode;
- int cluster_open;
-
- struct cue_entry *cue_list;
- unsigned int cues;
-
-};
-
-
-void Ebml_Write(EbmlGlobal *glob, const void *buffer_in, unsigned long len)
-{
- if(fwrite(buffer_in, 1, len, glob->stream));
-}
-
-
-void Ebml_Serialize(EbmlGlobal *glob, const void *buffer_in, unsigned long len)
-{
- const unsigned char *q = (const unsigned char *)buffer_in + len - 1;
-
- for(; len; len--)
- Ebml_Write(glob, q--, 1);
-}
-
-
-/* Need a fixed size serializer for the track ID. libmkv provdes a 64 bit
- * one, but not a 32 bit one.
- */
-static void Ebml_SerializeUnsigned32(EbmlGlobal *glob, unsigned long class_id, uint64_t ui)
-{
- unsigned char sizeSerialized = 4 | 0x80;
- Ebml_WriteID(glob, class_id);
- Ebml_Serialize(glob, &sizeSerialized, 1);
- Ebml_Serialize(glob, &ui, 4);
-}
-
-
-static void
-Ebml_StartSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc,
- unsigned long class_id)
-{
- //todo this is always taking 8 bytes, this may need later optimization
- //this is a key that says lenght unknown
- unsigned long long unknownLen = LITERALU64(0x01FFFFFFFFFFFFFF);
-
- Ebml_WriteID(glob, class_id);
- *ebmlLoc = ftello(glob->stream);
- Ebml_Serialize(glob, &unknownLen, 8);
-}
-
-static void
-Ebml_EndSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc)
-{
- off_t pos;
- uint64_t size;
-
- /* Save the current stream pointer */
- pos = ftello(glob->stream);
-
- /* Calculate the size of this element */
- size = pos - *ebmlLoc - 8;
- size |= LITERALU64(0x0100000000000000);
-
- /* Seek back to the beginning of the element and write the new size */
- fseeko(glob->stream, *ebmlLoc, SEEK_SET);
- Ebml_Serialize(glob, &size, 8);
-
- /* Reset the stream pointer */
- fseeko(glob->stream, pos, SEEK_SET);
-}
-
-
-static void
-write_webm_seek_element(EbmlGlobal *ebml, unsigned long id, off_t pos)
-{
- uint64_t offset = pos - ebml->position_reference;
- EbmlLoc start;
- Ebml_StartSubElement(ebml, &start, Seek);
- Ebml_SerializeBinary(ebml, SeekID, id);
- Ebml_SerializeUnsigned64(ebml, SeekPosition, offset);
- Ebml_EndSubElement(ebml, &start);
-}
-
-
-static void
-write_webm_seek_info(EbmlGlobal *ebml)
-{
-
- off_t pos;
-
- /* Save the current stream pointer */
- pos = ftello(ebml->stream);
-
- if(ebml->seek_info_pos)
- fseeko(ebml->stream, ebml->seek_info_pos, SEEK_SET);
- else
- ebml->seek_info_pos = pos;
-
- {
- EbmlLoc start;
-
- Ebml_StartSubElement(ebml, &start, SeekHead);
- write_webm_seek_element(ebml, Tracks, ebml->track_pos);
- write_webm_seek_element(ebml, Cues, ebml->cue_pos);
- write_webm_seek_element(ebml, Info, ebml->segment_info_pos);
- Ebml_EndSubElement(ebml, &start);
- }
- {
- //segment info
- EbmlLoc startInfo;
- uint64_t frame_time;
-
- frame_time = (uint64_t)1000 * ebml->framerate.den
- / ebml->framerate.num;
- ebml->segment_info_pos = ftello(ebml->stream);
- Ebml_StartSubElement(ebml, &startInfo, Info);
- Ebml_SerializeUnsigned(ebml, TimecodeScale, 1000000);
- Ebml_SerializeFloat(ebml, Segment_Duration,
- ebml->last_pts_ms + frame_time);
- Ebml_SerializeString(ebml, 0x4D80,
- ebml->debug ? "vpxenc" : "vpxenc" VERSION_STRING);
- Ebml_SerializeString(ebml, 0x5741,
- ebml->debug ? "vpxenc" : "vpxenc" VERSION_STRING);
- Ebml_EndSubElement(ebml, &startInfo);
- }
-}
-
-
-static void
-write_webm_file_header(EbmlGlobal *glob,
- const vpx_codec_enc_cfg_t *cfg,
- const struct vpx_rational *fps)
-{
- {
- EbmlLoc start;
- Ebml_StartSubElement(glob, &start, EBML);
- Ebml_SerializeUnsigned(glob, EBMLVersion, 1);
- Ebml_SerializeUnsigned(glob, EBMLReadVersion, 1); //EBML Read Version
- Ebml_SerializeUnsigned(glob, EBMLMaxIDLength, 4); //EBML Max ID Length
- Ebml_SerializeUnsigned(glob, EBMLMaxSizeLength, 8); //EBML Max Size Length
- Ebml_SerializeString(glob, DocType, "webm"); //Doc Type
- Ebml_SerializeUnsigned(glob, DocTypeVersion, 2); //Doc Type Version
- Ebml_SerializeUnsigned(glob, DocTypeReadVersion, 2); //Doc Type Read Version
- Ebml_EndSubElement(glob, &start);
- }
- {
- Ebml_StartSubElement(glob, &glob->startSegment, Segment); //segment
- glob->position_reference = ftello(glob->stream);
- glob->framerate = *fps;
- write_webm_seek_info(glob);
-
- {
- EbmlLoc trackStart;
- glob->track_pos = ftello(glob->stream);
- Ebml_StartSubElement(glob, &trackStart, Tracks);
- {
- unsigned int trackNumber = 1;
- uint64_t trackID = 0;
-
- EbmlLoc start;
- Ebml_StartSubElement(glob, &start, TrackEntry);
- Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber);
- glob->track_id_pos = ftello(glob->stream);
- Ebml_SerializeUnsigned32(glob, TrackUID, trackID);
- Ebml_SerializeUnsigned(glob, TrackType, 1); //video is always 1
- Ebml_SerializeString(glob, CodecID, "V_VP8");
- {
- unsigned int pixelWidth = cfg->g_w;
- unsigned int pixelHeight = cfg->g_h;
- float frameRate = (float)fps->num/(float)fps->den;
-
- EbmlLoc videoStart;
- Ebml_StartSubElement(glob, &videoStart, Video);
- Ebml_SerializeUnsigned(glob, PixelWidth, pixelWidth);
- Ebml_SerializeUnsigned(glob, PixelHeight, pixelHeight);
- Ebml_SerializeFloat(glob, FrameRate, frameRate);
- Ebml_EndSubElement(glob, &videoStart); //Video
- }
- Ebml_EndSubElement(glob, &start); //Track Entry
- }
- Ebml_EndSubElement(glob, &trackStart);
- }
- // segment element is open
- }
-}
-
-
-static void
-write_webm_block(EbmlGlobal *glob,
- const vpx_codec_enc_cfg_t *cfg,
- const vpx_codec_cx_pkt_t *pkt)
-{
- unsigned long block_length;
- unsigned char track_number;
- unsigned short block_timecode = 0;
- unsigned char flags;
- int64_t pts_ms;
- int start_cluster = 0, is_keyframe;
-
- /* Calculate the PTS of this frame in milliseconds */
- pts_ms = pkt->data.frame.pts * 1000
- * (uint64_t)cfg->g_timebase.num / (uint64_t)cfg->g_timebase.den;
- if(pts_ms <= glob->last_pts_ms)
- pts_ms = glob->last_pts_ms + 1;
- glob->last_pts_ms = pts_ms;
-
- /* Calculate the relative time of this block */
- if(pts_ms - glob->cluster_timecode > SHRT_MAX)
- start_cluster = 1;
- else
- block_timecode = pts_ms - glob->cluster_timecode;
-
- is_keyframe = (pkt->data.frame.flags & VPX_FRAME_IS_KEY);
- if(start_cluster || is_keyframe)
- {
- if(glob->cluster_open)
- Ebml_EndSubElement(glob, &glob->startCluster);
-
- /* Open the new cluster */
- block_timecode = 0;
- glob->cluster_open = 1;
- glob->cluster_timecode = pts_ms;
- glob->cluster_pos = ftello(glob->stream);
- Ebml_StartSubElement(glob, &glob->startCluster, Cluster); //cluster
- Ebml_SerializeUnsigned(glob, Timecode, glob->cluster_timecode);
-
- /* Save a cue point if this is a keyframe. */
- if(is_keyframe)
- {
- struct cue_entry *cue, *new_cue_list;
-
- new_cue_list = realloc(glob->cue_list,
- (glob->cues+1) * sizeof(struct cue_entry));
- if(new_cue_list)
- glob->cue_list = new_cue_list;
- else
- {
- fprintf(stderr, "\nFailed to realloc cue list.\n");
- exit(EXIT_FAILURE);
- }
-
- cue = &glob->cue_list[glob->cues];
- cue->time = glob->cluster_timecode;
- cue->loc = glob->cluster_pos;
- glob->cues++;
- }
- }
-
- /* Write the Simple Block */
- Ebml_WriteID(glob, SimpleBlock);
-
- block_length = pkt->data.frame.sz + 4;
- block_length |= 0x10000000;
- Ebml_Serialize(glob, &block_length, 4);
-
- track_number = 1;
- track_number |= 0x80;
- Ebml_Write(glob, &track_number, 1);
-
- Ebml_Serialize(glob, &block_timecode, 2);
-
- flags = 0;
- if(is_keyframe)
- flags |= 0x80;
- if(pkt->data.frame.flags & VPX_FRAME_IS_INVISIBLE)
- flags |= 0x08;
- Ebml_Write(glob, &flags, 1);
-
- Ebml_Write(glob, pkt->data.frame.buf, pkt->data.frame.sz);
-}
-
-
-static void
-write_webm_file_footer(EbmlGlobal *glob, long hash)
-{
-
- if(glob->cluster_open)
- Ebml_EndSubElement(glob, &glob->startCluster);
-
- {
- EbmlLoc start;
- int i;
-
- glob->cue_pos = ftello(glob->stream);
- Ebml_StartSubElement(glob, &start, Cues);
- for(i=0; i<glob->cues; i++)
- {
- struct cue_entry *cue = &glob->cue_list[i];
- EbmlLoc start;
-
- Ebml_StartSubElement(glob, &start, CuePoint);
- {
- EbmlLoc start;
-
- Ebml_SerializeUnsigned(glob, CueTime, cue->time);
-
- Ebml_StartSubElement(glob, &start, CueTrackPositions);
- Ebml_SerializeUnsigned(glob, CueTrack, 1);
- Ebml_SerializeUnsigned64(glob, CueClusterPosition,
- cue->loc - glob->position_reference);
- //Ebml_SerializeUnsigned(glob, CueBlockNumber, cue->blockNumber);
- Ebml_EndSubElement(glob, &start);
- }
- Ebml_EndSubElement(glob, &start);
- }
- Ebml_EndSubElement(glob, &start);
- }
-
- Ebml_EndSubElement(glob, &glob->startSegment);
-
- /* Patch up the seek info block */
- write_webm_seek_info(glob);
-
- /* Patch up the track id */
- fseeko(glob->stream, glob->track_id_pos, SEEK_SET);
- Ebml_SerializeUnsigned32(glob, TrackUID, glob->debug ? 0xDEADBEEF : hash);
-
- fseeko(glob->stream, 0, SEEK_END);
-}
-
-
-/* Murmur hash derived from public domain reference implementation at
- * http://sites.google.com/site/murmurhash/
- */
-static unsigned int murmur ( const void * key, int len, unsigned int seed )
-{
- const unsigned int m = 0x5bd1e995;
- const int r = 24;
-
- unsigned int h = seed ^ len;
-
- const unsigned char * data = (const unsigned char *)key;
-
- while(len >= 4)
- {
- unsigned int k;
-
- k = data[0];
- k |= data[1] << 8;
- k |= data[2] << 16;
- k |= data[3] << 24;
-
- k *= m;
- k ^= k >> r;
- k *= m;
-
- h *= m;
- h ^= k;
-
- data += 4;
- len -= 4;
- }
-
- switch(len)
- {
- case 3: h ^= data[2] << 16;
- case 2: h ^= data[1] << 8;
- case 1: h ^= data[0];
- h *= m;
- };
-
- h ^= h >> 13;
- h *= m;
- h ^= h >> 15;
-
- return h;
-}
-
-#include "math.h"
-
-static double vp8_mse2psnr(double Samples, double Peak, double Mse)
-{
- double psnr;
-
- if ((double)Mse > 0.0)
- psnr = 10.0 * log10(Peak * Peak * Samples / Mse);
- else
- psnr = 60; // Limit to prevent / 0
-
- if (psnr > 60)
- psnr = 60;
-
- return psnr;
-}
-
-
-#include "args.h"
-
-static const arg_def_t debugmode = ARG_DEF("D", "debug", 0,
- "Debug mode (makes output deterministic)");
-static const arg_def_t outputfile = ARG_DEF("o", "output", 1,
- "Output filename");
-static const arg_def_t use_yv12 = ARG_DEF(NULL, "yv12", 0,
- "Input file is YV12 ");
-static const arg_def_t use_i420 = ARG_DEF(NULL, "i420", 0,
- "Input file is I420 (default)");
-static const arg_def_t codecarg = ARG_DEF(NULL, "codec", 1,
- "Codec to use");
-static const arg_def_t passes = ARG_DEF("p", "passes", 1,
- "Number of passes (1/2)");
-static const arg_def_t pass_arg = ARG_DEF(NULL, "pass", 1,
- "Pass to execute (1/2)");
-static const arg_def_t fpf_name = ARG_DEF(NULL, "fpf", 1,
- "First pass statistics file name");
-static const arg_def_t limit = ARG_DEF(NULL, "limit", 1,
- "Stop encoding after n input frames");
-static const arg_def_t deadline = ARG_DEF("d", "deadline", 1,
- "Deadline per frame (usec)");
-static const arg_def_t best_dl = ARG_DEF(NULL, "best", 0,
- "Use Best Quality Deadline");
-static const arg_def_t good_dl = ARG_DEF(NULL, "good", 0,
- "Use Good Quality Deadline");
-static const arg_def_t rt_dl = ARG_DEF(NULL, "rt", 0,
- "Use Realtime Quality Deadline");
-static const arg_def_t verbosearg = ARG_DEF("v", "verbose", 0,
- "Show encoder parameters");
-static const arg_def_t psnrarg = ARG_DEF(NULL, "psnr", 0,
- "Show PSNR in status line");
-static const arg_def_t framerate = ARG_DEF(NULL, "fps", 1,
- "Stream frame rate (rate/scale)");
-static const arg_def_t use_ivf = ARG_DEF(NULL, "ivf", 0,
- "Output IVF (default is WebM)");
-static const arg_def_t *main_args[] =
-{
- &debugmode,
- &outputfile, &codecarg, &passes, &pass_arg, &fpf_name, &limit, &deadline,
- &best_dl, &good_dl, &rt_dl,
- &verbosearg, &psnrarg, &use_ivf, &framerate,
- NULL
-};
-
-static const arg_def_t usage = ARG_DEF("u", "usage", 1,
- "Usage profile number to use");
-static const arg_def_t threads = ARG_DEF("t", "threads", 1,
- "Max number of threads to use");
-static const arg_def_t profile = ARG_DEF(NULL, "profile", 1,
- "Bitstream profile number to use");
-static const arg_def_t width = ARG_DEF("w", "width", 1,
- "Frame width");
-static const arg_def_t height = ARG_DEF("h", "height", 1,
- "Frame height");
-static const arg_def_t timebase = ARG_DEF(NULL, "timebase", 1,
- "Stream timebase (frame duration)");
-static const arg_def_t error_resilient = ARG_DEF(NULL, "error-resilient", 1,
- "Enable error resiliency features");
-static const arg_def_t lag_in_frames = ARG_DEF(NULL, "lag-in-frames", 1,
- "Max number of frames to lag");
-
-static const arg_def_t *global_args[] =
-{
- &use_yv12, &use_i420, &usage, &threads, &profile,
- &width, &height, &timebase, &framerate, &error_resilient,
- &lag_in_frames, NULL
-};
-
-static const arg_def_t dropframe_thresh = ARG_DEF(NULL, "drop-frame", 1,
- "Temporal resampling threshold (buf %)");
-static const arg_def_t resize_allowed = ARG_DEF(NULL, "resize-allowed", 1,
- "Spatial resampling enabled (bool)");
-static const arg_def_t resize_up_thresh = ARG_DEF(NULL, "resize-up", 1,
- "Upscale threshold (buf %)");
-static const arg_def_t resize_down_thresh = ARG_DEF(NULL, "resize-down", 1,
- "Downscale threshold (buf %)");
-static const struct arg_enum_list end_usage_enum[] = {
- {"vbr", VPX_VBR},
- {"cbr", VPX_CBR},
- {"cq", VPX_CQ},
- {NULL, 0}
-};
-static const arg_def_t end_usage = ARG_DEF_ENUM(NULL, "end-usage", 1,
- "Rate control mode", end_usage_enum);
-static const arg_def_t target_bitrate = ARG_DEF(NULL, "target-bitrate", 1,
- "Bitrate (kbps)");
-static const arg_def_t min_quantizer = ARG_DEF(NULL, "min-q", 1,
- "Minimum (best) quantizer");
-static const arg_def_t max_quantizer = ARG_DEF(NULL, "max-q", 1,
- "Maximum (worst) quantizer");
-static const arg_def_t undershoot_pct = ARG_DEF(NULL, "undershoot-pct", 1,
- "Datarate undershoot (min) target (%)");
-static const arg_def_t overshoot_pct = ARG_DEF(NULL, "overshoot-pct", 1,
- "Datarate overshoot (max) target (%)");
-static const arg_def_t buf_sz = ARG_DEF(NULL, "buf-sz", 1,
- "Client buffer size (ms)");
-static const arg_def_t buf_initial_sz = ARG_DEF(NULL, "buf-initial-sz", 1,
- "Client initial buffer size (ms)");
-static const arg_def_t buf_optimal_sz = ARG_DEF(NULL, "buf-optimal-sz", 1,
- "Client optimal buffer size (ms)");
-static const arg_def_t *rc_args[] =
-{
- &dropframe_thresh, &resize_allowed, &resize_up_thresh, &resize_down_thresh,
- &end_usage, &target_bitrate, &min_quantizer, &max_quantizer,
- &undershoot_pct, &overshoot_pct, &buf_sz, &buf_initial_sz, &buf_optimal_sz,
- NULL
-};
-
-
-static const arg_def_t bias_pct = ARG_DEF(NULL, "bias-pct", 1,
- "CBR/VBR bias (0=CBR, 100=VBR)");
-static const arg_def_t minsection_pct = ARG_DEF(NULL, "minsection-pct", 1,
- "GOP min bitrate (% of target)");
-static const arg_def_t maxsection_pct = ARG_DEF(NULL, "maxsection-pct", 1,
- "GOP max bitrate (% of target)");
-static const arg_def_t *rc_twopass_args[] =
-{
- &bias_pct, &minsection_pct, &maxsection_pct, NULL
-};
-
-
-static const arg_def_t kf_min_dist = ARG_DEF(NULL, "kf-min-dist", 1,
- "Minimum keyframe interval (frames)");
-static const arg_def_t kf_max_dist = ARG_DEF(NULL, "kf-max-dist", 1,
- "Maximum keyframe interval (frames)");
-static const arg_def_t kf_disabled = ARG_DEF(NULL, "disable-kf", 0,
- "Disable keyframe placement");
-static const arg_def_t *kf_args[] =
-{
- &kf_min_dist, &kf_max_dist, &kf_disabled, NULL
-};
-
-
-#if CONFIG_VP8_ENCODER
-static const arg_def_t noise_sens = ARG_DEF(NULL, "noise-sensitivity", 1,
- "Noise sensitivity (frames to blur)");
-static const arg_def_t sharpness = ARG_DEF(NULL, "sharpness", 1,
- "Filter sharpness (0-7)");
-static const arg_def_t static_thresh = ARG_DEF(NULL, "static-thresh", 1,
- "Motion detection threshold");
-#endif
-
-#if CONFIG_VP8_ENCODER
-static const arg_def_t cpu_used = ARG_DEF(NULL, "cpu-used", 1,
- "CPU Used (-16..16)");
-#endif
-
-
-#if CONFIG_VP8_ENCODER
-static const arg_def_t token_parts = ARG_DEF(NULL, "token-parts", 1,
- "Number of token partitions to use, log2");
-static const arg_def_t auto_altref = ARG_DEF(NULL, "auto-alt-ref", 1,
- "Enable automatic alt reference frames");
-static const arg_def_t arnr_maxframes = ARG_DEF(NULL, "arnr-maxframes", 1,
- "AltRef Max Frames");
-static const arg_def_t arnr_strength = ARG_DEF(NULL, "arnr-strength", 1,
- "AltRef Strength");
-static const arg_def_t arnr_type = ARG_DEF(NULL, "arnr-type", 1,
- "AltRef Type");
-static const struct arg_enum_list tuning_enum[] = {
- {"psnr", VP8_TUNE_PSNR},
- {"ssim", VP8_TUNE_SSIM},
- {NULL, 0}
-};
-static const arg_def_t tune_ssim = ARG_DEF_ENUM(NULL, "tune", 1,
- "Material to favor", tuning_enum);
-static const arg_def_t cq_level = ARG_DEF(NULL, "cq-level", 1,
- "Constrained Quality Level");
-
-static const arg_def_t *vp8_args[] =
-{
- &cpu_used, &auto_altref, &noise_sens, &sharpness, &static_thresh,
- &token_parts, &arnr_maxframes, &arnr_strength, &arnr_type,
- &tune_ssim, &cq_level, NULL
-};
-static const int vp8_arg_ctrl_map[] =
-{
- VP8E_SET_CPUUSED, VP8E_SET_ENABLEAUTOALTREF,
- VP8E_SET_NOISE_SENSITIVITY, VP8E_SET_SHARPNESS, VP8E_SET_STATIC_THRESHOLD,
- VP8E_SET_TOKEN_PARTITIONS,
- VP8E_SET_ARNR_MAXFRAMES, VP8E_SET_ARNR_STRENGTH , VP8E_SET_ARNR_TYPE,
- VP8E_SET_TUNING, VP8E_SET_CQ_LEVEL, 0
-};
-#endif
-
-static const arg_def_t *no_args[] = { NULL };
-
-static void usage_exit()
-{
- int i;
-
- fprintf(stderr, "Usage: %s <options> -o dst_filename src_filename \n",
- exec_name);
-
- fprintf(stderr, "\nOptions:\n");
- arg_show_usage(stdout, main_args);
- fprintf(stderr, "\nEncoder Global Options:\n");
- arg_show_usage(stdout, global_args);
- fprintf(stderr, "\nRate Control Options:\n");
- arg_show_usage(stdout, rc_args);
- fprintf(stderr, "\nTwopass Rate Control Options:\n");
- arg_show_usage(stdout, rc_twopass_args);
- fprintf(stderr, "\nKeyframe Placement Options:\n");
- arg_show_usage(stdout, kf_args);
-#if CONFIG_VP8_ENCODER
- fprintf(stderr, "\nVP8 Specific Options:\n");
- arg_show_usage(stdout, vp8_args);
-#endif
- fprintf(stderr, "\n"
- "Included encoders:\n"
- "\n");
-
- for (i = 0; i < sizeof(codecs) / sizeof(codecs[0]); i++)
- fprintf(stderr, " %-6s - %s\n",
- codecs[i].name,
- vpx_codec_iface_name(codecs[i].iface));
-
- exit(EXIT_FAILURE);
-}
-
-#define ARG_CTRL_CNT_MAX 10
-
-
-int main(int argc, const char **argv_)
-{
- vpx_codec_ctx_t encoder;
- const char *in_fn = NULL, *out_fn = NULL, *stats_fn = NULL;
- int i;
- FILE *infile, *outfile;
- vpx_codec_enc_cfg_t cfg;
- vpx_codec_err_t res;
- int pass, one_pass_only = 0;
- stats_io_t stats;
- vpx_image_t raw;
- const struct codec_item *codec = codecs;
- int frame_avail, got_data;
-
- struct arg arg;
- char **argv, **argi, **argj;
- int arg_usage = 0, arg_passes = 1, arg_deadline = 0;
- int arg_ctrls[ARG_CTRL_CNT_MAX][2], arg_ctrl_cnt = 0;
- int arg_limit = 0;
- static const arg_def_t **ctrl_args = no_args;
- static const int *ctrl_args_map = NULL;
- int verbose = 0, show_psnr = 0;
- int arg_use_i420 = 1;
- unsigned long cx_time = 0;
- unsigned int file_type, fourcc;
- y4m_input y4m;
- struct vpx_rational arg_framerate = {30, 1};
- int arg_have_framerate = 0;
- int write_webm = 1;
- EbmlGlobal ebml = {0};
- uint32_t hash = 0;
- uint64_t psnr_sse_total = 0;
- uint64_t psnr_samples_total = 0;
- double psnr_totals[4] = {0, 0, 0, 0};
- int psnr_count = 0;
-
- exec_name = argv_[0];
- ebml.last_pts_ms = -1;
-
- if (argc < 3)
- usage_exit();
-
-
- /* First parse the codec and usage values, because we want to apply other
- * parameters on top of the default configuration provided by the codec.
- */
- argv = argv_dup(argc - 1, argv_ + 1);
-
- for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step)
- {
- arg.argv_step = 1;
-
- if (arg_match(&arg, &codecarg, argi))
- {
- int j, k = -1;
-
- for (j = 0; j < sizeof(codecs) / sizeof(codecs[0]); j++)
- if (!strcmp(codecs[j].name, arg.val))
- k = j;
-
- if (k >= 0)
- codec = codecs + k;
- else
- die("Error: Unrecognized argument (%s) to --codec\n",
- arg.val);
-
- }
- else if (arg_match(&arg, &passes, argi))
- {
- arg_passes = arg_parse_uint(&arg);
-
- if (arg_passes < 1 || arg_passes > 2)
- die("Error: Invalid number of passes (%d)\n", arg_passes);
- }
- else if (arg_match(&arg, &pass_arg, argi))
- {
- one_pass_only = arg_parse_uint(&arg);
-
- if (one_pass_only < 1 || one_pass_only > 2)
- die("Error: Invalid pass selected (%d)\n", one_pass_only);
- }
- else if (arg_match(&arg, &fpf_name, argi))
- stats_fn = arg.val;
- else if (arg_match(&arg, &usage, argi))
- arg_usage = arg_parse_uint(&arg);
- else if (arg_match(&arg, &deadline, argi))
- arg_deadline = arg_parse_uint(&arg);
- else if (arg_match(&arg, &best_dl, argi))
- arg_deadline = VPX_DL_BEST_QUALITY;
- else if (arg_match(&arg, &good_dl, argi))
- arg_deadline = VPX_DL_GOOD_QUALITY;
- else if (arg_match(&arg, &rt_dl, argi))
- arg_deadline = VPX_DL_REALTIME;
- else if (arg_match(&arg, &use_yv12, argi))
- {
- arg_use_i420 = 0;
- }
- else if (arg_match(&arg, &use_i420, argi))
- {
- arg_use_i420 = 1;
- }
- else if (arg_match(&arg, &verbosearg, argi))
- verbose = 1;
- else if (arg_match(&arg, &limit, argi))
- arg_limit = arg_parse_uint(&arg);
- else if (arg_match(&arg, &psnrarg, argi))
- show_psnr = 1;
- else if (arg_match(&arg, &framerate, argi))
- {
- arg_framerate = arg_parse_rational(&arg);
- arg_have_framerate = 1;
- }
- else if (arg_match(&arg, &use_ivf, argi))
- write_webm = 0;
- else if (arg_match(&arg, &outputfile, argi))
- out_fn = arg.val;
- else if (arg_match(&arg, &debugmode, argi))
- ebml.debug = 1;
- else
- argj++;
- }
-
- /* Ensure that --passes and --pass are consistent. If --pass is set and --passes=2,
- * ensure --fpf was set.
- */
- if (one_pass_only)
- {
- /* DWIM: Assume the user meant passes=2 if pass=2 is specified */
- if (one_pass_only > arg_passes)
- {
- fprintf(stderr, "Warning: Assuming --pass=%d implies --passes=%d\n",
- one_pass_only, one_pass_only);
- arg_passes = one_pass_only;
- }
-
- if (arg_passes == 2 && !stats_fn)
- die("Must specify --fpf when --pass=%d and --passes=2\n", one_pass_only);
- }
-
- /* Populate encoder configuration */
- res = vpx_codec_enc_config_default(codec->iface, &cfg, arg_usage);
-
- if (res)
- {
- fprintf(stderr, "Failed to get config: %s\n",
- vpx_codec_err_to_string(res));
- return EXIT_FAILURE;
- }
-
- /* Change the default timebase to a high enough value so that the encoder
- * will always create strictly increasing timestamps.
- */
- cfg.g_timebase.den = 1000;
-
- /* Never use the library's default resolution, require it be parsed
- * from the file or set on the command line.
- */
- cfg.g_w = 0;
- cfg.g_h = 0;
-
- /* Now parse the remainder of the parameters. */
- for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step)
- {
- arg.argv_step = 1;
-
- if (0);
- else if (arg_match(&arg, &threads, argi))
- cfg.g_threads = arg_parse_uint(&arg);
- else if (arg_match(&arg, &profile, argi))
- cfg.g_profile = arg_parse_uint(&arg);
- else if (arg_match(&arg, &width, argi))
- cfg.g_w = arg_parse_uint(&arg);
- else if (arg_match(&arg, &height, argi))
- cfg.g_h = arg_parse_uint(&arg);
- else if (arg_match(&arg, &timebase, argi))
- cfg.g_timebase = arg_parse_rational(&arg);
- else if (arg_match(&arg, &error_resilient, argi))
- cfg.g_error_resilient = arg_parse_uint(&arg);
- else if (arg_match(&arg, &lag_in_frames, argi))
- cfg.g_lag_in_frames = arg_parse_uint(&arg);
- else if (arg_match(&arg, &dropframe_thresh, argi))
- cfg.rc_dropframe_thresh = arg_parse_uint(&arg);
- else if (arg_match(&arg, &resize_allowed, argi))
- cfg.rc_resize_allowed = arg_parse_uint(&arg);
- else if (arg_match(&arg, &resize_up_thresh, argi))
- cfg.rc_resize_up_thresh = arg_parse_uint(&arg);
- else if (arg_match(&arg, &resize_down_thresh, argi))
- cfg.rc_resize_down_thresh = arg_parse_uint(&arg);
- else if (arg_match(&arg, &resize_down_thresh, argi))
- cfg.rc_resize_down_thresh = arg_parse_uint(&arg);
- else if (arg_match(&arg, &end_usage, argi))
- cfg.rc_end_usage = arg_parse_enum_or_int(&arg);
- else if (arg_match(&arg, &target_bitrate, argi))
- cfg.rc_target_bitrate = arg_parse_uint(&arg);
- else if (arg_match(&arg, &min_quantizer, argi))
- cfg.rc_min_quantizer = arg_parse_uint(&arg);
- else if (arg_match(&arg, &max_quantizer, argi))
- cfg.rc_max_quantizer = arg_parse_uint(&arg);
- else if (arg_match(&arg, &undershoot_pct, argi))
- cfg.rc_undershoot_pct = arg_parse_uint(&arg);
- else if (arg_match(&arg, &overshoot_pct, argi))
- cfg.rc_overshoot_pct = arg_parse_uint(&arg);
- else if (arg_match(&arg, &buf_sz, argi))
- cfg.rc_buf_sz = arg_parse_uint(&arg);
- else if (arg_match(&arg, &buf_initial_sz, argi))
- cfg.rc_buf_initial_sz = arg_parse_uint(&arg);
- else if (arg_match(&arg, &buf_optimal_sz, argi))
- cfg.rc_buf_optimal_sz = arg_parse_uint(&arg);
- else if (arg_match(&arg, &bias_pct, argi))
- {
- cfg.rc_2pass_vbr_bias_pct = arg_parse_uint(&arg);
-
- if (arg_passes < 2)
- fprintf(stderr,
- "Warning: option %s ignored in one-pass mode.\n",
- arg.name);
- }
- else if (arg_match(&arg, &minsection_pct, argi))
- {
- cfg.rc_2pass_vbr_minsection_pct = arg_parse_uint(&arg);
-
- if (arg_passes < 2)
- fprintf(stderr,
- "Warning: option %s ignored in one-pass mode.\n",
- arg.name);
- }
- else if (arg_match(&arg, &maxsection_pct, argi))
- {
- cfg.rc_2pass_vbr_maxsection_pct = arg_parse_uint(&arg);
-
- if (arg_passes < 2)
- fprintf(stderr,
- "Warning: option %s ignored in one-pass mode.\n",
- arg.name);
- }
- else if (arg_match(&arg, &kf_min_dist, argi))
- cfg.kf_min_dist = arg_parse_uint(&arg);
- else if (arg_match(&arg, &kf_max_dist, argi))
- cfg.kf_max_dist = arg_parse_uint(&arg);
- else if (arg_match(&arg, &kf_disabled, argi))
- cfg.kf_mode = VPX_KF_DISABLED;
- else
- argj++;
- }
-
- /* Handle codec specific options */
-#if CONFIG_VP8_ENCODER
-
- if (codec->iface == &vpx_codec_vp8_cx_algo)
- {
- ctrl_args = vp8_args;
- ctrl_args_map = vp8_arg_ctrl_map;
- }
-
-#endif
-
- for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step)
- {
- int match = 0;
-
- arg.argv_step = 1;
-
- for (i = 0; ctrl_args[i]; i++)
- {
- if (arg_match(&arg, ctrl_args[i], argi))
- {
- match = 1;
-
- if (arg_ctrl_cnt < ARG_CTRL_CNT_MAX)
- {
- arg_ctrls[arg_ctrl_cnt][0] = ctrl_args_map[i];
- arg_ctrls[arg_ctrl_cnt][1] = arg_parse_enum_or_int(&arg);
- arg_ctrl_cnt++;
- }
- }
- }
-
- if (!match)
- argj++;
- }
-
- /* Check for unrecognized options */
- for (argi = argv; *argi; argi++)
- if (argi[0][0] == '-' && argi[0][1])
- die("Error: Unrecognized option %s\n", *argi);
-
- /* Handle non-option arguments */
- in_fn = argv[0];
-
- if (!in_fn)
- usage_exit();
-
- if(!out_fn)
- die("Error: Output file is required (specify with -o)\n");
-
- memset(&stats, 0, sizeof(stats));
-
- for (pass = one_pass_only ? one_pass_only - 1 : 0; pass < arg_passes; pass++)
- {
- int frames_in = 0, frames_out = 0;
- unsigned long nbytes = 0;
- struct detect_buffer detect;
-
- /* Parse certain options from the input file, if possible */
- infile = strcmp(in_fn, "-") ? fopen(in_fn, "rb")
- : set_binary_mode(stdin);
-
- if (!infile)
- {
- fprintf(stderr, "Failed to open input file\n");
- return EXIT_FAILURE;
- }
-
- /* For RAW input sources, these bytes will applied on the first frame
- * in read_frame().
- */
- detect.buf_read = fread(detect.buf, 1, 4, infile);
- detect.position = 0;
-
- if (detect.buf_read == 4 && file_is_y4m(infile, &y4m, detect.buf))
- {
- if (y4m_input_open(&y4m, infile, detect.buf, 4) >= 0)
- {
- file_type = FILE_TYPE_Y4M;
- cfg.g_w = y4m.pic_w;
- cfg.g_h = y4m.pic_h;
-
- /* Use the frame rate from the file only if none was specified
- * on the command-line.
- */
- if (!arg_have_framerate)
- {
- arg_framerate.num = y4m.fps_n;
- arg_framerate.den = y4m.fps_d;
- }
-
- arg_use_i420 = 0;
- }
- else
- {
- fprintf(stderr, "Unsupported Y4M stream.\n");
- return EXIT_FAILURE;
- }
- }
- else if (detect.buf_read == 4 &&
- file_is_ivf(infile, &fourcc, &cfg.g_w, &cfg.g_h, &detect))
- {
- file_type = FILE_TYPE_IVF;
- switch (fourcc)
- {
- case 0x32315659:
- arg_use_i420 = 0;
- break;
- case 0x30323449:
- arg_use_i420 = 1;
- break;
- default:
- fprintf(stderr, "Unsupported fourcc (%08x) in IVF\n", fourcc);
- return EXIT_FAILURE;
- }
- }
- else
- {
- file_type = FILE_TYPE_RAW;
- }
-
- if(!cfg.g_w || !cfg.g_h)
- {
- fprintf(stderr, "Specify stream dimensions with --width (-w) "
- " and --height (-h).\n");
- return EXIT_FAILURE;
- }
-
-#define SHOW(field) fprintf(stderr, " %-28s = %d\n", #field, cfg.field)
-
- if (verbose && pass == 0)
- {
- fprintf(stderr, "Codec: %s\n", vpx_codec_iface_name(codec->iface));
- fprintf(stderr, "Source file: %s Format: %s\n", in_fn,
- arg_use_i420 ? "I420" : "YV12");
- fprintf(stderr, "Destination file: %s\n", out_fn);
- fprintf(stderr, "Encoder parameters:\n");
-
- SHOW(g_usage);
- SHOW(g_threads);
- SHOW(g_profile);
- SHOW(g_w);
- SHOW(g_h);
- SHOW(g_timebase.num);
- SHOW(g_timebase.den);
- SHOW(g_error_resilient);
- SHOW(g_pass);
- SHOW(g_lag_in_frames);
- SHOW(rc_dropframe_thresh);
- SHOW(rc_resize_allowed);
- SHOW(rc_resize_up_thresh);
- SHOW(rc_resize_down_thresh);
- SHOW(rc_end_usage);
- SHOW(rc_target_bitrate);
- SHOW(rc_min_quantizer);
- SHOW(rc_max_quantizer);
- SHOW(rc_undershoot_pct);
- SHOW(rc_overshoot_pct);
- SHOW(rc_buf_sz);
- SHOW(rc_buf_initial_sz);
- SHOW(rc_buf_optimal_sz);
- SHOW(rc_2pass_vbr_bias_pct);
- SHOW(rc_2pass_vbr_minsection_pct);
- SHOW(rc_2pass_vbr_maxsection_pct);
- SHOW(kf_mode);
- SHOW(kf_min_dist);
- SHOW(kf_max_dist);
- }
-
- if(pass == (one_pass_only ? one_pass_only - 1 : 0)) {
- if (file_type == FILE_TYPE_Y4M)
- /*The Y4M reader does its own allocation.
- Just initialize this here to avoid problems if we never read any
- frames.*/
- memset(&raw, 0, sizeof(raw));
- else
- vpx_img_alloc(&raw, arg_use_i420 ? VPX_IMG_FMT_I420 : VPX_IMG_FMT_YV12,
- cfg.g_w, cfg.g_h, 1);
- }
-
- outfile = strcmp(out_fn, "-") ? fopen(out_fn, "wb")
- : set_binary_mode(stdout);
-
- if (!outfile)
- {
- fprintf(stderr, "Failed to open output file\n");
- return EXIT_FAILURE;
- }
-
- if(write_webm && fseek(outfile, 0, SEEK_CUR))
- {
- fprintf(stderr, "WebM output to pipes not supported.\n");
- return EXIT_FAILURE;
- }
-
- if (stats_fn)
- {
- if (!stats_open_file(&stats, stats_fn, pass))
- {
- fprintf(stderr, "Failed to open statistics store\n");
- return EXIT_FAILURE;
- }
- }
- else
- {
- if (!stats_open_mem(&stats, pass))
- {
- fprintf(stderr, "Failed to open statistics store\n");
- return EXIT_FAILURE;
- }
- }
-
- cfg.g_pass = arg_passes == 2
- ? pass ? VPX_RC_LAST_PASS : VPX_RC_FIRST_PASS
- : VPX_RC_ONE_PASS;
-#if VPX_ENCODER_ABI_VERSION > (1 + VPX_CODEC_ABI_VERSION)
-
- if (pass)
- {
- cfg.rc_twopass_stats_in = stats_get(&stats);
- }
-
-#endif
-
- if(write_webm)
- {
- ebml.stream = outfile;
- write_webm_file_header(&ebml, &cfg, &arg_framerate);
- }
- else
- write_ivf_file_header(outfile, &cfg, codec->fourcc, 0);
-
-
- /* Construct Encoder Context */
- vpx_codec_enc_init(&encoder, codec->iface, &cfg,
- show_psnr ? VPX_CODEC_USE_PSNR : 0);
- ctx_exit_on_error(&encoder, "Failed to initialize encoder");
-
- /* Note that we bypass the vpx_codec_control wrapper macro because
- * we're being clever to store the control IDs in an array. Real
- * applications will want to make use of the enumerations directly
- */
- for (i = 0; i < arg_ctrl_cnt; i++)
- {
- if (vpx_codec_control_(&encoder, arg_ctrls[i][0], arg_ctrls[i][1]))
- fprintf(stderr, "Error: Tried to set control %d = %d\n",
- arg_ctrls[i][0], arg_ctrls[i][1]);
-
- ctx_exit_on_error(&encoder, "Failed to control codec");
- }
-
- frame_avail = 1;
- got_data = 0;
-
- while (frame_avail || got_data)
- {
- vpx_codec_iter_t iter = NULL;
- const vpx_codec_cx_pkt_t *pkt;
- struct vpx_usec_timer timer;
- int64_t frame_start, next_frame_start;
-
- if (!arg_limit || frames_in < arg_limit)
- {
- frame_avail = read_frame(infile, &raw, file_type, &y4m,
- &detect);
-
- if (frame_avail)
- frames_in++;
-
- fprintf(stderr,
- "\rPass %d/%d frame %4d/%-4d %7ldB \033[K", pass + 1,
- arg_passes, frames_in, frames_out, nbytes);
- }
- else
- frame_avail = 0;
-
- vpx_usec_timer_start(&timer);
-
- frame_start = (cfg.g_timebase.den * (int64_t)(frames_in - 1)
- * arg_framerate.den) / cfg.g_timebase.num / arg_framerate.num;
- next_frame_start = (cfg.g_timebase.den * (int64_t)(frames_in)
- * arg_framerate.den)
- / cfg.g_timebase.num / arg_framerate.num;
- vpx_codec_encode(&encoder, frame_avail ? &raw : NULL, frame_start,
- next_frame_start - frame_start,
- 0, arg_deadline);
- vpx_usec_timer_mark(&timer);
- cx_time += vpx_usec_timer_elapsed(&timer);
- ctx_exit_on_error(&encoder, "Failed to encode frame");
- got_data = 0;
-
- while ((pkt = vpx_codec_get_cx_data(&encoder, &iter)))
- {
- got_data = 1;
-
- switch (pkt->kind)
- {
- case VPX_CODEC_CX_FRAME_PKT:
- frames_out++;
- fprintf(stderr, " %6luF",
- (unsigned long)pkt->data.frame.sz);
-
- if(write_webm)
- {
- /* Update the hash */
- if(!ebml.debug)
- hash = murmur(pkt->data.frame.buf,
- pkt->data.frame.sz, hash);
-
- write_webm_block(&ebml, &cfg, pkt);
- }
- else
- {
- write_ivf_frame_header(outfile, pkt);
- if(fwrite(pkt->data.frame.buf, 1,
- pkt->data.frame.sz, outfile));
- }
- nbytes += pkt->data.raw.sz;
- break;
- case VPX_CODEC_STATS_PKT:
- frames_out++;
- fprintf(stderr, " %6luS",
- (unsigned long)pkt->data.twopass_stats.sz);
- stats_write(&stats,
- pkt->data.twopass_stats.buf,
- pkt->data.twopass_stats.sz);
- nbytes += pkt->data.raw.sz;
- break;
- case VPX_CODEC_PSNR_PKT:
-
- if (show_psnr)
- {
- int i;
-
- psnr_sse_total += pkt->data.psnr.sse[0];
- psnr_samples_total += pkt->data.psnr.samples[0];
- for (i = 0; i < 4; i++)
- {
- fprintf(stderr, "%.3lf ", pkt->data.psnr.psnr[i]);
- psnr_totals[i] += pkt->data.psnr.psnr[i];
- }
- psnr_count++;
- }
-
- break;
- default:
- break;
- }
- }
-
- fflush(stdout);
- }
-
- fprintf(stderr,
- "\rPass %d/%d frame %4d/%-4d %7ldB %7ldb/f %7"PRId64"b/s"
- " %7lu %s (%.2f fps)\033[K", pass + 1,
- arg_passes, frames_in, frames_out, nbytes, nbytes * 8 / frames_in,
- nbytes * 8 *(int64_t)arg_framerate.num / arg_framerate.den / frames_in,
- cx_time > 9999999 ? cx_time / 1000 : cx_time,
- cx_time > 9999999 ? "ms" : "us",
- (float)frames_in * 1000000.0 / (float)cx_time);
-
- if ( (show_psnr) && (psnr_count>0) )
- {
- int i;
- double ovpsnr = vp8_mse2psnr(psnr_samples_total, 255.0,
- psnr_sse_total);
-
- fprintf(stderr, "\nPSNR (Overall/Avg/Y/U/V)");
-
- fprintf(stderr, " %.3lf", ovpsnr);
- for (i = 0; i < 4; i++)
- {
- fprintf(stderr, " %.3lf", psnr_totals[i]/psnr_count);
- }
- }
-
- vpx_codec_destroy(&encoder);
-
- fclose(infile);
-
- if(write_webm)
- {
- write_webm_file_footer(&ebml, hash);
- }
- else
- {
- if (!fseek(outfile, 0, SEEK_SET))
- write_ivf_file_header(outfile, &cfg, codec->fourcc, frames_out);
- }
-
- fclose(outfile);
- stats_close(&stats, arg_passes-1);
- fprintf(stderr, "\n");
-
- if (one_pass_only)
- break;
- }
-
- vpx_img_free(&raw);
- free(argv);
- return EXIT_SUCCESS;
-}
diff --git a/y4minput.c b/y4minput.c
deleted file mode 100644
index 449afe8..0000000
--- a/y4minput.c
+++ /dev/null
@@ -1,881 +0,0 @@
-/*
- * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- *
- * Based on code from the OggTheora software codec source code,
- * Copyright (C) 2002-2010 The Xiph.Org Foundation and contributors.
- */
-#include <stdlib.h>
-#include <string.h>
-#include "y4minput.h"
-
-static int y4m_parse_tags(y4m_input *_y4m,char *_tags){
- int got_w;
- int got_h;
- int got_fps;
- int got_interlace;
- int got_par;
- int got_chroma;
- char *p;
- char *q;
- got_w=got_h=got_fps=got_interlace=got_par=got_chroma=0;
- for(p=_tags;;p=q){
- /*Skip any leading spaces.*/
- while(*p==' ')p++;
- /*If that's all we have, stop.*/
- if(p[0]=='\0')break;
- /*Find the end of this tag.*/
- for(q=p+1;*q!='\0'&&*q!=' ';q++);
- /*Process the tag.*/
- switch(p[0]){
- case 'W':{
- if(sscanf(p+1,"%d",&_y4m->pic_w)!=1)return -1;
- got_w=1;
- }break;
- case 'H':{
- if(sscanf(p+1,"%d",&_y4m->pic_h)!=1)return -1;
- got_h=1;
- }break;
- case 'F':{
- if(sscanf(p+1,"%d:%d",&_y4m->fps_n,&_y4m->fps_d)!=2){
- return -1;
- }
- got_fps=1;
- }break;
- case 'I':{
- _y4m->interlace=p[1];
- got_interlace=1;
- }break;
- case 'A':{
- if(sscanf(p+1,"%d:%d",&_y4m->par_n,&_y4m->par_d)!=2){
- return -1;
- }
- got_par=1;
- }break;
- case 'C':{
- if(q-p>16)return -1;
- memcpy(_y4m->chroma_type,p+1,q-p-1);
- _y4m->chroma_type[q-p-1]='\0';
- got_chroma=1;
- }break;
- /*Ignore unknown tags.*/
- }
- }
- if(!got_w||!got_h||!got_fps)return -1;
- if(!got_interlace)_y4m->interlace='?';
- if(!got_par)_y4m->par_n=_y4m->par_d=0;
- /*Chroma-type is not specified in older files, e.g., those generated by
- mplayer.*/
- if(!got_chroma)strcpy(_y4m->chroma_type,"420");
- return 0;
-}
-
-
-
-/*All anti-aliasing filters in the following conversion functions are based on
- one of two window functions:
- The 6-tap Lanczos window (for down-sampling and shifts):
- sinc(\pi*t)*sinc(\pi*t/3), |t|<3 (sinc(t)==sin(t)/t)
- 0, |t|>=3
- The 4-tap Mitchell window (for up-sampling):
- 7|t|^3-12|t|^2+16/3, |t|<1
- -(7/3)|x|^3+12|x|^2-20|x|+32/3, |t|<2
- 0, |t|>=2
- The number of taps is intentionally kept small to reduce computational
- overhead and limit ringing.
-
- The taps from these filters are scaled so that their sum is 1, and the result
- is scaled by 128 and rounded to integers to create a filter whose
- intermediate values fit inside 16 bits.
- Coefficients are rounded in such a way as to ensure their sum is still 128,
- which is usually equivalent to normal rounding.
-
- Conversions which require both horizontal and vertical filtering could
- have these steps pipelined, for less memory consumption and better cache
- performance, but we do them separately for simplicity.*/
-
-#define OC_MINI(_a,_b) ((_a)>(_b)?(_b):(_a))
-#define OC_MAXI(_a,_b) ((_a)<(_b)?(_b):(_a))
-#define OC_CLAMPI(_a,_b,_c) (OC_MAXI(_a,OC_MINI(_b,_c)))
-
-/*420jpeg chroma samples are sited like:
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
-
- 420mpeg2 chroma samples are sited like:
- Y-------Y-------Y-------Y-------
- | | | |
- BR | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- BR | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
-
- We use a resampling filter to shift the site locations one quarter pixel (at
- the chroma plane's resolution) to the right.
- The 4:2:2 modes look exactly the same, except there are twice as many chroma
- lines, and they are vertically co-sited with the luma samples in both the
- mpeg2 and jpeg cases (thus requiring no vertical resampling).*/
-static void y4m_42xmpeg2_42xjpeg_helper(unsigned char *_dst,
- const unsigned char *_src,int _c_w,int _c_h){
- int pli;
- int y;
- int x;
- for(y=0;y<_c_h;y++){
- /*Filter: [4 -17 114 35 -9 1]/128, derived from a 6-tap Lanczos
- window.*/
- for(x=0;x<OC_MINI(_c_w,2);x++){
- _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[0]-17*_src[OC_MAXI(x-1,0)]+
- 114*_src[x]+35*_src[OC_MINI(x+1,_c_w-1)]-9*_src[OC_MINI(x+2,_c_w-1)]+
- _src[OC_MINI(x+3,_c_w-1)]+64)>>7,255);
- }
- for(;x<_c_w-3;x++){
- _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[x-2]-17*_src[x-1]+
- 114*_src[x]+35*_src[x+1]-9*_src[x+2]+_src[x+3]+64)>>7,255);
- }
- for(;x<_c_w;x++){
- _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[x-2]-17*_src[x-1]+
- 114*_src[x]+35*_src[OC_MINI(x+1,_c_w-1)]-9*_src[OC_MINI(x+2,_c_w-1)]+
- _src[_c_w-1]+64)>>7,255);
- }
- _dst+=_c_w;
- _src+=_c_w;
- }
-}
-
-/*Handles both 422 and 420mpeg2 to 422jpeg and 420jpeg, respectively.*/
-static void y4m_convert_42xmpeg2_42xjpeg(y4m_input *_y4m,unsigned char *_dst,
- unsigned char *_aux){
- int c_w;
- int c_h;
- int c_sz;
- int pli;
- int y;
- int x;
- /*Skip past the luma data.*/
- _dst+=_y4m->pic_w*_y4m->pic_h;
- /*Compute the size of each chroma plane.*/
- c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
- c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
- c_sz=c_w*c_h;
- for(pli=1;pli<3;pli++){
- y4m_42xmpeg2_42xjpeg_helper(_dst,_aux,c_w,c_h);
- _dst+=c_sz;
- _aux+=c_sz;
- }
-}
-
-/*This format is only used for interlaced content, but is included for
- completeness.
-
- 420jpeg chroma samples are sited like:
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
-
- 420paldv chroma samples are sited like:
- YR------Y-------YR------Y-------
- | | | |
- | | | |
- | | | |
- YB------Y-------YB------Y-------
- | | | |
- | | | |
- | | | |
- YR------Y-------YR------Y-------
- | | | |
- | | | |
- | | | |
- YB------Y-------YB------Y-------
- | | | |
- | | | |
- | | | |
-
- We use a resampling filter to shift the site locations one quarter pixel (at
- the chroma plane's resolution) to the right.
- Then we use another filter to move the C_r location down one quarter pixel,
- and the C_b location up one quarter pixel.*/
-static void y4m_convert_42xpaldv_42xjpeg(y4m_input *_y4m,unsigned char *_dst,
- unsigned char *_aux){
- unsigned char *tmp;
- int c_w;
- int c_h;
- int c_sz;
- int pli;
- int y;
- int x;
- /*Skip past the luma data.*/
- _dst+=_y4m->pic_w*_y4m->pic_h;
- /*Compute the size of each chroma plane.*/
- c_w=(_y4m->pic_w+1)/2;
- c_h=(_y4m->pic_h+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
- c_sz=c_w*c_h;
- tmp=_aux+2*c_sz;
- for(pli=1;pli<3;pli++){
- /*First do the horizontal re-sampling.
- This is the same as the mpeg2 case, except that after the horizontal
- case, we need to apply a second vertical filter.*/
- y4m_42xmpeg2_42xjpeg_helper(tmp,_aux,c_w,c_h);
- _aux+=c_sz;
- switch(pli){
- case 1:{
- /*Slide C_b up a quarter-pel.
- This is the same filter used above, but in the other order.*/
- for(x=0;x<c_w;x++){
- for(y=0;y<OC_MINI(c_h,3);y++){
- _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[0]
- -9*tmp[OC_MAXI(y-2,0)*c_w]+35*tmp[OC_MAXI(y-1,0)*c_w]
- +114*tmp[y*c_w]-17*tmp[OC_MINI(y+1,c_h-1)*c_w]
- +4*tmp[OC_MINI(y+2,c_h-1)*c_w]+64)>>7,255);
- }
- for(;y<c_h-2;y++){
- _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[(y-3)*c_w]
- -9*tmp[(y-2)*c_w]+35*tmp[(y-1)*c_w]+114*tmp[y*c_w]
- -17*tmp[(y+1)*c_w]+4*tmp[(y+2)*c_w]+64)>>7,255);
- }
- for(;y<c_h;y++){
- _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[(y-3)*c_w]
- -9*tmp[(y-2)*c_w]+35*tmp[(y-1)*c_w]+114*tmp[y*c_w]
- -17*tmp[OC_MINI(y+1,c_h-1)*c_w]+4*tmp[(c_h-1)*c_w]+64)>>7,255);
- }
- _dst++;
- tmp++;
- }
- _dst+=c_sz-c_w;
- tmp-=c_w;
- }break;
- case 2:{
- /*Slide C_r down a quarter-pel.
- This is the same as the horizontal filter.*/
- for(x=0;x<c_w;x++){
- for(y=0;y<OC_MINI(c_h,2);y++){
- _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[0]
- -17*tmp[OC_MAXI(y-1,0)*c_w]+114*tmp[y*c_w]
- +35*tmp[OC_MINI(y+1,c_h-1)*c_w]-9*tmp[OC_MINI(y+2,c_h-1)*c_w]
- +tmp[OC_MINI(y+3,c_h-1)*c_w]+64)>>7,255);
- }
- for(;y<c_h-3;y++){
- _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[(y-2)*c_w]
- -17*tmp[(y-1)*c_w]+114*tmp[y*c_w]+35*tmp[(y+1)*c_w]
- -9*tmp[(y+2)*c_w]+tmp[(y+3)*c_w]+64)>>7,255);
- }
- for(;y<c_h;y++){
- _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[(y-2)*c_w]
- -17*tmp[(y-1)*c_w]+114*tmp[y*c_w]+35*tmp[OC_MINI(y+1,c_h-1)*c_w]
- -9*tmp[OC_MINI(y+2,c_h-1)*c_w]+tmp[(c_h-1)*c_w]+64)>>7,255);
- }
- _dst++;
- tmp++;
- }
- }break;
- }
- /*For actual interlaced material, this would have to be done separately on
- each field, and the shift amounts would be different.
- C_r moves down 1/8, C_b up 3/8 in the top field, and C_r moves down 3/8,
- C_b up 1/8 in the bottom field.
- The corresponding filters would be:
- Down 1/8 (reverse order for up): [3 -11 125 15 -4 0]/128
- Down 3/8 (reverse order for up): [4 -19 98 56 -13 2]/128*/
- }
-}
-
-/*Perform vertical filtering to reduce a single plane from 4:2:2 to 4:2:0.
- This is used as a helper by several converation routines.*/
-static void y4m_422jpeg_420jpeg_helper(unsigned char *_dst,
- const unsigned char *_src,int _c_w,int _c_h){
- int y;
- int x;
- /*Filter: [3 -17 78 78 -17 3]/128, derived from a 6-tap Lanczos window.*/
- for(x=0;x<_c_w;x++){
- for(y=0;y<OC_MINI(_c_h,2);y+=2){
- _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(64*_src[0]
- +78*_src[OC_MINI(1,_c_h-1)*_c_w]
- -17*_src[OC_MINI(2,_c_h-1)*_c_w]
- +3*_src[OC_MINI(3,_c_h-1)*_c_w]+64)>>7,255);
- }
- for(;y<_c_h-3;y+=2){
- _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(3*(_src[(y-2)*_c_w]+_src[(y+3)*_c_w])
- -17*(_src[(y-1)*_c_w]+_src[(y+2)*_c_w])
- +78*(_src[y*_c_w]+_src[(y+1)*_c_w])+64)>>7,255);
- }
- for(;y<_c_h;y+=2){
- _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(3*(_src[(y-2)*_c_w]
- +_src[(_c_h-1)*_c_w])-17*(_src[(y-1)*_c_w]
- +_src[OC_MINI(y+2,_c_h-1)*_c_w])
- +78*(_src[y*_c_w]+_src[OC_MINI(y+1,_c_h-1)*_c_w])+64)>>7,255);
- }
- _src++;
- _dst++;
- }
-}
-
-/*420jpeg chroma samples are sited like:
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
-
- 422jpeg chroma samples are sited like:
- Y---BR--Y-------Y---BR--Y-------
- | | | |
- | | | |
- | | | |
- Y---BR--Y-------Y---BR--Y-------
- | | | |
- | | | |
- | | | |
- Y---BR--Y-------Y---BR--Y-------
- | | | |
- | | | |
- | | | |
- Y---BR--Y-------Y---BR--Y-------
- | | | |
- | | | |
- | | | |
-
- We use a resampling filter to decimate the chroma planes by two in the
- vertical direction.*/
-static void y4m_convert_422jpeg_420jpeg(y4m_input *_y4m,unsigned char *_dst,
- unsigned char *_aux){
- int c_w;
- int c_h;
- int c_sz;
- int dst_c_w;
- int dst_c_h;
- int dst_c_sz;
- int tmp_sz;
- int pic_sz;
- int pli;
- /*Skip past the luma data.*/
- _dst+=_y4m->pic_w*_y4m->pic_h;
- /*Compute the size of each chroma plane.*/
- c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h;
- c_h=_y4m->pic_h;
- dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
- dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
- c_sz=c_w*c_h;
- dst_c_sz=dst_c_w*dst_c_h;
- for(pli=1;pli<3;pli++){
- y4m_422jpeg_420jpeg_helper(_dst,_aux,c_w,c_h);
- _aux+=c_sz;
- _dst+=dst_c_sz;
- }
-}
-
-/*420jpeg chroma samples are sited like:
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
-
- 422 chroma samples are sited like:
- YBR-----Y-------YBR-----Y-------
- | | | |
- | | | |
- | | | |
- YBR-----Y-------YBR-----Y-------
- | | | |
- | | | |
- | | | |
- YBR-----Y-------YBR-----Y-------
- | | | |
- | | | |
- | | | |
- YBR-----Y-------YBR-----Y-------
- | | | |
- | | | |
- | | | |
-
- We use a resampling filter to shift the original site locations one quarter
- pixel (at the original chroma resolution) to the right.
- Then we use a second resampling filter to decimate the chroma planes by two
- in the vertical direction.*/
-static void y4m_convert_422_420jpeg(y4m_input *_y4m,unsigned char *_dst,
- unsigned char *_aux){
- unsigned char *tmp;
- int c_w;
- int c_h;
- int c_sz;
- int dst_c_w;
- int dst_c_h;
- int dst_c_sz;
- int pli;
- int y;
- int x;
- /*Skip past the luma data.*/
- _dst+=_y4m->pic_w*_y4m->pic_h;
- /*Compute the size of each chroma plane.*/
- c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h;
- c_h=_y4m->pic_h;
- dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
- c_sz=c_w*c_h;
- dst_c_sz=c_w*dst_c_h;
- tmp=_aux+2*c_sz;
- for(pli=1;pli<3;pli++){
- /*In reality, the horizontal and vertical steps could be pipelined, for
- less memory consumption and better cache performance, but we do them
- separately for simplicity.*/
- /*First do horizontal filtering (convert to 422jpeg)*/
- y4m_42xmpeg2_42xjpeg_helper(tmp,_aux,c_w,c_h);
- /*Now do the vertical filtering.*/
- y4m_422jpeg_420jpeg_helper(_dst,tmp,c_w,c_h);
- _aux+=c_sz;
- _dst+=dst_c_sz;
- }
-}
-
-/*420jpeg chroma samples are sited like:
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | BR | | BR |
- | | | |
- Y-------Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
-
- 411 chroma samples are sited like:
- YBR-----Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- YBR-----Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- YBR-----Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
- YBR-----Y-------Y-------Y-------
- | | | |
- | | | |
- | | | |
-
- We use a filter to resample at site locations one eighth pixel (at the source
- chroma plane's horizontal resolution) and five eighths of a pixel to the
- right.
- Then we use another filter to decimate the planes by 2 in the vertical
- direction.*/
-static void y4m_convert_411_420jpeg(y4m_input *_y4m,unsigned char *_dst,
- unsigned char *_aux){
- unsigned char *tmp;
- int c_w;
- int c_h;
- int c_sz;
- int dst_c_w;
- int dst_c_h;
- int dst_c_sz;
- int tmp_sz;
- int pli;
- int y;
- int x;
- /*Skip past the luma data.*/
- _dst+=_y4m->pic_w*_y4m->pic_h;
- /*Compute the size of each chroma plane.*/
- c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h;
- c_h=_y4m->pic_h;
- dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
- dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
- c_sz=c_w*c_h;
- dst_c_sz=dst_c_w*dst_c_h;
- tmp_sz=dst_c_w*c_h;
- tmp=_aux+2*c_sz;
- for(pli=1;pli<3;pli++){
- /*In reality, the horizontal and vertical steps could be pipelined, for
- less memory consumption and better cache performance, but we do them
- separately for simplicity.*/
- /*First do horizontal filtering (convert to 422jpeg)*/
- for(y=0;y<c_h;y++){
- /*Filters: [1 110 18 -1]/128 and [-3 50 86 -5]/128, both derived from a
- 4-tap Mitchell window.*/
- for(x=0;x<OC_MINI(c_w,1);x++){
- tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(111*_aux[0]
- +18*_aux[OC_MINI(1,c_w-1)]-_aux[OC_MINI(2,c_w-1)]+64)>>7,255);
- tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(47*_aux[0]
- +86*_aux[OC_MINI(1,c_w-1)]-5*_aux[OC_MINI(2,c_w-1)]+64)>>7,255);
- }
- for(;x<c_w-2;x++){
- tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(_aux[x-1]+110*_aux[x]
- +18*_aux[x+1]-_aux[x+2]+64)>>7,255);
- tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(-3*_aux[x-1]+50*_aux[x]
- +86*_aux[x+1]-5*_aux[x+2]+64)>>7,255);
- }
- for(;x<c_w;x++){
- tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(_aux[x-1]+110*_aux[x]
- +18*_aux[OC_MINI(x+1,c_w-1)]-_aux[c_w-1]+64)>>7,255);
- if((x<<1|1)<dst_c_w){
- tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(-3*_aux[x-1]+50*_aux[x]
- +86*_aux[OC_MINI(x+1,c_w-1)]-5*_aux[c_w-1]+64)>>7,255);
- }
- }
- tmp+=dst_c_w;
- _aux+=c_w;
- }
- tmp-=tmp_sz;
- /*Now do the vertical filtering.*/
- y4m_422jpeg_420jpeg_helper(_dst,tmp,dst_c_w,c_h);
- _dst+=dst_c_sz;
- }
-}
-
-/*Convert 444 to 420jpeg.*/
-static void y4m_convert_444_420jpeg(y4m_input *_y4m,unsigned char *_dst,
- unsigned char *_aux){
- unsigned char *tmp;
- int c_w;
- int c_h;
- int c_sz;
- int dst_c_w;
- int dst_c_h;
- int dst_c_sz;
- int tmp_sz;
- int pli;
- int y;
- int x;
- /*Skip past the luma data.*/
- _dst+=_y4m->pic_w*_y4m->pic_h;
- /*Compute the size of each chroma plane.*/
- c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h;
- c_h=_y4m->pic_h;
- dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
- dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
- c_sz=c_w*c_h;
- dst_c_sz=dst_c_w*dst_c_h;
- tmp_sz=dst_c_w*c_h;
- tmp=_aux+2*c_sz;
- for(pli=1;pli<3;pli++){
- /*Filter: [3 -17 78 78 -17 3]/128, derived from a 6-tap Lanczos window.*/
- for(y=0;y<c_h;y++){
- for(x=0;x<OC_MINI(c_w,2);x+=2){
- tmp[x>>1]=OC_CLAMPI(0,(64*_aux[0]+78*_aux[OC_MINI(1,c_w-1)]
- -17*_aux[OC_MINI(2,c_w-1)]
- +3*_aux[OC_MINI(3,c_w-1)]+64)>>7,255);
- }
- for(;x<c_w-3;x+=2){
- tmp[x>>1]=OC_CLAMPI(0,(3*(_aux[x-2]+_aux[x+3])
- -17*(_aux[x-1]+_aux[x+2])+78*(_aux[x]+_aux[x+1])+64)>>7,255);
- }
- for(;x<c_w;x+=2){
- tmp[x>>1]=OC_CLAMPI(0,(3*(_aux[x-2]+_aux[c_w-1])-
- 17*(_aux[x-1]+_aux[OC_MINI(x+2,c_w-1)])+
- 78*(_aux[x]+_aux[OC_MINI(x+1,c_w-1)])+64)>>7,255);
- }
- tmp+=dst_c_w;
- _aux+=c_w;
- }
- tmp-=tmp_sz;
- /*Now do the vertical filtering.*/
- y4m_422jpeg_420jpeg_helper(_dst,tmp,dst_c_w,c_h);
- _dst+=dst_c_sz;
- }
-}
-
-/*The image is padded with empty chroma components at 4:2:0.*/
-static void y4m_convert_mono_420jpeg(y4m_input *_y4m,unsigned char *_dst,
- unsigned char *_aux){
- int c_sz;
- _dst+=_y4m->pic_w*_y4m->pic_h;
- c_sz=((_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h)*
- ((_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v);
- memset(_dst,128,c_sz*2);
-}
-
-/*No conversion function needed.*/
-static void y4m_convert_null(y4m_input *_y4m,unsigned char *_dst,
- unsigned char *_aux){
-}
-
-int y4m_input_open(y4m_input *_y4m,FILE *_fin,char *_skip,int _nskip){
- char buffer[80];
- int ret;
- int i;
- /*Read until newline, or 80 cols, whichever happens first.*/
- for(i=0;i<79;i++){
- if(_nskip>0){
- buffer[i]=*_skip++;
- _nskip--;
- }
- else{
- ret=fread(buffer+i,1,1,_fin);
- if(ret<1)return -1;
- }
- if(buffer[i]=='\n')break;
- }
- /*We skipped too much header data.*/
- if(_nskip>0)return -1;
- if(i==79){
- fprintf(stderr,"Error parsing header; not a YUV2MPEG2 file?\n");
- return -1;
- }
- buffer[i]='\0';
- if(memcmp(buffer,"YUV4MPEG",8)){
- fprintf(stderr,"Incomplete magic for YUV4MPEG file.\n");
- return -1;
- }
- if(buffer[8]!='2'){
- fprintf(stderr,"Incorrect YUV input file version; YUV4MPEG2 required.\n");
- }
- ret=y4m_parse_tags(_y4m,buffer+5);
- if(ret<0){
- fprintf(stderr,"Error parsing YUV4MPEG2 header.\n");
- return ret;
- }
- if(_y4m->interlace=='?'){
- fprintf(stderr,"Warning: Input video interlacing format unknown; "
- "assuming progressive scan.\n");
- }
- else if(_y4m->interlace!='p'){
- fprintf(stderr,"Input video is interlaced; "
- "Only progressive scan handled.\n");
- return -1;
- }
- if(strcmp(_y4m->chroma_type,"420")==0||
- strcmp(_y4m->chroma_type,"420jpeg")==0){
- _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h
- +2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2);
- /*Natively supported: no conversion required.*/
- _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=0;
- _y4m->convert=y4m_convert_null;
- }
- else if(strcmp(_y4m->chroma_type,"420mpeg2")==0){
- _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
- /*Chroma filter required: read into the aux buf first.*/
- _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=
- 2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2);
- _y4m->convert=y4m_convert_42xmpeg2_42xjpeg;
- }
- else if(strcmp(_y4m->chroma_type,"420paldv")==0){
- _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
- /*Chroma filter required: read into the aux buf first.
- We need to make two filter passes, so we need some extra space in the
- aux buffer.*/
- _y4m->aux_buf_sz=3*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2);
- _y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2);
- _y4m->convert=y4m_convert_42xpaldv_42xjpeg;
- }
- else if(strcmp(_y4m->chroma_type,"422jpeg")==0){
- _y4m->src_c_dec_h=_y4m->dst_c_dec_h=2;
- _y4m->src_c_dec_v=1;
- _y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
- /*Chroma filter required: read into the aux buf first.*/
- _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*_y4m->pic_h;
- _y4m->convert=y4m_convert_422jpeg_420jpeg;
- }
- else if(strcmp(_y4m->chroma_type,"422")==0){
- _y4m->src_c_dec_h=_y4m->dst_c_dec_h=2;
- _y4m->src_c_dec_v=1;
- _y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
- /*Chroma filter required: read into the aux buf first.
- We need to make two filter passes, so we need some extra space in the
- aux buffer.*/
- _y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*_y4m->pic_h;
- _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h;
- _y4m->convert=y4m_convert_422_420jpeg;
- }
- else if(strcmp(_y4m->chroma_type,"411")==0){
- _y4m->src_c_dec_h=4;
- _y4m->dst_c_dec_h=2;
- _y4m->src_c_dec_v=1;
- _y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
- /*Chroma filter required: read into the aux buf first.
- We need to make two filter passes, so we need some extra space in the
- aux buffer.*/
- _y4m->aux_buf_read_sz=2*((_y4m->pic_w+3)/4)*_y4m->pic_h;
- _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h;
- _y4m->convert=y4m_convert_411_420jpeg;
- }
- else if(strcmp(_y4m->chroma_type,"444")==0){
- _y4m->src_c_dec_h=1;
- _y4m->dst_c_dec_h=2;
- _y4m->src_c_dec_v=1;
- _y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
- /*Chroma filter required: read into the aux buf first.
- We need to make two filter passes, so we need some extra space in the
- aux buffer.*/
- _y4m->aux_buf_read_sz=2*_y4m->pic_w*_y4m->pic_h;
- _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h;
- _y4m->convert=y4m_convert_444_420jpeg;
- }
- else if(strcmp(_y4m->chroma_type,"444alpha")==0){
- _y4m->src_c_dec_h=1;
- _y4m->dst_c_dec_h=2;
- _y4m->src_c_dec_v=1;
- _y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
- /*Chroma filter required: read into the aux buf first.
- We need to make two filter passes, so we need some extra space in the
- aux buffer.
- The extra plane also gets read into the aux buf.
- It will be discarded.*/
- _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=3*_y4m->pic_w*_y4m->pic_h;
- _y4m->convert=y4m_convert_444_420jpeg;
- }
- else if(strcmp(_y4m->chroma_type,"mono")==0){
- _y4m->src_c_dec_h=_y4m->src_c_dec_v=0;
- _y4m->dst_c_dec_h=_y4m->dst_c_dec_v=2;
- _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h;
- /*No extra space required, but we need to clear the chroma planes.*/
- _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=0;
- _y4m->convert=y4m_convert_mono_420jpeg;
- }
- else{
- fprintf(stderr,"Unknown chroma sampling type: %s\n",_y4m->chroma_type);
- return -1;
- }
- /*The size of the final frame buffers is always computed from the
- destination chroma decimation type.*/
- _y4m->dst_buf_sz=_y4m->pic_w*_y4m->pic_h
- +2*((_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h)*
- ((_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v);
- _y4m->dst_buf=(unsigned char *)malloc(_y4m->dst_buf_sz);
- _y4m->aux_buf=(unsigned char *)malloc(_y4m->aux_buf_sz);
- return 0;
-}
-
-void y4m_input_close(y4m_input *_y4m){
- free(_y4m->dst_buf);
- free(_y4m->aux_buf);
-}
-
-int y4m_input_fetch_frame(y4m_input *_y4m,FILE *_fin,vpx_image_t *_img){
- char frame[6];
- int pic_sz;
- int frame_c_w;
- int frame_c_h;
- int c_w;
- int c_h;
- int c_sz;
- int ret;
- /*Read and skip the frame header.*/
- ret=fread(frame,1,6,_fin);
- if(ret<6)return 0;
- if(memcmp(frame,"FRAME",5)){
- fprintf(stderr,"Loss of framing in Y4M input data\n");
- return -1;
- }
- if(frame[5]!='\n'){
- char c;
- int j;
- for(j=0;j<79&&fread(&c,1,1,_fin)&&c!='\n';j++);
- if(j==79){
- fprintf(stderr,"Error parsing Y4M frame header\n");
- return -1;
- }
- }
- /*Read the frame data that needs no conversion.*/
- if(fread(_y4m->dst_buf,1,_y4m->dst_buf_read_sz,_fin)!=_y4m->dst_buf_read_sz){
- fprintf(stderr,"Error reading Y4M frame data.\n");
- return -1;
- }
- /*Read the frame data that does need conversion.*/
- if(fread(_y4m->aux_buf,1,_y4m->aux_buf_read_sz,_fin)!=_y4m->aux_buf_read_sz){
- fprintf(stderr,"Error reading Y4M frame data.\n");
- return -1;
- }
- /*Now convert the just read frame.*/
- (*_y4m->convert)(_y4m,_y4m->dst_buf,_y4m->aux_buf);
- /*Fill in the frame buffer pointers.
- We don't use vpx_img_wrap() because it forces padding for odd picture
- sizes, which would require a separate fread call for every row.*/
- memset(_img,0,sizeof(*_img));
- /*Y4M has the planes in Y'CbCr order, which libvpx calls Y, U, and V.*/
- _img->fmt=IMG_FMT_I420;
- _img->w=_img->d_w=_y4m->pic_w;
- _img->h=_img->d_h=_y4m->pic_h;
- /*This is hard-coded to 4:2:0 for now, as that's all VP8 supports.*/
- _img->x_chroma_shift=1;
- _img->y_chroma_shift=1;
- _img->bps=12;
- /*Set up the buffer pointers.*/
- pic_sz=_y4m->pic_w*_y4m->pic_h;
- c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h;
- c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v;
- c_sz=c_w*c_h;
- _img->stride[PLANE_Y]=_y4m->pic_w;
- _img->stride[PLANE_U]=_img->stride[PLANE_V]=c_w;
- _img->planes[PLANE_Y]=_y4m->dst_buf;
- _img->planes[PLANE_U]=_y4m->dst_buf+pic_sz;
- _img->planes[PLANE_V]=_y4m->dst_buf+pic_sz+c_sz;
- return 1;
-}
